Catalysis Today 69 (2001) 3–9

Monolith structures, materials, properties and uses夽

Jimmie L. Williams
Corning Incorporated, SP-DV-02, Corning, NY 14831, USA

Abstract
Extruded monolith substrates are widely used in automotive and stationary emission control reactors such as selective
catalytic reduction (SCR) units. Monoliths are increasingly being used, developed, and evaluated as catalyst supports in many
new reactor applications such as chemical and refining processes, catalytic combustion, ozone abatement, and others.
This paper gives a general overview of monolith fabrication, characteristics and typical use. Several commercial product
applications and new developments for use of monolith reactors in automotive, stationary and chemical industry are discussed.
© 2001 Elsevier Science B.V. All rights reserved.
Keywords: Monoliths; Honeycomb; Monolith reactor; Automotive emissions; Stationary emissions; Chemical process; Carbon; Gamma
alumina; Zeolites

1. Introduction Compositions based on cordierite possess a unique

Extruded monolith substrates are widely used in
automotive and stationary emission control reactors
for the selective catalytic reduction (SCR) of nitrogen
oxides. Monoliths are increasingly under development
and evaluation for many new reactor applications,
e.g. chemical process and refining industries, catalytic
combustion, ozone abatement, etc. [1–9]. Table 1 lists
various applications in which monoliths are used. In
automotive applications, ceramic monoliths are made
from synthetic cordierite, 2MgO·2Al2 O3 ·5SiO2 , a
material having a low thermal coefficient of expansion
[7]. Cordierite is a highly anisotropic crystalline phase
with concomitant high thermal expansion anisotropy
that leads to orientation during extrusion and low
overall expansion.
夽 For publication and presentation at the First International
Conference on Structured Catalyst and Reactors (ICOSCAR-1),
Delft, October 21–24, 2001.
E-mail address: williamsjl@corning.com (J.L. Williams).
combination of several critical characteristics: (a)
thermal shock resistance due to a low thermal ex-
pansion coefficient; (b) porosity and pore size distri-
bution suitable for ease of washcoat application and
good washcoat adherence; (c) sufficient refractoriness
because the melting point exceeds 1450◦ C; (d) suf-
ficient strength for survival in an automotive exhaust
environment; and (e) compatibility with washcoat and
catalysts [7].
The combination of high cell density, 31–186 cells/
cm2 (200–1200 cells/in.2 ), with thin walls, 0.051–
0.27 mm (0.002–0.0105 in.), give rise to low backpres-
sure in automotive exhaust systems. This combination
also yields high open frontal area (OFA), 72–87%,
which is a necessary condition for low backpressure.
Physical and chemical properties of cordierite raw
materials are controlled to a high degree necessary
for sophisticated quality control. For example, the
particle size of various commonly used raw materials,
silica, clay, alumina and talc, are closely controlled to
consistently yield the desired porosity and pore size
distribution as well as hold parts to close tolerances.

0920-5861/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 0 1 ) 0 0 3 4 8 - 0
4 J.L. Williams / Catalysis Today 69 (2001) 3–9
Table 1
Applications of cordierite cellular ceramic substrates
Automotive emission control
Diesel particulate filter
Stationary emission control
Woodstove combustor
Molten metal filter
Natural gas storage
Indoor air purification
Ozone abatement
Catalytic incineration
Industrial heat recovery
Ultrafiltration
Chemical process catalyst support
Water filtration
Examples of monolithic substrates are shown in
Fig. 1. The monolith parts itself can be produced in
a number of sizes and shapes, typically round or oval
cross-sectional areas for automotive applications, or
square for stationary emission uses. Cross-sectional
part diameters for single pieces up to 35 cm have been
produced commercially for heavy duty vehicle uses.
Much larger cross-sectional areas are made by stack-
ing blocks together for stationary emission reactor
systems. In addition to the overall part dimension, the
ρν 2
geometry of the monolith channels can be produced
in many forms including square, round, hexagonal,
and triangular.
Cell configurations and properties of monoliths are
described in terms of geometric and hydraulic param-
the monolith length (cm); ν the velocity in channel
eters [2,10]. These properties can be defined in terms
of cell spacing L, the distance measured from the
Fig. 1. Cellular ceramic monoliths.
center of one cell wall of a square channel to the next
wall, and wall thickness, t.
The cell density N is defined as the number of cells
per unit of cross-sectional area and is expressed in
units of cells per square inch (cpsi) or per square
centimeter.
1
N= 2 (1)
L
The OFA is defined as a function of wall thickness,
cell spacing and cell density, as shown in Eq. (2).
OFA = N (L − t)2 (2)
The hydraulic diameter defined by Eq. (3) decreases
as the cell density increases for a monolith. The
Dh = L − t (3)
hydraulic diameter is different for uncoated and
washcoated monoliths since washcoating with catalyst
or ceramic materials changes the wall thickness.
In designing monolithic catalysts, there is a balance
between geometric surface area and pressure drop. The
pressure drop across the monolith depends linearly on
flow velocity and length.
2f
P = ch (4)
Gc Dh
where f is the friction factor, dimensionless; Dh the
hydraulic diameter (cm); Gc the gravitational constant;
(cm/s); and ρ the gas density (g/cm3 ).
These fundamental equations allow one to design
monolith geometric parameters such as cell density
or wall thickness to meet the constraints of external
processing requirements such as space velocity, flow
rates, pressure drop, etc. For a more detailed discus-
sion see Refs. [1,2].
In the case of automotive catalytic converters, the
design and durability of advanced ceramics must pro-
mote fast light-off, improved conversion efficiency,
and reduced backpressure [1]. Current designs are fo-
cusing on thinner wall monoliths. Thermal shock re-
sistance and a low coefficient of thermal expansion
are important properties of automotive converter sub-
strates, and are a major reason why cordierite is the
ceramic material of choice for this application.
The success of cordierite monoliths as catalyst
substrates in the treatment of automotive exhaust
 J.L. Williams / Catalysis Today 69 (2001) 3–9
has given rise to interest in the chemical industry as
substrates for catalytic reactors. For many of these
applications, the use of other substrate materials and
designs can be utilized since the thermal shock re-
quirements are not as demanding as automotive. In
general, many inorganic catalyst support materials
used today in conventional chemical and refining ap-
plication can be extruded into a monolith form. This
has ignited interest in monolith reactors as potential
replacements for fixed bed and slurry reactors. Mono-
lith reactors offer the advantage of thinner walls, high
geometric surface area, low-pressure drop, good mass
transfer performance, and ease of product separation.
There may also be advantages in kinetics and mass
transfer in gas/liquid phase applications due to better
catalyst wetting, higher effectiveness factors, and thin
films that are conducive to high mass transfer rates.
However, the hydrodynamics and flow distribution of
multi-phase flow through a monolith reactor bed is
still not fully understood [3,4,11–13].
This paper gives a general overview of monolith
fabrication, characteristics and typical use. Several
commercial product applications and new develop-
ments for use of monolith reactors in automotive,
stationary and chemical industry are discussed.
2. Extrusion process
The extrusion of cordierite or other ceramic materi-
als into monolith shape follow the basic steps outlined
in Fig. 2 [7]. In the first step good mixing of raw
materials is necessary to insure complete reaction or
interaction of raw materials. Plasticizing with solvent,
usually water, is required to bring the batch materials
to a condition suitable for forming by extrusion. Ex-
trusion is done with a series of unique dies developed
for this process [14]. Uniform drying of monoliths is
Fig. 2. Fabrication of monolithic honeycomb substrates.
5
necessary to insure that enough moisture is driven off
without cracking the monolith body. In the final step,
the dry monolith part is calcined at elevated temper-
atures to complete solid state reactions and to obtain
the desired physical properties. For cordierite, calcin-
ing is conducted at 1400◦ C to transform the mixture
into a phase-oriented, low expansion sintered body.
Extruded carbon monoliths are fabricated in a sim-
ilar process except for the use of phenolic resins and
fillers in the extrusion step followed by curing, car-
bonizing, and activation steps [15].
3. Automotive
Future emission limits for gasoline- and diesel-
fueled vehicle require more efficient exhaust gas after
treatment devices. For gasoline engines, the catalytic
converter is one essential part of a complex system
design. As emission regulations become more strin-
gent, new developments in engine control, fuel man-
agement, catalyst technology, and advanced substrate
technology is needed.
The most widely used ceramic material for extruded
monolithic substrates is cordierite, 2MgO·2Al2 O3 ·
5SiO2 . For use as automotive catalytic converter,
cordierite has a set of unique properties including
high porosity, low thermal expansion, high thermal
shock resistance, and a tailored microstructure [10].
Advances in ceramic substrate technology are made
possible through judicious choice of cell shape and
size, wall porosity and microstructure of cordierite.
Table 2 lists some of the cell sizes and shapes of
monoliths that are generally used in commercial sub-
strates. Cell densities in the range 400–1200 cpsi are
of interest for automotive catalytic converters. The
trend is toward the use of ultra-thin-walled substrates
for automotive catalytic converters for light-off per-
formance. As part of the overall system design, more
durable washcoat formulations and high temperature
stable catalysts are needed to achieve longer lifetime
of the converter [1].
4. Diesel particulate filter
The diesel particulate filter (DPF) is the most ef-
fective method to remove soot particles from diesel
6 J.L. Williams / Catalysis Today 69 (2001) 3–9
Table 2
Monolith substrates of interest for automotive converters
Cell density (cpsi) Web thickness, 1 mil = 0.001 in.
25 (35 mil) 0.889 mm
50 (25 mil) 0.635 mm
100 (15 mil) 0.381 mm
200 (10.5 mil) 0.267 mm
300 (8 mil) 0.203 mm
400 (4 mil) 0.102 mm
600 (4 mil) 0.102 mm
900 (2 mil) 0.051 mm
engine vehicle exhaust with filtration efficiencies up
to 95% [10]. Regeneration of the filter is critical
for continuous engine and vehicle operation at low
backpressure. Controlled on-board regeneration of
diesel filters is an area that continues to be ex-
plored. Both passive and active techniques are under
investigation.
A DPF differs from automotive catalyst substrates in
that alternate channels are plugged on the entrance face
and the adjacent channels are plugged on the exit face
of the filter, forming a checkerboard pattern. Filtration
occurs when the particle laden exhaust gas is forced
through the wall between adjacent channels. Soot par-
ticles are trapped in the entrance channel and clean
exhaust passes out from the exit channel. Periodically
the soot is burned off, thereby cleaning the channel
and recovering backpressure. Cordierite and silicon
carbide are two materials that are commercially avail-
able for diesel exhaust after treatment. Development
of new materials for DPF type filters is in progress.
A high cell density cordierite DPF product, RC
200/19, was recently introduced at the 2001 SAE
conference held in Detroit. The geometry is 200 cpsi
with 0.019 in. (0.048 cm) wall thickness compared
to a standard EX-80 product having 100 cpsi with
0.017 in. wall (0.048 cm). RC 200/19 has a higher
thermal mass which leads to lower temperatures
during regeneration [16,17].
5. NOx adsorbers
The NOx storage catalyst was developed to meet
low emissions and fuel consumption targets for gaso-
line engine vehicles. The NOx adsorber is a washcoat
containing barium, potassium, or other alkali/alkaline
Geometric surface area (cm2 /ccm) OFA
6.51 0.681
9.17 0.678
13.39 0.723
18.98 0.725
23.51 0.742
27.09 0.846
31.01 0.787
43.11 0.834
earth metal oxides as adsorbents [18–20]. During
adsorption, a nitrate is formed with the alkali oxide
through conversion of NO to NO2 with O2 promoted
by platinum. It is known that potassium is a better
adsorbent than barium in high temperature conver-
sion of NOx , however, one problem with potassium
is its migration from the washcoat into the substrate
and subsequent loss of adsorption efficiency. Zeolites,
titanias, and silicas are being developed as barriers to
prevent potassium migration and sulfur poisoning of
the NOx adsorber [19].
6. Chemical process industry applications
Successful use of cordierite monolith catalysts in the
treatment of automotive exhaust has given rise to their
use in other catalytic applications. For many of these,
the use of other substrate materials and designs can be
utilized since the requirements are not as demanding
as automotive. This has resulted in an increase in the
potential use of monoliths in chemical and refining
catalyst applications.
Monolith catalyst substrates can be either metal or
ceramic. They can be washcoated with a high surface
area catalyst support material and impregnated, or ex-
truded directly into catalytic bodies using appropriate
batch materials. The use of monoliths in multi-phase
reactors is of interest as possible replacements for
slurry and fixed bed reactors. Monolith reactors offer
the advantage of thinner walls, high geometric surface
area, low pressure drop, good mass transfer, and ease
of product separation [4,8,11–13].
Hydrogen peroxide is presently the only large-scale
industrial application of monolithic catalysts in a
 J.L. Williams / Catalysis Today 69 (2001) 3–9
multi-phase process [21]. Anthraquinone is used to
facilitate the selective oxidation of hydrogen to hy-
drogen peroxide indirectly using a chemical redox ap-
proach in a two-step process. The quinone reduction
step utilizes a monolith catalyst containing palladium
on a washcoat in a gas/liquid multi-phase reactor.
Schanke et al. [24] describe the use of a multi-phase
monolith reactor system for Fischer–Tropsch synthe-
sis in a recent US patent. The monolithic catalyst
is comprised of ␥-alumina, silica, zeolite, or titania
and a conventional Fischer–Tropsch catalyst com-
bined with a precious metal promoter. The strongly
exothermic nature of Fischer–Tropsch synthesis re-
quires effective heat transfer for successful reactor
operation.
Vinyl acetate is synthesized by the vapor phase re-
action of ethylene, oxygen, and acetic acid over a
monolith catalyst. The monolith catalyst consists of Pd
and Au deposited on silica-coated cordierite or mullite
monoliths. Catalyst space–time yields for vinyl acetate
production of 434 g/l h at 192◦ C have been reported
[22,23].
Addiego et al. [25] have reported the extrusion of
iron oxide honeycomb catalysts for the vapor phase
dehydrogenation of ethyl benzene to styrene. They
indicate styrene selectivity >90% and ethyl benzene
conversion greater than 60%. The low pressure drop
characteristic of monolith shapes for this equilibrium
limited reaction is an important benefit.
Unreacted o-xylene and other intermediates are con-
verted to phthalic anhydride (PA) in a post-reactor
containing a monolith catalyst located downstream of
the main PA reactor in the Wacker low energy phthalic
anhydride process. The use of the proprietary catalyst
in the post-reactor provides higher overall conversion
to PA and increases operational flexibility around the
main reactor [26].
Table 3
Physical properties of coated carbon monoliths
Sample Surface Pore volume
area (m2 /g) t-Micro Meso Macro
1 698 0.27 0.32 0.02
2 1041 0.29 0.81 0.03
3 863 0.17 0.91 0.07
4 828 0.04 1.30 0.08
7
7. Carbon
Carbon monoliths have potential for use in a wide
variety of applications for which activated carbon
bodies are suited. Examples include residential wa-
ter purification, volatile organic compound emission
control, natural gas fuel storage for gas-powered
vehicles or equipment, indoor air purification, indus-
trial respirators, automotive cabin air filters, ventless
hoods, chemical separations, NOx , and SOx , control,
and exhaust traps for automotive cold start emissions
[15,19,27–30].
Preparation of carbon monoliths is either by
full-body extrusion or as a coating on cordierite
substrates. Activated carbon-coated monoliths are
fabricated using high yield polymeric phenolic resins
having low viscosity. A pore former polymer is added
for mesopore formation (30–100 Å). The manufac-
turing process includes dipping a porous ceramic
monolith in a resin, drying, curing, carbonizing in
nitrogen, and activating in carbon dioxide or steam.
Gadkaree and others have described more details of
this process elsewhere [15].
Table 3 shows the physical properties of activated
monoliths prepared by the above method [27]. At least
50% of the pores are in the mesopore range. Surface
areas range from 700 to 1000 m2 /g.
Activated carbon monolith catalysts are prepared by
the incipient wetness impregnation method or by in-
clusion of catalyst functionality in the polymeric resin
during process steps [27].
Extrusion of full-body carbon monoliths is simi-
lar to that described in Fig. 2. The basic process in-
volves forming a mixture of a thermosetting resin,
which can be either a solid or liquid resin and a filler,
which can be carbonizable or inorganic. The mixture
is extruded into a monolith, dried, and the resin is
Total pore Fraction Fraction Fraction
volume (cm3 /g) micro meso macro
0.61 0.45 0.52 0.04
1.12 0.26 0.72 0.02
1.16 0.15 0.79 0.06
1.41 0.03 0.92 0.06
8 J.L. Williams / Catalysis Today 69 (2001) 3–9
cured. The cured monolith is carbonized and activated
[15,18].
8. Zeolites
Zeolites, both natural and synthetic, have found
wide use for removal of NOx from coal-fired utilities
and gas turbine catalysts for SCR of NOx using NH3 .
Zeolites can be combined with vanadia/titania to pro-
duce a catalyst with a wide DeNOx operating window
that is more resistant to deactivation [7].
Zeolite catalysts can be extruded or coated onto
metal or ceramic substrates. Again, the extrusion
process is the same as that outlined in Fig. 2. The
coating process involves preparation of a zeolite
slurry using a binder, usually colloidal silica or alu-
mina, dipping the monolith, then drying and firing.
Ion exchange of zeolite can be done before or after
coating and/or extrusion. One potential disadvantage
of coated and extruded zeolites is the use of binder
that can block the zeolite pores and active metal sites.
As such, the amount of binder should be kept to
a minimum.
The in situ growth of a zeolite on a substrate is one
way to prevent blockage of pores and active material.
The substrate is covered by a thin layer of synthe-
sis mixture or immersed in a liquid that contains
the mixture inside an autoclave. Patil [31] discusses
preparation of mixtures of mordenite and MFI type
zeolites on cordierite support. The main disadvan-
tages of in situ growth of zeolites are complexity
compared to washcoating and growth of thin layer of
zeolite on the substrate surface. The main advantage
is complete accessibility of reactants inside the zeolite
pores.
9. γ-Al2 O3 monoliths
␥-Al2 O3 is used in many applications (e.g. au-
tomotive catalytic converter) as a high surface area
washcoat on the channel walls of monolith substrates
[1]. However, for many applications, porous wash-
coats are inadequate and catalyst supports made up
mostly or entirely of active high surface area ␥-Al2 O3
are used. Such applications include processes wherein
the kinetics of the chemical reaction on the catalyst
are slow relative to the diffusion and mass transfer
steps involved in the overall process.
Extruded monoliths of ␥-Al2 O3 consists of a mix-
ture of alumina powder, organic binder, and precur-
sors such as colloidal alumina solution as a permanent
binder. Narrow pore size distribution and high strength
are attained by proper selection of starting oxides and
processing variables. High strength and surface area
are needed to maintain structural integrity of the sup-
port in a reactor. Low to moderate firing temperatures
are required to preserve the high porosity and BET
surface area of the starting materials [32,33].
Many potential monolith applications require a
high BET surface area for effective catalyst function
and have 50–300 m2 /g. Higher BET surface areas of-
fer the possibility for a more compact reactor, which
could lead to significant cost reductions for the overall
reactor system [1].
10. Conclusions
Extruded monolith catalyst substrates for automo-
tive catalytic converters will continue to use cordierite
as the material of choice. Cordierite offers unique
properties in porosity, strength, thermal shock, ther-
mal expansion and robustness required for auto ex-
haust treatment. Advances in ceramic processing have
led to the development of higher cell density mate-
rials with improved strength and thermal properties.
With proper selection of raw materials and process-
ing control of microstructure, pore size distribution,
strength and other physical properties can be tailored
for specific applications.
The success of cordierite as catalyst support for
treatment of automotive exhaust emissions has given
rise to considerable interest in the chemical industry.
For many applications, the use of other materials and
designs can be utilized, since the requirements are
not as demanding as automotive. Hydrodynamics of
multi-phase flows in monolith beds need to be better
understood for wide use of monoliths in the chemi-
cal and refining industries. In particular, design will
be a key parameter to control the fluid flow inside
monolithic reactors and hydrodynamic knowledge is
being generated. Monolith reactors have potential to
replace fixed bed and slurry reactors for certain types
of reactions.
 J.L. Williams / Catalysis Today 69 (2001) 3–9
References
[1] R.M. Heck, S. Gulati, R.J. Farrauto, Chem. Eng. J. 82 (2001)
149–156.
[2] J.P. Day, L.S. Socha, SAE Paper No. 910371, 1991.
[3] A. Cybulski, J.A. Moulijn, Structural Catalysts and Reactors,
Marcel Dekker, New York, 1998.
[4] F. Kaptejn, J.J. Heiszwolf, T.A. Nijhuis, J.A. Moulijn, Cattech
3 (1999) 24–41.
[5] F.M. Dautzenberg, M. Mukherjee, Chem. Eng. Sci. 56 (2001)
25–267.
[6] R.M. Heck, M. Durilla, A.G. Bouney, J.M. Chen, in: Procee-
dings of the 81st Annual Meeting of Air Pollution Control
Association, Dallas, TX, 1988.
[7] I.M. Lachman, J.L. Williams, Catal. Today 14 (1992) 317–
329.
[8] T.A. Nijhuis, M.T. Kreutzer, A.C.J. Romijn, F. Kapteijn, J.A.
Moulijn, Chem. Eng. Sci. 56 (2001) 823–829.
[9] T.A. Collins, W.A. Cutler, D.L. Hickman, A.N. Mack, US
Patent 6 159 363 (2000).
[10] T.F. Cash, J.L. Williams, U.H. Zink, SAE Brazil Paper No.
982927, 1998.
[11] P.J.M. Lebens, Development and design of a monolith reactor
for gas–liquid countercurrent operation, Ph.D. Thesis, Delft
University of Technology, Delft, 1999 (in English).
[12] Th. Vergunst, Carbon-coated monolith catalysts — Prepara-
tion aspects and testing in the three-phase hydrogenation
of cinnamaldehyde, Ph.D. Thesis, Delft University of
Technology, Delft, 1999 (in English).
[13] P.M. Machado, D.J. Parrillo, R.P. Boehme, R.R. Brokuis, US
Patent 6 005 143 (1999).
[14] R.D. Bagley, US Patent 3 790 654 (1974).
9
[15] K.P. Gadkaree, Carbon 36 (7–8) (1998) 981–989.
[16] G.A. Merkel, D.M. Beall, D.L. Hickman, M.J. Vernacotola,
SAE Paper No. 2001-01-0193, 2001.
[17] G.A. Merkel, W.A. Cutler, C.J. Warren, SAE Paper No.
2001-01-0190, 2001.
[18] J.R. Asik, D.A. Dobson, G.M. Meyer, SAE Paper No.
2001-01-1299, 2000.
[19] K. Iwachida, et al., SAE Paper No. 2001-01-1298,
2001.
[20] S. Kojima, SAE Paper No. 2001-01-1297, 2001.
[21] C.T. Berglin, W. Hermann, US Patent 4 552 748 (1985).
[22] I. Nicolau, P.M. Colling, L.R. Johnson, US Patent 5 705 679
(1998).
[23] I. Nicolau, P.M. Colling, L.R. Johnson, US Patent 5 854 171
(1998).
[24] D. Schanke, E. Bergene, A. Holmen, US Patent 6 211 255
(2001).
[25] W.P. Addiego, W. Liu, T. Boger, ICOSCAR, Delft, October
2001, Catal. Today 69 (2001) 25.
[26] J. Eberle, J. Breimair, H. Domes, T. Gutermuth, Gas Technol.
(June 2000).
[27] Y.L. Peng, T. Tao, J.L. Williams, in: Proceedings of the 24th
Biennial Conference on Carbon, 1999.
[28] K.P. Gadkaree, US Patent 5 820 967 (1996).
[29] D.C. Bookbinder, E.M. De Liso, R.E. Johnson, K.P. Streicher,
US Patent 5 389 325.
[30] D.M. Beall, D. Chalasani, G.A. Merkel, Y.L. Peng, US Patent
6 207 101.
[31] M.D. Patil, J.L. Williams, US Patent 5 248 643 (1993).
[32] K.R. Brundage, S.H. Swaroop, European Application,
PCT/US99/29510 (1999).
[33] I.M. Lachman, L.A. Nordlie, US Patent 4 631 267 (1986).
 Catalysis Today 69 (2001) 11–15

A new technique for preparing ceramics for catalyst support

exhibiting high porosity and high heat resistance
Toshiyuki Yokota∗ , Yasuyuki Takahata, Tetsuo Katsuyama, Yoshihiro Matsuda
Department of Chemistry and Chemical Engineering, Yamagata University, 4-3-16, Jonan, Yonezawa, Yamagata 992-8510, Japan

Abstract
A new technique for preparing magnesia ceramics of high porosity and high temperature resistance has been developed.
Spray freeze drying of magnesium sulfate aqueous solution produced fine salt particles having open pores due to sublimation
of ice crystals. The particles were calcined to porous magnesium oxide and formed a green body. Highly porous magnesia
was produced by firing the green body. The porous magnesia exhibited a bimodal pore size distribution of macro-pores of
micron order and meso-pores smaller than 100 nm. Porosity was 87–90%. After addition of an aluminum additive with an
amount 3–5 mol%, the magnesia exhibited high heat resistance; surface area was greater than 20 m2 g−1 after 20 h exposure
in a 1573 K oven. Thus, the porous magnesia is expected to be very suitable for combustion catalyst support used in a high
temperature environment. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Catalyst support; High heat resistance; High porosity; Freeze drying

1. Introduction Si-stabilized alumina [4] have been developed as

Porous ceramics of open-cell structure have been
used in a wide range of applications, such as molten-
metal filters, diesel engine exhaust filters, industrial
hot-gas filters, and catalyst supports. In the present
study, a novel technique based on freeze drying has
been developed for producing high porosity, high heat
resistance ceramics. High porosity ceramics having
open pores are considered suitable for catalyst sup-
ports, because of low mass transfer resistance. High
heat resistance is a property required of combus-
tion catalyst supports, especially for application in
a high temperature range. Stabilized alumina-added
La and/or Ba added to alumina matrix [1–3] and
∗ Corresponding author. Present address: Department of Materi-
als Science and Engineering, 4-3-16 Jonan, Yonezawa, Yamagata
992-8510, Japan. Tel.: +81-238-26-3155; fax: +81-238-26-3130.
E-mail address: yokota@chem.yz.yamagata-u.ac.jp (T. Yokota).
inexpensive, stable, and durable support materials.
Such a support can withstand continuous operation at
temperatures between 1200 and 1300 K without ex-
cessive sintering.
In this study magnesia was selected as base ma-
terial, because it exhibits no crystal phase change at
high temperature. Addition of a second metal to the
bulk magnesia to suppress sintering was examined.
The porous magnesia was made of porous particles
prepared by freeze drying.
Freeze dry processing was first applied to ceramics
powder preparation by Schnettler et al. [5]. Yokota
and Ohto [6] developed a wetted-column-type freezer
for a spray freeze dry method which can produce fine
porous particles. In their method, metal salt porous
particles are calcined into porous oxide particles. By
forming the oxide particles into a green body and firing
it, a high porosity ceramic having open pores can be
produced.

0920-5861/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 0 1 ) 0 0 3 4 9 - 2
12 T. Yokota et al. / Catalysis Today 69 (2001) 11–15

2. Experimental in order to obtain porous magnesia with another metal

component.
Magnesium sulfate aqueous solution (2 mol dm−3 ) For characterization, the porous materials were
was prepared as a starting feed solution for freeze subjected to SEM observation (Nihon Denshi, JSM-
dry processing. The solution was sprayed through T330), measurement of porosity, pore volume, and
a two-fluid nozzle running nitrogen gas in a wetted pore size distribution (Yuasa-Ionics, ASP 60), spe-
wall-column freezer [6]. Fine droplets of the aque- cific surface area (by the BET method), and X-ray
ous solution were trapped by a refrigeration agent diffraction pattern analysis (Shimadzu, XD-D1).
of n-hexane cooled at 213 K, and solidified instan-
taneously into frozen particles. Average diameter of
the frozen particles was 42 ␮m. The frozen particles 3. Results and discussion
were transferred into a vacuum chamber, and its tem-
perature was controlled so as to sublimate the ice Fig. 1 shows SEM micrographs of the freeze-dried
crystals. The traces of the sublimation inside the dried and calcined particles. Spray freeze-dried particles
particles became open pores having a diameter of the
micron order.
The dried particles were calcined into magnesium
oxide in a tubular-type oven. The rate of tempera-
ture increase (10 K min−1 ), the highest temperature at-
tained (1473 K), and the vacuum atmosphere within
the oven were selected in order to attain large surface
area in a porous particle.
The porous oxide particles were formed to a cylin-
drical green body of 10 mm diameter and 3 mm thick-
ness, using ethyl alcohol (8 × 10−3 dm3 per 1 g oxide
particle) as the binder. A rather low compression pres-
sure (1.25 MPa) was used in the formation, so as to
not crush the pores in oxide particles and to obtain a
green body of uniform cylindrical structure.
Firing of the green body was carried out in an oven
containing an air atmosphere. The rate of temperature
increase was 10 K min−1 , and the attained tempera-
tures were controlled within a range 1573–1773 K. In
order to test heat resistance, the fired bodies were set
in an oven of 1573 K and held in the oven for up to
20 h. The effect of water vapor on sintering was also
examined in a tubular oven through which humidified
air flowed.
A second metal was added in order to suppress
sintering of the magnesia, so as to obtain a durable
combustion catalyst support. Aluminum, zinc, and
chromium metals were used as the additive. A cer-
tain volume of the additive metal sulfate solution was
added to the feed of magnesium sulfate aqueous solu-
tion. The two metal components were well mixed in
the feed solution and instantaneously solidified in the
freezer without segregation. The mixed frozen parti- Fig. 1. SEM observation of spray freeze-dried magnesium sulfate
cles were treated in the same manner described above particles and calcined magnesium oxide particles.
 T. Yokota et al. / Catalysis Today 69 (2001) 11–15
Fig. 2. Pore size distributions of freeze-dried magnesium sulfate particles and calcined magnesium oxide particles.
were porous spheres, and their average diameter was
42 ␮m. Calcined particles also retained a spherical
shape, but their size was reduced to an average diam-
eter of 20 ␮m. The porous structure in the particles
still remained.
Fig. 2 shows the pore size distributions as measured
by a mercury porosimeter. The freeze-dried particles
contain, as main pores, macro-pores in the order of
1 ␮m formed through ice crystal sublimation. The
calcined particles show a bimodal pore distribution.
The macro-pores are from the dried particle pores,
and other pores of sub-micron scale appeared after
thermal decomposition of metal sulfate salt into metal
oxide. Micro-pores formed through SOx gas ejection
must have enlarged into meso-pores during the calci-
nation process. Particles calcined in air contain larger
macro- and meso-pores as compared with those cal-
cined in vacuum. SOx gas ejected by thermal decom-
position of sulfate is easily evacuated from particles
in vacuum. In air, SOx gas might expand pores be-
fore escaping from particles. The smaller meso-pores
contribute to the large surface area of MgO particles.
Thus, the particles calcined in vacuum were used for
forming green bodies.
Cylindrical green body samples were fired in an
oven controlled within a range 1573–1773 K. In or-
der to examine heat stability, the fired samples were
placed in an oven controlled to 1573 K and held therein
13
for several hours. Fig. 3 shows the change in specific
surface area as a function of holding time. In the figure,
“fresh sample” means one that has not been subjected
to a stability test. The results indicate that considerable
sintering occurred for these samples. In most of the
samples, surface area was reduced to about 10 m2 g−1
after a 3 h test.
A second metal was added to magnesia in an effort
to attain high heat resistance. The aluminum addi-
tive worked effectively to suppress sintering. Fig. 4
illustrates the relationship between specific surface
area and holding time, and that between porosity and
holding time for samples with 3, 5 mol% aluminum
additive fired at 1573 K. The samples exhibit fairly
high surface area above 20 m2 g−1 and high poros-
ity above 86% even after 20 h exposure in an oven
maintained at 1573 K. In humidified air, the samples
with aluminum additive show high heat stability. The
samples of 1 mol% aluminum content show smaller
surface areas than do those displayed above, and
10 mol% content samples show almost the same
heat-resistance as illustrated in Fig. 4.
The X-ray diffraction pattern of the magnesia sam-
ple with aluminum additive shows small MgAl2 O4
spinel peaks among the MgO peaks. When the spinel
phase was formed at grain boundary of bulk mag-
nesia, this second boundary phase impeded diffusion
of bulk atom and prevented grain growth. Thus, the
14 T. Yokota et al. / Catalysis Today 69 (2001) 11–15

Fig. 3. Relationship between specific surface area and holding time in a 1573 K oven for magnesia samples without the second metal additive.

magnesia with aluminum additive exhibits high heat area as a function of holding time in the oven is almost
resistance. the same as that shown in Fig. 3. Zinc metal does not
The effect of zinc metal additive was examined, but form a spinel compound with magnesium. Therefore,
heat stability was not improved. The change in surface zinc atoms fused in bulk magnesia formed a solid

Fig. 4. Relationship between specific surface area, porosity and holding time in a 1573 K oven for magnesia samples with aluminum additive.
 T. Yokota et al. / Catalysis Today 69 (2001) 11–15
solution. When chromium metal was added in mag-
nesia, the fresh sample exhibited small surface area;
i.e., below 10 m2 g−1 . Therefore, further heat stability
testing was not carried out.
4. Conclusions
Porous heat-stable magnesia was prepared from
porous MgO particles produced by spray freeze dry-
ing. The magnesia shows a bimodal pore size distri-
bution of macro-pores in the order of microns and
meso-pores under 100 nm. Porosity was 87–90%. The
magnesia with 3–5 mol% aluminum additive exhibits
high heat resistance, with specific surface area be-
ing greater than 20 m2 g−1 after 20 h exposure in a
1573 K oven. Thus, the magnesia is expected to be
very suitable for combustion catalyst support used in
a high temperature operation.
15
Acknowledgements
Financial support for this work was provided by
Yoshino Gypsum Co., Ltd. and a Grant-in Aid for
Scientific Research (No. 08555181) from the Ministry
of Education, Science and Culture of Japan.
References
[1] S. Matsuda, A. Kato, M. Mizumoto, H. Yamashita, in:
Proceedings of the Eighth International Congress on Catalysis,
Berlin, Vol. 4, 1984, p. 879.
[2] S. Matsuda, Ceram. Jpn. 20 (1985) 189–195.
[3] M. Machida, K. Eguchi, H. Arai, Chem. Lett. (1986)
151–154.
[4] A.F. Ahlstrom-Silversand, Appl. Catal. A 153 (1997) 157–175.
[5] F.J. Schnettler, F.R. Monforte, W.W. Rhodes, Sci. Ceram. 4
(1968) 79–90.
[6] T. Yokota, Y. Ohto, Kagaku Kogaku Ronbun. 15 (1989) 877–
880.
 Catalysis Today 69 (2001) 17–24

Micro-engineered catalyst systems: ABB’s advancement in

structured catalytic packings
Robert E. Trubac a , Frits M. Dautzenberg a,∗ , Timothy A. Griffin b , Bettina Paikert b ,
Verena R. Schmidt b , Rudolf A. Overbeek a
a ABB Lummus Global Inc., 1515 Broad Street, Bloomfield, NJ 07003-3096, USA
b ABB Corporate Research Ltd., CH-5405 Baden-Dätwil, Switzerland 1

Abstract
ABB has advanced catalysis with micro-engineered catalyst (MEC) systems by providing a uniquely small particle size
on a formable catalyst support through the integration of catalysis and reaction engineering. A mechanically strong catalytic
web of micro-fibers has been engineered and shaped utilizing both computational fluid dynamics (CFD) and cold flow
experiments to optimize flow characteristics. This article discusses techniques used for the development of novel catalytic
structured packings for catalytic distillation applications. CFD models (verified through experiments performed on small-sized
structures) were shown to be of great utility in screening new structure ideas. Results will illustrate achievement of both high
gas–liquid contacting and bulk mixing at low pressure drop with the potential to provide enhanced catalyst utilization by
taking advantage of the intrinsic MEC properties, particularly its high porosity and exposed geometric fiber and catalyst
surface area. This was shown by the successful testing of one of these catalyzed structures in the selective hydrogenation of
C4 acetylenes. © 2001 Published by Elsevier Science B.V.
Keywords: Catalytic packings; Structured catalysts; Catalyst; Structured packings; Catalytic distillation

1. Introduction Catalysts are usually made as millimeter-sized par-

In recent years there has been much emphasis on
“process intensification”, i.e. reduction in the physical
size and/or cost of a chemical plant at a given produc-
tion rate [1]. In many processes the reaction section is
the major size/cost component. This has placed focus
on improving the integration of catalyst and reactor
systems not only at the molecular level, but also by
concentrating on the interaction between chemical
reaction and heat- and mass-transfer properties.
1 Currently
ALSTOM Power Technology.
∗ Corresponding
author. Tel.: +1-973-893-3319;
fax: +1-973-893-2745.
E-mail address: frits.m.dautzenberg@us.abb.com
(F.M. Dautzenberg).
ticles with the active catalytic agents dispersed on the
surface of/or throughout the particle. Reactants then
flow through a reactor containing the bulk-loaded
catalyst. Arrangement, shape and size of the cata-
lyst govern the flow dynamics and pressure drop.
The flow pattern is random making fluid dynamics
and heat-management difficult to predict and control.
The effective use of active catalytic agents is some-
times inefficient due to the occurrence of internal
mass-transfer limitations. These constraints result in
higher raw material and operating costs as well as the
production of less valuable by-products and waste.
ABB has invented a type of catalyst system
that maximizes both internal and external catalyst
mass-transfer while minimizing overall hydrodynamic

0920-5861/01/$ – see front matter © 2001 Published by Elsevier Science B.V.

PII: S 0 9 2 0 - 5 8 6 1 ( 0 1 ) 0 0 3 5 0 - 9
18 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24
constraints. Such “micro-engineered catalyst (MEC)
systems” consist of a mechanically strong catalytic
web of micro-fibers — thinner than a human hair —
in which microscopically small catalyst particles are
entrapped [2]. Fig. 1 shows two current MEC tech-
nology concepts. In both, a highly porous metal sheet
is used as a rigid catalyst support for micron-sized
catalyst particles. MEC systems achieve their advan-
tage by effectively utilizing catalyst particles on a
much smaller scale than currently applied in industry.
The size of the catalyst particles used in an MEC is
typically not greater than 50 ␮m, i.e. about two orders
of magnitude size reduction, thereby greatly reducing
internal mass-transfer limitations. The supports may
be formed into structures tailored for a specific appli-
cation, thereby optimizing the functions of catalysis
and reaction engineering and thereby also limiting
external mass-transfer limitations.
This paper describes the ongoing development of
catalytic structured packings for applications requiring
management of gas/liquid, liquid/solid and gas/solid
mass-transfer by employing computational fluid dy-
namics (CFD) in concert with detailed hydrodynamic
measurements. The development of structured MEC
catalyst packings for two-phase applications is ex-
plained in detail. One example of such an application
is catalytic distillation, which requires the integra-
tion of a chemical reaction and appropriate heat- and
mass-transfer characteristics with the goal of maxi-
mizing separation and reaction efficiencies [3,4].
2. Structured packing design — current
state-of-the-art
The generally accepted design criteria for non-
catalytic structured distillation packings are:
• a large surface area for high interfacial contact,
• an open structure for low resistance to the flow
(both phases),
• uniform liquid distribution across the surface,
• uniform vapor flow in the column cross-section
[5,6].
Surface characteristics to improve spreading of
the liquid on the surface have been found to be
of critical importance for enhancing mass-transfer.
Liquid spreading is effected via perforations and
embossments, which also enhance mass-transfer due
to turbulence in the liquid film improve “surface
wetting” characteristics (low surface tension) of the
packing material, and act to cause coalescence and
break-up of liquid droplets [6–11]. It was further found
that with porous materials, capillary effects have a
negative effect on liquid hold-up [7,12–14]; however,
liquid spreading and wetting, particularly at low liquid
flows are enhanced due to the capillary forces [12].
To achieve good mixing the flow should be
three-dimensional, encompass a high turbulence in-
tensity, have a high mean velocity gradient, and
exhibit a frequent change of direction of the main
streams [15,16]. High shear is desired, and created
by flow streams of different orientations crossing or
touching each other. A preferred packing geometry
should spread the flow in a three-dimensional manner
and should not have predominant characteristics in
specific directions.
3. Experimental
Initially, a series of experimental techniques was
applied to benchmark existing commercial packings
in single- and two-phase environments. The self-made
packings were manufactured from various materials,
viz. metal sheet, gauze (wire diameter 0.2 mm, open-
ing 1.0 mm), and a 90% void, 12 ␮m micro-fiber felt
used to manufacture the MEC structure.
In single-phase flow, laser doppler anemometry
(LDA) was applied to measure velocity gradients and
laser induced fluorescence (LIF) to measure mixing
[17–22].
3.1. CFD modeling
To develop new packing geometries, single-phase
flow CFD modeling tools (Fluent UNS with un-
structured gridding) were used and the results com-
pared and validated against experimental data from
cold-flow modeling.
Initially, a CFD grid was set up to model as closely
as possible to the small experimental test rig. The rig
inlet was configured to reduce any velocity non- uni-
formities, and consisted of a channel 30 mm in both
height and width, which contained a three layer sec-
tion of the packing. The overall length was 60 mm.
 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24 19

Fig. 3. Comparison between actual CFM experiment and

Fig. 1. MEC system concepts. CFD-simulated experiment of normalized velocity distributions for
a corrugated structured sheet packing with holes.

LDA measurements were made on planes close to the

exit of a common corrugated structured packing using
water at an F-factor (defined as uρ 1/2 and measured
in SI units) of around 130.
The CFD grids created were tetrahedral with around
300,000 cells. A top-hat velocity profile was assumed
at the inlet, which was positioned 10 mm in front of
the packing, and the domain outlet was 30 mm down-
stream of the packing exit plane. This allowed enough
space to measure and compare two planes. Two grids
of this kind were created, one without any perfora-
tions in the packing sheet, and the other with circular
holes 4 mm in diameter. The holes in the computa- Fig. 4. Comparison between actual CFM experiment and
tional grid were positioned identically to those in the CFD-simulated experiment for the dye distribution downstream of
a corrugated structured sheet packing with holes.
experimental rig.

Fig. 2. Experimental approach to quantify mixing performance of packings.

20 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24
3.2. Cold flow testing
For calibration tests, miniature corrugated packing
geometries were assembled. To determine the effect
of packing material on fluid dynamic performance,
structures were made from both sheet metal contain-
ing small (4 mm diameter) holes, and also from a
12 ␮m micro-fiber felt. These miniature geometries
were measured using the experimental approach pre-
sented in Fig. 2. Results of cold flow modeling (CFM)
measurements expressed as a normalized velocity
distribution in various directions were determined for
water at an F-factor of ca. 130. Measurements were
taken 10 and 30 mm downstream of the packing.
Fig. 3 shows the quantitative normalized comparison
between experimental CFM and the CFD simulations.
Represented are the results at 10 mm downstream of
the positions downstream of the corrugated structured
packing made of a solid foil with 4 mm circular holes.
Comparing the data from both the CFM and CFD
results at a position 10 as well as 30 mm downstream
of the packing clearly shows that the CFD simulation
accurately predicted the experimental result. This is
further demonstrated by comparing an actual camera
picture with a CFD-generated picture illustrating mix-
ing capability represented by injected dye distribution
(Fig. 4). A similar result was also seen with two other
benchmarked structures: structured packings made of
a solid foil and felt material without punched holes.
Oxygen desorption from water was used to measure
gas–liquid mass-transfer rates.
Fig. 5. Results of LDA measurements. Left: Sulzer Mellapak® 250 Y; right: self-assembled packing (250 m2 /m3 ) made of 90% void
micro-fiber felt.
4. Results and discussion
4.1. Characterization of standard structured
packings: velocity and liquid distributions
Fig. 5 shows a comparison between a Sulzer
Mellapak® 250 Y and a self-made packing from
a 90% void, 12 ␮m micro-fiber felt with a similar
geometry, but without holes or special surface char-
acteristics. Both grossly show similar undulations in
the velocity ratios indicating that the felt has similar
mixing/flow distribution characteristics as the solid
sheet with holes. Even though the felt is 90% porous,
it is quite likely that filling of the relatively narrow
pores (around 30 ␮m average) with liquid results in a
solid barrier to bulk flow.
The ability to distribute liquid was also measured.
Even though capillary forces may result in an im-
proved capacity, they are known to have a negative
effect on liquid hold-up [7,13,14,26]. However, liquid
spreading and wetting, particularly at low liquid flows
should be enhanced.
Sulzer Optiflow® packing provides excellent liquid
distribution compared to a Mellapak. We believe that
this is based on a number of positive design features
in the Optiflow packing:
• a large number of drip points,
• frequent liquid film renewal,
• the ability to properly disperse the liquid flow.
 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24
Thus, the performance of a corrugated structure can
be improved by:
• openings to allow gas distribution perpendicular to
the sheets,
• a large number of connecting points,
• adding drip points to transport liquid from one sur-
face to the other,
• good contact between adjacent sheets for improved
liquid distribution.
In the design of MEC structures, it was decided
that the above-mentioned features should be combined
with the other design criteria for structured packings.
The Optiflow design is not intended as a mixer, but
rather as a low pressure drop distillation packing (lack
of large scale vortices, low p [27]). This structure also
has a relatively low geometric surface area, and thus
may not be a good packing design for MEC applica-
tions, which require geometric surface area to allow
volumetric catalytic activity.
4.2. CFD modeling to design novel MEC
structured packings
Novel packing configurations were modeled to
determine appropriate packing characteristics for im-
proved mixing at low pressure drop. The tested com-
putational geometries were all periodic and, therefore,
a relatively small computational domain could model
an infinitely large packing structure. Each grid was
again generated using a tetrahedral mesh with each
domain containing approximately 50,000 cells.
Several novel packing designs were evaluated in the
developed, calibrated, single-phase CFD model. The
most desirable designs were selected on the basis of
large turbulent kinetic energy and high pressure drop
of the structure to enhance both mass-transfer through
the fiber felt (so called “cross-flow”) and forced
flow into the structure. However, the overall pressure
drop over the length of the packing had to remain
tolerable [5,7–9,23–25]. The goal was to enhance
bulk mass-transfer and to increase catalyst utilization
throughout the thickness of the porous medium.
We “tested” the novel structure concepts using the
previously mentioned benchmark tool. Two generic
design concepts (shown in Fig. 6) that yielded pos-
itive results from the CFD benchmarking tool were
selected on the basis of manufacturability and desired
21
performance criteria. Both showed a large turbulent
kinetic energy and high pressure drop across the faces
of the structure at acceptable overall pressure drop
over the length of the packing. CFD results for these
specific features are shown in Fig. 7. The calculation
showed that the intrinsic pressure drop of the vertical
channel design is lower than any of the corrugated
designs at equivalent geometric surface area. The ac-
tual pressure drop of the vertical channel design at
280 m2 /m3 surface area is about equal to the standard
corrugated packing design (with holes) with less sur-
face area (250 m2 /m3 ). The addition of the special
surface features (vortex generators) results in only a
minimal increase in pressure drop. Vortex-generating
punch-outs allow the addition of holes at no expense
of geometric sheet surface area, which is then avail-
able for the catalytic reaction. The sharp pointing
edges of the punch-outs will enhance continuous liq-
uid renewal and the increased localized eddies should
act to reduce flow channeling [23,24]. The punch-outs
can be modified to produce sufficient channel obstruc-
tion to minimize the occurrence of this undesirable
channeling. Multiple geometrical modifications are
thus possible [28,29].
Finally, the new MEC structure designs were man-
ufactured from micro-fiber felt with specially de-
signed tools [30]. Mass-transfer rates were measured
using oxygen desorption from water in two operating
regimes of low and high liquid flow rate in a counter-
current air stream. Fig. 8 shows the results in the case
of low flow rate, showing how comparable gas–liquid
mass-transfer rates and overall performance are ob-
tained for the porous structures (corrugations, vertical
channel) and a commercially available Mellapak®
250 Y structure. This is remarkable because the
frontal blockage of the MEC packings is about five
times larger than that of the Mellapak structure due to
the differences in sheet thickness. Also, the charac-
teristics of the micro-fibrous sheet, such as capillary
forces, would be expected to attribute to intrinsically
poorer performance. This would suggest that the
novel designs — if constructed from a flat metal sheet
— would have superior mass-transfer characteristics.
At higher liquid flow rates, the Mellapak® structure
has better oxygen desorption rates than the vertical
channel design; however, the capacity range of the
vertical channel design seems to be larger [26].
At high flow rate, the highest desorption rates are
22 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24

measured with the corrugated structure with vortex

Fig. 6. Two new families of structured packing designs. Left:
corrugated packing design with surface vortex generators; right:
vertical channel design.
generators. We ascribe this to the observed bubble
formation at the vortex generator points, which in-
crease the mass-transfer area and enhance gas–liquid
mass-transfer [5,9]. In addition to this positive feature,
the capacity of this packing seems to be enhanced, as
evidenced by it’s drop-off in performance occurring
at a higher gas rate.
One of these catalyzed structures was successfully
tested in the selective hydrogenation of C4 acetylenes,
wherein it was proven that both catalyst utilization
and hydrodynamic characteristics could be improved.
Due to their proprietary nature, detailed results cannot
be included herein. However, Fig. 9 shows how the

Fig. 7. CFD results for the two new families structured packing designs; increased darkening in color indicates larger differential in values.
 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24 23

Fig. 8. Mass-transfer rate comparison, measured using oxygen desorption from water at low liquid rate, between new MEC packing designs
and a Sulzer Mellapak® 250 Y packing.

Fig. 9. MEC concept applied to catalytic distillation (selective hydrogenation of C4 acetylenes).

selectivity of the triple bound hydrogenation compared
to double bond hydrogenation has been improved by
careful selection of catalyst activity and hydrodynamic
structure.
5. Conclusions
Novel structured catalyst packings have been de-
signed based on the use of verified computational fluid
dynamics and selected cold flow experiments. These
packing designs take advantage of the unique proper-
ties of a new type of catalyst support material: a highly
porous micro-fiber felt consisting of small metal fibers.
By being more than 90% porous, the felt offers sig-
nificantly greater exposed catalyst surface area and,
therefore, enhanced catalyst utilization. Using this ap-
proach, separately optimized bulk transport properties
can be combined with enhanced catalyst utilization.
CFD allowed both the determination of overall bulk
parameters (such as pressure drop and mixing) and
local parameters (such as local pressure drop across
24 R.E. Trubac et al. / Catalysis Today 69 (2001) 17–24
the MEC fiber felt). It was used to screen a number
of novel structural designs without performing expen-
sive, time-consuming experimental tests.
Two families of structural designs were found in
this study: a corrugated packing design with surface
vortex generators and a vertical channel design. Both
the designs can be manufactured with specially de-
signed tools without compromising the nature of the
micro-fiber felt substrate. Two-phase mass-transfer
testing was performed on the new designs, which
were compared with a commercially available prod-
uct (Sulzer Mellapak® 250 Y). In these tests at low
liquid flow rate, comparable gas–liquid mass-transfer
rates and overall performance were obtained for the
new packing designs made of the porous MEC mate-
rial. This is remarkable, since the frontal blockage of
the MEC packings is about five times larger than that
of the Mellapak structure and certain characteristics
of the micro-fibrous sheet, such as capillary forces,
would be expected to be attributed to intrinsically
poorer performance.
By introducing a novel catalyst support material
— the high surface area micro-fiber felt — optimal
bulk transport can be combined with enhanced catalyst
utilization.
Acknowledgements
The authors wish to thank the following co-workers
who contributed significantly to this paper: Rebei Bel
Fdhila and Ulrike Windecker of ABB Corporate Re-
search, Sweden, and Jonathan Lloyd of ALSTOM
Power, Switzerland.
References
[1] F.M. Dautzenberg, M. Mukherjee, in: Proceedings of the 16th
ISCRE, 2000, Chem. Eng. Science 56 (2) (2001) 251–267.
[2] R.A. Overbeek, T.F. Kellett, F.M. Dautzenberg, ABB Rev. 2
(2000) 48–55.
[3] K. Rock, G.R. Gildert, T. McGuirk, Chem. Eng. 104 (7)
(1997) 78.
[4] W.P. Stadig, Chem. Proc. (1987).
[5] M.H. de Brito, Ind. Eng. Chem. Res. 33 (1994) 647–656.
[6] J.M. Coulson, J.F. Richardson, Chemical Engineering, Vol.
6, Design, 2nd Edition, Pergamon Press, Oxford, 1993,
pp. 529–544.
[7] F. Stoter, Chem. Eng. J. 53 (1993) 55–66.
[8] R.F. Strigle Jr., Packed Tower Design and Applications:
Random and Structured Packings, 2nd Edition, Norton
Chemical Process Products Corporation, Gulf Publishing
Company, Houston, 1994.
[9] J.A. Rocha, Ind. Eng. Chem. Res. 32 (1993) 641–651.
[10] J.M. Coulson, J.F. Richardson, Chemical Engineering, Vol.
2, Particle Technology and Separation Processes, 4th Edition,
Pergamon Press, Oxford, 1991, pp. 132–173.
[11] J.R. Fair, J.L. Bravo, IChemE Symp. Ser. 104 (1988) A183–
A201.
[12] G. Nardini, Chem. Eng. Technol. 19 (1996) 20–27.
[13] R.K. Helling, M.A. DesJardin, Chem. Eng. Prog., 13 (October
1994) pp. 62–66.
[14] O. Becker, R. Steiner, Chem. Eng. Technol. 67 (7) (1995)
883–888.
[15] J.M. Ottino, The Kinematics of Mixing: Stretching, Chaos
and Transport, 1st Edition, Cambridge University Press,
Cambridge, 1989.
[16] H. Tennekes, J.L. Lumley, A First Course in Turbulence, MIT
Press, Cambridge, 1989.
[17] G. Eigenberger, Fortschrittberichte VDI Nr. 112, Reihe 15:
Umwelttechnik, VDI-Verlag, 1991.
[18] G. Eigenberger, Ullmann’s Encyclopedia of Industrial
Chemistry, Vol. B4, 1992, VCH Publishing, pp. 199–238.
[19] Separation Columns for Distillation and Absorption, Sulzer
Chemtech, 1991.
[20] V. Kottke, H. Blenke, Verfahrenstechnik 16 (6) (1982) 504.
[21] V. Kottke, H. Blenke, Chem. Eng. Technol. 50 (2) (1978) 81.
[22] W. Kühnel, V. Kottke, Proc. Eur. Sem. 46 (1995) 61.
[23] F.A. Streiff, J.A. Rogers, Don’t Overlook Static-Mixer-
Reactors, Information on Mixing and Reaction Technology
by Sulzer Chemtech, 1994.
[24] European Patent Application No. 82301389.1, March 18,
1982.
[25] A.M.J. Davis, D.F. James, Int. J. Multiphase Flow 22 (5)
(1996) 969–989.
[26] R.F. Strigle Jr., Chem. Eng. Prog., 1993, pp. 79–83.
[27] P. Süess, W. Meier, A New Transfer Structure for Distillation
and Absorption, Information by Sulzer Chemtech, 1995,
pp. 416–420.
[28] B. Paikert, J. Lloyd, T.A. Griffin, PCT Patent Application,
WO9934911, 1999.
[29] R.A. Overbeek, R.E. Trubac, V.A. Strangio, B. Paikert, J.
Lloyd, T.A. Griffin, R. Bel Fdhila, PCT Patent Application,
WO9962629, 1999.
[30] F. Buchi, E. Vogt, P. Dubach, T. Griffin, J. Lloyd, B. Paikert,
PCT Patent Application, WO0024506, 2000.
 Catalysis Today 69 (2001) 25–31

Iron oxide-based honeycomb catalysts for the

dehydrogenation of ethylbenzene to styrene
William P. Addiego∗ , Wei Liu, Thorsten Boger
Corning Incorporated, Sullivan Park-DV-01-9, Corning, NY 14831, USA

Abstract
Corning has recently developed a novel extrusion method to make bulk transition metal oxide honeycomb catalysts. One area
of effort has been iron oxide-based catalysts for the dehydrogenation of ethylbenzene to styrene, a major chemical process
that yields worldwide 20 MM tons/yr. In industry, the monomer is synthesized mostly in radial-flow fixed-bed reactors.
Because of the high cross-sectional area for flow and shallow depth of the catalyst bed in these reactors, low reactor pressure
gradients are maintained that favors the yield and selectivity for styrene formation. However, the radial-flow design has
inherent detractions, including inefficient use of reactor volume and large temperature gradients that decrease catalyst service
life. The overall economics of the process can be improved with parallel-channel honeycomb catalysts and axial flow reactors.
The simple axial flow design of honeycomb catalysts provides low-pressure drop, while making more efficient use of reactor
volume, with better heat and mass transfer characteristics compared to a conventional radial packed bed. An important part
of this concept is the ability to fabricate a wide family of dehydrogenation catalyst compositions into honeycombs with the
requisite chemical, physical, mechanical, and catalytic properties for industrial use. The ethylbenzene dehydrogenation (EBD)
honeycomb catalysts developed by Corning have compositions similar to those commonly used in industry and are prepared
with the same catalyst and promoter precursors and with similar treatments.
However, to enable extrusion of catalyst precursors into honeycomb shapes, especially at cell densities above 100 cell/in.2 ,
Corning’s process compensates for the high salt concentrations and the high pH of the batch material that would otherwise
prevent or impede honeycomb extrusion. The improved rheological characteristics provide the necessary plasticity, lubric-
ity, and resiliency for honeycomb extrusion with sufficient binder strength needed before calcination to the final product.
Iron oxide-based honeycombs after calcination are strong and possess macroporosity and high surface area. In bench-scale
testing, particular honeycomb catalyst compositions exhibited 60–76% ethylbenzene conversion with styrene selectivity of
95–91%, respectively, under conventional reaction conditions without apparent deactivation or loss of mechanical integrity.
© 2001 Elsevier Science B.V. All rights reserved.
Keywords: Iron oxide; Honeycomb; Catalysts; Dehydrogenation; Ethylbenzene; Styrene; Axial flow; Reactor

1. Introduction tons/yr of styrene monomer are made. Although this

The industrial process for the dehydrogenation of
ethylbenzene to styrene is used to make valuable com-
modity chemicals such as polystyrene and synthetic
rubber such as ABS and SB latex; as much as 20 MM
∗ Corresponding author. Fax: +1-607-974-3745.
E-mail address: addiegowp@corning.com (W.P. Addiego).
process was already commercialized in the 1930s, re-
search and development has continued to improve it,
including the development of new catalysts, reactor
designs and process routes [1–13].
The non-oxidative ethylbenzene dehydrogenation
(EBD) to styrene is endothermic, reversible and
equilibrium limited. Because of the thermodynamic
limitation, the formation of styrene is favored by the

0920-5861/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 0 1 ) 0 0 3 5 1 - 0
26 W.P. Addiego et al. / Catalysis Today 69 (2001) 25–31

low partial pressure of ethylbenzene and hydrogen. Table 1

To promote the forward reaction, the industrial pro- Pressure drop comparison of honeycomb and packed bead with
equivalent bed cross-section
cess uses steam, where the molar ratio of steam to
ethylbenzene is 6–13:1 at atmospheric or vacuum Bed properties Honeycomb Bead
conditions. Steam provides heat for the reaction, pre- Void fraction 0.5
vents over-reduction of the iron oxide-based catalyst, Linear velocity (m/s) 6.8
prevents coke formation, and decreases the partial Wall thickness (mm) 1 0.5 3.2φ
pressure of gases, thereby shifting the chemical equi- Cell density (cpsi) 55 220 N/A
Pressure drop (bar/m) 0.021 0.085 0.437
librium to higher styrene conversion [3,13]. Characteristic heat/mass 0.5 0.25 0.8
For low reactor pressure drop, industrial EBD is per- transfer dimension (mm)
formed in radial-flow reactors, as shown in Fig. 1(a).
Total ethylbenzene conversion through both reactors
is about 60–70%. An axial flow honeycomb reac-
tor, shown in Fig. 1(b), could improve overall effi- catalyst, while maintaining low-pressure drop [14].
ciency by incorporating heat exchangers to minimize With a higher geometric surface area relative to packed
thermal gradients along the catalyst bed, improving beads, the honeycomb also has a better mass/heat
the performance and extending the service life of the transfer dimension, as shown in Table 1.

Fig. 1. (a) Diagram of a first-stage radial-flow reactor with cylindrical shaped packed catalyst bed held in place by concentric screens
and with a central conical gas distributor. (b) Axial flow honeycomb catalyst reactor with interbed heat exchanger. In both configurations,
effulent is sent to a second-stage reactor to dehydrogenate remaining ethylbenzene.
 W.P. Addiego et al. / Catalysis Today 69 (2001) 25–31
Industrial dehydrogenation compositions are a
complex mixture of metal oxides and large amounts
of salts, principally iron oxide and potassium car-
bonate. During reaction start-up, under essentially re-
duction conditions, the catalytically active potassium
ferrite, KFeO2 , forms along with ferrite polymorphs
[16,20–22]. In addition, potassium carbonate also acts
as a catalyst to oxidize carbonaceous deposits to pre-
vent coking [3]. The literature is replete with studies
on EBD catalysts [2–6,13,15–24].
Controlling the rheological properties of the catalyst
compositions is important in order to provide the req-
uisite plasticity and lubricity during extrusion needed
for web formation and knitting, skin formation, and
stiffening as the honeycomb emerges from the die.
However, because of the large concentration of salts
in typical dehydrogenation catalyst compositions, ex-
trusion rheology is usually degraded with the loss of
plasticity and water migration. This occurs, in part, be-
cause salts can induce the precipitation of organics and
the loss of “bound” water layers associated with or-
ganics and distributed throughout the extrusion batch.
The proper combination of organic additives and ma-
terials processing is necessary to extrude honeycomb
catalysts containing high salt concentration and avoid
or attenuate adverse ionic shock. This paper reports on
a method of extruding industrial-type EBD catalysts,
their characterization and catalytic performance.
2. Experimental
2.1. Extrusion compositions
Extrusion batches were made in a manner simi-
lar to industrial methods. ␣-Fe2 O3 was mixed with
Table 2
EBD catalyst compositions
Component Batch component Oxide (wt.%)
Fe2 O3 Oxide 25–80
K2 O Carbonate 10–35
CeO2 Carbonate 0–5
MoO3 Ammonium molybdate 0–3
CaO Carbonate 0–3
MgO Carbonate 0–10
27
appropriate amounts of K2 CO3 , (NH4 )6 Mo7 O24 ,
CeCO3 , and MgCO3 or CaCO3 , based on the ana-
lyzed equivalent oxide of the salts, as listed in Table 2.
The catalyst precursors were dry-blended in a tur-
bular mixer with certain organic additives. After dry
blending, organic emulsions were added with water,
along with additional plasticizers and lubricants, and
mulled to form a plastic dough, that was subsequently
homogenized to ensure a thorough mixing of solid
and liquid components. Honeycombs were then ex-
truded with square cells of 100, 200, and 400 cell/in.2
(cpsi) with web thickness of 0.64, 0.38, and 0.18 mm,
respectively. Extrusion rheology was measured with
a parallel-plate rheometer.
2.2. Chemical and physical characterization
and catalyst testing
After extrusion, honeycombs were dried at 80◦ C
and calcined at 850◦ C/6 h. Calcined samples of
100 cpsi honeycombs were characterized for surface
area, porosity, pore size distribution, A-axis crush-
ing strength, and catalytic activity and selectivity for
styrene formation. As shown in Table 3, honeycomb
catalysts which had a BET surface area of 4 m2 /g
were macroporous with >50% porosity. Honeycombs
were very strong, exhibiting an A-axis crushing
strength (i.e., force applied parallel to cell walls) of
1300–2000 psi after calcination.
Honeycomb catalysts were also tested for catalytic
performance using an integral reactor. The reactor con-
sisted of a stainless steel tube within which a 1-in. di-
ameter quartz tube was inserted to support a 50 cm3
honeycomb catalyst, 2.5 cm (ID) × 10.0 cm long. −60
mesh SiC particles were packed above and below the
honeycomb, separated from the honeycomb by quartz
Function/promoter Compositions oxide (wt.%)
J-A, 2J-AB Q-B1
Activity 72 79
Activity 16 11
Activity 4 4
Selectivity 1 3
Chemical/mechanical stability 0 2
Chemical/mechanical stability 7 1
28 W.P. Addiego et al. / Catalysis Today 69 (2001) 25–31

Table 3
Physical and mechanical characteristics of 100 cpsi EBD honeycomb catalysts calcined at 850◦ C/6 h
Honeycomb Surface A-axis compression Apparent % Porosity Median pore
catalyst area (m2 /g) strength (psi) density (g/ml) size (nm)
J-A 4.0 1300 3.9 53 350
Q-B1 3.0 1800 4.4 56 380
2J-AB 3.5 2000 3.9 54 330

wool. Thermocouples were placed at the honeycomb
entrance and exit in the SiC packing for uniform tem-
perature measurement. Above the SiC packing, dense
␣-alumina beads were packed to help the heat feed
uniformly before entering the honeycomb.
Reagent grade ethylbenzene and de-ionized water
were delivered to the top of the reactor by two liquid
pumps in order to provide the appropriate steam to
ethylbenzene ratio. The mixture was vaporized in the
reactor as it flowed to the honeycomb catalyst. The ef-
fluent was cooled and liquid product and gas were then
separated. Hydrocarbons were separated from water
and measured by gas chromatography.
3. Results and discussion
3.1. Extrusion behavior
Extrusion compositions were prepared with various
polymer emulsions to affect rheology. For this part of
the study, one particular inorganic composition was
used, J-A listed in Table 2. These were compared
to extrusion performance without organics or with
soluble cellulose ethers. Certain polymer emulsions
were suitable for extruding honeycombs and superior
to water-soluble organic extrusion aids, providing
plasticity to an extent that honeycomb matrix forma-
tion was possible with a 100 cpsi/0.7 mm geometry
or greater. Systems with appropriate colloidal poly-
mer emulsions set and dried rapidly without crack-
ing and showed considerable green strength before
calcination.
The impact of polymer glass transition tempera-
ture, Tg , on extrusion rheology was measured, based
on changes in the batch material’s stiffness or rigid-
ity as a function of the complex modulus, G∗ . Extru-
sion batches containing a soft polymer with a Tg near
or below ambient stiffened much more rapidly dur-
ing processing at ambient temperatures than batches
containing glassy polymers with a Tg significantly
above the mixing or extrusion temperature, as shown
in Table 4. The elastic modulus, G , of a batch con-
taining a glassy polymer increased rapidly after 1 h, as
shown in Fig. 2, due to polymer cross-linking at the
typically high batch pH ≈ 9, rendering the batch brit-
tle and less capable of plastic deformation. With the
appropriate organic system, the batch remained suffi-
ciently plastic during extrusion. Honeycombs rapidly
set at temperatures near the Tg of the polymer and
dried with little or no cracking. This feature minimized
distortion of the honeycomb matrix.
Water-soluble extrusion aids, such as cellulose
ethers and polyethylene oxides, are very suscepti-
ble to ionic shock and tend to precipitate in the
presence of fairly low concentrations of alkali salts,
especially carbonate and sulfate salts [25,26]. In ex-
trusion batches containing K2 CO3 , these polymers
easily flocculate, lose the ability to disperse water,
and contribute to the overall degradation of extrusion
rheology. The yield stress of extrusion batches with
different types of organic systems was approximated
from the G∗ modulus at a shear frequency of 1 rad/s
and 5% strain and is listed in Table 5. The amount
of liquid to plasticize the batch was normalized to a
reference batch not containing any organic additive.
A significant amount of water was needed to plasti-
cize the reference, which showed a low yield stress of
19 kPa. Less water was required after adding methyl-
cellulose, raising the yield stress, but honeycombs
Table 4
Effect of organic additive Tg on setting time of EBD catalyst
extrusion batch
Polymer Tg (◦ C) Setting Complex
emulsion time (min) modulus (kPa)
EM101 <0 30 >350
VM501 <15 90 >350
RT502 >75 140 >350
 W.P. Addiego et al. / Catalysis Today 69 (2001) 25–31 29

Fig. 2. The elastic modulus, G , increases more rapidly during processing than the viscous modulus, G , as the organic binder sets,
stiffening the extrudate. The high salt concentration and pH of the extrusion batch promotes binder setting.

Table 5
Effect of organic additives on yield stress of EBD catalyst extrusion batch
Organic additives Extrusion quality Yield stress (kPa) % Liquida

No organic Matrix collapse soft, tearing 19 100

Methylcellulose Matrix distortion slumping 31 46
Organic pack 1 Matrix collapse slumping 9 69
Organic pack 2 No matrix distortion 61 57
a The amount of water and aqueous emulsion in the batch and normalized to the liquid content of the composition containing no

organic additives.

were easily deformed. After drying and calcination,
the material was brittle and weak. Another extrusion
batch exhibited a very low yield stress because of the
relatively high concentration of polymer emulsion in
the batch. Lowering the emulsion concentration and
decreasing the water content raised the yield stress to
61 kPa with sufficient stiffness under shear to extrude
a plastic material into honeycombs with little or no
matrix distortion. As shown in Fig. 3, an extrusion
using organic pack 2 had a significantly higher G∗
than all of the other extrusion batches. Consequently,
honeycombs were more resilient emerging from the
extrusion die than honeycombs of the other softer
compositions. Examples of the extrudates that had a
high yield stress are shown in Fig. 4 with cell densities
up to 400 cpsi.
3.2. Catalytic performance
The J-A honeycomb catalyst was tested and com-
pared with the literature, as shown in Table 6. As
expected, the ethylbenzene conversion increased with
an increase in the steam/EB ratio and decreased when
the bed temperature was lowered. Selectivity toward
styrene decreased with higher EB conversion and
increased with low steam/EB ratio and bed temper-
ature. Although not shown, ethylbenzene conversion
decreased with an increase in the space velocity, as
expected. In addition, the major by-products were
benzene and toluene, and phenylacetylene concentra-
tion was less than 200 ppm. Styrene selectivity was
as high as 93% with an EB conversion of 67% at a
low steam/EB ratio. Fig. 5 shows the selectivity and
yield as a function of EB conversion. The model was
calculated by determining a best-fit equation from ex-
perimental data. Experimental data fit well with that
reported in the literature [2,7–9].
During the reaction, catalyst surface area decreased
to about 2 m2 /g and porosity decreased to 45–50%.
These changes were associated with the formation of
potassium ferrites under reducing conditions. While
no catalytic deactivation was observed during the
30 W.P. Addiego et al. / Catalysis Today 69 (2001) 25–31

Fig. 3. The G∗ modulus measures the rigidity of the batch material. Initially, shear-thinning lowers G∗ as the shear rate (frequency)
increases. The lower the G∗ , the softer the material. G∗ is affected by the type and concentration of organic additives in the extrusion batch.

Fig. 4. Extruded EBD honeycomb catalysts.

Table 6
Catalytic performance of EBD honeycomb catalysts
J-A honeycomb catalysts Literature valuesa

T (◦ C) 609 605 588 584 593 593

H2 O/EB molar ratio 13.4 6.7 13.4 13.4 12.2 9.3
LHSV 0.48 0.48 0.48 0.48 0.48 0.48
Conversion (%) 76.3 66.7 68.2 59.5 68.1 64.8
Selectivity (%) 90.7 93.3 93.4 95.2 92.4 92.3
a Ref. [8].
 W.P. Addiego et al. / Catalysis Today 69 (2001) 25–31
Fig. 5. Selectivity and yield for styrene as a function of ethyl-
benzene conversion during dehydrogenation. Comparison of hon-
eycomb catalysts with literature values [2,7–9].
reaction, formation of catalytically inactive magnetite
and potassium hydroxide might have also occurred to
some extent [22].
4. Summary and conclusions
Highly active and selective industrial-type EBD
catalysts were extruded into strong honeycombs. With
the proper polymer emulsion and concentration, there
was sufficient plasticity and lubricity for extruding
compositions with high concentrations of potassium
carbonate and a high pH. The yield stress was high
enough for the honeycomb matrix to support itself
with cell densities up to 400 cpsi without slumping or
distortion. Processing time was extended by using a
polymer with a Tg significantly above ambient. Hon-
eycombs were rapidly set by drying the honeycomb
near the Tg of the polymer, minimizing distortions as
the honeycomb dried. Calcined honeycomb catalysts
were strong, with crushing strengths from 1300 to
2000 psi. The honeycombs were generally macrop-
orous with >50% porosity.
Catalytic performance was generally equivalent to
packed beads reported in the literature. Styrene selec-
tivity was 95–91% with 60–76% EB conversion.
The results have shown that dehydrogenation
catalysts with a complex set of precursors can be
31
extruded into honeycombs of various cell geometries
with proper rheological control. The honeycomb is
converted into an active catalyst under reaction con-
ditions, while retaining its structure without loss of
mechanical integrity.
References
[1] A.A. Savoretti, D.O. Borio, V. Bucala, J.A. Porras, Chem.
Eng. Sci. 54 (1999) 205–213.
[2] Q. Chen, X. Chen, L. Mao, W. Cheng, Catal. Today 51 (1999)
141–146.
[3] F. Cavani, F. Trifiro, Appl. Catal. 133 (1995) 219–239.
[4] G. Kolios, G. Eigenberger, Chem. Eng. Sci. 54 (1999) 219–
239.
[5] N. Mimura, M. Saito, Catal. Today 55 (2000) 173–178.
[6] N. Mimura, I. Takahara, M. Saito, T. Hattori, K. Ohkuma,
M. Ando, Catal. Today 45 (1998) 61–64.
[7] H.-J. Kremer, US 5 097 091.
[8] D.K. Kim, G.M. Longland Jr., US 5 503 572 (1991).
[9] D.L. Williams, K.J. Russ, E.K. Dienes, G.A. Laufer, US
5 023 225 (1991).
[10] C.-C. Chu, US 4 503 163 (1985).
[11] J.T. Smith, B.S. Masters, D.J. Smith, US 4 467 046
(1984).
[12] P.W. Dellinger, R.G. Moore, F.A. Sherrod, A.R. Smith, US
5 510 552 (1996).
[13] E.H. Lee, Catal. Rev. 8 (2) (1973) 285–305.
[14] T. Boger, W. Liu, W.P. Addiego, S.A. Campbell, (2001), 6th
World Congress to Chemical Engineering.
[15] T. Hirano, Appl. Catal. 26 (1986) 65–79.
[16] T. Hirano, Appl. Catal. 26 (1986) 81–90.
[17] T. Hirano, Appl. Catal. 26 (1986) 119–132.
[18] V.K. Kaushik, T.S.R.P. Rao, B.L.S. Yadav, M.S. Chhabra,
Appl. Surf. Sci. 32 (1988) 93–98.
[19] G.E. Vrieland, P.G. Menon, Appl. Catal. 77 (1991)
1–8.
[20] W. Weiss, D. Zscherpel, R. Schlögl, Catal. Lett. 52 (1998)
215–220.
[21] A. Trovarelli, C. deLeitenburg, M. Boaro, G. Dolcetti, Catal.
Today 50 (1999) 353–367.
[22] M. Muhler, R. Schlögl, A. Reller, G. Ertl, Catal. Lett. 2
(1989) 201–210.
[23] N.J. Jebarathinam, N.J.M. Eswaramoorthy, V. Krishnasamy,
Appl. Catal. A 145 (1996) 57–74.
[24] W.P. Addiego, C.A. Estrada, D.W. Goodman, M.P. Rosynek,
R.G. Windham, J. Catal. 146 (2) (1994) 407–414.
[25] MethocelTM Cellulose Ethers Technical Handbook, The Dow
Chemical Company, 1997.
[26] F.W. Stone, J.J. Stratta, in: H.F. Mark, N.G. Gaylord,
N.M. Bikales (Eds.), Encyclopedia of Polymer Science and
Technology, Vol. 6, Wiley, New York, 1967, pp. 103–145.
 Catalysis Today 69 (2001) 33–39

Permeability of sintered microfibrous composites

for heterogeneous catalysis and other chemical
processing opportunities
Donald R. Cahela, Bruce J. Tatarchuk∗
Chemical Engineering Department, Auburn University, 230 Rose Hall, Auburn, AL 36849, USA

Abstract
Microstructured materials have potential for enhanced mass and heat transfer compared to typical catalyst particulates
used in industrial processes. The pressure drop through catalyst-containing materials is a very important reactor design
consideration. A model equation to predict porous media permeability (PMP) over the entire range of possible bed voidages is
extended to predict properties of sintered metal meshes. A correlation of data for sintered meshes of nickel fibers is presented
in the form of a Kozeny constant form drag plot. Comparison of predictions by the PMP equation with data taken on a sintered
composite fiber/particle mesh is presented. Use of the PMP equation as a design tool for optimization of media for adsorbents,
catalysts, and filters is discussed. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Permeability; Heterogeneous catalyst; Catalysis; PMP

1. Introduction thermal conductivity of the micrometal fiber matrix

A new-patented class of composites materials made
by a wet lay paper-making/sintering process can incor-
porate particles as small as 10 ␮m into a micrometal
fiber matrix [1–4]. Sintered microfibrous composites
using a 16% Ni/alumina catalyst for toluene hydro-
genation in a trickle bed reactor have demonstrated
2–6 times higher specific activities than conventional
packed bed catalysts on a gravimetric basis, while
volumetric activities of 40 vol.% composite catalysts
were 80% higher than conventional extrudates [5].
Effectiveness factors for reactions employing such
small particles are typically close to unity. The high
∗ Corresponding author. Tel.: +1-334-844-2023;
fax: +1-334-844-2065.
E-mail addresses: tatarbj@auburn.edu, brucet@eng.auburn.edu
(B.J. Tatarchuk).
produces radial heat transfer coefficients that are about
twice those in packed beds. Enhanced heat transfer
due to the larger surface to volume ratio of small parti-
cles and the micrometal matrix results in more nearly
isothermal conditions in fixed bed reactors minimizing
hot spots, resulting in higher selectivity, and allowing
processing to be done with smaller safety factors.
The micrometal fiber matrix also acts as a micron
scale static mixer eliminating channeling. The wet lay
paper-making/sintering process allows continuous ad-
justment of void volume from upwards of 98% down
to values similar to packed beds of particles [6]. This
allows the composite catalyst/adsorbent material to be
optimized to the appropriate void volume for different
types of applications. Production of these materials is
based on standard paper-making techniques [7], which
can be scaled up to manufacturing on high speed
continuous paper machines resulting in relatively

0920-5861/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 0 1 ) 0 0 3 5 2 - 2
34 D.R. Cahela, B.J. Tatarchuk / Catalysis Today 69 (2001) 33–39

Nomenclature
a wetted surface per volume of
bed (cm2 /cm3 )
av surface to volume ratio of a solid
shape (cm2 /cm3 )
b/a aspect ratio of an ellipse
CD coefficient of drag for sphere in
turbulent flow = 0.6
Cf coefficient of friction for turbulent flow
CFD coefficient of form drag of sphere in
turbulent flow (CFD = CD − Cf )
D diameter of characteristic size of any
arbitrary shape (cm) Fig. 1. SEM of sintered composite material of 55–88 ␮m activated
k Kozeny constant (Eq. (2)) carbon and 2, 4 and 8 ␮m nickel fibers.
L geometrical unit cell spacing
P pressure drop across fixed bed
(g/cm2 ) or (cm H2 O) sharply from this equation at void volumes greater
T bed thickness (cm) than 80%, because it does not account for form drag.
v0 face velocity (cm/s) The viscous loss portion of Ergun’s equation is re-
xFD form drag parameter, xFD = ε2 /12(1 − ε) ferred to as the Carman–Kozney equation. Porosity
xi volume fractions of components in functions have typically been used to empirically cor-
bed of solids rect the Carman–Kozney equation for fibrous media
at high voidages where it becomes inaccurate. Sev-
Greek symbols
eral more complicated formulas have been developed
α solid volume fraction, 1 − ε
for fibrous media, but they lose accuracy at low void
ε void fraction of bed
volumes, and none of these formulas can be applied
θ angle of flow paths through bed (◦ )
directly to mixtures of fibers and particulates. A new
µ fluid viscosity (P) or (g/cm s)
theoretically based porous media permeability (PMP)
ρ fluid density (gm/cm3 )
equation, having a single media-dependent parameter,
τ tortuosity of cubic cell with one sphere
describes permeability across the entire range of void
inside, τ = 1 + 21 α
volumes, and is applicable to mixtures of objects of
φ shape factor for any arbitrary
any arbitrary shape.
shape, φ = 6/Dav
Almost every process uses porous media. Some
examples are: fixed bed reactors, packed distillation
inexpensive materials. For an example of these mi- columns, adsorbent beds, filters, heat pipes, etc. Equa-
crofibrous composite carriers, an SEM image of tions which predict permeability or pressure drop
55–88 ␮m activated carbon particles entrapped in a through the porous media are available, but all of
micrometal fiber matrix of 2, 4 and 8 ␮m nickel fibers them apply only over a restricted range of porosity,
is shown in Fig. 1. either low porosities or porosities approaching unity.
Dullien [8] reviewed over 300 studies published Macdonald et al. [9] reviewed equations describing
since 1927 dealing with the permeability of porous flow through porous media, and concluded that there
media, such as packed beds. None of these studies are no generally applicable equations. The Ergun
produced models to predict the permeability of porous equation is the best known, but it is only applicable to
media over the entire range of possible void volumes. media of 50% or less porosity. Kyan et al. [10] devel-
Ergun’s equation, which is fitted to data for packed oped a pressure drop equation based on a geometrical
beds of particulates, is typically used to predict per- model, but it is only applicable to high porosity. Their
meability of porous media. Experimental data deviate model is unnecessarily complicated, and includes a
 D.R. Cahela, B.J. Tatarchuk / Catalysis Today 69 (2001) 33–39
deflection energy loss term that is not well justified. In
this paper, we develop a generally applicable equation
for the permeability of porous media by combining
the friction losses estimated by the Hagen–Poiseuille
equation, and the form drag losses estimated from
Stokes’ law.
The equation developed here applies directly to a
new class of materials. These materials are fixed beds
of micron-sized metal fibers with entrapped particles.
Tatarchuk et al. [1–7] developed a method for mak-
ing these materials based on paper-making technology.
Materials are mixed in solution and cast on a screen
forming a paper preform. The most basic preform con-
sists of micron-sized metal fibers mixed with cellulose,
which acts as a binder in the preform. Preforms are
heated in a reducing environment at temperatures from
1073 to 1373 K sintering the metal fibers and gasify-
ing the cellulose. Sintering produces welded joints be-
tween the metal fibers resulting in a very mechanically
robust material. Fibers of very different properties
such as activated carbon or ceramic fibers can be easily
incorporated into the paper-making step. Particles can
also be added in the paper-making process, so this be-
comes a very flexible manufacturing process in which
materials of vastly different properties can be com-
bined, so the advantageous properties of each material
can be accessed in a manner previously unattainable.
Previously, a stacked screen model has been used to
estimate the permeability of a microfibrous mesh [5],
but this method cannot be easily extended to mixtures
of fibers and particles. Permeability of this new class
of materials can be estimated by the PMP equation de-
veloped here. The PMP equation is also applicable to
beds packed to any void fraction and fluidized beds.
2. Porous media permeability equation
Filter materials are composed of fibrous materials
and usually have a very high porosity greater than
80%. Usual design equations, such as the Ergun equa-
tion, for packed columns do not apply to beds with
porosity >50%. The reason these equations cannot be
used for high porosity materials is that they do not in-
clude the form drag losses, which are small in packing
material of low porosity.
Kozeny [11] first developed an equation describing
viscous friction losses for flow through packed beds
35
by considering the packed bed as a collection of cap-
illaries. In Kozeny’s model the hydraulic radius of the
porous media is inserted into the Hagen–Poiseuille
formula for laminar flow inside a pipe. Ergun cor-
related several hundred data points taken on packed
beds to improve the coefficients for the viscous losses
in Kozeny’s expression, and also fit the coefficient of
Burke and Plummer’s equation for inertial losses. Bird
et al. [12] give a derivation of Ergun’s equation for
pressure drop in a packed column. Calculating the vis-
cous form drag losses of the porous media by Stokes
law, and adding these to the viscous friction losses can
improve the Kozney/Ergun equation for high voidage
materials. The resulting PMP equation [13] is given by
the following equation for a homogeneous material.

Viscous
P τ 2 µv0 (1 − ε)2
= 72 [1 + xFD ]
L Losses cos2 (θ ) ε 3 (φD)2
(1)
The model parameters are defined in the nomencla-
ture. The angle of flow paths in the permeable media is
the only adjustable parameter varying from 0◦ to 45◦ ,
thus allowing only a twofold adjustment in the calcu-
lated permeability. The form drag parameter (xFD =
ε2 /12(1−ε)) is the ratio of viscous form drag losses to
viscous skin friction losses for a homogeneous media.
The tortuosity (τ = 1 + 21 α) is estimated for the PMP
equation by considering a cubic unit cell containing
one spherical particle. Taking this geometrical model
and calculating the extra length fluid has to traverse
because of the flow around the sphere allows estima-
tion of the tortuosity. This approach results in a very
simple result of the tortuosity being unity plus the solid
volume fraction divided by 2. The Kozeny constant
(k) is a rearrangement of the pressure drop equation
for viscous losses intended to eliminate effects of bed
porosity and particle size and shape effects. Kozney
constant corresponding to the PMP equation for a ho-
mogeneous material is given by the following formula:

2
P 1 ε3 φD
k ≡
L µv0 (1 − ε)2 6
2τ 2
= [1 + xFD ] (2)
cos2 (θ )
The derivation of the PMP equation was presented
at IEF2000 and is in press [13]. Comparisons of the
36 D.R. Cahela, B.J. Tatarchuk / Catalysis Today 69 (2001) 33–39

PMP equation to literature data taken on textile fibers of data to the M-E and determined best values of the
by Lord [14] and also data tabulated by Jackson and constants in the M-E equation. Values of angle of flow
James [15] was also given in the previous publica- path and coefficient of friction for the PMP equation
tion. Excellent agreement of the PMP equation with to equal the M-E at an average bed voidage of 45%
the literature data was shown. The data from Jackson are flow path angle of 35◦ and coefficient of friction
et al. extended the range of solid fraction covered of 0.20–0.54.
down to 3.45 × 10−4 or a form drag parameter of 240,
covering three and one-half decades of the form drag
parameter. Data for permeability from the highest 3. Pressure drop for meshes of sintered composite
solid fractions, similar to packed beds, to the lowest meshes of metal fibers and particles
solid fractions follow the trend of the PMP equation.
The inertial friction and form drag losses can be 3.1. Experimental methods
estimated by friction factor correlations for turbulent
flow as in the derivation of Ergun’s equation. Inertial Nickel fibers of 2, 4, 8 and 12 ␮m from Micrometals
form drag losses can be estimated similarly by a fric- were formed into preforms by standard paper-making
tion factor correlation for a sphere in turbulent flow. methods using 12 g of metal and 3 g of cellulose in
The coefficient of form drag is the coefficient of drag a 1 ft2 preform. Preparation of these types of materi-
minus the coefficient of friction. The drag coefficient als has been presented previously [1–7]. Rectangular
for a sphere in turbulent flow is approximately 0.6. The pieces 2.5 cm × 7.0 cm were sintered at 1223 K with
total pressure loss is given by the following equation: 300 cm3 H2 /min for 1 h, oxidized with 50% air/helium

2 mixture at 773 K for 1 h, and resintered at 1223 K
P τ 2 µv0 1−ε
= 72 2 (1 + xFD ) for 1 h. The oxidation step removes all traces of py-
T cos (θ ) ε 3 φD rolyzed cellulose from the metal meshes. The nickel


τ 3 ρv02 (1 − ε) CFD meshes typically have a voidage of about 95% after
+6 C f + ε (3) sintering. Twelve three-quarter inch circles were cut
cos3 (θ ) 2ε 3 φD 4
out of each of the sintered meshes. These were com-
The PMP equation can be generalized to mixtures of pressed by various amounts to obtain 10–12 samples
any shape by determining the hydraulic radius of the with solid volume fractions corresponding to a spread
mixture and summing the viscous form drag losses of form drag parameters on the x-axis of the Kozeny
over all of the solid shapes in the composite material. constant plots. An SEM image of a sintered mesh of
The total pressure drop for a mixture of solid parti- 4 ␮m nickel fibers is shown in Fig. 2.
cles or fibers of any shape is given by the following Pressure drop through each of the samples was
equation: measured using a sample cell consisting of a 43 in.
P τ 2 µv0 (1 − ε)2
= 72 2
T cos (θ ) ε3

 
 xi 2  xi
× + xFD
φi D i (φi Di )2
τ 3 ρv02 (1 − ε)
+6 3
cos (θ ) 2 ε3
 ε   xi
× Cf + CFD (4)
4 φi Di
Parameter values corresponding to the Ergun equation
can be found by equating the Ergun equation and the
modified Ergun (M-E) equation to the one developed
here. Macdonald et al. [9] fitted a very large amount Fig. 2. SEM of mesh of 4 ␮m nickel fibers.
 D.R. Cahela, B.J. Tatarchuk / Catalysis Today 69 (2001) 33–39
Swagelok Ultratorr fitting with 43 in. SS tubing inserted
into the ends of the Ultratorr fitting. Three-quarter
inch OD spacers with 58 in. ID were machined from
1
8 in. thick aluminum. Two spacers were placed be-
tween the sample and the tubing inserted into the
sample cell. Connection to an air cylinder with regu-
lator was made through 41 in. SS tubing. A clamp was
put around the outside of the sample holder to com-
press the spacer rings and prevent air from bypassing
the sample. A Cole-Parmer flow tube (NO34-39)
0.5–15 l/min, was used to measure the flow rate of
air through the sample. The flow tube was calibrated
using a 1000 cm3 bubble meter. Pressure drop was
measured by a DP cell (OMEGA PX154-010DI), and
readings were indicated by a four digit process meter
(OMEGA DP24E). The DP cell was calibrated from
0 to 1 in. H2 O using a Dwyer inclined manometer.
For each of the samples pressure drop was measured
at 15 flow rates. A Kozney constant form drag plot for
these four sets of samples is shown in Fig. 3. Meshes
composed of 8 ␮m nickel fibers, 100–120 mesh acti-
vated carbon powders, and pyrolyzed cellulose were
prepared to test against the prediction of the PMP
equation. Results of four measurements of permeabil-
ity against face velocity on meshes compressed from
33 to 69 solid volume percent are shown in Fig. 4A.
Values of permeability at 1 cm/s face velocity was
Fig. 3. Kozney constant form drag plot of data for meshes of 2, 4, 8 and 12 ␮m micrometals nickel fibers.
37
determined by a curve fit of the data, and are plotted
against the prediction of the PMP equation as a func-
tion of solid volume fraction in Fig. 4B.
3.2. Shape factor of sintered mesh of
elliptical fibers
Fibers from Micrometals are described by the
manufacturer as ribbon-shaped fibers with an as-
pect ratio of 3. The fibers appear to be more of an
elliptical shape. The 8 and 12 ␮m fibers also have
ridges that are about 18 as high as the diameter of
the fibers discernable in the SEM images. To cal-
culate Kozeny constants for these samples a shape
factor for the sintered meshes is required. The shape
factor for an elliptical fiber can easily be calculated
from formulas for the area and perimeter of an ellipse
[16]. Many different geometrical cell models can de-
scribe a mesh of sintered fibers. The model which
will be used to estimate the shape factor of sintered
meshes of metal fibers is a geometrical model of lay-
ers of fibers separated equidistantly in space which
are stacked at right angles to each other. Each set of
two perpendicular layers are separated by the same
distance as the spacing between fibers in each layer.
Considering the overlapping areas of the fibers as lost
during the sintering process allows estimation of the
38 D.R. Cahela, B.J. Tatarchuk / Catalysis Today 69 (2001) 33–39

Fig. 4. (A and B) Pressure drop through 8 ␮m nickel fiber/ACP meshes with comparison to PMP equation 8 ␮m nickel fibers, 100–120
mesh Calgon BPL, and 10 ␮m PC remnants; solid volume percentages of nickel, ACP and PC of: 7.1, 91.4 and 1.5 ␮m. Fiber diameters
used in PMP for nickel, ACP, and PC of: 7.6, 137 and 10.4 ␮m. Flow path angle of 20◦ was used.

shape factor of a sintered mesh. The shape factor for
this model is given by the following formula:
√ form drag parameters >0.7 corresponding to a solid
3 b/a
φMEF = √ 
[2 − (2(b/a)αM )/π] (1 + (b/a)2 )/2
(5)
For a solid fraction of zero this formula reduces to the
shape factor of an elliptical fiber, and for an aspect
ratio of 1 it reduces further to a value of 23 , which is
the shape factor for a cylindrical fiber.
The hydrodynamic diameter which is the value
which best fit the permeability data best with a flow
path angle of 45◦ was used in the Kozeny constant
form drag plot. The hydrodynamic diameters are close
to the average diameters calculated from the area of
an ellipse, but slightly smaller for 8 and 12 ␮m fibers.
The hydrodynamic diameters used for nominal 2, 4, 8
and 12 ␮m fibers were 2.30, 3.50, 7.60 and 9.85 ␮m.
These fibers all exhibit about 12 ridges with a height
about 18 the width of the fibers. These ridges can eas-
ily explain the slightly smaller diameter required to
fit the permeability data by calculating a shape factor
for a rough fiber model and calculating an equivalent
smooth fiber diameter with a shape factor of 1.5. The
details of this comparison will be omitted. A Kozeny
constant form drag plot for 2, 4, 8 and 12 ␮m nickel
fibers meshes is shown in Fig. 3. The permeability
data for 2 ␮m nickel fiber meshes, shown as solid cir-
cles, conform to the theoretical curve for all form drag
parameters within about 10%. Kozeny constants for
4, 8 and 12 ␮m fibers show a scatter of about 10% for
volume fraction of 10% or a void volume of 90%.
The scatter in the data is likely due to errors in mea-
suring sample thickness for thin layers used in these
measurements. Using thick plugs of material would
eliminate errors in the bed thickness measurements.
For higher solid volume fractions the data tends to
be above the theoretical line in the Kozeny constant
form drag plot. These points correspond to predicted
pressure drops that are lower than the experimental
data. The predicted pressure drop is lower than the
data due to loss of surface area not being as large as
predicted by the shape factor for a sintered mesh of
elliptical fibers for solid volume fractions greater than
10%. If meshes compacted to solid volume fractions
greater than 10% were sintered again, the pressure
drop through them would likely be much smaller
than that through the meshes sintered only once. This
lower pressure drop would result from loss of surface
area due to sintering of contact points between fibers.
So, the shape factor for sintered meshes of elliptical
fibers given here does not well represent meshes of
metal fibers sintered once and compacted to more
than 10% solid volume fraction.
Permeability for meshes composed of 8 ␮m nickel
fibers, 100–120 mesh activated carbon powders
(ACP), and pyrolyzed cellulose (PC) and the predic-
tion of using the PMP equation show a very good
 D.R. Cahela, B.J. Tatarchuk / Catalysis Today 69 (2001) 33–39
agreement in Fig. 4 for the four samples measured.
The shape factor for a sintered metal mesh was used
for the nickel fibers in the composite meshes based
on the solid volume fraction of the nickel component.
Shape factors for the ACP and PC were considered
as constant at 0.72 and 1.5. A flow path angle of 20◦
was used in the prediction by the PMP equation for
the composite meshes.
4. Optimization of adsorbent material using
the PMP equation
Previously there has not been a modeling tool
available to calculate pressure drop through mix-
tures of fibers and particles, so an optimization of
an adsorbent material for adsorption properties and
pressure drop had to be done solely by experimen-
tation. Combining this model with a prediction of
the concentration breakthrough allows for prediction
of the optimum size of the adsorbent particles. The
adsorption rate constant for a mass transfer limited
adsorption is inversely proportional to the square of
the particle diameter, and the pressure drop is pro-
portional to the square of the particle diameter. Low
pressure drop and high adsorption rate are desired. An
optimum particle size which gives minimum pressure
drop and minimum amount of adsorbent for a required
adsorption capacity can be determined given a break-
through concentration model such as the Bed Depth
Service Time equation first derived by Bohart and
Adams [17].
5. Summary
A model equation that predicts PMP over the entire
range of possible bed voidages has been extended to
predict properties of sintered metal meshes. A cor-
relation of data for sintered meshes of nickel fibers
has been presented in the form of a Kozeny form
drag plot. Comparison of the PMP equation with data
39
from a composite fiber/particle mesh has also been
presented. Use of this model for optimization of ad-
sorbent meshes was briefly discussed.
Acknowledgements
The authors gratefully acknowledge the financial
support of ONR and DARPA through contract ONR
N000014-00-10282.
References
[1] B.J. Tatarchuk, M.R. Rose, A. Krishnagopalan, J.N. Zabasajja,
D. Kohler, US Patent 5 080 963.
[2] B.J. Tatarchuk, US Patent 5 096 663.
[3] B.J. Tatarchuk, M.F. Rose, A. Krishnagopalan, US Patent
5 102 745.
[4] B.J. Tatarchuk, M.F. Rose, G.A. Krishnagopalan, J.N.
Zabasajja, D.A. Kohler, US Patent 5 304 330.
[5] M.W. Meffert, Preparation and characterization of sintered
metal microfiber-based composite materials for heterogeneous
catalyst applications, Ph.D. Thesis, Auburn University,
Auburn, AL, 1998.
[6] C.J. Marrion, D.R. Cahela, S. Ahn, B.J. Tatarchuk, J. Power
Sources 47 (1994) 297–302.
[7] D. Kohler, J. Zabasajja, A. Krishnagopalan, B. Tatarchuk, J.
Electrochem. Soc. 137 (1990) 136.
[8] F.A.L. Dullien, Porous Media: Fluid Transport and Pore
Structure, Academic Press, New York, 1992.
[9] I.F. Macdonald, M.S. El-Sayed, K. Mow, F.A.L. Dullien, Ind.
Eng. Chem. Fund. 18 (1979) 199.
[10] C.P. Kyan, D.T. Watson, R.C. Kinter, Ind. Eng. Chem. Fund.
9 (1970) 596.
[11] I.J. Kozeny, Wasserkraft und Wasserwerischaft 1 (1931) 67.
[12] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport
Phenomena, Wiley, New York, 1960, pp. 196–200.
[13] D.K. Harris, D.R. Cahela, B.J. Tatarchuk, Wet layup and
sintering of metal-containing microfibrous composites for
chemical processing opportunities, IEF2000, Italy, May 2000
and JCOMA 2001, in press.
[14] E. Lord, J. Text. Inst. 46 (1955) T191.
[15] G.W. Jackson, D.F. James, Can. J. Chem. Eng. 64 (1986) 364.
[16] W.H. Beyer, S.M. Selby (Eds.), Standard Mathematical
Tables, 24th Edition, CRC Press, Boca Raton, FL, 1976.
[17] G.S. Bohart, E.Q. Adams, J. Am. Chem. Soc. 42 (1920)
523–544.