Solutions to the
Diffusion Equation
3.205 L3 11/2/06
Solutions to Ficks Laws
Ficks second law, isotropic one-dimensional
diffusion, D independent of concentration
"c
"2 c
=D 2
"t
"x
Figure removed due to copyright restrictions.
See Figure 4.1 in Balluffi, Robert W., Samuel M. Allen, and W. Craig Carter.
Kinetics of Materials. Hoboken, NJ: J. Wiley & Sons, 2005. ISBN: 0471246891.
Linear PDE; solution requires one initial
!
condition and two boundary conditions.
3.205 L3 11/2/06
Steady-State Diffusion
When the concentration field is independent of
time and D is independent of c, Ficks second
law is reduced to Laplaces equation, "2 c = 0
For simple geometries, such as permeation
!
through a thin membrane, Laplaces
equation can
be solved by integration.
3.205 L3 11/2/06
(a)
Examples of steady-state profiles
Diffusion through a flat plate
Figure removed due to copyright restrictions.
See Figure 5.1 in Balluffi, Robert W., Samuel M. Allen, and W. Craig Carter.
Kinetics of Materials. Hoboken, NJ: J. Wiley & Sons, 2005. ISBN: 0471246891.
(b)
Diffusion through a cylindrical shell
Figure removed due to copyright restrictions.
See Figure 5.2 in Balluffi, Robert W., Samuel M. Allen, and W. Craig Carter.
Kinetics of Materials. Hoboken, NJ: J. Wiley & Sons, 2005. ISBN: 0471246891.
3.205 L3 11/2/06
Error function solution
Interdiffusion in two semi-infinite bodies
Figure removed due to copyright restrictions.
See Figure 4.2 in Balluffi, Robert W., Samuel M. Allen, and W. Craig Carter.
Kinetics of Materials. Hoboken, NJ: J. Wiley & Sons, 2005. ISBN: 0471246891.
Solution can be obtained by a scaling method
that involves a single variable, " # x 4Dt
# x &
)c # x &
c(") = c%
( = c + erf %
(
$ 4Dt '
$ 4Dt '
! 2
3.205 L3 11/2/06
erf (x) is known as the error function and is
2 x $% 2
defined by
erf x "
e d%
( )
&
# 0
An example:
c( x < 0, 0!
) = 0;c( x > 0, 0) = 1;c("#,t ) = 0; c(#,t ) = 1; D = 10"16
t = 102, 103, 104, 105
!
Application to
problems with
fixed c at surface
3.205 L3 11/2/06
Movie showing time dependence of erf solution
3.205 L3 11/2/06
Superposition of solutions
When the diffusion equation is linear, sums of
solutions are also solutions. Here is an example
that uses superposition of error-function solutions:
Figure removed due to copyright restrictions.
See Figure 4.4 in Balluffi, Robert W., Samuel M. Allen, and W. Craig Carter.
Kinetics of Materials. Hoboken, NJ: J. Wiley & Sons, 2005. ISBN: 0471246891.
Two step functions, properly positioned, can be
summed to give a solution for finite layer placed
between two semi-infinite bodies.
3.205 L3 11/2/06
Superposed error functions, contd
The two step functions (moved left/right by x/2):
# x + "x 2 &,
c0 )
cleft = +1+ erf %
(.
$ 4Dt '2*
# x / "x 2 &,
c0 )
cright = / +1+ erf %
(.
$ 4Dt '2*
and their sum
3.205 L3 11/2/06
clayer
# x ) "x 2 &c0 * # x + "x 2 &
= ,erf %
( ) erf %
(/
$ 4 Dt '.
2 + $ 4Dt '
9
Superposed error functions, contd
An example:
c("#,t ) = 0; c(#,t ) = 0; c( x $ "%x / 2, 0) = 0;c("%x / 2 < x < %x / 2) = 1;
c( x & %x / 2, 0) = 0;%x = 2 '10"6 ; D = 10"16
t = 101, 103, 104, 105
Application to
problems with
zero-flux plane
at surface x = 0
3.205 L3 11/2/06
10
Movie showing time dependence of superimposed
erf solutions
3.205 L3 11/2/06
11
The thin-film solution
The thin-film solution can be obtained from the
previous example by looking at the case where x
is very small compared to the diffusion distance,
x, and the thin film is initially located at x = 0:
N
# x 2 ( 4 Dt )
c( x,t ) =
e
4 "Dt
where N is the number of source atoms per unit
!
area initially
placed at x = 0.
3.205 L3 11/2/06
12
Diffusion in finite geometries
Time-dependent diffusion in finite bodies can
soften be solved using the separation of variables
technique, which in cartesian coordinates leads to
trigonometric-series solutions.
A solution of the form
c( x, y, z,t ) = X ( x) "Y ( y) " Z ( z) " T (t )
is sought.
!
3.205 L3 11/2/06
13
Substitution into Ficks second law gives two
ordinary-differential equations for onedimensional diffusion:
dT
= "#DT
dt
d2X
= "#X
2
dx
where is a constant determined from the
boundary conditions.
!
3.205 L3 11/2/06
14
Example: degassing a thin plate in a vacuum
c(0 < x < L, 0) = c0 ; c(0,t ) = c( L,t ) = 0
The function X(x) turns out to be the Fourier series
representation of the initial conditionin this
!
case, it is a Fourier sine-series representation of a
constant, c0:
)
&
#
x
Xn ( x) = * an sin%n" (
$ L'
n=1
with
!
3.205 L3 11/2/06
2c0
an =
L
$ #'
* sin&%n" L )( d#
0
15
degassing a thin plate, contd
The function T(t) must have the form:
2 2
'
$
n
#
Tn (t ) = Tn exp&" 2 Dt )
% L
(
and thus the solution is given by KoM Eq. 5.47:
4c0
!
c( x,t ) =
"
2
* 1
#
x & # (2 j +1) " 2 &1,, 2 j +1 sin%$(2 j +1)" L (' exp%) L2 Dt (//
$
'.
j=0+
0
!
3.205 L3 11/2/06
16
degassing a thin plate, contd
Example:
c(0,t ) = 0; c( L,t ) = 0; c(0 < x < L, 0) = 1
L = 10"5 ; D = 10"16
3.205 L3 11/2/06
17
Other useful solution methods
Estimation of diffusion distance from x " 4Dt
Superposition of point-source solutions to get
solutions for arbitrary initial conditions c(x,0)
!
Method of Laplace transforms
Useful for constant-flux boundary conditions,
time-dependent boundary conditions
Numerical methods
Useful for complex geometries, D = D(c), timedependent boundary conditions, etc.
3.205 L3 11/2/06
18
