Lecture 7

Fluid-solid catalytic reactions: Internal mass transfer effects

Copyright ©: All contents in this lecture, no matter the form it takes (whether digital, print,
or media) is protected under copyright law. The contents are only for the purpose of
academic use.

Internal Mass Transfer Limitations:

The total reaction rate for a surface reaction, in the absence of transport limitations, is proportional
to the total surface area. Catalyst materials usually contain many small pores, because this allows
them to have much greater surface area than if they were simply larger, solid pieces. The surface
to volume ratio of an object (with fixed shape) is inversely proportional to its size. For example,
consider the surface to volume ratio of spheres of different sizes. The surface area is 4 r 2 and the
 
volume is  4 / 3  r 3 , so the surface to volume ratio is 3 r or 6 d . So as we decrease the radius
of the sphere, the surface area per volume increases. Imagine that we start with a 3 cm diameter
sphere, and we repeatedly divide it into smaller pieces and roll them into spheres (keeping the
same total amount). The increase in surface area would be as follows:

Sphere Number of Total Total Surface

Radius (cm) Spheres Volume (cm3) Surface to Volume Ratio
Area (cm2) (cm-1)
3.00E+00 1.00E+00 1.13E+02 1.13E+02 1.00E+00
3.00E-01 1.00E+03 1.13E+02 1.13E+03 1.00E+01
3.00E-02 1.00E+06 1.13E+02 1.13E+04 1.00E+02
3.00E-03 1.00E+09 1.13E+02 1.13E+05 1.00E+03
3.00E-04 1.00E+12 1.13E+02 1.13E+06 1.00E+04
3.00E-05 1.00E+15 1.13E+02 1.13E+07 1.00E+05
3.00E-06 1.00E+18 1.13E+02 1.13E+08 1.00E+06
3.00E-07 1.00E+21 1.13E+02 1.13E+09 1.00E+07
3.00E-08 1.00E+24 1.13E+02 1.13E+10 1.00E+08

If we were carrying out a reaction on the surface of these spheres, then the total reaction rate would
scale with the total surface area. So, for example, with the same amount of catalyst material,
spheres of 3 µm diameter would give a reaction rate 10,000 times greater than spheres of 3 cm
diameter. The same effect is achieved by making the catalyst porous. Many catalysts have an
average pore size below 10 nm, which allows them to have surface areas of hundreds of square
meters per gram of catalyst. How many square meters of floor space do you have in your house
or apartment? Unless you are an unusually wealthy student, it is less than the surface area of a
gram of the catalyst that you would find in a catalytic cracking unit.
In order for us to take advantage of the large surface area provided by porous catalysts, reactants
must be able to diffuse into the pores and access all of the surface area. As a first step in analyzing
this diffusive transport, let us consider diffusion and reaction in a single pore of length l and
diameter dp. We will assume that reactant (A) and product (B) are diffusing in and out of the pore,
and that the reaction A → B occurs everywhere on the pore walls, with a rate given by r = kr CA.
This is shown schematically below.

At the mouth of the pore, the reactant concentration is CAs. Within the pore, it is CA(x), where x is
distance inside the pore, and x = 0 at the pore mouth. The diffusive flux of A inside the pore is
dC A
given by N A   DA . At the end of the pore (x = l), we will assume that either the wall is
dx
unreactive or that the pore goes all the way through the catalyst pellet, so that x = l is halfway
through the catalyst, and the concentration profile is symmetric about x = l. In either case, the
dC A
diffusive flux at x = l is zero, and this gives us the condition that  0 at x = l. We can do a
dx
shell balance on a differential length of pore inside the catalyst:

Flux in – Flux out + Production by reaction = 0

 d p2  dC   d p
2
 dC 
 DA  A   DA  A    d p dxkr C A  0
4  dx  x 4  dx  x  dx

 d p2
Dividing this equation through by DA dx and taking the limit as dx goes to zero gives us the
4
differential equation describing the diffusion inside the pore:

d 2 C A  4k r 
  CA  0
dx 2  d p DA 

This is often re-written as

 1
d 2C A  4kr  2

  2C A  0 , where   
dx 2  d D 
 p A

 dC 
We need to solve this 2nd order ODE with the boundary conditions CA (x=0) = CAs and  A   0
 dx  x l
. This equation has the general solution

where C1 and C2 are constants to be determined from the boundary conditions. At x = 0, the
boundary condition gives

CAs = C1 + C2

At x = l, the boundary conditions gives

from which

Substituting this into the 1st b.c. gives

C As C As el C As e l
C2   , so C1  l  l
1  e2l el  e l e e

So, the concentration profile is

We are not actually interested in the details of the concentration profile, but just in the overall
effect on the reaction rate. We therefore want to integrate this concentration distribution over the
pore length to get an average reaction rate that we can compare to the rate we would have with no
diffusion limitations. This is usually described by an effectiveness factor, which is defined as

actual rate k Cr A ( x)dx

1 cosh( (l  x))
l

l 0 cosh(l )
Effectiveness factor     0
 dx
ideal rate kr C As l

The integral of the hyperbolic cosine function is the hyperbolic sine function, so upon evaluating
this integral, we get

The quantity λl that appears in this expression is a dimensionless group that is generally called the
Thiele modulus. It is usually represented by the symbol, . Looking back at our definition of λ,
we see that the Thiele modulus is

A plot of the effectiveness factor vs. the Thiele modulus looks like

When the Thiele modulus is small, there are no diffusion limitations, and the effectiveness factor
is 1. When the Thiele modulus is large, pore diffusion is rate limiting, and the effectiveness factor
is inversely proportional to the Thiele modulus. At a Thiele modulus of 1, the effectiveness factor
is about 0.76 – the rate is about 24% below what it would be if there were no pore diffusion
limitations.

The analysis for a single, ideal pore developed above can be extended to treat diffusion and
reaction in a porous solid. For describing the diffusion of fluids in the pores of a solid catalyst, an
effective diffusion coefficient can be defined so that the flux per unit external surface area of the
pellet is given by the equation
 dC A
N A   DeA
dx

where NA is the flux, x is distance perpendicular to the pellet surface, and DeA is the effective
diffusivity. In most cases, we do not consider a convective component of the transport inside the
pores – the gas in them is assumed to be stagnant. This is a good approximation because of their
small size. However, if there is a change in mole number on reaction so that the moles of product
diffusing out differs from the moles of reactant diffusing in, then there will be a non-zero overall
molar velocity inside the pore. The mass-average velocity will still be zero – since mass is
conserved in the reaction. The effective diffusion coefficient is related to the ordinary diffusion
coefficient by the void fraction of the particle, εs, and the “tortuosity factor” of the pores, τ. The
void fraction gives the fraction of the pellet volume that consists of pores. One minus the void
fraction is the fraction of the pellet volume that consists of solid material. Since the flux above is
defined in terms of the total external surface area of the particle and diffusion in the total particle
volume, the diffusion coefficient needs to be multiplied by the void fraction to account for the fact
that only part of the external surface area consists of pores. Also, the concentration gradient in the
expression above is computed in terms of distance perpendicular to the catalyst surface. However,
the actual pores are not straight cylinders, but follow some sort of tortuous path that can be much
longer than the straight-line distance between points in the catalyst. This is accounted for by
dividing the diffusion coefficient by the ‘tortuosity factor’, τ. This tortuosity is the pore length per
unit change in the x coordinate, which goes in a straight line perpendicular to the external surface
of the catalyst. So, the effective diffusion coefficient may be related to the actual diffusion
coefficient by

and the flux per unit area of catalyst is given by

dC A
N A   DeA
dx

If our porous catalyst is in the form of a slab of thickness 2l, then the analysis that we performed
for the single pore can be applied to the catalyst slab, with the ordinary diffusion coefficient
replaced by the effective diffusion coefficient described above. In that case, the effectiveness
factor is still

tanh  



Where the Thiele modulus is now

Since the pores are not uniform cylinders with a well-defined diameter, it is useful to replace the
pore diameter with an effective pore diameter that is defined based on observable properties of the
catalyst. For an infinitely long cylinder with diameter d, the surface to volume ratio is 4 d . We
usually can measure the surface area per unit mass of our catalyst (Sg). The pellet volume per unit
mass is the inverse of the particle density (1/ρc). So, the average surface area to volume ratio of
the catalyst is Sg ρc. This allows us to define an effective diameter by equating this to the surface
to volume ratio of an infinite cylinder

So the effective diameter is

Substituting this for the particle diameter in the Thiele modulus gives

We could carry out the same analysis for a spherical catalyst pellet that we did for the single pore
and slab. In that geometry, the equation describing the concentration inside the pellet would be

with boundary conditions

dC A
 0 at r = 0, and CA = CAs at r = R (the pellet radius)
dr

This has the solution

 R  sinh( r ) S g  c kr
C A (r )  C As   , where  
 r  sinh( R ) DA s

which leads to an effectiveness factor

3  coth    1

 

where the Thiele modulus is now defined by

The variation of η with φ is qualitatively the same as for the previous case, but for large values of
the Thiele modulus, the effectiveness factor goes to 3/φ rather than to 1/φ. For a long cylindrical
catalyst pellet, the same analysis would lead to an effectiveness factor that can be expressed in
terms of modified Bessel functions. For large values of the Thiele modulus, that expression would
go to 2/φ. Based on these observations, it was hypothesized and later shown, that there is an almost
universal curve for the effectiveness factor vs. the Thiele modulus, if the Thiele modulus is defined
by

where V/S is the ratio of the total pellet volume to the external surface area. With this definition of
the Thiele modulus, the η vs.  curve goes asymptotically to 1/  for large values of  for any pellet
shape, including irregular shapes. The curve then looks like the curve shown before for the slab
geometry, with slight variations near  =1 for different pellet shapes. This curve is shown below:

The entire analysis presented above was based on a single first order reaction in the pore. In fact,
very few surface reactions will have simple first order kinetics, although we may be able to
approximate the kinetics as first order over a limited range of concentrations. A generalized Thiele
modulus can be defined that extends the above analysis (at least approximately) to arbitrary
kinetics. Computing the generalized modulus requires numerically integrating the rate expression,
and therefore it is only slightly simpler than numerically solving for the concentration profile
inside the catalyst. It does have the advantage of showing the relationship between the result for
the complex kinetics and the simple, first-order kinetics analyzed above.

For multiple reactions, the generalized modulus is generally not applicable (except for a few simple
situations like non-interacting parallel reactions). Then, the only rigorous approach is to solve the
continuity equations for the reactants and products to give the concentration profiles of all of the
species inside the pellet. These equations are of the form:

d 2C  dC 
 r  C  , with boundary conditions C  0   Cs , and   0
 dx  x l
2
dx

For a few cases, these can be solved analytically. In general, the equations will have to be solved
numerically. Approximating the derivatives with finite differences leads to a set of algebraic
equations for the concentrations at discrete points within the catalyst pore. These algebraic
equations can be solved using Newton’s method or a related technique. As a general rule, diffusion
limitations degrade selectivity to desired products. In many cases the catalyst used in a process has
been optimized to selectively produce one product rather than another. That is, it catalyzes some
reactions more than others, providing a higher rate for the desired reaction or reactions. As the
reactions become diffusion limited, the rates of different reactions become more nearly equal, since
diffusion coefficients for different species are all approximately the same. The alteration of the
apparent rate constant for a single reaction is discussed below. Qualitatively similar results would
be obtained for most cases of multiple reactions, but quantitative analysis requires solution of the
reaction-diffusion equations. Shape-selective and size-selective catalysts provide a very important
exception to this rule. These catalysts, mostly zeolites, have very small pores in which diffusion
coefficients for different molecules differ widely. If a desired product can diffuse in and out of the
pores much more easily than an undesired product, then diffusion limitations can be used to
actually improve selectivity.