Solid State Physics

(2005) H.A.
Atwater, Caltech
1
APh 114b Lecture 1
Solid State Physics
Electrons in Solids:
-Free Electron Gas - Electrical Conductivity
-Bloch's Theorem - Electron waves in Solids
Energy Band Structure:
-Nearly-Free Electron Approximation
-Tight-Binding Approximation
-kp Theory; Density Functional Theory
Transport Properties of Solids:
-Effective Mass Theorem
-Boltzmann Transport Equation
-Coherent Transport
-Semiconductors
- Equilibrium Properties, doping
- Transport in Homogeneous Semiconductors
- pn Junction
Optical Properties of Solids
-Dielectric Response
- Connection to Transport Properties
- Light Emission and Absorption
APh114b Winter Term Topics
(2005) H.A. Atwater, Caltech
2
APh 114b Lecture 1
Solid State Physics
Adiabatic Approximation:
-assume that electronic relaxation so rapid that we can consider the
atoms to be at rest, for purposes of band structure and transport
calculation
One Electron Approximation:
-solve for energy levels and dynamics of a single electron moving in
an effective potential. This neglects electron-electron interactions,
which can be treated perturbatively
Periodic Potential: we assume at first that electrons move in a perfectly
periodic potential. Later, we will examine the consequences for
energy level calculations and transport of departures from
periodicity.
Pauli Exclusion Principle: as we fill available electron states with
electrons, we do so according to the Pauli Exclusion Principle: no
two electrons with identical spatial and spin quantum numbers may
occupy a given quantum state.
A Quantum Mechanical Approach, but Quasi-Classical when Possible
Underlying Assumptions of Initial Analysis:
3
APh 114b Lecture 1
Solid State Physics
Quantum Mechanical Approach
Basis for quantum mechanical analysis of electron states, subject to
the underlying assumptions, is the time-independent Schrodinger
equation (TISE):
2
2m
Vr r Er
p
2
2m
Vr r Er
where p i
4
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
While in general we will assume that V(r) is an effective periodic
potential, we start by analysis of the truly free electron gas, where V=0
inside the solid. Even though electrons are free within the solid of
length L, their motion is constrained by the size of the solid:
V
0 for 0 x, y, z L
otherwise
The time-independent Schrodinger equation is then

2
2m
2
r Er
subject to boundary condition
x 0 x L 0
y 0 y L 0
z 0 z L 0
Suggested Reading:
Ziman: pp. 77-78
A&M: pp 30-49
Kittel: pp 141-156
Marder: pp 135-149
5
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
Normalization of the electronic wave function
where the boundary conditions dictate that
solid
d
3
r
rr 1
r
2
L
3/2
sink
x
x sink
y
ysink
z
z
gives
k
x

L
n
x
, k
y

L
n
y
, k
z

L
n
z
; n
x
, n
y
, n
z
1, 2, 3. . .
The number of states less than an energy E can be found by counting
the number of states n
x
, n
y
, n
z
up to E. This can be done by noting that
these state fill the positive octant of a sphere of radius k
where the volume per state is
E

2
k
2
2m
V
k

L
3
6
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
We treat the number of states as a quasi-continuous function:
including spin:
dN
1
8
4k
2 1
/L
3
dk
dE

2
k
m
dk
dN
2m
3/2
4
2
3
E
1/2
dE
dN
2m
3/2
2
2
3
E
1/2
dE
The density of states for a free electron gas is
DE
dN
dE

2m
3/2
2
2
3
E
1/2
7
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
Free Electron Gas at T=0 K:
For classical particles, the temperature-dependent occupation
probability is the Maxwell-Boltzmann Factor):
The electrons must occupy energy states so that their total energy
corresponds to the mean thermal energy. This means that there is a
temperature-dependent occupation probability f(T,E) that governs state
filling. The electron density per unit volume is
n

0
DE fT, E dE
fT, E e
E/kT
Quantum mechanical particles must obey the Pauli exclusion
principle. At T=0 K, states up to the characteristic energy
separating occupied and unoccupied states (called the Fermi
energy) are completely filled. Above the Fermi energy, states are
all empty.
8
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
Fermi Energy How big is E
F
? Let's note that n ~ 510/cm. This
sets E
F
~ 5eV . Note also that the Fermi temperature T
F
~ 50,000K(!!)
which is ~100 the melting point of solids. We can also define a
Fermi velocity for an electron
where
Values for v
F
~10
8
cm/s =0.01c.
v
F

k
F
m
k
F

2mE
F
2
9
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
Because of the Pauli exclusion principle, a Fermi gas, unlike a
classical gas, has non-zero internal energy at T=0 K . The internal
energy density is the average value over all states, so
U

0
E
F
DE EdE
3
5
nE
F
T>0K : Fermi Statistics

We now derive the distribution f(T,E) using thermodynamics. The
energy levels of the solid E
i
can have a degeneracy g
i
and occupation
number n
i
. The Pauli Exclusion Principle dictates that n
i
g
i
. In
thermodynamic equilibrium, the free energy F is invariant with respect
to relative occupation numbers of the states
F

i
F
n
i
n
i
0 (equilibrium)
i
n
i
0 (particle conservation)
10
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
Fermi Statistics (contd)
Consider exchanging electrons between levels k and l:
dF
dn
k
n
k

F
n
l
n
l
0
n
k
n
l
0
Thus
F
n
k
F
n
l
(chemical potential of electrons)
The free energy is F=U-TS, where
and entropy is S=k ln w. The quantity w is the possible number of ways
of distributing the electrons among the states. For one level E
i
, the
number of ways to put one electron is g
i
; the second electron has g
i-1
ways, and so on.
U

i
n
i
E
i
11
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
This is
g
i
g
i1
g
i2
. . . g
i
n
i
1
g
i
!
g
i
n
i
!
Quantum mechanical particles are not distinguishable from each other
upon exchange, so to avoid overcounting the number of distinguishable
states, we must divide by 1/n
i
!. For the whole solid, the number of ways
w is the product of possibilities for each energy level E
i
,
w

i
g
i
!
n
i
!g
i
n
i
!
and then
S k
i
lng
i
! lnn
i
! lng
i
n
i
!
n.b., Stirling's Approximation:
lnn
i
! nlnn n (for large n)
12
APh 114b Lecture 1
Solid State Physics
Free Electron Gas
Thus

F
n
i
E
i
kTln
n
i
g
i
n
i
n
i
g
i
1
e
E
i
/kT
1
and rearranging to obtain
and rearranging to obtain the Fermi function (temperature dependent
occupation probability for fermions
fE, T
1
e
E/kT
1
The Fermi function:
E
F
=
f(E)
E
0.5
1
13
APh 114b Lecture 1
Solid State Physics
Heat Capacity of Free Electron Gas
Why is the heat capacity of a electron gas so small? We would expect
naively that in a metal the specific heat would have lattice and electronic
contributions that are about equal. Thus we might expect
c
v,electronic
(3/2)nk
where n is the electron density and n N, the lattice atomic density. The
actual electronic heat capacity is only about 1% of this value! Let's find out
why:
Internal energy is changed upon heating from 0K to T:
UT

0
dEEDE fE
0
E
F
dEEDE
n

0
dEDE fE
The particle number is
E
F
n E
F
dEDE fE
14
APh 114b Lecture 1
Solid State Physics
Differentiate the U(T) and E
F
n(T) by T:
C
el

U
T

0
dEEDE
f
T
0
N
T
E
F

0
dEE
F
DE
f
T
The electronic heat capacity is
C
el

0
dEE E
F
DE
f
T
DE
dEE E
F
f
T
Since E
F
=(T=0), for low and moderate temperatures
f
T

E E
F
kT
2
e
EE
F
/kT
e
EE
F
/kT1
2
15
APh 114b Lecture 1
Solid State Physics
Specific Heat Capacity for a Fermi Gas:
C
v el

1
3
2
DE
F
k
2
T
where, n.b.,
dx x
2
e
x
e
x
1
2

2
3
Note that since in metals up to their melting points, T<< T
F
, so
n

0
DE fE dE

0
E
F
DE dE
We can write the density of states as
DE DE
F
E
E
F
1/2
so
n
2
3
DE
F
E
F
16
APh 114b Lecture 1
Solid State Physics
and C
vel

2
2
nk
kT
E
F

2
2
nk
T
T
F
Note that (T/(T
F
)<0.01, so this explains the low electronic heat
capacity! Physically, we can picture the electron density at finite
temperature as:
Thus as the temperature is changed, the only electrons that can change
their internal energy are those excited from ~2kT below E
F
to unoccupied
states ~2kT above E
F
. This follows from the Pauli Exclusion Principle,
which forbids transitions from filled states to other filled states.
1
APh 114b Lecture 2
Solid State Physics
Electrostatic Screening in a Fermi Gas
Suppose that an additional electron were "trapped" at some specific location
in a solid (i.e., a charged defect) within an otherwise free electron Fermi
gas. This would induce a change in the otherwise homogeneous electron
density that compensates the field of the trapped charge. A trapped charge
locally raises the density of states curve by an energy e, where is the
electrostatic potential.
If e << E
F
, the electron concentration fraction that is transferred to the
surrounding to produce screening is
onr DE
F
|e| r
2
APh 114b Lecture 2
Solid State Physics
Electrostatic Screening in a Fermi Gas
The Poisson equation gives
2
r
e
E
0
on
e
2
c
0
DE
F
r
z e
2
DE
F
/c
0

1/2
If we define
1
r
2
r
r
2
r
Noting that
in spherical coordinates. The Poisson equation has a solution
r
oe
zr
r
, o
e
4mc
0
z e
2
DE
F
/c
0

1/2
So we can identify
As a screening length in the Fermi gas.
3
APh 114b Lecture 2
Solid State Physics
Last time, we found that the reason for the surprisingly small heat capacity of an
electron gas is that the electron gas is a Fermi gas, and so the Pauli exclusion
principle must be obeyed, leading to a state occupancy factor given by the Fermi-
Dirac distribution function. Because only electrons near the Fermi energy can
change their internal energy with a temperature change, the heat capacity of an
electron gas was found to be 0.01 of the classical estimate for heat capacity.
We might as well ask: "How does assumption of a Fermi gas and the Fermi-Dirac
distribution function affect the electrical and thermal conductivity, relative to
classical estimates?" In many regards, the classical Drude Model of electrical
conductivity is very successful -- but why?
Drude Model for Electrical Conductivity (1900)
Key elements of the Drude model:
1.Fixed ion cores
2."sea" of free electrons which are classical particles
3.electrons collide with ion cores, and relax at rate 1/, which nesters
electrons toward thermal equilibrium

4
APh 114b Lecture 2
Solid State Physics
j oE
j nev
v jE
j et/m
o
ne
2
t
m
Ohm's Law:
Continuity:
Definition of Mobility:
Boltzmann Transport Equation
(in relaxation time approximation):
Drude Model for Electrical Conductivity
5
APh 114b Lecture 2
Solid State Physics
Hall Effect - Magnetoresistance
The Hall effect arises from
electron dynamics in
magnetic field:
dp
dt
e E
p
mc
H
p
t
At steady state:
dp
z
dt
0
dp
x
dt
0 eE
x
o
c
p
y

p
x
t
dp
y
dt
0 eE
y
o
c
p
x

p
y
t
o
c

eH
mc
I
+ -
V
V
current source
resistivity
Hall voltage
B
x
y
z
d
Where the cyclotron frequency is:
6
APh 114b Lecture 2
Solid State Physics
Hall Effect - Magnetoresistance
where:
I
+ -
V
V
current source
resistivity
Hall voltage
B
x
y
z
d
The Hall coefficient is:
multiply by net/m:
o
0
E
x
o
c
tj
y
j
x
o
0
E
y
o
c
tj
x
j
y
o
0

ne
2
t
m
E
y

o
c
t
o
0
j
x

H
nec
j
x
R
H

1
nec
The sign of Hall coefficient tells us the carrier type (electrons or holes)
and its magnitude gives the carrier density
7
APh 114b Lecture 2
Solid State Physics
Two Dimensional Electron Gases
Two dimension electron gases (2 DEGs) exist commonly in heterostructures
and field-effect devices such Si MOSFETs and GaAs/AlGaAs HEMTs
A triangular quantum well is form at the GaAs/AlGaAs interface (HEMT) or
Si/SiO
2
interface (MOSFET).
8
APh 114b Lecture 2
Solid State Physics

2
2m
2
z
2
eEz _
n
z
n
r_
n
z
Where E(z) is the transverse electric field.
Boundary conditions are:
_
n
0 0
_
n
0
Time Independent Schrodinger Equation for triangular quantum well:
_
n
z Ai
2m
2
e
2
E
2
1/3
eEz
n
Solutions are
n

e
2
E
2
2
2m
1/3
a
n

2
2m
1/3
3emE
2
n
3
4

2/3
the Airy functions with energy eigenvalues
9
APh 114b Lecture 2
Solid State Physics
For a two-dimensional electron gas (2DEG) confined in the z-
direction:
Now with an applied magnetic field in z-direction
where n = 1,2,and j = 1,2,. This indicates that with an applied
magnetic field along the z-direction, discrete energy levels emerge,
called Landau levels. Since the number of states remains the same
when applying the magnetic field, it can be shown that each
Landau level contains N
L
= eB/h states, so N
L
is the degeneracy of
the Landau level.
direction - z for numbers quantum
,
,... 2 , 1
2 2
2
*
2
2
*
2
=
+ + =
j
k
m
k
m
E
j y x j

E
j,n

j
n
1
2
o
1
2
gj
B
B
10
APh 114b Lecture 2
Solid State Physics
The 2DEG behaves like an electrical insulator (
xx
=
yy
= 0)
until, with increasing magnetic field, the Landau levels increase
in energy. As successive Landau levels cross the Fermi energy,
they are emptied and excess electrons are transferred to lower
Landau levels. During the crossing, a partially filled band exists
and oscillations in
xx
result.
If a Landau level lies completely below the Fermi level, it is
completely filled with N
L
electrons if and the 2DEG
exhibits semiconductor behavior with effective bandgap
Shubnikov-de Haas Oscillations
T
o
k
E
g
o
c
11
APh 114b Lecture 2
Solid State Physics
Shubnikov-de Haas Oscillations in GaN/AlGaN
Longitudinal resistance, R
xx
, as a
function of B at 1.2 K showing the
presence of well defined Shubnikov-de
Haas oscillations. (The SdH oscillations
became visible only after the samples
were exposed to UV radiation from a
395 nm LED.) After S. Elhamri, et.al., J.
Appl. Phys. 95(12), 2004, pp. 7982-
7989.
12
APh 114b Lecture 2
Solid State Physics
The Hall coefficient R
H
= -1/nec is related to the off-diagonal
conductance matrix element
xy
. There are singular points in the
Hall resistance R
H
B=U
H
/I where the 2D electron density which carries
the current is a multiple of the degree of degeneracy N
L
of the
Landau levels:
,... 2 , 1
1
2
= = =
= =
v
v e
h
e vN
B
I
U
ne
B
B R
I
U
L
H
H
H
,
where is the filling factor
of the Landau levels. The
signature of the quantum
Hall effect is the existence
of plateaus in the Hall
resistance
xy
.
Quantum Hall Effect
13
APh 114b Lecture 2
Solid State Physics
Quantum Hall Effect
-1
= h/e
2
c = 1/137.036
First discovered by von Klitzing in Si
MOSFETs. New standard for resistance
(sample shape and size independent,
material independent; accurate 1 part in
10
8)
. A measure of the fine structure
constant:
14
APh 114b Lecture 2
Solid State Physics
Plateaus in the Hall resistance
Assumed to be localized states
between the Landau levels. These states
accept electrons once the Fermi level
crossed a Landau level. These electrons,
are not free to move since they are
localized. The Hall resistance does
therefore not change upon a further shift
of the Landau level until the next Landau
level is reached: a series of plateaus
results.
Quantum Hall Effect
15
APh 114b Lecture 2
Solid State Physics
Periodic Potentials
Bloch Functions
We now assume a potential V0. Due to crystalline order, electrons
move in a periodic potential with translational symmetry
Ur R Ur
where R a lattice vector. Now the time-independent Schrodinger equation is
2
2m
U
E
Bloch Theorem: for U(r+R)=U(r)
r R e
ikR
r
As a consequence of Bloch's Theorem, we can choose Bloch wave
functions of a form: (plane wave) x (Periodic function)
16
APh 114b Lecture 2
Solid State Physics
Periodic Potentials
Bloch Functions
nk
r e
ikR
U
nk
r
where
r R U
nk
r
Apply Born-Von Karman Boundary Conditions:
r R r N
i
a
i
r
where a = primitive lattice vector, i=1,2,3
Bloch's Theorem:
nk
r N
i
a
i
e
iN
i
ka
i
nk
r
17
APh 114b Lecture 2
Solid State Physics
Periodic Potentials
Bloch Functions
e
iN
i
ka
i
1
so
for
k x
1
b
1
x
2
b
2
x
3
b
3
where b
j
a
i
2md
ij
Thus
e
i2N
i
x
i
1
x
i

m
i
N
i
, m
i
1, 2, 3. . .
Volume of k space/state is
k
1
N
2m
3
V

1
N
2m
3 N
V
k
2m
3
V
In words, this result implies: the number of allow wave vectors (states)
in a primitive cell of the reciprocal lattice is equal to the number of sites
in the crystal.
1
APh 114b Lecture 3
Solid State Physics
Electron Group Velocity and Effective Mass
For free electrons, with plane wave eigenfunctions and energy
eigenstates the momentum, k, is a constant of the motion.
In a periodic potential, the actual momentum is not in general
translationally invariant, and is not a constant of the motion. In
periodic potentials it is the crystal momentum which is a constant of
the motion. If we picture an electron as a localized wavepacket, then
the group velocity of the wavepacket is
Now suppose that the electron is in an applied electric field E. Its change
in energy is
where energy is
E .
e
ikz
E

2
k
2
2m
dE
n

E
n
k
dk eEdx eEv
g
dt
eE
E
n
k
dt
so
dk
eE
dt
v
g

n
k
k

1
E
n
k
k
2
APh 114b Lecture 3
Solid State Physics
and
where we now use to denote the crystal momentum.
The derivative of the group velocity with time is
dv
g
dt

1

t
E
n
k
k

1

n
2
Ek
k
2
k
t
dv
g
dt

eE
2

n
2
E
n
k
2

eE
m
where m* is termed the effective mass; these can be rearranged so that

1
m
2

2
E
n
k
k
2
dp
dt

dk
dt
eE
3
APh 114b Lecture 3
Solid State Physics
Notice the striking distinctions between the velocity and mass of an electron in
the Drude model and the present result. According to Drude, the velocity of an
electron would be zero except in the presence of an applied field. Here we see
that for a state with a wavevector k, the electron velocity is a time-independent
quantity!! Moreover, a Drude electron always has a mass m
e
, whereas these
results imply that electrons move as though possessed of an effective mass
which depends entirely on the value which we will see can be
positive, negative or zero, depending on the form of E
n
(k). These relations can
be formalized using time-independent perturbation theory. We expand E
n
(k+q) in
q
2
E
n
k
k
2
E
n
k q E
n
k
i
E
n
k
i
q
i

1
2

i j
2
E
n
k
i
k
j
q
i
q
j
Oq
3

using a Hamiltonian H
kq
n
Ek q
n
H
kq
H
k

2
m
q i k

2
q
2
2m
4
APh 114b Lecture 3
Solid State Physics
Time-independent perturbation theory for gives
H H
0
H
E
n
E
n
E
n
0

n
d
3
r
n

n
n
d
3
r
2
E
n
0
E
n
0
Ignoring quadratic and higher order terms in q,
i
E
n
k
i
q
i

i
u
nk

2
m
i k
i
q
i
u
nk
d
3
r
so
E
n
k

2
m

u
nk
i k u
nk
d
3
r
Since Bloch's theorem gives
r e
ikr
u
nk
r
E
n
k

2
m

nk
i
nk
d
3
r
5
APh 114b Lecture 3
Solid State Physics
Since the electron group velocity is by definition
v
g

i
m
E
n
k
k
v
g
For the effective mass,
i j
1
2

n
2
E
k
i
k
j
q
i
q
j

2
q
2
2m

n
u
nk

2
m
q i k u
nk
d
3
r
2
E
nk
E
n
k
See the second-order terms now:
i j
1
2

n
2
E
k
i
k
j
q
i
q
j

2
q
2
2m

n
u
nk

i
2
m
q u
nk
d
3
r
2
E
nk
E
n
k
6
APh 114b Lecture 3
Solid State Physics
The Effective Mass Theorem is thus:
where
nk|i
i
| n
nk
nk
d
3
r
This tells us that the effective mass is a tensor quantity, whose values
depend on the form of E(k) in a particular direction in momentum space. We
shall see that the orientation dependence in k-space of E(k) can lead to
physically measurable transport properties, e.g., anisotropy in conductivity of
low symmetry materials.
2
E
n
k
i
k
j

2
m

ij

2
m
2
nk|i
i
|n
kn
k|i
j
|nk nk|i
j
|n
kn
k|i
i
|nk
E
n
k E
n
k
7
APh 114b Lecture 3
Solid State Physics
Nearly Free Electron Approximation
Our knowledge of the properties of periodic lattices now bears fruit in
understanding electronic structure. We describe solutions to the time-
independent Schrodinger equation for a weak periodic potential (TISE):
H

2
2m
2
V
E
where
Vr Vr R
Since V(r) is periodic in the real lattice, we can expand it in a Fourier series
of reciprocal lattice components in terms of reciprocal lattice vectors G:
Vr

G
V
G
e
iGr
Since periodic potential is weak, the electrons are nearly free, and we can
use a plane wave basis. The most general plane wave-like solution is
r

k
C
k
e
ikr
8
APh 114b Lecture 3
Solid State Physics
so substituting this solution into the TISE yields
2
k
2
2m
C
k
e
ikr
k,G
C
k
V
G
e
ikGr
E
k
C
k
e
ikr
Rearranging yields:
k
e
ikr

2
k
2
2m
E C
k

G
V
G
C
kG
0
Thus since r is unconstrained, is nonzero in general, so
e
ikr
2
k
2
2m
E C
k

G
V
G
C
kG
0
which is a set of algebraic equations for C
k
coupled to C
k-G
, that yield values for
E(k).
9
APh 114b Lecture 3
Solid State Physics
Note that
k
r

k
C
k
e
ikr
and also

k

G
C
kG
e
ikGr

G
C
kG
e
iGr
e
ikr
Thus

k
u
k
r e
ikr
meaning that the solutions to the TISE for a weak periodic potential are
Bloch wave functions. Lets look at some other properties of periodicity.
Write:
G
kG
r

G
C
kGG
e
iG
r
e
i
kG
kG

G
C
kG
e
iG
r
e
ikr

k

10
APh 114b Lecture 3
Solid State Physics
kG

k
So Bloch functions are also periodic in reciprocal space.
also from the TISE, lets look at the Hamiltonian
H
k
r Ek
k
r
H
kG
Ek G
kG
H
k
Ek G
k
Ek Ek G
So the energy band structure is also periodic in G.
11
APh 114b Lecture 3
Solid State Physics
Note that E(k) for free electrons has a degeneracy at the
Brillouin zone boundaries:

G
2

a
and
G
2

a
The Brillouin zone boundary electronic states must be a
superposition of states of the form (in 1 dimension)
e
i
G
2
x
and e
i
G
2
x
so there are two possibilities:
e
i
G
2
x
e
i
G
2
x
cos
x
a
e
i
G
2
x
e
i
G
2
x
sin
x
a
Electronic charge
localized on ion cores
Electronic charge
localized away from
ion cores
Heuristically, we can
see that ,
Giving rise to bandgaps
at the zone boundaries
E
These are standing waves

with the same wavevector
that are one-half period out
of phase. The electronic
probability densities are,
relative to the lattice are
+
and

12
APh 114b Lecture 3
Solid State Physics
To formalize this idea, note that the Schrodinger equation translated by G is:
E

2
2m
|k G|
2
C
kG

G
V
G
C
kGG
V
G
G
C
kG
or
C
kG

G
V
G
G
C
kG
E

2
2m
|
k G
|
2
If we set E=k/2m, then C
k-G
is only large when k
2
|k G|
2
Note this is the Bragg condition! so deviations from the free electron band
structure are anticipated to be large at the diffraction condition. Thus since we
only need to focus on the C
k
and C
k-G
components at the zone boundary, we
can write a set of two equations:
E

2
2m
k
2
C
k
V
G
C
kG
0
E

2
2m
|k G|
2
C
kG
V
G
C
k
0
13
APh 114b Lecture 3
Solid State Physics
Setting
det

2
k
2
2m
E V
G
V
G

2
2m
|k G|
2
E
0
yields E

1
2
E
kG
0
E
k
0

1
4
E
kG
0
E
k
0

2
|V
G
|
2
where
E
kG
0

2
2m
|k G|
2
and E
k
0

2
k
2
2m
Thus at the zone boundary, k=G/2
E E
E
_
|2V
G
|
and we get bandgaps!
14
APh 114b Lecture 3
Solid State Physics
Tight Binding Approximation
We now take a very different approach to energy band structure, relative to
that seen in the nearly free electron approximation. We recognize that the
low lying core levels of a free atom are strongly localized in space. When
free atoms are brought together, the core electronic levels remain strongly
localized, and thus we don't anticipate large changes in the energy
eigenstates as we go from atom to solid. Thus we might expect that the
electronic wavefunction in the solid can be expressed as a superposition of
atomic wavefunctions, or linear combination of atomic orbitals (LCAO) as it
is often termed. Let's begin by assuming that solutions for the atomic
eigenstates and eigenfunctions, i.e., the solutions to
H
at
r R
n

i
r R
n
E
i
i
r R
n

are known. We then assume that these atoms form a solid where the
atoms are arranged on a lattice with lattice vectors R
n
. The total potential
of this Hamiltonian, which is the Hamiltonian of the solid expressed as a
perturbed atomic Hamiltonian,
H
solid
H
at
H

2
2m
2
V
at
r R
n

H
r R
n

15
APh 114b Lecture 3
Solid State Physics
The perturbation H is in this case the difference between the atomic potential
and the crystalline solid potential. For an atom localized at R
n
,
H
r R
n

mn
V
at
r R
m

Of course, what we ultimately seek are solutions to the Schrodinger
equation in a crystalline solid which are Bloch waves,
k
(r), so that
H
solid
k
r Er
k
r
16
APh 114b Lecture 3
Solid State Physics
But multiplying by
k
*(r), and integrating yields E(k)
Ek

k
H
solid
k
d
3
r
k
d
3
r
The variational principle states that if one uses instead of the true
wavefunction
k
(r) a trial wavefunction
k
then one obtains an approximate
energy E(k) that is always bigger than E(k). In this case, we try as an
approximation solution a linear combination of atomic eigenfunctions
k

k

n
a
n
i
r R
n

where the coefficient a
n
are chosen so as to construct a Bloch wave:
a
n
e
ikR
n
We can see this by noting
kG

n
e
ikR
n
1
e
iGR
n
i
r R
n

k
17
APh 114b Lecture 3
Solid State Physics
Now let's find E(k)
E
k

k
H
solid
k
d
3
r
k
d
3
r
The denominator is:

k
k
d
3
r

n,m
e
ikR
n
R
m

i
r R
m

r R
n
d
3
r
We now make our first physical approximation: we note that
i
(r-R
n
) only has
significant value near R
m
; thus we only retain terms with n = m, so
k
d
3
r

n
r R
n
i
r R
n
d
3
r N
where N is the number of atoms in the solids. Since we already know
the atomic eigenstate E
i
, we write
E

1
N

n,m
e
ikR
n
R
m
r R
m

E
i
H
r R
n

r R
n

d
3
r
18
APh 114b Lecture 3
Solid State Physics
where in the sums on n,m to form H we sum over nearest neighbors for
simplicity, and as before, for the term in E
i
nearest neighbor overlap is
neglected. In a simple case where i is a single spherically symmetry s-
orbital, we can write down the nearest neighbor tight binding band
structure as
Ek E
i
E
io
E
inn
mnearest neighbors
e
ikr
n
r
m

where
E
io
r R
n

H
r R
n

r R
n

d
3
r
E
inn
r R
n

H
r R
n

r R
n

d
3
r
19
APh 114b Lecture 3
Solid State Physics
Note that E
io
> 0 since H(r-R
n
) < 0. Now let's specialize to a simple cubic
lattice with
r
n
r
m
a, 0, 0
0, a, 0
0, 0, a
For the simple cubic lattice, E(k) is then
E
E
i
E
io
2E
inn

cosk
x
a cosk
y
a cosk
z
a

1
APh 114b Lecture 4
Solid State Physics
Band Structures of Real Materials
Energy Band Structures in Real Solids
What can we learn from study of the energy band structure of solids? Many
things! Here are some:
1.Is the material a metal, semiconductor (direct or indirect energy gap) a
semimetal or an insulator?
2.What molecular or atomic energy levels give rise to the bands that are
important in transport and optical properties? (i.e., which bands are near
the Fermi level?
3.Are the carriers electrons or holes? Are there many or few carriers?
Where are the carriers in the Brillouin zone? Is the carrier mobility high or
low? These are issues related to transport.
4.Where in the Brillouin zone is the threshold for optical transitions? At
what energy does the optical threshold occur? Is the threshold a direct or
indirect gap?
2
APh 114b Lecture 4
Solid State Physics
Band Structures of Real Materials
3
APh 114b Lecture 4
Solid State Physics
Band structure of GaAs at 300 K
4
APh 114b Lecture 4
Solid State Physics
Band structure of Ge at 300 K
5
APh 114b Lecture 4
Solid State Physics
T-dependence of the relative populations of the
, L and X valleys in GaAs
Blakemore, J. S., J. Appl.
Phys. 53, 10 (1982) R123-
R181
6
APh 114b Lecture 4
Solid State Physics
Band structure of Diamond (C) at 300 K
7
APh 114b Lecture 4
Solid State Physics
Band structure of InP at 300 K
8
APh 114b Lecture 4
Solid State Physics
Band structure of Si at 300 K
9
APh 114b Lecture 4
Solid State Physics
Band structure of GaP at 300 K
10
APh 114b Lecture 4
Solid State Physics
Band structure of InAs at 300 K
11
APh 114b Lecture 4
Solid State Physics
Band structure of InSb at 300 K
12
APh 114b Lecture 4
Solid State Physics
Band structure of GaN (Zincblende) at 300 K
13
APh 114b Lecture 4
Solid State Physics
Band structure of GaN (Wurtzite) at 300 K
14
APh 114b Lecture 4
Solid State Physics
Band structure of GaN (Wurtzite) at 300 K
15
APh 114b Lecture 4
Solid State Physics
Band structure of AlN (Wurtzite) at 300 K
16
APh 114b Lecture 4
Solid State Physics
Band structure of Carbon Nanotubes at 300 K
Armchair Zigzag
17
APh 114b Lecture 4
Solid State Physics
Alkali Metal Band Structures
18
APh 114b Lecture 4
Solid State Physics
Noble Metal Energy Band Structure
19
APh 114b Lecture 4
Solid State Physics
Band Structure of Aluminum
20
APh 114b Lecture 4
Solid State Physics
Energy Band Structure of Graphite
Number of carrier << 10
22
/cm
3
Examples: As(4s
2
4p
3
), Sb(5s
2
5p
3
),
Bi(6s
2
6p
3
); graphite
Semimetals:
21
APh 114b Lecture 4
Solid State Physics
Energy Band Structure of -Sn
22
APh 114b Lecture 4
Solid State Physics
Energy Band Structure in Insulators: Ionic Crystals
Ionic insulator bands are very narrow, approaching atomic orbitals. The band
gap in KCl is ~ 9 eV. For alkali halides energy gaps range from 7-14 eV.
1
APh 114b Lecture 5
Solid State Physics
Effective Mass Theorem Revisited
Connection to Transport and Applications
Having studied the various approaches to determination of the energy band
structure E(k), we seek to treat perturbations of the perfect periodic structure in
solids, due, e.g., to defects or externally applied fields, in the context of
perturbations of the states E(k) for the perfect crystal. These perturbations
could be either time-independent or time-dependent in the most general case.
We thus start with the time-dependent Schrodinger equation,
H
0
H
n
r, t i

t
n
r, t
where H
o
Hamiltonian for perfect crystal, and H is the perturbation. With
r, t

d
3
kA
nk
e
ikr
u
nk

2
APh 114b Lecture 5
Solid State Physics
give the Bloch wave solutions to the perfect crystal. Thus:
where
H
0
e
ikr
u
nk
r E
n
ke
ikr
u
nk
r
H
0
H
n
r, t

d
3
kE
n
k H
A
nk
t e
ikr
u
nk
r, t ih

t

n
r, t
i
d
3
k A
nk
t e
ikr
u
nk
r, t
Note that E
n
(k) is a periodic function of the reciprocal lattice, so
E
n
k

R
l
E
nl
e
ikR
l
3
APh 114b Lecture 5
Solid State Physics
with R
l
as lattice vectors l=1,2,...N. Now consider the operator
formed by replacing k by ,
Ei
i
Ei

R
l
E
nl
e
R
l
Consider the effect of on an arbitrary function f(r):

E
n
i
e
R
l
fr 1 R
l

1
2!
R
l
R
l
. . . fr
fr R
l
fr
1
2!
R
l,
R
l,

2
fr
r
dr
. . . fr R
l

4
APh 114b Lecture 5
Solid State Physics
Thus for a Bloch state

nk
r
E
n
i
nk
r

R
l
E
nl
nk
r R
l

R
l
e
ikR
l
e
ikr
u
nk
r
E
n
k
nk
r
since using Blochs Theorem
nk
r R
l
e
ikR
l
e
ikr
u
nk
r
substituting
E
n
i
nk
r e
ikr
nk
r
5
APh 114b Lecture 5
Solid State Physics
into the Schrodinger equation yields:
d
3
k
E
n
A
nk
e
ikr
u
nk
r, t

E
n
d
3
k A
nk
e
ikr
u
nk
r, t
so another statement of the effective mass theorem is:

E
n
i H
n
r, t ih

t
n
r, t
Thus once E(k) is known, we can treat perturbations through knowledge of
the bandstructure without resorting to the crystal potential itself!! Suppose
that we consider a wavepacket localized in k-space around k=k
o
, then
nk
e
ikk
0
r
e
ik
0
r
u
nk
e
ikk
0
r
nk
0

6
APh 114b Lecture 5
Solid State Physics
where we assume u
nk
(r)u
nko
(r). The wavepacket is then:
r, t

d
3
k A
nk
e
ikk
0
r
nk
0
r, t
nk
0
where F(r,t) is an envelope function:

F
r, t

d
3
k A
nk
e
ikk
0
r
Since the Fourier coefficients A
nk
(t) are only larger near k=k
o
, F(r,t) is a
slowly varying function of r, since e
ikk
0
r
1 ik k
0
r.
where
k k
0
i.
7
APh 114b Lecture 5
Solid State Physics
Application of Effective Mass Theorem to Impurities in a Semiconductor
Let's consider addition of a column V atom to a group IV semiconductor
(e.g. P, As or Sb in Si or Ge). The extra electron experiences a Coulomb field
of the positive fixed donor impurity charge, and is screened by the other
electrons in the solid. Assume a band structure of the form:
8
APh 114b Lecture 5
Solid State Physics
Using the effective mass theorem,
This is justified as long as F(r) is slowly varying compared to the unit cell. The
perturbation is the donor atom's Coulomb potential:
is the static (=0) dielectric constant. Thus the effective mass theorem gives:
which is precisely the Schrodinger equation of the hydrogen atom under the
identification:

E E
c

e
2
r

2
2m
e
2
r
F

E E
c
Hydrogen atom Semiconductor Impurity

m m
e
2
e
2
/
E E E
c
me
4
/2l
2
2
m
e
4
/2l
2
2
, l 1, 2, 3. . .
9
APh 114b Lecture 5
Solid State Physics
So the donor bound states are:
E
d
E
c

m
e
4
2
2
l
2
2
E
i

e
2
2l
2
a
0
where
a
effective Bohr radius

a
0

2
m
e
2
For l 1,
E
d
E
c

m
e
4
2
2
2
Ex: Germanium(Ge): 16, average m
0. 12m, then E
c
E
p
0. 0072eV, a
0
70.
Notes for hydrogen atom, E 13. 6eV, a
0
0. 5
1
APh 114b Lecture 6
Solid State Physics
Quasi-Classical Electron Dynamics
and Electrical Conductivity
According to the Correspondence Principle of quantum mechanics,
wavepacket solutions of Schrdinger's equation follow the trajectories of
classical particles and satisfy Newton's laws. The connection between the
classical Hamiltonian and the quantum mechanical Hamiltonian is made by
identification of So,
p ih
quant um mechanical
E
n
i H
r
classical
E
n
r Hp, r
In classical mechanics, Hamilton's equations give the velocity as
v r
H
p

p
H
1
Ek
k
which agrees with the previously derived expression for the group velocity.
Hamilton's equation for acceleration is
p

H
r

H
r
r
2
APh 114b Lecture 6
Solid State Physics
For an applied electric field E, the perturbation Hamiltonian is
H
r er E
so
p
eE
In an applied magnetic field,
p
eE
v
c
B
Let's apply this development to the electrical conductivity of a metal with
an arbitrary band structure and Fermi surface. The electron is
considered to be a wavepacket with momentum k, and moving in an
external electric field E in compliance with Newton's Laws. The
electric field can be thought to create a displacement of the electron
distribution (i.e. Fermi surface) in k space. The displacement is
k
e
Et
3
APh 114b Lecture 6
Solid State Physics
The volume of k space swept out in a time t due to the presence of the field is
3
V
k
d
2
S
F
n
k d
2
S
F
n
Et
The current density is:
current
unit area

# electrons
unit volume in real space
electronic charge velocity
4
APh 114b Lecture 6
Solid State Physics
so

j

2
2
3

3
V
k
e V
g
and the instantaneous rate of change of current density is:
j
t

e
2
4
3

Fermi Surface
V
g
n
Ed
2
S
F

e
2
4
3

Fermi Surface
V
g
V
g
E
|
V
g
|
d
2
S
F
since the group velocity is directed normal to the Fermi surface,
(n

V
g
|V
g
|
).
In steady state, scattering limits the electrons which have not made a
collision at time t is
nt n0 e
t/
5
APh 114b Lecture 6
Solid State Physics
The relaxation time is the average collision time
t
1
t e
t/
dt
Thus in steady state t= and the current density is
j
e
2
4
3

Fermi Surface
V
g
V
g
E
|
V
g
|
d
2
S
F
Define the conductivity tensor as
j E
then

e
2
4
3

V
g
V
g
|V
g
|
d
2
S
F
6
APh 114b Lecture 6
Solid State Physics
In the free electron limit, becomes a scalar:
E

2
k
2
2m
E
F

2
k
F
2
2m
V
g

k
F
m
V
g
V
g
V
x
2
V
y
2
V
z
2
V
2
3
d
2
S
F
4k
F
2
The electron density is n
1
4
3
4
3
k
F
3
7
APh 114b Lecture 6
Solid State Physics
j
e
2
4
3
k
F
m
1
3
E 4k
F
2
ne
2
m
E
when
Ek

2
k
2
2m

ne
2

8
APh 114b Lecture 6
Solid State Physics
Boltzmann Transport Equation:
Application to Electrical Conductivity
The Boltzmann equation is at the heart of collision-limited transport
calculations for solids. We will defer a formal derivation of the Boltzmann
equation to a more advanced course, but we will explain and use this
important equation. In statistical mechanics, we commonly work with a
distribution function, f(r,p,t); this gives the probability of finding the system at a
point (r,p) in phase space at time t. For electrons in equilibrium,
f
r, p, t
f
0
r, p
f
0

1
1 e
EE
F
/kT
The current density in a solid can be expressed as:
j
e
4
3

vk fr, k, t dk
Note the crystal momentum k plays the role of momentum in determining the
location in phase space. The carrier density is
n
1
4
3

fr, k, t dk
9
APh 114b Lecture 6
Solid State Physics
The distribution function evolves in space and time via
1) diffusion
2) action of external forces and fields
3) collisions
Diffusion
The electrons can move in real space, thus moving in phase space by
virtue of thin drift velocity:
f
r, k, t
t
diffusion
vk
f
r, k, t
r
External Forces and Fields
The electrons evolve from the element dk centered around k by
changes in crystal momentum caused by external forces acting on
electrons, such as the Lorentz force:
hk
hk
e E
1
c
v
B
10
APh 114b Lecture 6
Solid State Physics
so
f
r, k, t
t
fields and forces
k
r, k, t
k
Collisions
Collisions, which alter the carrier momentum, can occur via electron-electron
scattering, electron-phonon scattering, ionized impurity scattering, defect
scattering, etc. Define a scattering probability from state k to k by one of these
processes, S(k,k). Because of the Pauli Exclusion Principle scattering always
occurs from an occupied state to an unoccupied state. Lets define:
f
k
probability that an electron initially occupies a state k
1 f
k
probability that the state k is unoccupied

Sk, k
probability for electron scattering fromstate k to state k
Sk
, k probability for electron scattering fromstate k
to state k
11
APh 114b Lecture 6
Solid State Physics
Then:
f
r, k, t
t
collisions

dk
f
k
1 f
k
Sk
, k f
k
1 f
k
Sk, k
At equilibrium, f
k
= f
o
(E), and the principle of detailed balance applies
Sk
, kf
0
E
1 f
0
E Sk
, kf
0
E1 f
0
E
so,
f
r, k, t
t
collisions
0
For an elastic collision,
Ek
Ek
12
APh 114b Lecture 6
Solid State Physics
f
0
f
0
and for elastic collisions,

so Sk
, k Sk, k
In steady state,
fr, k, t
t
0
So, accounting for diffusion, transistions due to fields and forces and
collisions,
f
r, k, t
t

f
r, k, t
t
diffusion
r, k, t
t
fields and forces
r, k, t
t
collisions
0
13
APh 114b Lecture 6
Solid State Physics
Thus, upon substitution of the diffusion and the forces and fields terms,
vk
f
r, k, t
t
k
r, k, t
t

f
r, k, t
t
collisions
and since in steady state
the Boltzmann equation can be written as:
fr, k, t
t
0
f
r, k, t
t
vk
f
r, k, t
t
k
r, k, t
t

f
r, k, t
t
collisions
14
APh 114b Lecture 6
Solid State Physics
The Linearized Boltzmann Equation
For small departures from equilibrium, one can linearize the Boltzmann equation
fr, k f
0
E f
1
r, k
The Relaxation Time Approximation
Even in linearized form, the Boltzmann equation is difficult to solve. We can
further simplify by assuming that the rate of return to equilibrium, upon the
cessation of the applied forces and fields, is characterized by a single
parameter called the relaxation time k
f
t
collisions

f f
0

f
1

15
APh 114b Lecture 6
Solid State Physics
Solution to this equation leads to
ft f
0
f0 f
0
e
t/
The concept of a relaxation time is useful when does not depend on the
magnitude of the driving term in steady state. This is approximately valid
for elastic scattering, but for inelastic scattering, it may not be true. A
useful criterion is whether or not the energy transfer upon collision E is
less that kT. Let's write:
f
1
r, k r, k
f
0
E
E
so the linearized Boltzmann equation is:
v
f
r
k

f
k

f
0
E
, f f
0
f
1
, f
1
small
16
APh 114b Lecture 6
Solid State Physics
Electrical Conductivity of Metals
Assume an applied electric field along the x-direction, and no magnetic
field or thermal gradients. Since the first term is
v
f
r

f
0
r

f
0
T
T
r
and
T
r
0
v
f
r
0
For the second term
f
k

f
0
f
1
k

f
0
E
E
k

f
1
k
17
APh 114b Lecture 6
Solid State Physics
and
hk
eE
In the linearized Boltzmann equation, we can neglect
f
1
k
, so
k
f
k

f
0
E
So eE
f
0
E
vk
k
f
0
E
and k eE vk
Note from above:
f
1
k k
f
0
E
eE vk
f
0
E
E
18
APh 114b Lecture 6
Solid State Physics
The electric current density is found from the distribution function as the
average <nev>
j
1
4
3

evk fk dk
1
4
3

evk f
1
k dk
since

ev
f
0
dk 0
This equation states there is no net current flow without an electric field, a
sensible result that is a manifestation of equilibrium! Now substituting for
f
1
j
e
2
E
4
3

vv
f
0
E
dk
19
APh 114b Lecture 6
Solid State Physics
where k and v
1
E
k
dk

d
2
Sdk

d
2
SdE
E
k
where dS denotes a volume element on a constant energy (Fermi) surface
Note that -
f
0
E
is delta function like at E
F
:
20
APh 114b Lecture 6
Solid State Physics
and the electrical conductivity is
In component form,
j
e
2
E
4
3

Fermi surface
vv
d
2
S
F
v

j
E

e
2
4
3

Fermi surface
vv
d
2
S
F
v
ij

e
2
4
3

Fermi surface
v
i
v
j
d
2
S
F
v
21
APh 114b Lecture 6
Solid State Physics
In a cubic crystal, the
ij
have the form:

0 0
0 0
0 0
so
j E
v v E
x
v
x
2
E
1
3
v
2
E
The contribution from the x component, for example, is
v
1
E
k
1
APh 114b Lecture 7
Solid State Physics
Electrical Conductivity of a Semiconductor
The equation that we developed for conductivity:
j
e
2
4
3

vv
f
0
E
dk E
is general, so it should be able to describe electrical conductivity of
semiconductors. For an intrinsic (undoped) semiconductor, (E-E
F
)>>kT.
Thus we can approximate the Fermi function as
f
0

1
1 e
EE
F
/kT
e
EE
F
/kT
If we define E=0 at the bottom of the
conduction band, E
F
< 0 and we can write:
f
0
E e
|
E
F
|
/kT
e
E/kT
2
APh 114b Lecture 7
Solid State Physics
and
fE
E

e
|
E
F |
/kT
e
E/kT
kT
Consider a semiconductor with spherical, parabolic bands
, so
k E
E

2
k
2
2m
and
vv
1
3
v
2
and
v
g

1
E
k

k
m
dk 4k
2
dk 4 2
m
2
3/2
E
1/2
dE
3
APh 114b Lecture 7
Solid State Physics
so
j
e
2
4
3
e
|E
F
|/kT
2 2
3
m
1/2
4
3
kT

0
E
3/2
dEe
E/kT
E
Cultural note about Gamma Functions:
x
p
dx e
x
p 1
x 1 xx; 1/2
The conductivity of a semiconductor is then:

2e
2
kT
2
2
3/2
e
|E
F
|/kT
4
APh 114b Lecture 7
Solid State Physics
The semiconductor carrier density is:
n
1
4
3

f
0
Edk
1
4
3
e
|E
F
|/kT
e
E/kT
4k dk
n
4 2
4
3
m
2
3/2
e
|
E
F |
/kT
kT
3/2

2
E
1/2
dEe
E/kT
n 2
m
kT
2
2
3/2
e
|E
F
|/kT
yielding our familiar Drude form for the conductivity:

ne
2
for constant .
5
APh 114b Lecture 7
Solid State Physics
If the semiconductor has more than one carrier pocket, then the
conductivity contributions of each carrier pocket add

i
i
i carrier pocket index
Suppose now that (acoustic phonon scattering), then E 1/E
1/2
f
ne
2
f = numerical factor of order unity

Suppose next that the Fermi surface is ellipsoidal rather than spherical, i.e.,
Ek

2
k
x
2
2m
x

2
k
y
2
2m
y

2
k
z
2
2m
z
6
APh 114b Lecture 7
Solid State Physics
Constant energy surface for the
first conduction band in Silicon
The energy value is 0.25eV above
the conduction band minima.
7
APh 114b Lecture 7
Solid State Physics
Conductivity for Ellipsoidal Fermi Surfaces
Ellipsoidal Fermi Surfaces:
Substitution of
k
i
m
0
k
i
m
i
, i x, y, z
yields a form Ek

2
k
2
2m
0
where
dk
m
xx
m
yy
m
zz
m
0
3
1/2
dk
and the carrier density associated with one carrier pocket is

n
i
2m
xx
m
yy
m
zz
1/2
kT
2
2
3/2
e
|E
F
|/kT
8
APh 114b Lecture 7
Solid State Physics
For an ellipsoidal Fermi surface the conductivity tensor is written

ne
2
with
j

For an electric field applied in a arbitrary direction, it makes sense to
write E in terms of its components along the semi-major axes, because
along these directions, the conductivity tensor is diagonal

xx
0 0
0
yy
0
0 0
zz
9
APh 114b Lecture 7
Solid State Physics
When
xx

yy
, the conductivity is different along different directions
Let (x,y,z) be the coordinate system for major axes of the ellipsoid and (X,Y,Z)
be the lab frame coordinate system. Then
j
X
j
Y
j
Z

XX

XY

XZ
YX

YY

YZ
ZX

ZY

ZZ
E
X
E
Y
E
Z
10
APh 114b Lecture 7
Solid State Physics
and in the (x,y,z) system,
j
X
j
Y
j
Z
ne
2
1
m
x
0 0
0
1
m
y
0
0 0
1
m
z
Ecos
Esin
0
ne
2
cos
m
x
sin
m
y
0
E
11
APh 114b Lecture 7
Solid State Physics
We can go from the lab frame to the carrier pocket frame by a coordinate
transform
j
X
j
Y
j
Z
R
j
X
j
Y
j
Z
j
X
j
Y
j
Z
ne
2
cos sin 0
sin cos 0
0 0 1
cos
m
x
sin
m
y
0
E
12
APh 114b Lecture 7
Solid State Physics
j
X
E
j
Y
E
j
Z
E
ne
2
cos
2
m
x

sin
2
m
y
cos sin
1
m
y

1
m
x
0
cos sin
1
m
y

1
m
x
sin
2
m
x

cos
2
m
y
0
0 0
1
m
z
Note that semiconductors with ellipsoidal Fermi surfaces usually have several
such Fermi surfaces located in crystallographically equivalent locations. If the
semiconductor has cubic symmetry, the sum of the conductivity components
results in an isotropic conductivity even though the contribution of each
ellipsoid is anisotropic! Thus the measurement of electrical conductivity
provides no information about the anisotropy of the Fermi surface in cubic
materials!
So... does an ellipsoidal Fermi surface even matter? Yes! Because not all
systems of interest have high symmetry.
13
APh 114b Lecture 7
Solid State Physics
Example: Strained Si MOSFET with Enhanced Electron Mobility
Along <100> directions, m
l
= 0.92m , m
t
= 0.19m
Ek

2
k
y
2
2m
l

2
k
x
2
2m
t

2
k
z
2
2m
t
and in the absence of strain, all 6 pockets are equivalent.
14
APh 114b Lecture 7
Solid State Physics
Upon application of tensile strain in biaxial form, the equivalence of the
pockets is broken, the pockets along k
y
and k
x
rise in energy and those with
axes along k
z
fall in energy. Now the relative contributions of the carrier
pockets is unequal! This results in a larger contribution to the carrier density
from the carrier pockets that are characterized by smaller effective mass.
This principle is exploited in Si MOS field effect transistors to enhance the
electron and hole channel mobility. To do this, first a virtual substrate
consisting of a relaxed Si
1-x
Ge
x
buffer layer is grown, and then a tensile-
strained Si layer is grown epitaxially on the Si
1-x
Ge
x
layer.
15
APh 114b Lecture 7
Solid State Physics
Strained Si MOSFET
Both electron and hole mobilities are
enhanced in-plane directions due plane
directions due to suppressed
intervalley/interband scattering and
reduced effective mass.
n-FET and p FET mobility enhancements
have been demonstrated experimentally.
up to ~80%.
Reference: K. Rim, J. Hoyt, and J. Gibbons, Fabrication and
Analysis of Deep Submicron Strained-Si n-MOSFETs,
IEEE Trans. Electron Devices 47, 1406 (2000).
16
APh 114b Lecture 7
Solid State Physics
Strained Si MOSFET:
Mobility Enhancement with Strain
1
APh 114b Lecture 8
Solid State Physics
Carrier Density and Mobility :
Last time, we learned that we could characterize semiconductor electrical
conductivity by a conductivity tensor with components

ne
2
where the ellipsoidal Fermi surfaces gave rise to an effective mass tensor.
So we have learned something about the anisotropy of the conductivity, via
the effective mass tensorbut what about it's magnitude? To answer this,
we need to be able to find the carrier density. Our general expression is, for
electrons in the conduction band,
n

E
c
Fermi funct ion

f
0
E
densit y of st at es
conduct ion band
g
c
E dE
2
APh 114b Lecture 8
Solid State Physics
Similarly, for holes in the valence band,
p

0
E
v
densit y of st at es
valence band
g
v
E
1 Fermi funct ion
hole occupancy, i.e.
1 f
0
E dE
Writing in constitutive form,

p

n
pe
p
ne
n
where
p

e
p
m
p
n

e
n
m
n

3
APh 114b Lecture 8
Solid State Physics
Electrons:
Holes:
Density of States
State Occupancy
Product = (DOS)(State Occupancy)
Density of States
State Occupancy
Product = (DOS)(State Occupancy)
4
APh 114b Lecture 8
Solid State Physics
In the case of an intrinsic semiconductor (i.e., no dopant impurities added)
the valence band is completely filled at T=0 and the conduction band is
completely empty at this temperature. At finite temperatures, the carrier
density is finite.
n.b., on energy axis, increasing
hole energy is opposite w.r.t. to
increasing electron energy:
5
APh 114b Lecture 8
Solid State Physics
In an intrinsic semiconductor, therefore, n=p. If the valence and
conduction bands are parabolic near the band extrema,
E
n

2
k
2
2m
n
E
p

2
k
2
2m
h
then the densities of states can be written as

g
c
E
1
2
2
2m
n
2
3/2
E
n
1/2
where E
n
E E
c
g
v
E
1
2
2
2m
p
2
3/2
E
p
1/2
where E
p
E E
v

6
APh 114b Lecture 8
Solid State Physics
Substituting into the carrier density,
n
1
2
2
2m
n
2
3/2
e
E
F
/kT
n
1/2
E
n
1/2
e
E
n
/kT
dE
n
where we have approximated
f
0

1
1 e
EE
F
/kT
e
EE
F
/kT
and
n

0
g
c
E
n
f
0
E
n
E
F
dE
n
so
n 2
m
n
kT
2
2
3/2
e
E
F
/kT
7
APh 114b Lecture 8
Solid State Physics
n.b.
0
x
1/2
e
x
dx

2
Likewise,
p 2
m
p
kT
2
2
3/2
e
E
F
/kT
Their product np is np 4
kT
2
2
3
m
n
m
p
3/2
e
E
g
/kT
Recall for an intrinsic semiconductor, n = p, so
n p 2
kT
2
2
3/2
m
n
m
p
3/4
e
E
g
/2kT
We can now connect the Fermi level position to the energy gap,
n 2
m
n
kT
2
2
3/2
e
E
F
/2kT
2
kT
2
2
3/2
m
n
m
p
3/4
e
E
F
/2kT
8
APh 114b Lecture 8
Solid State Physics
giving:
E
F

1
2
E
g

3
4
kTln
m
p

m
n
Doped Semiconductors -- n type

Assume there are N
D
donors per unit volume with a bound state at energy
E
D
referenced to a single nondegenerate conduction band. The distribution
depends on their degeneracy. If we assume the usual case of
f
D
E
1
1
1
2
e
E
D
E
F
/kT
So the Fermi level is near the middle of the gap in an undoped semiconductor.
If the hole effective mass is a little larger than the electron effective mass,
the Fermi level is a little below midgap.
then the density of ionized donors
N
D
, is 1 f
D
E
D
N
D
or
N
D
N
D
1
1 2e
E
D
E
F
/kT
9
APh 114b Lecture 8
Solid State Physics
Note that when E
D
- E
F
is greater than several kT, all the donors are
ionized (empty)!! When E
F
E
D
from below, N
D
N
D
and the donor states are partially filled. The picture when E
F
in an n-type
semiconductor is a few kT below the donor (~all donors are ionized):
Note that the Fermi energy position in a doped semiconductor is also
determined by a requirement of overall neutrality. The charge density
is:
In general for n-type material,
qN
D
p n
q N
D
1
1 2e
E
D
E
F
/kT
N
v
e
E
i
E
F
/kT
n
10
APh 114b Lecture 8
Solid State Physics
If we assume complete donor ionization, it is then O.K. to use
Boltzmann statistics for n. Now =0 implies:
N
D
n p
and if N
D
N
D
p
n
i
2
n
and , we have
n
N
D
2

N
D
2
2
n
i
2
1/2
Two limiting cases:
i)
ii)
N
D
2
n
i
(This is hard to achieve) n n
i

N
D
2
N
D
2
n
i
(This is the typical case)
n N
D
p
n
i
2
N
D
11
APh 114b Lecture 8
Solid State Physics
However!! When N
D
gets to be large, several things occur.
1) N
D
+
< N
D
(partial donor ionization)
2) Bandgap narrowing: the donor states form a band, and the donor band
merges with the conduction band, leading to a decrease in E
c
and hence
in E
G
.
Doped Semiconductor -- p type
The mathematical analysis for p type material is similar, except that an
acceptor is ionized when occupied.
12
APh 114b Lecture 8
Solid State Physics
For Si, Ge, and GaAs, the valence band is twofold degenerate at k = 0, so
accounting for spin degeneracy, acceptors are four-fold degenerate:
N
A
N
A
1
1
1
4
e
E
F
E
A
/kT
When E
F
is more than a few kT above E
A
, we assume total acceptor
ionization, i.e.,
N
A
N
A
and if N
A
n
i
, we find p N
A
, n
n
i
2
N
A
.
When both types of dopants are present, neutrality requires
qN
D
N
A
p n 0
in equilibrium, and this condition is used to find E
F
.
13
APh 114b Lecture 8
Solid State Physics
Some convenient notation for carrier densities in the Boltzmann limit:
n
i

N
c
N
v
1/2
e
E
G
/2kT
n n
i
e
E
F
E
i

/kT
p n
i
e
E
F
E
i
/kT
1
APh 114b Lecture 9
Solid State Physics
Scattering Mechanisms
Electron Scattering Mechanisms
In development of a semiclassical model for transport such that

ne
2
ne,
e
m
the scattering time is the parameter that we currently know least about.
Electron scattering restores a system back toward equilibrium. The
scattering rate depends on both the collision frequency and the
effectiveness of a scattering event in returning a system to equilibrium
(thus a small angle scattering collision is not as effective as a large angle
scattering event in restoring equilibrium). We can define two parameters
D
time for systemto be restored to be equilibrium
c
time between collisions
2
APh 114b Lecture 9
Solid State Physics
where

D

c
1 cos
, scattering angle
The collision time is related to the mean free path as
v
c
and to the density of scattering centers N
c
as
1
c
N
c
v
where scattering cross section. Scattering centers include
phonons, impurities, crystal defects, interfaces, surfaces, etc.. For
both metals and semiconductors, electron-phonon scattering is the
most important above cryogenic temperatures.
3
APh 114b Lecture 9
Solid State Physics
However, electron-phonon scattering is physically very different in
semiconductors than in metals! Electrons must be scattered from
occupied states to unoccupied states, and hence electron scattering
occurs from one point to another on the Fermi surface. In metals, most of
the Brillouin zones occupied by electrons (i.e. Fermi surface fills Brillouin
zone). By contrast, in semiconductors, most of the Brillouin zone is
unoccupied by electrons (i.e. Fermi surface occupies small fraction of the
Brillouin zone). As a result, scattering in metals tends to result in large
momentum changes, whereas in semiconductors, scattering can occur
with small momentum changes.
Scattering Processes in Semiconductors
Electron-Phonon Scattering
Electron-phonon scattering predominates except at low temperatures.
Conservation of energy and momentum gives:
ENERGY:
E
i
E
f

2
2m
k
i
2
k
f
2
4
APh 114b Lecture 9
Solid State Physics
where E
i
initial energy
E
f
final energy
k
i
initial wavevector
k
f
final wavevector
q phonon wavevector
MOMENTUM:
k
i
k
f
q
In semiconductors, scattered electrons generally remain in the vicinity of
a single band extremum, with small changes in k (low q phonon
wavevectors).
Density of final states:
E
f

2m
3/2
E
f
1/2
2
2
3
E
f

2m
3/2
E
i
q
1/2
2
2
3
5
APh 114b Lecture 9
Solid State Physics
The probability that an electron makes a transition from an initial state i to
a final state f is proportional to:
1. The availability of electron final states ((density of
states)(Prob{unoccupied}) we assume Prob{unoccupied}1 in a
semiconductor conduction band.
2. The probability of absorbing or emitting a phonon:
E
f

2m
3/2
E
f
1/2
2
2
3

2m
3/2
E
i
q
1/2
2
2
3
absorption: n

1
e
q/kT
1
emission: 1 n

1
1 e
q/kT
6
APh 114b Lecture 9
Solid State Physics
3. The electron-phonon coupling strength, C(q). Thus
1

2m
2
3/2
4
2
3

q
C
phonon absorpt ion

E
i
q
1/2
e
q/kT
1
phonon emission
E
i
q
1/2
1 e
q/kT
The electron-phonon coupling mechanisms give the value of C(q):
1. Electromagnetic Coupling - oscillating E-field gives rise to oscillating
dipole moments; important if charge distributions different on
neighboring ion sites (III-V and II-VI semiconductors).
2. Piezoelectric Coupling - important in ionic or partially ionic
semiconductors lacking inversion symmetry. Acoustic mode vibration
cause compressions and rarefaction that produce electric fields which
scatter electrons.
3. Deformation Potential Coupling - associated with changes in energy
band extrema caused by compression and rarefaction of crystals
during acoustic mode vibration. Important in crystals with inversion
symmetry (i.e. no piezoelectric coupling) and having the same species
on each site (hence no electromagnetic coupling). Important in Si, Ge
and diamond.
7
APh 114b Lecture 9
Solid State Physics
For electromagnetic coupling, optical phonon scattering is most
important. For optical modes (q) is only weakly dependent on q
because electromagnetic excitation involves only small wavevectors.
In this case
0
kT and
0
E, so
1
m
3/2
0

1/2
1 e
0
/kT

8
APh 114b Lecture 9
Solid State Physics
and the mobility

e
m

1 e
/kT
m
5/2
0

1/2
The mobility decreases with increasing temperature is independent of
energy E.
For deformation potential coupling, it is acoustic phonons that are
important. Then kT and E.
For the acoustic phonons of interest
Cq q and q.
So
1
e
/kT
1

1
1

kT
1

kT

kT
q
Thus
are independent of q
Cq
e
/kT
1
and
1
1e
/kT
and
9
APh 114b Lecture 9
Solid State Physics
The mobility

e
m

1
m
5/2
E
1/2
kT
Other Scattering Mechanisms
So far we have discussed electron-phonon coupling, but there are other
important mechanisms for scattering that we should be aware of,
including:
Electron-electron scattering
This type of scattering redistributes energy and momenta among
carriers, and acts in conjunction with other scattering mechanism to
restore equilibrium.
Intervalley scattering
Scattering can occur within e.g. the conduction band from either:
1) one equivalent minimum to another (e.g. Si X point X point) or
2) from a lower lying to a higher lying conduction band minimum (e.g.
Gunn effect in GaAs: enables 100-200 GHz oscillators due to
negative differential mobility.)
10
APh 114b Lecture 9
Solid State Physics
Intervalley scattering:
GaAs Si
Defect scattering
Point defects, dislocations, interfaces, grain boundaries, etc., can limit mean
free path.
1
APh 114b Lecture 11
Solid State Physics
Thermal Conductivity and Thermoelectric Effects
Boltzmann Equation Approach to Thermal Conductivity
We now consider a parallel development to our expression for electrical
conductivity using the Boltzmann transport equation to determine the electrical
contribution to the thermal conductivity.
Application of a temperature gradient gives rise to heat flow, with a thermal
current, u, which is driven by an energy,
where f = f
0
+ f
1
(f
0
is the Fermi function).
At equilibrium, there is no net current u, i.e.,
so
u
1
4
3
vE E
F
f dk
E E
F
:
E E
F
f
0
dk 0
u
1
4
3

vE E
F
f
1
dk
2
APh 114b Lecture 11
Solid State Physics
Recall that f
1
is found from the Boltzmann equation:
At steady state,
and since in the absence of an electric field,
so the Boltzmann transport equation reduces to:
(Note also that this will come into play later.)
f
t
v
f
r
k
f
k

f
1
f/t 0
v
f
r

f
1
v
1
E
k
k
0, k
eE/,
3
APh 114b Lecture 11
Solid State Physics
For small perturbations, we can linearize f and expand the derivative
such that:
f
r

e
EE
F
/kT
1e
EE
F
/kT
2

1
kT
E
F
r

EE
F
kT
2
T
r
f
r
kT
f
0
E

1
kT
T
r
E
F
T

EE
F
T
f
r

f
0
E
E
F
T

EE
F
T
T
r
f
r

f
0
r

r
1
1e
EE
F
/kT
f
0
chain rule with
E
F
E
F
r, T Tr
f
0
E

1
1e
EE
F
/kT
2
e
EE
F
/kT
kT
simplify
4
APh 114b Lecture 11
Solid State Physics
Thermal Conductivity in Semiconductors and Metals
For semiconductors:
Plugging in reasonable values, if then
On the other hand, for a band gap of and
So we see that:
E
F

1
2
E
g

3
4
kTln
m
p
m
n
E
F
T

3
4
k ln
m
p
m
n
m
p
/m
n
2,
E
g
0. 5eV E E
F
E
g
,
EE
F
T

0.5eV
300K
k
8.610
5
eV/K
20k
EE
F
T

E
F
T
E
F
/T 0. 5k.
5
APh 114b Lecture 11
Solid State Physics
This is also true for metals:
At T = 300K and if
So now ,
becomes:
E
F
E
F
0

2
12
kT
2
E
F
0
E
F
T

2
6
kT
E
F
0
k
E
F
T
8 10
3
k
E
F
0
5eV,
EE
F
T

E
F
T
f
r

f
0
E
E
F
T

EE
F
T
T
r
f
r

f
0
E
EE
F
T
T
r
6
APh 114b Lecture 11
Solid State Physics
Now, from our simplified Boltzmann equation,
so
And so, for our definition of thermal current,
we get:
v
f
r

f
1
f
1
v
f
r
f
1
v
f
0
E
EE
F
T
T
r
u
1
4
3

vE E
F
f
1
dk
u
1
4
3
T
T
r
vvE E
F
2
f
0
E
dk
7
APh 114b Lecture 11
Solid State Physics
Thermal Conductivity
The thermal conductivity, , is defined by its relationship to the thermal
current, u:
We thereby obtain the electronic contribution to thermal conductivity,
where dk can be given by:
dk d
2
sdk
d
2
sdE
E
k
1
d
2
s
dE
v
v
1
E
k
K
u K
T
r
K u
T
r
1
K
1
4
3
T
vvE E
F
f
0
E
dk
8
APh 114b Lecture 11
Solid State Physics
Wiedemann-Franz Law
We can evaluate the thermal conductivity in the expression
by the expanding the integral about the Fermi energy:
And since in our case,
we need keep only the second derivative term,
GE
F
vvE E
F
2
E
F
0
2
G
E
2
2vv
|
E
F
K
1
4
3
T
vvE E
F
f
0
E
d
2
S
v
dE
GE
f
0
E
E GE
F

2
6
kT
2
2
G
E
2
E
F

9
APh 114b Lecture 11
Solid State Physics
Thus, the thermal conductivity is:
and, since electrical conductivity was:
we get the Wiedemann-Franz Law:
Physically:
In electrical conduction, each e
-
carries a charge e, is acted on by a force e,
and the current per electric field is proportional to e
2
.
In thermal conduction, each e
-
carries a thermal energy kT, is acted on by a
thermal force -kT, and the current/thermal gradient is proportional to k
2
T.
So we expect:

e
2
4
3
vv
d
2
S
F
v
K
k
2
T
12
vv
d
2
S
F
v
K T

2
k
2
3e
2
K

k
2
T
e
2
10
APh 114b Lecture 11
Solid State Physics
At high temperatures, the dominant scattering
mechanism is electron-phonon scattering: (1/T)
and thus constant, independent of temperature.
At low (but not very low) temperatures,
(
D
/T)
5
with T
D
, but the phonon density T,
so (1/T).
At lower temperatures, T
D
, impurity and
defect scattering dominates, and f(T), so T
and the Wiedemann-Franz Law holds again.
In metals, the Wiedmann-Franz Law works well at high temperatures (T >
D
)
and at very low temperatures (T
D
):
Note that the Wiedemann-Franz Law also breaks down when the lattice thermal conductivity
becomes comparable to the electronic thermal conductivity.
Typical thermal conductivity for a metal
K
K
K
11
APh 114b Lecture 11
Solid State Physics
In semiconductors, the conduction band energy, since
electrons in the conduction band carry the heat, so
For parabolic bands,
so:
E

2
k
2
2m
dk
0
2
2m
2
3/2
E
1/2
dE
v
1
E
k

k
m
Parabolic Bands
E E
F
E,
f
0
E

1
kT
e
E
F
/kT
e
E/kT
12
APh 114b Lecture 11
Solid State Physics
We can therefore express as:
The temperature dependence is:
vv v
x
2
E E
F
E
f
0
E

1
kT
e
E
F
/kT
e
E/kT
dk
0
2
2m
2
3/2
E
1/2
dE
105

8
K
xx

1
4
3
T

0
v
x
2
E
2
1
kT
e
E
F
/kT
e
E/kT
2
2m
2
3/2
E
1/2
dE
K
xx

kkT
3
2
m
2m
kT
2
3/2
e
E
F
/kT
x
7/2
e
x
dx
K
xx
T
5/2
e
E
F
/kT
K
xx
13
APh 114b Lecture 11
Solid State Physics
The temperature dependence is:
so since
we find that:
The Wiedemann-Franz Law also holds for (non-degenerate) semiconductors.
xx

2e
2
kT
2
2
3/2
e
E
F
/kT
K
xx
T
5/2
e
E
F
/kT
K
xx

35
2
k
2
e
2

xx
T
14
APh 114b Lecture 11
Solid State Physics
Thermoelectric Effects
There often exists a coupling between electrical and thermal currents.
Recall that the currents are given by:
Electric: Thermal:
where:
Using
we can obtain:
u
1
4
3
vE E
F
f
1
dk j
e
4
3
vf
1
dk
f
1
v
f
r
k
f
k
f
k

f
0
E
E
k
f
1
v
f
0
E
E E
F
T
T eE
f
1
v
f
0
E
EE
F
T
T eE
v
1
E
k
k

eE
15
APh 114b Lecture 11
Solid State Physics
In the case where there is both an applied electric field and a temperature
gradient, the electric current is:
where
and
The thermal motion of carriers across a temperature gradient therefore
results in an electric current given by:

e
2
4
3
vv
d
2
SF
v
K
0

1
4
3

vv
f
0
E
dk

e
2
K
1

1
4
3
vvE E
F

f
0
E
dk
j
e
4
3

vf
1
dk e
2
K
0
E
e
T
K
1
T
e
T
K
1
T
16
APh 114b Lecture 11
Solid State Physics
Seebeck Effect
A temperature gradient will be accompanied by an electric field in the
opposite direction. This thermoelectric field is conventionally written as:
where the proportionality constant, is the thermopower.
At steady state (under conditions of zero current flow):
Q,
E QT
j 0 e
2
K
0
E
e
T
K
1
T
E
1
eT
K
0
1
: K
1
T Q T Q
1
eT
K
0
1
: K
1
17
APh 114b Lecture 11
Solid State Physics
When Q is scalar (the material is isotropic),
this is also sometimes called the Seebeck coefficient.
In this system where there is both an applied electric field and a
temperature gradient, there is also a thermal current:
where is defined as above, and
u
1
4
3

vE E
F
f
1
dk eK
1
E
e
T
K
2
T
K
1
K
1
4
3
T
vvE E
F
f
0
E
dk
K
2

1
4
3
vvE E
F
f
0
E
dk T K
Q
1
eT
K
1
K
0
18
APh 114b Lecture 11
Solid State Physics
Peltier Effect
If an electric current is driven in a bimetallic circuit maintained at uniform T,
then heat will be evolved at one junction and absorbed at the other. This is
because an isothermal electric current in a metal is accompanied by a thermal
current.
In the absence of a thermal gradient, T = 0.
so
u j
j e
2
K
0
E
e
T
K
1
T
j e
2
K
0
E
u eK
1
E
e
T
K
2
T u eK
1
E j

1
e
K
1
: K
0
1
19
APh 114b Lecture 11
Solid State Physics
When the materials are isotropic,
Note that both and are proportional to 1/e, and thus the sign of the
charge carrier.
Measurement of can be used to determine if electrons or holes are the
majority carriers.
To determine the thermoelectric behavior of materials, we need to find
for the material.

1
e
K
1
K
0
Q
Q
K
1

1
4
3

vv
E E
F

f
0
E
dk
20
APh 114b Lecture 11
Solid State Physics
Metals: Thermoelectric Effects
In metals, we note that is delta-function-like, and
Recalling that,
We define here:
Where
f
0
/E
dk d
2
S
dE
V
GE
f
0
E
E GE
F

2
6
kT
2
2
G
E
2
E
F

GE gEE E
F
gE
1
4
3

vv
d
2
S
V
dE K
0
21
APh 114b Lecture 11
Solid State Physics
So,
and since
we find that:
For spherical parabolic bands,
and
2
G
E
2

2
g
E
2
E E
F
2
g
E
GE
F
0
2
GE
F
E
2
2
gE
F
E
E
2
k
2
/2m
gE
1
4
3

vv
d
2
S
V
dE

3
2
m
2m
2
3/2
E
3/2
22
APh 114b Lecture 11
Solid State Physics
In metals we therefore have,
and
So the Seebeck coefficient in a metal is:
K
0
gE

3
2
m
2m
2
3/2
E
F
3/2
K
1

1
4
3

vvE E
F

f
0
E
dk

6m
2m
2
3/2
E
F
1/2
k
2
T
2
Q
m

2
k
2e
kT
E
F
23
APh 114b Lecture 11
Solid State Physics
Semiconductors: Thermoelectric Effects
Again,
but since
so,
Since and
K
2

1
4
3

vv
E E
F

f
0
E
dk
E E
F
E,
f
0
E

1
kT
e
E
F
/kT
e
E/kT
E
2
k
2
/2m
, v k/m
,
dk 2
2m
2
3/2
E
1/2
dE
f
0
e
E/kT
e
|
E
F
|
/kT
24
APh 114b Lecture 11
Solid State Physics
Then, is:
and the Seebeck coefficient is found to be
where,
thus yielding:
K
1xx

5kT
m
kT
2
2
3/2
e
|E
F
|/kT
Q
s

1
eT
K
1xx
K
0xx
K
0xx

xx
e
2
Q
s

5
2
k
e
K
1xx
25
APh 114b Lecture 11
Solid State Physics
Above, we developed constitutive relations for the Seebeck coefficient, Q,
and the Peltier coefficient, . Let's consider some simple important examples
of these phenomena.
Seebeck Effect in a Thermocouple
In a metal subject to a thermal gradient T, there exists an effective
electric field,
under conditions of zero net current flow,
The Seebeck coefficient is
E Q|T|
j 0 e
2
K
0
E
e
T
K
1
T
Q
1
eT
K
1
K
0
26
APh 114b Lecture 11
Solid State Physics
In practice we typically measure the difference in thermopower,
Q
A
Q
B
, between two different metals A and B by measuring the
open circuit voltage V
AB
.
So at open circuit,
if T
1
T
2
, (metal 1 metal 2) a potential is developed, and it is independent of
T
0
. For a metal, on the other hand, note that:
V
AB

E dr
Q
T
r
dr
V
AB

T
0
T
1
Q
B
dT
T
1
T
2
Q
A
dT
T
2
T
0
Q
B
dT
Q

2
2
k
e
kT
E
F
T
V
AB

T
1
T
2
Q
A
Q
B
dT.
27
APh 114b Lecture 11
Solid State Physics
Peltier Heat Pump
In a metal subject to an electric field but no temperature gradient,
there will be an effective thermal current, so:
or for scalar
The Peltier effect is measured by the heat absorbed or generated at the
junction between two dissimilar materials. Sending an electric current around a
circuit of two dissimilar materials cools one junction and heats the other.
j e
2
K
0
E
e
T
K
1
T
j e
2
K
0
E
u
K
1
e
2
j K
0
1
K,
u eK
1
E,
u
K
1
eK
0
j j
28
APh 114b Lecture 11
Solid State Physics
A Final Thought
Given the close correspondence between the Seebeck and Peltier effects,
it is perhaps not surprising that materials with large Seebeck coefficients have
larger Peltier coefficients.
References
Ashcroft and Mermin pp. 20 25; 256 258.
Kittel Appendix F: Boltzmann Transport Equation.
An Introductory graduate course in Solid State Physics
(Mark Jarrell, U. of Cincinnati), Chapter 9: Electronic Transport
http://www.physics.uc.edu/~jarrell/COURSES/SOLID_STATE/Chap9/chap9.pdf
1
APh 114b Lecture 12
Solid State Physics
We now embark on a discussion of the optical properties of solids.
In optical experiments in which we measure reflectivity, transmission
coefficient, luminescence, or light scattering, we can learn a great deal
about energy band structure, impurities, phonons, defects and
magnetic excitations. In particular, optical measurements are used to
determine the complex dielectric or complex conductivity, which is
directly related to bandstructure.
H
1
c
D
t

4m
c
j
E
1
c
B
t
0
D 0
B 0
Faradays Law
Gauss Law
We start with
Maxwell's Equations:
Amperes Law
(in absence of space charge)
No Magnetic Monopoles!
2
APh 114b Lecture 12
Solid State Physics
and the constitutive relations for homogeneous media:
D cE
B jH
j oE
(Ohms Law)
Combining the Ampere and Faraday laws with the constitutive relations and
yields a wave equation:
2
E
jc
c
2
2
E
t
2

4moj
c
2
E
t
3
APh 114b Lecture 12
Solid State Physics
For optical frequencies the electric fields in homogeneous media
with planar boundaries have solutions of the form
E E
0
e
ikrot
where k is a complex propagation constant, so the real{k} gives the
wavevector and imaginary{k} gives the absorption. We can solve for
The complex propagation constant k is obtained by substitution in the
wave equation
k
2
jco
2
c
2

4miojo
c
2
We can write the propagation constant in terms of a complex dielectric
constant
c
k
o
c
c
c
j
c
c
c
4mio
o
c
1
ic
2
4
APh 114b Lecture 12
Solid State Physics
We can also define a complex conductivity
c
such that
c
c

4mi
o
o
co
4mi

4mi
o
o
c
o
c
o
co
4mi
i.e.,
Having defined
c
and
c
, we will soon relate these to
1) laboratory observables (reflectivity, absorption, etc.)
2) properties of solids (carrier density, band structure, etc.)
Before going on, there are some more definitions that you need to
know. Many handbooks describe the dielectric behavior of solids in
terms of n
c
the complex refractive index (c.f. Handbook of Optical
n
c
2
jc
c
n ik
2
Constants by E.D. Palik
Taking = 1 in nonmagnetic materials,
5
APh 114b Lecture 12
Solid State Physics
c
c
n
c
2
c
1
ic
2
n ik
2
so the real part is c
1
n
2
k
2
and the imaginary part is
c
2
2nk
For a beam incident on a solid, we can develop the picture below:
The wave inside the solid is
with E
x
E
0
e
ikzot
k
o
c
n
c
6
APh 114b Lecture 12
Solid State Physics
Since n = 1 in free space, the boundary condition E
0
E
1
E
2
gives the wave in free space as
E
x

incident beam
E
1
e
i
oz
c
ot

reflected beam
E
2
e
i
oz
c
ot
From
E
j
c
H
t

ijo
c
H
we get
E
x
z

ijo
c
H
y
so
E
0
k E
1
o
c
E
2
o
c
E
0
o
c
n
c
The normal incidence reflectivity is
R
E
2
E
1
2
7
APh 114b Lecture 12
Solid State Physics
Noting
E
2

1
2
E
0
1 n
c
E
1

1
2
E
0
1 n
c
so
R
1n
c
1n
c
2
1n
2
k
2
1n
2
k
2
where R < 1. Power continuity dictates:
1 R A T
where R = reflectivity, A = absorbance, T = transmittance.
8
APh 114b Lecture 12
Solid State Physics
We have now related an observable, R, to properties of solids such
as n
c
. If we can measure reflectivity at normal incidence over all
frequencies, we can obtain both n and . This is because n and
are not independent, but are related through the Kramers-Kronig
relations. In assessing the optical response of a solid, we will
examine both:
i) intraband transitions (e.g. Free carrier absorption)
ii) interband transitions
Free Carrier Contribution to Optical Properties: Drude Theory
For free carriers, we can apply the semi-classical equations
developed for transport:
m
dv
dt

m
v
t
eE
0
e
iot
where the time harmonic carrier motion is describable in terms of a carrier
velocity . v v
0
e
iot
9
APh 114b Lecture 12
Solid State Physics
m
io
m
t
v
0
eE
0
So
The current j nev
0
o
c
E
0
giving
v
0

eE
0
m
t
im
o
and a complex conductivity
o
c

ne
2
t
m
1iot
Free Carrier Absorption: Drude Theory (contd)
We recall that j E where
j
o
o
o[
E
[, with
1
m

o[

1
2
Ek
k
2
10
APh 114b Lecture 12
Solid State Physics
Suppose the only conduction mechanism we treat explicitly is that due
to free carriers. Then all other mechanisms can be included in a core
dielectric constant,
core
, so the dielectric constant is written as
c c
core

4mio
o
Without free carrier absorption, =0 and =
core
. In the Drude theory,
the complex dielectric constant is written as
c c
core

4mi
o
ne
2
t
m1iot
There are two limiting cases:
ot 1
ot 1
(low frequencies)
(high frequencies)
11
APh 114b Lecture 12
Solid State Physics
At low frequencies,
ot 1. Here we obtain
c c
core

4mine
2
t
mo
Since the free carrier term is proportional to 1/o,
this term dominates at low frequencies. For Si, c 11. 5
in Ge,
,
c 16 and c 100 in PbTe. To find n and k,
take the square root of
c
4mne
2
t
mo
i
c
n.b.:
i e
mi/4
1i
2
So in the low frequency limit, n k
and both are large.
12
APh 114b Lecture 12
Solid State Physics
The low frequency normal incidence reflectivity is
R
n1
2
k
2
n1
2
k
2

n
2
k
2
n
2
k
2
1
ot 1
So at low frequencies, materials with free carriers are very good reflectors.
The free carriers can respond rapidly, on the time scale of an AC cycle, so
as to screen the external applied AC field.
At high frequencies, ot 1
, and in this limit,
c c
core

4mne
2
mo
2
and
c c
core
A real number. Thus n 0 and k 0
13
APh 114b Lecture 12
Solid State Physics
So the reflectivity approaches a constant value
R
n1
2
n1
2
where
n c
core
.
Thus the Drude theory predicts that materials with high free carrier
densities become dielectrics at high frequencies. At high frequencies,
the conduction electrons can no longer move rapidly to screen the
external applied field on the timescale of an AC cycle. Only the core
electrons bound to the atoms can contribute to the dielectric response.
The characteristic frequency at which the metallic-to-dielectric transition
occurs is called the plasmon frequency, which is the frequency at which
the real part of c vanishes.
14
APh 114b Lecture 12
Solid State Physics
Free Carrier Absorption: Plasmon Frequency
In Drude theory, the complex dielectric constant is
c c
1
ic
2
c
core

4mi
o
ne
2
t
m
1iot
1iot
1iot
so the real part is
c
1
c
core

4mne
2
t
2
m
1o
2
t
2
c
1
o
p
0 c
core

4mne
2
t
2
m
1o
p
2
t
2
so
o
p
2
4mne
2
m
c
core
1
t
2
o
p
2
4mne
2
m
c
core
or
15
APh 114b Lecture 12
Solid State Physics
At the plasmon frequency, is also small since at c
2
o
p
,
c
2

4m
m
o
p
ne
2
t
1o
p
t
2
whereas above the plasmon frequency, where
o
p
t 1,
c
2
c
core
/o
p
t small
At o
p
, c
1
0 and n k
and thus c
2
2nk 2n
2
.
So near
o
p
, n tends to be small, and thus the reflectivity R
n1
2
k
2
n1
2
k
2
goes to a minimum at resonance, corresponding to a condition of
resonant absorption.
16
APh 114b Lecture 12
Solid State Physics
Silver
Complex
refractive
index real
part n and
imaginary
part for Ag.
Gold
Complex
refractive
index real
part n and
imaginary
part for
Au.
Aluminum
Complex
refractive
index real
part n and
imaginary
part for Al.
Copper
Complex
refractive
index real
part n and
imaginary
part for Cu
Source: n,k Database at Ioffe Institute (http://www.ioffe.ru/SVA/NSM/nk/index.html#Metals)
17
APh 114b Lecture 12
Solid State Physics
Gallium
Arsenide
Free carrier
absorption
versus
wavelength at
different
doping levels.
n-Ge.
T 296 K.
for conduction
electron
concentration:
1. 1.310
17
cm
-3
;
2. 4.910
17
cm
-3
;
3. 10
18
cm
-3
;
4. 410
18
cm
-3
Silicon
Absorption coefficient versus photon energy at
different doping levels, n-Si, T = 300 K
1 -- 1.410
16
cm
-3
,
2 -- 810
16
cm
-3
,
3 -- 1.710
17
cm
-3
,
4 -- 3.210
17
cm
-3
,
5 -- 6.110
18
cm
-3
,
6 -- 110
19
cm
-3
.
18
APh 114b Lecture 12
Solid State Physics
Germanium
Free carrier
absorption
versus
wavelength at
different
doping levels.
n-Ge.
T = 300 K.
for conduction
electron
concentration:
(cm
-3
):
1. 8.010
17
;
2. 4.810
18
;
3. 1.3510
19
;
4. 1.810
19
;
5. 3.610
19
.
Indium Antimonide
Absorption coefficient versus photon energy at
different doping levels, n-InSb, T = 130 K
no (cm
-3
):
1. 6.610
13
;
2. 7.510
17
;
3. 2.610
18
;
4. 610
18
;
19
APh 114b Lecture 12
Solid State Physics
Optical Transition Rate: Fermi's Golden Rule in Solids
Time-dependent perturbation theory enables us to treat excitations which
explicitly depend on time. This is of importance in discussing optical
transitions because we can view the solid, with unperturbed Hamiltonian
H
0
,as being perturbed by the time-dependent electromagnetic field H(t)
generated by an incident photon flux
H
solid
H
0

photon
H
t
If the matrix element <m|H|l> connects an initial state m to a final state l, a
very important result of time-dependent perturbation theory is Fermi's
Golden Rule, which states that the transition rate
ml
is given by
ml

2m
|m|H
|l|
2
oE
l
E
m
o
where is the photon energy o
20
APh 114b Lecture 12
Solid State Physics
In principle, we need to calculate the matrix element <m|H|l> for each set
of initial and final states, m and l. In practice in solids, the matrix element
does not vary strongly as we consider successive band states, so the total
transition rate in a solid can be approximated as

m,l

ml

m,l
2m
|m|H
|l|
2
oE
l
E
m
o
2m
|m|H
|l|
2
m,l
oE
l
E
m
o
Let's assume the state m and l are valence and conduction band
states, respectively. Then We thus define
the joint density of states as
m|H
|l v|H
|c.
vc
o
2
8m
3
dkoE
c
k E
v
k o
21
APh 114b Lecture 12
Solid State Physics
and thus
2m
|H
vc
|
2
vc
E
c
k E
v
k o
Because of the form of the joint density of states, optical measurements
provide information about particular points in the Brillouin zone, these
usually being high symmetry points near band extreme. To see this, let's
cast the integral above over dk in terms of energy. Let's introduce an
energy surface S in k-space such that E
c
E
v
o.
dk dSdk
n
We can also write:
dE |
k
E| dk
n
so
|
k
E
c
E
v
| dk
n
dE
c
E
v
and thus
dk dSdk
n
dS
dE
c
E
v
|
k
E
c
E
v
|
22
APh 114b Lecture 12
Solid State Physics
The joint density of states is then
vc
o
2
8m
3

k-space
dSdE
c
E
v
oE
c
E
v
o
|
k
E
c
E
v
|
Integrating over gives dE
c
E
v
vc
o
2
8m
3

k-space
dS
|
k
E
c
E
v
|
E
c
E
v
o
At critical points where especially large
contribution to occur. Critical points fall into 4 categories
depending on whether the band separations are increasing or decreasing
as we move away from the critical point. If we expand
in a Taylor series around
k
E
c
E
v
0,
vc
o
E
c
k E
v
k
k
0
,
E
c
k E
v
k E
g
k
0

i1
3
a
i
k
i
k
oi
23
APh 114b Lecture 12
Solid State Physics
where 3 indicate the directions x,y,z, the coefficients a
i
are
a
i

2
k
i
2
E
c
k E
v
k.
Consider example of an M
o
singularity, assuming simple
parabolic bands:
E
c
k
E
g
2

2
k
2
2m
c
E
v
k
E
g
2

2
k
2
2m
r
24
APh 114b Lecture 12
Solid State Physics
So
E
c
k E
v
k E
g

2
k
2
2
1
m
c

1
m
v
E
g

2
k
2
2m
r
Where the reduced effective mass is

1
m
r

1
m
c

1
m
v
Now the gradient of is

k
E
c
E
v

2
k
m
r
E
c
E
v
and the joint density of states is
vc
o
2
8m
3

dS
|
k
E
c
E
v
|
E
c
E
v
o

2
8m
3
4mk
2
2
k
m
r
E
c
E
v
o
m
r
m
2
2
k
25
APh 114b Lecture 12
Solid State Physics
To evaluate k, note that
E
c
E
v
o E
g

2
k
2
2m
r
so
k
2m
r
2
o E
g
1/2
and finally the joint density of states for a 2 band system with spherical, parabolic
bands is
vc
o
1
2m
2
2m
r
2
3/2
o E
g
1/2
1
APh 114b Lecture 13
Solid State Physics
Interband Transitions
While at low frequencies, the primary electronic and optical excitation
mechanism is via free carrier conduction, as the photon energy approach
the energy gap, a new conduction process can occur. An interband
transition occurs when a photon excites a carrier from an occupied state in
the valence band to an unoccupied state in the conduction band.
Points to consider:
1.We expect this process to have an energy
threshold equal to the energy gap.
2.Transitions are either direct (i.e., conserve crystal
momentum) or indirect (a phonon is involved, so
).
Note that but .
3.The transition rate depends on the coupling
between valence and conduction bands via the
momentum matrix element
k
valence
k
conduction
q
phonon
q
photon
G
q
phonon
G
Ex: Yellow light 600nm z k
2m
z
10
5
cm
1
, G 10
8
cm
1
.
2
APh 114b Lecture 13
Solid State Physics
Electromagnetic Hamiltonian
H
p
e
c
A
2
2m
Vr
p
2
2m
Vr
e
mc
A p
e
2
A
2
2mc
2
Since optical fields usually represent a perturbation, we keep only the
linear term, so
H

e
mc
p A
4.The Pauli Exclusion Principle requires that an interband transition
occurs from an occupied state below the Fermi level to an unoccupied
state above the Fermi level.
5.Photons of a certain energy are more effective in producing interband
transitions if the energy separation between the bands is nearly constant
over many k values, where the two bands have a large joint density of
state.
3
APh 114b Lecture 13
Solid State Physics
Large joint density of states from
E
c
E
v
o near k 0
Large joint density of states
from A to A.
4
APh 114b Lecture 13
Solid State Physics
Interband transitions contribute to the conductivity and to the dielectric
constant. For a current density in the direction due to a -directed
field, the interband conductivity is:
j
into
o
into[
E
[
where
o
into[

e
2
m
2
sum over all valence and conduction states
i,j
fE
i
fE
j
E
i
E
j
i|P
o
|jj|P
[
|i
io
1
t

i
E
i
E
j
The dielectric constant is then

c c
core

4mi
o
o
free carrier
o
int
5
APh 114b Lecture 13
Solid State Physics
Connection between Optical Transitions and Effective Mass:
kp Perturbation Theory
We will see that because of the connection between optical transition
matrix elements between valence and conduction states <v|p|c> and
the effective mass (1/)(E)/(k), band diagrams provide important
information about optical properties. Conversely, experimental
information about optical properties can provide information about E(k).
If we have a Bloch solution to the Schrodinger equation
nk
r e
ikr
u
nk
r
and thus
p
2
2m
Vr e
ikr
u
nk
r E
n
ke
ikr
u
nk
r
we also have
pe
ikr
u
nk
r e
ikr
p hku
nk
r
6
APh 114b Lecture 13
Solid State Physics
Therefore the equation for is u
nk
r
p
2
2m
Vr
k p
m

2
k
2
2m
u
nk
r E
n
ku
nk
r
or
H
0
p
2
2m
Vr
H
k p
m
u
nk
r
E
n
k
E
n
k

2
k
2
2m
u
nk
r
H
0
H
u
nk
r E
n
ku
nk
r
7
APh 114b Lecture 13
Solid State Physics
If we know the solution at a point k=k
o
, kp perturbation theory allows
us to find the E(k) for states near k
o
. Let's first set k=k
o
=0
E
n
k E
n
0 u
n,0
|H
|u
n,0

n
u
n,0
|H
|u
n
,0
u
n
,0
|H
|u
n,0
E
n
0 E
n
0
The first order term vanishes by inversion symmetry. (H is odd, u
n,0
is either even or odd). Since
H

k p
m
the matrix element is
u
n,0
|H
|u
n
,0

m
k u
n,0
|p|u
n
,0
8
APh 114b Lecture 13
Solid State Physics
Now assume
1.The bands n and n are strongly coupled (close to each other and far
away from other bands)
2.Interband transitions occur across an energy gap E
g
n.b., again:
E
n
k E
n
0

2
k
2
2m
E
n
k E
n
0

2
m
2
k
o
k
[
|v|p
o
|cc|p
[
|v|
E
g
E
n
k E
n
0

2
k
2
2m

2
m
2
k
o
k
[
define effective mass by
E
n
k

2
k
2
2m
E
n
0
9
APh 114b Lecture 13
Solid State Physics
then
1
m
o[

o
o[
m

2
m
2
|v|p
o
|cc|p
[
|v|
E
g
1
APh 114b Lecture 14
Solid State Physics
Optical Absorption in Solids
Optical Absorption
In optical absorption, we are concerned with the light intensity I(z) after
traversing a thickness of material z relative to the incident intensity I
o
.
The optical absorption coefficient is defined by
I(z) I
0
e
oz
o 2
ok
c
where
The factor of 2 comes from the fact that intensity is proportional to the
square of the field amplitude. We now consider various forms of
optical absorption including:
i) free carrier absorption in
a) semiconductors
b) metals at low frequencies
ii) direct interband transition
iii) indirect interband transitions
2
APh 114b Lecture 14
Solid State Physics
For free carrier absorption, we have
c c
core
i
4mo
o
,
o
ne
2
t
m
(i iot)
where
and we also know the plasmon frequency
o
p

4mne
2
m
c
core
For semiconductors,
core
is typically large and the free carrier
contribution is small at infrared and visible frequencies. For this
frequency range, >> 1, so
c c
core

i4mne
2
t(1 iot)
m
o 1 (ot)
2
c
core

ic
core
o
p
2
t(1 iot)
o 1 (ot)
2
Free Carrier Absorption in Semiconductors
3
APh 114b Lecture 14
Solid State Physics
c c
core

ic
core
o
p
2
t
2
(1 iot)
o
3
t
3

c
core
o
p
2
o
2
, for ot 1
Complex Dielectric Constant for Semiconductors
In the infrared and visible >>
p
, so
c
1
c
core
, c
2
small
c
1
n
2
k
2
c
2
2nk 2 c
core
k
c
2

c
core
o
p
2
t
2
o
3
t
3
So n is large; k is small.
4
APh 114b Lecture 14
Solid State Physics
The absorption coefficient is:
o(o)
2ok
c

2o
c
c
core
o
p
2
2 c
core
o
3
t
o(o)
c
0
o
p
2
co
2
t
So in semiconductors, for the free carrier contribution to optical absorption
o(o) is proportional to
1
o
2
or to z
2
; ot 1, o o
p
Free Carrier Absorption in Metals
For typical metals << 1, and <<
p
so
c
core

4mo
o
and n . Then both n and are large and
5
APh 114b Lecture 14
Solid State Physics
c i
4mo
o
i
4mne
2
t
m
c
2
2nk 2k
2
Complex Dielectric Constant for Metals
so
k
2mne
2
t
m
o
The absorption coefficient is:
o
2ok
c

8mone
2
t
m
c
2
So for metals the free carrier contribution to the optical absorption
o(o) is proportional to o
6
APh 114b Lecture 14
Solid State Physics
Unless the metal is a very thin semi-transparent film, it is more
convenient to measure reflectivity than absorption. The reflectivity is
R
(n 1)
2
k
2
(n 1)
2
k
2
n
2
2n 1 k
2
n
2
2n 1 k
2
1
4n
n
2
k
2
2n 1
1
4n
n
2
k
2
1
2
n
but for metals
n k
2mne
2
t
m
o
7
APh 114b Lecture 14
Solid State Physics
and the free contribution to reflectivity of the metal surface is
R 1 2
m
o
2mne
2
t
, o o
p
which is the Hagens-Rubens relation, valid for metals and
semiconductors below their plasma frequencies.
Absorption from Direct Interband Transitions
The absorption coefficient can be defined as
o
(o) (transition probability / volume time)
incident electromagnetic flux
8
APh 114b Lecture 14
Solid State Physics
The incident electromagnetic flux is:
S
c
2(4m)
ReE H
)
For chosen vector potential A,
E
1
c
A
t

io
c
A
For the choice of a Coulomb gauge with = 1 but c c
0
H B A i A
so
S
c
8m
Re
io
c
A (i A)
o
8m
Re(A - A) (A - )A)
Direct Interband Transitions
9
APh 114b Lecture 14
Solid State Physics
For a transverse plane wave, , so
A - 0
S
o
8m
no
c
|A|
2
||
The transition probability is calculated from Fermi's Golden Rule

2m
|H
vc
|
2
vc
(ho); H

e
mc
A - p
At finite temperatures, we consider also the state occupancy factors
given by the Fermi functions:
absorption
f(E
v
)|1 f(E
c
)]
emission
f(E
c
)|1 f(E
v
)] |f(E
v
) f(E
c
)]
|H
vc
|
2
e
mc
2
|A|
2
|v||c)|
2
10
APh 114b Lecture 14
Solid State Physics
Thus for direct interband transitions
o
(o)
2m
(
e
mc
)
2
|A|
2
|v||c)|
2
vc
(ho) |f(E
v
) f(E
c
)]
o
8m
(
no
c
)|A|
2
o
16m
2
e
2
m
2
cno
|v||c)|
2
vc
(ho)|f(E
v
) f(E
c
)]
Our previous result for a two-band model for the joint density of states
(JDOS) for a three-dimensional bulk semiconductor with spherical
parabolic bands was
vc
(ho)
1
2m
2
2m
r
2
3/2
|ho E
g
]
1/2
1
m
r

1
m
n

1
m
p
11
APh 114b Lecture 14
Solid State Physics
Assume low temperatures, so
f(E
v
) 1, f(E
c
) 0
We can estimate given the effective mass from the
effective mass theorem
|v||c)|
2
1
m

1
m

2
m
2
|v||c)|
2
E
g
In a typical semiconductor (especially for narrow gap semiconductors)
m
m so
1
m

2
m
2
|v||c)|
2
E
g
or
|v||c)|
2
mE
g
2
m
m

12
APh 114b Lecture 14
Solid State Physics
Putting everything together,
o
(o)
2m
(
e
mc
)
2
|A|
2
|v||c)|
2
vc
(ho) |f(E
v
) f(E
c
)]
o
8m
(
no
c
)|A|
2
o(o)
( o)
2 m
e
2
m
2
c
2
m
2
E
g
2 m
1
2 m
2
2 m
r
2
(hoE
g
)
1/2
no
2
8mc
o(o)
8
c
e
2
2
m
r
|o E
g
]
1/2
o
3D semiconductor; spherical
parabolic bands
so
is proportional to
o(o)
oE
g
o
13
APh 114b Lecture 14
Solid State Physics
Indirect Interband Transitions
Consider excitation from a valence band state at k=0. To conserve
crystal momentum it's necessary, for a indirect energy gap, to absorb
or emit a phonon of wave vector equal to the difference in k between
valence and conduction band extrema. If we consider photon
energies comparable to the fundamental energy gap
there is not enough energy to excite a direct optical transition at k=0.
In order not to violate energy conservation, the uncertainty principle
requires that the system not exist in the intermediate state longer than
a virtual photon state lifetime
l
where
ho E
c
E
v
,
t
l

E

|E
g
o|
k
E
E
c
photon
phonon
virtual
states
E
v
overall indirect
transition
During that time, an electron phonon
scattering event occurs, putting the electron
into its final state in the conduction band. So
for example, process (2) is more likely than
Process 2
Process 1
14
APh 114b Lecture 14
Solid State Physics
process (1) because the virtual state lifetime is longer. Indirect
transitions involve loss or gain of the phonon energy
o E
f
E
i
o
q
, (phonon absorption)
o E
f
E
i
o
q
, (phonon emission)
Considering total energy conservation gives
o E
g
o
q

2
(k
n
k
c
)
2
2m
n

2
k
p
2
2m
p
E
g
o
q
(E
n
E
c
)
where . k
n
k
p
q
k
E
E
c
E
p
E
v
E
n
- E
c
15
APh 114b Lecture 14
Solid State Physics
The density of final states is proportional to
c
(E
n
) (E
n
E
c
)
1/2
(o E
g
E
p
o
q
)
1/2
k
E
E
c
E
p
E
v
E
n
- E
c
k
n
crystal momentumof excited electron
k
p
crystal momentumof excited hole
E
p
can vary from:
E
p
0 maximumkinetic energy given to electron
E
n
E
c
0 maximumkinetic energy given to hole
16
APh 114b Lecture 14
Solid State Physics
Define the range of
The density of initial states is proportional to
E
p
by o o E
g
o
q
v
(E
p
) E
p
1/2
Thus the effective density of states for phonon absorption is formed
by summing over all E
p
values that conserve energy.
(o)

0
o
c
(E
n
)
v
(E
p
) dE
p
o
2
m/8
0
o
(o E
p
)
1/2
E
p
1/2
dE
p
17
APh 114b Lecture 14
Solid State Physics
So the joint density of states is
Finally, the phonon absorption probability is
(o)
m
8
(o E
g
o
q
)
2
.
n(o
q
)
1
e
o
q
/kT
1
so the absorption coefficient for indirect interband transitions for the
phonon absorption process is
o
a
(o)
C(o E
g
o
q
)
2
e
o
q
/kT
1
18
APh 114b Lecture 14
Solid State Physics
For phonon emission, probability is proportional to
n(o
q
) 1
n(o
q
) 1
1
e
o
q
/kT
1
1
1
1 e
o
q
/kT
so the indirect interband absorption coefficient for phonon emission
processes is
o
e
(o)
C(o E
g
o
q
)
2
1 e
o
q
/kT
Plotting the square root of
absorption gives:
19
APh 114b Lecture 14
Solid State Physics
GaAs Interband Transitions
Temperature Dependence Wide Energy Range
Doping Dependence n type
Doping Dependence p type
20
APh 114b Lecture 14
Solid State Physics
Si Interband Transitions
Temperature Dependence
Doping Dependence
n type
Wide Energy Range
21
APh 114b Lecture 14
Solid State Physics
Ge Interband Transitions
300 K Doping
Dependence
n type (As)
Wide Energy Range
22
APh 114b Lecture 14
Solid State Physics
Diamond Interband Transitions
23
APh 114b Lecture 14
Solid State Physics
GaN (Wurtzite) Interband Transitions
77K
5 K
300 K
300 K Doping
Dependence
1 - 2 x 10
16
cm-3,
2 - 2.8 x 10
17
cm-3,
3 - 5 x 10
17
cm-3,
4 - 2.3 x 10
18
cm-3,
5 - 2 x 10
19
cm-3
1
APh 114b Lecture 15
Solid State Physics
Kramers-Kronig Relations
In an optical absorption experiment, we measure only the imaginary part
of the complex index of refraction. To get a complete description of
complex dielectric media, we need to know both the real part
1
and the
imaginary part
2
. The Kramers-Kronig relations are integral transform
relations that connect the real and imaginary part of the index of
refraction. They are:
1
1
2
P

0
2

2
d
2

2
P

0
2

2
d
where P denotes the principal value of the integral. These relations

imply that if either of the functions
1
or
2
is known for all frequencies
the other one can be found.
2
APh 114b Lecture 15
Solid State Physics
Owing to the form of the Kramers-Kronig relations, it is apparent that
the main contribution to
1
() comes from
2
() near . So
although we should in fact know
2
() for all to obtain
1
() ,we
can get a good approximation to the optical response by looking at
finite frequency intervals. (Aside to E.E.'s: this is analogous to a
Hilbert transform that relates the real and imaginary parts of the
system function in a casual, linear system.)
Let's apply causality to polarization in a solid:
D E E 4P
or
P
1
4
E E
where () defines polarizability and P is polarization/volume.
3
APh 114b Lecture 15
Solid State Physics
We can write then the system function as:

r
i
i
Let be an impulse electric field at t=0. Then by

definition of a -function, we have:
Et E
0
t
Et E
0
t
E
0
cost d
The response to this impulse field yields an in-phase term
proportional
r
(), and an out-of-phase term proportional to
i
()
Pt
E
0
r
cost
i
sint d
Since P(t) is casual and bounded, the integral of is well-
behaved along the contour C as R, and no contribution to the integral
is made along C in the upper half plane.
e
it
4
APh 114b Lecture 15
Solid State Physics
Further, causality dictates that P(t) vanish
for t<0, requiring () to have no poles in
the upper half plane. To get an explicit
expression for (), we have to generate a
pole on the real axis. Then we can isolate
the behavior of () at a point
o
. Since
() has no poles in the upper half-plane,
the function will have a single
pole at =
o
, so an integral around the
closed contour
0
vanishes:

d

0
0
5
APh 114b Lecture 15
Solid State Physics
We can take this piece by piece:
0

d

0

C
d

0

R
d

0

C
d

0

R
d

0
The contribution over C vanishes since () remains finite while
as R. Along contour C, use Cauchy's theorem to
obtain
1
0
0
lim
0

C
0
d i
residue
(negative sign (-): clockwise path on contour)

Define the principal value P of the integral as
lim
R
0
d

0

R
d

0
P
d

0
6
APh 114b Lecture 15
Solid State Physics
Since
d
0
0,
0
i
i
0

1
i
P

r
i
i

0
d
Equating real and imaginary parts,
0

1

0
d
r
is even
0

1

0
d
i
is odd
To get integrals over positive frequencies, use even- and odd-ness,
and multiply by

0
0

1

i

0
2

0
2
d
2
2

0
2
d
0

1

r

0
2

0
2
d
2
2

0
2
d
7
APh 114b Lecture 15
Solid State Physics
Can subtract out the singularity at
o
by writing
0
i
i
0

1
i

0

0

0
d
Using even and odd-ness, we get
0

2

i

0
2

0
2
d
and
0

2
r

0
2

0
2
d
To get the dielectric constant, use 1 4,
to get
0
1
2

2

0
2

0
2
d
0

2
1

0
2

0
2
d
8
APh 114b Lecture 15
Solid State Physics
Light Scattering
Light scattering is very useful for studying fundamental excitations in
solids because light can be scattered inelastically (so incident and
scattered photons have difference frequencies). Because the scattered
intensity is often ~10
-6
-10
-7
of the incident intensity, laser excitation is
very useful (but not necessary).
Basic Picture:
Energy conservation:

0

q
Momentum conservation:
k k
0
q
Thus denotes either creation (-) or annihilation (-)of an excitation
9
APh 114b Lecture 15
Solid State Physics
Light Scattering
If , then Measurement of the
angular dependence of (q) gives dispersion relations near q = 0.
If q an acoustic phonon Brillouin scattering
If q an optical phonon Raman scattering
q

0
|k| |k
0
|
|q| 2|k
0
| sin

2
.
In an applied electric field,
E E
0
sin
0
t
a polarization results due to solid's polarizability, . The total
polarizability and polarization are
P E

atomic
0

lattice vibration
1
sin
q
t
P E
0
0

1
sin
q
t sin
0
t
E
0
Rayleigh
0
sin
0
t
Stokes
1
2
cos
0

q
t
Antistokes
1
2
cos
0

q
t
10
APh 114b Lecture 15
Solid State Physics
Light Scattering
For Brillouin scattering, light is inelastically scattered by acoustic
phonons. Near q = 0, we can make the Debye approximation that
Brillouin Scattering
q
v
phonon
q
For the Stokes scattering case (- sign),
q k k
o

nv
phonon
c
k
o
2
k
2
2kk
o
cos
k
o
k
2nv
phonon
c
1 cos
2
q

o

2n
o
v
phonon
c
sin

2
q
(q)
acoustic
optical
11
APh 114b Lecture 15
Solid State Physics
Raman Scattering
In Raman scattering, light is inelastically scattered by optical phonons.
Near q = 0,
q
is approximately constant for the optical phonon branch.
1
APh 114b Lecture 16
Solid State Physics
Optical Studies of Lattice Vibrations: Polaritons
We now examine the dynamics of lattice vibrations in an optical field.
Consider the equation of motion of an ion in the solid with normal mode
displacement coordinate r
Maxwell's equations also gives
H
1
c

t
D
1
c

t
E 4m

t
P
m

2
t
2
r kr eE o
2
mr; Et E
o
e
iot
E
1
c

t
H i
o
c
H
If the total number of optical modes/volume is N
opt
,and the electron
density is N
el
, then there is both an ionic and electronic contribution to
the polarization.
2
APh 114b Lecture 16
Solid State Physics
P N
opt
er N
el
oE
So now we have 4 equations in 4 unknowns: E, H, r, and P. We
assume two degrees of freedom: the ionic system and the electronic
system, and we assume that the polarizations of each are
independent. The electric field in all of these equations is the applied
electric field, as we assume that local field corrections can be
neglected (e.g., not true in ferroelectric materials). The plane wave
solutions for transverse waves are:
Et E
o
e
iotkz
H
y
H
y
o
e
iotkz
P
x
P
x
o
e
iotkz
r
x
r
x
o
e
iotkz
3
APh 114b Lecture 16
Solid State Physics
Typical values for k=(2/)=100 cm
-
which yields ~60 microns. Substitution
yields
ikH
y
i
o
c
E
x

4mio
c
P
x
0
ikE
y
i
o
c
H
y
0
o
2
r
x

kr
x
m

e
m
E
x
0
P
x
N
opt
er
x
N
el
oE
x
0
4
APh 114b Lecture 16
Solid State Physics
det
o
c
k
4mo
c
0
k
o
c
0 0
e
m
0 0 o
2
k
m
N
el
o 0 1 N
opt
E
0
E
x
H
y
P
x
r
x
In matrix form, we get
which yields a 4th order equation:
o
4
1 4mN
el
o o
2
c
2
k
2
k
m

4mN
opt
e
2
m

4mN
el
ok
m
c
2
k
2 k
m
0
5
APh 114b Lecture 16
Solid State Physics
c
high frequency (optical frequency) dielectric constant

c
1 4m
P
E
Let's define
From the equation of motion
kr
m

e
m
E o
2
r
we can conclude that the ionic displacement is small (otherwise

2
t
2
r
at high frequencies. Thus the ionic motion contributes less and less to
the polarization as frequency increases, so the electronic contribution
dominates and
c
1 4mN
el
o
6
APh 114b Lecture 16
Solid State Physics
Dielectric Constant at Low Frequencies
At = 0, the equation of motion gives r=(eE)/), so the polarization at = 0 is
P
o

N
opt
e
2
k
N
el
o E
and
o
1 4m
N
opt
e
2
k
N
el
o
At a general frequency ,
1 4m
N
opt
e
2
k mo
2
N
el
o
7
APh 114b Lecture 16
Solid State Physics
Let's also define a frequency
o
T

k
m
which depends on the harmonic forces in the crystal and not the applied
field. Substitution yields
o
4
c
o
2
c
2
k
2
o
T
2
c
o
c
2
k
2
o
T
2
0
The solutions to this equation are
o
2
1
2c
o
T
2
c
o
c
2
k
2

1
4c
2
o
T
2
c
o
c
2
k
2
2
o
T
2
k
2 c
2
c
Each solution is 2-fold degenerate, since E can be chosen in an arbitrary

direction perpendicular to the propagation vector. This coupled excitation
of a transverse optical phonon to electromagnetic radiation is called a
polariton. In the longitudinal direction the light and the lattice vibrations do
not couple.
8
APh 114b Lecture 16
Solid State Physics
We get a total of 6 modes for the polariton, since there are 3 optical
modes and 3 electromagnetic modes.
1. At small k vectors we get two solutions:
i) a positive solution
(k 10
4
cm
1
o
2
1
c
o
T
2
c
o
c
2
k
2
o
T
2
c
o
c
o
L
2
At k = 0, this is a phonon-like mode with finite frequency
(optical phonon branch).
ii) a negative solution
o
2
1
2c
o
T
2
c
o
c
2
k
2

1
2c
o
T
2
c
o
c
2
k
2
1
1
2
o
T
2
c
2
k
2
/c
1
2c
o
T
2
c
o
c
2
k
2
which gives
o
2
c
c
o
k
, a photon-like mode
9
APh 114b Lecture 16
Solid State Physics
At large k vectors (k 10
5
cm
-
),we get
i) a positive solution:
o
2
1
c
o
T
2
c
o
c
2
k
2

c
2
k
2
c
which is a photon-like mode with

o
2
c
c
k
ii) and a negative solution:
o
2
o
T
2
c
o
c
2
k
2
o
T
2
c
o
c
2
k
2
o
T
2
which is a phonon-like mode.
(k)
k
photon-like
mode
photon-like
mode
L
phonon-like
mode
phonon-like
mode
10
APh 114b Lecture 16
Solid State Physics
In the course of our analysis we have also obtained the
Lyddane-Sachs-Teller relation
o
L
2
o
T
2

c
o
c
Now let's look at (): Note that

P N
opt
er N
el
oE
mo
2
r kr eE
giving
r
eE
k mo
2

E
e
m
k
m
o
2

E
e
m
o
T
2
o
2
11
APh 114b Lecture 16
Solid State Physics
and
P
E

co 1
4m

N
opt
e
2
m
o
T
2
o
2
N
el
o
N
opt
e
2
m
o
T
2
o
2

c
1
4m
The dielectric constant is:
co c

4mN
opt
e
2
m
o
T
2
o
2
We note that
c
electronic polarizability
4mN
opt
e
2
m
o
T
2
o
2
lattice polarizability
12
APh 114b Lecture 16
Solid State Physics
The resonant frequency
T
= is called the restrahl frequency. The
frequency at which the dielectric constant vanishes is
L
= .
c
o
c

4mN
opt
e
2
m
o
T
2
so
co c

c
o
c
1
o
2
o
T
2
2
co
L
0 c

c
o
c
o
L
2
o
T
2
1
o
L
2
o
T
2

c
o
c
We note that
o
L
LO phonon frequency
o
T
TO phonon frequency
13
APh 114b Lecture 16
Solid State Physics
The has
1) a zero at =
L
2) a singularity at =
T
()
/
T
L
0
Real {()}
Im{()}
Reflectivity of a thick crystal of
NaCl at several temperatures; the
300
o
K values of
T
= 61 microns
and
L
= 38 microns
14
APh 114b Lecture 16
Solid State Physics
Recapping and looking at both the real and imaginary parts of the
complex wavevector, the polariton dispersion relations have a
1) Lower branch that looks photon-like
2) a singularity in dielectric function at =
T
3) A restrahl band between
L and
T
characterized by high
reflectivity
(k)

L
Real { k}
T
(k)

L
Im { k}
T
photon-like
mode
photon-like
mode
phonon-like
mode
Restrahl band
15
APh 114b Lecture 16
Solid State Physics
Luminescence
Luminescence refers to the emission of radiation by a solid in excess
of the amount expected in thermal equilibrium. As a non-equilibrium
phenomenon, luminescence requires excitation by light, electron
beams, current injection, etc.
photoluminescence: excitation by optical radiation
cathodoluminescence: excitation by electron bombardment
electroluminescence: excitation by current injection
Possible Transition:
Note that not all transitions are radiative. A luminescent material is one
where radiative transition predominate over nonradiative transitions.
16
APh 114b Lecture 16
Solid State Physics
Luminescence
Relation between Absorption and Emission
For a radiative emitter, the rate of radiative recombination equals the
rate of generation of electron-hole pairs by optical absorption
Rodo Pogodo
where R() = radiative recombination rate, P() = absorption
probability/time, g() = density of photons/frequency.
Planck Radiation Law:
godo
o
2
n
3
m
2
c
3
1
e
o/kT
1
do
17
APh 114b Lecture 16
Solid State Physics
Luminescence
where n n(). The absorption probability is related to the photon
lifetime in the solid:
Po
1
to
The mean free path (1/(()) is the product of the velocity v=(c/n)
and the lifetime (), so
to
1
oov

n
oovc
Po oo
c
n
18
APh 114b Lecture 16
Solid State Physics
Luminescence
The absorption/emission relation is then:
Rodo
ooo
2
n
2
m
2
c
2
e
o/kT
1
do
19
APh 114b Lecture 16
Solid State Physics
Excitons
An exciton denotes a system consisting of a bound electron and hole,
which are bound together by coulomb interaction. Since the exciton
consists of an electron and a hole, the binding is attractive and thus
an exciton represent a lower energy state than a free electron and a
free hole. Let's analyze the exciton spectrum for interband
transitions, using the effective mass approximation.
The Schrodinger equation for exciton wave packets is
written, using the effective mass approximation,
p
n
2
2m
n

p
p
2
2m
p

e
2
c|r
n
r
p
|
E
20
APh 114b Lecture 16
Solid State Physics
Excitons
We now introduce a change of coordinates:
a difference coordinate:
and a center-of-mass coordinate:
The Schrodinger equation for exciton wave packets is
written, using the effective mass approximation,
p
n
2
2m
n

p
p
2
2m
p

e
2
c|r
n
r
p
|
E
r |r
n
r
p
|

m
n
r
n
m
p
r
p
m
n
m
p
21
APh 114b Lecture 16
Solid State Physics
Excitons
The transformation of the kinetic energy of the electron and hole yields
a kinetic energy of:
i) center of mass motion and
ii) relative motion
p
n
2
2m
n

p
p
2
2m
p

p
p
2
2m
n
m
p

p
r
2
2j
where
j

m
n
m
p
m
n
m
p
Now separate the exciton wavefunction into center of mass and

relative motion terms, G() and F(r), respectively:
r
n
, r
p
FrG
22
APh 114b Lecture 16
Solid State Physics
Excitons
So
p
p
2
2m
n
m
p

p
r
2
2j

e
2
cr
FrG EFrG
Let's focus first on the center of mass component to obtain
p
p
2
2m
n
m
p
G zG
with eigenvalues
z

2
k
2
2m
n
m
p
Note that this is a free particle energy, in which the "potential" is the
exciton which is free to move about the crystal.
23
APh 114b Lecture 16
Solid State Physics
Excitons
Now let's examine the relative motion:
p
r
2
2j

e
2
cr
Fr E

2
k
2
2m
n
m
p
Fr
Note that this is a hydrogenic Hamiltonian, whose eigenvalues are
E
r

2
k
2
2m
n
m
p
e
4
2
2
c
2
n
2
Note that since
is smaller than m
n
, the
exciton binding energy is less than the
impurity ionization energy. For, e.g., Ge, the
exciton binding energy is about 3 meV, and
the Bohr radius is
r
c
2
e
2
j

0. 516
0. 05
160
Hydrogenic exciton
states in Cu
2
O
1
APh 114b Lecture 17
Solid State Physics
Photoemission
Photoemission is one of the most important techniques for mapping
energy band structure in solids. The incident excitation is a photon
and the excited particle is an electron ejected into vacuum and then
measured with an electron spectrometer. This provides a
measurement of the dielectric function. Photoemission is also
capable of measuring chemical shifts due to charge transfer which
alters the binding energy of photoelectrons. The photoemission
current can be written in the form
I IE,
e
,
e
, , ,
p
,
p
,
p
where E is the electron energy, are the polar and azimuthal

electron scattering angles, is the electron spin, is the photon
energy,
p
is the photon polarization, and
p
and
p
are the polar and
azimuthal incident photon angles.

e
,
e
2
APh 114b Lecture 17
Solid State Physics
Photoemission
Different photoemission techniques
measure different variables:
1. Energy Distributions I=I(E) with
all other variables held constant
2. Constant Initial State
Spectroscopy where
is kept constant, i.e.,
3. Constant Final State
Spectroscopy where
4. Angle Resolved
Photoemission, where all the
angles are allowed to vary but
photon energy and electron
kinetic energy are fixed.
E
I IE,
Econstant
I I
The photoelectric yield is the measurement of
the number of electrons emission per incident
photon, and energy distributions are measured
by spectroscopic resolution of the
photoemitted electrons in an electrostatic or
magnetic analyzer.
3
APh 114b Lecture 17
Solid State Physics
Photoemission
Historically, after ultraviolet
photon sources were developed,
ultraviolet photoemission
spectroscopy (UPS) was first
used to measure the work
function of solids and
development of X-ray sources
enabled study of core levels in
molecules and solids. In
photemission, light is absorbed
within a optical skin depth (~ few
microns) at ultraviolet or X-ray
energies, but electrons are
photoemitted into vacuum from a
much shallower electron escape
depth corresponding to the mean
free path for electron scattering.
Universalelectron escape depth
4
APh 114b Lecture 17
Solid State Physics
Photoemission
Angle-resolved photoemission can be used
to directly map out energy band structure by
Resolving photoelectron energy and momentum
5
APh 114b Lecture 17
Solid State Physics
Photoemission
z
E
n
e
r
g
y
E
o
(Vacuum Level)
e
h h
c
Measurements of photoemission spectra
provide information about band structure
through analysis of the structure in the
energy-dependent photoelectric yield.
For a metal surface oriented in the x-y
plane, with photoelectron momentum
normal to the surface p
z
,and electron can
escape and be photoemitted in vacuum if
p
z
2
2m
E
o
where E
o
is the vacuum level. The work function e represents the
minimum kinetic energy for electron escape from the solid into vacuum.
The photocurrent is proportional to the number of electrons escaping
from the solid surface
I e
o
p
z
np
z
Dp
z
dp
z
where D(p
z
) is the escape probability for
an electron of momentum p
z
and n(p
z
) is
the electron density.
metal surface
6
APh 114b Lecture 17
Solid State Physics
Photoemission
The electron density is then
np
z

2
h
3

dp
x
dp
y
1
e
p
2
2m
E
F
/kT
1
In a semiconductor, there is an energy gap,
and the Fermi energy lies in the energy gap.
The work function is also defined relative to
the Fermi energy but the threshold for
photoemission is now e + . The electron
affinity is a Fermi energy-independent
parameter equal to the energy difference
from the bottom of the conduction band the
the vacuum level. The photoemission
threshold is:
z
E
n
e
r
g
y
E
o
h

E
F
E
g
e
semiconductor surface
p
z
2
2m
e E
g
7
APh 114b Lecture 17
Solid State Physics
Photoemission
In semiconductors, valence band photoemission will dominate until the
photoemitted electrons are themselves energetic enough to form
electron-hole pairs via collisions with other electrons. There is then a
secondary interband transition threshold
p
z
2
2m
E
g
E
g
2E
g
Energy
Yield
+E
g
+2E
g
This illustrates that both primary photoelectrons and
inelastically scattered photoelectrons contribute to the
photoelectron yield.
The primary photoemission yield depends on three
factors
IE, I
p
E, I
s
E,
I
p
E, PE, TEDE
8
APh 114b Lecture 17
Solid State Physics
Photoemission
again, on this page,
I
p
E, PE, TEDE
where P(E,) is the energy distribution of photoelectrons and T(E) is
the transmission function of excited electrons in terms of the optical
skin depth
photon
() and the electron mean free path (escape depth)
electron
(E):
TE

electron
E
photon
1

electron
E
photon
The escape function D(E) is

DE
1
2
1
E
F
E
; E E
F
0; otherwise
9
APh 114b Lecture 17
Solid State Physics
Density of States Function in Wells, Wires and Dots
10
APh 114b Lecture 17
Solid State Physics
GaAs/AlGaAs Quantum Well Heterostructure
11
APh 114b Lecture 17
Solid State Physics
Energy Picture of Optical Pumping:
Quantum Well
Heterostructure
Bulk Semiconductor
12
APh 114b Lecture 17
Solid State Physics
Quantum Well Active Region in GRINSCH Laser
13
APh 114b Lecture 17
Solid State Physics
Analytic approx. for lowest 1s excited state:
Weak confinement (R>>a
e
, a
h
):
Coulomb term (~1/R) important
exciton confinement regime.
Strong confinement (R<<a
e
, a
h
):
Confinement term (~1/R
2
) dominates and e
-
and
h
+
quantized
molecular-type regime. Exciton Bohr
radius
Si: < 4.3 nm; Ge: < 11.5 nm; GaAs: < 12.4
R
e
(confined)
...
.
* *
+
+ + =
R
e 8 1
m
1
m
1
2R
E E(R)
2
2
h e
2
2 2
g
Excitons in Quantum Dots

a
e
(bulk)
a
h
(bulk)
R
h
(confined)
Quantum
Dot
Quantum confinement occurs when
R a
e
, a
h
(electron/hole Bohr radius)
14
APh 114b Lecture 17
Solid State Physics
Electronic States in
Quantum Dots
Envelope wave functions sample larger k-space:
Optical selection rules relax.
Optical absorption coeff. increases
due to wave function overlap
Energy levels become discrete.
Density of states become delta-functions.
Volume
1
) 0 ( U
2
p

(M.S. Hybertsen, PRL 72, 1514 (1994))
Si
15
APh 114b Lecture 17
Solid State Physics
Quantum Confinement Effects on Si and Ge
Band gap with confinement
(i.e., with nanocrystal diameter).
Radiative transition rate (1/) with confinement.
Si predicted to be quasidirect for d<~1.5 nm.
(ex. J.P. Proot et al., APL 61, 1948 (1992))
Zero phonon
Phonon-assisted
Si
(M.S. Hybertsen, PRL 72, 1514 (1994))
2 4 6 8 10
0.5
1.0
1.5
2.0
2.5
Ge
TD Tight-binding:
Hill, private comm.
EMA: Takagahara et al.,
PRB 46, 15578 (1992)
1
2
3
4
5
Bulk E
g
Si
Nanocrystal Diameter (nm)
Calculations:
TD Tight-binding: Hill et al., PRL 75, 1130 (1995))
Pseudopotential: Wang et al.,
J. Phys. Chem. 98, 2158 (1994)
3rd NN T-B: Delerue et al., PRB 48, 11024 (1993)
EMA: Takagahara et al., PRB 46, 15578 (1992)
C
a
l
c
u
l
a
t
e
d

B
a
n
d

G
a
p

(
e
V
)
16
APh 114b Lecture 17
Solid State Physics
Si Quant um Dot s:
XPS and HRTEM
106 104 102 100 98
106 104 102 100 98
600
o
C
800
o
C
1000
o
C
1100
o
C
Bulk Si (100)
400
o
C
As-Implanted
Thermal SiO
2
I
n
t
e
n
s
i
t
y

(
A
r
b
.

U
n
i
t
s
)
Binding Energy (eV)
Avg size 1~3 nm
As-implanted oxide is suboxide with
various Si
x+
(x=1~4) species.
Onset of precipitation ~600
o
C.
SiO
2
stoichiometric after annealing at
high T (T>1000
o
C).
Si 2p core level X-ray photoemission spectra :
5x10
16
Si/cm
2
17
APh 114b Lecture 17
Solid State Physics
Emission Energy Tuning by Oxidation:
PL vs. oxidation times:
PL emission energy continuously tunable from 1.4 eV to 2.2 eV
600 700 800 900 1000 1100
0.0
0.2
0.4
0.6
0.8
1.0
Ann. in O
2
at 1000
o
C
0 min
3 min
10 min
15 min
20 min
25 min
30 min
N
o
r
m
a
l
i
z
e
d

P
L

I
n
t
e
n
s
i
t
y

(
A
r
b
.

U
n
i
t
s
)
Wavelength (nm)
2.2 2 1.8 1.6 1.4 1.2
Bulk Si E
g
Energy (eV)
18
APh 114b Lecture 17
Solid State Physics
Colloidally Synthesized Quantum Dots
PL emission energy continuously tunable from
0.5 eV to 3.2 eV
19
APh 114b Lecture 17
Solid State Physics
Epitaxial InGaAs Quantum Dots

Solid State Physics

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Solid State Physics

Enviado por

Direitos autorais:

Formatos disponíveis

(2005) H.A.

T>0K : Fermi Statistics

where m* is termed the effective mass; these can be rearranged so that

(2005) H.A. Atwater, Caltech

These are standing waves

(2005) H.A. Atwater, Caltech

(2005) H.A. Atwater, Caltech

(2005) H.A. Atwater, Caltech

Consider the effect of on an arbitrary function f(r):

(2005) H.A. Atwater, Caltech

so another statement of the effective mass theorem is:

(2005) H.A. Atwater, Caltech

where F(r,t) is an envelope function:

Hydrogen atom Semiconductor Impurity

effective Bohr radius

(2005) H.A. Atwater, Caltech

probability that the state k is unoccupied

probability for electron scattering fromstate k to state k

, k probability for electron scattering fromstate k

and for elastic collisions,

(2005) H.A. Atwater, Caltech

f = numerical factor of order unity

and the carrier density associated with one carrier pocket is

Conductivity for Ellipsoidal Fermi Surfaces

Fermi funct ion

(2005) H.A. Atwater, Caltech

then the densities of states can be written as

(2005) H.A. Atwater, Caltech

Doped Semiconductors -- n type

phonon absorpt ion

(2005) H.A. Atwater, Caltech

Where the reduced effective mass is

Now the gradient of is

The dielectric constant is then

Direct Interband Transitions

where P denotes the principal value of the integral. These relations

Let be an impulse electric field at t=0. Then by

(negative sign (-): clockwise path on contour)

high frequency (optical frequency) dielectric constant

Each solution is 2-fold degenerate, since E can be chosen in an arbitrary

which is a photon-like mode with

Now let's look at (): Note that

Now separate the exciton wavefunction into center of mass and

where E is the electron energy, are the polar and azimuthal

The escape function D(E) is

Excitons in Quantum Dots

Você também pode gostar