Solution in Semi Infinite Diffusion Couples (Error Function Analysis)
Solution in Semi Infinite Diffusion Couples (Error Function Analysis)
C B = 0 and C B = C B+
However, we shall show later that the solution we are going to derive, can be
used only in the case where the concentration and the chemical potential
difference of the end members is not much. That means diffusion coefficients do
not vary significantly with the composition.
Semi infinite means that, we anneal for a certain annealing time such that the
end of the initial materials are not affected by the diffusion of elements.
Although the meaning of the semi infinite indicates that a good part of the end
of the diffusion couple should not be affected but actually even with one
unaffected atomic layer in the end is sufficient to consider the system as semi
infinite diffusion couple!
This is important since otherwise we can’t apply the relation derived here to
determine the diffusion parameter or to calculate the concentration profile from
the known diffusion parameter.
For the sake of solving the Fick’s second law, we divide the alloy with the
composition CB+ into n numbers of very thin slices.
We consider the distribution of element B from one slice. Suppose from the
slice, i, which is at the distance of ξi from the initial contact plane, x=0. Then the
distribution of element B following the solution for thin film, we can write
C B+ ∆ξ i (x − ξ i )2
C (x ) =
i
B exp −
2 πDB t 4 DB t
where
( )
M B mole / m 2 = C B+ ∆ξ i
Similarly, if we consider the distribution of elements B from all the slices and
superimpose, we shall get the total distribution of element B at a particular
position as
C B+ n ( x − ξ i )2
C B (x ) = ∑ exp− 4 D t ∆ξ i
2 πD B t i =1 B
CB+
∞
x −ξ
CB ( x) = ∫
2 πDB t 0
exp −
4 D t
B
dξ
x −ξ
Further consider η =
2 DB t
dξ
That means dη = −
2 DB t
x
When ξ =0 η= and when ξ = ∞ η = −∞
2 DB t
By replacing
−∞ (
x 2 DB t )
CB+ C B+
CB (x ) = − ∫ exp(−η ) dη = ∫ exp(−η )dη
2 2
π x (2 DB t ) π −∞
Since this integration is not very straightforward and takes very long time to
determine one single value, it is expressed in terms of error function. Advantage of
using this error function is that one can determine the values from a table.
Error function of Z is defined as
z
2
erf ( z ) = ∫ exp(−η )dη
2
π 0
Note here that z is actually the upper limit of the integration. Lower limit of the
integration should be zero.
So the expression in the previous slide should be rewritten such that it can be
replaced by the error function
x (2 DB t )
C B+
0
CB (x ) = ( 2
) ( )
∫ exp − η dη + ∫ exp − η 2 dη
π −∞ 0
C + x
C B (x ) = B − erf − ∞ + erf
( )
2 2 D t
B
C B+ x
C B (x ) = 1 + erf
2
2 DB t
C B (x ) 1 x
C B (x ) 1 x
= 1 + erf = 1 − erf
+
2 −
CB
2 DB t CB 2
2 DB t
So from the previous relation, it is apparent that the sign of x will depend on
which side of the x = 0, we are interested to calculate.
Further, if the composition profile is just the opposite compared to the first
example, it can be shown that the relation becomes
C B (x ) 1 x
= 1 − erf
−
2
CB
2 DB t
Note: We always use “+” at superscript for the concentration to denote right hand
side of the couple and “-” for the left hand side of the couple.
Tabulated error function values
z
2
erf (z ) = ∫ exp(− η )
2
dη
π 0
2 z3 z5 z7
= z − + − .....
π 3x1! 5x 2! 7x3!
erf (− z ) = −erf (z )
In the previous case, we considered the diffusion between the diffusion couple of
two different blocks with composition CB=0 and CB+ or otherwise CB=CB- and 0.
That means in both the cases, it was a couple between one pure element and an
alloy of AB.
Error function solutions given previously can be rewritten interms of normalized
concentration profiles as
CB (x ) − 0 1 x CB (x ) − 0 1 x
= 1+ erf = 1 − erf
+ −
−
CB − 0 2 2 DB t C B 0 2 2 DB t
C B ( x ) − C B− 1 x
C B+ + C B− C B+ − C B− x
= 1 + erf
or C B (x ) = + erf
+ −
2 2 D t
CB − CB 2 DB t 2 2 B
Further, the relation in the case of diffusion couple, where CB- and CB+ and CB- >
CB+ as shown in Fig. b (in the previous slide), it can be expressed as
C B ( x ) − C B+ 1 x C B− + C B+ C B− − C B+ x
= 1 − erf or C B (x ) = − erf
C B− − C B+ 2 2 D t 2 2
B
2 DB t
In the case of carburizing, we can see that the relation will be the same,
irrespective of the side on which carburizing is done, since in the case of Fig. a,
x is positive but in Fig. b, x is negative
C ( x ) − Co x C ( x ) − Co −x x
= 1 − erf = 1 + erf = 1 − erf
C S − Co 2 Dt C S − Co 2 Dt 2 Dt
x
C (x ) = C S − (C S − C0 )erf
2 Dt
x
C ( x ) = CS 1 − erf For Co=0
2 Dt
Few important notes
So from the previous relations, it is apparent that the sign of x (whether positive
or negative) will depend on in which side of the x = 0, we are interested to calculate.
Note again that D will be more or less the constant at any position it is calculated.
It was also one of the assumptions for this derivation.
C B ( x ) − C B− 1 x
C B ( x = 0 ) − C B− 1 1 +
from the relation, + −
= 1 + erf
2 D t
it comes + −
= ⇒ C B ( x = 0 ) = (C B + C B− )
CB − CB 2 B
CB − CB 2 2
x
from the relation, C (x ) = CS − (CS − C0 )erf it comes C (x = 0) = CS
2 Dt
It indicates that in a system, where two blocks with different concentrations are
coupled, at x =0, which is basically the initial contact plane, the concentration will
be always average of the concentration of the initial materials.
Note that we need to locate anyway the initial contact plane after the diffusion
process, since x in the equation is actually measured from the location of the initial
contact plane.
This also indicates that this equation can only be used when end parts of the
couple is not affected by the diffusion process, since otherwise at x =0, it will have
another average value depending on how much of the end members are affected.
P P
Q Q
In the case of carburizing, the surface concentration should always be the same.
Note again that D will be more or less the constant at any position it is calculated. It
was also one of the assumptions for this derivation.
Before proceeding further for calculation using error function, one needs to check
if indeed you can locate the position of x = 0 by taking average of the end member
concentrations.
This position should be possible to locate even by equalizing the areas P and Q
(since the loss from one side of the couple should be equal to the gain in another
side).
If the composition profile is not symmetric, it can be found mathematically by
x +∞
∫C ∫ (C )
+
B dx = B − CB dx
−∞ x
When for a particular x the integrals have the same values, it should be fixed as 0
position.
Although it looks very trivial, finding this plane may not be that easy in most of the
cases.
Since initial contact plane x=0 is an unique plane (only one value is possible), it
should be possible to find from any of the concentration profiles of elements A or B,
as shown in the previous slide.
When it deviates (positively or negatively) from the ideality, one finds two different
values from two different composition profiles.
It feels as if the system lost its initial contact plane. It will be discussed further in
details at later stage, with examples.
So the limitation of the error function analysis is not only that diffusion coefficient
is more or less the constant but also the molar volume does not change
much with the composition or changes ideally.