Mass Transfer

Estimation of Diffusivities
Lecture 8, 08.11.2017, Dr. K. Wegner

8. November 2017
8. ESTIMATION OF DIFFUSIVITIES
Diffusion coefficients in gases:
- Around 10-1 cm2/s
- Can be estimated theoretically

Diffusion coefficients in liquids:

- Around 10-5 cm2/s
- Not as reliably estimated as for gases

Diffusion in solids:
- Around 10-8 – 10-10 cm2/s
- Strongly dependent on temperature
- Strongly dependent on material (metals, glasses, polymers, etc.)

Mass Transfer – Diffusivities 9-2

8.1 Diffusion Coefficient (Diffusivity) of Gases
Background:
Kinetic theory of gases, proposed by Maxwell, Boltzmann, Clausius

A gas consists of molecules of diameter d [m], mass m [kg] and

number concentration c [#/cm3] that are in random motion.

The size of the molecules is negligible (diameters much smaller than

the average distance between collisions)

The molecules only interact through perfectly elastic collisions (no

energy transferred).

Mass Transfer – Diffusivities 9-3

Molecules have a distribution of speeds (“Maxwell distribution of
speeds”) with an average molecular velocity of:

8k B T M
v= where m = M = molar mass
π⋅m NA
m = molecular mass
They collide when their centers come within a distance σ of each
other where σ, the collision diameter, is of the order of the actual
diameters d of the molecules.
hit

v σ
d

A = πσ2
collision cross section
miss
Mass Transfer – Diffusivities 9-4
For calculating the frequency of such collisions, first imagine that the
positions of all molecules but one are frozen.

The number of molecules with centers inside the collision tube is

then given by the number concentration c and the tube volume:

(
z* = c ⋅ πσ2 ⋅ v ⋅ ∆t )
z*
The collision frequency is z= = c ⋅ πσ2 ⋅ v
∆t
Since the molecules are not stationary, the average relative
velocity must be used, which is
v rel = 2 ⋅ v

Thus, z = c ⋅ πσ2 ⋅ 2 ⋅ v
Mass Transfer – Diffusivities 9-5
Using the ideal gas law this can be expressed as:

p ⋅ NA 2
z= ⋅ πσ ⋅ 2 ⋅ v (c= (# of molecules) / V)
R⋅T
R
kB = Boltzmann’s constant
NA

If a molecule travels with mean speed v and collides with frequency z,

it spends time 1/z between collisions and travels the mean free path:

v kB ⋅ T 1
λ= = or λ=
z 2 ⋅ πσ2 ⋅ p 2 ⋅ πσ2 ⋅ c

Mass Transfer – Diffusivities 9-6

 Mean free path of air:

λair (1atm, 298K) ≈ 65 nm

Mass Transfer – Diffusivities 9-7
Diffusion coefficient from the kinetic theory of gases
z

c high c low
A

λ λ

Determine the diffusive flux of molecules from a region of high

concentration to low concentration through the area A.

Assuming 1/3 of the molecules have motion in z-direction, then 1/6

of the molecules have motion in positive z direction.

Mass Transfer – Diffusivities 9-8

If the concentration of molecules at the plane A is cA, the number
concentration (c+) of molecules moving towards plane A at a point
one mean free path away from plane A will then be 1/6 of the total
number concentration.

+ 1 dc 
c = c A − λ 
6 dz 

The number concentration of molecules that cross in the negative z-

direction per unit area is:

− 1 dc 
c = c A + λ 
6 dz 

Mass Transfer – Diffusivities 9-9

So the net flux per unit area (molecules / (cm2 s)) can be calculated
by multiplying with the average molecular velocity v:

+ −
(
j = j − j = v⋅ c −c + −
) 1
= − v ⋅λ⋅
3
dc
dz

1
Comparing with Fick’s 1. law gives: D = v ⋅λ
3
1 8k B T 1
or: D = ⋅ “self-diffusivity”
3 π⋅m 2π ⋅ σ 2 ⋅ c

1 8k B T k BT 2 k B3 T3 2
or: D = ⋅ = ⋅ 12 2
3 π⋅m 2π ⋅ σ ⋅ p 3 π m ⋅ σ ⋅ p
2 3

Mass Transfer – Diffusivities 9-10

For species “A” with mA=MA/NA and σA:

2 k B3 ⋅ NA T 3 ⋅ 1/ M A
D AA = ⋅
3 π 3
p ⋅ σ 2A

For diffusion of gas A and gas B:

Then the collision diameter is the arithmetic average of the collision
diameters of the two species present: 1
σ AB = (σ A + σB )
2

2 k B3 ⋅ NA T 3 ⋅ (1/ (2MA ) + 1 (2MB ))

D AB = ⋅
3 π 3
 σ A + σB 
2

p⋅ 
 2 
Mass Transfer – Diffusivities 9-11
However the standard equation is that of Chapman and Enskog:

T 3 ⋅ (1 MA + 1 MB )
D AB = 1.858 ⋅ 10 −3 in cm2/s
p ⋅ σ 2AB ⋅ Ω AB

with T (temperature in K), p (pressure in atm), M (molecular weight

in g/mol). σAB (collision diameter in Å) and ΩAB (collision integral,
dimensionless) are molecular properties obtained best from the
book by Poling et al. “The properties of gases and liquids”.

B.E. Poling, J.M. Prausnitz, J.P. O’Connell, “The properties of gases and liquids”
McGraw-Hill, 5th ed., 2000.

Earlier editions by R.C. Reid, J.M. Prausnitz and B.E. Poling

Mass Transfer – Diffusivities 9-12
The collision integral Ω can be obtained from tables when the energy
of interaction εAB (described by the Lennard-Jones potential, also
tabulated) is known.

ε AB = ε A εB

This equation applies best to non-polar gases (not to H2O and NH3)
and low pressures < 10 atm. For higher pressures, polar gases and
concentration-dependent diffusivity, check the book by Poling et al..

Mass Transfer – Diffusivities 9-13

Source: Cussler “Diffusion”, 3rd edition

Mass Transfer – Diffusivities 9-14

 Diffusion Coefficients from Empirical Correlations
The Chapman-Enskog theory requires the knowledge of Lennard-
Jones potential parameters which are not always known and assumes
non-polar molecules. Other estimates of diffusion coefficients are
based on empirical correlations, like the one of Fuller et al. (1966):
~ ~
10 -3 T1.75 (1/M1 + 1/M2 )1/2
D=
[
p (Σi Vi1 )1/3 + (Σi Vi2 )1/3 ] 2

~
T in K, p in atm, M in g/mol
Vij: Volumes of parts of the
molecule j according to the Table

Fuller, E.N., Schettler, P.D., Giddings, J.C. (1966), Ind. Eng. Chem. 58, 19.
Mass Transfer – Diffusivities 9-15
Diffusion Coefficients in Gases

Experimental values of diffusion coefficients in

gases at 1 atm. Source: Cussler “Diffusion”
Mass Transfer – Diffusivities 9-16
8.2 Diffusivity in Liquids
The estimation of diffusivity in liquids is far more complex and relies
heavily on correlations. We will articulate it here in the frame of the
Stokes-Einstein equation (which is also the basic framework for
particle diffusivity in gases).
This equation describes the diffusion of a spherical particle
undergoing Brownian motion in a quiescent fluid at uniform
temperature.
8.2.1 Particle Diffusivity in Gases
Particle diffusivity in gases is taken as a model system for
molecular diffusion in liquids.

D = f (particle size and gas properties)

A. Einstein (1905), Ann. d. Physik 17, 549.
Mass Transfer – Diffusivities 9-17
Consider particle transport in one dimension, x
t0
Release N0 equally sized particles
t0
t1 at t=0 and observe the distribution
x=0
of n in space and time
t2

∂n ∂2 n
=D 2 (1)
∂t
t3
∂x
x=0

For the boundary conditions at x = 0, x = ∞ and t = 0, the particle

concentration n(x,t) is (see Chapter on Fick’s 2nd law):

N0  x2 
n( x, t) = exp −  (2)
2 πDt  4Dt 
Mass Transfer – Diffusivities 9-18
The mean square displacement of the particles from x=0 at time t is:
+∞
1
x2 = ∫ n(x, t ) dx
2
x (3)
N0 −∞

+∞
− ax 2 1 π
∫
2
Noting that x e dx = (see math tables)
−∞
2a a

eq. (3) becomes using eq. (2):

1 N0 +∞
 x2 
x = ∫ x exp − 4Dt dx
2 2

N0 2 πDt −∞  

1 1 π
= = 2Dt (4)
2 πDt 2 1  1
 
 4Dt  4Dt
Mass Transfer – Diffusivities 9-19
We can measure x 2 by putting spheres in a liquid and follow their
motion through a chequered glass.
The goal is to relate the mean square displacement of a particle with
the energy required for this “job”.
Force balance on a particle in Brownian motion:
du
m = - f ⋅u + F( t ) (5)
dt
(Random) fluctuating
Force acting Frictional resistance force arising from the
on a particle (proportional to u) thermal motion of fluid
molecules
where m is the particle mass, u is the particle velocity, t is time and f
the friction coefficient. For spheres (Stokes law): f = 3π µ dp with the
dynamic viscosity μ of the fluid.
Mass Transfer – Diffusivities 9-20
Now multiply both sides of eq. (5) by the displacement x and divide by
m. For a single particle:
du f F(t )
x = − ux + x (6)
dt m m

define as β = f/m and A = F(t)/m and remember that:

d ( ux ) du dx du
=x +u =x + u2
dt dt dt dt

Using these expressions eq. (6) becomes

d ( ux )
+ β ux= u2 + Ax
dt
Mass Transfer – Diffusivities 9-21
Consider ux as the variable, say, y
dy
+ βy = u2 + Ax
dt

Apply the standard formula for ordinary differential equations and

integrate from t=0 to t
dy
y Q(x)
+ P ( x )= y e ∫
− Pdx  Qe ∫ Pdx dx + C 
dt
⇒ =  ∫ 

and obtain:

β t′
t t
ux = e − βt ∫ u2e dt ′ + e − β t ∫ A x eβ t ′ dt ′ (7)
0 0

where t´ is a variable of integration representing time.

Mass Transfer – Diffusivities 9-22
Average over all particles:
(ux)1 + (ux)2 + 
= ux
N0
Since the mean value of F(t) over a large number of particles
vanishes at any given time, A= F(t)/m → 0 and the second term of eq.
(7) vanishes:

[ ]
t 2
ux = e −βt u2 ∫ eβt d(β t′) = e −βt
1 ′ u t
eβt′ =
β0 β
0

=e −β t
β [
u2 β t
e −1 = ]
u2
β [
1 − e−β t ] (8)

x dx dx 2 1 d x 2
You can also write: ux = = = (9)
dt 2dt 2 dt
Mass Transfer – Diffusivities 9-23
Because the derivative of the mean over particles with respect to time
is equal to the mean of the derivative:

From eq. (8) & (9):

d x 2 u2
2dt
=
β [
1 − e− β t ]
Integrate over time from t = 0 to t

( )
2 2 2
x u u −βt u2 u2 −βt
= t+ 2 e = t + 2 e −1
2 β β β β
0

=
u2 
t + e
β  β
(
1 −β t 
−1

)
Mass Transfer – Diffusivities 9-24
For t >> 1/β (or βt >> 1):
x 2 u2  1  u2  1  u2
≈  t + (0 − 1) =  t −  ≈ t (10)
2 β  β  β  β β
Invoke the equipartition of energy, meaning that the kinetic energy of
particles is equal to that of the surrounding gas molecules:

mu2 k B T (11)
Wkin = = (1 dimensional space!)
2 2

x 2 u2 t u2m kB T
Considering (10) and (11) eq. (4) becomes: D = = = =
2t βt f f

The Stokes-Einstein expression for D relates D to the properties of

the fluid and the particle through the friction coefficient.
Mass Transfer – Diffusivities 9-25
The Stokes-Einstein equation is limited to cases in which the solute is
larger than the solvent. Thus investigators have developed correlations
for cases in which solute and solvent are similar in size, e.g.:

Mass Transfer – Diffusivities 9-26

The French physicist and 1926-Nobel laureate Jean Baptiste Perrin
verified Einstein's theoretical explanation of Brownian motion by
studying the motion of an emulsion.

J. Perrin (1909), "Mouvement Brownian et

Réalité Moléculaire", Annales de Chimie et
de Physique 18, 5-114.
Mass Transfer – Diffusivities 9-27
 x2 kB T R T
D= = = ⋅
2t 3πµdP NA 3πµdP

His experiments allowed Perrin to determine Avogadro's constant as

RT2t
NA = = 7 × 10 23 where R is the gas constant
3πµdP x 2

Thus, he gave an experimental proof of the kinetic theory by

measuring the net displacement. Modern methods show that

NA = 6.023 × 10 23 molecules/mol

Mass Transfer – Diffusivities 9-28

 Diffusion Coefficients in Liquids
Diffusion coefficients at infinite dilution in water at 25°C Diffusion coefficients at infinite dilution in non-aqueous liquids
at 25°C, unless noted

Source: Cussler “Diffusion”

Mass Transfer – Diffusivities 9-29