Chapter 3 Diffusion in concentrated solutions page 3-1

Chapter 3 Diffusion in concentrated solutions

1. Diffusion flux halfway up a capillary

The concentration profile is
z/l
1 - y1  1 - y1l 
= 
1 - y10 1 - y10  (1)

where y 1l = 0, => y 1 = 1 - (1 - y 10 )1 - z/l (2)

Dc
from Eq. (3.3-11) n 1 = - l ln(1 - y 10 ) (3)
Dc
and from Eq. (3.3-12) j 1 = -(1 - y 10 )1-z/l l ln(1 - y 10 ) (4)

 n 1 /j 1 = (1 - y 10 )z/l-1 (5)
(a) at -124 °C, y 10 = 5/760 = 0.006579,
 at halfway, z/l = 0.5, n 1 /j 1 = 1.0033
(b) at -86 °C, y 10 = 400/760 = 0.526316,
 at halfway, z/l = 0.5, n 1 /j 1 = 1.453
2. Gas-oil cracking
The reaction is A  3B
Basic equations:
dn1
0 = - dz (1)

Thus n 1 = constant for all z. In addition,

dc1
n 1 = -D dz + c 1 v0 (2)

Because of the reaction, 3n 1 = -n 2 , thus cv0 = n 1 + n 2 = -2n 1

dy1
Substitute into Eq. (2), -Dc dz = n 1 (1 + 2y 1 )

B. C. z = 0 y 1 = 0 (due to reaction)
z = l y 1 = y 1l (the bulk)
Integration yields ln(1 + 2y 1 ) = -2n 1 z/Dc
Dc
 n 1 = - 2l ln(1 + 2y 1l ) (3)

When y 1l is small, ln(1 + 2y 1l )  2y 1l , thus this equation reduces to

D
n 1 = - l c 1l (diffusing through a thin film)

For evaporation through stagnant solvent, Eq. 3.3-11 shows that (after changing coordinates)
Dc
n 1 = - l ln(1 - y 1l )
Chapter 3 Diffusion in concentrated solutions page 3-2

which is very similar to Eq. (3), the solution of this problem.

3. Interface movement of evaporating benzene

D   e  
2
1
From Eq. (3.3-39), N 1 =   c 1 (sat)
 t  1  V1c1 (sat)   1  erf  
and  is known from Eq. (3.3-38)
-1
 1 
V 1 c 1 (sat) = 1 + 2
 (1 + erf )e 
Here, V 1 c 1 (sat) = y 1 (sat) = 395/760 = 0.5197,
395/760
c 1 (sat) = (82.05)(333) = 1.910-5 mol/cm3

  = 0.3768
 when t = 1 sec, N 1 = 0.104/ 1.285 c 1 (sat) = 4.44710-6 mol/s-cm2
= 3.4710-4 g/cm2-s
At 60 °C, the density of benzene is 0.83g/cm3
 N 1 = 3.4710-4/0.83 = 4.1510-4 cm/s
4. Concentration profile in a membrane reactor
The basic equations are
dn1
0 = - dz (1)

Thus n 1 = constant for all z. In addition,

dc1
n 1 = -D dz + c 1 v0 (2)

B. C. z = 0 n 1 = 0 (at the membrane)


 0 c 1 dz = M/A (the catalyst injected per membrane area)
From B. C. 1, we have n 1 = 0 for all z, thus integrating Eq. (2), we have
0
ln(c 1 /c 10 ) = v0z/D => c 1 = c 10 ev z/D
M c10D Mv0 0
Applying B. C. 2, - A = v0 => c 10 = - AD (v < 0 because of the coordinates)
Mv0 v0z/D
 c 1 = - AD e

5. Diffusion Forms of diffusion coefficient

From Table 3.2-1,
j 1 = c 1 (v 1 - v0) = -Dc 1 => v 1 - v0 = -Dlnc 1
similarly, v 2 - v0 = -Dlnc 2
=> v 1 - v 2 = -Dln(y 1 /y 2 ) (1)
For molar velocity, j*1 = c 1 (v 1 - v*) = -D*cy 1
Chapter 3 Diffusion in concentrated solutions page 3-3

cy1 y1
=> v 1 - v* = -D* c = -D* y = -D*lny 1
1 1

similarly, v 2 - v* = -D*lny 2
=> v 1 - v 2 = -D*ln(y 1 /y 2 ) (1)
By comparing (1) and (1), we thus have D = D*.
6. Fick's law vs. Maxwell-Stefan equation
(a) make a steady mass balance on a shell

4 r 2
r 
0   4 r 2 j1  r   4 r 2 j1  r  r

0  
d 2
dr
r j1  
combine with Fick’s law
dc1
 j1  D
dr
to give
d  2 dc1 
0  D  r 
dr dr 
subject to
r  Rin c1  c1 sat 
r  Rout c1  0
To solve, integrate twice to find
A
c1   B
r
where A and B are integration constants.
Thus
1 1 
c1  r  r 
 out

c1 sat   1 1 
 R  R 
in out

But
Chapter 3 Diffusion in concentrated solutions page 3-4

dc1
j1 r  Rout  D r  Rout
dr
 Dc sat  1
  1 
 Rout   Rout  
 R 1
in 


(b) The mass balance for the Maxwell-Stefan equation is unchanged. Now assuming ideal
solutions with constant total concentration c, we find from Eq. 3.2-6 or 3.2-7,
c1  1 2 v2  v1 
c c
cD
1
 c1n2  c2 n1 
CD
1
 c1 j2  c2 j1 
CD
The solution is dilute and the polymer isn’t moving much so c 1 and j 2 are near zero. Thus c 2
is about equal to c, and
 j1  D c1

This is the same as (a) so the solution is unchanged.

(c) We now must replace Fick’s law and the Maxwell-Stefan relation with
 j1  D11 c1  D12 c2

or

'  2
v  v1   1 '3 v3  v1 
c1 c2 c c
c1 
c D12 c D13

Details of this replacement are in Chapter 7.

7. Oxygen permeation
ARTp0  l2H
From ex 3.5-2, p =  Vl  HDt - 6 
  
l2 (3510-4)2
=> 6D = 2.3,  D = (6)(2.3) = 8.87710-7 cm2/s
ARTp0
Vl HD = 88 mHg = 1.157910 atm
-4

(1.157910-4)(68)(3510-4)
 H= = 0.001038 mole/cm3-atm
17.3%(2.25)(82.05)(298)(8.87710-7)
Chapter 3 Diffusion in concentrated solutions page 3-5

8. Mass balance assumptions

(a) Steady flow and unsteady mass transfer in a tube.
(b) A symmetric washer-shaped region 2 r r z .

(c) The terms are

accumulation = (axial flow in – out)
+ (radial diffusion in – out)
+ (axial diffusion in – out).
Radial, diffusion-induced flow is sensibly neglected.

9. Porous solid desiccant

When using a semi-infinite slab model, from Eq. 3
2
(3.3-39), y = 0.0279x + 0.1759x - 0.011
2.5
D
2
D 1 e-2 R = 0.9994
N1 = c 10 =
t 1 - V1c10 1 + erf  t V1 2
The absorbed amount
M

1.5
M = 0t N1Adt = 2 Dt A/V 1
1
As observed on the right figure, this model does
not fit the experimental data. Instead, as the 0.5
problem states, we assume that the slab contains 0
cylindrical pores in which water can diffuse fast. 0 2 4 6 8
1/2
The flux out of these pores is given by Eq. 5.80 t

in Crank's book (p. 87):

 D D 
j1 =  + 2a - ... c 10
 t 
where a is the radius of the pore. Integration gives
 4Dt Dt
M = M 0 + A 0t j1dt = M0 + A  + 2a c 10
  
As the figure shows, these data fit very well to this equation. Thus,
4D
A c = 0.1759 g/min1/2 = 0.00126 mol/s1/2
 10
The surface area of the slab is
A = 2*202 + 4*0.5*20 = 840 cm2
At 45 °C, p(sat) = 0.0946 atm,
=> c 10 = 0.2*0.0946/82.05/318 = 7.2510-7 mol/cm3
Chapter 3 Diffusion in concentrated solutions page 3-6
2
 0.00126 
 D= -7 = 3.37 cm /s
2
2*840*7.2510 

10. Leaching of copper

c1 Deff  2 c1
Using Ex. 3.5-3, = r2 r
t r r
subject to t = 0, all r, c 1 = c 10
c1
t > 0, r = 0, =0
r
r = R0, c1 = 0
The solution of this problem is (From Carslaw and Jaeger)
c1 R0   (2n+1)R0 + r (2n+1)R0 - r
= 1 - 
r 1 erf - erf 
c10 4Dt 4Dt 

and the total amount leached out should be

t 4R03
R20 0 n1|r = R0dt = 0.8 3 c 10
Dt
From Crank, p. 93, => - R = 0.35
0
0.352*0.12
 t= = 471.2 s = 7.85 min
2.610-6
11. Concentration profile in a slab
The mass transfer equation is (from Eq. (C) in Table 3.4-2)
c1 2c1
=D 2 (1)
t z
c1
bc |
z z = 0
=0 (2)

c 1 (b, t) = 0
ic c 1 (z, 0) = c 10 (3)
Using Laplace transformation, these equations become
d2 
= (4)
d2 
where  = (c 1 - c 10 )/c 10 ,  = z/b, and  = tD/b2. Thus
d
bc | = 0 and (1, ) = 0
d  = 0
ic (, 0) = 0
The solution of Eq. (4) is
(, ) = Bcos e- 
2
Chapter 3 Diffusion in concentrated solutions page 3-7

(-1)n
To satisfy the bc's,  = (n + 1/2), and B =
(n + 1/2)
The final form of the solution is
 (-1)n
(, ) = n=0
 cos(n+1/2) exp(-(n+1/2)22)
(n+1/2)
(-1)n (n+1/2)z  (n+1/2)  Dt
2 2


or c 1 (z, t) = c 10 + c 10 n=0 cos exp- 
(n+1/2)  b   b2 
12. Graham's experiment
The basic equation is
c1 2c1
=D 2 (1)
t z
bc c 1 (0, t) = 0 (a simplified condition) (2)
c 1 (, t) = c 10
ic c 1 (z, 0) = c 10 (3)
This is the semi-infinite slab, and the solution is Eq. 2.3-18
j 1 | z=0 = D/t c 10
The concentration variation in the large beaker is thus
dc1
V dt = Aj1|z=0 = D/t Ac 10
2Ac10 Dt
=> c1 = V

The diffusion coefficient D can be determined from the slope of c 1 vs t.
13. Wool dyeing
This is a cylindrical diffusion problem, known as "Diffusion from a stirred solution of limited
volume". Thus from section 5.3.3 in Crank's book, the total amount of dyestuff in the wool
fiber, M(t), is

4(1 + )

M 2 2
=1- 2 exp(-Dq n t/R ) (5.33)
M n=1 4 + 4+ (qn )
where M  is M at infinite t,
V
=
HR2L
where V is dye bath's volume, and H is the partition coefficient, and q n are the positive roots of
q n J 0 (q n ) + 2J 1 (q n ) = 0 (5.34)
For large  at small Dt/R , Eq. (5.33) approaches to
2

M 2(1 + ) 2 1 2  2 1 1 2 4  3 
=  -  2 +  -  - - 2 + ... (5.40)
M      3  8    
where  = Dt/R2
Chapter 3 Diffusion in concentrated solutions page 3-8

The dye concentration in the bath is

M - M M
C1 = V and C 10 = V
M C1
 =1-C
M 10

Therefore, by plotting (1 - C 1 /C 10 ) vs t1/2, the initial slope is 4(1 + 1/) D/R2.

14. Truffles
This is a point source diffusion problem. The basic equation is
c1 D  2 c1
= 2 r
t r r r
subject to c 1 (r,0) = 0
c 1 (r 0 ,t) = 0
c1
J 0 = -4r2D at z = d,x = y = 0
r
The solution can be found in Carslaw and Jaeger (p. 261) as
J0 r
c1 = erfc
4Dr 4Dt
To meet the boundary condition of the ground surface, define
r + = x2 + y2 + (z + d)2, r - = x2 + y2 + (z - d)2
 r+ r- 

J0 
erfc
4Dt
erfc
4Dt
Thus c1 = - 
4D r+ r- 
c1 J0d erfc(r0/ 4Dt) exp(-r0/ 4Dt)
2
Finally, the flux is j 1 | z=0 = -D |= 
z z=0 r0   - 2
r0 +
r0 Dt 


15. Two immiscible substances in contact

This is a two semi-infinite slab problem.
c1A 2c1A
In slab A, = DA 2
t z
c1B 2c1B
In slab B, = DB 2
t z
These are subject to t = 0, z > 0, c 1A = c 1A0 ,
z < 0, c 1B = 0
t > 0, z = 0, c 1A = c 1B /H, j 1A = j 1B
z = -, c 1A = c 1A0
z = +, c 1B = 0
The solution of this problem is given in Carslaw and Jaeger, p. 88:
Chapter 3 Diffusion in concentrated solutions page 3-9

c1A DA  DB z 
c1A0 = DA + DB 1 + DA erf 4DAt
c1B DA  z 
c1A0 = 
DA + DB 1 + erf 
4DBt
16. Flux away from a dissolving drop
For a dissolving drop, the basic equation is
c1 D  2
= 2 (r n 1 )
t r r
c1
For the pseudo-steady state case where the drop has constant size r, is omitted, and thus
t
r2n 1 = R2N 1 (1)
where N 1 is the total surface flux. But from Fick's law for a stagnant solvent gas, n 2 = 0,
dc1 dc1 n1 + n2 dc1 c1n1
thus n 1 = -D dr + c1v0 = -D dr + c1 c = -D dr + c (2)
Dc dc1 R2N1
By combining (1) and (2), we have n 1 = - c - c dr = r2
1

Subject to r = R, c 1 = c 10
r = , c 1 = 0
c - c1R R2N1  1 1 
After integration, ln  = - Dc R - 
c - c1  
Dc 1 - c10
The surface flux is N 1 = - R ln c 
 
The flux drops as the distance r increases. This is similar to the diffusion through a stagnant
solvent, Eq. 3.3-11.