Sie sind auf Seite 1von 129

PH4211 Statistical Mechanics

Problem Sheet 1
1.1 Demonstrate that entropy, as given by the Boltzmann expression S = k ln Ω , is an
extensive property. The best way to do this is to argue clearly that Ω is multiplicative.

1.2 Demonstrate that gravitational energy is not extensive: show that the gravitational
energy of a sphere of radius r and uniform density varies with volume as Vn and find
the exponent n.

1.3 In investigating the conditions for the establishment of equilibrium through the
transfer of thermal energy the fundamental requirement is that the entropy of the
equilibrium state should be a maximum. Equality of temperature was established
from the vanishing of the first derivative of S. What follows from a consideration of
the second derivative?

1.4 Do particles flow from high µ to low µ or vice versa? Explain your reasoning.

1.5 In the derivation of the Boltzmann factor the entropy of the bath was expanded in
powers of the energy of the ‘system of interest’. The higher order terms of the
expansion are neglected. Discuss the validity of this.

1.6 The Boltzmann factor could have been derived by expanding Ω rather than by
expanding S. In that case, however, the expansion cannot be terminated. Why not?

1.7 Show that ln N ! = ∑n =ln


N
0
n . By approximating this sum by an integral obtain
Stirling’s approximation: ln N ! ≈ N ln N − N .

1.8 Show that the Gibbs expression for entropy: S = − k ∑ j Pj ln Pj , reduces to the
Boltzmann expression S = k ln Ω in the case of an isolated system.

1.9 What is the condition that the geometric progression in the derivation of the
Bose-Einstein distribution is convergent?

1.10 Show that the trajectory of a 1d harmonic oscillator is an ellipse in phase space.
What would the trajectory be if the oscillator were weakly damped.

1.11 Why can’t the evolutionary curve in phase space intersect? You need to
demonstrate that the evolution from a point is unique.

1.12 Starting from the expression for the Gibbs factor for a many-particle system,
write down the grand partition function Ξ and show how it may be expressed as the
product of Ξk, the grand partition function for the subsystem comprising particles in
the kth single-particle state.

PH4211 © B. Cowan 2005 page P1-1 Problem Sheet 1


1.13 (Difficult) This problem considers the probability distribution for the energy
fluctuations in the canonical ensemble. The moments of the energy fluctuations are
defined by
σn = ∑(Ej −ε ) e j
1 n βE

Z j
where β = − 1 kT and ε is an arbitrary (at this stage) energy.
Show that
∂n
n {
Zσ n = e βε Z e − βε }
∂β
and use this to prove that the fluctuations obey a normal distribution around ε. (You
really need to use a computer algebra system to do this problem.)

1.14 For a single-component system with a variable number of particles, the Gibbs
free energy is a function of temperature, pressure and number of particles:
G = G (T , p, N ) . Since N is the only extensive variable upon which G depends, show
that the chemical potential for this system is equal to the Gibbs free energy per
particle: G = Nµ.

1.15 Use the definition of the Gibbs free energy together with the result of the
previous question to obtain the Euler relation of Appendix 1.

1.16 The energy of a harmonic oscillator may be written as mω 2 x 2 2 + p 2 2m so it is


quadratic in both position and momentum – thus, classically, equipartition should
apply. The energy levels of the quantum harmonic oscillator are given by
ε n = ( 12 + n ) =ω . Show that the partition function of this system is given by
1 =ω
Z= cosech
2 2kT
and that the internal energy is given by
1 =ω =ω =ω
E = =ω coth = =ω kT + .
2 2kT e −1 2
Show that at high temperatures E may be expanded as
= 2ω 2
E = kT + +"
12kT
Identify the terms in this expansion.

PH4211 © B. Cowan 2005 page P1-2 Problem Sheet 1


PH4211 Statistical Mechanics

Problem Sheet 2
2.1 In Section 2.1 we saw that the density of free-particle states for a three
dimensional volume V was shown to be
1 V
g (ε ) = 2 3(
2m ) ε 1 2 .
32


This followed from counting the number of states in the octant of radius
R = nx2 + n y2 + nz2 .
By similar arguments show that in two dimensions, by counting the number of states
in the quadrant of radius
R = nx2 + n y2 ,
the density of states is given by
mA
g (ε ) =
2π 2
where A is the area. Note in two dimensions the density of states is independent of
energy.

And similarly, show that in one dimension the density of states is


12
L ⎛ m ⎞ −1 2
g (ε ) = ε .
π ⎜⎝ 2 ⎟⎠

2.2 In Sections 2.3.1 and 2.3.2 the ideal gas partition function was calculated
quantum-mechanically and classically. Although the calculations were quite different,
they both resulted in (different) Gaussian integrals. By writing the Gaussian integral
of the classical case as
∞ ∞ ∞
(
− x2 + y 2 + z 2 )
∫ dx ∫ dy ∫ dz e
−∞ −∞ −∞

and transforming to spherical polar coordinates, you can perform the integration over
θ and ϕ trivially. Show that the remaining integral can be reduced to that of the
quantum case.

2.3 The Sakur-Tetrode equation, discussed in Section 2.3.3,


3
S = Nk ln V − Nk ln N + Nk ln T + Nks0
2
is often interpreted as indicating different contributions to the entropy: the volume
contribution is in the first term, the number contribution in the second term and the
temperature contribution in the third term. Show that such an identification is
fallacious, by demonstrating that the various contributions depend on the choice of
units adopted – even though the total sum is independent. Discuss the origin of the
fallacy.

PH4211 © B. Cowan 2005 page P2-1 Problem Sheet 2


2.4 Show that the Fermi energy for a two-dimensional gas of Fermions is
2π 2 N
εF =
αm A
where A is the area of the system.

2.5 Show that the chemical potential of a two-dimensional gas of fermions may be
expressed analytically as
µ = kT ln {eε F kT − 1}

2.6 Calculate the low temperature chemical potential of a two-dimensional gas of


fermions by the Sommerfeld expansion method of Section 2.4.3. Observe that the
temperature series expansion terminates. Compare this result with the exact result of
the previous question. Discuss the difference between the two results.

2.7 The general formula for the Fermi integrals In of Section 2.4.3 was quoted as

ex
In = ∫ x n dx
−∞ ( e + 1)
x 2

= ( 2 − 22− n ) ζ ( n ) n !
Derive this result. (You might find the discussion in Landau and Lifshitz, Statistical
Physics, helpful.)

2.8 Obtain the chemical potential µ, the internal energy E and the heat capacity Cv for
system with general density of states g(ε) as in Section 2.4.4. I.e. show that these are
given in terms of the behaviour of the density of states at the Fermi surface.

2.9 Consider the Bose gas at low temperatures. You saw in Section 2.5.2 and 2.5.3
that when the occupation of the ground state is appreciable then the chemical potential
µ is very small and it may be ignored, compared with ε in the integral for the number
of excited states.

Show that when the ground state occupation N0 is appreciable then µ may be
approximated by
µ ~ − kT N 0 .
Now consider the more stringent requirement that µ may be neglected in comparison
with ε in the integral for the number of excited states. This will be satisfied if µ is
much less than the energy ε1 of the first excited state. The expression for ε1 is
π2 2
ε1 ~ .
2mV 2 3
Where does this expression come from?

Show that the condition µ ε1 is satisfied when N 1015 (approximately) when


considering 1cm3 of 4He (molar volume 27cm3) at a temperature of about 1K.

Thus show that the expression

PH4211 © B. Cowan 2005 page P2-2 Problem Sheet 2


⎧⎪ ⎛ T ⎞3 2 ⎫⎪
N 0 = N ⎨1 − ⎜ ⎟ ⎬
⎩⎪ ⎝ c ⎠ ⎭⎪
T
is then valid to temperatures below Tc right up to within ∼ 10−8Tc of the critical
temperature.

2.10 Liquid 4He has a molar volume at saturated vapour pressure of 27cm3. Treating
the liquid as an ideal gas of bosons, find the temperature at which Bose-Einstein
condensation will occur. How will this temperature change as the pressure on the fluid
is increased?

2.11 The superfluid transition temperature of liquid helium decreases with increasing
pressure. Very approximately ∂Tc ∂p ~ −0.015 K bar −1 . How does this compare with
the behaviour predicted from the Bose-Einstein condensation?

2.12 Show that below the transition temperature the entropy of a Bose gas is given by
32
5 I3 2 ⎛ T ⎞
S = Nk ⎜ ⎟ .
3 I1 2 ⎝ Tc ⎠
Since the number of excited particles is given by
32
⎛T ⎞
N ex = N ⎜ ⎟ ,
⎝ Tc ⎠
show that the entropy per excited particle is given by
S 5 I3 2
= k ≈ 1.28k .
N ex 3 I1 2
Discuss the connection between this result and the two fluid model of superfluid 4He.

2.13 Show that the Bose-Einstein transition temperature of a gas of bosons and the
Fermi temperature for a gas of ‘similar’ fermions are of comparable magnitude.
Discuss why this should be.

2.14 In Section 2.6 we studied a paramagnetic solid: a collection of essentially


distinguishable magnetic moments. If we were to consider a (classical) gas of
indistinguishable magnetic moments, how would the partition function be modified?
What would the observable consequences of this modification?

2.15 Show, using arguments similar to those in Section 2.1.3, that the energy levels of
an ultra-relativistic or a massless particle with energy-momentum relation E = cp are
given by

ε = 1 3 ( nx2 + n y2 + nz2 ) .
12

V
Hence show that the pressure of a gas of such particles is one third of the (internal)
energy density.

2.16 Evaluate the Fermi temperature for liquid 3He, assuming it to be a Fermi “gas”.
Its molar volume is 36 cm3. Calculate the de Broglie wavelength at T = TF and
show that it is comparable to the interparticle spacing as expected.

PH4211 © B. Cowan 2005 page P2-3 Problem Sheet 2


2.17 In Problem 2.1 we found the expression for the energy density of states g(ε) for
a gas of fermions confined to two dimensions and we saw that it was independent of
energy. What surface density of electrons is necessary in order that TF = 100 mK?
Show that, for a given area, the heat capacity is independent of the number of
electrons.

2.18 Use the Sommerfeld expansion method of Section 2.4.3 to show that the Fermi-
Dirac distribution function may be approximated, at low temperatures, by
1 π2
Θ ( µ − ε ) − ( kT ) δ ′ (ε − µ ) + …
2
( ε − µ ) kT
~
e +1 6
where Θ is the unit step function and δ ′ is the first derivative of the Dirac delta
function.

Can you write down the general term of the series?

PH4211 © B. Cowan 2005 page P2-4 Problem Sheet 2


PH4211 Statistical Mechanics

Problem Sheet 3
3.1 Show that the Joule-Kelvin coefficient is zero for an ideal gas.

3.2 Derive the second virial coefficient expression for the Joule-Kelvin coefficient

µJ =
RS b g
2T dB2 T
− 2
B T b g UV
5k TdT T W
and find the inversion temperature for the square well potential gas.
a 2a
3.3 For the van der Waals gas show that TB = and TI = .
bk bk
3.4 Show that for a van der Waals fluid the critical parameters are given by
a 8a
Vc = 3 Nb, pc = 2
, kTc = . Express these critical quantities in terms of the
27b 27b
microscopic interaction (Lennard-Jones) parameters ε and σ.

3.5 Show that, for a van der Waals fluid, the combination of critical parameters
PcVc/NkTc takes the universal value 3/8.

3.6 Show that for the Dieterici fluid the critical parameters are given by
a a
Vc = 2 Nb, pc = 2 2 , kTc = , and the universal combination PcVc/NkTc has the
4b e 4b
value 2/e2 = 0.271.

3.7 Show that the partition function for an interacting gas may be expressed as
− ∑U ( qi ,q j ) kT
1
Z = Z id N ∫ e i< j d3N q
V
where Zid is the partition function for a non-interacting gas. In terms of this expression
explain why the partition function of a hard sphere gas might be approximated by.
⎛ V − Nb ⎞
N

Z = Z id ⎜ ⎟ .
⎝ V ⎠

3.8 Show that the approximate partition function for the hard sphere gas in the
previous question leads to the equation of state p (V − Nb ) = NkT . This is sometimes
called the Clausius equation of state. Give a physical interpretation of this equation.

Show that the first few virial coefficients are given by B2 (T ) = b, B3 (T ) = b 2 ,


B4 (T ) = b3 , etc. These virial coefficients are independent of temperature. Discuss
whether this is a fundamental property of the hard sphere gas, or whether it is simply
a consequence of the approximated partition function.

PH4211 © B. Cowan 2005 page P3-1 Problem Sheet 3


3.9 For a general interatomic interaction potential U(r) we may define an effective
hard core dimension d by U(d) = kT. What is the significance of this definition?
Show that for the Lennard-Jones potential of Section 3.2.4, d is given by
16
⎧⎪ 2 ⎫⎪
d =σ ⎨ ⎬ .
⎪⎩1 + 1 + kT ε ⎭⎪
Plot this to demonstrate that d is a very slowly varying function of temperature.

If you have access to a symbolic mathematics system such as Mathematica or Maple,


show that at low temperatures
⎛ 1 kT 19 ⎛ kT ⎞
2

d ~ σ ⎜1 − + ⎜ ⎟ −…⎟ .
⎜ 24 ε 1152 ⎝ ε ⎠ ⎟
⎝ ⎠
So how high must the temperature be so that d differs appreciably from its zero
temperature value?

How does d vary for the hard core and the square well potentials of Sections 3.2.2 and
3.2.3?

3.10 The one dimensional analogue of the hard sphere gas is an assembly of rods
constrained to move along a line (the Tonks model). For such a gas of N rods of
length l confined to a line of length L, evaluate the configuration integral QN. Show
that in the thermodynamic limit the equation of state is
f ( L − Nl ) = NkT
where f is the force, the one dimensional analogue of pressure.

Comment on the similarities and the differences from the hard sphere equation of
state mentioned in Problem 3.8 (Clausius equation) and the van der Waals equation of
state.

3.11 Compare the square well and the van der Waals expressions for the second virial
coefficient. Show that they become equivalent when the range of the square well
potential tends to infinity while its depth tends to zero.

PH4211 © B. Cowan 2005 page P3-2 Problem Sheet 3


PH4211 Statistical Mechanics

Problem Sheet 4
4.1 Obtain an expression for the Helmholtz free energy for the Weiss model in zero
b g
external magnetic field, in terms of the magnetisation. Plot F M for T > TC ,
T = TC and T < TC .

4.2 Show that F =


Nk RSb g T
T − TC m2 + C m4 +
UV
2 T 6 W
for the Weiss model ferromagnet in

the limit of small m. Explain the appearance of TC in the m4 term.

4.3 Show that d 2 F dϕ 2 > 0 below Tc at the two roots ϕ = ± − F2 2 F4 in the Landau
model. Show that d 2 F dϕ 2 < 0 below Tc and d 2 F dϕ 2 > 0 above Tc at the single root
ϕ = 0. What is the physical meaning of this?

4.4 In the Landau theory of second order transitions calculate the behaviour of the
bg
order parameter below the critical point, ϕ T , when the sixth order term in the free
energy expansion is not discarded. What influence does this term have on the critical
exponent β ? Comment on this.

4.5 A ferroelectric has a free energy of the form


b g
F = α T − Tc P 2 + bP 4 + cP 6 + DxP 2 + Ex 2
where P is the electric polarisation and x represents the strain. Minimise the system
with respect to x. Under what circumstances is there a first order phase transition for
this system?

4.6 Consider a one-dimensional binary alloy where the concentration of A atoms


varies slowly in space: x = x(z). Show that the spatial variation of x results in an
additional term in the free energy per bond of 3a 2ε ( dx dz ) 2 , where a is the spacing
2

between atoms and ε is the energy parameter defined in Section 4.7.3.

4.7 Show that in the vicinity of the critical point the free energy of the binary alloy
may be written as
⎧ 2 16 ⎫
Fm = F0 + 2 Nk ⎨(T − Tc ) ( x − 12 ) + Tc ( x − 12 ) + Tc ( x − 12 ) + …⎬
2 4 6

⎩ 3 15 ⎭
Discuss the Landau truncation of this expression; in particular, explain at what term
the series may/should be terminated.

4.8 Plot some isotherms of the Clausius equation of state p (V − Nb ) = NkT . How do
they differ from those of an ideal gas? Does this equation of state exhibit a critical
point? Explain your reasoning.

4.9 The scaling expression for the reduced free energy is given in Section 4.1.9 by

PH4211 © B. Cowan 2005 page P4-1 Problem Sheet 4


⎛ B ⎞
f (T , B ) = A t
2 −α
Y ⎜D ∆ ⎟.
⎜ t ⎟
⎝ ⎠
Show that the heat capacity is given by
d2 f (t, B )
C~
dt 2
and hence identify α as the heat capacity critical exponent.

4.10 Using the scaling expression for the reduced free energy in the previous section,
show that the magnetisation is given by
df ( t , B )
M~
dB
and hence show that the order parameter exponent β is given by
β = 2 −α − ∆ .
Show that the magnetic susceptibility is given by
d2 f (t, B )
χ~
dB 2
and hence show that the susceptibility exponent g is given by
γ = 2 − α − 2∆ .

4.11 Show that the Landau free energy has the scaling form of Problem 4.9 above,
with α = 0.

PH4211 © B. Cowan 2005 page P4-2 Problem Sheet 4


PH4211 Statistical Mechanics

Problem Sheet 5
bg b g
5.1 A random quantity has an exponential autocorrelation function G t = G 0 e −γt .
Calculate its correlation time using the usual definition.

5.2 Show that the autocorrelation function of a periodically varying quantity


m ( t ) = m cos ωt is given by
m2
G (t ) = cos ωt .
2
Show that the autocorrelation function is independent of the phase of m(t). In other
words, show that if m ( t ) = m cos (ωt + ϕ ) , then G(t) is independent of ϕ.

5.3 The dynamical response function X(t) must vanish at zero times, as shown in
Fig. 5.13. What is the physical explanation of this? What is the consequence for the
step response function Φ(t)? Is this compatible with an exponentially decaying Φ(t)?

5.4 In Section 5.3 we examined the form of the dynamical susceptibility χ(ω) that
followed from the assumption that the step response function Φ(t) decayed
exponentially. In this question consider a step response function that decays with a
gaussian profile, Φ ( t ) = χ 0 e −t 2τ . Evaluate the real and imaginary parts of the
2 2

dynamical susceptibility and plot them as a function of frequency. The real part of the
susceptibility is difficult to evaluate without a symbolic mathematics system such as
Mathematica. Compare and discuss the differences and similarities between this
susceptibility and that deduced from the exponential step response function (Debye
susceptibility).

5.5 The Debye form for the dynamical susceptibility is


1
χ ′ (ω ) = χ 0
1 + ω 2τ 2
ωτ
χ ′′ (ω ) = χ 0 .
1 + ω 2τ 2
Plot the real part against the imaginary part and show that the figure corresponds to a
semicircle. This is known as a Cole-Cole plot.

5.6 Plot the Cole-Cole plot (Problem 5.5) for the dynamical susceptibility considered
in Problem 5.4. How does it differ from that of the Debye susceptibility.

5.7 The full quantum-mechanical calculation of the Johnson noise of a resistor gives
hf
v2 = 4 R hf kT ∆f .
∆f e −1

PH4211 © B. Cowan 2005 page P5-1 Problem Sheet 5


Show that this reduces to the classical Nyquist expression at low frequencies. At what
frequency will there start to be serious deviations from the Nyquist value? Estimate
the value of this frequency.

5.8 Show that for the Debye susceptibility, the relation


1 χ ′′ (ω )

χ0 = ∫ dω
π −∞
ω
holds. Demonstrate that χ ′′ vanishes sufficiently fast at ω = 0 so there is no pole in the
integral and there is thus no need to take the principal part of the integral in the
Kramers-Kronig relations.

5.9 In Section 5.3.6 we considered an electrical analogue of the Langevin Equation


based on a circuit comprising an inductor and a resistor. In this problem we shall
examine a different analogue: a circuit of a capacitor and a resistor. Show that the
equation analogous to the Langevin equation, in this case, is
dV ( t ) 1
C + V (t ) = I (t ) .
dt R
Hence show that the fluctuation-dissipation result relates the resistance to the current
fluctuations through

1 1
=
R 2kT ∫ I ( 0) I (t )
−∞
dt .

PH4211 © B. Cowan 2005 page P5-2 Problem Sheet 5


PH4210 Statistical Mechanics
Problem Sheet 1 — Answers

1 Demonstrate that entropy, as given by the Boltzmann expression S = k ln Ω , is an extensive


property. The best way to do this is to argue clearly that Ω is multiplicative.

We must prove that if one system has entropy S1 and another has S2 then when considered
together they have entropy S = S1 + S2. Note that we are not considering bringing the two
systems into contact, so the energy levels and their populations are not changed. We are
considering a composite system of two separate parts.

A microstate of the composite system is specified when the microstate of each of the systems is
specified. If the macrostate of system 1 has Ω1 microstates and the macrostate of system 2 has
Ω2 microstates, then when system 1 is in one of its microstates then system 2 can be in any one
of its Ω2 microstates. For every microstate of system 1 there are Ω2 microstates. But since
system 1 can exist in any of its Ω1 microstates it follows that there must be Ω1Ω2 microstates in
the macrostate of the composite system:
Ω = Ω1Ω 2
so that the entropy of the composite system is
S = kT ln Ω
= kt ln Ω1Ω 2
= kT ln Ω1 + kT ln Ω 2
= S1 + S 2 .
We have shown that the entropy for two isolated systems is additive, which means that entropy
is an extensive quantity.

2 Demonstrate that gravitational energy is not extensive: show that the gravitational energy of a
sphere of radius r and uniform density varies with volume as Vn and find the exponent n.

We shall calculate the work done against the force of gravity in assembling a sphere of matter
by bringing together the constituent parts from infinity. When all the matter is off at infinity we
take the gravitational energy to be zero. The gravitational energy of the body is the work done
in bringing the matter together.

We consider an intermediate state of the system: we have a spherical mass M of radius R and
we evaluate the work done in bringing up an extra mass dm from infinity to the surface.

The force between masses M and m separated by a distance r is given by Newton’s law of
gravitation
Mm
F = −G 2 .
r

Answers 1.1
Then the work done in moving a mass dM from infinity to distance R is
R
dE = − ∫ Fdr ,

where the minus sign means that we are doing work by applying a force against the force of
gravitation. So integrating up the Newton expression we obtain
GM
dE = − dM .
R
But in adding this extra mass the radius will have increased slightly. Since
4
M = π R3 ρ ,
3
where ρ is the density, it follows that
dM = 4π R 2 ρ dR .
We substitute for M and dM in the work expression:
G 4
dE = − π R 3 ρ 4π R 2 ρ dR
R 3
16
= −G π 2 ρ 2 R 4 dR .
3
The total gravitational energy is found by assembling the complete system, by building the
radius up from zero to its final value. Upon integration we find
16
E = −G π 2 ρ 2 R 5 .
15
Finally we must express this energy in terms of the volume of the sphere
4
V = π R3 ,
3
so that
13
⎛ 3 ⎞
R=⎜ V⎟ .
⎝ 4π ⎠
Then we obtain
53
16 ⎛ 3 ⎞
E = −G ⎜ ⎟ π ρV .
2 2 53

15 ⎝ 4π ⎠
Thus the gravitational energy varies as the 5/3 power of volume; the exponent n = 5/3. This
shows that gravitational energy is not extensive; extensivity requires energy to be proportional
to volume; the exponent n would be unity.

3 In investigating the conditions for the establishment of equilibrium through the transfer of
thermal energy the fundamental requirement is that the entropy of the equilibrium state should
be a maximum. Equality of temperature was established from the vanishing of the first
derivative of S. What follows from a consideration of the second derivative?

The total entropy of the composite system may be written S ( E ) = S1 ( E ) + S 2 ( Et − E ) . Setting


the first derivative of this to zero leads to the equality of the temperatures of the two systems.
Answers 1.2
The second derivative of the total entropy is easily expressed in terms of the first derivatives of
the temperatures of the two systems
d2S
=
d 1 FG IJ
+
d 1
.
FG IJ
dE 2
dE T1 H K
dE T2 H K
The requirement that in equilibrium the entropy be a maximum means that the second derivative
must be negative. If we evaluate the derivative of the reciprocals we find
1 dT 1 dT
− 2 1 − 2 2 <0
T1 dE T2 dE
or
1 1
+ >0,
C1 C2
since the temperatures are equal in equilibrium. Here the C are the thermal capacities of the
subsystems. The inequality must be satisfied for all allowed values of C1 and C2. In particular it
must be satisfied when either system becomes vanishingly small so that its heat capacity
approaches zero. Then the inequality requires that the thermal capacities must be positive.

4 Do particles flow from high µ to low µ or vice versa? Explain your reasoning.

The change in entropy, when particles flow is given by


⎛ ∂S ∂S ⎞
∆S = ⎜ 1 − 2 ⎟ ∆N1 ,
⎝ ∂N ∂N ⎠
which must be greater than or equal to zero by the Second Law. But since
∂S µ
=−
∂N T
it follows that
1
∆S = ( µ2 − µ1 ) ∆N1 ≥ 0
T
since temperature is the same in both systems.

We conclude that when µ2 > µ1 then ∆N1 will be positive; N1 will increase. In other words,
particles will flow from high µ to low µ.

5 In the derivation of the Boltzmann factor the entropy of the bath was expanded in powers of
the energy of the ‘system of interest’. The higher order terms of the expansion are neglected.
Discuss the validity of this.

We derived that
P ( E ) ∝ e S ( ET − E ) k
where E ET so that we expand S as a Taylor series:
∂S E 2 ∂ 2 S
S ( ET − E ) = S ( ET ) − E + −…
∂E 2 ∂E 2
Answers 1.3
The derivative in the second term is immediately identified as the inverse temperature
∂S 1
= .
∂E T
The derivative in the second term is then
∂2S ∂ 1
=
∂E 2
∂E T
1 ∂T
=− 2 .
T ∂E
But here ∂T ∂E is the inverse of the thermal capacity C (of the reservoir). Then the entropy
expansion becomes
E E2 1
S ( ET − E ) = S ( ET ) −− −…
T 2T 2 C
and the higher order terms involve derivatives of the thermal capacity.

The key point of the argument is that the thermal capacity of the reservoir is very large. The
assumption is that while the reservoir determines the properties of our “system of interest”, the
system of interest can have no effect on the reservoir. This is what we mean by a reservoir, and
it may be encapsulated by saying that its thermal capacity is essentially infinite. Another way
of looking at this is to say that since thermal capacity is an extensive quantity then in the limit
that the reservoir is large, its thermal capacity will be large. Then the inverse of the thermal
capacity will be small and then the third term in the entropy expansion can be neglected. The
higher-order terms involve derivatives of the thermal capacity so that these can also be ignored.

6 The Boltzmann factor could have been derived by expanding Ω rather than by expanding S.
In that case, however, the expansion cannot be terminated. Why not?

If we expand Ω we get
∂Ω E 2 ∂ 2Ω
Ω ( ET − E ) = Ω ( ET ) − E + −…
∂E 2 ∂E 2
Since S = k lnΩ, and ∂S ∂E = 1 T , it follows that
∂Ω Ω ∂ 2Ω Ω
= , = , etc.
∂E kT ∂E ( kT )
2 2

then
⎧⎪ E 1⎛ E ⎞
2
⎫⎪
Ω ( ET − E ) = Ω ( ET ) ⎨1 − + ⎜ ⎟ − … ⎬
⎩⎪ kT 2 ⎝ kT ⎠ ⎭⎪
or
⎪⎧ ⎪⎫
2
E 1⎛ E ⎞
P ( E ) ∝ ⎨1 − + ⎜ ⎟ − …⎬ .
⎩⎪ kT 2 ⎝ kT ⎠ ⎭⎪
So in this case you certainly can’t terminate the expansion. But you can sum the infinite series
(it is the exponential function) – and this gives the Boltzmann factor.

Answers 1.4
7 Show that ln N ! = ∑ n =1 ln n . By approximating this sum by an integral obtain Stirling’s
N

approximation: ln N ! ≈ N ln N − N .

The factorial is given by


N ! = 1× 2 × 3 × … × ( N − 1) × N
so that the logarithm is
ln N ! = ln {1× 2 × 3 × … × ( N − 1) × N }
= ln1 + ln 2 + ln 3 + … + ln ( N − 1) + ln N
N
= ∑ ln n ,
n =1

as required.

Now we approximate the sum by an integral:


N N

∑ ln n ≈ ∫ ln n dn ;
n =1 0

this corresponds to the lower dotted line of the figure below.

lnx

1 2 3 4 5 6 7 8 x

Evaluation of the integral gives


N

∫ ln n dn = N ln N − N
0

so that we obtain Stirling’s approximation as


ln N ! ≈ N ln N − N .

8 Show that the Gibbs expression for entropy: S = − k ∑ j Pj ln Pj , reduces to the Boltzmann
expression S = k ln Ω in the case of an isolated system.

Answers 1.5
The fundamental postulate of statistical mechanics states that for an isolated system all
microstates are equally likely. If this isolated system has Ω microstates then the probability of
any one of these microstates is 1/Ω. Then generalised entropy expression is then
S = − k ∑ Pj ln Pj
j

⎛1⎞ ⎛1⎞
= − k ∑ ⎜ ⎟ ln ⎜ ⎟
j ⎝Ω⎠ ⎝Ω⎠
⎛1⎞
= k ∑ ⎜ ⎟ ln Ω.
j ⎝Ω⎠

Now each term in the sum is a constant, and there will be Ω such terms. Thus the expression
for entropy becomes
S = k ln Ω
as required.

9 What is the condition that the geometric progression in deriving the Bose-Einstein
distribution is convergent?

The geometric progression is

( pV )k = kT ln ∑ {e−(ε −µ ) kT }
∞ nk
k
.
nk = 0

This will be convergent if


− ( ε k − µ ) kT
e <1.
In other words one requires
ε >µ
for all single-particle energies ε. Now the ground state energy will be zero (or very close to it),
so that the condition is on the chemical potential:
µ <0
the chemical potential must always be negative. Note that this requirement applies to Bose
particles; it does not apply to Fermions.

10 Show that the trajectory of a 1d harmonic oscillator is an ellipse in phase space. What
would the trajectory be if the oscillator were weakly damped.

The displacement of a simple harmonic oscillator evolves in time as


x ( t ) = A sin (ωt + ϕ ) .
The momentum is given by
p = mx = Amω cos (ωt + ϕ ) .
The pair
x ( t ) = A sin (ωt + ϕ )
p ( t ) = B cos (ωt + ϕ )

Answers 1.6
are seen to specify an ellipse in x-p space (phase space), since ( x A ) + ( p B ) = 1 .
2 2

In the case of damping there will be an exponential decay superimposed on the sine and cosine.
So the phase point will approach the point x = 0, p = 0. In the case of weak damping the phase
point will trace out many cycles of the ellipse before the “radius” changes appreciably. The
effect of weak damping is thus to cause the phase point to gradually spiral into the origin.

The point about weak damping is that any single cycle is still observed to be essentially
elliptical.

11 Why can’t the evolutionary curve in phase space intersect? You need to demonstrate that the
evolution from a point is unique.

Newton’s equations of motion are second order differential equations – there are up to second
derivatives of the spatial coordinates. A complete solution to the equations thus involves two
constants of integration for each degree of freedom. And these constants could be the position
coordinate and the momentum component at a given instant in time. (Hamilton’s formulation of
dynamics then extends the idea to a generalised view of coordinates and momenta.) Such a
complete solution of the equations of motion gives a unique solution; the evolution in time is
completely determined. Now a ‘point’ in phase space represents a given value of position and
momentum. So the future (and past) evolution of the system, and thus the phase space
trajectory, is completely determined from this point. It then follows that the evolutionary path
through a point in phase space is unique. Then two paths cannot pass through the same phase
point and so the evolutionary curve in phase space cannot intersect itself.

12 Starting from the expression for the Gibbs factor for a many-particle system, write down the
grand partition function Ξ and show how it may be expressed as the product of Ξk, the grand
partition function for the subsystem comprising particles in the kth single-particle state.

One starts from the Gibbs factor


1 −{ E ( N ,V ) − µ N } kT
PN , j (V , T , µ ) = e N,j .
Ξ (V , T , µ )
The normalization constant Ξ is the grand partition function:
−{ E ( N ,V ) − µ N }
Ξ (V , T , µ ) = ∑ e N , j
kT

N, j
th
where N, j specify the j quantum state of the system when it contains N particles.

Now for a system of identical particles


N = ∑ nk
k

E = ∑ nk ε
k

and a given state of the system is specified by the occupation of the single-particle states k:

Answers 1.7
( n1 , n2 , n3 , …) ≡ {nk } .
So in this case the grand partition function is expressed as
− ∑ ( ε k − µ ) nk kT
Ξ (V , T , µ ) = ∑ e k
{nk }

= ∑∏ e
− ( ε k − µ ) nk kT
.
{nk } k

Now the nk are all independent, so that


Ξ (V , T , µ ) = ∏ ∑ e
−( ε k − µ ) nk kT
.
k nk

Now we have the definition


Ξ k (V , T , µ ) = ∑ e
−( ε k − µ ) nk kT
,
nk

so that the grand partition function may be expressed as the product


Ξ (V , T , µ ) = ∏ Ξ k (V , T , µ )
k

as required.

It is important to note the logic of this answer. It is not good enough to start from Ξk and then
to argue that it must be multiplicative since the (pV)k contributions must be additive; this is not
what is asked for. — That was the way things were argued in the text.

13 This problem considers the probability distribution for the energy fluctuations in the
canonical ensemble. The moments of the energy fluctuations are defined by
1
σn = ∑(Ej −ε ) e j
n βE

Z j
where β = − 1 kT and ε is an arbitrary (at this stage) energy.
Show that
βε ∂
n

n {
Zσ n = e Z e − βε }
∂β
and use this to prove that the fluctuations in an ideal gas obey a normal distribution around ε.
(You really need to use a computer algebra system to do this problem.)

The partition function is given by


Z = ∑e
βEj

so that
β ( E j −ε )
Ze − βε = ∑ e .
j

If we differentiate this n times with respect to β then this brings down n factors (Ej − ε) on the
right hand side:

Answers 1.8
∂n n β ( E −ε )
n {
Ze − βε } = ∑ ( E j − ε ) e j .
∂β j

Multiply this by e βε , to obtain


∂n
n {
Ze − βε } = ∑ ( E j − ε ) e j
n βE
e βε
∂β j

whereupon we observe the right hand side to be Zσn. And thus we find
∂n
Zσ n = e βε
∂β n {
Z e − βε }
as required.

The nth moment of the energy fluctuation distribution is then given by


∂n
∂β n {
Z e − βε }
σn =
Z e − βε
and for the ideal gas the partition function Z is given by
3N 2 N
⎧ mkT ⎫ ⎧Ve ⎫
Z =⎨ 2⎬ ⎨ ⎬
⎩ 2π ⎭ ⎩N ⎭
3N 2
.
⎧ −m ⎫
N
⎧Ve ⎫
=⎨ ⎬ ⎨ ⎬ .
⎩ 2π β ⎭
2
⎩N ⎭
Then the nth moment may be written as
∂n
∂β n
{( −1/ β ) 3N 2
e − βε }
σn =
( −1/ β )
3N 2
e − βε
and, in particular, we find for the first moment
⎛ 3N ⎞
σ1 = − ⎜ ε + ⎟
⎝ 2β ⎠
3
NkT − ε . =
2
If we now choose the energy ε, about which the moments are evaluated, to be the mean energy
3NkT/2 then the first moment will vanish. Thus we adopt this value for ε. The moments may
then be evaluated as
σ0 =1
σ1 = 0
σ 2 = 3N 2β 2
σ 3 = −3N β 3
σ 4 = ( 27 N 2 + 36 N ) 4 β 2
etc.

Answers 1.9
The higher-order moments may be calculated in a straightforward, if tedious, manner.

The normal distribution (with zero mean) is characterised by the numerical value of the
dimensionless ‘reduced’ moments
mn = σ n σ 2n 2 .
In general these moments will be functions of the number of particles N. We find
m0 = 1
m1 = 0
m2 = 1
2 2
m3 = −
3N
4
m4 = 3 +
N
2⎛ 4 5 ⎞
m5 = −4 ⎜ 32 + 12 ⎟
3⎝N N ⎠
260 160
m6 = 15 + +
3N 3N 2
etc.
Then in the thermodynamic limit, N → ∞, we see that the odd reduced moments vanish while
the even ones tend to the values
m0 = 1
m2 = 1
m4 = 3
m6 = 15
m8 = 105
etc.
The even reduced moments of the normal distribution are given by
m2 n = 1× 3 × 5 × … × ( 2n − 1) .
We see that these evaluate to the numbers calculated above. And thus we conclude that in the
thermodynamic limit the energy fluctuations of an ideal gas obey a normal distribution.

14 For a single-component system with a variable number of particles, the Gibbs free energy is
a function of temperature, pressure and number of particles: G = G (T , p, N ) . Since N is the
only extensive variable upon which G depends, show that the chemical potential for this system
is equal to the Gibbs free energy per particle: G = Nµ.

If the system is increased by a factor x then the extensive variables G and N will be increased by
this factor:
xG (T , p, N ) = G (T , p, xN ) .
And in particular if x = 1/N then
Answers 1.10
G (T , p, N ) = NG (T , p,1) ;
the Gibbs free energy for a system of N particles is N times the Gibbs free energy per particle.
But the gibbs free energy per particle may be expresses as

G (T , p,1) = G ( T , p, N ) .
∂N
However we identify the derivative of the Gibbs free energy as the chemical potential. Thus we
obtain the required result
G = Nµ .

15 Use the definition of the Gibbs free energy together with the result of the previous question
to obtain the Euler relation of Appendix 1.

G is defined as
G = E − TS + pV .
But the previous Problem gives G = Nµ. Thus we have
E − TS + pV = µ N
and the Euler relation
E = TS − pV + µ N
then follows immediately.

16 The energy of a harmonic oscillator may be written as mω 2 x 2 2 + p 2 2m so it is quadratic


in both position and momentum – thus, classically, equipartition should apply. The energy
levels of the quantum harmonic oscillator are given by ε n = ( 12 + n ) ω . Show that the partition
function of this system is given by
1 ω
Z = cosech
2 2kT
and that the internal energy is given by
1 ω ω ω
E = ω coth = ω kT + .
2 2kT e −1 2
Show that at high temperatures E may be expanded as
2
ω2
E = kT + +
12kT
Identify the terms in this expansion.

The partition function is defined by



Z = ∑ e −ε n kT
.
n=0

Upon substitution for εn we then have

Answers 1.11

− ( 12 + n ) ω kT
Z = ∑e
n=0

∑ (e )
ω 2 kT − ω kT n
= e− .
n=0

We observe the second sum here to be a geometric progression, which is easily summed:

∑ ( e− ω kT ) = 1 − e1− ω kT .
n

n=0

Thus Z is given by
e − ω 2 kT
Z=
1 − e − ω kT
1
= ω 2 kT ω 2 kT
e − e−
which we identify to be
1 ω
Z = cosech ,
2 2kT
as required.

The internal energy is found from the partition function as


∂ ln Z
E = kT 2 .
∂T
Upon differentiation we find
∂ ln Z ω ω
= coth
∂T 2kT 2
2kT

so that
1 ω ω ω
E= ω coth = ω kT + .
2 2kT e −1 2
The high temperature expansion of this expression may be found using the expansion of the
hyperbolic cotangent:
1 x x3 2 x5
coth x = + − + +…,
x 3 45 945
giving
⎧⎪ 1 ω 1 ⎛ ω⎞
3
1 ⎛ ω⎞
5
⎫⎪
E = kT + ω ⎨ − ⎜ ⎟ + ⎜ ⎟ + … ⎬.
⎩⎪12 kT 720 ⎝ kT ⎠ 30240 ⎝ kT ⎠ ⎭⎪
The first term, kT, is the high-temperature equipartition contribution to the internal energy.
There follows a series in inverse powers of T. These give the corrections as the temperature gets
lower.

Answers 1.12
PH4211 Statistical Mechanics

Problem Sheet 2 — Answers


1 In Section 2.1 we saw that the density of free-particle states for a three dimensional
volume V was shown to be
1 V
g (ε ) = 2 3(
2m ) ε 1 2 .
32


This followed from counting the number of states in the octant of radius
R = nx2 + n y2 + nz2 .
By similar arguments show that in two dimensions, by counting the number of states
in the quadrant of radius
R = nx2 + n y2 ,
the density of states is given by
mA
g (ε ) =
2π 2
where A is the area. Note in two dimensions the density of states is independent of
energy.

And similarly, show that in one dimension the density of states is


12
L ⎛ m ⎞ −1 2
g (ε ) = ε .
π ⎜⎝ 2 ⎟⎠

In the two-dimensional case the energy states may be specified (similar to Eq. 2.2) by
π2 2
ε=
2mA
(n 2
x + n y2 ) ,

where A is the area. This may be written as


π2 2
ε= R2
2mA
where R = n + n .
2 2
x
2
y

Now the number of states of energy up to ε, denoted by N(ε), is given by the number
of points in the quadrant up to ε(R). (A quadrant is used since n x and n y are restricted
B B B B

to being positive). And the number of points in the quadrant is approximately equal
to the area of the quadrant:
1
N (ε ) = π R 2 .
4
But since
12
⎛ 2mA ⎞
R = ⎜ 2 2 ⎟ ε1 2 ,
⎝π ⎠
we then obtain

Answers 2-1
1 2mA
N (ε ) = π 2 2 ε
4 π
1 mA
= ε.
2π 2
Recall that the density of states g(ε) is defined by saying that the number of states
with energy between ε and ε + dε is g(ε). In other words
g ( ε ) dε = N ( ε + dε ) − N ( ε )
or, simply
dN ( ε )
g (ε ) = .

So differentiating N(ε) we obtain, in two dimensions
mA
g (ε ) =
2π 2
which is the required expression for the density of states. And indeed we observe this
is a constant, independent of energy.

In the one-dimensional case the energy states are specified (similar to Eq. 2.2) by
π2 2
ε= 2
nx2 ,
2mL
where L is the length. This may be written as
π2 2
ε= 2
R2
2mL
where here R = n .
2 2
x

Now the number of states of energy up to ε, denoted by N(ε), is given by the number
of points in the positive line up to ε(R). (The positive line is used since n x is restricted
B B

to being positive). And the number of points in the line is approximately equal to the
length of the line:
N (ε ) = R .
But since
12
⎛ 2mL2 ⎞ 1 2
R=⎜ 2 2 ⎟ ε ,
⎝π ⎠
we then obtain
12
⎛ 2mL2 ⎞ 1 2
N (ε ) = ⎜ 2 2 ⎟ ε
⎝π ⎠
L
( 2m ) ε 1 2 .
12
=
π
Recall that the density of states g(ε) is defined by saying that the number of states
with energy between ε and ε + dε is g(ε). In other words
g ( ε ) dε = N ( ε + dε ) − N ( ε )
or, simply

Answers 2-2
dN ( ε )
g (ε ) = .

So differentiating N(ε) we obtain, in one dimension
12
L ⎛m⎞
g (ε ) = ⎜ ⎟ ε
−1 2

π ⎝2⎠
which is the required expression for the density of states.

2 In Sections 2.3.1 and 2.3.2 the ideal gas partition function was calculated quantum-
mechanically and classically. Although the calculations were quite different, they both
resulted in (different) Gaussian integrals. By writing the Gaussian integral of the
classical case as
∞ ∞ ∞
(
− x2 + y 2 + z 2 )
∫ dx ∫ dy ∫ dz e
−∞ −∞ −∞
and transforming to spherical polar coordinates, you can perform the integration over
θ and ϕ trivially. Show that the remaining integral can be reduced to that of the
quantum case.

If we transform to spherical polar coordinates:


x = r sin θ cos ϕ
y = r sin θ sin ϕ
z = r cos θ
then the volume element is given by
dv = dx dy dz = r 2 sin θ dr dθ dϕ .
So the Gaussian integral is given by
∞ ∞ ∞ 2π π ∞
(
− x2 + y2 + z2 )=
∫ dx ∫ dy ∫ dz e ∫ dϕ ∫ sin θ dθ ∫ r e dr .
2
2 −r
I=
−∞ −∞ −∞ 0 0 0

We can evaluate the θ and φ integrals, to give



I = 4π ∫ r 2 e − r dr .
2

Now change variables to x = r 2 so that dr = dx 2 x1 2 . Then the integral beomes



I = 2π ∫ x1 2 e − x dx
0

and this is the integral that appeared in the quantum calculation of the partition
function.

3 The Sakur-Tetrode equation, discussed in Section 2.3.3,


3
S = Nk ln V − Nk ln N + Nk ln T + Nks0
2
is often interpreted as indicating different contributions to the entropy: the volume
contribution is in the first term, the number contribution in the second term and the
temperature contribution in the third term. Show that such an identification is
fallacious, by demonstrating that the various contributions depend on the choice of

Answers 2-3
units adopted – even though the total sum is independent. Discuss the origin of the
fallacy.

The entropy of the ideal gas is given by


⎡⎛ mkT ⎞3 2 V 5 2 ⎤
S = Nk ln ⎢⎜ 2 ⎟
e ⎥ .
⎣⎢⎝ 2π ⎠ N ⎦⎥
The argument of the logarithm is dimensionless. This may be checked explicitly, but
note that in terms of the thermal de Broglie wavelength Λ,
⎡ V ⎤
S = Nk ln ⎢ 3 e5 2 ⎥
⎣Λ N ⎦
so the argument is a volume divided by a volume – dimensionless.

By analogy with the Sakur-Tetrode equation, we may write the entropy as


5
S = Nk ln V − Nk ln N − Nk ln Λ 3 + Nk .
2
Now imagine that we decide to measure length in centimetres instead of meters. Then
the numerical value of the argument in lnV will increase by a factor of 10 6 . So the
P P

calculated value of lnV will increase by 6ln10 = 13.82…. So the size of the ‘volume
contribution’ to the entropy will change if the unit of volume is changed.

Of course the total entropy does not change. Clearly if the length unit is changed in
this way then the thermal de Broglie wavelength term will also change – and in just
such a way to cancel the change from the V term. Thus it makes no sense to identify
NklnV as the volume contribution to the entropy.

Another way of looking at this matter is to recognise that for any function that
contains different powers of its arguments (recall the power series for the logarithm),
the argument of the function should be dimensionless. If not, the analysis of
dimensions becomes senseless. Thus we conclude that it is acceptable to write S as
V 5
S = Nk ln 3 − Nk ln N + Nk ,
Λ 2
but any further decomposition gives non-unique apportionments of the different
contributions.

4 Show that the Fermi energy for a two-dimensional gas of Fermions is


2π 2 N
εF =
αm A
where A is the area of the system.

At zero temperature, where the Fermi distribution becomes a step function, the
number of particles in the system may be expressed as
εF
N = α ∫ g ( ε ) dε
0

and since in 2d the density of states is

Answers 2-4
mA
g (ε ) = ,
2π 2
independent of energy, N is given by
α mA
N= εF
2π 2
so that
2π 2 N
εF = ,
αm A
as required.

5 Show that the chemical potential of a two-dimensional gas of fermions may be


expressed analytically as
µ = kT ln eε F kT − 1 { }
At finite temperatures the number of particles is given by
α mA ∞ dε
2 ∫ ( ε − µ ) kT
N=
2π 0 e +1
and since
2π 2 N
εF =
αm A
the integral for N may be expressed


εF = ∫ (ε − µ ) kT
.
0 e +1
We shall change variables to x = ( ε − µ ) kT so that

dx
ε F = kT
− µ
∫ kT
e +1
x
.

The integral may be evaluated (using Mathematica, for example) as



dx µ
∫ = + ln {1 + e − µ kT }
− µ kT
e + 1 kT
x

however this may be simplified by writing the argument of the logarithm differently
µ
kT
+ ln {1 + e − µ kT
} = kTµ + ln {e − µ kT
(e µ kT
+ 1) }
µ
= + ln e − µ kT + ln {e µ kT
+ 1}
kT
= ln {e µ kT + 1} .
Thus we conclude
ε F = kT ln {e µ kT + 1} .
And this may be rearranged to give the chemical potential as
µ = kT ln {eε F kT − 1} ,
as required.

Answers 2-5
6 Calculate the low temperature chemical potential of a two-dimensional gas of
fermions by the Sommerfeld expansion method of Section 2.4.3. Observe that the
temperature series expansion terminates. Compare this result with the exact result of
the previous question. Discuss the difference between the two results.

We start from the expression for the Fermi energy for a two-dimensional system,
following from the expression for the total number of particles:

ε F = ∫ n ( ε ) dε .
0

This may be integrated by parts (the other factor of the integrand in this case is unity):

ε F = − ∫ ε n′ ( ε ) dε .
0

We have to expand the ε factor about ε = µ


ε = µ + (ε − µ )
so the Fermi energy is given by
∞ ∞
ε F = − µ ∫ n′ ( ε ) dε − ∫ ( ε − µ ) n′ ( ε ) dε
0 0

with no higher order terms. The first integral is –I 0 of Section 2.4.3 and the second
B B

integral is –kTI 1 . Now I 0 = 1, which may be shown quite simply. And within the
B B B B

approximation of the Sommerfeld expansion, that the lower limit of the integral be
extended to –∞, I 1 vanishes, as do all the odd-order terms. Thus the Sommerfeld
B B

expansion expression for the chemical potential is simply


µ = εF ;
the expansion terminates (rather early) and it completely misses the temperature
dependence of µ away from T = 0.

The previous Problem showed in 2d the chemical potential may be expressed exactly
µ = kT ln {eε F kT − 1} .
This can also be written as
µ = kT ln eε { (1 − e
F kT −ε F kT
)}
= εF + ln {1 − e } −ε F kT

which is more convenient at low temperatures. Then the logarithm term is the part
missed by the Sommerfeld series. And at low temperatures the exponential is small
and so the logarithm can be expanded in powers of this exponential
1 1 1
ln {1 − e −ε F kT } = −e−ε F kT − e− 2ε F kT − e −3ε F kT − e− 4ε F kT − …
2 3 4
so that
1 1 1
µ = ε F − e−ε F kT − e− 2ε F kT − e −3ε F kT − e − 4ε F kT − … .
2 3 4
It is the exponential terms that are missed by the Sommerfeld expansion.

Answers 2-6
7 The general formula for the Fermi integrals I n of Section 2.4.3 was quoted as
B B


ex
In = ∫ x n dx
−∞ ( e + 1)
x 2

= ( 2 − 22− n ) ζ ( n ) n !
Derive this result. (You might find the discussion in Landau and Lifshitz, Statistical
Physics, helpful.)

8 Obtain the chemical potential µ, the internal energy E and the heat capacity C v for B B

system with general density of states g(ε) as in Section 2.4.4. I.e. show that these are
given in terms of the behaviour of the density of states at the Fermi surface.

9 Consider the Bose gas at low temperatures. You saw in Section 2.5.2 and 2.5.3 that
when the occupation of the ground state is appreciable then the chemical potential µ is
very small and it may be ignored, compared with ε in the integral for the number of
excited states.

Show that when the ground state occupation N 0 is appreciable then µ may be
B B

approximated by
µ ~ − kT N 0 .
Now consider the more stringent requirement that µ may be neglected in comparison
with ε in the integral for the number of excited states. This will be satisfied if µ is
much less than the energy ε 1 of the first excited state. The expression for ε 1 is
B B B B

π 2 2
ε1 ~ .
2mV 2 3
Where does this expression come from?

Show that the condition µ ε1 is satisfied when N 1015 (approximately) when


considering 1cm 3 of 4 He (molar volume 27cm 3 ) at a temperature of about 1K.
P P P P P P

Thus show that the expression


⎧⎪ ⎛ T ⎞3 2 ⎫⎪
N 0 = N ⎨1 − ⎜ ⎟ ⎬
⎪⎩ ⎝ Tc ⎠ ⎭⎪
is then valid to temperatures below T c right up to within ∼ 10 −8 T c of the critical
B B P P B B

temperature.

10 Liquid 4 He has a molar volume at saturated vapour pressure of 27cm 3 . Treating


P P P P

the liquid as an ideal gas of bosons, find the temperature at which Bose-Einstein
condensation will occur. How will this temperature change as the pressure on the fluid
is increased?

The condensation temperature is given, from Eq. 2.50, by

Answers 2-7
2 23
⎧N ⎫
Tc = 3.313
⎨ ⎬ .
mk ⎩ V ⎭
The ‘atomic mass’ of helium is 4. And one atomic mass unit weighs 1.66 × 10 –27 kg P P

so a helium atom weighs 6.64 × 10 –27 kg. P P

The volume of one mole of liquid helium is 27 cm 3 , or 27 × 10 –6 m 3 . Now N/V is


P P P P P P

Avogadro’s number divided by the molar volume:


N 6.02 × 1023
=
V 27 × 10−6
= 2.23 × 1028
We need the 2/3 power of this:
23
⎛N⎞
⎜ ⎟ = 7.92 ×10 .
18

⎝V ⎠
Then, using the values for Planck’s constant and Boltzmann’s constant, we obtain

Tc = 3.313
(1.05 ×10 ) −34 2

× 7.92 × 1018
−27 −23
6.64 × 10 × 1.38 ×10
= 3.15 K.
The temperature at which Bose-Einstein condensation occurs is 3.15 K.

Since the formula for T c indicates that it increases with density, N/V, and since
B B

increasing the pressure will increase the density, it follows that increasing the pressure
will increase the Bose-Einstein condensation temperature.

11 The superfluid transition temperature of liquid helium decreases with increasing


pressure. Very approximately ∂Tc ∂p ~ −0.015 K bar −1 . How does this compare with
the behaviour predicted from the Bose-Einstein condensation?

The Bose-Einstein condensation temperature is given by


2 23
⎧N ⎫
Tc = 3.313
⎨ ⎬ ;
mk ⎩ V ⎭
this predicts that T c will increase with pressure (density). So the observed reduction
B B

of T c with pressure is in conflict with this model.


B B

12 Show that below the transition temperature the entropy of a Bose gas is given by
32
5 I3 2 ⎛ T ⎞
S = Nk ⎜ ⎟ .
3 I1 2 ⎝ Tc ⎠
Since the number of excited particles is given by
32
⎛T ⎞
N ex = N ⎜ ⎟ ,
⎝ Tc ⎠
show that the entropy per excited particle is given by

Answers 2-8
S 5 I3 2
= k ≈ 1.28k .
N ex 3 I1 2
Discuss the connection between this result and the two fluid model of superfluid 4 He.
P P

The entropy is most conveniently found from the heat capacity C V . Since
B B

∂S
CV = T
∂T v
it follows that
CV
S=∫ dT .
T
And since the heat capacity is given from Eq. 2.54 as
32
5 I3 2 ⎛ T ⎞
CV = Nk ⎜ ⎟ ,
2 I 5 2 ⎝ Tc ⎠
upon integration we find
32
5 I3 2 ⎛ T ⎞
S = Nk ⎜ ⎟
3 I1 2 ⎝ Tc ⎠
as required.

Now the number of particles out of the ground state, the number of excited particles,
is
32
⎛T ⎞
N ex = N ⎜ ⎟ ,
⎝ Tc ⎠
so dividing the total entropy we obtain the entropy per excited particle as
S 5 I3 2
= k ≈ 1.28k .
N ex 3 I1 2
This indicates that the entropy per excited particle is a constant, independent of
temperature. Now the two fluid model argues that the system is comprised of two
interpenetrating fluids – the ‘normal’ component and the ‘superfluid’ component. And
it asserts that the superfluid component carries no entropy. Here we see that the
entropy may indeed all be associated with the ‘normal’ or excited component, each
excited particle carrying the same ‘unit’ of entropy.

13 Show that the Bose-Einstein transition temperature of a gas of bosons and the
Fermi temperature for a gas of ‘similar’ fermions are of comparable magnitude.
Discuss why this should be.

The Bose-Einstein transition temperature for a gas of spin 0 bosons (α = 1) is given by


23
2π 2 ⎧ N ⎫
Tc = ⎨ ⎬
mk ⎩ 2.612V ⎭
2 23
⎧N ⎫
= 3.313 ⎨ ⎬ .
mk ⎩ V ⎭

Answers 2-9
The Fermi temperature of a gas of spin ½ fermions (α = 2), ε F k , is given by
23
2
⎧ 6π 2 N ⎫
TF = ⎨ ⎬
2mk ⎩ 2 V ⎭
2 23
⎧N ⎫
= 4.785 ⎨ ⎬ .
mk ⎩ V ⎭
Thus the Bose condensation temperature and the Fermi temperature are very similar.
They certainly depend on all the system variables in the same way and the the Fermi
temperature of a gas of spin ½ fermions is approximately 1.4 times the Bose-Einstein
transition temperature for a gas of spin 0 bosons.

The reason they are similar is because both temperatures are characterised by the
thermal deBroglie wavelength for the particles becoming comparable with the inter-
particle spacing. This is the condition for quantum effects to become important.

14 In Section 2.6 we studied a paramagnetic solid: a collection of essentially


distinguishable magnetic moments. If we were to consider a (classical) gas of
indistinguishable magnetic moments, how would the partition function be modified?
What would the observable consequences of this modification?

A full treatment of a gas of particles would require consideration of the particles’


kinetic energy. However if we neglect this then the only modification to the partition
function would be the inclusion of the N! term for indistinguishability. Then we
would have
{2 cosh ε kT }
N

Z=
N!
and since F = − kT ln Z the free energy is
F = − NkT ln {2 cosh ε kT } + NkT ln N − NkT .
The first term is the free energy calculated for the distinguishable solid case and the
second and third terms are the corrections for indistinguishability.

Since dF = − SdT − MdB , it follows that


∂F ∂F
M =− and S = − .
∂B ∂T
This means that while the magnetisation of the system will be the same as that of the
distinguishable system, there will be extra contributions to the entropy (and thus to the
heat capacity etc.).

15 Show, using arguments similar to those in Section 2.1.3, that the energy levels of
an ultra-relativistic or a massless particle with energy-momentum relation E = cp are
given by

ε = 1 3 ( nx2 + n y2 + nz2 ) .
12

V
Hence show that the pressure of a gas of such particles is one third of the (internal)
energy density.

Answers 2-10
16 Evaluate the Fermi temperature for liquid 3 He, assuming it to be a Fermi “gas”. Its
P P

molar volume is 36 cm 3 . Calculate the de Broglie wavelength at T = T F and show


P P B B

that it is comparable to the interparticle spacing as expected.

The Fermi temperature is given by


23
2
⎧ 6π 2 N ⎫
TF = ⎨ ⎬
2mk ⎩ 2 V ⎭
2 23
⎧N ⎫
= 4.785 ⎨ ⎬ .
mk ⎩ V ⎭
The ‘atomic mass’ of 3 He is 3. And one atomic mass unit weighs 1.66 × 10 –27 kg so a
P P P P

helium-3 atom weighs 4.98 × 10 –27 kg. P P

The volume of one mole of liquid helium-3 is 36 cm 3 , or 36 × 10 –6 m 3 . Now N/V is


P P P P P P

Avogadro’s number divided by the molar volume:


N 6.02 ×1023
=
V 36 ×10−6
= 1.67 ×1028
so that
23
⎛N⎞
⎜ ⎟ = 6.53 ×10 .
18

⎝V ⎠
Then, using the values for Planck’s constant and Boltzmann’s constant, we obtain

TF = 4.785 ×
(1.05 ×10 ) −34 2

× 6.53 × 1018
−27 −23
4.98 × 10 × 1.38 × 10
= 5.01 K.
Thus the Fermi temperature is calculated as 5.01 K.

17 In Problem 2.1 we found the expression for the energy density of states g(ε) for a
gas of fermions confined to two dimensions and we saw that it was independent of
energy. What surface density of electrons is necessary in order that T F = 100 mK? B B

Show that, for a given area, the low temperature heat capacity is linear in T and
independent of the number of electrons.

At zero temperature, where the Fermi distribution becomes a step function, the
number of particles in the system may be expressed as
εF
N = α ∫ g ( ε ) dε
0

and since in 2d the density of states is


mA
g (ε ) = ,
2π 2
independent of energy, N is given by
α mA
N= εF
2π 2

Answers 2-11
so that
2π 2 N
εF = ,
αm A
and the Fermi temperature TF = ε F k is
2π 2 N
TF = .
α mk A
The areal density corresponding to a given Fermi temperature is
N α mkTF
= .
A 2π 2
For electrons we have α = 2, m = 9.11×10 –31 kg, so for a Fermi temperature of 0.1 K
P P

we have
N 2 × 9.11× 10−31 × 1.38 ×10−23 × 0.1
=
2π × (1.06 × 10−34 )
2
A

= 3.56 ×1013.
So the number density is 3.56×10 13 electrons per square meter.
P P

The heat capacity is found by differentiating the internal energy.

The simple treatment of the next part is to use the argument that at low temperatures
only a fraction T/T F of the electrons will be excited so that the internal energy is
B B

essentially this fraction of the classical (equipartition) internal energy NkT in two
dimensions. Thus
T2
E ~ Nk
TF
and so, upon differentiation,
T
CV ~ Nk .
TF
Thus in the low-temperature limit the heat capacity is linear in T, as in the 3-d case.

The point here is that in 2-d the Fermi temperature is proportional to N (over A) so
this N cancels with that in the expression for C V , making the heat capacity
B B

independent of N:
2π 2 N
TF =
α mk A
so that
T
CV ~ Nk
TF
α mk 2 A
~ T
2π 2
and thus we see that the low-temperature heat capacity of a 2-d electron gas is
independent of the number of electrons.

A better treatment of the low-temperature heat capacity would use the Sommerfeld
expansion. This gives the internal energy as

Answers 2-12
π2 kT 2
E = const + N
6 TF
and the heat capacity is then
π2 T
CV = Nk .
6 TF
This is the correct low temperature limiting behaviour, but as we saw in Problem 2.6,
the Sommerfeld method is incapable of treating the higher-order terms in 2-d.
Observe this result demonstrated that the previous approximation gives the correct
result to within a numerical correction factor. Then that argument about the N from
the Fermi temperature cancelling still applies and again we find that the heat capacity
will be independent of N.

18 Use the Sommerfeld expansion method of Section 2.4.3 to show that the Fermi-
Dirac distribution function may be approximated, at low temperatures, by
1 π2
~ Θ ( µ − ε ) − ( kT ) δ ′ ( ε − µ ) + …
2
(ε − µ ) kT
e +1 6
where Θ is the unit step function and δ ′ is the first derivative of the Dirac delta
function.

Can you write down the general term of the series?

According to the results of Section 2.4.3 integrals of the form



ϕ (ε )
I =∫ (ε − µ ) kT
dε ,
0 e +1
where φ(ε) is an arbitrary function, are approximated by
( kT )
n

d nψ
I = ∑ In .
n=0 n ! dε n ε =µ

Here ψ is the integral of φ:


ψ ( ε ) = ∫ ϕ ( ε ) dε .
We have the expression for the Sommerfeld expansion as
ϕ (ε ) ( kT )
∞ n

d nψ
∫ e( ε − µ ) kT
dε = ∑ I n
0 +1 n=0 n ! dε n ε =µ

and we can write this out term by term


ϕ (ε ) ( kT ) dϕ ( kT ) ( kT )
∞ µ 2 4 n
d 3ϕ d n −1ϕ
∫ e(ε −µ ) kT + 1 ∫
dε = ϕ ( ε ) dε + I2 + I4 3 + … + I n n −1 + … .
0 0
2 dε 4! dε n! dε
The right hand side could be written as
( kT )
∞ 2 ∞

∫ Θ ( µ − ε ) ϕ (ε ) dε +
0
2
I 2 ∫ δ ′ ( ε − µ )ϕ ( ε ) dε +
0

( kT ) ( kT )
4 ∞ n ∞
I 4 ∫ δ ′′′ ( ε − µ )ϕ ( ε ) dε + … + In ∫ δ ( (ε − µ )ϕ (ε ) dε + …
n −1)
+
4! 0
n! 0

Answers 2-13
where Θ is the step function and δ (n) is the nth derivative of the delta function. Then
P P

since this result holds for arbitrary functions φ(ε) we can equate the insides of the
integral to give
( kT ) ( kT )
2 4
1
= Θ(µ −ε ) + I 2δ ′ ( ε − µ ) + I 4δ ′′′ ( ε − µ ) + … +
e(
ε − µ ) kT
+1 2 4!
( kT )
n

+ (ε − µ ) + …
I nδ (
n −1)

n!
If we substitute for the values of I n in the first few terms we obtain
B B

1 π2 7π 4
= Θ(µ −ε ) + ( kT ) δ (ε − µ ) + ( kT ) δ ′′′ (ε − µ ) + … ,
2 4
( ε − µ ) kT

e +1 6 360
as required. And the general term of the expansion is given by
In
( kT ) δ ( n−1) (ε − µ ) = ( 2 − 22−n ) ζ ( n )( kT ) δ ( n−1) ( ε − µ ) .
n n

n!
Observe that the first term of the expansion, the step function, is the zero-temperature
distribution function. The higher order terms give the finite-temperature corrections to
this.

Answers 2-14
PH4211 Statistical Mechanics

Problem Sheet 3 — Answers


1 Show that the Joule-Kelvin coefficient is zero for an ideal gas.

The Joule-Kelvin coefficient is given by


⎧⎪ ∂V
1 ⎫⎪
µJ = ⎨T −V ⎬ .
⎩⎪ ∂T p
cp ⎭⎪
For an ideal gas we have pV = NkT, so upon differentiating:
∂V Nk
=
∂T p p
so that
1 ⎧ Nk ⎫
µJ = ⎨T −V ⎬ .
cp ⎩ p ⎭
And since pV = NkT, the bracket is zero. Thus the Joule-Kelvin coefficient is zero.

2 Derive the second virial coefficient expression for the Joule-Kelvin coefficient

µJ =
2T dB2 T RS b g
− 2
B Tb g UV
5k dT T T W
and find the inversion temperature for the square well potential gas.

In the limit of small B 2 (T) we have the equivalent expression for the virial expansion
B B

NkT
V= + NB2 (T ) .
p
And we have
1 ⎧⎪ ∂V ⎫⎪
µJ = ⎨T −V ⎬ .
c p ⎪⎩ ∂T p ⎭⎪
So differentiating the expression for V gives
∂V Nk dB (T )
= +N 2 .
∂T p p dT
We substitute this into the expression for µ J to give
B B

NT ⎧ dB2 (T ) B2 (T ) ⎫
µJ = ⎨ − ⎬.
c p ⎩ dT T ⎭
Now for an ideal gas the thermal capacity at constant volume is given by
5
c p = Nk
2
and we adopt this as an acceptable approximation in this weakly interacting case. We
then obtain the result

Answers 3-1
2T ⎧ dB2 (T ) B2 (T ) ⎫
µJ = ⎨ − ⎬.
5k ⎩ d T T ⎭

The second virial coefficient for the square well gas is given by
2
{
B2 (T ) = πσ 3 1 − ( R 3 − 1)( eε kT − 1) .
3
}
To find the inversion temperature we require to solve the equation
dB2 (T ) B2 (T )
− =0
dT T
for T. Now using the square-well expression for B will result in an insoluble
equation. A numerical solution can be found if R is specified, or an approximate
solution may be found. We shall look for a solution for the inversion temperature in
the (realistic) limiting case where there is a long-range, but weak interaction. In other
words we are considering the case where R is large and ε is small. When R is very
much greater than unity then B may be approximated by
2
{
B2 (T ) = πσ 3 1 − R 3 ( eε kT − 1)
3
}
and if ε is small then we may expand the exponential so that
2 ⎧ R 3ε ⎫
B2 (T ) = πσ 3 ⎨1 − ⎬.
3 ⎩ kT ⎭
This expansion is valid for temperatures T much greater than ε/k. This will be justified
a posteriori.

This expression for B 2 (T) is actually equivalent to that obtained for the van der Waals
B B

gas; the general behaviour is


a
B2 (T ) = b − .
kT
We can then quote the result (derived in the next Problem)
2a
TI =
bk
2
= R 3ε .
k
This satisfies the requirement on the expansion of the exponential, that kT >> ε since
R is assumed to be large.

a 2a
3 For the van der Waals gas show that TB = and TI = .
bk bk

We obtained the second virial coefficient for the van der Waals gas to be
a
B2 (T ) = b − .
kT
The Boyle temperature is that for which B 2 (T) = 0. So solving the above expression
B B

for zero gives

Answers 3-2
a
TB = .
bk
The inversion temperature is that for which
dB2 (T ) B2 (T )
− =0.
dT T
Now differentiating B 2 (T) gives
B B

dB2 (T ) a
= 2
dT kT
so that we now require to solve
a b a
2
= − 2.
kT T kT
This has solution
2a
TI = .
bk

4 Show that for a van der Waals fluid the critical parameters are given by
a 8a
Vc = 3 Nb, pc = 2
, kTc = . Express these critical quantities in terms of the
27b 27b
microscopic interaction (Lennard-Jones) parameters ε and σ.

The van der Waals equation of state is


⎛ N2 ⎞
⎜ p + a ⎟ (V − Nb ) = NkT
⎝ V2 ⎠
and this can be rearranged to give
NkT N2
p= −a 2 .
(V − Nb ) V
Then differentiating this gives
∂p 2aN 2 NkT
= − =0
∂V (V − Nb )
3 2
V
∂2 p 2 NkT 6aN 2
= − 4 =0.
∂V 2 (V − Nb )3 V
Solve the simultaneous equations to give T c and V c : kTc = 8a 27b, Vc = 3 Nb and then
B B B B

substitute in to get pc = a 27b .


2

In the mean field discussion the repulsive and the attractive parts of the inter-particle
interaction were treated separately. Within the spirit of this let us consider how the
two parameters of the van der Waals equation might be related to the two parameters
of the Lennard-Jones inter-particle interaction potential. The repulsion is strong;
particles are correlated when they are very close together. We accounted for this by
saying that there is zero probability of two particles being closer together than σ.
Then, as in the hard core discussion of Section 3.2.1, the region of co-ordinate space
is excluded, and the form of the potential in the excluded region (U(r) very large)
does not enter the discussion. Thus just as in the discussion of the hard core model,
the excluded volume will be

Answers 3-3
2
Vex = Nπσ 3 ,
3
one half of the total hard core volume.

The attractive part of the potential is weak. Here there is very little correlation
between the positions of the particles; we therefore treat their distribution as
approximately uniform. The mean interaction for a single pair of particles E1 is then

1
E1 = ∫ 4π r 2U ( r ) dr

1

⎧⎪⎛ σ ⎞12 ⎛ σ ⎞6 ⎫⎪ 8πσ 3
= ∫ 4π r 4ε ⎨⎜ ⎟ − ⎜ ⎟ ⎬ dr = −
2
ε.
Vσ ⎩⎪⎝ r ⎠ ⎝ r ⎠ ⎭⎪ 3V
Now there are N(N – 1)/2 pairs, each interacting through U(r), so neglecting the 1, the
total mean energy per particle is
E = E1 N 2
4πσ 3 N
ε. =−
3 V
In the van der Waals equation it is the derivative of this quantity we require. Thus we
find
2
d E 4 3⎛ N ⎞
N = πσ ⎜ ⎟ ε .
dV 3 ⎝V ⎠
These results give the correct assumed N and V dependence of the parameters used in
the previous section. So finally we identify the van der Waals parameters a and b as
4
a = πσ 3ε
3
2
b = πσ 3 .
3

5 Show that, for a van der Waals fluid, the combination of critical parameters
p c V c /NkT c takes the universal value 3/8.
B B B B B B

The critical parameters for the van der Waals equation are given by
a 8a
Vc = 3 Nb, pc = 2
, kTc = .
27b 27b
Then p c V c /NkT c is given by
B B B B B B

pcVc a 1 27b
= 2
3 Nb
NkTc 27b N 8a
= 3 8.

6 Show that for the Dieterici fluid the critical parameters are given by
a a
Vc = 2 Nb, pc = 2 2 , kTc = , and the universal combination p c V c /NkT c has the
4b e 4b
B B B B B B

value 2/e 2 = 0.271. P P

Answers 3-4
The Dieterici equation of state is
Na

p (V − Nb ) = NkTe kTV

and this can be arranged to give


Na

NkTe kTV
p= .
V − Nb
Then differentiating this gives
Na

∂p
2 {
aN (V − Nb ) − kTV 2 }
Ne kTV
= 2
∂V V (V − Nb )
and
Na

∂2 p
∂V 2
=
Ne kTV
kTV (V − Nb )
4 3 {
2k 2T 2V 4 + N 2 a 2 (V − Nb ) − 2 NakTV ( N 2b 2 − 3 NbV + 2V 2 ) .
2
}
For the critical point, these must be set equal to zero and the equations solved to give
T c and V c . Solution of the equations thus gives
B B B B

a
kTc = , Vc = 2 Nb
4b
and we then substitute these into the equation of state to give:
a
pc = 2 2 .
4b e

Using these values for the critical parameters we than find p c V c /NkT c is given byB B B B B B

pcVc a 1 4b
= 2 2 2 Nb
NkTc 4b e N a
= 2 e 2 ≈ 0.271.

7 Show that the partition function for an interacting gas may be expressed as
− ∑U ( qi ,q j ) kT
1
Z = Z id N ∫ e i< j d3N q
V
where Z id is the partition function for a non-interacting gas. In terms of this
B B

expression explain why the partition function of a hard sphere gas might be
approximated by
⎛ V − Nb ⎞
N

Z = Z id ⎜ ⎟ .
⎝ V ⎠

The partition function for an interacting gas is given by


⎛ 2 ⎞
−⎜ ∑ i + ∑U ( qi , q j ) ⎟ kT
p
1 ⎜ i 2 m i< j ⎟
3N ∫
Z= e ⎝ ⎠
d 3 Np d 3 Nq .
N !h
This may be factorised, separating the p and q integrations:

Answers 3-5
⎛ ⎛ ⎞
pi2 ⎞
1 −⎜ ∑



kT ∑ (
− ⎜ U qi , q j
⎜ i< j
) ⎟⎟ kT

3N ∫
Z= e ⎝ i 2m ⎠
d p ∫e
3N ⎝ ⎠
d 3 Np d 3 Nq .
N !h
But the ideal gas partition function is
pi2
−∑
VN
3N ∫
2 mkT 3 N
Z id = e i
d p,
N !h
so that Z may then be written as
− ∑U ( qi , q j ) kT
1
N ∫
Z = Z id e i< j
d3 N q .
V

If the interaction energy U(q i , q j ) were zero, then the integral would become V N as
B B B B P P

one is simply integrating (unity) over the 3N dimensional space. Now for the hard
sphere interaction the integrand is unity so long as the particles do not approach closer
than the dimension σ. But if the particles did approach closer then the interaction
energy would be infinite and the integrand would be zero. So when the q integrals are
performed over all space the effect of this interaction appears simply to remove the
hard core volume. And in this way the integral would be (V − Nb ) so that the
N

partition function would then be


⎛ V − Nb ⎞
N

Z = Z id ⎜ ⎟ .
⎝ V ⎠

8 Show that the approximate partition function for the hard sphere gas in the previous
question leads to the equation of state p (V − Nb ) = NkT . This is sometimes called
the Clausius equation of state. Give a physical interpretation of this equation.

Show that the first few virial coefficients are given by B2 (T ) = b, B3 (T ) = b 2 ,


B4 (T ) = b3 , etc. These virial coefficients are independent of temperature. Discuss
whether this is a fundamental property of the hard sphere gas, or whether it is simply
a consequence of the approximated partition function.

To find the equation of state we must differentiate kTlnZ with respect to volume. Now
the logarithm of approximate partition is
ln Z = ln Z id + N ln (V − Nb ) − N ln V .
For the ideal gas partition function we have
⎡⎛ mkT ⎞3 2 Ve ⎤
ln Z id = N ln ⎢⎜ 2 ⎟ ⎥,
⎢⎣⎝ 2π ⎠ N ⎥⎦
which may be written as
ln Z id = N ln V + terms independent of V .
Then
ln Z = N ln (V − Nb ) + terms independent of V
as the NlnV terms cancel. We then have for the pressure

Answers 3-6
∂ ln Z NkT
p = kT = ,
∂V T ,N V − Nb
so that
p (V − Nb ) = NkT
as required.

The physical interpretation of this is as an the equation of state for an ideal gas, but
with a reduced volume available, because of the finite volume of the (otherwise non-
interacting) particles.

In order to find the virial coefficients we must expand p/kT in powers of N/V. Now
p N
= .
kT V − Nb
(Observe that the right hand side is independent of temperature, so we know that the
virial coefficients will be temperature-independent before we start.)

We write p/kT as
−1
p N⎛ N⎞
= ⎜1 − b ⎟
kT V ⎝ V ⎠
and this is then expanded as
p N⎛ ⎞
2 3
N 2⎛ N ⎞ 3⎛ N ⎞
= ⎜1 + b + b ⎜ ⎟ + b ⎜ ⎟ + … ⎟ .
kT V ⎜⎝ V ⎝V ⎠ ⎝V ⎠ ⎟

From this we identify directly
B2 (T ) = b, B3 (T ) = b 2 , B4 (T ) = b3 , etc .

The argument about excluded volume in obtaining the partition function for the hard
sphere gas is an approximation. In reality imposing the boundary condition
qi − q j > σ for all i and j is more complex. However, by examining the structure of
the configuration integral
⎛ ⎞
−⎜ ∑U ( qi , q j ) ⎟ kT
1 ⎜ ⎟
QN = N ∫ e ⎝ i< j ⎠
d3N q
V
we note that the exponent is independent of temperature since U/kT is either zero or
infinite. Thus Q N is independent of temperature and from this it follows that the virial
B B

coefficients will be temperature-independent. In other words, the virial coefficients


for the hard sphere gas will be temperature-independent, irrespective of any
approximations made.

⎡A full calculation of the virial coefficients of the hard sphere gas gives
B2 (T ) = b, B3 (T ) = 85 b 2 , B4 (T ) = 0.29b3
where b = 2πσ 3 3 . So the precise values of the virial coefficients are not quite those
predicted from the approximated Claussius equation of state. ⎦

Answers 3-7
9 For a general interatomic interaction potential U(r) we may define an effective hard
core dimension d by U(d) = kT. What is the significance of this definition? Show
that for the Lennard-Jones potential of Section 3.2.4, d is given by
16
⎧⎪ 2 ⎫⎪
d =σ ⎨ ⎬ .
⎪⎩1 + 1 + kT ε ⎭⎪
Plot this to demonstrate that d is a very slowly varying function of temperature.

If you have access to a symbolic mathematics system such as Mathematica or Maple,


show that at low temperatures
⎛ 1 kT 19 ⎛ kT ⎞
2

d ~ σ ⎜1 − + − … ⎟ .
⎜ 24 ε 1152 ⎜⎝ ε ⎟⎠ ⎟
⎝ ⎠
So how high must the temperature be so that d differs appreciably from its zero
temperature value?

How does d vary for the hard core and the square well potentials of Sections 3.2.2 and
3.2.3?

With d defined by U(d) = kT, this corresponds to the distance of closest approach of
particles with kinetic energy kT. So from equipartition, this is essentially the mean
distance of closest approach of particles of a gas at the temperature T.

For the Lennard-Jones potential, d then satisfies


⎪⎧⎛ σ ⎞ ⎛ σ ⎞ ⎪⎫
12 6

4ε ⎨⎜ ⎟ − ⎜ ⎟ ⎬ = kT .
⎪⎩⎝ d ⎠ ⎝ d ⎠ ⎪⎭
This may be solved for d by setting x = (σ d ) . Then x satisfies
6

x 2 − x − kT 4ε = 0 ,
which has solution
1 ± 1 + kT ε
x= .
2
We must keep the positive square root since this gives a positive value for x. Then we
find that
16
⎧⎪ 2 ⎫⎪
d =σ ⎨ ⎬
⎪⎩1 + 1 + kT ε ⎭⎪
as required. This is plotted below.
dês
kT
ÅÅÅÅÅÅ ÅÅ
20 40 60 80 100 e
0.95

0.9
0.85

0.8

0.75

Answers 3-8
The figure shows that d varies quite slowly, particularly at high temperatures.

Series expansion of d gives


⎛ 1 kT 19 ⎛ kT ⎞
2

d ~ σ ⎜1 − + ⎜ ⎟ −…⎟
⎜ 24 ε 1152 ⎝ ε ⎠ ⎟
⎝ ⎠
so that T must be many times greater than ε/k for d to differ appreciably from its zero
temperature value of σ.

For both the hard core and the square well potentials the effective hard core
dimension d is a constant, equal to the actual the hard core dimension σ, independent
of temperature.

10 The one dimensional analogue of the hard sphere gas is an assembly of rods
constrained to move along a line (the Tonks model). For such a gas of N rods of
length l confined to a line of length L, evaluate the configuration integral Q N . Show B B

that in the thermodynamic limit the equation of state is


f ( L − Nl ) = NkT
where f is the force, the one dimensional analogue of pressure.

Comment on the similarities and the differences from the hard sphere equation of
state mentioned in Problem 3.8 (Clausius equation) and the van der Waals equation of
state.

In one dimension the configuration integral is given by


⎛ ⎞
− ⎜ ∑U ( xi , x j ) ⎟ kT
1 ⎜ ⎟
QN = N ∫ e ⎝ i< j ⎠
dx1dx2 … dxN .
L
This may be simplified through the change of variables X i = x i – x i–1 . This gives
B B B B B B

⎛ ⎞
1 −⎜
⎜ ∑ U ( X i ) ⎟ kT

N ∫
QN = ⎝
e i ⎠
dX 1dX 2 … dX N
L
with an appropriate choice of boundary conditions (cyclic boundary conditions). But
in this form the integrals over the different variables are independent; the exponential,
and thus the integral, can be factorised. And since the integral over each X will be the
same, we obtain
N
⎛1 ⎞
QN = ⎜ ∫ e −U ( X ) kT dX ⎟
⎝L ⎠
When X > l we have U = 0 so that the exponential is unity. And when X < l then U is
infinite so that the exponential is zero. So in this case the excluded volume argument
holds true and
−U ( X ) kT
∫e dX = L − Nl
so that
⎛ L − Nl ⎞
N

QN = ⎜ ⎟ .
⎝ L ⎠

Answers 3-9
The partition function for the ideal one-dimensional gas is
N
1 ⎛L⎞
Z id = ⎜ ⎟
N !⎝ Λ ⎠
Then using the calculated configuration integral the partition function is
1 ⎛ L − Nl ⎞
N

Z= ⎜ ⎟ .
N !⎝ Λ ⎠
The logarithm of Z is
ln Z = N ln ( L − Nl ) + terms independent of L .
The force in the one dimensional gas is given by
∂ ln Z
f = kT
∂L T ,N

and thus we find


NkT
f =
L − Nl
or
f ( L − Nl ) = NkT .

This is the 1d analogue of the Clausius equation of state for the hard sphere gas in 3d.
However this is an exact equation of state; the Clausius equation is an approximation.
The van der Waals equation is similar. But in that case the long-distance attractive
interactions also are taken into account by modifying the pressure.

11 Compare the square well and the van der Waals expressions for the second virial
coefficient. Show that they become equivalent when the range of the square well
potential tends to infinity while its depth tends to zero.

The square well second virial coefficient is given by


2
{
B2sq (T ) = πσ 3 1 − ( R 3 − 1)( eε
3
kT
}
− 1) ,

while that for the van der Waals equation is


a
B2VW (T ) = b − .
kT

When ε is small the exponential of the square well second virial coefficient can be
expanded:
2 ⎧⎪ ⎛ ε 1 ⎛ ε ⎞2 ⎞ ⎫⎪
B2sq (T ) ≈ πσ 3 ⎨1 − ( R 3 − 1) ⎜ + ⎜ ⎟ + …⎟⎬
3 ⎜ kT 2 ⎝ kT ⎠ ⎟
⎪⎩ ⎝ ⎠ ⎭⎪
at the same time the range of the well become large so that the 1 may be neglected in
comparison with the R 3 . Then
P P

2 ⎧⎪ ⎛ ε 1 ⎛ ε ⎞2 ⎞ ⎫⎪
B2 (T ) ≈ πσ 3 ⎨1 − R 3 ⎜ + ⎜ ⎟ + … ⎟⎬ .
3 ⎜ kT 2 ⎝ kT ⎠ ⎟
⎩⎪ ⎝ ⎠ ⎭⎪
In the limit that ε → 0, R → ∞, while R 3 σ remains finite this then becomes
P P

Answers 3-10
2 ⎧ ε ⎫
B2 (T ) ≈ πσ 3 ⎨1 − R 3 ⎬.
3 ⎩ kT ⎭
And this has the form of the square well second virial coefficient, where
2 2
b = πσ 3 , a = πσ 3 R 3ε .
3 3
Thus we have shown that the second virial coefficient of van der Waals equation of
state is equivalent to that arising from the square well interaction in the limit of a
long-range but weak attraction.

Answers 3-11
PH4211 Statistical Mechanics

Problem Sheet 4 — Answers

1 Obtain an expression for the Helmholtz free energy for the Weiss model in zero
b g
external magnetic field, in terms of the magnetisation. Plot F M for T > TC ,
T = TC and T < TC .

The Helmholtz free energy is defined as


F = E − TS .
The internal energy is given by
E = − ∫ B.dM .
The magnetic field, in the Weiss model, is the sum of the applied field and the local
(mean) field
B = B0 + b .
We shall write the local field in terms of the critical temperature:
Nk
b = 2 Tc M .
M0
Integrating up the internal energy we obtain
2
NkTc ⎛ M ⎞
E = − B0 M − ⎜ ⎟ .
2 ⎝ M0 ⎠
For the present we will consider the case where there is no external applied field.
Then B0 = 0 , and in terms of the reduced magnetisation m = M M 0 (the order
parameter) the internal energy is
NkTc 2
E=− m .
2
Now we turn to the entropy. This is most easily obtained from the definition
S = − Nk ∑ p j ln p j
j

where p j are the probabilities of the single-particle states. It is simplest to treat spin
B B

one half, which is appropriate for electrons. Then there are two states to sum over:
S = − Nk ⎡⎣ p↑ ln p↑ + p↓ ln p↓ ⎤⎦ .
Now these probabilities are simply expressed in terms of m, the fractional
magnetisation
1+ m 1− m
p↑ = and p↓ =
2 2
so that the entropy becomes
Nk
S= ⎡ 2ln 2 − (1 + m ) ln (1 + m ) − (1 − m ) ln (1 − m ) ⎤⎦ .
2 ⎣
We now assemble the free energy F = E − TS , to obtain

Answers 4-1
Nk
F =−
2
{
Tc m 2 + T ⎡⎣ 2ln 2 − (1 + m ) ln (1 + m ) − (1 − m ) ln (1 − m ) ⎤⎦ . }
This is plotted for temperatures less than, equal to and greater than the critical
temperature.

Landau
free F T/Tc = 1.2 T /T c = 1
energy

T/Tc = 0.8

m
−1
 −0.5
 0.5 1

Landau free energy for Weiss model ferromagnet

The occurrence of the ferromagnetic phase transition can be seen quite clearly from
this figure. For temperatures above T c we see there is a single minimum in the Landau
B B

free energy at m = 0, while for temperatures below T c there are two minima either
B B

side of the origin. The symmetry changes precisely at T c . There the free energy has
B B

flattened, meaning that m may make excursions around m = 0 with negligible cost of
free energy – hence the large fluctuations at the critical point.

2 Show that F =
Nk RSb gT
T − TC m2 + C m4 +
UV
for the Weiss model ferromagnet in
2 T 6 W
the limit of small m. Explain the appearance of T C in the m 4 term.
B B P P

The free energy is given by


Nk
F =−
2
{
Tc m 2 + T ⎡⎣ 2 ln 2 − (1 + m ) ln (1 + m ) − (1 − m ) ln (1 − m ) ⎤⎦ . }
In order to expand this in powers of m we must expand the logarithms:
m 2 m3 m 4
log (1 ± m ) = ± m − ± − ±….
2 3 4
By expanding and collecting terms we obtain
Nk ⎧ T 4 T 6 ⎫
F= ⎨−2T ln 2 + (T − TC ) m + m + m + …⎬ .
2

2 ⎩ 6 15 ⎭
In the vicinity of the critical point the first term may be approximated by 2T C ln2; this B B

is a constant and so it may be ignored. By a similar argument the T in the m 4 and the P P

m 6 term should be replaced by T C as T is restricted to the vicinity of the critical point.


P P B B

Then we have the expression for the free energy:

Answers 4-2
Nk ⎧ TC 4 ⎫
F= ⎨(T − TC ) m + m + …⎬
2

2 ⎩ 6 ⎭
as required.

3 Show that d 2 F dϕ 2 > 0 below T c at the two roots ϕ = ± − F2 2 F4 in the Landau


B B

model. Show that d 2 F dϕ 2 < 0 below T c and d 2 F dϕ 2 > 0 above T c at the single
B B B B

root ϕ = 0. What is the physical meaning of this?

The Landau free energy is


F = F2ϕ 2 + F4ϕ 4
so that its second derivative is
d2 F
= 2 F2 + 12 F4ϕ 2 .
dϕ 2
For the two roots ϕ = ± − F2 2 F4 below the transition
ϕ 2 = − F2 2 F4
so that
d2 F F
= 2 F2 + 12 F4 2
dϕ 2
2 F4
= −4 F2 .
Now F 2 is negative for temperatures below the transition:
B B

F2 = a (T − TC ) ,
so we see that below the transition the second derivative of F is positive for the two
roots ϕ = ± − F2 2 F4 .

But at the root ϕ = 0 then


d2 F
= 2 F2 .
dϕ 2
And since F 2 is negative then the second derivative is negative for the root at ϕ = 0 .
B B

From this we conclude that below the critical temperature the roots ϕ = ± − F2 2 F4
are minima while the root at ϕ = 0 is a maximum.

Above the critical temperature there is the single root at ϕ = 0 . Then


d2 F
= 2 F2
dϕ 2
but now F 2 is positive. So above the critical point the single root at ϕ = 0 is a
B B

minimum.

4 In the Landau theory of second order transitions calculate the behaviour of the
bg
order parameter below the critical point, ϕ T , when the sixth order term in the free

Answers 4-3
energy expansion is not discarded. What influence does this term have on the critical
exponent β ? Comment on this.

The sixth order expression for the free energy is


F = F2ϕ 2 + F4ϕ 4 + F6ϕ 6 .
The equilibrium state is determined by minimising F: differentiate and set equal to
zero
dF
= 2 F2ϕ + 4 F4ϕ 3 + 6 F6ϕ 5 = 0 .

One root of this equation is φ = 0. The other roots are roots of the equation
F2 + 2 F4ϕ 2 + 3F6ϕ 4 = 0 .
And these roots are
− F4 ± F42 − 3F2 F6
ϕ =
2
,
3F6
but we must take the positive square root since φ 2 must be positive – note that F 2 is
P P B B

negative. Then the values of φ at the roots are


− F4 + F42 − 3F2 F6
ϕ =± .
3F6
The singular behaviour arises from the temperature dependence of F 2 , which goes B B

through zero at the critical point:


F2 = a (T − Tc ) .
Then the (singular part of the) temperature variation of the order parameter is given
by
− F4 + F42 − 3a (T − Tc ) F6
ϕ =± .
3F6
To find the order parameter critical exponent we must expand φ in powers of F 2 or B B

T – Tc:
B B

F2 3F6 F 3 63F6 F5
ϕ= − + − 25 + − 2 9 +….
2 F4 8 2 F4 128 2 F4
(This expansion may be obtained using a symbolic algebra system such as
Mathematica or Maple, or it may be evaluated by hand as indicated below.) The
temperature dependence of the order parameter is then given by
a (T − Tc ) 3F6 a 3 (T − Tc ) a 5 (T − Tc )
3 5
63F6
ϕ= − + − + − +… .
2 F4 8 2 F45 128 2 F49
The first term gives the leading order singularity. We see that it gives the order
parameter critical exponent as β = ½. Observe that this leading order term is
independent of F 6 ; indeed the series is seen to be an expansion in powers of F 6 . So
B B B B

the critical exponent is unaffected by the inclusion of the F 6 term.


B B

⎡ In order to perform the power series expansion of φ by hand we write the inner
square root as

Answers 4-4
12
⎛ 3F F ⎞
F − 3F2 F6 = F4 ⎜ 1 − 2 2 6 ⎟
4
2

⎝ F4 ⎠
⎛ 3F F 9 F 2 F 2
= F4 ⎜ 1 − 2 26 − 2 4 6 + …
⎝ 2 F4 8 F4
2 2
3F F 9 F F
= F4 − 2 6 − 2 3 6 + …
2 F4 8 F4
The expression for φ is then
F2 3F22 F6 9 F23 F62
ϕ =± − − − −…
2 F4 8F43 16 F45
F2 3F F
=± − 1 + 2 26 + … .
2 F4 4 F4
In the case that F 6 is zero we have the familiar expression
B B

F
ϕ =± − 2
2 F4
a (T − Tc )

2 F4
so we obtain the critical exponent β = ½ .

Now incorporating the higher order terms, we have


F 3F F
ϕ = ± − 2 1 + 2 26 +… . ,
2 F4 4 F4
and since F 2 is small we can expand the second square root:
B B

F ⎛ 3F F
ϕ = ± − 2 ⎜1 + 2 2 6 + … .
2 F4 ⎝ 8 F4

5 A ferroelectric has a free energy of the form


b g
F = α T − Tc P 2 + bP 4 + cP 6 + DxP 2 + Ex 2
where P is the electric polarisation and x represents the strain. Minimise the system
with respect to x. Under what circumstances is there a first order phase transition for
this system?

The equilibrium value of the strain is x is found by minimising F with respect to


variations in x:
∂F
= DP 2 + 2 Ex = 0 ,
∂x
the solution of which is
DP 2
x=− .
2E
Then

Answers 4-5
D2 P4 D2 P4
DxP 2 = − , Ex 2 =
2E 4E
and at the equilibrium strain the Landau free energy is
⎛ D2 ⎞ 4
F = α (T − Tc ) P 2 + ⎜ b − ⎟ P + cP .
6

⎝ 4E ⎠
There will be a first order transition when the coefficient of P 4 becomes negative, that
P P

is when
D2
>b.
4E

6 Consider a one-dimensional binary alloy where the concentration of A atoms varies


slowly in space: x = x(z). Show that the spatial variation of x results in an additional
term in the free energy per bond of a 2ε ( dx dz ) , where a is the spacing between
2

atoms and ε is the energy parameter defined in Section 4.7.3.

When the system is inhomogeneous then the concentration of A atoms varies in space.
In this case the calculation of the bond energy is a little more complicated. The
direction from the left atom to the right atom is chosen to be along the z axis and the
‘position’ of the bond is taken as its mid point.

z − a/2 z z + a/2

The concentration is a function of position. It is assumed to vary smoothly so that a


Taylor expansion is appropriate.
2
a ∂x 1 ⎛ a ⎞ ∂ 2 x
xl = x − + ⎜ ⎟ −…
2 ∂z 2 ⎝ 2 ⎠ ∂z 2
2
a ∂x 1 ⎛ a ⎞ ∂ 2 x
xr = x + + ⎜ ⎟ +… .
2 ∂z 2 ⎝ 2 ⎠ ∂z 2
where l and r stand for left and right. The expression for elr is then given by
2 2
a 2 ⎛ ∂x ⎞ a 2 ∂2 x a4 ⎛ ∂2 x ⎞
elr = e0 + ε ⎜ ⎟ + ( ε aa − ε bb + 2ε (1 − 2 x ) ) 2 − ε ⎜ 2 ⎟ + …
2 ⎝ ∂z ⎠ 8 ∂z 8 ⎝ ∂z ⎠
where e0 is the energy per bond in the homogeneous case.

It is a fundamental assumption that the concentration x varies slowly with position; in


other words, the spatial derivatives are small. The terms in a 2 are second order in
P P

smallness; these contain the leading order contributions from the spatial variation in x.
The term in a 4 is fourth order in smallness; this may be ignored. Both terms in a 2
P P P P

must be considered as these are of the same order. However it is possible to transform

Answers 4-6
the term in ∂ 2 x ∂z 2 through integration by parts, to a term in ( ∂x ∂z ) plus a surface
2

term that may be ignored.

The key point is that the expression for elr is a function of position z through the
dependence of x upon position. Thus elr is an energy density and the total energy is
found by integrating over the system, and since we are considering variations in the z
direction, this means integrating over z. The contribution to the total energy from
the ∂ 2 x ∂z 2 term is then
z2
∂2 x
E = ∫ f ( z) dz
z1
∂z 2
where
a2
f ( z) =
8
(
ε aa − ε bb + 2ε (1 − 2 x ( z ) ) )
and z 1 and z 2 are the extremities. Now the expression for E is integrated by parts to
B B B B

give
z2 z2
∂x ∂f ∂x
E = f ( z) −∫ dz .
∂z z1 z1
∂z ∂z
The first term is the surface term. This will be zero if we chose the extremities to be
away from any spatial variation in concentration. In the second term the first
derivative may be re-expressed using the chain rule for differentiation
∂f ∂f ∂x
=
∂z ∂x ∂z
and then E is given by
z2 2
∂f ⎛ ∂x ⎞
E = −∫ ⎜ ⎟ dz .
z1
∂x ⎝ ∂z ⎠
The integrand is the energy density. We see it has been transformed into a term in
( ∂x ∂z ) . Then since
2

∂f a2
=− ε
∂x 2
it follows that
z 2
a 2ε 2 ⎛ ∂x ⎞
2 ∫z1 ⎝ ∂z ⎠
E= ⎜ ⎟ dz ,

corresponding to an energy density of


2
a 2ε ⎛ ∂x ⎞
⎜ ⎟ .
2 ⎝ ∂z ⎠
We add this to the original term in ( ∂x ∂z ) to give the result
2

2
⎛ ∂x ⎞
elr = e0 + a ε ⎜ ⎟ . 2

⎝ ∂z ⎠

Answers 4-7
We now relax the restriction that the variation in concentration is in the z direction.
Then the derivative ∂x ∂z becomes the gradient of x and the second derivative
becomes the Laplacian, thus
elr = e0 + a 2ε ( ∇x ) .
2

This gives the additional free energy per bond when there is a spatial variation in the
concentration.

7 Show that in the vicinity of the critical point the free energy of the binary alloy may
be written as
⎧ 2 16 ⎫
Fm = F0 + 2 Nk ⎨(T − Tc ) ( x − 12 ) + Tc ( x − 12 ) + Tc ( x − 12 ) + …⎬
2 4 6

⎩ 3 15 ⎭
Discuss the Landau truncation of this expression; in particular, explain at what term
the series may/should be terminated.

We shall write the free energy (of mixing) for the binary alloy as
Fm ( x) = Nk {2 x(1 − x)Tc + T [ x ln x + (1 − x) ln(1 − x) ]} .
This is rather like the expression for the magnet treated in Problem 2. Now we must
expand the logarithms about the point x = ½.
log x = − log 2 + 2 ( x − 12 ) − 2 ( x − 12 ) + …
2

log (1 − x ) = − log 2 − 2 ( x − 12 ) − 2 ( x − 12 ) + …
2

This gives the expansion for the free energy as


⎧1 4 32 ⎫
Fm = Nk ⎨ Tc + T ln 2 + 2 (T − Tc ) ( x − 12 ) + T ( x − 12 ) + T ( x − 12 ) + …⎬
2 4 6

⎩2 3 15 ⎭
Except in the T – T c term, we may replace T by the constant T c as we are considering
B B B B

the behaviour only in the vicinity of the critical point. Then


⎧1 4 32 ⎫
Fm = Nk ⎨ Tc + Tc ln 2 + 2 (T − Tc ) ( x − 12 ) + Tc ( x − 12 ) + Tc ( x − 12 ) + …⎬ .
2 4 6

⎩2 3 15 ⎭
We now identify the constant F 0 as B B

⎧1 ⎫
F0 = Nk ⎨ Tc + Tc ln 2 ⎬
⎩2 ⎭
and this enables us to write the free energy in the vicinity of the critical point as
⎧ 2 16 ⎫
Fm = F0 + 2 Nk ⎨(T − Tc ) ( x − 12 ) + Tc ( x − 12 ) + Tc ( x − 12 ) + …⎬ .
2 4 6

⎩ 3 15 ⎭
Observe that the coefficient of the fourth order term is positive. This means that the
transition is second order (for the critical concentration) and the series may be
truncated at this term. That is, because the coefficient of the fourth order term is
positive, we do not need any higher order terms.

8 Plot some isotherms of the Clausius equation of state p (V − Nb ) = NkT . How do


they differ from those of an ideal gas? Does this equation of state exhibit a critical
point? Explain your reasoning.

Answers 4-8
9 The scaling expression for the reduced free energy is given in Section 4.1.9 by
⎛ B ⎞
f (T , B ) = A t Y ⎜ D ∆ ⎟ .
2 −α

⎜ t ⎟
⎝ ⎠
Show that the heat capacity is given by
d2 f (t, B )
C~
dt 2
and hence identify α as the heat capacity critical exponent.

10 Using the scaling expression for the reduced free energy in the previous section,
show that the magnetisation is given by
df ( t , B )
M~
dB
and hence show that the order parameter exponent β is given by
β = 2 −α − ∆ .
Show that the magnetic susceptibility is given by
d2 f (t, B )
χ~
dB 2
and hence show that the susceptibility exponent g is given by
γ = 2 − α − 2∆ .

11 Show that the Landau free energy has the scaling form of Problem 4.9 above, with
α = 0.

Answers 4-9
PH4211 Statistical Mechanics

Problem Sheet 5 — Answers


bg b g
1 A random quantity has an exponential autocorrelation function G t = G 0 e −γt .
Calculate its correlation time using the usual definition.

The correlation time is the ‘width’ of the correlation function – the area divided by the
height. Thus the definition

1
G ( t ) dt .
G ( 0 ) ∫0
τc =

So for the exponential correlation function



τ c = ∫ e −γ t dt
0

e −γ t 1
= = .
−γ 0
γ
The correlation time is simply 1/γ – the time constant of the exponential.

2 Show that the autocorrelation function of a periodically varying quantity


m ( t ) = m cos ωt is given by
m2
G (t ) = cos ωt .
2
Show that the autocorrelation function is independent of the phase of m(t). In other
words, show that if m ( t ) = m cos (ωt + ϕ ) , then G(t) is independent of ϕ.

The autocorrelation function is defined by


G ( t ) = m (τ ) m (τ + t ) .
It is convenient here to regard the average as a time average. That is, we average over
τ so that G(t) is calculated as
T
1
G (t ) = m (τ ) m (τ + t ) dτ .
2T −∫T
But actually we don’t need to do any integration, as we shall see. We star from
m ( t ) = m cos ωt , so that
G ( t ) = m (τ ) m (τ + t )
= m cos (ωτ ) m cos ω (τ + t ) .
We use the trig identity cos ( x + y ) = cos x cos y − sin x sin y , so that
G ( t ) = m 2 cos 2 ωτ cos ωt + m 2 cos ωτ sin ωτ sin ωt .
The average is evaluated over τ, thus
G ( t ) = m 2 cos ωt cos 2 ωτ + m 2 sin ωt cos ωτ sin ωτ .

Answers 5-1
The average of cos 2 is ½; the function varies smoothly between 0 and 1. The average
P P

of cos sin is zero; this function varies smoothly between –1 and +1. Thus we conclude
that
m2
G (t ) = cos ωt
2
as required.

Now consider the case m ( t ) = m cos (ωt + ϕ ) . Then


G ( t ) = m (τ ) m (τ + t )
= m cos (ωτ + ϕ ) m cos (ωτ + ϕ + ωt ) .
We use the same trig identity to obtain
G ( t ) = m 2 cos 2 (ωτ + ϕ ) cos ωt + m 2 cos (ωτ + ϕ ) sin (ωτ + ϕ ) sin ωt .
The average is evaluated over τ, thus
G ( t ) = m 2 cos ωt cos 2 (ωτ + ϕ ) + m 2 sin ωt cos (ωτ + ϕ ) sin (ωτ + ϕ ) .
Then, once again, the average of cos 2 is ½ and the average of cos sin is zero. And so
P P

in this case also we obtain


m2
G (t ) = cos ωt ;
2
this shows that G(t) is independent of φ as required.

3 The dynamical response function X(t) must vanish at zero times, as shown in
Fig. 5.13. What is the physical explanation of this? What is the consequence for the
step response function Φ(t)? Is this compatible with an exponentially decaying Φ(t)?

The step response function Φ(t) is proportional to the autocorrelation function of the
response variable
Φ (t ) ∝ M ( 0) M (t ) ,
from the Onsager hypothesis or the linear response derivation. Since, certainly in the
classical case, M(0) and M(t) commute, we may swap these around so that
Φ ( t ) = Φ ( −t ) .
Thus Φ(t) is an even function and its odd derivatives must vanish at t = 0. And then
since X(t) is the first derivative of Φ(t), it follows that X(0) = 0 as required.

If the odd derivatives of Φ(t) vanish at the origin, this is clearly incompatible with an
exponential decay. More precisely Φ(t) cannot decay exponentially in the vicinity of
t = 0; it can elsewhere.

4 In Section 5.3 we examined the form of the dynamical susceptibility χ(ω) that
followed from the assumption that the step response function Φ(t) decayed
exponentially. In this question consider a step response function that decays with a
gaussian profile, Φ ( t ) = χ 0 e−t 2τ . Evaluate the real and imaginary parts of the
2 2

dynamical susceptibility and plot them as a function of frequency. The real part of the
susceptibility is difficult to evaluate without a symbolic mathematics system such as

Answers 5-2
Mathematica. Compare and discuss the differences and similarities between this
susceptibility and that deduced from the exponential step response function (Debye
susceptibility).

The response function X(t) is given by minus the derivative of Φ(t), thus
d
X ( t ) = − χ 0 e −t 2τ
2 2

dt
t 2 2
= χ 0 2 e −t 2τ .
τ
The dynamical susceptibility is the Fourier transform of this

χ (ω ) = ∫ X (t ) e
iωt
dt
−∞

so that

χ
χ (ω ) = 20 ∫ te
− t 2 2τ 2 iωt
e dt .
τ −∞

Upon integration we obtain the real and imaginary parts as


⎧⎪ π ωτ ⎫⎪
χ ′ (ω ) = χ 0 ⎨1 −
2 2
ωτ e −ω τ 2 erfi ⎬
⎩⎪ 2 2 ⎭⎪
π
χ ′′ (ω ) = χ 0
2 2
ωτ e−ω τ 2 .
2
These are plotted in the figure.

χ /χ 0
1 χ'(ω)

0.8

0.6
χ''(ω)
0.4

0.2

1 2 3 4 ωτ
- 0.2

The imaginary part of the susceptibility is always positive, as required by energy


considerations. In this case the real part of the susceptibility goes negative. This is in
contrast to that of the Debye susceptibility, that remains positive at all frequencies.

Answers 5-3
5 The Debye form for the dynamical susceptibility is
1
χ ′ (ω ) = χ 0
1 + ω 2τ 2
ωτ
χ ′′ (ω ) = χ 0 .
1 + ω 2τ 2
Plot the real part against the imaginary part and show that the figure corresponds to a
semicircle. This is known as a Cole-Cole plot.

The Cole-Cole plot of the Debye susceptibility is shown below.


χ''(ω)

0.4

0.2

0.2 0.4 0.6 0.8 1 χ'(ω)

- 0.2

- 0.4

The upper half circle corresponds to positive frequencies and the lower half
corresponds to negative frequencies.

The real and imaginary parts pf the Debye susceptibility are seen to satisfy the
equation
2 2 2
⎛ χ ′′ ⎞ ⎛ χ ′ 1 ⎞ ⎛ 1 ⎞
⎜ ⎟ +⎜ − ⎟ = ⎜ ⎟ .
⎝ χ0 ⎠ ⎝ χ0 2 ⎠ ⎝ 2 ⎠
This is corresponds to a circle of radius ½ centred on χ ′ χ 0 = 1 2 .

6 Plot the Cole-Cole plot (Problem 5.5) for the dynamical susceptibility considered in
Problem 5.4. How does it differ from that of the Debye susceptibility.

In Problem 5.4 we obtained the expressions for the real and the imaginary parts of the
dynamical susceptibility
⎧⎪ π ωτ ⎫⎪
χ ′ (ω ) = χ 0 ⎨1 −
2 2
ωτ e −ω τ 2 erfi ⎬
⎪⎩ 2 2 ⎭⎪
π
χ ′′ (ω ) = χ 0
2 2
ωτ e−ω τ 2 .
2

Answers 5-4
From these we can make the Cole-Cole plot:
χ''(ω)

0.6

0.4

0.2

- 0.2 0.2 0.4 0.6 0.8 1 χ'(ω)

- 0.2

- 0.4

- 0.6

The main difference from the Debye form is that this curve passes to the left of the y
axis (high frequencies), where the real part of the susceptibility becomes negative.
Since in the high frequency limit the real and imaginary parts of the susceptibility
must both vanish, this gives the cardioid shape to the plot.

7 The full quantum-mechanical calculation of the Johnson noise of a resistor gives


hf
v 2 = 4 R hf kT ∆f .
∆f e −1
Show that this reduces to the classical Nyquist expression at low frequencies. At what
frequency will there start to be serious deviations from the Nyquist value? Estimate
the value of this frequency.

At low frequencies such that hf << kT we can expand the exponential so that
hf
v2 = 4R ∆f
∆f 1 + hf kT + … − 1
hf
= 4R ∆f
hf kT + …
and in the low frequency limit the hf cancels, giving
v2 = 4kTR∆f ,
∆f

the classical Nyquist expression.

For the Nyquist expression to be valid we require the frequency to satisfy f << kT/h,
so at room temperature (T ~ 300K) this means

Answers 5-5
1.4 ×10−23 × 300
f << −34
= 6.4 ×1012 Hz .
6.6 × 10

8 Show that for the Debye susceptibility, the relation


1 χ ′′ (ω )

χ0 = ∫ dω
π −∞
ω
holds. Demonstrate that χ ′′ vanishes sufficiently fast at ω = 0 so there is no pole in the
integral and there is thus no need to take the principal part of the integral in the
Kramers-Kronig relations.

In the Debye case


ωτ
χ ′′ (ω ) = χ 0
1 + ω 2τ 2
so that the integrand is
χ ′′ (ω ) τ
= χ0 .
ω 1 + ω 2τ 2
The ω in the denominator has been cancelled by the ω in the numerator. Thus there is
no pole in the integral.

We wish to evaluate the integral


1

χ ′′ (ω ) 1

τ
I= ∫
π −∞ ω
dω = χ 0 ∫
π −∞ 1 + ω 2τ 2
dω .

We can remove the τ by changing variables to x = ωτ, so that dω = dx/τ, and



1 dx
I = χ0
π ∫ 1+ x
−∞
2
.

The integral is standard; its value is π. Thus we have shown that I = χ 0 ; we have
B B

demonstrated that the Kramers-Krönig sum rule holds in the Debye case.

5.9 In Section 5.3.6 we considered an electrical analogue of the Langevin Equation


based on a circuit comprising an inductor and a resistor. In this problem we shall
examine a different analogue: a circuit of a capacitor and a resistor. Show that the
equation analogous to the Langevin equation, in this case, is
dV ( t ) 1
C + V (t ) = I (t ) .
dt R
Hence show that the fluctuation-dissipation result relates the resistance to the current
fluctuations through

1 1
=
R 2kT −∞ ∫ I ( 0 ) I ( t ) dt .

Answers 5-6
UNIVERSITY OF LONDON

MSci EXAMINATION 2000

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Calculators ARE permitted

PH4211A/40
© Royal Holloway and Bedford New College 2000
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js

= = h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1


Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. (a) The partition function Z for a system of N indistinguishable non-


interacting particles is often approximated by

1 N
Z= z
N!

where z is the partition function for a single particle.

Explain the arguments that lead to this approximation. [3]

(b) Given that the Helmholtz free energy F = E − TS is related to Z by

F = − kT ln Z

where the symbols have their usual meanings, show that the pressure is
given by

∂ ln Z
p = kT . [3]
∂V T ,N

(c) The partition function for a single particle moving freely in a box of
volume V may be written as
32
 2π mkT  V
z =V   =
 h
2
 Λ3

By evaluating the pressure for system of N such indistinguishable


particles, show that the equation of state corresponds to that for an ideal
gas. [3]

(d) In the van der Waals approach to an interacting gas the single particle
partition function may be approximated by

V − Vex − E
z=
kT
e
Λ3

Sketch the expected variation with separation of the inter-particle


interaction potential. Discuss how the various features of the interaction
are incorporated into the above expression for z through the quantities
Vex and E . [5]

(e) Why is this approach referred to as a mean field approximation? [2]

(f) Sketch the general form for a van der Waals isotherm and identify the
stable, metastable, and unstable regions. [4]

TURN OVER
− page 2 − PH4211A
PART
MARKS

2. (a) Outline the sequence of arguments by which one finds that two isolated
systems, when brought into thermal contact, end up in the thermodynamic
state for which

∂ ln Ω1 ∂ ln Ω 2
=
∂E ∂E

and define the terms Ω1 , Ω 2 and E in this expression. [4]

(b) Write down an expression for entropy in terms of Ω. How does the above
equation imply the equalisation of the temperatures of the two systems? [4]

(c) Now consider a small sub-system of a large isolated system. The total
energy of the isolated system is Et . The sub-system can exchange
thermal energy with the large system. When the sub-system is in a
microstate of energy E the entropy of the combined system may be
expressed as

∂S E 2 ∂ 2 S
S = S ( Et ) − E + − ....
∂E 2 ∂E 2

Justify the structure of this expression. [4]

(d) Show how the above result leads to the Boltzmann distribution function
(otherwise known as the Boltzmann factor). [4]

(e) The equilibrium state of an isolated system corresponds to a maximum of


the entropy. Discuss, in terms of the second derivative of S, how the
existence of an entropy maximum has a consequence for the heat capacity
of the system. [4]

TURN OVER
− page 3 − PH4211A
PART
MARKS

3. (a) A binary alloy contains two atomic species A and B with relative
proportions x and 1 − x . The Gibbs free energy for this system has the
form shown in the figure:

lo w T

h ig h T

0 x 1

In the low temperature case explain how, for some values of x, the system
may lower its free energy by separating into two phases of different
concentrations. [3]

(b) If the fraction of the system at concentration x1 is denoted by α then the


fraction at the other concentration x2 is 1 − α . Show that the fractions
α and 1 − α are given by the lever rule:
x −x x −x
α= 2 0, 1−α = 0 1
x2 − x1 x2 − x1

where x0 is the initial concentration of A atoms. [4]

(c) Sketch, on a T – x graph, the phase separation curve and the spinodal
curve. How are these determined from the above figure? What is the
meaning of the spinodal curve? [4]

(d) Identify, on the phase separation curve, the critical point. [2]

(e) It may be said that the first order transition becomes second order at the
critical point. Explain this. [3]

(f) Why are fluctuations important in the vicinity of the critical point?
Describe the nature of the fluctuations in this system. [4]

TURN OVER
− page 4 − PH4211A
PART
MARKS

4. (a) Explain what is meant by the order parameter in the context of phase
transitions and describe the difference in the behaviour of the order
parameter for first order and second order transitions. [3]

(b) When the Landau theory of phase transitions is applied to the ferroelectric
transition the free energy is expressed by a polynomial of the form

F = F0 + F2ϕ 2 + F4ϕ 4 + F6ϕ 6

What is the order parameter ϕ for this system? Give arguments for the
absence of odd-order terms in the expression. [3]

(c) By varying an external parameter, such as the strain, the ferroelectric


transition can be either first order or second order. Sketch the possible
variations of the free energy as a function of the order parameter that can
account for the first order and the second order transitions respectively. [4]

(d) Explain qualitatively how the order of the transition depends on the sign
of the F4 coefficient and why the F6 term may be neglected when the
transition is second order. [3]

(e) When the transition is first order show that the discontinuity in the order
parameter at the transition is given by

− F4
∆ϕ =
2 F6

and discuss the behaviour of the discontinuity as the transition becomes


second order. [3]

(f) Within the spirit of the Landau theory it is conventional to approximate


the temperature dependence of the F2 coefficient by

F2 = α (T − Tc ) .

Explain this by reference to the second order case. Using this temperature
dependence show that the latent heat at the first order transition is given
by

F4 [4]
L = αTtr
2 F6

and discuss the behaviour of L as the transition becomes second order.

TURN OVER
− page 5 − PH4211A
PART
MARKS

5. (a) The velocity v (t ) of a Brownian particle is a randomly-varying function


of time. Define Gv (τ ) , the autocorrelation function for the velocity and
explain its physical significance. [4]

(b) Show that the mean square displacement of the particle is given by
t
[3]
x 2 (t ) = 2 ∫ (t − τ ) Gv (τ ) dτ .
0

(c) What is meant by the correlation time of Gv (τ ) ? Discuss the limiting


behaviour of the mean square displacement for times shorter than, and
longer than the correlation time. In particular, show that in the long time
limit the mean square displacement is proportional to time whereas in the
short time limit the mean square displacement is proportional to time
[4]
squared.

(d) Give a physical explanation of the behaviour in the short and long time
limits. [4]

(e) Using the above results, show that the diffusion coefficient of the
Brownian particle may be expressed in terms of the area under Gv (τ ) . [3]

(f) Why is this called a fluctuation-dissipation theorem? [2]

END BPC/JS
UNIVERSITY OF LONDON

MSci EXAMINATION 2001

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Only CASIO fx85WA Calculators ARE permitted

PH4211A
© Royal Holloway and Bedford New College 2001
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js

= = h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1


Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. (a) By considering an isolated system containing a constraint, such as a [6]


dividing partition, explain clearly why the equilibrium state, upon
removal of the constraint, corresponds to that of maximum entropy.

(b) Two systems are brought into contact so that they may exchange thermal [8]
energy, mechanical energy and particles. By using the appropriate
definitions, show that the equilibrium state corresponds to that in which
the temperatures, pressures and chemical potentials of the two systems are
equalised.

(c) Since the equilibrium state has maximum entropy, discuss the implications [6]
of the behaviour of the second derivative of the entropy with respect to
energy of the composite system in terms of the heat capacity.

2. When the mean field theory of phase transitions is applied to the


ferromagnetic transition in the absence of a magnetic field, the free
energy in the vicinity of the transition is given approximately by

Nk Nk
F= (T − Tc )m 2 + Tc m 4 .
2 12

(a) Explain the structure of this expression and indicate what the various [4]
quantities are.

(b) By sketching the form of this expression, explain how the transition [4]
occurs as the temperature is lowered through Tc and discuss the
magnitude of the fluctuations in the vicinity of Tc.

(c) Using the above expression show that in the vicinity of the transition [5]

3(Tc − T )
m=± for T < Tc
Tc
=0 for T > Tc .

Sketch this behaviour.

(d) What is the order of this transition? Explain your reasoning. [3]

(e) Show that below the transition there is a contribution to the entropy given [4]
by
3 Nk T − Tc
2 Tc
and that this leads to a jump in the thermal capacity on cooling through
the transition of
3
∆C = Nk .
2

TURN OVER
− page 2 − PH4211A
PART
MARKS

3. (a) Outline the arguments by which the Ising hamiltonian

 = −J ∑
neighbours
S zi S zj
i, j

is approximated, in mean field theory, by a local magnetic field

b = λ M z zˆ

Where ẑ is the unit vector in the z direction, λ is a constant and Mz is the [4]
component of magnetisation in the z direction.

(b) The magnetisation of a non-interacting assembly of N spin ½ magnetic


moments, µ, is given by
M B 
M = M 0 tanh  0 
 N kT 

where the saturation magnetisation is M 0 = N µ and the directions of M


and the applied magnetic field B are parallel.

Show that when the Ising interaction mean field is incorporated, there can
be a spontaneous magnetisation given by
Mz M T 
= tanh  z c 
M0  M0 T 

where Tc = λ M 02 Nk . What is the interpretation of Tc? Sketch the [4]


behaviour of the Mz as a function of temperature and discuss the order of
the transition.

(c) Now consider the transition in the presence of a transverse magnetic field [2]
B = Bx xˆ . Write down the magnitude and the direction of the total
magnetic field.

(d) Hence show that the magnetisation in the z direction is given, within this
model, by
−1 2
M T [4]
Mz  Bx2  B2 
=  1 + 2 2  tanh  z c 1 + 2 x 2  .
M0  λ M z   M0 T λ M z 

Hint: consider the magnetisation in the direction of the total field.

(e) By rearranging this expression and considering the case where M z → 0 , [3]
obtain an expression for the locus of spontaneous Mz in the T–Bx plane
and sketch this.

(f) Discuss the behaviour of Mz at T = 0 as Bx is varied. Why is this referred [3]


to as a quantum phase transition?

TURN OVER
− page 3 − PH4211A
PART
MARKS

4. Write short notes on three of the following:

(a) conserved and non-conserved order parameters; [62/3]

(b) the cluster expansion treatment of non-ideal gases; [62/3]

(c) scaling theory and critical exponents; [62/3]

(d) the absence of long-range order in one dimensional systems; [62/3]

(e) The Ising model. [62/3]

5. (a) The velocity v (t ) of a Brownian particle is a randomly varying function [4]


of time. Define Gv (τ ) , the auto-correlation function for the velocity and
explain its physical significance.

(b) Show that the mean square displacement of the particle is given by
t
[4]
x (t ) = 2 ∫ (t − τ ) Gv (τ ) dτ .
2

(c) What is meant by the correlation time of Gv (τ ) ? Discuss the limiting [4]
behaviour of the mean square displacement for times shorter than, and
longer than the correlation time. In particular, show that in the long time
limit the mean square displacement is proportional to time whereas in the
short time limit the mean square displacement is proportional to time
squared.

(d) Give a physical explanation of the behaviour of the particle in the short [3]
and long time limits.

(e) Using the above results, show that the diffusion coefficient of the [3]
Brownian particle may be expressed in terms of the area under Gv (τ ) .

(f) Why is this result called a fluctuation-dissipation theorem? [2]

TURN OVER
UNIVERSITY OF LONDON

MSci EXAMINATION 2002

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Only CASIO fx85WA Calculators are permitted

PH4211A/31
© Royal Holloway and Bedford New College 2002
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js

= = h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1


Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. (a) Explain why a system in thermal equilibrium with a reservoir at a [3]


temperature T has fluctuations in its energy E.

(b) The quantity σE is given by

12
2
 E =  E − E  .
 
[3]
Why is σE a measure of the energy fluctuations?

(c) Show that


[2]
σ 2E = E 2 − E .
2

(d) The mean energy of a system in thermal equilibrium at a temperature T


may be written as

1
∑ − E / kT
E = E je j .
Z j
[3]
Explain the meaning of this expression, defining the quantity Z.

(e) By considering the expression for the mean square energy E 2 show
that the size of the energy fluctuations may be written as

σ E = kT 2 CV
[6]
where CV is the thermal capacity of the system.

(f) Discuss how the energy fluctuations depend on the size (number of
particles N ) of the system and show that the fractional energy fluctuations
tend to zero as N −1/2. [3]

TURN OVER
− page 2 − PH4211A
PART
MARKS

2. (a) What is Brownian motion? [2]

(b) The force on a Brownian particle may be written as [4]

1
F (t ) = f (t ) − v
µ

where f (t ) is a randomly fluctuating force, v is the velocity and µ the


mobility of the particle. Discuss the separation of the force into these two
parts.

(c) Show that the equation of motion for the Brownian particle may be [3]
written as

dv (t )
+ γv(t ) = A(t )
dt

and identify the terms.

(d) The solution to the equation of motion may be written [3]

t
v(t ) = v(0)e −γ t
+ ∫ eγ (u −t ) A(u )du .
0

Describe how this solution arises and explain its implications.

(e) The autocorrelation function for the random force is defined by the [3]
average

A(t ) A(t + τ) .

Discuss the physical meaning of this expression and explain why it is


independent of the time t.

(f) Show how the motion of the Brownian particle depends on the [5]
autocorrelation function of the velocity, and show how this leads to
diffusive behaviour. Give an expression for the diffusion coefficient in
terms of the velocity autocorrelation function.

TURN OVER
− page 3 − PH4211A
PART
MARKS

3. (a) A binary alloy contains two atomic species A and B with relative
proportions x and 1 − x of different concentrations x1 and x2. The Gibbs
free energy for this system has the form shown in the figure:

lo w T

h ig h T

0 x 1

In the low temperature case explain how, for some values of x, the system [3]
may lower its free energy by separating into two phases of different
concentrations.

(b) If the fraction of the system at concentration x1 is denoted by α then the


fraction at the other concentration x2 is 1 − α . Show that the fractions
α and 1 − α are given by the lever rule:

x2 − x0 x0 − x1
α= , 1−α =
x2 − x1 x2 − x1

where x0 is the initial concentration of A atoms. [3]

(c) Sketch, on a T – x graph, the phase separation curve and the spinodal [5]
curve. How are these determined from the above figure? What is the
meaning of the spinodal curve?

(d) Identify, on the phase separation curve, the critical point. [2]

(e) It may be said that the first order transition becomes second order at the [3]
critical point. Explain this.

(f) Why are fluctuations important in the vicinity of the critical point? [4]
Describe the nature of the fluctuations in this system.

TURN OVER
− page 4 − PH4211A
PART
MARKS

4. (a) Explain what is meant by the order parameter in the context of phase
transitions and describe the difference in the behaviour of the order
parameter for first order and second order transitions. [3]

(b) When the Landau theory of phase transitions is applied to the ferroelectric
transition the free energy is expressed by a polynomial of the form

F = F0 + F2ϕ 2 + F4ϕ 4 + F6ϕ 6

What is the order parameter ϕ for this system? Give arguments for the [3]
structure of this free energy expression.

(c) By varying an external parameter, such as the strain, the ferroelectric


transition can be either first order or second order. Sketch the possible
variations of the free energy as a function of the order parameter that can [4]
account for the first order and the second order transitions respectively.

(d) Explain qualitatively how the order of the transition depends on the sign
of the F4 coefficient and why the F6 term may be neglected when the
transition is second order. [3]

(e) When the transition is first order show that the discontinuity in the order
parameter at the transition is given by

− F4
∆ϕ =
2 F6

and discuss the behaviour of the discontinuity as the transition becomes [3]
second order.

(f) Within the spirit of the Landau theory it is conventional to approximate


the temperature dependence of the F2 coefficient by

F2 = α (T − Tc ) .

Explain this by reference to the second order case. Using this temperature
dependence show that the latent heat at the first order transition is given
by

F4
L = αTtr
2 F6

and discuss the behaviour of L as the transition becomes second order. [4]

TURN OVER
− page 5 − PH4211A
PART
MARKS

5. (a) Outline the sequence of arguments by which one shows that two isolated
systems, when brought into thermal contact, end up in the thermodynamic
state for which

∂ ln Ω1 ∂ ln Ω 2
=
∂E ∂E

and define the terms Ω1 , Ω 2 and E in this expression. [4]

(b) Write down an expression for entropy in terms of Ω. How does the above [4]
equation imply the equalisation of the temperatures of the two systems?

(c) Now consider a small sub-system of a large isolated system. The total
energy of the isolated system is Et . The sub-system can exchange
thermal energy with the large system. When the sub-system is in a
microstate of energy E the entropy of the combined system may be
expressed as

∂S E 2 ∂ 2 S
S = S ( Et ) − E + − ....
∂E 2 ∂E 2

Justify the structure of this expression. [4]

(d) Show how the above result leads to the Boltzmann distribution function [4]
(otherwise known as the Boltzmann factor).

(e) The equilibrium state of an isolated system corresponds to a maximum of


the entropy. Discuss, in terms of the second derivative of S, how the
existence of an entropy maximum has a consequence for the heat capacity [4]
of the system.

TURN OVER
UNIVERSITY OF LONDON

MSci EXAMINATION 2003

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Only CASIO fx85WA Calculators are permitted

PH4211A/33
© Royal Holloway and Bedford New College 2003
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js
η = h/2π = 1.05 × 10-34 Js
Boltzmann constant k = 1.38 × 10-23 J K-1
Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. (a) When the liquid and the gas phase of a fluid coexist in equilibrium the
temperature, pressure and chemical potential are the same in both phases.
Explain why this is the case. [3]

(b) Under what constraints is the equilibrium state of such a system


determined by minimising the Helmholtz free energy F = E − TS ? [3]

(c) Justify the double tangent construction on the F (V ) curve to determine


the equilibrium state and show that the volume fractions α1 and α2 of the
two coexisting phases may be written

V2 − V0 V0 − V1
α1 = , α2 =
V2 − V1 V2 − V1
[6]
where the symbols have their usual meaning.

(d) A van der Waals p-V isotherm, for a temperature less than the critical
temperature, is shown in the figure.
p

The true coexistence behaviour should be a horizontal straight line. Using


the result

∂F
p=−
∂V T

show how the coexistence pressure may be determined from the double
tangent construction. [5]

(e) Explain the connection with Maxwell’s ‘equal area’ construction. [3]

TURN OVER
− page 2 − PH4211A
PART
MARKS

2. (a) The ferromagnetic transition and the ferroelectric transition have


similarities and differences. For each case:

i) What is the order parameter?

ii) Is the order parameter conserved or non-conserved?

iii) What symmetry is broken at the transition?

iv) Is the broken symmetry continuous or discrete?

v) What is the order of the transition? [10]

(b) Outline the arguments by which the Heisenberg Hamiltonian


H = − J ∑ Si .S j
i, j

is approximated, in mean field theory, by a local magnetic field


b = λM
where M is the magnetisation and λ is a constant. [4]

(c) The magnetisation of a non-interacting assembly of N spin ½ magnetic


moments µ is given by

M ⎛M B ⎞
= tanh ⎜ 0 ⎟
M0 ⎝ N kT ⎠
where the saturation magnetisation is M 0 = N µ and the directions of M
and the applied magnetic field B are parallel.

Show that in the presence of a Heisenberg interaction between the spins,


in the mean field approximation, the spontaneous magnetisation is given
by

M ⎛ M Tc ⎞
= tanh ⎜ ⎟
M0 ⎝ M0 T ⎠
where Tc = λ M 02 Nk . What is the interpretation of Tc? [4]

(d) Sketch the behaviour of the spontaneous magnetisation as a function of


temperature and discuss the order of the transition. [2]

TURN OVER
− page 3 − PH4211A
PART
MARKS

3. (a) Show that when two isolated systems are brought into thermal contact,
they end up in the thermodynamic state for which

∂ ln Ω1 ∂ ln Ω 2
=
∂E ∂E

defining the terms Ω1 , Ω 2 and E in this expression. [5]

(b) Write down the Boltzmann expression for entropy in terms of Ω. How
does the above equation imply the equalisation of the temperatures of the
two systems? [4]

(c) Now consider a small sub-system of a large isolated system. The total
energy of the isolated system is Et . The sub-system can exchange
thermal energy with the large system. When the sub-system is in a
microstate of energy E the entropy of the combined system may be
expressed as

∂S E 2 ∂ 2 S
S = S ( Et ) − E + − ....
∂E 2 ∂E 2

Justify the structure of this expression. [3]

(d) Show how the above result leads to the Boltzmann distribution function
(otherwise known as the Boltzmann factor). [4]

(e) The equilibrium state of an isolated system corresponds to a maximum of


the entropy. Discuss, in terms of the second derivative of S, how the
existence of an entropy maximum implies that the heat capacity of the
system is positive. [4]

TURN OVER
− page 4 − PH4211A
PART
MARKS

4. (a) The partition function for a single particle moving freely in a box of
volume V may be written as

⎛ 2π mkT ⎞
32
V
z =V ⎜ ⎟ =
⎝ h
2
⎠ Λ3

where the symbols have their usual meaning.

The quantity Λ is known as the thermal de Broglie wavelength. Explain


the meaning of this quantity. [3]

(b) Show that the pressure of a gas of N indistinguishable such particles is


related to z by

∂ ln z
p = NkT . [4]
∂V T

(c) Evaluate the pressure for this system, hence deriving the equation of state
of an ideal gas. [3]

(d) In the van der Waals approach to an interacting gas the single particle
partition function may be approximated by

V − Vex − E
z=
kT
e
Λ3

Sketch the expected variation of the inter-particle interaction potential


with separation. Discuss how the various features of the interaction are
incorporated into the above expression for z through the quantities Vex and
E . [5]

(e) Why is this approach referred to as a mean field approximation? [2]

(f) Discuss qualitatively how this approach to the interacting gas is connected
with the law of corresponding states. [3]

TURN OVER
− page 5 − PH4211A
PART
MARKS

5. (a) The autocorrelation function of a random function of time x ( t ) is defined


by

G (t ) = x (0) x (t )

where the angled brackets indicate an average. Explain how


G ( t ) describes the mean decay of the fluctuations of x ( t ) . [4]

(b) The correlation time τc of the random function is defined by



1
G ( t ) dt .
G ( 0 ) ∫0
τc =

What feature of the fluctuating quantity x ( t ) does the correlation time


describe? [3]

(c) A fluctuating quantity is found to have an exponential correlation function

G ( t ) = G ( 0 ) e−t τ .

What is the corresponding correlation time? [2]

(d) For an isolated system the probability P ( x ) that a fluctuating quantity has
the value x is given by the Einstein expression

P ( x ) ∝ eS( x) k

where S is the entropy. Justify this formula. [3]

(e) Sketch and explain the functional form of S ( x ) in the vicinity of the mean
value x . [3]

(f) Hence discuss qualitatively why the fluctuations in x may follow a normal
distribution. [3]

(g) Discuss the way the diffusion coefficient of a particle is related to its
velocity autocorrelation function. [2]

END BPC
UNIVERSITY OF LONDON

MSci EXAMINATION 2004

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Only CASIO fx85WA Calculators or CASIO fx85MS Calculators are permitted

PH4211A/34
© Royal Holloway and Bedford New College 2004

2003-04
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js

= h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1


Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. (a) The Boltzmann expression for entropy is written S = k ln Ω . Identify the


terms in the equation, taking care to explain the meaning of Ω. [3]

(b) By considering an isolated system containing a constraint, such as a


dividing partition, explain clearly why the equilibrium state, upon
removal of the constraint, corresponds to that of maximum entropy. [5]

(c) Two systems are brought into contact so that they may exchange thermal
energy, mechanical energy and particles. By using the appropriate
definitions, show that the equilibrium state corresponds to that in which
the temperatures, pressures and chemical potentials of the two systems are
equalised [7]

(d) Using the fact that the equilibrium state has maximum entropy, state what
you can about the second derivative of the entropy with respect to energy
of the composite system. Derive the implications for the heat capacity. [5]

2. (a) What is meant by the term order parameter in a phase transition ? [2]

(b) Explain the difference between a conserved and a non-conserved order


parameter. Give an example of each. [4]

(c) In the Landau theory of phase transitions, the free energy is expanded in
powers of the order parameter. A key feature of this approach is the
truncation of the power series. Discuss what determines the power at
which the series is terminated. [2]

(d) Expansions (in even powers) up to fourth order can describe second order
transitions while expansions up to sixth order can describe first order
transitions. Explain qualitatively why this is. [4]

(e) In the vicinity of the critical point of a ferromagnetic transition the free
energy is expressed as a function of the order parameter ϕ as

F = F0 + F2ϕ 2 + F4ϕ 4 .

Sketch the form of this function for temperatures above, equal to, and
[4]
below the critical temperature.

(f) Show that the critical point corresponds to F2 = 0. [2]

(g) Show that below the critical temperature the order parameter has the form

ϕ ~ Tc − T . [2]

TURN OVER
− page 2 − PH4211A
PART
MARKS

3. The partition function Z for a gas of N interacting atoms is given by


⎛ p2 ⎞
−⎜ ∑ i + ∑U ( qi ,q j ) ⎟ kT
1 ⎜ ⎟
3N ∫
2 m i< j
Z= e⎝i ⎠
d3 N p d3 N q
N !h

where the symbols have their usual meaning.

(a) Show that Z may be expressed as

− ∑U ( qi ,q j ) kT
1
Z = Z id N ∫ e i< j d3N q Eq. 3.1
V

where Zid is the partition function for an ideal (non-interacting) gas.


[4]
Be sure to explain the appearance of the VN factor.

(b) The interaction potential for a pair of hard spheres with centres a distance
r apart is given by

U (r ) = ∞ r <σ
=0 r >σ

where σ is the hard core dimension.

Sketch and label this interaction potential and relate σ to the radius of an [2]
atom in the gas.

(c) The partition function for a gas of hard spheres might be approximated by

⎛ V − Nb ⎞
N

Z = Z id ⎜ ⎟ .
⎝ V ⎠

Give a justification for this from the structure of Eq. 3.1. [3]

Question 3 continued overleaf

TURN OVER
− page 3 − PH4211A
PART
MARKS

(d) The above approximation for the partition function leads to the equation
of state

p (V − Nb ) = NkT .

Express this equation in terms of the virial expansion


2 3
p N ⎛N⎞ ⎛N⎞
= + B2 (T ) ⎜ ⎟ + B3 (T ) ⎜ ⎟ + .
kT V ⎝V ⎠ ⎝V ⎠

Hence find expressions for B2 and B3 and write down the general Bn.
[4]

(e) The virial coefficients for the hard sphere gas are independent of
temperature. Why is this? [3]

(f) Exact calculations of the first few virial coefficients for the hard sphere
gas give:

B2 (T ) = b, B3 (T ) = 85 b 2 , B4 (T ) = 0.29b3

where b = 2πσ 3 3 .

What do you conclude, by comparing these values with those you


obtained in part (d) above? [4]

You may find the expansion below useful in part (d).

1
= 1 + x + x 2 + x3 + …
1− x

TURN OVER
− page 4 − PH4211A
PART
MARKS

4. (a) Describe the essential features of the Ising Model for a ferromagnet. What [4]
is the order parameter for the Ising transition? To what extent is it a
suitable model to describe the behaviour of real materials?

(b) In one dimension there is no transition at finite temperatures. By


considering the free energy for this system in the thermodynamic limit,
explain why this is so. [4]

(c) In two dimensions the Ising model has a phase transition at finite
temperatures. Sketch the temperature dependence of the order parameter
and the heat capacity. Describe and explain how these differ from the [5]
results of mean field / Landau theory.

(d) When a transverse high magnetic field is applied to the system it exhibits
a quantum phase transition. What is a quantum phase transition? Describe
the nature of the ordered phase a) in the case of low magnetic field and b)
high magnetic field. [3]

(e) Sketch the temperature – transverse magnetic field phase diagram and
label the phases. What would be the effect of a parallel field on the [4]
transition?

TURN OVER
− page 5 − PH4211A
PART
MARKS

5. The force on a Brownian particle may be written as

bg bg
F t = f t −
1
µ
v

bg
where f t is a randomly fluctuating force, v is the velocity and µ the
mobility of the particle.

(a) What is Brownian motion? [2]

(b) Discuss the separation of the force into the two parts. In particular,
explain qualitatively how the friction force, proportional to the velocity,
arises as a consequence of the random motion of the background fluid
atoms. [5]

(c) Show that the equation of motion for the Brownian particle may be
written as

dv ( t )
+ γ v (t ) = A(t )
dt
[4]
and identify the terms.

(d) The autocorrelation function for the velocity is defined by the average

G (τ ) = v ( t ) v ( t + τ )

Discuss the physical meaning of this expression and explain why it is


independent of the time t. Comment on the relative time scales of G (τ ) [4]
and v ( t ) .

(e) Show how the mean square displacement of the Brownian particle
depends on the velocity autocorrelation function. How does this lead to
the description of the motion as diffusive? [5]

END
UNIVERSITY OF LONDON

MSci EXAMINATION 2005

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Only CASIO fx85WA Calculators or CASIO fx85MS Calculators are permitted

PH4211A/29
© Royal Holloway and Bedford New College 2005

2004-05
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js

= = h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1


Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. The Helmholtz free energy of mixing of a binary alloy, a regular mixture


of two different atomic species, may be written as

Fm = Nk ⎡⎣ 2Tc x (1 − x ) + T { x ln x + (1 − x ) ln (1 − x )}⎤⎦

where x is the concentration of one of the species and

s ⎧ 1 ⎫
Tc = ⎨ε ab − ( ε aa + ε bb ) ⎬ .
2k ⎩ 2 ⎭

(a) Define the terms and explain the structure of these equations. Under what
constraints is the equilibrium state of such a system determined by
minimising the Helmholtz free energy F = E − TS ? [4]

(b) Discuss how the zero-temperature state of this system depends upon the
sign of the energy parameter ε ab − ( ε aa + ε bb ) 2 . [2]

(c) When this energy parameter is positive, the form of Fm for temperatures
below and above Tc is shown in the figure below.

Fm

0 x 1

Which curve corresponds to which case? [2]

(d) Explain why, in the low-temperature case, the system may lower its free
energy by assuming an inhomogeneous state and discuss the nature of this
state. [4]

(e) Show that the phase separation temperature as a function of x is given by


Tps Tc = 2 (1 − 2 x ) ln ( (1 − x ) x ) . Sketch and label this curve. [4]

(f) In a solid mixture of the isotopes 3He and 4He, Tc is 300mK. At what
temperature will a homogeneous mixture of concentration x0 = 0.01 phase
separate? There is another initial concentration of mixture that is
predicted to phase separate at the same temperature. What is this
concentration? In reality, for these two initial concentrations the phase
separation temperatures are observed to be somewhat different. What
could be the explanation of this? [4]

TURN OVER
− page 2 − PH4211A
PART
MARKS

2. (a) The ferromagnetic transition and the ferroelectric transition have


similarities and differences. For each case:

i) What is the order parameter?

ii) Is the order parameter conserved or non-conserved?

iii) What symmetry is broken at the transition?

iv) Is the broken symmetry continuous or discrete?

v) What can be the order of the transition? [10]

(b) Outline the arguments by which the Heisenberg Hamiltonian


H = − J ∑ Si .S j
i, j
neighbours

is approximated, in mean field theory, by a local magnetic field


b = λM
where M is the magnetisation and λ is a constant. [4]

(c) The magnetisation of a non-interacting assembly of N spin ½ non-


interacting magnetic moments µ is given by

M ⎛M B ⎞
= tanh ⎜ 0 ⎟
M0 ⎝ N kT ⎠
where the saturation magnetisation is M 0 = N µ and the directions of M
and the applied magnetic field B are parallel.

Show that in the presence of a Heisenberg interaction between the spins,


but no applied field, the spontaneous magnetisation in the mean field
approximation, is given by

M ⎛ M Tc ⎞
= tanh ⎜ ⎟
M0 ⎝ M0 T ⎠
where Tc = λ M 02 Nk . What is the interpretation of Tc? [4]

(d) Sketch the behaviour of the spontaneous magnetisation as a function of


temperature and discuss the order of the transition. [2]

TURN OVER
− page 3 − PH4211A
PART
MARKS

3. (a) State the ‘law’ of corresponding states for a liquid-gas system. Discuss
how this law arises from a consideration of the van der Waals equation of
state and explain the microscopic basis for the law. [5]

(b) Liquid-gas coexistence data for a number of substances are plotted in the
figure below, showing reduced temperature against reduced density. The
dashed line is that calculated from the van der Waals equation of state. Tc
and ρc are the critical temperature and critical density.

1.00
gas liquid
T/Tc

0.90

0.80 3
He
4
He
Ne
A
Kr
0.70 Xe
N2
O2
CO
CH4
0.60

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6
ρ/ρc

For the moment, ignore the helium data. Discuss the extent to which the
other data support the law of corresponding states. Discuss why the van
der Waals curve is not in agreement with the data, and why the critical
exponent β is closer to 1/3 than the mean field value of 1/2. [5]

(c) There are no data very close to the critical point, because of experimental
difficulties. However, the general behaviour of most of the materials is
well-described by the continuous ‘universal’ curve. Discuss the behaviour
of the helium isotopes very close to the critical point. In particular, to
what extent do you expect them to give data falling on the universal curve
and what do you expect the critical exponent β to be? Justify your [5]
speculations.

(d) The thermal de Broglie wavelength is given by


Λ== .
mkT

Explain the significance of this quantity. Discuss the importance of Λ in


relation to the helium data in the figure that deviate from the universal
curve. Why are the helium data on the right hand side of the curve falling
away from the universal curve, but not on the left hand side? [5]

TURN OVER
− page 4 − PH4211A
PART
MARKS

4. The force on a Brownian particle in one dimension may be written as


F ( t ) = f ( t ) − v µ where f ( t ) is a randomly fluctuating force, v is the
velocity and µ the mobility of the particle.

(a) Discuss the separation of the force into the two parts. In particular,
explain qualitatively how the damping force, proportional to the velocity,
arises as a consequence of the random motion of the background fluid
atoms. [5]

(b) Show that the equation of motion for the Brownian particle of mass M
(the Langevin equation) may be written as

dv ( t ) 1
M + v (t ) = f (t ) . [2]
dt µ

(c) The solution to the Langevin equation is given by


t
1
v ( t ) = v ( 0 ) e −t M µ + ∫ e ( u − t ) M µ f ( u ) du .
M 0

Show, using appropriate approximations, that the equilibrium mean



square velocity may be expressed as v 2 = ( µ 2 M ) ∫ f ( 0 ) f ( t ) dt and
−∞
invoking the equipartition theorem, it follows that

1 1
=
µ 2kT ∫ f (0) f (t )
−∞
dt . [5]

(d) Discuss how this may be regarded as an example of the fluctuation-


dissipation theorem. [2]

(e) The voltage across an L-R circuit is given in terms of the current I(t) by

dI ( t )
L + RI ( t ) = V ( t )
dt

where V(t) may be regarded as a fluctuating voltage. By making the


appropriate identifications, show that the resistance may be related to the
voltage fluctuations through

1
∫ V ( 0 )V ( t )
R= dt . [4]
2kT −∞

(f) How may such voltage fluctuations be observed? [2]

TURN OVER
− page 5 − PH4211A
PART
MARKS

5. (a) In the context of statistical mechanics explain the meaning of the term
ensemble. [4]

(b) Discuss and compare the Boltzmann and the Gibbs view of ensembles. [4]

(c) Explain the meaning of the term phase space, including a discussion of
the difference between Boltzmann’s view and that of Gibbs. [4]

(d) Liouville’s theorem implies that the flow of points in phase space is
incompressible. What assumptions go into the derivation of this theorem? [2]

(e) Paradoxically, Boltzmann’s H theorem seems to contradict Liouville’s


theorem. Give a statement of Boltzmann’s H theorem and explain the
paradox. Show how the paradox may be resolved. [6]

END
UNIVERSITY OF LONDON

MSci EXAMINATION 2006

For Internal Students of

Royal Holloway

DO NOT TURN OVER UNTIL TOLD TO BEGIN

PH4211A: STATISTICAL MECHANICS

Time Allowed: TWO AND A HALF hours

Answer THREE QUESTIONS only

No credit will be given for attempting any further questions

Approximate part-marks for questions are given in the right-hand margin

Only CASIO fx85WA Calculators or CASIO fx85MS Calculators are permitted

PH4211A/25

© Royal Holloway and Bedford New College 2006

2005-06
GENERAL PHYSICAL CONSTANTS

Permeability of vacuum µ0 = 4π × 10-7 H m-1


Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1
1/4πε0 = 9.0 × 109 m F-1
Speed of light in vacuum c = 3.00 × 108 m s-1
Elementary charge e = 1.60 × 10-19 C
Electron (rest) mass me = 9.11 × 10-31 kg
Unified atomic mass constant mu = 1.66 × 10-27 kg
Proton rest mass mp = 1.67 × 10-27 kg
Neutron rest mass mn = 1.67 × 10-27 kg
Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1
Planck constant h = 6.63 × 10-34 Js

= h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1


Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4
Gas constant R = 8.31 J mol-1 K-1
Avogadro constant NA = 6.02 × 1023 mol-1
Gravitational constant G = 6.67 × 10-11 N m2 kg-2
Acceleration due to gravity g = 9.81 m s-2
Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3
One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e ≅ 2.718 π ≅ 3.142 loge10 ≅ 2.303


− page 1 − PH4211A
PART
MARKS

1. The partition function for a single particle of mass m at temperature T, in


a box of volume V is given by

V
z=
Λ3

where the thermal de Broglie wavelength Λ is

2π 2
Λ= .
mkT

(a) Why is Λ known as the thermal de Broglie wavelength? [2]

(b) Explain why the partition function Z of a gas of N identical particles may
be written in terms of the partition function z of a single particle as

1 N
Z= z .
N!

In particular, discuss the origin of the N! term and the validity of this
expression. [5]

(c) Write down the Helmholtz free energy in terms of the partition function
and hence show that the free energy of the gas is given by

F = − NkT ln (Ve N Λ 3 ) . [5]

(d) What would the expression for the free energy be if the N! term in the
partition function were ignored? Why would such an expression be
physically unacceptable? (Hint: Think how F would change if both N and
V were doubled.) [4]

(e) Evaluate Λ for helium (4He) gas at a pressure of 1 bar for temperatures of
3 K and 300 K and compare it with the mean distance between atoms.
What physical meaning do you attach to these results? [4]

The mass of a 4He atom is 6.64×10–27 kg.


Stirling’s approximation: lnN! NlnN – N.

2. Write an essay on “The Arrow of Time”. You should include a


description of the problem in terms of phase space and you should
include a discussion of Liouville’s theorem and Boltzmann’s H theorem. [20]

TURN OVER
− page 2 − PH4211A
PART
MARKS

3. (a) The van der Waals equation of state for a fluid system is given by

⎛ aN 2 ⎞
⎜ p + ⎟ (V − Nb ) = NkT .
⎝ V2 ⎠

Explain the form of this equation; in particular discuss the significance of


the constants a and b, and how they account for the interactions between
the particles. [3]

(b) Sketch a van der Waals p - V isotherm corresponding to liquid-gas


coexistence. Indicate the stable, metastable and unstable regions of the
curve. Sketch, also, an isotherm that indicates the critical point. [3]

(c) Show that the temperature and volume at the critical point are given by

8a
kTc = , Vc = 3 Nb .
27b [4]

(d) Corrections to the ideal gas behaviour at low densities are conventionally
written in terms of the virial expansion
2 3
p N ⎛N⎞ ⎛N⎞
= + B2 (T ) ⎜ ⎟ + B3 (T ) ⎜ ⎟ + … .
kT V ⎝V ⎠ ⎝V ⎠

Show that, for the van der Waals gas the second virial coefficient is given
by B2 (T ) = b − a kT . [3]

(e) Experimental measurements of the second virial coefficient of Nitrogen


yield

B2 (T ) = −2.83 × 10−29 m3 at T = 300K


B2 (T ) = −5.03 × 10−30 m3 at T = 400K .

Calculate the critical temperature that is predicted from these values. [3]

(f) The experimental value for the critical temperature of Nitrogen is found to
be 126 K. Compare this with the value inferred from the second virial
coefficient measurements and account for any discrepancy. [2]

(g) At 77K liquid Nitrogen normally boils at a pressure of one atmosphere.


However it is possible, with care, to decrease the pressure below one
atmosphere without boiling the liquid. With reference to the p – V
diagram describe this process and explain, in outline, how one may
determine the minimum pressure at which this superheated liquid can
exist (you are not required to calculate this). [2]

TURN OVER
− page 3 − PH4211A
PART
MARKS

4. Write short notes on two of the following:

(a) Brownian motion; [10]

(b) universality and the classification of phase transitions; [10]

(c) the Ising model and its ubiquity. [10]

5. (a) What is meant by the term order parameter in the context of phase
transitions? [3]

(b) Describe the behaviour of the order parameter in a first order transition
and in a second order transition. [3]

(c) In the Landau treatment of phase transitions the appropriate free energy is
expanded in powers of the order parameter. Discuss what considerations
lead to the absence of terms from the expansion and what determines the
order at which the expansion will be terminated. [3]

(d) Sketch the form of the free energy as a function of order parameter for a
system exhibiting a second order transition, for temperatures above, equal
to and below the critical temperature. [3]

(e) Show that, within the framework of the Landau model, the critical
exponent, β, associated with the order parameter has the value 1/2. [6]

(f) How does this compare with real systems? [2]

END
Royal Holloway
UNIVERSITY OF LONDON

MSci EXAMINATION

STATISTICAL MECHANICS CP4211A

SUMMER 1999

Time Allowed: TWO AND A HALF HOURS

Answer THREE questions only. No credit will be given for attempting a further
question.

Each question carries 20 marks. The mark provisionally allocated to each section
is indicated in the margin.

TURN OVER WHEN INSTRUCTED

PH4211A/35
©Royal Holloway and Bedford New College 1999
GENERAL PHYSICAL CONSTANTS
Permeability of vacuum µ0 = 4π × 10-7 H m-1

Permittivity of vacuum ε0 = 8.85 × 10-12 F m-1

1/4πε0 = 9.0 × 109 m F-1

Speed of light in vacuum c = 3.00 × 108 m s-1

Elementary charge e = 1.60 × 10-19 C

Electron (rest) mass me = 9.11 × 10-31 kg

Unified atomic mass constant mu = 1.66 × 10-27 kg

Proton rest mass mp = 1.67 × 10-27 kg

Neutron rest mass mn = 1.67 × 10-27 kg

Ratio of electronic charge to mass e/me = 1.76 × 1011 C kg-1

Planck constant h = 6.63 × 10-34 Js

= = h/2π = 1.05 × 10-34 Js

Boltzmann constant k = 1.38 × 10-23 J K-1

Stefan-Boltzmann constant σ = 5.67 × 10-8 W m-2 K-4

Gas constant R = 8.31 J mol-1 K-1

Avogadro constant NA = 6.02 × 1023 mol-1

Gravitational constant G = 6.67 × 10-11 N m2 kg-2

Acceleration due to gravity g = 9.81 m s-2

Volume of one mole of an ideal gas at STP = 2.24 × 10-2 m3

One standard atmosphere P0 = 1.01 × 105 N m-2

MATHEMATICAL CONSTANTS

e = 2.718 π = 3.142 loge10 = 2.303


- page 1 - CP4211A
PART
MARKS

1. (a) By the use of probability arguments explain why the equilibrium state of an isolated
system corresponds to that of maximum entropy. [6]

(b) Two systems are brought into contact so that they may exchange thermal energy,
mechanical energy and particles. By using the appropriate definitions, show that the
equilibrium state corresponds to that in which the temperature, pressure and
chemical potential of the two systems are equalised. [8]

(c) Since the equilibrium state has maximum entropy, discuss the second derivative of [6]
the entropy of the composite system.

2. When the mean field theory of phase transitions is applied to the ferromagnetic
transition in the absence of a magnetic field, the free energy is given approximately,
in the vicinity of the transition, by

F=
Nk
2
 $
T − Tc m2 +
Nk
12
Tc m4 .

(a) Explain the structure of this expression and indicate what the various quantities are. [4]

(b) By sketching the form of this expression, explain how the transition occurs as the
temperature is lowered through Tc and discuss the magnitude of the fluctuations in
[4]
the vicinity of Tc.

(c) Using the above expression show that in the vicinity of the transition

m= ±

3 Tc − T $
Tc

and sketch this behaviour. [5]

(d) What is the order of this transition? Explain your reasoning. [3]

(e) Show that below the transition there is a contribution to the entropy given by
3 Nk
2

T − Tc $
and that this leads to a jump in the thermal capacity on cooling through the transition
[4]
of
3 NkTc
∆C = .
2

TURN OVER
- page 2 - PH4211A
PART
MARKS

3. (a) Explain why a system in thermal equilibrium with a reservoir at a temperature T has
fluctuations in its energy E. [3]

(b) A measure of the size of the energy fluctuations is given by

σE = E− E % 2 1/ 2
. [3]

What does this expression mean?

(c) Show that the mean square energy fluctuations are given by

σ 2E = E 2 − E .
2
[2]

(d) The mean energy of a system in thermal equilibrium at a temperature T may be


written as

1

− E / kT
E = E je j .
Z j [3]

In terms of probabilities, explain the meaning of this expression.

(e) By considering the expression for the mean square energy E 2 show that the size
of the energy fluctuations may be written as

σ E = kT 2 CV
[6]

where CV is the thermal capacity of the system.

(f) Discuss how the energy fluctuations depend on the size (number of particles) of the [3]
system.

TURN OVER
- page 3 - PH4211A
PART
MARKS

4. (a) What is ferroelectricity and what is the order parameter below the ferroelectric [2]
transition?

(b) By varying an external parameter, such as pressure, the ferroelectric transition can
be either first or second order.

Sketch the variation of the free energy as a function of the order parameter which
[3]
can account for the first order and the second order transitions.

(c) Explain how the features of the ferroelectric transition may be treated by an
expansion of the free energy in (even) powers of the order parameter up to the sixth
power.

F = F2ϕ 2 + F4ϕ 4 + F6ϕ 6

[3]
Give arguments for the absence of odd terms in the expansion.

(d) Show how the order of the transition depends on the sign of the F4 coefficient. [2]

(e) When the transition is second order explain why the F6 term may be neglected and [3]
show that at the transition the F2 term vanishes.

(f) When the transition is first order show that the discontinuity in the order parameter
at the transition is given by

− F4
∆ϕ = .
2 F6

[5]
Discuss the behaviour of the discontinuity as the transition becomes second order.

(g) There is no latent heat involved in a second order transition. By considering the
temperature variation of the F coefficients, discuss the relationship between the
[2]
latent heat and the discontinuity in the order parameter.

TURN OVER
- page 4 - CP4211A
PART
MARKS

5. (a) What is Brownian motion? [2]

(b) The force on a Brownian particle may be written as

$ $
F t = f t −
1
µ
v

$
[4]
where f t is a randomly fluctuating force, v is the velocity and µ the mobility of
the particle. Discuss the separation of the force into these two parts.

(c) Show that the equation of motion for the Brownian particle may be written as

$ $ $
dv t
+γv t = A t
dt
[3]
and identify the terms.

(d) The solution to the equation of motion may be written

$  $
v t = v 0 e − γt + 1
0
t
$
e γ  u − t $ A u du .

[3]
Describe how this solution arises and explain its implications.

(e) The autocorrelation function for the random force is defined by the average

$  $
A t A t +τ .

Discuss the physical meaning of this expression and explain why it is independent of [3]
the time t.

(f) Show how the motion of the Brownian particle depends on the autocorrelation
function of the velocity. Explain, in general terms, how the velocity autocorrelation
function follows from the solution to the Langevin equation and discuss how this
[5]
relates to the fluctuation- dissipation theorem.

END
PH4211 Outline Solutions 2005
Question 1

a) N: number of particles. k: Boltzmann’s constant. Tc: critical temperature. s: number


of nearest neighbours. εaa, εbb, εab: energy of aa, bb and ab bond. [1]

Structure of eqn for Fm: first term is energy of system, found by counting bonds and
their energies. Second term is entropy – has entropy of mixing form. Eqn for Tc – this
compares energy of an ‘unlike’ bond with the arithmetic mean for the two types of
‘like’ bonds. [2]

Helmholtz free energy is minimised at constant T and V. I.e. the system is being
considered at specified T and V. [1]

b) If the parameter is positive then aa and bb bonds are favoured over ab bonds at
T = 0. Then get phase separation at low temperatures. If the parameter is negative
then ab bonds are favoured over aa and bb bonds. In that case the low temperature
phase will be a ‘superlattice’ structure. [2]

c) Upper curve is high temperature case and lower curve is lower temperature case.
[2]

d) The curve corresponds to a homogeneous system. In regions where the curve is


concave it is possible to lower the free energy by dropping below the curve – when
the chord falls below the curve.

F0
F1

x1 x0 x2
By dropping below the (homogeneous) curve the system of initial concentration x0
can lower its free energy from F0 to F1. This happens when the system becomes
inhomogeneous, comprising regions of concentration x1 and regions of x2.

Lowest free energy would correspond to double tangent drawn on curve.


Conservation of numbers of particles gives mol-fractions of each phase in terms of the
lever rule. [4]

e) Phase separation transition found from minima in Fm curve, so differentiate Fm and


set equal to zero:
Fm = Nk ⎡⎣ 2Tc x (1 − x ) + T { x ln x + (1 − x ) ln (1 − x )}⎤⎦
so
dFm
= Nk ⎡⎣ 2Tc (1 − 2 x ) + T {ln x − ln (1 − x )}⎤⎦ .
dx
Set this equal to zero:
2Tc (1 − 2 x ) + T {ln x − ln (1 − x )} = 0
and this may be solved for T as a function of x, giving
T 2 (1 − 2 x )
=
Tc ln (1 − x ) x
as required. [2]
phase separation curve
1

critical point
T/Tc

0
0 0.5 x 1 [2]

f) Substitute in values. For x = 0.01, T = 0.427 Tc = 128 mK. [2]

Symmetry of phase curve about x = ½ implies that the same transition temperature
will apply for a concentration x = 1 – 0.01 = 0.99. [1]

Symmetry of the curve is a consequence of the regular solution model/assumption. So


if the concentrations x = 0.01 and x = 0.99 have different transition temperatures, this
means that the curve is not symmetric and so may be our specimen is not a regular
solution. (In reality there are crystallographic transformations to consider – but the
students are not expected to know that!) [1]

Question 2
a)
ferromagnet ferroelectric
Order parameter Magnetisation Polarisation [2]
Cons / non-cons Non-conserved Non-conserved [2]
symmetry rotational / time rev. inversion [2]
Cont / discrete continuous discrete [2]
order 2nd order 1st or 2nd order [2]

b) H = − J ∑ Si .S j looks like H = −∑ Si .Bi , in other words, where each spin


i, j i
neighbours

sees a magnetic field: Bi = J ∑


j:neighbours
S j . However each spin sees a different field. In
of i
the mean field approximation we take the field to be the same at each site, being the
mean value. Then b ∝ S and so b is proportional to the magnetisation so we can
write b = λ M . [4]

M ⎛M B ⎞
c) = tanh ⎜ 0 ⎟ and B = λ M as there is no external field. Thus
M0 ⎝ N kT ⎠
M ⎛ M λM ⎞
= tanh ⎜ 0 ⎟,
M0 ⎝ N kT ⎠
which we can write as
M ⎛ M Tc ⎞
= tanh ⎜ ⎟ [3]
M0 ⎝ M0 T ⎠
where Tc = λ M 02 Nk . Here Tc is the critical temperature. [1]

d)
1.0

M/M0

0.5

0.0
0.0 0.5 1.0 T /T c [1]
behaviour of magnetisation

The transition is second order – the magnetisation goes to zero continuously at the
critical point. [1]

Question 3

a) Law of corresponding states: equation of state takes similar ‘universal’ form if p,


V, T are scaled by their values at the critical point. [2]

If the variables of the van der Waals equation are scaled in this way then the equation
of state does indeed take on a universal form – the a and b parameters vanish (but
behaviour not in quantitative agreement with reality). [1]

On assumption that the law of interaction between molecules takes a universal form:
V ( r ) = ε f ( r σ ) – here f is a universal function and ε and σ scaling parameters,
different for different molecules, then in terms of reduced variables the partition
function, and thus all thermodynamic properties, would have universal form. This is
independent of the van der Waals equation of state. [2]
b) Apart from the heliums, the data do fall reasonably well on a single curve – this
supports the law of corresponding states.

The data do not fall on the van der Waals curve so this is not a good model for
describing the data. Thus corresponding states is not reliant on van der Waals.
[2]

Van der Waals is a mean field theory; thus the mean field β critical exponent of ½. If
the data exhibit a β critical exponent of ⅓, this is an indication of the inapplicability of
mean field theory. [3]

c) In the vicinity of the critical point the helium data should fall on the common
curve of the other data. This is because the critical behaviour is universal, not
depending on quantum effects. [3]

Thus the β critical exponent of for the helium data is expected to be ⅓. [2]

d) This is (essentially) the wavelength corresponding to the momentum


corresponding to the kinetic energy (of a free particle) kT. It represents the ‘quantum
size’, the extent of delocalisation of a particle due to quantum effects. [2]

The importance of this quantity is that when the mean inter-particle spacing is
comparable to or less than Λ then quantum effects are important. Then it is not
adequate to use classical mechanics in discussing such systems. This is true for the
helium liquids (and to a marginal extent to liquid neon – as can be seen in the data).
[2]

Data on right hand side are for liquids – there density is high, so if quantum effects
are important it will be there. Data on left hand side are for gas. There density is low
so quantum effects are not significant; data fall on universal curve. [1]

Question 4

a) A Brownian particle stationary in the centre of mass frame of the fluid experiences
random impacts from all directions; it experiences a random force of mean value zero.
A Brownian particle moving with respect to the centre of mass frame of the fluid
experiences impacts from the front which are more energetic than the ones from
behind. These have a greater momentum transfer. So in addition to the random force
there will be a mean force depending on the velocity of the Brownian particle (with
respect to the centre of mass frame of the fluid). Simple argument shows force
proportional to the velocity. [5]

b) Simple application of Newton’s law Mv = F , and using F = f − v µ . [2]


c)
2
⎡ 1 t ( u −t ) M µ ⎤
v ( t ) = ⎢ v ( 0 ) e −t M µ +
2
∫ e f ( u ) du ⎥
⎣ M 0

2 t u −t M µ
= v 2 ( 0 ) e − 2 t M µ + v ( 0 ) e − t M µ ∫ e ( ) f ( u ) du
M 0

1 t t
+ 2 ∫ du1 ∫ du2 e( 1 ) e( 2 ) f ( u1 ) f ( u2 ) .
u −t M µ u − t M µ

M 0 0

Now take average:


2 t
v 2 ( t ) = v 2 ( 0 ) e − 2 t M µ + e − t M µ ∫ e ( u − t ) M µ v ( 0 ) f ( u ) du
M 0
[2]
1 t t
+ 2 ∫ du1 ∫ du2 e ( u1 −t ) M µ ( u2 −t ) M µ
e f ( u1 ) f ( u2 ) .
M 0 0

First term is mean transient response; this dies to zero at long times. In second term
v(0) is uncorrelated with f(t) so this term vanishes. So we are left with
e − 2t M µ t t
v2 (t ) = 2 ∫ du1 ∫ du2 e( u1 +u2 ) M µ f ( u1 ) f ( u2 ) .
M 0 0

Assumption of short correlation time: f ( u1 ) f ( u2 ) = f 2δ ( u1 − u2 ) , where



f2 =∫ f ( 0 ) f ( t ) dt . This assumption forces u1 = u2 when the integral is done:
−∞

f 2 e − 2t M µ t
v2 (t ) = ∫ du e 1
2 u1 M µ

M2 0

µ
= f 2 e − 2t M µ ( e 2t M µ − 1)
2M
µ
= f 2 (1 − e − 2t M µ ) .
2M
So at long times the equilibrium value is
µ
v2 = f2
2M
or

µ
= ∫ f ( 0) f (t )
2
v dt . [2]
2M −∞

Now equipartition gives v 2 = kT M so that



kT µ
=
M 2M ∫ f ( 0) f (t )
−∞
dt

or

1 1
=
µ 2kT ∫ f (0) f (t )
−∞
dt . [1]

d) Left hand side is a dissipation coefficient. Right hand side involves fluctuations.
So this is an equation relating fluctuations and dissipation. Generalisation called the
fluctuation-dissipation theorem. [2]
e) By direct analogy M → L (not needed), v(t) → I(t), 1/µ → R, f(t) → V(t). [2]

Then

1
R=
2kT ∫ V ( 0 )V ( t )
−∞
dt . [2]

f) Can observe as the hiss (as opposed to the hum) from a loudspeaker when there is
no input connected to a hi-fi amplifier. [2]

Question 5

a) Ensemble: a collection of identically-prepared systems that can be used for


probabilistic calculations. The systems are envisaged as all manifesting the same
macrostate, while being in different microstates. [4]

b) Boltzmann regarded a gas as being an ensemble of single particles. Gibbs regarded


the gas as being one element of the ensemble. To Gibbs the ensemble was a collection
of boxes of the gas. Boltzmann view breaks down when the interactions between the
particles becomes significant. The Gibbs approach is the more general and it does not
suffer from this limitation. Thus for Boltzmann the ensemble is real; for Gibbs it is
imaginary. [4]

c) Phase space is a space of (generalised) position and momentum coordinates. A


microstate is specified as a point in phase space and its time evolution is specified as a
curve in phase space. According to the Boltzmann view a gas of N particles would be
specified by N points in a 6-dimensional phase space. According to the Gibbs view
the gas would be specified by a single point in 6N-dimensional phase space. An
ensemble would then be represented by a collection of points in the phase space. [4]

d) Assumption that goes into Liouville’s theorem is that the dynamics are governed
by Classical (Hamiltonian)mechanics – although an analogous result may be obtained
in quantum mechanics. Restriction on the flow of points in phase space is that no two
curves can intersect. [2]

e) Boltznann’s H theorem: Essentially Boltzmann’s H theorem says that the density


of points in phase space decreases as time evolves. Strictly it is specified in terms of
ρ ln ρ which, when integrated gives the mean value of lnρ – connection with entropy.
[2]

The paradox is that while Liouville’s theorem states that ρ remains constant, the H
theorem states that ρ decreases. [2]
Paradox may be resolved by considering the dendritic nature of the flow in phase
space and coarse-graining:

evolves to which
appears as

1 2 3

Initial state represented by picture 1. Some time later this has evolved to picture 2.
The density of points in the enclosed ‘volume’ has remained the same – in accordance
with Liouville’s theorem. One erects a grid; the reasoning is that one cannot
practically discern detail in phase space on a scale finer than the grid. Thus one
considers the mean density in each cell– called coarse-graining. And then the
apparent density is as in picture 3; the density ρ appears to have reduced and the
volume occupied appears to have increased. Ultimately the Uncertainty Principle
would provide an appropriate scale for coarse-graining. [2]
PH4211 Outline Solutions 2006

Question 1

(a) Λ is the de Broglie wavelength corresponding to the momentum corresponding to


the kinetic energy corresponding to the temperature:
Λ ~ h p where p ~ 2mE where E ~ kT [2]

(b) For distinguishable objects one multiplies the individual partition functions so
that Z = z N . For indistinguishable objects one must take account of permutations,
which result in identical states. Thus we would have
n !n !… N
Z= 1 2 z
N!
where N is the number of particles, n1 is the number of particles in state 1 etc.

When multiple occupancy may be ignored then


1 N
Z= z ,
N!
as required.

So the N! takes account of indistinguishability and the expression is valid when


multiple occupancy may be ignored – when its probability is very small. [5]

(c) Helmholtz free energy F:


F = − kT ln Z
But
ln Z = N ln z − ln N !
= N ln z − N ln N + N ln e
= N ln ( ze N )
using Stirling’s approximation.
Now use
z = V Λ3
so that
ln Z = N ln (Ve N Λ 3 )
and then
F = − NkT ln (Ve N Λ 3 ) . [5]

(d) If the N! term were ignored then we would have


ln Z = N ln z
so that
F = − NkT ln (V Λ 3 ) .
This expression is unacceptable because the argument of the logarithm is extensive.
This means that F would increase with volume in a strange way; F would not be
extensive.
(e) given
2π 2
Λ=
mkT
Substitute values in to get
Λ (T = 3K ) = 5.02 ×10−10 m
Λ (T = 300 K ) = 5.02 × 10−11 m

Mean distance between atoms is


λ ~ ( kT p )
13

Use 1 bar = 105 N m-2 so that


λ (T = 3K ) = 7.45 ×10−10 m
λ (T = 300 K ) = 3.46 ×10−9 m
So at 300K the interparticle spacing is much greater than Λ so quantum effects can be
ignored (and multiple occupancy is not a problem). But at 3K λ and Λ are similar and
so quantum effects become important.
[4]
Total [20]

Question 2

Points that should be made include:


• Connection with Second Law [2]
• Reversibility appearing to occur at the microscopic level [2]
• But irreversibility appearing to occur at the macroscopic level [2]
• Need for a statistical description [2]
• Discussion of a macrostate as a region in phase space [2]
• Evolution of the region determined by equations of motion [2]
• Liouville’s theorem saying that ‘volume’ in phase space is constant [2]
• Boltzmann’s H theorem arguing that ‘volume’ in phase space increases [2]
• Nature of the ‘flow’ in phase space [2]
• Resolution of paradox in terms of coarse graining. [2]
• Possible mention of cosmological connections.
Total [20]
Question 3

(a) Equation looks like ideal gas equation of state but pressure increased by
a N 2 V 2 and volume decreased by Nb. The a term relates to the long-distance
attractive tail of the inter-particle interaction. This has the effect of decreasing the
pressure on the walls of the container. The b term relates to the repulsive hard core of
the inter-particle interaction. This has the effect of reducing the allowed volume the
particles can occupy. [3]

(b)
p

critical
point
T1
liquid
T2
gas T3
forbidden
region
superheated supercooled V
liquid gas

Curves out of the coexistence region are stable


Superheated and supercooled regions are metastable
Forbidden region is unstable [3]

(c) The critical point is characterised by (2 marks for this)


∂p ∂V = 0 and ∂ p ∂V = 0 2 2

So write the equation of state as


NkT N2
p= −a 2
V − Nb V
and differentiate (twice), set equal to zero and solve the equations. Working through
the algebra will give
8a
Tc = , Vc = 3Nb .
27kb
[4]
Alternative derivation (following H E Stanley): write the equation of state as
⎛ kT ⎞ 2 2 a 3 ab
V 3 − N ⎜b + ⎟V + N V − N =0.
⎝ p ⎠ p p
In general the solutions to this equation give three values for V (possibly two are
complex). At the critical point the three values coalesce – the 3 roots of
⎛ kT ⎞ a ab
V 3 − N ⎜ b + c ⎟V 2 + N 2 V − N 3 =0
⎝ pc ⎠ p c p c

are equal. Then this equation must be equivalent to the cubic


(V − Vc ) = 0
3
or
V 3 − 3VcV 2 + 3Vc2 − Vc3 = 0 .
So equate coefficients of the two equations to get
⎛ kT ⎞ a ab
3Vc = N ⎜ b + c ⎟ , 3Vc2 = N 2 , Vc3 = N 3 .
⎝ pc ⎠ pc pc
Then these are easily solved to give
8a
Tc = , Vc = 3 Nb.
27 kb

(d) To do this one write the equation of state as


−1 2
p N⎛ N⎞ a ⎛N⎞
= ⎜1 − b ⎟ − ⎜ ⎟
kT V ⎝ V ⎠ kT ⎝ V ⎠
and then expand the first term using the binomial series
2
p N⎛ N ⎞ a ⎛N⎞
= ⎜1 + b + … ⎟ − ⎜ ⎟
kT V ⎝ V ⎠ kT ⎝ V ⎠
2
N ⎛ a ⎞⎛ N ⎞
= + ⎜b − ⎟⎜ ⎟ +…
V ⎝ kT ⎠ ⎝ V ⎠
thus we identify the second virial coefficient as
⎛ a ⎞
B2 (T ) = ⎜ b − ⎟.
⎝ kT ⎠
[3]

(e) From the above equation we find the parameters a and b as


B (T ) − B2 (T1 ) T B (T ) − T1 B2 (T1 )
a = kT1T2 2 2 , b= 2 2 2
T1 − T2 T1 − T2
and from the given data calculate
a = 3.86 × 10−49 J m3 , b = 6.48 × 10−29 m3
Then using Tc = 8a 27kb , this gives a Tc of 127.6.
[3]

(f) The ‘calculated’ value is 128K to 3 sig. figs, whereas the experimental value is
126K. This small discrepancy is related to the fact that the van der Waals equation is
only an approximation to the behaviour of a real gas. The fact that the discrepancy is
indeed small is an indication that the van der Waals equation is quite good. [2]

(g) This corresponds to moving along the curve of the ‘superheated liquid’ in the
figure above, rather than joining the (equilibrium) coexistence line. The curve is
described by the van der Waals equation of state. The limit of superheating (the
minimum pressure) is at the minimum of the isotherm. So knowing the parameters a
and b of the equation of state, one can find the pressure at which the derivative is a
minimum. (the algebra is a bit messy). [2]
Total [20]
Question 4

(a) History: Brown’s observation. Other examples. Randomness of the Brownian


motion. Equipartition applies to a Brownian particle. Langevin equation – friction
proportional to velocity – fluctuation-dissiation relation. Diffusive motion –
expressions for diffusion coefficient. [10]

(b) Critical exponents – behaviour in the vicinity of the critical point. Scaling
arguments – a single length parameter. Law of corresponding states. Discussion of the
Ehrenfest classification of phase transitions and the idea that it is not really
appropriate/useful. [10]

(c) Ising model as a description of magnetic phenomena. Has non-conserved order


parameter in this case. Mention of a scalar order parameter. Binary mixture can also
be expressed in this form. And lattice gas model of fluid systems. But both of these
have conserved order parameter. [10]

Total [20]
Question 5

(a) The order parameter is a measure of the order on the ordered phase. It is taken to
be zero in the disordered phase. The order parameter may be a real scalar, a complex
scalar, a vector or a tensor – depending on the nature of the order. [3]

(b) First order transition – order parameter is discontinuous at the transition. In a


second order transition the order parameter is continuous. [3]

(c) Symmetry might require odd terms to vanish. For instance if the order parameter
is a vector then you must take the dot product of it with itself to make the (scalar) free
energy.

The highest order term must be an even term and its coefficient must be positive so
the system is stable – the order parameter must be bounded; it can’t go off to infinity.

The order of the expansion must be sufficiently high that the free energy curve
exhibits the phenomena under investigation.

But the expansion must not go to higher order than strictly needed – the Landau free
energy is supposed to describe generic behaviour. And higher order terms would just
add superfluous detail. [3]

(d) Plot of free energy….


Landau
free F T/Tc = 1.2 T /T c = 1
energy

T/Tc = 0.8

m
-1 -0.5 0.5 1

[3]

(e) Write free energy as


F = F2ϕ 2 + F4ϕ 4 .
Equilibrium states found by finding minima of F:
dF
= 2 F2ϕ + 4 F4ϕ 3 = 0

so that
− F2
ϕ =0 or ϕ =± .
2 F4
We know that F4 is positive. So when F2 < 0 there are three stationary points while
when F2 > 0 there is only one. The critical point is where the roots coalesce – so this
means the vanishing of F2 at the critical point. Simplest assumption is a linear
variation
F2 = a (T − Tc )
while F4 is a constant b. Then in the vicinity of the critical point
a (T − Tc )
ϕ =± .
2b
Thus the order parameter critical exponent has the value 1/2. [6]

(f) Real systems often have a beta closer to 1/3. Importance of the role of
fluctuations. [2]

Total [20]