Você está na página 1de 11

Chemical Engineering Science 62 (2007) 35023512

www.elsevier.com/locate/ces
Multi-scale catalyst design
Wei Liu

Science and Technology Division, Corning Incorporated, Corning, NY 14831, USA


Received 10 July 2006; received in revised form 16 January 2007; accepted 7 February 2007
Available online 28 March 2007
Abstract
Catalyst design has long been sought in catalysis and reaction engineering research. In this work, multi-scale analysis and strategy is explored
to take a holistic view toward catalyst design perspective and elucidate impacts of designs at different scales to a catalytic reaction process
performance. A few promising design concepts are introduced to break the compromise that often needs to be made in the conventional design
approach. In the catalyst bed scale, micro- or mini-structured catalyst designs can be used to potentially eliminate all mass transfer resistance
and realize intrinsic catalytic performance. At the particle level, incorporation of membrane separation functions into the catalyst unit enables
regulation of mass transfer rate of individual reactant or product molecules that high reaction selectivity is achieved. At the level of intrinsic
catalyst structures, three-dimensional (3-D) catalyst design models are outlined here to outweigh limitations or constraints imposed by the
conventional way of thinking 2-D catalyst surface. Examples of exceptional catalytic activity or concerted effects are shown by incorporating
different materials into nano-composite catalysts and optimizing size and/or shape of a catalyst material at the nano-scale.
2007 Elsevier Ltd. All rights reserved.
Keywords: Multi-scale; Multi-dimension; Nano-scale; Catalyst design; 3-D model; Interface; Composite; Ceria; Copper; Gold; Membrane; Micro-reactor;
Structured catalyst
1. Introduction
Catalytic technologies are critical to present and future en-
ergy, chemical process, and environmental industries. Conver-
sion of crude oil, coal, and natural gas to fuels and chemical
feedstock, production of a variety of petrochemical and chem-
ical products, and emission control of CO, hydrocarbons, and
NO all rely on catalytic technologies. Catalysts are also essen-
tial components of electrodes for fuel cells that use either solid
oxide ionic or polymeric proton electrolyte. Sattereld (1991)
provided a comprehensive introduction to solid catalyst sys-
tems that are signicant to industrial application.
Rationalizing catalyst designs in terms of chemical compo-
sitions and physical structures has long been sought in catal-
ysis and reaction engineering research. A number of excellent
books, treaties, and reviewarticles have been published address-
ing almost every aspect of the catalyst science and technology.

Current address: Pacic Northwest National Laboratory, Richland, WA.


Tel.: +1 607 936 1262, +1 509 375 2524 (ofce).
E-mail addresses: weiliu2476@yahoo.com, wei.liu@pnl.gov.
0009-2509/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2007.02.057
Bell (1990) discussed catalyst designs based on molecular-
level understanding of catalyst structure, surface adsorption,
and reaction mechanisms. Signicant progress in characteriza-
tion capabilities (such as SEM, TEM, STEM, AFM, etc.), in
situ surface adsorption and reaction experimental techniques,
and molecular dynamics modeling enables such an approach.
Hegedus and Pereira (1990) presented some point of views to-
ward use of chemical reaction engineering principles in newcat-
alyst development. A catalyst design strategy through effective
combination of reaction chemistries of known catalysts with
known reaction engineering principles was outlined by Davis
(1994). Recently, Tian et al. (2004) illustrated micro-scale cata-
lyst development and design with bi-metallic Pt-SnO/-alumina
catalyst system for catalytic-reforming processes and Mo/
HZSM-5 catalyst for direct conversion of methane to aromat-
ics. Liu (2005) classied catalyst technologies into three levels,
macro-scale, micro- and mini-scale, and nano-scale, and pre-
sented an overview of catalyst development at each scale. Ying
(2006) reviewed recent progress in preparation of metal oxide
catalysts of nano-structures by use of novel catalyst synthesis
strategies and techniques.
W. Liu / Chemical Engineering Science 62 (2007) 35023512 3503
Aside from catalyst research, multi-scale science and tech-
nology has become an increasingly interesting research subject
or idea. Li and Kwauk (2003) proposed a multi-scale method-
ology for analysis of complex systems in chemical engineering,
and demonstrated it as an insightful approach toward revealing
and quantifying detailed ow structures of gassolid uidized
beds.
In this work, multi-scale catalyst design concepts are pro-
posed to delineate catalyst design attributes at different scales,
reactor level, catalyst particle level, and intrinsic catalyst site
or structure level. The multi-scale analysis claries impact and
limitation of catalyst designs at each scale and also allows
a holistic view toward catalyst research, development, and
practice.
2. Catalyst design at reactor level
Catalyst design at the reactor level addresses practical prob-
lems in implementation of catalytic technologies, such as
scale-up, durability, etc. At this scale, catalyst product design
becomes an integral part of reactor design. Hydrodynamics,
external mass transfer, and mechanical reliability are often the
major factors that determine the catalyst design characteristic
for given catalyst chemistries. For example, catalyst particles
used in the uidized bed need to be resistant to attrition, while
catalyst pellets in the xed bed need to be strong enough to
withstand large static pressure.
Desired performance criteria from the catalyst bed design
include:
(i) low pressure drop,
(ii) high catalyst loading per reactor volume,
(iii) high external mass transfer rate from bulk uid onto cata-
lyst surface, and high internal mass transfer rate from cat-
alyst surface to the catalyst inside.
Heat transfer and management are also important but it is
generally addressed through the reactor design and will not
be discussed here. A great amount of knowledge and how-
how in this area has been accumulated over the years from
commercialization of various catalytic technologies. Fixed bed
and uidized bed reactors are widely used in the industrial
process. Fulton (1986) presented a series of articles on cata-
lyst engineering for xed bed designs. Presently, the catalyst
bed design can be aided by extensive use of computer model-
ing tools such as computational uid dynamics (Nijemeisland
et al., 2004).
However, both uidized bed and xed bed design are associ-
ated with fundamental limitations. Three desired performance
attributes listed above could not be achieved at the same time
and need to be compromised in the xed or uidized bed design.
The uidized bed offers low pressure drop and uniform mix-
ing of catalyst particles. As illustrated in Fig. 1, there is often
not efcient contacting between the reaction uid and catalyst
particle due to clustering of the particles and/or bypass of the
reaction uid. The uid/catalyst contacting may be improved
in certain uidization regimes (Ge and Li, 2002) when the solid
particle loading fraction is low, such as less than 0.1. But the
low solid particle fraction reduces reaction conversion rate on
the reactor volume basis. The xed bed reactor can host signi-
cant fraction of the catalyst volume, greater than 0.5. However,
pressure drop becomes large when the reactor needs to be op-
erated at high space velocity and/or the catalyst particle size is
small.
The micro- or mini-structured catalyst bed looks very attrac-
tive to solve those long-standing problems with the conven-
tional bed designs. Fig. 1 shows that in the structured bed, the
reaction uid passes through straight, open channels with the
channel wall being catalyzed. Design characteristic and per-
formance attributes of the structured bed is compared to the
uidized bed and xed bed in Table 1.
For the uidized bed or xed bed, the primary catalyst design
dimension is particle size. In the uidized bed, spherical shape
is generally preferred. In the xed bed, the catalyst can be made
in various shapes (Afandizadeh and Foumeny, 2001), sphere,
cylinder, trilobe, wagon wheel, minilith, and so on. However,
impact of the catalyst shape is secondary as compared to the
catalyst size. By contrast, the structured catalyst bed offers one
more independent design dimension than the xed bed or u-
idized bed, which enables decoupling of hydrodynamics and
external mass transfer from the catalyst design itself. As a re-
sult, the above-listed three performance attributes do not need
to be compromised in the structured bed. Hydrodynamics and
external mass transfer can be optimized by adjusting channel
size and channel shape, while the catalyst loading and/or inter-
nal mass transfer can be controlled by the thickness and struc-
ture of catalyst channel wall.
Development of the structured catalyst bed can be traced
back to monolith catalytic converters in the 1970s. The ceramic
monolith structure of channel size in the order of 1 mm, as
shown in Fig. 2, has been successfully used in the automotive
emission control. Unfortunately, this catalytic materials tech-
nology has not been widely adopted in the catalytic conver-
sion processes beyond the emission control area. The structured
body can be made of ceramic, metallic, or glass materials by
extrusion process, which is suitable for large-scale application.
The recent progress in the micro-reactor area employs silicon
material for fabrication of more sophisticated structures. As il-
lustrated in Fig. 2, active catalytic material can be a coating
layer on the channel wall made of inert material or the whole
channel wall can be catalytic material itself.
One advantage with the structured system compared to the
xed bed is low pressure drop, because of its straight ow chan-
nels versus tortuous ow paths. Fig. 3 shows pressure drops
of different catalyst beds for dehydrogenation of ethyl benzene
to styrenea practical gas-phase catalytic process (Liu et al.,
2002). In an axial ow reactor conguration of the same reac-
tor diameter, i.e., the same supercial gas linear velocity, the
monolith catalyst bed provides nearly two orders of magnitude
lower P than the catalyst particle-loaded bed. The pressure
drop through the monolith-structured bed is comparable to that
through the particle-loaded bed in radial ow reactor cong-
uration. Although the radial ow reactor design enables use
of catalyst particles at low P, it is associated with several
3504 W. Liu / Chemical Engineering Science 62 (2007) 35023512
Fig. 1. Catalyst bed design for different types of reactor.
Table 1
Characteristic of micro-structured catalyst bed compared to xed bed and uidized bed
Performance Micro-structured bed Fixed bed Fluidized bed
Primary design Channel size Particle size Particle size
parameter Channel shape
Wall thickness
Secondary design Wall structure Particle shape Particle structure
parameter Channel pattern Particle structure
Hydrodynamics Low P High P Low P
Wide range of ow conditions Narrow range of ow condi-
tions
Narrow range of ow condi-
tions
External mass transfer (mixing) Unlimited by independent
control of channel size and
ow conditions
Limited by particle size and
ow conditions
Limited due to clustering of
particles and/or bypass such
as bubbling
Segregation at particle scale
for G/L multiphase ow
Internal mass transfer-pore diffusion Unlimited by independent
control of catalyst thickness
Limited by particle size Unlimited due to use of ne
particles
drawbacks, such as complicated internal structures, poor uti-
lization of reactor space, maldistribution of feed uids, etc.
Impact of the structured catalyst bed to the reaction perfor-
mance of gas/liquid reactions probably can be more pronounced
than to the gas-phase reaction (Liu, 2002), because multiphase
catalytic reactions involve complex ow hydrodynamics and
multiple mass transfer steps. Trickle beds are commonly used
for the gas/liquid reaction over a solid catalyst. Flow charac-
teristic of monolithic structured catalysts for gas/liquid two-
phase ow in comparison to the trickle bed was elucidated by
Sattereld and Ozel as early as 1977. Fig. 4 shows that under the
similar gas and liquid ux, at the channel opening comparable
to the inter-particle void, the pressure drop through the mono-
lith channel is two to three orders of magnitude lower than that
through the trickle bed. The low P advantage of the struc-
tured bed is clearly shown for both gas-phase and gas/liquid
two-phase ows.
Another advantage with the structured bed is high mass trans-
fer rate from the bulk uid onto the channel wall. This ad-
vantage is realized by use of small channel sizes, in the order
of 1 mm. For the gas-phase catalytic reaction, the fast mass
transfer has been well demonstrated in the automotive catalytic
converter where 99% reaction conversion is achieved at gas-
hourly space-velocity above 100,000 l/h. It is noteworthy that
the external mass transfer rate for gas/liquid multiphase reac-
tions is also fairly high inside small channels. Mass transfer
rate constant of the liquid reactant from the bulk to the channel
wall surface for gas/liquid catalytic reactions is mapped over
a wide range of ow conditions by Liu et al. (2005). The lit-
erature data from the trickle bed study is plotted in Fig. 5 for
W. Liu / Chemical Engineering Science 62 (2007) 35023512 3505
Fig. 2. Ceramic monolith as building block of mini-structured bed.
10
100
1000
10000
P
e
l
l
e
t

A
P
e
l
l
e
t

B
P
e
l
l
e
t

A
P
e
l
l
e
t

B
M
o
n
o
l
i
t
h

C
M
o
n
o
l
i
t
h

D
P
r
e
s
s
u
r
e

d
r
o
p
,

m
b
a
r
Radial
Flow
Reactor
Axial Axial
Fig. 3. Pressure drop of gas ow through different packed beds for dehydro-
genation of ethylbenzene to styrene.
Pellet A B Monolith C D
Diameter 3 mm 3.5 mm Wall 1.5 mm 1.05 mm
thickness
Shape Cylindrical Shaped Cell density 25 cpsi 50 cpsi
Void 0.33 0.33 Void 0.49 0.49
fraction fraction
Packing 1300 kg/m
3
1300 kg/m
3
Packing 1000 kg/m
3
1000 kg/m
3
density density
comparison. At the same liquid supercial linear velocity, the
mass transfer rate in the monolithic channel is nearly one order
of magnitude higher than that in the trickle bed. This can be
explained by intensive gas/liquid mixing inside the channel and
dispersed liquid thin lm on the channel surface, which results
Fig. 4. Pressure drop of co-current gas/liquid down ow through parti-
cle-loaded xed bed in comparison to monolith bed (data from Sattereld
and zel, 1977).
in intimate contact of gas/liquid/catalyst. Liquid mass transfer
rate constant in 1-mm ow channel can be estimated by the
following correlation equation developed by Liu et al. (2005):
ak
ls
(1/s) =0.094(U
L
+0.1U
G
)
0.788
,
where U
L
and U
G
are liquid and gas supercial linear velocity
inside the channel in unit of cm/s, respectively.
Mini- or micro-structured catalyst beds are a promising cat-
alyst design approach at the reactor level. It can break up the
3506 W. Liu / Chemical Engineering Science 62 (2007) 35023512
0.01
0.10
1.00
10.00
1.0
a
k
L
S
,

1
/
s
Gas/Liquid = 1 v/v
gas/Liquid = 5 v/v
Gas/Liquid = 10 v/v
Trickle bed
Liquid superficial linear velocity,cm/s
0.1 10.0 100.0
Fig. 5. Liquid reactant mass transfer rate constant from bulk to channel surface
in 1-mm channel in co-current down ow (data from Liu et al., 2005).
compromise that has to be made in the conventional catalytic
bed design. It offers various design options to eliminate any
mass transfer resistance and achieve intrinsic catalytic perfor-
mance. In addition, it presents a much better way for integra-
tion of reactor-level catalyst designs with catalyst particle (or
thickness)-level designs.
3. Catalyst design at particle level
The particle-level catalyst design deals with catalyst func-
tions and material structures of a catalyst unit. The catalyst unit
is the catalyst particle for particle-shaped catalysts and is the
catalyzed wall for structured catalysts. Optimization of catalyst
particle designs, such as pore size, pore distribution, catalytic
material distribution in its carrier, etc., has been extensively
studied, which is not the focus of the present discussion.
Geometries and some physical properties of the catalyst par-
ticle are determined by the design requirements at the reactor
level. Design at the particle scale essentially revolves around
the effectiveness of a catalyst unit for desired reactions, which
can be rationalized by the following mass balance equations:
Inside catalyst unit: (D
i
C
i
) = r
i
, D
i
the molecular dif-
fusivity of species i in the pore structure, C
i
the concentration
of species i, and r
i
the disappearance rate of species i due to
reaction.
Boundary condition:
D
i
C
i
|
S
=k
BS
(C
i,B
C
i,S
).
The effective diffusivity is affected by the pore structure and
the reaction rate is determined by the content of the active cata-
lyst material. Thus, spatial variation of diffusivity and reaction
rate constant can be controlled by pore structures and catalyst
distribution inside the particle. The boundary condition presents
another design dimension, which is advocated in this work.
Mass transfer enhancement through reactor-level design does
not discriminate individual molecules. For an actual catalytic
process, selectivity or yield is as important as the reaction con-
version rate itself. It is desirable to enhance the main reaction
rate while minimizing side reactions. To achieve this objec-
tive, a membrane layer can be coated on the external surface
of a catalyst unit so that mass transfer rate of individual re-
actant into the catalyst or individual product molecule out of
the catalyst can be regulated. In this way, reactants can be de-
livered to the place where the reaction occurs or products can
be removed out of the catalyst as soon as they are produced.
Thus, such a catalyst design is about manipulation of bound-
ary conditions between the catalyst unit and bulk uid. The
membrane-enhanced catalyst design is particularly useful for
those catalytic reactions that are limited by thermodynamics.
An excellent example of such a catalyst design is selective
disproportionation of toluene to paraxylene over ZSM-5 cata-
lysts, which was commercialized in the 1990s. Paraxylene is an
important chemical feedstock for polyester production. ZSM-5
catalyst is known to be active for conversion of toluene to ben-
zene and xylenes. However, over a catalyst surface without any
mass transfer restriction at the molecular level, toluene dispro-
portionation produces benzene and an equilibrium xylene mix-
ture consisting of about 24% para xylene, 26% ortho-xylene,
and 50% meta-xylene, as shown below.
Making high purity paraxylene from a mixture of xylene iso-
mers is a fairly difcult separation problem because of their
similarity in boiling point and molecular size. Breakthrough in
this catalytic process was achieved by modifying the zeolite
catalyst such that paraxylene-enriched product is produced via
toluene disproportionation. Early experimental work and the-
oretical models on this subject were reviewed by Olson and
Haag (1983). Later development showed that greater than 95%
paraxylene selectivity can be obtained over the modied ZSM-
5 catalyst (Chang and Rodewald, 1996). Compared to 24%
paraxylene selectivity with an unmodied catalyst, the selective
process gives dramatic improvement to paraxylene yield and
has large impact on the capital and operation costs of paraxy-
lene manufacturing process.
A catalyst design model in Fig. 6 is proposed here to explain
the working principle of this process. The modication occurs
at the individual zeolite crystal level (0.1100 m size) and
serves two purposes. First, external acidic sites available for
catalytic reactions are passivated so that the catalytic reaction
is conned only inside zeolite pores. Second, a membrane-like
coating layer is formed on external surface of the zeolite crys-
tal so that only linear-shaped molecules can go through, while
branched molecules are blocked. As a result, toluene molecule
diffuses into the zeolite channel and is converted into ben-
zene and xylene inside the zeolite crystal, and only paraxylene
comes out of the zeolite channel while meta- and ortho-xylene
are hindered. The commercial success of selective toluene dis-
proportionation process demonstrated feasibility of membrane-
functionalized catalyst designs for large-scale catalytic reaction
processes.
W. Liu / Chemical Engineering Science 62 (2007) 35023512 3507
Reaction inside zeolite pore
Pore opening is fine-tuned to significantly reduce mass transfer rate of
branched molecules into and out of the pore relative to linear ones
Zeolite crystal
Post coating to passivate external
catalytic sites and modify pore opening
Fig. 6. Selective disproportionation of toluene to benzene and paraxylene over surface-modied ZSM-5 catalyst crystal.
Catalytic reaction
zone, Pt/TiO
2
Silicalite membrane only allows linear
molecules into and out of catalyst core
+H
2
Fig. 7. Membrane-functionalized catalyst particle for selective hydrogenation
reaction (Nishiyama et al., 2004).
Recent research results show application of such a cata-
lyst design concept to other reactions. For example, Nishiyama
et al. (2004) coated a Pt/TiO
2
hydrogenation catalyst particle
of meso-pore structures with a sillicalite zeolite membrane on
its external surface, shown in Fig. 7. The resulting catalyst al-
lows only linear-shaped molecules to come in and out of the
catalyst core. As a result, olens of linear shape in a mixture
can be selectively hydrogenated. Another example is selective
alkylation of toluene with methanol for paraxylene production,
which is a different reaction route from toluene disproportion-
ation. Miyamoto et al. (2006) reported nearly 100% selectivity
toward paraxylene by coating the ZSM-5 catalyst particle with
a silicalite membrane lm. The membrane allows toluene and
methanol molecules permeate into the ZSM-5 catalyst core, re-
act to formxylenes, but only allowparaxylene product molecule
permeate out of the catalyst particle.
In summary, incorporating membrane separation functions
into the catalyst unit enables regulation of mass transfer rate of
individual reactant and/or product molecules, and is promising
achieve a step change to the product yield by use of known
catalysis chemistries. The micro- or mini-structured catalyst
bed design presents a good platform for engineering interfaces
of the catalyst with bulk uid.
4. The 3-D catalyst design at nano-scale
Catalyst design at the reactor and particle-size level is largely
based on mass transfer and hydrodynamic principles. Enhance-
ment of intrinsic catalytic activity and selectivity requires im-
provement of the catalyst material itself. Discovery of new cat-
alyst compositions and/or active structures traditionally relies
on a researchers intuition and the trial-and-error experimen-
tal approach. Rational design of catalysts has been a long-term
endeavor in catalysis research.
The 3-D catalyst design strategy is presented here that calls
for shift from conventional 2-D view of catalyst surface or
active sites to actual 3-D identity.
4.1. Optimizing 3-D shape and/or size of a catalyst material
This design approach aims to enhance density of active sites
per unit mass (or volume) of a given catalyst material. Recent
progress has shown that dramatic improvement of catalytic ac-
tivity can be realized by manipulating the material composi-
tion and/or structure at a level of nanometer. Different from
homogenous catalysts that are molecules, individual atoms do
not comprise a solid catalyst. It is known that for a large cat-
alyst particle or crystal (such as detected by XRD), catalytic
3508 W. Liu / Chemical Engineering Science 62 (2007) 35023512
A
c
t
i
v
i
t
y

Catalyst structure size
Individual
atom
Large crystal
Clusters or amorphous particles in
nano-meter size
Activity may be
affected by ways how
atoms are assembled
Fig. 8. Postulated variation of catalyst activity with its structure size.
activity per unit volume decreases with increasing particle size.
However, catalytic behavior between the atom and crystal size
has been blurry. The present author hypothesizes variation of
catalytic activity with the size of catalyst structure in Fig. 8.
Contrary to the conventional concept that high catalyst activity
is associated with high dispersion of a catalyst metal, activity
may increase from the size of individual atoms to the size of
certain types of clusters. The critical size may be at the nanome-
ter level and not detectable by XRD. The catalytic activity of
clusters may be affected by the way how individual atoms are
assembled together or by the supporting environment.
Such a catalyst design concept is shown with recent research
progress about Pt catalysts. Instead of the trial-and-error exper-
imental approach, Lee and Cho (2003) applied quantum chem-
istry calculation toward identication of active Pt catalyst size
and shape. At the nanometer scale, the properties of a mate-
rial are found dictated by the arrangement of individual atoms.
A conguration of 611 Pt atoms that measures 3.1 nanometers
in diameter, shown in Fig. 9, is identied as the most stable
and active candidate. Recent development of a new genera-
tion of hydrotreating catalysts, trade named of Nebula (Pappal
et al., 2003), presents a commercial success in this direction.
The hydrotreating process has been in commercial use for many
decades and incremental improvement was made by optimiza-
tion of catalyst bed structures, catalyst particle shape, pore size
and pore size distribution, and catalyst loading. A step change,
four times of hydrodesulfurization activity of the conventional
supported catalyst, is obtained by nano-scale catalyst engineer-
ing. Instead of supporting Co and Mo catalyst on a -alumina
support, a composite of Co and Mo metal oxides is prepared
of desired composition and size (Soled et al., 2000).
The above two examples show that even with well-known
catalyst materials, signicant improvement to the activity is
possible by manipulating the material structures at a scale of
atomic clusters.
Fig. 9. The most active Pt cluster with 611 atoms from computer simulation
(www.nanostellar.com).
4.2. Vertical integration of catalyst layers
In most previous catalyst models, catalyst distributions and
reaction mechanisms were treated in a 2-D conguration, as
illustrated in Fig. 10(a). Depth of a catalyst layer is another
design dimension that should be explored. Catalyst materials
of different functions can be vertically integrated to achieve
unique performance attributes (Fig. 10(b)). Hu et al. (1996)
applied 3-D catalyst design concept to the development of Pd-
only three-way catalysts for automotive catalytic converter. As
shown in Fig. 10(c), two layers of catalyst materials of differ-
ent compositions are deposited on a substrate surface. In such
an arrangement, Pd oxidation activity and ceria oxygen stor-
age function are effectively utilized and the catalyst structures
are stabilized at the same time. As a result, the Pd-only cata-
lyst shows higher conversion activities for oxidation of hydro-
carbons, CO oxidation, and reduction of NO than the more-
expensive Pt/Rh catalyst.
4.3. Interface catalyst model of nano-composite
The catalyst design principle here is to create new catalytic
properties from combination of different materials, i.e., cre-
ation of synergism. A catalytic reaction involves complex re-
arrangement of individual atoms and requires both suitable
electronic and geometric congurations of the catalyst surface.
One individual material can provide certain catalytic functions
but may not be an effective catalyst by itself. Combination of
materials of different properties would generate rich catalytic
functions. Thus, different materials must be in intimate contact
to show concerted effect in catalyzing a chemical reaction.
The interface catalyst model is illustrated in Fig. 11 with
transition metaluorite oxide composite catalysts, such as Cu
or Au with ceria, which was developed by Liu et al. (1994a, b)
and Liu (1995) for exceptional oxidation catalysis activities.
It is noted that the interface catalyst model is a different cat-
alyst design concept from the strong metal support interac-
tion (SMSI) (Baker et al., 1986). The SMSI theory led to
W. Liu / Chemical Engineering Science 62 (2007) 35023512 3509
Support
2-D catalyst distribution 3-D catalyst distribution
Catalyst A Catalyst B
Catalyst B
Top layer : Pd-BMO-Al
Bottom layer : Pd-Ce-BMO-Al
Substrate
HC + CO + NO + O
2
CO
2
+ N
2
+ H
2
O
Low-T catalytic reaction
High-T catalytic reaction
3-D design of Pd-based catalyst (BMO = base metal oxie, Al = alumina): 1
st
layer composite catalyst designed for reactions at low temperature, 2
nd
composite
catalyst layer for high-T reactions
Support
Catalyst A
Permeable to reactants and/or products
Fig. 10. The 3-D catalyst design of different catalyst layers and example of Pd-based catalyst for automotive catalytic converter.
x
Material B: Cu or Au Material A: ceria
I
n
t
e
r
a
c
t
i
o
n

s
t
r
e
n
g
t
h
Distance from interface
Fig. 11. Interaction of two kinds of materials through interface.
an important revelation that catalysis properties such as dis-
persion of a catalyst metal could be altered by the composi-
tion and structure of the support material where the catalyst
metal sits on. However, catalysis function was still centered
on the catalyst metal itself in the SMSI model. By contrast,
composite of different materials as a whole is viewed as an
active catalyst conguration in the present interface catalyst
model. The concerted catalytic effect results from electronic
and chemical interactions through the interface. Thus, intimate
contacting of different materials at the nano-size level is nec-
essary to make the interface catalyst model work. Mechani-
cal mixing at the macro-scale results in an interfacial area too
small to show signicant degree of electronic and chemical
interactions.
In the present example of transition metaluorite oxide
composite catalysts, those two different kinds of materials com-
plement each other in electronic properties, oxygen mobility,
and surface chemistry. Its unique catalytic function can be well
illustrated with (Cu, Au)/ceria composite for CO oxidation. CO
is a major air pollutant generated by burning carbon-containing
fuels. CO oxidation is also a good model reaction to study ox-
idation reactions because of its relatively simple kinetics.
Ceria has one stable, face-centered-cubic (fcc) lattice struc-
ture from room temperature up to 2400

Cits melting point.


Gold and Cu oxides are not miscible with ceria. The Cu-ceria
and Au-ceria catalysts show substantially higher activity and
stability for CO oxidation than ceria, Cu, or Au catalyst alone
(Liu and Flytzani-Stephanopoulos, 1995a). At a space velocity
of 45,000 v/v/h, complete CO conversion over the Cu-ceria and
Au-ceria catalyst occurs at about 80 and 20

C, respectively.
The Cu-ceria catalytic activity was found comparable or supe-
rior to the Pt catalyst. Although Au/FeO
x
or Au/TiO
2
system
was reported for low-temperature CO oxidation in the earlier
research (Haruta et al., 1993), Au-ceria has simple crystalline
structure and high stability, presenting a nice model system
clearly demonstrating synergistic effect of a composite catalyst.
It is noteworthy that active catalyst conguration comprises
nano-sized copper or Au particles/clusters embedded in ce-
ria matrix (Liu and Flytzani-Stephanopoulos, 1995b). Small
amounts of Cu or Au (a few wt% or less) are sufcient to make
an active composite catalyst (Liu and Flytzani-Stephanopoulos,
1996; Fu et al., 2003), which is consistent with the model pro-
jection. Large interfacial area resulting from nano-sized Cu or
Au clusters enables electronic and chemical interactions be-
tween the two kinds of materials to occur to a large extent. As
illustrated in Fig. 11, such interactions are postulated to extend
3510 W. Liu / Chemical Engineering Science 62 (2007) 35023512
Table 2
Kinetic parameters of CuCeO
2
composite catalyst for CO oxidation (data
from Liu and Flytzani-Stephanopoulos, 1995b)
Catalyst Relative k A E
a
Q
at 150

C (kJ/mol) (kJ/mol)
Cu
0.15
[Ce(La)]
0.85
O
x
9.4 10
4
7.0 10
8
78 28
Cu
0.01
[Ce(La)]
0.99
O
x
As prepared
a
7.8 10
3
1.4 10
7
73 37
+3 h in air at 860

C 1.7 10
4
1.6 10
9
87 61
CuOCrO
x
/-Al
2
O
3
1 3.0 10
5
91 5
a
This catalyst was prepared by 4-h calcination in N
2
gas at 650

C, while
the other composite catalyst was typically calcined in air at 650

C.
beyond the interface and may decline exponentially with dis-
tance away from the interface. Degree of the interaction is
clearly reected in the reaction kinetics.
CO oxidation over the Cu-ceria composite catalyst shows the
following reaction kinetics:
rate
CO
2
=
k
s
K
CO
P
CO
P
0.1
O
2
1 +K
CO
P
CO
Table 2 lists the kinetic parameters of three CuCuO
composite catalysts in comparison to the CuOCrO
x
/alumina
catalyst. The CuOCrO
x
catalyst was regarded as the most
active base metal catalyst for CO oxidation at high temper-
atures prior to discovery of the present composite catalyst.
The heat of adsorption of CO on the CuCeO composite
catalysts, derived from K
CO
, is substantially higher than that
on the CuOCrO
x
/alumina. Relative reaction rate constants
at 150

C for the composite catalysts are about 45 orders of


magnitude higher than that for the CuOCrO
x
catalyst. The
activation energies for the composite catalysts, derived from
k
s
, are slightly lower than that of the CuOCrO
x
catalyst. It is
interesting to note that calcination of the 1 at% CuCeO com-
posite catalyst at a higher temperature increases its catalytic
activity. This is due to formation of more copper clusters from
isolated Cu atoms.
To further elucidate the impact of catalyst composition on the
intrinsic reaction rate of the CuCeO composite catalyst, the
catalyst weight-based reaction rate constant is normalized by
BET surface area and correlated with surface atomic fraction
of Cu/(Ce +Cu +La) as measured by XPS. In this way, com-
plication due to variation in BET surface area among different
catalysts is eliminated. A 3.4th powder order equation is found
to t well the reaction rate constants at different temperatures:
k
s
/S
g
S
3.4
Cu
,
where k
s
is the rate constant (mol/g s), S
g
the BET surface area
(m
2
/g), and S
cu
the Cu/(Cu +Ce +La) 100% measured by
XPS.
Such a relationship is against the conventional wisdom that
the rate constant is proportional to the surface dispersion or
coverage of the active catalyst metal. If the catalytic reaction
is only limited at the interface or limited to the area of the
copper clusters, the reaction rate constant is expected to have a
linear relationship with the surface coverage by Cu. The super-
linear relationship can be explained with the interaction model
as depicted in Fig. 11. If the copper cluster size is assumed con-
stant, the distance between clusters decreases with increasing
surface coverage by the cluster that the extent of the interac-
tion increases exponentially. Impact of the electronic conduc-
tivity on the ceria catalysis properties was clearly shown with
non-stoichiometric CeO
2
(Tschope et al., 1995). The non-
stoichiometric cerium oxide, having both electronic and oxygen
ionic mobility/conductivity, shows oxidation catalytic activity
orders of magnitude higher than the fully calcined cerium ox-
ide that has only oxygen ionic conductivity.
Since exceptional catalytic performance of this composite
system for several reactions was reported in the mid-1990s,
rapid growth in research activity has occurred for recent years.
Various aspects of the above catalyst system, preparation,
characterization, mechanistic study, and new reaction appli-
cation have been studied. In addition to CO oxidation, strong
synergism of the composite catalyst has been demonstrated for
several other reactions as summarized below for the Cu-
ceria, Au-ceria, and Ni-ceria catalyst systems studied so far
(Liu, 2005):
Hydrogen production-related applications:
Steam reforming of methanol: CH
3
OH +H
2
O
CO
2
+H
2
,
Partial oxidation of hydrocarbon: CH
4
+O
2
CO +H
2
,
Watergas shift (WGS) reaction: CO +H
2
O CO
2
+H
2
,
Selective CO oxidation: CO +O
2
CO
2
,
H
2
+O
2
= H
2
O.
Air pollution control-related applications:
CO combustion: CO +O
2
CO
2
,
Combustion of hydrocarbon: CH
4
+O
2
CO
2
+H
2
O,
Reduction of NO by CO: CO +NO CO
2
+N
2
,
Reduction of NO by NH
3
: NH
3
+NO H
2
O +N
2
,
Reduction of SO
2
to elemental sulfur: CO +SO
2
.
CO
2
+S
Strong synergism shown for those seemingly different reac-
tions by the Cu-ceria, Au-ceria, and Ni-ceria composite cat-
alysts is explained with two common fundamental attributes.
First, all the reactions involve one oxidizing molecule and
another reducing molecule. Second, all the reactions involve
transfer of oxygen atom from one molecule to another. Thus,
a unied, mechanistic model is proposed in Fig. 12. The re-
ducing molecule reacts with the surface oxygen and creates a
surface oxygen vacancy. Then, the oxidizing molecule transfers
its oxygen to the vacancy.
Cerium oxide provides several surface oxygen species avail-
able for different kinds of reactions:
O
2
e
O

2
e
O
2
2
O

e
O
2
lattice
.
For example, oxidation of methane may involve lattice oxy-
gen, while CO oxidation uses super oxide species (O

2
). The
oxidation/reduction cycle involves both electron transfer and
oxygen atom transfer. Cerium oxide has little electronic con-
ductivity. Addition of transition metal species, such as Cu, Au,
W. Liu / Chemical Engineering Science 62 (2007) 35023512 3511
Reducing molecule
(CO) reacts with surface
oxygen specie
Surface capping
oxygen vacancy
Oxidizing molecule
transfers oxygen to
vacancy (O
2
)
Fig. 12. Unied reaction mechanism for (Cu, Au, Ni)/ceria nano-composite catalyst.
and Ni, facilitates the electron transfer and also provides surface
sites for reducing molecules to adsorb. As a result, synergistic
effects are displayed for enhancement of the catalytic activity.
5. Summary
Multi-scale analysis and approach is effective to elucidate
impacts of design perspective at different scales to a catalytic
reaction process performance and open up new design dimen-
sions to break up the compromise that often needs to be made
with the conventional approach. A few promising catalyst de-
sign directions at catalyst bed level, catalyst particle level, and
intrinsic catalyst structure level are proposed and elaborated
with specic examples.
In the catalyst bed scale, micro- or mini-structured catalyst
designs enable decoupling of hydrodynamics and external mass
transfer from internal mass transfer and catalytic reactions. As
a result, all mass transfer resistance can potentially be mini-
mized and intrinsic catalytic performance is realized. It is par-
ticularly promising to gas/liquid multiphase catalytic reactions
that involve complex mass transfer steps and ow hydrody-
namics. The micro-structured bed design also provides a good
platform for integration with catalyst-particle level designs.
In the catalyst particle scale, design of bulk uid/catalyst
boundary condition is proposed to control mass transfer rate of
individual molecules into and out of the catalyst, which could
not be realized by catalyst bed level designs. For example, ad-
dition of a membrane coating layer onto external surface of
a catalyst unit enables dramatic improvement in reaction se-
lectivity that can be as important as the conversion rate for a
practical process. Commercial process of selective dispropor-
tionation of toluene to paraxylene over a modied zeolite cat-
alyst and recent research progress for other reactions are used
to show feasibility of this design strategy.
At the level of intrinsic catalyst structures, 3-D catalyst de-
sign approaches are outlined to outweigh limitations or con-
straints associated with the conventional 2-D catalyst model.
The 3-D catalyst design is illustrated by (i) optimizing size
and/or shape of a catalyst material at nano-scale to enhance den-
sity of active catalysis sites, (ii) integrating catalyst materials of
different functions in vertical direction, and (iii) incorporating
different kinds of catalyst material to form nano-composite cat-
alysts. Exceptional catalysis properties of the nano-composite
catalyst made of common catalyst materials are clearly demon-
strated with (Cu, Au, or Ni)/ceria catalyst system for a range
of reactions in hydrogen processing and air pollution control
application.
Acknowledgments
The author would like to thank Institute of Process En-
gineering, Chinese Academy of Sciences for promoting and
supporting studies of multi-scale science and technology.
Scientic discussion with Dr. Jinghai Li in the institute there
about understanding uidization and particle clustering from
1st principles was inspiring. The author is grateful to Profs.
Flytzani-Stephanopoulos and Adel Sarom for their mentorship,
and his colleagues in the industry for technical discussion and
cooperation on various catalyst and reactor technologies.
References
Afandizadeh, S., Foumeny, E.A., 2001. Design of packed bed reactors: guides
to catalyst shape, size, and loading selection. Applied Thermal Engineering
21 (6), 669682.
Baker, R.T.K., Tauster, S.J., Dumesic, J.A. (Eds.), 1986. Strong metal-support
interactions. American Chemical Society, Washington, DC.
Bell, A.T., 1990. Impact of catalyst science on catalyst design and
development. Chemical Engineering Science 45 (8), 20132026.
Chang, C.D., Rodewald, P.G., 1996. Selectivating zeolites with organosiliceous
agents. U.S. Pat. 5 516 736. Mobil Oil Corp.
Davis, M.E., 1994. Reaction chemistry and reaction-engineering principles in
catalyst design. Chemical Engineering Science 49 (24A), 39713980.
Fu, Q., Saltsburg, H., Flytzani-Stephanopoulos, M., 2003. Active nonmetallic
Au and Pt species on ceria-based water-gas shift catalysts. Science 301,
935.
Fulton, J.W., 1986. Selecting the catalyst conguration. Chemical Engineering
May 12, 97101.
Ge, W., Li, J.H., 2002. Physical mapping of uidization regimesthe EMMS
approach. Chemical Engineering Science 57 (18), 39934004.
Haruta, M., Tsubota, S., Kobayashi, T., Kageyma, H., Genet, M.J., Delmon,
B., 1993. Low-temperature oxidation of CO over Gold Supported on TiO
2
,
-Fe
2
O
3
, and Co
3
O
4
. Journal of Catalysis 144, 175192.
Hegedus, L., Pereira, C.J., 1990. Reaction engineering for catalyst design.
Chemical Engineering Science 45 (8), 20272044.
Hu, Z., Wan, C.Z., Lui, Y.K., Dettling, J., Steger, J.J., 1996. Design of a
novel Pd three-way catalyst: integration of catalytic functions in three
dimensions. Catalysis Today 30 (13), 8389.
Lee, B., Cho, K., 2003. Hierarchical multiscale study of metal nanoparticles.
40th Annual Technical Meeting of the Society of Engineering Science,
Oct. 1215, Ann Arbor, Michigan.
Li, J., Kwauk, M., 2003. Exploring complex systems in chemical
engineeringthe multi-scale methodology. Chemical Engineering Science
58, 521535.
Liu, W., 1995. Development of novel metal oxide composite catalysts for
complete oxidation reactions. ScD Dissertation, Massachusetts Institute of
Technology, Cambridge, MA.
3512 W. Liu / Chemical Engineering Science 62 (2007) 35023512
Liu, W., 2002. Mini-structured catalyst bed for gasliquidsolid multiphase
catalytic reaction. A.I.Ch.E. Journal 48, 15191532.
Liu, W., 2005. Catalyst technology development from macro-, micro- down
to nano-scale. China ParticuologyScience and Technology of Particles
3 (6), 383394.
Liu, W., Flytzani-Stephanopoulos, M., 1995a. Total oxidation of carbon
monoxide and methane over transition metal-uorite oxide composite
catalysts I. Catalyst composition and activity. Journal of Catalysis 153,
304316.
Liu, W., Flytzani-Stephanopoulos, M., 1995b. Total oxidation of carbon
monoxide and methane over transition metal-uorite oxide composite
catalysts II. Catalyst characterization and reaction kinetics. Journal of
Catalysis 153, 317332.
Liu, W., Flytzani-Stephanopoulos, M., 1996. Transition metal-promoted
oxidation catalysis by uorite oxides: a study of CO oxidation over Cu-
CeO
2
. Chemical Engineering Journal 64, 283294.
Liu, W., Sarom, A.F., Flytzani-Stephanopoulo, M., 1994a. Complete
oxidation of carbon monoxide and methane over metal-promoted uorite
oxide catalysts. Chemical Engineering Science 49, 48714888.
Liu, W., Sarom, A.F., Flytzani-Stephanopoulos, M., 1994b. Reduction of
sulfur dioxide by carbon monoxide to elemental sulfur over composite
oxide catalysts. Applied Catalysis B: Environmental 4, 167186.
Liu, W., Addiego, W., Sorensen, C., Boger, T., 2002. Monolithic reactor for
the dehydrogenation of ethylbenzene to styrene. Industrial Engineering
Chemistry Research 41 (13), 31313138.
Liu, W., Roy, S., Fu, X., 2005. Gasliquid catalytic hydrogenation reaction
in small catalyst channel. A.I.Ch.E. Journal 51, 22852297.
Miyamoto, M., Dung, V.V., Nishiyama, N., Egasshira, Y., Ueyama, K., 2006.
A silicalite lm coating on ZSM-5 catalysts for selective production of
p-xylene. Proceedings of the 9th International Conference on Inorganic
Membranes, June 2529. Lillehammer, Norway, pp. 341344.
Nijemeisland, M., Dixon, A.G., Stitt, E.H., 2004. Catalyst design by CFD
for heat transfer and reaction in steam reforming. Chemical Engineering
Science (UK) 59 (2223), 51855191.
Nishiyama, N., Ichioka, K., Park, D.H., Egashira, Y., Ueyama, K., Gora,
L., Zhu, W.D., Kapteijn, F., Moulijn, J.A., 2004. Reactant-selective
hydrogenation over composite sillicalite-1-coated Pt/TiO
2
particles.
Industrial & Engineering Chemistry Research 43 (5), 12111215.
Olson, D.H., Haag, W.O., 1983. Structure-Selectivity Relationship in Xylene
Isomerization and Selective Toluene Disproportionation. Chapter 14, ACS
Series Catalytic Materials, pp. 275307.
Pappal, D.A., Plantenga, F.L., Tracy, W.J., Bradway, R.A., Chitnis, G., Lewis,
W.E., 2003. Stellar improvements in hydroprocessing catalyst activity.
NPRA Annual meeting, AM-03-59, Mar. 2325, San Antonio, TX.
Sattereld, C.N., 1991. Heterogeneous Catalysis in Industrial Practice. 2nd
ed. McGraw-Hill,
Sattereld, C.N., zel, F., 1977. Some characteristic of two-phase
ow in monolithic catalyst structures. Industiral Engineering Chemical
Fundamentals 16 (1), 6167.
Soled, S.L., Riley, K.L., Schleicher, G.P., 2000. Hydroprocessing using bulk
Group VIII/Group VIB catalysts. U.S. Pat. 6, 162350.
Tian, Z., Xu, Y., Lin, L., 2004. Multi-scale phenomena in heterogeneous
catalytic processes-impact of chemical phenomena under the micro-scale
level on catalyst development and design. Chemical Engineering Science
59 (89), 17451753.
Tschope, A., Liu, W., Flytzani-Stephanopoulos, M., Ying, J.Y., 1995. Redox
activity of nonstoichiometric cerium oxide-based nanocrystalline catalysts.
Journal of Catalysis 157, 4250.
Ying, J.Y., 2006. Design and synthesis of nanostructured catalysts. Chemical
Engineering Science 61 (5), 15401548.

Você também pode gostar