Materials Physics II

Contents:
1. Diffusion and concentration dependent diffusion
2. Role of crystal defects in diffusion
3. Solidification of pure materials: nucleation and crystal growth
4. Solidification of solid solutions and eutectic phase mixtures
5. Precipitation
6. Metastable phases and spinodal decomposition
7. Solid solution hardening
8. Plastic deformation of multicomponent systems
9. Diffusion free phase transformations
10. Ceramics, covalent (ionic) crystals and glasses
11. Composites
Atomic mechanisms of diffusion
Vacancy-atom interaction takes place in self diffusion

Diffusion process in a face centered cubic (close packed)

system

(
D B = D0 exp −
Δ H VD
kT ) Δ H VD =Δ H Vf +Δ H Vm
Atomic mechanisms of diffusion
Correlation effects example:
: vacancy
: A*

Diffusion of A* tracer atom in crystal A in the presence of

a V vacancy

A* jumps position 6 to 7
A* swaps with V
V jumps position 7 to 6

next jump:
V swaps with the same possibility with all the new neighbors,
but A* jumps back to position 6 with much higher possibility

This is a preferred back jump, i.e. a correlation between jumps

Atomic mechanisms of diffusion
Preferred back jump, i.e. correlation will change the squared
average of the displacement of atom A*

Therefore the assumption that D A ∗= D A will fail, where

2
α
D A= c V νV is the calculated self diffusion coefficient and
6
D A∗ is the experimentally measured isotope diffusion coefficient

To determine the ratio of the isotope diffusion and the calculated

self diffusion a correlation factor can be defined:
2 ∗
rm( A ) D A∗ correlated diffusion
f =lim 2 =
m → ∞ r (V )
m
DA not correlated diffusion
Atomic mechanisms of diffusion
For the previous example 2
f =1−
z

A more accurate calculation for cubic metals, where θ is the angle

between the successive jumps of atom A*

1+cos θ
f=
1−cos θ

All the possible jumps should be taken into account in the average

Typical values for f in different cubic structures

0.78 fcc, hcp structures
0.73 bcc structure
0.5 diamond structure
 ~
Concentration dependent diffusivity D= D(c)

Role of defects in diffusion

- concentration dependent diffusivity total diffusivity of
- movement of markers: Kirkendall effect the components
- asymmetric vacancy distribution

T mA <T mB B: material swelling

A: vacancy formation
faster diffusion removes material
~ ~ ∂ νA
v=( D A− D B ) ~ Darken I.
∂x
~ ~ ~
D=ν A D B + ν B D A Darken II.

different atom-vacancy interaction

for A and for B components
Concentration dependent diffusivity
vacancy flux is
not source free
j V = j B− j B

D B> DB
Kirkendall voids
Atomic mechanisms of diffusion
Diffusion of solute atoms, atom B in crystal A

Correlation becomes a decisive process, because V interacts

differently with B than with A atoms

For example, if B is a large atom, V has a strong interaction with B

→ f ≪1 Γ B ≫Γ A
1 2 ΓB 1 2
Instead of D B= α Γ B D B= D A f Γ = α f Γ B
6 A 6
2
r m ( B) 2 Di m
because f= 2 and r (i)= Γ
m
r ( A)
m
i

ΓB ( z−1) Γ A
For the previous example cos θ=− f≈
Γ B +( z−1) Γ A 2 Γ B +( z−1) Γ A
Atomic mechanisms of diffusion
( z−1) Γ A therefore f can be T dependent
f≈
2 Γ B +( z−1) Γ A for solute diffusion

In summary: 0 ≤ f ≤ 1

self diffusion: f is independent from T

diffusion in solid solution: f can be T dependent
Atomic mechanisms of diffusion
Other diffusion mechanisms

Interstitial diffusion (e.g. C in Fe),

no vacancy formation is required

Ring mechanism in
substitutional alloys

Other (more complicated) mechanism can also contribute to the

total diffusivity. Typically, these mechanism have smaller
contribution because of their higher activation energies.
Thermodynamic factor
Diffusion can take place in the absence
of a concentration gradients, because ∂ μi
chemical potential determines =0
∂x
the equilibrium condition

Thus, gradients in the chemical potential cause atomic currents

∂μA ∂μB ∂ μV
j A =−M AA − M AB − M AV
∂x ∂x ∂x
∂μA ∂μB ∂ μV
j B =−M BA − M BB − M BV
∂x ∂x ∂x
∂μA ∂μB ∂μ V
j V =−M VA − M VB − M VV
∂x ∂x ∂x

For binary A,B alloy with V vacancies

Thermodynamic factor
Onsager relation: Mij=Mji
If there is no vacancy formation/sink: j A + j B + j V =0

j A =−M AA ∂ (μ A−μV )− M AB ∂ (μ B −μ V )
∂x ∂x
j B =−M AB ∂ (μ A −μV )−M BB ∂ (μ B −μV )
∂x ∂x

Darken equation can be obtained, if the vacancy concentration is in

equilibrium and the non-diagonal elements are small
∂μ V
=0
∂x
Thermodynamic factor

~ ∂ cB ∂μB
j B =− D B =−M BB
∂x ∂x

reg 0 0
with activity
μ =μ +kT ln a B =μ + kT ln ( γ B ν B )
B B B or activity
coefficient
d ln a B ~ ∂ cB
j B =−M BB kT =− D B
dx ∂x
Thermodynamic factor
cB ~
Ideal solution a B =ν B = D B= D B
NV

− j B= D B
∂ cB
∂x
= M BB kT
d ln ν B
dx
= M BB kT [
d ln c B d ln N V
dx
−
dx ]
1
D B =kT M BB⋅
cB

Regular solution a B =γ B ν B
d ln γ B ν B d ln γ B ν B d ln ν B
− j B =kT M BB =kT M BB
dx d ln ν B dx

If
d ln N V
dx
=0
~
D B =kT M BB⋅
1
cB
1+
[
d ln γ B
d ln ν B ]
Thermodynamic factor
The thermodynamic factor is in
the brackets for the regular solutions
~
[
D B = D B 1+
d ln γ B
d ln ν B ]
γB is constant for small B concentrations (Henry law)
and equal to unity for large B concentration (Raoult law)
therefore for the extremes D ~ (concentration independent)
B= D B

But the two diffusion constant can be different in the two extremes of
the concentration region (for self diffusion and diffusion of B in A)

The thermodynamic factor of only one component appears in the

Darken equations because of the Gibbs-Duhem relation:
d ln γ A d ln γ B
νA =ν B N A d μ A + N B d μ B =0
d ln ν A d ln ν B
Thermodynamic factor
1) The diffusion constant can be even negative
~
(
D B = D B 1+
d ln γ B
d ln ν B )
d ln γ B ν A ν B ∂ 2 F
1+ =
d ln ν B kT ∂ ν 2B

Up-hill diffusion between spinodals ∂2 F

2
<0
(diffusion against the concentration gradient) ∂ νB

2) Two phase regions

(e.g. eutectic)
Thermodynamic factor
3) Multicomponent systems

Concentration gradient can increase in one component temporarily,

while the other components approach the equilibrium
concentration.
Diffusion along interfaces and dislocations
Diffusion in interfaces and along dislocations

∗
DL lattice (or volume or bulk) diffusion
∗
DG grain boundary diffusion

grain boundary ∗ ∗
D L ≪ DG

|
∗
∂c ∂ jy 2 ∗ ∂ 2
c D L ∂c
=− − j = DG +2 ⋅
∂t ∂y δ x ∂y
2 δ ∂x x=± δ
2
 Diffusion in interfaces and along dislocations
Quasi static distribution forms along the GB rapidly, then
thin film diffusion into the bulk material
c( x=0 , y)=cG

(
c( x , y , t )=c 0 exp − ∗ 1/4
y
∗ ∗ 1/ 2
(π D L t ) (δ D G / 2 D L ) )
[ ( √ )]
⋅ 1−erf
x
2 DL t
∗

−Q / kT
D= D0 e Q L >Q G >Q S
Q L : Q G : Q S ≈4 : 3 : 2
4: 3:1
D 0 , L > D 0 , G > D0 , S
Orientation dependent diffusion along grain
boundaries

Difference in the activation energy (bulk vs. grain boundary): QL>QG

ΔHVFG<ΔHVFL vacancy formation
ΔHVMG<ΔHVML vacancy motion (more important)
 Sintering
Powder aggregate is densified to a solid body by diffusion process

Important in powder metallurgy and in the production of solid

ceramics parts

Pores are eliminated by forming necks and surface vacancy diffusion

Sintering is driven by the surface energy excess in the porous

material

2
∂Gp ∂(4 π r p ) 2 γ S Ω
μ p= ≃γ S =
∂n ∂ (V p / Ω) rp
 Sintering
Copper wire sintering