PH 253 / LeClair Fall 2013

Problem Set 7 Solutions

1. The specific heat at constant volume per electron is defined as

dhEi
cV = (1)
dT

where hEi is the average energy of electrons. A classical electron would have an average energy of
3
2 kB T . Presuming the electrons follow a Fermi-Dirac distribution, estimate the specific heat for a
collection of free electrons. Presume low temperatures, i.e., kB T  EF . Hint: electrons that have
been thermally excited above the Fermi energy behave basically as free electrons. What fraction of
electrons are excited above EF , roughly?

Solution: Roughly speaking, only electrons within a few kB T of the Fermi energy EF will con-
tribute to the specific heat, because only those electrons will have available states at a nearby
energy. How many kB T should we take? We could characterize the ’width’ of the distribution f (E)
by taking df /dE (a function peaked at EF ) and finding its full width at half maximum, which is
about 3.5kB T . About half of that width (the upper half) ought to be electrons available. (Or you
could have just said “about three kB T ” and it amounts to the same thing.)The fraction of electrons
contributing then would be

3.5kB T
fraction = (2)
2EF

The average energy of these electrons would then be the classical result 32 kB T times the fraction of
electrons actually available we just found:

3

3.5kB T
 2 T2
21 kB
hEi = kB T = (3)
2 2EF 8 EF

The specific heat is then approximately

dhEi 2T
21 kB k2 T
cV = ≈ = 5.25 B (4)
dT 4 EF EF

We find cV linear in T , which agrees with experiments at low temperature. The exact result using
2 k2 T k2 T
the full Fermi-Dirac distribution is π2 EBF ≈ 4.9 EBF , off by about 7%. Of course, we could just
have easily decided to use 2 or 3 kB T as our fraction, so the agreement is a bit spurious, but the
facts that cV ∝ T and that we have the right order of magnitude still hold true.

2. Following section 10.6 in your textbook (find one), you can find the total energy of a photon gas
U as well as the total number of photons in the gas N .i Recall that a photon gas was our model
for blackbody radiation! (a) What is the total specific heat CV for the photon gas, noting

∂U
CV = (5)
∂T

(b) How does the average energy per photon (U/N ) vary with temperature? (c) Given this average
energy per photon, estimate the specific heat per photon using the expression given in problem 1.

Solution: The internal energy of a photon gas of volume V , using the link in the footnotes, is

8π 5 kB
4
U= V T4 (6)
15c3 h3

You’d find this by integrating the Planck formula for the energy density over all frequencies, by the
way. The total specific heat (or heat capacity) is then

∂U 32π 5 kB
4
CV = = V T3 (7)
∂T 15c3 h3

The same link gives us the expected number of photons in the gas:

16πkB3 ζ(3)
N= V T3 (8)
c h3
3

where ζ(n) is the Riemann zeta function, and ζ(3) ≈ 1.202. The average energy per photon then
goes as

U π 4 kB T
= (9)
N 30ζ(3)

The specific heat per photon is then readily found.

∂(U/N ) πkB
cV = = (10)
∂T 240ζ(3)

The heat capacity per photon is a constant, independent of temperature.

3. Use the free electron theory to determine the Fermi energy and density of states vs. energy for
a two-dimensional metal. Take N as the average number of electrons per unit area.

Solution: In two dimensions, we can write the energy as a function of ~

k in terms of the x and y
i
For N , use the expression for dN in Eq. 10.38 and integrate it over all energies. Or look here: http://en.
wikipedia.org/wiki/Photon_gas. Always read the footnotes.
components of the k vector:

~2  2 
E(~
k) = kx + ky2 (11)
2m

What we need to find is the density of allowed k values. Given a crystal of side L, our boundary
conditions dictate that the k components must come in integer multiples of 2π/L. If we made a
plot with axes kx and ky , there would be allowed states at every integer multiple of 2π/L along
each axis, making a square lattice of such points. The area each allowed state takes up is then
dA = (2π/L)2 . That’s how much “k-space” each allowed state takes up.

What about the total number of states and their area? We have as many states as it takes to fill
all of them up to the Fermi wave vector kF , so all the allowed states lie within a circle of radius kF .
Their total area is thus A = πkF2 . The number of states is then the total area of all states divided
by the area of a single state, times 2 to account for the fact that each state can have one spin up
and one spin down electron:

A πkF2 kF2 L2
N =2 =2 = (12)
dA (2π/L)2 2π

Inverting, we can find what kF must be given N electrons in a square of side L in two dimensions:
√
2πN
|~
kF | = kF = (13)
L

The Fermi energy is then readily found from the energy-wavevector relationship

~2 kF2 πN ~2
EF = = (14)
2m mL2

How about the density of states? From the preceding equation, at an energy E we know what the
energy is as a function of N by virtue of knowing E(k) and k(N ). Equation 13 tells us that for a
given N , k 2 = 2πN/L2 , thus

~2 ~2 2πN π~2 N
 
E(N ) = (k(N ))2 = = (15)
2m 2m L2 mL2

The density of states per unit volume just depends on dE/dN and the volume V = L2 :

1 dN 1 1 1 mL2 m
g(E) = = 2 = 2 2
= (16)
V dE L dE/dN L π~ π~2

In 2D, the density of states is a constant, independent of energy.

4. (a) Obtain an expression for the Fermi energy at T = 0 K for an electron gas in a three
dimensional metal in terms of the total number of electrons, the volume, and fundamental constants.
(b) At T = 0 K, what is the average speed, in terms of the Fermi energy, of a three-dimensional
electron gas in a metal?ii

Solution: The Fermi energy at T = 0 is given by

!2/3
~2 3π 2 N
EF T =0
= (17)
2m V

where N is the number of electrons in a volume V . A derivation may be found in your textbook,
or here:

http://en.wikipedia.org/wiki/Fermi_energy

This is the energy of the highest-energy electrons in a metal at T = 0. If all this energy is available
as kinetic energy,
s
1 2EF
Ef = mv 2 =⇒ v= (18)
2 m

5. Show that the average kinetic energy of an electron in a three-dimensional electron gas at 0 K
is Eav = 35 EF .iii

Solution: The Fermi energy, the energy of the highest occupied quantum state in a system of
fermions at absolute zero temperature, is a function of the number of fermions as found in the last
problem. If there are N particles in a volume V , then the highest occupied state has an energy
!2/3
~2 3π 2 N
EF (T = 0) = (19)
2m V

Thus, as more and more particles are added, it takes more and more energy to add the last particle.
The total energy of the system, if there are enough particles to consider the distribution of states
to be quasi-continuous, is given by

ZN
Etot = EF (N 0 ) dN 0 (20)
0
ii
Open your textbook to 10.7.
iii
Open your textbook to 10.7.
The average energy is simply the total energy divided by the number of particles:

ZN ZN !2/3 !2/3 
1 1 ~2 3π 2 N 0 1 ~2 3π 2 3 5/3 3

Eavg = EF (N 0 ) dN 0 = dN 0 = N = EF (21)
N N 2m V N 2m V 5 5
0 0

6. Now that you’ve followed the derivation for the three dimensional case from the textbook, repeat
the previous two problems for a two dimensional electron gas, using your results from problem 3.

Solution: We already found the Fermi energy in problem 3, conveniently enough.

~2 kF2 πN ~2
EF = = (22)
2m mL2

The average speed at the Fermi energy is the same as in three dimensions (in terms of EF anyway):
s s
2EF 2πN ~2 √ ~
v= = 2 2
= 2πN (23)
m m L mL

If the density of states is constant, the average energy is easy - it must be half the maximum value,
1
2 EF . As a sanity check, we can proceed as in the previous problem:

ZN ZN
1 0 0 1 πN 0 ~2 0 1 πN 2 ~2 πN ~2 1
Eavg = EF (N ) dN = 2
dN = 2
= 2
= EF (24)
N N mL N 2mL 2mL 2
0 0