You are on page 1of 15

Chapter 6

Phase transitions

6.1 Concept of phase

Consider a system described by the variables P and T . For a given P and T , the equilib-
rium state of the system does not necessarily have to be homogeneous. The system can
show different state forms (f. i., for water the possible state forms are liquid/solid/gas).
These state forms are called phases. In each phase certain macroscopic properties can
show very different values.

Example: water.

Figure 6.1: P − T diagram for water.

The P − T diagram for water shows phase boundaries, or lines of coexistence.

55
56 CHAPTER 6. PHASE TRANSITIONS

Critical point: at this point, the gas and the liquid have equal densities and specific
entropies (entropy per particle). The phase transition becomes
second-order at this point.

⋆ There is no critical point for the liquid-solid transition.

Triple point: at this point (line), gas, liquid and solid coexist.

Note: the melting curve has negative slope, which means that at constant temperature
T < T0 ice melts under increasing pressure (we observe this during ice-skating).

6.2 First-order phase transition


Since we have already started discussing the {P, T } phase diagram for water, let us
continue and concentrate first on phase transitions of first order.

When water starts boiling, it undergoes a phase transition from a liquid to a gas phase.
For both phases independently, the equation of state is a well-defined regular function,
continuous, with continuous derivatives. However, while going from liquid to gas one
function ”abruptly” changes to the other function. Such a transition is the first-order
phase transition.
The first-order phase transition can be defined more systematically by considering the
Gibbs thermodynamic potential. In a ”first-order phase transition”, the first derivative of
the Gibbs potential is discontinuous across the phase boundary. Also, according to the
relations:    
∂G ∂G
V = and S = − ,
∂P T ∂T P
volume V and entropy S are discontinuous as well.

Let us consider an isotherm drawn in the P − V diagram (Fig. 6.2).

Figure 6.2: An isotherm in the P − V diagram for water.


6.3. CONDITION FOR PHASE COEXISTENCE 57

This P − V diagram is a projection onto P − V of the equation of state surface for water
(Fig. 6.3).

Figure 6.3: The equation of state surface for water.

Compare Figure 6.3 with the cut P − T shown in Fig. 6.1.

Important properties: latent heat.


⇒ Since the coexisting phases have different entropies, the system
must absorb or release heat during a phase transition.
When an amount of phase 1 is converted into phase 2, an amount of latent heat ∆QL is
released:
l = T0 (s2 − s1 ), (6.1)
∆QL = lN = T0 (S2 − S1 ). (6.2)
In eq. (6.1) and (6.2), T0 is the absolute temperature at which the two phases coexist,
si is the entropy per particle of the ith phase (specific entropy) and N is the number of
particles that have undergone the transition.

6.3 Condition for phase coexistence


We consider a system composed by one species of particles at given conditions of P and
T (constant). Let us assume coexistence of two phases in our system. In this case, the
Gibbs potential is the sum of Gibbs potentials of the two phases:
G(T, P ) = G1 (T, P ) + G2 (T, P )
= N1 µ1 (T, P ) + N2 µ2 (T, P ),
where Ni and µi are the number of particles and the chemical potential in phase i. The
condition of equilibrium is that the Gibbs potential has to be at its minimum, i.e., dG = 0,
58 CHAPTER 6. PHASE TRANSITIONS

with N = N1 + N2 fixed. In other words, in equilibrium, G is at a minimum with respect


to particle transfer from one phase to the other. Therefore,

dG = µ1 dN1 + µ2 dN2 = 0,

with dN1 = −dN2 .


→ µ1 (T, P ) = µ2 (T, P ) (6.3)
The condition of coexistence, eq. (6.3), tells us that both phases must have equal chemical
potential (Fig. 6.4).

Figure 6.4: The equilibrium condition, µ1 (T, P ) = µ2 (T, P ), in the G − T diagram.

The derivatives of G are discontinuous across the phase transition:


∂(∆G)
∆S = S2 − S1 = − >0
∂T
∂(∆G)
∆V = V2 − V1 = >0
∂P

6.4 Clapeyron equation


Since the variables ∆G, ∆S, and ∆V are functions of V , T , and P , there must exist a
relation among them:
f (∆G, ∆S, ∆V ) = 0.
6.4. CLAPEYRON EQUATION 59

Let us derive such a relation. We apply the chain rule:


     
∂(∆G) ∂T ∂P
= −1 → (6.4)
∂T P ∂P ∆G ∂(∆G) T
 
∂(∆G)  
∂T ∂P
→  P =− .
∂(∆G) ∂T ∆G
∂P
T
∂P

In equilibrium, ∆G = 0 and ∂T ∆G= 0
gives the slope of the transition line in the P − T
diagram, i.e.,
 
dP ∂P
≡ . (6.5)
dT ∂T ∆G=0

P is called vapor pressure (Dampfdruck) for the gas-liquid transition.

We substitute eq. (6.5) into eq. (6.4) and obtain

dP ∆S
=
dT ∆V
or
dP ∆QL
= → Clapeyron equation
dT T ∆V
where ∆QL = T ∆S is the latent heat.

Depending on the sign of ∆V , the slope dP


dT
may be positive or negative. Figure 6.5 shows
the P − T diagram for CO2 , which contracts upon freezing.

Figure 6.5: The slope of the liquid(2)-solid(1) boundary is positive ⇒ the liquid contracts
upon freezing.
60 CHAPTER 6. PHASE TRANSITIONS

Figure 6.6 shows the P − T diagram for water, which expands upon freezing (due to the
Hydrogen bonding between molecules).

Figure 6.6: The slope of the liquid(2)-solid(1) boundary is negative ⇒ the liquid expands
upon freezing.

In a ”second-order” phase transition the first derivatives of G vanish, and the Clapeyron
equation is replaced by a condition involving second derivatives.

6.5 Ehrenfest classification of phase transitions


For the following discussion, let us denote the two phases in equilibrium at a given coex-
istence curve as α and β. Then, following Ehrenfest, we define the ”order of the phase
transition” as
the order of the lowest derivative of G, which shows a discontinuity
upon crossing the coexistence curve.

Explicitly, a phase transition between phases α and β is of order n if


1)
∂ m Gα
   m 
∂ Gβ
m
= for m = 1, 2, . . . , n − 1,
∂T ∂T m
 m P  m P
∂ Gα ∂ Gβ
m
= for m = 1, 2, . . . , n − 1;
∂P T ∂P m T

2)
∂ n Gα
   n 
∂ Gβ
n
6=
∂T ∂T n
 n P  n P
∂ Gα ∂ Gβ
n
6=
∂P T ∂P n T
6.5. EHRENFEST CLASSIFICATION OF PHASE TRANSITIONS 61

In practice, only phase transitions of first- and second -order are of importance. Their
properties are listed below.

1st order:

1) G(T, P ) continuous;
   
∂G ∂G
2) S = − and V = discontinuous;
∂T P ∂P T

3) ∃ latent heat.

2nd order:

1) G(T, P ) continuous;

2) S(T, P ) and V (T, P ) continuous;

3) response functions (susceptibilities) discontinuous:


 2 
∂ G
CP = −T ,
∂T 2 P

1 ∂ 2G
 
κT = − ,
V ∂P 2
 2 
1 ∂ G
β= .
V ∂P ∂T

Figure 6.7: Second-order phase transition following Ehrenfest.

There are not many phase transitions of the second order that, following Ehrenfest classifi-
cation, show a finite discontinuity in the specific heat. A superconductor at zero magnetic
field is an example of such 2nd -order phase transitions (Fig. 6.8).
62 CHAPTER 6. PHASE TRANSITIONS

Figure 6.8: Second-order phase transition in a superconductor at zero magnetic field.

Note: in practice, though, many systems with phase transitions which are not of the first
order show divergences (not finite discontinuities) in the response functions. For instance,
the susceptibility of a magnetic system is of the form
 
∂M 1
χ= , χ(T ) ∼ ,
∂H T (T − Tc )γ

with γ being called critical exponent. For systems with such properties, the Ehrenfest
classification is not anymore valid.

For a more general classification, one considers only the behavior of the entropy at the
phase boundary as a function of T . Within this classification, discontinuous and contin-
uous phase transitions are distinguished as discussed below.
1) discontinuous phase transitions are equivalent to phase transitions of the 1st order.

Properties:

(i) ∆S 6= 0; ∃ latent heat;


 2 
(ii) CP = −T ∂∂TG2 is finite for T 6= T0 ; no condition exists for T = T0 .
P

2) continuous phase transitions have the following properties:

(i) S continuous ⇒ no latent heat;


(ii) ∃ critical point Tc ;
(iii) singularities in CV , κT , χT
6.5. EHRENFEST CLASSIFICATION OF PHASE TRANSITIONS 63

Consider a fluid system whose phase transition is depicted in Fig. 6.9. Since, as seen from

Figure 6.9: ρ − T diagram for a fluid system.

Fig. 6.9, V = ∂G

∂P T0
shows a finite discontinuity, the phase transition is of 1st order. If
we now increase temperature, ∆ρ diminishes (Fig. 6.10) and at Tc the transition is not

Figure 6.10: ∆ρ − T0 diagram.

anymore discontinuous (1st order) but continuous (2nd order).

This phenomenon happens for many systems. As another example, let us also consider a
magnet (Fig. 6.11).
Since  
∂G
M =− , B 0 = µ0 H
∂B0 T0
is discontinuous, this transition is of 1st order. However, as in the previous example, by
increasing temperature the phase transition becomes second-order at T0 = Tc (Fig. 6.12).
64 CHAPTER 6. PHASE TRANSITIONS

Figure 6.11: Phase transition of a magnet.

Figure 6.12: MS − T0 dependence for a magnet.

6.6 Van der Waals equation of state


The equation of state for an ideal gas,

P V = N kB T // P V = nRT,

is only valid for very small densities of particles and, therefore, cannot describe the gas-
liquid phase transition. This phase transition is due to intermolecular interactions. In
this section, we will derive an equation of state for a gas which includes intermolecular
interaction in a phenomenological way.

For that purpose, we consider the following ansatz:

Pef f Vef f = nRT.

In order to derive explicit expressions for Vef f and Pef f , we assume that the potential
energy U (r) between two molecules as a function of their separation r0 has the form as
shown in Figure 6.13.
In particular,

- U (r) has a repulsive core with radius r0 (a few angstroms) due to the electrostatic
repulsion of the electron cloud;

- U (r) has an attractive tail at r > r0 due to the mutual electrostatic polarization of
two interacting molecules.
6.6. VAN DER WAALS EQUATION OF STATE 65

Figure 6.13: Intermolecular potential.

The depth of attractive U (r) is about 1 eV (changes with the gas species). This minimum
is responsible for the chemical valence and for the crystal structure of solids.
In order to consider these effects qualitatively, we separate them.
a) The hard core excludes a certain volume around a molecule so that other molecules
have less room in which to move. As a consequence, the effective volume of the gas
is reduced:
Vef f = V − bn .
V : total volume of the system;
b: total excluded volume, b ≈ N πr03 /6;
n: number of moles.
b) The pressure of the gas comes from the molecules hitting the walls of the container.
A real gas hits the walls with less kinetic energy than an ideal gas since in the real gas
the molecules are held back by the attraction of neighboring molecules (attractive
tail) (Fig. 6.14). The reduction in the pressure is proportional to the number of

Figure 6.14: Attraction between molecules in a real gas.

pairs of interacting molecules near the wall:


a 2
P = Pef f − n.
V2
A particle at the wall experiences a resulting force directed towards the container’s
interior, which reduces the pressure of the gas on the container’s wall,

Pef f > ”wall pressure” P.


66 CHAPTER 6. PHASE TRANSITIONS

Van der Waals made an assumption that, for n mole of gas,

Pef f Vef f = nRT,


  n 2 
(V − bn) P + a = nRT . (6.6)
V

6.6.1 Virial expansion


The van der Waals equation of state approaches that of the ideal gas in the low density
limit:
V → ∞, for fixed N .
The successive corrections to the ideal gas equation are obtained by expanding in powers
of V −1 . We define
an2 = A, B = bn, and R̄ = nR.
Then, from eq. (6.6),  
A R̄T
P+ 2 = ,
V (V − B)
 −1
R̄T B A
P = 1− − 2,
V V V
 −1
PV B A
= 1− − . (6.7)
R̄T V R̄T V
After expanding the first term in (6.7),
 −1  2
B B B
1− =1+ + + ...,
V V V
for large V , we get
    2  3
PV 1 A B B
=1+ B− + + + ... (6.8)
R̄T V R̄T V V
This expression has the form of a virial expansion:
PV C2 C3
=1+ + 2 + ..., (6.9)
R̄T V V
where Cn is the nth virial coefficient. Comparing eq. (6.8) and (6.9), we have

A
C2 = B − .
R̄T
This coefficient can be obtained experimentally by observing deviations from the ideal
gas law. From measuring these deviations at varying temperature, one can extract the
parameters A and B.
6.6. VAN DER WAALS EQUATION OF STATE 67

Figure 6.15: Isotherms for the van der Waals system

6.6.2 Critical point


The van der Waals isotherms have the shape given in Fig. 6.6.2.
The isotherms are solutions of the van der Waals equation:

an2 n3
 
3 2 nRT
V −V nb + +V − ab = 0 . (6.10)
P P P

Eq. (6.10) is a polynomial of the 3rd order and, hence, for given T and P there would
exist three real roots or only one real root. For T < Tc and P < Pc , we have three real
roots. As we increase T , these roots move closer and merge at T = Tc , the critical point.
For T > Tc , there is only one real solution and two complex conjugate ones.

At the critical point (Pc , Tc , Vc ), the equation of the state must be

(V − Vc )3 = 0

since all three solutions are the same. Then,

V 3 − 3Vc V 2 + 3Vc2 V − Vc3 = 0 . (6.11)

By compare eq. (6.10) with eq. (6.11), we find Vc , Pc , and Tc :


nRTc  solution :

3Vc = nb + 
Pc   Vc = 3bn

2

2 an a
3Vc = P c =
Pc  27b2
abn 3 8a


Vc3 =  RTc =


Pc 27b
From this solution we can define the parameter

Pc V c 3
Zc = = .
nRTc 8
68 CHAPTER 6. PHASE TRANSITIONS

Experimentally, one finds that almost all real gases have Zc < 83 , while for the ideal gas
Zc = 1. Therefore, the van der Waals equation is an improvement with respect to the
ideal gas equation.

6.6.3 Maxwell construction


The van der Waals isotherm is a monotonic function of V for T > Tc . For T < Tc , it has
a ”kink” with negative compressibility:
 
1 ∂V
κT = − <0 →
V ∂P T
 
∂P
→ > 0.
∂V T

Negative compressibility is not realistic since a decrease of volume dV < 0 would mean
a decrease of pressure dP < 0, which does not make sense since such a behavior leads to
the collapse of the system. This unphysical result originates from the implicit assumption
that the density of the system is always uniform. In fact, it is possible to show (and we
are going to do this now) that the system at T < Tc prefers to undergo a first-order phase
transition by breaking up into a mixture of phases of different densities.

Solution: according to the Maxwell relation,


 
∂F
P =− ,
∂V T

where F is the Helmholtz free energy, the free energy can be obtained as the area under
the isotherm: Z
F (V, T ) = − P dV.
isotherm

We perform the integration graphically (Fig. 6.16).


The volumes V1 and V2 are defined by the double tangent construction. At any point
along the tangent, like X, the free energy is a linear combination of the free energy at
1 and 2 and, therefore, represents a mixture of two phases. This non-uniform state has
the same P and T as the uniform state 3, but it has lower free energy as it is seen from
Figure 6.16.

The phase-separated state is the equilibrium state; the states 1 and 2 are defined by the
conditions:
two phases in equilibrium ⇒ G1 (T, PV ) = G2 (T, PV ),
where PV is the ”vapor pressure”;

⇒ U1 − T S1 + PV V1 = U2 − T S2 + PV V2
⇒ F1 − F2 = PV (V2 − V1 ).
6.6. VAN DER WAALS EQUATION OF STATE 69

Figure 6.16: Maxwell construction.

For T = const,
dF = −P dV,
Z V1 Z V2
F1 − F2 = − dV P (V ) = dV P (V ) ⇒
V2 V1
Z V2
⇒ P dV = PV (V2 − V1 ) .
V1

This means that the areas A and B are equal to each other. The horizontal line in Fig.
6.16 (F, V ) is the Maxwell construction.

You might also like