Homework 2: Solutions

1. Problem. We have approximated the conduction band of many semiconductors with an isotropic, ‘parabolic’
expression of the form:
h̄2 k2
E(k) ≈ . (1)
2m∗
However, this approximation fails at some large k. A first correction to this expression consists in adding also
terms of order k4 , so that the dispersion is usually written as:


h̄2 k2 h̄2 k2
E(k) ≈ 1 − α , (2)
2m∗ 2m∗

where α (of dimension of inverse energy) is called the ‘nonparabolicity parameter’.

a. Using Eq. (2) derive the density of states of the conducton band only to first-order in the parameter α.
b. Also to first-order in α, find the group velocity ∇E(k)/h̄ as a function of k. (Note that, like the energy,
the group velocity depends only on the magnitude k of k and not on its direction.)
Solution.
Let’s first derive a few useful relations: From Eq. (2), to first order in α:

h̄2 k2
≈ E (1 + αE) , (3)
2m∗
so that
[2m∗E(1 + αE)]1/2 (2m∗E)1/2 α
k ≈ ≈ (1 + E) . (4)
h̄ h̄ 2
From Eq. (3), differentiating:
m∗
kdk ≈ (1 + 2αE) dE . (5)
h̄2

ECE609 Spring 2010 1

 a. Then:
  ∞
2 1 2
ρ(E) = d k δ[E( k ) − E] = dkk δ[E(k) − E] ≈
(2π)3 π2 0
 ∞    
(2m∗)1/2 m∗   α   21/2 m∗3/2 1/2 5
dE (1 + 2αE ) 1 + E δ(E − E) ≈ E 1 + αE (6)
.
π2 h̄3 0 2 π2 h̄3 2

b. From Eq. (5), to first-order in α:

1 dE h̄k
v(k) = ≈ (1 − 2αE) . (7)
h̄ dk m∗

This is the ‘usual’ parabolic velocity h̄k/m∗ but corrected by the factor (1 − 2αE). In Si, α = 0.5 eV−1 , so
the velocity is reduced with respect to its ‘parabolic’ expression.
2. Problem. For a parabolic conduction band with effective mass m∗ , derive EF as a function of electron density
n at zero temperature.
Hint: At T = 0 the electrons in k-space fill a sphere of radius kF , the Fermi wavevector. Thus, the electron density will be the
volume of this sphere divided by the density of states in k-space.
Solution. The electron density is, by definition:

2  2
n = fF D (k) = dk fF D (k) . (8)
V (2π)3
k

But at T = 0 fF D vanishes for E(k) > EF , that is for k > kF , where kF = (2m∗EF )1/2 /h̄. Also,
FF D (k) = 1 for k ≤ kF . So 
4 3
dk fF D (k) = πkF , (9)
3

ECE609 Spring 2010 2

 (the volume of a sphere of radius kF in k-space) and

3
2 4 3 kF (2m∗)3/2 1/3
n = πk = = 3
EF , (10)
(2π)3 3 F 3π2 2
3π h̄
so
32/3 π4/3 h̄2 2/3
EF = n . (11)
2m∗
So the Fermi level is always above the bottom of the conduction band at T = 0.
3. Problem. In a band diagram, plot the position of the Fermi level for Si at 300 K as a function of donor
concentration, ND . The plot may be qualitative, but it should be qualitatively correct. Try to be as accurate
as possible.
Solution. We have discussed in class the behavior in the non-degenerate limit. The lecture notes (page 40)
show n vs.EF , so that the requested plot can be obtained from that figure by ‘turning it on its side’, so to
speak...
4. Problem. Prove in detail that the screend Coulomb potential has the form (in MKS units)

e2 e−βr
V (r) = .
4πs r
Use Fourier transforms and the fact that the Fourier components of the screened Coulomb potential are:

e2 1
V (q) = .
s q 2 + β 2

Solution. We must take the inverse Fourier transform of the potential energy:
 
1 iq·r e2 eiq·r
V (r) = dq V (q) e = dq 2 . (12)
(2π)3 (2π)3 s q + β2

ECE609 Spring 2010 3

 Going to polar coordinates with the z axis aligned with r, integrating trivially over the azimuthal angle φ and
noticing – as usual – that q · r = qr cos θ , where θ is the polar angle, noticing also that V depends only on
the magnitude of r:
 ∞  π
e2 q2 iqr cos θ
V (r) = dq dθ sin θ e =
(2π)2 s 0 q2 + β 2 0
 ∞
e2 sin(qr) e2 −βr
2
dq q 2 2
= e , (13)
2π s r 0 q +β 4πs r

since, from any book on tables of integrals:

 ∞
x sin(ax) π −βa
2 + β2
dx = e .
0 x 2

5. Problem. Looking at the figure on the next page, consider a system of electrons uniformly distributed over a
square area with a uniform background of positive charge compensating the negative electron charge, so that at
(mechanical) equilibrium the system is charge-neutral. Let n be the density of the sheet of electron charge, let
m∗ be the mass of each electron, and let s be the dielectric constant of the medium surrounding the system.
Now shift the electrons – assumed to be all rigidly connected among themselves, so as to form a rigid sheet of
negative charge – by a displacement u to the left. This will leave a ‘rectangle’ of positive charge exposed at the
right hand side of the square. This positive charge will exhert a force which will attempt to ‘call back’ the sheet
of electrons to their equilibrium position. Assuming that the ‘square’ extends to infinity up and down along the
vertical direction (so that edge effects should be neglected and the expression for the electrostatic force is simply
that between a point-charge and an infinite sheet of charge):
a. Write the equation of motion for the displacement of the sheet of electrons and show that it will lead to
harmonic oscillations.
b. Show that the frequency of these oscillations is the ‘plasma frequency’ (in MKS units) ωp =
[e2 n/(m∗ s )]1/2 .

ECE609 Spring 2010 4

 Solution.
a. The positive charge density (at the right in the figure) is enu. The force acting on any electron in the
negative sheet of charge will be independent of the distance from the positive charge and will be directed
opposite to u, that is, F = − e2 nu/s . Thus, the equation of motion for the displacement u will be:

2
∗ d u(t) e2 nu(t)
m = − . (14)
dt2 s

This is the equation of a harmonic oscillator: The sheet of electrons will move to the left (or in whatever
direction it was intially pushed) until the recalling force will grow so large as to stop it and invert its motion.
The force will disppear when u will reach its zero value – since there will be no ‘exposed charges’ – but the
inertia will drive the sheet to the right and the cycle will repeat itself.
b. Equation (14) can be written as:
d2u(t) 2
+ ω u(t) = 0 , (15)
dt2 p

where ωp = [e2 n/(m∗ s )]1/2 . This is the frequency of the oscillations which is, in a sense, the proper
eigenfrequency of the electron gas.

6. Problem. Show in detail the steps required to derive the expression for 1/τop given at page 70 of the notes,
using the matrix element in the same page. As usual, polar coordinates in k-space are handy.
Solution. By the Fermi Golden Rule:

1 2π 
= | < k |Hop|k > |2 δ[E(k) − E(k) ± h̄ωop] , (16)
τop (k) h̄  k

where the upper sign holds for emission and the lower sign for absorption of optical phonons. Using the

ECE609 Spring 2010 5

expression for the matrix element given at page 70 of the lecture notes,

2  
1 2π h̄(Dt K)op 1 1 1 
= nop + ± δ[E(k) − E(k) ± h̄ωop ] , (17)
τop (k) h̄ 2ρωop 2 2 V 
k

where, again,
 the upper signs hold for emission and the lower signs for absorption of optical phonons. The term
(1/V ) k δ(E  − E ± h̄ωop ) is just one half the density of states at energy E ∓ h̄ωop . Thus:

 
1 π(Dt K)2op 1 1
= nop + ± ρ[E(k) ∓ h̄ωop ] . (18)
τop (k) 2ρωop 2 2

Using the expression for the density of states, we finally get:

 
1 (Dt K)2opm∗3/2 1 1
= nop + ∓ [E(k) ∓ h̄ωop]1/2 . (19)
τop (k) 21/2 πh̄3 ρωop 2 2

ECE609 Spring 2010 6

ECE609 Spring 2010 7
7. Problem. Looking at the figure above, consider a linear chain of ions connected by springs with elastic constant
α. The ions have alternating masses, M and m, as shown, with M > m. The equibrium distance between
two adjacent atoms is a.
a. Write the equations of motion. Hint: Write two equations, one for the displacement u1 of the atoms of mass M and
another one for the displacement u2 of the atoms of mass m.
b. Find the frequencies of the possible modes, assuming that the displacement of each atom is of the form
∼ ei(2jqa−ωt) , where j is the index labeling the atom position, and q the wavevector of the mode. (Note the
use of 2a instead of a since the ‘lattice constant’ for this 1D lattice is indeed 2a.)
c. Sketch as accurately as you can ω vs. q for the possible modes and discuss their physical meaning.
Solution. Denoting by u1,j the displacement of the ion of mass M at location ja and by u2,j the displacement
of the ion of mass m at location (j + 1)a, than the equations of motion are:

d2 u1,j
M = α(u2,j − u1,j ) + α(u2,j−1 − u1,j ) . (20)
dt2

ECE609 Spring 2010 8

 d2 u2,j
m = α(u1,j+1 − u2,j ) + α(u1,j − u2,j ) (21)
dt2
Let’s define:
i(2jqa−ωt)
u1,j = 1 e , (22)
i(2jqa−ωt)
u2,j = 2 e . (23)
Then the equations of motion become:

−M ω 2 1 = α2 − α1 + α2 e−2iqa − α1 , (24)

−m ω 2 2 = α1 e2iqa − α2 + α1 − α2 , (25)

or:
(M ω 2 − 2α)1 + α(1 + e−2iqa)2 = 0 , (26)
2iqa 2
α(1 + e )1 + (M ω − 2α)2 = 0 . (27)
This 2 × 2 linear homogeneous system admits nontrivial solutions only if the determinant of the coefficients
vanish, that is:
2 2 2 2iqa −2iqa
(M ω − 2α)(M ω − 2α) − α (1 + e )(1 + e ) = 0. (28)

The two solutions are:


1/2 
2 α
ω± = (m + M ) ± (m + M )2 − 4mM sin2 (2qa) . (29)
mM

The choice of the ‘plus’ sign leads to an ‘optical’ branch:

 
1
2α m + M 1 for q → 0
2
ω+ →
2α π .
for q → 2a
m

ECE609 Spring 2010 9

The choice of the ‘minus’ sign yields an acoustic mode:

2 0 for q → 0
ω− → 2α π .
M for q → 2a

The optical solution corresponds to a mode in which the two different ions oscillate out of phase, the acoustic
solution to a mode in which the different ions oscillate in phase.

ECE609 Spring 2010 10