Statistical Physics HS 14

Solution Sheet 3 Prof. M. Sigrist

Exercise 3.1 Non-interacting Particles in the Gravitational Field

a) The canonical partition function Z is given by
Z
1
Z(N, T ) = 3N d3N qd3N p e−βH(p,q) , β = 1/kB T. (1)
h N!
With the Hamiltonian of the system
N
X
p2i /2m + mgzi

H(p, q) =
i=1

(note that we have set ~qi = (xi , yi , zi )) one obtains

Z
1 2
Z(N, T ) = 3N Πi=1 d3 qi d3 pi e−β(pi /2m+mgzi )
N
h N!
Z ∞ Z
1 −βmgzi 2
= 3N N
Πi=1 πR 2
dzi e d3 pi e−βpi /2m
h N! 0
N
1 2πm 3N/2 πR2
  
= . (2)
N ! h2 β βmg
This leads to the Helmholtz free energy
F (N, T ) = −kB T log Z(N, T )
" #
2πm 3/2 πR2

= −N kB T log (kB T )5/2 + N kB T log N − N kB T, (3)
h2 mg

using Stirling’s approximation log N ! ≈ N log N −N . Consequently, one finds the following
expression for the entropy,
 
∂F
S(N, T ) = −
∂T N
" #
2πm 3/2 πR2

5
= N kB + N kB log (kB T )5/2 − N kB log N + N kB . (4)
2 h2 mg

Finally, the internal energy is given by

∂ log Z 5
U (N, T ) = − = N kB T. (5)
∂β 2

b) We first note that the local density for just one particle hδ(zi − z)i is independent of the
phase-space coordinates of the other N −1 particles and uniform in the planar coordinates,
such that πR2 n(z) = N hδ(zi − z)i for any i. We have
Z
1 1
hδ(zi − z)i = d3N q̃d3N p̃ δ(zi − z)e−βH
Z h3N N !
1
=R e−βmgz
dzi e−βmgzi
= βmge−βmgz , (6)

1
 such that
βmg −βmgz
n(z) = N e . (7)
πR2
The local pressure p(z) exerted by the particles above z is obtained by adding up the force
f (z) = −∇Vgrav (z) = −mg and we find
3
R Z ∞
qz >z d q n(qz )|f (qz )|
Z
mg mg −βmgz
p(z) = = dqx dqy dqz n(qz ) = N e . (8)
πR2 πR2 z πR 2

Obviously, the (local) thermal equation of state holds,

p(z) = n(z)kB T. (9)

For the local energy density we find

 2 
3 δ(qz − z)
Z Z
N 1 3(N −1) 3(N −1) p
u(z) = d q̃d 3
p̃ d pd q + mgqz e−βH
Z h3N N ! πR2 2m
Z  2 
N Z p 2
= 3
d p + mgz e−β(p /2m+mgz) (10)
Z d3 pd3 q e−β(p2 /2m+mgqz )
R
2m
 3/2 Z  2 
β βmg p 2
=N 2
3
d p + mgz e−β(p /2m+mgz)
2πm πR 2m
 
βmg 3
=N kB T + mgz e−βmgz , (11)
πR2 2
such that d3 q u(qz ) = U . We find
R
 
3
u(z) = kB T + mgz n(z). (12)
2
The first term corresponds to a local caloric equation of state while the second term is the
potential energy.
c) i) Using equation (4) we obtain
   2 
∂S ∂ F 5
C=T = −T 2
= N kB . (13)
∂T N ∂T N 2

ii) We first rewrite the Hamiltonian as

N
( ! )
X X piµ ∂H ∂H
H(p, q) = + zi , (14)
µ
2 ∂piµ ∂zi
i=1

where i enumerates particles and µ runs over the coordinates. The equipartition law
is given by    
∂H ∂H
qiµ = δij δµν kB T = piµ , (15)
∂qjν ∂pjν
and with U = hHi we find
N
(    )
∂hHi ∂ X 1X ∂H ∂H
C= = piµ + zi
∂T ∂T 2 µ ∂piµ ∂zi
i=1
 
∂ 3 3
= N kB T + kB T = N kB + N kB . (16)
∂T 2 2

2
 The final result is the same as in i), but we are now able to identify the different
contributions: The first term stands for the kinetic contribution to the internal ener-
gy whereas the second term is due to the potential energy of the particles in the
gravitational field.
iii) Defining the specific volume via the local particle density, v(z) = N/n(z), the thermal
expansion coefficient takes the form
 
1 ∂v(z)
α(z) =
v(z) ∂T p(z)
 
−1 ∂n(z)
=
n(z) ∂T p(z)
−1 −p(z) 1
= 2
= , (17)
n(z) kB T T
where the thermal equation of state equation (9) has been used for the last step. With
the relation given on the exercise sheet we thus finds for the specific heat at constant
pressure
3
cp (z) = n(z)kB + n(z)kB = cV + n(z)kB . (18)
2
We see that a non-interacting particle ensemble subjected to an external force cor-
responds to the situation of an ideal gas with locally constant external pressure in a
specific volume. We can only make a statement about local variables, while in a free
ideal gas this relation holds globally. The C we calculated in i) and ii) can neither be
identified with a Cp or a CV but is a CN .
iv) If the particles are non-interacting and the potential energy only depends on the
spatial coordinates,
XN
H(p, q) = [Hkin (pi ) + Hpot (qi )],
i=1
the evaluation of the variance of H simplifies substantially,
N n
X
(∆H)2 = hH2 i − hHi2 =


[Hkin (pi ) + Hpot (qi )][Hkin (pj ) + Hpot (qj )]
i,j=1


o
− [Hkin (pi ) + Hpot (qi )] [Hkin (pj ) + Hpot (qj )]
N n
X


= Hkin (pi )Hkin (pj ) + Hpot (qi )Hpot (qj )
i,j=1




o
− Hkin (pi ) Hkin (pj ) − Hpot (qi ) Hpot (qj )

= (∆Hkin )2 + (∆Hpot )2 . (19)

It follows directly that the heat capacity C = (∆H)2 /kB T 2 separates in a similar
fashion, C = Ckin + Cpot . The kinetic contribution is identical to the case of an ideal
gas, and reads Ckin = 23 N kB .
The variance of the potential energy contribution is (∆Hpot )2 = N (mg)2 (∆z)2 , since
Hpot (q) = mgz. The total potential energy is given by N mghzi, where
πR2 ∞
Z
1
hzi = dz zn(z) = (20)
N 0 βmg

3
 is the mean height of the particles. Derivation with respect to temperature gives
∂
hzi = −kB β 2 {hzihHpot i − hHpot zi} = −kB β 2 hzimghzi − mghz 2 i

∂T
kB β 2 Cpot
= kB β 2 mg(∆z)2 = (∆Hpot )2 = , (21)
N mg N mg
such that
∂T N mghzi = Cpot .
The potential contribution to the heat capacity can thus be calculated in a similar
fashion as C = ∂T U for the total potential energy N mghzi. It follows directly that
3
C = Ckin + Cpot = N kB + N kB , (22)
2
as expected.

Exercise 3.2 Classical Ideal Paramagnet II

a) The partition function has the usual product form
N Z  Z N  N
Y
βmHcosθi βmH cos θi 4π N
Z= dΩi e = 2π 1d cos θi e = sinh(βmH) ≡ Zm .
−1 βmH
i=1
(23)
The link to thermodynamics is obtained via the Helmholtz free energy and the internal
energy,
F (T, H, N ) = −kB T log Z = N kB T log(βmH) − N kB T log(4π sinh(βmH)), (24)
∂
U (T, H, N ) = − log Z = N kB T − N mH coth(βmH)). (25)
∂β
From this we directly obtain
   2
∂U mH
CH = = N kB − N kB . (26)
∂T H kB T sinh(mH/kB T ))

In the zero temperature limit we have

C H → N kB , as T → 0, (27)
which is non-vanishing. It corresponds to the contribution of the potential energy of a
two-dimensional harmonic oscillator (the fluctuating magnetic moments nearly aligned
with the magnetic field feel a harmonic potential).
b) Because the partition function has a product form, the magnetization can be written as
Z
1
hmi
~ = dΩ m(φ,
~ θ)eβmH cos θ , (28)
Zm
leading to hmx i = hmy i = 0 and

2πm 1
Z  
mH kB T
hmz i = d cos θ cos θ eβmH cos θ = m coth − . (29)
Zm −1 kB T H
It is clear from equation (24) that the above expression stasfies the usual thermodynamical
relation.

4
b) The mean square magnetization hm2z i can be obtained in a similar fashion

2πm 1
Z  
2 2 βmH cos θ 2kB T kB T
hmz i = d cos θ cos θ e =− m coth(mH/kB T ) − + m2 .
Zm −1 H H
(30)
Deriving hmz i with respect to Hz , we directly get the magnetic susceptibility
" #
kB T 2
 2 
N m2

∂ F 1
χzz = − = − , (31)
∂Hz2 T,N kB T mH sinh2 (mH/kB T )

such that we finally obtain the fluctuation-dissipation relation

kB T
(∆mz )2 = hm2z i − hmz i2 = χzz . (32)
N
We see that the fluctuations of the magnetization are connected to its response function,
the magnetic susceptibility. This is a special case of the more general fluctuation-dissipation
theorem.