M Sc (Physics) SEM-II (2021)

Subject: Statistical Physics

Fifth State of matter

Bose-Einstein Condensate

?
Sixth State of matter
Fermi Condensate
Unit I:
Fundamentals of classical statistical mechanics,
microstate and macrostate, distribution function,
Liouville’s theorem, Gibbs Paradox, ensembles
(micro-canonical, canonical and grand-canonical),
partition function, free energy and connection with
thermodynamic quantities, energy and density
fluctuations
➢ Statistical Physics: The branch of Physics which deals with the study
of macroscopic properties of the systems in terms of their microscopic
properties.
➢ It predicts the average value of macroscopic properties by applying
theory of probability which are nearly correct for the system consisting
of a very large number of particles. It is useless for the system of small
number of particles.
➢ It bridges the gap between Classical Mechanics and Thermodynamics.
➢ Thermodynamics: It deals with the study of macroscopic properties of
the systems based on laws of thermodynamics without the information
of molecular theory of the system.
➢ Classical Physics: Mechanics based on classical or Newtonian concept
obtain complete information about a system by solving its equation of
motion. It is useful for the system with small number of particles or
small number of degrees of freedom. It is difficult to apply for the
system with large number of degrees of freedom.
➢ Quantum Mechanics: Mechanics based on quantum approach obtain
complete information about a microscopic system from its de Broglie
wavelength. It is difficult to apply for the macroscopic systems.
Statistical Physics is not concerned with actual motions or
interactions of individual particles but explores the most probable
behaviour of assembly of particles.
 Entropy and Probability
➢ Both thermodynamics and statistical mechanics deals with the study of
macroscopic properties of the systems, there must be correlation
between them.
➢ Thermodynamically, the entropy of the system is maximum when it is
in equilibrium state.
➢ Statistically, the probability is maximum when the system is in
equilibrium state.
➢ Boltzmann suggested that statistics through probability and
thermodynamics through entropy establish the correlation between
entropy and probability as

𝑺 = 𝒌 𝐥𝐧 𝑾
Where, S is entropy and W is thermodynamic probability
 Statistical Physics
Theory of equilibrium state Theory of nonequilibrium
(Probabilities and average state (Probabilities and
values are independent of average values are dependent
time) of time)

Every system undergoes from its nonequilibrium state

to equilibrium state

Universe is changing from its nonequilibrium state with

minimum entropy (low disorder) to equilibrium state with
maximum entropy (high disorder).

✓ Entropy of the system remains constant in all reversible

processes
✓ Entropy of the system increases in all irreversible processes
✓ Entropy of the universe is increasing
According to Big Bang theory, physicists and astronomers tried very hard to
measure the composition of the universe. A universe with too little matter would
expand forever, and its average density would eventually drop to zero.
 Statistical Physics

Classical Statistics Quantum Statistics

Maxwell-Boltzmann Bose-Einstein Fermi-Dirac

✓ Internal structure ✓ Based on Internal ✓ Based on Internal
ignored structure structure
✓ Particles are ✓ Particles are ✓ Particles are
distinguishable indistinguishable indistinguishable
(Boltzanos) (Bosons) (Fermions)
✓ No restriction on ✓ No restriction on ✓ Restriction on
occupancy of occupancy of energy occupancy of energy
energy levels levels levels (As per Pauli’s
✓ Spin of particles ✓ Particles have exclusive principle)
not considered integral spin ✓ Particles have half
✓ Wavefunction is not ✓ Wavefunction is integral spin
considered symmetric ✓ Wavefunction is
antisymmetric
 Example: I have 2 gifts to be distributed among 3 best students, what
would be nature of gifts according to MB, BE and FD statistics?

➢ According to MB, 2 gifts must be distinguished and may be simply a

PEN and a BOOK. There will be 9 ways with which these gifts can be
distributed.
➢ According to BE, 2 gifts must be indistinguished and may be simply
2 identical PEN. There will be 6 ways with which 2 PEN can be
distributed to 3 students.
➢ According to FD, 2 gifts must be indistinguished and may be simply
______________________. There will be 3 ways with which they can be
distributed to 3 students.
MB BE FD
X Y Z X Y Z X Y Z
PB -- -- PP -- -- O O --
P-PEN -- PB -- -- PP -- -- O O
B-BOOK -- -- PB -- -- PP O -- O
P B -- P P --
O-?
B P -- -- P P
-- P B P -- P
-- B P P-PEN
P -- B
B -- P
 Fundamentals of classical statistical mechanics
➢ Generalized coordinates: The state of the particle is described by
three position coordinates (x, y, z) and three momentum coordinates
(px, py, pz) in a dimensional space. These coordinates are called as
generalized coordinates. Thus, 6 generalized coordinates are required
to specify the state of the particle with 3 degree of freedom. This idea is
extended for the system of N-particles as 6N generalized coordinates
{qi, pi} required to specify the state of N-particles with 3N degree of
freedom.
➢ Phase space: The dimensional space made of positional space and
momentum space is called as phase space. 6-dimensional space is
required to specify the state of the particle by knowing its six
generalized coordinates (3 position coordinates, {x, y, z} and 3
momentum coordinates, {px, py, pz}). Such phase space is called as
micro space (µ-space) which deals with a single particle. 6N-
dimensional space is required to specify the state of the system of N-
particles by knowing its 6N generalized coordinates (3N position
coordinates qi and 3N momentum coordinates pi). Such phase space is
called as gamma space (Γ-space) which deals with all N-particles. The
smallest volume of phase space is micro space and largest volume of
phase space is gamma space.
Volume of micro space (µ-space), 𝒅𝝁 = 𝒅𝒙𝒅𝒚𝒅𝒛 𝒅𝒑𝒙 𝒅𝒑𝒚 𝒅𝒑𝒛

Volume of gamma space (Γ-space), 𝒅𝜞 = ම𝒅𝒙𝒅𝒚𝒅𝒛 ම𝒅𝒑𝒙 𝒅𝒑𝒚 𝒅𝒑𝒛

Unit cell or phase cell is the smallest volume of phase space (Γ-space)
and the smallest volume of the phase space is the volume of micro
space (µ-space).
Volume of unit cell = Volume of micro space (µ-space) = 𝒅𝒙𝒅𝒚𝒅𝒛 𝒅𝒑𝒙 𝒅𝒑𝒚 𝒅𝒑𝒛
Using Heisenberg’s uncertainty principle, Volume of unit cell = h3
Thus, the phase space is divided into number of unit cells each of volume h3.
Number of unit cells in the phase space is called as statistical weight (g).
𝒅𝜞 1 𝑽
𝒈 = 3 = 3 ම𝒅𝒙𝒅𝒚𝒅𝒛 ම𝒅𝒑𝒙 𝒅𝒑𝒚 𝒅𝒑𝒛 𝒈 = 3 ම𝒅𝒑𝒙 𝒅𝒑𝒚 𝒅𝒑𝒛
𝒉 𝒉 𝒉
Microstate and Macrostate:
The state of the system of N-particles is completely described by position
coordinates qi and momentum coordinates pi represented by a phase
point in the phase space. Such state of the system is called as
microstate which describes the state of individual particles of the
system. On the other hand, the state of the system defined by the
parameters N, V, and E, is called as macrostate.

Macrostate deals with external properties of the system, whereas,

microstate deals with internal properties of the system.
Example: The number of ways of filling 6 balls in 2 boxes
Macrostate Box Box Number of ways (Microstates) Mathematical
(Distribution) I II (Thermodynamic probability, W) probability, P
1 6 0 6C = 1
0 1/64 = 0.015
2 5 1 6C = 6
1 6/64 = 0.093
3 4 2 6C = 15
2 15/64 = 0.234
4 3 3 6C = 20
3 20/64 = 0.312
5 2 4 6C = 15
4 15/64 = 0.234
6 1 5 6C = 6
5 6/64 = 0.093
7 0 6 6C = 1
6 1/64 = 0.015
Total W = 64 Total P = 1
Thermodynamic probability (W): It is the number of microstates
corresponding to a macrostate or distribution.
Most probable distribution: It is the distribution or macrostate
corresponding to which the number of microstates or thermodynamic
probability is maximum.
Apriori probability: There is no any priority for all microstates or
thermodynamic probability corresponding to a macrostate. All probabilities
are equally accessible.
Thermodynamic probability (W) for MB, BE and FD statistics

W for MB statistics 𝑵 = 𝒏1 + 𝒏2 + 𝒏3 + ⋯ = ෍ 𝒏𝒊
N particles are 𝒊

distributed into n1, 𝑬 = 𝒏1 𝑬1 + 𝒏2 𝑬2 + 𝒏3 𝑬3 + ⋯ = ෍ 𝒏𝒊 𝑬𝒊

n2, n3, … ni groups 𝒊
corresponding to 𝑵! 𝒏
𝑾1 = 𝒈1 1
which unit cells 𝑵 − 𝒏1 ! 𝒏1 !
are g1, g2, g3,… gi
according to their 𝑵 − 𝒏1 ! 𝒏
𝑾2 = 𝒈2 2
respective energy 𝑵 − 𝒏1 − 𝒏2 ! 𝒏2 !
E1, E2, E3,…Ei.
𝑵 − 𝒏1 − 𝒏2 ! 𝒏
𝑾3 = 𝒈3 3
𝑵 − 𝒏1 − 𝒏2 − 𝒏3 ! 𝒏3 !
𝑾 = 𝑾1 𝑾2 𝑾3 … . .

𝑵! 𝒏 𝑵 − 𝒏1 ! 𝒏 𝑵 − 𝒏1 − 𝒏2 ! 𝒏
𝑾= 𝒈1 1 𝒈2 2 𝒈3 3 … . .
𝑵 − 𝒏1 ! 𝒏1 ! 𝑵 − 𝒏1 − 𝒏2 ! 𝒏2 ! 𝑵 − 𝒏1 − 𝒏2 − 𝒏3 ! 𝒏3 !
𝒏 𝒏 𝒏 𝒏
𝑵! 𝒈1 1 𝒈2 2 𝒈3 3 … . . 𝒈𝒊 𝒊
𝑾= = 𝑵! ෑ
𝒏1 ! 𝒏2 ! 𝒏3 ! … . 𝒏𝒊 !
𝒊
 W for BE statistics Considering an example of
indistinguished ni =15 particles filled in gi
= 6 unit cells. The number of ways of
filling ni particles in gi unit cells will be
dependent on transferring ni as well as
transferring the partitions (gi-1).
Therefore, thermodynamic probability
(Wi) will be given by
𝒏𝒊 + 𝒈𝒊 − 1 ! 𝒏𝒊 + 𝒈𝒊 − 1 !
𝑾𝒊 = =
𝒏𝒊 + 𝒈𝒊 − 1 − 𝒏𝒊 ! 𝒏𝒊 ! 𝒈𝒊 − 1 ! 𝒏𝒊 !
𝒏𝒊 + 𝒈𝒊 !
For very large ni and gi , 𝑾𝒊 =
𝒈𝒊 ! 𝒏𝒊 !
Therefore, 𝑾 = ෑ 𝑾𝒊 𝒏𝒊 + 𝒈𝒊 !
𝑾=ෑ
𝒊
𝒈𝒊 ! 𝒏𝒊 !
𝒊
W for FD statistics
According to restriction of occupancy of energy level in FD statistics, there
must be condition imposed for the indistinguishable particles as ni should
be less than or equal to gi. Using this condition, (Wi) will be given by
𝒈𝒊 ! 𝒈𝒊 !
𝑾𝒊 = Therefore, 𝑾 = ෑ 𝑾𝒊 𝑾=ෑ
𝒈𝒊 − 𝒏𝒊 ! 𝒏𝒊 ! 𝒈𝒊 − 𝒏𝒊 ! 𝒏𝒊 !
𝒊 𝒊
Example: I have 2 gifts to be distributed among 3 best students, verify
mathematically the number of ways to distribute gifts to students
according to MB, BE and FD statistics?
MB BE FD
X Y Z X Y Z
X Y Z
PB -- -- PP -- --
O O --
-- PB -- -- PP --
-- O O
-- -- PB -- -- PP
O -- O
P B -- P P --
B P -- -- P P
-- P B P -- P O-Vacancy
-- B P O-Vacancy
P-PEN, P-PEN
P -- B 𝒈𝒊 !
B -- P 𝒏𝒊 + 𝒈𝒊 ! 𝑾=ෑ
𝑾=ෑ 𝒈𝒊 − 𝒏𝒊 ! 𝒏𝒊 !
𝒈𝒊 ! 𝒏𝒊 ! 𝒊
P-PEN 𝒊
W = 10 W=3
B-BOOK
𝒏 Since ni and gi are very small
𝒈𝒊 𝒊 in this example, 1 can’t be
𝑾 = 𝑵! ෑ
𝒊
𝒏𝒊 ! neglected and hence;
W=9 𝒏𝒊 + 𝒈𝒊 − 1 !
𝑾=ෑ W=6
𝒈𝒊 − 1 ! 𝒏𝒊 !
𝒊
 Most probable distribution (ni) for MB, BE and FD statistics
Using Langrangian undermined multiplier method, 𝜕𝒍𝒏𝑾 + 𝜷𝜕𝑬 + 𝜶𝜕𝑵 = 0
Since W is
𝑵 = ෍ 𝒏𝒊 𝜕𝑵 = ෍ 𝜕𝒏𝒊 = 0 𝜶𝜕𝑵 = 𝜶 ෍ 𝜕𝒏𝒊 = 0 maximum for
𝒊 most probable
𝒊 𝒊
distribution,
𝑬 = ෍ 𝒏𝒊 𝑬𝒊 𝜕𝑬 = ෍ 𝜕𝒏𝒊 𝑬𝒊 = 0 𝜷𝜕𝑬 = 𝜷 ෍ 𝜕𝒏𝒊 𝑬𝒊 = 0
𝜕𝒍𝒏𝑾 = 0
𝒊 𝒊
𝒊 𝒏 𝒏
𝒈𝒊 𝒊 𝒈𝒊 𝒊
For MB statistics 𝑾 = 𝑵! ෑ 𝒍𝒏𝑾 = 𝒍𝒏 𝑵! ෑ
𝒏𝒊 ! 𝒏𝒊 !
𝒊 𝒊
𝒏 𝒏
But, 𝜕𝒍𝒏𝑾 = 0 𝒈𝒊 𝒊 𝒈𝒊 𝒊
𝜕 𝒍𝒏𝑵! + 𝒍𝒏 ෑ =0 𝜕 𝒍𝒏𝑵! + ෍ 𝒍𝒏 =0
𝒏𝒊 ! 𝒏𝒊 !
𝒊 𝒊
Using Stirling theorem, ln X! = X ln X - X, if X is very large
And,
𝒈𝒊
𝜕𝒍𝒏𝑾 + 𝜷𝜕𝑬 + 𝜶𝜕𝑵 = 0 𝒏𝒊 =
𝒆 𝜶+𝜷𝑬𝒊

𝒏𝒊 + 𝒈𝒊 ! 𝒈𝒊
For BE statistics 𝑾=ෑ 𝒏𝒊 =
𝒈𝒊 ! 𝒏𝒊 ! 𝒆 𝜶+𝜷𝑬𝒊 −1
𝒊
𝒈𝒊 ! 𝒈𝒊
For FD statistics 𝑾=ෑ 𝒏𝒊 = 𝜶+𝜷𝑬𝒊
𝒈𝒊 − 𝒏𝒊 ! 𝒏𝒊 ! 𝒆 +1
𝒊
Time average and Ensemble average
The phase space is a classical hypothetical concept which is the
combination of position and momentum space for representation of the
phase point (q, p) in the space. µ-space (6-dimensional space) is the
smallest volume of the phase space which deals with a single particle. Γ-
space (6N-dimensional space) deals with the system of N-particles.
In course of time, the phase
point moves in the space
traces a path or trajectory in
the phase space in
accordance with the laws of
mechanics by using equation
of motion in the Hamilton
canonical form as;
𝜕𝑯 𝜕𝑯
𝒒ሶ𝒊 = 𝒑ሶ𝒊 = −
𝜕𝒑𝒊 𝜕𝒒𝒊

Where, 𝒊 = 1, 2, … … , 3𝑵

H(q, p) is Hamiltonian
of the system
The phase point (q, p) has different phases depending on the variation of
time and the points on the phase trajectory gives different microstates of
the system. Classically, the average value of any physical quantity (say f)
is determined in the form of its time average which can be defined as
the average value of physical quantity corresponding to all phase
points lie on the trajectory over the period of time T.
1 𝑻
Time average, 𝒇ത = න 𝒇 𝒒, 𝒑 𝒅𝒕
𝑻 0
Determination of time average is very difficult as the system has N-
particles and posses very large degree of freedom.
Gibbs introduced the concept of ensemble with which the average value
can be determined easily called as ensemble average. According to Gibbs,
an ensemble is the collection of large number of systems which are
similar, independent, non-interacting with same macroscopic but
different microscopic states.
Classically, the average value of any physical quantity (f) is determined in
the form of its ensemble average which can be defined as the average
value of all macroscopically similar systems over an ensemble at a
fixed time.
Ensemble average, 𝒇 = න𝒇 𝒒, 𝒑 𝝆 𝒒, 𝒑 𝒅𝒒𝒅𝒑
Where, 𝝆 𝒒, 𝒑 is density distribution function
Ensembles
Ensemble is defined as the collection of a very large number of systems
which are similar, independent, non-interacting with macroscopically
same but microscopically different.
Micro-canonical Canonical Grand-canonical

According to Gibbs, these types of ensembles as statistical tools can be

used to study the macroscopic properties of the systems in terms of
microscopic properties.
Density distribution function or Probability density (ρ)
The function of position coordinates q, momentum coordinates p and
time t, i.e. 𝝆 = 𝝆 𝒒, 𝒑, 𝒕 represents the number density of the phase points
in the phase space is called statistical distribution function or density
distribution function or probability density.

Let dN be the number of phase points in the volume element dqdp of the
phase space, then
𝑵
𝒅𝑵 = 𝝆 𝒒, 𝒑, 𝒕 𝒅𝒒𝒅𝒑 By definition 𝝆 =
𝑽
Total number of phase points, 𝑵 = න𝝆 𝒒, 𝒑, 𝒕 𝒅𝒒𝒅𝒑

By definition of probability, let dW be the probability of finding the

number of phase points dN in volume element dqdp from total N.
𝒅𝑵 𝝆 𝒒, 𝒑, 𝒕 𝒅𝒒𝒅𝒑
𝒅𝑾 = =
𝑵 ‫𝒒 𝝆 ׬‬, 𝒑, 𝒕 𝒅𝒒𝒅𝒑
From normalization condition, න𝒅𝑾 = 1 This is possible only when
න𝝆 𝒒, 𝒑, 𝒕 𝒅𝒒𝒅𝒑 = 1
Thus, 𝒅𝑾 = 𝝆 𝒒, 𝒑, 𝒕 𝒅𝒒𝒅𝒑
𝝆 𝒒, 𝒑, 𝒕 satisfy the normalization condition and hence it is called as
probability density
Phase space and Quantum states
The phase space is a classical hypothetical concept which is the
combination of position and momentum space used for representation of
the phase point (q, p) in the space.
The quantum states is the quantum concept replacing unit cells or
phase cells used in classical theory.
The relation between classical concept of phase space and quantum
concept of quantum states can be obtained in quasi-classical case by
considering an example of a system as one dimensional harmonic
oscillator. Classically, Hamiltonian of one dimensional harmonic
oscillator is given by 𝒑2 1
𝑯 𝒒, 𝒑 = + 𝑲𝒒2
2𝒎 2
𝒑2 1
Therefore, the total energy of the system will be E= + 𝑲𝒒 2
2𝒎 2
𝒒2 𝒑2
2 + 2 =1 𝒙 2 𝒚2 2𝑬
2𝑬 2𝒎𝑬 + =1 𝒂= 𝒃 = 2𝒎𝑬
𝑲 𝒂2 𝒃2 𝑲
Elliptical phase trajectory as shown
2𝑬 2𝝅𝑬 2𝝅𝑬
𝑨 = 𝝅𝒂𝒃 𝑨=𝝅 2𝒎𝑬 𝑨= =
𝑲 𝝎
𝑲ൗ
𝒎
Where, 𝝎 is frequency of oscillator
Now, let us consider the oscillator in the energy range between E and
E+dE with the phase point confined to the region of phase space given by
2𝝅 2𝝅𝒅𝑬
𝒅𝑨 = 𝑬 + 𝒅𝑬 − 𝑬 =
𝝎 𝝎
According to quantum mechanics, the discreat energy of harmonic
oscillator is given by
1
𝑬𝒏 = ђ 𝒏 + Where, n = 0, 1, 2,.. is the vibrational quantum number
2
𝟏 𝟑 𝟓
𝑬𝟎 = ђ 𝑬𝟏 = ђ 𝑬𝟐 = ђ and so on..
𝟐 𝟐 𝟐
𝑬𝟏 − 𝑬𝟎 = 𝑬𝟐 − 𝑬𝟏 = 𝑬𝟑 − 𝑬𝟐 = ђ 𝒅𝑬 = ђ
Thus, the region of phase space between E and
E+dE is the volume of quantum state given by
𝒅𝑨 = 2𝝅ђ
Let the volume element of the phase space is dqdp and the number of
quantum states in this volume element is given by 𝒅𝒒𝒅𝒑
∆𝜞 =
2𝝅ђ
This quantization rule is extended for 3 degree of freedom of a particle
and then 3N degree of freedom for the system of N particles, we have
𝒅𝒒𝒅𝒑 𝒅𝒒𝒅𝒑 for 3N degree of freedom
∆𝜞 = for 3 degree of freedom; ∆𝜞 =
2𝝅ђ 3 2𝝅ђ 3𝑵
Liouville’s Theorem
The probability density 𝝆 = 𝝆 𝒒, 𝒑, 𝒕 is constant along the phase
trajectory of the phase points. This means the rate of change of probability
𝒅𝝆
density with respect to time is zero ( 𝒅𝒕 = 𝟎 ). This is also known as
“Principle of conservation of density distribution function in phase space”.
Another form of this theorem is “Principle of conservation of extension in
phase space”, this means that any extension in phase space with moving
𝒅
boundaries retains constant volume as time passes, i.e. (𝒅𝒕 𝝏𝜞 = 𝟎).
The total number of phase points in an
ensemble is conserved. Therefore, the total
number of phase points (N1) leaving any
volume ‘V’ in the phase space per second
must be equal to the rate of decrease of
the number of phase points (N2) in the
same volume.
ഥ ∙ 𝒅ഥ
𝒅𝑵1 = 𝝆 𝒗 ഥ∙𝒏
𝑺 =𝝆 𝒗 ෝ 𝒅𝑺
Now, Gauss’s divergence theorem

𝜕𝝆
Since, 𝑵1 = 𝑵2 + ∇ ∙ 𝝆ഥ
𝒗 =0
𝜕𝒕
 𝜕𝝆 This is the equation of continuity for the distribution
+ ∇ ∙ 𝝆ഥ
𝒗 =0 function in phase space. 𝛛𝑯 𝜕𝑯
𝜕𝒕 𝒒ሶ𝒊 = 𝒑ሶ𝒊 = −
𝛛𝒑𝒊 𝜕𝒒𝒊
Now,
𝜕 𝜕 𝜕 𝜕 𝜕𝝆 𝜕𝝆
∇ ∙ 𝝆ഥ𝒗 =෍ 𝝆𝒒ሶ 𝒊 + ሺ𝝆𝒑ሶ 𝒊 ൯ = ෍ 𝝆 𝒒ሶ𝒊 + ሺ𝒑ሶ𝒊 ൯ + 𝒒ሶ𝒊 + 𝒑ሶ𝒊
𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒒𝒊 𝜕𝒑𝒊
𝒊 𝒊
𝜕 𝜕𝑯 𝜕 𝜕𝑯 𝜕𝑯 𝜕𝝆 𝜕𝑯 𝜕𝝆 𝜕𝑯 𝜕𝝆 𝜕𝑯 𝜕𝝆
=෍ 𝝆 − + − =෍ − = 𝑯, 𝝆
𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒑𝒊 𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒒𝒊 𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒑𝒊 𝜕𝒒𝒊 𝜕𝒒𝒊 𝜕𝒑𝒊
𝒊 𝒊
𝑯, 𝝆 is Poisson-bracket
Therefore, 𝜕𝝆 + ∇ ∙ 𝝆ഥ
𝒗 =0
𝜕𝝆
+ 𝑯, 𝝆 = 0
𝜕𝒕 𝜕𝒕
Now, 𝜕𝝆 𝜕𝝆 𝜕𝝆
𝝆 = 𝝆 𝒒, 𝒑, 𝒕 Ξ 𝝆 𝒒𝒊, 𝒑𝒊, 𝒕 𝒅𝝆 = ෍ 𝒅𝒒 + 𝒅𝒑 + 𝒅𝒕
𝜕𝒒𝒊 𝒊 𝜕𝒑𝒊 𝒊 𝜕𝒕
𝒊
𝒅𝝆 𝜕𝝆 𝒅𝒒𝒊 𝜕𝝆 𝒅𝒑𝒊 𝜕𝝆 𝜕𝝆 𝜕𝝆 𝜕𝝆 𝜕𝝆 𝜕𝑯 𝜕𝝆 𝜕𝑯 𝜕𝝆
=෍ + + =෍ 𝒒ሶ𝒊 + 𝒑ሶ𝒊 + =෍ − +
𝒅𝒕 𝜕𝒒𝒊 𝒅𝒕 𝜕𝒑𝒊 𝒅𝒕 𝜕𝒕 𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒕 𝜕𝒒𝒊 𝜕𝒑𝒊 𝜕𝒑𝒊 𝜕𝒒𝒊 𝜕𝒕
𝒊 𝒊 𝒊

𝒅𝝆 𝜕𝝆 𝒅𝝆
Thus, = 𝑯, 𝝆 + =0
𝒅𝒕 𝜕𝒕 𝒅𝒕
𝒅
Verification of another theorem, 𝝏𝜞 = 𝟎.
𝒅𝒕
Let us consider 𝝏𝜞 be the region in the phase space and the number of
phase points in this region will be
𝒅 𝒅 𝒅 𝒅 𝒅 𝒅
𝜕𝑵 = 𝝆𝜕𝜞 𝜕𝑵 = 𝝆𝜕𝜞 = 𝝆 𝜕𝜞 + 𝜕𝜞 𝝆 = 𝝆 𝜕𝜞 = 0 𝜕𝜞 = 0
𝒅𝒕 𝒅𝒕 𝒅𝒕 𝒅𝒕 𝒅𝒕 𝒅𝒕
 Gauss’s divergence theorem

Thus,

𝜕 𝜕 𝜕
Divergence of a vector ഥ = 𝒊Ƹ
∇∙𝑭 + 𝒋Ƹ ෡
+𝒌 Ƹ 𝒙 + 𝒋𝑭
∙ 𝒊𝑭 ෡ 𝒛
Ƹ 𝒚 + 𝒌𝑭
𝜕𝒙 𝜕𝒚 𝜕𝒛

𝜕𝑭𝒙 𝜕𝑭𝒚 𝜕𝑭𝒛

ഥ=
∇∙𝑭 + +
𝜕𝒙 𝜕𝒚 𝜕𝒛

𝜕 𝜕
Thus, ∇ ∙ 𝝆ഥ
𝒗 =෍ + 𝝆 𝒒ሶ𝒊 + 𝒑ሶ𝒊
𝜕𝒒𝒊 𝜕𝒑𝒊
𝒊

𝜕 𝜕
∇ ∙ 𝝆ഥ
𝒗 =෍ ሶ
𝝆𝒒𝒊 + ሺ𝝆𝒑ሶ 𝒊 ൯
𝜕𝒒𝒊 𝜕𝒑𝒊
𝒊
Law of increase of Entropy
Let us consider the isolated system consists of two parts 1 and 2

If an isolated system is in non-equilibrium state, it does not have a

maximum value of entropy. The entropy of the system increases from the
state of non-equilibrium to equilibrium state at which it has maximum
entropy.
Thus, if the entropy of a closed or isolated system does not have its
maximum value at any time, then the entropy will increase or atleast
remain constant at a later time, but it will never decrease. This is said to
be law of increase of entropy.
Additive property of entropy
The number of quantum states in volume element dqdp of the phase
space for the system of N-particles with 3N degrees of freedom represents
the probability (∆𝜞) given by
𝒅𝒒𝒅𝒑
∆𝜞 =
2𝝅ђ 3𝑵
Entropy of the system satisfy the additive property as, 𝑺 = 𝑺1 + 𝑺2 + 𝑺3 + ⋯

Probability satisfy the multiplicative property as, ∆𝜞 = ∆𝜞1 ∙ ∆𝜞2 ∙ ∆𝜞3 ∙∙∙∙∙

According to Boltzmann, statistically probability and thermodynamically

entropy are analogous and corelated by S = k ln ∆Γ, in which (𝒍𝒏 ∆𝜞) is
used instead of (∆𝜞), as it satisfy additive property as that of entropy.

𝒍𝒏 ∆𝜞 = 𝒍𝒏 ∆𝜞1 ∙ ∆𝜞2 ∙ ∆𝜞3 ∙∙∙∙∙ 𝒍𝒏 ∆𝜞 = ෍ 𝒍𝒏 ∆𝜞𝒊

𝒊
𝑺 = 𝒌𝒍𝒏 ∆𝜞 𝑺 = 𝒌 ෍ 𝒍𝒏 ∆𝜞𝒊
𝒊

Thus, 𝑺 = 𝒌𝒍𝒏∆𝜞1 + 𝒌𝒍𝒏∆𝜞2 + 𝒌𝒍𝒏∆𝜞3 + ⋯ = 𝑺1 + 𝑺2 + 𝑺3 + ⋯

 Conditions for statistical equilibrium

The entropy is maximum when a closed system is in statistical

equilibrium. A system in thermal, mechanical and chemical equilibrium
is said to in statistical equilibrium. The entropy (S) of the system in
equilibrium depends on energy (E), number (N) and volume (V).
𝜕𝑺 𝜕𝑺 𝜕𝑺
S = S(E, V, N) 𝒅𝑺 = 𝒅𝑬 + 𝒅𝑽 + 𝒅𝑵
𝜕𝑬 𝜕𝑽 𝜕𝑵
𝝏𝑺𝟏 𝝏𝑺𝟏 𝝏𝑺𝟏 𝝏𝑺𝟐 𝝏𝑺𝟐 𝝏𝑺𝟐
𝒅𝑺𝟏 = 𝒅𝑬𝟏 + 𝒅𝑽𝟏 + 𝒅𝑵𝟏 and 𝒅𝑺𝟐 = 𝒅𝑬𝟐 + 𝒅𝑽𝟐 + 𝒅𝑵𝟐
𝝏𝑬𝟏 𝝏𝑽𝟏 𝝏𝑵𝟏 𝝏𝑬𝟐 𝝏𝑽𝟐 𝝏𝑵𝟐
𝜕𝑺1 𝜕𝑺1 𝜕𝑺1 𝜕𝑺2 𝜕𝑺2 𝜕𝑺2
𝒅𝑺 = 𝒅𝑺1 + 𝒅𝑺2 = 𝒅𝑬1 + 𝒅𝑽1 + 𝒅𝑵1 + 𝒅𝑬2 + 𝒅𝑽2 + 𝒅𝑵2
𝜕𝑬1 𝜕𝑽1 𝜕𝑵1 𝜕𝑬2 𝜕𝑽2 𝜕𝑵2
𝜕𝑺1 𝜕𝑺2 𝜕𝑺1 𝜕𝑺2 𝜕𝑺1 𝜕𝑺2
𝒅𝑺 = − 𝒅𝑬1 + − 𝒅𝑽1 + − 𝒅𝑵1
𝜕𝑬1 𝜕𝑬2 𝜕𝑽1 𝜕𝑽2 𝜕𝑵1 𝜕𝑵2
Since, 𝑺 = 𝑺1 + 𝑺2 = 𝑪𝒐𝒏𝒔𝒕𝒂𝒏𝒕 𝒅𝑺 = 𝒅𝑺1 + 𝒅𝑺2 = 0
𝜕𝑺1 𝜕𝑺2 𝜕𝑺1 𝜕𝑺2 𝜕𝑺1 𝜕𝑺2
− 𝒅𝑬1 + − 𝒅𝑽1 + − 𝒅𝑵1 = 0
𝜕𝑬1 𝜕𝑬2 𝜕𝑽1 𝜕𝑽2 𝜕𝑵1 𝜕𝑵2

𝜕𝑺1 𝜕𝑺2 𝜕𝑺1 𝜕𝑺2

− =0 = Thermal equilibrium
𝜕𝑬1 𝜕𝑬2 𝜕𝑬1 𝜕𝑬2

𝜕𝑺1 𝜕𝑺2 𝝏𝑺𝟏 𝝏𝑺𝟐

− =0 = Mechanical equilibrium
𝜕𝑽1 𝜕𝑽2 𝝏𝑽𝟏 𝝏𝑽𝟐

𝝏𝑺𝟏 𝝏𝑺𝟐 𝝏𝑺𝟏 𝝏𝑺𝟐

− =𝟎 = Chemical equilibrium
𝝏𝑵𝟏 𝝏𝑵𝟐 𝝏𝑵𝟏 𝝏𝑵𝟐
𝜕𝑺1 𝜕𝑺2
For Thermal equilibrium − 𝒅𝑬1 = 0
𝜕𝑬1 𝜕𝑬2
𝝏𝑺 𝟏 𝟏 𝟏
Let = − 𝒅𝑬𝟏 = 𝟎
𝝏𝑬 𝑻 𝑻𝟏 𝑻𝟐
𝟏 𝟏
If T1 > T2, then − 𝒅𝑬𝟏 > 𝟎 and it is possible when 𝒅𝑬𝟏 < 0,
𝑻𝟏 𝑻𝟐
This means that if T1 is increased, its energy decreased. Hence, T
must be temperature. 𝝏𝑬
𝑻=
𝝏𝑺
 𝜕𝑺1 𝜕𝑺2
For Mechanical equilibrium − 𝒅𝑽1 = 0
𝜕𝑽1 𝜕𝑽2
𝝏𝑺 𝝏𝑺 𝝏𝑬 𝟏 𝑷𝟏 𝑷𝟐
Let = = 𝑷 − 𝒅𝑽𝟏 = 𝟎
𝝏𝑽 𝝏𝑬 𝝏𝑽 𝑻 𝑻𝟏 𝑻𝟐
𝟏
If P1 > P2, and T1 = T2 then 𝑷 − 𝑷𝟐 𝒅𝑽𝟏 > 𝟎 and it is possible when
𝑻 𝟏
𝝏𝑬
𝒅𝑽𝟏 > 0, hence P must be pressure. 𝑷=
𝝏𝑽

𝜕𝑺1 𝜕𝑺2
For Chemical equilibrium − 𝒅𝑵1 = 0
𝜕𝑵1 𝜕𝑵2
𝝏𝑺 𝝏𝑺 𝝏𝑬 𝟏 µ𝟐 µ𝟏
Let = = ሺ−µ) − 𝒅𝑵𝟏 = 𝟎
𝝏𝑵 𝝏𝑬 𝝏𝑵 𝑻 𝑻𝟏 𝑻𝟐
𝟏
If µ1 > µ2, and T1 = T2 then µ − µ𝟏 𝒅𝑵𝟏 > 𝟎 and it is possible when
𝑻 𝟐
𝒅𝑵𝟏 < 0, hence µ must be chemical potential. 𝝏𝑬
µ=-
𝝏𝑵

The systems are said to be in statistical equilibrium, if their

temperature, pressure and chemical potential are same known as the
conditions for statistical equilibrium.
 Microcanonical Ensemble
Microcanonical Ensemble is defined as the
collection of a very large number of systems which
are microscopically different but macroscopically
similar with energy (E), volume (V) and number of
particles (N).
Since, the systems are thermally insulated, the
energy can not be exchanged among the systems. The
density distribution function (ρ) is constant along the
phase trajectory and zero outside the trajectory which
means it is confined by the energy surface of constant
energy Eo between the energies E and E+dE.
𝝆 𝒒, 𝒑 = 𝝆 𝑬 = 𝑪𝒐𝒏𝒔𝒕𝒂𝒏𝒕
It is lying in the range of energy from E to E+dE and has very
large value (ideally, ∞) at E = Eo and zero value at E ≠ Eo.

𝝆 𝒒, 𝒑 = 𝑪 𝜹 𝑬 − 𝑬0 𝜹 𝑬 − 𝑬0 is Dirac Delta function

This is Gibbs microcanonical distribution

න𝜹 𝑬 − 𝑬0 𝒅𝑬 = 1

න𝝆 𝒒, 𝒑 𝒅𝒒𝒅𝒑 = 1 𝜹 𝑬 − 𝑬𝟎 = ∞ at E=Eo
𝜹 𝑬 − 𝑬𝟎 = 𝟎 at E≠Eo
𝑪 න𝜹 𝑬 − 𝑬0 𝒅𝒒𝒅𝒑 = 1 1
= න𝜹 𝑬 − 𝑬0 𝒅𝒒𝒅𝒑
𝑪
 Entropy of perfect gas in Microcanonical Ensemble
In microcanonical ensemble, Hamiltonian system of perfect gas is the
kinetic energy of the system. 𝒑2
𝑯 𝒒, 𝒑 = 𝑬 𝒒, 𝒑 =
2𝒎
The number of quantum states in the volume element dqdp of the phase
space for the system of N particles
𝒅𝒒𝒅𝒑 𝒅𝒒𝒅𝒑 ‫𝒊𝒑𝒅 ׬ 𝒊𝒒𝒅 ׬‬ 𝑽𝑵
∆𝜞 = ∆𝜞 = = = න𝒅𝒑𝒊
2𝝅ђ 3𝑵 𝒉 3𝑵 𝒉 3𝑵 𝒉 3𝑵
R න𝒅𝒑1 is the space between two hyper spheres in 3-
dimensional space with radius p and p+dp
Volume of sphere
3
3
𝝅 ൗ2 3 4 𝝅 ൗ2 3
= 𝑹 = 𝝅𝑹3 න𝒅𝒑1 = 𝑷
3 3 3
2!
!
2
Like wise, the space between two hyper spheres in 3N-
dimensional space with radius 2𝒎𝑬 and 2𝒎 𝑬 + 𝒅𝑬
𝟑𝑵 3𝑵
𝝅 ൗ𝟐 𝟑𝑵 𝝅 ൗ2 3𝑵ൗ 2𝝅𝒎𝑬
3𝑵ൗ
2
න𝒅𝒑𝒊 = 𝟐𝒎𝑬 = 2𝒎𝑬 2 =
𝟑𝑵 3𝑵 3𝑵
𝟐 ! 2 ! 3𝑵 2 !
ൗ2
𝑽𝑵 2𝝅𝒎𝑬
∆𝜞 ==
3𝑵 𝒉2
!
2
Entropy of the system of perfect gas in microcanonical ensemble is given by
3𝑵ൗ
𝑽𝑵2𝝅𝒎𝑬 2
𝑺 = 𝒌 𝒍𝒏∆𝜞 = 𝒌 𝒍𝒏
3𝑵 𝒉 2
2 !
3ൗ
2𝝅𝒎𝑬 2 3𝑵
𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 − 𝒌𝒍𝒏 !
𝒉2 2

For very large number N, using Stirling theorem, lnN! = NlnN - N

3ൗ
2𝝅𝒎𝑬 2 3𝑵 3𝑵 3𝑵
𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 −𝒌 𝒍𝒏 −
𝒉2 2 2 2
3ൗ 3ൗ
2𝝅𝒎𝑬 2 3𝑵 2 3
𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 − 𝑵𝒌 𝒍𝒏 + 𝑵𝒌
𝒉2 2 2
3ൗ
2𝝅𝒎𝑬 2
𝑽 3 3ൗ 3ൗ
𝒉2 𝑬 2 4𝝅𝒎 2 3
𝑺 = 𝑵𝒌 𝒍𝒏 3ൗ + 𝑵𝒌
3𝑵 2 2 𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 + 𝑵𝒌
𝑵 3𝒉2 2
2
To check the additive property of this entropy expression, let us increase V,
N and E by the factor 2.

E’ = 2E
3ൗ 3ൗ
𝑬 2 4𝝅𝒎 2 3 V’ = 2V S’ = 2S
𝑺′ = 2𝑵𝒌 𝒍𝒏 2𝑽 + 2𝑵𝒌 N’ = 2N
𝑵 3𝒉2 2

3ൗ 3ൗ
𝑬 2 4𝝅𝒎 2 3
′
𝑺 = 2 𝑵𝒌 𝒍𝒏 𝑽 + 𝑵𝒌 + 2𝑵𝒌 𝒍𝒏2
𝑵 3𝒉2 2

𝑺′ = 2𝑺 + 2𝑵𝒌 𝒍𝒏2

This shows that the obtained expression of entropy does not satisfy
additive property of entropy. This is called as Gibbs paradox.

This expression of entropy is applicable for irreversible processes (mixing of

different gases) in which change in entropy dS > 0 but not applicable for
reversible processes (mixing of similar gases) as it gives dS > 0, instead of
dS = 0. Gibbs paradox can be resolved by correction in probability (∆𝜞).
 Correct entropy of perfect gas in Microcanonical Ensemble

∆𝜞 This is in accordance with the Boltzmann counting method for

∆𝜞 = indistinguishable particles due to which N! terms are repeated
𝑵!
and to be avoided in probability expression ∆𝜞.
𝟑𝑵ൗ
𝑵 𝟐
∆𝜞 𝑽 𝟐𝝅𝒎𝑬 𝟏
𝑺 = 𝒌 𝒍𝒏∆𝜞 = 𝒌 𝒍𝒏 = 𝒌 𝒍𝒏
𝑵! 𝟑𝑵 𝒉𝟐 𝑵!
!
𝟐
𝟑ൗ
𝟐𝝅𝒎𝑬 𝟐 𝟑𝑵
𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 − 𝒌𝒍𝒏 ! − 𝒌𝒍𝒏 𝑵!
𝒉𝟐 𝟐
𝟑ൗ
𝟐𝝅𝒎𝑬 𝟐 𝟑𝑵 𝟑𝑵 𝟑𝑵
𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 −𝒌 𝒍𝒏 − − 𝒌 𝑵𝒍𝒏𝑵 − 𝑵
𝒉𝟐 𝟐 𝟐 𝟐
𝟑ൗ 𝟑ൗ
𝟐𝝅𝒎𝑬 𝟐 𝟑𝑵 𝟐 𝟑
𝑺 = 𝑵𝒌 𝒍𝒏 𝑽 − 𝑵𝒌 𝒍𝒏 + 𝑵𝒌 − 𝑵𝒌 𝒍𝒏𝑵 + 𝑵𝒌
𝒉𝟐 𝟐 𝟐
𝟑ൗ
𝟐𝝅𝒎𝑬 𝟐 3ൗ
2
3ൗ
2
𝑽 𝟓 𝑽 𝑬 4𝝅𝒎 5
𝒉𝟐 𝑺 = 𝑵𝒌 𝒍𝒏 + 𝑵𝒌
𝑺 = 𝑵𝒌 𝒍𝒏 𝟑ൗ + 𝑵𝒌 𝑵 𝑵 3𝒉2 2
𝟑𝑵 𝟐 𝟐
𝑵 𝟐 This entropy satisfy additive property and has no
Gibbs paradox.
Since, 𝑬 = 3 𝑵𝒌𝑻 𝑽 2𝝅𝒎𝒌𝑻
3ൗ
2 5 Sackur-Tetrode
2 𝑺 = 𝑵𝒌 𝒍𝒏 + 𝑵𝒌 equation of S
𝑵 𝒉2 2
Canonical Ensemble
Canonical Ensemble is defined as the collection of a
very large number of systems which are
microscopically different but macroscopically similar
with temperature (T), volume (V) and number of
particles (N).
Since, the systems are thermally contacted, the
energy gets exchanged among the systems.
Let us consider an isolated system in equilibrium
with a heat reservoir (thermostat) with the constant
temperature.

For such composite system, microcanonical

distribution is applicable.
Let the probability dW that the system and reservoir
are in the volume element dΓ1 and dΓ2 respectively.

dW = C (E1+E2-Eo) dΓ1 dΓ2

Where, Eo =E1+E2 be the total energy of closed system

 The probability ρn that the system A is in definite quantum state of energy
En (nth quantum state). Therefore, E1=En and dΓ1 = 1.

𝝆𝒏 = 𝑪 න 𝜹 𝑬𝒏 + 𝑬𝟐 − 𝑬𝟎 𝒅𝜞𝟐 𝝆𝒏 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝜞𝟐

𝜞2 is the number of quantum state of energy E2 for the thermostat

𝑺2 𝑬2 ൗ
𝜕𝜞2 ∆𝜞2 𝒆 𝒌
𝜞2 = 𝜞2 𝑬2 𝒅𝜞2 = 𝒅𝑬2 = 𝒅𝑬2 = 𝒅𝑬2
𝜕𝑬2 ∆𝑬2 ∆𝑬2
𝑺𝟐 ሺ𝑬𝟐 )ൗ 𝑺𝟐 ሺ𝑬𝟎 −𝑬𝒏 )ൗ
𝒆 𝒌 𝒆 𝒌
𝝆𝒏 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝑬𝟐 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝑬𝟐
∆𝑬𝟐 ∆ሺ𝑬𝟎 − 𝑬𝒏 )
𝑺𝟐 ሺ𝑬𝟎 −𝑬𝒏 )ൗ 𝑺𝟐 ሺ𝑬𝟎 −𝑬𝒏 )ൗ
𝒆 From property
𝒌 𝒆 𝒌
𝝆𝒏 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝑬𝟐 = 𝑪 of Dirac-Delta
∆ሺ𝑬𝟎 − 𝑬𝒏 ) ∆ሺ𝑬𝟎 ) function
∆ 𝑬0 − 𝑬𝒏 = ∆𝑬0 = 𝑪𝒐𝒏𝒔𝒕𝒂𝒏𝒕
Now, using Taylor’s expansion, 𝑺𝟐 ሺ𝑬𝟎 − 𝑬𝒏 ) can be expanded about 𝑬𝟎
𝜕𝑺2 𝑬0 1
𝑺2 𝑬0 − 𝑬𝒏 = 𝑺2 𝑬0 − 𝑬𝒏 = 𝑺2 𝑬0 − 𝑬𝒏
𝜕𝑬0 𝑻
𝑬
ሾ𝑺2 𝑬0 − 𝒏 ቃ൘ 𝑺2 𝑬0 ൗ 𝑺2 𝑬 0 ൗ
𝑻 𝒆 𝒌
𝒆 𝒌 𝒆 𝒌 𝑬
− 𝒏 𝑬
− 𝒏 where, 𝑨 = 𝐶
𝝆𝒏 = 𝑪 =𝑪 𝒆 𝒌𝑻 𝝆𝒏 = 𝑨𝒆 𝒌𝑻 ∆ 𝑬0
∆ 𝑬0 ∆ 𝑬0
Gibbs canonical distribution
 From normalization condition,
𝑬 𝟏 𝑬
− 𝒏 − 𝒏 Quantum
෍ 𝝆𝒏 = ෍ 𝑨𝒆 𝒌𝑻 =𝟏 = ෍ 𝒆 𝒌𝑻
𝒏 𝒏
𝑨
𝒏

−
𝑬ሺ𝒒,𝒑) 𝟏 𝑬ሺ𝒒,𝒑) Classical
Or න 𝝆 𝒒, 𝒑 𝒅𝒒𝒅𝒑 = න 𝑨𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 =𝟏 −
= න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑
𝑨
𝟏
Reciprocal of normalization constant ( ) is called as canonical partition
𝑨
function (QN) which is defined as integral or sum over all the quantum
states. 𝑬
𝑬ሺ𝒒,𝒑) − 𝒏
−
𝑸𝑵 = න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 Or 𝑸𝑵 = ෍ 𝒆 𝒌𝑻
𝒏
In order to make canonical partition function (QN) to be dimensionless, and
in accordance with correct Boltzmann counting to avoid Gibbs paradox, it
is divided by h3N and N!
1 𝑬 𝒒𝒑 1 𝑬
− 𝒏
𝑸𝑵 = −
න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 Or 𝑸𝑵 = ෍ 𝒆 𝒌𝑻
𝑵! 𝒉3𝑵 𝑵! 𝒉3𝑵
𝒏
Classical canonical partition function Quantum canonical partition function

𝑬
− 𝒏
𝒆 𝒌𝑻 Gibbs canonical distribution in terms of
𝝆𝒏 = canonical partition function
𝑸𝑵
Grand Canonical Ensemble
Grand Canonical Ensemble is defined as the
collection of a very large number of systems which
are microscopically different but macroscopically
similar with temperature (T), volume (V) and chemical
potential (µ).
Since, the systems are thermally contacted and
permeable, the energy (E) and number (N) gets
exchanged among the systems.
Let us consider an isolated system in equilibrium
with a heat reservoir (thermostat) with the constant
temperature.
For such composite system, microcanonical
distribution is applicable.
Let the probability dW that the system and reservoir
are in the volume element dΓ1 and dΓ2 respectively.

dW = C (E1+E2-Eo) dΓ1 dΓ2

Where, Eo =E1+E2 be the total energy of closed system

 The probability ρnN that the system A is in definite quantum state of energy
En (nth quantum state) with the number of particles N. Therefore, E1=En ,
N1=N and dΓ1 = 1.
𝝆𝒏𝑵 = 𝑪 න 𝜹 𝑬𝒏 + 𝑬𝟐 − 𝑬𝟎 𝒅𝜞𝟐 𝝆𝒏𝑵 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝜞𝟐

𝜞2 is the number of quantum state of energy E2 for the thermostat

𝑺2 𝑬2 , 𝑵2 ൗ
𝜕𝜞2 ∆𝜞2 𝒆 𝒌
𝜞2 = 𝜞2 𝑬2 𝒅𝜞2 = 𝒅𝑬2 = 𝒅𝑬2 = 𝒅𝑬2
𝜕𝑬2 ∆𝑬2 ∆𝑬2
𝑺𝟐 ሺ𝑬2 , 𝑵2 )ൗ 𝑺𝟐 ሺ𝑬𝟎 −𝑬𝒏 , 𝑵𝟎 −𝑵)ൗ
𝒆 𝒌 𝒆 𝒌
𝝆𝒏𝑵 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝑬𝟐 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝑬𝟐
∆𝑬𝟐 ∆ሺ𝑬𝟎 − 𝑬𝒏 )
𝑺𝟐 ሺ𝑬𝟎 −𝑬𝒏 , 𝑵𝟎 −𝑵)ൗ 𝑺𝟐 ሺ𝑬𝟎 −𝑬𝒏 , 𝑵𝟎 −𝑵)ൗ
𝒆 𝒌 𝒆 𝒌 From property
𝝆𝒏𝑵 = 𝑪 න 𝜹 𝑬𝟐 − 𝑬𝟎 − 𝑬𝒏 𝒅𝑬𝟐 = 𝑪 of Dirac-Delta
∆ሺ𝑬𝟎 − 𝑬𝒏 ) ∆ሺ𝑬𝟎 ) function
∆ 𝑬0 − 𝑬𝒏 = ∆𝑬0 = 𝑪𝒐𝒏𝒔𝒕𝒂𝒏𝒕
Now, using Taylor’s expansion, 𝑺𝟐 ሺ𝑬𝟎 − 𝑬𝒏 , 𝑵𝟎 −𝑵) can be expanded about 𝑬𝟎 , 𝑵𝟎
𝜕𝑺2 𝑬𝟎 , 𝑵𝟎 𝜕𝑺2 𝑬𝟎 , 𝑵𝟎 𝑬𝒏 𝝁𝑵
𝑺𝟐 𝑬𝟎 − 𝑬𝒏 , 𝑵𝟎 −𝑵 = 𝑺2 𝑬𝟎 , 𝑵𝟎 − 𝑬𝒏 −𝑁 = 𝑺2 𝑬𝟎 , 𝑵𝟎 − +
𝜕𝑬0 𝜕𝑵0 𝑻 𝑻
𝑬 𝝁𝑵
𝑺2 𝑬0 ,𝑵0 − 𝒏 +
𝑻 𝑻 ൘
𝑺2 𝑬0 ,𝑵0 ൗ where,
𝒆 𝒌 𝒆 𝒌 𝑬
− 𝒏
𝝁𝑵
µ𝑵−𝑬𝒏 𝑺2 𝑬0 ,𝑵0 ൗ
𝝆𝒏𝑵 = 𝑪 =𝑪 𝒆 𝒌𝑻 𝒆 𝒌𝑻
∆ 𝑬0 ∆ 𝑬0 𝝆𝒏𝑵 = 𝑨𝒆 𝒌𝑻 𝒆 𝒌
𝑨=𝑪
Gibbs grand canonical ∆ 𝑬0
distribution
 From normalization condition,
𝝁𝑵−𝑬𝒏 1 𝝁𝑵−𝑬𝒏
෍ 𝝆𝒏𝑵 = ෍ 𝑨𝒆 𝒌𝑻 =1 = ෍𝒆 𝒌𝑻 Quantum
𝑨
𝒏,𝑵 𝒏,𝑵 𝒏,𝑵

𝝁𝑵−𝑬ሺ𝒒,𝒑) 1 𝝁𝑵−𝑬ሺ𝒒,𝒑)
Classical
Or න 𝝆 𝒒, 𝒑 𝒅𝒒𝒅𝒑 =න 𝑨𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 =1 = න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑
𝑨
𝟏
Reciprocal of normalization constant ( ) is called as grand canonical
𝑨
partition function (ZN) which is defined as integral or sum over all the
quantum states. 𝝁𝑵−𝑬𝒏
𝝁𝑵−𝑬ሺ𝒒,𝒑)
𝒁𝑵 = න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 Or 𝒁𝑵 = ෍ 𝒆 𝒌𝑻
𝒏,𝑵
In order to make grand canonical partition function (ZN) to be
dimensionless, and in accordance with correct Boltzmann counting to avoid
Gibbs paradox, it is divided by h3N and N!
1 𝝁𝑵−𝑬𝒏
1 𝝁𝑵−𝑬ሺ𝒒,𝒑)
𝒁𝑵 = 3𝑵 ෍ 𝒆
𝒌𝑻
𝒁𝑵 = න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 Or 𝑵! 𝒉
𝑵! 𝒉3𝑵 𝒏,𝑵
Classical grand canonical partition function Quantum grand canonical partition function

𝝁𝑵−𝑬𝒏
𝒆 𝒌𝑻 Gibbs grand canonical distribution in terms
𝝆𝒏𝑵 = of grand canonical partition function
𝒁𝑵
 Canonical and Grand Canonical Partition Function for perfect gas
𝒑2
Hamiltonian system of perfect gas, 𝑯 𝒒, 𝒑 = 𝑬 𝒒, 𝒑 =
2𝒎
Canonical partition function (QN) for perfect gas
1 −
𝑬 𝒒, 𝒑
𝑸𝑵 = 3𝑵 න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑
𝑵! 𝒉
𝒑2 2
σ𝒊 𝒊 1 𝒑
− σ𝒊 𝒊
1 − 2𝒎 𝑸𝑵 = න𝒅𝒒𝒊 න 𝒆 2𝒎𝒌𝑻 𝒅𝒑𝒊
𝑸𝑵 = න 𝒆 𝒌𝑻 𝒅𝒒𝒅𝒑 𝑵! 𝒉 3𝑵
𝑵! 𝒉3𝑵
2
𝑽𝑵 𝒑
− σ𝒊 𝒊
𝑸𝑵 = න 𝒆 2𝒎𝒌𝑻 𝒅𝒑𝒊
𝑵! 𝒉3𝑵
𝒑𝟐𝒙 𝟏
Let, =𝒖 𝒑𝒙 = 𝟐𝒎𝒖𝒌𝑻 𝒅𝒑𝒙 = 𝟐𝒎𝒌𝑻 𝒅𝒖
𝟐𝒎𝒌𝑻 𝟐 𝒖
𝒑2𝒙 ∞ 𝒑2𝒙 ∞ 1 ∞ 1
−
න 𝒆 2𝒎𝒌𝑻 𝒅𝒑𝒙 = 2න 𝒆−2𝒎𝒌𝑻 𝒅𝒑𝒙 = 2𝒎𝒌𝑇 න −𝒖 −2
𝒆 𝒖 𝒅𝒖 = 2𝒎𝒌𝑻 න 𝒆 𝒖2−1 𝒅𝒖
−𝒖
0 0 0

= 2𝒎𝒌𝑻 𝝅 = 2𝝅𝒎𝒌𝑇 1
𝜞 = 𝝅
Gamma Function 2
∞ ∞
𝜞 𝒏 + 1 = න 𝒆−𝒙 𝒙𝒏 𝒅𝒙 𝜞 𝒏 + 1 = 𝒏𝜞 𝒏 = 𝒏 න 𝒆−𝒙 𝒙𝒏−1 𝒅𝒙
0 0
𝜞 𝒏 + 1 = 𝒏!
 𝒑2𝒙
1ൗ
Therefore, න 𝒆− 2𝒎𝒌𝑻 𝒅𝒑𝒙 = 2𝝅𝒎𝒌𝑇 = 2𝝅𝒎𝒌𝑻 2

For 3 degree of freedom, For 3N degree of freedom,

𝒑2𝒙 +𝒑2𝒚 +𝒑2𝒛
− 3ൗ 𝒑2𝒊 3𝑁ൗ
න𝒆 𝒅𝒑𝒙 𝒅𝒑𝒚 𝒅𝒑𝒛 = 2𝝅𝒎𝒌𝑇 σ
2𝒎𝒌𝑻 2
න 𝒆− 𝒊2𝒎𝒌𝑻 𝒅𝒑𝒊 = 2𝝅𝒎𝒌𝑇 2

𝒑2
𝑽𝑵 − σ𝒊 𝒊 𝑽𝑵 3𝑁Τ
𝑸𝑵 = 𝑵!𝒉3𝑵 ‫𝒆 ׬‬ 2𝒎𝒌𝑻 𝒅𝒑𝒊 = 2𝝅𝒎𝒌𝑇 2
𝑵!𝒉3𝑵

3ൗ 𝑵
1 2𝝅𝒎𝒌𝑻 2
𝑸𝑵 = 𝑽
𝑵! 𝒉2

Grand Canonical partition function (ZN) for perfect gas

1 𝝁𝑵−𝑬𝒏
1 𝝁𝑵 𝑬 𝝁𝑵
𝒁𝑵 = ෍𝒆 𝒌𝑻 − 𝒏
𝑵! 𝒉3𝑵 = ෍𝒆 𝒌𝑻 ෍ 𝒆 𝒌𝑻 = ෍𝒆 𝒌𝑻 𝑸𝑵
𝒏,𝑵 𝑵! 𝒉3𝑵
𝑵 𝒏 𝑵

3ൗ 𝑵 3ൗ 𝑵
𝝁𝑵 1 2𝝅𝒎𝒌𝑻 2 2
𝒁𝑵 = ෍ 𝒆 𝒌𝑻 𝑽 1 𝝁 2𝝅𝒎𝒌𝑻
𝑵! 𝒉2 𝒁𝑵 = ෍ 𝑽𝒆𝒌𝑻
𝑵 𝑵! 𝒉2
𝑵
 Entropy of perfect gas in Canonical Ensemble
In thermodynamics, the entropy is fixed by energy independently of
whether the system is isolated as in microcanonical or in thermally contact
as in canonical. Hence, entropy of the system in canonical ensemble with
average energy <E> will be equal to the entropy of the system in
microcanonical ensemble with average energy <E>.
1
𝑺 = 𝒌 𝒍𝒏∆𝜞 𝝆 < 𝑬 > ∆𝜞 = 1 ∆𝜞 =
𝑬𝒏 <𝑬>
𝝆 <𝑬>
𝒆− 𝒌𝑻 𝒆− 𝒌𝑻 𝟏 <𝑬>
𝝆𝒏 = 𝝆 < 𝑬 > = ∆𝜞 = <𝑬> ∆𝜞 = 𝑸𝑵 𝒆 𝒌𝑻
𝑸𝑵 𝑸𝑵 −
𝒆 𝒌𝑻
𝑸𝑵
<𝑬> <𝑬>
𝑺 = 𝒌 𝒍𝒏 𝑸𝑵 𝒆 𝒌𝑻 𝑺 = 𝒌 𝒍𝒏𝑸𝑵 + 𝑻𝑺 −< 𝑬 > = 𝒌𝑻𝒍𝒏𝑸𝑵
𝑻
< 𝑬 > −𝑻𝑺 = −𝒌𝑻𝒍𝒏𝑸𝑵 𝑨 =< 𝑬 > −𝑻𝑺 𝒅𝑸 = 𝑻𝒅𝑺
𝑨 = −𝒌𝑻𝒍𝒏𝑸𝑵 𝒅𝑨 = 𝒅 < 𝑬 > −𝑻𝒅𝑺 − 𝑺𝒅𝑻 𝒅𝑸 = 𝒅𝑼 + 𝑷𝒅𝑽
𝒅𝑨 = 𝑻𝒅𝑺 − 𝑷𝒅𝑽 − 𝑻𝒅𝑺 − 𝑺𝒅𝑻 = 𝒅 < 𝑬 > +𝑷𝒅𝑽
𝒅𝑨 = −𝑺𝒅𝑻 − 𝑷𝒅𝑽 𝒅 < 𝑬 >= 𝑻𝒅𝑺 − 𝑷𝒅𝑽
𝝏𝑨 𝝏𝑨
𝑺=− 𝑷=−
𝝏𝑻 𝑽
𝝏𝑽 𝑻
 𝝏ሾ𝑻𝒍𝒏𝑸𝑵 ] 𝝏𝒍𝒏𝑸𝑵
𝑺=𝒌 𝑺 = 𝒌𝑻 + 𝒌𝒍𝒏𝑸𝑵
𝝏𝑻 𝝏𝑻
𝑽
3ൗ 𝑵
1 2𝝅𝒎𝒌𝑻 2
For perfect gas, 𝑸𝑵 = 𝑽
𝑵! 𝒉2
3ൗ 𝑵
1 2𝝅𝒎𝒌𝑻 2
3ൗ 𝑵
𝜕𝒍𝒏 𝑵! 𝑽 1 2𝝅𝒎𝒌𝑻 2
𝒉2
𝑺 = 𝒌𝑻 + 𝒌𝒍𝒏 𝑽
𝜕𝑻 𝑵! 𝒉2
𝟑𝑵ൗ 𝟑ൗ
𝝏𝒍𝒏 𝑻 𝟐 𝟐
𝟐𝝅𝒎𝒌𝑻
𝑺 = 𝒌𝑻 + 𝑵𝒌𝒍𝒏 𝑽 − 𝒌𝒍𝒏𝑵!
𝝏𝑻 𝒉𝟐
𝟑ൗ
𝟐
𝟑 𝝏𝒍𝒏𝑻 𝟐𝝅𝒎𝒌𝑻
𝑺 = 𝑵𝒌𝑻 + 𝑵𝒌𝒍𝒏 𝑽 − 𝒌 𝑵𝒍𝒏𝑵 − 𝑵
𝟐 𝝏𝑻 𝒉𝟐
3ൗ
3 1 2𝝅𝒎𝒌𝑻 2
𝑺 = 𝑵𝒌𝑻 + 𝑵𝒌𝒍𝒏 𝑽 − 𝑵𝒌𝒍𝒏𝑵 + 𝑵𝒌
2 𝑻 𝒉2
3ൗ
𝑽 2𝝅𝒎𝒌𝑻 2 5
𝑺 = 𝑵𝒌𝒍𝒏 + 𝑵𝒌 This is the same Sackur-Tetrode
𝑵 𝒉2 2 equation of entropy
 Other thermodynamic properties of perfect gas in Canonical Ensemble
𝝏𝑨 𝜕𝒍𝒏𝑸𝑵 𝑨 = −𝒌𝑻𝒍𝒏𝑸𝑵
Pressure, 𝑷=− 𝑷 = 𝒌𝑻
𝝏𝑽 𝑻
𝜕𝑽 𝑻
3ൗ 𝑵
1 2𝝅𝒎𝒌𝑻 2 𝑷𝑽 = 𝑵𝒌𝑻
𝜕𝒍𝒏 𝑵! 𝑽 𝟏
𝒉2 𝑷 = 𝑵𝒌𝑻 Equation of state
𝑷 = 𝒌𝑻 𝑽
𝜕𝑽 of system
𝑵
Energy, < 𝑬 > −𝑻𝑺 = −𝒌𝑻𝒍𝒏𝑸𝑵
𝟑ൗ
𝟏 𝑽 𝟐𝝅𝒎𝒌𝑻 𝟐
𝝏𝒍𝒏
𝜕𝒍𝒏𝑸𝑵 𝑵! 𝒉𝟐
< 𝑬 > = 𝑻 𝒌𝑻 + 𝒌𝒍𝒏𝑸𝑵 − 𝒌𝑻𝒍𝒏𝑸𝑵
𝜕𝑻 = 𝒌𝑻𝟐
𝝏𝑻
𝟑𝑵ൗ
𝝏𝒍𝒏 𝑻 𝟐 𝟑 𝟏 𝟑
𝟐 < 𝑬 > = 𝑵𝒌𝑻 𝟐 < 𝑬 > = 𝑵𝒌𝑻
< 𝑬 > = 𝒌𝑻 𝟐
𝝏𝑻 𝟐 𝑻

Specific heat,
𝜕<𝑬> 3
𝑪𝑽 = 𝑪𝑽 = 𝑵𝒌
𝜕𝑻 2
 Entropy of perfect gas in Grand Canonical Ensemble
In thermodynamics, the entropy is fixed by energy and number of particles
independently of whether the system is isolated as in microcanonical or in
thermally contact as in grand canonical. Hence, entropy of the system in
grand canonical ensemble with average energy <E> and average number
<N> will be equal to the entropy of the system in microcanonical ensemble
with average energy <E> average number <N>.
1
𝑺 = 𝒌 𝒍𝒏∆𝜞 𝝆 < 𝑬 >, < N > ∆𝜞 = 1 ∆𝜞 =
𝝆 < 𝑬 >, < N >
𝑬 µ𝑁 <𝑬> µ<𝑁>
− 𝒏 −
𝒆 𝒌𝑻 𝒆 𝒌𝑻 𝒆 𝒌𝑻 𝒆 𝒌𝑻 <𝑬> −µ<𝑁>
𝝆𝒏 = 𝝆 < 𝑬 >, < N > = ∆𝜞 = 𝒁𝑵 𝒆 𝒌𝑻 𝒆 𝒌𝑻
𝒁𝑵 𝒁𝑵
<𝑬> −µ<𝑁> <𝑬> µ<𝑁>
𝑺 = 𝒌 𝒍𝒏 𝒁𝑵 𝒆 𝒌𝑻 𝒆 𝒌𝑻 𝑺 = 𝒌 𝒍𝒏𝒁𝑵 + −
𝑻 𝑻
< 𝑬 > −𝑻𝑺 − µ < 𝑁 >= −𝒌𝑻𝒍𝒏𝒁𝑵 A − µ < 𝑁 >= −𝒌𝑻𝒍𝒏𝒁𝑵 𝑨 =< 𝑬 > −𝑻𝑺
Ω=A−µ<𝑁 >
Ω = −𝒌𝑻𝒍𝒏𝒁𝑵 𝒅Ω = 𝒅 < 𝑬 > −𝑻𝒅𝑺 − 𝑺𝒅𝑻− < 𝑵 > 𝒅µ
𝒅𝑸 = 𝑻𝒅𝑺
𝒅Ω = 𝑻𝒅𝑺 − 𝑷𝒅𝑽 − 𝑻𝒅𝑺 − 𝑺𝒅𝑻− < 𝑵 > 𝒅µ 𝒅𝑸 = 𝒅𝑼 + 𝑷𝒅𝑽
𝒅Ω = −𝑺𝒅𝑻 − 𝑷𝒅𝑽− < 𝑵 > 𝒅µ = 𝒅 < 𝑬 > +𝑷𝒅𝑽
𝒅 < 𝑬 >= 𝑻𝒅𝑺 − 𝑷𝒅𝑽
𝝏Ω 𝝏Ω 𝝏Ω
𝑺=− 𝑷=− < 𝑵 >= −
𝝏𝑻 𝝏𝑽 𝑻, µ
𝝏µ
𝑽, µ 𝑽, 𝑻
Thermodynamic properties of perfect gas in Grand Canonical Ensemble
3ൗ 𝑵
2
1 𝝁 2𝝅𝒎𝒌𝑻 𝑿𝑵 𝑿 𝑿2 𝑿3
𝒁𝑵 = ෍ 𝑽𝒆𝒌𝑻 ෍ =1+ + + + ⋯ = 𝒆𝑿
𝑵! 𝒉2 𝑵! 1! 2! 3!
𝑵 𝑵
𝝁 3ൗ
2𝝅𝒎𝒌𝑻 2
𝑽𝒆𝒌𝑻
=𝒆 𝒉2
3ൗ
𝝁 2𝝅𝒎𝒌𝑻 2 Ω = −𝒌𝑻𝒍𝒏𝒁𝑵
𝒍𝒏𝒁𝑵 = 𝑽𝒆𝒌𝑻
𝒉2
3ൗ
𝝁 2𝝅𝒎𝒌𝑻 2
𝝏 𝒍𝒏𝒁𝑵 𝜕 𝑽𝒆𝒌𝑻
𝝏Ω 𝒉2
< 𝑵 >= − < 𝑵 >= 𝒌𝑻 < 𝑵 >= 𝒌𝑻
𝝏µ 𝝏𝝁 𝜕𝝁
𝑽, 𝑻
3ൗ
3ൗ 2
1 𝝁 2𝝅𝒎𝒌𝑻 2 𝝁 2𝝅𝒎𝒌𝑻
< 𝑵 >= 𝒌𝑻𝑽𝒆𝒌𝑻 = 𝑽𝒆𝒌𝑻 = 𝒍𝒏𝒁𝑵
𝒌𝑻 𝒉2 𝒉2

Pressure, Energy, Specific heat,

𝟏 𝟑 3
𝑷 =< 𝑵 > 𝒌𝑻 𝑷𝑽 =< 𝑵 > 𝒌𝑻 < 𝑬 > = < 𝑵 > 𝒌𝑻 𝑪𝑽 = <𝑵>𝒌
𝑽 𝟐 2
Entropy,
𝝏ሾ𝑻𝒍𝒏𝒁𝑵 ] 𝝏𝒍𝒏𝒁𝑵
𝑺=𝒌 𝑺 = 𝒌𝑻 + 𝒌𝒍𝒏𝒁𝑵
𝝏𝑻 𝝏𝑻
𝑽 3ൗ 𝑵 3ൗ
2 𝝁 2𝝅𝒎𝒌𝑻 2
1 𝝁 2𝝅𝒎𝒌𝑻
𝒁𝑵 = ෍ 𝑽𝒆𝒌𝑻 𝒍𝒏𝒁𝑵 = 𝑽𝒆𝒌𝑻
𝑵! 𝒉2 𝒉2
𝑵 3ൗ 𝝁 3ൗ
2 𝜕{𝒆𝒌𝑻 3ൗ 2
2𝝅𝒎𝒌 𝑻 2} 𝝁 2𝝅𝒎𝒌𝑻
𝑺 = 𝒌𝑻𝑽 + 𝒌𝑽𝒆𝒌𝑻
𝒉2 𝜕𝑻 𝒉2
3ൗ 3ൗ
2𝝅𝒎𝒌 2 𝝁 𝟑 −µ 𝝁 𝝁 2𝝅𝒎𝒌𝑻 2
1ൗ 3ൗ
𝑺 = 𝒌𝑻𝑽 {𝒆𝒌𝑻 𝑻 2 + 𝑻 2 𝒆𝒌𝑻 } + 𝒌𝑽𝒆𝒌𝑻
𝒉2 𝟐 𝒌𝑻2 𝒉2
𝟑ൗ 𝟑ൗ 3ൗ
𝟑 𝝁 𝟐𝝅𝒎𝒌𝑻 𝟐 µ 𝝁 𝟐𝝅𝒎𝒌𝑻 𝟐2𝝅𝒎𝒌𝑻 𝝁 2
𝑺= 𝒌 𝑽𝒆𝒌𝑻 − 𝑽𝒆𝒌𝑻 + 𝒌𝑽𝒆𝒌𝑻
𝟐 𝒉𝟐 𝑻 𝒉𝟐 𝒉2
3 µ µ 5
𝑺 = < 𝑵 > 𝒌 − < 𝑵 >+< 𝑵 > 𝒌 = − <𝑵>+ <𝑵>𝒌
2 𝑻 𝑻 2
3ൗ 3ൗ
1 𝝁 2𝝅𝒎𝒌𝑻 2 𝑽 2𝝅𝒎𝒌𝑻 2
From, < 𝑵 >= 𝒌𝑻𝑽𝒆𝒌𝑻 𝝁 = −𝒌𝑻𝒍𝒏
𝒌𝑻 𝒉2 <𝑵> 𝒉2
3ൗ
𝑽 2𝝅𝒎𝒌𝑻 2 5
𝑺 =< 𝑵 > 𝒌𝒍𝒏 + <𝑵>𝒌 This is the same Sackur-
<𝑵> 𝒉2 2 Tetrode equation of entropy
with N=<N>
Fluctuations: Energy and density fluctuations
Fluctuations
It is the deviation of a physical quantity from its mean value.

Fluctuation of f ∆𝒇 = 𝒇 −< 𝒇 >

Average fluctuation of f < ∆𝒇 >=< 𝒇 −< 𝒇 > >= 𝟎

Mean square fluctuation of f < ∆𝒇 2 >=< 𝒇2 > −< 𝒇 >2

Root mean square fluctuation of f 𝝏𝒇 = < ∆𝒇 𝟐 >

Relative fluctuation of f 𝜕𝒇 < ∆𝒇 2 > 1

= ∝
<𝒇> <𝒇> 𝑵
f <f> ∆f <∆f> ∆𝒇 𝟐
< ∆𝒇 𝟐
> 𝝏𝒇 ∂f/<f>
1 15/5 -2 0 4 10/5 = 2 𝟐 𝟐/3
2 =3 -1 1
3 0 0
4 1 1
5 2 4
 Energy fluctuation in Canonical Ensemble
In canonical ensemble, the energy E of the system is not fixed and hence
the fluctuation in energy needs to be studied.
From the definition of ensemble average,

< 𝑬 >= න 𝝆 𝒒, 𝒑 𝑬 𝒒, 𝒑 𝒅𝒒𝒅𝒑 or < 𝑬 >= ෍ 𝝆𝒏 𝑬𝒏

𝒏
−𝑬𝒏 −𝑬𝒏 −𝑬𝒏
𝑬𝒏 𝒆 𝒌𝑻 𝑬𝒏 𝒆 𝒌𝑻 σ𝒏 𝑬𝒏 𝒆 𝒌𝑻
< 𝑬 >= ෍ < 𝑬 >= ෍ −𝑬𝒏 < 𝑬 >= −𝑬𝒏
𝑸𝑵
𝒏 𝒏 σ𝒏 𝒆 𝒌𝑻 σ𝒏 𝒆 𝒌𝑻
−𝑬𝒏 −𝑬𝒏 1 −𝑬𝒏 1 −𝑬 𝒏
𝜕<𝑬> σ𝒏 𝒆 𝒌𝑻 σ𝒏 𝑬𝒏 𝒆 𝒌𝑻 𝑬𝒏 2 − σ𝒏 𝑬𝒏 𝒆 𝒌𝑻 σ𝒏 𝑬𝒏 𝒆 𝒌𝑻
𝒌𝑻 𝒌𝑻2
=
𝜕𝑻 −𝑬𝒏 2
σ𝒏 𝒆 𝒌𝑻
−𝑬𝒏 −𝑬𝒏 −𝑬𝒏 2
1 2 1 −𝑬𝒏 −𝑬𝒏
σ
𝜕 < 𝑬 > 𝒌𝑻2 𝒏 𝒏𝑬 𝒆 𝒌𝑻 σ 𝑬
2 𝒏 𝒏 𝒆 𝒌𝑻 σ 𝑬
𝒏 𝒏 𝒆 𝒌𝑻 2
1 σ𝒏 𝑬𝒏 𝒆 𝒌𝑻 1 σ𝒏 𝑬𝒏 𝒆 𝒌𝑻
= − 𝒌𝑻 = − 2
𝜕𝑻 −𝑬𝒏 −𝑬𝒏 2 𝒌𝑻2 −𝑬𝒏 𝒌𝑻 −𝑬𝒏
σ𝒏 𝒆 𝒌𝑻 σ𝒏 𝒆 𝒌𝑻 σ𝒏 𝒆 𝒌𝑻 σ𝒏 𝒆 𝒌𝑻

𝝏<𝑬> 𝟏 𝟏 𝝏<𝑬> 𝟏 𝟐 >− <𝑬> 𝟐

𝟐 𝟐 = < 𝑬
= <𝑬 >− 𝟐 <𝑬>
𝝏𝑻 𝒌𝑻𝟐 𝒌𝑻 𝝏𝑻 𝒌𝑻𝟐
𝝏<𝑬> 𝟏 𝟐 >= 𝒌𝑻𝟐
𝝏<𝑬> 𝟐 𝑪
= < ∆𝑬 𝟐 > < ∆𝑬 = 𝒌𝑻 𝑽
𝝏𝑻 𝒌𝑻𝟐 𝝏𝑻
This is mean square fluctuation in energy in canonical ensemble
Relative fluctuation in energy

𝟑 𝟑
< ∆𝑬 > 𝟐 𝒌𝑻𝟐 𝑪𝑽 𝒌𝑻𝟐 𝟐 𝑵𝒌 𝟐𝑵
= = =
<𝑬> <𝑬> 𝟑 𝟑
𝟐 𝑵𝒌𝑻 𝟐𝑵

< ∆𝑬 𝟐 > 𝟏 𝟐 𝟏 1
= = ∝
<𝑬> 𝟑 𝟑 𝑵 𝑵
𝟐
𝑵

This shows that relative fluctuation in energy in canonical ensemble is

𝟏
proportional to 𝑵

Since the system has very large number of particles (N), therefore, the
relative fluctuation in energy in canonical ensemble is neglected. This
means the canonical ensemble is equivalent with microcanonical
ensemble and therefore, both the statistical tools provide identical
thermodynamic properties on account of neglecting relative
fluctuation in energy.
Density fluctuation in Grand Canonical Ensemble
𝝁𝑵−𝑬𝒏
σ𝒏,𝑵 𝑵𝒆 𝒌𝑻
From the definition of ensemble average, < 𝑵 >= 𝝁𝑵−𝑬𝒏
σ𝒏,𝑵 𝒆 𝒌𝑻

𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝑵 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝑵

𝜕<𝑵> σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝑵𝒆 𝒌𝑻 σ 𝑵𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻
= 𝒌𝑻 − 𝒏,𝑵 𝒌𝑻
𝜕𝝁 𝝁𝑵−𝑬𝒏 2 𝝁𝑵−𝑬𝒏 2
σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻

2
𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏
𝜕<𝑵> 1 σ𝒏,𝑵 𝑵2 𝒆 𝒌𝑻 1 σ𝒏,𝑵 𝑵𝒆 𝒌𝑻
= 𝝁𝑵−𝑬𝒏
− 𝝁𝑵−𝑬𝒏
𝜕𝝁 𝒌𝑻 𝒌𝑻
σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻

𝝏<𝑵> 𝟏 𝝏<𝑵>
= < 𝑵𝟐 > −< 𝑵 >𝟐 < 𝑵𝟐 > −< 𝑵 >𝟐 = 𝒌𝑻
𝝏𝝁 𝒌𝑻 𝝏𝝁
𝑵
Since, 𝝆=
𝝏<𝑵> 𝑽
< 𝝆𝟐 > −< 𝝆 >𝟐 = 𝒌𝑻 < 𝑵𝟐 >=< 𝝆𝟐 >
𝝏𝝁
< 𝑵 >𝟐 =< 𝝆 >𝟐
Pressure in grand canonical ensemble, 𝟑ൗ
𝝁 𝟐𝝅𝒎𝒌𝑻 𝟐
𝝏 𝑽𝒆𝒌𝑻
𝝏Ω 𝝏 𝒍𝒏𝒁𝑵 𝒉𝟐
𝑷=− = 𝒌𝑻 = 𝒌𝑻
𝝏𝑽 𝝏𝑽 𝝏𝑽
𝟑ൗ 3ൗ
𝟐 𝜕𝑷 𝝁 2𝝅𝒎𝒌𝑻 2 <𝑵>
𝝁 𝟐𝝅𝒎𝒌𝑻 <𝑵> = 𝒆𝒌𝑻 =
𝑷= 𝒌𝑻𝒆𝒌𝑻 = 𝒌𝑻 𝜕𝝁 𝒉2 𝑽
𝒉𝟐 𝑽
Let,
<𝑵> 𝟏 𝝏𝝁 𝝏𝝁 𝝏 < 𝑵 > 𝝏𝒙 𝝏𝝁 𝝏 < 𝑵 > 𝝏𝑷 𝝏𝑽
= =𝒙 =𝒙 = 𝒙 𝝏𝑽 𝝏𝒙
𝑽 𝒙 𝝏𝑷 𝝏 < 𝑵 > 𝝏𝒙 𝝏𝑷 𝝏 < 𝑵 > 𝝏𝒙
𝜕𝝁 𝑽 𝜕𝑷 𝝏𝝁 𝝏𝑷 𝒙𝟐
=𝒙 <𝑵> = −𝒙 <𝑵>
𝜕 < 𝑵 > −𝒙2 𝜕𝑽 𝝏<𝑵> 𝝏𝑽 𝑽
3
𝑽
𝝏𝝁 𝝏𝑷 𝒙𝟑 𝜕𝝁 𝜕𝑷 <𝑵>
=− <𝑵> =− <𝑵>
𝝏<𝑵> 𝝏𝑽 𝑽 𝜕<𝑵> 𝜕𝑽 𝑽

𝝏𝝁 𝝏𝑷 𝑽𝟐 𝝏<𝑵> 𝟏 𝝏𝑽
=− =− 𝟐 < 𝑵 >𝟐
𝝏<𝑵> 𝝏𝑽 < 𝑵 >𝟐 𝝏𝝁 𝑽 𝝏𝑷
𝟏 𝝏𝑽
𝝏<𝑵> <𝑵 >𝟐 𝑲𝑻 = − 𝑽 𝝏𝑷
𝝏<𝑵> 𝟏 𝝏𝑽 < 𝑵 >𝟐 = 𝑲𝑻
= − Isothermal
𝝏𝝁 𝑽 𝝏𝑷 𝑽 𝝏𝝁 𝑽 compressibility
This is mean square
𝒌𝑻
< 𝝆2 > −< 𝝆 >2 =< ∆𝝆 2 > = 𝑲𝑻 < 𝝆 >2 fluctuation in density in
𝑽 grand canonical ensemble.
 𝟏
To show that the relative fluctuation in density is proportional to
𝑵

𝟐
𝒌𝑻
< ∆𝝆 > = 𝑲𝑻 < 𝝆 >𝟐
𝑽
< ∆𝝆 𝟐 > 𝒌𝑻
= 𝑲𝑻
< 𝝆 >𝟐 𝑽

< ∆𝝆 𝟐 > < ∆𝝆 𝟐 > 𝒌𝑻

= = 𝑲𝑻
< 𝝆 >𝟐 <𝝆> 𝑽 𝑵𝒌𝑻
𝑷=
𝑽
< ∆𝝆 𝟐 > 𝒌𝑻 𝟏 𝝏𝑽 𝝏𝑷 𝑵𝒌𝑻
= − =− 𝟐
<𝝆> 𝑽 𝑽 𝝏𝑷 𝝏𝑽 𝑽

< ∆𝝆 𝟐 > 𝒌𝑻 𝟏 𝒌𝑻 𝟏 𝟏
= − = − =
<𝝆> 𝑽𝟐 𝝏𝑷 𝑽𝟐 𝑵𝒌𝑻 𝑵
𝝏𝑽 −𝑽 𝟐

This shows that the relative fluctuation in density in grand canonical

ensemble is neglected for the system of large number of particles N.
Energy fluctuation in Grand Canonical Ensemble
𝝁𝑵−𝑬𝒏
σ𝒏,𝑵 𝑬𝒏 𝒆 𝒌𝑻
From the definition of ensemble average, < E >= 𝝁𝑵−𝑬𝒏
σ𝒏,𝑵 𝒆 𝒌𝑻

𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝑵 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝑵

𝜕 < E > σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝑬𝒏 𝒆 𝒌𝑻 σ 𝑬 𝒆
𝒌𝑻 − 𝒏,𝑵 𝒏
𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻
𝒌𝑻
= 2 2
𝜕𝝁 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏
σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻

𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏

𝝏<𝑬> 𝟏 σ𝒏,𝑵 𝑬𝒏 𝑵𝒆 𝒌𝑻 𝟏 σ𝒏,𝑵 𝑬𝒏 𝒆 𝒌𝑻 σ𝒏,𝑵 𝑵𝒆 𝒌𝑻
= 𝝁𝑵−𝑬𝒏
− 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏
𝝏𝝁 𝒌𝑻 𝒌𝑻
σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻

𝝏<𝑬> 𝟏
= < 𝑬𝑵 >−< 𝑬 >< 𝑵 >
𝝏𝝁 𝒌𝑻
 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵 − 𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵 − 𝑬𝒏
𝜕 < 𝑬 > σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝑬𝒏 𝒆 𝒌𝑻 σ 𝑬 𝒆
−𝒌𝑻2 − 𝒏,𝑵 𝒏
𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻
−𝒌𝑻2
= 2 2
𝜕𝑻 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏
σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻

𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏 𝝁𝑵−𝑬𝒏

1 σ 𝑬𝒏 𝝁𝑵𝒆 𝒌𝑻 −σ𝒏,𝑵 𝑬𝒏 2 𝒆 𝒌𝑻 1 σ 𝑬𝒏 𝒆 𝒌𝑻 σ𝒏,𝑵 𝝁𝑵𝒆 𝒌𝑻 −σ𝒏,𝑵 𝑬𝒏 𝒆 𝒌𝑻
= −𝒌𝑻2 𝒏,𝑵 𝝁𝑵−𝑬𝒏 + 𝒌𝑻2 𝒏,𝑵 𝝁𝑵−𝑬𝒏 2
σ𝒏,𝑵 𝒆 𝒌𝑻 σ𝒏,𝑵 𝒆 𝒌𝑻
𝝏<𝑬> 𝟏 𝟐
𝟏
= 𝟐
𝝁 < 𝑬𝑵 >−< 𝑬 > + 𝟐
𝝁 < 𝑬 >< 𝑵 > −< 𝑬 >𝟐
𝝏𝑻 −𝒌𝑻 𝒌𝑻
𝝏<𝑬> 𝝁 𝟏
= 𝟐
< 𝑬 >< 𝑵 >−< 𝑬𝑵 > + 𝟐
< 𝑬 >𝟐 −< 𝑬𝟐 >
𝝏𝑻 𝒌𝑻 𝒌𝑻
𝝏<𝑬> 𝝁 𝝏<𝑬> 𝟏
= 𝟐
𝒌𝑻 + 𝟐
< ∆𝑬 𝟐 >
𝝏𝑻 𝒌𝑻 𝝏𝝁 𝒌𝑻
𝝏<𝑬> 𝝏<𝑬> 𝝏𝑬
< ∆𝑬 𝟐 >= 𝒌𝑻𝟐 − 𝝁𝒌𝑻 𝝁=−
𝝏𝑻 𝝏𝝁 𝝏𝑵
𝟐
𝝏 <𝑬> 𝝏 <𝑬>𝝏<𝑵>
< ∆𝑬 >= 𝒌𝑻𝟐 𝑪𝑽 − 𝒌𝑻 −
𝝏 < 𝑵 > 𝝏 < 𝑵 > 𝝏𝝁
𝟐 𝟐
𝝏<𝑬> 𝝏<𝑵> 𝟐
𝝏<𝑬>
< ∆𝑬 𝟐
>= 𝒌𝑻𝟐 𝑪𝑽 + 𝒌𝑻 < ∆𝑬 >= 𝒌𝑻𝟐 𝑪𝑽 +< ∆𝑵 𝟐
>
𝝏<𝑵> 𝝏𝝁 𝝏<𝑵>

𝝏𝑬
𝟐 This is mean square
𝟐
< ∆𝑬 >= 𝒌𝑻𝟐 𝑪𝑽 +< ∆𝑵 𝟐
> fluctuation in energy in grand
𝝏𝑵 canonical ensemble.
Thermodynamic equivalence of ensembles
With the idea of ensemble theory, microcanonical ensemble deals
closed isolated systems, canonical ensemble deals closed isothermal
systems and grand canonical ensemble deals open isothermal systems.
Since, fluctuation in E and N in case of microcanonical ensemble for
closed isolated systems is meaningless. But relative fluctuation in E in
case of canonical ensemble for closed isothermal systems is neglected
for the systems of very large number of particles. As well as relative
fluctuation in E and N in case of grand canonical ensemble for open
isothermal systems are neglected for the systems of very large number
of particles.

This shows that all three ensembles gives identical results for all
thermodynamic properties like S, E, P, CV etc. keeping in view of
considering the energy E as <E> and number N as <N>. This is known
as thermodynamic equivalence of ensembles.

But, on account of both E and N involved in grand canonical ensemble,

it is the most convenient to apply for the determination of
thermodynamic properties of the systems of ideal as well as real gases.