September 2, 2011 9:36 World Scientific Book - 9in x 6in han

Chapter 1

Drude Theory of Metals

(1) Four assumptions in the Drude theory

I A given conduction electron in a metal interacts with an ion only
when it makes a collision with the ion. Between two successive
collisions of the conduction electron, its interaction both with
other electrons and with ions is neglected.
II Collisions of conduction electrons with ions in the Drude model
are instantaneous events as in kinetic theory. The velocity of the
conduction electron that suffers the collision is abruptly changed.
III A conduction electron in a metal makes a collision with an ion
with probability 1/τ per unit time.
IV Local thermal equilibrium of conduction electrons with their en-
vironment is achieved only through their collisions with ions. Im-
mediately after a collision, the velocity of the conduction electron
is not related to its velocity before the collision, with the direc-
tion of the velocity random and the magnitude of the velocity
determined by the temperature at the collision place.

(2) Collision probability density (5) Drude AC conductivity

1 σ0
p(t) = e−t/τ . σ(ω) = .
τ 1 − iωτ
(3) Equation of motion (6) Wave equation for E in a metal
dp(t) p(t) ω 2 ǫr (ω)
=− + f (t). ∇ 2
E + E = 0.
dt τ c2
(4) Drude DC conductivity (7) Relative dielectric function
ne2 τ iσ
σ0 = . ǫr (ω) = 1 + .
m ǫ0 ω

1
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

2 Problems in Solid State Physics with Solutions

(8) Plasma frequency (11) Wiedemann-Franz law

2  2
ne κ 3 kB
ωp2 = . = T.
ǫ0 m σ 2 e
(9) Hall coefficient (12) Thermopower
1 cv kB
RH = − . Q=− =− .
ne 3ne 2e
(10) Thermal conductivity
1
κ = ℓvcv .
3

1-1 Electron speed distribution. An electron collided with an ion at

a place having a temperature T . Find the probability
p for the electron
having emerged with a speed greater than vm = 2kB T /m.

According to Assumption IV , the probability for the electron to emerge

with a speed falling within the interval between v and v + dv is given
by
 3/2
m 2
dP = fM (v)dv = 4π e−mv /2kB T v 2 dv,
2πkB T
where fM (v) is Maxwell’s speed distribution function. Thus,
 3/2 Z ∞ Z ∞
m −mv 2 /2kB T 2 4 2
Pv>vm = 4π e v dv = √ e−x x2 dx
2πkB T vm π 1
 Z ∞  Z ∞
2 2 ∞ 2 2 2 2
−x −x
= √ −xe + e dx = √ + √ e−x dx
π 1 1 πe π 1
2
= √ + erfc(1) ≈ 0.572,
πe
where erfc(x) is the complementary error function.
1-2 Average and standard deviation of the collision time interval.
For a randomly picked conduction electron in a metal at time t = 0,
the probability density function for the conduction electron to have a
collision with an ion at time t is given by
1
p(t) = e−t/τ .
τ
(1) Evaluate the average of the time interval for the electron to have
its next collision.
(2) Evaluate the average of the square of the time interval for the
electron to have its next collision.
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

Drude Theory of Metals 3

(3) Evaluate the standard deviation of the time interval for the elec-
tron to have its next collision.
(4) Taking the above standard deviation of the time interval as the
time uncertainty in the time-energy uncertainty relation in quan-
tum mechanics, estimate the energy uncertainty in quantum me-
chanics. Comment on the results of the energy uncertainty given
by classical and quantum theories.

(1) The average of the time interval for the electron to have its next
collision is given by
Z ∞
1 ∞
Z Z ∞
t= dt tp(t) = dt te−t/τ = τ dx xe−x = τ.
0 τ 0 0

(2) The average of the square of the time interval for the electron to
have its next collision is given by
Z ∞
1 ∞
Z Z ∞
2 2 −t/τ 2
2
t = dt t p(t) = dt t e =τ dx x2 e−x = 2τ 2 .
0 τ 0 0

(3) The standard deviation of the time interval for the electron to
have its next collision is then given by
1/2
∆t = t2 − (t)2 = (2τ 2 − τ 2 )1/2 = τ.


(4) From τ ∆Eq ∼ ~, we have ∆Eq ∼ ~/τ in quantum theory. In

classical theory, ∆Ec is given by ∆Ec = (3/2)1/2 kB T that can be
obtained by using Maxwell’s distribution. We make the following
comments on ∆Eq and ∆Ec .
• ∆Eq in quantum theory is intrinsic in nature and it is the
consequence of the wave property of microscopic particles,
while ∆Ec in classical theory is statistical in nature.
• While ∆Ec in classical theory depends on temperature, ∆Eq
in quantum theory does not.
• For τ ∼ 10−15 s−1 , ∆Eq ∼ 10−12 erg. At T = 273 K, ∆Ec ∼
10−14 erg. Thus, at not too high temperatures, ∆Ec < ∆Eq .
1-3 Two successive collisions. Assume that a conduction electron in a
metal experiences two successive collisions at times t1 and t2 , respec-
tively. Let T = t2 − t1 be the time interval between the two successive
collisions.
(1) Find the probability density function for T .
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

4 Problems in Solid State Physics with Solutions

(2) Evaluate the average T of the time interval between two successive
collisions.

(1) The probability density for the occurrence of the first collision
is given as usual by p(t1 ) = τ −1 e−t1 /τ . The probability den-
sity for the occurrence of the second collision is given by p(t2 ) =
τ −1 e−(t2 −t1 )/τ . Let g(T ) be the probability density function for
T . According to statistics, we have
Z ∞ Z ∞
g(T ) = dt1 dt2 δ(t2 − t1 − T )p(t1 )p(t2 )
0
Z ∞ 0 Z ∞
1
= 2 dt1 dt2 δ(t2 − t1 − T )e−t2 /τ
τ 0
Z 0
1 −T /τ ∞ 1
= 2e dt1 e−t1 /τ = e−T /τ .
τ 0 τ
(2) The average of T is given by
Z ∞
1 ∞
Z
T = dT T g(T ) = dT T e−T /τ = τ.
0 τ 0
1-4 Conductivity of a superconductor. Assume that the real part of
the conductivity of a superconductor is described by Re σ(ω) = Aδ(ω)
with δ(ω) the Dirac δ-function. This expression can be taken as the
τ → ∞ limit of the real part of σ(ω) = σ0 /(1−iωτ ) with σ0 = ne2 τ /m.
(1) Express A in terms of the electron density, mass, and charge.
R∞
(2) Evaluate the integral −∞ dω Re σ(ω).

(1) The real part of σ(ω) = σ0 /(1 − iωτ ) is given by

σ0 ne2 τ /m
Re σ(ω) = 2
= .
1 + (ωτ ) 1 + (ωτ )2
For the superconductor, we have
ne2 τ /m
Re σ(ω) = Aδ(ω) = lim
τ →∞ 1 + (ωτ )2

ne2 1/τ πne2

= lim 2 2
= δ(ω).
m τ →∞ ω + 1/τ m
Thus,
πne2
A= .
m
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

Drude Theory of Metals 5

(2) Making use of the property of the δ-function, we have

Z ∞
πne2 ∞ πne2
Z
dω Re σ(ω) = dω δ(ω) = .
−∞ m −∞ m

1-5 Relative dielectric function of a metal. In the Drude theory of

metals, the complex relative dielectric function of a metal is given by
iσ(ω)
ǫr (ω) = 1 + ,
ǫ0 ω
where σ(ω) is the optical conductivity (the AC electrical conductivity)
σ0 ne2 τ
σ(ω) = , σ0 = .
1 − iωτ m
(1) Find the real and imaginary parts, Re σ(ω) and Im σ(ω), of σ(ω).
Evaluate the integral
1 ∞ dω
Z
Im σ(ω).
π −∞ ω
(2) Find the real and imaginary parts, Re ǫr (ω) and Im ǫr (ω), of ǫr (ω).
(3) For σ0 τ /ǫ0 = 2, plot Re ǫr (ω) and Im
p ǫr (ω) as functions of ωτ .
(4) Evaluate the square root of ǫr (ω), ǫr (ω).
(5) The index of refraction n(ω) and the extinctionp coefficient k(ω)
are given byp p of ǫr (ω), respectively,
the real and imaginary parts
n(ω) = Re ǫr (ω) and k(ω) = Im ǫr (ω). Find the explicit
expressions for n(ω) and k(ω) in terms of the real and imaginary
parts of ǫr (ω). For simplicity in notations, use ǫ′r for Re ǫr and ǫ′′r
for Im ǫr .

(1) Separating the real and imaginary parts of σ(ω), we have

σ0 (1 + iωτ ) σ0 σ0 ωτ
σ(ω) = = +i .
1 + (ωτ )2 1 + (ωτ )2 1 + (ωτ )2
Thus,
σ0 σ0 ωτ
Re σ(ω) = 2
, Im σ(ω) = .
1 + (ωτ ) 1 + (ωτ )2
The value of the concerned integral is given by
1 ∞ dω σ0 τ ∞ dω
Z Z
Im σ(ω) = = σ0 .
π −∞ ω π −∞ 1 + (ωτ )2
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

6 Problems in Solid State Physics with Solutions

(2) Separating the real and imaginary parts of ǫr (ω), we have

 
iσ(ω) i σ0 σ0 ωτ
ǫr (ω) = 1 + =1+ + i
ǫ0 ω ǫ0 ω 1 + (ωτ )2 1 + (ωτ )2
σ0 τ /ǫ0 σ0 /ǫ0
=1− 2
+i .
1 + (ωτ ) ω[ 1 + (ωτ )2 ]
Thus,
σ0 τ /ǫ0 σ0 /ǫ0
Re ǫr (ω) = 1 − , Im ǫr (ω) = .
1 + (ωτ )2 ω[ 1 + (ωτ )2 ]
(3) For σ0 τ /ǫ0 = 2, we have
2 2
Re ǫr (ω) = 1 − 2
, Im ǫr (ω) = .
1 + (ωτ ) (ωτ )[ 1 + (ωτ )2 ]
The plots of Re ǫr (ω) and Im ǫr (ω) are given in Fig. 1.1.

1 5
(a) (b)
Re ǫr (ω )

Im ǫr (ω )

0 0

-1
-5
-2 -1 0 1 2 -2 -1 0 1 2
ωτ ωτ

Fig. 1.1 Plots of the real (a) and imaginary (b) parts of the relative dielectric function
ǫr (ω) as functions of ωτ .

(4) Let ǫ′r (ω) = Re ǫr (ω) and ǫ′′r (ω) = Im ǫr (ω). We then have
2 2
1/2 i tan−1 (ǫ′′ ′
ǫr = ǫ′r + iǫ′′r = ǫ′r + ǫ′′r e r /ǫr ) .

Taking the square root of ǫr , we have

p 2 2
1/4 i tan−1 (ǫ′′ ′
ǫr (ω) = ǫ′r + ǫ′′r e r /ǫr )/2

2 2
1/4 2 2
1/4
= ǫ′r + ǫ′′r cos tan−1 (ǫ′′r /ǫ′r )/2 + i ǫ′r + ǫ′′r
 

× sin tan−1 (ǫ′′r /ǫ′r )/2 .

 

Doing some algebras, we find that

1  1/2
cos tan−1 (ǫ′′r /ǫ′r )/2 = 1/2 1 + [1 + (ǫ′′r /ǫ′r )2 ]−1/2
 
,
2
1  1/2
sin tan−1 (ǫ′′r /ǫ′r )/2 = 1/2 1 − [1 + (ǫ′′r /ǫ′r )2 ]−1/2
 
.
2
We then have
p 1  2 2
1/2 1/2 i  2 2
1/2 1/2
ǫr (ω) = 1/2 ǫ′r +ǫ′′r +ǫ′r + 1/2 ǫ′r +ǫ′′r −ǫ′r .
2 2
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

Drude Theory of Metals 7

p
(5) From the above result for ǫr (ω), we obtain the following index
of refraction n(ω) and the extinction coefficient k(ω)
1  2 2
1/2 1/2
n(ω) = 1/2 ǫ′r + ǫ′′r + ǫ′r ,
2
1  2 2
1/2 1/2
k(ω) = 1/2 ǫ′r + ǫ′′r − ǫ′r .
2
1-6 Propagation of electromagnetic radiation in a metal. A beam
of electromagnetic radiation of high frequency ω0 is shed perpendicu-
larly on a flat surface of a metal. It is found experimentally that the
radiation can only penetrate a very small distance into the metal, with
its strength dropping to 1/e at a small distance λ0 . Using the Drude
model, determine
(1) the plasma frequency of the metal;
(2) the electron density of the metal;
(3) the expression of λ at an arbitrary frequency ω within a narrow
frequency window below ω0 .

(1) The equation for the electric field E inside the metal reads
d2 E ω2
2
+ 2 ǫr (ω)E = 0,
dx c
where ǫr (ω) = 1 − ωp2 /ω 2 at high frequencies. Since the radiation
can only penetrate a very small distance into the metal, ǫr (ω)
must be negative, ǫr (ω) < 0. Then the solution to the above
equation decays exponentially as x increases
1/2
E(x) = E(0)e−[−ǫr (ω)] ωx/c
.
From the fact that the strength of the radiation of frequency ω0
drops to 1/e at a small distance λ0 , we have
1/2 1
e−[−ǫr (ω0 )] ω0 λ0 /c = .
e
Solving for ǫr (ω0 ), we obtain ǫr (ω0 ) = −(c/ω0 λ0 )2 . From ǫr (ω0 ) =
1 − ωp2 /ω02 , we finally obtain
  2 1/2
c
ωp = ω0 1 + .
ω0 λ0
(2) From ωp2 = ne2 /ǫ0 m, we obtain the electron density of the metal
2 
ǫ0 mω02
 
c
n= 1 + .
e2 ω0 λ0
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

8 Problems in Solid State Physics with Solutions

(3) To find the expression of λ at an arbitrary frequency ω, we replace

λ0 in ωp with λ and ω0 with ω. We then have
  2 1/2
c
ωp = ω 1 + .
ωλ
Since ωp is an intrinsic property of the metal, it is independent of
ω. We thus have dωp2 /dω = 0 from which we obtain
 2 
dωp2 2c2 2c2 dλ
  
c dλ
0= = 2ω 1 + − λ + ω = 2ω −
dω ωλ ωλ3 dω λ3 dω
which leads to
 
1 1
d 2 = − 2 dω 2 .
λ c
Integrating yields
1 1 1
− 2 = 2 ω02 − ω 2 .


λ2 λ0 c
Solving for λ, we obtain
λ0
λ=  1/2 .
1 + (ω02 − ω 2 )λ20 /c2

1-7 Thermal conduction of a one-dimensional metal. The constant

temperature gradient dT /dx = −γ < 0 exists in a metal bar whose
left end is maintained at temperature Th and is taken to be the origin
of the x axis.
(1) Compute the thermal energy difference per electron between x−vτ
and x + vτ .
(2) Evaluate the thermal current density at x.

(1) From dT /dx = −γ, we have dT = −γdx. Integrating yields

T = Th − γx. The thermal energy of an electron is given by
ε(x) = 3kB T /2 = 3kB (Th − γx)/2. The thermal energy difference
per electron between x − vτ and x + vτ is given by
3  
ε(x − vτ ) − ε(x + vτ ) = kB Th − γ(x − vτ )
2
3  
− kB Th − γ(x + vτ )
2
= 3γvτ kB = 3γℓkB .
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

Drude Theory of Metals 9

 
(2) From jq = (nv/2) ε(x − vτ ) − ε(x + vτ ) , we have
nv 3
jq = · 3γℓkB = nvγℓkB .
2 2
1-8 Metal in a uniform static electric field. A metal at uniform
temperature is placed in a uniform static electric field E. A conduction
electron in the metal suffers a collision with an ion, and then it suffers
a second collision with another ion after a time t.
(1) Find the average energy lost to the ion in the second collision.
(2) Find the average energy lost to the ions per electron per collision.
(3) Suppose that the temperature in the metal is not uniform with
a constant temperature gradient given by ∇T . Find the average
energy loss to the ions per electron per collision when both the
applied static electric field and the constant temperature gradient
are taken into account.

(1) The speed the electron acquires between two collisions is v =

eEt/m. Thus, its kinetic energy right before the second collision
is
1 1
Ek = mv 2 = (eEt)2 .
2 2m
The electron loses its kinetic energy to the ion in its second colli-
sion. Therefore, the energy loss is given by
1
Eloss = Ek = (eEt)2 .
2m
(2) To find the average energy lost to the ions per electron per col-
lision, we need to evaluate the average of t2 . Making use of the
probability density function for t, p(t) = τ −1 e−t/τ , we have
Z ∞
1 ∞
Z
t2 = dt t2 p(t) = dt t2 e−t/τ = 2τ 2 .
0 τ 0
Thus, the average energy loss to the ions per electron per collision
is given by
(eE)2 t2 (eEτ )2
E loss = = .
2m m
(3) Assume that the second collision occurred at r. Then, the first
collision occurred at r + eEt2 /2m, where −eEt2 /2m is the dis-
placement of the electron between two collisions in the uniform
electric field E. In the presence of the temperature gradient, the
September 2, 2011 9:36 World Scientific Book - 9in x 6in han

10 Problems in Solid State Physics with Solutions

energy loss of the electron includes the thermal energy difference

between the two collision spots in addition to its kinetic energy.
The temperatures at the two collision spots are respectively given
by T (r + eEt2 /2m) and T (r). The energy loss of the electron is
then given by
et2 (eEt)2
  
Eloss = ε T r + E − ε[ T (r) ] +
2m 2m
2
(eEt)2
 
et
≈ ε T (r) + ∇T · E − ε[ T (r) ] +
2m 2m
2
et dε (eEt)2
≈ ε[ T (r) ] + ∇T · E − ε[ T (r) ] +
2m dT 2m
et2 dε (eEt)2
= ∇T · E + .
2m dT 2m
Making use of t2 = 2τ 2 , we obtain the average energy lost to the
ions per electron per collision
eτ 2 dε (eEτ )2
E loss = ∇T · E + .
m dT m