Solidification

Liquid cooled below its equilibrium melting temp (Tm) - there is a

driving force for solidification = GS - GL (< 0)
In reality the liquid does not spontaneously solidify - Why?
Liquid always needs to be cooled to a certain temp. below Tm for
solidification to begin - undercooling
e.g. under suitable conditions liquid nickel can be undercooled to ~
250 K below its melting temp. (1453°C) without any solidification
starting
Solidification has to start by the nucleation and subsequent growth
of solid particles in the liquid
If solid particles nucleate homogeneously (uniformly) throughout
the liquid then homogeneous nucleation - very high undercooling
required

Usually high undercoolings ~ 250 K would not be observed in
practice - walls of the container act a potent nucleation sites -
heterogeneous nucleation

Starting from a completely liquid state, when solid firsts starts
forming in the liquid
Consider volume of liquid at a temp. Tm-ΔT with G1 being the initial
free energy of the liquid
When a solid particle forms inside this liquid, the free energy
changes to G2 = G1 + ΔG

1
Phase Transformations
 Free energy change accompanying this transformation can be
written as below:
G2 = VSGVS + VLGVL + ASL" SL
G1 = (VS + VL ) GVL

( )
#G = G2 $ G1 = VS GVS $ GVL + ASL" SL
#G = $VS#GV + ASL" SL
#GV = GVL $ GVS
For undercooling of ΔT,
LV "T
"GV = GVL # GVS =
! Tm
Where LV is the latent heat of fusion per unit volume
For T < Tm, ΔGV is a positive quantity and therefore,
there is a negative contribution of ΔGV to the total free energy
!
change (ΔG)
However, there is a positive contribution to ΔG from the additional
interface created between the solid particle and the surrounding liquid

This excess free energy can be minimized by adopting the solid
particle of shape which has the minimum surface area to volume
ratio
If γSL is isotropic, this shape is a sphere
For a spherical solid nucleus of radius r,
4 3
"Gr = #"GV $r + % SL 4$r 2
3

2
Phase Transformations
Interfacial energy increases as r2 whereas volume free energy
decreases as r3
Leads to a critical radius (size) of nucleus, r*, corresponding to the
maximum excess free energy of the system, ΔG*
For, r > r*, it is favorable for the solid nucleus to grow since ΔG
reduces
For r < r*, it is unfavorable for the solid nucleus to grow since ΔG
increases - more favorable for the solid particle to dissolve back into
the liquid

3
Phase Transformations
 d"Gr
r = r*, =0
dr
2#
r* = SL
"GV
3
16$# SL
"G* = 2
3( "GV )

Substituting for ΔGV,

LV "T
! "GV = GVL # GVS =
Tm
% 2$ T ( 1
r* = ' SL m *
& LV ) "T
% 16+$ 3 T 2 ( 1
"G* = ' SL m
2 * 2
& 3LV ) ( "T )

ΔG* and r* reduce with increasing ΔT

The same relationship can be derived for the Gibbs-Thomson excess
! free energy concept (discussed previously in Thermodynamics)

A solid sphere of radius r will have a greater free energy as
compared to that of infinitely flat solid surface by an amount,
2#Vm Per mole
"G =
r
2#
"GV = Per unit volume
r
2#
$ r* =
"GV 4
Phase Transformations

!
What is the mechanism of the process of homogeneous nucleation of
solid precipitates within the liquid?

Dilatometry measurements indicate that at the melting point, liquid
has a 2-4% greater volume as compared with the solid
The atomic positions in the liquid, averaged over a period of time, are
completely random
However, at any particular instant if you freeze the liquid atoms in
their positions it will look like shown below:

Many small close-packed clusters of

atoms that resemble the solid structure

5
Phase Transformations
 On average, the number of such spherical clusters of radius r:
$ " #Gr '
nr = n0 exp& )
% kT (
4
"Gr = #"GV $r 3 + % SL 4$r 2
3
! n0 is the total number of atoms, ΔGr is the excess energy associated
with a cluster of radius r
For T > Tm, the above relationship holds for all r
!
But for T < Tm, this relationship only holds for r < r*
Because, r > r* clusters are already part of the solid and no longer
part of the liquid

As r increases, ΔGr increases very rapidly and the number of such
clusters decreases exponentially
e.g. 1mm3 of copper at its melting point ~ 1020 atoms has,
0.3 nm radius clusters (~ 10 atoms) = 1014
0.6 nm radius clusters (~ 60 atoms) = 10
Only approximate values, but gives you an idea

There is a maximum cluster size rmax~ 100 atoms that actually has a
finite probability of occurring in the liquid
This calculation was done at T = Tm, for T < Tm, with increasing ΔT
There is an increasing contribution to ΔGr from ΔGv and the
maximum cluster size (rmax) tends to increase,
While r* and ΔG* keeps decreasing with increasing ΔT

For a critical undercooling ΔTN, or greater there is a high probability
that some clusters will reach size r* and then nucleation of the solid
can take place

Therefore, homogeneous nucleation can take place when the liquid
undercooling reaches ~ ΔTN, or below 6
Phase Transformations
 The formation of a nucleus of critical size can be considered as a
thermally-activated process,
Solid-like cluster in a liquid has to overcome a nucleation barrier ΔG*
before it becomes a stable solid nucleus
Probability of achieving this energy is proportional to,
$ " #G * '
exp& )
% kT (
Based on this a homogeneous nucleation rate can be evaluated
Liquid contains C0 atoms per unit volume
Number of clusters that have reached the critical size = C*
!
$ " #G * '
C* = C0 exp& )
% kT (
$ " #G * ' Homogeneous nucleation rate
N hom = f0C0 exp& )
% kT ( Nuclei m-3 sec-1
f0 is the frequency at which an atom is attached to a critical sized
cluster - complex function depends on number of different
! parameters ~ 1011 7
Phase Transformations
 C0 ~ 1011 atoms m-3
For a reasonable nucleation rate of 1 cm-3 sec-1,
ΔG* ~ 78 kT
Another version of the nucleation rate,

$& (
"A &
N hom = f0C0 exp% 2
)
&' ( #T ) &*
3
16+, SLTm2
A= A is relatively insensitive to temperature
3LV2 kT

Due to the (ΔT)2 term in the denominator, the nucleation rate increases
very rapidly for undercoolings larger than ΔTN

Virtually no nuclei are formed for undercoolings less than ΔTN
and then there is sudden explosion of nuclei forming as soon as the
undercooling exceeds ΔTN
ΔTN ~ 0.2 Tm for most metals, .i.e. ~ 200 K

8
Phase Transformations
 In reality homogeneous nucleation rarely occurs in liquids
Instead heterogeneous nucleation occurs at crevices in mold walls
containing the liquid or other inhomogeneities in the liquid

ΔG* can be reduced and nucleation promoted at lower undercoolings by
reducing the interfacial energy term

If nucleus forms in contact with the mold wall, and the relative
interfacial energies as shown above,
" ML = " SM + " SL cos#

cos# =
(" ML $ " SM )
" SL

The total interfacial energy of the system is minimized if the embryo

has the shape of a spherical cap with a wetting angle θ
! The excess free energy associated with the embryo forming on the
mold wall is given by

"Ghet = #VS"GV + ASL$ SL + ASM$ SM # ASM$ ML

9
Phase Transformations
 The area of solid/liquid interface and solid/mold interface is added
while the area of liquid/mold interface is removed

The equation for heterogeneous nucleation can be re-written as
follows,
& 4 )
"Ghet = '# $r 3"GV + 4$r 2% SL * S(, )
( 3 +
where
2
S(, ) = ( 2 + cos, )( 1# cos, ) / 4

This equation for heterogeneous nucleation is the same as that for

homogeneous nucleation except the factor S(θ) which is often referred
! to as a shape factor
S(θ) <= 1 and depends only on the shape of the spherical cap

Comparison of ΔG*hom and ΔG*het shown below:

10
Phase Transformations
 Also for heterogeneous nucleation,
d"Gr
r = r*, =0
dr
2#
r* = SL
"GV
3
16$# SL
"G* = 2
.S(% )
3( "GV )
* *
"Ghet = "Ghom .S(% )

Activation barrier for heterogeneous nucleation is smaller than that

for homogeneous nucleation by a factor of S(θ) <= 1
! For θ = 10°, S(θ) ~ 10-4, barrier for heterogeneous nucleation can be
very small as compared to that for homogeneous nucleation
For θ = 30°, S(θ) = 0.02, and for θ = 90°, S(θ) = 0.5

When n1 atoms are in contact with the mold wall, then the number of
nuclei should be given by,
$ " #G* '
n* = n1 exp& het
)
% kT (

Heterogeneous nucleation favorable when ΔG*het is sufficiently small

Assuming that the critical value for ΔG*het ~ 78 kT (similar to
! homogeneous nucleation), the undercoolings required for heterogeneous
nucleation are much smaller than that required for homogeneous
nucleation

11
Phase Transformations
 Volume rate for heterogeneous nucleation would be,
$ "#G* '
N het = f1C1 exp& het
)
% kT (
Where f1 is the frequency of atoms attaching to the heterogeneously
nucleated cluster and C1 is the number of atoms in contact with the
heterogeneous nucleation site per unit volume of liquid
!

12
Phase Transformations
 Heterogeneous nucleation is enhanced by adding inoculants to the
molten metal in order to refine the grain size during solidification.
Typically the inoculant reacts with one of the components of the melt
and forms a solid compound which acts a heterogeneous nucleation site
Effectiveness of the inoculant depends on the ability of the soid nucleus
to wet the compound formed by the inoculant - depends on wetting angle
and surface roughness

Melting
Nucleation during solidification requires some undercooling, but melting
always occurs at the equilibrium melting temperature, even for high
heating rates - Why?
Reason lies in the relative values of the 3 interfacial energies of interest,
" SL + " LV < " SV

Therefore wetting angle θ = 0°

No superheating required for nucleation of the liquid
!
The actual atomic mechanism of melting is still not well understood.
Solid to liquid transformation in metals typically corresponds to an
increase in vacancy concentration by ~ 10% - difficult to explain based
on usual defect structures

Good model of melting - still to be developed!

13
Phase Transformations
Growth of Pure Solid

Two types of solid/liquid interfaces:
atomically rough or diffuse (typically metallic systems)
atomically flat or sharp defined interface (typically non-metals)
Difference in structure leads to differences in the migration
mechanisms of these two types of interfaces
Rough interfaces: continuous growth
Flat interfaces: lateral growth involving ledges

Continuous growth
Activation energy barrier for atom crossing the solid/liquid interface
is ~ that of diffusion in the liquid phase
Driving force for solidification,

LV "Tint
"G =
Tm
Net rate for solidification,

v = k1"Tint
k1 is related to the boundary mobility and is usually very high for
rough diffuse interfaces
Therefore, for normal solidification rates, ΔTint can be ignored -
solid/liquid interface at the equilibrium Tm

14
Phase Transformations
Solidification of metals is a diffusion controlled process - interface
mobility is very high, so not rate controlling

True only for diffuse interfaces where atoms can be attached with
equal probability at any location on the solid surface - no preferential
site for attachment of atoms at the solid/liquid interface - continuous
growth
Interface is disordered and atoms arriving at random positions on the
solid will not significantly disrupt the equlibrium configuration of the
interface

Lateral Growth
Atomically smooth, sharply defined solid/liquid interface
Materials with a high entropy of melting typically favor such
interfaces - usually close packed interfaces
Minimum free energy corresponds to minimum internal energy -
minimum number of broken solid bonds
Single atom arriving at
solid surface and
joining -

Attachment to flat
surface or terrace leads
to 4 broken bonds
increased

Attachment to ledge
leads to 2 broken bonds
increased

Attachment to jog leads
to no increase in
broken bonds 15
Phase Transformations
 Atoms have a much higher probability of attaching to a jog as
compared to a ledge or terrace
Atoms attaching to a terrace have a greater probability of re-
dissolving in the liquid
Smooth solid/liquid interface can therefore be expected to advance by
the lateral growth of ledges
Note that ledges and jogs are non-equilibrium features on the
interface - growth will be dependent on how jogs and ledges are
supplied - by 3 ways:
1. Repeated surface nucleation
2. By spiral growth
3. From twin boundaries

Surface nucleation
While single atom attaching to a smooth terrace is unstable, if a
sufficiently large number of atoms can cooperatively come together
and form a disc shaped layer it can become stabilized and continue to
grow
Disc formation is a 2D analogue of 3D cluster formation
Critical radius (r*) associated with the two-dimensional nucleus
decreases with increasing ΔT - very ineffective at low undercoolings
Growth velocity of disc-shaped nucleus:
v " exp(#k 2 / $Tint )

16
Phase Transformations
 Spiral growth
If solid has dislocations intersecting the solid/liquid interface -
automatic steps at the interface
e.g. screw dislocation in a solid block - step or ledge on surface (see
figure below) - addition of atoms to this ledge will cause it to rotate
about the point where it emerges from the solid block - spiral growth

Initially angular velocity will be fast in spiral growth near the

dislocation core - ledge develops in a growth spiral - finally the spiral
growth reaches a constant angular velocity
2
v " k 3 ( #Tint )

! 17
Phase Transformations
Comparing,
For a given solid growth rate,
Continuous growth requires the least undercooling while surface
nucleation requires the largest undercooling
At the same undercooling,
Fastest growth rate for a rough interfaces growing by continuous
growth while for smooth interfaces surface nucleation (or repeated
nucleation) is much slower than spiral growth

Growth from Twin Intersections

During solidification of materials showing faceting (smooth
interfaces), there is often a tendency for form two crystals with
twinned orientations together
Interfacial facets will therefore intersect at the twin boundary - acts as
permanent source for steps providing a growth mechanism similar to
spiral growth

18
Phase Transformations