JOURNAL OF CATALYSIS 79, 470-474 (1983)

Effect of Particle Size Distribution on Catalyst Effectiveness

Calculations of catalyst effectiveness in and r + dr. The volume fraction distribu-

heterogeneous catalysis are normally based tion is observed experimentally in particle
on an average particle size. This approach size analyses by conductometric tech-
is quite adequate for situations in which the niques (Coulter counter), for example. Pro-
catalyst particles are nearly uniform in size. vided that the particle density does not vary
In packed bed reactors, for example, the with radius, the volume fraction distribu-
catalyst particles are usually tabletted, ex- tion is identical to the weight fraction distri-
truded, or formed in some other fairly con- bution, fW(r), which is obtained in sedimen-
sistent manner. On the other hand, there tation measurements. Microscopic deter-
are cases in which a distribution of particle minations yield a number fraction dis-
sizes must be considered. In fluidized bed tribution, f(r), such that f(r)dr is the num-
catalytic cracking, the catalyst particles are ber fraction of particles with radii between r
typically distributed over a range of diame- and r + dr. The two distributions, f(r) and
ters from about 10 to 150 pm. Another ex- f”(r), are related according to
ample is catalyst microeffectiveness’ in
pilled zeolite catalysts. Synthetic zeolite fv(r)dr = 2(4n’)f(r)dr
crystals are typically distributed over a I o Eb-W?Wr
range of particle sizes from about 1 to 10
pm. Under such circumstances it is not or
clear how the average particle size should
be calculated. Neither is it clear at what .f&) = -=$ Ar),
point the use of an average particle size be-
comes unacceptable and the entire distribu- where
tion of sizes must be considered. These
questions will be addressed in the present
communication.
We will consider the case of a first-order that is, 7 is the expectation of r”.
catalytic reaction in spherical particles, but It is convenient to define a mean value of
before proceeding further it is important to the effectiveness factor, r),,,, applicable to a
distinguish between several different types distribution of particle sizes according to
of particle size distribution functions. Let
-RAW,obs
f”(r) be the volume fraction particle size dis- %I= -RAWS’ (3)
tribution such that f,(r)dr is the volume
fraction of particles with radii between r where -&w,obs is the reaction rate ob-
served for the distribution of particle sizes
1 Pilled zeolites are “bidisperse structured,” and the and -RAw,S is the rate that would be ob-
overall catalyst effectiveness will generally depend tained in the limit of infinite intraparticle
upon the rates of diffusion within the interstices be-
tween crystals (macroeffectiveness) as well as the dif- mass transfer, i.e., if the reactant concen-
fusion rate within the crystals (microeffectiveness). tration throughout all particles was uniform
See Ors and Dogu (I) for more details. at CA& the particle surface concentration.
470
0021-9517/83/020470-05%03.00/0
Copyright 0 1983 by Academic Press, Inc.
All rights of reproduction in any form reserved.
 NOTES 471

The observed reaction rate can also be ob-

(9)
tained by integrating over the distribution
of particle sizes, i.e.,
Upon substituting from Eqs. (I), (8) and (9)
into Eq. (5) we have
3
rim = 0’ r2f(r)dr.
and since the particle density is assumed to I
be independent of particle size, 7%
J-

-&w,obs = -RAW,S r qfvtrW. (4) and with the aid of Eq. (2)
I

3
It is appropriate to employ the volume dis-
tribution function fV(r) in this integration
since ~/p,(-R~w,s) is the observed reaction
rate per unit volume in particles of radius r. or
From Eqs. (3) and (4) it is immediately ap-
parent that 77m =$ WV
m

(5) where

We note that this result is quite general ppk

since no assumptions have been made re-
h, = rm
J 11)
DAM
garding reaction rate form or shape of the
and
particle size distribution.
For a first-order reaction in a sphere the Fi
kinetics expression on a catalyst weight ba- r, = - 12)
3
sis is
is the average particle radius. Defining the
-RAW = kCA (6) average particle radius in this fashion en-
and the effectiveness factor is given by sures that effectiveness factor plots (71, vs
h,,,) will converge to the same asymptotic
3 1 1 curve at large values of Thiele modulus for
q=h [ tanh(h)--h’ I (7)
all particle size distributions. Aris recom-
mended using a mean radius equivalent to
where h is the Thiele modulus, rm on the basis of an argument similar to
that employed here (2). Although Eq. (12)
h=r P&
J ip was obtained for a first-order reaction, it is
easily shown from an analysis based on the
Note that the nonsubscripted h and r) are generalized Thiele modulus (.?) that the
written for particles of a particular size, r. result is valid for any reaction rate form.

AVERAGE PARTICLE RADIUS EFFECT OF PARTICLE SIZE DISTRIBUTION

A convenient definition of average parti-
cle radius can be obtained by considering
the asymptotic region of strong intraparti-
cle diffusion resistance. For large h Eq. (7)
is closely approximated by
The use of a mean particle size is strictly
valid only in the asymptotic limit of strong
intraparticle diffusion resistance (or in the
trivial case where 77is unity in all particles).
Outside the asymptotic region calculations
472 NOTES

based on a mean particle size can only be

approximate. In order to explore the magni- 7)m = &
tude of this approximation, let us consider a
case in which the volume fraction particle From a rearrangement of Eq. (16) we have
size distribution is log-normal with parame- r = en+tipz
ters (Yand p, i.e., (18)
The Thiele modulus is obtained by substi-
1
f”W = 2/1;; p tuting for r in Eq. (8),

1 ev _ (In r - a)2 h = ea+tiPzeP&

0
; [ 2p2 ’ 1 r > 0. (13) J DA
The log-normal distribution is very com- or
mon in nature. It arises when an event can
be expressed as the product of independent h= ,a-,3212
causative factors, and under such circum-
stances Eq. (13) can be derived by applica-
tion of the Central Limit theorem of mathe- and upon combining this result with Eqs.
matical statistics (4). (14) and (11) we have
To express the mean radius, r,,,, in terms h = h, e\/5pZ+P212
of the distribution parameters, (Yand p, we (19)
first write Eq. (2) for II = 2 and eliminate Equations (19), (7), and (17) can be solved
f(r) with the aid of Eq. (1) to obtain to obtain r), as a function of only two pa-
rameters, h, and /3. The parameter (Yis con-
tained in h,.
The integral of Eq. (17) was evaluated
using the Gauss-Hermite quadrature for-
The integral in this expression can be evalu-
mula (5) for a range of parameter values and
ated after substituting from Eq. (13) for the the results are plotted in Fig. 1. For values
distribution function. Upon combining of /3 less than about 0.5 an average particle
these results with Eq. (12) we have
size can be employed in calculations of ef-
r,=e a-0212 . (14) fectiveness factor with little loss of accu-
racy. The maximum error in this approxi-
The mean effectiveness factor, nm, is ob- mation is only 9%. Errors as great as 32%
tained by substituting for f”(r) in Eq. (5). can result from using an average particle
Thus size when p is unity. For values of p greater
than unity it will be necessary to include the
r)m=
I0&jI entire particle size distribution in calcula-
tions of catalyst effectiveness. Note, how-
(In r - ~4)~ ever, that values greater than unity will sel-
dr. (15)
w2 dom be encountered in practice.
The mean, CL, and variance, cr2, of the
This expression is written in a more con- log-normal distribution are given by the (6)
venient form by making the transformation
of variables: p = eU+P212, (20)

Inr-cu u2 = p2(eP2 - 1). (21)

(16)
z= flP’
Consequently, for given (I! the width of the
Equation (15) can then be written as distribution varies roughly in proportion to
 NOTES 473

004 I I I Illlll I I I I lllll I

0.1 02 04 06 08 IO 2 4 6 8 IO 20
Thiele Modulus, hm
FIG. 1. Effectiveness factor plot for a first-order reaction in spherical particles having a log-normal
(volume fraction) size distribution. 0 = 0 corresponds to particles of uniform size.

the exponential of p2. Care must therefore tions based upon the average radius can be
be exercised in interpreting p as a measure employed for values of /? less than unity.
of distribution width.

NOMENCLATURE
CA = Reactant concentration, moYcm3
SUMMARY DAe = Reactant effective diffusivity,
When calculating the effectiveness factor cm2/s
for a distribution of catalyst particle sizes, h = Thiele modulus for sphere
the average particle radius defined by Eq. k = First-order rate constant, cm3/g . s
(12) should be employed. Then, in the as- n = Exponent in Eq. (2)
ymptotic limit of strong intraparticle diffu- r = Particle radius, cm
sion resistance all plots of effectiveness fac- -R..,w = Reaction rate (disappearance),
tor vs Thiele modulus, regardless of the mol/g . s
shape of the particle size distribution, will z = Variable defined by Eq. (16)
be identical to the plot for a uniform parti- (Y = Parameter in log-normai distribu-
cle size. Outside the asymptotic limit one tion, Eq. (13)
must take into account the distribution of fl = Parameter in log-normal distribu-
particle sizes, Eq. (5), for precise calcula- tion, Eq. (13)
tions. However, an analysis based upon the 7) = Effectiveness factor
log-normal distribution indicates that the p = Mean of log-normal distribution,
use of an average particle radius is quite Eq. (20), cm
reasonable for values of the parameter p pp = Particle density, g/cm3
less than about 0.5 (Fig. 1). If errors of the u2 = Variance of log-normal distribu-
order of 32% can be tolerated, then calcula-’ tion, Eq. (21), cm2
474 NOTES

Subscripts: 4. Chow, V. T., Proc. Amer. Sot. Civil Engrs. 80,536

(1954).
A= Reactant 5. Scheid, F., “Numerical Analysis,” Schaum’s Out-
m= Mean or average value line Ser. in Math. McGraw-Hill, New York, 1968.
obs = Observed quantity 6. Miller, I., and Freund, J. E., “Probability and Sta-
tistics for Engineers,” 2nd ed. Prentice-Hall,
S= Quantity at external particle sur- Englewood Cliffs, N.J., 1977.
face
V= Volumetric basis HENRY W. HAYNES, JR.
W= Weight basis
Chemical Engineering Department
REFERENCES University of Wyoming
1. Ors, N., and Dogu, T., AZChE J. 25, 723 (1979). Laramie, Wyoming 82071
2. Ark, R., Chem. Eng. Sci. 6, 262 (1957).
3. Bischoff, K. B., AZChE J. 11, 351 (1965). Received June 8, 1982; revised October 19, 1982