Thermodynamic Potentials

1
We introduce thermodynamic potentials. There is
a fundamental set of four, all of which are energies and hence
conserved. They are:
U the internal energy
H enthalpy ( old fashioned name is “heat content”)
Important when discussing heat capacities and latent
heats.
F Helmholtz Function. Often A is used as the symbol.
Important in statistical mechanics
G Gibbs Function. Important in phase transitions and
chemical reactions.

The thermodynamic potentials are also called generating functions.

These potentials, similar to potential energy in mechanics, are

introduced for convenience in the analysis of certain processes.
2
Some Mathematical Preliminaries

Thermodynamic properties are continuous point functions

and have exact differentials. A property of a single
component system may be written as general mathematical
function z = z(x,y). For instance, this function may be the
pressure P = P(T,v). The total differential of z is written as

3
where

Taking the partial derivative of M with respect to y and of N with respect to x yields

Since properties are continuous point functions and have exact differentials,
the following is true

The equations that relate the partial derivatives of properties P, v, T, and s of

a simple compressible substance to each other are called the Maxwell
relations. They are obtained from the four Gibbs equations. The first two of
the Gibbs equations are those resulting from the internal energy u and the
enthalpy h.
du  T ds  P dv
dh  T ds  v dP 4
The second two Gibbs equations result from the definitions of
the Helmholtz function a and the Gibbs function g defined as
a  u  Ts
da  du  T ds  s dT
da   s dT  P dv

g  h  Ts
dg  dh  T ds  s dT
dg   s dT  v dP
Setting the second mixed partial derivatives equal for these four
functions yields the Maxwell relations

5
Now we develop two more important relations for partial derivatives—the
reciprocity and the cyclic relations. Consider the function z = z(x,y)
expressed as x = x(y,z). The total differential of x is

Now combine the expressions for dx and dz.

Rearranging,

6
Since y and z are independent of each other, the terms in each bracket must
be zero. Thus, we obtain the reciprocity relation that shows that the inverse
of a partial derivative is equal to its reciprocal.

or

The second relation is called the cyclic relation.

7
Another way to write this last result is

8
• Legendre Transformation

This is a very useful mathematical transformation.

For example it is used in classical mechanics to go
from the Lagrangian to the Hamiltonian.
One often introduces the notion of canonically
conjugate pairs, one variable being extensive and the
other intensive. For example (V, -P), (S, T).
In an analysis of certain processes U is often not a
convenient function to work with. We have already
introduced another useful function (H) and now we
will introduce two new state functions:
 Helmholtz F
 Gibbs G

9
The thermodynamic potentials U, H, F and G are
state functions. If we know these functions then all
the thermodynamic properties of a system can be
calculated by differentiation alone.
Consider dU(S,V) = TdS - PdV
The extensive variables S,V are taken as
independent. We say that U is a generating function
having as its natural variables S and V. Canonically
conjugate pairs are (-P, V) and (T, S). The first pair
are mechanical variables and the second pair are
thermal variables.

10
 We have dU(S,V) = TdS - PdV
But we can write dU   US  V dS   UV  S dV
Comparing gives  US  V  T and  UV  S   P
These two equations are called the equations of state.
Hence if we know U(S,V) we can obtain T and P by
differentiation.
There are four possible pairings of a thermal variable and
a mechanical variable:
(S, V) (S, P) (T, V) (T, P)
We will obtain thermodynamic potentials for each of
these pairs. This will be done using a Legendre
Transformation.
11
 Consider Z = Z(x, y)
dZ = Xdx + Ydy _ _ _ _(1)
with (x, X) and (y, Y) as canonical pairs. Suppose
we wish to replace (x, y) by (X, Y). We use a
transformation to form a new function
M(X, Y) = Z - xX - yY
dM = dZ - Xdx - Ydy - xdX - ydY
dM = - xdX - ydY _ _ _ _(2)
 Z   Z   M   M 
dZ    dx    dy dM    dX    dY
 x  y  y  x  X Y  Y  X
The reciprocity relations are:

 Zx  y  X  
Z
y x Y
 MX  Y   x  MY  X  y 12
 To replace only one variable, say y by Y:
N(x,Y) = Z - yY _ _ _ _(3)
and using (1) dN = Xdx - ydY _ _ _ _(4)
 N   N 
dN    dx    dY
 x Y  Y  x
with reciprocity relations

 Nx  Y  X  NY  x   y
With this introduction we proceed to the thermodynamic
potentials. 13
Definition of the Thermodynamic Potentials
We had, for the first law of thermodynamics, for a
hydrostatic system:
dU(S, V) = TdS - PdV _ _ _ _(5)
We will replace U(S, V) by a new state function H (S,
P) (natural variables). In this case we are replacing
only one variable V by its conjugate -P.
H=U-(-P)V → H=U+PV
dH=dU+PdV+VdP
dH=TdS-PdV+PdV+VdP → dH=TdS+VdP

 H   H 
dH    dS    dP
 S  P  P  S 14
 Reciprocity relations:
 HS  P  T  HP  S  V

If H is written in terms of S and P differentiation permits

us to calculate T and V.

15
 The Helmholtz Function
Now we wish to replace U(S, V) by F(T, V)
We wish to replace S by T. The appropriate Legendre
transformation is then F(V,T)=U-ST
dF=dU-SdT-TdS=TdS-PdV-SdT-TdS
dF(V,T) = -PdV – SdT
 F   F 
dF    dT    dV
 T V  V T

 
F
V T
 P  
F
T V
 S
Again, if F is a known function of V and T (natural
variables), then we can obtain P and S. 16
 The Gibbs Function
Now we wish to replace U(S, V) by G(T, P).
Two variables are changed so we use
G(T, P) = U – ST + VP giving
dG(T, P) = -SdT + VdP
 GT  P   S  GP  T  V
Let us summarize the thermodynamic potential
functions.
dU(S,V) = -PdV + TdS _ _ _ _(a)
dH(S,P) = VdP + TdS _ _ _ _(b)
dF(T,V) = -PdV – SdT _ _ _ _(c)
dG(T,P) = VdP – SdT _ _ _ _(d) 17
 Helmholtz Function
In statistical mechanics the quantity that is usually
calculated is the Helmholtz Function. Other
thermodynamic properties are then calculated
from this.

Before pursuing a general discussion of this function,

let us consider an example to show how U and F differ by
means of a simple example. We consider a closed cylinder
in which there is a piston which separates two gases which
as not initially at the same pressure. The situation is shown
on the next slide.

18
 Bath at T

V2 Cylinder
V1
Piston (adiabatic)

We let the piston move until equilibrium is achieved.

Now we consider an infinitesimal reversible change about this
equilibrium position.
(a) Suppose that the wall of the cylinder is adiabatic.
đ Q  dU  PdV đ Q  0 dU   PdV dU  dU 1  dU 2   P1dV1  P2 dV2
Since the pressures are now the same and obviously dV2  dV1
we have dU=0. At equilibrium the internal energy has
its minimum value.
19
(b) Suppose that the wall of the cylinder is diathermal.
dF  SdT  PdV dT  0 dF   PdV dF  dF1  dF2   P1dV1  P2 dV2

Since the pressures are now the same and dV2  dV1
we have dF=0. In this situation, it is F, not U that is a minimum
at equilibrium.

(U is not minimized in this case because the total entropy of the

system and the reservoir must be maximized and the Helmholtz
Function takes this into account.)

In continuing our discussion of the Helmholtz Potential, we

consider isothermal processes.

 
F
V T  P 20
 Recall that if we have a system in contact with a reservoir
(surroundings at constant temperature) then
S (system) + S (reservoir)  0.
If an amount of heat Q is transferred from the reservoir to the
system, then
S(system) - Q
 0 giving (at T)
T

Q  T (S) (system)

The equality sign holds only for a reversible process. We have

F= U – ST (This is just the Legendre Transformation.)

At constant temperature: F = U - TS or

F = Q - W - TS

21
 If no new entropy is created in a process (reversible process) then
Q = TS and F = -W or
( F f  Fi )   W (constant T)
The change in the Helmholtz function is then equal to the work done.
If new entropy is created in a process (irreversible) then Q <TS
and F < -W or
( F f  Fi )   W (constant T)
The work is all the work on or by the system, including any done by
the system’s surroundings.

Conclusion:
If a system undergoes a reversible process at constant T and V (no
work) then the Helmholtz Function does not change.
If the system undergoes an irreversible process under constant T and
V, then the Helmholtz Function decreases. 22
Conclusion:
The change in F in an isothermal reversible process is the work done
on or by the system. In an isothermal process, the maximum
amount of work that can be done by a system is the decrease in the
Helmholtz function.

W  Fi  F f (constant T)
One can also say that the decrease in F gives the maximum amount of
energy that can be fixed in an isothermal process and made
available for work. For this reason F is often called the Helmholtz
free energy.

So F is intimately associated with work in an isothermal process.

23
 If we consider constant T and constant V and
no other types of work, W = 0 and F  0
and so (Ff - Fi) < 0 in an irreversible process.
Natural processes at constant T and V
undergoing change will move to lower values of
F. Equilibrium occurs at minimum F.

The Helmholtz potential is a function of T and V,

its natural variables. In statistical mechanics these
are the important variables, hence the importance
of the Helmholtz potential in statistical
mechanics.

The example starting on the next slide is to show that, if the

Helmholtz Function is known, many other thermodynamic
24
properties can be calculated.
Example: Some expressions involving F(T,V).
We previously obtained the expressions:
 
F
V T
 P  
F
T V
 S
F  U  TS  TS  F  U H  U  PV
G  H  TS  (U  PV )  ( F  U )  F  PV
 F 
G  F  V  
 V T
 F 
U  F  TS U  F  T  
 T V
 F   F 
H  U  PV H  F  T    V  
 T V  V T 25
  U   F 
CV    and we had U  F  T 
 T  V  T  V
  F    F    2
F
CV       T 2 
 T  V  T  V  T  V
 F 2

C V   T 2 
 T  V
Notice that the thermodynamic quantities are all expressed as
functions of
 F   F 
F ,V , T ,   ,  
 T V  V T
If one can theoretically calculate F(V,T) then any other
thermodynamic function can be obtained as a function of V and T.
26
EXAMPLE: A cylinder contains a piston on each side of which is one
kilomole of an ideal gas. The walls of the cylinder are diathermal and
the system is in contact with a heat reservoir at T=273K. The initial
volumes of the gases are 12 liters (left) and 2 liters (right). The piston
is now moved reversibly so that the final volume on both sides is 7
liters. What is the change in the Helmholtz potential? f

12 2
dF=-PdV-SdT=-PdV F    PdV
i
Since we have 1 kmole of each gas
1 kmole 1 kmole PV=RT and T is constant.
For the left side: Wl  RT dV  RT ln 7
7

273K 
12 
V  12 
 
7 7 7
dV  7
For the right side Wr  RT   RT ln 
V 2
1 kmole 1 kmole 2

 7 7 J  49 
F   RT ln    8.314  103 (273K ) ln 
 12 2  K  24 
(Some work is done and
F decreases.) F  1.62MJ 27
• Gibbs Function (very important in chemistry)
G = H – TS
To create a system one need not supply the entire
enthalpy, as some energy (TS) can flow in as heat
from the surroundings. Only the difference (H-TS)
must be supplied by work.

Consider a system in surroundings that constitute a T

and P reservoir. This is the usual environment for
chemical reactions and some phase changes. (T and
P natural variables for the Gibbs Function.)
Since H = U + PV
G = U - TS + PV
28
Considering a process which is both isothermal and
isobaric G = U - TS + PV and since
U = Q - W, then
G = Q - W - TS + PV
In general, besides any mechanical work, we can
have some other work, such as electrical work.
W = PV + W(other)
G = Q - TS -W(other)
Suppose W(other) = 0
As before Q  TS and so G  0 for constant T, P.
Again, in an evolving process, equilibrium will be
established when G reaches its minimum value. 29
 Suppose W (other) 0
If we have a reversible process Q = TS and
G = -W(other), so
Gf - Gi = -W (other) or W(other)  (G i  G f )
The change in the Gibbs function gives the maximum
energy that can be freed in an isothermal, isobaric
process and made available for non-mechanical work.
For this reason one often speaks of the Gibbs free
energy. This can be confusing as we have the
Helmholtz free energy also. It is best to avoid using
free in both cases.
N.B. The state variables T and P need not be fixed
throughout the process, but must have the same initial
and final values.
30
Reversible processes: Let us summarize some results
for some particular reversible processes.

Isochoric dU = đQ
Isobaric dH = đQ
Isothermal dF = -đW
Isothermal, isobaric dG = -đW(other)

31
The relationship amongst the thermodynamic
potentials can be summarized as follows:

- TS

U F

+ PV
H G

32
• Maxwell’s Relations

Let us apply the condition for an exact

differential to the differential forms of
the potentials.

dU = TdS - PdV  VT  S   PS  V

dH = TdS + VdP  TP  S   VS  P
dF = - SdT - PdV  VS  T   TP  V
dG = -SdT + VdP  PS  T   VT  P
33
 The relationships among the partials are Maxwell’s
relations. These relations are enormously useful.

Each partial involves a state variable that can be integrated

along any convenient reversible path between two states to
obtain the difference in the value of a state variable
between two equilibrium states.
They can be used to obtain values, or differences between
values, for quantities not easily measured from more easily
measured quantities, such as P, V or T.
They are often used to replace a partial, which cannot be
evaluated, by another partial which can be determined.

34
 The VFT-VUS diagram is a device for remembering
the differential forms of the potentials.
Notice the direction of the arrows.

If you go in a direction opposite to an arrow, the

quantity is negative.
The potentials are at the sides. They are functions of
the state variables at the corresponding corners.
The corresponding conjugate coordinate to the variable
is obtained by following the diagonal arrow.

35
 V F T

U G

S H P
Consider the Gibbs function, dG will consist of a dT
and a dP. The coefficient of dT will be -S and the
coefficient of dP will be V.

dG=-SdT+VdP

36
The Clapeyron Equation

The Clapeyron equation enables us to determine the enthalpy change asso­

ciated with a phase change, hfg, from knowledge of P, v, and T data alone.

37
Consider the third Maxwell relation

During phase change, the pressure is the saturation pressure, which depends on the
temperature only and is independent of the specific volume. That is Psat = f(Tsat).
Therefore, the partial derivative can be expressed as a total derivative
(dP/dT)sat, which is the slope of the saturation curve on a P-T diagram at a specified
state.
This slope is independent of the specific volume, and thus it can be treated as a
constant during the integration of the third Maxwell relation between two saturation
states at the same temperature.
For an isothermal liquid-vapor phase-change process, the integration yields

38
During the phase-change process, the pressure also remains constant.
Therefore, from the enthalpy relation

Now we obtain the Clapeyron equation expressed as

39
 • Clausius-Clapeyron Equation
This is an important relation describing how
the P varies with T for a system consisting of
two phases in equilibrium.
From the VFT-VUS diagram the molar Gibbs
function is given by:
dg = -sdT + vdP
Hence s   g

T P
and v  
g
P T
(reciprocity relations)

{Note that we are concerned with the variables P and T and

these are the natural variables for the Gibb’s Potential.} 40
 In a first order phase transition these
variables change as follows:

s
phase (i) phase (f) phase (i) phase (f)

T T

(In higher order phase transformations s

and v are the same.)

41
At the T of the phase transition g is
single valued but the slope is
discontinuous.

slope discontinuity
g
phase (f)
phase (i)

42
Two phases can exist in equilibrium at various
combinations of (P, T). On a P, T diagram
there will be a phase boundary (coexistence
curve).
Phase equilibrium line
phase (i) with g (i) = g (f)

P
b
• phase (f)
a
•

43
 We consider a reversible phase change at
constant T and P. As was discussed earlier,
(slide 31) the Gibbs function will remain
constant:
g(i) = g(f)
This is true for all phases in equilibrium, i.e. on
a phase equilibrium line.
Consider the two neighboring points a and b,
separated by dP and dT. The g’s will change as
we go from a to b, but they must remain equal,
so
dg(i) = dg(f)
-s(i) dT + v(i) dP = -s(f) dT + v(f) dP
[Note: P and T are not independent since we are moving
44
along a phase boundary.]
This yields dP
dT  s( f )  s( i )
v( f )  v( i )
Hence the slope of the phase boundary is
determined by the entropies and volumes of
the two phases .
đQ 
From dS   S 
T T
in which ℓ is the appropriate latent heat, we
therefore have
dP
dT
  
T v (f)  v ( i ) 
This is the famous Clausius-Clapeyron
equation. It applies to the slope of any
phase boundary line on a PT diagram.
45
Let us consider a sublimation or vaporization curve.
The vapor will be approximated as an ideal gas.
Since v(i) << v(f) we can make the excellent
approximation v(f) - v(i) = v(f).
From the ideal gas law v ( f )  RT
,P
the Clausius-Clapeyron equation now becomes
dP
dT
 P
RT 2
P2 T2
 dP
P1 P
 
R  dT
T1 T
2 , giving

ln   
P2
P1
 1
R T2
 1
T1

46
 • Example A pressure cooker
At P = 1 atmosphere the BP of water is 373k.
What is the BP at 2 atmospheres? The latent
heat of vaporization is 40.68 kJ/mole.
Using the above equation:
ln2    R  1
T2
 1
373k

giving T2 = 393.8k = 120.8 0C.
The experimental value is 120.6 0C

(Latent heat has some temperature dependence, but this

can usually be ignored if the range is not very great.) 47
EXAMPLE: Clausius-Clapeyron Equation. dP s(f )  s(i )
 (f )

dT v  v ( i ) 
s always Gas specific volume
increases Liquid usually increases
Solid
except 3
He below 0.3K
 
Consider a solid to liquid transition. s ( f )  s (i )  0

If v v 0
(f ) (i )
 dP
dT
0 curve has positive slope

If v v 0
(f ) (i )
 dP
dT
0 curve has negative slope

One of few substances with a negative slope is water

48
 S-L (water)
S-L
S-L slopes exaggerated

P L
S
G

L-V

V
S-V

49
EXAMPLE: Let us use the Clausius-Clapeyron equation to calculate
the decrease in temperature from the triple point to the NMP.

H2 0
magnified TP 0.01C 273.16K 612Pa
P fusion curve {NMP 0.00C 273.15K 1 .013  10 5
Pa }

NMP Apply the Clausius-Clapeyron

equation to fusion curve

TP  T( v  v )
dP
 dT  dP
dT
T( v  v ) 

T kg kg 1
 H 2O  10 3
 ice  916 3 v
m3 m 
3 3
m m J
v  10 3 v  1.09  10 3   3.34  10
5

kg kg kg 50
 m 3

273.16K  10  3  1.09  10  3 
kg
dT  (1.01  10 5  612)Pa
5 J
3.34  10
kg
dT=-0.0074K

Presence of dissolved air in ice-water mixture further reduces the

temperature by 0.0023K. Hence
dT=-0.0097K

51
 EXAMPLE:
Prove that, on a phase diagram, the slope of the sublimation
curve at the triple point is greater than that of the vaporization curve
at the same point.
We consider a substance, such as water, that expands upon
freezing. v  v ! v  v v  v
H 2O Fusion curve From VFT-VUS dH=TdS+VdP
At P=constant dH=TdS= đ Q
Liquid
P At a phase transformation

Solid rve đ Q   dS 
r cu T
a po dP s ( f )  s ( i ) 
V  (f ) 
 
e

Vapor
rv

dT v  v (i )
T v (f )  v ( i )
cu
lim
b

  h( f )  h( i ) h  h  h  h  h  h

Su

T
(h  h)  (h  h)  (h  h) 52
SV  LV  SL at the triple point

 dP   dP   dP 
T( v  v )   T( v  v )   T( v  v ) 
 dT  SV  dT  LV  dT  SL

Remember that, in this case, v  v

 dP   dP   ( v  v )   dP   ( v  v ) 

         
 dT  SV  dT  LV  ( v   v  )   dT  SL  ( v   v  ) 
>1 (-) (-)

 dP   dP 
    
 dT  SV  dT  LV

53
General Relations for du, dh, ds, Cv, and Cp

The changes in internal energy, enthalpy, and entropy of a simple, compress­

ible substance can be expressed in terms of pressure, specific volume, tem­
perature, and specific heats alone.

Consider internal energy expressed as a function of T and v.

Recall the definition of the specific heat at constant volume

Then du becomes

54
Now let’s see if we can evaluate in terms of P-v-T data only.
Consider the entropy as a function of T and v; that is,

Now substitute ds into the T ds relation for u.

Comparing these two results for du, we see

55
Using the third Maxwell’s relation
 s   P 
   
 v T  T v
 u   P 
  T   P
 v T  T v
Notice that the derivative is a function of P-v-T only. Thus the total
differential for u = u(T,v) is written as

Let’s evaluate the following partial derivative for an ideal gas.

56
For ideal gases

This result helps to show that the internal energy of an ideal gas does not depend
upon specific volume. To completely show that internal energy of an ideal gas is
independent of specific volume, we need to show that the specific heats of ideal
gases are functions of temperature only. We will do this later.

We could also find the following relations for dh and ds where h = h(T,P) and s =
s(T,v) or s = s(T,P)

57
Example

Determine an expression for the entropy change of an ideal gas when

temperature and pressure data are known and the specific heats are
constant.

For an ideal gas

For constant specific heat this becomes

58
Example

Show that Cv of an ideal gas does not depend upon specific volume.

For an ideal gas

59
Therefore, the specific heat at constant volume of an ideal gas is
independent of specific volume.

60
The Joule-Thomson Coefficient

The temperature behavior of a fluid during a throttling (h = constant)

process is described by the Joule-Thomson coefficient, defined as

The Joule-Thomson coefficient is a measure of the change in temperature of

a substance with pressure during a constant-enthalpy process, and it can
also be expressed as

61