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
 Diffusion
Concentration change occurs, therefore diffusion takes place in
almost all solid state transformations

1) Typical questions correspond to such solid state transformations

How fast is the atomic redistribution?
How long will be the heat treatment?

2) Lattice defects have a key role in materials characteristics and

also in diffusion. We will get answer to the following diffusion related
question:

How are the Å size crystal defects, random atomic motion and
macroscopic atom currents connected?
How do these phenomena work in solid state transformations?
Diffusion
Fick laws (Adolph Fick 1858): continuum description of diffusion
Diffusion in dilute, single phase alloy, e.g. B atoms in A matrix
(or A* atoms in A matrix or in AB matrix)

Fick I:
number of total number
in 1 dimension B atoms per of atoms per
volume volume
∂ cB −2 −1 −3
j B =− D B c B= N V νB [ j B ]=m s [c B ]=m
∂x 2
m
where D B is the diffusion constant for B atoms [ D B ]=
s

()
in 3 dimensions ∂ cB
∂x
∂c
j B =−D B grad (c B )=−D B B
∂y
∂ cB
∂z
Diffusion
Fick II:
in 1 dimension
∂ cB ∂ jB
=−
∂t ∂x
2

( )
∂ cB ∂ ∂ cB ∂ DB ∂ cB ∂ cB
= DB if =0 =D B
∂t ∂ x ∂x ∂x ∂t ∂x
2

i.e. when DB does not depend on cB

in 3 dimensions
∂ cB
∂t
=−div ( j B )=− (
∂ j x,B ∂ j y ,B ∂ jz ,B
∂x
+
∂y
+
∂z )
∂ cB ∂ DB ∂ cB
=div ( D B grad (c B )) if =0 =D B div (grad (c B ))
∂t ∂x ∂t
Diffusion

( )
2 2 2
∂ cB ∂ cB ∂ c B ∂ c B
=D B div (grad(c B ))= D B + +
∂t ∂x 2
∂y 2
∂ z2

∂ cB
=D B Δ c B
∂t

This diffusion equation is an approximation. It is valid if

+ diffusivity is concentration independent
+ distances are larger than an atom
+ time is longer than the time of an atomic jump
Examples for diffusion

1) Thin film solution A A

cB=
mB
2 √π DB t
exp −
( x2
4 DB t ) layer of B atoms

∞
m B = ∫ c B ( x ,t )dx c B ( x , 0) δ( x) function
−∞

sample (half plane) and its dimensions

2) Half plane 0≤ x<∞
cs c0 −∞< y <∞
−∞<t <∞

boundary conditions:
c B =c0 , if t=0 ; 0≤ x<∞
c B =c S , if x=0 ; 0<t <∞
Examples for diffusion
2) Half plane
solution:

where
2
∞
erf ( z)= √ π ∫ e− y dy
2
[ (
c B (t , x)=c0 +(c s −c0 ) 1−erf −

and
x
2 √ DB t
erf (0)=0
)]
lim erf ( z )=1
0 z →∞

c B (0 , t)=c S , independently from t erf (0)=0

c B ( x →∞ , t)=c 0 erf ( z)=1
x
The good parameter is not t, but
2 √ DB t
Therefore this solution is good for thin (d) samples if d ≫2 √ D B t

( )
c B −c0 x x
=1−erf − c B =const.=
c s−c 0 2 √ DB t 2 √ DBt
c B −c0 for n times larger penetration depth
=1/2 x=1.04 √ D B t
c s−c 0 n2 times longer time is needed
Examples for diffusion
3) Two half planes with different concentrations
This is a superposition of thin film solutions
Examples for diffusion
4) Homogenization
t=0 πx 2l is the wavelength
c B ( x)=c0 +c m cos
l of the inhomogeneity
If DB is independent from x and cB

(
c B ( x)=c0 +c m exp −
π2 D B t
l
2 ) cos
πx
l
2
l is the characteristic time
c B −c0 ∼exp(−t / τ) where τ= for homogenization
D B π2
Atomic mechanism of diffusion
Fick laws work also in liquids and gases, where atomic collisions
should be taken into account in the diffusion process
In solid materials, the jump of atoms to a neighboring position
leads to diffusion (vacancy motion enhanced atomic jumps)
atomic layers
#1 #2 #3 layer #1: n1 =c B 1 α B atoms per unit area
layer #2: n 2=c B 2 α B atoms per unit area
Γ : mean frequency of atomic jumps
α : distance of jumps (atomic layers)
α
z : number of neighboring atoms
1
δ t ≫ Γ during the period of several atomic jumps,
the number of jumps in layer 1 (per unit area): u1 =n1 Γ δ t
only 1 atom from z jumps toward layer 2
(z=6 in a simple cubic system)
Atomic mechanism of diffusion
1 1 1 1
ϕ 12= u 1= u 1 δ ϕ= ( 1 2 ) Γ δ t ( n1−n2 )
u −u =
6 z z z 1
1 1 1 j= Γ α ( c B 1 −c B 2 )
ϕ 21= u 2 = u 2 j= Γ ( n 1−n 2 ) z
6 z z
If the jump frequency is the same for the two layers and
the probability of the jumps is direction independent
1
j= Γ α ( c B (1)−c B (2) )
z
∂ cB 1 2 ∂ cB
c B (2)=c B (1)+α j= Γ α (Fick I)
∂x z ∂x
1 2 1 2
D B= Γ α and in simple cubic D B = Γ α
z 6
Diffusion is the natural consequence of atomic mobility and
the presence of concentration gradients
( DB is anisotropic if the system has no cubic symmetry )
Atomic mechanism of diffusion
Alternative picture: theory of 1D Brownian motion

Random motion of a B atom along x axis with m≫1 steps

each step has a length of α
2 2
The average of the displacement square is x m =α m

Compare it with the thin film solution

∫ B
( )
2
mB x
2 c ( x) x dx
cB= ⋅exp − x2= =2 D B t
2 √π DB t 4 DBt ∫ c B ( x)dx
α 2 m=2 D B t 1
D B = α2 Γ
m=Γ t 2

in 3D, only the 1/3 of these jumps give contribution to the x2

1
D B= α 2 Γ
6
Atomic mechanism of diffusion
Diffusion depends strongly on temperature
due to Γ , because α changes only 10-5 1/K

Γ=z ν exp − (
Δ Gm
kT )
α

ν : lattice vibration frequency

Thermally activated process

 Atomic mechanism of diffusion
Experiments: ( )
D= D 0 exp −
Q
kT
where Q=Δ G m =Δ H m−T Δ S m

( ) (
D= D 0 exp
Δ Sm
k
exp −
Δ Hm
kT )
Self diffusion measurement:
c0
concentration profiles at different T
tracer

The main micro process for diffusion: atom swaps with a vacancy

Γ=cV νV jump frequency of a vacancy

cV (T ) vacancy concentration
Atomic mechanism of diffusion
Atomic lattice vibration frequency ν 0 (Debey frequency: 1013 1/s),
but for a jump the energy of the lattice vibration is not enough.
Thermal activation is needed for the jump.

( ) ( ) ( )
ν V =ν 0 exp −
Δ G Vm
kT
=ν 0 exp
Δ S Vm
k
exp −
Δ H Vm
kT

( ) ( ) ( )
cV =ν 0 exp
Δ S Vf
k
exp −
Δ H Vf
kT
=c ∞ exp −
Δ H Vf
kT
1 2 '
(
D B = α ν 0 c∞ exp −
6 ) Δ H Vf +Δ H Vm
kT

D = D exp (− )
ΔH VD
B 0 Δ H =Δ H +Δ H VD Vf Vm
kT

Typically, temperature dependence is of a thermally activated process

vacancy formation and motion

Atomic mechanism of diffusion
Linear functions in log plot as a function
of Tm/T

(
D B = D0 exp −
Δ H VD
kT )
Δ H VD≈const T m Δ H VD≈16.5 L m
where L m is the latent heat per atom

16.5 atoms become liquid (not really true),

but a good rule of thumb
Atomic mechanism of diffusion
Examples:

Cu self diffusion vacancy mechanism

Δ H VD≈2 ev / atom T m =1350 K

( )
2
Δ S Vf −5 m
Δ S Vf ≈2 k in c∞ =exp D 0∼10
k s

( )
2
−6 m 200.3 kJ /mol
D Cu∼31⋅10 exp −
s RT
C diffusion in α-Fe interstitial mechanism
Δ H VD≈1 ev /atom Δ S m≈2−3 k

( )
2
α ν0 Δ Sm −6 m
2
D 0= exp D 0∼10
a k s

Multicomponent systems: concentration dependent D

Concentration dependent diffusivity ~
D= D(c)

Boltzmann-Matano analysis
x
=
∂t ∂ x (
∂c ∂ ~ ∂c
D
∂x ) let us use a new variable: η=
√t
x dc 1 d ~ dc
− 3/2 =
2t d η t d η
D
dη( )
−
η dc
=
2 dη dη
d ~ dc
D
dη ( ~ dc
→d D
dη) (η
=− d c
2 )
Concentration dependent diffusivity ~
D= D(c)

Boltzmann-Matano analysis

|
c
~ 1 dc x
D (c)=− ⋅∫ ηd c ' η=
2 dηc 0 √t
|∫
~ 1 dx
c but result depends on
D (c)=− ⋅ xd c' at const. t the position of x=0 !
2t d c c 0

Matano plane:
~ is the interdiffusion conservation law
D for material
coefficient of the two
components,
~ ~
D A and D B
c0

∫ x d c ' =0
0
Concentration dependent diffusivity ~
D= D(c)

Boltzmann-Matano analysis

Boundary between the two materials can be marked with

oxides or W wiskers (markers), however the markers can also
move regarding the ends of the sample
(toward the end with lower melting point component in excess)

A B T mA <T mB right side: material swelling

left side: vacancy formation,
xk
faster diffusion removes material

xk

xk
Concentration dependent diffusivity ~
D= D(c)

Kirkendall effect: the same problem in two coordinate systems

0 L x

x system
~ ∂ cB
j B (x=0 ,t )=− D B
∂x |x=0

x system ~ ~ ∂ cB
j B ( x = L ,t )=− D B ~
∂ x ~x = L|+ v c B|~x = L

The diffusion coefficient is determined by the sum of the two

constituents (A and B)
Concentration dependent diffusivity ~
D= D(c)

the same for A component

ci = N v ν i c A +c B = N v Nv

∂ N v ∂ cA ∂ cB
∂t
=
∂t
+
∂t (
∂ jA ∂ jB ∂
) [
~ ∂ c A ~ ∂ cB
=− ~ + ~ = ~ D A ~ + D B ~ −v (c B +c A )
∂x ∂x ∂x ∂x ∂x ]
∂Nv
Total density of atoms does not change =0
∂t
∂ c
at ~
i
x =0 =0 and v=0→ [ ... ] =0
∂x

v=
Nv [
1 ~ ∂ cA ~ ∂ cB
D A ~ + DB ~
∂x ∂x ]
~ ~ ∂ νA
v=( D A− D B ) ~ Darken I.
∂x
Concentration dependent diffusivity ~
D= D(c)

~ ~ ∂ cB
j B ( x = L ,t )=− D B ~ |
∂ x ~x = L
+ v c B|~x = L II.
∂ cB
=− ∂~ j B (~
x) III.
∂t ∂x

From Eqs I, II and III

∂ cB ∂ 1 ~
=
∂t ∂ x Nv
( [ ~ ∂ cB
D B c A+ D A c B ) ~
∂x ]
Compare it with the eq. of Matano analysis ∂t ∂ x(
∂c ∂ ~ ∂c
= D
∂x )
~ ~ ~
D=ν A D B + ν B D A Darken II.

DA and DB can be determined from D and v using Darken I. and II.