Sunteți pe pagina 1din 15

Chemical Engineering Science 55 (2000) 5231}5245

Jump conditions at non-uniform boundaries: the catalytic surface


Brian D. Wood , Michel Quintard, Stephen Whitaker *
Pacixc Northwest National Laboratory, Richland, WA 99352, USA
Institut de Me& canique des Fluides de Toulouse, Alle& e du Professeur Camille Soula, 31400 Toulouse, France
Department of Chemical Engineering and Material Science, University of California at Davis, Davis, CA 95616, USA
Received 22 November 1999; accepted 24 May 2000

Abstract

A spatially smoothed jump condition is developed for the process of di!usion and reaction at a catalytic surface where a "rst-order,
irreversible reaction takes place at isolated regions on the #uid}solid interface. The point jump condition for this process is given by
!n ) D
c "kc at the c}i interface,
AG A A A
in which the rate coe$cient k undergoes abrupt changes with position on the #uid}solid interface. The averaging procedure leads to
a spatially smoothed jump condition that takes the form
!n ) D
1c 2A"k 1c 2A at the c}i interface,
AG A A  A
in which the e!ective reaction rate coe$cient is determined by the solution of a closure problem. It is this e!ective reaction rate
coe$cient, times the interfacial area per unit volume, that is measured in a typical experimental study of di!usion and reaction in
a porous catalyst. The solution of the closure problem allows one to relate the intrinsic properties of the catalytic surface to k , and

the results are presented in terms of a surface e!ectiveness factor as a function of a Thiele modulus.  2000 Elsevier Science Ltd. All
rights reserved.

Keywords: Jump condition; Non-uniform boundary; Catalytic surface; Spatial averaging; E!ective reaction rate; Closure problem

1. Introduction tion and the associated closure problem solved for


some representative region of a non-uniform catalytic
Most phase interfaces are non-uniform, i.e., the proper- surface.
ties of the interface vary from point to point. The typical Determination of e!ective reaction rates associated
catalytic surface is clearly non-uniform (Haaland & Will- with non-uniform surfaces is important in a variety of
iams, 1982; Mukesh, Morton, Kennedy & Cutlip, 1984b; processes. Early work in this area was done by Solc and
Dumont, Poriaux & Dragonnier, 1986; Bell, 1990; Gun- Stockmayer (1971), who extended the classical theory of
ter, Niemantsverdriet, Ribeiro & Somorjai, 1997) and the biomolecular solution kinetics (Smoluchowski, 1917) to
chemical reaction rate will be non-zero only in those the case where the reactive sites on molecules were as-
regions where the catalyst has been deposited. Our analy- sumed to be spatially heterogeneous. This subject was
sis of non-uniform catalytic surfaces allows us to relate further explored by Schmitz and Schurr (1972), who
ewective reaction rate coe$cients to the structure of the compared numerical results with experimental measure-
surface and the intrinsic reaction rate coe$cients asso- ments of the e!ective reaction rate coe$cient, and by Hill
ciated with reactive regions. The connection between (1975), who studied the di!usion-controlled rate of
e!ective rate coe$cients and intrinsic properties is ligand}protein association.
achieved by means of a spatially smoothed jump condi- Non-uniform surfaces are important in biological pro-
cesses, and ligand binding to cell surfaces has been
studied in some detail. Berg and Purcell (1977) conducted
* Corresponding author. Tel.: #1-530-752-8775; fax: #1-530-752- the "rst such study and they developed an expression for
1031. the e!ective reaction rate parameter associated with
E-mail address: sfwhitaker@ucdavis.edu (S. Whitaker). chemoreceptors distributed on a microbial cell surface.

0009-2509/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 9 - 2 5 0 9 ( 0 0 ) 0 0 1 6 1 - 5
5232 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

Statistical mechanical methods for predicting the e!ec-


tive reaction rate parameter for ligand binding at cell
surfaces were studied by DeLisi (1980), DeLisi and
Wiegel (1981), Shoup and Szabo (1982), and Northrup
(1988). Zwanzig (1990) made use of e!ective medium
theory to study this biological process. Although the
results from these studies compare well with numerical
simulations (Northrup, 1988), all of these works assume
that the geometry of the system is spherical and that the
reaction sites do not interact with one another. The
catalytic system analyzed in this paper is analogous to
the model for di!usion-controlled ligand binding "rst
proposed by Berg and Purcell (1977), and experimental
results for the ligand-binding process can be interpreted
in terms of our spatially smoothed jump condition and
closure problem in order to obtain intrinsic reaction rate
coe$cients. In the area of supported metal catalysts, Fig. 1. Porous catalyst with a non-uniform catalytic surface.
Kuan, Davis and Aris (1983a,b) analyzed the e!ectiveness
of catalytic archipelagos using a model of surface diwu-
sion and reaction at the boundary between the catalyst
particles and the supporting solid; however, that process a volume-averaged di!usion}reaction equation for the
is quite di!erent from the process of bulk diwusion and porous catalyst (Jackson, 1977), and the jump condition
reaction at the catalyst particles which is studied in this at the catalytic surface plays a key role in the derivation
work. (Whitaker, 1999) of that equation. The development of
The purpose of this paper is to illustrate how jump the point jump condition (Whitaker, 1992) begins with
conditions can be developed for surfaces that are spa- the governing equation for the molar concentration of
tially heterogeneous. Although we focus on the problem species A
of a catalytic surface, the development is kept general so
*c
that the basic results are applicable to other types of  #
) N "R everywhere, (2)
surfaces. The analysis results in a jump condition of the *t  
form
in which the molar #ux can be represented explicitly in
!n ) D
1c 2A"k 1c 2A at the c}i interface, terms of the concentration and velocity according to
AG A A  A
(1) N "c v . (3)
  
where the ewective reaction rate parameter, k , can be
 We assume that Eq. (2) is valid everywhere, i.e., in the i-
predicted from knowledge of the point reaction rate para- phase, in the c-phase, and in the interfacial region where
meter and the structure of the reactive surface site ge- the reaction rate term will be interpreted as a heterogen-
ometry. The remainder of the paper is organized as eous reaction rate. Because Eq. (2) applies everywhere, the
follows: In Section 2, we outline the point equations that molar rate of reaction and the molar #ux will be complex
describe di!usion and reaction within the catalyst and we functions of position, and generally valid constitutive
introduce the jump condition at the point scale. In Sec- equations for R and N will not be available. In
tion 3, we develop the spatially smoothed version of the  
Fig. 2 we have shown a distribution of R in the neigh-
jump condition and in Section 4 we show how the aver- 
borhood of the dividing surface, and we have identi"ed
aged and point expressions for the reaction rate are the thickness of the interfacial region as d. The shaded
related. In Sections 5 and 6, the closure problem used to area under the curve represents the heterogeneous reac-
determine the e!ective reaction rate parameter is derived tion rate which is obtained by integrating R over the
and solved. Finally, in Sections 7 and 8 we present some 
interfacial region and subtracting the contribution from
results and conclusions. the homogeneous reaction rates, R and R . The distri-
A G
bution of R illustrated in Fig. 2 represents the following

special case: (1) there is no homogeneous reaction in the
2. The catalytic surface i-phase, i.e., R "0, (2) there is homogeneous reaction
G
in the c-phase, i.e., R O0, and (3) there is heterogeneous
A
To illustrate the process of spatial smoothing for reaction associated with the c}i interface.
a non-uniform catalytic surface, we consider the c}i In regions that are far enough away from the interface
system shown in Fig. 1. In general, we wish to develop so that the interface has no in#uence on the physical
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5233

The convention used here is that n represents the unit


AG
normal vector directed from the c-phase to the i-phase
and this requires the relation
n "!n . (6)
AG GA
The heterogeneous reaction rate in Eq. (5) is an excess
surface quantity de"ned by the relation

HAG
R
Q
dA"
 4H
R d<!
 4HA
R
A 
d<!
4HG
R
G
d<.
(7)
Here AH represents the area of the dividing surface
AG
illustrated in Fig. 3, while <H and <H are clearly identi-
A G
"ed in Fig. 4. To be useful, Eq. (5) requires a chemical
Fig. 2. Distribution of the reaction rate in the neighborhood of the
kinetic constitutive equation for R , and this is often
dividing surface. Q
determined by experimental observation of the average
behaviour of a catalyst.
In most catalytic processes, mass transfer occurs only
in the c-phase and the catalytic surface can be treated as
quasi-steady and reaction-controlled (Whitaker, 1986a).
For these conditions, the problem under consideration
can be simpli"ed to
*c
A #
) N "0 in the c-phase, (8a)
*t A
B.C. N ) n "!R at the c}i interface, (8b)
A AG Q
in which R can be represented in terms of the bulk
Q
concentration, c . For a "rst order, irreversible reaction
A
Fig. 3. Volume used for determination of the point jump condition. in which species A is consumed, the excess surface rate of
reaction, or the heterogeneous reaction rate, can be ex-
process of mass transfer and reaction, we express Eq. (2) pressed as
as
R "!kc . (9)
Q A
*c
A #
) N "R in the c-phase, (4a) Here we note that the rate constant, k, can be in#uenced
*t A A
by the adsorption rate constant, the desorption rate
*c constant, and a chemical kinetic rate constant. If the
G #
) N "R in the i-phase. (4b) catalytic surface is not quasi-steady and reaction con-
*t G G
trolled, one must replace Eq. (8b) with the complete form
These two equations are not valid in the interfacial region; of the jump condition and make use of an interfacial #ux
however, we will make use of both of them in that region. constitutive equation in order to determine the surface
In order to correct for the error associated with the use of concentration, c , that appears in Eq. (5). Under these
Eqs. (4a) and (4b) in the interfacial region, we need to Q
circumstances, the analysis becomes more complex
construct a jump condition (Truesdell & Toupin, 1960; (Whitaker, 1986a).
Slattery, 1990; Whitaker, 1992; Torres & Herbolzheimer,
1993) which requires that Eq. (2) be satis"ed on the
average in the volume <H illustrated in Fig. 3. The only
constraint placed on <H is that the parallel bounding
surfaces of <H lie in the homogeneous c and i-phases.
When surface transport (Ochoa-Tapia, del RmH o
& Whitaker, 1993) can be neglected, we show in Appen-
dix A that the jump condition is given by
*c
Q "N ) n #N ) n #R
*t A AG G GA Q
at the c}i interface. (5) Fig. 4. Volumes of the c- and i-phases.
5234 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

If the concentration of species A is small compared to l and note that the theoretical development will be
A
the total concentration, the molar #ux is di!usive in restricted by
nature and can be represented by
l l , (12)
N "!D
c . (10) A A
A A A
where l represents the characteristic length for the c-
The use of this expression for the molar #ux in Eqs. (8a) A
phase illustrated in Fig. 1. If this length-scale constraint is
and (8b) leads to not satis"ed, one can form the volume average of Eqs.
*c (11a) and (11b) for the averaging volume shown in Fig.
A "
) (D
c ) in the c-phase, (11a) 1 and proceed directly to the development of a closure
*t A A
problem. The theoretical development is straightforward;
B.C. however, the numerical solution of the closure problem
would be di$cult because of the extensive grid required
!n ) D
c "kc at the c}i interface. (11b)
AG A A A to capture the e!ect of the widely dispersed catalytic sites.
This represents a pore scale description of di!usion and
reaction in a porous catalyst in which the reaction rate 3.1. Volume averaging
coe$cient, k, is a function of position.
The "rst step of the pore-scale averaging process be-
gins by locating an averaging volume at every point in
3. Spatial smoothing at the pore scale space, and one of these averaging volumes is illustrated in
Fig. 5. The size of the averaging volume V illustrated in
For most practical situations, reactions will occur at Fig. 5 must be large enough so that the non-uniformities
catalytic sites such as we have illustrated in Figs. 1 and 5. associated with the surface are spatially smoothed.
Under these circumstances, the boundary condition This generally means that r must be much, much

given by Eq. (8b) or by Eq. (11b) must be spatially larger than the distance between catalytic sites, l , and we
A
smoothed in order to produce a useful pore scale descrip- express this idea by the constraint
tion of the process. This spatially smoothed jump condi-
r l . (13)
tion will contain information about the distribution of  A
catalytic sites illustrated in Fig. 5 and information about In terms of the averaging volume, V, the average of Eq.
the intrinsic reaction rate coe$cient associated with (2) is given by
those sites. A key assumption inherent in the theoretical

  
analysis is that the distance between the metal clusters is 1 *c 1 1
 d<#
) N d<" R d<. (14)
small compared to the pore diameter. Under these cir- V V *t V V
 V V

cumstances, the disparate length scales will allow us to
perform the averaging procedure at two di!erent length Use of the general transport theorem and the spatial
scales in exactly the way that hierarchical systems are averaging theorem leads to
treated (Cushman, 1990; Quintard & Whitaker, 1996).

     
We designate the distance between catalytic sites as * 1 1
c d< #
) N d<
*t V V
 V V


  
1
" R d< . (15)
V V


We can make use of the traditional nomenclature repre-


sented by


1
1t 2" t d<, (16)
 V V


in order to express Eq. (15) as

*1c 2
 #
) 1N 2"1R 2 everywhere. (17)
*t  

Here one must note that 1c 2, 1N 2 and 1R 2 are


  
intrinsic averages. In Fig. 6 we have illustrated averaging
Fig. 5. Averaging volume for smoothing surface non-uniformities. volumes that are located entirely in the homogeneous
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5235

Fig. 7. Volume used for determination of the volume averaged jump


Fig. 6. Averaging volumes located entirely in the homogeneous c and condition.
i-phases.

regions of the c and i-phases. For those averaging vol- process. To be explicit, we note that 1R 2 is dexned by
 Q
umes, Eq. (17) provides the following two transport equa-
tions:
*1c 2A
HH
AG
1R 2 dA"
 Q 4HH
1R 2 d<!
  4HH
A
1R 2A d<
A
A #
) 1N 2A"1R 2A in the c-phase,

(18a)
*t A A ! 1R 2G d<, (21)
HH G
4 G
*1c 2G
G #
) 1N 2G"1R 2G in the i-phase, (18b) where <HH and AHH are illustrated in Fig. 7. The parallel,
*t G A AG
bounding surfaces of <HH must lie in regions of the c- and
in which the intrinsic averages are de"ned by the i-phases that are homogeneous with respect to vol-
ume-averaged quantities, and this is indicated in Fig. 7,

 
1 1 where d represents the thickness of the interfacial region
1t 2A" t d<, 1t 2G" t d<.
A < A G < G and r represents the radius of the averaging volume, V.
A 4A G 4G 
For the typical porous catalyst, our governing di!eren-
(19)
tial equations and boundary condition given by Eqs.
(18a), (18b) and (20) reduce to
3.2. The jump condition Spatially smoothed equations for diwusion and reaction
in a catalyst:
Eqs. (17), (18a) and (18b) are completely analogous to
Eqs. (2), (4a) and (4b), thus by following an analysis *(c )A
A #
) (N )A"0 in the c-phase, (22a)
similar to that presented in Appendix A we can immedi- *t A
ately develop a jump condition analogous to Eq. (5).
B.C.
*1c 2
B.C.  Q "1N 2A ) n #1N 2G ) n #1R 2 1N 2A ) n "!1R 2 at the c}i interface (22b)
*t A AG G GA  Q A AG  Q
and these equations should be thought of as the spatially
at the c}i interface. (20)
smoothed pore-scale equations describing di!usion and
Eqs. (18a), (18b) and (20) form a complete description of reaction in a porous catalyst. When these equations are
the transport and reaction in the porous catalyst where averaged over the volume illustrated in Fig. 1, one ob-
the heterogeneous surface reaction has been spatially tains the classic di!usion}reaction equation that can be
smoothed. Here one must remember that Eqs. (18a) and used with experimental data to determine 1R 2 times
 Q
(18b) are not valid in the interfacial region; however, they the surface area per unit volume.
will be used in that region in conjunction with the jump
condition given by Eq. (20). The jump condition has been
constructed on the basis that Eq. (17) is satis"ed on the 4. Relation between 1R 2 and R
 Q Q
average in the interfacial region. The surface excess
quantities 1c 2 and 1R 2 that appear in Eq. (20) are In addition to obtaining measured values of 1R 2 , it
 Q  Q  Q
de"ned in a manner analogous to the point quantities, is important to relate this quantity to the point value, R ,
Q
c and R , appearing in Eq. (5) (see Appendix A) except since this allows one to determine the e!ect of the distri-
Q Q
for the change of scale that accompanies the averaging bution of the catalytic sites on the measured reaction
5236 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

rate. For the particular case under consideration, i.e., no


homogeneous reaction in either the c- or i-phase, the
de"nition of the point excess surface reaction rate de"ned
by Eq. (7) becomes
Point excess surface reaction rate:

HAG
R dA" R d<.
Q
4H

(23)

The excess surface reaction rate associated with the vol-


ume-averaged transport equation is de"ned by Eq. (21),
and when there is no homogeneous reaction in either the Fig. 8. Spherical averaging volume used to determine 1R 2"x .

c-phase or the i-phase this leads to
Volume-averaged excess surface reaction rate:

 HH
AG
1R 2 dA" 1R 2 d<.
 Q
4H


(24)

The key question at this point is: How do we connect


R as de"ned by Eq. (23) to 1R 2 as de"ned by Eq.
Q  Q
(24)? To answer this question, we note that the volume-
averaged rate of reaction at a point x is de"ned by


1
1R 2"x " R "x y d<. (25)
 V(x)  >
Vx

Here x is the position vector locating the centroid of the


averaging volume and y is the position vector relative to Fig. 9. Evaluation of the average rate of heterogeneous reaction,
the centroid as indicated in Fig. 8. Substitution of Eq. (25) 1R 2 .
 Q
into Eq. (24) leads to an expression relating 1R 2 to the
 Q
point value, R , according to

The magnitude of the averaging volume is a constant

    
1
1R 2 dA" R "x y d< d<. (26) given by
 Q V(x)  >
HH
AG 4HH Vx

V(x)"pr (28)
However, we need the relation between 1R 2 and R ,  
 Q Q
and in order to develop this relation with a minimum of and we can use this representation to express Eq. (27) as
e!ort we will ignore all e!ects of curvature so that Eq.

   
(26) can be expressed as 1 X>D Fp PP EX>(P \P
1R 2 "
 Q 4/3pr
   
X>D 1  X\D F P EX\(P \P
1R 2 " R "x y d< dz. (27)
 Q V(x)  >


X\D 
Vx
;R dg r dr dh dz, (29)
Here we have chosen <HH to be a right circular cylinder 
having a vanishingly small cross-sectional area, AHH, and
AG in which z is the distance measured from the dividing
we have represented the normal direction as the z-direc-
tion. The graphical representation of Eq. (27) is shown in surface to the centroid of the averaging volume. In this
Fig. 9 where the distance D is large enough so that the form of Eq. (27) it is understood that 1R 2 is associated
 Q
with the position x and that R is a function of r, h and
averaging procedure is carried into the homogeneous re-  
gions of the c- and i-phases. Neglecting the e!ects of z. We can immediately interchange the order of integra-
curvature means that d is small compared to the radius of tion in Eq. (29) to express the average rate of heterogen-
curvature, and it also means that the radius associated eous reaction as
with V(x) is small compared to the radius of curvature.

   
This is the type of situation that we have illustrated in 1 Fp PP X>D EX>(P \P
1R 2 "
Fig. 5.  Q 4/3pr
 F P X\D EX\(P \P

 The e!ects of curvature are examined in Appendix B.  


;R dg dz r dr dh. (30)
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5237

Fig. 11. Interfacial region within the domain of integration.


Fig. 10. Domain of integration with respect to distance from the
dividing surface.

When curvature e!ects are neglected, the de"nition of


R given by Eq. (23) can be simpli"ed to
Q


X>B
R " R dz (31)
Q 
X\B
and it is this integral that we must locate within Eq. (30) if
we are to relate 1R 2 to R . Locating this particular
 Q Q
integral in the representation for 1R 2 requires that the
 Q
order of integration in Eq. (30) be changed.
In order to interchange the order of integration in the
last two integrals in Eq. (30), we "rst identify the domain
of integration as indicated in Fig. 10 where we have
simpli"ed the nomenclature by using the function a de-
"ned by
Fig. 12. (a) Domain of integration for non-zero values of R . (b)

a(r)"(r !r, 0)a)r . (32) Transformed domain of integration for non-zero values of R .
  

The integration with respect to z can be simpli"ed on the


basis that the rate of reaction is constrained by
With no loss of generality, the domain of integration
R "0, z*"d/2". (33) represented in Fig. 12a can be transformed in the manner

indicated in Fig. 12b, and this allows us to express Eq.
This means that only values of z between !d/2 and (34) as
#d/2 need to be considered in the evaluation of the

   
integrals contained in Eq. (30), and this domain is illus- X>D EX>? E>? X>B
R dg dz" R dz dg. (35)
trated in Fig. 11.  
The limits of integration for the interfacial region are X\D EX\? E\? X\B
given in more detail in Fig. 12a, and on the basis of these Use of the de"nition of the surface excess rate of reac-
limits we can express the last two integrals in Eq. (30) tion given by Eq. (31) allows us to express Eq. (35) in the
according to form

   
X>D EX>? EB\? XE>?
  
R dg dz " R dz dg X>D EX>? E>?
  R dg dz" R dg"2a(r)R . (36)
X\D EX\? E\B\? X\B  Q Q
X\D EX\? E\?

 
E\B>? XB Substitution of this result into Eq. (30) and making use of
# R dz dg
 the de"nition of a given by Eq. (32) leads to
EB\? X\B

   
EB>? XB 1 Fp PP
# R dz dg. (34) 1R 2 " (2(r !r)R r dr dh. (37)
  Q 4/3pr  Q
E\B>? XE\?  F P
5238 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

When R is a constant over the c}i surface, this result B.C.


Q
correctly simpli"es to
!n ) D
c "kc at the c}i interface. (44b)
AG A A A
1R 2 "R "constant (38)
 Q Q The spatially smoothed pore scale description can be ob-
and for the more general case Eq. (37) indicates that tained from Eq. (10), Eqs. (22a) and (22b) and Eq. (43)
1R 2 is given by a weighted area average of R over the which lead to
 Q Q
c}i surface. We illustrate this idea by arranging Eq. (37)
as *1c 2A
A "
) 1D
c 2A in the c-phase, (45a)
*t A A

 
1 Fp PP
1R 2 " w(r)R r dr dh, (39)
 Q pr Q B.C.
 F P
in which the weighting function is given by !1D
c 2A ) n "1kc 2
A A AG A AG
w(r)"(1!(r/r ). (40) at the c}i interface. (45b)
 
The area average of the weighting function is one, i.e., At this point it is important to remember that 1c 2A and
A
1N 2A are de"ned in terms of the averaging volume
A
 
1 Fp PP illustrated in Fig. 6. For averages de"ned in the homo-
w(r)r dr dh"1 (41)
pr geneous c-phase, we have
 F P
and if R is uncorrelated with the radial position we can
Q 1D
c 2A"D
1c 2A (46)
A A A A
approximate Eq. (39) by
and this allows us to express Eqs. (45a) and (45b) as

 
1 Fp PP
1R 2 " R r dr dh. (42) *1c 2A
 Q pr Q A "
) (D
1c 2A) in the c-phase, (47a)
 F P *t A A
We can also express this result as
B.C.


1
1R 2 " R dA"1R 2 , (43) !n ) D
1c 2A"1kc 2
 Q A Q Q AG AG A A A AG
AG AG
in which A represents the area of the c}i interface at the c}i interface. (47b)
AG
occupied by the spherical averaging volume illustrated in
To develop the closure problem, we make use of the
Fig. 9 when the centroid of the averaging volume is
decomposition given by
located on the interface.
While Eq. (43) has great intuitive appeal, one must c "1c 2A#c (48)
remember that the e!ects of curvature have been ignored A A A
and we have assumed that R is not correlated with the and subtract the averaged equations from the point
Q
radial distance from the centroid of the averaging vol- equations to obtain
ume. This latter simpli"cation is entirely consistent with
*c
the restriction given by Eq. (13). A "
) (D
c ) in the c-phase, (49a)
*t A A

B.C.
5. Closure problem
!n ) D
c "kc !1kc 2
AG A A A A AG
Eq. (43) provides us with a relation between the point
value and the averaged value of the reaction rate. How- at the c}i interface. (49b)
ever, in order to explicitly develop the jump condition of
We now decompose the reaction rate coe$cient in
the form of Eq. (1) we must develop a relation between
terms of an area average and a spatial deviation
the e!ective reaction rate parameter and the point-scale
processes. This is accomplished by the development of k"1k2 #kI (50)
a closure problem. To begin, we return to the pore scale AG
description of the di!usion and reaction process given by and make use of the decomposition given by Eq. (48) to
Eqs. (11a) and (11b) which we repeat here as obtain the following form of Eqs. (49a) and (49b)

*c *c
A "
) (D
c ) in the c-phase, (44a) A "
) (D
c ) in the c-phase, (51a)
*t A A *t A A
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5239

in this study it is assumed to be a constant, k , at the



catalytic surface sites, and to be zero elsewhere on the c}i
interface.
We can follow the work of Ryan, Carbonell and
Whitaker (1981) and others (Carbonell & Whitaker,
1984; Whitaker, 1986b,1999) to represent the spatial devi-
ation concentration as

c "s 1c 2A, (54)


A A A
in which s is referred to as the closure variable or the
Fig. 13. Representative region of a non-uniform catalytic surface. A
mapping variable since it maps 1c 2A onto c . The
A A
boundary value problem that determines this variable is
B.C. given by

!n ) D
c "kc !1kI c 2
s "0, (55a)
AG A A A A AG A
#k 1c 2A at the c}i interface. (51b) B.C. 1
GHI A
   !n ) D
s !ks #1kI s 2 "kI at A , (55b)
 AG A A A A AG AG
Here we have used the simpli"cation indicated by B.C. 2 s "0, z"h. (55c)
A
11c 2A2 "1c 2A (52) Periodicity: s (r#l )"s (r), i"1, 2, 3. (55d)
A AG A A G A
and the length-scale constraints associated with this ap- Average: 1s 2A"0. (55e)
A
proximation are discussed elsewhere (Whitaker, 1999,
Chapter 1). We only need to solve Eqs. (51a) and (51b) in Here we have imposed the condition that the average of
some representative region associated with the catalytic the closure variable is zero, and this is necessary in order
surface, and such a region is illustrated in Fig. 13. In that to evaluate the area integral represented by 1kI s 2 .
A AG
region, we can ignore variations in the molecular di!us- Closure problems of this type are described by Quintard
ivity and we can treat the process as quasi-steady and Whitaker (1993,1995) and by Whitaker (1999, Chap-
(Whitaker, 1999, Chapter 1) so that the closure problem ter 2).
becomes
5.1. Closed form

 c "0, (53a)
A
B.C. 1 Our objective in this study is to develop a representa-
tion for 1kc 2 that can be used with Eqs. (47a) and
A AG
!n ) D
c "kc !1kI c 2 (47b) in order to determine 1c 2A. On the basis of the
AG A A A A AG A
decomposition given by Eq. (48), and the representation
#k 1c 2A at the c}i interface, (53b) given by Eq. (54), we obtain
GHIA
   1kc 2 "1k(1#s )2 1c 2A.
 A AG A AG A
(56)
B.C. 2 c "0, z"h. (53c)
A The decomposition represented by Eq. (50) can be used
Periodicity: c (r#l )"c (r), i"1,2,3. (53d) to express this result in the form
A G A
1kc 2 "(1k2 #1kI s 2 )1c 2A. (57)
Here we have treated the region shown in Fig. 13 as A AG AG A AG A
a spatially periodic model of a non-uniform catalytic
This suggests that we de"ne an e!ective reaction rate
surface. The boundary condition imposed at z"h is
coe$cient according to
based on the idea that the di!erence between c and
A
1c 2A is caused by the heterogeneous chemical reaction k "1k2 #1kI s 2 ,
A CDD AG A AG
(58)
at the c}i interface. This means that c will be zero at
A
some distance, h, from the catalytic surface and this so that our di!usion and reaction process can be de-
distance must be determined by trial and error; however, scribed by
for a di!usive process we know that h will be on the order
of, or less than, l . The reaction rate coe$cient, k, can be *1c 2A
A A "
) (D
1c 2A) in the c-phase, (59a)
an arbitrary function of position on the surface; however, *t A A
5240 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

B.C. B.C. 1 !n ) (D /1k2 )


S !KS #1"0
AG A AG A A
!n ) D
1c 2A"k 1c 2A at A , (64b)
AG A A CDD A AG
at the c}i interface. (59b) B.C. 2 S "0, z"h, (64c)
A
This represents a spatially smoothed, pore-scale descrip- Periodicity: S (r#l )"S (r), i"1,2,3. (64d)
A G A
tion of di!usion and reaction in a porous catalyst having
a non-uniform catalytic surface. In order to use this to Once these two boundary value problems have been
solved, we use the condition on the average of s given by
obtain the classic di!usion}reaction equation (Ryan et A
al., 1981; Whitaker, 1986b), Eq. (59a) must be averaged Eq. (61e) in order to determine U according to
over the averaging volume illustrated in Fig. 1. 1kI s 2 1s2A
U" A AG "! A . (65)
1k2 1S 2A
AG A
6. Solution of the closure problem In terms of this computed quantity and the average value
of the reaction rate coe$cient, 1k2 , the ewective reac-
AG
The closure problem represented by Eqs. (55a)}(55e) is tion rate coezcient can be expressed as
similar to the problem encountered in the study of heat
conduction in two-phase systems (Quintard & Whitaker, k 1s2A
 "1! A . (66)
1993). In order to develop an e$cient solution procedure 1k2 1S 2A
AG A
for Eqs. (55a)}(55e), we de"ne the following dimension-
Di!usion limitations will always cause k to be less than
less quantities: 
1k2 as indicated by the results presented in the next
AG
KI "kI /1k2 , K"k/1k2 , U"1kI s 2 /1k2 , (60) section.
AG AG A AG AG
so that the closure problem can be expressed as

s "0, (61a) 7. Results


A
B.C. 1 In order to characterize non-uniform catalytic surfa-
ces, we de"ne the following two parameters:
!n ) (D /1k2 )
s !Ks #U"KI at A , (61b)
AG A AG A A AG
total projected area covered by catalyst
B.C. 2 s "0, z"h, (61c) e " , (67)
A AG total projected surface area
Periodicity: s (r#l )"s (r), i"1, 2, 3, (61d)
A G A 4(total projected area covered by catalyst)
d " . (68)
Average: 1s 2A"0. (61e) N total perimeter of the catalyst
A
The solution to this problem can be represented in the Here d is reminiscent of the ewective particle diameter
N
form that is often used as a characteristic length scale for heat
and mass transfer in porous media, and in this case we
s "s#US , (62) refer to the quantity de"ned by Eq. (68) as the ewective
A A A
catalytic site diameter. One might argue that both e and
where the functions s and s are determined by the AG
A A d should be de"ned in terms of the actual surface area of
N
following two boundary value problems: the catalyst rather than the projected area; however, we
have chosen the latter for purely practical purposes.
Problem I In Fig. 14 we have shown values of k /1k2 as
 AG

s"0, (63a) a function of u where u is de"ned by


A
B.C. 1 !n ) (D /1k2 )
s!Ks"KI at A , u"1k2 l/d D (69)
AG A AG A A AG AG A N A
(63b) Fig. 14 represents an e!ectiveness factor * Thiele
modulus plot for non-uniform catalytic surfaces. For
B.C. 2 s"0, z"h,
A
(63c) values of u less than one-tenth, the e!ective reaction rate
coe$cient is essentially equal to the area average, 1i2 .
Periodicity: s(r#l )"s(r), i"1,2,3. (63d) AG
A G A For values of u greater than one, di!usion limitations
occur and k decreases with increasing value of the
Problem II 
Thiele modulus. An asymptotic condition is achieved for

S "0, (64a) small values of d /l , and the calculations represented in


A N A
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5241

c surface concentration of species A, mol/m


Q
D mixture molecular di!usivity for species A in
A
the c-phase, m/s
d diameter of circular, catalytic islands, m
A
d e!ective catalytic site diameter, m
N
k "rst-order, heterogeneous reaction rate coe$c-
ient, m/s
1k2 area averaged, "rst-order, heterogeneous reac-
AG
tion rate coe$cient, m/s
k e!ective, "rst-order, heterogeneous reaction

rate coe$cient, m/s
l characteristic length for the c-phase, m
A
l center-to-center distance between circular
A
catalytic islands, m
l i"1,2,3, lattice vectors, m
G
N c v , molar #ux of species A, mol/m s
  
N c v , molar #ux of species A in the homo-
@ @ @
geneous region of the b-phase, b"c,
Fig. 14. E!ective reaction rate coe$cient. i, mol/m
1N 2 intrinsic average of the molar #ux for species A,

mol/m s
Fig. 14 indicate that k /1k2 is independent of d /l for 1N 2@ intrinsic average of the molar #ux for species
 AG N A @
values of this ratio of length scales that are less than 0.2. A in the b-phase, b"c, i, mol/m s
Under these circumstances, there is no interaction be- n unit normal vector directed from the c-phase
AG
tween the catalytic sites illustrated in Fig. 13 and toward the i-phase
k /1k2 becomes a unique function of u. n !n , unit normal vector directed from the
 AG GA AG
i-phase toward the c-phase
r position vector, m
8. Conclusions r radius of the averaging volume, V, m

R molar rate of production of species A,

In this study we have shown how the method of vol- mol/m s
ume averaging can be applied to non-uniform surfaces. R molar rate of production of species A in the
@
The form of the spatially smoothed jump condition for homogeneous region of the
a non-uniform catalytic surface was derived, and a clos- b-phase, b"c, i, mol/m s
ure problem was developed that provides theoretical 1R 2@ intrinsic average, homogeneous reaction rate
@
values of the e!ective reaction rate coe$cient for a "rst (production) for species A in the b-phase,
order, irreversible reaction. The e!ective reaction rate b"c, i, mol/m s
coe$cient, k , is the quantity that one normally R excess surface rate of reaction (production) of
 Q
measures in a typical experiment, and in order to predict species A, mol/m s
this quantity one must have detailed information about 1R 2 spatially smoothed excess surface reaction rate
 Q
the catalytic activity of the surface. (production) of species A, mol/m s
s closure variable that maps 1c 2A onto c
A A A
<H <H#<H, volume used to determine the point
A G
Notation jump condition, m
<H volume of the c-phase, up to the dividing sur-
A
A area of the dividing surface, m face, contained in the volume <H, m
AG
AH area of dividing surface associated with <H, m <HH <HH#<HH, volume used to determine the
AG A G
AHH area of the dividing surface associated with average jump condition, m
AG
<HH, m <HH volume of the c-phase, up to the dividing sur-
A
c molar concentration of species A, mol/m face, contained in the volume <HH, m

c molar concentration of species A in the homo- V small-scale averaging volume used to smooth
@
geneous region of the b-phase, b"c, i, mol/m surface heterogeneities, m
1c 2@ intrinsic average concentration of species A in x position vector locating the centroid of V, m
@
the b-phase, b"c, i, mol/m y position vector relative to the centroid of V, m
c c !1c 2A, spatial deviation concentration z distance measured normal to the c}i interface,
A A A
of species A in the c-phase, mol/m m
5242 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

Greek letters This leads to

  
d
d thickness of the interfacial region, m c d<# N ) n dA" R d<. (A.5)
dt   
u (1k2 l/d D , Thiele modulus 4H H 4H
AG A N A
The volume <H can be expressed in terms of the volumes
of the c and i-phases according to
Acknowledgements <H"<H#<H, (A.6)
A G
while the surface area of <H can be expressed in terms of
Support for BW and SW was provided by the
the external portions of the bounding surfaces of the
ECRILDRD program at Paci"c Northwest National
c and i-phases to obtain
Laboratory. Paci"c Northwest National Laboratory is
operated for the DOE by Battelle Memorial Institute AH"AH#AH. (A.7)
A G
under contract DE-AC06-76RLO 1830. Integration of Eq. (A.2) leads to a variation of Eq. (A.5)
that contains the interfacial area as indicated by

  
d
Appendix A. Derivation of the jump conditions c d<# N ) n dA# N ) n dA
dt A A A A AG
4HA HA HAG
In order to develop the point jump condition at the c}i
interface shown in Fig. 1, we begin with the continuity
equation for species A that is valid everywhere and we
"
4HA

R d<
A
(A.8)

express this equation as and the analogous result for Eq. (A.3) has the form

  
d
*c c d<#
G
N ) n dA#
G G
N
G
) n dA
GA
 #
) N "R everywhere. (A.1) dt
*t   4HG HG HGA

By everywhere we mean that this equation is valid in the


c-phase, in the i-phase, and in the interfacial region. In
"
4HG

R d<.
G
(A.9)

the c-phase we determine the concentration of species Here we have used n to represent the unit normal
GA
A by means of the special form of Eq. (A.1) given by vector directed from the i-phase toward the c-phase, and
we have made use of the convention that
*c
A #
) N "R in the c-phase. (A.2) n "!n ,
GA AG
AH "AH .
GA AG
(A.10)
*t A A
The unit normal vectors, n , n and n are illustrated in
GA A G
By in the c-phase we mean everywhere in the c-phase Fig. 15.
including the interfacial region where Eq. (A.2) is not Our objective in the development of a jump condition
valid. We treat the i-phase in the same manner and is to satisfy the macroscopic equation represented by Eq.
express the governing di!erential equation as (A.5) and not the point equation given by Eq. (A.1). The
procedure consists of solving the point equations given
*c by Eqs. (A.2) and (A.3) subject to the appropriate jump
G #
) N "R in the i-phase. (A.3)
*t G G condition. To accomplish this, we "rst subtract Eqs. (A.8)
and (A.9) from Eq. (A.5) and arrange the result in the
The jump condition at the c}i interface will be construc- form

   
ted so that Eqs. (A.2) and (A.3) will provide a concentra- d
tion "eld that satis"es Eq. (A.1) on the average in the c d<! c d<! c d<
dt  A G
volume <H illustrated in Fig. 3. We can form the integral 4H 4HA 4HG

 
of Eq. (A.1) over <H to obtain
# N ) n dA! N ) n dA
 A A
H HA
  
*c
 d<#
 

) N d<" R d<. (A.4)
H *t   N ) n dA
4 4H 4H !
G G
HG

 
Since the limits of integration associated with <H do not
depend on time, we can use the general transport the- ! (N !N ) ) n dA
H A G AG
orem to interchange di!erentiation and integration in  AG

Eq. (A.4), and we can use the divergence theorem to


express the second volume integral as an area integral.
"
 4H
R d<!
 
4HA
R d<!
A  4HG
R dA .
G  (A.11)
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5243

and from this we extract the point form of this jump


condition given by

*c
Q #
) N "(N !N ) ) n #R
*t Q Q A G AG Q

at the c}i interface. (A.17)

This last step requires that the limits of integration in Eq.


Fig. 15. Unit normal vectors. (A.16) be arbitrary, and while this presents no problem for
a #at interface, complications arise for curved surfaces
that are discussed in Appendix B.
For the typical catalytic surface, one can ignore
This form immediately suggests the dexnitions of the surface di!usion and consider the i-phase to be
following excess surface quantities: impermeable. Under these circumstances Eq. (A.17) sim-
pli"es to
Excess surface concentration:
*c
Q "N ) n #R at the c}i interface.
   
(A.18)
c dA" c d<! c d<! c d<. *t A AG Q
Q  A G
HAG 4H 4HA 4HG
(A.12) When the catalytic surface is quasi-steady and reaction
controlled (Whitaker, 1986a) the jump condition takes
Excess surface yux: the form

  
N ) n "!R at the c}i interface. (A.19)
N ) n dp" N ) n dA! N ) n dA A AG Q
H Q Q H  A A
!  HA and when species A is consumed by a "rst order, irrevers-


ible heterogeneous reaction we have
! N ) n dA. (A.13)
G G
HG N ) n "kc at the c}i interface (A.20)
A AG A
Excess surface rate of reaction:
This is Eq. (7b) in the main body of the paper.

 H
AG
R dA"
Q  4 H
R d<!
  4H
A
R
A
d<
A.1. Volume-averaged jump condition

!
 4 H
G
R d<.
G
(A.14) Given Eq. (A.1) we can form the volume average to
obtain
The use of these three de"nitions in Eq. (A.11) leads to an

  
integral surface transport equation of the form 1 *c 1 1
 d<#
) N d<" R d< (A.21)
V V *t V V
 V V


 
d
c dA# N ) n dp
dt Q Q Q and use of the general transport theorem and the spatial
HAG !H
averaging theorem leads to
"
 (N !N ) ) n dA#
A G AG  R dA.
Q
(A.15)

     
HAG HAG * 1 1
c d< #
) N d<
*t V V
 V V

The surface transport theorem and the surface divergence
theorem (Ochoa-Tapia et al., 1993) can be used to express

  
this result in the form 1
" R d< . (A.22)
V V


 
*c
Q dA#
) N dA
H *t Q Q Use of the traditional nomenclature given by
 AG HAG

  
1
" (N !N ) ) n dA# R dA (A.16) 1t 2" t d<, (A.23)
H A G AG H Q  V 
AG  AG
V
5244 B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245

allows us to express Eq. (A.22) as than the radius of curvature of the surface, and when this
is not the case we encounter di$culties as indicated in
*1c 2 Fig. 17. There we have shown an interface for which the
 #
) 1N 2"1R 2 everywhere. (A.24)
*t   thickness of the interfacial region is larger than the local
radius of curvature. The volume <H illustrated in Fig. 17
From here one needs only to repeat the analysis from Eq. has been constructed using lines perpendicular to the c}i
(A.1) to Eq. (A.19) to obtain the volume-averaged jump interface; however, that volume does not include the
condition given by homogeneous i-phase. Other volumes can be chosen
that will include the homogeneous regions in both the
1N 2A ) n "!1R 2 at the c}i interface. (A.25) c and i-phases; however, they will not permit arbitrary
A AG  Q
values of AH thus we are prevented from extracting Eq.
This is Eq. (21b) in the main body of the paper. AG
(A.17) from Eq. (A.16). At this point of time, it is not clear
how to resolve this di$culty; however, it is clear that our
analysis is restricted to cases in which the mean radius of
Appendix B. E4ect of curvature curvature of the surface is large compared to the thick-
ness of the interfacial region. For the jump condition
In this appendix we examine the simplest case of a cur- given by Eq. (A.19), this thickness is d and it will often be
ved surface, i.e., the surface of a sphere. A two-dimen- small compared to the mean radius of curvature of the
sional representation of this surface is given in Fig. 16 surface. On the other hand, the thickness of the interfacial
where we have shown several possibilities for the volume region associated with the jump condition given by Eq.
<H. In this case it is clear that the limits of integration (A.25) is 2r #d and the restriction that this be small

associated with Eq. (A.16) are arbitrary and we encounter compared to the mean radius of curvature will be more
no di$culty in passing from Eq. (A.16) to Eq. (A.17). In di$cult to satisfy.
Fig. 16 the thickness of the interfacial region is smaller
B.1. Inyuence of the curvature on the heterogeneous rate of
reaction

In order to provide a precise example of the in#uence


of curvature on the surface excess properties, we again
consider the spherical surface illustrated in Fig. 16 and
make use of the de"nition of the heterogeneous rate of
reaction given by Eq. (A.14). For the special case in which
there is no homogeneous reaction, we have

HAG
R
Q 
dA"
4H
R d<.

(B.1)

For the special case of uniform conditions over the surface


of the sphere, no reaction in the i-phase, and R being

only a function of radial position, we can express Eq.
(B.1) as
Fig. 16. Interfacial region associated with a spherical surface.


P
R (4pR )" R 4pr dr. (B.2)
Q  
P0
As an example, we take the rate of reaction to be given by

R "R exp(!3g/d), (B.3)


 
in which g is the distance measured from the surface of
the sphere, i.e.,

g"r!R , r*R . (B.4)


 
Fig. 17. Interfacial region having a thickness greater than the radius of Using the simple expression given by Eq. (B.3) in Eq. (B.2)
curvature of the interface. allows us to express the heterogeneous rate of reaction
B. D. Wood et al. / Chemical Engineering Science 55 (2000) 5231}5245 5245

according to

  

K K
1#6(d/R ) exp(!3m)m dm#3(d/R ) exp(!3m)m dm
 
R "R d/3 K K . (B.5)
Q  GFFFFFFFFFFFFFHFFFFFFFFFFFFFI
             

When the thickness of the interfacial region is small Mukesh, D., Morton, W., Kennedy, C. N., & Cutlip, M. B. (1984b).
compared to the radius of curvature of the interface, Eq. Island models and the catalytic oxidation of carbon monoxide and
carbon monoxide-ole"n mixtures. Surface Science, 138, 237}257.
(B.5) reduces to the expected result that the heterogen- Northrup, S. H. (1988). Di!usion-controlled ligand binding to multiple
eous rate of reaction is independent of the radius of competing cell-bound receptors. Journal of Physical Chemistry, 92,
curvature. 5847}5850.
Ochoa-Tapia, J. A., del RmH o, P. J. A., & Whitaker, S. (1993). Bulk
R "R d/3, dR . (B.6)
Q   and surface di!usion in porous media: An application of the sur-
face averaging theorem. Chemical Engineering Science, 48,
It is important to remember that the thickness of the 2061}2082.
interfacial region associated with the average heterogen- Quintard, M., & Whitaker, S. (1993). One- and two-equation models
eous rate of reaction, 1R 2 , is not d but the larger value for transient di!usion processes in two-phase systems. In Ad-
 Q
represented by 2r #d. Under these circumstances, we vances in heat transfer, vol. 23 (pp. 369}465). New York: Academic

except 1R 2 to be independent of the curvature when Press.
 Q Quintard, M., & Whitaker, S. (1995). Local thermal equilibrium for
2r #dR .
  transient heat conduction: Theory and comparison with numerical
experiments. International Journal of Heat and Mass Transfer, 38,
2779}2796.
References Quintard, M., & Whitaker, S. (1996). Transport in chemically and
mechanically heterogeneous porous media I: Theoretical develop-
Bell, A. T. (1990). The impact of catalyst science on catalytic design and ment of region averaged equation for slightly compressible, single
development. Chemical Engineering Science, 45, 2013}2026. phase #ow. Advances in Water Resources, 19, 29}47.
Berg, H. C., & Purcell, E. M. (1977). Physics of chemoreception. Ryan, D., Carbonell, R. G., & Whitaker, S. (1981). A theory of di!usion
Biophysical Journal, 20, 193}219. and reaction in porous media. A.I.Ch.E. Symposium Series, No. 202,
Carbonell, R. G., & Whitaker, S. (1984). Heat and mass transfer in vol. 71 (pp. 46}62).
porous media. In J. Bear and M. Y. Corapciogla (Eds.), Funda- Schmitz, K. S., & Schurr, J. M. (1972). The role of orientation con-
mentals of Transport Phenomena in Porous Media (pp. 123}198). straints and rotational di!usion in bimolecular solution kinetics.
Dordrecht: Martinus Nijho!. Journal of Physical Chemistry, 76, 534}545.
Cushman, J. H. (1990). Dynamics of yuids in hierarchical porous media. Shoup, D., & Szabo, A. (1982). Role of di!usion in ligand binding to
New York: Academic Press. macromolecules and cell-bound receptors. Biophysics Journal, 40,
DeLisi, C. (1980). The biophysics of ligand-receptor interactions. Quar- 33}39.
terly Review of Biophysics, 13, 201}230. Slattery, J. C. (1990). Interfacial transport phenomena. New York:
DeLisi, C., & Wiegel, F. W. (1981). E!ect of nonspeci"c forces and "nite Springer.
receptor number on rate constants of ligand-cell bound-receptor Smoluchowski, M. v. (1917). Versuch einer mathematischen theorie der
interactions. Proceedings of the National Academy of Sciences, U.S.A., koagulationskinetick kolloider loK sungen. Zeitschrift fuer-
78, 5569}5572. Physikalische Chemie, 92, 129}168.
Dumont, M., Poriaux, M., & Dragonnier, R. (1986). On surface reaction Solc, K., & Stockmayer, W. H. (1971). Kinetics of di!usion-controlled
kinetics in the presence of islands. Surface Science Letters, 169, reaction between chemically asymmetric molecules, I, General the-
L307}L310. ory. Journal of Chemical Physics, 54, 2981}2988.
Gunter, P. L. J., Niemantsverdriet, J. W., Ribeiro, F. H., & Somorjai, Torres, F. E., & Herbolzheimer, E. (1993). Temperature gradients and
G.A. (1997). Surface science approach to modeling supported cata- drag e!ects produced by convection of interfacial internal energy
lysts. In Catalysis reviews: Science and engineering. New York: around bubbles. The Physics of Fluids A, 5, 537}549.
Marcel Dekker. Truesdell, C., & Toupin, R. (1960). The classical "eld theories. In: S.
Haaland, D. M., & Williams, F. L. (1982). Simultaneous measurement Flugge, Handbuch der physik, vol. III, Part 1, New York: Springer.
of CO oxidation rate and surface coverage on Pt /Al O using Whitaker, S. (1986a). Transient di!usion, adsorption and reaction in
 
infrared spectroscopy: Rate hysteresis and CO island formation. porous catalysts: The reaction controlled, quasi-steady catalytic
Journal of Catalysis, 76, 450}465. surface. Chemical Engineering Science, 41, 3015}3022.
Hill, T. L. (1975). E!ect of rotation on the di!usion-controlled rate of Whitaker, S. (1986b). Transport processes with heterogeneous reaction.
ligand-protein association. Proceedings of the National Academy of In S. Whitaker, & A.E. Cassano, Concepts and Design of Chemical
Sciences, U.S.A., 72, 4918}4922. Reactors. New York: Gordon and Breach (Chapter 1).
Jackson, R. (1977). Transport in porous catalysts. New York: Elsevier. Whitaker, S. (1992). The species mass jump condition at a singular
Kuan, D. Y., Davis, H. T., & Aris, R. (1983a). E!ectiveness of catalytic surface. Chemical Engineering Science, 47, 1677}1685.
archipelagos I: Regular arrays of regular islands. Chemical Engineer- Whitaker, S. (1999). The method of volume averaging. Dordrecht: Kluwer
ing Science, 38, 719}732. Academic Publishers.
Kuan, D. Y., Davis, H. T., & Aris, R. (1983b). E!ectiveness of catalytic Zwanzig, R. (1990). Di!usion-controlled ligand binding to spheres
archipelagos II: Regular arrays of random islands. Chemical Engin- partially covered by receptors: An e!ective medium treatment. Pro-
eering Science, 38, 1569}1597. ceedings of the National Academy of Sciences, 87, 5856}5857.

S-ar putea să vă placă și