Fall 2013 - Physics 143 b

Problem Set 1 Solutions

Shubhayu Chatterjee

1. Griffiths 5.15
In the ground state of the electron gas, all the momentum eigenstates with momentum less than the
Fermi momentum kF will be filled. Since electrons are spin half particles, there are two electrons
(of opposite spin) for each value of allowed momenta.
So, the total energy in the ground state is given by
Etotal = 2|~k|<kF E(~k)
For large lattice sizes, we can use the continuum approximation, where we assume that each allowed
3
point in ~k space occupies a volume of (2)
L3 , and replace the summation by an integral over momenta.
More precisely, we can evaluate the total energy as
Etotal

2|~k|<kF E(~k)
Z
d3 k
= 2
E(~k)
3
3
|~
k|<kF (2) /L
Z
L3
E(~k)
= 2
(2)3 |~k|<kF
=

2 2
k
Now, we note that E(~k) = h2m
depends only on the magnitude of the vector ~k, we move to spherical
polar coordinates in the momentum space to simplify evaluating the integral.
Z kF
Z
Z 2
L3
h2 k 2
2
Etotal = 2
dk
k
d
sin()
d
(2)3 0
2m 0
0
Z kF
L3 h2
4
dk k 4
= 2
(2)3 2m
0

L3
h2 kF5
10 2 m

The total number of electrons is simply equal to the number of occupied states. The volume of the
3
occupied sphere is 34 kF3 , and each allowed value of momenta in ~k space occupies a volume of (2)
L3 .
Multipying by a factor of 2 for spin,


4
(2)3
L3 kF3
N = 2 kF3 /
=
3
3
L
3 2
The allowed momenta are determined by the boundary conditions - Griffiths uses hard boundary conditions, but
I am going to use periodic boundary conditions as done in class. In the end, the static properties turn out to be
independent of the boundary conditions.

Thus, in the ground state of electron gas, the average energy per electron is
Etotal /N =

3
h2 kF2
3
= EF
10m
5

2. Griffiths 5.16
a) From the previous problem, we can find the Fermi momentum kF in terms of number density n

of electrons. Since the number of free electrons in Cu is the same as the number of atoms, this is
the same as the number density of Cu atoms.
n = N/L3 = kF3 /3 2
Hence, we have kF = 3 2 n

1/3

, so

EF =

2 2
h
k = 1.13 1018 J = 7.05eV
2m F

b) The corresponding velocity can be estimated by vF =

2EF /m = 1.57 106 m/s .

This is roughly 0.005 times the speed of light, so the non-relativistic approximation should give us
sensible results. We could see this in another way - the rest mass energy of the electron (mc2 ) is
around 511 keV, which is much larger than the maximum kinetic energy of a few electron volts. So,
we can treat the problem non-relativistically and still get sensible results.
c) The Fermi temperature is TF = EF /kB 81800K .
This is much larger than the room temperature, which has very important physical implications.
Thermal fluctuations can only excite electrons to states close to the Fermi surface, and the density
of states at the Fermi surface therefore determines a lot of physical properties of metals, such as
specific heat or susceptibility to magnetic fields.
d) The degeneracy pressure of Cu in the electron gas model can be evaluated by using equation 5.46
in Griffiths.
(3 2 )2/3 h2 n5/3
P =
= 3.83 1010 N/m2
5m
Note: This is the pressure at zero temperature, since we have only used the ground state energy to
calculate it. But since room temperature is very low compared to the typical temperature scale of
Cu (which is TF ), the thermodynamic properties are determined mainly by the ground state, and
this is still a good approximation.
Griffiths

uses for number density

3. Griffiths 5.17
The (isothermal) bulk modulus is defined as the fractional change of volume, per unit increase of
pressure on a system, keeping the number of particles and temperature constant.
B = V

P
|N,T
V

Equation 5.46 in Griffiths gives us the pressure at zero temperature in terms of density, which we
can easily convert to a function of volume and number of particles.
P =

(3 2 )2/3 h2 n5/3
(3 2 )2/3 h2 N 5/3 5/3
=
V
5m
5m

Hence, the bulk modulus at zero temperature is given by



5 (3 2 )2/3 h2 N 5/3 8/3
5
B = V
V
= P
3
5m
3
Plugging in the value of P from the previous problem, we find that the bulk modulus of Cu is
B = 6.38 1010 N/m2

4. Griffiths 5.18
a) Following Griffiths 5.59 and 5.60, we can use Blochs theorem for a particle moving in a periodic
potential to say
(x) = A sin(kx) + B cos(kx), (0 < x < a)
(x + a) = eiKa (x) = (x) = eiKa [A sin(k(x + a)) + B cos(k(x + a))] , (a < x < 0)
Using continuity of the wavefunction at x=0, we have
B = eiKa [A sin(ka) + B cos(ka)] = A = B

eiKa cos(ka)
sin(ka)

We can use this to write


(x)

= B

eiKa cos(ka)
sin(ka)


sin(kx) + B cos(kx)


= C sin(kx) + eiKa ( cos(kx) sin(ka) sin(kx) cos(ka))


Here C is a constant in terms of B and ka. By using trigonometric addition formulae, the above
can be re-written as


(x) = C sin(kx) + eiKa sin(k(a x)) , 0 < x < a

b) From figure 5.6 in Griffiths, we can see that at the top of a band, one of the following two holds.
1. ka = (2n + 1) and cos(Ka) = 1
2. ka = 2n and cos(Ka) = 1

If we use either of these pair of conditions in equation 5.60, we see that Blochs theorem is automatically satisfied for arbitrary coefficients A and B. In other words, equation 5.60 reduces to
(x) = A sin(kx) + B cos(kx), a < x < 0
This is automatically continuous at x = 0, so imposing continuity at x = 0 does not give us any new
condition.
But even more than that, we actually have the same functional form for both x < 0 and x > 0. This
means that unlike the generic case of the delta function potential, the derivative of the wavefunction
cannot have any discontinuity across x = 0. This is only possible if the wavefunction is zero at x = 0
(so that it cannot feel the effect of the delta function at all). Hence, the form of the wavefunction
must be
(x) = A sin(kx)

I argued physically, but you can show that the same answer comes out of explicitly evaluating the discontinuity
of the derivative across x = 0