PHYS 352 Homework 1 Solutions

Aaron Mowitz (1 and 2) and Nachi Stern (3, 4, and 5)

Problem 1
We will solve this problem using the microcanonical ensemble. The temperature of a thermody-
namic system is defined by  
1 ∂S
=
T ∂E N
Each link in the polymer either points left or right, i.e. has two possible states. If n← links are
pointing left and n→ are pointing right, the total number of possible configurations of the polymer
is the number of ways one can arrange the left and right facing links. This is given by the binomial
coefficient:
N!
Ω=
n← !n→ !
We then can write the entropy as
S = kB ln Ω = kB (ln N ! − ln n← ! − ln n→ !)
≈ kB (N ln N − n← ln n← − n→ ln n→ )
where Stirling’s approximation is used in the last line. We also have the following constraints on
the total number of links and the total energy:
N = n← + n→
E = −qEL = −qE (n→ − n← ) a
Since we are working in the microcanonical ensemble, both of these quantities are fixed. We can
use these constraints to rewrite the entropy in terms of E and N :
        
1 E E 1 E E
S = kB N ln N − N− ln N − − N+ ln N +
2 qEa qEa 2 qEa qEa
Now we can find the temperature using the relation between temperature and entropy we wrote
before:
E
N − qEa
      
1 kB 1 E 1 E kB
= ln N− − ln N+ = ln E
T 2qEa 2 qEa 2 qEa 2qEa N + qEa

The length of the polymer is related to the energy by E = −qEL, so we can invert the above
relation to express the length of the polymer in terms of the temperature:
!
2 qEa
L = Na 1 − 2qEa = N a tanh
1 + e kB T kB T

Since tanh x is an increasing function of x, the length of the polymer will indeed decrease as
temperature increases.

1
Problem 2
We will solve this problem using the canonical ensemble. Assuming the N systems in our collection
are distinguishable, we can write the partition function of the entire system as a product of the
partition functions of N three-level systems:
 N
Z = Z1N = 1 + e−β + e−2β

We can then find the average energy of the system using this partition function:

∂ ln Z e−β + 2e−2β
E=− =N
∂β 1 + e−β + e−2β
This can be inverted to find T in terms of the energy:

√ !!−1
 −3E 2 + 6EN  + N 2 2 + N  − E
T =− ln
kB 2(2N  − E)

(Note: when inverting, there are two possible solutions. However, the other yields an imaginary
temperature, which is unphysical) To find the entropy, we use the relation between the Helmholtz
free energy and canonical partition function:

F = E − T S = −kB T ln Z

This gives us
   
S = kB ln Z + E/T = kB N ln 1 + e−β + e−2β − E/ ln e−β
   
N   −β
= kB N ln e + 2e−2β − E/ ln e−β
E
   
N 2


= kB N ln x + 2x − E/ ln x
E
√
2 2 2
where x = e−β = −3E +6EN +N  +N −E
2(2N −E) . One can also solve this problem via the microcanonical
ensemble, similar to problem 1. However, since there are 2 constraints (total energy and total
number of systems) but 3 unknowns (number of systems in each of the three states), there will be
one free parameter (e.g. the number of systems with energy ). To find this, one must maximize
the entropy with respect to this parameter, then one can proceed as in the first problem.

2
3) Quantum-Classical Correspondence in a Harmonic Oscillator
𝒑𝟐 𝟏
i) For the harmonic oscillator 𝑯 = 𝟐𝒎 + 𝟐 𝒎𝝎𝟐 𝒙𝟐, find the number of energy levels with energy less
than 𝑬.

First consider the classical harmonic oscillator: Fix the energy level 𝐻 = 𝐸, and we may rewrite the
energy relation as

𝑝2 1 1 𝑚𝜔2 2
𝐸= + 𝑚𝜔2 𝑥 2 → 1=( ) 𝑝2 + ( )𝑥
2𝑚 2 2𝑚𝐸 2𝐸

In phase space, this equation corresponds to an ellipse, with semi-axes

2𝐸 1
𝑎 = √2𝑚𝐸 𝑏=√
𝑚𝜔

The volume of phase space contained within the ellipse is just its surface. As lower energy trajectories
are also ellipses contained within this ellipse, we have

2𝜋𝐸 𝑑𝑥𝑑𝑝 𝐸
𝑆=∫ 𝑑𝑥𝑑𝑝 = 𝜋𝑎𝑏 = → ∫ =
𝐻<𝐸 𝜔 𝐻<𝐸 2𝜋ℏ ℏ𝜔

Now for the quantum harmonic oscillator, whose energy level are given by
1
𝐸𝑛 = ℎ𝜔 (𝑛 + ) ≈ ℎ𝜔𝑛
2
Where the approximation is justified for high energy levels. The energy levels are equally spaced, so the
number of energy levels below 𝐸 is just

𝐸 𝑑𝑥𝑑𝑝
𝑁≈ =∫
ℏ𝜔 𝐻<𝐸 2𝜋ℏ

3
 𝒑𝟐
ii) For the a more general potential 𝑯 = 𝟐𝒎 + 𝑽(𝒙), find the number of energy levels with energy less
than 𝑬.

Setting once more an energy level 𝐻 = 𝐸, one might express the momentum as

𝑝 = √2𝑚(𝐸 − 𝑉(𝑥))

𝑉(𝑥) is said to be monotonically decreasing and increasing on either side of 𝑥 = 0, in addition to

diverging at infinity. For a fixed energy 𝐸 there are only two solutions for 𝑝 = 0. The trajectory in phase
space will be given by a closed curve, with momentum switching directions when 𝐸 = 𝑉. Moreover, for
lower energies, the trajectories again must be contained within the curve defined by energy level 𝐸. We
can thus integrate over the phase space surface, and the result must be proportional to the number of
energy states inside it:

𝑆=∫ 𝑑𝑥𝑑𝑝
𝐻<𝐸

Recall Green’s theorem

𝜕𝐿 𝜕𝑀
∮ (𝐿𝑑𝑥 + 𝑀𝑑𝑦) = ∫ ( − ) 𝑑𝑥𝑑𝑦
𝜕𝑦 𝜕𝑥
𝐶 𝑆

Set 𝐿 = 𝑦 = 𝑝, 𝑀 = 0 and find

𝜕𝐿 𝜕𝑀 𝜕𝑝
∫( − ) 𝑑𝑥𝑑𝑦 = ∫ 𝑑𝑥𝑑𝑝 = ∫ 𝑑𝑥𝑑𝑝 = ∮ 𝑝𝑑𝑥
𝜕𝑦 𝜕𝑥 𝜕𝑝 𝐻<𝐸
𝑆 𝑆 𝐻=𝐸

Bohr-Sommerfeld quantization gives us a relation between the integral over the trajectory and the
quantum energy level 𝑁. This condition is written as

∮ 𝑝𝑑𝑥 ≈ ℎ𝑁 = 2𝜋ℏ𝑁
𝐻=𝐸

Dividing by the prefactor, we retrieve the result of the first part

𝑑𝑥𝑑𝑝
𝑁≈∫
𝐻<𝐸 2𝜋ℏ

4
4) Thermodynamics of Oxygen Production

Air is well approximated by an ideal gas, so we may use the Sackur-Tetrode relation for the entropy:
𝑉 2𝜋𝑚𝑘𝑇 3⁄2 5
𝑆(𝑁, 𝑉) = 𝑘𝑁 [ln [ ( 2
) ]+ ]
𝑁 ℎ 2
Defining 𝑁𝑂 and 𝑁𝑁 as the number of oxygen and nitrogen particles respectively, we make use of the
fact that 80% of the mixture is nitrogen to write
𝑁 4𝑁
𝑁𝑂 = ; 𝑁𝑁 =
5 5
In the separated state, the volume is proportional to the number of particles
𝑉 4𝑉
𝑉𝑂 = = 1𝐿 ; 𝑉𝑁 = = 4𝐿
5 5
The entropy of the mixed system is
𝑉 2𝜋𝑚𝑘𝑇 3⁄2 5 𝑉 2𝜋𝑚𝑘𝑇 3⁄2 5
𝑆𝑖 = 𝑆(𝑁𝑂 , 𝑉) + 𝑆(𝑁𝑁 , 𝑉) = 𝑘𝑁𝑂 [ln [ ( 2
) ] + ] + 𝑘𝑁𝑁 [ln [ ( 2
) ]+ ]
𝑁𝑂 ℎ 2 𝑁𝑁 ℎ 2
The entropy of the separated state is
𝑉𝑂 2𝜋𝑚𝑘𝑇 3⁄2 5 𝑉𝑁 2𝜋𝑚𝑘𝑇 3⁄2 5
𝑆𝑓 = 𝑆(𝑁𝑂 , 𝑉𝑂 ) + 𝑆(𝑁𝑁 , 𝑉𝑁 ) = 𝑘𝑁𝑂 [ln [ ( 2
) ] + ] + 𝑘𝑁𝑁 [ln [ ( 2
) ]+ ]
𝑁𝑂 ℎ 2 𝑁𝑁 ℎ 2
The change of entropy between the states is
𝑉 𝑉 𝑘𝑁 5 1
∆𝑆 = 𝑆𝑖 − 𝑆𝑓 = 𝑘𝑁𝑂 ln ( ) + 𝑘𝑁𝑁 ln ( ) = [ln(5) + 4 ln ( )] ≈ 𝑘𝑁
𝑉𝑂 𝑉𝑁 5 4 2
Using the ideal gas equation of state

𝑝𝑉 105 [𝑃𝑎] × 5 ⋅ 10−3 [𝑚3 ] 5 𝐽

𝑘𝑁 = ≈ =
𝑇 300[𝐾] 3𝐾
No work is done in the process and all energy goes to separating the gasses. By the 2nd law of
thermodynamics
1 1 5 𝐽
∆𝐸 ≥ 𝑇∆𝑆 ≈ 𝑘𝑇𝑁 ≈ × [ ] × 300[𝐾] = 250𝐽
2 2 3 𝐾
In practice some energy will usually transform into heat, so the required energy to separate the gases is
typically substantially higher.

5
5) A Thermodynamic Identity

Show that
𝐶𝑝 (𝜕𝑉/𝜕𝑝) 𝑇
=
𝐶𝑉 (𝜕𝑉/𝜕𝑝)𝑆

Nearly all of you solved this problem correctly, but I suspect many spent a lot of time trying guessing
which relations should be used in order to get the right answer. Well, there is a rather easy way to do
these kind of problems, that require little to no guesswork. It also works also for harder problems that
contain more than 4 dynamic variables (in this case 𝑉, 𝑇, 𝑆, 𝑝). This general method is adapted from
Gene Mazenko’s textbook “Equilibrium Statistical Mechanics”, and is based on the properties of
Jacobians.

The Jacobian for a 3-variable transformation (𝑢, 𝑣, 𝑤) → (𝑥, 𝑦, 𝑧) is defined as:

𝜕𝑢 𝜕𝑢 𝜕𝑢
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕(𝑢, 𝑣, 𝑤) 𝜕𝑣 𝜕𝑣 𝜕𝑣
≡ det
𝜕(𝑥, 𝑦, 𝑧) 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝑤 𝜕𝑤 𝜕𝑤
( 𝜕𝑥 𝜕𝑦 𝜕𝑧 )
And it may be easily generalized to more variables. The property that makes Jacobians useful is
𝜕𝑢 𝜕(𝑢, 𝑦, 𝑧)
( ) =
𝜕𝑥 𝑦,𝑧 𝜕(𝑥, 𝑦, 𝑧)

By the properties of determinants, one can show that (replacing rows)

𝜕(𝑢, 𝑣, 𝑤) 𝜕(𝑣, 𝑢, 𝑤)
=−
𝜕(𝑥, 𝑦, 𝑧) 𝜕(𝑥, 𝑦, 𝑧)
And the generalized chain rule and reciprocal relations
−1
𝜕(𝑢, 𝑣, 𝑤) 𝜕(𝑢, 𝑣, 𝑤) 𝜕(𝑟, 𝑠, 𝑡) 𝜕(𝑢, 𝑣, 𝑤) 𝜕(𝑥, 𝑦, 𝑧)
= ; =[ ]
𝜕(𝑥, 𝑦, 𝑧) 𝜕(𝑟, 𝑠, 𝑡) 𝜕(𝑥, 𝑦, 𝑧) 𝜕(𝑥, 𝑦, 𝑧) 𝜕(𝑢, 𝑣, 𝑤)

Let us use these rules to show the identity in question:

𝜕𝑄 𝜕𝑆 𝜕(𝑆, 𝑝) 𝜕(𝑆, 𝑝) 𝜕(𝑇, 𝑉) 𝜕(𝑆, 𝑝) 𝜕𝑉
𝐶𝑝 = ( ) = 𝑇( ) = 𝑇 =𝑇 =𝑇 ( )
𝜕𝑇 𝑝 𝜕𝑇 𝑝 𝜕(𝑇, 𝑝) 𝜕(𝑇, 𝑉) 𝜕(𝑇, 𝑝) 𝜕(𝑇, 𝑉) 𝜕𝑝 𝑇
𝜕𝑉
Note: I chose the chain rule above because I wanted to find a relation containing (𝜕𝑝 ) . There is no
𝑇
guesswork required!

6
 𝜕𝑉
Similarly we want to find an equation for 𝐶𝑉 containing (𝜕𝑝 ) :
𝑆

𝜕𝑄 𝜕𝑆 𝜕(𝑆, 𝑉) 𝜕(𝑆, 𝑉) 𝜕(𝑆, 𝑝) 𝜕(𝑆, 𝑝) 𝜕𝑉

𝐶𝑉 = ( ) = 𝑇( ) = 𝑇 =𝑇 =𝑇 ( )
𝜕𝑇 𝑉 𝜕𝑇 𝑉 𝜕(𝑇, 𝑉) 𝜕(𝑆, 𝑝) 𝜕(𝑇, 𝑉) 𝜕(𝑇, 𝑉) 𝜕𝑝 𝑆

Dividing both expressions:

𝜕(𝑆, 𝑝) 𝜕𝑉 𝜕𝑉
𝑇 ( ) ( )
𝐶𝑝 𝜕(𝑇, 𝑉) 𝜕𝑝 𝑇 𝜕𝑝 𝑇
= =
𝐶𝑉 𝜕(𝑆, 𝑝) 𝜕𝑉 𝜕𝑉
𝑇 ( ) ( )
𝜕(𝑇, 𝑉) 𝜕𝑝 𝑆 𝜕𝑝 𝑆