The Maxwell-Stefan

Equations
ChEn 6603

Wednesday, February 1, 12 1
 Outline

Diffusion in ideal, binary systems

Particle dynamics
Maxwell-Stefan equations
Ficks Law
Diffusion in ideal multicomponent systems
Example: Stefan tube
Matrix form of the Maxwell-Stefan equations
Ficks Law for multicomponent systems
Reference velocities again

Wednesday, February 1, 12 2
 T&K 2.1.1-2.1.2

Particle Dynamics
Conservation of momentum: u1
m1 (u1 uf 1 ) + m2 (u2 uf 2 ) = 0
m2

For molecules, inelastic

Conservation of kinetic energy (elastic collision): collisions are known by
another name ... what is it? u2
m1 (u21 u2f 1 ) + m2 (u22 u2f 2 ) =0 m1

Solve for final particle velocities: Momentum exchanged in a collision:

u1 (m1 m2 ) + 2m2 u2 m1
m1 (u1 uf 1 ) = m1 u1 (u1 (m1 m2 ) + 2m2 u2 ) ,
uf 1 = , m1 + m2
m1 + m2
2m1 m2 (u1 u2 )
u2 (m2 m1 ) + 2m1 u1 = .
uf 2 = m1 + m2
m1 + m2

Rate of change of Momentum

Sum of forces acting Rate of 1-2
on particles of type
1 per unit volume
momentum of
particles of type 1
per unit volume
exchanged per
collision between
1 and 2
collisions per
unit volume

u1 u2 x1 x2

Wednesday, February 1, 12 3
 Momentum
u1 Sum of forces acting Rate of 1-2
m2
on particles of type
1 per unit volume
exchanged per
collision between
1 and 2
collisions per
unit volume
u2
m1

Assume: dS Z Z
a x1 p a dS = r(px1 ) dV
System pressure is constant S(t) V(t)
Collisions are purely elastic (kinetic energy is V(t) Z
conserved in collisions) = prx1 dV
V(t)
No shear stress (negligible velocity gradients)
Why the (-) sign?

So our force (momentum) balance becomes:

f12: drag coefficient for
px1 x1 x2 (u1 u2 ), drag that particle 1 feels
as a result of interactions
= f12 x1 x2 (u1 u2 ) with particles of type 2

p What is the binary

Define a binary diffusion coefficient as D12 = diffusivity a function of?
f12
x1 x2 (u1 u2 ) Maxwell-Stefan Equations
x1 = What about x2?
D12 for a binary, ideal mixture.

diffusion
driving force drag force on species 1
(resisting diffusion)
for species 1
Wednesday, February 1, 12 4
 Ficks Law - Binary Ideal System
x1 x2 (u1 u2 ) Maxwell-Stefan equations for a binary,
x1 = ideal system at constant pressure.
D12

x2 N1 x1 N2
rx1 =
ct D12
x2 J1 x1 J2
= can you show this?
ct D12

J1
rx1 =
ct D12

Ficks law for a binary, ideal

J1 = ct D12 rx1
system at constant pressure

Wednesday, February 1, 12 5
 Re-Cap

12 can be interpreted as an inverse drag coefficient.

12 = 21 (symmetric due to momentum conservation)
12 depends on the characteristics of species 1 and 2
(molecule shapes, etc.), but not on their relative
compositions.
12 may depend on temperature and pressure.
We call 12 the Maxwell-Stefan diffusivity or Binary
diffusivity.
There are no 1-1 interactions here - 11 is not defined.

Wednesday, February 1, 12 6
 T&K 2.1.3-2.1.4

Multicomponent Systems
x1 x2 (u1 u2 )
Binary system: x1 = u1
D12 m2

Ternary system: must consider 1-2, 1-3, and 2-3 interactions.

u2 u3
x1 x2 (u1 u2 ) x1 x3 (u1 u3 ) m1
x1 =
D12 D13
x1 x2 (u2 u1 ) x2 x3 (u2 u3 ) m3
x2 =
D12 D23

Multicomponent system: must consider i-j interactions.

n
X in general... n
X
xi xj (ui uj ) xi xj (ui uj ) What about i=j?
rxi = di =
j=1
ij j=1
ij
j6=i j6=i

Recall: Ni = xi cui Ji = Ni xi cu
n
X xj Ni xi Nj
di = xi di =
cDij
, Assumptions:
(so far)
j6=i
Constant Pressure
X n
xj Ji xi Jj Ideal mixture (elastic collisions)
=
cDij Conservation of translational energy.
j6=i
Where else could the energy go?
Wednesday, February 1, 12 7
 T&K Example 2.1.1

Example: Stefan Tube

Species balance equations (no reaction):
Air z= i i
= = ni ,
t t
cxi ci
= = Ni
Liquid t t
Mixture z=0 At steady state (1D),
ni = i Convection-
diffusion balance...
Ni = i
Given: ij, xi(z=0), xi(z=),
=0.238 m, T=328.5 K From the Maxwell-Stefan equations:
n
X
Find xi(z) dxi xj Ni xi Nj
=
Acetone (1), Methanol (2), Air (3) dz ct ij
j6=i
x1(z=0)=0.319, x2(z=0)=0.528
D12 = 8.48 mm2 /s
D13 = 13.72 mm2 /s
D23 = 19.91 mm2 /s

Wednesday, February 1, 12 8
 A semi-analytic solution
n
X
Maxwell-Stefan dxi xj Ni xi Nj Normalized z d d d 1 d
= , = =
Equations dz ct ij coordinate: dz d dz d
j6=i

n
X
1 dxi xi Nj x j Ni We need to eliminate xn from the equation
= , so that we have unknowns x1 ... xn-1.
d j=1
ct ij
j6=i

n n n
X1 n
X1
X Nj x n Ni X1 x j Ni Eliminate xn by x = 1
= xi n xj = 1 xi xj
ct ij ct in ct ij substituting:
j6=i j6=i j=1 j6=i

0 1
X n n
X1 nX1 xj Ni
1 dxi Nj Ni @ rearrange a bit, collecting
= xi 1 xi xj A ,
d ct ij ct in ct ij terms on xi, xj.
j6=i j6=1 j6=i
0 1
X n n
X 1
Ni Nj A Ni Ni Ni
= xi @ + + xj , move l over.
ct in ct ij ct in ct in ct ij
j6=i j6=i
0 1
X n n
X 1
dxi @ Ni + Nj A Ni Ni Ni
= = xi + xj
d ct in / ct ij / ct in / ct ij / ct in /
j6=i j6=i | {z } | {z }
| {z } ij i
ii

almost there...
Wednesday, February 1, 12 9
 0 1
dxi @ Ni
n
X
Nj A Ni X1
n
Ni Ni

= + xi + xj
d ct in / ct ij / ct in / ct ij / ct in /
j6=i j6=i | {z } | {z }
| {z } ij i
ii
n
X
d (x) =
Ni
+
Nk
,
= [ ] (x) + () ii
ct in / ct ik /
d k6=i

A system of linear ODEs 1 1
ij = Ni ,
with constant coefficients ct in / ct ij /
(ct, Nj are constant) Ni
i =
ct in /

Analytic solution
h i h ih i 1
(x) = exp [[ ] ] (x0 ) + [exp [[ ] ]] [I] ()
(assuming Ni are all constant)
see T&K 8.3 and Appendix B
Matrix exponential! Note: if we had not eliminated
exp[] [exp(ij)]! the nth equation, we could not
form the inverses required here.
In Matlab, use expm

Algorithm:
Note: we could also solve the
1. Guess Ni 4. If (xl) matches the known equations numerically in step 3 and
2. Calculate [], () boundary condition, we are done. eliminate step 2 (work straight from
3. Calculate (x) at =1 (z=l) Otherwise return to step 1. the original Maxwell-Stefan equations)

Later in the course, we will show another way of getting Ni.

Wednesday, February 1, 12 10
 T&K 2.1.5

Matrix Form of Maxwell-Stefan Equations

n
X xj Ni xi Nj Eliminate Jn from the set of n equations n-1 equations.
di = ,
cDij n 1 n 1
j6=i
Jn = Jj = Ji Jj
X n
xj Ji xi Jj j=1 j=1

= n
j=i

cDij X xj Ji xi Jj
j6=i di = ,
n
ct Dij
Easily shown for the j6=i
di = 0 case we have addressed Xn
xj Xn
Jj Split the summation
i=1 thus far, di = xi. ct di = Ji + xi , into individual terms.
Dij Dij
j=1
j6=i
j=1
j6=i
Recall that we dont
have a ii term!
Only n-1 of these Xn n
X 1
xj Jj Jn
equations are independent. = Ji + xi + xi , Isolate the nth
j=1
Dij j=1
Dij Din diffusive flux.
j6=i j6=i
0 1
For a binary system, we have: Xn
xj
n
X1 Jj xi B X C n 1
eliminated the = Ji + xi
x2 N1 x1 N2 @Ji + Jj A ,
x1 = , nth diffusive flux Dij Dij Din
j=1 j=1 j=1
ct D12 j6=i j6=i j6=i
0 1
x1 N2 x2 N1
x2 = Xn n
X 1
ct D21 Gather Ji B xi xj C 1 1
and Jj terms
= Ji @ + A + x i Jj
Show that these sum to zero. Din j=1
D ij j=1
Dij Din
j6=i j6=i

n
X1
Define diagonal
and off-diagonal = Bii Ji Bij Jj
matrix entries. j6=i
Wednesday, February 1, 12 11

n n 1
xi xj 1 1 n-1 dimensional form
ct di = Ji + + xi Jj , of the Maxwell-Stefan
Din j=1
Dij j=1
Dij Din equations
j=i j=i

n 1
= Bii Ji Bij Jj
j=i

n

xi xj
n-1 dimensional
matrix form:
ct (d) = [B](J) Bii =
Din
+
Dij
,
j=i

1 1
Bij = xi
Dij Din

d1 B1,1 B1,2 B1,n 1 J1
d2 B2,1 B2,2 B2,n J2
1
ct .. = .. .. .. .. ..
. . . . . .
dn 1 Bn 1,1 Bn 1,2 Bn 1,n 1 Jn 1

Note: we can write this in n-dimensional

form, but then [B]-1 cannot be formed.

Wednesday, February 1, 12 12
 T&K 3.2

Ficks Law
Maxwell-Stefan Equations Ficks Law
(matrix form) (matrix form)

ct (d) = [B](J) (J) = ct [B] (d)

1
so far,
= ct [D](x) di = xi.

Some Observations:
For an ideal gas mixture, the ij are largely independent of composition (but are
functions of T and p), while the Dij are complicated functions of composition.
The Fickian diffusion coefficients (Dij) may be negative, while ij 0.
The binary diffusivity matrix is symmetric (ij = ji) but the Fickian diffusivity
matrix is not symmetric (Dij Dji).
Note that ii never enter in to any expression, and have no physical meaning.
However, the Fickian Dii enter directly into the expression for the fluxes, and
represent the proportionality constant between the driving force and the
diffusion flux for the ith component.
ij are independent of reference frame. Dij is for a molar-averaged velocity
reference frame.

Wednesday, February 1, 12 13
 T&K 3.2

Binary/Ternary Comparison
n n
xi = 1 xi = 0
i=1 i=1

Binary Diffusion Ternary Diffusion

x3 = 1 x1 x2

x1
x1 x2 = 1 x1 x2 = 1 x1 x3

z=0 z=L z=0 z=L

x2 = x1 x1 = x2 x3

Wednesday, February 1, 12 14
 Diffusion Regimes
(J) = ct [D](rx)

Binary Diffusion Ternary Diffusion

J1 J1
w
aw

L a
's
's L

reverse k
diffusion Fic
k
Fic

diffusion osmotic
D12 barrier diffusion

x1 x1
"normal"
diffusion

J1 = ct Dx1 J1 = ct D11 x1 ct D12 x2 ,

J2 = ct D21 x1 ct D22 x2 .

Wednesday, February 1, 12 15
 Multicomponent Effects
Ternary Diffusion
J1
(J) = ct [B] 1
(d) s Law
reverse
i ck'
diffusion F
diffusion osmotic
barrier diffusion
J1 D1,1 D1,2 D1,n 1 d1
J2 D2,1 D2,2 D2,n d2
1 x1
.. = ct .. .. .. .. ..
. . . . . . "normal"
diffusion
Jn 1 Dn 1,1 Dn 1,2 Dn 1,n 1 dn 1

For multicomponent effects to be important, Dij xj

O(1)
Dijxj must be significant compared to Diixi. Dii xi

|Dij /Dii | O(1)

xj = 0
Wednesday, February 1, 12 16
 T&K 3.2.4, 1.2.1

Ficks Law & Reference Velocities

How do we write Ficks law in other reference frames?
(J) = c[D](x) Molar diffusive flux relative to a molar-averaged velocity.
(j) = t [D ]() Mass diffusive flux relative to a mass-averaged velocity.

(JV ) = [DV ](c) Molar diffusive flux relative to a volume-averaged velocity.

Option 1: Start with GMS equations and write them for the desired diffusive flux and
driving force. Then invert to find the appropriate definition for [D].

Option 2: Given [D], define an appropriate transformation to obtain [D] or [DV].

xk xn
uo
Bik = i
[D ] = [B uo ] 1 [ ][x] 1 [D][x][ ] 1 [B uo ] ik
k n

= [B ou ][ ][x] 1 [D][x][ ] 1 [B ou ] 1 ou
Bik = ik i 1
n xk
xn k

xi
Vu
Bik = ik Vk Vn
[D ] = [B
V Vu
][D][B Vu
] 1
Vt

= [B V u ][D][B uV ] Vk
uV
Bik = ik xi 1
Vn
Wednesday, February 1, 12 17
 T&K 3.2.3-3.2.4

T&K Example 3.2.1

Given [DV] for the system acetone (1), benzene (2),
and methanol (3), calculate [D].
m3
74.1 106
(J) = (j) = t [D ]()
V1 = V
c[D](x) (J ) =
V2 = 89.4 106
[DV ](c)
mol
m3
mol ,
m3
V3 = 40.7 106 mol
Diffusivities in units of 10 m /s
-9 2

x1 x2 V
D11 V
D12 V
D21 V
D22
[DV ] = [B V u ][D][B V u ] 1
0.350 0.302 3.819 0.420 -0.561 2.133
= [B Vu
][D][B uV
]
0.766 0.114 4.440 0.721 -0.834 2.680

0.533 0.790 4.472 0.962 -0.480 2.569

m3
V1 = 74.1 10 6
mol 0.400 0.500 4.434 1.866 -0.816 1.668

m3 0.299 0.150 3.192 0.277 -0.191 2.368

V2 = 89.4 10 6
mol 0.206 0.548 3.513 0.665 -0.602 1.948
m3
V3 = 40.7 10 6 0.102 0.795 3.502 1.204 -1.130 1.124
mol
0.120 0.132 3.115 0.138 -0.227 2.235

0.150 0.298 3.050 0.150 -0.269 2.250

Wednesday, February 1, 12 18