Chapter 1

=
T
=
(1.38 10 ) ( 373)
23

= 8.6510-9 m
2 AB 2 ( 3.635 10 ) (10 1.013 10 )
10 2
2
P 5

o
= 86.5 A

The Knudsen number is then:

Kn = /d = 86.5/4000 = 0.022 < 0.05.

Therefore, diffusion inside the pores occurs only by ordinary diffusion. At a total pressure of
10 atm, DAB = 0.86/P = 0.86/10 = 0.086 cm2/s. The effective diffusivity is then

DAB 0.4 0.086

DAB,eff = = = 0.014 cm2/s.
2.5

1.5-b Knudsen diffusion in porous solid: When the molecular mean free path is much
larger than the diameter of the pore in which the diffusing molecules reside, the molecules
collide with the wall rather than colliding with other molecules. This mode of transport is
known as Knudsen diffusion. In a straight cylindrical pore of diameter d and length L, the
Knudsen diffusivity, DK,i, is given by

1/ 2
d 8RT
DK,i = for Kn > 5 (1.5-4)
3 Mi

The Knudsen flux is given by

DK ,i ( pi1 pi 2 )
Ni = (1.5-5)
RTL

For binary gas mixtures, the species fluxes are related by

NB MA
= (1.5-6)
NA MB

For Knudsen diffusion in porous solids of porosity and tortuosity ,

DK,i
DK,i,eff = (1.5-7)

DK,i,eff ( pi1 pi 2 )
Ni = (1.5-8)
RTL

1-43
Example 1.5-21 ------------------------------------------------------------------------------

A mixture of O2 (A) and N2 (B) diffuses through the pores of a 2-mm-thick piece of unglazed
porcelain at a total pressure of 0.1 atm and a temperature of 293 K. The average pore
diameter is 0.1m, the porosity is 30.5%, and the tortuosity is 4.39. Estimate the diffusion
fluxes of both components when the mole fractions of O2 are 80% and 20% on either side of
o
the porcelain. The collision diameters for oxygen and nitrogen are 3.467 and 3.798 A ,
respectively.

Solution ----------------------------------------------------------------------------------------------
o
AB = 0.5(A + B) = 0.5(3.467 + 3.798) = 3.632 A

=
T
=
(1.38 10 ) ( 293)
23

= 6.8110-7 m
2 AB 2 ( 3.632 10 ) ( 0.10 1.013 10 )
10 2
2
P 5

o
= 6810 A

The Knudsen number is then:

Kn = /d = 6810/1000 = 6.81 > 0.05.

Therefore, diffusion inside the pores occurs mainly by Knudsen diffusion with diffusivity
given by

1/ 2
d 8RT 10 7 8 8314 293
1/ 2
-5 2 2
DK,A = = = 1.4710 m /s = 0.147 cm /s
3 Mi 3 32

The effective diffusivity is then

DK,A 0.305 0.147

DK,A,eff = = = 0.0102 cm2/s
4.39

The diffusion flux of oxygen is given by

DK,A,eff ( p A1 p A2 ) 0.0102 10 4 (0.8 0.2) 1.013 104

NA = = = 1.27410-3 mol/m2s
RTL (8.314)(293)(2 103 )

The diffusion flux o nitrogen is given by

NB MA 32
= NA = (1.27410-3) = 1.36210-3 mol/m2s
NA MB 28

1
Benitez, J. Principle and Modern Applications of Mass Transfer Operations, Wiley, 2009, p. 60
1-44
1.6 Solute Diffusion through Heterogeneous Medium
We need to know the diffusivity of solute in biological systems that are heterogeneous
medium. The discussion in the following section is general applicable to both the transport of
solute from blood across the capillary wall and the transport of solute through a synthetic
membrane.

Capillary pore
of radius r
r-a
r
Solvent flows across
solute in small pores
increases drag
Solute of
radius a

Figure 1.6-1 Diffusion of solute through capillary pore

Let D be the diffusivity of solute in plasma. When the plasma flows through the capillary
pore, the value of the diffusivity will change to Dm < D due to steric exclusion and
hydrodynamic drag. Steric exclusion refers to the fact that only a fraction of the pore volume
is available to the solute. This fraction is also known as the partition coefficient and is given
(r a)2
by the ratio K = = (1 a/r)2. As the solute diffuses through the pore it experiences
r 2

hydrodynamic drag since the solvent must flow over the solute surface to occupy the volume
left behind by the solute. The Renkin equation provides a relation between the pore
diffusivity Dm and the bulk diffusivity D

a a
3
a
5

Dm/D = (1 a/r)2 1 2.1 r + 2.09 r 0.95 r (1.6-5)

The first term in equation (1.6-5) is the partition coefficient

K(a/r) = (1 a/r)2

The second term in equation (1.6-5) represents the increase in hydrodynamic drag as the
solute diffuses through the pore.

a a
3
a
5

r(a/r) = 1 2.1 + 2.09 0.95

r r r

1-45
The pore diffusivity can be written in terms of the bulk diffusivity, the steric exclusion, and
the hydrodynamic drag as

Dm = DK(a/r) r(a/r)

A graphical depiction of Dm and D is given in Figure 1.6-2. Ficks law for the diffusion of
solute in pores is then given as

1 dC K dC
JS = DmAP = D AP (1.6-6)
dx dx

In this equation, is the tortuosity that accounts for the fact that the pore length is greater
than the membrane thickness.

Interstitial
Plasma fluid space Solute

D0
Dm

D Cell

Capillary Tissue space

Capillary
wall
Figure 1.6-2 Graphical depiction of Dm and D0

There is also a reduction of solute diffusivity as it diffuses through the interstitial fluid space
due to the presence of a variety of macromolecules. The correlation by Brinkman can be used
to estimate the diffusivity D0 of solute through the gel-like structure of interstitial fluid

1
D0/D = r(a/r) =
1
1 + a + (a ) 2
3

The above equation only accounts for the increase in drag as the solute diffuses through the
interstitial fluid where the parameter depends on the composition of the interstitial fluid or
the microstructure of the gel. must be determined from experiment. If the solute enters the
gel from a bulk solution, the steric exclusion must be included in the correlation

a
2
1
D0/D = K(a/r)r(a/r) = exp 1 +

a f 1 + a + 1 (a ) 2

3

1-46
 a
2

where the partition coefficient K(a/r) = exp 1 + is developed by Ogston. In this

a f

expression, af represents the radius of the macromolecules that are assumed to form very
long cylindrical fibers and is the volume fraction of the macromolecules within the gel.

Example 1.6-1. ----------------------------------------------------------------------------------

(Problem 2.8, Fournier)

The following data was obtained by Iwata, Morikawa, and Ikada (1996) for the effective
diffusivity of several solutes through a 5% agarose hydrogel. Show how well the Renkin
equation
a a
3
a
5

Dm/D = (1 a/r)2 1 2.1 r + 2.09 r 0.95 r

represent this data. What is the effective pore size of the gel based on the Renkin equation?

Solute Molecular weight Gel diffusivity, cm2/sec

Glucose 180 1.610-6
Vitamin B12 1,200 1.710-6
Myoglobin 17,000 4.010-7
BSA 69,000 1.010-7
IgG 150,000 1.310-7

The bulk diffusivity can be estimated from the correlation

D = 1.01310-4 (MW)-0.46 (cm2/sec)

The solute radius can be estimated from the formula

1/ 3 1/ 3
3MW 3MW
a = =
4N A 4N A

Solution ------------------------------------------------------------------------------------------

The solution to this problem requires knowledge of nonlinear curve fitting, a topic
reviewed in appendix A. We first need to evaluate the experimental Dm/D or D0/D using the
estimated solute bulk diffusivity.

1-47
 180 0.9294
1200 0.3883

MW = 17000 D = 1.01310-4 (MW)-0.46 = 10 5 0.1147

69000 0.0602
150000 0.0421

0.45 0.4842
0.17 0.4378

Dm,exp = 10 5 0.04 , therefore (Dm/D)exp = 0.3487

0.01 0.1661
0.013 0.3086

The solute radius is computed next

0.0415
0.0781
3MW
1/ 3

a = = 10-6 0.1889
4N A
0.3013
0.3903

a) Find the effective pore radius r in the Renkin equation

a a
3
a
5

Dm/D = (1 a/r)2 1 2.1 r + 2.09 r 0.95 r = f(r)

Since the expression depends on the ratio (a/r), we can multiply a by 107 and later divide the
value of r obtained by 107. Therefore, let

0.415
0.781

a = 1.889

3.013
3.903

We need an initial guess for r. Since the largest solute radius is 3.903(10-7), let r = 5.0. We
can then calculate (Dm/D)cal

1-48
 0.6955
0.4843

(Dm/D)cal = 0.1208

0.0184
0.0038

The goal of curve fitting is to minimize the error between the experimental data and the
calculated values. We need to minimize the following expression

Fobject = [(Dm/D)exp (Dm/D)cal ]2

The above equation is called the objective function. We need to choose a value of r so that
the objective function is a minimum.

0.8
Correlation
Experiment
0.7

0.6

0.5
Dm/D

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
a/r

Figure 1.6-2 Effective diffusivity in 5% agarose hydrogel.

We can use the MATLAB function fminbnd or fminsearch to curve fit the data. We first
must create a function that contains the expression to be minimized. In this case, the
minimizing or objective function is

[(Dm/D)exp (Dm/D)cal ]2

1-49
Table 1.6-2 lists the function poren used in fminbnd. We need to supply the name, poren, of
the Matlab file that contains the function to be minimized and the range of the variable (4,20)
to the function fminbnd.

______ Table 1.6-2 Matlab function for fminbnd ______

function y=poren(r)
a=[.415 .781 1.889 3.013 3.903];
DmD=[.4842 .4378 .3487 .1661 .3086];
x=a./r;
DmDc=((1-x).^2).*(1-2.1*x+2.09*x.^3-.95*x.^5);
y=sum((DmD-DmDc).^2);

>> rm=fminbnd('poren',4,20)
rm =
6.3134

Example 1.6-2. ----------------------------------------------------------------------------------

4
A molecule with a radius of 2.5 nm enters a membrane having pores with a radius of 7.5 nm.
For a temperature of 37oC, and assuming the fluid phase is water, what is the apparent
diffusivity of the molecule in the pores of the membrane?

Solution ------------------------------------------------------------------------------------------

The pore diffusivity can be calculated from

a a
3
a
5

2 1 2.1 + 2.09 0.95
Dm/D = (1 a/r) r r r

For a/r = 2.5/7.5 = 0.333333 Dm/D = 0.166

The molecule molecular weight can be estimated from

1/ 3
3MW 4
a = MW = NAa3
4N A 3

4
MW = (1)(6.0231023)(2.510-7)3 = 39,420
3

The diffusivity of molecule in water is given by

D (cm2/sec) = 1.01310-4(MW)-0.46 = 1.01310-4(39,420)-0.46 = 7.7910-7 cm2/sec

The pore diffusivity is then

Dm = (0.166)(7.7910-7) = 1.2910-7 cm2/sec

4
Fournier, R. L., Basic Transport Phenomena in Biomedical Engineering, Taylor & Francis, 2007, p. 219.
1-50