Problem Solutions in Transport Phenomena: Mass Transfer Problems

For theory relevant to the mass transfer problems below, please refer to the book:
Bird, R. B., Stewart, W. E., and Lightfoot, E. N., "Transport Phenomena", 2nd edition, John
Wiley, New York (2002).

The solutions below will also help you solve some of the problems in BSL (an abbreviation often
used for this classic textbook based on the initials of its authors).

Oxidation of silicon - a diffusion problem in microelectronics

Problem
A diffusion problem that occurs in the field of microelectronics is the oxidation of silicon
according to the reaction Si + O2 SiO2. Consider the the oxidation of a material M given by the
general reaction M + (1/2)x O2 MOx (with x = 2 for silicon oxidation). When a slab of the
material is exposed to gaseous oxygen (species A), the oxygen undergoes a first-order reaction
with rate constant k1" to produce a layer of the oxide (species B). The task is to predict the
thickness  of the very slowly-growing oxide layer as a function of time t using a quasi-steady-
state approach (which suggests that the rate of change of the dissolved oxygen content in the
layer is small compared to the rate of reaction). Let the oxygen, whose dissolved concentration is
cA0 at the free surface of the oxide layer at z = 0, diffuse through the layer as per Fick's law to
reach the reaction surface at z =  as in the figure below.

Figure: Diffusion of oxygen through an oxide layer formed by oxidation of silicon.

a) Write unsteady-state molar balances on the oxide and oxygen over the thickness of the layer.

b) Find the concentration profile of oxygen in the layer at steady-state.

c) Using quasi-steady-state arguments, show that the oxygen concentration at the reaction plane
(z = ) is given by

1
  k "  2k1"
c A0 
D  1 c A 

 AB  xD AB c B
where DAB is the oxygen diffusivity through the oxide film and cB is the molar density of the
oxide layer.

d) Derive an expression for the thickness of the very slowly-growing oxide layer when the
quadratic term on the right-hand side of the above equation is negligible.

e) Reduce the above results for the limiting case of diffusion - controlled oxidation where the
reaction rate constant k1" tends to infinity.

Solution
a)
Step. Unsteady-state molar balance on oxide over thickness of layer

The unsteady-state balance on the moles of oxide is

Rate of moles of oxide generated = rate of moles of oxide accumulated

Thus,

2 d 2
"
k1 c A S   c B S  or "
k1 c A 
x dt x
(1)

where cB is the molar density of the oxide layer and S is the cross-sectional area of the slab. Note
that (2/x) moles of oxide are generated for each mole of oxygen consumed in accordance with
the stoichiometry of the reaction.

Step. Unsteady-state molar balance on oxygen over thickness of layer

The unsteady-state balance on the moles of oxygen is

rate of moles of oxygen in by diffusion - rate of moles of oxygen consumed by reaction = rate of
moles of oxygen accumulated

Thus,

2
 d 
N Az S  k1
"
c A S  
dt 
(2)
where NAz is the molar flux of oxygen in the z-direction. Dividing throughout by S (since the
cross-sectional area is constant) and then using Leibniz formula on the right-hand side of the
above equation, we get


 k1   0
"
N Az c A
(3)

Since the rate of change of the dissolved oxygen content in the layer is small compared to the
rate of reaction, the first term on the right-hand side above is neglected. Then, on substituting
Fick's law for NAz,

dc A d
 D AB  k1 c A  c A
"

dz dt
(4)

where DAB is the oxygen diffusivity through the oxide layer.

b)
Step. Steady-state concentration profile of oxygen in layer

At steady-state in the absence of homogeneous reaction, the species continuity equation and the
diffusion equation simplify to .NA = 0 (where NA is the molar flux of species A) and 2cA = 0.
Since the oxygen concentration is a function of z only, the differential equation in Cartesian
coordinates may be simply integrated twice as follows:

d 2cA dc A
2
0   C1 or c A  C1 z  C 2 (5)
dz dz

The integration constants C1 and C2 are determined using the boundary conditions:

BC 1 : z  0, c A  c A0  C 2  c A0 (6)

c A  c A0
BC 2 : z   , c A  c A  C1  (7)


3
On substituting the integration constants,

dc A c A  c A0
 or c A  c A0   c A0  c A z (8)
dz  

Thus, the steady-state profile for the dissolved oxygen concentration is linear. The oxygen
concentration at the reaction plane is so far unknown and is determined next.

c)
Step. Expression for oxygen concentration at reaction plane
On eliminating dcA/dz (using Eq. 8) and d/dt (using Eq. 1), Eq. 4 yields

c A0  c A 2k1"
D AB  k1" c A  c2
A
 xc B
(9)
On rearranging the above equation, the oxygen concentration at the reaction plane is given by the
following expression.

 k1"  2k1"
c A0    1 c A  c 2A
 D AB  xD AB c B
(10)
d)
Step. Solution of differential equation for thickness of oxide layer
The quadratic term on the right-hand side of the above equation has been shown to be negligible
in the microelectronics literature. Thus, the oxygen concentration at the reaction plane is simply
given by

 k1" 

 D  1
 c A  c A0
 AB 
(11)

It must be emphasized that the above equation can be directly obtained from a simple steady-
state oxygen balance (rather than the unsteady-state balances in Eqs. 2 - 4). In other words, the
quasi-steady-state arguments could include a steady-state balance where the moles of oxygen in
by diffusion are exactly equal to the moles of oxygen consumed by the oxidation reaction.

4
On combining Eq. 1 with Eq. 11, the following ordinary differential equation for (t) is obtained.

 k1"  d 2 k1" c A0

 D  1
 dt 
 AB  xc B
(12)

Noting that  = 0 at t = 0, integration gives

2 D AB 4 D Ab c A 0
 2
 "
  t
k1 xc B
(13)

On solving the above quadratic equation, the oxide layer thickness is obtained as

 D 
2
4 D Ab c A
  

AB
"

 
 k xc B
 1 
(14)

e)
Step. Limiting solution for diffusion - controlled oxidation

For the limiting case of diffusion - controlled oxidation, the reaction rate constant tends to
infinity and the dissolved oxygen concentration tends to zero at the reaction plane. Thus, the
above results at the diffusion - controlled limit yield

"
k1    c A  

(15)

The oxide layer thickness is seen to be predicted by a quadratic law.

5
Diffusion with homogeneous chemical reaction of a solid sphere in a
liquid

Problem
A solid sphere (of radius R and density ) made of substance A (of molecular weight M) is
suspended in a liquid B. Solid A undergoes a first-order homogeneous chemical reaction with
rate constant k1''' being slightly soluble in liquid B. Let cA0 be the molar solubility of A in B, and
DAB be the diffusivity of A in B.

Figure: Diffusion with chemical reaction of sphere in liquid.

a) Establish the concentration profile for A at steady state (i.e., when the mass diffusion is in
exact balance with the chemical reaction).

b) Using a quasi-steady-state approach, derive an expression for the time t required for the sphere
radius to decrease from an initial radius R0 to R.

Solution
a)
Step. Differential equation for diffusion with homogeneous chemical reaction

The problem involves mass diffusion with homogeneous chemical reaction. At steady state, the
equation of continuity for species A in molar quantities simplifies to .NA = RA (where NA is the
molar flux of A and RA is the rate of homogeneous chemical reaction per unit volume). For a first
- order homogeneous reaction in spherical coordinates, the species continuity equation gives

d 2
dr
 
r N Ar  R A r 2   k1"' c A r 2
(1)

where NAr is the molar flux of A in the radial direction and cA is the molar concentration of A. The
above equation can also be simply obtained by a molar balance for A over a thin spherical shell
in the liquid.

6
Neglecting the convection term and describing the diffusion of A by Fick's law (NAr = - DAB
dcA/dr), we get

d  2 dc A  "'
  k1 c A r
2
D AB r
dr  dr 
(2)

where DAB is the diffusivity of A in B.

The differential equation may be rewritten as

1 d  2 dc A 
  a cA  0
2
2 r (3)
r dr  dr 

where a = (k1'''/DAB)1/2.

Step. Concentration profile by solving differential equation

The solution of the above ordinary differential equation is of the form:

C1 ar C 2  ar
cA  e  e (4)
r r

The integration constants C1 and C2 are determined using the boundary conditions:

BC 1: r   , c A  0  C1  0 (5)

c A0 R
BC 2 : r  R, c A  c A0  C2  (6)
exp( aR)

The first boundary condition states that species A is not present far from the sphere, whereas the
second boundary condition states that cA0 is the concentration at the sphere surface based on the
molar solubility of species A in liquid B.

On substituting the integration constants, the steady-state concentration profile is obtained as

c A R exp( ar )
(7)
c A0 r exp( aR )

b)
Step. Expression for time required for decrease in sphere radius

7
An unsteady - state molar balance on A over the sphere gives

- Rate of moles of A out by diffusion = Rate of change of moles of A (due to volume change)

d   4 
 N AR 4R 2   R 3  (8)
dt  M 3 

where  is the sphere density and M is the molecular weight of species A.

Now, NAR is the molar flux of A at the sphere surface (r = R). A quasi-steady-state approach
implies that the sphere radius is changing slowly enough for the system to be considered at
steady state at each moment in time. Therefore, the steady-state concentration profile (Eq. 7) may
be differentiated and NAR obtained from Fick's law can be substituted in Eq. 8. Thus,

dR M M  1  aR 
 N AR   D AB c A 0   (9)
dt    R 

Integration now gives

R R M
R0 1  aR dR   
D AB c A0 t (10)

Note that

R 1  1  1 1 
 1  aR dR  a  1  1  aR  dR  a  R  a ln1  aR   (11)

Using the limits of integration, we get

 1  aR  a 2 MD AB c A0
 a  R  R0   ln    t (12)
 1  aR0  
On rearranging, the time required for the sphere radius to decrease from R0 to R is finally
obtained as

 
t  a  R0  R   l
k1"' Mc A 0 
(13)

8
Oxygen metabolism of a spherical bacterial aggregate

Problem
Consider a spherical aggregate of bacterial cells (assumed homogeneous) of radius R. Under
certain circumstances, the oxygen metabolism rate of the bacterial cells is almost zero order with
respect to oxygen concentration and may be approximated by an effective volumetric reaction
rate rO2 = -k0'''. The diffusion may be described by Fick's law with an effective pseudobinary
diffusivity for oxygen in the bacterial medium DO2-M. Neglect transient and convection effects
because the oxygen solubility is very low in the system. Let 0 be the oxygen mass concentration
at the aggregate surface. It is then convenient to define the following dimensionless quantities:


  O2 ,r and N 
k 0"' R 2
0 R (  0 DO 2 M )
If N is large enough, then an oxygen - free core may exist in the aggregate such that  = 0 for  <
0. This requires that  and d/d both be zero at  = 0.

a) Establish the pseudo-steady-state oxygen concentration distribution ().

b) Derive an expression from which the dimensionless oxygen - free core radius 0 may be
determined as a function of N.

c) For what values of N does an oxygen - free core exist?

d) Sketch the oxygen concentration profile  for different values of N.

Solution
a)
Step. Differential equation from oxygen mass balance

On neglecting the transient effects, an oxygen mass balance over a thin spherical shell in the
bacterial aggregate gives

9
 d 2
dr
 
r nO2  rO2 r 2   k 0"' r 2 (1)

where nO2 is the mass flux of oxygen and rO2 is the effective reaction rate for oxygen metabolism
per unit volume.

Neglecting the convective oxygen transport and describing the oxygen diffusion by Fick's law
d O 2
nO2   DO2  M , we get
dr

d  d 
 DO2M  r 2 O2    k0"' r 2 (2)
dr  dr 
where DO2-M is the effective pseudo binary diffusivity for oxygen in the bacterial medium and O2
is the mass concentration of oxygen.

In dimensionless form, the ordinary differential equation is

d  2 d 
    N 2 (3)
d  d 

where the dimensionless quantities are defined as

  O2 , r and N  k 0"' R 2 /(  0 DO 2 M )

0 R
(4)

and 0 is the mass concentration of oxygen at the aggregate surface (r = R).

Step. Oxygen concentration profile by solving differential equation

On integrating,

d N C N 2 C1
   21 or     C2 (5)
d 3  6 

The integration constants are determined using the boundary conditions:

10
 d N 3
BC 1 :    0 , 0  C1   0 (6)
d 3

N 2
BC 2 :    0 ,   0  C 2   0 (7)
2

N N 3
BC 3 :  1,  1  C 2 1   0 (8)
6 3
where 0 is the dimensionless oxygen - free core radius. Note that if N is large enough, then an
oxygen - free core may exist in the aggregate such that  = 0 for  < 0. Since the concentration
and mass flux (i.e., concentration gradient) must be continuous, this requires that d/d = 0 (as in
BC 1) and  = 0 (as in BC 2) at  = 0. The third boundary condition simply states that the mass
concentration of oxygen O2 = 0 at the aggregate surface at r = R.

On substituting the integration constants, the oxygen concentration profile is obtained as

1 

6

N 2
  2 03 /   3 02  or  1 
N
6
 N

1   2   03   1
3 
for    0  0
 
(9)

The first form of the expression for  above is obtained on substituting the integration constants
from BC 1 and BC 2, whereas the second form results from substituting the integration constants
from BC 1 and BC 3. Note that  = 0 for  < 0.

b)
Step. Expression for dimensionless oxygen - free core radius

On eliminating C2 from Eqs. 7 and 8, we get

3 2 1 3
 03  0   (10)
2 2 N

The above expression may be utilized to determine the dimensionless oxygen - free core radius
0 as a function of N as discussed below.

c) Condition for existence of oxygen - free core

On rearranging the above equation, we get

6
N (11)
1  2 03  3 02

From the above equation, it is readily observed that N goes from 6 to infinity when 0 goes from
0 to 1. Thus, the oxygen - free core exists only when N > 6.

11
d) Sketch of oxygen concentration profile and oxygen - free core radius

On eliminating N from Eqs. 9 and 11, the dimensionless oxygen concentration profile is obtained
as

 2  2 03 /   3 02
 for    0  0 (12)
1  2 03  3 02

For a particular value of 0, Eq. 12 may be used to compute () and Eq. 11 to calculate N in
order to obtain the plot given below.

Figure: Oxygen concentration profiles for three different N values.

12