Cornell University, Physics Department Fall 2013

PHYS-3341 Statistical Physics Prof. Itai Cohen

Solutions to Problem Set 6
David C. Tsang, Woosong Choi
6.1 Harmonic Oscillator
Reif 6.1: A simple harmonic one-dimensional oscillator has energy levels given
by E
n
= (n +
1
2
), where is the characteristic (angular) frequency of the
oscillator and where the quantum number n can assume the possible integral
values n = 0, 1, 2, . . .. Suppose that such an oscillator is in thermal contact with
a heat reservoir at temperature T low enough so that kT/() 1.
(a) Find the ratio of the probability of the oscillator being in the rst excited
state to the probability of its being in the ground state.
(b) Assuming that only the ground state and rst excited state are apprecia-
bly occupied, nd the mean energy of the oscillator as a function of the
temperature T.
(a) We have
P
1
P
0
=
exp[E
1
]
exp[E
0
]
=
exp[(1 + 1/2)]
exp[(0 + 1/2)]
= e

(b) The average energy is given by

E =

r
e
Er
E
r

r
e
Er
=
E
o
e
Eo
+ E
1
e
E
1
e
Eo
+ e
E
1
which gives

E =
1
2
+
3
2
P
1
P
0
1 +
P
1
P
0
=

2
1 + 3e

1 + e

2
1
6.2 Two State System
Reif 6.6: A system consists of N weakly interacting particles, each of which
can be in either of two states with respective energies
1
and
2
, where
1
<
2
.
(a) Without explicit calculation, make a qualitative plot of the mean energy

E
of the system as a function of its temperature T. What is

E in the limit
of very low and very high temperatures? Roughly near what temperature
does

E change from its low to its high temperature limiting values?
(b) Using the result of (a), make a qualitative plot of the heat capacity C
V
(at
constant volume) as a function of the temperature T.
(c) Calculate explicitly the mean energy

E(T) and heat capacity C
V
(T) of this
system. Verify that your expressions exhibit the qualitative features dis-
cussed in (a) and (b).
(a) For a system in contact with heat reservoir we know that the probabilty of the system
being in a state with energy E is proportional to e
E
, the Boltzmann factor. Thus,
without detailed calculation, we can deduce that low T limit gives

E = N
1
and high
T limit

E = N

1
+
2
2
. (In low temperatures, particles will prefer to sit in a lower energy
state. In high temperatures, particles will ll both state equally likely.) Now, there is
one energy scale associated with the dierence between the two states =
2

1
.
Roughly, the transition of

E will occur at = 1.
T =

2

1
k
(b) Qualitatively, the energy should approach /2 as T and as T 0, energy goes
to 0. Also, the probability ratio of being in the two states is an exponential function,
the Boltzmann factor, this is not a linear graph. Thus it should be something like a
smooth step function changing its value at around kT.
C
V
is a rst derivative of E(T) thus is should be peaked at around kT according
to the argument and should approach zero for high and low temperatures.
(c) Using the Boltzmann factor,

E = N

1
e

1
+
2
e

2
e

1
+ e

2
=

1
e +
2
e
(
2

1
)
1 + e
(
2

1
)
We can easily verify that in the low temperature limit, , we get

E
1
. For
high temperature limit, 0, the Boltzmann factor approch one, thus

E

1
+
2
2
.
To nd C
V
we take the derivative of this
C
V
=


E
T
= N

2
e

(1 + e

) (
1
+
2
e

)(e

)
(1 + e

)
2

T
=
N
kT
2

2
e

(1 + e

)
2
2
Plotting these functions we get gures 1 and 2. And from these we can indeed see the
features we predicted. C
V
is peaked around some value , and goes to 0 as T is
greater or smaller than the value.
Figure 1: E/ vs kT/
2
3
Figure 2: C
V
/k vs kT/
6.3 Dilute Gas in an Enclosure
(a) By symmetry we see
v
z
0 .
By the maxwell-boltzmann distribution we have
|v
z
| =
_

|v
z
|e
mv
2
z
/2kT
dv
z
_

e
mv
2
z
/2kT
dv
z
=
2
_

0
v
z
e
mv
2
z
/2kT
dv
z
2
_

0
e
mv
2
z
/2kT
dv
z
=
2kT/m
_
2kT/m
=
_
2kT
m
v
2
=
2
_

0
v
2
z
e
mv
2
z
/2kT
dv
z
2
_

0
e
mv
2
z
/2kT
dv
z
=
1
2

_
2kT
m
_
3/2

_
2kT
m
_
1/2
=
kT
m
(b) Dene (v
z
) as the # of molecules escaping per unit time per unit z-velocity. That is
=
_

0
(v
z
)dv
z
. Now (v
z
) = Av
z
n(v
z
) where n(v
z
) is the number of molecules with
z-velocity v
z
per unit volume per unit z-velocity.
n(v
z
) =
N
V
e
mv
2
z
/2kT
_

e
mv
2
z
/2kTdvz
=
AN
2V
_

0
e
mv
2
z
/2kT
dv
z
_

0
e
mv
2
z
/2kTdvz
=
AN
2V
|v
z
| =
AN
V
_
kT
2m
4
(c) Now assume we cover the hole with a plate and assume elastic collisions on that plate.
Then the froce on that plate per unit time is
F = p = 2mv
z,esc
where p is the average change in momentum due to an elastic collision.
P =
F
A
=
2m
A
v
z,esc
(d)
v
z,esc
=
AP
2m
=
PV
Nm
_
m
2kT
=
kT
m
_
m
2kT
=
_
kT
2m
=

2
|v
z
|
For an ideal gas with maxwell-boltzmann distribution we have v
z,esc
> |v
z
|. Another
way of seeing this is to recognize that the escaping molecules are pushed preferentially
on one side, giving them a greater average speed.
2
6.4 Statistical Entropy
(a) For the microcanonical ensemble, we know S = k ln . So we need to show that
k

r
P
r
ln P
r
= k ln . For the microcanonical ensemble: P
r
= 1/ for E < E
r
<
E + E and 0 otherwise, thus we have
k

r
P
r
ln P
r
= k

r
1

ln
1

= k

r
1

ln 1 + k

r
1

ln
= k ln .
(b) Reif 6.13
S = k

s
P
rs
ln P
rs
= k

s
P
r
P
s
ln(P
r
P
s
) since P
r
and P
s
are independant
= k

s
P
s

r
P
r
ln P
r
k

r
P
r

s
P
s
ln P
s
= k(1)(S
1
/k) k(1)(S
2
/k)
= S
1
+ S
2
(1)
(b) Reif 6.15
S = k

r
P
r
ln P
r
S
o
= k

r
P
(o)
r
ln P
(o)
r
5
S S
o
= k

r
_
P
r
ln P
r
+ P
(o)
r
lnP
(o)
r

= k

r
_
P
r
ln P
r
+ P
r
ln P
(o)
r
P
r
ln P
(o)
r
+ P
(o)
r
ln P
(
r
o)

= k

r
P
r
ln
P
(o)
r
P
r
+ k

r
_
P
r
(E
r
ln Z) + P
(o)
r
(E
r
ln Z)

(Since ln P
(o)
r
= E
r
ln Z)
= k

r
P
r
ln
P
(o)
r
P
r
+ k

:
0
(

r
P
r
E
r

r
P
(o)
r
E
r
) + k lnZ

:
0
(

r
P
r

r
P
(o)
r
)
= k

r
P
r
ln
P
(o)
r
P
r
.
We also note that ln x x 1 which implies
S S
o
= k

r
P
r
ln
P
(o)
r
P
r
k

r
(P
(o)
r
P
r
) = 0
Thus we must have
S S
o
where equality represents the most likely canonical distribution.
6.5 Two-Dimensional Gas
Reif 7.7: Monoatomic molecules adsorbed on a surface are free to move on this
surface and can be traeted as a classical ideal two-dimensional gas. At absolute
temperature T, what is the heat capacity per mole of molecules thus adsorbed
on a surface of xed size?
In an ideal gas the particles are non-interacting , and there are no external potentials,
therefore V = 0. Hence we have the energy E = K the kinetic energy. The equipartition
theorem gives us

E =

K = 2N
_
1
2
kT
_
= NkT
where 2 comes from the number of dimensions and N is the number of particles.
Thus we have
C =


E
T
= Nk = R
C
n
= R
2
6
6.6 More Oscillators
Reif 7.10: A system consists of N very weakly interacting particles at a tem-
perature T suciently high so that classical statistical mechanics is applicable.
Each particle has mass m and is free to perform one-dimensional oscillations
about its equilibrium position. Calculate the heat capacity of this system of
particles at this temperature in each of the following cases:
(a) The force eective in resotring each particle to its equilibrium position is
proportional to its displacement x from this position.
(b) The restoring force is proportional to x
3
.
(a) We have an energy of the form
E =
N

i
p
2
i
m
+
1
2
k
o
x
2
i
Thus there are two seperate quadratic degrees of freedom per particle, hence we have
by equipartition

E = NkT C =
_
E
T
_
= Nk
(b) For this system we have the energy
E =
N

i
p
2
i
m
+
1
4
k
o
x
4
i
The classical partition function is then
Z =
1
h
N
0
N

i
_

e
p
2
i
/2m
dp
i
_

e
kox
4
i
/4
dx
i
=
_
2m
h
2
o

_
N/2
_

1/4
_

e
koy
4
/4
dy
_
N
(where y =
1/4
x)
=
_
2m
h
2
o
_
N/2
__
e
koy
4
/4
dy
_
N

3N/4
Since the rst two terms in the product do not depend on we must have

E =
ln Z

=
3
4
NkT C =
3
4
Nk
2
7
6.7 Compressibility of Copper
Reif 7.12: Consider a solid of compressibility . Assume that the atoms in this
solid are arranged on a regular cubic, lattice, the distance between their nearest
neighbors being a. Assume further that a restoring force
0
a acts on a given
atom when it is displaced by a distance a from its nearest neighbor.
(a) Use simple reasoning to nd an approximate relation between the spring
constant
0
and the compressibility of this solid. (Consider the force
needed to decrease the length of one edge of a solid parallelepiped by a
small amount.)
(b) Estimate roughly the order of magnitude of the Einstein temperature
E
for copper (atomic weight = 63.5) by assuming that it is a simple cubic
structure with density 8.9gcm
3
and compressibility 4.5 10
13
cm
2
dyne
1
.
(a) Consider a column of atoms in the solid pressed with a force F. Then, since all the atoms
are connected in series every nearest neighbor distance will change by a =
F

0
.
The compressibility is dened as =
1
V
_
V
p
_
T
. V for this column is a
3
N where N
is the number of atoms in the column. The change in volume is thus V = Na
2
a
and the pressure is p = F/a
2
. Now we have all the components. Putting in the values,
=
1
Na
3
Na
2
a

0
a/a
2
=
a

0
(b) The Einstein temperature
E
=

k
can be calculated by nding out what the is for
the oscillators in the solid. Classically,
2
= k/m where k is the spring constant. For
this model, we found what the spring constant
0
is, so we need to plug in numbers.

E
=

k
=

_

0
/m
k
=

_
a/m
k
=

k
_
(m/)
1/3
m
In c.g.s units, we are given 1.055 10
27
erg s, 1.38 10
16
erg K, m =
63.5g/6.022 10
23
, = 8.9g cot cm
3
, and = 4.5 10
13
cm
2
dyne
1
. Substitut-
ing the values

E
=
1.055 times10
27
1.38 10
16
erg K

(63.5/8.9/6.022 10
23
)
1/3
63.5/6.022 10
23
4.5 10
13
167

K
2
8
6.8 Ferro Fluids
Reif 7.14: Consider an assembly of N
0
weakly interacting magnetic atoms per
unit volume at a temperature T and describe the situation classically. Then
each magnetic moment can make any arbitrary angle with respect to a given
direction (call it the z direction). In the absence of a magnetic eld, the proba-
bility that this angle lies between and +d is simply proportional to the solid
angle 2sind enclosed in this range. In the presence of a magnetic eld H in
the z direction, this probability must further be proportional to the Boltzmann
factor e
E
, where E is the magnetic energy of the moment making this angle
with the z axis. Use this result to calculate the classical expression for the
mean magnetic moment

M
z
of these N
0
atoms.
Following the argument given in the problem we have
P()d 2sinde
Hcos
for one magnetic moment. Since they are only weakly interacting, we can consider the atoms
individually and sum up in the end. Since we have the proportionality of P() we can nd
the partition function as follows:
Z =
_

0
d2sinde
Hcos
= 2
_
1
1
d(cos)e
Hcos
= 2
_
e
Hcos
H
_
1
1
=
4
H
sinh(H)
To nd the average magnetization of the magnetic moment in z-direction,

z
=
ln Z
(H)
= coth(H)
1
H
thus for N
0
atoms

M
z
= N
0

_
coth(H)
1
H
_
2
9
6.9 Mean Values
Reif 7.19: A gas of molecules, each of mass m, is in thermal equilibrium at
the aboslute temperature T. Denote the velocity of a molecule by v, its three
cartesian components by v
x
, v
y
and v
z
, and its speed by v. What are the following
mean values:
(a) v
x
(b) v
2
x
(c) v
2
v
x
(d) v
3
x
v
y
(e) (v
x
+ bv
y
)
2
where b is a constant
(f) v
2
x
v
2
y
(a) v
x
= 0 since v
x
is an odd function.
(b) v
2
x
=
kT
m
by the equipartition thereom.
(c) v
2
v
x
= 0 since v
2
v
x
is an odd function
(d) v
3
x
v
y
= 0 since v
3
x
v
y
is odd in both v
x
and v
y
.
(e) (v
x
+ bv
y
)
2
= v
2
x
+ 2bv
x
v
y
+ bv
2
y
=
kT
m
+ 0 +
bkT
m
= (1 + b)
kT
m
.
(f) v
2
x
v
2
y
= v
2
x
v
2
y
=
_
kT
m
_
2
since the distributions of v
x
and v
y
can be considered independent.
2
10
6.10 Emissions and the Doppler Eect
Reif 7.22 A gas of atoms, each of mass m, is maintained at the absolute
temperature T inside an enclosure. The atoms emit light which passes (in the
x direction) through a window of the enclosure and can then be observed as a
spectral line in a spectroscope. A stationary atom would emit light at the sharply
dened frequency
o
. But, because of the Doppler eect, the frequency of the
light observed from an atom having an x component of the velocity v
x
is not
simply equal to the frequency
o
, but is approximately given by
=
o
_
1 +
v
o
c
_
where c is the velocity of light. As a result, not all of the light arriving at the
spectroscope is at a frequency
o
; instead it is characterized by some intensity
distribution I()d specifying the fraction of light intensity lying in the frequency
range between and + d. Calculate
(a) The mean frequency of the light ovserved in the spectroscope.
(b) The root-mean square frequency of the shift ()
rms
= [( )
2
]
1/2
(mea-
sured from the mean frequency) of the light observed in the spectroscope.
(c) The relative intensity distribution I()d of the light observed in the spec-
troscope.
We have (v
x
) =
o
(1 + v
x
/c) where v
x
is a random variable with a distribution
P(v
x
) e

m
2
v
2
x
This gives a distribution
P()d =

mc
2
2kT
2
o
e

mc
2
2kT
2
o
(o)
2
d
(a) We must then have
=
o
,
since the mean value of v
x
is 0.
(b) The distribution of gives us

=
kT
2
o
mc
2
(c) We must be a little careful here, and think carefully about how intensity depends on
frequency. It turns out that the intensity per photon divided by the cube of the
photons frequency is invariant under changes of reference frame. In other words
i()

3
=
i(

3
11
where

id =
Id
N
, is the intensity for a given emission photon, and N is the number of
such emissions.
We must therefore have
I()d =
_

o
_
3
I
o
e

mc
2
2kT
2
o
(o)
2
d
where I
o
is an arbitrary normalization. Note that for kT mc
2
, or the non-relativistic
regime, we have the (/
o
)
3
term falling o as (/
o
)
3
(1+3

/
o
) = (1+3
_
kT/mc
2
)
1 for most of the distribution.
Thus for gases that have temperature much less than mc
2
/k we can write
I()d = I
o
e

mc
2
2kT
2
o
(o)
2
d
as the Reif claims.
2
6.11 Isotope Separation
Reif 7.26: A vessel is closed o by a porous partition through which gases can
pass by eusion and then be pumped o to some collecting chamber. The vessel
itself is lled with a dilute gas consisting of two types of molecules which dif-
fer because they contain two dierent atomic isotopes and have correspondingly
masses m
1
and m
2
. The concentrations of these molecules are c
1
and c
2
, respec-
tively, and are matintained constant inside the vessel by constantly replenishing
the cupply of gas in it by a steady slow ow of fresh gas through the vessel.
(a) Let c

1
and c

2
denote the concentrations of the two types of molecules in the
collecting chamber. What is the ratio c

2
/c

1
?
(b) By using the gas UF
6
, one can attempt to separate U
235
from U
238
, the
rst of these isotopes being the one useful in initiation of nuclear-ssion
reactions. The molecules in the vessel are then U
238
F
19
6
and U
235
F
19
6
. (The
concentrations of these molecules, corresponding to the natural abundance
of the two uranium isotopes, are c
238
= 99.3 percent and c
235
= 0.7 percent.)
Calculate the corresponding ratio c

235
/c

238
of the molecules collected after
eusion in terms of their original concentration ratio c
235
/c
238
.
(a) The ratio of concentration of the molecules in the collecting chamber will be the ratio
of the molecules eused through the porous partition. From (7.11.13) average number
of particles that will hit, i.e. pass through, the holes per unit area is

0
=
p

2mkT
=
nkT

2mkT
12
where p is the mean pressure of the molecules, thus can be substituted to nkT using
the equation of state. T is the same for the two types of molecules thus
0

n

m
is
the relation relevant to the ratio.
c

1
c

2
=
c
1
c
2
_
m
2
m
1
(b) For UF
6
, using the result of (a) we get
c

235
c

238
=
c
235
c
238
_
m
238
m
235
=
c
235
c
238
_
352
349
= 1.004
c
235
c
238
Although the change in concentration ratio is small, by repeatedly using eusion we
can increase the concentration of U
235
in the mixture.
2
13