M.SC.

PHYSICS – I YEAR
DKP11 : CLASSICAL AND STATISTICAL MECHANICS
SYLLABUS

Unit 1: Lagrangian and Hamiltonian formulations

Hamilton’s principle - Derivation of Lagrange’s equations from Hamilton’s principle -
Principle of Least Action and its applications, Canonical Transformation : The Hamiltonian
Formalism, Canonical formalism, Hamiltonian equations of motion, Cyclic coordinates,
Rauthian procedure and equations, Derivation of Generating functions, examples, properties,
Derivation of Hamiltonian equations from variational principle.

Unit 2: Poisson bracket and theory of small oscillations

Poisson bracket, Special cases of Poisson bracket , Poisson theorem, Poisson bracket and
canonical transformation, Jacobi identity and its derivation, Lagrange bracket and its
properties, the relationship between Poisson and Lagrange brackets and its derivation, the
angular momenta and Poisson bracket, Liouville’s theorem and its applications; Theory of
small oscillations:
Formulation of the problem, Eigenvalue equation and the principle axis transformation,
frequencies of free vibration and normal coordinates, free vibrations of a linear triatomic
molecule

Unit 3: Two - body central force problem and H - J theory

Two body central force problem: Reduction to the equivalent one body problem, the equation
of motion and first integrals, classification of orbits, the virial theorem, the differential
equation for the orbit, integral power law in time in the Kelper’s problem ,scattering in
central force field;
H-J Theory: H-J equation and their solutions, use of H-J method for the solution of harmonic
oscillator problem, Hamilton’s principle function, Hamilton’s characteristic function and
their properties, Action angle variable for completely separable systems, the Kelper’s
problem in action angle variables

Unit 4: Classical Statistical Mechanics

Foundation of Statistical Mechanics: The macroscopic and microscopic states, postulate of

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

1
equal a priori probability, Contact between statistics and thermodynamics; Ensemble theory:
Concept of ensemble, phase space, Density function, Ensemble average, Liouville’s
theorem, Stationary ensemble; The microcanonical ensemble, Application to the classical
ideal gas; The canonical and grand canonical ensembles, Canonical and grand canonical
partition functions, Calculation of statistical quantities; Thermodynamics of a system of non-
interacting classical harmonic oscillators using canonical ensemble, and of classical ideal gas
using grand canonical ensemble, Energy and density fluctuations; Entropy of mixing and the
Gibb’s paradox, Sackur-Tetrode equation .

Unit 5: Quantum Statistical Mechanics

Quantum-mechanical ensemble theory: Density matrix, Equation of motion for density
matrix, Quantum- mechanical ensemble average; Statistics of indistinguishable particles,
Two types of quantum statistics- Fermi-Dirac and Bose-Einstein statistics, Fermi-Dirac and
Bose-Einstein distribution functions using microcanonical and grand canonical ensembles
(ideal gas only), Statistics of occupation numbers; Ideal Bose gas: Internal energy, Equation
state, Bose-Einstein Condensation and its critical conditions; Bose-Einstein condensation in
ultra-cold atomic gases: its detection and thermodynamic properties: Ideal Fermi gas:
Internal energy, Equation of state, Completely degenerate Fermi gas.

Books for Study and Reference

1.Classical Mechanics (3rd ed.,2002) by H. Goldstein, C.Poole and J. Safko, Pearson Edition
2. Classical Mechanics - J. C. Upadhyaya- Second Edition-2005-Himalaya Publishing House
3.Classical Mechanics - G. Aruldhas-2008-PHI Learning Pvt.Ltd.
4.Classical Mechanics-A Text Book-Suresh Chandra-Narosa Publications
5.Statistical Mechanics by R. K. Pathira (2nd edition)
6.Statistical Mechanics by R.K. Pathira and P.D. Beale (3rd edition)
7.Statistical Mechanics by K.Huang
8.Statistical Mechanics by L.D.Landau and I.M.Lifshitz

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

2
UNIT 1: LAGRANGIAN AND HAMILTONIAN FORMULATIONS
Hamilton’s principle - Derivation of Lagrange’s equations from Hamilton’s principle -
Principle of Least Action and its applications, Canonical Transformation : The Hamiltonian
Formalism, Canonical formalism, Hamiltonian equations of motion, Cyclic coordinates,
Rauthian procedure and equations, Derivation of Generating functions, examples,
properties, Derivation of Hamiltonian equations from variational principle.

LAGRANGIAN AND HAMILTONIAN FORMULATIONS

Hamilton’s principle - Derivation of Lagrange’s equations from Hamilton’s principle -
Principle of Least Action and its applications, Canonical Transformation : The Hamiltonian
Formalism, Canonical formalism, Hamiltonian equations of motion, Cyclic coordinates,
Rauthian procedure and equations, Derivation of Generating functions, examples, properties,
Derivation of Hamiltonian equations from variational principle.
1.1 HAMILTON’S PRINCIPLE:
The motion of the system from time t1 to time t2 is such that the line integral
t2

I=  Ldt
t1

Where L = T – V, is an extremum for the path of motion

t2

(or) I =   Ldt  0
t1

 is the variation symbol.

Deduction of Lagrange’s equations:
Consider a conservative system of particles. The integral can be written as

 T q , q   V q dt.
t2

j j j
t1

According to Hamilton’s variational principle, we have

  T q j , q j   V q j  dt = 0
t2

t1

t2
 T T  V 
 
t j  q j j q j j  q j j  dt = 0
  q  q  q
1   
 T V 
q j dt   T q j dt  0
t2 t2

t j  q j  q j  t j q j
1   1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

3
  T V 
q j dt   T d q j  dt  0
t2 t2

t j  q j  q j  t q j dt
1   1

Integrating by parts the second term, we get

t2
 T V  T
t2
d  T 
t j  q j  q j q j dt  j q j q j  q j dt  0
t2

1  
t1
t1
dt  q j 

There is no coordinate variation at end points and hence
t2
q j t1
 0.

Now equation reduces to

 T V   
q j dt   d  T
t2 t2

t j  q j  q j

 t j dt  q j q j dt  0

1   1  
t2
 T V d  T 
t  q j  q j  dt  q j
 q j dt  0

1   
each δqj are independent of each other, the coefficient of every δqj should be equated to
zero. And we get
 T V d  T 
      0

 j
q q j dt  q j 


 d  T   
  
 q
(T  V )  0
 dt  q j
  j 
V is not a function of q j and therefore

d  (T  V )  
  (T  V )  0
dt  q j  q j

T – V = L = Lagrangian for a conservation system

L = Scalar function of q j , q j and t

 d  L  L 
   0 J = 1, 2,......n.
 dt  q j  q j 
This set of equations is called Lagrange’s equations of motion and each is a second
order differential equation in terms of the time as independent variable.
1.2 DERIVATION OF LAGRANGE’S EQUATIONS FROM HAMILTON’S
PRINCIPLE:
The Lagrangian L is a function of generalized co-ordinates q j ′s and generalized
velocities 𝑞 j ′s and time t.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
4
 That is L  L[q j (t ), q j (t ), t ]

According to Hamilton’s variational principle, motion of a conservative system from finite

time t1 to time t2 is such that the line integral
t2

I=  L[q j (t ), q j (t ), t ]dt
t1

t2

is zero. That is I    L[q j (t ), q j (t ), t ]dt  0

t1

If the Lagrangian does not depend on time t explicitly, then the variation δL can be written as
n L L
L   q j  q j
j 1 q q j
j

Integrating both sides from t = t1 to t = t2

L L
t2 t2 2 t

t Ldt  t j q j
 q j dt  t j q j q j dt  0
1 1 1

But from Hamilton’s principle

t2

 Ldt  0
t1

L L
t2 2 t

Therefore t j q j j
 q dt  t j q j q j dt  0
1 1

d
q j  (q j )
dt
Integrating by parts the second term, we get
t2
t2
L  L  t2
d  L 
 q j dt
t j q j j
 q dt   
j q
q    

 t1 t1 j dt  q j
j
1  j 

There is no coordinate variation at end points and hence

t2
 L 
 q j   0.
j q
 j  t1

Now equation reduces to

L d  L 
t2 t2
 q j dt  0
t j q j j t j dt  q j
 q dt  

1 1  
(or)

d  L 
 q j dt   L q j dt  0
t2 t2

t j dt  q j

 t j q j
1   1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

5
 t2
 d  L  L 
t j  dt  q j
 
 q  j

q dt  0
1    j 

For holonomic system the generalized co-ordinates δq j are independent of each other.
Therefore the coefficients of each δq j must vanish. And we get
 d  L  L 
   0 J = 1, 2,......n.
 dt  q j  q j 
This set of equations is called Lagrange’s equations of motion and each is a second
order differential equation in terms of the time as independent variable.
1.3 PRINCIPLE OF LEAST ACTION:
The time integral of twice the K.E is called the action. The principle of least action
states that
t2

  2T dt  0
t1

But in systems for which H remains constant

2T   p j q j
j

t2

    p j q j dt  0
j
t1

 represents variation of the path which allows time

as well the position coordinates to vary.
In  variation
1. time as well as the position coordinates are allowed to vary.
2. time t varies even at the end points.
3. the position coordinates are held fixed at the end points. ie q j = 0
Let APB be the actual path and AP′B be the varied path. The end points A and B after
time t take the positions A and B such that A and B are fixed while time is not fixed.
A point P on the actual path gives P on the varied path.
q jq j= q j + q j
If  is the variational parameter, then in δ process t is independent of . But in  process t is
a function of  even at the end points. ie t = t (α).
q j depends on t and 
dq j
q j  d
d

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

6
  q q t 
 j  j d
  t  

q j t
 d  q j d
 
 q j  q j  t

q j t
Since  q j  d and q j d  q j  t
 
Any function f = f (q j , q j , t )

 f f f 
f =   q j  q j  t 
j q q j t 
 j

f
 q j  q j t    f q j  qj t   f t
j q j q j t

f
 f  t
t

 =   t ....(1)
t
Here  operation and time differentiation cannot be interchanged.
Proof:
t2

A =   p q
t1
j
j j dt

t2

=  ( L  H ) dt
t1

t2

=  L dt  H (t
t1
2  t1 ) [since H is conserved] ....(2)

t2

A =   L dt  H  (t 2  t1 )
t1

t2 t2

=   L dt  H  t ....(3)
t1 t1

t2

To Solve   L dt :
t1

t2

Let  L dt
t1
=I so that I  L

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

7
 Now  I = δI + I t
t2 t2 t2
ie   L dt   L dt  L t .....(4)
t1 t1 t1

Substituting equation (4) in equation (3) we get

t2 t2 t2
A   L dt  L t |  H t | .......(5)
t1 t1 t1

t2

  L dt cannot be zero.
t1

t2 t2
 L L 
  L dt     q j  q j  dt
t1 t1
j
 q j q j 
 d  L  
q j  L d (q j ) dt
t2

=   
j dt  q  q j dt
t1    j  

L d  L 
[Since from Lagrange’s equation of motion   ]
q j dt  q j 

t2 t2
 d  L 
  L dt  t j  dt  q j j  dt
   q
t1 1   

Putting q j  q j  q j t
t2
 d  L L
t2

  L dt      q j  q j t  dt
j dt  q q j 
t1 t1    j 
t2
 L L 
=  q j  q j t 
j  q q j 
 j t1

At end points q j = 0 . Therefore

L
t2 t2
  L dt    q j t |
t1
j q j t1

t2
=   p j q j t |
j t1

Now equation (5) becomes

t2 t2 t2
A    p j q j t |  L t |  H t |
j t1 t1 t1

 
=  H  L   p j , q j  t |
2 t

 j  t1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

8
 = 0.
Since H =  p j , q j  L
j

t2

Thus A     p j , q j dt  0
t1

Which proves the principle of least action.

t2

Here   p, q
t1
j  Hamiltion’s characteristic function

(i) Principle of least action in terms of arc length of the particle trajectory.
Let a system contain only one particle of mass m
2

Kinetic energy T  m  
1 ds
2  dt 
ds element of arc traversed in time dt
1

 m
2

dt =   ds
 2T 
The principle of least action
1
t2 t2
m
2

  2T dt    2T   ds  0
t1 t1  2T 
t2

ie   2m T 2 ds  0
1

t1

t2 1

  [2m( E  V )] 2 ds  0 Since T +V = E
t1

t2 1

  [2m( H  V )] 2 ds  0 Since E = H
t1

t2 1

  [ H  V ] 2 ds  0 Since m = Constant ....(6)

t1

The above equation represents the principle of least action in term of arc length of the
particle trajectory.
(ii) Jacobi’s form of the principle of least action:
The K.E of the system
1
T   a jk q j q k
2 j ,k
2T   a jk q j q k
j ,k

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

9
 dq j dqk
  a jk
j ,k dt 2

d 
2

=   ....(7)
 dt 

Where d 2   a jk d qj dqk , a differenti al

j ,k

From equation (7) T   

1 dp
2  dt 
d
dt =
2T 
1
2

The principle of least action

t2

  2T dt  0
t1

d
t2

  2T 1 0
t1 (2T ) 2
t2

  2T 2 d  0
1

t1

t2

  2E  V 2 d  0
1

t1

t2

  2H  V 2 d  0
1

t1

t2

  H  V 2 d  0
1
....(8)
t1

This equation gives the Jacobi’s form of principle of least action.

(iii) Fermat′s principle:
It states that the time taken by a light ray to travel between two points is extremum.
According to principle of least action
t2

  2T dt  0
t1

t2

  T dt  0
t1

If T is conserved, then

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

10
 t2

  dt  0
t1

 (t2 – t1) = 0
(t2 – t1) = extremum.
ie, the time taken by the light ray to travel between two points is extremum.
1.4 CANONICAL TRANSFORMATION:
Transformation is one to change one set of position and momentum coordinates into
another set of position and momentum coordinates.
We assume p j and q j are the old momentum and position coordinates and P j and Q j
are new one related by
P j =P j (p j, q j,t)
Q j = Q j (p j,q j,t) .…(1)

Then if there exists a new Hamiltonian H in the new coordinates such that
H H
Pj  and Q j  ….(2)
Q j Pj

These equations are known as canonical (or) contact transformations.

Q j, P j are canonical coordinates.
By definition we have H = p jq j - L

and H   p j Q j  K
…(3)
Here the position and momenta coordinates are independent.
Canonical transformations are the transformations of phase space. They are
characterized by the property that they leave the form of Hamilton’s equations of motion.
example : Cartesian to polar coordinate is an example of co-ordinate transformation.
The transformation of one set of position coordinates to new set of coordinates is
called point transformation. They are the transformations of configuration space.
1.5 HAMILTONIAN FORMALISM:
Lagrangian equations of motion are invariant in form with respect to the set of any
generalized coordinates. In the new set Q j, Lagrange’s equations will be

d  L  L
 0
dt  Q
  Q
j j

ie, Lagrange’s equations are covariant with respect to point transformations. If we define Pj

as

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

11
 Pj 
L
Q j

Q j , Q j 
The Hamilton’s canonical equation will also be covariant. ie,
H
Q j  Q j , Pj 
Pj

H
P j   Q j , Pj 
Q j
Therefore, this transformation is extended to Hamiltonian formulation. In
Hamiltonian formulation, we admit the existence of one more independent variable called
momentum. Consequently the simultaneous transformation of the independent coordinates
and momenta q j, p j to a new set Q j, P j can be represented in the form
Q j  Q j q, p, t 
….(1)
Pj  Pj q, p, t 

For Q j, P j to be canonical, they should be able to be expressed in Hamiltonian form

of equations of motion.
K K
ie Q j  P j  
Pj Q j
Where K is a function of (Q, P, t) and is a substitute for H of old set in new set of
coordinates. If Q j, P j are to be canonical coordinates, they must satisfy the modified
Hamilton’s principle of the form

 
t2

   Pj Q j  k Q, P, t  dt  0 ....(2)
t1

The old coordinates p j, q j are already canonical.

Therefore

 
t2

   p j q j  Hq, p, t  dt  0 ....(3)
t1

The simultaneous validity of equations (2) and (3) does not mean that the integrands
of the two integrals are equal. We can therefore write

 
   p j q j  H    Pj Q j  K  dt  0
t2

....(4)
t1

Equation (4) will not be affected if we add to or subtract from it a total time derivative of a
function F = F(q,p,t).Now we can write equation (4) as

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

12
    p j q j  k    Pj Q j  K    dt  0
t2
 dF 
t 
1
dt 

is follows that

 p qj j  
 H    Pj Q j  K 
dF
dt
....(5)

1.6 CANONICAL FORMALISM

DERIVATION OF GENERATING FUNCTIONS:
The first term in equation (5) is regarded as a function of q j, p j and t and the
second term as a function of Q j, P j and t. F is in general as a function of (4n +1) variables q j,
p j, Q j, P j and t. The two sets of variables are connected by the 2n transformation equations
and besides t, only 2n are independent. Now F is a function of both old and new set of
coordinates and four forms of F are possible. F1(q, Q, t), F2(q, P, t) F3(p, Q, t) and F4 (p, P, t)
and F is termed as the generating function.
(A) FIRST FORM F1 (q, Q,t)
We can write equation (5) as

 p j q j  H   Pj Q j  K  1 q, Q, t 
dF
....(6)
dt
F1 = F1 (q, Q, t)
dF1 F F F
  1 q j   1 Q j  1
dt j q j j Q j t

Now equation (6) becomes

F F F
 p j q j  H   Pj Q j  K   1 q j   1 Q j  1
q j Q j t

 F   F  F
 1  p j  q j   Pj  1  Q j  K  H  1  0 .... (7)
 q   
Q j  t
 j  
Since q j and Q j are to be treated as independent variables, equation (7) can hold only
 separately vanish.
if q j and Q j

F1
ie pj  q, Q, t  ....(8)
q j

F1
Pj   q, Q, t  ....(9)
Q j

F1
and K = H q, Q, t  ....(10)
t

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

13
Solving equation (8), we get Q j = Q j (q j, p j, t)
Which when substituted in (9) gives P j = P j (q j, p j, t)
(B) SECOND FORM F2(q,P,t)
F2(q j, P j, t) = F1 (q j, Q j, t) + P j Q j
F1(q j, Q j, t) = F2 (q j, P j, t) P j Q j .....(11)
putting equation (13) in equation (6), we get

 p j q j  H   Pj Q j  K 
d
dt

F2 q j , Pj , t    Pj Q j 
F F F
  Pj Q j  K   2 q j   2 Pj  2   Pj Q j   Pj Q j
q j Pj t

F2 F F
   Pj Q j  K    2 q j   2 Pj
t q j Pj

 F   F  F
  2  p j  q j    2  Q j  Pj  H  2  K  0 ....(12)
 q   P  t
 j   j 
Since q j and P j are independent variable, equation (12) can be satisfied only when
F2
pj  q j , Pj , t  .....(13)
q j

F2
Qj  q j , Pj , t  .....(14)
Pj

F2
and K  H 
t
q j , Pj , t  .....(15)

Equation (13) can be solved to give P j = P j (q j, p j, t)

which when substituted in (14) gives Q j = Q j (q j, p j, t)
(C) THIRD FORM F3 (p,Q, t)
F3 can be obtained from F1 by replacing q j by p j
F3 (p j, Q j, t) = F1 (q j, Q j, t) - p j q j
F1 (q j, Q j, t) = F3 (p j, Q j, t) + p j q j ....(16)
putting equation (16) in equation (6), we get

 p j q j  H   Pj Q j  K 
d
dt

F3  p j , Q j , t    p j q j 
F F F
  Pj Q j  K   3 p j   3 Q j  3   p j q j   p j q j
p j Q j t

F3  F   F 
 H   p j Q j  K    3  q j  p j   3  Pj  Q j
t  p   Q 
 j   j 
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
14
  F   F  F
  3  q j  p j    3  Pj  Q j  H  3  K  0 ....(17)
 p   Q  t
 j   j 
Since p j, Q j are independent variables, equation (17) can be satisfied only when
F3
qj  p j , Qj , t  ....(18)
p j

F3
Pj   p j , Q j , t  ....(19)
Q j

F3
k H  p j ,Qj ,t  ....(20)
t
equation (18) gives Q j = Q j (q, p, t)
and equation (19) gives P j = P j (q, p, t)
(D) FOURTH FORM F4(p, P, t)
F4 can be obtained from F3 by replacing Qj, by Pj
F4( p j, P j, t ) = F3 (p j, Q j, t) + P j Q j
= F1 (q j, Q j, t) -p j q j - +P j Q j
F1 (q j, Q j ,t) = F4(p j, P j, t) + p j q j - P j Q j ....(21)
putting equation (23) in equation (6) we get

 p j q j  H   Pj Q j  K 
d
dt

F4  p j , Pj , t    p j q j   Pj Q j 
F F F
  PjQ j  K   4 p j   4 Pj  4   p j q j   p j q j   PjQ j   PjQ j
p j Pj t

 F   F  F
 H   K   4  q j  p j   4  Q j  P j  4
 p   P  t
 j   j  .… (22)

Since p j and P j are independent variables, equation (22) can be satisfied only when
F4
qj    p j , Pj , t 
p j
.…(23)
F4
Qj   p j , Pj , t )
Pj
….(24)
F4
K H   p j , Pj , t 
t ….(25)

Equation (23) gives Q j = Q j (q, p, t)

and equation (24) gives P j = P j (q, p, t)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

15
(i) Condition for a transformation to be canonical:

If the expression  (P j dQ j- p j dq j)

(or)  (p j dq j- P j dQ j)

be an exact differential then the transformation from (q j, p j) set to (Q j, P j) set is

canonical.

Proof:

We know that for a transformation to be canonical, equation


  p j q j  H   Pj Q j  K 
dF
dt

….(1)

must be satisfied.

Suppose generating function F does not include time explicitly then

F
KH H
t

Now equation (1) becomes

F
 p j q j   Pj Q j 
dt

(or)
 p j dq j  Pj dQ j   dF

Where dF is the exact differential of F.

Exercises:

1. Show that the transformation

Q  2q e cos p

P  2q e  sin p

is a canonical transformation.

Solution:

Q  2q e cos p

dQ  2q  e cos p dq  2q  e sin pdp

1/ 2 1/ 2

PdQ  pdq  2q  e  sin p 2q  e cos p dq  2q  e sin p 2q  e sin pdp  pdq
1/ 2 1/ 2 1/ 2 1/ 2

PdQ  pdq  sin p cos p dq  2q sin 2 p dp  pdq

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
16
 PdQ  pdq  (sin p cos p  p)dq  2q sin 2 p dp

1 
 sin 2 p  p dq  2q sin 2 pdp
= 2 
 1   1 
 q sin 2 p  pq  dq   q sin 2 p  pq dp
= q  2  p  2 
F F
dq  dp
= q p
= dF
Which shows that the RHS is an exact differential of the function
1
F q sin 2 p  pq
2
and hence the transformation is canonical.
2. Show that the transformation

q  2 p sin Q p  2 p cos Q

(or)
P
2

1 2
p  q2  Q  tan 1
q
p
is canconical

Solution:
dq  (2 p)1/ 2 cos Q dQ  2 p 
1/ 2
sin Q dP

pdq  (2 p)1/ 2 cos Q(2 p)1/ 2 cos Q dQ  (2 p)1/ 2 cos Q2 p 

1/ 2
sin Q dP
pdq = 2P cos2 Q dQ + sin Q cosQ dP
pdq -PdQ = 2P cos2 Q dQ + sin Q cosQ dP-PdQ
1
= (2P cos2 Q - P) dQ + 2sin 2Q dP

 1   1 
  P sin 2Q dQ   P sin 2Q dP
Q  2  P  2 
F F
= dQ  dP
Q P
1
= dF where F  P sin 2Q
2
3. Show that the transformation
P = q cot p
 sin p 
Q = log  is canonical. show that the generating function is
 q 


F  e Q 1  q 2 e 2Q 1/ 2
 q sin 1 (qe Q )

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

17
Solution:
Q Q
dQ  q  dp
q p

   sin p     sin p 
 log  dq  log  dp
q   q  p   q 

1  sin p  1  cos p 
   2 dq   dp
 sin p   q   sin p   q 
   
 q   q 
dq
  cot p dp
q

 dq 
pdq  PdQ  pdq  q cot p   cot p dp 
 q 
  p  cot p  dq  q cot 2 p dp

= qp  q cot pdq   qp  q cot p dp
q p
F F
 dq  dp
q p
 dF  exact differential
F = qp +q cot p
and hence the transformation is canonical.
 sin p 
Let us put Q  log 
 q 
sin p  q eQ  p  sin 1 qeQ  
cos p = 1  q 2e 2Q  1/ 2

cot p =
1  q e 
2 2Q 1/ 2

qe Q
Now the generating function F  q sin 1 q eQ   eQ 1  q 2e1  1/ 2
1.7 HAMILTONIAN:
 
The quantity  p j q j  L is a constant of motion with the condition that L does not

involve time explicitly. This constant was designated by H.

ie H   p j q j  Lq j , q j 

H as Hamiltonian H = H (q j, p j)
=  p j q j  L(q j , q j )

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

18
If H does not involve time, it is said to be a constant of motion.
(i) Hamilton’s canonical equations of motion:
Hamiltonian is in general as a function of the position coordinates q j, the momenta p j
and the time t.
H = H (q j, p j, t)
H H H
dH   dq j   dp j  dt …(1)
q j p j t

H =  p j q j  L

So that H   q j dp j   p j dq j  dL ….(2)


But Lagrangian L  L q j , q j , t 
L L L
dL  dq j   dq j  dt ....(3)
q j q j t
Substituting equation (3) in equation (2) we get
L L L
dH   q j dp j   p j dq j   dq j   dq j  dt
q j q j t
….(4)
L L
 p j and  p j
q j q j

Now equation (4) becomes

L
dH   q j dp j   p j dq j   p j dq j   p j dq j  dt
t
L
dH   q j dp j   p j dq j  dt
t
….(5)
Comparing coefficients in equation (5) and equation (1), we arrive
H
q j 
 j

H
p j  
q j
….(6)
L H
 
t t
Equations (6) are known as Hamilton’s canonical equations of motion and of a set of 2n first
order equations of motion.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

19
 (ii) Physical Significance of H
Hamiltonian H also possesses the dimensions of energy but in all circumstances
HE
E = H equality has some restrictions. That are
1. The system be conservative one. ie. Potential energy is coordinate dependent and not
velocity dependent.
2. Coordinate transformation equations be independent of time so that  p j q j  2T

Let us write H = H (p1, p2,……p j, q1, q2,……..q j, t)

dH H H H
 q j   p j 
dt q j p j t

From Hamilton′s equations of motion

H H
  p j  q j
q j p j

dH H
Therefore    p j q j   q j p j 
dt t
H
= ….(1)
t
L
=
t
H L
  .…(2)
t t
L
If L is not an explicit function of time, 0
t
H dH
 0 ie 0
t dt
H = constant .…(3)
Thus if L is not an explicit function of time, H is constant of motion.
For conservative systems, the Potential energy does not depend upon generalized
velocity, ie
V
0 .…(4)
q j

We know H   p jq j  L

L
 q j  L
q j

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

20
   
  q j  T  V  - L Since L = T – V
 q j 

 T V 
  q j    L
 q j q j 

 T  V
  q j    L since 0
 q j  q j

 1 2
H   q j  m q j   L
q j 2 

  m q j  L
2

= 2T – L = 2T – (T –V)
= T + V = K.E + P.E = Total energy.
 H represents the total energy of the system for conservative system.
1.8 CYCLIC (OR) IGNORABLE COORDINATE:
We know that the Lagrangian L is a function of generalized coordinate q j,
generalized velocity 𝑞 j and time t. If the Lagrangian of a system does not contain a
L
particular coordinate q k, then  0. such a coordinate is referred to as an ignorable or
qk
cyclic coordinate.
(i) Generalised momentum: [conjugate (or) canonical momentum]
Consider a system of mass points acted upon by forces derived from potentials
dependent on position only.
Now Lagrange’s equations of motion are

d  L  L
 0
dt  q j  q j

Suppose q j is cyclic.ie it does not occur in Lagrangian L, then for this coordinate Lagrange’s
equation reduces to

d  L 
 0
dt  q j 

L
 constant
q j
Generalized momentum = constant.

L T V T
  
xi xi xi xi
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
21
 
 1

 ml xi  y i  zi
xi 2
2 2 2

 mi xi  pix
= x component of the linear momentum associated with the ith particle.
Generalising the concept, a momentum associated with the coordinate q j shall be
T L
pj   = generalized momentum.
q j q j

Thus p j  constant

The generalized momentum conjugate to a cyclic coordinate is conserved.

L
If we put p j in Lagrange’s equation we get p j 
q j
1.9 ROUTHIAN PROCEDURE AND EQUATIONS :
In Kepler’s problem

L
1
2

m r 2  r 2  2 
k
r

 does not occur in L and is therefore an ignorable coordinate so that corresponding
momentum
L
p   mr 2  l  a constant

It appears from the expression L that we can solve Kepler problem without
considering  which is ignorable. But this is not so in Lagrangian formation because  which
requires how  varies with t. Thus we consider . However Routhian procedure which
eliminates this consideration. We want to find a function R called Routhian function such
that it does not contain generalized velocities corresponding to ignorable coordinates.
L  Lq1 , q 2 ,.....q n , q 1 , q 2 .....q n , t 
If coordinates q1......q k are ignorable then

L  Lq k1......q n , q 1.....q n , t 

n
L n
L L
L  
j  k 1 q j
q j 
j 1 q
j
q j  t
t

 k
L  n
L n
L L
(or) L   q   q j   q j  t . … (1)
 j 1 q
j j  j  k 1 q j j  k 1 q
j t

Routhian function R in which velocities q 1......q k corresponding to ignorable coordinates

q1......q k are eliminated, can be written as

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
22
 R  Rqk 1....qn ,...qk 1....qn , t 
n
R n
R R
so that R  
J  k 1 q j
q j  
J  k 1 q
j
q j 
t
t …(2)

We can also define the Routhian function as

k
R  L   q jp j
J 1

k k
R  L   q j p j   q j p j
J 1 J 1

k k
 L   p jq j   q j p j
J 1 J 1

 k
L  k
 L   q j    q j p j
 J 1 q
j  J 1
n
L n
L L k
 
J  k 1 q j
q j  
J  k 1 q
j
q j  t   q j p j
t J 1
….(3)

Comparing equations (2) and (3), we get

L R

q j q j
….(4)
L R
 J = k+1 …n
q j q j

putting equation (4) in Lagrange’s equations

n  d  L  L 
  
J 1  dt q

 q 
0
  j  j 
we get
n  d  R  R 
    0
 j  q j 
J  K 1  dt  q
….(5)
 
in which Routhian function has replaced Lagranian function. These are only (n-k) second
order equations in the non-ignorable variables. Thus we can eliminate the ignorable
coordinates through Routhian procedure.
(i) Kepler’s problem:
we know that p  m r 2 

so that R = L  p 

 L  m r 2  2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

23
 1
2
 k

 m r 2  r 2  2   mr 2  2
r

=
1
2
 
m r 2  r 2  2 
k
r
R
 mr
r
R k
 mr  2  2
r r
Now we get the equation of motion as
d
mr   mr2  k2  0 [from equation (5)]
dt r
k
mr  mr2  2  0
r
k
mr  mr 2 
r2
mv 2 
  f (r )
r
Which is the equation of motion of a particle under central force.
1.10 DERIVATION OF HAMILTON’S CANONICAL EQUATIONS FROM
VARIATIONAL PRINCIPLE:
Hamilton’s principle is stated as
t2

I    Ldt  0
t1

H   p j q j  L

L   p j q j  H

 
t2

and hence I     p j q j  H q j , p j ,t  dt  0 …(1)

t1

Eqn. (1) is termed as modified Hamilton’s principle.


The  variation can be expressed as   d

I
I  d



 p q 
t2

 d  H q j , p j , t  dt  0

j j
t1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

24
 
end point’s times are same for every path. Limits are independent of α and hence can be

taken inside the integral.
 p j
t2
q H q j H p j H t 
I  d    q j  p j j     dt  0 .…(2)
t1    q j  p j  t  

t
But  0 since time of travel along every path is same. Also

t2
q j t2
d  q j 
 pj
t1

dt   p j
t1
  dt
dt   

q j t2 t2
q j
 pj |   p j dt
 t1 t1

t2
q j
   p j dt
t1


q j
vanishes at limits t1 and t2,

Now equation (2) becomes
t2
 p j q H q j H p j 
I  d    q j  p j j    dt  0
t1 
   q j  p j  
q j
Putting d  q j

p j
d  p j

 t2
H H 
We get I    p j q j  p jq j  q j  p j  dt  0
t1  q j p j 


t2
  H   H  
    p j q j    q j  p j    dt  0
t1 
  p j   q j  

Since p j and q j are independent variables, the above equation can be satisfied only if

H H
qj   0  q j 
p j p j

H H
p j   0  p j  
q j q j
which are the desired canonical equations of motion.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

25
1.11 APPLICATIONS OF HAMILTON’S EQUATIONS OF MOTION:
1. Simple pendulum :
1 22
The kinetic energy of the bob T  ml 
2
and the potential energy r = mg l(1cos)
Lagrangian L = TV
1 2 2
= m l   mgl(1cos) …. (1)
2
L
p = = ml2  2 …. (2)

Hamiltonian H =  p j q j  L

= p   L

1 
= ml 2  2   ml 2  2  mgl (1  cos )
2 
1 2 2
= m l   mgl (1  cos  ) ….(3)
2
=T+V
Then the system is conservative.
Putting equation (2) into equation (3), we get
2
1  p 
H  ml 2  2   mgl (1  cos  )
2  ml 
giving
H p
 2
p ml
….(4)
H
 mgl sin 

Thus Hamilton’s equations of motion for this system will be
H p
   2 _....(5)
 ml
 H
p   mgl sin  ….(6)

From equation (5), we have
p  ml 2
and hence equation (6) becomes

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

26
 ml 2   mgl sin 

l  g sin 

l  g sin   0

l   g   0

  g   0
l
which represents the equation of motion of a simple pendulum with period 2 l / g

2. Compound pendulum :
1 2
T I
2
V = mglcos
 2
Now the Lagrangian L = I  mgl cos 
2
L
Then p 

= I  …. (1)

Hamiltonian H = p j q j  L

= p  L

1
 I  2  I 2  mgl cos 
2
1 2
= I   mgl cos 
2
2
1 p 
= I     mgl cos 
2  I 
2
p
=   mgl cos  ….(2)
2I
=T+V
Then the system is conservative.
The Hamilton’s equations of motion for  and p are
H
 
P
….(3)
 H
p  

From equation (2) we find

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

27
 H p 

p  I

H
 mgl sin 

p
Now equation (3) becomes    ….(4)
I
and p    mgl sin  ….(5)

From equation (4) p   I

Now equation (5) becomes
I   mgl sin 
mgl
  sin   
I
   2   

mgl
which is the equation of motion of compound pendulum with  
I
3. Linear Harmonic Oscillator :
The system is conservative and constraint is independent of time. Hamiltonian will
represent the total energy of the system. The Lagrangian
L  T V
1 1
 mx 2  kx2
2 2
L T
  p  mx
x x
p
x 
m
giving Hamiltonian
H  T V
2
1  p 1
 m   kx2
2 m 2
p2 1
   kx2
2m 2

Equations of motion are

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

28
  H
p   kx
x
H p
x  
p m
p  kx
x  
m m
mx  kx  0
This relation shows that motion is simple harmonic and is the desired equation.
4. Particle in a central field of force :
The system is conservative and hence the Hamiltonian represents the total energy.
1

T  m r 2  r 2  2
2

V  V (r )
Lagrangian L  T  V
1

 m r 2  r 2 2  V (r )
2

L p
pr   mr  r  r
r m
L p
p   mr 2    θ 2
 mr
Hamiltonian H= T + V
1

 m r 2  r 22  V (r )
2

2 2
p p
 r   2  V (r )
2m 2mr
H V
2
p
p r    3 
r mr r
H
p    0
 …. (1)

and
H p r
r  
p r m
H p
   2
p  mr …. (2)

Which are the desired equations of motion.

From equation (2) we can write

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

29
 1 V
2
p r p
r   2 3 
m m r m r
V
2
p
mr  
mr 3
r …. (3)
We put
V
  F (r )  radial force and
r
[mr 2]2
2
p
3
 3
 mr 2 [ From equation (2)]
mr mr
m[r ]2

r
2
mv
  centrifuga l force
r
From equation (3)
2
mv 
mr   F (r )
r
2
mv
.
gives an equation of motion involving the actual force F(r) and a centrifugal force r
5. Hamiltonian for a charged particle in an electromagnetic field :
 1  
Lagrangian L  T  q   v  A 
 c 

1  vA 
  mv j  q    j j 
2
…. (1)
2  c 

L L
pj  
q j v j

q
 mv j  Aj
c
Hamiltonian H   q jp j  L

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

30
  q 
  v j  mv j  A j   L
 c 
q
  mv j   v j A j  L
2

c
q  
 
 mv 2  v  A  L
c

 
q     1   
 2T  v  A  T  q   v  A 
c   c 
T  q
1
 mv 2  q
2
such that H can be interpreted as the sum of kinetic and electrostatic potential energies of the
particle.

6. Particle moving hear the surface of earth :

Let z axis be along upward vertical direction, then kinetic energy is

T
1
2

m x 2  y 2  z 2 
The applied force on the body is its weight acting in negative z direction.ie
V
F  Fz  mg  
z
V  mgz
Lagrangian
L  T V


1
2
 
m x 2  y 2  z 2  mgz

L T
  p x  mx
x x
px
giving x  .
m
py pz
Similarly y  and z 
m m
Hamiltonian for such a system is conserved. ie
H  T V


1
2
 
m x 2  y 2  z 2  mg z

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

31
 
1
2m
 
p x  p y  p z  mgz
2 2 2

giving equations of motion

H
p x   0
x
H
p y   0
y
H
p z    mg …. (1)
z
H p x
and x  
p x m

H p y
y  
p y m

H p z
z   ….(2)
p z m
From equation (2) we get
p x
x  0
m
p y
y  0
m
p z
z   g
m
which shows that the acceleration along z direction is the acceleration due to gravity and is
true.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

32
CLASSICAL AND STATISTICAL MECHANICS

UNIT 2

POISSON BRACKET AND THEORY OF SMALL OSCILLATIONS

Poisson bracket, Special cases of Poisson bracket , Poisson theorem, Poisson bracket
and canonical transformation, Jacobi identity and its derivation, Lagrange bracket and its
properties, the relationship between Poisson and Lagrange brackets and its derivation, the
angular momenta and Poisson bracket, Liouville’s theorem and its applications; Theory of
small oscillations:
Formulation of the problem, Eigenvalue equation and the principle axis transformation,
frequencies of free vibration and normal coordinates, free vibrations of a linear triatomic
molecule

2.1 POISSON BRACKET : DEFINITION

Let F be any dynamical variable of a system.

Suppose F is function of conjugate variables qj ,pj and t, then
dF dF X ∂F ∂F ∂F
= (qj , pj , t) = q˙j + p˙j +
dt dt j
∂qj ∂pj ∂t

X ∂F ∂H 
∂F ∂H ∂F
= − +
j
∂qj ∂pj ∂pj ∂qj ∂t
on using Hamiltonts canonical equations of motion.
The first bracketted term is called Poisson Bracket of F with H.
In general if X and Y are two dynamical variables then
X  ∂X ∂Y ∂X ∂Y

[X, Y ]q,p = − (1)
j
∂q j ∂p j ∂pj ∂qj

2.2 SPECIAL CASES OF POISSON BRACKET

(a) [X, Y ] = −[Y, X]

(b) [X, X] = 0

(c)[X, Y + Z] = [X, Y ] + [X, Z] (2)

(d) [X, Y Z] = Y [X, Z] + [X, Y ]Z

32
CLASSICAL AND STATISTICAL MECHANICS

Also

(e) [qi , qj ]q,p = 0 = [pi , pj ]q,p

(f ) [qi , pj ]q,p = δij = 0 if i 6= j (3)

= 1 if i = j

Equation (3) are known as fundamental Poisson brackets.

Take the property
(c)[X, Y + Z] = [X, Y ] + [X, Z]

Proof :

X  ∂X ∂(Y + Z) ∂X ∂(Y + Z) 
[X, Y + Z] = −
j
∂qj ∂pj ∂pj ∂qj

X ∂X ∂Y  X  ∂X ∂Y 
∂X ∂Z ∂X ∂Z
= + − +
j
∂q j ∂p j ∂q j ∂p j j
∂p j ∂q j ∂pj ∂qj
X  ∂X ∂Y ∂X ∂Y
 X
∂X ∂Z ∂X ∂Z

= − + −
j
∂qj ∂pj ∂pj ∂qj j
∂qj ∂pj ∂pj ∂qj
= [X, Y ] + [X, Z]

Similarly
X  ∂qi ∂qj ∂qi ∂qj

[qi , qj ]q,p = −
k
∂qk ∂pk ∂pk ∂qk

∂qj ∂qi
= =0
∂pk ∂pk
and hence
[qi , qj ]q,p = 0 = [pi , pj ]q,p

X  ∂qi ∂pj ∂qi ∂pj


[qi , pj ]q,p = −
k
∂q k ∂pk ∂pk ∂qk
X ∂qi ∂pj
=
k
∂qk ∂pk

since
∂qi ∂pj
and =0
∂pk ∂qk

33
CLASSICAL AND STATISTICAL MECHANICS

Also
∂qi ∂pj
and =0
∂qk ∂pk
sothat
X
[qi , pj ]q,p = δik δjk

= δij

= 0 if i 6= j

= 1 if i = j

Excercises

1.If [φ, ψ] be the Poisson bracket of φ and ψ prove that

   
∂ ∂φ ∂ψ
(a) [φ, ψ] = , ψ + φ,
∂t ∂t ∂t
   
d dφ dψ
(b) [φ, ψ] = , ψ + φ,
dt dt dt

We have
X  ∂φ ∂ψ ∂ψ ∂φ

[φ, ψ] = −
i
∂qi ∂pi ∂qi ∂pi
 
∂ X ∂ ∂φ ∂ψ ∂ψ ∂φ
[φ, ψ] = −
∂t i
∂t ∂q i ∂p i ∂qi ∂pi

X ∂  ∂φ  ∂ψ X ∂φ ∂  ∂ψ 
= +
i
∂q i ∂t ∂p i i
∂qi ∂pi ∂t
X ∂    
∂ψ ∂φ X ∂ψ ∂ ∂φ
− −
i
∂qi ∂t ∂pi i
∂qi ∂pi ∂t
X ∂    
∂φ ∂ψ ∂ψ ∂ ∂φ
= −
i
∂qi ∂t ∂pi ∂qi ∂pi ∂t
X ∂φ ∂  ∂ψ 

∂
 
∂ψ ∂φ

+ −
i
∂qi ∂pi ∂t ∂qi ∂t ∂pi

   
dφ dψ
= , ψ + φ,
dt dt

34
CLASSICAL AND STATISTICAL MECHANICS

2.If {pl , qi }, {ql , qi } are the Lagrange’s brackets and [pl , pj ], [ql , pj ] are the Poisson brackets,
then prove that

n
X n
X
{pl , qi }[pl pj ] + {ql , qi }[ql , pj ] = 0
i=1 i=1

We have
{pl , qi } = −{qi , pl }

= −δil

and
{ql , qi } = 0

[pl , pj ] = 0

[ql , pj ] = δlj

Substituting these values we get

n
X
{pl , qi }[pl , pj ] + {ql , qi }[ql , pj ] = −δil × 0 + 0 × δlj
i=1

=0

3.If {ql , qi }, {pl , qi } are the Lagrange’s brackets and [ql , qj ], [pl , qj ] are the Poisson brackets,
then prove that

n
X n
X
{ql , qi }[ql , qj ] + {pl , qi }[pl , qj ] = δij
i=1 i=1

We know
{ql , qi } = 0

[ql , qj ] = 0

Also
{pl , qi } = −{qi , pl }

= −δil

[pl , qj ] = −[qj , pl ]

= −δjl

35
CLASSICAL AND STATISTICAL MECHANICS

n
X n
X
{ql , qi }[ql , qj ] + {pl , qi }[pl , qj ] = 0 × 0 + [−δil × −δjl ]
i=1 i=1

= δij

Hence proved.

2.3 POISSON’S THEOREM

For a dynamical variable F (q, p, t)

dF ∂F
= [F, H] +
dt ∂t

If F is a constant of motion so that dF/dt = 0, then by Poisson’s theorem

∂F
[F, H] + =0
∂t

Furthermore if F does not contain time explicitly, that is ∂F /∂t = 0 then

[F, H] = 0

This is the required condition for F to be a constant of motion.

2.4 POISSON BRACKET AND CANONICAL TRANSFORMATION :

Poisson brackets are invariant under a canonical transformation.That is

[X, Y ]q,p = [X, Y ]Q,P

Proof :
X  ∂X ∂Y ∂X ∂Y

[X, Y ]Q,P = −
i
∂Qi ∂Pi ∂Pi ∂Qi
X  ∂X  ∂Y ∂qj ∂Y ∂pj
 
∂X ∂Y ∂qj ∂Y ∂pj

= + − +
i,j
∂Q i ∂q j ∂P i ∂p j ∂P i ∂P i ∂q j ∂Q i ∂pj ∂Qi
X ∂Y X ∂X ∂qj   
∂X ∂qj X ∂Y X ∂X ∂pj ∂X ∂pj
= − + −
j
∂qj i ∂Qi ∂Pi ∂Pi ∂Qi j
∂pj i ∂Qi ∂Pi ∂Pi ∂Qi
X  ∂Y ∂Y

= [X, qj ]Q,P + [X, pj ]Q,P
j
∂q j ∂p j

36
CLASSICAL AND STATISTICAL MECHANICS

Further

[X, qj ]Q,P = −[qj , X]Q,P

X  ∂qj ∂X ∂qj ∂X

= − −
m
∂Qm ∂Pm ∂Pm ∂Qm
X  ∂qj  ∂X ∂qk ∂X ∂pk
 
∂qj ∂X ∂qk ∂X ∂pk

= − + − +
m,k
∂Q m ∂q k ∂Pm ∂p k ∂Pm ∂P m ∂q k ∂Qm ∂pk ∂Qm
X ∂X X ∂qj ∂qk   
∂qj ∂qk X ∂X X ∂qj ∂pk ∂qj ∂pk
= − − + −
k
∂qk m ∂Qm ∂Pm ∂Pm ∂Qm m
∂pk i ∂Qm ∂Pm ∂Pm ∂Qm
X  ∂X ∂X

= − [qj , qk ]Q,P + [qj , pk ]Q,P
k
∂qk ∂pk
X ∂X ∂X
= − δjk = (4)
k
∂pk ∂pj

Similarly
X  ∂Y ∂X ∂Y ∂X

[X, Y ]Q,P = − +
j
∂q j ∂p j ∂pj ∂qj
= [X, Y ]q,p

Hence Poisson brackets are invariant under canonical transformation.

A canonical transformation can be generated from functions F1 (qj , Qj , t),
F2 (qj , Pj , t), F3 (pj , Qj , t) and F4 (pj , Pj , t).
In the case of generating function F1 , we have obtained
∂F1
pj =
∂qj
and
∂F1
Pj = −
∂Qj
Using the above two relations we get
∂pj ∂ 2 F1 ∂Pi
= =− (5)
∂Qi ∂Qi ∂qj ∂qj

Similarly in the case of generating function F2 , we have obtained

∂F2
pj =
∂qj
and
∂F2
Qj =
∂Pj

37
CLASSICAL AND STATISTICAL MECHANICS

Using the above two relations we get

∂pj ∂ 2 F2 ∂Qi
= = (6)
∂Pi ∂Pi ∂qj ∂qj

Similarly in the case of F3 and F4 we have seen that

∂qj ∂ 2 F3 ∂Pi
=− = (7)
∂Qi ∂Qi ∂pj ∂pj

and
∂qj ∂ 2 F4 ∂Qi
= =− (8)
∂Pi ∂Pi ∂pj ∂pj

X  ∂Qi ∂Pj ∂Qi ∂Pj


[Qi , Pj ]q,p = −
k
∂qk ∂pk ∂pk ∂qk

Using eqns(5) and (7) we get

X  ∂Qi ∂qk ∂Qi ∂pk

[Qi , Pj ]q,p = +
k
∂q k ∂Qj ∂pk ∂Qj
∂Qi
= = δij = [Qi , Pj ]Q,P
∂Qj

and similarly

[Qi , Qj ]q,p = 0 = [Qi , Qj ]Q,P

[Pi , Pj ]q,p = 0 = [Pi , Pj ]Q,P (9)

Thus we have proved the affirmation for the fundamental brackets.

Excercises

1.Using Poisson bracket show that the transformation defined by

√
q = 2P sinQ
√
p = 2P cosQ is canonical.
q,p can be rewritten as

q
tanQ =
p
1
P = (q 2 + p2 )
2

38
CLASSICAL AND STATISTICAL MECHANICS

If the transfomation is canonical, it must satisfy the conditions

[Q, Q] = [P, P ] = 0
and
[Q, P ] = 1 (1)
Already we know
[Q, Q] = [P, P ] = 0
Hence we can show that
[Q, P ] = 1 From eqn.(1)

∂Q 1 ∂Q −q
sec2 Q = sec2 Q = 2
∂q p ∂p p
∂P ∂P
= q =p
∂q ∂p
Then

 
∂Q ∂P ∂Q ∂P
[Q, P ] = −
∂q ∂p ∂p ∂q
    
1 2 q 2
= cos Q p + cos Q q
p p2
q2
 
2
= cos Q 1 +
fflp2
cos2 Q 1 + tan2 Q


=

= cos2 Q × sec2 Q = 1

Hence the transformation is canonical.

2.Using Poisson bracket show that the transformation defined by

Q = e−q (1 − p2 e2q )1/2

e−q (1 − p2 e2q )1/2
P = tan−1
p
is canonical.
Q and P can be rewritten as

Q = e−q (1 − p2 e2q )1/2

= (e−2q − p2 )1/2

39
CLASSICAL AND STATISTICAL MECHANICS

and
Q
P = tan−1
p
Now
Q
tanP =
p
Q2
1 + tan2 P = 1 +
p2
Q2 + p2
1 + tan2 P =
p2
Q2 + p2
sec2 P =
p2
p
Q2 + p2
sec P =
p
p
cos P = p
Q2 + p2
p
=p
e−2q − p2 + p2
p
=√
e−2q
p
= −q
e
= p eq

P = cos−1 (p eq )

So now the transformation is

Q = (e−2q − p2 )1/2

P = cos−1 (p eq )

Then

∂Q 1 (−e−2q )(−2) −e−2q

= =
∂q 2 (e−2q − p2 )1/2 (e−2q − p2 )1/2
∂Q 1 (−2p) −p
= −2q
= −2q
∂p 2 (e − p2 )1/2 (e − p2 )1/2
∂P −peq −p −p
= 2 2q 1/2
= −q = −2q
∂q (1 − p e ) e (1 − p2 e2q )1/2 (e − p2 )1/2
∂P −eq −1 −1
= 2 2q 1/2
= −q = −2q
∂p (1 − p e ) e (1 − p2 e2q )1/2 (e − p2 )1/2

40
CLASSICAL AND STATISTICAL MECHANICS

Therefore
 
∂Q ∂P ∂Q ∂P
[Q, P ] = −
∂q ∂p ∂p ∂q
−e−2q
   
−1
=
(e−2q − p2 )1/2 (e−2q − p2 )1/2
  
−p −p
−
(e−2q − p2 )1/2 (e−2q − p2 )1/2
e−2q p2
= −2q −
(e − p2 ) (e−2q − p2 )
(e−2q − p2 )
= −2q =1
(e − p2 )
Furthermore [Q.Q] = 0 and [P, P ] = 0
Hence the transformation is canonical.

2.5 EQUATIONS OF MOTION IN POISSON’S BRACKET FORM :

The total time derivative of a dynamical variable F (qj , pj , t) can be expressed as

∂F
Ḟ = [F, H] +
∂t
If F does not involve time t explicitly then

Ḟ = [F, H] (1)

If the Poisson bracket of F with H vanishes then F =constant of motion. This requirement
does not however require that H should be a constant of motion. Suppose such dynamical
variables are qj and pj , then
q˙j = [qj , H]

and
p˙j = [pj , H] (2)

The above equations are identical with Hamilton’s canonical equations of motion
X  ∂qj ∂H ∂qj ∂H 
[qj , H] = −
i
∂H ∂p i ∂pi ∂qi

Since
∂qj
=0
∂pi

41
CLASSICAL AND STATISTICAL MECHANICS

we get
∂H
[qj , H] = δij
∂pi
∂H
= (f or i = j)
∂pj
Therefore
∂H
q˙j = = [qj , H]
∂pj
Similarly
∂H
pj = −
= [pj , H]
∂qj
Equations (2) can thus be known as equations of motion inpoisson bracket form.
If Poisson bracket [pj , H] vanishes, then

p˙j = 0

pj = constant

That is the linear momentum is conserved and hence the corresponding co-ordinate is cyclic.
Thus all functions whose Poisson bracket with Hamiltonian vanish will be constants of
motion and conversely Poisson brackets of all constants of motion with H must vanish.

2.6 JACOBI’S IDENTITY AND ITS DERIVATION :

[X, [Y, Z]] + [Y, [Z, X]] + [Z, [X, Y ]] = 0

We have
∂[X, Y ] ∂Z ∂Z ∂[X, Y ]
[X, [Y, Z]] = −
∂qj ∂pj ∂qj ∂pj
   
∂ ∂X ∂Y ∂Y ∂X ∂Z ∂Z ∂ ∂X ∂Y ∂Y ∂X
= − − −
∂qj ∂qj ∂pj ∂qj ∂pj ∂pj ∂qj ∂pj ∂qj ∂pj ∂qj ∂pj
 2 2 2
∂Y ∂ 2 X ∂Z

∂ X ∂Y ∂X ∂ Y ∂ Y ∂X
= + − −
∂qj 2 ∂pj ∂qj ∂qj ∂pj ∂qj 2 ∂pj ∂qj ∂qj 2 ∂pj
 2
∂ 2 Y ∂X ∂ 2 X ∂Y

∂Z ∂ X ∂Y ∂Y ∂X
− + − −
∂qj ∂pj ∂qj ∂pj ∂p2j ∂qj ∂pj ∂qj ∂pj ∂p2j ∂qj
∂ 2 X ∂Y ∂Z ∂ 2 X ∂X ∂Z ∂ 2 Y ∂X ∂Z ∂ 2 Y ∂X ∂Z
 
= + − +
∂qj 2 ∂pj ∂pj ∂p2j ∂qj ∂qj ∂qj 2 ∂pj ∂pj ∂pj 2 ∂qj ∂qj
∂ 2X ∂ 2Y
   
∂Y ∂Z ∂Y ∂Z ∂X ∂Z ∂X ∂Z
− + + +
∂pj ∂qj ∂pj ∂qj ∂qj ∂qj ∂pj ∂qj ∂qj ∂pj ∂pj ffl∂qj

42
CLASSICAL AND STATISTICAL MECHANICS

Similarly
∂ 2 Z ∂X ∂Y ∂ 2 Z ∂X
 2
∂ 2 X ∂Z ∂Y

∂Y ∂ X ∂Z ∂Y
[[Z, X], Y ] = + − +
∂qj 2 ∂pj ∂pj ∂p2j ∂qj ∂qj ∂qj 2 ∂pj ∂pj ∂pj 2 ∂qj ∂qj
∂ 2Z ∂ 2X
   
∂X ∂Y ∂X ∂Y ∂Z ∂Y ∂Z ∂Y
− + + +
∂pj ∂qj ∂pj ∂qj ∂qj ∂pj ∂pj ∂qj ∂qj ∂pj ∂pj ffl∂qj

∂ 2 Y ∂Z ∂X ∂ 2 Y
 2
∂ 2 Z ∂Y ∂X

∂Z ∂X ∂ Z ∂Y ∂X
[[Y, Z], X] = + − +
∂qj 2 ∂pj ∂pj ∂p2j ∂qj ∂qj ∂qj 2 ∂pj ∂pj ∂pj 2 ∂qj ∂qj
∂ 2Y ∂ 2Z
   
∂Z ∂X ∂Z ∂X ∂Y ∂X ∂Y ∂X
− + + +
∂pj ∂qj ∂pj ∂qj ∂qj ∂pj ∂pj ∂qj ∂qj ∂pj ∂pj ffl∂qj
Adding all we get
[X, [Y, Z]] + [Y, [Z, X]] + [Z, [X, Y ]] = 0

2.7 LAGRANGE’S BRACKETS AND ITS PROPERTIES:

Lagrange’s bracket u, v with respect to (qj , pj ) is defined as

X  ∂qj ∂pj ∂pj ∂qj

{u, v}q,p = − (1)
j
∂u ∂v ∂u ∂v

(a) Lagrange bracket is invariant under canonical transformation :

Poincare’s theorem states that the integral

Z Z X
J1 = dqj dpj (2)
S j

taken over an arbitrary two dimensional surface S of the 2n dimensional (q,p)phase space is
invariant under canonical transformation.
Position of a point on any two dimensional surface is expressed as

qj = qj (u, v)

pj = pj (u, v)

(3)
Transforming the integral (2) in terms of (u,v), we write
∂(qj , pj )
dqj dpj = dudv (4)
∂(u, v)

43
CLASSICAL AND STATISTICAL MECHANICS

with
 
 ∂qj ∂pj 
 
 
 ∂u ∂u 
 
∂(qj , pj )  
=  (5)
∂(u, v) 


 
 ∂qj ∂pj 
 
 ∂v ∂v 
as the Jacobian.
Further
Z Z X Z Z X
dqj dpj = dQj dPj (6)
S j S j

where (Qj , Pj ) is another set of canonical co-ordinates to which the set(qj , pj )has been trans-
formed.
Now relation (6) becomes
Z Z X Z Z X
∂(qj , pj ) ∂(Qj , Pj )
dudv = dudv
S j ∂(u, v) S j ∂(u, v)

S is arbitrary and area dudv is arbitrary. Therefore expressions on boyh the sides will be
equal only when
X ∂(qj , pj ) X ∂(Qj , Pj )
=
j
∂(u, v) j
∂(u, v)

   
 ∂qj ∂pj   ∂Qj ∂Pj 
X  ∂u ∂u 
  ∂u ∂u 
X 
=
  
  
   
j  ∂q ∂pj  j  ∂Q ∂Pj 
 ∂vj ∂v
  ∂vj ∂v

or
X  ∂qj ∂pj ∂pj ∂qj
 X  ∂Qj ∂Pj ∂Pj ∂Qj

− = −
j
∂u ∂v ∂u ∂v j
∂u ∂v ∂u ∂v

{u, v}q,p = {u, v}Q,P (7)

Thus Lagrange’s bracket is invariant under canonical transformation.

44
CLASSICAL AND STATISTICAL MECHANICS

(b) Lagrange brackets do not obey the commutative law :

X  ∂qj ∂pj
∂pj ∂qj

{u, v} = −
j
∂u ∂v ∂u ∂v
X  ∂pj ∂qj ∂qj ∂pj

=− −
j
∂u ∂v ∂u ∂v
X  ∂qj ∂pj ∂pj ∂qj

=− −
j
∂v ∂u ∂v ∂u

= {v, u} (8)

(c) Proof : {qi , qj } = 0 {pi , pj } = 0 {qi , qj } = δij :

{qi , qj } = 0

{pi , pj } = 0

{qi , qj } = δij
X ∂qk ∂pk ∂qk ∂pk

{qi , qj } = − (9)
k
∂qi ∂qj ∂qj ∂qi
q’s and p’s are independent and hence

∂pk ∂pk
= 0 and =0
∂qj ∂qi

Then
{qi , qj } = 0

Similarly we can prove

{pi , pj } = 0 (10)

Now
X  ∂qk ∂pk ∂qk ∂qk

{qi , pj } = −
k
∂qi ∂pj ∂pj ∂qi
q’s and p’s are independent and hence

∂qk
=0
∂pj

45
CLASSICAL AND STATISTICAL MECHANICS

Then
X ∂qk ∂pk
{qi , pj } = (11)
k
∂qi ∂pj
We have
∂qk ∂pk
= δki and = δkj
∂qi ∂pj
Hence
X
{qi , pj } = δki δkj
k

= δij (12)

2.8 RELATIONSHIP BETWEEN LAGRANGE AND POISSON BRACKET

We can show that

2n
X
{ul , ui } [ul , uj ] = δij
i=1

Here{ul , ui } is Lagrange bracket and [ul , uj ] is Poisson bracket. Now

2n 2n
( n  )
X X X ∂qk ∂pk ∂pk ∂qk 
{ul , ui } [ul , uj ] = −
l=1 l=1 k=1
∂ul ∂ui ∂ul ∂ui
( n  )
X ∂ul ∂uj ∂ul ∂uj
−
m=1
∂qm ∂pm ∂pm ∂qm

(1)

The first four terms on R.H.S on multiplication is

X ∂pk ∂uj X 2n
∂qk ∂ul X ∂pk ∂uj ∂qk
· = ·
k,m=1n
∂u i ∂pm
l=1
∂u l ∂qm
k,m
∂u i ∂pm ∂qm

X ∂pk ∂uj
= · δkm
k,m
∂ui ∂pm
But
∂pm
δkm =
∂pk
The first four terms on R.H.S on multiplication is

X ∂pk ∂uj X 2n
∂qk ∂ul X ∂pk ∂uj ∂pm
· = ·
k,m=1n
∂ui ∂pm l=1 ∂ul ∂qm k,m
∂ui ∂pm ∂pk

46
CLASSICAL AND STATISTICAL MECHANICS

X ∂pk ∂uj
=
k,m
∂ui ∂pk

(2)
The last four terms will be
X ∂qk ∂uj X 2n
∂pk ∂ul X ∂qk ∂uj ∂pk
· = ·
k,m=1n
∂u i ∂qm
l=1
∂u l ∂pm
k,m
∂u i ∂qm ∂pm

X ∂qk ∂uj
= · δkm
k,m
∂ui ∂qm
X ∂qk ∂uj ∂qm
= ·
k,m
∂ui ∂qm ∂qk
X ∂qk ∂uj
=
k,m
∂ui ∂qk

(3)
The second term is
X ∂pk ∂uj X 2n
∂qk ∂ul
− · =0
k,m=1n
∂u i ∂qm
l=1
∂u l ∂pm

since
2n
X ∂qk ∂ul ∂qk
= =0
l=1
∂ul ∂pm ∂pm
Similarly the third term will be zero.
Hence R.H.S of equation (1) is
X ∂pk ∂uj X ∂qk ∂uj X  ∂uj ∂pk ∂uj ∂qk

+ = +
k
∂ui ∂pk k
∂ui ∂qk k
∂pk ∂ui ∂qk ∂ui

∂uj
= (qk , pk )
∂ui
∂uj
= = δij
∂ui
(4)
Therefore
2n
X
{ul , ui } [ul , uj ] = δij
i=1

Which gives the relation between Lagrange and Poisson brackets.

47
CLASSICAL AND STATISTICAL MECHANICS

2.9 THE ANGULAR MOMENTUM AND POISSON BRACKETS :

The angular momentum l can be expressed in terms of linear momentum p and radius
vector r as
l=r×p

= (ix + jy + kz) × (ipx + jpy + kpz )

= i(ypz − zpy ) + j(zpx − xpz ) + k(xpy − ypx ) (1)

giving
lx = ypz − zpy ; ly = zpx − xpz ; lz = xpy − ypx

∂lx ∂ly ∂lz

=0 =z = −y
∂px ∂px ∂px
∂lx ∂ly ∂lz
= −z =0 =x
∂py ∂py ∂py
∂lx ∂ly ∂lz
=y = −x =0
∂pz ∂pz ∂pz
(2)
We know that
X  ∂F ∂pj ∂F ∂pj

[F, pj ] = −
k
∂qk ∂pk ∂pk ∂qk
∂F
= (f or j = k)
∂qj
so that
∂lx ∂ly ∂lz
[lx , px ] = =0 [ly , px ] = = −pz [lz , px ] = = py
∂x ∂x ∂x
∂lx ∂ly ∂lz
[lx , py ] = = pz [ly , py ] = =0 [lz , py ] = = −px
∂y ∂y ∂y
∂lx ∂ly ∂lz
[lx , pz ] = = −py [ly , pz ] = = px [lz , pz ] = =0
∂z ∂z ∂z
(3)
Equations (2) and (3) give Poisson brackets of angular and linear momentum components.
We know that
[pj , px ] = 0

Therefore

[px , py ] = [py , pz ] = [pz , px ] = 0 (4)

48
CLASSICAL AND STATISTICAL MECHANICS

Now we find out the Poisson bracket of components of l

X  ∂lx ∂ly ∂lx ∂ly

[lx , ly ] = −
k
∂qk ∂pk ∂pk ∂qk
     
∂lx ∂ly ∂lx ∂ly ∂lx ∂ly ∂lx ∂ly ∂lx ∂ly ∂lx ∂ly
= − + − + −
∂x ∂px ∂px ∂x ∂y ∂py ∂py ∂y ∂z ∂pz ∂pz ∂z
Using equations (2) and (3), we get

[lx , ly ] = 0 − 0 + 0 − 0 + (−py )(−x) − (y)(px )

= (xpx − ypx ) = lz

Similarly we can prove

[ly , lz ] = lx

and
[lz , lx ] = ly

2.10 LIOUVILLE’S THEOREM :

The theorem consists of two parts

(1) The first part states the conservation of density in phase space

dρ
i.e =0
dt

2) The second part states the conservation of extension in phase space

d
i.e (δΓ) = 0
dt

(1) First Part :

Consider any fixed element of volume of phase space located between q1 and q1 + δq1 , ...qf
and qf + δqf , p1 and p1 + δp1 , ...pf and pf + δpf .
If ρ is the density of phase points, the number of phase points in this volume element at any
instant t is
δN = ρ.δΓ = ρ(δq1 ...δqf δp1 ...δpf )

49
CLASSICAL AND STATISTICAL MECHANICS

q´1
[ ( ) ]
q´1 +
q´1
q1
δ q1

p
D C
p1 + δ p1

δ p1
p1
A B

δ q1

q1 q1 + δ q1

q
The number of phase points located in the volume (δq1 ...δqf δp1 ...δpf ) changes as the co-
ordinates and momenta vary. The change in the number of phase points within this volume
of phase space in time dt is
d(δN ) d(ρ.δΓ) dρ
= = dt(δq1 ...δqf δp1 ...δpf )
dt dt dt
This change is due to the number of phase points entering and leaving this volume in time
dt.
Consider two faces of hypervolume normal to the q-axis with coordinates q1 and q1 + δq1 .
Number of phase points entering the first phase in time dt is

ρq˙1 dtδq2 ...δqf δp1 ...δpf (1)

ρ and q˙1 are the density and velocity component at (q1 ...qf ; p1 ...pf )
Number of phase points entering the second phase in time dt is
  
∂ρ ∂ q˙1
ρ+ δq1 q˙1 + δq1 dtδq2 ...δqf δp1 ...δpf (2)
∂q1 ∂q1

Neglecting higher order terms we have

   
∂ q˙1 ∂ρ
ρq˙1 + ρ + q˙1 δq1 dtδq2 ...δqf δp1 ...δpf (3)
∂q1 δq1

Eqn(1)-Eqn(3) gives
 
∂ q˙1 ∂ρ
− ρ + q˙1 dtδq1 ...δqf δp1 ...δpf (4)
∂q1 δq1

50
CLASSICAL AND STATISTICAL MECHANICS

Similarly for p1 co-ordinate

 
∂ p˙1 ∂ρ
− ρ + p˙1 dtδp1 ...δqf δp1 ...δpf (5)
∂p1 δp1

The net increase in the number of phase points in time dt is in this volume of phase space
is obtained by summing the net number of phase points entering the volume through all the
faces labelled by q1 ...qf and p1 ...pf . Hence
f     
d X ∂ q˙j ∂ p˙j ∂ρ ∂ρ
(δN ) = − ρ + + q˙j + p˙j dtδq1 ...δqf δp1 ...δpf (6)
dt j=1
∂qj ∂pj ∂qj ∂pj

Already we have seen that

d ∂ρ
(δN ) = dtδq1 ...δqf δp1 ...δpf
dt ∂t
and hence
∂ρ
dtδq1 ...δqf δp1 ...δpf
∂t
f     
X ∂ q˙j ∂ p˙j ∂ρ ∂ρ
=− ρ + + q˙j + p˙j dtδq1 ...δqf δp1 ...δpf
j=1
∂q j ∂p j ∂q j ∂p j

f     
∂ρ X ∂ q˙j ∂ p˙j ∂ρ ∂ρ
= − ρ + + q˙j + p˙j (7)
∂t j=1
∂q j ∂p j ∂q j ∂p j

The equations of motion in canonical form are

∂H ∂H
q˙j = and p˙j = −
∂pj ∂qj
Now
∂ q˙j ∂ 2H
=
∂qj ∂qj ∂pj
∂ p˙j ∂ 2H
=−
∂pj ∂pj ∂qj
The order of differentian is immaterial and hence

f  
X ∂ q˙j ∂ p˙j
+ =0 (8)
j=1
∂qj ∂pj

51
CLASSICAL AND STATISTICAL MECHANICS

Now Eqn(7) becomes

  f  
∂ρ X ∂ρ ∂ρ
=− q˙j + p˙j (9)
∂t q,p j=1
∂qj ∂pj
  f  
∂ρ X ∂ρ ∂ρ
+ q˙j + p˙j =0 (10)
∂t q,p j=1
∂qj ∂pj

This result is known as Liouville’s theorem.This equation is identical with the equation of
continuity in hydrodynamics.
If ρ is a function of q, p and t and q, p are functions of t, then

dρ ∂ρ ∂ρ dq ∂ρ dp
= + +
dt ∂t ∂q dt ∂p dt

On generalization we get
f f
dρ ∂ρ X ∂ρ dqj X ∂ρ dpj
= + + (11)
dt ∂t j
∂q j dt j
∂pj dt

Comparing eqns (10) and (11) we get

dρ
=0
dt

This form is called the principle of the conservation of density in phase space.

(2) Second Part :

Here we have to prove that

d
(δΓ) = 0
dt
We know
δN = ρ δΓ

Now

d dρ d
(δN ) = δΓ + ρ (δΓ) (12)
dt dt dt

The number of phase points δN in a given region must remain fixed.

d
(δN ) = 0
dt

52
CLASSICAL AND STATISTICAL MECHANICS

dρ d
δΓ + ρ (δΓ) = 0 (13)
dt dt
Already we have proved that
dρ
=0
dt
It follows that
d
ρ (δΓ) = 0 (14)
dt
But ρ 6= 0, we get
d
(δΓ) = 0 (15)
dt
This equation gives the principle of conservation of extension in phase space.

Excercises :

1.If the transformation eqations between two sets of co-ordinates are

P = 2(1 + q 1/2 cos p)q 1/2 sin p

Q = log (1 + q 1/2 cos p)

then show that (i) the transformation is canonical and (ii)the generating function of this
transformation is
F3 = −(eQ − 1)2 tan p

Solution :

(i) For the trasformation to be canonical (pdq − P dQ) must be an exact differential.

q 1/2 (−sin p dp) + cos p( 21 q −1/2 dq)

pdq − P dQ = pdq − 2(1 + q 1/2 cos p)q 1/2 sin p ·
(1 + q 1/2 cos p)
q(−sin p dp) + cos p( 21 dq)
pdq − P dQ = pdq − 2(1 + q 1/2 cos p)q 1/2 sin p ·
q 1/2 (1 + q 1/2 cos p)
= p dq + 2qsin2 p dp − sin p cos p dq
1
= p dq + q(1 − cos 2p)dp − ( sin 2p dq)
2
1
= (p − sin 2p) dq + q(1 − cos 2p) dp
2

53
CLASSICAL AND STATISTICAL MECHANICS

∂ 1 ∂ 1
= [q(p − sin 2p)]dq + [q(p − sin 2p)]dp
∂q 2 ∂p 2
1
= d[q(p − sin 2p)]
2
= an exact dif f erential

(ii) We have

Q = log (1 + q 1/2 cos p)

eQ = (1 + q 1/2 cos p)

eQ − 1 = q 1/2 cos p
eQ − 1
q 1/2 =
cos p
(eQ − 1)2
q= (1)
(cos p)2

Now
P = 2(1 + q 1/2 cos p)q 1/2 sin p
(eQ − 1)
= 2eQ sin p
cos p
= 2eQ (eQ − 1)tan p (2)

We know that
∂F3
q=−
∂p
∂F3
P =−
∂Q
Now
∂F3 (eQ − 1)2
=− = −(eQ − 1)2 sec2 p (3)
∂p cos2 p
∂F3
= −2eQ (eQ − 1)tan p (4)
∂Q
Integrating eqn(3) we get
Z
F3 = − (eQ − 1)2 sec2 p dp + constant

54
CLASSICAL AND STATISTICAL MECHANICS

Taking the constant of integration to be zero we get

F3 = −(eQ − 1)2 tan p (5)

Integrating eqn(4) we get Z

F3 = 2 eQ (eQ − 1)tan p dQ

Again taking the constant of integration to be zero we get

F3 = −(eQ − 1)2 tan p (6)

Since
Equation(5) = Equation(6)

means F3 is the generator of the given transformation.

2.Show that the generating function for the transformation

1
p= , p = P Q2
Q

is
q
F =
Q
Solution :

1
p= , p = P Q2
Q
We find
1
pdq = (2P QdQ + Q2 dP )
Q
= 2P dQ + QdP

Now
pdq − P dQ = 2P dQ + QdP − P dQ

= P dQ + QdP
∂ ∂
= (P Q)dQ + (P Q)dP
∂Q ∂P
∂F ∂F
= dQ + dP
∂Q ∂P

55
CLASSICAL AND STATISTICAL MECHANICS

= dF

= an exact dif f erential

The generating function is

F = PQ

But generating function involves both sets of co-ordinates. As

q
P =
Q2
we have
q
F = Q
Q2
q
=
Q
Hence the result.

THEORY OF SMALL OSCILLATIONS :

2.11 FORMULATION OF THE PROBLEM :EIGEN VALUE EQUATION :

Consider a system with n degrees of freedom and whose total energy is conserved.Suppose
the system is in an equilibrium stste at the point having the co-ordinates q01 , q02 ...q0n . Thus
the potential energy satisfies the condition
 
∂V
=0
∂qj q=q0
Here q = q0 represents qj = q0j for all values of j. let ηj denotes a small displacement in the
corresponding co-ordinate qj from the equilibrim position. Sothat

qj = q0j + ηj

with j=1,2,...n
Now the potential energy becomes

V (q1 , q2 , ...qn ) = V (q01 + η1 , q02 + η2 , ...q0n + ηn )

n  
X ∂V
= V (q01 , q02 , ...q0n ) + ηj
j=1
∂qj q=q0
n n
∂ 2V
 
1 XX
+ ηj ηk + ... (1)
2 j=1 k=1 ∂qj ∂qk q=q0

56
CLASSICAL AND STATISTICAL MECHANICS

The departure from the equilibrium position is small and we truncate the series after the
first non-vanishing term in ηj .
The first term on the R.H.S is constant, the second term is zero and we consider upto the
third term.
The first term gives the shift the reference of potential energy to the equilibrium point.
Thus the equation (1) becomes
n n
∂ 2V
 
1 XX
V (q1 , q2 , ...qn ) = ηj ηk
2 j=1 k=1 ∂qj ∂qk q=q0
n n
1 XX
= Vjk ηj ηk (2)
2 j=1 k=1

Where
∂ 2V
 
Vjk =
∂qj ∂qk q=q0

Here Vjk = Vkj . For the stable equilibrium, the potential energy should be minimum at the
equilibrium position and hence Vjk must be positive. The kinetic energy of the system is
n n n n
1 XX 1 XX
T = cjk q˙j q˙k = cjk η˙j η˙k
2 j=1 k=1 2 j=1 k=1

cjk be the functions of the co-ordinates qj ’s and we can expand in Taylor series form about
the equilibrium position :

cjk (q1 , q2 , ...qn ) = cjk (q01 + η1 , q02 + η2 , ...q0n + ηn )

n  
X ∂cjk
= cjk (q01 , q02 , ...q0n ) + ηj + ...
j=1
∂q j q=q0

T has the terms of second order in η’s and we retain only the zeroth order term. Thus we
have

cjk (q1 , q2 , ...qn ) = cjk (q01 , q02 , ...q0n ) = Tjk

and the kinetic energy is

n n
1 XX
T = Tjk η˙j η˙k (3)
2 j=1 k=1

Here Tjk = Tkj .Using eqns (2) and (3) the Lagrangian of the system is
n n n n
1 XX 1 XX
L = Tjk η˙j η˙k − Vjk ηj ηk
2 j=1 k=1 2 j=1 k=1

57
CLASSICAL AND STATISTICAL MECHANICS

The Lagrangian equations of mtion are

 
d ∂L ∂L
− =0
dt ∂ η˙j ∂ηj

where j=1,2,...n.Therefore we have

n
X
(Tjk η¨k + Vjk ηk ) = 0 (4)
k=1

where j=1,2,...n.Equation (4) can be expressed in the matrix form

       
T T ... T η¨ V V ... V η 0
 11 12 1n
  1   11 12 1n
  1  
       
 T12 T22 ... T2n   η¨2   V12 V22 ... V2n   η2   0 
   +    =  
       
 ... ... ... ...   ...   ... ... ... ...   ...  ...
       
T1n T2n ... Tnn η¨n V1n V2n ... Vnn ηn 0

or
T η̈ + V η = 0

Here T and V represent the square matrices of order n × n.Both of these matrices are
real as well as symmetric.Each of the n equations (4) involves all the n variables and can
be simplified by transforming them into another set of n equations each of which involves
only one variable. This is possible with the help of the normal co-ordinates and normal
frequencies.Let us try the oscillatory solutions of the form

ηk = Ak eiωt (5)

here ω is the frequency of oscillations. The real part Ak cos ωt corresponds to the actual
motion. Using eqn (5) in (4) we get
n
X
(Vjk − ω 2 Tjk )Ak = 0 (6)
k=1

where j=1,2,...n. Thus we have

    
V11 − ω 2 T11 V12 − ω 2 T12 ... V1n − ω 2 T1n A1 0
    
 2 2 2    
 V21 − ω T21 V22 − ω T22 ... V2n − ω T2n   A2   0 
   =  
    
 ... ... ... ...   ...  ...
    
Vn1 − ω 2 Tn1 Vn2 − ω 2 T22 ... Vnn − ω 2 T2n An 0

58
CLASSICAL AND STATISTICAL MECHANICS

This set of equations has anon-trivial solution when

 
2 2 2
 V11 − ω T11 V12 − ω T12 ... V1n − ω T1n 
 
 
 2 2 2 
 V21 − ω T21 V22 − ω T22 ... V2n − ω T2n 
 =0
 
 ... ... ... ... 
 
2 2 2
 
Vn1 − ω Tn1 Vn2 − ω T22 ... Vnn − ω T2n 

(7)
This equation is known as the secular or characteristic equation.
Equation (7) may also be wrtten as

(V A − ω 2 T A) = 0

called eigen value equation. Here A is a column matrix of n components:

 
A
 1
 
 A2 
A=
 

 ... 
 
An

Expansion of this determinant of order n × n gives a polynomial of degree n in ω 2 .This

polynomial has roots ω12 , ω22 , ...ωn2 which are the characteristic or eigen frequencies of the
system. These frequencies are known as the normal frequencies of the system. The eigen
values are real as they correspond to a real symmetric matrix. For each of these frequencies
equation (6) can be solved to get the eigen function Ak
To distinguish between various modes,let us put the suffix l and then the eigen function
is Akl and the frequency eigen value isωl . Then a general solution for the displacement ηk
consists of a linear combination of all the modes ;
n
X
ηk = Ak1 eiω1 t + Ak2 eiω2 t + ... + Akn eiωn t = Akl eiωl t
l=1

Each mode associated with an eigen frequency is known as the principal or natural
mode.Each of the normal frequencies must be real.The eigen functions corresponding to
different eigen values must be orthogonal to each other. Thus we have
n
X
Akl Akl0 = δll0
k=1

59
CLASSICAL AND STATISTICAL MECHANICS

2.12 PRINCIPAL AXES TRANSFORMATION ;

We can choose a certain system of body axes with respect to which the off-diagonal
elements should disappear and only the diagonal elements remain in the expression for I.
Such axes are called principal axes of transformation. The corresponding moments of inertia
are called principal moments of inertia.
if we denote this form of inertia tensor by I0 and I1 , I2 , I3 stand for the principal values,
 
I 0 0
 1 
0
I =  0 I2 0  (1)
 
 
0 0 I3

where we have denoted Ix0 x0 = I1 , Iy0 y0 = I2 , Iz0 z0 = I3 . If ωx , ωy , ωz are the components of

angular velocity and Lx , Ly , Lz are the angular momentum about the principal axes, then
    
L I 0 0 ω
 x  1   x
Ly  =  0 I2 0  ωy 
    
    
Lz 0 0 I3 ωz
 
I1 ωx + 0 + 0
 
= 0 + I2 ωy + 0
 
 
0 + 0 + I3 ωz
or

Lx = I1 ωx
Ly = I2 ωy

Lz = I3 ωx (2)

That is each of angular momentum component along a principal axis is a function of cor-
responding angular velocity component only related to it through the principal moment of
inertia about that direction.
Thus in general if a rigid body is rotating about a principal axis, the angular momentum L
and ω are directed along any of the principal axes and therefore

L=Iω

60
CLASSICAL AND STATISTICAL MECHANICS

where I is the scalar, the moment of inertia about this axis.The angular momentum L and
angular velocity ω are along the principal axes and hence

L = Lxbi + Lybj + Lz k
b = I(ωxbi + ωybj + ωz k)
b

where bi, bj, k

b are unit vectors along X,Y,Z axes respectively.

Thus
Lx = Iωx , Ly = Iωy , Lz = Iωz (3)

using the symmetry property of inertia tensor we get

Lx = Ixx ωx + Ixy ωy + Ixz ωz = Iωx

Ly = Iyx ωx + Iyy ωy + Iyz ωz = Iωy

Lz = Izx ωx + Izy ωy + Izz ωz = Iωz

or   
Ixx − I Ixy Ixz ωx
  
  ωy  = 0 (4)
  
 Iyx Iyy − I Iyz
  
Izx Izy Izz − I ωz

For these equations to have non-trivial solutions, the determinant of the coefficients must
vanish. That is
 
 
Ixx − I Ixy Ixz 
 
 Iyx Iyy − I Iyz  = 0 (5)
 
 
 
 Izx Izy Izz − I 

This is called secular equation of inertia and its solutions the secular values
or eigenvalues. We solve the determinantal equation which will be cubic in I and therefore
will furnish three values for I through I1 , I2 , I3 which are desired principal moments of inertia
:
The direction of any one principal axis is determined by substituting for I = I1 and determine
the ratios for ωx : ωy : ωz as
ωy ωz
= λ1 , = λ2
ωx ωx
Thus

ω = ωxbi + ωybj + ωz k
b

61
CLASSICAL AND STATISTICAL MECHANICS

and
q
2
ω
b = (bi + λ1bj + λ2 k)/
b 1 + λ21 + λ2

Hence we can determine the direction of ω or the direction of principal axis corresponding
to I1 .Similarly we may find the direction of the corresponding pricipal axis if we substitute
I2 or I13 .
Example:
If the symmetry axis of the body is taken as axis of rotation and the origin of body axes
lies on this then the principal axes are the symmetry axis and any two perpendicular axes
normal to the symmetry axis. In the case of a sphere, every axis through the centre is
symmetry axis and hence any three orthogoal axes through the centre are principal axes.

2.13 FREQUENCIES OF FREE VIBRATIONS AND NORMAL CO-

ORDINATES:

The co-ordinates in the solution of equations where only one single frequency is involved
in the solution are known as the normal co-ordinates.Thus the normal co-ordinates are
defined as the generalized co-ordinates wher each of them execute oscillations with a single
frequency.On transformation from the co-ordinates uj into the normal co-ordinates denotted
by ηk the lagrangian as well as equations of motion are changed.
Let the new co-ordinates ηk be related to the original co-ordinates as
X
uj = ajk ηk
k

In the matrix form this relation can be expressed as

u = aη (1)

Here u and η are column matrices of order n × 1 and a is a square matrix of order n × n.We
shall express potential and kinetic energy in terms of η.The potential energy is
1X
V = Vjk uj uk
2 j,k

1X
= uj Vjk uk
2 j,k

62
CLASSICAL AND STATISTICAL MECHANICS

In the matrix form this relation can be expressed as

1
V = uT V u
2

Using eqn(1) we get

1
V = (aη)T V aη
2
1
= η T aT V aη
2
1
= ηT ∧ η
2
Where
AT V A = ∧ = diagonal matrix

= diag(ω1 2 , ω2 2 ...ωn 2 )

Further the above eqn is quardratic in η so that

1X
V = λl ηl 2
2 l

1X 2 2
V = ωl ηl (2)
2 l
The kinetic energy is given by

1 XX
T = Tjk u˙j u˙k
2 j k

1 XX
= u˙j Tjk u˙k
2 j k
1
= u̇T T u̇
2
1
= (aη̇)T T (aη̇)
2
1
= η̇ T aT T aη̇
2
1
= η̇ T η̇
2
Where
aT T a = 1
1X 2
= η̇l (3)
2

63
CLASSICAL AND STATISTICAL MECHANICS

The Lagrangian in new co-ordinates system will be

n n
1 X 2 1X 2 2
L = ffl η̇l − ωl ηl
2 i=1 2 i=1

Giving
n n
∂L X ∂L X
= η̇l and =− ωl 2 η̇l
∂ η̇l i=1
∂η l i=1

Now  
d ∂L ∂L
− =0
dt ∂ η̇l ∂ηl
gives
n
X
(η̈l + ωl 2 η̇l ) = 0
i=1

Therefore the equations of motion in new co-ordinates are

η¨1 + ω1 2 η˙1 = 0

η¨2 + ω2 2 η˙2 = 0

... ... ...

η̈l + ωl 2 η̇l = 0

(4)
Thus each co-ordinate executes only one single frequency oscillation and therefore η1 , η2 etc
are termed as normal co-ordinates.
The solution of equation
η̈l + ωl 2 η̇l = 0

is
ηl = Al cos ωl t + Bl sin ωl t if ωl 2 > 0

ηl = Al t + Bl t if ωl 2 = 0

ηl = Al eωl t + Bl eωl t if ωl 2 < 0

(5)
For ωl 2 > 0, all co-ordinates remain finite and the equilibrium is stable.
But for ωl 2 = 0 and ωl 2 < 0, the co-ordinates become infinite and the equilibrium is unstable.

64
CLASSICAL AND STATISTICAL MECHANICS

Normal modes of vibration :

Since
ω = 2πν

The solutions become

ηl = Al cos 2πν1 t + Bl sin 2πν1 t

η2 = Al cos 2πν2 t + Bl sin 2πν2 t

... ... ... ... ... ...

ηn = Al cos 2πνn t + Bl sin 2πνn t

(6)
Here A’s and B’s are 2n arbitrary constants determined by the initial conditions.
Suppose all constants except A1 and B1 are zero. Then only η1 will vary periodically
with time. This situation corresponds to a normal mode of vibration. That is the system
is vibrating in a normal mode.Therefore there will be n normal modes of vibration and n
normal frequencies ν1 , ν2 , ν3 ...νn corresponding to each normal co-ordinate η1 , η2 , η3 ...ηn Now
eqn(6) will take the form
ηl = Al cos (ωl t + δl )

Her δl is the phase factor.

Now the old co-ordinates are given by
X
uj = ajk Ak cos (ωk t + δk )
k

2.14 FREE VIBRATIONS OF A LINEAR TRIATOMIC MOLECULE :

We consider a linear triatomic molecule of the type Y X2 (e.g CO2 ). Y is a central

atom.There exists an elastic bond between the central atom and the end atoms with force
constant k. Let the mass of each end atom be m and that of cental atom be M.Let the
displacement of atoms from their equilibrium position be q1 , q2 , q3 .
The kinetic energy of the system

1 1
T = m(q˙1 2 + q˙3 2 ) + M q˙2 2
2 2

65
CLASSICAL AND STATISTICAL MECHANICS

Figure 1: LINEAR TRIATOMIC MOLECULE

  
m 0 0 q˙
h i   1
2T = q˙1 q˙2 q˙3  0 M 0  q˙2 
  
  
0 0 m q˙3

giving  
m 0 0
 
T = (Tij ) =  0 M 0 
 
 
0 0 m

The potential energy V of the system is

1 1
V = k(q2 − q1 )2 + k(q3 − q2 )2
2 2
1
V = k(q1 2 + 2q2 2 + q3 2 − 2q1 q2 − 2q2 q3 )
2
or   
k −k 0 q
h i   1
2V = q1 q2 q3 −k 2 k −k q2 
  
  
0 −k k q3

giving the matrix V as  

k −k 0
 
V = (Vij ) = −k 2 k −k
 
 
0 −k k

Then write secular equation

|V − ω 2 T | = 0

66
CLASSICAL AND STATISTICAL MECHANICS

That is  
k − ω 2 m
 
−k 0 
 
2 −k  = 0
 
 −k 2k−ω M
 
k − ω 2 m
 
 0 −k
After expansion we get

(k − ω 2 m)[(2k − ω 2 M )(k − ω 2 m) − k 2 ] + k[−k(k − ω 2 m)] = 0

(k − ω 2 m)[(2k − ω 2 M )(k − ω 2 m) − k 2 ] − k 2 (k − ω 2 m) = 0

(k − ω 2 m)[(2k − ω 2 M )(k − ω 2 m) − 2k 2 ] = 0

giving
ω1 = 0
r
k
ω2 =
m
s  
1 2
ω3 = k +
m M
Now we want to express η1 , η2 and η3 in terms of the generalized co-ordinates q1 , q2 and q3
as
X
qj = Ajk ηk
k
giving
    
q1 A11 A12 A13 η
     1
q2  = A21 A22 A23  η2  (1)
    
    
q3 A31 A32 A33 η3

Our problem is to find the components of eigen vectors A1 , A2 and A3 . For this purpose we
apply the relation
3
X
(Vij − ω 2 Tij )Aij = 0 i = 1, 2, 3
j=1

First case :
ω = ω1 = 0 and calculation of the components A11 , A21 and A31 of A1 . We have
  
k −k 0 A
   11 
−k 2k −k A21  = 0
  
  
0 −k k A31

67
CLASSICAL AND STATISTICAL MECHANICS

giving
kA11 − kA21 = 0

−kA11 + 2kA21 − kA31 = 0

−kA21 − kA31 = 0

It gives
A11 = A21 = A31 = α

Second case :

q
k
ω = ω2 = m
and calculation of the components A12 , A22 and A32 of A2 . We have
  
0 −k 0 A
   12 
kM
−k  A22  = 0
  
−k 2k − m
  
0 −k 0 A32

giving
−kA22 = 0
kM
−kA12 + (2k − )A22 − kA32 = 0
m
−kA21 − kA31 = 0

It gives
A22 = 0 A12 = −A32 = β

Third case :

q
1 2


ω = ω3 = k m
+ M
and calculation of the components A13 , A23 and A33 of A3 . Then
 
2 2 1 2 k 2m
ω = ω3 = k + = (1 + )
m M m M

We have   
− 2mk
M −km 0 A13
  
− kM −k  A23  = 0
  
 −k m
  
0 −k − 2mk
M A33

68
CLASSICAL AND STATISTICAL MECHANICS

giving
2mk
− A13 − kA23 = 0
M
kM
−kA13 − A23 − kA33 = 0
m
2mk
−kA23 − A33 = 0
M
It leads to
2mk
A13 = −kA23
M
2m 2m
A13 = −A23 ⇒ A23 = − A13
M M
Take
A13 = γ

Then
2m
A23 = − γ
M
Similarly
2mk
−kA23 − A33 = 0
M
That is
2mk
kA23 + A33 = 0
M
Which leads to
2mk 2mk 2m
A33 = −kA23 ⇒ A33 = −k(− γ)
M M M
A33 = γ

It gives
2m
A13 = A33 = γ A23 = − γ
M
Thus the matrix A is  
α β γ
 
A = Aij = α 0 − 2mγ
 
M

 
α −β γ

The components can be calculated by applying

AT T A = I

69
CLASSICAL AND STATISTICAL MECHANICS

     
α α α m 0 0 α β γ 1 0 0
     
   2mγ  =  
β 0 −β   0 M 0  α 0 − M  0 1 0
     
γ − 2mγ
M γ 0 0 m α −β γ 0 0 1
    
α α α αm βm γm 1 0 0
    
=
    
β 0 −β  αM 0 −2mγ  0 1 0
    
γ − 2mγ
M γ αm −βm γm 0 0 1
   
α2 (2m + M) 0 0 1 0 0
   
2β 2 m  = 0 1 0
   
 0 0
   
0 0 2γ 2 m(1 + 2m/M ) 0 0 1

It gives
1
α2 (2m + M ) = 1 ⇒ α = √
2m + M
1
2β 2 m = 1 ⇒ β = √
2m
1
2γ 2 m(1 + 2m/M ) = 1 ⇒ γ = p
2m(1 + 2m/M )
Thus the matrix A is
 
√ 1 √1 √ 1
 2m+M 2m 2m(1+2m/M ) 
 1 1

A = Aij = 
 √2m+M 0 √  (2)
2m(1+2m/M ) 
 
√ 1 √1 √ 1
2m+M 2m 2m(1+2m/M )

The normal co-ordinates can be obtained by using equation (2) in equation (1).

70
Unit III - TWO - BODY CENTRAL FORCE PROBLEM AND H - J THEORY
Two body central force problem: Reduction to the equivalent one body problem, the
equation of motion and first integrals, classification of orbits, the virial theorem, the
differential equation for the orbit, integral power law in time in the Kelper’s problem,
scattering in central force field;
H-J Theory: H-J equation and their solutions, use of H-J method for the solution of
harmonic oscillator problem, Hamilton’s principle function, Hamilton’s characteristic
function and their properties, Action angle variable for completely separable systems, the
Kelper’s problem in action angle variables

TWO - BODY CENTRAL FORCE PROBLEM :

3.1 REDUCTION TO THE EQUIVALENT ONE BODY PROBLEM:

Consider a conservative system of two mass
points m1 and m2. Let their instantaneous position

vectors in an inertial frame with origin O are r1 and

r2 respectively.
Hence the vector distance of m2 relative to m1 is
  
r  r2  r1
....(1)
The Lagrangian for the system is
L= T–V
1 2 1  2
= m1 r1  m2 r2  V(r ) ....(2)
2 2

Let us choose the three components of the position vector of the centre of mass R
  
and three components of the relative vector r  r2  r1 to describe the state of the system.
The position vector of the centre of mass is defined by
  
m1r1  m2 r2
R  ....(3)
m1  m2
  
 m1 r1  m2 r  r1 
R 
m1  m2
  
m1 r1  m 2 r  m 2 r1
=
m1  m 2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

71
 =
m1  m2 r1  m2 r
m1  m2

 m2 r
 r1 
m1  m2
 
 m2 r
r1  R  ….(4)
m1  m2
 m1 r2  r   m2 r2
Similarly R 
m1  m2
  
m 1 r2  m1r  m2 r2

m1  m2

=
m1  m2  r2  m1r
m1  m 2

 m1 r
= r2 
m1  m 2

  m1r
r2  R  ….(5)
m1  m2

  m2 r
Therefore r1  R  and
m1  m2

  m1r
r2  R  ....(6)
m1  m2
 2  2
1   m 2 r  1   m1 r 
Hence L = m1 R    m 2 R    V( r )
2  m1  m 2  2  m1  m 2 

2
 
1   2 m 2 r 2 2Rm 2 r 
= m1 R   
2 
 m 1  m 2 2
m 1  m 2  

2
 
1   2 m1 r 2 2Rm1r 
 m 2 R     V( r )
2 
 m 1  m 2 2
m1  m2 


m1  m2  R 2  1 m1m2 m1  m2 2 r 2  V(r)

1 
=
2 2 m1  m2 

=
1
m1  m2  R 2  1 m1m2 r 2  V(r) ….(7)
2 2 m1  m2 

1  2 1  2
L = M R   r  V( r ) ....(8)
2 2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

72
 m1 m2
Where M  m1  m2 and  
m1  m2
 
Lagrangian equations of motion in terms of the two variables R and r will be

d  L  L
   0 ….(9)
dt  R  R

d  L  L
    0 ….(10)
dt  r  r
L  L
  M R and   0
R R

L  V
  r 
r r
Hence equations (9) and (10) become
d
dt
 

M R 0 ....(11)

dt
 
d  V
r 
r
0 ....(12)


Equation (11) giving M R  constant

(or) R  constant
That is velocity with which the centre of mass moves is constant.
 V
Equation (12) giving  r   = f (r)
r
representing equation of motion for the system under consideration. Consequently
we can ignore the first term in equation (8) and write
1 
L =  r 2  V(r ) ....(13)
2
which is effective in describing the motion of the components of r. But L is the
same as a single particle of mass µ moving at a distance r from a fixed centre of force
which gives rise to the potential energy V(r). Thus two body problem can be reduced to
the equivalent one body problem.
3.2 EQUATIONS OF MOTION AND FIRST INTEGRALS:
Let us describe the position of the particle in the plane poloar co-ordinates r and 
1

T = m r 2  r 2 2
2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
73
 V = V (r)
The Lagrangian L = T –V

=
1
2
 
m r 2  r 2 2  V(r )

As  is cyclic coordinate so that its conjugate angular momentum Pθ given by

L
P =  mr 2   l


d  L  d
P    

dt    dt
mr 2  0   .…(1)

Integrating, m r 2  constant = l

l first integral and represents the magnitude of angular momentum.
d
dt

m r 2  0 
d 2
dt
 
r  0

d  1 2 
 r   0
dt  2 
1 2
r   constant .…(2)
2
ie, Areal velocity = constant
The rate at which the area swept out by the radius vector is constant which is
Kepler‟s second law of planetory motion.
Suppose dA is the area swept out by the radius vector in time dt.

r rd 
1
Then dA 
2
1 2
= r d
2
dA 1 2 d 1 2 
 r  r
dt 2 dt 2
Therefore from equation (2), we write
1 2
r  = constant (or)
2
dA
= constant .…(3)
dt

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

74
Here we have two Lagrangian equations of motion. ie,
d  L  L
  0 .…(4)
dt  r  r
d  L  L
and   0 .…(5)
dt    
L L dV
We have  mr and  mr 2 
r r dr
L L
 mr 2  and 0

 
Now equations (4) and (5) become
d
mr    mr 2  dV   0 .…(6)
dt  dr 

And
d
dt

mr 2  0 .…(7)

Equations (6) and (7) are the equations of motion

(i) Expression for r(t) and  (t):
From equation (6) we write
dV
m r  m r  2  
dr
l
putting   2 we get
mr
2
 l  dV
m r  mr 2   
 mr  dr

l2 dV
m r  3

mr dr
d  l2 
m r   
dr  2mr 2 
V

d  l 
mr r   V r
dr  2mr 2 

d 1 2 d  l 2  dr
(or)  mr     V  
dt  2  dr  2mr 2  dt

d  l2 
  V  
dt  2mr 2 

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

75
So that

d 1 2 l2 
 mr 
  V ( r ) = 0
dt  2 2mr 2 
1 2 l2
(or) mr   V (r ) = constant
2 2mr 2
putting l  mr 2  , then

1 2 l2 1 2 m2r 4 2
mr   V (r )  mr   V (r )
2 2mr 2 2 2mr 2
1 1
 mr 2  mr 2 2  V (r )
2 2
1
 
 m r 2  r 2 2  V (r )
2
 TV  E
Thus we write
1 2 l2
mr 
  V (r )  E
2 2mr 2 ….(8)
E is a constant of motion. This is another first integral of motion.
1

2  l 2 
2
dr
r    E  V   
2 
m  2mr  dt

dr
dt  1

2  l 2 
2

  E  V  
m  2mr 2 

On integration we get
r
dr
t 1 .…(9)
2 l 
2
r0 2

  E  V  
m  2mr 2 

where r  ro at t  0.
This equation gives r as a function of time. ie r (t).
l
We have  
mr 2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

76
 d l

dt mr 2
t
l dt
   2  o .…(10)
m o r (t )

Thus integration of equations (9) and (10) provides us r (t) and  (t). Then we can
locate the position of the particle on the path at any time t and the solution.
3.3 CLASSIFICATION OF ORBITS:
We have already derived
d  l2 
m r   V  
dr  2m r 2 

dV l2
= 
dr m r 3

l2
= f (r )  3
mr
= f ′ = effective force.

f= f (r ) 
mr  
2 2

mr 3
= f (r )  m r  2

= f (r ) 
m r 
2

r
2
m v
= f (r ) 
r
= Central force + Centrifugal force.
Thus equation takes the form
m r  f 
It is an equation of motion for a particle subjected to actual force f (r) and pseudo
l2 dV 
centrifugal force 3
. Also f   
mr dr
V  = effective potential energy
 l2 
V  =   f  dr     f r   3  dr
 mr 

l2
=   f (r ) dr 
2mr 2
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
77
 l2
= V (r ) 
2mr 2
The second term is the potential energy related to the centrifugal force.
(i) Orbits under inverse square law of force :
The effective potential energy
k l2
V (r )    dr 
r2 2mr 2
k l2
= 
r 2mr 2
as a function of r has been plotted.
(ii) Motion with different values of k:

Case (i) : E > 0:

There is a minimum radial distance r3 but no maximum. The motion is
unbounded. The particle comes in from r = to a turning point and travels out to
infinity again; Thus the motion is not periodic and with a single turning point.

Case (ii) : 0 > E > Vmin = E m :

corresponds to energy E1

The radial motion is confined to the values r = r1= rmin and r = r2= rmax

r1 and r2 are the turning points.

1 l2
E1  mr 2   V (r )
2 2mr 2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

78
 1 2
 mr  V (r )
2

1 2
E1  V (r )  mr  T
2

Though r =0 at these points   0

The particle will not be at rest at these points. The motion is confined between the
areas of two circles of radii r1 and r2.

A possible shape for an attractive inverse square law of force is an ellipse with the
focus at the centre.

When r varies from r1 to r2 and back, the radius vector turns through an angle  obtained
by 

l
 dt  0
mr 2

dr
dt  1
2  l 2  2
  E  v  
m  2mr 2 

 l 
 2  dr
 r 
1
2 l 2  2
m   E  V  
m  2mr 2 

 l 
 2 dr
 r   o
1
 l 2 2

 2m( E  V )  2 
 r 

2m
When   , the path is a closed orbit. When  is not a rational fraction of 2π, the
n
path has the shape of a rosette. Such an orbital motion is processing motion.

 : r  0 and   finite, the particle moves in a circle.

Case (iii) : E  Em  Vmin

Energy Em corresponds to the minimum value of V(r)

dV  d   k l2 
   
dr dr  r 2mr 2 
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
79
 k l2
= 
r 2 mr 3
Putting r = ro for which V(r) is minimum.

k l2 k l2
2
 3
 o  2
 3
ro mro ro mro

l2
ro 
mk
E4 = V (r ) |min

k l2
= 
ro 2mro 2

mk 2 1 mk 2 1 mk 2
=   
l2 2 l2 2 l2
In this case two bounds coincide
r3 = r2 = ro
l2
Orbit is a circle of radius .
mk
 : r will be imaginary and therefore no physically meaningful motion is
Case (iv) : E< Vmin
possible.
(iii) Stability of orbits and conditions for closure:
(a) Stability:
The condition of stability in radial motion is given by the existence of local
minimum in V(r), the effective potential. That is we require that
 2V (r )
0 at r = ro
r 2
 2V (r )
given by 0
r
For any central force, the potential energy function is given by,
V (r )   krn 1 (for bounded motion)

l2
V (r )   k r n 1 
2mr 2

V (r ) l2
 k (n  1)r 
n

r mr
3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

80
 V (r ) l2
 k (n  1)r0  0
n

r r  r0 mr0
3

l2
k (n  1)r0  
n
3
mr0
 ( 3 n)
l 2r0
k (n  1)  
m
 2V (r ) n 1 3l 2
  kn( n  1) r 
r 2 r  r
0 4
0
mr0

l 2  (3 n ) n 1 3l 2  4
n ro ro  ro
m m
l 2  4 3l 2  4
n r0  r0
m m
l2 4
 (n  3)r0
m
Which is positive if n >3
Therefore any circular orbit with r = r0 under any central force can satisfy the
stability condition if n >3.
(b) Conditions for closure:
(iii) All bound orbits are closed only if for the inverse square law of force of
electrostatic attraction or gravitational type and for Hooke‟s type linear law of
force.
(iv) The condition for bound motion is that there is a bounded domain of r in
which V(r)  E. The condition for stability of circular orbits is n >3 where f
(r)  r n. That is for n = 1 and n =  2 only closed orbits exist.
3.4 VIRIAL THEROEM:
We consider a system of mass points with position vectors ri and applied forces Fi .

Fundamental equations of motion will be

p i  Fi
….(1)
Let us write a quantity
G   pi . ri
dG
  pi .ri   p i .ri
dt ….(2)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

81
But  ri . pi   mi ri . ri

=  mi vi
2

= 2T
and  p i . ri   Fi . ri

G
So that  2T   Fi . ri
t
The time average over a time interval  will be
z
1 dG
 0 dt
dt  2T   Fi . ri

dG
(or) 2T    Fi . ri 
dt


1
G( )  G(0) ....(3)

If the motion is periodic and  is its period then G() = G(0) and hence the RHS of
equation (3) is zero. If the motion is not periods but there is an upper bound to G and the
co-ordinates and velocities remain finite, then  sufficiently large, RHS again be
approximated to zero. Hence

2T   Fi . ri  0
1
T   Fi . ri .…(4)
2
Equation (4) is known as the virial theorem and RHS is called virial of claussius.
This theorem is used in kinetic theory of gases to prove Boyle‟s law and to obtain
equation of state for imperfect gases etc.
It the forces are derivable from a potential, then the theorem becomes.
1
T    iV ri
2

1  V 
   rˆ . r
2  r  i

1 V
= r
2 r
If the potential energy is
V  k r n 1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

82
 V
 (n  1) kr n
r
1 1 1
Then T  (n  1)k r n r  (n  1)k r n 1  (n  1)V
2 2 2
For inverse square law n = 2, hence
1
T   V  2T  V  0
2
3.5 DIFFERENTIAL EQUATION OF THE ORBIT:
Here we wish the obtain an equation that may provide a relation between r and .
i.e r  r()
Such an equation will be the equation of the orbit.
It is convenient to introduce the new variable
1 1
u (or ) r 
r u
1 du 1 du d
so that r    2
2
u dt u d dt
du 
  r2 
d
l du   l 
r   since 
m d mr 2 

l d  du 
And r    
m dt  d 

l d  du  d
  
m d  d  dt

l d 2u 
 
m d 2

l 2u 2 d 2 u

m 2 d 2
We know that the equation of motion of a particle is
dV
m r  m r  2 
dr
l2
  f (r )
mr 3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

83
  l 2u 2 d 2u  l 2 3 1
m  2 2
 u  f( )
 m d  m u
d 2u l 2 3   m   1   m 
 u  2 2   f   2 2 
d 2
m l u   u  l u 
m 1
= u  f 
l u u
2 2

Which is the differential equation of the orbit.

In case l = 0, equation is absurd but from
l  mr 2

m r 2  0

  0
 = constant
giving a straight line through the origin.
3.6 INTEGRABLE POWER LAW IN TIME IN KEPLER’S PROBLEM:
The inverse square law of force is most important of all the central force laws. It
results the deduction of Kepler‟s laws.
1. All the planets move in an elliptical orbit with sun at one of its foci.
2. The radius vector connecting the sun and the planet sweeps at equal areas is equal
intervals of time. ie, areal velocity is constant.
3. The square of the period of revolution of any planet about the sun is proportional
to the cube of the semi major axis.
(i) Deduction of first law:
The central force varies inversely as the square of the distance. That is
k
f (r )  k is a constant
r2
The corresponding potential energy will be
k
V (r ) 
r
d 2u m 1
Equation of the orbit is u  2 2 f  
d 2
l u u

1 k
putting f (r )  f    2  ku2 we get
u r

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

84
 d 2u mk
u 2 ….(1)
d 2
l
mk
Let us put y  u  so that
l2
d2y d 2u

d 2 d 2
Equation of the orbit then becomes
d 2u
 y0
d 2
The general solution of this equation is
y  u cos   

 u cos   
mk
(or) u
l2

u   2  u cos    
1 mk
r l
Where u and  are constants.
If we orient our co-ordinate system so that =0, then
1 mk
 2  u  cos 
r l
mk  u  l 2 
 2 1  cos 
l  mk 
l2
(or) r mk .…(2)
 u l 2 
1  mk cos 
 
Thus  is a turning point.
We show that equation (2) represents a conic section, coincident with Kepler‟s
first law of planetary motion. That is the orbits are conic sections with the centre at one of
the foci.
We define the conic section as a curve for which the distance from a fixed point to
that from a fixed line is a constant. That is
r
 constant = 
d
 = eccentricity

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

85
From figure, We see that
P =d + r cos θ
p
Let p = P  P 

p
 d  r cos 

r
=  r cos 

p = r + r cos
= r (1+ cos)
p
r = .…(3)
1  cos 
Which represents a conic section.
Equation (2) is of the form of equation (3) orbit under an inverse square force is always a
conic section.
Comparing equations (2) and (3), we get
l2 ul 2
p and  
mk mk
r
Further from   , we must   0 or r must be positive.
d
(ii) To explore the shapes of orbit:
Equation (2) is the equation of the conic and consequently it should have been
possible to explore the shape of orbits using this equation. Here  involves one unknown
constant u. Hence we express  in terms of known constants. Then we put the value of 
in one equation of the conic which yield the desired information about the shape of the
conic.
1 l2
Already we have E = m r 2   V (r )
2 2mr 2
k
Here V (r) =
r
1 l2 k
E = m r 2  2
 ....(4)
2 2mr r
Suppose we take the turning point at which r is minimum say rmin. Then r  0.
Now equation (4) becomes.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
86
 l2 k
E = 2
 ....(5)
2m rmin rmin

p l2
From equation (3) rmin  
1   mk (1  )
Now equation (5) becomes

l 2  mk 1     mk 1   
2

E    k 
2m  l 2
  l2 

mk 2 mk 2
= 1   2
 1  
2l 2 l2
mk 2
= 1   1    2
2l 2
mk 2
= 1    (  1)
2l 2

=
mk 2 2
2l 2

 1 
2l 2E 2El 2
giving 2 1    2
 1 
mk 2 mk 2
1

 2El 2 
2

such that  = 1  2 
.…(6)
 mk 
now  is a known one. After putting the value of  and p in equation (3) we get
l2
r mk ….(7)
1/ 2
 2E l 2 
1  1  cos 
 mk 2 

If E > 0 giving  > 1 – conic is hyperbola

If E = 0 giving  = 1 – conic is parabola
If E < 0 giving  < 1 – conic is ellipse
 mk 2
If E = giving  = 0 – conic is circle.
2l 2
(iii) Case of elliptic orbits:
Relation between energy and semi major axis:
In the case of elliptic orbits, semi major axis is given by:-
when θ = 0, r = r1 = perihelion

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

87
 p
Then r1 =
1 
r = r2 = aphelion
p
Then r2 =
1 
a is the sum of one half of perihelion r1 and aphelion r2
r1  r2
a
2
p 1 1 
=  
2 1   1   
p 1    1   
=
2  1   2 
p
 
=
1  2

l2
But p =
mk

l2
Hence a =

mk 1   2  .…(8)

l2
(or) 1 - 2 =
mka

Now E
mk 2 2
2l 2

  1 becomes 
mk 2 l2 
=   From equation (8)
2l 2  mka 
k
E =
2a
which shows that all ellipses with the same major axis have the same energy.
(iv) Deduction of Kepler’s second law:
We know

p 
d  L  d
 
dt    dt
mr 2  0  
d 2
dt
 
r  0

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

88
 d  1 2 
 r 0
dt  2 
1 2
r   constant
2
ie Areal velocity = constant
The area swept out by radius vector per unit time = constant.
If r describes an angle d in time dt, the area swept out by r in time dt is

dA  r rd   r 2 d
1 1
2 2
dA 1 2 d
 r
dt 2 dt
1 d
 r2
2 dt
1 2
= r  = constant
2
which is Kepler‟s second law.
(v) Deduction of third law : (period of elliptic motion )
 is the ratio of the total area of the ellipse to the rate at which the area is
swept out. Suppose dA is the area swept out by radius vector is time dt, then
the rate will be dA / dt.
area
Hence 
dA / dt
Area of the ellipse =  a b
dA 1 2  l
And  r 
dt 2 2m
ab 2mab
  
l / 2m l

b  a 1  2 
1
2
putting

2ma 2
 
1

  1  2
2
Then
l
4 2 m 2 a 4
2 
l2

1  2 
4 2 m 2 a 4 l 2
=
l2 mka

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

89
  4 2 m  3
=   a
 k 
2  a3
Which is Kepler‟s third law. It states that the square of period is proportional to cube of
the semi-major axis.
3.7 SCATTERING IN CENTRAL FORCE FIELD:

(i) Rutherford scattering:

Assumptions:

1. The heavy nucleus and the positively charged particle to be point nucleus so that their
dimensions are not taken into account.
2. The nucleus of the atom is so heavy that it is at rest during collision.
3. The mass of the positively charged particle may be taken as constant because the
velocity is very small compared with the velocity of light.

Let a positively charged particle of charge z′e approach a heavy nucleus N of

charge ze. There will be a force of repulsion between them. The force increases as the
particle gets closer to the nucleus. The positively charged particle of initial velocity vo is
repelled by the heavy nucleus and changes from a straight line to a hyperbola PAQ
having one focus at N. The asymptotes PO and OQ give the initial and final directions of
the particle. As the initial and final directions are not the same, the particle is said to be
scattered.

The perpendicular distance of PO from N =MN = s. This is the shortest distance from the
nucleus to the initial direction called impact parameter.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

90
The equation of the orbit

 1   cos(   )
1 1
r p

1/ 2
l2  2 El 2 
p and   1  2 
mk  mk 

In this case k=-zz′e2 and hence

1 mzze2
 2
p l

mzze2
1
 1   cos(   )
r l2

If the initial line is set such that    0 , then

mzze2
1
 1   cos  
r l2

1/ 2
 2 El 2 
  1  2 
 mk 

1/ 2
 2 El 2 
 1  2 2
 m( zze ) 

1/ 2
 2 El 2 
 1  2 2 4 
 mz z e 

If the initial velocity is vo, then

1
E and mv o  2mE
2
mv o
2
l
l  mv o s and mv o 
s

According to principle of conservation of angular momentum

mvo s  mr 2   l
l
mvo 
s

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

91
 l
 2mE
s

l  s 2mE


 2 E s 2mE 2 
  1  
 1/ 2

 mz 2 z 2 e 4 

1/ 2
  2 Es  2 
 1   
2 

  zz  e  

From this equation it is clear that   1 .

mzze2
1
  1   cos  
r l2
represents a hyperbola.

(ii) Angle of scattering:

The angle between initial and final directions of the positively charged particle is
called angle of scattering. That is the angle between the asymptotes is called angle of
scattering. Here  is the angle of scattering.

The asymptotic directions are those for which r is infinite and    ,

1   cos   0

1
cos   


From figure   2  

 
 
2 2

   1
 cos     
 2 2 

 1
sin 
2 


cos ec  
2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

92
 
cos ec 2 2
2


2
 2 Es 
1  cot
2
   1 
2
2
2  zze 

  2 Es 
2

cot 2 
2  zze 2 

  2Es 
cot 
2  zze 2 

 2Es 
  2 cot 1 
 zze 
2

This equation gives  in terms of impact parameter s, energy E, the charge on the
nucleus ze and the charge on the particle z e.

(v) Rutherford scattering cross section:

The scattering cross section is defined as

number of particles scattered into a solid angle dω per unit time

()d 
incident intensity

dω is the element of solid angle

 ( ) is the differential scattering cross section.

The incident intensity is defined as the number of particles crossing unit area normal to
the incident beam in unit time.

The differential of solid angle dω in the plane whose azimuth lies between  and
 + d  is d  2 sin  d

The scattering cross section through angle  in any plane is

2
 ( ) d    ( ) sin  d d  2 () sin  d
o

The number of particles scattered into solid angle dω per unit time  2 I () sin  d

I is the incident intensity.

The cross section for the particles having collision parameter between s and s+ds is the
area of ring of radius s and width ds so that
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
93
  (s, ds)  2sds

The number of particles lying between s and s+ds  2 Is ds

Now 2 Is ds   2 I () sin  d 

The negative sign is due to the decrease in d  as the increase in ds.

  
sds
sin d

 2 Es
cot 
2 zze 2
zze 2 
gives s  cot
2E 2

zze2  d
Then ds   cos ec 2
2E 2 2

 zze 2     zze 2   d
   cot   cos ec 2
 ()  
2E  2  2E  2 2
 
2 sin cos d
2 2
2
1  zze 2  
   cos ec 4
4  2E  2

This is Rutherford scattering cross section.


The scattering cross section must be proportional to i) cos ec 4 , ii) the square of (ze),
2
iii) the square of (ze) and iv) inversely proportional to square of kinetic energy (E).

HAMILTON – JACOBI THEORY

In canonical transformation, the method involves the transformation of old set of

coordinates ( qk ) to new set of coordinates ( Qk ) which are all cyclic and hence all
momenta are constants provided the Hamiltonian is conserved.

In case the Hamiltonian involves time, an alternate approach is used to a canonical

transformation which leads to the new Hamiltonian H   0 so that the new coordinates
and momenta Qk and Pk are constants. This procedure is due to Jacobi which is a

transformation as well as a method itself and applied when Hamiltonian involves time.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

94
3.8. HAMILTON – JACOBI EQUATION:

If we make a canonical transformation from old set of variables ( qk , pk ) to a new

net of variables ( Qk , Pk ), then the new equation of motion are

H  H 
Pk   and Q k  …(1)
Qk Pk

If we require that the transformed Hamiltonian H′ =0, then equations of motion (1)
assume the form

Pk  0 and Q k  0

Pk  constant and Qk  constant …(2)

Thus the coordinates and momenta are constants in time and they are cyclic.

The new Hamiltonian H  is related to old Hamiltonian H by

F
H  H 
t

Which will be zero when F satisfies the relation

F
H ( qk , pk , t )  0 ...(3)
t

Where H (qk , pk , t ) is written for H (q1 , q2 ,...qn , p1 , p2 ,... pn , t )

We take the generating function F as a function of the old coordinate qk , the new

constant momenta Pk and time t ie, F2 (qk , Pk , t ) . Then

F2
pk  …(4)
qk

F2 F
Therefore H (qk , , t)  2  0 …(5)
qk t

F2  F2 (qk , Pk , t )

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

95
The total time derivative of F2 is

dF2 n F n F F
  2 qk   2 Pk  2
dt k 1 qk k 1 P
k t

F F2
Here Pk  0, 2   H and  pk
t qk

dF2 n
Therefore   pk qk  H  L
dt k 1

(or) F2   Ldt  S …(6)

=Hamilton‟s principal function

S S
Put F2  S in equation (5), we get H (qk , ,t)  0 …(7)
qk t

This is known as Hamilton – Jacobi equation. It is a partial differential equation of first

order in (n+1) variables q1 , q2 ,...qn , t.

Solution to Hamilton – Jacobi Equation :

Let the solution of equation (7) be of the form

S  S (q1 , q2 ,...qn ,1 , 2 ,... n , t ) …(8)

Where 1 ,  2 ,... n are n independent constants of integration.

In equation (8), the solution S is a function of n co-ordinates of qk ,time t and n

independent constants. We can take these n constants of integration as the new constant
momenta

ie Pk   k …(9)

Now the n transformation equation are

S (q1 , q2 ,...qn ,1 , 2 ,... n , t )

pk  …(10)
qk

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

96
 These are n equations at t = t0 give the n values of in terms of  k in terms of qk and pk .
The other n transformation equations are

s
Qk   k
Pk

S (q1 , q2 ,...qn , 1 ,  2 ,... n , t )

or k  …(11)
 k

Similarly one can calculate the constants  k by using initial condition at t = t0, the

known initial values of qk . Thus  k and  k constants are known and equation (10) will

give qk in terms of  k ,  k and t. that is

𝑞𝑘 = 𝑞𝑘 (𝛼1 , 𝛼2 , … 𝛼𝑛, 𝛽1, 𝛽2 … 𝛽𝑛 , 𝑡) …(12)

On differentiation of equation (10), equation (12) may be substituted for 𝑞𝑘 to obtain

momenta 𝑝𝑘 . Thus 𝑝𝑘 will be obtained as function of constants 𝛼𝑘, 𝛽𝑘 and t That is

𝑝𝑘 = 𝑝𝑘 (𝛼1 , 𝛼2 , … 𝛼𝑛, 𝛽1, 𝛽2 … 𝛽𝑛 , 𝑡) …(13)

Thus we see that the Hamilton‟s principal function S is the generaor of a canonical
transformation to constant coordinates (𝛽𝑘 ) and momenta (𝑞𝑘 ). Also in solving the
Hamilton- Jacobi equation, we obtain simultaneously a solution to the mechanical
problem.

3.9 SOLUTION OF HARMONIC OSCILLATOR PROBLEM BY HAMILTON –

JACOBI METHOD:-

Consider a one - dimensional Harmonic oscillator. For such a system forces are
conservative. The force acting on the oscillator at a displacement q is

𝑓 = −𝑘𝑞

k =Force constant.

q
1
Potential energy V    kqdq  kq2
o
2

1 𝑝2
Kinetic energy 𝑇= 𝑚𝑣 2 =
2 2𝑚
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
97
 Hamiltonian 𝐻 =𝑇+𝑉

𝑝2 1
= + 𝑘𝑞 2
2𝑚 2

𝜕𝑆
But 𝑃 = and therefore
𝜕𝑞

1 𝜕𝑆 2 1
𝐻= + 𝑘𝑞 2
2𝑚 𝜕𝑞 2

Hence the Hamilton – Jacobi equation corresponding to the Hamiltonian is

𝜕𝑆
𝐻+ =0
𝜕𝑡

1 𝜕𝑆 2 1 𝜕𝑆
+ 𝑘𝑞 2 + =0 …(1)
2𝑚 𝜕𝑞 2 𝜕𝑡

Since we can separate the variables, solution will be of the form

𝑆 𝑞, 𝛼, 𝑡 = 𝑊 𝑞, 𝛼 − 𝛼𝑡 …(2)

α is a constant.

𝜕𝑆 𝜕𝑊 𝜕𝑆
= 𝑎𝑛𝑑 = −𝛼
𝜕𝑞 𝜕𝑞 𝜕𝑡

Now equation (1) takes the form

1 𝜕𝑊 2 1
+ 𝑘𝑞 2 = 𝛼 …(3)
2𝑚 𝜕𝑞 2

1 𝜕𝑊 2 1
= (𝛼 − 𝑘𝑞 2 )
2𝑚 𝜕𝑞 2

𝜕𝑊 2 1
= 2𝑚 (𝛼 − 𝑘𝑞 2 )
𝜕𝑞 2

𝜕𝑊 1
= 2𝑚 (𝛼 − 𝑘𝑞 2 )
𝜕𝑞 2

On integrating we get

1
W =  2m (  kq2 ) dq  c
2

c = constant of integration
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
98
 1
Then S =  2m (  kq2 ) dq  t  c
2

c is an additive constant and will not affect the transformation. Because to obtain the
𝜕𝑆
new positon coordinate 𝛽 = only partial derivative of S w. r. to α is required. Hence
𝜕𝛼

c is dropped. Thus

1
S= 2𝑚 (𝛼 − 𝑘𝑞 2 ) 𝑑𝑞 − 𝛼𝑡 …(4)
2

Now 𝛼 = new momentum P.

The new constant co - ordinate is obtained by the transformation.

S 2m dq
2 
  t
 1 2
  kq
2

m dq
2
 t
1 2
  kq
2

m dq
2
 t
k 2
 kq2

2 k

m dq

k  2
t
 kq2

m dq
k
t 
2
 kq2
k

𝑚 𝑘
= sin−1 𝑞
𝑘 2𝛼

𝑘 𝑘
sin−1 = (𝑡 + 𝛽)
2𝛼 𝑚

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

99
 𝑘
𝜔= 𝑚

𝑘
sin−1 𝑞 = 𝜔 (𝑡 + 𝛽)
2𝛼

2𝛼
𝑞= sin 𝜔 (𝑡+ 𝛽) …(5)
𝑘

= 𝑞0 sin 𝜔 (𝑡 + 𝛽)

Which is the familiar solution for the harmonic oscillator.

𝜕𝑆 𝜕𝑊
𝑁𝑜𝑤 𝑝 = =
𝜕𝑞 𝜕𝑞

1
= 2𝑚(𝛼 − 𝑘𝑞 2 )
2

= 2𝑚 𝛼 − 𝑚2 𝜔 2 𝑞 2 …(6)

2𝛼
𝑝= 2 𝑚 𝛼 − 𝑚2 𝜔 2 sin2 𝜔 (𝑡 + 𝛽)
𝑘

= 2 𝑚 𝛼 − 2 𝑚 𝛼 sin2 𝜔 (𝑡 + 𝛽)

= 2 𝑚 𝛼 cos 𝜔 𝑡 + 𝛽 …(7)
= 𝑝0 cos 𝜔 𝑡 + 𝛽

The constants 𝛼 and 𝛽 are to be known from initial conditions.

At t = 0 the particle is at rest

ie, p = 0 and it is at the displacement 𝑞 = 𝑞0 from the equilibrium position .

From (6) 2𝑚 𝛼 − 𝑚2 𝜔 2 𝑞 2 = 0

2 𝑚 𝛼 = 𝑚2 𝜔2 𝑞 2

1
𝛼 = 𝑚𝜔2 𝑞 2
2

1
= 𝑘 𝑞02 …(8)
2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

100
 2𝛼 2𝛼
𝑞0 = =
𝑚𝜔 2 𝑘

and hence the solution (5) takes the form

𝑞 = 𝑞0 sin 𝜔 (𝑡 + 𝛽) …(9)

at 𝑡 = 0 𝑞 = 𝑞0 , cos 𝜔 𝛽 = 0 𝑎𝑛𝑑 sin 𝜔 𝛽 = 1

𝜋 𝜋
𝜔𝛽= (𝑜𝑟) 𝛽 =
2 2𝜔

Thus the new constant canonical coordinate measures the initial phase angle and in the
𝜋
present initial conditions the initial phase 𝜔𝛽 = .
2

Now equation (9) become

𝑞 = 𝑞0 cos 𝜔𝑡. …(10)

Hamilton‟s characteristic function W (q,α) and principal function S are related by

𝜕𝑠
𝑆 = 𝑊 𝑞, 𝛼 − 𝛼𝑡; 𝐴𝑙𝑠𝑜 𝐻 ′ = 𝐻 + 𝜕𝑡 = 𝐻 − 𝛼 = 0

𝐻 = 𝛼 But the system is conservative.

Hence 𝐻 = 𝐸 Thus the new canonical momentum (𝑃) is the total energy of the oscillator

H.J method to Harmonic oscillator :

Hamilton‟s principal function S can he obtained as

𝑆= 𝑝𝑑𝑞 − 𝛼𝑡
𝑃= 2 𝑚 𝛼 cos 𝜔 (𝑡 + 𝛽)

𝑞 = 𝑞0 sin 𝜔 (𝑡 + 𝛽)

2∝
= sin 𝜔 𝑡 + 𝛽
𝑘

2𝛼
𝑑𝑞 = 𝜔 cos 𝜔 𝑡 + 𝛽 𝑑𝑡
𝑘

2𝛼
𝑆= 2𝑚𝛼 cos 𝜔 𝑡 + 𝛽 𝜔 cos 𝜔 𝑡 + 𝛽 𝑑𝑡 − 𝛼𝑡
𝑘

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

101
 2𝛼
= 2𝑚𝛼 𝜔 𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 𝑑𝑡 − 𝛼𝑡
𝑘

= +2𝛼 𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 𝑑𝑡 − 𝛼𝑡

1
= 2𝛼 [𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − 2] 𝑑𝑡

The Lagrangian L is given by

𝑝2 1
𝐿= − 𝑘𝑞 2
2𝑚 2

1
= 𝛼𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − 𝑘 𝑞0 2 𝑠𝑖𝑛2 𝜔 𝑡 + 𝛽
2

1 2𝛼
= 𝛼𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − 𝑘 𝑠𝑖𝑛2 𝜔 𝑡 + 𝛽
2 𝑘

= 𝛼 𝑐𝑜𝑠 2 𝑤 𝑡 + 𝛽 − 𝛼 𝑠𝑖𝑛2 𝜔 𝑡 + 𝛽

= 𝛼[𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − 𝑠𝑖𝑛2 𝜔 𝑡 + 𝛽 ]

= 𝛼[𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − (1 − 𝑠𝑖𝑛2 𝜔 𝑡 + 𝛽 )]

= 𝛼[2𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − 1]

1
= 2𝛼[𝑐𝑜𝑠 2 𝜔 𝑡 + 𝛽 − ]
2

There fore 𝑆= 𝐿𝑑𝑡

Thus for Harmonic oscillator we prove that the Hamilton‟s principal function is the time
integral of Lagrangian.

3.10 HAMILTON’S PRINCIPAL FUNCTION :

From the solution of Hamilton Jacobi equation, we recognize S, the

Hamilton‟s principal function, as the generating function which gives rise to a canonical
transformation involves constant momenta and constant co - ordinates.

Consider the total time derivative of

𝐹2 = 𝐹2 (𝑞𝑘, 𝑃𝑘, 𝑡)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

102
 dF2 n F n F F
  2 qk   2 Pk  2
dt k 1 qk k 1 P
k t

F F2
Here Pk  0 , 2   H and  pk
t qk

dF2 n
Therefore   pk qk  H  L
dt k 1

(or) F2   Ldt  S

S is known as the Hamilton‟s principal function in relation to the variational principle

3.11 HAMILTON’S CHARACTERISTIC FUNCTION:

Conservative systems:-
In a particle moving under central force in which H does not depend on time t explicitly
and hence H = constant = 𝛼1 = E. In such cases, it is Hamilton‟s characteristic function
W. In Hamilton‟s principal function S, an explicit dependence on time is involved.

Hamilton – Jacobi equation for Hamilton‟s principal function S (𝑞𝑘, 𝛼𝑘, 𝑡) becomes
𝜕𝑆 𝜕𝑆
𝐻 𝑞𝑘, 𝜕𝑞 + =0 …(1)
𝑘 𝜕𝑡

We can assume the solution S in the form

𝑆 𝑞𝑘, 𝛼𝑘, 𝑡 = 𝑊(𝑞𝑘, 𝛼𝑘 ) − 𝛼1 𝑡 …(2)

𝜕𝑆 𝜕𝑊 𝜕𝑆
There fore = 𝑎𝑛𝑑 = −𝛼1
𝜕𝑞 𝑘 𝜕𝑞 𝑘 𝜕𝑡

and hence the Hamilton – Jacobi equation take the form

𝜕𝑊
𝐻 𝑞𝑘 , 𝜕𝑞 = 𝛼1 …(3)
𝑘

𝜕𝑊 𝜕𝑊 𝜕𝑊
(or) H [𝑞1, 𝑞2,… 𝑞𝑛 , 𝜕𝑞 , 𝜕𝑞 , … 𝜕𝑞 ] = 𝛼1 …(4)
1 2 𝑛

This is time - independent Hamilton – Jacobi equation.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

103
For conservative system H = α1= E =total energy of the system. Now Hamilton – Jacobi
equation is written as
𝜕𝑊
𝐻 𝑞𝑘, 𝜕𝑞 =𝐸 …(5)
𝑘

Equation (4) can also be obtained directly by taking W as the generating

function W (𝑞𝑘 , 𝑃𝑘 ) independent of time. The transformation equation are

𝜕𝑊 𝜕𝑊
𝑝𝑘 = 𝑎𝑛𝑑 𝑄𝑘 = 𝜕𝑃 …(6)
𝜕𝑞 𝑘 𝑘

Now if the new momenta 𝑃𝑘 are all constant of motion 𝛼𝑘 , where 𝛼1 in particular is the
𝜕𝑊
constant of motion H, then 𝑄𝑘 = 𝜕𝛼
𝑘

The condition to determine W is that

𝐻 𝑞𝑘 , 𝑝𝑘 = 𝛼1

𝜕𝑊
Using 𝑝𝑘 = , we obtain
𝜕𝑞 𝑘

𝜕𝑊
𝐻 𝑞𝑘, = 𝛼1
𝜕𝑞 𝑘

which is identical to equation (4).

𝜕𝑊
Also 𝐻 ′ = 𝐻 + 𝜕𝑡

But 𝑤(𝑞𝑘 , 𝑝𝑘 ) does not involve time and hence

𝐻 ′ = 𝐻 = 𝛼1 (= 𝐸) …(7)

𝑊 = Hamilton‟s characteristic function.

It generates a canonical transformation where all new coordinates 𝑄𝑘 are cyclic because
𝐻′ = 𝛼1, depending on one of the new momenta 𝑃1 = 𝛼1, and does not contain any 𝑄𝑘 .

𝜕𝐻 ′
Now 𝑃𝑘 = − 𝜕𝑄 = 0 𝑜𝑟 𝑃𝑘 = 𝛼𝑘
𝑘

𝜕𝐻 ′
and 𝑄𝑘 = 𝜕𝛼 = 1 𝑓𝑜𝑟 𝑘 = 1 …(8)
𝑘

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

104
 𝜕𝐻 ′
𝑄𝑘 = 𝜕𝛼 = 0 𝑓𝑜𝑟 𝑘 ≠ 1
𝑘

Hence the solution are

𝜕𝑊
𝑄1 = 𝑡 + 𝛽1 = 𝜕𝛼 𝑓𝑜𝑟 𝑘 = 1
1

𝜕𝑊
and 𝑄𝑘 = 𝛽𝑘 = 𝜕𝛼 𝑓𝑜𝑟 𝑘 ≠ 1 …(9)
𝑘

Thus out of all the new coordinates 𝑄𝑘, 𝑄1 is the only coordinate which is not a constant
of motion. Here we observe the conjugate relationship between the time as the new
coordinate and Hamiltonian as the conjugate momentum.

Physical significance of the Hamilton’s characteristic function W:-

The function W has a physical significance similar to the Hamilton‟s principal function S.
since W (𝑞𝑘, 𝑃𝑘 ) does not involve time t explicitly, its total time derivative is

dW n W n W
 q k   Pk
dt k 1 qk k 1 P
k

Since Pk   k  Pk  0 and therefore

dW n
  pk q k
dt k 1

W    pk q k dt    pk dqk …(10)
k k

Which is the action.

and S   Ldt   [ pk q k  H ] dt
k

= W   Hdt

When H does not involve time t explicitly

 Hdt   t
1

Now S  W  1t
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
105
 S (qk , k , t )  W (qk , k )  1t

S (qk , t )  W (qk )  1t …(11)

Here pk   k are constants.

1  E = total energy.

When the Hamiltonian does not involve time explicitly, one can solve a
mechanical problem by using either Hamilton‟s principal function (or) Hamilton‟s
characteristic function.

3.12 ACTION AND ANGLE VARIABLES:

In the motion of a system with many degrees of freedom the Hamilton- Jacobi
equation is completely separable in coordinate variables. We consider conservative
system in which the Hamiltonian does not involve time explicitly. The Hamilton-Jacobi
equation is given by

 W W W 
H  q1 , q2 ,........qn ; , ,.........   1 ….(1)
 q1 q2 qn 

The variables qk are separable, if a solution of the form

W  Wk qk ; 1 ,  2 ,.....  n  …(2)

k

splits the equation into n equations:

 Wk 
H k  qk ; ; 1 ,  2 ,.... n   1 …(3)
 qk 

Wk
Each of equations (3) involves only one of the coordinates qk and the corresponding
qk

The equations of canonical transformation has the form

Wk qk ; 1 ,  2 ......  n 

pk  …(4)
qk

it gives pk  ( pk ;1 , 2 ,..... n ) …(5)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

106
Equation (5) represents the orbit equation of the projection of the representative point of
the system on the ( qk , pk ) plane in the phase space. Now action-angle variables can be

defined if the orbit equations for all the ( pk , qk ) pairs describe either closed orbits or
periodic functions of q k.

The action variables J k are defined as J k   pk dqk …(6)

Here the integration is to be carried out over a complete period.

Suppose q k is cyclic coordinate, then p k =constant.

If q k is angle coordinate, then integral for action variable is to be taken from 0 to 2

giving.
2
J k  pk  dqk  2pk …(7)
0

Since action A    pk qk dt    pk dqk

k k

Using equation (4), we obtain

Wk
Jk   qk ; 1, 2, ........n  dqk …(8)
qk

Since q k is a variable of integration and it will be out, when integration is over. Thus J k

is a function of the n constants α k. Thus

J k  J k (1 ,  2 , ......  n ) …(9)

(or)  k   k ( J1 , J 2 ,...... J n ) …(10)

Thus the Hamilton‟s characteristic function is

W  W q1, q2 ,....... qn ; J1, J 2 ,.... J n 

 Wk (qk ; J1, J k , ....... J n ) …(11)

k

H  H   1 and 1  1 ( J1, J 2 ,.......J n )

H   H  ( J1 , J 2,.......J n ) …(12)

The generalized coordinate ωk conjugate to J k is defined as angle variable, given by

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

107
 W n Wl
k   q1; J1, J 2 ,........ J n  (13)
J k l 1 J k

Now the equation of motions for the angle variables are

H

k  J1 , J 2 , ....... J n 
J k

  k ( J1, J 2 ,.....J n ) …(14)

This gives solution of the form

k   k t  k …(15)

k s are constants and functions of action variables only and are frequencies of the
periodic motion. Here the frequencies of the periodic motion can be obtained without
complete solution of the problem.

Suppose that the change in angle variable ω k with the completion of one cycle by q l is 
l ω k, then

k
l k   dql
ql

W
Using k  we get
J k

 2Wk
 l k   dql
ql J k

 W

J k  q dq l
l



J k  p dq
l l

J l
   kl
J k

Thus  k k  1 k=l

=0 k≠l …(16)

η k is the period corresponding to q k then

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

108
using equation (15),  k k   k tk   k k …(17)

But  k k  1 hence

1
k k 1 (or ) k   k …(18)
k

Thus k s are identified to the frequencies of the periodic motion.

2.13 KEPLER’S PROBLEM IN ACTION-ANGLE VARIABLES:

k
For a particle of mass m, moving in an inverse square force field [V (r )   ], the
r
Hamiltonian of the system is given by

1  2 p  k
2
H  r
p  
2m  r2  r

Determine the frequency by the method of action-angle variables and discuss degeneracy.
Show that the period of the orbit is given by

m
  k 
2E 3

Solution :

The action variable Jk are given by

J k   pk dq k

In the case of Kepler‟s problem, the action variables are given by

J    p d

and J r   pr dr

W
p   2


and therefore
2
W
J   d     2 d     2 d   2 2
 0

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

109
 W 2mk  2
2
pr   2m1   2
r r r

Where we have replaced E by α1

W
 J r   pr d r   dr
r

2mk  2 2
  2m1   2 dr
r r

The motion is bounded and is elliptical path for negative value of the total energy
E. Further the limits are given by rmin and rmax values of r.

These values are determined by the zero of the quadratic equation in

 2 dr
 2mk  22
   2 .
r2 2m E   2
r r

2mk  22
ie 2m1   2 0
r r

2m 1r 2  2mkr  22  0

k 2
r2  r 2 0
1 2m1

k 1 2  2
r  k2  1 2
21 21 m

k  2  2 
=  1  1  1 2 2 
21  mk 

k  1  21 22 
ie rmin   1   and
21  mk 2 

k  1  21 22 
rmax   1  
21  mk 2 

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

110
In a complete cycle of the co-ordinate r, it varies from rmin to rmax and back to rmin.

Now

rmax
2mk  2
2
Jr = 2 
rmin
2m1 
r
 2 dr
r

2mk  2
2
rmax 2m1   2
=2  r r dr
2mk  2
2
2m1   2
rmin

r r

rmax
2m1r  2mk dr
=2 
rmin 2m1r  2mkr  
2 2
dr  2 22 
r 2m1r  2mkr   22
2
2

2mk
Jr =  2 2
 2m1 

J + Jr =
2mk
 2m1 
2mk  2

 2m1 J   J r 2

22 mk 2
1  
J   J r 2

But H = H = 1= E

22 mk 2
Therefore H   E  1  
J   J r 2

Now the frequencies  and r are given by

H     22 mk 2  42 mk 2
    2

J  J   J   J r   J   J r 
3

H     22 mk 2  42 mk 2
and r    
J r J r  J   J r 2  J   J r 3

42 mk 2
Thus   r 
J 0  J r 2
The two frequencies are equal and the motion of the system is said to be
degenerated.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
111
The period of the orbit:

1 J   J r 
3
 
r 42 mk 2
22 mk 2
E
J   J r 2
22 mk 2
J   J r 2

E
1

 22 mk 2  2
( J   J r )  
 E 
3

 22 mk 2 
2
1
 2 2  E 
4 mk  
1

 22 mk 2   22 mk 2 
2
1
=
42 mk 2  E   E 
 
1

1  22 mk 2 
2


2E  E 
=

 m 
2

= k  3 
 2E 
k
This formula agrees with Kepler‟s third law that the semi-major axis a   .
2E

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

112
UNIT IV : CLASSICAL STATISTICAL MECHANICS
Foundations of Statistical Mechanics: The macroscopic and microscopic states, postulate
of equal a priori probability, Contact between statistics and thermodynamics; Ensemble
theory: Concept of ensemble, phase space, Density function, Ensemble average,
Liouville’s theorem, Stationary ensemble; The microcanonical ensemble, Application to
the classical ideal gas; The canonical and grand canonical ensembles, Canonical and
grand canonical partition functions, Calculation of statistical quantities;
Thermodynamics of a system of non-interacting classical harmonic oscillators using
canonical ensemble, and of classical ideal gas using grand canonical ensemble, Energy
and density fluctuations; Entropy of mixing and the Gibb’s paradox, Sackur-Tetrode
equation.
4.1 FOUNDATIONS OF STATISTICAL MECHANICS:
 Statistical mechanics is the branch of science which gives the interpretation of the
macroscopic behavior of a system in terms of its microscopic properties.
 Statistical mechanics is not concerned with the actual motion of individual particle
but investigates average or most probable or statistical properties of the system.
 The larger is the number of particles in the physical system considered, the more
nearly correct are the statistical predictions. The smaller is the number of
particles in the mechanical system, statistical mechanics cease to have meaning.
 Statistical mechanics is applicable for a system consisting large number of
particles.
 There are two statistical methods known as classical statistics and Quantum
statistics.
 Classical statistics explained many observed physical phenomenon like
temperature , pressure ,energy etc., but could not explain several experimentally
observed phenomenon like black body radiation, specific heat at low temperature
etc.
 For explaining such phenomenon Bose-Einstein and Fermi-Dirac made new
approach known as quantum statistics.
 Quantum statistic can be classified as,
i. Bose-Einstein statistics
ii. Fermi-Dirac statistics

4.2 MICROSTATE AND MACROSTATE:

Microstate:
 The specification of individual position of phase points for each system or
molecule of the ensemble.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
113
Macrostate:
 The specification of the number of phase points in each cell of phase space. Many
different microstates may corresponds to same macrostate.
 Consider a system consists of four particles a,b,c, and d and two cells A and B.
The distribution of 4 particles in two cells can be illustrated here.

Cell A Cell B
4 0
3 1
2 2
1 3
0 4

(2,2)-macrostate (3,1 macrostate)

ab cd abc d

bc da bcd a

cd ab cda b

da bc dab c

 If we interchange any two phase points from different cells we have different
microstates but the same macrostate.
 If we interchange any two phase points in the same cell , we have same microstate
and same macrostate.
 The number of microstates corresponding to a given macrostate in called the
thermodynamic probability of the macrostate.
 The probability that the ensemble possessing energy E is proportional to Ω(E).
𝑃 𝐸 = 𝑐 𝛺(𝐸)
c→proportionality constant
Ω(E)→thermodynamic probability

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

114
4.3 POSTULATE OF EQUAL A PRIORI PROBABILITY:
Statement:
 The probability of finding the phase point for a given system in any one region of
phase space is identical with that for any other region of equal volume.
 The necessity of this postulate arises due to incompleteness of our knowledge
concerning the system of interest.
 This postulate appears to be reasonable in character with the principles of
statistical mechanics derived from Liouville‟s theorem.
 According to the principle of conservation of density, the density of a group of
phase points remains constant.
 At any time the phase points are distributed uniformly in the phase space.
 There is no crowding of phase points in any particular region of phase space.
 Any arbitrary element of volume in the phase space bounded by a moving surface
and containing a definite number of phase points does not change with time.
 The property of no crowding of phase points in any particular region of phase
space and the constancy of volume element of phase space with time indicate the
validity of the postulate.
 That is the probability of finding a phase point in any particular region of phase
space is directly proportional to the volume of that region.
 The postulate replaces the postulate of equal priori probability when different
volumes in the phase space are considered.
4.4 CONTACT BETWEEN STATISTICS AND THERMODYNAMICS :
(BOLTZMANN RELATION BETWEEN ENTROPY AND PROBABILITY)
 Boltzmann used the idea that the probability of the system in equilibrium state is
maximum.
 Thus in equilibrium state both the entropy and thermodynamical probability have
their maximum values.
 Boltzmann concluded that the entropy „S‟ is a function of thermodynamic
probability Ω.
ie, 𝑆 = 𝑓(𝛺) ….(1)
 Consider two independent systems A and B having entropies S1 and S2 and
thermodynamic probabilities Ω1 and Ω 2.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

115
  Entropy is an additive quantity and hence the entropy of systems together must be
equal to the sum of their individual entropies.
𝑆 = 𝑆1 + 𝑆2 ….(2)
 The probability Ω of finding both systems will be the product of the two
probabilities Ω 1 and Ω 2.
ie 𝛺 = 𝛺1 𝛺2 ….(3)
Substituting equations (2) and (3) in equation (1) we get,
𝑆 = 𝑓(𝛺) = 𝑓(𝛺1 𝛺2 ) ….(4)
𝑆 = 𝑆1 + 𝑆2
𝑓 𝛺1 𝛺2 = 𝑓 𝛺1 𝛺2 …..(5)
 Differentiating with respect to Ω1 we get,
𝛺2 𝑓 ′ 𝛺1 𝛺2 𝑓 ′ 𝛺1 ….(6)
 Differentiating with respect to Ω2 we get,
𝛺1 𝑓 ′ 𝛺1 𝛺2 = 𝑓 ′ 𝛺2 … . (7)
 Divide equation (7)/(6), we get
𝛺1
= 𝑓 ′(𝛺2 )/ 𝑓 ′(𝛺1 )
𝛺2
𝛺1 𝑓 ′(𝛺1 ) = 𝛺2 𝑓 ′(𝛺2 )
𝛺 𝑓 ′(𝛺) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑘
𝑘
𝑓 ′(𝛺1 ) =
𝛺
Integrating, 𝑓(𝛺) = 𝑘 𝑙𝑜𝑔 𝛺 + 𝑐
𝑆 = 𝑘 𝑙𝑜𝑔 𝛺 + 𝑐 .…(8)
 For a thermo dynamical system at absolute zero Ω=1 and S=0 so that c=0.
𝑆 = 𝑘 𝑙𝑜𝑔 𝛺.
 This gives the Boltzmann‟s relation between entropy and probability.
(a) Identification of constant ‘k’:
 Consider the expansion of one mole of an ideal gas at pressure p1 and volume V1
into an evacuated chamber of volume V2.
 The find pressure is p2 and the final volume is V1+ V2.
 The probability of finding one molecule in the first container with volume V1 is,
𝑉1
𝑉1 + 𝑉2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

116
  There are N molecules and hence the probability of finding one mole of the gas in
the first container with volume V1 is,
𝑉1 𝑁
𝛺1 = ….(10)
𝑉1 +𝑉2

 The probability of finding one mole of the gas in the container has volume V1+V2
is
𝑉1 +𝑉2 𝑁 𝑁
𝛺2 = = 1 .…(11)
𝑉1 +𝑉2

From Boltzmann relation

∆𝑆 = 𝑆2 – 𝑆1
= 𝑘 𝑙𝑜𝑔 𝛺2 – 𝑘 𝑙𝑜𝑔 𝛺1
𝛺
= 𝑘 𝑙𝑜𝑔 (𝛺2 )
1

𝑁
1
= 𝑘 𝑙𝑜𝑔 𝑉1
𝑉 1 +𝑉 2

𝑉1 +𝑉2 𝑁
∆𝑆 = 𝑘 𝑙𝑜𝑔 𝑉1

𝑉1 +𝑉2 𝑁𝑘
= 𝑙𝑜𝑔 ….(12)
𝑉1

 The change in entropy when the gas changes from one state with volume V1 and
temperature T1 to another state with volume V2 and temperature T2 is given by,
𝑇2 𝑉1 + 𝑉2
∆𝑆 = 𝐶𝑣 𝑙𝑜𝑔 + 𝑅 𝑙𝑜𝑔 [ ]
𝑇1 𝑉1
𝑇
For isothermal change 𝑇2 = 𝑇1 and hence 𝐶𝑣 𝑙𝑜𝑔 𝑇2 = 0
1

𝑉1 + 𝑉2
∆𝑆 = 𝑅 𝑙𝑜𝑔 [ ]
𝑉1
𝑉1 +𝑉2 𝑅
= 𝑙𝑜𝑔 .…(13)
𝑉1

Comparing equation (12) and (13), we get

𝑁𝑘 = 𝑅
𝑘 = 𝑅/𝑁
= 1.03 × 10−23 𝐽/𝐾 = Boltzmann‟s constant
4.5 ENSEMBLE THEORY: CONCEPT OF ENSEMBLES:
 A system is defined as a collection of identical particles.
 An ensemble is defined as a collection of macroscopically identical, but
essentially independent systems.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
117
  Macroscopically identical means each system satisfies the same macroscopic
conditions ex: volume, energy, pressure etc.
 Independent systems mean the systems are non interacting.
 There are three most commonly used ensembles namely,
(i) micro canonical ensemble
(ii) canonical ensemble
(iii) grand canonical ensemble.
Micro canonical ensemble:
 Collection of large number of
essentially independent systems with same energy
E, volume V, and the number of particles N.
 All the particles are identical.
 The individual systems are separated
by rigid , impermeable and well insulated walls.
 No exchange of heat energy as well
as the number of particles between the systems
takes place.

Canonical ensemble:

 Collection of large number of

essentially independent systems with same
temperature T, volume V and the number of
particles N.
 All the particles are identical.
 The individual systems are separated by
rigid, impermeable but conducting walls.
 Exchange of heat energy between the
system takes place. But not the particles.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

118
Grand canonical ensemble:

 Collection of large number of

essentially independent systems with same
temperature T, volume V, and the chemical
potential μ.
 All the particles are identical.
 The individual systems are separated
by rigid, permeable and conducting walls.
 Exchange of heat energy as well as the
particles between the system takes place.
 The state of any ensemble can be completely specified by a large number of phase
points in the phase space called dust cloud.
 Thus the behavior of an ensemble can be represented by a large number of
trajectories (or) phase lines (or) streaming motion of the dust cloud.
4.6 PHASE SPACE:
 The instantaneous position of a single particle is described by three independent
co-ordinates 𝑥, 𝑦 𝑎𝑛𝑑 𝑧.
 The instantaneous motion of a particle is described by momentum co- ordinates
𝑝𝑥 , 𝑝𝑦 , 𝑝𝑧 .
 Thus the state of a single particle is completely specified by position co-ordinates
𝑥, 𝑦, 𝑧 and momentum co-ordinates 𝑝𝑥 , 𝑝𝑦 , 𝑝𝑧
 We may imagine a six dimensional space with volume 𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝𝑥 𝑑𝑝𝑦 𝑑𝑝𝑧 .
 The position of a point particle in this space can be described by a set of 6 co-
ordinates 𝑥 𝑦 𝑧 𝑝𝑥 , 𝑝𝑦 , 𝑝𝑧 .
 This 6 dimensional space for a single particle is damped as phase space.(µ=space)
 If the system contains a large number of particles such that f independent position
co-ordinates 𝑞1 , 𝑞2 , … … … . , 𝑞𝑓 and f momentum co-ordinates 𝑝1 , 𝑝2 , … … . , 𝑝𝑓 ,
then 2f combined position, momentum co-ordinates may be allowed to define 2f-
dimensional space called phase space(Γ-space)
 The Γ-space is considered to be a conceptual Euclidean space having 2f
rectangular axes and an element of volume represented by
𝑑𝑞1 , 𝑑𝑞2 , … … , 𝑑𝑞𝑓 𝑑𝑝1 , 𝑑𝑝2 , … … , 𝑑𝑝𝑓
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
119
  The instantaneous state of a particle in the phase-space is represented by a point
known as phase point (or) representative point.
 The number of phase points per unit volume is known as phase density.
4.7 COUNTING THE NUMBER OF MICROSTATES IN THE ENERGY RANGE
ε TO ε+dε:
 For a single particle we have six dimensional phase space.
 Three position co-ordinates (𝑥, 𝑦, 𝑧) and three momentum
co-ordinates (𝑝𝑥 , 𝑝𝑦 , 𝑝𝑧 ) specify the microstate of a particle
in the phase space.
 An element of volume in phase space is,
𝛿𝑥 𝛿𝑦 𝛿𝑧 𝛿𝑝 𝑥 𝛿𝑝 𝑦 𝛿𝑝 𝑧 = 𝑕3

 The total volume of phase space is 𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝 𝑥 𝑑𝑝𝑦 𝑑𝑝 𝑧

 We have 𝑑𝑥 𝑑𝑦 𝑑𝑧 = 𝑉
 So the volume in phase space = 𝑉 𝑑𝑝 𝑥 𝑑𝑝 𝑦 𝑑𝑝 𝑧

 Volume of momentum space containing momentum between p and 𝑝 + 𝑑𝑝 will

be given by the volume of a spherical cell with radius p and thickness dp.
 Therefore,
𝑑𝑝 𝑥 𝑑𝑝 𝑦 𝑑𝑝 𝑧 = 4𝜋𝑝2 𝑑𝑝

𝑝2
𝜀= → 𝑝2 = 2𝑚𝜀
2𝑚
2𝑝𝑑𝑝 = 2𝑚𝑑𝜀
𝑚
𝑑𝑝 = 𝑑𝜀
𝜑
𝑚 𝑚
= 𝑑𝜀 = 𝑑𝜀
2𝑚𝜀 2𝜀

 Now volume of phase space

= 𝑉. 4𝜋𝑝2 𝑑𝑝
𝑚
= 𝑉 × 4𝜋(2𝑚𝜀) 𝑑𝜀
2𝜀
= 4𝜋𝑉 2 𝑚3/2 𝜀 1/2 𝑑𝜀
 The number of cells within the phase space.
4𝜋𝑉 2
ie, 𝛺 𝜀 𝑑 𝜀 = 𝑚3/2 𝜀 1/2 𝑑𝜀
𝑕3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

120
  For a single particle the number of accessible microstates will be equal to the
number of cells in phase space.
 Hence the number of microstates in this energy range ε to ε+dε is given by,
4𝜋𝑉 2
𝛺 𝜀 𝑑𝜀 = 𝑚3/2 𝜀1/2 𝑑𝜀
𝑕3

4.8 TIME AND ENSEMBLE AVERAGE:

 An ensemble consists of a large number of independent systems.
 It may be represented by a particular point in phase space.
 A gas containing a large number of molecules forming a system.
 The gas molecules move constantly and hence they change the position and
momentum with time.
 The entire gas shows a time independent property (eg: temperature , energy etc.)
which may be considered as the average of the specified property of the
constituent gas molecules .
 Here, we discuss this type of average property of the ensemble.
 Let the state of the ensemble changes with time.
 Let u be the property of the ensemble.
 u takes values 𝑢1 , 𝑢2 , … … . . , 𝑢𝑚 having probabilities 𝑃1 , 𝑃2 , … … . , 𝑃𝑚 .
𝑃1 𝑢1 + 𝑃2 𝑢2 + ⋯ … … + 𝑃𝑖 𝑢𝑖 + ⋯ … … 𝑃𝑚 𝑢𝑚
𝑢=
𝑃1 + 𝑃2 + ⋯ … … + 𝑃𝑖 + ⋯ … … … 𝑃𝑚
𝑚
𝑖=1 𝑃 𝑖 𝑢 𝑖
= 𝑚 ….(1)
𝑖=1 𝑃 𝑖

 The sum of the probabilities of the all possible state must be equal to one.
𝑚
ie, 𝑃1 + 𝑃2 + ⋯ … . . +𝑃𝑖 + ⋯ … … . +𝑃𝑚 = 𝑖=1 𝑃𝑖 =1 ….(2)
 This called normalization condition.
 Now equation (1) becomes
𝑚
𝑢= 𝑖=1 𝑃𝑖 𝑢𝑖

 If the ensemble consists of N systems, u can be expressed as the function of all

position and momentum co-ordinates of the systems.
 If the probability distribution function is continuous , then equation(1) can be
expressed as,
𝑢 𝑞,𝑝 𝑃 𝑞 ,𝑝 𝑑𝛤
𝑢=
𝑃 𝑞,𝑝 𝑑𝛤

𝑑𝛤 = 𝑑𝑞1 , 𝑑𝑞2 , … … … , 𝑑𝑞𝑓 𝑑𝑝1 , 𝑑𝑝2 , … . , 𝑑𝑝𝑓

 According to normalization condition
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
121
 𝑃 𝑞, 𝑝 𝑑𝛤 = 1
 Hence 𝑢= 𝑢 𝑞, 𝑝 𝑃 𝑞, 𝑝 𝑑 𝛤
 This gives the ensemble average.
4.9 LIOUVILLE’S THEOREM:

 Liouville‟s theorem gives information about the rate of change of phase density in
the phase space. The theorem may be stated in two parts.
 The rate of change of density of phase points in the neighborhood of a moving
phase point in the Γ space is zero. This part represents the principle of
conservation of density in the phase space.

𝑑𝜌/𝑑𝑡 = 0 ….(1)

 Any arbitrary element of volume or extension in phase in the Γ space bounded by

a moving surface and containing a number of phase points does not change with
time. This part represents the principle of conservation of extension in the phase
space.

𝑑 𝑑 𝑓
𝛿𝛤 = 𝑑𝑡 𝑖 𝑑𝑞𝑖 𝑑𝑝𝑖 = 0 ….(2)
𝑑𝑡

(i) The principle of conservation of density in the phase space:

 Consider any arbitrary hyper volume
𝛿𝛤 = 𝛿𝑞1 𝛿𝑞2 … … . . 𝛿𝑞𝑓 𝛿𝑝1 𝛿𝑝2 … … . . 𝛿𝑝𝑓
in the phase space located between
𝑞1 and 𝑞1 + 𝛿𝑞1 … … … . . 𝑞𝑓 and 𝑞𝑓 + 𝛿𝑞𝑓 ,
𝑝1 and 𝑝1 + 𝛿𝑝1 , … … … . 𝑝𝑓 and 𝑝𝑓 + 𝛿𝑝𝑓 . The
number of phase points in this volume element
changes with time due to the motion of phase
points.
 If ρ is the density of phase points, the number
of phase points in this volume element at any
instant t is ,

𝛿𝑁 = 𝜌. 𝛿𝛤 = 𝜌𝛿𝑞1 𝛿𝑞2 … … . 𝛿𝑞𝑓 𝛿𝑝1 𝛿𝑝2 … … … 𝛿𝑝𝑓 ….(3)

 The change in number of phase points in volume element per unit time,

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

122
 𝑑 𝛿𝑁 𝑑
= 𝑑𝑡 (𝜌. 𝛿𝛤) = 𝜌𝛿𝛤 = 𝜌 𝛿𝑞1 𝛿𝑞2 … … . 𝛿𝑞𝑓 𝛿𝑝1 𝛿𝑝2 … … … 𝛿𝑝𝑓 …(4)
𝑑𝑡

 This change in the number of phase points in the given hyper volume is due to the
difference between the number of phase points entering the hyper volume through
any face and the number of those leaving the opposite face per second.
 Consider two faces of hyper volume with co-ordinates 𝑞1 and 𝑞1 + 𝛿𝑞1 . If 𝑞1 is
the component of velocity of phase point at 𝑞1 , 𝑞2 , … … . 𝑞𝑓 , 𝑝1 , 𝑝2 , … … … 𝑝𝑓 ,then
the number of phase points entering the first face AD per second
= 𝜌𝑞1 𝛿𝑞2 … … . . 𝛿𝑞𝑓 𝛿𝑝1 … … . 𝛿𝑝𝑓 ….(5)
 As density ρ changes with change in position and momentum co-ordinates and at
the opposite face BC the co-ordinate q1 changes to 𝑞1 + 𝛿𝑞1 and the density ρ
𝜕𝜌
changes to (𝜌 + 𝜕𝑞 𝛿𝑞1 ) at the face BC. The velocity 𝑞 1 changes to (𝑞1 +
1

𝜕𝑞 1
𝛿𝑞1 ). Therefore the number of phase points leaving the opposite face BC at
𝜕𝑞 1

q1+δq1 per second.

𝜕𝜌 𝜕𝑞
= (𝜌 + 𝜕𝑞 𝛿𝑞1 ) (𝑞1 + 𝜕𝑞 1 𝛿𝑞1 ) 𝛿𝑞2 … … 𝛿𝑞𝑓 𝛿𝑝1 , … … … 𝛿𝑝𝑓
1 1

 Neglecting higher order differentials, we get

𝜕𝑞 1 𝜕𝜌
= 𝜌𝑞1 + (𝜌 + 𝑞1 𝜕𝑞 )𝛿𝑞1 𝛿𝑞2 … … 𝛿𝑞𝑓 𝛿𝑝1 , … … … 𝛿𝑝𝑓 ....(6)
𝜕𝑞 1 1

 Subtracting (6) from (5) we get the expression for change in the number of phase
points per second corresponding to q1.

𝜕𝑞 1 𝜕𝜌
= −(𝜌 + 𝑞1 𝜕𝑞 )𝛿𝑞1 𝛿𝑞2 … … 𝛿𝑞𝑓 𝛿𝑝1 , … … … 𝛿𝑝𝑓 ….(7)
𝜕𝑞 1 1

 Similarly, the expression for the change into the number of phase points per
second corresponding to p1 is

𝜕𝑝 1 𝜕𝜌
= −(𝜌 + 𝑝1 𝜕𝑝 )𝛿𝑞1 𝛿𝑞2 … … 𝛿𝑞𝑓 𝛿𝑝1 , … … … 𝛿𝑝𝑓 ….(8)
𝜕𝑝 1 1

 Since the change in number of phase points per second corresponding to all
position and momentum coordinates are like equation (7) and (8), then they are
summed up.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

123
  The net increase in the number of phase points in the given hyper volume per
second is given by,

𝑑(𝛿𝑁) 𝑓 𝜕𝑞 𝑖 𝜕𝑝 𝜕𝜌 𝜕𝜌
=− 𝑖=1 𝜌 + 𝜕𝑝 𝑖 + ( 𝑞1 𝜕𝑞 + 𝑝𝑖 ) 𝛿𝑞1 … 𝛿𝑞𝑓 𝛿𝑝1 , … 𝛿𝑝𝑓 .....(9)
𝑑𝑡 𝜕𝑞 𝑖 𝑖 𝑖 𝜕𝑝 𝑖

 using equation (4)we get,

𝜕𝜌 𝑓 𝜕𝑞 𝑖 𝜕𝑝 𝜕𝜌 𝜕𝜌
=− 𝑖=1 𝜌 + 𝜕𝑝 𝑖 + ( 𝑞1 𝜕𝑞 + 𝑝𝑖 ) ….(10)
𝜕𝑡 𝜕𝑞 𝑖 𝑖 𝑖 𝜕𝑝 𝑖

 From canonical equation,

𝜕𝑞𝑖 𝜕2𝐻 𝜕𝑝𝑖 −𝜕 2 𝐻

= 𝑎𝑛𝑑 =
𝜕𝑞𝑖 𝜕𝑞𝑖 𝜕𝑝𝑖 𝜕𝑝𝑖 𝜕𝑝𝑖 𝜕𝑞𝑖

 Since the order of differentiation is immaterial i.e,

𝜕2𝐻 𝜕2𝐻
=
𝜕𝑞𝑖 𝜕𝑝𝑖 𝜕𝑝𝑖 𝜕𝑞𝑖

𝜕𝑞 𝑖 𝜕𝑝
We get = − 𝜕𝑝 𝑖
𝜕𝑞 𝑖 𝑖

𝑓 𝜕𝑞 𝑖 𝜕𝑝
𝑖=1 𝜕𝑞 𝑖 + 𝜕𝑝 𝑖 = 0 ….(11)
𝑖

 Now equation (10) becomes

𝜕𝜌 𝑓 𝜕𝜌 𝜕𝜌
=− 𝑖=1 𝑞1 𝜕𝑞 + 𝑝𝑖
𝜕𝑡 𝑞,𝑝 𝑖 𝜕𝑝 𝑖

𝜕𝜌 𝑓 𝜕𝜌 𝜕𝜌
+ 𝑖=1 𝑞1 + 𝑝𝑖 = 0 ….(12)
𝜕𝑡 𝑞,𝑝 𝜕𝑞 𝑖 𝜕𝑝 𝑖

 This equation represents Liouville‟s theorem.

𝑑𝜌
(𝑞 , … . , 𝑞𝑓 , 𝑝1 , … … . , 𝑝𝑓 , 𝑡) = 0
𝑑𝑡 1

𝑑𝜌
ie, 𝑑𝑡 = 0 .…(13)

 This expression represents the principle of conservation of density in phase space.

(ii) The principle of conservation of extension in phase space:

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

124
  Consider a very small region of hyper volume δΓ in the Γ space, so that the
density of phase points ρ can be taken as uniform throughout the hyper volume.

The number of phase points in this hyper volume, 𝛿𝑁 = 𝜌. 𝛿𝛤

𝑑 𝑑
(𝛿𝑁) = (𝜌. 𝛿𝛤)
𝑑𝑡 𝑑𝑡

𝑑𝜌 𝑑(𝛿𝛤 )
= 𝛿𝛤 + 𝜌 ....(14)
𝑑𝑡 𝑑𝑡

 As each phase point represents a definite system and systems can neither be
created nor destroyed, the number of phase points δN must remain fixed.

𝑑
i.e 𝛿𝑁 = 0
𝑑𝑡

𝑑𝜌 𝑑(𝛿𝛤)
𝛿𝛤 + 𝜌 =0
𝑑𝑡 𝑑𝑡

𝑑𝜌
 from equation (14) =0
𝑑𝑡

𝑑(𝛿𝛤 )
𝜌 =0
𝑑𝑡

𝑑(𝛿𝛤)
=0
𝑑𝑡

𝛿𝛤 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ….(15)

 This expression represents the principle of conservation of extension in the phase

space.

STATIONARY ENSEMBLE:
4.10 MICRO CANONICAL ENSEMBLE (ISOLATED SYSTEM):
 An ensemble in which each system has the same fixed energy as well as the same
number of particles is called micro canonical ensemble.
 In this ensemble, density ρ, for a closed isolated thermo dynamical system is a
function of energy and we take

ρ(E) = constant between the energy shells E and E+δE of phase space.

=0 outside the region of phase space.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
125
  We call this region in which ρ(E) = constant as accessible region dΓ of phase
space.
 The above choice of ρ(E) being constant in dΓ and zero outside dΓ indicating
accessibility can be justified as follows:
 Suppose we consider a gas of volume V, separated into smaller volumes V1 and
V2 by a thin perfectly conducting wall of negligible heat capacity through which
the particles of the gas can diffuse very slowly, hut through which energy can be
exchange freely.
 Let at a particular instant, we determine the pressure in the two volumes, and let
at this instant n out of total n′ particles be in volume V1. The particles in volume
V2 will be then (n‟-n). Now,
 (i) For an experiment of short duration, it would not be appropriate to take all
particles could be found with equal probability anywhere within the volume V
and therefore accessible region is the region of phase space in which all the first n
particles are in V1 and remaining (n‟-n) are in V2.
 (ii) For an experiment of long duration in which a considerable amount of
diffusion could occur, the whole of phase space is accessible.
 Thus for short duration experiments dΓ is accessible and it is inappropriate to
include in the ensemble, the assembly lying outside this region dΓ, which means
ρ(E) = constant for dΓ while zero outside dΓ.
 In general, all accessible regions of phase space are given equal weightage in
averaging over a microcanonical ensemble. This is known as the „Principle of
equal a priori probabilities‟.

(i) Partition Function:

 Consider an assembly of ideal gas obeying classical statistics.
 Let the distribution of gas molecules be such that 𝑛𝑖 molecules occupy the 𝑖 𝑡𝑕
state with energy between 𝜀𝑖 and 𝜀𝑖 + 𝑑𝜀𝑖
 Let 𝑔𝑖 be the degeneracy of the 𝑖 𝑡𝑕 state.
 According to M-B distribution law,
𝑛𝑖 = 𝑔𝑖 𝑒 −𝛼 𝑒 −𝛽𝜀𝑖
= 𝑔𝑖 𝑒 −𝛼 𝑒 −𝜀 𝑖 /𝑘𝑇 [β=1/kT]
𝑒 −𝛼 = 𝐴
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
126
 Then 𝑛𝑖 = 𝐴𝑔𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇
 Let the total number of gas molecules be N.
𝑁= 𝑖 𝑛𝑖 .
= 𝑖 𝐴 𝑔𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇 .

=𝐴 𝑖 𝑔𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇
𝑁
= 𝑖 𝑔𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇
𝐴

𝑍= 𝑖 𝑔𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇
 Z is known as partition function and Z indicates how the gas molecules of an
assembly are distributed (or) partitioned among the various energy levels.
 If the energy of the 𝑖 𝑡𝑕 level is 𝜀𝑖 then the weight of an individual level is unity.
ie, 𝑔𝑖 = 1
𝑍= 𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇
 Here the energy term may contain the rotational, vibrational and electronic
components in addition to translational component.
 „Z‟ can be used for calculating the various thermodynamic properties of
ensembles.
 In classical treatment the energy distribution is continuous.
 The number of energy levels of the momentum interval p and 𝑝 + 𝑑𝑝 is given by,
𝑉 4𝜋𝑝2 𝑑𝑝
𝑔(𝑝)𝑑𝑝 =
𝑕3
𝑝2 = 2𝑚𝜀
2𝑝𝑑𝑝 = 2𝑚𝑑𝜀.
𝑚
𝑑𝑝 = 𝑑𝜀.
𝑝

𝑚 𝑚
= 𝑑𝜀 = 𝑑𝜀
2𝑚𝜀 2𝜀

 Now the number of energy levels in the energy range 𝜀 and 𝜀 + 𝑑𝜀 is obtained as,
𝑉 𝑚
𝑔 𝜀 𝑑𝜀 = 4𝜋 2𝑚𝜀 𝑑𝜀
𝑕3 2𝜀
2𝜋𝑉 3/2
= 2𝑚 𝜀 1/2 𝑑𝜀
𝑕3

𝑍= 𝑖 𝑔𝑖 𝑒 −𝜀 𝑖 /𝑘𝑇 .
∞
= 0
𝑔(𝜀) 𝑒 −𝜀/𝑘𝑇 𝑑𝜀
∞ 2𝜋𝑉 3/2
= 0 𝑕3
2𝑚 𝜀1/2 𝑒 −𝜀/𝑘𝑇 𝑑𝜀

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

127
 2𝜋𝑉 3/2 ∞ 1/2
= 2𝑚 𝜀 𝑒 −𝜀/𝑘𝑇 𝑑𝜀
𝑕3 0

2𝜋𝑉 3/2 1 𝜋
= 2𝑚
𝑕3 2 𝛽3

2𝜋𝑉 3/2 1
= 2𝑚 𝜋(𝑘𝑇)3
𝑕3 2
𝑉
𝑍= (2𝜋𝑚𝐾𝑇)3/2
𝑕3
 This is the translational partition function for a gas molecule.
4.11 CLASSICAL IDEAL GAS USING MICRO CANONICAL ENSEMBLE:
 Consider a micro canonical ensemble of a perfect gas.
 Let there be n point particles with mass m confined in a volume V with total
energy u within the energy range 𝛿𝑢.
 The corresponding volume

∆𝛤 = 𝑑 𝑞1 … … … . 𝑑𝑞3𝑛 𝑑 𝑝1 … … … . 𝑑𝑝3𝑛

𝑑 𝑞1 … … … 𝑑𝑞3𝑛 = 𝑉 𝑛 .
Hence ∆𝛤 = 𝑉 𝑛 𝑑 𝑞1 … … … … 𝑑𝑞3𝑛
 The momentum space integral is to be evaluated subject to the constraint of the
ensemble
𝑢 − 𝛿𝑢 ≤ 𝑢𝑟 ≤ 𝑢.
𝑛 2
𝑢𝑟 = 𝑖=1 𝑝𝑖 /2𝑚 .
1 𝑛 2
𝑢 − 𝛿𝑢 ≤ 2𝑚 𝑖=1 𝑝𝑖 ≤ 𝑢.

 The accessible volume in momentum space is the volume of a spherical shell of

𝑚 1
radius (2𝑚𝑢)1/2 and thickness (2𝑢 )2 𝛿𝑢.

 The volume of three dimensional sphere of radius „R‟ is,

4 𝜋 3/2 𝜋 3/2
𝑉3 (𝑅) = 3 𝜋𝑅 3 = 3 𝑅3 = 3 𝑅 3 = 𝐶3 𝑅 3
𝛤( +1) !
2 2

𝜋 𝑓/2
𝑉𝑓 (𝑅) = 𝑅 𝑓 = 𝑐𝑓 𝑅 3
𝑓/2 !

𝜋 𝑓/2
where 𝐶𝑓 = 𝑓/2 !

 Therefore for 3n dimensional hyper-sphere of radius (2𝑚𝑢)1/2 , the volume is,

𝜋 3𝑛 /2 3𝑛/2
𝑉3𝑛 𝑅 = 2𝑚𝑢
3𝑛 /2 !

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

128
  The volume coupled between hyper spheres of radii (2𝑚𝑢)1/2 to [2𝑚 𝑢 −
𝛿𝑢]1/2 is
𝜋 3𝑛 /2
𝑑𝑝1 … … … … 𝑑𝑝3𝑛 = [(2𝑚𝑢)3𝑛/2 – {2𝑚(𝑢 − 𝛿𝑢)}3𝑛/2 ]
3𝑛/2 !

𝜋 3𝑛 /2 𝛿𝑢 3𝑛/2
= (2𝑚𝑢)3𝑛/2 [1 − (1 − ) ]
3𝑛/2 ! 𝑢

𝜋 3𝑛 /2 3𝑛 𝛿𝑢
= (2𝑚𝑢)3𝑛/2 [1 − 𝑒𝑥𝑝(− . )]
3𝑛/2 ! 2 𝑢
3𝑛 𝛿𝑢
 For a macroscopic system 3n=1023 ; >>u.
2 𝑢

 And hence we can drop the exponential term.

𝜋 3𝑛 /2
𝑑𝑝1 … … … … … … 𝑑𝑝𝑛 = (2𝑚𝑢)3𝑛/2
3𝑛/2 !

∆𝛤 = 𝑉 𝑛 𝑑𝑝1 … … … … … … . . 𝑑𝑝3𝑛
𝜋 3𝑛 /2
= 𝑉𝑛 (2𝑚𝑢)3𝑛/2
3𝑛/2 !

 According to classical statistical mechanics , the entropy ζ in statistical

equilibrium is given by,
𝜍 = 𝑙𝑜𝑔 ∆𝛤

𝑛
𝜋 3𝑛 /2
= 𝑙𝑜𝑔 [𝑉 (2𝑚𝑢)3𝑛/2 ]
3𝑛/2 !
= 𝑛 𝑙𝑜𝑔[𝑉 𝜋 3/2 2𝑚𝑢 3/2
] − 𝑙𝑜𝑔 (3𝑛/2)!
= 𝑛 𝑙𝑜𝑔[𝑉 𝜋 3/2 2𝑚𝑢 3/2
] − (3𝑛/2) 𝑙𝑜𝑔 (3𝑛/2) + 3𝑛/2
= 𝑛 𝑙𝑜𝑔 [𝑉 𝜋 3/2 2𝑚𝑢 3/2
] − 𝑛 𝑙𝑜𝑔 3𝑛/2 3/2
+ 3𝑛/2
𝑉 𝜋 3/2 2𝑚𝑢 3/2
= 𝑛 𝑙𝑜𝑔 + 3𝑛/2
3𝑛 /2 3/2

4𝜋𝑚 3/2 𝑢 3/2 3𝑛

𝜍 = 𝑛 𝑙𝑜𝑔 [𝑉 ]+
3 𝑛 2

 We know that the entropy should not depend on the unit of hyper volume ∆Γ. To
make it dimensionless we divide it by 𝑕3𝑛 .

𝜍 = 𝑙𝑜𝑔[∆𝛤/𝑕3𝑛 ]

4𝜋𝑚 3/2 𝑢 3/2

3 𝑛 3𝑛
= 𝑛 𝑙𝑜𝑔 𝑉 +
𝑕3 2

 The above equation does not satisfy the additive property and hence to satisfy the
additive property we must divide by n!

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

129
 ∆𝛤
𝜍 = 𝑙𝑜𝑔 𝑕 3𝑛 𝑛!

4𝜋𝑚 3/2 𝑢 3/2

3 𝑛 3𝑛
= 𝑛 𝑙𝑜𝑔 𝑉 + − 𝑙𝑜𝑔𝑛!
𝑕3 2

4𝜋𝑚 3/2 𝑢 3/2

3 𝑛 3𝑛
= 𝑛 𝑙𝑜𝑔 𝑉 + – 𝑛 𝑙𝑜𝑔 𝑛 + 𝑛
𝑕3 2

𝑉 4𝜋𝑚 3/2 𝑢 3/2

𝑛 3 𝑛 5
𝜍 = 𝑛 𝑙𝑜𝑔 +2n
𝑕3

 This expression satisfies the additive property because instead of V and u we have
V/n and u/n.
 We shall now establish the connection of statistical quantities with corresponding
thermodynamic quantities.
(a) Internal energy(U):

By the definition of statistical temperature 𝜏,

1 𝜕𝜍
=
𝜏 𝜕𝑢 𝑇,𝑛

𝑉 4𝜋𝑚 3/2 𝑢 3/2

𝜕 𝑛 3 𝑛 5
= 𝜕𝑢 𝑛𝑙𝑜𝑔 +2n
𝑕3
𝑇,𝑛

𝜕 4𝜋𝑚 3/2 𝑢 3/2 𝜕 5

= 𝑛 𝑙𝑜𝑔 𝑣 – 𝑛 𝑙𝑜𝑔 𝑛 + 𝑛 𝑙𝑜𝑔 + 𝑛 𝑙𝑜𝑔 − 𝑛𝑙𝑜𝑔 𝑕3 + 𝜕𝑢 (2 𝑛)
𝜕𝑢 3 𝑛

𝜕 4𝜋𝑚 3/2 3 3
= [𝑛 𝑙𝑜𝑔 𝑣– 𝑛 𝑙𝑜𝑔 𝑛 + 𝑛 𝑙𝑜𝑔 + 2 𝑛 𝑙𝑜𝑔 𝑢 − 2 𝑛 𝑙𝑜𝑔 𝑛 – 𝑛𝑙𝑜𝑔 𝑕3 ] +
𝜕𝑢 3
𝜕 5
( 𝑛)
𝜕𝑢 2

𝜕 3
= ( 𝑛 𝑙𝑜𝑔 𝑢)
𝜕𝑢 2

1 3 1
= 𝑛
𝜏 2 𝑢

3 3
𝑢 = 2 𝑛𝜏 (or) 𝑢 = 2 𝑛 𝑘𝑇

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

130
  Which is the well known result for the internal energy of a perfect mono atomic
gas.
(b) Relation between 𝝉 and T:

The statistical temperature

𝜏 = 𝑘 × 𝑡𝑕𝑒𝑟𝑚𝑜𝑑𝑦𝑛𝑎𝑚𝑖𝑐 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒.

𝜏 = 𝑘𝑇

(c) Relation between 𝝉 and p:

𝑝 𝜕𝜍
We have =
𝜏 𝜕𝑉 𝑛.𝑢

3
𝜕 4𝜋𝑚 2 3 3
= 𝑛 𝑙𝑜𝑔 𝑣 − 𝑛 𝑙𝑜𝑔 𝑛 + 𝑛 𝑙𝑜𝑔 + 2 𝑛 𝑙𝑜𝑔 𝑢 – 2 𝑛 𝑙𝑜𝑔 𝑛 −
𝜕𝑉 3

𝑛 𝑙𝑜𝑔 𝑕3+ 𝜕𝜕𝑉(52𝑛) 𝑛,𝑢

𝜕
= [ 𝑛 𝑙𝑜𝑔 𝑉]
𝜕𝑉

= 𝑛/𝑉

𝑃𝑉 = 𝑛 𝜏 (or) 𝑃𝑉 = 𝑛𝑘𝑇

 Which is well known ideal gas equation for a perfect mono atomic gas.

(d) Thermodynamic entropy (S): (Sackur - Tetrode equation)

The relation between thermodynamic entropy and statistical entropy is given by,

𝑆 = 𝑘𝜍

𝑉 4𝜋𝑚 3/2 𝑢 3/2

𝑛 3 𝑛 5
= 𝑛𝑘 𝑙𝑜𝑔 + 𝑛𝑘
𝑕3 2

𝑉 4𝜋𝑚 3/2 3 5
= 𝑛𝑘 𝑙𝑜𝑔 (2 𝑘𝑇)3/2 + 2 𝑛𝑘
𝑛𝑕 3 3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

131
 3
Since 𝑢 = 𝑛𝑘𝑇
2

𝑉 3/2
5
= 𝑛𝑘 𝑙𝑜𝑔 (2𝜋𝑚𝑘𝑇) + 𝑛𝑘
𝑛𝑕3 2

3/2
𝑉 2𝜋𝑚𝑘𝑇 5
= 𝑛𝑘 𝑙𝑜𝑔 + 𝑛𝑘
𝑛 𝑕2 2

𝑉 2𝜋𝑚𝑘𝑇 3/2
= 𝑛𝑘 𝑙𝑜𝑔 𝑒 5/2
𝑛 𝑕2

 This is the famous Sackur - Tetrode equation for the entropy of a perfect gas.
This formula is valid for the mono atomic gas of atoms with zero total angular
momentum.
 The thermal de-broglie wavelength associated with a molecule may be defined as,

𝜆 = 𝑕/𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑡𝑕𝑒𝑟𝑚𝑎𝑙 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑜𝑓 𝑎 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒.

𝜆 = 𝑕/(2𝜋𝑚𝑘𝑇)1/2

𝜆3 = 𝑕3 /(2𝜋𝑚𝑘𝑇)3/2

1
= (2𝜋𝑚𝑘𝑇/𝑕2 )3/2
𝜆3

𝑉 1 5
Now 𝜍 = 𝑛 𝑙𝑜𝑔 +2𝑛
𝑛 𝜆3

𝑉 1 5
𝑆 = 𝑛𝑘 𝑙𝑜𝑔 + 2 𝑛𝑘
𝑛 𝜆3

 Thus the entropy of a perfect gas is determined essentially by the ratio of the
volume per particle to the volume λ3 associated with de-Broglie wavelength.
(e) Chemical potential of a perfect gas:

The chemical potential of a perfect gas is given by,

−𝜇 𝜕𝜍
=
𝜏 𝜕𝑛 𝑢,𝑉

𝜕 𝑉 1 5
= 𝜕𝑛 𝑛 𝑙𝑜𝑔 + 𝑛 𝑢,𝑉
𝑛 𝜆3 2

𝜕 𝜕 5
= 𝜕𝑛 [𝑛 𝑙𝑜𝑔 𝑉 − 𝑛 𝑙𝑜𝑔 𝑛 − 𝑛 𝑙𝑜𝑔 𝜆3 ]𝑢,𝑣 + 𝜕𝑛 𝑛
2 𝑢,𝑣
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
132
 5
= 𝑙𝑜𝑔 𝑉 − 1 − 𝑙𝑜𝑔𝑛 – 𝑙𝑜𝑔 𝜆3 + 2

𝑉 3
= 𝑙𝑜𝑔 +2
𝑛𝜆 3

𝜇 𝑛𝜆 3 3
= 𝑙𝑜𝑔 −2
𝜏 𝑉

𝑛 𝑝
=
𝑉 𝜏

𝜇 𝑝𝜆 3 3
= 𝑙𝑜𝑔 −2
𝜏 𝜏

𝜆3 3𝜏
µ = 𝜏 𝑙𝑜𝑔 𝑝 + 𝜏 𝑙𝑜𝑔( 𝜏 ) − .
2

= 𝜏 𝑙𝑜𝑔 𝑝 + 𝑓(𝜏)

 Where 𝑓(𝜏) is the function of the temperature alone.

4.12 GIBB’S CANONICAL ENSEMBLE:

(i) System in contact with heat reservoir:

 The micro canonical ensemble describes the systems which are perfectly insulated
and have given energy.
 In thermodynamics we do not know the exact value of energy as we usually deal
with systems kept in thermal contact with a heat reservoir at a given temperature.
Thus we know only its temperature i.e its average energy.
 The energy varies from instant to instant but the time average is known.
 On the other hand the canonical ensemble describes those systems which are not
isolated, but are in thermal contact with a heat reservoir.
 In this situation the system of interest together with a heat reservoir forms a large
closed system and the system of interest is treated as a subsystem.
 If the energy of the large closed system is constant, then it would represent a
microcanonical system where as the subsystem which can exchange energy with a
heat reservoir would represent canonical system.
 Thus any part of sub system of an isolated system in thermal equilibrium can be
represented by a canonical ensemble.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

133
  Consider a micro canonical ensemble representing a very large isolated system.
Imagine that each system of the ensemble is made up of large number of
subsystems which are in mutual thermal contact and can exchange energy.
 Choose a sub system s. The rest of the subsystem is denoted by r called heat
reservoir. The total sub system is denoted by t. As the total system is a member of
the microcanonical ensemble, it is isolated and Et is constant.
 Let the energies of the sub system and heat
reservoir be Es and Er so

𝐸𝑡 = 𝐸𝑟 + 𝐸𝑠

 As s can exchange energy but not the particles,

it is a member of the canonical ensemble. s is
comparatively small but usually macroscopic containing 1024 particles. In the
case of a gas, the sub system may be a single molecule.
(ii) Thermodynamical functions and partition function:
 We shall calculate the entropy, energy, Helmholtz free energy and partition
function of the canonical ensemble.
 Consider an isolated system with total energy E0. This system is a part of a micro
canonical ensemble.
 The micro canonical ensemble minus system is heat reservoir. Our system is in
thermal equilibrium with the heat reservoir in such a way that 𝐸 =E0.
 Let our system in the microcanonical ensemble be defined in the energy range
between E0 and E0+δE. But in microcanonical ensemble δE is unimportant.
 Then we may choose δE to be equal to the range of reasonably probable values of
the energy in the canonical ensemble.
 Therefore we define the entropy of the canonical ensemble with the mean energy
E to be equal to the entropy of a microcanonical ensemble with energy 𝐸 .
 Consider the volume δΓ of the phase space corresponding to the energies
between 𝐸 and 𝐸 +δE

𝜕𝛤 𝐸
∆Γ== 𝛿𝐸
𝜕𝐸 𝐸

To estimate δE:

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

134
  Let ω(E) dE represent the canonical ensemble probability for a system to have
energy in the range E and E+dE.
 ρ(E) is the probability density of unit volume of the phase space at energy E, then
the probability of volume ∆Γ in the range E and E+dE will be ρ(E) ∆Γ(E).

So that,

𝜔 𝐸 𝑑𝐸 = 𝜌 𝐸 ∆𝛤(𝐸) ….(1)
𝜕𝛤 𝐸
= 𝜌(𝐸) 𝑑𝐸
𝜕𝐸 𝐸

 Figure represents the variation ω(t) as a function of E.

 The normalization condition is,

ω(E) dE=1 ….(2)

 The simply means that the area under the curve

ω= ω (E) is equal to unity.
 Since the mean energy of the canonical ensemble is
𝐸 the function ω(E) will have an extremely sharp
maximum at E=𝐸 differing appreciably from zero only in the immediate
neighbourhood of this point.
 So on normalizing the plot we can introduce the width δE of the curve of ω= ω(E)
defining it as the width of the rectangle whose height is equal to the value of the
function ω(E) at its maximum and whose area is equal to unity.
 Thus the width δE is determined by the normalization condition.

ω(𝐸 ) δE = 1

 Comparing equations (1) and (2) with E = 𝐸 , we get

Now, ρ(𝐸 )∆Γ = 1

1
∆Γ = 𝜌(𝐸 )

 But we have, 𝜌(𝐸 ) = 𝐴𝑒 −𝐸 /𝜏

∆𝛤 = 𝐴−1 𝑒 𝐸 /𝜏

= 𝐴−1 𝑒 𝑈/𝑘𝑇 ….(3)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

135
  where 𝐸 =U=energy of the system and η=kT
 So that the statistical entropy ζ is given by,

𝜍 = 𝑙𝑜𝑔 ∆𝛤

= 𝑙𝑜𝑔[𝐴−1 . 𝑒 𝑈/𝑘𝑇 ]
𝑈
= − 𝑙𝑜𝑔 𝐴 + 𝑘𝑇

𝑈
𝑙𝑜𝑔 𝐴 = −𝜍
𝑘𝑇

𝑈−𝑘𝜍𝑇
= 𝑘𝑇

𝑈−𝑆𝑇
= ….(4)
𝑘𝑇

 where 𝑆 = 𝑘𝜍 = thermodynamic entropy.

Helmholtz free energy:

𝐹 = 𝑈 − 𝑇𝜍 = 𝑈 − 𝑆𝑇

𝐹
𝑙𝑜𝑔 𝐴 = 𝑘𝑇

𝐴 = 𝑒 𝐹/𝑘𝑇

 So that the canonical distribution function takes form,

𝜌(𝐸) = 𝐴𝑒 −𝐸/𝜏

= 𝑒 𝐹/𝑘𝑇 𝑒 −𝐸/𝑘𝑇

= 𝑒 (𝐹−𝐸)/𝑘𝑇

 Now applying the normalization condition

𝜌 𝐸 𝑑𝛤 = 1

𝑒 (𝐹−𝐸)/𝑘𝑇 𝑑𝛤 = 1

𝑒 −𝐹/𝑘𝑇 = 𝑒 −𝐸(𝑝,𝑞)/𝑘𝑇 𝑑𝛤

 Now the partition function is defined as,

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

136
 𝑍= 𝑒 −𝐸(𝑝,𝑞)/𝑘𝑇 𝑑𝛤 (classical)

𝑍= 𝑖 𝑒 −𝐸𝑖/𝑘𝑇 (quantum)

So, 𝑒 −𝐹/𝑘𝑇 = 𝑍

−𝐹/𝑘𝑇 = 𝑙𝑜𝑔 𝑍

−𝐹 = 𝑘𝑇 𝑙𝑜𝑔 𝑍

−𝐹 = 𝜏 𝑙𝑜𝑔 𝑍→ 𝐹 = −𝜏 𝑙𝑜𝑔 𝑍. ….(5)

 This equation represents the expression for the Helmholtz free energy in terms of
Z.
 Suppose N independent identical spinless particles, we must correct the classical
partition function dimensionally and take into account the indistinguishability of
the particles so that the correct expression is,

1
𝑍 = 𝑁!𝑕 3𝑁 𝑒 −𝐸(𝑝,𝑞)/𝑘𝑇 𝑑𝛤 (classical) ….(6)

Entropy of a system:

 The statistical entropy of a system in canonical ensemble is given by,

𝜕𝐹
𝜍=− 𝜕𝜏 𝑉

𝜕
= (𝜏 log 𝑍)
𝜕𝜏

𝜕
= log 𝑍 + 𝜏 (log 𝑍) ….(7)
𝜕𝜏

 If E i is the i th energy eigen value of a system, we have

𝑍= 𝑖 𝑒 −𝐸𝑖 /𝜏

log 𝑍 = log( 𝑒 −𝐸𝑖 /𝜏 )

𝑖

𝜕 𝜕 −𝐸𝑖 /𝜏
(log 𝑍) = 𝜕𝜏 log 𝑖𝑒
𝜕𝜏

𝐸𝑖
𝑒 −𝐸 𝑖 /𝜏 )
𝜏2
=
𝑖 𝑒 −𝐸𝑖 /𝜏
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
137
 1 𝐸𝑖 𝑒 −𝐸𝑖 /𝜏
=
𝜏2 𝑒 −𝐸𝑖 /𝜏

1 𝑈
= 𝐸= ….(8)
𝜏2 𝜏2

𝑈
 Now statistical entropy 𝜍 = log 𝑍 + ….(9)
𝜏
𝑈
 The thermo dynamic entropy 𝑆 = 𝑘𝜍 = 𝑘 log 𝑧 + 𝜏

𝑈
𝑆 = 𝑘 log 𝑍 + 𝑇 ….(10)

4.13 A SYSTEM OF NON-INTERACTING CLASSICAL HARMONIC

OSCILLATORS USING CANONICAL ENSEMBLE:
 We now take up the quantum-mechanical situation, according to which the energy
eigenvalues of a one-dimensional harmonic oscillator are given by
 1
 n   n    ; n  0,1,2,.... .…(1)
 2
 Accordingly, we have for the single-oscillator partition function
n 
Q1    exp   n 
n 0
n 
  1 
  exp   n  
n 0   2 
 1 
exp    
  2 
1  exp  
1
 1 
= 2 sinh    …(2)
 2 
 The N – oscillator partition function is then given by
N
  1 
 exp   2   
QN   Q1 
N
  
1  exp  
 

 N 
exp    
  2 
  e  N / 2  1  e  
N
.…(3)
1  exp  N
 For the Helmholtz free energy of the system, we get

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

138
 A  kT ln QN

 
A  kT ln e  N / 2  1  e  
N


 NkT ln[e1 2 ]  NkT ln[1  e  ]

1

 N    kT ln 1  e    ….(4)
2 
 whereby
µ = A/N. .… (5)
P=0 .… (6)
S  kU  k lnQ N

1  
S  kN    
2

(e  1) 

 k ln e  N / 2  1  e   
N

1  
S  kN     
(e  1) 

 k ln e  N / 2   k ln 1  e   
N

2

S  kN

2

 kN 
(e  1)
 kN

2
 k ln 1  e   
N


 
S  kN    
 ln 1  e   
 (e  1) 
 Then
 
S  kN    
 ln 1  e    .…(7)
 (e  1) 
Also
1 1   1 
S  Nk   coth     ln 2 sinh  
2 2   2 

U  ln QN





 
ln e  N / 2  1  e  
N


  

  ln e  N / 2   ln 1  e 
 

N

 e  N / 2   N 2   N 1  e 

  

 N 1 

e () 


 e  N / 2 
  1  e 
N
  

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

139
 N N
  
2 
e 1  e  
N N
  
2 (e  1)

1   1 1 
U  N       N coth    .… (8)
2 (e  1)  2 2 

dU
and C p  Cv 
dT
d 1  
 N    
dT  2 (e  1) 

d   

dT  (e kT  1) 
N

 kT     
 e  
 N kT
kT 2  
 (e  1) 2 
 
 

   e
2

 Nk  
 kT  e  12

 Then
e
C p  Cv  Nk 
2

e 
 2 .… (9)
1
 Also
2
1  1 
Cp  Cv  Nk   cos ech 2   
2  2 

U = standard deviation in energy distribution

 U 
U  kT 2 
2

 T V

e
 kT Nk 
2 2

e 
 2 [From equation (9)]
1

e
 Nk 2T 2
1
2

k 2T 2 e  1
2
 
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
140
 e
 N 
2

e 
1 2

e  2
 Now U  N 
 
.… (10)
e   1

 Thus U  N

 Formula (8) is especially significant, for it shows that

the quantum-mechanical oscillators do not obey the
equipartition theorem.
 The mean energy per oscillator is different from the
equipartition value kT; actually, it is always greater than
kT; see curve 2 in figure.
 Only in the limit of high temperatures, where the
thermal energy kT is much larger than the energy quantum , does the mean
energy per oscillator tend to the equipartition value.
 It should be noted here that if the zero-point energy 1 2  were not present, the
limiting value of the mean energy would be kT  1 2 , and not kT- we may
call such an oscillator the Planck oscillator, see curve 1 in figure .
 In passing, we observe that the specific heat which is the same for the Planck
oscillator as for the Schrodinger oscillator, is temperature dependent; moreover, it
is always less than, and at high temperatures tends to, the classical value.
 Indeed, for kT   , formulae (2) through (9) go over to their classical
counterparts, respectively.
4.14 Grand Canonical Ensemble:
(i) System in contact with a particle reservoir:
 In microcanonical ensemble each system contains same fixed energy as well as
same number of particles.
 The microcanonical ensemble would not be applied to thermodynamics because
we deal with systems kept in contact with heat reservoir. Then we know only the
time average of energy.
 In canonical ensemble we relaxed the condition of constant energy and allowed
the subsystem to exchange energy with heat reservoir. But this model could not be

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

141
 applied to processes where not of particle varies. e.g. chemical processes and
quantum processes.
 Therefore, we seek an ensemble which allows the subsystem to exchange energy
as well as the particles with reservoir.
 Such an ensemble which allows the subsystem to exchange energy as well as the
number of particles with the heat reservoir is called grand Canonical Ensemble.
 In grand canonical ensemble the independent variables are T,V and µ. Then we
have the grand potential.

𝛺 = 𝑈 − 𝑇𝑆 − µ𝑛

which is minimal when T,V and µ are held fixed.

 Consider a microcanonical ensemble representing very large isolated system.
 Each system is made up of large number of sub system which are in mutual
thermal contact and can exchange energy as well as particles with each other.
 Choose a sub system s, heat reservoir r and the total system t.
 𝐸𝑠 and 𝐸𝑟 represent the energies of the sub system and the reservoir.
 𝑛𝑠 and 𝑛𝑟 represent the number of particles in the sub system and the reservoir.
 Then the subsystem and the reservoir may exchange energy and particles subject
to the conditions

𝐸𝑠 + 𝐸𝑟 = 𝐸𝑡 and 𝑛𝑠 + 𝑛𝑟 = 𝑛𝑡

 Now we find the probability 𝑑𝜔𝑠 𝑛𝑠 of finding the sub systems in a state in
which sub system S contains 𝑛 particles and is found in the element 𝑑𝛤 𝑛𝑠 of its
phase space.
 𝑑𝛤 𝑛𝑠 indicates that the nature of the phase space of the subsystem changes with
𝑛𝑠 .
 For grand canonical ensemble

𝑑𝜔 𝑛 = 𝐴 exp 𝑛µ − 𝐸 /𝜏 𝑑𝛤 𝑛

𝛺
 Normalization constant 𝐴 = exp 𝜏

 Ω is called grand potential or thermodynamic potential.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

142
 𝑑𝜔 𝑛 = exp 𝛺 + 𝑛µ − 𝐸 /𝜏 𝑑𝛤 𝑛
𝑑𝜔 𝑛 = 1

exp 𝛺 + 𝑛µ − 𝐸 /𝜏 𝑑𝛤 𝑛 = 1

 From the normalization condition 𝜌 𝑛 𝑑𝛤 𝑛 = 1

 𝜌(𝑛) is the density of distribution of phase points in the phase space.
 Now we get 𝜌 𝑛 = exp 𝛺 + 𝑛µ − 𝐸 /𝜏
 An ensemble characterized by the probability distribution 𝜌(𝑛) given by the
above equation is called grand canonical ensemble.
(ii) Partition function and thermodynamic function for grand canonical
ensemble :
 In grand canonical ensemble sub system is allowed to exchange energy and the
particles with the heat reservoir under the condition,

𝐸𝑠 + 𝐸𝑟 = 𝐸𝑡 𝑎𝑛𝑑 𝑛𝑠 + 𝑛𝑟 = 𝑛𝑡 .…(1)

𝐸𝑠 , 𝑛𝑠 → the energy and the number of particles of the sub system.

𝐸𝑟 , 𝑛𝑟 →the energy and the number of particles of the heat reservoir.

𝐸𝑡 , 𝑛𝑡 →the energy and the number of particles of the total system.

 The probability distribution 𝜌(𝑛) is given by,

𝜌 𝑛 = exp 𝛺 + 𝑛µ − 𝐸 /𝜏 ….(2)

Ω→grand potential

µ→chemical potential

 The grand partition function is defined as,

𝒵 = exp −𝛺/𝜏 ….(3)

= 𝑒 µ𝑛/𝜏 𝑒 𝐸/𝜏 𝑑𝛤(𝑛) (classical)

= 𝑛 𝑖 exp µ𝑛 − 𝐸𝑛,𝑖 /𝜏 (quantum)

𝒵= 𝑒 µ𝑛 𝑍𝑛
𝑛

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

143
 𝑍𝑛 = 𝑖 𝑒 −𝐸/𝜏 𝑑𝛤 𝑛 (classical)

= 𝑒 −𝐸𝑛 ,𝑖 /𝜏 (quantum)

is the canonical partition function .

 i.e the grand partition function 𝒵 is the sum of canonical partition functions Z i
for ensemble with different n′ s with weighing factor 𝑒 −µ𝑛/𝜏 .
 From equation (3) the partition function is given by,

𝒵 = exp −𝛺/𝜏

𝛺 = −𝜏 log 𝒵 ….(4)

 The entropy ζ may be written as,

𝜍 = log ∆𝛤

1
= log
𝜌 𝑛, 𝐸

= − 𝛺 + 𝑛µ − 𝐸 /𝜏

= − 𝛺 + 𝑛µ − 𝑈 /𝜏 ….(5)

𝜏𝜍 = − 𝛺 + 𝑛µ − 𝑈

𝑈 − 𝜏𝜍 = (𝛺 + 𝑛µ)

Helmholtz free energy 𝐹 = 𝑈 − 𝜏𝜍 ….(6)

= 𝛺 + 𝑛µ ….(7)

𝐺 = 𝐹 + 𝑝𝑉 ….(8)

= 𝑈 − 𝜏𝜍 + 𝑝𝑉 ….(9)

𝑑𝐺 = 𝑑𝑈 − 𝜏𝑑𝜍 − 𝜍𝑑𝜏 + 𝑝𝑑𝑉 + 𝑉𝑑𝑝

But 𝑑𝑈 = 𝜏𝑑𝜍 − 𝑝𝑑𝑉 + µ𝑑𝑛

Hence 𝑑𝐺 = 𝜏𝑑𝜍 − 𝑝𝑑𝑉 + µ𝑑𝑛 − 𝜏𝑑𝜍 − 𝜍𝑑𝜏 + 𝑝𝑑𝑉 + 𝑉𝑑𝑝

= µ𝑑𝑛 − 𝜍𝑑𝜏 + 𝑉𝑑𝑝

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
144
 𝑑𝐺
=µ ….(10)
𝑑𝑛 𝜏,𝑝

Hence 𝐺 = µ𝑛 for fixed p and η.

In this case 𝐺 = µ𝑛

𝛺 + 𝑛µ + 𝑝𝑉 = 𝑛 µ

𝛺 = −𝑝𝑉

𝐹 + 𝑝𝑉 = µ𝑛 ….(11)

𝑈 − 𝜏𝜍 + 𝑝𝑉 = µ𝑛

𝑈 − 𝜏𝜍 − 𝛺 = µ𝑛

𝑈 − 𝜏𝜍 − µ𝑛 = 𝛺 ….(12)

𝑑𝛺 = 𝑑𝑈 − 𝜏𝑑𝜍 − 𝜍𝑑𝜏 − µ𝑑𝑛 − 𝑛𝑑µ

= 𝜏𝑑𝜍 − 𝑝𝑑𝑉 + µ𝑑𝑛 − 𝜏𝑑𝜍 − 𝜍𝑑𝜏 − µ𝑑𝑛 − 𝑛𝑑µ

= −𝑝𝑑𝑉 − 𝜍𝑑𝜏 − 𝑛𝑑µ ….(13)

𝜕𝛺
Then 𝑝=− 𝜕𝑉 𝜏,µ

𝜕𝛺
𝜍=− 𝜕𝜏 𝑉,𝜇

𝜕𝛺
𝑛=− 𝜕𝜇 𝑉,𝜏

 From the above three relations we can evaluate thermodynamic quantities for the
grand canonical ensemble.

4.15 CLASSICAL IDEAL GAS USING GRAND CANONICAL ENSEMBLE:

 The grand partition function is given by,

𝒵= 𝑛 𝑒 𝜇𝑛 /𝜏 𝑍𝑛 ….(1)

 Canonical partition function

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

145
 1 𝑓𝑁
𝑍𝑛 = 𝑛!𝑕 3𝑛 𝑒 −𝐸(𝑛)/𝜏 𝑑𝛤(𝑛) = ….(2)
𝑛!

2𝜋𝑚𝑘𝑇 3/2 2𝜋𝑚𝜏 3/2

where 𝑓 = 𝑉 = 𝑉
𝑕2 𝑕2

 ∴ The grand partition function

𝑛
𝜇𝑛 /𝜏 𝑓
𝑛 𝑒 𝜇 /𝜏 𝑓
𝒵= 𝑛𝑒 = 𝑛 ....(3)
𝑛! 𝑛!

𝑥𝑛
 From the series expansion 𝑛 𝑛! = 𝑒𝑥

𝒵 = exp 𝑒 𝜇 /𝜏 𝑓 ….(4)

𝛺 = −𝜏 log 𝒵 = −𝜏𝑒 𝜇 /𝜏 𝑓

2𝜋𝑚𝜏 3/2
𝛺 = 𝜏𝑒 𝜇 /𝜏 𝑉 ....(5)
𝑕2

(a) Chemical potential per particle (𝝁):

𝜕𝛺
𝑛=− 𝜕𝜇 𝑉,𝜏

𝜕 2𝜋𝑚𝜏 3/2
= − 𝜕𝜇 −𝜏𝑒 𝜇 /𝜏 𝑉
𝑕2
𝑉,𝜏

𝜕 2𝜋𝑚𝜏 3/2
= 𝜕𝜇 𝜏𝑒 𝜇 /𝜏 𝑉
𝑕2
𝑉,𝜏

2𝜋𝑚𝜏 3/2 1
=𝜏 𝑉 𝑒 𝜇 /𝜏 𝜏
𝑕2

2𝜋𝑚𝜏 3/2
𝑛= 𝑉𝑒 𝜇 /𝜏 ….(6)
𝑕2

𝛺
=−𝜏

𝛺 = −𝜏𝑛 ….(7)

2𝜋𝑚𝜏 3/2
𝛺 =−𝜏 𝑉𝑒 𝜇 /𝜏 ….(8)
𝑕2

2𝜋𝑚𝜏 3/2
−𝜏𝑛 = −𝜏 𝑉𝑒 𝜇 /𝜏
𝑕2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

146
 2𝜋𝑚𝜏 3/2 𝑉 𝑓
This gives 𝑒 −𝜇 /𝜏 = =𝑛
𝑕2 𝑛

𝜇 𝑓
− 𝜏 = log 𝑛

𝑓
𝜇 = −𝜏 log ….(9)
𝑛

(b) Entropy: (Sackur-Tetrode equation)

 The statistical entropy ζ is given by,

𝜕𝛺
𝜍=− 𝜕𝜏 𝑉,𝜇

𝜕 2𝜋𝑚𝜏 3/2
= 𝜕𝜏 𝜏𝑒 𝜇 /𝜏 𝑉
𝑕2

2𝜋𝑚 3/2 𝜕
= 𝑉 𝜏 5/2 𝑒 𝜇 /𝜏
𝑕2 𝜕𝜏

2𝜋𝑚 3/2 5 3/2 𝜇 /𝜏

= 𝑉 𝜏 𝑒 + 𝜏 5/2 𝑒 𝜇 /𝜏 −𝜇/𝜏 2
𝑕2 2

2𝜋𝑚 3/2 5 𝜇
= 𝑉𝜏 3/2 𝑒 𝜇 /𝜏 −𝜏
𝑕2 2

2𝜋𝑚𝜏 3/2 5 𝜇
𝜍= 𝑉𝑒 𝜇 /𝜏 −𝜏 ….(10)
𝑕2 2

5 𝜇
𝜍=𝑛 −𝜏 ….(11)
2

𝜇 𝑓 𝑓
 From equation (9) we have = − log → 𝜇 = −𝜏 log
𝜏 𝑛 𝑛
𝑛
2𝜋𝑚𝜏 3/2 𝑙𝑜𝑔 5 𝑓
 Now equation (10) becomes 𝜍 = 𝑉𝑒 𝑓 + log
𝑕2 2 𝑛

2𝜋𝑚𝜏 3/2 𝑛 5 𝑓
= 𝑉 + log
𝑕2 𝑓 2 𝑛

2𝜋𝑚𝜏 3/2 2𝜋𝑚𝜏 −3/2 1 5 𝑓

= 𝑉𝑛 + log
𝑕2 𝑕2 𝑉 2 𝑛

5 𝑓
=𝑛 + log
2 𝑛

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

147
 5
𝜍=𝑛 +
2

𝑙𝑜𝑔2𝜋𝑚𝜏𝑕23/2𝑉𝑛 ….(12)

 Thermodynamic entropy S = 𝑘𝜍

5 2𝜋𝑚𝜏 3/2 𝑉
= 𝑛𝑘 + 𝑙𝑜𝑔
2 𝑕2 𝑛

2𝜋𝑚𝜏 3/2 𝑉
= 𝑛𝑘𝑙𝑜𝑔 𝑒 5/2 ….(13)
𝑕2 𝑛

 This gives the famous Sackur-Tetrode equation for the entropy of a perfect gas
and this is in agreement with the equation for microcanonical ensemble and
canonical ensemble.
(c) Internal energy:

We have 𝛺 = 𝑈 − 𝜏𝜍 − 𝜇𝑛

𝑈 = 𝛺 + 𝜏𝜍 + 𝜇𝑛

𝛺 = −𝑛𝜏 [from equation (7)]

5 𝜇
𝜍=𝑛 −𝜏 [from equation (11)]
2

5 𝜇
Now 𝑈 = −𝑛𝜏 +𝜏 𝑛 2
− 𝜏 + 𝜇𝑛

5
= −𝑛𝜏 + 2 𝑛𝜏 − 𝜇𝑛 + 𝜇𝑛

3
= 2 𝑛𝜏.

3
𝑈 = 2 𝑛𝑘𝑇.

 which is well known relation for the internal energy of a perfect gas.
4.16 ENERGY AND DENSITY FLUCTUATIONS IN ENSEMBLES:

(a) Canonical ensemble:

 In canonical ensemble the systems are in thermal equilibrium with the heat
reservoir and so energy fluctuations take place.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

148
  For canonical ensemble partition function,

𝑍= 𝑖 𝑒 −𝛽𝐸𝑖 ….(1)

1
where 𝛽 = 𝑘𝑇

𝜕𝑍
= 𝑖 𝑒 −𝛽𝐸𝑖 −𝐸𝑖
𝜕𝛽

= 𝑖 𝐸𝑖 𝑒 −𝛽𝐸𝑖

𝜕𝑧
𝑖 𝐸𝑖 𝑒 −𝛽 𝐸𝑖 = − 𝜕𝛽 .

−𝛽 𝐸 𝑖
𝑖 𝐸𝑖 𝑒
 Mean energy 𝐸= −𝛽 𝐸𝑖 .
𝑖𝑒

−𝛽 𝐸 𝑖
𝑖 𝐸𝑖 𝑒
= .
𝑍

1 𝜕𝑧
𝐸=− ….(2)
𝑧 𝜕𝛽

𝜕2𝑧
= 𝑖 𝐸𝑖 2 𝑒 −𝛽𝐸𝑖
𝜕𝛽 2

𝜕2𝑧
𝐸𝑖 2 𝑒 −𝛽𝐸𝑖 = 𝜕𝛽 2

2 𝑒 −𝛽 𝐸 𝑖
𝑖 𝐸𝑖
𝐸2 = −𝛽 𝐸 𝑖
𝑖𝑒

2 𝑒 −𝛽 𝐸 𝑖
𝑖 𝐸𝑖
= .
𝑍

1 𝜕2𝑧
𝐸2 = 𝑧 ….(3)
𝜕𝛽 2

𝜕𝐸 𝜕 1 𝜕𝑧
 Using equation (2) = 𝜕𝛽 − 𝑧 𝜕𝛽
𝜕𝛽

𝜕 1 𝜕𝑧 1 𝜕2𝑧 1 𝜕𝑧 2
= − 𝜕𝛽 =− − 𝑧2
𝑧 𝜕𝛽 𝑧 𝜕𝛽 2 𝜕𝛽

= − 𝐸2 − 𝐸 2

2
= − 𝛿𝐸 ….(4)

 The molar heat at constant volume

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
149
 𝜕𝐸
𝐶𝑉 = 𝜕𝑇 𝑉

𝜕𝐸 𝜕𝛽
= 𝜕𝛽 𝜕𝑇

2 1
= − 𝛿𝐸 − 𝑘𝑇 2

1 2
𝐶𝑉 = 𝛿𝐸 ….(5)
𝑘𝑇 2

2
𝛿𝐸 = 𝑘𝑇 2 𝐶𝑉

2 1/2
𝛿𝐸 = 𝑇 𝑘𝐶𝑉 .…(6)

 The energy fluctuation is measured by the ratio

2
∆𝐸 𝛿𝐸
=
𝐸 𝐸

𝑇 𝑘𝐶𝑉 1/2
= ….(7)
𝐸

 For large values of T, C v and 𝐸 are proportional to the number of molecules N

and hence fluctuation is proportional to 𝑁 −1/2 .
 For an ideal gas 𝐸 = 𝑁𝑘𝑇 and 𝐶𝑉 = 𝑁𝑘

∆𝐸 𝑇 1/2
= 𝑘𝑁𝑘
𝐸 𝑁𝑘𝑇

𝑁 1/2
= = 𝑁 −1/2 ….(8)
𝑁

 For a macroscopic system N=1022.

 So the fluctuations are very small in the order 10-11.
 Therefore in canonical ensemble the distribution of energies is so peaked about
the ensemble average energy that in practice regarded as a microcanonical
ensemble.

(b) Grand canonical ensemble:

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

150
  In ground canonical ensemble the fluctuations take place in concentration as well
as energy.
 The energy fluctuation can be calculated as for canonical ensemble.
 For Grand canonical ensemble the partition function,

(𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏

𝒵= 𝑛,𝑖 𝑒 ….(9)

𝜕𝒵 𝑛
= 𝑛,𝑖 𝑒 (𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏
𝜕𝜇 𝜏

1
=𝜏 𝑛,𝑖 𝑛 𝑒 (𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏

𝜕𝒵
𝑛𝑒 (𝜇𝑛 −𝐸𝑛 ,𝑖 )/𝜏 = 𝜏 𝜕𝜇

(𝑛𝜇 −𝐸 𝑛 ,𝑖 )/𝜏
𝑛 ,𝑖 𝑛𝑒
 Mean concentration 𝑛= (𝑛𝜇 −𝐸 𝑛 ,𝑖 )/𝜏
𝑛 ,𝑖 𝑒

(𝑛𝜇 −𝐸 𝑛 ,𝑖 )/𝜏
𝑛 ,𝑖 𝑒
=
𝒵

𝜏 𝜕𝒵
𝑛= … . (10)
𝒵 𝜕𝜇

𝜕2𝒵 1
= 𝜏2 𝑛,𝑖 𝑛2 𝑒 (𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏
𝜕𝜇 2

𝜕2𝒵
𝑛,𝑖 𝑛2 𝑒 (𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏 = 𝜏 2 𝜕𝜇 2

2 (𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏

2 𝑛,𝑖 𝑛 𝑒
𝑛 = (𝑛𝜇 −𝐸𝑛 ,𝑖 )/𝜏
𝑛,𝑖 𝑒

2 𝑒 (𝑛𝜇 −𝐸 𝑛 ,𝑖 )/𝜏
𝑛 ,𝑖 𝑛
=
𝑍

𝜏2 𝜕2𝒵
𝑛2 = . … (11)
𝑧 𝜕𝜇 2

𝜕𝑛 𝜕 𝜏 𝜕𝒵
= 𝜕𝜇
𝜕𝜇 𝒵 𝜕𝜇

𝜏 𝜕2𝒵 𝜏 𝜕𝒵 2
= 𝒵 𝜕𝜇 2 − 𝒵 2 𝜕𝜇

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

151
 𝜕𝑛 𝜏2 𝜕2𝒵 𝜏2 𝜕𝒵 2
𝜏 𝜕𝜇 = − 𝒵2
𝒵 𝜕𝜇 2 𝜕𝜇
𝜕𝑛 2
𝜏 𝜕𝜇 = 𝛿𝑛 … . (12)

2𝜋𝑚𝜏 3/2
 For an ideal classical gas 𝑛 = 𝑒 𝜇 /𝜏 𝑉 𝑕2

𝜕𝑛 1 2𝜋𝑚𝜏 3/2
= 𝜏 𝑒 𝜇 /𝜏 𝑉
𝜕𝜇 𝑕2

𝜕𝑛 1
= 𝑛
𝜕𝜇 𝜏
𝜕𝑛
𝜏 𝜕𝜇 = 𝑛
2
𝛿𝑛 =𝑛

2
𝛿𝑛 = 𝑛

 The concentration fluctuation is measured by the ratio

2
∆𝑛 𝛿𝑛 𝑛 1
= = =
𝑛 𝑛 𝑛 𝑛

𝑝𝑉 = 𝑛 𝑘𝑇

𝑝𝑉
𝑛 = 𝑘𝑇
∆𝑛 𝑘𝑇 1/2
=
𝑛 𝑝𝑉

 Smaller of the volume greater is the fractional fluctuation.

4.17 ENTROPY OF MIXING AND THE GIBB’S PARADOX:

 The partition function of a perfect gas is given by,
𝑉 3/2
𝑍 = 𝑕 3 2𝜋𝑚𝑘𝑇 ….(1)

 The entropy of a perfect gas is given by,

3
𝑆 = 𝑁𝑘 log 𝑍 + 𝑁𝑘 ….(2)
2
𝑉 3
= 𝑁𝑘 𝑙𝑜𝑔 (2𝜋𝑚𝐾𝑇)3/2 + 2 𝑁𝑘
𝑕3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

152
 3 3
= 𝑁𝑘 𝑙𝑜𝑔 𝑉 + 2 𝑙𝑜𝑔 𝑚 + 2 𝑙𝑜𝑔𝑇 + 𝐶 ….(3)

C is a constant term including h,k

 The entropy given by equation (3) does not satisfy the
additive property and giving paradoxial results.
Explanation:
 Consider two systems a and b at the same
temperature Ta=Tb=T
 a and b are partitioned by a barrier as shown in figure.
 The particles of the two system are identical and distinguishable.
 The entropies of a and b are given by,
3 3
𝑆𝑎 = 𝑁𝑎 𝑘[𝑙𝑜𝑔 𝑉𝑎 + 𝑙𝑜𝑔 𝑚𝑎 + 𝑙𝑜𝑔𝑇 + 𝐶]
2 2
3 3
𝑆𝑏 = 𝑁𝑏 𝑘[𝑙𝑜𝑔 𝑉𝑏 + 𝑙𝑜𝑔 𝑚𝑏 + 𝑙𝑜𝑔𝑇 + 𝐶] ….(4)
2 2

 Here Na , ma and Va represent the number of particles, the mass of each particle
and volume of system a.
 Here Nb , mb and Vb represent the number of particles, the mass of each particle
and volume of system b.
 Entropy is an extensive quantity and satisfy the additive property.
 If the entropy given by equation (3) had satisfied the additive property , then by
removing partition and allowing the gas molecules to mix freely, the entropy
of the joint system would be
𝑆𝑎𝑏 = 𝑆𝑎 + 𝑆𝑏

3 3 3
= 𝑁𝑎 𝑘[𝑙𝑜𝑔 𝑉𝑎 + 𝑙𝑜𝑔 𝑚𝑎 + 𝑙𝑜𝑔𝑇 + 𝐶] + 𝑁𝑏 𝑘[𝑙𝑜𝑔 𝑉𝑏 + 𝑙𝑜𝑔 𝑚𝑏 +
2 2 2
3
𝑙𝑜𝑔𝑇 + 𝐶] ….(5)
2

 If the particles of the two system are the same and for convenience we take
𝑉𝑎 = 𝑉𝑏 = 𝑉, 𝑁𝑎 = 𝑁𝑏 = 𝑁 and 𝑚𝑎 = 𝑚𝑏 = 𝑚 ,then the entropy of the
individual system be,
𝑆𝑎 = 𝑆𝑏
3 3
= 𝑁𝑘 𝑙𝑜𝑔 𝑉 + 𝑙𝑜𝑔 𝑚 + 𝑙𝑜𝑔𝑇 + 𝐶 ….(6)
2 2

 Now the entropy of the combined system be,

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

153
 3 3
𝑆𝑎𝑏 = 2𝑁𝑘 𝑙𝑜𝑔 𝑉 + 𝑙𝑜𝑔 𝑚 + 𝑙𝑜𝑔𝑇 + 𝐶 ….(7)
2 2

 Now we shall find actual entropy. Let the partition is removed.

 Allow the molecules of the gas to mix freely.
 Now we have a system with 2N particles and volume 2V.
 Then the entropy of the joint system ab is given by,
3 3
𝑆𝑎𝑏 = 2𝑁𝑘[𝑙𝑜𝑔 2𝑉 + 𝑙𝑜𝑔 𝑚 + 𝑙𝑜𝑔𝑇 + 𝐶]
2 2
3 3
= 2𝑁𝑘[𝑙𝑜𝑔 𝑉 + 𝑙𝑜𝑔 𝑚 + 𝑙𝑜𝑔𝑇 + 𝐶] + 2𝑁𝐾 𝑙𝑜𝑔 2
2 2

= 𝑆𝑎 + 𝑆𝑏 + 2𝑁𝐾 𝑙𝑜𝑔 2 ….(8)

 Equation (8) is not equal to equation (7), but has an additional factor 2Nk log 2.
 Thus by mixing of two gases with each containing N molecules and by removing
a partition between them , then the entropy of the joint system increases by 2Nk
log 2.
 This additional entropy is called entropy of mixing.
 Thus if we use equation (3) for entropy we got the paradoxial results.
 This peculiar behavior of the entropy is called Gibb’s paradox.
To resolve Gibb’s paradox:
 Gibb‟s solved this paradox by considering the two systems with the molecules are
identical and distinguishable.
 If two systems containing same number N are mixed by removing the partition
then the diffusion takes place unnoticeably.
 In this situation N molecules of each system cannot be distinguished in N! ways.
𝑔𝑖𝑛𝑖
 Hence the weight of the configuration 𝑊 = 𝑁! 𝑛𝑖!

𝑔𝑖𝑛𝑖
can be replaced by 𝑊= 𝑛𝑖!

𝑙𝑜𝑔 𝑊 = 𝛴𝑛𝑖 𝑙𝑜𝑔 𝑔𝑖 − 𝛴 𝑙𝑜𝑔 𝑛𝑖 !

= 𝛴𝑛𝑖 𝑙𝑜𝑔𝑔𝑖 − 𝛴 𝑛𝑖 𝑙𝑜𝑔𝑛𝑖 + 𝛴𝑛𝑖
= 𝛴𝑛𝑖 𝑙𝑜𝑔 𝑔𝑖 − 𝛴 𝑛𝑖 𝑙𝑜𝑔 𝑛𝑖 + 𝑁
 From Maxwell- Boltzmann law,
𝑛𝑖 = 𝑔𝑖 𝑒 −𝛼 𝑒 −𝛽𝜀𝑖
𝑙𝑜𝑔 𝑊𝑚𝑎𝑥 = 𝛴𝑛𝑖 𝑙𝑜𝑔𝑔𝑖 − 𝛴 𝑛𝑖 𝑙𝑜𝑔[𝑔𝑖 𝑒 −𝛼 𝑒 −𝛽𝜀𝑖 ] + 𝑁
= 𝛴𝑛𝑖 𝑙𝑜𝑔 𝑔𝑖 − 𝛴𝑛𝑖 𝑙𝑜𝑔𝑔𝑖 + 𝛴 𝑛𝑖 𝛼 + 𝛴𝑛𝑖 𝛽𝜀𝑖 + 𝑁
𝑙𝑜𝑔 𝑊 𝑚𝑎𝑥 = 𝛼𝑁 + 𝛽𝐸 + 𝑁

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

154
 Let us substitute 𝐴 = 𝑒 −𝛼 ; 𝑖𝑒, 𝛼 = − 𝑙𝑜𝑔 𝐴
𝑙𝑜𝑔 𝑊𝑚𝑎𝑥 = −𝑁 𝑙𝑜𝑔 𝐴 + 𝛽𝐸 + 𝑁
= 𝑁 − 𝑁 𝑙𝑜𝑔 𝐴 + 𝛽𝐸
= 𝑁[1 − 𝑙𝑜𝑔𝐴] + 𝛽𝐸
𝑆 = 𝑘 𝑙𝑜𝑔 𝑊𝑚𝑎𝑥
= 𝑁𝑘[1 − 𝑙𝑜𝑔 𝐴] + 𝛽𝐸𝑘
1 3
= 𝑁𝑘 1 − 𝑙𝑜𝑔 𝐴 + 𝑁𝑘𝑇 𝑘
𝑘𝑇 2
3
= 𝑁𝑘[1 − 𝑙𝑜𝑔 𝐴] + 𝑁𝑘 .
2
3
= 𝑁𝑘 − 𝑁𝑘 𝑙𝑜𝑔𝐴 + 2 𝑁𝑘 .
5
= −𝑁𝑘 𝑙𝑜𝑔𝐴 + 2 𝑁𝑘
𝑁 5
= −𝑁𝑘 𝑙𝑜𝑔 𝑍 + 2 𝑁𝑘
𝑍 5
= 𝑁𝑘 𝑙𝑜𝑔 𝑁 + 2 𝑁𝑘
𝑉 (2𝜋𝑚𝐾𝑇 )3/2 5
𝑆 = 𝑁𝑘 𝑙𝑜𝑔 + 𝑁𝑘
𝑕3 𝑁 2

𝑉 2𝜋𝑚𝐾𝑇 3/2 5
= 𝑁𝑘 𝑙𝑜𝑔 + 𝑁𝑘. ....(9)
𝑁 𝑕2 2

 The entropy given by this equation satisfies the additive property since here in the
argument of logarithm we have V/N in place of V.
 In equation (9) replacing N by 2N and V by 2V , the entropy of the combined
system be given by,
2𝑉 2𝜋𝑚𝐾𝑇 3/2 5
𝑆𝑎𝑏 = 2𝑁𝑘 𝑙𝑜𝑔 + 2𝑁𝑘 .
2𝑁 𝑕2 2

𝑉 2𝜋𝑚𝐾𝑇 3/2 5
= 2 𝑁𝑘 𝑙𝑜𝑔 + 2 𝑁𝑘
𝑁 𝑕2

= 2𝑆 = 𝑆𝑎 + 𝑆𝑏 ….(10)
 Thus Gibb‟s paradox is resolved .The resolution of Gibb‟s paradox is an example
of the success of the quantum theory.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

155
UNIT V : QUANTUM STATISTICAL MECHANICS
Quantum-mechanical ensemble theory: Density matrix, Equation of motion for density
matrix, Quantum- mechanical ensemble average; Statistics of indistinguishable particles,
Two types of quantum statistics- Fermi-Dirac and Bose-Einstein statistics, Fermi-Dirac
and Bose-Einstein distribution functions using microcanonical and grand canonical
ensembles (ideal gas only), Statistics of occupation numbers; Ideal Bose gas: Internal
energy, Equation state, Bose-Einstein Condensation and its critical conditions; Bose-
Einstein condensation in ultra-cold atomic gases: its detection and thermodynamic
properties: Ideal Fermi gas: Internal energy, Equation of state, Completely degenerate
Fermi gas.
QUANTUM MECHANICAL ENSEMBLE THEORY

5.1 THE DENSITY MATRIX :

 A pure quantum state of a system is represented by a single eigenvector   .

 When the system is described by non-negative probabilities p , p ,....... for being

in states   ,  ......... , a statistical approach is necessary.

 A pure classical state is represented by a single moving point in phase space, that
have definite value of coordinates q1 , q 2 ,...........q f and canonical momenta
p1 , p2 ,....... p f at each instant of time.

 The statistical state can be described by a non-negative density function

f (q1,…qf, p1…..pf,t)
 The probability that the system is found in the interval dq1,,...dqf... dp1...dpf at
time t is
dq1....dq f , dp1.....dp f 
 The quantum analogue of the classical density function is known as density
operator.
 We know that operators can be expressed by matrices and hence the density
operator expressed as matrix is known as density matrix.
 The density matrix expresses the result of taking quantum mechanical matrix
elements and ensemble averages in the same operation.
 Consider an ensemble consisting of N systems in the normalized states
 i , i  1,2,3.....N. Let i be the probability that an assembly will be in the state  i .

 We then define the density matrix in the {  i } representation as

ij  i ij …(1)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

156
  Suppose now we wish to calculate the probability that if a measurement is made

on an observable whose operator is A having Eigen functions i , the result will
be the Eigen value an corresponding to the Eigen function  n .
 If the assembly is in the state  i and we express  i as a linear combination of the
 i we obtain

 i   cij  j …(2)
j


 The probability that a measurement of A will give an is then simply cin*cin.
 But the probability that the assembly is in the state  i is i . Therefore the
probability that the measurement of A will yield an is just
i cin * cin   i ij cin * cin
i j

  ij c in c in*
i j

^
  c in c in *   i *  j dq
i j

 
    cin  *   cin *  j dq
 i 


=   n *  n dq …(3)


 Therefore the probability that the measurement on A will give an is just


 n n dq  nn
 *  …(4)

 Now suppose we wish to calculate the average value of A. This will be simply



A   an  nn


= 
n
n *  an n dq

 
    n *  A n  n dp

  
 Trace  A .
 

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

157
 
We note that the average A is a double average, quantum mechanical average and a

statistical mechanical average.

5.2 EQUATION OF MOTION FOR DENSITY MATRIX

(A QUANTUM MECHANICAL VERSION OF THE LIOUVILLE’S

THEOREM ):

 From the definition of ρ in {θk} representation, we write

1 𝑁 ∗
𝜌𝑚𝑛 = 𝑁 𝑖=1 𝑐𝑖𝑛 𝑐𝑖𝑚 ….(1)

∗
𝜕𝜌 𝑚𝑛 1 𝜕𝑐𝑖𝑛 𝜕𝑐 𝑖𝑚
 Again =𝑁 𝑁
𝑖=1 𝜕𝑡
∗
𝑐𝑖𝑚 + 𝑐𝑖𝑚 ….(2)
𝜕𝑡 𝜕𝑡

 The schroedinger time dependent wave equation is

ħ 𝜕
𝑖 2𝜋 𝜕𝑡 𝛹𝑖 = 𝐻 𝛹𝑖

ħ 𝜕
𝑖 𝑘 𝜕𝑡 𝑐𝑖𝑘 𝜙𝑘 = 𝑘 𝑐𝑖𝑘 𝐻 𝜙𝑘 ….(3)
2𝜋

 Multiplying equation (3) by θj* and integration over q.

ħ 𝜕
𝑖 2𝜋 𝑘 𝜕𝑡 𝑐𝑖𝑗 𝜙𝑗 ∗ 𝜙𝑘 𝑑𝑞 = 𝑘 𝑐𝑖𝑘 𝜙𝑗 ∗ 𝐻 𝜙𝑘 𝑑𝑞

ħ 𝜕
𝑖 2𝜋 𝜕𝑡 𝑐𝑖𝑗 = 𝑘 𝑐𝑖𝑘 𝐻𝑗𝑘 [𝐻𝑗𝑘 = 𝜙𝑗 ∗ 𝐻 𝜙𝑘 𝑑𝑞] ….(4)

 Taking complex conjugate we have

ħ 𝜕
−𝑖 2𝜋 𝜕𝑡 𝑐𝑖𝑗 ∗ = 𝑘 𝑐𝑖𝑘 ∗ 𝐻𝑗𝑘 ∗ ….(5)

 Substituting equations (4) and (5) in equation (2) and taking Hij=Hji*

ħ 𝜕𝜌 𝑚𝑛 ħ 1 𝑁 𝜕𝑐 𝑖𝑛 ∗ 𝜕𝑐 𝑖𝑚
𝑖 2𝜋 = 𝑖 2𝜋 𝑁 𝑖=1 𝑐𝑖𝑚 + 𝑐𝑖𝑛 ∗
𝜕𝑡 𝜕𝑡 𝜕𝑡

1 1
= −𝑁 𝑘 𝑐𝑖𝑘 ∗ 𝐻𝑛𝑘 ∗ 𝑐𝑖𝑚 + 𝑁 𝑘 𝑐𝑖𝑛 ∗ 𝑐𝑖𝑘 𝐻𝑚𝑘

= − 𝜌𝑚𝑘 𝐻𝑛𝑘 ∗ −𝜌𝑘𝑛 𝐻𝑚𝑘

= −(𝜌𝑚𝑘 𝐻𝑘𝑛 − 𝐻𝑚𝑘 𝜌𝑘𝑛 )

= − 𝜌, 𝐻 𝑚𝑛
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
158
 Thus the time development of density matrix is given by

ħ 𝜕𝜌
𝑖 2𝜋 𝜕𝑡 = − 𝜌, 𝐻

This equation is analogous to the Liouville‟s theorem in classical mechanics.

5.3 QUANTUM MECHANICAL ENSEMBLE AVERAGE :

 An ensemble in quantum statistical mechanics is assumed to be a collection of a
very large number of perfectly isolated (conservative) and hence independent
systems in a variety of quantum mechanical state(q, t).
 Now the quantum mechanical version of Liouville's theorem is
 mn    
i  H  mn
t  
 mn
 and for the element mn to be independent of time, = 0, so that [H,] = 0, this
t
 
means  and H commute.
 Therefore,  is a matrix associated with some constant of the motion of the
system.
    
  is some function of H,  =  ( H ) where ( H ) can be expanded in a power

series in H .
 Taking k‟s as the basic set of eigenvectors, the matrix element of  can be written
as

 mn    n *  H   n dq …(1)
 
 In the special case that the energy eigenfunctions n's are chosen as basic vectors,
equation (1) becomes

 mn    n *H  m dq …(2)
 
 If we consider ρ( H )as a power series in H ,then
H   a0  a1 H  a2 H 2  ........

mn  a0 mn  a1Emmn  a2 Em mn  ........

2

as H mn  Em mn 
 
 a0  a1Em  a2 Em  ..... mn
2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

159
   E m  mn
 and  H m  E m  m …(3)

 Thus any ρ ( H ) is a diagonal matrix. Also, ρ(En) is the probability of observing
the eigenvalue En :
(En) =n ...(4)
 From this equation it follows that in case of degenerate level, all the basic states
m have the same probability n i.e.
m1 =m2 = m3... = (Em1). ...(5)
 Now the state of a system known to have the energy E, within a range E very
small compared to L, must be represented by a superposition of basic states m
belonging to eigen values Em in the range E ≤ Em≤ E +E.
 Let  be the number of basic states m belonging to eigen values Em in the range
E ≤ Em ≤ E + E, then from equation (5), we have
m1  m1 1   m1 2 ..........m1  1   E  ... (6)

 Thus the probability of observing the eigen value En in the range E and E+E is
proportional to 1 /  .
(a) MICRO-CANONICAL ENSEMBLE
 For a closed, isolated thermodynamic system i.e. a system with assigned values
for the independent variables E; n1, n2 ... nr ; x1, x2 ... xs, using energy eigen
functions as basic vectors, we write
mn   mn n , …(7)
1
 Where n  for E  E n  E  E ...(8)

= 0 otherwise,
 where E is a very small range in E, and  is the number of basic states n
belonging to eigen values En in the range E≤ En≤ E + E. The constant 1 / 
result from the normalization


n
n 1 ...(9)

 From equation (8) it is clear that for a system known to have an energy between E
and E + E, all basic states n belonging to an eigen value En in this range have

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

160
 the same probability n. This is usually called the postulate of “equal a priori
probabilities”.
(b) CANONICAL ENSEMBLE:
 For a closed, isothermal thermodynamic system i.e., a system with assigned
values for independents variables T, ... n1 … nr, x1… xs using arbitrary basic
vectors k


ρ  constant  e  H ...(10)
1
 Where  = constant = . From equation (1) we can write
kT


mn  constant   n * e  H m dq ....(11)

 In general, mn= n is the probability of observing an eigen value an.

 Hence we can write


 nk 1 constant    n * e  H m dq

n n

1
 or constant = 


n
n * e  H  m dq


1
= where Z    n * e  H m dq …(12)
Z n


  H
e
 and  …(13)
Z
 where Z (, x1,… xs, n1…nr ) is called the partition function.

p,q
 The classical partition function is  e  H dp dq …(14)

 Comparing equations (12) and (14), the integration over  in classical mechanics
is replaced by a summation over quantum states.
 Z may be written in alternative ways

Z    e  H   trace e  H
 

n   nn


 sum of eign values of e H

 Z is obviously invariant under a change of basic vectors.
 The most invariant special choice of basic vectors is the set of energy eigen
function n.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
161
  Therefore we can write

1
   n e  m dq
*  H
mn
Z

1
   n  m e Em dq
*
...(15)
Z

1  2
 H
 Because e  1   H  ( H )
2!

and H  m  E m m
 Now we obtain

  m dq    n  meEm dq
*  H *
n e

1
=  mn e En
Z
= mn n
 Where n is the probability that a system, chosen at random from canonical
ensemble, will be found in the energy state En
 The partition function is written as


Z    n e  H  n dq
*

   n eE n dq
2

  e En ....(16)
n

(c) GRAND CANONICAL ENSEMBLE:

 Now we consider the case of an ensemble composed of members which can differ
not only in the state but also in the amounts of material of various kinds which
they contain. Such an ensemble is called the grand canonical ensemble.
 Let us suppose that a system is composed of r independent kinds of components
and n1, n2...nr be the number of molecules in any member system of grand
ensemble.
 For an open, isothermal thermodynamic system i.e, a system with assigned values
for the independent variables T; (µ1-µr; x1-xs) we can define the grand canonical
ensemble corresponding to such a system by the formula,
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
162
 

  constant  e - H  e n …(1)

 where n  1n1   2 n2  ....   r nr


 mn  constant  e -n   n e  H  m dq
*
 …(2)
 The total probability of finding a system in one or another state N will be taken as
normalized to unity. Thus we can write,


N ,n
nn 1 …(3)

 Now from equation (2), we have

1
Constant  

en  n e H mdq

*

1
= …(4)
Z
 where Z is called the „grand partition function‟. Therefore


e  H e n
 …(5)
Z
 From equation (5), we get
en 

Z 
*  H
 nn   n e n dq

e n
 Z(n )
Z
=  (n) say. …(6)
 Combining equations (5) and (6), we get


 (n) e  H ( n )
 …(7)
Z ( n)
 If the basis vectors are a set of energy eigen functions, then, the probability of
observing the state n (N, x) is
e n e En
n N    nn  …(8)
 e n e En
(d) CONDITION FOR STATISTICL EQUILIBRIUM:
 When a system is in equilibrium, its corresponding ensemble must be
stationary.i.e

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

163
 mn  0 .This is possible, when
i. the density matrix is constant, or
ii. the density matrix is a function of a constant of motion
(i) If the density matrix is constant, its element will be given by
mn  0mn …(1)
 i.e., all the non-diagonal elements of the matrix will be zero and all the diagonal
elements will be equal to a constant 0

 In the energy representation, the basic function n are the eigen functions of the

Hamiltonian H .So, the matrices H and  are diagonal. Thus
mn  n mn …(2)

 In the representation, the density operator  may be written as

  
    n n n dq  H, 
*

  …(3)
n

 To verify this, consider an element  kl . Now

 

 kl    k  l dq      k  n dq   n  n  l dq
* *

n  

  kn nnl  k kl …(4)

n

 This agrees with equation (2)

  
 Therefore i mn    H , 
l  


  H ml l n  ml H l n 
l


  0 H ml  l n  0ml H l n 
l

 0 Hmn  Hmn  =0 …(5)

 The distribution does not change with time. So, the system under consideration is
in equilibrium.
 
(ii) From eq. (5), it is obvious that  commutes with H . Therefore,  must be a
function of
a constant of motion.
5.4 INDISTINGUISHABILITY AND QUANTUM STATISTICS:

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

164
  By indistinguishable particles we mean that if the position and spin coordinates of
two of them are interchanged, there is no physical way of measuring that a change
has been made.
 In classical mechanics identical particles do not lose their individuality despite the
identity of their physical properties.
 If a pair of particles is completely equivalent even then it is possible to identify
them by the continuity of their trajectories because this property enables an
observer to follow each particle.
 This is due to the fact that their wave packets do not overlap, and the particles
move in separate, distinguishable continuous orbits.
 As an example, consider the molecules in a gas at N.T.P.
Molecular density = 1019 mole./cm.2
Volume available to each molecule = 10-19 cm.3
Molecular radius = 10-8 cm.
Molecular volume = 10-24 cm.3
 Because the molecule is smaller than the volume available, we can identify every
molecule of the gas. The molecules are thus distinguishable.
 The situation is quite different in quantum mechanics as follows at once from the
uncertainty principle.
 Due to the uncertainty principle, the concept of the path ceases to have any
meaning. If the path of an electron is exactly known at a given instant, its
coordinates have no definite values even at an infinitely close subsequent instant.
 By localising and numbering the identical particles at some instant, at some other
instant we cannot say which of particle arrived at that point.
 In quantum mechanics there is no way of keeping track of each particle separately
when the wave functions of two identical particles overlap.
 Thus in quantum mechanics there is, in principle, no possibility of separately
following each of a number of similar particles and thereby distinguishing them.
 As an example, we consider the conduction electrons of a metal:
Density of electrons = 1022 per cm3
Volume available to each electron = 10-22 cm3
Momentum px = (2mE)l/2 for 1 eV. = 0.5 x 10-19 erg-sec cm-1
Uncertainty in position x = h/px = 13 x 10-8 cm.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

165
 Volume of conduction electron = (13 x 10-8)3 cm3
=2 x 10-21 cm3.
 Comparing the volume of conduction electron to the volume available, we
conclude that the electron wave functions overlap considerably and hence they are
indistinguishable.
 Thus we have two categories of particles:
(i) Classical, which are identical but distinguishable.
(ii) Quantum, which are identical and indistinguishable.
 When quantum particle density is low, i.e., uncertainty is small in comparison to
the volume available, the particles obey classical statistics otherwise we use
quantum statistics.
5.5 ILLUSTRATION OF CLASSICAL AND QUANTUM STATISTICS
 Consider a gas consists of only two particles a and b.
 Assume that each particle can be in one of the possible quantum states S = 1,2,3.
 Let us calculate the possible states of the whole gas.
Maxwell –Boltzmann statistics:
 The particles are considered distinguishable and any number of particle can be any
one state.
1 2 3
ab - -
- ab -
- - ab
a b -
b a -
a - b
b - a
- a b
- b a
 Total states 32 = 9 possible states for the whole gas.
Bose-Einstein statistics:
 Particles are indistinguishable , i.e b=a
1 2 3

aa - -

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

166
 - aa -

- - aa

a a -

a - a

- a a

 3+3=6 possible states for the whole gas.

Fermi-Dirac statistics:
 Particles are indistinguishable and no more than one particle can be in any one
state.
1 2 3

a a -

a - a

- a a

 3 possible states for the whole gas.

𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑡𝑕𝑎𝑡 𝑡𝑕𝑒 𝑡𝑤𝑜 𝑝𝑎𝑟𝑡𝑐𝑙𝑒𝑠 𝑎𝑟𝑒 𝑓𝑜𝑢𝑛𝑑 𝑖𝑛 𝑡𝑕𝑒 𝑠𝑎𝑚𝑒 𝑠𝑡𝑎𝑡𝑒
Let 𝜀 = 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑡𝑕𝑎𝑡 𝑡𝑕𝑒 𝑡𝑤𝑜 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑎𝑟𝑒 𝑓𝑜𝑢𝑛𝑑 𝑖𝑛 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡 𝑠𝑡𝑎𝑡𝑒𝑠

 Thus for the three cases,

3
𝜀𝑀−𝐵 = 6 = 1/2
3
𝜀𝐵−𝐸 = 3 = 1
0
𝜀𝐹−𝐷 = 3 = 0

 Thus in B-E statistics, there is a greater tendency for the particles to bunch
together in the same states in comparison to M-B statistics.
 On the other hand, in the F-D statistics, there is a greater relative tendency to
particles to remain apart in different states than there is in classical statistics.
5.6 BOSE-EINSTEIN STATISTICS:

 Consider a system having n identical and indistinguishable particles.

 These particles be divided into quantum groups such that there are 𝑛1, 𝑛2, … , 𝑛𝑖 , …
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
167
 number of particles with energies 𝜀1 , 𝜀2 , … , 𝜀𝑖 , … respectively.
 𝑔𝑖 be the number of eigen states in the i th level.

Conditions:

 Particles are identical and indistinguishable.

 Particles do not obey Pauli‟s exclusion principle.
 The total number of particles in the system is constant.
𝑁= 𝑛𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
 The total energy of the system is constant.
𝐸= 𝑛𝑖 𝜀𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

Bose-Einstein distribution function:

 Consider the i th level

 𝑔𝑖 cells and 𝑛𝑖 particles. Here we have to find out how 𝑛𝑖 particles can be divided
into 𝑔𝑖 cells.
 First the choice that which cell will head the sequence can be found as 𝑔𝑖 ways.
 Then the total number of permutations among 𝑛𝑖 particles and the remaining
(𝑔𝑖 − 1) cells is 𝑛𝑖 + 𝑔𝑖 − 1 !
 Now the total number of possible ways in which 𝑛𝑖 particles can be distributed in
𝑔𝑖 cells is
𝑔𝑖 𝑛𝑖 + 𝑔𝑖 1 ! ....(1)
 Since the particles are indistinguishable the permutations of the particles among
themselves will not give rise to different arrangements. Hence equation (1) must
be divided by 𝑛𝑖 !

𝑔𝑖 𝑛𝑖 + 𝑔𝑖 − 1 !
𝑛𝑖 !

 Similarly the permutations of cells among themselves will not give rise to
different arrangements. Hence equation (1) must also be divided by 𝑔𝑖 !

𝑔𝑖 𝑛𝑖 + 𝑔𝑖 − 1 ! 𝑛𝑖 + 𝑔𝑖 − 1 !
=
𝑛𝑖 ! 𝑔𝑖 ! 𝑛𝑖 ! 𝑔𝑖 − 1 !

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

168
  Considering all the available groups such that n1 particles with energy 𝜀1 , n2
particles with energy 𝜀2 and so on. Then the total number of possible
arrangements is given by,

𝐺
𝑛𝑖 + 𝑔𝑖 − 1 !
= . … (2)
𝑛𝑖 ! 𝑔𝑖 − 1 !

𝑛𝑖 and 𝑔𝑖 >> 1, hence one may be neglected.

𝑛𝑖 + 𝑔𝑖 !
𝐺=
𝑛𝑖 ! 𝑔𝑖 !

 The probability Ω of the system is proportional to the total number of eigen states.

𝛺 = 𝐺 × 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝑛 𝑖 +𝑔 𝑖 !
𝛺= × 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 . … (3)
𝑛 𝑖 !𝑔 𝑖 !

log 𝛺 = log 𝑛𝑖 + 𝑔𝑖 ! − log 𝑛! − log 𝑔𝑖 ! + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

 Using stirling′s approximation, we get

log 𝛺 = 𝑛𝑖 + 𝑔𝑖 log 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 log 𝑛𝑖 + 𝑛𝑖 − 𝑔𝑖 log 𝑔𝑖 + 𝑔𝑖

+ 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

= 𝑛𝑖 + 𝑔𝑖 log 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 log 𝑛𝑖 − 𝑔𝑖 log 𝑔𝑖 + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 . . . . (4)

1 1
𝛿 log 𝛺 = 𝑛𝑖 + 𝑔𝑖 𝛿𝑛𝑖 + log 𝑛𝑖 + 𝑔𝑖 𝛿𝑛𝑖 − 𝑛𝑖 𝛿𝑛𝑖 −
𝑛 𝑖 +𝑔 𝑖 𝑛𝑖

log𝑛𝑖 𝛿𝑛𝑖+0

= 𝛿𝑛𝑖 + log 𝑛𝑖 + 𝑔𝑖 𝛿𝑛𝑖 − 𝛿𝑛𝑖 − log 𝑛𝑖 𝛿𝑛𝑖

= 𝑙𝑜𝑔 𝑛𝑖 + 𝑔𝑖 − log 𝑛𝑖 𝛿𝑛𝑖

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

169
 𝑛 𝑖 +𝑔 𝑖
= 𝑙𝑜𝑔 𝛿𝑛𝑖
𝑛𝑖

=
𝑛𝑖
− 𝑙𝑜𝑔 𝛿𝑛𝑖 . … (5)
𝑛 𝑖 +𝑔 𝑖

𝛿(log 𝛺) = 0.

𝑛𝑖
− 𝑙𝑜𝑔 𝛿𝑛𝑖 = 0
𝑛 𝑖 +𝑔 𝑖

𝑛𝑖
𝑙𝑜𝑔 𝛿𝑛𝑖 = 0 … . (6)
𝑛 𝑖 +𝑔 𝑖

𝑛𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝛿𝑛𝑖 = 0 ….(7)

𝜀𝑖 𝑛𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝜀𝑖 𝛿𝑛𝑖 = 0 . … (8)

 Multiply the equation (7) by 𝛼 and (8) by β and then adding to equation (6),we
get,

𝑛𝑖
log + 𝛼 + 𝛽𝜀𝑖 𝛿𝑛𝑖 = 0 ….(9)
𝑛 𝑖 +𝑔 𝑖

𝛿𝑛𝑖 ≠ 0

𝑛𝑖
∴ log + 𝛼 + 𝛽𝜀𝑖 = 0
𝑛 𝑖 +𝑔 𝑖

𝑛𝑖
log + 𝛼 + 𝛽𝜀𝑖 = 0
𝑛 𝑖 +𝑔 𝑖

𝑛𝑖
log = −(𝛼 + 𝛽𝜀𝑖 )
𝑛 𝑖 +𝑔 𝑖

𝑛𝑖
= 𝑒 − 𝛼+𝛽 𝜀 𝑖
𝑛 𝑖 +𝑔 𝑖

𝑛 𝑖 +𝑔 𝑖
= 𝑒 𝛼+𝛽 𝜀 𝑖
𝑛𝑖

𝑔𝑖
1+ = 𝑒 𝛼+𝛽 𝜀 𝑖
𝑛𝑖

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

170
 𝑔𝑖
= 𝑒 𝛼+𝛽 𝜀 𝑖 − 1
𝑛𝑖

𝑔𝑖
𝑛𝑖 = 𝛼 +𝛽 𝜀 𝑖 −1 ….(10)
𝑒

 This equation represents the most probable distribution for a system obeying Bose
- Einstein statistics and known as Bose - Einstein distribution law.

5.7 FERMI-DIRAC STATISTICS:

 Consider a system having n identical and indistinguishable particles.

 These particles be divided into quantum groups such that there are 𝑛1 ,𝑛2 ,…,𝑛𝑖 ,…
number of particles with energies 𝜀1 ,𝜀2 ,…,𝜀𝑖 ,… respectively.
 𝑔𝑖 be the number of eigen states in the 𝑖 𝑡𝑕 level.

Conditions:

 Particles are identical and indistinguishable

 Particles obey Pauli‟s exclusion principle. Hence each cell contains 0 (or) 1
particle. Obviously𝑔𝑖 ≥ 𝑛𝑖 .
 The total number of particles in the system is constant.
𝑁= 𝑛𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
 The total energy of the system is constant.
𝐸= 𝑛𝑖 𝜀𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

Fermi-Dirac distribution function:

 Consider the i th level.

 There are 𝑔𝑖 cells and 𝑛𝑖 particles.
 Each cell must be occupied by zero or one particle.
 Among 𝑔𝑖 cells, only 𝑛𝑖 cells are occupied by one particle and the remaining
𝑔𝑖 − 𝑛𝑖 cells are empty.
 The possible number of such a distribution is given by 𝑔𝑖 ! ....(1)
 Since the particles are indistinguishable the permutation of the particles among
themselves will not give rise to different arrangements. Hence equation (1) must
be divided by 𝑛𝑖 !

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

171
 𝑔𝑖 !
𝑛𝑖 !

 𝑔𝑖 − 𝑛𝑖 ! permutations of empty cells among themselves will not give rise to

different arrangements. Hence equation(1) must also be divided by 𝑔𝑖 − 𝑛𝑖 !

𝑔𝑖 !
𝑛𝑖 ! 𝑔𝑖 − 𝑛𝑖 !

 Considering all the available groups such that 𝑛1 particles with energy 𝜀1 , 𝑛2
particles with energy 𝜀2 and so on. Then the total number of possible
arrangements is given by
𝑔𝑖 !
𝐺= . … (2)
𝑛 𝑖 ! 𝑔 𝑖 −𝑛 𝑖 !

 The probability Ω of the system is proportional to the total number of eigen

states.

𝛺 = 𝐺 × 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝑔𝑖 !
𝛺= × 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 . … (3)
𝑛 𝑖 ! 𝑔 𝑖 −𝑛 𝑖 !

𝑙𝑜𝑔𝛺 = log 𝑔𝑖 ! − log 𝑛𝑖 ! − log 𝑔𝑖 − 𝑛𝑖 ! + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

 Using stirling′ approximation, we get

𝑙𝑜𝑔𝛺 = 𝑔𝑖 log 𝑔𝑖 − 𝑔𝑖 − 𝑛𝑖 log 𝑛𝑖 + 𝑛𝑖 − 𝑔𝑖 − 𝑛𝑖 log 𝑔𝑖 − 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖

+ 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

= 𝑔𝑖 log 𝑔𝑖 − 𝑛𝑖 log 𝑛𝑖 − 𝑔𝑖 − 𝑛𝑖 log( 𝑔𝑖 − 𝑛𝑖 ) + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

1 𝑔𝑖 − 𝑛𝑖
𝛿(log 𝛺) = −𝑛𝑖 𝛿𝑛𝑖 − log 𝑛𝑖 𝛿𝑛𝑖 − −𝛿𝑛𝑖 − log 𝑔𝑖 − 𝑛𝑖 (−𝛿𝑛𝑖 )
𝑛𝑖 𝑔𝑖 − 𝑛𝑖

= −𝛿𝑛𝑖 − 𝑙𝑜𝑔𝑛𝑖 𝛿𝑛𝑖 + 𝛿𝑛𝑖 + log⁡

(𝑔𝑖 − 𝑛𝑖 )𝛿𝑛𝑖 .

= 𝑙𝑜𝑔𝑛𝑖 𝛿𝑛𝑖 + log⁡

(𝑔𝑖 − 𝑛𝑖 )𝛿𝑛𝑖 .

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

172
 𝑔 𝑖 −𝑛 𝑖
= log 𝛿𝑛𝑖 .
𝑛𝑖

𝑛𝑖
=− log 𝛿𝑛𝑖 . … (5)
𝑔 𝑖 −𝑛 𝑖

𝛿(log 𝛺) = 0

𝑛𝑖
− log 𝛿𝑛𝑖 = 0.
𝑔 𝑖 −𝑛 𝑖

𝑛𝑖
log 𝛿𝑛𝑖 = 0 . … (6)
𝑔 𝑖 −𝑛 𝑖

𝑛𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝛿𝑛𝑖 = 0 . … (7)

𝜀𝑛𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.

𝜀𝛿𝑛𝑖 = 0 . … (8)

 Multiplying equation (7) by 𝛼 and (8) by 𝛽 and then adding to equation (6), we
get

𝑛𝑖
log + 𝛼 + 𝛽𝜀𝑖 𝛿𝑛𝑖 = 0
𝑔 𝑖 −𝑛 𝑖

𝑛𝑖
log + 𝛼 + 𝛽𝜀𝑖 = 0
𝑔 𝑖 −𝑛 𝑖

𝑛𝑖
log + 𝛼 + 𝛽𝜀𝑖 = 0
𝑔 𝑖 −𝑛 𝑖

𝑛𝑖
log = − 𝛼 + 𝛽𝜀𝑖
𝑔 𝑖 −𝑛 𝑖

𝑛𝑖
= 𝑒 − 𝛼+𝛽 𝜀 𝑖
𝑔 𝑖 −𝑛 𝑖

𝑔 𝑖 −𝑛 𝑖
= 𝑒 𝛼+𝛽 𝜀 𝑖
𝑛𝑖

𝑔𝑖
− 1 = 𝑒 𝛼+𝛽 𝜀 𝑖
𝑛𝑖
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
173
 𝑔𝑖
= 𝑒 𝛼+𝛽𝜀 𝑖 + 1
𝑛𝑖

𝑔𝑖
𝑛𝑖 =
𝑒 𝛼+𝛽𝜀 𝑖 + 1

 This equation represents the most probable distribution for a system obeying
Fermi-Dirac statistics and known as Fermi-Dirac distribution law.
5.8 AN IDEAL GAS IN A QUANTUM MECHANICAL MICRO CANONICAL
ENSEMBLE:
 We consider a gaseous system of N non-interacting, indistinguishable particles
confined to a space of volume V and energy E.
 Let 𝛺(𝑁, 𝑉, 𝐸) be the number of distinct microstates accessible to the system
under the macro state 𝛺(𝑁, 𝑉, 𝐸).
 Let 𝜀𝑖 denote the average energy of a level and 𝑔𝑖 be the number of levels in the
𝑖 𝑡𝑕 cell.𝑔𝑖 ≫ 1.
 We have n1 particles in the first cell, n 2 particles in the second cell and so on. The
distribution set 𝑛𝑖 must confirm to the conditions,

𝑖 𝑛𝑖 = 𝑁 ….(1)

𝑖 𝑛𝑖 𝜀𝑖 = 𝐸 ….(2)
 Then 𝛺 𝑁, 𝑉, 𝐸 = 𝛴 ′ 𝑊{𝑛𝑖 } ….(3)
 𝑊{𝑛𝑖 } is the number of distinct microstates associated with the distribution
set{𝑛𝑖 }. The primed summation goes over all distribution sets that confirm to
conditions (1) and (2),
 𝑊{𝑛𝑖 } = 𝑖 𝑤(𝑖) ….(4)
 𝑤(𝑖)is the number of distinct ways in which the ni identical and indistinguishable

particles can be distributed among the g i levels of the i th cell.

𝑛 𝑖 +𝑔 𝑖 −1 !
 In B-E case 𝑤𝐵𝐸 𝑖 = 𝑛 𝑖 ! 𝑔 𝑖 −1 !
𝑛 𝑖 +𝑔 𝑖 −1 !
and hence 𝑊𝐵𝐸 𝑛𝑖 = ….(5)
𝑛 𝑖 ! 𝑔 𝑖 −1 !

 In the F-D case, no single level can accommodate more than one particle.
𝑔𝑖 !
𝑤𝐹𝐷 𝑖 = 𝑛
𝑖! 𝑔 𝑖 −1 !
𝑔𝑖 !
and hence 𝑊𝐹𝐷 𝑛𝑖 = 𝑛 ....(6)
𝑖 ! 𝑔 𝑖 −1 !
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
174
  In M-B case the particle are distinguishable.
𝑔𝑖 𝑛 𝑖
 𝑊𝑀𝐵 𝑛𝑖 = ….(7)
𝑛𝑖!

 Now the entropy of the system would be given by,

𝑆 𝑁, 𝑉, 𝐸 = 𝑘 ln 𝛺(𝑁, 𝑉, 𝐸)
′
= 𝑘 ln 𝑛𝑖 𝑊(𝑛𝑖 )
≈ 𝑘 ln 𝑊(𝑛𝑖∗ ) ….(8)
 (ni ) is the distribution set that maximizes the number 𝑊 𝑛𝑖 . 𝑛𝑖∗ is the most
*

probable value of the distribution number ni .

 Our condition for determining the most probable distribution set 𝑛𝑖∗ now turns
out to be,
𝛿𝑙𝑛𝑊 𝑛𝑖 − 𝛼 𝑖 𝛿𝑛𝑖 + 𝛽 𝑖 𝜀𝑖 𝛿𝑛𝑖 =0 ….(9)
𝑙𝑛𝑊 𝑛𝑖 = 𝑖 ln 𝑤(𝑖)
 In B-E case,
(𝑛 𝑖 +𝑔 𝑖 −1)!
𝑊𝐵𝐸 𝑛𝑖 =
𝑛 𝑖 ! 𝑔 𝑖 −1 !

𝑛 𝑖 +𝑔 𝑖 !
≈ 𝑛 𝑖 !𝑔 𝑖 !

ln 𝑊𝐵𝐸 𝑛𝑖 = ln 𝑛𝑖 + 𝑔𝑖 ! − ln 𝑛𝑖 ! − ln 𝑔𝑖 !
= 𝑛𝑖 + 𝑔𝑖 𝑙𝑛 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 ln 𝑛𝑖 + 𝑛𝑖 −
𝑔𝑖ln𝑔𝑖+𝑔𝑖
= 𝑛𝑖 + 𝑔𝑖 𝑙𝑛 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 𝑙𝑛 𝑛𝑖 −
𝑔𝑖ln𝑔𝑖
= 𝑛𝑖 ln 𝑛𝑖 + 𝑔𝑖 − 𝑛𝑖 ln 𝑛𝑖 + 𝑔𝑖 ln 𝑛𝑖 + 𝑔𝑖 − 𝑔𝑖 ln 𝑔𝑖
𝑛 𝑖 +𝑔 𝑖 𝑛 𝑖 +𝑔 𝑖
= 𝑛𝑖 ln + 𝑔𝑖 ln
𝑛𝑖 𝑔𝑖
𝑔𝑖 𝑛𝑖
= 𝑛𝑖 ln + 1 + 𝑔𝑖 ln 1 − 𝑎
𝑛𝑖 𝑔𝑖

 In general,
𝑔𝑖 𝑔𝑖 𝑛𝑖
ln 𝑊 𝑛𝑖 = 𝑛𝑖 ln −𝑎 − ln 1 − 𝑎 ….(10)
𝑛𝑖 𝑎 𝑔𝑖

a= -1 BE case
a=+1 FD case
a=0 MB case

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

175
 𝛿𝑛 𝑖 𝛿𝑛 𝑖
𝛿 ln 𝑊𝐵𝐸 𝑛𝑖 = 𝑛𝑖 + 𝑔𝑖 + ln 𝑛𝑖 + 𝑔𝑖 𝛿𝑛𝑖 − 𝑛𝑖 −
𝑛 𝑖 +𝑔 𝑖 𝑛𝑖

ln𝑛𝑖𝛿𝑛𝑖

= ln 𝑛𝑖 + 𝑔𝑖 − ln 𝑛𝑖 𝛿𝑛𝑖
𝑛 𝑖 +𝑔 𝑖
= ln 𝛿𝑛𝑖
𝑛𝑖
𝑔𝑖
= ln + 1 𝛿𝑛𝑖
𝑛𝑖

 In F-D case,
𝑔𝑖 !
𝑊𝐹𝐷 = 𝑛 𝑖 ! 𝑔 𝑖 −𝑛 𝑖 !

ln 𝑊𝐹𝐷 = ln 𝑔𝑖 ! − ln 𝑛𝑖 ! − ln 𝑔𝑖 − 𝑛𝑖 !
= 𝑔𝑖 ln 𝑔𝑖 − 𝑔𝑖 − 𝑛𝑖 𝑙𝑛 𝑛𝑖 + 𝑛𝑖 − 𝑔𝑖 − 𝑛𝑖 ln 𝑔𝑖 − 𝑛𝑖 +
𝑔𝑖−𝑛𝑖
= 𝑔𝑖 ln 𝑛𝑖 − 𝑛𝑖 ln 𝑛𝑖 − 𝑔𝑖 − 𝑛𝑖 ln(𝑔𝑖 − 𝑛𝑖 )
𝛿𝑛 𝑖 −𝛿𝑛 𝑖
𝛿 ln 𝑊𝐹𝐷 = −𝑛𝑖 − ln 𝑛𝑖 𝛿𝑛𝑖 − 𝑔𝑖 − 𝑛𝑖 − ln 𝑔𝑖 −
𝑛𝑖 𝑔 𝑖 −𝑛 𝑖

𝑛𝑖−𝛿𝑛𝑖

= −𝛿𝑛𝑖 − ln 𝑛𝑖 𝛿𝑛𝑖 + 𝛿𝑛𝑖 + ln 𝑔𝑖 − 𝑛𝑖 𝛿𝑛𝑖

𝑔 𝑖 −𝑛 𝑖 𝑔𝑖
= ln 𝛿𝑛𝑖 = ln − 1 𝛿𝑛𝑖
𝑛𝑖 𝑛𝑖

 In M-B case
𝑔𝑖 𝑛 𝑖
𝑊𝑀𝐵 𝑛𝑖 = 𝑛𝑖!

ln 𝑊𝑀𝐵 = 𝑛𝑖 ln 𝑔𝑖 − ln 𝑛𝑖 !
= 𝑛𝑖 ln 𝑔𝑖 − 𝑛𝑖 𝑙𝑛 𝑛𝑖 + 𝑛𝑖
𝛿𝑛 𝑖
𝛿 ln 𝑊𝑀𝐵 = 𝑛𝑖 . 0 +ln 𝑔𝑖 𝛿𝑛𝑖 − 𝑛𝑖 − ln 𝑛𝑖 𝛿𝑛𝑖 + 𝛿𝑛𝑖
𝑛𝑖

= ln 𝑔𝑖 𝛿𝑛𝑖 − 𝛿𝑛𝑖 − ln 𝑛𝑖 𝛿𝑛𝑖 + 𝛿𝑛𝑖

𝑔𝑖 𝑔𝑖
= ln 𝛿𝑛𝑖 = ln − 𝑎 𝛿𝑛𝑖
𝑛𝑖 𝑛𝑖

 Now equation (9) becomes,

𝑔𝑖
𝑖 ln − 𝑎 − 𝛼 − 𝛽𝜀𝑖 𝛿𝑛𝑖 = 0
𝑛𝑖 𝑛=𝑛 ∗
𝑔𝑖
ln − 𝑎 − 𝛼 − 𝛽𝜀𝑖 = 0
𝑛𝑖∗
𝑔𝑖
ln − 𝑎 = 𝛼 + 𝛽𝜀𝑖
𝑛𝑖∗
𝑔𝑖
− 𝑎 = 𝑒 𝛼+𝛽𝜀 𝑖
𝑛𝑖∗
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
176
 𝑔𝑖
= 𝑒 𝛼+𝛽 𝜀 𝑖 + 𝑎
𝑛𝑖∗
𝑛𝑖∗ 1
= ….(11)
𝑔𝑖 𝑒 𝛼 +𝛽 𝜀 𝑖 +𝑎
𝑔𝑖
𝑛𝑖 ∗ = ….(12)
𝑒 𝛼 +𝛽 𝜀 𝑖 +𝑎
𝑠 𝑔𝑖 𝑔𝑖 𝑛𝑖∗
≈ ln 𝑊 𝑛𝑖 ∗ = 𝑖 𝑛𝑖 ∗ ln ∗ −𝑎 − ln 1 − 𝑎
𝑘 𝑛𝑖 𝑎 𝑔𝑖
𝑔𝑖 𝑎
= 𝑖 𝑛𝑖 ∗ 𝛼 + 𝛽𝜀𝑖 − ln 1 −
𝑎 𝑒 𝛼 +𝛽 𝜀 𝑖 +𝑎

𝑔𝑖 𝑒 𝛼 +𝛽 𝜀 𝑖
= 𝑖 𝑛𝑖 ∗ 𝛼 + 𝛽𝜀𝑖 − ln
𝑎 𝑒 𝛼 +𝛽 𝜀 𝑖 +𝑎

𝑔𝑖 1
= 𝑖 𝑛𝑖 ∗ 𝛼 + 𝛽𝜀𝑖 − ln
𝑎 1+𝑎𝑒 − 𝛼 +𝛽 𝜀 𝑖
𝑔𝑖
= 𝑛𝑖 ∗ 𝛼 + 𝛽𝜀𝑖 + ln 1 + 𝑎𝑒 −𝛼−𝛽𝜀 𝑖 ….(13)
𝑎

 The first sum on RHS of (13) is αN while the second sum is βE. For the third
sum, we have
1 𝑆
𝑔𝑖 ln 1 + 𝑎𝑒 𝛼+𝛽 𝜀 𝑖 = 𝑘 − 𝛼𝑁 − 𝛽𝐸 ….(14)
𝑎
𝜇 1
𝛼 = − 𝑘𝑇 and 𝛽 = 𝑘𝑇

 The RHS of equation (14) is equal to,

𝑆 𝜇𝑁 𝐸 𝐺−(𝐸−𝑇𝑆) 𝑃𝑉
+ 𝑘𝑇 − 𝑘𝑇 = =
𝑘 𝑘𝑇 𝑘𝑇

 The thermodynamic pressure of the system is given by

𝑘𝑇
𝑃𝑉 = 𝑖 𝑔𝑖 ln 1 + 𝑎𝑒 −𝛼−𝛽𝜀 𝑖 ….(15)
𝑎
𝑃𝑉
= 𝑎 −1 𝑖 𝑔𝑖 ln 1 + 𝑎𝑒 −𝛼−𝛽 𝜀 𝑖 ….(16)
𝑘𝑇

 In M-B case (a→0), equation (15) takes the form

𝑃𝑉 = 𝑘𝑇 𝑔𝑖 𝑒 −𝛼−𝛽𝜀 𝑖
= 𝑘𝑇 𝑖 𝑛𝑖∗
= 𝑁𝑘𝑇 ….(17)
 Which is the familiar equation of state of the classical ideal gas. Equation (17)
for the M-B case holds irrespective of εi. The RHS of equation (16) is to be
identical to the q-potential of the ideal gas.
5.9 AN IDEAL GAS IN A QUANTUM MECHANICAL GRAND CANONICAL
ENSEMBLE:
 In canonical ensemble, the thermodynamics of a given system is derived from its
partition function

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

177
 −𝛽𝐸
𝑄𝑁 𝑉, 𝑇 = 𝐸𝑒 ….(1)
E → energy eigen values of the system.
and 𝛽 = 1/𝑘𝑇
 E can be expressed in terms of the single-particle energies ε. For instance
𝐸= 𝜀 𝑛𝜀 𝜀 ….(2)
𝑛𝜀 = number of particles in the single-particle energy
state ε.
Also 𝜀 𝑛𝜀 = 𝑁 ….(3)
 Now equation (1) can be written as,
′ −𝛽 𝜀 𝑛 𝜀 𝜀
𝑄𝑁 𝑉, 𝑇 = {𝑛 𝜀 } 𝑔{𝑛𝜀 }𝑒 ….(4)
𝑔{𝑛𝜀 } = statistical weight factor appropriate to the distribution set {𝑛𝜀 }
′→goes over all distribution sets that conform to the restrictive condition(3).
 The statistical weight factor is given by
𝑔𝐵−𝐸 {𝑛𝜀 } = 1 ….(5)
𝑔𝐹−𝐷 {𝑛𝜀 } = 1 if all 𝑛𝜀 =0 (or) 1.
= 0 otherwise ….(6)
and
1
𝑔𝑀−𝐵 {𝑛𝜀 } = 𝜀𝑛 ! ….(7)
𝜀

 Here we are dealing with single-particle states as individual states without

requiring them to be grouped into cells.
 Take M-B case and substituting equation (7) into (4) we get
′ 1 𝑛𝜀
𝑄𝑁 𝑉, 𝑇 = {𝑛 𝜀 } 𝜀𝑛 ! 𝜀 𝑒 −𝛽𝜀
𝜀

1 ′ 𝑁! 𝑛𝜀
= 𝑁! {𝑛 𝜀 } 𝜀 𝑒 −𝛽𝜀 ….(8)
𝜀 𝑛𝜀 !

 Evaluated with the help of the multinomial theorem

1 𝑁
𝑄𝑁 𝑉, 𝑇 = 𝑁! 𝜀 𝑒 −𝛽𝜀
1 𝑁
= 𝑁! 𝑄1 𝑉, 𝑇 ….(9)

 The number of single particle states with energies lying between ε and ε+dε is,
𝑄1 𝑉, 𝑇 = 𝜀 𝑒 −𝛽𝜀
2𝜋𝑉 3/2 ∞ −𝛽𝜀 1/2
≈ 2𝑚 𝑒 𝜀 𝑑𝜀 ….(10)
𝑕3 0

2𝜋𝑉 3/2 1 𝜋
= 2𝑚
𝑕3 2 𝛽3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

178
 2𝜋𝑚𝑘𝑇 3/2 𝑉
=𝑉 = 𝜆3
𝑕3
1/2
where 𝜆 = 𝑕/ 2𝜋𝑚𝑘𝑇
= mean thermal wavelength of the particles.
𝑉𝑁
 Hence, 𝑄𝑁 𝑉, 𝑇 = 𝑁!𝜆 3𝑁 ….(11)

 From which complete thermodynamics of this system can be derived.

 We obtain for the grand partition function of this system,
∞ 𝑁
Q 𝑧, 𝑉, 𝑇 = 𝑁=0 𝑍 𝑄𝑁 𝑉, 𝑇
∞ 𝑁 1 𝑉
𝑁 1 ∞ 𝑍𝑉 𝑁
= 𝑁=0 𝑧 𝑁! 𝜆 3𝑁 = 𝑁! 𝑁=0 𝜆 3

= exp 𝑧𝑉/𝜆3 ….(12)

 B-E and F-D cases:
′
𝑄𝑁 𝑉, 𝑇 = {𝑛 𝜀 } 𝑒 −𝛽 𝜀 𝑛𝜀 𝜀 ….(13)
 Now complete thermodynamics of this system can be derived.
 The Grand partition function Q turns out to be
∞ ′
𝐐 𝑧, 𝑉, 𝑇 = 𝑁=𝑜 𝑧𝑁 {𝑛 𝜀 } 𝑒 −𝛽 𝜀 𝑛𝜀 𝜀

∞ ′ 𝑛𝜀
= 𝑁=0 {𝑛 𝜀 } 𝜀 𝑧𝑒 −𝛽𝜀 ….(14)
 The double summation in (14) over the number 𝑛𝜀 constrained by a fixed value of
the total number N and then over all possible values of N, which is equivalent to a
summation over all possible values of 𝑛𝜀 independently of one another.
 Hence,
𝑛0 𝑛1
𝐐 𝑧, 𝑉, 𝑇 = 𝑛 0 ,𝑛 1 ,…… 𝑧𝑒 −𝛽 𝜀 0 𝑧𝑒 −𝛽𝜀 1 …
𝑛0 𝑛1
= 𝑛0 𝑧𝑒 −𝛽𝜀 0 𝑛1 𝑧𝑒 −𝛽𝜀 1 ….(15)
 In B-E case 𝑛𝜀 can be either 0 or 1 or 2 or ……
 In F-D case 𝑛𝜀 can be only 0 or 1.
1
𝐐 𝑧, 𝑉, 𝑇 = 𝜀 (1−𝑧𝑒 −𝛽𝜀 ) in B-E case with 𝑧𝑒 −𝛽𝜀 <1.

𝜀 (1 + 𝑧𝑒 −𝛽𝜀 ) in F-D case ….(16)

The q potential of the system is thus given by,
𝑃𝑉
𝑞 𝑧, 𝑉, 𝑇 = 𝑘𝑇 = 𝑙𝑛𝑄 𝑧, 𝑉, 𝑇

=∓ 𝜀 (1 ∓ 𝑧𝑒 −𝛽𝜀 )
ln⁡ ….(17)
with 𝑔𝑖 = 1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

179
  The identification of the fugacity z with the quantity 𝑒 −𝛼 is quite natural. 𝛼 =
𝜇
− 𝑘𝑇

 The upper sign in equation (17) corresponds to Bose case.

 The lower sign in equation (17) corresponds to Fermi case.
 Now in general,

𝑃𝑉 1
𝑞 𝑧, 𝑉, 𝑇 = 𝑘𝑇 = 𝑎 𝜀 (1 + 𝑎𝑧𝑒 −𝛽𝜀 )
ln⁡ ….(18)

 Where 𝑎 = −1, +1 𝑜𝑟 0 depending on the statistics governing the system.

 In classical case a → 0 gives,

𝑞𝑀−𝐵 = 𝑧 𝜀 𝑒 −𝛽𝜀 = 𝑧𝑄1 ….(19)

𝜕𝑞
𝑁≡𝑧 𝜕𝑧 𝑉,𝑇

1 𝑎𝑒 −𝛽𝜀
= 𝑧𝑎 1+𝑎𝑧 𝑒 −𝛽𝜀
𝑧
= 𝑒 𝛽𝜀 +𝑎𝑧
1
= ….(20)
𝑧 −1 𝑒 𝛽𝜀 +𝑎
𝜕𝑞
𝐸≡− 𝜕𝛽 𝑧,𝑉

1 𝑎𝑧 𝑒 −𝛽𝜀 (−𝜀)
= −𝑎 1+𝑎𝑧 𝑒 −𝛽𝜀
𝜀
= ….(21)
𝑧 −1 𝑒 𝛽𝜀 +𝑎

 At the same time, the mean occupation number 𝑛𝜀 of level ε turns out to be,
1 1 𝜕𝐐
𝑛𝜀 = 𝐐 − 𝛽 𝜕𝜀 𝑧,𝑇,𝑎𝑙𝑙 𝑜𝑡𝑕𝑒𝑟 𝜀

1 𝜕𝑞
= −𝛽 𝜕𝜀 𝑧,𝑇,𝑎𝑙𝑙 𝑜𝑡𝑕𝑒𝑟 𝜀

1 1 𝑎𝑧 𝑒 −𝛽𝜀 (−𝛽)
= −𝛽 𝑎 1+𝑎𝑧 𝑒 −𝛽𝜀

𝑧𝑒 −𝛽𝜀
= 1+𝑎𝑧 𝑒 −𝛽𝜀
1
= 𝑧 −1 𝑒 𝛽𝜀 +𝑎 ….(22)

Thus the mean value <n> and the most probable value n* of the occupation number n
of a single particle state are identical.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

180
5.10 STATISTICS OF THE OCCUPATION NUMBERS :
 The mean occupation number of a single particle with energy ε as an explicit
function of the quantity (  ) kT :
1
n  (   ) kT
….(1)
e a
 The functional behaviour of this number is shown in figure.

 In the Fermi-Dirac case (a= +1), the mean

occupation number never exceeds unity, for the
variable nε, itself cannot have a value other than 0
or 1.

 Moreover ε < μ and     kT , the mean

occupation number tends to its maximum possible

value 1.
 In the Bose-Einstein case (a= -1) we must have μ < all ε. When μ becomes equal
to the lowest value of ε (εo), the occupancy level becomes high which leads to
Bose-Einstein condensation.
 For μ < εo all (  ) are positive and the behaviour of all n is nonsingular.

Finally in Maxwell-Boltzmann case (a= 0), the mean occupation number takes the
form
n MB
 exp (  ) kT  exp(   kT ) ….(2)

 We note here that the distinction between the quantum statistics and the classical
statistics becomes imperceptible when for all values of ε ,
exp (  ) kT>>1. ….(3)

 Now equation (1) reduces to (2) and we may write n  1. ….(4)

 Condition (4) implies that the probability of any of the nε being greater than unity
is quite negligible.
 The distinction between the classical treatment and the quantum-mechanical
treatment then becomes rather insignificant.
 Correspondingly for large values of (  ) kT the quantum curves 1 and 2
essentially merge into the classical curve 3.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
181
  Condition (3) also implies that μ, the chemical potential of the system must be
negative and large in magnitude. This means that the fugacity z  exp( kT ) of

the system must be much smaller than unity.

N3
 This is further equivalent to  1 ….(5)
V
 Now we examine statistical fluctuations in the variable nε. We have

1  1   
2

n 2
    Q ….(6)
Q     

z ,T ,all other 

 It follows that
 1   2 
n  n   
2
2
  ln Q 
     z ,T ,all other 

 1   
    n  ….(7)
     z ,T
 For the relative mean-square fluctuation we obtain

n2  n
2
 1   1  1 
    z e ….(8)
n
2
     
n

 The actual value of this quantity will depend on the statistics of the particles
because for a given particle density (N/V) and a given temperature T, the value of
z will be different for different statistics.
 Equation (8) can be written in the form

n2  n
2
1
2
 a ….(9)
n n

 In the classical case (a= 0), the relative fluctuation is normal.

 In the Fermi-Dirac case it is given by 1 n  1 , which is below normal and tends

to vanish as n  1.

 In the Bose-Einstein case , the fluctuation is above normal .

 This result would apply to a gas of photons and hence to the oscillators states in
the black-body radiation.
 To understand the statistics of the occupation numbers, we evaluate the quantity
p (n) , the probability that there are exactly n particles in a state of energy ε.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
182
  We know that p (n)  ( ze ) n . On normalization, it becomes in the Bose-
Einstein case

p (n) B.E  ( ze ) n 1  ze 
n
 n  1 ( n ) n
    ….(10)
n
   1 
 n  1 n
  1
n1

 In the Fermi-Dirac case, we get


p (n) F .D  ( ze ) n 1  ze  1

= 1  n for n0

n for n 1 ….(11)

 In the Maxwell-Boltzmann case, we have p (n)  ( ze ) n / n!

 On normalization we get

( ze ) n / n! ( n )  n
n

p ( n) M . B   e ….(12)
exp( ze ) n!
 Equation (12) is clearly a Poisson distribution for which the mean square
deviation of the variable is equal to the mean value itself.
 It also resembles the distribution of the total particle number N in a grand
canonical ensemble consisting of ideal, classical systems.
 Here we seen that the ratio p (n) p (n  1) varies inversely with n, which is a
“normal” statistical behaviour of uncorrelated events.
 The distribution in the Bose-Einstein case is geometric with a common ratio

n n  1 .

 This means that the probability of a state ε acquiring one more particle for itself is
independent of the number of particles already occupying the state.
 In comparison with “normal” statistical behaviour, bosons exhibit a special
tendency of “bunching” together. That is a positive statistical correlation among
them. Fermions exhibit negative statistical correlation.
5.11 IDEAL BOSE-EINSTEIN GAS:

 Consider a perfect Bose-Einstein gas of n bosons.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
183
  These particles be divided into quantum groups such that there are 𝑛1 , 𝑛2 , … , 𝑛𝑖 , …
number of particles with energies 𝜀1 , 𝜀2 , … , 𝜀𝑖 , … respectively.
 𝑔𝑖 be the number of eigen states in the i th level.
 For the most probable distribution

𝑔𝑖 −𝜇 1
𝑛𝑖 = here 𝛼 = and 𝛽 = 𝑘𝑇
𝑒 𝛼 +𝛽 𝜀 𝑖 −1 𝑘𝑇

𝑔𝑖
= 1 𝛽𝜀 where, 𝐴 = 𝑒 −𝛼
𝑒 𝑖 −1
𝐴

 Since the number of particles cannot be negative, we must always have

1
𝑛𝑖 ≥ 0 𝑒 𝛽𝜀 𝑖 ≥ 0.
𝐴

 The constant α can be determined by the condition

𝑔𝑖
𝑛= 𝑛𝑖 = 𝛼 +𝛽 𝜀 𝑖 −1
𝑒
𝑔𝑖
= 1 𝛽𝜀 . … (1)
𝑒 𝑖 −1
𝐴

 Since the particles in a box are normal size and the translational levels are closely
spaced and hence now the summation is replaced by integration.
 The number of particles states 𝑔 𝑝 𝑑𝑝 between momentum p and p+dp is given
by
4𝜋𝑝 2 𝑑𝑝 4𝜋𝑉 𝑝 2 𝑑𝑝
= 𝑔𝑠 = 𝑔𝑠
𝑕 3 /𝑉 𝑕3

 𝑔𝑠 = degenaracy factor = 1

4𝜋𝑉 𝑝 2 𝑑𝑝
Now 𝑔 𝑝 𝑑𝑝 = ….(2)
𝑕3

 Then equation (1) can be rewritten as

𝑔 𝑝 𝑑𝑝
𝑛 𝑝 𝑑𝑝 =
𝑒 𝛼 +𝛽 𝜀 𝑖 −1

4𝜋𝑉 𝑝 2 𝑑𝑝 1
= . … (3)
𝑕3 𝑒 𝛼 +𝛽 𝜀 𝑖 −1

𝑝2
𝜀 = 𝜀𝑖 = 2𝑚

𝑝2 = 2𝑚𝜀

2𝑝𝑑𝑝 = 2𝑚𝑑𝜀

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

184
 𝑚
𝑑𝑝 = 𝑑𝜀
𝑝

𝑚
= 𝑑𝜀
2𝑚𝜀

𝑚 1/2
= 𝑑𝜀
2𝜀

 Now equation (3) becomes

4𝜋𝑉 𝑚 1/2 1
𝑛 𝜀 𝑑𝜀 = (2𝑚𝜀) 𝑑𝜀 𝑒 𝛼 +𝜀/𝑘𝑇 −1
𝑕3 2𝜀

4𝜋𝑚𝑉 1/2 𝑑𝜀
= 2𝑚𝜀 . … (4)
𝑕3 𝑒 𝛼 +𝜀/𝑘𝑇 −1

𝜀
=𝑥
𝑘𝑇

𝜀 = 𝑘𝑇𝑥

𝑑𝜀 = 𝑘𝑇𝑑𝑥

4𝜋𝑚𝑉 1/2 𝑘𝑇𝑑𝑥

𝑛 𝜀 𝑑𝜀 = 2𝑚𝑘𝑇𝑥
𝑕3 𝑒 𝛼 +𝑥 −1

𝑉 1/2 𝑑𝑥
3/2 2 𝑥
= 𝑕 3 2𝜋𝑚𝑘𝑇 . … (5)
𝜋 𝑒 +𝑥 −1
𝛼

 The total number of particles is given by

∞ 𝑉 1/2 𝑑𝑥
3/2 2 𝑥
𝑛= 0 𝑕3
2𝜋𝑚𝑘𝑇 𝜋 𝑒 +𝑥 −1
𝛼

1/2 𝑑𝑥
𝑉 3/2 2 ∞ 𝑥
= 𝑕 3 2𝜋𝑚𝑘𝑇 𝜋 0 𝑒 +𝑥 −1
𝛼

𝑉 3/2
𝑛 = 𝑕 3 2𝜋𝑚𝑘𝑇 𝑓1 𝛼 . … (6)

2 ∞ 𝑥 1/2 𝑑𝑥
where 𝑓1 𝛼 = . … (7)
𝜋 0 𝑒 𝛼 +𝑥 −1

 The total energy is given by

∞
𝐸= 0
𝜀 𝑛 𝜀 𝑑𝜀

∞ 𝑉 1/2 𝑑𝑥
3/2 2 𝑥
= 𝑘𝑇𝑥 𝑕 3 2𝜋𝑚𝑘𝑇
0 𝜋 𝑒 +𝑥 −1
𝛼

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

185
 𝑉 3/2 2 ∞ 𝑥 3/2 𝑑𝑥
= 𝑕 3 𝑘𝑇 2𝜋𝑚𝑘𝑇 0 𝑒 𝛼 +𝑥 −1
𝜋

3 𝑉 3/2 4 ∞ 𝑥 3/2 𝑑𝑥
= 2 𝑕 3 𝑘𝑇 2𝜋𝑚𝑘𝑇 3 𝜋 0 𝑒 𝛼 +𝑥 −1

3 𝑉 3/2
= 2 𝑕 3 𝑘𝑇 2𝜋𝑚𝑘𝑇 𝑓2 𝛼 . … (8)

4 ∞ 𝑥 3/2 𝑑𝑥
𝑓2 𝛼 = 3 0 𝑒 𝛼 +𝑥 −1
. … (9)
𝜋

 For 𝐴 < 1, 𝑓1 𝛼 𝑎𝑛𝑑 𝑓2 𝛼 may be evaluated as follows:

2 ∞ 𝑥 1/2 𝑑𝑥
𝑓1 𝛼 = 𝜋 0 𝑒 𝛼 +𝑥 −1

2 ∞ 𝑥 1/2 𝑑𝑥
= 𝑥
𝜋 0 𝑒 −1
𝐴

2 ∞ 1/2 𝑒𝑥 −1
= 𝑥 −1 𝑑𝑥
𝜋 0 𝐴

2 ∞ 1/2
= 𝑥 𝐴𝑒 −𝑥 1 − 𝐴𝑒 −𝑥 −1
𝑑𝑥
𝜋 0

2 ∞ 1/2
= 𝑥 𝐴𝑒 −𝑥 1 + 𝐴𝑒 −𝑥 + 𝐴2 𝑒 −2𝑥 + ⋯ 𝑑𝑥
𝜋 0

2 ∞ 1/2 ∞ 1/2 2 −2𝑥

= 𝑜
𝑥 𝐴𝑒 −𝑥 𝑑𝑥 + 0
𝑥 𝐴 𝑒 𝑑𝑥 +⋯
𝜋

2 𝜋 𝐴2 𝐴3
= 𝐴+ + + ⋯..
𝜋 2 23/2 33/2

𝐴2 𝐴3
= 𝐴 + 23/2 + 33/2 + ⋯ … … …

∞ 𝐴𝑟
𝑓1 𝛼 = 𝑟=1 𝑟 3/2 . … (10)

4 ∞ 𝑥 3/2 𝑑𝑥
 Similarly 𝑓2 𝛼 = 3 −(𝛼 +𝑥 ) −1
𝜋 0 𝑒

𝐴2 𝐴3
= 𝐴 + 25/2 + 35/2 + ⋯

∞ 𝐴𝑟
𝑓2 𝛼 = 𝑟=1 𝑟 5/2 . … (11)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

186
 𝑉 3/2
Now 𝑛 = 𝑕 3 2𝜋𝑚𝑘𝑇 𝑓1 (𝛼)

𝑉 3/2 𝐴2 𝐴3
= 𝑕 3 2𝜋𝑚𝑘𝑇 𝐴 + 23/2 + 33/2 + ⋯ . … (12)

3 𝑉 3/2
𝐸 = 2 𝑘𝑇 2𝜋𝑚𝑘𝑇 𝑓2 (𝛼).
𝑕3

3 𝑉 3/2 𝐴2 𝐴3
𝐸 = 2 𝑘𝑇 2𝜋𝑚𝑘𝑇 𝐴 + 25/2 + 35/2 + ⋯ . … 13
𝑕3

−1
𝐸 3 𝐴2 𝐴3 𝐴2 𝐴3
= 2 𝑘𝑇 𝐴 + 25/2 + 35/2 + ⋯ 𝐴 + 23/2 + 33/2 + ⋯
𝑛

−1
3 𝐴 𝐴2 𝐴 𝐴2
= 2 𝑘𝑇 1 + 25/2 + 35/2 + ⋯ 1 + 23/2 + 33/2 + ⋯

3 𝐴 𝐴2
= 2 𝑘𝑇 1 − 25/2 + 35/2 + ⋯

3 𝐴 𝐴2
𝐸 = 2 𝑛𝑘𝑇 1 − 25/2 + 35/2 + ⋯ . … (14)

 The value of α (or) A can be determined by equation (6) as

𝑉 3/2
𝑛 = 𝑕 3 2𝜋𝑚𝑘𝑇 𝑓1 (𝛼)

𝑛 𝑕3
𝑓1 𝛼 = 𝑉 2𝜋𝑚𝑘𝑇 3/2

 Here 𝑓1 (𝛼) is directly proportional to particle density n/v and inversely

proportional to temperature as 𝑇 3/2 .
 For 𝐴 << 1, 𝑓1 𝛼 = 𝐴

𝑛 𝑕3
𝐴=𝑉 . … (15)
2𝜋𝑚𝑘𝑇 3/2

 Obviously A would be small for high temperatures (low density).

M-B distribution as a limiting case of B-E distribution:

 For 𝐴 << 1, 𝑒 𝛼 +𝛽𝜀𝑖 becomes very large compared to 1.

𝑔𝑖 𝑔𝑖
 Now B-E distribution 𝑛𝑖 = becomes 𝑛𝑖 = which is well known M-
𝑒 𝛼 +𝛽 𝜀 𝑖 −1 𝑒 𝛼 +𝛽 𝜀 𝑖

B distribution.
 For 𝐴 << 1, 𝑓1 𝛼 = 𝑓2 𝛼 = 𝐴
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
187
 𝑉 3/2
 Equation (12) becomes 𝑛 = 𝑕 3 2𝜋𝑚𝑘𝑇 ×𝐴
3 𝑉 3/2
 Equation (13) becomes 𝐸 = 2 𝑘𝑇 2𝜋𝑚𝑘𝑇 ×𝐴
𝑕3

𝐸 3 3
= 2 𝑘𝑇 ⇒ 𝐸 = 2 𝑛𝑘𝑇
𝑛

 Which is well known expression for energy in M-B statistics.

5.12 BOSE-EINSTEIN CONDENSATION AND ITS CRITICAL CONDITIONS:

 The degeneracy parameter A is given by,

𝑛 2𝜋𝑚𝑘𝑇 3/2
𝐴 = 𝑒 −𝛼 = 𝑉 .…(1)
𝑕3
𝑉 3/2
 And 𝑛 = 𝑕 3 2𝜋𝑚𝑘𝑇 𝑓1 𝛼 .…(2)

 If the particle density is increased or the temperature is decreased then A

increases.
 Now the behaviour of perfect gas departs from the classical perfect gas.
 This is due to the fact that the velocities of the particles are subjected to quantum
statistics.
 The gas under this condition is said to be degenerate gas and A is called the
degeneracy parameter.
 ′A′ contains three variables n/V - the particle density , m - mass of each boson
and
T -temperature of the gas.
𝑛/𝑉
 The degeneracy criterion will be based on the magnitude
𝑚𝑇 1/2

 Thus the degree of degeneracy will be large when T is low, n/V is large and m is
small.
 For low energy values A=1 and α=0.
1 1
 Then 𝑓1 (𝛼) 𝑚𝑎𝑥 = 𝑓1 0 = 1 + 23/2 + 33/2 + ⋯

= 2.612
𝑛 2𝜋𝜋𝑚𝑘𝑇 3/2
 Now = 2.612 . … (4)
𝑉 𝑚𝑎𝑥 𝑕3

 Equation (4) corresponds to the limiting case of Bose-Einstein degeneration.

 The solution of equation (2) can exist for

𝑛 2𝜋𝑚𝑘𝑇 3/2
> 2.612 . … (5)
𝑉 𝑕3
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
188
  n/v can be alternatively expressed in terms of critical temperature 𝑇𝑜 defined as
𝑛 2𝜋𝑚𝑘 𝑇0 3/2
= 2.612 . … (6)
𝑉 𝑕3

3/2 𝑛 1
2𝜋𝑚𝑘𝑇0 = 𝑕3 𝑉 2.612

𝑛 1 2/3
2𝜋𝑚𝑘𝑇0 = 𝑕2 𝑉 2.612

𝑕2 𝑛 1 2/3
𝑇0 = 2𝜋𝑚𝑘 𝑉 2.612

 T0 is the lower temperature for which a solution of equation (2) is possible.

 There is no solution for 𝑇 < 𝑇0
 i.e. the degeneracy starts at T0.
 A graph is drawn between the energy E and temperature
T of the gas.
 Why there is no solution 𝑇 < 𝑇0
 Because we have assumed continuous distribution and
replaced the summation by integration.
 But at low temperature, the number of particles begin to
crowd into lower energy levels.
 Hence a large number of particles may occupy the
ground state 𝜀0 = 0.
 The number of particles between the energy range 𝜀 and 𝜀 + 𝑑𝜀 is given by,

𝑔 𝜀 𝑑𝜀
𝑛 𝜀 𝑑𝜀 = 𝑒 𝛼 +𝜀/𝑘𝑇 −1 . … (8)

4𝜋𝑚𝑣 1/2
And 𝑔 𝜀 𝑑𝜀 = 2𝑚𝜀 𝑑𝜀 . … (9)
𝑕3

 For ground state actually 𝜀0 = 0 𝑎𝑛𝑑𝑔 𝜀 = 1 𝑏𝑢𝑡 𝑛𝑜𝑡 𝑔 𝜀 = 0

 For 𝜀 ≠ 0 and 𝑔(𝜀) ≠ 0, the distribution given by equation (8) is correct.
 But for 𝜀 = 0, 𝑔 𝜀 = 0 the law gives incorrect result.
𝑔𝑖
 For a single state , 𝑛𝑖 =
𝑒 𝛼 +𝛽 𝜀 𝑖 −1

𝜀𝑖 = 𝜀0 = 0

𝑔𝑖 = 𝑔 𝜀𝑖 = 1
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
189
 1
𝑛 = 𝑛0 = 𝑒 𝛼 −1 . … (10)

 This gives the number of particles in the ground state.

 Now the total number of particles for the degenerate gas,
𝑛 = 𝑛0 + 𝑛 𝜀 𝑑𝜀
∞ 4𝜋𝑚𝑉 1/2 𝜀 1/2 𝑑𝜀
= 𝑛0 + 2𝑚
0 𝑕3 𝑒 𝛼 +𝜀/𝑘𝑇 −1

= 𝑛0 + 𝑛′ . … (11)

4𝜋𝑚𝑉 ∞ 𝜀 1/2 𝑑𝜀
𝑛′ = 2𝑚 1/2
….(12)
𝑕3 0 𝑒 𝛼 +𝜀/𝑘𝑇 −1

𝑉
𝑛′ = 𝑕 3 2𝜋𝑚𝑘𝑇 3/2
𝑓1 𝛼 . . . . (13)

𝑛 2𝜋𝑚𝑘 𝑇0 3/2
 From equation (6) = 2.612
𝑉 𝑕3

𝑉 3/2 1
2𝜋𝑚𝑘𝑇0 =𝑛
𝑕3 2.612

𝑇 3/2 𝑓1 (𝛼)
 Now 𝑛′ = 𝑛 .…(14)
𝑇0 2.612

 As 𝑓1 𝛼 < 𝑓1 0 , 𝑛′ acquires its maximum value when 𝛼 = 0.

 Hence 𝑓1 𝛼 = 2.612 for maximum value.
𝑇 3/2
𝑛′ = 𝑛 𝑓𝑜𝑟 𝑇 < 𝑇0
𝑇0

𝑛 = 𝑛0 + 𝑛′
𝑇 3/2
𝑛 = 𝑛0 + 𝑛 𝑇0

𝑇 3/2
𝑛𝑜 = 𝑛 − 𝑛 𝑇0

𝑇 3/2
𝑛0 = 𝑛 1 − 𝑓𝑜𝑟 𝑇 < 𝑇0 . … (15)
𝑇0

 𝑛0 gives the number of particles condensed in the ground state.

 When the temperature is lowered below T0, the number of particles in the ground
state rapidly increases.
 This rapid increase in the population of the ground state
below the critical temperature 𝑇0 for a Bose - Einstein gas is
called Bose – Einstein condensation.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

190
  Obviously T0 depends on the particle density n/V.
 Equation (15) is plotted in figure which represents the fraction of
particles condensed in the ground state 𝑓𝑜𝑟 𝑇 < 𝑇0 .
 At the ground state 𝜀 = 0, the particles condensed in the ground state do not
contribute to the energy.
 For above To, 𝛼 ≠ 0 there is negligible number of particles in the ground state
and the gas is said to be classical or non - degenerate gas.
 Example: For Helium T0 can be calculated to have the value 3.12 K.
 Therefore the degeneration and condensation of Helium must start at 3.12 K.
 But experimental observation shows that the condensation of Helium starts at 2.19
K.
 i.e. the lambda point transition observed in liquid helium at 2.19 K is essentially a
Bose-Einstein condensation.

5.13 BOSE – EINSTEIN CONDENSATION IN ULTRACOLD ATOMIC GASES :

 The first demonstration of Bose- Einstein condensation in ultracold atomic gases
came in 1995.
 Since 1995, many isotopes have been Bose condensed including 7Li, 23Na, 41
K,
52
Cr, 84Sr, 85Rb, 87Rb, 133Cs and 174Yb.
 The first molecular Bose –Einstein condensates were created in 2003 by the
research groups of Rudoif Grimm at the University of Innsbruck.
 The first step of the cooling of the atomic vapour uses three sets of counter-
propagating laser beams oriented along cartesian axes that are tuned just below the
resonant frequency of the atoms in the trap.
 Atoms that are stationary are just off resonance and so rarely absorb a photon.
 Moving atoms are Doppler shifted on resonance to the laser beam that is
propagating opposite to the velocity vector of the atom.
 Those atoms preferentially absorb photons from that direction and then reemit in
random directions, resulting in a net momentum kick opposite to the direction of
motion.
 This results in an "optical molasses" that slows the atoms.
 This cooling method is constrained by the "recoil limit" in which the atoms have a
minimum momentum of the order of the momentum of the photons used to cool
the gas.
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
191
  This gives a limiting temperature of (hf)2/2mc2k  lµK, where  is the frequency of
the spectral line used for cooling and m is the mass of an atom.
 In the next step of the cooling process, the lasers are turned off and a spatially
varying magnetic field creates an attractive anisotropic harmonic oscillator
potential near the center of the magnetic trap
1

V (r )  m 12 x 2  22 y 2  32 z 2
2
….(1)

 The frequencies of the trap  are controlled by the applied magnetic field. One
can then lower the trap barrier using a resonant transition to remove the highest
energy atoms in the trap. If the atoms in the vapor are sufficiently coupled to one
other, then the remaining atoms in the trap are cooled by evaporation.
 If the interactions between the atoms in the gas can be neglected, the energy of
each atom in the harmonic oscillator potential is

 l1 ,l2 ,l3  1l1  2l2  3l3  1  2  3 

1
.…(2)
2
 where l (= 0,1,2,.. .∞) are the quantum numbers of the harmonic oscillator. If the
three frequencies are all the same, then the quantum degeneracy of a level with
energy   l  3 / 2 is l  1l  2 / 2
 For the general anisotropic case, the smoothed density of states as a function of
energy (suppressing the zero point energy and assuming    ) is given by

2
a        1l1  2l2  3l3  dl1dl2 dl3  … (3)
20 
3
0 0 0

 where 0  12 3 1/ 3 ; this assumes a single spin state per atom. The

thermodynamic potential  for bosons in the trap is then given by

, T   
kT 
4 
x ln 1  e e  dx 
kT 
4
g z 
20  0
2  x 
… (4)
3
0 3 4
 where z = exp () is the fugacity. Volume is not a parameter in the
thermodynamic potential since the atoms are confined by the harmonic trap. The
average number of atoms in the excited states in the trap is

     kT 
N T        g 3 z  .… (5)
  T  0 

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

192
  For fixed N, the chemical potential monotonically increases as temperature is
lowered until Bose-Einstein condensation occurs when = 0 (z = 1). The critical
temperature for N trapped atoms is then given by
1/ 3
kTc  N 
  .… (6)
0  3 

 where (3) = g3(1)  1.202. While the spacing of the energy levels is of order 0 ,

the critical temperature for condensation is much larger than the energy spacing of
the lowest levels for N>> 1.
 For T <Tc the number of atoms in the excited states is
3 3
N excited (3)  kT   T 
     .… (7)
N N  0   Tc 

 so the fraction of atoms that condense into the ground state of the harmonic
oscillator is
3
N0 T 
 1    ; .… (8)
N  Tc 
(a) Detection of the Bose-Einstein condensate:
 The linear size of the ground state wave function in Cartesian direction  is


a  , .…(9)
m

 while the linear size of the thermal distribution of the noncondensed atoms in that
direction is

kT kT
athermal   a .…(10)
m2


 At trap frequency f= 100 Hz and temperature T =100 nK, these sizes are about 1
m and 5m, respectively.
 Instead of measuring the atoms directly in the trapping potential, experimenters
usually measure the momentum distribution of the ultracold gas by a time-of-
flight experiment.
 At time t = 0, the magnetic field is turned off suddenly, eliminating the trapping
potential.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

193
  The atomic cloud then expands according to the momentum distribution the atoms
had in the harmonic trap. The cloud is allowed to expand for about 100
milliseconds.
 The speed of the atoms at this temperature is a few millimeters per second, so the
cloud expands to a few hundred microns in this period of time.
 The cloud is then illuminated with a laser pulse on resonance with the atoms,
leaving a shadow on a CCD in the image plane of the optics.
 The size and shape of the light intensity pattern directly measures the momentum
distribution the atoms had in the trap at t = 0.
 The expanding cloud can be divided into two components, the No atoms that had
been Bose-condensed into the ground state and the remaining N -No atoms that
were in the excited states of the harmonic oscillator potential.
 The Bose-condensed atoms have smaller momenta than the atoms that were in the
excited states.
 After time t, the quantum evolution of the ground state has a spatial number
density

N0 3    r2 
n0 r , t   N o  0 r , t  
1
    .…(11)
2

3 / 2 1  a 1  2 t 2

exp
  
 a 2 1  2 t 2
  

 The atoms that are not condensed into the ground state can be treated semi
classically, that is, the position-momentum distribution function is treated
classically while the density follows the Bose-Einstein distribution function:
1
f (r , p,0)  ….(12)
 p m 2 2 
 
2
exp   1 x  22 y 2  32 z 2  u   1
 2m 2 
 After the potential is turned off at t = 0, the distribution evolves ballistically:
 pt 
f (r , p, 0)  f  r  , p, 0  ….(13)
 m 
 The spatial number density of atoms in the excited states is
1  
nexcited r , t  
pt
3  
f r  , p, t  dp ….(14)
h  m 
 which can be integrated to give

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

194
 1  ej  3    jm2 r2  
nexcited r , t  
1
  
3 j 1 j 3 / 2  1  1  2 t 2
exp 
 
 
2 1  2 t 2  
….(15)
   

 where = h / 2mkT is the thermal deBroglie wavelength. The integrals over the
condensed state and the excited states correctly count all the atoms:

N 0   no r , t  dr , ….(16a)

N  N o   nexcited (r , t )dr  N excited . ….(16b)

 Note that at early times t 1 both the condensed and the excited distributions
are anisotropic due to the anisotropic trapping potential.
 However, at late times t 1 , the atoms from the excited states form a
spherically symmetric cloud because of the isotropic momentum dependence of
the t = 0 distribution function.
 By contrast, the atoms that were condensed into the ground state expand
anisotropically due to the different spatial extents of the ground state
wavefunction at t = 0.
 The direction that has the largest 0 is quantum mechanically squeezed the most at
t = 0; so, according to the uncertainty principle, it expands the fastest.
 This is an important feature of the experimental data that confirms the onset of
Bose-Einstein condensation.
(b) Thermodynamic properties of the Bose-Einstein condensate :
 The temperature, condensate fraction, and internal energy can all be observed
using time- of-flight measurements. The internal energy can also be written in
terms of the function gv(z):

U , T   

3 1
d  3
kT  g z 
4

….(17)
0 20  0 3 4
3    
e 1

 The heat capacity at constant number can be written as

 U   U   U    
C N T           
 T  N  T    T  T  N

 U   N 
   
 U   T   T 
  
 T   N 
 
  T ….(18)
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
195
  Equations (5) and (6) can be used to determine the fugacity z numerically, as
shown in Figure (a). The fugacity can then be used in equation (17) to obtain the
scaled internal energy. [Figure (b)]

  T  4 4
 3   for T  Tc ,
 T 3
   3 4
U
NkTc   T  g z 
3 T  3 for T  Tc;
4

  c ….(19)
 The scaled specific heat is given by
 124  T 3
   for T  Tc ,
C N  3  Tc 

Nk  1  T  9 g 2 z  
 12 g 4 z   3  for T  Tc ,
 3  Tc  g 2 z  
 .(20)
and shown in figure (c).

 Unlike the case of Bose-Einstein condensation of free particles in a box the

specific heat of a condensate in a harmonic trap displays a discontinuity at the
critical temperature.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

196
 124 

CN  3  10.805 as T  Tc
 
Nk 124 9 3
   4.228 as T  Tc


 3 2 ..(21)

 Equation (15) is called the Virial equation of state.
 This equation can also be written in terms of the pair correlation function and is
also used in computer simulations to determine the pressure of the system.
5.14 IDEAL FERMI-DIRAC GAS:

 Consider a perfect Fermi-Dirac gas of n Fermions.

 These particles be divided into quantum groups such that there are
𝑛1 , 𝑛2 , … , 𝑛𝑖 , … number of particles with energy is 𝜀1 , 𝜀2 , … , 𝜀𝑖 , . .. respectively.
 𝑔𝑖 be the number of eigen states in the i th level.
 For the most probable distribution

𝑔𝑖 −𝜇 1
𝑛𝑖 = here 𝛼 = and 𝛽 = 𝑘𝑇
𝑒 𝛼 +𝛽 𝜀 𝑖 +1 𝑘𝑇

𝑔𝑖
= 1 𝛽𝜀 where, 𝐴 = 𝑒 −𝛼
𝑒 𝑖 +1
𝐴

𝛼 may be positive or negative.

 The constant α can be determined by the condition

𝑔𝑖
𝑛= 𝑛𝑖 =
𝑒 𝛼 +𝛽 𝜀 𝑖 +1
𝑔𝑖
= 1 𝛽𝜀 . … (1)
𝑒 𝑖 +1
𝐴

 Since the particles in a box are normal size and the translational levels are closely
spaced and hence now the summation is replaced by integration.
 The number of particle states 𝑔 𝑝 𝑑𝑝 between momentum p and p+dp is given by
4𝜋𝑝 2 𝑑𝑝 4𝜋𝑉 𝑝 2 𝑑𝑝
= 𝑔𝑠 = 𝑔𝑠
𝑕 3 /𝑉 𝑕3

 𝑔𝑠 = degenaracy factor = (2𝑠 + 1)

4𝜋𝑉 𝑝 2 𝑑𝑝
Now 𝑔 𝑝 𝑑𝑝 = 𝑔𝑠 ….(2)
𝑕3

 Then equation (1) can be rewritten as

𝑔 𝑝 𝑑𝑝
𝑛 𝑝 𝑑𝑝 =
𝑒 𝛼 +𝛽 𝜀 𝑖 +1

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

197
 4𝜋𝑉 𝑝 2 𝑑𝑝 1
= 𝑔𝑠 . … (3)
𝑕3 𝑒 𝛼 +𝛽 𝜀 𝑖 +1

𝑝2
𝜀 = 𝜀𝑖 = 2𝑚

𝑝2 = 2𝑚𝜀

2𝑝𝑑𝑝 = 2𝑚𝑑𝜀

𝑚
𝑑𝑝 = 𝑑𝜀
𝑝

𝑚
= 𝑑𝜀
2𝑚𝜀

𝑚 1/2
= 𝑑𝜀
2𝜀

 Now equation (3) becomes

4𝜋𝑉 𝑚 1/2 1
𝑛 𝜀 𝑑𝜀 = 𝑔𝑠 (2𝑚𝜀) 𝑑𝜀
𝑕3 2𝜀 𝑒 𝛼 +𝜀/𝑘𝑇 +1

4𝜋𝑚𝑉 1/2 𝑑𝜀
= 𝑔𝑠 2𝑚𝜀 . … (4)
𝑕3 𝑒 𝛼 +𝜀/𝑘𝑇 +1

𝜀
=𝑥
𝑘𝑇

𝜀 = 𝑘𝑇𝑥

𝑑𝜀 = 𝑘𝑇𝑑𝑥

4𝜋𝑚𝑉 1/2 𝑘𝑇𝑑𝑥

𝑛 𝜀 𝑑𝜀 = 𝑔𝑠 2𝑚𝑘𝑇𝑥
𝑕3 𝑒 𝛼 +𝑥 +1

𝑉 1/2 𝑑𝑥
3/2 2 𝑥
= 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 . … (5)
𝜋 𝑒 𝛼 +𝑥 +1

 The total number of particles is given by

∞ 𝑉 1/2 𝑑𝑥
3/2 2 𝑥
𝑛= 𝑔
0 𝑠 𝑕3
2𝜋𝑚𝑘𝑇 𝜋 𝑒 +𝑥 +1
𝛼

1/2 𝑑𝑥
𝑉 3/2 2 ∞ 𝑥
= 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 𝜋 0 𝑒 𝛼 +𝑥 +1

𝑉 3/2
𝑛 = 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 𝑓1 𝛼 . … (6)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

198
 2 ∞ 𝑥 1/2 𝑑𝑥
where 𝑓1 𝛼 = . … (7)
𝜋 0 𝑒 𝛼 +𝑥 +1

 The total energy is given by

∞
𝐸= 0
𝜀 𝑛 𝜀 𝑑𝜀

∞ 𝑉 1/2 𝑑𝑥
3/2 2 𝑥
= 𝑘𝑇𝑥 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇
0 𝜋 𝑒 𝛼 +𝑥 +1

𝑉 3/2 2 ∞ 𝑥 3/2 𝑑𝑥
= 𝑔𝑠 3
𝑘𝑇 2𝜋𝑚𝑘𝑇 0 𝑒 𝛼 +𝑥 +1
𝑕 𝜋

3 𝑉 3/2 4 ∞ 𝑥 3/2 𝑑𝑥
= 2 𝑔𝑠 𝑕 3 𝑘𝑇 2𝜋𝑚𝑘𝑇 3 𝜋 0 𝑒 𝛼 +𝑥 +1

3 𝑉 3/2
= 2 𝑔𝑠 𝑕 3 𝑘𝑇 2𝜋𝑚𝑘𝑇 𝑓2 𝛼 . … (8)

4 ∞ 𝑥 3/2 𝑑𝑥
𝑓2 𝛼 = 3 0 𝑒 𝛼 +𝑥 +1
. … (9)
𝜋

 𝑓1 𝛼 and 𝑓2 𝛼 must be evaluated for both positive and negative values of α.

 Now we introduce the Fermi-Dirac distribution function 𝑓(𝜀) defined by,

𝑛 𝜀
𝑓 𝜀 =𝑔 𝜀

1
= 𝑒 𝛼 +𝜀/𝑘𝑇 +1

1
= 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1 . … (10)

𝜇 𝜀
 Where 𝛼 = − 𝑘𝑇 = − 𝑘𝑇𝐹

 At T=0 K, 𝑓 𝜀 = 1 𝑓𝑜𝑟 𝜀 < 𝜀𝐹 0

= 0 𝑓𝑜𝑟 𝜀 > 𝜀𝐹 0
1
= 2 𝑓𝑜𝑟 𝜀 = 𝜀𝐹

 This is shown in the figure.

 𝜀𝐹 is determined by the condition that the total
number of particles is constant at a given
temperature T.
 The number of fermions in the energy range between 𝜀 and 𝜀 + 𝑑𝜀 is given by,

𝑛 𝜀 𝑑𝜀 = 𝑓 𝜀 𝑔 𝜀 𝑑𝜀
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
199
 4𝜋𝑚𝑉 1/2 𝜀 1/2 𝑑𝜀
= 𝑔𝑠 2𝑚 (𝜀−𝜀 𝐹)/𝑘𝑇 . . . . (11)
𝑕3 𝑒 +1

 At absolute zero, all states with 0 < 𝜀 < 𝜀𝐹 (0) are completely filled and all states
with 𝜀 > 𝜀𝐹 (0) are empty.
 𝜀𝐹 at T=0 K i.e. 𝜀𝐹 (0) is determined by,

𝜀 𝐹 (0)
𝑛= 0
𝑛 𝜀 𝑑𝜀

𝜀 𝐹 (0)
= 0
𝑓(𝜀)𝑔 𝜀 𝑑𝜀

𝜀 𝐹 (0)
= 0
𝑔 𝜀 𝑑𝜀

𝜀 𝐹 (0) 4𝜋𝑚𝑉 1/2 1/2

= 0
𝑔𝑠 𝑕 3 2𝑚 𝜀 𝑑𝜀

4𝜋𝑚𝑉 1/2 𝜀 𝐹 (0) 1/2

= 𝑔𝑠 2𝑚 𝜀 𝑑𝜀
𝑕3 0

4𝜋𝑚𝑉 1/2 2 3/2

𝑛 = 𝑔𝑠 2𝑚 𝜀𝐹 0
𝑕3 3

3/2 3𝑛 𝑕3
𝜀𝐹 0 = 4𝜋𝑉 𝑔
𝑠 2𝑚 3/2

𝑕2 3𝑛 2/3
𝜀𝐹 0 = 2𝑚 … . (12)
4𝜋𝑉 𝑔𝑠

 This gives the Fermi-energy at T=0K.

 For particles with spin equal to 1/2, 𝑔𝑠 =2 ie, one particle with spin up and another
with spin down.
 Now we define Fermi temperature 𝑇𝐹 as,

𝜀 𝐹 (0)
𝑇𝐹 = 𝑘

𝑕2 3𝑛 2/3
= 2𝑚𝑘 4𝜋𝑉 𝑔𝑠

𝑚𝑛 𝑛 𝜌
𝜌= ⇒𝑉=𝑚
𝑉

𝑕2 3𝜌 2/3
𝑇𝐹 = 2𝑚𝑘 4𝜋𝑚 𝑔𝑠

𝑕2 3𝜌 2/3
𝑇𝐹 = 2𝑚 5/3 𝑘 … . (13)
4𝜋𝑔𝑠
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.
200
  𝑇 << 𝑇𝐹 strong degeneracy α is negative
 𝑇 > 𝑇𝐹 weak degeneracy α is positive
 𝑇 >> 𝑇𝐹 non degeneracy α is positive
(A) Degeneracy:
i. Weak degeneracy: (High temperature and low density)
𝑇 > 𝑇𝐹 ; 𝜀𝐹 is negative ; α is positive and hence 𝐴 < 1.
For 𝐴 < 1, we can write
1 1 −1
1 = 𝑒𝑥 + 1
𝑒 𝑥 +1 𝐴
𝐴

= 𝐴𝑒 −𝑥 1 + 𝐴𝑒 −𝑥 −1

= 𝐴𝑒 −𝑥 1 − 𝐴𝑒 −𝑥 + 𝐴2 𝑒 −2𝑥 + ⋯
2 ∞ 𝑥 1/2 𝑑𝑥
𝑓1 𝛼 = 𝜋 0 𝑒 𝛼 +𝑥 +1

2 ∞ 𝑥 1/2 𝑑𝑥
= 𝜋 0 1 𝑒 𝑥 +1
𝐴

2 ∞ 1/2
= 𝑥 𝑑𝑥 𝐴𝑒 −𝑥 1 − 𝐴𝑒 −𝑥 + 𝐴2 𝑒 −2𝑥 + ⋯
𝜋 0

2 ∞ 1/2 −𝑥 ∞ 1/2 ∞ 1/2 −3𝑥

= 𝐴 0
𝑥 𝑒 𝑑𝑥 − 𝐴2 0
𝑥 𝑒 −2𝑥 𝑑𝑥 + 𝐴3 0
𝑥 𝑒 𝑑𝑥 +⋯
𝜋
𝐴2 𝐴3
= 𝐴 − 23/2 + 33/2 + ⋯ . … (1)
4 ∞ 𝑥 3/2 𝑑𝑥
 Similarly 𝑓2 𝛼 = 3 0 1 𝑥
𝜋 𝑒 +1
𝐴

𝐴2 𝐴3
= 𝐴 − 25/2 + 35/2 + ⋯ . … (2)
𝑉 3/2
 Now 𝑛 = 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 𝑓1 𝛼
𝑉 3/2 𝐴2 𝐴3
= 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 𝐴 − 23/2 + 33/2 + ⋯ … . (3)
3 𝑣 3/2
𝐸 = 2 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 𝑘𝑇 𝑓2 𝛼
3 𝑣 3/2 𝐴2 𝐴3
= 2 𝑔𝑠 𝑕 3 2𝜋𝑚𝑘𝑇 𝐴 − 25/2 + 35/2 + ⋯ … . (4)
𝐸 3 𝐴2 𝐴3 𝐴2 𝐴3 −1
= 2 𝑘𝑇 𝐴 − 25/2 + 35/2 + ⋯ 𝐴 − 23/2 + 33/2 + ⋯
𝑛
3 𝐴 𝐴2
= 2 𝑘𝑇 1 + 25/2 − 35/2 + ⋯ … . (5)

 To find first approximation we can get,

𝐸 3
= 2 𝑘𝑇
𝑛

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

201
 3
𝐸 = 2 𝑛𝑘𝑇 … . (6)

 which is well known relation for E in classical statistics.

 Comparison of equations (5) and (6) shows that the ideal Fermi-Dirac gas deviates
from a classical perfect gas and this deviation is known as degeneracy.
 A is the degeneracy function and greater the value of A, more will be the
degeneracy.
ii. Strong degeneracy: (low temperature and high density)
 Here we discuss the degeneracy case in two temperature ranges.
 (a) At absolute zero(T=0 K)
 (b) Above absolute zero
(a) At absolute zero(T=0 K):
 In this case the Fermi-dirac gas is completely degenerate.
1
 At T=0 K, 𝑓 𝜀 = 1 𝜀/𝑘𝑇
𝑒 +1
𝐴

1
= 1 𝜀−𝜀 /𝑘𝑇
𝑒 𝐹 +1
𝐴

=1 𝑓𝑜𝑟 0 ≤ 𝜀 ≤ 𝜀𝐹 (0)

=0 𝑓𝑜𝑟 𝜀 > 𝜀𝐹

4𝜋𝑚𝑉 1/2 1/2

 So that 𝑛 𝜀 𝑑𝜀 = 𝑔𝑠 2𝑚 𝜀 𝑑𝜀 𝑓𝑜𝑟 0 ≤ 𝜀 ≤ 𝜀𝐹 (0)
𝑕3

=0 𝑓𝑜𝑟 𝜀 > 𝜀𝐹 (0)

 The total internal energy of Fermi-Dirac gas at T=0 K gives zero point energy.

𝜀 𝐹 (0)
𝐸0 = 0
𝜀 𝑛 𝜀 𝑑𝜀

𝜀 𝐹 (0) 4𝜋𝑚𝑉 1/2 1/2

= 𝑔𝑠 2𝑚 𝜀 𝑑𝜀
0 𝑕3

4𝜋𝑚𝑉 1/2 𝜀 𝐹 (0) 3/2

= 𝑔𝑠 2𝑚 𝜀 𝑑𝜀
𝑕3 0

4𝜋𝑚𝑉 5/2
1/2 𝜀 𝐹 (0)
= 𝑔𝑠 2𝑚
𝑕3 5/2

2 4𝜋𝑚𝑉 1/2 5/2

= 5 𝑔𝑠 2𝑚 𝜀𝐹 (0)
𝑕3

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

202
 2 4𝜋𝑚𝑉 𝑕2 3𝑛 2/3 5/2
1/2
= 5 𝑔𝑠 2𝑚
𝑕3 2𝑚 4𝜋𝑣𝑔𝑠

3𝑛 4𝜋𝑉 𝑔𝑠 2𝑚 3/2 𝑕5 3𝑛 5/3

= 5 3𝑛 𝑕2 2𝑚 5/2 4𝜋𝑣𝑔𝑠

3𝑛 𝑕 2 3𝑛 2/3
= 5 2𝑚 4𝜋𝑉 𝑔𝑠

3𝑛
𝐸0 = 𝜀𝐹 0 … . (7)
5

2 𝐸0
 Zero point pressure 𝑝0 = 3 𝑉

2 1 3𝑛
=3 𝜀𝐹 (0)
𝑉 5

2 𝑛
𝑝0 = 5 𝜀𝐹 0 … . (8)
𝑉

 Thus a strongly degenerate Fermi-Dirac gas possesses energy and pressure even
at absolute zero.
(b) Above absolute zero: 𝑨 >> 1 𝑎𝑛𝑑 𝑇 << 𝑻𝑭
 The Fermi gas is strongly degenerate and 𝜀 is still positive.
 The number of particles in the energy range between 𝜀 and 𝜀 + 𝑑𝜀 is given by,

1
1
4𝜋𝑚𝑉 𝜀 2 𝑑𝜀
𝑛 𝜀 𝑑𝜀 = 𝑔𝑠 2𝑚 2 … . (9)
𝑕3 𝑒 𝜀−𝜀 𝐹 /𝑘𝑇 +1

1/2
4𝜋𝑉 𝑔𝑠 2𝑚 3 𝜀 1/2 𝑑𝜀
= 3𝑛
3𝑛 𝑕6 𝑒 𝜀−𝜀 𝐹 /𝑘𝑇 +1

 We have Fermi energy

𝑕2 3𝑛 2/3
𝜀𝐹 0 = 2𝑚 4𝜋𝑣 𝑔𝑠

3𝑛 2/3 2𝑚 𝜀 𝐹 (0)
=
4𝜋𝑣 𝑔𝑠 𝑕2

3𝑛 2𝑚 𝜀 𝐹 (0) 3/2
=
4𝜋𝑣 𝑔𝑠 𝑕2

4𝜋𝑣 𝑔𝑠 2𝑚 𝜀 𝐹 (0) −3/2

=
3𝑛 𝑕2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

203
 3/2
𝑕2
= 2𝑚 𝜀 𝐹 (0)

3/2 1/2
𝑕2 2𝑚 3 𝜀 1/2 𝑑𝜀
𝑛 𝜀 𝑑𝜀 = 3𝑛 2𝑚 𝜀 𝐹 (0) 𝑕6 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1

3 𝑛 𝜀 1/2 𝑑𝜀
= … . (10)
2 𝜀 𝐹 (0) 3/2 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1

∞
𝑛= 0
𝑛 𝜀 𝑑𝜀

3 𝑛 ∞ 𝜀 1/2 𝑑𝜀
=2 … . (11)
𝜀 𝐹 (0) 3/2 0 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1

∞
𝐸= 0
𝜀 𝑛 𝜀 𝑑𝜀

3 𝑛 ∞ 𝜀 1/2 𝑑𝜀
= 3/2 0 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1
… . (12)
2 𝜀 𝐹 (0)

 To solve the integrals in equation (11) and (12), we consider the general integral

∞ 𝜑 𝜀 𝑑𝜀
𝐼= 0 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1
… . (13)

 Here, 𝜑(𝜀) is a function of 𝜀 such that 𝜑 𝜀 = 0 𝑎𝑡 𝜀 = 0.

 By Taylor series expansion,

∞ 𝜑 𝜀 𝑑𝜀 𝜀𝐹 𝜋𝑘𝑇 2 7
0 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1
= 0
𝜑 𝜀 𝑑𝜀 + 𝜑′ 𝜀=𝜀 𝐹 + 360 𝜋𝑘𝑇 4
𝜑′′′ 𝜀=𝜀 𝐹 +⋯
6

Take 𝜑 𝜀 = 𝜀 1/2

1
𝜑′ 𝜀=𝜀 𝐹
= 2 𝜀𝐹 −1/2

1
𝜑 ′′ 𝜀=𝜀 𝐹
= − 4 𝜀𝐹 −3/2

3
𝜑 ′′′ 𝜀=𝜀 𝐹
= 8 𝜀𝐹 −5/2

3 𝑛 ∞ 𝜀 1/2 𝑑𝜀
𝑛=2 3/2 0 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1
𝜀 𝐹 (0)

3 𝑛 2 𝜋𝑘𝑇 2 1 7 43
=2 3/2
𝜀𝐹 3/2 + 𝜀𝐹 −1/2 + 360 𝜋𝑘𝑇 𝜀𝐹 −5/2 + ⋯
𝜀 𝐹 (0) 3 6 2 8

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

204
 𝜀𝐹 3/2 1 𝜋𝑘𝑇 2 7 𝜋𝑘𝑇 4
𝑛=𝑛 1+8 + 640 + ⋯……
𝜀 𝐹 (0) 𝜀𝐹 𝜀𝐹

𝜀𝐹 3/2 1 𝜋𝑘𝑇 2 7 𝜋𝑘𝑇 4

= 1+8 + 640 + ⋯……
𝜀 𝐹 (0) 𝜀𝐹 𝜀𝐹

 To the first approximation we get ,

𝜀𝐹 −3/2 1 𝜋𝑘𝑇 2
= 1+8
𝜀 𝐹 (0) 𝜀𝐹

−2/3
𝜀𝐹 1 𝜋𝑘𝑇 2
= 1+8
𝜀 𝐹 (0) 𝜀𝐹

1 𝜋𝑘𝑇 2
= 1 − 12 … . (14)
𝜀𝐹

1 𝜋𝑘𝑇 2
𝜀𝐹 = 𝜀𝐹 0 1 − 12 … . (15)
𝜀𝐹

 By applying Crude approximation we get,

1 𝜋𝑘𝑇 2
𝜀𝐹 = 𝜀𝐹 0 1 − 12 … . (16)
𝜀𝐹 0

Take 𝜑 𝜀 = 𝜀 3/2

𝜀𝐹 2
0
𝜀 3/2 𝑑𝜀 = 5 𝜀𝐹 5/2

3
𝜑 ′ (𝜀) 𝜀=𝜀 𝐹
= 2 𝜀𝐹 1/2

3
𝜑 ′′ (𝜀) 𝜀=𝜀 𝐹
= 4 𝜀𝐹 −1/2

3
𝜑 ′′′ (𝜀) 𝜀=𝜀 𝐹
= − 8 𝜀𝐹 −3/2

3 𝑛 ∞ 𝜀 3/2 𝑑𝜀
𝐸=2 3/2 0 𝑒 (𝜀−𝜀 𝐹 )/𝑘𝑇 +1
𝜀 𝐹 (0)

3 𝑛 2 𝜋𝑘𝑇 2 3 7 43
= 𝜀𝐹 5/2 + 𝜀𝐹 1/2 − 𝜋𝑘𝑇 𝜀𝐹 −3/2 + ⋯
2 𝜀 𝐹 (0) 3/2 5 6 2 360 8

3 𝜀 𝐹 5/2 5 𝜋𝑘𝑇 2 7 𝜋𝑘𝑇 4

=5 𝑛 3/2
1+8 − 384 +⋯
𝜀 𝐹 (0) 𝜀𝐹 𝜀𝐹

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

205
 3 𝜀𝐹 5/2 5 𝜋𝑘𝑇 2 7 𝜋𝑘𝑇 4
= 5 𝑛𝜀𝐹 0 1+8 − 384 +⋯
𝜀 𝐹 (0) 𝜀𝐹 𝜀𝐹

 To the first approximation we get,

3 𝜀𝐹 5/2 5 𝜋𝑘𝑇 2
𝐸 = 5 𝑛𝜀𝐹 0 1+8 … . (17)
𝜀 𝐹 (0) 𝜀𝐹

 By applying Crude approximation we get,

3 𝜀𝐹 5/2 5 𝜋𝑘𝑇 2
𝐸 = 5 𝑛𝜀𝐹 0 1+8 … . (18)
𝜀 𝐹 (0) 𝜀𝐹 0

 Using equation (16) we get,

3 1 𝜋𝑘𝑇 2 5/2 5 𝜋𝑘𝑇 2

= 5 𝑛𝜀𝐹 0 1 − 12 1+8
𝜀 𝐹 (0) 𝜀 𝐹 (0)

3 5 𝜋𝑘𝑇 2 5 𝜋𝑘𝑇 2
= 5 𝑛𝜀𝐹 0 1 − 24 1+8
𝜀 𝐹 (0) 𝜀 𝐹 (0)

3 5 𝜋𝑘𝑇 2 5 𝜋𝑘𝑇 2
= 5 𝑛𝜀𝐹 0 1 − 24 +8
𝜀 𝐹 (0) 𝜀 𝐹 (0)

3 5 𝜋𝑘𝑇 2
𝐸 = 5 𝑛𝜀𝐹 0 1 + 12 … . 19
𝜀𝐹 0

 The corresponding pressure is

2 𝐸
𝑝= 3 𝑉

2 𝑛𝜀 𝐹 0 5 𝜋𝑘𝑇 2
=5 1 + 12 (20)
𝑉 𝜀𝐹 0

 Equations (19) and (20) give the approximate energy and pressure of a strongly
degenerate Fermi gas and also known as equation of state of an ideal Fermi gas.

Prepared by
Dr. L. RAVISANKAR
Assistant Professor of Physics
Sri KGS Arts College, Srivaikuntam – 628 619

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

206