d!

s
dCL
CL

Cs=kCo
kCo

Solute Redistribution
Topic 6
M.S Darwish
MECH 636: Solidification Modelling
 Introduction
• Local Equilibrium
– Solidus and liquidus curves
– Equilibrium distribution coefficient, k0
– Tie–line “jump”
• Macrosegregation at Planar Interfaces
– Gulliver–Scheil equations
» Assumptions
» Mass balance
» Limitations
– Transport–limited segregation
» Initial Transient
» Steady–state solute distribution
» Terminal Transient
• Effects of Stirring & Convection
– Burton–Prim–Slichter Theory, kBPS
Lever Rule
 Introduction
T
In practice equilibrium is
To CS maintained at the solid liquid
CL k=CS/CL
TL
interface where composition
T agrees with the phase diagram
Temperature

TS

solid liquid

Cs*
k=
Cl*
CL
C

Composition
A B
?
kCo Co Cmax Co/k Ceut
?
Composition
CS

Distance
 Equilibrium at S/L
T

liquidus
Molecula
r scale
solidus

C
Tie line “jump”

CB
S L Continuu
m scale
 Equilibrium Solidification
T
To CS
CL k=CS/CL
TL
T
Complete Diffusion in
TS Solid and Liquid
Temperature

Respectively

C χ sCs + χ l Cl = Co Lever Rule

A kCo Co Csm Co/k Ceut B
Composition

Start of Solidification At Temperature T End of Solidification

Composition

Composition
Composition

CL

CL* = CL! Co C*S = Co

Co Co
CS
kCo kCo

0 L 0 L 0
Distance Distance Distance
Schiel Rule
 Gulliver–Scheil Differential Mass Balance
x

solid liquid
Although local interfacial equilibrium
generally holds during solidification, the dx
system cannot remain in “global” 0 L
equilibrium because diffusion rates in ( s l ) d x = (L − x )dCl
C *
− C *

the solid are too slow.

⎛ C * ⎞
Cl ⎜ * −1⎟ d x = ( L − x ) dCl
* s
G. Gulliver (1921) and E. Scheil (1942) ⎝ Cl ⎠
both assumed:
Perfect mixing in liquid (C l = Cl* ) Cl* ( k −1)dx = ( L − x ) dCl
No diffusion in solid
dx ′ d Cl′
x l C

(k −1) ∫ = ∫
Local equilibrium at S/L interface
0 ( L − x ′) C 0 Cl′
x
note that χ s =
L
 Gulliver–Scheil Segregation Equation

• The Gulliver–Scheil mass balance

eventually leads to non–physical
predictions as χs→1.

• The phase diagram provides limits to

how high the concentrations, Cl or Cs
can rise due to segregation. Second
( k0 −1) Cl
phase formation (generally from eutectic (1− χ s ) =
reaction) then occurs. C0

x
Where χ s =
L
Gulliver–Scheil is independent
of the shape of the S/L interface!
 No Solid Diffusion
T (Non-Equilibrium Lever Rule)
To CS solidification of dχ s
(C − Cs* ) dχ s = (1− χ s ) dCl
CL k=CS/CL
TL l
T d!s
dCL
TS rejected solute from solute increase in
Temperature

CL
solid liquid
χs Cs=kCo
dχ
C
l
dC
∫ 1− χ = ∫ C (1−l k ) kCo

0 Co l

C
Cl = Co (1− χ s ) Cs* = kCo (1− χ s )
k−1 k−1
A kCo Co Csm Co/k Ceut B Scheil relation
Composition
End of Solidification
Start of Solidification At Temperature T

CE
Composition
Composition

CL

Composition
* CL! Co
CL=
Co Co
Csm
C*S
kCo kCo Co

0 L 0 Distance L kCo
Distance
0 L
Distance
Gulliver–Scheil Segregation Equation
( k0 −1) Cl
(1− χ s ) =
C0
5

4 k0<1 k0 =0.01
0.1
0.2
3 0.3
0.4
Cl /Co

0.5
0.6
2
0.7
0.8

0
0 0.2 0.4 0.6 0.8 1

Fraction Solid, fs
Gulliver–Scheil Segregation Equation
( k0 −1) Cs
k0 (1− χ s ) =
C0

1.5 k0<1
Cs /C0

0.8
0.7
0.6
0.5
0.4
0.3 0.2
0.5 0.1
k0=0.03

0
0 0.2 0.4 0.6 0.8 1
Fraction Solid, fs
Gulliver–Scheil Segregation Equation

( k0 −1) Cl
(1− χ s ) =
C0

1 1.05

0.8
k0>0 1.2

0.6 1.5
Cl /C0

0.4 3

0.2 k0=10

0
0 0.2 0.4 0.6 0.8 1
Fraction Solid, fs
Gulliver–Scheil Segregation Equation

( k0 −1) Cs
k0 (1− χ s ) =
C0

k0=10

4 k0>1
5

3
Cs /C0

2
2

1.5

1 1.05
1.2

0
0 0.2 0.4 0.6 0.8 1
Fraction Solid, fs
 Al-Cu Phase Diagram

Maximum solubility
of Cu in Al is 2.5 at.% or 4.5 wt%.
 Application of Gulliver–Scheil
Segregation
3

2.5 Dendritic Microsegregation

Al–2.5
(Al-4.5at%wt%
Cu Cu)
Conc.entration, Cu (at.%)

1.5 Segregation limit set

by solubility limit.
1

0.5

0 2 4 6 8 10
Distance, X (microns)
 Plane Front
Solidification
 Moving S/L
T
Cs* (T ) Cl* (T )

liquidus

solidus

solid liquid Solute boundary layer

develops ahead of the
x advancing S/L interface.
Cl* (T )
CB
Cl(x) C∞
v
Cs(x) Cl* (T )
Steady State Solute
Boundary Layer

Solid Cl* Liquid

Cl ( x )
C0
!
Cs*
Cs ( x ) v
!
!
!
!

JB
Solid Liquid
Computing the Steady-State
Solute Profile

• Consider Geometry

• Apply Fick’s second law

v • Find the general SS solution

• Apply boundary conditions

• Find solute gradient at the

interface
 Apply Fick’s Law and SS Condition

dC ∂2C
=D =0
dt ∂z

dC
= ∇ ⋅ D∇C ≈ D∇ 2C
dt
v
dC ∂2C
In 1-D =D
dt ∂z
 Apply a Flux balance at Interface

J1 J2
Cl
dC
dz

v
A
J1 = = (Cl − Cs )
* dz
= v (Cl − Cs* )
Cs dt dt
dc
dz J2 = − D
dz z=0 D dC
J1=J2 v =−
( l s ) dz z= 0
C − C *
 Example 1
c = c(Z , t) = c(z(t))
∂c ∂c
dc = dZ + dt
∂z ∂t

Z dc ∂c dZ ∂c
= +
dt ∂z dt ∂t
∂ 2c ∂c ∂c
D 2 = − v+
z = Z − vt dz ∂z ∂t
dz 1 ∂c ∂ 2c v ∂c
Cγ = −v = 2+
dt D ∂t dz D ∂z
∂ 2c v ∂c
+ =0
dz D ∂z
2

⎛ v ⎞
c(z) = a0 + a1 exp⎜ − z⎟
Cα
z ⎝ D ⎠

Z
 ⎛ v ⎞
c(z) = a0 + a1 exp⎜ − z⎟
⎝ D ⎠
Apply Boundary conditions to evaluate the constants:

c(z = ∞ ) = C 0 c(z = 0) = C γ
Z a0 = C 0 a1 = C γ − C 0

⎛ v ⎞
c(z) = C 0 + (C γ − C 0 )exp⎜ − z⎟
⎝ D ⎠
z = Z − vt Flux Balance at SS:
⎛ ∂c ⎞
v(C γ − Cα )= − D ⎜ ⎟
dz ⎝ ∂z ⎠ z=0
Cγ = −v ∂c v
dt Evaluate dc/dz: = − (C γ − C 0 )
∂z z=0 D

⎡ v ⎤
v(C γ − Cα )= − D ⎢− (C γ − C 0 )⎥
⎣ D ⎦
Cα (C − C0 )
=1 = Ω
γ
z (Supersaturation)
(Cγ − Cα )
Z Must equal 1 for SS!
 Z

Destabilization
requires a
Cγ
certain growth
distance.
C0

Solute buildup
increases with time
Cα (or growth distance).
 Sequential
Crystallization Concepts
• First step of crystallization:

• r th step of crystallization:

Where
 Limited Liquid Diffusion
T
To CS At the interface far from the interface
CL k=CS/CL
Cl = Co
TL Cl = Co /k
T
⎛ dC ⎞∗
−D⎜ ⎟ = v (Cl∗ − Cs∗ )
TS
Temperature

⎝ dx ⎠
dC d 2C dC dx dC
=D 2 =v
dt dx dx dt dx
⎛ dC ⎞∗ v ∗
⎜ ⎟ = − Cl (1− k ) dC d 2C ⎛ 1− k − x ⎞
⎝ dx ⎠ D v =D 2 Cl = Co ⎜1+ e D /v ⎟
dt dx
C ⎝ k ⎠
A kCo Co Csm Co/k Ceut B Characteristic Distance
Composition After Solidification
Start of Solidification

Final
CEut
Transient
Co/k
Initial
Composition

Transient
Composition

Steady State

Csm
Co
D /v Co

0 L
Distance
0 Distance L
 After Solidification

T1 b
a c d
Final
T2 f T3
CEut
Transient e
Initial
Transient A B
Composition

Steady State

Csm

Co
d
b
c
a L
0 Distance L f
b
a L
⎛ x ⎞ ⎛ 1− k − x ⎞
−k
Cs∗ = Co ⎜1+ (1− k )e D / v ⎟ Cl = Co ⎜1+ e D /v ⎟ f
⎝ ⎠ ⎝ k ⎠ b
a L
 Solute Boundary Layer

Steady-State Dimensionless Solute Field in Melt Note the characteristic

diffusion length, ζ,
0.8
[Cl(x)-C0]/[C0 /k0-C0]

0.6

0.4
1/e

0.2
1/e2
1/e3
0
0 0.5 1 1.5 2 2.5 3
ζ=-(vX)/Dl
 Diffusion Limited Steady State
Segregation
• Diffusion is negligible beyond 3 e-folding lengths from the S/L
interface.
• The linear boundary layer approximation is equal to 1 e-folding,
1
ζ=1.
Steady-State Dimensionless Solute Field in Melt

0.8
[Cl(x)-C0]/[C0/k0-C0]

0.6

0.4
1/e

0.2
1/e2
1/e3
0
0 0.5 1 1.5 2 2.5 3
ζ=-(vX)/Dl
 Terminal Solidification
Transient
Terminal freezing was studied by V.G. Smith, W.A. Tiller, and J.W. Rutter, Canadian
Journal of Physics, 33, p. 723, (1955). They obtained the following infinite series
solution. Here Δx is the distance to the end of the ingot:
 25

• As a practical note, terminal

20 Terminal Transient transients are seldom of
interest in real crystal growth
Scaled Solid Concentration, Cs /C0

processes. The growth process

15 is usually halted before this
transient begins.
k0=0.05
10 • Note, little if any interaction
occurs when the S/L interface
0.1 is more than a distance Dl/v
5 0.2 from the end of the crucible.
0.5
1
3
0
1 0.8 0.6 0.4 0.2 0

Dimensionless Distance from End, Δx/(Dl /v)

Example 2: SS Spherical Growth
dc
r Fick’s 2 nd 2
law: dt = ∇ • D∇c ≈ D∇ c

In spherical coordinates:
1 ⎡ ∂ ⎛ 2 ∂c ⎞ 1 ∂ ⎛ ∂c ⎞ 1 ⎛ ∂ 2c ⎞⎤
R
2
∇ c= 2 ⎢ ⎜r ⎟+ ⎜⎜ sinφ ⎟⎟ + 2 ⎜⎜ 2 ⎟⎟⎥
r ⎣ ∂r ⎝ ∂r ⎠ sinφ ∂φ ⎝ ∂φ ⎠ sin φ ⎝ ∂θ ⎠⎦

∂c ∂c ∂ 2c
For spherical symmetry: = = 2 =0
∂φ ∂θ ∂θ
dc 1 ⎡ ∂ ⎛ ∂c ⎞⎤
= ∇ 2c = 2 ⎢ ⎜ r 2 ⎟⎥
dt r ⎣ ∂r ⎝ ∂r ⎠⎦

∂ ⎛ 2 ∂c ⎞ 2 ∂c
For steady-state: ⎜ r ⎟ = 0 thus r = −a
∂r ⎝ ∂r ⎠ ∂r
a
and c(r ) = +b
r
where a and b are constants.
 r General solution:

R where a and b are constants.

Boundary conditions:

a
C ( r >> R) = C0 = +b
r

a = R(Cl* − C0 )
a
C ( r = R) = Cl* = +b
R

Particular solution:

C ( r ) = C 0 + (C l − C 0 )
R *
r
 C ( r ) = C 0 + (C l − C 0 )
r R *
r
Controlling solute gradient:
R
dC
= − 2 (Cl* − C0 )
R dC
=−
( Cl* − C0 )
dr r dr r= R R

D dC
Flux balance: J1=J2 v =−
( l s ) dz r= R
C − C *

Rv (Cl − C0 ) ⎛ 1 ⎞ (C l − C 0 )
Substituting yields: = or v = D⎜ ⎟
D (Cl − Cs* ) ⎝ R ⎠ (Cl − Cs* )

NOTE: diffusivity
shape factor
supersaturation
Example 3: SS Needle growth

Needle with hemispherical cap

A Ah dC ( Cl* − C0 )
=−
dr r= R R

Flux balance
r
A(Cl* − C0 )
dz dC
= −Ah D
dt dr r= R
⎛ Ah ⎞ (Cl − C0 )
*

v = D⎜ ⎟
⎝ AR ⎠ (Cl* − Cs* )
R
v ⎛ 2 ⎞ (C l − C 0 )
*

v = D⎜ ⎟ *
⎝ R ⎠ (Cl − Cs* )
Ah 2πR 2
= =2
A πR 2
 v

Left for homework…

 The Shape Factor

Plane Sphere Plate Needle

Effect of Convection
 Effect of Convection and Stirring

C^l

solid melt
C0
Cs(x)

C^s diffusion
no diffusion region convective mixing region
δ
0 x

During crystal growth and plane–front solidification stirring occurs, which

affects the segregation process. This was analyzed by Burton, Prim and
Slichter (BPS Theory). They suggested three important zones:
Solid zone (no diffusion)
A “static” layer, δ, (diffusion transport)
Liquid zone (complete mixing)
 BPS Segregation Theory
BPS approximated the fluid dynamic momentum boundary layer as a “static
zone,” and bulk liquid as fully mixed.
They derived segregation curves (“inner” & “outer”) consistent with these
approximations in the fluid mechanics.
⎛ v ⎞
−⎜ x⎟
Cl ( x ) = Cs* + (Cl* − Cs* )e ⎝ Dl ⎠
, (x ≤ δ )
δ

Cl ( x ) = C0 , (x > δ )
Cl*

Cs* Cs*
kBPS ≡
C0
Solid Liquid
 BPS results (ko<1)
Rapid Slower stirring
1
stirring
0.8 0.5
0.3
0.15

0.6 0.05
0.02
KBPS

0.01
k =0.005
0
0.4

0.2

k0<1
0
0 2 4 6 8 10

δ/(Dl/v)
Rapid stirring Slower stirring
100

80 k0=0.01

0.0133

60
o
/k
BPS

0.02
k

40

0.03

20 0.05

0.1
0.3
0
0 2 4 6 8 10

δ/(Dl/v)
 BPS results (ko>1)
Rapid stirring Slower stirring
1

1.1

1.2
0.8

1.5

0.6
kBPS/k0

0.4
3

0.2 k =5
0

k0>1
0
0 1 2 3 4 5

δ/(Dl/v)
 BPS Limits
C0
k0
i. no mixing: δ→∞ ⎧⎪ Ĉs → C0
No natural solid melt δ →∞ ⎨
convection ⎪⎩ kBPS → 1
Cs C0

Ĉs

ii. perfect mixing: δ→0 Ĉl C0

Very rapid stirring
solid ←δ ~0 melt ⎧⎪ Ĉs → k0C0
⎨
Cs ⎪⎩ kBPS → k0
Ĉs
 BPS Summary

As summarized below, stirring provides kinetic control on the

amount of macrosegregation segregation during steady–state
plane front growth.

Faster stirring
δ kBPS Cs
0 k0 k0 C0
0<δ<∞ k0< kBPS<1 k0 C0<Cs<0
∞ 1 C0
 Key Points
• Equilibrium (lever-law) at S/L interfaces requires:
– Tie-line “jumps”
– Solidus & liquidus slopes (ms & ml)
– Segregation coefficient, k0
• Motion of the interface causes mass transport
– Diffusion boundary layer
– Solute rejection
• Gulliver-Scheil behavior leads to macrosegregation.
• Transients arise during solidification:
– Initial
– Steady-state
– Terminal
• Bouyancy convection treated with Burton-Prim-Schlichter theory