10.

Basic Theories for Mass Transfer Coefficients

Aim: Connect the mass-transfer correlations (e.g. Tables 8.3-2 and -3
in Cussler’s book) to fundamentals (first part of this class). The
MTCs connect complex mass transfer and fluid flow.

GOAL: Predict the mass transfer coefficient k as a function of the

diffusion coefficient D and the fluid velocity v.

First, we look at MTC’s for fluid-fluid systems that are VERY

important in industrial applications.
Then we investigate models for simple fluid-solid interfaces that can
be rather mathematically elegant, elaborate and detailed but have
limited application in industry.
Fluid-Fluid Interfaces, e.g.
Gas bubbles in Liquid in
Falling film
tank packed tower

Source: Cussler, Chapter 2.5.2 Source: Wikipedia, “Blasensäule” Source: Büchi Glas, Uster
 Gas bubbles in Liquid in
Falling film
tank packed tower
Evaporation, Aeration, Extraction,
Gas scrubbing Gas ab- and desorption, Distillation
Bioreactors
From Table 8.3-2:
0 .5 13 13 13 0.67 0 .5
kz  z  v0  kd  d3  g   /       1  v0  D
 0.69     0.31     k    0.0051    ad0.4
  g   a  
2
D  D  D   D

k ~ D0.5, ~v0.5 k ~ D2/3 k ~ D0.5, ~v0.67

Why is k ~ D1/2 or k ~ D2/3 or k ~ v0.67 ??

19.1.1
TheThe FilmTheory
Film Theory (Nernst,
(Nernst,1904)
1904)

Assumptions:
- All action (fluid flow and mass transfer) occurs in a thin film at the
interface
- Bulk fluid (e.g. gas) – FILM – bulk fluid (e.g. liquid)
- Steady-state flux across film
 n1 z 0  N1  k c 1i  c 1  (1)

This flux can be obtained also in terms of D

(for dilute concentrations)
D
N1  j1 z 0  (c 1i  c 1 ) (2)

D
Comparing equations (1) and (2) gives k (3)

k
Or by rewriting gives  1  Sh (4)
D
This simple theory gives k  D1 BUT all fluid characteristics (e.g.
fluid velocity due to stirring) are in the unknown film thickness  .
This simple theory provides the FRAMEWORK of most MTC’s
as follows:

 mass transfer   characteristic 

     other 
 coefficient   length   F  system 
Sh   
 diffusion   variables 
   
 coefficient 

Applications:

The film theory is used in some practical cases to determine the  .

Example:
CO2 is being scrubbed out of a gas by water flowing through a
packed bed. Calculate the film thickness if 2.3ꞏ10-6 mol/(cm2∙s) of
CO2 are adsorbed when
pCO2= 10 atm, H = 600 atm and DCO2/H2O= 1.9ꞏ10-5 cm2/s.

Solution:
First find the interfacial concentration c1i :
c1i
p1  H  x1  H
c
 c1i 
10atm  600atm  3 

 (1 mol)/(18 cm ) 
c1i = 9.3  10-4 mol/cm3
Calculate k from equation (1):
N1  k(c1i  c1 )

2.3ꞏ10-6 mol/(cm2 s) = k (9.3ꞏ10-4 mol/cm 3 - 0)

k = 2.5ꞏ10-3 cm/s

Now
D 1.9  10 5 cm2 / s 2
  3
 0.76  10 cm
k 2.5  10 cm / s
2

Typically   10 2 cm  VERY IMPORTANT

2 The Penetration Theory (Higbie, 1935)
Assumptions:

• Same as in “Film-theory” but the film

is VERY thick
• Diffusion is important in z-direction
• Convection is important in x-direction

Equation (1): N1  k(c 1i  c 1 )

The solution to this problem was given before in the context of the
semi-infinite slab (Cussler p. 46):
N1  j1 z 0  Dv max /  x  (c 1i  c 1 ) (5)
where N1 is the flux and vmax is the velocity of the liquid at the interface
Note that this flux at the interface is valid at a specific x. To find the
average flux, N1(x) has to be averaged over the entire surface:
L W
1
N1   
W L 0 0
n1 z0 dy  dx

where L is the length of the film in x and W is its width in y. Since n1

does not vary in y, inserting eq. 5:
L
L
1 D  v max 2 D  v max  x
N1  
L0 x
 c dx → N1 
L 
c
0

2 D  v max  L
N1   c
L 
 D  v max
or N1  2   c1i  c1 
 L
D  v max
so k  2 (6)
 L

The L/vmax is called contact time and is not known a priori in

complex situations, as was  in the film theory.

Compare: k  D1/2 (penetration theory)

k D (film theory)
These two theories bracket the experimental data (Table 8.3-2) very
well, almost too well to be accepted.
Equation 6 can be rewritten, assuming that the average velocity is
v0 = 2/3 vmax (true for a laminar slit flow of a Newtonian fluid).
1 12 1 1
L L 3 2  Dv0
 6 2  Lv 0
 2
k 2        
D D2  L      D 
1 1 1
6 2  Lv 0
 2 2
     
    D
1
6 2
   Re Sc
12 12

   1

Pe 2
3 The Surface Renewal Theory (Dankwerts, 1951)
The success of the penetration theory with data despite its restrictive
assumption motivated scientists and engineers to propose alternative
and more realistic models leading, however, to the same variable
dependencies at the end.
Assumption:
The thick film framework is replaced
by TWO regions: interface and bulk.
In the interfacial region mass transfer
takes place according to penetration
theory. Then elements of this region
ARE EXCHANGED with the bulk
region. This is the so-called surface-
renewal process.
The issue is how long the fluid elements stay in the interfacial region
“exposed to penetration”.

 probability of a surface 
 
E(t)dt=  element to be at the 
 surface for time t 
 


E(t) is the residence time distribution, RTD, and 
0
E(t) dt  1

The transfer of interfacial elements into the bulk is random and any
surface element is equally likely to be withdrawn.
By definition the fraction of surface elements at time t is :

  exp   t /  

where  is a characteristic constant equivalent to the average

residence time of an element in the interfacial (surface) region.

Now the fraction  is also the sum of probabilities    E(t)dt
t

Thus the residence time distribution of surface elements is:

 
 
  t 
d d 1  t d
  Et dt     exp     E( t )dt   E( t )dt   Et )
dt  t  dt
     dt   
 
0
 0

 1 
 1  t
So, E( t )  exp  
  

The mass transfer coefficient at the interfacial region is obtained

from the semi-infinite slab model (see eq. 9.1-5 and Cussler p. 46)
as:
D
n1 z 0  N1   c 1i  c 1 
t

The semi-infinite slab model is used here even though the

interfacial region is not infinite. But if the surface is rapidly renewed
and the  is small, then the interfacial region appears as infinite.
The average flux (over all surface elements) is:

N1,av   E( t )  n1 z 0 dt
0

 exp t  D 
     c1i  c 1  dt
0
 t 

D c 1i  c 1  exp t 

  dt
  0 t



D c 1i  c 1     erf t  

 

   1  
 0
 D c 1i  c 1   
N1,av 
 
 
  
 
    erf    erf 0  

 1 0 
D
N1,av  c1i  c1  Thus, k D /


As in the penetration theory, here k  D1/2

Again the residence time  is as unknown as the  in film theory
or the L/vmax (contact time) in the penetration theory.
The major contribution of the surface renewal theory is that it
gives a more REALISTIC physical situation. This gives a better
starting point for development of effective correlations and better
models.
Summary:
Advantages Disadvantages
The Film Theory
Simple; Film thickness 
k good base for is unknown
1 extension
D
The Penetration Theory
Simplest Contact time
D  v max
k  2 including flow (L/vmax) usually
 L unkown

The Surface-Renewal Theory

Similar math to Surface-renewal
k D / penetration theory, rate () is
but better physical unknown
picture
The dependencies of k on D and v, like k ~ D1/2, k ~ D2/3 or k ~ v0.67,
observed in the experimentally-based MTCs are typically not well
reflected by the simple mass transfer models.

These simple models for fluid-fluid interfaces pretend that fluid

motion is incorporated in diffusion and everything is treated as a
thin film or semi-infinite slab problem.

In principle, these two extreme cases should bracket all possible

geometries. Yet, especially the effect of flow (velocity) is usually
not well reflected.

One reason is that the simple theories assume a homogeneous

system while real systems are heterogeneous with respect to
concentration and flow (Schlünder, 1977).
Schlünder E.U., Chem. Eng. Sci. 32, 845-851.