1 Thermodynamic functions

1.1 INTRODUCTION
Thermodynamics facilitates the linking together of “the many observable properties so that they can
be seen to be a consequence of a few” [1]. It provides a firm basis for the rules that macroscopic
systems follow at equilibrium. When combined with phenomena associated with the approach to
equilibrium, it forms the foundations of kinetic theory. It was on this basis that Zener attempted
to rationalise the transformations that occur in steels [2, 3] so that the effect of alloying elements,
atomic mobility, nucleation and mechanism could all be incorporated into a single hypothesis.
After an introduction to some essential concepts, the remainder of this chapter deals with theory
that is relevant particularly to iron and its solutions.

1.2 DEFINITIONS
1.2.1 INTERNAL ENERGY AND ENTHALPY
The change in the internal energy ∆U of a closed system can be written as

∆U = q − w (1.1)

where q is the heat transferred into the system and w, the work done by the system. The sign
convention is that heat added and work done by the system are positive, whereas heat given off and
work done on the system are negative. Equation 1.1 may be written in differential form as

dU = dq − dw. (1.2)

For the special case where the system does work against a constant atmospheric pressure, this be-
comes
dU = dq − PdV (1.3)

where P is the pressure and V the volume.

The specific heat capacity of a material represents its ability to absorb or emit heat during a unit
change in temperature. Heat changes the distribution of energy amongst the particles in the system
(atoms, electrons, . . .) and it is these fundamental mechanisms that control the heat capacity, defined
formally as dq/dT . Since dq = dU + PdV , the specific heat capacity measured at constant volume
is given by:
 
∂U
CV = .
∂T V
It is convenient to define a new function H, the enthalpy of the system:

H = U + PV.

A change in enthalpy accounts for both the heat absorbed at constant pressure, and the work done
by the P∆V term. The specific heat capacity measured at constant pressure is therefore given by:
 
∂H
CP = .
∂T P

1
2 Course on thermodynamics

Heat capacity can be measured using a variety of calorimetric methods. The data can then be used
to estimate enthalpy changes as a function of temperature and pressure:

ZT2
∆H = CP dT. (1.4)
T1

1.2.2 ENTROPY, FREE ENERGY

In the reversible Carnot cycle, a gas is placed in contact with a heat reservoir at temperature T2 ,
expands isothermally on absorbing a quantity of heat q2 , in the process doing work −w1 . The gas
is then insulated and expands adiabatically, does work −w2 as its temperature drops to T1 . It is then
placed in contact with a heat reservoir at T1 , compressed reversibly and isothermally with work w3
done upon it and giving up heat −q1 to the reservoir. To complete the cycle, the gas is insulated,
compressed reversibly and adiabatically by doing work w4 upon it, causing its temperature to rise
back to T2 [4]. The change in internal energy ∆U = q1 + q2 + w on completion of the cycle is
therefore zero, where w = Σ4i=1 wi . The work output of the engine, −w, is the difference between
heat taken and heat returned to the reservoirs, i.e. −w = q2 − (−q1 ) so the maximum efficiency is
defined as the ratio of the work output to the heat absorbed:
−w q2 + q1
efficiency = = .
q2 q2

Kelvin used this to define the absolute temperature,

q2 + q1 T2 − T1 q2 q1
efficiency = = so that + = 0. (1.5)
q2 T2 T2 T1

By considering a cyclic process in terms of infinitesimal parts [4], it can be demonstrated that the
following relationship holds for any reversible cycle,
I
dq2 dq1 dq
+ =0 with =0
T2 T1 T

making the quantity dq/T a function of state S with

dq
dS = (1.6)
T
Clausius during the 19th century named this function S as entropy; in the absence of any change
in enthalpy, a reaction can occur spontaneously and irreversibly in an isolated system if it leads
to an increase in entropy, i.e., ∆S > 0. It is evident that in general, neither the enthalpy nor the
entropy change can in isolation be assumed to reliably indicate the whether a reaction can occur
spontaneously. The Gibbs free energy G is therefore defined as a combination of these two terms,

G = H − T S. (1.7)

The Helmholtz free energy F is the corresponding term at constant volume, when H is replaced by
U in Equation 1.7. A process can occur spontaneously if it leads to a reduction in the free energy.
Quantities such as H, G and S are all functions of state.
From an experimental perspective, a change in entropy can be measured via the heat capacity:
Z T2
CP
∆S = dT.
T1 T
Thermodynamic functions 3

1.2.3 CONFIGURATIONAL ENTROPY

Figure 1.1a shows a mixture of two kinds of atoms, with like atoms segregated with no mixing; there
is only one way of achieving this arrangement. On the other hand, if they are allowed to mix ideally
then there are many more ways of configuring them, three of which are illustrated in Figure 1.1c-d.
A mixing of the atoms is obviously more probable.

(a) (b)

(c) (d)

Figure 1.1 Four different configurations of a mixture of two kinds of atoms. (a) The two kinds of atoms
are partitioned into their own spaces, without mixing. (b-d) If the atoms are allowed to mix then many more
arrangements are possible, here only three of the many are illustrated.

Suppose there are N sites amongst which are distributed n atoms of type A and N − n of type B.
The first A atom can be placed in N different ways and the second in N − 1 different ways. These
two atoms cannot be distinguished so the number of different ways of placing the first two A atoms
is N(N − 1)/2. The number of distinguishable ways of placing all the atoms in this way, the number
of distinguishable ways of placing all the A atoms is

N(N − 1) . . . (N − n + 2)(N − n + 1) N!
= . (1.8)
n! n!(N − n)!

So if the atoms behave ideally, i.e., they do not have a preference for the type of neighbour, then the
probability of a uniform distribution is much much more likely than the ordered distribution.
For a real system for which the number of atoms is very large, a parameter is needed that ex-
presses the likelihood as a function of the correspondingly large number of configurations (wc )
possible. Suppose that a term S is defined such that S ∝ ln wc , where the logarithm is taken because
it may be necessary to add two different kinds of disorder (after Boltzmann), then the S is identi-
fied as the configurational entropy S = k ln wc , where k, the proportionality constant, is known as
the Boltzmann constant which for a mole of atoms is the gas constant R. The entropy is a thermo-
dynamic function of state and it is additive. When comparing scenarios, the one that is favoured
on the basis of the degree of disorder is that which has the greater entropy. In terms of solutions,
entropy favours mixing over separation. On this basis, it can be shown quite simply that the change
in entropy when atoms mix is given by

j
∆S = −R Σi=1 xi ln xi

where i = 1 . . . j represents the atomic species and xi its mole fraction.

4 Course on thermodynamics

1.2.4 RELATIONSHIP BETWEEN CLAUSIUS AND BOLTZMANN ENTROPIES

The Carnot engine illustrates the spreading of energy whereas the Boltzmann approach is about
mixing. The relationship between these is quite straightforward – both involve mixing. As Denbigh
stated with admirable elegance, “As soon as it is accepted that matter consists of small particles
which are in motion it becomes evident that every large-scale natural process is essentially a process
of mixing” [p. 110, 5]. Energy transfer involves the motion of the atoms; it was not, at the time of
Carnot, known that matter consists of atoms. It really isn’t necessary to say much more.

1.3 MORE FUNCTIONS

We have seen that heat capacity data can be related directly to the thermodynamic functions of
state H, G and S. Its variation with temperature and chemical composition is therefore important in
determining the relative stabilities of the phases. A number of factors can contribute independently
to the ability of a material to absorb energy. It has been found useful to factorise the specific heat
capacities of each phase in iron, into three components with different origins.
The bulk of the contribution comes from lattice vibrations, the electrons themselves contributing
in a relatively minor way because the Pauli exclusion principle prevents all of them from partici-
pating in the energy absorption process. The third contribution, which is particularly significant for
iron, comes from temperature-induced magnetic changes. The net specific heat capacity at constant
pressure is therefore:  
L TD
CP {T } = CV C1 + Ce T + CPµ {T }
T
where TD is the Debye temperature and CVL is the Debye specific heat with the function C1 correcting
CVL to a specific heat at constant pressure. Ce is the electronic specific heat coefficient and CPµ the
component due to magnetism.
The Debye specific heat has its origins in the vibrations of atoms, which become increasingly
violent as the temperature rises [6]. These elastic waves (phonons) take discrete, quantised wave-
lengths consistent with being bound by the periodic lattice of atoms in the solid, although the Debye
model described here is a continuum model. The atoms do not all vibrate with the same frequency,
so a spectrum of vibrations is considered in deriving their contribution to the internal energy U.
The maximum in the spectrum is designated the Debye frequency ωD , which is proportional to the
Debye temperature TD at which the highest frequency mode is excited:

h ωD
TD =
2π k
where h and k are the Planck and Boltzmann constants respectively. With the approximation that the
phonon frequency is proportional inversely to the wavelength, the internal energy due to the atom
vibrations is:
9NkT 4 xmax x3
Z
U= dx (1.9)
TD3 0 (ex − 1)
where x = hωD /(2π kT ) and N is the total number of lattice points in the specimen. Since CVL =
dU/dT , it follows that the lattice specific heat capacity at constant volume can be specified in terms
of the Debye temperature and the Debye function (equation 1.9).
At low temperatures (T ≪ TD ), U → 3NkT 4 π 4 /(5TD3 ) so that CVL → 12π 4NkT 3 /(5TD3 ) and the
lattice specific heat thus follows a T 3 dependence. For T ≫ TD , the lattice heat capacity can similarly
be shown to become temperature independent and approach a value 3Nk, as might be expected for
N classical oscillators, each with three degrees of freedom (Figure 1.2).
Thermodynamic functions 5

Figure 1.2 The Debye function showing how the

heat capacity due to phonons varies as a function of the
absolute temperature normalised by the Debye temper-
ature.
0 1 2
 2 Equilibrium state
2.1 CHEMICAL POTENTIAL
In a single-phase equilibrium diagram such as that for iron as a function of temperature and pressure,
the boundaries between the phase fields represent the locus of all points along which the adjacent
phases are in equilibrium, i.e., they have an identical free energy. For example, the α/γ phase
boundary is defined by setting (Figure 2.1):
Gα = Gγ . (2.1)
This is because allotropic transitions are considered here as a function of variables such as temper-
ature and pressure, where the crystal structure changes but not the chemical composition.

transition
temperature
Free energy

ferrite Figure 2.1 The transition temperature for

an allotropic transformation.

austenite

Temperature

A different approach is needed when the chemical composition is variable. Consider a single-
phase alloy consisting of two components A and B. The molar free energy G{x} of that phase will
in general be a function of the mole fractions (1 − x) and x of A and B respectively, written as a
weighted mean of the free energy contributions from each component:
G{x} = (1 − x)µA + x µB . (2.2)
| {z } |{z}
contribution from A atoms contribution from B atoms
The terms µB and µB , known as the chemical potentials per mole of A and B respectively, in effect
partition the free energy G{x} into a component purely due to A atoms and another due to B atoms
alone. This equation is illustrated in Figure 2.2 by the tangent at the coordinate [G{x}, x]. Consistent
with Equation 2.2, the intercepts of this tangent on the vertical axes give µA and µB . Since the slope
of the tangent depends on the composition, so do the chemical potentials. Note that the free energies
of the pure components are written µA◦ and µB◦ .
It should be obvious from Figure 2.2 that
∂G
µA = G{x} − x
∂x
∂G
and µB = G{x} + (1 − x)
∂x
where ∂ G/∂ x is the slope of the tangent so the product on the right-hand side of the equations
simply represents the difference in µ and G. In general, for a system with n components [97, p. 57]:
n
∂G
µi = G{xi } + ∑ (δi j − x j ) (2.3)
j=2 ∂xj

7
8 Course on thermodynamics

where δi j is the Kronecker delta (δi j = 0 for i 6= j and δi j = 1 for i = j).

Figure 2.2 Illustration of the chem-

ical potential µ for a binary solution,
with µ ◦ representing the free energy of
the pure component.

The chemical potential µ {x} of a component is known also as its partial molar free energy,
describing a part of the integral molar free energy G{x}. There are in fact many quantities which
can be expressed using relationships of the form implied by Equation 2.2. Thus, the volume of a
solution might be written in terms of the partial molar volumes of the components:

Vm = V A xA + V B xB (2.4)

where V i refers to the partial molar volume of component i =A,B.

2.2 EQUILIBRIUM BETWEEN SOLUTIONS

Consider now two phases α and γ that are placed in intimate contact in a binary steel. The phases
will only be in equilibrium with each other if the carbon atoms in γ have the same chemical potential
as the carbon atoms in α, and if this is true also for the Fe atoms:

µCα = µCγ
γ
µFe
α
= µFe . (2.5)

In fact, in a binary solution, the chemical potentials of A and B when sharing a tangent are not
independent so this last condition is redundant. This is apparent from Figure 2.2, where the two
potentials are connected by the tangent.
If the atoms of a particular species have the same chemical potential in both the phases, then
there can be no tendency for them to migrate across the phase boundaries. The system will be in
stable equilibrium if this condition applies to all species of atoms. The way in which the free energy
of a phase varies with concentration is unique to that phase, so the concentration of a particular
species of atom need not be identical in phases which are at equilibrium. Therefore, in general,

xαγ
C 6 = xγα
C
xαγ
Fe 6= xγα
Fe (2.6)

where xαγ
i describes the mole fraction of element i in phase α which is in equilibrium with phase γ
etc.
The condition that the chemical potential of each species of atom must be the same in all phases
at equilibrium is general. For the binary alloy, two phase case, it follows that the equilibrium com-
positions can be found on a plot of free energy versus composition, by constructing a tangent that is
common to the two free energy curves as illustrated in Figure 2.3.
Equilibrium state 9

Figure 2.3 The common tangent

construction giving the equilibrium
compositions xαγ and xγα of the two
phases at a fixed temperature.

2.3 ACTIVITY
The chemical potential µAα of the A atoms in the α phase may be expanded in terms of a contribution
from the pure component A and a concentration dependent term as follows:

µAα = µA◦α + RT ln aα
A (2.7)

where µA◦α is the free energy of pure A in the structure of α, and aA is the activity of atom A in the
solution of A and B.
The activity of an atom in a solution can be thought of as its effective concentration in that
solution. For example, there will be a greater tendency for the A atoms to evaporate from solution,
when compared with pure A, if the B atoms repel the A atoms. The effective concentration of A in
solution will therefore be greater than implied by its atomic fraction, i.e., its activity is greater than
its concentration. The opposite would be the case if the B atoms attracted the A atoms.
The atom interactions can be expressed in terms of the change in energy as an A-A and a B-B
bond is broken to create 2(A-B) bonds. An ideal solution is formed when there is no change in
energy in the process of forming A-B bonds. The activity is equal to the mole fraction in an ideal
solution (Figure 2.4). If, on the other hand, there is a reduction in energy than the activity is less
than ideal and vice versa. The activity and concentration are related via an activity coefficient Γ:

a = Γx. (2.8)

The activity coefficient is in general a function of the chemical composition of all the elements
present in the solution, but tends to be constant in dilute solutions (i.e., in the Henry’s law region).
In this discussion, the activity of the solute was defined with respect to a Raoultian reference
state, i.e., a = 1 for x = 1. Other definitions are sometimes convenient. A common alternative for
dilute solutions being that the activity tends to unity as the concentration tends to 1 wt%.
Note that solutions where the enthalpy of mixing is positive tend to exhibit clustering at low
temperatures whereas those with a negative enthalpy of mixing will tend to exhibit ordering at low
temperatures. The effect of temperature is to mix all atoms since both clustering and ordering cause a
reduction in entropy (i.e., a reduction in entropy). The product −T ∆S becomes increasingly positive
at high temperatures, so much so that it eventually overcomes the enthalpy effects and causes the
mixing of all atoms.
10 Course on thermodynamics

Figure 2.4 Variation in Raoultian activity as a

function of its concentration in a binary solution.
The ideal solution represents the case where the
enthalpy of mixing is zero; the atoms are indif-
ferent to the specific nature of their neighbours.
The case where the activity is larger than the con-
centration is for solutions where the enthalpy of
mixing is greater than zero, with like atoms pre-
ferred as near neighbours. When the activity co-
efficient is less than unity, unlike atoms are pre-
ferred as near neighbours, the enthalpy of mixing
being negative.

2.4 IDEAL SOLUTION

An ideal solution is one in which the atoms at equilibrium are distributed randomly; the interchange
of atoms within the solution causes no change in the potential energy of the system. For a binary
(A-B) solution the numbers of the different kinds of bonds can therefore be calculated using simple
probability theory:
1
NAA = N(1 − x)2
2
1 2
NBB = Nx
2
NAB = N(1 − x)x

where NAB represents both A-B and B-A bonds which cannot be distinguished. N is the total number
of atoms and x the fraction of B atoms. The factor of 12 avoids counting A-A or B-B bonds twice.
For an ideal solution, the entropy of mixing is given by Equation 3.3 with mB = mA = 1. There
is no enthalpy of mixing since there is no change in energy when bonds between like atoms are
broken to create those between unlike atoms. This is why the atoms are randomly distributed in the
solution. The molar free energy of mixing is therefore:

∆GM = Na kT [(1 − x) ln{1 − x} + x ln{x}]. (2.9)

Figure 2.5 shows how the configurational entropy and the free energy of mixing vary as a function of
the concentration. ∆GM is at a minimum for the equiatomic alloy because that is when the entropy
of mixing is at its largest; the curves are naturally symmetrical about x = 0.5. The form of the
curve does not change with temperature though the magnitude at any concentration scales with the
temperature. It follows that at 0 K there is no difference between a mechanical mixture and an ideal
solution.
From Equation 2.7, the chemical potential per mole for a component in an ideal solution is given
by:
µA = µA◦ + Na kT ln{1 − x}
and there is a similar equation for B. Since µA = µA◦ + RT ln aA , it follows that the activity coefficient
is unity.
Equilibrium state 11

Figure 2.5 The entropy of mixing

(kJ mol−1 K−1 ) and the free energy of
mixing (kJ mol−1 ) as a function of
concentration in an ideal binary solu-
tion where the atoms are distributed at
random. The free energy is for a tem-
perature of 1000 K. The data are plot-
ted as dots rather than curves because
concentration is strictly a discrete vari-
able. So the slope at the vertical axes is
not ±∞ as implied by Equation 2.9, but
finite though very large.

2.5 REGULAR SOLUTIONS

There are no solutions of iron that are ideal. The iron-manganese liquid phase is close to ideal,
though even that has an enthalpy of mixing which is about −860 J mol−1 for an equiatomic solu-
tion at 1000 K, which compares with the contribution from the configurational entropy of about
−5800 J mol−1. The ideal solution model is nevertheless useful because it provides a reference. The
free energy of mixing for a non-ideal solution often is written with an additional term, the excess
free energy (∆e G = ∆e H − T ∆e S) that indicates the deviation from ideality:

∆GM = ∆e G + Na kT [(1 − x) ln{1 − x} + x ln{x}]

= ∆e H − T ∆e S + Na kT [(1 − x) ln{1 − x} + x ln{x}] (2.10)

One of the components of the excess enthalpy of mixing comes from the change in the energy when
new kinds of bonds are created during the formation of a solution. This enthalpy is, in the regular
solution model, estimated from the pairwise interactions between adjacent atoms. The term regular
solution was proposed by Hildebrand [98] to describe mixtures, the properties of which when plotted
varied in an aesthetically regular manner; he went on to suggest that a regular solution, although not
ideal, would still contain a random distribution of the constituents.1 Following Guggenheim [99],
the term regular solution is now restricted to cover mixtures that assume an ideal entropy of mixing
but have a non-zero interchange energy.
In the regular solution model, the enthalpy of mixing is obtained by counting the different kinds
of near neighbour bonds when the atoms are mixed at random; this information together with the
binding energies gives the required change in the enthalpy on mixing. The binding energy may be
defined by considering the change in energy as the distance between a pair of atoms is decreased
from infinity to an equilibrium separation (Figure 2.6). The change in energy during this process is
the binding energy, which for a pair of A atoms is written −2εAA . It follows that when εAA + εBB <
2εAB , the solution will have a larger than random probability of bonds between unlike atoms. The
converse is true when εAA + εBB > 2εAB since atoms then prefer to be neighbours to their own kind.
Notice that for an ideal solution it only is necessary for εAA + εBB = 2εAB , and not εAA = εBB = εAB
[99].
Suppose now that the approximation that atoms are randomly distributed is retained, even though
the enthalpy of mixing is not zero. The number of A-A, A-B and B-B bonds in a mole of solution
is then 12 zNa (1 − x)2 , 21 zNa x2 and zNa (1 − x)x respectively, where z is the co-ordination number. It
12 Course on thermodynamics

Figure 2.6 Change in energy as a

function of the distance between a pair
of A atoms. −2εAA is the binding en-
ergy for the pair of atoms. There is a
strong repulsion at close-range.

follows that the molar enthalpy of mixing is given by:

∆HM ≃ Na z(1 − x)xω where ω = εAA + εBB − 2εAB . (2.11)
The product zNa ω is often called the regular solution parameter, which in practice will be temper-
ature and composition dependent. A composition dependence also leads to an asymmetry in the
enthalpy of mixing as a function of composition about x = 0.5. For the nearly ideal Fe-Mn liquid
phase solution, the regular solution parameter is −3950 + 0.489T J mol−1 if a slight composition
dependence is neglected.
A positive ω favours the clustering of like atoms whereas when it is negative there is a tendency
for the atoms to order. This second case is illustrated in Figure 2.7, where an ideal solution curve is
presented for comparison. Like the ideal solution, the form of the curve for the case where ∆HM < 0
does not change with the temperature, but unlike the ideal solution, there is a free energy of mixing
even at 0 K where the entropy term ceases to make a contribution.

Figure 2.7 Free energy of mixing at

1000 K, as a function of concentra-
tion in a binary solution where there
is a preference for unlike atoms to be
near neighbours. The free energy curve
for the ideal solution (∆HM = 0) is in-
cluded for reference.

The corresponding case for ∆HM > 0 is illustrated in Figure 2.8, where the form of the curve
is seen to change with the temperature. The contribution from the enthalpy term can largely be
neglected at high temperatures where the atoms become randomly mixed by thermal agitation so the
free energy curve then has a single minimum. However, as the temperature is reduced, the opposing
contribution to the free energy from the enthalpy term introduces two minima at the solute-rich and
solute-poor concentrations. This is because like-neighbours are preferred. On the other hand, there
is a maximum at the equiatomic composition because that gives a large number of unfavoured unlike
atom bonds. Between the minima and the maximum lie points of inflexion which are of importance
in spinodal decomposition, to be discussed later.
For a regular solution, Equation 2.2 shows that the chemical potential per mole is given by:
µB = µA◦ + zNa x2 ω + Na kT ln{1 − x} (2.12)
and that the activity coefficient is exp{zx2 ω /kT }. Some of the properties of the different kinds of
solutions are summarised in Table 2.1.
Equilibrium state 13

Figure 2.8 Free energy of mixing as

a function of concentration and tem-
perature in a binary solution where
there is a tendency for like atoms to
cluster. The free energy curve for the
ideal solution (∆HM = 0) is included
for reference.

Table 2.1
Elementary thermodynamic properties of solutions

Type ∆SM ∆HM

Ideal Random 0
Regular Random 6 0
=
Quasichemical Not random 6= 0
 3 Case study: mechanical
alloying
All steels are solutions. The distinction between a compound and a solution is that the free energy of
the former increases sharply with a change in its chemical composition. For a solution, the variation
in free energy with composition is much more gentle so that the range of composition over which
it can exist is greater. It is useful to examine the nature of a solution by considering its evolution as
the components are mixed together. For when is a mixture, a solution?

3.1 ALLOYING BY DEFORMATION

Mechanical alloying is a process invented by Benjamin [95], in which mixtures of fine powders
consisting of elemental metals or master alloys are changed into solid solutions, apparently without
any melting (Figure 3.1). The powders are forced to collide with each other and with much larger,
hardened steel balls whilst contained in a ball mill. The collisions are energetic, involve large contact
pressures, and lead eventually to the formation of an intimate solid solution. Refractory oxides can
also be introduced into the mechanically alloyed powder for dispersion strengthening. The alloyed
powder is finally extruded to form full density bulk samples in rod, sheet or other useful shapes. The
process has been used commercially to make iron alloys containing large amounts of aluminium for
oxidation resistance, and yttria particles to guard against creep.

Figure 3.1 Mixture of metallic

powders and compounds ball-milled
together until alloying occurs. The top
part shows a cylindrical drum contain-
ing a mixture of elemental powders
and large steel balls. When the drum
is rotated the balls collide causing the
powder particles to coalesce and frag-
ment repeatedly. The resulting pow-
ders are then canned and hot-extruded
to produce solid metal.

3.2 CHEMICAL STRUCTURE

An alloy can be created without melting, by violently deforming mixtures of different powders. The
intense deformation associated with mechanical alloying can force atoms into positions where they
may not prefer to be at equilibrium.
A solution which is homogeneous will nevertheless exhibit concentration differences of increas-
ing magnitude as the size of the region which is chemically analysed decreases. These are random
fluctuations which obey the laws of stochastic processes, and represent the real distribution of atoms
in the solution. These equilibrium variations cannot usually be observed directly because of the lack
of spatial resolution and noise in the usual microanalytical techniques. The fluctuations only become
apparent when the resolution of chemical analysis falls to less than about a thousand atoms block.

15
16 Course on thermodynamics

Figure 3.2 illustrates the variation in the iron and chromium concentrations in fifty atom blocks,
of the ferrite in a commercial alloy. There are real fluctuations but further analysis is needed to show
whether they are beyond what is expected in homogeneous solutions

Figure 3.2 The variation in the iron and chromium concentrations of 50 atom samples of an alloy

For a random solution, the distribution of concentrations should be binomial since the fluctua-
tions are random; any significant deviations from the binomial distribution would indicate either the
clustering of lik–atoms or the ordering of unlike pairs.
The frequency distribution is obtained by plotting the total number of composition blocks with a
given number of atoms of a specified element against the concentration. Figure 3.3 shows that the
experimental distributions are essentially identical to the calculated binomial distributions, indicat-
ing that the solutions are random.

Figure 3.3 Frequency distribution curves for iron, chromium and aluminium in a mechanical alloy.

This does not mean that the solutions are thermodynamically ideal, but rather that the alloy
preparation method which involves intense deformation forces a random dispersal of atoms. Indeed,
Fe–Cr solutions are known to deviate significantly from ideality, with a tendency for like atoms to
cluster. Thus, it can be concluded that the alloy is in a mechanically homogenised nonequilibrium
state, and that prolonged annealing at low temperatures should lead to, for example, the clustering
of chromium atoms.
Case study: mechanical alloying 17

3.3 SOLUTION FORMATION

The preparation of a binary alloy by this route can be considered in terms of the two elemental
powders (‘A’ and ‘B’) which are mixed such that the mole fraction of B is x. The pure powders
have the molar free energies µA◦ and µB◦ respectively, Figure 3.4. The free energy of this mechanical
mixture of powders is given by:

G{mixture} = (1 − x)µA◦ + xµB◦ + ∆SM (3.1)

where ∆SM is the accompanying change in configurational entropy. It has been assumed here, that
there is no change in enthalpy in the process, i.e., the atoms in the context of bonding, are indifferent
to the type of neighbouring atom.

Figure 3.4 Plot of free energy ver-

sus composition, both for mechanical
mixtures and a solid solution. ∆GM is
the free energy of mixing when the
mechanical mixture turns into a solid
solution.

The change in configurational entropy as a consequence of mixing can be obtained using the
Boltzmann equation S = k ln{wc } where wc is the number of configurations. Suppose that there are
mA atoms per powder particle of A, and mB atoms per particle of B; the powders are then mixed in
a proportion which gives an average concentration of B which is the mole fraction x.
There is only one configuration when the heaps of powders are separate. When the powders are
mixed randomly, the number of possible configurations for a mole of atoms becomes [96]:


Na ([1 − x]/mA + x/mB ) !
. (3.2)
(Na [1 − x]/mA )! (Na x/mB )!
The numerator in equation 3.2 is the total number of particles and the denominator the product
of the factorials of the A and B particles respectively; Na is Avogadro’s number. Using Stirling’s
approximation, the molar entropy of mixing is:
 
∆SM (1 − x)mB + xmA (1 − x)mB + xmA
= ln Na
kNa mA mB mA mB
 
1−x Na (1 − x)
− ln
mA mA
 
x Na x
− ln (3.3)
mB mB

subject to the condition that the number of particles remains integral and non-zero.1
The largest reduction in free energy occurs when the particle sizes are atomic, Figure 3.5, which
shows the molar free energy of mixing for a case where the average composition is equiatomic.
18 Course on thermodynamics

Such a composition maximises configurational entropy. When it is considered that phase changes
often occur at appreciable rates when the accompanying reduction in free energy is just 10 J mol−1 ,
Figure 3.5 shows that the entropy of mixing cannot be ignored when the particle size is less than a
few hundreds of atoms. In commercial practice, powder metallurgically produced particles are typi-
cally 100 µm in size, in which case the entropy of mixing can be entirely neglected, though solution
formation must be considered to be advanced when the processing reduces particle dimensions to
some 102 atoms. These comments must be qualified due to the neglect any enthalpy change during
mixing.

Figure 3.5 The molar Gibbs free energy

of mixing, ∆GM = −T ∆SM , for a binary al-
loy, as a function of the particle size when
all the particles are of uniform size in a mix-
ture, the average composition of which is
equiatomic. T = 1000 K.

3.4 ENTHALPY OF MIXING FOR PARTICULATE MIXTURES

The enthalpy of mixing will not in general be zero as was assumedabove. Equation 2.11 gives the
molar enthalpy of mixing for atomic solutions. For particles which are not monatomic, only those
atoms at the interface between the A and B particles will feel the influence of the unlike atoms. It
follows that the enthalpy of mixing is not given by Equation 2.11, but rather by

∆HM = zNa ω 2δ SV x(1 − x)

where SV is the amount of A-B interfacial area per unit volume and 2δ is the thickness of the
interface, where δ represents a monolayer of atoms.
A further enthalpy contribution, which does not occur in conventional solution theory, is the
structural component of the interfacial energy per unit area, σ :

∆HI = Vm SV σ

where Vm is the molar volume.

Both of these equations contain the term V, which increases rapidly as the inverse of the particle
size m. The model predicts that solution formation is impossible because the cost due to interfaces
overwhelms any gain from binding energies or entropy. And yet, as demonstrated by atom-probe
experiments, solutions do form during mechanical alloying, so there must be a mechanism to re-
duce interfacial energy as the particles are divided. The mechanism is the reverse of that associated
with precipitation (Fig. ??). A small precipitate can be coherent but the coherency strains become
unsustainable as it grows. Similarly, during mechanical alloying it is conceivable that the particles
must gain in coherence as their size diminishes. The milling process involves fracture and welding
of the attrited particles so only those welds which lead to coherence might succeed.
Case study: mechanical alloying 19

Figure 3.6 The change in coherence as a function of particle size. The lines represent lattice planes which
are continuous at the matrix/precipitate interface during coherence, but sometimes terminate in dislocations for
the incoherent state. Precipitation occurs in the sequence a→c whereas mechanical alloying is predicted to lead
to a gain in coherence in the sequence c→a.

3.5 SHAPE OF FREE ENERGY CURVES

There are many textbooks which emphasise that free energy of mixing curves such as that illustrated
in Fig. 2.2 must be drawn such that the slope is either −∞ or +∞ at x = 0 and x = 1 respectively.
This is a straightforward result from equation 19 which shows that
 
∂ ∆SM x
= −kNa ln
∂x 1−x

so that the slope of −T ∆SM becomes ±∞ at the extremes of concentration. Notice that at those ex-
tremes, any contribution from the enthalpy of mixing will be finite and negligible by comparison, so
that the free energy of mixing curve will also have slopes of ±∞ at the vertical axes corresponding
to the pure components. Note that the intercepts at the vertical axes representing the pure compo-
nents are nevertheless finite, with values µA0 and µB0 . It follows that the free energy of mixing of any
solution from its components will at first decrease at an infinite rate.
However, these conclusions are strictly valid only when the concentration is treated as a contin-
uous variable which can be as close to zero or unity as desired. The discussion here emphasises that
there is a discrete structure to solutions. Thus, when considering N particles, the concentration can
never be less than 1/N since the smallest amount of solute is just one particle. The slope of the free
energy curve will not therefore be ±∞ at the pure components, but rather a finite number depending
on the number of particles involved in the process of solution formation. Since the concentration is
not a continuous variable, the free energy ‘curve’ is not a curve, but is better represented by a set
of points representing the discrete values of concentration that are physically possible when mixing
particles. Obviously, the shape approximates to a curve when the number of particles is large, as is
the case for an atomic solution made of a mole of atoms. But the curve remains an approximation.
 4 Computer calculation of
phase diagrams
The thermodynamic methods described thus far in this chapter are revealing and have been applied
towards the understanding and modelling the behaviour of iron and its alloys. It nevertheless is too
complicated in the context of multicomponent steels where individual solute concentrations can vary
over a large range. Therefore, methods have been developed for doing this in a seamless manner;
these methods have been so successful that they now represent the first step in any alloy development
project. The subject of the computer calculation of phase diagrams based on experimental data has
been reviewed extensively, e.g., [124–129]. The focus here is on the framework for such generic
calculations, which necessarily involves a degree of educated, clever empiricism. The process has
also led to the systematic compilation and assessment of experimental data on a scale that is perhaps
unique in science. All this was initiated by like-minded scientists long before “big data” or computer
modelling became fashionable.
One possibility is to represent thermodynamic quantities by a series expansion with sufficient ad-
justable parameters to adequately fit the experimental data. There has to be a compromise between
the accuracy of the fit and the number of terms in expansion. However, such expansions do not
generalise well when dealing with complicated phase diagram calculations involving many com-
ponents and phases. Experience suggests that the specific heat capacities for the pure elements are
better represented by a polynomial with a form that describes most of the known experimental data:

b11
CP = b8 + b9 T + b10T 2 + . (4.1)
T2
If the fit with experimental data is found not to be good enough, the polynomial is applied to a
range over which the fit is satisfactory, and more than one polynomial is used to represent the full
dataset with care exercised to ensure continuity over the range. A standard element reference state
is defined with a list of the measured enthalpies and entropies of the pure elements at 298 K and one
atmosphere pressure, for the crystal structure appropriate for these conditions. With respect to this
state, the Gibbs free energy is obtained by integration to be:

b17
G = b12 + b13T + b14T ln{T } + b15T 2 + b16T 3 + . (4.2)
T
This free energy is defined with respect to a reference (included in b12 ) i.e., relative to the enthalpy at
298.15 K and entropy at 0 K of the stable states of the element(s) concerned at 298.15 K. Allotropic
transformations can be included if the transition temperatures, enthalpy of transformation and the
CP coefficients for all the phases are known.
Any specific contributions to CP , such as due to magnetic transitions, are dealt with separately, as
are the effects of pressure. Once again, the equations for these effects are chosen carefully in order
to maintain generality.
The excess Gibbs free energy for a binary solution with components A and B is written:
j
∆e GAB = xA xB ∑ LAB,i (xA − xB )i . (4.3)
i=0

For i = 0 this gives a term xA xB LAB,0 which is familiar in regular solution theory, where the coeffi-
cient LAB,0 is, as usual, independent of chemical composition and to a first approximation describes

21
22 Course on thermodynamics

the interaction between components A and B. If all other LAB,i are zero for i > 0 then the equation
reduces to the regular solution model with LAB,0 as the regular solution parameter. Further terms
(i > 0) are added to allow for any composition dependence not described by the regular solution
constant.
In the first approximation, the excess free energy of a ternary solution can be represented purely
by a combination of the binary terms in Equation 4.3:
j
∆e GABC = xA xB ∑ LAB,i (xA − xB )i
i=0
j j
+xB xC ∑ LBC,i (xB − xC )i + xC xA ∑ LCA,i (xC − xA )i .
i=0 i=0

The advantage of the representation embodied in Equation 4.3 is clear, that for the ternary case, the
relation reduces to the binary problem when one of the components is set to be identical to another,
e.g., B≡C [130].
Experimental data may indicate significant ternary interactions, in which case a term
xA xB xC LABC,0 is added to the excess free energy. If this does not adequately represent the devia-
tion from the binary summation, then it can be converted into a series which properly reduces to a
binary formulation when there are only two components:
 1
xA xB xC LABC,0 + (1 + 2xA − xB − xC )LABC,1
3
1 1 
+ (1 + 2xB − xC − xA )LBCA,1 + (1 + 2xC − xA − xB )LCAB,1 .
3 3
This method can be extended to any number of components as long as appropriate thermody-
namic data are available, with the advantage that few coefficients have to be changed when the data
due to one component are improved. The information necessary to derive the coefficients become
sparse for systems with more than three components.
 5 Thermodynamics of
irreversible processes
Thermodynamics as a subject is limited to the equilibrium state. Properties such as entropy and
free energy are, on an appropriate scale, static and time-invariant during equilibrium. There is an
extension of the subject to systems that are close to equilibrium so that they can be divided into
subsystems where the rules of equilibrium can be applied locally [139]. Parameters not relevant
to the discussion of equilibrium, such as thermal conductivity, diffusivity and viscosity, then enter
the picture because they can describe a second kind of time independence, that of the steady state.
For example, the concentration profile does not change during steady-state diffusion, even though
energy is being dissipated during diffusion.
The thermodynamics of irreversible processes deals with systems that are not at equilibrium but
are nevertheless stationary. The theory in effect uses thermodynamics to deal with kinetic phe-
nomena. There is nevertheless, a distinction between the thermodynamics of irreversible processes
and kinetics. The former applies strictly to the steady-state, whereas there is no such restriction on
kinetic theory.

5.1 REVERSIBILITY
A process, the direction of which can be changed by an infinitesimal alteration in the external condi-
tions is called reversible, because an exact reversal leads to no net dissipation of energy. Figure 5.1
shows the response of an ideal gas contained at uniform pressure within a cylinder, any change be-
ing achieved by the motion of the piston. For any starting point on the pressure-volume curve, the
application of an infinitesimal force may cause the piston to move to an adjacent position still on
the curve, while the removal of the infinitesimal force restores the system to its original state. This
process is reversible because there is no net dissipation in displacing and recovering the frictionless
piston.
If the motion of the piston in the cylinder entails friction, then deviations occur from the P/V
curve as illustrated by the cycle in Figure 5.1. An infinitesimal force cannot move the piston because
energy must be dissipated to overcome the friction; this energy is the area enclosed by the cycle
on the P/V plot. A process such as this, which involves the dissipation of energy, is classified
as irreversible with respect to an infinitesimal change in the external conditions. More generally,

Figure 5.1 The curve represents the

gas variation in pressure within the cylin-
der as the volume of the ideal gas is al-
tered by the frictionless positioning the
piston. The cycle represents the dissi-
pation of energy when the motion of
the piston causes friction.

reversibility means that it is possible to pass from one state to another without appreciable deviation
from equilibrium. Real process are not reversible so equilibrium thermodynamics can only be used

23
24 Course on thermodynamics

approximately, though the same principles define whether or not a process can occur spontaneously
without ambiguity.
For irreversible processes the equations of classical thermodynamics become inequalities. For
example, at the equilibrium melting temperature, the free energies of the liquid and solid are identi-
cal (Gliquid = Gsolid ) but not so below that temperature (Gliquid > Gsolid ). Such inequalities are much
more difficult to deal with though they indicate the natural direction of change. For steady-state
processes however, the thermodynamic framework for irreversible processes as developed by On-
sager [140] is particularly useful in obtaining relationships even though the system may not be at
equilibrium.

5.1.1 LINEAR LAWS

There is no change in entropy or free energy at equilibrium. An irreversible process dissipates energy
and entropy is created continuously. In the example illustrated in Fig. 5.1, the dissipation was due to
friction; diffusion ahead of a moving interface is dissipative. The rate at which energy is dissipated
is the product of the temperature and the rate of entropy production:

T Ṡ = JX (5.1)

where J is a generalised flux of some kind, and X a generalised force. In the case of an electrical
current, the heat dissipation is the product of the current (J) and the electromotive force (X).
As long as the flux-force sets can be expressed as in Equation 5.1, the flux must naturally depend
in some way on the force. It may then be written as a function J{X} of the force X. At equilibrium,
the force is zero. J{X} can be expanded in a Taylor series about equilibrium (X = 0):
∞
J{X} = ∑ an X n
0
X X2
= J{0} + J ′{0} + J ′′ {0} . . . (5.2)
1! 2!
In this expansion, J{0} = 0 because there is no flux in the absence of force. If the high order terms
are neglected then a proportionality between the force and flux is revealed:

J ∝ X.

Therefore, the forces and their conjugate fluxes are linearly related whenever the dissipation can
be expressed as in Equation 5.2, at least when the deviations from equilibrium are not large. This
caveat is illustrated nicely by the relationship between the rate at which an interface moves and the
driving force. In Chapter ??, Equation ??, which is limited to small driving forces, shows a linear
relationship between the two quantities, whereas Equation ?? which is derived without limits on
the magnitude of the driving force, shows that the rate and driving force are not in general linearly
related.
In another example, consider a closed system in which a quantity dH of heat is transferred in a
time interval dt across an area A in a direction z normal to that area, from a region at temperature Th
to a lower temperature Tℓ . The receiving part increases its entropy by dH/Tℓ whereas the depleted
region experiences a reduction dH/Th, so the change in entropy is
 
1 1
dS = dH − .
Tℓ Th
The rate of entropy production per unit volume is therefore
 
1 dS 1 dH 1 1
Ṡ = = − . (5.3)
V dt V dt Tℓ Th
Thermodynamics of irreversible processes 25

The flux of heat J is defined as A−1 dH/dt so Equation 5.3 becomes

   
A 1 1 1 dT
Ṡ = J − ≡J − 2
V Tℓ Th T dz
 
1 dT
or T Ṡ = |{z} J −
T dz
flux | {z }
force
Some examples of forces and fluxes in their generic form are listed in Table 5.1.

Table 5.1
Examples of forces and their conjugate fluxes. z is the distance over which the gra-
dient exists, φ is the electrical potential and µ the chemical potential.

Force Flux

− ∂∂φz Electrical Current

− T1 ∂∂Tz Heat flux
− ∂∂µzi Diffusion flux
Stress Strain rate

5.1.2 MULTIPLE IRREVERSIBLE PROCESSES

There are circumstances whereby a number of irreversible processes occur together. In a ternary
Fe-Mn-C alloy, the diffusion flux of carbon depends not only on the gradient of carbon, but also
on that of manganese. A uniform distribution of carbon will tend to become inhomogeneous in
the presence of a manganese concentration gradient. Similarly, the flux of heat may not depend on
the temperature gradient alone; heat can be driven also by an electromotive force (Peltier effect).1
Electromigration involves diffusion that is driven by an electromotive force. When there is more
then one dissipative process, the total energy dissipation rate can still be written

T Ṡ = ∑ Ji Xi . (5.4)
i

In general, if there is more than one irreversible process occurring, it is found experimentally that
each flow Ji is related not only to its conjugate force Xi , but is also linearly related to all other forces
present. Thus,
Ji = Mi j X j (5.5)
with i, j = 1, 2, 3 . . .. Therefore, a given flux depends on all the forces causing the dissipation of
energy.

5.1.3 ONSAGER RECIPROCAL RELATIONS

Equilibrium in real systems is dynamic on a microscopic scale. It seems obvious that to maintain
equilibrium under these dynamic conditions, a process and its reverse must occur at the same rate on
the microscopic scale. The consequence is that provided the forces and fluxes are chosen from the
dissipation equation and are independent, Mi j = M ji . This is known as the Onsager theorem, or the
26 Course on thermodynamics

Onsager reciprocal relations. It applies to systems near equilibrium when the properties of interest
have even parity, and assuming that the fluxes and their corresponding forces are independent. An
exception occurs with magnetic fields in which case there is a sign difference Mi j = −M ji [141].
 6 Quasichemical solution
The regular solution model assumes a random distribution of atoms even though the enthalpy of
mixing is not zero. whereas in reality a random solution is only expected at high temperatures when
the entropy term overwhelms any tendency for ordering or clustering of atoms. It follows that the
configurational entropy of mixing should therefore vary with the temperature. The quasichemical
solution model has a better treatment of configurational entropy which accounts for a non-random
distribution of atoms. The model is so-called because it has a mass-action equation that has similar-
ity to chemical reactions [100]. However, the presentation below follows derivations by Christian
[101] and Lupis [97].
Recalling that zNAB represents the number of A-B bonds, the total energy of the assembly for a
particular value of NAB is UNAB = −z(NA εAA + NB εBB − NAB ω ) where ω = εAA + εBB − 2εAB . In a
non-random solution there are many values that NAB can adopt; each value corresponding to one or
more arrangement of atoms with an identical value of U is therefore associated with a degeneracy
gNAB which is the number of arrangements possible for a given value of U. The partition function is
therefore the sum over all possible NAB :
 
UN
Ω = ∑ gNAB exp − AB
NAB kT
 
z(NA εAA + NB εBB − NAB )ω
= ∑ gNAB exp . (6.1)
NAB kT

For a given value of NAB , the different non-interacting pairs of atoms can be arranged in the follow-
ing number of ways (N = NA + NB )

( 12 zN)!
gNAB ∝ (6.2)
( 21 z[NA − NAB ])! ( 21 z[NB − NAB ])! ( 12 zNAB )! ( 12 zNBA )!
where the first and second terms in the denominator refer to the numbers of A-A and B-B bonds
respectively, and the third and fourth terms the numbers of A-B and B-A pairs respectively. This
is not an equality because the various pairs are not independent, as illustrated in (Figure 6.1); the
distribution of pairs is not random. Guggenheim addressed this difficulty by using a normalisation

Figure 6.1 Why pairs of atoms cannot be distributed at random on lattice sites
which are marked as small dots. Once the bonds connecting the coordinates (i, i +
1), (i + 1, i + 2), (i + 2, i + 3) are made as illustrated, the final bond connecting
(i, i + 3) is necessarily occupied by a pair AB. Adapted from Lupis [97].

factor such that the summation of all possible degeneracies equals the total number of possible
configurations as follows.
Suppose that the number of arrangements of pairs of atoms possible in a random solution is
identified with an asterix, then from the proportionality 6.2, it is seen that

( 21 zN)!
g∗ ∝ . (6.3)
∗ ])! ( 1 z[N − N ∗ ])! ( 1 zN ∗ )! ( 1 zN ∗ )!
( 12 z[NA − NAB 2 B AB 2 AB 2 BA

27
28 Course on thermodynamics

This again will overestimate the number of possibilities (Figure 6.1), but for a random solution it is
known already that
N!
g∗ = . (6.4)
NA ! NB !
It follows that gNAB can be normalised as
∗ ])! ( 1 z[N − N ∗ ])! ( 1 zN ∗ )! ( 1 zN ∗ )!
( 12 z[NA − NAB 2 B AB 2 AB 2 BA N!
gNAB = × . (6.5)
( 12 z[NA − NAB ])! ( 21 z[NB − NAB ])! ( 21 zNAB )! ( 21 zNBA )! NA ! NB !
With this, the partition function Ω is defined explicitly and the problem is in principle solved. It
is usual however, to simplify first by assuming that the sum in Equation 6.1 can be replaced by
its maximum value. This is because the thermodynamic properties that follow from the partition
function depend on its logarithm, in which case the use of the maximum is a good approximation.
e of A-B bonds may then be obtained by setting ∂ ln{Ω}/∂ N
The equilibrium number NAB AB = 0
[97, 101]:
e 2Nzx(1 − x)
NAB = (6.6)
βq + 1
with βq being the positive root of the equation

βq2 − (1 − 2x) = 4x(1 − x) exp{2ω /kT }, (6.7)

so that
e 2Nzx(1 − x)
NAB = 1
.
[1 − 2x + 4x(1 − x) exp{2ω /kT }] 2 + 1
The percentages of the different pairs are plotted in Figure 6.2. Equation 6.6 obviously corresponds
to the regular solution model if βq = 1 with a random arrangement of atoms. As expected, the
number of unlike pairs is reduced when clustering is favoured, and increased when ordering is
favoured.

Figure 6.2 Calculated percentages of

pairs for the quasichemical model with
x = (1 − x) = 0.5. The result is indepen-
dent of the coordination number z.

The free energy of the assembly is

G = F = −kT ln{Ω} = UNAB
e − kT ln gN e
AB
(6.8)
so that the free energy of mixing per mole becomes
e
∆GM = zNAB ω − Na kT ln gNAB
e

2zω Na x(1 − x)
= −RT ln gNAB
e . (6.9)
βq + 1
| {z }
molar enthalpy of mixing
Quasichemical solution 29

The second term on the right-hand side has the contribution from the configurational entropy of
mixing. By substituting for gNAB
e , and with considerable manipulation, Christian has shown that this

can be written in terms of βq so that the molar free energy of mixing becomes:

2zω Na x(1 − x)
∆GM =
βq + 1
 
+RT (1 − x) ln{1 − x} + x ln{x}
 
1 βq + 1 − 2x βq − 1 + 2x
+ RT z (1 − x) ln + x ln
2 (1 − x)(βq + 1) x(βq + 1)

The second term in this equation is the usual contribution from the configurational entropy of mixing
in a random solution, whereas the third term can be regarded as a quasichemical correction for the
entropy of mixing because the atoms are not randomly distributed.
It is not possible to give explicit expressions for the chemical potential or activity coefficient
since βq is a function of concentration. Approximations using series expansions are possible [97]
but the resulting equations are not as easy to interpret physically as the corresponding equations for
the ideal or regular solution models.
The expressions in the quasichemical (or first approximation) clearly reduce to those of the reg-
ular solution (or zeroth approximation) model when βq = 1. Although a better model has been
obtained, the first approximation relies on the absence of interference between atom-pairs. How-
ever, each atom in a pair belongs to several pairs so that better approximations can be obtained by
considering larger clusters of atoms in the calculation. Such calculations are known as the “cluster
variation” method proposed originally by Kikuchi [102]. The improvements obtained with these
higher approximations are usually rather small though there are cases where pairwise interactions
simply will not do.
It is worth emphasising that although the quasichemical model has an excess entropy, this comes
as a correction to the configurational entropy. The excess entropy from this model is always negative;
as Lupis pointed out [97], there is more disorder in a random solution than in one that is biased.
Therefore, the configurational entropy from the quasichemical model is always less than expected
from an ideal solution. Thermal entropy or other terms such as magnetic or electronic are additional
contributions.
The procedure in the development of the quasichemical models is illustrated in Figure 6.3.
30 REFERENCES

Categorise and count the variety of atom-

atom or atom-vacancy pairs possible

Estimate the configurational energy aris-

ing from the pairwise interactions

Write the configurational partition function

Discover the degeneracy for each configuration

Normalise the degeneracy function to ensure that the grande summation

of all degeneracies is the total number of possible configurations

Replace the summation in the partition function by its largest term

Derive required thermodynamic functions from partition function

Deal with any excess thermodynamic quantities

not described by a consideration of configurations

Figure 6.3 Steps in the construction of a quasichemical solution model.

REFERENCES
1. A. B. Pippard: Classical Thermodynamics: Cambridge, U. K.: Cambridge University Press,
1981.
2. C. Zener: ‘Kinetics of the decomposition of austenite’, Trans. Am. Inst. Min. Metall. Engng.,
1946, 167, 550–595.
3. C. Zener: ‘Equilibrium relations in medium alloy steels’, Trans. A.I.M.E., 1946, 167, 513–534.
4. W. J. Moore: Physical Chemistry: 5th ed., New Jersey, USA: Prentice-Hall, 1972.
5. K. G. Denbigh: The principles of chemical equilibrium: 4th ed., Cambridge, U. K.: Cambridge
University Press, 2018.
6. P. Debye: ‘Zur theorie der spezifischen wärmen’, Annalen der Physik, 1912, 344, 789–839.
REFERENCES 31

7. L. Kaufman, S. V. Radcliffe, and M. Cohen: ‘Thermodynamics of the bainite reaction’, In:

V. F. Zackay, and H. I. Aaronson, eds. Decomposition of Austenite by Diffusional Processes.
New York, USA: Interscience, 1962:313–352.
8. F. A. Lindemann: ‘Ueber die berechnung molekularer eigenfrequenzen (the calculation of
molecular Eigen-frequencies)’, Physikalische Zeitschrift, 1910, 11, 609–612.
9. L. Kaufman: ‘Estimation of the entropy of NaCl type compounds’, Trans. Metall. Soc. AIME,
1962, 224, 1006–1009.
10. L. Kaufman, E. V. Clougherty, and R. J. Weiss: ‘The lattice stability of metals – iii iron’, Acta
Metallurgica, 1963, 11, 323–335.
11. L. D. Blackburn, L. Kaufman, and M. Cohen: ‘Phase transformations in iron-ruthenium alloys
under high pressure’, Acta Metallurgica, 1965, 13, 533–541.
12. F. H. Herbstein, and J. Smuts: ‘Determination of the Debye temperature of alpha-iron by
X-ray diffraction’, Philosophical Magazine, 1963, 8, 367–385.
13. G. L. Stepakoff, and L. Kaufman: ‘Thermodynamic properties of hcp iron and iron-ruthenium
alloys’, Acta Metallurgica, 1968, 16, 13–22.
14. R. L. Clendenen, and H. G. Drickamer: ‘The effect of pressure on the volume and lattice
parameters of ruthenium an iron’, Journal of the Physics and Chemistry of Solids, 1964, 25,
865–868.
15. H.-K. Mao, W. A. Bassett, and T. Takahashi: ‘Effect of pressure on crystal structure and lattice
parameters of iron up to 300 kbar’, Journal of Applied Physics, 1967, 38, 272–276.
16. A. H. Wilson: The theory of metals: Cambridge, U. K.: Cambridge University Press, 1958.
17. W. H. Keesom, and B. Kurrelmeyer: ‘The atomic heat of iron from 1.1◦ to 20.4◦ K’, Physica,
1939, 6, 633–672.
18. J. A. Hofmann, A. Paskin, K. J. Tauer, and R. J. Weiss: ‘Analysis of ferromagnetic and
antiferro-magnetic second-order transitions’, Journal of Physics and Chemistry of Solids,
1956, 1, 45–60.
19. P. Wilkes: Solid state theory in metallurgy: Cambridge, U. K.: Cambridge University Press,
1973.
20. J. M. Ziman: Models of disorder: Cambridge, U. K.: Cambridge University Press, 1979.
21. I. M. K. Billas, A. Chatelain, and W. A. de Heer: ‘Magnetism from the atom to the bulk in
iron, cobalt, and nickel clusters’, Science, 1994, 265, 1682–1684.
22. A. J. Freeman, and R. Q. Wu: ‘Electronic structure theory of surface, interface and thin film
magnetism’, Journal of Magnetism and Magnetic Materials, 1991, 100, 497–514.
23. I. M. K. Billas, J. A. Baker, A. Chatelain, and W. A. de Heer: ‘Magnetic moments of iron
clusters with 25 to 700 atoms and their dependence on temperature’, Physical Review Letters,
1993, 71, 4067–4070.
24. L. C. nd X. L. Wan, and K. M. Wu: ‘Three-dimensional analysis of ferrite allotrimorphs
nucleated on grain boundary faces, edges and corners’, Materials Characterization, 2010, 61,
580–583.
25. L. Patrick: ‘The change in ferromagnetic Curie points with hydrostatic pressure’, Physical
Review, 1954, 93, 384–392.
26. J. A. Hoffmann, A. Paskin, K. J. Tauer, and R. J. Weiss: ‘Analysis of ferromagnetic and
antiferromagnetic second-order transitions’, Journal of the Physics and Chemistry of Solids,
1956, 1, 45–60.
27. C. Kittel: Introduction to Solid State Physics: 4th ed., New York, USA: John Wiley and Sons
Inc., 1971.
28. L. Kaufman: ‘Condensed-state reactions at high pressures’, In: W. M. Mueller, ed. Ener-
getics in Metallurgical Phenomena, vol. 3. New York, USA: Gordon and Breach Publishers,
1967:55–84.
29. D. C. Wallace, P. H. Sidles, and G. C. Danielson: ‘Specific heat of high purity iron by a pulse
32 REFERENCES

heating method’, Journal of Applied Physics, 1960, 31, 168–176.

30. R. J. Weiss, and K. J. Tauer: ‘Components of the thermodynamic functions of iron’, Physical
Review, 1956, 102, 1490–1495.
31. H. A. Gersch, C. G. Shull, and M. K. Wilkinson: ‘Critical magnetic scattering of neutrons by
iron’, Physical Review, 1956, 103, 525–534.
32. G. Inden: ‘Determination of chemical and magnetic interchange energies in bcc alloys’,
Zeitschrift für Metallkunde, 1975, 66, 577–582.
33. M. Hillert, and M. Jarl: ‘A model for alloying effects in ferromagnetic metals’, CALPHAD,
1978, 2, 227–238.
34. A. P. Miodownik: ‘The calculation of magnetic contributions to phase stability’, C, 1977, 1,
133–158.
35. H. Ino, K. Hayashi, T. Otsuka, D. Isobe, K. Tokumitsu, and K. Oda: ‘Appearance of fer-
romagnetism in fcc solid solutions of binary and ternary Fe-Cu based systems prepared by
mechanical alloying techniques’, Materials Science & Engineering A, 2001, 304-306, 972–
974.
36. H. H. Ettwig, and W. Pepperhoff: ‘On magnetism of gamma-Fe-Ni-Mn alloys’, Physica Status
Solidi A, 1974, 23, 105–111.
37. J. M. D. Coey: ‘Magnetic localization and magnetoresistance in mixed-valence manganites
and related ferromagnetic oxides’, Philosophical Transactions of the Royal Society A, 1998,
356, 1519–1541.
38. T. G. Perring, G. Aeppli, S. M. Hayden, Y. Tokura, Y. Moritomo, J. P. Remeika, and S.-W.
Cheong: ‘Magnetic dynamics in colossal magnetoresistive perovskite manganites’, Philo-
sophical Transactions of the Royal Society, London, 1998, 356, 1563–1575.
39. K. J. Tauer, and R. J. Weiss: ‘Unusual magnetic structure of face centered cubic Fe’, Bulletin
of the American Physical Society, 1961, 6, 125.
40. R. J. Weiss, and A. S. Marotta: ‘Spin-dependence of the resistivity of magnetic metals’,
Journal of the Physics and Chemistry of Solids, 1956, 9, 302–308.
41. S. C. Abrahams, L. Guttman, and J. S. Kasper: ‘Neutron diffraction determination of antifer-
romagnetism in face-centered cubic (gamma) iron’, Physical Reviews, 1962, 127, 2052–2055.
42. U. Gonser, C. J. Meechan, A. H. Muir, and H. Wiedersich: ‘Determination of the néel temper-
atures in fcc iron’, Journal of Applied Physics, 1963, 34, 2373–2378.
43. R. J. Weiss: ‘The origin of the ‘Invar effect’, Proceedings of the Physics Society, 1963, 82,
281–288.
44. J. G. Wright: ‘Ferromagnetism in epitaxial FCC iron films’, Philosophical Magazine, 1971,
24, 217–233.
45. U. Gonser, K. Irischel, and S. Nasu: ‘Ferromagnetic ordering in fcc gamma-iron precipitates
in Cu-Au alloys’, Journal of Magnetism and Magnetic Materials, 1980, 15-18, 1145–1146.
46. S. D. Bader, and E. R. Moog: ‘Magnetic properties of novel epitaxial films’, Journal of Applied
Physics, 1987, 61, 3729–3734.
47. J. Z. Jiang, and F. T. Chen: ‘A study of the microstructure of mechanically alloyed fcc FeCu’,
Journal of Physics – Condensed Matter, 1994, 6, L343–L348.
48. J. Z. Jiang, Q. A. Pankhurst, C. E. Johnson, C. Gente, and R. Borman: ‘Magnetic properties
of mechanically alloyed FeCu’, Journal of Physics – Condensed Matter, 1994, 6, L227–L232.
49. J. Z. Jiang, C. Gente, and R. Bormann: ‘Mechanical alloying in the Fe-Cu system’, Materials
Science & Engineering A, 1998, 242, 268–277.
50. K. J. Tauer, and R. J. Weiss: ‘Magnetic second-order transitions’, Physical Reviews, 1955,
100, 1223–1224.
51. J. Kübler: ‘Magnetic moments of ferromagnetic and antiferromagnetic bcc and fcc iron’,
Physics Letters A, 1981, 81, 81–83.
52. G. L. Krasko: ‘Metamagnetic behaviour of fcc iron’, Physical Review B, 1987, 36, 8565–8569.
REFERENCES 33

53. P. Mohn, K. Schwarz, and D. Wagner: ‘Magnetoelastic anomalies in Fe-Ni invar alloys’,
Physical Review B, 1991, 43, 3318–3324.
54. E. F. Wassermann: ‘The invar problem’, Journal of Magnetism and Magnetic Materials, 1991,
100, 346–362.
55. M. Nicol, and G. Jura: ‘Mössbauer spectrum of iron-57 in iron metal at very high pressure’,
Science, 1963, 141, 1035–1038.
56. D. N. Pipkorn, C. K. Edge, P. Debrunner, G. De Pasquali, H. G. Drickamer, and H. Frauen-
felder: ‘Mössbauer effect in iron under very high pressure’, Physical Review A, 1964, 135,
1604–1612.
57. J. S. Kouvel: ‘Magnetic properties of solids under pressure’, In: W. Paul, and D. M.
Warschauer, eds. Solids under pressure. New York, USA: McGraw Hill, 1963:277–301.
58. K. Shimizu, T. Kimura, S. Furomoto, K. Takeda, K. Kontani, Y. Onuki, and K. Amaya: ‘Super-
conductivity in the non-magnetic state of iron under pressure’, Nature, 2001, 412, 316–318.
59. V. Thakor, J. B. Staunton, J. Poulter, S. Ostanin, B. Ginatempo, and E. Bruno: ‘Ab initio cal-
culations of incommensurate antiferromagnetic spin fluctuations in hcp iron under pressure’,
Physical Review B, 2003, 67, 180405.
60. S. Fox, and H. J. F. Jansen: ‘Structure and properties of trigonal iron’, Physical Review B,
1996, 53, 5119–5122.
61. S. Peng, and H. J. F. Jansen: ‘Electronic structure of face-centered tetragonal iron’, Journal of
Applied Physics, 1990, 67, 4567–4569.
62. E. C. Bain: ‘The nature of martensite’, Trans. AIME, 1924, 70, 25–46.
63. C. M. Wayman: ‘The growth of martensite since E C Bain (1924) - some milestones’, Mate-
rials Forum, 1990, 56-58, 1–32.
64. S. S. Peng, and H. J. F. Jansen: ‘Structural and magnetic properties of tetragonally distorted
iron’, Ultramicroscopy, 1992, 47, 361–366.
65. B. T. Jonker, K. H. Walker, E. Kisker, G. A. Prinz, and C. Carbone: ‘Spin-polarized pho-
toemission study of epitaxial Fe(001) films on Ag(001)’, Physical Review Letters, 1986, 57,
142–143.
66. G. L. Krasko, and G. B. Olson: ‘Ferromagnetism and crystal lattice stability of bcc and fcc
iron’, Journal of Applied Physics, 1990, 67, 4570–4572.
67. H. C. Herper, E. Hoffmann, and P. Entel: ‘Ab initio investigations of iron-based martensitic
stystems’, Le Journal de Physique IV, 1997, 7, C5–71 – C5–76.
68. H. C. Herper, E. Hoffmann, and P. Entel: ‘Ab initio full-potential study of the structural and
magnetic phase stability of iron’, Physical Review B, 1999, 60, 3839–3848.
69. C. Li, A. J. Freeman, H. J. F. Jansen, and C. L. Fu: ‘Magnetic anisotropy in low-dimensional
ferromagnetic systems: Fe monolayers on Ag (001), Au(001) and Pd(001) substrates’, Physi-
cal Review B, 1990, 42, 5433–5422.
70. H. J. Haugan, B. D. McCombe, and P. G. Mattocks: ‘Structural and magnetic properties of
thin epitaxial Fe films on (110) GaAs prepared by metalorganic chemical vapor deposition’,
Journal of Magnetism and Magnetic Materials, 2002, 247, 296–304.
71. A. S. Edelstein, J. S. Murday, and B. B. Rath: ‘Challenges in nanomaterials design’, Progress
in Materials Science, 1997, 42, 5–21.
72. J. H. van Vleck: ‘On the anisotropy of cubic ferromagnetic crystals’, Physical Review, 1937,
52, 1178–1198.
73. C. Stamm, F. Marty, A. Vaterlaus, V. Weich, S. Egger, U. Maier, U. Ramsperger, H. Fuhrmann,
and D. Pescia: ‘Two-dimensional magnetic particles’, Science, 1998, 282, 449–451.
74. S. T. Purcell, B. Heinrich, and A. S. Arrott: ‘Perpendicular anisotropy at the (001) surface of
bulk iron single crystals’, Journal of Applied Physics, 1988, 64, 5337–5339.
75. R. F. Willis, J. A. C. Bland, and W. Schwarzacher: ‘Ferromagnetism in ultrathin metastable
films of fcc Fe, Co, Ni’, Journal of Applied Physics, 1988, 63, 4051–4056.
34 REFERENCES

76. S. D. Bader, E. R. Moog, and P. Grunberg: ‘Magnetic hysteresis of epitaxially deposited iron
in the monolayer range: a Kerr effect experiment in surface magnetism’, Journal of Magnetism
and Magnetic Materials, 1986, 53, L295–L298.
77. T. Baykara, R. H. Hauge, N. Norem, P. Lee, and J. L. Margrave: ‘A review of containerless
thermophysical property measurements for liquid metals and alloys’, High Temperature and
Materials Science, 1994, 32, 113–154.
78. R. I. L. Guthrie, and T. Iida: ‘Thermodynamic properties of liquid metals’, Materials Science
& Engineering A, 1994, 178, 35–41.
79. R. I. Orr, and J. Chipman: ‘Thermodynamic functions of iron’, TMS-AIME, 1967, 239, 630–
633.
80. W. W. Anderson, and T. J. Ahrens: ‘An equation of state for liquid iron and implications for
the Earth’s core’, Journal of Geophysical Research, 1994, 99, 4273–4284.
81. L. Kaufman: ‘Thermodynamics of martensitic fcc⇋bcc and fcc⇋hcp transformations in the
iron-ruthenium system’, In: Physical properties of martensite and bainite, Special Report 93.
London, U.K.: Iron and Steel Institute, 1965:49–52.
82. G. P. Tiwari: ‘Modification of Richard’s rule and correlation between entropy of fusion and
allotropic behaviour’, Metal Science, 1978, 12, 317–320.
83. J. Mo, H. Liu, Y. Zhang, M. Wang, L. Zhang, B. Liu, and W. Yang: ‘Effects of pressure on
structure and mechanical property in monatomic metallic glass’, Journal of Non-Crystalline
Solids, 2017, 464, 1–4.
84. C. V. Thompson, and F. Spaepen: ‘On the approximation of the free energy change on crys-
tallisation’, Acta Metallurgica, 1979, 27, 1855–1859.
85. J. D. Hoffman: ‘Thermodynamic driving force in nucleation and growth processes’, Journal
of Chemical Physics, 1958, 29, 1192–1193.
86. D. R. H. Jones, and G. A. Chadwick: ‘An expression for the free energy of fusion in the
homogeneous nucleation of solid from pure melts’, Philosophical Magazine, 1971, 190, 995–
998.
87. L.-G. Liu, and W. A. Bassett: ‘The melting of iron up to 200 kbar’, Journal of Geophysical
Research, 1975, 80, 3777–3782.
88. J. H. Nguyen, and N. C. Holmes: ‘Melting of iron at the physical conditions of the Earth’s
core’, Nature, 2004, 427, 339–342.
89. O. L. Anderson, and D. G. Isaak: ‘Another look at the core density deficit of earth’s outer
core’, Physics of the Earth and Planetary Interiors, 2002, 131, 19–27.
90. H. Ichikawa, T. Tsuchiya, and Y. Tange: ‘The P-V-T equation of state and thermodynamic
properties of liquid iron’, Journal of Geophysical Research, 2014, 119, 240–252.
91. D. Gerlich, and S. Hart: ‘Pressure dependence of the elastic moduli of three austenitic stainless
steels’, Journal of Applied Physics, 1984, 55, 880–884.
92. M. Müller, P. Erhart, and A. Karsten: ‘Analytic bond-order potential for bcc and fcc iron—
comparison with established embedded-atom method potentials’, Journal of Physics – Con-
densed Matter, 2007, 19, 3262220.
93. H. K. Mao, Y. Wu, L. C. Chen, and J. F. Shu: ‘Static compression of iron to 300 GPa and
Fe0.8 Ni0.2 alloy to 260 GPa: implications for composition of the core’, Journal of Geophysical
Research, 1990, 95, 21737–21742.
94. M. W. Guinan, and D. N. Beshers: ‘Pressure derivatives of the elastic constants of α -iron to
10 kb’, Journal of the Physics and Chemistry of Solids, 1968, 29, 541–549.
95. J. S. Benjamin: ‘oxide dispersion strengthened ods superalloys directional recrystallisation’,
Metallurgical Transactions, 1970, 1, 2943–2951.
96. A. Y. Badmos, and H. K. D. H. Bhadeshia: ‘Evolution of solutions’, Metallurgical & Materials
Transactions A, 2000, 28, 2189–2193.
97. C. H. P. Lupis: Chemical thermodynamics of materials: New York, USA: North-Holland,
REFERENCES 35

1983.
98. J. H. Hildebrand: ‘Solubility. xii. regular solutions’, Journal of the American Chemical Soci-
ety, 1929, 51, 66–80.
99. E. A. Guggenheim: ‘The statistical mechanics of regular solutions’, Proceedings fo the Royal
Society A, 1935, 148, 304–312.
100. E. A. Guggenheim: Mixtures: Oxford, U. K.: Oxford University Press, 1952.
101. J. W. Christian: Theory of Transformations in Metals and Alloys, Part I: 2 ed., Oxford, U. K.:
Pergamon Press, 1975.
102. R. Kikuchi: ‘A theory of cooperative phenomena’, Physical Review, 1951, 81, 988–1003.
103. R. B. McLellan, and W. W. Dunn: ‘A quasichemical treatment of interstitial solid solutions:
its application to carbon in austenite’, Journal of Physics and Chemistry of Solids, 1969, 30,
2631–2637.
104. W. W. Dunn, and R. B. McLellan: ‘The application of a quasichemical solid solution model
to carbon in austenite’, Metallurgical Transactions, 1970, 1, 1263–1265.
105. S. Ban-Ya, J. F. Elliott, and J. Chipman: ‘Activity of C in Fe-C alloys at 1150 C’, Trans.
Metall. Soc. AIME, 1969, 245, 1199–1206.
106. S. Ban-Ya, J. F. Elliott, and J. Chipman: ‘Thermodynamics of austenitic Fe-C alloys’, Metal-
lurgical Transactions, 1970, 1, 1313–1320.
107. K. Alex, and R. B. McLellan: ‘A zeroth order mixing treatment of interstitial solid solutions’,
Journal of the Physics and Chemistry of Solids, 1970, 31, 2751–2753.
108. H. K. D. H. Bhadeshia: ‘Quasichemical model for interstitial solutions’, Materials Science
and Technology, 1998, 14, 273–276.
109. W. W. Dunn, and R. B. McLellan: ‘The thermodynamic properties of carbon in body-centered
cubic iron’, Metallurgical Transactions, 1971, 2, 1079–1086.
110. R. B. McLellan: ‘The thermodynamics of dilute interstitial solid solutions’, In: P. S. Rudman,
and J. Stringer, eds. Phase Stability in Metals and Alloys. New York, USA: McGraw Hill,
1967:393–417.
111. L. Kaufman, and H. Nesor: ‘Coupled phase diagrams and thermochemical data for transition
metal binary systems - IV’, CALPHAD, 1978, 2, 295–318.
112. K. Oda, H. Fujimura, and H. Ino: ‘Local interactions in carbon-carbon and carbon-M (M: Al,
Mn, Ni) atomic pairs in fcc gamma-iron’, Journal of Physics – Condensed Matter, 1994, 6,
679–692.
113. A. L. Sozinov, A. G. Balanyuk, and V. G. Gavriljuk: ‘C-C interaction in iron-base austenite
and interpretation of Mössbauer spectra’, Acta Materialia, 1997, 45, 225–232.
114. V. M. Nadutov, L. A. Bulvain, and V. M. Garamus: ‘Analysis of small-angle neutron scatter-
ing from Fe–18Cr–10Mn–16Ni–0.5N and Fe–21Cr–10Mn–17Ni–0.5C austenites’, Materials
Science & Engineering A, 1999, 264, 286–290.
115. H. K. D. H. Bhadeshia: ‘Application of first-order quasichemical theory to transformations in
steels’, Metal Science, 1982, 16, 167–169.
116. Y. Mou, and H. I. Aaronson: ‘The carbon carbon interaction energy in alpha Fe-C alloys’,
Acta Metallurgica, 1989, 37, 757–765.
117. J. C. Fisher, J. H. Hollomon, and D. Turnbull: ‘Kinetics of the austenite to martensite trans-
formation’, Metals Transactions, 1949, 185, 691–700.
118. J. H. Jang, H. K. D. H. Bhadeshia, and D. W. Suh: ‘Solubility of carbon in tetragonal ferrite
in equilibrium with austenite’, Scripta Materialia, 2012, 68, 195–198.
119. A. G. Khachaturyan, and G. A. Shatalov: ‘Elsatic-interaction potential of defects in a crystal’,
Soviet Physics - Solid State, 1969, 11, 118–123.
120. A. G. Khachaturyan, and G. A. Shatalov: ‘Theory of ordering of carbon-atoms in a martensite
crystal’, Physics of Metals and Metallography, 1971, 32, 5–13.
121. A. G. Khachaturyan: Theory of phase transformations and structure of solid solutions: New
36 REFERENCES

York, USA: John Wiley & Sons, Inc., 1983.

122. D. A. Mirzayev, A. Mirzoev, and P. V. Chirkov: ‘Ordering of carbon atoms in free martensite
crystal and when enclosed in elastic matrix’, Metallurgical & Materials Transactions A, 2016,
47, 637–640.
123. P. V. Chirkov, A. A. Mirzoev, and D. A. Mirzaev: ‘Role of stresses and temperature in the
Z ordering of carbon atoms in the martensite lattice’, Physics of Metals and Metallography,
2016, 117, 1138–1143.
124. L. Kaufman: ‘The stability of metallic phases’, Progress in Materials Science, 1969, 14, 57–
96.
125. T. G. Chart, J. F. Counsell, G. P. Jones, W. Slough, and P. J. Spencer: ‘Provision and use of
thermodynamic data for the solution of high-temperature practical problems’, International
Metals Reviews, 1975, 20, 57–82.
126. M. Hillert: ‘Prediction of iron base phase diagrams’, In: D. V. Doane, and J. S. Krikaldy,
eds. Hardenability concepts with applications to steels. Warrendale, Pennsylvania, USA: The
Metallurgical Society of AIME, 1977:5–27.
127. I. Ansara: ‘Comparison of methods for the thermodynamica calculation of phase diagrams’,
International Metals Reviews, 1979, 1, 20–53.
128. G. Inden: ‘The role of magnetism in the calculation of phase diagrams’, Physica B, 1981, 103,
82–100.
129. K. Hack, ed.: The SGTE Casebook: Thermodynamics at work: The Institute of Materials,
London, 1996.
130. M. Hillert: ‘Empirical methods of predicting and representing thermodynamic properties of
ternary solution phases’: Tech. Rep. 0143, Royal Institute of Technology, Stockholm, Sweden,
1979.
131. M. Hillert, and L. I. Staffansson: ‘The regular solution model for stoichiometric phases and
ionic melt’, Acta Chimica Scandinavica, 1970, 24, 3616–3626.
132. M. Temkin: ‘Mixtures of fused salts as ionic solutions’, Acta Phsicochimica URSS, 1945, 20,
411–417.
133. M. H. M. Waldenström: ‘A thermodynamic analysis of the Fe-Mn-C system’, Metallurgical
& Materials Transactions A, 1977, 8, 5–13.
134. B. Uhrenius, and H. Harvig: ‘A thermodynamic evaluation of carbide solubilities in the Fe-
Mo-C, Fe-W-C, and Fe-Mo-W-C systems at 1000 C’, Metal Science, 1975, 9, 67–82.
135. B. Sundman, and J. Ågren: ‘A regular solution model for phases with several components
and sublattices, suitable for computer applications’, Journal of the Physics and Chemistry of
Solids, 1981, 42, 297–301.
136. L. F. Bates: Modern Magnetism: Cambridge, U. K.: Cambridge University Press, 1963.
137. M. V. Nevitt, and A. T. Aldred: ‘Ferromagnetism in V-Fe and Cr-Fe alloys’, Journal of Applied
Physics, 1963, 34, 463–468.
138. G. Inden, and W. O. Meyer: ‘Approximate determination of the Curie temperatures of bcc
Fe-Co alloys’, Zietschrift für Metallkunde, 1975, 66, 725–727.
139. K. G. Denbigh: The thermodynamics of the steady state: New York, USA: John Wiley &
Sons, Inc., 1955.
140. L. Onsager: ‘Reciprocal relations in irreversible processes – i’, Physical Review, 1931, 37,
405–426.
141. D. G. Miller: ‘Thermodynamics of irreversible processes: The experimental verification of the
Onsager reciprocal relations.’, Chemical Reviews, 1960, 60, 15–37.
142. J. Z. Jiang, U. Gonser, C. Gente, and R. Bormann: ‘Mössbauer investigations of mechnical
alloying in the Fe-Cu system’, Applied Physics Letters, 1993, 63, 2768–2770.
143. J. Z. Jiang, U. Gonser, C. Gente, and R. Bormann: ‘Thermal stability of the unstable fcc-FeCu
phase prepared by mechanical alloysing’, Applied Physics Letters, 1993, 63, 1056–1058.
NOTES 37

144. J. G. Gay, and R. Richter: ‘Spin anisotropy of ferromagnetic films’, Physical Review Letters,
1986, 56, 2728–2731.
145. J. G. Gay, and R. Richter: ‘Spin anisotropy of ferromagnetic slabs and overlayers’, Journal of
Applied Physics, 1987, 61, 3362–3365.
146. J. H. Perepezko, and J. S. Paik: ‘Thermodynamic properties of undercooled liquid metals’,
Journal of non-crystalline solids, 1984, 61, 113–118.
147. G. J. Shiflet, J. R. Bradley, and H. I. Aaronson: ‘A re-examination of the thermodynamics of
the proeutectoid ferrite transformation in steels’, TMS-AIME, 1978, 9A, 999–1008.
148. A. H. Cottrell: ‘Constitutional vacancies in NiAl’, Intermetallics, 1995, 3, 341–345.

Notes
1 Hildebrand’s definition: “A regular solution is one involving no entropy change when a small amount of one of its

components is transferred to it from an ideal solution of the same composition, the total volume remaining unchanged” [98].
1 This equation reduces to the familiar

∆SM = −kNa [(1 − x) ln{1 − x} + x ln{x}] (6.10)

when mA = mB = 1.
1 In the Peltier effect, the two junctions of a thermocouple are kept at the same temperature but the passage of an electrical

current causes one of the junctions to absorb heat and the other to liberate the same quantity of heat. This Peltier heat is
found to be proportional to the current.