3.

Thermodynamics Of Phase Transitions

3.A. Introductory
3.B. Coexistence of Phases: Gibbs Phase Rule
3.C. Classification of Phase Transitions
3.D. Pure PVT Systems
3.E. Superconductors
3.F. Helium Liquids
3.G. Landau Theory
3.H. Critical Exponents
3.A. Introductory
3.B. Coexistence of Phases: Gibbs Phase Rule
Coexisting phases are in thermal & mechanical equilibrium.
They can also exchange matter.
Hence, they have the same T, Y, and μ.

YXT System of 1 Kind of Particles

The condition for the coexistence of 2 phases I and II is

 I Y , T    II Y , T  (3.1)

where, according to the Gibbs-Duhem equation (2.62), μ is a function of intensive

variables only.

Eq(3.1) can be solved to give the coexistence curve of the 2 phases

Y  Y T  (3.2)

in the Y-T plane.

The conditions for the coexistence of 3 phases I , II and III are

 I Y , T    II Y , T    III Y , T  (3.3)

which can be solved to give the locations of triple points in the Y-T plane.

There cannot be a coexistence of more than 3 phases because the coexistence

conditions will be over-determined.

YXT System of m Kinds of Particles

For a mixture of m different types of particles in a YXT system, the independent
variables can be chosen as Y , T , ni , where ni is the number of moles of the ith type
particles, and i  1, , m . For a closed system, the total number n of moles is a
constant and we can write
m m

 ni  n
i 1
→ x
i 1
i 1

ni
where xi  is the molar fraction of particles of type i.
n
Thus, only m  1 of the xi 's are independent and the system itself has only m  1

independent (intensive) variables Y , T , x , where x   x1 , , xm 1  .

The conditions for the coexistence of r phases are

1I Y , T , x I   1II Y , T , x II     1r Y , T , x r  (3.4)

… … …

iI Y , T , x I   iII Y , T , x II     ir Y , T , x r  (3.5)

… … …

mI Y , T , x I   mII Y , T , x II     mr Y , T , x r  (3.6)

where is is the chemical potential of the i-type particles in the sth phase whose

composition is x s   x1s , , xms 1  .

The number of independent equations in (3.4-6) is m  r  1 .

The number of unknowns is 2  r  m  1 .

The maximum number R of phases that can coexist simultaneously is therefore given
by

m  R  1  2  R  m  1

i.e.,
R m2
3.C. Classification of Phase Transitions
For a system of m kinds of particles,
m
G   n j j (3.7)
j 1

Since
m 
 G   G  G 
dG    dT    dY     dn j
 T Y n j   Y T n j  
j 1  n j 
TY n | i  j
i

we have, for constant T and Y,

m  G  m
 dG YT    dn j    j dn j (3.8)

j 1  n j  j 1
TY ni  j 

At a transition point, 2 or more phases can coexist in equilibrium. The absence of

matter exchange means that, for every kind of particles, the values of its chemical
potential in the coexisting phases are equal, i.e.,

ir  is for i  1, , m (a)

where r, s denotes the coexisting phases.

Without loss of generality, we shall restrict our discussion to cases where the entire
system is always in a single homogeneous phase.

Eqs(a) and (3.7) mean that, at a transition point, G of all coexisting phases are equal.

In other words, the function G Y , T , x  is continuous at a transition point.

The manifest differences in the properties of the phases must then appear as
discontinuities in some derivatives of G. If the 1st order derivatives are
discontinuous, the transition is called 1st order. Otherwise, it is called continuous.

In an older scheme introduced by Ehrenfest, the order of the transition is taken to be

the lowest order of the derivatives that are discontinuous.

Typical behavior of G of a PVT system near a 1st order transition is shown in Fig.3.2.

 G 
Discontinuity in    V means a finite difference of volumes of the 2 phases,
 P T n j 
  G   G 
I II

V  V  V  
I II
   (3.9)
 P T n j   P T n j 

 G 
Discontinuity in     S means a finite difference of entropy of the 2
 T  Pn j 

phases,

 G   G 
II I

S  S I  S II      (3.10)
 T  Pn j   T  Pn j 

which means the heat capacities are not defined.

Since G is continuous, the difference in enthalpy (latent heat) is given by

H    G  TS   T S (3.11)

The case for a continuous transition is shown in Fig.3.3.

The distinguishing feature is the peak in the heat capacity at the transition point.
3.D. Pure PVT Systems
3.D.1. Phase Diagrams
3.D.2. Coexistence Curves: Clausius-Clapeyron Equation
3.D.3. Liquid-Vapor Coexistence Region
3.D.4. The van der Waals Equation
3.D.1. Phase Diagrams
The coexistence curves of a typical pure PVT system are shown in Fig.3.4.
In general, there are 3 possible phases, namely, vapor, liquid, and solid.
All transitions are 1st order with accompanying latent heat.
Coexisting phases Name of curve Symmetry change
vapor-liquid vaporization no
liquid-solid fusion yes
vapor-solid sublimination yes
Point A is the triple point at which all 3 phases coexist.
Point C is the critical point at the end of the vaporization curve.

The vaporization curve AC actually denotes a vapor-liquid coexistence region as

shown in the P-v diagram in Fig.3.5. Note that the isotherms are also isobars in the
coexistence region.

The 3-D P-v-T diagram is shown in Fig.3.6.

3.D.2. Coexistence Curves: Clausius-Clapeyron Equation
G
The molar Gibbs energies (chemical potentials), g    , of 2 coexisting phases I
n
and II, must be equal. Thus,
dg I  dg II along the coexistence curve

For a PVT system, this means

v I dP  s I dT  v II dP  s II dT (3.12)
i.e.,
 dP  s I  s II s
    (3.13)
 dT  coex v  v v
I II

h
 (3.14)
T v
where h  T s is the molar latent heat. Eq(3.14) is called the Clausius-
Clapeyron equation.

Exercise 3.1

Prove that h  0 in a transition from a low to a high T phase.

Answer
Let II be the low and I be the high T phase.
Since the stable state has the lowest G, we have

G I  G II for T  TC

G I  G II for T  TC

Coupled with the stability conditions of section 2.H.3, we obtain a situation as

depicted in the Figure. Thus,

 G I   G II 
     for all T
 T  Pn j   T  Pn j 

so that
S I  S II for all T
and

H  TC S  TC  S I  S II   0

3.D.2.a. Vaporization Curve

3.D.2.b. Fusion Curve
3.D.2.c. Sublimation Curve
3.D.2.a. Vaporization Curve

Consider the coexistence curve between the liquid and gaseous phases of a pure
substance. This is indicated as the vaporization curve between the triple A and
critical point C in Figure 3.4.

If we neglect the volume changes of the liquid phase, we have

RT
vlg  v g  vl  v g 
P
where the last equality assumes the vapor to be an ideal gas.
The Clausius-Clapeyron equation thus simplifies to

 dP  hlg Phlg
    (3.15)
 dT  coex T vlg RT 2

where hlg is the molar latent heat of vaporization. Treating hlg as a constant,

we get, along the vaporization curve,

dP hlg dT
 P  R  T2
h
ln P   lg  const
RT

 h 
P  P exp   lg  (3.16)
 RT 
where P is the pressure for T   .

Exercise 3.2

Find the molar heat capacity ccoex along the vaporization curve.
Answer
 s 
ccoex  T  
 T  coex
With T, P as independent variables, we have
 s   s 
ds    dT    dP
 T  P  P T
On the vaporization curve, we have
 dP  Phlg
dP    dT  dT
 dT  coex RT 2
so that
  s   s   P  
 ds coex         dT
 T  P  P T  T  coex 

 s   s  Phlg 
     2 
dT
 T  P  P T RT 

Thus
 s  Phlg
ccoex  cP   
 P T RT
Using
 s   v  R
     
 P T  T  P P
where the last equality is valid for an ideal gas, we have
hlg
ccoex  cP  (2)
T
hlg
which means for T  , we have ccoex  0 so that the vapor gives off heat as T is
cP
raised.
3.D.2.b. Fusion Curve

Consider now the coexistence curve between the solid and liquid phases.
This is depicted as the fusion curve AB in Fig.3.4.
Note that the fusion curve does not terminate.
It can have either positive or negative slope, as depicted with a solid and dotted line in
Fig.3.4, respectively.
The relevant Clausius-Clapeyron equation is
 dP  hsl
   (3.17)
 dT  coex T vsl
where hsl  0 is the molar latent heat of fusion. Hence, the sign of vsl  vl  vs
determines that of the slope of the fusion curve. A well known example of a fusion
curve that has a negative slope is that between ice and water.
3.D.2.c. Sublimation Curve

The coexistence curve between the solid and the gas phases is called the sublimation
curve [see Fig.3.4]. The relevant Clausius-Clapeyron is

 dP  hg  hs hsg
    (3.18)
 dT  coex T  vg  vs  T vsg

where hsg is the molar latent heat of sublimation.

Neglecting vs and using the ideal gas approximation for vg , we have

RT
vsg  v g 
P
so that
 dP  Phsg
   (3.19)
 dT  coex RT 2

Thus, along the sublimation curve,

T 2 dP d ln P
hsg  R   R (3.20)
dT P 1
d 
T 

Exercise 3.3

Near triple point of NH3, we have

3726
sublimation curve: ln P  27.79 
T
3005
vaporization curve: ln P  24.10 
T
1. Find T and P of triple point.
2. Find latent heats of sublimation and vaporization.
Answer
1. The sublimation and vaporization curves meet at the triple point. Hence,
3726 3005
27.79   24.10   Tt  195.4 K
Tt Tt
3726
ln Pt  27.79   6.13kPa
195.4
2. For the sublimation curve,
 dP 3726 RT 2  dP  3726
 2  hsg      31 kJ / mol
PdT T P  dT  coex R
For the vaporization curve,
dP 3005 3005
 2  hlg   25 kJ / mol
PdT T R

Low Temperature Case

Using
 s   s 
ds    dT    dP
 T  P  P T
c  s   v 
 P dT  v P dP where        v P
T  P T  T  P
the molar enthalpy can be written as
dh  Tds  vdP

 cP dT  v 1  T  P  dP (3.21)

Now, at low temperatures, the vapor pressure along the sublimation curve is very low.
Thus, we can neglect the pressure variation and write
dh  cP dT on sublimation curve
so that
T
h  h   cP dT
0

T0

where h  h 0 at T0 .

The latent heat of sublimation is therefore

T
hsg  hg  hs  h    cPg  cPs  dT
0
sg (3.22)
T0

Eq(3.20) can thus be integrated to give

P 1 h
T
ln   2sg dT
P0 R T0 T

T  0 T' g 
 hsg    cP  cP  dT '' dT '
1 1
R T0 T '2
 s

 T0 

hsg0  1 1  1 T 1 T ' g 
   cP  cP  dT '' dT '
R  T0 T  R T0 T '2
   
s
(3.23)
 T0 
which can be used to extrapolate experimental measurements.
3.D.3. Liquid-Vapor Coexistence Region
Consider the phase diagram in the P-v plane depicted in Fig.3.7. [cf. Fig3.4-6].
Along an isothermal T0  TC , the straight line segment AB inside the vapor-liquid
coexistence region indicates a mixture of the vapor and liquid phases. Here, both
phases are at the same temperature T0 and pressure P0 as given by a point on the
vaporization curve. The molar volumes of the pure vapor and liquid phases are

given by v g  v A and vl  vB , respectively. The state D of molar volume vD thus

represents a mixture of phases obeying the lever rule

vD  x g v g  xl vl (3.24)

where x indicates a molar fraction. With the help of the constraint x g  xl  1 , we

can rewrite (3.24) as

x g  xl  vD  x g v g  xl vl

xl v g  vD
  (3.25)
x g vD  vl

Metastable States

According to eq(2.179), mechanical stability requires

1  v   v 
T      0    0 (a)
v  P T  P T

Consider again the isotherm T0  TC in Fig.3.7. If we extend the gaseous state

beyond point A into the coexistence region (see dashed line), mechanical stability can
still be maintained as long as condition (a) is satisfied. However, since the free
energy is not a minimum, the system is only metastable. These states are called
supercooled (vapor) states. Similarly, metastable states resulting from the
continuation of the liquid state past point B into the coexistence region are called
superheated (liquid) state.

It is possible to extend the superheated liquid states into the negative pressure region.
In which case, no wall is required to contain the system.

Note that at the critical point C, the molar volumes of both phases are equal. Since
the relevant symmetries are also the same, these phases become indistinguishable.
Also, there are no metastable states.

Law of Corresponding States

The liquid-vapor coexistence curve in the T-ρ plane can be plotted in terms of reduced
T 
quantities and , where the subscript C denotes a critical value.
TC C
Guggenheim found that such curves more or less coincide for a large number of pure
substances (see Fig.3.8). This is an example of the law of corresponding states,
which postulates that all pure classical fluids obey the same equation of state
involving reduced quantities.
The curve in Fig.3.8 is given by Guggenheim as the solutions to the following
equations

l   g 3 T 
 1  1   (3.26)
2C 4  TC 

l   g 7 
1/ 3
T 
 1   (3.27)
C 2  TC 

Heat capacities

In the coexistence region, the total internal energy of a mixture (point D in Fig.3.7 ) is

U tot  ng ug  v g , T0   nl ul  vl , T0  (3.28)

where ni is the number of moles of phase i substance present and ui  v, T  is the

molar energy of the substance when it is in the pure phase i with molar volume v and
temperature T. Dividing by the total number of moles, we obtain the molar total
internal energy

utot  x g ug  v g , T0   xl ul  vl , T0  (3.29)

so that the molar heat capacity at constant volume is

 u 
cv   tot 
 T  vD

 u   u   x 
 xg  g   xl  l    ul  ug   l  (3.30)
 T  coex  T  coex coex
 T  vD

where we've used dx g  dxl  0 .

The following steps are required to related (3.30) with directly measurable quantities.

1. From ul  ul  vl , T0  , we have

 ul   ul   ul   vl 

       
 T  coex  T  vl  vl T  T  coex

 u   v 
 cvl   l   l  (3.31)
 vl T  T  coex

Similarly,

 ug   ug   vg 

 T   c      (3.32)
 vg T  T  coex
vg
  coex

2. u  u 
l g coex  hl  P0vl   hg  P0v g 

 hlg  P0 vlg

  dP  
  T0    P0  vlg (3.34)
  dT  coex 

 x 
3. Finally, to calcualte  l  , we begin with
 T  vD

vD  xl vl  1  xl  vg

so that

 x   v   x   v 
0   l  vl  xl  l    l  vg  1  xl   g 
 T  vD  T  coex  T  vD  T  coex

 xl  1   vl   v g  
i.e.,     xl    1  xl    
 T  vD vlg   T  coex  T  coex 

1   vl   v g  
  xl    xg    (3.36)
vlg   T  coex  T  coex 

Putting everything into (3.30) gives

  u   vg     u   v  
cv  xg  cvg   g      xl  cvl   l   l  
 vg
  T  T  coex    vl T  T  coex 
   dP     vl   v g  
  T0    P0   xl    xg    (3.37)
  dT  coex    T  coex  T  coex 

  u   dP    v g  
 x g cvg   g   T0    P0   
 v  dT  coex T  coex 
  g T   

  u   dP    vl  
 xl cvl   l   T0    P0   (3.37a)
  vl T  dT  coex   T  coex 

Now, from
du  Tds  Pdv
we have
 u   s   P 
  T  PT  P (3.38)
 v T  v T  T  v

Also, using the Maxwell relation

 x   x   x   w 
 y    y    w   y  (2.8)
 z  w  y  z

we can write
 P   P   P   v 
        (3.39)
 T  coex  T  v  v T  T  coex

Putting (3.38-9) into the terms included in one of the square brackets in (3.37a), we
get, in the coexistence region,
 u   P   P   P 
   T0    P0  T0    T0  
 v T  T  coex  T  v  T  coex
 P   v 
 T0    
 v T  T  coex
Hence, (3.37) simplifies to
  P   vg  
2
  P   vl  
2

cv  xg cvg  T0       xl cvl  T0  v   T   (3.40)

 vg
  T  T  coex    l T   coex 

with all quantities on the right side measurable.

Since the isotherm is also an isobar in the coexistence region, cP is infinite for a
mixture in the coexistence region. Thus, adding heat at constant pressure to a
mixture in the coexistence region only converts some liquid into vapor. The
temperature remains constant until the entire mixture is turned into vapor.
3.D.4. The van der Waals Equation
The (molar) van der Waals equation
 a
 P  2   v  b   RT (2.12)
 v 
can be written in various forms such as
a RT
P  (3.41a)
v 2
vb
or
 RT  2 a ab
v3   b  v  v  0 (3.41)
 P  P P
An isotherm with T  TC in the v-P plane is shown in Fig.3.9.

In order to simplify the notations, we shall replace all quantities x with their reduced
x
values x  , where the subscript C denotes value at the critical point. Using the
xC
fact that the isotherm T  TC has an inflection point at the critical point, we get

 P  2a RTC
   3  0 (3.42)
 v T TC vC  vC  b 
2

 2P  6a 2 RTC
 2  4  0 (3.42a)
 v T TC vC  vC  b 3

Eliminating TC gives
4a 6a
 4  vC  3b
vC  vC  b  vC
3

so that (3.42) becomes

2a 8a
TC  3  vC  b  
2

RvC 27bR
which turns (3.41a) into
a 4a a
PC   2
 2

9b 27b 27b 2
Eq(2.12) thus becomes
 aP a  8a
 27b2  9b2v 2   3bv  b   27b T
 
 3 
  P  2   3v  1  8T (3.44)
 v 
 3 8T
P  (3.44a)
v 2
3v  1
1  8T  2 3 1
v 3  1   v  v  0 (3.44b)
3 P  P P
Note that the independence of these equations on a and b is another example of the
law of corresponding states.

Maxwell Construction

An isotherm with T  TC in the v-P plane is shown in Fig.3.9.

Now, for equilibrium states, an isotherm v  P  must be single- valued except at

transition points. Thus, if one tries to bring the system quasi-statically and
isothermally from A to I, the actual path followed must be one similar to ABCEGHI.
The task is to determine the position of the vertical line segment CEG, which indicates
a coexistence of phases [cf. Fig.3.7].

Note that one implication of the above discussion is that the states on the segment
DEF are not in equilibrium. This is indeed the case since the slope of the curve there
is positive so that
1  v 
T      0  mechanical instability
v  P T

Consider now the molar Gibbs free energy g T , P  along the isotherm in Fig.3.9.

Since T  const , we have dg  vdP or

PX

gX  gA   v  P  dP
PA
(3.46)

 g A  I  X , A

PX

where I  X , A   v  P  dP . The values of g as X moves from A to I are shown in

PA

Fig.3.10. The salient features are as follows:

1. From A to D: g X  g A  I  X , A increases monotonically from g A to

g D  g A  I  D , A .
2. From D to F: g X  g D  I  X , D   g D  I  D, X  decreases monotonically

from g D to g F  g D  I  D, F  .

3. From F to I: g X  g F  I  X , F  increases monotonically from g F to

gI  gF  I  I , F  .

As shown in Fig.3.10, the segments AD and FI intersects in the g-P plane. Since the
equilibrium state has the lowest energy, the quasi-static isothermal path in the g-P
plane runs as A intersect I . Referring back to the v-P plane, we see that C
and G must be the position of this intersect on the segments AD and FI, respectively.

Now, the condition gC  gG means

g A  I  C , A  g F  I  G , F 

 g D  I  D, F   I  G , F 

 g A  I  D , A  I  D , F   I  G , F 

 I  C , A  I  D , A   I  D , F   I  G , F  (a)

Using

I  C , A  I  D , A   I  D , C 

and

I  D, F   I  D, E   I  E , F 

eq(a) becomes

I  D, C   I  D, E   I  E , F   I  G , F  (3.48)

i.e., Area 2  Area 1 [see shaded areas in Fig.3.9]

This is called the Maxwell construction.
3.D.4.a. Cubic Equation

The cubic equation

x 3  ax 2  bx  c  0
has 3 roots given by
1
x1  S  T  a
3
1 1 3
x2   S  T   a  i S  T 
2 3 2
1 1 3
x3   S  T   a  i S  T 
2 3 2
where

S  3 R  Q3  R2 T  3 R  Q3  R2

2 3
b a a b c a
Q   R    
3  3 3 2 2  3

The character of the roots is determined by the discriminant

D  Q3  R2
so that
D0  1 real & 2 complete conjugates
D0  all real with at least 2 equal
D0  all real & all equal
3.E. Superconductors
For some metals, the resistance drops to zero when the temperature falls below a
critical value TC . The metal is then said to be in the superconducting state.

Naively, one may be tempted to associate zero resistance with an infinite conductivity.
However, the superconducting state cannot be so described, as will be shown below.

Substituting the Ohm's law for a metal,

J  E (3.50)
into the Faraday's law
1 B
E   [Gaussian units] (3.51)
c t
we get
B c
  J
t 
Hence,
B
0 for   
t
i.e., B is time independent inside a metal of infinite conductivity, which leads to
hysteresis ( B depends on sample history) as shown in Fig.3.12.

However, the superconducting state was found experimentally to be a thermodynamic

state with perfect diamagnetism ( B  0 regardless of sample history). This is called
the Meissner effect.

According to the BCS theory, transition to the superconducting phase is a Bose-

Einstein condensation, in momentum space, of Cooper pairs. Here, a Cooper pair is
a boson representing a bounded state of 2 electrons of opposite spins interacting under
an attractive effective potential caused by electron- phonon interactions.

Thus, the superconducting state is a macroscopic quantum state. The onset of

superconductivity is an order- disorder phase transition with an effective wave

Here, ns  
2
function Ψ as order parameter. gives the density of

superconducting electrons. The vanishing of electrical resistance is due to the

ineffectiveness of the scattering offered by individual impurities against the entire
electron condensate.
Superconductivity can be destroyed by an applied field H  H C T  (see Fig.3.13).

Thus, H C  T  defines the coexistence curve in the H-T plane (see Fig.3.14). It was

found empirically that,

 T2 
H C T   H 0  1  2  (3.53)
 TC 

where H 0  H  0  . Note that

dH C 2 H 0T dH C
  0
dT TC2 dT
dH C
with  0.
dT T 0

Using
1
dg   sdT  BdH (3.57)
4
we have, along the coexistence curve,
1 1
 sn dT  Bn dH   ss dT  Bs dH (3.54)
4 4

 dH   sn  ss   sn  ss coex
    4    4 (3.55)
 dT  coex  Bs  Bn  coex H C T 

where we've used Bs  0 and Bn  H C T  on the coexistence curve.

Eq(3.55) is the Clausius- Clapeyron equation for superconductivity.

Using (3.53), eq(3.55) becomes

2 H 0T s  s 
  4 n s coex
TC2
H C T 

H 0T H 02  T 2 
  sn  ss coex  H C T   1  T
2 TC2 2 TC2  TC2 

so that
    sn  ss   H 02  T 2 2T 2 
 cn  cs coex  T   T 1  2  T 2 
T  coex 2 TC  TC
2
 C 

H 02  T 3T 3 
   3  (3.56)
2 TC  TC TC 

For sufficiently low T, the cubic term will be smaller than the linear one so that
cn  cs in the coexistence region. At T  TC , we have
H 02
 cn  cs T T 
C
 TC
Note also that

 sn  ss T 0  0
in agreement with the 3rd law.

Integrating (3.57) at fixed T gives

H
1
g  T , H   g T ,0    BdH (3.58)
4 0

Using Bn  H and Bs  0 , we have

1 2
g n  T , H   g n T ,0   H (3.59)
8

g s T , H   g s T ,0  for H  H C (3.60)

Finally, with the help that, on the coexistence curve,

gn T , H C   g s T , H C  (3.61)

we can combine (3.59-60) into

g s T ,0   g s T , H C   g n T , H C 

1 2
 g n  T ,0   HC (3.62)
8
Thus, the condensation energy is
1
g n  T ,0   g s  T ,0    H C  T 
2

8
with

g n TC ,0   g s TC ,0   0
3.F. Helium Liquids
3.F.1. Liquid He4
3.F.2. Liquid He3
3.F.3. Liquid He3-He4 Mixtures
3.F.1. Liquid He4
The phase diagram of He4 is shown in Fig.3.15 and should be compared with that of a
typical classical fluid shown in Fig.3.4. Salient points of interest are
1. The solid phase appears only for pressures above 25atm even as T  0 .
2. There are 2 liquid phases separated by a λ-line near T  2 K . A λ-line is a line
of λ-points, so called because the graph of C vs T near a λ-point looks like the
letter λ. (see Fig.3.16)
3. The high temperature liquid phase is called the He I or normal phase. Its
behavior is similar to a classical fluid.
4. The low temperature liquid phase is called the He II or superfluid phase. It’s
the first discovered example of a quantum fluid.
5. He II exhibits frictionless flow and can leak through cracks impermeable to He
gas. Hence the name superfluid. [cf. superconductor]
6. The He I-II transition is continuous with broken gauge symmetry. The order
parameter is the macroscopic superfluid wave function.
7. The slopes of g-l and s-l coexistence curves goes to zero as T  0 , in
accordance with the 3rd law.
3.F.2. Liquid He3
He3 has only 3/4 of the mass of He4. Hence,
3 4
TBoiling [ He3 ]  TBoiling [ He 4 ] PSolidification [ He3 ]  PSolidification [ He 4 ]
4 3
The phase diagram of He3 is shown in 2 scales in Figs.3.17-8.

The minimum in the l-s coexistence curve in Fig.3.17 is due to the spin contribution

to the entropy which makes S  Sliquid  S solid change sign at T  0.3K . Since

V  const along the coexistence curve, the minimum follows directly from the
 dP  T S
Clausius- Clapeyron equation    .
 dT  coex V

The superfluid phases (see Fig.3.18) appears only for T  0.0027 K so that they are
not discernible in Fig.3.17. Since He3 is a fermion, a (Bose-Einstein) condensation
into the superfluid phase requires a pairing mechanism. Depending on the angular
momenta of the bounded pairs, different superfluid phases results.
1. The high temperature A-phase is anisotropic.
2. The low temperature B-phase is more or less isotropic.
3. A third phase appears under magnetic field.
4. Transition between normal & superfluid is continuous.
5. Transition between the A and B phases is 1st order.
3.F.3. Liquid He3-He4 Mixtures
In a mixture of He3-He4 , one expects to find two types of phase transitions, namely,
the superfluid and the binary. This is shown in the phase diagram in Fig.3.19 with x3
indicating the molar fraction of He3.

The -line, extending from ( T  2.19 K , x3  0 ) to the tricritical point at ( T  0.87 K ,

x3  0.67 ), denotes the 2nd order superfluid phase transition.

The tricritical point is also the critical point of the 1st order binary phase transition.
Thus, below T  0.87 K , the mixture can segregate into coexisting He3 rich and He4
rich regions. This can be observed using NMR technique.
The tricritical point is so called because it is actually the intersect of 3 lines of 2nd
order transitions. [see R. B. Griffith, Phys. Rev. Lett. 24, 715 (1970)]
3.G. Landau Theory
1st order transition Continuous transition
st
1 derivatives of G Discontinuous at transition point Continuous
Order parameter Discontinuous across coexistent curve Continuous
(except at the critical point)
Symmetry May or may not be broken Aways broken

The Ginzburg-Landau theory is a mean field theory that describes an order-disorder

transition. Thus, the low temperature, or ordered, phase is characterized by a
nonzero order parameter η, which vanishes in the high temperature, or disordered
phase. In a continuous transition, there is also an accompanying change of
symmetries with the ordered phase having the lower symmetries. The molar free
energy φ is assumed to be analytic (i.e., a Taylor series expansion exists) so that near
the transition point where η is small, we can write

 T , Y ,    0 T , Y    2 T , Y  2   3 T , Y  3   4 T , Y  4 (3.63a)

where terms O  5  are neglected and we have set 1  0 so that the disordered

phase   0 is always a local minimum. In the presence of an external force

conjugate to η , (3.63a) is generalized to

 T , Y , f    T , Y ,   f 

  0 T , Y    2 T , Y  2   3 T , Y  3   4 T , Y  4  f  (3.63)

3.G.1. Continuous Phase Transitions

3.G.2. First Order Transitions
3.G.1. Continuous Phase Transitions
Since φ is a real scalar, if η is a complex number, vector, or tensor of odd rank, we

must have  j  0 for all odd j because any term containing odd powers of η can

never be a real scalar. In which case, eq(3.63a) simplifies to

 T , Y ,    0 T , Y    2 T , Y  2   4 T , Y  4 (3.64)

where, to ensure (formal) global stability, we must have

 4 T , Y   0 (3.66)

[  4  0 implies global minima of φ at    ]

For fixed Y and T, each equilibrium state (phase) is a minimum of φ so that

     2 
    0  2 0 (3.65)
 TY   TY

i.e.,

2 2  4 4 3  0 2 2  12 4 2  0 (3.65a)

There are 2 solutions to eq(3.65a), namely,

0 with  2  0 (3.65b)
and
2
2   with 2  0 (3.65c)
2 4
Obviously, (3.65b) and (3.65c) correspond to the disordered ( T  TC ) and ordered
( T  TC ) states, respectively. Thus, the transition, or critical, point satisfies

 2 T , Y   0 (3.65d)

For a given Y, let TC Y  be the solution of eq(3.65d), i.e.,

 2 TC Y  , Y   0 (3.65e)

Thus, TC Y  describes a coexistence curve in the Y-T plane. The conditions

(3.65b,c,e) are guaranteed if we write

 2  T , Y    0  T , Y  T  TC Y  (3.67)

where  0  0 is expected to be slowly varying in the neighborhood of the transition
point. Hence, eq(3.65c) becomes,

0
 TC  T  (3.69)
2 4

Putting everything into (3.64) gives

  T , Y 
 T  TC
 T , Y ,     02 for (3.70)
   T , Y   4 TC  T  T  TC
2

 4

Heat Capacity

Since φ is the Gibbs free energy, the molar heat capacity is given by,

  2 
cY  T  2  (3.71)
 T Y

so that

 cY 0
 T  TC
cY    02 for (3.71a)
c 
 Y 0 2 T  TC
 4

where

  2 0 
cY 0  T  2 
 T Y
Thus, there is a discontinuity at TC of magnitude
 02
cY  cY TC   cY TC   (3.72)
2 4
which gives the cY vs T plot a shape of a λ (see Fig.3.21) and hence the reason for
calling the critical point a λ-point.

Superfluid

The normal- superfluid transition of He 4 is continuous.

The order parameter is the macroscopic superfluid (complex) wave function  so
that

  T , P,     0  T , P    2    4 
2 4
(3.73)

with  2   T  TC   0 . Similar to (3.69), we have

 0 0
  TC  T     ei TC  T 
2

2 4 2 4

where the phase θ is a real number that can be set to zero when current flow is absent.

External Force

In the presence of an external force f that couples to the order parameter η, the
relevant free energy is [see (3.64)]

 T , Y , f    0 T , Y    2 2   4 4  f  (3.74)

which implies the Legendre transform (for equilibrium states),


 T , Y ,    T , Y , f   f
f

  T , Y , f   f 


where   . Typical plots of  can be found in Fig.3.22.
f

For fixed T,Y, and f, the equilibrium phases are minima of ψ so that

  
    2 2  4 4  f  0
3
(3.75)
 TYf

  2 
 2   2 2  12 4  0
2

  TYf

See Fig.3.22 for a few typical solutions. Note that (3.75) is simply the assertion that

  
for equilibrium states, f is indeed the force conjugate to η, i.e., f    .
  TY

More interesting is the susceptibility

     2 
f      2
 f TY  f TY

with the implicit assumption that all quantities take on their equilibrium values.
Taking the partial derivative of eq(3.75) gives

     
2 2    12 4 2   1  0
 f TY  f TY
 1
 f  (3.76)
2 2  12 4 2
For f  0 , eq(3.65b,c) gives

 0
 T  TC
 2 for (3.76a)
   2 T  TC
 4

so that with  2   T  TC   0 , we have

 1 1
 2  2  T  T   T  TC

 0  
2 C 0
for (3.77)
 1  1 T  TC
 4 2 4  TC  T   0

which exhibits a divergence at T  TC .

Paramagnetic-Ferromagnetic Transition

In a para- to ferro- magnetic transition, critical values are called Curie values.
The order parameter is the magnetization vector M.
The symmetry that is broken is the rotational symmetry. Thus,

 T , M    0 T    2M  M   4  M  M 
2

 T , H    0 T    2M  M   4  M  M   H  M
2
(3.78)

For H  0 , the analog of (3.76a) is

 0
 T  TC
M   ˆ 0 for
  M 2  TC  T  T  TC
 4

where M̂ is a unit vector.

The heat capacity exhibits the λ shape as shown in Fig.3.23.

3.G.2. First Order Transitions
If η is a scalar, or tensor of even rank, odd power terms of η can also be scalars if
properly constructed. Transitions in such systems are 1st order. For simplicity, we
shall assume η to be a scalar and write,

 T , Y ,    0 T , Y    2 2   3 3   4 4 (3.79)

where  4  0 for global stability. The equilibrium conditions are

  
    2 2  3 3  4 4  0
2 3
(a)
 TY

  2 
 2   2 2  6 3  12 4  0
2
(b)
  TY

Solutions to (a) are

0
and

2 2  3 3  4 4 2  0 (c)

 
1
8 4
 3 3  D  (d)

where the discriminant is D  9 32  32 2 4 .

As before, the disordered case   0 requires  2  0 or T  TC to satisfy (b).

If we follow the procedure of the last section, we can obtain from (d) an expression

for  for the ordered phase valid for  2  0 or T  TC . Obviously, this would be a

continuous phase transition.

On the other hand, as shown in Fig.3.24, we can adjust 3 to produce another
minimum at   D lower than that at   0 for the case  2  0 . Obviously,
transition to this new minimum will result in a discontinuous change in the slope of 
and hence signifies a 1st order phase transition. Furthermore, since we are still in the
region T  TC , this transition preceeds that for the continuous one at T  TC so that
the latter is never observed. Again with reference to Fig.3.24, the onset of this 1st
order transition (curve D) happens when both minima have the same value, i.e.,

 T , Y ,D    0 T , Y 

  2 2   3 3   4 4  0   2   3   4 2  0 (e)
Thus, the new phase is determined by conditions (b,c,e).
Now, 4(e)  (c) gives
2 2
2 2   3  0  
3
3
(c)  2(e)   3  2 4 2  0  
2 4
 32
Combining these 2 expressions, we also have 2  .
4 4
Combining (e) with condition (b) gives
5 2  3 3  0

or, simply,  2  0 . Since  4  0 , the sign of  is determined by that of 3 .

3.H. Critical Exponents
3.H.1. Definition Of Critical Exponents
3.H.2. The Critical Exponents For Pure PVT Systems
3.H.3. Exercise 3.4
3.H.1. Definition Of Critical Exponents
The "distance" from a critical point is usually measured in terms of the reduced
parameter
T  TC
 (3.80)
TC
The critical exponent λ of a thermodynamic function f is defined as

ln f   
  lim (3.82)
 0 ln 

which implies f can be written in the form

f     A  1  B y   y0 (3.81)

since
ln f   ln  as   0

Thus, as   0 ,

1.  0  f 

2.  0  f 0

3.   0 . A modified exponent λ' is defined as

ln f  j
 
 '  j  lim (3.83)
 0 ln 

d jf  j
where j is the smallest integer such that  f diverges.
d j

For example, if f    A  B y , we have

ln A ln A
  lim  lim 0
 0 ln  x   x

If y is a positive fraction, we have j  1 and f '  By y 1 so that

y 1
ln 
 '  1  lim y
 0 ln 

Another example is the logarithmic divergence f     A ln   B which gives

 ln ln  ln x x 1
  lim  lim  lim 0
 0 ln  x   x x  1

A
where we've used the L'Hospital rule. Since f '  , we have j  1 and

1
ln 
 '  1  lim 0
 0 ln 

Typical plots of f vs ε can be found in Fig.3.26.

3.H.2. The Critical Exponents For Pure PVT Systems
There are 4 critical exponents of interest.

(a) Degree of the Critical Isotherm, δ


P  PC   C
 A sign     C  T  TC (3.84)
PC0
C

where the subscript 0 denotes properties of the ideal gas.

Experimentally, 6    4 .

(b) Degree of the Coexistence Curve, β

l   g
 A   

T  TC (3.85)
C

where  l   g is the order parameter. Experimentally,   0.34 .

(c) Critical Exponent of the Heat Capacity, α, α'

 A '    ' T  TC

CV   for   C (3.86)
 A   

T  TC

Experimentally,    '  0.1 .

(d) Isothermal Compressibility, γ

 T  A '   
 '
T  TC
 for (3.87)
 T0  A    T  TC

Experimentally,  '  1.2 and   1.3 .

Exponent Inequalities

Using
 P  1  v  m m 
    and     2
 v T v T m T T T     T
where ρ the mass density and m is the molar weight, we can rewrite eq(3.40) as

x T   
2
x T   
2

cv  x g cvg  xl cvl  gg 03  g   ll 03  l  (3.88)

 T  g  T  coex  T  l  T  coex

where c now denotes specific heat (molar heat capacity /m). Since all terms on the
right are positive, we have

x T   
2

cv  gg 03  g  (3.89)
 T  g  T  coex

Now, when the critical point is approached from below, we have

cv    
 '
[from (3.86)]

  C

 T    
 '
[from (3.87)]

   l   g  
    
 1
 [from (3.85)]
 T  coex

Taking the square of the last gives

  l  g 
2

 T  T     
2  2

  coex

   
2

On the other hand,  l  g  is expected to be regular, [see (3.26)], so that

 T T  coex

1   l  g    l  g   1   l    g  
2 2 2 2

         
4  T T   T T   2  T   T  
  coex   coex

  
2

    
2  2

 T  coex

Putting everything into (3.89), we have

A   
2

cv  A1      A3   
 '  '  2  2
 2 (3.90)
 T  T coex

where the Aj's are constants. Since 0    1 , we have

ln      ln     0

Thus, the singular part of the logarithm of (3.90) gives

 ' ln       ' 2   2  ln   

  '  ' 2   2 (3.92)

which is known as the Rushbrook inequality.
3.H.3. Exercise 3.4
Compute α, β, δ and γ for a van der Waals fluid.

Answer

For convenience, we set

T  TC v  vC P  PC
  T 1   v 1   P 1
TC vC PC
so that the reduced van der Waals equation
 3 
 P  2   3v  1  8T
 v 
can be rewritten as

 3 
2 
  1  3  2   8   1 (1)
    1 
 
Solving for π gives

3 8    1
  1  
  1 3  2
2

Using

   1  3  2   2  7  8 2  3 3
2

3  3  2   9  6

8   1    1  8  16  8 2  8  16  8 2

2

we have
8  16  8 2  3 3
 (2)
2  7  8 2  3 3

The Degree of the Critical Isotherm δ

Rewriting (3.84) in the present notation gives


 1
   1   1   at   0
v

Eq(2) thus gives

3 3
  3
2  7  8 2  3 3
so that   3 .

The Isothermal Compressibility Exponent γ

Rewriting (3.87) in the present notation gives

 T    at 0

where
1
1  v   w    
T          
v  P T      

Using (2), we have

   16  16  9 2  8  16  8  3  7  16  9 

2 3 2

   
   2  7  8  3  2  7  8 2  3 3 
2 3 2

 8  14  6 for   0

Hence,  T   1 so that   1 .

The Degree of the Coexistence Curve β

Rewriting (3.85) in the present notation gives

1 1 v  v   l
 A   

l   g    g l  g
vl v g vl v g vl v g
or

g  l    

(a)

along the coexistence curve. Now, the intercepts (points C and G in Fig.3.9) of an
isotherm T0  TC with the coexistence curve is given by the Maxwell construction as
G

 vdP  0
C
(6)

where the integral is along the isotherm T  T0 . In terms of the present notation, we
have
G

 1    d  0
C
(6a)

T0
where π is given by (2) with    0  1 . In the vicinity of the critical point, (2)
TC
can be written as
 1
 3  7 3 
   4  8  4 2   3   1    4 2   3 
 2  2 2 

 3   7  49   343 3  3 
  4  8  4 2   3  1      4   2        
 2   2  4   8 2 
3 7 33
 4  8  4 2   3    4  8    2  4   
2 2 4
3
 4  6  9 2   3   (5)
2

where only terms  n m with n  m  3 are kept. For fixed    0 , we have

 9 
d   6 0  18 0   2    d  (5a)
 2 
so that (6a) becomes
G
 9 
 1     6
C
0  18 0   2    d   0
2 

G
 9 
   6
C
0  12 0   2    d   0
2 

G
 3 3 
 6 0  6 0  2     0
2

6 0 g  l   6 0 g2  l2   g  l3     0

3 3
(9)
2

where, by definition,   g at G and   l at C [see Fig.3.8-9].

Now, both G and C are at the same temperature and pressure so that (5) gives
3 3
4 0  6 0g  9 0g2  g3    4 0  6 0l  9 0l2  l3  
2 2

6 0 g  l   9 0 g2  l2    g  l3     0

3 3
 (8)
2
This is compatible with (9) only if

9 0 g2  l2   6 0 g2  l2 

 g2  l2 (8a)

 vg  vC vl  vC
Since g   0 and l   0 , eq(8a) implies
vC vC

g  l  0 (8b)

Either (8) or (9) then gives

12 0g  3g3  0

 g  0 or g2  4 0

Thus, slightly away from the critical point, we have

g  2  0 (11)

1
where  0  0 since T0  TC . Comparing with condition (a) gives   .
2

The Heat Capacity Exponent α

Rewriting (3.86) in the present notation gives

   '  0
cv   for at 0
   

 0

Now, as one approaches the critical point along the isochore   0 , one gets [see
(3.88)],
 x gT0   g 
2
xlT0   l 
2

 x g cvg  xl cvl  g 3    0
cv    T  g  T  coex  Tl  l3  T  coex for
  0
 cvg

Very close to the critical point, the liquid and vapor phases merges so that
  T0    2 
cv   3      0
cv   g   T   T   coex for
   0
 cvg

Thus, there is a discontinuity at the critical point,

 T0    2 
  cv  T   cv T   
 
3    with v  vC
  T   T   crit
C C

Now,
  C  C   1  C   1 
      
T TC T TC T  v  TC     1 
 C  C 2
 
TC   1  TC   1 
2 2

where eq(11) was used to obtain the last equality. Thus

2
  1  2
3
1    C
2
4
3      
  T   C3  TC   1  
2
T  C   1  2
C
2

Similarly,
1  v  1  v 
T        
v  P T PC v  P T
1
1    1   
     
PC   1    PC   1   

Using (5a), we have

   9 2
   6  18    
   2
so that
1  9 
 PC 1     6  18   2   
T  2 
Putting everything together, we have

 4  9 
  TC    1 2 2 C 
P 1     6  18   2    
 TC  C   1   2   crit

where crit means   2  with   0 . Thus

4 PC    9 
  6 2  18    
TC  C    2 
4 PC  3 9 
    9     
TC  C  2 2 
PC 3
Using  R , we have
TC  C 8
9
 R (14)
2
Since the jump is finite, we have    '  0 .