Lecture 3: Solidification

Driving Force
Solidification is undoubtedly the most important processing route for
metals and alloys. Consider a pure metal (Fig. 1). At the fusion temperature Tf , G = 0 so that
G = Hf Tf Sf = 0

or

Hf = Tf Sf

where Hf is the latent heat of fusion and Sf is positive for melting.

For any temperature other than Tf ,
G = H T S
" Hf T Sf
= Sf (Tf T ) = Sf

T
!"#$

undercooling

The driving force is therefore proportional to the undercooling provided

that the latent heat and the entropy of fusion do not vary much with
temperature.

Fig. 1: Driving force for solidification

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Nucleation and Growth

Homogeneous nucleation occurs only at very large T , for example,
in containerless experiments where a pure liquid is isolated from its
environment. In general, solidification is by heterogeneous nucleation,
either on impurity particles or wherever the liquid comes into contact
with the container surface.
The velocity v of the transformation front is related to the difference in the rate of liquidsolid atom jumps and the solidliquid atom
jumps:

i.e. for small G

&'
%
&(
%
G
Q
1 exp
v exp
kT
kT

v G
v T

The last two proportionalities assume that the undercooling is small, in

which case exp{x} " 1 + x.

Fig. 1: Barrier to interface motion

Geometry of Solidification
Fig. 2 shows the grain structures possible. The chill zone contains
fine crystals nucleated at the mould surface. There is then selective
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

growth into the liquid as heat is extracted from the mould. If the liquid
in the centre of the mould is undercooled sufficiently there may also be
equiaxed grains forming.

solid

chill
zone

mould

liquid
T

+ve T gradient
columnar growth

solid liquid

- ve T gradient
equiaxed growth

Fig. 2: Geometry of solidification

Equiaxed growth in a pure metal shows morphological instabilities,

i.e. thermal dendrites (Fig. 3). This is because a small perturbation at
the interface ends up in even more supercooled liquid so the interface
becomes unstable. Dendrites have preferred growth directions, almost
always < 1 0 0 > in cubic metals.

Fig. 3: Thermal dendrite formation when the temperature gradient in the liquid is negative.

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Alloys: Solute Partitioning

Dendrite formation is extremely common in alloys, where solute
partitions between the solid and liquid phases (Fig. 4). By convention,
we shall label the composition of the solid phase which is in equilibrium
with the liquid as C SL and a similar interpretation applies to C LS . C0
represents the average composition of the alloy. The partition coefficient
k is written
C SL
k = LS
C

frequently < 1

Under equilibrium conditions the compositions of the solid and liquid at

all stages of solidification are given by a tieline of the phase diagram,
and the proportions of the phases at any temperature by the lever rule.
Temperature

liquid

solid

CSL C0 CLS
Concentration

concentration

solid

liquid
CLS

CSL

C0

Distance

Fig. 4: Solidification under equilibrium conditions

In practice, equilibrium will only be maintained at the interface

(Fig. 5), where the compositions agree with the phase diagram. The
composition of the solid follows the solidus. Steady state solidification
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

occurs when the temperature is T , when C SL = C0 and there is no

further solute partitioned into the remaining liquid.
From AH8,

%
&
x
C = C + (C C ) exp
D/v
%
&
x
C (1 k)
exp
C0 + 0
k
D/v
SL

LS

SL

Note that D/v has dimensions of length, approximately the diffusion

distance into the liquid. It is typically about 0.5 mm but can be just a
few micrometres in rapid solidification processing.

Fig. 5: Solidification under nonequilibrium conditions

1602, 1603

Levitation

T607

dendrite, cast aluminium

Video of succinonitrile

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Lecture 4: Solidification
Mixing in the Liquid
Convection can cause mixing, reducing solute gradients in the liquid.

A limiting case is where the liquid is uniform in composition

throughout solidification (Fig. 6).

For liquid of composition C L (note change in terminology since it is
no longer in equilibrium with solid), the composition of the solid phase
at the interface is given by C S = kC L . If fs is the fraction solidified,
then from AH9,
(C L kC L )df = (1 f )dC L
) CL
) fs
df
dC L
=
L
1f
0
C0 C (1 k)
so that

C L = C0 (1 fs )k1

and

C S = kC0 (1 fs )k1

This last relation is known as the Scheil Equation.

Fig. 6: Mixing in the liquid

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

The redistribution of solute is larger when there is mixing in the liquid

(Fig. 7). This can be used in zone refining, e.g. of silicon, with a repeated
sweeping in one direction, leading to purification (elements for which
k > 1 would be swept in the other direction). By moving the furnace to
and fro, one can obtain zone levelling.
convection

concentration

concentration

diffusion only

C0
kC 0

C0

kC 0

S
L

S
moving furnace, zone refining

Fig. 7: Distribution of solute

Constitutional Supercooling (AH10)

Solute is partitioned into the liquid ahead of the solidification front.
This causes a corresponding variation in the liquidus temperature (the
temperature below which freezing begins). There is, however, a positive
temperature gradient in the liquid, giving rise to a supercooled zone
of liquid ahead of the interface (Fig. 8). This is called constitutional
supercooling because it is caused by composition changes.
A small perturbation on the interface will therefore expand into a
supercooled liquid. This gives rise to dendrites.
It follows that a supercooled zone only occurs when the liquidus
temperature (TL ) gradient at the interface is larger than the temperature
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Fig. 8: Diagram illustrating constitutional supercooling.

gradient:

*
TL **
T
>
x *x=0
x

i.e.,

*
CL **
T
m
>
x *x=0
x

where m is the magnitude of the slope of the liquidus phase boundary

on the phase diagram. From AH8 we note that
*
CL **
C LS C SL
=
x *
D/v
x=0

so that the minimum thermal gradient required for a stable solidification

front is

mC0 (1 k)v
T
<
x
kD
It is very difficult to avoid constitutional supercooling in practice because the velocity required is very small indeed. Directional solidification with a planar front is possible only at low growth rates, for example
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

in the production of silicon single crystals. In most cases the interface

is unstable (Fig. 9):
Dendrites

Cells

Fig. 9: Cells and dendrites

Microsegregation
There is little or no segregation ahead of dendrite tip because the
main partitioning is of solute trapped between the dendrite side arms
(metallography specimen M2). The scale of this segregation can be
reduced by increasing the solidification rate because the dendrite arm
spacing then decreases. A rough explanation of this phenomenon is
that a large undercooling permits the creation of a greater surface, i.e.
smaller dendrite spacing. Segregation can also be reduced by inducing
a greater number density of nuclei, for example, by adding inoculants
such as titanium boride in aluminium (the TiB2 particles coat with
Al3 Ti which in turn nucleates aluminium. Anything which refines the
scale of the microstructure will lead to less segregation.
This chemical segregation can lead to the formation of phases (such
as eutectic) which are not expected under equilibrium conditions.
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

806

Cell formation

807

Plan view of cells

786

Dendrites

1281

Solute segregation

1287

Dendrite arm spacing

828,829

Al-Si, Al-Si-Na

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Lecture 5: Solidification
Solidification Processing
A key phenomenon in solidification processing is the transfer of heat
by radiation, direct contact with the mould, conduction through air and
convection in the air gap between the mould and ingot. Casting situations may be divided according to whether or not significant thermal
gradients are set up in the solidifying metal.
The transfer of heat across a gap (an interface) is given by
q = hT
where q is the heat flux, h is the thermal conductance of the interface
and T the temperature difference between the source and the sink.
Values of the heat transfer coefficient vary widely: 10 102 for large
air gap; 103 104 for normal castings with gravity contact; 105 106

for excellent contact as in pressure casting. The temperature profile

obtained depends on the a comparison of the heat conductance of the
interface with that of the whole casting. The thermal conductance of
a casting is written K/L where where K is the thermal conductivity
of a casting of length L in the direction of heat flow. The ratio of the
thermal conductance of the interface (h) to that of the casting, termed
the Biot number:
Bi =

hL
h
=
K/L
K

.
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

temperature

temperature

mould

liquid
q

6T

T
Small Bi

T
0
distance
T

Large Bi

T
0
distance

Fig. 10: The two different heat transfer situations

that arise in casting.

For small Bi the thermal resistance of the interface dominates that

of the casting, which therefore remains at approximately a constant
temperature. This is called Newtonian cooling.
Newtonian cooling is favoured by high conductivity metals such as
Al and Cu or for small castings.
Chill Mould Castings
Chill casting uses a massive metal mould with a large thermal capacity
or a water cooled metal mould so that Bi & 1. When the metal is

poured in at a temperature well above its melting temperature Tm , it

first cools and begins solidification when T = Tm . For the first stage,
illustrated in Fig. 11, it follows that
q = hT = T CV L

so that

hT
T =
LCV

where CV is the heat capacity of the liquid (J m3 K1 ) and T is the

cooling rate of the liquid. Note that hT is the heat flow into the mould
and T CV L is the heat liberated by the liquid.
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Fig. 11: Chill mould. The liquid is at a temperature

higher than Tm , the melting temperature.
This cooling of the liquid will be followed by the second stage which
is isothermal solidification (Fig. 12). For this case, the speed v of the
solidification front can be obtained by balancing heat evolution against
heat extraction:
q = hT = vHF

i.e.,

v=

hT
HF

where q is the heat flux and HF is latent heat released on solidification.

Fig. 12: Chill mould. The casting is now at the melting temperature Tm

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Casting Processes

CASTING PROCESSES

Permanent Mould

Temporary Mould

Gravity die
Pressure die
Centrifugal
Continuous

Sand
Plaster
Lost wax

Fig. 12: Procedure for sand casting. This process is

usually automated.

Sand Casting
In sand casting a mould is made by packing sand around a pattern
(Fig. 13), removing the pattern and hardening the sand with polymer
or silicate. The metal is then poured into the resulting mould in a
controlled manner so that solidification proceeds in an orderly manner
without leaving any holes or porosity. To avoid the blockage of channels, chill blocks may be used to permit certain parst to solidify first.
Risers are used to feed metal into the mould as it freezes and contracts.
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

Examples of sand cast components include automobile engine blocks,

ship propellers, huge steel components for power plant boilers, nodes
for North Sea oil platforms.
Die Casting
Die casting involves the injection of a lowmelting temperature alloy into a steel mould. Zn-4Al wt%, which is a eutectic composition, is
a classic diecasting alloy melting at 419 C. The process is usually for
nonstructural components since the casting inevitably contains internal pores and because Zn does not have particularly good mechanical
properties. However, productivity can be high and the surface finish is
good. Toys, boxes etc.

Fig. 13: The metal may be injected in at pressure

Centrifugal Casting
Centrifugal casting has a rotating cylindrical mould; molten matter
poured into the mould solidifies to give a pipe. The quality is excellent
and the process is used for making long pipelines for the chemical industry.
Continuous Casting
Continuous casting is illustrated in AH12 and is the primary casting
Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia

process in the manufacture of steel and aluminium.

Others
Lost wax process, rapid solidification.
1444

Continuous casting
Continuous casting video (Scunthorpe Steel)

865

Melt spinning

Part IB Mat. Sci. and Metall.: Metals & Alloys, by H. K. D. H. Bhadeshia