Statistical Mechanics

Chapter 1: Thermodynamics
Yongle Li
物理系 & 量子与分子结构国际中心
上海大学

2025 年 6 月 24 日

1 Introduction
It should be kept in mind that there is no distinct boundary between statistical physics
and condensed matter physics. Indeed, progress in the former was made through efforts in the
latter.
−→Kubo et al. , Statistical Physics II, Springer-Verlag, 1992 (translated from 1978 Japanese
edition)

2 Review of Thermodynamics
2.1 Known knowledges
1. Loschmidt’s constant:

p0
n0 =
k B T0

p0 = 1.01325 × 105 Pa
T0 = 273.15K
n0 = 2.687 × 1025

2.2 Basic assumptions

The thermodynamics is the abstraction and condensation of the knowledge appears first
in the course of thermal physics. So in the first class, we need to collect what we have learned

1
during thermal physics, and reformulate them by using as much as mathematics.
At the beginning, we need to differentiate following concepts:

2.2.1 Physical concepts

1. Idealized containers: adiabatic wall, diabatic wall. The terminology of ’wall’ means it
can ideally separate things we want to separate, such as heat and matter.

2. System and Environment. The ’system’ means a part of thing what we are interested in,
and ’environment’ means other parts we don’t want to think about.

3. Equiliberium. That means no matter when we observe it, its macroscopic property does
not change.

2.3 Basic Laws

From the course Thermal Physics, we have exploited four laws in thermodynamics. In
the course of Statistical Physics, we need to revise them and summarize them again. Although
one can introduce statistics first, and leave these basic laws as the derivations from statistics,

1. The zeroth law: Equilibrium is transitive: If system A is in equilibrium with B and B is

in equilibrium with C, then A is in equilibrium with C.

A ⇔ B, B ⇔ C, → A ⇔ C

This law implies the existence of an intensive state variable known as temperature. Tem-
perature exists as a property of equilibrium systems as does an equation of state, a
functional relation among state variables.

2. The first law （Generalized conservation of energy）Internal energy U is established as

a state variable. Internal energy is conserved; work and heat are interconvertible forms
of energy.

dU = đQ + đW (1)

The equation (1) is sometimes named the basic equation of thermodynamics.

2
 Here one needs to pay attention to the direction of đW .

The Eq. (1) uses the convention as đW is positive when obtaining work, while
negative when doing work. Someone would like use the reversed convention, that
is, doing work is positive while obtaining work is negative. If so, the equation (1)
is written as:

dU = đQ − đW, or
đQ = dU + đW
（Enrico Fermi, Thermodynamics, Dover）

Here đ denotes both the heat and work are related to the concrete pathway，and are not
real differentiation.

3. The second law

(a) Clausius: It is impossible to devise a process that produces no effect other than the
transfer of heat from a colder to a hotter object.
(b) Kelvin: It is impossible to devise a process that produces no effect other than the
extraction of heat from a single reservoir and the performance of an equal amount
of work.

4. The third law

At T = 0, the entropy change accompanying any thermodynamic process vanishes.

lim (∆S)T = 0
T →0

Corollary: One cannot cool an object to the absolute zero of temperature via finite steps.

2.3.1 First law in detail

Since in experiments of gas and liquid phases, the pressure is easy to control, when we
measure the heat under fixed pressure:

QP = ∆U − p∆V

it’s convenient to define a new quantity with energy dimension, to describe the heat transferred
in such isobaric process:
H = U + pV = Qp (2)

3
2.3.2 Generalized force and generalized displacement

See Table 1.

表 1: Generalized forces and displacements

System Force Displacement

Wire tension F length L
Film surface tension S area A
Fluid presure −P volume V
Magnet magnet fleld H magnetization M
Dielectric electric field E polarization P
Chemical reaction chemical potential µ particle number N

Example of generalized force/displacement couple: liquid droplet

When a liquid droplet reaches thermal equilibrium with its environment, the inner and
outer pressure will not be equal. Suppose its surface tension for this droplet is σ, and
dV and dA is its volume and surface element separately, there would be (See Fig.1):

∆pdV = σdA

And since   r

4 3 r
dV = d πr = 4πr2 dr = 8πrdr = dA
3 2 2
So
2σ
∆pdV = dV
r
2σ
⇒ ∆p =
r

4
 图 1: Plot of surface tension for a droplet.

2.3.3 Joule’s free expansion

The gas expand freely to vacuum, so that there is no heat transition nor work, therefore
the change of the inner energy is 0. This experiment indicates the inner energy is only a
function of T . It is a state variable.

E = E(T ) (3)

2.3.4 Legendre transform and Maxwellian relations

In practical usages, one can change the hard-to-measure term T dS into other self variables,
orelse one can also change −pdV into other self variables. This technique is called Legendre
transform. An example is, if one wants to use variables (S, p) other than (S, V ), one can
introduce a multiplication of discarding variable and introduced variable, and make the original
U minuse it, denoting as H:
H = U + pV

Then, its differentiation is:

dH = dU + d(pV ) = T dS − p dV + (p dV + V dp) = T dS + V dp

From this, one can also introduce two free energies: F = U − T S and G = H − T S. Then,
using the fact of multiple differentication:
∂f (x, y) ∂f (x, y)
fx = , fy =
∂x ∂y
∂fx ∂fy
⇒ =
∂y ∂x

5
 One can obtain four relations:
   
∂T ∂p
=− (Internal energy)
∂V S ∂S V
   
∂T ∂V
= (Enthalpy)
∂p S ∂S p
   
∂S ∂p
= (Helmholtz free energy)
∂V ∂T V
 T  
∂S ∂V
=− (Gibbs free energy)
∂p T ∂T p

Further discussion for Legendre transform is collected in the Appendix 1.

2.3.5 Heat capacity: isovolume and isopressure

In this small section, we want to obtain a general relation between CV and Cp , using
p, V , and T as variables. By the way, the heat capacities are typical example of response
functions. Response functions describe how one thermodynamic function responds to a change
of another thermodynamic function under controlled conditions. They are received special
interests because they can experimentlly measured. Other examples of response functions are:
   2 
1. Isothermal compressibility: κT ≡ − V1 ∂V
∂p
= − 1
V
∂ G
∂p2
T T
   2

2. Adiabatic compressibility: κS ≡ − V1 ∂V
∂p
= − V1 ∂ U
∂p2
S S


3. Thermal expansivity: αp ≡ 1
V
∂V
∂T p


4. Thermal pressure coefficients βV ≡ 1
p
∂p
∂T V

5. magnetic susceptibility χ ≡ ∂M
∂B

   
đQ đQ
CV = , Cp = (4)
dT V dT p

General relation between Cp and CV :

6
From the total differentiation:
 
∂E
dE = CV dT + dV
∂V T

And from the fundamental equation:

dE = T dS − p dV
h ∂S   ∂S  i
=T dT + dV − p dV
∂T V ∂V T
 ∂S  h  ∂S  i
=T dT + T − p dV
∂T V ∂V T

Compare these two equations, we have:

 
 ∂S 
T = CV
∂T V (5)
    ∂S 
 ∂E =T −p
∂V T ∂V T
   
∂S
∴ dE = CV dT + T − p dV
∂V T
      
∂E ∂S ∂V
= CV + T −p (6)
∂T p ∂V T ∂T p

∂H ∂(E + P V )
Cp ≡ =
∂T 
 ∂T
 
∂E ∂V
= +p
∂T p ∂T p
        
∂E ∂S ∂V ∂V
= + T −p +p
∂T V ∂V T ∂T p ∂T p
   
∂S ∂V
= CV + T
∂V T ∂T p
   
∂p ∂V
≡ CV + T Using Maxwell′ s relation
∂T V ∂T p
    
∂E ∂V
≡ CV + p + FromEq.(5)
∂V T ∂T p

We can find:    

 ∂E ∂V

 + p
∂T p ∂T p
Cp =   (7)

 ∂S
T

∂T p

7
 
  
∂p ∂V
Cp = CV + T
∂T V ∂T p
     (8)
∂E ∂V
= CV + p +
∂V T ∂T p

For ideal gas,

   
∂p ∂V
Cp = CV + T
∂T V ∂T p
N kB N k B
= CV + T
V p
= CV + N k B

Cpideal gas = CVideal gas + N R (9)

Consider: What’s the results for van der Waals gas?
aν 2
(p + )(V − νb) = νRT
V2

2.3.6 The second law in detail

Definition

1. Kelvin’s statement No process is possible whose sole result is complete conversion of

heat into work.

2. Clausius’s statement No precess is possible whose sole result is the transfer of heat
from a colder to a hotter body.

3. Caratheodory’s statement In the neighborhood of any arbitary initial state J0 of a

physical system, there exist neighboring states J which are not accessible from J0 along
adiabatic paths. (Am. J. Phys. 17, 41 (1949).)

Further discussion

1. Carnot’s theorem No engine operating between two reservoirs is more efficient than a
Carnot engine operating between them. The best efficient is:
T1
η =1− (10)
T2

8
图 2: Proof of Carnot’s theorem.

9
 Proof:
From both the isothermal lines, we can obtain the works W1 and W2 :
Z V2
W1 = − pdV
V1
Z V2
N RT1
=− dV
V1 V
V2
= −N RT1 ln
V1
Z V4
W2 = − pdV
V3
Z V4
N RT2
=− dV
V3 V
V4
= −N RT2 ln
V3
V3
= N RT2 ln
V4
From both the adiabatic lines:

T1 V2γ−1 = T2 V3γ−1
T2 V4γ−1 = T1 V1γ−1
V1 V4
⇒ =
V2 V3

V2
⇒ W2 = N RT2 ln
V1

W = W1 + W2
V1
= N R(T1 − T2 ) ln
V2
W W
η= =
Q W1
T1 − T2 T2
= =1−
T1 T1

2. Corollary of Carnot’s theorem All reversible Carnot’s engines have the same universal
efficiency η, since each can be used to run any other one backward. Now there is a new
work of Carnot’s engine working between a source with negative temperature and a
normal source. One who is interested in can resort to the literature: Jing-Yi Xi and
Hai-Tao Quan Commun. Theor. Phys. 68, 347 (2017) And a work for single atom

10
 Carnot engine: J. Roßnagel et al. Scinece, 352, 325, (2016)

3. Hurricanes can be looked as Carnot’s engine. There is an excellent discussion in

K. A. Emanuel, Science, 326, 483 (1987).
That is, when the air is heated on the surface of the ocean, with temperature Thigh ∼
300 K, and then rises up adiabatically and quickly from the surface of the oceans to sky
about ∼ 12 km. The temperature of such high in the sky is about 200 K. Then the cold
air loss its kinetic energy and down back to the surface of the ocean adiabatically. Such
a Carnot cycle provides the spinning motion of hurricanes.

图 3: Hurricane Felix hovers over the Caribbean Sea, 2007.

Photo ID: Astronaut photograph ISS015-E-25054.
([Link]

H đQ
4. Clausius’s theorem For any cyclic transformation (reversible or not), T
≤ 0, where
đQ is the heat increment supplied to the system at temperature T .

5. Thermodynamic potentials U , H, F , G are called thermodynamic potentials. They

all have dimensionality of energy.

6. Characteristic functions ”Among these tools, the thermodynamic potentials present

a major practical importance, as all macroscopic properties of a material at equilibrium
can thereby be derived from such a single function. This idea was discovered in 1869 by
François Massieu, again in the prospect of improving steam engines.” (C. R. Physique
18, 526-530 (2017); F. Massieu, Sur les fonctions caractéristiques des divers fluides, C.
R. hebd. Séances Acad. Sci. 69, 858–862 and 1057–1061 (1869).)

11
 图 4: The mature stage of hurricane.
From the yellow arrows we can extract a Carnot’s cycle.
([Link]/how-thunderstorms-form-3444271)

7. Stability conditions

Here are major consequences of Clausius’s theorem:

đQ
1. is path-independent.
T
2. One can introduce a new thermodynamic variable, Entropy:
Z B
đQrev
S(B) − S(A) ≡
A T
here rev means reversible.

3. Clausius inequity: I
đQ
≤0
T
The proofment needs Carnot’s cycle, and the readers should have learned it in the course
of ’Thermal Physics’, so here I ignore the proof. One can refer to 刘川《热力学与统计
物理》for an indept discussion.
P
4. For a reversible transformation, đQ = T dS and đW = i Ji dXi , and the first law
implies:

12
 X
dE = T dS + Ji dXi (11)
i

so we can see S is like general displacement, and T is like general force.

The equation (11) is the most fundamental and useful in thermodynamics.

5. From Eq. (11) one can also deduce the independent variables necessary to describe a
thermodynamics system.

6. For an irreversible change,

Z B
đQ
S(B) − S(A) ≥
A T

that is dS ≥ đTQ . So, once a system undergoes an adiabatic process, that is, đQ = 0,
dS ≥ 0.
So once an adiabatic system reaches its equilibrium state, the entropy reaches its maxi-
mum correspondingly.

2.3.7 The third law

The law The entropy of all systems at zero absolute temperature is a universal constant
that can be taken to be zero. (Nernst’s theorem) This law is obtained from the investigations
of condensed matters, and then is extended to gases. （Enrico Fermi, Thermodynamics, Dover）

lim S(X, T ) = 0 (12)

T →0

Consequences:

1. Since S(T = 0, X) = 0 for all coordinates X,

∂S
lim T =0 (13)
T →0 ∂X

2. Heat capacities must vanish as T → 0 since

Z T
CX (T ′ )
S(T, X) − S(0, X) = dT ′ (14)
0 T′
and the integral diverges as T → 0 unless

lim CX = 0 (15)
T →0

13
 3. Thermal expansivities also vanish as T → 0 since

1 ∂x 1 ∂S
αJ = = (16)
x ∂T J x ∂J T

4. It’s impossible to cool any system to absolute zero temperature in a finite number of
steps.

Appendix: Proof of Eq. (16)

Using exterior differential.

dE = T dS + J dX
⇒ dS ∧ dT = dJ ∧ dX

 S = S(T, J)
X = X(T, J)
    
 ∂S ∂S

 dS = ∂T dT + dJ
∂J T
 J  

 ∂X ∂X
dX = dT + dJ
∂T J ∂J T
    
∂S ∂X
⇒ dJ ∧ dT = dJ ∧ dT
∂J T ∂T J
□

2.4 Applications
2.4.1 Joule-Thompson’s throttling

H1 = H2 (17)

Proof:

∵Q=0
∴ ∆E − W = 0
⇒(E2 − E1 ) + (p2 V2 − p1 V1 ) = 0
E 2 + p2 V 2 = E 1 + p1 V 1
□

14
 Since H = H(T ), for ideal gas, the temperature is unchanged.
But for non-ideal gas, there is a parameter to control whether the gas is heated or cooled
in a throttling process:  
∂T
µ≡ (18)
∂p H

∵ dH = T dS + V dp
T dS + V dP = 0 iso − enthalpy
"    #
∂S ∂S
T dT + dp + V dp = 0 (dH = 0) (19)
∂T p ∂p T

And since from Maxwell relation:

∵ dE = T dS − p dV
dS ∧ dT = − dp ∧ dV
  "  #
∂S ∂V
⇒ dp ∧ dT = − dp ∧ dT
∂p T ∂T p

   
∂S ∂V
=− (20)
∂p T ∂T p

And we have already have (7):

 
∂S
Cp = T
∂T p

Put Eq. (20) into the Eq. (19), we can obtain:

"   # 
∂S ∂S
T dT + dp + V dp = 0
∂T
p ∂p T
"   #
Cp ∂V
T dT − dp + V dp = 0
T ∂T p
"   #
∂V
Cp dT + V − T dp = 0
∂T p

"   #
∂V
  T −V
∂T ∂T p
∴µ≡ = (21)
∂p H Cp

15
3 The physical meaning of enthalpy, Helmhotz free energy
and Gibbs free energy
The beginners always feel confused for the three Legendre transformed inner energies:
enthalpy, Helmholtz free energy and Gibbs free energy. Here I provide a brief summary.

3.1 Enthalpy
From the first law and the fundamental equation:

dU = T dS − p dV = đQ + đW (22)

If the system does only volumic work (đW = −p dV and dp = 0) and undergoes isobaric
process,
dH = d(U + p dV )p = đQp + (−p dV ) + p dV = đQp

So, the change of enthalpy of a system undergoing an isobaric process and not doing
non-volumic work, is the heat of this process.

3.2 Helmholtz free energy

Firstly, we divide the work into two parts: volumic work and non-volumic work.

đW = đWn + đWV = đWn − p dV (23)

图 5: The system and environment of the discussion.

For a system reaching equilibrium to its environment, if it changes a bit, from the second
law, the total entropy of the super system constructed by the system and environment should

16
increase:
dS tot = dS system + dS environment ≥ 0 (24)

The ∆S for the system would be complicated, but that for the environment comes simply
from the heat:
đQ
dS environment = − (25)
T
So we have:
đQ
dS system ≡ dS ≥
T

T dS ≥ đQ (26)

Add the work:

đW + T dS ≥ đQ + đW = dU (27)

We need to discuss this equation in two cases separately:

isothermal and isochoric (dT = 0 and −p dV = 0):

đWn + T dS ≥ dU
(28)
đWn ≥ (dU − T dS)V = d(U − T S)V = dF
Since in the convention, both đWn and dF are smaller than zero, the Helmholtz free energy
is the maximum non-volumic work done by the system during the isothermal and isochoric
process.

3.3 Gibbs free energy

The second case is isothermal and isobaric process. now:

(đWn − p dV )p + T dS ≥ dU
(29)
đWn ≥ (dU − T dS + p dV )p = d(U − T S + pV )p = dG

Again, since both đWn and dG are both smaller than zero, the Gibbs free energy is the
maximum non-volumic work done by the system during the isothermal and isobaric process.
In chemistry, the pressure and temperature are relatively easier to control, so during the
chemical reactions, one always discuss the enthalpy and Gibbs free energy. The enthalpy is the
pure heat during such process, while the free energy is the work. From the view of energetics,
free energies contain also the contribution of entropy.

17
3.4 Thermodynamic potential and Massieu’s principle
The quantities from Legendre transformed fundamental equation of thermodynamics, H,
F (A), and G, with original inner energy U (E), are called thermodynamic potential. They are
all with dimension of energy. Their quotients with T are called charasteristic functions, and
characteristic functional are with dimension of entropy.
Now we can feel that, once we choose a set of proper thermodynamic variables, we can use
only one thermodynamic potential to determine the thermodynamic property for an equilibrium
closed system. This is proposed by Francois J. D. Massieu, a French physicist and engineerer.
For detailed descriptions, one can refer to 赵柳《统计热物理学》

(The main content of the Chapter 1 finished here. Following is appendices.)

18
4 Appendix 1: Mathematics: Legendre transformation
Introduce a Legendre function:

L = L(a1 , a2 , · · · , an )
dL = A1 da1 + A2 da2 + · · · + An dan

If one wants to discard ai , one can define a new Legendre function:

dL̄ = dL − Ai ai (30)

So we can obtain

dL̄ = A1 da1 + A2 da2 + · · · + Ai dai + · · · + An dan − Ai dai − ai dAi

dL̄ = A1 da1 + A2 da2 + · · · − ai dAi + · · · + An dan

Example:
dE = T dS − p dV
Introduce : H ≡ U + P V
dH = dU + p dV + V dp
= T dS + V dP
Further discussion:
Hilbert & Courant, Mathematical Physics, Vol. 2, Ch. 1: 6.
Using coordinates of tangent plane to describe the integral surface instead of the point
coordinates.
For a given surface u, one can either consider it is a set of points determined by a function
u = u(x, y):
(x, y, u), u = u(x, y)

Or else one can look it as the envelope of its tangent planes determined by a set of (x̄, ȳ, ū)
(running coordinates):

ū − ξ x̄ − η ȳ + ω = 0

Then we call ξ, η, ω the coordinates of this plane.

From the constraint: the plane is tangent to the surface u(x, y), we have:

∂u(x, y) ∂u(x, y)
(ū − u) − (x̄ − x) − (ȳ − y) =0
∂x ∂y

19
 So we can find the plane coordinates (That is, solve the (ξ, η, ω)):

 ∂u

 ξ=

 ∂x

 ∂u
η=

 ∂y



 ∂u ∂u
ω = x +y −u
∂x ∂y
If w is a function of (ξ, η), the set of the tangent surfaces can be determined, so that the
original surface u(x, y).
Actually, from equations 

 ∂u
ξ =
∂x

 ∂u
η =
∂y
we can determine x = x(ξ, eta) and y = y(ξ, η), then insert this pair into

∂u ∂u
ω=x +y − u = xξ + yη − u
∂x ∂y
We can obtain the ω(ξ, η) from u(x, y).
Reversibly, we can determine u(x, y) from ω(ξ, η):


 u = xξ + yη − w



 ∂w
=y
 ∂η



 ∂w
 =x
∂ξ
But from the condition, if the plane is so-called developable surfaces, the Legendre trans-
formation will fail. Since the Jacobian must be finite:

∂(ξ, η)
J≡
∂(x, y)
∂u ∂u
∂( , )
∂x ∂y
=
∂(x, y)
∂2u ∂2u ∂2u
= −2 ≡ρ
∂x∂x ∂y∂y ∂x∂y
Exercise:
What’s the Jacobian of the inverse transformation (ξ, η) → (x, y)?

20
5 Appendix 2: Jacobian under canonical transformation
is equal to 1
If we introduce canonical transformation:
∂H(p, q, t) ∂K(P , Q, t)
q̇i = Q̇i =
∂pi ∂pi
⇒
∂H(p, q, t) ∂K(P , Q, t)
ṗi = − Ṗi = −
∂qi ∂Qi
with:
Z thX i Z thX i
δ pi q̇i − H(p, q, t) dt = δ Pi Q̇i − K(P , Q, t) dt = 0
0 i 0 i

We can find:
∂(P , Q)
J(P , Q) =
∂(p, q)
∂p ∂q ∂q ∂p
= ( )Q ( )P − ( )Q ( )P
∂P ∂Q ∂P ∂Q
Now introduce generating function, such as:
∂f3 (p, Q, t)
q=−
∂p
∂f3 (p, Q, t)
P =−
∂Q
Consider a simple example:
∂p ∂p ∂q −1
( )Q = ( )−1
Q ( )
∂p ∂p ∂Q p
(The last equality follows using f3 (p, Q))
And
∂Q ∂q
( )p = −( )p
∂p ∂p
(The last equality follows using f4 (p, P ))

∂q ∂Q ∂q ∂p ∂q
J =( )p ( )p + ( )Q ( )p = ( )p = 1
∂Q ∂q ∂p ∂q ∂q

6 Appendix 3: Mathematics: Exterior differential (1)

In thermodynamics, according to Gibbs’ phase law, for single ingredient equilibrium sys-
tem, one can choose a pair of thermodynamic variables as basic variables, and other thermo-
dynamic variables can be represented as the functions of them.

21
 Such as, there are four common thermodynamic variables: T , S, P , and V . One can
choose two of them as basic, and other two can be expressed by the previous two.
And from the basic equation of thermodynamics, we can see they are with some direction,
like the surface/path integral in calculus. So their calculations are related to the sequence.
(In mathematics, this part of knowledge is related to the differentiation form in differen-
tiation geometry. There are two references:

1. 余扬政、冯承天, 《物理学中的几何方法》

2. 常庚哲、史济怀的《数学分析教程（下册）》

Considering a differentiation with direction:

ZZ ZZ
g(x, y) dx dy = g(u(x, y), v(x, y))J du dv
∂(x, y)
J≡
∂(u, v)
,
So once u = y and v = x, → J = −1.
This means: ZZ ZZ
g(x, y) dx dy ̸= g(x, y) dy dx

.
So we introduce the wedge product:

dx ∧ dy = − dy ∧ dx

Properties:

1. Besides the symbol, the product rule is the same as normal product.

2. Since it’s anti-symmetric, the diagonal terms are zeros:

dx ∧ dx = dy ∧ dy = 0

So once x = x(u, v), y = y(u, v), we have

   ∂x    ∂y   ∂y  
∂x
dx ∧ dy = du + dv ∧ du + dv
∂u ∂v ∂u ∂v
= J du ∧ dv

22
6.1 Example 1
From the second law,
dE = T dS − p dV
⇒ dT ∧ dS = dp ∧ dV
.
If we choose:
T = T (S, V )
P = P (S, V )
,
insert the functional form into the previous equation,

dT (S, V ) ∧ dS = dp(S, V ) ∧ dV
         
∂T ∂T ∂p ∂p
dS + dV ∧ dS = dS + dV ∧ dV
∂S ∂V S ∂S ∂V S
V
    V
∂T ∂p
dV ∧ dS = dS ∧ dV
∂V S ∂S V
   
∂T ∂p
⇒ =−
∂V S ∂S V

6.2 Example 2
If we choose:
T = T (P, S)
S = S(T, P )
P = P (T, S)
,
then we have
 
∂T
dT ∧ dS = dp ∧ dS
∂p S
   
∂T ∂S
= dp ∧ dT
∂p S ∂T p
     
∂T ∂S ∂p
= dS ∧ dT Divide the dS ∧ dT from both sides.
∂p S ∂T p ∂S T
     
∂T ∂S ∂p
⇒ = −1
∂p S ∂T p ∂S T

7 Homework
1. Please use Legendre transformation to obtain all the thermodynamic variables: F , H,
and G.

23
2. Please try to obtain all the Maxwell relations.

24