Substitutional diffusion
Previously we discussed about the interstitial atomic mechanism of diffusion
Now we shall discuss atomic mechanism of substitutional diffusion, which is
very different in nature.
As explained earlier tracer diffusion method that is monitoring the diffusion of
atoms that are detectable is the way to measure self diffusion coefficient
First we need to understand the tracer diffusion method and then we shall
develop the relation between tracer and self diffusion coefficient.
As mentioned earlier, we actually determine the tracer diffusion coefficients
to understand the self diffusion.
Unlike previous case, there is difference between these two diffusion
coefficients in substitutional diffusion.
So first let us discuss about the tracer diffusion mechanism, following which
self diffusion will be discussed.
Tracer diffusion coefficient measurements
Suppose we want to measure tracer diffusion of A in
the same material A.
Radio tracer atoms A* are deposited on the surface
of an element A and then allowed to diffuse in the
substrate.
1 2
ln I R = ln − x
πD * t 4 D *A t
A
As explained previously, from the values of
slope, we can determine the tracer diffusion
coefficient
Substitutional diffusion of tracer atoms: concept of correlation factor
We have discussed before that even almost near the melting point, there could be
maximum one vacancy available out of 1000 atoms.
That means one vacancy is available in a cube of 10 atoms each side. So, we can
neglect the interaction between vacancies (That means we are neglecting the
contribution from divacancies).
Now the vacancy can exchange position with any atoms surrounding it. We
consider that the tracer atom, with its own probability has changed the position with
vacancy.
Further question is that what will be the next position for the tracer atom.
The next most probable jump is jumping back to the previous position, since
vacancy is available at that site and vacancies are not available at any other sites!
Then two successive jumps will cancel the diffusion of the tracer atoms.
However, our experiments reflect that the tracer atoms jump to different distance
after a particular time. This further indicates that there should be a possibility for
the tracer atom to go forward.
This is possible because, the jump of the tracer atoms is dependent on the
availability of vacancies, however, a vacancy can exchange position with another
atoms
So the vacancy has an equal probability to exchange position with other atoms,
for example 1 or 5
So the next possible jump of the tracer atom is to 1 or 5, on the condition that
vacancy has already exchanged the position with atoms at 1 or 5.
The next possible jump for the tracer atom is to 6 or 8, on the condition that the
vacancy has arrived there.
The chance for tracer atom to reach to the point 7 is the least since first the
vacancy after few jumps has to reach to that position.
So on broader sense, we can say that the jump of tracer atom is correlated
(unlike jump of interstitial atoms). That means, the jump of a tracer atom is
dependent on the other jumps. That further means that the tracer atom does not
have true random jump. So the average jump length of the tracer atom should be
written as
2 n −1 n − j
L = nl 1 + ∑∑ cosθ i ,i + j Second part cannot be taken as zero
2
t
2
n j =1 i =1
On the other hand vacancies can go through a random jump process, since it
can exchange position with any atom. So the average jump distance of the
vacancies can be written as 2 2
Lv = nl
If we take the ratio of the square of the average jump distance, the value
indicates the fraction of the jump of tracer atoms compared to the jump of
vacancies.
The ratio is called the correlation factor and expressed as
Lt2 2 n−1 n − j
f = lim 2 = lim1 + ∑∑ cosθ i ,i + j
n j =1 i =1
n →∞ L
n →∞
v
Since diffusion coefficient is proportional to the square of the jump distance,
we can write
D* = fDV
Further since a single vacancy even involves jump of many atoms and again
because we calculate the diffusion rate from the jump of many atoms together,
we can say that the self diffusion coefficient of the elements is the same as the
diffusion coefficient of the vacancies. So we can write
D * = fD S
The correlation factor in different crystal structure is calculated as
Crystal structure Z f
Simple cubic 6 0.65549
Body centered cubic 8 0.72149
Face centered cubic 12 0.78145
Note: Correlation factor in interstitial diffusion is 1, since we consider
that the concentration of interstitial atoms is very low to make random
jump possible because of availability of free sites to jump.
Now, let us do the similar analysis what we did for the interstitial diffusion to find
the relation between the self diffusion coefficient with the jump frequency.
However, the starting point will be again the tracer diffusion, since we measure the
diffusion of tracer atoms.
We consider the (111) plane of a FCC crystal. Further we assume that the tracer
atom and the atoms 4-9 are sitting on the plane 1
Similarly, atoms in voids 1,2 and 3 are sitting on plane 0. That means we can
consider that atoms in voids 10,11 and 12 are sitting on plane 2
Let us consider the diffusion between plane 1 and 2 only, in the x direction.
It must be clear that if we consider the jump of the tracer atom from the plane 1
to the plane 2, then it can jump to any of the positions occupied by the three atoms
10, 11 and 12, on the condition that vacancy is available there.
So the flux of tracer atoms from the plane 1 to the plane 2 can be written as
r* 3 1
J A = n1 f p v Γ1 = n1 f p v Γ1
12 4
Γ1: Jump frequency per second from plane 1
Γ2: Jump frequency from plane 2
n1 and n2 are the number of tracer atoms/area in plane 1 and 2
f is the correlation factor,
pv is the probability that tracer atom will find a vacancy to exchange the
position, which should be equal to the fraction of vacancy, Xv present
Flux of tracer atoms from plane 2 to 1
s* 3 1
J A = n 2 f p v Γ2 = n 2 f p v Γ2
12 4
Following similar analysis, as we did for interstitial diffusion, we can write the
resultant flux as
1
J *A = f X v Γ(n2 − n1 )
4
1 dc
J *A = − f p v Γ∆x 2
4 dx
1
D A* = f p v Γ∆x 2
4
1
D A* = f X v Γ∆x 2
4
Further, if we consider the jump to any of the nearest neighbor and
considering many atoms together, we can write
Z
1
∑ Γi ∆xi
* 2
D = fX v
A
Z Z =1
So the self diffusion coefficient can be written as
Z
1
∑ Γ ∆x
S 2
D = Xv
A i i
Z Z =1
