Você está na página 1de 15

Applied Catalysis A: General 201 (2000) 225239

Catalytic steam reforming of bio-oils for the production


of hydrogen: effects of catalyst composition
Luca Garcia
1
, Richard French, Stefan Czernik

, Esteban Chornet
National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401, USA
Received 8 October 1999; received in revised form 5 January 2000; accepted 5 January 2000
Abstract
Catalytic steam reforming of condensable vapors (i.e. bio-oils) derived from pyrolysis of biomass is a technically viable
process for hydrogen production. In this study the aqueous fraction of bio-oil, generated from fast pyrolysis, was catalyti-
cally steam reformed at 825 and 875

C, high space velocity (up to 126,000 h


1
) and low residence time (26 ms). Using a
xed-bed micro-reactor interfaced with a molecular beam mass spectrometer (MBMS), a variety of research and commercial
nickel-based catalysts were tested. The catalysts were prepared by impregnation of an -Al
2
O
3
support with nickel and
additives. Since the main constraint in reforming bio-oils is catalyst deactivation caused by carbon deposition, two strategies
were applied to improve the performance of the catalysts. The rst approach aimed at enhancing steam adsorption to facili-
tate the partial oxidation, i.e. gasication of coke precursors. The second one attempted to slow down the surface reactions
leading to the formation of the coke precursors due to cracking, deoxygenation, and dehydration of adsorbed intermediates.
Magnesium and lanthanum were used as support modiers to enhance steam adsorption while cobalt and chromium additives
were applied to reduce coke formation reactions. The cobalt-promoted nickel and chromium-promoted nickel supported on
MgO-La
2
O
3
--Al
2
O
3
catalysts showed the best results in the laboratory tests. At the reaction conditions progressive catalyst
deactivation was observed leading to a decrease in the yields of hydrogen and carbon dioxide and an increase in carbon
monoxide. The loss of activity also resulted in the formation of higher amounts of methane, benzene and other aromatic
compounds. Commercial catalysts that were developed for steam reforming of natural gas and crude oil fractions proved to
be more efcient for hydrogen production from bio-oil than most of the research catalysts mainly due to the higher watergas
shift activity. 2000 Elsevier Science B.V. All rights reserved.
Keywords: Nickel catalyst; Hydrogen production; Steam reforming; Biomass pyrolysis; Molecular beam mass spectrometry; Cobalt;
Chromium; Magnesium; Lanthanum; Catalyst deactivation

Corresponding author. Tel.: +1-303-384-7703;


fax: +1-303-384-6363.
E-mail addresses: luciag@posta.unizar.es (L. Garcia),
stefan czernik@nrel.gov (S. Czernik)
1
Present address: Chemical and Environmental Engineering
Department, University of Zaragoza, 50009 Zaragoza, Spain.
1. Introduction
Hydrogen is an important raw material for the
chemical industry and is a clean fuel that can be used
in fuel cells and internal combustion engines [1,2].
The main process for hydrogen production is cur-
rently the catalytic steam reforming of methane, light
hydrocarbons, and naphtha. Partial oxidation of heavy
oil residues, and coal gasication are also alternative
0926-860X/00/$ see front matter 2000 Elsevier Science B.V. All rights reserved.
PII: S0926- 860X( 00) 00440- 3
226 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
processes to produce hydrogen. Renewable lignocel-
lulosic biomass can be used as an alternative feedstock
for hydrogen production. Two possible technologies
that have been explored in recent years are steam
gasication [37] and catalytic steam reforming of
pyrolysis oils [8,9]. The latter route begins with fast
pyrolysis of biomass to produce bio-oil, which (ei-
ther as a whole or using specic fractions) can be
converted to hydrogen via catalytic steam reforming
followed by, if necessary, a shift conversion step.
Fast pyrolysis has made signicant advances in
the past 20 years [1013]. It requires rapid heat-
ing of biomass particles to temperatures between
450 and 550

C and short residence times of the


volatile vapors in the reaction zone. In such con-
ditions the yield of the liquid product, bio-oil, can
reach 75 wt.% [14]. Bio-oil is a mixture of sim-
ple aldehydes, alcohols, and acids as well as more
complex carbohydrate- and lignin-derived oligomeric
materials emulsied with water. By simply adding
water the bio-oil separates into: (a) a water-rich phase
containing mostly carbohydrate-derived compounds,
and (b) a hydrophobic phase, pyrolytic lignin, com-
Table 1
Composition of commercial steam reforming catalysts [21]
Producer Catalyst Feedstock NiO Al
2
O
3
MgO MgAl
2
O
4
CaO SiO
2
K
2
O
BGC CRG B Naphtha 79 2021 0.753.3
ICI 46-1 Naphtha 22 26 11 13 16 7
ICI
a
46-1 Naphtha 1025 1.5
ICI
b
46-1 P Naphtha 22 26 11 13 16 7
ICI
a,c
46-4 C
2
/C
3
+ + +
ICI
b
57-3 Natural gas 12 78 10 (0.1)
ICI
a
25-4M Natural gas 15 70 13 1.5
Topse RKS-1 Natural gas 15 85 (0.1) <500 ppm
Topse R67 Natural gas/LPG 15 85 (0.1) <500 ppm
Topse RKNR Naphtha 34 12 54
UCI C-11-9 Natural gas 1120
UCI G-56 Light hydrocarbons 1525
UCI G-90 Light hydrocarbons 715
UCI
a
G-90C C
3
+C
4
hydrocarbons 15 7076 58
UCI
a
G-90B C
3
+C
4
hydrocarbons 11 7682 69
UCI
a
G-91 C
3
+C
4
hydrocarbons 11 7682 69 Unknown
BASF
a
G1-25S Natural gas 15
BASF
b
G1-25/1 Naphtha 25 66 8 <0.2 1
BASF
b
G1-50 Naphtha 20 32 11 16 14 7
a
[8].
b
[22].
c
ZrO
2
is present in its composition.
posed mainly of lignin-derived oligomers [14,15].
One possible application of bio-oil being considered
is the use pyrolytic lignin as a substitute for phe-
nol in phenolformaldehyde resin formulations [16]
while the aqueous, carbohydrate-derived fraction is
converted to hydrogen by steam reforming [17].
Previous work on catalytic steam reforming of
biomass-derived liquids to produce hydrogen perfor-
med in our group included: mechanisms of reforming
model compounds (acetic acid and hydroxyacetalde-
hyde) [18]; catalyst screening [8]; and bench scale
xed-bed and uidized-bed reformer tests [9,19,20].
Commercial catalysts were mainly used in those
studies.
Table 1 presents a list of commercial catalysts de-
veloped for steam reforming of hydrocarbons. Nickel
is the main active component of the most of these cat-
alysts [8,2123]. Nickel-based catalysts are also used
for gasifying tarry compounds and for CO
2
reforming
of natural gas [2426]. Although noble metals (Ru,
Rh) are more effective than Ni and less susceptible
to carbon formation, such catalysts are not common
in industrial applications because of their cost [27].
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 227
Catalytic reforming of hydrocarbons is assumed
to proceed according to the following mechanism
[28]: (1) methane, or any other hydrocarbon, is dis-
sociatively adsorbed on the metal sites; (2) H
2
O
is also dissociatively adsorbed on the Al surface
sites, hydroxylating the surface; (3) metal-catalyzed
dehydrogenation takes place, creating adsorbed
hydrocarbon-derived fragments; (4) the OH surface
groups migrate to the metal sites, activated by the
temperature, and they eventually form intermediates
leading to carbon oxides. An unwanted side effect
of these reactions is the formation of carbon (coke)
on the catalyst surface. There are different types of
carbon deposits: whisker-like, encapsulating lm, and
pyrolytic carbon. Fundamentals of carbon formation
as well as strategies for its minimization have been
thoroughly discussed in the literature [27,29]. The
most signicant factor is the ratio of carbon in the
feedstock to steam. It is also known that carbon for-
mation increases with unsaturation, molecular weight,
and aromaticity of the feed [27].
Two strategies can be employed to decrease carbon
deposits on the catalyst. The rst one is based on en-
hancing steam adsorption on the catalyst with the ob-
jective of gasifying any carbon or carbon precursors
formed on the catalyst surface. The second strategy
aims at modifying the surface reactions via the pres-
ence of other metals. It is still uncertain whether the
reduction of carbon formation is due to the effect of
altering the crystallite size when several metals are
present in the catalyst formulation or to the formation
of alloys that interfere with the process of carbon dis-
solution in the nickel [27].
Magnesium signicantly changes Ni/Al
2
O
3
cat-
alytic properties and its presence has been reported
in several commercial reforming catalysts [21,30,31]
as well as in research formulations [25,26,3237].
Magnesium enhances steam adsorption capability
and solid solutions of NiO/MgO stabilize nickel
and prevent catalyst sintering [35]. The formation
of magnesium aluminate spinel considerably in-
creases mechanical strength of the catalyst [38].
Wang and Lu [25], in a comparative study of nickel
catalysts aimed at CO
2
reforming of methane us-
ing different supports, found that Ni/MgO showed
long-term stability, which they attributed to low sin-
tering of nickel crystallites, thus limiting carbon
formation.
The addition of lanthanum improves the stability
of the catalyst and decreases carbon formation [38].
Bangala et al. [34] observed an important decrease of
carbon deposits on the catalyst when they included
lanthanum in its formulation. This metal is also found
in several patented formulations on steam reforming
of hydrocarbons [4042]. Cheng et al. [39] relate the
promotion effect of lanthanide oxide to its interaction
with nickel. Several other studies [4345] also docu-
mented the decrease of carbon deposition in the pres-
ence of rare earth oxides.
The main effect of chromium in co-precipitated
Ni/Al
2
O
3
catalyst is the improvement of cata-
lyst stability rather than enhancement of the ac-
tivity [38]. Cr
2
O
3
is used as a promoter having
hydrogenationdehydrogenation and dehydrocycliza-
tion activity [46]. According to Bangala et al. [34]
5% seems to be the optimum Cr loading. Above this
level, the conversion decreases and the coke formation
increases. In another study [47], a four-component
Ni-Mg-Cr-La catalyst showed high activity and se-
lectivity for partial oxidation and CO
2
reforming
of methane to syngas as well as remarkable coking
resistance.
Cobalt-promoted Ni/Al
2
O
3
is signicantly more re-
sistant to deactivation by carbon than nickel alone [48].
The addition of Co to NiO-MgO catalysts has a ben-
ecial effect by eliminating or signicantly reducing
the formation of lamental carbon on the catalyst dur-
ing CO
2
reforming of methane [49].
We studied the inuence of magnesium, lanthanum,
chromium and cobalt on Ni/Al
2
O
3
catalyst used
for steam reforming the carbohydrate-derived frac-
tion of bio-oil. The objective was to test a series of
catalyst formulations having reasonably high activ-
ity and stability for hydrogen production reactions.
The performances of the research formulations were
compared to those for commercial steam reforming
catalysts tested under the same conditions. Reform-
ing tests were performed at space velocities signi-
cantly higher that those used for processing natural
gas; since we expected oxygenated organics to be
more reactive than hydrocarbons. This should allow a
high reactor throughput during hydrogen production
from bio-oils. In addition, at high space velocities
differences in catalyst performance, deactivation pat-
tern, and intermediates formation can be observed at
shorter reaction times.
228 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
2. Experimental
2.1. Apparatus
The experiments have been carried out in a dual bed
quartz reactor to which a molecular beam mass spec-
trometer (MBMS) was interfaced to analyze the prod-
ucts. A schematic of this installation, similar to that
used in a previous study [8,18], is shown in Fig. 1.
A tubular furnace with ve independently controlled
temperature zones enclosed the reactor comprising
two inner tubes surrounded by a large outer housing.
The catalyst beds were placed in the inner tubes (diam-
eter of 0.785 cm) while the outer ow was mainly used
Fig. 1. Schematic of the micro-reactor/molecular beam mass spectrometer system.
for the dilution of samples with helium to stabilize the
molecular beam and improve the signal-to-noise ratio.
The MBMS application for studying thermal and
catalytic processes was described previously [50]. Its
advantage is the rapid real-time sampling and mea-
surement of volatile products in the presence of tars
and particulates. The hot gases exiting the reactor at
ambient pressure are sampled through an orice into a
low-pressure region of the MBMS to form the molec-
ular beam. Electron impact (EI) ionization of neutral
molecules is used to produce ions. The mass spectrum
of positive ions in the mass range of interest (m/z 13
and 12250) is obtained by a scanning mass spectro-
meter (an Extrel triple quadrupole mass analyzer)
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 229
every 2 s. An IBM-compatible Pentium II

computer
is used to control the instrument and acquire data
with a Teknivent Vector Two

data system.
2.2. Catalysts
2.2.1. Commercial catalysts
The commercial catalysts used in this study were:
G-91 and C11-NK, from United Catalyst (UCI,
Louisville, KY) and 46-1 and 46-4, provided by ICI
Katalco (Cleveland, UK).
2.2.2. Research catalysts
The catalysts were prepared using the support im-
pregnation method. -Al
2
O
3
support provided by
United Catalyst (G-90 Carrier) was ground and sieved
to select particles in the size range of 355600 m.
Different nitrate solutions (nickel nitrate hexahydrate,
magnesium nitrate hexahydrate, lanthanum nitrate
hexahydrate, cobalt nitrate hexahydrate and chromium
nitrate non-ahydrate) were used for the support im-
pregnation. The catalyst preparation included drying
in an oven at 115

C overnight, then calcining in air.


In order to control more carefully the nitrate decom-
position, the precursors were maintained at 300

C for
3 h and then the Ni/Al
2
O
3
catalyst was calcined at
the nal temperature of 850

C for 5 h.
Two catalysts, Ni/MgO-Al
2
O
3
and Ni/MgO-La
2
O
3
-
Al
2
O
3
, were prepared to modify the support, follow-
ing the procedure recommended by Cheng et al. [39].
The impregnation with magnesium nitrate or magne-
sium nitrate and lanthanum nitrate was carried out
prior to the impregnation with nickel nitrate.
Cobalt and chromium were introduced to obtain
Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
and Ni-Co/MgO-La
2
O
3
-
Al
2
O
3
. The support preparation was similar to that
for the Ni/MgO-La
2
O
3
-Al
2
O
3
catalyst. The impreg-
nation of nickel and the promoter was carried out
in two steps: rst, impregnation with nickel nitrate
then impregnation with chromium nitrate or cobalt
nitrate. After that the precursors were calcined at a
nal temperature of 850

C for 2 h.
Catalyst compositions having a NiO content of
15 wt.%, and mole ratios of Mg/Ni=1, Ni/La=8,
Ni/Cr=3, and Ni/Co=3 were selected based on
known commercial and research formulations
[24,31,32,34,47,51]. Before reforming tests all the
catalysts were reduced using hydrogenhelium ow
Table 2
Surface area of commercial and research catalysts
Catalyst Surface area (m
2
/g)
Carrier G-90 (-Al
2
O
3
) 2.5
G-91 16.0
ICI 46-1 12.8
ICI 46-4 11.8
C11-NK 17.7
Ni/Al
2
O
3
2.7
Ni/MgO-Al
2
O
3
3.1
Ni/MgO-La
2
O
3
-Al
2
O
3
3.6
Ni-Co/MgO-La
2
O
3
-Al
2
O
3
4.1
Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
4.9
MgO-Al
2
O
3
2.9
MgO-La
2
O
3
-Al
2
O
3
3.1
with a H
2
content of 41 vol.%. The reduction was
carried out for 2 h at 700

C.
2.2.3. Characterization of the catalysts
The catalysts (before reduction) and the sup-
ports (before metal loading) were characterized by
single-point surface area measurement. A stream of
30.4% N
2
in helium was used to carry out these
tests in a FlowSorb II (Micromeritics) apparatus. The
results obtained are shown in Table 2.
All the catalysts had surface areas of less than
18 m
2
/g. Commercial catalysts surface areas were
greater than that of the -Al
2
O
3
support and of the
catalysts prepared in the laboratory. For the labora-
tory preparations only a slight increase in the surface
area compared with the -Al
2
O
3
support was seen
for Ni/Al
2
O
3
, MgO-Al
2
O
3
, Ni/MgO-Al
2
O
3
, and
MgO-La
2
O
3
-Al
2
O
3
but there were larger increases for
Ni/MgO-La
2
O
3
-Al
2
O
3
, Ni-Co/MgO-La
2
O
3
-Al
2
O
3
and, especially, Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
.
Atomic composition of the catalyst surface was
determined using X-ray photoelectron spectroscopy
(XPS). XPS analyses were carried out using a Physical
Electronics 5600 photoelectron spectrometer operat-
ing at a base pressure of 210
10
Torr. Photoelectrons
were excited by monochromatic Al K X-radiation
(anode power of 300 W at 15 kV) then collected at
45

from the sample normal and resolved with a


hemispherical analyzer at a pass energy of 58.7 eV
for narrow scans or 93.9 eV for survey spectra. A
low-energy electron ood gun was used to compen-
sate for sample charging. The results of XPS charac-
terization showed that the surface Ni concentration in
230 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
the commercial catalysts ranged from 4.6 to 7.6% and
the Ni/Al ratio varied from 0.2 to 0.3. The additives
detected were Ca (G-91), Ca and K (ICI 46-4), Ca,
Mg, and Si (C11-NK), and Ca, Mg, Si, and K (ICI
46-1). The Ni concentration in the research catalysts
prepared for this study varied from 10 to 14% and the
Ni/Al ratio ranged from 0.6 to 1.5. The highest Ni/Al
ratio corresponded to Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
and
Ni-Co/MgO-La
2
O
3
-Al
2
O
3
. It was also determined
that the presence of lanthanum in the catalyst caused a
signicant decrease in the surface magnesium content.
2.3. Bio-oil: carbohydrate-derived aqueous fraction
Bio-oil was generated from poplar wood using
the NREL fast pyrolysis vortex reactor system [12].
Its elemental composition comprised 46.8% carbon,
7.4% hydrogen, and 45.8% oxygen, with water con-
tent of 19.0 wt.%. It was separated into an aqueous
(predominantly carbohydrate-derived) and an organic
(lignin-derived) fraction by adding water to the oil
at a weight ratio of 2:1. The nal aqueous solution
(55% of the whole oil) contained 22.9% organics
(CH
1.34
O
0.81
) and 77.1% water. Though less complex
than the whole bio-oil, the aqueous extract contains
over 100 compounds. The most abundant (at the
level of several percent) are acetic and formic acids,
hydroxyacetaldehyde, acetol, and levoglucosan [14].
Aromatic compounds (phenols, guaiacols, syringols)
account for approximately 5% of the organic ma-
terial. About 25% of the solutes (mostly mono and
oligosaccharides) are non-volatile. These non-volatile
materials are difcult to reform because they tend
to pyrolyze to carbonaceous material which gasies
slowly and, when transported to the catalyst, blocks
the surface of the catalyst, thus deactivating it.
2.4. Steam reforming tests
A 3 cm high catalyst bed was placed in the quartz
reactor between quartz wool layers. In some tests the
catalyst was diluted with quartz chips. A K-type ther-
mocouple was placed inside the catalyst bed to moni-
tor its temperature. The experiments were essentially
isothermal at 825 or 875

C, though a small temper-


ature drop of about 3

C was observed when feeding


started. The Ni/Al
2
O
3
catalyst was used in each ex-
periment as a reference. It was always placed in one
of the two tubes of the dual reactor, while the other
tube contained the catalyst tested.
Methanol (28% aqueous solution) and the aqueous
fraction of bio-oil were catalytically steam reformed
in the micro-reactor. The mixture of methanol and
water of similar carbon weight content as the aqueous
fraction of bio-oil was always used at the beginning
of each test to check the catalyst performance.
The liquids were fed continuously by a syringe
pump (SAGE model 355) through the side ports of
the reactor using a needle injector. A small ow rate
of argon helped transport the liquid to a preheated
helium stream that was introduced in the inner tube.
The feeding rate of liquids was about 150 mg/min and
the feed-catalyst contact time was 0.026 s (calculated
as the voidage volume of catalyst bed (0.4) divided by
the total volumetric ow of input gases at the reaction
temperature and at ambient pressure).
Helium, argon, hydrogen, carbon dioxide, and car-
bon monoxide, at purity >99.99%, were obtained from
commercial suppliers.
Most of the experiments were performed at 825

C
and a steam-to-carbon ratio (S/C, mole steam/mole
carbon in feed) of 4.92. Several tests were carried out
at a higher temperature, 875

C, or higher S/C ratio


(close to 11) to study the inuence of these variables
on the process performance. In the latter experiments,
in order to maintain similar residence times and space
velocities, the aqueous fraction was diluted with water
and the catalyst bed height and the feeding rate of
liquids were adjusted accordingly.
2.5. Quantitative and semi-quantitative analysis
using MBMS
A quantitative analysis of H
2
, CO
2
and CO was
performed based on the intensities of selected mass
spectrum lines. Due to the linearity of the response
of these gases, only one concentration level, close to
that generated by steam reforming, was used for cali-
bration that was performed each experimental day. A
30 ml
STP
/min ow of argon (m/z 40) was used as an
internal standard. For the quantitative analysis of H
2
(m/z 2), CO
2
(m/z 44), and CO (m/z 28) the signal
response was calculated subtracting the background.
The ionization energy (EI) of 33.5 eV was used in the
mass range of m/z 13 to obtain more reliable H
2
re-
sponse, while in the mass range of m/z 12250 it was
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 231
25 eV in order to decrease interference. The accuracy
of the analysis was estimated to be 5%. More de-
tailed information about the quantitative analysis can
be found in Wang et al. [18].
The relative amounts of methane (m/z 16), methanol
(fragment ion m/z 31), benzene (m/z 78), toluene (m/z
92), and naphthalene (m/z 128) in the product gas
were calculated as a ratio of the intensity of the res-
pective spectrum line (decreased by the intensity of
this line in the background) to the intensity for the in-
ternal standard (argon). These values are considered
semi-quantitative.
3. Results and discussion
3.1. Initial experiments
The steam reforming of any oxygenated or-
ganic compound proceed according to the following
reaction:
C
n
H
m
O
k
+(n k)H
2
O = nCO +

n +
m
2
k

H
2
(1)
This reaction is followed by the watergas shift
reaction:
nCO +nH
2
O = nCO
2
+nH
2
(2)
Fig. 2. Gas production during catalytic steam reforming of bio-oil aqueous fraction using Ni/Al
2
O
3
catalyst, H
2
(), CO
2
(), and CO
(). T=825

C, S/C=4.92, and G
C1
HSV=126,000 h
1
.
Therefore, the overall process can be represented as
follows:
C
n
H
m
O
k
+(2n k)H
2
O
= nCO
2
+

2n +
m
2
k

H
2
(3)
Initial steam reforming experiments were carried out
at 825

C and an S/C ratio of 4.92 using Ni/Al


2
O
3
cat-
alyst, which served as a reference for comparison with
the commercial and research catalysts. Even at a very
high space velocity of 126,000 h
1
(G
C1
HSV dened
as C
1
equivalent volume of feed per hour per unit vol-
ume of catalyst), total conversion of methanol (100%
carbon conversion to CO and CO
2
) was achieved. The
amounts of H
2
, CO, and CO
2
were close to the equilib-
rium values, which correspond to 0.583 mol of CO
2
,
0.413 mol of CO, and 2.583 mol of hydrogen per mol
of methanol. In equilibrium, hydrogen yield would be
86.1% of that possible for the stoichiometric conver-
sion according to Eq. (3). When an inert bed (quartz
chips) was used at the same conditions instead of the
catalyst, less than 56% of the carbon from methanol
reacted to produce CO and CO
2
, and hydrogen yield
was only 40% of the stoichiometric potential. Also
signals of m/z 15, 31 and 32 were observed, which
indicated an incomplete conversion of methanol.
The production of H
2
, CO
2
and CO from the bio-oil
fraction at the above process conditions is shown in
Fig. 2. H
2
yield is expressed as the percentage of the
232 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
stoichiometric potential (conversion of feed to H
2
and
CO
2
), while the amounts of CO
2
and CO correspond
to the conversion of carbon in the feed to these com-
pounds. The changes in product gas composition dur-
ing the reforming of the aqueous fraction of bio-oil
were observed after 10 min of the experiment per-
formed at G
C1
HSV of 80,000 h
1
, and after 3 min of
the experiment at G
C1
HSV of 126,000 h
1
. H
2
and
CO
2
production decreased and CO increased as a re-
sult of the catalyst deactivation.
The hydrogen yield calculated from the thermo-
dynamic equilibrium at the reaction temperature of
825

C and the S/C ratio of 4.92 is 83.8% of the sto-


ichiometric (Eq. (3)), with 69.9% of the carbon con-
verted to CO
2
and 30.1%to CO. These values are close
to the results obtained at the beginning of the experi-
ment when approximately 90% of the carbon from the
bio-oil fraction was converted to CO and CO
2
, while
the remaining 10% formed methane and carbon de-
posits.
As shown in Fig. 2, after 25 min of feeding the
bio-oil fraction the H
2
yield decreased to 30% of the
stoichiometric potential, with only 8% of the feed car-
bon converted to CO
2
and 68% to CO. The com-
parison of these values with those obtained for the
non-catalytic reforming shows that the catalyst still
has some activity.
Fig. 3. Minor products observed during catalytic steam reforming of bio-oil aqueous fraction using Ni/Al
2
O
3
catalyst, methane (m/z 16)
(), methanol (m/z 31) (), benzene (m/z 78) (), toluene (m/z 92) () and naphthalene (m/z 128) (). T=825

C, S/C=4.92, and
G
C1
HSV=126,000 h
1
.
With the catalyst deactivation an increase in the pro-
duction of CH
4
(m/z 16), methanol (fragment ion m/z
31), benzene (m/z 78), toluene (m/z 92), and naphtha-
lene (m/z 128) was observed. The amounts of these
components measured as signal intensity versus time
is shown in Fig. 3. We estimate that the amount of ben-
zene in the product gas was less than 1%. It can be seen
that the production of these components stabilized af-
ter 15 min of feeding at a signicantly lower level, ex-
cept for CH
4
, than that observed for the non-catalytic
process, in which other components such as phenol,
styrene, xylene, cresol, and methylnaphthalene were
also detected.
These results indicate that under the conditions used
for processing methanol and the bio-oil fraction both
thermal decomposition and catalytic reaction occur.
Though the rate of the latter is signicantly greater,
thermal decomposition cannot be avoided. Carbon is
formed via both pathways. For volatile compounds
this is due to gas-phase reactions. The non-volatile
material present in the bio-oil pyrolyzes in the hot
reactor producing carbonaceous material that, when
transported to the catalyst blocks access to the surface,
deactivating the catalyst. The enhancement of water
(steam) adsorption on the catalyst surface may favor
the gasication of carbon and thus improve the overall
reforming process.
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 233
Fig. 4. Hydrogen yield obtained by catalytic steam reforming of bio-oil aqueous fraction, (a and b) the reference catalyst, Ni/Al
2
O
3
(,
two experiments)
,
(a) commercial catalysts, G-91 (), 46-1 (), 46-4 () C11-NK (), (b) research catalysts, Ni/MgO-Al
2
O
3
(),
Ni/MgO-La
2
O
3
-Al
2
O
3
(), Ni-Co/MgO-La
2
O
3
-Al
2
O
3
(, two experiments), and Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
(). T=825

C, S/C=4.92,
and G
C1
HSV of about 126,000 h
1
.
3.2. Commercial catalysts
All commercial catalysts, G-91, ICI 46-1, ICI 46-4,
and C11-NK, were tested at 825

C and a G
C1
HSV of
approximately 126,000 h
1
. About 0.08 g of catalyst
was mixed in the bed with 1.69 g of quartz chips. The
nominal gas residence time in the bed was 0.026 s and
the molar S/C ratio was 4.92. Fig. 4a shows hydrogen
yield as a function of time for all commercial cata-
lysts and for the reference catalyst (Ni/Al
2
O
3
). At the
beginning of the experiment, H
2
yields for the com-
mercial and for the reference catalysts were almost
the same. However, commercial catalysts performed
better (higher H
2
yield) than the reference catalyst at
longer times on stream. After 25 min the H
2
yields
were approximately 65% of the stoichiometric poten-
tial for commercial catalysts while a decrease to 35%
was observed for the Ni/Al
2
O
3
reference. CO
2
yields
paralleled those of H
2
, for commercial catalysts after
25 min on stream they were about 48% of the value for
the stoichiometric conversion and signicantly less for
the Ni/Al
2
O
3
catalyst. At the beginning of the exper-
iments the amounts of carbon converted to CO were
similar for both commercial and the reference cata-
lysts. However, after 25 min on stream CO production
increased slowly to around 0.3 mol CO/mol C fed us-
ing commercial catalysts. The increase was greater for
the reference Ni/Al
2
O
3
catalyst.
Fig. 5a presents the conversion of carbon in the
bio-oil fraction to CO and CO
2
. Commercial cat-
alysts and the Ni/Al
2
O
3
reference show similar
performances with a decrease in carbon to gas con-
version from approximately 9080% after 25 min
on stream. In light of the higher hydrogen yields
234 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
Fig. 5. Carbon conversion to CO and CO
2
for catalytic steam reforming of bio-oil aqueous fraction, (a and b) the reference catalyst,
Ni/Al
2
O
3
(, two experiments), (a) commercial catalysts, G-91 (), 46-1 (), 46-4 () and C11-NK (), (b) research catalysts,
Ni/MgO-Al
2
O
3
(), Ni/MgO-La
2
O
3
-Al
2
O
3
(), Ni-Co/MgO-La
2
O
3
-Al
2
O
3
(, two experiments) and Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
().
T=825

C, S/C=4.92, and G
C1
HSV of about 126,000 h
1
.
obtained using the commercial catalysts it is clear that
the commercial preparations catalyzed the watergas
shift reaction more effectively than did the reference
catalyst.
The evolution of methane, methanol, benzene,
toluene, and naphthalene during steam reforming of
the bio-oil fraction was very similar for all commercial
catalysts. Fig. 6a shows the data obtained using G-91
commercial catalyst. A slight increase in methane in-
tensity and a larger increase in benzene intensity (up
to values of 0.0036 after 25 min) were observed. The
other components were detected at even lower levels.
A comparison of Figs. 3 and 6a shows that although
initial amounts of hydrocarbons generated are similar
for all the catalysts, they increase much faster for
the reference catalyst than for the commercial ones.
This is consistent with the slower deactivation of
commercial catalysts shown by the hydrogen yields.
3.3. Research catalysts
In addition to the reference Ni/Al
2
O
3
catalyst, four
other catalysts (Ni/MgO-Al
2
O
3
, Ni/MgO-La
2
O
3
-
Al
2
O
3
, Ni-Co/MgO-La
2
O
3
-Al
2
O
3
and Ni-Cr/MgO-
La
2
O
3
-Al
2
O
3
) were tested for the efciency of steam
reforming the aqueous bio-oil fraction at 825

C,
G
C1
HSV of about 126,000 h
1
, and S/C molar ratio
of 4.92.
The hydrogen yield as a function of time for the
laboratory catalysts is shown in Fig. 4b. Compared
to the reference catalyst, the Ni/MgO-Al
2
O
3
catalyst
performed better (higher hydrogen yield and slower
deactivation). Even more improvement was observed
for the Ni/MgO-La
2
O
3
-Al
2
O
3
catalyst for which the
hydrogen yield was approximately 55% of the sto-
ichiometric potential after 25 min on stream. Addition
of cobalt or chromium to this catalyst further
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 235
Fig. 6. Minor products observed during catalytic steam reforming of bio-oil aqueous fraction using (a) G-91 commercial catalyst or (b)
Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
research catalyst, methane (m/z 16) (), methanol (m/z 31) (), benzene (m/z 78) (), toluene (m/z 92) (),
and naphthalene (m/z 128) (). T=825

C, S/C=4.92, and G
C1
HSV=124,000 h
1
.
improved its performance and also changed the shape
of the deactivation curve for these catalysts for
about 12 min on stream the hydrogen yield was main-
tained at the initial level and then started decreasing.
A slightly higher hydrogen yield was observed for
the Ni-Co/MgO-La
2
O
3
-Al
2
O
3
preparation than for
the Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
catalyst, though the
difference was within the precision of the analyti-
cal method used. The changes in CO
2
yields during
steam reforming tests using these catalysts showed
a tendency similar to that observed for H
2
while
the carbon conversion to CO increased during the
tests. Additions of magnesium, lanthanum, cobalt,
and chromium resulted in less CO production than
that obtained using the reference catalyst. The carbon
conversion to gas (CO and CO
2
) versus time using
the research and the reference catalysts is presented
in Fig. 5b. The decrease from about 90 to 80% after
25 min on stream was observed for all the catalysts.
As for the commercial catalysts, methane and
benzene were the most important minor compounds
observed in the reforming product gas. Figs. 7 and 8
present the evolution of methane and benzene (in ar-
bitrary units) during reforming of the bio-oil fraction
using the research catalysts. Lower methane produc-
tion was observed for Ni/MgO-Al
2
O
3
than for the
Ni/Al
2
O
3
catalyst. It decreased even more signi-
cantly in the presence of lanthanum in the catalysts
(Ni/MgO-La
2
O
3
-Al
2
O
3
, Ni-Co/MgO-La
2
O
3
-Al
2
O
3
and Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
). Although reform-
ing using the Ni/MgO-Al
2
O
3
catalyst generated less
benzene than Ni/MgO-La
2
O
3
-Al
2
O
3
at short times
on stream (less than 10 min), at longer times the ad-
dition of lanthanum proved benecial. The lowest
236 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
Fig. 7. Methane produced using research catalysts for catalytic steam reforming of bio-oil aqueous fraction, Ni/MgO-Al
2
O
3
(),
Ni/MgO-La
2
O
3
-Al
2
O
3
(), Ni-Co/MgO-La
2
O
3
-Al
2
O
3
(, two experiments), Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
() and the reference catalyst,
Ni/Al
2
O
3
(, two experiments). T=825

C, S/C=4.92, and G
C1
HSV of about 126,000 h
1
.
benzene concentrations in the reforming gas were
observed when Ni-Co/MgO-La
2
O
3
-Al
2
O
3
and Ni-Cr/
MgO-La
2
O
3
-Al
2
O
3
catalysts were employed.
Comparing the efciency of the research and the
commercial catalysts it can be seen that the hydrogen
yield obtained using the former was lower than that
observed for the latter after 25 min on stream. How-
ever, the carbon conversion to CO and CO
2
was very
Fig. 8. Benzene produced using research catalysts for catalytic steam reforming of bio-oil aqueous fraction, Ni/MgO-Al
2
O
3
(),
Ni/MgO-La
2
O
3
-Al
2
O
3
(), Ni-Co/MgO-La
2
O
3
-Al
2
O
3
(, two experiments), Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
(), and the reference catalyst
(, two experiments), Ni/Al
2
O
3
. T=825

C, S/C=4.92, and G
C1
HSV of about 126,000 h
1
.
similar for all the catalysts. Carbon conversion to CO
and CO
2
was less than 100% mostly due to the forma-
tion of coke and char, but also because of the reactions
leading to methane, benzene, and other aromatic hy-
drocarbons. Coke deposits have been observed on the
catalyst surface during steam reforming of even such
a simple compound as acetic acid [18]. At reaction
temperatures thermal decomposition of the organics
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 237
present in the aqueous fraction of bio-oil could occur
before the feed reached the catalyst bed. Some of
the oxygenated products derived from the carbohy-
drate component of biomass, especially anhydrosug-
ars, are known to dehydrate rapidly and form char
[52].
The improvement in performance of the research
preparations and the commercial catalysts compared
to the reference nickel catalyst were shown as higher
hydrogen and CO
2
yields and lower COyield. It seems
that the main difference in the catalyst effectiveness
was related to the watergas shift reaction. We ob-
served increased water (m/z 18) in the product gas at
longer catalyst times on stream. Possible reasons for
the higher efciency of the commercial catalysts could
be their surface area (four to six times greater than
that for the research formulations) as well as the lower
surface Ni/Al ratio. Either of these could enhance
steam adsorption on the surface, thus favoring carbon
oxidation reactions. In addition to the watergas shift,
the best laboratory (Ni-Co/MgO-La
2
O
3
-Al
2
O
3
and
Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
) and commercial catalysts
more efciently prevented the formation of methane
and benzene than the reference Ni/Al
2
O
3
catalyst. The
XPS analysis showed that Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
and Ni-Co/MgO-La
2
O
3
-Al
2
O
3
, which produced the
highest amounts of hydrogen and the lowest ben-
zene from reforming the bio-oil fraction, had also
the greatest surface Ni/Al ratio. The commercial cat-
Fig. 9. Gas production during catalytic steam reforming of bio-oil aqueous fraction using Ni-Co/MgO-La
2
O
3
-Al
2
O
3
research catalyst, H
2
(), CO
2
(), and CO (). T=825

C, S/C=4.92, and G
C1
HSV=62,300 h
1
.
alysts, however, provided similar performances with
lower surface nickel concentrations and Ni/Al ratios
than the research formulations.
3.4. Inuence of temperature, steam-to-carbon ratio,
and space velocity
The commercial and research catalysts tested at a
higher temperature, 875

C, or at a steam-to-carbon ra-
tio of about 11, surprisingly did not show a signicant
increase in the hydrogen production. The changes in
these two variables, temperature and S/C ratio, were
expected to decrease the catalyst deactivation and to
extend the duration of its stable performance. Higher
S/C ratio ought to improve coke gasication and fa-
vorably shift the watergas shift equilibrium. Higher
temperatures should favor the former but negatively
affect the latter. The reason for the insignicant effects
is not clear at this time yet.
A signicant increase in hydrogen production and
an improvement in catalyst stability were observed
when the laboratory catalyst, Ni-Co/MgO-La
2
O
3
-
Al
2
O
3
, was tested at lower space velocity. Fig. 9
shows H
2
, CO
2
and CO production as a function of
time during the steam reforming of the bio-oil aque-
ous fraction at 825

C, G
C1
HSV of 62,300 h
1
, and
S/C ratio of 4.92. H
2
, CO and CO
2
yields remained
constant for more than 22 min. Benzene and toluene
were detected only after 20 min of the experiment and
238 L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239
at much lower levels than those measured at higher
space velocity.
4. Conclusions
We have demonstrated that high space velocities
(up to 126,000 h
1
) and low residence times (26 ms)
can be used when reforming bio-oils (825875

C,
4.911.0 molar steam/carbon ratio).
Gradual deactivation of the catalysts was observed
as a function of time on stream. This resulted in
a decrease of H
2
and CO
2
and an increase of the
CO production. Total carbon conversion to CO and
CO
2
decreased at a slower rate. Also more methane,
benzene, and other aromatics were detected in the
reforming gas at longer times on stream.
The presence of lanthanum in the research cata-
lysts resulted in a signicant decrease in methane
and benzene production compared to that observed
for the other preparations. The lowest levels of ben-
zene in reformed gases were detected when Ni-Co/
MgO-La
2
O
3
-Al
2
O
3
and Ni-Cr/MgO-La
2
O
3
-Al
2
O
3
catalysts were used. The improved performance of
those catalysts was due to the benecial effects of
the additives. Magnesium and lanthanum enhanced
steam adsorption that facilitated the gasication of
surface carbon. Cobalt and chromium modied the
metal sites forming alloys with nickel and possibly
reducing the crystallite size.
Although carbon conversion to CO and CO
2
de-
creased slightly with reaction time, the main impact
on the decrease of H
2
and CO
2
yields appears to
be linked to the ability of the catalyst to enhance
the watergas shift reaction. Therefore, we intend
to study the inuence of nickel concentration on
the overall catalyst performance. We believe that
more hydrophilic sites surrounding nickel crystallites
are needed to extend the duration of the catalysts
activity.
Though catalyst deactivation upon steam reform-
ing of complex bio-oils occurs much faster than with
natural gas or naphtha, the catalyst can be efciently
regenerated by steam or CO
2
gasication. A process
for hydrogen production from biomass liquids will
thus require a reactor conguration that allows for
the regeneration and reuse of the catalyst such as
alternating xed-beds or a uidized bed.
Acknowledgements
The authors are grateful to Mr. Neil Rossmeissl and
Ms. Catherine Grgoire-Padr, the managers of the
U.S. DOE Hydrogen Program, for nancial support of
this research. The acknowledgement is also extended
to the Secretara de Estado de Universidades, Investi-
gacin y Desarrollo (Spain) for a postdoctoral fellow-
ship awarded to Luca Garcia that made possible her
participation in this project.
References
[1] J.L. Cox, A.Y. Tonkovich, D.C. Elliott, E.G. Baker, E.J.
Hoffman, in: Proceeding of Second Biomass Conference of
the Americas, Portland, OR, 2124 August 1995, p. 657.
[2] E.D. Larson, R.E. Katofsky, in: A.V. Bridgwater (Ed.),
Advances in Thermochemical Biomass Conversion, Blackie,
London, 1994, p. 495.
[3] M.P. Aznar, J. Corella, J. Delgado, J. Lahoz, Ind. Eng. Chem.
Res. 32 (1993) 1.
[4] M.P. Aznar, M.A. Caballero, J. Gil, A. Olivares, J. Corella,
in: R.P. Overend, E. Chornet (Eds.), Making a Business form
Biomass, Pergamon Press, New York, 1997, p. 859.
[5] S. Rapagna, N. Jand, P.U. Foscolo, Int. J. Hydrogen Energy
23 (1998) 551.
[6] S. Turn, C. Kinoshita, Z. Zhang, D. Ishimura, J. Zhou, Int.
J. Hydrogen Energy 23 (1998) 641.
[7] L. Garcia, M.L. Salvador, J. Arauzo, R. Bilbao, Energy and
Fuels 13 (1999) 851.
[8] D. Wang, S. Czernik, D. Montan, M. Mann, E. Chornet,
Ind. Eng. Chem. Res. 36 (1997) 1507.
[9] D. Wang, S. Czernik, E. Chornet, Energy and Fuels 12 (1998)
19.
[10] D.S. Scott, J. Piskorz, D. Radlein, Ind. Eng. Chem. Process
Des. Dev. 24 (1985) 581.
[11] R.G. Graham, B.A. Freel, M.A. Bergougnou, in: A.V.
Bridgwater, J.L. Kuester (Eds.), Research in Thermochemical
Biomass Conversion, Elsevier, London, 1988, p. 629.
[12] J. Diebold, J. Scahill, in: E.J. Soltes, T.A. Milne (Eds.),
Pyrolysis Oils from Biomass: Producing, Analyzing and
Upgrading, ACS Symposium Series 376, American Chemical
Society, Washington, D.C., 1988, p. 31.
[13] A.M.C. Janse, W. Prins, W.P.M. van Swaaij, in: A.V.
Bridgwater, D.G.B. Boocock (Eds.), Developments in
Thermochemical Biomass Conversion, Blackie, London,
1997, p. 368.
[14] J. Piskorz, D.S. Scott, D. Radlein, in: E.J. Soltes, T.A. Milne
(Eds.), Pyrolysis Oils from Biomass: Producing, Analyzing
and Upgrading, ACS Symposium Series 376, American
Chemical Society, Washington, D.C., 1988, p. 167.
[15] S. Czernik, D. Wang, D. Montan, E. Chornet, in:
A.V. Bridgwater, D.G.B. Boocock (Eds.), Developments
in Thermochemical Biomass Conversion, Blackie, London,
1997, p. 672.
L. Garcia et al. / Applied Catalysis A: General 201 (2000) 225239 239
[16] S.S. Kelley, X.-M. Wang, M.D. Myers, D.K. Johnson,
J.W. Scahill, in: A.V. Bridgwater, D.G.B. Boocock (Eds.),
Developments in Thermochemical Biomass Conversion,
Blackie, London, 1997, p. 557.
[17] M.K. Mann, P.L. Spath, K. Kadam, in: Proceedings of the
1996 U.S. DOE Hydrogen Program Review, Miami, FL, 12
May 1996, NREL/CP-430-21968, National Renewable Energy
Laboratory, Golden, CO, 1996, p. 249.
[18] D. Wang, D. Montan, E. Chornet, Appl. Catal. A. 143 (1996)
245.
[19] S. Czernik, R. French, C. Feik, E. Chornet, in: R.P. Overend,
E. Chornet (Eds.), Biomass, A Growth Opportunity in Green
Energy and Value-Added Products, Elsevier, Amsterdam,
1999, p. 827.
[20] S. Czernik, R. French, C. Feik, E. Chornet, ACS Division
of Fuel Chemistry, Preprints of Symposia, Vol. 44, No. 2,
Anaheim, CA, March 1999, p. 240.
[21] C.N. Sattereld, Heterogeneous Catalysis in Industrial
Practice, McGraw-Hill, New York, 1991, p. 427.
[22] J. Corella, A. Oro, P. Aznar, Ind. Eng. Chem. Res. 37 (12)
(1998) 4617.
[23] L.K. Mudge, E.G. Baker, M.D. Brown, W.A. Wilcox, in: A.V.
Bridgwater, J.L. Kuester (Eds.), Research in Thermochemical
Biomass Conversion, Elsevier, London, 1988, p. 1141.
[24] S. Wang, G.Q. Lu, Appl. Catal. A 169 (1998) 271.
[25] S. Wang, G.Q. Lu, Appl. Catal. B 16 (1998) 269.
[26] S. Wang, G.Q. Lu, Energy and Fuels 12 (2) (1998) 248.
[27] D.L. Trimm, Catal. Today 37 (1997) 233.
[28] J.R.H. Ross, M.C.F. Steel, A. Zeini-Isfahani, J. Catal. 52
(1978) 280.
[29] J.R. Rostrup-Nielsen, Catal. Today 37 (1997) 225.
[30] P.B. Tottrup, B. Nielsen, Hydrocarb. Process. (1982) 89.
[31] J.R. Rostrup-Nielsen, US Patent RE 028,655 (1975), to Haldor
Topsoe A/S.
[32] J. Arauzo, D. Radlein, J. Piskorz, D.S. Scott, Ind. Eng. Chem.
Res. 36 (1997) 67.
[33] D.N. Bangala, N. Abatzoglou, J.-P. Martin, E. Chornet, Ind.
Eng. Chem. Res. 36 (1997) 4184.
[34] D.N. Bangala, N. Abatzoglou, E. Chornet, AICHE J. 44
(1998) 927.
[35] V.R. Choudhary, B.S. Uphade, A.S. Mamman, J. Catal. 172
(1997) 281.
[36] V.R. Choudhary, A.S. Uphade, A.S. Mamman, Appl. Catal.
A 168 (1998) 33.
[37] E. Ruckenstein, Y.H. Hu, Appl. Catal. A 154 (1997) 185.
[38] J.R.H. Ross, Catalysis, Vol. 7, The Royal Society of
Chemistry, London, 1985, p. 1.
[39] Z. Cheng, Q. Wu, J. Li, Q. Zhu, Catal. Today 30 (1996) 147.
[40] R.M. Sambrook, J.R.H. Ross, US Patent 4,469,815 (1984),
to Dyson Refractories Limited.
[41] H.J. Setzer, US Patent 4,414,140 (1983), to United
Technologies Corporation.
[42] D.N. Bangala, E. Chornet, US Patent 5,679,614 (1997), to
University of Sherbrooke.
[43] . Slagtern, U. Olsbye, R. Blom, I.M. Dahl, H. Fjellvg,
Appl. Catal. A 165 (1997) 379.
[44] S. Wang, G.Q. Lu, Appl. Catal. B 19 (1998) 267.
[45] L. Wenying, F. Jie, X. Kechang, Reac. Kinet. Catal. Lett.
64 (2) (1998) 381.
[46] A.L. Hensley, T.D. Newitt, US Patent 3,893,908 (1975), to
Standard Oil Company.
[47] P. Chen, H.B. Zhang, G.D. Lin, K.R. Tsai, Appl. Catal. A
166 (1998) 343.
[48] C.H. Bartholomew, G.D. Weatherbee, G.A. Jarvi, Chem. Eng.
Commun. 5 (1980) 125.
[49] V.R. Choudhary, A.S. Mamman, J. Chem. Technol.
Biotechnol. 73 (1998) 345.
[50] R.J. Evans, T.A. Milne, Energy and Fuels 1 (1987) 123.
[51] J.S. Choi, K.I. Moon, Y.G. Kim, J.S. Lee, D.L. Trimm, Catal.
Lett. 52 (1998) 43.
[52] M.J. Antal, in: K.W. Boer, J.A. Dufeld (Eds.), Advances in
Solar Energy, Solar Energy Society, New York, 1982, p. 61.

Você também pode gostar