Submitted to Phys. Rev.

Volume Effect in the Landau Theory of Martensitic

Phase Transitions in Cubic Crystals

M.A.Fradkin†
arXiv:cond-mat/9406030v2 7 Jun 1994

Institute of Crystallography, Russian Academy of Sciences

59 Leninsky prospect, Moscow, 117333, Russia

(June 5, 1994)

Abstract

An effect of the volume change upon proper ferroelastic (martensitic) phase

transitions in cubic crystals is considered. Corresponding terms in the

Ginzburg-Landau expansion of the Gibbs free energy are analyzed for the

first- as well as second-order phase transitions from cubic to tetragonal lat-

tice under the action of uniaxial and hydrostatic pressure. The pressure effect

on the critical temperature as well as on the phase transition anomalies of

isothermal compressibility and linear thermal expansion coefficient are stud-

ied and recent experimental data on thermal expansion anomalies in V3 Si,

In-Tl and Ni-Al are discussed. The non-linearity of thermal expansion leads

to the special relation between the shear strain and volume change as a re-

sult of the elastic energy minimization. This phenomenon can provide the

transformation from FCC lattice to BCC one, observed in the iron alloys.

PACS: 05.70.Fh, 64.10.+h, 81.30.Kf

Typeset using REVTEX

 I. INTRODUCTION

There is a growing attention in recent years to the physics of martensitic phase transfor-

mations in metal alloys [1,2], though these phenomena were known by materials scientists for
many years as diffusionless transformations characterized by specific transformation kinetics
[3] not described by the classical nucleation theory [4]. Two different kinds of martensitic
transformations are known, i.e., an athermal and isothermal ones. In the athermal case

the transformation begins at some start temperature Ms , but the parent phase still exists
until the temperature goes down to Mf - a martensite finish point. In the isothermal case
the transformation proceeds in time at a constant temperature and, generally speaking,
could be completed in some finite time, which might be very long and depends in turn
on temperature. There is a particular sub-class of a so-called ”thermo-elastic” martensite

within this class of transformation, which is characterized by the reversibility of a structure

change - the alloy regains its high-temperature structure upon heating from a martensitic
low-temperature state through the transition point. This means that the lattice of product
phase is coherent with respect to the parent one. This phenomenon gives the ground for the

”shape-memory effect” [5] and is, thus, of a great practical importance.

Due to a spontaneous strain release during the transformation some elastic distortions
appear in the matrix of the parent phase surrounding the martensite nuclei. The mini-
mization of elastic long-range energy determines the shape of new phase precipitates and

provides an equilibrium state that has complicated heterogeneous structure which involves
multiple-twin bands for deeper relaxation of the elastic strain [6]. Thus, a real development
of transformation takes place in complicated conditions of apriory unknown external pres-
sure from the lattice of the parent phase upon the regions under transformations. Thus,
the heterogeneity of the system makes it difficult to analyze the thermodynamic of phase

transition, and idealized single-crystal systems should be considered in order to study an

equilibrium structure of the martensite phase as well as an equilibrium development of the
transformation.

2
 The reversibility of the transformation means that a phase transition between equilibrium
phases takes place and the structure coherence as well as the symmetry breaking at a critical
temperature implies that the transition can be analyzed within the frame of the Landau

theory. In this approach the difference in free energy between the parent and product
phase is considered as a function of some order parameter which is equal to zero in a high-
temperature symmetrical phase and becomes non-vanishing below the critical temperature
[7]. The fundamental difference of the structure of a martensitic phase with respect to the

parent lattice is well known to be spontaneous strain [6], so this phase transition belongs to
the class of proper ferroelastics [8]. It means that the structure change takes place through
the elastic instability of the crystalline lattice of a parent high-temperature phase with
respect to a spontaneous homogeneous strain of a special kind [9]. In other words, some

combination of the elastic modulii vanishes at the critical temperature.

It gives rise to the drastic lowering of the frequency of a corresponding mode of the
acoustic vibration leading to a central peak of the inelastic neutron scattering [10] and this
is the reason why this sort of phase transition is often referring to as a ”soft-mode” one.

Though, the modes never become completely soften [11] and a finite (albeit – small) jump of
the order parameter is usually observed, the martensitic phase transition still may be treated
as a weakly-discontinuous and considered in the frame of the Landau theory of a continuous
phase transition. A corresponding Ginzburg-Landau expansion of the free energy in series
of the symmetrized strain components has been developed [12–14] and the heterogeneous

fluctuations [15], nucleation of the martensite phase around the defects [16] and some other
phenomena [17] were studied in such a formalism.
In the present paper the role of hydrostatic strain in the martensitic phase transitions
is considered for the case of a cubic lattice of a high-temperature phase. A cubic point

symmetry leads to a coupling between the shear strain and volume change in the elastic
energy expansion near the critical temperature, and the volume change can be considered
as an additional order parameter not related with the symmetry breaking. The coupling is
shown to lead to the transition anomaly in the thermal expansion coefficient as well as in

3
the isothermal compressibility. For the case of the first-order transition some effect is found
to occur in the low-temperature phase even at the temperature region outside fluctuation-
induced singularities.

The effect of the uniaxial pressure, that preserves the symmetry of the low-temperature
tetragonal phase is studied as well. It is an external field conjugated to the order-parameter
which is known [7] to suppress the continuous (second-order) phase transition for arbitrary
small field value, however, the weakly-discontinuous transition is preserved under the field

lower than the critical one. The dependence of the transition temperature on the external
uniaxial as well as hydrostatic pressure is derived and the critical pressure is found.
The non-linear analysis of the volume change shows, that the elastic energy minimization
can provide the FCC structure of low-temperature phase for the BCC parent lattice through

the special relationship between the shear strain and volume change. Hence, non-linear elas-
tic effects might be responsible for the challenging FCC – BCC martensitic transformation
in pure iron and some ferrous alloys.
The paper is organized as follows. A brief description of a weakly-first-order transition

within the Landau theory is a content of the Section II. The Landau theory of a proper
ferroelastic (martensitic) transformation is analyzed in Section III. The volume change due
to such a phase transition is considered in Section IV. Non-linear thermal expansion and its
possible contribution to the martensitic phase transition is studied in Section V.

II. THE LANDAU THEORY OF THE FIRST-ORDER PHASE TRANSITION

A. Continuous phase transition

Let us recall briefly main ideas of the Landau theory of continuous phase transitions [7].
As Landau supposed, if the symmetry group of low-symmetry phase G1 is a subgroup of

the symmetry group G0 of the high-symmetry one, than there is some variable η, called as
an ”order parameter” which is invariant under all the transformations from the G1 whereas

4
some transformations from G0 change it. The thermodynamic potential – the Gibbs free
energy could be expanded as a power series in η near the critical temperature. As the
thermodynamic potential should not change under the symmetry transformations which do

not change the structure, the η in the high-temperature phase should vanish.
General expression for the (Ginzburg-Landau) expansion of the difference in Gibbs free
energy between the phases has the following form [7]

α C
∆G = (T − Tc )η 2 + η 4 (2.1)
2 4

where Tc is a critical temperature and the coefficients α and C should be positive. The
equilibrium value of η is determined by the minimization of ∆G with respect to η:

∂∆G ∂ 2 ∆G
=0 and > 0. (2.2)
∂η ∂2η

The solutions are the high-symmetry phase with η = 0, stable for T > Tc and low-

temperature phase with η 2 = α(Tc − T )/C that is stable for T < Tc . The dependence
of η on T is continuous in the critical point Tc , hence this model describes the second-order
phase transition. The Gibbs free energy as well as entropy changes continuously trough the
transition temperature Tc

α2
∆G = − (T − Tc )2 (2.3)
4C

∂∆G α2
∆S = − = (T − Tc ) , (2.4)
∂T 2C

but the derivative manifests discontinuity

∂∆S ∂ 2 ∆G α2 Tc
∆CP = T = −T = . (2.5)
∂T ∂T 2 2C

Vanishing of the coefficient of second degree in the Ginzburg-Landau expansion (2.1)

when the temperature approaches Tc leads to the critical fluctuations of the order parame-
ter. Mean square of the homogeneous order parameter fluctuations is given by well-known
expression [7]

5
 kB T
hη 2i ∝ . (2.6)
α |T − Tc |−1

Inhomogeneous fluctuations appear to be crucial in many cases, however, for the purposes
of present study they can be neglected due to long-range nature of elastic interactions in
solids [18].

B. External field effect

Let us consider the effect of the external field E conjugated to the physical variable of the
order parameter. In what follows it is an external pressure conjugated to the symmetrized
spontaneous strain. The Ginzburg-Landau expansions takes the form

α C
∆G = (T − Tc )η 2 + η 4 − Eη (2.7)
2 4

and the minimization of ∆G with respect to η leads to the cubic equation

∂∆G
= α(T − Tc )η + Cη 3 − E = 0 (2.8)
∂η

with discriminant
!3 2
α(T − Tc ) E

Q= + (2.9)
3C 2C

It is seen that for any value of external field E the high-symmetry phase with η = 0 no longer

provides the stable solution of the Eq.(2.2). Instead, we get η 6= 0 for any temperature. It
is known [19], that the cubic equation has one solution in real numbers for Q > 0 and three
ones for the case of Q < 0. It means that for temperatures below some critical one that now
depends on E

3  1
T0 (E) = Tc − 2CE 2 3 (2.10)
2α

the additional minimum of the Gibbs free energy appears that corresponds to new phase.
However, the initial phase described by the high-temperature solution of (2.8) provides the
minimum with lower value of the free energy and is, thus, stable. The free energy behavior

6
as well as the order parameter dependencies on the temperature for different values of the
external field are shown in the Fig.1 and Fig.2.
It is seen from the Fig.1 and might be proven rigorously that different minima of the

∆G(η) curve have different energies for any temperature T < T0 . Thus, the high-temperature
state remains stable throughout all the region of the phase co-existence. Only the condition
of E = 0 leads to the degeneracy with respect to sign of η that implies the equal energies
of different minima. It leads to a phase transition of the first order under the variation of

external field at constant temperature T < Tc . In other words, the variation of temperature
and external field act on the systems described by the expression (2.7) in a different way,
because the field variation may lead to the phase change but the temperature one may not.
The external field suppress the transition, however some decrease in

∂ 2 ∆G
(T, E) = α(T − Tc ) + 3Cη 2 (T, E)
∂η 2

leads to enhanced fluctuations of the order parameter around Tc with smooth peak instead
of divergence (2.6) shown in the Fig.(3). So, some transition anomalies are preserved in
sufficiently small external fields E.

C. Weakly-discontinuous phase transition in the Landau theory

The first-order phase transition arises in the Landau theory when the symmetry of the
system allows to have non-vanishing third-degree invariant composed by the order-parameter

component [7,20]. Corresponding term should be taken into account in the Ginzburg-Landau
expansion:

α B C
∆G = (T − Tc )η 2 + η 3 + η 4 − Eη. (2.11)
2 3 4

Choosing the case of B < 0, that gives positive η in the low-temperature phase, we can
write the free energy expansion in the following form

C3 τ 2 ζ3 ζ4
∆G̃ = ∆G = ζ − + − σζ , (2.12)
B4 2 3 4

7
with

B αC C2
η=− ζ , τ = 2 (T − Tc ) , and σ = − 3 E .
C B B

1. First-order transition in the absence of external field

For the σ = 0 case minimization of the Gibbs free energy (2.12) with respect to ζ implies

the equation

∂∆G̃
= τζ − ζ2 + ζ3 = 0 . (2.13)
∂η
1
For τ > τ0 = there is the only minimum at ζ = 0 corresponding to a high-temperature
4
 √ 
undistorted phase. The second minimum at ζ = 12 1 + 1 − 4τ , or
 
− 21
B  4αC

η=− 1 + 1 − 2 (T − Tc′ )  (2.14)
2C B

appears at T0 and corresponds to a low-temperature distorted phase which initially has

higher free energy. The phase energies become equal at τ∗ = 29 , though the supercooling of
the high-temperature state as well as superheating of the low-temperature one are possible.
It means that the first-order phase transition takes place at the temperature T∗ . As well as
in the second-order case, high-symmetry phase becomes unstable at τ = 0.

At the temperature of the first-order transition T∗ the order parameter jumps from the
η = 0 to the

2B
η=− , (2.15)
3C

overcoming the activation energy barrier

1 B4
∆Gb = . (2.16)
324 C 3

The entropy now has a finite change at transition temperature

2 αB 2
∆S = − , (2.17)
9 C2

8
that correspond to latent heat of the phase transition

2 αB 2
∆Q = T∗ . (2.18)
9 C2

In order the weakly-first-order transition to be properly identified and clearly separated

from the background of critical fluctuations around Tc (2.6), the transition jump of the order
parameter should be greater than its mean fluctuation. In other word, the energy scale of the
problem should be larger than the thermal fluctuation energy kB T . It implies the condition
for the value of the third-order coefficient in the Ginzburg-Landau expansion
 1
B ≥ C 3 kB Tc 4
. (2.19)

If this condition is not satisfied, the phase transition is ”too weak” to be first-order and will
be seen in the experiments as a continuous one.

2. Effect of the external field on the weakly-discontinuous phase transition

Substituting ζ = ζ̃ + 13 into the Ginzburg-Landau expansion (2.12), we get the third-order

term excluded [21]:

τ̃ 2 ζ˜4
∆G̃ = ζ̃ + − σ̃ ζ̃ + ∆G̃0 , (2.20)
2 4

where

1 τ 2
τ̃ = τ −; σ̃ = σ − +
3 3 27
τ σ 1
and ∆G̃0 = − − . (2.21)
18 3 108

This is equivalent to the free energy expansion (2.7) for the second-order phase transition
under the external field, the only difference consisting of the term ∆G̃0 that is independent

on ζ̃. It appears because the free energy is counted with respect to the (ζ = 0) or (ζ̃ = − 13 )
state, that implies ∆G̃0 = ∆G̃(ζ̃ = 0) 6= 0.
The condition (2.2) leads to cubic equation (2.8) with the effective temperature τ̃ and
field σ̃ instead of the real ones. The sign of discriminant

9
  3  2
τ̃ σ̃
Q= + ∝ 4σ + 27σ 2 − 18στ − τ 2 + 4τ 3 (2.22)
3 2

of this equation indicates, whether it has one root or three ones in real numbers. The latter
case corresponds to the appearance of different minima on ∆G̃(ζ̃), second minima of the free
energy appearing when Q(τ, σ) < 0.

Hence, (2.20) can be considered as the Ginzburg-Landau expansion for the phase transi-
tion between the states, related with different minima of the Gibbs free energy. The minima
have non-zero values of order parameter, because the symmetry is broken already by the
applied field for any temperature. As there is no symmetry breaking, it is not a true phase

transition, described by the Landau theory, however, the undistorted phase with ζ = 0
can be treated as an analog of ideal high-symmetry ”praphase” [22] that would allow the
symmetry reduction to both of the phases which provide minima of the free energy. It is
interesting to note that the first-order phase transition in absence of external field appears
to be equivalent to the second-order one under the action of ’effective’ external field, the

only feature of this situation is zero value of η for one of two minima of ∆G̃(ζ̃).
The phase diagram in (τ, σ) plane is shown at the Fig.4. Additional minimum of the free
energy appears for σ1 ≤ σ ≤ σ2 with

2  3
 τ
σ1,2 = − 1 ± (1 − 3 τ ) 2 + , (2.23)
27 3
3
4
that leads to the hysteresis with respect to the external field ∆σ = 27
(1 − 3 τ ) 2 .
According to an analogy with the second-order phase transition described by (2.7), the
different minima of the ∆G̃(ζ̃) have equal energy only at σ̃ = 0. This is the condition of

the first-order phase transition between corresponding phases and it determines the effect
of applied field on the transition temperature τ∗

2
τ∗ (σ) = 3 σ + . (2.24)
9

For σ = 0 we get naturally τ∗ (0) = 29 . The Eq.(2.24) corresponds to the straight line on
1
(τ, σ) plane (Fig.4). For τ > 3
on this line the equilibrium phase has ζ = 31 . This state is an

10
analog of the undistorted high-symmetry phase of the Landau theory without an external
1
field which is unstable for τ < 3
and becomes a maximum of ∆G̃, i.e. the energy barrier
with a height of

9 2 σ 1
Eb = σ − + ,
4 6 324
√
for the first-order transition between two different minima with ζ̃1,2 = ± −τ̃ , separated by
the order parameter discontinuity

2√
∆ζ̃ = ∆ζ = 1 − 27σ . (2.25)
3

As this line of the first-order transition in the phase diagram at (τ, σ) plane sep-
arates states without symmetry-breaking relationship [7], it terminates in critical point
1 1
(τc = 3
, σc = 27
). The discontinuity in order parameter as well as the potential barrier
separating different minima of the free energy vanishes when approaching this critical point.

There is no transition for σ > σc or τ > τc , that means suppressing the weakly-first-order
phase transition under the fields greater than the critical one. In a contrast with the second-
order case where arbitrary small external field destroys the phase transition, here we find
that the fields lower than σc preserve the transition. The critical point is an analog of

the continuous phase transition from state with ζ̃ = 0 corresponding to the breaking of
symmetry with respect to change of the ζ̃ sign.

III. PROPER FERROELASTIC PHASE TRANSITION

A phase transition characterized by the appearance of spontaneous strain at the tran-

sition temperature is called ferroelastic [8]. When this spontaneous strain describes the
symmetry breaking at the transition, and is, thus, an order parameter, the proper ferroelas-

tic transition takes place. For the case of improper ferroelastics the spontaneous strain is a
complimentary order parameter, coupled with the primary one.
The free energy difference between the parent and product phases for the case of proper
ferroelastic transition is due to the elastic strain and corresponds to the Ginzburg-Landau

11
expansion of the elastic energy in series of the strain components [12]. The second-order
term in the Ginzburg-Landau expansion is a linear combination of the second-order elastic
constants that vanishes at the critical temperature. It is the eigenvalue of the lattice stiffness

matrix corresponding to the relevant irreducible representation of the symmetry group of

the high-temperature phase [7]. The strain tensor components transforming with respect to
this representation form the order parameter and the phase transition belongs to so-called
”soft-mode” class, because it is accompanied by a noticeable softening of the corresponding

acoustic mode of atomic vibrations [18], visible as a central peak of the inelastic neutron
scattering.

A. Spontaneous strain in cubic lattice

In what follows the case of cubic symmetry of a high-temperature phase will be consid-

ered, that describes the martensitic transformations in the A15 compound [23] as well as in
some metallic alloys. The spontaneous strain tensor has only diagonal components and the
order parameter is composed by their symmetrized linear combinations [12]

1
η1 = √ (−ǫxx − ǫyy + 2ǫzz ) (3.1)
6
1
η2 = √ (ǫxx − ǫyy ) . (3.2)
2

The η1 corresponds to the extension along z axis without the volume change and η2 describes
the strain non-tetragonality in XY plane. The critical acoustic mode is a transverse phonons
distributing in h110i directions.

The elastic free energy expansion can be written in the Ginzburg-Landau form [24]

A 2 B C
∆G = (η1 + η22 ) + η1 (η12 − 3η1 η22 ) + (η12 + η22 )2 . (3.3)
2 3 4

with the following combination of the elastic constants as the coefficients [13]

A α 1
= (T − Tc ) = (C11 − C12 ) (3.4)
2 2 2

12
 1
B = √ (C111 − 3C112 + 2C123 ) (3.5)
6 6
1
C= (C1111 + 6C1112 − 3C1122 − 8C1123 ) . (3.6)
48

Substituting

η1 = η sin θ and η2 = η cos θ ,

we get the Ginzburg-Landau expansion in the form

α B C
∆G = (T − Tc ) η 2 − η 3 sin(3θ) + η 4 . (3.7)
2 3 4

The minimization with respect to θ

∂∆G ∂ 2 ∆G
= −Bη 3 cos(3θ) = 0 and >0
∂θ ∂2θ

implies sin(3θ) = ±1 depending on the sign of B. In the case of B < 0 we get three solutions
√ √
1 3 1 3
(η, 0) , (− η, η) and (− η, − η) , (3.8)
2 2 2 2

corresponding to the low-symmetry tetragonal phases obtained by the extension of parent

cubic lattice along three coordinate axis. The free energy dependence on η for these minima is

the single-component Ginzburg-Landau expansion (2.11) for the weakly-discontinuous phase

transition. As the solutions are related through the symmetry transformation from the cubic
point symmetry group and, thus, are completely equivalent, we can consider further only
one of them, e.g. (η, 0), without loss of generality. Hence, we can use the Landau theory for

the case of single-component order parameter.

The cubic symmetry allows the third-order term in the Ginzburg-Landau expansion to
appear, hence, the Landau theory says that the phase transition should be of the first order.
Indeed, for the case of Ni-Al and some other systems partial mode softening takes place and

the finite strain appears at the transition, though the shear modulus decreases considerably
near the phase transition temperature. Some other so-called ”pre-transformation” phenom-
ena were found in a number of alloys [25–28]. This case respects to the weakly first-order
transition mentioned in Section II C.

13
 However, this third-degree term appears to be very small in In-Tl, V3 Si and some other
alloys [29] where the critical mode becomes almost complete soften, i.e. shear modulus
(C11 − C12 ) vanishes as temperature goes to Tc . Central peak of the inelastic neutron

scattering as well as other critical phenomena appear and the order parameter undergoes
very small change at the critical temperature Tc [30]. So, the transition is very weakly
discontinuous, being sometimes of the second order within the experimental accuracy. The
considerable enhancement in the acoustic wave magnitude occurs in the vicinity of the critical

temperature Tc . Hence, this case corresponds to the the second-order phase transformation
considered in Section II A.

B. Effect of external stress on the transition

An applied pressure gives rise the ’external’ stress tensor Ê corresponding to linear term

−ǫ̂ Ê in the free energy expression [31]. The symmetry-breaking strain components are
sensitive only to diagonal components of Ê, hence, relevant external stress is superposition
√
of hydrostatic pressure P = Tr(Ê)/ 3 with ones applied (uniaxially) along z axis

√
E1 = (−Exx − Eyy + 2Ezz )/ 6 (3.9)

and within XY plane

1
E2 = √ (Exx − Eyy ).
2

Hence, E1 and E2 are external fields, conjugated to the primary order parameter compo-
nents η1 and η2 , respectively, whereas hydrostatic pressure will be shown below affect the
ferroelastic phase transition through the volume change η0 related with the order parameter
by the coupling term proportional to η0 (η12 + η22 ).
Applied non-hydrostatic pressure E1,2 breaks the symmetry of undistorted phase, moving

out corresponding minimum of ∆G(η1 , η2 ) from the origin and lifts the degeneracy between
three low-symmetry phases (3.8). The pressure which is co-aligned with the spontaneous

14
strain corresponding to one of these solutions, e.g. (η1 , 0), preserves the tetragonal symmetry
of distorted phase and the pressure value E1 is an external field conjugated to the value of
symmetrized strain η1 as a single-component order parameter, considered in Section II C 2.

Otherwise the stable low-temperature state of the system has rhombohedral lattice with
three different lattice parameters that is characterized by the pressure-dependent η1 and
η2 . In what follows the effect of uniaxial pressure E1 conjugated to η1 , which preserves the
tetragonal symmetry of low-temperature phase is analyzed.

In the agreement with the analysis of Section II B the uniaxial pressure was found [32]
to suppress the ferroelastic phase transition from cubic to tetragonal lattice in the V3 Si
compound, where third-order term is very small and the transition is of the second-order. For
Ni-Al alloy, where the first-order phase transition takes place, the uniaxial pressure appears

[33] to shift the transition temperature linearly in complete agreement with the Eq.(2.24).
It should be noted that though this alloy exhibit martensitic transition where spontaneous
homogeneous strain is accompanied by so-called shuffle [11] related with q 6= 0 critical mode,
the central peak of the inelastic neutron scattering as well as noticeable softening of C11 −C12

elastic constant appear well above the transition temperature. Hence, this case can also be
analysed in the frame of the Landau theory of ferroelastic phase transition.

IV. VOLUME CHANGE IN THE ELASTIC ENERGY EXPANSION

If the phase transition is sensitive to applied external hydrostatic pressure then there is

a difference in the volume of elementary cell of the parent and product phases. The volume
change, indeed, takes place in some cases and it was shown [34] that virtual volumetric
strain could reduce the potential barrier for the system to overcome in the phase transition
development. In order to analyze the volume change one needs to include corresponding
terms into the Ginzburg-Landau expansion of the Gibbs free energy. Then the equilibrium

state of the system should provide the minimum of the free energy with respect to both
shear strain and volumetric one.

15
 A. Linear energy of the thermal expansion

In the linear elasticity theory [31] the trace of strain tensor is a measure of the volume
change, the corresponding symmetrized linear combination of the strain components having
a form

Tr(ǫ̂) 1
η0 = √ = √ (ǫxx + ǫyy + ǫzz ). (4.1)
3 3

Due to thermal strain as well as external pressure, the expansion of the elastic energy in
series of η0 should start with the linear terms [31]:

A0 2
∆G0 (η0 ) = −κ0 A0 (T − TR )η0 + η + P η0 , (4.2)
2 0

where TR is some reference temperature for volume change, κ0 and A0 are the volume thermal
expansion coefficient and bulk modulus, respectively and P is external pressure applied. The
bulk modulus A0 = √1 (C11 + 2C12 ) is always positive.
3

The minimization of ∆G0 (η0 ) with respect to η0 implies equilibrium values of η0 (T ) and

∆G0 (T, P )

P
η0 = κ0 (T − T0 ) − (4.3)
A0

A0 2 P2
∆G0 = − α0 (T − T0 )2 − + κ0 P (T − T0 ) (4.4)
2 2A0

It is easy to see that usual thermodynamical expressions for isothermal compressibility

βT and thermal expansion coefficient

∂ 2 ∆G0 ∂η0 1
βT = − =− = (4.5)
∂P 2 ∂P A0

∂ 2 ∆G0 ∂η0
κ0 = = (4.6)
∂P ∂T ∂T

are satisfied for such an expression of the Gibbs free energy if one bears in mind that P is
opposite of the pressure inside the system, which is a thermodynamical variable (we also
put the volume of whole system equal to 1 for convenience).

16
 B. Effect of the volume change on ferroelastic phase transition

The lowest-order term of coupling between the shear strains η1,2 and volume change η0
for the case of a cubic symmetry of the high-temperature phase is [13]

∆Gint = Dη0 (η12 + η22 ), (4.7)

1
where D = √
2 3
(C111 − C123 ). Substituting this term into the expansion of elastic energy in
a power series of the volume change, we get the expression (4.3) for the equilibrium value of
η0 in the following form
P D 2
η0 = κ0 (T − T0 ) − − (η1 + η22 ) . (4.8)
A0 A0
The coupling term does not induce any anisotropy in the (η1 , η2 ) plane, hence, three
equivalent minima of the Gibbs free energy at low temperature have a form (3.8). We
consider one particular solution of the form (η1 , 0). The uniaxial pressure E1 applied along

Z axis does not break the tetragonal symmetry of corresponding low-temperature phase.
Thus, it is an external field conjugated to the η1 order parameter and we can use the results
of Section II C 2 to study its influence on the phase transition. The volume change η0 as well
as its derivatives – thermal expansion coefficient α and isothermal compressibility βT does

not depend directly on E1 , but only follows the dependence of symmetrized shear strain η1
through Eqs.(4.8).
The general expression for the free energy has the following form

∆G = ∆G0 + ∆G1 + ∆Gint (4.9)

where ∆G1 is the Ginzburg-Landau expansion (2.11) with respect to η1 as the single-
component order parameter with included effect of applied uniaxial pressure E1 (3.9)
α1 B1 3 C1 4
∆G1 = (T − Tc )η12 + η + η − E1 η1 . (4.10)
2 3 1 4 1
Substituting (4.8) with η2 = 0 into Eq.(4.9) and taking Tc as a reference temperature TR
for the volume change, we get the following renormed Ginzburg-Landau expansion of ∆G in
the power series of η1

17
 A′ (T, P ) 2
∆G(T, P, η1 ) = ∆G0 + η1 +
2
B′ 3 C ′ 4
η + η − E1 η1 , (4.11)
3 1 4 1

with the coefficients

A′ (T, P ) = α′ (T − Tc′ ) =
2P D
 
= (α1 + 2κ0 D)(T − Tc ) − , (4.12)
A0

2D 2
B ′ = B1 and C ′ = C1 − . (4.13)
A0

Here ∆G0 (T, P ) is the energy of a high-symmetry phase with η1 = 0, given by the (4.4).
The critical temperature

2DP
Tc′ = Tc + (4.14)
A0 (α1 + 2κ0 D)

is shifted by the applied hydrostatic pressure

dTc 2D
= . (4.15)
dP A0 (α1 + 2κ0 D)

in agreement with experimental studies, e.g. in In-Tl alloys [35]. New stiffness is given by
the formula

α′ = α1 + 2κ0 D . (4.16)

In order the Eq.(4.11) be considered as an analog of Eq.(2.11), the α′ and C ′ should be

positive.
The inequality C ′ < C1 corresponds to the smoothening of the potential relief when
additional degrees of freedom appear, which allows the system to relax easily. It should be

noted that the inclusion of the terms corresponding to the possible volume change into the
Ginzburg-Landau expansion (4.10) does not affect the third-order term B1 , hence, the order
of phase transition could not be changed by applied hydrostatic pressure.

18
 As the hydrostatic pressure change the transition temperature according to Eq.(4.15),
the line (2.24) of the first-order phase transition becomes a surface in the 3D (T, P, E1) phase
diagram given by a formula

2B12 2D 3C ′
T∗′ (P, E1) = Tc + ′ ′ + ′ P − ′ E1 . (4.17)
9α C α A0 α B1

The transition can be induced by variation of hydrostatic pressure under the fixed values of
temperature T and uniaxial pressure E1 at the point

α′ A0 A0 B12 3A0 C ′
P∗ (T, E1 ) = (T − Tc ) − + E1 . (4.18)
2D 9DC ′ 2DB1

The critical point of the end of transition line (2.24) is now a line in the (T, P, E1 ) phase
diagram given by the uniaxial pressure Ec = −B13 /(27C ′ 2 ) and critical hydrostatic one that

depends linearly on the temperature

!
A0 B2
Pc (T ) = α′ (T − Tc ) − 1′
2D 3C

and vanishes when T goes to Tcp = Tc + B12 /(3α′ C ′ ). It means that some values of hydro-
static pressure suppress the transition, caused by the change of uniaxial pressure E1 . The

closer temperature is to the Tcp the smaller hydrostatic pressure needed for the transition
to disappear.

C. Transition anomalies of isothermal compressibility, thermal expansion coefficient

and specific heat

The isothermal compressibility, thermal expansion coefficient and specific heat are
expressed by Eqs.(4.5), (4.6) and (2.5), respectively, through the second derivatives of
∆G(T, P, η1 ) with respect to temperature and hydrostatic pressure for an equilibrium value

of η1 given by the condition (2.2) in the form

F (A′ , η1 ) = A′ η1 + B1 η12 + C ′ η13 − E1 = 0 (4.19)

For the first derivative we have the following formula

19
 ∂ ∂∆G ∂A′ ∂∆G ∂η1
(∆G − ∆G0 ) = + (4.20)
∂T ∂A′ ∂T ∂η1 ∂T

Second term in this expression vanishes as equilibrium η1 is given by the Eq.(2.2) and taking
into account Eq.(4.12) we get the formula

∂ η2
(∆G − ∆G0 ) = α′ 1 , (4.21)
∂T 2

that leads to the following expression for the transition anomaly of the specific heat

∂2 2 ∂η1
∆CP = −T (∆G − ∆G0 ) = −α′ T η1 (4.22)
∂T 2 ∂A′

For isothermal compressibility and thermal expansion coefficient we get analogous expres-
sions

∂2 4D 2 ∂η1
∆βT = − (∆G − ∆G0 ) = − 2
η1 (4.23)
∂P 2 A0 ∂A′
∂2 2Dα′ ∂η1
∆κ = (∆G − ∆G0 ) = − η1 . (4.24)
∂P ∂T A0 ∂A′

It is easy to see that the Keesom-Ehrenfest relationships [7]

dTc ∆βT T ∆α
= = (4.25)
dP ∆α ∆CP

are satisfied.
Differentiating both sides of Eq.(4.19) we get
!−1
∂η1 ∂F ∂F
=− ′
∂A ′ ∂A ∂η1

and resolving (4.19) with respect to A′ , we can finally obtain the following expression

∂η1 η13
η1 = − . (4.26)
∂A′ E1 + B1 η12 + 2C ′ η13

D. Second-order case

Let us consider the case of the martensitic phase transition of the second order for which
B1 = 0. In absence of the uniaxial pressure (4.26) does not depend on η1 and, hence, neither
on temperature nor on hydrostatic pressure:

20
 ∂η1 1
η1 ′
=− ′ .
∂A 2C

Low-temperature phase appears at Tc′ with continuous evolution of the order parameter
that gives the volume difference between the parent and product phases

α′ D
∆η0 = − (T ′ − T ) (4.27)
A0 C ′ c

and the discontinuities of the isothermal compressibility, thermal expansion coefficient and
specific heat at phase transition take the following forms
2
2 D 2D 2

∆βT = ′ = (4.28)
C A0 A20 C1 − 2A0 D 2

α′ D (α1 + 2κ0 D)D

∆κ = ′
= (4.29)
C A0 A0 C1 − 2D 2

Tc′ (α′)2 ′ A0 (α1 + 2κ0 D)

2
∆CP = = Tc (4.30)
2 C′ 2(A0 C1 − 2D 2 )
D
Thus, to find three independent parameters of the model - α′ , C ′ and A0
, we have four
measurable values – ∆CP , ∆βT and ∆α along with

dTc 2 D
= ′ , (4.31)
dP α A0

which are related by Eq.(4.25).

As was mention above, for the case of second-order ferroelastic phase transition consid-
erable critical fluctuations take place due to the softening of C11 − C12 shear modulus. The
inhomogeneous fluctuations η1 (x) appear to be relevant only in very close vicinity of the
critical temperature [18]. From Eq.(2.6) that describes the homogeneous order parameter

fluctuations taking into account Eq.(4.8) we get the following expressions for the critical
fluctuations of volume

D
η0 ∝ − |T − Tc |−1 (4.32)
A0

and thermal expansion coefficient

21
 ∂∆η0 D
κ= ∝ |T − Tc |−2 . (4.33)
∂T A0

in the temperature interval around Tc , where the fluctuations in η1 are important.

There are some experimental data available on the anomalies of thermal expansion in

the single-crystal specimens near the martensitic phase transition [36], which are in an
agreement with the above results. In most of the alloys the thermal expansion coefficient
increases near Tc′ , that implies the positiveness of D. Outside the temperature region of
thermal fluctuations the thermal expansion coefficient does not depend on the temperature.
Besides, linear model of Eq.(4.2) leads to the independence of α on the applied hydrostatic

pressure.
The second-order phase transition disappears under the applied external field E1 , hence,
CP , βT and κ manifest continuous behavior near the temperature Tc under arbitrary small
external field. However, from Eq.(4.26) we get in such a case
−1
∂η1 η13 E1

η1 =− ′ + η13 . (4.34)
∂A′ 2C 2C ′

For sufficiently small values E1 of external uniaxial pressure the difference in the isothermal

compressibility, thermal expansion coefficient and specific heat outside close vicinity of the
transition temperature appears to be described by Eqs.(4.28) - (4.30). The critical diver-
gencies given by Eqs.(4.32) and (4.33) disappear and the smeared peaks around Tc′ appear
instead.

E. The first-order transition

1. Absence of external uniaxial pressure

If the third-order coefficient in the Ginzburg-Landau expansion (4.10) has a non-zero

value, then the Eq.(4.11) in the absence of external uniaxial pressure describes the first-

order phase transition at the temperature

2 B12 2DP 2 B12

T∗′ = Tc′ + = Tc + + (4.35)
9 α′ C ′ A0 α′ 9 α′ C ′

22
The shift of the transition temperature by the applied hydrostatic pressure has the same
form (4.14) as for the second-order transition. The volume difference between the phases
appears to depend on the temperature through the temperature dependence of η1 given by

Eq.(2.14)
 !− 1 2
DB12  4α′C ′ 2
∆η0 = − 1 + 1 − (T − Tc′ )  . (4.36)
4A0 C ′ 2 B12

This leads to finite volume change at the phase transition temperature

4 B12
∆η0 (T∗′ ) = − (4.37)
9 (C ′ )2

that can be observed in diffraction as well as dilatometric studies.

Eq.(4.26) takes the form

∂η1 η1
η1 =− =
∂A ′ B1 + 2C ′ η1
 !− 1 
1 4α′ C ′ 2
= − ′ 1 + 1 − (T − Tc′ )  (4.38)
2C B12

and along with Eq.(4.24) gives the temperature dependence of difference in thermal expan-
sion coefficients between the low- and high-symmetry phases in the form
 !− 1 
α′ D  4α′C ′ 2
∆κ = ′ 1+ 1− (T − Tc′ )  . (4.39)
C A0 B12

The change of thermal expansion coefficient at the transition point T∗′ now has a form

4α′ D α1 + 2κ0 D
∆κ(T∗′ ) = ′
= 4D . (4.40)
C A0 A0 C1 − 2D 2

The value of ∆κ decreases to that given by Eq.(4.29) as T goes down from T∗′ to Tc′ .
For the difference in isothermal compressibility between low-temperature phase and high-

temperature one we can find similarly

 !− 1 
2D 2 4α′ C ′ 2
∆βT = ′ 2 1 + 1 − 2
(T − Tc′ )  (4.41)
C A0 B1

that gives us the phase transition discontinuity in the form

23
 8D 2
∆βT (T∗′ ) = . (4.42)
C ′ A20
It should always be positive according to general thermodynamical arguments [7].
The specific heat has the following temperature dependence
 !− 1 
T α′ 2  4α′ C ′ 2
∆CP = 1 + 1 − (T − Tc′ )  (4.43)
2 C′ B12
with the jump at the transition temperature
′2
′α A0 (α1 + 2κ0 D)2
∆CP (T∗′ ) = 2T∗ ′ = 2T∗′ (4.44)
C A0 C1 − 2D 2
The phase transition discontinuities of volume and entropy are related through the
Clapeyron-Clausius relationship with the slope (4.15 of the equilibrium line at phase di-

agram
dTc ∆η0 (T∗′ ) 2D
= = ′ . (4.45)
dP ∆S α A0
Both the thermal expansion coefficient and isothermal compressibility of undistorted
phase with η = 0 do not depend on temperature, so, the expressions (4.39) and (4.41) de-
scribe the temperature dependence of these quantities in the low-symmetry phase that can

be observed experimentally below the transition temperature T∗′ . It should be noted, how-
ever, that the effects can be seen for T∗′ being sufficiently far from Tc , outside the temperature
region where thermal fluctuations 2.6 are important, because the fluctuation-induced singu-
larities of the thermal expansion coefficient as well as the other quantities become larger in
critical region than the jumps in their equilibrium values. Thus, the third-order coefficient

B1 should satisfy condition (2.19).

Such a situation occurs in the case of Ni-Al and some other alloys where shear modulus
at T∗′ is soften only slightly – by 10 ÷ 20%, and the experimentally measured tempera-
ture dependence of shear modulus can be interpreted as pointing even to negative Tc [11].

However, for In-Tl system where third-order term appears to be very small and the shear
modulus almost vanishes at the transition temperature, the volume discontinuity is so small
[35] that it is hidden by thermal fluctuations (4.32). The similar effect occurs with respect
to other discontinuities at the transition temperature T∗′ , which is very close to Tc .

24
2. The effect of external uniaxial pressure on the anomalies around the first-order phase transition

For the case of the first-order phase transition the dependence of the discontinuity in the
order parameter on the uniaxial pressure E1 follows from general expression (2.25)
!1
2B1 27C ′2 E1 2

∆η1 = − ′ 1 + .
3C B13

that leads to the following volume change at T∗

! 12
′ D 2 4 DB12 27C ′ 2 E1
∆η0 (T∗ ) = − ∆(η1 ) = − 1+ (4.46)
A0 9 A0 C ′ 2 B13

Both parent and product phases have η1 6= 0 and ∆G 6= 0 under applied external uniaxial
pressure, thus, the differences in the isothermal compressibility, thermal expansion coefficient

and specific heat are proportional to

!
∂η1 ∂η1,1 ∂η1,2
∆ η1 = η1,1 − η1,2 ,
∂A′ ∂A ′ ∂A′

where η1,1 (T, P ) and η1,2 (T, P ) correspond to two different minima of the free energy given
by the different solutions of Eq.(4.19). Using the dimensionless variables (2.21) we can write
Eq.(4.26) in the form

∂η1 1 (ζ̃ + 13 )3
η1 = −
∂A′ C ′ σ − (ζ̃ + 31 )2 + 2(ζ̃ + 31 )3

At the transition point we have σ̃ = 0 and

√ 1√
ζ̃ = ± −τ̃ = 1 − 27σ,
3

that gives the expression

∂η1 (ζ̃ + 13 )2
η1 = . (4.47)
∂A′ 2C ′ ζ̃ 2

Taking into account that ζ̃ 2 has the same value −τ̃ at the transition point for both phases,

we can finally obtain

!
∂η1 2 − 12
∆ η1 =− (1 − 27 σ) . (4.48)
∂A′ C′

25
 From the Eqs.(4.22) - (4.24) we get the expressions for the phase transition discontinuities
of the specific heat, isothermal compressibility and thermal expansion coefficient as follows
!− 1
α′ 2 27C ′ 2 E1 2

∆CP (T∗′ ) = 2T∗′ ′ 1+ (4.49)

C B13

!− 1
8D 2 27C ′2 E1 2

∆βT (T∗′ ) = ′ 2 1+ (4.50)

C A0 B13

!− 21
4α′ D 27C ′ 2 E1
∆α(T∗′ ) = ′ 1+ . (4.51)
C A0 B13

In the limit of small values of external uniaxial pressure we get the Eqs.(4.44), (4.42)
and (4.40), derived from their temperature dependence in the Section IV E 1. When E1 goes
1
to the value of the critical point Ec , these discontinuities diverge as ∝ |Ec − E1 |− 2 .

V. TRANSFORMATION FROM FCC INTO BCC LATTICE VIA

SPONTANEOUS STRAIN

A. FCC – BCC transformation through the Bain strain

There is the case of martensitic transformation of especial interest, namely FCC – BCC
transformation in Fe and some ferrous alloys. Since, there is no group-subgroup relationship

for the symmetry breaking, the Landau theory is, generally speaking, inapplicable to this
case. However, there is an orientational relationship between lattices of the parent and
product phase, and the transition could be described in terms of spontaneous strain of
so-called Bain type [6].

The Bain strain is the single-axis shear of the same kind as an order parameter of
the ferroelastic phase transition from cubic to tetragonal lattice. It is accompanied by the
volume change, that is approximately 1.5% in the case of pure Fe, where transformation from
austenite FCC γ-phase to ferrite BCC α-one takes place at 910o C. If the lattice periods

26
for austenite and martensite (ferrite) are aγ and aα respectively, then the strain tensor
components have the form

√ aα aα
ǫxx = ǫyy = 2 − 1 and ǫzz = −1 . (5.1)
aγ aγ

The fundamental feature of this case as compared with the above considered phase
transition from cubic lattice to the tetragonal one, is the fixed value of the spontaneous
strain needed to get the symmetry properties of the low-temperature phase. In the above

considered case for any non-zero value of the order parameter η1 the symmetry of the lattice
was tetragonal, whereas in the case of the FCC – BCC transformation the peculiar value of
η1 is needed to get the low-temperature BCC lattice. If we separate the shear strain from
the volume change by taking the latter equal to zero, then we get single (and very large)

value of symmetrized shear strain (3.1)

√ √ 
6 2−1
η1 = − √ ≈ − 0.256 (5.2)
2 2+1

Hence, this phase transition is completely different from the continuous ones, which the
Landau theory describes, where the value of the order parameter changes with temperature

in low-symmetry phase according to the minimization of its Gibbs free energy (2.2).
However, coupling with the volume change η0 makes it possible to have the η1 variation
in low-temperature phase without breaking of its symmetry. Indeed, the strain tensor (5.1)
implies the following expressions for the symmetrized combinations used above as the order

parameter components
√
(2 2 + 1) aα √
η0 = √ − 3 (5.3)
3 aγ
s
2 √ aα
η1 = − ( 2 − 1) ; η2 = 0 , (5.4)
3 aγ

and we get the relationship between shear strain and volume change in the form
√
2 2+1 √
η0 = − √ η1 − 3. (5.5)
2− 2

27
Hence, the variation in value of η1 preserves the BCC structure of the low-temperature
phase, if η0 is changed in such a way that this relationship is satisfied. It should be noted
that (5.5) is meaningful only in restricted region of η0 and η1 . For example, η1 = 0 implies

unreal result aα = 0 from (5.4). Thus, (5.5) is justified only in some vicinity of the transition
√
that is characterized by small volume change 3 η0 .
Having supposed FCC – BCC transformation to be ferroelastic one, we should get the
minimum of elastic energy for the values of η0 and η1 obeying the condition (5.5). Let us

study what are the coefficient in the expansion which provide such a minimum. Without
careful analysis, it should be noted, however, that the linear approximation used above
gives another kind of relationship (4.8) between the shear strain and volume change, thus,
non-linear approximation for the thermal expansion energy should be used.

B. Non-linear elasticity for the volume change

Non-linearity arises naturally when taking into account large value of the strain tensor
component. The Tr(ǫ̂) for Bain strain in pure iron is approximately three times larger than
real value of the volume change for this transformation given by direct multiplication of the

lattice periods of low-temperature phase

δV
= (1 + ǫxx )(1 + ǫyy )(1 + ǫzz ) − 1. (5.6)
V

It could be expressed through the symmetrized combinations η0 , η1 and η2 as follows

δV √ η3 η 2 + η22
= 3 η0 + η02 + √0 − 1 −
V 3 3 2
η0 (η12 + η22 ) η1 (η12 − 3 η22)
− √ + √ . (5.7)
2 3 3 6

For the FCC – BCC transition we have η2 = 0 and proper account for the volume
change should, thus, involve the terms of higher order in η0 and η1 . Terms of the first
and second order in the volume change η0 within non-linear approximation have no longer

simple relation with the thermal expansion and isothermal compressibility that was obtained

28
in Section IV A. Similarly, other terms in both ∆G1 and ∆Gint should be changed. General
non-linear elastic energy expansion near the elastic instability with respect to η1 now has
the form

A0 2 B0 3
∆G = L0 η0 + η + η + D η0 η12
2 0 3 0
A1 B1 3 C1 4
+ η12 + η + η , (5.8)
2 3 1 4 1

where the coefficients B0 and C1 in highest-order terms should be positive and we again
consider the particular expression with η2 = 0.
The minimization with respect to η0 implies

∂∆G
= L0 + A0 η0 + B0 η0 2 + D η1 2 = 0 (5.9)
∂η0

that leads to the following relationship between η0 and η1 in distorted phase with η1 6= 0 :
2
A0 L0 A2 D 2

η0 + + − 02 = − η . (5.10)
2B0 B0 4B0 B0 1

For the high-symmetry phase we have

 !1 
A0  4B0 L0 2
η0 = − 1∓ 1−  (5.11)
2B0 A20

As there is no co-existing high-symmetry equilibrium states with different values of η0 , the

condition

A20 − 4L0 B0 = 0 (5.12)

should be satisfied. Substituting this expression into Eq.(5.10), we get

2
B0 A0

η12 =− η0 + (5.13)
D 2B0

In order the right-hand side of this expression to be positive condition D < 0 must be
satisfied, because of positiveness of B0 .

Finally, we can get the following expression for the minimum of the free energy (5.8)
s
D A0
η0 = − − η1 − (5.14)
B0 2B0

29
and Eq.(5.5) along with (5.12) lead to the following relations between the coefficients in the
elastic energy expansion (5.8)

L0 = 3B0 (5.15)

√
A0 = 2 3 B0 (5.16)
√
9+4 2
D=− √ B0 . (5.17)
3−2 2
These relations could be, generally speaking, satisfied only in isolated points on phase dia-
gram and the phenomenological approach used in the present study is unable to find their

origin. It could be done only in some microscopic theory beyond the scope of the paper.
However, as far as these relations are satisfied, we can try to find their consequences for the
elastic properties of the system under phase transition.
Substituting these expressions into the free energy (5.8) and excluding the volume change

through Eq.(5.5) we get renormed expansion of the elastic energy with respect to sym-
metrized strain η1
√ !
√ √18 + 8 2 η12
∆G = − 3 B0 + A1 + 3 B0 √
3−2 2 2
√ ! 3
815 + 580 2 η1 C1 4
+ B1 + B0 √ + η , (5.18)
116 − 41 2 3 4 1
which can be considered as a Ginzburg-Landau expansion for ferroelastic phase transition.
The first term does not depend on η1 , the critical temperature Tc is defined by the condition
√
√ 18 + 8 2
A1 + 3 B0 √ =0.
3−2 2
and the temperature T∗ of the first-order transition with finite jump in η1 and η0 is given

by an equation

9 A3 − 3 A B 2 + C1 B 2 = 0

where A and B are the expressions in brackets of the first- and second-degree terms in
Eq.(5.18). In order some transition line to exist on phase diagram, the B0 coefficient should
be temperature- and pressure-dependent.

30
 VI. CONCLUSIONS

We have analyzed the volume change effect on ferroelastic (martensitic) phase transitions

and considered the case of cubic lattice of high-symmetry phase as an example. The mini-
mization of elastic energy with respect to hydrostatic strain as a secondary order parameter
is shown to renorm the second- and fourth-order coefficients of the Ginzburg-Landau ex-
pansion of elastic free energy in powers of symmetrized shear strain. The coupling between

shear strain and volume change appears to shift the transition temperature under applied
external hydrostatic pressure and lead to the finite volume effect of the weakly discontinuous
ferroelastic phase transition.
The isothermal compressibility as well as thermal expansion coefficient is shown to di-
verge near the critical temperature of the second-order ferroelastic phase transition due to

the homogeneous fluctuations of the order parameter. The difference between their values in
parent and product phases outside the fluctuation region appears to be proportional to cou-
pling coefficient. For the case of first-order transition isothermal compressibility and thermal
expansion coefficient depend on the temperature in the low-symmetry phase according to

the square root law.

The uniaxial pressure conjugated to the symmetrized shear strain is shown to suppress
the second-order transition, leading to the change of divergencies for smeared peaks in the
temperature dependencies of isothermal compressibility and thermal expansion coefficient

around critical temperature. We have found the first-order transition surface at the phase
diagram in coordinates of the temperature and hydrostatic as well as uniaxial pressure.
This terminates at the line of critical point and the uniaxial pressure of magnitude lower
than critical, shifts the transition temperature, but preserves the transition. The critical
hydrostatic pressure that suppresses phase transition has linear temperature dependence.

The order parameter discontinuity and the volume effect diverge at the critical line as well
as the difference in isothermal compressibility and thermal expansion coefficient between the
parent and product phases.

31
 The coupling between the volume change and shear strain is shown to lead to the FCC
– BCC martensitic transformation for some special relations between the coefficients in the
free energy expansion. Though some fixed value of the Bain strain is needed to get the

low-temperature BCC lattice, the volume change as a secondary order parameter makes it
possible to have some temperature variation of the shear strain preserving the BCC lattice
and changing its period only. The non-linear expression for the elastic energy of thermal
expansion is shown to lead to proper relation between the shear strain and volume change

for the minima of elastic energy.

ACKNOWLEDGMENTS

The author is grateful to Prof. A.L. Roytburd for useful discussions. This work was
supported, in part, by a Soros Foundation Grant awarded by the American Physical Society.

The final part of the work has been done at Carleton University under the hospitality of
Prof. J. Goldak.

32
 REFERENCES

†
Present address: Department of Mechanical and Aerospace Engineering, Carleton Uni-
versity, Ottawa, Ont., K1S 5B6, Canada, e-mail: [email protected]

[1] L.E. Tanner, D. Schryvers and S.M. Shapiro, Mater. Sci. Eng. A 127, 205 (1990).

[2] S.M. Shapiro, in Competing Interactions and Microstructures: Statics and Dinamics,
edited by R. Lesar, A.R. Bishop and R. Heffner, Springer Proceedings in Physics, Vol.27
(Springer, Berlin, 1988).

[3] J.W. Christian, Theory of Transformations in Metals and Alloys, Pergamon, Oxford,
1965.

[4] B. Ya. Ljubov, Kineticheskaja Teorija Fazovykh Prevraschenij (in Russian), Moscow,
Metallurgizdat, 1969.

[5] K. Otsuka and K. Shimizu, Int. Metal Review, 31, n.3, p.93, (1986).

[6] A.L. Roitburd, in Solid State Physics: Advances in Research and Applications, edited
by H. Ehrenreich, F. Seitz and D. Turnbull, Vol.33, (Academic Press, New York, 1978),

p.317.

[7] L.D. Landau, E.M. Lifshitz, Statistical Physics, 3rd edition, Oxford, Pergamon, 1981.

[8] E.K.H. Salje, Phase transitions in ferroelastic and co-elastic crystals, Cambridge Uni-
versity Press, Cambridge, 1990.

[9] N. Rusovic and H. Warlimont, Phys. Status Solidi A 44, 609, (1977).

[10] S.M. Shapiro, Mater. Sci. Forum 56 - 58, 33, (1990).

[11] J.A. Krumhansl, R.J. Gooding, Phys. Rev. B 39, 3047, (1989).

[12] N. Boccara, Ann. Phys. (N.Y.) 40, 40, (1968).

[13] J. Liakos, G.A. Sounders, Phil. Mag. A 46, 217, (1982).

33
[14] F. Falk and P. Konopka, J. Phys.: Condens. Matter 2, 61, (1990).

[15] W. Cao, J.A. Krumhansl and R.J. Gooding, Phys. Rev. B 41, 11319 (1990).

[16] A.C.E. Reid and R.J. Gooding, Physica D 66, 180, (1993).

[17] J. Pouget, Phys. Rev. B 48, 864 (1993).

[18] R.A. Cowley, Phys. Rev. B 13, 4877 (1976).

[19] G.A. Korn, Mathematical handbook for scientists and engineers, 2d ed., 1968

[20] J.C. Toledano and P. Toledano, The Landau theory of phase transitions, World Scientific,
Singapore, 1987.

[21] M.A. Fradkin, Preprint MAF – 13.3/94 (1993).

[22] A.P. Levaniuk and A.S. Sigov, Defects and structural phase transitions, Gordon and

Breach, New York, 1988.

[23] M. Weger and I.B. Goldberg, in Solid State Physics: Advances in Research and Appli-

cations, edited by H. Ehrenreich, F. Seitz and D. Turnbull, Vol.28, (Academic Press,

New York, 1973), p.1.

[24] P.W. Anderson and E.I. Blount, Phys. Rev. Lett. 14, 217 (1965).

[25] T.R. Finlayson, Aust. J. Phys. 36, 553 (1983).

[26] L.E. Tanner, A.R. Pelton and R. Gronsky, J. Physique 43, C4-169, (1984).

[27] S. Muto, R. Oshima and F.E. Fujita, Acta Metall. Mater. 38, 685, (1990).

[28] A. Saxena and G.R. Barsch, Physica D 66, 195, (1993).

[29] M.P. Brassington and G.A. Sounders, Proc. Roy. Soc. A387, 289, (1983).

[30] G.A. Sounders, Physica Scripta T1, 49, (1982).

[31] L.D. Landau and E.M. Lifshitz, Theory of Elasticity, 3rd edition, Oxford, Pergamon,

34
 1981.

[32] J.R. Patel and B.W. Batterman, J. Appl. Phys. 37, 3447 (1966).

[33] S.M. Shapiro, E.C. Svensson, C. Vettier, B. Hennion, Phys. Rev. B 48, 13223 (1993).

[34] R.J. Gooding, Y.Y. Ye, C.T. Chan, K.M. Ho, B.N. Harmon, Phys. Rev. B 43, 13626
(1991).

[35] G.A. Sounders, J.D. Comins, J.E. Macdonald, E.A. Sounders, Phys. Rev. B 34, 2064
(1986).

[36] T.F. Smith and T.R. Finlayson, Thermochimica Acta 218, 153, (1993)

35
 FIGURES

~
0.1 ∆G

0.08 τ1 = 0.2 τ1
τ2 = − 0.1
0.06
τ3 = − 0.33
0.04
τ2
0.02

-0.4 -0.2 0.2 0.4

η
-0.02
σ = 0.04 τ3
τ0 ≈ − 0.221 -0.04
FIG. 1. The dependence of the Gibbs energy on the order parameter η under the applied field

for different temperatures τ1 > τ2 > τ0 > τ3 in the case of the second-order phase transition.

36
 η
0.8

0.6

0.4
σ1
σ3 σ2
0.2

-0.4 -0.2 0.2 0.4 τ

σ3
σ2 -0.2 σ1= 0.1
σ1 -0.4 σ2= 0.05
-0.6 σ3= 0.01
FIG. 2. The order parameter dependence on the temperature in various fields for the case of

the second-order phase transition. Dashed line corresponds to the absence of external field, σ = 0.

37
 < η2 >,
arb. un. 14
σ1 = 0.05
σ2 = 0.025
12

10
σ3 = 0.01
8 σ3
6 σ2
4

2 σ1
τ
-0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
FIG. 3. Mean square of the homogeneous order parameter fluctuations around Tc (τ = 0) under

different external fields.

38
 σ
0.1
Q>0 (σc,τc)

-0.6 -0.4 -0.2 0.2 τ

-0.1
Q<0
-0.2
Q>0
-0.3

-0.4

-0.5
FIG. 4. The region of the phase coexistence. The dashed line corresponds to points of the

first-order phase transition. It terminates in the critical point (τc = 1/3, σc = 1/27).

39