Chapter 5

Thermodynamic potentials

Thermodynamic potentials are state functions that, together with the corresponding equa-
tions of state, describe the equilibrium behavior of a system as a function of so-called
natural variables. The natural variables are a set of appropriate variables that allow to
compute other state functions by partial differentiation of the thermodynamic potentials.

5.1 Internal energy U

The basic relation of thermodynamics is given by the equation

m
X
X
dU = T dS + Fi dqi + j dNj , (5.1)
i=1 j=1

where {F, q} denote the set of conjugate intensive and extensive variables that characterize
a system. For instance, for a gas

{F, q} {P, V },

for a magnetic system

{F, q} {B, M}.
In (5.1), N is the number of particles in the system (an extensive variable); the index j
(j = 1, . . . , ) denotes different sets of particles that may constitute the system.
Chemical potential (an intensive variable): is defined as the energy needed to add a
particle to a thermally and mechanically isolated system.
The last term in eq. (5.1), dN , is needed if the number of particles in the system is not
kept constant, i.e. if particles enter or leave the system.

For a gas, eq. (5.1) is written as

X
dU = T dS P dV + j dNj ,
j=1

39
40 CHAPTER 5. THERMODYNAMIC POTENTIALS

which means that

U = U (S, V, N ) .
Since dU is a total differential, through differentiation of U as a function of S, V , and N
one obtains thermal and caloric equations of state:
 
U
T = ,
S V,N
 
U
P = ,
V S,N
 
U
j = .
Nj S,N,Ni,i6=j

The experimentally important response functions are obtained by second-order differen-

tiation:
 2    "  #1
U T S T
2
= = =
S V,N S V,N T V,N CV

"  #1
2U
CV = T ,
S 2 V,N

   
2U P 1
= =
V 2 S,N V S,N V kS

"  #1
1 2U
kS = . (5.2)
V V 2 S,N

Maxwell relations
A Maxwell relation follows from the differentiability of U :
   
U U
=
V S S V
   
T P
= (Maxwell relation). (5.3)
V S S V

Example: monoatomic ideal gas. The behavior of a monoatomic ideal gas obeys the
following relations:
P V = nRT,
5.1. INTERNAL ENERGY U 41

3
U = nRT,
2
3
CV = nR.
2
Let us use this information and derive an expression for U in terms of its natural variables:
dU + P dV dT dV
dS = = CV + nR ,
T T V
   
T V
S(T, V ) = CV ln + nR ln + S0
T V
 0  0
U V
= CV ln + nR ln + S0 ,
U0 V0
   
S S0 U nR V
= ln + ln . (5.4)
CV U0 CV V0
Since
nR = CP CV
nR CP CV 5 2
= =1= 1= ,
CV CV 3 3
we can write eq. (5.4) as
   CnR
SS0 U V V
e CV
= ,
U0 V0
from which
 1
V0 SS0
U (S, V ) = U0 e CV
(5.5)
V
follows. Eq. (5.5) is the fundamental equation for the ideal gas, with U (S, V ) as the
thermodynamic potential and S, V as independent natural variables.

Corollary: the natural variables for U are S and V , which means that if the function
U (S, V ) is known for a given system we can obtain all the thermodynamic properties of
the system through the differentiation of U (S, V ).

On the contrary, in the equation of state

U = U (T, V, N ) (5.6)

U is not an appropriate thermodynamic potential any more since from the first derivatives
of eq. (5.6),  
U
= CV
T V
and    
U P
=T P,
V T T V
42 CHAPTER 5. THERMODYNAMIC POTENTIALS

we do not obtain the dependent variables S and P .

If we rewrite eq. (5.1) as

1 1 1X
dS = dU + P dV i dN ,
T T T j=1

it becomes clear that

S = S(U, V, N )

is also a thermodynamic potential.

Note the analogy between classical mechanics and thermodynamics:

Classical mechanics Thermodynamics

Potential V (x, y, z) U (S, V, N )

Independent variables x, y, z S, V, N

V
Dependent variables Fx = ,... Maxwell relations:
x  
U
obtained differentiating P = ,...
V S

5.2 Legendre transformation

A disadvantage of using U (S, V, N ) as a thermodynamic potential is that the natural
variable S is difficult to control in the lab. For practical purposes, it is more convenient
to deal with other thermodynamic potentials that can be defined by making use of the
Legendre transformation.

A thermodynamic potential for a system with variable number of particles should depend
on {, N } as well as a thermal variable and a mechanical variable, which can be (for a
gas):

thermal: S (extensive) T (intensive)

mechanical: V (extensive) P (intensive)

5.3. HELMHOLTZ FREE ENERGY 43

There are four possible combinations of these variables:

Intensive variables Extensive variables

P Enthalpy H(S, P ) S

Gibbs enthalpy Internal energy

G(T, P ) U (S, V )

T Free energy F (T, V ) V

Legendre transformation (Reminder):

Consider a function f = f (x) with the differential
df
df = dx = udx.
dx
We want to find a function g = g(u) such that
dg
= x.
du
Function g(u) can be found in the following way:

df = udx = d(ux) xdu

d(f ux) = xdu

d
(f ux) = x.
du
We define
df
g(u) = f (x) ux = f (x) x
dx
as the Legendre transformation of f (x).

5.3 Helmholtz free energy

By replacing the independent variable S by T in U (S, V, N ), we define the free energy F :
 
U
F =U S = U TS ,
S V,N

F = F (T, V, N ) Free energy.

44 CHAPTER 5. THERMODYNAMIC POTENTIALS

In
dF = dU d(T S) = dU SdT T dS
we substitute dU from eq. (5.1) and get

X
dF = SdT P dV + j dNj .
j=1

The natural variables of the free energy are {T, V, N }. The dependent variables are
obtained out of the first derivatives:
   
F F
S = , P = thermal state equations
T V,N V T
and    
S P
= Maxwell relation.
V T T V

5.4 Enthalpy
Enthalpy
H = H(S, P, N )
is obtained when we substitute the variable V by P in U (S, V, N ):
H = U + PV .
The total differential of H is derived as follows.
 
U
H =U V ,
V S,N
| {z }
P

dH = dU + d(P V ) = dU + V dP + P dV
X
= T dS P dV + j dNj + V dP + P dV
j=1

X
dH = T dS + V dP + j dNj .
j=1

The natural variables of the enthalpy are {S, P, N }. Out of the first order derivatives we
obtain  
H
T = ,
S P,N
 
H
V = ,
P S,N
and the Maxwell relation    
T V
= .
P S,N S P,N
5.5. GIBBS ENTHALPY 45

5.5 Gibbs enthalpy

We obtain the Gibbs enthalpy
G = G(T, P, N )
by replacing S by T and V by P in U (S, V, N ):
   
U U
G=U S V
S V,N V S,N

G = U TS + PV .
The total differential of the Gibbs enthalpy:
dG = dU T dS SdT + P dV + V dP,

X
dG =
T dS PdV

+ j dNj SdT +
T dS
PdV

+ V dP,
j=1
P
dG = SdT + V dP + j=1 j dNj .
Out of the first derivatives of G, we obtain
 
G
S= ,
T P,N
 
G
V = ,
P T,N
and the Maxwell relation    
S V
=
P T,N T P,N

5.6 Homogeneity relations

The thermodynamic potentials vary as extensive state properties if the volume and num-
ber of particles vary by a factor , i.e.,

V V
U U. (5.7)
Nj Nj
The property (5.7) is a true one only if we neglect interactions between particles as
well as surface effects.

Since dU is extensive and T is intensive, from

X
dU = T dS P dV + j dNj
j=1

we see that the other involved variables are:

46 CHAPTER 5. THERMODYNAMIC POTENTIALS

S, dS: extensive
j : intensive
P: intensive
V, dV : extensive
It follows then that dF , dG, and dH, i.e., F , G, and H are all extensive state properties,
and the following homogeneity relations should be fulfilled:

F (T, V, N ) = F (T, V, N ),
H(S, P, N ) = H(S, P, N ),
G(T, P, N ) = G(T, P, N ). (5.8)

Let us now use the extensivity of G. If we differentiate eq. (5.8) with respect to and
set = 1:
d
G(T, P, N ) = G(T, P, N )|=1
d !
 
X G
= Nj ,
j=1
(Nj ) T,P,Ni6=j
=1

we obtain:

X
G(T, P, N ) = j Nj Gibbs-Duhem relation. (5.9)
j=1

When there is only one class of particles ( = 1),

G(T, P, N ) = N .

The chemical potential can be interpreted as Gibbs enthalpy per particle. is an

intensive variable.
We can use the definition of G(T, P, N ) as the Legendre transform of U ,

G = U T S + P V,

and eq. (5.9) to get the following relation:

P
U TS + PV j=1 j Nj = 0 .

5.7 Grand canonical potential

In a system that interchanges particles with a reservoir, the chemical potential is constant
and the number of particles changes accordingly. In order to describe such a system,
we introduce the grand canonical potential (T, V, ), where the natural variable N is
replaced by :
(T, V, ) = F (T, V, N ) N .
5.8. EQUILIBRIUM CONDITIONS 47

We define as the Legendre transform of F , with

 
F
(T, V, N ) = .
N T,V
The total differential of is then
d = dF dN N d
d = SdT P dV N d .
Through differentiation of we can calculate S, P and N :
 

S= ,
T V,
 

P = ,
V T,
 

N = .
T,V
Since G = N and G = F + P V ,
= F N = F G = P V,
from which follows

= P thermal equation of state.
V

5.8 Equilibrium conditions

An important property of the thermodynamic potentials is that, by keeping constant
certain variables and changing others, one can calculate how the energy exchange with
the environment happens.

Example: internal energy of a gas.

a)S = const : dU = P dV (work)
dU = T dS P dV
b)V = const : dU = T dS (heat)
The second law of thermodynamics,

X
T dS Q = dU + P dV j dNj , (5.10)
j=1

can be written in very simple forms for the various exchanges between the system and its
environment by using various thermodynamic potentials. The thermodynamic potentials
allow for the description of the development of a system towards equilibrium state itself.
Depending on a specific experimental situation, an adequate potential is to be used.

Let us now consider several examples.

48 CHAPTER 5. THERMODYNAMIC POTENTIALS

5.8.1 Isolated system

In an isolated system

dU = 0 (Q = 0),
dV = 0,
dN = 0.

Then, according to eq. (5.10),

dS 0
dS = 0 in equilibrium.

In all irreversible processes, under the conditions

U = const,
V = const,
N = const,

the entropy increases and is maximum in the stationary equilibrium. S evolves to

a maximum under conditions of fixed U , V , N . This is an extremum principle.

Consider two isolated systems 1 and 2 separated by a wall (Fig. 5.1).

Figure 5.1: Isolated systems 1 and 2.

The wall is movable and allows exchange of energy and particles. V1 , V2 , U1 , U2 , N1 , and
N2 are variable but must obey the following constraints:

U = U1 + U2 = const dU1 = dU2 ,

V = V1 + V2 = const dV1 = dV2 ,

N = N1 + N2 = const dN1 = dN2 .

The total entropy of 1 and 2 is a sum of their individual entropies:

S = S(U1 , V1 , N1 ) + S(U2 , V2 , N2 ) = S1 + S2 .
5.8. EQUILIBRIUM CONDITIONS 49

Both systems will interchange particles and energy until equilibrium is attained. In the
equilibrium state

0 = dS = dS1 + dS2
(    )
S1 S2
= dU1
U1 V1 ,N1 U2 V2 ,N2
(    )
S1 S2
+ dV1
V1 U1 ,N1 V2 U2 ,N2
(    )
S1 S2
+ dN1
N1 U1 ,V1 N2 U2 ,V2
   
1 1 P1 P2
= dU1 + dV1
T T2 T1 T2
 1 
1 2
+ + dN1 .
T1 T2

Since U1 , V1 , and N1 are independent variables, each curly bracket {} has to be zero.
Therefore,
T1 = T2 = T ; P1 = P2 = P ; 1 = 2 = .

In an isolated system in equilibrium everywhere there is the same temperature,

pressure and chemical potential.

5.8.2 Closed system in a reservoir without work interchange

We consider now a closed system in a reservoir without work interchange. Such a system
has the following features according to the attributes that we gave it:

closed Nj = const dNj = 0,

in a reservoir T = const dT = 0,
without work interchange V = const dV = 0.

According to the second law of thermodynamics,

T dS dU,

if
T = const,
then
T dS = d(T S) d(U T S) 0 .
50 CHAPTER 5. THERMODYNAMIC POTENTIALS

The latter expression tells us that in this case the free energy F ,

F = U T S,

is the appropriate potential.

dF 0
dF = 0 in equilibrium

In all irreversible processes, under the conditions

T = const,

V = const,
N = const,
the free energy decreases and reaches a minimum at equilibrium.

Consider a composite system in a heat reservoir at temperature T (Fig. 5.2), whose

particles can pass through the wall.

Figure 5.2: System in a heat reservoir.

It obeys such conditions:

V = V1 + V2 = const dV1 = dV2 ,

N = N1 + N2 = const dN1 = dN2 ,

F = F (T, V1 , N1 ) + F (T, V2 , N2 ) = F1 + F2 .
In equilibium,

0 = dF = dF1 + dF2
(    )
F1 F2
= dV1
V1 N1 ,T V2 N2 ,T
(    )
F1 F2
+ dN1
N1 V1 ,T N2 V2 ,T
= {P1 + P2 } dV1 + {1 2 } dN1 . (5.11)
5.8. EQUILIBRIUM CONDITIONS 51

From eq. (5.11) follows that, since V1 and N1 are independent variables,

P1 = P2 = P,

1 = 2 = .

In a closed system with constant volume V in contact with a heat bath at constant
temperature T , the equilibrium state is characterized by the same pressure
and chemical potential everywhere.

One more example: closed system in a heat reservoir at constant pressure P (will be
worked out in the exercises).

dT = 0, dNj = 0, dP = 0

dG 0,
dG = 0 in equilibrium.

The Gibbs energy decreases in an irreversible process under the above conditions
and reaches its minimum at equilibrium. In this case, the equilibrium state is
characterized by the same chemical potential everywhere.
52 CHAPTER 5. THERMODYNAMIC POTENTIALS

5.9 Summary of thermodynamic potentials

Potential Natural Conjugated Maxwell relations

independent dependent and others
variables variables
 
U
Internal energy T =
S
 V,N    
U T P
U S, V , N P = =
 V S,N V S S V
U
=
N S,V
 
1 S
Entropy =
T  U V,N
P S
S U, V , N =
T  V U,N
S
=
T N U,N
     
F S P
Free energy S= =
 T V,N V
T T
 V
F 2 F
F = U TS T, V , N P = U = T
 V  T,N T T
F
=
N T,V
 
H
Enthalpy T =
 S P,N    
H T V
H = U + PV S, P , N V = =
 P  S,N P S S P
H
=
N S,P
     
G S V
Gibbs enthalpy S= =
 T  P,N P
T T  P
G G
G = H T S = N T, P, N V = H = T 2
 P T,N T T
G
=
N T,P
5.9. SUMMARY OF THERMODYNAMIC POTENTIALS 53

Potential Natural Conjugated Maxwell relations

independent dependent and others
variables variables
 

Grand canonical S=
 T V,

potential T, V , P =
V
  T,

= F N = P V N =
T,V
 
S
Chemical potential T s= =
N  T P
G V
= P v= =
N N P T