Вы находитесь на странице: 1из 114

Natural Sciences Tripos Part IB

MATERIALS SCIENCE
Course A: Phase Transformations
Dr E.R. Wallach
Michaelmas Term 2013-14
IB
AH1 Course A: Phase Transformations AH1



rob.wallach@msm.cam.ac.uk 2013-14
INTRODUCTION

Improving metallic materials is a vital activity at the leading edge of science and technology.
Metals and their alloys have offered unrivalled combinations of properties and reliability over
many centuries, generally at affordable costs. They are versatile because subtle changes in
their composition (alloying) or microstructure can cause significant variations in their properties,
e.g. the strengths of commercial steels range from as low as 50 MPa to up to 5500 MPa.

It is possible to specify and obtain metals with very specific properties and so metals continue to
be used in many different applications despite the increased competition from other classes of
material such as polymers, ceramics and composites. Hence an understanding of the
development of microstructures in metals and alloys is essential for materials scientists.

Course A builds on the coverage of metals and alloys in Part IA. Whereas Part IA dealt mainly
with the thermodynamic aspects, kinetics are included and emphasised in this course in the
treatment of phenomena such as diffusion, solidification and solid-state transformations.

Objectives
At the end of this course you should understand:
- kinetics of transformations in materials by considering diffusion in terms of mechanisms,
rate equations and the effects of microstructure and alloy composition;
- how grain structures in cast materials arise, how the distribution of solute in liquids affects
the microstructures and concentration profiles in solidified products, simple analysis of
heat-flow in castings, the relative advantages of different commercial casting processes,
microstructural evolution in cast irons and aluminium cast alloys;
- nature of grain boundaries, the processes of recovery, recrystallisation and grain growth in
cold-worked metals, the effects of alloy additions on the kinetics of these processes;
- mechanisms by which precipitation occurs in two-phase systems including the importance
of metastable phases, precipitate free zones, precipitate coarsening, real alloy systems
based on aluminium alloys and nickel-base superalloys;
- combination of diffusion controlled and diffusional transformations in steels leading to the
wide range of commercial steels and different mechanisms by which properties can be
optimised;
- sustainability issues based on life cycle analyses.













Dendrites in water (left) and in a Cu-P cast alloy (right).
AH2 Course A: Phase Transformations AH2



rob.wallach@msm.cam.ac.uk 2013-14
CONTENTS

Introduction 1
Contents 2
Resources 3
Guide to Symbols 6

1. Diffusion 8
1.1 Introduction 8
1.2 Atomistic Approaches to Diffusion 9
1.3 Continuum Approaches to Diffusion 11
1.4 Diffusion in Metallic Alloys 16
1.5 Diffusion and Microstructure 18
1.6 Thermodynamics of Diffusion 21
1.7 Diffusion in Polymers 25

2. Solidification 26
2.1 Introduction 26
2.2 Solidification Microstructures of Pure Metals 29
2.3 Solidification Microstructures of Metallic Alloys 31
2.4 Heat Flow in Solidification 40
2.5 Solidification Processing (Casting Methods) 44
2.6 Peritectic Transformations 49

3. Microstructure and Properties 50
3.1 Dislocations 50
3.2 Grains and Grain Boundaries 55
3.3 Deformation, Recovery, Recrystallisation, Twinning and Grain Growth 57
3.4 Precipitates 66

4. Alloy systems 77
4.1 Ternary Phase Diagrams 77
4.2 Cast Irons 80
4.3 Steels 84
4.4 Aluminium Alloys 91
4.5 Titanium Alloys 96
4.6 Nickel-base Superalloys 97

5. Sustainability of Metals 101
5.1 Life-Cycle Analysis 101
5.2 Lifetime: Degradation of Metals 104
5.3 Notes on using Cambridge Engineering Selector software 105

Glossary 107
AH3 Course A: Phase Transformations AH3



rob.wallach@msm.cam.ac.uk 2013-14
ADDITIONAL RESOURCES

A. Web and computer resources
Specific course material is available from the website of the Department of Materials Science:
www.msm.cam.ac.uk/Teaching/mat1b/index.html


Additional information to assist with understanding of the course is as follows.

1. DoITPoMS Micrograph Library has many images of microstructures, together with
descriptions and explanations:
www.msm.cam.ac.uk/doitpoms/miclib/index.php

2. DoITPoMS Teaching & Learning Packages
www.msm.cam.ac.uk/doitpoms/tlplib/index.php
include the following packages which are directly relevant:
Casting
Diffusion
Electromigration
Introduction to Dislocations
Microstructural Examination
Optical Microscopy
Phase Diagrams and Solidification
Solidification of Alloys
The Jominy End Quench Test

3. MATTER software www.matter.org.uk
Several excellent resources (select using pull-down menu on left-hand side of page)
3a. www.matter.org.uk/solidification especially the two sections called:
Redistribution and
Cell, dendrite and grain structure
3b. aluMATTER http://aluminium.matter.org.uk has considerable information on
aluminium alloys, including strengthening mechanisms and processing:

3c. SteelMATTER and also SteelUniversity have information on steels, including processing
and hardness testing
www.matter.org.uk/steelmatter
www.steeluniversity.org/

4. Additional resources related to this course are available on the department website at
www.msm.cam.ac.uk/phase-trans/teaching.html



Please send Rob Wallach any other good sites or sources that you find!
AH4 Course A: Phase Transformations AH4



rob.wallach@msm.cam.ac.uk 2013-14
B. Books available in Departmental library for reference and borrowing

Porter D.A., Easterling K.E. and Sherif M.Y., Phase Transformations in Metals and Alloys,
2
nd
edition, (Chapman and Hall, 1992) OR
3
rd
edition, (CRC Press, 2009) Need to check it is the corrected 2009 3
rd
edition.
The correct version has 520 pages and NOT 500 pages.
This is a key book for the course.

Newey C. and Weaver G., Materials Principles & Practice (Butterworths, 1990).
A good book, originally commissioned for the Open University materials courses.

Weidmann G., Lewis P. and Reid N., Structural Materials (Butterworths, 1990).
Lots of information on steels and microstructures.

Abbaschian R., Abbaschian L. and Reed-Hill R.E., Physical Metallurgy Principles, 4
th
edition,
(PWS-Kent Publishing Co., Boston, Mass., 2010).

Smallman R.E. & Bishop R.J., Modern Physical Metallurgy and Materials Engineering
Science, 6
th
edition, (Butterworth-Heinemann, 1999).

Honeycombe R. W. K. and Bhadeshia H. K. D. H., Steels: Microstructure and Properties,3
rd

edition, (Butterworth-Heinemann 2006).

Polmear I. J., Light Alloys From Traditional Alloys to Nanocrystals, 4
th
edition, (Elsevier
Butterworth-Heinemann, 2006).



C. Question sheets
There is one revision question sheet which covers material from Part IA and then four weekly
question sheets are available from the website below. Answers to the question sheets will be
posted on the website normally about seven days after the week in which the question sheet
should be attempted.
www.msm.cam.ac.uk/teaching/partIB/courseA.php



D. Metallographic specimens
Metallographic specimens, labelled M1 to M25, are an integral part of this course as well as
providing experience which will be useful during the Manufactured Materials Project. The table
on the next page specifies the alloy and processing conditions corresponding to each specimen.

Questions pertaining to the specimens are included on the questions sheets (see C above). In
some cases, photographs of the specimens are provided on the question sheets. For others, as
identified on the question sheet, you will be expected to look at the actual specimens
themselves. This might be conveniently undertaken before or after your practical session in the
Department (time may be also provided during the actual practical sessions AP1 to AP4).
Phase diagrams in the IB Data Book provide additional information and it is essential to refer to
these when looking at and interpreting the microstructures.

The specimens themselves can be examined using the optical microscopes in the Part IB
Laboratory. Choose a variety of magnifications to examine the prepared mounted and polished
specimens, and record your observations carefully. Your supervisor will go over these with you.

AH5 Course A: Phase Transformations AH5



rob.wallach@msm.cam.ac.uk 2013-14
D. The 'M' Series of Metallographic Specimens

The column marked QS shows the question sheet on which reference is made to the
individual specimen and the number in parenthesis refers to the question number.

Code QS (#)
Composition
(wt %)
Condition Etchant
M1 AQ1(4) eutectoid steel heat-treated at 1200C nital
M2 AQ2(5) Cu-30Ni as-cast (metal mould) FeCl
3

M3 AQ2(5) Al-5Cu as-cast (metal mould) NaOH
M4 AQ2(5) Al-4Cu as-cast (sand mould) NaOH
M5 AQ2(5) Sn-21Cu as-cast (sand mould) HCl
M6 AQ2(5) Al-33Cu as-cast (directionally solidified) NaOH
M7 AQ2(4) Al-12Si as-cast (metal mould) as-polished
M8 AQ2(4) Al-12Si-0.02Na as-cast (metal mould) as-polished
M9 AQ3(3) Cu-30Zn annealed + cold rolled 50% FeCl
3

M10 AQ3(3) Cu-30Zn M9 + 30 min.@ 600C FeCl
3

M11 AQ3(3) Cu-30Zn M9 + 30 min.@ 800C FeCl
3

M12 AQ3(3) Cu-30Zn
anneal. + cold roll 70% +
30min.@ 800C
FeCl
3

M13 AQ4(2) Fe-18Cr-8Ni cast + rolled HCl/HNO
3

M14 AQ4(2) Zn cold rolled HCl
M15 AQ4(2) Fe-0.2C explosively deformed nital
M16 AQ4(2) Fe-30Ni-0.3C quenched to -80C as-polished
M17 AQR(14) Fe-0.8C water quenched from 800C nital
M18 AQR(14) Fe-1.0C oil quenched + 23 hr. @ 700C nital
M19 AQR(14) Fe-0.8C
annealed + 10 min. @ 700C +
quenched
nital
M20 AQ4(2) Fe-3.2C-Si... as-cast nital
M21 AQ4(2) Fe-3.2C-Si...(SG) as-cast nital
M22 AQ3(3) Al-4Cu solution treated + over-aged NaOH
M23 AQ4(2) Cu-40Zn
annealed + quenched + 1 hr.
@ 300C
FeCl
3


M24 AQ4(2) Cu-40Zn annealed 850C + air cooled FeCl
3


M25 AQ2(4) Al As-cast (macrostructure) Poulton's







AH6 Course A: Phase Transformations AH6



rob.wallach@msm.cam.ac.uk 2013-14
GUIDE TO SYMBOLS

The following is a guide to symbols used in the handout. An attempt has been made to use
each symbol to represent just one quantity where possible. However there are some instances
in which a particular symbol is so commonly used for different quantities (i.e. G for Gibbs free
energy and shear modulus) that it would be more confusing to use something else. In these
cases, it should be clear which interpretation is intended from the context.

A Area (m
2
)
b Burgers vector (m)
B
i
The Biot number
c
V
volumetric heat capacity (J m
-3
K
-1
)
C concentration (composition) (at%, wt% or fraction)
C
0
overall composition (m
-3
)
C
!
composition of phase ! (m
-3
)
C
!"
composition of phase ! in equilibrium with phase " (m
-3
)
D diffusion coefficient (diffusivity) (m
2
s
-1
)
E
V
energy per unit volume (J m
-3
)
d diameter (m)
F force (N)
f fraction
f
S
fraction solidified
G Gibbs free energy (J mol
-1
) OR shear modulus (J m
-3
)
h interfacial heat transfer coefficient (W m
-2
K
-1
)
H enthalpy (J mol
-1
)
J flux (m
-2
s
-1
)
k partition coefficient
k
B
Boltzmann constant (J K
-1
)
K thermal conductivity (W m
-1
K
-1
)
L length (m)
m gradient of the liquidus line on the phase diagram (at.% K
-1
or wt.% K
-1
)
N number
n number per unit volume (m
-3
)
P Drag pressure (force per unit area on a grain boundary) (N m
-2
)
q heat flux (J m
-2
s
-1
)
Q activation energy (J mol
-1
) OR rate of heat generation (J m
-3
s
-1
)
Q
f
activation energy for vacancy formation (J mol
-1
)
r radius (m)
r
*
critical radius (m)
S entropy (J mol
-1
K
-1
)
AH7 Course A: Phase Transformations AH7



rob.wallach@msm.cam.ac.uk 2013-14
T temperature (K)
T
L
liquidus temperature (K)
T
m
melting temperature (K)
t time (s)
U stored energy per unit volume (J m
-3
) OR strain energy per unit volume (J m
-3
)
v velocity (m s
-1
)
W work per unit volume (J m
-3
)
W
n
work of nucleation per unit volume (J m
-3
)
x distance (m)
X concentration (composition) (at.%, wt.% or fraction)
z atomic spacing (m)
! thermal diffusivity (m
2
s
-1
)
" grain boundary width (m)
# strain
$ surface energy (J m
-2
)
% jump distance (m)
& jump frequency (s
-1
)
' stress (N m
-2
)
( misorientation angle ()
chemical potential (J mol
-1
)
) dislocation density (m
-2
)
AH8 Course A: Phase Transformations AH8



rob.wallach@msm.cam.ac.uk 2013-14
1. DIFFUSION

1.1 Introduction
Diffusion is a transport phenomenon caused by the physical motion of chemical species
(molecules, atoms or ions), heat or similar properties of a medium (gas, liquid or solid).
Diffusion is a consequence of concentration differences (or, strictly, chemical potential
differences see section 1.7.1). In general, species move from high concentration areas to low
concentrations areas until uniform concentration is achieved.
Mass transport can generally involve:
fluid flow dominant in gases or liquids (e.g. convection currents);
viscous flow flow of a viscous material, generally amorphous or semi-crystalline (e.g.
glasses and polymers) due to the forces acting on it at that moment;
atomic diffusion principal mechanism in solids and in static liquids (as occurs in
solidification).
Atomic diffusion occurs during important processes such as:
solidification of materials [section 2];
precipitation strengthening, e.g. Al-Cu alloys [MMS IA and section 3];
annealing of metals to reduce excess vacancies & dislocations formed during working
[section 3];
creep [MMS IA and section 4.6]
manufacture of doped silicon as used in many electronic devices [MMS IB: course F].
Atomic diffusion in crystalline solids is:
jumping of atoms on a fixed network of sites within a crystalline lattice;
generally the consequence of a thermodynamic driving force, i.e. chemical potential
gradient (often, but not exclusively, a concentration gradient).
Diffusion can be regarded and analysed atomistically or
on a macroscopic (continuum) scale.
AH9 Course A: Phase Transformations AH9



rob.wallach@msm.cam.ac.uk 2013-14
1.2 Atomistic Approach to Diffusion
1.2.1 Random walk approach
Consider an atom jumping with a frequency & (number of jumps per sec) and each jump has a
magnitude or average distance %. The jump frequency & is given by:
& = &
0
exp (G/RT) where &
0
is the Debye frequency,
G is the activation energy,
T is temperature (in Kelvin).
For a number of random jumps, n, the mean distance x after time t is given is given by:

x = ! n = ! " t
Hence the diffusion distance is proportional to t
- shows the time dependence of diffusion
- slightly oversimplified as random jump directions
are not likely in crystalline lattices.

1.2.2 Mechanisms of diffusion in crystalline solids
Atoms in the solid-state migrate by jumping into either interstitial or substitutional vacant lattice
sites.
Interstitial diffusion








direct exchange ring vacancy

Distortion of the lattice during atomic movement results in a barrier to motion. The magnitudes
of the activation energies G will be quite different for all of the above mechanisms.

Substitutional
diffusion
AH10 Course A: Phase Transformations AH10



rob.wallach@msm.cam.ac.uk 2013-14
1.2.3 Evidence for vacancy diffusion: the Kirkendall experiment
The vacancy diffusion mechanism (AH9) was confirmed by Smigelsklas & Kirkendall in 1947. In
their experiment, two different alloys (copper and brass) were joined together with inert markers
(fine molybdenum wires) at the interface to form a diffusion couple.
This is shown more simply (schematically) below as a diffusion couple comprising just A atoms
on the left and B atoms on the right.





time t = 0







time t > 0
In the frame of the sample, A atoms move to the right faster than B atoms move to the left.
Hence there is a net flux of atoms to the right (whilst the inert marker wires remain stationary).
Alternatively, the situation can be viewed in terms of vacancies. The equilibrium number of
vacancies in A is low and in B it is high. Hence vacancies tend to be created in B and then
diffuse from B to A where they are destroyed; hence there is a net vacancy flux to the left.
This leads to void formation in the alloy on the left-hand side, and these can be observed using
a microscope.
Observations are, in reality, made from outside the block (i.e. the block remains stationary from
an external perspective and does not move as shown pictorially above) with the result that the
inert marker wires appear to have moved to the left.

This experiment confirms that diffusion occurs via vacancies - other mechanisms such as the
ring (see page AH9) would not generate a net movement of the markers in either direction.
A B
J
A
J
B
A B
J
A
J
B
J
A
J
A
J
B
Inert fine
wires
AH11 Course A: Phase Transformations AH11



rob.wallach@msm.cam.ac.uk 2013-14
1.3 Continuum Diffusion
1.3.1 Ficks First Law of Diffusion in a uniform & constant concentration gradient
Consider two adjacent planes of atoms in a simple cubic crystal with lattice parameter a.














The number of impurity atoms contained in a plane of unit area is the impurity concentration in
the plane multiplied by a [since concentration is the number of atoms per unit volume].
So number of impurity atoms in plane 3 = C a atoms / m
2

number of impurity atoms in plane 4 = (C + !C) a atoms / m
2

Consider the atomic flux J of impurity atoms crossing in each direction


J
L!R

and


J
R!L

where the atomic flux is the number of atoms passing through a square metre per second:


J
L!R
=
1
6
" C #
$
%
&
a and

J
R!L
=
1
6
" C +#C $
%
&
'
a



Hence net flux from left to right (down the concentration gradient) is the difference

J
L!R
" J
R!L



J = !
1
6
" #C $
%
&
'
a = !
1
6
" a
2
(C
(x
OR J = !D
"C
"x

This is Ficks first law of diffusion where D is the diffusion coefficient with units of m
2
s
-1
.
It applies to steady state diffusion, when the concentration gradient does NOT vary with time.
Reconciling the continuum & atomistic approaches, the random walk approach (AH9) gave:
mean diffusion distance,

x = ! " t #

x = a ! t as the atomic spacing a replaces $
and from the above derivation of Ficks 1
st
law

D =
1
6
! a
2

Hence

x = 6Dt ,

x ! 2 Dt , or simply

x ! Dt

This approximation is useful for estimating diffusion distances.
distance x
Concentration C
of impurity
atoms (dark)

C
1 2 3 4 5 unit area
AH12 Course A: Phase Transformations AH12



rob.wallach@msm.cam.ac.uk 2013-14
1.3.2 Ficks Second Law of Diffusion results in a change in concentration gradient
Consider two nearby planes of atoms, denoted by A and B and the movement of atoms across
the element between them due to an imposed concentration gradient:







Flux into element J
A


= !D
"C
"x
#
$
%
&
'
(
A


Flux out of element J
B


= !D
"C
"x
#
$
%
&
'
(
B


= ! D
"C
"x
#
$
%
&
'
(
A
+)x
"
2
C
"x
2
#
$
%
&
'
(
*
+
,
,
-
.
/
/





Hence numbers of atoms entering and leaving element of area A in a time increment %t are:



N
in
= J
A
A!t and

N
out
= J
B
A!t

The change in concentration %C in this time is given by:



!C =
N
in
" N
out
volume


=
J
A
! J
B
( )
A "t
A "x


Rearranging the above equation gives:



!C
!t
= "
!J
!x
=
#
#x
D
#C
#x
$
%
&
'
(
)



In many situations, the diffusion coefficient D can be regarded as independent of concentration
and so:

Ficks second law is:

!C
!t
= D
!
2
C
!x
2




A B A B
concentration C
distance x
A B
AH13 Course A: Phase Transformations AH13



rob.wallach@msm.cam.ac.uk 2013-14

1.3.3 Solutions to Ficks second law

For general cases, numerical solutions are used (e.g. see Crank
1
for reference only).
Consider now two special cases for which analytical solutions are possible.

1.3.3.1 Fixed quantity of solute, i.e. finite source into a semi-infinite bar

Let there be a total number of atoms (per square metre of cross section), B, at one end of the
bar








Ficks 2
nd
law can be solved using the following two boundary conditions
i. initially, the solute concentration away from the end of the bar is zero


C x,t = 0
{ }
= 0
ii. as time increases, the total amount of solute is fixed, hence the area under the curve
remains constant


C x,t
{ }
!
dx = B

The solution, which satisfies both Ficks 2
nd
law and also the above boundary conditions, is
given by:












1
Crank J.C., The Mathematics of Diffusion, Oxford University Press, 2
nd
edition 1979.
time t
0

distance x
C

distance x
t
1
t
2
t
3
C
x

C x,t
{ }
=
B
! D t
exp "
x
2
4Dt
#
$
%
&
'
(
AH14 Course A: Phase Transformations AH14



rob.wallach@msm.cam.ac.uk 2013-14

1.3.3.2 Constant surface concentration, i.e. infinite source at end of a semi-infinite bar

Solute concentration at surface (C
s
) remains constant while initial concentration in the bar is C
0
.











The profile that develops is slightly less straightforward than previously.
As the source of solute at x = 0 is constantly replenished, the result is the equivalent of adding
together many exponential functions of different widths, i.e. integrating the exponential function.
The solution is known as the error function. This is a mathematical function (like sine or
cosine) and has the form

erf x
} {
=
2
!
exp
0
x
"
- u
2
} {
du
with properties as shown graphically and numerically:


at x = ! erf (x) = 1
at x = 0 erf (x) = 0
at x = + ! erf (x) = +1





After time, the profile in the above bar develops according to the error function












+1


erf 0

* 1
-& x = 0 &
distance x
C
0
t
1
t
2
S
C
C
C
C
s
C
0
distance x
Constant surface
concentration Cs
AH15 Course A: Phase Transformations AH15



rob.wallach@msm.cam.ac.uk 2013-14
The general solution to Ficks 2
nd
law is given by the equation:


C x,t
{ }
= A+ B erf
x
2 Dt
!
"
#
$
%
&

In order to solve Ficks 2
nd
law, the following boundary conditions need to be met:
i.

C x,t = 0
{ }
= C
0

ii.

C x = !,t
{ }
= C
0

iii.

C x = 0,t
{ }
= C
S

and so

C x,t
{ }
= C
S
! C
S
!C
0
( )
erf
x
2 Dt
"
#
$
%
&
'

This solution applies when both the initial surface and end concentrations, C
S
and C
o
, are kept
constant and the concentration inside changes with time. Examples include:
(i) diffusing from a fixed partial pressure of gas into a metal e.g. carburising steels see AQ1;
(ii) initial doping of semiconductors, i.e. prior to drive-in;
(iii) heat flow into a bar when one end is kept at a constant temperature (see Practical AP2).
The solution also is used when two semi-infinite bars are joined together. In this case, the
surface where the concentration remains constant is the interface between the two bars.















An application of this occurs during the fabrication of A15 type superconductors, e.g. Nb
3
Sn.
Niobium wires are embedded in bronze (Cu-Sn alloy), co-extruded and heated to encourage
interdiffusion and the formation of superconducting Nb
3
Sn intermetallic.

C
A
distance
x
C

at time t
o

t
0
C
B
distance
x
C

t
0
C
B

t
1
t
2

AH16 Course A: Phase Transformations AH16



rob.wallach@msm.cam.ac.uk 2013-14
1.4 Diffusion in Metallic Alloys
1.4.1 Interdiffusion
The net effect can be described using Ficks laws and the interdiffusion coefficient, D, is used. It
is the weighted average of the diffusion coefficients of the individual components D
A
and D
B
with
respect to an alloys composition defined in terms of the mole fractions of the two elements X
i
.
The relationship is called the Darken equation and is:
D =

X
B
D
A
+ X
A
D
B

In a system where there is complete solubility across the composition range (such as Cu-Ni
alloys), the diffusivities of the various species are as indicated below.








1.4.2 Temperature Dependence of Diffusion
If an atom in a potential well of height G vibrates with a frequency &
0
, the frequency & with which
it will make a successful jump into an adjacent position is given by:


! = !
0
exp "
G
k
B
T
#
$
%
&
'
(


= !
0
exp
S
k
B
"
#
$
%
&
'
exp (
H
k
B
T
"
#
$
%
&
'
given that G = H T S
It was shown in section 1.3 that
2
6
1
a D ! = , where a is the interplanar distance and, in practice,
is the smallest lattice vector, and so will be the Burgers vector, b, in many crystalline materials.

Hence

D =
1
6
!
0
b
2
exp
S
k
B
"
#
$
%
&
'
exp (
H
k
B
T
"
#
$
%
&
'
or

D = D
0
exp !
H
k
B
T
"
#
$
%
&
'


An Arrhenius graph, i.e. plot of log (D) versus 1/T, gives a straight line, see next page.
~
~
AH17 Course A: Phase Transformations AH17



rob.wallach@msm.cam.ac.uk 2013-14









Arrhenius plot of ln (D) versus 1/T for the self-diffusion coefficients of various metals
and including the diffusion coefficient for interstitial diffusion of carbon in ferritic iron.

The gradient of each line is H/k
B
. The enthalpy term, H, is the activation energy for diffusion
and is more generally denoted by Q.
For substitutional diffusion, this activation energy Q can be separated into two components:
enthalpy of migration (due to lattice distortions), Q
m

enthalpy of formation of vacancy in adjacent lattice site, Q
f

Hence D = D
o
exp(--
Q
m
kT
) exp(--
Q
f
kT
)

For interstitial diffusion, the energy of formation of a vacant site is not relevant (there always
are available sites for diffusing interstitial atoms), and so the second exponential for atom
migration is not needed, hence
Q
interstitial
<< Q
substitutional
and so D
interstitial
>> D
substitutional
.
Note that D has similar value (around 10
-12
m
2
s
-1
) for many metals at their melting points.
Materials dependence. For materials other than metals, D depends on the bonding in the
material. For instance, silicon has directional covalent bonding and hence there is a much
higher activation energy term for diffusion and consequently much lower values of D.
C in !-Fe
Al
Cu
Fe
Si
W
D
(
m
2
s
-
1
)
10
-10
10
-15
10
-20
10
-25
1000/T (K
-1
)
0 1 2 3
C in !-Fe
Al
Cu
Fe
Si
W
D
(
m
2
s
-
1
)
10
-10
10
-15
10
-20
10
-25
1000/T (K
-1
)
0 1 2 3
ln(D)
AH18 Course A: Phase Transformations AH18



rob.wallach@msm.cam.ac.uk 2013-14
1.5 Diffusion and Microstructure
If a lattice is locally disrupted, it will have a more open structure. The activation energy for
diffusion Q will be lower and so diffusion will be faster in such disrupted regions.
In metals, lattice disruption occurs in the vicinity of vacancies, dislocations and grain boundaries,
all of which provide easier diffusion paths.

1.5.1 Effect of grain boundaries (gbs)
A grain can be represented by a cylinder of diameter d with grain boundary width %.





Area of grain = Area of half grain boundary = as shared by 2 grains
Hence ratio of areas
Consider the total number of atoms per second that can diffuse through a grain (denoted now
by the term lattice) and its associated boundary. The overall flux J is given by:
sum of the flux through each ' area of each = J
lattice
A
lattice
+ J
gb
A
gb



Since A
lattice
>> A
gb
can assume A
lattice
( A

And so the flux (number of atoms per unit area per second) is
J = J
lattice
+ J
gb


2!
d
"
#
$
%
&
'

Using Ficks first law

J = !D
"C
"x
and assuming a constant concentration gradient

!C
!x


D
measured
= D
lattice
+ D
gb
2!
d
"
#
$
%
&
'

Grain boundaries have more open structures than the corresponding lattice, so Q
gb
< Q
lattice

These different Q values result in faster overall diffusion in polycrystalline samples than in single
crystals.

grain
grain

!
d
2
"
#
$
%
&
'
2

2!r
"
2
#
$
%
&
'
(

=
area of grain boundary
area of grain or lattice
=
A
gb
A
lattice
=
2!
d
simplified


representation
AH19 Course A: Phase Transformations AH19



rob.wallach@msm.cam.ac.uk 2013-14
In many polycrystalline alloys, the grain boundary area is just a small fraction of the total area.
Hence the contribution of grain boundary diffusion to the total diffusion is significant only at low
temperatures, when D
lattice
<< D
gb
. This can be seen on an Arrhenius plot.






The effect is very dependent on grain size d. For smaller grains, the lower dashed line moves
upwards so the significant contribution from grain boundary diffusion increases and is seen over
a greater temperature range.

1.5.2 Effect of dislocations
Dislocations, which are line defects, can also provide fast diffusion paths. Thinking of an edge
dislocation, there is a more open structure where the extra half-plane terminates.





By a similar analysis to that for grain boundaries (gb), it can be shown that


D
measured
= D
lattice
+ D
disl
A
disl
!
( )
(assume A
lattice
>> A
disl
, similar to gb analysis)

where A
disl
dislocation core area (typically ~1 nm
2
)
) dislocation density (typically 10
10
-10
16
m/m
3
)

Hence since A
disl
) ( 10
-3
- 10
-8
, the effect of dislocations (like grain boundaries) is significant
only at low temperatures, when the rate of diffusion within the lattice itself is very small.
This is relevant to creep deformation time dependent deformation at a stress < yield stress "
y

ln D
1/T
D
lattice
D
measured

D
gb
D
gb

AH20 Course A: Phase Transformations AH20



rob.wallach@msm.cam.ac.uk 2013-14
1.5.3 Effect of vacancies
The greater the number of vacancies in a sample, the faster substitutional diffusion will occur.
The equilibrium vacancy concentration is given by n = n
0
exp !
Q
f
k
B
T
"
#
$
%
&
'

Consequences of the above.
(i) An excess of vacancies can arise in a sample if it is quenched from a high temperature.
The equilibrium number of vacancies will be high at the high temperature and, if the sample
is cooled rapidly enough (quenched), there will not be time for the vacancies to diffuse to
sinks in order to reach the lower equilibrium number of vacancies. Hence, they become
frozen-in at the lower temperature, increasing overall diffusivity so that it is greater than
would be expected.
(ii) The number of vacancies also increases significantly during mechanical deformation, e.g.
when dislocation sessile jogs are dragged through a lattice.
(iii) Dislocations also act as vacancy sources or sinks, resulting in the climb of the
dislocations.




Vacancy source Vacancy sink
[The white arrows show the directions in which the vacancies diffuse.]
Note that the dislocation climb is relevant to creep deformation, i.e. the time-dependent
deformation at a stress less than the yield stress "
Y

(see Pt IA Course F and also practical FP2, creep of Pb-Sn solder).


AH21 Course A: Phase Transformations AH21



rob.wallach@msm.cam.ac.uk 2013-14
1.6 Thermodynamics of Diffusion
The ideas developed so far are insufficient for systems which favour the formation of multiple
phases. Consider a two-phase alloy of A and B atoms, which exists as the ! and " phases.




For two phases in equilibrium, there can be no driving force for either the A or B atoms to
diffuse from one phase to the other even though their compositions will be quite different..

1.6.1 Chemical potential


To explain the above, it is necessary to use the concept of the chemical potential, .
If a small amount dX
A
(where X
A
is in units of mole fraction) of species A is added to a system,
the change in the overall free energy of the system dG is:
dG =
A
dX
A

The chemical potential
A
of a species A is defined as:

A
=
!G
!X
A

and is the rate of change of free energy with composition.


For a single phase: G =
A
X
A
+
B
X
B

and dG =
A
dX
A
+
B
dX
B
(see Porter & Easterling p.16 for a proof)
Using the above two expressions, it can be shown that the chemical potential of each species in
the alloy (for some overall composition, X) can be found by drawing a tangent to the free energy
curve as shown on the next page.

!
"
!
"
!
"
!
"
!
"
!
"
AH22 Course A: Phase Transformations AH22



rob.wallach@msm.cam.ac.uk 2013-14
Single phase: continued









Two phases in equilibrium: requires the A atoms in the ! phase to have the same energy (or
chemical potential) as the A atoms in the + phase (and similarly for B atoms in the two phases).

This can be achieved only if the tangent to the two free energy curves is the same. Hence the
common tangent construction introduced in Part IA determines the equilibrium compositions
because it ensures that the chemical potential of each species is the same in both phases.








The consequence is that there is no driving force for an A atom to diffuse from the ! phase to
the + phase as the chemical potential
A
is the same in both phases


!
+
X
A B
G
!
+
Gibbs free energy
Composition X
A B
G
A B

A (X)

B (X)
X
i

AH23 Course A: Phase Transformations AH23



rob.wallach@msm.cam.ac.uk 2013-14
1.6.2 Diffusion down chemical potential gradients
1.6.2.1 Diffusion down a concentration gradient






Consider, a material in which initially there are two regions with different compositions P and Q.
Diffusion occurs down the concentration gradient, leading to a single uniform, homogeneous
phase of composition R.

1.6.2.2 Diffusion up a concentration gradient






In this system, a material which initially has a single region of composition S will unmix to form
discrete regions with compositions U and V. Hence diffusion occurs up the concentration
gradient. Note that this can be called spinodal decomposition or uphill diffusion.
Note that in this second case (1.6.2.2), diffusion is not down a concentration gradient, but
actually creates a concentration gradient by the formation of discrete regions (phases) of
different compositions. Equilibrium is reached when the chemical potential of a given element is
the same in each of the two different phases.
In each of the two examples, diffusion occurs down chemical potential gradients in order
to lower the overall free energy.
A B
G
X
P
Q
R
A B
G
X
S
U
V
AH24 Course A: Phase Transformations AH24



rob.wallach@msm.cam.ac.uk 2013-14
1.6.3 Diffusion in chemical potential gradient other than concentration
1.6.3.1 Diffusion in an elastically bent bar
If a bar is bent elastically, a strain gradient develops.
If the bar is made from a uniform solid solution of A and B atoms, where B atoms are larger than
A atoms, the B atoms will diffuse to where there is more space in the A-lattice (i.e. the tensile
region) in order to reduce the overall strain energy.








In this case, the chemical potential gradient, down which diffusion of the larger atoms occurs, is
a result of the variation in elastic strain across the bar thickness. The consequence is that a
concentration gradient forms as the strain decreases.

1.6.3.2 Electromigration (see DoITPoMS Teaching & Learning package)
Electromigration is the transport of material in a conductor under the influence of an applied
electric field, and can result in the failure of integrated circuits.
The applied electric field results in a metal ion experiencing an electrostatic force, the extent of
which depends on the charge on the ion core, modified by screening effects.
This results in a net flux of metal ions towards either the positive or negative terminals. Note
that the overall flux is extremely small but, over time, can cause failure of an electronic device.

AH25 Course A: Phase Transformations AH25



rob.wallach@msm.cam.ac.uk 2013-14
1.7 Diffusion in Polymers: non-Fickian behaviour
Diffusion in polymers can be complex due to possible interactions between the polymer matrix
and the diffusing species, as well as nature of the polymer structure itself. Polymers may be
amorphous (glassy) or have varying amounts of crystallinity.
Hence there may be departures from Ficks laws since, for cases considered previously, it has
been assumed that there is no interaction between the matrix and the diffusing species.
When a polymer is immersed in a liquid, the liquid may:
- not penetrate at all;
- dissolve the polymer;
- swell the polymer as it enters (a sharp interface forms at swollen and unreacted polymer).
In the last case, the diffusion coefficient in the swollen layer is significantly higher due to the
more open polymer structure there.
Hence liquid penetration is governed by the kinetics of liquid-polymer interaction at the interface.
One consequence can be linear migration of the sharp interface and an almost linear weight
gain with time, rather than parabolic (as has been seen in metals).
This property can be exploited for controlled drug delivery in medical applications.
When a drug is taken in the form of a tablet, the drug release can be:
- instantaneous, assuming the tablet dissolves almost immediately;
- time dependent, in which case a constant rate or sustained release is beneficial as the
same dose is available over a long period without being either excessive or too low.







Concentration of drug in the body as a function of time after administration in various ways.
AH26 Course A: Phase Transformations AH26



rob.wallach@msm.cam.ac.uk 2013-14
2. SOLIDIFICATION

2.1 Introduction
Solidification is a critical processing route for many materials e.g. metals and alloys, polymers.
Use of liquid phase # easy mixing of different elements and/or compounds (alloying)
# shaping to give final product directly by casting into mould i.e. container
which typically is sand, metal or ceramic.
Solidification and casting methods (of which there are many, see section 2.5) include:
continuous casting: around a billion tonnes of steel per annum and can form steel sheet
directly by including rolling mills as part of steel making plant;
sand casting: cheap and good for mass-producing items such as car engine
cylinder blocks in cast irons or aluminium-silicon alloys;
die casting: forcing of liquid metal into intricate moulds leads to high definition
products that can be used directly with no further machining;
single crystals: silicon single crystals grown for electronic devices, and also turbine
blades in Ni-base superalloys for aero engines since elimination of
grain boundaries enhances creep resistance;
welding: localised melting and solidification enables joining of materials;
zone refining: enables metal purification.

Solidification problems can lead to poor or variable properties, including:
poor initial microstructure: often coarse and non-uniform resulting in variable properties

partitioning of solute during solidification: coring and/or segregation see AH39

porosity: gases generally have higher solubility in liquids than in solids. Hence gases will
come out of solution during solidification and this causes porosity.


AH27 Course A: Phase Transformations AH27



rob.wallach@msm.cam.ac.uk 2013-14
2.1.1 Driving force for solidification




G = H !TS



!G = !H "T!S


At the melting point T
m

"G = 0 and so "H = T
m
"S
"S = "H / T
m


At any other lower temperature T i.e. T < T
m
then "G # 0 and let "S = "H / T
m

"G = "H -T ("H / T
m
)
=

!H (T
m
"T)
T
m

=

!H
T
m
"T OR = "S "T where )T = undercooling and
)H $ latent heat
i.e. the greater the undercooling, )T, the greater the driving force, )G, for solidification.


2.1.2 Review of nucleation and growth
On cooling, solidification does not happen instantaneously upon cooling below T
m
, but takes
place via:
nucleation homogeneous: rare and only if very large )T
heterogeneous: on mould walls, impurities etc.

growth which may be affected by temperature gradients and preferred crystal
growth directions
solid
liquid
T
m
Gibbs
energy G
temperature T
AH28 Course A: Phase Transformations AH28



rob.wallach@msm.cam.ac.uk 2013-14






Atoms will be able to move in either direction across the energy barrier.
Rate of transfer L # S R
LS
* exp !
Q
kT
"
#
$
%
&
'


Rate of transfer S # L R
SL
* exp !
Q+ "T
( )
kT
#
$
%
&
'
(

Hence the growth velocity, v * exp !
Q
kT
"
#
$
%
&
'
1! exp !
"G
kT
#
$
%
&
'
(
)
*
+
,
-
.


If )T is small, then )G is small and as exp(x) ( 1 + x for small x, so above equation becomes
v * exp !
Q
kT
"
#
$
%
&
'

(G
kT
"
#
$
%
&
'

hence v * )G and since )G is proportional to )T
then v * )T
In practice, heterogeneous nucleation occurs with undercoolings )T of a few milli Kelvin.
Growth rates of 50 m s
-1
are usually regarded as very fast but still are very slow compared to
solid-state martensitic transformations in which no diffusion occurs and which advance at much
greater speeds e.g. ~ 6000 m s
-1
in steels (the speed of sound in air is ~ 340 m s
-1
).
Growth rates - different result in different microstructures in the solid that forms, and
- faster rates result in finer grains (higher strength and simultaneously toughness)

G
Q
)G
liquid solid
AH29 Course A: Phase Transformations AH29



rob.wallach@msm.cam.ac.uk 2013-14
2.2 Solidification Microstructures of Pure Metals

2.2.1 Grain structures in castings
Consider the solidification sequence when molten metal is poured into a cold mould.






Chill zone:
fine grains heterogeneously nucleate on the mould surface, with random orientations;
many nuclei form due to large initial undercooling on cold mould walls;
some swept into liquid and survive to assist formation of equiaxed zone see below.

Columnar zone:
growth of some chill crystals with optimal crystallographic growth direction, <100> in
many metallic alloys, and also influenced by direction of maximum temperature gradient;
may form either dendrites or cellular interface, depending on the casting conditions.

Equiaxed zone:
formed by chill crystals, detached dendrite arms swept by convection to centre of casting
or formed by deliberate addition to molten metal of grain refiners/inoculants;
desirable because a small grain size
(i) reduces the extent of microsegregation and
(ii) improves mechanical properties (both strength and toughness) of final product.
chill



columnar


equiaxed


AH30 Course A: Phase Transformations AH30



rob.wallach@msm.cam.ac.uk 2013-14
2.2.2 Pure metals: thermal dendrites
Consider a molten metal solidifying in a mould, and assume unidirectional solidification.
Dendrites can grow from small perturbations on the growing solid if the conditions in liquid
ahead of the interface are more favourable for solidification.








Hence two possible scenarios:







Initially and close to mould wall: Nearer centre of mould:
dT
dx
> 0 so planar growth occurs
dT
dx
< 0 so non-planar growth occurs
as any perturbation will melt back. as perturbations will be stable and grow.
Results in planar or cellular interface Results in non-planar interface and
and columnar grains. dendrites.

T
distance
liquid
0 >
dx
dT
T
distance
liquid
0 <
dx
dT
solid solid
0 <
dx
dT
x
T
m m
AH31 Course A: Phase Transformations AH31



rob.wallach@msm.cam.ac.uk 2013-14
2.3 Solidification Microstructures of Metallic Alloys

2.3.1 The partition coefficient, k
When a solid and liquid are in equilibrium in an alloy system, the amount of solute (or impurity)
atoms in each phase will not generally be the same. Consider the part of a phase diagram
shown below (which, for instance, might represent one end of a eutectic system).







In the above figure, the nomenclature C
!"
represents the composition of phase ! when it is in
equilibrium with phase ", i.e. C
SL
means solid of that composition in equilibrium with liquid of
composition C
LS
at the given temperature. The starting composition of the liquid is C
0
.
In the example above, as the alloy cools in equilibrium, there is always considerably more
solute in the liquid than there is in the solid. The partition coefficient, k, is defined as
k =
C
SL
C
LS
=
C
solid
C
liquid


The partition coefficient is constant for a range of temperatures if the solidus and liquidus are
assumed to be straight lines. Whilst these lines on many phase diagrams generally are not
strictly straight, this approximation is often reasonable.
Note that the above phase diagram could have been drawn with the two elements (A and B)
reversed, i.e. as
Hence, for the same system k =
C
solid
C
liquid
> 1


temperature T
C
0 C
SL
C
LS
S L S + L
composition (%B)
T
composition (%B)
100% A
AH32 Course A: Phase Transformations AH32



rob.wallach@msm.cam.ac.uk 2013-14
2.3.2 Equilibrium solidification (assuming one dimensional heat flow)
Let a solidifying bar cool sufficiently slowly that there is plenty of time for diffusion in both the
liquid and solid. The concentrations of each phase will be those given by the phase diagram.
time t
1
time t
2
> t
1




The final solid bar would be the same uniform composition, C
0
, as the initial liquid if there was
sufficient time for diffusion of solute in the solid bar.
In practice, cannot occur as solute diffusion in a solid is much too slow, by orders of magnitude.

2.3.3 Equilibrium only at the interface: diffusion only in liquid and none in the solid
Assume no turbulence (mixing) in liquid. Again, diffusion in solid phase relative to that in liquid
is incredibly slow and generally can be neglected completely. Complete equilibrium is unlikely.
Consider the situation when the profile is determined solely by diffusion in the liquid. When the
solid begins to form, solute is rejected into the liquid. Within the liquid, this solute is distributed
into the liquid only by diffusion; this results in a build-up of this solute ahead of the interface.







The first solid to form from liquid originally with composition C
0
has a composition given by kC
0

(where k is the partition coefficient).
C
0
S
L
S + L
C
0
/k kC
0
T
L
composition
C
distance
liquid
C
0
C
LS
C
SL
C
distance
liquid
C
LS
solid solid
C
0
C
SL
T*
AH33 Course A: Phase Transformations AH33



rob.wallach@msm.cam.ac.uk 2013-14
As the solute builds up ahead of the interface, the compositions of both liquid and solid at the
interface increase. As equilibrium is maintained at the interface, the ratio of compositions will be
maintained at k. A steady state is reached when the composition of the solid at the interface
reaches the original (or average) composition C
0
.
initial build-up steady state




When the steady state is reached the solute profile does not change. No further build-up occurs
and the solid-liquid interface and solute profile simply move forward at a constant velocity.
The rate at which the interface advances is balanced by the rate at which solute can diffuse
away into the liquid. Using Ficks 1
st
and 2
nd
laws of diffusion, the composition at a distance x
ahead of the interface is given by an equation of the form

C
x
= C
0
+
C
0
(1! k)
k
exp !
x
D
v
"
#
$
$
%
&
'
'

In this equation, D/v is a measure of the width of the liquid zone that is enriched by solute.

The diffusivity of a liquid is typically around 10
9
m
2
s
-1
and so:
- at slow solidification rates (v ( 10 m s
-1
) D/v ( 0.1 mm
- at fast solidification rates (v ( 1 mm s
-1
) D/v ( 1 m
In the final stages of solidification, the solute profile in the liquid alters since there is insufficient
liquid remaining into which solute can diffuse. Hence the value of C
o
/k increases with a
corresponding increase in the solute concentration of the final solid that solidifies. The final
profile of the solidified bar is as shown below.
C
distance
liquid
C
0
distance
liquid solid
kC
0
C
0
/k
solid
C
C
0
kC
0
C
C
0
distance
AH34 Course A: Phase Transformations AH34



rob.wallach@msm.cam.ac.uk 2013-14
2.3.4 Equilibrium only at the interface: perfect mixing in the liquid and no mixing in solid
Now assume that solute that is ejected from the growing solid does not build up at the interface
but instead is mixed uniformly in the liquid, e.g. by convection and turbulence in the liquid.






The crucial aspect here is that the solute profile in the liquid is always uniform due to the mixing
and so solid progressively solidifies from higher solute enriched liquid.

To determine the profile, consider what happens as the interface advances a small amount, df.






Equating the shaded areas to conserve solute gives:


C
L
! kC
L ( )
df = 1! f
( )
dC
L
and separating variables


df
1! f
"
#
$
%
&
'
0
f
s
(
=
dC
L
C
L
1! k
( )
"
#
$
%
&
'
C
0
C
L
(

Hence

C
L
= C
0
1! f
s
( )
k !1 ( )
or

C
S
= kC
0
1! f
s
( )
k !1 ( )

This relationship is known as the Scheil Equation.

The equation suggests that as f
s
tends to 1, the composition tends to infinity. But in practice,
there is a limit on the amount of solute ejected into the liquid. For example, the limit would be
reached in a simple binary eutectic when the composition of remaining liquid attains the eutectic
composition since any remaining liquid will then solidifies to form eutectic solid. See next page.
C
distance
liquid
C
0
C
distance
liquid
C
0
kC
0 kC
0
solid solid
fraction solidified f
kC
0
0 1
df
C
kC
L
C
L
dC
L
AH35 Course A: Phase Transformations AH35



rob.wallach@msm.cam.ac.uk 2013-14
2.3.5 Summary: solute profiles within solidified bars, assuming I-D heat extraction

Case A: equilibrium
(see section 2.3.2)




Case B: diffusion control
(see section 2.3.3)






Case C: perfect mixing in the liquid
(see section 2.3.4)




Zone refining and levelling

It is possible to purify a material by producing the profile shown in case C, where the solute is
swept to the right hand end of the bar. If the last part of the bar to freeze is cut off, then the
overall impurity concentration in the remaining material is lowered. The whole process can be
repeated as many times as is wanted to significantly purify the bar.

This principle is called zone refining, i.e. a hot small (liquid) zone is passed along a bar,
causing the impurities to concentrate in the small liquid zone which is carried to the end of the
bar. The sequence then is repeated to enable purification of the bar to the level required.

In zone levelling, the hot small (liquid) zone is passed backwards and forwards along a bar to
provide a uniform solute concentration and so eliminates microsegregation.
C
f
C
0
kC
0

C
f
C
0

kC
0

C
C
0
f
AH36 Course A: Phase Transformations AH36



rob.wallach@msm.cam.ac.uk 2013-14
2.3.6 Constitutional undercooling
Return now to the case of limited mixing in the liquid (AH32, section 2.3.3), i.e. where the solute
ejected into the liquid is carried away only by diffusion (no convection leading to complete
mixing in the liquid) and thus will build up near to the interface over some distance given
approximately by D/v.











As the solute diffuses away gradually, the solute concentration in the liquid is non-uniform, i.e. is
high at the interface and decreases to the value C
0
, as shown by the curve A-B above. Hence
the liquidus temperature (the temperature at which can solidification begin) in the liquid will also
be non-uniform (lower at the interface, increasing to some constant value, curve P-Q).
There also is a real temperature gradient within the liquid, as shown by the dashed line in the
lower figure above. This gradient depends on the actual heat extraction through the mould
around the casting, and hence is affected by the casting method and casting conditions.
If the actual temperature locally is below the equilibrium freezing temperature, then that liquid is
described as being undercooled (note that this sometimes is referred to as supercooled).
In pure metals (when the liquidus temperature is constant), a negative temperature gradient will
also cause undercooling, as was shown in section 2.2.2. In the above case, however,
undercooling is achieved near the interface even when there is a positive temperature gradient
in the liquid. This is illustrated on the next page.
T
composition
C
SL
= C
0
S L S + L
T
2
T
1
C
LS


A


B


Q



P
C
distance
T
L
T
1
liquid
T
2
distance
solid
C
LS


C
SL
(=C
0
)

P
AH37 Course A: Phase Transformations AH37



rob.wallach@msm.cam.ac.uk 2013-14






This phenomenon is known as constitutional undercooling as it is undercooling caused by
the effect of composition (constitution). Perturbations on a planar interface next to undercooled
liquid are stable and grow faster into the liquid and can develop into cells or dendrites.
Interface morphology
From the above figure, dendrite formation will be avoided (and a planar growth front maintained)
only if the actual temperature gradient in the liquid is high enough that there is no undercooled
region. Conversely, if the temperature gradient is lower than this critical value, a stable planar
interface will break down into cells or dendrites.
Undercooling
Undercooling occurs if at the interface (x=0)

dT
L
dx
x=0
>
dT
dx

Let the magnitude of the slope of the liquidus line on the phase diagram be denoted as m and
note that m will be equal to dT
L
/dC in the region of the phase diagram being considered.
Undercooling occurs if

!m
dC
dx
x=0
>
dT
dx
or

dT
dx
< ! m
dC
dx
x=0


From page AH32, the rate at which an interface advances v is balanced by the rate at which
solute can diffuse into the liquid, i.e.
flux of solute away from interface = (C
LS
C
SL
) v
using Ficks first law,

C
LS
!C
SL
( )
v = !D
dC
dx
x=0

Hence undercooling occurs if

dT
dx
<
m C
LS
!C
SL
( )
v
D



dT
dx
<
m C
0
/ k !C
0
( )
v
D



dT
dx
<
mC
0
(1! k)v
kD

T
x=0 distance, x
T
L
Dotted line shows actual temperature of liquid
undercooled
region
P
AH38 Course A: Phase Transformations AH38



rob.wallach@msm.cam.ac.uk 2013-14
To avoid undercooling and maintain a planar growth front during solidification, the temperature
gradient must be greater than the value determined using the equation on the previous page.

In practice, this is difficult to achieve, as is indicated below. Assume:


dT
dx
= 10 K mm
-1
D
L
= 5 x 10
9
m
2
s
-1

k = 0.2
C
0
= 0.2 wt.%
m = 10 K wt.%
-1

Using the above values, the maximum velocity v
max
for a planar interface equals 6 x 10
-6
m s
-1

or 2 cm per hour. This clearly is much slower than would be experienced in practice.
Hence undercooling generally occurs during the solidification of metallic alloys and so dendritic
growth is experienced rather than a planar interface and columnar growth. This can be seen in
the resulting solidification microstructures.
Note that exceptions are important, e.g. turbine blades (AH48)
silicon single crystals (AH48)




AH39 Course A: Phase Transformations AH39



rob.wallach@msm.cam.ac.uk 2013-14
2.3.7 Segregation in castings
a. Macrosegregation
If a planar solidification front moves along a bar, solute concentrated in the liquid will be swept
to one end of the bar (this is known as macrosegregation - solute is segregated on a
macroscopic scale). It also is important to consider what happens to this solute when dendritic
growth occurs.




b. Microsegregation
Solute is rejected at all solidification fronts and will be trapped between growing dendrite arms.
This leads to composition variations in the final solid on the scale of the dendrite arm spacing.
This is called microsegregation, and is shown below for a Cu-Ni alloy. The light areas which
have solidified first are Ni-rich, and Cu-rich liquid has been trapped between the dendrite arms
before solidifying. Note that there is complete solid solubility across the Cu-Ni phase diagram
and no discrete second phases can form; the below contrast arises from the etching procedure.



Cu 70, Ni 30 (wt%), cored dendrites (etched in NH
4
OH, H
2
O
2
)

Microsegregation in solidified alloys can be a problem as variations in composition can lead to
variable properties (mechanical, electrical, chemical).
Non-equilibrium phases may also form similar to cases B and C on page AH35 when local
solute enrichment in the liquid can result in eutectic formation in the final stages of solidification.
Homogenisation, i.e. annealing at an elevated temperature for sufficient time to allow diffusion,
will reduce segregation although it generally is difficult or uneconomic to achieve complete
homogenisation. Mechanical working results in smaller grains and more defects, both of which
enhance homogenisation in subsequent heat-treatments.
mould

heat extraction
AH40 Course A: Phase Transformations AH40



rob.wallach@msm.cam.ac.uk 2013-14
2.4 Heat Flow in Solidification
2.4.1 Heat extraction from a casting
The rate of heat extraction during casting is a critical factor that strongly affects the
microstructure of the final solid. The contributions to the heat that must be removed are:
superheat )T
s
: due to the need to pour liquid metal at a temperature above its melting
point;
latent heat )H
s
: heat released as the metal solidifies;
heat capacity of the system as it cools.

Ignoring the superheat, the equation governing the heat flow has the form


dT
dt
= !
d
2
T
dx
2
+
Q
c
V
where ! is the thermal diffusivity (equal to K/c
V
)
K is the thermal conductivity (W m
-1
K
-1
)
c
V
is the volumetric heat capacity (J m
-3
K
-1
)
Q is the rate of latent heat generation (J m
-3
s
-1
)
Analytical solutions to this are generally not possible except in some very simple cases.
Heat transfer across the interface between the mould and the casting will affect significantly the
rate of cooling.
Heat transfer at the mould/casting interface can occur by:
direct contact (initially)
conduction through the air gap
convective air currents
radiation



air
gap
mould mould
casting
solid
AH41 Course A: Phase Transformations AH41



rob.wallach@msm.cam.ac.uk 2013-14
2.4.2 The Biot number
Heat is extracted from a casting in several stages, one of which may be rate-determining:
a. thermal conduction through the mould;
b. interfacial heat transfer at the mould-casting interface;
c. thermal conduction through the casting itself.








a. Thermal conduction through the mould can be rate-determining (e.g. very slow in sand
casting) but is not considered further in this analysis.
b. The heat flux across the interface between the mould and casting is
q
int
= h )T
int

where h is the heat transfer coefficient, describes heat conduction across interface/air gap.

c. The heat flux (in W m
-2
) within the casting by thermal conduction is given by


q
cast
=
K!T
cast
L

where K is the thermal conductivity and )T
cast
/L is the thermal gradient in the casting (in
the heat flow direction). Hence heat conduction in the casting is characterised by K/L.

The Biot number is the ratio of the heat fluxes (b and c) at the interface. It indicates whether
heat extraction is dominated by conduction through the casting or across the interface.


B
i
=
h
K / L
=
hL
K


mould
casting
heat flow
L
liquid
T
1
distance
2
AH42 Course A: Phase Transformations AH42



rob.wallach@msm.cam.ac.uk 2013-14
2.4.3 Newtonian cooling
Newtonian cooling occurs when the limiting factor is the interfacial heat transfer, i.e. the transfer
of heat across the interface is much slower than heat conduction through the casting and hence
the Biot number is very small (<<1). This situation is most likely if the metal has a very high
thermal conductivity (Al, Cu) or the casting is very small (thin).
It leads to the temperature being approximately uniform within the casting itself and the
temperature drop takes place at the interface between the casting and the mould.







When Newtonian cooling occurs, the analysis of heat flow is relatively simple.
Assuming unit cross-sectional area, the rate of heat removal from the mould is balanced with
the rate at which heat flows across the interface:
q
int
= h #T
int

The heat to be removed = #T c
v
per unit volume
Hence, rate of heat extraction, or heat flux (heat removed per unit area per second) out of the
cooling liquid is:


which is balanced by the heat flow across the casting/mould interface:


q =
dT
dt
c
V
L = h!T
int

or
dT
dt
=
h!T
int
Lc
v

Obviously, in this case, )T will change (decrease) as the liquid cools and so the rate of cooling
of the liquid actually slows down.
Note that L often is taken to be half of the width of the casting since a mould encloses a casting
resulting in two solidification fronts.


T
distance
liquid
L
metal mould

q =
dT
dt
c
V
L
AH43 Course A: Phase Transformations AH43



rob.wallach@msm.cam.ac.uk 2013-14
A liquid cooled to its solidification temperature cannot solidify instantly as the latent heat )H
F

which is produced during solidification has to be removed. If the solidification front moves
forward at velocity v, the rate at which heat is generated per unit area of casting is given by:


q = v!H
F

where )H
F
is the latent heat per unit volume (or enthalpy of fusion).
Balancing this with the heat flow across the casting/mould interface:


q = v!H
F
= h!T


v =
h!T
!H
F

2.4.4 Superheat
For all castings, in order for molten metal to completely fill a mould, its temperature initially has
to be significantly higher than its melting point - this is known as superheat. Thus some time
elapses while the liquid cools down to its melting point and solidification can begin.
The effect of this superheat on the above equation is that more heat needs to be removed, and
so )H
F
is replaced by the expression
()H
F
+ c
v
"T
s
)
where "T
s
is the temperature difference between the liquid metal at the time of pouring and its
solidification temperature, and c
v
is the specific heat of the liquid metal (taken generally to be
the same value as for the solid).
2.4.5 Finite element modelling of solidification
The above equations are for relatively simple ideal cases. Also, modern castings may have
very complex and non-symmetrical shapes, and consequently analyses of heat and fluid flow
would be very difficult.
In practice, finite element modelling now is used to provide accurate simulations of
solidification to optimise the conditions to be used in practice to ensure optimal microstructures.
AH44 Course A: Phase Transformations AH44



rob.wallach@msm.cam.ac.uk 2013-14
2.5 Solidification Processing (Casting Methods)
[Processes described in Cambridge Engineering Selector (CES) see page AH104, 5c]
Criteria for selecting a casting method include:
nature of object to be made and properties required in service;
metal to be cast (especially melting point);
size and shape of casting;
dimensional accuracy and surface finish: determine extent of subsequent finishing steps;
number of castings to be made;
economics.
AH45 Course A: Phase Transformations AH45



rob.wallach@msm.cam.ac.uk 2013-14
2.5.1 Sand casting




Sand is packed around a wooden pattern in 2 half-moulds.
Pattern is removed and the two half-moulds are placed together.
Require








Alternatively, a polystyrene pattern may be used and the sand packed around it using a full
mould. When the molten metal is poured, the polystyrene pattern is burned-out. This allows far
more complex shapes to be cast - re-entrant shapes are possible as the pattern does not have
to be removed prior to pouring.
Problems associated with sand-casting include:
poor surface finish;
inability to produce thin sections;
often significant porosity due to water which is present to bind sand.


runner to
pour in the
molten metal
risers (often more
than one) to ensure
liquid fills mould and
air can escape
AH46 Course A: Phase Transformations AH46



rob.wallach@msm.cam.ac.uk 2013-14
2.5.2 Die- casting
In die-casting, the mould is permanent and reusable. There are two common forms:






Gravity Pressure
Advantages of die-casting include:
fast cycle time;
high definition and good surface finish arising from metal mould;
fast solidification (due to metal mould) results in fine grain size;
thin sections are possible.
Disadvantages of die-casting include:
moulds are expensive as they must be precision-machined so use for mass production;
it is difficult for metals with high melting points mainly used with Al and Zn alloys;
complex shapes cannot be ejected from the mould.


2.5.3 Centrifugal casting
Castings produced by allowing molten metal to solidify in rotating moulds. The speed of the
rotation and metal pouring rate vary with the alloy and also size and shape being cast. Can
make long stainless steel pipes that are used in chemical and oil refining plants.






AH47 Course A: Phase Transformations AH47



rob.wallach@msm.cam.ac.uk 2013-14
2.5.4 Continuous casting
This is the method used for large-scale steel production. The metal is cast and rolled
continuously in one process. The mould may oscillate in order to improve surface quality.












2.5.5 Investment casting
A wax pattern is made with precise dimensions.
Slurry is used to make a ceramic mould around the wax pattern.
The mould is heated to remove the wax and a hollow ceramic mould remains.








(i) High accuracy allows usage for dental crowns (using gold).
(ii) Use of a ceramic mould allows metals with high melting temperatures to be cast, such as
Ni-based superalloys for aircraft turbine blades. Results in very high-definition castings
their shape is very close to final requirement so can use electrochemical machining to
finish shaping (advantageous due to very hard nature of alloy).
Blades with internal cooling channels are produced by including silica (or equivalent) fibres in
the original wax pattern in such a way that their position is fixed even when the wax melts. After
casting, the fibres be chemically etched out of the turbine blade to leave air-cooling channels.
Alternatively, laser drilling may be used.
ladle





tundish to ensure constant head (pressure) of liquid metal

water cooled mould, lubricated to prevent sticking



rollers prior to flame cutting into billets which
then are rolled, extruded to final
products



AH48 Course A: Phase Transformations AH48



rob.wallach@msm.cam.ac.uk 2013-14
2.5.6 Melt spinning
Cooling rates (Newtonian) of up to 10
7
K s
-1
can be achieved in thin ribbons by casting onto a
water-cooled, rotating Cu wheel. Heat extraction is very rapid and metallic glasses can be
produced, i.e. no time for crystallisation. Ribbon can be produced at a rate of up to 40 m s
-1
.







Process also used to make powders used to fabricate ODS alloys (see AH76)



2.5.7 Single crystal growth
2.5.7.1 Turbine blades: using investment casting
Achieved by adding a pig tail to the bottom of the cast
turbine blade mould.
Chill grains initially nucleate in the rectangular section below the
pig tail and only a few survive to grow as columnar grains, as a
consequence of their preferred orientation (Section 2.1.1) and
the imposed temperature gradient.
Competitive growth through the pig tail ensures just one grain
enters the main part of the mould. In this way, a single crystal
turbine blade is fabricated.

2.5.7.2 Silicon single crystals: Czochralski process
Silicon single crystals underpin electronic devices, as will be covered in IB Course E, and are
fabricated by pulling a single crystal seed from very pure silicon melt at slow speed.
In this way, nucleation is avoided and very large defect-free (including very low dislocation
densities) single crystals can be fabricated: up to 400 mm in diameter and up to 2 m in length.
droplets/jet of
molten metal
ribbons:
~ 50 !m thick
~ 300 mm widths possible

rotating copper wheel
(water cooled)
AH49 Course A: Phase Transformations AH49



rob.wallach@msm.cam.ac.uk 2013-14
2.6 The Peritectic Transformation
Similar to the eutectic and eutectoid reactions, the peritectic happens at a well-defined
temperature and composition. It is the formation of a single solid phase (") by interdiffusion
between an original different solid phase (!) plus a liquid phase (L), i.e.
! + L # "









An important aspect of solidification in a peritectic system in real systems is that the new solid
phase " nucleates and grows heterogeneously on the original solid ! phase.
Given the relatively fast rate of cooling of a melt and the time that would be needed for diffusion
in the solid state to attain full equilibrium, not all the original ! phase transforms to ". As full
equilibrium is not reached, a peritectic sample generally shows some residual original ! which is
surrounded by ". This is shown clearly above for a Sn - 21 wt.%Cu alloy.


The peritectic reaction is utilised commercially to grain refine aluminium cast products. For
example, small amounts (0.01 0.1 wt.%) of titanium Ti are used as an alloying addition:
- TiAl
3
first nucleates at ~1300C, well above the melting point of Al alloys (~600 - 660C);
- these small TiAl
3
nuclei provide heterogeneous nucleation sites for subsequent Al
solidification;
- result is a more extensive fine grain, equiaxed structure in the final casting with improved
properties (a smaller grain size results in improved toughness and also strength).

!
"
! + "
! + L
"
L
T
A B
!
! + "
! + L?
L
T
P
T
" + L
"
.
Sn -21 wt.%Cu
Metallographic sample M5
AH50 Course A: Phase Transformations AH50


rob.wallach@msm.cam.ac.uk 2013-14
3. MICROSTRUCTURE AND PROPERTIES
3.1 Dislocations
A dislocation is a line defect in a crystalline solid. Typical dislocation densities are:
in annealed metals is ~ 10
10
m m
-3
or lines m
-2

in heavily-deformed metals is ~ 10
16
lines m
-2

Associated with each dislocation is a local region of high elastic strain energy due to the disruption
of the regular crystalline packing. The energy per unit dislocation line length is ~ !Gb
2
J m
-1
.
Impeding dislocation movement results in strengthening. Various strengthening mechanisms in
metallic alloys were covered in Part IA.



3.1.1 Review of dislocation types
Edge: Burgers vector ! is perpendicular to line vector "
Screw: Burgers vector ! is parallel to line vector "
Mixed: In practice, virtually all dislocations lines in materials are mixed, i.e. can be
described as having a combination of edge and screw components.










From http://courses.eas.ualberta.ca/eas421/lecturepages/microstructures.html
AH51 Course A: Phase Transformations AH51


rob.wallach@msm.cam.ac.uk 2013-14
3.1.2 Partial dislocations
A single dislocation in an fcc alloy often will dissociate into two partial dislocations providing the
overall energy of the two partials is lower than that of the original perfect dislocation:




The first partial moves some atoms in B sites into C sites to create a local packing change called a
stacking fault. The second partial restores the correct ABC packing in the alloy.
A similar change in local packing also may occur during the recrystallisation of f.c.c. alloys and can
result in annealing twins (see AH59). In both cases, the separation of the two partials will depend
on the stacking fault energy of the alloy (as discussed in annealing twins AH59).
3.1.3 Order strengthening and superdislocations
Some metallic alloys can have ordered structures, e.g Fe-Co and Cu
3
Au
Consider an edge dislocation moving through a simulated 2-D ordered alloy under a shear stress.






Before passage of 1 dislocation After passage of 1 dislocation After passage of 2 dislocations
Movement of a dislocation through the lattice causes similar atoms to become neighbours and
results in the formation of an anti-phase boundary APB (see AH 53) with a consequent increase
in energy of the lattice. Hence dislocation motion is more difficult than in a disordered structure.
To minimise this energy increase associated with the APB, two dislocations generally travel
together, one after the other. This pair of dislocations, called a super-dislocation, maintains the
lattice in its preferred ordered (and lower energy) state. As a higher stress is required to move the
dislocation pair (super-dislocation), the ordered alloy is stronger than if the alloy was disordered.
AH52 Course A: Phase Transformations AH52


rob.wallach@msm.cam.ac.uk 2013-14
Crystallography of partial dislocations in fcc lattice















Consider the (111) plane:
















Dislocation with Burgers vector b
0
is equivalent to PQ = a/2 [ 10 ]


Dislocation with Burgers vector b
p1
equals ! PV = " PT = a/3 (PU + UT)
= a/3 ( [01 ] + # [ 01] )
= a/6 ( [0 2 ] + [ 01] )
= a/6 [ 2 ]


Dislocation with Burgers vector b
p2
can be obtained using PS + SR = PR
i.e. b
p2
is equivalent to " SR = a/3 (PR PS)
= a/3 ( [ 1 0] # [01 ] )
= a/6 [ 2 ]


Now can check: b
0
= b
p1
+ b
p2
using vector addition, and also that
all direction lie in (111) plane using Weiss zone law.

R
Q
P

T
S



U
z







y


x

AH53 Course A: Phase Transformations AH53


rob.wallach@msm.cam.ac.uk 2013-14
Examples of ordered lattices.
Consider the passage of dislocations in two types of unit cell which can be disordered or ordered.
(i) A-B unit cell which can be disordered (bcc or I) or ordered (P). An example is Fe-Co.






(ii) A
3
B unit cell which is disordered (fcc or F) or ordered (P) e.g. Cu
3
Au or Ni
3
Al (see AH96).




In both disordered unit cells, the Burgers vector of a dislocation is associated with the shortest
lattice vector. In each of the ordered unit cells, the overall Burgers vector of the super-dislocation,
the pair of dislocations that ensure the lattice stays ordered, is no longer associated with the
shortest lattice vector. This is due to the change in lattice type as a consequence of ordering.
3.1.4 Anti-phase boundary APB
A cubic close packed (fcc) metal slips on {111} planes and the packing of these {111} planes can
be described as A, B and C as introduced in Part IA.
Consider a disordered fcc A
3
B alloy which orders on cooling to form a P (primitive) lattice.
Two {111} planes of a perfectly ordered A
3
B lattice are shown below. Let atoms in the bottom plane
be described as being in A positions and those in the top plane as occupying B positions.




! !
AH54 Course A: Phase Transformations AH54


rob.wallach@msm.cam.ac.uk 2013-14
If an edge dislocation is introduced into the structure, it effectively introduces a missing half-plane
of atoms, perpendicular to the plane of the paper, in the centre of the structure. Atoms in the top
plane remain in B positions (rather than C sites), even those that have undergone slip (which are
shown below on the RHS). However because this is an ordered alloy, those that have slipped are
no longer in the perfectly ordered positions but now have similar neighbours.



A second edge dislocation is needed to restore the structure back to its lowest energy ordered
configuration. Hence the structure, due to the passage of a super-dislocation can be described by:



In the ordered A
3
B alloy, each dislocation in the superdislocation can itself dissociate into partials:




Hence, an ordered alloy, containing a pair of dislocations a superdislocation is described by:




Therefore, when slip occurs on the slip plane by dislocation motion in an ordered A
3
B alloy:
- anti-phase domain boundaries must move and anti-phase domains are created and destroyed;
- domain boundaries interact with other microstructural features, on which they can become pinned.

To overcome this requires additional work (analogous to the motion of ferroelectric domains as in
Part IA Course B). This leads to enhanced mechanical strength which is order strengthening.
This is exploited to optimise the strength of nickel-base superalloys used in gas turbine engines.
SF SF APD SF SF APD
AH55 Course A: Phase Transformations AH55


rob.wallach@msm.cam.ac.uk 2013-14
3.2 Grains and Grain Boundaries
Grain size and grain boundaries are important microstructural features which affect properties.
Mechanical properties
Small grains generally strengthen an alloy, as described by the Hall-Petch relationship:

!
Y
= !
o
+
k
d
1/2

and also simultaneously improve toughness, but have an adverse effect on creep behaviour.
Physical properties
Large grains are advantageous for: soft magnets since domain wall movement is easier;
higher electrical conductivity
Small grains are advantageous for: hard magnets

3.2.1 The nature of grain boundaries
The description of grain boundaries on an atomic scale is generally complex, but the simple case of
a low-angle symmetrical tilt boundary can be represented as shown below.







The energy (J m
-1
) of a dislocation is !Gb
2
per unit length, where G is the shear modulus.
The energy per unit area of the grain boundary (!) is given by the energy per unit length of each of
the edge dislocations multiplied by the number of dislocations per metre, namely 1/d, and hence



! =
1
2
Gb
2
d




=
1
2
G b !
i.e. the energy associated with the grain boundary increases linearly with the misorientation
angle, ! , due to the fact that the spacing between dislocations decreases.
d = b/!
b
AH56 Course A: Phase Transformations AH56


rob.wallach@msm.cam.ac.uk 2013-14
3.2.2 Grain boundary energy and the coincidence site lattice
Initially the grain boundary energy increases approximately linearly with misorientation angle !, but
this then saturates at around 15. At this angle, the edge dislocations become so close together
that the strain fields surrounding them begin to overlap and the simple analysis of adding together
the energies is no longer valid. This is illustrated by the dotted line in the figure below; the energy
of a grain boundary plateaus at misorientations higher than ~15.






Adapted from Miura H. et al, Colloque de Phys., C151, (1990), page 293.
However, it is found experimentally that there are several misorientation angles for which the
energy decreases sharply, and this is shown by the solid line in the above figure. This means that
there are some high-angle grain boundaries which have reduced energies.
This can be explained by a coincidence site lattice (CSL) model, where a significant fraction of
atomic sites in the two adjacent grains coincide or match at the interface. This is illustrated below
where one lattice has been rotated by an angle of 36.9 (equal to cos
-1
0.8).



A "5 boundary:



In the above figure, as one in five lattice points coincide exactly, this is known as a "5 boundary.

21
grain
boundary
energy !
misorientation !
AH57 Course A: Phase Transformations AH57


rob.wallach@msm.cam.ac.uk 2013-14
3.3 Deformation, Recovery, Recrystallisation, Twinning and Grain Growth
When a metal or alloy is deformed, the shapes and orientations of the grains within it are altered
and also substantial increases in internal defects occur, e.g. dislocations from 10
10
to 10
16
m m
-3
.
Subsequent heating annealing results in changes in the defect density and the grain structure
by processes called recovery and recrystallisation. Further continued heating may then lead to
the onset of grain growth or grain coarsening.

3.3.1 The effect of cold-work on metals or alloys
Plastic deformation or cold work occurs when the shape of a metal or alloy is changed permanently
at room temperature, for instance by rolling, forging or extrusion. It results in:
an overall macroscopic shape change;
changes in the shapes and crystallographic orientations of grains within the body;
increased internal defect densities ! of dislocations and also of vacancies.

!
cold work

(a) annealed state (equiaxed grains) (b) cold-worked state (elongated grains)

The higher defect densities result in an increase in the stored energy within the body and this can
be estimated as follows:




Ignoring work-hardening and, by analogue to energy equals the product of force and distance, the
work done W when a stress "
y
(force per unit area) results in a shape change (strain #) is given by:
W

= "
y
#
For a steel with yield stress "
y
= 300 MPa and subjected to a strain # = 0.1, W = 30 MJ m
-3

Most (~95%) of this work is released as heat, as is readily noticed when a large object is plastically
deformed and becomes hot. Only around 5% (in this case ~1.5 MJ m
-3
) is stored as internal
defects - mainly the increase in dislocation density "$ due to work hardening.
) )
stress #
strain $
"
y

AH58 Course A: Phase Transformations AH58


rob.wallach@msm.cam.ac.uk 2013-14
3.3.2 Recovery
When a cold-worked sample subsequently is annealed by heating to ~ 0.3 T
m
, its stored internal
strain energy is reduced as excess dislocations glide and climb to form networks of sub-grains
within the original elongated deformed grains (also, some dislocations of opposite sign may
annihilate each other).





cold-worked state recovered state
random dislocations dislocations in arrays forming sub-grains
As the overall dislocation density $ is not decreased markedly, the strain energy U is not reduced
substantially and so the mechanical properties are not significantly altered.
However, electrical conductivity is recovered. The resistivity decrease is due to a reduction in
point scattering sites and the formation of defect-free channels between the dislocation arrays
within each grain. Hence electrons have greater mean free paths when accelerated by a voltage.

3.3.3 Recrystallisation
The process of recrystallisation occurs at a temperature above ~ 0.6 - 0.7 T
m
. Favourably oriented
sub-grains grow, sweeping out both the polygonised sub-grain structure and also the elongated
grain boundaries in order to form new, equiaxed grains. Note the new grains are not nucleated but
grow from a number of suitable sub-grains.
The driving force for the transformation is the reduction in internal energy U, e.g. the dislocation
density decreases from around 10
16
m m
-3
to 10
10
m m
-3
. The mechanical properties change as the
effects of work hardening are eliminated, e.g. the yield strength decreases.
The final recrystallised grain size is determined by:
a) the extent of the original cold-work, as this determines the dislocation density. The strain prior to
annealing affects the driving force for recrystallisation and also the size of the sub-grains formed
during recovery. Less original cold work will act to inhibit recrystallisation.
AH59 Course A: Phase Transformations AH59


rob.wallach@msm.cam.ac.uk 2013-14









b) extent of alloying additions as increased alloying acts to inhibit recrystallisation. Alloy additions
will either be present in solid solution (leading to solute drag) or will form second-phase
precipitates (leading to Zener drag see page AH62).






3.3.4 Effect of recovery and recrystallisation on properties






Resistivity falls dramatically upon recovery, whilst yield strength and toughness change when
recrystallisation occurs.
T
recrystallisation

(C)
% Al
99.97 99.98 99.99 100
400
300
200
99.97 99.98 99.99 100
400
300
200
T
annealing
/ T
m 0.3 0.6
21
property,
measured
at room
temperature
Strain before annealing
From http://aluminium.matter.org.uk/content/html/eng/default.asp?catid=68&pageid=1238100209
AH60 Course A: Phase Transformations AH60


rob.wallach@msm.cam.ac.uk 2013-14
The graph below shows microhardness data for a variety of aluminium alloys, where the reduction
in hardness is indicative of the onset of recrystallisation. There is a general trend to higher
recrystallisation temperatures with increasing alloying additions due to solute drag.







3.3.5 Twinning
Two completely distinct forms of twinning may be observed in different crystal lattices:
- deformation or mechanical twinning in lattices with insufficient slip systems, e.g. hcp or bcc
- annealing or recrystallisation twin but only in fcc lattices
3.3.5.1 Deformation twins or mechanical twins
These may be seen in an alloy which has a crystal structure that does not have sufficient
independent slip systems to be able to deform by slip (e.g. hcp. at lower temperatures).

For a general shape change by dislocation movement, at least 5 independent slip systems are
needed. If fewer operate, the strain resulting from an imposed stress may be taken up by twinning
and not dislocation movement. Mechanical twins are shown below for a cold-rolled Zn sample.




Pure zinc deformed (from DoITPoMS micrograph library)
Note that deformation twins have a lenticular (lens-like) shape, tapering towards the edges of the
grain boundary. This is to minimise the overall strain energy associated with the edges of the
mechanically twinned regions where they link to grain boundaries.
microhardness
at room
temperature
(a measure of
yield strength)
annealing temperature C (30 minute soak)
AH61 Course A: Phase Transformations AH61


rob.wallach@msm.cam.ac.uk 2013-14
3.3.5.2 Annealing twins (fcc alloys only)
These may form during recrystallisation of some fcc alloys. Within a single grain, one or more
regions commonly grow in twinned orientations, where one twin is a mirror image of the other.
Annealing twins are characterised by their straight or facetted edges, and so are distinct from the
lenticular shapes observed as a result of mechanical twinning of hcp alloys shown on page AH60.




Cu-30wt.%Zn alloy - single phase a-brass (from DoITPoMS Micrograph library)
Annealing twins arise in fcc alloys when atoms diffuse across sub-grain boundaries as sub-grains
grow during recrystallisation . Some of these atoms may join {111} close packed planes in the
recrystallised regions incorrectly, i.e. ABA atomic packing (local hexagonal) may occur in the <111>
direction rather than the expected ABC as in fcc. This is called a stacking fault (as AH54).
Note that the probability of forming annealing twins depends on the amount of the extra energy
introduced by the incorrect local ABA packing. Hence in aluminium alloys, which have intrinsically
high stacking fault energy, annealing twins are not seen. However, they are observed in alloys with
low stacking fault energies such as %-brass, shown above, and austenitic stainless steels.

3.3.6 Grain growth and its control
The major driving force for recrystallisation is the reduction in excess stored energy associated with
the defect density arising from the original cold-work, i.e. it is determined by the amount of elastic
strain present in the alloy due to the higher defect concentration before annealing.
Prolonged further heating after recrystallisation is complete often results in further reductions in
energy by decreasing the recrystallised grain boundary area by grain growth. This process is
controlled solely by diffusion and so depends on temperature and time only, not prior deformation.
The driving force for grain growth is addressed in section 3.3.6.3 and also is compared with the
forces acting to restrict grain boundary motion.
m m
AH62 Course A: Phase Transformations AH62


rob.wallach@msm.cam.ac.uk 2013-14
Both mechanical and physical properties are affected by grain size and hence the control of grain
size is important in many materials to optimise the required properties.
Solute additions in alloys can restrict grain growth in two ways:
by solute atoms solute drag;
by precipitates Zener drag.

3.3.6.1 Solute drag
In a solid solution, the sizes of solute atoms differ from those of the matrix atom which causes:
local lattice distortion and so a region of higher local strain energy around a solute atom;
preferentially solute segregation to sub-grain and grain boundaries to reduce strain energy.
This in turn results in less mobility of the boundaries since either more energy is needed to move
the boundaries, or the boundaries move but more slowly as the solute atoms are dragged along.
Hence pure metals are easier to recrystallise than alloys, and do so at lower temperatures.






3.3.6.2 Zener drag
Zener drag is the term used to describe the pinning of grain boundaries by precipitates or particles
embedded in the matrix, rather than atoms as in solid solution strengthening. A grain boundary has
a certain amount of surface energy associated with it. If part overlaps with a particle that bit of grain
boundary no longer exists in the matrix and this will lower the total energy of the system.
Assuming spherical particles, the largest reduction in energy occurs when the grain boundaries
intersect the equators of the particles and so there will be a force which attracts the boundary to the
centre of such particles this is shown schematically on the next page.
T
recrystallisation
(C)
% Al
99.97 99.98 99.99 100
400
300
200
99.97 99.98 99.99 100
400
300
200
AH63 Course A: Phase Transformations AH63


rob.wallach@msm.cam.ac.uk 2013-14
The surface energy associated with the grain boundary gb (J m
-2
) is equivalent to a force per unit
length (N m
-1
) which acts where it intersects the particles. This surface energy is denoted by !.







2-D representation of a gb plane perpendicular to the page intersecting a spherical partical
The boundary is moving upwards, but experiences an attractive force (a drag force) back towards
the centre of the particle. The site where the drag force acts is a line where the grain boundary and
particle intersect. This is a circle with circumference 2& (r cos').
The force per unit length exerted by the boundary on the particle resolved in the direction of
movement of the boundary is ! sin'.
Hence the total force, F, which acts to pull back the boundary is
F = ! sin' 2&r cos'
i.e. F = r & ! sin2'
Differentiating with respect to ' dF = r & ! 2cos2' d'
Hence there will be a maximum in this pull-back force when '=45, and so
F
max
= & r !
Note that there is no net force in either direction for:
' = 0 as the boundary is located at the lowest energy position and
' = 90 as the area of contact is zero.
'
! sin ' ! sin
'
! ! surface energy
r
particle
(spherical)
grain
boundary
grain 2
grain 1
AH64 Course A: Phase Transformations AH64


rob.wallach@msm.cam.ac.uk 2013-14
A grain boundary will intersect many particles, as shown below.



Consider n particles per unit volume, each of radius r, then the volume fraction of particles is:




Particles whose centre lies within r of the grain boundary will intersect it.
Considering a boundary of unit area, the number of particles which will intersect it (n
A
) is simply the
number of particles per unit volume (n) multiplied by the volume lying within 2r of the boundary.
Hence the number of particles intersecting unit area of boundary:


n
A
= 2r n =
3f
2! r
2

and so the total force per unit area exerted on the grain boundary (the drag pressure P) is :
P = F
max
n
A
= ! r " n
A
=
3 " f
2 r

This shows that the drag pressure P increases as r decreases and
as f and ! increase,
that is P will be large if there is a fine scale dispersion of small incoherent (high !) particles.
Note that the net pressure P = 0 if the boundary is planar, since statistically, equal numbers of
particles will be pushing and pulling. In practice, the boundary curves such that there are a larger
number pulling it back than pushing it forward.



f =
4
3
! r
3
n i.e. n =
3f
4! r
3
AH65 Course A: Phase Transformations AH65


rob.wallach@msm.cam.ac.uk 2013-14
3.3.6.3 Effectiveness of particles in restricting grain growth
The driving force for grain growth
Let a grain be represented by a sphere of radius r, in which case its surface area A = 4&r
2
and its
volume V = 4/3&r
3
.
Hence the energy associated with a grain boundary (since the surface is shared between two
adjacent grains) is half of 4&r
2
! (where ! is the surface energy).
The energy per unit volume of a grain will be given by:


E
V
=
2!r
2
"
4
3
!r
3
=
3"
2r
=
3"
D
(where D is the grain diameter)
Hence the energy associated with grain boundaries can be reduced further after recrystallisation by
grain growth. This is the driving force (or grain growth pressure) for that process.
For a material in which the grain diameter is 10 m and the grain boundary energy is 0.3 J m
-2
:


P
G
=
3!
D
=
3 " 0.3
10 " 10
#6
$ 0.1 MJ m
-3


The Zener pinning force
Consider an alloy with grain boundary energy ! = 0.3 J m
-2
:
small spherical precipitates r = 30 nm
which make up 1 volume % f = 0.01

From the previous page, P
Z
=
3 ! f
2 r
=
3 ! 0.3 ! 0.01
2! 30 !10
"9
= 0.15 MJ m
-3

As this is similar to the driving force for grain growth (as above), this fine dispersion of precipitates
will inhibit grain growth.
In section 3.3.1, the driving force for recrystallisation by a reduction in dislocation density was
estimated as 1.5 MJ m
-3
. Hence in most alloys which contain fine particles or precipitates:
- recrystallisation occurs readily (as long as the temperature is high enough to enable diffusion);
- grain growth beyond around 10 m will, however, be restricted by Zener pinning.
AH66 Course A: Phase Transformations AH66


rob.wallach@msm.cam.ac.uk 2013-14
3.4 Precipitates
3.4.1 Importance of precipitates
The presence of precipitates in an alloy can inhibit grain growth via the mechanism of Zener drag,
see section 3.3.6.2. Precipitates also are important in raising yield strengths of materials by pinning
dislocations (see Part IA Course D).
Dislocation pinning requires the precipitates to be finely dispersed. If they are able to grow, or
coarsen (by a mechanism known as Ostwald ripening, see section 3.4.8), dislocations can easily
move past them (a mechanism known as Orowan bowing) and the alloy strength will decrease.
A system where precipitation hardening is useful is that of Al-Cu (used in aircraft alloys). This was
introduced in Part IA Course C and will be looked at again in more detail in section 4.4.
3.4.2 Precipitate formation - revision of thermodynamics.







Sequence to optimise formation of fine precipitates:
- hold at temperature (1) to attain complete solid solution;
- quench to room temperature (2) to maintain solid solution;
- age at slightly elevated temperature (3), which is below the solvus temperature, to allow
sufficient atomic diffusion to nucleate and grow precipitates in a controlled manner to ensure
optimal precipitate size, coherence and spacing to maximise strength.
The temperatures in practice for an Al-4 wt.% Cu alloy areT
age
~160C and T
solvus
~550C.

T
%
liquid
% + L
C
o
% + !
composition %B
AH67 Course A: Phase Transformations AH67


rob.wallach@msm.cam.ac.uk 2013-14
3.4.3 Precipitate formation - revision of kinetics







Thermal fluctuations create embryos of the new phase, which only become stable nuclei on
exceeding the critical radius r*.
The driving force (G
V
depends on the undercooling "T. This driving force is opposed by:
- strain energy, U, (if the nucleating solid phase has different volume to the matrix), and
- surface energy, !, (which depends on the coherency of the interface).
For a spherical nucleus, the work of nucleation "G
n
is:
!G
n
= !
4
3
!r
3
(!G
v
!U) + 4!r
2
"
and by differentiating this with respect to r, we can show that critical values for a stable spherical
nucleus are:


r * =
2!
"G
v



!G
n
* =
16"
3
#
3
!G
v
2

Because the surface energy, %, opposes the transformation to the new phase, more coherent
metastable precipitates may form more readily as an intermediate stage because "G
n
* will be lower
for these. This occurs in the Al-Cu system.

r
4 & r
4
/
3
& r
3
( G
v
W
n
r
area x surface energy !
r
volume x ((G
v
U)
radius of precipitate r
AH68 Course A: Phase Transformations AH68


rob.wallach@msm.cam.ac.uk 2013-14
3.4.4 Nucleation of metastable precipitates
When considering the thermodynamics of non-equilibrium phase growth, free energy curves and
the common-tangent construction, as Part IA Course C, are still the approach for determining the
solvus for metastable precipitates.
Consider quenching an % solid solution to a temperature T*














On quenching from a solid solution to T*, the free energy of the supersaturated solid solution is G
1
.
This can be lowered to G
2
if the supersaturated solution decomposes to form the two phases % and
metastable ).
The free energy can be further lowered to G
3
if diffusion is allowed to continue and the metastable )
phase changes to the stable phase ) in an % matrix.


G
T
solvus for
stable ) phase
solvus
solvus for metastable ) phase
metastable phase )
stable phase )
solid
solution %
solid
solution
composition
% solute
.
.
composition
T*
AH69 Course A: Phase Transformations AH69


rob.wallach@msm.cam.ac.uk 2013-14
3.4.5 Diffusion-controlled precipitate growth
Precipitates () phase) will generally have a different chemical composition to that of the matrix (%)
and hence their growth is usually by a diffusion-controlled mechanism.
As an individual precipitate grows, the extent of the associated diffusion field increases. This slows
down the rate of growth as the solute needed for growth has diffuse from a further distance.
Consider the isothermal growth, after its nucleation, of a precipitate in a binary system:







The solute profile for half of a growing precipitate would ideally be represented by:






A number of assumptions are involved:
concentrations at the interface between % and ) are those given by the phase diagram;
concentration gradient in the matrix is constant and linear;
far-field concentration, C
o
, is constant, i.e. assume growing precipitates are far apart.
Note that this breaks down at long times i.e. as X and Y converge and so dC/dx decreases.
This results in a decrease in the ) growth rate. It is called solute impingement.

T
% B
% % + ) )
C
%
C
)


C
o
distance x
concentration
(%B)
(C
&'
C
%
C
)

(x
C
o
(C
ss
AH70 Course A: Phase Transformations AH70


rob.wallach@msm.cam.ac.uk 2013-14
The extra solute required by the precipitate ) comes from the depleted region, hence by equating
areas in the diagram on the previous page:


C
!
" C
0
( )
x =
1
2
#x C
0
" C
$
( )
[A]
As the precipitate grows, the interface advances dx in a time dt.






The rate at which solute is incorporated into the precipitate (shaded grey above) equals the
diffusive flux of solute down the concentration gradient towards the interface:


C
!
" C
#
( )
dx
dt
= D
C
0
" C
#
( )
$x
[B]
Eliminate (x using the two equations [A] and [B] gives:
v =
dx
dt
=
C
0
!C
!
( )
2
C
!
!C
"
( )
C
!
!C
0
( )

D
2x

If C
)
>> C
%
then (C
)
- C
%
) * (C
)
- C
o
) which can be denoted by (C
%)
.
Similarly, let (C
ss
represent (C
0
- C
%
) and so the above equation can be rewritten as:
v =
dx
dt
=
!C
ss
!C
!"
"
#
$
%
&
'
2
D
2x


distance x
concentration
(%B)
(C
%)
C
%
C
)
(x
C
o
(C
ss
AH71 Course A: Phase Transformations AH71


rob.wallach@msm.cam.ac.uk 2013-14
Rearranging this gives

2x dx =
!C
ss
!C
"#
$
%
&
'
(
)
2
D dt
and integrating

2
!
x dx =
"C
ss
"C
#$
%
&
'
(
)
*
2
D
!
dt


x
2
=
!C
ss
!C
"#
$
%
&
'
(
)
2
D t
Hence the precipitate thickness

x =
!C
ss
!C
"#
$
%
&
'
(
)
D t
By differentiation, the growth velocity is given by


v =
dx
dt
=
1
2
!C
ss
!C
"#
$
%
&
'
(
)

D
t


In practice, the ) precipitate initially grows faster than this analysis predicts, due to the fact that the
diffusion profile is non-linear as shown below.






Due to this, more accurate expressions turn out to be twice as large:
x = 2
!C
ss
!C
!"
"
#
$
%
&
'
Dt and

v =
dx
dt
=
!C
ss
!C
"#
$
%
&
'
(
)

D
t


distance
concentration
(%B)
AH72 Course A: Phase Transformations AH72


rob.wallach@msm.cam.ac.uk 2013-14
3.4.6 Nucleation of precipitates
When precipitates form in a solid matrix, heterogeneous (rather than homogeneous) nucleation is
much more likely because of the introduction of strain energy and surface energy. Thus
precipitation tends to occur preferentially at:
grain boundaries;
dislocations;
vacancies;
near impurities (or other precipitates) due to their local strain fields.

Consider the role of vacancies in an alloy. Under equilibrium, the number of vacancies n is given by:

n = n
0
exp !
Q
f
kT
"
#
$
%
&
'

where Q
f
is the activation energy for vacancy formation. Excess vacancies are trapped when an
alloy is quenched from high temperature and these lead to an enhanced diffusion rate at the lower
temperature.
These excess vacancies will diffuse to sinks, e.g. grain boundaries, in order to return to the
equilibrium vacancy concentration for the given temperature. This will lead to a variation in vacancy
concentration, especially near to a grain boundary.
As precipitates nucleate preferentially on vacancies (lowers the activation energy), there will be a
critical vacancy concentration at a given temperature to allow nucleation.









distance
vacancy
concentration
grain
boundary
vacancy concentration
needed for precipitation
AH73 Course A: Phase Transformations AH73


rob.wallach@msm.cam.ac.uk 2013-14
3.4.7 Precipitate free zones (PFZs)
a. Formation: vacancy depletion
Precipitation is less probable for lower vacancy concentrations and so the lower vacancy
distributions at grain boundaries result in precipitate-free zones adjacent to grain boundaries.
At the grain boundaries themselves, precipitates nucleate very readily due to the local lattice misfit
energy of the boundary itself, and are often larger than those in the nearby matrix






b. Formation: solute depletion
Another cause of PFZs can be due to solute depletion near the grain boundary as a consequence
of preferential heterogeneous nucleation and growth of precipitates on the boundary itself.
c. Effects of precipitate-free zones:
softer, weaker zone in which any subsequent deformation may be localised (can lead to
failure adjacent to grain boundaries);
composition variation adjacent to grain boundaries (can cause localised corrosion as an
electrochemical cell is set up).

The widths of precipitate-free zones can be minimised by optimising precipitation heat-treatments:
quench to retain supersaturated solution with excess vacancies;
heat at low ageing temperature to promote initial metastable precipitates as formation of
these is less dependent on number of vacancies because precipitates are coherent (and so
have smaller nucleation barrier);
raise ageing temperature slightly to grow more stable precipitates and to optimise strength
by control of precipitate size, coherence and spacing.
grain boundary
fine precipitates
within grains
fine precipitates
within grains
AH74 Course A: Phase Transformations AH74


rob.wallach@msm.cam.ac.uk 2013-14
3.4.8 Stability of precipitates Ostwald ripening
For high yield strength, it is advantageous to have an array of fine-scale precipitates so that they
are effective in pinning dislocations. Hence it is important that they do not coarsen at the
temperatures at which the material is used.
Smaller precipitates have a greater ratio of surface area to volume, hence have higher free energy.
This provides a driving force for the diffusion of solute from the small precipitates to larger ones.



Consider a precipitate of radius r containing n moles of solute each of molar volume V
m
. Then

n V
m
=
4
/
3
! r
3
(A)

The free energy of formation of the precipitate !G
n
(see section 3.4.3)
!G
n
= !
4
3
!r
3
(!G
v
! U) + 4 ! r
2
" (B)
Differentiating (A) and (B)


! "G
n
( )
!n
=
! "G
n
( )
!r

!r
!n
= # 4! r
2
("G
v
# U) + 8! r !
$
%
&
'

V
m
4! r
2




= V
m

2!
r
! ("G
v
! U)
#
$
%
&
'
(



Let n moles be transferred from particle of smaller radius r
2
to particle of larger radius r
1
.
The net change in the free energy of the system is:




If r
1
> r
2,
!G is negative, i.e. there is a thermodynamic driving force for transfer of atoms from
small to large particles (from r
2
to r
1
).
Hence the driving force for coarsening is proportional to

2!
r
where r is precipitate radius
% is surface energy.
Accordingly, coherent precipitates, with low surface energy with respect to the matrix, are stable
and do not coarsen.
solute
diffusion
small ' larger '
radius r
2
radius r
1

!G = V
m

2!
r
1
" !G
v
"U
( )
#
$
%
&
%
'
(
%
)
%
" V
m

2!
r
2
" !G
v
"U
( )
#
$
%
&
%
'
(
%
)
%
= 2! V
m

1
r
1
"
1
r
2
*
+
,
-
.
/
AH75 Course A: Phase Transformations AH75


rob.wallach@msm.cam.ac.uk 2013-14
Coarsening will depend on:
surface energy, ! - which depends on the precipitate coherence;
temperature - which governs the rate of diffusion;
solubility of solute in matrix - need solubility to allow atom transfer between precipitates
- if precipitate is insoluble, coarsening could not occur
(this is the basis of ODS alloys, see page AH76)


Coarsening can thus be controlled by:
the nature of the precipitates, hence the use of coherent ! precipitates Ni
3
(Ti,Al) in nickel-
based superalloys (see section 4.6);
strict control of temperature (hence the need to monitor temperatures of aircraft engines).

The overall alloy strength is determined by
precipitate density, size and coherency.

AH76 Course A: Phase Transformations AH76


rob.wallach@msm.cam.ac.uk 2013-14
3.4.9 Dispersion strengthening to avoid coarsening: ODS alloys
Dispersion strengthening is based on introducing fine, insoluble particles into an alloy during alloy
fabrication. The insoluble particles cannot coarsen since the third condition on the previous page is
not satisfied and hence problems associated with precipitate coarsening are avoided.
The insoluble particles provide good strengthening, including at elevated temperatures, since they
inhibit dislocation movement as a consequence of their fine size and uniform distribution.

Oxide dispersion strengthening involves:
use of insoluble particles in the matrix (i.e. Al
2
O
3
in Al, and Y
2
O
3
in Ni or Fe-based alloys);
fabrication of powders from complex alloys by rapid solidification;
manufacture of alloys of specific compositions by initially mechanically mixing powders and
then mechanically compressing to form green compact;
sintering of green compact to consolidate the material this involves diffusion, so requires
holding at high temperature, pressure for considerable times.
The high temperature stability of such alloys is better than the equivalent precipitation-hardened
alloys although the oxide distribution may not be as fine or the particles as small as precipitates.





strength
temperature T
ODS alloy

precipitation
hardened alloy
AH77 Course A: Phase Transformations AH77


rob.wallach@msm.cam.ac.uk 2013-14
4. ALLOY SYSTEMS
4.1 Ternary Phase Diagrams
4.1.1 Representing three components and temperature
To date in Parts IA and IB, only phase diagrams for binary systems (those in which there are only
two components) have been considered. Many alloys of practical importance have a greater
number of alloying additions and it often helpful to use a phase diagram in which there are three
components. In order to represent three components plus the temperature, it is convenient to
construct a triangular-based prism.







In this 3-dimensional space, volume elements represent phase regions.
A vertical section is the same as a binary phase diagram; for example if A-B is a eutectic system,
the phase boundaries on the front face would be shown as:








!!
T
C

A



B
T
C

A



B
AH78 Course A: Phase Transformations AH78


rob.wallach@msm.cam.ac.uk 2013-14
4.1.2 Constant temperature sections in ternary phase diagrams
A useful representation of ternary phase diagrams is a horizontal section which shows the effect of
composition at a single temperature, an isothermal section. This produces an equilateral triangle
with the 3 components at the corners, and each point within the triangle represents a unique
combination of the components.







Clearly the lower left corner represents 100% A, whilst the line BC represents 0% A. Each line
parallel to BC is a constant composition of A, which simply increases linearly across the triangle.
Hence, compositions can easily be read from the grid.
There is also another way to find the compositions. If lines are drawn, parallel to the axes, through
the constitution point, these lines will split the axes into parts which are proportional to the
compositions of each component. Clearly the operation of working out the composition can be done
by drawing just two lines.







A
B
C
A
B
C
50% A 20% B 30% C
AH79 Course A: Phase Transformations AH79


rob.wallach@msm.cam.ac.uk 2013-14
A good example of a simple ternary diagram is that for Fe,Ni,Cr shown below representing a
temperature of 900C. It is evident that Fe and Ni (both f.c.c.) form a complete solid solution and
that a reasonable amount of Cr can be dissolved into this structure. Meanwhile b.c.c. Cr can
dissolve a large amount of Fe, but any significant amount of Ni results in 2 phases.











Stainless steels
The Fe, Cr and Ni ternary system is the basis of many stainless steels, all of which contain at least
10 wt.% Cr and can go up to 25 wt.% Cr (although the majority of alloys have less than 20 wt.% Cr).
These amounts of chromium ensure that stable and coherent chromium oxide surface layers form
and these protect stainless steels in aggressive environments in which other steels corrode rapidly.
Cr is a b.c.c or ferrite # stabiliser in steels while Ni is a f.c.c or austenite $ stabiliser (see AH 83).
Hence there are four major classes of stainless steel:
ferritic # (low Ni) austenitic $ (~10 wt.%Ni) dual phase martensitic
Two stainless steels are:

a. 18 wt.%Cr, 8 wt.% Ni, balance Fe which is austenitic at room temperature;
b. 25 wt.%Cr, 4 wt.% Ni, 4 wt.% Mo balance Fe which is a duplex stainless steel (in AP1) and
is a 50:50 mix of $ and #. It solidifies as fully ferritic and partially transforms to $ on cooling.
900C
section
AH80 Course A: Phase Transformations AH80


rob.wallach@msm.cam.ac.uk 2013-14
4.2 Cast Irons
4.2.1 Introduction to cast irons
The phase diagram for the Fe-C system, which was introduced in Part IA, is usually depicted for
iron-cementite (Fe-Fe
3
C). In fact, iron-graphite is the thermodynamically stable system although
Fe
3
C, a metastable phase, is usually shown rather than graphite because kinetic considerations
mean that Fe
3
C generally predominates in steels.
Cast irons, which have carbon contents in the range of 2 - 4 wt.%, often can form the stable
graphite phase. The dotted lines in the phase diagram below denote the phase boundaries for iron-
graphite.













Cast irons usually contain Si as well as other alloying additions. These alloying additions have a
significant effect on the phases which are found in the final alloy (often causing graphite to be
formed even for alloys which contain less than 4.3 wt% C). There are many different types of cast
iron with characteristic microstructures and properties. Two common families of cast iron are grey
cast iron and white cast iron.
AH81 Course A: Phase Transformations AH81


rob.wallach@msm.cam.ac.uk 2013-14
4.2.2 Grey cast irons
Grey cast irons are favoured by high C or Si content, and if the cooling rate is slow
If the alloy is hyper-eutectic, i.e. C > 4.3 wt% or lower if significant Si > 1 wt.% is present:
- graphite flakes form when the temperature falls below the liquidus line;
- at the eutectic temperature, a graphite ! eutectic forms from the remaining liquid phase;
- on cooling further below the eutectic temperature, C is rejected from the ! solid-solution to
grow on to existing pro-eutectoid flakes and also may nucleate new graphite flakes;
- at the eutectoid temperature (~723C), the ! matrix normally transforms to pearlite.

The presence of additional alloying elements can affect the above. For instance, sufficient Ni
stabilises the ! phase and hence the final alloy can be graphite flakes in a ! matrix. This is due to
the thermodynamic effect of the Ni in Fe (see section 4.3.2).







Graphite flakes are desirable because they have a low density leading to a volume expansion on
solidification (good for surface definition of castings). Also, the alloys have good machinability as
graphite acts as a lubricant for cutting tools as well as being intrinsically brittle.
A negative aspect is that graphite flakes act as stress concentrators, leading to poor toughness in
service applications. Moreover, the graphite flakes are interconnected so that any cracks can
propagate readily.

Common applications of grey cast irons include car cylinder heads and the bases of machining
tools such as lathes.
Fe 3.6 wt.% C 2.1 wt.% Si (from DoITPoMS Micrograph library, 62 and 63)
AH82 Course A: Phase Transformations AH82


rob.wallach@msm.cam.ac.uk 2013-14
4.2.3 Spheroidal graphite (SG) cast irons
One variant of the grey cast irons is formed by Mg or Ce additions (~0.1 wt.%) to the melt which
cause spherical graphite nodules to form rather than the interconnected flakes.
Two theories have been proposed to account for the effect of these elemental additions.
i) The Mg or Ce alloying additions poison the preferred growth sites by attaching to the
edges of the graphite planes, slowing growth in that direction. Hence the graphite grows
in a more isotropic manner, leading to the development of approximately spherical
particles.
ii) Alternatively, the surface energies of the graphite and the liquid melt are altered such
that the surface area of the graphite is minimised when growing in the melt.







The absence of the graphite flakes leads to increased toughness. This effect is similar to that of
sodium in Al-Si casting alloys. Brittle interconnected Si plates are prevented from forming if small
amounts of the Na modifier are added to the melt (see question sheet AQ2).
Applications of spheroidal graphitic cast irons are generally those in which the increased
toughness/ductility is important and include car crankshafts, ductile iron pipes, fittings, valves and
manhole covers.
Fe 3.6 wt.% C 2.1 wt.% Si 0.07 wt.% Mg (from DoITPoMS Micrograph library 64 and 65)

AH83 Course A: Phase Transformations AH83


rob.wallach@msm.cam.ac.uk 2013-14
4.2.4 White cast irons
This family of cast irons are favoured by low C or Si contents (<1 wt.%) and fast cooling rates, so
that the non-equilibrium cementite, Fe
3
C, forms rather than graphite.
If the alloy is hypo-eutectic (C < 4.3 wt%):
- ! dendrites form as the primary phase, followed by the ! and Fe
3
C eutectic intergrowth;
- on further cooling below the eutectic temperature, C is rejected from the ! phase to form more
Fe
3
C;
- at the eutectoid temperature A
1
(~723C), any remaining ! (both the primary dendrites plus that
formed in the ! and Fe
3
C eutectic) transforms to pearlite.
.







White cast irons are harder and more wear-resistant than grey cast irons. Applications include rolls
for crushing minerals.

Notes.
(i) The names grey and white used to describe cast irons are not associated with the etched
microstructures, but come from the appearance of the fracture surfaces.
White cast irons are brittle and so form smooth fracture surfaces, whereas grey cast irons,
which are more ductile, have less reflective surfaces when fractured.
(ii) There are many other cast irons depending on precise composition and/or cooling conditions.
All build on basic steel metallurgy, e.g. fast cooling can promote martensite formation.
!"" $ !"" $ !" $ !" $
Fe ~3 wt.% C (from DoITPoMS Micrograph library 66 and 67d)

AH84 Course A: Phase Transformations AH84


rob.wallach@msm.cam.ac.uk 2013-14
4.3 Steels
4.3.1 Overview of steels [phase diagram on page AH78]
Steels are crucially important due to their widespread use in construction and other applications.
The extent to which steel dominates the metals market in terms of annual production
(approximately 1.4 billion tonnes) is associated both with the low cost (typically around 0.3 /kg)
and also with the excellent mechanical properties (yield strengths often in excess of 1 GPa).
Additionally, the properties can often be tailored to suit a particular requirement by careful choice of
alloying additions together with appropriate themomechanical processing.
Steels are iron-based alloys with C content up to 1.5 wt.%, but almost all steels produced and used
commercially have C << 0.5 wt.% and so are hypo-eutectoid. Many modern high-strength steels
have C contents < 0.1 wt.% and suitable other alloying additions.
Major equilibrium phases in steels are:
ferrite # bcc phase with up to 0.035 wt.% C and stable up to 910C
cementite Fe
3
C orthorhombic structure with composition ~6.7 wt.% C
pearlite eutectoid of # - Fe
3
C stable up to 723C
austenite ! fcc phase with extensive C solubility and stable above 723 - 910C

Categories of steel include:
Plain carbon steels low carbon C < 0.1 wt.%
medium carbon C < 0.25 wt.%
high carbon C > 0.25 wt.%

Alloy steels containing Cr, Mo, Ni, V, Nb, Ti, Al in varying amounts

Stainless steels contain sufficient Cr to form coherent surface Cr
2
O
3
oxide layers which provide
corrosion resistance. There are four major classes of stainless steel, characterised by lattice type:
ferritic (Cr with low Ni and C, e.g. 13 wt.% Cr)
austenitic (Cr with high Ni, e.g. 18 wt.%, Cr, 8% Ni wt.%)
martensitic (as ferritic, but higher C, e.g. 0.1 < C < 0.6 wt.%)
duplex (# and $)
AH85 Course A: Phase Transformations AH85


rob.wallach@msm.cam.ac.uk 2013-14
4.3.2 Thermodynamics
The relative stabilities (free energies #G) of ferrite and austenite are shown below on a graph of
free energy versus temperature). The metastable phase martensite, $, is included see AH85.









If austenite $ is quenched sufficiently fast, there is insufficient time for the reconstructive (diffusional)
transformation to ferrite #. When there is a sufficient undercooling below T
E
, martensite # starts to
form, i.e. at the temperature M
S
. The amount of martensite that forms depends on the driving force
from the undercooling and so only at M
F
will all the steel have transformed to martensite.
The undercooling %T is required since opposing the transformation are:
- the increase in energy due to the surface energy of the martensite plates or laths, and
- the strain energy (volume of martensite is 4% greater than the austenite from which it forms)

Different alloying additions alter the nature of the Fe-C equilibrium:
Cr and Si stabilise the ferrite phase (hence % can exist stably at high temperatures), whilst
C, Ni and Mn stabilise the austenite phase (hence ! can exist stably at low temperatures).






Note: thermodynamic changes also affect nature of TTT curves see page AH87.
!
T
Concentration of % stabilisers
%
T

Concentration of ! stabilisers
!
%
M
F
M
S
T
E

ferrite #
martensite #
austenite $
AH86 Course A: Phase Transformations AH86


rob.wallach@msm.cam.ac.uk 2013-14
4.3.3 Kinetics
4.3.3.1 Diffusional transformations.















4.3.3.2 Alloy addition partitioning and TTT curves
As well as affecting thermodynamics (as on page AH83), alloying additions often affect kinetics of
transformations, and this can be seen by shifts in the lines on TTT curves.
Different additions prefer to exist in one phase rather than the other, and will partition by diffusion
during transformation of the austenite phase to ferrite and cementite.
At the transformation interface, ! to % , diffusion occurs, in the $ ahead of the interface, of both C
and other alloying additions.



Carbon occupies interstitial sites in both % and ! phases and therefore can diffuse very rapidly.
Other alloying additions are substitutional in !, % and Fe
3
C and so diffuse much more slowly. This
makes transformations of ! happen more slowly and hence displaces the TTT curves to the right.




!
%
Fe
3
C
Fe
3
C
%
Fe
3
C

2
Pro-eutectoid phase (% or Fe
3
C) initially
nucleates on ! grain boundaries on cooling
below T
A3
, the ! solvus.

When temperature falls below T
eutectoid
or T
A1

the eutectoid reaction commences to form
intergrowth of fine Fe
3
C and % plates.

Relative amounts of phases estimated using
lever rule on equilibrium diagram.

Grain refinement possible due to nucleation
of % grains on ! grain boundaries
T
!
1
log (time)
1% 99%
M
S

M
F


M
s


M
f

AH87 Course A: Phase Transformations AH87


rob.wallach@msm.cam.ac.uk 2013-14
4.3.3.3 Diffusionless transformations: martensite
If a steel is cooled from the austenite (!) phase, the products depend on the cooling rate:
- slow cooling allows the equilibrium transformation to ferrite (%) and possibly pearlite;
- moderately quickly can lead to Widmansttten ferrite (%) or bainites (see AH88);
- quenching rapidly to a temperature typically less than around 250C, results in the formation of
martensite by a displacive (diffusionless) transformation at the speed of sound. Martensite is a
non-equilibrium metastable phase (see AH85). The carbon remains as a supersaturated solid
solution in a bct lattice because there is no time for C to diffuse to form Fe
3
C.
Martensite is typically hard and brittle due to the volume increase of ~ 4% when it forms and high
internal strain energy. Hence entirely martensitic steels usually are avoided but the phase can be
beneficial as a surface layer to promote wear resistance.
The hardness and brittle nature of martensite increases with the carbon content due to the
associated increase in the c/a lattice parameter ratio, hence strain, in the martensitic bct lattice.





Tempering of martensite
Martensite can be tempered by heating to a temperature less than the eutectoid temperature A
1

and at which diffusion can be rapid. As a result, equilibrium phases (# and Fe
3
C) form which
restores some ductility and toughness.
As with any other supersaturated solution, this results in precipitate formation via a series of
metastable phases, e.g. as in the Al-Cu system (AH92).

Pearlite is not formed as the steel has not been cooled from the austenite phase sufficiently slowly
through the eutectoid temperature A
1
, and the tempering itself is at temperature less than A
1
.
Precipitation from a supersaturated solid-solution generally is associated with strengthening, as
occurs in the Al-Cu system, due to increased interaction of dislocations with the small precipitates.
However, when tempering martensite, there is a reduction in strength even though Fe
3
C
precipitates are formed. This is because the significant reduction in the internal strain energy,
present in the original martensite, has a greater effect than the formation of the Fe
3
C precipitates.

% C in martensite
c




a

lattice
parameter
AH88 Course A: Phase Transformations AH88


rob.wallach@msm.cam.ac.uk 2013-14
Tempering a martensitic steel allows control of yield strength, ductility and toughness.






Secondary hardening occurs in some alloy steels containing strong carbide formers, and:
- causes yield strength to increase when tempering around 600C dotted line in above figure;
- is due to the formation at around 600C of stable carbides of Cr, Mo, W, Nb or V;
- these steels have better wear-resistance than plain-C steels and are used as cutting tools, i.e.
are able to retain strength when dull red hot, and also for creep resistance in power plants.
4.3.4 Bainites
The (reconstructive diffusional) transformation to ferrite and pearlite during slow cooling and the
(displacive shear) transformation to martensite when quenching have been described. However, it
also is possible to produce different steel microstructures, in addition at intermediate cooling rates.
Bainites form by combined shear and diffusional transformations at temperatures above that at
which martensite forms. The resulting microstructures are extremely fine and can be resolved only
at high magnification using transmission electron microscopy (TEM).
Upper bainite:
ferrite % forms at austenite ! grain boundaries on cooling and grows into the ! by a shear
mechanism;
the % is plate-like in form, but on a very fine scale;
C diffuses to the %/! interface where fine Fe
3
C precipitates nucleate and grow.



%
!
C diffusion
$ grain
boundary
Tempering temperature (C)
100
100 700
Hardness or
yield strength
alloy steel: secondary
hardening

plain C steel
AH89 Course A: Phase Transformations AH89


rob.wallach@msm.cam.ac.uk 2013-14
Lower bainite:
forms at slightly lower temperatures where there is insufficient time for C to diffuse to the %/!
interface;
fine Fe
3
C precipitates nucleate and grow with an orientation relationship with the ferrite in
which they are growing.





4.3.5 TTT diagrams
As was seen in section 4.3.2, different alloying additions in steels alter the thermodynamics of
phase stability in the Fe-C system. Not only are the presence of the equilibrium phases affected
but also the kinetics of their formation. The kinetics of phase formation can be illustrated using TTT
diagrams (Part IA Course C).
For steels, alloying additions can affect TTT diagrams as follows:
- temperatures at which phases first nucleate change;
- rates of nucleation and growth are changed, and phase growth is very dependent on the
relative diffusion coefficients of the elements present;
- relative stabilities of the various phases change, depending on the alloying additions present;
- M
S
and M
F
temperatures are altered.





%
!
$ grain
boundary
T
time (log scale)
T
time (log scale)
# and
pearlite
bainites
start end start end
Ms
Mf
Ms
Mf
# and pearlite




bainites
0.3 wt.% C plain C steel alloy steel (e.g. 3 wt.% Ni, 1 wt.% Cr)

AH90 Course A: Phase Transformations AH90


rob.wallach@msm.cam.ac.uk 2013-14
4.3.7 Steels: summary of transformations
Cooling rate:
slow ferrite % and then pearlite form by equilibrium diffusional transformations
% nucleates at ! grain boundaries and this leads to grain refinement
pearlite forms by eutectoid reaction (coupled growth of % and cementite Fe
3
C)
moderate Widmansttten % (strain energy minimised) grows from ! grain boundaries,
pearlite grows as for slow cooling rate above.
fast upper bainite forms - as shear increases (at lower transformation temperatures)
then changes to
lower bainite formation instead (both types form with coupled shear and diffusion)
pearlite does not form
quench martensite forms by non-equilibrium displacive shear transformation (no diffusion)

Tempering:
Heat a steel below the A
1
temperature (see section 4.2.1), i.e. not into the # plus ! region, hold.
martensite/bainites: classical nucleation and growth from supersaturated martensitic solid-
solution
spheroidal Fe
3
C nucleates and grows in % matrix no pearlite possible
alloy steels: as above but possibility of secondary hardening (see section 4.3.3.3)

Normalise:
Heat a steel above A
3
temperature (see section 4.2.1), i.e. just into the ! region, hold for a short
time and then cool but do not quench.
Resulting nucleation and growth, firstly of ! grains when heating and then of % grains when
cooling, leads to significant grain refinement and consequent improvements in both yield
strength and toughness.



AH91 Course A: Phase Transformations AH91


rob.wallach@msm.cam.ac.uk 2013-14
4.4 Aluminium alloys
4.4.1 Introduction
Al alloys are used in either as-cast or in the wrought (deformed and heat-treated) state.
For cast products:
cooling rate and the use of grain refiners ( see section 2.6.2) are important;
many alloys are based on Al-12 wt% Si (eutectic), and Al-Zn particularly for die-casting;
Si has the diamond cubic structure which is low density and therefore compensates for
shrinkage, reducing porosity in the casting;
modification of silicon is achieved by small additions of Na (~0.2 wt%) to the melt, resulting
in the coarse Si plates becoming fine needles.
For wrought products (rolled, extruded, forged):
final properties rely on appropriate heat treatment/working;
alloys are either precipitation-hardened or work-hardened (marked with * in table);
precipitation-hardened alloys get strength by controlling precipitate distribution, size and
coherency;
worked alloys get strength from small final grain size and high defect density.
Al alloys can be classified according to the US system:

Main additions Strengthening Applications
1xxx pure Al work-harden* Foil and electrical conductors
2xxx Cu ' CuAl
2
pptes Al-Cu-Mg for aircraft wing sheet
3xxx Mn work-harden* General purpose cookware
4xxx Si Si needles Casting alloy
5xxx Mg work-harden* Very common, corrosion resistant, structural, boats
6xxx Mg & Si Mg
2
Si pptes Extruded products, construction, golf clubs
7xxx Mg & Zn MgZn
2
pptes High strength alloy
8xxx Li (also others) + Li
3
Al pptes Low density, increased stiffness - aircraft

Possible strengthening mechanisms include:
work-hardening
solid solution hardening;
grain size control;
two-phase alloys;
precipitates of optimised size and coherency;
oxide-dispersion strengthening;
order hardening.
AH92 Course A: Phase Transformations AH92


rob.wallach@msm.cam.ac.uk 2013-14
4.4.2 Precipitation in Al-Cu alloys
This was introduced in Part IA and the relevant part of the Al-Cu phase diagram is shown below









The solid lines show the equilibrium phase diagram, with an f.c.c. solid solution (denoted by # or ,)
based on Al and the intermetallic Al
2
Cu phase, '. A solvus is also shown for each of the
metastable phases










C
0
C
o
free
energy

composition
100
200
300
400
500
600
700
2 4 0
T
,
L
,

, + '
' '
' ''
GP
1
6
0
wt.%
Cu
+ L
AH93 Course A: Phase Transformations AH93


rob.wallach@msm.cam.ac.uk 2013-14
A supersaturated solid solution is obtained by annealing the Al-4 wt% Cu alloy just below the
eutectic temperature of 548C, then quenching to room temperature. During subsequent heat-
treatment (ageing), a sequence of precipitates is observed: in order, they are
GP1 zones, ''', '' and, finally, '.
The stable (equilibrium) ' phase precipitate is shown on the equilibrium phase diagram. These
precipitation reactions progressively lower the free energy of the system; the metastable
precipitates form preferentially due to lower activation energies.










4.4.3 Structures of metastable precipitates
The first metastable precipitate to form is the Guinier-Preston (GP1) zone. This takes the form of a
Cu plate fully coherent with the Al matrix. There is a high degree of strain due to the differences in
lattice parameter: a
Cu
= 0.9 a
Al
. The plates are ~10 nm in diameter and ~0.5 nm thick




composition, wt.% Cu
,
, +

, + '
, " , + ' (
G
G
0
G
1
G
2
G
3
G
4
,
, + GPZ
, + '"
, + '
, + '
, " , + ' (direct precipitation)

G
G
o
G
1
G
2
G
3
G
4
"G
-
AH94 Course A: Phase Transformations AH94


rob.wallach@msm.cam.ac.uk 2013-14
Next is ''' (also known as GP2). This is a tetragonal precipitate as shown below, fully coherent
with the Al matrix on the {001} plane. These form plates ~15 nm in diameter and 0.8 nm thick which
create some local strain in the z direction.






'' is also tetragonal and also coherent on the {001} plane, but fits less well on the other interfaces,
leading to greater strain.





', the final precipitate to form, is the equilibrium phase. It is incoherent with the Al matrix and has a
body centred tetragonal (bct) unit cell.





0.580 nm
0.404 nm
0.404 nm
0.607 nm
0.487 nm
0.607 nm
0.404 nm
0.404 nm
0.768 nm
Al atoms

Cu atoms
AH95 Course A: Phase Transformations AH95


rob.wallach@msm.cam.ac.uk 2013-14
4.4.4 Optimisation of metastable precipitates for highest yield strength
For the highest yield strength, it is necessary to inhibit dislocation motion.








When the precipitate dispersion is fine and the precipitates are coherent, they can easily be cut by
dislocations. Such an alloy is underaged.
If the dispersion is too coarse, the dislocations can bow between the precipitates (as they will be
widely spaced). Such an alloy is overaged.
In practice, alloys are usually used in a slightly overaged condition, i.e. heat-treated to just past the
peak hardness.
Note that, as was presented in Part IA Course D, the variation in yield strength with ageing time is
consequence of a number of changes occurring during ageing.
cutting
stress
bowing
stress
yield
stress
annealing time
AH96 Course A: Phase Transformations AH96


rob.wallach@msm.cam.ac.uk 2013-14
4.5 Titanium alloys
4.5.1 Overview
Titanium and its alloys have:
extremely high strength and stiffness to weight ratios;
excellent mechanical properties;
very good corrosion resistance, and good oxidation resistance but only up to ~ 600C even
though Ti melts at 1660C.

Hence there are extensive applications including:
aerospace: compressor blades and other parts in gas turbine engines, airframes;
energy generation: nuclear fuel plants, heat exchangers;
chemical: food processing plants;
medical: prostheses such as hip implants.

Titanium exists as a number of phases. T < 882C: % phase hcp
T > 882C: ) phase bcc
Fast-cooled: martensitic phases e.g. orthorhombic %
Alloying additions can stabilise these phases to different temperature regimes (just as for steel
phases in section 4.2.2).
Interstitial elements include O, N, C, which have a high solubility and are mainly % stabilisers. They
provide good solid-solution strengthening.
Substitutional additions usually are %-stabilisers if they have <4 bonding electrons (i.e. Al, Ga) but
are likely to stabilise the ) phase or may even form a eutectoid if they have >4 bonding electrons.
# stabilised & stabilised eutectoid
O, N, C, Al, Ga Mo, V, Nb, Ta Cu, Mn, Fe, Ni
T
) % + )
solute content
%
)
% % + )
solute content
T
)
% + !
!
) + !
solute content
%
T
AH97 Course A: Phase Transformations AH97


rob.wallach@msm.cam.ac.uk 2013-14
4.5.2 Titanium 6:4 (Ti- 6 wt.% Al, 4 wt.% V)
This is one of the most important commercially used Ti alloys. The final microstructure and grain
size depend on the amount of hot-working and the heat-treatment cycle.
The microstructures can be predicted using a similar approach to TTT diagrams for steels but using
a continuous cooling or CCT diagram, as below.







In Ti - 6 Al - 4 V, both the % and ) phases are stable at room temperature. Slow cooling from the
) phase results in the nucleation and growth of Widmansttten % laths inside the ) matrix. This is a
basket-weave structure with good strength, toughness and fatigue resistance.





Ti 6 wt.% Al 4 wt.% V (from DoITPoMS Micrograph library 54)
This alloy also shows superplasticity: the ability to deform very extensively up to ~800% at
temperatures ~0.6 T/T
m

For superplastic behaviour, a fine grain size (20 m) and stable microstructure at the deformation
temperature are required. This allows grain boundary sliding plus instantaneous annealing out of
any work hardening due to dislocation motion.
Superplastic forming is used to fabricate the large fan blades at the front of jet engines.
time (log scale)
T
M
S

M
F

' to & start

' to & finish

AH98 Course A: Phase Transformations AH98


rob.wallach@msm.cam.ac.uk 2013-14
4.6 Nickel-base superalloys
4.6.1 Introduction
Ni-base superalloys are useful at high temperature due to the fact that fcc metals tend to be more
stable than bcc metals and also Ni has a reasonably high melting point (1455C). Such alloys are
used extensively for turbine blades for aircraft engines which operate at gas temperatures up to
~1700C which is above the melting point of the alloy itself!
High operating temperatures increase engine efficiencies (Carnot cycle see Pt IA Course F) and
simultaneously reduce CO
2
emissions. Principal metallurgical concerns under these conditions are
creep resistance and oxidation resistance.
Creep deformation was introduced in Pt IA Course F and the mechanisms summarised on a
deformation-mechanism map. Two such maps based on nickel alloys are shown below.








Map for pure Ni with 100 m grain size Map for superalloy with 10 mm grain size
so similar to single crystal turbine blade

The contours on the maps indicate creep rates (strain rate in units of s
-1
). In the MARM200 nickel-
base superalloy
-
, it is evident that alloying and the absence of grain boundaries dramatically reduce
the creep rate for the operating conditions under which turbine blades in an aircraft engine are likely
to be used (indicated by the rectangular box in the figures).
The elimination of grain boundaries by casting single crystal blades (see AH48) reduces the
diffusion rates very considerably (see AH18).

-
Composition of MARM200 in weight % is 5 Al 2 Ti 9 Cr 12 W 10 Co 1 Nb 0.15 C balance Ni
AH99 Course A: Phase Transformations AH99


rob.wallach@msm.cam.ac.uk 2013-14
4.6.2 Ni (Al, Ti)
The major solutes for precipitate formation in Ni-based superalloys are Al and Ti, which are usually
present in concentrations of less than 10%. They form fine !' precipitates Ni
3
(Ti-Al).






For the above isothermal sections of the Ni-Ti-Al ternary phase diagram, consider the composition
marked by the red dot, which is ~ 90% Ni and approximately equal amounts of Al and Ti.
The single ! phase is formed by annealing at higher temperatures.
A very fine dispersion of !' precipitates in the ! matrix is obtained by cooling and ageing.
The ! phase is based on Ni and therefore has the fcc structure, with Al and Ti in solid solution
(occupying random substitutional sites on the lattice). The !' phase is approximately 75% Ni and is
an ordered phase of Ni
3
(Al,Ti) - i.e. Ni atoms occupy the face centre positions of the cell with Al or
Ti atoms at the corners. Hence this ordered structure is no longer cubic F, but cubic P as
introduced on page AH53 and shown below.





! phase disordered fcc !' phase ordered cubic P
In addition to solid solution strengthening from the alloy additions such as W, Co and Cr, the alloy
relies on the strengthening from these ordered !' phase precipitates. Dislocations in the ! phase
AH100 Course A: Phase Transformations AH100


rob.wallach@msm.cam.ac.uk 2013-14
find it difficult to penetrate into the ordered !' phase as that would upset the ordering (see order
strengthening, AH52). The passage of super dislocations, required to maintain order, is
considerably more difficult and the presence of partial dislocations (AH51) together result in the
high strength of the superalloys.

If the !' phase precipitates coarsened rapidly at the high operating temperatures, the alloy would
rapidly lose its strength. This does not occur since both ! and !' phases are cubic and have similar
lattice parameters so the !' precipitates grow coherently within the ! matrix with a simple cube-to-
cube orientation relationship. Because the !-!' interfacial energy is low, there is only a very minimal
driving force for coarsening (see Ostwald ripening, AH74) of the !' precipitates and the precipitates
remain small and evenly distributed even when used at the high operating temperatures (>1000C
in the body of the turbine blade).

Modern Ni-based superalloys are, in practice, far more complex than this 3 component system, as
shown by the composition of the MARM200 on the previous page. The micrograph below is of a
different superalloy containing Ni 53, Fe 19, Cr 18, Nb 5 (wt.% approximately) plus small amounts
of Ti, Mo, Co, Al. It contains three different precipitate phases which strengthen the ! matrix.
The Cr and Al additions also allow the formation on the alloys surface of a very stable and coherent
stable oxide. This minimises oxidation at elevated temperatures. Oxide formation is covered in
Course C.








Nickel-based superalloy IN718 held for 72 hours at 850C
(from DoITPoMS Micrograph library 720)

AH101 Course A: Phase Transformations AH101


rob.wallach@msm.cam.ac.uk 2013-14
5. SUSTAINABILITY OF MATERIALS
5.1 Life Cycle Analysis (LCA)
5.1.1 Stages of life
There are 4 stages in the life cycle of a manufactured product, as shown in the diagram below.








Each of these stages may involve stresses on the environment, to different extents, including:
energy consumption;
water consumption;
gas emissions;
particulates;
toxic residues.
It can be difficult to assess the levels of each of these stresses and to judge accurately the relative
importance of each, or to combine them to come up with a single overall index of sustainability.
Currently, single measures of environmental impact are widely used, e.g. based on
energy consumption or gas emissions (especially CO
2
).
Manufacturers increasingly are required to provide numeric data on energy usage and CO
2
emissions for making and using their products.
In practice, life cycle analysis, LCA, can produces similar results using one or both indices.
AH102 Course A: Phase Transformations AH102


rob.wallach@msm.cam.ac.uk 2013-14
When selecting and using materials, the four main phases of the product life cycle considered to be
important are:
production, manufacture, usage, disposal
In practice, the use of energy in one phase often predominates over the others, as shown for some
examples below. This suggests where attention should be focussed in order to decrease overall
energy usage.











5.1.2 Applications to manufactured products
As an example of how such energy life-cycle analysis can be carried out, consider the case of
drinks containers, which could easily be made using metals (either Al or steel), plastics (PET or PE)
or glass. A key factor in selecting the optimum material for this application is the amount of material
required to hold a litre of liquid (as is tabulated overleaf together with the energy needed to make
the mass shown), while other energy criteria might include:
production of raw material
manufacture
use
disposal
AH103 Course A: Phase Transformations AH103


rob.wallach@msm.cam.ac.uk 2013-14
Material Mass required (g)
(g)
Energy (MJ)
Aluminium 45 9.0
Steel 102 2.4
Glass 433 8.2
PET 62 5.3
PE 38 3.2
This approach allows identification of the material, from those listed, with the lowest energy impact.
Steel has the lowest impact for the drink can market while aluminium alloys also are widely used.


5.1.3 Energy Inputs for Metallic Products
This course has dealt with how metals are turned into useful products, by solidification processing
and also by deformation processing. A necessary prior stage, however, usually is the extraction of
the metal from its ore. An introduction to the thermodynamics underlying materials extraction will
be included in Part IB Course C (i.e. the Ellingham diagram).
Consider aluminium as an example. It is the most abundant metal on earth, found as the oxide
bauxite, but is relatively expensive due to the fact that it takes a large amount of energy to extract
the aluminium from bauxite. This involves removal of the ore from the ground followed by further
processing in order to achieve the pure oxide alumina (Al
2
O
3
). Thermodynamic calculations show
that the free energy change for the reaction:

3
2 2 3 2
2 Al O Al O + ! is -1576 kJ mol
-1
(at 298 K).
As also will be evident from the Ellingham diagram, this confirms that Al
2
O
3
is one of the most
stable metallic oxides and hence a large amount of energy is required to extract metallic Al.
Processing of metals also requires a significant amount of energy.
- casting clearly involves taking the alloy into its molten state a typical value of energy usage in
casting processes is around 3 MJ kg
-1

- deformation processing such as rolling or forging requires an energy input of around 6 MJ kg
-1
.
(Typical values for steel and Al from Cambridge Engineering Selector).
AH104 Course A: Phase Transformations AH104


rob.wallach@msm.cam.ac.uk 2013-14
See notes in section 5.3 on how to use Cambridge Engineering Selector software.
5.1.4 Reuse and recycling
Metals have a large advantage over other materials when it comes to reuse and recycling. Due to
their good mechanical properties (strong and ductile), products are generally durable and suitable
for reuse. Furthermore, even when used in disposable products, recycling is usually relatively
straightforward because metals (unlike polymers, currently) are generally easy to separate.
The ability to recycle has a clear impact on the life-cycle analysis that was carried out earlier. Using
an energy sustainability index, the dominant factor for both steel and aluminium in drink cans turns
out to be production of the raw material. Hence, if a significant fraction of aluminium cans are
recycled, the impact in terms of energy use is reduced significantly. This is the reason that
aluminium is able to compete with steel for this application.

5.2 Lifetime: degradation of metals
Most metals react chemically in their environments, as will be covered in Part IB Course B. Many
metals in service form oxide layers and their individual nature will dictate the extent to which they
can protect the metal from continued degradation. As will be seen in Course B, aluminium oxide
often acts to protect underlying alloy from further reaction, whereas the different nature of the
oxides on steels can lead to the familiar problems associated with continual oxidation and rusting.
Finally, note that the degradation of metals is not always problematic and can also be beneficial.
When a metal oxidises, some of the energy used in its initial production can be recovered. This is
the basis of many batteries.
AH105 Course A: Phase Transformations AH105


rob.wallach@msm.cam.ac.uk 2013-14
5.3 Notes on using Cambridge Engineering Selector software
Cambridge Engineering Selector (CES) software is available on the computers in the department.
It also can be downloaded for your own computer from:
www.msm.cam.ac.uk/teaching/ces.php

Instructions for use.
1. Open CES Edupack 2012.
2. The Configuration box appears; choose Level 2.
3. From the pull-down File menu, choose New Project.
4. Under 1. Selection Data Pick a selection template on the left-hand side of your screen, use the
pull-down menu and choose Edu level 2: Materials with Eco properties.
5. You then have several options, including plotting a graph, see (a) below, obtaining data on
individual materials as in (b) below, or finding out about fabrication processes as in (c) below .
a. Under 2. Selection Stages, you can choose Graph. This will bring up a dialogue box which
allows you to specify the data to be plotted on both x and y axes of your graph. If you are not
familiar with this, select the Help option at the left-hand bottom corner of the dialogue box and
information on the Graph Stage wizard will be shown. An example is provided over the page.
For the data needed for the answer on question sheet BQ4, the graph stage is not strictly
required.
b. On the left-hand side of the screen is a list of 97 different materials. Scroll down and then
double click on a material of interest to show the full data stored on that material.
c. To find out about different fabrication processes (also useful for the Artefact Project next term),
change the selection template (showing Edu level 2: Materials with Eco properties) in
Selection Data to Edu level 2: Processes Shaping and then choose individual processes
of interest from the 60 shown on the left-hand side of the screen.
6. The CES Help menu (top line) also provides full information on the software.


Background information on Eco Selection.

A full and helpful paper:
'The CES Eco Audit tool a white paper', Ashby M.F., Ball N., and Bream C.
on choosing materials on the basis of eco considerations can be downloaded by selecting the CES
Help menu. The Welcome page of CES Help has Contents displayed and the second bullet point is
White Papers. Choose this option and then Download the above paper (the PDF option does not
work in all situations). The paper will download in a new browser window.

[Note that the above paper provides information on transportation costs.]


Plotting a graph using CES see next page


AH106 Course A: Phase Transformations AH106


rob.wallach@msm.cam.ac.uk 2013-14
Plotting a graph using CES
1. Follow the instructions in 5(a) on the previous page.
2. In the pull-down menu Category, there are twelve distinct groups plus All - alphabetically, and
each Attribute has been put into one of these to aid their identification and selection. Hence
select a Category and then, using the pull-down list below, select the particular Attribute that
you wish to plot on the x-axis. Note that you can name the axis by entering data into the Axis
Title box.
3. Repeat this for the y-axis by selecting the y-axis tab at the top of the selection box.
4. To plot the graph, choose OK at the bottom of the dialogue box
5. Additional options.
a. It is possible to combine Attributes by using the Advanced button. Select the first Attribute
and one of the various operators listed, followed by the second Attribute.
b. In addition, the selection of all materials in one or more of four main classes can be made
using the Advanced button. This allows the relative properties of the Attribute selected for
the y-axis to be compared for the material class/es selected. This is most conveniently done
by displaying the family along the x-axis. Hence select the x-axis tab followed by Advanced
and then the Trees tab. Select one or more of the four families in turn by either highlighting
the family and using Insert, or by double-clicking on the family. Choose the Attribute for the
y-axis as in 3 above and then plot the graph as in 4 above.


AH107 Course A: Phase Transformations AH107


rob.wallach@msm.cam.ac.uk 2013-14
GLOSSARY
Activity Chemical activity generally is proportional to the concentration of a substance taking part
in a reaction and the activity is a measure of this, often linearly proportional to concentration.
Ageing The heat-treatment of a supersaturated alloy to promote precipitation, hence at a
temperature below the solvus on the relevant equilibrium phase diagram.
Anneal Heat a work-hardened alloy to induce recovery and recrystallisation.
Annealing Twins Those which arise as a result of a growth accident during the recrystallisation of
cubic close-packed materials such as %-brass, copper or nickel. They have flat edges.
Anti-Phase Boundary The boundary between anti-phase domains.
Anti-Phase Domain In an ordered alloy, a region where the ordering is in a different sequence.
Atomic Flux The number of atoms crossing unit area in unit time.
Bainite A phase produced in steels by a combination of shear and diffusional transformations.
Binary alloy One consisting of 2 components or phases.
Biot Number A dimensionless quantity which indicates whether heat extraction from a casting is
dominated by conduction across the mould/casting interface or by conduction through the casting.
Chemical Potential The amount by which the free energy of a system is altered by a change in
the concentration of a species, the temperature, pressure, etc.
Constitutional undercooling Undercooling (or supercooling) arising due to variations in
composition near a solidifying interface.
Deformation Twins (also Mechanical Twins) These arise within a single grain in a material in
which there are insufficient slip systems to allow plastic deformation by dislocation movement. The
need to minimise the strain energy at grain boundaries results in the twins having a lenticular shape.
Error Function A mathematical function which describes the solution to the diffusion equation,
Ficks second law, when one end of a semi-infinite bar is maintained at a constant concentration.
Eutectic A phase transformation at a given temperature in which liquid . solid
1
+ solid
2

Ficks 1
st
Law describes diffusion when concentration is independent of time, i.e. a constant
concentration gradient.
AH108 Course A: Phase Transformations AH108


rob.wallach@msm.cam.ac.uk 2013-14
Ficks 2
nd
Law describes diffusion when the concentration changes with time.
Grain Growth Following recrystallisation, a mechanism by which the overall energy of a material
can be decreased by the reduction in grain boundary area.
GP Zones The first metastable precipitates that form when ageing an Al-Cu alloy.
Interdiffusion Diffusion of an atomic species into a material comprising different atomic species.
Kirkendall Experiment An experiment which shows that diffusion in solids happens by a vacancy
mechanism.
Mechanical Twins see Deformation Twins.
Martensite. A metastable supersaturated solid-solution phase in steels formed by a displacive
(shear) transformation when quenching austenite.
Microsegregation Segregation of solute on a microscopic scale.
Normalise Heat a steel above A3 temperature i.e. just into the region, hold and then cool but do
not quench. Leads to significant grain refinement and improvements in mechanical properties.
Ostwald Ripening The growth of large precipitates at the expense of smaller ones to reduce total
precipitate surface area and hence lower the overall energy of a material.
Oxide Dispersion Strengthening The introduction of very fine, stable oxide powders into the
microstructure to prevent or minimise Ostwald ripening.
Partition Coefficient The ratio (at equilibrium) of the composition of solute in a solid to that in the
liquid with which it coexists at a given temperature.
Partitioning The preferential distribution of solute (impurities) into one phase rather than another.
Peritectic A phase transformation at a given temperature in which solid
1
+ liquid . solid
2

Precipitate Free Zones Regions surrounding grain boundaries which often contain many fewer
precipitates than the remainder of the material.
Quench Rapid cooling from higher temperatures by immersion of an alloy in a light oil or water so
that there is insufficient time for equilibrium phase transformations.
Recovery The process by which a cold-worked metal rearranges its dislocation structure at around
0.3 T
m
to form new sub-grains (the overall dislocation density is virtually unchanged).
AH109 Course A: Phase Transformations AH109


rob.wallach@msm.cam.ac.uk 2013-14
Recrystallisation Following recovery, the formation of new, strain-free, equiaxed grains at a
temperature of ~ 0.6-0.7 T
m
(the driving force is a substantial reduction in dislocation density).
Scheil Equation describes the composition profile that forms in the solid if there is no diffusion in
the solid but perfect mixing in the liquid during solidification.
Secondary Hardening is the formation of stable carbides in a martensitic steel containing carbide-
forming alloying elements when tempering at around 600C.
Segregation The non-uniform distribution of solute in a material which arises via partitioning.
Self Diffusion Diffusion of an atomic species within a material composed of that atomic species.
Solute Impurity atoms in solution in an alloy.
Solute Drag The mechanism by which grain boundary movement is restricted by solute atoms in a
solid solution.
Stacking Fault A region in which the atomic stacking sequence is disrupted.
Supercooling See undercooling.
Superdislocation A combination of two dislocations which travel together to minimise lattice
disorder.
Superheating is the amount by which the temperature of a liquid (usually in a casting) is above the
equilibrium melting/fusion temperature.
Supersaturated solid solution is formed by cooling a binary alloy below its solvus at a rate fast
that enough there is insufficient time for the second phase to precipitate out from the matrix.
Ternary Alloy One consisting of three components.
Temper Heat a steel which has been quenched in order to form equilibrium phase.
Undercooling of a liquid is when its actual temperature is below the equilibrium melting/fusion
temperature (also called supercooling). Can promote formation of dendrites.
Zener Drag (or Zener Pinning) The mechanism by which grain boundary movement is restricted by
particles or precipitates embedded in the matrix.
AH110 Course A: Phase Transformations AH110


rob.wallach@msm.cam.ac.uk 2013-14
Lecture 1






Lecture 2






Lecture 3






AH111 Course A: Phase Transformations AH111


rob.wallach@msm.cam.ac.uk 2013-14
Lecture 4






Lecture 5






Lecture 6






Lecture 7

AH112 Course A: Phase Transformations AH112


rob.wallach@msm.cam.ac.uk 2013-14





Lecture 8






Lecture 9






AH113 Course A: Phase Transformations AH113


rob.wallach@msm.cam.ac.uk 2013-14
Lecture 10






Lecture 11






Lecture 12

Вам также может понравиться