Carbon and Coal Gasification

NATO ASI Series

Advanced Science Institutes Series

A Series presenting the results of activities sponsored by the NATO Science

Committee, which aims at the dissemination of advanced scientific and technological
knowledge, with a view to strengthening links between scientific communities.

The Series is published by an international board of publishers in conjunction with the

NATO Scientific Affairs Division

A Life Sciences Plenum Publishing Corporation

B Physics London and New York

C Mathematical and D. Reidel Publishing Company

Physical Sciences Dordrecht and Boston

D Behavioural and Martinus Nijhoff Publishers

Social Sciences DordrechtiBoston/Lancaster
E Applied Sciences

F Computer and Springer-Verlag

Systems Sciences Berlin/Heidelberg/New York
G Ecological Sciences

Series E: Applied Sciences - No. 105

Carbon and Coal
Gasification
Science and Technology

edited by

Jose L. Figueiredo
Faculty of Engineering
University of Porto
4099 Porto Codex
Portugal

Jacob A. Moulijn
Institute for Chemical Technology
University of Amsterdam
Nieuwe Achtergracht 166
1018 WV Amsterdam
The Netherlands

1986 Martinus Nijhoff Publishers

Dordrecht / Boston / Lancaster
Published in cooperation with NATO Scientific Affairs Division
Proceedings of the NATO Advanced Study Institute on Carbon and Coal Gasification
- Science and Technology, Alvor, Portugal, May 20-31, 1985

Library of Congress Cataloging in Publication Data

NATO Advanced Study Institute on Carbon and Coal
Gasification--Science and Technology (1985 :
Alvor, Portugal)
Carbon and coal gasification.

(NATO ASI series. Series E, Applied sciences

v. 105)
"Proceedings of the NATO Advanced Study Institute
on Carbon and Coal Gasification--Science and Technology,
Alvor, Portugal, May 20-31, 1985"--T.p. verso.
"Published in cooperation with NATO Scientific
Affairs Division."
Includes index.
1. Coal gasification--Congresses. 2. Carbon--
Congresses. I. Fiegueiredo, Jose Luis. II. Moulijn,
Jacob A. III. North Atlantic Treaty Organization.
Scientific Affairs Division. IV. Title. V. Series.
TP759.N28 1985 665.7'72 85-32080
ISBN-J3:978-94-010-8448-2

ISBN-I3:978-94-010-8448-2 e-ISBN-I3:978-94-009-4382-7
001: 10.10071978-94-009-4382-7

Distributors for the United States and Canada: Kluwer Academic Publishers,
190 Old Derby Street, Hingham, MA 02043, USA
Distributors for the UK and Ireland: Kluwer Academic Publishers, MTP Press Ltd,
Falcon House, Queen Square, Lancaster LA1 1RN, UK
Distributors for all other countries: Kluwer Academic Publishers Group, Distribution
Center, P.O. Box 322, 3300 AH Dordrecht, The Netherlands

All rights reserved. No part of this publication may be reproduced, stored in a

retrieval system, or transmitted, in any form or by any means, mechanical,
photocopying, recording, or otherwise, without the prior written permission of the
publishers,
Martinus Nijhoff Publishers, P.O. Box 163, 3300 AD Dordrecht, The Netherlands

Copyright © 1986 by Martinus Nijhoff Publishers, Dordrecht

Softcover reprint ofthe hardcover 1st edition 1986
PREFACE

Carbon gasification reactions form the basis of many important

industrial processes, such as the combustion of coal and the produc-
tion of synthesis gas, fuel gases and activated carbons. They are
also involved in metallurgical processes and in the regeneration of
coked catalysts.

Thus, understanding the fundamentals of carbon gasification is

of vital importance for further technological development. Moreover,
the subject is of interdisciplinary nature, involving chemistry, ma-
terials science and chemical engineering. Therefore, it was thought
that an Advanced Study Institute would be fruitful in establishing
the state of the art, in bringing together experts from the various
sectors involved and in identifying areas of required research and
industrial development. Such a meeting was held at Alvor, Portugal,
from the 20th to the 31st May 1985, and the lectures presented the-
reof are collected in this volume.

The present volume is organized in seven chapters. The Intro-

duction presents the carbon gasification reactions an~ their rele-
vance for particular processes and industrial uses. In the second
chapter, the structures of carbon and coal are reviewed, together
with methods of structural, chemical and textural characterization.
Catalysis of carbon gasification is addressed next, and the mecha-
nisms of catalysis by alkali, alkaline-earth and transition metals
are discussed in the context of C-C02, C-H20 and C-H2 reactions. In
the fourth chapter, chemical engineering aspects relevant for gasi-
fication technology are covered. The remaining chapters are concer-
ned with the major industrial applications involving gas-carbon re-
actions:
- Coal gasification to produce fuel gases or synthesis gas;
- The production of cokes for metallurgy and the removal of coke
from catalysts;
- The production of activated carbons.

In addition to the main lectures published here, a number of

shorter communications were presented during the Institute, which
could not be included in this volume for lack of space. Their titles
are listed separately.

All lecturers are to be thanked for their excellent contribu-

tions and for their cooperation in setting up the scientific pro-
gramme of this Institute. In particular, we are indebted to Profes-
sor Philip L. Walker,Jr., for his support in the early stages of the
organization of the meeting. Unfortunately, health problems preven-
ted him from attending. He was nevertheless able to send one of his
lectures, which constitutes the introductory chapter of this book.

The collaboration of Dr. C.A.Bernardo in all stages of the or-

ganization is acknowledged with gratitude. Thanks are also due to
J.J.Orfao, J.I.Martins, S.Alves, Alcina Bernardo, Helena Silva and
Maria Idalina, who made an important contribution to the smooth run-
ning of the meeting, and to Mr. L.Santana and his staff at the Tor-
ralta complex, who made our stay a very pleasant one.

The support of the Scientific Affairs Division of NATO, through

the Advanced Study Institutes Programme, is gratefully acknowledged.

Jose L. Figueiredo
Jacob A. Moulijn
NATO ADVANCED STUDY INSTITUTE "CARBON AND COAL GASIFICATION -
SCIENCE AND TECHNOLOGY"

Organizing Committee

Director: J.L. Figueiredo

Associate Director: C.A. Bernardo
Scientific Advisors: H. Marsh, J.A. Moulijn, K.H. van Heek

Lecturers

R.T.K. Baker, Exxon, NJ, USA

J.L. Figueiredo, Univ. Porto, Portugal
K.H. van Heek, Bergbau-Forschung, F.R. Germany
K.J. Hlittinger, Univ. Karlsruhe, F.R. Germany
A. Linares, Univ. Alicante, Spain
A. Marchand, Univ. Bordeaux I, France
H. Marsh, Univ. Newcastle upon Tyne, England
J.A. Moulijn, Univ. Amsterdam, The Netherlands
A.E. Rodrigues, Univ. Porto, Portugal
F. Rodriguez Reinoso, Univ. Alicante, Spain
K.M. Thomas, British Gas, England
D.L. Trimm, Univ. New South Wales, Australia
T. Wigmans, Norit, The Netherlands
LIST OF CONTRIBUTED PAPERS PRESENTED DURING THE A.S.I. BUT NOT
INCLUDED IN THIS VOLUME:

Characterization of carbons using the TPR technique, J.G.McCarty

- Kinetics of the alkali carbonate catalysed C02 gasification of

activated carbon, F.Kapteijn, M.B.Cerfontain, O.Peer, J.A.Moulijn

- Reactivities and active surface areas of steam-gasified coal chars,

P.D.Causton, B.McEnaney

- Influence of temperature, pressure and gasifying agent on the

KzC03 catalysed and uncatalysed gasification of char, H.J.MUhlen,
W.Schumacher, K.H. van Heek

Coal gasification processes in Turkey, S. UlkU

- The status of coal classification, M.J.Lemos de Sousa

- The potential impact of catalysts in coal gasification, R.Shinnar,

C.A.Feng, A.Avidan

- Catalytic COz activation of almond shells, M.Almela-Alarcon, A.

Linares-Solano, F.Rodriguez-Reinoso, C.Salinas-Martinez de Lecea,
A.Sepulveda-Escribano

- Possible use of C02 adsorption as a "theoretical pre-adsorbate",

M.Almela-Alarcon, A.Linares-Solano, F.Rodriguez-Reinoso, C.Salinas
Martinez de Lecea, A.Sepulveda-Escribano

- Recirculating fluidized bed gasifier, J.B.Escudero, J.J.Pis

- Structural variations and kinetics of coal gasification with car-

bon dioxide, S.Balci

- Experiences with catalan lignites gasification in a fluidized bed

reactor, A.Lucas,J.Arnaldos, J.Casal, L.Puigjaner

- The characterization of the plastic properties of a medium volatile

bituminous coal by means of hot-stage microscopy, P.J.J.Tromp,
J .A.Moulijn

Interaction between potassium and calcium catalysts and mineral

matter during coal gasification, R.E.Quartermaine
TABLE OF CONTENTS

Part I. Introduction

P.L. Walker, Jr.

Carbon gasification - A ubiquitous reaction of great
importance to mankind 3

Part II. Structure and characterization

H. Marsh
Structure in carbons 27

K.M. Thomas
Coal structure 57

A. Marchand
Physico-chemical and structural characterization of carbons 93

A. Linares-Solano
Textural characterization of porous carbons by physical
adsorption of gases 137

Part III. Catalytic gasification

J.A. Moulijn, F. Kapteijn

Catalytic gasification 181

K.J. HUttinger, R. Minges

Alkali metal catalyzed water vapour gasification of
carbon using mineral catalyst raw materials 197

K.J. HUttinger, J. Adler, G. Hermann

Iron catalyzed water vapour gasification of carbon 213

R.T.K. Baker
Metal catalyzed gasification of graphite 231

J.L. Figueiredo, C.A. Bernardo, J.J.M. 6rfao

Gasification of carbon deposited on metallic catalysts 269
Part IV. Chemical Engineering principles

F. Kapteijn, J.A. Moulijn

Kinetics of catalysed and uncatalysed coal gasification 291

A.E. Rodrigues
Single particle models for fluid-solid reactions 361

Part V. Coal gasification

K.H. van Heek

General aspects and engineering principles for technical
application of coal gasification 383

K.H. van Heek

Technical status and new developments of coal gasification
processes 403

K.M. Thomas
Characterisation of coal for gasification processes 421

K.M. Thomas
Some scientific aspects of the development of the slagging
gasifier 439

J.A. Moulijn, P.J.J. Tromp

Coal pyrolysis 455

K.J. HUttinger, A. Michenfelder

Hydrogasification of brown coal - Chemism, kinetics and
technical aspects 485

Part VI. Coke production and removal

H. Marsh
The making of coke in the delayed coker and in the slot
coking oven 511

D.L. Trimm
Coke formation and gasification in refineries 523

D.L. Trimm
The regeneration of coked catalysts 543

Part VII. Activated carbons

T. Wigmans
Fundamentals and practical implications of activated carbon
production by partial gasification of carbonaceous materials 559
F. Rodriguez-Reinoso
Preparation and characterization of activated carbons 601

List of participants 643

Subject index 645

PART I. INTRODUCTION
 3

CARBON GASIFICATION -- A UBIQUITOUS REACTION OF GREAT IMPORTANCE

TO MANKIND

Philip L. Walker, Jr.

Department of Materials Science and Engineering, The

Pennsylvania State University, University Park, PA,
16802

It is generally not recognized how important the many types of

carbon are to our society. Carbon comes in two main crystallograph-
ic forms (Figure 1): graphite and diamond. Graphite has trigonal
bonding of carbon atoms; diamond has tetrahedral bonding. The car-
bon artifacts with which we will be concerned will be polycrysta1-
line, their properties determined by their crystallite size (and
distribution), degree of crystallite alignment, and extent of de-
fect character within the crystallites (vacancies and dislocations).
We will be concerned with carbonaceous solids having trigonal bond-
ing within their crystallites and in some cases, possibly, tetrahe-
dral bonding between crystallites.

There are an infinite number of organic precursors which can be

used to produce carbonaceous solids. These precursors contain

tZ ·5:~
I
I
I
I
"1"
I I I I
I
I
I
I
I
I
I I I I I I I I I

±
I I I I I I I I I
I I I r I I I I I
I I I I I I I

~2p:M I I I I I I
I I I I r I I I I
II II I
I
I
I
I
I
I
I
I
I
I
I
I
I

~r5E:V-Y
Diamond arrangement Graphite arrangement

Fig. 1 - The arrangement of carbon atoms in the graphite and dia-

mond lattices.
 4

heteroatoms and possibly inorganic elements in addition to carbon.

Upon their thermal treatment at elevated temperatures, they more or
less lose their heteroatoms and inorganic elements, yielding a solid
higher in carbon content. The structure of the carbonaceous solid
produced depends upon the structure or structures of the organic
compound(s) in the precursor and the exact thermal treatment to
which the precursor is exposed. If part of the thermal treatment
consists initially of taking a solid (at room temperature) through
a carbonization or charring step in the temperature range from about
400 to 600°C, the nature of the mesophase (or intermediate phase)
formed is of overwhelming importance (1,2). If part of the thermal
treatment consists initially of pyrolysis of a gaseous or vapor
species (the so-called chemical vapor deposition process), the
temperature and pressure used is of overwhelming importance (3).
The structure of the carbonaceous solid produced and the presence
of heteroatoms and inorganic species will determine the ease with
which the solid is gasified. In some processes and uses of carbon-
aceous solids, it is desirable that gasification rates be rapid; in
others, it is equally important that gasification rates be very slow.

We will briefly consider the global chemistry of important

gasification reactions and then their relevance for particular pro-
cesses and uses.

1. CARBON GASIFICATION REACTIONS

1.1 C-0 2 Reaction

Arguably, the combustion of carbon to produce heat is the most

important chemical reaction available to mankind. However, as we
will see shortly, carbon gasification by 02 is most undesirable
when carbon is used as a material. It has been shown that upon the
reaction of carbon with 02, two primary gasification products, CO
and C02, are formed (4).

C + 02 + CO 2 (1)

C + 1/2 02 + CO (2)

Following these reactions, secondary reactions can occur, as follows

CO + 1/2 02 + CO 2 (3)

C + CO 2 + 2CO (4)

1.2 C-H 2 0 Reaction

As we will see, carbon gasification by steam is a most impor-

tant reaction. Like the C-02 reaction, there are instances where
 5

the reaction is desirable; there are other instances where it is un-

desirable. Following the gasification of carbon by steam, other re-
actions can occur which can be of great importance. Some are listed.

C + HZO 7 CO + Hz (5)

CO + HZO 7 CO Z + Hz (6)

C + CO Z 7 ZCO (4)

C + ZHZ 7 CH 4 (7)

3HZ + CO 7 CH4 + HZO (8)

1.3 Gasification of Carbon by Other Reactants

In addition to 0Z, HZO, COZ, and HZ being important molecular

reactants for the gasification of carbon (and the ones which will
be considered in detail in later chapters), other reactants should
be mentioned. Interaction of several of these reactants with car-
bon has considerable significance to the air pollution field. For
example, oxides of nitrogen formed during the combustion of coal
can subsequently react with carbon to produce NZ -- a highly desir-
able reaction (5-11).

C + NO 7 CO + l/Z NZ (9)

C + Z NO + CO Z + NZ (10)

C + NZO 7 CO + NZ (11)

Sulfur dioxide formed during the combustion of coal can , in turn,

react with carbon, giving a number of products (lZ,13)

C + SOZ + CO, CO Z ' COS (lZ)

The interaction of F Z with carbon is used to produce a product of

commercial interest (14).

(13)

Carbon reacts with NZ at a significant rate above l400°C to form

cyanogen (15)

(14)

Gasification of carbon by atomic species (0, H, N) and hydroxyl (OR)

radicals is rapid (16,17).
 6

2. RELEVANCE OF CARBON GASIFICATION FOR PARTICULAR PROCESSES AND

USES

2.1 Coal and Coal Char Combustion

Along with escalating oil prices and the advent of synfuels,

there has been accelerated interest in the combustion of coal and
coal chars. Fine pulverized fuel (pf) particles « 0.2 mm in size)
are used in many power generating plants. At the high particle
temperatures achieved during combustion in air (1500-2000°C), par-
ticle burn-out times are short; and combustion rates are determined
to a large extent (or completely) by the rate at which O2 can dif-
fuse to the exterior surface of the particles (18-20). We recently
studied the reactivity of 60-80 ~m particles of five coal chars,
obtained as residues from various synfuel processes (21), to air.
Three of the chars had particle densities of about 0.90 glcc (high
porosity chars); two of the chars had particle densities of about
1.50 glcc (low porosity chars). At low temperatures (Figure 2)
where combustion rates were determined by the intrinsic reactivity
of chars, there was a 60-fo1d difference in mass reactivities, de-
pending upon the rank of the starting coals and their thermal his-
tory in the particular synfuel process (21). At elevated tempera-
tures (Figure 3), similar to those found in pf combustion, mass re-
activities were inversely proportional to particle density, which

Tp (K)
700 650 600 550
10- 3

\
10-5

\
. \\ \ .
\0
N
~
~
!. ~

:i
~
1: 10.6
10-4 00

D~
~

. ..!
II<
~
II<

l o Cho, A
1
0\ \
o Cho, B \ w
.6. Cho,C
0 Char 0
\1 Char E

10-5 L-J'--_ _- ' -_ _--L.._ _- - - '_ _ _- ' - - - - ' 10-7

1.4 1.5 1.6 1.7 1.8
Particle Temperature (10 3 IT p' K-l)

Fig. 2 - Reactivities of residual chars from coal conversion pro-

cesses in air at low temperatures.
Reprinted from Ref. 21 by courtesy of Elsevier Science
Publishers.
 7

2100 1900 1500

-!: -4:0~
~ :~ 10.1 I- ---~~LJ _ _ _ _ _
J~ o Ch"A
LQcP-:

1
..(

uJ
0
6.
0
Ch"B
Char C
Char 0
'1 Char E

10. 2 I::::-----+----+----f-----+-----j

{"
~

.-
.~
~ 10

.
~

'"'

4.5 5.0 5.5 6.0 6.5 7.0

Particle TemperatlJre (104/T p' K91)

Fig. 3 - Reactivities of residual chars from coal conversion pro-

cesses in air at high temperatures (22).

was expected for chars of similar particle sizes (22). Further, ex-
ternal area reactivities were independent of intrinsic char reac-
tivities, again as expected.

2.2 Production of Hydrogen and/or Carbon Monoxide

The production of HZ and/or CO via coke or char gasification

is now an important commercial process and promises to become of
much greater importance in the future. As seen in Figure 4, if a
product HZ/CO mixture is desired, a blend of coal (or other carbon-
aceous solid), steam, and 0z is fed into the gasifier. The exact
ratio of these reactants charged depends on reactivity of the car-
bonaceous solid, whether the gasifier is to be operated in a slag-
ging or nons lagging mode, and the composition of product gas desired.
The higher the maximum temperature at which the gasifier is to be op-
erated the lower is the feed HZO/OZ ratio. In the limit as the ratio
approaches zero, the gasifier operates at a very high temperature
yielding primarily CO as the product gas. The product gas is fre-
quently passed through a CO-shift converter [reaction (6)] to pro-
duce a final desired HZ/CO ratio. The desired ratio depends upon
 ""
C02

j
H2/N2 =3/1
Ammonia
HZ/CO = 2/1
Methanol
HZ/CO =3/1 ~

a j Methane (SNG)

--
.~ HZ/CO =1/2 ~ 2/1
HydrocarbOns

-
$ HZ/CO =1/1
Coal ;E Oxoalcohols
.!! ~
:.: 6
Oxygen ~ u
CJ
Steam
Hz >97\
Hydrogen

,1 l
Sulphll

co + H2 (N2. CO2) Reduci nl gas

(for producinr
reduced iron)
CO + HZ (N2. CO2)
Fuelga

Fig. 4 - Production of CO and H2 by coal gasification.

the ultimate use to be made of the synthesis gas, as seen in Figure

4. Fundamentals of coal gasification and description of various
gasification processes are well covered elsewhere (23,24).

Gasifiers can be classified as operating essentially in one of

three modes: fixed bed, fluid bed or entrained flow. It appears
that, commercially, gasifiers increasingly operate in either a fixed
bed (for example, the Lurgi) or entrained flow (for example, the
Texaco or Koppers-Totzek) mode. The four main reaction zones in a
nons lagging fixed bed gasifier are shown in Figure 5. The nons1ag-
ging gasifier suffers from the basic disadvantages that (i) a large
stoichiometric excess of steam is required to moderate the combus-
tion zone temperature above the grate and thereby prevent ash clink-
ering and fusion and (ii) relatively large amounts of aqueous liquor
are produced. By decreasing the H2 0/02 ratio from about 9 to 1.3,
the gasifier can operate in a slagging mode, thereby increasing
thermal efficiency, lowering liquor production, and increasing CO
production at the expense of C02.

Entrained-flow gasifiers use pulverized fuel, which is intro-

duced into the gasifier either in a dry mode (Koppers-Totzek) using
high gas velocities or in a wet mode (Texaco)' as a solid-water
slurry. Figure 6 shows the Koppers-Totzek two headed gasifier,
 9

drying and
devolatilization zpne

reduct ion zone

Fig. 5 - Four main reaction zones in a nonslagging fixed bed gasi-

fier.
Reprinted from Ref. 24 by courtesy of Wiley Interscience.

coal dust crude gas

coal dust

t t t

=:JL...:l~.!~tt:gj-- slag quench tank

Fig. 6 - The Koppers-Totzek entrained flow gasifier.

Reprinted from Ref. 24 by courtesy of Wiley Interscience.
10

which operates in a slagging mode at essentially atmospheric pres-

sure. Two burners are positioned directly opposed to improve tur-
bulence and to ensure continuous ignition. Because of rapid heating,
devolatilization of coal takes place so rapidly that agglomeration
of bituminous coal particles during their plastic stage does not
occur; therefore, any coal can be gasified irrespective of coking
properties, ash content, or ash fusion temperature. Further, since
the gasifier can operate at very high temperatures (when the HZO/OZ
ratio becomes small), unreactive fuels such as anthracite and
petroleum coke can be readily gasified. Shell is now developing a
pressurized version of the Koppers-Totzek gasifier, which is thought
to hold great promise (Z3).

It is thought that, at present, the only commercially available

entrained-flow gasifier is the Texaco unit. It operates at elevated
pressures. Recently Eastman Kodak's Chemical Division unveiled its
chemicals from coal plant, using the Texaco gasifier to produce
synthesis gas for the production of acetic anhydride (Z5). More
recently the Cool Water coal gasification-combined-cycle power
plant was brought on-line, using the Texaco gasifier (Z6). The
gasifier is followed by a gas (combustion) turbine and then a stearn
turbine. It has been said that "this plant is the first fundament-
ally new technology for making electricity from coal since the
19Z0s." Since coal is fed to the gasifier in slurry form, it means
that a high HZO/coal imput ratio is required (particularly for low
rank coals which adsorb considerable HZO into their pore system).
The resultant additional heat requirement is reflected in a high 0z
demand.

It is most advantageous to have the reaction(s) of HZ and CO

produce the desired gaseous fuels or chemicals, to as great an ex-
tent as possible, within the gasifier. Since synthesis reactions
are highly exothermic, their occurrence in the gasifier helps sup-
ply the heat required for the endothermic C-HZO reaction and there-
by reduces or eliminates the need for OZ, Entrained flow gasifiers
produce little hydrocarbons since they operate at high temperatures
(producing equilibrium restrictions) and short residence times (pro-
ducing kinetic limitations). Fixed bed reactors, operating at lower
temperatures and longer residence times, produce some hydrocarbons.
Exxon, through its pioneering Catalytic Coal Gasification Process,
has demonstrated the possibility of producing CH4 entirely within a
fluid bed gasifier (Z7). Figure 7 presents a schematic of the pro-
cess. All product CO and HZ leaving the gasifier is recycled. The
sum of the following reactions are involved to make the process es-
sentially thermo neutral:
 11

make-up catalyst

recovered CO-H2
catalyst steam recycle

residue

Fig. 7 - Schematic flow plan for the Exxon catalytic coal gasifica-
tion process.
Reprinted from Ref. 24 by courtesy of Wiley Interscience.

2C + 2H 20 + 2H2 + 2CO (5)

CO + H2 0 + H2 + CO 2 (6)

3H2 + CO + CH 4 + H20 (8)

2C + 2H20 + CH4 + CO 2 (15)

Since reduction in temperature favors the formation of CH4 (at

equilibrium), a potassium catalyst is introduced to accelerate the
gasification rate such that reaction (5) goes at a suitable speed
at 600°C and 3.5 MPa pressure. Fortuitously, potassium is also
found to be a reasonably good catalyst for the methanation reaction
and, perhaps even more important, is not strongly poisoned by H2S
produced during gasification (28).

2.3 Metallurgical Processes

2.3.1 The blast furnace. In many ways, the blast furnace re-
sembles the air blown fixed bed gasifier. Air introduced through
the tuyeres towards the bottom of the furnace reacts rapidly with
coke producing CO which is in turn rapidly oxidized to C02 producing
a temperature of about l800°C. At this temperature, the C02 then
reacts rapidly with coke to produce CO. The CO in turn reduces
iron ore passing down the furnace, again producing CO 2 • The CO/C02
ratio leaving the top of the blast furnace depends upon the reac-
tivity of the coke to C02 and the ease with which CO reduces a par-
ticular iron ore. A coke of high reactivity to C02 can be benefi-
cial or detrimental depending upon the thermal and chemical require-
ments of the process (29). Cokes of intermediate reactivity to C02
are generally desired. Coke reactivity to C02 decreases the higher
 12

the coking temperature and rank of bituminous coal from which the
coke is made (30).

Because of a shortage of good coking coals in some parts of

the world, there recently has been an interest in form coke -- a
composite made from a filler char and binder pitch coke obtained
from low temperature coal carbonization (31). Here it is important
that the binder coke is not more reactive to C02 than is the filler
char. If it is more reactive, serious loss of strength and disin-
tegration of the form coke occurs, leading to fines formation and
restriction of gas flow up through the blast furnace. To achieve
a low reactivity in the binder carbon, it is imperative that the
metals content in the pitch precursor be low so that catalyzed car-
bon gasification be held to a minimum.

2.3.2 The electric furnace. Increasingly, the electric fur-

nace is being used for the smelting of ores and the production of
alloys. For example, in the United States, approximately 30% of the
steel is produced in electric furnaces using graphite electrodes.
Part of the electrode close to the top of the steel bath is at a
sufficiently elevated temperature to undergo slow gasification in
air. Roughly, 1 kg of electrode is consumed for each 200 kg of
steel produced, with loss primarily being due to carbon gasification
in air. The rate of electrode gasification in air decreases with
increasing crystallite alignment in the synthetic graphite (32) and
decreasing metals content, particularly Ca and V.

2.3.3 The Hall cell for aluminum production. Aluminum is pro-

duced in a cell consisting of an electrolyte of A1203 dissolved in
cryolite operating at about 960°C. Carbon anodes are used both to
carry current to the cell and to serve as a reductant of A1 2 03.
Important reactions to be considered are:

A1 2 0 3 + 1.5 C 7 2Al + 1.5 CO 2 (16)

1.5 C + 1.5 CO 2 7 3CO (4)

Or adding reactions (16) and (4)

(17)

At equilibrium for these reactions at the anode, the gaseous product

would be about 99% CO. It is generally agreed that the primary
anodic product, at the current densities used in practice, is close
to 100% C02 (33,34). For this case, essentially 0.33 kg of carbon
anode would be consumed per kg of Al produced through reaction (16).
In practice, carbon consumption is higher, or about 0.5 kg of anode
per kg of Al produced (35). This is attributed, in part, to the
secondary carbon gasification reaction (4). The fact that this
reaction is relatively slow, however, means that equilibrium is not
 13

attained and that, fortuitously, carbon consumption is less than that

which would exist at equilibrium.

2.4 Production of Activated Carbon

Activated carbon is used extensively for the cleaning up of

air and water. Thermosetting precursors like coal, wood, coconu,t
hulls, acid sludge coke, and olive pits are pyro1yzed at about 800-
900°C, liberating volatile matter, and producing a porous char con-
taining a significant open and closed (to He) micropore volume.
Either concurrent with or following the pyrolysis step, the char is
gasified (usually in steam or a steam/C02 mixture) to enlarge the
average size of open micropores and open micropores which are closed.

Figures 8-10 summarize pertinent results for the production of

activated carbon from an anthracite coal (36). The anthracite was
first devolatilized in N2 at 950°C, and then the 230 ~m particle
size was gasified in 0.1 MFa of C02 to different extents of burn-off
at both 850°C and 950°C. As seen in Figure 8, as the amount of car-
bon gasified increases, the apparent density of the particle (Pa),
as measured by Hg displacement, decreases essentially linearly; and
the He density (PHe) increases. The decrease in Pa is due to the
continuous removal of carbon atoms from the interior of the particle.
The increase in PHe is due to the opening up of closed pore volume.

..., 2.20 ..ci

..; ..;
E E
E E
u
v
--.,.
0>

'0 2.00 G
:r
':l
...>
;;;
~
~
Z ;;;
w
c
zw
... C

5'"
Z
w
1.80
'"....
~ w
....
~
:r
0.5 '--__--'_ _--'-_ _-'-_ _.....L.._ _- ' -_ _" - - _ - - - '
o 10, 20. 30, 40 50 60 70
BURN-OFF, %, m.m.tb.

Fig. 8 - Variation of densities with burn-off of anthracite char in

C02 at 850°C.
Reprinted from Ref. 36 by courtesy of Pergamon Press.
 14
1000

.lJ 850°C.·
e
e
"e o o
<
w 500
'"-<
w

...-<
U

'"
:::>
IJ)

0
0 10 20 30 40 50 60 70
BURN.OFF, ~o, m,m.l.b.

Fig. 9 - Variation of surface area with burn-off of 1 g of anthra-

cite char in CO 2 ,
Reprinted from Ref. 36 by courtesy of Pergamon Press.

0.3

.lJ

Ii
e 0

u
u
w'
~
....
:::> 0.2
0
>
w

...'"
0
....
-<
I-
0I-

30 40 50 60 70
BURN.OFF, %, m.m.f.b.

Fig. 10 - Variation of total open pore volume with burn-off of 1 g

of anthracite char in C02'
Reprinted from Ref. 36 by courtesy of Pergmaon Press.
 15

At high carbon burn-offs, the PHe approaches closely the true den-
sity of the carbon crystallite.

In both Figures 9 and 10, activation was started with 1 g of

anthracite char. Surface area was measured by N2 adsorption at
77 K, and total pore volume (accessible to He) was calculated from
values of Pa and PHe (37). It is seen that surface area and pore
volume, remaining, initially increase with extent of carbon gasified.
Both parameters go through maxima at about 40-50% burn-off a~d then
decrease. However, the extent of surface area and pore volume de-
velopment is seen to increase with decreasing temperature used for
activation. This is attributed to a balance between the rate of
gasification and the rate of diffusion of reactant into the interior
porosity of the char. The lower the gasification temperature, the
less the concentration gradient of C02 in the radial direction of
the particle and, as a consequence, the more uniform is the extent
of activation through the particle. Producers of activated carbon
optimize their process by deciding what total burn-off to give the
char and at what temperature to activate. The higher the tempera-
ture they use the greater is the rate of throughput in the reactor
but the less area and pore volume produced per unit of burn-off.

Recently, there has been interest in the production of molecu-

lar sieves from microporous carbons to be used for the separation
of molecules. In this case, the extent of carbon gasification is
kept relatively low so that the micropores remain small -- compar-
able to the sizes of molecules to be separated. Figure 11 summar-
izes results for the production of a molecular sieve carbon from an
anthracite char (38). Small differences in amounts of burn-off
drastically change the accessibility of CH4 into the char particles,
as measured by the diffusion parameter and its change with tempera-
ture.

2.5 Regeneration of Catalysts

In typical refinery operations for the upgrading of petroleum

feedstocks, catalyst deactivation, as a result of coke deposition
in the pores of the catalyst, is a continuing problem. The problem
has become of increasing seriousness as feedstocks of higher molecu-
lar weights and aromaticities and catalysts of increasing Lewis
acidities are used (39,40). Catalysts are intermittently regenerat-
ed by gasifying away the coke deposits using air, air/H20, or air/C02
as reactants (41). Intrinsic reactivities of the coke deposits are
expected to vary markedly depending upon their structure and amount
of heteroatoms present, which in turn depend upon the nature of the
feedstock, temperature at which the coke was deposited (or reacted
off the catalyst if it is higher), and nature of the catalyst sur-
face on which coke deposition occurred. If the catalyst is regen-
erated at sufficiently low temperatures in air (Zone I kinetics are
16

T/K

1.4
E:::.O

1.2

r
...

S
1.0
.~l
-I

___
- 0.8 •

O~kJmOrl
0.6 ~

0.4
~O
10 3 KIT

Fig. 11 - Effect of burn-off of an anthracite char on ease of dif-

fusion of CH4 into the micropores. Burn-offs (%): () ,
6.9; • 8.0; I::l., 9.l.
Reprinted from Ref. 38 by courtesy of the Royal Society
(London) .

observed (42», regeneration time (t) is determined by the following

parameters.
m
t - C jknC o
c 2

where Cc is coke concentration on the catalyst, k is intrinsic coke

reactivity, n is the atoms of carbon gasified per molecule of 02
consumed, CO 2 is the 02 concentration at the exterior of the gran-
ule, and m is the order of reaction. If the catalyst is regenerated
at higher temperatures, the combustion process in a porous catalyst
granule may assume a shell-progressive model (43). In this case
regeneration time is independent of intrinsic coke reactivity and
is determined by the following parameters (43):
2
t - R C jnDC o
c 2
where R is the granule radius and D is the effective diffusivity of
the porous granule for 02. Usually regeneration is conducted at as
 17

high a temperature as possible, consistent with negligible catalyst

deactivation caused by sintering.

Recently there has been interest in carbon as a catalyst sup-

port for metals (44), due in some cases to its low activity for pro-
moting coke formation, compared to alumina for example (45). Ob-
viously, if carbon is to be used as a support, its rate of gasifi-
cation either during production runs or during catalyst regeneration
must be small. The former is true in the case of most reactions in-
volving H2 , both because the C-H2 reaction inherently proceeds at a
slow rate (42) and one or more of the reactants or products inhibits
the gasification rate. Far example, CO inhibits carbon support gasi-
fication when H2/CO mixtures are used in methanation or Fischer-
Tropsch synthesis. Likewise H2 S, produced during hydrocarbon de-
sulfurization in H2 , is an inhibitor of carbon gasification. The
minimization of gasification of the carbon support in air, during
catalyst regeneration to remove deposited coke, is still a challenge
to researchers.

2.6 Graphite Nozzles for Rocket Motors

Because of its strength at high temperatures, good thermal

shock resistance, and low density, carbon is the preferred nozzle
material in rocket motors. Upon the firing of a typical propellant
consisting of 75% ammonium perchlorate (oxidizer), 10% hydroxy
terminated polybutadiene (binder), and 15% aluminum powder (fuel),
a complex mixture of propellant combustion species is produced.
This combustion gas enters the nozzle throat at a velocity of about
1000 mis, a pressure of 4.5 MFa, and temperature of 3l00°C. Of major
concern is the extent to which the combustion gases remove carbon
from the nozzle by gasification, causing throat recession. Throat
recession affects motor performance by reducing the delivered spe-
cific impulse and by altering the thermostructural behavior of the
nozzle. The ability to predict the amount of recession a nozzle
will experience as a function of graphite used and operating condi-
tions is highly desirable. For the typical propellant described
above, the major reactant which causes graphite gasification (re-
cession) is steam (mole fraction of about 0.23 in combustion prod-
ucts). The major products which are expected to inhibit carbon
gasification in steam are H2, C12, and HCl (42,46). Inhibition is
caused by the strong chemisorption of Hand Cl atoms on carbon ac-
tive sites, thereby blocking these sites from attack by steam.

Recently reactivities in H2 0 and H20/H2 mixtures of carbon com-

posites and bulk graphite used as nozzle materials have been studied
(47). It was found that composites made of PAN (polyacrylonitrile)
fibers were more reactive than those made of petroleum pitch fibers.
It was concluded that carbon consumption under nozzle conditions was
in the regime between Zones II and III (42), where carbon gasifica-
tion occurs in a thin layer close to the exterior surface of the
 18

nozzle. Consumption of carbon in this layer reduces graphite den-

sity and can lead to reduction in strength and modulus, resulting
in mechanical erosion of carbon from the nozzle under the high
velocity conditions encountered. For example, Figure 12 presents
data for the loss of modulus in a PAN fiber-based composite as
weight loss due to gasification occurs (48).

The importance of understanding gasification of graphite rocket

nozzles was brought out dramatically during a recent mission of the
space shuttle, Challenger. The launching of two communication
satellites went awry because of the failure of the small rocket that
was to loft them into proper orbit in space. They ceased firing
after only a few seconds rather than the intended 88 s burn. Ex-
perts suspect that the nozzle, consisting of a carbon fiber compo-
site, burned through prematurely.

2.7. Graphite in Nuclear Reactors

Graphite is used in gas-cooled (C02 or He is used) nuclear re-

actors both as a moderator of fast neutrons and a structural material.
It is important that the graphite does not gasify away. Unfortun-
ately, the presence of fast neutrons and gamma radiation in the re-
actor promotes carbon gasification. Collision of fast neutrons with

DENSITY (g/cm 3 1
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7

°O~------~IO--------~20~------~30~----~~40;==-====-5~O~--~
WEIGHT LOSS (%1

Fig. 12 - Effect of carbon gasification on Young's modulus of PAN

fiber-based composite (48).
 19

the graphite produces damage in the form of vacancies and intersti-

tials (49). Gamma radiation of C02 or of trace impurities in the
He coolant produces reactive free radical and ionic species (50).
Figure 13 presents results for the gasification of graphite in air
at 300°C (51). Damage to the graphite and gamma radiation produce
an additive effect on graphite gasification rates -- in the case
shown the rate is increased some 20 times over the base case.

The problem of gasification of moderator graphite by trace

amounts of H20 found in the He stream in the high temperature re-
gion of the nuclear reactor is compounded by the existence of the
Boudouard reaction over steel heat exchanger tubes in the cooler
part of the reactor (50). Enhanced rates of carbon transport from
the warmer to the cooler part of the reactor occur as the concentra-
tion of CO builds up in the reactor, as follows:

high
-+ CO + H2 (5)
T

- 1.14 =SLOPE

.75

C/)
C/)
0
...J
>l! .50
0
l-
I
(!)
W
::=
...J .25
~
~
.10

160

Fig. 13 - Gasification of nuclear graphite in air at 300°C under

different conditions. Slopes give % loss in 100 days •
.. , Unirradiated sample, oxidized in absence of radia-
tion; () , irradiated sample, oxidized in absence of
radiation; ~ , unirradiated sample, oxidized in 2xl0 5
r/h gamma flux; 0, irradiated sample, oxidized in
2xl0 5 r/h gamma flux; ~ , unirradiated sample, oxidized
in 6.lxl0 5 r/h gamma flux.
20

low
2CO + C + CO 2 (18)
T

high
C + CO 2 + 2CO (4)
T

3. CONCLUSIONS

The gasification of carbon is a most important process in our

society. In some cases, it is a highly desirable reaction which we
want to proceed at a maximum rate. In other cases, it is a highly
undesirable reaction, which we want to proceed at a negligible rate.
In both cases, understanding the fundamentals of carbon gasification
is vital to our making the desired technological advances.

In this chapter, we have considered some of the main cases

where carbon gasification is of prime importance. However, in fact,
the number of cases which we would have considered is almost endless.
The following is a partial list of additional areas of importance:

(i) Carbon in batteries, fuel cells, and industrial electroly-

sis (in addition to its use in aluminum production) where carbon
corrosion in acidic and basic electrolytes is of concern (52).

(ii) Carbon oxidation when used as a refractory (53).

(iii) Carbon oxidation when used as a thermal insulator.

(iv) Carbon oxidation when used in aerospace applications (in

addition to its use as a nozzle).

(v) Carbon fiber etching to roughen its surface (54).

(vi) Carbon oxidation when used as bearings and seals.

(vii) Gasification of carbons in various oxidizing media to

analyze for amounts of various structural components (55).
 21

REFERENCES

1. Brooks, J.D. and G.H. Taylor, in P.L. Walker, Jr., ed., Chem-
istry and Physics of Carbon, Vol. 4 (New York, Marcel Dekker
Publishers, 1968) pp. 243-286.
2. Marsh, H. and P.L. Walker, Jr., in P.L. Walker, Jr. and P.A.
Thrower, ed., Chemistry and Physics of Carbon, Vol. 15 (New
York, Marcel Dekker Publishers, 1979) pp. 229-286.
3. Bokros, J.C., in P.L. Walker, Jr., ed., Chemistry and Physics
of Carbon, Vol. 5 (New York, Marcel Dekker Publishers, 1969)
pp. 1-118.
4. Walker, P.L., Jr., F.J. Vastola, and P.J. Hart, in Fundament-
als of Gas-Surface Interactions (New York, Academic Press
Publishers, 1967) pp. 307-317.
5. Beer, J.M., A.F. Sarofim, L.K. Chan, and A.M. Sprouse, in
Proc. 5th International Conference on Fluidized Bed Combustion,
Vol. 2,
6. West, L.A. and G.A. Somorjai, J. Vacuum Sci. Tech., 2, 668
(1972).
7. Smith, R.N., D. Lesnini, and J. Mooi, J. Phys. Chern., 61, 81
(1957).
8. Bedjai, G., N.K. Orbach, and F.C. Riesenfeld, Ind. Eng. Chern.,
50, 1165 (1958).
9. Arthur, J.R., H.F. Ferguson, and K. Lauber, Nature, 178, 206
(1956).
10. Watts, H., Trans. Faraday Soc., 54, 93 (1958).
11. Edwards, H.W., A.I.Ch.E. Symposium Series 68 (1971) pp. 91-
101.
12. Sappok, R.J. and P.L. Walker, Jr., J. Air Poll. Control Assoc.,
19, 856 (1969).
13. Sinha, R.K. and P.L. Walker, Jr., A.I.Ch.E. Symposium Series
68 (1971) pp. 160-167.
14. Rosner, D.E. and J.P. Strakey, J. Phys. Chern., 77, 690 (1973).
15. Cullis, C.F. and J.G. Yates, Trans. Faraday Soc~ 60, 141
(1964). --
16. McCarroll, B. and D.W. McKee, Carbon, 2, 301 (1971).
17. Fenimore, C.P. and G.W. Jones, J. Phys. Chern., 71, 593 (1967).
18. Essenhigh, R.H., in M.A. Elliott, Ed., ChemistrY-of Coal Uti-
lization (New York, Wiley Interscience Publishers, 1981)
pp. 1153-1312.
19. Smith, I.W., in 19th Symposium (International) on Combustion
(Pittsburgh, The Combustion Institute Publishers, 1982)
pp. 1045-1065.
20. Timothy, L.D., A.F. Sarofim, and J.M. Beer, in 19th Symposium
(International) on Combustion (Pittsburgh, The Combustion
Institute Publishers, 1982) pp. 1123-1130.
21. Radovic, L.R. and P.L. Walker, Jr., Fuel Proc. Tech., ~, 149
(1984).
22

22. Wells, W.F., S.K. Kramer, L.D. Smoot, and A.U. Blackham, "Re-
activity and Combustion of Coal Chars," presented at 20th
Symposium (International) on Combustion, August 1984.
23. Vogt, E.V. and M.J. van der Burgt, Phil. Trans. R. Soc. Lond.,
A300, III (1981).
24. Hebden, D. and H.J.F. Stroud, in M.A. Elliott, ed., Chemistry
of Coal Utilization (New York, Wiley Interscience Publishers,
1981) pp. 1599-1752.
25. Chem. Eng. Prog., 80, No.5, 24 (1984).
26. EPRI Journal, 9, N~ 10, 16 (1984).
27. Nahas, N.C., F~el, ~, 241 (1983).
28. Kapteijn, F. and J.A. Moulijn, J. Chem. Soc., Chem. Commun.,
No.5, 278 (1984).
29. Kirkaldy, J.S. and R.G. Ward, Aspects of Modern Ferrous Metal-
lurgy (Toronto, University of Toronto Press Publisher, 1964).
30. Hippo, E. and P.L. Walker, Jr., Fuel, 54, 245 (1975).
31. Moran, R.F. and R.T. Joseph, Trans. AIME, 260, 29 (1976).
32. Walker, P.L., Jr., F. Rusinko, Jr., J.F. Rakszawski, and L.M.
Liggett, Proc. 3th Carbon Conference (New York, Pergamon Press
Publishers, 1959) pp. 643-658.
33. Thonstad, J., J. Electrochem. Soc., Ill, 959 (1964).
34. Cammarota, V.A., Jr. and D. Schlain,~ Electrochem. Soc.,
117, 282 (1970).
35. Reese, K.M., A.F. Garcia, and R.A. Lewis, Ind. Eng. Chem.,
47, 2066 (1955).
36. Kawahata, M. and P.L. Walker, Jr., Proc. 5th Carbon Confer-
ence (New York, Pergamon Press Publisher, 1963) Vol. 2,
pp. 251-263.
37. Mahajan, a.p. and P.L. Walker, Jr., in C. Karr, Jr., ed.,
Analytical Methods for Coal and Coal Products, Vol., 1 (New
York, Academic Press Publisher, 1978) pp. 125-162.
38. Walker, P.L., Jr., Phil. Trans. R. Soc. Lond. A300, 65 (1981).
39. Franklin, J.L. and D.E. Nicholson, J. Phys. Chem. , 60, 59
(1956). -
40. Scaroni, A.W., R.G. Jenkins, J.R. Utrilla, and P.L. Walker,
Jr., Fuel Proc. Tech., 9, 103 (1984).
41. Pansing, W.F., A.I.Ch.E~ Journal, 1, 71 (1956).
42. Walker, P.L., Jr., F. Rusinko, Jr., and L.G. Austin, in D.D.
Eley, P.W. Selwood, and P.B. Weisz, eds., Advances in Cataly-
sis, Vol. 11 (New York, Academic Press Publisher, 1959)
pp. l33-221.
43. Weisz, P.B. and R.D. Goodwin, J. Catalysis, 2, 397 (1963).
44. Walker, P.L., Jr., Proc. 5th International Carbon Graphite
Conference (London, Soc. Chem. Industry Publisher, 1979)
pp. 427-436.
45. Scaroni, A. W., Ph.D. Thesis, The Pennsylvania State Univer-
sity, 1981.
46. Day, R.J., P.L. Walker, Jr., and C.C. Wright, First Industrial
Carbon and Graphite Conference (London, Society Chemical In-
dustry Publisher, 1958) pp. 348-370.
 23

47. Jones, L.E., M.S. Thesis, The Pennsylvania State University,

1984.
48. Zhao, J.X., R.C. Bradt, and P.L. Walker, Jr., "Effect of Air
Oxidation at 873 K on the Mechanical Properties of a Carbon-
Carbon Composite," Carbon, in press.
49. Reynolds, W.N., in P.L. Walker, Jr., ed., Chemistry and Phys-
ics of Carbon, Vol. 2 (New York, Marcel Dekker Publisher,
1966) pp. 121-196.
50. Everett, M.R., D.V. Kinsey, and E. Romberg, in P.L. Walker,
Jr., ed., Chemistry and Physics of Carbon, Vol. 3 (Ne~ York,
Marcel Dekker Publisher, 1968) pp. 289-436.
51. Hennig, G.R., G.J. Dienes, and W. Kosiba, Second United Nations
International Conference on the Peaceful Uses of Atomic Energy,
Paper A/Conf. 15/P/1778 (1958).
52. Proceedings of the Workshop on "The Electrochemistry of Car-
bon," S. Sarangapani, J.R. Akridge, and B. Schumm, eds. (Pen-
nington, NJ, The E1ectrocqemica1 Society, Inc. Publisher,
1984).
53. Park, H.K. and L.R. Barrett, Trans. Brit. Ceram. Soc., ~,
No.2, 26 (1979).
54. McKee, D.W. and V.J. Mimeau1t, in P.L. Walker, Jr. and P.A.
Thrower, eds., Chemistry and Physics of Carbon, Vol. 8 (New
York, Marcel Dekker Publisher, 1973) pp. 151-245.
55. Donnet, J.B., P. Ehrburger, and A. Voet, Carbon, 10, 737
(1972) •
PART II. STRUCTURE AND CHARACTERIZATION
 27

STRUCTURE IN CARBONS

Harry Marsh

Northern Carbon Research Laboratories, School of Chemistr~

Bedson Building, University of Newcastle upon Tyne,
Newcastle upon Tyne NE1 7RU, U.K.

ABSTRACT

Although diamond has a greater emotional appeal than graphite

as an allotrope of carbon structure, it is the graphite lattice
which dominates discussions of structure within carbons.

Structure within all carbons uses descriptions of the stacking

perfection of lamellae of carbon atoms in hexagonal arrangement,
structural perfection within such lamella and the size, shape and
extent of stacking of these lamellae.

In broad terms, carbons are separated into being either non-

graphitizab1e or graphitizab1e, i.e. a measure of their converti-
bility to 3-dimensiona1 graphitiC-Carbon on heating to -3000 oC.

The non-graphitizab1e carbons, on the one hand. are

essentially chars having been made from organic materials e.g. wood
which do not become fluid during the carbonization process-.--Such
carbons or chars tend to be porous with pore volumes of -1 cm3g- 1
contained dominantly within microporosity. The sizes of stacked
units of lamellae are in dimensions of a few nanometers (10- 9 m)
these "crystallites" being randomly arranged. This makes such
carbons isotropic when polished surfaces are viewed by polarized
light optical microscopy.

On the other hand, the graphitizab1e carbons are cokes and are
made essentially by the carbonization of fusible parent materials
such as petroleum pitches and residues, coal-tar pitch and some
model organic compounds. During the fluid phase of carbonization,
discotic nematic liquid crystals or mesophase are homogeneously
 28

precipitated from the fluid phase ultimately replacing the parent

fluid pitch. These liquid crystals have 2-dimensional order and
are capable of being graphitized to graphitic carbon. The stacking
of lamellae is in dimensions of micrometers (10- 6 m) and exhibits
parallelism so making these carbons anisotropic in the optical
microscope. Such anisotropic carbons/cokes have low internal pore
volumes.

The range of carbonaceous parent materials available for

carbonization is such that there exists a continuum of structure
within carbons, from being non-graphitizable to graphitizable.

Structure in carbons is usually assessed by X-ray and electron

diffraction, by, high resolution transmission electron microscopy
and by polarized light optical microscopy.

1. INTRODUCTION

Carbon is the naturally occurring element of atomic number 6

and atomic weight 12 a.m.u. The element exists in three structured
allotropes, Diamond, Graphite and possibly Carbyne (1, 2). However
it is as the chemically heterogeneous, structurally less-ordered
non-graphitic state as well as graphitic states (rather than
diamond) that the element carbon is most commonly found, e.g. in
coals, cokes and chars.

The carbon atom possesses six electrons which in the ground

state have the configuration of ls2 2s2 2p2.

The Trigonal or sp2 hybrid has three of the outer electrons

forming equally spaced, two-dimensionally distributed, a-bonds
at 120 0 bond angles with the remaining p-electron in an-bond,
e.g. ethene -CH=CH-. It is this form of hybridisation that is
present in graphite and in cokes and chars which have their structures
based on the graphite lattice.

The principal group of organic chemicals considered a~ graphite

precursors are those whose structure is formed from the sp hybri-
dised carbon atom, i.e. aromatic or benzenoid structures. The
chars, cokes and synthetic graphites which constitute the non-
graphitic and graphitic carbons can be considered as transitional
forms, intermediate between the organic chemical precursor and
single crystal graphite. In diamond, each carbon atom is linked
by sp3 hybridised bonding to its four neighbours resulting in an
extensive network of atoms possessing the characteristic strength
and hardness (see Figure 1). The structure of diamond was first
described by Bragg and Bragg (3). It has the face-centred cubic
lattice, the unit cell containing eight atoms on each side of the
cube which is 0.3567 nm in size.
 29

Figure 1. The tetrahedral arrangement of carbon atoms in

diamond.

2. STRUCTURE IN GRAPHITE

Ergun (4) provides a review of X-ray studies of carbon dated to

about 1967. Ergun (4) points out that the valence electrons of
carbon permit trigonal- and TI-bonding between each atom and its
three neighbours. The first structural analysis of graphite was
proposed by Debye and Scherrer in 1917 (5) but was incorrect as
the stacking was descr~bed in A-B-C-A-B-C sequences. Five years
later Hull (6) re-examined the graphite lattice and stated that
adjacent layers are translated by a bond length, but alternative
layers have identical projections, these being A-B-A-B sequences
with the unit cell of a hexagonal lattice. In 1924 Bernal (7)
published his careful analyses of single crystal graphite and
confirmed the postulates of Hull (6). The unit cell was 0.682 nm
1
high and 0.245 nm on each side. The interla er spacing is assigned
as 0.3354 nm, the density being 2.267 g cm- ,see Figure 2.
 30

~ I
I
I
I
I
I "
'I
J
1/ I
I /
, I,' I
, I
I
I
d0 02 3.35 ~
:
I
I
I
I
I
I
I
:
" I
, I
,',,' I
n I'~'
I
II'
" A I
'I
I

I 1. 42
Q:i\j
I

-~~
~i I I I,
II~l"'IIII
)2
d1 I I I, I : I,
II
I
I

g.
, I III,' I I ,
I
I
I ':
I I
·,03 I,
"11
I
I
I I
I

~
~
"~ ',a.,
c-axis
a2 dloo =
2.13 ~
al a3

Figure 2. The hexagonal structure of graphite.

Later, in 1942 Lipson and Stokes (8) observed lines in X-ray

diffractograms which could not be assigned to the hexagonal lattice.
These lines Here attributed to the presence of about 10% of A-B-C-
A-B-C sequences, i.e. to rhombohedral graphite. This latter form
of graphite is thermodynamically less stable than hexagonal
graphite (9) and reverts to hexagonal graphite on heat treatment
to between 1500 0 and 3000 0 c (10).

2.1 Defects in Hexagonal Graphite

Single crystal graphite, of high quality, is very rare. Some

of the natural flake graphites, e.g. Ticonderoga, are of high orders
of crystallinity. Probably the most perfect of graphites (and the
most expensive) are the highly orientated synthetic pyrolytic
graphites as described by Moore (11).

The so-called single-crystal graphite invariably contains

defects within its structure, i.e. dislocations, stacking faults,
vacancies and interstitial defects (12).

Dislocations in graphite are defined as line imperfections as

distinct from point defects. Imperfections occur in graphite
either in the stacking order between the layer planes (non-basal)
or as defects in the layer planes (basal). Knowledge of dislocations
 31

in graphites is important, not only because of structural conside-

rations but also because of its relevance to mechanisms of
gasification (13-16).

Stacking faults involve overcoming the weak forces of the van

der Waals type which bind together successive layers.

Basal plane dislocations are essentially stacking faults

occurring between the boundaries of the layer such that at the
edges the stacking sequence is unaltered: A-B-A-B. The fault is
apparent as a narrow ribbon of rhombohedral graphite, with the bound-
aries between it and the hexagonal graphite being the actual
dislocation lines, known as a partial dislocation. The carbon
hexagons in the region of the dislocation must be under some strain
although no carbon bonds are actually broken. Basal plane disloca-
tions are easily introduced into graphite by slip motion brought
about by mild mechanical action.

Interstitial and vacancy loops occur where a small section of

a layer is either inserted between the layers (interstitial) or is
removed from a layer (vacancy), see Figure 3. These dislocations
are introduced to the lattice on quenching rapidly from heat treat-
ment at 3000 0 e. Upon annealing, after the above treatment, at
1000 0 e, point defects (see below) migrate to form the loops.

:::
-b c---t----b-
d
=~----L b
=~
-a
, ~E
a----
~::
a--
-b b b--

-b b b--
-a a a-
-b b-
-a
-b
-a
-b
3~Rb b
a--
b--
a--
b--
-a a a-

Figure 3. Vacancy (above) and interstitial (below) loops

in single crystal graphite (12).
 32

Point defects occur as vacancies either in the crystal lattice

or interstitial loops. Should 'impurities' become bonded at
vacancy sites, a distortion of the adjacent hexagon network may
accommodate these impurities. Vacancy defects become mobile at
temperatures >1000 oC. This facilitates the re-integration of
damaged sites by annealing. Again, this is a property of consider-
able importance in discussions of mechanisms of gasification of
carbon at temperatures above -1400 oC (17). At these high tempera-
tures, vacancies caused by gasification of carbon atoms can be
annealed out by surface mobility of carbon atoms.

Non-basal edge dislocations result from the insertion of an

additional row of carbon atoms into a layer (18). Bond breakage
occurs with extreme distortion of hexagons in the surrounding
lattice.

Screw dislocations in graphite occur by a displacement of a

region of a layer in a direction perpendicular to the basal plane.
The outcome is a spiral ramp of layer planes with the actual
dislocation line being a pole about which the spiral ramp revolves.
Figure 4 shows a typical screw dislocation.

Figure 4. A screw dislocation within single crystal

graphite (12).
 33

Non-basal edge and screw dislocations can be distinguished by

mild oxidation followed by surface examination using optical
microscopy (18). Non-basal dislocations give the same pattern of
etch pits on both sides of cleaved surfaces of the graphite. Screw
dislocations give rise to a cleavage step.

3. NON-GRAPHITIZABLE AND GRAPHITIZABLE CARBONS

The structures of carbons (as distinct from single crystal

graphite) are all based on the graphite lattice to various extents.
As a consequence of this, these carbons have been classified and
provided with a nomenclature (See Carbon, 20, 445-449 (1982) and
1l, 517-519 (1983)). --

3.1 Nomenclature of Solid Carbons

The term 'solid carbon' (usually referred to in texts as

'carbon') covers all natural and synthetic substances consisting
mainly of atoms of the element Carbon and with the structure of
graphite or at least with two-dimensionally ordered layers of
carbon.

Carbonization is a process of formation of material with

increasing carbon content from organic material, usually by
pyrolysis ending with an almost pure carbon residue at temperatures
of up to about 1600 K.

A coke is a highly carbonaceous product of pyrolysis of

organic material at least parts of which have passed through a
liquid or liquid-crystalline state during the carbonization process
and which consists of non-graphitic carbon.

A char is the carbonization product of a natural or synthetic

organic material which has not passed through a fluid stage during
carbonization.

'Non-graphitic carbons' are all varieties of substances

consisting mainly of the element Carbon with two-dimensional long-
range order of the carbon atoms in planar hexagonal networks, but
without any measureable crystallographic order in the third direc-
tion (c-direction) apart from more or less parallel stacking.

'Graphitic-carbons' are all varieties of substances consisting

of the element Carbon in the allotropic form of graphite irrespec-
tive of the presence of structural defects.

'Single-crystal graphite' is a form of natural graphite which

occurs as large flakes of single crystals of high crystallographic
perfection.
34

'Synthetic graphite' is a material consisting of graphitic

carbon which has been obtained by means of graphitization heat
treatment of non-graphitic carbon or of chemical vapour deposition
(CVO) from hydrocarbons at temperatures above 2100 K or of
decomposition of thermally unstable carbides.

'Polygranular graphite' is a graphite material composed of

grains which can be clearly distinguished by optical microscopy.

'Graphitic material' is a material consisting essentially of

graphitic carbon.

'Nuclear graphite' is a polynuclear graphitic material for use

in nuc~ear reactor cores consisting of graphitic carbon of very
high chemical purity. High purity is needed to avoid absorption
of low-energy neutrons and activation of the impurities.

'Anisotropic' carbon is a carbon whose properties are a

function of its crystallographic direction (graphitizable carbon).

'Isotropic' carbon is a carbon whose properties are not a

function of its crystallographic direction (non-graphitizable
carbon).

'Non-graphitizable carbon' is a non-graphitic carbon which

cannot be transformed into graphitic carbon solely by hi~h
temperature treatment up to 3000 K at atmospheric or low pressure.

'Graphitizable carbon' is a non-graphitic carbon which, upon

graphitization heat treatment, converts into graphitic carbon.

'Graphitization' is the solid state transformation of thermo-

dynamically unstable non-graphitic carbon into graphite by thermal
activation.

'Natural graphite' is a mineral consisting of graphitic carbon

regardless of its crystalline perfection.

'Soft' carbon is synonymous with graphitizable/anisotropic

carbonS:--

'Hard' carbon is synonymous with non-graphitizable/isotropic

carbonS:--

The term 'turbostratic' is not recommended as it does not

clearly distinguish non-graphitizable from graphitizable carbons.
 35

4. ORIGIN OF CARBONS

4.1 Non-Graphitizable Carbons

The non-graphitizable carbons are essentially chars and

include charcoals which is a traditional term for a Char obtained
from wood and some related natural organic materials. Because the
precursor has not passed through a fluid stage, the Char usually
retains the shape of the precursor (although becoming of smaller
size). For such material the term 'pseudomorphous' has been used.
Some simple organic compounds, e.g. sugar, melt at an early stage
of decomposition and then polymerise during Carbonization to produce
Chars. The behaviour of sugar is therefore exceptional.

Activated carbon (charcoal) is a porous carbon material, a

Char, which has been subjected to reaction with gases, sometimes
adding chemicals, e.g. ZnC12' during or after Carbonization to
increase its porosity. Activated Carbons have a large adsorption
capacity and are used for purification of liquids and gases. By
controlling the process of Carbonization and activation a variety
of active carbons having different pore sizes can be obtained.

The majority of non-graphitizable carbons originate from wood,

nut-shells and non-fusing coals, e.g. anthracites or sub-bituminous
coals. Almost all vegetable matter will produce some form of non-
graphitizable carbon, it being woods, nut-shells etc. which
produce industrially acceptable products. It is the polymeric
cellulose or lignin within these natural-occurring materials which
are pyrolysed. These polymers lose molecules of water, carbon
dioxide and small organic degradation molecules with the basic
back-bone of the polymer remaining and eventually cross-linking, so
preventing fusing. The void age created by loss of small molecules,
including hydrogen and carbon monoxide at higher temperatures (500-
10000C) constitutes the porosity (often microporosity 0.5 to
2.0 nm diameter) and high surface areas (-1000 m2g- l ) within the
carbons (19, 20).

The carbonization behaviour of the sub-bituminous coals to

produce active carbons is comparable to that of woods, even
though the detail of the pyrolysis chemistry differs. With
the anthracites, the coalification process has produced lamellar-
type aromatic molecules with extensive 'cross-linking' to produce
a rigid non-fusing structure. Again, loss of small molecules,
e.g. CH4' CO, CO 2 , H20 and H2 during carbonization does not disrupt
the basic structure of the anthracite and creates the microporosity
within these materials.

4.2 Graphitizable Carbons

The graphitizable carbons are essentially cokes and include
 36

materials resulting from the Carbonization, in an inert atmosphere,

of pitch materials derived from coal and petroleum as well as the
fusing bituminous coals (caking and coking). Model organic
compounds, e.g. acenaphthylene, decacyclene, anthracene will form
graphitizabre-carbons if suitably carbonized. Certain naturally
occurring resins (e.g. Gilsonite) also form graphitizable carbons.

The mechanism of formation of graphitizable carbons is very

distinct from that of the non-graphitizable carbons. The
graphitization process (21, 22) whereby graphitic carbon
(exhibiting 3-dimensional X-ray diffractograms) is produced from
the non-graphitic but graphitizable carbon does not involve any
major re-crystallization process. Rather graphitization is the
perfection of an already existing 'embryonic' graphite lattice
within the graphitizable carbon, almost totally by single atom
movement.

The graphitizable carbons owe their inherent embryonic

graphitic structure to the formation of discotic, nematic, aromatic,
liquid crystals, otherwise known as mesophase from within the
isotropic fluid phase of pyrolysing pitch or coal.

5. MESOPHASE

Mesophase 1S the intermediate material formed during carboni-

zation of parent pitch and leading to the resultant coke. The
discovery and development of the concept of mesophase over the
last twenty years must represent one of the most significant
advances in carbon science over this period. Mesophase is a term
borrowed from the area of science of conventional liquid crystals
and means "intermediate phase". A major difference between the
conventional liquid crystals and those found in pyrolysing pitch
systems is that the former are chemically stable whilst the latter
still undergo further condensation reactions which eventually
produce the solid semi-coke.

5.1 Discotic Aromatic Nematic Liquid Crystals

There already exist several review articles describing in

some detail how the graphitizable (anisotropic) carbon originates
from the isotropic melt of the pitch (23-29). Brooks and Taylor
(30) initially advanced the proposal that from within the isotropic
melt of pitch there grows a new phase, more viscous than the parent
pitch, but of anisotropic texture, and yet able to flow. Nematic
liquid crystals are distinguished from other forms of liquid crystal
systems (31) in that the discotic (lamellar) constituent molecules
of pitch precursors or the rodlike molecules of conventional liquid
crystals are stacked approximately parallel to each other, other
types of order being absent (see Figure 5).
 37

Pole

Trace of
¥' lamellae
direction

¥' Edge of
disk of
sphere

1
c - axis
Pole

Mesophase sphere. Section through a mesophase sphere.

Figure 5. Structure of anisotropic nematic liquid crystals,

as spheres of mesophase within pyrolyzing isotropic
pitch (30).

Mesophase is observed by optical microscopy using polarized

light reflected from polished surfaces of the material. Use is
made of interference colours produced within the microscope to
describe the shape, size and perfection of growth units of meso-
phase within the isotropic pitch (32, 33). The growth units in
pitch are often spherical (30), the use of optical microscopy and
electron diffraction indicating that the discotic nematic arrange-
ment is of molecules lying parallel to each other and parallel to
an equatorial plane rather than being parallel to the external
surface of the sphere, as in an onion, see Figure 6.

5.2 Growth and Coalescence of Mesophase

Structure within the wide variety of cokes which are available

today, e.g. shot-, needle- and metallurgical-cokes, is dominantly
established from the mesophase. However, it has to be recognised
that mesophase itself exhibits a wide range of physical properties
of which viscosity is a major controlling influence.

It is established now without doubt that mesophase is a poly-

merized form of discotic aromatic, nematic liquid crystal. The
evidence for this, based on optical microscopy and electron diffraction,
is overwhelming (30, 32, 33). Currently, using high resolution,
fringe-imaging transmission electron microscopy, it is possible to
 38

(a) (b)

Figure 6. Structure in mesophase (a) the most common Brooks and

Taylor structure of equatorial stacking of lamellae,
(b) the less common onion-structure of lamellae
parallel to the external spherical surface.

reproduce the lattice of most forms of carbons as fringe images.

Thus structure can be seen 'directly' as it were, Figure 7. Of
the many studies now available in this area of structural analysis
the earlier studies of Ban et a1. (34) and studies of Crawford
et a1. (35, 36) and Marsh e~ (37) were specifically devoted to
structural analyses of the non-graphitic, non-graphitizab1e and
graphitizab1e carbons. Auguie et a1. (38) examined the lattice
structures specifically found in spheres of mesophase using the
high resolution fringe-imaging technique (HREM). Millward et al.
(39) have reviewed the use of HREM in several carbon forms.

Pyrolysis chemistry, in general terms is of dehydrogenative

po1ymerisation processes whereby there is a progressive increase
in the average molecular weight of what essentially are po1y-
aromatic lamellar molecules (26). This growth process occurs
within the fluid isotropic medium of the pyrolysis pitch.

However, with continuous growth in average molecular weight

of such molecules a new phenomenon becomes manifest. As a result
of molecular motion there is a finite probability of two lamellar
molecules approaching each other face to face. As a result of the
now adequate van der Waals forces between these molecules, the
residence time of these molecules together far exceeds that
associated with a molecular collision. Because these lamellar
'molecu1es'may in fact be reactive radicals, there also exists a
 39

'Figure 7. High resolution electron micrograph of structure in a

carbon black to resolve individual constituent
molecules as fringes.

finite probability that some form of bond is established across

the lamellar molecules and this prevents the separation of the
pair of molecules.

The consequence of this continuous growth of molecular weight

etc. is the growth of these associations of lamellar molecules in
a stacking sequence. This growth is the initiation of the formation
of nematic liquid crystals, the extreme stability of which is due
to the large cumulative van der Waals forces between the layers.
This is the process of homogeneous nucleation of the nematic
liquid crystals or mesophase from the pyro1ysing melt.

When studying the growth of mesophase by hot-stage optical

microscopy, the first appearance is usually of spheres of highly
anisotropic material (40). These spheres then grow in size at the
expense of the surrounding isotropic fluid. Should two spheres
touch during this stage of pyrolysis then there occurs immediate
coalescence (.<1 second) with a resultant re-orientation of the
material to fo~ a new larger sphere. This process continues until
the entire pitch matrix has been converted to mesophase. The extent
to which coalescence and re-orientation takes place depends on the
viscosity of the mesophase (Section 5.4).

Figure 6 illustrates the arrangement or structures of the

40

lamellar molecules within a sphere. With coalescence of spheres

these rather simple structures are lost and the bulk mesophase
(i.e. coalesced but still fluid mesophase) and eventually the
semi-coke (hardened mesophase) adopts a quite complex arrangement
of lamellae.

The structure of coalesced mesophase possesses defects which

are known as "disclinations" (24, 29). These are low energy defects
some of which anneal out during the transformation of coalesced
bulk but fluid mesophase to the stage of semi-coke (hardened
mesophase).

However, not all pyrolysing systems of pitch, coal or model

compounds provide mesophase which grows and coalesces during
pyrolysis (25). With e.g. coals which undergo caking during carbon-
ization, the initial growth units of mesophase are not necessarily
spherical but are of a more irregular shape. Qian et al. (41) and
Hays et al. (42) studied structure in cokes from coals of different
rank. With these materials, the growth units of mesophase
apparently grow independently to diameters of -<5 ~m. No further
growth occurs and the entire pyrolysing coal system is converted
into growth units of mesophase of size -1-5 ~m dia. On touching,
such growth units of mesophase adhere to each other, in fact may
'interlock' their surfaces, but no coalescence occurs with
annihilation of the initial pair of mesophase spheres.

5.3 Structure in Mesophase

Structures within coalesced mesophase can thus vary from

that of what is essentially a deformed "single-crystal" (domains)
to a composite of smaller individual "crystals", -1 ~m diameter,
called fine-grained mosaics. Table 1 lists the nomenclature used
to describe these structures of what overall is termed THE OPTICAL
TEXTURE of anisotropic graphitizable carbons (24, 25, 28, 30, 33).

It is this wide variation in size and shape of the optical

texture, or crystal arrangement, which accounts for the observed
variations in structure and properties found in cokes, e.g. small
size, <2 ~m dia., found in shot cokes to the elongated crystals,
e.g. 100 x 10 ~m, found in needle-cokes.

Additionally there exists a relationship between the size of the

optical texture and the shape, size and perfection of stacking
arrangements within the lamellar molecules which constitute
these materials, as studied using HREM (43), see Figure 8(1-111)
Figure 8-1 from acridine (Figure 9) which gives a large size of
optical texture (domains) suggests that the parallel stacking extends
at least 30 nm in the c-direction. The degree of alignment is
relatively good and it is possible to follow single fringes over a
distance of about 8 nm. Assuming that the a-b dimension of the unit
 41

cell of graphite is 0.245 nm, these fringes are produced from a

'sheet' of condensed acridine molecules having a diameter of about
30 ;ings. The shortest of the fringes has a length equivalent to
a diameter of 4 to 5 rings. Figure 8-11, from 1,10-phenanthroline
which forms a coke with an optical texture of small domains
(Table 1) shows a strip of material,S nm width,
orientated w~tn approximately parallel c-axes as examined by HREM.
In Figure 8-11 the degree of alignment is lower than in Figure 8-1.
An average fringe length of about 2.5 nm is equivalent to about 10
hexagonal rings in the polymer molecule. Figure 8-111, of compound
III which forms a carbon of optical texture of fine-grained mosaics
(Table 1) indicates that the size of the fringes and the
degree of alignment are much lower than observed for the other two
carbons. There is much less linearity within the fringe. An
average length appears to be about 1.5 nm. The degree of ordering
within a carbon and hence its graphitizability are thus related to
size of optical texture.

I II III

f'i?ure 8
2.5 )lm

I II III

Figures 8,9. Drawings from high resolution electron micrographs

which image fringes of constituent molecules in
carbons from acridine (I), 1,lO-phenanthroline (II)
and compound III containing N,S and 0 (43).
42

Table 1

Nomenclature to Describe Optical Texture

in Polished Surfaces of Cokes

Isotropic No optical activity

Very fine-grained mosaics (VMf) <0.5 ~m in diameter

Fine-grained mosaics (Mf) <1.5 > 0.5 ~m ~n diameter

Medium-grained mosaics (Mm) <5.0 > 1.5 ~m in diameter

Coarse-grained mosaics (Mc) <10.0 > 5.0 ~m in diameter

Supra mosaics (SM) Mosaics of anisotropic carbon

orientated in the same
direction to give a mosaic
area of isochromatic colour.

Medium-flow anisotropy <30 ~m in length,

elongated (MFA) <5 ~m in width

Coarse-flow anisotropy <60 > 30 ~m in length,

elongated (CF) <10 > 5 ~m in width

Acicular flow domain >60 ~m in length,

anisotropy (AFD) <5 ~m in width

Flow domain anisotropy >60 ~m ~n length,

elongated (FD) >10 ~m in width

Small domains,
isochromatic (SD) >60 > 10 ~m in diameter

Domains, -isometric (D) >60 ~m in diameter

Db is from basic anisotropy of low-volatile

coking vitrains and anthracite.
D is by growth of mesophase from fluid phase
m

~ibbons (R) Strands of mosa~cs inserted into an isotropic texture

 43

5.4 Viscosity of Mesophase and Optical Texture

The degree of coalescence of growth units of mesophase

(Section 5.2) is governed by the viscosity of the many different
systems of mesophase. Cokes with
a large size of optical texture result from mesophase of relatively
low viscosity while cokes with small sizes of optical textures,
e.g. fine-grained mosaics, result from mesophase of relatively high
viscosity (25), see Figure 10. Curve A shows a mesophase system
with a relatively high viscosity and a short range of temperature
of minimum viscosity.
Here, growth units of mesophase lack the ability to
flow and to coalesce on contact, and fuse on contact. The high
viscosity precludes extensive flow of the bulk mesophase. Solid-
ification occurs relatively rapidly and creates cokes of optical
texture of mosaics (Figure 10). Curve C shows a mesophase system
with relatively low viscosity and an extended range of temperature
at mlnlmum viscosity. Mesophase growth units coalesce on contact
and large sizes of liquid crystal result, seen in optical microscopy
as isochromatic areas. Flow in the bulk mesophase of low viscosity
is able to remove defects and disclinations introduced during the
coalescence process.

A C

Viscosity

Pyrolysis temperature

Figure 10. Diagram of variation between viscosity of a pyrolyzing

system and pyrolysis temperature leading to formation
of mesophase.
Curve A - carbon having small mosaics of optical
texture; Curve B - carbon having large mosaics of
optical texture; Curve C - 'needle-coke' carbon
having large sizes of optical texture.
 44

5.5 Viscosity of Mesophase and Molecular Reactivity

Structure within a coke is determined by its optical texture

which in turn is determined by the viscosity of mesophase. The
viscosity of mesophase, in turn, depends on the relative reactivities
of the constituent molecules of the mesophase. In this sense, the
term reactivity refers to the ability of the constituent molecules
to polymerise. Mesophase of high viscosity
therefore results from the high reactivity of constituent molecules
undergoing cross-linking reactions. Because the viscosity is
relatively high at the beginnings of the growth process, no coal-
escence is possible. With relatively less reactive molecular
constituents, low viscosity mesophase results.

The detail of the chemistry of pyrolysis is reviewed (25, 26,

44). The presence of oxygen and sulphur tends to reduce the sizes
of optical texture. Nitrogen does not always act this way and may
enhance the size of optical texture. The presence of alkyl groups
or polar groups may reduce growth of mesophase (45). Additions of
sulphur to systems or alkylation can result in carbons which are
isotropic to optical microscopy. The use of dehydrogenation
catalysts can result in the formation of anisotropic coke from material
which otherwise had vaporised (46, 47). Mochida et a1. (46, 47)
examined the co-carbonization of the heterocyclic compounds
carbazole, phenazine and acridine with the Lewis acid-a1uminium-
chloride. Whereas in the absence of the catalyst these compounds
would have vaporized from the system, in the presence of the
catalyst, carbons of needle-coke appearance were formed from carbazole
and phenazine, with fine-grained mosaics (Table 1) being observed
in the coke from the acridine. In these systems, the compromise
between increases in average molecular weight and the rate of the
process favours the production of a liquid phase in which mesophase
can grow. However, if naphthalene, anthracene or pyrene are carbon-
ized with aluminium, sodium or potassium, the catalytic action of
these catalysts is so severe that isotropic carbon is formed,
because the rapid increase in molecular weight precludes the form-
ation of a liquid phase.

Mochida and Marsh (48) indicate that time and temperature are
not interdependent parameters for formation of mesophase, as in
established kinetics. The situation differs because of the contro-
lling influence of viscosity on growth and coalescence of mesophase.
Enhanced heating rates appear to be able to increase the size of
optical texture (49).

Structures of mesophase are influenced significantly by

carbonization under pressure, to produce the 'botryoidal' shapes
resembling microcarbon beads (50, 51).

A chemical consideration of major importance in determining

 45

resultant structures in cokes is the role of hydrogen transfer

reactions. Marsh and Neavel (52) compared the processes of
carbonization of coals and pitches with coal liquefaction processes.
The role of hydrogen in liquefaction is one of transfer to reactive
radicals so stabilizing the radical/molecule and preventing the
retrogressive reactions of coke formation. In carbonization
systems, transfer of hydrogen to reactive (transient) free radicals
could prevent those polymerization reactions leading to isotropic
carbon. The fluidity of the carbonizing system is therefore
enhanced allowing the development of anisotropic carbon (of larger
optical texture). See also refs. 53-56.

6. STRUCTUFE IN CARBONS

As discussed above, structures in carbons are based on the

graphite lattice, these structures (it must be stressed) not being
graphitic. Carbons differ from each other in essentially four ways
when compared with the graphite lattice:-

(a) The size of the constituent molecules, layers or lamellae,

varies from carbon to carbon. This is, the La dimension of a
'notional' crystallite as determined from broadening of the
(100) X-ray diffraction bond is a variant.
(b) The perfection of the hexagonal arrangements of carbon atoms
within the layer varies from carbon to carbon. These defects
constitute single atom vacancies (point defects), lines of
vacancies or of mismatch within the layer, and the presence of
heteroatoms such as oxygen, nitrogen and sulphur.
(c) The extent of planarity within constituent molecules varies
from carbon to carbon. The greater the number of defects
and heteroatoms within a molecule, the less likely is the
molecule to be planar.
(d) The distances between layers, i.e. the so-called 'd o02 ' spacing
of a notional crystallite and the extent of stacking perfection
between the layers or parallelism, vary widely from carbon to
carbon. The concept of a regular sequence of -A-B-A-B- stacking
cannot be applied to such imperfect structures. The extent of
parallel perfection is the Lc dimension of a notional
crystallite obtained from broadening of the (002) diffraction
line of graphite. In the porous, active carbons, with effective
surface areas >600 m2 g- l and pore volumes of -1 cm3 g- l the
availability of such large areas is a consequence of quite
irregular stacking of the non-planar constituent molecules, the
microporosity being the d-spacings of 0.5 to 2.0 nm between
the constituent molecules.
46

6.1 X-ray Diffraction Studies

Detailed analyses of X-ray diffraction methods as applied to

carbons are published by Ergun (4) and Gay and Gasparoux (57) for
graphitic carbons. Franklin (58) compared structural parameters
derived from X-ray data for a wide range of carbons from different
sources and so divided carbons into the two groups of non-
graphitizable and graphitizable. Blayden, Gibson and Riley (59)
made a comprehensive study of structure within a rank range of
coals, and the chars (non-graphitizable) and cokes (graphitizable)
derived from these coals. Riley (66) summarized studies of
structure within carbons and graphites. Ghosh and Riley (61)
investigated by X-ray methods the structural changes observed in
the carbonization of a coal-tar pitch.

It has to be mentioned that the use of the X-ray diffraction

technique has severe limitations of interpretation in so far as it
does not provide an accurate model of structure in carbons. There
are two principal reasons for this. Firstly, the broadening of the
major diffractions of graphite, e.g. (002), (004), (100), as
measured for the non-graphitic carbons, is the result of two
structural phenomena, i.e the small size of the
'notional crystallite'-r<-lO nm) and the strain within the lattice
resulting from the defective nature. Size broadening is not usually
separated from strain broadening (it may be impossible to do it in
a meaningful way) and hence the diffraction broadening is inter-
preted in terms of the sizes of the notional crystallite, L
(height) and L (width). Secondly, the X-ray diffractions fesult
from only thos~ parts of the carbon structure which contribute to
the Debye diffraction conditions. A part of the carbon which may
exist as single layers, or, as layers stacked in a manner whereby
they do not contribute to Debye diffraction, are'not included in
the model.
The contributions which X-ray diffraction studies have made to
studies of carbon structure lie in comparing, e.g. the relative
positions of plots of crystallite size, L ,L and d(002) with heat
treatment temperature (HTT) of the carbon~ inCthe HTT range -lOOOoC
to 3000 0 C (58, 59). Changes in structure within the same carbon
can likewise be compared, e.g. the effects of irradiation with
0.5 MeV electrons on carbon fibres (62).

6.2 Electron Microscopy Studies

The X-ray diffraction method is unsuitable to distinguish

clearly between non-graphitizable and graphitizable carbons of
HTT -lOOaoC. However, electron diffraction studies, making use of
dark-field imaging, are able to resolve the more crystalline
components of the graphitizable carbons. Oberlin (63) discusses how
new dark-field electron microscope techniques allow a radial or
complete exploration of reciprocal space permitting characterization
 47

of microstructures of carbons. Examples described include the

imaging of aromatic molecules in coals, kerogens and chars and
structural studies in pitch cokes, anthracene cokes and PAN and
pitch fibres.

In dark-field imaging, the more ordered regions of graphitizable

carbons show up as bright spots on the microscope screen. High
resolution fringe imaging clearly distinguishes between the non-
graphitizable isotropic carbons with a quite random arrangement of
constituent lamellae as in the porous active carbons (64), the
relatively more ordered but still non-graphitizable pitch-based
carbon fibres (65) and graphitizable carbons (37). Oberlin (66)
reviews electron microscope observations related to carbonization
and graphitization of a wide range of carbonaceous precursors.
Conclusions which relate to structure and drawn from this review
include:-

(i) The initial carbonaceous material which can graphitize, i.e.

a mesophase, must be made up of small basic structural units
(BSU) less than 1 nm in size.
(ii) These BSU must align approximately parallel in large areas of
size >1 ~m. There appears to develop a zig-zag texture
which allows defect mobility, with minimum energy during
graphitization (See Section 6.4).
(iii) From the many carbons studied a continuous process without a
gap has been found as far as the ability to graphitize is
concerned. This implies that the two groups of carbons, being
either non-graphitizable or graphitizable represent extremes
of a continuous availability of carbons in terms of
graphitization properties.

6.3 Raman Spectral Studies

As indicated above in discussions of the concept of the

notional crystallite, carbons contain disordered as well as ordered
material. The use of Raman spectra based on the vibrational charact-
eristics of carbonaceous lamellae has added to the understanding of
carbon structure. For example Mernagh et al. (67) suggest that the
Raman spectrum of Graphon carbon black (graphitized Spheron 6)
behaves like the comb~nation of two distinct resonance Raman spectra
arising from the disordered and ordered regions of the material.

6.4 Graphitization

Graphitization is the conversion process of structure from

that found in mesophase, green coke and calcined coke to the more
ordered structures of carbon heated to about 3000 0 c and which
exhibit 3-dimensional X-ray diffractograms (21, 22). This
conversion process can be brought about by thermal treatment alone
 48

or by use of graphitization catalysts (68-70). (See Section 3.1).

Figures 11, 12, 13 represent three ways of describing the

graphitization process.

Graphitizable
.'
24
I
Crystallite

-- ---
height Lc
.;
/ nm

------ --Non-graphitizable
12

1000 2000

Heat treatment temperature / °c

Figure 11. The general form of a Lc / HTT curve for a
non-graphitizable and graphitizable carbon (59,61).

Figure 11, illustrates the general form of the Lc/HTT curve for
a graphitizable carbon (59). The average height of the notional
crystallite apparently increases from about 1 nm to 2 nm, HTT to
-800 oC, subsequently decreases to about the 1 nm level and then
progressively increases with increasing HTT (see also Ref. 61).
The maximum and minimum in Lc are essentially spurious in that the
broadening of the (002) band associated with the decrease in Lc is
not the result of a decrease in height of the notional crystallite
but due to the strain introduced into the carbon lattice during the
growth process of the constituent lamellae. At this stage the
smaller sized constituent molecules become incorporated into the
carbon lattice (71). The non-graphitizable carbon exhibits similar
(slightly lower) values of L which do not rise as sharply with
c
 49

increasing HTT as L for graphitizing carbon (Figure 11).

c
Figure 12 illustrates the general form of the curve relating
height of the notional crystallite expressed as number of layers
per crystallite (M) to heat treatment temperature (HTT) according
to Franklin (58). Note that at values of HTT below 10000C all forms
of carbon merge together such that this method of analysis is
insensitive to the major differences in detail of the structures
found in non-graphitizable and graphitizable carbons.

30 Graphitizing

Number of
layers per
crystallite Partially graphitizing
(anthracite)
/ M 20

10
Non-~raphitizins

1000 :woo 3000

Ten~craturc / °c
Figure 12. The general form of the ~rystallit2 height
(number cf layers (M)) / HTT curve for all
forms of carbon.
50

Franklin (58) distinguished clearly the non-graphitizab1e

from the graphitizab1e carbons and included the intermediate
anthracite-derived carbons. However, as discussed by Oberlin (66),
carbon forms exhibit a continuum in structure between the two
extremes of Figure 12 without the gaps of Figure 12.

Figure 13 is a sketch or model of the graphitization process

as studied by high resolution phase contrast electron microscopy.
The model is intended to show the complexity of the structures
within graphitizab1e carbons and the way these structures change
with increasing heat treatment temperature of the carbon. The use
of modern electron microscopy techniques, of fringe-imaging, of
micro-micro diffraction, of dark-field viewing, etc., currently
represents areas of major advance into knowledge of structure
within carbons. The use of X-ray diffraction techniques, alone
are of limited value.

Figure 13. Diagram modelling the stages of the

process of graphitization (72).
 51

Figure 13 shows model structures at five stages of the process

of ultimate graphitization (72).

Firstly, HTT <1500 K, the structure of the coke shows

approximate parallel alignment of molecules as essentially absorbed
into the mesophase from the pyrolysing pitch system. The structure
within an individual layer is far-removed from the regular hexagonal
arrangement of carbon atoms of the graphite layer. Major vacancies
occur, together with chemisorbed hydrogen.

Secondly, HTT 1500 - 1700 K, there occurs lateral growth of

the individual molecules, but the extents of parallel alignment of
those layers are still limited. Structure within an individual
layer shows progression towards the graphite layer but gross defects
still occur of multiple vacancies and the presence of hydrogen and
heteroatoms.

Thirdly, HTT 1700-2000 K, lateral growth of the individual

molecules continues and this results in extensive bending (zig-zag)
and interlocking of the layers (see Ref. 66). These zig-zag
structures appear in all graphitizing carbons to different extents.
The rigidity in structures associated with this zig-zag alignment
may relate to puffing phenomena exhibited by needle-cokes containing
nitrogen and sulphur (73).

Fourthly, HTT -2000-2500 K, the constituent layers or molecules

within the coke material lose their zig-zag arrangements and become
progressively planar.

Fifthly, HTT >2500, the planar constituents of structure

adjust into the A-B-A-B alignments of the graphite single crystal.

All of the synthetic graphites are polycrystalline in nature,

a state far removed from single crystal graphite.

The graphitization process can be followed not only

structurally, as indicated above, but also by electronic (band
theory) studies (74, 76).
Strength or fracture properties of graphite relate to structure
(77, 78). The formation and properties of intercalation compounds
(lamellar reactions) of graphitizable carbons are also relevant
studies (79).

7. CONCLUSIONS

7.1 Although the basic structural unit within all carbons is the
sheet, layer, lamella, etc. based on the graphitic sheet, the number
of resultant carbon forms is indeed large.

7.2 Overall, carbons exhibit a continuous range of structure from

 52

the non-graphitic, isotropic carbons to the graphiti~able,

anisotropic carbon.

7.3 Predominantly, carbons appear to fall into one of the two

possible extremes of characterization, non-graphitizable or
graphitizable.

7.4 The non-graphitizable carbons are formed from parent precursors

which do not pass through a fluid phase, e.g. wood, cellulose, PAN,
thermosetting resins, etc. The carbon is thus a pseudomorph of
original structure. Extents of parallel stacking of constituent
molecules are limited.

7.5 The graphitizable carbons are formed from parent precursors

which pass through a fluid phase of pyrolysis, where the crystalline
order or high extents of parallel stacking of molecules, necessary
for subsequent graphitization, are established via the formation of
discotic, aromatic,nematic liquid crystal systems.

7.6 The variation in "chemical reactivity" of the molecules of the

parent feedstock accounts for variations in sizes and shapes of the
liquid crystals these properties being transferred to resultant
cokes. A wide variety of coke-types exists.

7.7 The low relative extent of parallel stacking in non-graphitizable

carbons, with non-planarity of molecules and cross-linkages, creates
the microporosity found in the adsorbing ac§ive carbons of surface
area >600 m2 g- l , pore volumes of up to 1 cm g-l.

7.8 The technique of X-ray diffraction provides relative values

of structural informations, useful in comparing series of carbons,
effects of HTT and other effects, etc. This technique does not
produce accurate models of true structure. The modern approaches
of electron microscopy and diffraction are much more informative,
providing accurate modelling of structure in the wide range of
availability of carbons and of the graphitization process. The use
of optical microscopy with polarized light to characterise optical
texture is extremely useful.

7.9 Carbons are initially classified into chars (non-graphitizable)

and cokes (graphitizable). However, a range of carbon forms exists
with specific morphologies, structures and applications in industrial
processes. The more important of these carbon forms are as below:-

7.9.1 Carbon blacks - acetylene, furnace, lamp and thermal.

7.9.2 Carbon fibres - isotropic pitch-based

mesophase pitch-based
PAN-based
gas-phase grown
 53

7.9.3 Activated carbon - highly microporous, used for adsorption

from the gas phase and solution.

7.9.4 Pyrolytic carbon - carbon deposited from the gas phase on a

suitable substrate - quite graphitizable.

7.9.5 Delayed coke - made within a delayed coker and comprising

the shot, sponge and needle-cokes
needle-cokes can be made from filtered coal-tar pitches.

7.9.6 Metallurgical coke - a hard, refractory unreactive coke

prepared by carbonization of coal blends for specific use in the
blast furnace as a reductant of iron ore.

7.9.7 Glass-like carbon - this is a monolithic non-graphitizable

carbon with a very high isotropy of structural and physical
properties and with a very low permeability for liquids and gases.
The original surfaces and the fracture surfaces have a pseudo-
glassy appearance.

8. ACKNOWLEDGEMENTS

The author wishes to acknowledge the support of Professor J.

L.Figueiredo in providing for this NATO Institute. Further HM is
appreciative of the secretarial support of Miss B.A.Clow in the pre-
paration of this manuscript and of Mrs. P.M.Wooster for diagrams.
Miss C.S.Latham assisted with proof-reading.

REFERENCES

1. Gilliam H.K. and M.P. Whittaker, Ext. Abs. 11th Biennial Conf.
on Carbon, American Carbon Society, Gattlinburg, 211 (1973).
2. Kasatochkin V.I., V.V. Korsha~ Yu.P. Kudryavtsev, A.M. Sladkov
and I.E. Sterenberg, Carbon, 11, 70 (1973).
3. Bragg W.H. and W.L. Bragg, Proc. Roy. Soc., (London) 89, 277
(1913). --
4. Ergun S., Chemistry and Physics of Carbon, Ed. P.L. Walker Jr.,
Marcel Dekker N.Y., 3, 211 (1968).
5. Debye P. and P. Scherrer Physik Z., 18, 291 (1917).
6. Hull A.W., Phys. Rev., 20, 113 (1922)-.-
7. Bernal J.D., Proc. Roy.:Soc., AlOO, 749 (1924).
8. Lipson H. and A.R. Stokes, Proc. Roy. Soc., A18l, 101 (1942).
9. Kochanovska A., Czech. J. Phys., 3, 193 (195~
10. Boehm H.P. and R.W. Coughlin, Carbon, 2, 1 (1964).
11. Moore A.W., Chemistry and Physics of Carbon, Eds. P.L. Walker
Jr., P.A. Thrower, Marcel Dekker, N.Y., ~, 223 (1981).
54

12. Amelinckx S., P. Delavignette and M. Heerschap, Chemistry and

Physics of Carbon, Ed. P.L. Walker Jr., Marcel Dekker, N.Y. !,
1 (1965).
13. Yang R.T., Chemistry and Physics of Carbon, Ed. Peter A. Thrower,
Marcel Dekker N.Y., 19, 163 (1984).
14. Thomas J.M., Chemistry and Physics of Carbon, Ed. P.L. Walker
Jr., Marcel Dekker, N.Y., 1, 121 (1965).
15. Hennig G.R., Chemistry and-Physics of Carbon, Ed. P.L. Walker
Jr., Marcel Dekker, N.Y., 2, 1 (1966).
16. Baker R. T .K., J. Catal.., 78-; 473 (1982).
17. Marsh H., Special Publication No. 32, Chemical Society, London,
(1st BOC Priestly Conference, 1977), 133 (1978).
18. Thomas J.M. and C. Roscoe, Chemistry and Physics of Carbon,
Ed. P.L. Walker Jr., Marcel Dekker N.Y., 3, 1 (1968).
19. 'Active Carbon', by M. Smisek and S. Cerny, Elsevier Publishing
Co., Amsterdam, (1970).
20. 'Activated Carbon' Ed. A. Capelle and F. de Vooys, Norit N.V.,
Anersfoort, The Netherlands, (1983).
21. Fischbach D.B., Chemistry and Physics of Carbon, Ed. P.L. Walker
Jr., Marcel Dekker, N.Y., 7, 1 (1971).
22. Pacault A., Chemistry and Physics of Carbon, Ed. P.L. Walker
Jr., Marcel Dekker N.Y., 7, 107 (1971).
23. Marsh H., Proc. 4th Intern. Conf. on Ind. Carbon and Graphites,
1974, Society of Chemical Industry, London, 2 (1976).
24. White J.L. and J.E. Zimmer, 'Surface and Defect Properties of
Solids', Chemical Society of London,S, 16 (1976).
25. Marsh H. and P.L. Walker Jr., Chemistry and Physics of Carbon,
Eds. P.L. Walker Jr. and P.A. Thrower, Marcel Dekker, N.Y.,
15, 229 (1979).
26. Lewis I.C. and L.S. S~nger, Chemistry and Physics of Carbon,
Eds. P.L. Walker Jr. and P.A. Thrower, Marcel Dekker N.Y.,
17, 1 (1981).
27. Forrest M.A. and H. Marsh, 'Coal and Coal Products: Analytical
Characterization Techniques', Ed. E.L. Fuller Jr., American
Chemical Society, Washington D.C., Symposium Series, 205, 1
(1982).
28. Marsh H. and C. Cornford, 'Petroleum Derived Carbons', Eds.
M. Deviney and T. O'Grady, American Chemical Society,
Washington D.C., Symposium Series, 21, 266 (1976).
29. Zimmer J.E. and J.W. White, AdvanceS-in Liquid Crystals, ~,
157 (1982).
30. Brooks J.D. and G.H. Taylor, Chemistry and Physics of Carbon,
Ed. P.L. Walker Jr., Marcel Dekker N.Y., 4,
31. Gray G.W. and J .W. Goodby, "Smectic Liquid Crystals", Ed.
Leonard Hill, Blackie Publishing Group, Glasgow (1984).
32. Marsh H. and J. Smith, 'Analytical Methods for Coal and Coal
Products, Ed. Clarence Karr Jr., Academic Press N.Y., 371
(1978) .
 55

33. Forrest R.A., H. Marsh, C. Cornford and B.T. Kelly, Chemistry

and Physics of Carbon, Ed. P.A. Thrower, Marcel Dekker N.Y.,
19, 211 (1984).
34. Ban L.L., D. Crawford and H. Marsh, J. App1. Cryst., ~, 415
(1975) .
35. Crawford D. and H. Marsh, Fuel, 55, 251 (1976).
3"6. Crawford D. and H. Marsh, J. Microscopy, 109, 145 (1977).
37. Marsh H. and D. Crawford, Carbon, 22, 413-(1984).
38. Auguie D., M. Oberlin, A. Oberlin and P. Hyvernat, Carbon, ~,
337 (1980).
39. Millward G.R. and D.A. Jefferson, Chemistry and Physics of
Carbon, Eds. P.L. Walker Jr. and P.A. Thrower, Marcel Dekker
N.Y., 14, 1 (1978).
40. Davis~, D.S. Hoover, L.D. Wakeley and G.D. Mitchell, J.
Microscopy, 132, 315 (1983).
41. Qian Z., D.E:iClarke and H. Marsh, Fuel, 62, 1084 (1983).
42. Hays D., J.W. Patrick and A. Walker, Fue1:-61 , 232 (1982).
43. Marsh H., D. Augustyn, C. Cornford, D. Crawford and G. Hermon,
Ext. Abs. 12th Biennial Conf. on Carbon, Pittsburgh, American
Carbon Society, 117 (1975).
44. Fitzer E., K. Mueller and W. Schaefer, Chemistry and Physics
of Carbon, Ed. P.L. Walker Jr., Marcel Dekker N.Y., 2, 237
(1971) •
45. Greinke R.A. and I.C. Lewis, Carbon, 22, 1 (1984).
46. Mochida I., E. Nakamura, K. Maeda andlK. Takeshita, Carbon, li,
341 (1976).
47. Mochida I., T. And 0 , K. Maeda and K. Takeshita, Carbon, ~,
453 (1978).
48. Mochida I. and H. Marsh, Fuel, 58,809 (1979).
49. Marsh H., I. Gerus-piasecka and-X. Grint, Fuel, 59, 343 (1980).
50. Marsh H., F. Dachi11e, J. Melvin and P.L. Wa1ker-rr., Carbon,
9, 159 (1971).
51. Forrest M.A. and H. Marsh, J. Mat. Sci., 18, 991 (1983).
52. Marsh H. and R.C. Neave1, Fuel, 59, 511 (1980).
53. Yokono T., H. Marsh and M. Yokon~ Fuel, 60, 607 (1981).
54. Yokono T., T. Obara and Y. Sanada, Carbon:-22, 169 (1984).
55. Korai Y. and I. Mochida, Fuel, 62, 893 (1983).
56. Mochida I., K. Tamaru, Y. Korai~H. Fujitsu and K. Takeshita,
Carbon, ~, 231 (1982).
57. Gay R. and H. Gasparoux, Les Carbones, Vol. 1, Masson and Cie
120 Boulevard St. Germain, Paris, 62 (1965).
58. Frankin R.E., Act. Cryst., 4, 253 (1951).
59. B1ayden H.E., J. Gibson and-H.L. Riley, Proc. Conf. on U1tra-
fine Structure of Coals and Cokes, BCURA, 1943, Leatherhead,
176 (1944).
60. Riley H.L., Quart. Rev. Chern. Soc., 1, 59 (1947).
61. Ghosh S.P. and H.L. Riley, Fuel, 27,-56 (1948).
62. Kent M., K. Wolf, J.D. Memory, R.~ Fornes and R.D. Gilbert,
Carbon, 22, 103 (1984).
63. Ober1in~, Carbon, ~, 7 (1979).
56

64. Evans M. and H. Marsh, Characterization of Po~ous Solids,

Eds. S.J. Gregg, K.S.W. Sing and H.-F. Steock1i, SCI London,
67 (1979).
65. Bennett S.C. and D.J. Johnson, Carbon, ~, 25 (1979).
66. Oberlin A., Carbon, 22, 251 (1984).
67. Mernagh T.P., R.P. Cooney and R.A. Johnson, Carbon, ~, 39
(1984).
68. Marsh H. and A.P. Warburton, J. of App1. Chern., 20, 133 (1970).
69. Oya A. and H. Marsh, J. Materials Science, 17, 309 (1982).
70. Holstein W.L., R.D. Moorhead, H. Poppa and MI. Boudart, Chemistry
and Physics of Carbon, Ed. P.A. Thrower, Marcel Dekker, N.Y.,
18, 139 (1982).
71. Marsh H. and D. Crawford, Carbon, 22, 413 (1984).
72. Marsh H. and J. Griffiths, Ext. Ab~ of Int. Symp. on Carbon,
Toyohashi, 1982, Kagaku Gijutsu-sha, Tokyo, 81 (1982).
73. Mochida I. and H. Marsh, Ext. Abs. 17th Biennial Conf. on
Carbon, Amer. Carbon Soc., June 1985, Lexington, Kentucky, USA,
276 (1985).
74. Spain I.L., Chemistry and Physics of Carbon, Eds. P.L. Walker Jr.
and P.A. Thrower, Marcel Dekker, N.Y., 16, 119 (1981).
75. Tsukuku T. and K. Sugihara, Chemistry and Physics of Carbon,
Eds. P.L. Walker Jr. and P.A. Thrower, Marcel Dekker, N.Y.,
12, 109 (1976).
76. Dehhaes P., Chemistry and Physics of Carbon, Ed. P.L. Walker Jr.
Marcel Dekker, N.Y., 7, 193 (1971).
77. Brocklehurst J.E., Chemistry and Physics of Carbon, Eds.
P.L. Walker Jr. and P.A. Thrower, Marcel Dekker, N.Y., ~,
145 (1977).
78. Zhao J., J.L. Wood, R.C. Bradt and P.L. Walker Jr., Carbon,
19, 405 (1981).
79. Robert M.C., M. Oberlin and J. Mering, Chemistry and Physics
of Carbon, Ed. P.L. Walker Jr., Marcel Dekker, N.Y., 10,
141 (1974).
 57

COAL STRUCTURE

K. Mark Thomas

British Gas
London Research Station,
Hichael Road, London, SW6 ZAD., U.K.

INTRODUCTION

The properties of a solid depend on the arrangement of the

atoms as well as the number and type of atoms of which it is
composed. The empirical formula of a solid compound describes the
latter but does not reveal the arrangement of atoms or how the
compound will react. Structure can be defined as the way in which
the component parts of a complex whole are interrelated. Clearly a
knowledge of the structure of a solid is important for understanding
its pro pert ies.

If pure compounds are characterised it is usually possible to

define the structure precisely. Structural determinations can be
achieved by a variety of chemical and spectroscopic methods and in
the case of crystalline solids, by X-ray diffraction methods. A
crystalline solid is packed regularly in three dimensions and it is
poss ible to def ine a unit cell which is the bas ic build ing block
which translates into a three dimensional solid. The size, shape,
symmetry and content of the unit cell represent the structure of the
solid. This is usually expressed in the atomic dimensions of the
molecule and the way the molecules pack within the unit cell. This
gives a precise description of the solid including bond lengths,
angles and conformation as well as intramolecular distances and
or ientation. This is true for even quite complex molecules 1 ike
proteins which have high molecular weights and are made up of
repeating amino acid sub-units which form 'building blocks' but can
nevertheless be highly ordered crystalline solids.

The structural characterisation of coal is however much more

difficult largely because of its heterogeneous and macromolecular
 1.11
00

TABLE 1

COAL STRUCTURE - THE PROBLEM

Structure Definition Description

Primary Identification of chemical Macromolecules built up of

(0.1 -lOOnm) 'sub units' present and the condensed aromatics and
nature of the bonds between bridges.
the units.

Secondary Local orientation of the Reflection of the underlying

(lnm -l0lJ.m) structures. molecular structure and
orientation of macromolecules
in the coal structure-pore
structure, fractures, etc.

Tertiary Long range orientation of Structure within macerals and

(llJ.m -lO-lm) locally parallel structures. aggregates of these macerals
(lithotypes). Cross-linking
and entanglement of macro-
molecules.
 59

nature. Coal is generally considered (1,2) to be derived from plant

material which has decayed and been incorporated in strata and then
altered by ctemical and physical processes existing under geological
conditions occurring over millions of years. The starting material
for the coal, being plant material, was heterogeneous and combined
with the natural variation in geological processes, leads to coal
having a wide range of properties and being visibly heterogeneous.
Clearly it is not poss ible to def ine the s truc ture of coal' in the
same precise way as for a pure crystalline compound.

The definition of structure needs to be considered in the

widest sense when discussing coal. It has regular and repeating
features which have a definite chemical and physical structure but
the degree to which these features can be defined is much less than
for a crystalline solid. A knowledge of these properties is
important for unders tand ing coal behav iour under var ious condit ions.
The problem of defining coal structure lies in the many different
features and aspects which need to be considered (3). This is
illustrated in Table 1 by considering various aspects of coal
structure on the basis of size.

The heterogeneous nature of coal means that it has to be

sampled in order to provide a representative sample for laboratory
experiments. The involvement of size reduction techniques in this
process means that some information about the macroscopic structure
can be lost (2) in this process.

The approach which has been made in this review is to divide it

into three main topics, accord ing to the type of informat ion
provided, as follows:

1) Physical structure.
2) Molecular structure.
3) Macromolecular structure.

There will be some overlap between sections because of the

nature of the topic. The molecular structure is mainly concerned
with the primary structure whilst the physical and macromolecular
structures relate to different aspects of the secondary and tertiary
structures. The objective of the review is to provide an
introduction to the many different aspects of coal structure while
referring to reviews of various topics where more detailed
information can be obtained.

PHYSICAL STRUCTURE

The physical properties of coal and the changes in these

properties during heating are to a large degree a consequence of the
physical structure which represents the spatial arrangement of large
macromolecules and smaller molecules and the extent of cross
60

TABLE 2

ASPECTS OF THE PHYSICAL STRUCTURE OF COAL (4)

Description of features

Heterogeneity Macerals
Distribution of inorganic material

The porous structure of coal Porosity

Pore size distribution
Internal surface area

Effects of heat on coal Changes in pore structure during

structure heating
Devolatil izat ion
Plastic properties

TABLE 3,

MACERALS IN COAL AND THEIR ORIGIN (6)

Group Maceral Origin

Vitrinite Telinite Humic gels

Collinite Wood, bark or cortical
tissue.

Exinite Sporinite Fungal and other spores

Resinite Resin, plant secretions
Cutinite Leaf cuticles, etc.
L iptodetr inite
Alginite Algae remains

Inertinite Micrinite Unspec if ied

Macrinite detrital matter
Semifusinite Carbonised woody
Fusinite tissues
Scle rotinite Fungal sclerotia and mycelia.
Inertodetrinite
 61

Fig 1 A crossection through a lump of coal showing vitrain

with durain bands
62

linking. The physical properties often vary systematically with rank

and can be studied (4) without a detailed knowledge of the chemical
structure being available. The physical structure of coal has been
reviewed recently in detail (4).

The main aspects of the physical structure which need to be

considered are given in Table 2. The heterogeneity in coal is of
prime importance in the structural characterisation of coal. It is
clearly heterogeneous at the macroscopic level and it may have a
banded structure as shown in Figure 1. These bands are lithotypes
and there exist several methods of describing coals megascopically.
Figure 1 also shows that a coal lump is cracked and fissured which
is another aspect of the physical structure. Clearly the
heterogeneity in coal is a major problem in the structural
characterisation. At the microscopic level, coal can be considered
to be made up of macerals which are the organic microscopic
constituents and minerals. The former may be readily identified by
optical microscopy. The constituents and their distribution may be
considered as being responsible for determining the properties of a
given coal. Coal petrology has been reviewed (5) in detail previously.

The macerals can be divided into three main groups - vitrinite,

exinite and inertinite, and various subgroups, and these together
with their original composition (6) are shown in Table 3.
Photomicrographs of various macerals are shown in Figure 2. The
minerals are widely distributed in coal as discrete particles which
vary considerably in size and distribution. The heterogeneous
distribution of some minerals in coal is shown in Figure 3.
Inorganic atoms, for example oxygen, sulphur and nitrogen, are also
incorporated in heterocycles and bridges in the organic structure.
The distribution of mineral matter in coal has been reviewed (7,8)
in detail previously and a list of common minerals found in coal is
given in Table 4. The inorganic and mineral matter in coal may arise
from several sources. The material can arise from the plants which
formed the coal or may have been introduced at a later date from
various sources by geological processes.

Trace amounts of virtually all the elements in the Periodic

Table are found in the various coals (9). Although the percentage
quantities of trace elements are small, the use of large amounts of
coal in conversion processes means that their presence can be
significant in terms of utilization, for example corrosion and
environmental considerations.

X-ray diffraction analysis is the preferred technique for

studying the structure of a pure crystalline solid. However coal is
heterogeneous and non-crystalline and this results in a diffuse
diffraction pattern (10) so that the structure of coal cannot be
determined in the same manner. Nevertheless it can provide much
 63

(a) (~

(c) ~)

(e) (f)
Fig 2 Optical micrographs of various coal macerals

a) Vitrinite b) Sporinite c) Micrinite

d) Sclerotinite e) Semifusinite f) Fusinite
64

TABLE 4

SOME MINERALS FOUND IN COALS (7,8)

Kaolinite (Al 4 Si 40 l0 (OH)a)

Illite (Kl-loSAl4Si7-60SA11-loS020(OH)4)
Montmorillonite Al2Si40l0(OH)2oH20

Chlorite (Mg,Fe,Al)6 (SiAl)4 010 (OH)a

Silica Quartz (S10 2 )

Sulphides pyrites

Marcasite (FeS 2 ) (orthorhombic)

Pyrrhotite (Fel-XS)

Sphalerite (ZnS)

Chalcopyrite (CuFeS 2 )

Galena (PbS)

Carbonates Calcite (CaC0 3 )

Aragonite (CaC0 3 )

Dolomite (CaMg(C0 3 )2)

Siderite (FeC0 3 )

Strontianite (SrC0 3 )

Witherite (BaC0 3 )

Magnesite (MgC0 3 )

Ankerite (Ca(FeMg)C0 3 )

Sulphates Gypsum (CaS0 4 2H 20)

Anhydrite (CaS0 4 )

Bassanite (CaS0 4!..2H 20)

Barytes (BaS0 4 )
 65

(a)

(b)

(c)
Fig 3 Scanning electron micrographs of minerals in coals

a) Pyrites b) Pyrites c) Gypsum

66

useful information on the arrangement of the carbon atom skeleton.

The broad peaks in the diffraction patterns of a wide range of coals
correspond in position to the sharp peaks obtained from crystalline
graphite. Broad peaks in X-ray diffraction patterns are usually
associated with a small crystallite size and marked disorder.
Similar X-ray diffraction curves are observed for highly disordered
carbons. The latter can be explained by the presence of small
graphite-like planar layers which are not stacked regularly with
respect to each other. The X-ray scattering arises from the regular
array of scattering centres in a planar hexagonal network. The
layer is terminated by deviations from planarity whereas substituent
atoms will extend the planarity and increase the apparent
crystallite size. Therefore the histograms of layer sizes obtained
(10) from diffraction data are not strictly quantitative. The
distribution of stacks of parallel layers can be obtained from the
002 reflection in the X-ray diffraction curve. Studies of a series
of coals shows that there are mainly single layers in low rank coals
but rarely more than three layers. In comparison more layers are
stacked in coking coal.

Investigations (10) of a series of coals using X-ray

diffraction methods has resulted in the proposal of three main types
of structure for vitrinites-

1. Low rank coals - an open structure of randomly oriented

and cross linked condensed aromatic
layers.
2. Medium rank - a structure with fewer cross-links,
coking coals reduced porosity and a moderate degree of
ring orientation.
3. High rank coals- larger layers with a greater degree of
orientation and an oriented pore system.

Models of these structures are illustrated in Figure 4. Exinites

and inertinites have slightly different X-ray diffraction curves.
Inertinites appear to have larger and better stacked layers. The
diffraction profiles of exinites show a band (y-band) with a
spacing of 0.4-0.Snm which is not found in vitrinites. This is
thought (11) to be due to non-aromatic groupings for example
alicyclic layers or aliphatic side chains.

Coal lumps are cracked and fissured showing another aspect of

structure (see Figure 1). The microscope also shows that coal
consists of holes, cracks and macropores which may represent gaps
between particles. At the atomic scale, coals may have an ultra-fine
structure with pore sizes down to a few nanometres in size. The
pores represent holes in or between macromolecules. Some of the
pores may be open giving gases access to the internal surface area
of the coal whilst others are closed but may open during reaction.
The pore structure can be quantified by a variety of methods. The
 67

Disordered
material a) 80% C

approximately parallel

Hydrogen
bond b) 89% C

Disordered
material
Lamella ~
X~---IV ~
~~-;::v-~
~
~~~ .... IV c) 94%C
---v. ~ vv----'V'

Fig 4. STRUCTURAL MODELS OF COAL a) OPEN STRUCTURE

b) "LIQUID" STRUCTURE c) ANTHRACITIC STRUCTURE
AFTER HIRSCH (10)
68

X-ray diffraction profiles of coals show marked diffuse scattering

at low angles which has been attributed to the presence of porosity.
The intens ity of this scattering is at a minimum for coking coals
which is expected from other methods of measuring porosity. The pore
structure can also be quantified by adsorption of gases (12) and
penetration of liquids and these are the more commonly used methods.
Tqe sorption-desorption equilibria of gases is a function of the
properties of the sorbate and the temperature of adsorption as well
as the surface area of the coal and therefore the surface areas are
used in a comparative way since the absolute values are uncertain.
The pore s truc ture of coal is formed by a wide range of pore sizes.
The IUPAC definitions (13) are based on their width-macropores
>sOnm; meso pores 2-s0nm ; micropores <2nm. The macropores act as
transport pores affecting the rate of diffusion into the meso and
micropores. The mesopores branch off the macropores and lead to the
micropores which constitute the major part of the internal surface
area. This can be visualised like a tree branching out.

The structural changes which occur in coal when it is heated

are also of considerable interest since all coal conversion
technologies involve the application of heat. The major changes
which occur involve decomposition to coke. The primary
devolatilization involves tar and gas evolution which is accompanied
by caking and swelling for some coals and leads to the formation of
coke. This is discussed in detail in the accompanying review on coal
characterisation for gasification processes. An important aspect is
the change in pore structure during the carbonization process. Pore
structure development occurs mainly in the plastic range. The
process has been d iv ided into four stages (14):

1. Pore nucleation within large particles.

2. Fus ion of part icles with the concomitant swell ing caus ing
fill ing of intrapart icle' vo ids.
3. Post fusion pore growth to a maximum size.
4. Compaction of the solid involving a decrease in pore size.
This occurs at around the resolidification temperature.

These stages together with corresponding dilatometry, Gieseler

plastometry and mean pore sizes versus temperature are shown in
Figure 5. Dilatometry and plastometry involve the release of
volatiles, the changes in pore structure will accompany these
phenomena and be related in a complex manner.

MOLECULAR STRUCTURE

The molecular structure of coal has been reviewed (15)

recently. The organic part of coal is composed of macromolecules and
a knowledge of the chemical sub-units of the macromolecules is
 69

T.G.A. curve

RUh~ilatometer
curve
Gieseler fluidity
curve

Q)
c
0 Q)
C N c

- ...
.2 tn
0
N

-'"
c tn
.-otn
:J

--
:J 0 C) Q)

:J Q) 0 .5 .::s:.
0
tn C. CJ CJ
tnj! - Q)
:J CJ
0 ; .- a. -
'" - E >- a.
...
. . . .-. . E .r:. E
0'"
C)
0 0 '" '" 0 ._ 0 Q)
a. a.a.o:r:::O en

Pores/cm

I
-- ----------

300 400 500 600

Fig 5
THE RELATIONSHIP BETWEEN COKING CHARACTERISTICS
AND PORE STRUCTURE DEVELOPMENT (14)
70

important for understanding the reactions of coal and visualizing

the composition of a heterogeneous material. The problem in giving a
molecular structure to coal is one of assigning a statistical
average structure to a mixture of macromolecules which are not
composed of repeating monomeric units. The main information
required concerning the molecular structure is as follows:

1. Size distribution of macromolecules.

2. Number, type and distribution of:
a) functional groups
b) heteroatoms
c) aromatics
d) aliphatics
3. Nature and distribution of bridging structures between
aromatics.
4. Cross linking in the macromolecular structure.

The separation and characterisation of coal macromolecules is

extremely difficult.

Attempts have been made to separate the macromolecules by

extraction and solubilisation using a variety of chemical methods,
for example, depolymerisation, oxidation, reduction etc. The
chemical methods provide information on the size and composition of
coal fragments rather than macromolecules. These fragments result
from the mild breakage of cross-links or more drastic conditions
involving extensive degradation of the coal macromolecules. The
problem of breaking up coal into fragments in a consistent manner
means that the significance of the results can be questioned.
However there does appear to be broad agreement with other
techniques. Even in the case of extensive thermal degradation of
coal to yield tars, the nitrogen and sulphur heterocyclic compounds
observed and also retained in the coke are similar to those found
under much milder conditions. It is likely (16) that the
heterocyclic aromatic structures are also found in coal although
some may be present as NH, SH and SS groups etc. Oxygen is the
predominant heteroatom in coal and the distribution of oxygen
functional groups has been estimated (17-19). Data on the
distribution of sulphur groups has been obtained (20). The
heteroatom and functional group content of coal has been reviewed
(21) recently.

3.1 Methods of Investigating the Organic Structure of Coal

The chemical and physical methods that can be used (22) for
investigating the organic structure of coal and the information
obtained from these methods are given in Tables 5 and 6
respectively. The problem is one of piecing together information
from a wide variety of techniques (23-57) on a number of
heterogeneous coals which are often poorly characterised.

MAIN CHEMICAL METHODS USED FOR INVESTIGATING
THE ORGANIC STRUCTURE OF COALS (22)
Chemical Method
Depolymerisation
with phenol (23-28)
via reductive alkylation
(29-33)
via non-reductive
alkylation (34)
Oxidation
- aqueous sodium dichromate
(35)
- Chromic acid (36)
- nitric acid (37,38)
- H20 2 , trifluoracetic
acid, H2 SO 4 (39,40)
- sodium hypochlorite
(41-44)
Reduction
- hexamethylphosphoramide,
sodium t-butanol (45)
- sodium hydroxide/
methanol (46)
Derivatisations
(silylation/acetylation)
(17)(47)
Dehydrogenation (48)
71
TABLE 5
principal Information Obtained
Molar mass distributions of coal
fragments
Details of some aromatic and
aliphatic groups
Molar Mass distributions of coal
fragments
Details of some structures
present
Details of some aliphatic
structures
Details of aromatic structures
Details of aliphatic structures
Details of some aromatic and
aliphatic structures
Details of aliphatic
structures
Estimate of aromaticity
Molar Mass distribution of coal
fragments
Estimate of hydroxyl group
concentrations
Details of hydroaromatic
structures
72
MAIN PHYSICAL METHODS USED FOR INVESTIGATING
THE ORGANIC STRUCTURE OF COALS (22)
Physical Method
Solid state 13 C NMR (49-53)
Infrared spectroscopy
including FTIR (54,55)
Size Exclusion Chromato-
graphy (56)
Field ionisation mass
spectrometry (57)
Densimetric constitution
analysis (1)
Statistical approaches
(58-64)
TABLE 6
Principal Information Obtained
Estimate of aromaticity
Distribution of aliphatic carbon
Estimate of fraction of aromatic
hydrogen of total hydrogen
Identification of carbonyl
functions
Estimate of hydroxyl group
concentrations
Molar Mass distribution of
extracts
Molar Mass distribution of
solvent extracts
Estimate of aromaticity
Model structures for coal
 73

The characterisation of coal fragments obtained by a variety

of extraction methods is particularly significant in determining
the organic structure of coal and the characterisation of the
extracts presents severe difficulties. The reasons for this are the
wide molar mass distributions and the problems associated with the
characterisation of mixtures. The molar mass distributions of the
extracts are usually obtained by size exclusion chromatography
(SEC) (56). This provides useful information on the degree of
degradation of the coal by the extraction techniques and the
macromolecular structure of the coal. It is also a possible
chromatographic method of fractionation. The technique is based on
the fact that the size of a wide range of aromatic molecules is
directly related to the retention volume for certain types of
chromatographic column. There is a linear response between the
total permeation and exclusion limits and this can be calibrated by
suitable standards. Columns with various molar mass ranges are
available. Field ionization mass spectrometry can also give (57)
molar mass distributions but the mass range is limited.

Nuclear magnetic resonance (NMR) is a very important technique

for characterising coal extracts. This technique provides details
of the functional groups, and aromatic and aliphatic structures in
the extracts and estimates of carbon aromaticity can be obtained.
Information can also be obtained from cross polarisation magic
angle spinriing (CPMAS) NMR on solid samples as a comparison with
that obtained for extracts. This technique gives an estimate of
carbon aromaticity. Figure 6 shows the (CPMAS) NMR spectra for the
whole coal and the vitrinite, exinite and inertinite fractions
derived from the coal. The results show that the carbon aromaticity
decreases in the order inertinite > vitrinite > exinite.
Aromaticity in coal has been reviewed (49-51) in detail recently.
Infrared spectroscopy is also a useful technique for estimating
HAr/HAl ratios and hydroxyl and carbonyl concentrations. A
wide range of other chromatographic and spectroscopic techniques
are used to characterise coal extracts. Those outlined above are
intended as an illustration of the methods used in the
investigation of the organic structure of coal.

3.2 Model Structures for Coal

Three main types of model (65) have been proposed for the
organic structure of coal:
1. The aliphatic/polyamantane model (66).
2. The aromatic/hydroaromatic model (58-64).
3. The molecular sieve model (65).
The aliphatic/polyamantane model has been discounted (15). The
molecular sieve model considers coal as consisting of relatively low
molecular weight, hydrogen-rich molecules trapped in a hydrogen
deficient matrix of high molecular weight material. In the
74

fa 0·54

I I i i i i i
250 200 150 100 50 o -50
ppm

Fig 6. THE AROMATIC CARBON CONTENT (fa) OF A

COAL AND ITS MACERALS
 75

molecular sieve model the extractable molecules fill the pores of

the macromolecular network which binds these trapped molecules by
donor-acceptor or hydrogen bonds with the network. Solvents may
displace these molecules by forming stronger donor acceptor bonds.
The amount of extractable material correlates (67) with the electron
donor and acceptor properties of the solvent.

The concept of the trapped or mobile phase which consists of

relatively small molecules partly trapped or physically held in the
macromolecular network has been proposed. This raises questions
(68) as to the nature of the molecules, the bonding interactions
with network and whether or not there are significant amounts of not
readily extractable small molecules trapped in coals. Attempts have
been made to distinguish between the mobile and immobile network
phases by NMR spectroscopy but this is still the subject of
considerable discussion (68).

The aromatic/hydroaromatic model is generally accepted. It

contains possible sites which can act as donor-acceptor sites as in
the molecular sieve model. The average size of the aromatic rings in
this model are quite small (2-3) and generally range from 1 to 5
aromatic rings for high volatile coals (~80C% dmmf).

The information produced by the various techniques can be used

to derive a model molecular structure for coal. These structures
have been criticized by Larsen (69,70) on the grounds, that "for most
purposes a knowledge of the extremes of the range of structures
present will be much more useful in predicting the chemical
behaviour of coals". Given (68) has also expressed reservations
concerning the usefulness of the average molecular structures.
However, they do form a convenient way of visualizing the types of
structures expected in coal and the way in which it may behave in
chemical terms. It would be unwise to place great reliance on model
structures. They should be regarded as discussion points for
understanding the behaviour of coal.

The important aspect of model building is that it should not

only fit the characterisation data but also have predicted behaviour
similar to that observed for coal.

In 1960,Given (58) reviewed the chemical, spectroscopic and

crystallographic information available at that time to construct a
molecular model of coal (see Figure 7). The basis for the model was
the X-ray work of Hirsch and van Krevelens statistical
constitutional analysis. It contained one and two membered rings
linked together by dihydroaromatics, pyridine type rings, quinones
and hydroxyl and carbonyl groups. The tropolone and trytycene
molecules were unusual features of the model. The model of Hill and
Lyon (59) was quite different and involved small aromatic ring
systems held together by ether, methylene and various other sorts of
76

Fig 7 MODEL STRUCTURE PROPOSED BY GIVEN (58)

(Reproduced with permission)
 77

80% Vitrain
o 0
H (a)

90% Vitrain

.. points where dimerisation could occur

Fig8 MODEL STRUCTURES PROPOSED BY PITT (11)

(Reproduced with permission)
78

H H
~~" H H °
H
H.....
°I
H
I ..... H
~ 1 ,I ~
I- C-~' ~ o-~, Ji ti
H
/ ~ ~ 'HAW
H ~ ..... H2 6=0
,S
I H~'C-
,...... H
H NH2
H2 1:1 1:1
H'' ,
'"'0
2 I,'H Ii sI H ~
H-
" ci
,- ,-
°
~ I I H I H-C-H H I H H
H
H~'
I 'H-y-H t:I
,
t:I H I H H2
, ..... I I
t' . . . O"C
I
H3
~
, - cr-Yii'~-O-" ' 0 H ~ H H
~
H
~ ~ , H HH{~ 'c' I I
'ceo'
°
I'
H-C-H
I
H-t-HS
J.
" I t;t2...
H2H H I ,
2
I
H,
1:1'..... :::...
N
I
.OH
'H
°
H-C-H H,~\,,)/0 H-C- H H H' I
H-C-H
I . . . ~H 1 I I
H-C-H H I HO'(fX?0 ~ "OH H
I H H H I I H
'c""" ~
"O.....H
cr
I H '0 I
~" IjI H I
'!PI
H" ~
I
t;t2
_H
..... ~
H,
n
H"')~""'H
...... H
I
H
I
..... ~ ..........
OH
HO/~, ~ .....H2 H H H I I
HH H ,,~,
H H" H2 H I
20H H
H

Fig. 9 MODEL STRUCTURE PROPOSED BY WISER (61)

 79

Fig 10 The coal structure model of Shinn (64) showing the

build up of aromatics, functional groups and
aliphatics to form the structure
80

bridges. Ladner (60) modified Givens model to take NMR data into
account and proposed phenanthrene type ring systems. Pitt
proposed (11) similar models for 80% and 90%C vitrain based on 9-10
dihydrophenanthrene type of linkages between aromatic units. These
are shown in Figure 8. It was suggested that these types of
molecules exist in an entangled state in coal. This model was
slightly hydrogen-rich but this would be reduced by dimerisation at
the sites shown in Figure 8.

The model of Wiser (see Figure 9) is based (61) on small

aromatic rings linked together by a wide variety of bridges, for
example, ether and methylene. Hence it is straightforward to
conceive how this may react to produce smaller fragments which are
basically condensed aromatic systems derived from breaking the weak
bridge bonds. Other models have been constructed by Solomon (62) and
Kosky and Spiro (63). The latter utilized space filling models to
show the three-dimensional characteristics of the molecules which
are not apparent from the two-dimensional representations normally
used. Recently, Shinn (64) has reviewed the literature on the
functional group, aromatic and aliphatic distributions in coal and
also data from coal liquefaction processes to derive his model.
Figure 10 shows the building of the model starting from the
aromatics (based on data from Winans et al) adding in the functional
group distribution (Blom, Cronauer, Liotta et al for oxygen groups
and Attar et al for the sulphur group) and finally adding in the
aliphatics (Calkins and Deno et al). This clearly shows the
generally accepted roles for the components of the coal structure.

The models (58-64) which have been discussed show marked

differences in structure and thes'e together with some other models
are compared in Table 7. This Table shows that there is substantial
agreement on the composition of coal in terms of functional groups,
heteroatom content, bridges, and heteroaromatic, aromatic and
hydroaromatic structures. This is summarised below:

Functional groups O-H, COOH,>O,C=O, ~N: ,>N-H

Bridges ether, methylene, -o-C, hydrogen bonds
Aromatic ring size 1-5
Hydroaromatics dihydroanthracene,
9-10 dihydrophenanthrene

A general picture of coal with the functional groups being

mainly situated on small aromatic structures held together by
various types of bridges (mainly ether and methylene) or
hydroaromatic structures is commonly accepted. The aliphatic part of
coal is usually considered to be side chains or involved in bridges.
Some molecules involving long methylene chains have also been
suggested (71) and this is supported by the presence of long chain
paraffins in tars.
 TABLE 7 DETAILS OF PROPOSED MODEL STRUCTURES FOR COAL

Hetero- Type of Main

Functional Bridges Aromatic Aromatic Hydroaromatic Comments
Author groups Structures Ring Size Structures

Given (1960) ~N: Quinone dihydro- Based on Hirschs' X-ray

(58) -O-H - Pyridine 1-2 anthracene data and van Krevelens
~C=O Tropolone constitutional analysis I

Hill and ~N: -R-S-

Lyon (1962) -O-H,>NH -0- Pyridine
(59) -COOH -(CH 2 )n- Pyrrole 2-6 -
R-S;:;:C=O -c-o-
Ladner (1963) -O-H Quinone 9-10 dihydro 82%C
(60) >c=o -0- Pyridine 1-3 phenanthrene Vitrinite
:;:=N: fluorene
Wiser (1973) :;:C=O,>S;OH -(CH2 )n- Pyridine
(61) -S-H,>,N: -S- Thiophene
-NH 2 ,>0 -S-S- Furan 1-5 -
-COOH,>NH -O-,-O-C- pyrrole
Pitt (1979) -OH - Quinone 9-10 dihydro 80%C
( 11) ~N: pyridine 1-3 phenanthrene Vitrain
~C=O
Solomon(1981) -OH,:;:?N: -O--H-O Furan 9-10 dihydro PSOC 170
(62) -S-H;>NH -O-C- pyrrole 1-5 phenanthrene C = 81%
-COOH>S, -(CH 2 )n- Thiophene + dihydro-
Pyridine anthracene
Spiro and -OH -S- Furan 1-4 dihydro- 82.5%C
Kosky (1982) ~N-H?S -(CH2 )n- Thiophene anthracene
(63) :;:=N:!>O -O-C- Pyridine
Shinn (1984) -OH, ~N: -(CH2 )n- Pyridine
(64) -COOH,-SH (-;.CH-) Quinone 1-3 dihydro-
00
-NH2 ,>NH -NH- Furan anthracene t-'

-C=O,>S -0- pyrrole ---

82

The models illustrated above are mainly guides to possible

structures that represent a statistical average of composition
data. They should be used with care, bearing in mind the
heterogeneous nature of coal. The main emphasis has been in the
construction of models for coals with C ~ 80(% dmmf). Models have
also been proposed for low rank coals (63) and high rank coals
(11,63).

MACROMOLECULAR STRUCTURE

The evidence for the existence of macromolecular nature in coal

has been discussed for many years. The properties of coal which
provide strong evidence for the macromolecular structure of coal
were outlined by van Krevelen (1). The evidence is given below:

1. Coals are not soluble but part may be extracted with a

solvent.
2. Coals swell in various solvents without being decomposed by
them even at high temperatures unless thermal degradation or
reaction occurs.
3. Coals are viscoelastic. When subjected to a stress they
deform until a limit determined by the cross-links is
reached. They return to their original shape if the force is
removed. This is evidence for covalent cross-links being
involved. Weak associative forces such as hydrogen bonding
would not explain the results.
4. Hydrogenolysis increases the amount of soluble material from
coals and this material is similar to the material extracted
from untreated coal. Also mild oxidation produces
simultaneously aliphatic and aromatic carboxylic acids of
low molecular weight.

The heterogeneous nature of coal is a major obstacle in the

investigation of its macromolecular structure and this structure
will vary for different macerals. The application of polymer science
techniques, in particular solvent induced swelling, to the study of
the macromolecular structure of coal has received considerable
attention and been reviewed recently (69,70). Models for coal
networks have been proposed (69) with either various types of
covalent cross-links through atoms and aromatic groups or also
entanglements of macromolecules (11). It is also possible that a
whole range of inter- and intramolecular bonding interactions,
including hydrogen bonding, donor-acceptor interactions (72,74)
dispersion forces, dipole-dipole interactions (75) etc., are
involved in a complex heterogeneous substance like coal. The
important question is the relative importance of the cross-links,
entanglements, trapped molecules, hydrogen bonding, etc., in
determining coal properties and behaviour. The main problems are:
 83

1. To establish a theory for understanding the polymeric

properties of coal.
2. To estimate accurately the degree of cross-linking and
the molecular weight between cross-links.
3. To distinguish between the effects of entanglements and
cross-links and understand the bonding interactions between
macromolecules.

The Flory theory (73) has been used extensively with polymers
and attempts have been made (69,70) to use this theory for the study
of the macromolecular structure of coal. This model was developed
(73) for a non porous polymer with a small degree of cross-linking
of a single type between one type of monomers. No account is taken
of entanglements of macromolecules or weaker interactions such as,
hydrogen bonding, donor-acceptor interactions etc. and these being
probably the weakest may break first in solvent swelling studies. A
coal network model (69) illustrating cross-links and entanglements
is shown in Figure 11.

A macromolecular network will absorb solvents and swell without

dissolving to an extent controlled by the size of the interactions
between the solvent and network and also the cross-link density. The
average molecular weight between cross-links can be calculated from
measurements of swelling and the solvent-molecule interaction
parameter using the Flory-Rehner equation.

Where Me is the number average molecular weight between

cross-links, P2 is the original density of the network,
VI the molar volume of the solvent, V2 the volume fraction
of polymer at equilibrium, and X the Flory polymer-solvent
interaction parameter. The equation does not include the
controversial logarithmic term in the elastic free energy.

The equation assumes the following

1. The origin of the elastic restoring force on the network
is entropy
2. Individual network chains deform in a related manner
3. The solvation process can be described by a modified
regular solution theory.

The number average molecular weight between cross-links, Mc

depends on the Flory-Huggins interaction parameter X and this is
shown in Figure 12. However the problem is to estimate a value for X
which is both concentration and temperature dependent. This is
usually achieved (69,70) by measurements of the osmotic pressure
of a solution of the polymer but these are not possible for coal
since one cannot rely on only the soluble part of coal being the
same as the coal itself. There are many problems with estimating
84

D ,Aromatic cluster
- Bridge bonds
(-O-,-(CH 2 )-n, -S-S- etc)

Fig 11 COAL NETWORK MODEL ILLUSTRATING (1) CLUSTER

AS CROSS - LINK (2) TETRA FUNCTIONAL (IDEAL)
CROSS-LINK BASED ON CARBON ATOM
(3) ENTANGLEMENT BETWEEN NETWORKS
 85

3OO0~------------------------------------,

1000

OL-------~---------L--------~------~
0-5 0-6 0-7 o-a 0-9
X
Fig 12_ VARIATION OF THE AVERAGE MOLECULAR WEIGHT
BETWEEN CROSS-LINKS Me WITH FLORY-HUGGINS
INTERACTION PARAMETER X AFTER LUCHT AND
AND PEPPAS (79)
86

X from regular solution theory and it is difficult to measure X

experimentally. The need to determine X accurately is shown in
Figure 12 which shows (79) the variation in Me with X. At
present values of Me are best considered as a relative
measurement.

Problems also exist with experimental determinations of the

solvent induced swelling of coals. The experiments have been
carried out on both whole and pre-extracted coals (76,77). The
latter have been investigated to avoid complications of extractable
material but the extraction process will change the coal structure.
Further work is needed on the role of extractable material in
solvent induced swelling. In view of the heterogeneous nature of
coal it would also be advisable to study individual macerals.

Recent studies (74) of the swelling of coal by a range of twenty one

solvents has shown that the swelling correlated with solvent
electron-donor number whereas there was little correlation with
solubility parameters. The solvents with donor number <16 only
induced slight swelling in the coal while solvent swelling rapidly
increased for solvents with a donor number >16. It was proposed
that the mechanism involved solvent induced breakage of electron
donor-acceptor interactions between coal macromolecules. These
interactions could also involve hydrogen bonds, dipole-dipole
interactions, dispersion forces etc. (75).

The molecular weight distribution of coal extracts may also

provide information (78) on the nature of the coalification process.
If a condensation polymerisation from a mixture of monomers in gel
formation occurs during coalification then as the molec~lar weight
of the components increase their concentration should decrease. The
opposite behaviour is observed (78) for Illinois 6 and Bruceton
coals indicating a more complex process. Further work is needed on a
wider range of coals to confirm this work.

Larsen has pointed out (69) that cross-link density showed a

minimum at 86(%dmmf) carbon and this should correlate with
reactivity but the existence of the minimum has been questioned.
The reactions of a coal with fewer cross-links would involve
breakage of fewer bonds and it should be more reactive. It was
pointed out that many properties show a peak at 86(%dmmf) carbon.
Some of the parameters which show this phenomenon are listed
below:

1. Pyridine extraction yield.

2. Yield of quinoline or pyridine soluble material produced on
heating coals with quinoline.
3. Short time conversion of coals in tetralin solution.
4. Conversion of coals by heating in pyrene.
5. Reactivity (tar yield).
 87

Various physical properties also show the same phenomenon:

1. Fluid ity (Gieseler).

2. Swell ing •

In compar ison dens ity and poros ity show (1) a minimun for a
coal with a carbon content of 86(% dmmf).

These data illustrate the uniqueness in terms of coal

properties at 86(%dmmf) carbon content but it has not been proved
beyond doubt that it is due to a minimun in cross-linked density.
The evidence in support of the minimum in cross-linked density comes
from solvent swelling and X-ray crystallographic studies. It is
noteworthy that there is probably a change over in that the extent
of oxygen cross-links will decrease and carbon cross-links increase
with increase in rank.

The involvement of covalent cross-links and entanglement of

macromolecules either of which would give rise to a three
dimensional network, is still the subject of considerable
discussion. It has been suggested (69) that it would be difficult to
envisage entangled molecules having the mechanical and elastic
propert ies of coal. It is clear that the molecules found in coal are
very large but the way they are associated remains the subject of
discuss ion. Bear ing in mind the wide var iety of coals found it is
possible that both are involved in varying degrees and this may be
partly responsible for some coal properties. It is becoming
increasingly accepted that vitrinite consists to a major extent of a
cross-linked macromolecular network (68). The inclusion of the
mobile phase, hydrogen bonds, donor-acceptor (72) and other weak
interactions (75) is also significant. The important question is -
can techniques be developed which are capable of distinguishing
between different types of forces and cross-links which are involved
in coal structure?

CONCLUSIONS

The structure of coal cannot be defined to the same extent as a

pure crystalline solid because of its heterogeneous nature. However
it does contain definable units and features which repeat throughout
the coal and these have distinct physical and chemical properties
which can be defined to varying degrees. A knowledge of these
structural features is important for understanding coal properties
and behaviour during gasification, carbonisation, and combustion.
Direct evidence for the structure of coal has been obtained from a
wide variety of methods.
88

Primary Structure
Spectroscopic methods - Nuclear magnetic resonance
Photoelectron spectroscopy
Mass spectrometry (field ionization)
FTIR
Chemical methods Solvent extraction
Oxidation
Dehydrogenat ion
Depolymerisation
Derivatisation (silylation, acetylation
etc.)

Secondary and Tertiary Structure

Optical microscopy
Electron microscopy
X-ray diffraction

In addition information has been obtained from indirect

methods, for example, solvent swelling, and studies of physical
proper ties.

Qualitative models of the distribution of various bond types in

coal have been produced. These models are based on chemical,
analytical and spectroscopic data which provide statistical data on
aromatic ring condensations, aliphatics, functional groups etc. and
should not be taken as representing precise structures found in
coal. They represent possible forms of a statistical average
s.tructure. The models are useful starting points for the discussion
of chemical processes.

It is generally agreed that coal has a macromolecular

structure. This may involve cross-links and entanglements of
heterogeneous macromolecules. The type of cross-links may vary
significantly and could involve oxygen, carbon and various other
types of cross-links. The application of techniques used in polymer
science to the study of coal is in its early stages and considerable
further work is needed on both the theoretical and experimental
aspects.

A considerable amount of factual information has been generated

on various aspects of coal structure. However, a great deal of work
is not directly comparable and this in itself is a problem in
deriving useful information and providing understanding of coal
structure.

'Science is built up with facts, as a house with stones. But a

collection of facts is no more a science than a heap of stones
is a house.'
Poincare
 89

There have been a considerable number of hypotheses on coal

structure but there is a lack of vital information in many areas.
It is fair to say that a start has been made on rudimentary models
of coal structure and these will form a basis for future
developments. The models are illustrative rather than definitive and
a considerable amount of further work is needed to understand the
various diverse aspects of coal structure, in particular in the
nature of the macromolecular structure of coal.

ACKNOWLEDGEMENTS

The author would like to thank British Gas for permission to

publish the paper. Thanks are also due to J.E. Williamson for the
NMR data, R.B. Sinha for the electron micrographs and A.D. Carr and
C.A. Johnson for the optical micrographs.
 90

REFERENCES

1. van Krevelen, D.W. 'Coal Typology-Chemistry-Physics -

Constitution'. Elsevier, Amsterdam, 1961.
2. Neavel, R.C. in 'Coal Structure' Eds. (Gorbaty, M.L. and Ouchi,
K.), Advances in Chemistry Series (192), ACS, Washington, 1981,
pl.
3. Myers, R.A. in 'Coal Structure' (Ed. Myers, R.A.) Academic
Press, New York 1982, 1
4. Grimes, W.R. in 'Coal Science'. Eds. (Gorbaty, M.L., Larsen
J.W. and Wender, I.), Academic Press, New York 1982, 1, 28.
5. Ting, F.C. in 'Coal Structure' (Ed. Myers, R.A.) Academic
Press, New York, 1982 p.8.
6. Mackowsky, M-Th, Microsc Acta. 77, 1975, 114.
7. Renton, J.J. in 'Coal Structure' (Ed. Myers, R.A.) Academic
Press, New York 1982, p283.
8. Jenkins, R.G. and P.L. Walker Jr. in 'Analytical Methods for
Coal and Coal Products, (Ed. Karr, C. Jr.) Academic Press, New
York, 1978, Vol.2, p265.
9. Wong, J., C.L. Spiro, D.H. Maylotte, F.W. Lytle and R.B.
Greegor. Proceedings of the International Coal Science
Conference, Dusseldorf, 1981, p362.
10. Hirsch, P.B. Proc. Roy, Soc. A226(1954), 143 and in "Res.ConLon
Science in the Use of Coal", Institute of Fuel, London, 1958,
A29-A33.
11. Pitt, G.J. in 'Coal and Modern Coal Processing. An Introduction.'
(Eds. Pitt, G.J. and Millward, G.R.) Academic Press, London, 1979
p.27.
12. Mahajan, O.P. and P.L. Walker Jr. in 'Analytical Methods for
Coal and Coal Products (Ed. Karr, C. Jr.) Academic Press, 1978,
Vol. 1 p125
13. Sing, K.W.S., Pure Appl. Chem. 54,(1982), 2201.
14. Hays, D., J.W. Patrick and A. Walker, Fuel 55, 1976, 297.
15. Davidson, R.M. in 'Coal Science' (Eds. Gorbaty. M.L. Larsen,
J.W. and Wender, I.) Academic Press, New York, 1982,1,83.
16. Grainger, L. and J. Gibson, 'Coal Utilisation: Technology,
Economics and Policy', Graham and Trotman, London, 1981 p.28.
17. Blom, L., L. Edelhausen and D.W. van Krevelen, Fuel,36,
1957,l35.
18. Cronauer, D.C. and R.G. Ruberto, EPRI Report AF-913, Palo Alto,
California 1979.
19. Liotta, R. and G. Brons, J. Am. Chem. Soc. , 103, 1981, 1735.
20. Attar, A. and W.H. Corcoran. Ind. Eng. Chem. Prod. Res. Dev.
16(2), 1977 , 168.
21. Attar, A. and G.G. Hendrickson in 'Coal Structure' (Ed. Myers,
R.A.). Academic Press, New York, 1982, p132.
22. Snape, C.E. , Ph.D. Thesis, University of Leeds, 1982.
23. Heredy, L.A. , A.E. Kostyo and M.B. Neuworth, Fuel, 44, 1965,
125.
24. Heredy, L.A. and M.B. Neuworth, Fuel 41, 1962, 221.
 91

25. Ouchi, K., K. Imuta and Y. Yamashita, Fuel, 44, 1965, 29 and
205.
26. Ouchi, K., Fuel 46, 1967, 319.
27. Heredy, L.A. in 'Coal Structure' Advances in Chemistry 192 (Ed.
Larsen J.W.) ACS, Washington 1981, p.179.
28. Hooper, R.J. and D.G. Evans, ACS Div. of Fuel Chem., 24, 1979,
131.
29. Sternberg, H.W., C.L. Delle Donne, P. Pantages, E.C. Moroni and
R.E. Markby, Fuel, 50. 1971, 432.
30. Sternberg, H.W. and C.L. Delle Donne, Fuel, 53, 1974, 172.
31. Ignasiak, B.S. and M. Gawlak, Fuel, 56, 1977, 216.
32. Wachowska, H., Fuel, 58, 1979, 99.
33. Stock, L.M. in 'Coal Structure' Eds. (Gorbaty, M.L., Larsen,
J.W. and Wender, I.), Academic Press, New York, 1982, Vol.l,
p.161.
34. Ignasiak, B.S., D. Carson and M. Gawlak, Fuel 58, 1979, 833.
35. Hayatsu, R., R.E. Winans, R.G. Scott and M.H. Studier, Fuel,
57, 1978, 541.
36. Hayatsu, R., R.E. Winans, R.G. Scott and R.L. McBeth, Fuel, 60,
1981, 158.
37. Hayatsu, R., R.G. Scott, R.L. McBeth, L.J. Basile, E.H.
Appelman and R.E. Winans. Proceedings Int. Coal Science Conf.,
Dusseldorf, 1981, 139.
38. Winans, R.E., R. Hayatsu, R.L. McBeth, R.G. Scott, L.G. Moore
and M.H. Studier in 'Coal Structure' Advances in Chemistry 192.
(Ed. Larsen, J.W. ) ACS Washington 1980, p.161.
39. Deno, N.C., B.A. Greigger and J.G. Stroud, Fuel 57, 1978, 455.
40. Deno, N.C., B.A. Greigger, A.D. Jones, W.D. Ratkitsky, D.D.
Whitehurst and T.O. Mitchell, Fuel, 59, 1980, 701.
41. Chackrabatty, S.K. and H.O. Kretschmer, Fuel 53, 1974, 132.
42. Mayo, F.R., Fuel, 54, 1975, 273.
43. Mayo, F.R. and N.A. Kirschen, Fuel, 58, 1979, 698.
44. Aczel, T., M.L. Gorbaty, P.S. Maa and R.H. Schlosberg, Fuel 58,
(1979), 228.
45. Ouchi, K., Y. Hirano, M. Makabe and H. Itoh, Fuel 59, 1980,
751.
46. Ouchi, K. and M. Makabe, Fuel, 58, 1979, 43-47.
47. Friedman, S., M.L. Kaufman, W.A. Steiner and I.Wender, Fuel,
40, 1961, 33.
48. Reggel, L., I. Wender and R. Raymond, Fuel 43, 1964, 229.
49. Retcofsky, H.L. in 'Coal Science' (Eds. Gorbaty, M.L., Larsen,
J.W. and Wender, I.) Academic Press, New York, 1982, Vol.l.
p.43.
50. Gerstein, B.C., P. Dubois Murphy and L.M. Ryan in 'Coal
Structure' (Ed. Myers, R.A.) Academic Press, New York, 1982,
p.87.
51. Gerstein, B.C, L.M. Ryan and P.D. Murphy in 'Coal Structure'
(Eds. Gorbaty, M.L. and Ouchi, K.) Advances in Chemistry
Series 192, ACS, Washington, 1981, p.13.
92

52. Pugmire, R.J., W.R. Woolfenden, C.L. Mayne, J. Karas and D.M
Grant, in 'Chemistry and Characterisation of Coal Macerals',
Eds. (Winans, R.E. and Crelling, J.C.), ACS Symposium Series 252,
ACS Washington 1984, p.79.
53. Hagaman, E.W. and M.C. Woody, Fuel 61, 1982, 53.
54. Kuehn, D.W., A. Davis and P.C. Painter in 'Chemistry and
Characterisation of Coal Macerals', Eds. (Winans, R.E. and
Crelling, J.C.), ACS Symposium Series 252, ACS, Washington 1984,
p.99.
55. Painter, P,C" R.W. Snyder, M. Starsinic, M.M. Coleman, D.W.
Kuehn and A. Davis, Applied Spectroscopy, 35, 1981, 475.
56. Bartle, K.D., M.J. Mulligan, N. Taylor, T.G. Hartin and C.E.
Snape, Fuel, 63, 1984, 1556.
57. Bodzek, D. and A. Marzec, Fuel, 60, 1981, 47.
58. Given, P.R., Fuel, 39, 1960, 147.
59. Hill, G.R. and L.B. Lyon, Ind. Eng. Chem., 54, 1962,37.
60. Ladner, W.R. in Gibson, J., J.Inst.Fuel, 51,1978,67
61. Wiser, W.R., 'Proceedings of the Electric Power Research
Institute Conference on Coal Catalysis', Palo Alto, California
1973, p.3.
62. Solomon, P.R. in 'New Advances in Coal Chemistry' (Eds.
Blaustein B.D., Bockrath B.C. and Freidman S.), ACS Symposium
Series 169, 1981, p.61.
63. Spiro, C.L. and P.G. Kosky, Fuel, 61, 1982, 1080.
64. Shinn, J.R., Fuel, 63, 1984, 1187.
65. Aczel, T., M.L. Gorbaty, P.S. Maa and R.H. Schlosberg, Prepr.
Coal Chem. Workshop, 1976, CONF-760885, paper 12, p.165.
66. Chakrabartty, S.K. and N. Berkowitz, Fuel, 53, 1974, 240.
67. Marzec, A., M. Juzwa, K. Betlej and M. Sobkowiak. Fuel Process.
Technol. 2,1979,35.
68. Given, P.H., in 'Coal Science', (Eds., Gorbaty, M.L., Larsen, J.W.
and Wender, I.), Academic Press, New York, 3,1984,63.
69. Green, T., J. Kovac, D. Brenner and J.W. Larsen, in 'Coal
Structure', (Ed. Myers, R.A.) Academic Press, 1982, p.199.
70. Larsen, J.W. and J. Kovac in 'organic Chemistry of Coal', (Ed.
Larsen J.W.) ACS Symposium Series 71, ACS, Washington, 1978,
36.
71. Calkins, W., Am. Chem. Div. Fuel. Chem. Prepr. 28 (5), 1983,
85.
72. Marzec, A. and W. Kisielow, Fuel, 62, 1983, 977.
73. Flory, P.J., 'Principles of Polymer Chemistry', Cornell Univ.
Press. Ithaca, New York, 1953.
74. Szeliga, J. and A Marzec, Fuel, 62, 1983, 1229.
75. Hall, P.J. and H. Marsh, Fuel, (In Press).
76. Larsen, J.W., and D. Lee, Fuel, 64,1985,981.
77. Weinberg, V.L. and T.F. Yen, Fuel 59,1980,287.
78. Larsen, J.W., M. Mohammadi, I. Yiginsu and J. Kovac, Geochimica
and Cosmochimica Acta, 48, 1984, 135.
79. Lucht,L.M. and N.A. Peppas. Conf. Phys. Chem. Coals 1980 in
reference 69.
 93

PHYSICO-CHEMICAL AND STRUCTURAL CHARACTERIZATION OF CARBONS

A. Marchand

Universite de Bordeaux I
Centre de Recherche Paul Pascal (CNRS)
Domaine universitaire, 33405, Talence Cedex, France

1. FROM COALS TO GRAPHITE

1.1. Definitions: the World of Carbons

Any coal or organic matter heated to a sufficient temperature in

an inert atmosphere will progressively lose its atoms other than
carbon and will become an artificial carbon material or "carbon". A
similar process takes place to some extent in nature, where
substances of biological origin are buried and may be transformed
into coals of progressively higher carbon content under the action
of temperature and also pressure and chemical agents. Coals and
carbons can then be considered as members of a same family, which
generate each other, the ultimate offspring being the graphite
crystal.

It is however a very numerous family, and its genealogical tree

is extremely complex, with many different places of birth, a lot of
dead branches, and even some mutants •••. A wide variety of physico-
chemical methods are necessary to characterize the chemical
constitution, structure, and properties of these solids, and to
monitor their evolution along the path which may eventually lead
them to more or less perfect graphite.

It is not even easy to give a precise definition of who belongs

to the family. The term "carbon" usually refers to materials which
are at least constituted of, let us say, 95 % atoms of carbon. But
this dividing line is obviously very arbitrary and, although it keeps
some coals in the carbon family, it leaves most of them outside.
Since we are going to discuss the characterization of coals AND
 94

carbons, this lack of an exact definition will not create any real
difficulty.

Let us however note that diamonds are not generally considered

as "carbons": carbon atoms in a carbon material are assumed to be
predominantly linked by hybridized sPz bonds, in spite of possible
exceptions in carbon films.

But it is necessary to start by reviewing briefly the main

branches of the family.

First there are different starting materials:

1) LIQUID AND SOLID PRECURSORS (coals, natural biological

materials or hydrocarbons, industrial organic substances, textiles)
can be submitted to heat-treatments up to 1000-1100°C in reducing,
oxidizing, or neutral environments. These CARBONIZATION processes
yield COKES, GLASS-LIKE CARBONS, or FIBRES. During the manufacture
of cokes, the material may remain in the solid state throughout the
whole treatment, and it may also undergo a transitory melting in the
vicinity of 450-500°C: the resulting intermediate liquid is a
MESOPHASE, with interesting anisotropic properties.

At higher heat-treatment temperatures (H.T.T.) the PRE-

GRAPHITIZATION (up to 1600°C) and GRAPHITIZATION (up to 3000°C)
processes yield PREGRAPHITIC CARBONS or ARTIFICIAL GRAPHITE.

2) GASEOUS PRECURSORS (evaporated carbon or gaseous hydrocarbons)

can be pyrolyzed at various temperatures ranging from near room-
temperature up to more than 1600°C and a very wide variety of
PYROLYTIC CARBONS and CARBON BLACKS as well as FIBRES are obtained,
according to the precise operating conditions (temperature, pressure,
gas composition, gas flow, gradients, etc .•. ). In the case of
pyrolytic carbons, the process is called "chemical vapor deposition"
(C.V.D.): it may be performed at relatively lower temperatures with
the help of an electric discharge ("plasma-assisted CVD").

All these materials can of course be heat-treated at temperatures

higher than their "temperature of deposition" (T D) through a
subsequent pre-graphitization or graphitization.

But the ultimate stage is not always graphite. Those carbon

materials which can be eventually purified and organized in crystals
of three-dimensionnally ordered graphite are labeled "GRAPHITIZING
CARBONS" or "SOFT CARBONS". The other carbons also undergo a
progressive purification and ordering under the effect of the
graphitization treatment, but, even at the highest H.T.T., they
never reach the three-dimensional graphite stage: they are called
"NON-GRAPHITIZING" or "HARD CARBONS".
 95

It is then clear that, in spite of their diversity, most or all

carbons, (if not coals) can be assigned a position on a "graphitiz~
tion scale" which indicates how far they are located on the
evolutionary path which goes from organic or biological molecules to
graphite. Such information can of course be obtained from many
physico-chemical measurements. But this "degree of graphitization" is
not the only "index" which must be used in order to characterize a
carbon material or coal and to determine its location among the
multiplicity of branches and species which constitute the "tree".

1.2. The Building of Graphitic Structures

We will assume that the hexagonal structure of graphite, with

its characteristic ABABA ..• stacking, is known. After this first
determination by Bernal in 1924, and particularly after 1941, the
structures of various carbon materials and their evolution during
the carbonization-graphitization process have been investigated by a
multitude of workers. But only recently did we acquire a reliable
and general (although still quite incomplete) model of the way
graphitic structures are built. We will summarize it briefly, without
giving any description of the many studies which led to our present
understanding.

The building blocks of all carbon materials are small groups of

condensed aromatic rings.

BELOW 1000-1100°C (CARBONIZATION RANGE) the carbon materials (and

coals) can still be considered as molecular solids, with a large
proportion of heteroatoms and non-aromatic C-C bonds. Their structure
is characterized (2) by "structural units" (S.U.), which are
molecules constituted of less than 10-12 condensed aromatic rings:
their size is in the order of 8-10 Angstrom units and they are
sfacked by two or three. There is no general organization of these
units, except a possible preferred orientation, usually local but
sometimes extending more or less through the whole solid. The effect
of increased H.T.T. is mainly a progressive "purification" due to
the loss of heteroatoms.

Carbons manufactured from a gas phase in this temperature range

have the same characteristics.

At increasing heat-treatment temperatures the carbon materials

lose their heteroatoms.

IN THE 1000-1600°C RANGE (PRE-GRAPHITIZATION) there are very few

heteroatoms left and most of them are probably bound laterally to
the aromatic molecules. The S.U. are thus able to orient themselves
parallel to one another and to build columnar structures like stacks
of plates, but these stacks are bent and very irregular because they
 96

are constituted of units of slightly different shapes and diameters.

The stack height is also limited since there are still some
disoriented units, but the number of aromatic layers associated in
such parallel stacking increases with H.T.T. as the proportion of
disoriented units decreases. This is shown on figure lB, which
illustrates the graphitization of carbon films (3).

~RAPHITIZATION OCCURS BETWEEN 1600 AND 3000°C IN TWO STEPS: a two-

dimensional growth from 1600°C to 2000-2100°C, and a three-
dimensional organization from 2100°C upwards. Hard carbons usually
deviate from the graphitizing behavior at this stage: they do not
acquire the complete three-dimensional lattice of graphite even at
the highest treatment temperatures.

The two-dimensional growth of the aromatic systems results from

the gradual removal of the various defects (heteroatoms, SP3 bonds,
twists and dislocations) located at the periphery of the structural
units. This allows the coalescence of the columnar associations,
which then constitute very large "wrinkled" parallel sheets of
aromatic carbons (fig. lC). The continuing elimination of defects
progressively rubs out the S.U. boundaries, and very extended rigid
two-dimensional parallel graphitic layers are obtained at 2000-2100°C
(fig. lD).

It has been suggested to name "graphenes" these two-dimensional

structures in order to avoid confusion with the "true" graphite
structures, which is three-dimensional. "Graphenes" have no three-
dimensional organization. In pregraphitic carbons their parallelism
may be nearly perfect, but they also are translated or rotated
arbitrarily relative to each other, so that there is no fixed
relationship between the positions of two carbon atoms belonging to
adjacent graphenes: this kind of structure is called "turbostratic".

Pyrolytic carbons deposited near 2100°C also show this two-

dimensional structure built from turbostratically stacked graphenes.

The final step consists in ordering this turbostratic stacking

into the regular ABABAB •.. sequence of hexagonal graphite.
There are hardly any defects left in the solid to prevent the
necessary shifts of the graphenes relative to one another, PROVIDED
ALL GRAPHENES STACKS HAVE THE SAME ORIENTATION: non-graphitizing
carbons usually do not meet this condition because of either a
remaining microporosity (where the graphitic layers are aligned
parallel to the pore walls) or too small grain sizes (e.g. carbon
blacks).

Although temperature is certainly the most important factor in

these transformations, it is not alone. The effect or high pressures
has been demonstrated, not only in the natural processes (deep burial
 97

==
:::: ~
~
§ === =~ @

II
~
~

rigid
~ plane
layers
El0~
~8E3 ® layers
~~~
0 8
@

s.u.
Fig. 1: Schematic building of graphitic structures (3)

of organic matter which results in coal formation) but also in the

laboratory (pressure induced graphitization). The chemical
characteristics of the environment (reducing or oxidizing) may have
a long term effect on the carbon, particularly during the "mesophase"
formation: there is undisputable evidence that the action of oxygen
at this stage can prevent the graphitizability of the resulting coke
and that hydrogenation has the opposite effect. Most of all, the
heat-treatment time is very important: a number of studies have shown
that increasing H.T.T. from Tl to T z is equivalent to multiplying
the treatment time t by a constant factor k(T1,T z )' so that the
curves representing at Tl and T z the variations of a particular
property P as a function of the logarithm of time can be superimposed
by a translation log k along the time axis (1). A "single curve" IS
then obtained by plotting P as a function of a "composite time
variable" D which incorporates both H. T. T. and t. Figure 2 shows such
a single curve, where the Hall coefficient A (see paragraph 4.3) vs.
D behavior of a pitch coke is presented: it incorporates experimental
results from various H.T.T. in the 1420-2850°C range.

We will not examine further this question: for our purpose it

will be usually sufficient to characterize approximately each step
in the carbonization-graphitization by a specific temperature range
(treatment times are assumed to be in the order of one hour).
 98

.' ,:' :,. ':. ;·,::1-

A . 10· m' ~-' I

'OI--~-i-,L..~",g"~";I-.-:-~ E~ - i 1.._

5-: 1:~1-- ...... " 1 : ' 0--1

oll---.---II---+-----+-+:-.-'-+---t-.. - -----+--+---t----- - .......-.... ---1-
. '" '../i ...... 1

SL-~1dO-~1JOr-~~b).r-~10h--1~O~~~d.-~1~C--~~r--1..0~~1~0~'D~1~0w--.1~0'~'-~D~~;i~

Fig. 2: Pitch coke: the single curve A f(log D) obtained

from isothermal curves (1)

2. STRUCTURAL CHARACTERIZATION:INTERACTION WITH X RAYS AND ELECTRONS

This schematic description of the carbonization-graphitization

processes summarizes information which mostly results from studies
performed with the usual tool of the crystallographer, namely X-ray
diffraction. But an other technique of structural characterization
has recently demonstrated its efficiency in the carbon field: the
electron microscope. A large part of our present understanding of
carbonization comes from electron microdiffraction studies. We will
show first what kind of information can be obtained from X-rays, and
the following section will deal with electrons.

2.1. X-Ray Diffraction

Most carbon materials are constituted of powders or poorly

oriented microcrystals. There are a few exceptions, including
graphite and very well oriented highly graphitized samples, where
single crystal techniques can be used. In most cases only Debye-
Scherrer diagrams are available.

The powder diagrams of coals or low temperature chars usually

show more or less diffuse rings corresponding approximately to the
002, and possibly the 10, and 11 beams of carbon. With more strongly
carbonized or graphitized carbons, the rings become sharper and new
reflexions appear. The best way to analyze this kind of data is to
record the intensity distribution as a function of the diffusion
angle (fig. 3). This technique does not require more than the
equipment usually available in a solid state Physics or Chemistry
laboratory. It is possible to deal separately with the DOt reflexions
and with the hk diffraction "bands".
 99

2.1.1. The 00£ reflexions: with a graphite crystal the 002,004,006,

.•• etc reflexions can be observed (fig. 3): the d oo2 distance is
that between two adjacent carbon planes (0.3354 nm). The 002
reflexion is the only one which is nearly always present in the
powder diagram of a carbonaceous substance. In coals and low
temperature chars it is usually very broad and diffuse and corresponffi
to the average distance between adjacent aromatic molecules in a
structural unit or in a stacking of such units. In carbons heat-
treated above 1600°C it measures the average distance between
adjacent carbon sheets and its decreasing from about 0.344 nm to the
graphite value can be used as a "graphitization index" of the carbon.

Analysis of the profile of 00£ lines can yield an evaluation of

the average thickness of the stacks of parallel graphitic layers,
usually designated by Lc' This can be done by computing the Fourier
transform of two successive 00£ line profiles (e.g. 002 and 004)(4).
If this is not possible, the 002 experimental line shape may be
compared to theoretical profiles: Lc is easily determined by trials
and errors (5). It may even be approximately obtained by measuring
only the width 6 at half-maximum of the 002 line: Lc is inversely
proportional to 6 cos 8 (where 8 is the Bragg diffraction angle).

2.1.2. Band profile analysis: in many powder diagrams of carbons,

in addition to 00£ lines, it is also possible to detect some "bands"

002

10

looo'e

j
11
oo~
20

101
oo~

002
11.0 112
10.0 103
,r.,IIit.

10.2

Fig. 3: Typical X-ray diffractogram of carbons (3)

 100

corresponding to hk reflexions of the graphitic structure (e.g. the

10 band). These band shapes may also be compared to theoretical
profiles (6) and an average "crystallite dimension", usually called
La, is very often determined in this way. But the physical
significance of the La parameter is theoretically valid only in the
case of real graphite microcrystallites: it is only an APPARENT
crystallite dimension in the case of disordered carbons, because the
various defects which are present in the layers cause a broadening
of the hk bands. Moreover the "wrinkles" of the layers may break the
periodicity of the two-dimensional lattice, and so do the undulations
of the not quite planar layers: then La is only the "dimension of
coherence" of the aromatic layers. It must be used with caution.

2.1.3. Band modulations: before the onset of a true three-

dimensional organization, the carbon structure is essentially made
up of two-dimensional units or layers. This is why only hk "bands"
and no hk-R. "lines" are visible in the powder diagram (fig. 3). But
as graphitization proceeds to its last stage (H.T.T. > 2100°C) the
shape of these bands change with the appearance of "modulations"
which are an indication of the progressive ordering of the graphenes
into the regular ABABA ... sequence. Figure 4 is an illustration of
this phenomenon, where a shoulder first grows on the 10 band tail,
and then develops at higher H.T.T. into a new line, so that the
former 10 band is replaced by two lines (100 and 101) which are
evidence of the existence of a true three-dimensional graphite
structure.

Fig. 4: Progressive modulation of the (10) band ofpyrolytic

carbons deposited at 2100°C - Onset of the three-
dimensional periodicity at H.T.T. > 2600°C (20)
 101

2.1.4. Radial Distribution of interatomic distances: the

statistical distribution of C atoms can be determined from the
Fourier transform of the intensity of the complete diffraction
spectrum. This method is the only one which can be used for very
poorly crystallized substances and a number of crystallographers did
try it to obtain the radial density distribution curves of many kinds
of carbon materials: it yields the average environment of each
individual atom. But this proved rather disappointing since only the
expected distances for aromatic structures (0.142, 0.246, and
0.284 nm) were found. Specific point defects however can be
identified in some cases: in pitch cokes heat-treated around 2000°C
for instance, interstitial carbon atoms can be detected in positions
above or below the centres of graphitic hexagons. The corresponding
C-C distance (0.190 nm) disappears from the distribution curves of
these cokes at H.T.T. > 2500°C (7).

2.1.5. Anisotropic Carbons: the structural units (S.U.) or carbon

layers are not always randomly oriented. In many cases they tend to
lie parallel to a common direction (e.g. the axis of a fibre) or
plane (e.g. the plane of deposition of pyrolytic carbons deposited
above 2000°C). This preferred orientation is easily detected from
the X-ray diffractogram, since in this case the OO~ reflexions do not
generate complete rings in the Debye-Scherrer diagram: the diffracted
intensity is concentrated along the ring diameter perpendicular to
the preferred direction of the aromatic planes. The stronger the
orientation, the sharper is the corresponding intensity maximum, and
the degree of preferred orientation can be estimated from an analysis
of the line profile.

2.2. Electron Diffraction

During recent years A. Oberlin and her coworkers demonstrated

that a wealth of information can be extracted from electron
diffraction (2,3). There are several good reasons behind this new
development.

The electronic intensity scattered by carbon atoms is nearly 10 8

times the scattered X-ray intensity, so that less than 0.001 ~m3 of
matter (a piece of carbon a few tenths of micron wide) is sufficient
to record in a few seconds a satisfactory diffraction picture.

2.2.1. Selected Area Diffraction: it means that it is possible to

select a very small fraction of a sample to study its structure:
such a "microscopic" fragment may be for instance a nearly single
microcrystal, or it may be a group of aromatic molecules with a
common orientation. In both cases the diffraction pattern will be
specific and easy to identify. Instead of observing an average
diffractogram, resulting from a very large number of more or less
randomly oriented elements, as is the case with X-rays, this
 102

technique of "selected area (electron) diffraction" (S .A.D.) allows

a very detailed exploration of the different domains constituting a
carbon sample.

2.2.2. Dark Field Images: a second advantage of electron over

X-ray diffraction is the fact that an electron beam can be deviated
by a lens, while an X-ray beam cannot. Interesting new possibilities
are then open with the electron microscope. Figure 5 shows schema-
tically various electron beams scattered by the objet in different
directions through the microscope. The normal image of the object is
displayed in the "Gaussian" plane G, but its diffraction pattern is
formed in the back focal plane A of the objective lens: there
converge towards the same point all the beams emitted by all the
points of the object in a given direction, so that each point of
plane A corresponds to a particular hkt reflexion of the object. If
a screen is placed in A, with one aperture just wide enough to allow
a single reflected hkt beam to pass through, only the regions of the
object which have emitted this hkt beam will give a lighted image in
G. They will appear BRIGHT on a DARK FIELD. It is thus possible to
obtain specific images of those regions of the object which have
lattice planes with a given orientation relative to the incident
electron beam.

Let us suppose the object is a powdered sample: the Debye-

Scherrer (D-S) diffraction pattern will be located in plane A. The
position of the small aperture in A can be selected at any convenient
point of the D-S diagram (fig. 6) and only the corresponding
diffracted beams will pass through it to give in G an image of the
sample (fig. 5). If the aperture is located on the 002 ring for
instance, there will only appear in G a bright image (on a dark field)
of all the regions in the sample where the incident electron beam
can be reflected towards the aperture by the graphitic or aromatic

Fig. 5: Formation of a dark field image (3)

 103

Fig. 6: Various positions of the aperture in plane A (2)

planes at the 002 Bragg angle. Such images are shown on figures 7a
and 7b: both are dark field 002 images from the same char sample,
but the aperture position in 7b is rotated 90° (around the electron
microscope axis) from the position in 7a. It is obvious that the
dark regions of 7b are the bright regions of 7a, and vice versa: this
shows that there is an average preferred orientation of the
"structural units" (aromatic molecules) in small domains scattered
throughout the sample ("clouds" of bright dots), and that the
preferred orientation of the bright domains in 7b is rotated 90° from
that of the bright domains in 7a.

2.2.3. Lattice Fringes: if the aperture in plane A is larger, the

beam transmitted along the incident beam direction can be let through
to interfere in plane G with the selected hkt diffracted beam
(fig. 8): diffracted fringes with the dhkt period appear on a bright
field at the corresponding location in the image of the sample. If
the 002 beam has been selected in A, there will be in G an image
which LOOKS LIKE a projection of the family of aromatic or graphitic
layers seen "edge on", BUT IS ACTUALLY A SYSTEM OF FRINGES. This
technique allows the observation of "lattice fringes" for any family
of lattice planes which happens to lie at the Bragg angle to the
incident beam, and the fringes will reflect (provided suitable
precautions are taken) the dimensions and orientations of the
original family of planes.

Figure 9 shows an exemple of 002 lattice fringes for carbons of

increasing H.T.T.: the evolution from disordered (a) to stacked (b)
S.U., then to graphenes Cd) and to a graphite crystal (e) is
104

(a) (b)
Fig. 7: Dark field 002 images of a char sample (2)-
the position of aperture in (b) is rotated
90° relative to position in (a)

Fig. 8: Formation of lattice fringes (bright field) -

interference of direct beam (0) with selected
hk£. beam (p,p')
 105

Fig. 9: Lattice fringes (002) from a carbo~ film heat-

treated at 800°C (a), 1510°C (b), 1820°C (c),
1975°C (d) and 2840°C (e) - from A. Oberlin (3)

visualized in a spectacular way.

The image really LOOKS LIKE a real representation of the lattice

planes. The only drawback, which however must always be remembered,
is that it is only a two-dimensional projection of a real three-
dimensional object, and that such an image may be very misleading
if the observation is limited to one direction. It is always wise to
observe from several directions.

2.3. Conclusion

This brief review shows that, while structural characterization

of carbonaceous materials may be quite straightforward and easy when
large and reasonably well crystallized samples are available, it is
often very tricky and necessitates sophisticated equipments and
experienced operators because of the usual poor crystallinity and
small dimensions of the structural units.
 106

X-ray diffraction nearly always yields easily an average d oo2

distance, possibly a Lc value, and also some indications on global
preferred orientation (if any), which can be used as characterization
parameters for both coals and carbonized substances: for this kind
of materials further information cannot be obtained without an
electron microscope and a skilled operator. Pregraphitic carbons
(H.T.T. > 1000°C) may be characterized also (with caution) through
La determinations. Only in the graphitization range (H.T.T. > 1600°C)
can a complete X-ray diagram be usefully interpreted (band
modulations and profiles). But in all cases electron diffraction
alone will give the details of the local organization and
orientation of the small « l~m) domains constituting the sample.

3. INTERACTION WITH VISIBLE LIGHT

Perhaps the most "visible" characteristics of carbonaceous

matter is that it is black, so black indeed that coal is often used
in the common language as a symbol of blackness. The strong
absorption of visible light by carbons is therefore a dominant
property which must be taken into account: the depth of penetration
of these radiations is at'most a few hundred nanometers. Visible
light cannot interact with anything but the most superficial outer
"skin" of a carbon material. The use of optical methods is limited
to the characterization of the surface of specimens.

Since most of the visible radiation falling on a carbon surface

is either absorbed or scattered (possibly reflected), the obvious
techniques which can be used are the study of the absorption and
diffusion spectra, and the measure of the reflectance.

The visible (and ultra-violet) absorption spectra of carbon

materials proved of very little use, as absorption increases
regularly with frequency and shows no conspicuous features. Infrared
absorption however is weaker and can yield interesting results,
which will be examined later.

The' spectrum of incoherently scattered visible light, i.e. the

Raman effect, of carbons and coals has been intensively studied
during the last ten or fifteen years, and Raman effect was shown to
be an efficient technique of characterization.

Reflectance measurements have been now for years a standard

method of evaluation of coals and of studying the texture of carbons.

3.1 Reflectance

Measuring the intensity of light reflected on a polished carbon

surface, when the incident and reflected beams are perpendicular to
the surface, is a rather straightforward operation. It can be done
 107

through a microscope, which allows a detailed study of the

reflectance of the various components of a sample, at the sub-
millimeter scale.

The reflectance R (ratio of the reflected to the incident

intensity) is, for an isotropic substance:

(1) R = (v o - n)2 + k~
(v o + n) 2 + k~

where n is the refraction index of the transparent medium (air or

oil), Vo the refraction index of the carbon, and ko its absorption
index.

But many carbons are anisotropic substances, since their

structural units are either aromatic stacks or graphitic layers: the
movement of n electrons is much easier along the aromatic or
graphitic plane directions than perpendicular to them. In other words,
the 0a "in plane" conductivity is much higher than the 0c
conductivity along the c-axis of the crystallites. A light beam with
its polarization vector parallel to the aromatic or graphitic layers
("a" direction in figure 10) is not reflected in the same way as a
beam polarized parallel to the c-axis ("c" direction in figure 10).
There is a maximum value Ra and a minimum value Rc of the reflectance,
and a "reflectance anisotropy" Ra/Rc'

When the reflecting surface is randomly oriented relative to the

aromatic or graphitic layers, there are still principal directions
of polarization corresponding to the projections of the "a" and "c"
axes on the surface, and a measurable (although smaller) reflectance
anisotropy. In a polarizer-equipped microscope, the incident light
is polarized in a plane which may make any random angle with the

·~1....

Fig. 10: Reflectance anisotropy

108

aromatic or graphitic layers of the sample surface: the reflected

light is depolarized, but from each one of the various domains cons-
tituting the surface the light is reflected with a different
polarization and a different reflectance. Examination of the
reflected light through an analyzer gives a microscopic image whose
multi-colored and contrasted areas are clues to the domains
orientations.

The reflectance of all the macerals of coal can be measured, but

the vast majority of results have been obtained with vitrinite. The
reflectance of vitrinite has been found to increase quite regularly
with the carbon content. Usually the values do not depend very much
on the direction of the incident light polarization, as coals are
generally rather isotropic at the optical microscope scale. However
a small difference can be observed for coals of high carbon content
when the reflectance is measured parallel or perpendicular to the
bedding plane.

Other carbon materials have a carbon content higher than that

of most coals, and their reflectance is consequently higher. For the
same reason their anisotropy is also larger, except in the case of
quite low H.T.T. (lower carbon content) or grains of very small size
(anisotropic but randomly oriented).

Anisotropic carbons show very characteristic images when the

polarizer and analyser are at right angle: no reflected light can
be received from domains where the aromatic or graphitic layers

Fig. 11: Cross-section of a fibre with deposited pyrocarbon

as seen through polarizer-analyzer equipped microscope.
Bright areas correspond to depolarized
reflected light
 109

("a" direction) lie parallel either to the polarizer or to the

analyzer polarization vector since in these cases the reflected and
incident beams are similarly polarized. For all other orientations
the reflected light is depolarized and the corresponding areas of
the image are bright. By rotating the sample, the "dark areas" can
be moved around and a complete exploration of crystallite
orientations can be performed (figure 11).

From this kind of microscopic observation with polarized light,

various "textures" of carbon have been defined:

- general orientation of the grains (e.g. natural graphite or

anthracite)

- parallel strata (e.g. pyrolytic carbons)

- fibrous textures (e.g. some petroleum or pitch cokes)

- granular (e.g. some other cokes)

- isotropic.

3.2. Raman effect

Interaction of an incident visible radiation (photons hVo ) with

the phonons (hv) of a target substance results in an incoherent
light scattering, the Raman effect. Measurement of the frequencies
v' of the scattered light yields the so-called "Raman frequencies"
vR = Ivo - v'l. The "first order" Raman spectrum is constituted of
the frequencies of single phonons which satisfy the selection rules
(phonons with zero wave-vector). The "second order" spectrum may
correspond to the combination of several phonons with a zero total
wave vector.

Incoherent light scattering from carbon materials could not be

studied before the advent of the laser as a light source. After the
first Raman spectrum of graphite was reported in 1970 (9), a number
of laser Raman spectroscopic investigations were conducted on coals
and carbons. Simultaneously the phonon dispersion theory for graphite
was developed and extended to many pregraphitic carbons and
intercalation compounds. We are now well equipped, both experimentaly
and theoretically, to use Raman spectroscopy as a characterization
technique for carbonaceous substances. Moreover the Raman microprobe
helps give a new insight into the microscopic structure of
heterogeneous and composite materials: spectra from areas as small
as 1 square micrometer can be recorded.

Two vibration modes only of the perfect graphite lattice are

Raman active: at 42 and 1582 cm- 1 (figure 12). But observation of
 110

o
I I
I I I

rM2l
~
iP2, E!.2)
2,
Fig. 12: Raman active vibration modes of graphite
E(l) at 42 cm- l and E(2) at 1582 cm- l
2g 2g

the first one is extremely difficult, so that for all practical

purposes the Raman spectrum of perfect graphite consists of a single
line (E 2g symmetry) at 1582 cm- l •

In addition to this "first order spectrum", a "second order

spectrum" can be observed near 2700 cm- l •

The other carbon materials, including coals, generally show more

complex spectra because their numerous structural imperfections
result in a relaxing of the selection rules:"forbidden lines" may
be observed.

The most visible features of these imperfect carbons spectra are

a new "band" in the 1350 cm- l region, a slight shift of the E2g line
from 1582 cm- l to higher frequencies, and a broadening of the E2
and 2700 cm- l lines. As an example, figure 13 shows the spectra of
a pitch coke heat-treated at increasing temperatures.

A measure of the structural disorder in the carbon or coal is

the ratio of the intensity of the 1350 cm- l band to that of the E2g
line: it decreaseB to zero as graphitization proceeds. Some authors
even claim an inverse linear relationship between this ratio and the
value of La.

But a Raman frequency or a line or band-width is easier to

measure accurately than intensities: it was suggested (10) to use
the following "parameters" to determine the position of carbons on
the "graphitization path":
 III

1400 1600 1600 2400 2600 2800 3000

I I I I I I I

,.....". ' .. 1 Acr• d'

. . . ..A,. . .,. .

Fig. 13: First (left) and second (right) order Raman spectra
of a pitch coke heat-treated at 1100°C (bottom),
2250°C (middle), and 3000°C (top) - upper scale is
in cm- 1
 112

a) parameters indicating two-dimensional ordering (i.e. showing the

progressive building of "graphenes"):
- the frequency of the E2g line, '0E ,which shifts from 1600
to 1580 cm-1 2g
- the E2g line-width, 6'0E2g' which decreases to 20 cm- 1

b) parameter indicating the onset of three-dimensional order:

- the 2700 cm- 1 line-width, 6'0 2700 : it decreases down to a
minimum during the two-dimensional ordering, then increases
with complete graphitization.

Figure 14 illustrates the meaning of these indicators: it shows

the "graphitization path" in a space of coordinates 6'0E and
6v 2700 . 2g

3.3. Conclusion

The "opticat' properties of carbon materials are actually those

of carbon surfaces, and they can be used for materials
characterization only if the surface is assumed to be similar to the
inside. This distinction may prove important in some cases: for
instance polishing a sample may alter its Raman spectrum.

150
x

• •
... +
o

"'q{.~606ox·• •
50
\
\ ... xo
• anthracene cokes
o pitch cokes
+ saccharose cokes
x fibers
.. pyrocarbons

Fig. 14: "Graphitization ~ath" of carbons as monitored through the

E2g and 2700 cm- line-widths
 113

Quantitative use of reflectance is in practice limited to coals

(and mainly vitrinite), for an evaluation of their rank or carbon
content. But qualitative microscopic examination of thin plates or
polished surfaces under polarized light is a very convenient way of
characterizing the anisotropy or texture of most diverse carbon
materials.

From the point of view of Raman scattering, all carbons,

including coals, behave very similarly. Carbonization has no great
influence on the spectra, but pre-graphitization and graphitization
are easy to follow by measurements of several "Raman parameters"
which show the progressive two-dimensional and three-dimensional
orderings of the materials.

4. FREE ELECTRONS - INTERACTIONS WITH ELECTRIC AND MAGNETIC FIELDS

AND THERMAL GRADIENTS

A common feature of both aromatic molecules and graphitic layers

is the "hexagonal ring" structure with its 'IT electrons. The
corresponding specific properties can be used to monitor the
evolution of the carbon materials.

These properties are those of phonons (from the carbon atoms

network) and nearly free 'IT electrons. The phonons may be investigated
by Raman and infra-red spectroscopy, and measurements of thermal
conductivity and specific heat. The most useful electronic properties
are electrical conductivity, magnetic susceptibility, Hall effect,
magnetoresistance, and thermo-electric power.

We have already reviewed the results from Raman scattering, and

will later examine infra-red absorption. Thermal properties are
actually of little use for the characterization of carbons, except at
very low (near liquid helium) temperatures, when their measurement
~s most delicate.

We will now exam~ne the properties of nearly free 'IT electrons.

4.1. Electrical conductivity

Measuring an electrical conductivity is apparently one of the

easiest things in the world. Unfortunately most carbon materials are
either powders or solids with a high porosity, and they very often
are not even homogeneous . ... The measured conductivity value may
prove meaningless, since it may reflect mainly the resistance
between carbon particles or through pores.

A significant value of the electrical conductivity, that is a

value which is indicative of the actual structure of the aromatic or
graphitic units, can only be obtained if the parasitic contact
 114

resistances between particles are reduced to a negligible minimum.

This obviously could be achieved by applying a sufficient external
pressure to the sample. A number of trials and studies have been
conducted on the effects of pressure upon the measured conductivity
of carbon powders (11,12): let us summarize the main findings.

a) The resitivity p decreases as P-s with applied pressure P, and no

definite limit was found even at pressures as high as 1 kilobar.

b) The effect of pressure is not reversible: the p = f(P) curve

obtained during the first application of increasing pressure is
different from that obtained at decreasing pressure and also from
the curve corresponding to the subsequent similar experiments.

c) The measured resistivity decreases with increasing compression

time.

d) The resistivity ~s a function of the particle s~ze of the carbon

log P (Q.em)

Fig. 15: Evolution of the resistivity of carbons during

heat-treatment. An insulator-conductor
transition occurs in the H.T.T. = 600-700°C
range
 115

powder: it increases with decreasing size. Moreover it may be

influenced by the grinding operation itself, through the formation
of a surface oxide layer (which can be removed by heating).

This brief summary is sufficient to show how careful one must be

in using resistivity measurements to characterize a carbon material.
Comparison between two samples is significant only to the extent
that measurements have been performed in identical conditions. Even
so, the reproductibility of experimental values may not be better
than an order of magnitude.

Fortunately the conductivities of carbons vary on a large scale.

Aromatic molecules are practically insulating substances with
resistivities in the order of 10 10 ~.cm, coals and chars are poor
conductors, and the resistivity of carbons decreases with increasing
carbonization or graphitization: the polycrystalline graphite
resistivity is about 10- 5 ~.cm. Figure 15 shows schematically that
evolution: it is clearly during the carbonization process that the
conductivity increases most rapidly (by 12 orders of magnitude), and
a real transition between an insulating and a conducting state occurs
when the material is heat-treated around H.T.T. ~ 600-700°C.

The cause of this transition is obviously the loss of hetero-

atoms from the parent organic material: oxygen and hydrogen are first
eliminated as water vapor and carbon oxide or dioxide, and at higher
temperatures low molecular weight hydrocarbons are lost as the side
chains are cut out and cyclization takes place in the carbon skeleton.
Nitrogen and sulfur are more difficult to drive out and may remain
in the carbon up to 1800°C. This correlation between heteroatom loss
and increased conductivity is paralleled by a similar decrease of
the resistivity of coals of increasing carbon content (8).

But graphite is also a very anisotropic crystal: its conductivity

anisotropy ratio Ga/G c is in the order of 10 3 - 105 at room
temperature, depending upon the crystal perfection. When a poly-
crystalline graphite or a coal or carbon powder is studied (and also
a compact carbon with unknown preferred orientation of the graphitic
layers), the measured resistivity reflects mainly the resistivity
along the layers. This must be kept in mind for any interpretation
of the results.

Figure 16 shows the temperature dependence of the a-axis

resistivity Pa of a graphite single crystal: it is a typical
"metallic behavior" since Pa increases with temperature. On the
contrary all pregraphitic carbons, chars, and coals have a
CONDUCTIVITY which increases with temperature, at least below 1000°C
(figure 17). It is a "semi-conductor behavior", although it would be
a mistake to assume that such materials are true semi-conductors:
they usually do not have a sufficiently well defined crystal structure.
116

P, (l1 Q cm)

Fig. 16: Temperature dependence of a-axis resistivity

of a graphite single crystal

P (Q .cm)_10 3

1000 2000 T (-C)

Fig. 17: Temperature dependence of the resistivity of

soft carbons at various graphitization stages
 117

log P (Q.em)

HTT= ~oo ·c
~

Fig. 18: Temperature dependence of the resistivity of

a natural coal heat-treated at increasing
temperatures

~owever a plot of the log of resist~v~ty vs. reciprocal temperature

generally shows a linear variation in the lower temperature range,
as illustrated by figure 18 for a carbonized coal (13): the higher
the carbonization temperature, the smaller the slope of the linear
part of the curve, so that the "apparent activation energy" which
can be measured for these carbons decreases with increasing H.T.T.
It can be used as a characterization parameter, but one should be
aware of its lack of real physical significance: carbonized substances
(H.T.T. < 1000°C) are essentially heterogeneous materials and any
meaningful interpretation of their behavior must take that fact into
account (14).

4.2. Diamagnetism

Diamagnetic anisotropy is one of the most prominent properties

of aromatic n electrons: it is already remarkably large in benzene
because of the very wide radii of the electronic orbits in the
aromatic layer plane, but it becomes extremely large in the graphite
crystal. If X is the diamagnetic susceptibility measured along the
aromatic 1aye~s and Xc the susceptibility parallel to the c-axis,
the anisotropy is defined as Xc - Xa : in electromagnetic C.G.S.
units per gramme, its value is about - 1.3x10- 6 for benzene, and
- 21x10- 6 for the graphite single crystal at room temperature.
118

Ixc -x.lx106 (emu g-l)

o~ ____
~ ______ ~ ______ ~

o
10 3 /T

Fig. 19: Temperature dependence of the diamagnetic

anisotropy of a graphite single crystal (the
actual values are negative)

Moreover, the graphite susceptibility Xc is strongly temperature

dependent, while Xa = - 0.3x10- 6 is completely temperature
independent: the temperature variation of Xc - Xa is shown on figure
19. It is now quite well understood and explained thanks to the work
of McClure and coworkers (15).

It is impossible to measure the anisotropy of coals and carbons

when the structural units of these materials are not conveniently
oriented parallel to a same direction. It is however possible to
indirectly determine it from measurements of the average diamagnetic
susceptibility X = (2Xa + Xc)/3 since Xa is never very different from
its graphite single crystal value. The behavior of X reflects very
closely that of the anisotropy. Figure 20 shows the evolution of X
during the carbonization-graphitization process: the rapid increase
of the diamagnetism corresponds to the two-dimensional growth of
graphitic layers, which allows for larger and larger orbits of the
TI electrons, and the small maximum at the end of this growth phase
 119

XX 106 (emu CCS.g-')

6
craPhiteJ
5

0~ __L -_ _ ~ _ _~_ _~_ _- L_ _~_ _~_ _- '

< carbonization> <.g rap hit i z a t ion > log 0

Fig. 20: Schematic evolution of the average diamagnetic susceptibi-

lity of carbons at room temperature (actual values are
negative)

c::
.-
~
C)

It)

.~
.--c:
~

Q..
Ib
(.,.
~

solid
O~ __ ~ ____ ~ __ ~ __ ~ __ ~~ 103 /T
______ ~

o 2
!ig. 21: Temperature dependence of the average diamagnetic suscepti-
bility of carbons of increasing graphitization stage
120

is an indication of the existence of "graphenes", which have a

diamagnetic anisotropy actually larger thant that of graphite, as
theoretically shown by McClure (15)

Simultaneous with the increase of X is the appearance of its

thermal dependence: figure 21 shows various curves where X is plotted
vs. the reciprocal temperature. A gradual change from the organic
molecules temperature-independent diamagnetism to the graphite
behavior can be observed. All these curves are quite well represented
above 77 K by the same equation:

(2)

where Xo and T can be used as adjustable parameters. The X

and T
va 1ues are conven1ent
0 to c h aracter1ze
'h "1
t e progress1ve bU10
d1ng of 0
two-dimensional order in the graphitic materials. However the above
formula is derived from the theoretical value of the diamagnetic
anisotropy of a two-dimensional electron gas (16): it lacks
sufficiently strong foundations in the case of n electrons in the
graphitic lattice and has been proved inadequate to account for the
experimental magnetism of carbons below 77 K (17,18).

Pyrolytic carbons deposited at high temperature are a special

case in the evolution of magnetism from carbons to graphite. Their
deposition temperature (TD> 2000°C) is so high that the corresponding
two-dimensional structure is already quite extended: its diamagnetic
anisotropy is actually larger than that of graphite since it is very
close to that of theoretical "graphenes" (15). When these materials

+ HTT: 2 550 ·C
HTT: 2600·C
o HTT : 2650"C
20

15

10 10 10 4 o (min)
Fig. 22: Evolution of the room temperature diamagnetic anisotropy
of a pyrocarbon (TD = 2100°C) during its graphitization
(single curve resulting from 3 isothermal studies)
 121

are heat-treated at still higher temperatures, their diamagnetic

anisotropy DECREASES as shown by figure 22 AND GOES THROUGH A
MINIMUM before increasing again to the graphite value.

4.3. Galvanomagnetic Effects - Hall Effect and Magnetoresistance

Galvanomagnetic effects are observed when a magnetic field and

an electric field are simultaneously applied: a Hall voltage and a
change in electrical resistance ("magnetoresistance") can then be
measured. The studies are generally performed in the simplest case,
where the fields are mutually perpendicular. Figure 23 shows the
principle of such measurements. The Hall voltage is measured in the
direction perpendicular to both fields, and its value is given by

(3) V=AIH/t

where I is the d.c. electrical intensity, H the magnetic field, t

the length of the sample along the magnetic field direction, and A
is called the "Hall coefficient". The magnetoresistance ("transverse
magnetoresistance" in that case) is defined as

(4) fiR / R = [R(H) - R(O)] / R(O)

where R(H) is the d.c. resistance of the sample when field H is

applied.

The "low field theory" of Hall effect and magnetoresistance

(which is usually valid for magnetic fields generated with

I
I
I

----------------4" I

,~ ""'"
I

~ [
Fig. 23: Principle of Hall coefficient measurement
 122

conventional electro-magnets) shows that:

1) If conductivity results from only one type of charge carriers

(for instance Tf electronsor alternatively holes in the Tf band of
graphite), the magnetoresistance is zero, and the Hall coefficient
is given by:

(5) A = l/Nq

where q is the electric charge of the carriers and N their

concentration (number per volume unit). Then the sign of the charge
carriers and their concentration N can be directly computed from the
Hall coefficient value A.

2) If conductivity results from several types of charge carriers

(e.g. electrons and holes) the Hall coefficient may be a function of
temperature and of magnetic field, and magnetoresistance is positive
and increases as the square of H.

The graphite single crystal Hall coefficient (measured with H

and I perpendicular to the c-axis) is dependent upon both magnetic
field and temperature. At room temperature it is negative (the
majority carriers are electrons) but at low temperatures it may be
positive at low fields and negative at high fields (figure 24). That
electrical conduction is simultaneously carried out by electrons and
holes is confirmed by the very large positive magnetoresistance of
the graphite single crystal, which may be as high as 10 4 • With
imperfect crystals however, much smaller magnetoresistance values

(cm 3 . C·,)
0,5
•
•
o _e _ _ _ _ _ _

Fig. 24: Single crystal of graphite: magnetic field

dependence of the Hall coefficient at 4.2 K
 123

15

10

< carbonizatiort> < graphitization > log D

Fig. 25: Schematic Hall coefficient behavior during
evolution of carbons
are observed, since this property is extremely sensitive to crystal
defects.

Contrary to graphite, in most carbons a single type of charge

carriers is responsible for the electrical conduction, and the Hall
coefficient is independent of temperature and magnetic field. Its
behavior along the carbonization-graphitization process is
illustrated by figure 25 (or see also fig. 2), which shows that this
property is a very sensitive indicator of the evolution of the
carbon. During carbonization the Hall coefficient is positive
(conduction by holes), but its measurement is rather difficult
because of the high resistivity of the material. The pregraphitization
stage corresponds to a first change of sign (electrons become
predominant), but a second change occurs at the beginning of
graphitization. During most of this process the conduction is carried
out predominantly by holes (electrons are missing at the top of the IT
valence band), and the maximum of the positive Hall coefficient
corresponds exactly to the end of the two-dimensional organization
and the onset of three-dimensional order. During this l~st stage
(from "graphenes" to graphite) the Hall coefficient is decreasing and
temperature dependent.

The graphitization can also be monitored through the magneto-

resistance changes, as shown by figure 26. During the carbonization
and pregraphitization stages the magnetoresistance mayor may not be
124

80K

3Dim ~
-0,005
Fig. 26: Transverse magnetoresistance of a pitch coke during its
graphitization (3 hours at H.T.T. - 0.65 Tesla)

negligible, but it is always NEGATIVE as soon as the two-dimensional

organization starts setting in (H.T.T. = 1600°C), reaches a minimum
at the beginning of three-dimensional ordering (H.T.T. around 2000°C)
and then increases continuously up to very high values if the
material is completely graphitizable. It is also much stronger at low
temperatures.

In spite of its obvious usefulness for characterization purposes,

the negative magnetoresistance of pregraphitic carbons is still
somewhat of a mystery: it is by no means an extraordinary property,
since many other inhomogeneous materials display a similar behavior,
but no really satisfactory theory has been put forward to explain it
(19).

4.4. Thermoelectric Power

When a closed circuit is constituted of two different conducting
materials, A and B, there are two junctions between A and B. If these
junctions are not at the same temperature, a voltage difference ~V
 125

can be measured between them: let ~T be the temperature difference,

the thermoelectric power of B relative to A is

(6) T.E.P. = ~V/~T

and the absolute T.E.P. of B can be determined, provided that of A

is known.

These measurements, which can be perturbed by many other effects

or artifacts, are difficult. They however have been performed for a
number of various carbons, mainly materials deposited or heat-treared
above 1500°C since it was hoped such carbons could be used for the
manufacture of thermocouples for high temperature utilization. The
sign of the thermoelectric power reflects that of the majority
carriers (as does the Hall coefficient). Figure 27 shows for instance
that materials in the two-dimensional ordering stage have a positive
T.E.P., which increases with temperature, while the graphite
thermoelectric power is negative and a decreasing function of
temperature. Moreover the figure also shows a strong negative peak
on the graphite curve around 100 K, which has been the subject of a
great many theoretical papers (phonon drag effect), thus illustrating
how difficult the interpretation of such measurements may be.

TEP (tJ-V/K) To =
1600°C

~1900°C
----~

o
Fig. 27: Temperature dependence of the thermoelectric power of
pyrolytic carbons deposited at various temperatures (Tn)
 126

4.5. Conclusion

The properties of the ~ electrons yield some the easiest and

most efficient methods of characterization of coals and carbons.

Although electrical conductivity and thermoelectric power

measurements are somewhat difficult to perform accurately and to
interpret, a large amount of information can be drawn from
diamagnetism, Hall effect, and magnetoresistance. It is possible to
monitor:

- the PREGRAPHITIZATION stage through diamagnetism and Hall

effect,

- the TWO-DIMENSIONAL GRAPHITIZATION through magnetism, Hall

effect, and magnetoresistance,

- the THREE-DIMENSIONAL GRAPHITIZATION through Hall effect and

magnetoresistance.

And finally magnetoresistance is a choice method for assessing

crystalline imperfections and the "true" anisotropy of poorly
oriented materials.

5. MONITORING OF THE CHEMICAL EVOLUTION: INFRA-RED AND MAGNETIC

RESONANCE SPECTRA

The progressive transformation of an organic substance into

graphite actually results from successive chemical reactions (loss
of side chains or heteroatoms, cyclization, etc •• ). Since the outcome
of these reactions can be observed through modifications of the
infra-red absorption spectra and magnetic resonance signals (ESRand
NMR), these techniques enable us to monitor the chemical evolution
of carbons.

5.1. Electron Spin Resonance

Transitions between magnetic electronic energy levels can occur

with the absorption of suitable photons: this is "electron para-
magnetic resonance" or "electron spin resonance" or E.S.R. The
relation between the magnetic field intensity H and the electro-
magnetic frequency V is:

(7) hv = gBH

where B is the Bohr magneton, and g the so-called "g factor", which
characterizes the nature and symmetry of the paramagnetic species
responsible for the resonance. The g value in carbons may be slightly
anisotropic (gc measured with the magnetic field parallel to the
 127

c-axis and ga measured with H parallel to the graphitic layers), but

is always very close to the free electron value: 2.0023. Because of
these very small variations, interpretation of the experimental g
values is very delicate in all cases.

ESR has been used extensively since 1954 for the study of carbons
and coals. Three kinds of informations can easily be obtained from
the E.S.R. signal (21,22,23) :

- the integrated intensity of the signal, which is proportional

to the paramagnetic susceptibility of an average spin and to the
total number of spins in the whole sample,

- the line-width of the signal, which is inversely proportional

to the "relaxation tine" of the spins, and can be translated in terms
of "spin-spin interactions" and "spin-lattice interactions",

- the variations of the signal intensity with temperature, which

can be translated either as variations of the spin concentration or
as variations of the average spin susceptibility: if the para-
magnetic species are free electrons, their total susceptibility
(Pauli paramagnetism) is independent of temperature, but "localized
centers" such as organic free radicals have a susceptibility which
follows the Curie or Curie-Weiss laws.

Organic free radicals are created in coals and carbons during the
first stages of the carbonization process, mainly through breaking
of C-O, C-H, and C-C bonds. Consequently the ESR intensity increases
strongly and can be very easily detected. Since the samples are solid,
these free radicals have a very low mobility, and being (at least at
the lower H.T.T.) widely separated from one another, they cannot
recombine and are very stable: they can be observed during any length
of time. The intensity increases with the carbon content in coals
and chars. At H.T.T. around 600°C the free radicals have become so
numerous that they are no longer widely separated, and they start
recombining: a maximum is observed in the paramagnetic susceptibility
Xp (ESR intensity) which corresponds to free radical concentrations
in the order of 10 19 - 10 20 per gramme of carbon. Figure 28 shows
schematically the evolution of Xp (measured at room temperature) as
a function of the heat-treatment variable D: after its maximum, X
decreases slowly during the end of carbonization and during the p
pregraphitization, and more rapidly during graphitization. But during
the pregraphitization stage the signal may be so broadened that it
very often seems to disappear completely (dashed lines on figure 28).

A study of the temperature dependence of the ESR intensity yieMs

results similar to those presented in figure 29: the variation of Xp
is linear as a function of 1/T. This means that the paramagnetic
susceptibility originates from two kinds of species: free electrons
128

,
,
I
I

1 I
I
I
I

Or~ ____ ~ ______ ~ ____ ~~ ____- L_ _ _ _ _ _ ~~

< carbonization > <: graphitization > D

Fig. 28: Schematic evolution of Xp with heat-treatment

~o

20

o~ __ ~ ______
~ __________ ~~ ________~~ ________ ~~_

a 10 50 1 7
1031T (K-I)
Fig. 29: Temperature dependence of Xp (anthracene chars) (25)
 129

<carbonization> <graphitization >

2

-
"'0-
)(

::--
,
Xp (Paull)
Xp (Ioc)
1
-
"'-
0
)(

tJ)
~,
tJ)
V) I
(,!) V)
C,,) (,!)
/ C,,)
::)
E: 1 I \ ::)
~ \ E:

-
,5 ~
~ \
g \
~
: :)
I» \ <II

, ... ~
...
... _- --------
'<11 \
"(} ><
oS!
"-
...
>< 0 0
3000
> HTT ("C)
<carbonization ~ <graphitization
Fig. 30:
Schemat~c evolution of the "free" and "localized"
electrons contributions
and localized radicals which add up their contributions. The free
electrons contribution is independent of temperature (Pauli para-
magnetism) and the localized radicals have a Curie temperature
dependence: each contribution can be separately determined from the
experimental results (figure 29) by extrapolating the straight lines
to infinite temperature.

The "localized" and "Pauli" susceptibilities are schematically

represented in figure 30 as a function of H.T.T. The end of the
carbonization process corresponds to the near-complete disappearance
of the localized free radicals, while the free electron concentration
starts building up from 700°C on, and reaches a maximum towards the
end of pregraphitization. The disappearance of the free radicals is
easy to explain through the recombination processes in a solid medium
where these radicals are densely packed (H.T.T. between 700 and
1000°C) or where electrons move more and more freely (H.T.T. above
1000°C). But the free electrons are more puzzling because the carbon
materials are very inhomogeneous and poorly crystallized at that
stage. Some kind of hopping conductivity is assumed in these
disordered solids: the carrier mobility (rather than their number)
probably increases with crystalline order.

But figure 30 shows that the free electrons concentration

eventually starts decreasing as soon as graphitization proceeds:
"graphenes" are bui It and progressively acquire tridimensional order,
 130

so that there is now a true crystalline band structure, with less

and less TI electrons trapped in the defects. The positive Hall effect
shows that the carriers are holes (see § 4.3) but the electron
deficit in the TI valence band of the graphitic solid is progressively
reduced and the number of holes falls to the small graphite value.

5.2. Nuclear Magnetic Resonance

Using N.M.R. for the study of solids is much more difficult than
for liquids. Coals and carbons are no exception, be it proton (lH)
or 13 C resonance. The main reason is an extreme broadening of the
resonance "lines" in solids, which used to prevent any attempt at
high resolution work. Two factors (which are averaged out in liquids)
contribute to this broadening:

- dipole-dipole interaction between neighbouring nuclear spins

(lH-IH or IH_ 13 C interactions since 13 C nuclei are statistically too
far from one another to interact significantlY);

- chemical shift anisotropy: the chemical shift is a tensor and

the resonance line POSItIon varies according to the orientation of
the magnetic field relative to the local molecular frame of the
solid.

Moreover, in the case of 13 C resonance, the signal intensity is

very weak because the very long spin-lattice relaxation time results
In a very easy saturation and the necessity of using small R.F. power.

This is the maIn reason why N.M.R. studies in carbons were fora
long time restricted to proton work. Only during the last decade did
13C resonance become important, thanks to recent technical
improvements.

Classical proton resonance gives broad lines spectra with very

little useful details. Compared with the band widths, chemical shifts
are too small to allow any identification of specific features. But
aromatic and aliphatic hydrogen atoms give two bands which can be
separated and an estimation of the aromatic/aliphatic ratio in the
carbon is thus obtained (8).

The study of the 13C resonance in carbons and coals was given an
interesting new potential, first by the use of "proton decoupling"
and Fourier transform (F.T.) spectrometers, and later by the advent
of the so-called "cross polarization-magic angle spinning" technique
(C.P.-M.A.S.)(26,27).

- Proton decoupling (excitation of the proton resonance

simultaneously with the 13C resonance) averages out the dipole-
dipole interaction between IH and 13 C•
 131

- Cross polarization allows, through an adequate timing of the

proton and 13 C R.F. excitation pulses, an energy transfer between
the IH and 13 C nuclei which is equivalent to a cooling of the 13 C
spins to a lower temperature, and consequently enhances the 13 C
signal.

- Magic angle spinning consists in rotating the sample around an

axis at an angle a = 54°.7 with the static magnetic field
(cos 2 a = 1/3): such a rotation averages out to zero the chemical
shift anisotropy, provided its frequency is sufficiently high. But
it also generates "side-bands" corresponding to harmonics of the
rotation frequency, which have to be eliminated from the "true" N.M.R.
spectrum.

Using these very sophisticated techniques and interpreting the

results are, at least up to the present time, the privilege of
skilled and experienced N.M.R. specialists. But a great deal of work
is going on in this field, and it may be expected that 13 C N.M.R.
will become an important method of characterization of coals and
carbons, particularly for hydrogen-poor materials where 13 C is the
only "probe" left in the carbon skeleton.

Distinguishing between aromatic and aliphatic structures is

already easier with 13 C than IH because the chemical shifts differ
much more in 13 C resonance. Aromatic and carboxylic carbons, C=O,
C=C, and C=C on one hand, aliphatic carbons on the other hand,
generate two widely separated bands (by 100 ppm), as shown on figure
31, which presents their evolution with coal rank (28).
Ar AI

Fig. 31: 13 C NMR spectra (CP-MAS)

of coals of various ranks. The
younger ones (top) display a much
stronger aliphatic band (28)
 132

5.3. Infra-Red Spectroscopy

In spite of their strong absorption of visible light, thin slices

of carbon or coal can transmit some infra-red radiation, particularly
in the longer wave-lengths, and yield useful infra-red spectra. The
carbon material is usually diluted in KBr pellets. It is however
difficult to obtain good spectra and many precautions are necessary
in order to avoid excessive radiation scattering or absorption, as
well as absorbed water vapor which may give very strong bands, not
to be confused with those of the carbon material. With conventional
spectrometers this is a time consuming procedure and it requires
experienced operators. Moreover it is often necessary to use signal
enhancement by computer averaging (C.A.T.) and this is still more
time-consuming.

The recent advent of Fourier transform Infra-Red spectrometers

(F.T.I.R.) has made possible the recording of very detailed spectra
in quite short times: but these spectra, although much more easily
obtained, are usually no more informative than good quality
conventional ones.

More strongly carbonized or graphitized materials, which contain

very few specific functional groups, give of course very "bare"
spectra. Most of the carbon atoms being part of the aromatic network,
the main features of such spectra are an "aromatic carbon band"
around 1600 cm- I and sometimes aromatic C-H frequencies in the
3000 cm- I region.

The materials heat-treated at the lowest temperatures, such as

coals, cokes, or carbon blacks, display much more complex spectra,
which may be quite useful for the characterization of their chemical
constitution, especially their surface functional groups. The
following approximate frequencies may be usually observed (but their
complete correct interpretation requires an experienced infra-red
specialist):

C-H stretching: aromatic (>3000 em-I) and aliphatic (2900 cm- I )

- CHzand CH 3 deformation: around 1350-1450 cm- 1

- C=O: 1700 cm- 1

- Aromatic and graphitic C=C: 1600 cm- 1

- C-O and C-O-C: from 900 to 1300 cm- 1

- Aromatic carbon skeleton: between 750 and 850 cm- I .

As an illustration, figure 32 shows the spectrum of a naphtalene

 133

4000

_VENUMBER (em"')

Fig. 32: Infra-red absorption spectrum of a naphtalene pitch

(H.T.T. = 500°C) (29)

pitci1 (29) where several of these features are clearly visible.

The CH aromatic and aliphatic stretching frequencies can be used

to determine the aromatic vs. aliphatic H content ratio of the carbo~
by measuring the integrated intensities of the respective infra-red
bands. This procedure has shown that the proportion of aromatic H in
a coal increases very rapidly with the C content above 90 % (8).

The aromatic H proportion also increases with H.T.T. while the

I1700/IloOQ intensity ratio of the bands at 1700 cm- 1 (C=O) and
1600 cm- 1 tC=C) decreases.

The whole evolution from an organic substance to a pregraphitic

carbon, as well as chemical modifications of the surface of the
material (e.g. oxidation) are thus reflected in the infra-red
spectra.

5.4. Conclusion

Infra-red spectroscopy is an interesting technique to monitor,

at least qualitatively, the chemical evolution of a low H.T.T. carbon
Reliable quantitative results (in terms of infra-red band intensity
ratios) are however much more difficult to obtain.

Nuclear Magnetic Resonance, although showing some very promising

prospects, does not seem yet to have more potentiel than determining
 134

the aromatic vs. aliphatic contents.

Electron spin resonance (which does not necessitate very

expensive nor sophisticated equipment) actually seems the best method
for a precise quantitative evaluation of the chemical changes which
take place in a carbon material submitted to heat-treatments or
chemical actions. Its information may be quite complementary to the
qualitative results of infra-red spectroscopy and a joint study by
both techniques can be very efficient.

6. CONCLUSION

This brief and incomplete survey already shows a great diversity

of physico-chemical methods which may be of interest for the
characterization of carbons and coals. Some of them, although very
informative, can only be used by skilled and experienced operators,
but many necessitate quite ordinary equipment and do not require any
particular training. We tried to show in each case the nature of the
information which can be obtained about the material under study.
Selecting any particular technique must be a matter of knowing what
particular information is seeked.

Generally speaking optical microscopic observation, E.S.R., and

infra-red studies are the most "rewarding" techniques in the
carbonization range, while X-ray determination of d oo2 ' diamagnetic
susceptibility and Hall effect measurements, and possibly Raman
effect (if a microprobe is available) are most efficient in the pre-
graphitization and graphitization stages.

Finally one must be aware that these measurements are not

performed on idealized "carbons", but on real carbon samples: only
carbon SAMPLES can be characterized, and the definition of
REPRODUCIBLE SAMPLES is the first problem to solve if meaningful
result.s, ~re' to be obtained. Carbons and coals being what they are,
this is not an easy task ••••

REFERENCES

(1) FISCHBACH D.B. (chapter 1) and A. PACAULT (chapter 2) in vol. 7

"Chemistry and Physics of Carbon" (New York, Marcel Dekker -
1971)

(2) OBERLIN A., Carbon, 12 (1979), 7-20

(3) OBERLIN A., J. GOMA, J.N. ROUZAUD, Proc. Intern. Carbon. Conf.
Bordeaux 1984 - in print. J. Chim. Phys. (1985)~, 701

(4) MAIRE J. and J. MERING, "Chemistry and Physics of Carbon"

(New York, Marcel Dekker - 1970), vol. 6, p. 125
 135

(5) MERING J., Acta Cryst., 1 (1949), 371-7

(6) WARREN B.E., Phys. Rev., 2 (1941),693

(7) ROUCHY J.P. and L. GATINEAU, Carbon, J2 (1975), 267

(8) VAN KREVELEN D.W., "Coal" (Elsevier 1961)

(9) TUINSTRA F. and J.F. KOENIG, J. Chern. Phys., 53 (1970), 1126

(10) LESPADE P., A. MARCHAND, M. COUZI, and F. CRUEGE, Carbon, 22

(1984),375-385

(11) MROZOWSKI S., Proc. 3rd. Carbon Conf. (Buffalo 1957) p. 495

(12) G.F.E.C., "Les Carbones" (Masson 1965) !, 340

(13) WATERS P.L., Proc. 5th. Carbon Conf. (Penn State 1961) ~, 131

(14) DELHAES P. and F. CARMONA, Chern. and Phys. of Carbon (New York,
Marcel Dekker - 1981) vol. 12, p. 89

(15) McCLURE J.W., Phys. Rev., 119 (1960) 606 -

SHARMA M.P., L.G. JOHNSON and J.W. McCLURE, Phys. Rev., B9,
(1974), 2467

(16) PACAULT A. and A. MARCHAND, Cornptes Rendus Acad. Sc. 241, (1955)
489 - MARCHAND A., Ann. Chirnie, J2, (1957), 469

(17) POQUET E. and A. PACAULT, Carbon, 1, (1963), 71

(18) MAAROUFI A., S. FLANDROIS, C. COULON, and J.C. ROUILLON, J.

Phys. Chern. Solids, 43, (1982), 1103-09

(19) DELHAES P., Chern. and Phys. of Carbons (New-York, Marcel Dekker,
1971), vo 1. ~, p. 193

(20) POQUET E., J. Chirn. Phys. 60 (1963), 566

(21) INGRAM D.j.E., G.G. TAPEY, R. JACKSON, R.L. BOND and

A.R. MURNAGHAN, Nature, 174 (1954) 797

(22) UEBERSFELD J., J.Phys. Rad., ~ (1954) 126

(23) G.F.E.C., "Les Carbones" (Masson 1965) !, 318-338

(24) MROZOWSKI S., Carbon ~ (1966) 841

(25) CARMONA F. and P. DELHAES, J. Appl. Phys. 49 (1978) 618

 136

(26) GERSTEIN B.C., "Analytical methods for coal and coal products"
(Academic Press 1980)

(27) GERSTEIN B.C., P. DUBOIS MURPHY, and L.M. RYAN, Adv. in

Chemistry, Am. Chem. Soc. (1981) 192

(28) LEGRAND A.P., "Resonance Magnetique Nuc1eaire pour Solide"

L'Actua1ite Chimique l (1985) 27

(29) GREINKE R.A., and I.C. LEWIS, Carbon 22 (1984) 305

 137

TEXTURAL CHARACTERIZATION OF POROUS CARBONS

BY PHYSICAL ADSORPTION OF GASES

Angel Linares-Solano

Department of Inorganic Chemistry. University of Alicante.

Alicante. Spain.

1. INTRODUCTION
The carbons involved in many industrial applications
take many diverse forms ranging from well-developed crystalline
structures, such as graphite and diamond, to the so-called
amorphous varieties such as carbon blacks, cokes, chars,
acti vated carbons.

The porous texture of carbons has an immediate relevance

to the rates of many heterogeneous reactions, e.g., gasification
reactions, because mass transport of the reacting gas from
the exterior surface to an active site beneath the surface
and transport of the products in the opposite direction (1,2)
takes place through the porous network. If the escape of
gas products is controlled by in -pore diffusion, then the
number, the shape and the size of the pores greatly affect
the extent of the accessible internal area of the carbon.
Increases in porosity and on accessibility will increase
both, mass transport and gasification rates. Walker and his
School (3-9) have related the decrease in reactivity of American
coal-chars and the decrease in acc€-ssibili ty to the internal
porosity of the chars to increasing rank of the coal from
which the char is produced.

Active carbons are produced almost. entirely through

the activation of carbonaceous material (10,11). The development
of their porous texture (micro, meso and macropores) and
as resul t, their surface properties will depend upon the
extent of the activation process. In order to understand
the adsorption capacity of an activated carbon and to r-elate
it with its industrial applications a knowledge about its
138

pore size distribution and its surface area is essential.

There are numerous techniques and methods for analysing

pore structure and internal surface area (11-18). Mercury
porosimetry and physical adsorption measurements are widely
used to obtain a good description of the texture of a porous
carbon. However, because of the various physical forms of
carbon. the characterization of porous carbon may be complex.
A wide range of pores needs to be covered, from those with
visible dimensions -gross cracks- to those with widths of
molecular dimensions -micropores- with an interconnecting
aperture-cavi ty pore system with pore entrances that are
smaller than the pores. Therefore, it may be difficult to
select the most appropriate method and, in almost all cases,
a combination of methods is necessary.

The purpose of this chapter is to give a summary of

the essential theoretical back-ground of the methods most
used in the characterizat~on of porous carbon solids, mainly
by physical adsorption technique; methods of analysing mesoporous
structure (e.g. by analysis of hysteresis loo~s associated
with capillary condensation) are not discussed. In the descrip-
tion which follows, no attempt is made to describe details
of recent work or to account for all the methods available
for the characterization of porous carbons. Rather, an outline
is given of some of the simpler ideas which underline the
study of gas-solid adsorption with the intention of discussing
their principles and applications as well as recognising
their limitations.

Special attention will be paid to the principal theories

applied to microporous carbons; the presence of micro~ores
makes interpretation of gas adsorption difficult, because
theories of microporous adsorption are not so well known
as with adsorption on non porous solids.

There is an extensive literature in this area (19-29)

which can be cor.sul ted by those requiring further information.

2. GAS-SOLID INTERACTIONS
When a gas or vapour (the adsorptive) is confined in
a closed space, at a given pressure and temperature, in the
presence of an outgassed solid (the adsorbent), an adsorption
process begins. The adsorptive molecules are transferred to,
an accumulate in, the interfacial layer, as a consequence
of an attractive force between the surface of solid and the
adsorptive (21,25).

In general, two types of adsorption are distinguished,

physical adsorption and chemisorption. Henceforth we will
 139

be concerned mainly with physical adsorption, that is with

systems in which specific chemical interactions between adsorbent
and adsorptive are absent, or almost so.

Physical adsorption includes attractive dispersion forces

and, at very short distances, repulsive forces, as well as
contribution from polarization and electrostatic forces between
permanent electrical moments and the electrical field of
the solid, if the adsorptive or the adsorbent have a polar
nature.

The dispersion forces were shown by London in 1930 (30)

to result from the rapid fluctuations in electron density
wi thin one atom which induce an electrical moment in a neigh-
bouring atom. If only dipole-dipole interaction is considered
the potential enegy, U d(r) ~f a pair of interacting atoms
is given by the well known equatlon of London

where r is the interatomic distance an~ C is a constant related

to the dipole-dipole interactions. The repulsion forces due
to the interpenetration of the electron clouds of the atoms
is often expressed in the form (21)

where Band m are empirical constants; usually m=12. The total

potential energy between the two atoms bec~mes
-6 -12
U(r)= -C.r + B r

Equation 131 can also bE ~ritten as

r r
U = £'1 (~)12 _ 2 (~)61
(r) r r
where ro is the separation for which dU/dr is zero, and
£' is the value of U when the separation is ro .
Equation 131
is usually known as the Lennard-Jones 6:12 potential (31);
although values of n=9 and m=14 can also lead to a reasonable
fit of data.

Assuming the property of additivity of the pairwise

interactions and regarding each atom in the solid as a force
centre, it is possible to obtain the potential energy of
adsorption of a gas molecule at a solid surface
140
where 4> (z) is the energy of adsorption of a molecule (i) as
a function of its distance z from the surface plane, and
I'ij is the distance betweoen- i and each force centre j in
the solid. Replacing the summation by an integral (21,25)
and assvming a uniform numcer of solid atoms per unit of
volume (p), the 6: 12 potential function between isolatec
atoms leads to a 3:9 potential function between an adatom (ad-
sorbed atom) and a plane solid surface.

P C 1(~)3 1 (zo )9 1
4>(z)= -
'IT

z 3 z
161
6 z 3
0

where Zo represents the value of z for which d4>/dr is zero.

In other wordE, the attracti ve force falls off

much less rapidly with distance than for two isolated atoms.

The potential energy of two isolated atoms and of a

molecule in the solid as a function of their distances apart
are sketched in Fig. 1. Both curves 4> ) against z and U ) aga-
inst r have roughly the same shape bu~Zthe equilibrium JEstance
z0 is smaller than the corresponding value of r 0 ' (being
Zo = 0.765 ro in the case of a continuous solid), the
4>(z) is higher than the U(r)' and 4>(z) falls to zero at a larger

solid molecule (i)

or Z ,r
b

Figure 1.- Potential energy curves. a) of two isolated atoms

and b) of a molecule in a solid as a function of distance apart
 141

distance from the surface than between the two isolated atoms.

Depending on the nature of the molecule and the solid,

specific interactions may contribute to the adsorption energy.
The overall interaction energy (I/>T) for a polar adsorbate
molecule on a heteropolar surface is given by the equation
(29,32,33).

I/>T= I/>D + I/>p + I/>F + I/>FQ + I/>R

where I/>D energy due to dispersion forces.

I/>p energy due to polar nature of adsorbent and/or adsorp-
tive.
I/>F and I/>FQ= energy due to permanent dipole and quadrupole moments
of the adsorptive molecules, respectively.
I/> = energy due to repulsion forces.
R
The terms I/>D and I/>R are the non-specific contribution
which are always present on physisorption, whereas the other
terms provide the specific contribution which may be impc.rtant
to the overall adsorption energy according to the ~olar nature
of the molecules and the adsorbents (34-35).

When adsorption on a microporous carbon is considered

the potential energy of adsorption will be enhanced. The
specific adsorbate-adsorbent interactions with microporous
carbons will be dealt with later on.

3. ADSORPTION-DESORPTION ISOTHERM SHAPES

The adsorption of a gas or vapour by a solid, among
other things, permits information about porous texture, surface
area, interaction energy adsorbent-adsorbate etc. to be ol:tai-
ned.

Many different procedures have been developed to measure

the amount of gas adsorbed on a conveniently outgassed adsorbent
(22,28). Those which involve the measurement of the amount
of gas leaving the gas phase, as happens in a volumetric
system, and those which involve the measurement of the uptake
of the gas by the adsorbent, as occurs in a gravimetric system,
are in practice the most extensively used procedures.

The amount of gas adsorbed (n) on the surface of the

solid is a function of the mass of the adsorbent, pressure
(P), temperature (T) and nature of the gas and of the solid
surface.

n f (P, T, gas, solid)

142

n, may be expressed in different units (moles, grams, cubic

centimeters at S.T.P or liquid volume), however reccmmendations
have been mc.de to express it as moles per gram of outgassed
solid (36).

If pressure and temperature are r.eld constant for a

given adsorbent-adsorbate system

n = f (P)T, gas, solid

The amount of gas adsorbed at equilibrium is only a

function of pressure. The representation of the amo~nt adsorbed
versus equilibrium pressure (or relative pressure PIP o ' where
Po is the liquid vapour pressure at the adsorption temperature)
gives the adsorption isotherm.

In 1940, Brunauer et al. (37) classified physical adsorp-

tion isotherms into five well known grOl':ps in which can be
included most of the experimental adsorption isotherms publi-
shed. Recently a sixth group (step-shaped isotherms), for
very homc'geneous solids has been added (29). The six types
of adsorption isotherms are schematically represented in
Figure 2.

Depending on the different physical forms of carbcn

adsorbents and the nature of the adsorptives, the six groups
of isotherms of Fig. 2 are observed on porous carbon solids .

•
1.0 1.0
p / po

Figure 2.- Schematic representation of different types of

adsorption isotherms (redrawn from 29)
 143

Type I isotherms are characteristic of microporous

carbons. These isotherms are very useful for obtaining a good
approximation of the total micropore volume of the solid,
by direct reading of the amount adsorbEd at the plateau of
the isotherm. They are widespread, for example adsorption
on activated carbons of mc.st gases and vapovrs.

Type II isotherms are observed for macro~orous and

non-porous carbons, for example adsorption isotherms of nitrogen
(77K) on many carbon blacks and graphitized carbon blacks.
Type II isotherm represer.ts unrestricted monc,layer-mul tilayer
adsorption.

Type III isotherms are characteristic of very weak

adserbate-adsorbent interactions (lower than adsorbate-adsorbate
interactions), for example adsorption of ammonia or water
vapour on graphitized r-arbon blacks. They are typical of
co-operative adsorption; the more molecules that are already
adsorbed, the easier it is for further molecules to become
adsorbed.

Type IV isotherms reSEmble isotherms of type II but

instead of approaching the line PIP o =1 asymptotically they
turn towards the ~ressure axis giving a nearly horizontal
part. It is characteristic of carbons which possess a wide
proportion of mesopores, as happens on compacting carbon
blacks under pressure; the adsorption of n-hexane vapour
gives an isotherm of type IV.

From type II and IV it is possible to determine the surface

area of the solid.

- Type V isotherms are characteristic of weak adsorbate-

adsorbent interactions and, as happens with type III,
again they are typical of a co-operative adsorption. Examples
of this type cf isotherms are very current on water vapour
adsorption on charcoal and on activated carbons.

Type VI isotherms are seldom encountered unless the

carbon has a very uniform non-porous surface. Experimentally
the step-shapEd isotherms are well established using noble
gas as adsorpti ve on somE graphitized carbon blacks. The'
step-he'ight represents the monolayer capacity for each adsorbed
layer.

The isotherm shape by itself can give a good idea of the

porous texture of the carbon under study, however, very often,
different types of pores are ~resent and, as a result, a
mixed isotherm is obtained, which is much more difficult
to analyse.
144

Another way to get information on the porous texture

of the adsorbent is to look at the shape of the desorptior.
branch. It is commonly found for porous adsorbents that the
adsorption and desorption branches are not coincident over
the whole pressure range. The hysteresis loop may te limited
only to the high relative pressure range, or may also persist
to the lowest relative pressure range. The terms high pressure
hysteresis and low pressure hysteresis are used (29).

The hysteresis loops may exhibit a wide variety of shapes

which have often been associated with specific pore structures.
The classi fication originally put forward by de Boer in 1958
(38) (from A to E) has been recently modified (HI to H4)
(29,36) as illustrated in Fig. 3. The essential features
of these loops were explained by de Boer in terms of pore
shape and the location 2nd form of the liqt.:id meniscus. All
of these hysteresis types are found on porous carbon solids.

- Type HI hysteresis (or type A of de Boer I s classifica-

tion) is often &ssociated with capillary condensation in
open-endend cylindrical-shaped pores. The formation of a
cylindrical meniscus occurs at a higher PIP o than the emptyin~
process, which proceeds through the evaporation from a hemisphe-
rical meniscus.

Type H2 hysteresis (or type E) is well established

in the literature but is difficult to interpret. It corresponds
to spheroidal cavities or void spaces between spherical close-
packed particles, as w6ll as to "ink bottle" pores. The

o 1.0
p / po

Figure 3.- Types of adsorption-desorption hysteresis loops

(Redr awn from 29)
 145

liquid is trapped in the body of the pore until P /Po is reduced

to allow evaporation from the neck, therefore the release
of condensate is limited by the neck radius.

- Type H3 hysteresis (or type B) which exhibits no limiting

adsorption at P/P o of unity is indiCative of slit-shaped
pores. The meniscus is not formed until a high P/Po is obtained.
On the desorption process the pore space is err.ptied by evapora-
tion from cyclindrical meniscus.

Type H4 hysteresis is associated to type I isotherms,

that is, with microporous adsorbents.

Low pressure hysteresis (39,40), as commented earlier,

may be observed.This phenomenon has been related to the swelling
of non-rigid porous structures and its corresponding distortion
by an irreversible uptake of the adsorptive within pores
of about molecular dimensions. For low pressure hysteresis
to occur the molecules need to be similar in size to the
widths of the pore entrances. Therefore, it is expected that
low pressure hysteresis decreases or disappears with increasing
the extent of activation, as ~as found by McEnaney (41).
However, different behaviour was observed by Martin-Martinez
et a1. (42) studying adsorption-desorption of different hydro-
carbons on a series of activated carbons, which were reacted
with air to various degrees of burn-off.The original activated car-
bons which have no low pressure hysteresis once reacted with

15 n/mmole.g- 1
C-3 (72 0,.)

10 C-2 (46°1.)

A-2(42·1o)

5
A-3 (72 0,.)

0
0 0.5 1.0
p / Po

Figure 4.- Adsorption-desorption isotherms of benzene on

gasified active carbons: A, gasified in air (350°C) and C,
gasified in CO~(825°C). (reduced from 42).
146

air, (350 0 C) , have low pressure hysteresis which increase s

with the extent of the air-reaction. Figure 4 shows the adsorp-
tion-desorption isotherms of benzene at 298K for two air
reacted activated carbons (carbons A) and two carbon dioxide
reacted-·activ&ted carbons (carbons C), with comparable burn-
off. The behaviour of the serie C is the expected one. Much
work has to be done in order to relate the nature of the
gasification reaction to the r.:resence of low pressure hyste-
resis.

4. ADSORPTION-ISOTHERM EQUATIONS
Because carbon forms rar:ge from non porous carbon such
as graphite to more complex systems like activated carbons
which also contain meso and rr.acropores, a brief critical
review of the mcst used adscrption equations developed for
non-porous as well as for microporous carbons is presented.

The knowledge of the starting assumptions used in each

equation is important for a good understanding of its 1 imi ta-
tions and correct use in physical characterization of porous
carbons.

A.- The Langmuir equation.

By means of molecular -kinetic arguments Langmuir
(1918) developed an adsorption equation which describes very
well the shape of type I isotherms (43) . The equation has
also been derived from thermodynamic principles (44) and
by statistical mechanical reasoning (45).

The starting point of the molecular -kinetic derivation

is the dynamic concept of adsorption equilibrium in which
the rates of adsorption and des,orption are equal. The following
simplifying assumptions are made.

i) only one molecule per site can be adsorbed on the

free surface of the adsorbent and saturation is reached on
completion of the monolayer.

ii) The surface of a solid is ccmposed of a two-dimensional

array of energetically homogeneoLs sites.

iii) The forces (repulsive or attractive) between the

adsorbed molecules c1re negligible in comp2rison to adsorbent-
adsorptive interactions.

The final equation can be expressed in various forms.

n BP
n 1 + BP
m
 147

where n is the amount adsorbed per gram of solid at equili-

brium pressure P,
n is the amount adsorbed per gram of solid to complete a
m
monolayer.
B is an adsorption coefficient depending on the adsor-
bate-adsorbEnt system.

In order to test that equation against experimental data

it is preferable to rewrite in the form

P 1 P
n B n
+ n
Illl
m m

When experimental data fit the Langmuir eqL:ation a linear

plot of Pin versus P is obtained, the slope of which gives
the value of lln m , and its intercept with the Pin axis, the
value of the term 1/n mB.

At very low equilibrium pressure or at relatively high

adsorption temperatures BP« 1, equation 1101 reduces to the
simple Henry's law:

n = n m BP 1121
At high equilibr ium pressure BP >"> 1, equation 110 I reduces
to n/n =1.
m
The linearity of Pin versus P cannot be taken as a proof
ei ther that the adsorption process follows the assumed mecha-
nism or that the assumptions used in the derivation of that
equation are actually true.

B.- The BET equation.

The best known theory of physisorption is undoubtedly
that of Brunauer, Emmett and Teller (46) who developed the
earlier Langmuir I s ideas to describe mul timolecular adsorption.
By introduction of some simplifying assumptions the BET theory
provides an extem;ion of the Langmuir monolayer model. The
first adsorbed layer is similar to the Langmuir model. Each
adsorbed molecule in the monolayer is assumed to be an adsorption
site for second layer molecules, and so on as the relative
pressure increases, until bulk condens&tion occurs. The energy
of adsorption of second and all higher layer molecules is
postulated equal to the heat of liquefaction of the adsorbate
at the adsorption temperature.

The reodel (19) for formation of an infinite number of

layers at the saturation pressure, Po leads to the well known
BET equation.
148

P 1 C-1 P
n(P -p) C t C 1131
n n P
o m m o

where n and nm are the amount adsorbed, at the equilibrium

pressure PiP, and the monolayer capacity respectively,
C is a cons~ant, which is related exponentially to
the first layer heat of adsorption by the equation

C = expl(q 1-q L)/RT 1

where q1 is the heat of adsorption of the first layer, qL'

is the heat of adsorption of the second and subsequent
layers (equal to the latent heat of vaporization of
the adsorptive). Therefore, q1 -q L is the net heat of
adsorption.

Equation 1131 gives a linear relation between P/n(P -p) and

P/P o ' with a slope equal to 1 (C-l)/nmcl and in intgrcept of
(lin C); thus values of nand C can be calculated. The stronger
the madsorbate-adsorbent fnteractions, the higher thE C-value.
As shown in Fig. 5, the C value greatly affects the shape of
the isotherm in the low relative pressure range, independently
of the n value (27,47).
m

Furthermore, as expected from equation 1131 the C value

1.5

0.5

o
o 0.2 0.4

Figure 5.- The influence of the value of the BET C constant on

the shape of adsorption isotherms, calculated according
to equation 13
 149

depends on the relative pressure range used. When the solid

is heterogeneous, as happens with an activated carbon, the
pip range used to determine C can be very important. In
the0 author's opinion this dependence should be borne in mind
when values of the C constant are used to deduce adsorbent-
adsorbate interaction Fig. 6 is an example (48). A value
of C is calculated for each experimental N2 adsorption data
pair on an activated carbon following the proposed method
of Dollimore (49). The plot of In C vs pip exhibits a minimum
near the nm value of the isotherm. 0

Equation 1131 gives an adequate description of type

II isotherms over a limited pip0 range only, usually wi thin
PIP = 0.05-0.30 due to the limi tations of some of the assump-
tiogs on which the BET theory is based. At relative pressures
below Pip = 0.05 the assumption that the surface is an energetically
homogeneo3s surface does not apply to the great majority
of adsorbents. The lack of success at higher relative pressures.
pip 0 > 0.35 is due to failure to take into account lateral
interaction between neighbouring adsorbed molecules.

Indeed strong limitations Exist when applying BET to

microporous carbons (type I isotherms), because the assumption
of multilayer adsorption in very fine pores is obviously
inappropiate. The physical meaning of n m has been very much
debated (26,50-53) and Dubinin's idea (26,29), about
the filling up of the micropore volume with the adsorbate
in a liquid like condition, seems much more reliable.

9 ~ In C

7
1.50

5
1.00

3 ~~--~--~--~--~~--~--~--~~0.50
0.0 0.2 0.4 0.6 0.8 1.0
p/po

Figure 6.- Dependence of CBET with P/Po range on an activated

carbon (Redrawn from 48)
 150

If the number of mul tilayers is restricted to a finite

number x then, using the same assumptions and an analogous
procedure, the BET equation becomes (37).

x x+1
n C(P/Po) 1-(x+1) (P/Po) + x(p/Po)
x+1
n l-(P/P o ) 1+(C-1)(P/P o ) - C(P/P o )
m
x being the total number of layers which, partly or fully
filled, can be adsorbed.

The complex distribution of size and shape of the pores

of activated carbons can be hardly satisfied by a constant
x, which is better a correction factor than an effective
mean number of adsorbed layer. Furthermore, McEnaney (54)
has found for an activated series of carbons that the x value
decreases wi th increasing activation. That apparently is
indicative of a decrease in pore width with '3.ctivation whicr.
is contrary to the known effect that activation increases
micropore and mesopores widths. Therefore equation \15\ is
seldom used.

In general, despite of its limitations and because of

its simplicity, equation \13\ is still used, even to analyse
adsorption in microporous adsorbents. Of course, the use
of BET equation should be applied to microporous carbons
with care.

Its use on these adsorbents is only recommended, as

suggested by Barrer (55), as a measurement of the monolayer
equivalent area (area which would result if the amount of
adsorptive required to fill the micropores were spread as
a close-packed monolayer of molecules). Figure 7 represents
a comparison of the equivalent monolayer, of a series of activa-
ted carbons (56), deduced from Langmuir, BET and DRK (Dubinin-
Radushkevich-Kaganer) (57). The DRK equation is a modified
form of the DR equation (see below, section 6C) by replacing
the concept of micropore filling by that of surface coverage.

log ~ '" - D 1 og 2(~)

P
n
m
where D is a characteristic constant.

C.- The specific surface area of solids.

Once the monolayer capacity is known from equation
\11\ or equation \13\, the specific surface area S(m 2 .g- 1 ) of the
adsorbent may be calculated, provided that the area effectively
occupied by an adsorbed molecule (Am) in the complete monolayer
is known.
 151

Cross-sectional
Number of molecules
Specific surface
in the monolayer
x area of each \17\
molecule

That is,

S = nm \18\
where the monolayer capacity has been expressed in mole/g
and NA is the Avogadro constant.

In general, the molecular surface area of a given adsorptive

may not be constant because it depEnds somewhat on the nature
of the adsorbent. In other words, the conventional picture
of Am value which corresponds to a monolayer completely filled
with adsorbate molecules in a liquid-like packing does not
correspond with reality. Problems related with the choice of
A m have been comprehensively summarized elsewhere (29,58).

Associated with this problem is also the large number

of gases and vapours which have been used as adsorptive, from
which only a small proportion fulfill the minimum conditions
to be suitable. The use of nitrogen, selected by Brunauer
(59) a long time ago, and of argon, desr-i te of the different
values of A which have been published (60-65), are recommended
m
as standard reference gaees (29).

500
"BET

3 -1
Figure 7.- Comparison of the monolayer capacity, n(cm.g ,
NTP) , deduced from Langmuir equation (n L ), from BET equation
(n BET ) a~d from D~binin-Radushkevich-Kaganer equation (n DRK ),
for a ser~es of act~vated carbons; (Redrawn from 56).
 152

5. CONCEPT AND USE OF STANDARD ISOTHERMS IN TEXTURAL CHARACTERI-

ZATION
A number of attempts have been made to express adsorption
isotherm data in a normalized from (66-70). For a large number
of non-porous solids a plot of nlnm against plP o can be represen-
ted by a single curve called the "standard isotherm". However,as
can be seen in Fig. 8 (71), some discrepancies appear between
the different published versions of "standard isotherms".
It is evident from the Fig. 8 that a single standard isotherm
can not be found and therefore some authors (71-74) have sugges-
ted the need of different common isotherms for different groups
of substances. These discrepancies may be caused because the
solid studied may not be free of pores and because the adsorption
energy for a given adsorptive is dependent on the adsorbent
structure.

Three different but related approaches can be used for

detecting deviations from the standard isotherms which will
now be considered.

A.- The t-Method of Lippens and de Boer (69).

Assuming that the multimolecular layer of adsorbed
nitrogen is formed freely on the surface, the thickness of
the layer increases with the relative pressure by the equation

t ~ 0.354 (n/n ) ~ f (pip)

m 0

where the value of 0.354 nm was used (75) as t for nitrogen,

Pierce
3
Cranston-In kley

o 0.2 0.6 1.0

Figure 8.- Comparison of different proposed standard isotherms

of nitrogen at 77K. (Redrawn from 29)
 153

b)
2.0

....
'"
c::
is
---c::
-
~
0
c
c

0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6
P/ Po P / Po

Figure 9.- Standard isotherm~ plots; a) t-method orO( -method

and b) Lecloux method: 1 - C >- 300; z - 300~ C ~100; 3 - 100 ~
C:?-40; 4 - 40>--Cb30; 5 - 30kC)/20. (Redrawn from 47).

assuming hexagonal close-packing of the molecules in the adsorbed

film. Thi s relation can be expressed in a universal t-curve,
as represented schematically in Fig. 9.

By ccrr;paring the nitrogen adsorption isotherm of a given

porous solid with the t-curve of a standard non porous solid
(replacing the relative pressure axis by a layer-thickness
axis) valuable information about porous structure of the adsorb~nt
can be obtained.

In the case of an unhindered adsorption the n-t-plot

is a straight line passing through the origin, the slope of
which is a direct measure of the surface area of the adsorbent.
However if the adsorbent under test is porous, deviations from
linearity will occur. Fig. 10,a shows schematic t-plots for
different types of solids.

An upward deviation (curve 2) is indicative that capillary

condensation takes place in mesopores at high relative pressures,
so the edsorption increases much more than the standard non-
porous solid. If a downward deviation is observed (curve 3),
micropores or slit-shaped pores are present in the solid.

Fig. 10 a:, also shows how the t-plot can be used to obtain
information about surface homogeneity (76-77). The negative
deviation at low pressure has been related to a homogeneity
factor (H).

The t-method has been widely used in the li terature to

 154

a) b)

lIIesopDrDs i ty

'- 1000
free surface

....
I

C')
"!I
E

--c:
()

500

.
c:

5 10 15
t

Figure 10.- a) Schematic t-p1ot shapes showning the deppendence

on the adsorbent porous texture.
b) t-plots applied to a Saran carbon activated
to different % burn-off; n as cm 3 .g- 1 (NTP).
(Redrawn from 81).

analyse mi croporous carbons (78-80). Micropore volume, external

surface area and total surface area are the data more often
calculated by this method. The intercept on the adsorption
axis of the extrapolated linear branch gives the micropore
contribution while the slopes (OC) and (AB) give the total
and external surface area, respectively (see Fig. lOa).

Fig. lOb illustrates the variation on the porous texture

of a Saran Carbon and the surfac e area deduced from the t-
method as a function of the extent of the activation (81).

However, the earlier use of the t-method for adsorption

on a microporous soli d has been criticized by sEveral authors
because it is assumed that monolayer adsorption takes place
on the walls of micropores in the same manner as on a non-
porous solid (24,71).

B.- The a-Method of Sing (78,82).

A different way of comparing an experimental isotherm
to a standard one is used by Sing. Instead of the thickness
of the adsorbed layer, he proposed the ratio n/nO. 4 ' so-called
a, where nO. 4 is the amount adsorbed by a non porous reference
solid at PIP o = 0.4. PIP o = 0.4 was justified with nitrogen
 155

isotherms at 77K since monolayer coverage and micropore filling,

if present, is complete, and capillary condensation has not
yet begun. By plotting a versus PIP o he defines a new tYIJe
of reduced standard isotherm.

The a-curve is used in a manner closely analogous to

that for the t-curve. If there is neither capillary condensation
nor micropore filling on the test sample the a-plot is a straight
line through the origin. The equation to the line may be written

n =b a 1201
where b is the slope of the straight line.

Since the surface of the standard is known (usually,

but not necessary, from the BET nitrogen isotherm) the specific
surface (5 ) of the test sample can be obtained by direct
comparison a of its b value with that of the reference isotherm:

n O• 4 (sample) 5 (sample)
Slope (sample) a
1211
Slope (reference) n O. 4 (reference) S (reference)

Slope (sample)
5
a
(sample) x S (reference) 1221
Slope (reference)

Sing claimed several advantages of the a-Method over

the t-Method. i) It can be used even when the standard isotherm
does not possess a sharp knee, as happens with type III iso-
therms. ii) Its applicability is not restricted to nitrogen
as adsorptive. iii) The a-Method can be independent of the
BET-Method.

On the other hand, there is a noticeable difference between

the t- and the a -method about the choice of the standard
isotherm. Sing chooses the standard isotherm according to
the chemical nature of the sample to be studied, while for
de Boer the t-curve is almost indepent of the nature of the
non porous- adsorbent.

C.- The Method of Lecloux (47,83).

Lecloux introduced a general criterion in the selection
of standard isotherms which is different from a-method; the
thickness of the adsorbed layer on a non porous solid is not
only dependent on the relative pressure but also on the adsor-
bent-adsorbate interactions. Consequently, the standard isotherm
must be chosen according to these interactions expressed by
the C-BET constant value.
 156

a) b)
200 600

-. I
m 100 -I
~
400

-
"'e" "'"e
-- c c
200

0
I I I I I I I I
2.5
I
0 0.2 0.6
... 1.0

--
0 0.2 ~.4 0.6 0.8
t nm)--+
I I I I I I I I
0 0.4 0.8 1.2 1.6
.,
Figure 11.- a) Comparison of the t, a and Lecloux method showing
the scale factor on the X-axis. (Reduced from 47)
b) a-plots for C02 adsorption at 195K on I) an ac-
tivated carbon, II) the same carbon heat treated
to 1800 0 C and III) the 2000 0 C heat treated carbon
used as reference material. (Reduced from 54).

From the results they derjved five standard isotherms

characterized by their C values which are illustrated in Figure
9.

The main advantage claimed by Lecloux is that the whole

set of standard isotherms obtained from nitrogen adsorption
is also suitable for other adsorbates.

If a correct choice of C-BET constant value is taken

for the standard isotherm, then a straight line passing through
the origin is found. On the other hand, if the right reference
data are not used, the l:ine makes either a posi ti ve or negati ve
intercept with the n axis.

According to Lecloux if the right standard isotherm is

chosen, the only difference between the plots of t-method,
a-method and Lecloux method (L) is a scale factor on the
X-axis. Fig. 11,a shows a comparison of the t, a and L methods
for nitrogen adsorption on an amorphous silica-alumina and
its scale factor on the X-axis.

This method is also subjet to some limitations which

have been pointed out by Sing (29) j the presence of micropores
 157
distorts the type II isotherm increasing the C-BET constant
value. In such cases the value of C may not be a guide to
follow the course of the isotherm on the external surface.

The use of standard isotherms as a tool to characterize

the porosity of solids (even microporous solid) by means
of one of the three methods above described may be appropriate.
However, it is necessary to keep in mind tha't all of them
are subject to the same limitation; that is, the appropriate
choice cf a non-porous reference material. Graphitized carbon
blacks have been used as star.dard reference material but
their structure is quite different from most microporouE
carbons. McEnaney et al. (54,84) have used a heat treated
microporous carbon as an alternative reference material;
heat-treatment loses off microporosity. This approach merits
further consideration because it may be very useful when
applied to microporous carbons. The authors claim that the
non-microporcus surface of a high heat-treated carbon will
more closely resemble that of an open microporous carbon
than does that of graphitized carbon black. Fig. 11,b represents
the a-plots for an activated carbon prepared from cellulose
triacetate and for the same carbon after heat treatment to
1800 0 C and 2000 o C; the 20000C carbon was used as reference
adsorbent. The a.-plots show clear evidence for microporosi ty
in the active carbon which is closed off by 1800°C, since
the later a.-plot is similar to a non porous carbon. It is
interesting to note that McEnaney I s method for selecting
a reference material is quite different from that of Lecloux,
for whom the most important point in selecting a reference
material is the C-BET constant. Of course, further work is
needed to clarify this important point.

6. PECULIARITIES OF ADSORPTION IN MICROPOROUS CARBONS.

As it was stated earlier adsorption in microporous solids
is not so well understood as on non-porous or mesoporous
solids. Pore sizes of the same order of magnitude as the
sizes of the adsorbate molecules lead neither to the progressive
completion of a monolayer nor to multilayer adsorption but
to the filling up of the micrope,re volume with the adsorbate
in a liquid like condition' (24,26). There are a number of
problems associated with adsorption in micropores including
the following important points:

i) BET surface areas are unr ealistically high (79,85,86)

compared with the calculated area for one gram of carbon
in the form of an extended graphite layer plane, counting
both sides (2630 m2 /g) (87); e.g. see Fig. 10. Also urn values
obtained either by BET or Langmuir equation, are similar
to the limiting uptake of the type I isotherlT1s, which impl ies
that micropore filling is occurring rather than surface area
 158

coverage of the adsorbent.

ii) Adsorption energy within micropores of a given chemical

nature is much higher than on a non-porous solid of the same
chemical nature. This enhancement of the adsorption potential
is shown in Fig. 12, which compareE. the net heat of adsc.rption
curves c.f a carbon black with that of an active carbon (26).
The net heat of adsorption on the active carbon is three
times higher than that on the carbon black.

iii) The adsorption of benzene and water vapc,ur (88)

(expressed as liquid adsorbate per gram of adsorbent as a
function of the relative pressure) on graphitized and on
ungraphi tized charcoal are presented in Fig. 13. These results
provide an excellent example confirming that the overall adsorp-
tion process on microporous carbon is occurring by a volume
filling mechanism rather than surface coverage. Note that
the liquid volumes of the two adsorbates taken up on the
same adsorbent are really coincident, even though benzene
and water vapour adsorption give type I and type V isotherms,
respectively.

Associatec with point iii) is the fact that very often

the Gurvitsch rule is obeyed by systems showing Type I isothErms
(29,89,90). In other words, the amounts of different adsc:rptives

10

o~~--~--~~--~
0.2 0.6

Figure 12.- Comparison of the net heat of adsorption (qst_~ ;Kcal

mole-I) for a) an active carbon and b) a carbon black; nO 2 a p is
the adsorption at P/P o = 0.2 (Redrawn from 26). .
 159

a) b)

0.5

Ori 9 inal
M

--
::; 0.3 F":'=;':':':;=;;:J

c:

0.1

1 0.2 0.6
P / Po

Figure 13.- Adsorption isotherms of a) water vapour and b) benzene

on graphitized and ungraphitized charcoal (Redrawn from 88, with
omission of experimental points).

(expressed as volume of liq~id) taken up at a relative pressure

near unity, by a given microporouiO. adsorbent are very similar.
Furthermore, the amount of liquid adsorbate at saturation
is similar to the total pore volume obtained by a completely
different method, that is measurements of apparent density
of the adsorbent, by immersion in mercury ar:d some other suitable
fluid (e.g. helium) (86,91).

iiii) Adsorption isotherms sl-·ould always represent

the equilibrium amc·unt adsorbee as a function of the equilibrium
pressure. However, with some micropcrous materials like coals
and carbonized chars, where the widths of the pore entrance
are commensurate with the diameter of the adsorptive molecules,
the adsorption equilibrium is very difficult to reach (92-
93) with very long equilibrium times.

In these cases, contrary to the exothermic nature cf

the adsorption process ,an increase in the temperature of adsorp-
tion leads to an increase of the amount adsorbed due to activated
diffusion (94-96). The pore entrance which may be narrower
than the cavity presents an energy barrier to the passage
160

of the molecules through the constrictions (28). If the adsorp-

tion temperature is low, the rrolecules will have insufficient
kinetic energy; an increase in the temperature of adsorption
will have an irrnnediate effect on the rate of approach to equili-
brium. The number of molecules entering to the cavity during
the adsorption equilibrium time will, therefore, increase
with rise of temperature.

Figure 14, take'n frorr. reference 54, shows the influence

of the adsorption temperature upon the micropore volume (deduced
from the DR equation, see below). Note that the micropore
volume of carbon A is independent of adsorption temperature
whereas the micropore volume of carbon B increases wi th the
adsorption temperatur e (97). This indicates that the micropores
of carbon A are wide enough to adsorb CO 2 molecules at 195K
(minimum dimension 2.8 R), but the micro{::ore size of carbon
B is narrower than carton A.

iiiii) Owing to the different pore size distribution

which can be found on a microporous carbon, the adsorption
of adsor~,tives with different molecular size may be selective
according to both shape and size. This phenomenon of molecular
sieve action (12), can be exploited as a method for measuring
micropc,re size distributions. This method, terrried molecular
probe technique, will be briefly discussed later.

A.- Adsorbate-adsorbent interaction in micropores.

A marked increase in the dispersion energy is predicted
for a molec~le which is able to enter into a cavity whose
size is similar to molecular dimensions (98,99).

M<:J> 0.4 .........- - - -••- - -...1--- A

--
E
u

,.o
~ 0.2

~ ~B
.....E
ao:
Cl
o
200 220 240
Adsorption Temperature/ K

Figure 14.- Effect of activated diffusion on narrow-pore

carbon B (Redrawn frorr, 54).
 161

Detailed calculations of the potential energy in very narrow

pores indicate that there is an overlap of the force fields
of opposite pore walls which will enhance the energy of interac-
tion of aJ" adsorbate molecule comr,ared with that correspcnding
to adsorption on a plane surface (87,100,101). The enhanced
adsorption potential in the rr.icropore (¢m) relative to that
in a free surface (¢f) depends, alT'or:g other factors, on the
ratio of the size of the pore to that of the adsorbate molecule,
as well as on the pore model used for calculations. Both effects
on the relative potential energy profiles (¢m!¢f) can be seen
in Fig. 15 (87). The curves correspond to the variation of
(¢m!¢f) as c, function of the relative size (x/r o ) of the pore
to that of the molecule for slit-shaped pore (curve a) and
cylindrical pore model (curve b); x is the half-width of
the slit-shaped pore and/or the radius of the cylinder in
a cylindrical pcre model, and ro is the collision radius of
the molecule.

From the Fig. it is clear than i) the enhancement of

the pc·tential energy (¢m!¢f) is considerably greater for cylin-
drical pore model thaD for slit porES, ii) the (¢m/¢f) persists
for much higher value of x/ ro on the cylindrical pore model
than on the slit model (the enhancement disappears at x/ro > 3 and
x/ro ..,. 2 respectively) iii) the ratio of the size of the pore
to that of the adsorbate molecule may be more critical parameter
than the pore size itself, that means that the pore mayor
not act as micropore depending on the diameter of the adsorptive
molecule. Pores with enhanced adsorp tim potential have been

2 3

Figure 15.- Enhanced potential energy, a) for a slit-shapEd pore

and b) for a cylindrical pore (Reduced from 87).
162

termed ul tramicropores while supermicropores is used for larger

micropores (26).

All the aforementioned points about peculiarities of

adsorption in microporous carbons shoW that a special atten-
tion is needed when characterising microporous carbons, such
as activated carbons and coals, by adsorption methods.

B.- The Polanyi Potential Theory

A major development in understanding adsorption of
gases and vapours on microporous carbons was provided by the
potential theory of adsorption of Polanyi (98,102).

The potential theory assumes that at the adsorbent surface

the molecules of gases are compressed by attractive forces acting
between the surface and the molecules and these forces of
attraction decrease with increasing distance from the surface.
The force of attraction at any given point near the surface
is measured by the adsorption potential £, which can be defined
as the work done by the surface adsorption forces on transferring
a molecule from the gaseous phase to the given point above
the surface.

Polanyi described the adsorption space as a series of

equipotential surfaces, each with a given adsorption potential
£i, and each enclosing a volume, Wi. As one moves away from
the surface the values of adsorption potential decrease until
it falls to zero and the adsorption space increases up to
a limiting value, Wo (zero potential). At the surface, W = 0 and
£i = £max The building up of the volume enclosed wi thin
the adsorption space may be described by a function of the
type £ = f(W).

Polanyi assumed, that since dispersion and electrostatic

forces (30) are independent of temperature that the adsorption
potential at constant volume filling is also temperature-indepen-
dent,

o \23\

Equation 23, means that the curve £ =f(W) should be the

same for a given gas and a given adsorbent at all temperatures.
This relationship between £ and W is called the Characteristic
curve.

Polanyi expressed the adsorption potential £ for a volume

filling W as the amount of work necessary to compress the
adsorptive from its equilibrium pressure, P 1 , to the compressed
adsorbate pressure, P2.
 163

£ = j P2

P
1
RT
--
P PI
P2
d P = RT In - -

Thus, £= -~G, where 6G is the equivalent free energy change.

The state of the compressed adsorbate in the adsorption

space depends on the temperature. Polanyi distinguished three
different cases. i) when the temperature is well below the
critical temperature of the adsorption, Tc' the adsorbed vapour
may be considered as liquid-like, ii) when the temperature
is just below the Tc, most of the adsorbate will be as liquid-
like, but also the adsorbate may be as compressed gas (for
those layers which have less £ than is necessary for compressing
the vapour to liquid) iii) when the temperature is above the
critical temperature, the adsorbate will be as compressed
gas.

The first case is, by far, the most used one. Therefore
the adsorption potential will take the form
Po
e: = RT In Cp ) 1251

Equation 25 now represents the work needed for compressing

the adsorptive from its equilibrium pressure in the gas phase
to its saturated vapour pressure, Po'

From equation 25, since it is assumed than the liquefied

adsorbate is incompressible and has the normal density of
the liquid at the given adsorption temperature, it is possible
to obtain the volume of filled adsorption space by

w = ~= n.v 1261
p m
where n is the amount adsorbed, expressed in moles j M is the
molecular weight of the adsorptive and p and vm are the liquid
density and the molar volume of the adsorptive at the adsorption
temperature, respectively.

The temperature-invariance of the adsorption potential,

which is the fundamental postulate of the Polanyi's theory,
has been widely proved, mainly by Dubinin and Co-workers (24,26,
103). Isotherms for a given vapour on a given adsorbent at
different adsorption temperatures can lie on a single curve
for many adsorptive-adsorbent systems. An example, is represented
in Fig. 16 (26,104) which illustrates the temperature invariance
of adsorption of benzene on an active carbon at five different
temperatures.

Dubinin and Co-workers during the course of their extensive

 164

20°C
0.5 o 50°C
o 80°C
0.4 x 11 0 °C
• 140°C
.-;
I
"!' 0.3
C'l
E
0
-....
0.2

0.1

0 5 10
E ,Kcal.mole- 1

Figure 16.- A Characteristic curve of benzene adsorption on

active carbon (Redrawn from 26).

studies of active carbons have developed the so-called Theory

of volume Filling Micropores (TVFM) (24,26). Some of the relevant
equations in the development of that Theory will be briefly
discussed.

C.- The Dubinin-Radushkevich equation.

Based on numerous experimental data Dubinin and Collabo-
rators (24) have added a second postulate, which complements
the PolanYi Theory: for an identical degree of filling of the
volume of adsorption space, the ratio of adsorption potentials
for any two vapours is constant.

1271
S, called the affinity (also similarity) coefficient, should
be regarded as the relative differential molar work of adsorption
of a given vapour (£) relative to the differential molar work
of adsorption of the vapour chosen as standard (£0); for benzene
(chosen as standard adsorptive by Dubinin) S = 1. The affinity
coefficient is closely related to the ratio of the molar volume
(105), or of the parachors for the liquid-like phases of
the vapour and benzene (106).
 165

According to experimental data and assuming that the

pore size distribution is Gaussian, Dubinin and Radushkevich
(107) arrived at an expression which relates the degree of
filling of the micropores, (e), with the differential molar
work of adsorption

e=-w- = exp
Wo
where K is a constant dependent on the pore structure. By
combining equations 1251 and 12sl, the well known DR equation
is obtained:
P
W Wo exp 1-
K
(RT In
0 )2
S2 P

For plotting, equation 1291 may be transformed into

log W = log Wo - D log

2 P
(-2)
P
where D = 2.303 K (RT/S)2 ; a plot of log W against log2(Po/p)
will be a straight line having an intercept equal to the total
micropore volume Wo. From its slope the value of D is obtained.
Because it is related to K and so to the microporous structure
of the adsorbent, D decreases as the size of the pores decreases
(lOS) •

The agreement between experimental and equation 1301

is substantially oeserved in the literature. Fig. 17 is an
illustration, from Nikolaev's data (109), where experimental
results, obtained at different adsorption temperatures and
wi th different adsorptives, fit on straight lines over a very
wide range cf relative pressures and all intersect the ordinate
axis at one point, indicating the constancy of the limi ting
volurre of adsorption space, WOo

By introducing S into the potential equation, all the

characteristic curves of different adsorptives, on a given
adsorbent, should coincide. This is illustrated by a typical
example in Fig. IS, for two narrow-micropores activated
carbons (110).

Unfortunately deviations of the DR equation from linearity

are often observeg (111); these deviations are due to the
w.icroporous structure itself; to the adsorbent-adsorbate
system, to the range cf P/P o studied, etc. Different types
of deviations are sketched in Fig. 19, together with a straight
line (I) fitting the DR equation. The DR plot (II) displays
an upward turn as saturation pressure is approached. This
is explained in terms of multilayer adsorption and capillary
condensation in mesopores. If the contribution of mesopcres
166

10 20 30 40 50
o~--~--~--~--~--~-----
1 nitrogen
2 krypton
3 xenon
4 tetrafluoroethylene
-1.0 5 hexafluoropropylene
(V")

"'"
C1
o

Figure 17. - DR plots on an active carbon; E, erg/mole. ( Redraw

from 109) •

o N2 77 K
-2 A Xe 323 K
• CSHS 273-323 K

-4

-6

figure 18.- Adsorption of various gases and benzene by carbosieve

carbon; y=(T/a)2 log2 (Po/P). (Reduced from 110).
 167

...,...
III
...,
.
.....
.&>

III v

\ ~"
~ "

Figure 19.- Schematic representation of a DR plot and some

comrron types of deviations.

appreciably exceeds 50 m 2.g -1, Dubinin (24) claims that the

extrapolated value of micropore volume would then need to be
corrected for mesopcre adsorptior: which would have cor:tributed
to the uptake at lower relative pressures.

Occasionally the DR plot gives two straight lines

(III) as is frequently observed for coals, carbonized materials
and activated carbons with very fine micropores (54,92,111-
113). It is then not clear whether the micropore volume is
obtained by extrapolation from the linear low-pressure rart
or the linear high-pressure part of the DR plot. Such deviations
can be linearized to some extent by using the Dubinin-Astakhov
equation with n > 2 (see below). Since the downward deviation
at low pressures can be eliminated by extending the equilibration
times, it has been attributed to activated diffusion in the
narrow pores of the adsorbent (54).

Sometimes the DR plot is convex to the log2(Po Ip) axis

ct high-pressures while the low-pressure part is linear
(curve IV); this type of deviation is typical for active
carbons with high degree of activation and, consequently,
with a wide micropore distribution. Dubinin and Polstyanov
168

(104) visualize that type of microporous structure as a superpo-

si tion of two different microporous structures, each one
wi th different values of E: and Wo. Their two term equation
is:

W= W
o 01
exp [ - (K
1
E:
01
) 2 1+ W
02
exp [- (K
2
E:
02
) 2I
'
\31\

where 1 and 2 correspond to the first and second microporous

structure respectively.

DR plots like curve V may also be observed which fit the DR

equation fairly well at high pressures but with an upward
deviation at low-pressures (110) . This type of deviation
is also associated with strongly activated carbons with a complex
microporous structure (heterogeneous system of micropores).

Finally, intermediate to DR plots IV and V and related

to them, is plot VI; in this case the experimental data do
not fit the DR equation at all. Occasionally, some Saran
active carbon with a very high level of activation may exhibit
this behaviour (81).

It is evident that when such deviations occur the extrapola-

tion to the ordinate to obtain Wo becomes uncertain. In attempts
to overcome this problem and to extend the DR equation a
number of more general equations have been put forward.

D.- The Dubinin-Astakhov equation.

Dubinin and Astakhov (114-115) assuming a Weibull
distribution of pore sizes, rather than a Gaussian, obtained
the expression:
E:
9 exp \ _ ( _ ) n \32\
E
W

where n is a small integer (n < 4-6); E is a characteristic

free energy of adsorption, equivalent to E: when 9 = 1/ e. By
combining equation \32\ and \25\

W
o RT n n Po
9 = exp \ - ( - ) ln (-) \ \33\
E P
W

Equation \33\ may be transformed into

p
n
log W = log Wo-D' log (~)
P
\34\
 169

where D I = 2.303 n-1 (RT /E) n . I t follows from equation 1331

that the DR equation is a special case, when n=2.

The exponent n mainly affects the breadth and the mode

of the Weibull distribution function as can be seen in Fig.
20, a (deduced from (26), for E=5 Kcal. /mole); a small value
of n denotes a heterogeneous micropore structure, with a
broad distribution of the adsorption volume over the free
adsorption energies. In this case, a rational approximation of the
process is given by the two-term adsorption equation (31),
which leads to a differentiation of the micropore structure
into micro and supermicropores (116).

The influence of E on the Weibull distribution is illustra-

ted in Fig. 20,b, for n=3; the lower the E value is, the
more homogeneous is the microporous structure.

Transformation of equation 1321 gives:

Wo
In In(-) = n In (£) - n In(E)
W
A plot of In In (W /W) against In (£) enables n to be determined
as well as E. Us<Gally the n value obtained is rounded to
an integer; W may be obtained from the "plateau" of a type
I isotherm or by application of the DR equation.

For microporous carbonaceous adsorbents the n parameter

is usually equal to 2 (116). Only for active carbons with
particularly fine micropores can n be 3.

It has been suggested (117-118) that non-integral values

of n can provide a better fit of experimental results to

aJ bJ

0.4 0.4
E= 5 Kcat/,"ote

'"
0
,..< '"
0
,..<

.., .;,
-0 -0
...... 0.2 ...... 0.2
<D <D
.,; -0

E ,Kcal.mole- 1

Figure 20.- Influence of a) n and b) E upon the slope of the

Weibull's distribution function (Redrawn from 26).
170
equation 1331, in particular for carbons with high degrees
of burn-off. However, no clear correlations have been found
between the degree of activation and the variable n (110).
Furthermore a physical meaning for a non-integral value has
not been proposed. On the other hand, Dubinin and Collaborators
(116) assigning fractional values of the n parameter (from
n=1.5 to 2.5) have observed no improvement in the agreement
between the experimental calorimetry data and the heats of
adsorption deduced from the adsorption isotherm (see below,
equation 1401 and Fig. 21). On the contrary, a better agreement
between the calculation results and experiment is found for
n=2.

E.- The Stoeckli equation.

Stoeckli and Col. (119-120) have pointed out that
equation 1291 is followed only by activated carbons with
molecular sieve properties, if wide ranges of temperature
and relative pressure are used. However for strongly activated
carbons with a heterogeneous collection of micropores, the
overall adsorption isotherm is a sum of contributions from
individual pore groups.

W L W . 1361
o,J
j

Integrating the summation and assuming a normal uaussian equation

for the distribution of the micropore volume with respect
to the B parameter of the DR equation, Stoeckli obtained.

2 2 1
W(y) wo expl-Boyl expl y 21:!. 1 "2 11-erf(x) 1 1371

where x = (y-B /1:!.2)

o
MI2 y= IT/f31og(P /P)
o
12 and erf(x)= error
function.

The three parameters of Stoeckli equation are: W (micropore

volume), Bo (a structural constant) and I:!. (the 0 spread of
size distribution).

F.- Entha1pies of immersion of microporous carbons.

Stoeckli and Kraehenbuehl ( 121-122) have recently
provided important experimental support for the TVFM. They
have established a link between the enthalpy of immersion
( I:!. H.) and the parameters of the microporous carbon obtained
fromlDubinin Theory.

When an outgassed solid is immersed in a liquid, heat

is evolved. In the case of microporous solids, without external
 171
surface area, the enthalpy of immersion is related to the

O:.:d:::"~ionr~~:S(:;:;~'_'H
isosteric heat
(T)
l)~ vap
As shown by Dubinin, the net differential heat of adsorption
net st
q q -tlH
vap
can be derived from DA or DR equations; i.e. for n=2 the
resulting equation is:

1 ) Yo
q net(T.·. e) =,..,DEI
0 (1 ns 2
+

where a is the thermal expansion coefficient of the adsorptive

at temperature T. This equation permits calculation of the
differential heat of adsorption from one isotherm of adsorption,
for the region of temperature invariance of E. As an example,
Fig. 21, deduced from Dubinin (24), compares differential
heats of adsorption of benzene (continuous curve) computed
by equation (40) with experimental ,isosteric heats of adsorption
(denoted by circles) for an active carbon.

The contribution of equations 1381 and 1401, followed

by integration (122), leads to the relation

-I
Q/kcal..ole
20

15

o
10 '--_ _...I-_ _---L_ _ _- ' -_ _- ' - _
2 3 4
n/ •• ole.g- I

Figure 21.- Comparison of experimental(continuous curve) and cal-

culated (circles) differential heat of adsorption on an activated
carbon (Redrawn from 24).
172

where vm is the molar volume of the liquid adsorbate.

Experimental enthalpies of immersion of benzene for

10 activated carbons, measured with a Cal vet type calorimeter
and corrected for external surface area, were compared with
the calculated from equation 1411. Fig. 22,a clearly shows
the validity of the equation. On the other hand, Fig. 22g,
illustrates the proportionality which exists between AH. (J!cm )
and the characteristic energy E of the correspondinl carbon
and, consequently, its dependen~e on the micropore system.

G.- Independent experimental assessment of microporosity.

It has been shown previously the characterization
of microporous solids by means of gas adsorption is complex.
All the above theory related to the Polanyi Potential Theory
may also be submitted to some criticisms (12,29,123). Perhaps
the most direct method of evaluating microporosity is to
fill up only the micropores with some suitable adsorbate
(i.e., n-nonane). The external surface area (meso, macropore
and open surface) is measured by gas adsorption; i.e. by
ni trogen at 77K (48,124). A few preadsorbates have been used
(125-127). The main advantage of the preadsorption method
is that micropore volumes are evaluated without any previous
theoretical assumption and without the need to use any adsorption
equation.

On the other hand, a reliable method of obtaining micropore

size distributions from a single adsorption isotherm is still
lacking. A number of attempts, have been made (78, 128, 129),

-AH j exp./Jg-l -A H./J c m-3

400 I

150

300
100

50 -A H j calc./Jg- 1 Eol KJmol- 1

I I
200 , , ,
50 100 150 16 20 24 211

Figure 22.- Enthalpies of immersion of microporous carbon

into benzene. a) Relationship between calculated (equation)
and experimental data. b) Relationship betweem f1 H. and E
1 0
(Redrawn from 122).
 173

however much more work is required. A more realistic micropore

size distribution can be obtained by the laborious method
of using different molecular size adsorbates. Differences
in uptake with the molecular dimension of the adsorbate may
reflect a micropore size distribution of the adsorbent (130).
Examples of application of the preadsorption technique and
of the molecular probe technique in the characterization
of activated carbon will be dealt with in chapter 7 of this
volume.

7. EXTERNAL SURFACE AREA OF MICROPOROUS CARBONS

A microporous carbon should be characterized mainly
by its micropore volume. However, in the case of some activated
carbons and coals their external surface area (consisting
of meso-and macropores) may be important. Furthermore, some
of the equations used in section 6E and 6F should be corrected
for external surface area contribution (26,121). Consequently
an assessment of the external surface area is also needed.

Analysis of adsorption isotherms are useful for this

goal, however a major difficulty is to obtain a suitable
separation of adsorption in micropores from that on the non-
microporous surface. N2 adsorption at 77K on prefilled micropore
volume (i.e., n-nonane) has been compared with many techniques,
including immersion calorimetry (127).

Space does not allow a detailed consideration of other

suitable attempts to separate adsorption in micropores from
non-microporous adsorption, like the t/F method of Dubinin
and Kadlec (131) or the more direct method proposed recently
by Ali and McEnaney (132), and termed the isotherm subtraction
(1.S.) method.

* * * * * * * * * *
It has been shown from physical adsorption measurements
that it is possible to get enough information to give a coherent
picture of the porous texture of a carbon solid. A summary
of the different types of information, obtained by physical
adsorption measurements, and the way they are connected to
the methods of analysis are summarized in Fig. 23, which
is slightly modified from the original idea of Lecloux.
174

Figure 23.- Different types of information obtained by physical

adsorption data and the way they are connected to
the methods of analysis.

Aknowledgements:

The author wishs to thank Dr. B. McEnaney for his English

assistance.

REFERENCES

1. Walker, P.L., Jr., Rusinko, F., Jr. and Austin, L.G.; in "Advances in
Catalysis", Vol. 11 (Eds. D.O. Eley, P.W. Selsood and P.B. Weisz), Academic
Press, New York, p. 164 (1959).
2. Ergun,· S. and Meutser; H.; in "Chemistry and Physics of Carbon", Vol.
1 (Ed. P.L. Walker, Jr.), Dekker, New York, p. 2D3 (1966).
3. Jenkins, R.G., Nandi, S.P. and Walker, P.L., Jr.; Fuel 52, 288 (1973).
4. Hippo, E. and Walker, P.L. Jr.; Fuel 54, 245 (1975).
5. Tomita, A., Mahajan, D.P. and Walker, P.L., Jr.; Fuel 56, 137 (1977).
6. Seledade, L.E.B., Mahajan, D.P. and Walker P.L., Jr.; Fuel 57, 56 (1978).
7. Mahajan, D.P. and Walker, P.L., Jr.; in "Analytical Methods for coal
and· coal products". (Ed. C. Karr, Jr.) Academic Press, New York, Vol.
2, p. 465 (1978).
 175

B. Linares-Solano, A. , Mahajan, O.P. and Walker, P. L., Jr. ; Fuel 58, 327
( 1979) •
9. Walker, P. L., Jr. ; Matsumoto, S. , Hanzawa, T. , Muira, T. and
I. M.K.; Fuel 62, 140 (19B3).
10. Hassler, J .W.; "Activated Carbon". Chemical Publishing Co. I n c. , New
York (1963) •
11. Smisek, M. and Cerny, S.; "Active Carbon Manufacture, Properties and
Applications". Elsevier, New York (1970).
12. "Porous Carbon Solids". (Ed. R.L. Bond) Academic Press, New York (1967).
13. Jenkins, R.G. and Walker, P.L. Jr.; Carbon 14, 7 (1976).
14. Janosi, A. and Stoeckli, H.F.; Carbon, 17, 465 (1979).
15. Fryer, J.R.; Carbon 19, 431 (19Bl).
16. Giles, C.H., D'Silva, A.P. and Trivedi, A.S., in "Surface Area Determination"
p. 317. Butterworths, London (1970).
17. Giles, C.H., Smith, D. and Huitson, A.; J. Colloid Interface Sci., 47,
755 (1974).
lB. Mattson, J.S. and Mark, H.B. Jr.; "Activated Carbon". Marcel Dekker,
New York (1971).
19. Brunauer, S.; "The Adsorption of Gases and Vapours". Princenton University
Press. Princenton 1945.
20. de Boer, J.H.; "The Dynamical Character of Adsorption". Oxford University,
London (1953).
21. Young, D.H. and Crowell, A.D.; "Physical Adsorption of Gases". Butterworth,
London (1962).
22. Ross, S. and Olivier, J.P.; "On Physical Adsorption", Interscience Publishers
London (1964).
23. Gregg, S.J.; "The Surface Chemistry of Solids". Chapman and Hall, LTD.
London (1965).
24. Dubinin, M.M.; in "Chemistry and Physics of Carbons". Vol. 2, (Ed. P.L.
Walker, Jr.) M. Dekker, New York, p. 51 (1966).
25. Flood, E.A.; "The Solid-Gas Interface". Vol. 1, M. Dekker, New York (1967).
26. Dubinin, M.M.; "Progress in Surface and Membranes Science", Vol. 9, (Ed.
Dacadenhead) Academic Press, New York (1975).
27. Adamson, A.W.; "Physical Chemistry of Surfaces" 3::i edition. J. Wiley
and Sons, New York (1976).
28. Gregg, S.J. and Sing, S.K.W.; in "Surface and Colloid Science", Vol.
9, (Ed. E. Matijevic) J. Wiley and Sons, p. 231 (1976).
29. Gragg, S.J. and Sing, S.K.W.; "Adsorption, Surface Area and Porosity",
2~ edition, Academic Press, London (1982).
30. London, F.; Z. Physik, 63, 245 (1930).
31. Lennard-Jonas, J.E.; Physica 4, 941 (1937).
32. Sing, K.S.W.; in "Characterization of Powder Surfaces" (Ed. G.D. Parfitt
and K.S.W. Sing) Academic Press, London (1975).
33. Barrer, R.M.; "Zeolites and Clay Minerals as Sorbents and Molecular Sieves".
Academic Press, p. 174, New York (1978).
34. Kiselev, A.V. and Yashin, Y.I.; "Gas-Adsorption Chromatography". Plenum
press. New York (1969).
35. Curthoys, G., Davydov, V.Y., Kiselev, A.V., Kiselev, S.A. and Kuznetsov,
B.V.; J. Colloid Interface Sci., 48, 58 (1974).
176
36. Sing, K.S.W.; Pure and Appl. Chern., 54, NQ 11, 2201 (1982).
37. Brunauer, S., Deming, L.S., Deming, W.S., and Teller, E.; J. Amer. Chern.
Soc. 62, 1723 (1940).
38. de Boer, J.H.; in "The Structure and Properties of Porous Materials"
(Ed. Everett D.J. and Stone F.S.), p. 68, Butterworths, London (1958).
39. Arnell, J.C. and Mc Dermott, H.L.; in "Proceeding of the Second International
Congress on Surface Activity", II, p. 113, Butterworths, London (1957).
40. Bailey, A., Cadenhead, D.A., Davies, D., Everett, D.J. and Miles, A.;
Trans. Faraday Soc. 67, 231 (1971).
41. McEnaney, B; Trans. Faraday Soc. 70, 84 (1974).
42. Martin-Martinez, J.M., Garrido, J., Molina, M., Torregrosa, R., Linares-
Solano, A. and Rodrfguez-Reinoso, F.; Carbone 84. Bordeaux, France, p.30
(1984).
43. Langmuir, I.; J. Amer. Chern. Soc. 40, 1361 (1918).
44. Volmer, M.; Z. Phys. Chern. 115, 253 (1925).
45. Fowler, R.M.; Proc. Camb. Phil. Soc. 31, 260 (1935).
46. Brunauer, S., Emmet, P.H. and Teller, E.; J. Amer. Chern. Soc., 60, 309
(1938).
47. Lecloux, A.J.; in "Catalysis Science and Technology". (Ed. J.R. Anderson
and M. Boudart) Springer-Verlag, New York (1981).
48. Linares-Solano, A., Lopez-Gonzalez, J.D., Martin-Martinez, J.M. and Rodri-
guez-Reinoso, F.; Adsorption Science & Technology 1, 123 (1984).
49. Ainscough, A.N., Dollimore, D. and Spooner, P.; in "Studies in Surface
Science and Catalysis". (Ed. Rouquerol, J. and Sing, K.S.W.) Vol. 10,
p. 187. Elsevier Publishing, Amsterdam (1982).
50. Brunauer, S.; J. Colloid Interface Sci., 39, 435 (1972).
51. Brunauer, S.; J. Colloid Interface Sci., 41, 612 (1972).
52. Bering, B.P., Dubinin, M.M., Serpinsky, V.V.; J. Colloid Interface Sci.,
38, 185 (1972).
53. Rand, B., Marsh, H.; J. Colloid Interface Sci., 40, 121 (1972).
54. McEnaney, B. and Master, K.J.; Thermochimica Acta, 82, 81 (1984).
55. Barrer, R.M.; in "Structure and Properties of Porous Materials" (Ed.
D.H. Everett and F.S. Stone) Butterworths, London, p. 50 (1958).
56. Rodriguez-Reinoso, F., Lopez-Gonzalez, J.D. and Berenguer, C.; Carbon
20, 513 (1982).
57. Kaganer, M.G.; Zhur, Fiz. Khim.; 32, 2202 (1959).
58. McClellan, A.L., Harnsberger, H.F.; J. Colloid Interface Sci., 23, 577
(1967).
59. Emmett, P.H. and Brunauer, S. ; J. Amer. Chern. Soc. , 59, 1553 (1937).
60. Zettlemoyer, A.C.; J. Colloid Interface Sci., 28, 343 (1968).
61. Rouquerol, J. , Partyka, S. and Rouquerol, F. ; J. Chern. Soc. , Faraday
Trans. 1., 73, 306 (1977).
62. Carruthers, J.D., Payne, O.A., Sing, K.S.W. and Stryker, L.J.; J. Colloid
Interface Sci., 36, 206 (1971).
63. Pierce, C. and Ewing, B.; J. Phys. Chern., 68, 2562 (1964).
64. Aristov, B.C. and Kiselev.; Russ. J. Phys. Chern. (transl.) 37, 1359 (1963).
65. Brown, C.E. and Hall, P.G.; Trans. Faraday Soc., 67, 3558 (1971).
66. Shull, C.G.; J. Amer. Chern. Soc., 70, 1410 (1948).
67. Crauston, R.W., Inkley, F.A.; Advan. Catal., 9, 143 (1957).
 177

68. Pierce, C.; J. Phys. Chern., 63, 1076 (1959).

69. Lippens, B.C., de Boer, J.H.; J. Catal., 4, 319 (1965).
70. Kiselev, A.V.; in "Structure and Properties of Porous Materials" (Ed.
D.H. Everett and F.S. Stone) p. 200. Butterworths, London (1958).
71. Pierce, C.; J. Phys. Chern., 72, 3673 (1968).
72. de Boer, J.H.; in "Surface Area Determination", (Eds. D.H. Everett and
R.H. Ottewill) p. 24. Butterworths, London (1970).
73. Carruthers, J.D., Cutting, P.A., Day, R.F., Harris, M.R., Mitchell,
S.A. and Sing, K.S.W.; Chern. and Ind., 1772 (1968).
74. Bhambhami, M.R., Cutting, P.A. Sing, K.S.W. and Turk, D.H.; J. Colloid
Interface Sci., 38, 109 (1972).
75. Lippens, B.C., Linsen, B.G. and de Boer, J.H.; J. Catalysis 3, 32 (1964).
76. Griffiths, D.W.L., Thomas, W. and Walker, P.L., Jr.; Carbon 1,515 (1964).
77. Salinas-Martinez de Lecea, C., Linares-Solano, A., Lopez-Gonzalez, J.D.,
and Rodriguez-Reinoso, F.; Carbon 17, 65 (1981).
78. Mikhail, R. Sh., Brunauer, S., and Bodor, E.E.; J. Colloid and Interface
Sci., 26, 45 (1968).
79. Van der Plas, Th.; in "Physical and Chemical Aspect of Adsorbents and
Catalyst". (Ed. B.G. Linsen) p. 427. Academic Press.
80. Parfitt, G.D., Sing, K.S.W. and Urwin, D.; J. Colloid Interface Sci.,
53, 187 (1975).
81. Rodriguez-Reinoso, F., Linares-Solano, A., Garrido Torres-Puchol, A.,
and Lopez-Gonzalez, J.D.; An. Quim., 78, 166 (1982).
82. Sing, K.S.W.; in "Surface Area Determination". (Ed. D.H. Everett and
R.H. Ottewill) p. 25. Butterworth, London (1970).
83. Ledoux, A. and Pirard, J.P.; J. Colloid Interface Sci., 70, 265 (1979).
84. McEnaney, B. and Dovaston, N.G.; Carbon 13, 515 (1975).
85. Marsh, H. and Lamond, T.G.; Carbon 1, 293 (1963).
86. Culver, R.V. and Heath, N.S.; Trans. Faraday Soc., 51, 1569 (1955).
87. Everett, D.H. and Powl, J.C.; J. Chern. Soc., Faraday Trans. I., 72,
619 (1976).
88. Kiselev, A.V.; in "Proceeding of the Second International Congress on
Surface Activity". II. p. 219. Butterworths, London (1957).
89. Gregg, S.J. and Stock, R.; Trans. Faraday Soc., 53, 1355 (1957).
90. Brown, M.A., Foster, A.G.; J. Phys. Chern., 56, 733 (1952).
91. Dollimore, D. and Shingles, T.; J. Appl. Chern., 19, 218 (1969).
92. Toda, J., Hatami, M., Toyoda, S., Yoshida, Y. and Honda, H.; Carbon
8, 565 (1970).
93. Masters, K.J. and McEnaney, B.; J. Colloid and Interface Science, in
press.
94. Gregg, S.J. and Pope, M.I.; Fuel 38, 501 (1959).
95. Marsh, H. and Wynne-Jones, F.K.; Carbon 1, 281 (1964).
96. Gregg, S.J., Olds, R.M. and Tyson, R.F.S.; in "Third Conference on Indus-
trial Carbon and Graphite". (Soc. Chern. Ind.) p. 184. Academic Press,
New York (1970).
97. McEnaney, B. and Willis, M.A.; in "High Temperature Chemistry of Inorganic
and Ceramic Materials". (Ed. F.P. Glasser and P.E. Potter) p. 102. Chemical
Society. London (1977).
98. Polanyi, M.; Trans. Faraday Soc., 28, 316 (1932).
99. de Boer, J.H. and Custers, J.F.H.; Z. Phys. Chern. 258, 225 (1934).
178
100. Gurfein, N.S., Dobychin, D.P. and Kopliensko, L.S.; Russ. J. Phys.
Chern. 44, 411 transl. (1970).
101. Stoeckli, H.F.; Helv. Chim. Acta, 57, 2195 (1974).
102. Polanyi, M.; Verh. Deut. Phys. Ges. 16, 1012 (1914).
103. Oubinin, M.M.; Quart. Rev. Chern. Soc., 9, 101 (1955).
104. Dubinin, M.M. and Polystyanov, E.F.; Izv. Akad. Nauk. SSSR, Servo Khim.
p. 793 (1966).
105. Dubinin, M.M. and Timofeev, D.P.; Dokl. Akad. Nauk. SSSR., Servo Khim.,
53, 137 (1947).
106. Dubinin, M.M. and Zaverina; Z.H. Fig. Khim., 72, 319 (1950).
107. Dubinin, M.M. and Radushkevich, L.V.; Proc. Acad. Sci., SSSR., 55,
331 (1947).
th
108. Dubinin, M.M., Proc. 5 Conf. on Carbon, 1, 81 (1962).
109. Nikolaev, K.M. and Dubinin, M.M.; Izv. Akad. Nauk. SSSR., Otd. Khim.
Nauk, 1165 (1958).
110. Stoeckli, H.F., Houriet, J.Ph., Perret, A. and Huber, U.; in "Characteri-
zation of Porous Solids". (Eds. S.J. Gregg, K.S.W. Sing and H.F. Stoeckli)
p. 31 (1978).
111. Marsh, H. and Rand, 8.; J. Colloid and Interface Sci., 33, 101 (1970).
112. Walker, P.L., Jr. and Patel, R.L.; Fuel 49, 91 (1970).
113. Lamond, T.G. and Marsh, J.; Carbon 1, 281 (1964).
114. Dubinin, M.M. and Astakhov, V.A.; "Molecular-Sieve Zeolithes-II". American
Chemical Society, Washington (1971).
115. Dubinin, M.M. and Astakhov, V.A.; Izv. Nauk. SSSR., Ser. Khim. 1, 5
(1971) •
116. Dubinin, M.M.; in "Characterization of Porous Solids". (Eds. S.J. Gregg,
K.S.W. Sing and H.F. Stoeckli) p.1, Soc. Chern. Ind. London (1979).
117. Kadlec. D.; in "Physical Adsorption". p. 140, Prague: Heyrowski Institute
of Physical Chemistry and Electrochemistry, 1975.
118. Rand, J.; J. Colloid and Interface Sci., 56, 337 (1976).
119. Huber, U., Stoeckli, H.F., Houriet, J.Ph.; J. Colloid Interface Sci.,
67, 195 (1978).
120. Stoeckli, H.F.; J. Colloid Interface Sci., 59, 184 (1977).
121. Stoeckli. H.F. and Kraehenbuehl, F.; Carbon 19, 353 (1981).
122. Kraehenbuehl, H.F. Ph.D. Thesis, University of Neuchatel, Switzerland
1983.
123. Hobson, J.P.; Can. H. Phys., 43, 1934 (1965).
124. Gregg, S.J. and Langford, J.F.; Trans. Faraday Soc., 65, 1394 (1969).
125. Eltekov, V.A., Kiselev, A.V. and Novikova, U.N.; Dokl. Akad. Nauk,
149, 131 (1963).
126. 8arrer, R.M.; in "Surface Area Determination". p. 90. 8utterworths.
London (1970).
127. Stoeckli, H.F. and Kraehenbuehl; Carbon 22, 291 (1984).
128. Medek, J.; Fuel 56, 131 (1977).
129. Spitzer, Z., 8iba, V. and Kadlec, 0.; Carbon 14, 151 (1976).
130. Martin-Martinez, J.M., Linares-Solano, A., Rodriguez-Reinoso, F., and
Lopez-Gonzalez, J.D.; Adsorption Science & Technology, 1, 195 (1984).
131. Kadlec, O. and Dubinin, M.M.; Carbon 13, 263 (J~75).
132. Ali, S. and McEnaney, 8.; Abstract of 16 Amer. Carbon Conference.
San Diego p. 377 (1983).
PART III. CATALYTIC GASIFICATION
 181

CATALYTIC GASIFICATION

J.A. Moulijn and F. Kapteijn

Institute for Chemical Technology, University of Amsterdam
Nieuwe Achtergracht 166
1018 WV AMSTERDAM
The Netherlands

I NTRODUCTI ON
Catalysed gasification of carbonaceous materials is important
both from a practical and a theoretical point of view. Sometimes,
it is an undesired reaction:
- corrosion of carbon is often due to gasification catalysed by
impurities [1]
- carbon is a promising catalyst support or even a catalyst in it-
self [2-8]. However, the stability of the carbon applied is
amongst other things dependent on the reactivity towards cataly-
sed gasification
In other cases it is a desired reaction:
- catalytic coal gasification enables a one-step process for the
conversion of coal into methane [9-13]
- catalytic coal gasification enables the coupling of coal gasifi-
cation with an external heat source, e.g. a nuclear reactor
[ 12]

- solid catalysts that are deactivated by deposition of coke can

be regenerated by gasification of this coke.
A number of materials have been shown to exhibit catalytic activi-
ty, the most active being transition metals, alkali metals and
alkaline-earth metals.
182

Gasification of carbon can take place by several reactions:

C + 2 H2-~ CH4 (1)
C + 02 -+ CO 2 (2)
C + H2O ----;... CO + H2 (3)
C + CO2 -+ 2 CO (4)

It is clear that the mechanism is not the same for all these
catalyst systems and reactions. Transition metal catalysis isdealt
with by Baker [14) and HUttinger [15) and will not be discussed
here.
In this chapter alkali metal and alkaline earth metal catalysis
is emphasised. These metals are active towards the gasification by
02, C02 and H20, whereas they show little, if any, activity towards
H2. Because of its simplicity, the C02 gasification has been stu-
died most extensively.
Of the various alkali and alkaline-earth metal catalysts
K2C03 has been paid most attention to, mainly because it is be-
lieved to be the most promising catalyst for practical application.
In this introductory chapter the discussion will be focussed on
the K2C03-catalysed gasification by C02'

EXPERIMENTAL FACTS (K2C03 catalysed C02-gasification)

Potassium carbonate catalyses not only the gasification of carbon

by C02 but also gasification by H20 and 02' The order of the reac-
tivity of the alkali metals is according to that of the periodic
tab 1e:
Cs > Rb > K> Na > Li
However, note that pretreatment conditions are critical and can
change this sequence [16, 17). The reactivity increases with the
potassium contents up to a saturation point of ~ K/C = 0.1
Below this saturation point some authors have reported a linear
relationship, whereas others found S-shaped behaviour [18-20).
Careful recent data from Shadman [21] , however, show convincingly
that a linear relationship exists (figure 1)
 183

06
Figure 1.
Initial gasification rate
as function of K-loading

02

oL-~--~~--~~--~
o 02 0.4 0.6 0.8 10 12
KIC • 103

In contrast to most examples of solids acting catalytically.

the initial dispersion is ofminor importance; physical mixtures and
impregnated samples show comparable activity [22]. In most studies
potassium is added in the form of K2C03. therefore. it is not sur-
prising that much attention has been given to the possible decom-
position of this salt [11. 16.23.27]. It has generally been
found that decomposition takes place below the melting temperature
of bulk K2C03. Detailed information is given by "outgassing ex-
periments" in which the sample is heated. using a linear rate of
heating, in a flow of inert gas. Figure 2 gives a typical result
of such1an experiment, the production of CO and C02 being plotted
15

Production
mmols· 1moliC.1
I ' ---- CO 2
10
-co Figure 2.
Outgassing pattern of a
K2C03/NORIT RX 1 extra
sample (heating rate
05 10 K min-1, Heflow 10-4
mol s-1. C/K = 51.8)
-----, C02; -----CO

400 f;()() HOD 1000 1200 Isothermal

184

as a function of the temperature (heating rate 10 K min- 1).

Between ~ 500 and 1000 K decomposition takes place according to:
(5 )

in which 'K20/C' denotes carbon supported potassium oxide of a yet

unknown structure. Between 900 and 1200 K reduction of the active
species 'K20/C' to the metal takes place:
'K20/C' + C~ 2 'K/C' + CO (6)

It is significant that catalytic activity is already observed at

temperatures where most of the potassium is present in an oxidic
form.
Fundamental information is obtained from kinetics [25].
It is highly significant that CO is an inhibitor, the initial rate
being [26]:

k2 NT
r = (7)
PCO
1+
PC02 K1
This equation has been explained by:
CO2 + Cf 0( k-1l> CO + C(O) (8)
k1

C(O) ~ CO (9)
k2 is the forward rate constant of equation (9); PCO and PCOz are
the partial pressures of CO and C02. K1 (=~) is the equili-
brium constant of equation (8) and it is assu~ed that equation (9)
is rate determining
Experiments with labeled compounds. show that the O-exchange
reaction
*C02 + CO ~
~ *CO + C02 ( 1 0)

is much faster than gasification [28]. Moreover it was found that

the exchange rate has a positive order in PCO in contrast to the
gasification rate, where CO is an inhibitor. By in-situ infrared
spectroscopy it was shown that a key property is the propensity of the
alkali ion to form either a surface compound or crystalline alkali
carbonate [29]. In the former case a highly dispersed catalytic
system is formed whereas in the latter case inactive crystals are
 185

present. It appeared that the fraction present as inactive crystals

decreases in the sequence Na- K- Cs
Some direct information on the structure of the active sites
is given by poisoning studies, with HCl appearing to be a poison
in contrast to H2S [1S, 30]. In this context it is noteworthy that
CH3I reacts with the system after gasification [31], showing its
nucleophilic character.
In a recent study Tromp and Cordfunke [32] investigated the
thermodynamic stability of CSK at various temperatures. A typical
result of high-temperature X-ray diffraction experiments is given
in figure 3. Up to ~ 545 K only the reflections of CaK are obser-
ved. Between 545 and 610 K CSK and C24K coexist. At higher tempe-
ratures there is a stepwise transition from lower to higher stage
intercalation compounds. It is clear that> 675 K no bulk inter-
calation compounds are observed. The same authors also checked the
stability of intercalation compounds prepared from activated car-
bon. They found that these are even less stable than those prepa-
red from graphite. Therefore, considering the fact that < 900 K
no potassium metal is present during the heating up of a K2C03/
carbon sample, it can be concluded that during gasification no
bulk intercalation compounds exist. Similar results have been ob-
tained by others [33]

05

~ Figure 3.
t.:: 620
X-ray diffraction patterns
645
&50
I of CSK during heating at a
1i nea r1hea t i ng ra te (he 1i urn ,
685 II 5 Kh )

It is significant that the K2C03/C system itself is a catalyst

for many reactions. An obvious example is the disproportionation
reaction:
2 CO ~ C02 + C ( 11 )
186

being the reverse of gasification by C02. Recently Kapteijn and

Moulijn [34] showed that alkali metal/carbon systems catalyse the
reaction:
( 12)

Simultaneously the Boudouard and probably the water gas-shift re-

action took place. Although the system was not fully optimised it
exhibited significant catalytic activity at 723 K. Also for this
reaction the sequence in activity was:
Cs > Rb > K > Na
A commercially, highly interesting, reaction is the reduction of
nitrogen oxides, e.g.
(13)

and alkali metal/carbon systems proved to be very active.

To obtain information on K2C03/carbon at high temperatures,
vapour pressure measurements under equilibrium conditions at
typical gasification tempratures were performed [35,36].
Tromp and Cordfunke [36] found that for K2C03/graphite, at all K/C
ratios, a fixed K-vapour pressure is observed, whereas for K2C03/
activated carbon the vapour pressure is dependent on the K/C ratio.
The vapour pressure for the graphite case was slightly below the
theoretical value associated with the equilibrium:
K2C03 + 2C t K + 3CO (14)

It was concluded that this small deviation is because, in addition

to the CO some C02 is formed.
For the activated carbon case the potassium vapour pressure
is considerably lower than theoretically expected, especially at
low K/C ratios. Hence, significant interaction exists in this case.
In addition to the outgassing experiments, step-response expe-
riments give,significant information on the nature of the potassium
species during gasification and a number of results have been re-
ported by Cerfontain et al. [24,37]. Figure 4 gives some examples.
In these experiments helium flows at a certain temperature through
the reactor, and, within a short interval of time, a C02-flow is
substituted for the He-flow. In Figure 4, the resulting CO and C02
concentrations are plotted for various temperatures. As expected,
CO is present in the effluent, but the concentration patterns con-
sist of two parts. Initially, a peak is observed followed by
steady state concentration level.
 187

LL
(.-- f~-- f:------ f;;---------· r-;-----------

tl:
,iL... "
Figure 4. C02 and CO response for He to C02 step-response
experiments at various temperatures after heat
treatment in He for 1073 K for 10 min. a, 773 K;
b, 873 K; c, 973 K; d, 1023 K; e, 1073 K.
-'-'- , C02; - - , CO
The interpretation is as follows. The finite amount of potassium
present in the metallic phase is oxidised and, as a consequence,
a peak is observed. This peak will be called an lIovershoot".
The constant CO level corresponds with steady-state gasification.
The area of the overshoot increases with the temperature.
From the peak area it can be calculated that at 1073 K the potus-
sium is formally present as "K01.i l although part of the over-
shoot is probably due to oxidation of the carbon surface.

MECHANISMS PUBLISHED
Figure 5 summarises the mechanisms published. Taylor and
Neville [38] were among the first to propose a mechanism for the
catalysed gasification of carbon in C02. Essentially they consi-
dered catalysis to be due simply to an increased adsorption of
C02. Fox and White [39] believed that the rate-determining step is
not a chemical one but that the acceleration is due to an increase
in rate of external mass transfer. In their view Na2C03 is reduced
at the surface of the carbon. Because of its high vapour pressure
the Na-metal formed evaporates into the bulk .gas phase.
After reaction with C02 the Na2C03 is drawn back to the surface,
etc. In essence, they consider the catalyst as a pump between bulk
gas phase and carbon surface.
Long and Sykes [40] were the first to explain catalysis in
terms of electronic interaction between catalyst and carbon.
They proposed that a non-stoichiometric oxide with excess of metal
is formed leading to covalent ~1-C bonds. By this action:thearoma-
tic character would be decreased and, as a consequence, accelera-
tion of the reaction occurs.
188

Taylor and Neville CatalySIS due to Increa!ed adsorption of C02 Mlms and Pabst
C0i(0-KJUcCO'
further CO
oxidiZed
F"ox and White cluster
COl No~Na2Co, 2Na.2CO: .. Na~Ol·CO

---~Cb-----~~L-I--·----- No.,C01.2C~ZNO. 3eD Wigmans

Long ond Sykes o

MT:H_O_C_t M\~"2H_O_C

without catalyst
6.
M
with catalyst
2M'-~~~:;?11~ M-O-C-. H2
MCsur ' _ 2MC x M(g)
Franke ond Merolklb Catolyst donates electrons Into lattice s.... tace Intercalation evaporation
CO J • e- CO. O~o.l~ stabilization
0:'1,. c #CO. e
KapteIJ". CerfontOln

CO~XC02 parallel wIth mechanIsm Of McKee

KOH CO

6~6~(5~002-6 meta
stobie

Figure 5. Survey of published mechanisms for potassium

catalysed C02 gasification
Franke and Meraikib [41] concluded that the alkali metal intro-
duces extra electrons into the carbon lattice. Catalysis is as-
summed to be due to an increase in concentration of ion radicals
at the surface. r-1cKee et al. [42, 43] proposed a kinetic scheme
(related to the chemistry proposed by Fox and \~hite) consisting of
three elementary reactions:
Carbothermic reduction
K2C03 + 2C ---l>- 2K + 3 CO (15 )

Oxidation
2K + C02 t K20 + CO (16)

Carbonate formation
K2 0+ C02 ~ K2C03 ( 17)
 189

Yokayama et al. [44] assumed that the catalytic cycle only involves
the metal and K20!
A possible involvement of intercalation compounds has often
been suggested [40,45]. Wen [46] published a detailed mechanism
basically following the McKee mechanism but instead of alkali metal
he assumed intercalation compounds. Mims and Pabst [47] were the
first to propose that surface complexes of the potassium phenolate
type, are the active species. Later they concluded that the active
species contain 4 - 5 potassium ions [31]. Catalysis is assumed to
take place by a combination of a C02 dissociation equilibrium and
a rate determining reductionbyCarbon. ~vigmans [17] integrated
mechanisms of McKee, Wen, and Mims and Pabst by presenting a de-
tailed scheme involving surface complexes, metal oxide and metal
in various forms (bulk vapour, physi- and chemisorbed at the sur-
face, intercalation compounds). The relative abundance of the
various species was supposed to depend on the composition of the
surface and the reaction conditions. Kapteijn and Cerfontain [24,
26] formulated the catalytic elementary steps by the cycle:
[KxOy + 11 + CO (18)

[KxO y + 1 ] + C ~ KxOy + CO ( 19)

Reaction (19) was assumed to be rate determining. The reaction
rate is proportional to the concentration of [KxO y + 1] species,
and, consequently, governed by the "oxidation power" of the gas
phase. Recently, Cerfontain [24] showed that x/y is greater
than 0.5, so the species are oxidic rather than metallic.
Spiro et al. [48] suggested that in char gasification a hy-
dride-hydroxide cycle is operative besides the "classical" ~4cKee
cycle (Equation (15) - (17)) because of the high H- contents of
char compared to graphite. Wood et al. [35] concluded that the
carbonate is transformed into a molten layer of non-stoichiometric
oxides with an excess of the metal, this liquid film serving as
the oxygen transfer medium between the gas phase and the carbon
surface.
An alternative way to classify the literature on mechanisms
is by considering the proposed active species as is done in Table
1.
Table 1 Suggested active species

K gasphase K+ K20. K20 2• K20 3• etc

physisorbed K 2C0 3
chem isorbed KOH
intercalation compounds
@-OK
@-K 101
KxOy
190
In an excellent review Wood [49] critically summarizes the litera-
ture, basically in an analogous way.

DISCUSSION
In principle, there are many possibilities to explain cataly-
sis in carbon gasification. The following scheme for gasification
in C02 suggests itself:
C02:t CO + (0) (20)
(0) + Cf t C(O) ( 21)

C(O) :t CO (22)
It is assumed that the dissociation reaction (20) is in equilibrium
during gasification. With this assumption, catalysis can be ex-
plained as follows:
1. C02 dissociation, reaction (20) is accelerated., e.g. by a redox
cycle, or the equilibrium is shifted to the right due to thein-
volvement of catalytic species in the reaction stoichiometry.
2. The surface reaction (21) is accelerated by increasing the con-
centration of (0) or Cf by increasing the intrinsic rate con-
stant.
3. The decomposition of the surface complexes C(O) is accelerated
by destabilisation of these complexes.
All the mechanisms proposed in the literature fit this scheme:
* Taylor-Neville
The concentration of (0) is increased by the catalysts
* Fox and White
This mechanism is not a chemical one. Because external mass
transfer limitations are not important in many studies, this
mechanism is incorrect.
* Long and Sykes
The catalyst accelerates the decomposition step (22)
* Franke and Heraikib
The dissociation of C02 is accelerated, step (20) and the con-
centration of Cf is increased(step (21) is accelerated)
* McKee
Tneiaissociation of C02 is accelerated (step 21)
 191

* Wen
See McKee.
* Mims and Pabst
The surface reaction is accelerated by increasing the concen-
traction of (O)(step 21)
* Wigmans
This mechanism is a combination of McKee, Wen, and Mims and
Pabst.
* Kapteijn and Cerfontain
The concentration of (0) is increased and, as a consequence,
(step 21) is accelerated.
* Spiro
See McKee.
* Wood
rne-dissociation of C02 is accelerated, thus increasing the
concentration of (0).
A different approach is to elucidate the chemical structure of
the active species, (Table 1). Of course, some of the structures
in Table 1 are not per se at variance. For instance, chemisorbed K,
might well be identical with an intercalation compound. Also KxOy
is a more general description K20, K202, K203 and even
~-O- K+

The mechanism can be classified according to the structures in

Table 1. Metallic potassium as the active species is assumed by Fox
and White, Franke and Meraikib, McKee, Wen, Yokoyama, Wigmans
(partially) and Wood. However, from the experimental facts it is
concluded that the active species is not metallic, although after
heating up in an inert atmosphere metallic potassium canbe present.
However, bulk intercalation compounds are not present. More proba-
ble schemes are those involving only oxidic potassium species.
C02 + KxOy ~ CO + KxOy + 1 (23)
KxOy + 1 + C ~ KxOy + C(O) (24)

C(O) + CO (25)
However, recent kinetic results show that this scheme is too
simple [28, 25]. Especially the observation that CO promotes the
exchange rate (equation 9) whereas it inhibits gasification proves
that the gasification sites are different from the exchange sites.
 192

The inhibition by CO shows that at least one additional step has to

be incorporated in the kinetic scheme. e.g.:
CO + C(O) ---+- C02 + C (26)
Nearly all experimental results can be explained by the above
mechanism. Reaction (23) is not only possible for C02 but also for
H20 and 02. This explains the catalytic activity for gasification
in C02. H20 and 02. Similarly. the catalytic activity for the dis-
proportionation. the methanation and the NO reduction can be
explained. x
The most important conclusion from the kinetic studies is that
in alkali metal catalysis the catalyst only increases the number of
active C(O) species; the rate-determining gasification step is the
same both for the uncatalysed and catalysed gasification. viz. step
(22). The activity sequence Na < K < Cs is explained by the different
number of active sites. Apparently. Cs has the highest dispersion
propensity.

REFERENCES

1. Morgan. W.C. and M.T. Thomas.

Carbon 20. 71 (1982)
2. Beer. V.H.J. de. F.J. Derbyshire. C.K. Groot. R. Prins.
A.W. Scaroni and J.M. Solas.
Fuel 63. 1095 (1984)
3. Kapteijn. F. and J.A. Moulijn.
J. Chem.Soc.Chem.Commm •• 278 (1984)
4. Kapteijn. F.• A.J.C. Mierop. G. Abbel and J.A. Moulijn.
J. Chern. Comm .• 1085 (1984)
5. Thomas. C.L .•
Catalytic Processes and Proven Catalysts.
Acad. Press. NY. U.S.A. 1970
6 Boehm. H.P .• G. Mair. T. Stoehr. A.R. de Rincon and R. Terec-
zki.
Fuel 63, 1061 (1984)
7. Klein, J. and K-D Henning,
Fuel 63. 1064 (1984)
8. Rodriguez-Reinoso, F., J. de Dios L6pez- Gonzalez, C. Moreno-
Castilla, A. Guerrero-Ruiz and I. Rodriguez-Ramos,
 193

Fuel 63, 1089 (1974)

9. Walker, P.l., F. Rusinko and l.G. Austin,
Adv. Catal. 11, 133 (1959)
10. Nahas, N.C.
Fuel 62, 239 (1983)
11. McKee, D.W.,
Fuel 62, 170 (1983)
12. JUntgen, H. and K.H. van Heek,
Kohlevergasung, Karl Thiemig, MUnchen, 1981
13. Wigmans, T.,
Ph.D. thesis, University of Amsterdam, Amsterdam, 1982
14. Baker, T.,
this book
15. HUttinger, K.,
this book
16. Kapteijn, F., G. Abbel and J.A. Moulijn,
Fuel 63, 1036 (1984)
17. Wigmans, T., J.C. Goebel and J.A. Moulijn,
Carbon ~, 295 (1983)
18. Mims, C.A. and J.K. Pabst,
Am.Chem.Soc. Div. Fuel Chern. 25, 1258 (1980)
19. Wigmans, T., P. Tromp and J.A. Moulijn,
Carbon 22, 319 (1984)
20. Huhn, F. J. Klein and H. JUntgen,
Fuel 62, 197 (1983)
21. Shadman, F.,
Personal communication
22. Wigmans, T., R. Elfring, E.P.H. Cordfunke and J.A. Moulijn,
Prepr. 15th Conf. on Carbon, Philadelphia, USA, 407 (1981)
23. Wigmans, T., J. van Doorn and J.A. Moulijn.
Fuel 62, 190 (1983)
24. Cerfontain, M.B., and J.A. Moulijn,
Proc. Int. Conf. Coal Science, Pittsburgh, USA, 419 (1983)
 194
25. Kapteijn, F. and J.A. Moulijn,
this book
26~ Kapteijn, F. and J.A. Moulijn,
Fuel 62, 221 (1983)
21. Kapteijn, F., O. Peer and J .A. Moul ijn,
to be published
28. Cerfontain. M.B., R. Meijer and J.A. Moulijn,
Proc. Int. Coal Science Conf., Sydney, Australia, in press
29. Cerfontain, M.B. and J.A. Moulijn,
Submitted for publication
30. HUttinger, H.J.,
Personal communication
31. Mims, C.A. and J.K. Pabst,
Fuel 62, 176 (1983)
32. Tromp, P.J.J. and E.P.H. Cordfunke,
Thermochimica Acta 77, 49 (1984)
33. Ferguson, E.,R. Schlogl and W. Jones,
Fuel .§l, 1049 (1984)
34. Kapteijn, F. and J.A. Moulijn,
J. Chern. Soc. Chern. Comm, 278 (1984)
35. Wood, B.J. , R.H. Fleming and H. ~Jise,
Proc. Int. Conf. Coal Science, Pittsburgh, USA, 415 (1983)
36. Tromp, P.J.J. and E.P.H. cordfunke,
Thermochimica Acta, ~ 113 (1984)
37.Cerfontain M.B., F. Kapteijn and J.A. Moulijn,
8th Int. Congress on Catalysis, Berlin, FRG, 1984
38. Taylor, H.A. and H.A. Neville,
J.Am.Chem. Soc. 43, 2005 (1921)
39. Fox, D.A. and H.H, White,
Ind. Eng. Chern. 23, 259 (1931)
40. Long, F.J. and K.W. Sykes,
Proc. Roy. Soc. (London) Series A, A215, 100 (1952)
41. Franke, F.H. and M. Meraikib,
Carbon ~, 423 (1970)
 195

42. McKee, D.W. and D. Chatterji,

Carbon ..!l, 381 (1975)
43. r~cKee, D.W.,
Carbon 20, 59 (1982)
44. Yokoyama, S.Y., K. Mujahara, K. Tanaka, J. Tashiro, and I
Takakuwa,
Nippon Kagaku Kaishi ~, 974 (1980)
45. Wigmans, T, A. Hoogland, P. Tromp and J.A. r~oulijn,
Carbon ~, 12 (1983)
46. Wen, W. - Y. ,
Catal. Rev. Sci. Eng. ~, 1 (1980)
47. Mims, C.A. and J.K. Pabst,
Am.Chem.Soc.Div. Fuel Chem. 25, 263 (1980)
48. Spiro, C.L. , D.W. McKee, P.J. Kosky and E.J. Lamby,
Fuel g, 181 (1983)
49. Wood, B.J. and K.M. Saucier,
Catal. Rev. Eng. 26 233 (1984)
 197

ALKALI METAL CATALYZED WATER VAPOUR GASIFICATION OF

CARBON USING MINERAL CATALYST RAW MATERIALS

K.J. Hlittinger, R. Minges

Institut fUr Chemische Technik

Universitat Karlsruhe
Fed. Rep. Germany

1. INTRODUCTION

Alkali metals have been shown to be the most promi-

sing catalysts of water vapour gasification of carbon or
coke (1,2). The advantages in comparison with other cata-
lysts like iron or earth alkali metals are low tempera-
ture activity and selectivity for carbon monoxide metha-
nation and the shift reaction. The most impressive example
represents the potassium catalyzed coal gasification pro-
cess of Exxon, which makes use of all these catalyst pro-
perties (3,4). However, it is also known, that the ex-
cellent low temperature activity is only achieved with
such potassium salts like potassium hydroxide or potas-
sium carbonate, which under in situ conditions of gasi-
fication are easily forming the K-O complex as catalytic
active species.
High catalyst amounts up to 30 weight % of the coal
and high catalyst losses by reaction with the silicates
of the coal minerals make the use of expensive catalyst
raw materials like potassium hydroxide or potassium car-
bonate questionable (5,6).
HUttinger et al. have shown (7) that the activity of
a catalyst for coal gasification depends on (1) activa-
tion of the catalyst raw material, (2) intrinsic activity
of the active species, and (3) inactivation of the active
species. In so far, it should be possible to overcome the
economic problem of catalyzed coal gasification by intro-
ducing low cost catalyst raw materials of mineral origin
like chlorides or sulfates I if the activation problem
198

can be solved. Such catalysts could be used without re-

covery of the metal from the gasification residue on a
once through basis.
The problem was studied in two steps, namely (1) by
fundamentally investigating the activation of common po-
tassium salts, and (2) by elaborating possibilities for
accelerating the activation process of alkali chlorides
and sulfates. The latter problem was treated in an ex-
emplary manner by following three routes:
(1) Lowering the melting point, studied with the systems:
KCI / MgCl 2 ' KCI / NaCI, KCI / NaCI / MgC1 2 .

(2) Applying an in situ anion exchange forming more easily

reducible compounds. This route was studied with the
following systems:
++ ++ -- ++ ++ --
KCI/(Mg , Ca
) C0 3 ' KCI/(Mg , Ca )S04 .
(3) Catalyzing the activation. This route was followed
with the systems:
K2 S0 4 /Fe
++/+++ --
(S04
-
' CI , ° ,S ).
All points were studied by separately analyzing the
activation process and the catalytic activity during ga-
sification of a natural graphite as a model carbon and a
subbituminous coal for comparison.

2. EXPERIMENTAL
2.1. Raw materials

As model carbon a CEYLON graphite was used (ash content

2.5 wt%, main components silicon and iron). Surprisingly,
no catalytic activity of iron below 950°C was found. To
follow the activity of the various catalyst systems fur-
thermore, a German subbituminous coal (LEOPOLD) was used.
The ash content of the coal was determined to amount to
4.8 wt%, further analytical data are given in earlier
papers (8,9). Iron as constituent of mineral matter
(0.53 wt% of coal, maf) may be catalytically active under
reducing gasification conditions and also in hydrogasifi-
cation (8-11).
All catalyst raw materials were received from MERCK in
an analytical grade. They were added to the graphite or
coal by dry mixing; the desired particle size between 0.4
and 1.0 mm was produced by pressing of pellets without a
binder, which then were crushed. The catalyst amount was
 199

5 mass parts potassium as catalytic active metal per

100 mass parts graphite or coal (mf). In all cases where
KCI was employed, the double amount was taken in order
to compensate losses by evaporation. Variation of the
catalyst amount in this range has only a minimum effect.

2.2. Methods

The activauion process was studied by applying various

methods like thermogravimetric analysis, differential
thermoanalysis, X-ray photoelectron spectroscopy, scan-
ning electron microscopy and electron probe microanaly-
sis (12,13). Melting and wettability were investigated
using a sessile drop arrangement (14). For gasification
experiments a fixed bed flow reactor was used which is
described elsewhere (8,9). The gasification atmosphere
was either an equimolar argon/ water vapour mixture (oxi-
dizing conditions) or an equimolar hydrogen/water vapour
mixture (reducing conditions), total pressure mainly was
2 MPa.

3. ACTIVATION OF PURE POTASSIUM SALTS

The activation of pure potassium salts, namely potas-

sium hydroxide, potassium carbonate, potassium nitrate,
potassium sulfate, potassium halides, was studied by thermo·
dynamic calculations, kinetic and physico-chemical mea-
surements using the above cited methods. The results will
be presented in a separate paper (15). They are summarized
in a schemeJshown in Fig.1.
The decisive point of these studies is the finding
that the activation of all compounds is leading by hydro-
lysis to potassium hydroxide as the key compound, after
decomposition and reduction reactions have occurred (po-
tassium nitrate/potassium carbonate, potassium carbonate/
potassium oxide ?, potassium sulfate/potassium sulfide)
(12,16,17). Potassium hydroxide is reduced on the carbon
surface to elemental potassium which may further react
with potassium hydroxide, whereby potassium oxide is
formed. From these single reactions KxOy (y<x) as the
catalytic active species is resulting; in the scheme it
is marked as K for simplification. As generally assumed,
the active species catalyzes hydrolysis of water finally
yielding a carbon oxygen surface complex (C(O)). This
complex may be desorped as carbon monoxide or further
reacts after desorption to carbon dioxide or methane. The
scheme explains the catalytic influence of potassium on
both reactions. The shift and methanation reaction have
200

been proven in separate studies using potassium impreg-

nated active carbon and synthesis gas (18,19).

GENERAL REACTION SCHEME OF THE

ICT POTASSIUM CATALYZED WATER VAPOUR GASI- 1985
FICATION OF COAL INCLUDING AC fI VATION
OF VARIOUS POTASSIUM SALTS Fig. 1

The studies on the activation process have confirmed

that the missing low temperature activity of potassium
catalyzed gasification with potassium chloride and po-
tassium sulfate results from the more difficult conver-
tibility to potassium hydroxide. In so far, the problem
of introducing these minerals as raw materials for po-
tassium catalyzed water vapour gasification consists in
accelerating the activation. Three possible routes as
described in the introduction will be discussed in the
following.

4. IMPROVED ACTIVATION BY LOWERED MELTING POINTS

The system used as catalyst raw materials in this part
of the study have the following melting points: potassium
chloride/magnesium chloride (stoichiometric) : 488 °c
 201

(dystectic); potassium chloride/sodium chloride (stoichio-

metric) 650 DC (eutectic); potassium chloride/sodium
chloride/magnesium chloride (21.6 mole % potassium chlo-
ride, 27.6 mole % sodium chloride, 50.8 mole % magnesium
chloride) : 397 DC (eutectic), whereas potassium chlo-
ride is melting at 770 DC. Phase diagrams of all systems
are available in literature (20-22). Sodium chloride and
especially magnesium chloride are more easily hydrolyzable
than potassium chloride, but the melting points of the
systems given above are not increased at least by partial
hydrolysis of these salts (20,22).
Gasification studies were performed with the graphite
at a linear rate of heating (argon/water vapour mixture,
total pressure 1 MPa). Under these conditions, carbon di-
oxide represents the main product.

40'---~-----------r---------------'

u;
b
x
'c
, Een 20 KCII NaCI/MgCI 2

"0
E KCI/NaCI
0'
u
Without Cot

0
900
, I
T :C
a I, min 60
Ceylon Graph~e I KCI ; 10 K ; Ar/H 2 0 ; ~,o=O.5; 1 MFb;
leT 1985
4 K· min-1 ; 900·C; 1h.
Fig. 2

As follows from Fig. 2 lowering the melting point re-

markably shifts the beginning of the gasification reac-
tion from approx. 860°C with pure potassium chloride to
770 °c , the maximum gasification rates are also higher
as compared to pure potassium chloride. Without catalyst,
the gasification rate of the graphite is negligible. The
sequence of activity corresponds to the melting points,
which means that the low temperature activity is the
higher the lower the melting temperature is. In so far,
the application of these systems is successful, but the
increase in maximum gasification rate as compared to
pure potassium chloride is not satisfying.
202

5. IMPROVED ACTIVATION BY IN SITU ANION EXCHANGE

In the systems used for this part of the study forma-

tion of potassium sulfate or potassium carbonate was ex-
pected by anion exchange during activation or gasifi-
cation according to the following equations:
++
2 KCl + (Mg ,Ca ++ ) C0 3 ~ K2 C0 3 + (Mg ++ ,Ca ++ )C1 2 -

2 KCl + ++ ca++) S04

(Mg,

The free reaction enthalpies of the reactions are shown

on the left hand side of Fig.3 in comparison to hydroly-
sis of potassium chloride. The~G~-values of all anion
400.-.------,,------.-. ~o.-.------.-------.-.

=>
~ 200
2KCf ... M9c03~K?C03 +MgCI;

2KCI+CaCOl --K1CO]+ Cacrz

~ 100 2KCI • MgSo, -K,SO,.MgCI;
2 KCI ... CoS04 K1S0 4 ... Co CI]

o --------------------------
o
400 800 1200 400 800 1200
T,K T,K

lIGR-VALUES FOR o. ANION EXCHANGE

I CT 1985
, b. ANION EXCHANGE AND HYDROLYSIS Fig. 3

exchange reactions are positive, but lower than the~G~­

values for hydrolysis of potassium chloride. Besides hy-
drolysis of potassium chloride, hydrolysis of magnesium
chloride and calcium chloride as consecutive reactions
of the anion exchange reactions have to be considered.
As follows from the right hand side of Fig. 3 hydrolysis
of magnesium chloride looks promising. By combining the
anion exchange reactions with the hydrolysis reactions
of the primary reaction products magnesium chloride and
calcium chloride the potassium chloride/magnesium sulfate
system becomes the most attractive one. Beyond 600 °c the
~G~ - values are negative. Removal of gaseous hydrogen
chloride in a flow system will further shift the equili-
brium.
Potassium sulfate is less easily activated than po-
tassium carbonate formed in the carbonate systems.
 203

The reducibility of potassium sulfate and potassium car-

bonate by carbon is shown in Fig. 4. It represents the
free reaction enthalpies of overall reactions consisting
of anion exchange, hydrolysis and reduction by carbon.
WOr-,------,------~-, 600r--"l---------,--------,----,

400

o 200
a b
400 800 1200 400 800 1200
T. K T. K
lIGR - VALUES FOR ANION-EXCHANGE, HYOROLYSIS AND
ICT CARBOTHERMAL REDUCTION FOR 1985
. ..::.S-,-UL::..;P-,H-,--A-,-,T...;::E,--...::S-,-Y-,-ST,-,E,--M-,S,--.=.b_:C,--A-'CR-,B,-,O,-N,---A'-C.T-=.E_-.=.SY-'--S=-T:...:E::..:.M-'S'--____-'--__-'
L..-_ _--L.::0:..;..· Fig. 4

A comparison between the sulfate systems (left hand side)

and carbonate systems (right hand side) reveals the su-
periority of both sulfate systems. In the case of the
carbonate systems the overall reactions mentioned above
are additionally compared with overall reactions consi-
sting of anion exchange, hydrolysis of the earth alkali
chlorides and of the potassium carbonate, which gives
~G~ - values in the same order of magnitude as calcu-
lated for the former ones. Therefore, it may be expected
that the sulfate systems should exhibit an improved low
temperature activity as compared to the carbonate systems.
The catalytic activity of the anion exchange systems
was studied by gasifying the graphite at 2 MPa using an
equimolar argon/water vapour mixture. The pure salts and
the primary anion exchange products potassium sulfate and
potassium carbonate were tested for reference. Fig. 5
shows the formation rates of carbon dioxide and carbon
monoxide as the main products of gasificationj as the
ratio of the formation rates is between 5 to 10 it is
possible to regard carbon dioxide formation rates only.
Potassium carbonate, potassium chloride/magnesium sulfate
and potassium chloride/calcium sulfate are most active,
followed by potassium chloride/calcium carbonate and pure
calcium sulfa.te. Compared to these catalyst systems the
activity of potassium chloride is small. The sequence of
activity found in gasification is in agreement with the
204

':'S? 100
"-c
E
"-0>
(5
E 50

a
u

20r---,-------,--------.----------~

'0
x
"c
E
70, 10
(5
E CaSO,
0 WITHOUT CAT ~,
u
CaC0),Mg so,

0
500 700 9QO
T. °C 60
CEY~ON GRAPHITE/K~ CacMg-S~LTS; 1~K.5KAriH:lO 1985
ICT
xH,o-0.5; 2MPa. 4K·mln; 900 C; 1h. Fig. 5

conclusions drawn from the thermodynamic calculations,

confirming that the potassium chloride/magnesium sulfate
system really is the superior one. In comparison to the
results of part 1 (lowered melting points) the starting
temperature of gasification is lowered to a similar ex-
tent or even more, but the gasification rates are at
least three times higher and comparable to those with po-
tassium carbonate.
Analogous gasification studies were performed with
the coal. With two exceptions, the results are in prin-
ciple identical to those with the graphite (23). (1)
With potassium carbonate the gasification reaction is
starting earlier than with the potassium chloride/magne-
sium sulfate and the other anion exchange systems. The
reason is the promoted activation only of potassium car-
bonate on the coal surface (24,25). (2) Potassium chlo-
ride/calcium carbonate exhibits a similar activity like
potassium chloride/magnesium sulfate. This improved
 205

activity may also be explained by the promoted activation

of the anion exchange product potassium carbonate on the
coal surface. In spite of the unique activity of potas-
sium carbonate in gasification of a coal the anion ex-
change systems represent valuable alternatives, complete
gasification of the coal at isothermal gasification at
700°C was shown to be possible (26).
The results of this part of the study may be summarized
in an activation scheme. This is shown for the most pro-
mising systems, namely potassium chloride/magnesium sul-
fate and potassium chloride/calcium carbonate, in Fig.6.

CO

CH 4 ~ C' ~ 1--~-=::::---'" C02

H2 KID} K KID) K
C(O)

KC\+CaC03
GENERAL REACTION SCHEME OF THE POTASSIUM
I CT CATALYZED WATER VAPOUR GASIFICATION OF 1985
COAL INCLUDING ACTIVATION OF K CI VIA
ANION - EX~IjAN~G~E_ _ _ _ ~ ___ ~ _____--'--_-----1 Fig. 6

The primary step consists in an anion exchange, combined

with hydrolysis of the resulting earth alkali chlorides.
Potassium sulfate formed by the anion exchange is re-
duced by carbon to potassium sulfide obviously at a much
higher rate than preformed potassium sulfate. Otherwise,
the low temperature activity of the
206

anion exchange system in comparison to potassium sulfate

would not be understandable. The surprising point is the
instantaneous hydrolysis of potassium sulfide to potas-
sium hydroxide, which was established by kinetic studies
and X-ray photoelectron spectroscopy (12,13). With the
potassium chloride/calcium carbonate system potassium
carbonate is formed in a similar way like potassium sul-
fate. Its conversion to potassium hydroxide may proceed
by primary dissociation and hydrolysis or by direct hy-
drolysis.

6. IMPROVED ACTIVATION BY APPLYING A CATALYST

Potassium sulfate, also available as a mineral, shows

a poor low temperature activity. This behaviour is sur-
prising, because the free enthalpies of potassium sulfate
reduction either by hydrogen and carbon monoxide or by
carbon are strongly negative as follows from Fig. 7.
100r-.---------.--------.-.

-100

-~~~~O~----~OO~O------~1=20~O~
T, K

~GR -VALUES FOR THE REDUC

ICT TION OF K2SO" WITH C, CO, 1985
H2.Fe.Ni AND Co. Fig. 7
 207

Therefore, it has to be assumed that these reactions are

kinetically inhibited. This kinetic inhibition should be
overcome by applying a catalyst. A screening test with
various catalysts showed that iron was most active. There-
fore, the action of this metal in catalyzing potassium
sulfate reduction was studied in detail.
Simultaneous thermogravimetric and differential thermo-
analytic studies on the reduction of various iron com -
pounds as raw materials during linear heating with
4. K·min- 1 in hydrogen showed that elemental iron is
formed between 400 and 700 DC. Reduction of iron sulfate
and iron oxide by carbonis thermodynamically favoured,but
may imply kinetic limitations. Analoqous studies with
potassium sulfate and potassium sulfate/iron sulfate mix-
tures in hydrogen or in the presence of a carbon powder
(CEYLON graphite) revealed that the formation of potas-
sium sulfide is lowered by addition of iron sulfate by
approx. 100 K ( 7 I 1 2 , 1 3) •
Further information on the influence of iron on the
reduction of potassium sulfate was received from study-
ing the melting behaviour (Fig. 8).

u ~?~.~.
0
750
w
0::: 0
~

«0:::
~ K2S04/Fe504
w
a...
~
w
~ 700
C)
z
5w K~04/FeCI2,
~

K2S04/Fe203
/\
650
Fe(SOi-) wt% METAL K( 504-)

MELTING TEMPERATURES OF
ICT K2SOl, AND K2SOl,/ Fe - SALT 1985
MIXTURES IN H21H20; XH1o=O.25;
0.1 MPa; 4 K m',n- 1 Fig. 8
 208

If potassium sulfate is heated on a graphite substrate

(polished surface) in a hydrogen/water vapour atmosphere
(XH20 = 0.25) with 4 K·min- 1 a melting point of 755 °c
is observed instead of 1069 °c in an inert atmosphere.
The reason is the formation of potassium sulfide, which
forms with potassium sulfate an eutectic with a melting
temperature of 610 °c (27). With potassium sulfate/iron
sulfate melting points below 755 °c were observed, being
the lower the smaller the iron amount was. This correla-
tion underlines a catalytic effect of iron. With the po-
tassium sulfate/iron oxide mixture the melting point is
still lower, because iron oxide is more easily reduced
to elemental iron than iron sulfate.
Results of gasification studies (2 MPa, hydrogen/water
vapour mixture) using the coal are given in Fig. 9.
100

K2SQ/FeSO,

"'52
~
E
.,- 50
m
0
E

I
u

60
"'~
"c
E
b 30
0
E
o·
u
0
500
T.oC I
I,min 60
COALLEOPOLDIK.-Fe-SALTS ; 5K,2Fe; H2/H 20 1985
ICT
xH,o;0.5 ; 2MPa ; 4K min-I; 850°C; 1h. Fiq. 9

Under the experimental conditions methane and carbon mon-

oxide are formed to a similar extent. A comparison between
the activities of potassium sulfate and potassium sulfate/
iron sulfate (5 mass parts potassium, 2 mass parts iron
per 100 mass parts coal) and potassium carbonate for re-
ference reveals that the low temperature activity of po-
tassium sulfate is improved and nearly comparable to that
 209

of potassium carbonate. The iron itself is inactive as

follows from the methane and carbon monoxide formations
with pure iron sulfate. Identical studies, but using an
argon/water vapour atmosphere, yielded the same gradua-
tion between potassium sulfate and potassium sulfate/iron
sulfate, but the low temperature activity of potassium
carbonate was decisively higher. This special low tempe-
rature activity of potassium carbonate may again be at-
tributed to the accelerated activation on the coal sur-
face (24,25).
In summary, it may be concluded that the technique of
catalyzing the activation process has also been shown to
be useful. For representation of the activation processes
with different iron salts in an oxidizing or reducing
atmosphere (argon/water vapour or hydrogen/water vapour)
the scheme shown in Fig. 10 was developed.

GENERAL REACTION SCHEME OF THE POTASSIUM

ICT j CATALYZED WATER VAPOUR GASIFICATION OF COAL
1985
INCLUDING ACTIVATION OF K2 Sq VIA CATALYSIS
1 Fig. 10
210

The first stage of activation represents the formation of

elemental iron. All reaction paths to elemental iron are
well established. For the interaction between iron and
potassium sulfate an oxygen transfer mechanism is pro-
posed. A spill-over mechanism with dissociative adsorp-
tion of hydrogen on the metal has been excluded, because
nickel and cobalt are rather inactive. On the other hand,
they are not able to act as oxygen acceptor, because the
oxygen affinity is too low (see Fig.7). A further argu-
ment is delivered by the catalytic effect of iron in an
argon/water vapour atmosphere, where hydrogen is missing.

7. CONCLUSION
Low temperature catalytic activity of potassium in
water vapour gasification of carbon or coal requires po-
tassium salts like potassium hydroxide or potassium car-
bonate, which are easily reducible to elemental potassium.
However, such compounds represent a high cost factor. The
use of low cost mineral raw materials like potassium
chloride and potassium sulfate implies the need to acce-
lerate their activation. This problem was studied by
three routes, all exclusively based on mineral raw mate-
rials. Route (1)makes use of such systems, which have a
lowered melting point as compared to pure potassium chlo-
ride. The basis of route (2) are anion exchange reac-
tions whereby more easily reducible potassium salts are
formed from potassium chloride, and route (3) is consi-
sting of a catalytic acceleration of the activation of
potassium sulfate.
All routes showed to be more or less successful. Lowe-
ring the melting point improves the low temperature acti-
vity by 100 K, but does not effect a satisfying accele-
ration of the gasification reaction. The anion exchange
reaction is most successful with the systems potassium
chloride/magnesium sulfate and potassium chloride/calcium
carbonate. Low temperature activity and catalytic acce-
leration of the gasification reactions are comparable
with those of potassium carbonate. Catalyzing the acti-
vation was successful in a similar way like the anion ex-
change principle. In spite of the promising results, it
should be pOinted out that this study is an exemplary
one, but it should open a broad field of research. The
usefulness of route (3) is studied in cooperation with
LURGI and financially supported by the Federal Ministry
of Research and Development.
 211

REFERENCES

1. McKee, D.W. 'The Catalyzed Gasification Reactions of

Carbon' in Chemistry and Physics of Carbon ~ (1981)1,
Eds. P.L. Walker, P. Thrower, Marcel Dekker, New York.
2. Pullen, J.R. 'Catalytic Coal Gasification', lEA Rep.
ICTIS/TR26, lEA Coal Research, London, 1984.
3. Gallagher, J.E., C.A. Euker. 'Catalytic Coal Gasifi-
cation for SNG Manufacture', Energy Res. 4 (1980)137.
4. Nahas, N.C. 'Exxon Catalytic Coal Gasification Pro-
cess', Fuel 62 (1983) 239.
5. Leonhardt, P~ A. Sulimma; K.H. van Heek, H. Jlintgen.
'Steam Gasification of German Hard Coal using Alka-
line Catalysis: Effects of Carbon Burn-Off and Ash
Content'. Fuel 62 (1983) 200.
6. Klihn, L., H. Plogmann. 'Reaction of Catalysts with
Mineal Matter during Coal Gasification',
Fuel 62 (1983) 205.
7. Adler-,-J., K.J. Hlittinger, R. Minges. 'Mixtures of
Potassium Sulphate and Iron Sulphate as Catalysts
for Water Vapour Gasification of Carbon, 2. Wettabi-
lity phenomena and reactions of the catalyst pre-
cursors.' Fuel 63 (1984) 1397 .
8. Hlittinger, K.J~ W. KrauB. 'Catalytic Activity of
Coal Minerals in Hydrogasification of Coal',
Fuel 60 (1981) 93.
9. Hlittinger, K.J., W. KrauB. 'Catalytic Activity of
Coal Minerals in Water Vapour Gasification of Coal'
Fuel 61 (1982) 291.
10. Hlittinger, K.J., P. Schleicher. 'Kinetics of Hydro-
gasification of Coke Catalyzed by Fe, Co and Ni'
Fuel 60 (1981) 1005.
11. Hlittinger, K.J. 'Fundamental Problems in Iron-Cata-
lyzed Coal Gasification - a Survey' Fuel 62 (1983)166.
12. Minges, R. 'Katalyse der Wasserdampfvergasung von
Kohlen durch Kalium - Untersuchungen zur Einflihrung
von Katalysatorrohstoffen mineralischen Ursprungs'
Ph.D.Thesis, University of Karlsruhe (FRG) 1985.
13. Hlittinger, K.J., R. Minges. 'Mixtures of Potassium
Sulphate and Iron Sulphate as Catalysts for Water
Vapour Gasification of Carbon. 3.Chemistry of the
in situ Activation of the Catalyst. Fuel 64 (1985)364.
14. Hlittinger, K.J., R. Minges. 'Catalytic Water Vapour
Gasification of Carbon, Importance of Melting and
Wetting Behaviour of the catalyst'. Fuel 64 (1985)491.
15. Hlittinger, J.J., R. Minges. Publication in-preparation
16. Hlittinger, K.J., R. Minges. 'Catalytic Activity of
Potassium Halides in Water Vapour Gasification of
Graphites', Fuel 63 (1984) 9.
212

17. Hlittinger, K.J., R. Minges. 'Entwicklung von Kataly-

satoren flir die Wasserdampfvergasung von Kohle und
Koksen'. Teil II: Kalium und Mischkatalysatoren.
Erd61 und Kohle - Erdgas - Petrochemie 38 (1985).
18. Kapteijn, F., J.A. Moulijn. 'Methanation-of CO over
Alkali Metal-Carbon Catalysts'. J. Chern. Soc. Chern.
Commun., 1984, 278.
19. Masling, B. 'untersuchungen zur Kohlenmonoxid-Kon-
vertierung und -Methanisierung an Aktivkohle-Kalium-
Katalysatoren'. Diploma thesis, University of
Karlsruhe (FRG) 1985.
20. Klemm, W., K. Beyersdorfer, J. Oryschkewitsch.
'Das Verhalten von NaCl, KC1, RbCl, NaJ und KJ zu
den entsprechenden Magnesiumhalogeniden'.
z. anorg. Chern. 256 (1948) 25.
21. Scheil, E., M. Stadelmaier. 'Untersuchungen liber die
Entmischung in Kaliumchlorid-Natriumchlorid-Misch-
kristallen'. z. Metallkd. 43 (1952) 227.
22. Syoltiro, Imatomi. 'Melting-Point of the System
MgC1 2 -Kcl-NaCl'. Bull. Inst. Phys. Chern. Research
(Tokyo) 19 (1940) 167
23. Hlittinge~ K.J., R. Minges. 'Water Vapour Gasifica-
tion of Carbon, Improved eatalytic Activity of Po-
tassium Chloride using Anion Exchange'.
Fuel 64 (1985) 486.
24. Mims,-C.A., J.K. Pabst. 'Role of Surface Salt Com-
plexes in Alkali-Catalyzed Carbon Gasification'
Fuel 62 (1983) 176
25. Huhn,F.M. 'Katalysatoren flir die Wasserdampfver-
gasung von Kohle und ihr EinfluB auf die Backfahig-
keit'. Dissertation, University of Essen, 1983
26. Kleinmann, D. 'Untersuchung zur Niedertemperatur-
katalyse der Wasserdampfvergasung von Kohle durch
Alkalimetalle'. Diploma thesis, University of
Karlsruhe (FRG) 1985
27. SO, C.W., D. Barham 'The System K2 S0 4 -K 2 S. A new
Experimental Technique'. J. Therm. Anal. 20 (1981)275.
 213

IRON-CATALYZED WATER VAPOUR GASIFICATION OF CARBON

K.J. Hlittinger, J. Adler, G. Hermann

Institut flir Chemische Technik

Universitat Karlsruhe
Fed. Rep. Germany

1. INTRODUCTION

Besides the alkali metals the iron metals are doubt-

less the most active catalysts of water vapour gasifica-
tion of carbon or coal. From a technical point of view,
iron seems to be the most promising iron metal combining
high activity with nearly unlimited availability. Waste
acids from titanium dioxide production and metal pickling
plants are preferred sources of iron salts, mainly of
iron sulphate.
The catalytic activity of iron has been described by
an electron transfer mechanism (1), by an oxygen trans-
fer mechanism (2-4), by a dissolution/precipitation me-
chanism (5) and a carbide mechanism (6). Activation ener-
gies of iron catalyzed gasification are often found to be
in the same order of magnitude as of non-catalyzed ga-
sification (approx. 350 kJ/mole). This may lead to the
conclusion that iron is not a true catalyst, increasing
only the concentration of active sites as discussed for
the earth alkali metals (7).
The application of iron as a catalyst of carbon or
coal gasification implies numerous problems, the primary
one being the dispersion of the catalyst raw material
(salt) on the carbon or coal surface. Agglomeration during
activation by reduction and during gasification has been
detected to occur (8,9), the consequence being a loss of
active sites. Catalytic activity requires a reducing
atmosphere to keep iron in the elemental, catalytically
active state. Sulfur present in coal either as
 214

constituent of mineral matter (disulfides pyrite and mar-

kasite) or of the orqanic substance (organic or hetero-
cyclic bound sulfur) is a strong pOison, even at high
gasification temperatures. Pressure has a negative ef-
fect on water vapour gasification but enhances methane
formation by hydrogasification.
The present paper is based on fundamental studies of
all these problems. Desorption measurements give some
insights into the mechanisms and rate limiting pteps of
non-catalyzed and iron-catalyzed water vapour gasifica-
tion. Gasification studies using various carbonaceous
materials as model carbons demonstrate technological a-
spects and problems of iron-catalyzed water vapour gasi-
fication.

2. EXPERIMENTAL
Gasification studies were performed using PVC-chars with-
out and with sulfur (prepared by pyrolysis of PVC*) or
PVC/sulfur mixtures), a calcined petroleum coke, a syn-
thetic graphite powder and subbituminous coals. All ma-
terials are described in literature (10-19). As catalyst
raw material iron sulfate was preferentially employed.
It was added by a vacuum impregnation technique or by
an ion exchange (10,16,20). Catalyst amounts were varied
between 0.04 and 2 wt%. Reduction of the iron sulfate
in a reducing gasification atmosphere (H2/H20) was
studied by thermogravimetric analysis showing that ele-
mental iron is formed at about 600°C (16,21).
Desorption measurements were performed by pregasify-
ing the coke (PVC cokes of different HTT) without and
with iron in H2/H20 atmospheres followed by quenching
with 300 K'min- 1 or 300 K'S- 1 to temperatures between
500°C and room temperature and measuring carbon mon-
oxide and hydrogen formation during temperature pro-
grammed heating with 20 K'min- 1 in an inert atmosphere
(22) .
Kinetics of gasification were studied with pure hydrogen
and hydrogen/water vapour mixtures of various composi-
tions using a fixed bed flow reactor operated at total
pressures between 0.1 and 2 MPa. The gasification reac-
tor and apparatus have been described in earlier papers
(17,18). Temperature programmed heating (4 and 15 K'min- 1 )
allowed to identify the starting temperature of the
gasification reactions and to follow the gasification

*)polyvinyl chloride
 215

rate over the temperature range up to 1000 DC by simul-

taneously analyzing the gaseous reaction products with
the aid of IR analyzers.

3. DESORPTION STUDIES AND MECHANISMS

3.1 Non-catalyzed

Desorption studies were performed with cokes produced

by pyrolysis of PVC at 600 DC and further heat treat-
ments at 800, 900, 1000 and 1500 DC. These cokes have the
advantage to be free of heteroatoms and ash forming ele-
ments.
Desorption studies with pure cokes by measuring carbon
monoxide formation rate after pregasification showed a
single peak for each coke, which was practically inde-
pendent of the gasification temperature and quench con- -1
ditions (300 K·min- 1 in hydrogen/water vapour or 300 K·s
in nitrogen), but shifted to higher temperatures with
increasing heat treatment temperature. After applying a
Fast FOURIER Transformation (FFT) for eliminating the
peak broadening caused by the flow system and the IR ana-
lyzer for measuring carbon monoxide *), activation ener-
gies E and frequency factors ko were calculated on the
basis of a first order rate law. The values given in
Table 1 let recognize that the activation energy and the
frequency factor increase with increasing heat treatment
temperature of the coke. Carbon monoxide measured results
from oxygen of ether bridges which are formed by conden-
sation of phenolic hydroxyl groups (Fig.1).
The activation energy thus describes the thermal decom-
position of these surface groups which are extremely
stable (decomposition temperature above 800 to 850 DC).
Gasification of cokes in a hydrogen/water vapour atmo-
sphere (XH20 = 0.1) is starting at about 700 to 750 DC,
nearly independent of the heat treatment temperature of
the coke. Activation energies of gasification calculated
from non-isothermal kinetics (linear heating) are also
shown in Table 1. They are independent of the heat treat-
ment temperature of the coke. The average value of about
335 kJ·mole- 1 is in good accordance with the values
known from literature (24-26). During gasification in the
presence of water the ether bridges are splitted above
700 to 750 DC, which leads to the gasification cycle

*)Method and results are given elsewhere (22,23)

216

temperature
H

H20 ~(O)
' T "~rgJ-n~~.j--
}@ CO

. ) lb~~'
-('),
_ ,
H20

8dC(H2) ~ (OH)
. "~C(H)
~ dissociatIon of water..
.. atether bridges
desorption
of CO
... at active sites
H (slow) (fast)
formation of active formation of
sites with dissociation stable ether
ofwaterlcoal or char) bridg~

Fig.1
I C T -) Reaction schema of non- catalyzed water vapour. gasification of cool or char 11985

shown in the right hand side of Fig.1. As a consequence

of the experimental results and ki~etic calculations it
may be assumed that dissociation of water represents
the rate limiting step of non-catalyzed water vapour
gasification.

-1
Activation energies E (kJ mole ) and frequency factors
ko (min- 1 )of co formation by desorption in N2 and
gasification in H2/H20 (X H20 = 0.1).

1)
HTT of PVC-cokes Desorption Gasification
°c E ko E

600 393 1.3.10 18 301 2)

900 418 8.5.10 18 359
1000 481 1.2.10 21 340
1500 543 4.4.10 22 343

1)non-isothermal
2)HTT below starting temperature of gasification
Tab.1
 217

3.2 Catalyzed
Desorption studies with the same cokes as above but
with iron (approx. 2 wt%, added by vacuum impregnation
with aqueous iron sulfate solution) yielded much more
complex carbon monoxide formation patterns. After quen-
ching in hydrogen/water vapour atmosphere (XH20 = 0.1)
with 300 K·min- 1 three further carbon monoxide formation
peaks were detected (peak maximum between 550 and 600°C,
650 and 700 °C, 750 and 800°C), besides the high tempe-
rature carbon monoxide desorption peak of non-catalyzed
gasification experiments; the intensities of the carbon
monoxide formations in the three ranges varied with the
quench temperature. At lower temperatures (350°C) the
peak of the first range was maximum and that of the third
rate was minimum. At higher temperature (500 °C) it was
inverse. After quenching with 300 K·s-1 only the peaks of
range two (with very small intensity) and range three be-
sides that of non-catalyzed gasification experiments were
found.
Carbon monoxide desorptions are ascribed to three dif-
ferent kinds of iron oxygen surface complexes. Gasifica-
tion experiments in hydrogen/water vapour after quenching
showed that carbon monoxide formation may start at 600°C
corresponding to the second desorption peak, but this
gasification is of transitionary nature. Steady state
gasification is possible only above 700°C corresponding
to the third carbon monoxide desorption peak. This be-
haviour lets assume that oxygen transfer from the iron
to the carbon surface represents the rate limiting step
of iron-catalyzed gasification, because the less stable
iron oxygen surface complexes of range two are trans-
formed to the more stable complexes of range three in a
water containing atmosphere.
Experimental data were mathematically treated in the
same manner as described in the case of non-catalyzed
gasification. The results are given in Table 2. It follows
that the activation energies E and frequency factors ko
determined for carbon monoxide desorption in all three
ranges are not depending on the quench temperature, but
the actual energies increase'from range to range. The
activation energy of carbon monoxide desorption in range
three amounts to 345 kJ·mole- 1 , it is lower than the
activation energies of gasification, calculated from non-
isothermal and isothermal gasification experiments (360
or 390 kJ·mole- 1 ). This relationship between activation
energies of desorption and gasification again confirms
that oxygen transfer plays the dominating role in iron
catalyzed water vapour/gasification, because the iron
218

Activation energies E (kJ·mole- 1 ) and frequency factors k (min- 1 )

of CO formation by desorption in N2 and gasification in H~/H20
XH20 = 0,1), PVC-coke (900°C) with 2 wt% Fe

Desorption
Quench Temp., °c
(Residence Time Range 1 Range 2 Range 3
in N2 1h) .
E ko E ko E ko

350 255 3.5.10 17 280 6.0.10 16 347 1.0.10 18

400 247 1.4.10 17 276 3.3.10 16 360 6.2.10 18
450 251 2.2.10 17 288 1.0.10 17 356 4.8.10 18
500 238 1.4.10 16 276 3.5.10 16 318 1.3.10 17

248 280 345

Gasification

E E

non-isothermal 310 360 Tab.2

isothermal 390

oxygen surface complexes are stabilized in the water va-

pour atmosphere in comparison to a nitrogen atmosphere.
The results may be summarized in a reaction scheme of
iron-catalyzed water vapour gasification (Fig.2), which is
based on an oxygen transfer mechanism. This mechanism is
supported by the experimental data as well as by the
kinetic calculations.

temperature 600°C 700°C

I (fast)

K
C(O)

(SIOW)
H20 H2 H20 H, H20 H2
\ ~ : A . (fast) A
Fe ~ Fe(O)I~Fe(O)n~ Fe(O)m Fe

\C""~. ~ ~ (fast)
f'\
H, ' {

H2 0 J\.. Fe A C(O)

t
CO
H2 H2 0

transitionary steady state

low temperature high temperature
gasification gasif icat i on

I CT IReaclion scheme of iron-catalyzed water vapour gasification of coal or char

(oxygen transfer mechanism)
1 1985
Fig.2
 219

4. GASIFICATION STUDIES AND TECHNOLOGICAL PROBLEMS

4.1 Influence of water vapour partial pressure and the

aspect of auto-catalysis

From the mechanism shown in Fig.2 it is obvious that

iron is catalytically active for dissociation of water only
in the elemental state. Oxygen transfer being the rate
limiting step of iron-catalyzed gasification requires a
minimum temperature between 700 and 750°C. As a conse-
quence, gasification is possible only in a reducing atmo-
sphere, for example by using hydrogen/water vapour mix-
tures. Equilibrium constants and conversions have been
calculated (3,27,14). Recycling of hydrogen and carbon
monoxide formed during gasification has shown to be
possible (28).
The influence of water vapour partial pressure was ex-
perimentally studied by gasification of PVC cokes of
different heat treatment temperature (iron content
1.5 wt%, impregnation with iron sulfate solution) as
model carbons (20) with hydrogen/water vapour mixtures
of various compositions. The temperature was raised at a
rate of 4 K·min- 1 up to final temperatures between 750
and 950°C, which were hold for 1 h. Fig.3 shows the re-
lative mass losses after gasification as function of the
water vapour mole fraction.

-E
E
1
~-.\ \
<l

0.5

~O ° °L~ O~
~9~V
00 0.2 0.4 0.6 0.8
X H20

PVC - coke (60Qoc) ; Fe ~ 1.5 wt % (impr.); \1981

I CT\ H 2 /H 20; 0.1 MPo; 4Kmin-'; lh. Fig.3

It follows, that the range of total gasification (6m/mo =1)

is decisively extended to higher water vapour mole frac-
tions by increasing the final gasification temperature.
This means that gasification is also possible with such
hydrogen/water vapour mixtures in which iron would be
oxidized to wustite or even to magnetite. These results
220

are understandable if it is assumed that the activation

energy of oxygen transfer from the iron to the carbon
surface is higher than that of water dissociation on the
iron surface. The high activation energies of oxygen
transfer have been demonstrated by desorption measure-
ments. On the other hand, it is obvious that the reaction
products hydrogen and carbon monoxide, formed at a higher
rate ,are stabilizing elemental,catalytically active iron.
This feedback gives to the system an autocatalytic charac-
ter. Isothermal oscillations have been observed in the
experiments. In endothermic reactions like water vapour
gasification they are only possible if the mechanism is
an autocatalytic one (irreversible thermodynamics).
In comparison to cobalt and nickel iron is the most ac-
tive catalyst as far it may exist in the elemental state.
The reason for that is a higher oxygen affinity of iron
and thus a higher activity in dissociation of water. Co-
balt and nickel, which are not oxidized by water, are
applicable and active also in gasification with pure
water vapour.

4.2 Influence of metal dispersion and catalyst requirements

Catalyst addition by impregnation of coals or cokes

with aqueous iron salt solutions gives a sufficient but
not an optimum dispersion of the metal (5,14). Agglomera-
tion during gasification has been found to occur and be-
ing the stronger the worse the original dispersion was.
As found by gasification of a calcined petroleum coke with
a negligible content of ash forming elements (hydrogen/
water vapour mixture, XH~O = 0.2, 0.1 MPa, 940°C, 1 h),
0.5 wt% metal (Fe, Co, N1) are sufficient for effectinq
maximum gasification rates (Fig.4). Higher catalyst
amounts have neither a positive nor a negative effect.

0,4

E ,-.c." . .• .. • " . Fe
Co
E
<I 0,2 l! . - - - Ni
rr~
00 1 2 3
catalyst metal, wt %

I CT petroleum coke (calc.); H2/H2 0; XH20= 0.2;1 1982

0.1 MPa; Tmax= 940°C; 1h. Fig.4
 221

An ion exchange yields a much better dispersion of the

metal. Studies were performed using a synthetic graphite
powder(LONZA T 44), which was pretreated with boiling
60 wt% HN03 (16 h) to produce exchangeable carboxylic
surface groups (16). Ion exchange with iron sulfate so-
lution (pH 7) (11) may occur equivalent or equimolar
(Fig.5).

ion exchange
(acc. to Boehm)

COO-Fe2. sot equlmolor

H·

equivalent
FeS0l, SOlution>
(0002 molor)

OH

leT ion exchange at the oxidized graphite surface 1962

Fig. 5

Iron contents up to 0.04 wt% were achieved. Gasification

of such a graphite with 0.04 wt% iron (hydrogen/water
vapour, xH20 = 0.1, 0.2 MPa, 4 K.min- 1 , 900 °C, 1 h) was
much more effective than that of a graphite vacuum im-
pregnated with iron sulfate solution yielding an iron
content of 2 wt% (Fig.6).

..m/ lT'o=0.540

J 40
'"E
.. m/mo= 0.522
-;~ 30
g
1.9 Fe
8 (vac. impr.)
20
0.04 Fe
I ion exchange)

10

leT graphite I Lanza T44); H,I H,O ; XH,O =0.1 ; 02 MPa ;

1961
4 K min- 1; Tmo,= 900°C; 1h . Fig. 6
 222

Besides the low temperature activity of the ion exchanged

graphite shown in Fig.6, the excellent atomic dispersion
of the iron also effected an improved activity at high
water vapour partial pressures, as follows from gasifi-
cation studies under the same experimental conditions as
above, but with increasing water vapour mole fractions
(Fig. 7).

30 0 0.04 Fe (ion exchange)

~ __-----LI--~i~~
7: ,(I- 0

~ I
o 15 , "
(.)
,
I,

pure graphite
/;. /
0.25 0.50 0.75

graphite (Lanzo T44), Hz' HzO , 0.2 MPo ,

1981
ICT 4 Kmin-1 , Tma x=900oc; 1h. Fig. 7

An ion exchange with coals is restricted to brown

coals due to the high amount of carboxylic groups. Coals
of higher rank have to be activated, but it is question-
able if strong acid carbocylic groups, necessary for an
ion exchange, can economically be produced, for instance
by a low temperature treatment in humid air

4.3 Influence of sulfur and catalysis with iron from

disulfides of mineral matter

Sulfur known to be a strong catalyst poison is con-

tained in coals as organic, preferentially heterocyclic,
and inorganic bound sulfur. The main sources of inorganic
bound sulfur are the iron disulfides pyrite and marka-
site.
Studies on the pOisoning effect of organic bound sul-
fur were performed with sulfur containing PVC chars as
model materials prepared by co-pyrolysis of PVC and ele-
mental sulfur at 600°C (residence time 2 h). It was
found that the sulfur content of the char is proportional
to the sulfur content of the mixture. The influence of
sulfur on the microstructure of chars was studied by
optical microscopy (16). The basic reactivity of the chars
 223

during gasification without iron was not changed up to

approx. 0.3 wt% sulfur.
Results of gasification studies with chars of diffe-
rent sulfur contents and 2 wt% iron (hydrogen/water vapour
XH20 = 0.5, 1 MPa, 4 K.min- 1 , 880°C, 1 h) are shown in
Fig. 8. It follows that the catalytic activity of iron
found with the sulfur free char is strongly diminished
by 0.06 wt% sulfur and completely inhibited with 0.3 wt%
sulfur only.

30

"0
E, 10
:r;"
u

O~~~ __- L_ _ _ _- L_ _ _ _L -_ _ _ _ _ _ _ _ _ _ _ _ ~

100
Owl%S

0~~600~--~7=00~--~~~~~~~~~~;J
T. 'C
30 I.min 60
PVC coke;PVClS cokes; 2 wt% FoISO,'); H,/H,O; XH 0=0,5; 1MPo;
leT 4Kmin'';Tm,,= 8800C;lh. ' 1983
Fig. 8

Further studies under the same conditions but increasing

temperatures showed that the inhibiting effect of sulfur
may partially be compensated. Gasification studies with
900°C final temperature, a total pressure of 1 MPa and
varied water vapour mole fractions using the char with
0.3 wt% sulphur,demonstrated that the inhibiting effect
is decreasing by lowering the water vapour content of the
gasification. medium. A most tremendous compensation of
the inhibiting effect is achieved by increasing the total
pressure (studied with xH20 = 0.25) as follows from Fig.9.
The reason for the effects of temperature and especially
of hydrogen partial pressure and total pressure is doubt-
less a hydrodesulfurization, which is favoured by hydroger
224

and especially total pressure. Sulfur determinations of

gasification residues confirmed this interpretation.

60

51
x 40
'c
E
....0.
"0
E 20 ... -"-- ...........
i!
u "",.", '--- ----- ......................
/ ... _--
-..,,::::'...-:.::.:.~ ....... ~""""-""'''' ........... ~ ...... ~
O~~=---~--~----~~~~------~~
1~r-.-----.----,-----.--------------.

100
'5x?
'c
E
-'-en
~ 50
ci
u

o 600 700 BOO

o 30 t.min 60
leT PVC/S coke(600"C),0.3%S,2Fe(SO~-),H2Ifi20,xH,o·0.25, 4Kmin~,
Tm.,·900 ·C, lh ...... ·0.2MPo, ---1MPo, -2MPa 1984 Fig. 9

Nevertheless, it is astonishing that sulfur is such a

strong poison at temperatures beyond 800 °c. The expla-
nation directly follows from thermodynamic considerations
on the stability of iron sulfur surface complexes. As de-
monstrated by Fig.10 the stability of iron sulfur surface
complexes is two orders of magnitude greater than that of
bulk iron sulfide at 800 °c. This well explains the inhi-
biting effect of sulfur on the catalytic activity on
iron, namely by blocking of active sites. Poisoning of
iron by a gas phase transport of hydrogen sulfide was si-
mulated by gasifying a mixture of a sulfur free coke
(80 wt%) and a sulfur containing coke (20 wt%, 1.5 wt%
sulfur) resulting in an average sulfur content of 0.3 wt%.
The effect of inorganic bound sulfur could easily be
detected by studying the catalytic effect of iron as con-
sti tuent of min.era.l metal of coal. For this purpose a
subbituminous coal (LEOPOLD) with 0.50 wt% iron (existing
 225

-2.-.---.-----.-------~
800 600 400 T."C

0.8 1.2 1.6 2.0

lIT. K-'.1O-3

= ==
stability ranges of Fe/Fe -Sods / FeS
leT Fe + H2 S Fe-SodS FeS + H2 1985
19 • coverage degr.... I
Fig. 10

namely as iron disulfide), 4.8 wt% ash and 1.3 wt% total
sulfur was gasified at different temperatures in hydrogen/
water vapour mixtures of different compositions (total
pressure 1 MPa) without and with pretreatment for desul-
furization. As follows from Fig.11 it takes at 880°C
more than 90 min until (catalyzed) gasification is star-
ting. This time period is obviously necessary in order
to form under these conditions catalytically active iron.
From the standpoint of thermodynamics, this desulfuriza-
tion of the iron disulfides should be possible also at
500°C .

..~ l,
,<>
E
ci
u 2

30 120 150 11Klt.llW\

leT Leopold coal. H2/H20. XrlzO= 0 5,01 MPo.4 K minwl. li.J0II:= B80oe. wlthoutAvith residence time 1980 Fig. 11
226
An interruption of heating at 500°C for 90 min effec-
ted that the gasification reaction started even before
880 °c and the gasification rate was found to be higher
than that observed after 90 min residence time at 880°C.
Desulfurization by iron trichloride according to the
equation
Fe + 3 FeCl 2 + 2 S2
before gasification was as successful as the ther-
mal desulfurization at 500°C (Fig.12)

E
i
u 2

,...... -~
\
.' .~

~~ 4
,, ;,.- ... ,
"'e:n
E ./ \
\ I
I ....
\

a ,/ \ I \
u 2 ,.i v
r' \
J
I
\
---.. ,-----:..:.~~~.~>/.~....
1OQO r.,"c
800
o 30 60 90 120 150
,eot.min
C Leopold coal; H, iH, 0; XH20~ 0.5; 0.1 MPo; 4 K min- 1 ; 1;.., ~ 880·C, 1981
I T ___ untreated; _._._ FeCI,; _ FeCI, ,washed;· FeCI ,woshed,HF/HCI. Fig. 12

In this case, an additional acceleration of the methane

formation but not of the carbon monoxide formation was
found. This improved methane formation results from the
additionally added iron (iron dichloride is easily re-
duced to elemental iron under gasification conditions) ,
because the necessary iron amounts for maximum methane
formation by hydrogasification are twice as high than
those for carbon monoxide formation by water vapour gasi-
fication. Finally, it should be noted that the inhibi-
ting effect of sulfur may not be compensated by enlar-
ging the iron content.
In summary,it follows that sulfur in coal represents
a strong poison for iron catalysis. The problem of organic
bound sulfur can best be governed by applying a hydrogen
rich gasification atmosphere and a high pressure, condi-
tions favouring hydrodesulfurization. Temperature has
only a minor effect. The inorganic bound sulfur mainly
of iron disulfides can successfully be removed by chemical
 227

treatment with iron trichloride or by a thermal treat-

ment at about 500 °c in a reducing (hydrogen rich) gasi-
fication atmosphere. After these operations, the iron of
mineral metal represents a useful catalyst.

5. CONCLUSION

Iron represents a useful catalyst for water vapour

gasification of carbon. Methane is formed by hydrogasi-
fication and not by methanation of carbon monoxide. The
catalytic activity of iron can completely be described
by an oxygen transfer mechanism with oxygen transfer as
rate limiting step. Due to this limitation iron may only
represent a high temperature catalyst with a minimum
temperature above 700 °C. Reducing conditions are funda-
mentally necessary in order to keep iron in the elemental,
catalytically active state,which is achievable by recyc-
ling carbon monoxide and hydrogen. Accelerating the oxy-
gen transfer by increasing the temperature allows the
application of a gasification medium with more than 50 %
water vapour. This is another reason why iron may only
be used as a high temperature catalyst.
In gasification of coals sulfur represents a strong
poison. The problem of organic bound sulfur may be solved
by applying a high pressure and a hydrogen rich gasifica-
tion medium in order to favour hydrodesulfurization.
Temperature has an additional effect. Inorganic bound sul-
fur mainly connected with the iron disulfides pyrite and
markasite may be removed by chemical desulfurization with
iron trichloride or by thermal desulfurization in a re-
ducing gasification medium at approx. 500 °C. After such
a treatment the iron present in most of the coals (approx.
0.5 wt% of the coal) is sufficient for a maximum cataly-
tic effect.

Acknowledgement
The research of iron-catalyzed water vapour gasifica-
tion was supporied by the Federal Ministry of Research
and Development.
228

REFERENCES
1. Long, F.W. and K.W. Sykes. The catalysis of the oxi-
dation of carbon. J. Chern. Phys. 47 (1950) 361.
2. Kroger, C. Die Vergasung von Kohlenstoff durch Luft,
Kohlendioxid und Wasserdampf und ihre Beeinflussung
durch anorganische Katalysatoren. Angew. Chern. 52
(1939) 129.
3. McKee, D.W. Effect of metallic impurities of the ga-
sification of graphite in water vapour and hydrogen.
Carbon 12 (1974) 453.
4. Tarnai, Y., H. Watanabe, and A. Tomita. Catalytic ga-
sification of carbon with steam, carbon dioxide and
hydrogen.Carbon 12 (1974) 453.
5. Hahn, R. Kinetische Untersuchungen und thermodyna-
mische Berechnungen zur Entwicklung eines technischen
Verfahrens fUr die eisenkatalysierte Wasserdampf-Ver-
gasung von Koks. Ph.D.Thesis, Univ. Karlsruhe, 1979.
6. Holstein, W.L. and M. Boudart. Transition metal and
metal oxide catalyzed gasfication of carbon by oxygen,
water, and carbon dioxide. Fuel 62 (1983) 162.
7. Otto, K., L. Bartosiewicz and M.-Shelef. Effects of
calcium, strontium, and barium as catalysts and sul-
phur scavengers in the steam gasification of coal
chars, Fuel 58 (1979) 565
8. Rewick, R.T.-,-P.R. Wentrcek, and H. Wise. Carbon
gasification in the presence of metal catalysts.
Fuel 53 (1974) 274.
9. HUttinger, K.J. and J. Adler. Untersuchungen zur Ver-
wendung von eisenhaltigen Abfallsauren fUr die kata-
lytische Vergasung von Kohlen und Koksen mit Wasser-
stoff und Wasserdampf. Research Report T 83-240,
Federal Ministry of Research and Development (BMFT),
1983
10. HUttinger, K.J. and R. Minges. Entwicklungen von Kata-
lysatoren fUr die Wasserdampfvergasung von Kohlen und
Koksen, Teil 1: Eisen und Eisenmetalle.Erdol und
Kohle - Erdgas - Petrochemie 37 (1984) 493
11. Hlittinger, K.J. and H. Kirrmann. Modelle fUr die hy-
drierende Vergasung von Steinkohlen III - Chemie der
Druckvergasung von Aromaten. Erdol und KOhle - Erdgas-
Petrochemie 35 (1982) 17.
12. Kirrmann, H. Modelluntersuchungen zur Methanbildung
bei der hydrierenden Verga sung von Steinkohlen.
Ph.D.Thesis, Univ. Karlsruhe, 1980.
13. HUttinger, K.J. and P. Schleicher. Kinetics of Hydro-
gasification of Coke Catalyzed by Fe, Co and Ni.
Fuel 60 (1981) 1005.
 229

14. Schleicher, P. Katalyse der Koksvergasung mit Wasser-

stoff und Wasserdampf durch Fe, Co und Ni. Ph.D.Thesis,
Univ. Karlsruhe, 1980.
15. Adler, J., K.J. Hlittinger and R. Minges. The cataly-
tic activity of iron in water vapour gasification of
sulphur containing cokes. Fuel (in press)
16. Adler, J. Untersuchungen zur Katalyse der Vergasung
von Kohlenstoffen durch Eisen, Kalium und Mischkata-
lysatoren aus Eisen und Kalium. Ph.D.Thesis, Univ.
Karlsruhe, 1984.
17. Hlittinger, K.J. and W. KrauB. Catalytic activity of
coal minerals in hydrogasification of coal., Fuel 60
(1981) 93. -
18. Hlittinger, K.J. and W. KrauB. Catalytic activity of
coal minerals in water vapour gasification of coal.
Fue 1 6 1 (1 982 ) 29 1 .
19. KrauS-,-W. tiber den katalytischen EinfluB von anorga-
nischen Kohlebestandteilen bei der Verga sung von
Kohlen mit Wasserstoff und Wasserdampf. Ph.D.Thesis,
Univ. Karlsruhe, 1982.
20. Hlittinger, K.J. Fundamental problems in iron-catalyzed
coal gasification - a survey. Fuel 62 (1983) 166.
21. Minges, R. Katalyse der Wasserdampf-Vergasung von
Kohlen durch Kalium - Untersuchungen zur Einflihrung
von Katalysatorrohstoffen mineralischen Ursprungs -.
Ph.D.Thesis, Univ. Karlsruhe, 1985.
22. Hermann, G. Mechanismen der nichtkatalysierten und
durch Eisen katalysierten Wasserdampfvergasung von
Kohlenstoff. Ph.D.Thesis, Univ. Karlsruhe, 1985.
23. Hermann, G. and K.J. Hlittinger. Correction of measured
gas formation peaks in non-isothermal kinetics by
FTT for evaluation kinetic parameters (in preparation) .
24. Long, F.J. and K.W. Sykes. The mechanism of steam-
carbon reaction. Proc. Roy. Soc. A 193 (1948) 397.
25. Rossberg, M. Experimentelle Ergebnisse liber Primar-
reaktionen bei der Kohlenstoffverbrennung. Z. Elektro-
chern. 60 (1956) 952.
26. Walker, P.L., F. Rusinko and L.G. Austin. Gas Reac-
tions of Carbon. Adv. in Catal. 11 (1959) 133.
27. Hahn, R. and K.J. Huttinger. Die~asgleichgewichte
der Koksvergasungsverfahren in geschlossener Dar-
stellung. Chem.-Ing.-Tech. 51 (1979) 501.
28. Hahn, R., K.J. Hlittinger an~W. Vohmann. Thermodyna-
mik der Koksvergasung bei Kreislaufflihrung von CO
und H2-Katalyse durch Ubergangsmetalle. Erd6l und
Kohle - Erdgas - Petrochemie 32 (1979) 179
 231

METAL CATALYZED GASIFICATION OF GRAPHITE

R. T. K. Baker
Exxon Resea rch and Engi neeri ng Company
Corporate Research Science Laboratories
Cl i nton Townshi p, Route 22 East
Annandale, New Jersey 08801

1. INTRODUCTION
One of the most interest i ng and i nt ri gu i ng topi cs in the
field of heterogeneous catalysis is the mechanism by which
metallic particles influence the gasification of graphite. The
gasification of carbon by oxidizing gases is often the limiting
factor in the performance of graphite structures at hi gh tem-
peratures. Currently, one of the most important areas involves
the function of metals in the removal of carbonaceous deposits
from spent catalysts and steam cracker tubes. Supported metal
catalysts are employed in a variety of commercial hydrocarbon
conversion processes. Such catalysts consist, in general. of
small metal crystallites (0.5 to 2.5 nm diameter) dispersed on
nonmetallic oxide supports. In the absence of such poisons as
sulphur, arsenic, phosphorous, and heavy metals, the major way
in which the catalyst loses activity is due to accumulation of
carbonaceous residues on its surface. catalyst regeneration is
usually achieved by gasification of the deposit in an oxidizing
environment at about 500°C. The efficiency of this process is
variable and so there is an increasing demand to learn more
about the process so that gasification can be optimized.

In recent years research acti vity in graphite gasifi cati on

experienced a tremendous resurgence because of its relevance to
catalyzed coal gasification processes. However. this area has
tended to lose its popularity of late due to the changing world
supply of oi 1.
232

In this paper, the discussion of the gasification behavior

will be restri cted to graphite. Graphite was selected as the
carbonaceous medium, as it can be obtained as natural single
crystals and prepared for study in a highly pure form. From a
practical standpoint, it is a useful material to investigate
since it is the most oxidation resistant component in most
carbonaceous deposits.

<1010"

Zig-Zag
Face
{1010}Planes

Armchair Face
{1120} Planes

Basal

® Plane Basal
Plane

Twin Band

FIG. 1 Schematic illustration of the graphite basal plane

with crystallographic directions and atomic planes,
and the formation of a twin band.
 233

Before proceeding further, it is felt worthwhile to remind

the reader of some of structural features of graphite. The
structure of graphite consists of layers, parallel to the basal
plane, of hexagonally bonded carbon atoms. where the stacki ng
pattern of the I ayers follows the sequence ••• a b a b ••• •
The distance between carbon atoms in the basal plane is 0.142
nm and the distance between equivalent layers is 0.669 nm. If
we consider the basal plane, Figure lA, it can be seen that
there are two arrangements of carbon atoms, the "armchai r"
configuration, which is aligned along the <laTa> crystallo-
graphic directions. and the "zig-zag" configuration which is
parallel to the <1120> crystallographic directions. The c-axis
is per~endicular to the basal surface, as are the {laTa}
and {1120} family of planes.

When graphite specimens are examined in the transmission

electron microscope, the orientation of various features
created by catalytic impurities can usually be determined by
reference to the position of twin bands which are always
present in the crystal. Twinning is a result of a rotation
about an axis in the basal plane and along the -C-C-bonds. as
shown in Fi gure lB. When vi ewed in the mi croscope, the twi n
band appears as a black line and is always aligned along
<laTa> directions.

2. EXPERIMENTAL PROCEDURES
The main limitation of a conventional transmission elec-
tron microscope is that materials can only be viewed in a
vacuum condition. In an attempt to combat this situation the
tactic is to examine a specimen before and after reaction. or
over periods of time. with the result that the investigator is
confronted with the problem of trying to decide from a compari-
son of mi crographs what has taken pI ace duri ng the i nterva Is.
While this procedure can yield extremely valuable information.
some caution must be exercised in interpretation, since the
specimen has been removed from the reaction environment and, as
a consequence. undergone cooling. Moreover. during examination
periods there is the possibility of the specimen becoming con-
tami nated and subsequently exhi biti ng unpredi ctab 1e behavi or.
There is also the more mundane problem of relocating areas of
interest in the electron microscope between successive periods
of reaction.

These di ffi culti es and drawbacks are readily overcome by

controlled atmosphere electron microscopy (CAEM). This tech-
nique enables the researcher to continuously follow the changes
234

in appearance of a specimen while it is being heated in the

presence of a gas. Details of the CAEM technique are already
well documented (1)(2) and will not be dealt with here. The
key design feature is the ability to operate the microscope at
high gas pressure in the specimen region while maintaining very
low pressure in the rest of the column. This aspect is accom-
plished by use of a cell which allows for differential pumping
around the specimen.

In order to capture the dynamic information generated in

these studies, sophisticated continuous recording facilities
have been incorporated into the system. Part of the transmis-
sion image is viewed by a Phillips Plumbicon TV camera equipped
with a fiber optics lens system. The magnified image is dis-
played on a monitor and continuously recorded on video tape.

It should be appreciated that CAEM is not just a technique

which provides a qualitative picture of the changes in appear-
ance of a specimen undergoing reaction, but in many cases it is
possible to measure the rates at which such changes occur.
Moreover, since there is close control of gas pressure, speci-
men temperature, and reaction time, these measurements are
amenable to kinetic interpretation. CAEM is therefore a power-
ful tool for gaining a unique insight into gas-solid reaction
mechanisms.

All the experiments reported in this study were performed

using natural single crystal graphite obtained from Ticon-
deroga, New York State. The crystals were extracted from the
pyroxene mineral in which they were embedded by the method
described by Hennig (3). They were first cleaved between glass
slides coated with a polystyrene cement and then released and
washed with acetone. These precleaved sections were then
mounted on glass slides with aqueous polyvinylpyrrolidone (PVP)
adhesive. The PVP was allowed to harden under a heat lamp and
the crystals further cleaved with Scotch tape. Successive
layers were removed until an optically transparent portion of
the crystal remained affixed to the slide. Such crystals,
which varied in thickness from 15 to 100 nm, were found to be
excellent transmission specimens. The cleaved crystals were
released from the slide onto a clean water surface and picked
up on platinum heater ribbons.

Two methods were used for the i nt roduct i on of controlled

amounts of metals onto the graphite surface. Vacuum evapora-
tion of spectrographically pure metal wire from a 6clean tungs-
ten filament at a residual pressure of about 10- Torr. The
amount of metal and distance separating the filament from the
 235

specimen were selected so as to produce an approximate mono-

layer coverage on the graphite. fJn alternative procedure was
to spray the graphite with a dilute solution of a metal salt
usi ng an atomi zer. Unfortunately, thi s approach di d present
difficulties with the control of metal loading on the graph-
ite. Subsequent treatment of both types of specimens at tem-
peratures of about 500°C generally resulted in nucleation and
growth of discrete particles from the deposited metal films.

3. FUNDAMENTAL ASPECTS OF CATALYTIC GASIFICATION

3.1 The Metal-Graphite-Gas Interface

In general, for particles to become activated they must be

located at edges or steps on the graphite, since these are
regions of high co-ordination number particles tend to accumu-
1ate at such sites. Regi ons of hi gh co-ordi nat i on can a 1so be
found at imperfections on the graphite basal plane, e.g.,
vacancies, and can also serve as centers for catalytic
attack. However, since the number of such sites is generally
quite small compared to those available at edges and steps,
they only make a minor contribution to the overall reactivity
of a given system.

8·90"
YMG

NON· WETTING WETTING SPREADING

NO REACTION REACTIVE VERY REACTIVE

FIG. 2 Changes in particle shape as a function of the degree

of wetting of the catalyst on graphite.

The mode by whi ch a cata lyst operates is governed to a

large degree by the strength of the interaction between the
metal and the graphite edge atoms. The factors controlling the
degree of interaction of metal particles with a graphite
support can be most easily understood by consideration of the
particle morphological situations depicted in Figure 2, and
described mathematically by Youngls equation. For a metal-gas-
236

support system at equilibrium:

(1)

where e is the contact angle between the metal particle and the
support, Y, is the surface energy, and the subscripts s, m and
g refer to support, metal and gas, respectively.

Equation (1) can be written in the form

cos e = (2)

Ymg

If Ymg is 1a rger than Ygs ' the contact angl e wi 11 be greate r

than 90° and the particle will be in a non-wetting configura-
tion; if the reverse relationship holds, then e will be less
than 90° and wetting wi 11 occur. and if Y = 1. + 1. • then
gs ms mg
the particle will spread out over the support surface. This is
a simple thermodynamic approach to the morphological behavior
of particles on supports and it will be demonstrated later that
chemical factors can ultimately dictate the particular particle
geometry that will prevail under reaction conditions. The
ability of particles to undergo a transformation from a non-
wetting to a wetting state and, in certain cases, to a spread
condition, suggests a significant degree of atomic mobility
exists within the particle at temperatures well below that of
the bulk melting points of the metals or metal oxides.

It has been known for many yea rs that when small meta 1
crystallites are heated on a graphite support they will exhibit
mobility at a particular temperature (4)(5). Our early studies
wi th the CAEM techni que showed that when iron parti cl es sup-
ported on graphite were heated in various gas environments, the
particles became mobile on the basal plane at 700°C. During
this period no interaction was observed between the metal and
the graphite, i.e., these particles were not catalytically
gas ifyi ng the graphite. Subsequent studi es with other metals
dispersed on graphite revealed that this was a general phenome-
non and that there was a characteristic mobility temperature
which was ~elated to the bulk melting point of the particular
metal. From the plot presented in Figure 3. it can be seen
that the experimentally observed temperature for the onset of
mobil i ty of 10 nm sized pa rt i c 1es of va ri ous meta 1sand oxi des
is approximately half their respective bulk melting points when
the temperature is expressed in absolute units. For the most
 237

part, these mobil ity temperatures coi nci de with the Tammann
temperature of these materi a 1s, whi ch is ca 1cul ated from the
bulk melting point [x 0.52] K.

This is a purely empirical relationship based on Tammann's

discovery in 1932 (6) that there was a minimum temperature at
which a solid would undergo a solid-solid interaction. Since
that time other workers reported that the rate of sintering of
oxides increased markedly at about half the value of the
absolute melting temperature. It was also found that under
these circumstances defects in the surface of a solid became
mobile enabling the surface migration of ions to take place.
It shoul d be appreci ated that the Tammann temperature does not
represent a discontinuity of behavior, but rather a temperature
in the vicinity of which a rapid change in the rate of mot·ion
of ions or atoms occurs. It is, therefore, not surprising to
find that in a system where there is a weak interaction between
a particle and support, that particle motion is observed at the
Tammann temperature of the particulate material.

caD
1500

Y
Pd
NIOFO
~u Co

oAu
°
Ag

Melting Point (K)

FIG. 3 Relationship between mobility temperature of

particles supported on graphite and their bulk melt-
ing points.

CAEM studies have shown that prior to catalytic attack,

particles located on the graphite edges undergo a transforma-
tion from a non-wetting to wetting state. For the graphite-
oxygen reaction this change in particle morphology can take
place at temperatures below that of the Tammann temperature of
the catalytic material, suggesting that such particles are
238

hotter than the surroundi ng graphi te surface. It is possi b1e

that extra heat is imparted to the particles from the C-0 2
o
reaction, which is highly exothermic, ~298 = -94 kcal/mole.

1500
g
j
Su 1000

i
'0

J 500

2000 2500 3000

Molting Point (K)

FIG. 4 Relationship between onset temperature of catalytic

oxidation of graphite and bulk melting points of
catalysts.

In an attempt to obtai n an esti mate of thi s extra heat the

following approach was adopted. The observed onset tempera-
tures for catalytic oxidation by various catalysts has been
plotted in an analogous manner to the data in Figure 3, and is
shown in Fi gure 4. If one accepts that the correct i dent it i es
of the catalytic species have been chosen, then it can be seen
that a linear relationship exists between these two parameters,
which is similar to that given in Figure 3. From an overlay of
the two plots, Figure 5 a temperature separation of 160 ± 20°C
is obtained which could be the extra heat generated by the
exothermic reaction at active particle surfaces. This is suf-
ficient to raise the temperature of such particles to the
Tammann poi nt and induce the requi red mobil Hy to enable parti-
cles to wet the graphite edge regions.

A logical extension to this argument is to question what

happens in the caseo of an endothermi c react i on such as graph-
He-steam, where ~H298 = + 31.4 kcal/mole. Although there is
only a limited amount of data to address this point, from
Figure 6, it would appear that using the same criteria as for
the previous system leads one to tentatively conclude that the
active catalyst particles are significantly cooler than the
 239

surro~ndi ng gr~phi te ina steam envi ronment. From the overl ay

of thlS plot wlth that derived for motion of inactive particles
it is possible to calculate a temperature decrease of about 50
± 10oe, Figure 7.

3000

FIG. 5 Determination of extra heat generated at particle

surface due to the exothermi c graphi te-oxygen reac-
t; on.

2000 2500 3000

Molting Point (K)

FIG. 6 Re 1ati onshi p between onset temperature of catalyzed

steam gasification of graphite and bulk melting
points of catalysts.
240

1500 .

g
j 1000 -
I
!
i
°5'-OO----:I-:':OOO.,------'I500"----20.LOO----,t-2..J.50-0---13000
Melling Polnl (K)

FIG. 7 Determination of amount of heat lost from catalyst

particle surface due to endothermic graphite-steam
reacti on.

3.2 Modes of Catalytic Attack

In the graphite-oxygen reaction, the catalyst may operate

by facilitating basal plane penetration in imperfect regions or
by i ncreas i ng the rate of remova 1 of ca rbon atoms at edges and
steps. Catalysts operati ng by the former mode are termed pit-
ting catalysts, and those operating by the latter mode are
referred to as edge-specific catalysts.

3.2.1 Catalytic Pitting Action

An example of pitting action by platinum particles on the

surface of graphite after exposure to 5 Torr oxygen at 500°C is
shown in the scanning micrograph, Figure 8. Pits are initiated
when particles come into contact with surface defects, such as
vacanci es, whi ch are most commonly encountered at the i nter-
section of twin bands. fJt this point, the active particle pro-
ceeds to attack the graphite crystal in a direction perpendicu-
lar to the basal plane, thus producing a pit. The pit is sub-
sequently expanded by edge recession due to uncatalyzed attack
and takes the form of a hexagonally shaped hole. With con-
tinued oxidation, the pit increases in depth, and the uppermost
graphite layers become progressively more circular in shape.
This type of reaction reveals one of the limitations of the use
of the CAEM technique - the inability to quantitatively assess
the rate of change of features that occur perpendicular to the
 241

plane of the specimen. It is well established that the rate of

attack by pitting is several orc;lers of magnitude slower than
that which occurs by catalytic removal of edge carbon atoms.
Moreover, since the concentration of surface defects in a given
sample is usually quite small, catalytic attack by the pitting
mode does not make a major contribution to the overall gasifi-
cation rate.

FIG. 8 Scanning micrograph of a pit .produced during the

platinum-catalyzed oxidation of graphite at 500°C.

There is an exception to this rule, which is found in the

sil ver/graphite-oxygen system. From the work of Harris et al
(7) it appears that silver is capable of extracting carbon
atoms from the perfect regions of the graphite basal plane and
thereby creating its own vacancies in the surface. ~ a conse-
quence, a pit is formed at the site of each silver particle and
this behavior accounts for the observed increase in the oxida-
tion rate of graphite in the presence of silver.
3.2.2 Catalytic Olanneling Pction
In the case of edge speci fi c attack the preci se mode by
which the catalyst functions is governed by the degree to which
it wets the graphite, i.e., the strength of the catalyst/graph-
ite interaction. With reference to Figure 2, for the inter-
mediate wetted state where the catalyst is present as a dis-
242

FIG. 9 Sequence showing the propagation of a channel across

the graphite surface by a pt-Ir particle (i ndi cated)
at l030°C.
243
244

crete cap-shaped particle, then the mode of attack would be

expected to proceed by a channeling action.

Figure 9 is a sequence of stills taken from the T.V. moni-

tor showi ng the propagati on of a channel at 1030°C in 5 Torr
oxygen; the electron-dense platinum-iridium catalyst particle
is clearly evident at the leading tip of the channel. Once
activated, a particle would remain so for as long as contact
was mai ntai ned wi th the step whi ch constituted the ti p of the
channe 1 • Loss of channel i ng act i vity occu rred if thi s contact
was lost. The most common situation in which this occurred was
if the particle crossed a region of channel which it had
a 1 ready formed and proceeded to attack the island of graphite
left behind, Figure 10.

~. (a) (b)

(c) (d)

FIG. 10 Schematic representation of the deactivation of cata-

lyst particles by loss of contact with edges.

During the development of a channel, the catalyst particle

appea rs to be qui te fl ui d and, provi ded that it exh i bi ts pre-
ferential wetting with the steps which constitute the channel
tip, it will remain unaltered in this location throughout the
propagation period. However, in some cases there is a tendency
for active particles to become depleted in size as material is
depos i ted on the wall s of the channels. fts a consequence, the
width of the channel decreases· and the channel walls start to
undergo expansion causing channels to acquire a fluted appear-
ance. This sequence of events is shown schematically in Figure
 245

11. When all the material from a given particle has been de-
posited on the walls, linear propagation of the channel
ceases. Fi gure 12 is a hi gh-magnifi cati on mi crograph of a
molybdenum particle in the act of cutting a channel at 635°C in
10.0 Torr oxygen. In thi s system the acti ve parti cles leave
material on the sides of the channel, a feature which is par-
ticularly noticeable in this micrograph.

lal fbi

FIG. 11 Effect of wetting and spreading action on the cata-

lytic channeling mode.

FIG. 12 Catalytic channels from molybdenum/graphite in oxygen

at 635°C.
246

Some catalyst particles tend to propagate channels that

are predominantly straight and bend through angles characteris-
tic of the underlying graphite, whereas other particles produce
channels which follow a random path. Both types of channels
share many common features:

(a) From scanning electron micrographs, such as the example

presented in Figure 13, one can conclude that the sides of the
channel are strai ght rather than curved. Furthermore, si nce
channels are never seen to get perceptably deeper on areas of
uniform thickness the channel can be considered as operating on
a base 1eve 1 and change in depth is then determi ned by the
surface relief of the graphite crystal.

FIG. 13 Scanni ng mi crograph of a channel created by an iron

particle in hydrogen at 1000°C.

In another series of experiments a shadowing procedure was

used to determine the dimensions of channels produced by nickel
in the graphite/hydrogen reaction (8). This operation was
accomplished by evaporating uranium atoms onto the reacted
specimen at an angle of 25° with respect to the plane of the
graphite surface. Regions of the specimen shielded from the
impinging beam of atoms by surface relief or depression fea-
tures are not coated, and, as a consequence, appear as enlarged
shadows when the specimen is examined in the transmission elec-
tron microscope. Using this approach it was possible to calcu-
late the depth of channels created by different sized catalyst
 247

particles, and the ratio of channel depth to width was found to

be about 0.25 to 0.3.
(b) Detailed quantitative kinetic analysis of channeling
sequences show that the 1i near rates of channel propagation in
different crystallographic directions across the graphite basa 1
plane are identical.
It is also found that the linear rates of propagation of
channels is proportional to the width of the catalyst particle
and the depth of the channel. For the oxi dat i on react i on at a
given temperature the rate of channel propagation decreases
with increasing particle size; provided that the channels are
of similar depth (9), Figure 14. These data can be expressed
in the form of:
(channel propagation rate) a 1/(particle sizell2 •

70

'.~.
60
'7
III

!
UJ
l-
e(
e-e_
'" 50

4~~5----~3bO~----~35'-----A4~0----~4n5~--~50
PARTICLE SIZE (nm)

FIG. 14 Variation of channeling rate as a function of par-

ticle size for the graphite-oxygen reaction, at con-
stant temperature.

In contrast, for the hydrogenation reaction at a given tempera-

ture, it is found that the larger the catalyst particle the
faster the rate of cata lyti c attack for channel s of simi 1ar
depth (10). This dependence, plotted in Figure 15, can be
expressed according to the relationship:
(channel propagation rate) a (particle size)2.
248

,
~

'"

~LO----~30L---~40~--~50~---6~O----~7~O----~80
PARTICLE SIZE (nm)

FIG. 15 Relationship between channel propagation rate and

catalyst particle size for the graphite-hydrogen
reaction at constant temperature.

This variation clearly points to a fundamental difference in

the reaction mechanisms of catalyzed oxidation and catalyzed
hydrogenation of graphite.

TI8LE 1

VARIATION IN CHANNEL PROPAGATION RATE OF CATALYST PARTICLES MOVING

FROM THIN AREAS OF GRAPHITE, "A", TO THICKER AREAS, "B"

Channelin~ Rate
Particle Width (nm s- ) Rate on "A" Thi ckness Ratio
(nm) Area II A" Area" BII Rate on "B" Area "B" / Area "A"

36 2389.0 352.5 6.8 7.1

49 150.8 71.9 2.1 2.0
56 556.5 53.5 10.4 10.8
139 255.1 97.4 2.6 2.8
 249

Tab 1e 1 shows the effect of channel depth on the channel

propagation rates for palladium particles in oxygen at dif-
ferent temperatures (11). These values have been measured from
a situation depicted schematically in Figure 16, where a par-
ticle has been active on a relatively thin area of graphite,
"A", and then encountered a step on the surface causi ng the
depth of the channel to increase as the particle moved through
a thicker region of graphite, "B". Inspection of Table 1 shows
there is good agreement between the ratio of channel propaga-
tion rates and the relative thicknesses of the graphite when
compared to the remaining thickness at the base of the
channels, indicating that channel propagation rate is inversely
proportional to the channel depth; furthermore, this relation-
ship is independent of temperature.

DIIWCtIon of Motion
•

FIG. 16 Schematic diagram showing the effect of surface re-

lief of graphite on channel depth.

A combi nat i on of these two dependenci es suggests that all

active particles, irrespective of their size, gasify the same
amount of carbon per unit time under given experimental condi-
tions.

By measuring the rates of channel propagation of similar-

sized particles cutting channels of uniform depth at several
temperatures, it is possible from an Arrhenius plot to obtain
the apparent activation energy for the catalyzed process. The
advantage of the CAEM technique is that it allows one to obtain
rate data for catalyzed and uncatalyzed gasification from adja-
cent regions of the same graphite crystal.
250

3.2.3 catalyzed Edge Recession

Finally, for the situation where catalyst particles ex-

hibit a strong interaction with the graphite edge atoms and as
a result undergo spreading to form a thin film, then the sub-
sequent catalytic mode of attack is by edge recession. The
sequence of events preceding catalytic edge recession are given
schematically in Figure 17, and an example of this form of
attack is presented in the sequence of stills, ngure 18. Edge
recession, like channel propagation, may exhibit a very ordered
motion if the catalyst material exhibits preferential wetting
of a particular graphite face. Spreading of the catalyst in
this manner results in the most efficient use of the additive
in that the contact area between catalyst and graphite edge
atoms is maximized.

Wetting

Graphite

Ca) (b)

Spruellnll EdgaR_ion

~_ ___ Jf-------
Ce) Cd)

FIG. 17 Sequence of events preceding catalytic edge reces-

si on.

FIG. 18 Sequence showing catalyzed edge recession by calcium

in steam at 715°C (receding edges are indicated).
 251

The following example is taken from a study of the barium

catalyzed steam gasification of graphite (12) to demonstrate
the relative efficiencies of catalytic channeling and catalyzed
edge recession. Jlt 825°C the catalyst exhibits a slow trans-
formation in its wetti ng properties of graphite and as a con-
sequence, both forms of catalytic attack can be observed on the
specimen. From the quantitative data it is possible to obtain
an est i mate of the re 1at i ve cont ri but ions of the two modes of
catalytic attack toward the overall steam gasification rate.
The number of moles of carbon gasified per second, dn/dt, is
gi ven by
dn 2 !I. D E; p
=
dt M

where f is the rate of channel propagation or edge recession

(nm.s-), D is the width of the channel or receding edge
(nm), E; is the depth of the cha~rel or 3edge (nm), p is the
density of graphite (2.25 x 10- g.nm-) and M the atomic
weight of carbon. At 825°, 30 nm diameter particles create
channels at a linear propagation rate of 3.30 nm.s-. If one
assumes that the depth of the channel is 10 nm, then the number
of mo19s of carbon gasified per second by this particle is 1.86
x 10- • Under these conditions, the particle would typically
move a distance of 750 nm before complete dissipation as a thin
film along the channel walls, as depicted in Figure 11. At
825°C, the amount of carbon gasified per second by a receding
edge of total leng\h, 1500 nm and 11rickness11O nm, moving at a
rate of 1.11 nm.s- is 3.09 x 10- mol.s-. It is therefore
apparent that edge recession is the more important mode of
catalytic attack in the barium/graphite-steam reaction.

4. CATALYZED GRAPHITE-OXYGEN REACTIONS

Over the years catalyzed oxidation of graphite has been an
area of considerable research activity and the reader is re-
ferred to some excellent reviews covering this subject (13-
17). Here, studies which, in the opinion of the author repre-
sent 1andmarks in our understandi ng of the mi croscopi c detai 1s
of the topographical effects induced by catalysts will be high-
lighted. This will be followed by a discussion of some of the
recent developments relating to the factors which control the
mode by which a particular catalyst operates in the graphite-
oxygen reaction.

Hennig (18) used transmission electron microscopy to

examine replicas of graphite specimens which had been treated
with various colloidal metal suspensions and then reacted in
 252

oxygen. He reported that on crystals which were relatively

free of defects, the meta 1 cat a lysts attacked ina di rect ion
parallel to the basal plane to produce channels, but that par-
ticles which were initially located on the basal plane were
inactive. He suggested that the motion of a particle along its
catalytic channel was due to attractive forces between the
particle and carbon atoms at the Ghannel tip. Since these
carbon atoms were constantly being oxidized, only the leading
surface of the catalyst particle could maintain contact with
edge atoms. He also claimed that metallic particles did not
induce catalysis by action at a distance, i.e., by altering the
Fermi level of conduction electrons in the graphite.

Thomas and Walker (19) were the first workers to directly

observe the influence of catalysts on the oxidation of graphite
using controlled atmosphere optical microscopy. The results of
these studies were later described in detail in Thomas's review
article (20), and led to the realization of the dangers in
attempting to establish the identity of a catalytic species
using thermodynamic data to a system where one is concerned
with the stability of phases at interfaces.

The first in-situ electron microscopy investigation of

catalyzed oxidation of graphite was performed by Presland and
Hedley (21), who allowed air to leak into the microscope in the
vicinity of heated specimen of graphite onto which they had
evaporated a thin film of platinum. During the course of their
experiments they observed both uncatalyzed and catalytic forms
of attack. They claimed that uncatalyzed oxidation occurred by
relatively slow attack at the crystal edges in an irregular
fashion. On the other hand, catalytic attack appeared to have
a strong dependence on the mobility of catalyst particles,
which was manifested by the creation of channels which were
initiated at edges. A similar experimental approach was used
by Fryer (22) to study the interaction of palladium/graphite
specimens with air. He also observed the catalytic channeling
behavior, and argued that the liquid characteristics of some of
the palladium particles indicated that they were at a consider-
ably higher temperature than the surrounding graphite.

Evidence that catalytic attack could also take place by

the edge recession mode was first presented by Harris and co-
workers (23). Using controlled atmosphere electron microscopy,
they found that at temperatures below 545°C the edge recessiog
rate of lead/graphite specimens exposed to oxygen was about 10
ti mes faster than the uncata lyzed rate. They suggested that
under these conditions the catalytic entity was present as an
extremely thin film occupying the edge and step regions of the
 253

graphite substrate.

The author has used the same techni que to exami ne the
catalytic influence of a variety of materials on the graphite-
oxygen reaction. Although these experiments have allowed us to
gain some unique insights into the qualitative and quantitative
features of a gi ven cata lyst. the factors whi ch control the
mode by whi ch the catalyst operates have for the most part
remained unknown. e.g •• it has not been possible to predict
whether cat a lyt i c attack wi 11 occur by channel i ng or edge re-
cession.
Recent ly. a rat i ona 1e has been developed to account for
the behavioral pattern of a given catalyst and also an argument
presented for the relative lack of activity of the ferro-mag-
netic metals on the graphite-oxygen reaction. If we examine
the behavior of the various elements which have been studied.
with respect to their position in the Periodic Table. Figure
19. then certain features become evident.

FIG. 19 catalytic behavior of various metals on the graphite-

oxygen reaction as a function of their position in
the Periodic Table.

Metals labelled by the shaded areas adsorb oxygen dis-

sociatively and form stable oxides (24). In-situ electron
microscopy observations of the behavior of these metal/graphite
systems shows that catalytic attack tends to take place by the
edge recession mode and any channeling action follows the form
shown in Fi gu re 11. In cont ra st. meta 1sin the open boxes
adsorb oxygen non-dissociatively and as a consequence remain in
the noble state (24). In these systems. the major mode of
catalytic action is by channeling with some pitting. but no
catalyzed edge recession.
 254

It can be seen from Fi gure 19 that three of the metals

whi ch have been studi ed, rutheni um, rhodi um and i ri di um, ex-
hibit both forms of attack; edge recession at temperatures
be low 1000 0 C and channe 1i ng at hi gher temperatures. It is
interesting to note that in these systems the oxides, Ru02'
Rh20~ and Ir02 are the respective stable solid phases below
1000 C and at higher temperatures they undergo dissociation
leaving the metal as the only stable solid phase (25)(26).

The final classification of metals are those labelled by

the cross-hatching, which do not exhibit sustained catalytic
activity for the graphite-oxygen reaction and do not wet graph-
ite to any significant degree under these conditions.

Let us first address the issue of why some catalysts

choose to attack the graphite by the edge recession mode while
others operate by the channel i ng mode. Strong i nterfaci a 1
bondi ng wi 11 occur between the group of metals whi ch readi 1y
form stable oxi des and the oxygenated carbon edge atoms. The
strength of this interaction will be sufficient to induce
sp readi ng of the catalyst materi ala long graphi te edges and
subsequent attack by edge recessi on. On the other hand. the
group of meta 1s wh i ch rema in in the zero va 1ent state do not
exhi bit such a strong i nteracti on with the oxygenated carbon
surface so that the catalyst remains in a particulate form, and
attack occurs by the channeling action. Consistent with these
arguments is the finding that catalysts which undergo a transi-
tion from the oxide to the metallic state at high temperatures
also exhibit a corresponding change in catalytic action from
edge recession to channeling.

The relative inactivity of iron, cobalt and nickel for the

graphite-oxygen reaction is at first sight difficult to under-
stand since all three metal s form stable oxi des. and based on
the reasonings presented above, would be expected to -form a
strong interaction with the carbon-oxygen surface. However,
when one examines alternate reaction pathways for these metals,
a possible explanation for the observed inactivity emerges.

Consider the possibility of converting the "active" metal

oxide to an "inactive" metal carbide during interaction with
graphite at 1000 K according to the following reaction:

Mx 0y + C + Mx C + Y CO 2

o 0 o
6GREACTION at 1000 K = Y (MiCO) + MiM C -
2 x
 255

The equilibrium constant, K. can be evaluated as follows:

°
$1000 -RT £n K
Thi s equi 1i bri urn constant is di rect ly related to the equi 1i _
brium pressure of carbon dioxide:

K =
(P CO )y
2
Hence, one can readily calculate the conditions where the reac-
tion is likely to occur. Values of the equilibrium partial
pressures of carbon dioxide have been calculated for a number
of systems where the relevant thermodynamic data is available
(27) and are listed in Table 2.

TABLE 2

EQUILIBRIUM VALUES OF TliE CO2 PMTIAL PRESSURES IN VMIOUS

METAL OXIDE/GRAPHITE - ~ SYSTEMS

Likelihood of carbide Formation

Oxide PCOzmm Under CAEM Conditions

T;Q2 5.66 x 10-8 No

Cr203 6.69 x 10- 6 No

Mo02 2.51 x 10-2 No

V6013 5.98 Yes

W03 1.58 x 104 Yes

Fe203 7.67 x 10 3 Yes

NiO 1.74 x 109 Yes

Inspection of this table shows that carbide formation is

probab 1e with not on ly i ron and ni cke 1, but a 1so wi th vanadi urn
and tungsten. Exami nat i on of the experi menta 1 data from these
1atter systems does indeed i ndi cate that both el ements tend to
lose their catalytic activity for the graphite-oxygen reaction
256

at hi gh temperatures (28) (29) • One can therefore tentatively

conclude that the most efficient catalysts for the carbon-
oxygen reaction will be those elements that form stable oxides,
but do not readily react with carbon to form carbides. Recent
studies by Yang and Wong (30) demonstrate that metal carbides
exhibit only marginal catalytic activity for graphite oxida-
tion. Furthermore, the carbides did not function by the well
established modes of edge recession or channeling, but appeared
to operate by a long-range action, catalyzing the removal of
carbon at edges of pits which were remote from the catalyst
particle locations. These results were interpreted according
to the notion that the metal carbides served as dissociation
centers for molecular oxygen and reaction at the edge regions
of the pits was due to oxygen spillover.
Detailed quantitative kinetic analysis of the reaction
sequences observed during CAEM experiments yields information
on the intrinsic activity of a particular catalytic species.
The characteristics of the catalytic channeling action have
been dealt with earlier in this paper. Examination of systems
where the catalyst operates by the edge recession mode shows
that at any gi ven temperature all edges whi ch are undergoi ng
gasification recede at the same rate, regardless of their
thickness or length. In general, the apparent activation ener-
gies derived from the behavior of individual catalyst particles
show excellent agreement wi th those obtained from bu 1k experi-
ments (31) (32) • One must exerci se cauti on in equati ng the
ki netic data obtai ned from CAEM studi es of a given catalyst
with its overall activity on a bulk scale. In addition to a
knowledge of the intrinsic catalytic activity such a determina-
tion requires an estimate of the fraction of the additive which
participates in the reaction and the degree of wetting of the
graphite by the catalyst.

Despite this limitation, the data are valuable for deter-

mining the temperature range where a given catalyst is active
and also for comparing the catalytic activities of materials
which operate by the same mode of attack. Figure 20 is a col-
lective plot of the catalytic influence of a number of oxides
on the graphite-oxygen reaction, together with data for the
uncata lyzed reacti on. These catalysts were found to operate
predomi nantly by the edge recession mode. It is clear that of
this group of oxides, Mo0 3 and V205 are the most active cata-
lysts; however, the temperature ranges over whi ch they exert
their influence are quite different. An analogous plot, Figure
21, has been constructed for the effect of various noble metals
on the graphite-oxygen reaction and in these cases catalysis
occurs by the channeling mode. From this dependence it appears
 257

that the order of catalytic activity follows the sequence: Pd

> Pt > Ir .. Rh .. Ru. It shou 1d be noted that at temperatu res
below lOOO°C, the latter three metals are present as oxides and
although they exhibit catalytic activity for the graphite-
oxygen reaction they fall far short of that di splayed by either
palladium or platinum, which remain in the metallic state
throughout the reaction.

IGO"C 700"C IGO"C

"0.7 0.1
1fT (K)x10'

FIG. 20 Arrhenius plots of the catalytic influence of various

metal oxides on the graphite-oxygen reaction.

1DOO"C 100"(; IOO"C

t.o

0"

1 .•

.• ---'.:'::.'--+".O:--~'.'~
0·'0L,-.7--:."""
11t(K)x10'

FIG. 21 Arrhenius plots of the influence of noble metals on

the graphite-oxygen reaction.
258

5. CATALYZED GRAPHITE - HYDROGEN REACTIONS

The first detailed study of catalyst particle behavior
during the hydrogenation of graphite was performed in an
optical microscope by Tomita and Tamai (33). These workers
made periodic examinations of graphite specimens containing
various transition metal additives, following reaction cycles
in hydrogen at 870°C. They determined that the catalytic
attack took place by channeling and found no evidence for pit
formation in the basal planes. Furthermore, it appeared that
the majority of channels were straight and preferentially
orientated parallel to the <1i20> crystallographic di rections
of the graphite.

In more recent years both controlled atmosphere (10,34,35)

and high resolution (36) electron microscopy techniques have
been used to investigate catalyzed graphite-hydrogen reac-
tions. In addition to confirming many of the optical micro-
scopy observations, these studi es have revea 1ed a number of
other features about the characteristics of the catalyst be-
havior.

Continuous observation of the reactions of a number of

metal/graphite systems in 1.0 Torr hydrogen shows that parti-
cles located on edge sites can under certain conditions, under-
go spreading; however, this action does not appear to be accom-
panied by edge recession. Gasification by single metal systems
is only observed when the catalyst is present as di screte par-
ticles and then attack occurs exclusively by the channeling
mode. Experiments with bimetallic catalysts have shown that
the addition of a second metal to the catalyst can result in
attack by the edge recession mode (37).

As pointed out earlier, the channels created in the

presence of hydrogen tend to be aligned parallel to the
<1120> di recti ons and even in the cases where they moved in
directions parallel to the <1010> orientations the facets of
the catalyst particles at their tips were parallel to the
<1120> directions. Although we do not have a clear under-
standing of this phenomenon, the results imply that there is a
preferential wetting of the zig-zag faces «1120> directions)
of the graph ite by the meta 1 and that the pa rt i c 1es move in
such a manner to maintain contact with these faces.

The channels showed many similar characteristics to those

formed under oxidation conditions, but it was evident that even
at temperatures in excess of 1000°C, they remained parallel-
sided throughout their propagation period, indicating that the
 259

rate of uncatalyzed attack, which would result in edge reces-

sion, was insignificant under the prevailing reaction condi-
tions. Weak beam dark field observations of platinum/graphite
samples which had been treated in hyrogen at temperatures up to
lOOO°C showed that the particles at the end of the channels
were rather flat in the form of a platelet of cubo-octahedron
with irregular edges (36).

Quantitative kinetic analysis of many channeling sequences

shows that at any given temperature large particles propagate
channels at a faster rate than small ones. The details of this
dependence are gi ven in Fi gure 15. There does appear to be a
critical upper temperature for many systems where the catalytic
activity starts to decline. Furthermore, if this temperature
is exceeded then on subsequent cooling and reheating in hydro-
gen the particles exhibit a new lower level of activity. This
effect, whi ch was fi rst reported by Rewi ck and co-workers from
thei r bu 1k studi es of va ri ous platinum doped ca rbons (38) can
in certain cases lead to complete loss of catalytic activity
(34)( 35) •

The differences in the deactivation behavior of various

metals at temperatures in excess of 950°C is probably related
to variations in the metal-carbon chemistry of these systems.
The formation of metal-carbon bonds with subsequent reaction of
carbidic carbon with atomic hydrogen to form methane are key
steps in the proposed mechanism of catalytic hydrogenation of
carbon (39). Surface carbides are readily formed when metals
are heated in the presence of a source of carbon. If the car-
bidic species are not removed fast enough to form gaseous pro-
ducts, then a decrease in catalytic activity will be observed
and under extreme circumstances where total blocking of the
particle surface occurs then the catalyst will completely lose
its activity. For the case where the catalyst particle becomes
partially blocked by carbon (graphitic) overlayers then a
uniform drop in the rate of channel i ng of a gi ven si zed par-
ticle during a subsequent cooling cycle would be expected,
i.e., a partially deactivated large particle would exhibit the
same activity as that displayed by a "fresh" smaller par-
ticle. Consequently, one would not expect to find large dif-
ferences in the values of the activation energies during the
heating and cooling cycles, which is consistent with the re-
ported data (38).

The plots presented in Figure 22 indicate the catalytic

influence, based on intrinsic channeling rates, of several
metals on the graphite-hydrogen reaction. To allow for the
effects of particle size and channel depth on rate, these
260

measurements were obtained from 2.5 nm diam. particles cutting

channels of similar depth. The data shown in Figure 22 are
based on the behavior of "fresh" catalyst particles and at
900°C the order of activity follows the sequence Re > Ru > OJ >
Ir > Pt > Ni > Fe. Comparison of this trend with that obtained
by Tomita and Tarnai (40) from bulk studies shows excellent
agreement: Rh ~ Ru > Ir > Pt > Ni > Fe.

FIG. 22 Arrhenius plots of the effect of various metals on

the graphite-hydrogen reaction.

Finally, although most of these catalysts have been found

to be susceptible to deactivation during high temperature reac-
tion in hydrogen, treatment of the deactivated catalyst in
oxygen at 635°C restores the initial high activity when the
systems are subsequently reacted in hydrogen.

6. CATAlYTIC GASIFICATION OF GRAP~ITE BY BIMETAlLIC PARTICLES

The purpose of this investigation was to establish if it
was possible to determine in a bimetallic system which com-
ponent preferentially segregated to the particle surface under
oxidizing and reducing conditions. The strategy was to compare
the catalytic behavior of an alloy with that of its pure con-
stituents on the graphite-oxygen and graphite-hydrogen reac-
tions. Controlled atmosphere electron microscopy has been used
to study the catalytic effect of tungsten and rhenium on the
 261

graphi te-oxygen and graphi te-hydrogen reacti ons. The i nvesti-

gation has also been extended to cover the influence of tungs-
ten-rhenium on the same reactions (29).

When the si ngle component/graphite systems were heated in

5.0 Torr oxygen, the observed patterns of behavior were very
similar, differing only in the temperatures where the various
events occurred, Table 3. In both cases the catalyst particles
showed a strong tendency to spread' along the graphite edge and
as a consequence recession of edges was the predominant form of
attack. In contrast, the bimetallic particles did not wet tile
graphite to the same degree and attack was limited to the
propagation of channels. This difference in behavior is
probably a result of strong bonding between tungsten and
rhenium, which is reflected in a reduction in the strength of
the interaction of each component with the graphite substrate.

TMlE 3

A COMPMISON OF 1l1E BEHAVIOR OF W, Re AND W-Re ON THE

GRAPHITE-02 AND GRAPHITE-H2 REACTIONS

catalyst Graphite-02 Graphite-H 2

W Pitting at 705·C
Channeling at 770·C
Spreading of active No reaction
material to produce
edge recession at 910·C
Ea (channeling) = 58.7
kcal.mole-

Re Channeling at 550·C Channeling at

Spreading of active 71S·C
material to produce Ea (channeling) -1
edge recession at 63S·C = 3S.9 kcal.mole
Ea (channeling) = 47.7
kcal.mole-

W-Re Channeling at 635·C No reaction

No spreading action
Ea (channeling) = 47.7
kcal.mole-
262

The similarity in the derived values of the apparent acti-

vat ion energi es for the mi xed system and the rheni um cat a lyzed
reaction indicates that under oxidizing conditions the bimetal-
lic particle surface is enriched with rhenium.
When rheni um/ graphite speci mens were heated in 1.0 Torr
hydrogen, catalytic attack of the graphite by the channeling
mode commenced at 715°C. The channels had all the characteris-
tics generally found under hydrogenation conditions. The chan-
neling action remained the exclusive form of attack up to
1050°C, the highest temperature studied. On the other hand, no
catalytic action was observed when either the tungsten/graphite
or tungsten-rhenium/graphite sampl es were exposed to hydrogen
at temperatures up to 1050°C. It is therefore concluded that
following reduction, the surfaces of the bimetallic particles
are preferentially enriched with tungsten.
The concept of using catalyzed gasification of graphite as
an in-situ microanalytical probe of bimetallic particles
appears to be a viable approach. It enables one to gain infor-
mation about the surfaces of such particles during reaction and
offers distinct advantages over many conventional surface
science tools, which suffer from a common limitation, namely
that speci mens have to be exami (led under hi gh-vacuum condi-
tions' nominally 10- 10 Torr. fJ.s a consequence, the observa-
tions are confined to the post-reaction state and may give
misleading inform3tion.

7. CATALYZED STEAM GASIFICATION OF GRAPHITE

Alkali, alkaline earths and certain transition metals,

particularly those of the iron sub-group are known to be active
metals for the steam gasification of graphite (14,17,41-44).
It is not the intention here to give an extensive review of the
literature pertaining to this area, but instead to focus on the
difference in behavior of metals to that of alkali and alkaline
earth elements on the graphite-steam graphite reacti on. The
specific examples of catalysts chosen for this purpose are
calcium and nickel.
One of the major breakthroughs in the understanding of why
calcium is such an efficient steam gasification catalyst is
attributable to Otto and co-workers (45). From a combination
of kinetic data and electron microscopy observations they con-
cluded that the increase in gasification rate was associated
with the ability of the alkaline earth to maintain extensive
and intimate contact with the graphite surface. It is in-
 263

teresting that in the first CAEM study of the ca1cium/graphite-

steam reaction, Coates et a1 (46) reported that they were un-
able to observe any catalytic action of calcium. They were,
however, basing their conclusion on the expectation that cata-
lytic attack would be manifested by channel formation. Recent
studies by the author (47) confirm the absence of channel form-
ation and in addition demonstrate that the high activity of
calcium is associated with the ability of the alkaline earth to
wet and spread along the graphite edges, which subsequently
gasify by the edge recession mode.

The CAEM studi es showed that in the presence of steam,

calcium coated graphite edges would occasionally deactivate.
This behavior, which was also observed in the potassium-cata-
lyzed steam gasification reaction (48), became more pronounced
as the temperature was raised to 900°C. Deactivation may arise
from chemi ca 1 changes in the catalyst or the graphite surface
or loss of bondi ng contact between the cata 1yst and the graph-
ite. It is believed that the catalyst-carbon interfacial bond-
ing which is initially responsible for the wetting interaction
of calcium at low temperatures (600°-700°C) involves surface
oxi de groups. It is poss i b1e that the graph i te edges cannot
support a suffi ci ent inventory of oxygen groups at hi gh tem-
perature and the interfacial bonding is broken thus allowing
the catalyst to sinter.

The steam gasification of graphite in the presence of

nickel followed the same pattern of behavior as that with cal-
cium, differing only in the temperatures where the various
events occurred (37). Catalytic attack by the edge recession
mode commenced at 715°C and continued in a systematic fashion
as the temperature was progressively raised to 1075°C. flt this
temperature, the recession rate tended to decline, and in some
cases, edges stopped moving, indicating that catalyst deactiva-
tion was occurring.

It has been proposed by Fi guei redo and Tri mm (49) that

nickel catalyzed steam gasification of carbon is controlled by
carbon diffusion through the metal particles. Support for this
postulate is seen from the excellent agreement between the
measured activation energy f~ the reaction performed in the
CAEM of 33.8 ± 4 kcal.mo1e- and t~at for carbon diffusion
through nickel of 33.0-34.8 kca1.mo1e- (50).

When graphite specimens containing a 1:1 mixture of nickel

and calcium were reacted in steam it was necessary to raise the
temperature to 795°C to induce catalytic attack·. Gasifi cati on
occurred by the edge recession mode, and as with the single
 264

components, the reaction proceeded in a very uniform manner,

the edges being aligned in directions parallel to the <1120>
crystallographic orientations of the graphite. It was signifi-
cant that with the mixed catalyst no evidence of deactivation
was apparent even at 1120°C.

The variation in the edge recession rate was measured as a

functi on of temperature and is presented in the form of an
Arrhenius plot, Figure 23. Also included on this plot for
comparative purposes are the data for calcium and nickel sys-
tems. From the slopes of these lines it is found that the
apparent activation energy of the mixed system is identical to
that of calcium, 51.3 ± 6 kcal.mole- •

:~'\
.\
1.5 \\

1.0 '\\ \_-Clilcium

\ \
~ 0.5
~\\ \\
:\\ \
-~ l I-----':~~\,\_>.;':----\-\-\--1
0

·. \ \

··..\
\\ \
. y....... \ \
\ \ \
\ \
\ \

\\
.1.0 L.-_~_ _:':--'--~:-'--'
0.7 0.1 0.' 1.0
1fT (KIa 10>

FIG. 23 Arrheni us plot of ni cke l-ca 1ci urn catalyzed .graphite-

steam reacti on. Al so i ncl uded for compan son pur-
poses are the plots for calcium and nickel alone for
the same reaction.

Although it was not possible to determine the surface

composition of the mixed catalyst in these experiments simple
thermodynami c cons i derat ions poi nt to the en ri chment of the
surface by calcium, which is probably present during the reac-
tion as oxide, whereas nickel will remain in the noble state
and tend to concentrate at the graphite interface.

Based on this rationale one can then understand the de-

crease in rate of gasification found when nickel is added to
 265

calcium by reference to the fundamental differences in the

mechanisms by which these two elements operate in the graphite-
steam reaction. Kapteijn and Moul ijn (51) have proposed the
following sequence of reaction steps to account for the in-
fluence of calcium on the graphite-steam reaction:

CaO + H2O -+ Cao .0 + H2 (I)

CaO.O + C -+ CaO + C [OJ ( I1)

C [OJ + CO ( I I I)

In thi s scheme the rate determi ni ng step is c 1aimed to be

the release of CO from the carbon structure.

A somewhat different reaction scheme was presented by

Holstein and Boudart (52) to explain the metal catalyzed
carbon-steam interaction.

C-C-C + M ! C-C + M-C ( IV)

M + H20 -+ M-O + H2 ( V)

M- C + M-O + 2M + CO (VI)

The fi rst step in the mechani sm is the formation of car-

bidic carbon on the metal, which depends upon the rupture of a
C-C bond in the bulk carbon source, followed by diffusion of
carbon species through or over the metal surface. The second
step i nvo 1ves the di ssoci at i ve adsorption of water, and the
final step the formation of CO by reaction of carbidic carbon
wi th adsorbed oxygen speci es. The rate determi ni ng step in the
nickel catalyzed steam gasification of graphite is believed to
be carbon diffusion through the metal, i.e., one of the factors
instep (IV).

When only the calcium salt is present the catalytic

species are in close contact with the graphite steps and edges
and there is direct access to the carbon source. However, as
pointed out above, in the mixed system calcium loses its inti-
macy with the graphite as nickel tends to concentrate at these
regi ons. fJ5 a consequence, the ca rbon supply to ca 1ci urn is
attenuated by the presence of a nickel diffusion barrier. It
is interesting to find that the apparent activation energy for
the nickel-calcium/graphite-steam reaction is almost the same
as that found for the i nfl uence of cal ci urn alone on the same
reaction, suggesting that the same rate controlling step is
266

operative in both systems. It would appear that in the mixed

catalyst nickel merely serves as an obstacle to the smooth
supply of carbon necessary for the efficient conversion of the
calcium peroxide species to calcium oxide and this aspect re-
sults in a lowering of the catalytic activity of the calcium
component.

REFERENCES
1. H. Hashimoto, T. Naiki, T. Eto and K. Fujiwara, Jap. J.
Appl. Phys. 7, 946 (1968).
2. R. T. K. Baker and P. S. Harris, J. Phys. E., ~, 793
(1970) •
3. G. R. Hennig, in "Chemistry and Physics of Carbon", Vol. 2
(ed. P. L. Walker, Jr.), Marcel Dekker, New York, 1966.
4. G. A. Bassett, Proc. European Regional Conf. Electron
Microscopy, Delft, 1960, I, 270 (1961).
5. J. M. Thomas and P. L. Walker, Jr., J. Chem. Phys. i!.., 587
(1964) •
6. G. Tammann, "Lehrbuck A. Metallkunde", 4th ed., Leipzig
(1932).
7. P. S. Harris, F. S. Feates and B. G. Reuben, Carbon 12,
189 (1974). -
8. R. T. K. Baker, R. D. Sherwood and E. G. Derouane, J.
Catal. 75, 382 (1982).
9. R. T. To Baker, R. D. Sherwood and J. A. Dumesic, J.
Catal. 62, 221 (1980).
10. R. T. To Baker, R. D. Sherwood and J. A. Dumesic, J.
Catal. 66, 56 (1980).
11. R. T. K.Baker, J. A. France, L. Rouse, and R. J. Waite,
J. Catal. 41,22 (1976).
12. R. T. K. Baker, C. R. F. Lund and J. J. Chludzinski, J.
Cata 1. 87, 255 (1984).
13. P. L. Walker, Jr., F. Rusinko, Jr., and L. G. Jlustin, in
"Advances in Catalysis", Vol. 11 (D. D. Eley, P. W.
Selwood. and P. B. Weisz, Eds.), Pcademic Press, New York,
1959, p. 133).
14. P. L. Walker, Jr., M. Shelef, and R. A. Mderson, in
"Chemistry and Physics of Carbon", Vol. 4 (P. L. Walker,
Jr., Ed.), Marcel Dekker, New York, 1968, p. 287.
15. J. B. Lewis, in "Modern Aspects of Graphite Technology"
(L. C. F. Blackman, Ed.), Pcademic Press, New York, 1970,
p. 129.
 267

16. H. Marsh, Special Publication of the Chemical Society, No.

32, 1977, p. 133.
17. D. W. McKee, in "Chemistry and Physics of Carbon", Vol. 16
(P. L. Walker, Jr., and P. A. Thrower, Eds.), Marcel
Dekker, New York, 1981.
18. G. R. Hennig, J. Inorg. Nucl. Chem. 24, 1129 (1962).
19. J. M. Thomas and P. L. Walker, J~ First Symposium on
Carbon, Tokyo, 1964.
20. J. M. Thomas, in "Chemistry and Physics of Carbon", Vol. 1
(P. L. Walker, Jr., Ed.), Marcel Dekker, New York, 1965.
21. A. E. B. Presland and J. A. Hedley, J. Nucl. Mat. 10, 99
(1963). -
22. J. R. Fryer, Nature 220, 1121 (1968).
23. P. S. Harris, F. S:--Feates and B. G. Reuben, Carbon l!.,
565 (1973).
24. N. D. S. Canning, D. Outka and R. J. Madix, Surf. Sci.
141, 240 (1984).
25. -;r:-C. Chaston, Plat. Metals Rev. 9,51 (1965).
26. G. V. Samsonov (Ed.), "The Oxide Handbook" (C. C. M.
Turton and T. J. Turton, transl.), Plenum, New York, 1973.
27. C. E. Wicks and F. E. Black, Bulletin 605, Bureau of
Mines.
28. R. T. K. Baker, R. B. Thomas and M. Wells, Carbon ll, 141
(1975) •
29. R. T. K. Baker, J. J. Chludzinski, Jr., N. C. Dispenziere
and L. L. Murrell, Carbon 21, 579 (1983).
30. R. T. Yang and C. Wong, J.c:atal. 85,154 (1984).
31. E. A. Heintz and W. E. Parker, Carbon 4, 473 (1966).
32. D. W. McKee, Carbon 8, 623 (1970). -
33. A. Tomita and Y. Tamai, J. Phys. Chem. 76, 2254 (1974).
34. C. W. Keep, S. Terry and M. Wells, J. Catal. ~, 451
(1980).
35. R. T. K. Baker and R. D. Sherwood, J. Catal. ~, 198
(1981).
36. J. Santiesteban, S. Fuentes and N. J. Yacaman, J. Vac.
Sci. Technol. AI, 1198 (1983).
37. R. T. K. Baker-;- J. J. Chludzinski and R. D. Sherwood,
Carbon in press.
38. R. J. Rewick, P. R. Wentrcek and H. Wise, Fuel.§., 274
(1974).
39. W. L. Holstein and M. Boudart, J. Catal. 72, 328 (1981).
40. A. Tomita and Y. Tamai, J. Catal. 27, 293\1972}.
41. F. J. Long and K. W. Sykes, Proc:Roy. Soc., London A215,
100 (1952). --
42. W. M. Tuddenham and G. R. Hill, Ind. Eng. Chem. 47,2129
(1955). -
43. K. Otto and M. Shelef, Proc. 6th Int. Congr. Catalysis,
London, 1977, Vol. 2, p. 1082, publ. The Chemical Soc.
268

44. F. Kapteijn and J. A. Moulijn, Fuel 62,221 (1983).

45. K. Otto, L. Bartosiewicz and M. Shelef, Carbon 17, 351
(1979). -
46. D. J. Coates, J. W. Evans and H. Heinemann, Jlppl. Catal.
7, 233 (1983).
47. R. T. K. Baker and J. J. Chludzinski, Jr., Carbon in
press.
48. C. A. Mims, J. J. Chludzinski, Jr., J. K. Pabst and R. T.
K. Baker, J. Catal. 88,97 (1984).
49. J. L. Figueiredo and D. L. Trimm, J. Catal. iQ., 154
(1975).
50. S. Di amond and C. Wert, Trans. AIME 239, 705 (1967). A.
B. Shavensin, A. H. Minkevitch, and--c. B. Scherbinski,
Izv. Vyssh. Uchebn. laved. Chern. Met. 1,95 (1965).
51. F. Kapteijn and J. A. Moulijn, private C:ommunication.
52. W. L. Holstein and M. Boudart, J. Catal.}2, 337 (1982).
 269

GASIFICATION OF CARBON DEPOSITED ON METALLIC CATALYSTS

J.L.Figueiredo*, C.A.Bernardo. and J.J.M.6rfao*

* Faculdade de Engenharia, Porto, Portugal

• Universidade do Minho, Braga, Portugal

1. INTRODUCTION

The catalytic gasification of carbon and coal is a subject of

great industrial importance, with applications in the production of
fuel gases, regeneration of coked catalysts and decoking of heat
transfer surfaces and reactor walls. On a more theoretical basis,it
is also relevant in the investigation of the mechanisms of important
catalytic processes, like the methanation, steam-reforming and Fis-
cher-Tropsch reactions. Thus, it is not surprising that a recent
computer search produced 1567 references, for the period 1982-1984
only.
However, most of those references referred mainly to the gasi-
fication of coal, graphite and carbons obtained under well control-
led conditions. Much less attention has been focussed on the gasifi-
cation of carbons formed on metallic surfaces, namely nickel and
iron, by decomposition of carbon-bearing gases. Only about one hun-
dred of the above mentioned references covered this topic. Neverthe-
less, the importance of this subject is undeniable, as this type of
carbon is responsible for great operational problems in the hydro-
carbon processing industry: catalyst deactivation and breakdown, in-
creased pressure drop in reactors, and fouling of heat exchangers.
Much of the earlier research in this area concentrated on the
regeneration of catalysts, such as the work of Dart et al.(l), Hager-
baumer et al.(2) and an important publication by Gilliland and Har-
riott (3). In recent years, however, most of the publications are
related to the methanation reaction, namely with the identification
of the reactive intermediate carbon species (4-12). Amongst these,
primus inter pares, is the work of McCarty and Wise on the hydroge-
nation of surface carbon formed on Ni/A1203 by CO decomposition at
270

low temperatures (4). This was later extended to hydrogen and stearn
gasification of deposits formed from CO and C2H4 at temperatures up
to 1273 K (13). The special interest of this work lies in the attempt
to relate the different types of carbon with the deactivation pat-
terns of the catalysts and the corresponding reactivities for gasi-
fication (regeneration), as will be shown later.
In this chapter, we will review the gasification of carbon de-
posits on nickel and iron catalysts. It is thus appropriate to start
by considering the various routes to carbon deposition on metals and
the different types of carbon present in such deposits.

1.1 Carbon deposition on metals

"Coke" deposits on metals may contain pyrolytic and catalytic

carbons:
a) Pyrolytic or gas phase carbons, include soot and tars (polycyclic
aromatic species) which are formed by free radical chain reactions
in the gas phase, and may deposit on any surface where they undergo
further rearrangement and dehydrogenation (36).
b) Catalytic carbons are formed by hydrocarbon adsorption and dehy-
drogenation, producing chemisorbed carbon atoms capable of:
- dissolving and diffusing through the metal to precipitate at some
active growth areas, such as grain Qoundaries, where they form
carbon whiskers or filaments; or
- nucleating and growing on the surface to form "encapsulating" car-
bon.
This mechanism was originally proposed for carbon formation on ni-
ckel (23); on iron, essentially the same mechanism is thought to ap-
ply, with the additional complication that stable iron carbides are
involved (30,37).
"Encapsulating" carbon leads to catalyst deactivation, while
carbon whiskers, carrying metal particles at their tips, may grow
without loss of activity. They are however responsible for catalyst
breakdown and increased pressure drop (23).

1.2 Identification of surface carbons

The nature of carbon deposirnformed on nickel supported cata-

lysts were characterized in terms of their reactivities towards gasi
fication by H2 and H20 using the technique of temperature-programmed
surface reaction (TPSR) by McCarty, Wise and coworkers (4,13).
It was found that carbon deposits formed by exposure to hydro-
carbons or carbon monoxide exhibit several states during TPSR, as
identified by maxima in the rate of CH 4 production (Figure 1). The
peak temperature, (Tp) was ta~en as char~cteristic of the reactive
state of carbon ~n the depos~t. Increas~ng the temperature of depo-
sition produced carbon states of lower reactivity (higher Tp).
 271

Table 1 - TPSR states for carbon deposited from C2H4

on Ni/A1203 catalyst (Adapted from ref.13)

State Identification T (K)* C2H4 exposure

p
temperature (K)

ex' Chemisorbed carbon 4l0±15 573 - 1073

ex Chemisorbed carbon 480±25 573 - 1073
Y Nickel carbide 550 573
6 Polymer carbon 660±30 573 - 773
0' Filamentous carbon 875±20 773 - 873
0 Encapsulating carbon 960±15 873 - 1273
E Platelet (?) carbon 1120±20 873 - 1273

* Peak temperature for heating rate 0.9 K/s. TPSR product

CH4. reactant H2.

0.08 r--r----,--,-----,,--,----,----,-----,--,-----.

•
0.06
n
'0
E
iI
,I
~ , I
w
t-
«
a:
0' ''',
"
,'I10
z
0
0.04
/
, ,, ," II
i=
()
:0
! \! \
, \I \
0
0
g: / Ii I

/ I, I/\\
v
I
()
0.Q2 I " \ \ €
I \ I ,......~'\',
I,'V ""
I
I
I
/ I
/ ~ "
' ,

I
/'
/
_/
1\' ....- ','.......
300 400 500 600 700 800 900 1000 1100 1200 1300
TEMPERATURE (K)

Fig. 1 - TPSR with hydrogen of carbon deposited on Ni/A1203 by

exposure to ethylene at: 573K ( - ) ; 773K (---); 1073K
( - - - ) ; and l273K ( - - ) . (Reprinted with permission
from ref.13 by courtesy of American Chemical Society).
Copyright 1982 American Chemical Society.
272

Seven types of surface carbon where identified, as summarized

in Table 1.
The a, a' states have been identified as chemisorbed carbon
atoms, and are sometimes referred to in the literature as "carbidic
carbon" and "surface carbide" (8,38). They are highly reactive to-
wards hydrogenation and they are thought to be the intermediates in
methanation and similar reactions (12,13). Therefore, it is not sur
prising that many recent publications deal exclusively with the gasi
fication of this type of carbon (e.g. 11,17). -
The S carbon state has been associated with the polymeric carbon
film found on commercial stearn-reforming catalysts, as referred by
Rostrup-Nielsen (39). It should be noted that the morphology of Cs
is somewhat uncertain (13,38). Under conditions of low hydrogen
concentration and high temperature, C can be transformed into the
less reactive Cs (4,13). a
The 8 and 8' states were identified as filamentary carbon,and
correspond to the whisker carbon reported by Lobo et al.(14) and
Baker et al.(15) and also to encapsulating carbon (16).
The present authors have been studying for a number of years
the gasification of carbon deposited on nickel and iron under con-
ditions where filamentary carbon is known to predominate (21-27).
Methane, propylene and n-hexane have been mostly used as the parent
hydrocarbons, and hydrogen, water and carbon dioxide as the gasify-
ing agents. On nickel, common characteristics were observed for the
gasification of the deposits, irrespective of gas or parent hydro-
carbon:
a) Reactivities much higher than those of carbons impregnated with
the same metal;
b) Long periods of constant gasification rate. Kinetic parameters
could thus be determined in these periods.
To explain the results, a mechanism was advanced, involving the
diffusion of carbon through nickel particles dispersed in the carbon
deposits (21). The same mechanism was later supported by other au-
thors (20,28). With iron, the same general features were observed,
but the interpretation of the results is more difficult than in the
case of nickel, as a number of solid phases can be present at the
temperatures considered (26,30). These results are described in mo-
re detail in the following sections.

2. EXPERIMENTAL DETAILS

The experimental arrangement consisted of a microbalance and

associated flow reactor, allowing the continuous recording of the
catalyst weight. Water vapour was produced in a bubbler system,where
a carrier nitrogen stream was saturated at known temperatures. The
system is described in more detail elsewhere (31). Polycrystalline
 273

metal foils of 0.10 rnm thickness were used (Ni: 99.7% purity,23 cm~g
Metals Research; Fe: 99.5% purity, 25 cm 2 /g, Goodfellow Metals). Ni-
ckel on alumina catalysts ( -18% Ni) with total and metal surface
areas of 8.5 and 1.1 m2 /g, respectively, were also used. Carbon was
deposited on these catalysts by decomposition of n-C6H14' C3H6 and
CH4·
Since gasification was found to be dependent on the characte-
ristics of the deposits, a cornmon deposition procedure was used in
all experiments. A mixture of hydrocarbon, hydrogen and nitrogen,in
fixed proportions, was passed over the pre-reduced catalysts at about
770-850 K. Under these conditions, the deposition proceeded essen-
tially at constant rate, i.e., without catalyst deactivation. (This
was easier to obtain on nickel). In this way, the deposits contain
mostly filamentary carbon, similar to those observed by Baker (15)
and McCarty (13). After deposition of a pre-determined amount of car
bon, the reactor was flushed with nitrogen and brought to the gasi--
fication temperature. The gasification mixture was then admitted
(H 2 /N 2 , H2 0/N2 or C02/N2) and the weight loss of the sample conti-
nuously recorded, while the composition of the effluent could be ana
lyzed by gas-chromatography.
In addition to isothermal runs, gasification thermograms could
be obtained by heating up the sample at constant heating rate under
flow of the reactant gas. This is similar to a TPSR experiment, and
the derivative thermogravimetric curve yields the same information
on the reactivity of the deposits.
The morphology of the deposits and metal surfaces (before and
after gasification) were examined by stereoscan electron microscopy
(JEOL JSM-35C) and associated X-ray microanalysis systems (JSC-35
and TRACOR TN 2000).

3. RESULTS

3.1 Morphology of the carbon deposits

The most striking feature of the carbon deposits obtained under

conditions such as described is their size. For example, 225 mg of
carbon were deposited on a supported nickel pellet of 4x6.5 mm and
60 mg weight (27). The deposit, shown in Figure 2, reached about 15
xlO rnm, its size being only limited by the sample holder and micro-
balance capacity.
Chemical analysis showed a considerable amount of metal in the
carbon which, in the case of foils was proportional to the total
weight of the deposit. A constant metal concentration of 1.6±O.4 wt%
was determined in the case of nickel for deposits of 30-120 mg. In
the case of supported catalysts, most of the metal was transferred
to the carbon deposit. Thus, the amount of metal in the deposit was
constant, not the concentration (21).
 274

Fig. 2 - Carbon deposit on a supported nickel catalyst

Fig. 3 - Morphology of carbon deposited on an iron foil. (Line=lO ~m).

Fig. 4 - Filamentary carbon on iron foil.(Line 1 ~m).

 275

Fig. 5 - Morphology of carbon deposited on Ni/A1203catalyst.(Line=10~m

Fig. 6 - Filamentary carbon on nickel foil. (Line= 10 ~m).

Fig. 7 - Iron concentration profile along a carbon filament.

276

Electron microscopy observations showed that the deposits con-

sisted of bundles of filaments (whisker carbon) as shown in the SEM
micrographs of Figures 3-6. The presence of metal particles on top
of the filaments is clearly visible, and was confirmed by the ener-
gy dispersive X-ray microanalysis, as shown in Figure 7. The depo-
sits were quite similar on both metals, with the exception of the
filament diameters, which were larger in the case of iron.

3.2 Gasification on nickel: kinetic results

Figure 8 represents typical gasification curves, where three

zones can be identified:
- An initial acceleration period, temperature dependent;
- A periou of constant gasification rate, lasting from about 15 to
75% of the total weight loss;
- A final period where the rate gradually diminishes.

Fig. 8 - Gasification of carbon on nickel foils by C02. Weight

v. time curves at temperatures shown. PC02 = 57.6 kPa

The long period of constant rate allows kinetic studies to be

made, by varying the temperature or the composition of the reacting
mixture. One such study is shown in Figure 9, representing the rate
of steam gasification of carbon deposits on Ni/A1203 (22). Similar
plots were obtained for the gasification with hydrogen and carbon
dioxide, both on foils and supported catalysts.
The orders with respect to the various reactant gases were de-
termined at constant temperature for the different catalysts. The
kinetic parameters obtained in the temperature range 675-1125 K are
summarized in Table 2 (21,22,24,25). The rate in the constant rate
period was found to be proportional to the initial amount of carbon
(m=l) up to a certain weight. Beyond that, it became independent of
the initial weight of the deposit. This "critical" weight of carbon
was therefore used to normalise the rates for each type of catalyst.
 277

TEMPERATURE (OC)
800 700 600 500
I I I

5.

.5

.1 r
IL
o .05

.01 r -

0.9 1.0 1.1 1. 2 1.3

3 -1
RECIPROCAL TEMPERATURE xl0 (K )

Fig. 9 - Arrhenius-type plot of the carbon gasification rates by

steam on a Ni/A1203 catalyst (PH 20= 26 kPa). (Reprinted
with permission from ref.22,by courtesy of Pergamon Press).

Table 2 - Kinetic data for the gasification of carbon deposited on

nickel catalysts. Average of independent sets of results.

Gas CO 2 H2 H2 O

T (K) <875 >875 <825 825-950 >950 <800 800-900 >1000

Ni foil 169±19 85±23 134±4 185±8
E
Ni/A1 2 03 140±8 66± 8 265±17 153±35 -82±5 164±9 68±7 .. 0
Ni foi 1 ,,0 0.5-1 2 ..0
n
Ni/A1 2 03 ,,0 0.5-1 0-0.5 1.4 1.8 0.5 0.5-1

E - Activation energy (kJ/mol) n - reaction order

278

Inspection of the results in Table 2 shows that, in general

terms, there are two kinetic regimes for gasification, corresponding
to:
- low temperatures and high activation energies;
- high temperatures and low activation energies.
In the case of hydrogen, a third zone is detected above 950 K,
in which the rate decreases as the temperature increases. The same
effect was observed by Nishiyama and Tarnai (28) and can be explained
in terms of the approach to the equilibrium for the reaction
C + 2 H2 = CH4
In the case of steam, the data in Table 2 indicate that, above
1000 K, the gasification of carbon on supported catalysts is almost
certainly limited by film diffusion. Further evidence for this is
the dependence of the gasification rate on the total flow rate (at
constant temperature and partial pressure) detected at high tempera-
tures.
Disregarding the effects observed at high temperatures (above
950 K), the kinetic data can be divided in two main groups:
i) Gasification by H20 and C02- The transition from the low to the
high temperature zone is accompanied by about 50% reduction in the
activation energy. At the same time, the reaction orders increase
from very low values to 0.5-1. These observations are consistent
with the onset of diffusion limitations in the pores of the carbon
deposits at higher temperatures.
ii) Gasification by H2- The decrease in the activation energy from
the first to the second temperature zone is now smaller, and the
order w.r.t. hydrogen increases to values higher than one. In this
case, diffusion limitations cannot explain the kinetic parameters
determined above 825 K. This is in agreement with the higher diffu-
sivity and lower reactivity of hydrogen.
In spite of the differences pointed out, gasification with the
three gases presents two major similarities:
a) High reactivities - The observed gasification rates are far su-
perior to those reported for other systems, namely carbons impre-
gnated with equivalent concentrations of metal. For instance, Marsh
et al.(32) measured a rate of 10- 5 s-l for the C02 gasification of
ex-PFA carbon with 2 wt% nickel at 883 K. At the same temperature,
the gasification of catalytic carbon obtained by decomposition of
C3H6 was 28 times faster (25). The same high reactivity of deposits
such as these has been reported, as early as 1954, by Gilliland and
Harriott (3) and later by Nishiyama et al.(28) and by Audier et al.
(20).
b) The same characteristic gasification curves - The gasification
curves for the three gases are similar, namely showing a lengthy
period of constant rate.
 279

3.3 Gasification on iron: kinetic results

The gasification with C02 of carbon deposited on iron foils

follows the same pattern as with nickel; however, a fourth zone can
be identified in the weight vs. time curves of Figure 10, namely a
final period of weight uptake, corresponding to the oxidation of the
metal. In fact, FeO crystals similar to those observed by Baker et
al.(40) were clearly visible under the microscope, as shown in Fig.
11, and their nature was confirmed by X-ray diffraction.
Kinetic parameters were determined in the period of constant
gasification rate, and are summarized in Table 3 (26). These data
are also consistent with the onset of diffusion limitations in the
pores of the deposit at temperatures higher than 997 K,as the acti-
vation energy is halved, and the reaction order changes from n=O to
(n+l)/2=0.s.

100~--------------~

--
•~
934K

943K
-550
~
1046K
5 10
Time (min)
Fig.lO - Gasification of carbon on iron foils by C02. (P C02 =sO kPa).

Table 3 - Kinetic data for the gasification with C02 of carbon

deposited on iron foils.

T (K) 943-997 >997

E (kJ/mol) 204 105

n 0* O.S§

* For PC02 >sO kPa, the rate decreases as PC02 increases, as a

result of the simultaneous gasification of carbon and oxida-
tion of iron.
§ In this temperature range, the rate of iron oxidation is not
significative,as compared to the rate of carbon gasification.
 280

Fig.ll - FeO crystals on iron foil oxidised by C02 .(Line 10 ~m).

Fig.12 - Metallic structure on Ni foil after carbon gasification

with hydrogen. (Line = 1 ~m).

Fig.13 - Metallic structure on Fe foil after carbon gasification

with hydrogen. (Line = 10 ~m).
 281

3.4 Surface structure

Upon gasification of the deposits, the surfaces of iron and

nickel foils were examined under the electron microscope. A granu-
lar metallic structure was observed on top of the original surface,
as shown in Figures 12 and 13.
It has also been observed that the metal surfaces become dis-
rupted after a deposition-gasification cycle, i.e., the surface
roughness increases. After one such cycle, carbon deposition on the
same foil is about 80% faster.

3.5 Temperature programmed gasification

Thermograms under reactive atmospheres were obtained with 5 mg

carbon deposits at the heating rate of 0.1 K/s. Typical results are
shown in Figures 14 and 15, for iron and nickel. Carbon dioxide is
more reactive than hydrogen, but oxidizes the iron surface well be-
fore complete gasification. With nickel, oxidation does not affect
significantly the gasification process. Mixtures of H2 + C02 show
essentially the same reactivity as C02 alone, but the oxidation of
iron is nearly suppressed. This is explained by the fact that the
rate of FeO reduction by hydrogen is higher than the rate of Fe oxi-
dation by C02' Separate experiments at 1043 K confirmed this.
Two sets of thermograms were obtained under different conditions:
i) with intermediate cooling to room temperature before starting the
temperature programme (Figures 14,15); ii) without intermediate
cooling, where the programme started at deposition temperature. The
thermograms with H2 exhibit two inflection points (peaks in the deri-
vative curve) showing the presence of two types of carbon. With C02,
this second peak cannot be observed due to oxidation of the metal.
The corresponding data are shown in Table 4.

Tab~e 4 - Thermogravimetric data for carbon gasification

Gas mixture

50% C02 + 50% N2 50% H2 + 50% N2

Metal
T (K) T (K) r (K- I ) T (K) r (K- I )
p PI PI P2 P2

Ni 875 1.76xlO- 2 906 7.09xlO- 3 1206 7.57xlO- 4

Fe 1031 1.96xlO- 2 1045 2.42xlO- 3 >1263 >2.27xlO- 3

r is the gasification rate at T , expressed as dX /dT, where X
p P c c
is the carbon conversion: X weight loss/initial weight.
c
 282

/
/
/
1ooI--......,.:":.-===::::-----===: /
" /
,
Fel
""\\ ,,
",/
\\
\ Ni
\Ni

\ \

\ \
\
\\
\

\ -------------------------------------
-------------------------
0~--~8±OO~---------ru900~~~~~~10~0~0----------~II00~--------~1f,200~----~
TEMPERATURE (K)

Fig.14 - TG curves of carbon gasification on iron and nickel foils

PH2 50 kPa ( - ) ; P C02 50 kPa (---).

·3
5xl0

Fe

0~--~80;0-----=====;90~0~-------~10~00~--------~110~0~--------~12~OO~-----J
TEMPERATURE (K)

Fig.15 - DTG curves for carbon gasification with PH2 50 kPa.

These results may be compared with those of McCarty et al.(13)

shown in Table 1. It is quite clear that the 1st peak of Figure 15
corresponds to the 8' state of McCarty, or filamentary carbon. The
2nd peak corresponds to a carbon more resistant to gasification; it
is possible that this is the same as the € state of McCarty.
This last form of carbon produces some deactivation. Thus, ga-
sification of carbon deposits on nickel foils was found to be 32%
slower if there was intermediate cooling, and it is reasonable to
assume that this type of deposit results from the precipitation of
 283

dissolved carbon (13,33). Indeed, graphite is known to form by the

dissolution-precipitation of carbon in cobalt and nickel (41). In
addition, carbon deposits removed from the parent foils could not be
gasified at 903 K unless they were previously heated up to 1223 K,
possibly to re-dissolve this type of surface carbon.
De Deken et al.(19) have also observed the slow burning of part
of the carbon in deactivated nickel catalysts, which was attributed
to dissolved carbon or nickel carbide.
The reactivity of carbonaceous species deposited from CO and
synthesis gas on a Fe/ZnO catalyst was studied by Kieffer et al.(42)
by isothermal and temperature programmed reaction with hydrogen. They
observed surface states similar to the a, a', and S states found on
nickel. Deposition temperatures lower than 550 K were used in this
study, and no filamentary carbon was reported.

4. GASIFICATION MECHANISMS

As a result of the mechanism involved in the deposition of car-

bon on iron and nickel, the deposits contain metal particles detached
from the original surface (23). Therefore, upon gasification, the
carbon deposits will behave as a supported catalyst, in which the
carbon is simultaneously the support and a reactant, the dispersed
metal particles being the active phase. These metal particles may
sinter in the course of gasification, originating the metallic struc-
tures seen in Figures 12 and 13.
Two general types of gasification mechanism may then be envisa-
ged, depending on the species interacting primarily with the catalyst:
a) Interaction between the reactant gas and the catalyst - The reac-
tant gas adsorbs on the metal surface and dissociates into atoms,
which migrate to the carbon-metal interface where reaction takes
place. Thus, the metal acts mainly as a dissociation center where the
gas molecules produce active species.
b) Interaction between the carbon and the catalyst - This involves
dissolution of carbon at the carbon-metal interface, followed by
diffusion through the particles to the metal-gas interface. This
process will supply carbon atoms (C ) to the surface, where they react
with the adsorbed gas species. a
This last mechanism, which is the reverse of the deposition me-
chanism, was proposed for the first time in 1975 (21) and explains
the two general kinetic features referred to above (Cf. § 3.2):
- The high reactivity of the deposits can be due to the enhanced in-
teraction of metal and carbon. In fact, the carbon contains highly
dispersed metal particles which are themselves saturated with carbon
(31,33). The recent work of Froment et al.(19) demonstrates the pre-
sence of carbon in the metal and supports this view.
 284

- The lengthy periods of gasification at constant rate may be explai-

ned by the continuous supply of carbon to the surface, where it pre-
cipitates and is gasified (19,21,23,28,31).
It is fairly well established that C can be transformed into
the less reactive forms of carbon, C~ andaC o (4,13,38). Recently,
Dost and co-workers demonstrated that the reverse process is also
feasible under gasification conditions (12). Using an XPS technique,
these authors have shown that the conversion of C~ to Ca occurs at
C~ hydrogenation temperatures, and postulated that the rate determi-
ning step in the gasification of C~ on nickel could be the detach-
ment of carbon atoms from C~ islands.
Thus, it is also reasonable to admit that carbon atoms (C a ),
supplied via the bulk, can be involved in the gasification. Any of
the steps in the above reaction scheme may be rate determining, de-
pending on the operating conditions and the gasifying agent. If
diffusion through the metal is rate controlling, then the observed
reaction orders and activation energies should agree with those ex-
pected for diffusion (n = 0; Eobs= Ediff - ~Hdiss)' This could be the
case for gasification by H20 and C02 on nickel at low temperatures
(24), but not for H2 on nickel (cf. Table 2).
Perhaps more important in order to test the feasibility of this
mechanism, is to check whether the rate of carbon diffusion through
the metal particles is comparable with the observed gasification ra-
tes. In the case of nickel, this has been shown to be the case, by
calculation of the rate of diffusion (21,24).
In the case of iron, if carbon diffusion were the controlling
step, an activation energy of about 180 kJ/mol should be observed,
corresponding to the dissolution (100 kJ/mol) and diffusion (80 kJ/
mol) of carbon through a-Fe. The situation, however, is more compli-
cated than in the case of nickel, as various solid phases may be in-
volved, including carbides and oxides (34). Thus, the process might
include the diffusion of carbon through Fe3C, with an activation
energy of 175 kJ/mol (35).
Recent publications, namely the work of McCarty et al. (13),
allow the discussion of this mechanism in a more realistic way. Not
only the different types of carbon formed on a metallic surface under
reaction conditions have now been identified, but also their reacti-
vity towards some gases is known. It is now possible to understand
many deposition and gasification results, interpreted so far mainly
in terms of kinetic arguments. For instance, it is now clear that,
at the temperatures of interest to this work, mainly 0'- and o-car-
bons are involved in the gasification. The growth of surface encap-
sulating carbon at temperatures above 875 K was proposed to explain
the decrease in the deposition rates observed during the decomposi-
tion of hydrocarbons (23,31). This can now be ascribed to the more
graphitic Co , that can ~ither be obtained by direct deposition, or
by slow transformation of the of-state (13). Gasification of this
 285

type of carbon will not be significant below 900 K. McCarty et al.

(13) speculated that Co and Co' correspond, respectively, to the
hard outer shell and to the inner, less dense and more reactive core
of the filaments observed by Baker et al.(15). This speculation is
in line with the observations of this author during the oxidation of
filaments carrying a nickel particle at the top. At lower temperatu-
res, the core was oxidised first, leaving the outer tube behind. In
some cases, the particles were observed to fall down the tubes. If
the temperature was increased, the outer tube would also gasify (15).
This may explain the incomplete gasification observed in Figure 8 at
the lower temperatures.
These observations can be used to put forward a realistic model
to assess the possibility of bulk-diffusion control during gasifica-
tion below 875 K.
The situation, as shown in Figure 16, can be understood consi-
dering that carbon dissolves in the lower part of the particle in
contact with the filament, and diffuses to the upper part, thereby
causing the metal to fall down the tube.
On the basis of this assumption, it was possible to derive an
equation for the lineap (and not massic) rate of gasification (25).
Unfortunately, linear rates of gasification have never been reported,
and the model could not be tested.
At higher temperatures, geometrical calculations loose meaning
as the metallic particles soften and change shape (33).
It is also interesting to mention that the diffusion of carbon
through nickel has been considered to explain the features observed
in the gasification of graphite by channelling particles (33).

Q) c)

Fig. 16 - Sequences in the gasification of a carbon filament.

ACKNOWLEDGEMENTS

The assistance of Mr. Carlos M.Sa with the SEM work is grateful-
ly acknowledged. Parts of the present work were carried out under
Research Grant 52/84 of Porto University and NATO R.G. 406/84.
286

REFERENCES

1. Dart, J.C., R.T. Savage and C.G. Kirkbride, Chern. Eng. Progr.,
102 (1949)
2. Hagerbaumer, W.A. and R. Lee, Trans. Am. Soc. Mech. Eng., 69,
780 (1947) --
3. Gilliland, E.R. and P. Harriott, Ind. Eng. Chern., 46, 2195 (1954)
4. McCarty, J.G. and H. Wise, J. Cata1., 57, 406 (1979)
5. Wentrcek, P.R., B.J. Wood and H. Wise,-Y. Cata1., 43, 363 (1976)
6. Araki, M. and V. Ponec, J. Cata1., 44, 439 (1976) --
7. Ho, S.V. and P. Harriott, J. Cata1.-,-64, 272 (1980)
8. Goodman, D.W., R.D. Kelley, T. Madey and J.M. White, J. Cata1.,
64, 479 (1980)
9. Bell, A.T., Cata1.Rev.-Sci.Eng., 23, 203 (1981)
10. Gardner, D.C. and C.H. BartholomeW; Ind. Eng. Chern. Fund., 20,
229 (1981)
11. Ozdogan, S.Z., P.D. Gochis and J.L. Falconer, J. Cata1., 83, 257
(1983)
12. Dost, A.A., V.R.Dhanak and S. Buckingham, J. Cata1., 89, 159
(1984)
13. McCarty, J.G., P.Y. Hou, D. Sheridan and H. Wise, Am. Chern. Soc.
- Symp. Ser., 202, 253 (1982)
14. Lobo, L.S., D.~Trimm and J.L. Figueiredo, Proe. 5th Int. Congr.
Catalysis, 1972, 2, 1125 (1973)
15. Baker, R.T.K., M.A. Barber, P.S. Harris, F.S. Feates and R.J.
Waite, J. Cata1., 26, 51 (1972)
16. Audier, M., A. Oberlin, M. Oberlin, M. Coulon and L. Bonnetain,
Carbon, 19, 217 (1981)
17. Bartholomew, C.H. and C.K. Vance, J. Cata1., 91, 78 (1985)
18. Kuijpers, E.G.M. and J.W. Geus, Proc. Int. Symp. Fundamentals of
Catalytic Coal and Carbon Gasification, Amsterdam, 1982, 92 (1982)
19. De Deken, J., P.G. Menon, G.F. Froment and G. Haemers, J. Cata1.,
70, 229 (1981)
20. Audier, M., M. Coulon and L. Bonnetain, Carbon~, 391 (1979)
21. Figueiredo, J.L. and D.L. Trimm, J. Cata1., 40, 154 (1975)
22. Bernardo, C.A. and D.L. Trimm, Carbon 17, 11S-(1979)
23. Figueiredo, J.L., in "Progress in Catalyst Deactivation", NATO
ASI Series, Ser. E54, 45 (1982)
24. Figueiredo, J.L., Carbon, 19, 146 (1981)
25. Bernardo, C.A. and J.L. Figueiredo, Proe. 8th Iberoameriean Symp.
Catalysis, Hue1va, 1982, 2, 596 (1982)
26. Figueiredo, J.L. and J.J.M. Orfao, Proe. 9th Iberoamerican Symp.
Catalysis, Lisbon, 1984, 1, 711 (1984)
27. Bernardo, C.A., I. A1strup and J. Rostrup-Nie1sen, unpublished
results
28. Nishiyama, Y. and Y. Tarnai, Carbon, 14, 13 (1976)
29. Baker, R.T.K., P.S. Harris, R.B. Thomas and R.J. Waite, J. Cata1.,
30, 861 (1973)
30. Cooper, B.J. and D.L. Trimm, J. Cata1., 62, 35 (1980)
31. Bernardo, C.A., Ph.D. Thesis, University-of London (1977)
 287

32. Marsh, H. and R.R. Adair, Carbon, 13, 327 (1975)

33. Keep, C.W., S. Terry and M. We11s,-Y. Cata1., 66, 451 (1980)
34. Sacco, A., P. Thacker, T.N. Chang and A.T.S. Chiang, J. Cata1.,
85,224 (1984)
35. Hi11ert, M., Metal. Trans. A, 15A, 245 (1984)
36. Trimm, D.L., in "Progress in Catalyst Deactivation", NATO ASI
Series, Ser. E54, 31 (1982)
37. Trimm, D.L., in "Progress in Catalyst Deactivation", NATO ASI
Series, Ser. E54, 65 (1982)
38. Bartholomew, C.H., Cata1. Rev.- Sci. Eng., 24, 67 (1982)
39. Rostrup-Nie1sen, J.R., in "Progress in Catalyst Deactivation",
NATO ASI Series, Ser. E54, 127 (1982)
40. Baker, R.T.K., J.R. Alonzo, J.A. Dumesic and D.J.C. Yates, J.
Cata1., 77, 74 (1982)
41. Derbyshire, F.J., A.E.B. Pres1and and D.L. Trimm, Carbon, ll,
III (1975)
42. Kieffer, E. and H.S. van der Baan, App1. Cata1., l, 245 (1982)
PART IV. CHEMICAL ENGINEERING PRINCIPLES
 291

KINETICS OF CATALYSED AND UNCATALYSED COAL GASIFICATION

F.Kapteijn, J.A.Moulijn

Institute for Chemical Technology, University of Amsterdam

Nieuwe Achtergracht 166, 1018 WV Amsterdam
The Netherlands

I INTRODUCTION

The world coal reserves constitute an important source of energy

and base chemicals. After the 1973 oil crisis and warning reports
of the Club of Rome for the limited world supplies of oil and natu-
ral gas the importance of coal for political, economical and energy
preservation reasons was acknowledged again. Increasing efforts in
coal research could be observed from that time.

Transformation of the solid material coal into gaseous consti-

tuents is one of the major topics. Gas is of easier transport than
coal and can, in principle, substitute natural or liquid petroleum
gases. Also, a central conversion of coal into gaseous products,
gasification, has many environmental advantages over a highly dis-
persed use of coal.

The subject of coal gasification covers the conversion of coal

to light gases, condensable liquids and tars and solid products in
the presence of reactive gaseous atmospheres, e.g. H20, CO2 , H2 and
°2 , and with or without help of a catalyst. DependenE on tfie reac-
t10n systems used the coal is converted into relatively low-energy
gases suitable for power generation and other industrial uses as
well as into high energy gases that can be used as a substitute for
natural gas (SNG).

To design and control gasification plants, knowledge about the

chemical and physical behaviour of the gaseous and solid reactants
is of fundamental importance. Chemical kinetics is one of the first
subjects to be investigated. Essentially this results in the develop-
ment of rate equations [1] expressing the reaction rates as a
292

function of the parameters that determine these rates, such as tem-

perature, partial pressures of reactants and products, type of coal
used, catalyst concentration and rate- and equilibrium constants:

r f ( T, p .••• , c, catalyst, k .••• , K ••• ) (1)

~ ~ eqi

For practical purposes, often power-law rate expressions are used,

but, although applicable over the range of experimental con~tions
the expression was derived for, extrapolation outside this range is
often impossible. Large deviations may occur. Therefore, rate ex-
pressions, based on a series of elementary steps pertaining to the
overall gasification process, are being preferred. Moreover, dis-
crimination between different kinetic models, can yield additional
information on the gasification process. This is nicely expressed
by Michel Boudart [1] "The primary goal of kinetics is to describe
the chemical transformation. A good description possesses permanent
value, so the results of qood chemical kinetics remain unchanged,
whereas the underlying reaction mechanism may be revised numerous
times in future. Therefore'kinetics is not only a tool of pure and
applied research, but also a very satisfying avocation".

The kinetics of coal and carbon gasification has some special

aspects that deserve mentioning:
- Gasification is a heterogeneous reaction between a gas and a solid
phase, the latter being consumed continuously.
- The reactivity of a coal depends on the number of sites available
for chemical attack ("active sites"). During gasification carbon
is being removed, thereby increasing or decreasing both the poro-
sity and the specific surface area, exposed to the gaseous phase.
The number of active sites may change continuously, thus hampering
a kinetic description. This is in contrast with heterogeneous
catalysis in which the solid is not consumed and the number of
active sites, in principle, remains constant. [1,2]
- Coal is a heterogeneous material consisting of maceral and mineral
constituents, its chemical composition depending on history and
coal rank. This implies that, although the kinetic description of
the gasification process can apply to more than one coal, the
kinetic parameters (rate constants etc.) will have to be deter-
mined for every coal individually.
Mineral matter can have a catalytic influence on gasification
rates [3] and consequently results of the kinetic experiments
should be carefully examined in this respect.
- Addition of catalysts in gasification processes adds an extra
dimension to the kinetic description. [4] Catalysts may increase
the number of active sites, lower activation energy barriers or
even completely change the gasification mechanism. Complications
 293

that may arise here are the contact between coal and catalyst,
catalyst dispersion and sintering, and deactivation due to
poisoning, changes of the gas phase or reaction with the mineral
matter in the coal.

The fundamental aspects of the kinetics of coal gasification

that will be dealt with in this chapter ,include thermodynamics, heat
and mass transfer phenomena, heterogeneous kinetics, catalysis,
experimental methods and data analysis. This covers a broad area
and the selection of the different topics has been subjective,
depending on personal interests, but also with the idea in mind to
present a useful survey for those who will be or are dealing with
kinetic aspects of coal gasification.

The interested reader can certainly be provided with more infor-

mation by refs. 1-7.

Considered will be the reactions of carbon/coal with CO 2 , H2 0

and H2 . Although the reaction with 02' generally referred to as
combustion, falls also in the class of gasification, this is not
included in this chapter.

II GLOBAL REACTION RATE

In gas-solid reactions the chemical process takes place at the

surface of the solid. Since under gasification conditions most of
the surface area is present within the carbon particles, the reac-
tants have to move into the porous particle.

The gasification process can be described by seven consecutive

steps as envisaged in fig. 1.
These steps are:

Fig. 1 Steps involved in gasification reactions.

294

1. Transport of reactants A, B, ... from the main stream to the coal

particle surface.
2. Transport of reactants within the pores.
3. Adsorption of reactants on an active site.
4. Surface chemical reaction between adsorbed molecules or atoms
with each other or with the carbon.
5. Desorption of products, R, S, ...
6. Transport of the products within the pores.
7. Transport of products from the particle surface back to the main
gas stream.

Steps 1, 3, 4, 5 and 7 are strictly consecutive processes and can

be studied separately and then combined into an overall rate, some-
what analogous to a series of resistances in heat transfer through
a wall. Steps 2 and 6 introduce some complications since active
sites are spread all through the particle so that the molecules
have to travel and, therefore, the resistance they encounter is not
the same for all of them.

It is obvious that transport processes may influence the overall

rate; heat and mass transfer between the gas and the solid,
generally, is not infinitely fast, so that the conditions locally
at the reaction site are not equal to those in the bulk gas phase
around the carbon particle.

In general, the rate of all these processes is proportional to

a driving force (V.F.), divided by a resistance term R:

D.F.
r "'T (2)

The simplified expressions for the different steps are given below.

II -1 EXTERNAL MASS TRANSFER

The transfer of material over a stagnant film around a solid

particle in fluid flow (steps (1) and (7»is:

k (c
B -cS ) (3)
J
g

where: J mol flux from bulk phase to particle surface

(molls m2 ). 3 2
kg mass transfer coefficient (m Is m )
. 3
CB concentration in the bulk phase (mol/m )
3
cS concentration at the external surface (mol/m )
 295

The mass transfer coefficient depends on the properties of the

gas phase gas flow rate and the shape of the particles and can be
obtained from correlations, e.g., in terms of the jo-factor. [21

11-2 PORE OIFFUSION

For diffusion in a single pore Fick's law for steps (2) and (6)
can be expressed as:

J= -~
dz
(4)

where: J mol flux in the pore cross section(mol/s m2 )

D the diffusivity (m2 /s) 3
c concentration in the pore (mol/m )
z distance in pore (m)
For large pore diameters, the diffusivity D is the ordinary mole-
cular gas diffusivity and can be found in Handbooks. D varies as
T1 . 50 and p-l. When the pore sizes are so small that its dimensions
are less than the mean free path of the gas, however, gas-gas colli-
sions are no longer dominant. Instead, gas-wall collisions are
important and the mode of diffusion transport is altered. The
Knudsen diffusivity DK takes the place of D:

(5)

where: M molecular weight of the diffusing species (kg/kmol)

r pore radius (m) (1 < r < 100 nm)

SO,D K varies with T~ and linearly with r, but is independent of

the pressure.

In a transition region between gas and Knudsen diffusion the

effective diffusivity is obtained by the semi-empirical relation:

(6)

When the pore size is in the order of magnitude of the molecules

themselves the so-called "configurational" diffusion takes place
with even lower values of the parameter D. In this case D is strongly
dependent on the temperature ("activated diffusion"). This kind of
diffusion is very complicated since specific details of the force
296

fields of the molecules and the walls have to be accounted for.

It might certainly be important in carbonaceous materials since
they have often pronounced micro-porosity. [8] Fig. 2 gives a
typical example of the diffusivity as a function of pore size.

11-3 SURFACE PHENOMENA

Generally, the steps 4, 5 and 6 are combined to derive a rate

expression including the adsorption and desorption phenomena and
the chemical transformation. Rate expressions often have the form:

kc (D.F;)
r (7)
R

where: r net reaction rate per amount of carbon present

(mol/mol. s)

k overall reaction rate constant, including the concen-

c
tration of active sites

R resistant term, which is a compound function of par-

tial pressures of reactants and products, rate- and
adsorption equilibrium constants

Regular
10-4
1 bar
10
1D
~10-8

10-12
ConfIgurational

10-16

0.1 m 10000
nm'-
Fig. 2 Diffusivity and size of pore aperture.

For an overall reaction aA+bB~cC+dD (8)

D.F. takes the form:

 297

(9)

and is in fact a measure of how far the reaction is away from

thermodynamic equilibrium.
Keq is the thermodynamic equilibrium constant:

c d

K
(P:: (?;)
f)

(p::S (P::f~
eq (10)

in which Pi are the partial pressures of the ideal gases i (bar),

Pref is the pressure of the thermodynamic standard reference state
for gases (1.01 bar) and v. the stoichiometric number in the reaction
expression (positive for pfoducts and negative for reactants) .

The equilibrium constant is a function of temperature and reac-

ting molecules and can be easily calculated from tabulated thermo-
chemical data such as the JANAF tables [ 9] :

RT inK (11)
eq

~o (T) (12)

where: free enthalpy change of the reaction

(kJ/mol) at temperature T.
free enthalpy of formation of compound i
(kJ/mol) at temperature T.
Apparently, the reaction rate (7) is a function of thermodynamic
constraints, i.e. chemical equilibrium (section III), the number of
active sites at the internal surface of the carbon (section IV and
V) and adsorption, desorption and intrinsic chemical phenomena,
which depend mainly on temperature and pressure (section IV and V) .

11-4 LIMITING PROCESS

The rate of the process (fig. 1) will be determined by the res-

pective rates of the individual steps. Incorporation of these steps
into a model that can be used in the description of a gasifier, is
needed for a practical design and falls outside the scope of this
chapter. Examples are given in ref. 2, chapter 4, and ref. 10.
298

In studying the kinetics of coal gasification preferably the

steps 1, 2 and 6, 7 are much faster than 3, 4 and 5, i.e. do not
determine the rate of the overall process.

The influence of internal and external mass transfer processes

can be verified experimentally (section VI), but also criteria have
been derived to calculate the presence of transfer limitations,
based on observed reaction rates [2]. For an isothermal situation
these are:

External mass transfer limitations are absent if:

L¥>i < 0.01 (13 )

P

where ~P. is the difference in partial pressure of reactant i

in the bUlk phase and at the particle surface and P the total pres-
sure in the stagnant gas film.

d 5c 2/3
p r(obs) (14)
p 6(1-e)jd F

with: d particle diameter (m)

p
Sc Schmidt number
E: porosity of the coal bed
J the J d factor
d
2
F superficial molar flux (molls m ) of the gas mixture
r(obs) the observed reaction rate (molls m3 )
Internal mass transfer is not controlling if the generalized Weisz-
Prater criterion is satisfied:
2 S
r(obs)L r(c ) « 1 ( 15)
S

f
c
2 Der(c)dc
c
eq
 299
3
where: observed reaction rate (mol/s m )
s 3
rate expression value for C (mol/sm)
CS concentration at the particle surface (mol/m 3 )
3
C equilibrium concentration (mol/m )
eq
L characteristic particle size (m) Vp/S the ratio
between particle volume and external particle
surface.

For an n-th order irreversible reaction the criterion simplifies

to:

2
r(obs)L n+l
-- « (16)
S 2
D c
e

In this respect the gasification rate expressions are frequently

considered to have a reaction order between 0 and 1. Neglection of
the strong inhibiting effect of CO on the reaction rate, however,
leads to a too optimistic conclusion about the absence of pore
diffusion limitations. This is clearly demonstrated by Petersen
[11] for the uncatalysed char gasification. Therefore, the genera-
lized Weisz-Prater critenonis presented above. It is noted that
it can only be used if the parameter values in the rate expression
are known since the criterion does not only contain observable
quantities.

Heat effects by endo- or exothermal reactions can cause tempe-

rature differences when the transfer of heat inside and to/from the
particle is insufficiently fast. It is obvious that this affects the
reaction rate, complicating the analysis of experimental results.
Criteria can be found in ref. 2.

Summarizing, temperature and concentration gradients throughout

the coal particles should be avoided. In the modeling of gasifiers
all these aspects must be included. In that case parameters for the
different heat and mass transfer phenomena have to be obtained
preferably by separate experiments. This paper is restricted to the
kinetic analysis of the surface phenomena.
300

III THERMODYNAMICS

Thermodynamic considerations related to the gasification of coal

or coal chars are important with respect to the theoretical evalua-
tion of limiting performance characteristics of process concepts
and to the practical design of experimental or commercial reactor
systems since the equilibrium constant and reaction heats are in-
cluded in the design expression. For these purposes the thermo-
chemical properties of the reaction components must be known. The
data of the gaseous components are accurately established [9],
whereas those of coal or coal char are difficult to measure due to
the complex and heterogeneous nature of these materials, as compared
to graphite. This aspect is discussed below.

The three main reactions involved in coal (char) gasification

are:

CD. C + 2H2 -+
-+-
CH 4

@ C + CO 2 -+
-+-
2CO

Q). C + H2 O -+
-+- H2 + CO

Other reactions can be derived from this set of independent reac-

tions by addition or subtraction. For instance, the water-gas-shift
reaction is obtained by subtracting 2 from 3:

The equilibrium constants of reactions CD to Q) can be expressed

as (the reference pressure is omitted for matter of convenience):

PCF4
x CF4
K (17)
eql 2 2
PH2· a C
x H • (p.a )
2 C

2
K
eq2
PCO X~o (; ) (18)
P CDl .a c x CDl c

K
eq3
~CO·PH2
P H20 ·a C
x CO ·xH2
x H2D (; ) c
(19)

K
eq3
K (20)
eq4 K
eq2
 301

where: x. mol fraction of component i

1.

p total pressure of the reaction system (bar)

a the "activity" of a particular carbon, referred

c
to graphite (for graphite a c 1)

The value a for coal chars is discussed in detail by Johnson

[5]. Researchefs have tried to establish values of a by measuring
the thermodynamic equilibrium composition for a partIcular coal
char and comparing this with the one calculated for graphite. Values
between 1 and 20 have been reported, but Johnson has shown that
kinetic effects are responsible for values exceeding unity. By
excluding other reactions Birch et al. [12] found for the gasifica-
tion of a brown coal with hydrogen that a = 1 at different pressures.
From the equilibrium measurements done byCBrowning and Emmett [ 13] ,
using a carbon formed from iron carbide decomposition, it could
also be calculated that a c = 1 for reaction lover a wide range of
temperatures.

In general the coal is transformed by heat treatment into a coal

char, which resembles a highly disordered structure of graphitic-
like domains, in analogy to the "glassy carbons" derived from poly
furfuryl alcohol by pyrolysis. Their structure is often indicated
as "poly crystalline-graphite" [7] . Therefore, it is not surprising
that coal chars in gasification reactions have the same free
enthalpy of formation, as graphite: ~Gf = 0 (graphite is the
reference substance in thermodynamic cdlculations). This assumption
enables the calculation of equilibrium gas compositions over a wide
range of conditions relevant to coal char gasification systems,
employing a value for a c of unity.

111-1 EQUILIBRIUM COMPOSITIONS

Equilibrium calculations have been made by numerous authors. The

calculations presented here are based on the JANAF tables [9],
equations (11) and (12) and (17) through (20). For convenience corre-
lations are given here for the Keq of reactions CD to @ in table
1. The equilibrium constant is expressed by a polynomial of the
form:

In K (21)
eq
 302

The parameters a O' a 1 and a have been estimated by non-linear

regression The accuracy of 2 the value of Keq is better than 1% over
the temperature range 400-1500 K, which covers the gasification
temperatures encountered in practice.

Table 1 Parameter values for the correlation of equilibrium con-

stants of gasification and related reactions.

Reaction aO a1 a2 average
% dev
C + 2H2 -+
+-
CH 4 - 13.63 - 11.816 - 0.521 0.5

C + CO 2 -+
+-
2CO 20.69 - 19.841 - 0.320 0.7

C + H2O -+
+- H2 + CO 17.375 - 16.590 + 0.1223 0.4

CO + H2 O -+
+- H2 + CO 2 3.49 3.563 0.313 0.9

C + 2H 2 O -+ CO 2 + 2H2 13.82 12.907 0.392 1.0

+-
CO + 3H 2 -+
+-
CH 4 + H2 O - 31.006 28.407 0.643 0.2

In fig. 3 the equilibrium mol fraction of methane is presented

as a function of temperature and pressure for reaction CD , assu-
ming ideal gas behaviour. Deviation of this behaviour can be
corrected by multiplying the partial pressure with a fugacity
coefficient. Values of this coefficient or correlations can be
obtained in various handbooks.

Fig. 4 presents the mol fraction CO as a function of temperature

and pressure for reaction @
The equilibrium trends of reaction CD and @ can be summarized as
follows:

- temperature increase:

- pressure increase

Equilibrium compositions for reaction Q) can be calculated, but

generally at lower temperatures CO 2 formation is observed and there-
fore, reaction @ should be included. This implies that the water
gas shift reaction ® is in equilibrium too.
Fig. 5 represents the equilibrium compositions. The CO 2 formation
goes through a maximum that shifts to higher temperatures and de-
creases with increasing pressure. H2 and CO are the only products
at high temperatures. At lower temperatures both decrease. Pressure
 303

_ _ .01 bar
...••...._._.1
_••••••• 1
_._._10
._._._._. 100

.4

.2

500

Fig. 3 Equilibrium mol fraction of methane in H2 -gasification as

a function of temperature and pressure.

.8 a~/--
.6
f !i
, , I '
,I

.4 !j i !
!: I ; _ _ .01 bar

)4/
................... 1
•••••••• 1
.2 _._._10
._._._._. 100

500 1000 1500 T 2000

K--:-

Fig. 4 Equilibrium mol fraction of carbon monoxide in CO 2 -gasifi-

cation as a function of temperature and pressure.
_ _ .01 bar
•••••••. 1 _ _ .01 bar
•. _._._._ 100

I
•••••••. 1
•. _._._._ 100
X .8 .8
X

.6
H 1.6 co
.4 .4 i
,.".,......
i
i
i
.2 .2 i

500 1000 1500 L 2000 1000 T 2000

K-
---i(-
Fig. 5 Equilibrium compositions for the H2 0, CO 2 , CO, H2 -C system
as a function of temperature and pressure.
 304

increase decreases also the H2 and CO formation, since the fraction

of H2 0 present increases.

Under practical conditions in the carbon-oxygen-hydrogen system

usually the following species exist in the gas phase: CO, CO 2 , H2 ,
H2 0 and CH 4 . For the equilibrium composition it must be assumed
tIiat reactlons CD, Q) and Q) (and therefore ® , too) are in
equilibrium.
The equilibrium composition of the O-C-H system can be expressed in
terms of temperature, pressure and the H/o ratio. Starting with pure
H2 0 and C (H/O = 2), fig. 6 can be obtained. The trends in fig. 6
are summarized in table 2. The arrow indicates whether a variable
increases (t) or decreases (+), or goes through a maximum (,,).

Table 2 Equilibrium trends for the C-H-O system

Temperaturet Pressuret H/o ratiot

x
H2 O + t
"
x
H2 t + t

x
CO t + ~

x
CO 2 n 1) +
x CH t 2) t
4

1) maximum fairly constant, but shifts to higher temperature

2) maximum shifts to higher temperatures

111-2 HEAT EFFECTS

During gasification of coal several phenomena occur that involve

enthalpy changes:

- Relatively cold coal particles are heated to reaction temperature

after introduction in the gasifier. The associated heat change can
be expressed as:
 305

t 1 _ _ .01 bar
•••••••. 1 t
_ _ .01 bar
•••••••. 1
X •. _._._._ 100 X •. _._._._ 100
.8 .8

.8 .6
co
.4

500 1000 1500 2000 500 1000 1500 2000

___ L _
K ---f-
Fig. 6 Equilibrium compositions for the allothermic H-O-C system
(H/O = 2) as a function of temperature and pressure.

'V
LYI= (22)

Correlations of Cp are given by Kirov [ 14]. These include devola-

tilization phenomena at lower temperatures < 500 o C.
- High temperature devolatilization and carbonization reactions,
leading to char formation. Lee [15] and Eiserman et al. [16] present
correlations for estimating thermodynamic properties.
- Enthalpy changes associated with gasification reactions. Table 3
lists some heats of reaction at different temperatures for
graphite as carbonaceous material (from 9).

Table 3 Heats of reaction /jJl (kJ/mol)

Reaction 800 K 1000 K 1200 K
C + 2H2 or CH 4 87.3 89.9 91.5
C + CO 2 or 2 CO 172.5 + 170.7 168.7
C + H2 O or CO + H2 135.6 136.0 135.9
CO + H2 O or CO 2 + H2 36.9 34.7 32.8

C + 02 or CO 2 - 394.5 - 394.9 - 395.3

2C + 02 or 2 CO - 222.0 - 224.2 - 226.6

306

Johnson [5,6] already observed that the heat effects associated

with the gasification reactions are much larger than those of the
devolatilization and carbonization reactions. The latter are not
further considered. Heats of formation of the coal char may differ
from those of graphite. Values up to 11 kJ/mol have been reported
for amorphous carbon [ 16]. Generally, heats of formation of coal
chars can be calculated from heats of combustion.

The main gasification reactions ~ and Q) are fairly endother-

mic. In gasifiers the heats necessary for these reactions can be
supplied either externally (allotherm reaction system [17]) or
internally by addition of oxygen to the gasifier as is done in most
gasifiers (autotherm reaction system). The oxygen reacts nearly
instantaneously with the carbon and provides the heat for the
endothermic gasification reactions.

Fig. 7 shows the heat changes associated with the equilibration

of the reaction system C-H-O. Gasifiers operating above 1500 K,
producing mainly CO and H2 , require considerable amounts of heat,
whereas at lower temperatures, where exothermal methane formation
becomes important, the reaction system is nearly thermo-neutral,
and oxygen has hardly to be added. Lower temperatures, however,
lower the reaction rates and in this region catalysts are needed
to maintain sufficiently high reaction rates. This concept is used
in the EXXON catalytic coal gasification process [ 18] .
Addition of oxygen not only lowers the calorific value of the product
gas but also changes the equilibrium composition. More CO is being
produced. This is envisaged in fig. 8 for an autothermic reaction
system. At a certain temperature H2 0 and 02 are added to the carbon
in such amounts that no net heat effect is observed upon reaching
chemical equilibrium. Fig. 9 represents the feed ratio 02/H ZO. It
is noted that at higher pressures less oxygen is needed. ThlS is
mainly caused by the increased formation of methane. The methane mol
fractions are hardly affected, since at low temperatures the reaction
system was already thermo-neutral. Only above 1000 K the allothermic
reaction system produces somewhat higher methane yields.
 307

140
"' ........ "",. . ....... -._.-
120 1 /' ,/' ,.,...
4H :~ /10 /
kJ lmol H:pconv 100 :' i /100 bar

80 / / ./
: I ;'
60
: i i
,/ i ./
4 :' / /
/ / /
20
. :;:;;:~',.,/
500 1000 1500 2000

----+-
Fig. 7 Enthalpy changes associated with the equilibration of the
allothermic H-O-C system of fig. 6 as a function of tempe-
rature and pressure. ~H is given per mol H20 converted.
_ _ .01 bar
•••••••. 1 _ _ .01 bar

I I
•. _._._._ 100 •••••••. 1
.'_'_'_'_ 100

i i·
co
..... -.-.
i'/
i
;
i
i .4
i

.2

500 1000 1500 2000

---f-
500 1000 1500 2000
---f-
Fig. 8 Equilibrium compositions for the autothermic H-O-C system
as a function of temperature and pressure.

Feed ratio .8

(~)

I .4

.2

500 1000 1500 2000

---~-
Fig. 9 The 02/H 20 (mol/mol) feed ratio for the autothermic system
of fig. 8 as a function of temperature and pressure.
 308

IV CARBON REACTIVITY

IV-l INTRODUCTION

Generally, the fate of a coal particle entering a gasifier can

be described in two steps:

- A fast pyrolysis step in which the coal particle is heated, vola-

tile matter (gases and tar) is released and carbonization reactions
take place, leading to the formation of the coal char.

- A slow gasification step in which the coal char reacts with the
gaseous environment of H20, CO 2 , H2 and, if present, 02.

For a given carbon Walker et al. [19] estimated the relative

rates at 1073 K and 0.1 atm for the different gases as:

(23)

3 1 3.1O~3

The absolute reactivity, however, depends on several parameters,

such as coal rank, volatile and mineral matter content and the
pyrolysis conditions.
During pyrolysis volatile matter is released and the porous char
structure is being formed. Some coals swell highly upon heating,
thus creating even more porosity and more carbon will be exposed to
the gaseous molecules. Mild preoxidation at 500 K can remove the
swelling properties and reduces the reactivity [ 20], so the history
of the coal before pyrolysis can influence the final reactivity. [21]

Many experimental observations have revealed that the char reacti-

vity often decrease with prolonged heating at elevated temperatures
(fig. 10 from ref. 22). This "annealing" of the char structure has
been attributed to the disappearance of structural defects, loss in
activity of specific mineral sites and to a decreasing specific
surface area. [23]

Under practical conditions the pyrolysis phenomena can play an

important role in the overall performance of the gasifier. No attempt
will be made here to treat this subject. [24] This chapter is
restricted to the relative slow gasification process of the coal
char.

The reactivity of a coal char or carbon will depend on the active

site density at the carbon surface, the total specific surface area,
the gas phase conditions, and temperature. Moreover, during gasifi-
cation the carbon is consumed and the site density or surface area
may change.
 309

2
fw
10-tis-1

1000 1200 7i 1400

___ ..:.IL_
K

Fig. 10 Effect of pretreatment temperature on gasification rates

of wood chars in CO 2 at 1.2 bar and 923 K (from [6]).

The reactivity can be expressed as follows:

(24)

s a .x (25)

where: total number of active sites per amount of carbon

present (mol/mol)
2
S specific surface area (m /mol C) (C0 2 adsorption)
a
2
surface concentration of active sites (mol/m )

Surface areas of porous carbons can be obtained by the BET method

(N 2 adsorption at 77 K) or the Dubinin Polanyi method (C0 2 adsorp tion
at 300 K). In the case of high micro porosity CO 2 adsorpt~on leads
to higher values since N2 molecules cannot enter these micro pores
at 77 K due to activated diffusion. Johnson (ref. 5, ch. 4) has
shown for four chars of different coal rank origin, that the reacti-
vity correlated excellently with the CO 2 surface area. In this
respect Walker et ale [19] note that for an efficient use of the
micro-porosity region in gasification a well-developed network of
wider feeder pores must be present.

As mentioned above the heat treatment prior to gasification

strongly influences NT The effects that cause this phenomenon are
310

usually lumped into one equation, originally developed by Blackwood

et al. [25] and adopted by Johnson et al. [6 ]:

(26)

where: deactivation energy (kJ/mol)

T char pretreatment or preparation temperature (K)

p

This is a rather empirical relation; it does not take into account

the soaking time at the pretreatment temperature. Nevertheless it
has been utilized ·successfully by many authors like Johnson [ 5] ,
McCarthy [21] and Sy and Calo [ 26] .

The decrease in NT with increasing pretreatment temperature can

potentially result from a decreasing Sa and/or decreasing A.
Johnson [ 6], however, shows that the Sa does hardly change for a
lignite char treated in N2 at temperatures between 1200 and 1373 K,
and concludes that only tne active site concentration A is changed.
Generally, however, above 1250 K the reactivity changes level off,
as can also be seen from fig. 10.

IV-2 UNCATALYSED GASIFICATION

During gasification of the char under chemically controlled con-

ditions, the reactivity changes due to changes in NT' which is
generally concluded to be caused by changes in the specific sur-
face area. The variation of the surface area with char conversion
("burn-off") is governed by pore growth, pore coalescence, smoothe-
ning of the pores and opening of closed pore space. Also thermal
annealing during burn-off may occur when the coal char is prepared
at the gasification temperature. Heat treatment at higher temperatures
removes this effect.

In experimental studies it has been observed that, during burn-

off, the specific surface area remains constant (ref. 6, ch. 4),
or passes through a maximum. The latter phenomenon is attributed to
the opposing effect of pore growth and their progressive collapse
by intersection. Models that account for these effects have been
developed by various authors, including Petersen [ 27], Hashimoto
and Silveston [28], Bhatia and Perlmutter [29], Gavalas [30],
Simons [31] and Chornet [32].
 311

In these models assumptions are necessary regarding poresize, pore-

length or poreshape. Sometimes probability functions are used,
which are hard to measure experimentally. The model of Bhatia and
Perlmutter [26] can explain both a monotonically decreasing char
reactivity, and a maximum reactivity as a function of burn-off.
Their "random pore" model requires only one parameter to predict the
rate-conversion behaviour:

(27)

471'L (1- € )
o 0
(28)

where: x fraction of carbon converted

ljJ "initial pore structure parameter"

o< ljJ < 2 monotonical decrease
ljJ ~ 2 maximum
L = ini tial pore length per unit volume (m/ m. 3)
o
S = initial surface area per unit volume (m 2/m 3 )
o
£ = initial porosity of the carbon
o

This model is incapable of explaining a maximum surface area at

conversion levels above 0.39. Nevertheless, it provided reasonable
fits to data of several investigators. [10, 29, 33]
Values of ljJ can be determined experimentally by fitting eq. (27)
with the surface area development during burn-off. Reported values
of ljJ vary between 1 and 14 [10, 29, 33].

Some limiting cases exist, corresponding with classical models

of the form:

r 'V (l_X)n (29)

312

For: 1. n o The reaction is zero order with respect to the

carbon, very unusual in the uncatalysed gasifica-
tion.

2. n The reaction is first order with respect to the

carbon, implying that the specific surface area
of the carbon is constant during burn-off.

3. n 2/3 This corresponds to the situation of shrinking

spheres. The char particle can be considered to
be either a solid sphere or to be composed of
several smaller spheres (grain model). It is
easily shown that the external surface of the
spheres varies with the sphere volume to the
power 2/3.

4. n = 1/2 Cylindrical grain model.

In general, only cases 2 and 3 are encountered and correspond

to values for ~ of 0 and 1, respectively, as can be seen in figs.
11 and 12 from ref. 29.

Summarizing, the gasification reaction rate can be generally

expressed as:

1 dNc
r
n NC • dt = X.S a (1-X){1-1/Iln(1-X)}~.f(T, Pi···' k i ···) (30)
o
i

th
or in terms of a n -order model:

r X.Sa (l-X)n f (T, Pi···' k i ···) (31)

n
o

where: r normalized reaction rate mol C gasified per second

n
and per amount carbon initially present
(mol/s mol C.)
~

A active site concentration (mol/m 2 )

2
Sa·
o
= specific surface area at X = 0 (m /mol C.)
~

X fractional conversion (-)

~ initial pore-structure parameter (-)

 313

f function expressing the intrinsic chemical reaction

rate as a function of temperature and pressure (s-l)

Alternatively the rate is often defined per amount of carbon actually

present. In the notation used here:

(32)

where: r specific reaction rate (s-l)

w

It is noted that the active site concentration A depends strongly

on the type of carbon used, the pretreatment temperature, and changes
sometimes also during gasification. For example, this is observed
during hydrogen gasification [6, 34].

By oxygen chemisorption Radovic et al. [35] found a reasonable

agreement between the chemisorption capacity, expressed as ASA (the
active surface area), and the reactivities of different chars.
Although this method provides higher values of the active site
density [36], it is useful in comparing surface reactivities of
different carbons. The only available method to obtain the true
active site density is by applying transient kinetic methods as
will be discussed in section V.

IV-3 MINERAL MATTER EFFECTS

It is well-known that the reactivity of chars towards CO 2 , H20

or 02 is increased by the presence of calcium [23, 37, 38].
Especially in lignites [38] calcium is highly dispersed and bound
to surface carboxylic groups. In this form calcium is a good gasi-
fication catalyst [ 39] .
Also other inorganic elements present in the coal char are poten-
tial catalysts that influence the reactivity [3].

Tomita et al. [40] studied the effects of several minerals added

to a polymer char in hydro-gasification. The iron containing minerals
pyrite and siderite showed increased activities. Both compounds had
been reduced to iron. In air these minerals showed little or no
activity.

Huttinger and Krauss [41] demonstrated that interaction of CaO

with FeS0 4 enhances the reduction to metallic Fe in hydrogen, indi-
cating that interaction between mineral constituents also can in-
 314

fluence their catalytic action. Generally, in hydrogen, metals are

present in the reduced state, and ,therefore, it is not surprising
that they can be catalysts in hydro-gasification.

Sakawa et al. [42] found that the gasification rate in CO 2 was

related to the alkali content of the ash of Japanese coals.

The above mentioned results indicate that the presence of the

mineral matter in the coal can influence the reactivities of chars.
Gasification should be considered as a catalysed or uncatalysed
reaction, depending on the relative contributions.

IV-4 CATALYSED GASIFICATION

Catalysts for gasification can be classified into three groups:

alkali metals, earth-alkaline metals and the transition metals.
Ref. 4 provides a good overview of this field.
The catalyst can enhance the gasification rate in two ways:
- Increasing the active site concentration.
- Changing the apparent activation energy by lowering energy barriers
of elementary steps or by providing an alternative route for
gasification.

Provided the uncatalysed reaction rate is negligible, compared

to the catalysed reaction, the porous structure of the coal char
only affects the dispersion of the catalyst.
If the surface area of the char is sufficiently large, the catalyst
concentration determines the concentration of active species.
Consequently, the reaction rate will be zero order with respect to
the carbon:

r (33)
n

where: number of catalyst active sites (mol/mol C initially

present)

This is observed in the CO 2 and H2 0 gasification, catalysed by

alkali metals [43,44] and In the H2 -gasification, catalysed by Fe,
Co and Ni [45] .
Increasing the catalyst concentration increases the activity without
 315

further affecting the apparent activation energy, so the number of

active sites is proportional to the catalyst content [ 44,46] .

In the case of high burn-off levels or a char with an initially

low surface area the carbon surface will be saturated with catalyst.
The reaction rate is determined by the carbon surface and will have
a positive order in the carbon present. Indeed Van Heek and co-
workers [46,47] have shown that for a Westerholt char the reaction
order amounted to 2/3 (~ = 1), independent of the catalyst content,
above 10-20% burn-off. A Leopold char [46] showed higher orders,
dependent on the catalyst loading.

Analogously to the uncatalysed gasification these orders can be

explained in terms of surface area changes. Wigmans et al. [48] have
shown that the specific surface area goes through a maximum during
the K-catalysed gasification of an activated carbon. If sufficient
catalyst is present which isab1e to disperse itself over the whole
surface, similar trends in gasification reactivity can be observed,
as envisaged in fig. 11.

Shadman and coworkers [49,50] proposed a model to describe this

effect of surface saturation by the catalyst during burn-off of
the carbon.

Several phenomena can further change the number of catalytically

active sites and complicate the description of reactivity:
- Sintering (earth alkaline metals [39] and transition metals [51] ) .
- Poisoning (sulfur destroys the activity of Fe [52] and Ni [53]}.
- Chemical transformation (Transition metals are only active in a
reducing atmosphere. Too much oxidizing agent transforms them into
their oxides [52]}.
- Reaction with mineral matter (Alkali metals react irreversibly
with aluminosilicates and become inactive for gasification [54]}.
- Evaporation of volatile catalyst compounds [ 55] .

In conclusion, the effect of catalysts on the description of the

reactivity of char in gasification reactions depends on the specific
surface area of the carbon, the catalyst concentration and disper-
sion and its mobility under gasification conditions to maintain
dispersion and contact with the carbon.
Two limiting cases exist for the rate expression:
- The reaction is zero order in carbon and expressions (30) and (31)
simplify to (33). The specific char surface is sufficiently large
to accomodate the whole amount of catalyst.
- The specific surface area of the char determines the catalyst
dispersion, and, as a consequence, the number of active catalytic
sites. ~'herefore expressions (30) and (31) remain applicable.
It is noted that, when the gasification mechanism is fundamentally
 0.5 1.0 1.5 2.0
---'1'-
02 0.4 0.6 0.8 1.0
--x-
Fig. 11 Development of the Fig. 12 Best fit values of n
reaction surface with in the grain model as
carbon conversion accor- a function of the pore
ding to the random pore structure parameter ~
model, compared with the in eqs (27) and (29).
grain model for n = 2/3
(from [ 29] ) .

changed by the catalyst, the function f also changes compared to

the uncatalysed reaction.
 317
V KINETIC MODELING

V-l INTRODUCTION

In kinetic modeling reaction rate expressions are derived from

kinetic models, consisting of a series of elementary reaction steps
(adsorption, surface reactions and desorption) pertaining to the
overall reaction. Generally, the kinetic models are based on proposed
reaction mechanisms for which evidence is provided by methods used
in surface science, such as the variety of spectroscopic techniques (IR,
UV, XPS, AES, UPS, etc.). It often occurs that reaction mechanisms
change, due to more recent information, but that the kinetic model
remains essentially the same, and the rate expression is unaltered.

Examples of kinetic modeling are given by Froment and Bischoff

[2], Boudart [1] and reviews related to gasification reactions are
given by Lowe [56], Ergun and Mentser [57], Walker et ai. [7,19],
Laurendau [58] and Johnson [5,6].

The concepts frequently encountered in kinetic modeling such as

steady- state and transient-state approximation, rate determining
steps, etc., will be introduced below by discussing the carbon-
carbon dioxide reaction as an illustrative example. The other reac-
tions can be treated in an analogous way.

V~2 THE CARBON-CARBON DIOXIDE REACTION

Reif [59] and Ergun[60] proposed model I, based on a two-step

oxygen exchange mechanism:
k1
CD Cf + CO 2 -+
+-
C(O) + CO
Model I k_1

@ C (0) k2 CO + Cf
-+

In step CD a carbon dioxide molecule dissociates at an active site

Cf,releasing a carbon monoxide molecule and forming a solid carbon-
oxygen complex. The actual carbon gasification occurs in step @
with the dissociation of the C-(O) complex from the carbon matrix,
leading to the formation of CO and a new active site.

The reaction rates of the steps can be expressed as:

(34)

k2 [ C(O)] (35)
 318

which are equal to the overall rate r.

Moreover it can be written for the free and the oxygenated carbon
sites Cf and C(O):

d[ Cfl = -d [ C ( 0) 1
dt dt (36)

V-2.1a ~!~~~y:~!~!~_~~~~!!!~~~
Under steady-state conditions and the assumption that the total
number of active carbon sites NT is constant:

(37)

(38)

with:

() (39)

the fraction oxygenated surface sites

To simplify the rate equations it is often assumed that one step

determines the rate, while the other steps are in (quasi) equili-
brium. This is equivalent to:

(40)

and (38) + (39) become:

r (41)

where: the equiJibrium constant (42)

of step CD
 319

Rate expressions (38) and (41) are of the familiar Langmuir type,
in which it is assumed that the heat of adsorption is independent
of the surface coverage, i.e. the surface species have no inter-
action. This is not always valid and attempts have been made to in-
corporate enthalpy distributions in the description of the adsorption
phenomena, leading to the Freundlich and Temkin isotherms [1]. This
introduces one or more additional parameters in the rate expression
and increases the number of unknown variables that have to be deter-
mined.
As is shown by Boudart [1] often the assumption of a distribution
of adsorption enthalpies leads to a rate expression identical to
that of a uniform surface.
Apparently a non-uniform surface behaves more like a uniform surface
than might have been expected from a knowledge of its thermodynami-
cal behaviour. This justifies the common practice of neglecting non-
uniformity of surfaces in kinetic studies. [1].

Assuming that in model I step (6) is reversible, a more general

expression is obtained of the form anticipated in section II-3:

r (43)

Here, Keq is the equilibrium constant of the overall reaction,

calculated in section III.

From expressions 39, 41 and 43 it appears that CO can have

an inhibiting influence on the gasification rate. Correlating the
rate expressions with experimental data, thereby evaluating values
for k. and K., finally reveals which of the rate expressions describes
the g~sification process best. One should, however, not restrict
oneself to one model.

Gadsby [61] assumed a competitive CO-adsorption on a free carbon

site:

Q) Cf + CO 2 -+ C(O) + CO
Model II
@ Cf + CO -+ C(CO)
+

® C(O) -+ CO + Cf
 320

leading to:

o (44)

r (45)

As is often observed, rate expressions derived from different models,

are mathematically identical, only the parameters k. and K. have a
different meaning. As will be illustrated below, a further 1 examina-
tion of these parameters offers a possibility for discrimination
between these models [62].

Mechanistic evidence might reveal that e.g. model I consists of

more elementary processes. Including these in the rate expression
yields more complicated mathematical forms and contains more unknown
parameters, whereas these processes may be kinetically insignificant.
Generally in expressions of the form like (43), using the concept
of the rate determining step, the terms in the denominator represent
the relative concentrations of the reaction intermediates. Boudart
[1] introduced the concept of the mari (most abundant reaction
intermediate) in combination with the rate determining step. The
denominator of the rate expression then only contains the terms
representing the concentration of the empty sites and the mario
To investigate which surface species are mari's, isotope labelling
is an appropriate method. Although less relevant for the Co 2 -gasi-
fication, it is significant for the H20 and H2 -gasification. In
those cases more reaction-intermediates can be proposed.

The above mentioned procedure simplifies the kinetic expressions

of reactions, for which detailed mechanisms have been proposed but
which are a disaster for engineers operating industrial gasifiers.

It should be kept in mind that rate expressions and kinetic

models are not unique. Depending on the experimental conditions
(temperature, pressure) another step might become rate determining
or products may be formed via an alternative path way.

V-2.1b ~~~E~~~~~~~_~~E~~~~~~X
The rate constants k. and equilibrium constants K. of the elemen-
tary steps in the kinetIc models vary with temperatu?e, according
to Arrhenius-type relations [1,63]:

k k exp(-E /RT) (46)

o a
 321

and:

K exp(~/R).exp(~/RT) (47)

where: k pre-exponential factor

0

E activation energy (kJ/mol)

a
R gas constant (8.314 .J/mol K)

/:c,S entropy change, associated with the reaction step

(J/mol K)

M enthalpy change, associated with the reaction step

(kJ/mol)

According to the absolute rate theory for a unimolecular reaction

the pre-exponential factor is given by (1,73):

k (48)
o

-23
where: k Boltzmann constant (1.38.10 J/K)
B
-34
h Planck's constant (6.6256.10 Js)
:j:
f f partition function of the transition and the
initial state, respectively
:1=
M entropy change associated with the formation
of the transition state from the initial state
(J/mol K)

When the obtained parameters k. and/or K. exhibit a different tem-

perature dependency, the assum~tions mad~ in deriving the rate
expression should be reconsidered or even the kinetic model should
be rejected.

V-2.1c ~~E~~~~~~!~~_~~~~~!~
In fig. 13 some literature results are presented for the specific
322

gasification rates, rw in CO 2 at 1 atmosphere pressure (adapted

from Johnson [ 5]). .
The apparent activation energies are generally clustered between
220-260 kJ/mol [5,10,37,64,65], but deviations of 60 kJ/mol are no
exception. As is apparent from fig. 13 the pre-exponential factor
NT.k ) can vary considerably. If k is considered to be a fundamen-
tal ~ate constant, similar for dif~erent carbon solids, major diffe-
rences in activity can be attributed to the total concentration of
active sites, NT.

Analysis of the kinetic data for the CO 2 gasification reveals

generally that model I adequately describes the experimental results
[5,19,60]. In fig. 14 values of K1 are plotted as a function of the
temperature (adapted from [ 5, 60 and 66]). The values of K1 for
different carbons show a surprising similarity, and give support to
the supposition that K1 is in fact an equilibrium constant for the
oxygen exchange reaction and is relatively independent of the carbon.
Erguns data were correlated by:

3
4.15·10 exp(-96300/RT) (49)

In this expression ~H = 96.300 kJ/mol can be considered as the heat

of the endothermic reaction CD in model I. strange and Walker [65]
pointed out that, if the carbon-oxygen bond in the C(O) complex has
about a two-thirds double bond character, this corresponds to an
enthalpy change of this magnitude. This supports the kinetic model
I and indicates the importance of examining the temperature depen-
dency of the parameter k. and K. in rate expressions.
~ ~

V-2.2 TRANSIENT CONDITIONS

A rate expression like (43) contains 5 independent terms of which,

in principle, only four parameters can be estimated:

Sometimes independent information is available that can be used for

a better determination of the other parameters (K from thermodyna-
mics, adsorption equilibrium constants from adsor~~ion measurements,
etc.), but values of the individual rate constants can hardly be
obtained by steady state measurements.
Moreover, the number of active sites NT' characterizing the reacti-
vity, can neither be determined.
 323

10 • Activalecl graphite
• Ceylongntpf'lile
• Activated carbon

0.1

0.6 0.8 1.0 a6

r-1
- - - 1 0 -3 K-1-
Fig. 13' Fig. 14
Specific gasification rates in Oxygen exchange equilibrium con-
CO 2 at 1 bar for a wide range stant as determined by various
of coal chars (from [6]). investigators [5,6,60,65]. Data
pOints are from Ergun [60].

By application of transient techniques information on the individual

rate constants and site densities can be obtained [67].

V-2.2a ~~~~_~~~~~~X
Consider model I and a reaction system that is in a steady state
gasification condition. If suddenly the gas phase is changed into
an inert atmosphere, a step-change, the C(O) surface species, present
at the surface, are no longer produced but only consumed by reaction
CD • Assuming that the CO produced is efficiently removed:

d [ C(O)] (50)
-k 2 [ C(O)]
dt

From integration between t= ° and t:

[ e(O)] [e(o)] exp (-k 2 t)

66 (51)
324
<
where: the suffix ss refers to the steady statE:! (t 0) .

The number of sites decays exponentially with time, and can be

measured by the production rate of co. The total amount of co relea-
sed equals the total number of e(O) species in the sample. So, this
method yields two parameter values:
- the intrinsic constant k2
- the steady state e(O) concentration.
Since [e(O)] = B.NT the value of NT follows directly from combi-
nation with steady state measurements, which yield e.

A drawback of this method is the following. An abrupt change of

the gaseous environment often alters the properties of the system
and the results obtained do not reflect the response under steady
state conditions. This is clearly demonstrated for the methanation
reaction of co over Nickel catalysts [67] .

In these cases the use of isotopes is to be preferred. For in-

stance, under steady state gasification conditions the gas atmos-
phere is changed from e 16 0 2 to e 18 0 or vice versa. The steady state
is maintained, but for the·e( 16 0) species can be written:

(52)

resulting in:

[ee 60)] = [e(16 0 )] exp {-(k 2 + k 1P 18 )t} (53)

ss - C 0

Under conditions where k2 » k 1P e 18o(10W PC 18 0) this is identical to

(51). Otherwise, the Pe 18 o' thus the conversion level, should be
varied, or independent measurements of k_l should be made.

V-2.2b Rate constants

Isotope exchange under steady state conditions can also be applied
to determine the rate constants kland k_ 1 • In this case either carbon
labeled cO 2 and co are used, since the 13 e only plays a role in
the first step of model I, and step CD does not interfere, or
oxygen labeled gases under the condition that the rate of step CD
is negligible (low T) .
 325
13 12 .
- Using mixtures of CO 2 and CO the stElad~\ ~tate for~tJ.on of.
12cO or 13 CO yields directly the exchange reaction rate:
2
13 12
CO + CO 2

Since the reaction proceeds via a surface intermediate the mea-

sured rate includes the site density (model I is assumed to be
valid) :

13 kl 13
CO 2 + C f -+ CO + C(O)

12 k_l 12
C(O) + CO -+ Cf + CO 2

r (54)

Analogously to steady state kinetic measurements pressure varia-

tions can reveal the pressure dependency of the rate and the
values of (k-1NT) and (k 1NT ). This product cannot be separated
by switching experiments from labeled to unlabeled gases as the
carbon is not present in a surface intermediate, ablk to relax to
a new situation. Only the equilibrium constant Kl = k:l is ob-
tained, and can be compared with steady state gasification results.

Oxygen labeling again can provide the specific rate constants k_l
and k 1 •

In the following experimentaCl~2/c160 mixture is abruptly re-

placed by a C16 0 /c 1 6 0 mixture.
The reactions th~t occur when no gasification takes place (low
temperature) :
18 kl
C O2 + Cf -+ c 18 0 + C(18 0 )

16 k_l
C 0 + C (18 0 ) -+ Cf + c 160 180

16 kl
C O2 .+ Cf -+ c 16 0 + C(16 0 )

At t>O the concentration of labeled oxygenated surface species

varies as:
326

(55)

SO:

(56)

The rate of decay in the C(180 ) species can be followed by the

production rate of C16 0 180, and leads to k_ 1 • The total amount of
C16 0 18 0 released equals the steady state concentration of oxyge-
nated surface species, analogously to (51).

V-2.2c ~~~~~!~~~~~~_~~~~~!~
Relatively little work has been published using isotope labe-
ling [68-72] or transient methods [26] in the field of uncatalysed
gasification.

By carbon labeling Bonner and Turkevich [68] established that

the forward rate of step <D (model I) is faster than step ~ ,
later confirmed by Mentser and Ergun [72]. The latter found reac-
tion enthalpy of 71 kJ/mol, in reasonable agreement with their
gasification data [60].
In a later evaluation Lowe [56] points out that the rate data of
Mentser and Ergun for isotopic exchange are incompatible with the
gasification results and that their high exchange rate is possibly
due to a parallel exchange reaction, not involved in the gasifi-
cation cycle.
Brown [69] measured a carbon transfer to the surface of graphite
and sugar char from both CO and CO 2 • The sugar char was found more
active than the graphite. Carbon deposition from CO may occur
according to:

<D CO + Cf + e(o)

CO + C (0)

Attempts of Rao et al. [71] to study the oxygen exchange between

e 16 0 2 and c 18 0 2 over a graphite failed, because their reactor
system turned out to be more active than the graphite.
 327

Sy and Calo [26] measured the response of activated coconut

char on a step change of Ar to CO 2 . They used model I to evaluate
their results, but assumed that both steps were irreversible, which,
in view of discussion above, is questionable.
The rate parameters they obtained are:

10 -1 -1
1.03·10 exp(-232400/RT) s bar (57)

8 -1
1.66·10 exp (-187600/RT) s (58)

The number of active sites turned out to be temperature dependent,

too, decreasing with increasing temperature:

-9
1. 70·10 exp (113000/RT) mol/mol C (59)

Evaluation of these data leads to the conclusion that the apparent

activation energy of the process is about 63 kJ/mol, an unreasonable
result. This can be explained in several ways:
- the backward reaction of step CD should be taken into account;
- mass transfers limitations, which explain the decreasing site
density with increasing temperature;
- the reaction system does not reflect the system under steady state
conditions. Indeed, Cerfontain et al. [73] observed for this type
of experiments "overshoot" effects. Initially, the carbon surface
is oxidized and the associated heat effects may influence the
data.

The reversed step change CO 2 ~ Ar has been applied by Freund

[34] for estimating rate parameters and site density. His comparison
of catalysed and uncatalysed gasification is dealt with in section
V-4.

In view of the discussion above it will be clear that the results

of transient experiments can confirm or invalidate assumptions made
in the derivation of kinetic expressions. On one hand it will provide
useful information on the active site density and the rate constants
of individual steps of the model under consideration, but on the
other hand new questions may arise that require more research.
 328

V-3 OTHER REACTIONS

In table 4 some frequently encountered kinetic models for the

gasification in carbondioxide,hydrogen and steam are given, together
with their respective rate expressions. By the assumption of a
rate determining step @ and quasi equilibrium of the other CD,
and by neglecting backward reactions (initial reaction rate), sim-
plifications of the expressions (43), (60) and (61) are obtained,
of course.

In the hydrogen gasification one observes frequently a burst in

methane production during the initial gasification stages [6,34],
followed by an approach to a steady state. For the steady state Shaw
[77] reinterpreted results of other investigators and found that eq.
(60) gives a good description of the rate data. Apparent activation
energies for this reaction generally vary between 150-200 kJjmol
[5,34,40,79] .

Fig. 15 is a comparison of specific gasification rates in pure

steam, at 1 atm. (adapted from ref. 5). Again this shows the variety
in activity for different carbons, but comparable activation energies.
These are lower than for the Co 2 -gasification and usually vary
between 140-200 kJjmol [5,10,64,80].

Under practical conditions in gasifiers the gas atmosphere con-

tains mainly H2 0, CO 2 , H2 , CO and CH 4 , The gasification rate will
be governed by the partial pressures of these components,

At low pressures methane will be absent, An obvious kinetic model,

describing the reactions that take place, is a combination of model
I and IV, assuming that the active sites are identical for both
reactions:

model V CD CO 2 + Cf -+
+
CO + C(O)

@ H2 O + Cf -+
H2 + C(O)
+
Q) C(O) -+ CO + Cf
+

This leads to the following expression:

r
 Process Kinetic Model Rate Expression

k k N j CO
p2\
Cn... "f" t" I CD CO2 + Cf--CO + C(O) 1 2 T P C02 -
V'~ gas~ ~ca ~on ..... r =
Keq (43)
(6) C(O) ~CO+Cf k + (k +k) +k
2 -2 -1 PCO 1P C02

k k N 2 Cf4
"f" t" III CD + H2 -C(H2) 1 2 T P H2P
~j -
H Cf
2 gas~ ~ca ~on =- r = eq
I (60)
~ C (H2) + H2 _ Cf4 + Cf k + (k + k ) P + k _P
-1 1 2 H2 -rCf4

k k N CePH2\
" f" t" IV H2O - H2 C (0) 1 2 T K
CD Cf + +
jPH2O _P
H~f""I
CJ;V gas~ ~ca ~on - = eq (61)
@ C (0) -co + C r
- f k2 + k -1PH2 + k 1P H7.0 + k_ 2Pco

Table 4 Common reaction models and rate expressions for the C~, H2 and H,o
gasification [5, 25, 58-60, 74-78]
Vol
N
\0
 330

10-4

\_~
coconut Char, char

5K-'-
0.8 1 -1 1.2
--10-
Fig. 15 Comparison of specific gasification rates in pure steam at
1 bar, adapted from [6].

If the oxygen exchange steps <D and @ are in equilibrium, this

implies that the water gas shift is in equilibrium:

+
+

K (63)
eq

Ergun and Mentser [56,76] found that eq. (63) was satisfied for the
chars investigated by them, (table 5) and model V can describe their
results. The forward reaction rates of (62) can then be simplified
to:

(64)

The latter expression is equivalent to (41).

 331

Table 5 Comparison of equilibrium constants Kl for oxygen exchange

in C-C0 2 reaction with values calculaEed from H2 0-C reac-
tion (from [75,76]).

Temperature H2 O-C reaction CO 2 -C reaction

(K)
K2 K (K 1 =K 2 K1
eq /K
eSl
1273 0.23 0.58 0.40 0.54

1373 0.39 0.47 0.83 1.0

1473 0.80 0.39 2.0 1.8

Mlihlen et al. [64] found from independent measurements of the CO 2

and the H ° gasification that the site density was nearly equal for
both reaction systems, providing support that both reactions take
place at identical sites.
Bliek [10] observed, however, that k3NT determined from H2 0 and from
CO 2 gasification experiments of chars derived from lignites to anthra-
cite, differ considerably, both in value and temperature dependency.
This result seemed to indicate that the C-C0 2 and C-HZO reactions
proceed at different sites as had been suggested ear11er by Gadsby
et al. [61] and Long and Sykes [78]. Some direct evidence for a
different behaviour between CO 2 and H2 0 in gasification arises from
E.M.-studies of graphite by Yang et al. [81]. They observed that
H2 0 gasified the graphite according to the hexagonal crystal struc-
ture, whereas CO 2 -gasification occurred in a non-oriented manner.
Bliek [10] used a rate expression being the sum of the two indepen-
dent reaction rates (43) and (61) to describe his data.

At elevated pressures the gasification of carbon with gases con-

taining steam becomes even more complicated because of the formation
of methane.

Blackwood and McCrory [82] suggested that the following reaction

steps occur in addition to steps CD - CD of model V:

+ -+
+-

+ -+
+-

-+
+-
332

Given this, the overall gasification rates can be dQscribed.h~ the

sum of two complex expressions [5], one representing the net gasi-
fication rate leading to CH 4 , and the other leading to (CO + CO 2 )
formation:

r = r( CH4) + r(CO + CCll) (6S)

(66)

cf>.r(CO + CCll) (k- 4P H2 + k SP H2 + k 6P HiJ )NT·[k3k2{PHiJ - P coPH2/Keq4}

2
+ k 1k 3 {P CCll - pco/KeqS}]

+ k3k4k6NT{PHiJPH2 - P CoPCH4 /Keq3 }

(67)

(k_ 4 +kSPfI2) (k 1PCCll + k_1PCO + k3 + k_ 3PCO +k~HiJ + k_~H2)

+ (k4P~ + k -SPCH4) (k1PCO + k3 + k -~H2)

+ k 6P HiJ (k 4P H2 + k_ SP CH4 + k1PCCll + k -lPCO + k3 + k -#CO + k2FfI20

(68)
 333

Where the equilibrium constants K apply to the following overall

eq
reactions:

rn 2H2 + C +-+ CH 4

[lJ co + H2 + H20 + C -+
+
CO 2 + CH 4

[] C + H2 + H2 O -+ CO + CH 4
+

@] C + H2 O -+ CO + H2
+

I3J C + CO 2 -+ 2CO
+
The form i.n which these expressions are given suggests that gasi-
fication b~H4 production occurs via overall reactions IT] , [l] and []
(pathways~, ~,(3X4X6) and fi)(4X6), and by CO and C02 prod-
uction via overall reacti-;n";'l]~ and-~(pathwaysOO, ~, ~
and®@X9.
Equations (65) through (68) reduce to the simpler forms in table 4
if gasification is carried out in a single gas. These expressions
have not been applied to conditions where all the major gas species
are present [ 5]. This is of course understandable in view of the
large number of kinetic parameters involved, which tend to dilute
the fundamental significance of the numerical evaluations made of
such parameters. with a sufficient number of parameters every ex-
pression is able to correlate experimental data. In kinetic modelling
it is a rule of thumb to use as few parameters as is possible.
It is also apparent that even a relatively slight alteration of the
kinetic model described, can significantly alter the form of the
resulting rate expressions.
Because of such complications, kinetic data obtained under conditions
of the simultaneous formation of methane and carbon oxides have been
correlated in what should be considered as semi-empirical expressions
[5, 82-85]. As Johnson has pointed out [5] these models satisfactori-
ly predict gasification behaviour in practice.

V-4 CATALYTIC REACTIONS

V-4.1 INTRODUCTION

The use of catalysts in coal gasification has received increasing

attention the last decennium. The lower gasification temperature
offers the possibility to produce gases with a high methane concen-
tration in a one step process. EXXON has designed a process based
on this concept, employing potassium as a catalyst [ 18] . Also Bergbau
Forschung investigates the application of potassium in an allothermic
 334

process [86,87]. Most kinetic data concern the alkali metals, al-
though also considerable research has been done on the transition
metal and earth alkaline metal catalysed gasification.

Alkali metals and earth alkaline metals catalyse the H20 and CO 2
gasification; the transition metals also catalyse the H2-gasificat~on
(table 6). with respect to the steam gasification it has been repor-
ted that the alkali metals also enhance the formation of CH 4 from
CO and hydrogen [ 88], a property usually exhibited by the transition
metals

The earth alkaline catalysts do not affect the apparent activation

energy of CO 2 and H20 gasification [37,39,89,90] indicating that
their catalyEic act~on is due to an increase of the number of active
sites.

The alkali metals lower the apparent activation energy for these
reactions about 20-60 kJ/mol, and usually vary between 190-230 kJ/
mol for the CO [66,89,91,92] and 140-180 kJ/mol for the H °
gasifi-
cation [44,47,~1]. Increasing the catalyst loading does not further
change Ea, indicating that also in this case only the number of
active sites changes [44,66,89,93]. In agreement with this there
are no apparent differences between K, Rb and Cs [44,92] except for
differences in the number of active sites.

Reported apparent activation energies of the transition metals

for the different gasification reactions range from 140 to 230 kJ/
mol [45,94,95] .

It seems evident that all catalysts increase the number of active

sites, or provide an alternative reaction pathway.

Furthermore, it is noted that the catalytic gasification of coal

has a strong correspondence with the formation of carbon deposits
on catalysts and their removal with 02 or H20. See for an overview
ref. 96.

Table 6 Catalysts for gasification reactions [7,45,94,95,97,98].

C + CO 2 -+ 2CO Li, Na, K Rb,Cs

+
Ca, Sr, Ba
C + H2O -+
+
CO + H2
pt-group metals, Fe, Ni, Co

C + 2H2 -+ CH 4 Fe, Co, Ni, Pt- group metals

+
Cu
 335
V-4.2 CO 2 GASIFICATION

The majority of the articles on the kinetic behaviour of the

catalysts conclude that a two-step model adequately describes the
rate data:

model VI <D CO 2 + * +
-+-

~ 0* + C +
-+- CO + *
Where: * [99]
represents a
.
catalytically active site of unknown nature

Generally, the first step is considered to be in quasi equilibrium

and the second is rate determining. These models closely resemble
models I and IV for the uncatalysed reaction, so the rate expressions
will have the same mathematical form as expressions (43) and (61).
Only the meaning of the parameters can differ. NT will merely repre-
sent the concentration of active sites of the catalyst, and not
carbon active sites.

The kinetic approach, however, does not give detailed information

on the mechanism of the reaction. These may be different for the
different catalysts but the consecutive steps in these mechanisms
may all lead to the same kinetic model.
Kapteijn and Moulijn [43] considered several kinetic models for the
CO 2 gasification catalysed by potassium and concluded that model VI
represents the best model.
The rate expression is given by:

r (69)

Figures 16 and 17 show recent results on the temperature depen-

dency of the parameters Kl , and (k 2NT ).
Freund reports that the temperature dependency of the parameters Kl
was identical for different carbons with and without potassium or
calcium, and equal to the correlations of Ergun [60]. The values of
E 2 are essentially the same both for the catalysed and uncatalysed
g~sification (240 + 15 kJ/mol) in agreement with results of Kapteijn
et al. [100] for the different alkali metals Na, K, Rb and Cs.
Therefore, it is concluded that the catalyst only increases the
active site density of the carbon, but does not strongly interact
with the rate determining desorption step.

The observation that the apparent activation energy is lowered

by the alkali catalyst must be ascribed to the inhibiting effect of
 336

1.0
10-3

10-4 I 0.1

N T k2 K1
s-l

I 10-5 I 0.D1

10-6
O.OOl--L..--r-~-r---r-~-r-"""T--'
0.9 1.0 1.1 1.2 0.8 1.0 1.2 1.4
T-1 r-1
- 1Q-3K-1 - -
103 K-1

Fig. 16 Fig. 17
Comparison of (k 2 N T ) Comparison of K 1 values as a
as a function of temperature function of temperature for
for different alkali metals different alkali metals [ 100]
[100] and carbons [66]. and carbons [66].

co. The temperature dependency of the denominator in rate expression

(69) lowers the activation energy associated with k 2 • Since the
inhibition is strongest at lower temperatures, where catalysts are
being used, this effect will be observed for the catalysed gasifi-
cation. At higher temperatures, during the uncatalysed gasification,
this influence is much smaller and the observed activation energy
will approach that of k 2 •

The site density NT of the catalyst can be established by tran-

sient experiments, as stated in section V-2. For the potassium cata-
lysed gasification this was first applied by Cerfontain et al. [73].
They measured the responses to step-changes from CO 2 to He. This
will result in a decreasing evolution of CO, produced via step CD
in model VI.
For low CO-pressures the reversed reaction <D is neglected and the
CO-production rate is given by:

d [ C(O)]
(70)
dt

Moreover, from steady state measurements the value of the product

k2NT is obtained. Combined, this will yield values for NT. For a
10 wt. % K2CO containing activated carbon NT is about 1.4 .10- 2
mol active sifes/mol carbon initially present [ 73], corresponding
with ~ 70% of the potassium being active. Their results also indicate
 337

that the number of active sites remained fairly constant between

973 and 1023 K.

Freund [36J applied the same method to Carbosieve S containing

ion exchanged Ca and to the pure carbon. His results (fig. 18)
revealed that k2 had the same value for the catalysed and uncatalysed
reaction and was correlated by:

-1
6.64'10 9 exp (-224800/RT) s (71 )

in which the 95% confidence region of Ea am~unts to + 40 kJ/mol.

According to the same method he found that for the pure carbon NT
is constant over the temperature region studied and accounts to 4.7 .
• 10- 5 mol sites per mol carbon and is by a factor 100 higher for
the calcium-containing sample. These results support the idea that
Ca increases only the number of active sites.

Experiments with another carbon (spherocarb), pure and loaded

with potassium or calcium, however, did hardly show a decay, as
predicted by eq. (70) [66J. The CO-production nearly immediately
dropped completely. Possibly the rate constants kl and k_l are much
higher than for the carbosieve.
This result indicates that not only the catalyst but also the nature
of the carbon has a strong influence on the gasification phenomena.

Catalysed oxygen exchange rates over carbons (step CD) have been
measured by Mims and Pabst [101] and Cerfontain et al. [102] for
alkali/carbon systems. Results of Cerfontain et al. are given in
fig. 19. These rates are much higher than the overall gasification
rates. If this step is part of the overall process it can be assumed
to be in equilibrium under gasification conditions. Salient detail
of these experiments: the reaction rates are first order in Co-
pressure but independent of Co 2 -pressure, suggesting that the oxygen
exchange is more complex than the assumed one step process.

This prompted Cerfontain et al. [ 102] to describe the gasification

by the following reaction scheme:

CD * + CO 2 -+
+- *0 + CO

@ *0 + C -+
* + c(o)

Q) C(O) -+ CO

® CO + C(O) -+ CO 2 + C
 338

1(j4 eO Ca/carbon

~~
10-1
10-5
10-2
10-6
NT k2 10-3 k2
7 10-7 "7
I 10-4 I
1o-a
10-5

1.0 1.1 1.2

-~
10-31("1

Fig. 18 Comparison of k2 and k2NT values for a carbosieve S with

and without Ca, obtained from transient and steady state
experiments [36].
-7

In
I
fOJ( -8

1-9
-10

-11 * Nori!

1.0 1.1 1.2 r-1 1.3

--10-3W'-

Fig. 19 Oxygen exchange rate between co and cO 2 as a function of

temperature over Na, K and Cs containing activated carbon
[ 102] .

The oxygen exchange between co and cO 2 takes place via step CD

over the catalyst at a high rate, the equilibrium far to the right.
In step CD the oxygen is transferred to the carbon, either by a
spill-over mechanism or by a mobile catalyst species. The desorption
of the co from the carbon matrix is the final gasification step.
Since this reaction is probably not affected by the catalyst, Co can
also remove the oxygen in the C(O) complex via step ~ at comparable
rates, analogously to the uncatalysed reaction, and can explain the
CO-inhibition.

This example illustrates that transient studies provide answers

but also raise even more new questions.
 339

V-4.3 H2 0-GASIFICATION

Analogously to the gasification in CO 2 a two step model also has

been applied here for most catalysts [97,103]:

model VII +-
-+

in which the second step is considered to be rate determining.

Although gasification processes have been designed based on the

application of catalysts [ 18,87], the reported results on the kine-
tic modeling of the catalysed stearn gasification are relatively few
in number.

Mims and Pabst [103] indicated that the rate depends strongly
on the PH 2 0/pH 2 ratio. Eq. (69) can be linearized for this reaction
(model VII) to:

(72)

Their data are used to construct fig. 20 and show the good corre-
lation of eq. 72. Kl is calculated to be 0.1 (at 977 K). Corrected
for the water gas shift equi librium constant this yields a value of
0.06 for the co 2 -gasification, and agrees well with results of the
CO 2 -gasification (fig. 17). This indicates a similarity between the
CO 2 - and H2 0-catalysed gasification.

The formation of methane during the stearn gasification occurs

only at high CO and H2 -pressures [18]. Kapteijn et al. [88] propo-
sed the catalysed formation of reactive carbon which is subsequently
hydrogenated:
cat
2CO -+

Only Lewis et al. [104] have presented some rate data. Under their
conditions the water gas shift was found in equilibrium. CO , H
and CO depress the gasification rate in increasing order. T~e e!fect
of CO is explained by the formation of CO 2 and H2 via the water
gas shift, thus inhibiting the reaction rate. The rate data were
correlated by the empirical relation:
340

2
1
r,;-

Fig. 20 H2o-gasification rate data at 977 K of [103] represented

according to eq. (72).

(73)

in which the water gas shift is in equilibrium.

V-4.4 H2 -GASIFICATION

Hardly any data are reported in the literature on the kinetics

of the reactions of the catalysed reaction, probably due to deacti-
vation of the catalyst during gasification, which hampers the assess-
ment of consistent rate data.

Huttinger and Schleicher [45] observed that the H2 -gasification

of a low reactivity petroleum coke, catalysed by Fe, Ni or Co pro-
ceeded at a constant rate up to 50% burn-off. The number of active
centers remained constant in absence of mass transfer limitations.
The following rate equation described the rate data satisfactorily:

r (74)
n

H2-concentration at thermodynamic equilibrium. The

 341

temperature dependency yields the following values for the activa-

tion energy (k 1 ) and the heat of H2 adsorption (K 2 ):

Ea (kJ/mol) f'lH(kJ/mol)

Fe 152 92
Ni 165 50
co 201 82

They point out that the lowest heat of adsorption corresponds to the
highest reactivity (Ni). The hydrogen is the most loosely bound and
can spill-over the surface to the centers where methane is being
formed.

The second order in the denominator of a rate equation generally

implies a bimolecular surface reaction or desorption step (1); this
might agree with the spill-over mechanism.
The driving force used by Huttinger and Schleicher is a simplified
form of the general form: (P~ - PCH IKe ),as mentioned in section
V-3. 2 4 q

It appears that the rate expression has no underlying kinetic

model, but without any mechanistic significance the following two-
step model leads to a similar rate equation as used by Huttinger
and Schleicher:

model VIII +

+
+-

with: N k { 2 I}
r = T 2 PH2 - PCH4 Keq
(75)
(1 + K1P ) 2
H2
V-5 TEMPERATURE PROGRAMMING

Sometimes temperature programming offers the possibility to ob-

tain relatively rapidly rate data over certain temperature range.
To satisfy the condition of steady state at each temperature the
heating rate cannot be too high. The reaction system must be able to
follow the change in temperature, also the rate of heat transfer to
the sample becomes important. An easy method to check if these con-
ditions are satisfied is cycling between desired temperatures and
determining the difference in the results of heating and cooling
step, and adjusting the heating rate, if necessary.
 342

A second application of temperature variations is temperature

programmed desorption (TPD) measurements:A certain amount of reactant
is adsorbed at the surface at a certain temperature. Then the sample
is heated at a constant rate and the desorption is followed. By
variation of heating rates the desorption process can be studied and
activation energies determined.

Since in coal gasification a kind of desorption step is proposed

to be rate determining, this method looks promising in studying the
rate determining step. By treating the sample in a reactive atmos-
phere the surface can be covered with C(O) species and subsequently
the experiment is carried out.

For desorption processes without readsorption the following rela-

tions can be derived (see for instance refs 1 and 100):

- First order desorption:

(76)

(77)

where: T temperature at maximum of desorption peak (K)

m

B heating rate (K/s)

desorption activation energy(kJ/mol)

8·314 J/mol K
-1
pre-exponential factor for desorption (s )

Varying B and plotting the left hand side of eq. (77) vs. l/ T will
yield Ed/R as slope and Ad from the intercept. m
It is noted that T at a constant value of B is independent of the
surface coverage, ~ypical for first order desorption kinetics.

- Second order desorption:

2 In T - ln~ (78)
M
 343
where Gois the initial coverage (molecules/m 2 or molecules/mol C).
Again varying 8 at constant Go or by changing 00 at constant 8 will
yield Ed and Ad' Note, however, that T depends on initial coverage
Go' which is proportional to the area ~f the desorption peaks. This
verifies that the desorption is indeed a second-order process. An
extensive treatment can be found in ref. 105.

TPD is nowadays common practice in heterogeneous catalysis and

the closely related surface science. Applying this method to coal
can give specific problems because of the residual oxygen content
of coal chars. Therefore, Keleman and Freund [106,107] used a pure
poly crystalline glassy carbon and studied CO-TPD after treatment
with CO 2 and with 02 at 300 0 C. The experiments were carried out in
a U.H.V.-chamber. By Auger Electron Spectroscopy (AES) the oxygen
surface concentration of the carbon was measured during adsorption
and during TPD.
The following conclusions could be drawn:

The adsorption activation energy of CO 2 increases with increasing

surface loading at 573 K: 60 kJ/mol at e = 0.1, 145 kJ/mol at e
= 0.5 and ~210 kJ/mol at e = 0.9.
- The CO-desorption was for 02 and CO 2 loaded surfaces identical.
The Ed varied from 370 kJ/mol at e = 0.1, via 320 kJ/mol at e = 0.4
to 290 kJ/mol at e = 0.9.

The overall activation energy for CO 2 gasification was determined to

be 280 kJ/mol. The authors conclude that under gasification condi-
tions the surface has a high oxygen surface coverage compared with
the U.H.V.-chamber, and that the CO-desorption apparently determines
the rate of CO 2 gasification.

Temperature-programmed desorption cannot be applied to determine

NT and k in the alkali catalysed gasification. Chemical transfor-
mation ot the alkali catalyst at the carbon surface by reduction in-
terferes too strongly, or even completely mimics the CO-evolution of
the gasification process [92] .

V-6 THERMODYNAMIC CONSIDERATIONS

In section V-2.1 it was pointed out that the rate and equilibrium
constants of elementary steps must follow Arrhenius-type relation-
ships (eqs. (46)-(48) ). Temperature variation thus may yield enthal-
py- and entropy changes over elementary steps and/or activation
energies and -entropies of elementary steps.

For a rate expression and the underlying kinetic model to be of

fundamental significance these thermodynamic properties must obey
certain rules or guidelines [1,108,109]. These can be used to test
the validity of the presumed elementary steps.
 344

In the case of adsorption equilibrium the rule is that ~Ho

should be negative, as adsorption is in general exothermic, ~gfle
entropy normally decreases upon adsorption, by loss of translational
degrees of freedom, although this loss cannot exceed what was avai-
lable to the gas phase molecule (SO ). This is valid even in case
of dissociative adsorption [ 109]. gas1t is obvious that for desorption
the reversed reasoning is valid.

Other equilibria can be treated analogously. For the first step in

kinetic model I can be written:

(79)

(80)

The unknown variables are the properties of the surface intermediate

C(O). Methods to estimate these unknown properties are given by
Benson [ 63]. The C (0) complex contains a localized c-o bond, so one
expects in the entropy change C ~ C(O) only vibrational contributions
of the c-o bond and possibly some contributions due to weakening of
the C-C bonds adjacent to the complex [65]. An upper limit of the
entropy loss can be estimated by assuming the formation of gaseous
species CO and o. Since 0 has only translational degrees of freedom
its entropy is given by:

R (1. 5 ln M + 2.5 ln T - 1. 16) (81)

and amounts to (at 1000 K) 168.5 J/mol K. This yields as upper limit
at 1000 K 160 J/mol K. Neglecting the entropy of the O-species gives
a lower limit.
Therefore,

-8 J/mol K < ~S~(1000 K) < 160 J/mol K (82)

Returning to the correlation of Ergun [ 60], later confirmed by Strange

and Walker [65], eq. (49) gives:

exp ( ~S/R) (83)

 345

and hence ~S = 69 J/mol K, well satisfying the conditions imposed by

thermodynamics. This supports the proposed kinetic model.

The activation energies should be positive and must be larger

than the enthalpy change for the overall step. In fact the activation
energy represents the potential energy barrier between the initial
and the final state, so at least the difference between the final
and initial state must be overcome.(fig. 21»:

(84)

....;::0..,..__ Final state

Inilial state _ _ _""

Fig. 21 Schematic reaction energy diagram.

This reasoning is also valid for the overall reaction, and implies
that the apparent activation energies must be larger than 170, 135
and -90 kJ/mol for the CO , H 0 and H2 gasification. One should
however be cautious, sinc~ th~ number of active sites may change
this value. If this number is temperature dependent and decreases
(or increases) with temperature, a lower (or higher) apparent acti-
vation energy will be found. Also mass transfer limitations result
in lower values (section VI).
Boudart [1] lists some order of magnitude estimates for the pre-
exponential factor of elementary steps which are given in table 7,
together with other physicochemical constraints.

Application of these constraints should occur with caution, in

view of the assumptions that have been made in deriving the rate
expression. If they are satisfied it should be considered as a support
for the model. If not, don't panic if the model describes the rate
data well, a non-uniform surface is always a good excuse.
346

Table 7: Physicochemical constraints and order of magnitudes for

the pre-exponential factor A.

k., K. > 0
~ ~

adsorption:
I'1H d < 0
a s
o <-l'1s < SO
ads gas
4 -1
A '\.. 10 s

unimolecular desorption or reaction:

-1
s

bimolecular desorption or surface reaction:

-2 2-1
A '\.. 10 cm s

generally
E > 0
Ea > I'1H (guideline)
a
 347

VI EXPERIMENTAL ASPECTS

VI-1 REACTORS AND DATA ACQUISITION

Laboratory-scale reactors for kinetic studies are not necessarily

scaled down replicas of reactor type ultimately used in the industrial
operation. The laboratory reactor has to furnish the kinetic data
with the greatest possible accuracy and it should be devised with
that sole purpose in mind.

In gas-solid reactions, kinetic experiments are preferably carried

out in flow reactors, generally of the tubular type, or in gradient-
less reactors. Weekman [110] has reviewed the merits and shortcomings
of different laboratory reactors.

To limit the complexity of the data treatment the flow in tubular

reactors has to approximate as much as possible the ideal conditions
of plug flow. A rule of thumb requires the ratios of the tube dia-
meter to the particle diameter larger than 10 to avoid short-circui-
ting along the wall.

The following relations can be derived under chemically controlled

steady state conditions:

Plug flow reactor: x

f~
W
F (8S)
o x
o

Perfectly mixed flow reactor:

W x - x
o (86)
F r{x)
o

where: F input flow rate of reactant at zero conversion (molls)

o
W amount of carbon in the reactor (kg or mol)

xo conversion of reactant at inlet

x conversion at outlet

r reaction rate expression as derived in section V

348

The expressions (85) and (86) constitute the relations between

independent variables (the experimental settings as partial pressures,
temperature, flow rates, carbon amounts) and the dependent variables
(fractional conversions).

Plug flow reactors can be operated. either at integral or differen-

tial conditions. At low conversion differences between in- and outlet
the reaction rate is constant over the carbon bed and eq. (85) redu-
ces to (86).

The strong inhibition by CO, however, invalidates this assumption

for the cO 2 gasification. This is envisaged in fig. 22, representing
a residual plot as a function of the temperature for a differential
and integral analysis of cO 2 gasification data. Although the conver-
sion level of the pure CO 2-feed is below one percent it is clear
that the differential approach is erroneous, whereas the integral
analysis yields a good description.

Since in coal gasification the solid phase disappears,so-called

micro-balances are frequently used to measure the gasification rate
by following the weight chahge of the sample. As pointed out by
Holstein [ 111], microbalances should be used under differential con-
ditions, employing mixtures of reactants and products. When using
pure reactants, the product partial pressure varies throughout the
carbon sample, and since the gas flows around, rather than through
the carbon bed, the plug flow analysis cannot be used. There is no
method available to calculate true kinetic parameters from such
experimental data. Furthermore, mass transfer limitations are more
severe in microbalance systems. This aspect requires always a
thorough examination of the data [ 112]. To avoid this problem Kapteijn
et ale [113] developed the so-called packed-bed balance-reactor,
which can be analysed as a plug flow reactor.

VI-2 DATA HANDLING AND PARAMETER ESTIMATION

VI-2.1 DIFFERENTIAL METHOD

With pure reactant feed and low conversions, the backward reaction
rate and products formed after the rate determining steps can be
neglected in the rate expression, which reduces to the "initial
reaction rate". For the H20 gasification (model III) this yields:

r step CD rate determining (87)

r step GP rate determining (88)

 349

5.10 .- 10.,'3

I I
..
. ...... .... --.. .
X-Xobs
X-Xobs

I 0
... .
\ ....~!'.:
.... ..-. I 0
.
.:.
......4,. ..... c.~:
.
..
.-.
.. -.

-5.10- -1
900 1000 L _
___ 1100 900 1000 1100
_ __ _ T
K K

Fig. 22 Comparison of residual conversions as a function of temperature.

Rate data of a K-catalysed CO 2 -gasification experiment at
1.5 bar interpreted, according to the differential method
(left) and to the integral method (right). CO 2 conversions
below 1%.

Since VI-2.1 suggests no influence of ,the hydrogen pressure, the

rate determining step can be easily established [2].

Parameter values are estimated upon minimizing an objective function

(O.F.) [114], which is generally the sum of squares of residuals,
i.e. the difference between the observed and calculated dependent
variable (SSR).
In this case:

{x. (obs) - x (calc)}2

~
OF SSR t'"• i
-~-----~~-------
w. (89)
~

with: wi a weighing factor in the minimization procedure.

Since most of the rate expressions are non-linear functions of

the conversion this must be achieved by non-linear regression methods
[2,115,116].

Linear least squares minimization can be applied by taking the

reciprocal of the rate expression, transforming it into a linear
form. Usually this gives the same parameter estimates, but for
reasons of statistics the confidence regions of these estimates are
incorrect [ 114]. The confidence intervals are given by:

~
parameter value = b i + tv I 0.025 [ v(b i)] (90)
 350

where: b, parameter estimate

~

variance of b i

t student's t-value at V degrees of freedom and 95%

v,0.025
confidence level

V n - n number of data points minus the number of parameters.

i p

Usually parameter estimates are correlated, and the confidence

intervals are approximations. Confidence contours of the SSR-value
as a function of the parameter estimates take this into account and
can be constructed from [114]:

n
SSR O• 95 SSRm.;n
...
, .-R
+ SSRm~n JJ
F(npJJ" 0. 95) (91)

where: F Fischers F-value at (n ,v) degrees of freedom at a 95%

p
significance level.

VI-2.2 INTEGRAL METHOD

Integration of expression VI-l.l leads to:

w
F = f (x, k i ···, K i ···) (92)
o

,E" w 2 E
E~ther , (F (calc) - F'. (exp) ) or , (x(calc) - x(exp»
2 can be m~n~-
mized. The~re~ression i~ generally Aon-linear and in the second case
the computations are more complicated because the equation is impli-
cit in x. For convenience the first method is most frequently applied
and yields usually proper parameter estimates, although not a depen-
dent variable is minimized. Correct confidence intervals are obtained
by the second method.

Kinetic analysis of more complex systems in which more than one

conversion can be measured should be performed by multi-response
methods [ 2] •
 351

In order to select an adequate rate expression several aspects of

regression analysis must be considered [114,117,118J :

- Lack-of-fit-test
The SSR is composed of a pure error sum of squares, SSE, due to
the errors in the experimental data, and a lack-of-fit sum of
squares, SSL, the incapability of the rate expression to correlate
the experimental data. SSE can be estimated from replicated data
at the same experimental settings. An expIEssion is inadequate if
[ 117] :

mean-SSL
mean-SSE> F (v l' v 2: 0.95) (93)

- Residual analysis
The distribution of the residual variable (that is minimized) as
a function of an independent variable should be random with zero
mean in case of a proper model. Trending effects in this residual
distribution indicate the incompatibility of the rate expression
with the experimental data, or can result from a wrong data inter-
pretation (integral vs. differential) as shown in fig. 22,

- Parameter limits
If the confidence interval of the parameter estimates, eq. (90),
includes zero it is doubtful whether the p~rameter is necessary
in the rate expression. For strongly correlated parameters it is
better to examine their confidence region given by eq. (91).

- Physico-chemical constraints
To have a fundamental significance the parameter estimates must
satisfy certain rules and guidelines indicated in section V-6.

In a first approach the above mentioned will be sufficient to

remove inconsistent rate expressions. Advanced methods exist to dis-
criminate between rival kinetic expression. Based on the information
of previous experiments, new experimental settings are determined
for the next experiment, in order to obtain the maximum divergence
between the expressions. This sequential design method [2,114] to
discriminate between rival models can also be applied to optimal
parameter estimation by increasing their accuracy. This is envisaged
in fig. 23 (from Froment [2D. As a rule of thumb the experimental
settings, thus determined, are usually located at the boundaries of
 352

EXPERIMENTATION _____ Theoretical

I
_ _ Observed

--
DISCRIMINATION LOOP

D_.
....._lwrlSt"""ulnu' ..... "'odIIl
In,

I
Docrll'll""UOII

DeslgncnlerlOl'l

External
transfer

--1/1-

Fig. 23 Sequential ~rocedure for Fig. 24 Idealised three tem-

optimal des1gn]of experi- perature-zone scheme
ments (from L 2 • for the overall rate
as a function of
temperature.

the operability region.

The sequential method for the design of an experimental program

can permit a substantial saving in experimental effort for equal or
greater significance, with respect to classical procedures. Automatic
application of these methods, no matter how powerful they are, should,
however, not be substituted for sound judgement. Simple visual in-
spection of the rate equations may already reveal regions of maximum
divergence, although this may become difficult or perhaps impossible
with complex, multi-variable models.

VI-3 MASS TRANSFER LIMITATIONS

Throughout the kinetic analysis it was assumed that the gasifi-

cation reaction was chemically controlled. Kinetic analysis requires
verification of this prerequisite. Several experimental tests can
be performed to check this:

a. External mass transfer limitations are diminished in plug flow

reactors by increasing the flow rate through the reactor. This
decreases the stagnant gas region around the particle and mass
transfer is increased. If this results in higher conversions at
equal ~ ratios, but different flow rates, mass transfer was indeed
contro{ling.

b. Pore diffusion limitations can be verified by varying the par-

ticle size. The length over which reactants have to travel is
changed and changes in observed reaction rates indicate the pre-
sence of such limitations.
 353

c. Binary gas diffusion limitations, when using mixtures of reactants

and inert gases, can be detected by changing the inert gas.
Diffusion rates in He are higher than in A or N2 , resulting in
faster transport of reactants and products: Especially in micro-
balances this method can detect transport limitations, as is
nicely demonstrated by Wigmans et al. [112].

d. Variation of the temperature yields a picture like fig. 24. At

low temperatures the rate is chemically controlled. The reaction
rate has a stronger temperature dependency than diffusion and at
higher temperatures pore diffusion is rate limiting. It can be
shown that the apparent activation energy decreases by a factor
two [ 2]. At even higher temperatures external mass transfer be-
comes controlling and the apparent activation energy becomes
virtually zero.

It is good to emphasize that in practice the three - temperature -

zones picture is not necessarily obeyed:
- A direct transition from chemical control to external diffusion
control can occur, e.g. in microbalances, where the stagnant film
thickness is relatively large compared with the pore diffusion
length.
- The rate controlling step can change, or the relative influence of
other steps (equilibrium constants) can change with temperature,
so the overall- or apparent activation energy will change.
-·Also poisoning, sintering and dispersion phenomena can affect this
picture resulting in deviating temperature behaviour, especially
in catalysed gasification.

VII CONCLUDING REMARKS

This chapter describes several aspects of the kinetics of coal

char gaSification ranging from rate expressions based on kinetic
models, carbon reactivity, catalysts and parasitic phenomena as
transfer limitations and catalyst deactivation. Moreover, it provides
information and suggestions for experimental approaches. Although
some subjects may have been omitted or underemphasized, the tools
and suggestions provided here for theoretical and experimental approach
in this field probably are sufficient for further research in this
area. It is hoped that both beginning and more advanced researchers
can take profit of this overview.

~~~~~~!~~~~~~~!
The authors gratefully acknowledge the financial aid of the Project
Office for Energy Research (PEa), financed by the Ministry of Economic
Affairs, within the framework of the Dutch National Coal Research
Programme.
 354

VIII References

1. M. Boudart and G. Djega-Mariadassou, in: "Kinetics of Heteroge-

neous Catalytic Reactions", Princeton University Press, 1984,
Princeton N.J.

2. G.F. Froment and K.B. Bischoff, "Chemical Reactor Analysis and

Design", John Wiley and Sons, 1979, New York.

3. Robert M. Davidson, "Mineral effects in coal conversion", lEA

Report ICTIS/TR 22, London 1983.

4. J.R. Pullen, "Catalytic Coal Gasification", lEA Report ICTIS/

TR 26, London 1984.

5. J.L. Johnson, "Fundamentals of Coal Gasification", in:

M.A. Elliot, "Chemistry of Coal Utilization", John Wiley and
Sons, 1981, ch. 23.

6. J.L. Johnson, "Kinetics of Coal Gasification", John Wiley and

Sons, 1979 .

.,. P.L. Walker jr., M. Shelef and R.A. Anderson, in: "Chemistry
and Physics of Carbon", vol. 4 (1968) 287-383.

8. P.L. Walker jr., Fuel 59 (1980) 809-10.

9. JANAF Thermochemical Tables, edited by D.R. Stull and H. Prophet,

NSDRS, NBS-37, US Government Printing Office, Washington D.C.
(1971) .

10. A. Bliek, "Mathematical Modeling of a Co-Current Fixed Bed Coal

Gasifier", PhD.Thesis, Twente University, The Netherlands, 1984.

11. E.E. Petersen, "Chemical Reaction Analysis",

Prentice Hall, Englewood Cliffs N.J., 1965, ch. 4.

12. T.J. Birch, K.R. Hall and R.W. Urie, J. Inst. Fuel 33 (1960)
422-35.

13. L.C. Browning and P.H. Emmett, J. Am. Chern. Soc. 73 (1951)
581-3.

14. N.Y. Kirov,BCURA, Month. Bull. 29 (2) (1965) 33-57.

15. A.L. Lee, Prep. Am. Chern. Soc. Div. Fuel Chern. 12 (3) (1968)
19-31.
 355

16. O.A. Hougen and K.M. Watson, Industrial Chemical Calculations

2nd ed., Wiley and Sons, New York (1936), p. 487.

17. H. Kubiak, H.J. Schroter, A. Sulimma and K.H. van Heek, Fuel 62
(1983) 242-5.

18. N.C. Nahas, Fuel 62 (1983) 239-41.

19. P.L. Walker jr., F. Rusinko jr. and L.G. Austin, Adv. Catal. XI
(1959) 133-221.

20. K. Gomi and Y. Hishinuma, Fuel £l (1982) 77-80.

21. D.J. McCarthy, Carbon 9 (1981) 297.

22. J.D. Blackwood and A.J. Ingeme, Austr. J. Chern. 13 (1960) 194-
209.

23. R.G. Jenkins, S.P. Nandi and P.L. Walker jr., Fuel 52 (1973) 288.

24. J.A. Moulijn and P.J.J. Tromp, this book.

25. J.D. Blackwood, D.J. McCarthy and B.D. Cullis, Austr. J. Chern.
20 (1967) 2525-8 and 1561-70.

26. O. Sy and J.M. Calo, Proc. Int. Conf. Coal Sci. 1983, Pittsburgh,
p. 445.

27. E.E. Petersen, AJChE-J. 3 (1957) 443.

28. K. Hashimoto and P.L. Silveston, AIChE-J. 19 (1973) 259.

29. S.K. Bhatia and D.D. Perlmutter, AJChE-J. 26 (1980) 379.

30. G.R. Gavalas, AIChE-J. ~ (1980) 577.

31. G.A. Simons, Fuel 59 (1980) 143,.

32. E. Chornet, J.M. Baldasano and H.D. Tarki, Fuel 58 (1979) 395-6.

33. K. Hashimoto and K. Miura, Report. Spec. Proj., "Research on

Coal Liquefaction and Gasification", Min. of Education, Science
and Culture, Japan, SPEY 12, 1984, 229-38.

34. H. Jlintgen, Carbon ~ (1981) 167-173.

35. L.R. Radovic, P.L. Walker jr. and R.G. Jenkins, Fuel 62 (1983)
849.

36. H. Freund, Fuel, submitted.

 356

37. J.T. Sears, H.S. Muralidhara and C.Y. Wen, Ind. Eng. Chem.
Process. Des. & Dev. ~ (1980) 358.

38. L.E.B. Soledade, O.P. Mahajan and P.L. Walker jr., Fuel 57 (1971)
421.

39. F. Kapteijn, H. Porre and J.A. Moulijn, AIChE Meeting San

Francisco 1984, paper 144 h.

40. A. Tomita, O.P. Mahajan and P.L. Walker jr., ACS Fuel Div. 22
(1971) 4-6, and Fuel 56 (1977) 137-144.

41. K.J. Huttinger and W. Krauss, Chemie-Ingenieur-Technik 53 (1981)

122 and Fuel ~ (1982) 291-7, Fuel 60 (1981) 93-102.

42. M. Sakawa, Y. Sakurai and Y. Hara, Fuel ~ (1982) 717-20.

43. F. Kapteijn and J.A. Moulijn, Fuel 62 (1983) 221-5.

44. F. Huhn, J. Klein and H. Juntgen, Fuel 62 (1983) 196-9.

45. K.J. Huttinger and P. Schleicher, Fuel 60 (1981) 1005-12.

46. P. Leonhardt, A. Sulimma, K.B. van Beek and H. Juntgen, Fuel 62

(1983) 200-4.

47. H. Juntgen and K.H. van Heek, ACS Div. Duel Chem. Prep. 29 (1984)
195-205.

48. T. Wigmans, A. Hoogland, P. Tromp and J.A. Moulijn, Carbon 21

(1983) 13-22.

49. F. Shadman and D.A. Sams, Prep. ACS Div. Fuel Chem. 29 (1984)
160-4.

50. R.T. Hamilton, D.A. Sams and F. Shadman, Fuel 63 (1984) 1008-12.

51. K.S.Colle,K. Kim and A. Wold, Fuel 62 (1983) 155-7.

52. K.J. Huttinger, Fuel 62 (1983) 166-9.

53. A. Tomita, A. Kimura, Y. Ohtsuka and Y. Tamai, Carbon 21 (1983)

225-9.

54. L. Kuhn and H. Plogmann, Fuel 62 (1983) 205-8.

55. D.A. Sams, T. Talveridan and F. Shadman, AIChE Meeting San

Francisco 1984, paper 144e.

56. A. Lowe, Carbon 12 (1974) 335.

 357

57. S. Ergun and M. Mentser, in: "Chemistry and Physics of Carbon".

(ed. P.L. Walker jr.), vol. 1 (1965) 203.

58. N.M. Laurendeau, Progr. Energy and Combust. Sci. 4 (1978) 221.

59. A.E. Reif, J. Phys. Chern. 56 (1952) 778.

60. S. Ergun, J. Phys. Chern. 60 (1956) 480.

61. J. Gadsby, F. Long, P. Sleightholm and K.W. Sykes, Proc. Roy.

Soc. (London) A 193 (1948) 357.

62. F. Kapteijn, PhD Thesis University of Amsterdam, 1980.

63. S.W. Benson, "Thermochemical Kinetics", John Wiley, New York

1976.

64. B.J. Muhlen, K.B. van Beek and B. Juntgen, Proc. Int. Coal
Science Conf., Dusseldorf 1981, 197-202.

65. J.F. Strange and P.L. Walker jr., ,Carbon 14 (1976) 345-50.

66. B. Freund, Fuel, accepted for publication.

67. P. Biloen, J. Mol. Catal. ~ (1983) 17-24.

68. F. Bonner and J. Turkevich, J. Am. Chern. Soc. 73 (1951) 561.

69. F. Brown, Trans. Far. Soc. 58 (1951) 1005-14.

70. A.A. Orning and E. Sterling, J. Phys. Chern. 58 (1954) 1044-7.

71. M.B. Rao, F.J. Vastola and P.L. Walker jr., Carbon 21 (1983)
401-7.

72. M. Mentser and S. Ergun, Carbon i (1967) 331.

73. R.B. Cerfontain, F. Kapteijn and J.A. Moulijn, Proc. Vlllth Int.
Congr. on Catalysis, Berlin (1984), vol. III, 593-603.

74. J.D. Blackwood, D.J. McCarthy, Austr. J. Chern. 19 (1966) 797-

813.

75. M. Mentser and S. Ergun, US Bur. Mines Bull. No. 644 (1973) p. 42.

76. S. Ergun, US Bur. Mines Bull. No. 598 (1962) p. 38.

77. J.T. Shaw, Proc. Int. Conf. Coal Science, Dusseldorf 1981, 209-
14.
358

78. F.J. Long and K.W. Sykes, Proc. Roy. Soc. (London) A 193 (1948)
p. 377.

79. J.R. Cao and M.H. Back, Carbon 20 (1982) 505-12.

80. N.R. Marcilio, J.L. Castellan, J.L.F. Monteiro and M. Schmal,

Proc. Int. Coal Sci. Conf. Dusseldorf, 1981, 203-8.

81. R.T. Yang and C. Wong, J. Catal. ~ (1983) 245-51.

82. J.D. Blackwood and F. Mc Grory, Austr. J. Chern. l! (1958) 16-33.

83. G.P. Curran, C.E. Fink and F. Mc Grory, Prepr. Am. Chern. Soc.
Div. Fuel Chern. 12 (1968) 62-82.

84. H.J. Muhlen, A. Sulimma, K.H. van Heek and H. Juntgen, Proc.
Int. Coal Sci. Conf., Pittsburgh 1983, 554-7.

85. K.H. van Heek, H. Juntgen and W. Peters, J. Inst. Fuel 46 (1973)
249-58.

86. H. Juntgen, Fuel 62 (1983) 234-8.

87. H. Kubiak, H.J. Schroter, A. Sulimma and K.H. van Heek, Fuel
62 (1983) 242-5.

88. F. Kapteijn and J.A. Moulijn, J.C.S. Chern. Commun., 1984, 278-
9.

89. C.L. Spiro, D.W. McKee, P.G. Kosky, E.J. Lamby and
D.H. Maylotte, Fuel 62 (1983) 323-30.

90. K. Otto, L. Bartosiewicz and M. Shelef, Fuel 58 (1979) p. 565,

and Carbon 12 (1979) 351.
91. D.W. McKee, C.L. Spiro, P.G. Kosky and E.J. Lamby, Fuel 62
(1983) 217-20.

92. F. Kapteijn, G. Abbel and J.A. Moulijn, Fuel 63 (1984) 1036-42.

93. F. Kapteijn, G. Abbel and J.A. Moulijn, Carbon Conf. Lexington

U.S.A. 1985.

94. D.W. McKee, General Electric Report No. 81 CRD 056 (1981).

95. R.T.K. Baker and J.J. Chludzinsky, Carbon ~ (1981) 75-82.

96. J.L. Figueiredo, "Progress in Catalyst Deactivation", Proc.

NATOjASI, M. Nijhoff, Den Haag 1982.
 359

97. W.L. Holstein and M. Boudart, Fuel 62 (1983) 162-5.

98. A. Tomita, Y. Ohtsuka and Y. Tamai, Fuel 62 (1983) 150-4.

99. B.J. Wood and K.M. Sancier, Catal. Rev. Sci. Eng. 26 (1984)
233-79.

100. F. Kapteijn, o. Peer and J.A. Moulijn, to be published.

101. C.A. Mims and J.K. Pabst, Prepr. ACS Div. Fuel Chern. 25 (1980) 263.

102. M.B. Cerfontain, R. Meijer and J.A. Moulijn, Proc. Int. Coal
Science Conf., Sydney, in press.

103. C.A. Mims and J.K. Pabst, Fuel 62 (1983) 176-79.

104. W.K. Lewis, E.R. Gilliland and H. Hipkin, Ind. & Eng. Chern. 45
(1953) 1697-1703.

105. J.L. Falconer and J.A. Schwarz, Catal. Rev. Sci. Eng. 25 (1983)
141-227.

106. S. Kelemen and H. Freund, Carbon, submitted.

107. S. Kelemen and H. Freund, Prep. ACS Div. Fuel Chern., Miami
Beach 1985.

108. M. Boudart, D.E. Mears and M.A. Vannice, Ind. Chim. Belg.,
special issue, ~ (1967) p. 281.

109. M.A. Vannice, S.H. Hyum, B. Kalpacki and W.C. Liauh, J. Catal.
56 (1979) p. 358.

110. V.W. Weekman, AIChE-J. 20 (1974) p. 833.

111. W.L. Holstein, Fuel 62 (1983) 259-61.

112. T. Wigmans, H. van Cranenburgh, R. Elfring and J.A. Moulijn,

Carbon ~ (1983) 23-31.

113. F. Kapteijn, P. Coevert and J.A. Moulijn, J. Phys. E. Sci.

Instrum. 15 (1982) 1064-7.

114. J.R. Kittrell, in: "Advances in Chemical Engineering", vol. 8,

Acad. Press, New York (1970) p. 97.

115. F. James and M. Roos, Compo Phys. Comm. ~ (1975) p. 343.

116. J.A. Nelder and A. Mead, J. Comput. 7 (1965) p. 308.

360

117. N.R. Draper and H. Smith, "Applied Regression Analysis", John

Wiley, New York (1967).

118. G.F. Froment, Proe. 4th ISCRE, Heidelberg 1976, vol. 2, p. 421.
 361

SINGLE PARTICLE MODELS FOR FLUID-SOLID REACTIONS

Alirio E.Rodrigues

Department of Chemical Engineering

University of Porto,Porto,Portugal

INTRODUCTION
The paper is divided into four parts.In the first section we will
review some general concepts in reaction-diffusion problems.Mechanisms
of gas transport through porous media will be discussed in the second
part. Then models for fluid-solid reactions are presented and finally
applications to the combustion of coke deposits in a catalyst pellet
and modelling of moving bed gasifiers are addressed.

SOME GENERAL CONCEPTS IN REACTION-DIFFUSION PROBLEMS

Transport and reaction processes taking place in an isothermal
catalyst particle are shown in Figure 1.

Figure 1- Transport and reaction processes in a catalyst particle.

I-film diffusion of reactant A;2-pore diffusion of A;
3- adsorption of A;4-reaction;5-desorption of product B;
6- pore diffusion of B;7- film diffusion of B.
362

Two limiting situations can occur III

a) diffusion rate» reaction rate
In this case the concentration profile inside the pellet is al-
most flat and the effectiveness factor of the catalyst particle (de-
fined as the ratio between the observed rate and the intrinsic rate
of reaction calculated at the surface conditions) is around unity
b) diffusion rate «reaction rate
In this situation we can get steep concentration profiles and
the effectiveness factor is n <1.
The model parameter governing the behaviour of an isothermal
particle is:
reaction rate
¢ 2 = "77"-::-::----;---- (1)
diffusion rate
where ¢ is the Thiele modulus.It can also be shown that ¢2 is the
ratio between time constants for diffusion and chemical reaction,i.e.

(2)

The behaviour of nonisothermal catalyst particles in steady state

situations is governed by three parameters:the Thiele modulus ¢ ,the
Prater thermici ty factor j3 and the Arrhenius number y.
Concentration and temperature in a point inside the particle are
related by the Damkholer equation:
(-llH) D
e
T-T
s
= A (cAs-c) 0)
e
The maximum temperature inside the pellet will be:

T
max =T s (1+ B)
. (4)

With regard to pressure gradients inside particles some results

should be kept in mind 121:
a) Single reactions E v. A.=O
. ~ ~

For isobaric systems ~e have:

L: v. /}f." =0 (5)
i ~ ~

We should note that for multicomponent systems this is not equi-

valent to = v.=O.
. ~

On the ~other hand the mass conservation equation for binary

systems is:
(6)
 363

and the Graham's relation for isobaric diffusion is:

(7)

Obviously if there is a change in number of moles Eq.(6) can not be

verified under isobaric conditions.So there must be a pressure gra-
dient inside the particle.
b) A -+ vB
For this single reaction the pressure bounds for the particle
are:
and IV p s in the Knudsen regime

and p + ()lD /B )In y\) in the bulk diffusion regime

s e 0

c) Nonisothermal particles; ~v.;O

i ~
Pressure bounds are now:
and
and

Before concluding these introductory remarks let us consider

an example 13! which illustrates the limiting working regimes:
kinetic and diffusional regimes.
Two carbon cylinders are being burnt by oxygen.Cylinder dimen-
sions are D, ,HI and D2 ,H 2 respectively.Moreover D2=2D l .Figure 2
shows a plof of the ratio of the rates of weight losses for cylin-
ders 2 and 1,i.e.,W 2 /W l as a function of temperature.

I
i W2 !W!

t t"
8
j~~
.~
.,j6 Cd
f:J I\~
0;'
\\
\,'
6
," \
~~} 4
:j, ~2 ~\~

y
1 \\\
\ '\
"" 2
~.-"::

t ,.
" \
" .I I ~:>\~,
,
' \I

10 2 600 700 800

T

Figure 2- W2 /W l as a function of T
364

At low temperature the reaction takes place in the whole volume of

the cylinders so the kinetic regime prevails.The plateau is then:

At high temperature the diffusional regime prevails and the reaction

takes place almost at the cylinder surfaces;as a result

W2 /W I =4

MECHANISMS OF GAS TRANSPORT THROUGH POROUS MEDIA 12,41

The main mechanisms of gas tranport through porous media are
summarized in Figure 3. They are:
a) Knudsen flow : the gas density is so low that collisions be-
tween molecules can be ignored compared to collisions of molecules
with the pore walls.
b) Viscous flow(convective or bulk): molecule-molecule collisions
dominate over molecule-wall collisions;the gas acts as a continuum
driven by a pressure gradient
c) Continuum diffusion: species of a mixture move relative to
each other under concentration gradient,thermal gradient or external
forces (forced diffusion)
d) Surface diffusion

Transport properties of a multicomponent mixture are specified

by a number of coefficients.In order to evaluate these coefficients
we need flux models.Two main classes of flux models are available:
a) In class I the medium is viewed as a network of interconnec-
ted capi llaries
b) In class II the medium is considered as an assembly of sta-
tionary obstacles dispersed in the gas on a molecular scale (dusty
gas model).

Knudsei
IN. surf
Ordinary ~d~f .
I

t.
1.
total
Figure 3- Mechanisms of gas transport through porous media
 365
Models of class I are closely related to the physical structure of
the medium. Solutions exist for Knudsen regime (dcapillary « mean
free path length) and ordinary diffusion (d c » mean free path length).
For Knudsen regime in a capillary we have:

dp
r
(8)
dZ

where A is the area and the mean speed of molecules 1S 18RT/rrM .

r
For isothermal systems:
dc
r
Nr=-D r dZ (9)

and so we can calculate D :

r
D = rrA /8RT (10)
r 2p rrM
r
The porous medium is considered as a network of capillaries.Intro-
ducing a local average flux vector N we have:
-I'

N De grad c (11)
-r r r

and we get an effective quantity for the porous medium De defined

r
as:
De=K /8RT (12)
r 0 rrM
r

For ordinary diffusion at constant pressure we can write:

_._=
dP r
- z:
px x (v -v )
r s r s
(13)
dz D
S'fr rs

Taking into account that N =c v =cx v we get:

r r r r r

dp x N -x N
r s r r s
dZ - RT z: (14)
sFr D
rs
or
dx x N - x N
r s r r s
dZ = - z: c D
(15)
s'i'r rs
Equation (14) or (15) are the Stephan-Maxwell relationships.For bi-
366

nary mixtures the flux of species 1 is:

dX l
Nl = -cD 12 ~ + xlN (16)
and the total flux is N=N l +N 2 .
In porous media where d » mean free path we have:
pore
dp x N -x N
r s r r s (17)
-~
De
rs

Again we can define for the porous medium an effective quantity

as:

De =K D (18)
r8 1 T8

Fo!" viscous flow of a pure substance in a tube we have:

Bo p dp
N = -]JRT ~
(19)

For gas mixtures we can write

The total flux N is then the sum of the diffusive flux Nd and
v r. r
the viscous flux Nr ,~.e.,

N =N d + NV (21)
r r r

where Nd accounts for Knudsen and ordinary diffusion contributions.

r
Models of pore network type are very common. One example is the
Wakao and Smith model lsi . However ,we have to assume a smooth field
model to allow that PDEs relating composition and space position can
be written down.
The dusty gas model comes out from the idea of Maxwell (1860)
quoted here:"We may suppose the action of the porous material to be
similar to that of a number of particles fixed in space and obstruc-
ting the motion of the particles of the moving system". So,dust par-
ticles are treated as one component of mixture consisting of giant
molecules that are motionless in space. The advantages of this
approach are that the structure of the porous medium is isolated as
an independent problem and it is possible to obtain the general
solution of the diffusion and viscous flow problems without having
to say anything about the structure of the medium.
 367

The starting point is:

x N' -x N' x -w
s-r r-s p r r)grad
-L: RT grad x r +( p -
D RT
sFr rs
1
[c F -w L: c F 1 (22)
RT r-r r s-s-
s

and
N =N' - grad(lnT) (23)
-r -r

Then we apply these equations to a pseudo mixture of n+l species

when the n+l species is the dust.Moreover we should have c 1 uni-
n+
form in space. The dust should be at rest so N =0 and M ~ 00 (mas-
-n+l -n+l
sive dust particles).
After some manipulations we end up with diffusive flux relations:

n x N -x N N
s-r r-s 1
L: + -r grad Pr ( 24)
s=l De De RT
s#'r rs r

where De =
rs
- cc' D
rs
and De =D
r
lx'
r,n+l n+l
These effective quantities do not provide physical basis for es-
timation.However if dust concentration c' increases then c'lc in-
creases as well as De ;at the same time De decreases and we approach
. rs r
the Knu d sen reglme.
The unknown coefficients are obtained from experiment or by cal-
culation if the pore structure is known. The unknown coefficients are
K ,K l and B already introduced in Eqs.(12),(18) and (20).
o 0
It should be stressed that

De 0: d 0 Tl/2
K p
pore
De 0: dO . -1 T3/2
rs pore' P
0
B 0: d 2 ;p
0 pore

MODELS FOR FLUID-SOLID REACTIONS

A large number of models for fluid-solid reactions are availa-
ble 16,71 .Let us briefly review some of the most popular models for
reactions

A(g) + b B(s) ~ R(g) + S(s) (25)

368

and spherical particles of unchanging size.

Shrinking core model (SCM): the idea is shown in Figure 4.
The time for complete conversion and relations between conver-
sion,time and unreacted core radius are shown in Table 1 for diffe-
rent working regimes.

Table 1 - Relations between conversion, time and unreacted core

radius for particles of unchanging size: SCM

3
Film diffusion controls tit cf =~=l-(r
--B c IR p )

tCf=PBRp/3bkfcAb
2/3
Ash diffusion controls tit =1-3(1-X) +2(1-X)=
ca B B
2 3 2
=1-3(r /R ) +2(r /R )
c pcp ,. tcxRp

2
t ca =PBRp 16bD e cAb

1/3
Reaction controls tit cr =l-(l-x_)
-~
=l-(r c /Rp ) ., tcxRp

t cr =PBRp /bk s cAb

t - time for complete conversion of solid B

c
chemical reaction is first order with regard to reactant A

_ _ _...... time

ash

Figure 4- The shrinking core model (SCM)

 369

Uniform conversion model (UCM)-Figure 5:

(~ (/~~
\ '
'" // ------+ '- -
, -..----- time
\. ~.
.' .../
Figure 5 - Uniform conversion model

Grainy pellet model (GPM)-Figure 6:

." /0.00. . ,.( > grain

10°0°0 \
!O
\0 0 () )-i' pellet
\ ,
~
.'
.•.
/

Figure 6 - The grainy pellet model

Each grain behaves according to the SCM,Two characteristic times

have to be considered: T . and Td'f(time for complete conver-
sion of a particle if T g:a~B),Two lim1.E1.ng situations can be found:
gra1.n
a) Td"1. f «T gra1.n
"
The observed time for complete conversion is T =T " and
the GPM behaves as the UCM, obs gra1.n
b) Td"f
1. »T gra1.n
"
The observed time for complete conversion is T b =Td"f (pro-
portional to R2) ,The model behaves as the SCM in the O ~sh 1. diffusion
control regime~

Crackling core model (CCM)-Figure 7

Figure 7- The crackling core model

Here we have to define T as the time needed for the crackling

core
front to reach the center of the pellet; we also have to consider
370

T . • If T «T . the model behaves as the UCM;if T »

T gra : n the mg8~I isg~r~~lar to the SCM. core
gral.n
Reactions in nonporous solids: In the particular case of non porous
solids and reactions A(g)+bB(s)--?cC(g) shown in Figure 8 it is po-
ssible to derive relations between the extent of reaction or conver-
sion,time and the radius of particle.
Under chemical reaction control the rate of disappearance of the
gaseous reactant A is

r = k cAs (26)

and the rate of disappearance of the solid reactant B is

dr
c
br = - PB ~ (27)

Introducing dimensionless quantities:

r* r /s(V /A ) s-shape factor (28a)

c p p
and
t* = bkcA t/PBs(V /A )
s P p

Combining Eqs.(26) and (27) we simply get:

dr*/dt* = -1 , t*=O and r*=l (29)

Finally we obtain:

r* =1- t* t* < 1 (30a)

r* =0 t* > 1

The conversion is then:

xB = 1- r*s (31)

and the time for complete conversion is:

t =P s(V /A ) /bkc A (32)

crB p P s
. b etween tl.·me and conversion can be written:
Finally a re1atl.on
(33)
 371

For film mass transfer control Eq.(27) should be written as:

dr
-P B dt C = bkf(c Ab - cis) (34)
"t-
O
If we take into account that:

kfr c
1 + C (r /R )1/2 (35)
D c p

we obtain:

dr* 1 (1+Cr*)1/2
(36)
dt*
2 r*
with r,bl ,t*=O and where t i '=2bDcAb t/PBR2 . Solving Eq. (36) we get
the solution t*(r*).For a stagnant mediumP t*=l-r*.At low Re 2nolds
number the tiTe for complete conversion is proportional to R ;at
high Re , t c a::Rp .5 . P
Changes in particle size can also be accounted for by considering
that the actual particle radius Rt is:

Z R + (l-Z) r (37)
c
where Z is the ratio between the volume of products formed and the
volume of reactant.For ash diffusion control we have:

t { Z-IZ+(l-Z) (I-X) 12/3 _ (l_X)2/3}

--
t
3 (38)
ca Z-l

where
t(X=l) Z - Z2/3
3 (39)
t Z - 1
ca

Change in pore structure during gasification: Several models are

available to describe changes in pore structure during coal/carbon
gasification !8-l31 .Here we briefly recall the simple Petersen model
for particles with enlarging pores;it assumes uniform cylindrical
pores with random intersections.
The starting equation is:

dr
(40)
b
crt
372

where r is the pore radius,S is the surface area per unit volume
of the particle,£ is the pa~ticle porosity and PB is the molar den-
sity of the pore-Free solid.
From Eq.(40) we get:

S =d£ /dr (40a)

v p

For first order reactions

(41)

Moreover S =2nrL-Kr 2 where K depends on the number of intersec-

tions and angl~s. Finally one can obtain:
2 G-(rlr~
£
p
/£ po =(rlr) (42)
0 G-l

where G=3nL/Kr o .We can calculate G by taking into account that when
S + 0 , £ + l;then G is the root of the cubic equation
v p
4£
---E£ G3 - G + 1 =0 (43)
27

The solid conversion is related through porosity change through

£ -£
P po
x= 1 -£
(44)
po

>
time

Figure R- Shrinking of a non porous particle.

 373

APPLICATION TO THE COMBUSTION OF COKE DEPOSITS IN A CATALYST PELLET

AND MODELLING OF GASIFIERS.
Modelling of the combustion of coke deposits in a catalyst pellet
is an important step for modelling fixed bed regenerators.Coke can
be considered as CH with 1> n> 0.5 .Most of the existing models
conS1'd er a s1ng
. 1 e react10n:
n .

3+n
2CHn + -2- °2 -+ (45)

Sotirchos et al. 1141 developed a model which included film mass

and heat transfer.It also assumes that temperature is uniform inside
particles and that coke is uniformly deposited as a monomolecular
layer throughout the catalyst.Moreover they assume that products
of carbon oxidation are not constant with temperature and vary so
that the CO/C0 2 ratio is :

A = 2.5l2xlO 3 exp(-6240/T) (46)

The interesting aspect of this model is that coke was considered as

two solid phases: a graphitic carbon C and an aliphatic hydrocar-
bon phase CH 2 .Consequently two reac,tions have to be written:

C + a0 2-+ 2(1-a) CO + (2a -1) CO 2 (47)

1 1
a= 2" (1+ l+A)
and

-+ (48)

with rates
2
r l =5.0l4xlO exp(-37600/RT) c kgrnole of C/m2 s
r '" 0 r
2 1 ; 0 =5
The model was tested by using experimental data from Massoth 1151.
The dimensionless model equations are:

o Mass balance in a volume element of a particle

(49)

o Boundary conditions

-du
- = B1. ( u-u ) p=l (50)
dp m b
 374

du =0 p=o (51)
dp
o Heat balance for the particle

2
Le dv
d8
= L: ~. ¢: (1 l} 1
Y - vo J u(1-W.)p2 dp -Bi (v-v)
o h b
(52)
i=l 1 1 1

o Initial condition
0
v=v 8=0 (53)
o Balance for the solid phase

dW.1 1
(1- -)
e v u(1- W·) i=1,2
d"8 = 1 ¢:1 (54)
G.
1

w.=o
1
; 8=0 (55)

W2=1- (l-W )w
1
with

In the model equations p=r/R is the radial coordinate. 8=D t/R2

is the reduced time ,u=c/c f ~s the reduced gas phase concenErat~on,
v=T/Tf is the reduced temperature, y=E/RT f is the Arrhenius number,
~ =(-~R)D cf/A T is the Prater thermicity factor • ¢~=R2p?S.r~f/D c f •
e es 1 p111 e
Bi =kfE R ID is the mass Biot number, Bih=R h/~ is the thermal
m pp e p e
Biot number, Le=c D 13A is the Lewis number, W.=(p? -p.)/p? is
the solid convers~o~ and O.=M.cf/ p? . 1 1 1 1
1 1 1
The influence of several parameters :oxygen concentration,coke
weight fraction and Hlc ratio can be analysed through a pseudo stea-
dy state approach. The effects of these parameters are shown in Figu-
res 9 to 11.
The burning time can be also predicted by this model. The influen-
ce of the initial coke weight fraction and pellet size is shown in
Figures 12 and 13.
Application to the regeneration of a solid catalyst by coke bur-
ning has also been made by Bilbao et al 1161 and Hashimoto et al. 1171.
Another important aspect is the modelling of gasifiers. There
are several types of gasifiers summarized by Groeneveld 1181.Figure
14 shows four types of gasifiers: moving bed countercurrent. moving
bed cocurrent.fluidised bed and entrained suspension bed.Temperature
and fuel conversion profiles in these reactors are also shown.
 375

Pellet
temperature
x10- 2 (K) 11
Oxygen concentration
ub

1 - 2%
2 - 10%
3 - 15%

7·----~--~-------L _____
6.5 7.5

Figure 9- Influence of the oxygen concentration (ref 14)

11 6
Tb(K)
Tx10- 2 ---- 5
4 1- 650K
(K) 3 2- 660K
3 3- 690K
-2 4- 707K
--- 1 5- 730K
4 3 2
5 6- 780K
Wo x102
CK
Figure 10- Effect of coke weight fraction on the
initial PSS (ref 14)
376

12

10

7 -6·----~~----------~~1
-2 8.5
Tbx10 (K)

Figure 11 - Effect of H/c ratio on the pseudo steady

state (ref 14)
1- H/c =1; 2- H/c=o.S (u b =lS%)

----
(/J

r')' - ' 3.8. .6 8.4

I
o
r-<
1.4
~
Q)
S
'M
-1..1

......
......
o
e3.4
~
7.6
.D

0.02 0.08 0.01 0.07

Figure 12- Effect of the initial coke weight fraction on

the burn-off time.(ref 14)
1 - u b =lS%;T b =70SK 2- u b =lS%;T b =720K
3 - u b=2% ;Tb =760K 4- ub =10%;T b =760K
 377

,......
III
'-'
C"l
I
0 1.46 .• 47
roof 1.87
><
~
.,-1
.I.J
\!-l
\!-l
0
P
~
;j
,0

.,.,IN! 1. 79 2.43
<Xl

.1 .3 .5 .1 .3 .5
d (cm) d (cm)
p p
Figure 13- Effect of pellet size on the burn-off time (ref 14)
1- Tb =760K;u b =2% 2- Tb =760K;u b =10%
3- Tb =760K;u b =6% 4- Tb =760K;u h =8%

rroving tHrd moving tHrd fluldiSlld b!Id CHlttalned

countar-Clrrent co- current suspensiOn bed

1!500
Te K)

f900

~t-------~----~~=======r------~
75
fiRl
conwrslon
(Ofo)

t 25

top top
bottom top

Figure 14- Reactor types with corresponding temperature and

conversion profiles (ref.18).
378

Baker et all 19 1studied oxygen/steam gasification of wood in a fi-

xed bed gasifier.Coal combustion in fluidised beds was reviewed by
La Nauze 1201 and multiplicity patterns in such reactorswere analy-
sed by Congalidis and Georgakis 1211.
Modelling of Lurgi moving bed gasifiers was carried out by Yoon,
Wei and Denn i221 .However moving bed cocurrent gasifiers were exten-
sively,studied from modelling and experimental point of view by
Bliek 231 .The reactor can be divided in four zones as shown in
Figure lS.Bliek provides a complete model for the reactor using ex-
perimental data for coal gasification obtained either under chemical-
ly controlled conditions or diffusional controlled conditions.He
used the dusty gas model to write down flux equations.The model was
successful in predicting axial profiles of CO+H 2 ,fuel gas composition
as a function of time,carbon uptake by the gas phase,etc.

air

~
1

I
""""...,....,.,
1

I
r 11 d- drying zone

--f--\-\ :f; 1
p- pyrolysis zone (>2S0 C)
d
0- oxidation zone
p
o (1200-1600 C)
r- reduction zone

Figure IS - The moving bed cocurrent gasifier (ref 18)

 379

NOTATION
A - area
A.- chemical species
1
B - permeability
o
c - gas phase concentration
D - diffusivity
D - effective diffusivity
e
Dt- thermal diffusivity
r
F - external force acting per mole of species r
r
(-~H)-heat of reaction
k f - film mass transfer coefficient
k - kinetic constant for the chemical reaction
h - film heat transfer coefficient
Le- Lewis number
M.- molecular weight of species A.
N1 _ molar flux 1
p - pressure
R - ideal gas constant
r - pore radius;rate of reaction
R - particle radius
p
r - radius of the unreacted core
Sc_ surface area/unit volume
v
t - time
t - time for complete conversion of the solid
c
T - temperature
u - reduced gas concentration
x - molar fraction
v - velocity
X - conversion
z - axial coordinate
Z - ratio between the volume of products formed and the volume of
reactant.
Greek symbols
8 - Prater thermicity factor
y - Arrhenius number
v - stoichiometric coefficient
].l - viscosity
Ae - thermal conductivity of the particle
€ - particle porosity
p
~ - solid conversion
p - reduced radial coordinate for the pellet
PB- molar density of the so~id
380

¢ - Thiele modulus
T - time constant
e- reduced time (=D t/R2)
e p
Subscripts
a - ash diffusion
b - bulk conditions
cr- chemical reaction
d - diffusion
f - film'
s - surface conditions

REFERENCES
1. R.Aris .The Mathematical Theory of Diffusion and Reaction in
Permeable Catalysts. (Clarendon Press,Oxford,1975).
2. R.Jackson.Transport in Porous Catalysts. (Elsevier,New York,1977).
3. X.Duval. J.Chim.Phys.58(1961)5
4. Mason,E. and A.Malinauskas.Gas Transportation in Porous Media:
The Dusty Gas Model. (Elsevier,New York,1983)
5. Wakao,N. and J.Smith.Chem.Eng.Sci.17(1962)825
6. O.Levenspie1.Chemical Reaction Engineering(J.Wi1ey,New York,1972)
7. Szeke1y,J.,Evans,J.and H.Sohn.Gas Solid Reactions(Academic Press,
New York,1972)
8. E.Petersen. AIChEJ 3(1957)4
9. Hashimoto,K.and P.Si1veston.AIChEJ 19(1973)259
10. Bhatia,S. and D.Perlmutter.AIChEJ 26(1980)379
11. Bhatia,S. and D.Perlmutter.AIChEJ 27(1981)247
12. G.Gava1as.AIChEJ 26(1980)577
13. G.Simons.Prog.Energy Comb.Sci.9(1983)269
14. Sotirchos,S.,Mon,E. and N.Amundson.Chem.Eng.Sci.38(1983)55
15. F.Massoth.Ind.Eng.Chem.Proc.Des.Dev.6(1967)200
16. Bilbao,J,; ;Romero,A. and J . Arandes .Chem.Eng. Sci.38(1983) 1356
17. Hashimoto,K.,Takatani,K.,Iwasa,H.and T.Masuda.The Chem.Eng.J.
27(1983)177
18. M.Groeneve1d.The Cocurrent Moving Bed Gasifier.PhD Thesis.Univ.
Twente,The Nether1ans(1980)
19. E.Baker.Ind.Eng.Chem.Proc.Des.Dev.23(1984)725
20. R.La'Nauze.Chem.Eng.Res.Dev.63(1985)3
21. Congalidis,J. and C.Georgakis.Chem.Eng.Sci.36(1981)1529
22. Yoon,H.,Wei,J. and M.Denn.AIChEJ 24(1978)885
23. A.B1iek. Mathematical Modelling of a Cocurrent Fixed Bed Coal
Gasifier.PhD Thesis,Univ.Twente,The Netherlands(1984)
PART V. COAL GASIFICATION
 383

GENERAL ASPECTS AND ENGINEERING PRINCIPLES FOR TECHNICAL APPLICA-

TION OF COAL GASIFICATION

K. H. van Heek
Bergbau-Forschung GmbH, Essen, F. R. of Germany

1. INTRODUCTION

1.1 Motives for coal gasification

Coal will have to play an important role in future energy con-

cepts of the industrial nations. This becomes obvious from the fact
that coal contributes only 34 % to the world-wide consumption of
fossil energy sources although its reserves account for approxi-
mately 83 % of all primary energy sources. Considerable efforts
must be made, however, to utilize coal as a substitute of oil and
gas so that the supply of coal products will be able to meet to-
day's standards in energy utilization. Furthermore, high levels
of environmental protection must be considered in the conversion
of the primary energy "coal".

Among the potential of the different coal conversion proces-

ses, the generation of gas offers special advantages due to the
high efficiency of the process, comfortable handling of the pro-
duct gas, easy and relatively cheap distribution on the basis of
present infrastructure by the distribution of natural gas, envi-
ronmentally acceptable combustion as well as various fields of app-
lication in the energy and raw materials sector.

Thus among the tasks the energy engineer is faced with, impro-
vement of existing and development of new processes for the gene-
ration of gas from coal are of outstanding importance. This task
is not a recent one. Corresponding technologies were developed al-
ready in the past and are applied today for coal refining on a
world-wide scale.
384

1.2 Basic processes and reactions

Two different approaches - as mentioned in table 1 - have to

be considered to produce gas from coal:

Table 1 - Fundamental processes for gas production from coal

Oevolatilization (pyrolysis. thermal decomposition) :

cool + heat (400 -12000 C) ----. coke (char)

+ liquids (tar!
+ gases

Gasification:

coal + gasifying agent (steam) + heat (700-1400 0 C)

~ gases (H 2 • CO. 0000) + minerals (ash)

- In the case of generation by devolatilisation, the gas which

evolves from the coal upon heating to temperatures above 400 DC
is employed. Coke occurs as a byproduct and is used separately.
- In the gasification of coal the organic component of the coal
is completely converted to gas, and so only ash remains. This
is achieved by subjecting the coal to a reaction with a gasify-
ing agent, such as steam, at a temperature above 700 DC.

This lecture is concerned exclusively with the gasification

of coal. However, devolatilisation or pyrolysis always occurs, when
coal is heated up. Thus it is a pre-reaction to or it is occuring
simultaneously with gasification.

The fundamental gasification reactions are indicated in table 2.

The main reaction is the heterogeneous water gas reaction (1),

in which the carbon reacts with steam to form hydrogen and carbon
monoxide. The products of the reaction, in turn, can participate
in two further reactions: the homogeneous water gas reaction, or
shift-reaction, (2) and that of methanation (3). They thereby react
either with the gasifying agent or with one another, thereby ge-
nerating carbon dioxide and methane. The subsequent reactions (2)
and (3) are exothermic; however, the endothermic main reaction (1)
 385

is dominant. Hence the overall reaction, which is the sum of the

three reactions included under steam gasification, is endothermic.

Table 2 - Fundamental reactions for coal gasification

--
Steam Gasification: ~H
kJ/mol
(1) C+H 2 O H2 +CO + 118,5

--
(2) CO+H2 0 H2 +C0 2 - 42,3
(3) CO+3H 2 CH 4 +H2 0 - 206,0
Hydrogasification :

(4) C +2H2
Combustion:
- CH 4 87,5

(5) C+ 02 CO 2 - 406,0
(6) C+ ~02 - CO - 123,0

lIH: "." endothermic = heat consuming

exothermic = heat producing

To produce methane from coal as a natural gas substitute, the

hydrogasification (4), i.e. direct reaction of hydrogen with car-
bon, is of outstanding importance. Finally, the reactions with
oxygen (5) an~ (6) must also be considered in gasification proces-
ses, both serving as heat-generating processes in connection with
endothermic steam gasification, and leading to CO at incomplete
combustion (6) or "partial oxidation".

2. POTENTIAL OF COAL GASIFICATION

Figure 1 gives a schematic view on processes and potential app-

lications connected with coal gasification by steam as applied to-
day or as achievable by today's technology.

By use of steam and air, the nitrogen simultaneously brought

into the process results in a strong dilution of the product gas
so that the final product will be a lean gas with a calorific va-
lue in the range of 4000 to 5000 kJ/m 3 , which has to be used with-
out long transport as an industrial fuel for heating purposes. Ex-
amples for this field of application are metallurgical plants of
metal-working industries, glass works, brick works, lime mills and
similar branches of industry, which used appropriate gas generators
in earlier times also in Germany. Today, this technology is applied
in countries where natural gas is not available or cannot be di-
 386

stribu ted due to lackin g gas supply system s. Espec ially

numerous
plants of this kind are operat ed in develo ping countr ies.
Wester n Europe , too, the constr uction of such plants has Yet in
again be-
come the subjec t of consid eratio n.

Coal

Heat (allothermal)
Steam Gasification
Oz}
or (autothe rmai)
Air

Synthesis gas SNG

Town gas

Heating Electricity NH3 Crude iron Heating

Methanol
Petrol

Fig. 1 - Proces s and applic ation of steam gasifi cation

The use of lean gas in power statio ns makes it possib le

move in an interm ediary purifi cation stage all pollut ants, to re-
and
especi ally sulphu r, origin ally presen t in coal, thus allowi
ng a
more enviro nment ally accept able electr icity genera tion from
coal.
Furthe rmore, the conver sion of coal in gas offers the possib
of using gas turbin es, which in turn entail s a consid erable ility
crease in efficie ncy. For this reason , this techno logy is in-
studie d
in variou s countr ies as a subjec t of resear ch and develo
pment pro-
jects.

Gasifi cation using pure oxygen yields an undilu ted gas which
basica lly consis ts of hydrog en, carbon monoxide, methan e
and carbon
dioxid e. By purifi cation and conver sion of this gas it is
possib le
to obtain synthe sis gas for the produc tion of ammonia, methan
and petrol . At presen t, most coal gasifi cation plants are ol
operat ed
to this purpos e. A furthe r field of applic ation of the produc
the direct reduct ion of ores. This techno logy is in the t is
first place
of intere st to countr ies produc ing coal with poor coking
ties, as it can be used for reduct ion and smelti ng of ironproper -
are via
coal gasifi cation . Finall y, it is possib le to conver t the
CO and
H2, produc ed by autoth ermal gasifi cation , to methan e by
cataly tic
reacti on accord ing to equati on (3) and to distri bute this
methan e
as substi tute natura l gas or town gas via the gas supply
system .
 387

3. ENGINEERING ASPECTS OF COAL GASIFICATION

3.1 General

Taking in mind the many applications of coal gasification, it

is already quite understandable that practice has created a large
number of different processes. Moreover, the tasks which have to
be performed for a successful gasification can be solved in dif-
ferent ways. Among these are
- the contact between gasifying agent and coal,
- the means by which heat is generated and transferred to the reac-
tor,
- nature and pressure of the gasifying agent,
- feeding of the coal and removal of the ash.

Moreover coal gasification in general is completely unsensitive

to the nature of the solid feedstock and is in its different varia-
tions employed for conversion of the whole range of solid fuel from
anthracite over high volatile and subbituminous coal to lignite
and even peat and wood. Thus coal properties can play different
roles in the separate processes. This leads in the end to the re-
quirement that a gasification process has to be chosen which best
suits the properties of a particular coal and has been also the
reason for the development of the various processes.

3.2 Types of gas generators

The main reactor types used are listed in table 3. The moving
bed reactor is operated in counter current flow of coal and gasi-
fication agent. In the bed a temperature profile is formed, so that
the coal subsequently is preheated, dried, devolatilized, gasified,
and finally combusted. This reactor is characterised by a high den-
sity of solids. Moreover, no back-mixing of fresh coal and partly
gasified char takes place, so that the highest possible burn-off
can be reached.

In fluidized bed reactors an uniform temperature distribution

and a high rate of back-mixing of the different burn-off stages
occurs. The latter is responsible for the relatively low burn-off
of the coal in fluidized bed reactors.

In entrained phase reactors the coal usually is transported

cocurrently together with the gasificiation agent. As in this case
the coal has to react in short residence times, until now these
reactor types are operated for partial oxidation. In the reactor
very high temperatures and temperature peaks are formed. By tur-
bulence a high rate of back-mixing of the solid material occurs.
The negative consequences of this on the conversion rate are how-
ever partly compensated by high temperatures, so that overall a
388

moderate up to a high burn-off can be reached.

Table 3 - Reactor Types for Coal Gasification

MOVING BED FLUIDIZED BED ENTRAINED PHASE

COUNTERCURRENT FLOW UNIFORM TEMPERATURE COCURRENT FLOW

DISTRIBUTION
TEMPERATURE PROFILE HIGH RATE OF BACKMIXING VERY HIGH TEMPERATURES (PEAKS)
FORMATION OF ZONES MODERATE BURN-OFF HIGH RATE OF BACKMIXING
- REHEATING AND DRYING MODERATE UNTIL HIGH RATE OF
- DEVOLATILIZATION BURN-OFF
- GASIFICATION
- COMBUST Iorl
HIGH DENSITY OF SOLIDS
NO BACKMIXING
HIGH BURN-OFF

The possibilities of heat transfer to the gasifier are shown

in figuroe 2.

Heat

autothermal allothermal aUothenool

with with combustion with nuclear
oxygen chamber reactor
Fig. 2 - Mode of heat supply to reactor for coal gasification
by steam

In autothermal coal gasification the reaction heat is genera-

ted in the gasifier itself by using as gasifying agent a mixture
 389

of oxygen and steam, whereby a certain amount of coal is burnt and

the remainder reacts with steam. In other processes heat is gene-
rated outside the gasifier by burning either coal or its products
in a separate combustion chamber with air. This heat can be trans-
ferred to the gasifier either by superheated steam, steam/gas mix-
tures or by solid heat carriers. As the reaction heat is not ge-
nerated in the gasifier itself these processes are called allother-
mal. A special allothermal process which after development in the
future shall take advantage of a high temperature nuclear reactor
as a heat source is shown on the right hand side. This concept sa-
ves coal since none is burnt for heat generation.

Principles of autothermic reactors are shown in figure 3, which

simultaneously explains the way of gas solids contact in the three
"classical" reactor types in more detail.

fixed bed fluidized bed entrained bed

product gas

coal dust
• oxygen
• steam

OIygen liquid slag

.steam

Fig. 3 - Types of reactors for autothermic coal gasification

In a fixed-bed or moving-bed gasifier a bed of coal particles

is moving downwards in a countercurrent flow of oxygen and steam.
In order to attain reasonable gasification rates the gas veloci-
ty must be high and the pressure drop low. This implies limitations
on the feedstock coal as will be discussed later. The product gas
leaving the gas generator at the top contains pyrolysis products
which are formed in the upper layer of the fixed bed. Most proces-
ses use a dry ash removal, thus the reaction temperatures must be
below the fusion point of the ash. The best known process of this
kind is the Lurgi which operates at pressure up to 35 bar and is
used on a commercial scale.
390

In a fluidized bed gas generator for coal gasification the ve-

locity of the gasifying agents, oxygen and steam, is just high
enough to suspend coal particles. Coal can be continuously fed
whilst ash is withdrawn. Fluidized bed technology was applied to
coal gasification nearly 60 years ago by Winkler. The process has
been operated on an industrial scale too, and is especially suited
for lignite and highly reactive hard coal.

Using entrained bed reactors the feed must be coal dust a~d
high linear velocities of oxygen and steam are employed. This re-
sults in temperatures of about 1500 DC and correspondingly high
reaction rates. At these temperatures liquid slag is formed and
is withdrawn at the bottom of the reactor. To ensure the high ra-
tes for the burn-off desired the amount of oxygen has to be high,
so that reaction (6) in table 2 is dominating. Therefore this mode
of operation is named as "partial oxidation" also. The product gas
does not contain any tar. The best known entrained bed reactor is
the Koppers-Totzek gasifier, which is used especially to make syn-
thesis gas for production of ammonia and fertilizers.

Figure 4 illustrates three different possibilities of allother-

mic reactors. At the left hand side the gasifying agent steam, which
can be mixed with other gases, e.g. product gases, has been super-
heated so high that its heat content can cover the heat required
by the reaction. Based on this idea industrial processes were de-
veloped and used in the past. Their inherent disadvantage is that
due to the great excess of superheated gases, much energy is lost.

gaseoos heat cmrier solid heat carrier heating surfaces

product gllS

ash ash

superheClleII steam steam

Fig. 4 - Types of reactors for allothermal coal gasification

 391

In the gasifier of the middle of the figure the reaction heat

is fed via solid heat carriers. These may be inert material or char
which are heated externally up to very high temperatures and lose
their excess heat within the gasifier where it is used as reaction
heat. However, in this case the amount of solid heat carriers cir-
culated is about twenty times higher than the amount of coal gasi-
fied. Because of the disadvantages mentioned, allothermic gasifi-
cation is no longer in use on industrial scale.

The possibility of heat transfer from nuclear reactors to the

gasifier, is shown in the right hand side of figure 4. Tubes through
which hot gases are flowing are immersed in a fluidized bed of coal
and steam. The technology of this so-called "immersion heater pro-
cess" is being developed for the gasification of coal using nuclear
heat from high temperature reactors. This development is currently
being carried out in the Federal Republic of Germany. The heat is
produced in the helium cooled core of a nuclear reactor at high
temperature and transferred to the gasifier via an intermediate
circuit of helium. Within the gasifier helium flows through the
immersion heater and its sensible heat replaces all the coal which
has to be burnt in other processes to generate the reaction heat
with consistent savings in terms of fuel.

3.3 Elements of a gasification plant

Figure 5 shows the arrangement of the elements of a plant for

gasification of coal: The raw coal after preparation is introduced
by a feeding system into the gasification reactor. The coal mine-
rals, which remain as ash or slaq are discharged and used for pur-
poses such as the construction of buildings or roads, or disposed.
The solids in the raw gas are either treated like the ash or - espe-
cially, if they contain considerable amounts of un-gasified car-
bon - recycled into the gasifier. Coal also serves as fuel in a
power station for raising of electricity and steam. In the case
of an authothermic process, as shown in the example, gasification
is performed with steam and oxygen, the latter being produced by
a cryogenic air separation process. The raw gas is dedusted cooled
down and finally purified and thus converted into a feedstock for
following synthesis or other utilisation. Thereby especially sul-
fur compounds can be completely removed. The latter are converted
into elemental sulphur, which is sold or may be disposed without
any problems for the environment. Gas purification and processing
consists of a high number of washers and other chemical apparatus.
The heat normally is used for steam raising. The waste water is
treated before it is disposed or recycled.
392

Coal
Depot

Milling
Drying
Screening

Coal Gas Product

Conditioning Gas
I
I I
I I
I
I I
I I
I I
I I
" I I I
, A' I I I
" , '-______________________________
Ir Steam lI ________ -': I
I
I
---------------------------------------------~
Waste Water

Fig. 5 - Arrangement of a coal gasification plant

Figure 5 shows that the gasifier itsel~ the technique of

which has been discussed in the previous chapter, of course is the
heart of a gasification plant. However, its other parts contribute
a considerable amount to the total effort. If the investments are
concerned for a special plant using moving bed technique the sha-
res of the single elements are as follows:
coal preparation 8 %
gasification 20 %
- gas processing 40 %
supply of energy 23 %
rest 9 ~~.

From this list it can be seen that the processes for gas pu-
rification and conditioning for the following synthesis processes
are of high importance. Therefore a survey shall be given on this
part of a gasification plant in the following chapter.

3.4 Gas purification and conditioning

Due to the various gasification processes and the different

types of feedstock gasified, the raw gas as produced in the ga-
sifier has different qualities. This is both the fact for the main
constituents, which are desired for the adjacent utilization,and
for the byproducts, which can be more or less disturbing in the
down stream steps of the process. The gas processing therefore has
the task to reach a quality which meets the requirement for gas
utilisation.
 393

As shown in the previous chapters the raw gas contains among

the main components CO, H2, C02 and CH4 undecomposed steam and so-
lids in form of carbonaceous dust. In moving bed technology tar
has to be expected in the raw gas, of course dependent on the rank
of the coal. Moreover the raw gas contains nitrogen components like
ammonia, hydrogen cyanide and nitrous oxide as well as sulphur com-
ponents like H2S and organic sulphur components. All these consti-
tuents are disturbing the synthesis process and have to be remo-
ved to a low concentration as will be shown later. Also the pres-
sure of the raw gas is important for the purification processes.

Gas Gas
Process Utilisation
components processing steps

Fuel gas
Steam gasification N2 ,CO,C02 ,H 2 Removal of tar,
-Industrial furnaces
using air CH 4 , H2 S dust and S
-Power stations

Remowl of tar, - Reducing gas for

Steam gasification dust and S iron ore reduction
H2 , CO, CO 2 - synthesiS gas _~
using oxygen production of NH3 ,
CH 4 , H2S + Shift conversion methanol + gasoline
(or allothermic) - TownQCls
• MetnC nation - ProduCtion Of ~Nl;

Removal of tar,
Hydrogasification CH 4 , H2 , CO
dust and S SNG
(in future) CO 2 ,H 2 S
• Gas separation

Table 4 - Gas treatment dependent on gasification processes

and final utilization

Utilization of the gas requires different qualities as it is

summarized in table 4. If the gas is just burned e.g. in a combi-
ned cycle process, then the removal of possible air pollutants at
high temperature is of importance. Gases for transport in pipe-
line have to be free of corrosive and poisoning material and re-
quire a standardization concerning their combustion conditions.
Special efforts are needed for chemical syntheses concerning the
correct relationship between the CO-, C02- and H2-concentrations
for the different product desired. Also the influence of the trace
products with respect to poisoning of catalysts has to be taken
into account. The requirements for feed gases for some syntheses
are listed in table 5.
394

Table 5 - Requirements for feed gases for different synthesis

processes

Synthesis for Ammonia Methanol Methane Gasoline

pressure (bar) 300-700 50-300 40-100 20

CO 0,001 30 25 0 27
H2 75 66 75 55
CO 2 > in 0'
,0 2 ppm < 0,8 -0 15
CH 4 1 < 1,5 any 1
N2 25 <2 - 2
~
H2S u.org.S mg/m~ <2 <2 <2 <2
NH3 " - - ? -

Two examples for typical combinations of possible processes

for purification and conditioning shall be discussed: Figure 6
deals with the processing of raw gas containing tar, which is pro-
duced under pressure of 30 bar. Dust and condensable liquids are
removed by cooling and condensation of a part of the un decomposed
steam, whereby a mixture of solids, tar and water is achieved, which
goes to the water make-up. The gas after cooling and condensation
contains still a lot of steam, so that it can be fed into a shift
conversion unit, the catalyst of which is working at high tempe-
ratures and is unaffected by the sulphur. In this step also a par-
tial hydrocracking of the higher hydrocarbons is possible. After
that the gas is again washed with water, whereby the remaining li-
quid hydrocarbons, nitric compounds and part of the sulphur com-
pounds are separated. Further down stream the gas is washed to se-
parate the H2S and the C02 and is obtained in a high pure quality
for synthesis.

The example as shown in figure 7 deals with the conditioning

of raw gas, which contains no tar and is obtained under atmosphe-
ric pressure. In the first step the raw gas is dedusted, which can
be done for instance with a cyclone at high temperature. Adjacently
it is cooled indirectly under production of steam. Also it is pos-
sible that the gas is firstly cooled and after that dedusted by
electric precipitators. After cooling, the gas is washed with water,
whereby the remaining dust, the ammonia and other nitrous compounds
are separated. After that it normally is compressed for special
gas washing processes and gas processing steps, which of course
 395

Claus
plant

Steam

raw gas Water H2S CO2 Synthesis

shift wash removal

Water
make -up

~t Tar

Figure 6 - Purification and conditioning of raw gas from coal

gasification containing tar

are oriented to the final utilization. In the scheme shown in

figure 7 the gases are firstly desulphurized, then they go through
a shift conversion and through a C02 wash to get the correct ratio
of H2 and CO for the synthesis. Finally also the CH4 can be sepa-
rated in a cryogenic process.

Both the flow sheets in figure 6 and 7 show a Claus plant. The
process is based on a catalytic conversion of the H2S to elemental
sulphur.

For the removal of H2S, organic sulphur compounds and C02 from
the gasification gas there exists a broad variety of processes a
survey of which is shown in table 6. Physical ad- and absorption,
condensation in liquid nitrogen and chemical absorption are app-
lied.

Also the catalytic conversion of CO in the shift reaction pro-

cess has been mentioned.
 396

Claus
plant

Water

Steam Compression
H2S
(C021

Row gas Water HZS C02

Cooling Shift
without tar wash lIIIIOYal removal

Solids -..er Steam

Wat@r Cryogenic S,nthesis

make-up

Dust Ammonia

Fig. 7 - Purification and conditioning of a raw gas from coal

gasification without tar

4. ROLE OF COAL PROPERTIES IN DIFFERENT GASIFICATION PROCESSES

A selection of a certain coal gasification process requires

a thorough analysis of the situation at the plant site, of the pro-
cess scheme itself, of the products desired and last not least of
the available coal. Coal properties which are applied today in coal
technology and science and measured by standard methods are those
which mainly have a bearing on production of coke from coal. For
this technology it is possible to describe the behaviour of the
coal during the process and the properties of the products - mainly
the coke - taking a relatively small set of parameters. In the case
of coking this is possible because the production of metallurgi-
cal coke today is based on one reactor, the horizontal coke-oven,
and is limited to a relatively small range of coal rank. The si-
tuation for coal liquefaction by direct hydrogenation seems to be
similar: It is also based mainly on one process - the Bergius-Pier-
process - and also on a narrow range of high volatile coal and lig-
 397

Table 6 - Processes for purification and conditioning of raw

gas to synthesis gas

f>rocess subsl:ances removed washinq aqent/catalyst name of process

~dsorpt1on traces of H2S+S org activated carbon

CO 2 + traces of H2O mOlecular sieves
CO 2+ traces of H2O silica
bhysical CO 2 + H2S water under pressure
dsorption
CO 2 + H2 S + COS dlmethvlether of poly- Selexol
• ethylene glycol
CO 2 + H2S + S ~rg. methanol Rectisol
CO~ + H2 S N-methylpyrrolidone Purisol
CO 2 ,,. H2S ~ropylene carbonate Fluor Solvent
t:onden-
sa~ion
'1 ~o + Ar + CH 4 liquid nitroqen

=hemical CO 2 + H2S ethanola,mine (mono-, Girbitol

bsorptioll di-, tri-, )
CO~ + H2S + COS diisopropanolamine Shell-halp
H2S dimethylglycine Alkazid OrR
CO 2 + 11 2S N-mcthylamine Alkazid M
CO 2 + H2S potassium carbonate
CO 2 + H2S .;. COS potassium carbonate B.:!nfield
with addi.tive
H2!": sodium thioar.senate Giammarco-
Vetrocoke
CO 2 + 11 2S + COS tetrahydrothiophene- Shell Sulfinol
diocide + diiso!lro-
panolamine
11 2S sodium carbonate/ Stretford
anthraquinone
CO copper formate/
copper chloride
'traces of CO 2 lammonia water
traces of CO 2 sodium hydroxide
S-compounds zinc oxide
HiS bog iron oretluxmasse
~atalytic CO Fe-/Cu-Zn-catalyst

i
~onvcrsion
CO; CO 2 Ni-catalyst
org. S-compounds Co-Mo-catalyst

nite.

With regard to coal gasification it is much more difficult to

describe a complete picture. As said before, this is due to the
fact that there are various gasification processes differing from
each other for reasons as discussed. Of course there are coal pro-
perties which have the same meaning for all processes and others,
398

which have to be taken into account only for special cases.

Coal properties as listed in table 7 have in like manner im-

portance for all gasification processes. The ultimate analysis of
the coal and also the analysis of the mineral components are impor-
tant for the overall mass balance and for the estimation of the
side products and the products with environmental consequences.
The water content of the coal affects mainly the mass and the heat
balance. The calorific value of the solid feed fuel is needed for
the calculation of the energy balance. Chemical structure, maceral
composition, and pore structure of the feed materials influence
coal properties such as reactivity and caking behaviour.

Table 7 - Significance of Selected Coal Properties for all Ga-

sification Processes

PROPERTY: SIGNIFICANCE FOR:

ULTIMATE ANALYSIS MASS BALANCE
SIDE PRODUCTS
COMPONENTS WITH ENVIRONMENTAL
ASPECTS
WATER CONTEKt MASS BALAWCE. HEAT.BALANCE
ASH CONTENT HEAT CONSUMPTION
tATAlH1C EFFECT
~NGINEERING ASPECTS

CALORIFIC VALUE HEAT BALANCE

CHEMICAL STRUCTURE INDIRECTLY FOR REACTIVITY AND
CAKING BEHAVIOR
MACERAL COMPOSITION AS CHEMICAL STRUCTURE
PORE STRUCTURE INDIRECTLY FOR REACTIVITY
i

The other coal properties have for the gasification processes

and reactor types just discussed different consequences as shown
in table 8. In this it is distinguished between the steam gasifi-
cation (I) in moving bed and fluidized bed reactors, the partial
oxidation (II) in entrained phase reactors, and the hydrogasifi-
cation (III) in fluidized bed and entrained phase reactors.

For the autothermic steam gasification in a moving bed reac-

tor the volatile matter content has great importance as in the de-
volatilization zone the pyrolysis products, e.g. mainly tar, higher
gaseous hydrocarbons and CH4 can be formed. The tar products have
to be taken into account as side products, as discussed above un-
 399

der the aspects of further processing, but also concerning envi-

ronmental considerations. In each case, methane is an acceptable
product, if SNG is desired.

Table 8 - Significance of Coal Properties for Special Gasifi-

cation Properties

GASIFICATION PROCESS COAL PROPERTIES

Gasification Reactor Typ Volatile Matter Ash Particle Size Grindibil ity Caking Behavior Reactivity
Agent Hard Coal

Moving Bed Primary CH4 fr,om Heat Requirement 5 - 80 '!Ill Formation of Finelnc Feeding of Gasification
Devolatil ;zat;oo Mel ting Point no Fine Gralns Grains During t:'!Strongly Caking Throughput
(Ash Removal) Mining and Trans- Coals
port
I
CatalYtic In-
fluence

Fluidized Bed little Signifi- Heat Requirement < 5nm Production of Feeding of Attainable
canee Narrow Particle Strongly Caking Burn-off
Meltino Point Narrow Distri- Size Distribu- Coals with Jet-
. (high) bution tion Feeder Possible
Catalytic In-
fluence
Degree of Inter-
growth (Dust
Formation)

II. Oxygen/ Entrained N.S. Hea t Requ i rement Fines Enerqy Consump- N.S. little Signifi-
(H 20) (hioh) < 0.1 nm tion by Gri ndi ng cance
~h~S~400 DC
Mel ting Point
(Partial
Oxidation) (low)
Corrosion

II 1. Hydrogen/ Fluidized Bed CH4 by Hydroge- Similar to I Fines as I and II Enhancement by Gasification
nation of Vola- H2' Oecaking Throughput and
(H20) :~~rained tiles lfigher Possible in En- Attainable Con-
Reactivity of trained Phase version Rate
~ha~e 1000 0c young Coals by Hydrogena t i on

N.S. = No Significance

The mineral matter, in the table named "ash", is primarily a

ballast material. It determines part of the heat demand. Further-
more the melting point of the ash is important, as the level of
this temperature determines the kind of ash removal in the gas ge-
nerator. For a coal with a very low melting point, e.g. between
1100 and 1200 DC, a slagging gasifier will be taken into account
from which the ash can be removed as a fluid. At higher ash mel-
ting points, for example between 1400 and 1500 °C~ a dry ash re-
moval is more advantageous. Finally, the ash can be active as a
catalyst on gasification reactions.

The method of operation of moving bed reactors requires a remo-

val of the fraction of fine particles from the feed coal to gua-
rantee an undisturbed gas flow through the height of the bed. There-
fore, for this reactor type such coals are ideal which give,in
mining, transports and also in subsequent processing, a minimum of
fines. An undisturbed uniform gas flow is also prevented, if strong
caking coals cause reactor plugging. Finally, the reactivity of
the coal determines the gasification capacity.
400

In fluidized bed reactors the volatile matter components are

of minor importance. Because of the good mass transfer in the flui-
dized bed, the volatiles are mixed with the solid particles and
with the gasification agent and almost completely converted by ga-
sification into the gasification products.

On the other hand the ash content influences the heat demand.
The meltinq point should be as high as possible, because removal
of the ash is only technically feasible in the form of dry particles.
Furthermore, the catalytic influence of the ash components has to
be taken into account. Of special importance for the operation of
the fluidized bed is the kind of intergrowth between coal and mi-
neral matter: when the coal particle and the mineral matter are
completely separate the mineral particles can be discharged at the
bottom because of their higher density. More complex is, however,
the behavior of the mineral part of the coal which is in tight in-
tergrowth with the organic matter content. As in the temperature
region of the fluidized bed between 800 and 1000 °C the gasifica-
tion with steam is controlled by the chemical reaction and by pore
diffusion, and therefore also takes place in the inner pores of the
coal particle, with increasing burn-off of the organic part fi-
nally a selection of ash is left over. This residue is destroyed
in the fluidized bed and can therefore lead to dust. Thus, for flui-
dized bed processes some attention has to be drawn to the inter-
growth between mineral matter content and organic part of the coal.

Generally, into fluidized bed reactors particles with a size

below 5 mm are fed, whereby because of its desired smooth opera-
tion and for limiting the rate of carry-over of fines, a narrow
distribution of the particle size is wanted. A specially suitable
coal would be that which would lead after grinding to the highest
possible amount of equal particle size.

Also, feeding of strong caking coals into a fluidized bed is

possible, if a suitable feeding system - for example a jet-feeder -
guarantees that the fresh feed coal is mixed intensively and fast
with the already partly gasified char particles, so that the caking
coal particles cannot form agglomerates. The reactivity determines
primarily the highest possible burn-off.

During partial oxidation in the entrained phase reactor the

reaction with oxygen at high temperatures occurs so intensively,
that the volatile matter content of the coal has no further impor-
tance. They will be rather completely cracked, whereby together
with the gasification progress the gas composition is controlled
by thermodynamics.

In this process the ash acts mainly as ballast material, be-

cause of the high heat demand which is necessary for heating up,
 401

and its conversion corresponding to the high process temperatures.

For the fluid slag removal the melting point should be as low as
possible. Furthermore,it has to be investigated if the ash contains
components which could lead to corrosion in the reactor insulation.

Entrained phase reactors can be operated only with fine ground

dust, whereby the coal preferably has to be ground to a particle
size below 0,1 mm. Therefore the grindibility gains more importance
as it determines the energy consumption during crushing. Because
of the aggressive reaction conditions in partial oxidation,also
a strong caking hard coal can be processed without special problems.
A high chemical reactivity is also of lower importance.

With regard to the points discussed above, the gasification

with hydrogen in fluidized bed or entrained phase reactors shows
no new aspects concerning ash, particle size, and grindibility.
Certainly, in this case the volatile matter components are of great
importance, because a great part of the primary methane is formed
by hydrogenating the pyrolysis products, and a higher reactivity
of the high volatile coals to hydrogen is found. Concerning the
caking behavior, firstly it has to be taken into account that by
the influence of the hydrogen the preference for caking is increa-
sed. Because of hydrogenation the formation of liquid products is
increased, which favours agglomeration. One possibility is to rea-
lize in the reactor such conditions which render exclusively the
hydrogenation to methane. Such kind of hydrogenation seems possible
in an entrained phase reactor. The reactivity of the different coals
in relation to hydrogen finally influences the gasification capacity
and the attainable throughput.

5. CHOICE OF LITERATURE

1. Juntgen, H. and K. H. van Heek, Coal gasification (in German)

Thiemig Taschenbucher, 94, Munchen 1981 (especially chapters 1,
2, 4 and 5 of the book)
2. Peters, W., H. Juntgen and G. Kolling, Coals as an oil and
gas replacement, Proceedings of lOth World Petroleium Congress,
Vol. 3, London 1980
3. Juntgen, H., J. Klein, K. Knoblauch, H. J. Schroter and
J. Schulze, Conversion of Coal and Gases Produced from Coal into
Fuels, Chemicals and other Products, Chemistry of Coal Utilization,
2nd Supplementary Volume. John Wiley and Sons, Inc., New York, 1981
4. Schilling, H.D., B. Bonn and U. Krauss, Coal Gasification:
Existing Processes and New Developments, Graham & Trotman, London,
1981
5. Coal gasification technology, Status report prepared by a
joint EEC-India study group, Comission of the European Communities,
Brussels, 1980 - XII/1450/80-EN
 403

TECHNICAL STATUS AND NEW DEVELOPMENTS OF COAL GASIFICATION PROCESSES

K. H. van Heek

Bergbau-Forschung GmbH, Essen, F. R. of Germany

1. HISTORICAL REVIEW OF COAL GASIFICATION AND SCOPE OF THE LECTURE

The Italian priest and scientist Felice Fontana is regarded

to be the father of coal gasi fication. In 1780 he noted: "i f one
quenches glowing coal with distilled water, one obtains ignitable
air". Two or three years later similar results have been found by
the French chemist Antoine Lavoisier. These two scientists descri-
bed for the first time the conversion of carbonaceons material in-
to hydrogen and carbon monoxide by the heterogeneous steam-carbon
reaction, which forms the basis for gasification of coal.

This invention was used within a rather short time, whereby

several periods can be distinguished up to now: In the first period
starting in 1830 the gasification of coke had been performed. It
was reacted with steam in retorts heated from outside, whereby the
heat necessary for the reaction had to be transferred through the
walls. Thereafter, in a second period starting at about 1870 dis-
continuous processes had been applied, whereby in a first stage
a bed of coke was treated with air. Thereby a part of coke was bur-
ned and the other part was heated up. In a second step steam was
applied to this hot coke whereby its sensible heat was used for
the supply of the heat necessary for the performance of the hete-
rogeneous steam carbon reaction. Todays gasification technology
and thereby the third period of the technical development and in-
dustrial application of coal gasification started in 1930, when
the first continuously operating process using mixtures of oxygen
and steam was developed. This had been the fluidized bed gas gene-
rator invented by Winkler. Moving bed technology for gasification
started in about 1935, whereas the entrained phase process was de-
 404

veloped after the war and applied in the early fifties. Finally,
after 1973 an intensive development for the improvement of gasi-
fication techniques and the finding of new possibilities is under-
way.

The lecture will firstly describe the three main gasification

processes in use today. Afterwards the new developments will be
reviewed covering examples for the improvement of the existing tech-
nology and such developments as hydrogasification and gasification
using nuclear heat. Also some remarks will be made with regards
to the potential of gasification in the field of coal based energy
supply.

2. PRESENT STATUS OF COAL GASIFICATION

2.1 Description of existing processes

The coal gasification plants currently used worldwide in in-

dustry dispose of gas generators, as shown in figure 1. These pro-
cesses are exclusively of the autothermal type, coal being gasi-
fied with steam and oxygen.

LurgL Winkler K~rs - Totzek

coal
t

jJlJlftl
raw gas

II ~
gas
slean
.0, J;l;0
f:h)\~,
:/'~'\
j ~/:J,\-~
) \..
L
coal-~~
slag
t
.0,
steam
.0, ~

ash

Fixed bed Fluidized bed Entrained bed

Fig. 1 - Types of Industrial Gasifiers

 405

In the Lurgi gas generator, gasification is carried out in a

moving bed characterized by the counterflow of coal and gasifying
agent. This leads to the formation of a temperature profile and
various zones in the gasifier in which coal is dried and devolati-
lized and finally gasified and burned with increasing temperature.
The process is carried out at pressures in the range of 25 - 35 bar.
For this reason, coal is fed into and the ash drawn off the gas
generator via lock hopper systems. A stirrer is arranged at the
top of the gasifier to prevent agglomerations by caking of the coal.
The gas generator is a watercooled pressure vessel with diameters
of 3 - 4 m.

The fluidized bed process, named after its inventor, Winkler,

was already developed for industrial coal gasification on a large
scale during the 1920's. Coal is thereby gasified with steam and
oxygen in a fluidized bed in a cylindrical gas generator whose lower
end is tapered. The coal is introduced near the middle of the bed
by means of a screw-feeder. The gasifying agent is introduced both
below and above the fluidized bed. The latter measure ensures that
the entrained particles of coal are subsequently gasified. This
results in more thorough burn-off of the carbon and thorough reduc-
tion of the byproducts at the high temperature which consequently
prevails above the fluidized bed. The Winkler process has hitherto
been carried out only at atmospheric pressure.

In the entrained bed process according to Koppers-Totzek, fine

coal is gasified in entrained phase with oxygen and steam in the
reactor. Because of the brief time of contact, very high tempera-
tures must be attained by the addition of relatively large quanti-
ties of oxygen, and thus a high rate of carbon conversion is achie-
ved. The ash occurs in molten form and is removed as slag from the
bottom of the gas generator. A waste-heat boiler for recovering
process heat is situated immediatley above the gas generator.

The three processes currently employed for coal gasification

are compared in table 1. The most important data for the gas gene-
rators have already been presented in the description. As far as
the coal used is concerned, it should be remarked that the Lurgi
and Winkler processes are appropriate for at most moderately caking
coal, whereas these properties are of no consequence for the Kop-
pers-Totzek process. Moreover, fine-grained coal is incompatible
with the fixed bed process; hence only coarse coal can be used here.
The fine coal must therefore be diverted to other purposes, such
as firing in a power plant for furnishing electrical energy and
steam for the coal gasification. Nevertheless, it should be emphasi-
zed that further developmental efforts are aimed at improving the
input range of coal types.
 406

Table 1 - Comparative data for industrially applied processes

of coal gasification

Gasg~ Lurgl Winkler !!!~~ers -Totzek

Type fixed bed fluidized bed entrained phase
Temperature ac 700-1000 1000 1300
Pressure bar 35
Capacity m3gas/h 35-50000 17-20000 20000 (2 units)
50000 (4 units)

Coal:
Kind brown and brown and all kinds
hard coal hard coal of coal
moderately moderately
caking coal coking cool
Particle Size mm 6-40 0-8 0,1

Crude Gas:
"10 per Vol. H2 37-39 35-46 31
CO 20-23 30-40 58
CO2 27-30 13-25 10
CH4 10 -12 1-2 0,1

With regard to the gas analysis it should be pointed out that

a ~igh content of methane is evident in the crude gas from the fixed
bed process; this is due to the separate zone of devolatilisation,
on the one hand, and to the higher pressure, on the other hand.
Hence the Lurgi process is particularly well suited for generating
gas with a high heating value, and was originally developed for
manufacturing town gas. The crude gas from the Winkler and Koppers-
Totzek processes is characterized by a very low content of methane.
It is therefore well suited for synthesis.

2.2 Fields of application and examples for gasification plants

Table 2 sums up the present state of the application of coal

gasification processes. Quite naturally, it is difficult to present
a complete list of all applications, e.g. including gasification
plants in Eastern Europe or China. A further problem arises from
the fact that some plants may already be shut down by today for
age or economic reasons.
 407

One of the main fields of application of coal gasification con-

sists in the generation of ammonia to produce fertilizer. To this
end, some 3.8 x 106 tce/year are currently used in various coun-
tries. Most plants are based on hard coal. Koppers-Totzek is the
most frequently used process. In South Africa, the Lurgi-process
is applied for hard-coal gasification mainly to produce synthesis
gas which subsequently is converted into petrol. A plant with 17
Lurgi-gasifiers has been operating since the beginning of the 60's,
annually gasifying 2,5 x 10 6 t of hard coal to produce 0,27 x 106 t
of petrol by means of the Fischer-Tropsch synthesis. At the same
time, this plant yields chemical raw materials, town gas and other

Table 2 - Examples for current application of coal gasification

Ammonia:
process plants gasifier coal 106 tcely NH3 tid country

Lurgi 2 5 hard coal 0.15 230 Pakistan, Korea

Yugoslavia, India,
Winkler 3 6 brown coal 0.35 530
Turkey

Koppers- 6 14 brown coal 0.69 1 030 Zambia, Greece, South

Totzek 6 14 hard coal 2.6 3 920 Africa, India etc.

Total 17 39 3.8 5 710

Petrol:
Petrol
t/y
106

South Africa,
Lurgi 3 95 hard coal 16.6 3.13
Sasol I, II and III

Town gas I SNG:

Gas
109 m3 /y

USA, Great plains

Lurgi 1 14 lignite 4.2 1.2 (SNG)
(since 1984)

Fixed bed 8* 4.2 GDR

4 86 lignite
(Lurgi) (town gas) CSSR

*estimated
 408

products. In a second plant with double capacity, which was com-

missioned in 1980, further 7 x 10 6 t of hard coal are gasified per
year while the resulting gases are further processed mainly to pet-
rol. A third one of equal size has been commissioned in 1982. Fixed
bed gasification technology is applied for the production of town
gas especially in the GDR and the CSSR on the basis of lignite since
many years and recently in the USA for the production of CH4 as
substitute natural gas (SNG).

3. FURTHER DEVELOPMENT OF GASIFICATION PROCESSES

3.1 General

In view of the future tasks of coal gasification, intensive

work is being carried out in the Federal Republic of Germany, in
Great Britain, in the USA and in other countries mainly with the
following objectives:
Further development of conventional processes in view of an enlar-
ged coal basis, increased effinciency and, in case of entrained
and fluidized bed gasification, in view of an operation at higher
pressures.
- Generation of methane as substitute natural gas by application
of hydrogasification.
- Coal conservation by use of heat from ~igh-lemperature nuclear
reactors (HTR).
- Application of gasification for electricity generation.
- Application of catalysts for gasification.
- Underground gasification.
A detailed description of all development objectives and of
the corresponding research projects would go beyond the scope of
this lecture. As examples firstly the developments shall be revie-
wed mainly dealing with the improvement of conventional technology.
After that hydrogasification and nuclear aided gasification will
be discussed. A broader review is given in the literature as menti-
oned at the end of the text. A tabulated compilation of the pre-
sent international status has been published by Brecht and Hoffmann
1983.

3.2 Development of conventional gasifiers

Table 3 shows a list of the pilot plants in the F. R. of Ger-

many. Column 1 identifies the projects in short and names the com-
panies carrying out the individual projects. It can be seen that
the coal producers make an essential contribution to the develop-
ment. With the exception of the nuclear aided hydrogasification
and the Winkler-process, all activities are performing in view of
using hard-coal. As to the characterization of the gas generators~
it can be noticed that two fixed-bed processes, three fluidized-
 409

bed processes and four entrained phase processes are being develo-
ped. Most of the plants listed in the table have already terminated
operation. The projects were supported by public fund, except for
the Shell-Koppers-process.
Table 3 - Pilot plants for coal gasification in the F. R. of
Germany
.. _.. - . ~--------

Project/Operator character istic data location/ throughput

operation t/h

Ruhr 100 fixed bed, Dorsten 4,5-10

Ruhr gas/Ruhrkohle/St eag 100 bar, 1000-1300 K 1979-83

High-temperature Winkler fluidized bed Frechen 1

Rheinische Braunkohlenwerke 10 bar, 1370 K 1978-84

Shell-process entrained phase Hamburg 6

Shell 30 bar, 1750-2200 K 1979-82

Texaco-process entrained phase Oberhausen 6

Ruhrchemie/Ruhrkoh Ie 40 bar, 1550-1750 K 1978-82 I)

Saarbe rg-Ot to-process

Saarbergwerke/Dr. c. Otto

KGN-process
slag bath
25-30 bar, 1750-1950 K

fixed bed,
Volklingen
1979-82

Hlickelhoven
10

1-2
l
Kohlegas Nordrhe in 1-6 bar, 1220-1420 K 1979-84 \
_-1
vrw- c a a1
converSlon entral.ned phase, partlal Werne 10 1
Vereinigte Elektrizitiitswerke air gasification, 1984-87 ~
10 bar, 1650 K
r-Hu-m-b-o-ld-t---p-ro-c-e-s-s---------g-·a-s-i-f-ic~a-t-i-on-i-n-a-b-a-th--- --sw-e-d-e-n----·"--------
I
10

.,:--l
Kliickner-Humboldt-Deutz of iron, 3 bar 2)

CGT -Combi gasi fication combined, reactor types HUnxe 4

Carbon Gas Technology 15 bar 1983-
Nuclear aided gasi fication: .
Hydrogasification fluidized bed, with H2, Wesseling 10
Rheinische Braunkohlenwerke 120 bar, 1150 K 1983-86

J
steam gasi fication fluidized bed Essen 0,2
Bergbau-Forschung 40 bar, 1100 K 1976-84

1) now in operation for gasification of hydrogenation residues

2) under construction

A few processes are described more closely in the following:

Figure 2 shows the gas generator for the fixed-bed gasification
under a pressure up to 100 bar which has been under develpment by
Ruhrgas, Ruhrkohle and Steag with the aim to improve production
of SNG from coal. The gas generator operates essentially according
to the conventional Lurgi principle. The most obvious modification,
410

however, is the use of two separate gas exits. Thus clear gas can
be drawn from the outlet above the zone of gasification, whereas
the gas from low-temperature devolatilization can be drawn from
the upper outlet. The latter contains the impurities from the pro-
cess of pyrolysis; these comprise essentially the tars and can be
more easily separated from the smaller quantity of carbonization
gas and subsequently processed, than from the total gas output,
as in the conventional Lurgi process. A specially designed agitator
should also permit the use of German coal with a strong tendency
towards agglomerating in the gas generator. The pilot plant has been
constructed in Dorsten and has commenced operation in the autumn
of 1979.

Two other developments of the Lurgi gasification reactor, which

are in progress outsid of the German projects, should also be men-
tioned:

Steam

(~rate

'Water

Steam ..
Oxygen

Fig. 2 - Gas generator for gasification of coal in a fixed bed

under a pressure of up to 100 bar - Lurgi Ruhr 100
type
 411

In the past years a fixed-bed generator operating according

to the "slagging" principle, has been tested in the former high-
pressure gasification plant in Westfield, Scotland, by British Gas.
A cross section of the gasifier is shown in Figure 3. By means of
an increased input of oxygen such a high temperature is thereby
attained in the lower section of the gas generator that the ash
occurs as a molten slag and can therefore be tapped.

A substantial increase in turnover is expected from this reac-

tor. A gas generator of his type is planned for a demonstration
plant in the United States. The second development is an increase
in the diameter of the reactor to 5 m; this measure would result
in a gas output up to 100 000 m3/h. A first prototype is being in-
stalled at Sasol, South Africa.

Coal
~
Coal Lock Hopper

Gas Quench Coal Distributor/

Stirrer

Refractory Lining
Water Jacket
Gas Outlet
Steam/Oxygen
It"'" Supplies

Circulating
Quench Water

Fig. 3 - "Slagging gasi fier" for fixed bed gasi fication of coal

On the basis of the conventional Winkler process Rheinische

Braunkohlenwerke have developed the fluidized bed gasification tech-
nique for German lignite, applying higher pressures and some higher
bed temperatures. Thereby mainly the throughput has been improved.
412

As the conventional one, this gasifier is mainly for production

of synthesis gas. Also the Shell process (it started under the name
Shell-Koppers-process) is based on a conventional gasifier, the
Koppers Totzek one. The aim of the development mainly was the ap-
plication of elevated pressure for the conversion of coal into syn-
thesis gas using an entrained phase gasifier.

The scheme of the Texaco coal gasification process is illustra-

ted in Figure 4. The coal is first pulverized and a suspension with
water is prepared. The slurry is then injected into the gas gene-
rator, which is under pressure and where the coal is gasified in
an entrained phase with oxygen and the steam arising from the sus-
pension. The reaction proceeds at temperatures above 1300 DC. The
molten slag which thereby forms falls into a water bath and is sub-
sequently removed through a lock system. The crude gas emerges from
the reactor at a high temperature; the heat is first recovered by
steam generation in a waste-heat boiler; the gas is then purified
prior to further processing and utilization. The pilot plant has
been constructed on the premises of the Ruhrchemie in Oberhausen-
Holten and has been in operation since the end of 1977.

Quench Cooler

Fig. 4 - Fluidized coal-dust gasification under pressure, accor-

ding to the Texaco process

The Saarberg-Otto-development was based on the former Rummel

gasifier, which applied a bath of liquid slag for the production
of synthesis gas by coal gasification. The aim of the project was
to perform the process unter elevated pressure. The KGN-process
mainly has been investigated to develop a fixed bed gasifier, which
 413

can be used in smaller units to produce fuel gas from coal. The
VEW gasification project aims at the improvement of coal based elec-
tricity generation in a combined cycle process. Coal is devolati-
lized and partially gasified in air at high temperatures. The hot
char produced is fed into the combustion chamber for steam raising
and the gas, after cleaning, is used in a gas turbine. Based on
a 1 t/h test plant operated in the 70ties at atmospheric pressure
a big pilot plant is erected.

A fairly new approach is being tested in the Humboldt process,

whereby coal shall be converted into a synthesis gas in a bath of
liquid iron. The project's philosophy is that the carbon is reac-
ted in a solution in the liquid iron and that the minerals as well
as the impurities are trapped in the slag. Finally, in the CGT
process the possibility of the combination of different reactor
types shall be tested. The projects mentioned under "nuclear ai-
ded gasification" will be discussed in the next chapters.

As to the future prospects, it should be stated that most pro-

cesses would still have to undergo the demonstration stage before
they can be operating on a commercial scale. For the time being
there are difficulties to continue the development, mainly for the
reason that coal gasification cannot compete economically with oil
or natural gas in Germany. Nevertheless some further steps have
already been done. The Rheinische Braunkohlewerke AG, is commis-
sioning the first line of a commercial plant based on the "Hochtem-
perature Winkler". It shall convert 72 t/h of raw brown coal into
synthesis gas for the production of 14 t of methanol. Also Ruhrkohle
and Ruhrchemie have announced the construction of the "Synthesegas-
Anlage Ruhr" on the basis of the Texaco process jointly developed
by these two companies. Starting in 1986 some 250000 t/a of hard
coal shall be converted into gas for the oxosynthesis.

3.3 Application of hydrogasification for producing methane

Endeavours to produce methane as a substitute for natural gas

(SNG) have already been in progress in the United States since the
beginning of the 1960's. Since that time many proposals and ideas
have been submitted and have reached various stages of development,
ranging from the laboratory phase all the way to large pilot plants.
For this purpose the indirect approach of steam gasification follo-
wed by catalytic methanation is one alternative used.

The applicational possibilities of gasification by direct hydro-

genation are illustrated in Figure 5. In the two versions employed,
which differ in the method of furnishing the necessary hydrogen,
the coal is first converted predominantly to methane in an exother-
mic reaction with hydrogen. The char which is thereby produced is
subjected to steam gasification, in order to produce the required
 414

hydrogen, as depicted on the left-hand side of the figure. This

combination of steam gasification with hydrogasification offers
the following advantage over steam gasification with subsequent
methanation: Only half as much coal must be converted in the strong-
ly endothermic steam gasification as with the latter combination.

SNG SNG

heating

Fig. 5 - Hydrogasification of coal combined with steam gasifi-

cation of char or with steam reforming of methane

In the version depicted on the right-hand side of Figure 5 the

required hydrogen is obtained by feeding only a part of the pro-
duced methane to the supply network and allowing the remainder to
react with steam. The hydrogen thus generated is supplied for hydro-
gasification along with the converted carbon monoxide. Because of
the diminishing reactivity of the coal towards hydrogen with pro-
gressing gasification, the process yields residual char, for which
further uses, such as combustion or gasification, must be found.
According to this basic scheme a process has been proposed for utili-
zing heat from a nuclear reactor for gasifying coal. Heat from high-
temperature nuclear reactors is thereby employed for steam reforming
methane.

With the combination of steam gasification and hydro-gasifi-

cation shown on the left-hand side of the figure, a few projects
have been tested at the pilot plant level in the United States.
Of these, as an example the Hygas process is explained in more de-
tail: Figure 6 shows longitudinal section through the gas generator
used. Four different zones of reaction are situated one above the
other in a tubular gas generator with countercurrent flow: The solid
matter thereby moves downward, whereas the reactive gas moves upward.
Coal is first introduced as a coal-oil suspension into a dry zone
and subsequently gasified with hydrogen, first in an entrained phase
and then in a fluidized bed; it is thus a two-stage operation. The
 415

crude gas

coal-oil
suspension -..*==~
oil
precipitation

--{
tSt:stage
'hYdrogenating
'gasification
----t- 675°C
semicoke __ 0
2nd stage
hydrogenating
gasification

residual --f9550(
coke

steam +
oxygen

Fig. 6 - Gas generator for the HYGAS process

residual char finally reaches a second fluidized bed at the bottom,

where it is gasified autothermically with steam and oxygen to gene-
rate hydrogen. A pilot plant with a coal throughput of 3 t/h has
been in operation for a number of years since 1972.

These and other gasification projects in the United states are

to a certain extent in a state of crisis. On the one hand, serious
technical difficulties have become evident during the operation of
the pilot installations. On the other hand, economic studies have
not yet indicated whether or not the use of the processes offers
advantages in comparison to conventional methods. Hence no predic-
tions can yet be made as to whether or which of the new processes
will be developed to technical maturity in demonstration installa-
tions.

3.4 Gasification of coal with the use of heat from nuclear reactors

In all conventional processes the coal serves two purposes:

On the one hand it constitutes the raw material for generating the
gas. On the other hand it furnishes the energy required for dri-
ving the process, that is, the heat of reaction and the heat for
 416

raising the temperature of the coal and steam, as well as all other
process heat necessary for operating a gasification plant. In prac-
tice this means that 30 to 40 % of the coal must be burned in order
to convert the rest into gas. With the use of heat from high-tem-
perature nuclear reactors (HTR), however, the energy requirement
of a gasification plant can be satisfied in such a way that the
coal can be totally converted to gas. The advantages of such an
approach include the conservation of the coal reserves and better
utilization of the available coal, reduction of emissions specific
to coal since less coal is burned, improved utilization of nuclear
energy by coupling of the heat into a chemical reaction, and lower
costs of manufacturing gas from coal.

With the aim of developing the corresponding processes all the

way to the stage of large-scale technical application, companies
of the nuclear reactor and coal industries are cooperating on the
PNP-project CPrototypanlage Nukleare Proze8warme C ~ Prototype Plant,
Nuclear Process Heat)). The objective is to appropriately design
and construct the high-temperature nuclear reactor, heat exchangers
and associated gasification processes. On the one hand, the combined
processes of gasification by hydrogen and methane reforming induced
by heat from the nuclear reactor, as discussed in conjunction with
Figure 5, are thereby contemplated.Un the other hand, the operation
with steam gasification as endothermic process directly driven by
the heat from the nuclear reactor is likewise feasible. The latter
alternative is illustrated in the following:

Figure 7 depicts the coupling of a high-temperature nuclear

reactor with steam gasification. The high-temperature nuclear reac-
tor is of the type developed in Germany with spherical fuel ele-
ments situated in a bed through which helium flows. The latter is
thus heated to a temperature! up to 950°C. For reasons of safety
the heat is then transferred by way of a heat exchanger to a" secon-
dary circuit through which helium also is circulated.

This in turn flows through a heat exchanger in the gas gene-

rator, whose heat-exchange tubes are immersed into a fluidized bed
of coal and steam, thus transferring the heat to the endothermic
reaction. At a lower level of temperature the heat of the helium
in the intermediate circuit is subsequently employed to generate
steam, which is supplied both to the gasification process and to
a power plant which furnishes electrical energy. The raw gas which
is generated is converted in the conventional manner and utilized.
Gasification in a fluidized bed under a pressure up to 40 bar and
with the application of heat from a helium circuit at a tempera-
ture between 900 and 1000 °C at entry has been conducted for the
first time in a semitechnical facility at Bergbau-Forschung. The
gas generator is a representative section from the large-scale
 417

1 HTR CO 2 • H2S
2 heat exchanger
3 steam generator
4 gas generator
90°C
rrh-Product
5 gas cleaning He

950°C

3000C •••
••
•
4-
electricity
He steam

Fig. 7 - Coupling of high-temperature nuclear reactor with

steam gasification

technical gas generator, and is designed in such a way that the

height of the fluidized bed, as well as the length and arrangement
of the heat-exchange tubes correspond to those of the large-scale
concept. The fluidized bed itself has a cross-sectional area of
about 1 m2 and can be operated up to a height of 4 m. The installa-
tion has been in operation from 1976 to 1984. As a next step a
pilot plant at a bigger scale - some tons of coal per hour - is
under consideration.

4. CONCLUSION AND OUTLOOK

Commercial gasification processes are available and used world-

wide for the production of heating gas, fertilizer, gasoline and
chemical products from coal. For new plants a construction period
of 4 to 5 years has to be taken into account. Most of the discussed
development projects, however, would still have to undergo the de-
monstration stage before they can be operated on a commercial scale.
Nuclear gasification will finally reach large-scale technical appli-
cation only towards the end of the 90's.

Table 4 tries to assess qualitatively gasification processes

for coal with respect to the gas which can be produced, the quality
requirements of the feedstock as well as the suppression of by-
products.
 418

Table 4 - Evaluation of Gasification processes taking as ex-

amples German developments
Kind of gas Low CAL.
fuel gas
Synthesis
gas
Med.CAL.
fuel gas
SNG Tolerance
to coal
Suppression
of
I
H* 4.0 11.5 18.0 38.0 by-products
Fixed Bed
Lurgi (dry ash) + + ++ ++ - -
Ruhr 100 - - ++ ++ !. +
KGN - + ++ !. !. +

Fluidized Bed
Winkler ++ ++ !. !. - +
High Temp. Winkler !. ++ !. !. - +
HTR, steam-gasifi-
cation - ++ + + + +

HTR,hydrogasifi-
cation - !. !. ++ + +

Entrained p:hase
Koppers-Totzek - ++ !. !. ++ ++
Shell-Koppers + ++ !. !. ++ ++
Texaco - ++ !. !. ++ ++
Saar-Otto + ++ !. !. ++ ++
VEW

*
++

H= appro cal. value in MJ/ml

- - -
- not suitable
+
I -
± possible
+ suitable
++ specially developed for

It grows evident from this table that the fixed-bed pressure

gasifiers working in counterflow are predominantly suited for the
production of gases with high methane contents. Fluidized bed and
entrained phase processes are particularly suited for the generation
of both synthesis gas and heating gas as it can find broad applica-
tion in chemical and other industrial plants. The gases resulting
from these processes are generally also well suited for reduction
purposes in metallurgy. Entrained phase processes are not limited
regarding the potential fields of application of various types of
coal since the fast conversion of the fine-grained coal prevents
it from caking together and getting deposited. This is not so with
fluidized-bed processes, so that in most cases only moderately caking
coal can be used. However special feeding systems are under develop-
ment, with which also strong caking coals can be fed into gasifiers.
On the whole coal gasification processes will be available, which
can meet the requirements of the product market.

Moreover, it can be stated that coal gasification is, to a

great extent, a nonpolluting process. From the experiences with
existing plants it may be concluded that gasification exhibits the
following advantages compared with other processes of coal conver-
sion:
 419

Gasification has high over-all efficiency and in this way emis-

sions related to the end-energy produced are small. Thereby the
use of nuclear energy for gasification processes has high poten-
tial in comparison with autothermal processes.
- Sulphur and particulate matter are completely removed from the
product gas, therefore coal gas used as town gas or SNG can be
burnt practically without S02 and dust emission.
- A technique is available by which all off-gas streams of the gasi-
fication plant may be cleaned effectively, especially waste water
treatment is highly developed.
According to the present state of knowledge, trace elements in
the ash are no problem for the environment because they are scar-
cely soluble in water.

To sum up, coal gasification is ready to playa decisive role

in ensuring a sufficient and environmentally acceptable supply of
energy in the future.

5. CHOICE OF LITERATURE

1. Juntgen, H. and K. H. van Heek, Coal gasification (in Ger-

man) Thiemig Taschenbucher 94, Munchen 1981 (especially chapters 6,
7 and 8)
2. Schilling, H. D., B. Bonn and U. Krauss, Coal Gasification:
Existing Processes and New Developments. Graham & Trotman, London,
1981
3. Brecht, Chr. and G. Hoffmann, Coal gasification - A tabu-
lated schedule of development at home and abroad and processes
used on an industrial scale, Gaswarme International 32 (1983),
p. 7 - 23
4. Bogner, F. and K. Wintrup, The fluidized bed (Winkler) Gasi-
fication in Handbook of Synfuels Technology Mc Graw Hill Book
Comp., New York, 1983
5. Hiller, H., Kohle-Druckvergasung. Erdol und Kohle, Erdgas,
Petrochemie 28 (1975), S. 74
6. Hoogendorn, J.C., Die neue Anlage zur Erzeugung flussiger
Kohlenwasserstoffe aus Kohle in Sudafrika. Gas, Warme International
25 (1976), S. 283
7. British Gas/Lurgi, Proving the commercial viability of the
slagging gasifier concept, Energy world, March 1982, p. 2-5
8. Toriter, K., R. L. Margolis and A. K. Kuhn, The Great Plains
coal gasification project, Coal Internat. 2 (1983), p. 7-8
9. Staege, H., Kohlevergasung im Flugstrom, Krupp Mitt. 67
(1983), p. 4-5
10. Chari, K., Coal based ammonia plants. Chern. Age India 28
(1977), S. 698
11. Lohmann, C., Robke, G., Das Kohlevergasungsprojekt Dorsten,
Lurgi - Ruhr 100 - Bisherige Betriebserfahrungen. Gas- und Wasser-
fach Gas 121 (1980), S. 359
 420

12. Langhoff, I., W. Schafer, C. Lohmann and H. Peyrer,

Ruhr 100 - New Results on the advanced development of the pressu-
rized Lurgi gasification, 9th Annual Int. Conf. on Coal Gasifica-
tion, Pittsburgh, 1982
13. Theis, K. A. and U. Femmer, The High-Temperatur-Winkler
demonstration plant, Braunkohle, April 1984, p. 120-124
14. Cornils, B., J. Hibbel, P. Ruprecht, R. DUrrfeld, J. Lang-
hoff, RCH/RAG's Version of Texaco coal gasification, Hydrocarbon
Processing, Jan. 1981, p. 149-156
15. RoB bach , M., Meyer, A., Hornung V., Das Saarberg/Otto-Koh-
levergasungsverfahren. Stahl und Eisen 100 (1980), S. 383
16. Volkel, H. K., Der Shell-Koppers-ProzeB. Ein Verfahren zur
Vergasung von Kohle unter Druck. Energie 30 (1978), S. 196
17. Heek, K. H. van, H. JUntgen and W. Peters, Steam gasifi-
cation of coal by means of process heat from high-temperature nuc-
lear reactors. Atomkernenergie/Kerntechnik 40 (1982), p. 225/246
18. Kichhoff R., K. H. van Heek, H. JUntgen and W. Peters,
Operation of a semi-technical pilot plant for nuclear aided steam
gasification of coal, Nuclear Engineering and Design, 78, (1984),
p. 233-239
19. JUntgen H., K. H. van Heek, R. Kirchhoff, J. Klein, Quanti-
tative Untersuchungen zur Umwelbeeinflussung durch Anlagen zur
Kohlevergasung und -hydrierung, Chemie und Umwelt, Chern. Ind.,
Februar 1983, S. 96-97
 421

CHARACTERISATION OF COAL FOR GASIFICATION PROCESSES

K. Mark Thomas

British Gas Corporation, London Research Station, Michael Road,

London SW6 2AD, UK.

INTRODUCTION

The characterisation of coal for gasification processes has

several functions and these are given below:

1. To provide data for the calculation of heat and mass

balances and efficiency,

2. To predict coal behaviour during gasification,

3. To obtain optimum conditions for gasification,

4. To provide specifications for the purchase of coal

feedstocks.

The aim of coal characterisation is to determine the minimum

number of tests or measurements required to provide the
information on coal properties required for a particular
gasification process. The alternative would be to test each
unknown coal empirically on a commercial or pilot scale reactor
which is neither economic nor realistic.

This review describes problems associated with the

characterisation of coals for high pressure gasifiers and
examines the effects of process variables such as heating rate,
pressure etc. on coal properties.
 422

COAL STRUCTURE

Coal is found (1) in widely different geographical and

geologically diverse environments and is derived from a wide
variety of complex plant materials which have been subjected to
chemical and physical processes under various conditions of
temperature and pressure over periods of millions of years. It
consists of mainly organic with small quantities of inorganic
materials, and these are distributed in a heterogeneous manner
when considered at the macroscopic and microscopic levels. The
inorganic material can take the form of discrete mineral species
or form a chemical complex with the coal. As a result the
properties of coal obtained from a mine vary on a day-to-day
basis leading to the need for careful sampling procedures in
order to obtain a representative sample.

When pure compounds are characterised it is usually

possible to specify the structure precisely. Structural
determinations can be achieved by a variety of chemical and
spectroscopic methods or in the case of crystalline solids by
X-ray diffraction methods. The latter provide accurate bond
lengths and the conformation in the solid state. In comparison,
the structural characterisation of coal is much more difficult
and the reason for this is its heterogeneous and macromolecular
nature. It is generally recognised that the structure of
macromolecules varies considerably. Coal is generally
considered (1,2,3) to be an assembly of units which vary in
composition and whose average size increases with rank. These
units may be cross-linked in a three-dimensional manner making
them difficult to separate and investigate. A small percentage
of the coal consisting of smaller molecules can be fairly easily
extracted. The physical and chemical structure of the organic
macromolecules are responsible for the coal and properties.

From the point of view of regarding coal as a macromolecule

so that it can be envisaged as reacting in conventional chemical
terms, models have been constructed based on experimentally
determined parameters such as chemical composition (C, R, N, 0
etc.); functional groups (-OR, -COOR, = 0); aromatic/aliphatic
ratios and other structural data (1,4). This method gives
statistically-averaged structures which while not existing as
such do provide a framework for discussion of possible chemical
reactions during pyrolysis, combustion and gasification. This
approach has limited practical use because it is difficult to
relate closely measurements of coal behaviour and physical
properties to the structures.
 423

General structural models for coal, based mainly on X-ray

data have been proposed (5). The imperfections in coal give
rise to diffuse diffraction patterns in comparison with
graphite. These have been used to estimate layer size and
stack height distributions of crystallites for a range of
coals. This gives a qualitative description of the structure
of coals in terms of layers without details of atomic positions
and has proved useful from a conceptual point of view.

The physical properties of coal and the changes in these

properties during heating are to a large degree a consequence
of the physical structure which represents the spatial
arrangement of the large macromolecules and smaller molecules
in coal and the extent of crosslinking. The physical
properties often vary systematically with rank and can be
studied (6) without a detailed knowledge of the chemical
structure being available.

Coal is clearly heterogeneous at the macroscopic level and

it may have a banded structure." These bands are lithotypes and
there are several nomenclatures for describing coals
megascopically. All coal lumps are cracked and fissured
showing another aspect of the physical structure. The organic
detrital matter of coal includes identifiable plant remains,
these being recognised by optical microscopy and termed
macerals. The macerals which are the microscopic constituents
of coal can be identified according to their morphology and
optical properties. The three groups of macerals (vitrinite,
exinite and inertinite) differ in chemical and physical
properties. The microscope also shows that coal consists of
holes, cracks or macropores which may represent gaps between
particles. At the atomic scale coals may have an ultrafine
structure with pore sizes down to a few nanometres in size.
These pores represent holes in or between macromolecules. Some
of the pores may be open giving gases access to the internal
surface area of the coal whilst others may be closed but may
open up during reaction. The pore structure can be quantified
by adsorption of gases, penetration of liquids and by small
angle X-ray scattering.

The structure of coal can be described in a qualitative

way but it cannot be defined precisely because of its
heterogeneous and macromolecular structure. The important
question is the extent to which coal structure needs to be
understood in order to provide the information necessary for
predicting optimum gasification conditions. The relative
importance of the various properties in relation to
gasification behaviour needs to be assessed and ways devised of
relating the information to actual plant operations. This will
 424

lead to the efficient use of resources and the optimization of

gasification processes.

BASIC COAL CHARACTERISATION DATA

The initial stage is to provide basic coal characterisation

data based on standard tests. This information will be useful
for most processes and allows a particular coal to be classified
in several commonly used coal classification schemes. Details
of the basic coal characterisation scheme are given below:

1. General data - sampling, origin, storage, particle size

distribution etc.

2. Proximate analysis.

3. Ultimate analysis including oxygen, forms of sulphur,

chlorine and carbon dioxide.

4. Gross calorific value.

5. Ash analysis - Si, Al, Fe, Ca, Mg, Na, K, Ti, Mn, S, P and
Silica ratio.

6. Petrographic data - Maceral analysis and reflectance

measurements.

7. Physical data
a) Caking and swelling Swelling number,
dilatometry and Gray-
King coke type.

b) Ash fusion characteristics Oxidising and reducing

environments.

This basic characterisation which is based on British

Standard tests (BS1016), provides sufficient data on coal
properties for use in the calculation of heat and mass balances;
it allows an approximate prediction of coal behaviour during
gasification and determination of gasifier conditions. It also
enables a wide range of coals to be selected for gasification
tests. However, it is noteworthy that some of the tests are
empirical and are carried out under arbitrary conditions which
are not appropriate for gasification. An example of this is
dilatometry (BS 1016, Part 12) which is measured at 3 C min- 1
and atmospheric pressure.
 425

GASIFICATION PROCESSES

Optimization of the coal gasification process in terms of

operability requires a more detailed examination of the coal
properties and their relationship to gasifier conditions. The
selection of these properties will depend on the type of
gasification process. Gasification processes can be classified
(7) according to the methods of contacting gases and solids as
follows:-

1. Fixed bed

2. Entrained-flow

3. Fluidized-bed

4. Molten bath

and the removal of the solid residue as

5. Ash

6. Slag

The efficient removal of the inorganic residue is necessary

for the continuous operation of the gasifier.

It is apparent from the widely different designs of

gasifiers and methods of contacting gases and solids that the
choice of a particular suite of special test methods which need
to be used to predict the behaviour of coal during gasification
will depend to a certain extent on the particular process and
the prevailing gasification conditions. For example, in an
entrained flow gasifier the study of behaviour of small coal
particles during rapid heating to very high temperatures will
provide useful information. In this case the gasification is
usually completed rapidly and all the product gases have low
molecular weight. In contrast, fluidized bed gasifiers have a
relatively uniform bed temperature throughout because of their
good heat and material transfer properties and in this case the
reactivity of the char under appropriate conditions and ash
fusion temperatures may be important considerations. However,
rapid heating can occur on the introduction of material in the
fluidized bed leading to problems with caking and swelling.
Fixed bed gasifiers have several well defined zones - drying,
low temperature carbonization (involving primary
devolatilization, caking and swelling and coke formation),
gasification and combustion, although these zones may overlap.
In their simplest operational configuration, graded coal
 426

(-1/4" - 2") is used as feedstock. However, there is a

considerable economic benefit to utilizing fines which can
involve feeding a limited quantity in with the graded coal,
entrained flow injection through the tuyeres and the manufacture
of briquettes with pitch binders. Since the characterisation of
coals for fixed bed gasifiers illustrates a wide variety of
problems associated with coal characterisation, such gasifiers
will be considered in more detail.

In this type of gasifier there is a counter current flow

of gases and solids. The coal introduced into the upper part of
the gasifier is carboni sed by the sensible heat of gases
produced in the gasification zone. In the carbonization zone
gases and liquid products are evolved and the coal agglomerates
and swells. Large caked lumps which may interfere with the flow
of gases and solids aFe broken down by the stirrer. The coke
formed slowly degrades in size as it passes down the gasifier
and reacts to a certain extent with the gases flowing upwards.
The main gasification occurs (8) at the bottom of the gasifier,
for example, in the raceway of the slagging gasifier. Optimum
conditions are achieved when there is a smooth flow of coal down
the gasifier and gas flows evenly through the bed. The coal
properties which significantly affect this are:

1) Caking and swelling.

2) Devolatilization.

3) Coke properties (strength, reactivity etc).

Caking and swelling and the primary devolatilization occur

simultaneously and lead to the formation of coke and hence the
inter-relationship between these phenomena is important. Most
of the measurements of these parameters are based on empirical
techniques which are very dependent on the experimental
conditions. Therefore an important aspect of the study of coal
properties in relation to gasification is the effect of changing
process variables such as pressure, heating rate, gaseous
atmosphere, particle size etc and to simulate, as far as
possible the conditions actually occurring in a commercial
gasifier. These properties may also provide the basis for
determining relationships which can be used to obtain a better
understanding of industrial gasifier processes. In addition a
knowledge of properties of the ash or slag produced is required
for continuous operation.
 427

CAKING AND SWELLING PROPERTIES

An appraisal of coal agglomeration and related rheological

properties is not straightforward (9). It requires measurement
of several fundamentally different empirical parameters and also
the effect of experimental conditions on these parameters.
Swelling-index and dilatometry are important measurements of
swelling whereas plastometry measures the extent of
agglomeration (variable torque) and fluidity (Gieseler).

The caking and swelling behaviour of coals are normally

measured at atmospheric pressure and in comparison very little
work has been carried out at elevated pressures. Consideration
of widely held views of the mechanism of caking and swelling
which involve the release of volatiles suggests that the effect
of pressure would be significant. There have been several
studies of the effect of experimental parameters such as,
pressure, heating rate, particle size and gaseous atmosphere on
dilatometry and swelling number measurements, and these have
shown complex relationships. Studies of the effect of pressure
and heating rate on dilatometry parameters on coals with a wide
range of rank have shown (10-13) that three types of
dilatometry-pressure curve have been observed (see Fig 1):-

a) Dilatation increases with increase in pressure.

b) Dilatation increases to a maximum at 1-2 MPa followed by
a decrease to a limiting value.
c) Dilatation decreases with increase in pressure.

The shapes of these curves obtained for a given coal may be

modified by changing the heating rate, with increased rate
increasing the dilatation. The large differences in dilatation
observed at atmospheric pressure are correspondingly smaller at
high pressures. Above 4MPa there appears to be only a slight
change in dilatation. If dilatation is used as a technique for
comparing the swelling behaviour of coals for gasification it
has been shown (10,12) that this order can change significantly
with pressure and this mainly occurs in the region below 2.5
MPa. It was not possible to predict the shape of
dilatometry-pressure curves from measurements made under
standard conditions or characterisation data. This was clearly
illustrated (12) by two coals having very similar basic
characterisation data as outlined previously but a
dilatometry-pressure curve which decreased with increase in
pressure for one coal whereas the curve for the other coal had a
distinct maximum at 1.5 MPa. The shapes obtained for
dilatometry pressure curves can be explained qualitatively in
terms of the increased pressure:
 428

1) Reducing volume of gas trapped and

2) Increasing the residence time of the tar/bitumen volatiles
in the fluid phase.

The former reduces the swelling whereas the latter

increases it. The balance between the two competing processes
can give rise to the three types of pressure-dilatometry curve.

300

Dilatation -1
Heating rate 40"cmin

--
c
.2
IV
!!
Q 100

A 902

-100~ _ _ _---::L:-_ _ _ _-+::-_ _ _---:=-::!

o 20 40 60
Pressure MPa (x10)
Fig. 1 Typical Dilatation - Pressure Curves for a range of coals

It has been shown (10) that dilatometry measurements are

not markedly affected by a hydrogen atmosphere but some
differences are observed (10,11) for different particle size
distributions.
 429

Recent studies (14-17) of the fluidity of coals at elevated

pressures has shown that the gas used can have a significant
effect. The coals become more fluid in a hydrogen atmosphere
than under nitrogen. However, the characteristic temperatures
determined in the Gieseler p1astometer were not markedly changed
by pressure and composition. Kaiho and Toda reported (14) that
the maximum fluidity of a weakly caking coal increased rapidly
in hydrogen and this was attributed to reaction with hydrogen
rather than suppression of volatile matter evolution. Below 265
p.s.i.g. (1.8 MPa) the maximum fluidity of the coal under
atmospheres of argon, carbon dioXide, nitrogen and helium,
follows the trend that the larger the diameter of the gas
molecules, the larger the fluidity, and this was thought to be
due to retention of volatile matter in the system. Lancet et a1
(15-17) also found that an increase in pressure caused a marked
increase, and hydrogen a moderate increase in fluidity for a
range of six U.S. coals. There appears to be general agreement
that high pressure inert gases promote condensation and
repo1ymerisation whereas hydrogen reduces repo1ymerisation.

DEVOLATILIZATION

A knowledge of the effects of changing parameters such as

heating rate, pressure, gaseous atmosphere, particle size etc on
coke, gas and tar yields and composition is of significance in
relation to caking and swelling properties and gasifier
behaviour. In addition, the properties of the tar produced are
also of interest in relation to tar recycling and processing and
the gases evolved during pyrolysis contribute to the overall
product gas.

The experimental equipment employed in coal pyrolysis

experiments often influences (18) the results obtained. This is
the result of the introduction of heat or mass transfer effects
on conversion rates thought to be kinetically controlled. It is
evident that experiments should as far as possible reflect the
conditions in the particular process. It is noticeable that
there is considerably less information available on pyrolysis at
high pressures and for large particles than for pulverised
fines.

The weight loss during pyrolysis of coal in an inert

atmosphere at elevated pressures is less than at atmospheric
pressure. This corresponds to an increased char/coke yield and
reflects the increased residence time of volatiles within the
particles leading to greater amount of secondary reactions
taking place. Data obtained (19) at pressures between 0.001 and
100 atm suggest (O.lKPa and 10 MPa) asymptotic behaviour at the
pressure extremes. The limiting yield at high pressure probably
 430

reflects the effect on volatility of the decomposition products

and the loss of significant quantities of species capable of
undergoing secondary reactions. An increase in pressure also
causes the production of less tar, increased methane and carbon
oxides, reflecting the enhancement of secondary reactions. The
water yield is decreased by increase in pressure for lignites
but the converse is true for bituminous coals. The pyrolysis of
coal at temperatures up to 1600 C and pressures (0.1 to 1.5 MFa)
has been studied (20) using a thermogravimetric reactor. The
results were consistent with a three stage mechanism for the
evolution of volatiles.

The influence of temperature, heating rate and pressure on

the kinetics of coal pyrolysis under inert gases and hydrogen
has been investigated (21). This study involved the
measurement of rates of formation, the yields of liquid and
gaseous products, and the amount of residual char under
pressures of 1 to 100 bar and heating rates between 1K min- i and
2000K sec-i. Two types of apparatus, a heated wire gauze for
rapid heating rates «2000K sec-i) and small quantities (10mg)
of coal (0.2 - 0.315mm) and a thermobalance for larger
quantities (~ 19 with particle size (0.8 - 1mm» and heating
rates up to 10K min-i. High heating rates result in slightly
higher yields of tar under an inert atmosphere whilst the yields
of hydrocarbon gases remain the same. An increase in pressure
decreases the tar yield. In a hydrogen atmosphere, increased
yields of high aromatic products are obtained but this reaction
is less effective at high heating rates due to the shorter
residence time and hydrogen diffusion problems. The yields of
methane and ethane increase markedly due to gasification of the
char. These results have been used (22) in design calculations
for entrained flow and fluidized bed reactors.

The effect of increasing particle size (0.1 - 1mm) is to

decrease (18) the weight loss. This behaviour is consistent
with large particles providing a greater resistance to the
escape of volatiles leading to an increase in the extent of
secondary reactions. The trends with respect to increase in
pressure for tar, methane and carbon oxide yields are those
expected from an increase in secondary reactions. The data
available on the effect of heating rates does not allow any firm
conclusions to be drawn.

In a fixed bed gasifier, pyrolysis takes place in the

product gases which have high concentrations of carbon
monoxide and hydrogen, and the particle size is quite large (5
to 50mm) and these factors will be important in determining the
pyrolysis behaviour. Hydrogen reacts at a considerably faster
rate with decomposing coal or some of the primary volatiles,
than with residual char. Reactive fragments produced during
 431

pyrolysis are hydrogenated with the products generated being

controlled by the thermal decomposition of the coal. The
transfer of hydrogen into, and volatiles out of, coal particles
is not well understood. The weight loss of small particles
(70~m) increases markedly in a hydrogen atmosphere at high
pressure and high heating rates. However, there is little
information available for conditions which resemble those found
in fixed bed gasifiers.

CHAR/COKE PROPERTIES

The coke formed at the top of a fixed bed gasifier degrades

in size as it passes down the gasifier. The coke should not be
too strong so that it is not broken up by the stirrer. On the
other hand if it degrades too much as it passes down the
gasifier, there may be excessive dust carried over (23).
Thermal stresses arising when particles are heated rapidly may
also cause degradation in particle size. In addition, char
particles moving around in the raceway may suffer degradation
which will be a function of the blast momentum (23). A
knowledge of the factors affecting strength and abrasion
resistance of the char is an important consideration because
the fuel degradation is dependent on these factors.
Accumulation of fine particles in the gasifier reduces the
permeability of the bed.

The BCURA constructed (24) an apparatus for carbonizing

coal under the conditions found in the carbonization zones of a
fixed gasifier at pressures up to 2.5 MPa and temperatures up to
1200 C. The chars produced in the apparatus were assessed for
strength and abrasion resistance by drum tests. A series of
coals between 700 and 900 in rank in the NeB classification were
studied under simulated dry ash (800 C) and slagging (1200 C)
conditions. It was found that under dry ash conditions the
strength index. was independent of rank but the abrasion
resistance was at a maximum for coals with a Gray-King coke of B
or BC. However, carbonisation under simulated slagging
conditions showed that the strength index varied with rank in an
inverse manner to the abrasion index which showed the trend as
under dry ash conditions.

Lancet et al (15-17) prepared cokes by shock heating coal

to 800 C at total pressures of atmospheric, 215 p.s.i.g.
(1.46MPa) and 365 p.s.i.g. (2.5 MPa) to study the effect on coke
properties. Similar tests were carried out with 65 p.s.i.a.
(0.44 MPa) and 115 p.s.i.a. (0.78 MPa) hydrogen partial pressure
in nitrogen. The coke strength was determined by a mini-drum
test. The mercury density of the cokes was also measured. The
strength of the coke made from low and medium fluidity coals
 432

decreased with an increase in pressure, but the effect for high

fluidity coal was negligible. The effect of hydrogen partial
pressure on coke strength showed no systematic variation except
for high fluidity coals where it decreased slightly. The coke
density remained the same or decreased very slightly with an
increase in total pressure. A study of cokes prepared from high
fluidity coal under a hydrogen partial pressure showed that the
density of the cokes remained the same or increased slightly
with an increase in hydrogen partial pressure.

The reactivity of char/coke is not an important parameter

in the slagging gasifier because of the high temperatures
produced in the raceway. However, in the dry ash Lurgi gasifier
where the temperatures are considerably lower, the process is
much more sensitive to changes in reactivity (25). The
reactivity of a char is difficult to define and measure
precisely (26). It is a function of several parameters and
decreases with an increase in rank and the mineral distribution
and composition; porosity, surface area, degree of burn-off and
particle size are important considerations.

ASH/SLAG PROPERTIES

Ash fusion temperatures and ash composition are carried out

as part of the basic coal characterisation. The ash fusion
temperature gives an approximate indication of the behaviour in
a gasifier but it is affected by iron content. This effect has
been attributed (27,28) to the fluxing· action between the
ferrous and ferric ions leading to higher ash fusion
temperatures in an oxidising atmosphere. However, the basic
coal characterisation does allow a prediction of the viscosity
of a slag produced from the ash as a function of the
temperature. At a given temperature the viscosity of the slag
varies with the composition of the ash/slag. Early correlations
were based on silica ratio and these have now been superseded
(29,30) by more complex empirical correlations. The aim is to
produce slag with a viscosity which is suitable for tapping.
This is achieved by a combination of fluxing and adjusting the
temperature of the slag by varying the steam/oxygen ratio.

THE EFFECT OF PROCESS OPERATIONAL CONSIDERATIONS

In·additien t~studylng the effect of process variables on

dilatometry and plastometry parameters it is also of interest to
include investigations of the effect of mineral content, tar and
p~tcb add1ti~ and pretreatment procedures because of their
tmportan~'iB large scale gasification proeasses. The mineral
content of a coal feedstock may vary quite significantly,
 433

affecting the behaviour of the feedstock. Similarly a flux is

added to the feedstock. Lancet et al (15-17) investigated the
effect of mineral addition on Gieseler fluidity and swelling
number. They found that 10% addition decreased the fluidity and
swelling significantly. Kahn and Jenkins (31) concluded from
investigations of some of the same coals with a high pressure
microdilatometer, that mineral addition reduced swelling at
atmospheric pressure but had very little effect at elevated
pressures (2.8 MPa). It was also noted that the composition of
the atmosphere had very little effect and this was ascribed to
the rapid heating rate used.

The effect of tar and pitch are also of interest because

tar is recycled at the top of the slagging gasifier, and pitch
may be used as a binder in the manufacture of briquettes for
utilization of fines. The addition of 10% tar to a range of
U.S. coals has led (15-17) to a marked increase in Gieseler
fluidity, usually an increase in swelling number, no effect on
coke yield and may cause some increase in the coke strength and
density. The effect of tar, coal tar pitch and a petroleum
pitch on dilatometry parameters have also been investigated
(13). All the additives enhanced the swelling significantly
below 1.5 MPa but had little or no effect above this value. The
optical texture of the resultant cokes was also investigated.
The addition of tars and pitches enhanced the anisotropic
content especially at low pressure, without affecting the
composition of the anisotropy and the content reached a limiting
value.

The addition of mineral matter (20%wt) to coal resulted

(15-17) in an increase in density of the coke produced at high
pressures. However, the results of doping mineral matter into
clean coal on the coke strength, were inconclusive. The effect
of tar addition on the coke properties is dependent on the rank
of the coal. The effect of tar, pitch and mineral addition and
experimental conditions on various parameters such as
dilatometry, optical texture etc are summarised in table 1.

The effect of preoxidation on the thermoplastic properties

of Pittsburgh Seam HVAb coal has been studied (32) using a high
pressure microdilatometer. The results showed that the maximum
swelling parameters of this preoxidized coal were always less
than those of the untreated coal at pressures < 1.0 MPa. At
pressures greater than 2.0 MPa, the maximum swelling parameters
of the preoxidized samples were in general greater than those of
the untreated coal. The conclusions of the study were that
carbonization at elevated pressures may partly restore the
thermoplastic properties of preoxidized coal and that the
behaviour of the coal must be investigated under actual
utilization conditions.
 -I=-
Parameter Gieseler Swelling Dilatometry Tar Coke Coke Coke Optical ~
Fluidity Number Yield Yield Strength Density Texture

Increase in
variable

Total + -0+ -0+ + ++ 0- +

Pressure

Hydrogen ++ 0 0+ 0+ 0- 0- -0+ ?
Pressure

Heating ++ +0 +0 + ? ? ? +
Rate

Tar ++ + +0 ? 0 0 0+ +
Addition

Pitch ++ ? +0 ? ? ? ? +
Addition

Mineral ? ? ? 0+
Content

References 14-17 lO,15-17 10-13 21 15,17-19 15-17 15-17 13

+ increase
0 unchanged
decrease
? unknown

Table 1
The Effect of Process Variables on Caking and Swelling, Devolatilization and Coke Properties
 435

GASIFIER SIMULATIONS IN THE LABORATORY

Lurgi have developed (33) a laboratory scale reactor which

operates· under simulated gasifier conditions and provides
information on some of the important coal properties. In this
case the gas composition is varied in line with the zones of the
commercial gasifier; the temperature-time patterns also simulate
movement down the gasifier, and the particle size range of the
coal is the same as for a commercial gasifier. The parameters
measured were product gas volume and composition. The bed
temperature profile was measured continuously and condensable
products integrally. Also, the coke sample remaining in the
pressure vessel can be examined by standard methods. This
laboratory apparatus provided the yields and compositions of
pyrolysis products and information on the disintegration
behaviour of the coal for use in predicting gasification results
in commercial size gasifiers with improved accuracy.

CONCLUSIONS

The characterisation of coals for gasification processes

has inherent difficulties related to the heterogeneous and
macromolecular nature of coal. A knowledge of coal behaviour
under gasifier conditions is of greater importance than a
detailed knowledge of the coal macromolecular structure. There
is comparatively little work reported in the literature on the
study of coal behaviour under high pressures. The results that
are available, highlight the inherent dangers in extrapolating
trends from measurements at atmospheric pressure and the
complexity of the problem. It is clear that measurements must
be carried out as far as possible under conditions found in the
gasifier.

It must be remembered that the coal parameters measured in

the laboratory only simulate some of the conditions. It is
noteworthy that factors such as particle size, quantity,
stirring speed, etc are often very different in the gasifier.
However, changing gasification parameters on a plant and using a
wide range of coals will provide operational experience as to
the relative importance of the various tests and laboratory
measurements, and from this information it should be possible to
develop a coal classification scheme specifically for a
particular process. Coal characterisation as a predictive
technique for gasifier behaviour can only follow extensive
correlation between plant performance and measured coal
properties.
 436

ACKNOWLEDGEMENTS

The author would like to thank British Gas for permission

to publish this paper and colleagues for helpful discussions.

REFERENCES

1. Van Krevelen, D.W. "Coal Typology - Chemistry - Physics -

Constitution", Elsevier, Amsterdam, 1961
2. Green, T., J. Kovak, D. Brenner, and J. Larsen, in 'Coal
Structure' (Ed. Myers R.A.) Academic Press, New York, 1982
p. 199.
3. Grainger, L. and J. Gibson, 'Coal Utilization: Technology,
Economics and Policy, Graham and Trotman, London 1981
p.22.
4. Davidson, R.M., in 'Coal Structure' (Eds. Gorbaty, M.L.
Larsen, J.W. and Wender, I.) Vol. 1 Academic Press, New
York, 1982 p.84.
5. Hirsch, P. B. Proc. Roy. Soc. A226, 1954, 143-169.

6. Grimes, W.R. in 'Coal Structure' (Eds. Gorbaty, M.L.,

Larsen, J.W. and Wender, I.) Vol.1 Academic Press, New
York, 1982, p.21.
7. Hebden, D. and H.J.F.S. Stroud in 'Chemistry of Coal
Utilization' (Ed. Elliott, M.A.) Second Supplementary
Volume, Wiley-Interscience, New York, 1981, 1599.
8. Brooks, C.T. in 'Oxygen and the Conversion of Future
Feedstocks' Proceedings of the Third BOC Priestley
Conference. Sept. 1983. Special Publication No. 48, The
Royal Society of Chemistry, London, p.76.
9. Habermehl, D., F. Orywal, and H.D. Beyer, in 'Chemistry of
Coal Utilization' (Ed. Elliott, M.A.) Second Supplementary
Volume, Wiley-Interscience, New York, 1981, p.317.
10. Beyer, H.D. 'Caking and Coking Power of Bituminous Coals
under High Pressure' Report BMFT-FB-T. 82-055
Erdol und Kohle-Erdgas Petrochemie 1981, 34, 409.
11. Kahn, M.R. and R.G. Jenkins, Fuel, 83 (I), 1984, 108.
12. Green, P.D. and K.M. Thomas, Fuel (In Press).
13. Green, P.D., J.W. Patrick, K.M. Thomas and A. Walker, Fuel
(In Press).
14. Kaiho, M. and Y. Toda, Fuel 58, 1979, 397
15. Lancet, M.S and Sim, F.A. ACS Div. Fuel, Chem., Prepr 26
(3), 1981, 167.
16. Lancet, M.S, G.P. Curran and F.A. Sim, EPRI report AP 2535,
Research Project 1267-7, August 1982.
 437

17. Lancet, M.S. F.A. Sim, and G.P. Curran, ACS. Div. Fuel,
Chem., Prepr 1982, 27 (1), l.
18. Howard, J.B. in 'Chemistry of Coal Utilization' (Ed.
Elliott, M.A.) Second Supplementary Volume,
Wiley-Interscience, New York, 1981, p.665.
19. Anthony, D.B. J.B. Howard, H.C. Hottel, and B.P. Meissner.
15th Sump. (Int.) Combustion, Combustion Institute,
Pittsburgh, 1975, p.1303.
20. Tamhankar, S.S, J.T. Sears and C.Y. Wen, Fuel, 63, 1984,
1230.
21. Arendt, P and K.H. van Keek, Fuel, 60, 1981, 779.
22. van Heek, K.H. Ger. Chem. Eng. 7, 1984, 319
23. Sparham, G.A. and I.G.C. Dryden, BCURA Monthly Bull. 17,
1963, l.
24. 'Carbonization Characteristics of Coals under Conditions
Simulating those in Fixed Bed Gasifiers Operating at High
Pressure.' BCURA Report A/CON/2624. A Test for Determining
the Suitability of Coals for Fixed Bed Gasifiers Operating
at High Pressure. BCURA Report APGN 25.
25. Tart, K.R. and T.W.A. Rampling, Coal Chem. 2000
The Institution of Chemical Engineers Symposium Series 62,
p.5!.
26. Mahajan, O.P. and P.L. Walker Jr, in 'Analytical Methods
for Coal and Coal Products' (Ed. Karr, C.) Academic Press,
New York 1978, Vol. 2, p.465.
27. Ely, F.G. and D.H. Barnhart, in 'Chemistry of Coal
Utilization', (Ed. Lowry, H.H.). Supplementary Volume,
Wiley, New York, 1963 p.820.
28. Sherman, R.A. and B.A. Landry, in 'Chemistry of Coal
Utilization' (Ed. Lowry, H.H) Supplementary Volume, Wiley,
New York, 1963, p. 773.
29. Riboud, P.V., Y. ROux, H. Lucas, H. Gaye and H. Fachber,
Huttenpraxis Metallwertuer, 19. 1981, 859.
30. Urbain, G., F. Cambier, M. Delatter and M.R. Anslan,
Trans. J. Brit. Ceram. Soc. 80, 1981, 139.
31. Kahn, M.R. and R.G. Jenkins, Fuel Processing Technology 8,
1984, 307 and EPRI report AP-2337, 1982.
32. Kahn, M.R. and R.G. Jenkins, Fuel, 64 1985, 189.
33. Stoldt, E., G. Schaub and C. Hafke, Erdol und Kohle, Erdgas
Petrochemie, 37 (1), 1984, 19.
 439

SOME SCIENTIFIC ASPECTS OF THE DEVELOPMENT OF THE SLAGGING

GASIFIER

K. Mark Thomas

British Gas, London Research Station, Michael Road, London,

SW6 2AD, UK.

INTRODUCTION

At present, there is no lack of energy supplies. However,

world events could change this situation very rapidly, as in the
so-called 'oil crisis' of 1973-74. The proven reserves of coal
far exceed those of other fossil fuels and they are fairly
widely distributed throughout the world. These reserves will
last several hundred years at current rates of consumption and
could fill a shortfall in energy supplies when oil and gas
supplies become scarce.

Coal tr~ification and gas transmission have several

advantages over using coal to generate electricity. In
particular, it is more economic being 60-80% efficient in
comparison with 30-35% for electricity generation, which after
allowing for other considerations gives a factor of two
advantages in efficiency. Environmental considerations are also
important. It is both expensive and difficult to remove
completely oxides of sulphur from the stack gas produced in coal
combustion. During the gasification of coal, the sulphur
present is converted mainly to H2 S and this can be easily
removed by conventional means. Recently, there has been
considerable interest in electrical power generation using a
gasifier in a combined cycle of gas and steam turbines. This
leads to an increased overall thermal efficiency for power
generation (~ 40%) compared with combustion processes. In
addition, the use of gasification processes for the production
of synthesis gas for use in the production of chemicals is also
of considerable interest. Clearly there are strong economic
reasons for continued research into coal gasification processes
 440

and the long lead times for developing the new technologies
means that this has to be carried out several decades before it
is needed commercially.

The objective of this review is to give a general overview

of fixed bed gasification processes and highlight the areas of
interest and problems associated with coal properties and
characterisation.

CHEMICAL PROCESSES OCCURRING DURING GASIFICATION

In comparison with liquid and gaseous hydrocarbons, coal

has a relatively low hydrogen content. Therefore the conversion
of coal to these products requires addition of hydrogen or
rejection of carbon or a combination of both these processes to
increase the Hlc ratio. This is achieved either by direct
hydrogenation or by reaction with steam to produce synthesis gas
(CO+H 2 ). In fixed bed gasifiers, the latter is the primary
process but a small amount of hydrogenation of the char by the
product gas may occur in the upper part of the gasifier. The
synthesis gas can then be converted to hydrocarbons by catalytic
methods.

In most gasification processes, steam and oxygen are used

to gasify the coal. The reason for this is the highly
endothermic reaction with steam.

~HIOOOK = +136 KJmol- 1

~GIOOOK = -7.61 KJmol- 1

The oxygen present reacts with the coal in a highly exothermic

manner supplying heat to keep the reaction with steam going.

C + 02 ~ CO 2 ~HIOOOK = -394 KJmol- 1

~GIOOOK = -396 KJmol- 1
C + ~2 ~ CO ~HIOOOK = -112 KJmol- 1
~GIOOOK = -200 KJmol- 1

In a commercial gasifier temperatures are usually much higher

than 1000K and carbon monoxide is the major product.

C + CO 2 ~ 2CO ~HIOOOK = +171 KJmol- 1

~GIOOOK = +5.34 KJmol- 1

Combination of the above equations gives the net reaction

3C + O2 + H2 0 ~ 3CO + H2 ~HIOOOK -108 KJmol- 1

~GIOOOK -183 KJmol- 1
 441

Several other reactions can take place which may alter the final
composition of the product gas. The Sabatier reaction can
occur:

co + 3H 2 + -225.8 KJmol- 1
+27KJmol- 1

The CO shift reaction which will be limited by the amount of

steam present in the reactor:

CO + H2 0 CO 2 + H2 -34.6 KJmol- 1
-3 KJmol- 1

Hydrocarbon volatiles and char may also be hydrogenated to a

small extent.

C + 2H2 + -89.9 KJmol- 1

+19.3 KJmol- 1

The relative importance of these reactions will depend on

the concentrations of the gases and the conditions prevailing in
the region of the gasifier. The reactions outlined above do not
reveal any of the complex mechanisms of the processes they
represent or the kinetics.

FIXED BED GASIFIERS

The physics and ch~fi~try of a fixed bed coal gasifier is

complex and multistaged~ - ). The basic design involves
maintaining a constant bed depth and temperature profile with
the fuel gradually moving down the bed to the gasification zone
and hence these gasifiers have a high fuel inventory. In a
fixed bed gasifier there is a countercurrent flow of gases and
solids. The hot gases produced in the combustion and
gasification zones at the bottom of the gasifier by reaction
with oxygen and steam pass upwards and the sensible heat of the
gases carbonises the coal descending in the upper part of the
gasifier. The coal cakes and swells, the primary
devolatilization occurs with release of tar and gaseous products
and coke is formed during the carbonization. The agglomeration
of coal into large caked masses could interfere with the smooth
flow of coal down the gasifier and these are broken down by the
stirrer. If the coke produced is too weak, degradation in fuel
size occurs. This leads to a wide size range of material which
could affect the flow of coal and high dust levels may reduce
the efficiency and throughput of the gasifier. The flow of
gases and solids and the method of mineral matter removal are
critical factors in determining the operational characteristics
in this type of gasifier.
442

. . Feed coal

(a)

Drive--~)=;

Distributor -~R::::::::l~---

"'::' --
Grate drive / -
Steam & oxygen ~~~~)

~ Coal lock
Hopper
. . Feed coal

Coal distributor/
stirrer
(b)
Refractory linir.i:I
Gas outlet ...-~. -. 1.....::Ii_m/oxygen
supplies
Pressure shell

-
slag tap ------.:~{£:W
Circulating quench water
Slag quench
Slag lock hopper
chamber

Fig 1 (a) The dry-ash Lurg; Gasifier

(b) The British Gas/Lurgi Siagging Gasifier
 443

The dry ash Lurgi gasifier is well established(4) and

commercially proven. It has been operational for over fifty
years, and has been modified considerably during this period by
the introduction of a rotary coal distributor, steam cooled
grate and mechanical stirrer. The early development of the
Lurgi gasifier with slagging operation took place from 1950 to
1964 maig16)in the U.K. at the Gas Councils, Midlands Research
Station(' and the British Coal Utilization Research
Association (B.C.U.R.A.)(7,8). Following the discovery of
natural gas in the North Sea and the relatively low price of
oil, this work was suspended in 1964. It was restarted ten
years later as a result of support from U.S. oil and gas
companies.

The essential difference between the dry ash (see fig. 1a)
and slagging gasifiers (see fig. 1b) is the method for removal
of mineral matter and the conditions this produces in other
zones of the gasifier. In the dry ash Lurgi a steam/oxygen
ratio of typically 6-10 is used to keep the temperature below
the clinkering range(9). This means that there are large
amounts of undecomposed steam in the gasification prot~}ts.
Since the output of a gasifier is essentially limited by the
rate at Which product gases can pass through it without
significant carryover of fuel, a higher gasifier output can be
achieved if the undecomposed steam is kept to a minimum. The
ash produced in the process is removed by a rotating grate and
discharged through an ash lock. Above the grate there is a
layer of ash which protects a relatively shallow (~100mm in
depth) combustion zone immediately above this followed by a
gasification zone which extends up to a devolatilization zone at
the top of the bed. The coal cakes and swells and tar and
gaseous products are given off in the devo1atilization zone.

A schematic diagram(10) of the temperature profile of gases

and solids is shown in fig. 2(a). The effect of different char
reactivities is also shown on this curve. A low reactivity coal
produces a higher temperature plateau in the gasification zone
because the smaller extent of the carbon gasification reactions
with steam which are endothermic absorbing the heat produced in
the combustion zone. Highly reactive brown coals with high
moisture contentsmayproduce(lO)low off-take temperatures which
may cause condensation. The maximum temperature is naturally
also affected by the steam/oxygen ratio and gasification rate.

In the slagging gasifier the steam to oxygen ratio is

typically in the range 1-2 and is injected into the gasifier
using tuyeres. A region of low density solids entrained in high
velocity turbulent gases called the raceway is produced in front
of the tuyeres where the temperature is in the region of 2000°C.
 444

(a)
Carbonisation .
Pyrolysis. Gasification .
! Heat exchange .
.a Maximum fuel ~.
~ temperature 14000 C ....-! ~
CD _--;..'7
Gas temperature. - . - -__ -
; _--------- Low reactivity
~ ,. Final reaction 0 Fuel-H20/0i
-T--r-550 0 C. temperature---750 C. ratio increased
Average temperature ._- - - to give
of particle. identical peak
temperatue.
lORY-ASH I
. . .
( b ) ---. Depth from top of fuel bed:--+c omjustlon
Carbonisation. . . 41 /.. • 41 • 41 f •
P rol sis. GasIfIcation. Slag.

Final reaction
temperature .-z......
14000 C .......- ... - - -

// "
.~,---,Gas temperature.
/
/
// Average temperature of particle.
-350 0 C.
ISLAGGINGI

--+ Depth from top of fuel bed :--+

Fig 2 Comparison of temperature profiles in fuel beds
of slagging and non - slagging gasifiers
(from theoretical considerations)
 445

This produces a temperature in the hearth of 14-1500°C which is

sufficiently high to produce a molten slag which with the
addition of small amounts of limestone flux can be made of the
correct viscosity for tapping. Since the output of a gasifier
is ultimately limited by the rate at which gases can be passed
through the bed without excessive dust carryover, the lack of
undecomposed steam and higher temperatures in the slagging
gasifier gives a greater output than a dry ash gasifier of the
same size. The temperature decreases from the raceway by
reaction with steam absorbing heat to a plateau around ~1400°C,
and this is shown in fig. 2(b). The lack of excess steam and
the higher temperature means that the CO shift reaction is not
significant and higher yields of CO and less of CO 2 are obtained
from the gtagging gasifier compared with the dry ash Lurgi
gasifier( • Above this zone is the pyrolysis or
devolatilization zone where swelling and agglomeration takes
place. These phenomena can affect the flow of coal down the
gasifier.

Typical outputs and gasifict9!r~)parameters for Pittsburgh

8 coal and a wide range of coals are given in table 1.
The product gas compositions do not vary markedly for a wide
range of coal feedstocks (see table 1) or gasification
conditions (see fig. 3). There is an inverse correlation
between the methane composition and the oxygen content or rank
of the coal but very little correlation with volatile(m~tter.
The lower rank coals with higher oxygen contents have 11) in
general, higher reactivities. The mechanism of methane
formation is not completely understood.

Operational characteristics (see figure 4) are also of

importance. Fixed bed gasifiers usually use a graded coal as a
feedstock which means that special provision has to be made for
the utilization of the fines normally produced by modern mining
methods. Fines can be utilized in the following ways:

1. For steam raising in a separate boiler.

2. A proportion cfY1fe fed in with graded coal. The precise
amount depends on the swelling characterisation of a
given coal but up to 35% have been fed in with strongly
caking coals.
3. Fines can be injected as pf or coal water slurry through the
tuyeres. This method is not available in the dry ash
Lurgi gasifier.
4. Converted to briquettes using a pitch binder derived from
the gasifier tar and fed in as graded coal.
446

100 540

-
90
-...
520 ct>

-
G>
:::J
«S
80 500 ~
c.

70 • • 480
E
-G>

--
G>
•
- •
~

• • «S
I
~
-60 4600

-
c
0
Ul
050
c.
E
0
(,)

Ul 40
ca
e"

30 H2

20

10

1 1·5 2
Steam / Oxygen ratio.
Fig 3 Variation in gas composition
and off - take temprature with
steam/oxygen ratio.
 Graded Coal Crude Gas

+Fines

Sulphur
Recovery CO,
CO 2
Briquette REMOVAL
Manufacture

Sulphur

Oxygen
Plant Waste Gas Drying I SNG
Water and
Compression

Fig.4 A Schematic Diagram of the Siagging Gasifier and the Route to SNG using the
HICOM Process
.p..
.p..
-...J
 448

Parameter Pittsburgh 8 Typical Ranges

Dry Ash Slagging Dry Ash Slagging

Gasifier
Conditions

Pressure (bar) 24.0 22.5 23-32 22.5-24

Steam/Oxygen 9.85 1.22 6-10 1. 2-1.3
Gas off take 646 516 376-660 410-510
temp.(OC)

.9.!!.
Com:eosition
(% vol.)

CO 16.9 56.4 15-18 53-58

H2 39.4 28.0 38-41 27-29
CO 2 31.5 3.0 27-31 2-5.5
CH 4 9.0 7.1 8-11 6.7-7.1
C2 H6 ,C 2H4 0.8 0.4 0.5-1 0.4-0.7
N2 1.6 4.2 0.2-2.4 3.9-4.4
H2S 0.8

Coal
~acterisation
Data (% w/w)

Moisture 4.58 5.0 4.6-24.7 4.2-9.5

Ash 7.74 10.7 4.4-19.2 4.4-18.9
Volatile Matter 37.37 34.1 29.2-37.4 31.2-35.4
F.S.I. 7\ 7 0-7\ 0-7\
Gray King G8 G8 A-G8 B-G8
Ash Fusion
Oxidising 1360 1260-1400
Reducing 1170 1060-1400

Table 1 - Performance Data(9-11)for the Lurgi Dry Ash and

British Gas/Lurgi Slagging Gasifiers

Tar is also produced by both gasifiers and this can be

utilized in the following ways:

1. Sold as a by-product.
2. Recycled via the top of the gasifier.
3. Injected through tuyeres. This method is not available for
the dry ash Lurgi gasifier.
4. Converted to pitch for use as a binder in the utilization of
fines.
 449

Recycling of tar is standard operational practice on both

gasifiers. Tar and pf injection through the tuye~es were
successfully tested in phase 2 of the programme for the 6'
British Gas/Lurgi slagging gasifier at Westfield.

The dry ash Lurgi gasifier is best suited(9)to coals with a

high moisture, reactivity and ash fusion temperatures. The
British Gas/Lurgi slagging gasifier offers comparable efficiency
on these coals but considerable advantages for low reactivity
coals with low ash fusion temperatures and these represent a
significant proportion of U.S. and European coals. In
addition, it has the advantage of lower steam requirements and
greater throughputs and flexibility. The dry ash Lurgi gasifier
is well proven commercially and produces a gas which is more
readily converted into SNG and other commercial feedstocks
because of its high H2 /CO ratio.

The econo~ft~ and efficiency of the slagging gasifier have

been discussed~ in detail previously. Thermal efficiencies
are typically 80%.

COAL PROPERTIES IN RELATION TO THE SLAGGING GASIFIER

Clearly, although the gasifiers have many similarities, the

different mechanisms of mineral content removal have altered the
coal gasification behaviour sufficiently so as to present
different problems for coal characterisation although there are
still many similarities. An example of the former is the
importance of reactivity towards steam in the dry ash Lurgi
process and the insensitivity of the slagging gasifier to this
factor.

Gasification tests on coals with a range of properties

provide a starting point for developing an understanding of the
gasification of coal in a particular gasifier, but these tests
usually raise many questions which need to be answered. In
addition to heat and mass balance data, a knowledge of factors
affecting gasifier operational characteristics is required.
Hence laboratory methods need to be developed for measuring coal
properties and behaviour under appropriate conditions. The
objective is to develop a model of gasifier behaviour to test
hypotheses concerning the relative importance of certain
properties in determining gasifier behaviour under various
conditions. This can be approached by varying gasifier
conditions and using a knowledge of the change in coal
properties with gasifier conditions such as heating rate,
pressure etc.
 450

The type of information normally available on large scale

gasifiers is as follows:

Standard Performance Data

1. Product gas composition

2. Liquid product yield (tar and liquor)
3. Steam, oxygen and coal consumption

Operational Characteristics

4. Bed differential pressures

5. Gas off-take temperatures
6. Stirrer torque
7. Variations in gas composition

Possible Additional Information

8. Dynamic tests with tracers

9. The use of special probes
10. The digout of the gasifier bed after a controlled shut-
down and the detailed characterisation of samples taken
from specific positions
11. Continuous analysis of representative samples from each
batch of coal entering the gasifier

In contrast, laboratory measurements comprise of the following:

1. Representative analytical data - C,R,N,O,S, etc.

2. Coal behaviour studies under specified conditions
3. Physical properties

The role of gasifier modelling and detailed coal

characterisation is to bridge the gap between the real gasifier
situation where there is much engineering information available,
and sophisticated laboratory measurements where the relevance of
the measurements to a given gasifier is in question. The
measurement of appropriate coal properties under simulated
gasifier conditions in the laboratory is an important
contribution to this research.

DEVELOPMENT PROGRAMME OF THE SLAGGING GASIFIER

The development of the 6' (1.8m) slagging gasifier has been

outlined(IU)previously and was completed in early 1984. This
has involved the gasification of a range of coals and the
 451

(ZSM5 Mobil process

synthetic fuels)

/-(CH 2)n- - Fischer-Tropsch

CO+3H2 synthesis

Synthesis CO+3H CH4 (Substitute natural

Gas 2.
CO+H2 gas manufacture)

RCH=CH+CO+H 2
~RCH~H2CHO - (Hydroformylation
process)

Ammonia synthesis

Fig 5 Applications of synthesis gas in the chemical industry

 452

development of proprietary equipment. Fines, hydrocarbon and

liquor handling techniques have been demonstrated successfully.
The commercial potential of the gasifier has been demonstrated
by a 90 day run consuming over 27000 tonnes of coal.

The next phase of the development will involve the 7'6"

(2.3m) diameter gasifier and will include tests on foreign
coals, a long demonstration run and a demonstration of the RICOM
process as a route to SNG.

An experimental gasifier with a 4' (1.2m) diameter and

capable of being operated at pressures up to 100 bar is also
being designed. This gasifier is seen as an economic way of
investigating various novel gasification techniques and
extending the proven technology to higher pressures.

USES OF SYNTHESIS GAS IN THE ENERGY AND CHEMICAL INDUSTRIES

There are many uses(9) for the product gas from a fixed bed
gasifier and these are given below:

a) After removal of hydrogen sulphide and tar, it is an

excellent fuel for industrial use.
b) It can be converted to SNG by catalytic methanation and
distributed to domestic users.
c) Electricity generation by coal gasification, together with
combined cycle of a gas turbine followed by a steam turbine
has benefits in efficiency as well as environmentally due to
the lower pollution levels over conventional combustion
methods. The rapid response time of the slagging gasifier
to load changes is one of the most important features of
the method.
d) It is an excellent feedstock for the production of
chemicals. Some potential applications of synthesis gas are
shown in figure 5.

CONCLUSIONS

Experience has been obtained with operating the 6' gasifier

over a period of ten years, involving 8600+ hours on running
with a consumption of 100,000 tonnes. It is a commercial
proposition which is flexible in operation. A larger version
(7'6" in diameter) capable of taking 500 tonnes per day is being
commissioned.
 453

ACKNOWLEDGEMENTS

This paper is published with the permission of British Gas.

The author would like to thank his colleagues for many helpful
discussions.

REFERENCES

1. Sharman, R., J.A. Lacey, and J.E. Scott. The British Gas/
Lurgi Slagging Gasifier: A Springboard to Syn Fuels. Paper
presented to the 8th International Conference on Coal
Gasification, Liquefaction and Conversion to Electricity
4-6th August, 1981.
2. Grainger, L. and J. Gibson. Coal Utilization: Technology,
Economics and Policy, (London, Graham and Trotman, 1981),
p.37.
3. Sparham, G. and I.G.C. Dryden. B.C.U.R.A. Monthly Bull. 18,
1963, 1.
4. Rudolph, P.F.H. Coal Chem - 2000. Institute of Chemical
Engineers Symposium Series No. 62, 1980 paper C1.
5. Hebden, D. and R.F. Edge. J.I.G.E. 108 (1958) 492 Gas
Council Research Communication GC50.
6. Hebden, D., J.A. Lacey, and A.G. Horsler, J.I.G.E., 5,
(1965) 367. Gas Council Communication GCl12.
7. Hoy, H.R., A.G. Roberts, and D.M. Williams, Proc.Jt.Conf. on
Gasification Processes, I.G.E./lnst. of Fuel, 1962,C11-C22.
B.C.U.R.A. Report A/CON/2645.
8. Hoy, H.R., A.G. Roberts, and D.M. Williams, J.I.G.E. 5
(1965) 444 I.G.E. Communication 672.
9. Hebden, D. and C.T. Brooks, Westfield - The Development of
Processes for the Production of SNG from Coal, 1976,
Institute of Gas Engineers Communication 988.
10. Tart, K.R. and T.W.A. Rampling, Coal Chem - 2000 Institute
of Chemical Engineers Symposium Series No. 62, 1980 paper
J1.
11. Brooks, C.T. Proceedings of the Third BOC Priestly
Conference Special Publication No. 48, (London, The Royal
Society of Chemistry) 1983, p.78.
 455

COAL PYROLYSIS

J.A. Moulijn and P.J.J. Tromp

Institute of Chemical Technology, University of Amsterdam

Nieuwe Achtergracht 166
1018 WV Amsterdam
The Netherlands

In most gasification research model materials are gasified, viz.,

graphite, activated carbon or coal char. In many respects these
model compounds fundamentally differ from "real" coal. The major
difference is that upon heating coal undergoes a so-called
pyrolysis step. It is the aim of this chapter to introduce this
subject and to discuss the relevant aspects.
The name "pyrolysis" stems from Greece: IIup (gen.: IIupocr) means
fire and AU€iV (noun: Aucricr) means loose, untie. So, pyrolysis
means decomposition due to exposure to high temperature.
The processes involved are very complex and highly dependent on
the type of the coal. Coal is a naturally-occurring organic rock
and, therefore, it is not surprising that "coal" is not a unique
indication. There are many types of coal; table 1 gives a classi-
fication.
Because of the practical relevance of pyrolysis phenomena, coals
traditionally have been divided according to their behavior upon
heating, namely in coking and non-coking coals. Analogous
designations are agglomerating and non-agglomerating, caking and
non-caking, and plastic and non-plastic coals. A coal belongs to
the second category if it exhibits thermoplastic behavior, i.e.,
it becomes viscoelastic upon heating while simultaneously often
swelling is observed; upon prolonged heating resolidification
takes place. The temperature a coal becomes plastic strongly
depends on the experimental conditions, especially the heating
rate and the particle size. At low heating rates « 1 K/min)
thermal softening is typically observed in the temperature range
of 325-450 oC, whereas at high heating rates (> 103 -10 4 K/s)
softening occurs above 1500 C. The physical and chemical changes
a caking coal can undergo are tremendous. A typical example is
given in figure '1, which shows scanning electron micrographs of
456

T~bh, 1. Classification of codis (after Elliot and Yoke (I) )

1-_--'A.;:S:.T:,:M.:..:....C=18;,:SS.:.i_fi..:,C8.:.t.:.io.:.n_......_ _--1lnternational Scientific Properties of Vitrinite

Volatile Btullb CI8ssificatia Elemental Volatile MOisture MJ/kg Raflec!~
Class Group Matter MOist Stages Rank Carbon Matter 03aturatedl _A of Vitrinite
DMMF" MMF"" OAF", DAF'C wl'4 MAF"iI '4 ~

Peat

- - - Undeftned _ _ _ 60 I- 75
B
Soft Brown
lIg"'tlc Coal
A ca53 f- 35 16.8 caO.3
8300
C Hard Dull
Sub- 9500 9 Brown
B
Bttumlnous 10500 Coal - ca 71 ca49 25 23
A
11500'·"q.-----1 Bnght
C 8
HIgh I- 13000 -I-____-I-'TT-~ca77 ca42 8-10 29.4 caO.S

f,!oIatile B 7
I-- 14000 4------1 '"Low
BituminOUS >31 !!
Hard Rank
A 6
>14000 Coal.!!'

Medium
31 -
l:::=::s:::==1 ~ f-- 87 29 36.3 1.1
Volatile 4 ~
Low
22 - i High
3
Volatile
14 -
~ Rank
2
8 -
f 91 8 36.3 2.5
Anthractic

Anthracite Anthracite
o

• Dry, mineral-free basis, weight percentage

"" Mineral-free. but with saturated pore moisture content
": Dry. ash-free basis, weight percentage

*
':-Moist. ash-free
Bituminous if agglomerating. sub-bituminous if non-agglomerating
 457

a e

b f

c g

d g

Fig. 1. Scanning electron micrographs of a medium-volatile bitu-

minous coal. The coal particles are mounted on a ferromagnetic
wire and are heated according to the Curie-point tecbni~e. In 0.1
the final temperature is reached and the samples are herd at this
temperature for 4 s. 0
a, e: after heating to 510 C; these micrographs are identical with
those of part1cles that have not been heated
b, f after heating to 6100 C
c, g after heating to 7700C
d, h after heating to 9BOoC
 458

coal particles exhibiting softening and swelling phenomena. These

particles were heated in N2 according to the so-called Curie-point
technique.
Upon heating up to 510 0 C no changes are visible, but it should be
stressed that this is dependent on the heating rate. In separate
experiments, at much lower heating rate « 15 K/s) ,softening and
swelling took place already at 450 o C.
Figure lb and if show that during heating to 610 0 C softening has
occurred without marked swelling. After heating to 770 0 C the ~ar­
ticles clearly have expanded. Some of them form large hollow
spheres (called cenospheres) .
Upon heating to 980 0 C resolidification occurs with collapse of the
big particles, probably due to the fact that complete devolatili-
zation has taken place.
Pyrolysis has drawn a lot of attention. The subject has been
reviewed extensively, see e.g. van Krevelen in his classical text
(2), and, more recently, Howard (3), van Heek (4), Furfari (5),
and Gavalas (6).
The pyrolysis literature can be divided into studies involving low
and those involving high rates. Early pyrolysis studies were aimed
towards improvement of the manufacture of coke. Therefore, the
conditions were close to those in this process: low heating rates,
densely packed beds and an inert atmosphere (1 bar). More recently,
high heating rates have been applied, because this is more realistic
for many coal utilization processes:
- in combustion, powdered coal combustors are widespread; in these
combustors basically coal particles are injected in a flame and,
as a consequence, are heated up within a time scale of typically
1 S;
- in gasification, in many processes coal particles are heated up
at high rates, e.g., in entrained-flow reactors as well ~s in
fluidized-bed reactors heating rates generally exceed 10 K/S.

Chemistry of coal pyro~sis

In order to describe pyrolysis in any detail, a complex network of

reactions, in combination with simultaneously occurring physical
changes, has to be analyzed and understood. In fact, considering
the state-of-the-art, a good description is only possible to a
limited extent. The chemistry of pyrolysis will be discussed on
the basis of the following subjects:
- (slightly speculative) coal models
- results of pyrolysis of model compounds
- analysis of coal pyrolysis products, viz., gas phase compositions
and char compositions.

A macroscopic view of pyrolysis

It is generally agreed upon that a bituminous coal can be described
as a network of macromolecules with trapped molecules in it
(2, 7, 8). Many authors distinguish a "mobile" and an "immobile"
 459

phase. The mobile phase consists of (part of) the trapped molecules
and fragments of the network.
During pyrolysis bonds are broken, the originally-solid material
softens, molecules escape to the gas phase and finally resolidifi-
cation takes place. The processes occurring are, at least partially,
heat- and mass-transfer controlled, especially at high heating
rates. From the literature it is concluded that the chemistry
involved is mainly radical chemistry. In a macroscopic sense
pyrolysis can be envisaged as illustrated in figure 2.

Figure 2. Schematic illustration of coal pyrolysis (4)

The native coal consists of lamellae with more or less flat aromatic
planes, interrupted by aliphatic protrusions; also many functional
groups are present (see below). The lamellae are cross-linked and
also held together by van der Waals forces.
Upon heating, many bonds undergo thermolysis and the planar units
- that can be conceived as the building blocks of the coal - can
slide over each other, more or less freely. This is a liquid state
in which the coal exhibits viscoelastic properties. The initially-
trapped molecules are now free and can act as lubricants. The
460

fragments, produced by thermolysis reactions, can have the same

function. When the heating is prolonged, gas bubbles are formed in
the viscous mass, causing swelling. This can be compared with the
rise of dough. Swelling of the particles will occur if bubble
formation and fluidity take place simultaneously. After dis-
appearance of the gaseous molecules, immobilization will result
due to a lack of lubricants and the occurrence of condensation
reactions.
From the model described above, softening and swelling phenomena,
in principle, can be understood. However, much more detailed infor-
mation is needed in order to predict pyrolysis phenomena observed
in practice. For instance, why is it that only coals of inter-
mediate rank exhibit thermoplastic behavior? Which processes occur?
What is the effect of pressure, of the composition of the gas phase,
of inorganic additives?
The starting point for any discussion, aimed to understand pyro-
lysis phenomena on a microscopic scale, should be the chemical
constitution of coal. E.g., the chemical structure of the lamellae
in figure 2 should be known as well as the nature of the cross-
links between these lamellae. Although the chemical constitution of
coal has been studied by a large number of scientists, it is still
far from understood. In the present chapter, without much discus-
sion, some models from literature are given. As pyrolysis behavior
is highly dependent on the rank of the coal, it is obvious that
models have to be used, which adequately describe characteristics,
typically for the rank.
Lignites and subbituminous coals are the geologically younger coals
and, as a consequence, it is to be expected that they have struc-
tural characteristics of their precursors. Indeed, it has been
found that they contain lignin-like polymers (8). Figure 3 gives
a typical structure of lignin. However, even a lignite will not be
described exactly by this structure. Moreover, apart from the in-
trinsic heterogeneity of coal (there are several different precur-
sors), during the coalification many reactions take place, including
dehydration, decarboxylation and demethoxylation. Consequently,
dependent on the extent these reactions have taken place, the struc-
ture will be different and, in particular, the number of functional
groups will be less.
Nevertheless, as will be shown at the end of this chapter, the model
of lignin is a useful model in understanding pyrolysis products for
a subbitumino~s coal.
For bituminous coals many models have been proposed. It is generally
agreed that these coals contain highly-substituted aromatic ring
clusters, also containing hetero-atoms (O,N,S). These clusters are
linked together by relatively weak bridges, e.g. aliphatic or ether
bridges.
In figure 4 a model is given that recently has been proposed by
Shinn (10) for a high-volatile bituminous coal. The high-volatile
coal Shinn considered for his model had a stoichiometry of
C661HS61074Nl1S. It should be stressed that the stoichiometry of
 461
coals differs profoundly. E.g., the medium-volatile coal studied in
figure 1 has the stoichiometry C661H410013N13S2. Obviously, that
coal has, compared to the coal considered by Sninn, much less OH-
groups and probably more ether bridges, due to dehydration of OH-
groups.

Eig. 3. Structure of lignin

Figure 4. Model of the chemical structure of a

high-volatile bituminous coal (10)
462

Model Compounds
In coal pyrolysis the dominant chemistry is radical chemistry. All
types of radical reactions occur, but in specific cases also con-
certed reactions are possible, e.g., the decomposition of phenol.
Much can be learned from experiments in which model compounds are
pyrolyzed under well-defined conditions. Very interesting results
have been published by Cypres and coworkers (11-14) on pyrolysis
of many relevant compounds including phenol and cresols. They found
that in phenol thermolysis (1000-1150 K) inter- and intramolecular
reactions take place. Apart from many secundary reactions they
propose the following scheme:

We undertook a systematic study of the thermal stability and the

product distributions arising from heat treatment of relevant
classes of model compounds (15). Figure 5 gives a schematic of the
apparatus used.

THERMOSTATED DIFFUSION CELL FURNACE

FTIR
15~o~t__-r__~GASCa
:~~~~__
VENT
~ __~~~~~+-+-~~ COILED TUBE
110C

REACTOR

GC NT
COMPOUND UNDER STUDY

50ml/min

250ml/min

Ar
•
Figure 5. Schematic of apparatus for studying pyrolysis of vola-
tile compounds.
 463
3
The reactor used was a coiled tube (volume 17 cm ). The temperature
of the diffusion cell is chosen as such that the concentration of
the model compound was 500 ppm {50 Pal and the residence time was
5 s.
Figures 6-8 give typical results.

100

80
Xlmol%
I
60

40

20

O~--P-----~------~------~------~---
773 873 973 1073 1173
T/K-

Figure 6. Thermal stability of benzene and benzene derivatives

(residence time 5 s, partial pressure 50 Pa, balance
Ar) •

In figure 6 data are shown on the thermal stability of benzene and

derivatives of benzene. Clearly, benzene is the most stable one,
as should be expected. Functional groups result in lower stabili-
ties, dependent on the type of functional group. The interpretation
in many cases is rather straightforward: the lower the binding
energy, the less stable a bond will be. E.g., the relative stability
464

of the (alkylated) benzenes can be understood from the binding

energies given in table 2.

C6 HS - H 428 kJ/mol

C6 HS - CH 3 365 kJ/mol

C6 HSCH2 - H 326 kJ/mol

C6 H5 CH 2 - CH 3 265 kJ/mol

Table 2. Binding energies of some single bonds

From this table it can be explained that the stability of (alkyl)

benzenes decreases in the order:

benzene > toluene > ethylbenzene

Figure 6 shows further that methoxy benzene has the lowest stabi-
lity, whereas fenol is relatively stable and does not decompose
below about 1000 K.

Coals contain many "hetero-atoms" and, as a consequence, also model

compounds containing the most abundant hetero atoms, being O,S,
and N, are relevant. From literature it is clear that 0 is present
in the form of functional groups (especially OH and COOH), in
aromatic structures (furans), and in mineral matter (carbonates,
sulfates, oxides). Sulphur is present in functional groups (thiol,
mercapto), in aromatic structures (thiophenes) and, of course, in
mineral matter (pyrite, sulfates). Nitrogen is present in aromatic
structures and, perhaps, in amino- and cyano groups (16).
Figure 7 focuses on oxygen. The thermal stability of fenol and
several substituted fenols are shown. Fenol is the most stable one.
Cresols are less stable and, as has also been shown by others (13),
ortho-cresol exhibits the lowest stability.
Substitution with CH 3 -group s decreases the stability, especially
in the ortho-position. The stability decreases in the order:

phenol> m,p-cresol > O-cresol > 2,4-xylenol

A similar conclusion has been drawn by Cypres and coworkers (12).

When the side-group is an aliphatic alcohol the stability is ~uch
less than fenols.
 465

100 I
,
i
I
i "
/
0'

80 i /
.I 'H
'"
j "
i
!
X/mol% / /,©I
.'fa
1'0 p~
f

I i i " f
i

60 i ~I
i "
•i ," !
!
/ rry/
:

3! " CH
!
i l8J i,' I ., :
f
~!
~ / CH3!f

!! rry
I, I:
40
hyH2 /! l8J,
,,
.' I
.I
@/
I ,
i /
20 / ,; /
i // ... ""/
/ /'
/,1 "
,' ......
....

/ • I ••·•

o~~----~~----~------~----~~-
773 873 973 1073 1173
-T/K-
Figure 7. Thermal stability of (substituted) fenols and
ethanolbenzene (residence time 5 s, partial
pressure 50 Pa, balance Ar).

Figure 8 summarizes data on the thermal stability of aromatic com-

pounds, focusing on the influence of hetero-atoms within the ring
st.ructure. The most stable ones are (benzo) thiophene, benzene and
naphtalene, whereas furan exhibits the lowest stability. The
results of figure 8 can be summarized in the following sequence in
decreasing stability:

(benzo)thiophene > benzene, naphtalene > pyridine> indene,

benzofuran, pyrrole > furan

A rough explanation can be given considering aromatic stabilization.

Recently, detailed information has been published on the thermal
466

stability of bibenzyl (17). At 650-700 K thermolysis reactions take

place. However, it should be noted that, because these studies were
directed towards liquefaction, the residence time in these experi-
ments was 15-1200 min. Therefore, i t is not surprising that the
reported temperature range in which they found C-C cleavage, is far
below the temperature range we found for ethylbenzene (> 900 K),
that is less stable than bibenzyl.
In literature also very significant results have been reported on
hydropyrolysis of model compounds (17-20). These will not be treated
here.

100
©o
/©D
I ,,
I
I

,,, I
/lQ)
.
80 N

i
I • I

Xlmol% I
I
I
? I
I

/00
I
I
I
I

60
/ !//~/©o
I
I
I
I

II 01
I
I

ii
I
.. ,
1/
I
I
I
I :! " I·
i
I I
40 I
.' !,' I·

/
I

, I
I !,' .
,,, I
I
.' !,'
.' i
, I·
I

.' !,'
, .' I.
i,
20 I
I
"
il
,. ,f' /. 'n
I
1[1,' 1[1/
«NY:

"I. /
'(0» " , ' I • '(5»
I
I
I H ..
...:
.. ,
I
I
...
"
o~------~------~------~------~--
973 1073 1173 1273

T/K
Figure 8. Thermal stability of (heterocyclic) aromatics.
 467

Product distributions in coal pyrolysis

Product distributions are important for two reasons. First the
economics of commercial processes strongly depend on the type and
amount of products.
Secondly, product distributions give fundamental information on the
processes occurring during pyrolysis; in fact, they also considerably
help in better understanding the structure of coals.
A typical plot of the product distribution as function of the tem-
perature is given in figure 9, for a lignite at high heating rate.

0
UJ
~
UJ
0
UJ
a:::I
(/)
« 40
"e
0

~
'-
0 30
~
UJ
>-
t-
O 20
~
0
0
a:::
0..
10

O~ __. . ~~~ __. .~~ _______

273 673 1073 1373
- PEAK TEMPERATURE / K -

Figure 9. Pyrolysis product distributions from Montana

lignite heated to different peak temperatures
(heating rate> 1000 K/s) (after Suuberg (22) ).
468

Up to about 700 K only water is produced. Subsequently the evolution

of CO 2 and tar starts. At slightly higher temperatures CO hydro-
carbons and H are observed. The hydrocarbons mainly consist of
methane and e€hylene. It is interesting that the production curves
exhibit a two-step behavior: CO 2 evolution increases enormously at
ca 700 K, while, subsequently, lt levels off to an asymptotic value,
but at ca 1050 K suddenly an increased production takes place. CO
and the hydrocarbons show analogous behavior. From a detailed
analysis (not given here) it appeared that this does not apply to
H2 : evolution starts at 800 K and it increases continuously up to
aBout 1150 K. It is worthwhile to note that in our own experiments
it appeared that at still higher temperatures extra H2 production
is observed.
Figure 10 gives the product distributions for a bituminous coal.

0
l.U
> 40
W
0
l.U
II:
I
Cf)
« 30

-
;,!i2
0

~
.......
0
...J
l.U 20
>-
I-
0
:::>
0
0 10
II:
c..

O~--·I--"-P--~~~--~----
673 873 1073 1273

- PEAK TEMPERATURE - - -

Figure 10. Pyrolysis product distribution from Pittsburgh

Seam bituminous coal heated to different peak
temperatures (heating rate > 1000 K/s) (after
Suuberg (22) ).
 469

Figures 9 and 10 give results which can be compared directly with

each other. It is obvious that the bituminous coal behaves quite
different than the lignite. Whereas in the former case the product
distribution is dominated by H2 0,CO and CO 2 , in the case of the
bituminous coal tar is by far the most important product. Figure 10

°
shows that tar production takes place over the whole temperature
range. H is formed at relatively low temperature, viz. below
700 K, w~ereas at higher temperatures no additional water is formed,
in sharp contrast to the case of lignite.
Analysis of the chars produced in pyrolysis gives information,
complementary to chemical analysis of the gaseous products.Figures
11 and 12 give results for a lignite and a bituminous coal.

UJ
::::J
...J
100
<{
>
...J
50 c
<{
Z o
~ 100
a:
o 50 -
LL H ~
o o
100 ...............
~o
50 N

-
ci
::;:
o o

-
() 100
...J

~ 50 S

-
Z
UJ
::;: o
UJ
...J 100
UJ
50 -
o
O~ l-
I I I I

873 1073 1273

PEAK TEMPERATURE/K

Figure 11. Elemental composition of lignite char for

different peak temperatures (heating rate
> 1000 K/s) (after Suuberg (22j" ) .

In this figure the retention in the char is shown for C,H,N,S and
0. It is clear that the char becomes relatively enriched in C:
even at the highest temperature only about 20% of the original
 470
carbon is lost. The reason, of course, is that the volatiles con-
tain large amounts of H2 0 and H2 , while also the evolution of
molecules like CH 4 and CO 2 lead to enrichment of C. The decrease
of H compares well with tne foregoing. At low temperature H2 0,
which is adsorbed or trapped in the coal, is evolved, whereas at
higher temperatures defunctionalisation takes place. N-loss does
not start before 1000 K and even at the highest temperature this
does not exceed 30%. S-loss takes place over the whole temperature
range. Compared to N less S remains behind in the char.
Q-volatilization exhibits two stages: at 700 K already about 40%
of the 0 is lost, obviously due to H 2 0- and co 2 -evolution; at higher
temperature oxygen volatilization continuously takes place, while
at about 400 K a sudden increase in O-loss occurs. At 1300 K vir-
tually no oxygen is retained in the char.

Figure 12 gives analogous information for a bituminous coal. Com-

paring these results with the previous ones, it is clear that

W
::J
...J
<t:
> 100 ~2tOl0S
Isothermal
50 - C
...J
<t: '!!.1~3~
z 50%
Q 0-
a:
0 100 -
~
u.
50 -
0
#-
>=
a: 0- ---
11%

0 100 -
z
Q
I-
50- ~ -64 0;

iii
It:2: 0-
-
0
(,)
...J
100
50 - S -- ---
45%
~
z 0
w
:2: 100
w
...J
W 50 Z.
0 - t I I I I I
---
0"

673 873 1073 1273

- PEAK TEMPERATURE/K -

Figure 12. Elemental compositions of bituminous coal char

for different peak temperatures (heating rate
1000 K/s); the dashed lines represent condi-
tions at 1273 K during 2-10 s (after Suuberg
(23) ).
 471

especially the C-profile is fundamentally different. Up to 50% of

the carbon, originally present, is volatilized eventually. It is
interesting that upon keeping the sample at 1275 K, additional
H-loss takes place, whereas the C-content of the char does not de-
crease further. This indicates that more extended poly-aromatic
hydrocarbon structures are formed.
Over the whole temperature range the carbon content decreases con-
tinuously. The H-profile shows an analogous pattern, which is in
good agreement with the predominant evolution of tar (figure 10).

Interpretation
Although a detailed analysis is not yet possible, a rough interpre-
tation of the chemical reactions involved in pyrolysis can be given.
At the lowest temperatures desorption takes place of molecules,
held by physi- or chemi-sorption forces. At higher temperature
pores are formed and, consequently, trapped molecules can escape
to the gas phase. These processes are more or less of a physical
origin, except that the porosity at higher temperature is generated
by chemical reactions. Therefore, it has to be expected that trapped
molecules can be evolved up to relatively high temperature.
At about 700 K (heating rate < 1000 K/s), in the case of lignite,
CO 2 is observed. This is associated with decomposition of carboxylic
groups. In agreement with this, at these temperatures CO 2 is not
observed for bituminous coals, obviously because they do not con-
tain significant amounts of carboxylic groups.
At slightly higher temperatures for lignites H2 0 evolution dominates.
The major reaction probably is dehydration of OR-groups, leading to
ether bridges.

Whereas for lignites a-containing products dominate, for bituminous

coals tar and hydrocarbons are the major species observed. This is
hardly sUrprising, considering the fact that bituminous coals con-
tain much less 0 than lignites. In bituminous coals the initiation
reactions probably are mainly C-C bond breaking reactions, because:
- modelcompounds, containing C-C groups decompose at about 900 K,
the temperature above which extensive tar evolution occurs;
- bituminous coals contain many C-C bonds (figure 4).

At very high temperature reactions take place forming H2 and

products containing Nand s. This can be well understood from the
model compounds studies. The aromatic ring clusters containing
these hetero-atoms are very stable and decompose only at high
temperature, giving probably highly aromatic structures of the P~B
(Poly Aromatic Hydrocarbons) type. Furans have the lowest stability
and, consequently, react most extensively. N- and S- aromatics are
more stable. Therefore, it can be explained that chars heated to
high temperature do not contain 0, but do contain considerable
amounts of Nand S.
472

It is noteworthy that the C/H ratio of chars produced at high tern -

perature is in the order of 8,suggesting a higher degree of conden-
sation than PAH.
In this interpretation the initiation part of the radical chains is
emphasized. Of course, also secondary reactions have to be considered
It is well-known that in using packed beds or applying the captive
sample technique (6), volatile yields are lower for larger sample
sizes. The reason is that secondary reactions are favoured in the
case of large samples:
- the heating time is larger;
- residence time in the sample bed is larger.
In the same way it can be understood why large particles give lower
volatile yields than small particles.

Effect of inorganic additives

Inorganic materials present in the coal or added to it, can drasti-
cally influence the pyrolysis behavior. E.g., it has been reported
that K1 C0 3 and KOH reduce swelling and caking of several coals (24).
Also tne composition of the gas phase is changed substantially
upon addition of (earth)alkali cations (25). For a bituminous coal
it is shown in the latter study that alkali ions significantly
reduce the yield of hydrocarbon volatiles and tar under flash pyro-
lysis conditions. Wigmans et al. (26) reported that K2 C0 3 shifts
the onset of pyrolysis to lower temperatures and leads to an in-
creased oxygen retention (the analysis was based on TGA measurements)
In the following some results will be given of a study of the in-
fluence of K2 C0 3 and Na 2 co 3 on the pyrolysis product spectrum of
flash pyrolysis (27).
The techniques used were Curie-point mass spectrometry, gas chro-
matography, and gas chromatography/mass spectrometry.
Because it has to be expected that functional groups able to bind
alkali ions playa major role, two coals with highly different
contents of oxygen were studied, viz. a subbituminous and a bitu-
minous coal.
Figure 13 gives a schematic of the Curie-point mass spectrometry
instrument.
A detailed description of this technique is given in the literature
(28) •
Figure 14 shows the pyrolytic mass spect~a of the native coals
pyrolyzed with a peak temperature of 510 C.
 473

....bbll........ ,51OC

,alkene'S
\
\
\
\

HVpump I
(ion ~1I.r)

Figure 13. Schematic of the Curie-point mass spectrometry

instrument.
474

subbituminous ; 510"C

;iI
>=
I-
en
Z
W
I-
~
z
Q
...J

o«
l-
I

-m/z_

18
,\
~ , \
1 \\
i
16
! \,\
!
bituminous; 51dc

i \
i
i i
;iI i
i \ alkanes
>= i
i
i
!::: 12
~
i
i
\i
w i "
i i
~ i i
zl
Q
i
i
i
\i
i i

g8 ! i
...J
I
i ,
\
I- i \

I
i
i
i
i \,
i \
6 i
i
i
\\
i
i ,
4
i
i \i
! \

20 40 60 80 100 120 140

-m/Z-

Figure 14. Pyrolysis mass sgectra of the native coals. Final

temperature (510 C) reached in 0.1 s, total pyro-
lysis time 0.8 s. Upper part: subbituminous coal,
lower part: bituminous coal.
 475

It is clear that the spectra are profoundly different. The spectrum

of the subbituminous coal shows many oxygen-containing compounds
(H 2 0,C0 2 ,CO, alkylated phenols, dehydroxybenzene and methoxyphenols),
whereas the spectrum of the bituminous coals is dominated by alkanes
and naphtalenes.
Figure IS gives the pyrolysis mass spectra for the coals investi-
gated by the same technique, but with a peak temperature of 770 o C.
In comparison with the S10 0 C experiment, the spectrum of the subbitumi-
nous coal shows an increased production of CO,CO and alkenes,
whereas the amounts of dehydroxy benzenes and methoxybenzenes are
lower, relatively.
While the spectrum of the subbituminous coal after pyrolysis at
770 0 C has similar characteristics to the low-temperature one
(SlOoC), the spectrum of the bituminous coal is completely different
than the low-temperature spectrum. Clearly, between SlO and 770 0 c
extensive pyrolysis reactions take place.
It is striking that, in the higher region, the spectrum of the
bituminous coal is dominated by benzenes/naphtalenes, whereas the
spectrum of the subbituminous coal shows mainly oxygen-containing
aromatics.
To get more information on the origin of the alkanes, observed in
the SlOoC spectrum of the bituminous ~oal, experiments were carried
out at a lower temperature. Figure 16 gives results of Curie-point
pyrolysis gas chromatography with a peak temperature of 3SS o C. It
is clear that C o-C S n-alkanes are already evolved at this low
temperature, in~lcatlng that this process can best be described as
desorption of trapped molecules. The same experiment, performed
with the subbituminous coal, revealed that essentially no products
were formed, except minor amounts of alkanes/alkenes (> C 29 ).
In the case of the sUbbituminous coal at SlOoC already pyrolytic
reactions occur, in which H2 0,C0 2 ,CO,H 2 S, alkanes, alkenes, dienes,
phenols, dihydroxybenzene, and methoxynenzenes are formed. From
the model of lignin (figure 3) and the results of the model com-
pounds stability study, this product distribution can roughly be
understood. At SlOoC part of the aliphatic C-C and c-o bonds will
be broken, and consequently (alkylated) phenols, methoxybenzenes,
and dehydroxybenzene are not unexpected.
The H2 S is probably formed by decomposition of thiol- or mercapto-
groups and not by cracking of thiophenic structures, because these
are extremely stable. The alkenes observed must be generated by
cracking of alkyl substituents, in agreement with the results of
the stability studies with model compounds.
In the case of the bituminous coal the SlOoC spectrum is dominated
by desorption of alkanes, whereas between SlO and 770 0 c pyrolytic
reactions occur. This is well understood from the model compounds
stability studies: aromatic compounds containing alkyl (~C2) or
aliphatic alcohol side chains exhibit decomposition reactions at
temperatures from SOOOC (residence time S s). The total pyrolysis
time in the Curie-point pyrolysis experiments was significantly lower,
476

subbituminous ; 770°C

alkenes

-m/z-

cp
bituminous ; 770'(;

Figure 15. Pyrolysis mass sgectra of the native coals. Final

temperature (770 C) reached in 0.1 s, total pyro-
lysis time 0.8 s. Upper part: sUbbituminous coal,
lower part: bituminous coal.
 nCf15

BITUMINOUS; 358'c
~

.-- I -- .-- I I I
50 100 150 200 250 300
-T/'C--

Figure 16. Pyrogram of the bituminous coal with a final temperature of 35SoC. Final
temperature reached in 0.1 s, total pyrolysis time 4 s.
.".
-...J
-...J
 478

C92

subbituminous +20wt% K2co3 : 770t

alkenes

160
-m/z-

bituminous+20wt% KA: 770t

20 40
-m/z_
Fig. 17. Pyrolysis mass spectra of the coals after addition of 20%
K2 C0 3 · Final temperature (770 0 C) reached in 0.1 s, total pyrolysis
0.8 s.Upper part: sUbbiturninous coal, lower part: bituminous coal.
 479

but not orders of magnitude lower (0.8 s). Therefore, it is reaso-

nable to expect that C-C and c-o cleavage in these Curie-point
pyrolysis experiments take place somewhat above SOOoC.
It is remarkable that for the subbituminous coal pyrolytic reac-
tions already at S10 0 C are observed. The reason probably is that
this coal contains a large number of oxygen-containing functional
groups that crack easily under formation of relatively large
amounts of radicals that accelerate the radical reactions involved
in pyrolysis.
It is interesting that in both coals no NH is observed. This in-
dicates that N is only present in condensect aromatic structures,
because these have been shown to be extremely stable.
The influence of K2 C0 3 is shown in figure 17, that gives the pyro-
lysis mass spectra for both coals after addition of 20% K2 C0 3 .

Comparison with the spectra of the native coals (fig. 15) reveals
that the spectrum for the sUbbituminous coals is profoundly dif-
ferent, whereas the spectrum for the bituminous coal is similar.
The subbituminous coal will be discussed first. The peaks corres-
ponding with H2 0 and CO 2 strongly increase, whereas the peaks due
to dihydroxybenzene and to a lesser extent due to the phenol series
decrease. To understand these results better, the same experiment
was repeated in the Curie-point gas chromatography mode. Figure 18
compares the pyrogram of the native coal with that of the coal with
K2 C0 1 . These results can be summarized by:
- (alkylated) benzenes, alkanes, alkenes, are not influenced;
- (alkylated) phenols, dihydroxybenzene, and methoxyphenols, except
ortho-methoxy phenol, are drastically suppressed.
At first sight it is surprising that ortho-methoxy phenol is not
influenced. The explanation probably lies in the fact that only
functional groups that can react with alkali ions are influenced.
Ortho-methoxyphenol probably is formed from structures of the
type (compare the structure of lignin, fig. 3):

OMe
o
~
Alkali-ions do not interact with ether linkages and, as a conse-
quence, the formation 0"" ortho-methoxyphenol is not influenced by
alkali-ions. Similarly, it can be explained that the phenols,
dihydroxybenzene and the other methoxyphenols are influenced,
because they are probably formed from functional groups that can
react with alkali-ions.
480

Comparison of fig. 15 and 17 shows that in the case of the bituminous

coals only minor changes are observed: H S,CO and H2 0 slightly in-
crease upon K2 co 3 addition, while the ot~er ploducts, including the
phenols, are not influenced. To check this conclusion, Curie-point
gas chromatography was also in this case applied. Figure 19 shows
the pyrogram for the native coal; the pyrogram for the coal after
addition of K2 CO is not given because it is not significantly
different from t~e one presented for the native coal. These chroma-
tograms not only confirm the conclusion that K C0 3 has no influence,
but they give also interesting information on the pyrolytic reactions
that occur: the only oxygen-containing products that are observed
are (alkylated) phenols. This can be understood from the chemical
constitution of the coal. In comparison to the model of the high-
volatile coal given in figure 4 the subbituminous coal in the study
descr~bed here (C G8 H425018N13S) contains much less -COOH and -OH
functlonal groups. ~herefore, the phenols observed probably are formed
from the decomposition of ether linkages directly bonded to an aro-
matic ring. This also explains why they are not influenced by alkali-
ions, because these do not interact with ether-linkages.

Acknowledgement
Thanks are due to Dr. J.J.Boon (FOM, Amsterdam) for his kind permis
sion to use the Curie-Point Temperature Pyrolysis instruments and
his help in the interpretation of the results.
 ~
y @a
@ C$ @I SUBBITUMINOUS : 77r1C
'~II~? ex><,
@ @'
I . .
@ I I I
, ? "'I ....,;"""", .... ~
'@
J§1,
@' ""33.

o 50 100 150 200 250 320

-T/C---
@ @
,
~ SUBBITUMINOUS+20wt%~ : 770·C

I I I I
"!?01 "'/""""'''~

o 50 m 250 320
-TlC--
."..
Figure 18. pyrograms for the subbituminous coal with and without addition of K2C0 3 . Final 00
temperature (770 o C) reached in 0.1 s, total pyrolysis time 4 s.
.....
 .p.
(Xl
N

@ I
I I
@ I@ BITUMINOUS; 770t
I

@'
I
tlI
,@
tl
rro u 1.
I
@'

~,~
I I .- I I -------.
o 50 100 150 200 250

-T/t-

Figure 19. Pyrogram of the bituminous coal. Peak temperature (770°C) reached in 0.1 S,
total pyrolysis time 4 s.
 483

References

1. Elliot, M.A. and G.R. Yoke, in: "Chemistry of Coal Utilization",

sec. suppl.vol. (ed. M.A. Elliot), Wileg, 1981, p. 1.

2. Van Krevelen, D.W., "Coal: Typology, Chemistry, Physics and

Constitution", Elsevier Sci. Publ. Co., Amsterdam 1981.

3. Howard, J.B., in: "Chemistry of Coal Utilization", sec. suppl.

vol. (ed. M.A. Elliot), Wiley 1981, p. 665.

4. Von Heek, K.H., "Druckpyrolyse von Steinkohlen", VDI-Verlag,

Dusseldorf, Germany, 1982.

5. Furfari, S., "Hydropyrolysis of Coal", lEA Coal Research,

London 1982.

6. Gavalas, J.R., "Coal Pyrolysis", Elsevier, Amsterdam 1982.

7. Given, P.H., in: "Coal Science", vol. 3 (eds. M.L. Gorbaty,

J.W. Larser, I. Wender), p. 65, Academic Press, 1984.

8. Davidson, R.M., in: "Coal Science", vol. (eds. M.L. Gorbaty,

J.W. Larren, I. Wender), p. 84, Academic Press, 1982.

9. Spiro, C.L., "Space-Filling Models for Coal: A Molecular Des-

cription of Coal Plasticity", General Electric, Schenectady,
N. Y ., 1981.

10. Shinn, J .H., Fuel .§l, 1187 (1984) .

11. Cypres, R. and B. Bettens, Proc. Int. Conf. Coal Science, p. 204,
Pittsburgh, U.S.A. , 1983.

12. Cypres, R. and B. Bettens, Tetrahedron 30, 1253 (1974) .

13. Cypres, R. and B. Bettens, Tetrahedron 31, 353 (1975) .

14. Cypres, R. and B. Bettens, Tetrahedron 31, 359 (1975) .

15. To be published.

16. Attar, A. and G.G. Hendrickson, in: "Coal Structure" (ed.

R.A. Meyers), Academic Press, N.Y., 1982, p. 132.

17. Pontsma, M.L., Fuel 59, 335 (1980).

18. Grader, W.-D. and K.J. Huttinger, Fuel §l, 499 (1982).

19. Grader, W.-D. and K.J. Huttinger, Fuel §l, 505 (1982).
484
20. Grader, W.-D. and K.J. Huttinger, Fuel §!, 510 (1982).

21. Penninger, J.M.L. and K. Versluis, Fuel §!, 209 (1982).

22. Suuberg, E.M., W.A. Peters and J.B. Howard, Ind. Eng. Chern.,
Process Des. Dev. 12, 37 (1978).

23. Suuberg, E.M., Sc. D. thesis, MIT, 1977, cited in ref. 3.

24. Kan, M.R. and R.J. Jenkins, Proc. 1983 Int. Conf. Coal Science,
p. 495, 1983.

25. Franklin, H.D., R.G. Gosway, W.A. Peters and J.B. Howard, Ind.
Eng. Chern. Process Des. Dev. ~, 39 (1983).

26. Wigmans, T., H. Haringa and J.A. Moulijn, Fuel 63, 870 (1984).

27. Tromp, P.J.J., J.J. Boon and J.A. Moulijn, to be published.

28. Meuzelaar, H.L.C., J. Haverkamp and F.D. Hileman, "Pyrolysis

Mass Spectrometry of Recent and Fossil Biomaterials", Elsevier,
Amsterdam 1982.
 485

HYDROGASIFICATION OF BROWN COAL - CHEMISM, KINETICS AND

TECHNICAL ASPECTS

Klaus J. Hlittinger, Alexander Michenfelder

Institut flir Chemische Technik

University of Karlsruhe
Fed. Rep. Germany

1. INTRODUCTION
Brown coals and lignites are known to be the most re-
active coals and are therefore the preferred materials
for hydrogasification. Birch et al. (1,2), Wen and
Huebler (3), and Blackwood and McCarthy (4) studying
hydrogasification of brown coals or lignites in fluidized
bed reactors at various temperatures and pressures, showed
that gasification is occurring in two stages. In the first
stage up to a few minutes gasification rate is very high
and caused by pyrolysis and hydropyrolysis reactions of
the coal. The lower gasification rates in the second
stage are ascribed to hydrogasification of the residual
coke from stage one. The authors also found that increa-
sing the hydrogen pressure has the most significant ef-
fect on the yields of volatiles, ethane and methane. Ac-
cording to Blackwood, maximum yields are achieved at a
hydrogen pressure of approx. 5 MPa (4). On the basis of
the experimental results of Birch et al. and Wen and
Huebler a pilot plant for hydrogasification of Rheinische
Braunkohle was successfully calculated (5,6).
From further studies with very high heating rates sum-
marized by Anthony and Howard (7,8) and Johnson (9,10) it
is known, that varying the heating rate in the range
between 10 K·s-1 and 10 4 K·s-1 has only a minor influence
on the yields of volatiles and gaseous products. Increa-
sing the temperature from 500 to 900°C causes a nearly
linear rise of the yields of these products. A most sig-
nificant increase of volatiles and gaseous products un-
der these heating conditions again results from increasing
486
the hydrogen pressure.
The present paper is concerned with the chemism, the
kinetics and technical aspects of hydrogasification of
Rheinische Braunkohle in order to elaborate optimum re-
action conditions for gasification. For analyzing decom-
position and gasification reactions of the coals in inert
or hydrogen atmosphere, studies were preferentially per-
formed at a linear rate of heating (4 K·min- 1 ), whereby
the pressure was varied between 0.2 and 5 MPa. A high
heating rate (100 K·s-1, realized by charging the sample
into the pressurized, pre-heated reactor) followed by
isothermal gasification (1 h) was used for comparison,
but especially in order to simulate heating conditions in
a fluidized bed reactor. Special attention was given to
the moisture contents of the coals and the hydrogen used
for gasification. The influence of mineral matter was
studied with two types of Rheinische Braunkohle with dif-
ferent contents of mineral matter (ash contents 3.5 or
17 wt%).

2. EXPERIMENTAL
2.1. Coals
The ultimate analyses, the ash and moisture contents
of the two coals are given in Table 1.

Table 1 - Ultimate analyses (daf) , ash contents (wf) and

moisture contents of the coals, wt%

ultimate analysis coal I coal II

C 64.9 66.1
H 4.8 5.3
N 0.7 0.8
S 0.9 0.6
o (diff.) 28.7 27.2

ash content 3.5 17 .0

moisture
as received 1) 15 12
after drying 2) 1.5 1.0

1) Predried from appro 50 wt% moisture in air at 105°C by

Rheinische Braunkohlewerke
2) Dried at 1 Pa and room temperature (16 h).
 487

2.2. Apparatus and Procedure

The gasification apparatus and reactor have been de-

scribed in earlier papers (11,12). The fixed bed pres-
sure reactor was modified by adding a piston at the head
of the reactor for charging the sample (100 to 200 mg)
under pressure at the desired temperature (isothermal ex-
periments). A condenser was installed in order to sepa-
rate the tar from the gaseous products (13). The latter
ones (CH4, C2H6, CO, C02' H20 *) were continuously ana-
lyzed by the aid of IR analyzers. The gas formation rates
and the total mass losses given in the following are ge-
nerally related on the water free raw materials (wf). The
gas yields are given as carbon conversion to the actual
gas Amc/mc/o. The volatiles are calculated by difference
on the basis of a carbon balance.

mC(volatiles) = mC(coal) - mC(gases) - mC(residue) .

Yields are presented as carbon conversion like with the
gases. The gas flow (hydrogen, hydrogen/water vapour
mixtures) was generally as high as to avoid boundary film
diffusion.

3. MOLECULAR STRUCTURE
Understanding the chemism of hydrogasification requires
ideas on the molecular structure of the coal. Attempts
to represent the structure or structure units of brown
coal were made by Kroger analyzing the thermal decomposi-
tion products of lignine (14,15), Imuta and Ouchi by ana-
lyzing compounds of depolymerized brown coal (16) and
Wolfrum, who studied functional groups and their relation
to the gasification behaviour (17,18).
The results of these and some further studies (19,20,
21,22,23) were used as a basis for developing a two-
dimensional structure model of Rheinische Braunkohle.
Functional oxygen containing groups were determined by
titration with NaHC03, Na2C03, NaOH and NaOC2H5' Thermal
decomposition of the brown coal in argon and hydrogen was
studied with temperature programmed heating (4 K'min- 1 )
at pressures between 0.2 and 5 MPa by continuously ana-
lyzing the gaseous products and determining the elemental
composition of the residues as function of heat treatment
temperature (24). As a result the structure shown in
Fig.1 is proposed. It is based on the assumptions given
in Table 2.

*) Results on H2 0 are not presented in this paper

488

o
H,G

c--c..8 1

CHJO~O
OVCH,

STRUCTURE MODEL OF RHEINISCHE BRAUNKOHLE 1985

ICT
C270 H240 N3 S, 090
Fig.1

Table 2 Assumptions for a structure model of

Rheinische Braunkohle

Elemental composition

Caromatic' 55 % (ratio between one-, two- and three

ring aromatics, 6:5:2)

caliphatic, 45 % (1 n-paraffin C26; 1 terpene C25;

3 cycloaliphatLcs C6; 5 C3-, 4 C2-'
2 C1 - bridges; 6 CHr, 2 C2H5-'
1 C3H7- groups; residue, C in oxygen
containing groups)

Oxygen ( 9 COO--, 4 C-O-C-, 16 C=O-,34 C-OH-,

2 o-cH 3 -groups; ~ 12 - 0 - bridges)

Nitrogen (2 heterocyclics; 1 amine)

sulfur (1 heterocyclic)
 489

4. KINETICS AND PRODUCT YIELDS AS INFLUENCED BY HYDROGEN

PRESSURE

4.1. Temperature programmed heating

Results on hydrogasification of dried brown coal I at

linear temperature increase up to 850°C are shown in
Fig.2.

0
u

'c'E 2

......:.,.Cl Iu
N
0
E
~.

's? 0
80

I. 40
u

qobo~--3~0~0--~~~--~~~~~~~
b t.min 60
I CT COAL I. dry. O.4-1mm. H2• 4K'min-1, 850·C. 1h. 1('85
- 0.2MPa, .... 0.5MPa.---1MPa.--·-2MPa.-·--'-·5MPa ~
Fig. 2
490

The influence of pressure up to 5 MPa has a very strong

effect on the methane-, but only a small effect on the
carbon monoxide- and carbon dioxide formation rates. Me-
thane is mainly formed at 550 and 800°C, whereby the
small peak at 550 °c is increased with increasing pres-
sure. The peak at 800°C first raises and than decreases
with pressure with the consequence of an additional me-
thane formation between 800 and 550 DC. The reason for
the variation of the methane formation is an improved
hydrogenation of the coal, which favours the formation
of volatile products (tar).
Formation of carbon dioxide seems to have no influence
on methane formation as could be expected. The decrease
of carbon monoxide formation at about 550°C is already
occurring at low hydrogen pressures, therefore methana~
tion of primary formed carbon monoxide may not be respon-
sible for that. Dehydroxylation seems to be more probable.
The results with coal I I with a higher ash content are
very similar, which means, that mineral matter has onLY
a minor influence. This also holds for the following
studies.
The results of Fig. 2 are summarized in Fig. 3 .

E."I ~.
: .....
] 0.8-/
1

4 PH" MPa
COAL!. dry.0.4-1mm. H2 . 4K·min-1 ,
ICT 850·C, lh. 1985
• total. I> CH,.. 0 C2 Hs' oGO· 0 c~ .•volatiles Fig. 3
 491

It confirms that raising pressure leads to higher yields

of volatiles at the cost of the methane yields. The
yields of ethane are slightl" increased and those of
carbon monoxide are decreased by hydrogen pressure, where-
as the carbon dioxide yields are nearly uneffected by
pressure.

4.2. Constant temperature

Hydrogasification of dried coals at constant tempera-

ture was studied at pressures of 1 and 5 MPa and tempera-
tures of 800, 850, 900 and 950°C (24). The gas forma-
tion rates at 800 and 900°C in Fig. 4 show that the
maximum gas formation is occurring within a few minutes
as observed by other authors, but the influence of hydro-
gen pressure between 1 and 5 MPa seems to~be poor as com-
pared to a temperature increase from 800 to 900°C.

b. b.

600

~
,
300 ;
i
I. i
~~
"c :~ '\ "
'\
~~">,_ .-.
..E
~ 0
\~:-.::-::.~-.--
a.
i -
a.
-0
E
<n'
'0
600

300 ~

, \\,
!.'~"~':)~.....".-. '-"_.
10 t, min 20 0 10 t. min 20
COAL I, dry, 0.4 -lmm, Hz, 1h Isothermal,
ICT a.1MPa, b.5MPa, left 800°C, right 900°C 1985
- CH4 ---- CZH6 -.-,- CO ...... COz
Fig. 4
492

Deviating from the results at 1 MPa, 800 and 900°C, a

double peak of methane- and ethane formation is found at
5 MPa and 900°C. This kind of methane or ethane forma-
tion is not caused by a catalytic effect of the mineral
constituents, it is correlated with an intrinsic reacti-
vity of carbonaceous materials under these conditions
of unknown nature (see discussion). It was also found
during gasification of PVC char prepared at 600°C (24).
All results at constant temperature (reaction time 1 h)
are summarized in Fig. 5 (1 MPa) and Fig. 6 (5 ~Pa). A
temperature increase obviously has the most tremendous
effect on the methane yields, especially between 800 and
850°C. As found at linear heating, high pressure favours
formation of volatiles, which occurs at the cost of me-
thane. A high pressure is only favourable if high ethane
yields are desired.

jo
l'°Lc ._.1
0,8
•

O,6~===========:=====:::::::
_

0,6

0,4

o
u'
E
} 0.2
<l
,_ _ _- 0 -_ _ _ ' ~o

II: _tfl
o
o 800 900 T, DC
COAL I. dry, 0.4-1mm, H2 .1MPa,lh,
I CT isothermal, 1985
.total. ~CH4' OC2H6'OCO, DC0 2,.volatiles Fig. 5
 493

f ~:f ~. ------ 1
I

0,6

.~
0,6

E
0
<.i

"-u
0,4

/
•

-
~ 0,2

--0
:~:
a •
a 0
0 0
0

0
800 900 T,oC
COALI, dry, O,4-1mm, H2 , 5MPa.
ICT 1h, isothermal, 1985
.total. 6CH4 , ~C2H6'oCO, DC~,.volatiles Fig. 6

5. KINETICS AND PRODUCT YIELDS AS INFLUENCED BY MOISTURE

OF COAL
5.1. Temperature programmed heating

Hydrogasification of both coals as received (15 or

12 wt% moisture) was studied at pressures of 1, 2 and
5 MPa. Fig. 7 gives a comparison between the methane for-
mation rates found with a wet coal (dotted lines) and the
former results with the dry coal. The moisture of coal is
negligible at 1 MPa, but its influence becomes remarkable
at 2 and especially at 5 MPa, where the methane formation
rate is decisively lowered. The formation rates of the
other gases not shown here are more or less unaffected
by the moisture.
 494

20

c.
~0~0~--~30~0~=-~5+'00~---=WhO~T-.-C--~~==~
- o"[liiin6o

.. 20
c

o
E
b.
"''52
100 300 500 700 T.·C 0----.,,--,---;:0-
r.'" I.min 60
u

40

20

o.
300 700 T .C
. 0 I. min 6b
COALI. 0.4-1 mm. 4K·min-1• 650 ·C. lh.
leT - dry.---wet. 0: 1 MPo. b: 2M Po c :5MPa 1984
Fig. 7

From the summary of the results in Fig. 8 it follows

that the coal is nearly completely gasified at all pres-
sures as found with the dry coal (Fig.3). The main ef-
fect of water consists in lowering the methane yield as
also follows from a comparison with the results of Fig.3.
As the influence of water is enhanced by high pressure,
it may be suggested that evaporation of water is dimi-
nished; water is thus available at higher temperatures,
where radicals are formed by thermal bond cleavages, and
able to inactivate these radicals.
 495

0.6

0.4 .-----
o

r!
"-
u
E
<l 02

COALI.wel.04-1mm.H2.4K·min-l,
ICT B50°C. 1 h. 19B5
.10101 "CIi. ,C2H6 oCO oC02 *vololiles Fig. 8

5.2. Constant temperature

Experiments at constant temperature were performed at
1 MPa and 850°C (residence time 1 h). Results are given
in Table 3.

Table 3 Comparison of relative mass losses and product

yields at 1 MPa and 850 DC received with dry
and wet coal I

dry coal wet coal

83.1 80.4

mc/mc,o

CH 4 55.4 49.9

C2 H6 10.8 11 .5

CO 5.4 6.4

CO 2 4.3 4.5

volatiles (by diff.) 2.2 0.6

496

The data confirm the results obtained during temperature

programmed heating, namely a small effect of moisture on
the relative mass loss and a strongest effect on the me-
thane yield.

6. KINETICS AND PRODUCT YIELDS AS INFLUENCED BY MOISTURE

IN HYDROGEN
6.1. Temperature programmed heating

If the suggestion is correct, that the inhibiting effect

of water on the methane formation consists in a blocking
of radicals or active sites, it should be possible to si-
mulate this effect by gasification of a dry coal with
humid hydrogen. Experiments at constant rate o~ heating
were performed at 1 MPa using hydrogen/water vapour mix-
tures of various compositions.
The results on the formation of methane, carbon dioxide
and carbon monoxide presented in Fig. 9 demonstrate that
the inhibition of methane formation is already existing
with 2 % water vapour in hydrogen. It is astonishing
that higher water vapour contents do not induce a water
vapour gasification as may be derived from the low car-
bon monoxide formation rates at high temperature. The
reason for that is the inhibition of water vapour gasifi-
cation by hydrogen (13,25). It was shown in further
studies that the brown coal is reacting with water vapour
in the absence of hydrogen very violently (13,24).
 497

'c
E
'C<5ll 5
E
on
'0

0 c.
u

0100
700 T DC
. 6 t.mln 60
6
"c
E
'C<5ll
E 3
'l's;?

<3
u
700 T DC
, 6 t, min 6b

.,.
c
E
<5
.nE 20
'0

a
I
u

300 700 T DC
, O~-;-t.-m-,n-""60

COALI.wet. 0,4-1mm, H2/H20~1MPa. 4K-min-'.

ICT 850 DC. 1h, 1984
0=0 - - -"H,o=O,02 ---xH,(T01 . xHf;f0.5 -,-xH,o=0}5 Fig. 9

A summary of the results given in Fig. 10 demonstrates

by the relative mass losses and the methane yields an over-
whelming effect of small water additions, a continuous
decrease of the methane yields and a comparable increase
of the carbon monoxide and carbon dioxide yields with
higher water additions. These results confirm the previous
suggestions that water is blocking active sites and
therefore poisoning the carbon surface.
498

q
~
"' 0.2
~
<l

0.2 0.4 XH,O 0.6

COAL! . wet. 04-1mm. H2 ' H2 0. 1MPa.

I CT 4 K min". 850°C. 1h 1985
.total. oCH,.. oCO. oC02' .volat,les Fig. 10

6.2 Constant temperature

The influence of moisture in isothermal gasification
was studied at total pressures of 1 and 5 MPa and tempe-
ratures of 800, 850, 900 and 950°C using hydrogen/water
vapour mixtures with 10 % water vapour. At 1 MPa and
850°C the water vapour content was additionally varied.
Gas formation rates at 800 and 900 °c and pressures of
1 and 5 MPa during gasification with hydrogen containing
10 % water vapour are shown in Fig. 11. It is remarkable,
that total pressure has a tremendous effect on the me-
thane formation rate already at 800 °c, which was not the
case in gasification with dry hydrogen (Fig.4). The
temperature influence is small at 1 MPa, but strong at
5 MPa. A further point of interest concerns the carbon
monoxide formation at 1 MPa and 900°C starting after
20 min reaction time. This carbon monoxide formation and
the simultaneously occurring weak increased methane for-
mation result from the catalytic influence of the iron,
which is formed by reduction of the iron disulfides of
mineral matter.
 499

400 b b.

200

a a.
"''0

f\i
100

/-.-.-

'"
40 0 40

COAL!. dry.04-1mm. H2/Hp.XH,o=0.1. lh isotherm

ICT a.1MPa. b.5MPa. left BOOoC. right900°C 1985
- CH 4 • - - C2H6' -·-CO. .. ••• CO2 Fig. 11

Fig.s 12 and 13 give the relative mass losses and the

yields of the gases and volatiles at 1 or 5 MPa. At the
lower pressure, the relative mass loss below 900°C only
amounts to 0.6 which is the lowest mass loss measured in
all experiments. The strong increase between 850 and 900°C
results from carbon monoxide and not from methane
formation, but the raising carbon monoxide yields are
mainly caused by the catalytic water vapour gasification
described above. To demonstrate this catalytic in-
fluence,the carbon monoxide yields in the first range up
to 20 minutes and in the second catalytic range between
20 and 60 minutes are plotted separately. The results
mean that coal conversion at 1 MPa as caused by the in-
fluence of water may be improved by raising the tempera-
ture above 850 or 900°C.
The situation drastically changes at 5 MPa as reflec-
ted by Fig. 13. Raising the temperature causes a nearly
linear increase of the relative mass loss, methane
500

1.2~
0__ o /0........" stage2

s":o -=~:~~:§/~=====1,
- - -CGALI
lMPa '1dr
• total , ',CH .Isotherm~1 ~IHzO,XH,o= 0.1
h Y '. O,4-1mm
4,OCzHSPCQ, aCOz, "VVlo\iles
._, Fig. 12

800 900
COAL I dr T DC
• total ',CH ' Isotherm~1
5MPo ' 1:' .0.4 -lmm H I '
z HzO,XH,o=O,1
_--,-,-.J .. , 4'oCzHs,oCO'
,oCOz,·volotiles
--'-- Fig. 13
 501

represents the overwhelming gaseous product, whereas the

carbon monoxide yield is really small as compared
to 1 MPa. Nevertheless, for compensating the poisoning
effect of water, temperature has to be at least 950°C.
A smimilar relative mass loss and a similar methane yield
are achievable already at 850°C and 1 MPa only, if a
dry coal is gasified with dry hydrogen (Fig. 5). In so far,
it is obvious that water either as moisture of coal or
hydrogen has a catastrophic effect on coal conversion
and methane yield.
This conclusion is completely confirmed by the results
obtained at 1 MPa and 850°C using hydrogen water vapour
mixtures of different compositions (Fig.14). Under these
conditions, the methane yields are lowered from 55 % car-
bon conversion with dry coal and dry hydrogen to 20 %,
if only 2 % of water vapour are present in the hydrogen.

f ::~ .
0.6 - - .

O.
1

0.4

~ 0.2 ~---------
<)

E
<l

0.2 xH,O 0.4

COALI.dry.0.4-1mm.H2/H20.1MPa.
ICT 850°C. 1h. isothermal. 1985
.totol. bCH4' OC2t\;.oCO. oCCh .•volo Fig. 14
502

7. DISCUSSION

7.1. Chemism of hydrogasification

Studies on pyrolysis and hydrogasification of the brown

coals at linear heating at different pressures up to 5 MPa
yielded fundamental informations on the formation of ga-
seous and volatile products. The results are summarized
in a scheme given in Fig. 15. Carbon dioxide and carbon
monoxide are both formed in two different temperature
ranges during pyrolysis in argon. Carbon dioxide may
mainly result from originally existinq carboxylic groups,
whereby neighbouring groups, which again may be carboxy-
lic groups, are responsible if they are decomposed at
lower or higher temperatures. Lactones and carboxylic an-
hydrides may intermedia.tely be formed. Esters may be a
further source of carbon dioxide.

i i i i i i
300 400 500 600 700 000
temperature, ·C

COAL-~~::::"'...,-,:~=~===:::::::::-..L:~~-"",,"""--__ residue

•
HZI(>lMPo)
I
:1
11«lMPo)
I
volatiles ~ H,

~
1Hz

(by char hydrogenation)

PRESENTATI()I OF HYDROOASIFICATION OF BROWN COAL

ICT AT LINEAR HEATING (4K·min-1l; WATER OMITTED 1985
Fig. 15

A correlation between oxygen containing groups

and carbon monoxide formation is more complex. In the
lower temperature range carbon monoxide formation may be
correlated with carbon dioxide formation, for example by
decomposition of lactones and carboxylic anhydrides.
Another possibility are aliphatic keto groups. In the
higher temperature range carbon monoxide should predo-
minantly be formed by the decomposition of phenolic hy-
droxylic groups, quinones and oxygen bridges.
 503

The formation of both gases in the lower temperature

range is not influenced by hydrogen pressure, whereas
that in the higher temperature range is completely sup-
pressed at 5 MPa. Analysis of water was not sensitive
enough in order to decide whether dehydroxylation is
occurring or primarily formed carbon monoxide is me-
thanated. Methanation is less probable because the carbon
monoxide formation is decisively lowered already at
0.2 MPa and especially at 0.5 MPa.
Ethane formed with a maximum rate between 400 and 450 D C
may primarily result from ethyl radicals. The same con-
clusion may be drawn for the methane formation at 450 D C,
namely from methyl radicals. At lower hydrogen pressures
up to approx. 1 MPa methane is further formed at about
550 and 800 DC. Methane at 550 DC is attributed to the
reaction of weakened ring structures,especially aliphatic
ring systems, and carbon bridges with hydrogen. Methane
at 800 DC results from the reaction of a highly disordered
reactive char with molecular hydrogen. A high amount of
volatiles is formed by thermal decomposition. The effect
of hydrogen may be seen in preventing condensation of
radicals, which are primarily formed by thermally induced
bond rupture. Strong hydrogenation especially at high
hydrogen pressures keeps the coal in the coal state up to
higher temperatures. As a consequence, increasing hydrogen
pressure effects a strong increase of the volatiles yields.
This process is the reason why methane formation is shif-
ted to lower temperatures. Missing methane formation at
800 DC and 5 MPa is caused by volatilization of the coal
at lower temperatures. The previous considerations are
also valid during gasification at constant temperature
after rapid heating up. The main difference consists in the
preferred formation of a reactive char instead of the
hydrogenation of coal. That is the reason why a tempera-
ture increase has the most dramatic effect, namely by re-
ducing the yields of volatiles and increasing the yields
of methane. It is worthwhile to note that the methane
yields at high temperatures are still higher at 1 MPa
than at 5 MPa, whereas the volatiles yields are extremely
lower at 1 MPa than at 5 MPa (Fig.s 5,6).

7.2. Influence of moisture

The discussion on the influence of moisture on hydro-
gasification, fed to thereactor either by the coal or by
the hydrogen, may be restricted to the influence of moi-
sture in general, because it was shown by the experiments
that moisture of coal is only inhibiting hydrogasification
504

at high pressure, where evaporation of water is dimi-

nished (Fig.7). It was also shown by the results, that
the inhibiting effect of water is maximum already with
very small amounts of water vapour in hydrogen as
followed from Fig. 10, Table 3, and especially from Fig.14
The experiments at constant temperature revealed that
the inhibiting effect of water is maximum at temperatures
up to 850°C and the lower pressure of 1 MPa. By increa-
sing the pressure and additionally the temperature it is
possible to compensate the inhibition of water
(Fig.s 12,13). The inhibition is maximum with small
amounts of water, therefore it may be suggested that
water acts as a poison by blocking of· active sites. Ac-
cording to HERMANN (26,27) studying water vapour gasifi-
cation with hydrogen/water vapour mixtures, water is dis-
sociatively adsorbed at active sites by formation of hy-
droxyl surface groups, which are able to condensate for-
ming extremely stable ether bridges (Fig.16).

+
. 2H,O. -H,
+H,,-2H,O

ICT i
DEMONSTRATION OF THE INHIBITING )1
EFFECT OF WATER IN HYDROGASIFICATIO 1985
BY FORMATION OF ETHER BRIDGES Fig. 16

By desorption measurements in inert gas he found that

temperatures decisively above 800 to 850°C are necessary
to decompose these structures. It may therefore be con-
cluded that these ether structures are responsible for
inhibition of hydrogasification.
It is another question, why water vapour gasification
is not occurring even if high water vapour contents are
used (Fig.14), because water is able to react with these
ether structures already above 750°C (26,27). This
question may roughly be explained by a molecular sieve
effect, which means that hydrogen is preferentially dif-
fusing into the pores and especially into the micropores
 505

with the consequence that the water vapour partial pres-

sure in the pores is extremely low (13).
In so far it is understandable that the inhibiting ef-
fect of water in hydrogasification may be eliminated by
increasing the temperature beyond 850°C, whereby at low
pressures high carbon conversions are caused by a cata-
lyzed water vapour gasification (Fig.12). Really high me-
thane yields which are comparable to those of gasifica-
tion in a dry atmosphere (Fig.s 5,6) are only achievable
if temperature and pressure are simultanously raised.
This result may be explained by assuming that ether
structures are reacting with hydrogen or that their for-
mation is inhibited, as also shown in Fig. 16.

Acknowledgement
Financial support by the Federal Ministry of Research
and Technology (BMFT) is gratefully acknowledged.

8. REFERENCES

1. Birch, T.J., K.R. Hall and R.W. urie. Gasification

of Brown Coal with Hydrogen in a Continuous Fluidized
Bed Reactor. J. Inst. Fuel 33 (1960) 422
2. Birch, T.J., P. Casamento, N.C. Grave and D.J. McCarthy.
Hydrogasification of ,Brown Coal Char in a 20-ft Con-
tinuous Fluidized Bed Reactor. J.Inst. Fuel 42 (1969)4
3. Wen, C.Y. and J. Huebler. Kinetic Study of Coal Char
Hydrogasification. Ind.Eng.Chem. Process Design De-
velopment 4 (1965) 143.
4. Blackwood, J.d. and D.J. McCarthy. The Mechanism of
Hydrogenation of Coal to Methane. Aust. J. Chern. 18
(1966) 797.
5. Schrader, L., H.Teggers and K.-A. Theis. Hydrierende
Vergasung von Kohle. Chem.-Ing.-Techn. 52 (1980) 794.
6. Teggers, H., K.-A. Theis and L. Schrader. Synthese-
gase und Synthetisches Erdgas aus Braunkohle. Erdal
und Kohle, Erdgas, Petrochemie. 35 (1982) 178.
7. Anthony, D.B. and J. B. Howard. Coal Devolatilization
and Hydrogasification. A.I. Ch. E. J. 22 (1976) 625.
8. Howard, J.B. Fundamentals of Coal Pyrolysis and Hydro-
pyrolysis. Chemistry of Coal Utilization, Ed. Elliott,
John Wiley, New York (1981) 665.
9. Johnson, J.L. Gasification of Montana Lignite in Hydro-
gen and in Helium during Initial Reaction Stages.
Am. Chern. Soc. Div. Fuel Chern. Preprints 20 (1975) 61.
10. Johnson, J.L. Kinetics of Coal Gasification. John Wiley
New York (1979).
506

11. Huttinger, K.J. and W. KrauB. Catalytic Activity of

Coal Minerals in Hydrogasification of Coal. Fuel 60
(1981) 93.
12. Huttinger, K.J. and W. KrauB. Catalytic Activity of
Coal Minerals in water Vapour Gasification of Coal.
Fuel 61 (1982) 291.
13. Huttinger, K.J. and A. Michenfelder. The Inhibition
of Hydrogasification of Brown Coals by Moisture:
Influence of Heating Rate, Temperature and Pressure.
Fuel 64 (1985) in press.
14. Kroger, C. Die Chemische Konstitution der Braunkohle
Humussubstanzen. Brennstoff-Chemie 47 (1969) 249.
15. Kroger, C. Der Wachs/Harzkomplex und das Kohlebitumen.
Brennstoff-Chemie 49 (1968) 35.
16. Imuta, K. and K. Ouchi. Some Compounds Separated from
Depolymerized Brown Coal. Fuel 52 (1973) 174.
17. Wolfrum, E.A. Correlations between Petrographical
Properties, Chemical Structure and Technological Be-
haviour of Rhenisch Brown Coal. The Chemistry of
Low Rank Coals, Ed. Schobert, ACS Symposium series
264, p.15
18. Wolfrum, E.A. and W. Assmann. Der EinfluB Sauerstoff-
haltiger Grupp~n auf das Entgasungsverhalt~n einiger
Braunkohletypen des Rheinischen Reviers. Communication
of Rheinische Braunkohlenwerke AG.
19. Kube, W.R., H.H. Schobert, S.A. Benson and F.R.Karner.
The Structure and Reactions of Northern Great Plains
Lignites. The Chemistry of Low Rank Coals. Ed. Schobert,
ACS Symposium series 264, p.39
20. Evans, D.G. and R.J. Hooper. Deduction of the Struc-
ture of Brown Coal by Reaction with Phenol. Coal
Structure, Ed. Ouchi, ACS Symposium series 192, p.191
21. Nelson, P. Chemically-Bond-Polymethylen Groups in
Coal. To be published in Nature.
22. Schmiers, H. untersuchungen uber die Verkokung von
Braunkohleteeren nach dem Delayed-Coking-Prinzip.
Brennstoff-Chemie 50 (1961) 13.
23. Mallya, N. and R.A. Zingaro. Some Structural Features
of a Wilcox Lignite. The Chemistry of Low Rank Coals.
Ed. Schobert, ACS Symposium series 264, p. 133.
24. Michenfelder, A. Chemismus der Hydrierenden Vergasung
Rheinischer Braunkohle. Ph.D. thesis University of
Karlsruhe (1985).
25. Hedden, K. and G. Mienkina. Zur Reaktionskinetik der
Druckvergasung von Kohlenstoff mit H20-H2-Gemischen.
Brennstoff-Chemie 46 (1965) 366.
26. Hermann, G. Mechanismen der Nichtkatalysierten und
durch Eisen Katalysierten Wasserdampfvergasung von
Kohlenstoff. Ph.D. thesis University of Karlsruhe
(1985 )
 507

27. HUttinger, K.J., J. Adler and G. Hermann. Iron Cata-

lyzed Water Vapour Gasification of Carbon. Proceedings
of this Conference.
PART VI. COKE PRODUCTION AND REMOVAL
 511

THE MAKING OF COKE IN THE DELAYED COKER AND IN THE SLOT COKING OVEN

Harry Marsh

Northern Carbon Research Laboratories, School of Chemistry,

Bedson Building, University of Newcastle upon Tyne,
Newcastle upon Tyne NEI 7RU, U.K.

1. INTRODUCTION

Coke is the highly carbonaceous product of pyrolysis of

organic matter at least parts of which have passed through a liquid
or liquid-crystalline state during the carbonization process and
which consists of non-graphitic carbon.

Cokes are of commercial interest for several reasons, a major

reason being that some are graphitizable, i.e. upon graphitization,
heat treatment to about 3000 o C, they are converted into poly-
crystalline graphitic carbon.

Some hundreds of millions of tonnes of cokes are produced

annually, internationally, almost exclusively by the petroleum and
coal-associated industries. All industrial cokes therefore are
produced from feedstocks based exclusively upon either petroleum
sources or on selected coals.

In terms of tonnages used, the major user of coke is the

aluminium industry. Here carbon is the reductant of aluminium
oxide (alumina) to aluminium in the electrolysis cell, as the pre-
baked anode. This anode is made from a petroleum coke and
a coal-tar coke derived dur1ng manufacture from a coal-tar pitch.
The second major use of petroleum coke is for electroeraphite, that
is the graphite electrodes used to carry electrical energy into the
crucibles of the steel making industries. Here the graphite is not
consumed as in the aluminium-making industries. The electro-
graphites are made from a superior grade of petroleum coke as well
as coal-tar pitch.
512

Although the major use of coal is still for combustion about

10% of mined hard coal is used to make metallurgical coke. This is
a porous strong, brittle solid of low reactivity, eontaining several
percent of mineral matter. In the blast furnace the coke produces
thermal energy and carbon monoxide for reduction of iron oxides.

There are specialist applications for cokes in carbon/graphite

artifacts, e.g. as in the many forms of electrodes and brushes for
electric motors, for torch batteries and as the moderator of
neutrons in the Advanced Gas Cooled (Nuclear) Reactors (AGRs).

The role of coke in maintaining established industry in a

developed society is therefore significant.

This Chapter summarises the origins, manufacture and essential

properties of the several cokes used by industry.

2. NOMENCLATURE

An 'International Committee for the Characterization and

Terminology of Carbon' exists (Carbon, 13,251 (1975)). The mandate
for this Committee is to establish definitions for the many
processes and carbon forms which are known to carbon science and the
carbon industry.

Tentative definitions are available in Carbon, 20, 445 (1982)

and Carbon, ~, 517 (1983).

2.1 Coalification This is the geological process of formation of

materials with increasing carbon content (peats and coals) from
organic materials which occur in the earth's crust by gradual
transformation ab moderate temperature (about 500 K) and under high
pressure.

2.2 Carbonization This is the process of formation of material

with increasing carbon content from organic material, usually by
pyrolysis, ending with an almost pure carbon residue at temperatures
up to about 1600 K.

2.3 Coke This is a highly carbonaceous product of pyrolysis of

organic material at least parts of which have passed through a
liquid or liquid-crystalline state during carbonization.

2.4 Delayed Coke This is a commonly used term for a primary

carbonization product (Green or Raw Coke) of high boiling hydro-
carbon fractions (heavy residues of petroleum or coal processing)
produced by the Delayed Coking Process (see Section 3.1).

2.5 Petroleum Coke This is a carbonization product of high

 513

boiling hydrocarbon fractions obtained in petroleum processing

(heavy residues). It is the general term for all special Petroleum
Coke products such as Green (Raw) and Calcined products.

Note. Three grades of petroleum coke, as produced in the

delayed coker, exist within the industry:-

Shot Coke is a har~ spherical coke (1-2 mm dia.), rich in mineral

impurity, of low commercial value.

Sponge (Regular) Coke is a coherent coke with adequate strength and

electrical conductivity for principal use as a Filler Coke as the
main constituents of carbon artifacts, e.g. carbon anodes for
aluminium production and for manufacture-0f electrographite.

Needle-coke is the commonly used term for a special type of

(delayed) coke with extremely high Graphitizability resulting from
a strong preferred parallel orientation of its microcrystalline
layer structure.

2.6 Metallurgical Coke This is produced by Carbonization of coals

or coal blends (usually) at temperatures up to 1400 K to produce a
macroporous carbon material of high strength. Metallurgical coke
must have high strength to support heavy weights (the burden) in
the blast furnace without crumbling.

3. THE DELAYED COKER

The delayed coker was developed and introduced by petroleum

companies in the 1940s and in subsequent years, as the demand
increased for petrol, aviation and heating fuels.

The delayed coker was developed to convert the heavy, pitch-

like residues of petroleum distillation into a coke, that is, a
solid material which could be disposed of more easily than the pitch
residues, and at the same time (and of more importance) to obtain
further high value products from the residues (low boiling-point
fractions).

Over the past decades, the use of the delayed coker has
developed from being almost a 'waste-bin' of the petroleum industry
to being a commercial process in its own right producing sponge
(regular) cokes and needle-cokes with an adequate commercial value
on the markets. This demand for delayed coke was initially
associated with the demand f!'lr aluminium, (the .conversion of
Soderberg processes to use of the baked anode) and for electro-
graphi tes.
 514

3.1 Delayed Coking Processes

In simplistic terms, the heavy crudes, residues, tars and

decant oils of the petroleum distillation processes are fed to a
heater where they are rapidly heated with some thermal decomposition
to gases, and fuel liquids. The undecomposed liquids are then fed
into the base of a coke drum and the coke drum filled. The
conversion of the petroleum residues to coke and overhead vapours
is completed within the drum over a period of 24 hours at
temperatures of about 425 0 C to 450 0 C (800 0 -8400 F) and pressures of
0.1 to 0.3 MPa (15 to 45 PSIG). The fluid within the drum is
subject to severe conditions of turbulence from both the injection
of the feedstock and the evolution of volatiles.

Under these conditions, the liquid feedstock residues are

converted to coke via the intermediate formation of the discotic,
aromatic, nematic liquid crystals (as described in the Chapter on
Structure in Carbons). Because of the wide range of chemical
composition of feedstocks and of the physical conditions within
the drum, wide variations of delayed coke structure then result
from that of the shot cokes to that of the premium-value needle-
cokes.

Recent developments within the petroleum industry involve the

preparation of feedstocks especially suitable for the production
of sponge-cokes or needle-cokes. The delayed coker has thus
changed its role somewhat from being a disposer of waste fluids
(which it still is) to a specialised processing unit.

The delayed coking process operates on a batch system. As one

drum is being filled with pre-heated fluid feedstock so a second
drum is being emptied of coke. Delayed coking units may comprise
four drums. At the end of the pitch-coke conversion process the
contents of the drum are steamed and cooled. The upper and lower
flanges of the coke drum are removed giving access to
the coke within the drum, this coke being firmly contained within
the drum (except for shot coke). It is therefore necessary to cut
out the coke. Initially a central hole is cut throughout the bed
of coke using a high pressure water lance. Then a hydraulic
cutting tool loosens the coke which falls out of the drum.

Althouvh the principles of delayed coking are described in

the above simplistic terms, the engineering and safe operation of
delayed cokers represent a major achievement.

The delayed coker is not capable of converting any given

feedstock to a desired coke. This is the operators' dream.
Probably the most critical of the process variables is feedstock
quality. If this is not within speci£ied limits then the desired
coke will not be produced. The other process variables available
 515

to an operator are heater outlet temperature and coke drum pressure.

However, the range of variables available is limited.

Too high a heater outlet temperature can lead to coking in

transfer lines. Too low a heater outlet temperature creates too
high a volatile content within the coke. Low drum pressures
favour increased yields of the desirable liquid distillation
products but with reduced coke yield.

3.2 Feedstock Quality

Shot Coke This is coke in the form of individual spheres or

agglomerates of spheres, 1-2 mm diameter usually. This coke is
unreactive and of low or zero commercial value. Its optical
texture (see Chapter on Structure) is of fine-grained mosaics
(-1 vm dia.). Feedstocks are usually of lowest value, high in
sulphur, asphaltenes and metals.

Sponge Coke This is the 'regular' type of coke from the

delayed coker. The individual spheres are not apparent and the
coke has a continuum of structure. it is porous, hence the name
sponge-coke. The optical texture is dominantly that of coarse-
grained mosaics (-10 vm dia.) with some flow anisotropy (10-50 vm).
These structures ensure the required electrical conductance for
anode usage. This coke is produced from high resin-asphaltene
feedstocks.

Needle-Coke This coke is quite anisotropic in structure with

elongated 'crystals' of coke as formed from the mesophase. The
porosity is also elliptical. The coke is of relatively low
strength but maximises in terms of electrical conductance. The
feedstocks are usually highly aromatic thermal tars, pyrolysis tars,
(ethylene cracker tars) and decant oils.

4. FORMATION OF DELAYED COKES

Structures found in the shot, sponge and needle-cokes are

closely related to the structures, i.e. size and shape, of the
mesophase as developed from that particular feedstock. An under-
standing of the origins of these structures must therfore be based
on an understanding of the origins and properties of mesophas'e(s)
with a wide range of size and shape of optical texture (see
Chapter on Structure in Carbons).

Size and shape of optical texture is intimately associated

with the viscosity of the developed mesophase or nematic liquid
crystal. At temperatures of 400 0 -425 0 C, i.e. those of the delayed
coker, the intermediate liquid crystal can adopt a variety of
structures dependent upon its viscosity at that stage and the
 516

magnitude of the deforming forces. But viscosity relates

to the pyrolysis chemistry of the parent feedstock. Because of
this, little latitude exists within the process variables to
produce different grades of delayed coke, at will.

The presence of oxygen, sulphur, of high molecular weight

asphaltenes with associated organometallics and low contents of
aromaticity constituent a feedstock composition such that any
developed mesophase is formed from relatively 'reactive' molecules
which ensure a (relatively) high viscosity mesophase. This in turn
prevents growth and coalescence of the mesophase and the optical
texture of resultant cokes is that of the small growth units of
mesophase, i.e. fine-grained mosaics (-1 jlm dia.) as in shot coke.

On the other hand, for highly aromatic feedstocks, the meso-

phase is much less reactive chemically, viscosity is relatively
low and the mesophase is therefore capable of having its structure
(optical texture) modified by external physical forces.

In summary, the chemical composition of the feedstock dictates

the size and shape of mesophase which can be formed in the delayed
coker. But the physical conditions within the delayed coker can
superimpose forces which create the unique appearances of the
various delayed cokes available. Shot coke and needle-cokes are
the extremes of effects of pyrolysis chemistry, fluid turbulence
and volatile release within a delayed coker.

4.1 Shot Cokes

Shot coke formation is detrimental to delayed coking. It

reduces operational efficiency, creates safety hazards for operating
personnel and is of little commercial value.

Although the origins of shot coke remain dbscure in detail,

they are associated with 'reactive' feedstocks and relatively high
production of volatiles (e.g. see Marsh R., C. Calvert and J. Bacha,
J. Mat. Sci., 20, 289 (1985)).

The feedstock is capable of producing mesophase of small size,

fine-grained mosaics, and so the optical texture of the coke is
established. The shot-like appearance must be associated with the
rate and extent of volatile release. The system of pitch/mesophase
within the delayed coker can become so dispersed with bubbles or
pores of volatile matter that eventually it becomes physically
unstable and 'inverts' to the more stable system of droplets of
pitch/mesophase dispersed within the vapour phase (i.e. phase
inversion occurs). The droplets of pitch/mesophase-eventually
convert to mesophase totally and then to the green shot coke which
settles downwards into the coke drum.
 517

The creation of individual spheres or of agglomerates of

spheres with different depths of agglomeration is a function of the
viscosity of the mesophase. There must be a continuum of viscosity
of mesophases because there must exist a continuum of 'reactivity'
of feedstocks.

For the sponge cokes, the interdependent relationships of

viscosity with rates and extents of volatile release are such that
the liquid pitch/mesophase syst.em is not diluted with vapours in
extent sufficient to cause a phase inversion. Thus a coherent coke
is created. The larger size of the optical texture indicqtes a
more stable mesophase from a relatively less reactive feedstock,
also indicative of lower extents of volatile release.

This sequence of changes is a continuum from the shot cokes to

the needle-cokes. All intermediates are possible in the same sense
as all intermediates of chemical reactivity within feedstocks are
possible.

4.2 Needle-Cokes

The optical textures of needle-cokes result from deformation

of mesophase formed from the feedstock in the delayed coker (For
further details see Markovic V., S. Ragan and H. Marsh, J. Mat.
Sci., 19, 3287 (1984) and Markovic V. and H. Marsh, J. Mat. Sci.,
19, 3301 (1984)). The microstructure (optical texture) of the
needle-cokes is much more complicated than that of the shot cokes
and hence it must have been established towards the later stages
of the coking process. Otherwise the unique acicular structures of
the mesophase would have coalesced into a continuum of uniform
optical texture. The complex folding of lamellae seen in needle-
cokes must have resulted from significant deformation during the
coking cycle within delayed cokers.

It is possible that the initial stages of mesophase development

consist of growth and coalescence to give a continuum of mesophase.
However, as filling of the drum continues, the mesophase viscosity
increases and the turbulent flow of bubbles and incoming feedstock
through the mesophase results in a permanent deformation which
cannot relax away. The upward movement of bubbles or rapid movement
of liquid through the mesophase appears to be responsible for this
deformation. Mesosphase is deformed due to displacement around the
bubble and often pairs of bubbles act to squeeze the mesophase
resulting in the folded lamellae characteristic of the acicular
structures of needle-cokes. Parts of the mesophase are not so
affected and this two-phase structure of needle-coke becomes frozen
into the semi-coke and recently formed mesophase.

The needle-cokes therefore result from a unique combination of

chemical composition of feedstock (to give a stable mesophase) and
 518

limited extents of volatile release from within the mesophase,

associated with bulk flow, to give the acicular, or "pulled-out"
structures of the needle-coke and the limited porosity. The needle-
cokes have a suitable combination of sizes of crystals, to give
good electrical conductance and of porosity and to give low values
of coefficient of thermal expansion (CTE).

5. THE SLOT COKING OVEN

Coke was used successfully by Abraham Darby in 1709 for the

first time, for iron smelting. These cokes were made in heaps
following the traditions of charcoal making. However, with the
advancement of the industrial revolution, the demand for coke far
exceeded that which could be supplied via the heap method. The
beehive oven was developed in the l750s to be eventually replaced
by the rectangular oven introduced about one century later.

These rectangular, or slot-like ovens, with by-product

recovery were built in batteries, the rectangular chambers being
separated by hollow walls in which the heating gas was burnt. The
rectangular ovens are filled with coal and the resultant coke is
discharged by automated machinery. The volume of an oven is about
37 m3 , containing 30 tonnes of coal, with the rectangular (slot)
shape, 6 m high, 16 m deep and 0.45 m width.

Whereas the earliest coke makers used single coals, i.e. the
prime coking coals, the use of blends of several coals foY--
carbonization is now well established and almost universal practice.
Coking practice also makes use of pre-heated coals fed to the oven
in order to improve coke quality. The development of modern coking
practice is a subject area in its own right and further reference
should be made to: "Modern Coking Practice, Volume 1, The 'Coke Oven
Managers' Association, 1984, Ed. G.E. Hall, (1984).

6. COKE IN THE BLAST FURNACE

Of the rank range of coals available, only the fusing coals of

high rank are suitable for coke making by traditional methods. The
carbonization process has to pass through a fluid phase in order to
create the necessary porosity, the necessary strength and the
necessary low reactivity to oxidising gases, for use in the blast
furnace.

Coke in the blast furnace has several roles to play:-

(i) it acts as a diluent for the iron ore charge so permitting

permeability of gases throughout the stack of the blast
furnace. As such, it has to withstand the pressures of the
 519

burden of the furnace and accordingly has to have high

strength.
(ii) it has to be resistant to gasification by the carbon dioxide
in the gaseous atmosphere of the blast furnace. This
minimises formation of carbon monoxide and 'solution-loss'
of carbon from the top of the furnace.
(iii) it has to withstand attack by alkali vapour in the blast
furnace. These vapours tend to cause comminution of the
coke into smaller pieces so reducing permeability.
(iv) coke is combusted in the raceways of the blast furnace to
produce carbon monoxide, the reducing agent for iron ore,
and most (or all) of the thermal energy within the furnace.

It is an act of geological kindness that the medium- and low-

volatile high rank coals carbonize to give coke structures which
fulfill, more or less, these four roles of metallurgical cokes in
iron making.

7. STRUCTURE IN COKE

Metallurgical coke is a porous solid with cell wall material

co~posed mainly of anisotropic carbon. Pore sizes are critical to
strength, a high porous coke, as produced from a high-volatile
softening coal, being too weak. The anisotropic carbon of the pore
walls is dominantly of small size optical texture, fine to coarse-
grained mosaics (1-10 ~m dia.). Such structures give coke its
required strength. These anisotropic carbons have lower
reactivity to oxidising gases than isotropic carbons and this is to
the advantage of the metallurgical coke. Further, small but
significant contents of flow anisotropy exist. These develop
shrinkage cracks on heating and assist with the thermal stability
of the cokes when heated to >1500 oC in the raceways.

Coke structure is complex, but this complexity allows it to

perform adequately within the blast furnace.

The coke structures are established via a liquid crystal

intermediate (mesophase). (See Chapter: Structure in Carbons).
The same considerations apply to coal and coal blends as to
petroleum and coal-tar pitches. The high content of heheroatoms
in coals, their mineral matter and cross-linkages tend to promote
mesophase(s) of relatively high viscosity. The judicious use of
coal blends has been successful in maintaining the necessary coke
structures.

7.1 Optical Textures of Cokes from Coals of Different Rank

There are major differences in the si2l.e and shape of optical

textures of cokes from coals of different rank, the size of the
 520

optical texture decreasing with decreasing coal rank. The high-

rank, low-volatile anthracites and semi-anthracites carbonize
without significant softening such that the resultant product is
highly anisotropic only because of the preservation of the inherent
anisotropy of the parent coal. The prime coking coals produce
cokes with the largest size of optical texture, ranging from very
fine-grained mosaics to medium-flow anisotropy. Fracture surfaces
examined using scanning electron microscopy (SEM) indicate a
layered arrangement within the cokes (probably associated with
inherent anisotropy) with a sub-structure of growth units of meso-
phase.

The cokes from low-rank, high-volatile, weakly caking coals

contain approximately 25% anisotropic carbon with an
optical texture of fine-grained mosaics, 0.5 - 1.5 ~m diameter.
The low-rank, non-caking bituminous and sub-bituminous coals
produce cokes which appear to be isotropic in the optical microscope.
In some cokes of this type, however, anisotropic units may be
present of size <0.5 ~m diameter, i.e. beyond the resolving power
of the optical microscope. These can
be distinguished using an
etching technique followed by examination using SEM.

Thus, the coking process of the prime coking coals represents

an optimization of several evenD.s. The coalification process is
not so advanced as to prevent thermolysis and fluidity. The extent
of aromatization is sufficiently advanced to permit the formation
of discotic aromatic nematic liquid crystals. The thermal
stability of the molecular constituents of the pyro1ysing coal or
blend leads to the decrease in volatile matter. This subsequently
prevents the formation of a highly porous coke. Instead, a coke of
optimum porosity from the point-of-view of strength and density is
created, a co-operating factor being a significant decrease in rate
of loss of volatile matter during the period of decreasing fluidity
of the system associated with increasing carbonization temperature.

The residual heteroatom content and functionality of the

coking coals permit the formation of the mosaics of optical texture
(0.5 to 10 ~m diameter) and some medium-flow anisotropy «30 ~m ~n
length). Should the heteroatom content and functionality have
decreased further, and if the same extent of thermolysis was
possible, then cokes resembling the petro1eum--derived cokes as
used in graphite manufacture would have resulted. Clearly, these
would be quite unsuitable for the blast furnace because of their
low mechanical strength.

7.2 Co-carbonizations to Produce Cokes with Suitable Optical

Texture

Currently, it is common practice to produce metallurgical

cokes from blends of coals of different rank. Considerable care is
 521

exercised in blend formation. Little or no interaction occurs

between prime coking and high rank coals. A strong interaction
occurs between prime coking coals and the lower rank medium- artd
high-volatile caking coals. Little interaction occurs on co-
carbonization of prime coking coals with the weakly caking high-
volatile coals. Where interactions occur it is observed that the
optical texture of resultant coke is clearly not that of either of
the coal-cokes but is different and is developed from the fluid
blend of the two parent coals. The prime coking coal up-grades the
size of the mosaics beyond that observed in cokes from the lower
rank coals of the acceptable blend.

Within recent years attention has been given to the use of

petroleum and coal-based pitch-like products to replace prime
coking coals in blend formulations.

In an examination of the co-carbonization of one coal with

several industrially supplied petroleum pitches it has been noted
that pitches differ considerably in their ability to modify the
carbonization behaviour of the coal. Some pitches (passive) had
little effect on the coal, the product being a composite of coal-
coke and pitch-coke. With the active pitches, the latter
interacted totally with the coal to produce a coke of different
optical texture, no pitch-coke being detectable in the resultant
coke composite. The need for a fluid phase is again stressed.

7.3 Aspects of the Chemistry of Coal Carbonization to Produce

Acceptable Cokes

As the chemical compositions of coals and pitches are extremely

complex so no exact explanation exists of differences in size and
shape of optical texture of resultant cokes in terms of the pyrolysis
chemistry of the carbonization process. However, it is known that
pitch materials differ widely in their ability to modify coal
carbonization systems so some explanation must be sought. It has
been suggested that naphthenic groups and aromatic ring systems
promote the growth of larger sized optical textures whereas the
presence of heteroatoms and functional groups promote the growth of
~maller sized optical textures. Recently, the concept of 'hydrogen
shuttling' has been introduced in discussions of the development of
mesophase during co-carbonization of coals with pitches. It is
suggested that the hydrogen may stabilize certain of the free-
radicals produced during pyrolysis and that coals should be
considered to be good hydrogen donors and acceptors. The benefici-
ation of hydrogenation of coal additives is reported and relates
closely to such treatments of the feedstocks used to make needle-
cokes in delayed cokers.

Thus, the process of modification'"of carbonization processes

by pitch additions is probably associated with hydrogen transfer
522

reactions (hydrogen shuttling) involving movement of hydrogen from

the pitch to radicals of the carbonization process. The
stabilization of otherwise reactive radicals by hydrogen acceptance
prevents formation, at too early a stage in the process, of coal-
coke and allows the pitch to interact chemically with the coal
ultimately to create a new beneficial optical texture via formation
of nematic liquid crystals.

8. CONCLUSIONS

8.1 The Delayed Coker

The chemical composition and treatments of feedstocks going

into the delayed coker primarily dictate coke quality. The process
variables of temperature and pressure within the coke drum are less
important. Structures of cokes formed within the delayed coker
are created by various extents of turbulence and volatile release
in mesophase(s) of different sizes and viscosity.

8.2 ¥etallurgical Coke

Metallurgical cokes exhibit anisotropy of structure of pore-

wall material. This anisotropy, like that of delayed cokes, is
formed via the mesophase in the fluid phase of carbonizing coking
coal blends. The operational specifications of metallurgical cokes,
related to strength and reactivity, are dependent upon porosity
size and the optical texture of mosaics and of flow anisotropy of the
pore wall material, properties unique to coking coal blends.
 523

COKE FORMATION AND GASIFICATION IN REFINERIES

D.L. Trimm

School of Chemical Engineering & Industrial Chemistry

University of New South Wales
Kensington, NSW 2033, Australia

1. INTRODUCTION

The processing of crude oil in a petroleum refinery is a complex

operation involving many steps. The actual processes involved
depend on the average molecular weight of the feed, with heavier oils
requiring more complex treatment. This results not only from the
presence in heavier oils of significant amounts of inorganic
impurities (S,N,V,Ni) but also from the need to adjust the C:H
ratio - either by removing carbon or by adding hydrogen - during
refining.

The potential for coke formation exists in any system in which

hydrocarbons undergo chemical reaction. Coking may be undesired as,
for example, during naphtha reforming or may be a desired route to
changing the C:H ratio as, for example, in delayed coking. Coke
removal is necessary in both cases but, in the first instance, it
will be an irregular occurrence that is unwanted while, in the
second, it will be a regular planned operation leading to the
production of potentially useful material. The principles are the
same but the objectives and the economics are very different.

The importance of the two operations can be seen from a line

diagram of a refinery designed to process heavy oils (Figure 1).
Steam cracking, not normally considered as a refinery operation, is
included since it provides a good example of one type of coke depo-
sition - removal during hydrocarbon processing (see below). For the
present purposes, the processing of light to heavy fractions may be
considered as desirable reactions which involve unwanted coking,
while the processing of very heavy fractions are based upon reactions
designed to take advantage of coking.
 \J'I
N
~
TYPICAL
REFINERY PRODUCTS

STEAM! r-
CRUDE CRACK'iNG~
I
UNIT
STABll-
-r LI GHT OlEF I NS
I ZER I LPG AND GAS rvAPO~~ J
SECTION ~RECOVERY REFINERY FUEL
STRAIGHT RUN GASOLINE .f. .f. GAS
CRUDE LPG
UNIT NAPHTHA [I HYDRO. 1 I J REFORM.:) - REF~RMATE t C!J REGULAR
I
~I TREATINGI I liNG 1 I Z~
CRUDE ATMOS. PREMIUM
TWR. MIDDLE DISTIllATES HYDRO I ~ GASOLINE
SECTION I T 1TREATING 1 I III SOLVENTS
I HV~ ATMOS. GAS Oil CATALYTIC I GASOLINE a:
1 CRACKING' L---.
I J. AVIATION
Q
VAC. GAS Oil T E
t- FUELS
CRUDE r- Z DIESELS
UNIT
~RACTIONATOR BOTTOMS .c
VACUUM SOLVENT L C!J HEATING OilS
SECTION EXTRACTIONI' ! lUBE OilS
lUBE. I
BASE STOCKS . ...
GASOLINE/NAPHTHA AND MIDDLE DISTlltAT S~ GREASES

H PROPANE I f1ffiL~"1~~~~------~----------~--~~ ASPHALTS

IlEASPHAlTER INDUSTRIAL
FUELS
AND MIDDLE DISTillATES REFINERY
11 FUEL Oil

~DElAYED ~I GASOLINE/NAPHTHAJ AND MIDDLE DISTilLATES J~

- - - , COKER COKE

Figure 1. Flow sheet of a typical refinery

 525

De-coking may involve the physical removal of coke or a series

of chemical reactions represented as

C + °2 CO 2 (1)
C + H2O CO + H2 (2)
C + 2H2 CH 4 (3)
C + CO 2 2CO (4)

Removal of coke in a refinery is based mainly on reactions 1

and 2, the object of the exercise being to remove coke whiie leaving
the catalyst/equipment in a state close to the original. As a
result, problems such as sintering or corrosion must be avoided, and
this may introduce some limitations on coke removal. Such
difficulties have been discussed previously.

Coke formation and removal is very dependent upon the particular

process under consideration. In some cases, coke removal is un-
economic and catalyst/coke dumping is preferred. In others, coke
removal may be an irregular or a regular operation, and it is useful
to consider briefly the fate of the gases produced.

The actual nature of the gas depends on the gasifying agent and
the operating conditions. Disregarding higher hydrocarbons that
may be produced by heating the coke during gasification, reaction
with oxygen or steam produces carbon oxides, hydrogen and water

2C + 02 2CO (5)
C + 02 CO 2 (1)
C + H2O CO + H2 (2)
CO + H2O CO 2 + H2 (6)
2CO + 02 2C0 2 (7)

Carbon monoxide production is often preferred as a result of

the exothermicity of carbon dioxide production, and the consequent
possibility of catalyst sintering during regeneration. Carbon
monoxide/hydrogen mixtures are then fed to a waste heat boiler or
are flared, depending on the regularity of production. Carbon
dioxide, if produced, is vented through a stack. In no case is the
gas used as a chemical feedstock for the production, for example, of
methanol or ammonia. Whether or not this situation will change with
the increasing coke deposits resulting from the future use of
heavier oils remains to be seen.

In view of the particular problems associated with different

refinery processes, it is useful to consider coke formation and
removal in individual systems.
 526

2. STEAM CRACKING

Although steam cracking of naphtha or of ethane to light

hydrocarbons is not normally considered as a refinery process, it
is a useful example on which to base the discussion of some general
problems arising from coke formation and coke removal. This results
partly from the fact that it is an extreme example of a general
problem and partly from the fact that the process has been the
subject of detailed scientific examination (1,2,3).

The process is a gas phase high temperature pyrolysis reaction

in which naphtha-steam or ethane-steam mixtures are converted mainly
to light olefins. Hydrocarbon and steam (ca 0.5 wt ratio) are fed
to a flame heated tubular furnace where cracking occurs at
temperatures of up to 900°C. The process is unselective and a
typical yield pattern is given in Table 1.

On an industrial scale, the reaction is carried out to produce

light olefins, although aromatics are valuable by-products (1).
Coke formation on the reactor tubes is unwanted and requires regular
removal and eventual tube replacement. Full details of the chemistry
and chemical engineering of the proc~ss have been published
recently (1).

The high temperatures of operation require the use of nickel

based alloys for furnace construction with HK40, the traditional
alloy, having a composition of 24.5% Cr, 21.6% Ni, 1.8% Si, 1.1% Mn
and the remainder as Fe. Coke is deposited on the surface both as
a result of gas phase reactions and of surface promoted reactions
(2,3). Eventually the deposit accumulates until the tube blocks or
heat transfer becomes too inefficient (4).

Product Yields
wt %
Hydro.gen 1.2
Metha~e 15.2
Ethylene 33.0
Propylene 11.6
Butadiene 4.7
C+ 26.5
5

Outlet T = 927°C : steam/oil (w/w) = 0.5 : contact time - 0.05 s.

Table 1. Yield of products from naphtha steam cracking.

 527

(fJ

S
M
M
cU
r::o
§
'M
+J
'M
(fJ
o 0
0..00
(1)00
'"CI-<tN
r:: ~ ~
o
..0
H'-''-'
cU cU ..0
U '-' '-'
 528

Regrettably, the carbon also carburises the tubes. Carbon

diffuses into the metallic body and forms metallic carbides which
weaken the structure. The process is easily seen by metallurgical
examination of the near-surface regions (Figures 2 and 3). Penetra-
tion of the carbide into the alloy is strongly dependent on alloy
composition, and comparative carburisation resistances have been
reported.

Decoking is accomplished by removing the tube from service and

using a steam-air mixture to burn the carbon out of the coil. The
mixture is heated to 900-1000oC and is slowly reacted with the carbon
to produce hydrogen and carbon monoxide (better temperature control).
The amount of air in the mixture is slowly increased as the amount
of carbon decreases until decoking is complete. It is not uncommon
for the procedure to take 72 hours, as a result of the need to avoid
too rapid coke removal with associated temperature overrun and tube
melting.

Not surprisingly, the removal of coke and carbides can result

in considerable disruption of the surface. Most of the coke is
removed and part or all of the carbides (Figure 4), production of
the metal oxides occurring. This process results in metal dusting
(metal dropping away from the surface), in severe disruption of

Figure 3. Carburisation of alloys (x 500)

 529

(a)

(b)

Figure 4. Coke removal from alloys

a) Coke deposition on a fresh alloy (x 500)
b) The surface of the alloy after coke gasification (x 500)
 530

Figure 5. Steam cracker tube failure

exterior scale and in enlargement of grain boundaries/cracks in the

metal (Figure 4). As a result, the tube is weakened and, eventually,
will fail (Figure 5).

The solution to the problem lies partially in the correct

selection of the alloy and partially in the possibility of depositing
protective layers on the tube surface. Aluminium (5) and silicon
(6) layers have been found to be efficaceous in reducing carburisa-
tion.

Although steam cracking is an extreme example, the problems

involved are typical of many in a petroleum refinery. High tempera-
ture coking and carburisation is possible in a variety of equipment,
and the associated corrosion problems present an expensive hazard
to the operator.

3. CATALYTIC CRACKING

Catalytic cracking is by far the most important catalytic

process in commercial operation, with nearly one million tons of oil
processed daily in the United States alone. The process is used to
convert heavy oils into more valuable gasoline and lighter products,
with the unwanted deposition of small amounts of coke on the catalyst.
Cracking is carried out using a fluidised bed reactor, with coked
catalyst being circulated through a gasifier for regeneration.

Two types of catalyst are used - amorphous silica-alumina and

crystalline alumino-silicate zeolites - their mode of action
 531

involving acid catalysis. The main desired reaction can be

represented as

RCHZCH = CHCHZR' + H+A-~ RCHZCH Z - CHCHZR'A-

RCHZCH Z - CHCHZR' ~R'CH = CH Z + CHZCHZR

± + -
CHZCHZR + A ~CHZ = CHR + H A

As a result of the high temperatures (ca SOODC) required for

cracking, a variety of other acid catalysed reactions are also
promoted (7). Amongst these is the sequence of polymerisation/
cyclisation/dehydrogenation reactions leading to coke. A typical
product spectrum is given in Table Z.

The desired reactions are endothermic and, as a result, coke

is burnt in an associated gasifier to produce carbon oxides, steam
and heat. A typical reactor configuration is shown in Figure 6,
from which it can be seen that the balance between heat lost during
reaction and heat gained during regeneration is maintained by con-
trolled circulation of the catalyst. Considerable detail as to the
chemistry (7,8) and engineering (9.10) of cracker plant has been
published.

It is this balance between heat loss during the endothermic

cracking reactions and heat gain during exothermic regeneration

Component wt %

Gas 4.5

C3 ane and ene 3.3

Isobutane Z.6

C4 ane and ene 3.5

Cs + gasoline 40.Z

Light cycle oil 33.Z

Decant oil 7.7

Coke 5.0

Reactor temperature at ca 500°C and lS-ZO psig

Table Z. Typical yields from cat. cracking a topped crude feed

 532

that controls much of the operation of a cat. cracker. Regenera-

tion is carried out by combustion in air, and the regenerator
temperature and coke burn off are controlled by the air feed rate,
the value of the control temperature being dictated by the heat
balance (usually ca 620-730°C).

Two types of regenerator operation are used industrially.

Older units were designed to combust coke to carbon monoxide since
this minimised air compression costs for the combustion. Carbon
monoxide was then burnt in a waste heat boiler. Newer units tend
to be designed for the combustion of coke directly to carbon
dioxide. This results in less residual coke on the catalyst
(0.05 wt% compared with up to 0.4 wt%) and in more efficient use of
the heat evolved, but may cause problems if the regenerator tempera-
ture limits are exceeded. As a result, some design changes in the
plant are required (10).

Modification of the catalyst may also be advantageous if direct

combustion is desired. Addition of small amounts of platinum to
the catalyst (ca 0.05 wt%) improves efficiency of regeneration by
complete combustion (11). Traces of arsenic, added to the feed,
are also used (12) but these are designed to passivate nickel and
vanadium deposited on the catalyst during cracking rather than to
improve cracking or regeneration.

Although overheating may damage the catalyst, attrition during

cracking and regeneration is a much more common cause of catalyst
failure. The cracking-regeneration sequence works well, and coke
removal is an efficient and necessary operation.

Waste heat boiler Products

ing steall
If-~.;.g;;=

Regen.
catalyst

~_"-'" Air

Figure 6. A line diagram of a catalytic cracker

 533

4. CATALYTIC REFORMING

Modern automobile engines require high octane gasoline and

this. in turn. requires more alkyl aromatics and branched chain
hydrocarbons in the product. This is achieved by catalytic
reforming (13).

The process involves several reactions which are catalysed by

precious metal(s) supported on alumina. These reactions include:

Desired: Paraffin isomerisation and cyclisation

Olefins are converted to paraffins
Naphthenes are converted to aromatics
Aromatics are left essentially unchanged

Undesired: RemovAl of side chains from aromatics or

naphthenes
Cracking of paraffins and naphthenes
Coke formation

Both components of the catalyst are required. Paraffins are

dehydrogenated to olefins on the metal. and olefin isomerisation
and cyclisation are promoted on the acidic support. The isomerised
products are then re-hydrogenated on the metal.

Industrial operations are carried out under conditions dictated

by the thermodynamics of the system. Typical reactions are
considered in Figure 7 and Table 3. from which it can be seen that
the desired reactions are favoured at high temperatures and low
pressures - exactly the conditions that also favour coke formation.
As a result. it is necessary to choose operating conditions such
that a balance is maintained between the desired reactions an.d coke
formation. Although the use of bi-metallic or multi-metallic
catalysts improves the situation (14). periodic shut down for
regeneration is still required.

Catalytic reforming in the refinery may be a continuous. a

cyclic or a semi-regenerative operation (10). A line diagram for
platforming. a typical industrial process operated in the semi-
regenerative mode. is given in Figure 8. Feed (heavy straight run
gasoline and 80-lS0oC cut naphtha) is pretreated to minimise sulphur
and nitrogen. mixed with hydrogen and heated to ca 500 Q C before
entering the first reactor. The major reaction in this reactor is
the endothermic dehydrogenation of naphthenes to aromatics, and the
gases must be re-heated before entering the second reactor. A.s the
gases pass through the reactors. the reaction rates decrease. the
catalyst charge becomes larger and the need for re-heat decreases.
Product leaving the last reactor is cooled to cond,ense liquids.
the gases being fractionated and hydrogen recycled through the'
 534

H]C-CH=CH2 1S
+ J!L.
CH
I
3
H)C-CHr CH 3 _ 3( + 4H2
10
CD, ,-
-
C~- CH-CHf (Hi (H 3 tt<Z> " /'
S ".Y,'
~ ID
© / "

(~CH3 ; 1' . C6 H14=

~H2
+ 3H2
0
• I

~ttJ teD ®U .I
a.
CH3-CH- CH-CH3 ¥

.;;'" - S ®"/ "

ij~ 0
~ i

-10

Figure 7. a) (above) Typical reaction

paths in catalytic
reforming.
b) (right) The thermodynamics -20&"'z;"'4~0'"""0~60~0~8~00=--~10~OO""""
of reforming reactions. Temperature. oK

system. Typical operating conditions and conversions are summarised

in Table 4.

Coke removal is usually required at 3-6 month intervals and must

be carefully controlled to avoid sintering the metal or the support.
At the same time, chloride - needed to increase the acidity of the
support (13) - can be replaced, the chloride also being believed to
aid in re-distribution of the metal (15). A typical sequence of
regeneration involves

a) The bulk of the coke is removed by passing air (oxygen partial

pressure 0.007-0.175) through the reactor at <425°C.

b) Residual coke is removed by heating to ca 425-485°C in the

presence of the same partial pressure of oxygen.

c) The oxygen partial pressure is increased to greater than 0.175

and the temperature increased from 485°C to 650°C. Halide,
together with steam, is admitted during the final stages to ensure
even distribution of metal.

d) The catalyst is reduced at 310°C and brought back on line.

The critical parameter in this regeneration procedure is temp-

erature, and control by control of burn off conditions is vital.
 535

Reaction
Reaction I1H Effect Effect Effect Effect
Rate
of high of high of high on
pressure tempera- space octane
ture velocity

Hydro- Slowest Exothermic + + +

cracking

Isomeri- Rapid Mild exo- nil + +

zation thermic

Cycliza- Slow Mild exo- + +

tion thermic

Naphthene Rapid Mildly

isomeriza- exothermic nil + slight -
tion

Naphthene V.Fast Quite endo- + +

dehydroge- thermic
nation

Coking Slow Endothermic +

Table 3. Some basic relationships in Catalytic Reforming

HYDROGEN RECYCLE NET HYDROGEN

COMPRESSOR
2400-4500 kPa 1700 - 3400 kPa

500-525"( SOO-525°C FRACTIONATION

NAPHTHA

Figure 8. A line diagram of a semi- regenerative platforming unit

(13) •
536

Component Feed Vol % Product Vol %

Paraffins 45-55 30-50

Olefins 0-2 0

Naphthenes 30-40 5-10

Aromatics 5-10 45-60

RON 90 98

Pressure 100-350 psig H2 /feed 4,000-8,000 scf/bbl

LHSV 2-3 T l50-200°C (final bed)

Table 4. Typical reformer operations

The coking and regeneration of reforming catalysts has been the

subject of more detailed examination (15,16), partially as a result
of industrial interest in the system and partially as a result of
the multi-component nature of the catalyst. Coke is formed both on
the metal and on the support and, indeed, is a welcome phenomenom
to a limited extent in that it limits excessive cracking (17). The
deposition and removal of more massive deposits depends on the type
of coke (which is dictated by the surface on which it is formed)
and the rate of coke formation and removal - which is affected by
the catalytic action of that surface and by the possibility of, for
example, spillover of oxygen from the metal to the support during
gasification (18). As a result, removal of coke on or near to
metallic particles can be expected to be faster than gasification
of coke on the support remote from the metal. Valuable studies of
the mechanism and kinetics of the processes involved are currently
in progress (16).

Coke removal is an infrequent operation and product gases -

often rich in carbon monoxide as a result of the need to limit
temperature during regeneration - are usually flared.

5. TREATMENT OF HEAVY ENDS : PYROLYSIS

Many refineries rely on pyrolytic processes to upgrade heavy

ends and these processes produce coke to be sold as a valuable
product or coke to be gasified. The most widely used processes are
delayed coking, flexicoking/fluid coking and visbreaking.
 537

The delayed coking process was developed to mlnlmlse refinery

yields of residual fuel oil by severe thermal cracking of stocks
such as vacuum residuals or thermal tars. The process is, in fact.
a derivative of operating experience for which little scientific
information exists. Early operations showed that severe thermal
cracking led to coking in the heaters. By gradual evolution. it
was found that heaters could be designed to raise residual stock
temperatures to above the coking point without significant coke
formation. This required minimal retention time in the h~aters.
If the hot stock was then held in a drum for longer periods. coking
occurred. together with the production of lighter material that
could be used. for example. as hydrocracker feedstocks. Once
sufficient coke had accumulated, the solid material was removed from
the drum. heated to ca 1100-1200 o C to remove residual volatile matter.
and sold as petroleum coke. Typical operating conditions and yields
are given in Table 5 for a decant oil feed. A more complete descrip-
tion of t~e process is given in references 10, 19 and 20.

Vis breaking is a relatively mild thermal pyrolysis used mainly

to reduce the viscosities and pour points of vacuum tower bottoms
to meet fuel oil specifications and to increase gasoline yields
or cat. cracker feedstocks. The prime target for visbreaking are
long paraffinic side chains attached to aromatic rings. since these
are the main cause of high pour points and viscosities in paraffinic
base residua. As a result. thermal cracking during visbreaking is
milder. and coke formation is less. Visbreaking may be carried out
in a coil furnace or a soaker drum (10,21). coke being deposited
either in the tubes or in the drum. Coke removal from the coil is
by gasification (compare, for example. with steam cracking) or by
physical means from the soaker drum. Typical yields are summarised
in Table 6. Further information is presented in references 10 and
21.

Weight percent yields based on feed

Gas 18%
Naphtha 12%
Gas oil 9%
Steam out oil etc. 14%
Coke 45%

Based on a 24 h run with the coke drum maintained at 90 psig and

the drum inlet temperature at 515°C.

Table 5. Delayed coking of a decant oil (10.19.20).

 538

Yields (wt %)

<C 4 2.5%
C5+ naphtha 5.9%
Gas oil 13.5%
Tar 78.1%

Based on a 6-18 month soaker visbreaking run maintained at

pressures up to 750 psig and drum inlet temperature of ca 420°C.

Table 6. Visbreaking Kuwait Long Residue (10,21)

In delayed coking and soaker vis breaking, coke removal is by

physical means and is an unwanted irregular occurrence. In Flexi~
coking or Fluid coking, this is not the case, and operations have
much in cornmon with catalytic cracking.

A simplified flow diagram of the F1exicoking process is shown

in Figure 9. A heavy oil such as vacuum resid is preheated to
ca 300-370 o C and sprayed into a reactor where it is contacted with
a hot f1uidised bed of coke particles (ca 500~5500C). Pyrolysis
occurs to produce coke deposition onto coke and lighter products
which are separated and processed. The coked coke is stripped with
stearn to remove volatile products and taken through a heater (600 0C)
to a gasifier, where some coke is removed by gasification in air
and stearn. Coke is then circulated to the reactor for continuation
of the process.

Fluid coking involves basically the same process but uses only
a reactor and a burner. Coke produced in the reactor is combusted
only to the extent required to satisfy the heat requirements of the
reactor (about 25% of coke produced), the remainder being withdrawn,
from the system.

Typical yields from the two systems are presented in Table 7.

Coke gas produced by the F1exicoker has the approximate composition
shown in Table 8. Further information about the two processes is
presented in references 10 and 22.

The actual gasification of coke is carried out using a mixture

of air and steam in a third f1uidised bed (Figure 9). The kinetics
of the process are difficult to predict owing to the diverse nature
of the coke deposited and the possibility of nickel or vanadium,
originally present in the feed, depositing with the coke and
 539

F1exicoking Fluid coking

<C 4 (wt %) 13.1 13.1

C5 +(LV%) 65.1 65.1
Gross coke (wt %) 30.7 30.7
Net coke (wt %) 0.61 24.9
Coke gas (LV %) 18.8

Table 7. Comparison of F1exicoking and Fluid coking yields from

an Arabian Heavy 1050°F + oil (10)

mol %
H2 15
CO 20
CH 4 2
CO 2 10
N2 53

Table 8. Coke gas composition after H2 S removal dry basis (10)

Coke gas

Air

Figure 9. A line diagram of a Flexicoker (22).

 540

catalysing gasification.

6. OTHER PROCESSES

A refinery is used to process hydrocarbons, often at high

temperatures and variant pressures. As a result, there are a
variety of reactions that involve coke formation and the necessity
of coke removal. Hydrotreating, hydrocracking, alkylation, poly-
merisation and lubricating oil processing are all cases in point.
The reactions involved are, however, typified by those discussed
above, the choice between catalyst replacement and catalyst regenera-
tion depending mainly on the cost of the catalyst. In no case is
the gas produced by coke removal used for any other purpose than
waste heat recovery or flaring, largely as a result of the in-
frequency of the operation. Whether or not the increased coking
resulting from the use of heavier feedstock will change this
situation remains to be seen, but the feedstocks processed are
unlikely to change dramatically before the 21st Century.
 541

REFERENCES

1. Albright, L.F., Crynes, B.L. and Corc.oran, W.H. (Eds) ,

Pyrolysis : Theory and Industrial Practice: Academic press:.
NY (1983).
2. Trimm, D.L., Ibid P 203.
3. Albright, L.F. and Tsai, T.C., Ibid P 233.
4. Trimm, D.L., Chem. Eng. Processing 18 (1984) 137.
5. Nap, C., Van Haarlem, A. and Emile, a.M.H., U,S. Paten.t
3,827,967 (1974).
6. Brown, D.E., Clark, J.T.K., McCarroll, J.J. and Sims, M.D.,
Ger. Patent 2,613,787 (1976).
7. Germain, J.E., The Catalytic Conversion of Hydrocarbons,
Academic NY (1969). .
8. Gruse, W.A. and Stevens, D.R., Chemical Technology' of
Petroleum, 3rd Edn., McGraw Hill NY (1960) ..
9. Saxton, A.L. and Worley, A.C., Oil Gas Jni. 82 (May 1970).
10. Gary, J.H. and Handwerk, G.E., Petroleum Refining, Technolog¥
and Economics, 2nd Edn., Marcel Dekker NY (1984).
11. Thomas, C.L., Catalytic Processes and Proven Catal¥sts,
Academic NY (1970).
12. Dale, G.H. and McKay, D.L., Hydrocarbon Process. 56 (1979) 97.
13. Gates, B.C., Katzer, J .R., Schuit, G.C.A., Chemistr¥ of
Catalytic Processes, McGraw Hill NY (1979).
14. Dowden, D.A., Specialist Periodical Reports, Catalysis (The
Chemical Society, London) 2 (1979) 1.
15. Bishara, A., Murad, K.M., Stanislaus, A., Ismail, M. and
Hussain, S.S., Appl. Cat. 7 (1983) 337, 351.
16. Parera, J.M. et aI, Private Communication.
17. Parera, J.M., Jablonski, E.t., Cabrol, R.A., Figoli, N.S.~
Musso, J.C. and Verderone, R.J., Appl. Cat. in press.
18. Dowden, D.A., Specialist Periodical Reports: Catalysis,
(The Chemical Society, London) 3 (1980) 136.
19. Dielsi, D.P., Miller, D.J. and Cox, N.R., Energy Progress 3
(1983) 193. .
20. Rose, K.E., Hydrocarbon Process 50 (1971) 85.
21. Rhoe, A. and de Blignieres C., Hydrocarbon. Process. 131 January
(1979).
22. Allan, D.E., Chem. Eng. Progr. 77 (1981) 40.
 543

THE REGENERATION OF COKED CATALYSTS

D.L. Trimm

School of Chemical Engineering & Industrial Chemistry

University of New South Wales
Kensington, NSW 2033, Australia

1. INTRODUCTION

Any catalyst that promotes reactions involving hydrocarbons has

the potential to coke. Coking is unlikely in reactions such as
oxidation but occurs frequently with other types of reaction.
Depending on the speed of coking and the value of the catalyst,
coke removal or catalyst replacement is required. For a reasonably
costly catalyst and for coke formation which is other than very slow,
coke removal will be preferred.

This may be achieved by gasification of coke, and four

possibilities are known

C + °2 CO 2 (1)

C + H2O CO + H2 (2)
C + 2H2 CH 4 (3)
C + CO 2 2CO (4)

In practice, gasification with air or with air/stearn mixtures

is used almost universally. The reactions present problems, for
example, as a result of the fact that reaction 1 is very exothermic,
but careful control of reaction conditions leads to effective
catalyst regeneration.

Coke deposited on catalysts is a collective name for carbon-

aceous deposits that can be formed in several ways (1). As a result,
their nature varies and the rates of gasification vary. It is useful
to consider briefly the formation and nature of the deposits before
attention is focused on means of gasification.
 544

2. TYPES OF COKE

Three g'enera1 types of coke can be identified (2,3). (Figure 1)

At high temperatures, free radical reactions in the gas phase occur,
and high molecular weight polymers and carbon may be produced. These
may deposit on any surface and rearrange/dehydrogenate further. For
clarity, these deposits are known as gas phase carbon (1,2,3).

Hydrocarbon intermediates can also adsorb or condense on inert

surfaces, where dehydrogenation/growth/ordering processes can occur
to produce carbon. This surface carbon is independent of the
chemical nature of the surface.

Finally, it is well known that catalytic carbon - carbon

produced as a result of reactions promoted by a catalyst - may be
formed on different materials (1,2,3). Carbon production can be
favoured at temperatures much less than those at which gas or
surface carbon is produced, and is promoted - to greater or lesser
extent - by the nature of the catalyst.

The situation is made lUore complex by the possibility of; inter-

action. Thus, for example, a catalytic surface may produce inter-
mediates which desorb to react further and form gas phase coke (3).
Various types of coke, once deposited, can dehydrogenate and order
on a surface, leading to material which has very similar chemical
composition. As a result, it is often difficult to establish the
origin of the deposit.

There are differences, however, in the as-deposited material,

with larger amounts of tars being favoured in the gas phase (3) or
on acidic catalysts (4). This results from the po1ymerisation
reactions involved, where free radicals or ionic species combine to
build up high molecular weight tars and carbon. (5). Whether or n.ot
the processes occur concurrently or consecutively is open to
question,(6) but the tar like materials, although difficult to
distinguish from coke, possess properties which make them behave
differently during gasification. Thus, for example, high tempera-
tures produced during gasification may cause tars to vo1ati1ise,
to pyro1ise (3) or to melt (1). Volatilisation leads to catalyst
regeneration. Pyrolysis leads to gaseous products and coke (3),
which may be removed by gasification. Melting may cause the tar to
flow across the surface to block pores and to give totally diff;erent
gasification kinetics. The net result is that the gasification
rates can be very different between different tars and between
tars and coke.

3. COKE FORMATION ON DIFFERENT SYSTEMS

Since the various constituents of coke can behave so differently,

it is rewarding to consider which systems favour which kinds of; coke.
 545

dehydrogenation/ordering

t
gas phase tars and carbon

t
free radical reactions

t
hydrocarbons

free radical reacti~ ~ catalyst promoted

adsorption?
reactions

\
condensation?

surface carbon
j catalytic coke

t
dehydrogenation/ordering
~
dehydrogenation/ordering

Figure 1. Different types of coke

Metallic catalysts, considered elsewhere, produce essentially only

ordered or disordered carbon, as a result of the fact that coke is
formed at temperatures lower than those that favour gas phase
reactions and of the fact that the metals promote dehydrogenation
of and coke formation from adsorbed intermediates (1).

Oxidic or su1phidic catalysts are responsible for more complex

deposits, largely as a result of the feedstocks and of the reactions
by which they form products or coke. In brief, the solids are
generally acidic catalysts that can promote po1ymerisation by
reactions such as (8)
+ -
H3 C - CH CH 2 + H+A- = H3 C CH - CH 3 + A
+
H3 C CH CH 2 + H3C - CH - CH3 = H3 C,
+
CH - CH2 - CH - CH3
/
H3 C
and cyc1isation, for example, by reactions such as

CH - CH 2 - CH 2 - CH 2 - CH2 - CH3
carbonium ion + A
 546

A + carbonium +HA
+-

Coke formation results from the steady increase in molecular

weight of products by these types of reaction, carbon being formed
eventually by the dehydrogenation of a high molecular weight polymer.
It is not surprising, then, to find that coke deposits contain both
polymers and carbon, the polymer often being of sufficien.tly high
molecular weight to condense under normal operating conditions (7).

Variations of feedstock are particularly important for coke

formation on sulphided catalysts, which are used to hydrotreat heavy
oils (8). In this case, the molecular weight of the feed itself is
high, and the production of polymers and carbon is easier. Coking,
as a result of tar or carbon deposit, occurs easily and is widely
observed (8).

Reactions proceeding at high temperatures or in situations

where there is a high void volume in the catalyst bed (e.g. a fluid-
ised bed) may also be accompanied by the formation of tars and carbon
in the gas phase (3). The coke is formed in the gas phase as a
result of free radical polymerisation and cyclisation reactions and
tars and/or coke are produced (9). There is some disagreement as to
whether the coke is formed only from the polymer or whether there
are two routes - one via polymer and one direct (6) - but both tars
and carbon are formed in such systems. The exact temperature of
formation depends on the reactivity of the feedstock, acetylene -
for example - forming coke at much lower temperatures than methane.
The coke will deposit on the catalyst, on which catalytic coke may
also be formed by a separate route.

These arguments illustrate the point that, although it is

possible to set out general rules for the production of a particular
type of coke, particular systems have their own characteristics.
Metallic catalysts do favour the production of carbon but, if the
system has a high void volume, tarry material will also be produced.
Gas phase coke tends to contain high molecular weight polymers but,
if these condense on a metallic surface, dehydrogenation to form
carbon can easily occur. As a consequence, coke formation in
individual systems must be regarded as a particular rather than a
general problem, and one in which all factors affecting the nature
of the coke deposited must be considered.

4. THE KINETICS OF COKE REMOVAL

The occurrence of the different types of coke can have further

effects on the kinetics of regeneration. Coke deposits on a catalyst
 547

at a rate affected by the catalyst and subsequently dissolves in,

is gasified from or orders and encapsulates the surface (1). The
catalytic action of the underlying material continues while the
reacting gas is in contact with the material. As soon as the surface
is totally encapsulated - by catalytic, surface or gas phase coke,
the kinetics of deposition is governed by that of coke deposition
on coke (10,11). During gasification, the reverse occurs. The
kinetics of gasification are essentially uncatalysed until the
point where underlying material is exposed, at which point any
catalysis of gasification begins to be significant. As a result,
it is not unexpected to observe two kinetic regimes during gasifica-
tion - the uncatalysed gasification of coke followed, eventually,
by the catalysed gasification. In this context, it is useful to
remember that most transition and precious metals are good catalysts
for coke gasification (12) as are some metal oxides (13). Other
metal oxides (e.g. alumina) do not appear to catalyse gasification.

In addition, of course, coke may have been deposited in the

pores of a catalyst or carbonaceous tars may flow over the surface
to block pores. This may cause the kinetics of gasification to be
further complicated by diffusional effects that disappear as gasifica-
tion proceeds.

The regeneration of catalysts is affected, then, by the nature

of the coke, the availability of a catalytic surface and the
possibility of diffusional resistances which, in turn, depend on
the texture of the catalyst and the nature of the coke. As a
result, it is necessary to consider coke gasification for individual
systems, bearing in mind the general problems that may arise. These
may include related problems which are not directly connected to
regeneration, but which have a profound effect on the conditions
chosen for regeneration. It is useful to consider these briefly.

5. PROBLEMS RELATED TO GASIFICATION

Catalyst regeneration by combustion with oxygen is a very exo-

thermic reaction. As a result, the temperature of the catalyst may
rise, and sintering may occur (13).

Heterogeneous catalysts are prepared with high surface areas

in order to maximise gas-solid contact. High dispersion is a
thermodynamically unstable state for any material and a driving
force to minimise surface area exists. Migration of a solid may
occur by different means - volume or surface diffusion and/or
evaporation - condensation (14) - but solids will sinter given
sufficient energy to overcome the activation barrier.

Perhaps the most important promotion of sintering results from

too high temperatures. Tamman (15) observed that sintering becomes
significant at temperatures greater than one-half of the melting
 548

point of a solid. Huttig (16), based on predictions made on the

basis of surface diffusion effects, suggested that one-third of the
melting point temperature was a more realistic temperature for
sintering to become important. A catalyst and a support is carefully
chosen with this in mind (13) but high temperatures generated, for
example, during oxidative regeneration can promote rapid sintering
of the catalyst or the support. As a result, the surface area and
the porosity of the catalyst may decrease, resulting in less gas-
solid contact and in a less active catalyst.

The nature of the gas atmosphere may also have a profound

effect on sintering. Thus, for example, the sintering of metals is
accelerated in different ambient atmospheres. Silver sinters faster
in oxygen (17) while nickel and iron sintering may result from
evaporation - condensation of carbonyls formed from contact between
carbon monoxide and the metal (18). Again, oxides - in particular -
are prone to excessive sintering in atmospheres containing steam (13).
Such sintering may result from the formation and migration of
hydroxides (e.g. silica)(19), or from phase transformations accelera-
ted by steam (e.g. from y- to a-alumina (20». Luckily, such
sintering generally occurs only at reasonably high temperatures and
over long periods of time (13,14), but excessive catalyst-steam
contact should be avoided.

Although it is possible to establish general rules such as the

avoidance of excessive temperature or the avoidance of too long
catalyst-steam contact, particular problems may arise in individual
systems. Carbonyl formation, or the sintering of silver in oxygen
are good cases in point. As a result, individual systems must be
considered carefully before regeneration by gasification is attempted.

It should also be emphasised that the regeneration of coked

catalysts does not necessarily mean that the original catalyst
activity is recovered. Catalyst coking, particularly during the
hydroprocessing of heavy oils, is accompanied by the deposition
of S, N and metal containing compounds which, on gasification of
coke, may be converted to gaseous products (sulphur or nitrogen
oxides) or to solid deposits (metal oxides) (8). These later
deposits foul the catalyst and reduce the activity even after
removal of coke. At least some of these deactivating effects may be
avoided by careful control during coke deposition and removal (1),
but some permanent deactivation is often observed. Procedures for
minimising such deactivation are beyond the scope of this paper,
but have been reported in the literature (1).

One further difficulty may arise during regeneration, as a

result of the possible carburisation - decarburisation of alloys
used to fabricate reactors or heat exchangers. Hydrocarbon
processing is often carried out at high temperatures and pressures,
and this requires that equipment be manufactured from nickel
 549

containing alloys. HK40 is the classic example, the alloy

containing 21.5% Ni, 24.5% Cr, 2.7% additives, the remainder being
iron. During the processing, coke is deposited on reactor
surfaces and, if the temperature is high enough, the alloys will
carburise (3). That is to say that carbon will migrate into the
alloy and form metal carbides. Subsequent gasification of coke
results in the destruction of these carbides, thereby weakening the
surface and causing metal dusting (2). This problem will be
discussed in more detail in the context of steam cracking as a
refinery process, but the carburisation - decarburisation cycle is
a potential problem related to coke removal, and one which must be
considered in the design and operation of catalyst regeneration
procedures.

Finally, it is interesting to comment on the fate of gases

generated during gasification. Carbon monoxide and hydrogen, which
could be generated by gasification

2CO
CO + H2

are valuable feedstocks for the production of methanol (19), or of

other petrochemicals (21). Such gases, even if produced during regu-
lar catalyst regenerations, are rarely used in this way. Instead
the gases are either vented or used to fuel a waste heat boiler.
The major reason for this lies with the relative infrequency of
catalyst regeneration and the intermittent nature of the supply of
gas. Even in catalytic cracking, a process in which continuous
regeneration is required, emergent gas is not used as a chemical
feedstock. Whether or not this situation will change depends on
the regularity of supply of gas and the requirements for heat
generation. Processing heavier feedstocks will cause more coking
problems and may result in a more constant supply of potential
feedstock.

6. THE REGENERATION OF COKED CATALYSTS

It is obvious from the discussion above that removal of coke

from catalysts may be achieved by gasification (reactions 1-4).
The kinetics of the process depend, however, on the nature of the
coke which, in turn, depends on the mechanism of formation of the
coke.

Further, it is clear that the procedure adopted is dependent

upon the nature of the catalyst on which coke is deposited. Coke
removal cannot be carried out at temperatures at which the catalyst
will sinter.
 550

It is not surprlslng, then, to find little general information

on the regeneration of coked catalysts and much information which
is pertinent only to particular systems. Particular systems have
been explored in depth (1,2), but the findings are relevant only to
those systems, and different systems - unless they are very similar -
require detailed investigation of their own.

The background information exists in the literature. Conditions,

for example, for catalyst sintering are known for different
materials (13) and procedures for decoking are established. By the
nature of the problem, detailed investigation is required for each
system.

It is useful, therefore, to consider individual reactions and

this is done in the context of catalysts used in a petroleum
refinery. Before moving to these, however, it is interesting to
consider one system as an example of the kinds of problem that can
arise. This is the regeneration of copper based catalysts used for
the selective hydrogenation of acetylenes in olefin feedstreams.

7. THE SELECTIVE HYDROGENATION OF ACETYLENES

One very interesting example of different types of coking and

their effect on the regeneration of the catalyst has been reported
in connection with the selective hydrogenation of acetylenes in an
industrial butadiene feedstock (22). 1,3 butadiene is required for
the production of synthetic rubber by the anionic solution polymeri-
sation route (23), but acetylenes present at levels above 100 ppm
are efficient poisons for the catalyst. As a result, it is
necessary to hydrogenate the acetylenes, while leaving the
butadiene unchanged. A typical feedstock contains 1,3 butadiene
(44-55%), 1 butene (25-35%), 2-methyl propene (ca 10%), 2-butenes
(ca 8%), butanes (ca 60%), 1,2 butadiene (ca 0.2%), vinyl acetylene
(ca 1.3%) and butynes (ca 0.4%). The catalyst must be active and
selective and yet may easily be poisoned by polymers of acetylenes
or by coke produced from these polymers.

Copper based catalysts were found to be effective for the

reaction but, if conversion was plotted as a function of time on
line, deactivation was observed in all cases (Figure 2).
Following traditional practice, temperature was increased after
initial deactivation (Figure 3). The activity could be regained,
but subsequent deactivation was rapid and, eventually, permanent
deactivation occurred.

The effect of increasing hydrogen was then examined, with the

object of promoting polymer hydrogenation (Figure 4). It is seen
that deactivation is much less and, in fact, permanent deactivation
was avoided. As would be expected, the selectivity of the reaction
decreased, but deactivation was minimised.
 551

~100 ~ 100
>' >
z
~ 80
'-'
0
w 80
UJ
UJ
Z z
u.J
~60 60
...
..J
~ ~
>-
UJ
~40 w 40
« «
..J
~ 20 ~ 20
:J:
>- .....
Z
UJ >
0 0
0 20 40 60 80 100 0 20 40 60 80 100
TIME ON II NE (h) TI ME ON II NE (h)

15 30
~ ;e
~
~
;> 12 :> 20
z z
0
0 w
w
9 u.J 10
UJ
z z
a
UJ
UJ

... 6 « 0
Cl
« >-
~
~
co
, 3 ~ -10
".,
N
-
0 -20
0 20 40 60 80 100 0 20 40 60 80 100
TIM E ON LIN E (h) TIME ON LINE (h)

Figure 2. Deactivation of a copper based catalyst.

Catalyst = 15% Cu/yA1 203

T = 220°C
H = 10.4
2/acetylenes
Space velocity = 415 v/v/h
552

~o100
!!...~ ~
:>
o ,II "II
II
II
~~ 80 -6
II
~ I
00 u
Uu I
UJ
z -12 I
!!:!
o I I I

N
I

NN
«
~ -18
co
I I I
-24 I I
I
N
II 11 I
i II II I
-30
30 60 90 120 150 o 30 60 90 120 150
TIME ON LINE (hI. TIME ON LINE (hI

~
o
>
z
9
I

,,
~I
II"
·,~N',,
I
• I
I 140

130
II
"

I
II
II
/I"
il
II IU
,
II,
1\ I
II I

8 6 I
n
II '
I u
120 'I L_JI
~I"I
II I
I
I , ,I H I
0_

, I
~I
UJ

·,
Z Cl!
, I
H
, II 'I I
UJ
is 3 I , ~

« I
I
I
", UJ
I-
110
I
"II I
II II I
i~
I- I
~ II I
co
I
0
Vi,, II I 100 ,I II " I I II
""
II II I
......
II I I ,
I
m
I I 90 _____IIIIL-...&I....II'----_
II I
-3
~~

o 30 60 90 120 150 o 30 60 90 120 150

TIME ON LINE (hI TIME ON LINE (h)

Figure 3. Effect of temperature on deactivation.

Catalyst as in Figure 1.
H = 7.2
2/acetylenes
Space velocity = 460 v/v/h
Temperature = variant (see plot)
 553

~ ;e 100
'l1II
_0 11,\1 0'-10
" II ~
~~
uS
80
II
II I
:>
~-14
I
"II! I
II
w
z:w
II I
u
w
II
wZ
...Jw 60 II I Z
w
~¢ II I II 0- 18
~ ~ 40 II I II «.....
Uu
««
II
II
I II
I
IIII
;:)

- I 22
...J...J II I
CD-
~~ 20 1\
....-
Zx:: I I N
>w ..... II I
0 -26
0 10 20 30 40 SO 60 0 10 2030 40 SO 60
TIME ON LINE [hI TIME ON LINE [hI

5 II I 11 II
;e II I a. II II
0
4- II I 10 II II
II I w
:>
z 3 II z 9 II 'II"'
11
~I
w
0 II ...J II
"
u
..... 8
~
~
2 w tt II
w
Z u I II
I «...... 7 IH
w
II\J I
II
0
« II z 6
w II I II
..... 0 II I \!J II I II
I 0 5
II II I
;:)
al
-1 II I
0: II
I
1'1"1
.- I I
0
~
x::
4 II I II
-2 3
0 10 20 30 40 50 60 0 10 20 30 40 50 60
TIME ON LINE [hI TIME ON LINE [h)

Figure 4. Effect of hydrogen on deactivation.

Catalyst as in Figure 1.
T = 133°C
Space velocity = 800 v/v/h
HZ/ ace t y 1 enes = variant (see plot)
 554

When permanent deactivation occurred the catalyst could not be

regenerated by hydrogen but could be regenerated by heating in air/
nitrogen mixtures (ca 10% oxygen) at 400°C. Conditions were
carefully controlled to avoid over-heating and copper sintering.

It is clear that deactivation may involve two types of species -

polymer and coke. Polymer was formed at low temperatures and could
be removed by increasing temperature (part volatilisation) or by
hydrogenation. Coke was formed at higher temperatures and could only
be removed by oxidation.

As an example of the various types of coke and their effect on

catalyst deactivation/regeneration, the system is ideal. A plant
operator, on the other hand, would describe the system differently!
 555

REFERENCES

1. Trimm, D.L., Appl. Catalysis 5 (1983) 263.

2. Trimm, D.L., Chern. Eng. Proces. 18 (1984) 137.
3. Pyrolysis: Theory and Industrial Practice, Ed Albright L.F.,
Crynes B.L. and Corcoran, W.H., Academic NY (1983).
4. Figoli, N.S. Beltramini, J.N., Martinelli, E.E. and Parera,
J.M., J. Chern. Tech. Biotechnol. 32 (1982) 445.
5. Appleby, W.G., Gibson, J.W. and Good, G.M., Ind. Eng. Chern.
Proc. Des. & Dev. 1 (1962) 102.
6. Graham, S.C., Homer, J.B. and Rosenfeld, J.L.J., Proc. 10th
Internat. Shock Tube Symp. (Kyoto) (1975) 621.
7. La Cava, A.I., Bernardo, C.A. and Trimm, D.L., Carbon 20
(1982) 219.
8. Weisser, O.,and Landa, S., Sulphide Catalysts: their
properties and applications. Pergamon (NY) and F. Vieweg
(Braunschweig) (1973).
9. Turner, C.J. and Trimm, D.L., J. Chern. Tech. Biotechnol. 31
(1981) 195.
10. Rostrup Nielsen, J.R. and Trimm, D.L., J. Cat. 48 (1977) 155.
11. Trimm, D.L., Cata1. Revs. Sci. Eng. 16 (1977) 155.
12. Walker, P.L. Jr., She1ef, K. and Anderson, R.A., Chemistry and
Physics of Carbon 4 287, Marcel pekker NY (1968).
13. Trimm, D.L., Design of Industrial Catalysts, Elsevier, The
Netherlands, (1980).
14. Flynn, P.C. and Wanke, S.E., J. Catal. 34 (1974) 390, 400.
15. Tamman, G. and Mansuri, O.A., Z. Anorg. A1lg. Chern. 126
(1923) 119.
16. Huttig, G.F., Disc. Farad. Soc. 8 (1950) 215.
17. Pres1and, A.E.B., Price, G.L. and Trimm, D.L., Prog. in
Surface Sci. 3 (1972) 63.
18. Anderson, J.R., Elmes, P.S., Howe, R.F. and Mainwaring, D.E.,
J. Cat. 50 (1977) 508.
19. Catalyst Handbook, Wolf Scientific Texts (1970).
20. Udaja, P., Young, D.J. and Trimm, D.L., Studies in Surface
Science & Catalysis, 6 (1980) 331.
21. Thomas, J.M. and Thomas, W.J., Introduction to the principles
of Heterogeneous Catalysis, Academic Press, NY (1967).
22. Kummerow, I., Trimm, D.L. and Wainwright, M.S., Proc. 11th
Aust. Chern. Eng. Conf. 103 (1983).
23. Billmeyer, F.W. Jr., Textbook of Polymer Science, Wiley, NY
(1971).
PART VII. ACTIVATED CARBONS
 559

FUNDAMENTALS AND PRACTICAL IMPLICATIONS OF ACTIVATED CARBON

PRODUCTION BY PARTIAL GASIFICATION OF CARBONACEOUS MATERIALS.

T. Wigmans

NORIT Activated Carbon, P.O. Box 105, 3800 AC AMERSFOORT,

The Netherlands

1. INTRODUCTION.

Carbon in different forms plays an important role in many

industries. Most types of industrial carbons are produced from
naturally occurring carbonaceous materials like coal, oil, peat
and wood, by some form of thermal process. An important class of
materials involves pure carbons, either in a crystalline
(diamond, graphite) or a non-crystalline form (e.g. glassy
carbons, cellulose carbon, carbon fibres, carbon-carbon
composites), which distinguish themselves by a high purity and a
poorly developed porosity. Carbon blacks are a class of
compounds, prepared by a partial combustion of organic substances
of high carbon content in a restricted air supply and thus having
very low inorganic constituents. The carbon content of carbon
blacks ranges from 90 to over 99%, whereas they can be either
porous or non-porous, depending upon the method of production.
The third important class of compounds (chars) are obtained as a
residue after removal of the volatile components of a
carbonaceous material. Usually their structure is only slightly
porous (coke, wood charcoal), unless a suitable activation
procedure has been applied. In that case the name activated
carbons is used, denoting materials with a well developed and
accessible internal pore structure.

Although it is now recognised that the pore structure is the

most important property of activated carbon, it was formerly
believed that the carbon had to be activated by chemical and heat
treatment before it could remove colour, hence the name ACTIVATED
CARBON.
560

It is now known that the removal of impurities from gases and

liquids by activated carbon is by adsorption, and the activation
process simply increases the internal surface area of the carbon
and hence the number of sites available for adsorption.

Adsorption is a term used to describe the existence of a

higher concentration of a substance at the interface between a
fluid and a solid than is present in the fluid. It can be divided
into physical adsorption and chemisorption. In physical
adsorption the impurities are held on the surface of the carbon
by weak van der Waals forces whereas in chemisorption the forces
are relatively strong and occur at active sites on the surface.
The efficiency of the carbon will, therefore, depend upon its
accessible surface area, and also upon the presence of active
sites upon the surface at which chemisorption may occur.

Activated carbons can be prepared in the laboratory from a

large number of materials but those most commonly used in
commercial practice are peat, coal, lignite, wood and coconut
shell. The residues from carbonization and activation are found
to have a large pore volume, and as this is derived from very
small diameter pores the internal surface area is high. Activated
carbons have internal surface areas in the range 500-1500 square
metres per gram and it is this enormous-area which makes them
effective adsorbents.

The internal surface area of activated carbons can be

determined by the adsorption of e.g. nitrogen but there is no
guarantee that all this surface is available for the adsorption
of organic compounds. Since organic molecules are much larger
than a nitrogen molecule and as the pores are of molecular
dimensions it is possible for a sieving action to occur. Thus it
is essential for an adsorptive carbon to have a large accessible
surface area which in turn is directly related to pore size
distribution.

2. PRINCIPLE OF THE ACTIVATION PROCESS.

2.1 Chemical and physical activation.

During carbonization of a carbonaceous material, i.e.

pyrolysis in the absence of air and without addition of chemical
agents, most of the non-carbon elements (hydrogen, oxygen, traces
of sulphur and nitrogen) are first removed in a gaseous form by
pyrolytic decomposition.
 561

The freed atoms of elementary carbon are grouped into organized

crystallographic formations known as elementary graphitic
crystallites. The mutual arrangements of these crystallites is
irregular, so that free interstices remain between them, and
apparently as the result of deposition and decomposition of tarry
substances, these become filled or blocked by disorganized
carbon. The resulting carbonized product has generally only a
small adsorption capacity. Activation of the carbonized material
is possible via two routes:
- Carbonization after addition of substances which restrict the
formation of tar. In this way a carbonized product with the
properties of a good activated carbon can be obtained in a
single operation. Yet, the activation agent like ZnC12,
H3P04, KOH, K2S or KCNS that has to be applied in large
quantities, has to be removed via washing in order to relieve
the porous structure and to make the material practically
applicable. This route is generally called "chemical
activation".
- Partial gasification, generally called "physical activation".
The activation agents most often used are steam, carbon
dioxide and air or a combination of these. During activation
of the intermediate product first the disorganized carbon
(depending on the carbonization procedure 10-20% burn off) is
removed, and by this the surface of the elementary
crystallites becomes exposed to the action of the activation
agent. The further development of the porosity on increasing
burn off depends on the mechanism of carbon removal via
active site formation and the relative rate of reaction in
the direction parellel with the plane of the graphitic layers
compared to carbon removal in the direction perpendicular
to this plane. Phenomena related with this will be a major
subject of further discussion. Phenomena occurring during
chemical activation are beyond the scope of this paper.

2.2 Gasification.

The basic reaction of carbon with water vapour is

endothermic and of the following stoYchiometric form:

~H = + 117 kJ mol- 1 ( 1)

Similarly, the reaction of carbon with carbon dioxide can be

expressed as:

C + C02 -+ 2CO ~H + 159 kJ mol- 1 (2 )

562

As under practical conditions (above 800°C) the water gas shift

reaction is at equilibrium:

with an equilibrium constant between 0.5 and 1, both reaction (1)

and (2) are important despite the fact that in most cases super
heated steam is the primary activation agent.
For the C-C02 reaction the following general reaction mechanism
has been postulated: (1)

C + C02++1 C(O) + CO
2 t h (3 )
CO ++ C(CO)
3

as this explains the measured rate expression:

k PCO
2 (4)
r

in which r is a reaction rate, k is a rate constant, K1 and K2

adsorption constants and PCO and PC02 pressures of CO and C02'
Two limiting cases are:
A. the reverse reactions in step 1 and 2 can be neglected. This
leads to: K ; i11 K1 ; i3/j3 ; Kc01 K2 ; i1/i2 in which in and
jn are rate constants of resp. the forward and reverse
reaction of the nth equilibrium step and Keo the adsorption
constant of CO. In this mechanism, inhibition by CO is
established by a decrease of the fraction surface coverage of
oxygen atoms in steady state by CO chemisorption.
B. the reverse reaction in step 2 and the forward reaction in
step 3 can be neglected. This leads to K ; i11 K1 ; j1/j21 K2
; i1/j2' In this mechanism, the fractional surface coverage by
oxygen atoms is regulated by the reaction of the surface atoms
with CO to C02'
From experimental results i t is still not possible to
discriminate between mechanism A and B.

For the C-H20 mechanism a similar rate equation holds,

except that it is still a point of controversy whether CO shows
inhibition or not:

(5)
 563

The general reaction mechanism can be expressed by:

1 is
C(O) + H2 (6)

A simplification, by stating that j 1 « is and i"6 « is leads to

the two possible mechanisms A and B, analogous to the C-C02
reaction. The fact that hardly co inhibition is observed suggests
that in this approach mechanism A is most likely, although still
discussion is going on about the validity of this approach. E.g.
the radical mechanism, postulated by Long and Sykes (2)

2 C + 2 H2O -+ C(H) + C(OH)

.j-

C(i2 ) + C(f) (7)

H2 CO

also predicts a rate equation, similar to eq. (5) •

So both types of gasification reactions show strong product

inhibition. Further differences between the reaction with C02 and
with H20 involve:
- the larger dimension of the C02 molecule compared with H20
resulting in:
+ slower diffusion into the porous system of the carbon (3).
+ restricted accessibility towards the micropores (4).
- a significantly slower reaction rate for reaction 2 (1).

As the gasification reactions are endothermic, heat has to be

supplied in order to maintain an isothermal situation. In
practice this is achieved by in situ burning off the product
gases CO and H2 via air supply:

flH -285 kJ mol- 1 (8)

-238 kJ mol- 1 (9)

564

Besides an improved heat balance this facilitates the

gasification reactions by increasing the effective partial
pressure of the activation agent and by decreasing the partial
pressure of the inhibiting product gases. It was shown by e.g.
Katta et al (5) that under practical conditions for various coal
chars the inverse rate of the gasification reaction appears to be
proportional to the ratio H2/H20 or CO/C02'

An implication of the foregoing procedure is that,

independent of the type of activation agent used, effectively
carbon burning takes place:

~H = -406 kJ mol- 1 (10)

This means that H20 and C02 are not effectively used during the
activation process as a continuous back-formation occurs.

At temperatures above BOOoC, being practical activation

conditions, oxygen is known to react more than 100 times faster
with carbon than steam or carbondioxide do (1). This inevitably
leads to a completely diffusion controlled reaction involving a
shrinking core type burn off and therefore hardly any
contribution to the porosity development of the carbon. This
means that under practical activation conditions no direct
contact between 02 and C occurs, which implies that the
activation process is governed by the dynamic processes occurring
in the gas film surrounding the carbon, and that therefore in
essence activation is only possible under mass transfer limiting,
and thus strongly product inhibiting conditions. In the case of a
chemically controlled competition between 02 and H20 or C02 only
carbon burning would occur. If reaction with C02 and H20 would
also be film diffusion controlled the same type of shrinking core
mechanism during burn off might be expected (3). As this is
obviously not the case the dynamic processes in the film, that
include CO and H2 diffusion into the direction of the gas bulk
and C02 and H20 back-formation, guarantee that near the external
carbon surface under practical conditions not-limiting amounts of
H20 and/or C02 are present, explaining the foregoing paradox.

An extra implication of mass transfer limiting conditions is

the fact that heat consumption by the endothermic gasification
reactions and heat production via CO and H2 burning occur in
different parts of the reactor. The analogy between mass and heat
transfer predicts that the temperature of the carbon surface at
which gasification takes place will be significantly lower than
the temperature of the flowing gas phase. Differences as high as
50°C can be found both during laboratory experiments and in
practice.
 565

An illustration via laboratory activation results of experiments

carried out in a packed bed balance reactor (6) applying C02
(99.995%, N 45 AGA) as the activation agent (0,7-2 mmol s-1) and
using extruded peat semi-coke, pre-pyrolysed in nitrogen at 1215
K during 30 minutes (diameter 1.5 mm, ash content 3.1%), as the
carbonaceous precursor, is given in table 1 (C02 conversions in
all cases below 5%, being a condition for differential reactor
behaviour). It shows that in the case of a rapid removal of
inhibiting products via high gas flows the activation energy of
the activation process remains on the level of the chemically
controlled reaction (approx. 300 kJ mol- 1 (7». The slight
decrease upon increasing burn off may be attributed to some
concentration of catalytic impurities (mainly Fe, Ca and Mg).
Lowering the gas flow, however, leads to a drastic decrease of Ea
upon increasing burn off and a decrease of the level of
reactivity compared with the high gas flow experiments. Yet the
high initial value of Ea at 20% burn off in combination with the
highest value for rn excludes a significant contribution of film
diffusion limitation (8). So the decrease of Ea upon increasing
burn off must be attributed to an enhanced accumulation of
inhibiting CO into the porous structure of the carbon (9).
Furthermore these results indicate that no proportionality can be
found between reactivity and BET surface area. The explanation
for this can be two-fold:
- Upon increasing burn off the ratio between the active surface
area (ASA) and the total surface area diminishes, indicating an
effective destruction of active sites, viz. edge atoms or
surface groups (10).
The limiting radius of pores accessible for C02 increases upon
decreasing Ea (11). This means that active sites in micropores
that are accessible for C02 at low burn off levels become
ineffective upon increasing burn off due to enhanced product
inhibition. As shown by Otto et al (12) not the equilibrium
oxygen uptake as a measure for ASA but the dynamic adsorption
during a fixed time of e.g. C02 at 273 K might be
representative for the accessible surface area of microporous
solids.

It is easily imaginable that not only the rate of activation

is determined by product inhibition but that also the type of
porosity created by gasification will depend on the H2/H20 or col
C02 ratio within the pores. After discussing the microstructure
(section 4) and characterization (section 5) of activated carbon
this will be a major topic in section 6.
 566

Table 1. Gasification results of peat semi-coke in C02; influence

of gas flow rate.

~ Burn off 20 40 60 80
~pprox. BET surface area
(m2 g- 1 ) 600 900 1200 1500

rn (% h- 1 at 1175 K) - rw
(= rn per unit weight
of carbon present)
- chemically controlled
reaction
( C02 = 2 mmol s-1) 2.4-3.0 2.1-3.5 1.6-4.0 1.0-5.0
internal mass transfer
limitation
( C02 = 1 mmol s-1) 1.8-2.25 1.5-2.5 1.3-3.25 0.8-4.0

~a (kJ mol- 1 )
r chemically controlled
reaction 320 311 308 268
r internal mass transfer
limitation 335 271 222 179

~emperature drop towards

setpoint (OC) 40 33 25

3. PHYSICAL FORMS OF AC.

Activated carbons are produced in both powdered and'granular

forms. Powdered carbons are usually ground to give a distribution
of particle sizes in the range of 5-150 ~m, but coarser and finer
grades are also available for specific applications. Granular
carbons are available as irregularly shaped particles obtained by
crushing and sieving or as regularly shaped particles as a result
of a specific moulding procedure prior to activation. The most
common appearance is the cylindrical form (diameter 0.8-4 mm)
produced by an extrusion process but in some cases spheres or
tri-quadrilobes are required for specific applications. Granular
coal based carbons are generally in an irregularly shaped form as
a result.of the required milling-compacting (briquetting)-
breaking-preoxidation sequential procedure necessary for making
the feedstock accessible for the activation agent (3).
 567

Wood, peat and lignite (or the equivalent chars) may be activated
as such, which results in irregularly shaped granulates that have
not been compacted prior to activation. Extrusion is the most
common compacting procedure for these feedstocks.

The manner in which powdered and granular carbons are used

is entirely different. Powdered carbon is mainly used for liquid
phase adsorption where it is intimately mixed with the liquid to
be purified. When the impurities have been adsorbed, the carbon
is removed from the solution by filtration or sedimentation, and
then discarded. The particle size distribution is important as it
can affect the rate of adsorption, filtrability and sedi-
mentation. Granular carbon, on the other hand is used in
adsorbers through which the liquid or gas stream to be purified
is passed. Generally granular carbons can be regenerated and
reused. As in practice the adsorber volume is fixed, the
applicability of a granular carbon is mainly determined by its
adsorption capacity per unit of volume, which often means that
carbon in the compacted form has to be used.

In gas phase adsorption, granular carbon is almost always

used because the pressure drop through the carbon bed is of prime
importance and only granular carbons will allow sufficient flow
with an acceptable pressure drop.

In liquid phase applications, on the other hand, the choice

between powdered and granular is complicated by the fact that
often either method could be used. For example, in the
purification of water supplies, some water works are based on
powdered carbon and others on granular carbon. This is also the
case in the sugar and glucose industries where both powdered
carbon and granular carbon-based refineries are in existence.
Generally granular carbon is used when a single product or grade
is being continuously produced in large quantities. Normally
granular carbon can be regenerated which involves washing or
burning off the impurities and restoring the pore structure.
Washing or steaming granular carbon can usually be done in situ
but if thermal regeneration is necessary, only the largest of
installations will be able to justify the investment in a
regeneration kiln. Other users will normally rely on the supplier
to carry out the regeneration for them.

If regeneration is possible it is obvious that the quantity

of carbon required per unit weight of product is less than the
amount of powdered carbon which would be used on a throwaway
basis. It is therefore necessary to examine the savings in carbon
costs against regeneration costs and the fact that granular
carbon applications involve the use of purpose built plants.
568

Powdered carbons are more versatile, in that the carbon

dosage can be easily altered making product quality control very
easy, especially when input quality is variable. Powdered carbon
can be used for most applications and can be used with standard
chemical plant equipment such as tanks, stirrers and filters. It
is, therefore, possible to use the same unit to make different
products, a flexibility not found with granular installations.
Therefore, for most liquid phase carbon treatments an examination
of the advantages and disadvantages of the two types of carbon
and system has to be made.

4. PHYSICAL AND CHEMICAL MICROSTRUCTURE OF ACTIVATED CARBONS.

4.1 Elementary microcrystalline structure.

Activated carbons, together with other types of chars and

carbon blacks, form a group of carbonaceous materials of which
the structure and the properties depending on it are more or less
similar to graphite, although the structure is less perfectly
ordered. Graphite is composed of layer planes formed by carbon
atoms ordered in hexagons. The interatomic distance between
carbon atoms in the individual layer planes is 1.42 ~. The layer
~lanes are in parallel array with an interlayer spacing of 3.35
~. On the basis of X-ray analyses Riley (13) proposed two types
of structure for activated carbons (and other chars and carbon
black). The first type of structure consists of elementary
crystallites~ these are in two dimensions analogous to graphite.
The structure differs from graphite, however, in that the
parallel planes are not perfectly oriented with respect to their
common perpendicular axis, the angular displacement of one layer
with respect to another is random and the layers overlap one
another irregularly (figure 1). For this arrangement Biscoe and
Warren (14) proposed the term "turbostratic structure". The
mutual orientation of the elementary crystallites is completely
~t random~ their dimensions depend primarily on the temperature
of carbonization. The following values are most frequently given:
height, Lo, from 9 to 12 R~ width, Ls (or diameter~ if a circular
cross section is assumed), from about 20 to 40 A. From this it
follows that the elementary crystallites are composed of about
three parallel plane graphite layers, the diameter of which is
about 9 times the width of one carbon hexagon.
 569

b a

Figure 1. Ordering of graphitic layers (adapted from (24»

a. regular (graphite)
b. irregular (activated carbon)

A chemical proof for the graphitic structure of activated

carbon was obtained via the production of intercalation compounds
of e.g. alkali metals (15) and fluorine (16) with activated
carbon. The interlayer spacing in activated carbons, however, is
considerably higher than in the case of graphite (3.44-3.65 ~)
(17). The second type of structure, Riley describes as a
disordered, cross-linked space lattice of the carbon hexagons,
which results from their deflection from the planes of graphitic
layers. This structure is probably stabilized by hetero-atoms, in
the first place by oxygen, as it has been found in chars prepared
from materials of a high oxygen content. According to Riley a
number of chars display both types of structure. On the basis of
numerous X-ray investigations of different carbonaceous materials
(coal, coke, activated carbon prepared from sucrose, PVDC, PVC
and others), Franklin (18) divides all carbonaceous materials,
with the exception of diamond, into two distinct classes:
graphitizing (soft), and non-graphitizing (hard). Carbonaceous
materials graphitize under the action of temperatures higher than
about 1000°C. The process consists first in the consumption of
the non-organized carbon, which is used up in the temperature
range from 1000 to 2000 0 C for increasing the width (diameter) of
the graphitic layers in the elementary crystallites.
570

PVDC carbonized at 1000°C contains about 35% of non-organized

carbon, and the diameter of its graphitic layers is approximately
16 E.
After it has been heated to 2000°C, the non-organized
carbon becomes less than 10% and the diameter of the layers
increases to 22 R. In this range of temperatures the height of
the elementary crystallites, i.e., the number of parallel
graphitic layers, changes very little. When all the non-organized
carbon has been spent in providing for the growth of the diameter
of the graphitic layers, upon a further increase in temperature
both the diameter of the layers and also the number of parallel
lines, i.e., the hight of the crystallite, increase. Upper
limits for carbonized PVDC seem to be 12 parallel layers with a
diameter of ca. 70 g (3000°C). In contrast with this, PVC
exhibits at much lower temperatures a continuing growth of the
number of parallel layers, indicating the formation of a truly
three dimensional graphite structure. This is explained by a
relatively weak cross linking between the elementary crystallites
and therefore a considerable mobility. Therefore, the charcoal
prepared from PVC is relatively soft, whereas the crystallites
sbow a high degree of parallel mutual orientation leading to a
low porosity (figure 2b).

a b

Figure 2. Schematic representation of (a) non-graphitizing and

(b) graphitizing structure of activated carbon (adapted
from (18».

Crosslinking leads to a rigid immobile mass with a well developed

microporous structure which is preserved even at very high
temperatures (figure 2a). It has been demonstrated that oxygen
rich precursors inhibit graphitization whereas high C/H ratios
enhance the formation of a three dimensional graphitic structure
( 19) •
 571

From investigations of carbonaceous adsorbents prepared from

various raw-materials it has been found (20,21) that differences
between graphitizing and non-graphitizing carbonaceous materials
are not limited to differences in their crystallographic
structure as determined by X-ray analysis, but are also
demonstrated by other properties. For the group of materials
examined, it has been found that the helium density of non-
graphitizing materials is less than 2 g.cm- 3 , whereas for
graphitizing substances it exceeds 2 g.cm- 3 and approaches that
of graphite, which is the limiting value (usually given as 2.26
g.cm- 3 ). Also the mercury density of non-graphitizing sUbstances
is lower (1.2-1.5 g.cm- 3 ) than that of graphitizing substances
(1.8-2.0 g.cm- 3 ). The dependence of both densities on the
temperature of heat-treatment of the carbonized intermediate
products is also different. Furthermore, the porous structure of
the two classes of materials is probably not the same:
significant differences are found in their magnetic
susceptibility, in the concentration of unpaired electrons and in
optical properties (non-graphitizing materials are optically
isotropic, graphitizing materials are anisotropic). Finally it is
worthwhile mentioning the fact that the addition of certain Lewis
acids (like Alel3) (22) or alkali compounds (23) enhances
graphitization of non-graphitizing substances. In conclusion,
although the temperatures applied during the production of
activated carbon are much lower than during graphite production,
the extent of graphitization highly influences the character of
the activated carbon produced. Therefore the type of precursor,
specific additives and the temperature level are of predominant
importance. An illustration of the latter is given by the
specific electrical resistance of birch wood charcoal, carbonized
at various temperatures (24): at 500 De a value of over 10 6 Ohm.cm
is found, whereas going from 600 to 900 D e a drop occurs from 44
to 0.11 Ohm.cm.

4.2 Porosity.

During the process of activation the spaces between the

elementary crystallites become cleared of various carbonaceous
compounds and non-organized carbon, and carbon is also removed
partially from the graphitic layers of the elementary
crystallites. The reSUlting voids are termed pores. The area of
the outer surface - the geometrical surface - is practically
negligible in comparison with the internal pore wall surface
area. Results seem to indicate that there are pores with a
contracted entrance (ink-bottle-shaped), pores in the shape of
capillaries open at both ends or with one end closed, pores in
the shape of more or less regular slits between two planes, v-
shaped, tapered pores, and other forms.
572
In most cases however it is difficult reliably to determine for a
given carbonaceous material what pore shapes are present and in
what proportion. For calculations it is usual to disregard the
actual pore shape and to look upon the pores as being either
bottle-shaped, or straight, non-intersecting cylindrical
capillaries. The calculation of radii of pores (assumed being in
the shape of cylindrical capillaries) or of radii of their necks
(if the ink-bottle model is assumed) obviously yields values
which approach more nearly the actual dimensions of the pores,
the more realistic the assumed shape. Activated carbon usually
has pores belonging to several groups, each group having a
certain range of values for the effective radius.

Micropores, according to Dubinin's classification (25) are

pores of such a small size that they are already completely
filled at the relative pressure corresponding to the beginning of
the hysteresis loop, which means that in these pores capillary
condensation does not take place. According to this criterion,
the effective radius of micropores is less than about 10 g and
the order of their size corresponds to that of molecules. For
activated carbons the micropore volume is generally from 0.15 to
0.50 ml/g and, usually, the specific surface area in micropores
amounts to at least 95% of the total specific surface (i.e. the
surface area per gram of adsorbent). A more precise
classification distinguishes the group of submicropores with an
effective radius of less than 2 R. Adsorption in these pores is
considered to be irreversible under practical conditions.

Transitional pores (also called mesopores), according to

Dubinin's classification, are those in which capillary
condensation with the formation of a meniscus of the liquefied
adsorbate can take place1 this phenomenon usually produces the
hysteresis loop on the adsorption isotherm. The size of
transitional pores is therefore in the range of effective radii
from approximately 10 to 250 i. The volume of transitional pores
is usually from 0.02 to 0.10 ml/g and their specific surface area
does not exceed 5% of the total surface of the activated carbon.
By employing special procedures, however, it is possible to
prepare activated carbons in which transitional porosity is
exceptionally developed, the volume of transitional pores
attaining 0.30 to 0.85 ml/g, and their surface area amounting up
to 200 m2 /g or even more.

Pores of an effective radius larger than about 250 ~, which

at this size cannot be filled by capillary condensation, are
classified by Dubinin as macropores. In activated carbon the
effective radius of macropores is frequently in the range from
5000 to 20000 R, their volume is between 0.2 and 0.5 ml/g, and
their surface area is from 0.5 to 2 m2 /g.
 573
Each of these three groups of pores has its specific
function in the process of adsorption on activated carbon. Of
greatest significance for adsorption are micropores which,
because of their very large specific surface area, attaining
values of several 100 m2 /g and of their large specific volume,
decide to a considerable degree the adsorptive capacity of a
particular activated carbon (provided, of course, that the
molecules are not too large to enter the micropores). The
function of transitional pores is twofold. First, in adsorption
under sufficiently high pressure vapours are entrapped in them by
the mechanism of capillary condensation. The second function of
transitional pores is that they can serve as a passage fot the
adsorbate to the micropores, for, according to Dubinin (25), only
few of the micropores lead directly to the outer surface of the
particle of activated carbon, its porous structure being
predominantly arranged in the following pattern: the macropores
open out directly to the external surface of the particle,
transitional pores branch off from the macropores, and
micropores, in turn, branch off from the transitional pores.
Thus, the significance of macropores is mainly that they enable
the molecules of the adsorbate to pass rapidly to smaller pores
situated deeper within t~e particle of activated carbon. When
activated carbon is used as a catalyst support, the pores of
larger sizes can be important as places of deposition of the
catalyst. As will be shown later, macropores are hardly formed
during activation. This means that they are already present in
the carbonaceous precursor. Examples are:
- Cell channels in materials of biological origin that have been
undergoing no or hardly any coalification (renewable materials
like wood, fruit shells, fruit stones etc. or peat).
- Interparticle space in extruded chars or powdered materials
(26).
- Bubble expansion during coal pyrolysis (27).
It is easily imaginable that the transport phenomena that govern
the accessibility of the micro/mesoporous structure during
application, also play an important role during activation. For a
good development of the essential porous structure the transport
of the activation agent into the interior of the precursor has to
be guaranteed.

This means that a well developed macroporous structure of the

carbonaceous material is essential. For this reason materials
like pitch coke, anthracite and caking coals are difficult to
activate without precautions that facilitate the accessibility of
the gaseous activation molecules.
574
To a certain extent, manipulation of the ratio of micro- and
mesopores formed during activation is possible (see section 6).
No success has yet been recorded, however, in the technology of
preparation of activated carbon with a porous structure
containing transitional pores and macropores, and only a very
limited proportion of micropores; one of the reasons for this is
that the needed extreme degree of activation would largely
destroy the mechanical strength of the adsorbent.

As has already been mentioned, the specific surface area and

volume of the different classes of pores depend on the method of
preparation of the activated carbon and can vary over a very wide
range of values. The suitability of activated carbon for a
particular application, however, depends on the ratio in which
pores of different sizes are present. For the adsorption of
vapours and gases from mixtures in which they are present in
small concentrations, markedly microporous activated carbons are
the most suitable. But activated carbons used for the recovery of
vapours of industrial organic solvents from waste gases should
contain a certain fraction of transitional pores. Frequently,
even a small variation in the porosity produces a marked change
in the functional properties of an activated carbon. Because of
the close relationship between the adsorptive and other
properties of activated carbon and its porous structure,
information about these parameters is of fundamental importance,
and much attention has therefore been given to the development of
methods for their investigation; a discussion of some of these
methods is given in section 5.

4.3 Chemical structure of activated carbons.

The adsorptive properties of activated carbon are determined

not only by its porous structure, but also by its chemical
composition. The decisive component of the adsorption forces on a
highly-ordered carbonaceous surface is the dispersion component
of the van der Waals forces. Disturbances in the elementary
microcrystalline structure as, for example, by the presence of
imperfect (partially burnt off) graphitic layers in the
crystallites, obviously changes the arrangement of the electron
clouds in the carbon skeleton (28). As a result, incompletely
saturated (residual) valencies or unpaired electrons appear, and
this influences the adsorptive properties of the activated
carbon, especially for polar or polarizable substances. Another
type of disturbance is the presence of hetero atoms in the carbon
structure. A distinction can be made between chemically bonded
elements, present as surface groups at edges and corners of the
elementary graphitic crystallites and therefore forming an
organic part of the chemical structure of activated carbon, and
inorganic ash components.
 575
The ash content and its composition vary widely with the
type of feedstock used. Generally the major components are
silicates and aluminates with lesser amounts of calcium,
magnesium, iron, potassium and sodium. Trace amounts of zinc,
lead, tin, copper, boron, arsene or vanadium generally give
additional information on the type of feedstock used. In
adsorption of electrolytes and non-electrolytes from solutions
and of gases, the adsorption characteristics of activated carbon
are significantly influenced even by small amounts of ash,
possibly due to a changed polarity of the elementary structure
(29). Furthermore specific ash components like iron, calcium and
alkali compounds exhibit a catalytic effect in steam activation,
that manifests itself by both an enhanced affinity of the
carbonaceous feedstock towards water and by a selective pore
formation. Alkali compounds like hydroxides or carbonates of
potassium or sodium are known to promote the formation of slit-
shaped micropores via intercalation (30), whereas transition
metals and earth alkali compounds enhance mesopore formation via
channeling of metallic particles (31).

Of greater effect on the adsorptive and other properties of

activated carbons is the presence of oxygen and hydrogen, being
the major chemically bonded elements in surface groups. With
respect to this the activated carbon surface may be considered as
being a hydrocarbon surface modified by oxygen compounds rather
than being graphitic (32). Values of up to 30 mol% Hand 15 mol%
o are not unusual. The origin of oxygen and hydrogen can be
threefold: 1. Originally present in the feedstock.
2. Selective removal or introduction during activation
3. Selective introduction via post-treatment.

Ad 1.
Activated carbon prepared from sucrose may contain up to 30 mol%
Hand 9 mol% 0 (33), whereas activated carbon free from oxygen
and with only traces of hydrogen can be prepared by the pyrolyses
of PVDC, Saran of phenolaldehydic resins. In contact with air,
however, these carbons bind oxygen via chemisorption. In general
the oxygen and hydrogen content of the starting material has also
a considerable influence on the arrangement and size of the
elementary crystallites formed (19), as discussed in 4.1.

Ad 2.
Apart from the composition of the feedstock, the temperature
level and the residence time during activation have a consider-
able influence on the elementary composition and therefore the
microstructure of the carbon. Practical activation conditions
(temperature above aooDc, residence time several hours, fossile
carbonaceous materials as feedstock) exclude oxygen contents over
1 mol% in the activated product prior to exposition to air.
Hydrogen contents vary, however, significantly.
576

Table 2 gives some results of extruded peat semi coke samples,

described earlier in section 2.2, that have been carbonized and
activated in a batch-wise operated internally recirculated kiln
(34) according to procedures that will be described in more
detail in section 6.

Table 2. Elementary composition, true density, microporosity and

crystallite dimension of peat semi-coke after
carbonization and partial steam activation.

Irempera- C/Hc wt% NC rue densityd Fmicro e df

I,-ure( °C) (mol/mol) (g/cm 3 ) (nro)

Ca
-
Ab
-C -A -C -A -A -C -A
860 9.6 13.7 1. 07 0.48 1. 79 1. 81 0.58 4.2 6.5
900 13.0 15.9 0.95 0.37 1.82 1.84 0.62 5.9 7.6
1000 19.0 28.9 1. 0 1 0.30 1.86 1.90 0.78 9.1 14.1
1040 22.6 37.5 1.06 0.29 1.86 1.91 0.82 10.9 18.4

a: Carbonization during 30 minutes in nitrogen flow (details in

section 6).
b: Activation during 10-12 hours in steam-nitrogen flow (PH 0
= 57-82 kPa), % burn off: 40-50 (details in section 6). 2
c: Determined with a Carlo Erba CHN analyser.
d: Determined as the density in water.
e: Ratio micropore volume/micro + mesopore volume.
f: Diameter of the primary graphitic layers, assuming square
planes without introversions.

I t is obvious that a higher temperature, both during

carbonization and activation leads to higher C/H ratios in the
product. Remarkable is the fact that the nitrogen content
decreases with increasing reaction temperature during activation
whereas during carbonization the temperature has hardly any
influence. The enhanced graphitization at higher reaction
temperatures is illustrated by both the higher true densities and
an increasing diameter of the primary graphitic layers. The
latter is determined assuming these layers being purely graphitic
with the edge C atoms being associated with hydrogen.
 577

The hypothetical square is then characterized by its surface area

to periphery ratio, in which the number of C atoms in the plane
will be proportional to its surface area, whereas the periphery
will be proportional to the number of H atoms in the plane.
Introversions in the primary plane will cause a lowering of the
surface area to periphery ratio. Therefore, introverted planes
with an equal C/H ratio compared with closed planes will be
larger. Table 3 indicates that the diameter of the introversion
strongly influences the diameter of the plane with a given C/H
ratio.

Table 3. Influence of the introversion diameter on the diameter

of the primary plane at given C/H ratio (= 25).

IFraction voids diameter of primary plaqe (nm)

(in primary plane)
1 nm* 3.5 nm* 10 nm*

0 13 13 13
0.10 150 18 16
0.20 > 1000 30 19
0.25 45 22
0.30 91 26
0.33 228 29

* diameter of introversion.

This means that introversions with the dimension of micropores in

the primary plane are very unlikely which means that micro pores
will be situated mainly in-between grafitic layers, making them
in essence slit- or V-shaped. One should keep in mind that the
diameter of the primary crystallites is 3-4 nm (fig. 2). This
means that the two-dimensional structure from figure 2 consists
of highly rumbled primary planes, that exhibit therefore a
considerable three-dimensional extension, and that the
crosslinking of the primary crystallites is partly due to this
extension. The merits of this for the pore development model will
be discussed later. It is stressed that the analogy between high
activation temperatures and microporous carbons, that follows
from table 2 will be essential in this discussion, apart from the
fact that activation is inherent in true density increase and
graphitic plane growth.
578

Ad 3.
In accordance with the theory of heterogeneity of solid surfaces,
first formulated by Taylor (35), oxygen, hydrogen and other
hetero atoms in carbonaceous materials are bonded to those carbon
atoms which form the edges and corners of the elementary
graphitic crystallites, because the valencies of these atoms are
not fully saturated by interaction with the surrounding carbon
atoms, and their reactivity is therefore higher. Similarly,
carbon atoms in defect positions of the crystal lattice, for
example in distorted or incomplete carbon hexagons, are more
reactive and tend therefore to reduce their potential energy by
forming bonds with oxygen, hydrogen and other elements. These
energetically richer carbon atoms cause also other effects: they
saturate their residual valencies by bonding to one another
neighbouring elementary crystallites; they bind products of
thermal decomposition during the process of carbonization, they
form centres where impurities concentrate by diffusion when the
carbonaceous material is subjected to a high temperature, they
probably provide sites where molecules of polar substances are
adsorbed, and they reduce the mobility of physically adsorbed
species.
In the foregoing section it was assumed that after activation the
major part of these sites is occupied by hydrogen. Yet many
specific features of activated carbon are ascribed to surface
complexes of oxygen with carbon, mainly formed after activation
by expos ion to air or by specific post-treatment. Surface
coverages of up to 20% have been reported in the latter case
(36). Low temperature air oxidation ("aging") yields surface
groups that can neutralize acids (H carbons). These groups
decompose generally at high temperatures (above 600°C) under
formation of mainly CO and are probably of the chinon-type
bonding (37). The surface coverage is limited to approx. 2%. High
temperature post-oxidation by air (300-500°C) or by C12, 03'
KMn04 or HN03 generally yields much larger quantities of surface
groups that react acidic (L carbons). Lactone-, carbonyl-,
carboxyl-, phenol- and quinone groups have been identified.
Generally, but not as a rule, these groups decompose at lower
temperatures as the basic groups do, primarily via C02 formation.
Several correlations between the presence of basic or acidic
surface groups and the adsorptive capacity for e.g. methanol,
benzene, non-polar organics, water, phenol etc. have been found.
However, this lies beyond the scope of this paper.

Besides oxygen containing surface groups, the presence of

sulphur-, nitrogen-, chlorine- and other elements containing
groups has been confirmed. For bulk phenomena that determine the
behaviour during activation, however, these groups are of no
significance.
 579

5. CHARACTERIZATION OF ACTIVATED CARBON.

Methods of quality control and of testing the suitability of

activated carbon for a given purpose are of great importance for
both technical and commercial practice.

For production control rapid orientational tests on final

and intermediary products should serve for controlling plant
operation. Among these are, for example, the determination of the
apparent density of granular carbon and the decolourizing
efficiency of powdered carbon as a rough measure for the degree
of activation. On the results the conditions of steam activation
may be controlled by adjusting the steam and/or air supply, by
regulating the temperature profile of the kiln or by adjusting
the feed rate of carbonaceous precursor and/or the rate of
withdrawal of carbon from the kiln, in this way regulating the
residence time and/or kiln content. Furthermore special
requirements are placed on acceptance tests, the results of which
serve for arbitration purposes.
It will be clear that insight into correlations between these
orientational or acceptance tests and more fundamental aspects of
activated carbons that determine their applicability, is
absolutely necessary for a successful production control.

5.1 Physical and mechanical properties.

In literature, methods for the characterization have been

described extensively (24). Therefore only a survey of the most
important properties will be given here in table 4. This table
also mentions some aspects of relevance for application and
production purposes and for the pore structure.

5.2 Chemical properties.

Among the most important chemical properties are:

- water content
- ash content
- extractable cations, both in water and in ·acid
- extractable anions like sulphate, phosphate, halide
- pH and neutralisation
- sulphur content
- self-ignition temperature
- volatile content (decomposition of surface groups)
- character and amount of surface groups

It lies beyond the scope of this paper to discuss the merits

of these for the various fields of application further. More
detailed information is abundant (24).
Table 4. Physical and mechanical properties. VI
0:>
o
Physical and mechanical properties. Relevance for

application production pore structure

- Particle size distribution (P,G)* filtrability (P) accessible possible

accessibility external heterogeneity
surface area,
- Grain length and diameter distri- flow resistance, viz. rate of
bution (G) apparent density, activation
accessibility

- Behaviour in back flow (G) classification in

filters (liquid
applications)

- Attrition and abrasion (G) pressure drop in dust losses, decreasing overall
behaviour filters due to measure for degree of activation
crushing, dust degree of due to attrition of
formation mesoporosity highly active outer
layer

- Apparent and true density (P,G) adsorptive plant control overall degree of
capacity per activation
unit of volume

Void fraction in meso/macropores accessibility, accessibility meso/macropore size

via mercury porosimetry (P,G) apparent density distribution

* P powdered activated carbon

G granular activated carbon
 581
5.3 Adsorptive properties.

In describing the adsorptive properties one should

distinguish two different goals:
- Determination of the usability for a specific application.
- Determination of the pore size distribution and the effective
surface area.

5.3.1 Application tests. General application tests involve,

among others:
- Molasses decolourization, specifically directed towards the
use in the sugar and glucose industry.
- Methylene blue adsorption, generally used as a test for
medicinal carbons and a measure for adsorption of larger
molecules.
Phenol value, as a measure for applicability in preparing
drinking water.
- Iodine adsorption, as a measure for the BET surface area.
- Chlorine decomposition in water, as a measure for specific
application in drinking water preparation.
- Carbontetrachloride and-butane adsorption from the gas phase,
as a measure for applicability in solvent recovery,
deodourization of air and other gas phase applications of
activated carbon.
It should be noticed that in general the buyer wishes to be
certain of the suitability for his purpose and therefore uses
tests in which the conditions simulate as closely as possible
his operational conditions. Therefore, whenever possible it is
preferable to test directly in industrial plants or at least in
a comparable pilot scale unit. Evaluation by standard tests is
done only when unavoidable or in the case of a proven
convertability of results from standard tests to practical
conditions. In the latter case a thorough knowledge of the
character of the process stream that has to be treated and the
pore structure of the carbon is vital.

5.3.2 Pore size distribution and effective surface area. In

literature, numerous methods for calculating pore size
distributions and effective surface areas are described
(24-26). Among them are the Kelvin equation, the t-method, the
potential theory of adsorption, more specifically Dubinin's
approach, and the BET equation. These are all based on data
from equilibrium adsorption tests at various partial pressures
under isothermal conditions and assume (except for the BET
equation) capillary condensation and volume filling of the
pores. In this study (section 6.) only a rough discrimination
will be made between micro-, meso- and macropore volume, based
on benzene adsorption data (according to the Kelvin equation)
and total pore volumes calculated from apparent and true
densities.
582

In some cases, specific pore size distributions will be given

that are based on nitrogen adsorption/desorption data (-196°C).
A further quantification and specification is considered to be
premature, as only trends will be evaluated.

One should keep in mind that during application in

practice equilibrium is established seldom ., mainly due to
diffusion limitations. This strongly hinders the interpretation
of pore size distribution data from equilibrium measurements in
terms of applicability. Application tests, as a consequence of
this, are therefore mostly adapted in such a way that they give
combined information on maximum adsorption capacity and rate of
adsorption. Generally the maximum adsorption capacity may be
derived from equilibrium adsorption data, but the kinetics of
adsorption are much more difficult to quantify, especially for
adsorption from the liquid phase. Both diffusion to the
external surface of the carbon particles and diffusion to the
internal surface area can play an important role. The first
type depends on the rheological conditions (external diffusion,
convective transport via either laminar or turbulent flow), the
second type on the extent of internal diffusion (either normal
diffusion or molecular Knudsen diffusion, depending on the pore
diameter), surface migration and viscous flow in the adsorbed
phase. It will be clear that especially for this internal
transport the accessibility of the porous system and the
molecular dimensions of the adsorbate play a predominant role
(see also section 7.2.2).

6. INFLUENCE OF PROCESS VARIABLES DURING PRODUCTION ON PORE

STRUCTURE DEVELOPMENT.

6.1 General literature survey.

An evaluation of the state-of-the-art knowledge of

phenomena governing the pore structure development during
activation is troubled by two major drawbacks:
- The information from the open literature is mainly based on
laboratory scale work. Due to the confidential character of
commercial production data the possible effects of upscaling
are poorly described.
Experimental conditions, viz. types of feedstock used and
type of equipment, vary widely. Moreover, the interpretation
of results is often directed towards a desired application of
activated carbon, which means that implicit demands are made
towards the optimum pore structure, that may differ greatly
for various applications.
 583

Still some general views may be distilled from the available

information:
- The pore volume created during activation increases invariably
with increasing burn off. Yet an optimum in surface area or
micropore volume is observed (34,38). This means that pore
enlargement and therefore a shift from micro- to meso- or even
macroporosity occurs (39).
- Besides pore enlargement, exclusion of constricted pores is
often considered to contribute to pore volume increase during
activation, especially in the case of coal as precursor
(40,41).
- Preoxidation via low temperature (upto 400°C) treatment with
air (42) or via addition of H3B03, H3P04 (43) or KOH (44), is
mostly performed with caking coals as the carbonaceous
feedstock. The reason for this may be twofold:
+ increased reactivity, ensuring acceptable production rates in
practice. It is believed that a selective reactivity increase
of the edge atoms of the microcrystallites via formation of
oxygen containing functional groups is achieved, that
enhances pore width enlargement during activation (41).
+ approved accessibility towards the gaseous reactant,
visualized by an increased surface area of the precursor
(45). Without precautions in activating caking coals (40) or
anthracites (46) the microporous structure of the precursor
remains inaccessible, which results in an increase of the
meso/macropore volume only during activation, without a
notable development of the micropore structure.
So a certain degree of oxygen functionality in the starting
material is essential for an optimum pore structure development
during activation (47). This means that high coking
temperatures are unwanted in the case of coke being the
carbonaceous precursor. In general oxygen rich feedstocks show
a more narrow pore size distribution after activation than
oxygen poor feedstocks (46,48).
- Closely related with the foregoing is the effect of the type of
activation agent. Due to differences in reactivity and
molecular dimensions c.q. diffusion rate, the diameter of
effectively accessible pores decreases in the range 02-C02-H20.
This means that the best developed micropore structure may be
expected in the case of steam activation. A less developed
micropore structure seems to be a result of higher partial
pressures of steam (49).
584

- The effect of the temperature seems to be somewhat confusing

because of the predominant effect of the measuring equipment on
especially external mass transfer limitations. Under
differential conditions 800-850 o C seems to be the optimum
achievable temperature level without notable shrinking core
behaviour (38,50). No systematic studies concerning the effect
of the activation temperature under product inhibiting
conditions have been reported. Although the effective
accessible pore diameter increases on increasing reactivity
(and thus increasing temperature), leading to a less developed
micropore structure, product inhibition (CO, H2) might overrule
this.
- The strength of compacted carbons is negatively influenced by
mesoporosity (4) (see also figure 2) and by a selective
gasification of binder material due to high temperatures or
reactivity enhancing additives like alkali or earth alkaline
compounds (51).
- As a result of low temperature carbonisation experiments of
both cellulose (52) and coal (53) it follows that charry
materials like activated carbon may be considered as highly
condensed hexagonal structures (figure 3). As stated before,
partial graphitization may influence the porosity development.
Table 2 illustrated that microporosity is enhanced by high
temperatures and is accompanied by high C/H ratios (and
therefore a higher degree of graphitization) and higher true
particle densities.

H H

H
H

H H

H H

Figure 3. Average char "molecule" with a C/H ratio of 4.4

(adapted from (53».
 585

6.2 Experimental.

Results will be reported on experiments carried out in a

batch-wise operated bench scale kiln, fed with approx. 1.5 1 of
carbonaceous precursor. Steam-nitrogen mixtures (PH20: 40-82 kPa)
at atmospheric pressure were applied as the gaseous reactant in
such quantities that integral reactor behaviour is achieved.
Water conversion levels ranged typically between 50 and 95%. In
all cases, the exhaust gas appeared to be at water gas shift
equilibrium (54). Unless stated otherwise, materials were
applied in the extruded form (diameter 1.5 mm), the conditions of
extrusion and drying being of no relevance for this study. Prior
to activation, carbonization in a nitrogen flow was performed
during half an hour at the temperature of isothermal activation.
In most cases peat semi coke was used as the carbonaceous
precursor. For making comparison possible, some results on
australian brown coal coke and german anthracite will be given.
The extent of activation will be given as % burn off, determined
on a volatile and ash free basis (; fixed carbon). The various
pore volumes are given per unit weight of starting material,
whereas true particle densities were determined by the replaced
volume in water. Table 5 summarizes proximate and ultimate
analyses of materials used.

Table 5. Proximate and ultimate analyses of materials used.

High moor peat Brown coal Anthracite

0* semi coke 0* coke 0*

~roximate analysis:
- % volatiles 65.0 9.0 52.4 5.0 6.0
- % ash 1.5 3.8 0.9 2.5 3.5
- % fixed C 33.5 87.2 46.7 92.5 90.5

pltimate analysis:
(% daf)
- C 57.1 89.9 68.0 94.4 94.9
- H 5.6 2.1 4.5 1.0 3.0
- N 1.3 1.6 0.3 0.5 1.2
- 0 35.8 6.1 27.0 3.8 0.2
- S 0.2 0.3 0.2 0.3 0.7

True density
(g cm- 3 )
- 1.79 - 1.84 1.44

* 0: starting material
586

6.3 Influence of activation temperature (peat semi coke).

Figure 4 gives the void fraction in the carbon particles as

a function of burn off at various temperatures. No systematic
effect of the activation temperature can be observed, which means
that the pore volume created does not depend on the temperature
of activation upto 1040°C in the case of integral reactor
behaviour.

Figure 5 gives the macropore volume per unit weight of

starting material as a function of burn off. It is obvious that
due to activation no extra macropore volume is created. At
temperatures above 900°C even an effective destruction takes
place at increasing burn off, possibly due to shrinking core
behaviour or shrinkage of the carbon.

I I
Toe I I
0860 Toe
x 900 o 860
0.9 f-- 0 91.0 macro pore volume )t 900
v 1000

,
(cm J ,g-1) 0940
'" 1040 ,/ v 1000
0.8 -- 0.1.0
void fraction ~'IU I '" 101.0
_x/o.
0.7
x/
0.6 /.
0.30
o-<r' -~ 8o:2!
x -0_
<f~x~-- ~_--x

4
~x ~~ ) x
i1- _ w low T

•
O.~ o¢

O.l. .
I

0.20
--"'Q ~_o
-- ~-highI T-
I

0.3- -

0.2 --; 0.10

0:1
,

10 20 30 1.0 % burn off 80 10 20 30 1,0 % burn off 80

Figure 4. Void fraction as a function of burn off during steam

activation of extruded peat semi coke (PH 2 0: 40-82
kPa) •

Figure 5. Macropore volume per unit weight starting material as

a function of burn off.
 587

The effect of this on t·he micro- and mesopore volume increase

(figure 6) appears to be of minor importance: independent of the
activation temperature the pore volume created due to activation
increases upto approx. 50% burn off, after which a leveling off
occurs, that is more pronounced for lower activation
temperatures. The latter means that the transformation of macro-
into mesopores due to shrinkage at high temperatures is likely.

Interesting is the influence of the activation temperature

on the development of the micropore volume (figure 7). At all
temperatures a distinct maximum is observed that is higher for
higher activation temperatures. Low temperature activation
exhibits a rapid increase of the micropore volume upto approx. 20%
burn off, possibly due to opening of restricted pores, whereas
further burning off results in only minor changes in micropore
volume. The mesopore volume development (figure 8) on the other
hand exhibits a steady course that is more pronounced at lower
activation temperatures.

I I I I Toe
0860
micro + mesopore l( 900
volume (cm 3 .g-1) 091.0
v 1000
O.1.0 --+---+---+-- '" T 0C
101.0 -+---+--1
. ...--'%-v'"
high T 0 860
.~~_ 0
~
I /0->< . l l( 900
mlcropore YO ume 091.0
160C1C x ~ low T I (cm'.g-l) v 1000
0.30
>.// x I 0.30 A 10l.O

,~V1 I I I V- ;; "~I
~/ I I ~o't '" (\1,,\
~ I
. :9..

I(
, I i 020 x x I
0.20 l-T-r~-- l(PV'"oV C 00 x~~ I hi9,hT

i j, I

'0
I I I ! flOWT

v
i 10 20 30
I
1.0 % /nlIUirn off
i:
80
V
v
i
i i ,
10 20 30, L.O % bur n off

Figure 6. Micro/mesopore volume per unit weight starting

material as a function of burn off.

Figure 7. Micropore volume per unit weight starting material

as a function of burn off.
588

In conclusion, high activation tempera.tures lead to 1IIOre

microporous carbons than low activation temperatures without
influencing the total pore volume created due to activation. In
principle, two extreme mechanisms of pore volume increase may
occur:
- pore drilling, leading to a steady increase in pore diameter.
- pore deepening, that affects the pore diameter hardly.
Both foregoing results and additional information on mesopore
size distributions suggest pore drilling to be the predominant
mechanism at low temperatures whereas pore deepening is favoured
by high temperatures.

6.4 Influence of residence time (peat semi coke).

Via steam feed regulation, the residence time needed for

establishing a certain degree of activation may be manipulated
under isothermal conditions. From figure 7 and 8 it follows that
with increasing burn off the ratio mesopore volume/micropore
volume increases, the extent however being determined by the
activation temperature. Figure 9 shows that relatively long
residence times (= low PH20) enhance microporosity. Furthermore,
this figure shows the lack of dependency of the degree of
activation with the true particle density. Figure 10 shows on the
other hand that residence time and true particle density are
closely related, which relationship depends mainly on the
temperature of activation and clearly not on the degree of
activation. Combining these results with those from section 6.2
leads to the conclusion that high temperatures and long residence
times enhance microporosity whereas low temperatures enhance
mesoporosity. From figure 10 it follows that high particle
densities cannot be achieved by long residence times below 900°C.
Clearly, particle density increase and therefore an enhanced
graphitization during burn off, is a strongly activated process.
Furthermore, microporosity development seems to be associated
with rearrangement of graphitic layers and growth of these
layers, as visualized by table 2 and figure 2.
 589

I I i
OC
meso pore votune T B
(cm'.g" ~ 9gg 1.95
I I
part icled ens1ittw
(g.cmm::JI)
J
x:4'----j.o~~:1
091.0 I
0.30 v 1000 -+--+---+-
'" 101.0 I
1.90 """.\0):11] 0.25
0.460 I ,

b.77 I
0.20 -..;.---t---+-----+-
i !

ilD515
l.B5~ I· PHzO (kPa) -
01.0

+
0.10 I x 57

'"
o~~ '"
'.0 I ... 1_ .1 ~ ~i
20 30 J.O % burn off 80 10 20 30 lIOl'!!o burn off 80

Figure 8. Mesopore volume per unit weight starting material as a

function of burn off.

Figure 9. True particle density as a function of burn off during

steam activation of extruded peat semi coke at 940°C;
the numbers indicate the ratio mesopore/micropore
volume.

30~------+------+--~
residence time
( h)

204---7-+----+-~~+---
/860
/ /
, I

/ /900
10 ~~-/----~~/------~~----~---
/ I I
/ I
./v""",'
/1
/ /1
I II/»,/

I;' _-.I
'lBO I 1:8'5----- 190 1.95
particledens:ity (g.cm3 )
Figure 10. True particle density as a function of residence time
during steam activation of extruded peat semi coke; the
numbers indicate the activation temperature in °c (for
symbols, see figure 9).
590

6.5 Influence of shapening and extrusion diameter (peat semi

coke).

The effect of shapening (granulation, extrusion) or the

extrusion diameter on the development of micro- and mesoporosity
appears to be comparable and associated with accessiblity of the
solid phase towards the gaseous reactant. From a simple
qualitative evaluation of figure 7 and 8 it follows that
preferential activation in the outer layer of bigger particles
due to a hindered internal diffusion and resulting in
heterogeneous burn off, leads to an enhanced mesoporosity
development and therefore less microporous carbons than in the
case of high effectiveness factors. This is in agreement with
results from Chiche et al (55), who concluded that the optimal
pore structure for a starting material would be bimodal,
containing micropores to provide surface area and large
macropores to provide access to the interior of the particle.

6.6 Influence of the gas phase composition (peat semi coke).

From figure 9 it has become clear that the residence time

has a predominant influence on the porosity development during
activation. Due to the integral 'reactor behaviour short residence
times are associated with high water partial pressures. This
phenomenon might explain the reported influence of the water
partial pressure, stating that high pressures hinder
microporosity development (49).

Furthermore, although not studied in detail, the

mesoporosity enhancing effect of activation in C02 was confirmed:
in 69 kPa C02 at 940°C and approx. 50% burn off the meso- versus
micropore volume ratio is 0.69, whereas under comparable
conditions during steam activation this ratio is 0.44.
Apparently, the slow diffusion rate of C02 prevents an effective
development of the micropore structure. Marsh and Rand (56)
observed that CO added to the reactant gas enhanced
microporosity. Thus by lowering the affinity of carbon towards
C02 via CO inhibition, the micropore structure may become more
accessible.
 591

o.8 I I I
vaid fraction /
.,,"
Figure 11. Void fraction as
.EY a function of
o.7 -- ----
burn off for
o.6 /:/ ~v X
various extruded
feedstocks.
V
//

o.5 / r--- geat semi cok e

x/ f-o- rown coal co ke
v f--lC- anthracite
0 0'7
I
x I

O. 3 I I I I I
m acropore volume .
I
(cm1 .g -1) ---- peat semi coke
o.2 I
I ~ brown coal coke
-x- anthracite
o.1 I o. 30-+--+-+--+-+--+-~
I x -- -- ----low T
10. 20 30 1.0 % burn off 80 I
I
Figure 12. Macropore volume per O.201--~
I - highT
.. --+---+--=-="kc~~---'I--=--t--
unit weight starting
material as a function
of burn off for
various extruded
feedstocks. 0.10 +----1---+--+---"'..+--+--1---+-
high T

micro+mesoporl) I
I
volume (cm 3.g 10 20 )0 1.0 % burn off 80
0.1.0
--- high T Figure 13. Micro/mesopore

~
~ --- - lowT
I
volume per unit
weight starting
, material as a
0.30 "" function of burn
""
,
,/ " / off for various
extruded feed-

/'
x
/
/
stocks.
I

V
0.20
I _____ peat semi coke
I -0-- brown coal coke
I
I -0- idem low T hig hT
I
I - x - anthracite high T

V 0
592
6.7 Influence of alkali additives (peat semi coke).

For the production of specific types of carbon the addition

of potassium or sodium compounds is common practice (24). Besides
a reactivity enhancing effect, these compounds are known to
establish high surface areas and therefore promote microporosity
(30). This appears to occur also under integral conditions: in
the presence of potassi~m, activation in 69 kPa H20 at 940°C
results in a meso- versus micropore volume ratio of 0.25 at 50%
burn off, whereas in the comparable case in the absence of
potassium 0.44 is found. Again the relationship between
microporosity and high true particle densities holds: in the
presence of potassium, values of over 2.0 occur (compare figure
9).

6.8 Influence of the type of carbonaceous feedstock.

Figure 11 shows the development of the void fraction within

the carbon particles on increasing burn off for anthracite and
brown coal coke, compared with peat semi coke. It is clear that
for anthracite the initial porosity is much lower than for the
other two feedstocks, which may influence the accessibility. The
development of the macropore volume (figure 12) indicates a
considerable destruction of the macroporous structure in
activating anthracite. To a lesser extent this also holds for
brown coal coke activation. As in the case of anthracite this
lowering of the macropore volume is not accompanied by higher
micro- and mesopore volumes (figure 13), shrinkage is not a
predominant phenomenon, as might be the case for brown coal coke.
Therefore, anthracite shows considerable shrinking core
behaviour, which is likely when the accessibility is poor and,
due to the low reactivity, the necessary activation temperature
is high. The temperature effect on the total pore volume created
during activation (figure 13), on the development of micro-
(figure 14) and of mesoporosity (figure 15) in the case of brown
coal coke appears to be fairly comparable with peat semi coke
activation, although mesoporosity seems to be somewhat less
developed. In the case of high temperature anthracite activation,
the low temperature pattern of microporosity development for peat
semi coke is followed (figure 14), without exhibiting the corres-
ponding enhanced mesoporosity development (figure 15). So from
these results it becomes clear that a lack of accessibility
strongly hinders a desired porosity development during
activation. In essence, carbons based on anthr"acite remain
microporous. Singh (57) concluded that in the gase of anthracite
actiyation pores with diameters less than 12 A are only created
by volatile loss during heating, whereas activation occurs mainly
in larger pores.
 593

I I I I
micropore volume
0.30 (cm 3.g-1) --I---'°'GDI---+--+------+--
I / /' ~~---- 'high T
~/ ,~
0.20 /1 _~ --9(-' f>.--_
;' /" x ........
~/~./ 00 .... 'Low T
,
I ' I
""
O.jl~ +----+----+------+_
obrown coal coke
o idem low Thigh T

",
/' x anthracite high T
--- peat semi cOKe
I
I I I
~ 10 20 3r1ll JIll) % burn, off SO

Figure 14. Micropore volume per unit weight starting material as

a function of burn off for various extruded
feedstocks.

mesopore vo~ume
[ (cm"!J"')
0.30 -+----l---+- --- peat semi coke_
a brown coal colke-
o idem-lowl IhIliglhl T
x anthracite-lhI,iglhl T

0.20 _+--+__
/
J L
low T
I I
, higlhlll'
0.1 0 f---_+-'7f'---:F:Q(D:8-~~/__l'-_+--
",,,

80

Figure 15. Mesopore volume per unit weight starting material as a

function of burn off for various extruded feedstocks.
594

7. MODEL FOR PORE DEVELOPMENT AND PRACTICAL IMPLICATIONS.

7.1 Evaluation of foregoing information.

Activation of carbonaceous materials apparently is only

possible when an accessible initial porosity occurs. Pore volume
increase may proceed either via drilling or deepening of existing
pores, the extent of both mechanisms being determined by the type
of feedstock and the process conditions. Due to the graphitic
character of activated carbon, removal of carbon via gasification
is thought to occur in a parallel mode preferentially.
Introversions in the graphitic layers with the dimension of
micropores appear to be highly unlikely, which follows from the
C/H ratio of activated carbons. Therefore, the orientation of the
graphitic layers and consequently the extent of graphitization
during activation, highly influences whether slit- or V-shaped
micropores, being the void fraction in-between separate layers,
or more irregularly shaped mesopores, being the void fraction in-
between primary crystallites, are the result of carbon removal.
A consequence of micropores being the void fraction in between
graphitic layers is that tgeir diameter is restricted to distinct
values (approx. 4 and 8 A, (26» and that pore deepening is the
predominant mechanism for pore formation, whereas the diameter of
mesopores varies more randomly (figure 16) but increases with
increasing burn off due to pore drilling.

In this study it was shown that true density increase via

high temperatures and long residence times involves microporosity
predominantly. Essential is, however, a good accessibility of the
initial micropores. A means to establish this at high
temperatures is strong product inhibition. Probably strong
analogies occur with the socalled "rotten apple" phenomenon (59).

As a result of the foregoing pore development model it

follows that the pore size distribution in predominantly
microporous carbons will be more or less monodisperse (large
micropores, small mesopores), whereas predominantly mesoporous
carbons will be bidisperse (small micropores, large mesopores).
The application merits of this will be discussed further.
 595

15 I
.Q..Y.. I
dD % burn off
13.7 -
10 67.8 ----
I,
I
I

·
I

··
a
5

"
, ,~"'~
~"-'

" \.. 10
'"
"
" ....... - - .... _.. ----....... _--
100
slt tdiamet er (om)

150----~------~-------------r---­
% burn off
Scum
(m 2.g") 13.7
67.9

100------~--~~-------------r----

50----~----~-------------_+----

100
sti tdlilCll'1'1lete-r trtm)
Figure 16. Volume distribution (a) and cumulative mesopore
surface area (b) at two burn off levels for steam
activated extruded peat semi coke.

7.2 Practical implications.

7.2.1 Production of activated carbon. The process of granular

activated carbon production generally involves the following
steps:
- Crushing and grinding of the carbonaceous feedstock.
- Kneading with specific binders.
- Extrusion or briquetting.
- Drying and carbonization.
- Activation.
- Classification and handling.
 596
For the production of powdered activated carbons the specific
steps that are associated with pre-sizing of the material lapse.
Generally, activation occurs using the carbonaceous feedstock as
such, after which grinding establishes the desired particle size
distribution.

The activation process, being the most essential step, is

invariably performed in atmospheric equipment. Temperature
regulation generally takes place via control of secondary air
supply. Although a large variety of apparatuses may be used (58),
rotary kilns, shaft kilns (and more specifically: multiple hearth
furnaces) and fluidised bed reactors are the most common
practice. Without further discussing advantages and disadvantages
of these types of kilns, specific features affecting pore
structure development during activation are:
- Differences in residence time of the carbon in the hot stage of
the kiln. A drastic decrease is observed in the range rotary
kiln (days) - multiple hearth furnace (hours) - fluidised bed
(minutes), due to differences in bed height and gas flow
profile.
Differences in maximum temperature and temperature profile in
the kiln. Generally fluidised bed reactors exhibit lower
maximum temperatures than the other two kiln types, whereas a
more uniform temperature profile is established.
Temperature regulation in multiple hearth kilns is possible for
every hearth separately, whereas in rotary kilns the
temperature profile is more a result of the various process
parameters, like steam feed and the rate of carbon dosage and
withdrawal.
Foregoing means that every type of kiln will produce more or less
specific types of activated carbon. In general, a decreasing
mesoporosity may be observed in the range fluidised bed-multiple
hearth furnace-rotary kiln. It is the skill of the producer to
choose the right kiln in combination with the optimum feedstock
to produce activated carbons specifically for desired
applications.

7.2.2 Application of activated carbon. As stated before, the

applicability of an activated carbon is, among others, determined
by both the maximum adsorption capacity (often specified per unit
of volume) and the rate of adsorption c.q. desorption (the latter
e.g. in the case of solvent recovery via steaming of loaded
carbons). These two'factors are inherent in the pore structure of
the carbon and sizing of the particles. Other factors like
purity, hardness, filtration behaviour etc. are beyond the scope
of this paper.
 597

The maximum adsorption capacity is strongly related with

both the degree of activation (with the price-performance ratio
as the main criterion) and the molecular dimensions of the
adsorbate, as the accessible surface area may differ dramatically
for small and large molecules. For the removal of large organic
molecules, the carbon requires a high surface area present in the
mesopores as the micropores remain inaccessible (e.g. colouring
matters in base sugar). Yet, a high mesopore surface area is not
essentially equivalent with a high mesopore volume, as was shown
in section 6, due to pore enlargement on increasing burn off. It
was proven experimentally that at equal mesopore volume molasses
decolourization improves with higher activation temperatures or
longer residence times. The micropore surface area as a function
of micropore volume appears to be less critical with respect to
the activation conditions. This means that high micropore volumes
are always inherent in high surface areas, although also in this
case molecular sieving phenomena may occur. Furthermore,
irreversible adsorption (unwanted in the case of solvent recovery
purposes) is a specific feature of adsorption in micropores (=
adsorption at low relative pressures). These sites of high
adsorption energy are probably equivalent with the sites that
enhance propagation of carbon removal during activation.

The rate of adsorption is closely related with the

accessibility of the porous structure and, again, the molecular
dimensions of the adsorbate. Moreover, the medium from which
adsorption occurs (gas or liquid) is predominant. Generally,
during adsorption from the gas phase in filters, the rate of
diffusion is not limiting. This means that in these cases the
maximum adsorption capacity is predominant, although e.g. for
solvent recovery purposes the rate of desorption from the carbon
is sometimes considered to be a bottle-neck. For liquid
applications however, the accessibility of the porous systems is
of predominant importance, especially for large molecules that
exhibit slow diffusion transport in narrow pores. This urges the
need for a short distance between the wider pores that facilitate
transport and more narrow pores in which effective adsorption
takes place. In practice this may be achieved by establishing a
well developed mesopore structure via the pore drilling
mechanism. However, this type of pore formation excludes large
mesopore surface areas. This means that optimum carbons for this
type of application are always a compromise between high
adsorption capacities and good accessibility. It will be clear
that small variations in process conditions during application
may lead to completely different optimum compromise conditions
during the production of activated carbon. Again, the skill of
the producer is predominant in these matters.
598

8. LITERATURE CITED.

1. Walker, P.L. Jr., F. Rusinko and L.G. Austin. Advan. Catal •

.!l., 133 (1959).
2. Long, F.L. and K.W. Sykes. Proc. Roy. Soc. ~215, 100 (1952).
3. Juntgen, H. Carbon~, 297 (1968).
4. Thrower, P.A., J.C. Bouquet and G.K. Matthew. Carbon ~, 457
(1982) •
5. Katta, S. and D.L. Keairns. Proc. 180th Nat. Meeting Am.
Chern. Soc., Div. Fuel Chern., 101, Las Vegas (1980).
6. Kapteijn, F., P. Coevert and J.A. Moulijn. J. Phys. E. ~,
1064 (1982).
7. Otto, K. and M. Shelef. Carbon~, 317 (1977).
8. Wigmans, T., F. van Dam and J.A. Moulijn. Carbon ~, 309
(1981).
9. Overholster, L.G. and J.P. Blakely. Carbon~, 385 (1965).
10. Radovic, L., P.L. Walker Jr. and R.G. Jenkins. Fuel ~, 849
(1983).
11. Born, M. and E. Klose. Fuel ~, 482 (1983).
12. Otto, K., H. Sorek, L. Bartosiewicz and M. Shelef. Fuel ~,
477 (1984).
13. Riley, H.L. Quart. Revs 1, 59 (1947).
14. Biscoe, J. and B.E. Warren. J., Appl. Phys • .!l, 364 (1942).
15. Fredenhagen, K. Z. Anorg. allg. Chern. 208, 2 (1932).
16. Ruff, O. ibid. 217, 1 (1934). ---
17. Arnell, J.C. and W.M. Barss. Can. J. Res. A26, 236 (1948).
18. Franklin, R.E. Acta Cryst. ~, 107 (1950).
19. Smith, T.D. J. Chern. Soc. 923 (1952).
20. Kipling, J.J., J.N. Sherwood, P.V. Shooter and N.R. Thompson.
Carbon 1, 315 (1964).
21. Blayden, H.E. and D.T. Westcott. Proc. 5th Conf. Carbon, 97
( 1963) •
22. Mochida, I., T. Ando, K. Macda, K. Fujitsu and K. Takeshita.
Carbon~, 319 (1980).
23. Marsh, H. and P.L. Walker Jr. Fuel Process. Techn. ~, 61
(1979).
24. Kienle, H. von and E. Bader. Aktivkohle und ihre industrielle
Anwendung (Enke Verlag, Stuttgart, 1980).
25. Dubinin, M.M. Uspekhi Khim 3!, 3 (1955).
26. In: "Activated Carbon, ••• a Fascinating Material" (Eds. A.
Capelle and F. de Vooys), 13, Norit Amersfoort, the
Netherlands (1983).
27. Miura, S. and P.L. Silverstone Carbon~, 93 (1980).
28. Weiss, D.E. Proc. 5th Conf. Carbon, 65 (1962).
29. Blackburn, A. and J.J. Kipling. J. Chern. Soc. 4103 (1955).
30. Wigmans, T., A. Hoogland, P. Tromp and J.A. Moulijn. Carbon
~, 13 (1983).
31. Capon, A. and F.A.P. Maggs. Carbon~, 75 (1983).
32. Kipling, J.J. Quart. Revs. 10, 1 (1956).
33. Smith, R.N. and J. Mooi. J.-Phys. Chern. 59, 814 (1955).
 599

34. In: "Activated Carbon, ••• a Fascinating Material" (Eds. A.

Capelle and F. de Vooys), 48, Norit Amersfoort, the
Netherlands (1983).
35. Taylor, H.S. Proc. Roy. Soc. ~, 105 (1925).
36. Kraus, G. J. Phys. Chem. 59, 343 (1955).
37. Mattson, J.S., L. Lee, H.B. Mark and W.J. Weber. J. of
Colloid and Interface Sci. 33, 284 (1970).
38. Rist, L.P. and D.P. Harriso~ Fuel 64, 291 (1985).
39. Rodriguez-Reinoso, F., A. Linares-Solano, J.M. Martin-
Martinez and J.D. Lopez-Gonzalez. Carbon~, 123 (1984).
40. "Effect of porous structure on carbon activation", EPA
project 17020 DOC, june 1971.
41. Donnet, J.B., J. Baptiste and P. Couderc. J. Chem. Phys.
Physiochim. BioI. ~, 1399 (1968).
42. Kaji, R., Y. Hishinuma and Y. Nakamura. Fuel 64, 297 (1985).
43. Wilson, J. Fuel 60, 823 (1981).
44. Gouker, T.R. and:R. Liotta. Fuel 64, 200 (1985).
45. Ludvig, M.M., G.L. Gard and P.E. Emmet. Fuel 62, 1393 (1983).
46. Fryer, J.R. Carbon 19, 431 (1980).
47. Huck, G. and J. Ka~il. Brennst. Chemie l!, 97 (1953).
48. Dubinin, M.M. and G.M. Plavnik. Carbon~, 183 (1968).
49. Sen, S., N.B. Bhattacharyya and S.R. Naidu. Technology, Q.
Bull. Fertil. Corp. India~, 69 (1966).
50. Peters, W. and H. Juntgen. Brennst. Chemie 46, 175 (1965).
51. Thrower, P.A., G.K. Matthew and N.J. McGinnis. Carbon~, 465
(1982).
52. Shafizadeh, F. and Y. Sekiguchi. Carbon 24, 511 (1983).
53. Spiro, C.L., D.W. McKee, P.G. Kosky, E.J. Lamby and D.H.
Maylotte. Fuel 62, 323 (1983).
54. In: "Activated Carbon, ••• a Fascinating Material" (Eds. A.
Capelle and F. de Vooys), 92, Norit Amersfoort, the
Netherlands (1983).
55. Chiche, P., J. Coue, S. Durif and S. Pregermain. Carbon 2,
297 (1969).
56. Marsh, H. and B. Rand. Carbon~, 47 (1971).
57. Singh, D.O. Chem. Ind. (London) 1968, 620 (1968).
58. Yehaskel, A. Activated Carbon, manufacture and regeneration
(Noyes Data Corporation, Park Ridge, New Jersey, U.S.A.,
1978).
59. Erk, H.F. and M.P. Dudukovic. Ind. Eng. Fundam. ~, 55
(1983).
 601

PREPARATION AND CHARACTERIZATION OF ACTIVATED CARBONS

F. Rodriguez-Reinoso.

Department of Inorganic Chemistry. University of Alicante.

Alicante. Spain.

1. INTRODUCTION

The name activated carbon is given to carbonaceous materials

wi th large adsorption capacity manufactured from a large number
of materials such as coconut shells, wood, peat, and coal. X-ray
analysis of activated carbons shows a structure which is much
more disordered than that of graphite, having crystallites only a
few layers in thickness and less than 10 nm in width. The spaces
between the crystallites of activated carbon constitute the
microporous structure with a large internal surface to which it
owes its remarkable adsorption properties. The structure of
activated carbon is dealt with in more detail in another chapter.

The total porous structure of an activated carbon is formed

by a wide range of pore sizes, although for practical reasons
they are classified in three groups, following the IUPAC
definitions (1), according to their width:

macropores: more than 'V 50 nm

mesopores: between 'V' 2 and /V 50 nm
micropores: less than 2 nm

The macropores of an activated carbon act as transport

pores, enabling the molecules of the adsorptive to reach the
smaller pores situated in the interior of the carbon particle.
Thus, the macropores are not important from the point of view of
the amount adsorbed in them, since their surface area is very
low, but they affect the rate of diffusion into the meso and
micropores.
602

The micropores constitute the largest part of the internal

surface and, consequently, most of the adsorption takes place
within them; thus, at least 90% of the total surface area of an
activated carbon corresponds to the micropores. The mesopores,
which branch off from the macropores, serve as passages to the
micropores for the adsorptive besides acting as pores where
capillary condensation can take place.

As mentioned above, activated carbons have, in general,

pores belonging to the three groups and, in this sense, they can
be considered general adsorbents, provided the micropores are
accessible to the adsorptive molecules. The usual carbons are not
able to adsorb selectively molecules of different dimensions.
However, considerable development work is taking place on the
preparation of so-called carbon molecular sieves for the
selective separation of adsorptives with different molecular
sizes.

The porous structure of an activated carbon is a function of

the raw material used in its preparation, the activation method
used and the extent of activation. This is the reason why the
surface area or pore volume of activated carbons can vary so
widely from one kind to another. However, the use of an activated
carbon in a given application will be determined by the
proportion in which the pores of different sizes are present.
Thus, whereas for the adsorption of gases and vapours an
essentially microporous carbon would be adequate, for the
adsorption of solutes from solution a well developed meso and
macroporosity is, in addition, required. Of course, the adsorp-
tive capacity of an activated carbon is not only defined by its
porous structure but also by the chemical nature of its surface.

As previously stated, many raw materials can be used for the

preparation of activated carbon. This chapter will concentrate on
the preparation and characterization of activated carbons from
agricultural by-products, especially olive stones and almond
shells, which are very abundant in Mediterranean countries (2,3).
Especial attention will be given to the different activation
processes used and their effect on the porous texture of the
final product; a somewhat detailed analysis of the interesting
problem of adsorption on the micropores will also be presented.

2. PREPARATION

The carbonaceous starting materials used for the preparation

of the activated carbons described here have been agricultural
by-products such as olive stones, almond shells and fruit stones
(apricot, peach, etc.); however, for the sake of simplicity the
results will be mainly referred to olive stones and almond
 603

shells. These materials were crushed and washed in a 10% solution

of sulphuric acid for 6 h and refluxed in distilled water to zero
acid removal; they later were sized to about 1.6-2.0 mm diameter
(3) •

Most of the commercial activated carbons are prepared by

steam activation in a two-stage process (carbonization followed
by activation), although some are prepared by CO2 activation
(also in a two-stage process) and by chemicar activation
(carbonization of a mixture of the carbonaceous material with a
chemical such as phosphoric acid and zinc chloride solutions).

Since the activated carbons described here are prepared on a

laboratory scale and the basic interest is to learn about the
activation process and its effect on the development of porosity,
activation with CO 2 has been selected because of its simplicity,
although some rererences to steam activation will also be
introduced.

The conventional activation procedure includes the carboni-

zation of the starting material in an inert atmosphere at a given
temperature. In order to learn about the most appropriate
conditions of the carbonization process the following experiment
may be carried out (the results given are referred to almond
shells (3)). The clean raw material is carbonize~ a::: 1 different
temperatures in a continuous nitrogen flow (1.5 ~r's ) for 2 h
using the same heating rate in all runs (5K.min ). The plot of
the yield of each carbonization as a function of the temperature
is given in Figure 1 which shows that above 773K the yield of the
process is almost constant at around 25%. Thus, 773K may be
considered to be the lowest temperature to be used in the
carbonization process if this is to be carried out before any
activation process. Olive stones exhibit a very similar behaviour
in the carbonization process (4).

The carbonized 3ma!rrial is later activated using a conti-

nuous flow (1. 5 cm . s ) of carbon dioxide; although a large
number of temperatures has been used both for the carbonization
(always above 773K) and the activation processes, temperatures
between 1073 and 1173K have been used as a general rule.

A possible modification of this conventional process is to

carry out a direct activation in CO 2 under the same experimental
conditions, without using the previous carbonization of the
precursor. A further modification would be to carry out a heat
treatment of the clean raw material in air at 573K (a temperature
at which the evolution of gases and tarry substances is
important) followed by activation in a CO 2 flow.

In order to compare the three different activation pro-

604

100

60

20

273 673 1073

Temperature (K)

Figure 1.- YieLd of the carbonization process for aLmond

sheLLs as a function of temperature; after (3).

cesses, the overall yields (the product of the yields for the
different stages used in any preparation method) obtained in the
case of a flow of CO 2 at 1123K have been plotted in Figure 2 for
almond shells as a function of activation time (4). It should be
noted first that for each series the experimental points are in a
straight line; this indicates a gradual and uniform gasification
rate (in the burn-off range covered here) which seems independent
of the poros i ty development during the activation process. A
second point is that the rate is influenced by the different
treatments carried out. In the case of the air-C0 2 activation
process, the yield is always larger than in the other two series
for a given activation time; the behaviour of the other two
process is very similar, giving comparable yields. It seems then
that the direct reaction of almond shells with CO 2 during heating
from room temperature to 1123K is not important and does not
imply a noticeable activation; in other words, its effect is
similar to carbonization carried out in nitrogen (5,6); similar
behaviour has been found for olive stones (4).

The results obtained with the air-C0 2 activation process are

also very interesting since the yield (see Figure 2) is larger
 605

o 5 10 15 20

Time (h)

Figure 2.- Overall yield as a function of activation time, after

(4); A :N2-C0 2; 0 :direct CO 2 ; 0 :air-C0 2

than in the other two series for comparable activation times, the
difference decreasing with increasing activation time. Thus,
extrapolation to t=o of the experimental data.for the CO 2 direct
activation and for the carbonization-activation processes give
yields of 24% similar to the yield for carbonization in nitrogen.
On the other hand, extrapolation of the data for the air-CO
process gives a yield of 32%, indicating that the air treatmen€
at 573K has modified the decomposition mechanism of the precursor
(4) •

A typical example of the effect of activation time on the

gasification of a carbonized material is in Figure 3, which
includes plots of burn-off as a function of time for activation
in CO 2 at 1098K for olive stones and almond shells carbonized in
nitrogen at 1123K (7,8).

The plots of Fig~e 3, which are practically linear in the

range covered (8-80% burn-off), indicate that the reaction rate
is almost constant up to about 80% burn-off and that very
probably the diffusion of CO 2 in the carbon particle is not a
rate determining factor since it is the same for low burn-off
(narrow porosi ty) as for large burn-off (wider porosi ty). An
approximately similar behaviour has been found for other agricul-
tural by-products (9,10).
606

60
......
......
~

c...
c...
0
I
r::: 40
1-0
::1
I:Q

20

Time (h)

Figure 3.- Burn-off as a function of activation time;

o:carbonized olive stones; tJ. :carbonized aLmond sheLLs (7,8).

Another modification introduced in the preparation of

activated carbons is the use of the catalytic effect of iron in
the C-C0 2 reaction (11,12). Almond shells were impregnated with
Fe(N0 3 )3 solutions (0.1, 1 and 5% of iron) and later activated in
CO 2 a~ different temperatures and lengths of reaction. To show
the long-term effect of iron, the extent of activation (burn-off)
has been plotted in Figure 4 as a function of activation time for
samples impregnated with 5% iron solution (13).

As Figure 4 shows, there is initially a considerable

increase in the extent of the activation, but after 8 h of
reaction the catalytic effect seems to have been lost and the
rate of activation becomes comparable with that of the uncataly-
zed reaction (i.e., direct activation in CO 2 ), Furthermore,
whereas the uncatalyzed reaction (as mentioned above the direct
activation with CO 2 is equivalent to the conventional carboniza-
tion followed by activation) at 800 c C produces a 4% burn-off
after 8 h (referred to a theoretically carbonized material) the
 607

80 ~---------------...,

60

......
......
~

c,..
c,..
0 40
I
E
~
CO

20

"....../
".....
".,..""'"
r'
,,""'"
0
0 4 8 12 16 20

Time (h)

Figure 4.- Burn-off against activation time for iron catalyzed

C-C0 2 reaction (0) and the uncatalized reaction (e) (13).

5% iron-impregnated material produces a 67% burn-off in the same

time.

Although activation with CO 2 is simple and convenient on the

laboratory scale it is interesting to compare it with steam
activation (used in most of commercial plants). The comparison
presented here is between samples prepared by CO 2 activation of
carbonized plum stones in the 800-850 o C temperature range and
samples prepared by steam activation of the same precursor in the
900-950 o C temperature range (10); the temperatures and periods of
activation were selected in order to have comparable activation
rates and burn-off. Figure 5 includes a plot of burn-off as a
function of activation time similar to those for CO 2 activation
(Figures 3 and 4). Again, the plot is linear in file burn-off
range covered for the steam activation at 900°C and a partial
pressure of 12.3kPa.
608

80

20 40
Time (h)

Figure 5.- Burn-off as a function of steam activation time

(plum stones)

3. CHARACTERIZATION

3.1. Experimental Techniques

The effects of the different activation conditions on the

porous structure of activated carbons have been characterized by
vapour and solution adsorption techniques and mercury porosime-
try. The adsorptives used. have been (i): from the vapour phase:
- N2 (77K), CO (273 and 298K) , n-butane and iso-butane (273K),
benzene, 2, 2-dlmethYlbutane and iso-octane (298K); (ii): from
aqueous solution at 298K:- p-nitrophenol, methylene blue, Orange
II, Crystal Violet and Victoria Blue.

Adsorption of gases and vapours has been measured in a

conventional gravimetric system using silica spring balances (2);
adsorption from solution has been determined using a spectropho-
 609
tometric method (3). The mercury porosimetry experiments were
carried out using a Carlo Erba 200 mercury porosimeter which
covers the pore size range from 7.5 to 15.000 nm diameter.

The n-nonane preadsorption technique has also been used to

complete the microporosity evaluation; the experimental procedure
(14) can be summarized as follows: after determination of the N2
(77K)_gdsorp tion isotherm the carbon was outgassed (523K for 4 n
at 10 torr) and n-nonane was admitted to the system at 77K for
30 min and later evacuated at room temperature for 12 h; then,
the adsorption of N (77K) was determined on the carbon with the
n-nonane preadsorbe~

3.2. Adsorption of N2

Since activated carbons are essentially microporous, the

adsorption isotherms are of type I having a very steep initial
branch followed by a more-or-less gradual approach to the linear
branch at higher relative pressures. The shape of the isotherm
will be a function of the pore size distribution; in general
terms, the adsorption isotherm for microporous carbons can belong
to one of the three types shown in Figure 6. Type a corresponds
to a carbon which is exclusively microporous and having
micropores with very narrow dimensions; once the micropores are
filled wi th the adsorbate the amount adsorbed does not increase
for higher relative pressures, thus giving a well-defined plateau
almost paralell to the relative pressure axis. Type b isotherm
corresponds to a carbon in which the micropore size distribution
is wider than for type a so that the linear branch is reached
gradually, the isotherm showing a more rounded knee. Carbons
glvlng isotherms of type b will then have narrow micropores
(responsible for the steep initial branch) and wider micropores
-supermicropores so-called by Dubinin- in which the adsorption
mechanism (of a cooperative nature) is different to that of the
narrow micropore filling (enhancement of the enthalpy of
adsorption) (15). The basis of adsorption is dealt with in more
detail in another chapter. Type c isotherms are found for typical
activated carbons in which a large and widely distributed
microporosi ty is accompanied by a well-develop mesoporosi ty so
that the linear branch at higher relative pressures is no longer
paralell to the pressure axis.

The shape of the adsorption isotherm is, of course, also a

function of the molecular dimensions of the adsorptive molecule,
since the critical parameter is not the pore size itself but
rather the ratio of the size of the pore to that of the adsorbed
molecule. According to Gregg and Sing (15), for slit-shaped
pores, as one could expect in activated carbons, the first stage
in the adsorption in micropores (enhancement of the enthalpy of
610

-
a b c

Figure 6.- Different types of isotherms on microporous carbons.

adsorption) would be expected to be limited to widths below two

molecular diameters and that of wider micropores (giving rise to
the rounded knee of the isotherm) to widths between two and five
molecular diameters. For nitrogen at 77K these values correspond
to 0.3 to 0.7 nm, and from 0.7 to 1.8 nm, respectively.

The adsorptive most commonly used for the characterization

of activated carbons is N2 at 77K, but the use of this adsorptive
may present some problems in carbons with low burn-off and
carbons with a very large degree of activation. Some results
obtained with carbons prepared from almond shells and olive
stones will be presented to illustrate this point.

Figure 7 includes the N2 (77K) adsorption isotherms on

carbonized almond shells (16); the adsorption process is very
slow so that the isotherms of Figure 7a required more than 100 h
allowing around 10 h for each experimental point to reach what
appeared to be equilibrium. This is only an apparent equilibrium
since the adsorption isotherms carried out in two of the carbons
(see Figure 7b) with a large contact time to ensure equilibrium
(more than 6 weeks) show that the amount adsorbed is far greater
than those of Figure 7a. These results indicate that the
carbonized materials are almost exclusively microporous and that
the pore dimensions are similar to those of the N2 molecule,
making the adsorption a very slow process.

In order to test this hypothesis the adsorption of N2 was

 611

140 a C-3

~ C-1
C-S

~
100 C-2
C-6

C-4

r
60

r-t 1
I
.
QO
Q.. 160 b C-S
C-6
E-<
CIl
C")
e()
.......
> 120

80
C-S
c C-6
C-4

r
140
C-2
C-1
C-3
100
I 1
200 400 600 800
p(torr)

Figure 7.-Adsorption of N2 on carbonized almond shells. a: apparent

equilibrium (77K); b: equilibrium (77K); c: 90K.; redrawn from (16).

carried out at a higher temperature (90K) and it was found that

the time needed to reach equilibrium was much shorter and that
the amount adsorbed (Figure 7c) was, in general, larger than that
determined in conditions of apparent equilibi'ium but smaller than
that corresponding to the equilibrium conditions. Furthermore,
the isotherms for all carbonized materials are almost coincident
and this is in agreement with the similarity of the two isotherms
determined under equilibrium conditions (Figure 7b).

It seems then that the adsorption of N2 on these materials

must take place through very narrow constrictions in the
micropores so that the entry into the cavities is a rate
612

-controlled process with a positive temperature coefficient.

Thus, an increase in adsorption temperature leads to an increase
of the rate of diffusion of the molecule through the constric-
tions and to an increase of the amount adsorbed. Consequently,
the width of the constrictions in the micropores must be close to
the molecular diameter of nitrogen (16).

When these carbonized materials are activated in CO 2 the N2

(77 and 90K) adsorption isotherms (Figure 8) show a completely
different behaviour. Now the time needed to reach equilibrium is
much shorter and the amount adsorbed at 90K is lower than at 77K
(16) . The activation has eliminated the constrictions of the
micropores and the adsorption process is no longer controlled by
gaseous diffusion. The amount adsorbed increases with increasing
activation and there is a noticeable change in the shape of the
isotherms: the knee becomes more rounded and the linear branch
becomes steeper. This is a consequence of the widening of the
microporosity and the development of the mesoporosity produced by
the increasing activation.

At this point it is interesting to note the effect of the

different activation methods described above on the adsorptive
capacity of the resulting activated carbons. Since the adsorption
isotherms of N2 at 77K are of more-or-less distorted type I, a
good way of comparing the different carbons is to compare the
volumes adsorbed at P/p =0.1 (as an approximate measure of the
micropore volume) and P/Bo=0.95 (as an approximate measure of the
micropore plus mesopore volume). Table 1 includes such values for
activated carbons prepared from a common precursor using
different activation methods but having comparable burn-off.

Table 1
3 -1
Volume (cm STP.g ) at P/PO=O.l and P/PO=0.95 for
different activated carbons
burn-off
Precursor method VO. l VO. 95
(%)

Almond shells CO 2 (direct) 46 324 362

Almond shells N2-C0 2 46 336 379
Almond shells air-CO 2 50 422 543
Almond shells CO 2 (Fe catalyzed) 43 156 184
Plum stones N2-C0 2 47 314 380
Plum stones steam 51 300 443
 b ~ .-25
a - - • -25

260

300 ~ / -

.-24
~ .l-23
..... 250 .....
I
,,·r~ •
I.., ..,
Q. c: 1'0
I- I-
en A-23 en
.-22
"" u• "". u
~ r ~ · : : .-21
> >
140
A-Z2

A-21
150 I- • 100

100 10
o 100 lOO 500 700
leo 300 500 700
P(torr}
p( torr}

Figupe 8.- N2 adsopption isothepms on activated capbons fpom almond shells; a: ??K; b: 90K.

0'\
t-'
VJ
614
The results of Table 1 show that for carbons prepared from
almond shells direct CO 2 activation and conventional carboniza-
tion followed by CO 2 actlvation produce carbons with very similar
adsorption behaviour; however, the air-C0 2 activation leads to a
larger adsorption capacity, caused by the more extended develop-
ment of both micro and mesoporosity. The iron catalyzed CO 2
activation does not affect much the development of both
microporosi ty and mesopore volume under the experimental condi-
tions used (13). However, steam activation does not especially
develop the microporosi ty but considerably increases the
mesopore volume, as seen in Table 1 for activated carbons
prepared from plum stones using both CO 2 and steam activation.

3.3. Adsorption of CO 2

The adsorption on carbonized, unactivated materials has been

a typical example of the possible problems found with the
adsorption of N2 (77K) on carbons with very narrow microporosity
(a similar but ~ess pronounced problem is found when the degree
of ac ti vat ion is low, up to less than 5% burn-off (17)). The
problem found with carbons having a large degree of activation
will be presented later.

Since the problem with very narrow microporosity seems to be

due to the restricted diffusion of N2 at the low temperature of
77K, a possible solution would be to use an adsorptive such as
CO 2 at a much higher temperature. In this sense, some authors
(18,19) have recommended the use of CO 2 at around room
temperature as the ideal adsorptive to be usee in the characteri-
zation of activated carbons and coals. The relative pressure
range covered by CO 2 at 273 and 298K is much lower than for N2 at
77K in conventional adsorption systems since the saturation
pressures are 26142 torr and 48250 torr, respectively.

Table 2
3 -1
Micropore volume (cm.g ) for carbonized (C) and
activated (A) almond shells.(16,17)

V o V o
Carbon N2 (77K) CO 2 (273K)

C-2 0.15 0.29

C-4 0.11 0.20
A-1 0.17 0.19
A-2 0.18 0.19
A-3 0.19 0.21
 615

Table 2 shows some results for both carbonized materials and

activated carbons with very low degree of activation (16,17).
Since the term surface area has little physical meaning in
microporous solids (this point will be discussed below), Table 2
contains micropore volumes deduced from the application of the
Dubinin-Radushkevich equation to the adsorption data of both N2
at 77K and CO at 273K (the equations of adsorption are dealt
wi th in more §etail in another chapter). These results confirm
that the constrictions of the micropores in these materials are
similar to the dimensions of the N2 and CO 2 molecules, since the
latter, with molecular dimensions very similar to those of N2
(15), is adsorbed much faster and yields larger micropore volumes
because of the higher adsorption temperature.

These results can be extended to explain similar results

obtained by other authors for carbonaceous materials but that
have not been correctly interpreted. Lower amounts of N2 adsorbed
have been attributed to a molecular sieve effect of the
microporosi ty which adsorbs CO 2 but not N2 , because the latter
has a slightly larger minimun dlmension. The difference in amount
adsorbed is very probably due to activated diffusion effects and
not to a molecular sieve effect. If the adsorption of N2 at 77K
would have been carried out under real equilibrium conditions
allowing the adsorptive to reach the cavities, the differences in
micropore volume would not be as large.

In fact, when the adsorption of N2 at 77K is compared with

the adsorption of CO 2 at 273K for a series of activated carbons
covering a wide range of burn-off (4,5,16,17) as in Tables 2 and
3, the carbons can be grouped in three different types:

VO(N 2 ) i.. VO(C0 2 )

VO (N 2 ) rv VO (C0 2 )

VO (N 2 ) V O (C0 2 )

It should be understood that the data included in Tables 2

and 3 correspond to adsorption isotherms determined under normal
conditions of contact time between the carbon and the adsorptive
and, consequently, equilibrium is not always ensured for
adsorption of N2 at 77K, since in narrow micropores the nitrogen
molecules will not reach all of the cavities in the normal time
of measurement.

The case in w~iCh VO (N 2 ) < VD (C0 2 ). can be attributed to lack

of equilibrium In the N2 ads orp1;l on as discussed above.
616
Table 3
3 -1
Micropore volumes (cm g ) of activated carbons
prepared from almond shells (16,17)

Carbon Vo Vo
Burn-off N2 (77K) CO 2 (273K)
(%)

A- 2 5 0.17 0.19
A- 5 10 0.21 0.21
A-12 11 0.25 0.23
A-IS IS 0.34 0.2S
A-19 25 0.41 0.31
A-37 46 0.53 0.34
, A-40 60 0.66 0.34

Consequently, the other two cases are more interesting in

relation to activated carbons, which have a relatively wide
microporosity. For carbons with medium burn-off it is found that
VO(N 2 ) is similar to Vo(CO ). The D-R plots for these carbons
are 11near in a wide range o~ relative pressures; in the case of
N2 the upward deviation of the plots occurs at relatively high
relative pressures, whereas for CO 2 there is a deviation from
linearity at very low relative pressures. Since the temperatures
of adsorption are very different it is much better to use the
characteristic curves. Figure 9 shows the characteristic curves
for the adsorption of N2 (77K) and CO 2 (273K) on an activated
carbon prepared from olive stones with medium burn-off (19%). The
low pressure experimental points (below P/PO=O.Ol) for the
adsorption of N2 at 77K extrapolate towards the adsorption data
of CO 2 at 273K. ~his means that the micropore volume deduced from
the aasorption of both gases is the same. Also, the relative
pressure range of the experimental points of the extrapolated
linear portion is very similar for both adsorbates. Carbons with
this type of adsorption behaviour have a relatively narrow
microporosity. The data of Table 4, in which the percentages of
pore volume filled at P/P o =O.01 and pip 0=0.1 are included (7),
show that for carbons with up to medium Durn-off at least 77% of
the pore volume is filled at very low relative pressure
«P/PO=O.01) and almost 90% filled at P/Po=o.l. This is a clear
indication of a relatively narrow microporosity.

For carbons wi th low burn-off (but not too low to have

problems of restricted diffusion) the D-R plots for N2 (77K) and
CO 2 (273K) are linear over almost all the relative pressure range
ana the experimental points (even if no data for N2 at P/P o <
0.01 are available) fit the characteristic curve and, consequen-
 617

o 100 200
or--------r--------~------~--------~--------~

-1

> -2
t:
.-!

-'3

-4

Figure 9.- Characteristic curves for the adsorption of:

O,N2(77K);~, CO 2 (2?3K) and!::" CO 2 (298K)

tly, the micropore volume is the same for both gases. The
microporosity of these carbons is very narrow.

Things are different for carbons with large percentages of

burn-off; if low relative pressure data (below P/P O=O.Ol) are not
available for the adsorption of N at 77K -as is the case for
most of conventional adsorption sys~ems- the extrapolation of the
experimental points is not coincident (see Figure 10) with that
of the adsorption data for CO 2 at 273K. In this case, the
relative pressure range for the experimental points extrapolated
is not the same and the micropore volume calculated for N2 at 77K
is larger than for CO 2 at 273K. These carbons have a much wider
microporosity (see dlfferences of percentages of pore volume
filled at pip 0=0.01 and pip 0=0.1 in Table 4) and, consequently,
the adsorption of N~ at 77K is occurring in micropores of larger
dimensions than is aBsorption of CO 2 at 273K.

The problem is more complex if the adsorption of CO is

determined at 298K, as suggested by some authors (18,~9).
Althou§~ the amount of CO adsorbed at 298K, expresse§ ~£
mmol.g , is lower than at 2~3K, the micropore volume (in cm g
of liquid) is larger at the higher temperature. This latter
behaviour has sometimes been attributed to restricted diffusion
618

Table 4

Fractions of pore volume filled at different

relative pressures (7)

Carbon
nO.01/0.95 nO.1/0.95

D -9 0.85 0.94
D-20 0.80 0.90
D-34 0.77 0.88
D-52 0.72 0.87
D-70 0.64 0.84
D-80 0.60 0.81

o 20 40 60 80 100
o

-1

-2

-3

Figure 10.- Characteristic curves for adsorption of:., N2 (77K)

and. ,C02 (273KJ for a highly activated carbon
 619

at 273K, but this seems unlikely at temperatures near the

cri tical temperature of CO . When the characteristic curves for
the adsorption of N2 at 77~, and CO 2 at 273K and 298K in highly
activated carbons are plotted (see Flgure 9) it is found that the
plot for CO 2 at 298K lies above and almost paralell to that of N2
(77K) and CO 2 (273K), leading to a larger micropore volume.
However, if tne density of the adsorbate is corrected, as shown
in a previous paper (21), the micropore volume is very similar to
that for the same gas at 273K.
The above discussion indicates that both adsorption of N2
and CO 2 have to be used with caution and care must be taken to
ensure that the experimental conditions are adequate.

3.4. Adsorption of n-butane

Addi tional information may be obtained using other adsorp-

tives which can be adsorbed at around room temperature, but which
have a lower saturation pressure than CO 2 , An example of such
adsorptives is n-butane at 273K, although the minimum dimension
of the molecule (0.43 nm) is larger than for N2 (0.30 nm) and CO 2
(0.28 nm).

The shapes of the adsorption isotherms of n-butane at 273K

(Figure 11) on carbonized almond shells are qualitatively similar
to those of N2 at 77K (Figure 7). However, the isotherms show
clear differences between the carbons in contrast with the
results obtained from the adsorption of N2 ; only a small
percentage of the micropores are accessible to n-butane (Table 5)
and the differences between the samples indicate that the
constrictions of the micropores are not of the same dimensions in
all t~e carbonized samples, rendering the entry of the N2
molecules at 77K more-or-less difficult. Since only few micro-
pores are accessible to n-butane this indicates that although
microporous constrictions in the range 0.30- 0.43 nm are probable
in these materials, the largest contribution must be of width
just above 0.30 nm.

This molecular sieve effect is not as clear in the case of

activated carbons prepared from carbonized almond shells (17). As
Table 5 shows the micropore volumes calculated from the
adsorption of n-butane are similar to (although slightly lower
than) those obtained from N2 at 77K.

3.5. Preadsorption of n-nonane

Since, as described above, there are some uncertainties in

the characterization of the microporosity of activated carbons it
is advisable to use a number of al ternati ve methods of
620

30 ~-------------------------------------,

C-3
20 C-6
C-5
C-l

10 C-2

:::=__~~__~QQ---~Q~----~~
V- .
C-4
= ,... Q

o L-__ ~ __ ~ __ ~ ____L -__ ~ __- L__ ~ ____ ~

o 100 300 500 700

P{torr)

Figupe 11.- Adsopption isothePms of n-butane (2?3K) on capbonized

aZmond sheZZs, aftep (1?).

Table 5
3 -1
Micropore volume (cm g ) of carbonized and activated
almond shells (17)

Vo Vo
Burn-off
Carbon N2 (77K) n-C 4 H10 (273K)
( %)
C- 2 -- 0.15 0.04
C- 4 -- 0.11 0.03
C- 5 -- 0.26 0.07

A-28 9 0.27 0.23

A-29 14 0.31 0.28
A-30 19 0.32 0.29
A-31 27 0.44 0.41
A-32 32 0.50 0.46
 621

300
6 0673 K
493 K

.-t
I 200
bO
Q.,
Eo<
tI.l
(I')
et)
-->
100
0 0 0 C 393 K
0

298 K
0
0 0.2 0.6

P/PO

Figure 12.- N2 (77K) adsorption isotherms on a preadsorbed

activated carbon, after (14).
evaluation. A particularly direct method is preadsorption of an
adsorbate which covers the micropores of the carbon. By way of
illustration the results obtained using preadsorption of n-nonane
to evaluate the micropore volume of activated carbons prepared
from agricultural by-products (14,20) will be presented. Follo-
wing the experimental procedure described in section 3.1 the
adsorption of N2 was determined at 77K on the pradsorbed carbons.
In order to test whether the adsorption of n-nonane had produced
any modification of the porous structure of the carbons, the
samples with n-nonane were outgassed at 673K for 5 h in high
vacuum and the adsorption of N2 (77K) was again determined. Figure
12 shows the the N2 (77K) adsorption isotherms for one activated
carbon prepared from almond shells (14) with n-nonane preadsorbed
and later outgassed at increasing temperatures. The original
isotherm and that measured after outgassing at 673K are
coincident and this indicates that the microporous texture of the
carbon has not been altered by the preadsorption of n-nonane.

The N2 adsorption isotherm for the pre adsorbed carbon

ou tgassed at 298K (i. e., under the standard conditions of the
preadsorption technique) is much lower than that of the original
carbon, with a low amount adsorbed and a linear branch which is
almost paralell to the isotherm for the carbon without n-nonane
preadsorbed. On the other hand, there is an increase in the
amount adsorbed as the outgassing temperature increases and the
resulting N2 (77K) adsorption isotherms have aproximately paralell
linear branches and only differ in the more-or-less rounded knee
at P/P o< 0.1. This may be taken as an indication of the fact that
n-nonane covers the microporosi ty of the carbon and does not
affect its meso and macroporosity.
622

Table 6

Micropore volume (cm 3 .g- 1 ) obtained from different methods

for a series of activated carbons with different burn-off (7)

2 -1
Burn-off(%) VO(N 2 ) VO(C0 2 ) V' V S t(m g )
0 n ex
8 0.25 0.26 0.25 0.16 22
19 0.31 0.30 0.31 0.24 38
34 0.39 0.36 0.38 0.31 50
52 0.50 0.41 0.50 0.41 83
70 0.57 0.48 0.52 0.47 224
80 0.62 0.51 0.52 0.50 376

It is difficult to know whether the n-nonane covers all of

the microporosi ty in solids such as activated carbons, with a
wide micropore size distribution. It is possible, however, to
compare the results of n-nonane preadsorption with those obtained
using different experiment~l methods. Furthermore, the analysis
of the N2 adsorption isotherm on the pre adsorbed sample may be of
help. Taole 6 includes the micropore volumes, deduced by
different methods, for a series of activated carbons (7,20) with
increasing degree of activation in which the microporosity is
progresively developed. The BET external surface areas deduced
from the N2 (77K) adsorption isotherms on the preadsorbed carbons
are also included in Table 6.

The micropore volumes of Table 6 have been calculated as

follows:

- VO(N 2 ): by application of the D-R equation to the N2 (77K)

adsorption isotherms on the original actlvated
carbons.

- V0 (CO 2) : by application of the D-R equation to the CO

(273K) adsorption isotherms on the originaf
activated carbons.

by measuring the vertical separation of the linear

branches of the N2 (77K) adsorption isotherms
before and after the n-nonane preadsorption.

- Vn : by calculating the volume of n-nonane retained by

the activated cc:.~on after evacuation at 298K,
taking 0.717 g.cm as the density.
 623

The data of Table 6 show that there is a very good

agreement between the V; V
and o (N 2 ) values for carbons with up to
52% burn-off even though these values have been obtained using
quite different procedures. It seems then that n-nonane occupies
or covers all the micropores of the carbon when the microporosity
is relatively narrow -compare the VO(N 2 ) and VO (C0 2 ) values. This
is confirmed by the external surface area of the porosity not
covered by n-nonane which is low and slightly increases with
activation, as expected in essentially microporous carbons.

For carbons with higher percentages of burn-off (70 and 80%)

for which the microporosity is wider -now VO (N 2 ) » VO(C0 2 )- the
resul ts seems to indicate that the n-nonane is not retalned in
all micropores and some of the n-nonane is removed by evacuation
at 298K; this implies that V; is lower than V
O (N 2 ) and the
surface area deduced for the rest of the porosity is now rather
large. Since the V0 (N 2 ) values have been deduced from the D-R
plots in which low pressure data are not available, the V~ values
will be similar to the V0 (C0 2 ) values for these two carbons, as
discussed in section 3.3.

The comparison of V and V' values indicates that the former

is lower for carbons wi£h low gurn-off, but both values approach
each other for increasing activation. This trend is expected
since the microporosi ty of carbons with low burn-off is very
narrow (there are problems of kinetic diffusion in the carbon
wi th 8% burn-off) and very probably n-nonane does not fill all
the micropores, but is just blocking the entrance of some of
them. This would not affect the V~ value but would lead to a
value of V which is too low (20). As the microporosity widens by
increasingnactivation, the V value approaches the V;
value when
all the microporosity is acce~sible to n-nonane molecules.

In general terms, V~ is similar to V0 (CO ), although the

latter is slightly lower in some cases for t~e reasons given
above. This behaviour, plus the above considerations, indicate
that the n-nonane preadsorption technique leads to the determina-
tion of the volume of relatively narrow microporosity; this means
that the micropore volume V~ will correspond to all the
micropores in carbons with narrow and uniform microporosity,
whereas in carbons with a wide micropore size distribution the
n-nonane preadsorption technique will not determine the total
micropore volume, but only that of narrow micropores. In the
latter carbons the technique fails because the n-nonane is
removed from the larger micropores upon outgassing at 298K prior
to the adsorption of N2 at 77K (14,20).
624

3.6. Adsorption of Molecular Probes

Another excellent and direct method to evaluate the

microporosity of activated carbons is the use of molecular
probes, i. e., adsorpti ves with different molecular dimensions.
The results presented here are for a series of carbons prepared
by reaction of an activated carbon (from almond shells) with dry
air at 623K (22,23); these carbons cover a wide range of burn-off
and, consequently, a wide range of microporosi ty. As a typical
example, Figure 13 shows the adsorption isotherms (the desorption
data are not included for the sake of clarity) of n-butane (273K)
and benzene, cyclohexane, n-hexane, 2,2dimethylbutane and iso-
octane at 298K on the original activated carbon (22)
-1
The amount adsorbed (in mmol. g ) for the different
adsorpti ves decreases in the order: benzene> n-butane > cyclo-
hexane > n-hexane )0 2, 2-dimethylbutane > iso-octane and this
sequence was found for all the samples of the series (22) •
Benzene and n-butane with the smallest minimum dimensions (0.37
and 0.43 nm, respectively) are adsorbed to a greater extent and
the amount of benzene adsorbed is about twice that of iso-octane,
the largest molecule (0.59 nm). These differences in amount
adsorbed for the different hydrocarbons are a possible indication
of a molecular sieving effect by the activated carbon of Figure
13.

The possible molecular sieving could be analyzed from the

point of view of the micropore volumes. The application of the
D-R equation to the adsorption of hydrocarbons on these
air-reacted activated carbons (22) leads to the micropore volumes
listed in Table 7. Benzene, n-butane, n-hexane and iso-octane
yield similar values of Vo; cyclohexane and 2,2-dimethylbutane
Give similar values of VO ' although slightly lower than the
other hydrocarbons. This sequence does not correspond with that
seen i~l Figure 13, in which the amount adsorbed was expressed in
mmol.g . This different behaviour could be due either to the use
of a density for the adsorbate which is not correct or to the
different sizes and shapes of the hydrocarbons molecules.
However, the existence of a molecular sieving effect would be
doubtful in carbons with a large degree of ac ti vat ion such as
these (22). In any case, if this effect were really present in a
carbon, there would be differences in the amount adsorbed
independently of the units in which it is expressed. Thus, for an
activated carbon with a lower degree of burn-off (7), the results
of Table 8 show that there is a clear molecular sieving effect
for the large molecules.

The differences in the kinetics (7) and amount adsorbed by

the carbons may supply further information about the maximum and
 Table 7

Micropore volumes (cm 3 .g- l ) from adsorption of hydrocarbons; after (22).

carbon n-butane benzene cyclohexane 2,2-dimethylbutane n-hexane iso-octane

A- 0 0.42 0.40 0.35 0.36 0.39 0.40

A- 5 0.44 0.43 0.41 0.43
A-I0 0.47 0.46 0.43 0.43
A-20 0.49 0.45 I
0.49 0.49
A-33 0.51 0.46 I
0.50 0.48 0.51 0.55
!
A-42 0.46 0.50 0.45 0.45
A-50 0.28 0.30 0.27 0.27
A-60 0.26 0.26 0.23 0.22
A-72 0.16 0.18 0.11 0.10 0.17 0.15
'-------------- ----- - - - - - - - - - -- ------ L

C'
N
V1
626

~bt!nZt!nt!
• __ _ n-but~nt!

o 0.2 0.6

P/PO

Figure 13.- Hydrocarbon adsorption isotherms on an activated

carbon. Redrawn from (22).

minimum dimensions of the micropores. The disk-like molecules

such as benzene and cyclohexane are adsorbed on the carbons
included in Table 7 more easily than the spherical molecules such
as 2,2-dimethylbutane and iso-octane with minimum dimensions
smaller than the maximum dimension of benzene. If it is assumed
that the molecules enter the pores according to their minimum
dimensions (22), these facts point to the existence of sli t-
shaped pores with minimum dimensions ranging from 0.37 nm
(minimum dimension of benzene) to 0.59 nm (minimum dimension of
iso-octane) and a minimum depth of 0.68 nm (maximum dimension of
benzene) .

Besides this direct application of the adsorption of

molecules with different dimensions, the use of hydrocarbons
allows a complete study of the desorption process in order to
determine the volume of meso and macropores of the carbons and
their evolution as a function of the activation process. Figure
14 includes the benzene adsorption-desorption isotherms obtained
on some carbons prepared by reaction of an activated carbon with
air at 623K (23).
 627

Table 8

Equivalent micropore volume (different units) for a

slightly activated carbon (7)

Units N2 Benzene Cyclohexane iso-butane iso-octane

3. -1
cm g 0.26 0.24 0.14 0.18 0.06
-1
mmol.g 7.45 2.63 1.28 1.82 0.35
g/100 g 20.86 20.51 10.75 10.56 3.99

All isotherms of Figure 14 exhibit type H4 hysteresis loops,

characteristic of slit-shaped pores. The adsorption and desorp-
tion branches are almost paralell but, except in carbon A-O, the
hysteresis is mantained even at low relative pressures. The size
of the loop increases with gasification, especially for large
burn-off; the increase in the high pressure hysteresis loop (HPH)
is related not only to the development of the meso and
macroporosity of the carbon but also to the contribution of low
pressure hysteresis (LPH). The LPH cycles and their evolution
with gasification must be related to more-or-Iess elastic
distortions or deformations of the porous structure produced
during the adsorption process. Such deformations enable adsorp-
ti ve molecules to enter previously inaccessible micropores in
which they remain during the desorption process unless the out-
gassing temperature is raised (23). Such deformations seem to be
produced more easily in carbons with increasing burn-off, because
they have a less rigid structure.

When the same material was reacted in CO 2 at 1098K (instead

of air at 623K) to produce carbons with comparable burn-off, the
LPH was only present on carbons with low burn-off (7);
consequently, the existence of LPH seems to be also conditioned
by the gas used in the activation reaction.

As a typical example of the use of the adsorption-desorption

isotherms to calculate the macropore volume of a carbon, Table 9
includes values calculated from the adsorption of benzene,
cyclohexane and 2, 2-dimethylbutane on carbons prepared by
reaction in air at 623K of an activated carbon (23). These values
are in very good agreement with those obtained by mercury
porosimetry.
628

-~

I
en 5
....:0
••
....",
t:

Figure 14.- Adsorption-desorption of benzene (298K) on activated

carbons;O:A- 0 ,"V:A-l0;~:A-33;O:A-60;A:A-72. Open sym-
boLs (adsorption); cLosed symboLs (desorption).
Redrawn from (23).
Table 9
3. -1
Macropore volume (cm g ) for some activated carbons; after (23)

Carbon benzene Cyclohexane 2,2-dimethylbutane porosimetry

A- 0 0.11 0.07 0.12 0.10

A- 5 0.17 0.09 0.10 0.10
A-10 0.15 0.11 0.11 0.12
A-20 0.15 0.10 0.13 0.12
A-33 0.11 0.18 0.15 0.12
A-42 0.10 0.17 0.08 0.13
A-50 0.11 0.06 0.12 0.14
A-60 0.12 0.11 0.14 0.11
A-72 0.12 0.08 0.07 0.11
 629

3.7. Surface Area

Since activated carbons are essentially microporous the term

surface area does not have much physical meaning and it is much
better to refer to pore volume. However, since the concept of
surface area is still in widespread use,the term can be used but,
following Barrer (24), i t should be regarded as a monolayer
equi valent area. For this reason it will be used here as an
apparent surface area.

The apparent surface area can be deduced from the adsorption

isotherms by applying equations such as BET, Langmuir or DRK (15)
to determine the equivalent monolayer, n , from which the surface
area can be calculated if the area cove~ed by a molecule (A ) is
known. The Langmuir and BET equations are not strictly appli~able
to microporous solids, but they will be compared with the DRK
equation for activated carbons with a wide range of burn-off. As
a first approximation, in order to avoid uncertainties about the
best A value for the different adsorbates at the temperature of
adsorp~ion, the comparison will be based on the n values deduced
from the use of the different equations. m

Figures 15 and 16 include the plots showing the relationship

between the n values deduced from the application of the three
adsorption eqrlations to the isotherms of several adsorpti ves on
acti vated carbons prepared from a common precursor and covering
the 8-80% burn-off range (7). The plot for the n values deduced
from the Langmuir and BET equations is linear i~ almost all the
range covered and is given by n (BET)= 0.95 n (L). The plot for
the n values deduced from them BET and DRK "kquations is also
linearm and given by n (DRK)= 1.13 n (BET). However, the basic
point of Figures 15 afrd 16 is that ~ny of the three adsorption
equations can be used to follow the evolution of the apparent
surface area of a series of activated carbons. The quantitative
differences found between the different methods for a given
carbon are not large and the surface area values will, in
general, decrease in the order DRK > Langmuir> BET for an average
activated carbon.

At this point it is convenient to recall the differences

found between the adsorption of N2 (77K) and CO 2 at temperatures
near ambient (273 or 298K). As aiscussed in section 3.3 the
differences in micropore volume deduced from the adsorption of N2
and CO 2 are a function of the micropore size distribution. When
these values are converted to apparent surface areas there is the
additional problem of ascribing the correct A value to the CO 2
molecule, especially at 298K. In order to prJllsent more clearly
these differences, Table 10 includes the apparent surface areas
for a series of carbons with a wide range of activation.
 Q'\
W
20 20 , o
,ll'

/
1 ,/ . /,,1
I ,
, ,-"
,,:""
,,,,
'
, //G:I
15 15 ,, ' ,
;// ";,,tg
""
.......
I ... ,//a!
~
;-:'/ I
~ I
0
-.• 0 /
10 10 ,'"
.....
....•
... I~'/
CO ,/
-.
.....
••
ao:
Q
....... ,f
.....0: +1.- ... CZI ,Ill
Q '/ " .....0: III N2
gi~/ \!) n.BulaM
III N2
J.,/ I Ii) i. Bulan •
\!) n-Dulan.
5 .+ B.nun.
G)
~/ i -Butan.
.l e dm_But.
~ B.nz.n. .. iso-O.cl.
e dm-Bul.
CD iso-Oct.

0 1
o
/' 10 15
01L
/o 5 10 15
-1 -1
n(BET)( ••ol.g ) n (L) ( •• ol.g )

Figure 15.- ReLationship between Figure 16.- ReLationship between

the BET and DRK equations. the BET and Langmuir equations.
 631
Co.mparing surface areas determined by adso.rptio.n o.f N2 (77K)
and CO 2 (273K), the general behavio.ur is paralell to. that
described when co.mparing the micro.po.re vo.lumes (sectio.n 3.3) and
three gro.ups can be distinguished:

S(N2 ) < S(C0 2 ): carbo.nized materials, activated carbo.ns with

very lew burn-o.ff and co.als, all with very
narro.w micro.po.ro.sity.

S(N 2 )NS(C0 2 ): activated carbo.ns with lew to. medium burn-o.ff

in which the micro.po.ro.sity is relatively
narro.w.

S(N 2 ) >S(C0 2 ): highly activated carbo.ns, with a wide micro.-

po.re size distributio.n.

The explanatio.n o.f these differences are the same as these

given when discussing micro.po.re vo.lumes, since the o.nly· facto. 2
intro.duced is the A value ~r beth adso.rbates, i.e., 0.162 nm
fer N2. at 77K and ~ .187 nm fer CO at 273K. Ho.wever, if the
Co.mparlSo.n is made with adso.rptio.n o.f2C0 2 at 298K, using the mere
generally accepted A value o.f 0.253 nm fer this adso.rbate, it
is fo.und (see Table fb) that the resulting apparent surface areas
are, fer all carbo.ns, larger than these calculated fro.m beth the
adso.rptio.n o.f N2 at 77K and CO 2 at 273K. Since it is difficult to.
accept that there co.uld be any pro.blem o.f restricted diffusio.n in

Table 10

N2 and CO 2 apparent surface areas (m 2g -1 ) o.f

activated carbo.ns (7)

Burn-o.ff N2 (77K) CO 2 (273K) CO 2 (298K) CO 2 (298K)Co.rr.

8 722 687 850 691

19 883 795 939 764
34 1104 936 1073 873
52 1407 1068 1228 999
70 1603 1245 1387 1128
80 1737 1327 1422 1156
632
highly activated carbons, the large surface areas determined by
adsorption of CO 2 at 298K must be due to an incorrect A value.
As mentioned in section 3.3., the density of CO adsorbedmat 298K
was corrected to fit the characteristic curve for the adsorption
of N2 at 77K and CO 2 at 273K. When the corresponding correctio~
is applied to the A of CO 2 at 298K, a value of around 0.21 nm
results (21) and, ifTIthis value is used to calculate the apparent
surface areas of the carbons, it is found (see Table 10) that
they are closer to those calculated from the adsorption of CO 2 at
273K.

3.8 Adsorption from Solution

As deduced from all the above considerations, the characte-

rization of the microporosity of activated carbons is a difficult
task because of the existence of micropores of different
dimensions. The use of molecular probes, i.e., adsorptives with
different molecular dimensions, can help in this characterization
as shown in section 3.6. However, since the microporosity
includes pores of up to about 2.0 nm diameter, larger adsorptive
molecules should be used to complete the characterization. The
adsorption of solutes with different molecular sizes (from just
over the size of nitrogen to molecules of large dimensions) may
improve the understanding of the microporosity of activated
carbons.

The adsorption of solutes from solution is a relatively

simple experimental technique, but the interpretation of the
results is rather complex since the phenomenon is not as
well-understood as gas adsorption. There are some excellent
reviews on the subject (25,26).

The first point to consider when studying the adsorption

from solution is the kinetics of the process. Figure 17 shows the
adsorption kinetics for five solutes -p-nitropheno1 (PNP),
methylene blue (MB), Orange II (O-II), Crystal Violet (CV) and
Victoria Blue (VB)- upon two carbons with similar degrees of
activation but prepared from two different precursors (27). The
plots of Figure 17 show that the percentage of solute adsorbed in
the first stage of the adsorption process increases with
decreasing molecular size, clearly showing the role of solute
diffusion through the pore system of the carbon particles. The
plots, in turn, show that one carbon adsorbs more rapidly than
the other because of the larger development of the meso and
macroporosity (27).

The next stage in the process of adsorption from solution is

to determine (usually by spectrophotometric methods) the adsorp-
tion isotherms by plotting the amount adsorbed as a function of
 100 • A-II PNP
3
~:-. OJ-II
SO
.-4
! I
...... bO
0 .
..... .-4

'0
i 2
M8 011
-
f0 100 '0
GI
i of0
.... SO
~(II
§
j 0
r ....
§
ev VB
100'
j rOo CV

~ ••• VB
SO

- . - ' - ___ n _ _ _ ----'--- _ _ _ _ , I I I I I o 2 4 6

0 200 o 200
r r Time (h)
C.l03 (M)

Figure 17.- Kinetics of adsorp- Figure 18.- Adsorption isotherms for

tion from aqueous soLution on severaL soLutes on one activated ear-
two activated carbons; A-16,J-18; bon; after (27).
after (27). 0-
W
W
634

1500

1000

500

o 5 10
D(nm)
Figure 19.- Apparent surface area distpibution curves fop
capbons with incpeasing degpee of activation.
a, J-11j~, J-14jC, J-18ja, J-21. Redpawn fpom
(27).

the equilibrium concentration. Figure 18 shows, as a typical

example, the adsorption isotherms of the same five solutes from
aqueous solution on a given carbon (27). The amount adsorbed by
the carbon decreases with increasing size of the solute because
of the molecular sieve effect through the pore entrances and also
because the dyes are adsorbed as ionic micelles with an
aggregation number which increases in the order MB < O-II < CV < VB
(28), which implies an increase in their minimum dimensions.
Because of this, rather than comparing the amounts adsorbed by
the carbons (as determined from the plateau of the isotherm or
from use of the Langmuir equation) it is more convenient to
convert these to apparent surface areas to compare the results
obtained. Figure 19 includes the curves of apparent surface areas
determined with the five solutes (the N2 surface areas are also
 635
included for the sake of comparison) for several activated
carbons as a function of the minimum pore dimension accessible to
the adsorptive molecule. These plots give the surface area of
pores larger than a given value and show the general decrease in
accessible area with increasing molecular dimension; the decrease
is very pronounced for the smaller sizes levelling thereafter and
showing the highly microporous texture of these activated
carbons. Since the carbons of Figure 19 have been prepared from
the same precursor but have different degrees of activation, the
results indicate that the surface area increases upon activation,
and the development and widening of the microporosity is clearly
shown.

3.9. Mercury Porosimetry.

Although activated carbons are essentially microporous,

their industrial applications require an adequate development of
the meso and macroporosity to facilitate access of the adsorptive
to the interior of the carbon particle; the macropores act as
admission pores and the mesopores as diffusion pores toward the
micropores where most of the adsorption takes place.

The technique of mercury porosimetry consists essentially in

measuring the extent of mercury penetration into an evacuated
solid as a function of the applied pressure. Automatic porosi-
meters are available to cover the pore size range from 3. 7nm
(corresponding to a maximum pressure of around 4000 bar) to
15.000 nm (corresponding to atmospheric pressure).

Figure 20 shows some plots of cumulative pore volume as a

function of the pore dimensions for some carbons prepared from
almond shells and olive stones with comparable degrees of
activation. It is clearly seen that the meso and macroporosity of
carbons produced from both precursors are different. The first
point to be noted is that the total pore volume is much larger in
activated carbons prepared from olive stones for the same
burn-off reached during the activation process. Another diffe-
rence is the range of porosity being developed. In carbons
prepared from olive stones the development of macroporosi ty is
very marked and not much pore volume is opened upon increasing
activation for pore sizes below around 100 nm. In carbons
prepared from almond shells the development of porosity is more
uniform in all the range covered by the mercury porosimeter, with
very little opening of the macroporosi ty. This is one of the
basic reasons why the kinetics of adsorption from solution is
faster in carbons prepared from olive stones, as shown in section
3.8.

A similar comparison can be made with carbons prepared using

different activation procedures. Thus, the mercury porosimetry
 636

0.6

H-23

..... 0.4
"""I H-ll
tIC
C')
et)
--> 0.2
C-23

C- 11

oL----------L----------~--------~~--~~
1 10 100 1000

D(nm)
Figure 20.- Cumulative pore volume (mercury porosimetry)
of activated carbons from olive stones (H) and
almond shells (C).

0.6

-----
------- ..- ......
"""I tIC 0.4 ---- -
C')
et)
--
>
0.2

0.0 L------------Jl0-----------l~0-0--------~1~0~0~0==--~~

D(nm)

Figure 21. - Mercury porosimetry plots for carbons prepared

from olive stones (top) and almond shells (bottom).
(-) N2,.-C0 2 ;(---) direct activation; (------) air-C0 2 ;
after 14).
 637
results are very similar for carbons having comparable extents
of activation prepared by direct activation with CO 2 , by
conventional carbonization followed by acti vation with CO 2
and by the two-stage air-C02 activation (see Figure 21).
The small differences found do not seem to be significant
(4), except for the slightly larger development of the mesoporo-
sity in the two-stage air-C02 activation process.

When the catalytic effect of iron on the C-C02 reaction

is used in the preparation of activated carbon, as described
in section 2, the cumulative plots obtained from the use
of mercury porosimetry are very different to those of carbons
prepared by direct activation with CO2 . Thus, Figure 22 includes
the cumulative plots for a series of activated carbons prepared Qy
direct activation in CO 2 at 800°C for 8 h (A-O-8) or by

1.4

A-S-8

.-I

.
I tiC

A-0.1-8

0.2
A-0-8
o
o 2 3 4 5

log D(nm)

Figure 22.- Mercury porosimetry results for carbons prepared

by direct CO 2 activation (A-O-B) and iron catalyzed
C02 activation.
638

activation, in the same conditions, of the same precursor (almond

shells) which had been previously impregnated with solutions of
Fe(NO) containing 0.1, 1 and 5% iron (carbons A-0.1-S, A-1-S
and it-&-s,
respectively) • The plots show that the catalytic
gasification by iron is very effective in the development of meso
and macroporosity; this development increases with increasing
iron content of the almond shells.

It is also interesting to compare the results found for

activated carbons prepared by CO 2 and steam activation. Figure 23
shows some cumulative pore volume plots for some of the carbons
prepared from plum stones according to the experimental procedure
described in section 2. The activation with CO 2 develops the
macroporosity, especially in the 600-1500 nm pore size range; for
pore sizes below 600 nm there is very little development of
porosi ty and the values of these pores are more-or-Iess similar
for the different carbons.

In the case of steam activation the extent of burn-off is

not the only factor to affect the development of porosity (as
discussed in section 3.2) but also the activation temperature
(13). The porosity development is also very different to that of
CO 2 activation since, as Figure 23 shows, pore development occurs
ma~nly in the mesoporous size range, in agreement with the

54
\
0.6 \
\
\
\
.-I \
I

(I')
.
00
0.4
47
\
\
e
u
32
..... 41
> 14
0.2
16

0
0 2 4

log D(nm)

Figure 23.- Mercury porosimetry results for CO 2 (----) and

steam (- - -) activated carbons.
 639

results obtained from the adsorption of N2 at 77K.

3.10. Total Porosity

When the pore volume of the different pore size ranges of an

activated carbon hav.e been calculated by the methods described
above, their addition will constitute the total pore volume of
the carbon. The total pore volume can also be determined if the
helium and mercury densities are known, since it will be given by
V=_l_ 1
t d Hg d He
This is so because helium will be accessible at room temperature
to the smallest micropores, whereas mercury will not enter the
pore system of the carbon at atmospheric pressure. The only
problem found when calculating the total pore volume of
microporous carbons is that helium may be adsorbed on the
micropores even at room temperature, leading to erroneous density
values. In most cases the increase in the temperature of
measurement may be enough to avoid the problem of helium
adsorption.

0.8
H-O

0.4

.-I
I

(I')
.
tIC 0

e(J 0.8
......
> C-O

0.4

0
-1 0 2 3 4

log r(nm)

Figure 24.- Cumulative pore volume of two activated carbons:

H-O (fY'om olive stones); c-o (fY'om almond sheUs).
640

H-O

II V e-o
II log r 2

oL-_____~~~=L~~--===¥~~~~~
-1 o 3

log r (run)

Figure 25.- Pore size distribution of two activated carbons:

H-O (from 0 live stones); c-o (from a z,mond she Us) •

Figure 24 shows the cumulative pore volume plots for two

activated carbons prepared from two different precursors. The
micropore volume has been deduced from the adsorption of benzene
at 298K (by applying the D-R equation); the volume of the
mesoporosity has been calculated by applying the Kelvin equation
to the desorption branch of benzene at 298K and the macropore
volume has been determined by mercury porosimetry. The total pore
volume for both carbons comes very close (+ 10%) to that
determined by measuring the helium and mercury densities.

The pore size distributions corresponding to the two carbons

of Figure 24 have been plotted in Figure 25 and clearly show the
different development of porosity in the two activated carbons
prepared from olive stones and almond shells. The former has a
less developed microporosi ty but a more developed macroporosi ty
and a more uniformly opened mesoporosity, confirming its better
behaviour in adsorption from solution. The plots show the
trimodal pore size distributions typical of activated carbons;
when the degree of activation is larger the trimodal distribution
is more clearly observed (7).
 641

4. CONCLUSIONS

The analysis of both the preparation and characterization of

activated carbons shows that both aspects are very complex. In
respect to the preparation using agricultural by-products as the
starting material, it has been shown that the direct activation
in CO 2 yields carbons with similar properties to those prepared
in the more conventional two-stage carbonization followed by
activation in CO 2 , with the subsequent cost saving. The two-stage
air-CO activat~on produces carbons with larger yield and
adsorp~i ve properties. Both the iron ca talysed activation with
CO 2 and the steam activation lead to a larger development of the
meso and macroporosity but similar microporosity.

The analysis of the different adsorptives that can be used

to characterize activated carbons has shown that no single
adsorptive can provide a complete information on the porous
texture, especially at very low or very high degrees of
activation. The basic problem is the wide micropore size
d~s!fibution found in activated carbons with around or above 1000
mg of apparent surface area. The relative pressure range
covered by the adsorptive at the temperature of adsorption will
determine the size range of the microporosi ty measured. Conse-
quently, the use of several gases complemented with the
adsorption of molecular probes and n-nonane preadsorption can
provide complete information about the microporosi ty. Capillary
condensation and mercury porosimetry may safely be used to
complete the knowledge of the total porosity of an activated
carbon and obtain the pore size distribution.

Acknowledgements

Financial support from CAleYT (P.795/81) is gratefully

acknowledged.

REFERENCES

1. Sing, K.S.W.; Pure Appl. Chern., 54, 2201 (1982).

2. Iley, M., Marsh, H. and Rodriguez-Reinoso, F.; Carbon 11, 633
(1973) .
3. Linares-Solano, A., L6pez-Gonzalez, J.D., Molina-S~bio, M.
and Rodriguez-Reinoso, F.; J. Chern. Techn. Biotechnol., 30,
65 (1980).
4. Rodriguez-Reinoso, F., Linares-Solano, A., Molina-Sabio, M.
and L6pez-Gonzalez, J.D.; Ads. Sci. Techn., 1, 211 (1984).
5. L6pez-Gonzalez, J.D., Martinez-Vilchez, F. and Rodriguez-
Reinoso, F.; Carbon 18, 413 (1980).
642

6. Gonzalez-Vilchez, P., Linares-Solano, A., Lopez-Gonzalez,

J.D. and Rodriguez-Reinoso, F.; Carbon 17,441 (1979).
7. Garrido-Segovia, J.; Ph. D. Thesis. University of Alicante
(Spain) 1984.
8. Torregrosa-Macia, R.; Ph.D. Thesis. University of Alicante
(Spain) 1984.
9. Rodriguez-Reinoso, F., Martin-Martinez, J.M., Molina-Sabio,
M., Perez-Lledo, I., and Prado-Burguete, C.; Carbon, in
press.
10. Rodriguez-Reinoso, F.; Linares-Solano, A., Molina-Sabio, M.
and Perez-Lledo, 1.; XVII Biennial Conference on Carbon.
Kentucky (U.S.A.). 1985.
11. Walker, P.L. Jr., Shelef, M. and Anderson, R.A.; "Chemistry
and Physics of Carbon" (Walker, P. L., Jr., Ed.). Vol. 4.
Marcel Dekker. New York (1966).
12. McKee, D.W.; "Chemistry and Physics of Carbon" (Walker, P.L.
Jr., Ed.). Vol. 16, p.1. Marcel Dekker. New York (1981)
13. Rodriguez-Reinoso, F., Almela-Alarcon, M., Salinas-Martinez
de Lecea, C. and Sepulveda-Escribano, A.; XVII Biennial
Conference on Carbon, Kentucky (U.S.A.). 1985.
14. Linares-Solano, A., Lopez-Gonzalez, J.D., Martin-Martinez,
J.M. and Rodriguez-Reinoso, F.; Ads. Sci. Techn., 1,123
(1984) .
15. Gregg, S.J. and Sing. K.S.W.; "Adsorption, Surface Area and
Porosity". Academic Press. London (1982).
16. Rodriguez-Reinoso, F., Lopez-Gonzalez, J.D. and Berenguer,
C.; Carbon 20, 513 (1982).
17. Rodriguez-Reinoso, F., Lopez-Gonzalez, J. D. and Berenguer,
D.; Carbon 22, 13 (1984).
18. Marsh, J. and Wynne-Jones, W.F.K.; Carbon 1, 269 (1964).
19. Walker, P.L., Jr. and Shelef, M.; Carbon 5, 7 (1967).
20. Rodriguez-Reinoso, F., Martin-Martinez, J .M. Molina-Sabio,
M., Torregrosa-Macia, R. and Garrido-Segovia, J.; J. Colloid
Interf. Sci., in press.
21. Rodriguez-Reinoso, F., Linares-Solano, A., Martin-Martinez,
J.M. and Lopez-Gonzalez, J.D.; Carbon 22, 123 (1984).
22. Martin-Martinez, J.M., Linares-Solano, A., Rodriguez-Reinoso
F. and Lopez-Gonzalez, J.D.; Ads. Sci. Techn., 1, 195 (1984).
23. Linares-Solano, A., Rodriguez-Reinoso, F., Martinez-Martinez
J.M. and Lopez-Gonzalez, J.D.; Ads. Sci. Techn., in press.
24. Barrer, R.M.; "Surface Area Determination". p. 80; Butter-
worth, London (1970).
25. Mattson, J.S. and Mark, H.B., Jr.; "Activated Carbon". Marcel
Dekker. New York. (1971).
26. Puri, B.R.; "Chemistry and Physics of Carbon". (Walker, P.L.,
Jr., Ed.). Vol. 6, p. 191. Marcel Dekker. New York (1970).
27. Linares-Solano, A., Rodriguez-Reinoso, F., Molina-Sabio, M.
and Lopez-Gonzalez, J.D.; Ads. Sci. Techn., 1, 223 (1984).
28. Giles, C.H., D'Silva, A.P. and Trivedi, A.S.; "Surface Area
Determination". p. 317. Butterworths, London (1970).
 643

LIST OF PARTICIPANTS

S.S. Alves, Instituto Superior Tecnico, 1096 Lisboa, Portugal

A. Avidan, City College of New York, NY, 10031, USA
N. Ba~, Middle East Technical University, Ankara, Turkey
R.T.K. Baker, Exxon, Annandale, NJ, 08801, USA
S. Balci, Middle East Technical University, Ankara, Turkey
W.A.A. van Barneveld, Netherlands Energy Research Foundation, 1755
ZG Petten, The Netherlands
C.A. Bernardo, Universidade do Minho, 4719 Braga, Portugal
H.T.P. Bos, VEG-Gasinstituut, 7300 AC Apeldoorn, The Netherlands
D.G. Carson, CERCHAR, 60550 Verneuil-en-Halatte, France
M.M.B. Carvalho, Faculdade de Ciencias, 1000 Lisboa, Portugal
P.D. Causton, School of Materials Science, Bath BA2 7AY, UK
J. Coca, Facultad de Quimica, 33007 Oviedo, Spain
J.M. Diez-Tascon, Instituto Nacional del Carbon,33080 Oviedo,Spain
S. Din~er, Yildiz University, ~i~li, Istanbul, Turkey
J.L. Figueiredo, Faculdade de Engenharia, 4099 Porto, Portugal
G. Gut, Technisch-Chemisches Lab.,ETH-Zentrum,CH-8092 ZUrich,
Switzerland
P. Hall, School of Chemistry, Newcastle upon Tyne, NEI 7RU, UK
K.H. van Heek, Bergbau Forschung, 4300 Essen 13, F.R. Germany
K.J. HUttinger, Institut fUr Chemische Technik,7500 Karlsruhe, F.R.
Germany
F. Kapte~jn, Laboratorium voor Chemische Technologie, 1018 WV
Amsterdam, The Netherlands
A. Linares-Solano, Facultad de Ciencias, Alicante, Spain
L.S. Lobo, Universidade Nova de Lisboa, 2825 Monte da Caparica,
Portugal
A. Marchand, Centre de Recherches Paul Pascal,33405 Talence, France
H. Marsh, School of Chemistry, Newcastle upon Tyne NEI 7RU, UK
J.I. Martins, Faculdade de Engenharia, 4099 Porto, Portugal
J.G. McCarty, SRI International, Menlo Park, CA 94025, USA
O. Mohammad, Kuwait Institute for Scientific Research,Safat, Kuwait
J.A. Moulijn, Laboratorium voor Chemische Technologie, 1018 WV
Amsterdam, The Netherlands
H.J. MUhlen, Bergbau Forschung , 4300 Essen 13, F.R. Germany
Z.I. Onsan, Bogazi~i University, Istanbul, Turkey
J.J.M. 5rfao, Faculdade de Engenharia, 4099 Porto, Portugal
B. Ossewaarde, Delft University of Technology, 2600 GA Delft, The
Netherlands
644

M. Paolucci, Universita di Roma., 00185 Roma, Italia

F. Pepe, Universita di Roma, 00185 Italia
D. Pinheiro, Siderurgia Nacional, 2840 Seixal, Portugal
J.J. Pis, Instituto Nacional del Carbon, 33080 Oviedo, Spain
W. Prins, Twente University of Technology, 7500 AE Enschede, The
Netherlands
L. Puigjaner, Universidad Politecnica de Catalunha, E.T.S.I.I.B.,
08028 Barcelona, Spain
R. Quartermaine, Department of Metallurgy and Materials Science,
Cambridge CB2 3QZ, UK
M.A. Queiroz, Universidade do Minho, 4719 Braga, Portugal
S. Ragan, BP Research Center, Sunbury on Thames,TW16 7LN, UK
A.E. Rodrigues, Faculdade de Engenharia, 4099 Porto, Portugal
F. Rodriguez-Reinoso, Facultad de Ciencias, Alicante, Spain
A. Sepulveda-Escribano, Facultad de Ciencias, Alicante, Spain
I.F. Silva, Universidade Nova de Lisboa, 2825 Monte da Caparica,
Portugal
M.J.L. Sousa, Faculdade de Ciencias, 4000 Porto, Portugal
T. Takekawa, School of Chemistry, Newcastle upon Tyne NEl 7RU, UK
M.T. Tavares, Universidade do Minho, 4719 Braga, Portugal
K.M. Thomas, British Gas Corporation, London SW6 2AD, UK
D.L. Trimm, University of New South Wales, Kensington, NSW 2033
Australia
J. Trommel, Delft University of Technology, 2600 GA Delft, The
Netherlands
P.J. Tromp, Laboratorium voor Chemische Technologie, 101S-WV,
Amsterdam, The Netherlands
S. UlkU, Dokuz EylUl University, Izmir, Turkey
T. Wigmans, Norit, 3800 AC Amersfoort, The Netherlands
Y. YUrUm, Weizmann Institute of Science, Rehovot, Israel
 645

aliphatic, 130, 131, 132, 133, 134

Index aliphatic ring, carbon bridges, 503
aliphaticlpolyamantane model, 73
alkali metals, 181, 182, 197, 336
alkaline-earth metals, 181, 182
abrasion, 431 allotherm, 308
absorbtion, 106, 107 allothermal, 388, 390
AC,566 allothermic 307, 309
accessible internal area, 137 almond shells, 602, 610, 640
activated, 603 Am value, 632
activated carbon, 13, 15, 53, 494, 559, 560, amount adsorbed, 624
561,565,566,568,569,570,571,572, analysis, 424
573,574,575,578,579,581,582,584, anisotropic, 28, 34, 36, 37, 39, 45, 52, 101,
594,596, 597,598, 599,601,6~ 609, 107, 108, 115
615,621,624,627,629,631,632,635 anisotropic coke, 44
activated coals, 35 anisotropy, 42,107,108, 113, 115,117, 118,
activated diffusion, 159 120, 121, 126, 130, 131, SIS, 519, 520,
activation, 197,606,623 522
activation energies, 324, 336, 347 apparent equilibrium, 610
activation energy, 117,213,337,338,345 apparent surface area, 629, 629, 634
activation methods, 602, 612 application of coal gasification, 383, 407
activation rates, 607 applying a catalyst, 206
activation reaction, 627 aqueous solution, 608
activation time, 605 armchair configuration, 233
active carbons, 45, 52 aromatic, 130, 131, 132, 133, 134
active site concentration, 315, 316 aromatic compounds, 465
active sites, 213, 311, 317 aromatic crystals, 514
active species, 189, 191 aromaticlhydroaromatic model, 75
active surface area, 315 aromatic ring size, 80
activity, 213 ash analysis, 424
activity of carbon, 303 ash diffusion controls, 368
adsorbent-adsorbate interaction, 149, 160 ash fusion, 424, 432
adsorption, 138 attractive force, 138
adsorption behaviour, 614 autocatalytic character, 220
adsorption branches, 627 autotherm, 308
adsorption-desorption isotherms, 626 autothermal, 388
adsorption equations, 146,629 autothermic, 309
adsorption equilibrium, 159 autothermic coal, 389
adsorption from solution, 632
adsorption in microporous carbons, 157 band modulations, 100, 106
adsorption isotherms, 632 band width, 110
adsorption on the micropores, 602 barrer, ISO
adsorption properties, 601 benzene, 608, 624
adsorption space, 162 benzene derivatives, 463
adsorption temperature, 615 BET, 149, 629
adsorptive properties, 641 BET equation, 147
adsorptives, 610, 629, 641 BET external surface, 622
affmity coefficient, 164 bituminous coal, 468, 470
agglomeration, 213, 427, 441 blocking of active sites, 497, 504
agricultural by-products, 602, 605 bonding interactions, 82
air-CCh,614 boudouard, 186
air-CCh activation, 604, 637 boudouard reaction, 19
646
bridges, 80
bright field, 103, 104
briquettes, 445
brown coals, 485
Brunauer, 142
bum-off, 607, 617, 623, 627
bum-off time, 377
13C, 130, 131
C.V.D.,94
CAJE~,233, 234, 236,237, 240, 249,256,
260,263
caking, 36,40
caking and swelling, 424, 426
calcined petroleum coke, 214
calcium, 315
capillary condensation, 602, 641
carbide, 254, 255, 256, 259
carbide mechanism, 213
carbidic carbon, 259, 265
carbon, 181,294
carbon blacks, 52, 94, 96, 132
carbon deposited, 271, 275
carbon deposited on iron, 279
carbon deposited on nickel catalysts, 277
carbon deposition on metals, 270
carbon deposits, 270, 273, 274, 276, 283, 336
carbon deposits on nickel and iron catalyst,
270
carbon dioxide, 490, 502
carbon fibres, 52
carbon filament, 275, 285
carbon gasification, 294
carbon mocular sieves, 602
carbon monoxide, 214, 490, 502
carbon reactivity, 310
carbon reactivity, 355
carbon surface, 497
carbon whiskers or filaments, 270
carbon whiskers, 270
carbonaceous, 565, 571, 575, 583, 585, 592,
595, 596
carbonaceous feedstock, 583
carbonaceous material 181, 559, 560, 568,
569, 571, 572, 573, 575,578, 594
carbonization, 4, 12,27,33,35,36,40,44,
45, 47, 53, 94, 95, 97, 98 ,113, 114,
115, 117, 118, 119,123, 127, 128,
129, 426, 511, 512, 518,520, 521,
522,603
carbonization-activation, 605
carbonization, x-ray, 134
carbonized almond shells, 610, 620
carbonized materials, 631
carbons, 616, 641
carboxylic groups, 222, 502
carburisation, 528, 530
catalyst deactivation, 269, 273
catalyst regeneration, 543, 549
catalysts for gasification, 408
catalysts, 275, 276, 283, 294, 355
catalytic carbons, 270, 544
catalytic coal gasification, 10
catalytic coal gasifier, 11
catalytic coke, 6, 545, 547
catalytic effect of iron, 606, 637
catalytic gasification of carbon, 269
catalytic influence of the iron, 498
catalytic, surface and gas phase coke, 547
catatyzed carbon gasification, 12
catalyzed coal gasification process, 231
catalyzed gasification, 181,316
catalyzing the activation process, 209
cavaties, 615
cenospheres, 458
Ceylon graphites, 198
change in pore structure during gasification,
371
change in pore structure during the carboniza-
tion process, 68
channeling, 241, 244,245,247,251,252,
253, 254, 256, 258, 259, 261, 262
char/coke properties, 431
characteristic curves, 162, 616
characterisation, 424,426,435, 601, 632
characterisation of activated carbons, 641
characterisation of coal, 421
characterisation of porous carbon, 138
charcoals, 35
charge carriers, 122, 123
chars, 27, 28,33, 35,46,47, 52, 302, 33
chemical activation, 603
chemical evolution, 126, 133
chemical methods, 70
chemical shift anisotropy, 130, 131
chemical vapor deposition, 4, 34, 94
chemism, 486
chemisorption, 138
class-like carbons, 94
classificcation of coals, 456
co-operative adsorption, 143
C~, 608, (614)
C~ activation, 603
C~ adsorbed at 2981{, 617
 647

COz at 273K, 615 cross cmorific vmue, 424

COz gasification, 337 cross-link density, 86
com, 181, 294, 455, 631 cross polarization, 130, 131
com agglomeration, 427 crystru violet, 608, 632
com caracterisation, 424 cumulative plots, 637
com char gasification, 355 cumulative pore volume plots, 640
com chars, 303 cumulative pore volume, 635
com classification, 424 Curie, 129
com classification scheme, 435 Curie laws, 127
com extracts, 86 Curie-point gas chromatography, 479
com gasification processes, 403 Curie-point mass spectrometry, 472
com gasification, 293 , 300, 344, 403 Curie-point pyrolisis, 475
com models, 458 Curie-point technique, 458
com networks, 82 cyclohexane, 624
com properties, 396, 449
com properties in different gasification d(r) equation of London, 139
processes,396 ~ 120, 123, 127, 128
coruification, 35,512,520 D-R plots, 616
cobmt, 220 dooz, 99, 106, 134
coke deposition, 15, 547 Damkho1er equation, 362
coke deposition and removm, 548 dark field, 102, 103, 104
coke deposits, 270 data handeling; 350
coke formation and removal, 525 de Boer's classification, 144
coke formation, 17,533,546 deactivated nickel catalysts, 283
coke properties, 426 deactivation, 270,282
coke removm, 543, 546, 549 Debye-Scherrer (D-S), 98, 101, 102
cokes, 27, 28, 33, 35, 40, 41, 42, 43, 44, 45, decoking, 528
46,47,51,52,94, 101, 109, 110, 111, defects,3~4~43,45
112, 124, 132, 511, 512, 514, 543, 544, defmition of structure, 59
545, 549, 550, 554 deformations of the porous structure, 627
coking, 36, 535,546 degradation, 431
combined cycle, 439, 452 degree of activation, 615
combustion of coke deposits in a catruyst pellet, degree of graphitization, 95, 635
373 dehydrogenation, 71
composite materims, 109 delayed coke, 53, 512, 513, 514
composite time variable (0), 97 delayed coker, 511, 513, 514, 515, 516, 517,
concentration of active sites, 324, 337 521, 522
conditioning of raw gas, 395 delayed coking, 523, 536, 537, 538
conductivity, 107, 113, 114, 115, 122, 126 density, 329
constrictions of the micropores, 619 density for the adsorbate, 624
contact angle, 236 density of the adsorbate, 619
continuum diffusion, 364 depolarized, 108, 109
controlled atmosphere, 258 depolymerisation, 71
controlled atmosphere electron microscopy deposition of carbon on iron and nickel, 283
(CAEM), 233, 243, 263, 237, 240, 249, deposition temperature, 120
256, 260,263 deposits, 272, 273,276,281
convertibility, 200 deposits and metal, 273
covment cross~links, 82, 87 derivatisations, 71
CP,130 desorption branch, 144
CP-MAS, 131 desorption branches, 627
cracking, 526, 530, 531, 532, 533, 537,538 desorption measurement, 214
crackling core model, 369 desulfurization, 226
648
development, 612, 635
deviations, 153
devolatilization, 384, 426, 429
diamagnetism, 117, 118, 120, 126
diamonds, 94
differential method, 350
differential thermo analysis, 199
diffraction, 98, 102, 106
diffusion, 106
diffusion effects, 615
diffusion of C~ 60S
dilmometty,427,432
dilatometty parameters, 427
2,2-dUnethylbutane, 608, 624
dipole-dipole interaction, 139
direct activation in ~, 603
direct activation with ~ 637
disclinations, 40, 43
discotic, 36, 37, 52
discotic aromatic nematic liquid crystals, 514,
520
discotic crystals, 514, 514
disk-like molecules, 626
dislocations, 30, 31
disoteric nematic liquid crystals, 27
dispersion forces, 139
dispersion of the catalyst, 213
dispersion phenomena, 355
disproportionation, 193
disproportionation reaction, 185
dissociative adsorption, 210
dissolution/precipitation mechanism, 213
Dollimore, 149
DRK., 629
dry ash Lurgi gasifier, 443
dubinin, 149, 164
Dubinin-Astakhov, 167
Dubinin-Radushkevich equation, 164
Dubinin-Radushkevich-Kaganer, 150
dusty gas model, 366
dystectic,201
earth a1k:aline metals, 316, 336
edge, 33
edge dislocations, 32
edge recession, 240, 250, 251, 252, 253, 254,
256,258,259,263,264
effects of heat on coal structure, 60
eleclron diffraction, 101
eleclron microscope, 98, 102, 103, 106
electron probe microanalysis, 199
eleclron spin resonance, 126, 134
eleclron lransfer mechanism, 213
eleclrons,98, 102, 106
n electrons, 107, 113, 117, 118, 120, 122,
126,130
electrostatic forces, 139
elemental composition, 469, 470
elements, 62
enerxY-dipole and quadrupole moments, 141
enerxY-dispersian forces, 141
energy-polar nature of adsorbent, 141
energy-repulsion forces, 141
enhance the enerxY of interaction
enhancement of potential energy, 161
enhancement of the adsorption potential, 158
enthalphies of immersion, 170
enthalpy- and entropy changes, 345
entrained bed, 389, 405
entrained flow, 8, 10,425
entrained phase, 387
equilibrium, 615
equilibrium compositions, 303, 304, 305, 306,
307, 308, 309
equilibrium concentration, 634
equilibrium conditions, 615
equilibrium constants, 299, 302, 304, 322,
333,335
equimolar, 221
equipotential surfaces, 162
equivalent, 221
equivalent free energy, 153
ESR, 126, 127, 134
ethane, 503
ether bridges, 215, 504
ethyl radicals, 503
eutectic, 201
evidence for macromolecular nature, 82
exinite, 60
experimental aspects, 349
extent of activation, 602
extemal mass lransfer limitations, 354
external mass lransfer, 296, 300
external surface area of microporous carbons,
173
extraction and solubilisation, 70
fast Fourier transformation, 215
Fe (N03)3 solutions, 606
feedstock, 582
fibre, 94, 101, 108, 112
filamentary carbon, 271, 272, 273, 274, 275,
282, 283
filaments, (whisker carbon), 276, 285
 649

film diffusion controls, 368 117, 118, 119, 120, 121, 122, 123, 125,
films,94,96, lOS 126, 130
fme pulverized fuel, 6 graphite single crystal, 116
fines, 426, 445 graphitic layers, 96, 99, 103, 107, lOS, 113,
fixed bed, 8, 10, 389, 425, 441 115, 118, 127
fixed bed flow reactor, 214 graphitic stIUctures, 95, 97
fixed bed gasifier, 11,441 graphitiZability, 41
flexi-coking, 538, 539 graphitizab1e, 27, 28, 34, 35, 36, 38, 46, 47,
flexicokinglfluid , 536 48, 50, 51, 52, 53
Flory theory, 83 graphitizab1e carbons, 33
fluid bed, 8 graphitization, 34, 36, 47, 48, 50, 51, 95, 96,
fluid bed gasifier, 10 97, 99, 100, 106, 110, 112, 113, 114,
fluid coking, 538, 539 115, 116, 118, 119, 120, 123, 124,
fluidity, 427, 429 126, 127, 128, 129, 134, 511
fluidized bed, 387, 389, 405, 416, 425 graphitized, 28
form coke, 12 graphitizing, 49
formation of the coke, 549 graphitizing carbons, 94
free electrons, 113, 127, 129 gravimetric system, 608
free radicals, 127, 129 growth, 235
frequencies, 109, 110, 126, 132, 133 Gurvitsh rule, 158
fruit stones, 602
functional groups, 80, 464 Hrgasification, 342
fundamental aspects of activated carbon, 579 H2O-gasification, 341
fundamentals, 559 hall coefficient, 97, 121, 122, 123, 125
hall effect, 121, 126, 130, 134
galvomagnetic effects, 121 hard,34
gas fase tars and carbon, 545 hard carbons, 94, 96
gas phase carbon, S44 heat, 294
gas phase coke, 546, 547 heat effects, 306
gas purification, 392 heat treatment in air, 603
gasification, 272, 293, 295,304, 307, 308, heat-1reatment temperatures (H.T.T.), 94, 100,
559,561,563, S64,565,566 103, lOS, 117, 124, 127, 129, 133
gasification mechanism, 317 heat-treatment time, 97
gasification of carbon deposited on nickel and heat-treatments, 94, 134
iron, 272 heavy ends, 536
gasification of coal, 302, 306 helium, 639
gasification of coal with the use of heat from helium adsorption, 639
nuclear reactors, 415 Heny's law
gasification on catalysts, 278 hetero cyclic aromatics, 466
gasification plant, 391 heteroatoms, 95, 96, 115, 126, 464
gasification process, 425 heterogeneity, 60, 62
gasification rates, 325, 330, 334 heterogeneous, 59, 422
gasification reaction rate, 314 heterogeneous water gas reaction, 384
gasification reactions, 384 high heating rates, 430
gasifier modelling, 450 high pressure hysteresis, 144
gasifier simulations in the laboratory, 435 high pressure hysteresis loop, 627
gasifier, 7, 8, 441 high-temperature nuclear reactors, 414, 417
general adsorbents 602 high-volatible bituminous coal model, 461
generation of amonia, 407 highly activated carbons, 631
Gieseler plastometer, 429 hkbands,l00
graphenes, 96, 100, 103, 112, 120, 123, 129 hk diffraction "bands", 98
graphite, 94, 96, 98, 99, 103, 109, 110, 115, holes, 122, 123, 130
650

homogeneous water gas reaction, 384

Humboldt process, 413 5% iron solution, 606
hydro aromatics, 80 iron sulfate, 207, 213
hydrocarbons, 626 iron sulfide, 225
hydrocracking, 535 iso-butane.. 608
hydrogasification, 485, 502 i~ctane, 608, 624
hydrogen, 214 isothermal gasification, 498
hydrogen gasification, 330 isothermal oscillations, 220
hydrogen shuttling, 521, 522 isotope labelling, 322
hydrogenation, 440 isotopes, 326
hydropyrolisis, 485 isotropic, 27, 36, 37, 39, 42, 44, 45, 47, 52
hydrolysis, 199 isotropy, 53
hygas process, 415
hysteresis, 144 K-catalysed gasification, 317
hysteresis loop, 144 KZC03,l84,472,479,480
kinetic and diffusion regimes, 363
ideal adsorptive, 614 kinetic model, 322, 324, 330, 331, 335, 343,
implications,594, 595 347
improved activation, 200 kinetic modeling, 319
in-situ anion exchange, 202 kinetic models, 293, 330,337, 355
in-situ infrared spectroscopy, 183 kinetic scheme, 193
inactivation, 197 kinetics, 214, 294, 300, 355, 430, 486, 624
inertinite, 60 knee, 612
influence of moisture, 503 Knudsen flow, 364
infra-red,l06, 113, 126, 132, 133, 134 Koppers-Totzek, 390, 405
inhibiting effect, 223
inhibiting effect of CO, 301 La, 100, 106, 110
inhibiting effect of water, 496 laboratory-scale reactors, 349
inhibiting influence, 321 Langmuir, 629
inhibition, 193, 338 Langmuir equotation, 146
inhibition by CO, 350 laser, 109
inhibiton of water, 504 lattice fringes, 103, 104, 105
ink: bottles pores, 144 Le,99,106
integral method, 352 lean gas, 386
intercalation, 109 lignite char, 469
intercalation compounds, 185, 189, 191 lignites, 485
internal mass transfer, 300 limitations, 149
internal surface, 602 line width, UO, U2, 127
interstitial, 101 linear branch, 612
interstitial defects, 30 liquid crystals, 28, 36, 43, 519
interstitial loops, 31, 32 localized, 127, 129
intrinsic activity, 197 low buro-·<>ff, 616
ion exchange, 214 low cost catalyst raw materials of mineral
IR analyzers, 215 origen, 197
iron, 207, 213, 269, 272, 275, 279, 280, 315, low pressure, 616
638 low pressure hysteresis, 145, 627
iron and nickel, 281, 282 low temperature activity, 197
iron catalyzed COlo (614) lowered melting points, 200
iron foil, 274 Lurgi, 389, 405
iron metals, 213 Lurgi-Ruhr 100,410
iron oxide, 207
iron oxygen surface complexes, 217 maceral analysis, 424
 651

~r.Us, 62,108,423 microporosity, 622, 641

macromolecular, 422 microporous, 609
macromolecular structure, 82, 435 microporous carbons, 143, 623
macropore volume, 627 microporous solids, 629
macropores, 601, 626 microporous structure, 601
macroporosity,635 microprobe, 109, 134
macroporous carbons, 143 microscope, 107, 108
macroscopic, 59 microscopic level, 62
magic angle spinning, 130, 131 microstructure, 222
magnetic, 113, 121, 122, 123, 130, 131 mineral 8ddition, 433
magnetic field, 126 mineral content, 432
magnetic, Hall effect, 113 mineral matter, 214, 315, 433
magnetic resonance, 126 minerals, 62
magnetoresistance, 113, 121, 122, 123, 124, minimum dimension, 619, 624
126 mobile phase, 75
MAS, 130 mobility, 236, 237, 238, 252
mass transfer limitations, 294, 350, 354 model, 461
maximum dimension, 624 model compounds, 462
McEnaney, 150 model structures for coal, 73
mechanisms, 187, 188, 189, 190, 214 modelling of gasifiers, 373
mechanisms of gas transport through porous modification, 603
media, 364 moisture of coal, 493
medium, 38 molecular dimensions, 609, 635
medium burn-off molecular probe technique, 160, 173
melting, 199 molecular probes, 624
mercury densities, 639 molecular sieve carbon, 15
mercury porosimetry, 608, 609, 627,635,640 molecular sieve effect, 504, 615
mesophase, 4, 27, 36, 37, 38, 39, 40, 42, 43, molecular sieve model, 73
44,47,51,94,97, SIS, 517, 518, 519, molecular sieves, 15
520,521,522 molecular sieving effect, 624
mesopore volume, 612 molecular structure, 68, 487
mesopores, 143, 153, 601, 602, 626 molecular weight distribution, 86
metal, 273 molten bath, 425
metal films, 235 monolayer, 147
metal surfaces, 281 monolayer capacity, 148
metallic surfaces, 269, 284 monoxide, 498
metallurgical-cokes, 37, 53, 513, 519, 520, monoxide methanation, 197
522 moving bed, 387, 405
methanation, 193, 384
methane, 503 n-butane, 608, 619, 624
methane formation, 226 n-hexane, 624
methane monoxide, 490 n-nonane, 621
methane yield, 494 n-nonane preadsorption, 609
method of Lec1oux, 155 N2,608
a-method of Sing, 154 N2 at 77K. 615
methods,l99 narrow constrictions, 610
methyl radicals, 503 narrow microporosity, 616
methylene blue, 608, 632 needle-coke, 37, 40, 43, 44, 51, 53, 513, 514,
micropore size, 641 515,516,517,518,521
micropore size distribution, 609, 622 nematic, 36
micropore volume, 612, 615, 617, 624 nematic crystals, 514
micropores, 157,601,602, 612,632 nematic liquid crystals, 37, 39, 52, SIS, 522
652

net heat of adsorbtion, 148 paramagnetic susceptibility, 127

nickel, 220, 269, 270, 272, 273, 275, 276, 285 parameter estimation, 350
-p-nitrophenol, 632 partial oxidation, 385
NMR, 126, 130, 131 particle size, 430
non-graphitic, 52 passages, 602
monoxide, 498 pauli, 129
non-graphitizable, 27, 28, 34, 35, 36, 38, 46, pauli para-magnetism, 127, 129
48, SO,53 Petersen model, 371
non-graphitizable carbons, 33 petrographic data, 424
non-graphitizing, 49 petroleum coke, 512
non-graphitizing carbons, 94, 96 phase inversion, 516
non-porous reference material, 157 phenol, 462
non-porous surface, 143 phenolic hydroxyl groups, 215
nonslagging, 8 phooons, 109, 113
nonslagging fixed bed gasifier, 9 physical adsotption, 138
nonslagging mode, 7 physical adsotption isotherms, 142
notional, 48 physical methods, 70
notionalaystallire,45,46,47,48,49 physical structure, 59
NOx reduction, (193) physicochemical constraints, 347
nozzles, 17, 18 pilot plants in the F.R. of Germany, 409, 409
nuclear magnetic resonance (NMR), 73, 130, pitch,433
133 pitch addition, 432
nucleation, 235 pitting, 240, 241, 253, 261
number of active sites, 299, 324, 329, 336, plasma, 94
347 plastometry, 427, 432
number of sites, 294 plum stones, 614
point defects, 31, 32, 45
olive stones, 602, 610, 635,640 poisoning, 317, 497
optical, 106, 108, 112, 134 poisoning effect, 222
optical microscopy, 222 poisoning phenomena, 355
optical rexture, SIS, 516, 517, 519, 520, 521, poisoning, sintering and dispersian
522 phenomena, 355
optimization of the coal gasification process, Polanyi, 162
425 Polanyi Porential theory, 162
orange II, 608, 632 polarization, 107, 108, 109
organic, 214 polarization forces, 139
outgassing remperature, 621 pore diffusion, 296
overlap of the force fields, 161 pore diffusion limitations, 354
overshoot, 187, 329 pore dimensions, 610
oxidation, 71 pore netwoIic type, 366
oxidic potassium species, 191 pore size distributions, 138, 640
oxygen affinity, 210 pore sizes, 601, 609, 635
oxygen containing groups, 502 pore structure, 423
oxygen exchange, 325 pores, 66
oxygen exchange mechanism, 319 porosity, 68
oxygen exchange rates, 339, 340 porous texture, 137
oxygen exchange reaction, 324 potassium carbonate, 197, 199
oxygen transfer mechanism, 210, 213 potassium halides, 199
potassium hydroxide, 197, 199
p-nitrophenol, 608 potassium nitrate, 199
pan, 17, 18 potassium salts, 197
paramagnetic, 126 potassium sulfate, 199

potassium sulfide, 206, 207
potential energy of adsorption, 139
power generating plants, 6
practical implications, 559
pre adsorption method, 172
pre adsorption of an adsorbate, 621
pre adsorption of n-nonane, 619
precursor, 60S, 607
preferred orientation, 95, 101, 103, 106, 115
pregraphitic carbons,94, 96, 106, 109, 115,
124, 133
pregraphitization, 95, 113, 123, 126, 127,
129,134
preoxidation on the thermoplastic properties,
433
preparation, 601
preparation of activated carbons, 641
pressure, 96, 97, 114,214
pressure gradients, 362
pretreatment procedures, 432
primary energy SOUICeS, 383
problems, 214
problems associated, 157
process variables, 426, 432
product distribution, 467, 468
profiles, 99, 100, 101, 106
properties, 425
proton decoupling, 130
pulverized fuel, 8
PVC-chars, 214
pyrocarbons, 108, 112
pyrolysis, 384, 430, 455, 459, 460, 462, 471,
485, 502
pyrolysis mass spectra, 474, 475, 478
pyrolysis of model compounds, 458
pyrolysis reactions, 475
pyrolytic carbons, 53, 94, 96, 100, 109, 120,
125
pyrolytic reactions, 480
raceway, 443
radial distribtion, 101
radicals, 494
raman, 106, 109, 110, 111, 112, 113, 134
random pore model, 313, 318
rate, 604
rate constants, 294, 322, 324, 325, 329, 339
rate-controlled process, 610
rate determining factor, 605
rate equation, 293, 342
rate expression, 297, 319, 322, 337, 343, 345,
351
653
rate limiting steps, 214
rate of diffusion, 601
raw materials, 198, 602
reaction controls, 368
reaction-diffusion problems, 361
reaction mechanism, 319
reaction rate, 301
reaction with mineral matter, 317
reactive char, 503
reactivity, 294, 315, 317, 324, 425, 432, 443
reactor types, 377, 388
reactor types with temperature and conversion
profiles, 377
recession of edges, 261
reduction, 71
reduction of nitrogen oxides, 186
reflectance, 106, 107, 108, 113
ool-reflexions, 98, 99, 101
reforming, 523, 533, 534, 536
refraction, 107
regenerated, 554
regeneration, IS, 16, 17, 546, 550, 554
regeneration of catalysts, 259, 547
regeneration of coked catalysts, 549
regeneration of coked crystals, 550
relative differential molar work of adsorbtion,
164
relatively narrow microporosity, 623
relaxation time, 127, 130
removal of coke, 548
repulvise forces, 139
resistivity, 114, 115, 116, 117, 123
resolidilication,458
restricted diffusion, 616, 631
rheological, 427
Saarberg-Otto development, 412
scanning electron micrographs, 455
scanning electron microscopy, 199
screw dislocations, 32, 33
selected area (electron) diffraction (S.A.D.),
101,102
selectivity, 197
semi-conductor, 115
semi-coke, 36, 40
shape of the isotherm, 609
shell-progressive model, 16
shift reaction, 197, 384
shot-cokes, 37, 40, 53, 513, 514, 515, 516,
517
shrinking core model, 368
silica spring balances, 608
654

similarity coefficient, 164 supermicropores, 162,609

single crysta~ 98, 115, 116, 117, 118, 122 surface, 106, 108, 112, 132, 133
single microcrystal, 101 surface area, 66, 615, 629, 631
single particle models for fluid-solid reactions, surface carbon, 544, 545
361 surface coke, 547
sintering, 317 surface diffusion, 364
sintering phenomena, 355 surface homogeneity, 153
site density, 325, 327, 333, 338 surface saturation, 317
size exclusion chromatography, 73 susceptibility, 117, 118, 119, 127, 134
slag, 432 swelling, 458
slagging,9 swelling index, 427
slagging gasifiers, 411,443 swelling phenomena, 460
slagging mode, 7, 8 synfuel,6
slit-shaped, 153 synthesis, 407
slit-shaped pores, 145, 609, 626, 627 synthesis gas, 8, 10, 386, 440, 452
slot coking oven, 511, 518 synthesis processes, 394
SNG,452 synthetic graphite powder, 214
soft, 34
soft carbons, 94, 116 t-method of Lippens and de Boer, 152
softening, 458 t-plots, 153
softening phenomena, 460 T.E.P.,l25
solute diffusion, 632 Tammann point, 238
solution adsorption, 608 Tammann, temperature, 237
solvent induce swelling, 82 tar yield, 430
spectrophotometric, 608 tar,432,433,448,546
spherical molecules, 626 Td, (125)
spill-over mechnism, 210 technical aspects, 486
spin, 127 technological aspects and problems, 214
sponge coke, 53, 514, 515, 517 temperature invariance, 163
sponge (regular) coke, 513 temperature of deposition (fD), 94
stacking faults, 30, 31 temperature profile, 443
standard isotherms, 152, 157 temperature-programmed desorption,344, 345
steady-state conditions, 320 temperature programmed heating, 214
steam activation, 603, 607, 614, 638 temperature programming, 343
steam cracking, 526, 530, 537 Texaco process, 412
steam forming methane, 414 textures, 106, 109, 113
step-respons experiments, 186 the grainy pellet mode~ 369
Stephan-Maxwell relationships, 365 the porous structure of coal, 60
strength, 431 the specific surface area of solids, 150
structural characterisation, 62 the Stoeckli equation, 170
structural defects, 33 theory of volume Filling micropores, 164
structural units, 95, 96, 99, 101, 103, 105, thermal expansion coefficient, 171
107,118 thermal stability, 463, 464, 465, 466
structure, 422, 423 thermodynamics, 294
structure model of Rheinische Braunkohle, thermoelectric power,113, 124, 125, 126
488 thermogravimetric analysis, 199,214
structure of lignin, 461 thermoplastic behavior, 455
structured models, 423 thermoplastic properties, 433
subbituminous coal (LEOPOLD), 198, 214 thickness of the layer increases, 152
substitute natural gas, 386 Thiele modulus, 362
substituted phenols, 464 three-dimensional, 96,100, 105, 112, 113, 123,
sulfur,213 124
 655

three-dimensional graphitization, 126 water-gas-shift-reaction, 186,302

total pore volume, 638 water vapour gasification, 213, 496
total porosity, 639, 641 water vapour mole fraction, 219
tranition, 114 Weibull dislribution, 169
transient experiments, 338 wettability, 199
transient methods, 328 weUiQg, 235,236,237, 244, 245,251, 256,
transient techniques, 325 258,263
transition, 115 widening, 635
transition metals, 316, 336 wider micropores, 609
Irapped molecules, 471 width of the cons1ructions, 612
turbostratic, 34, 96 Winkler, 390, 405
tuyeres, 443
twinning,233 x-ray, 98, 99, 101, 106
tw<Hlimensional, 96, 100, 105, 113, 120, 123, x-ray diffraction, 102, 106,422
124,125 x-ray diffraction analysis, 62
tw<Hlimensional graphitization, 126 x-ray photoeleclron spectroscopy, 199
two-dimensional ordering, 112
type a, 609 yield, 603, 604
type b, 609 yield properties, 641
type c, 609 Young's equation, 235
type H4 hysteresis loops, 627
type I isotherms, 146 zig-zag, 258
type 1,609 zig-zag configuration, 233
zones, 425
ullramicropores, 162
uncatalysed gasification, 312
uncata1yzed reaction, 606
uniform conversion model, 369
uniform gasification rate, 604
Yo. 622
Vrl..CQV,622
u
Vo(N ,622
Vn,622
vacancies, 30, 32, 45, 51
vacancy loops, 31
vacuum impregnation technique, 214
vapour, 608
very weak adsorbate-adsorbent interactions,
143
vibration, 109, 110
victoria blue, 608, 632
visbreaking, 536, 537,538
viscosity, 432
viscous flow, 364
visible, 106, 109, 132
vilrinite, 60, lOS, 113
volatile products (tar), 490
volatiles, 503
volatilization of the coal, 503
volume filling mechanism, 158

Wakao and Smith model, 366

water, 497