structural phase transitions and landau's theory
TRANSCRIPT
FOREWORD
Investigations of crystalline materials and phase transitions of them have been in
the focus of scientists for quite a long time. Those materials fascinate because of
their shape, transparency, color, hardness, and many other physical properties.
Especially in the vicinity of phase transitions the dependences of certain material
coefficients on temperature and/or pressure are often remarkable, technically
useable, and can also be well described by theoretical means.
Such materials - in particular single crystals - are in many cases fairly easy to
obtain and can be used to prove theoretical predictions.
The classical tool of choice is the theory named after the famous Russian
scientist Landau. This phenomenological approach is stringently based on group-
theoretical and symmetry-based considerations, and can be vastly applied to
model phase transitions.
I worked on the attached collection of papers since about year 2000, exclusively
during my leisure time. Some of the papers have been repeatedly refined or
slightly modified over the years, because after a while I always needed “a certain
distance” to the topics in order to be able to see the bigger picture again and not
to get lost in too many side aspects. Also some calculations were cumbersome.
That’s why I lost sometimes the passion to continue, but it luckily re-flourished
after a while.
One of my focus points was on ferroelastic / ferroelastoelectric transitions,
because they are just my favorites. On the other side I wanted to describe step-
by-step and very practically how to compile the energy expressions and how to
derive from them the temperature dependences of the material coefficients and
alike.
Hempel
December 2021
PS: Interested parties shall feel free to make use of the attached considerations,
but referencing to this source is highly appreciated.
Contact: [email protected]
Table of Contents
1. The Landau Theory of Symmetry Changes at Phase Transitions
(10 pages)
2. The Landau Theory of Phase Transitions - Case Discussions
(49 pages)
3. Calculation of Material Coefficients at Phase Transitions
(25 pages)
4. The Landau Theory of Ferroelastic Phase Transition of Rb4LiH3(SO4)4
(43 pages), with Annex for Improper Case (22 pages)
5. The Landau Theory of Ferroelastoelectric Phase Transition of
(NH4)2CuCl4 . 2H2O, (42 pages)
6. The Landau Theory of successive Phase Transitions of LiKSO4
(57 pages)
Hempel, 2017 Page 1 of 10
The Landau Theory of Symmetry Changes Phase Transitions
1. Introduction
It is assumed that a crystal undergoes a symmetry change at a critical point of temperature
TC and/or pressure pC.
The change of symmetry from a high temperature phase to a low temperature phase and
vice versa is called Phase Transition (PT).
Typically such PT can be described by the appearance of an Order Parameter (OP) at TC
when entering the low temperature phase.
In particular cases the OP can be spontaneous polarization, magnetization, elastic strain,…
Practically PTs are observed, where either the OP changes continuously in the vicinity of TC
(PT of 2nd kind) or discontinuously (PT of 1st kind).
2. Basic Assumptions
• The basic idea of Landau’s theory is to describe PTs of 2nd kind on the level of a
phenomenological theory
• Thereby the crystal exhibits the symmetry group Go in the high temperature phase and
transforms at TC to symmetry group G1 by losing certain symmetry elements (g) of Go.
• G1 is to be a subgroup of Go and the crystal transforms from one to another thermo-
Hempel, 2017 Page 2 of 10
The Landau Theory of Symmetry Changes Phase Transitions
• dynamic equilibrium state.
• Even at PTs of 2nd kind the symmetry of the crystal itself changes
discontinuously at TC
• Generally the symmetry elements (g) of a group G of a crystal leaves the
function ρ(x) (represents the probability density that an atom is located at
position x) invariant:
(R|t) ρ(x) = ρ ((R|t)-1 x) = ρ(x) with (R|t)=g G (1)
• Whereby: R represents the rotational part of space group element g
t represents the translation part of space group element g
• For the high temperature phase we can write:
(R|t) ρo(x) = ρo ((R|t)-1 x) = ρo(x) with (R|t)=g Go T≥TC (2)
• And for the low temperature phase:
(R|t) ρ1(x) = ρ1 ((R|t)-1 (x) = ρ1(x) with (R|t)=g G1 T<TC (3)
• Thus the density function in the low temperature phase can be written:
ρ1(x) = ρo(x)+δρ(x) (4)
Hempel, 2017 Page 3 of 10
The Landau Theory of Symmetry Changes Phase Transitions
• If we assume that δρ(x) is small, we can develop it according to terms of
basis functions of the irreducible representations (IRs) of group Go:
(5)
• The transform according to the respective IR D
• Next step is the introduction of a potential function compatible with symmetry
Go above, below and at the transition point.
• If we assume that is small - especially in the vicinity of the PT-point -
a Taylor expansion looks like:
F = F (T, p, ρo, ρ1) = F(o) + F(1) + F(2) + F(3) + F(4) + … (6)
• Depending on F has to be at minimum both, in the high and low
temperature phases
• F(o) stands for the identity representation of Go and depends on
F(o) cannot describe any PT.
( ) ( )xcx lki
k l i
lki
,,*
*
,,*
=
Basis
function i
of one
IR D
Multiplicity
of DAll IRs D
related to
one star *k
All stars *k
ooo c =
lki
,,*
( )x
( )x
Hempel, 2017 Page 4 of 10
The Landau Theory of Symmetry Changes Phase Transitions
• The other terms of eq. (6) have to be invariant under the symmetry operations
of Go too
• The transformations among the of within an IR D can be considered
to happen among the coefficients instead (i.e. acc. to the same rules) (7)
• Since must hold for T≥TC the are related to the OP (8)
• The 1st order term F(1) depends on (for fixed ,
and d being the dimension of D). The respective invariant must be ruled out
because:
a) such term cannot be related to an ID, but always to a
reducible representation (for simple explanation see e.g. /3/)
b) if a 1st order term is present, the potential F cannot assume
any minimum value, depending on the
• The term for fixed relates to the
2nd order invariant. There can be only one! (9)
• Together with eq. (9) a general expression of the invariant F(2) of eq. (6) is:
0),(,,* =pT c lki
( )( ) ( )=
=
d
i
lki
lki ccf
1
2,,*,,*2 lk ,,*
lkic ,,*
)(,,* x lki
lk
ic ,,*
( )( ) =
=
d
i
lki
lki ccf
1
,,*,,*1 lk ,,*
lkic ,,*
0
,,* lkic
F i.e.
Hempel, 2017 Page 5 of 10
The Landau Theory of Symmetry Changes Phase Transitions
(10)
• whereby the coefficients are not known. Generally their number
equals the number of the IRs.
• Typically only one ID and therefore one set is active / responsible for
a particular phase transition. Thus we put the respective
• Considering also higher order terms, finally F (ref. to eq. (6)) can be written for
the specific PT
(11)
• Looking now at eq. (11) it becomes obvious that all can only be zero (in
high temperature phase!) if the are all positive definite
• On the other side (below critical point), the can become unequal zero only
if the related is negative definite → has to change its sign
at the critical temperature TC (or pressure pC)
• Taking into account these requirements, the easiest function can be
(12)
Note: Generally the energy function F (eq. (11)) must be invariant under all symmetry operations (g) of the high
temperature phase Go. The Subscripts v and s ensure that the sums stretch over all possible invariants
),(,,* pT A lk
( )Co TTp AA −= )(
( )lki
lk
k llk
cf pT AF ,,*)2(,,*
*,,*
)2( ),(
=
lk
ic ,,*
ApT A lk ),(,,*
),(,,* pT A lk
),(,,* pT A lk
lkic ,,*
ilk
i cc ,,*
( ) ( ) ( ) ( ) ( ) ( ) ( ) ...,,,,,, )4()3(
1
2)( ++++== =
i
ss
si
vv
v
d
ii
o i cf pTCcf pTBcpTApTFcpTFF
A lk ,,*
Hempel, 2017 Page 6 of 10
The Landau Theory of Symmetry Changes Phase Transitions
• 3rd order terms*) in eq. (11) can be on principle allowed provided the
symmetrized cube [D]3 of ID D contains the identity representation of
group Go. The so called Landau Condition stipulates that for PTs of 2nd
order, 3rd order invariants must not be allowed in F
• This can be quickly checked by calculating the corresponding reduction
coefficient (see e.g. /02/):
(13)
• Existence of the 4th order terms*) can be formally checked through inspection
of the related reduction coefficient
• Important is to note: if the energy expression shall describe the PT under
equilibrium/minimum conditions, then F(4) (see eqs. (6) & (11)) has to be (14)
positive definite in total
*) Notes:
• The explicit form of the 3rd and 4th order invariants can be derived manually, which might be cumbersome sometimes
• Existence of a 3rd order invariant always means that the PT must be of 1st order. However, the Landau Theory and its
principles are usually applicable, as shown elsewhere (e.g. /03/).
• If a PT is of 1st kind, but no 3rd order invariant is allowed, then the energy expression has to be developed up the 6th
degree F(6), and 4th-term coefficient C has to be set negative, 6th order coefficient positive.
( ) ( ) possibleinvariant order 3rd
possibleinvariant order 3rd no gD
GoGgo 0
01 3
=
Hempel, 2017 Page 7 of 10
The Landau Theory of Symmetry Changes Phase Transitions
• Another important precondition of the Landau Theory is the requirement for
spatial homogeneity of the crystal. This is called Lifshitz Condition.
• If the related reduction coefficient calculates to be zero then spatial
homogeneity is expected.
(15)
• Also then any invariant containing terms like cannot
appear in eq. (11) (16)
• Generally, Landau stated that the real IDs (or physically IDs, being the sum
of a complex ID and its conjugated complex ID: ) which fulfil the
Landau- and Lifshitz-Conditions are named active or acceptable represen-
tations are able to trigger continuous phase transitions (2nd kind)**) (17)
*) Notes:
• If such gradient term(s) should be allowed by symmetry, the Landau approach can be applied as well. Then the
energy expression depends also on spatial changes of the OP, measurement dimensions, and not just on the unit cell.
• The OP(s) can be modulated and described by wave lengths which are not commensurable with the dimensions of the
crystal lattice (unit cell). Such phases are called incommensurate phases.
**) Note:
• Since ρo and δρ are real values (ref. to eqs. (1)&(4)), the basis functions i and the coefficients ci are also real.
( ) ( ) ( ) 0ggDG
vP
Ggo o
=
)(21
x
cc-
x
cc
P
ij
P
ji
)
DD
Hempel, 2017 Page 8 of 10
The Landau Theory of Symmetry Changes Phase Transitions
3. Special Case: Proper Phase Transitions
• PTs, described by the star (-point of the Brillouin zone of the high temperature
phase) are called ferrodistortive because the number of atoms of the primitive unit
cell doesn’t change at the transition.
• Also the translational symmetry doesn’t change.
• Proper PTs mean that specific physical properties like electric polarization,
magnetization, elastic strain can play the role of the order parameter itself.
• In case of polarization the related PT is to be ferroelectric and in case of strain
ferroelastic, respectively.
• The representations of the physical properties (e.g. polarization: polar vector ,
strain: polar tensor of rank 2: ) exhibit specific values for the characters
for individual symmetry elements g (see e.g. /2/).
• To check whether a certain active ID D can lead to a ferroelectric, ferroelastic or
other low temperature phase, the related reduction coefficients need to be looked at
(whereby represent the character of the ID of the concerned symmetry element
g):
(18)
0k =
)( v PD
2)( v PD
( ) ( )
=oGg
v P
o
E ggG
m )(1
( )( )gv P
)(
( )g
Hempel, 2017 Page 9 of 10
The Landau Theory of Symmetry Changes Phase Transitions
• The next topic is how to define in detail the components of polarization, strain, etc. that
they can be utilized to describe the PT under consideration
• As shown elsewhere (e.g. /2/) certain projection operators are capable to generate the
searched basis functions of an ID
(19)
• Application onto an arbitrary function f(x) (e.g. polarization, strain, …) yield to
a decomposition of the function according to its irreducible parts:
(20)
( ) ( )
=oGg
ij
o
ij gTgDG
d
Conjugated complex
element ij of D
Dimension of D
Operator related
to symmetry
element g
( ) ( )xfxf ijij =
Hempel, 2017 Page 10 of 10
The Landau Theory of Symmetry Changes Phase Transitions
4. Literature
/01/ Landau, L.D, Lifshitz, E.M., Lehrbuch der Theoretischen Physik, Statistische Physik,
Akademie-Verlag Berlin, 1984
/02/ Kocinsky, J., “Theory of Symmetry Changes at Continuous Phase Transitions”,
Elsevier, Amsterdam-Oxford-New York, 1983
/03/ Toledano P., Toledano J.-C., “The Landau Theory of Phase Transitions“, World
Scientific Publishing, 1987
/04/ Izyumov Yu. A., Syromyatnikov V. N., “Phase Transitions and Crystal Symmetry“,
Kluwer Academic Publishers, Dordrecht / Boston / London, 1990
/05/ Gufan, Yu. M., “Thermodynamic Theory of Phase Transitions”, Publisher: University of
Rostov on Don, 1982
Hempel, 2015 - 2016 Page 1 of 49
Landau Theory of Phase Transitions
Landau Theory – Case Discussions & Phase Diagrams
Table of Contents
1. Case: Q2 – Q4
2. Case: Q2 – Q3 – Q4
3. Case: Q2 – Q4 – Q6
4. Case: 2-component Order Parameter w/o cubic term
5. Case: 2-component Order Parameter with cubic term
6. Case: 2 coupled 1-component Order Parameters
7. Literature
Hempel, 2015 - 2016 Page 2 of 49
Landau Theory of Phase Transitions
1. Case: Q2 – Q4
This case deals with the Free Energy Expansion (FEE) in dependence of the Order
Parameter (OP) Q.
F = AQ2 + BQ4 *) (1)
with A = Ao(T-To), To- Transition Temperature, Ao and B being positive constants.**)
(2)
The equation of state requires: (3)
This yields two solutions for Q
Solution I: Q = 0 (describes Paraphase) (4)
Solution II: Q2 = -A/(2B) (describes Ferrophase) (5)
The stability condition of Solutions I & II requires:
(6)
For Solution I (ref. to eq. 4) follows: (7)
For Solution II (ref. to eq. 5) follows: (8)
*) Note: In FEE a linear term of Q is not allowed. Otherwise the equation of state (eq. 3) can never be fulfilled.
**) Note: B must be positive to guarantee that FEE stays positive for large values of OP Q.
042 3 =+=
BQAQ
Q
F
02
2
+
212BQ2A 0Q
F
oTT for stable 0A
oTT for stable 0A
Hempel, 2015 - 2016 Page 3 of 49
Landau Theory of Phase Transitions
Next task is to investigate the validity of Solutions I & II.
Solution I: Q = 0 → no limitations (9)
Solution II: Q2 = -A/(2B) → since B>0 must be A<0 (10)
To construct the Phase Diagram, the Phase Transition Line (PTL) has to be calculated
acc. to condition that the FEE of Para- and Ferrophase has to be equal.
F (Q = 0) = F (Q2 = -A/(2B)) (11)
0 = A (12)
The Phase Transition Line is characterized by A = 0, the two phases don’t overlap but
coincide and acc. to eq. (5) the OP has no jump at PTL, which mean that the investigated
Phase Transition (PT) is of 2nd kind. The Phase Diagram looks like:
PTL: A = 0
A
B
Q2 = -A/(2B)
Q = 0
Hempel, 2015 - 2016 Page 4 of 49
Landau Theory of Phase Transitions
2. Case: Q2 – Q3 – Q4
This case deals with the Free Energy Expansion (FEE) in dependence of the
Order Parameter (OP) Q.
F = AQ2 + CQ3 + BQ4 *) (1)
with A = Ao(T-To), To- Transition Temperature, Ao and B being positive constants.
The constant C can be positive or negative. (2)
The equation of state requires: (3)
*) (4)
This yields two principal solutions for Q:
Solution I: (describes Paraphase) (5)
Solution II: (6)
(describes Ferrophase)
*) Note: In FEE and in the equation of state (term in brackets of eq. (4)) it becomes visible, that the term (7) CQ3 respectively CQ is invariant if simultaneously the signs of Q and C change.
(+C)(+Q) = (-C)(-Q) or respectively (-C)(+Q) = (+C)(-Q).
0432 32 =++=
BQCQAQ
Q
F
( ) 0432 2 =++ BQCQAQ
0=Q
22
2
9
321
8
3
8
3
264
9
8
3
C
AB
B
C
B
C
B
A
B
C
B
CQQ −−=−−==
Hempel, 2015 - 2016 Page 5 of 49
Landau Theory of Phase Transitions
The stability condition of Solutions I & II requires:
(8)
For Solution I (ref. to eqs. 3,5) follows: (9)
For the Solutions II follows with eqs. (8) and (4):
(10)
which together results in (11)
Eq. (11) can be re-written as: (12)
Stability is guaranteed in 2 cases:
1st case: and (13)
and
2nd case: and (14)
062
2
++
212BQCQ2A 0Q
F
oTT for stable 0A
06 ++ 212BQCQ2A
0432 2 =++ BQCQA
083 2 + BQCQ
( ) 083 + BQCQ
0Q 083 + BQC
0Q 083 + BQC
Hempel, 2015 - 2016 Page 6 of 49
Landau Theory of Phase Transitions
Summary of stability constraints after discussion of solution II (see eq. (6)):
(15)
Summary:
a) Q+ is stable and positive if (16)
b) Q- is stable and negative if (17)
Another feature of the solutions Q+ and Q- as it can be easily shown using
eq. (6):
(18)
Q+
Q-
B
CA
32
9 2
possiblenot
0A
possiblenot
possiblenot
( ) ( )00
)0()0(
=−
=−
−+
−+
C QC Q
and
C QC Q
B
CA
32
9 2
B
CA
32
9 2
B
CA
32
9 2
B
CA
32
9 2
B
CA
32
9 2
00
00
−
+
Q C if
and
Q C if
Hempel, 2015 - 2016 Page 7 of 49
Landau Theory of Phase Transitions
Next task is to investigate the validity of Solutions.
Solution I: Q = 0 (19)
Solution II: and (20)
Both solutions are real and valid if the root radicand stays positive.
This requires:
(21)
Simultaneous consideration of eqs. (16), (17), (18), (21) leads finally to:
→ For : (22)
→ For : (23)
To construct the Phase Diagram, the Phase Transition Line (PTL) has to be
calculated acc. to condition that the FEEs of Para- and Ferrophase have
to be equal.
F (Q = 0) = F (Q ≠ 0) (24)
B
CA
32
9 2
29
321
8
3
8
3
C
AB
B
C
B
CQ −+−=+ 29
321
8
3
8
3
C
AB
B
C
B
CQ −−−=−
+Q
−QB
CA C
32
9,0
2
B
CA C
32
9,0
2
Hempel, 2015 - 2016 Page 8 of 49
Landau Theory of Phase Transitions
From eqs. (1) and (4) we obtain:
(25)
The solution Q2=0 can only occur provided A≡0 in Q+ (ref. to eq. (6)). This is not
possible because Q+ is stable already for bigger values of A:
The solution (A+CQ+BQ2)=0 leads together with eqs. (4), (23) to A = C2/(4B) (26)
which is the equation of PTL.
Since the stability limits of Para- and Ferrophase do overlap (ref. to eqs. (9),
(22), (23)) the Phase Transition is of 1st kind.
Also from eqs. (6), (26) we calculate for the jump of OP at the transition line:
or with eq. (25) (27)
The actual phase transition temperature calculates from eq. (2) together
with eq. (26) as:
(28)
( )220 BQCQAQ ++=
4320 BQCQAQ ++=
C
AQPTL
2−=
B
CQPTL
2−=
o
oCPTLBA
CTTT
4
2
+==
B
CA
32
9 2
Hempel, 2015 - 2016 Page 9 of 49
Landau Theory of Phase Transitions
A
B
PTL: A = C2/(4B)
Stability limit
Paraphase A = 0
Stability limit
Ferrophase A = 9C2/(32B)Q = 0
Q ≠ 0
Since the stability regions of both phases do overlap, the OP exhibits a jump at the PTL and
the investigated Phase Transition (PT) is of 1st kind.
This is always the case if a 3rd order OP-Term in FEE appears.
A
C
PTL: A = C2/(4B)
Stability limit
Paraphase A = 0
Q = 0
Q-(C>0) ≠ 0Q+(C<0) ≠ 0
Stability limit
Ferrophase A = 9C2/(32B)
Hempel, 2015 - 2016 Page 10 of 49
Landau Theory of Phase Transitions
3. Case: Q2 – Q4 – Q6
This case deals with the Free Energy Expansion (FEE) in dependence of the Order
Parameter (OP) Q.
F = AQ2 + BQ4 + DQ6 (1)
with A = Ao(T-To), To- Transition Temperature, Ao and D being positive constants.
The constant B can be positive or negative. (2)
The equation of state requires: (3)
(4)
This yields two solutions for Q:
Solution I: (describes Paraphase) (5)
Solution II:
(describes Ferrophase) (6)
0642 53 =++=
DQBQAQ
Q
F
( ) 0642 42 =++ DQBQAQ
0=Q
D
A
D
B
D
BQQ
393 2
222 −−==
2
22 31
33 B
AD
D
B
D
BQQ −−==
Hempel, 2015 - 2016 Page 11 of 49
Landau Theory of Phase Transitions
The stability condition of Solutions I & II requires:
(7)
For Solution I (ref. to eq. 5) follows: (8)
For the Solutions II eqs. (7) and (4) are valid:
(9)
(10)
Subtraction: eq. (9) – eq. (10) (11)
Therefore stability is guaranteed if is valid. (12)
Since the expression has to be positive definite too.
For the Paraphase (Q=0) follows from eq. (9) that A≥0. (13)
With the solution of eq. (6) certain cases can be discussed and are summarized
for the Ferrophase in the following table:
030 4
2
2
++
DQ12BQ2A 0
Q
F 2
CTT for stable 0A
( ) 022 + QQ3DB
0642 42 =++ DQBQA
030 4 ++ DQ12BQ2A 2
( ) 03 22 + QDQB
( )23DQB +02 Q
Hempel, 2015 - 2016 Page 12 of 49
Landau Theory of Phase Transitions
(14)
Summary:
is stable if (15)
is stable if (16)
Next task is to investigate the validity of Solutions.
Solution I: Q = 0 → no limitations (17)
Solution II: and (ref. to eq. (6)) are valid if the root radicand remains
positive. This calculates:
02 Q 03 2 + DQB
02 +Q
02 −Q
0,0
0,0
AB
AB
ok always B
possiblenot B
,0
,0
possiblenot B
ok always ,B
,0
0
ok always B
possiblenot ,B
,0
0
2+Q 0,0 AB
0B2−Q
2+Q 2
−Q
Hempel, 2015 - 2016 Page 13 of 49
Landau Theory of Phase Transitions
(18)
Simultaneous consideration of eqs. (15), (16), (18) leads finally to:
→ For : B>0 and A≤0 (19)
→ For : B<0 and (20)
To construct the Phase Diagram, the Phase Transition Line (PTL) has to be
calculated acc. to condition that the FEE of Para- and Ferrophase has to be equal.
F (Q = 0) = F (Q ≠ 0) (21)
(22)
Solution Q2 = 0: this requires A = 0 and materializes only for (23)
(ref. to eqs. (6), (19))
→ describes a Phase Transition of 2nd kind.
D
BA
B
AD
30
31
2
2−
2
−Q
2
+Q
D
BA
3
2
( )22 20 BQAQ +=
2
+Q
Hempel, 2015 - 2016 Page 14 of 49
Landau Theory of Phase Transitions
Solution (2A+BQ2) = 0: this requires A = B2/(4D) (equation of PTL) and
can be only materialized for (ref. to limitation
for A as stipulated in eqs. (18, 20)).
→ describes a Phase Transition of 1st kind. (24)
Inserting eq. (24) into Q- (ref. to eq. (6)) yield for the jump of the OP at the PTL:
(25)
With eq. (24) this can be re-written to be:
(26)
The actual phase transition temperature calculates from eq. (2) together
with eq. (24) as:
(27)
2
−Q
D
BQ PTL
22 −=
B
AQ PTL
22 −=
o
oCPTLDA
BTTT
4
2
+==
Hempel, 2015 - 2016 Page 15 of 49
Landau Theory of Phase Transitions
The Phase Diagram for solution looks like:
It represents a Phase Transition of 2nd kind.
The Phase Diagram for solution looks like:
It represents a Phase Transition of 1st kind.
2
+Q
PTL: A = 0
A
B
Q+2 ≠ 0
Q = 0
2
−QA
PTL: A = B2/(4D) B
Q-2 ≠ 0
Stability limit
Ferrophase A = B2/(3D)
Stability limit
Paraphase A = 0
Q = 0
Hempel, 2015 - 2016 Page 16 of 49
Landau Theory of Phase Transitions
4. Case: 2-component Order Parameter w/o cubic term
This case deals with the Free Energy Expansion (FEE) in dependence of two Order
Parameter (OP) Components Q1, Q2.
F = A(Q12 +Q2
2 ) + B1(Q14 +Q2
4 ) + B2Q12Q2
2 (1)
with A = Ao(T-To), To- Transition Temperature, Ao being positive.
The equations of state write:
(2)
(3)
This yields to totally 4 possible phases, depending on combinations of (Q1, Q2) :
Phase 0: (0, 0) → describes Paraphase (4)
Phase 1: ( , 0) → describes Phase 1 *) (5)
Phase 2: ( , ) → describes Phase 2 (6)
Phase 3: (Q1,Q2) → describes Phase 3 (7)
*) Note: Another possible Phase (0, Q‘) is identical to Phase 1. This originates from the symmetry of FEE
regarding Q1, Q2.
( ) 022 2
12
2
212
2
=++=
QBQBAQ
Q
F
( ) 022 2
22
2
111
1
=++=
QBQBAQ
Q
F
Q
Q Q
Hempel, 2015 - 2016 Page 17 of 49
Landau Theory of Phase Transitions
Phase 1: ( , 0)
From eqs. (2) & (3) follows
(8)
Phase 2: ( , )
From eqs. (2) & (3) follows
(9)
Phase 3: (Q1,Q2)
From the difference of eqs. (2) & (3) follows
(10)
Eq. (10) can only be fulfilled if , which results either in
and describing Phase 2 (as above) or in describing
Phase 3. Other values are not permitted.
The calculation of the stability condition requires exploitation of 2nd derivatives
of FEE:
02 21 =+ QBA
1
2
2B
AQ −=
0)2( 2
21 =++ QBBA
21
2
2 BB
AQ
+−=
22
2
1 QQ =
Q
Q Q
( ) ( ) 02 2
2
2
121 =−− QQBB
''21 QQQ ==
121 , QQ QQ 2 −=−=
Hempel, 2015 - 2016 Page 18 of 49
Landau Theory of Phase Transitions
(11)
(12)
(13)
For OPs with 2 components the stability requires:
and (14)
Phase 0: (0, 0) Inserting Q1=Q2=0 into eqs. (11)-(14) leads to
the condition A>0 for phase 0 (15)
Phase 1: ( , 0) Inserting from eq. (8) into eq. (11) leads to
A<0 (16)
( )
( )
212
21
2
2
12
2
212
2
2
2
22
2
112
1
2
4
62
62
QQBQQ
F
QBQBAQ
F
QBQBAQ
F
=
++=
++=
021
2
Q
F
0
22
2
12
221
2
21
2
Q
F
FQQ
F
Q
F
Q Q
Hempel, 2015 - 2016 Page 19 of 49
Landau Theory of Phase Transitions
Inserting from eq. (8) into eq. (14) leads to
(17)
Phase 2: ( , ) Inserting from eq. (9) into eq. (11) leads to
& also AB1≤0 (18)
Phase 3: Inserting from eq. (9) into eq. (14) leads to
(19)
Next task is to investigate the validity of Solutions.
Phase 0: (0, 0) no restrictions (20)
Phase 1: ( , 0) From eq. (8) follows that either
A<0 and B1>0 or A>0 and B1<0 must hold.
But eq. (16) requires A<0 which results together
with B1>0 and eq. (17) that (21)
Phase 2: From eq. (9) follows
& also A>0 and 2B1+B2<0 or (22)
Phase 3: A<0 and 2B1+B2>0 (23)
( )0
2
2
1
21 −
B
BB
( ) ( ) 022 2121 +− BBBB
0)2( 21 −BB
Q Q
Q
Q
( )QQ ,
Q
Q
−=
−=
12
21
−=
−=
12
21
Hempel, 2015 - 2016 Page 20 of 49
Landau Theory of Phase Transitions
Eqs. (22, 23) together with eq. (19) yield
A>0 and 2B1+B2<0 and 2B1-B2<0 or (24)
A<0 and 2B1+B2>0 and 2B1-B2>0 (25)
This can be re-written to be
A>0 and 2B1<-B2 and 2B1<B2 or (26)
A<0 and 2B1>-B2 and 2B1>B2 (27)
Eqs. (26, 27) can be rewritten as:
A>0, 2B1<B2<-2B1 → not possible since B1>0 or (28)
A<0, -2B1<B2<2B1 → possible since B1>0 (29)
Summary:
Phase 0: (0, 0) A>0, no further restrictions (30)
Phase 1: ( , 0) A<0, B1>0, and 2B1<B2 (31)
Phase 2: ( , ) A<0, B1>0, and -2B1<B2<2B1 (32)
Phase 3: A<0, B1>0, and -2B1<B2<2B1 (33)
Q
Q Q
−=
−=
12
21
Hempel, 2015 - 2016 Page 21 of 49
Landau Theory of Phase Transitions
To construct the Phase Diagram, the Phase Transition Line (PTL) has to be
calculated acc. to condition that the FEE of Para- and Ferrophase has to be equal.
F (0, 0) = F ( , 0) and
F (0, 0) = F ( , ) (34)
Both expressions of eq. (34) are fulfilled if A=0, where the stability limits of the
adjacent phases coincide. Therefore those Phase Transitions are of 2nd Kind.
The transition between Ferrophases 1 and 2 cannot be described within this
model since there is no solution of the form A = A(B1, B2). To do this, higher
order terms of (Q1, Q2) would have to be considered in FEE (ref. to eq. (1)).
PTL: A = 0A
B2
+2B1-2B1
Phase 0
Phase 2 / 3 Phase 1
Q Q
Q
Hempel, 2015 - 2016 Page 22 of 49
Landau Theory of Phase Transitions
5. Case: 2-component Order Parameter with cubic term
This case deals with the Free Energy Expansion (FEE) in dependence of two Order
Parameter (OP) Components Q1, Q2.
F = A(Q12 +Q2
2 ) + C(Q13 - 3Q2
2Q1) + B(Q14 +Q2
4) (1)
with A = Ao(T-To), To- Transition Temperature, Ao being positive, C can be positive
or negative, and B must be positive.
The equations of state write:
(2)
(3)
This yields to totally 6 possible phases, depending on combinations of (Q1, Q2) :
Phase 0: (0, 0) → describes Paraphase (4)
Phase 1: ( , 0) → describes Ferrophase 1 (5)
Phase 2: (0, ) → describes Ferrophase 2 (6)
Phase 3: ( , ) → describes Ferrophase 3 (7)
Phase 4: ( , ) → describes Ferrophase 4 (8)
Phase 5: (Q1, Q2) → describes Ferrophase 5 (9)
( ) ( ) 0432 2
2
2
11
2
2
2
11
1
=++−+=
QQBQQQCAQ
Q
F
( ) 0462 2
2
2
12212
2
=++−=
QQBQQCQAQ
Q
F
Q
Q
Q Q
Q Q −
Hempel, 2015 - 2016 Page 23 of 49
Landau Theory of Phase Transitions
Phase 1: ( , 0)
From eqs. (2) & (3) follows
(10)
Phase 2: (0, )
From eqs. (2) & (3) follows
solvable only if C = 0 (3rd order term ruled out!)
and yields (11)
Phase 3: ( , )
From eqs. (2) & (3) follows
solvable only if C = 0 (3rd order term ruled out!)
and yields (12)
Phase 4: ( , ) Same situation as for phase 3. (13)
0432 2 =++ QBQCA
B
A
B
C
B
CQQ
264
9
8
32
2
−−==
03 2 =− QC
B
AQ
22 −=
042 3 =+ QBQA
04 2 =+ QBA
043 2 =+− QBQCA
B
AQ
42 −=
Q
Q
Q
Q Q −
Q
Hempel, 2015 - 2016 Page 24 of 49
Landau Theory of Phase Transitions
Phase 5: (Q1, Q2)
From eqs. (2) & (3) follows → (14)
Inserting eq. (14) in eq. (2) yields:
(15)
This expression looks similar like the state equation
for phase 1 (ref. to eq. (10)). Indeed eq. (15) can be
transformed exactly into the state equation by putting
→ (16)
Result: the solutions (Q’, 0), and from eq. (14)
yield the same results, except some changes in the
coefficients.
The related Ferrophases differ only in their arrangements
with regard to the Paraphase.
These sets of combinations of the OP-components are
summarized as type which describes phase 1. (17)
Therefore in the following chapters only phase 1 needs
to be investigated.
22
213 QQ =03 2
221 =−QQ
083 211 =+− BQCQA
QQ −=2
11 0432 2 =++ QBQCA
),3( QQ
0,Q
Hempel, 2015 - 2016 Page 25 of 49
Landau Theory of Phase Transitions
The calculation of the stability condition requires exploitation of 2nd derivatives
of FEE:
(18)
Phase 0: (0, 0) A>0 (19)
Phase 1: ( , 0) Mind that only phase 1 needs to be investigated
Inserting (Q1=Q‘, Q2=0) in eq. (18) and making use of
the state equation eqs. (10, 16) yield:
(20)
212
21
2
2
2
2
112
2
2
2
2
2
112
1
2
86
)3(462
)3(462
QBQCQQQ
F
QQBCQAQ
F
QQBCQAQ
F
+−=
++−=
+++=
0
9
34
21
2
22
2
21
2
=
−=
−−=
F
QCQ
F
QCAQ
F
Q
Hempel, 2015 - 2016 Page 26 of 49
Landau Theory of Phase Transitions
Stability requires:
and (21)
and results in: and (22a,b)
Eq. (22a) reformed to and multiplied by itself yields:
(23)
If the equation of state (eq. (16)) is rearranged to be
and if this expression is introduced into eq. (23) the 1st stability condition
comes out to be:
(24)
If eq. (22a) holds then eq. (22b) requires: → (25)
Eq. (25) therefore demands that Q’<0 if C>0 and Q’>0 if C<0. (26)
Furthermore, following relations must hold:
021
2
Q
F0
22
2
12
221
2
21
2
Q
F
FQQ
F
Q
F
034 −− QCA ( ) ( ) 0934 −−− QCQCA
09 − QC 0QC
B
CA
32
9 2
034 + QCA
024916 222 ++ QACQCA
22 322416 QABQACA −=+
Hempel, 2015 - 2016 Page 27 of 49
Landau Theory of Phase Transitions
(27)
which mean that changes the sign if C changes its sign. This can be
easily proven, using eq. (10).
Next task is to investigate the validity of Solutions.
Phase 0: (0, 0) no restrictions (28)
Phase 1: ( , 0) From eq. (10) follows that either the root radicand needs
to be positive. The requires:
(29)
and coincides with one stability limit (ref. to eq. (24)).
To construct the Phase Diagram, the Phase Transition Line (PTL) has to be
calculated acc. to condition that the FEE of Para- and Ferrophase has to be equal.
F ( , 0) = F (0, 0) (30)
B
CA
32
9 2
Q
Q
Q
00
00
'
'
−
+
Q C if
and
Q C if ( )
)0()0(
0)0(
''
''
=−
=−
−+
−+
C QC Q
and
C QC Q
Hempel, 2015 - 2016 Page 28 of 49
Landau Theory of Phase Transitions
(31)
This can be reformed to be
Inserted into the state equation (eq. (16)) yields
(32)
Now the jump of the OP at the PTL calculates:
(33)
Inserting eq. (33) into eq. (31) yields the expression of the PTL:
(34)
If this is reinserted into eq. (32) another expression for the OP-jump
is obtained:
(35)
The actual phase transition temperature calculates from eq. (2) together
with eq. (34) as: (36)
0432 =++ QBQCQA
QCQAQB −−= 22
04432 =−−+ QCAQCA
C
AQPTL
2−=
B
CA
4
2
=
B
CQPTL
2−=
o
oCPTLBA
CTTT
4
2
+==
Hempel, 2015 - 2016 Page 29 of 49
Landau Theory of Phase Transitions
Since the stability regions of the Para- and Ferrophases do overlap, the OP exhibits a jump
at the PTL and the investigated Phase Transitions (PTs) are of 1st kind. This is always
the case if a 3rd order OP-Term in FEE appears.
A phase transition between the two Ferrophases [ ] cannot be described with this
model and would require the addition of higher OP-terms in FEE.
Stability limit
Ferrophase A = 9C2/(32B)
A
C
PTL: A = C2/(4B)
Q‘ = 0
Stability limit
Paraphase A = 0 Q‘-(C>0) ≠ 0Q‘+(C<0) ≠ 0
Hempel, 2015 - 2016 Page 30 of 49
Landau Theory of Phase Transitions
6. Case: 2 coupled 1-component Order Parameters
This case deals with the Free Energy Expansion (FEE) in dependence of two coupled
Order Parameters (OPs) – Q and P.
F = AQ2 + BQ4 + CQ6 + DP2 + EP4 + FQ2P2 (1)
with A = Ao(T-ToQ), ToQ- Transition Temperature regarding Q, Ao, C, E being positive, and
B, F can be positive or negative, D=Do(T-ToP), ToP- Transition Temperature regarding P.
The equations of state write:
(2)
(3)
This yields to totally 3 possible phases, depending on combinations of (Q, P) :
Phase 0: (0, 0) → describes Paraphase (4)
Phase 1: ( , 0) → describes Ferrophase 1 (5)
Phase 2: (0, P) → describes Ferrophase 2 (6)
Phase 3: ( , P) → describes Ferrophase 3 (7)
02642 253 =+++=
FQPCQBQAQ
Q
F
0242 23 =++=
PFQEPDP
P
F
Q
Q
Hempel, 2015 - 2016 Page 31 of 49
Landau Theory of Phase Transitions
Phase 1: ( , 0)
From eqs. (2) & (3) follows
(8)
Phase 2: (0, P)
From eqs. (2) & (3) follows
(9)
Phase 3: ( , P)
From eqs. (2) & (3) follows
(10)
with (11,12)
032 42 =++ QCQBA
2
22 31
33 B
AC
C
B
C
BQQ −−==
02 2 =+ EPD
E
DP
22 −=
032 242 =+++ FPQCQBA
02 22 =++ QFEPD
2
222 2448
11212
−−
−==
FDECCAE
ECECQQ
22
22Q
E
F
E
DP −−= 24 FBE −=
Q
Q
Hempel, 2015 - 2016 Page 32 of 49
Landau Theory of Phase Transitions
The calculation of the stability condition requires exploitation of 2nd derivatives
of FEE:
(13)
(14)
(15)
In particular *):
and (16a,b)
Phase 0: (0, 0) From eqs. (13-15) follows A>0and D>0 for phase 0 (17a,b)
Phase 1: ( , 0) From eq. (13, 16a) together with eq. (2) follows
(ref. also to case 3 of this paper) and (18)
yield that stability is given for if A<0 and for
if (19)
*) Note: As shown in the literature (ref. to Gufan et al.) the inequations are usually evaluated, but in
several cases it is appropriate to investigate the expressions as equations, to find the stability limits.
FQPPQ
FF
FQEPDP
FF
FPCQBQAQ
FF
QP
PP
4
2122
230122
2
22
2
2
242
2
2
=
=
++=
=
+++=
=
02
2
=
QQF
Q
F0
2
22
2
2
2
P
F
QP
FPQ
F
Q
F
02 + QBA2
−Q2
+Q
C
BA
3
2
Q
Hempel, 2015 - 2016 Page 33 of 49
Landau Theory of Phase Transitions
Eq. (16b) as calculates
(20)
Since the 1st factor is negative (ref. to eq. (18))
the expression also has to be negative or (21)
zero, too.
Using eq. (8) this yields the requirement:
(22)
From eqs. (18, 21) follows that:
→ if A>0 then B<0 and therefore ∆<0 (ref. to eq. (12)) (23)
→ if B>0 then A<0 (24)
→ if D>0 then F<0 (25)
→ if F>0 then D<0 (26)
Phase 2: (0, P) From eq. (13, 16a) together with eq. (3) follows
and requires (27)
Eq. (16b) as calculates
(28)
02 − QPPPQQ FFF
( ) ( ) 0022 22 −++ QFDQBA
)22( 2QFD +
DF
BD
F
CA
23 2
2+−=
022 2 + FPA DE
FA
2
02 − QPPPQQ FFF
( ) ( ) 0012222 22 −++ EPDFPA
Hempel, 2015 - 2016 Page 34 of 49
Landau Theory of Phase Transitions
Since the 1st factor is positive (ref. to eq. (27))
the expression also needs to be positive (29)
or zero. Using eq. (9) this yields the requirement:
(30)
From eqs. (27, 29) follows that:
→ if A<0 then F>0 (31)
→ if F<0 then A>0 (32)
→ if D<0 then E>0 (E is always positive ref. to eq. (1)) (33)
Phase 3: ( , P) First step is to investigate FQQ (ref. to eq. (16a)).
From eq. (13), but easier from eq. (2) via application
of the chain rule the 2nd derivative FQQ is obtained
(34)
The 1st factor of eq. (34) requires to be:
(ref. to validity of solution below – eqs. (48,49)) (35)
The 2nd factor of eq. (34) together with eq. (10) yields for:
with ∆<0 being always stable if root radicand
of eq. (10) is positive (36a)
with ∆>0 being stable if following relation holds
(36b)
0D
( )2122 EPD +
( ) 03 22 + QCBQ
02 Q
2
−Q
2
−Q
EC
B
C
BD
E
FA
632
2 +−
Q
Hempel, 2015 - 2016 Page 35 of 49
Landau Theory of Phase Transitions
with ∆>0 being always stable if root radicand
of eq. (10) is positive (37a)
with ∆<0 being stable if following relation holds
(37b)
Moreover from the 2nd factor of eq. (36) follows:
→ if B<0 then C>0 (but C is always positive ref. to eq. (1)) (38)
Next step is the investigation of .
This leads to the expression
(39)
The 1st factor of eq. (39) to be positive or zero leads to the
stability equation (which equals eq. (22)).
(40)
whereby the 2nd factor of eq. (39) to be also positive or
zero yields:
(41)
( ) 01216 222 + QCEPQ
2
2
32D
F
CD
F
BA −=
02 − QPPPQQ FFF
2
2
482 CED
E
FA
+=
2
+Q
2
+Q
EC
B
C
BD
E
FA
632
2 +−
Hempel, 2015 - 2016 Page 36 of 49
Landau Theory of Phase Transitions
Next task is to investigate the validity of Solutions.
Phase 0: (0, 0) no restrictions (42)
Phase 1: ( , 0) From eq. (8) follows that the root radicand needs
to be positive for . The requires: (43)
Moreover the expression for has to be positive definite.
→ For : either B>0 and A<0 or B<0 and A>0 (44)
→ For : B<0 is required (45)
Phase 2: (0, P) From eq. (9) follows that D<0 (46)
Phase 3: ( , P) From eq. (10) follows that the root radicand of has
to be positive:
(47)
Moreover the expression for has to be positive definite.
→ For : either ∆>0 and or ∆<0 and (48)
→ For : ∆<0 is required (49)
C
BA
3
2
2
2
482 CED
E
FA
+
2
Q
2
+Q
2
−Q
2
Q
DE
FA
2 D
E
FA
2
2
Q
2
+Q
2
−Q
2
Q
Hempel, 2015 - 2016 Page 37 of 49
Landau Theory of Phase Transitions
Additionally – to ensure that P2 is also positive definite -
from eq. (11) follows that:
→ has to be positive definite (ref. to eqs. (48,49)) (50)
→ from eq. (11) follows D+F <0, which yields: (51)
→ D<0 and F<0 or (52)
→ D<0 and F>0 or (53)
→ D>0 and F<0 (54)
To construct the Phase Diagram, the Phase Transition Lines (PTLs) have to be
calculated acc. to condition that the FEEs of the adjacent phases have to be equal.
PTL 0-1: - Phase 0 exists if A≥0 (ref. to eqs. (17a,b,42) (55)
- if B>0 than Phase 1 is stable for A<0 (ref. to eq. (24)
and the OP is (ref. to eq. (44)). (56)
→ the stability limits of Phases 0 and 1 coincide at A=0
and the PT is therefore of 2nd kind. (57)
- if B<0 than Phase 1 is stable if (ref. to eq. (19)),
→ the stability limits overlap and the PT is of 1st kind,
described by OP (ref. to eq. (45)). (58)
- the respective PTL calculates according to
F ( , P=0) = F ( , P=0) (59)
2
+Q
C
BA
3
2
2
−Q
2Q
Q
2Q
0=Q
Hempel, 2015 - 2016 Page 38 of 49
Landau Theory of Phase Transitions
and yield A = B2/(4C) (ref. also to case 3of this paper) (60)
PTL 1-2: From F ( , P=0) = F ( , P) follows after cumbersome
calculations for the PTL:
(61)
This PTL shall go continuously over to the PTL as derived
for the PT between Phases 0 and 1 (ref. to eqs. (55-58))
To check this out, D will be set to zero. From eq. (61)
follows: (62)
Rewriting eq. (62) results in:
(63)
and has two different solutions for A:
1. A=0 → describes the case where B>0 (ref. to eqs. (56,57))
The OP in Phase 1 is (64)
2. A = B2/(4C) → describes case where B<0 (ref. to eq. (58))
The OP in Phase 1 is (65)
032
23222432334
22524243246
=++
+−−
CDCDEB
ADCBECAEBACE
022 24243246 =− CAEBACE
( ) 04 22 =− BCAA
2
+Q
2
−Q
Q 0=Q
Hempel, 2015 - 2016 Page 39 of 49
Landau Theory of Phase Transitions
PTL 2-3: From F ( , P) = F ( , P) follows after cumbersome
calculations for the PTL:
(66)
PTL 1-3: Since the stability limits of Phases 1 and 3 coincide
at (ref. to eqs. (22,41))
the PT is therefore of 2nd kind. (67)
PTL 0-3: From F (Q=0, P=0) = F ( , P) follows after intense
calculations for the PTL (ref. to Gufan et al.):
(68)
with E
FDAA
2
~−=
2
2
32D
F
CD
F
BA −=
03
~32
~~2
222232
222223226
=+
−−+
DEBD
ADCACACE
2
2
642 CED
E
FA
+=
0=Q Q
Q
Hempel, 2015 - 2016 Page 40 of 49
Landau Theory of Phase Transitions
Summary:
Looking at the stability condition for Phase 3 (ref. to eq. (39)):
It turns out that in case the phase transition lines can only be obtained from
. The accompanied PTs are all of 2nd kind. Remembering that
(ref. to eq. (12)) it requires B>0 and (“weak coupling”).
The case (“strong coupling” between OPs and P) materializes if B>0 and
or respectively if and B>0 or B<0. In this case new PTLs appear.
One word on the symmetry effects on PTs described by two 1-component OPs with a
lowest order interaction term like .
This term is of course compatible with all symmetries. If we assume that and belong
to different irreducible representations of the high symmetry group of the Paraphase,
depending on the magnitude of interaction of the two Order Parameters they can either
appear successively or simultaneously (“triggered PT”), when coming from the Paraphase.
In the Phase where the two OPs are both present the symmetry is characterized by the
intersection of the symmetry elements accompanied by the onsets of and .
This means that for triggered PTs the lowering of symmetry is greater than the lowering of
symmetry connected with one OP only.
( ) 01216 222 + QCEPQ
0
022 == PQ24 FBE −=
BEFBE 22 +−
BEF 2+ EBF 2−
0 Q
22 PQ Q P
Q P
Hempel, 2015 - 2016 Page 41 of 49
Landau Theory of Phase Transitions
On the following summary slide an overview is provided about all stability limits and validity
constraints for the individual Phases, followed by graphical representations of different
cases – with particular focus on the cases with .
Let’s look at interesting examples:
1. In all discussed cases the PTL 1-3 describes completely (sometimes partly) a PT of 2nd
kind. This is because the stability limits of Phases 1 and 3 coincide there( ).
PT of 2nd kind means, that there is no “jump-like” appearance of the OP at the PTL
1-3 when coming from Phase 1 (decreasing D). Indeed both OPs ( for Phase 1 and
for Phase 3) exactly coincide at PTL 1-3 (same numerical values), and is going
to grow like a typical 2nd order PT OP on top of - as D further decreases. When
starts to exist in Phase 3 simultaneously OP P becomes present (acc. to eq. (11)).
2. If looking at sequential PTs – e.g. 0→1→2 (ref. to slide 41) the 1st transition is of 2nd kind
( appears smoothly). If A and D further decrease the subsequent PT 1→2 is of 1st
kind (provided this transition happens above the Three-Phase Point Q). Exactly at the
PTL 1-2 drops to zero (because of lost stability) and appears with a jump.
Remarkable is the variety of possible PTs in dependence on the coefficients B and F. Even
direct PTs between Phases 0 and 3 can be described (“triggered PT”). Assuming that
coefficients A and D depend (or not) only on temperature (T) then all relevant PTs lie on the
thermodynamic path in the A-D-Plane, described either by A(T) & B=const. or by B(T) &
A=const. or by A(T) & B(T).
0
2
2
32D
F
CD
F
BA −=
Q 2
Q
2Q
2Q
2Q
2Q
2+Q
2P2
+Q
Hempel, 2015 - 2016 Page 42 of 49
Landau Theory of Phase Transitions
Phase
Sta
bili
ty
1 Not stable Not stable Not stable
2 Not stable Not stable Not stable Not stable
3 Not stable Not stable
---
Stable if: and
Va
lidity
1
--- --- ---
--- --- ---
2 --- --- --- ---
3
--- --- ---
--- --- ---
2
+Q
2
2
32D
F
CD
F
BA −=
2
−Q 2
+Q 2
−Q2P
C
BA
3
2
0D ,2
DE
FA
0A
limits no & 0
EC
B
C
BD
E
FA &
6320
2 +−
limits no & 0
EC
B
C
BD
E
FA &
6320
2 +−
2
2
482 CED
E
FA
+=
0) (means
AB
or AB
0&0
0&0
0) (means
B
0
0D
C
BA
3
2
2
2
482 CED
E
FA
+=
DE
FA &
or DE
FA &
20
20
0
2
2
32D
F
CD
F
BA −=
Hempel, 2015 - 2016 Page 43 of 49
Landau Theory of Phase Transitions
The PTL 2-3 (1st kind) goes over at Point Q to PTL 1-2 (also 1st kind) and then approaches smoothly the PTL 0-1 at A=0.
At The Three Phase Point Q PTL 2-3 (ref. to eq. (66)) and PTL 1-3 (2nd kind, ref. to eq. (67)) meet. Also the PTL 1-2 (eq.
(61)) goes through Point Q.
CE
FD
24
)12( =
2
22
192
3)12(4
CE
FA
+=
Note: ‘+‘ is valid
2+Q
22 &PQ−
2P
Hempel, 2015 - 2016 Page 44 of 49
Landau Theory of Phase Transitions
The PTL 2-3 (1st kind) goes over at Point Q to PTL 1-2 (also 1st kind) and then approaches smoothly the PTL 0-1 at
A=B2/(4C). At The Three Phase Point Q PTL 2-3 (ref. to eq. (66)) and PTL 1-3 (2nd kind, ref. to eq. (67)) meet. Also the
PTL 1-2 (eq. (61)) goes through Point Q.
CE
FD
24
)12( =
2
22
192
3)12(4
CE
FA
+=
Note: ‘+‘ is valid
2−Q
22 &PQ−
2P
Hempel, 2015 - 2016 Page 45 of 49
Landau Theory of Phase Transitions
The PTL 1-3 (2nd kind) goes over at the Tricritical Point T to PTL 1-3 (1st kind) and then becomes at Point L the PTL 0-3
(1st kind, “triggered” PT, ref. also to Holakovsky). Further at Point N this PTL becomes the PTL 2-3 (1st kind).
Since B>0 the PT between Phases 0 and 1 is always of 2nd kind (PTL 0-1).
CE
FD
12
=
2
22
48
2
CE
FA
+=
2+Q
22 &PQ−
2P
Hempel, 2015 - 2016 Page 46 of 49
Landau Theory of Phase Transitions
The PTL 1-3 (2nd kind) ends at the Tricritical Point T and goes over into the 1st kind stability line of Phase 3 (blue dots). The
Point T lies above PTL 0-1 (1st kind, A = B2/(4C)) if F>-(2|B|E)1/2 and also above the Three-Phase Point L. This means that
in this case the transition between Phases 1 and 3 will be always of 2nd kind (PTL 1-3). The PTL 0-3 (1st kind) is a
“triggered” PT (ref. also to Holakovsky). Further at Point N the PTL 2-3 (1st kind) begins.
CE
FD
12
=
2
22
48
2
CE
FA
+=
2−Q
22 &PQ−
2P
Hempel, 2015 - 2016 Page 47 of 49
Landau Theory of Phase Transitions
With the settings B>0, F>0 and ∆>0 all PTs are of 2nd kind. In case C=0 the PTL 1-3 become straight a line.
2+Q
22 &PQ+
2P
Hempel, 2015 - 2016 Page 48 of 49
Landau Theory of Phase Transitions
With the settings B>0, F<0 and ∆>0 all PTs are of 2nd kind. In case C=0 the PTL 1-3 become straight a line. This behavior
includes the PT-sequence 0→2→3 (path: A=constant and D decreasing), where all PTs are of 2nd kind (ref. to Holakovsky).
2+Q
22 &PQ+
2P
Hempel, 2015 - 2016 Page 49 of 49
Landau Theory of Phase Transitions
7. Literature
/01/ Izyumov Yu. A., Syromyatnikov V. N., “Phase Transitions and Crystal Symmetry“,
Kluwer Academic Publishers, Dordrecht / Boston / London, 1990
/02/ Gufan, Yu. M., “Thermodynamic Theory of Phase Transitions”, Publisher: University of
Rostov on Don, 1982
/03/ Kocinsky, J., “Theory of Symmetry Changes at Continuous Phase Transitions”,
Elsevier, Amsterdam-Oxford-New York, 1983
/04/ Gufan Yu. M, Larin E. S., Sov. Solid State Phys. 22(2), 270(1980)
/05/ Gufan Yu. M., Torgashev V. I., Sov. Solid State Phys. 22(6), 951(1980)
/06/ Holakovsky, J., phys. stat. sol. (b) 56, 615(1973)
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 1 of 25
Temperature Dependences of Material Coefficientsat Structural Phase Transitions
Table of Contents
1. Theory
2. Elastic Coefficents
3. Dielectric Impermeabilities
4. Piezoelectric Coefficients
5. Clarifying Example Calculation
6. Calculation of Temperature Dependences
7. Summary
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 2 of 25
Issue
Provided the Thermodynamic Potential function, suitable to describe
the Phase Transition under consideration, has been properly derived –
How can the temperature dependences of certain material coefficients
be predicted ?
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 3 of 25
1. Theory
• Basing on standard group theoretical principles the Free Energy Power Expansion
has been derived as
F = F (Qi, Sj, Pk) with (1)
Qi – Order Parameter (OP) components
Sj – Deformation Tensor components (j=1...6, Voigt notation)
Pk – Polarization Vector components (k=1...3)
• For the sake of generality let‘s assume that external „forces“ are applied to the
crystal – according to:
Tj = F/ Sj (External Stress) (2)
Ek= F/ Pk (External Electrical Field) (3)
Note: For a mechanically and electrically „free crystal“ Tj and Ek has to be set = 0!
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 4 of 25
• Solving the equation system (2), (3) we obtain
Sj = Sj (Qi, Ek, Tj) (4)
Pk = Pk (Qi, Ek, Tj) (5)
• Inserting (4) and (5) into (1) leads to the Free Energy Expression in the form:
F = F (Qi, Ek, Tj) with Ek, Tj being the external „forces“ (6)
Note: Qi are the only internal parameters the Free Energy is depending on.
• That Energy Expression (eq. 6):
• Is only suitable to determine the equilibrium values of the Order Parameter
Components (Qi), and therefore the possible changes of the crystal symmetry
• Has no „dynamic“ relevance because the Sj and Pk are not able to follow the
fast changes („vibrations“) of the Order Parameter („soft mode“)
(7) 0,0
Q
P
Q
S ki
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 5 of 25
2. Elastic Coefficients
• To calculate the elastic coefficents, formulae (1) has to be re-considered:
F = F (Qi, Sj, Pk) (8)
taking into account the following conditions:
1. Via certain couplings the Qi can be dependent on Sj [Qi = Qi (Sj)]
2. Via electromechanical couplings Pk can be dependent on Sj
[Pk = Pk (Sj)]
• The elastic coefficients shall be calculated under the condition that the Order Parameter
(as the „driver“ of the phase transition) can freely move, i.e.
Ki = F/ Qi = 0 (9)
with Ki being the conjugated force with regard to Qi
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 6 of 25
• Equation (8) has dynamic relevance because the Order Parameter can quasistatically
follow the „slow“ changes of the strain Sj
• When putting together the Free Energy Expression according to (1) or (8), the
corresponding material coefficients have to be entered under following conditions:
1. Bare elastic stiffnesses coij – at Qi = 0 and Pk = 0
2. Bare dielectric impermeabilities oij - at Qi = 0 and Sj = 0
3. Bare piezoelectric coefficients homn - at Qi = 0
• Finally the elastic stiffnesses calculate according to:
(10)
with cij being the stiffnesses under the influence of the order parameter‘s action (ref. to
(9) above) when the Order Parameter can freely move.
ji
ij SS
Fc
=
2
0
j
i
S
Q
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 7 of 25
• The mechanical stresses calculate:
(11)
= 0 (ref. to (9))
• Finally the stiffnesses calculate:
(12)
influence of OP‘s influence of piezoelectric
action coupling
m
i
i im
m S
Q
Q
F
S
FT
+
=
+
+
=
j n
j
jmi n
i
imnm
mn S
P
PS
F
S
Q
QS
F
SS
Fc
222
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 8 of 25
Goal is now to express:
From eq. (9) follows that the Order Parameter can freely move:
(13.1)
Suggested ‘Ansatz‘: (13.2)
Interchange of subscripts j k leads to (13.3)
Intersting into eq. (13.1) yield: (13.4)
?=
m
i
S
Q
0==
j
j
KQ
F
0=
jn Q
F
dS
d0
22
=
+
n
i
i ijjn S
Q
F
QS
F
ijj jnn
i RQS
F
S
Q
−=
2
−=
ik
k knn
i RQS
F
S
Q
2
0222
=
−
+
k
ik
kni ijjn
RQS
F
F
QS
F
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 9 of 25
(13.5)
Comparison of coefficients results in:
(13.6)
and can be compactly re-written as:
(13.7)
Note: The term describing the piezoelectric coupling in eq. (12) can be calculated likewise.
jnkn
ikk iji QS
F
QS
FR
F
=
222
=
=
jk
jkR
Fik
k iji 1
02
kjiki ij
RQQ
F=
2
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 10 of 25
• Compact, final expression (see e.g. /1, 2/) :
(14)
Note: The stiffnesses are those to be measured at constant electric Field at isothermal conditions
with
(15)
being the equations to determine the R-matrix components in (14), xy = 0 if x y,
xy = 1 if x = y, represents the Kronecker symbol
vn
2
kvvk, km
2
kn
2
ikki, im
2
nm
2
mn PS
FR
PS
F
QS
FR
QS
F
SS
Fc
−
−
=
kvk jk
vjiki ji
kj RPP
FR
F
=
=
22
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 11 of 25
3. Dielectric Impermeabilities
• The calculation of the impermeabilities is analogue to the calculation of the elastic
coefficients as shown in Chapter 2.
• The tensor of the dielectric impermeabilities is related to the tensor of the dielectric
susceptibility like () = ()-1
(16)
• The electricl fields calculate:
(17)
= 0 (rf. to (9))
m
i
i im
m P
Q
Q
F
P
FE
+
=
ji
ij PP
F
=
2
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 12 of 25
• Finally the impermeabilites calculate:
(18)
influence of OP‘s influence of piezoelectric
action coupling
• Compact expression after some treatment (see /1, 2/):
(19)
Note: The impermeabilities are those to be measured at const. mech. Stress at isothermal conditions
with
(20)
+
+
=
j n
j
jmi n
i
imnm
mn P
S
SP
F
P
Q
QP
F
PP
F 222
vn
kvvk kmkn
ikki imnm
mn SP
FR
SP
F
QP
FR
QP
F
PP
F
−
−
=
2
,
22
,
22
kvk jk
vjiki ji
kj RSS
FR
F
=
=
22
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 13 of 25
4. Piezoelectric Modules
• The calculation of the piezomodules is analogue to the calculation of the elastic
coefficients and the impermeabilities as shown in chapters 2 and 3.
(21)
• The 1st derivatives of the Free Energy Expansion with regard to the polarization
components gives the electrical fields (see eq. (17))
(22)
= 0 (rf. to (9))
m
i
i im
m P
Q
Q
F
P
FE
+
=
ji
ij SP
Fh
=
2
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 14 of 25
• The 2nd derivatives with regard to the strains give the piezoelectric modules:
(23)
influence of OP‘s
action
(24)
with
(25)
+
=
i n
i
imnm
mn S
Q
QP
F
SP
Fh
22
kn
ikki imnm
mn QS
FR
QP
F
SP
Fh
−
=
2
,
22
iki ji
kj RQQ
F
=
2
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 15 of 25
5. Clarifying Example Calculation
• Let‘s assume the phase transition under consideration is driven by a 2-component
order parameter Qi (i = 1,2)
• Exemplarily the matrix components Rik (ref. to (13.7)) are calculated as follows:
1st Case kj
using the abbreviations: follows
iki ji
kj RQQ
F
=
2
k
j
k
j
RQQ
FR
F2
2
2
1
1
2
0
= +
...11
2
11 osa QQ
FF
=
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 16 of 25
(a)
(b)
2nd Case k=j
(c)
(d)
Solution of this system of 4 equations (a)...(d) yields:
21221112
22211211
0
0
RFRF
RFRF
+=
+=
j
j
j
j
RQQ
FR
F2
2
2
1
1
2
1
= +
22221212
21211111
1
1
RFRF
RFRF
+=
+=
212122211
1212
2122211
1122
2122211
2211
RFFF
FR
FFF
FR
FFF
FR
=−
=
−=
−=
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 17 of 25
6. Calculation of Temperature Dependences
• Depending on the coupling term between the OP-components (Qi) and the state
parameters (Si, Pj) in the Free Energy Expansion (1), following typical cases shall
be considered (e.g. ref. to /3, 4/):
1. True proper (ferroelastic) phase transition (2nd kind, OP identical to strain component S)
(30)
(31)
(32)
The stiffness calculate now acc. to eqs. (10) and (14) as:
(33)
o2
o
oo
TT for B
A-S
and TT for 0S to leads 0S
F
TTAA with BSASF
=
==
−=+= )(4
1
2
1 42
oooS
oooS
TT for )T-(T -2A-2Ac
and TT for )T-(T AAc
==
==
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 18 of 25
2. Pseudo-proper (ferroelastic) phase transition (2nd kind)
(34)
(35)
(36)
(37)
Next step is the calculation of the free energy F’ of the stress free crystal by
inserting of eq. (35) into eq. (34). After some calculation yields:
(38)
BQAQ
F Stability
DSBQAQQ
F F of Mininum
Qc
D-S DQSc
S
F crystal free
TT AA with DQSSc BQAQF
oS
oS
oo2o
S
2
2
2
3
42
3:
:
0:""
)(2
1
4
1
2
1
+=
++=
=+==
−=+++=
)(4
1)(
2
1'
242
2
oS
'oS c
DA A mit BQQ
c
DA F −=+−=
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 19 of 25
Reforming the expression for A’ results in a change of the transition temperature:
(39)
Searching for the minimum of F’ (eq. (38)) with respect to Q yield:
(40)
(41)
Now the elastic coefficients are calculated acc. to eq. (14).
(42)
For the Paraphase (T>TC) follows from eq. (42), using Q=0 (in eq. (37)) and
with reformed eq. (39) with regard to D:
(43)
)('2
ooS
oCCo Ac
DT T with )T-(TAA +==
C
C
TT for B
AQ
TT for Q
−=
=
'
0
2
2
oSQQ
oSS F
Dc
QS
FF
QS
Fcc
221
2
−=
−= −
)()( CoC
CoSS TT TT
TTcc
−+−
−=
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 20 of 25
For the Ferrophase (T<TC) follows from eq. (42), using Q=-A’/B (in eq. (37)) and
with reformed eq. (39) with regard to D:
(44)
3. Improper (ferroelastic) phase transition (2nd kind)
It is well known that an improper transition is accompanied by at least two order
parameter components (ref. to /5/). For simplicity we confine our calculations
here to one order parameter component Q.
(45)
(46)
(47)
(48)
)(2)(
)(2
TT TT
TT cc
CoC
CoSS
−+−
−=
2DSBQAQ
F Stability
DSQBQAQQ
F F of Mininum
Qc
D-S DQSc
S
F crystal free
TT AA with SDQSc BQAQF
oS
oS
oo22o
S
++=
++=
=+==
−=+++=
2
2
2
3
22
42
3:
2:
0:""
)(2
1
4
1
2
1
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 21 of 25
Next step is the is the calculation of the free energy F’ of the stress free crystal
by inserting of eq. (46) into eq. (45). After some calculation yields:
(49)
As it can be seen the transition temperature is not changed: To=TC
Searching for the minimum of F’ (eq. (49)) with respect to Q yield:
(50)
(51)
For the strain in the phases (ref. to eq. (46)) is valid:
(52)
)2
()2
(4
1
2
1'
24
22
oS
oS c
DB B' with Q
c
DB AQF −=−+=
C
oS
2
C
TT for
c
2D-B
A-
B
AQ
TT for Q
=−=
=
'
0
2
2
2Qc
DS
oS
−=
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 22 of 25
Now the elastic coefficients are calculated acc. to eq. (14).
(53)
For the Paraphase (T>TC) follows from eq. (53), applying Q=0:
(54)
For the Ferrophase (T<TC) follows from eq. (53), using Q2 from eq. (51) and
with FQQ (see eq. (48)) where the expressions for Q2 and S (see eqs. (51), (52))
have been inserted:
(55)
oSQQ
oSS F
QDc
QS
FF
QS
Fcc
2221
2 4−=
−= −
oSS cc =
B
Dcc o
SS
22−=
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 23 of 25
Proper Transition
Pseudo-proper Transition
Improper Transition
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 24 of 25
7. Summary
• The formalism presented firstly by Slonczewski et al. to predict the temperature
dependences of isothermal elastic coefficients has been extended with respect to:
- Dielectric Impermeabilities
- Piezoelectric Modules
as well as with respect to:
- Influences of measuring conditions (constant Pj or Ej respectively
constant Tk (“free” crystal) or Sk (“clamped” crystal))
• The outlined formalism is fully applicable at phase transitions that are characterized
by multi-component order parameters.
Temperature Dependences of Material Coefficients
at Structural Phase Transitions
Hempel, 2007 & 2017 Page 25 of 25
Literature
1. Rehwald, W., Adv. In Physics, Vol. 22, 1973, No.6, p. 721 ff.
„The study of structural phase transitions by means of ultrasonic experiments“
2. Slonczewski, J. C. and Thomas, H., Phys. Rev. B1, 3599 (1970),
„Interaction of elastic strain with the structural transition of SrTiO3“
3. Bulou, A., Rousseau, M., Nouet, J., Key Engineering Materials, Vol. 68, 133-186 (1992)
„Ferroelastic Phase Transitions and Related Phenomena“
4. Carpenter, M. A., Salje, E. K. H., et al. American Mineralogist, Vol 83, 2-22 (1998)
5. Wadhawan, V. K., Phase Transitions, Vol. 3, 3-103 (1982)
„Ferroelasticity and Related Properties of Crystals“
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 1 of 43Hempel, 2002 & 2016
Table of Contents
1. Observations made and models discussed so far
2. Crystal physical Basics
3. Landau Modelling
a. Proper Ferroelastic Transition
b. Pseudo-Proper Ferroelastic Transition
4. Comparison with experimental results and Summary
5. Literature
Annex: Investigation of an Improper Ferroelastic
Transition Model (4mmFmm2)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 2 of 43Hempel, 2002 & 2016
1. Observations made and models discussed
Rb4LiH3(SO4)4, abbreviated RLHS, was firstly reported to be ferroelastic by Wolejko
et al. /01,02/ and by Mroz et al. /03/.
They found that RLHS:
• Is of tetragonal symmetry by habitus (as grown), confirmed by X-ray investigation
• Exhibits a 2nd Order Phase Transition (PT) at 137 K
• Shows twinning in planes perpendicular to the c-axis. Two types of perpendicular
domain walls were observed along [010]- and [100]-directions
• Can be transferred to a mono-domain state by application of external stresses
along [010]- and [100]-directions, respectively
• Has a polar axis parallel to c-axis above and below the PT-point
• Shows a -type specific heat anomaly and almost no anomaly of the dielectric
permittivities at the PT-point
• It was concluded that RLHS should undergo a ferroelastic PT according to Aizu’s
species 4mmFmm2 /04/.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 3 of 43Hempel, 2002 & 2016
Other authors as of Hempel et al. /05/, Zúñiga et al. /06/ and later also Mroz et al.
/07/ discarded the possibility of Species 4mmFmm2 and proposed that the
symmetry change is correctly described by 4F2 (Space Groups (SGs): P41= C42,
P21=C22). The PT is of ferrodistortive (zellengleich, equitranslational) type which
means that the number of molecules per unit cell Z (Z=4) does not change through
the PT /08,09/.
This finding bases on following observations:
• X-ray and neutron diffraction investigations of RLHS
• The crystal exhibits a rotary power (optical activity) of -0,28°/mm along its c-axis
in the paraelastic phase (at T=293 K), which is generally not allowed for a
paraphase belonging to Point Group (PG) 4mm.
• Ferroelastic domain walls (“W’-walls” according to Sapriel’s classification /10/)
are present in planes perpendicular to c-axis. They are mutually perpendicular
and are rotated around c-axis by 35° (which contradicts to the observations
made by Wolejko et al. /01/ because in case 4mmFmm2 the domain walls would
be fix “W-walls”, mutually perpendicular and along either [100]- and [010]- or
[1-10]- and [110]- directions).
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 4 of 43Hempel, 2002 & 2016
Symmetry elements of PG 4:
1, 413, 4
23, 4
33
Order of PG 4mm = 4
(symmetry generators are marked red)
Coordinate System (x(1), y(2), z(3)):
(a, c are the lattice parameters, tetragonal a=b)x
y
z
As result of the PT 4 → 2 the fourfold symmetry along z is lost, but c-axis remains
the polar axis. Both phases are piezoelectric.
Since the PT is of ferrodistortive type, the Landau modelling can be done just within
the framework of the Point Groups – here PG 4 and PG 2.
All physical quantities and coefficients are related to the coordinate system x, y, z.
2. Crystal physical Basics
z
y
x
ecc
eab
eaa
=
=
=
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 5 of 43Hempel, 2002 & 2016
Since the ferroleastic phase exhibits symmetry of PG 2 (with two symmetry
elements:1, 423) the number of possible ferroelastic domains („states“) calculates as:
Species 4F2 leads to the spontaneous deformation tensors for the two states /04/:
According to Sapriel /10/ two mutually perpendicular domain walls are possible:
Those walls are denominated W‘-walls, because their orientation depends on the
spontaneous strain components and they again on the temperature.
22
4
2===
PG of order
4 PG of ordern
122 Sb and SS 2
1a with ab
ba
S ab
ba
S =−=
−
−−
=
−= )(
000
0
0
000
0
0
11221
( )a
bab pwith -y/p xypx
2 2/12++===
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 6 of 43Hempel, 2002 & 2016
3.a Landau Modelling - Proper Ferroelastic Transition
First step to compile the Free Energy Expansion (FEE) for RLHS is to look up the
Irreducible Representations (IDs) of PG 4. According to Kovalev /11/:
ID 1 413 42
3 433
T1 1 1 1 1
T2 1 i -1 -i
T3 1 -1 1 -1
T4 1 -i -1 i
IDs T2 and T4 are complex. To get
physically meaningful IDs they
have to be transformed into real
form /08/. Combination (here
addition) of complex T2 with its
conjugate complex T4 leads to:
ID 1 413 42
3 433
A=Γ1 1 1 1 1
B=Γ3 1 -1 1 -1
E=E1+E2
(=Γ2+Γ4)
2 0 -2 0
ID A is the identical representation
and not of interest here.
Next step is to check which ID can
induce PG 2 symmetry.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 7 of 43Hempel, 2002 & 2016
As it can be easily shown (e.g. Kocinski /12/) that in case of equitranslational phase
transitions only those IDs will lead to a certain low symmetry phase (subgroup of PG
4) where the characters ( ) of the symmetry elements (g) of the subgroup (here
PG 2) are either:
=1 for one-dimensional IDs
or
= 0, 2 for two-dimensional IDs
→ Only ID B is able to induce the PG 2 Symmetry, consisting of the symmetry
elements 1, 423.
Next step is the inspection of the reduction coefficients to derive information
regarding permitted appearance of ferroelasticity and ferroelectricity. Acc. to /12/,
the calculation of
yields mE=2 regarding ferroelasticity (i.e. allowed!) and mE=0 regarding ferroelec-
tricity (i.e. not allowed!). Of course this is just a formal group-theoretical justification
of the obvious behaviour. Meaning, that ferroelasticity is always allowed if the crystal
system of the paraphase changes as a result of a PT.
)( g
)( g
)()(4
1g g m B
gPE =
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 8 of 43Hempel, 2002 & 2016
Whether ID B is an active representation can be decided after consideration of the
Landau- and Lifshitz-Conditions.
It turns out that the Lifshitz-Condition is always fulfilled if the PT is of ferrodistortive
type (characterized by a PT-mode condensing at the Γ–Point of the Brillouin zone
and accompanied by a wave-vector ).
The Landau-Condition is always fulfilled for ferrodistortive PTs if the ID under
consideration is one-dimensional and real /12/.
Alternatively the following calculation should be done:
In the present case the result is zero, which confirms that the Landau-Condition isn’t
violated.
→ The ID B is an active representation and can induce the PT 4→2.
0=k
with g G
oGgB
o
0)(1 ?3
=
( )3233 )(6
1)()(
2
1)(
3
1)( g g g g g BBBBB ++=
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 9 of 43Hempel, 2002 & 2016
Next step to get Free Energy Expansion (FEE) is to decompose the (ordinary)
strain tensor and (ordinary) polarisation vector into their irreducible parts regarding
PG 4.
Utilisation of the projection operator (see e.g. /12/)
with = P resp. S and
= operator related to
symmetry element gk
yield the following results:
Strain S: ½(S11+S22), S33, ½(S11-S22), S23, S13, S12 respectively
(in Voigt’s Notation /13/ and
normalized*)
*) Note: ½(S11+S22)=S1/2+S2/2 → |S1/2+S2/2|=(0,5²+0,5²)1/2=1/(2)1/2→ ½(S1+S2)/|S1/2+S2/2|=(S1+S2)/(2)1/2
=
3
2
1
P
P
P
P
=
332313
232212
131211
SSS
SSS
SSS
S
)()()()( xfgTgG
dxf
Gg
kijkij
k
=)(xf
)( kgT
65421
321 ,,,
2
)(,,
2
)(SSS
SSS
SS −+
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 10 of 43Hempel, 2002 & 2016
Polarisation P: P1, P2, P3
When decomposing the strain tensor into its symmetry adjusted irreducible parts it
turns out that ID B induces two parts of the irreducible strain tensor, namely
In case of a proper ferroelastic transition one part plays the role to the primary OP
and couples linearly to the other secondary OP /14/. It is assumed that in both
possible cases the coupling between the two OP-parts is weak (i.e. the coefficient I
in eq. (2) below is small). Up to now it’s not clear which one represents the primary,
respectively secondary OP.
Each term of the FEE has to be invariant under the action of all the symmetry
elements of PG 4, but it is sufficient to prove this invariance under the action of the
“generators” of PG 4 only. These are here just the symmetry operation 413.
To this end the transformation properties of Si, Pk, and Q (Q to be Order Parameter
within a pseudo-proper model which will be discussed later) have been compiled:
621 ,
2
)(S
SS −
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 11 of 43Hempel, 2002 & 2016
(1)
IDIrreducible
Part
After action
of
413
A
B
E
6
21 )(2
1
S
SS −
IDIrreducible
Part
After action
of
413
A
B , Q , -Q
E
3
0
0
P
5
4
S
S
3
21 )(2
1
S
SS +
3
21 )(2
1
S
SS +
6
21 )(2
1
S-
SS −−
4
5
S
S−
3
0
0
P
02
1
P
P
0
0
0
0
0
0
−
01
2
P
P
ID A induces the identical representation of PG 4 whereby ID B (one component, two
parts) induces PG 2 (incl. OP Q), and ID E induces PG 1 (two components).
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 12 of 43Hempel, 2002 & 2016
The Free Energy Expansion (FEE) – reflecting the dynamics of the crystal - for a
proper ferroelastic transition can now be written:
F = FS + FP + FS-P
FS = A(S1-S2)2 + B(S1-S2)
4 + C(S1+S2)2 + DS3
2 + E(S42 +S5
2) +
+ GS62 + HS6
4 + I(S1-S2)S6 + J(S1+S2)S3 + K(S1-S2)2(S1+S2) +
+ L(S1-S2)2S3 + MS6
2(S1+S2) + NS62S3 + n(S1-S2)
2S32 (2)
FP = O(P12+P2
2) + PP32
FS-P = RS3P3 + T(S1+S2)P3 + U(P1S4+P2S5) + V(P1S5-P2S4) +
+ W(S1-S2)2(P1
2+P22) + XS6
2(P12+P2
2) + Y(S1-S2)2P3 + ZS6
2P3
We can distinguish between 2 cases where an acoustic mode is softening:
1.The driving/primary OP is the strain difference (S1-S2) and therefore the only
temperature–dependent parameter of eq. (1) is A = Ao(T-To). (3)
2.The driving/primary OP is the shear strain S6 and therefore the only
temperature–dependent parameter of eq. (1) is G = Go(T-To). (4)
Note: (S1-S2) and S6 cannot appear as primary OPs simultaneously, since they do not exhibit a common
coefficient in FEE eq. (2) (A ≠ G, see overleaf).
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 13 of 43Hempel, 2002 & 2016
The following material coefficients are used in FEE:
(5)
A straightforward method to elaborate the respective irreducible parts of the material
coefficients can be referred to in /15/. Third order elastic constants are listed at
Brugger /16/ and 4th order constants at Chung /17/.
It has to be noted that the coefficients in FEE have to satisfy following conditions:
1. coij - bare elastic stiffnesses measured at constant polarization (and at
constant OP Q as discussed later in the pseudo-proper model)
( )
hV hU hT hR P
cc 24
1n O c
6
1N cM
oooo
o
o
oo
o
ooo
15143133
33
33
12331133
11
11366166
2
1
2
1
2
1
2
1
6
1
======
−=====
)()(
24
1
2
1
2
1
2
1
)(4
1)34(
192
1)(
4
1
1231131121111316
6666664433
012111122111211111211
oooooo
oooo
oooooo
cc 12
1L cc
24
1K cJ cI
cH cG cE cD
cc C ccc B cc A
−=−===
====
+=+−=−=
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 14 of 43Hempel, 2002 & 2016
1.ßoij - bare dielectric impermeabilities measured at constant strain (and at
constant OP Q as discussed later in the pseudo-proper model)
2.hokl – bare piezoelectric modules are those measured at constant OP Q (as
discussed later in the pseudo-proper model)
Considering now that no external electrical (Ek) or mechanical (Ti) fields are applied,
the solution of the equation system:
(6)
yields the State Parameters in dependence of the OP for the “free crystal”:
1st case: OP = (S1-S2)
(7)
Note: Ki are constants, not explicitly shown here.
0
0
==
==
kk
ii
EP
F
TS
F
( )
( ) ( ) ( )
( ) ( ) ( ) ( )
( ) 02
22
1
22
1
0
5216
2
212
212
2
212
211
2
213
2
21221
2
21121
==−−=
−+−−=−+−=
−=−=+
−===
SS SSG
IS
SSK
SSS SSK
SSS
SSKS SSKSS
SSKP PP
4
3
3
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 15 of 43Hempel, 2002 & 2016
Inserting the state parameters into FEE (eq. (2)) the following expression is obtained,
which is suited to find the equilibrium states of the crystal:
F = A‘(S1-S2)2 + B’(S1-S2)
4 (8)
with renormalized coefficients and B’. (9)
As result the actual PT-temperature is changed from To to (10)
It is assumed that the coupling coefficient I= is small and
therefore yields: ToTc.
The equilibrium values of the OP are easily derived from eq. (8):
(11)
Solution 1: Paraphase (12)
Solution 2: Ferrophase (13)
G
IAA
4'
2
−=
GA
ITT
ooc
4
2
+=
( )( ) ( )32121
21
'4'20 SSBSSASS
F−+−==
−
( ) 021 =−SS
( )'2
'2
21 B
ASS −=−
oc16
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 16 of 43Hempel, 2002 & 2016
The material coefficients are now calculated acc. to /18,19/:
(14)
Note: - Stiffnesses are those at constant electric field (E), Impermeabilities at constant mech. stress (T).
- circled expressions exhibit throughout this paper “piezoelectric coupling terms” that describe the
differences between cEmn and cP
mn respectively between ßTmn and ßS
mn.
vn
2
kvvk, km
2
nm
2Emn PS
FR
PS
F
SS
Fc
−
=
vn
kvvk kmnm
Tmn SP
FR
SP
F
PP
F
−
=
2
,
22
nm
mn SP
Fh
=
2
kvk jk
vj RPP
F with
=
2
kvk jk
vj RSS
F with
=
2
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 17 of 43Hempel, 2002 & 2016
Throughout the complete paper following abbreviations are used to describe the
piezoelectric coupling terms (acc. to eq. (14)):
(15)
It needs to be remembered that the expressions:
(case 1), and
(case 2) are temperature dependent!
( ) ( )
( ) ( ) 2131211331211
2131211121133
22112
1
oooooo
oooooo
cccccc
ccccccRR
−+−
−−++
==
22112112 RRRR −=−==
( ) 213121133
121133
2 oooo
oo
cccc
ccR
−+
+=
( ) 213121133
1332233113
2 oooo
o
cccc
cRRRR
−+
−====
0366326621661 ====== RRRRRR
)(441211 oooo TT AAcc −==−
oo cR
cRR
66
66
44
5544
11===
)(2266 ooo TT GGc −==
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 18 of 43Hempel, 2002 & 2016
( )( ) ( )
( ) ( )
( )
( ) ( ) ( )( ) ( )
( ) ( ) ( )( ) ( )
( ) ( ) ( )
( )
( )
( )
−−−=
−−=
−==−
−
+−+−++−−=
−
+−++−−−++−=
−
+−++−−+++−=
−+−=−+=
−=−=−=
−
+−++−=
−
=
−
−
−
21
66
1661626
21
66
1661616
1211
2
2111221112111141211121112
2
211122111211114211121111211121122
2
211122111211114211121111211121111
122211
1
12111211
1
2
21112211121111412112
21
21
31
31
)(222
1 with
)34(16
1
2
1
2
1
)34(16
1
6
1
2
1
2
1
)34(16
1
6
1
2
1
2
1
)(2)(424
1
)(22
1
2
1
)34(16
1
2
1
SSc
ccc
SSc
ccc
TTAAcc
SScccKccccc
SScccKSSccccccc
SScccKSSccccccc
TTATTAccc
TTAcccc
SScccKccSS
F
o
ooE
o
ooE
oo
oo
oooooooE
oooooooooE
oooooooooE
ococo
EEE
OP
oo
ooEE
OP
ooooo
OP
(16)
Paraphase (note: TcTo)
Ferrophase (note: TcTo)
Inverse OP-Susceptibility
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 19 of 43Hempel, 2002 & 2016
( ) ( )
( )
( )
( ) ( )
SSc
cZSSKcc
hhcc
hhcc
SSc
chZ SS
c
ccc
oo
ooE
o
oooE
o
oooE
oo
oo
o
ooE
2
2126633
21622
2156666
11
215
214
4455
11
215
214
4444
21
6633
163321
66
1636636
4
2
3
−−−+=
+−=
+−=
−+−−=
( )( )
( )( ) ( )
( )( ) ( )21
33
33211231131332
21
33
33211231131331
33
2332
21123311333333
2
6
1
2
6
1
SSh
YSScccc
SSh
YSScccc
hSScc
12
1cc
o
ooooE
o
ooooE
o
ooooE
−+−−−=
−−−−+=
−−−+=
morphic coefficient!
(17)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 20 of 43Hempel, 2002 & 2016
( )
( )
( ) ( ) 23333333113
2
21366
216
22
212
112
31113333
44
215
2142
21266
216
1122
44
215
2142
21266
216
1111
24
12
2
2
ooo
o
oooT
o
oo
o
ooT
o
oo
o
ooT
hRhhRSSc
cZSSYRhR
c
hhSS
c
cXW
c
hhSS
c
cXW
−−−−−−−=
+−−
++=
+−−
++=
( )
( )
( )21
66
1636
1425
1524
1515
1414
3333
213132
213131
2
2
2
SSc
cZh
hh
hh
hh
hh
hh
SSYhh
SSYhh
o
o
o
o
o
o
o
o
o
−−=
=
−=
=
=
=
−−=
−+=
morphic coefficient!
(18)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 21 of 43Hempel, 2002 & 2016
2nd case: OP = S6
(19)
Note: Ki are constants, not explicitly shown here.
Inserting the state parameters into FEE (eq. (2)) the following expression is obtained,
which is suited to find the equilibrium state of the crystal:
F = G’S62 + H’S6
4 (20)
with renormalized coefficients and H’. (21)
As result the actual PT-temperature is changed from To to (22)
It is assumed that the coupling coefficient I= is small and
therefore results: ToTc.
The equilibrium values of the OP is easily derived from eq. (20):
(23)
A
IGG
4'
2
−=
AG
ITT
ooc
4
2
+=
( )
( ) 02
4242
0
521
6
2
67
26
2
67
1
2
68
2
6721
2
6621
===−
+=−=
==+
===
SS S A
I- SS
SA
IS
KS S
A
IS
KS
SKS SKSS
SKP PP
46
3
3
3
66
6
'4'20 SHSGS
F+==
oc16
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 22 of 43Hempel, 2002 & 2016
Solution 1: Paraphase (24)
Solution 2: Ferrophase (25)
(26)
06 =S
'2
'26 H
GS −=
( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
( ) ( ) 26911221112111141211122211
26911221112111141211121112
26911221112111146
1211
161121111211121122
26911221112111146
1211
161121111211121111
'33
2
66
1
6666
1
26
33
22610662
6
21
)34(16
1
2
12
4
1
)34(16
1
2
1
2
1
)34(16
1
3
1
2
1
2
1
)34(16
1
3
1
2
1
2
1
)(22
1)(4
2
4
SK ccc Kccccc
SK ccc Kccccc
SK ccc KScc
cccccccc
SK ccc KScc
cccccccc
TTGH
ZTTGc
TTGcc
SZ
SKcS
F
oooooEEE
oooooooE
ooo
oo
oooooooE
ooo
oo
oooooooE
ocoocoE
OP
oooE
OP
o
oOP
+−++−=−+
+−+−++−−=
+−++
−
−+++−=
+−++
−
−−++−=
−+
−−==
−===
−+=
=
−
−
−
Paraphase (note: TcTo)
Ferrophase (note: TcTo)
Inverse OP-Susceptibility
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 23 of 43Hempel, 2002 & 2016
( )( )
( )( )
( )( )
( )
( )o
oooE
o
oooE
o
ooE
o
o
oo
ooooE
ooo
oo
oo
ooooE
ooo
oo
oo
ooooE
ooE
ooE
hhcc
hhcc
Sh
ZScc
hS
cc
ccccc
Scc
chYS
cc
ccccc
Scc
chYS
cc
ccccc
Sc
cc
Sc
cc
11
2
15
2
144455
11
2
15
2
144444
6
33
33636636
33
2
332
62
1211
2
16123311333333
6
121133
16336
1211
161231131332
6
121133
16336
1211
161231131331
6166
1626
6166
1616
2
3
1
12
1
4
3
1
4
3
1
3
3
+−=
+−=
−=
−−
−+=
−−
−
−+=
−+
−
−−=
+−=
+=
morphic coefficient!
(27)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 24 of 43Hempel, 2002 & 2016
( )
( )
( )2
333333311326
66
2262
1211
2162
112
31113333
44
215
2142
62
1211
216
1122
44
215
2142
62
1211
216
1111
24
48
42
42
ooo
ooo
oooT
o
oo
oo
ooT
o
oo
oo
ooT
hRhhRSc
ZS
cc
cYRhR
c
hhS
cc
cWX
c
hhS
cc
cWX
−−−−
−−=
+−
−++=
+−
−++=
morphic coefficient!6363333
142515246
1211
163132
141415156
1211
163131
2
4
4
ZSh hh
hh hh Scc
cYhh
hh hh Scc
cYhh
o
oo
oo
oo
oo
oo
oo
==
=−=−
+=
==−
−=
(28)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 25 of 43Hempel, 2002 & 2016
Now the Free Energy Expansion (FEE) for a pseudo-proper ferroelastic transition
(case 3) can be compiled (similar to Quirion /22/ and David /23/):
F = FOP + FS + FP + FS-P + FS-Q + FP-Q
FOP = A1Q2 + B1Q
4
FS = A(S1-S2)2 + B(S1-S2)
4 + C(S1+S2)2 + DS3
2 + E(S42+S5
2) +
+ GS62 + HS6
4 + I(S1-S2)S6 + J(S1+S2)S3 + K(S1-S2)2(S1+S2) +
+ L(S1-S2)2S3 + MS6
2(S1+S2) + NS62S3 + n(S1-S2)
2S32
FP = O(P12+P2
2) + PP32 (29)
FS-P = RS3P3 + T(S1+S2)P3 + U(P1S4+P2S5) + V(P1S5-P2S4) +
+ W(S1-S2)2(P1
2+P22) + XS6
2(P12+P2
2) + Y(S1-S2)2P3 + ZS6
2P3
FS-Q = r(S1-S2)Q + sS6Q + tS3Q2 + u(S1+S2)Q
2 + vS4S5Q + w(S42+S5
2)Q2
FP-Q = x(P12+P2
2)Q2 + yP32Q2
The driving OP is now Q (e.g. related to a soft optical mode) and couples linearly
with both (S1-S2) and S6 (acoustic modes). The only temperature–dependent
parameter in eq. (29) is A1 = A1o(T-To). (30)
.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 26 of 43Hempel, 2002 & 2016
Solution of the equation system (ref. to eq. (6)) yield the state parameters of the
“free crystal” in dependence of the OP Q.
(31)
Inserting the state parameters into FEE (eq. (29)) the following expression is for the
free crystal is obtained:
F = A1‘Q2 + B1‘Q
4 (32)
with renormalized coefficients
and B1’ (not explicitly shown here). (33)
As result the actual PT-temperature is changed from To to (34)
( )
( )
QK
QK S Q
KQ
K S
QKQ IAG
sArI S QKQ
IAG
rGsI SS
S S QKS QKSS
QKP PP
19618
43
3
2222
4
2
4
2
00
0
182162
182161
2221
52
172
1621
21521
−=+=
=−
−==
−
−=−
====+
===
( )2
22
114
'IAG
AsGrrsI A with AAA
−
+−=+=
ooc A
ATT
1
−=
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 27 of 43Hempel, 2002 & 2016
• The equilibrium values of the OP is easily derived from eq. (32):
(35)
Solution 1: Paraphase (36)
Solution 2: Ferrophase (37)
• To calculate the temperature dependences of the material coefficients (ref.
to eq. (40, overleaf) also the 2nd derivatives FQQ (= inverse order parameter
susceptibilities) have to be used. They can be derived from FEE (eq. (29)),
whereby the equilibrium OP and state parameters (ref. to egs. (31, 36, 37))
have to be used:
FQQ = 2A1 + 12B1Q2 + 2tS3 + 2u(S1+S2) + 2yP3
2
FQQ = 2A1 + (12B1 + 2tK17 + 2uK16 + 2yK15)Q2 (38)
• This yields finally:
FQQ = 2A1o(T-To) for the Paraphase, and
FQQ = 4A1o(Tc-T) + 2A1o(Tc-To) for the Ferrophase (39)
(see also /20,21/)
311 '4'20 QBQA
Q
F+==
0=Q
'2
'
1
12
B
AQ −=
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 28 of 43Hempel, 2002 & 2016
• The material coefficients which will be derived are those valid for isothermal
conditions.
• The stiffnesses are those at no applied electric field (Ei) and freely moving OP.
• The dielectric impermeabilities are at no applied mechanical stress (Tij) and
freely moving OP.
• The piezoelectric coefficients are those at freely moving OP.
vn
2
kvvk, km
2
n
2
m
2
nm
2
mn PS
FR
PS
F
QS
F
F
QS
F
SS
Fc
−
−
=
−12
kvk jk
vj RPP
F with
=
2
vn
kvvk kmnmnm
mn SP
FR
SP
F
QP
F
F
QP
F
PP
F
−
−
=
− 2
,
221222
kvk jk
vj RSS
Fwith
=
2
(40)
QS
F
F
QP
F
SP
Fh
nmnm
mn
−
=
− 21222
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 29 of 43Hempel, 2002 & 2016
• Since (S1-S2) and S6 couples linearly with OP Q, as well as inter alia, it doesn’t
become easily obvious which combination of elastic stiffnesses has to go to
zero when approaching Tc.
• According to Boccara /24/ and others /20/ the eigenvalues of the stiffness matrix
of the PG 4, related to the eigenvector, which describes the transition to the
ferrophase with PG-symmetry 2, must become zero at Tc.
• First step is to look at the transformation matrix Dkl (k,l =1…6) which generates
the symmetry adapted strain vector (decomposition into irreducible parts) from
the ordinary strain, expressed as a 6 component vector. As shown on page 9
this is:
−
+
=
−
6
5
4
21
3
21
6
5
4
3
2
1
2
2
100000
010000
001000
00002
1
2
1000100
00002
1
2
1
S
S
S
SSS
SS
S
S
S
S
S
S
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 30 of 43Hempel, 2002 & 2016
• Application of transformation matric Dkl onto the matrix of stiffnesses of PG 4
leads to the stiffness matrix in form of symmetry adapted (irreducible) parts, like:
• As shown previously the specific eigenvector being in charge of the PT 4F2
reads like:
cc
c
c
ccc
cc
ccc
c
oo
o
o
ooo
oo
ooo
okl
−
+
=
6616
44
44
161211
3313
131211
00200
00000
00000
20000
00002
00002
~
T
k SSS
S
−= 6
21 002
00~
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 31 of 43Hempel, 2002 & 2016
• The eigenvalue equation:
leads to the specific eigenvalue (see also /24/):
whereby the solution with ‘+’ has to be discarded because the eigenvalue needs
to become zero at Tc.
It has to be noted that the stiffnesses and (as discussed in cases
1 and 2 above) will individually not become zero at the phase transition point.
* Note: The expression derived by Quirion et al. /22/ appears to be incorrect.
02
00200
00000
00000
20000
00002
00002
6
21
6616
44
44
161211
3313
131211
S
0
0
SS0
0
cc
c
c
ccc
cc
ccc
oo
o
o
ooo
oo
ooo
=
−
−
−
−
−−
−
−+
( )*
16
2
661211661211
2
82
1
+−−+−=
ooooooo ccccccc
oo cc 1211 −oc 66
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 32 of 43Hempel, 2002 & 2016
( ) ( )
( ) ( )
( )( )
( ) ( )
( ) ( )
( )
( )
( )o
o
oooE
QQo
ooE
QQo
ooE
ooEEE
ooEEE
oE
oooE
oooE
h
F
QtQKcc
12
1cc
F
suQrQK
c
ccc
F
suQrQK
c
ccc
F
rQKcc
2
1ccc
4
1
F
r cc
2
1ccc
4
1
F
uQruQrQKcc
F
uQrQKQKcccc
F
uQrQKQKcccc
33
233
2222
18123311333333
18
66
1661626
18
66
1661616
22
201211122211
2
1211122211
2201212
22
20181121111122
22
20181121111111
4
2
31
2
31
2
2
22
2
6
1
2
6
1
−−−+=
+−−
−−=
+−
−=
−+−=−+
−−=−+
+−+−−=
+−−+−−=
+−+−+=
(41)
Paraphase
Ferrophase
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 33 of 43Hempel, 2002 & 2016
( ) ( )
( ) ( )
( )
( )
QQo
oE
E
o
oooE
o
oooE
QQoo
oo
o
ooE
QQo
ooooE
QQo
ooooE
F
sQ
KZKcc
vQc
hhwQcc
hhwQcc
F
2tsQ- QK
c
chZ
c
ccc
F
tQuQrQK
hYcccc
F
tQuQrQK
hYcccc
22
33
2192
216666
45
11
215
2142
4455
11
215
2142
4444
18
6633
1633
66
1636636
18
33
331231131332
18
33
331231131331
4
2
2
2
3
222
6
1
222
6
1
−
−+=
=
+−+=
+−+=
+−=
+−−
−−−=
+−
−−+=
morphic coefficient!
morphic coefficient!
(42)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 34 of 43Hempel, 2002 & 2016
( )
( )
oooo
o
o
o
oooo
o
oT
o
oo
o
ooT
o
oo
o
ooT
F
QysKQZKh
hh hh hh hh
F
QytKhh
F
uQrQyKQYKhh
F
uQrQyKQYKhh
hRhhRhRQKc
ZQKYRyQ
c
hhQK
c
cXWxQ
c
hhQK
c
cXWxQ
315
1936
1425152415151414
415
3333
4315
183132
4315
183131
23333333113
23111
2219
66
222
182
112
3333
44
215
21422
18266
2162
1122
44
215
21422
18266
2162
1111
42
8
242
242
24
122
22
22
−=
=−===
−=
+−−−=
+−+=
−−−−−+=
+−
+++=
+−
+++=
morphic coefficient!
(43)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 35 of 43Hempel, 2002 & 2016
S3, S1+S2, P3
Case 1: OP = S1-S2 Case 2: OP = S6
Case 3: OP = Q (pseudo-ferroelastic)
(Coefficients r, s, t, u, y of eq. (29 ) are
assumed to be positive. Arrows mean jumps.)
Qualitative summary of results:
S1, S2
1(2)
2(1)
11(22)
22(11)
c12
(S1-S2), S6
1-2(6)
6(1-2)
S3, S1+S2, P3
S1, S2
1(2)
2(1)
c12
(S1-S2), S6
1-2(6)
6(1-2)
S3, S1+S2, P3
c11, c22
S1, S2
1(2)
2(1)
c12
(S1-S2), S6
1-2(6)
6(1-2)
¼(c11+ c22 -2c12)
c33
c11, c22
TcTcTc
11(22)
22(11)1−
OP½(c11-c12)
¼(c11+ c22 -2c12)
¼(c11+ c22 -2c12)
¼(c11+ c22 -2c12)
½(c11-c12)
½(c11-c12)
¼(c11+
c22 -2c12)
c11
c22
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 36 of 43Hempel, 2002 & 2016
c33, 11=22, 33
c16
c26
c44, c55
h33, h14, h15,
h24, h25,
Case 1: OP = S1-S2 Case 2: OP = S6 Case 3: OP = Q (pseudo-ferroelastic)
c31, c32
h31, h32
31(32)
c66
32(31)
c36, h36
c33, 11=22, 33
c16
c26
c44, c55
h33, h14, h15,
h24, h25,
c31, c32
h31, h32
31(32)
c66
32(31)
c36, h36
c16
c26
c44=c55,
11=22,
33, h33
c36, h36
c66
c45
c31, c32
h31, h32
h14, h15,
h24, h25
Tc
TcTc
( )1−
OP
31(32)
32(31)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 37 of 43Hempel, 2002 & 2016
4. Comparison with experimental results and summary
To compare the predictions with experimental results it has to be kept in mind that:
• In absence of external forces RLHS exhibits always a multidomain state in
the Ferrophase /01,02,03/, whereby the current predictions base on the
fact that the crystal is in a monodomain state of a well defined orientation
(ref. to slide 4). This means that the experimental results for the
Ferrophase need to be individually considered case by case, whether
applicable.
• In the Paraphase such particularities don’t appear, and therefore data
measured there, are assumed to be more useful to check theoretical
predictions.
• The predictions have been derived taking not into account the ordinary
pyroelectric- as well as the ordinary thermal expansion- effects.
• The measuring conditions can play an important role. E.g. the elastic
coefficients are usually measured at adiabatic conditions, whereby the
predictions are done for isothermal coefficients.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 38 of 43Hempel, 2002 & 2016
The measured state parameters (S1, S2, S3, P3) (ref. to /03,05,07/) fit nicely with the
predictions, but do not differ for the 3 cases discussed.
The elastic coefficient c66 (ciiρvii2) is in fair agreement with case 3, as well as
c44=c55. c33 shows a small jump downward at Tc which can also only be explained
within case 3 (see also overleaf).
ref. to Quirion /22/ref. to Hempel /05/
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 39 of 43Hempel, 2002 & 2016
ref. to Hempel /05/
The elastic coefficient c11=ρv112 is nicely in line with the predictions of case 3 (see
also overleaf the data from Wolejko /02/ measured with pendulum devise).
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 40 of 43Hempel, 2002 & 2016
ref. to Wolejko /02/ ref. to Hempel /05/
The coefficient-combination (c11-c12) follows qualitatively only the predictions derived
for case 3.
The dielectric permittivity 33 is in good agreement with all 3 cases.
33
( ) 11
1
11 cs −
( ) 33
1
33 cs −
( )1211 cc −
33
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 41 of 43Hempel, 2002 & 2016
Summarizing it can be stated that – as proposed by Quirion et al. /22/ - the pseudo-
proper ferroelastic Landau model indeed describes best the so far obtained
experimental results.
Major arguments are:
• The (small) jump of c33 at Tc which can only be explained with this model.
• No linear but hyperbolic temperature-dependences have been experimentally
observed so far for the elastic coefficients c11, c12, c66 in the Paraphase which
contradicts the proper ferroelastic models
• The measured temperature dependence of c11 matches well with the prediction
of the pseudo-proper model.
To prove the model and to get more clarity on the correct Free Energy Expansion,
especially the temperature dependences of - incl. c11, c12, (c11-c12), c31, c66, c16, c36
as well as of selected piezoelectric coefficients should be investigated in more
detail.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 42 of 43Hempel, 2002 & 2016
/01/ Wolejko, T., et al., Ferroelectrics, 81, 175(1988)
/02/ Wolejko, T., et al., Ferroelectrics, 81, 179(1988)
/03/ Mroz, B., et al., J. Phys. Cond. Matter, 1, 4425(1989)
/04/ Aizu, K., J. Phys. Soc. of Japan, 28, 706(1970)
/05/ Hempel, H., et al., phys. stat. sol. a(110), 459(1988)
/06/ Zúñiga, F. J., et al., Acta Cryst., C46, 1199(1990)
/07/ Mroz, B., et al., Phys. Rev. B55, 11174(1997)
/08/ Janovec, V., et al., Czech J. Phys., B25, 1362(1975)
/09/ Boyle, L. L., et al., Acta Cryst., A28, 485(1972)
/10/ Sapriel, J., Phys. Rev. B12, 5128 (1975
/11/ Kovalev, O. V., “Representation of Crystallographic Space Groups”, Taylor &
Francis Ltd., 1993
/12/ Kocinsky, J., “Theory of Symmetry Changes at Continuous Phase Transitions”,
Elsevier, Amsterdam-Oxford-New York, 1983
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Page 43 of 43Hempel, 2002 & 2016
/13/ Sirotin, Yu. I., Shaskolskaya, M. P., “Fundamentals of Crystal Physics”,
Mir Publishers, Moscow, 1982
/14/ Toledano, P., et al., Phys. Rev. B27, 5717(1983)
/15/ Hempel, H., PhD Dissertation, 1886
/16/ Brugger, K., J. of Appl. Phys., 36, 759(1965)
/17/ Chung, David, Y., Acta Cryst. A30, 1(1974)
/18/ Slonczewski, J. C., Thomas, H., Phys. Rev. B1, 3599(1970)
/19/ Hempel, H., internal paper
/20/ Bulou, A., et al., Key Engineering Materials 68, 133(1992)
/21/ Carpenter, M., A., American Mineralogist 83, 2(1998)
/22/ Quirion, G., J. Phys. Cond. Matter, 21, 455901(2009)
/23/ David, W. I. F., J. Phys. C: Solid State Phys. 16, 5093(1983)
/24/ Boccara, N., Annals of Physics, 47, 40(1968)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 1 of 22
Annex
Investigation of an Improper Ferroelastic Transition Model
(Aizu Species: 4mmFmm2)
Preface:
• Initially a symmetry change according to Aizu Species 4mmFmm2 (/01/, /02/,
/03/) was assumed
• This was later discarded (/05/, /06/, /07/), but in current paper a symmetry
change 4mm→mm2 is exemplarily investigated, based on an improper model
to demonstrate the differences between the proper / pseudo-proper and an
improper transition behaviour.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 2 of 22
Primitive 4mm Symmetry of the Paraelastic Phase
• 8 corresponding space groups do exist: Ci4v with i=1...8 (e.g. ref. to /08, /09/)
• Reciprocal lattice is calculated according to :aj x ak
bi = 2 i, j, k = 1, 2, 3ai (aj x ak)
• At tetragonal lattices there is bi || ai (i=1...3) reciprocal lattice is
primitive, too
• Now the BRILLOUIN Zone can be constructed
• Change of translational symmetry is described by certain lattive vectors (k-
vectors), that need to be in ‚symmetric position‘, i.e.:
kj = -kj + K with K being an entire reciprocal lattice vector (Kocinsky /09/)
• The complete set of those vectors can be identified either by trying out or,
more conveniently, by inspecting of Kovalev‘s tables /08/:
½ b1, ½ b2, ½ b3, ½ (b1+b2), ½ (b2+b3), ½ (b1+b3), ½ (b1+b2+b3), 0
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 3 of 22
• In order to describe the transition from 4mm to mm2, the k-vector with
inherent mm2-symmetry has to be selected via following procedure:
All 8 symmetry elements gk of 4mm (each represented as matrix
R(gk)) will be applied to all k-vectors in symmetric position. The set of
those symmetry elements that transform the selected vector into itself
or into an equivalent vector (acc. to the relation kj = kj +K) constitute
a point symmetry group. We are searching for mm2!
Example: ½ b1 = (/a, 0, 0)
-1 0 0 /a -/a
Symmetry 42 0 -1 0 0 = 0
element: 0 0 1 0 0
• Finally four k-vectors exhibit mm2-symmerty, namely:
½ b1, ½ b2, ½ (b2+b3), ½ (b1+b3)
• The fist two vectors belong to Brillouin-Zone point “X”, and the others to “R”
(/08/)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 4 of 22
• In order to identify the interrelation between these vectors, the
corresponding ‘stars‘ have to be determined (e.g. ref. to Kocinski /09/)
• the star of an arbitrarily selected k-vector is obtained when all symmetry
elements (of the point group 4mm) act on it like k’ = Rk + K, e.g.:
0 -1 0 /a 0
1 0 0 0 = /a a.s.o.
0 0 1 0 0
= 41 = ½ b1 = ½ b2
• The corresponding star consists now of all nonequivalent k’-vectors resulting
from this operation; particularly we get:
Star (K1): ½ b1, ½ b2 (has got two arms)
Star (K4): ½ (b2+b3), ½ (b1+b3) (has got two arms, too)
Note: All other k-vectors in symmetrical position as mentioned above possess symmetry
4mm, and therefore cannot describe a transition to mm2. Consequently we have toexclude them from further considerations.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 5 of 22
• Star K1 describes a doubling of the elementary cell in either x- or y-direction,
or in x- and y- direction simultaneously (causing a quadruplication of the
elementary cell in the ferroelastic phase)
• Star K4 behaves like K1 but with an additional doubling along z-direction
• Since nothing specific is known regarding the definitive symmetries of the para-
and ferrophase, we will confine our subsequent considerations to star K1
• Next step is the provision of the Full Irreducible Representations (FIRs) of the
space groups Ci4v with regard to star K1 (all FIRs of the Space Groups are
listed at Kovalev /08/). Their dimensions is two because the projective (small)
representations are all one-dimensional, but star K1 has two arms
• The structure of the related matrixes are identical for Ci4v with i=1,3,5,7 and
i=2,4,6,8 respectively
• Again for the reasons, that we don‘t know the exact space groups, we confine
the investigations to Ci4v with i=1,3,5,7
• In the following we will look at the FIRs (here provided for the generators, which
is sufficient)
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 6 of 22
Generators 41 m[100]
0 1 1 0
D1 1 0 0 1
0 1 -1 0
D2 1 0 0 -1 (1)
0 -1 1 0
D3 1 0 0 -1
0 -1 -1 0
D4 1 0 0 1
Notes: - The matrix structure is identical for all 4 FIRs for the space groups Ci4v with i=1,3,5,7 and
therefore the related Free Energy Expressions (FEE) are identical.
- D1-D4 are active representations, because they are real and the star K1 consists of two
arms and consequently the Landau condition is fulfilled. The Lifshitz is also obeyed since
the projective (small) representations are all one-dimensional (see e.g. /09/).
FIRs
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 7 of 22
Expression of the Free Energy Expression (FEE) with regard to Order
Parameter:
All Terms of the FEE) have to be invariant under the action of all
symmetry operations (incl. integral translations) of the high
temperature phase (para-phase)! We will consider here FIR D1 only.
• Since the above mentioned 4 FIRs are all 2-dimensional, we have to
consider an Order Parameter (OP) made up of 2 OP-Components
• Therefore we interpret the OP as a vector with 2 components p, q
p
OP =
q
1st Step - invariance under the symmetry generators (sufficient):
p p‘ q
(i) e.g D1(41) = =
q q‘ p
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 8 of 22
p p‘ p
(ii) D1(m[100]) = =
q q‘ q
2nd step - invariance under complete translations (translation operator T(tN)):
Vectors describing complete translations: tn= n1a1 + n2a2 + n3a3 with n1 n2 n3 –
integer and - basis vectors of the high
temperature unit cell (4mm)
(iii) T(tN) p = p‘ = e-ik1tn p = ± p (+ if n1=2, 4,…| - if n1=1, 3,…)
(iv) T(tN) q = q‘ = e-ik2tn q = ± q (+ if n2=2, 4,…| - if n2=1, 3,…)
k1, k2 are the arms of K1, whereby k1 relates to OP-component p and k2 to q. Since
the OP-terms have to be invariant under all possible complete translations tn –
this means for all values (and combinations) of n1 and n2.
This demand can only be satisfied with quadratic terms: p2, q2, p2q2, …
zyx eca eaa eaa === 321 ,,
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 9 of 22
3rd step – construction power terms of p and q (n-exponent, δi–constants):
!
(i) δ1 pn + δ2 qn = δ1 p‘n + δ2 q‘n → invariant if n even (see (iii), (iv)
above) and δ1=δ2 (see (i), (ii))
!
(ii) δpnqm = δp‘nq‘m → invariant if n,m even (see (iii) and (iv))
The Order Parameter Part (FOP) of the Free Energy Expression reads:
FOP = A(p2+q2) + B(p4+q4) + Cp2q2 + ... (2)
Remarks:
1. Expansion up to 4th power of p, q shall be sufficient because of
2nd order phase transition (experimentally evidenced)
2. According to Landau‘s theory is A=Ao(T-To) while B, C ... are
usually temperature independent coefficients
3. „Conditions of Landau and Lifshitz“ are obeyed (D1-D4 are
“active representations”) which allows a 2nd order and
commensurate phase transition
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 10 of 22
4th Step – Couplings of order parameter (OP) and macroscopic parameters:
• The order OP itself represents a microscopic parameter which is related to a
certain point of the BRILLOUIN Zone (at k0 in the present case – “X”-point)
• All macroscopic parameters are directly related to the -point‘s symmetry (which
reflects the symmetry of the high temperature point group under consideration –
here 4mm)
• The -Point of the BRILLOUIN Zone relates to k=0
• All Parameters related to k=0 are ‘per se’ and always invariant under complete
lattice translations of the high temperature symmetry
• All coupling terms of OP and macroscopic parameters must be simultaneously
invariant regarding the action of the symmetry generators as well as of complete
translations
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 11 of 22
Macroscopic Strain and Polarization (tensor of 2nd rank and ploar vector)
Application of projection operator ρTij (see /09/) on a function f(x):
with:
dT - Dimension of irreducible representation at -point
|G| - Order of group 4mm (here =8)
gk - Symmetry elements of 4mm
T*(gk)ij - complex conjugated matrix elements of irreducible representation
R(gk) - Symmetry operator connected with element gk
with f(x) = P (polarization vector) and Sij (strain tensor):
)()()()( x fg Rg G
dx f
Ggkijkij
k
=
=
3
2
1
P
P
P
P
=
332313
232212
131211
SSS
SSS
SSS
S
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 12 of 22
Result: 1. Irreducible parts of the polarization: P1, P2, P3
2. Irreducible parts of the strain tensor (in Voigt notation (/10/) and
normalized (see also /11/):
To finally construct the FEE in which all terms have to be invariant under the
action of the symmetry generators of the point Group 4mm (in particular the
elements: 41 and m[100]) and under complete translations related to star K1.
Therefore a transformation table for all parameters (OP, polarization, strain)
has been compiled.
65421
321 ,,,
2
)(,,
2
)(SSS
SSS
SS −+
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 13 of 22
Parameter After action
of 41
After action of
m[100]
After complete
translation
p q p ±p
q p q ±q
3
2
1
P
P
P
−
3
1
2
P
P
P
−
3
2
1
P
P
P
3
2
1
P
P
P
−
+
6
5
4
21
3
21
2
)(
2
)(
S
S
S
SSS
SS
−
−
−−
+
6
4
5
21
3
21
2
)(
2
)(
S
S
S
SSS
SS
−
−
−
+
6
5
4
21
3
21
2
)(
2
)(
S
S
S
SSS
SS
−
+
6
5
4
21
3
21
2
)(
2
)(
S
S
S
SSS
SS
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 14 of 22
Now the FEE can be finally compiled:
FEE = FOP + FP + FS + FOP-P + FOP-S with
FOP = A(p2+q2) + B(p4+q4) + Cp2q2 + (3)
FP = D(P12+P2
2) + EP32 +
FS = F(S1+S2)2 + GS3
2 + H(S42+S5
2) + IS62 + J(S1-S2)
2 +
FOP-P = K(P12+P2
2)(p2+q2) + L(P12+P2
2)p2q2 + MP32(p2+q2) + NP3
2p2q2 +
O(P12-P2
2)(p2-q2) +
FOP-S = P(S1+S2)(p2+q2) + Q(S1+S2)p
2q2 + RS3(p2+q2) + TS3p
2q2 +
US62(p2+q2) + VS6
2p2q2 + W(S42+S5
2)(p2+q2) + X(S42+S5
2)p2q2 +
Y(S42-S5
2)(p2-q2) + Z(S1-S2)(p2-q2) + a(S1-S2)
2(p2+q2) +
b(S1-S2)2p2q2 +
FS-P (piezoelectric coupling terms), FP-S-OP and other higher order energy terms have
been omitted.
The term characteristic for an improper ferroelastic transition is the bold-marked
expression in eq. (3). The elastic coefficient related to the spontaneous strain that
appears at the transition point is )(4
11211oo cc J −=
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 15 of 22
As shown in another article regarding improper transitions within this compilation,
essential features are provided here without calculation:
• Spontaneous strain (S1-S2) (p2-q2) (4)
• The transition temperature is unchanged To=TC
• All macroscopic state parameters and coefficients are independent of
temperature in the paraelastic phase (if ordinary thermal expansion and
pyroelectric effect is excluded from the model)
• The elastic coefficient related to the spontaneous strain looks like:
Tc=To
oo cc 1211 −
Tc=To
21 SS −
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 16 of 22
As discussed in another article on the Landau Theory, the Order Parameter
part of FEE FOP has got different solutions, namely:
Phase 0: (p=0, q=0) Paraelastic Phase
Phase 1: (p≠0, q=0) or (p=0, q≠0) Phase 1, Ferroelastic Phase
Phase 2: (p=q, q) Phase 2
Phase 3: (p=-q, q) Phase 3
Inspecting in eq. (3) the coupling term Z(S1-S2)(p2-q2), it turns immediately out,
that no spontaneous strain (ref. to eq. (4)) can appear in the Ferro Phases 2 and 3.
Only Ferro Phase 1 can induce such strain, whereby the two options reflect two
possible domains.
The solution (p≠0, q=0) characterizes the ferroelastic domain doubled along y-
direction, and (p=0, q≠0) the other one doubled along x-direction. Along z-direction
no doubling is taken into consideration within this model (star K1).
Be reminded, that the number of macroscopically distinguishable domains calculates
as the ratio of the orders of the point group of the paraelastic phase (4mm, order 8)
and of the ferroelastic phase (mm2, order 4) which yields 2.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 17 of 22
From geometrical considerations it becomes obvious (and can be easily proven by
application of the formalism to calculate the orientation of domain walls in
ferroelastics (see e.g. /12/)), that the walls are fixed ones, crossing x- and y-
directions under 45° and run parallel to the z-direction. Pairs of walls stand
perpendicular to each other.
As discussed by several authors (see e.g. /14/), in case of improper transitions
(which are always accompanied by an increase of the primitive unit cell in the ferroic
phase compared to the para phase), the number of different ferroic “terrains”
calculates as product of the number of macroscopic domains (here 2) and the factor
which represents the enlargement of the ferroic unit cell (here 2). Consequently there
are 2*2=4 terrains in general which differ in the (microscopic) order parameter. Those
are:
(-p, q=0), (+p, q=0), (p=0, -q), (p=0, +q)
Each ferroleastic domain is composed of 2 terrains which are separated from each
other by so called anti-phase boundaries.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 18 of 22
On the other hand, the solutions (p=q, q) and (p=-q, q), i.e. the simultaneous
appearance of p and q at the transition point, does not induce any orthorhombic
distortion (see eq. (4)), thus the tetragonality is preserved and no ferroelasticity can
be observed (i.e. just 1 macroscopic domain exists). Nevertheless, the volume of the
primitive unit cell is doubled along x- and y-directions, meaning an increase of the
unit cell by a factor of 4.
The number of terrains is therefore 4, which differ in the order parameter as follows:
(p=q, q=+|q|), (p=q, q=-|q|), (p=-q, q=+|q|), (p=-q, q=-|q|)
In case the star K4 is considered, the above made considerations are fully applicable
with the only difference that there is an additional doubling along the z-direction of
the unit cells in the ferro phase. The behavior with regard to the phase transition in
general, the order parameter(s), the state parameters and the physical coefficients
does not show differences.
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 19 of 22
z
x
y
paraphase,
p=q=0
⚫⚫
⚫⚫
ferroelastic phase
domain 2, p=0, q≠0
Doubling along y-axis
⚫ ⚫ ⚫
⚫ ⚫⚫
ferro-
elastic
phase
domain 1,
p≠0,
q=0
Doubling along x-axis
a
b
⚫ ⚫
⚫
⚫
⚫
⚫
Lattice
parameters:
a=2 atetra (1+S1)
b=atetra (1+S2)
c=ctetra (1+S3)
90°
domain walls
Lattice
parameters:
a=atetra
b=atetra
c=ctetra
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 20 of 22
z
x
y
paraphase,
p=q=0
⚫⚫
⚫⚫
(p=q, q)
and
(p=-q, q)
Doubling along y-axis
a
b
⚫
⚫
⚫
⚫
⚫
Lattice parameters:
a=2 atetra (1+S1)
b= a=2 atetra (1+S1)
c=ctetra (1+S3)
⚫
⚫
⚫
⚫
Dou
blin
g a
lon
g x
-axis
Lattice parameters:
a=atetra
b=atetra
c=ctetra
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 21 of 22
/01/ Wolejko, T., et al., Ferroelectrics, 81, 175(1988)
/02/ Wolejko, T., et al., Ferroelectrics, 81, 179(1988)
/03/ Mroz, B., et al., J. Phys. Cond. Matter, 1, 4425(1989)
/04/ Aizu, K., J. Phys. Soc. of Japan, 28, 706(1970)
/05/ Hempel, H., et al., phys. stat. sol. a(110), 459(1988)
/06/ Zúñiga, F. J., et al., Acta Cryst., C46, 1199(1990)
/07/ Mroz, B., et al., Phys. Rev. B55, 11174(1997)
/08/ Kovalev, O. V., “Representation of Crystallographic Space Groups”, Taylor &
Francis Ltd., 1993
/09/ Kocinsky, J., “Theory of Symmetry Changes at Continuous Phase Transitions”,
Elsevier, Amsterdam-Oxford-New York, 1983
/10/ Sirotin, Yu. I., Shaskolskaya, M. P., “Fundamentals of Crystal Physics”,
Mir Publishers, Moscow, 1982
Phase Transition & Landau Theory of Rb4LiH3(SO4)4
Hempel, 2002 & 2017 ANNEX Page 22 of 22
/11/ Janovec, V., et al., Czech J. Phys., B25, 1362(1975)
/12/ Sapriel, J., Phys. Rev. B12, 5128 (1975
/13/ Bulou, A., et al., Key Engineering Materials 68, 133(1992)
/14/ Wadhawan, V. K., Phase Transitions, 3, 3-103(1982)
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 1 of 42Hempel, 2007 & 2016
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Table of Contents
0. Abstract
1. About Non-Primary Ferroics
2. About Higher Order Ferroic Phase Transitions
3. Observability of Effects
4. Case (NH4)2CuCl4•2H2O – Ferroelastoelectrics
4.1. Symmetry Change at 200 K
4.2. Landau Modelling
4.3. Calculation of Temperature Dependence of State
Parameters and Material Coefficients
5. Comparison with Experimental Findings
6. Summary
7. Literature
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 2 of 42Hempel, 2007 & 2016
0. Abstract
(NH4)2CuCl4•2H2O (shortly ACC) seems to be a candidate for a ferroelastoelectric
phase transition according to Aizu’s Species .
Basing on symmetry considerations and the comprehensive usage of the Landau
Theory, the temperature dependences of State Parameters as well as of essential
Material Coefficients have been derived. All experimental data measured so far by
others are in good agreement with these results.
Moreover, is was theoretically investigated, how the crystal should behave under the
action of “external forces” (like electrical fields, mechanical stresses) imposed to the
crystal.
It is shown that the character of the phase transition should change from
ferroelastoelectric to ferroelastic, respectively to ferroelectric, or even to another
type of ferroelastoelectric behaviour – accompanied by all typical properties and
temperature dependences. Four respective scenarios have been discussed in detail.
mmmmF 24/4
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 3 of 42Hempel, 2007 & 2016
1. About Non-Primary Ferroics
• Non-Primary (or higher order) Ferroics are crystals that exhibit a certain
domain structure – usually as result of a ferroic Phase Transition from a
Prototype- or Paraphase (which are identical denominations)
• Order Parameters (OP) have the symmetry of macroscopic tensors (rank
greater than 2), e.g.
➢ Elastic coefficients (rank 4) – Ferrobielasticity
➢ Piezoelectric coefficients (rank 3) – Ferroelastoelectricity
➢ a.s.o /1/, /2/
• At a non-primary Phase Transition the crystal usually lowers its symmetry,
the Point Group (PG) changes, but the Crystal System remains (if the hexagonal and rhombohedral system are treated as one system)
• The number of non-primary ferroic domains equals the quotient of the
order of the Point Group of the Paraphase divided by the order of the
Ferroic Phase
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 4 of 42Hempel, 2007 & 2016
• Switching from one to another domain state should be possible in
principle if suitable external Forces are applied, e.g. at
• Ferroelastoelectrics - simultaneously mechanical stress and
an electric field
• Ferrobielastics - simultaneously mechanical stresses
2. About Higher Order Ferroic Phase Transitions /1/, /2/
• Proper Phase Transitions are those where the translational symmetry is
not changed (also called: equitranslational or ferrodistortive) transitions
• In general, proper transitions are related to the - Point ( =0) of the
Brillouin Zone of the Prototype Phase and the Order Parameter has the
symmetry of macroscopic tensor components
• Landau-Modelling is in principle feasible, provided there is a Group-
Subgroup relation between the two phases, the transition is of 2nd kind,
and the Landau- as well as the Lifshitz- Condition is satisfied. (But also
transitions of 1st kind and cases where the Landau-/Lifshitz – Conditions
are violated can be described by Landau’s theory satisfactorily.)
k
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 5 of 42Hempel, 2007 & 2016
• To describe proper ferroic transitions, consideration of the respective
Point Group Symmetries is sufficient
• In contrast, improper transitions change the translational symmetry. They
are also called antiferrodistortive
• Improper transitions are related to other points of the Brillouin Zone than
the -Point, and the Order Parameter is necessarily of microscopic nature
and
• Landau-Modelling requires to consider the symmetry changes of the
respective Space Groups (SG)
3. Observability of Effects
• In case a Primary- and a Non-Primary Ferroic behaviour occurs
simultaneously in a crystal, the Primary Ferroicity dominates
• Higher Order Ferroic Effects are in principle difficult to detect
• The energy input needed for higher order ferroics to switch among the
domain states is generally much higher than for primary ferroics
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 6 of 42Hempel, 2007 & 2016
• Therefore the investigation of switching effects of Non-Primary Ferroics
should be preferably undertaken in the vicinity of the Phase Transition
Temperature
4. Case (NH4)2CuCl4•2H2O – Ferroelastoelectrics
4.1. Symmetry Change at 200 K
• (NH4)2CuCl4•2H2O (abbreviated ACC) single crystals have been subject of
several experimental investigations /3/, /4/, /8/ as well as of theoretical
considerations /5/
• Firstly, Suga et al. /3/ discovered a phase transition at about 200 K, the
heat capacity shows a “-type” behaviour, thus indicating a transition of
2nd kind, which was also confirmed by Slaboszewska et al. /8/
• Bansal et al. /4/ carried out structural investigations and found a structural
change from P42/mnm ( ) to P421m ( ) at 200 K (at cooling). In
notation named after Aizu /9/, the species is called
14
4hD3
2dD
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 7 of 42Hempel, 2007 & 2016
4/mmmF42m (within the framework of Point Groups, “F” stands here for
“Ferroelastoelectric”)
• According to Toledano /2/ and Kovalev /7/, 4/mmmF42m can only occur at
the -Point of 4/mmm, which represents the prototypic phase
• -Point means that the transition is equitranslational ( =0), and of proper
respectively of pseudoproper nature
• If a transition is equitranslational than it is sufficient to describe it within
the framework of Point Groups /2/
• The Prototypic Phase 4/mmm (Symmetry Go) consists of the following
symmetry elements (gk) :
• The Generators of 4/mmm are:
• The Order of 4/mmm is |Go| =16 (total number of symmetry elements gk)
• If looking just at the Point Groups, generally there are two options (refer to
cases A, B below) how the transition can occur:
31
]110[]110[]001[]010[]100[
1
]110[
1
]110[
1132
,
1 4,4,,,,,,1,2,2,2,2,44,4,1 zzyxzzz mmmmm −−
1,2,4 11
xz
k
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 8 of 42Hempel, 2007 & 2016
Generators: Generators:
Order: |G1|=8 Order: |G1|=8
Case A:
42m ( ), Z=24...1
2dD
Case B:
42m ( ), Z=28...5
2dD
4/mmm ( ), Z=214
4hD
x
z y
]110[]110[
211131
,
),4(2,2,2,4,4,1
mm
zzyxzz
−
=
]010[]100[
211
]110[
1
]110[
31
,
),4(2,2,2,4,4,1
mm
zzzz =−
11 2,4 xz ]100[
1 ,4 mz
Note: Throughout this
entire paper one and
the same fixed Coor-
dinate System (x,y,z)
is used. It may differ
from usually applied
systems for the Point
Groups.
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 9 of 42Hempel, 2007 & 2016
• As stated before, in case of ACC it will be sufficient to undertake the
theoretical investigations within the framework of the Point Groups
• Moreover it was found (Bansal et al. /4/) that the symmetry G1 of the
ferroelastoelectric phase shall be , which corresponds to “case A”
• Note: Within the framework of Point Groups, cases A and B are equivalent
– all following results can be transferred from one to the other case by a
45°-Rotation around the z-axis. We are going to deal with case A only
• The number of possible ferroelastoelectric domains equals
• These two domains differ (at the lowest rank) in the matrix of the
Piezoelectric Constants hijk as well as in the Gyration Tensor’s
Components gkl, which appear at the transition temperature (so called
“morphic” coefficients). Above the transition temperature all hijk and gkl are
zero (centrosymmetric prototypical phase)
3
2dD
28
16
42
/4==
mofOrder
mmmofOrder
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 10 of 42Hempel, 2007 & 2016
• Piezoelectric Coefficients’ Tensor (Voigt’s Notation), e.g. Sirotin et al. /10/
Domain 1 Domain 2
• Gyration Coefficients’ Tensor (Voigt’s Notation), e.g. /10/
Domain 1 Domain 2
• The Symmetry elements that map domain 1 into domain 2 are all which
were lost at the transition, i.e.:
36
14
14
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
h
h
h
−
−
−
36
14
14
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
h
h
h
−
000
00
00
11
11
g
g
−
000
00
00
11
11
g
g
]001[]010[]100[1
]110[1
]110[31 ,,,1,2,2,4,4 mmmzz −
Note: both domains differ by sign in
their optical activity along x- resp. y-
axis. This property is also called
„Ferrogyrotropy“
Note: both domains differ by sign in
their piezolectric modules. Switching
between the domain states should be
possible via simultaneous action of
stresses and electr. fields like: S4/E1
or S5/E2 or S6/E3
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 11 of 42Hempel, 2007 & 2016
4.2. Landau Modelling
As listed at Kovalev /7/ there are 10 Irreducible Representations (IDs) of
Point Group 4/mmm (at the -Point) - Table of Characters ( )
0002-2000-210=Eu
000-22000-29=Eg
-111-1-11-1-118=B2u
1-1-1111-1-117=B2g
11-1-1-1-1-1116=B1u
-1-1111-1-1115=B1g
1-11-1-1-11-114=A2u
-11-111-11-113=A2g
-1-1-1-1-111112=A1u
1111111111=A1g
24 z
12,
12 yx
34,
14 zz
1]110[2
,1
]110[2 −1 ]001[m
]010[
,]100[
m
m
34,14 zz ]110[
,]110[
m
m −
ID
)(gi
gk
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 12 of 42Hempel, 2007 & 2016
• As it can be easily shown (e.g. Kocinski /11/) that in case of equitranslational
phase transitions ( =0) only those IDs will lead to a certain low symmetry
phase (subgroup of 4/mmm) where the characters of the symmetry elements of
the subgroup (here ) are either:
=1 for one-dimensional IDs
or
= 0, 2 for two-dimensional IDs
• → Only IDs 6 and 8 are able to induce the - PG Symmetry
Note: 6 corresponds to case A as mentioned above and 8 to case B
• Now it will be checked that 6 really can induce a ferroelastoelectric phase. To
this end it is necessary to calculate the respective reduction coefficients (e.g.
/11/):
m24
m24
)()(1 )(
ggG
mi
oGg
vP
oE =
)(gi
)(gi
k
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 13 of 42Hempel, 2007 & 2016
ID Ferroelastoelectricity Ferroeleasticity Ferroelectricity
1=A1gNo (mE=0) Not checked for
2=A1uYes (mE=1) No (mE=0) No (mE=0)
3=A2gNo (mE=0) Not checked for
4=A2uYes (mE=3) No (mE=0) Yes (mE=1)
5=B1gNo (mE=0) Not checked for
6=B1uYes (mE=2) No (mE=0) No (mE=0)
7=B2gNo (mE=0) Not checked for
8=B2uYes (mE=2) No (mE=0) No (mE=0)
9=EgNo (mE=0) Not checked for
10=EuYes (mE=5) No (mE=0) Yes (mE=1)
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 14 of 42Hempel, 2007 & 2016
• Next task is to check for the Landau- and Lifshitz Conditions:
• The Lifshitz Condition is automatically fulfilled because the transition is related
to =0
• Fulfilment of Landau’s Condition is checked by calculating the expression /11/:
with
(1+1+1+1-1-1-1-1-1-1-1-1+1+1+1+1)+
(1+1+1+1-1-1-1-1-1-1-1-1+1+1+1+1)+
(1+1+1+1-1-1-1-1-1-1-1-1+1+1+1+1) = 0
• The Landau Condition is also fulfilled → ID 6 is an “active” representation
0)(1 ?3
6=
gG
oGgo
3233 ))((6
1)()(
2
1)(
3
1)(
66666ggggg ++=
3
1)(3
6=g
6
12
1
k
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 15 of 42Hempel, 2007 & 2016
• Next step is to decompose the Strain Tensor S and Polarization Vector P
into irreducible parts with regard to PG 4/mmm
• To this end the ordinary representation of the symmetry elements T(g)
(with regard to the Cartesian coordinate system (x=1, y=2, z=3) and the
Projection Operator will be utilized:
with = S resp. P
=
3
2
1
P
P
P
P
=
332313
232212
131211
SSS
SSS
SSS
S
)()()()( xfgTgG
dxf
oGg
ijo
ij
=
ij
)(xf
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 16 of 42Hempel, 2007 & 2016
• This exercise yields the following irreducible parts which form the “basis
functions” for the Free Energy Expansion (FEE):
Strain S:
½(S11+S22), S33, ½(S11-S22), S23, S13, S12 respectively
(in Voigt’s Notation /10/ and normalized)
Polarization P:
P1, P2, P3
• Next step is to figure out the terms that will constitute the FEE. Each term
entering FEE has to be invariant under all symmetry operations of 4/mmm. But
it is sufficient to ensure the invariance just for the generating symmetry
elements (Generators)
• Following table shows how the Order Parameter Q and the State Parameters S
and P behave if the Generators act on them. For S and P this has to be done in
the Cartesian Space (x, y, z) and for Q the corresponding ID 6 has to be
applied
65421
321 ,,,
2
)(,,
2
)(SSS
SSS
SS −+
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 17 of 42Hempel, 2007 & 2016
1
Initial Irreducible Parts,
written as normalized
Vectorsafter action of after action of after action of
−
+
6
5
4
21
21
2
)(32
)(
S
S
S
SSS
SS
3
2
1
P
P
P
12x14z
−
−
−−
+
6
4
5
21
21
2
)(32
)(
S
S
S
SSS
SS
−
−
−
+
6
5
4
21
21
2
)(32
)(
S
S
S
SSS
SS
−
+
6
5
4
21
21
2
)(32
)(
S
S
S
SSS
SS
−
−
−
3
2
1
P
P
P
−
−
3
2
1
P
P
P
−
3
1
2
P
P
P
Q−QQ−Q
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 18 of 42Hempel, 2007 & 2016
• The FEE describes the dynamics and can now be written (ref. also to /5/):
F=F(Q,S,P)=FOP+FS+FP+FP_S+FOP_P+FOP_S+FOP_S_P
with
FOP = AQ2+BQ4
FS = C(S1-S2)2+D(S1+S2)
2+ES32+F(S4
2+S52)+GS6
2+H(S1+S2)S3+
r(S1+S2)S62+xS3S6
2
FP = J(P12+P2
2)+KP32
FP_S = L(P12+P2
2)(S1+S2)+M(P12+P2
2)S3+NP32 (S1+S2)+OP3
2 S3+ (1)
RP1P2S6+ g(P12-P2
2)(S1-S2) +h(P2P3S4+P1P3S5)
FOP_P = tQP1P2P3+UQ2(P12+P2
2)+VQ2P32
FOP_S = WQ2(S1+S2)+XQ2S3+YQ2(S42+S5
2)+ZQ2(S1-S2)2+bQ2S6
2
FOP_S_P = dQP3S6+eQ(P1S4+P2S5)+ fQP3S4S5
whereby all term-factors are assumed to be constant, except A=Ao(T-To).
Factor B has to be positive B>0.
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 19 of 42Hempel, 2007 & 2016
• The transformation matrix within the “Voigt’s Space” that maps
into has got the form:
and can be equally used to calculate the irreducible parts of the elastic
coefficients, entering the FEE
• This yields:
(2)
• The upper index “o” means that the elastic stiffnesses are those at constant Q
and P and the impermeabilities are those at constant Q and S.
−
+
6
5
4
21
21
2
)(32
)(
S
S
S
SSS
SS
6
5
4
3
2
1
S
S
S
S
S
S
−
100000
010000
001000
00002
1
2
1000100
00002
1
2
1
oooooo
ooooooo
KJccHcG
ccFcEccDccC
332211231366
5544;3312111211
2
1;
2
1
2
1;;
2
1
2
1
2
1
2
1);(
4
1);(
4
1
======
===+=−=
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 20 of 42Hempel, 2007 & 2016
4.3. Calculation of Temperature Dependences of State Parameters and
Material Coefficients
(3)
k
k
i
i
EP
F
S
F
=
=
• Solving this system of equations if non or certain values of „external forces“ Ti
(mechnical stress) resp. Ek (electric field) are applied yield following 4 selected
scenarios. Inserting the calculated expressions for Si and Pk into FEE gives
the Free Energy just as a function of the OP Q.
• The actual transition temperature Tc and the temperature dependence of Q
can be derived. Moreover, according to the Curie Principle (ref. e.g. /11/) the
symmetry of the Para- as well as of the Ferroic Phase reduces if certain
“external forces” are applied. Resulting Aizu’s Species are denominated.
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 21 of 42Hempel, 2007 & 2016
Applied External Forces Symmetry Change (Species)
(based on Case A)
Kind of Ferroicity
Ek=0, Ti=0 Ferroelastoelectric (Scenario 1)
E30 Ti=0 Ferroelastic (Scenario 2)
E10 Ti=0 Ferroelastic
E20 Ti=0 Ferroelastic
Field along 45° between x,y
(→ E1=E20, E3=0 Ti=0)
Ferroelastic
E1E20, E3=0 Ti=0 Ferroelastic
T10, T2=0 (or T1T20) Ferroelastoelectric (Scenario 4)
Stress |2T| along 45° between x,y
(→ T1=T2=T6=T)
Ferroelectric (Scenario 3)
Stress along arbitrary angle bet-
ween x,y, but 45° (→ T1T2T60)
Ferroelectric
T3 0, Ek=0 → no symmetry
breaking
Ferroelastoelectric
mmmmF 24/4
zyxxymmmFm 24
xxzy Fmm 22
yyzx Fmm 22
yxxyzyxFmmm 2
1Fmz
zyxyzx Fmmm 222
zyxxyzyxxy mFmmmm 2
zz mF2/2
mmmmF 24/4
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 22 of 42Hempel, 2007 & 2016Page 22 of 42
Scenario 1: Ek=0, Ti=0, , ferroelastoelectric
- The behaviour in the Paraphase is
included when putting Q0 (4)
- Since the term-factor A in FEE is not
renormalized, the transition temperature
where Q appears, is unchanged Tc=To
mmmmF 24/4
2
2
21
2
23
654321
4
2
2
1
4
2
0
SQDEH
HXEWS
QDEH
HWDXS
SSSPPP
=
−
−=
−
−=
======
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 23 of 42Hempel, 2007 & 2016
Scenario 2: E30, Ti=0, , ferroelastic
- Higher order terms regarding Q
have been neglected
- The behaviour in the Paraphase is
included when putting Q0
- Since S6 is proportional to Q, the
transition can be classified to be
of pseudo-proper ferroelastic nature
(5)
• Through the terms dependent on E3 the term-factor A in FEE renormalizes and
the actual transition temperature Tc does not coincide with To.
Tc depends on the absolute value of E3.
Note: S1=S2 because of orientation of the mirror planes in Ferrophase (ref. to slide 8, case A).
zyxxymmmFm 24
3
2
3
36
5421
22
1
4
0
EGK
bQ
KP
EQGK
dS
SSPP
+=
−=
====
2
2
3222
2
21
2
3222
2
23
164
2
2
1
4
2
2
1
164
2
4
2
SEDEKHK
HNENQ
DEH
HXEWS
EDEKHK
HNDNQ
DEH
HWDXS
=
−
−+
−
−=
−
−+
−
−=
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 24 of 42Hempel, 2007 & 2016
Scenario 3: T1=T2=T6=Ti0, Ek=0, , ferroelectric
- Higher order terms regarding Q
have been neglected
- The behaviour in the Paraphase is
included when putting Q0
- P3 is proportional to Q, and the
transition can be classified to be
of pseudo-proper ferroelectric nature
(6)
• Through the terms dependent on Ti the term-factor A in FEE renormalizes and
the actual transition temperature Tc does not coincide with To.
Tc depends on applied mechanical stress Ti.
Note: S1=S2 because of orientation of the mirror planes in Ferrophase (ref. to slide 8, case A).
zyxxyzyxxy mFmmmm 2
i
i
QGK
dP
KG
Qd
GS
SSPP
−=
+=
====
4
82
1
0
3
2
22
6
5421
22
2
21
2
2
23
44
2
2
1
44
2
SEDH
EQ
EDH
HXEWS
EDH
HQ
EDH
HWDXS
i
i
=
−−
−
−=
−+
−
−=
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 25 of 42Hempel, 2007 & 2016
Scenario 4: T1=Ti0, Ek=0, , ferroelastoelectric
- Higher order terms regarding Q
have been neglected
- The behaviour in the Paraphase is
included when putting Q0
(7)
• Through the terms dependent on Ti the term-factor A in FEE renormalizes and
the actual transition temperature Tc does not coincide with To.
Tc depends on applied mechanical stress Ti.
zyxyzx Fmmm 222
0
0
654
321
===
===
SSS
PPP
i
i
i
TEDH
E
CQ
EDH
HXEWS
TEDH
E
CQ
EDH
HXEWS
TEDH
HQ
EDH
HWDXS
−+−
−
−=
−−+
−
−=
−+
−
−=
44
1
2
1
4
2
2
1
44
1
2
1
4
2
2
1
42
1
4
2
2
2
22
2
2
21
2
2
23
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 26 of 42Hempel, 2007 & 2016
• Slonczewski et al. /12/ derived a compact formalism to calculated the
temperature dependence of the elastic stiffness coefficients cij resulting from
the motion of the order parameter, and at constant external electric fields (Ek).
• A very similar formalism is valid for the description of dielectric impermea-
bilities kl at constant external mechanical stresses (Ti).
• The piezoelectric modules hmn are those measured at freely moving OP.
• In the following the isothermal behaviour of material coefficients will be derived,
taking into account that:
- the OP Q can freely move
- there might be additional contributions, originating from the piezo-
electric effect if allowed by symmetry
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 27 of 42Hempel, 2007 & 2016
vn
2
kvvk, km
2
n
2
m
2
nm
2
mn PS
FR
PS
F
QS
F
F
QS
F
SS
Fc
−
−
=
−12
kvk jk
vj RPP
F with
=
2
vn
kvvk kmnmnm
mn SP
FR
SP
F
QP
F
F
QP
F
PP
F
−
−
=
− 2
,
221222
kvk jk
vj RSS
Fwith
=
2
QS
F
F
QP
F
SP
Fh
nmnm
mn
−
=
− 21222
(8)
• As result for the 4 scenarios the temperature dependences of the material
coefficients are explicitly provided on the slides 30-34. The appearing constants
are the same as summarized on slide 19 (eq. (2)).
• The equilibrium state of ACC is obtained if the state parameters regarding the 4
scenarios (ref. to eqs. (4-7)) are introduced in FEE (eq. (1).
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 28 of 42Hempel, 2007 & 2016
• As result the FEE (eq. (1)) of ACC under defined external forces reads like:
Scenario 1: F=AQ2 + B’Q4 with A=Ao(T-To) (9)
Scenarios 2-4: F=A’Q2 + B’Q4 with A’=Ao(T-Tc) (10)
• From this the equilibrium OP-expressions can be calculated:
Scenario 1: Q2=-A/(2B’) (11)
Scenarios 2-4: Q2=-A’/(2B’) (12)
• As depicted in eqs. (8), for the temperature dependences of the material
coefficients also the 2nd derivatives are needed. From FEE (eq. (1)) we get:
Scenario 1: FQQ=2A+12BQ2+2W(S1+S2)+2XS3 (13)
Scenario 2: FQQ=2A+12BQ2+2VP32+2W(S1+S2)+2XS3+2bS6
2 (14)
Scenario 3: FQQ=2A+12BQ2+2VP32+2W(S1+S2)+2XS3+2bS6
2 (15)
Scenario 4: FQQ=2A+12BQ2+2W(S1+S2)+2XS3+2Z(S1-S2)2 (16)
• For the individual scenarios the explicit expressions for the state parameters
(ref. to eqs. (4-7)) have to be introduced into eqs. (13-16), as well as eqs.
(11,12). The paraphase is described if put Q=0.
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 29 of 42Hempel, 2007 & 2016
• Finally the OP-expressions (eqs. (11,12)), the state parameters (eqs. (4-7)) as
well as the 2nd derivatives- expressions (eqs. (13-16)) are the ones which need
to be introduced into the formulas in eq. (8) and in the resulting expressions on
slides 30-34 in order to describe the temperature dependences of the
coefficients.
• Furthermore following additional abbreviations are used throughout the paper
to describe the piezoelectric contributions (acc. to eq. (8)):
( ) ( )
( ) ( )
( ) ( ) 213121133
13322331132
13121133
121133
213121133
212
211
12332
132112
213121133
212
211
2131133
2211
22
2
2
oooo
o
oooo
oo
oooooo
ooo
oooooo
ooo
cccc
cRRRR
cccc
ccR
cccccc
cccRR
cccccc
cccRR
−+
−====
−+
+=
−−−
−==
−−−
−==
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 30 of 42Hempel, 2007 & 2016
Scenario 1: Ek=0, Ti=0, , ferroelastoelectric
F
XQWQHc
F
WQZQDCc
K
dQxSSSrbQGc
cc
J
eQYQFc
F
XQEc
cc
F
WQZQDCc
)2)(2(
)2(222
2
)(2)(222
2
)(22
)2(2
)2(222
13
22
12
2
321
2
66
4455
22
44
2
33
1122
22
11
−=
−−+−=
−++++=
=
−+=
−=
=
−++=
dQh
eQhh
G
dQVQOSSSNK
F
eQUQMSSSLJ
=
==
−++++=
=
=
−++++=
36
2514
22
32133
12
1122
22
32111
2
)(22)(22
0
2
)(22)(22
mmmmF 24/4
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 31 of 42Hempel, 2007 & 2016
Scenario 2: E30, Ti=0, , ferroelastic
K
NOP
F
XQWQHc
K
NP
F
WQZQDCc
K
dQ
F
dPbQS
xSSSrbQGc
cc
J
hPeQYQFc
K
OP
F
XQEc
cc
K
NP
F
WQZQDCc
2
4)2)(2(
2
)2()2(222
2
)()4(
2)(222
2
)()(22
2
)2()2(2
2
)2()2(222
23
13
23
22
12
2236
3212
66
4455
23
22
44
23
2
33
1122
23
22
11
−−=
−−−+−=
−+
−
++++=
=
+−+=
−−=
=
−−++=
F
dSVQPdPbQSdQh
hh
F
dSVQPWQNPh
F
dSVQPXQOPh
hPhh
eQhh
G
dQROPRNOPRRNP
F
dSVQPVQOSSSNK
F
ehQPtQPRS
F
hPeQUQMSSSLJ
)4)(4(
)4(22
)4(22
2
)()2(16)()2(2
)4(22)(22
2
2
2
)()(22)(22
633636
3132
63331
63333
32415
2514
2
33
2
313
2
31211
2
3
2
632
32133
33612
1122
2
3
22
32111
++−=
=
+−=
+−=
==
==
++++−
+−++++=
−+=
=
+−++++=
zyxxymmmFm 24
Circled term induces
temperature dependence
already in Paraphase
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 32 of 42Hempel, 2007 & 2016
Scenario 3: T1=T2=T60, Ek=0, , ferroelectriczyxxyzyxxy mFmmmm 2
K
NOP
F
XQWQHc
K
NP
F
WQZQDCc
K
dQ
F
dPbQS
xSSSrbQGc
cc
J
hPeQYQFc
K
OP
F
XQEc
cc
K
NP
F
WQZQDCc
2
4)2)(2(
2
)2()2(222
2
)()4(
2)(222
2
)()(22
2
)2()2(2
2
)2()2(222
23
13
23
22
12
2236
3212
66
4455
23
22
44
23
2
33
1122
23
22
11
−−=
−−−+−=
−+
−
++++=
=
+−+=
−−=
=
−−++=
F
dSVQPdPbQSdQh
hh
F
dSVQPWQNPh
F
dSVQPXQOPh
hPhh
eQhh
G
dQROPRNOPRRNP
F
dSVQPVQOSSSNK
F
ehQPtQPRS
F
hPeQUQMSSSLJ
)4)(4(
)4(22
)4(22
2
)()2(16)()2(2
)4(22)(22
2
2
2
)()(22)(22
633636
3132
63331
63333
32415
2514
2
33
2
313
2
31211
2
3
2
632
32133
33612
1122
2
3
22
32111
++−=
=
+−=
+−=
==
==
++++−
+−++++=
−+=
=
+−++++=
Circled term induces
temperature dependence
already in Paraphase
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 33 of 42Hempel, 2007 & 2016
Scenario 4: T10, Ek=0, , ferroelastoelectric
• In order to properly describe this scenario – in particular to prove that c11c22,
c44c55 and h14h25 - the following additional (higher order) terms*) have to be
introduced into the FEE (ref. to eq. (1)):
m(S1-S2)2(S1+S2)
2 and n(S42-S5
2)(S1-S2) and p(P1S4-P2S5) (S1-S2)Q
QQQQ
F
SSZQWQSSZQWQZQSmSDCc
J
eQYQSSnFc
J
eQYQSSnFc
K
dQxSSSrbQGc
F
XQSSZQWQHc
F
XQEc
F
SSZQWQZQSSmDCc
F
SSZQWQZQSSmDCc
))(42))((42(2822
2
)(2)(22
2
)(2)(22
2
)(2)(222
)2))((42()2(2
))(42(2)412(22
))(42(2)412(22
21212
2112
22
2155
22
2144
2
321
2
66
2113
2
33
2
2122
1
2
222
2
2122
2
2
111
−−−+−−−+−=
−+−−=−+−+=
−++++=
−+−=−=
−−−+−++=
−+−+−++=
zyxyzx Fmmm 222
*) these terms are not needed
to describe scenarios 1-3
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 34 of 42Hempel, 2007 & 2016
Scenario 4: T10, Ek=0, , ferroelastoelectric, continued
dQh
QSSpeQh
QSSpeQh
G
dQVQOSSSNK
F
QSSpeUQSSgMSSSLJ
F
QSSpeUQSSgMSSSLJ
=
−−=
−+=
−++++=
=
−−−+−−+++=
−+−+−++++=
36
2125
2114
22
32133
12
22
212
2132122
22
212
2132111
)(
)(
2
)(22)(22
0
2
))((2)(22)(22
2
))((2)(22)(22
zyxyzx Fmmm 222
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 35 of 42Hempel, 2007 & 2016
S1=S2, S3
h14=h25, h36
(morphic)
c11=c22, c12
TcTo
S1=S2, S3, P3, h15=h24
c44=c55, 11=22
S6, h14=h25, h36, 12
(morphic)
c11=c22, c33, c12, c13, 33
h31=h32, h33
Tc=To
c33, c13
c44=c55, c66, 11=22, 33
Scenario 1: , ferroelastoelectricmmmmF 24/4 Scenario 2: , ferroelasticzyxxymmmFm 24
Qualitatively, the temperature dependences can now be depicted, taking not into
account the ordinary pyroelectric and thermal expansion effect, as well as the
multidomain state in the ferroic phases:
c66
11, 22, (12)
12, (11, 22)
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 36 of 42Hempel, 2007 & 2016
S1=S2, S3, S6
c44=c55, 11=22, 12
P3, h14=h25, h15=h24
(morphic)
c11=c22, c33, c66, c13, c12
33
h33, h31=h32, h36
(morphic)
TcTo
S1S2, S3
c44c55, c66, 1122, 33
h14 h25, h36
(morphic)
c11 c22, c12
Tc≠To
c33, c13
Scenario 3: , ferroelectriczyxxyzyxxy mFmmmm 2Scenario 4: , ferroelastoelectriczyxyzx Fmmm 222
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 37 of 42Hempel, 2007 & 2016
5. Comparison with Experimental Findings
Up to now no investigations of ACC are known where a constant electric field or
mechanical stress was applied when passing through the Phase Transition Point.
The experimental data available so far /13,14/ are all measured according to the
Scenario 1 (see page 35), under not defined domain conditions in the Ferrophase.
iiii /1=
Predicted behaviorThe temperature dependence of the dielectric
impermeabilities (which are within Scenario 1
exactly the inverse of the permittivities) is predicted
to show no jump at the phase transition point, but a
different slope in the Ferroic Phase. This coincides
well with the data measured.
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 38 of 42Hempel, 2007 & 2016
Despite the constant (and comparable) slope in
the Para- and -Ferroic Phase, c33 is well
described by the predicted behavior (although
the jump is somehow smeared).
c11 = c22 is less in coincidence with the
predictions but similar. The jump (if at all) is
even more smeared and the slope in the
phases are different - as it is predicted.
Predicted behavior
c11=c22, c12
c33, c13
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 39 of 42Hempel, 2007 & 2016
Predicted behavior
c44=c55, c66
h14=h25, h36
c44 = c55 and c66 show now jump but just a weak
change of slope at the transition point. This is in
line with the predictions, and can be explained
with the smallness of respective coefficients in the
FEE influencing the stiffnesses.
h36 is in agreement with the predicted behavior. It
reflects the temperature dependence of the order
parameter. As expected, above the transition point
h36 is identical to zero (‘morphic coefficient’).
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 40 of 42Hempel, 2007 & 2016
5. Summary
The transition of (NH4)2CuCl4•2H2O at 200 K is a candidate of a classical
ferroelastoelectric phase transition.
Moreover, if certain constant external field (e.g. electric field, mechanical
stress) is applied to the crystal throughout the entire temperature range, the
type of Ferroicity should change. It has been shown that the transition can then
be either ferroelastic or ferroelectric, accompanied by all typical dependences
(incl. domain structures).
Especially interesting is the predicted temperature dependence of c66 in the
Paraphase, if an electric field E3 is applied (scenario 2). But without that field
(e.g. “switched off” at a defined temperature) c66 should return to the
temperature-independent value as described by scenario 1.
A similar behaviour is predicted for 33 within scenario 3, where a constant
mechanical stress is applied, which could also be “switched off”.
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 41 of 42Hempel, 2007 & 2016
6. Literature
/1/ Wadhawan, V.K., Phase Transitions, 3, 3 (1982)
/2/ Toledano, P., et al., Phys. Rev. B16, 386 (1977)
/3/ Suga, H., et al., Bull. Chem. Soc. Japan, 38, 1007 (1965)
/4/ Bansal, M.L., et al., J. Phys. Chem. Solids (GB), 40, 109 (1979)
/5/ Hempel, H., et al., Ferroelectrics, 104, 361 (1990)
/6/ Janovec, V., et al., Czech. J. Phys., B25, 1362 (1975)
/7/ Kovalev, O.V., “Irreducible and Induced Representations and Corepresentations of
Space Groups”, Moscow, Nauka, 1986
/8/ Slaboszewska, M., et al., Ferroelectrics, 302, 55 (2004)
/9/ Aizu, K., J. Phys. Soc. of Japan, 27, 387 (1969)
/10/Sirotin, Yu.I., Shaskolskaya, M.P., “Fundamentals of Crystal Physics”,
Mir Publishers, Moscow, 1982
/11/Kocinsky, J., “Theory of Symmetry Changes at Continuous Phase Transitions”,
Elsevier, Amsterdam-Oxford-New York, 1983
/12/Slonczewski, J.C., et al., Phys. Rev. B1, 3599 (1970)
Phase Transition & Landau Theory of (NH4)2CuCl4 • 2H20
Page 42 of 42Hempel, 2007 & 2016
/13/Slaboszewska M., Tylczynski Z., Pietraszko A., Z., Karaev A., Ferroelectrics 302,
55-58(2004)
/14/Tylczynski Z., Wiesner M., Materials Chemistry and Physics 149-150, 566-573(2015)
Hempel, 2008-2015 Page 1 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase Sequence and Landau Theory of LiKSO4
Table of Contents
1. Introduction
2. Basics
3. Symmetry Considerations (Phase II omitted)
4. Landau Modelling
5. Phase Diagram and Stability
6. Temperature Dependences of Material
Coefficients
7. Discussion of Results and Comparison
with Experimental Findings
8. Conclusion
9. Literature
Hempel, 2008-2015 Page 2 of 57
Phase Sequence and Landau Theory of LiKSO4
1. Introduction
LiKSO4 (abbreviated LPS) exhibits a series of Phase Transitions (PTs), if temperature
(or/and pressure) is changed.
Numerous investigations have been made so far, but there is still no consistent view how
this crystal behaves.
The most recent summary regarding the phase sequence is as follows:
P63/mmc
or
P63mcP63P31cCc?????
Pmcn
or
Pc21n
Temperature (K)
IIIIIVVVIVIIVIIIIXX II
no. of formulae units in primitive elementary cell Z = 2 2 2 4 2
708205
(260 heating)
190
(195 heating)
165135806030 943
Fig. 1 Phase Transition Sequence acc. to /1/, /2/, /3/ (and references therein)
Hempel, 2008-2015 Page 3 of 57
Phase Sequence and Landau Theory of LiKSO4
The essential findings can be summarized:
• Phase I shall exhibit Point Group (PG) symmetry 6/mmm or 6mm (see /2/, /3/)
• Symmetry of phase II seems to be Pmcn (centro-symmetric PG: mmm) or Pc21n
(polar PG: mm2) (see /2/, /3/)
• No ferroelectricity nor ferroelastoelectricity was experimentally found in the entire
temperature interval (see e.g. /4/)
• Ferroelasticity was observed in phases II and V (see e.g. /4/)
• All PTs observed are of 1st Order, PT III→IV shall be reconstructive /1/
• Phases IV and V are not „pure“ but „mixed“ phases, depending on the crystal‘s
„individuality“, i.e. depending on growth- and thermal treatment conditions (see /5/)
• LPS exhibits certain twinning in phases III and IV (see /5/, /6/)
• Depending on the a.m. „individuality“, and actual twinning composition,
measurements‘ results are often less reproducible (or comparable), and the PT
between phases IV and V is sluggish (see e.g. /5/)
• The few theoretical models which have been established to describe the ferro-
elastic IV–V PT are not comprehensive enough to describe LPS‘ behaviour in this
temperature region satisfactorily (see /7-13/). The most advanced model was
proposed by Quirion et al. /14/
• Also, no attempt has been undertaken yet to model the (unusual)
series of PTs III–IV-V
Hempel, 2008-2015 Page 4 of 57
Phase Sequence and Landau Theory of LiKSO4
The aim of the present paper is to undertake an extended approach to model the phase
sequence I-III–IV-V, and to compare it with existing data and temperature dependences.
2. Basics
The following basic assumptions have been made:
1. The Landau Theory will be consequently applied
2. Phase I is assumed to exhibit 6mm symmetry (because of the 1st and 2nd finding
above, and because of the fact that PG 6mm is the common supergroup of the
phases III, IV, and V (see e.g. /15/). Therefore 6mm (to be precise - Space Group
P63mc) will be considered to represent the Prototypic Phase, as it was also
assumed by Quirion et al. /14/ and Willis et al. /11/.
In this context, there is no need to argue about reconstructive transition(s) as
suggested in /1/, despite the fact that the atomic displacements accompanied with
the III → IV PT shall be significant (see e.g. /5/).
3. Phase II will not be considered (but „skipped“) – in order to reduce by far the
complexity of the calculations.
Hempel, 2008-2015 Page 5 of 57
Phase Sequence and Landau Theory of LiKSO4
4. All phases of the targeted sequence I–III–IV–V do have the same translational
symmetry, thus the PTs are all of the „ferrodistortive“ type where the translational
symmetry is preserved, and the modelling can be done within the framework of
PGs.
5. No „mixture“ of phases will be taken into account, only „pure“ phases.
6. No kind of twinning will be taken into account for the Landau Modelling – with other
words – the results of the following theoretical considerations are valid for the case
that each PT leads to a defined single orientation state of the crystal (ref. to
following slide).
Note:
Modelling of the mentioned phase sequence under the assumption that the Prototypic Phase
exhibits 6/mmm instead of 6mm symmetry should be possible as well, but then the concept of
PTs, induced by „reducible representations“, needs to be applied (see e.g. /13/, /17/).
The author believes that in this case the complexity of the calculations is not reasonably
manageable, and the expected additional information is assumed to be marginal and will
primarily be related to the (theoretical) possibility of polarization reversal along the 6-fold axis
(i.e. ferroelectricity) within certain phases (which was not observed experimentally!).
Hempel, 2008-2015 Page 6 of 57
Phase Sequence and Landau Theory of LiKSO4
3. Symmetry Considerations (Phase II omitted)
Throughout all the phases one fixed rectangular system of coordinates x(1), y(2), z(3) is
used (mind the orientation of the symmetry elements in the stereographs below)
I (6mm) III (6) IV (3m) V (m)
Symmetry elements gi:
(see e.g. /16/)
(PG-Generators in red)
Order G of group: 12 6 6 2
Number of orientation
states n=12/G6,3m,m 1 2 2 6
Kind of expected none ferroelastoelectri- ferroelastoelectri- ferroelasticity,
ferroicity with regard (prototypic phase) city (and ferrogyro city and ferrobi- ferroelectricity
to the prototypic phase tropy*)) elasticity
see e.g. /17/, /18/):
*) Ferrogyrotropy represents an implicit form of ferroicity (see /19/)
y(2)
x(1)
z(3)
+ +
1, C61, C6
2, C63, C6
4, C65,
m[010], m[-110], m[100],
m[210], m[110], m[120]
1, C61, C6
2, C63, C6
4, C65 1, C6
2, C64 , m[210], m[-110],
m[120](=my)1, m[120](=my)
Hempel, 2008-2015 Page 7 of 57
Phase Sequence and Landau Theory of LiKSO4
III (6) IV (3m) V (m)
Number of orientation
states n=12/G6,3m,m 2 2 6
Lost symmetry elements
of PG 6mm - being the
twin operations between
the orientation states
Tensor components in Piezoelectric Piezoelectric Strain Tensor Components
which the orientation coefficients coefficients (see next slide)
states differ
Gyration Elastic coefficients Polarization Vector Components
(see next slide)
kind of expected ferroelastoelectricity ferroelastoelectricity ferroelasticity and
ferroicity: (and ferrogyrotropy) and ferrobielasticity ferroelectricity
)1()2(
ijij GG −=
)1(
14
)2(
14 hh −=)1(
11
)2(
11 hh −=
)1(
15
)2(
15 cc −=
m[010], m[-110], m[100],
m[210], m[110], m[120]
C61, C6
3, C65,
m[010], m[100], m[110]
C61, C6
2, C63, C6
4, C65,
m[010], m[-110], m[100], m[210], m[110]
(upper index defines the
orientation state)
Hempel, 2008-2015 Page 8 of 57
Phase Sequence and Landau Theory of LiKSO4
Application of the lost symmetry elements yield following 6 orientation states for phase V.
Depicted are the spontaneous values only, calculated e.g. after Aizu /20/:
(1) (2) (3) (4) (5) (6)
a) Polarisation
b) Deformation
Notes: 1. In the prototypic phase I (6mm) all spontaneous components (a, c, P1) are zero.
2. The abbreviations a=½(S11-S22), c=S13 are used.
3. Orientational states (1), (3) and (5), and the related spontaneous strains and polarizations
represent the a „stand alone“ PT 3m → m.
0
0
1P
−
02
32
1
1
1
P
P
−
−
02
32
1
1
1
P
P
−
02
32
1
1
1
P
P
−
0
0
1P
02
32
1
1
1
P
P
−
00
00
0
c
a
ca
−
−−
−−
02
3
2
2
3
22
322
3
2
cc
ca
a
ca
a
−−
−
−−
02
3
2
2
3
22
322
3
2
cc
ca
a
ca
a
−
−
−−
−
02
3
2
2
3
22
322
3
2
cc
ca
a
ca
a
−
−
−
00
00
0
c
a
ca
−
02
3
2
2
3
22
322
3
2
cc
ca
a
ca
a
Hempel, 2008-2015 Page 9 of 57
Phase Sequence and Landau Theory of LiKSO4
Sapriel /21/ calculated the permissible domain wall orientations for all ferroelastic species.
For phase V (PG m) it turns out that there are 19 different walls allowed if we refer to the
prototypic phase I (PG 6mm).
(a and c represent the spontaneous tensor components as outlined on the previous slide,
x, y, z the coordinates)
z=0 x=0 y=0
x=3 y x=-3 y x=y/3
y=cz/a3 y=-cz/a3 x=-y/3
x=cz/a x=-cz/a (3x+3 y)+2cz/a=0
(3x+3 y)-2cz/a=0 (3x-3 y)+2cz/a=0 (3x-3 y)-2cz/a=0
(x+3 y)+2cz/a=0 (x+3 y)-2cz/a=0 (x-3 y)+2cz/a=0
(x-3 y)-2cz/a=0
Note:
The expressions marked in purple describe the permissible domain walls regarding a “stand
alone ferroelastic PT” from a phase with symmetry 3m to m only. This particular PT constitutes a
subset of the complete series of transitions I-III-IV-V. In the limit a>>c they coincide with the
green equations which were found experimentally (see e.g. /4/).
Hempel, 2008-2015 Page 10 of 57
Phase Sequence and Landau Theory of LiKSO4
4. Landau Modelling
The Landau Theory can describe the PTs that shall be permitted form a high symmetry
group. The maximal subgroups in which the respective Irreducible Representations (IRs)
exhibit identity representations, represent the allowed low symmetry groups. The low
symmetry phases are characterized by the presence of certain Order Parameters (OPs) /17/.
The Landau Theory imposes to look up the „active“ or „acceptable“ IRs only, which can lead
to a PT of second order between strictly periodic crystalline structures /22/. This means:
(a) The anti-symmetrized square, noted {}2, of n(k*) must not have any IR in common with
the vector representation of Go (k* specifies the relevant star of k-vectors) – Lifshitz
Condition (absence of asymmetric gradient terms of OP).
(b) The symmetrized third power of n(k*), noted []3, must not contain the totally
symmetric IR of Go – Landau Condition (absence of cubical OP terms).
Nevertheless is has been proven that many 1st order PTs can be successfully described
within the Landau Theory (see /22/). If the Landau Condition is violated, cubical OP-term(s)
are allowed and lead necessarily to a PT of 1st order. If the Landau Condition is obeyed and
a 1st order PT shall be described, as it is the case of LPS (will be shown later), the Free
Energy Expression (FEE) has to be expanded up to the 6th power of the order parameter(s).
To get started the Irreducible Representations of PG 6mm need to be inspected.
Hempel, 2008-2015 Page 11 of 57
Phase Sequence and Landau Theory of LiKSO4
Irreducible Representations of PG 6mm acc. to /23/, /24/:
Note: The IRs 5 and 6 are originally complex ones. They have been transformed into real, physical
IRs by a suitable unitary transformation (e.g. ref. to /25/ and /26/):
IRreal= U+ IR U with U= 1/2 and U+=1/2
IR
acc. to
/21/
IR
acc. to
/15/
1 C61 C6
2 C63 C6
4 C65 m[010] m[-110] m[100] m[210] m[110] m[120]
1 A1 1 1 1 1 1 1 1 1 1 1 1 1
2 A2 1 1 1 1 1 1 -1 -1 -1 -1 -1 -1
3 B1 1 -1 1 -1 1 -1 1 -1 1 -1 1 -1
4 B2 1 -1 1 -1 1 -1 -1 1 -1 1 -1 1
5 E1
6 E2
−2
1
2
32
3
2
1
−2
1
2
32
3
2
1
−−
−
2
1
2
32
3
2
1
−
10
01
−
−−
2
1
2
32
3
2
1
−
2
1
2
32
3
2
1
−10
01
−−
−
2
1
2
32
3
2
1
−
−
10
01
−
−−
2
1
2
32
3
2
1
10
01
−
−−
2
1
2
32
3
2
1
−
2
1
2
32
3
2
1
−10
01
−
−−
2
1
2
32
3
2
1
−−
−
2
1
2
32
3
2
1
−10
01
−−
−
2
1
2
32
3
2
1
10
01
−
−−
2
1
2
32
3
2
1
10
01
−
−−
2
1
2
32
3
2
1
−
i
i
0
1
− ii
11
−
2
1
2
32
3
2
1
−
2
1
2
32
3
2
1
Hempel, 2008-2015 Page 12 of 57
Phase Sequence and Landau Theory of LiKSO4
• Since 1-4 are one-dimensional they give rise to one-dimensional Basis Functions (BF)
(one-dimensional OPs), whereas 5 and 6 are interrelated to two-dimensional BFs
(two-dimensional OPs).
• The identification of the PG-symmetries induced by the individual IRs can be easily
derived by checking under which symmetry elements of PG 6mm the BFs are invariant.
This can be conveniently done in the space where the OPs move. Those elements
constitute the PGs of the respective low symmetry phases.
• Corresponding investigations lead inter alia to:
• By checking the respective reduction coefficients (see e.g. /26/) it can be shown that the
IRs 2, 4 and 5 fulfil both, the Landau- and the Lifshitz- conditions. Therefore, neither
gradient- nor cubical OP terms are allowed to enter the FEE.
IR Basis Function (BF),
written as OP-vector
BF are invariant under action of symmetry
elements gi of PG 6mm
Induced low temperature
symmetry (PG)
2 Q6 6
4 Q3m 3my
5 my
1, C61, C6
2, C63, C6
4, C65
1, C62, C6
4 ,
m[210], m[-110], m[120] (=my)
=
0
0
2
1
m
m
Q
Q1, m[120] (=my)
Hempel, 2008-2015 Page 13 of 57
Phase Sequence and Landau Theory of LiKSO4
• Next step towards the FEE is to decompose the (ordinary) strain tensor and (ordinary)
polarisation vector into their irreducible parts regarding PG 6mm.
• Utilisation of the projection operator (see e.g. /26/)
with = P resp. S and
– operator related to symmetry
element gk
yield the following results:
Strain S: ½(S11+S22), S33, ½(S11-S22), S12, S13, S23 respectively
(in Voigt’s Notation /16/ and normalized)
Polarisation P: P1, P2, P3
=
3
2
1
P
P
P
P
=
332313
232212
131211
SSS
SSS
SSS
S
)()()()( xfgTgG
dxf
Gg
kijkij
k
=)(xf
65421
321 ,,,
2
)(,,
2
)(SSS
SSS
SS −+
)( kgT
Hempel, 2008-2015 Page 14 of 57
Phase Sequence and Landau Theory of LiKSO4
• Each term of the FEE has to be invariant under the action of all the symmetry elements
of PG 6mm
• It is sufficient to prove this invariance under the action of the “generators” of PG 6mm,
namely C61 and m[100]
• To this end the transformation properties of Si, Pk, Q6, Q3m, Qm1,2 have been
compiled:
irreducible part after action of C61 after action of m[100]
Q6 Q6 -Q6
Q3m -Q3m -Q3m
2
1
m
m
Q
Q
+−
+
21
21
2
1
2
32
3
2
1
mm
mm
− 2
1
m
m
Q
Q
−
+
6
5
4
21
3
21
)(2
1
)(2
1
S
S
S
SS
S
SS
3
2
1
P
P
P
+−
+
3
21
21
22
32
3
2
1
P
PP
PP
−
3
2
1
P
P
P
+−−
+
−
−
−−
+
621
45
54
621
3
21
)(32
1
)3(2
1
)3(2
1
2
3
2
)(
2
1
)(2
1
SSS
SS
SS
SSS
S
SS
−
−
−
+
6
5
4
21
3
21
)(2
1
)(2
1
S
S
S
SS
S
SS
Hempel, 2008-2015 Page 15 of 57
Phase Sequence and Landau Theory of LiKSO4
• The FEE, reflecting the full dynamics of the system, can now be written as:
F = FOP+FOP-OP-OP+FS+FP+FS-P+FS-P-OP+FOP-P+FOP-S with
FOP = A1Q62 + A2Q6
4 + A3Q66 + B1Q3m
2 + B2Q3m4 + B3Q3m
6 + C1(Qm12+Qm2
2) +
C2(Qm12+Qm2
2)2 + C3(Qm13-3Qm2
2Qm1)2 + C4(Qm1
2+Qm22)3
FOP-OP-OP = 1Q62Q3m
2 + 2Q62(Qm1
2 + Qm22) + 3Q3m
2(Qm12+Qm2
2)
FS = D[(S1-S2)2+S6
2] + E(S1+S2)2 + FS3
2 + G(S1+S2)S3 + H(S42+S5
2)
FP = I(P12+P2
2) + JP32
FS-P = K(S1+S2)P3 + LS3P3 + M(P1S5+P2S4)
FS-P-OP = N(S1+S2)P3Q62 + OS3P3Q6
2 + P(P1S5+P2S4)Q62 + Q(S1+S2)P3Q3m
2 + RS3P3Q3m2 +
S(P1S5+P2S4)Q3m2 + U(S1+S2)P3(Qm1
2+Qm22) + VS3P3(Qm1
2+Qm22) +
W(P1S5+P2S4)(Qm12+Qm2
2) + xQ6(P1S4-P2S5) + yQ3m[(S1-S2)P1-S6P2]
FOP-P = XQ62P3 + YQ3m
2P3 + Z(Qm12+Qm2
2)P3 + b(P12+P2
2)Q62 + d(P1
2+P22)Q3m
2 +
e(P12+P2
2)(Qm12+Qm2
2) + f(P1Qm2-P2Qm1)
FOP-S = hQ62[(S1-S2)
2+S62] + iQ6
2S3 + jQ62(S1+S2) + kQ6
2(S42+S5
2) + lQ3m2[(S1-S2)
2+S62] +
mQ3m2S3 + nQ3m
2(S1+S2) + oQ3m2(S4
2+S52) + p(Qm1
2+Qm22)[(S1-S2)
2+S62] +
q(Qm12+Qm2
2)S3 + r(Qm12+Qm2
2)(S1+S2) + s(Qm12+Qm2
2)(S42+S5
2) + t(Qm2S5-Qm1S4) +
u[(S1-S2)(Qm22-Qm1
2)-2S6Qm1Qm2)] + v[(S1-S2)S5-S4S6]Q3m + w[(S1-S2)S5-S4S6]Qm1Qm2
Note:
The OP-terms have been provided up to the power of six to reflect that all PTs are of 1st order
(1)
Hempel, 2008-2015 Page 16 of 57
Phase Sequence and Landau Theory of LiKSO4
whereby all term-factors, as usual in Landau’s Theory, are assumed to be constant,
except A1=Ao(T-T6), B1=Bo(T-T3m), C1=Co(T-Tm1m2).
• The transformation matrix within the “Voigt’s Space” that mapped
into looks like
and can be equally used to calculate e.g. the irreducible (strain) parts of the elastic
coefficients, resp. of the piezoelectric coefficients. The results are summarized below.
• The following equalities regarding eq. (1) hold:
( ) ( )
( )
ooo
o
o
o
o
ooooo
ooooo
hMhLhKJI
cHcGcFccE
CrystalshexagonalincccnoteccD
153331
33
33
11
11
4413331211
1211661211
,,,1
,1
2
1,,
2
1,
4
1
holds2
1:
4
1
=======
===+=
−=−=
−
+
6
5
4
21
21
2
)(32
)(
S
S
S
SSS
SS
6
5
4
3
2
1
S
S
S
S
S
S
−
100000
010000
001000
00002
1
2
1000100
00002
1
2
1
Hempel, 2008-2015 Page 17 of 57
Phase Sequence and Landau Theory of LiKSO4
• The elastic stiffnesses coij in the FEE (eq. (1)) are those defined at constant polarisation
and at constant OPs.
• The dielectric impermeabilities oij in eq. (1) are those at constant strains and at
constant OPs.
• The piezoelectric coefficients hokm in eq. (1) are those at constant OPs.
• Moreover, it must be kept in mind that the ordinary thermal expansion / contraction
(i.e. the linear temperature dependence of the symmetry allowed strains as well as the
ordinary pyroelectric effect (linear temperature dependence of the polarisation) is not
covered by the present model. This must be taken into consideration when comparing
the theoretical results with experimental data.
• Next step is the solving of the following system of 9 state equations for the „free“ crystal:
(2)
0
0
==
==
k
k
i
i
EP
F
S
Fi=1…6
k=1…3
Hempel, 2008-2015 Page 18 of 57
Phase Sequence and Landau Theory of LiKSO4
• At no „external forces“ Ti (mechanical stresses) nor Ek (electric fields) the OP-
dependences of Si and Pk are obtained.
• Neglecting higher orders of the OPs we get:
(3)
(4)
(5)
(6)
(7)
(8)
)]))(2((
))2(())2([(1
)(
2
)(
2
1
)(
2
)(
2
1
22
21463
23462
26461
543
4
22
213
232
261
21
22
2
4
22
213
232
261
21
22
1
mm
m
mmmmm
mmmmm
QQJqLZKKK
QJmLYKKKQJiLXKKKKK
S
K
QQKQKQK
D
QQuS
K
QQKQKQK
D
QQuS
+−+−
+−+−+−+−
++++
−+
++++
−−
)(
4
2
4
2
2
2
2
1
216
225
124
mm
mm
m
m
QQpD
QuQS
QMHI
tIfMS
QMHI
fMtIS
++
−
−
−
−
Hempel, 2008-2015 Page 19 of 57
Phase Sequence and Landau Theory of LiKSO4
)]))(2()((
))2()((
))2()([(2
1
4
2
4
2
2
2
2
1463435
2
3462425
2
6461415
54
3
122
221
mm
m
m
m
QQJqLZKKKZKKKK
QJmLYKKKYKKKK
QJiLXKKKXKKKKKJK
P
QMHI
MtfHP
QMHI
fHMtP
+−+−+
+−+−+
+−+−+−
−
−
−
−
• The following abbreviations have been used:
Note:
Inserting the equilibrium values of the OPs for the individual phases (ref. to eqs. (16-19)) into
eqs. (3-11) give the respective dependences – ultimately on the temperature.
It should be noted that S5, (S4), P1 and (P2) show „pseudoproper-ferroelastic“- resp.
„pseudoproper-ferroelectric“ behaviour, because they depend linearly on the OPs.
(9)
(10)
(11)
(12)
KLGJKLJFK
KLJFKEJKJrKZKLZJqKK
JnKYKLYJmKKJjKXKLXJiKK
−=−=
−−−=−+−=
−+−=−+−=
2 4
)4)(4( )2()2(
)2()2( )2()2(
6
2
5
2
6
22
4563
562561
Hempel, 2008-2015 Page 20 of 57
Phase Sequence and Landau Theory of LiKSO4
• Inserting the expressions for Si (Q6, Q3m, Qm1, Qm2) and Pk (Q6, Q3m, Qm1, Qm2)
into the FEE (eq. (1)) gives the Free Energy Expression F‘ of the mechanically and
electrically “free” crystal just as a function of the OPs:
F‘= A1Q62 + A‘2Q6
4 + A‘3Q66 + B1Q3m
2 + B‘2Q3m4 + B‘3Q3m
6 + C‘1(Qm12+Qm2
2) +
C‘2(Qm12+Qm2
2)2 + C‘3(Qm13-3Qm2
2Qm1)2 + C‘4(Qm1
2+Qm22)3 +
‘1Q62Q3m
2 + ‘2Q62(Qm1
2 + Qm22) + ‘3Q3m
2(Qm12+Qm2
2)
• The renormalized term-factors are now: A‘2, A‘3, B‘2, B‘3, C‘1, C‘2, C‘3, C‘4,
‘1, ‘2, ‘3, but A1 and B1 are unchanged.
• This yields inter alia the following expressions for the transition temperatures
of the „free“ crystal:
OP: Q6 T6 - unchanged with regard to eq. (1)
OP: Q3m T3m - unchanged with regard to eq. (1)
OP: Qm1, Qm2 T‘m1m2 - renormalized as
(13)
(14)
22
222
2
22
21
'
21)4(
2)4( MHIC
tfMHfItM
MHIC
HfItTT
oo
mmmm−
−+−
−
++=
Hempel, 2008-2015 Page 21 of 57
Phase Sequence and Landau Theory of LiKSO4
• Minimizing of F’ with regard to the OPs according to the system of equations
yield the specific equilibrium values of the OPs:
Phase I (PG 6mm):
Phase III (PG 6):
Phase IV (PG 3m):
0'
0'
0'
0'
2136
=
=
=
=
mmm Q
F
Q
F
Q
F
Q
F
(16)
(17)
(18)03
133
0 212'2
'31
'3
'2
'3
'22
36 ==−−−== mmm QQ B
BB
B
B
B
BQ Q
000 2136 ==== mmm QQ Q Q
003
133
2132'2
'31
'3
'2
'3
'22
6 ===−−−= mmm QQ Q A
AA
A
A
A
AQ
(15)
Hempel, 2008-2015 Page 22 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase V (PG m):
Notes:
The coefficients A‘3, B‘3, C‘4 have to be positive and A‘2, B‘2, C‘2 are assumed to be negative.
Other Solutions where the different OPs do exist simultaneously would imply to have PTs related
to so-called “reducible representations” which are not considered here because their induced low
symmetry phases are beyond the scope of this work.
• Next task is to calculate the 2nd derivatives of FEE (ref. to eq. (1)) which are needed
later (ref. to chapter 6., eqs. (29-31)) to derive the temperature dependences of the
material coefficients
• After creation of the derivatives, the OP-dependence of the strain- and polarization
components have to be used (eqs. (3)-(11)) and then the respective expressions for
the equilibrium OPs (ref. to eqs. (16-19) above) have to be applied as well
(19) C
CC
C
C
C
CQ Q Q Q mmm 2'
2
'41
'4
'2
'4
'22
2136
31
33000 −−−====
Hempel, 2008-2015 Page 23 of 57
Phase Sequence and Landau Theory of LiKSO4
• The 2nd derivatives of FFE (eq. (1)) for the individual phases calculate like:
Note:
Phase V (PG m) is characterized by equilibrium values of Qm2≠0 and Qm10 (see e.g. eq. (19)).
The expressions for the 2nd derivatives are provided up to the OP power of three. Higher order
terms have been omitted. K10-K22 are constants.
(20)
for
Para-
phasePhase PG 6 Phase PG 3m Phase PG m
Q6Q6 2A1 2A1+(12A2+K10)Q62 2A1+(2ɣ1+K14)Q3m
2 2A1+(2ɣ2+K18)Qm22
Q3mQ3m 2B1 2B1+(2ɣ1+K11)Q62 2B1+(12B2+K15)Q3m
2 2B1+(2ɣ3+K19)Qm22
Qm1Qm1 2C1 2C1+(2ɣ2+K12)Q62 2C1+(2ɣ3+K16)Q3m
2 2C1+(4C2+K20)Qm22
Qm2Qm2 2C1 2C1+(2ɣ2+K13)Q62 2C1+(2ɣ3+K17)Q3m
2 2C1+(12C2+K21)Qm22
Qm1Qm2 0 0 0 K22Qm23
ki QQ
F
2
Hempel, 2008-2015 Page 24 of 57
Phase Sequence and Landau Theory of LiKSO4
5. Phase Diagram and Stability
• The FEE - F’ (eq. (13)) is principally suited to investigate the possible equilibrium
states of the LPS-phase sequence, but the calculations are too complex to handle
• Particularly we are here only interested in the appearance of the phases III, IV and V
• Since ID 5 (refer to slide 12) gives rise to a two-component OP (Qm1, Qm2), but
phase V is induced by the special combination Qm1=0 and Qm20, we are going to
investigate a simplified “effective” FEE, as proposed by Gufan /27/. Also the terms
with B‘3, C‘3 and C‘4 in eq. (13) have been omitted:
F‘= A1Q62 + A‘2Q6
4 + A‘3Q66 + B1Q3m
2 + B‘2Q3m4 + C‘1Qm2
2 + C‘2Qm24 +
‘1Q62Q3m
2 + ‘2Q62Qm2
2 + ‘3Q3m2Qm2
2
• Minimizing of F’ with regard to the OPs according to the system of equations
yield:
(21)
0'
0'
0'
236
=
=
=
mm Q
F
Q
F
Q
F(22)
Hempel, 2008-2015 Page 25 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase I (6mm):
Phase III (6):
Phase IV (3m):
Phase V (m):
The coefficients B‘2, C‘2 and A‘3 have to be positive and A‘2 is assumed to be negative.
Note:
Other Solutions where the different OPs do exist simultaneously would imply to have PTs
related to “reducible representations” which are not considered here because their induced
low symmetry phases are beyond the scope of this work.
(23)
(24)
(25)
(26)
0 2
0 21'
2
12
36 ==−== mmm QQB
BQQ
'
2
'
12
21362
0 0 0C
CQQQQ mmm −====
0 0 0 2136 ==== mmm QQQQ
0 0 3
133
2132'
2
'
31
'
3
'
2
'
3
'
22
6 ===−−−= mmm QQQA
AA
A
A
A
AQ
Hempel, 2008-2015 Page 26 of 57
Phase Sequence and Landau Theory of LiKSO4
The schematic temperature dependences of the OPs are depicted in Fig. 2
• To check whether a phase sequence I-III-IV-V is possible at all and the solutions, as
given above, are stable, we shall look at the phase diagram
• To this end the formalism developed by Gufan /28/ for two coupled OPs was expanded
to three coupled OPs
• First finding is that in fact the phases I-III, III-IV, III-V, IV-V are adjacent (ref. to Fig. 3).
• Determination of the “loss of stability limits” for the considered phases turns out that
they do not coincide, but do overlap. This means that all related PTs are necessarily of
1st order (see e.g. /25/) – which is well proven by all experimental findings so far
• As usual for 1st order PTs, the actual phase boundaries are determined by the equality
of the FEEs of the concerned neighboring phases. In case of three coupled OPs as in
LPS, the PTs are depicted by planes in the 3-dimensional phase diagram spanned
by A1, B1, C’1 (ref. to Fig. 3)
• Since we stick to the basics of the Landau Theory, we have in eqs. (21), (23-26)
consequently assumed a linear temperature dependence of the coefficients A1, B1, C’1and weak or no temperature dependence of the remaining coefficients
Hempel, 2008-2015 Page 27 of 57
Phase Sequence and Landau Theory of LiKSO4
T (K)
Q6
Q3m
Qm2
0
0
0
0Qm1
m (V)
3m (IV)
6 (III) 6mm (I)
T“m1m2=195 T“3m=260 T“6 708heating:
T“m1m2=190 T“3m=205 T“6 708cooling:
T‘m1m2
T3m
T6
Fig. 2 Qualitative temperature dependences of the Order Parameters Q6, Q3m, Qm1, Qm2
Hempel, 2008-2015 Page 28 of 57
Phase Sequence and Landau Theory of LiKSO4
• Moreover, A‘3, B‘2 and C‘2 in eq. (21) must be positive definite (because FEE must be
positive for large OP-values), and A‘2 shall be negative definite (1st order transition)
• The OP-coupling coefficients ‘1, ‘2 and ‘3 must be all positive, because otherwise the
phase diagram looks different and the phases of interest are not adjacent - which
contradicts the above mentioned phase sequence
• Within the phase diagram we have now to search for a thermodynamic path along a
straight line passing through the phases of the sequence I-III-IV-V, like a vector:
with A1, B1, C’1 being the coefficients from eq. (21) and
unit vectors along the axes
Note: Indeed, this path reflects the practical procedure i.e. the observation of the mentioned phase
sequence at changing temperature.
• It also turns out that such a sequence is feasible provided the path-vector proceeds not
too far from the origin of the coordinate system – namely above a Three-Phase Point.
In particular the following inequalities have to be satisfied (see also /28/):
'
111
'111 CBA eCeBeA
++axese
( ) ( ) ( )
( ) ( ) ( )2'
23'
22'
2
'
22'2'
2
2'
2
'
22'
'
1'
23'
'
2
2'
2
'
22'
1
2
2'
3'
22'
12'
2'2'
1
2'
12'
2'
12
'3'
'
1
2'
12'
2'
1
64
444
8
4
64
444
8
4
CA
CACACand
CA
CAA
BA
BABABand
BA
BAA
−+−
−
−+−
−
(27)
Hempel, 2008-2015 Page 29 of 57
Phase Sequence and Landau Theory of LiKSO4
B1
C‘1
A1
I-III(straight plane ||
B1-C‘1-plane)
IV-V(straight plane ||
A1-axis)
I-V(A1-B1-semi-plane)
III-IV (bent plane ||
C‘1-axis)
III-V (bent plane
|| B1-axis)
Phase I
(PG 6mm)
Phase V
(PG m)
Phase IV
(PG 3m)
Phase III
(PG 6)
I-IV(A1-C‘1-semi-plane)
Fig. 3 Section of Phase Diagram
Hempel, 2008-2015 Page 30 of 57
Phase Sequence and Landau Theory of LiKSO4
• If in eq. (27) the equality sign is fulfilled, the equations describe lines in the phase
diagram which are parallel to axis C’1 respectively B1 where 3 adjacent phases start
to coexist. One of the three is always accompanied by mixed different OPs (which
is not of interest here)
• Looking again at the path vector and Fig 3. It turns out that the sequence I-III-IV-V
materializes if T6>T3m>T’m1m2. Interestingly, the sequence I-IV-III-V is theoretically also
possible if T6>T’m1m2>T3m. However, series like I-V-III-IV or I-V-IV-III are not possible.
• When coming from high temperatures and moving along the path-vector we shall
calculate those temperatures where the vector crosses the planes which represent the
phase boundaries of the relevant adjacent phases in the phase diagram. These
actual transition temperatures shall be equivalent to those which are observed
experimentally. In fair approximation one obtains:
I(PG 6mm) → III(PG 6)
III(PG 6) → IV(PG 3m) with A’20 (28)
oAA
ATT
'3
2'2
6"
64
+=
oo
moo
m
BA
BA
TBTA
BA
T
−−
−−
2'
2'2
362'
2'2
"3
3
3
Hempel, 2008-2015 Page 31 of 57
Phase Sequence and Landau Theory of LiKSO4
• Finally the predicted temperature behaviour of the OPs, basing on the above made
stability considerations, can be summarized as follows (see once more Fig. 2):
➢ Coming from high temperatures, at T”6 Q6 suddenly appears with a jump
(just like for an ordinary PT of 1st order)
➢ On further cooling at T”3m Q6 suddenly disappears and at the same
temperature Q3m appears with a jump, although Q3m could (theoretically)
exist already above T”3m (see dashed line in Fig. 2, which shows the
typical 2nd order behavior reflected by eq. (25)), if this were not forbidden
by stability reasons
➢ Qm2 behaves similar like Q3m. The equilibrium value of Qm1 is always
equal to zero
IV(PG 3m) → V(PG m) (14) eq. from Twith
CB
BC
TCB
BCT
T mm
o
o
mo
omm
mm'
21
2'2
2'2
32'2
2'2'
21
"21
1−
−
=
Hempel, 2008-2015 Page 32 of 57
Phase Sequence and Landau Theory of LiKSO4
6. Temperature Dependences of Material Coefficients
• The FEE as provided in eq. (1) does have dynamical significance and is suited to derive
the temperature dependences of the material coefficients).
• Basing on the work of Slonczewski /29/ and Rehwald /30/, the material coefficients
calculate from FFE (eq. (1)) as follows:
Pkv
k jk
vjQik
i ji
kj RPP
FR
Fwith
=
=
=
3
1
22
vn
Pkv
vk kmkn
Qik
ki imnm
Emn
PS
FR
PS
F
QS
FR
QS
F
SS
Fc
−
−
=
2
,
22
,
22
(29)
Hempel, 2008-2015 Page 33 of 57
Phase Sequence and Landau Theory of LiKSO4
(31)kn
Qik
ki imnm
mnQS
FR
QP
F
SP
Fh
−
=
2
,
22
Qik
i ji
kj RQQ
Fwith
=
2
Notes:
1. The coefficients cEmn,
Tmn to be determined according to the equation above are those at constant
electric field, and at constant mechanical stress, respectively. These boundary conditions are usually
realized within practical measurements.
2. In equations (29-31) “Qi”, “Qk” stand symbolically for Q6, Q3m, Qm1, Qm2 (to shorten the expressions).
(30)vn
Skv
vk kmkn
Qik
ki imnm
Tmn
SP
FR
SP
F
QP
FR
QP
F
PP
F
−
−
=
2
,
22
,
22
Skv
k jk
vjQik
i ji
kj RSS
FR
Fwith
=
=
=
6
1
22
Hempel, 2008-2015 Page 34 of 57
Phase Sequence and Landau Theory of LiKSO4
For phase V (PG m) the following relations hold for which govern the OPs’ influence on
the material coefficients via eqs. (29-31). Approximations for small OP-values are provided
in form of truncated Taylor Series:
For the phases III (PG 6) and IV (PG 3m) yield:
The expressions for FQiQi are collected in the table (eq. (20)) on page 23.
Now, first the stiffnesses at constant polarisation and the dielectric impermeabilities at
constant strain as well as the piezoelectric coefficients are calculated and presented on the
following 4 slides.
Note: In the following tables the “Qi” are those from eqs. (16)-(19) and the expressions for the state
parameters are those taken from eqs. (3)-(11)
+−
−= 2
2
1
202211
2
41
212211
22
m
1QQQQQQ
QQQ QC
KC
2C
1
FFF
FR
mmmmmm
mm
( ) ( ) 1
3
1
6 3366
−−==
mmQQQ
mPGQQQPG FR ,FR
(32)
+−
−= 2
2
1
212222
2
121
212211
11
m
1QQQQQQ
QQQ QC
KC
2C
1
FFF
FR
mmmmmm
mm
322
1
2222112
2212211
21
m
QQQQQQ
QQQQ QC
K
FFF
FRR
mmmmmm
mm −
==
QikR
Hempel, 2008-2015 Page 35 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
c11= co11
c22= co11
c33= co33
c44= co44
c55= co44
66
26
232
6
)(42
QQF
QjNP hQ
+−
mmQQ
mm
F
QnQP lQ
33
23
232
3
)(42
+−
++−++
++−++−
Qm
Qm
Qm
mRQurSS pUP wS
RQurSS pUP RQwSpQ
122
22135
222
22
213112
2522
))(2(4
))(2(4)(2
66
26
232
6
)(42
QQF
QjNP hQ
+−
mmQQ
mm
F
QnQP IQ
33
23
232
3
)(42
+−
66
26
23 )(4
QQF
QiOP +−
mmQQ
m
F
QmRP
33
23
23 )(4 +
−Q
m RQqVP 222
22
3 )(4 +−
262kQ 2
32 moQQ
m RtsQ 1222
22 −
262kQ
232 moQ ( )
++−−
+++−−
Qmmm
Qmm
Qm
mRtQsSQWP QSS w
RtQsSQWP RQSS wsQ
122521221
222
2521112
2
2
2122
)42()(2
)42()(2
−+−−−
−+−−+−−
Qm
Qm
Qm
mRQurSS pUP wS
RQurSS pUP RQwSpQ
122
22135
222
22
213112
2522
))(2(4
))(2(4)(2
Hempel, 2008-2015 Page 36 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
c66= co66
c12= co12
c13= co13
c23= co13
c15= 00 vQ3m
(ferrobielastic)
66
26
232
6
)(42
QQF
QjNP hQ
+−−
232 mlQ
Qmm RQupQ 112
222
2 42 −262hQ
mmQQ
mm
F
QnQP IQ
33
23
232
3
)(42
+−− ( )
+−−
+−−+−
−
−−
Qm
Qm
Qm
m
RQuSS p wS
RQuSS p rUP
RQSw
pQ
122
2215
222
22
212
3
112
225
2
22
))(2(4
))(2(4)(42
( )
66
2633 )(4
QQF
QiOPjNP ++− ( )
mmQQ
m
F
QmRPnQP
33
2333 )(4 ++
−( )
++
+++−+−
Qm
Qm
RQqVP wS
RQqVPurSS pUP
122
235
222
23213
2
)())(2(4
( )
66
2633 )(4
QQF
QiOPjNP ++−
( )
mmQQ
m
F
QmRPnQP
33
2333 )(4 ++
−
( )
( )
( )
( )
++−+−+
+++
++
++−++−
−
Q
m
mmm
Qmmm
Qm
RQurSS pUPSS w
QtQsSQWPwS
RtQsSQWPQ
urSS pUPRQSS Sw
122221321
225215
2225212
213112
22152
)(2)(2
42
42
)(22)(
( )
+−
+−+−−−
Qm
Qm
RQqVP wS
RQqVPurSS pUP
122
235
222
23213
2
)())(2(4
Hempel, 2008-2015 Page 37 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
11= o11
22= o11
33= o33
h31= ho31
h32= ho31
h33= ho33
262bQ 2
32 mdQ
66
26
2321 ))((2
QQF
QXOSSS N +++−
mmQQ
m
F
QYRSSS Q
33
23
2321 ))((2 +++
−Q
m RQZVSSS U 222
22
321 ))((2 +++−
( )
66
263321
26
))((4
QQF
QjNPXOSSS N
NQ
++++
−
( )
mmQQ
m
m
F
QnQPYRSSS Q
33
233321
23
))((4 ++++
−
( )
( )
++++
++−++++
−
Qm
Qm
m
RQZVSSS UwS
RQurSS pUPZVSSS U
UQ
122
23215
222
2213321
22
)(2
)(2))((4
( )
66
263321
26
))((4
QQF
QiOPXOSSS N
OQ
++++
−
( )
mmQQ
m
m
F
QmRPYRSSS Q
RQ
33
233321
23
)2)((4 ++++
−
( ) Qm
m
RQqVPZVSSS U
VQ
222
23321
22
))((4 ++++
−
262bQ 2
32 mdQ
Qm RfeQ 11
2222 −
Qmmm RfQePQWS eQ 22
22125
22 )42(2 ++−
( )
66
263321
26
))((4
QQF
QjNPXOSSS N
NQ
++++
−
( )
mmQQ
m
m
F
QnQPYRSSS Q
33
233321
23
))((4 ++++
−
( )
( )
+++−
−+−−+++
−
Qm
Qm
m
RQZVSSS U wS
RQurSS pUPZVSSS U
UQ
122
23215
222
2213321
22
)(2
)(2))((4
Hempel, 2008-2015 Page 38 of 57
Phase Sequence and Landau Theory of LiKSO4
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
h15= ho15
h24= ho15
h14= 0xQ6
(ferroelastoelectric)0
h11= 0 0yQ3m
(ferroelastoelectric)
26PQ
23mSQ ( )
−+++
++++
−
Qmmm
Qmmmm
m
RQSS wfQePQWS
RtQsSQWPfQePQWS
WQ
122212125
2225212125
22
)()42(
42)42(
2
6PQ 23mSQ
Qm RtfWQ 112
2 −
• The terms surrounded by circles give rise to expressions that are present in the
entire temperature region – from phase I through V, i.e. even there where the
OPs are identical to zero (in particular there holds acc. to eq. (32): ).
• Subsequently the stiffnesses at constant electric fields and the dielectric
impermeabilities at constant mechanical stresses are calculated
( )
( )
+++
++−+++−
Qmm
Qmmm
RfQePQWSwS
RQurSS pUPfQePQWS
1221255
2222132125
42
)(2)42(2
( ) Qmm RfQePQWSt 122125 42 +++
( ) 1
12211 2 −== CRR QQ
Hempel, 2008-2015 Page 39 of 57
Phase Sequence and Landau Theory of LiKSO4
To this end the terms standing most right in equations (29) and (30) are
calculated and result in*):
and*)
(33)
Note: *) OP-related terms have been considered to be small and are therefore omitted.
( ) 1
112211 2−
= oPP RR
( ) 1
3333 2−
oPR
( )( ) ( )( )oooooo
oooSS
cccccc
cccRR
3312112
131211
2133311
22112 +−−
−−=
( ) ( )( )oooooo
oooSS
cccccc
cccRR
3312112
131211
2133312
21122 +−−
−=
( ) oooo
oSSSS
cccc
cRRRR
3312112
13
1332233113
2 +−===
( )( ) oooo
ooS
cccc
ccR
3312112
13
121133
2 +−
+−
024421441544552255115 ========== SSSSSSSSSS RRRRRRRRRR
( ) 1
445544
−= oSS cRR ( ) ( ) 1
1211
1
6666 2−−
−= oooS cccR
Hempel, 2008-2015 Page 40 of 57
Phase Sequence and Landau Theory of LiKSO4
stiffnesses at
constant
polarisation
„electromechanical“ contributions in the individual phases
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
cE11= c11
cE22= c22
cE33= c33
cE44= c44
cE55= c55
Pm RVQL 33
222 )( +−
Pm RWQM 22
222 )( +−
Pm RUQK 33
222 )( +−
Pm RWQM 11
222 )( +−
Pm RUQK 33
222 )( +− PRK 33
2− PRNQK 3322
6 )( +−
++−
Pm
Pm
RQQK
RyQ
3322
3
112
3
)(
)(
PRK 332− PRNQK 33
226 )( +−
++
−−
Pm
Pm
RQQK
RyQ
3322
3
222
3
)(
)(
PRL 332− PROQL 33
226 )( +− P
m RRQL 3322
3 )( +−
PRM 222−
++−
P
P
RPQM
RxQ
2222
6
112
6
)(
)( P
m RSQM 2222
3 )( +−
PRM 112−
−+
+−
P
P
RxQ
RPQM
222
6
1122
6
)(
)( P
m RSQM 1122
3 )( +−
Hempel, 2008-2015 Page 41 of 57
Phase Sequence and Landau Theory of LiKSO4
stiffnesses at
constant
polarisation
„electromechanical“ contributions in the individual phases
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
cE66= c66
0 0 0
cE12= c12
cE13= c13
cE23= c23
cE15= c15 0 0 0
Pm RUQK 33
222 )( +−
( )
+
+−
Pm
m
RVQL
UQK
332
2
22 )(
( )
+
−
Pm
m
RSQM
yQ
1123
3 )(
Pm RyQ 22
23 )(−
PRK 332− PRNQK 33
226 )( +−
++
−−
Pm
Pm
RQQK
RQy
3322
3
1123
2
)(
PKLR 33− ( )
+
+−
PROQL
NQK
3326
26 )(
( )
+
+−
Pm
m
RRQL
QQK
3323
23 )(
PKLR 33− ( )
+
+−
PROQL
NQK
3326
26 )(
( )
+
+−
Pm
m
RRQL
QQK
3323
23 )(
( )
+
+−
Pm
m
RVQL
UQK
332
2
22 )(
Hempel, 2008-2015 Page 42 of 57
Phase Sequence and Landau Theory of LiKSO4
Impermea-
bilities
at
constant
strain
„electromechanical“ contributions in the individual phases
Phase I
(6mm)Phase III (6) Phase IV (3m) Phase V (m)
T11= 11
T22= 22
T33= 33
Sm RWQM 55
222 )( +−
( ) Sm RWQM 44
222+−
( )
( )( )
+
+
++
++
++
+
−
SS
m
m
Sm
SSS
m
RR
VQL
UQK
RVQL
RRR
UQK
2313
22
22
3322
2
122211
222
)(2
)(
2
)(
SRM 552−
++−
S
S
RPQM
RxQ
5522
6
442
6
)(
)( ( )
++
−+
−
Sm
SSS
m
RSQM
RRR
yQ
5522
3
122211
23
)(
2
)(
SRM 442−
( )
+
+−
S
S
RxQ
RPQM
552
6
44
226
)(
( )
+
+−
Sm
Sm
RyQ
RSQM
662
3
44
223
)(
( )
( )
++
+
++
−
SS
S
SSS
RRKL
RL
RRR
K
2313
332
122211
2
2
2( )
( )( )
++
++
++
++
+
−
SS
S
SSS
RROQL
NQK
ROQL
RRR
NQK
231326
26
3322
6
122211
226
)(2
)(
2
)(( )
( )( )
+
+
++
++
++
+
−
SS
m
m
Sm
SSS
m
RR
RQL
QQK
RRQL
RRR
QQK
2313
23
23
3322
3
122211
223
)(2
)(
2
)(
Hempel, 2008-2015 Page 43 of 57
Phase Sequence and Landau Theory of LiKSO4
7. Discussion of Results and Comparison with Experimental Findings
• In this chapter the qualitative temperature behaviour of the state parameters and
material coefficients will be depicted and compared with experimental findings
• In order to do this into the expressions of the previous slides the precise dependences
have to be introduced, i.e. for strains eqs. (3)-(8), for polarisations eqs. (9)-(11), for
eqs. (32), for and eq. (33). Everywhere where the OPs appear, their temperature
dependences have to be inserted from eqs. (16)-(19).
• It must be noted here that the curves shall be correct for phases III and IV, because
there are no ferroelectric nor ferroelastic domains, but care has to be taken because
ferroelastocelectricity respectively ferrobielasticity are present which do have an effect
on certain material coefficients
• experimentally found behaviour for phase V is influenced by the always existing
multidomain state and can therefore hardly be compared with theoretical predictions.
Only exemptions are the “longitudinal properties” along the c-axis, because the domains
behave all equally in this direction.
• Nevertheless the predicted behaviour in phase V is shown as well – but under the
assumption, that the crystal undergoes the transition IV → V to a (defined) single
domain state as stipulated earlier (see slide 6)
QijR
SijR P
ijR
Hempel, 2008-2015 Page 44 of 57
Phase Sequence and Landau Theory of LiKSO4
• The temperature dependences around the I→III PT are not shown, because phase II
has been omitted, and therefore no such PT (i.e. skipping of phase II) can materialize
in practical experiments.
• Wherever possible, experimental data measured at increasing temperatures have been
chosen, because those data are more reproducible then in cooling runs (see e.g. /5/)
• Some more remarks have to be made:
- Since some coupling terms of the general form X*(OP)2 appear in FEE (see
eq. (1)), the affected material coefficients (cij, kl, hmn) show jumps downwards
at the transitions temperatures (T“6, T“3m, T“m1m2) where the respective OPs
appear suddenly (e.g. on cooling).
- Also the opposite is true: If at a transition temperature an OP becomes
instable (i.e. disappears) the jump of the material coefficient is directed
upwards.
- It has been shown that exactly at T“6 Q6 disappears and Q3m appears
(the same applies at T“m1m2 for Q3m and Qm2). The accompanied jumps of
the coefficients compete and yield a “resulting” jump which can be positive or
negative or even zero (latter in case of exact compensation of contributing
jumps)
- Precise predictions would required to know the signs and values of the
relevant term-factors of FEE (see eq. (1))
Hempel, 2008-2015 Page 45 of 57
Phase Sequence and Landau Theory of LiKSO4
• Another comment has to be made if stiffnesses cij are compared with compliances sij,
The relation sii=1/cii holds only if all cij with ij are equal zero (or very small). The
relation between impermeabilities and permittivities is analogue.
• The relation between the piezocoefficients hmj and the usually measured dki reads
like /31/:
(34)
• As mentioned earlier the present calculations do not cover the ordinary thermal
dilatation/contraction nor the ordinary pyroelectric effect.
• Since in particular the signs of the majority of the term-factors of FEE (see eq. (1))
are not known, different temperature dependences - depending on the signs chosen –
are possible. Wherever reliable experimental data are available, only those signs have
been considered, where the curve fit is the best.
Snm
Eijminj
Smn
Eijminj cdhrespshd == .
Hempel, 2008-2015 Page 46 of 57
Phase Sequence and Landau Theory of LiKSO4
orthorhombic
setting, b=3 a
T“m1m2T“3m
S1
S2 S1=S2
T“m1m2T“3m
S3
T“m1m2T“3m
S4=S6=0
T“m1m2T“3m
S5
Taken from /32/
• There is a good agreement between predicted and experimental data (see also /5/ and /33/)
• Since strain S5 linearly depends on OP Qm2 (see eq. (13)), pseudoproper ferroelastic behaviour should
be observed in phase V
• As found in /5/, LPS is monoclinic, but pseudo-orthorhombic in phase V. The angle between the polar
c-axis and axes a (and b) amounts to 90° → therefore strain S5 is in fact zero (but can theoretically be
unequal zero as schematically depicted above).
• As an explanation from the phenomenological point of view it can be assumed that the responsible
coupling coefficients t and f in eq. (1) and in eq. (7) are very small.
Hempel, 2008-2015 Page 47 of 57
Phase Sequence and Landau Theory of LiKSO4
T“m1m2T“3m
P1
T“m1m2T“3m
P2=0
Taken from /34/
• There is a good agreement between predicted and experimental data
• Since polarisation P1 linearly depends on OP Qm2 (see eq. (9)), pseudoproper ferroelectric behaviour
should be observed in phase V as well
• As found experimentally in /12/, /35/, LPS doesn’t exhibit any spontaneous polarisation component P1
nor a P1- (or P3-) polarisation switching in phase V
• Nevertheless, P1 is allowed by group theory (similar like for S5 on the previous slide)
• As an explanation from the phenomenological point of view it can be assumed that the coupling
coefficients t and f in eq. (1) and in eq. (9) are very small.
T“m1m2T“3m
P3
Hempel, 2008-2015 Page 48 of 57
Phase Sequence and Landau Theory of LiKSO4
Taken from /12/
Taken from /8/
• Relation ii=1/ii holds exactly for phases III,
IV, but in phase V only if ij <<1 for iǂj
• There is a good agreement regarding 33, but
the experimentally found slope in phase IV is
not predicted. Experiments by others (see /35/, /12/) revealed no slope in phase IV nor any
singularity at the PT IV III.
• There is some principal agreement regarding 11 and 22, but the predicted temperature
dependence (with negative overall slope) above (!) T‘‘m1m2 has never been observed experimentally.
• A possible explanation is that coefficient f in eq. (1) and in the respective expressions on slide no. 37
is very small, as already assumed to explain the observed behaviour of P1
T“m1m2T“3m
33
T“m1m2T“3m
11=22
11≠22
Hempel, 2008-2015 Page 49 of 57
Phase Sequence and Landau Theory of LiKSO4
• Depicted curves for the hij coefficients are just possible examples. Depending on the magnitude
and sign of the relevant term-factors (which are not known from theory) other curve-forms are
possible as well (ref. to comments on slides nos. 44 and 45).
• Also it must be noted that usually the piezocoefficients dij are practically measured, which cannot be
directly compared with the hij (see eq. (34)).
• Interesting is the fact, that h15 and h24 should be temperature dependent (with a positive overall
slope) already in phases (I), III and IV, i.e. above T‘‘m1m2.
• No measurable slope would again imply that the coupling coefficients f and t in eq. (1) are very
small.
• Experimentally measured piezocoefficients dij can be found e.g. in /12/, /36/.
T“m1m2T“3m
h31=h32,h33
h31≠h32,h33
T“m1m2T“3m
h15=h24
h15≠h24
Hempel, 2008-2015 Page 50 of 57
Phase Sequence and Landau Theory of LiKSO4
Taken from /37/
Taken from /14/
• There is a good agreement for c11, c22 –
meaning that the observed dependences can be
described by the formulas derived earlier.
• c33 shows at T‘‘3m a jump which can be explained
if, on e.g. heating, at T‘‘3m Q6 appears accompa-
nied by a jump downwards and Q3m disappears
accompanied by a jump upwards. Obviously the
first dominates and the resulting jump is directed
downwards (see also explanation on slide no. 44)
• The same explanation holds in principle for c13
and c23.
T“m1m2T“3m
c11=c22
c11≠c22
T“m1m2T“3m
c33,
c13=c23
c33,
c13≠c23
Hempel, 2008-2015 Page 51 of 57
Phase Sequence and Landau Theory of LiKSO4
Taken from /37/
• There is less agreement for c44, c55 which should
show a positive overall slope in phases (I), III and
IV, i.e. above T‘‘m1m2. Only possibility to explain
the negative slope is that coefficient t is very small,
and the terms with Q2 dominate, i.e. they provide
negative (slope) contributions – with coefficients
k, o (see eq. (1) being positive.
• The direction of the experimentally found jump at
T‘‘3m can be explained similarly as for c33.
• The predicted jump at T‘‘m1m2 was experimentally
not observed, indicating small values of coeffici-
ents t, f and opposite signs of coefficients o and s.
T“m1m2T“3m
c44≠c55
c44=c55
T“m1m2T“3m
c66
• c66 is in fair agreement with measured data
• The negative slope above T‘‘3m (i.e. in phase
III) is explainable within the model if coeffi-
cient h is positive definite
Hempel, 2008-2015 Page 52 of 57
Phase Sequence and Landau Theory of LiKSO4
• The depencence of c12 is closely related to those of c11, c22, c66 (ref. to the expressions derived
earlier)
• c15 is a „morphic“ coefficient, i.e. it is zero above T‘‘3m. It fits well with RUS measurements /11/
but it must be noted that another coordinate system (rotated by 30° around c-axis) was used. This
mean that our c15 coincides with c14 of /11/.
T“m1m2T“3m
c12
T“m1m2T“3m
c15
Note: There are numerous experimental data regarding the elastic stiffnesses of LPS available in the
literature, e.g. /38/-/41/, but many of them are contradictorily.
Hempel, 2008-2015 Page 53 of 57
Phase Sequence and Landau Theory of LiKSO4
8. Conclusion
Basing on Symmetry Considerations, the Landau Theory of Phase Transitions and the
related Phase Diagram an attempt has been made to theoretically describe the phase
sequence of LPS.
It turned out that the majority of experimental data found up to now can be satisfactorily
explained and even reasonably fitted qualitatively.
The predicted appearances of S5 and P1 in phase V are not evidenced by experiments.
Also the expected overall temperature dependences of the material the coefficients 11,
22, c44, c55 in phases I, III and IV have not been experimentally observed.
These substantial deviations can be explained qualitatively well with the model if certain
coefficients in the Free Energy Expansion are assumed to be very small.
Hempel, 2008-2015 Page 54 of 57
Phase Sequence and Landau Theory of LiKSO4
9. Literature
/1/ Perpetuo G. J. et al., Phys. Rev. B45, 5163 (1992)
/2/ Scherf, Ch., Dissertation, RWTH Aachen, 2000
/3/ Park, H. M., et al., Acta Cryst. A61, C373 (2005)
/4/ Sorge, G. et al., phys. stat sol. (a)97, 431 (1986)
/5/ Bhakay-Tamhane, S. et al., Phase Transitions 35, 75 (1991)
/6/ Klapper, H. et al., Acta Cryst. B43, 147 (1987)
/7/ Zeks B., et al., Phys. Stat. Sol. b122, 399 (1984)
/8/ Fujimoto S., et al., J. Phys. D: Appl. Physics 18, 1878 (1985)
/9/ An Tu et al., Solid State Commun., 61, 1 (1987)
/10/ Mroz B., et al., J. Phys.: Condensed Matter 1, 5965 (1989)
/11/ Willis F., et al., Phys. Rev. B54, 9077 (1996)
/12/ Sorge, G., Hempel, H., Ferroelectrics 81, 167 (1988)
/13/ Hempel, H., phys. stat. sol. (b)163, K77 (1991)
/14/ Quirion G., J. Phys. C: Condensed Matter 15, 4979 (2003)
Hempel, 2008-2015 Page 55 of 57
Phase Sequence and Landau Theory of LiKSO4
/15/ Boyle L. L., et al. Acta Cryst. A28, 485 (1972)
/16/ Sirotin, Yu. I., Shaskolskaya, M. P., “Fundamentals of Crystal Physics”,
Mir Publishers, Moscow, 1982
/17/ Janovec V., et al., Czech J. Phys. B25, 1362 (1975)
/18/ Toledano P., Toledano J.-C., Phys. Rev. B16, 386 (1977)
/19/ Wadhawan, V. K., Phase Transitions 3, 3 (1982)
/20/ Aizu, K., J. Phys. Soc. Japan 28, 706 (1970)
/21/ Sapriel, J., Phys. Rev. B12, 5128 (1975)
/22/ Toledano P., Toledano J.-C., “The Landau Theory of Phase Transitions“, World
Scientific Publishing, 1987
/23/ Kovalev, O. V., “Representation of Crystallographic Space Groups”, Taylor & Francis
Ltd., 1993
/24/ Cracknell, C. J., Bradley, A. P., “The Mathematical Theory of Symmetry in Solids”,
Clarendon Press, Oxford, 1972
/25/ Izyumov Yu. A., Syromyatnikov V. N., “Phase Transitions and Crystal Symmetry“,
Kluwer Academic Publishers, Dordrecht / Boston / London, 1990
Hempel, 2008-2015 Page 56 of 57
Phase Sequence and Landau Theory of LiKSO4
/26/ Kocinsky, J., “Theory of Symmetry Changes at Continuous Phase Transitions”,
Elsevier, Amsterdam-Oxford-New York, 1983
/27/ Gufan et al., Sov. Solid State Phys. 22, 951 (1980)
/28/ Gufan et al., Sov. Solid State Phys. 22, 270 (1980)
/29/ Slonczewski, J. C. and Thomas, H., Phys. Rev. B1, 3599 (1970)
/30/ Rehwald, W., Adv. In Pysics, Vol. 22, 721 (1973)
/31/ Xu, Y., „Ferroelectric Materials and Their Applications“, Elsevier Science
Publishers, 1991
/32/ Tomaszewski, P. E., et al., Phase Transitions 4, 37 (1983)
/33/ Desert, A. et al., J. Phys. Cond. Matter 7, 8445 (1995)
/34/ Breczewski, T., et al., Ferroelectrics 33, 9 (1981)
/35/ Cach, R., et al., Ferrolectrics 53, 337 (1984)
/36/ Mroz, B., et al., Ferroelectrics 42, 71 (1982)
/37/ Kabelka, H., et al., Ferrolelectrics 88, 93 (1988)
/38/ Borisov, B. F., et al., phys. stat. sol. (b)199, 51 (1997)
Hempel, 2008-2015 Page 57 of 57
Phase Sequence and Landau Theory of LiKSO4
/39/ Drozdowski, M., et al. Ferroelectrics 77, 47 (1988)
/40/ Ganot, F., et al., J. Phys. C: Solid State Phys. 20, 491 (1987)
/41/ Kassem, E. M., et al., Acta Physica Polonica A63, 449 (1983)