Chapter 3.
Introduction to The p
Quantum Theory of Solids
1
3.1 Allowed and Forbidden Energy gy
Bands
2
Allowed and Forbidden Energy Bands gy
Formation of Energy Bands
The wave functions of the two atom electrons overlap, which means that
the two electrons will interact.
Th b l h d d This interaction or perturbation results in the discrete quantized energy
level splitting into two discrete energy levels.
The splitting of the discrete state into t o states is consistent ith the Pauli The splitting of the discrete state into two states is consistent with the Pauli
exclusion principle.
Fi r 3 1 ( ) P b bilit d it f ti f i l t d h d t
3
Figure 3.1 (a) Probability density function of an isolated hydrogen atom.
(b) Overlapping probability density functions of two adjacent hydrogen
atoms. (c) The splitting of the n = 1 state.
Formation of Energy Bands gy
If we start with a regular periodic arrangement of hydrogentype atoms
that are initially very far apart, and begin pushing the atoms together, the
initial quantized energy level will split into a band of discrete energy
l l levels.
This effect is shown schematically in Figure 3.2, where the parameter r
0
represents the equilibriuminteratomic distance in the crystal represents the equilibrium interatomic distance in the crystal.
Figure 3.2 The splitting
of an energy state into a
band of allowed energies.
4
Formation of Energy Bands gy
At the equilibrium, interatomic distance, there is a band of allowed
energies, but within the allowed band, the energies are at discrete levels.
The Pauli exclusion principle states that the joining of atoms to form a
system (crystal) does not alter the total number of quantum states regardless
of size.
H i l h h b h However since no two electrons can have the same quantum number, the
discrete energy must split into a band of energies in order that each electron
can occupy a distinct quantum state. can occupy a distinct quantum state.
At any energy level, the number of allowed quantum states is relatively small.
In order to accommodate all of the electrons in a crystal, we must have many y y
energy levels within the allowed band.
The energy difference between two energy levels is extremely small, so
5
that for all practical purposes, we have a quasicontinuous energy
distribution through the allowed energy band.
Example 3.1 p
Objective: To calculate the change in kinetic energy of an electron when the
velocity changes by a small value.
Consider an electron traveling at a velocity of 10
7
cm/s. Assume the velocity
i b l f 1 / Th i i ki i i i b increases by a value of 1 cm/s. The increase in kinetic energy is given by
Let v
2
= v
1
+∆v. Then
2 2 2 2
2 1 2 1
1 1 1
( )
2 2 2
E mv mv m v v A = ÷ = ÷
Let v
2
v
1
+∆v. Then
But ∆v << v
1
, so we have that
2 2 2 2
2 1 2 1
( ) 2 ( ) v v v v v v v = + A = + A + A
Solution
1 1
1
(2 )
2
E m v v mv v A ~ A = A
Substituting the number into this equation, we obtain
which may be converted to units of electron volts as
3 5 28
(9.11 10 )(10 )(0.01) 9.11 10 J E
÷ ÷
A = × = ×
6
which may be converted to units of electron volts as
28
9
19
9.11 10
=5.7 10 eV
1.6 10
E
÷
÷
×
A = ×
×
Comment
A change in velocity of 1 cm/s compared with 10
7
cm/s results in a change
in every of 5.7×10
9
eV, which is orders of magnitude larger than the
h i f 5 7×10
19
V b i h ll d change in energy of 5.7×10
19
eV between energy states in the allowed
energy band. This example serves to demonstrate that a difference in
adjacent energy states of 10
9
eV is indeed very small, so that the discrete adjacent energy states of 10 eV is indeed very small, so that the discrete
energies within an allowed band may be treated as a quasicontinuous
distribution.
7
The case for containing more than
one electron one electron
Suppose the atom contains electrons up through n = 3 energy level.
If the aqtoms are initially very far apart the electrons in adjacent atoms will If the aqtoms are initially very far apart, the electrons in adjacent atoms will
not interact and will occupy the discrete energy levels.
If these atoms brought closer together the outermost electrons in the n = 3 If these atoms brought closer together, the outermost electrons in the n 3
energy shell will begin to interact initially, so that this energy level will
split into a band of allowed energies.
If the atoms continue to mover closer together, the electrons in the n = 2
shell and n = 1 shell may begin to interact and also split into a band of
allowed energies.
Finally, if the atoms become sufficiently close together, the innermost
l i h 1 l l i h hi l l l li i electrons in the n = 1 level may interact, so that this level may also split into
a band of allowed energies. (Fig.3.3)
If the equilibrium interatomic distance is r then we have bands of allowed
8
If the equilibrium interatomic distance is r
0
, then we have bands of allowed
energies that the electrons may occupy separated by bands of forbidden
energies.
The case for containing more than
one electron one electron
This energyband splitting and the formation of allowed and forbidden
bands is the energyband theory of single crystal materials
9
Figure 3.3 Schematic showing the splitting of three energy states
into allowed bands of energies.
Band Splitting in Silicon Atom p g
The actual band splitting in a crystal is much more complicated than
indicated in Figure 3.3. A schematic representation of an isolated silicon
atom is shown in Figure 3.4a.
Ten of the fourteen silicon atom electrons occupy deeplying energy levels
close to the nucleus. The four remaining valence electrons are relatively
weakly bound and are the electrons involved in chemical reactions weakly bound and are the electrons involved in chemical reactions.
Figure 3.4b shows the band splitting of silicon. We need only consider n =
3 level for the valence electrons, since the first two energy shells are 3 level for the valence electrons, since the first two energy shells are
completely full and are tightly bound to the nucleus.
The 3s state corresponds to n = 3 and l = 0 and contains two quantum p q
states per atom. This state will contain two electrons at T = 0 K. The 3p
state corresponds to n = 3 and l = 1 and contains six quantum states per
10
atom.
As the interatomic distance decreases, the 3s and 3p states interact and
overlap. At the equilibrium interatomic distance, the bands have again split,
but now four quantum states per atom are in the lower band and four
i h b d quantum states per atom are in the upper band.
At absolute zero degrees, electrons are in the lowest energy state, so that all
states in the lower band (the valence band) will be full and all states in the states in the lower band (the valence band) will be full and all states in the
upper band (the conduction band) will be empty.
The bandgap energy E between the top of the valence band and the The bandgap energy E
g
between the top of the valence band and the
bottom of the conduction band is the width of the forbidden energy band.
We have discussed qualitatively how and why bands of allowed and q y y
forbidden energies are formed in a crystal. The formation of these energy
bands is directly related to the electrical characteristics of the crystal, as we
will see later in our discussion.
11
Band Splitting in Silicon Atom p g
Fi r 3 4 ( ) S h ti f i l t d ili t (b) Th plitti f th
12
Figure 3.4 (a) Schematic of an isolated silicon atom. (b) The splitting of the
3s and 3p states of silicon into the allowed and forbidden energy bands.
The KronigPenny Model g y
Figure 3.5a – The potential function of
a single, noninteracting, oneelectron
atom
Fi 3 5b S l i l Figure 3.5b – Several atoms are in close
and the potential functions overlap.
Figure 3.5c – The net potential function Figure 3.5c The net potential function
for onedimensional singlecrystal
material
KronigPenny Model : idealized
periodic potential representing a one
dimensional single crystal
13
Figure 3.6 The onedimensional periodic
potential function of the KronigPenny model.
Figure 3.5
To obtain the solution to Schroedinger equation, we make use of a
mathematical theorem by Bloch. The theorem states that all oneelectron
wave functions for problems involving periodically varying potential
energy functions wave function must be of the form k energy functions, wave function must be of the form
: a constant of motion, u(x): periodic function with period (a+b)
( ) ( )
jkx
x u x e v =
(3.1)
k
From Chapter 2, (total solution) = (timeindependent solution) ⅹ(time
dependent solution)
( )
(The travelingwave solution)
If id i Ⅰi Fi 3 6 (0 ) i hi h V( ) 0 k h
( / ) ( ( / ) )
( , ) ( ) ( ) ( ) ( )
jkx j E t j kx E t
x t x t u x e e u x e v o
÷ ÷
+ = = · =
(3.2)
(3.3)
If we consider regionⅠin Figure 3.6 (0<x<a) in which V(x) = 0, take the
second derivative of Eq.(3.1), and substitute this into the Eq.(2.13),
2
2 2
1 1
( ) ( )
2 ( ) ( ) 0
d u x du x
jk k u x o + ÷ ÷ =
(3 4)
14
1
2
2 ( ) ( ) 0 jk k u x
dx dx
o + ÷ ÷ =
2
2
2mE
o =
(3.4)
(3.5)
In region Ⅱ( )
0
0, ( ) b x V x V ÷ < < =
If d fi
2
2 2
0 2 2
2
2 2
2 ( ) ( )
2 ( ) ( ) 0
mV d u x du x
jk k u x
dx dx
o + ÷ ÷ + =
(3.6)
If we define
Eq.(3.6) can be written as
2 2
0
0
2 2
2 2
( )
mV m
E V o  ÷ = ÷ =
(3.7)
Eq.(3.6) can be written as
2
2 2
2 2
2
2
( ) ( )
2 ( ) ( ) 0
d u x du x
jk k u x
dx dx
 + ÷ ÷ =
(3.8)
The solution to Eq.(3.4), for region Ⅰ
( ) ( )
1
( ) (0 )
j k x j k x
u x Ae Be x a
o o ÷ ÷ +
= + < <
(3.9)
The solution to Eq.(3.8), for region Ⅱ
( ) ( )
2
( ) ( 0)
j k x j k x
u x Ce De b x
  ÷ ÷ +
= + ÷ < <
(3.10)
15
Since V(x) is everywhere finite, both the ψ(x) and ∂ψ(x)/∂x must be
continuous.
At the boundary (x = 0), the continuity condition
1 2
(0) (0) u u =
1 2
(0) (0) u u =
0 A B C D + ÷ ÷ =
1 2
du du
(3.11)
(3.12)
(3 13)
1 2
0 0 x x
du du
dx dx
= =
=
( ) ( ) ( ) ( ) 0 k A k B k C k D o o   ÷ ÷ + ÷ ÷ + + =
(3.13)
(3.14)
The periodicity and the continuity condition
1 2
( ) ( ) u a u b = ÷
( ) ( ) ( ) ( ) j k j k j k b j k b  
(3.15)
( )
( ) ( ) ( ) ( )
0
j k a j k a j k b j k b
Ae Be Ce De
o o   ÷ ÷ + ÷ ÷ +
+ ÷ ÷ =
1 2
du du
dx dx
=
(3.16)
(3.17)
The result is
x a x b
dx dx
= =÷
(3.18)
( ) ( ) ( ) ( )
( ) ( ) ( ) ( ) 0
j k a j k a j k b j k b
k Ae k Be k Ce k De
o o  
o o  
÷ ÷ + ÷ ÷ +
÷ ÷ + ÷ ÷ ÷ + =
16
The result is
2 2
( )
(sin )(sin ) (cos )(cos ) cos ( )
2
a b a b k a b
o 
o  o 
o
÷ +
+ = + (3.19)
When E < V
0
, the electron bound within the crystal. β is then imaginary.
Eq.(3.19) can be written in terms of γ
j  ¸ =
(3.20)
Let the barrier width b →0 and the barrier height V →∞ but such that the
2 2
(sin )(sinh ) (cos )(cosh ) cos ( )
2
a b a b k a b
¸ o
o ¸ o ¸
o¸
÷
+ = +
(3.21)
Let the barrier width b → 0 and the barrier height V
0
→∞, but such that the
product bV
0
remains finite. Eq.(3.21) then reduces to
sin mV ba a o
 
We may define a parameter P’ as
0
2
sin
cos cos
mV ba a
a ka
a
o
o
o
 
+ =

\ .
(3.22)
'
0
2
mV ba
P =
(3.23)
17
Finally, we have the relation
'
sin
cos cos
a
P a ka
a
o
o
o
+ =
(3.24)
The kSpace Diagram p g
For the case of V
0
= 0, from Eq.(3.24) For the case of V
0
0, from Eq.(3.24)
cos cos a ka o =
(3.25)
k o =
(3.26)
Since the potential is zero, the total energy is equal to the kinetic energy.
2 1
2
2 ( )
2
m mv
mE p
k o = = = =
(3.27)
P is the particle momentum. Therefore parameter k is related to the particle
f h f l d i l f d b
2 2
k o = = = =
(3.27)
momentum for the free electron, and is also referred to as a wave number.
2 2 2
2 2
p k
E
m m
= =
(3.28)
Figure 3.7 The parabolic
E versus k curve for the
18
E versus k curve for the
free electron.
The Ek Relation
As the P’ increases, the particle becomes more tightly bound to the potential
well or atom. We may define the left side of Eq.(3.24) as
'
sin
( ) cos
a
f a P a
o
o o = +
(3 29)
Now, Eq.(3.24) becomes
( ) cos f a P a
a
o o
o
= +
(3.29)
( ) f k
(3 30)
f(αa) must be bounded between +1 and 1.
The parameter α is related to the total energy E of the particle
( ) cos f a ka o =
(3.30)
The parameter α is related to the total energy E of the particle.
A plot of the energy E of the particle as a function of the wave number
can be generated from Figure 3.8c. can be generated from Figure 3.8c.
19
Figure 3.8 A plot of (a) the first term in Eq.(3.29), (b) the second term
20
in Eq.(3.29), and (c) the entire f(αa) function. The shaded areas show the
allowed values of (αa) corresponding to real values of k.
Figure 3 8 The E versus k diagram generated
21
Figure 3.8 The E versus k diagram generated
from Figure 3.8. The allowed energy bands and
forbidden energy bandgaps are indicated.
Example 3.2 p
To determine the lowest allowed energy bandwidth.
Assume that the coefficient P’ = 10 and that the potential width a = 5 Å. p
Solution
To find the lowest allowed energy bandwidth, we need to find the
diff r i l k h fr m 0 t (S Fi r 3 8 ) F r k difference in αa values as ka changes from 0 to π (See Figure 3.8c). For ka
= 0, Eq.(3.29) becomes
i l d fi d 6 d h f k
sin
1 10 cos
a
a
a
o
o
o
= +
By trial and error, we find αa = 2.628 rad. We see that for ka = π, αa = π.
For αa = π, we have
2
2mE
a t · =
or
2
a t · =
2 2 2 34 2
19
2
2 31 10 2
(1.054 10 )
2.407 10 J 1.50 eV
2 2(9.11 10 )(5 10 )
E
ma
t t
÷
÷
÷ ÷
×
= = = × =
× ×
For αa = 2.628, we find that E
1
=1.68×10
19
J =1.053 eV.
The allowed energy bandwidth is then
1 50 1 053 0 447 V E E E A
22
Comment ; We see from Figure 3.8c that, as the energy increases, the
width of the allowed bands increase from this KronigPenny model.
2 1
1.50 1.053 0.447 eV E E E A = ÷ = ÷ =
The cosine function is periodic.
(3 31)
Therefore, we can displace portions of the curve by 2π. And the entire E
r k pl t i nt in d ithin / <k< /
(3.31)
cos cos( 2 ) cos( 2 ) ka ka n ka n t t = + = ÷
versus k plot is contained within –π/a<k<π/a.
Fi 3 10 Th E k di
2t
23
Figure 3.10 The E versus k diagram
showing 2π displacements of several
sections of allowed energy bands.
Figure 3.11 The E versus k diagram in
the reducedzone representation.
Nearly Free Electron Model (Kittel, Chap 6) Ne y ee ec o ode ( e , C p 6)
The free particle Schrodinger equation in onedimension is
2 2
If the electrons are confined to a linear solid of length L, the wave function
2 2
2
( )
( )
2
x
E x
m x
v
v
÷ c
· =
c
(6)
is the standing wave
( ) sin( / )
n
x A nx L v t =
(7)
where n is a positive integer.
It is convenient to introduce wave functions that satisfy periodic boundary
conditions. We require the wave functions to be periodic in x with period
L. Thus
( ) ( ) x x L v v + (8)
Wave functions satisfying the freeparticle Schrodinger equations and the
periodicity condition are of the form of a traveling plane wave:
( ) ( ) x x L v v = + (8)
periodicity condition are of the form of a traveling plane wave:
24
( ) exp( ) x jkx v =
(9)
Figure 2 6 Particle in an infinite potential well: (a) Four lowest discrete energy levels (b) Figure 2.6 Particle in an infinite potential well: (a) Four lowest discrete energy levels. (b)
Corresponding wave functions. (c) Corresponding probability functions.
25
provided that the component of the wavevector k satisfy
Any component of k is of the form 2nπ/L, where n is a positive or
2 4
0; ; k
L L
t t
= ± ± · · · ·
(10)
negative integer. The components of k are the quantum numbers of the
problem. We confirm that these values of k satisfy (8), for
( ) exp[ ( )] exp[ 2 ( ) / ] exp( 2 / ) exp( 2 )
exp( 2 / ) exp( ) ( )
x L ik x L j n x L L j n x L j n
j n x L jkx x
v t t t
t v
+ = + = + =
= = =
(11)
On substituting (9) in (6) we have the energy E
k
of the orbital with
wavevector k:
2
2
E k =
(12)
The magnitude of the wavevector is related to the wavelength z by k = 2π/z.
2
k
E k
m
=
26
Nearly Free Electron Model (Kittel, Chap7) Nearly Free Electron Model (Kittel, Chap7)
On the free electron model, the allowed energy values are distributed
essentially continuously from zero to infinity. We saw in Chapter 6 that
2
2
2
k
E k
m
=
(1)
where, for periodic boundary conditions over a linear solid of length L,
2 4
0; ; k
L L
t t
= ± ± · · · · (2)
The free electron wave functions are of the form
L L
( ) exp( ) x jkx v = (3)
They represent running waves and carry momentum p = ħk.
The band structure of a crystal can often be explained by the nearly
free electron model for which the band electrons are treated as perturbed
only weakly by the periodic potential of the ion cores.
Thi d l l ll h li i i b h This model answers almost all the qualitative questions about the
behavior of electrons in metals.
27
We know that Bragg reflection is a characteristic feature of wave propagation
in crystals. Bragg reflection of electron waves in crystals is the cause of
energy gaps. (At Bragg reflection, wavelike solutions of the Schrodinger
i d i i Fi 2 ) Th f d i i equation do not exist, as in Fig. 2.) These energy gaps are of decisive
significance in determining whether a solid is an insulator or a conductor.
We explain physically the origin of energy gaps in the simple problem of a linear We explain physically the origin of energy gaps in the simple problem of a linear
solid of lattice constant a. The low energy portions of the band structure are shown
qualitatively in Fig. 2, in (a) for entirely free electrons and in (b) for electrons that are
nearly free, but with an energy gap at k = ±π/a. The Bragg condition (k+G)² = k²
for diffraction of a wave of wavevector k becomes in one dimension
1 n
k G
t
= ± = ± (4)
where G = 2πn/a is a reciprocal lattice vector and n is an integer. The first
reflections and the first energy gap occur at k = ±π/a. The region in kspace
2
k G
a
± ± (4)
between π/a and π/a is the first Brillouin zone of this lattice. Other energy gaps
occur for other values of the integer n.
28
Brillouin Zones Brillouin Zones
A linear lattice in one dimensions in Fig.11.
The zone boundaries of the linear lattice are at k = π/a, where a is
the primitive axis of the crystal lattice.
Figure 11 Crystal and reciprocal lattices in one dimension. The basis vectors in
the reciprocal lattice is b, of length equal to 2π/a. The shortest reciprocal
l i f h i i b d b h di l bi f
29
lattice vectors from the origin are b and –b. the perpendicular bisectors of
these vectors form the boundaries of the first Brillouin zone. The boundaries
are at k = +π/a.
The wave functions at k = ±π/a are not the traveling waves exp(jπx/a) or exp(jπx/a)
of free electrons. At these special values of k, the wave functions are made up of
equal parts of waves traveling to the right and to the left.
When the Bragg reflection condition k = ±π/a is satisfied by the wave vector a When the Bragg reflection condition k = ±π/a is satisfied by the wave vector, a
wave traveling to the right is Braggreflected to travel to the left, and vice versa. Each
subsequent Bragg reflection will reverse the direction of travel of the wave. A wave that
travels neither to the right nor to the left is a standing wave: it doesn’t go anywhere.
The timeindependent state is represented by standing waves. We can form two
different standing waves from the two traveling waves exp(±jπx/a) namely different standing waves from the two traveling waves exp(±jπx/a), namely
( ) exp( / ) exp( / ) 2cos( / );
( ) exp( / ) exp( / ) 2 sin( / ).
j x a j x a x a
j x a j x a j x a
v t t t
v t t t
+ = + ÷ =
÷ = ÷ ÷ =
(5)
The standing waves are labeled (+) or () according to whether or not they change
sign when –x is substituted for x. Both standing waves are composed of equal parts of
i h d l f di d li right and leftdirected traveling waves.
30
Figure 2 (a) Plot of energy ε versus wave vector k for a free electron.
(b) Plot of energy versus wave vector for an electron in a monatomic
linear lattice of lattice constant a. The energy gap E
g
shown is
associated with the first Bragg reflection at k = ±π/a; other gaps are gg g p
found at ±nπ/a, for integral values of n.
Origin of The Energy Gap Origin of The Energy Gap
The two standing waves ψ(+) and ψ() pile up electrons at different regions, and
therefore the two waves have different values of the potential energy. This is the
origin of the energy gap. The probability density ρ of a particle is ψ*ψ=ψ².
For a pure traveling wave exp(jkx) we have ρ = exp( jkx)exp(jkx) = 1 so that the For a pure traveling wave exp(jkx), we have ρ = exp(jkx)exp(jkx) = 1, so that the
charge density is constant. The charge density is not constant for linear
combinations of plane waves. Consider the standing wave ψ(+) in (5); for this we have
The function piles up electrons (negative charge) on the positive ions centered at x = 0,
2 i Fi 3 h th p t nti l n r i l t
2
2
( ) ( ) cos / x a p v t + = + ·
a, 2a, . . . in Fig. 3, where the potential energy is lowest.
Figure 3a pictures the variation of the electrostatic potential energy of a conduction
electron in the field of the positive ion cores. The ion cores bear a net positive charge p p g
because the atoms are ionized in the metal, with the valence electrons taken off to form
the conduction band. The potential energy of an electron in the field of a positive ion
is negative so that the force between them is attractive is negative, so that the force between them is attractive.
32
Figure 3 (a) Variation of potential energy of a conduction electron in the field of the ion
cores of a linear lattice. (b) Distribution of probability density ρ in the lattice for ψ()² ∝
i ² / ψ(+)² ∝ ² / d f t li Th f ti ψ(+) il sin² πx/a; ψ(+)² ∝cos² πx/a; and for a traveling wave. The wave function ψ(+) piles up
electronic charge on the cores of the positive ions, thereby lowering the potential energy in
comparison with the average potential energy seen by a traveling wave. The wave function
ψ( ) piles up charge in the region between the ions thereby raising the potential energy in ψ() piles up charge in the region between the ions, thereby raising the potential energy in
comparison with that seen by a traveling wave. This figure is the key to understanding
the origin of the energy gap.
For the other standing wave ψ(), the probability density is
which concentrates electrons away from the ion cores.
2
2
( ) ( ) sin / x a p v t ÷ = ÷ ·
In Fig. 3b we show the electron concentration for the standing waves
ψ(+), ψ(), and for a traveling wave.
When we calculate the average or expectation values of the potential
energy over these three charge distributions, we find that the potential
f (+) i l h h f h li h h energy of ρ(+) is lower than that of the traveling wave, whereas the
potential energy of ρ() is higher than the traveling wave. We have an
energy gap of width E
g
if the energies of ρ() and ρ(+) differ by E . energy gap of width E
g
if the energies of ρ( ) and ρ( ) differ by E
g
.
Just below the energy gap at points A in Fig. 2 the wave function is ψ(+),
and just above the gap at points B, the wave function is ψ(). j g p p ψ( )
34
Magnitude of The Energy Gap Magnitude of The Energy Gap
The wave functions at the Brillouin zone boundary k = π/a are √2 cos πx/a and
√2 sin πx/a, normalized over unit length of line. We write the potential energy of
an electron in the crystal at point x as
( ) cos 2 / U x U x a t =
(6)
The firstorder energy difference between the two standing wave states is
( ) cos 2 / U x U x a t =
1
2 2
( )[ ( ) ( ) ] E d U
í
(6)
0
2 2
( )[ ( ) ( ) ]
2 cos(2 / )(cos / sin / )
g
E dxU x
dxU x a x a x a U
v v
t t t
= + ÷ ÷
= ÷ =
í
í
We see that the gap is equal to the Fourier component of the crystal potential.
35
3.2 Electrical Conduction in Solids
36
The energy band and the bond model The energy band and the bond model
At T = 0 K, each silicon atom is surrounded by eight valence electrons that are
in their lowest energy state and are directly involved in the covalent bonding
and all of the valence electrons are in the valence band.
Th b d h d i b d i l l The upper energy band, the conduction band, is completely empty.
The 4N states in the lower band, the valence band, are filled with the valence
electrons electrons.
Figure 3.12 Twodimensional
representation of the covalent
bonding in a semiconductor at
T = 0 K
37
Band Splitting in Silicon Atom p g
Fi r 3 4 ( ) S h ti f i l t d ili t (b) Th plitti f th
38
Figure 3.4 (a) Schematic of an isolated silicon atom. (b) The splitting of the
3s and 3p states of silicon into the allowed and forbidden energy bands.
The energy band and the bond model The energy band and the bond model
As the temperature increases above 0 K, a few valence band electrons may
gain enough thermal energy to break the covalent bond and jump into the
conduction band.
Figure 3.13 (a) Twodimensional representation of the breaking of a covalent bond.
39
g ( ) p g
(b) Corresponding line representation of the energy band and the generation of a
negative and positive charge with the breaking of a covalent bond.
The energy band and the bond model The energy band and the bond model
The semiconductor is neutrally charged. This means that, as the negatively y g g y
charged electron breaks away from its covalent bonding position, a positive
charged “empty state” is created in the original covalent bonding position
in the valence bond.
As the temperature further increases, more covalent bonds are broken,
l t j t th d ti b d d iti “ t more electrons jump to the conduction band, and more positive “empty
states” are created in the valence band.
We can also relate this bond breaking to the E versus k energy bands We can also relate this bond breaking to the E versus k energy bands.
Figure 3.14a shows the E versus k diagram of the conduction and valence
bands at T = 0 K. The energy states in the valence band are completely full gy p y
and the states in the conduction band are empty.
40
Figure 3.14b shows these same bands for T > 0 K, in which some electrons
have gained enough energy to jump to the conduction band and have left
empty states in the valence band.
We are assuming at this point that no external forces are applied so the
electron and “empty state” distribution are symmetrical with k.
Fi 3 14 Th E k Figure 3.14 The E versus k
diagram of the conduction
and valence bands of a
semiconductor at (a) T = 0 K semiconductor at (a) T = 0 K
and (b) T > 0 K.
41
Drift Current
Current is due to the net flow of charge. If we had a collection of
positively charged ions with a volume density N (cm
3
) and an average
drift velocity v
d
(cm/s), then the drift current density is
2
A/ J N
(3 32)
If, instead of considering the average drift velocity, we considered the
individual ion velocities, then we could write the drift current density as
2
A/cm
d
J qNv =
(3.32)
individual ion velocities, then we could write the drift current density as
where v
i
is the velocity of the ith ion. The summation in Eq.(3.33) is taken
1
N
i
i
J q v
=
=
¯
(3.33)
i
y q ( )
over a unit volume so that the current density J is still in units of A/cm
2
.
Since electrons are charged particles, a net drift of electrons in the
conduction band will give rise to a current. The electron distribution in the
conduction band, as shown in Fig. 3.14a, is an even function of when no
t l f i li d
42
external force is applied.
Recall that k for a free electron is related to momentum so that, since there
are as many electrons with a +k value as there are with a k value, the
net drift current density due to these electrons is zero.
This result is certainly expected since there is no externally applied force.
If a force is applied to a particle and the particle moves, it must gain energy.
Thi ff i d This effect is expressed as
where F is force, dx is the differential distance the particle moves, v is the
dE Fdx Fvdt = =
(3.34)
where F is force, dx is the differential distance the particle moves, v is the
velocity, and dE is the increase in energy.
If an external force is applied to the electrons in the conduction band, there pp
are empty energy states into which the electrons can move; therefore,
because of the external force, electrons can gain energy and a net
momentum.
43
The electron distribution in the conduction band may look like that shown in
Fig. 3.15, which implies that the electrons have gained a net momentum.
We may write the drift current density due to the motion of electrons as
n
¯
(3 35)
where e is the magnitude of the electronic charge and n is the number of
l i l i h d i b d A i h i i
1
i
i
J e v
=
= ÷
¯
(3.35)
electrons per unit volume in the conduction band. Again, the summation is
taken over a unit volume so the current density is A/cm
2
.
Figure 3.15 The asymmetric g y
distribution of electrons in the
Ek diagram when an external
force is applied.
44
pp
Electron Effective Mass
The movement of an electron in a lattice will, in general, be different from
that of an electron in free space In addition to an externally applied force that of an electron in free space. In addition to an externally applied force
there are internal force in the crystal due to positively charged ions and
negatively charged electrons which will influence the motion of electrons g y g
in the lattice. We can write
total ext int
F F F ma = + = (3.36)
where F
total
, F
ext
, and F
int
are the total force, the externally applied force,
and the internal forces, respectively, acting on a particle in a crystal. The
i h l i d i h f h i l parameter a is the acceleration and m is the rest mass of the particle.
Since it is difficult to take into account all of the internal forces, we will
write the equation write the equation
where the acceleration a is now directly related to the external force. The
*
ext
F m a = (3.37)
45
where the acceleration a is now directly related to the external force. The
parameter m
*
is called the effective mass, takes into account the particle
mass and also takes into account the effect of the internal forces.
We can also relate the effective mass of an electron in a crystal to the Ek
curves, such as was shown in Fig. 3.11.
In a semiconductor material, we will be dealing with allowed energy bands
th t r lm t mpt f l tr n nd th r n r b nd r f ll f l tr n that are almost empty of electrons and other energy bands are full of electrons.
To begin, consider the case of a free electron whose E – k curve was shown in
Fig 3 7 Recalling the Eq (3 28) the energy and momentumare related by E Fig. 3.7. Recalling the Eq. (3.28), the energy and momentum are related by E
= p
2
/2m = ħ
2
k
2
/2m, where m is the mass of the electron.
The momentumand wave number k are related by p =ħk. If we take the y p
derivative of Eq. (3.28) with respect to k, we obtain
2
dE k p
dk
= =
(3.38)
Relating momentum to velocity, Eq.(3.38) can be written as
dk m m
1 dE p
v
dk m
= =
(3.39)
where v is the velocity of the particle. The first derivative of E with respect to
k is related to the velocity of the particle.
46
dk m
If we now take the second derivative of E with respect to k we have,
We may rewrite Eq. (3.40) as
2 2
2
d E
dk m
=
(3.40)
The second derivative of E with respect to k is inversely proportional to the
2
2 2
1 1 d E
dk m
=
(3.41)
mass of the particle.
For the case of a free electron, the mass is a constant, so the second
derivative function is a constant.
We may also note from Fig. 3.7 that d
2
E/dk
2
is a positive quantity, which
i li h h f h l i l i i i implies that the mass of the electron is also a positive quantity.
If we apply an electric field to the electron and use Newton’s classical equation
of motion we can write
47
of motion, we can write
F ma eE = = ÷
(3.42)
Solving for the acceleration, we have
eE ÷
(3 43)
The motion of the electron is in the opposite direction to the applied electric
field because of the negative charge
eE
a
m
=
(3.43)
field because of the negative charge.
We may now apply the results to the electron in the bottom of an allowed
energy band Consider the allowed energy band in Fig 3 16a energy band. Consider the allowed energy band in Fig. 3.16a.
48
Figure 3.16 (a) The conduction band in reduced k space, and the parabolic approximation.
(b) The valence band in reduced k space, and the parabolic approximation.
The energy near the bottom of this energy band may be approximated by a
parabola, just as that of a free particle. We may write
Th E i h h b f h b d Si E > E h
2
1
( )
c
E E C k ÷ =
(3.44)
The energy E
c
is the energy at the bottom of the band. Since E > E
c
, the
parameter C
1
is a positive quantity.
Taking the second derivative of E with respect to k from Eq (3 44) we obtain Taking the second derivative of E with respect to k from Eq.(3.44), we obtain
2
1
2
2
d E
C
dk
=
(3.45)
We may put Eq.(3.45) in the form
2
1
2 2 2
2 1 C d E
dk
=
(3.46)
Comparing Eq.(3.46) with Eq.(3.41), we may equate ħ
2
/2C
1
to the mass of the
particle However the curvature of the curve in Fig 3 16a will not in general
2 2 2
dk
49
particle. However, the curvature of the curve in Fig. 3.16a will not, in general,
be the same as the curvature of the freeparticle curve. We may write
2
1
2 2 2 *
2 1 1 C d E
= =
(3.47)
The effective mass relates the quantum mechanical results to the classical
force equations. If we apply an electric field to the electron in the bottom
2 2 2 *
dk m
(3.47)
q w pp y d b
of the allowed energy band, we may write the acceleration as
*
eE
a
m
÷
=
(3.48)
where m
*
n
is the effective mass of the electron. The effective mass m
*
n
of the electron near the bottom of the conduction band is a constant.
n
m
50
Concept of The Hole p
When a valence electron is elevated into the conduction band, a positively
charged “empty state” is created.
For T > 0 K, all valence electrons may gain thermal energy; if a valence
electron gains a small amount of thermal energy, it may hop into the empty
state.
Th f l l i h i i l The movement of a valence electron into the empty state is equivalent
to the movement of the positively charged empty state itself.
Figure 3 17 shows the movement of valence electrons in the crystal Figure 3.17 shows the movement of valence electrons in the crystal
alternately filling one empty state and creating a new empty state, a motion
equivalent to a positive charge moving in the valence band. q p g g
This charge carrier is called a hole and, as we will see, can also be thought
of as a classical particle whose motion can be modeled using Newtonian
51
mechanics.
The drift current density due to electrons in the valence band, such as shown
in Fig. 3.14b, can be written as
(filled)
i
i
J e v = ÷
¯
(3.49)
where the summation extends over all filled state. This summation is
inconvenient since it extends over a nearly full valence band and takes into
nt r l r n mb r f t t W m r rit Eq (3 49) in th f rm account a very large number of states. We may rewrite Eq. (3.49) in the form
( total) (empty) (total) (empty)
( ) ( )
i i i i
i i i i
J e v e v e v e v = ÷ ÷ ÷ = ÷ +
¯ ¯ ¯ ¯
(3.50)
52
Figure 3.17 Visualization of the movement of a hole in a semiconductor.
If we consider a band that is totally full, all available states are occupied by
electrons. The individual electrons can be thought of as moving with a
velocity as given by Eq. (3.39):
1 1 dE p dE
The band is symmetric in k and each state is occupied so that, for every
l i h l i h i di l i h l i
1 1
( ) ( )( )
dE p dE
v v E
dk m dk
= = ÷ =
electron with velocity v, there is a corresponding electron with a velocity v.
Since the band is full, the distribution of electrons with respect to k cannot
be changed with an externally applied force. The net drift current density be changed with an externally applied force. The net drift current density
from a completely full band, then, is zero, or
(t t l)
0
i
i
e v ÷ ÷
¯
(3.51)
We can write the drift current density from Eq.(3.50) for an almost full
band as
(total) i
(empty)
i
i
J e v = +
¯
(3.52)
53
where the v
i
in the summation is the associated with the empty
state.
(empty) i
1
( ) ( )( )
dE
v E
dk
=
Eq.(3.52) is entirely equivalent to placing a positively charged particle in the
empty states and assuming all other states in the band are empty, or neutrally
charged. This concept is shown in Fig 3.18. Fig. 3.18a shows the valence
band with conventional electronfilled states and empty states while Fig band with conventional electron filled states and empty states, while Fig.
3.18b shows the new concept of positive charges occupying the original
empty states. This concept is consistent with the discussion of the positively
charged “empty state” in the valence band as shown in Fig. 3.17.
54
Figure 3.18 (a) Valence band with conventional electronfilled states and empty
states. (b) Concept of positive charges occupying the original empty states.
The v
i
in the summation of Eq. (3.52) is related to how well this positively
charged particle moves in the semiconductor.
Now consider an electron near the top of the allowed energy band shown
in Fig. 3.16b. The energy near the top of the allowed energy band may
again be approximated by a parabola so that we may write.
2
( ) ( ) E E C k (3 53)
The energy E
v
is the energy at the top of the energy band. Since E < E
v
for
electrons in this band then the parameter C mist be a positive quantity
2
2
( ) ( )
v
E E C k ÷ = ÷ (3.53)
electrons in this band, then the parameter C
2
mist be a positive quantity.
Taking the second derivative of E with respect to k from Eq.(3.53), we
obtain obtain
We may rearrange this equation so that
2
2
2
2
d E
C
dk
= ÷
(3.54)
55
We may rearrange this equation so that
2
2
2 2 2
2 1 C d E
dk
÷
=
(3.55)
Comparing Eq.(3.55) with Eq.(3.41), we may write
where m
*
is again an effective mass. We have argued that C
2
is a positive
(3.56)
2
2
2 2 2 *
2 1 1 C d E
dk m
÷
= =
quantity, which now implies that m
*
is a negative quantity.
If we again consider an electron near the top of an allowed energy band
and use Newton’s force equation for an applied electric field, we will have
H
*
i i i i
*
F m a eE = = ÷
(3.57)
However, m
*
is now a negative quantity, so we may write
A l i h f ll d b d i h
* *
   
eE eE
a
m m
÷ +
= =
÷
(3.58)
An electron moving near the top of an allowed energy band moves in the
same direction as the applied electric field.
56
The net motion of electrons in a nearly full band can be described by
considering just empty states, provided that a positive electronic charge is
associated with each state and that the negative of m* from Eq. (3.56) is
i d i h h associated with each state.
We now can model this band as having particles with a positive electronic
charge and a positive effective mass charge and a positive effective mass.
The density of these particles in the valence band is the same as the density
of empty electronic energy states. of empty electronic energy states.
This new particle is the hole.
The hole, then, has a positive effective mass (denoted by ) and a
*
p
m The hole, then, has a positive effective mass (denoted by ) and a
positive electronic charge so it will move in the same direction as an applied
field.
p
57
Holes (Kittel, Chap. 8) Holes (Kittel, Chap. 8)
The properties of vacant orbitals in an otherwise filled band are important in
semiconductor physics and in solid state electronics. Vacant orbitals in a band are
commonly called holes. A hole acts in applied electric and magnetic fields as if it has
a positive charge +e. The reason is given in five steps in the boxes that follow. a positive charge e. The reason is given in five steps in the boxes that follow.
1. k
h
=  k
e
. (17)
The total wavevector of the electrons in a filled band is zero: Σk = 0. This result follows
from the geometrical symmetry of the Brillouin zone: every fundamental lattice type has
symmetry under the inversion operation r →– r about any lattice point; it follows that symmetry under the inversion operation r → r about any lattice point; it follows that
the Brillouin zone of the lattice also has inversion symmetry. If the band is filled all
pairs of orbitals k and –k are filled, and the total wavevector is zero.
If an electron is missing from an orbital of wavevector k
e
, the total wavevector of the a e ect o s ss g o a o b ta o wavevecto k
e
, t e tota wavevecto o t e
system is – k
e
and is attributed to the hole. This result is surprising: the electron is
missing from k
e
and the position of the hole is usually indicated graphically as situated at
k
e
, as in Fig. 7. But the true wavevector k
h
of the hole is – k
e
, which is the wavevector of
58
e
, g
h e
,
the point G if the hole is at E. The wavevector – k
e
enters into selection rules for
photon absorption.
Fi 7 Ab i f h f ħ d li ibl k l Figure 7. Absorption of a photon of energy ħω and negligible wavevector takes an electron
from E in the filled valence band to Q in the conduction band. If k
e
was the wavevector of
the electron at E, it becomes the wavevector of the electron at Q. The total wavevector of
the valence band after the absorption is k and this is the wavevector we must ascribe to the valence band after the absorption is  k
e
, and this is the wavevector we must ascribe to
the hole if we describe the valence band as occupied by one hole. Thus k
h
=  k
e
; the
wavevector of the hole is the same as the wavevector of the electron which remains at C.
For the entire system the total wavevector after the absorption of the photon is k + k = 0
59
For the entire system the total wavevector after the absorption of the photon is k
e
+ k
h
= 0,
so that the total wavevector is unchanged by the absorption of the photon and the creation
of a free electron and free hole.
The hole is an alternate description of a band with one missing electron, and we
either say that the hole has wavevector – k
e
or that the band with one missing electron
has total wavevector k has total wavevector – k
e
.
2. E
h
(k
h
) = – E
e
(k
e
) . (18)
Let the zero of energy of the valence band be at the top of the band. The lower in
the band the missing electron lies, the higher the energy of the system. The energy of
the hole is opposite in sign to the energy of the missing electron, because it takes more the hole is opposite in sign to the energy of the missing electron, because it takes more
work to remove an electron from a low orbital than from a high orbital. Thus if the
band is symmetric, E
e
(k
e
)=E
e
( k
e
)=E
h
( k
e
)=E
h
(k
h
). We construct in Fig. 8 a band
scheme to represent the properties of hole. This hole band is a helpful representation p p p p p
because it appears right side up.
60
Th h lf f h f h h h l b d h l h Figure 8 The upper half of the figure shows the hole band that simulates the
dynamics of a hole, constructed by inversion of the valence band in the origin.
The wavevector and energy of the hole are equal, but opposite in sign, to the
d f h l bi l i h l b d W d
61
wavevector and energy of the empty electron orbital in the valence band. We do
not show the disposition of the electron removed from the valence band at k
e
.
¹Bands are always symmetric under the inversion k→  k if the spinorbit
interaction is neglected. Even with spinorbit interaction, bands are always
symmetric if the crystal structure permits the inversion operation. Without a center
of symmetry, but with spinorbit interaction, the bands are symmetric if we of symmetry, but with spin orbit interaction, the bands are symmetric if we
compare subbands for which the spin direction is reversed: E (k,↑)=E (k,↓).
3 v = v (19) 3. v
h
= v
e
. (19)
The velocity of the hole is equal to the velocity of the missing electron. From Fig. 9
we see that ∇E (k )=∇E(k ) so that v (k )= v (k ) we see that ∇E
h
(k
h
)=∇E
e
(k
e
), so that v
h
(k
h
)= v
e
(k
e
).
4. m
h
= – m
e
. (20)
We show below that the effective mass is inversely proportional to the curvature
d²E/dk², and for the hole band this has the opposite sign to that for an electron in the
valence band. Near the top of the valence band me is negative, so that m
h
is positive.
62
p g ,
h
p
Figure 9. (a) At t = 0 all states are filled except F at the top of the band, the velocity v
x
, is
t F b dE/dk 0 (b) A l t i fi ld E i li d i th ± di ti Th zero at F because dE/dk
x
= 0 (b) An electric field E is applied in the ±x direction. The
force on the electrons is in the –k
x
direction and all electrons make transitions together in
the k
x
directtion, moving the hole to the state E. (c) After a further interval the electrons
move farther along in k space and the hole is now at D
63
move farther along in k space and the hole is now at D.
1 dk
5. (21)
This comes from the equation of motion
1
( )
h
h
d
e
dt c
= + ×
k
E v B
q
(CGS) (22)
1
( )
e
e
d
e
dt c
= ÷ + ×
k
E v B
that applies to the missing electron when we substitute –k
h
for ke and vh for ve. The
equation of motion for a hole is that of a particle of positive charge e. The positive
charge is consistent with the electric current carried by the valence band of Fig. 9: the
current is carried by the unpaired electron in the orbital G:
J = (e)v(G) = (e)[v(E)] = ev (E) , (23)
which is just the current of a positive charge moving with the velocity ascribed to the
missing electron at E. The current is shown in Fig. 10.
64
Fi 10 M i f l i h d i b d d h l i h Figure 10. Motion of electrons in the conduction band and holes in the
valence band in the electric field E. The hole and electron drift velocities
are in opposite directions, but their electric currents arc in the same
direction the direction of the electric field
65
direction, the direction of the electric field.
Metals, Insulators, and Semiconductors Metals, Insulators, and Semiconductors
Each crystal has its own energyband structure.
Insulator : energy bands are ether completely empty or completely full. The
bandgap energy E
g
is usually on the order of 3.5 to 6 eV or larger, so that at
room temperature, there are very few thermally generated electrons and
holes in an insulator.
S i d h l i l f l h l h b Semiconductor : has relatively few electrons or holes near the bottom or top
of the bands. The bandgap energy is usually on the order of 1 eV. If an electric
field is applied, the electrons or holes move through the crystal. The resistivity field is applied, the electrons or holes move through the crystal. The resistivity
can be controlled and varied over may orders.
Metal : partially full band in which there are many electrons available for p y y
conduction or conduction and valence bands overlap at the equilibrium
interatomic distance. There are large numbers of electrons as well as large
66
numbers of empty energy states into which the electron can move, so the
material can exhibit a very high electrical conductivity.
Figure 3.19 Allowed energy bands showing
(a) an empty band (b) a completely full
Figure 3.20 Allowed energy bands showing
(a) an almost empty band (b) a almost full
67
(a) an empty band, (b) a completely full
band, and (c) the bandgap energy between
the two allowed bands.
(a) an almost empty band, (b) a almost full
band, and (c) the bandgap energy between
the two allowed bands.
Figure 3.21 Two possible energy bands of a metal showing
(a) a partially filled band and (b) overlapping allowed
68
energy bands.
3.3 Extension to Three Dimensions
69
Extension to Three Dimensions
The basic concept of allowed and forbidden energy bands and the basic
concept of effective mass have been developed in the last sections.
In this section, we will extend these concepts to three dimensions and to real
crystals.
We will qualitatively consider particular characteristics of the three
di i l l i f h E k l b d d dimensional crystal in terms of the Ek plots, bandgap energy, and
effective mass.
We must emphasize that we will only briefly touch on the basic three We must emphasize that we will only briefly touch on the basic three
dimensional concepts; therefore, many details will not be considered.
One problem encountered in extending the potential function to a three One problem encountered in extending the potential function to a three
dimensional crystal is that the distance between atoms varies as the
direction through the crystal changes.
70
Extension to Three Dimensions
Figure 3.22 shows a facecentered cubic structure with the [100] and [110]
directions indicated.
Electrons travelling in the different directions encounter different potential
patterns and therefore different kspace boundaries.
The E versus k diagrams are in general a function of the kspace direction
i l in a crystal.
Figure 3.22 Two (100) plane of a
facecentered cubic crystal showing face centered cubic crystal showing
the [100] and [110] directions.
71
The kSpace Diagrams of Si and GaAs The k Space Diagrams of Si and GaAs
Figure 3.23 shows an Ek diagram of gallium arsenide and of silicon.
These simplified diagrams show the basic properties considered in this text,
but do not show many of the details more appropriate for advancedlevel
courses.
Note that in place of the usual positive and negative k axes, we now show
diff l di i two different crystal directions.
The Ek diagram for the onedimensional model was symmetric in k so
that no new information is obtained by displaying the negative axis that no new information is obtained by displaying the negative axis.
It is normal practice to plot the [100] direction along the normal +k axis
and to plot the [111] portion of the diagram so the +k point to the left. and to plot the [111] portion of the diagram so the k point to the left.
In the case of diamond or zincblende lattices, the maxima in the valence
band energy and minima in the conduction band energy occur at k = 0 or
72
gy gy
along one of these two directions.
The kSpace Diagrams of Si and GaAs The k Space Diagrams of Si and GaAs
73
Figure 3.23 Energy band structures of (a) GaAs and (b) Si.
The kSpace Diagrams of Si and GaAs The k Space Diagrams of Si and GaAs
Figure 3.23a shows the Ek diagram for GaAs.
The valence band maximum and the conduction band minimum both
occur at k = 0. The electrons in the conduction band tend to settle at the
minimum conduction band energy which is at k = 0.
Similarly, holes in the valence band tend to congregate at uppermost valence
b d band energy.
In GaAs, the minimum conduction band energy and maximum
valence band energy occur at the same k value As semiconductor with valence band energy occur at the same k value. As semiconductor with
this property is said to be direct bandgap semiconductor.
Transitions between the two allowed bands can take place with no change in p g
crystal momentum.
This direct nature has significant effect on the optical properties of the
74
material. GaAs and other direct bandgap materials are ideally suited for use
in semiconductor lasers and other optical devices.
The kSpace Diagrams of Si and GaAs The k Space Diagrams of Si and GaAs
The Ek diagram for silicon is shown in Fig. 3.23b.
The maximum in the valence band energy occurs at k = 0 as before.
The minimum in the conduction band energy occurs not at k = 0, but
along the [100] direction.
The difference between the minimum conduction band energy and the
maximum valence band energy is still defined as the energy bandgap E
g
.
A semiconductor whose maximum valence band energy and
i i d i b d d h k l minimum conduction band energy do not occur at the same k value
is called an indirect bandgap semiconductor.
When electrons make a transition between the conduction and valence When electrons make a transition between the conduction and valence
bands, we must invoke the law of conservation of momentum.
A transition in an indirect bandgap material must necessarily include an
75
A transition in an indirect bandgap material must necessarily include an
interaction with the crystal so that crystal momentum is conserved.
Additional Effective Mass Concepts Additional Effective Mass Concepts
The curvature of Ek diagrams near the minimum of the conduction band
i l d h ff i f h l energy is related to the effective mass of the electron.
We may note from Fig. 3.23, that the curvature of conduction band at its
minim m val e for GaAs is larger than that of silicon so the effective mass of minimum value for GaAs is larger than that of silicon, so the effective mass of
an electron in the conduction band of GaAs will be smaller than in silicon.
For the onedimensional E versus k diagram the effective mass was defined For the one dimensional E versus k diagram, the effective mass was defined
by Eq.(3.41) as
2
2 2
1 1 d E
m dk

=
A complication occurs in the effective mass concept in a real crystal.
A threedimensional crystal can be described by three k vectors.
m dk
y y
The curvature of Ek diagram at the conduction band minimum may not be the
same in the three k directions.
76
In later sections and chapters, the effective mass parameters used in calculations
will be a kind of statistical average that is adequate for most device calculations.
A Linear Lattice in One Dimension e ce O e e s o
L = 6a
L is the length of a linear lattice and a is the lattice constant of the lattice.
L = 6a
a
k = 2n¡z = u k 2n¡z u
k = 2n¡z = 2n¡6a = n/3a = 2n¡L
ψ(k) = expjkx : travelling wave
k = 2n¡z = 4n¡6a = 2n/3a = 4n¡L
ψ(k) = expjkx : travelling wave
k = 2n¡z = 6n¡6a = n/a = 6n¡L
ψ(k) = 2coskx : standing wave
k = 2n¡z = 8n¡6a = 4n/3a = 8n¡L
k = 2n¡z = 6n¡6a = n/a = 6n¡L
ψ(k) = 2jsinkx : standing wave
k = 2n¡z = 1un¡6a = 5n/3a = 1un¡L
k = 2n¡z = 12n¡6a = 2n/a = 12n¡L
ψ(k) = 2coskx : standing wave
∆k = 2n/3a  n/3a = n/3a = 2n/6a = 2n/L
If L = Na and N = 10
7
, ∆k = 2n/L = 2n/Na is very small. → The k is quasi
continuous,
The total number of wave vector k in an allowed band is N.
Ek Diagram in The ReducedZone
Representation Representation
3.4 Density of States Function y
We wish to describe the current–voltage (JV) characteristics of
semiconductor devices.
Since the current is due to the flow of charge, an important step in the
i t d t i th b f l t d h l th t ill b process is to determine the number of electrons and holes that will be
available for conduction.
The number of carriers that can contribute to the conduction process is a p
function of the number of available energy or quantum states since, by the
Pauli exclusion principle, only one electron can occupy a given quantum
state state.
When we discussed the splitting of energy levels into bands of allowed and
forbidden energies, we indicated that the band of allowed energies was g g
actually made up of discrete energy levels.
We must determine the density of these allowed energy states as a
f i f i d l l h l d h l
81
function of energy in order to calculate the electron and hole
concentrations.
3.4.1 mathematical Derivation
We will consider a free electron confined to a threedimensional infinite
potential well, where the potential well represents the crystal.
The potential of the infinite potential well is define as
( ) 0 for 0 V L < < ( , , ) 0 for 0
0
0
V x y z x L
y L
z L
= < <
< <
< <
(3.59)
where the crystal is assumed to be a cube with length L.
( , , ) elsewhere V x y z = ·
y g
Schrodinger’s wave equation in three dimensions can be solved using the
separation of variables technique.
2 2
where n
x
, n
y
, and n
z
are integers and n
2
= n
x
2
+n
y
2
+n
z
2
.
( )
2 2
2 2 2 2 2 2 2 2
2
2 2 2
x y z x y z
mE
k k k k n n n n
L L
t t
   
= = + + = + + =
 
\ . \ .
(3.60)
82
x y z
g
x y z
A Cubic Lattice in Three Dimensions A Cubic Lattice in Three Dimensions
The energy of electrons in a three dimensional lattice can be written by
2 2 2 2 2 2 2 2 2
( / 2 ) ( / 2 )( ) ( / 2 )(2 / ) E m k m k k k m L n t = = + + = (3 60)
where n
x
, n
y
, and n
z
are integers and n
2
= n
x
2
+ n
y
2
+ n
z
2
.
The energy E and the values of allowed k
x y z x y z
n n n n n n
( / 2 ) ( / 2 )( ) ( / 2 )(2 / )
x y z
E m k m k k k m L n t = = + + = (3.60)
2 2
0, 0 0 0 2, n n = =
0
1 1 0
E = 0, the total number of quantum states (k) = 1 1 1 0
1 1 0
2
1, 1 0 0 1 1 0
1 0 0
n =
1 0 1
0 1 0 1 0 1
0 1 0 1 0 1
0 0 1 1 0 1
0 0 1 0 1 1
2
2 2
1
2
E = / 2 ( ) ; the total number of quantum states m
L
t
= 6 0 1 1
0 1 1
0 1 1
83
0 1 1
2 2
2
2
E = / ( ) ;
the total number of quantum states = 12
m
L
t
Figure 3.24 (a) A twodimensional array of allowed quantum states in k
84
space. (b) The positive oneeight of the spherical k space
The distance between two quantum states in the k direction for example is given The distance between two quantum states in the k
x
direction, for example, is given
by
1
2 2 2
( 1) k k n n
t t t
   
÷ = + ÷ =
 
(3 61)
Generalizing this result to three dimensions, the volume V
k
of a single quantum
1
( 1)
x x x x
k k n n
L L L
+
+
 
\ . \ .
(3.61)
state is
3
2
k
V
L
t
 
=

\ .
(3.62)
A differential volume in k space is shown in Figure 3.24b and is given by 4nk
2
dk,
so the differential density of quantum state in k space can be written as
L
\ .
( )
2
3
4
( ) 2
2 /
T
k dk
g k dk
L
t
t
= (3.63)
85
( )
2 / L t
Th f 4 k
2
dk i i h diff i l l d h f (2 /L)
3
i h l The factor 4nk
2
dk is again the differential volume and the factor (2n/L)
3
is the volume
of one quantum state. Equation (3.63) may be simplified to
2
k dk
2
3
3
( )
T
k dk
g k dk L
t
t
=
(3.64)
Equation (3.64) gives the density of quantum states as a function of momentum,
through the parameter k. We can now determine the density of quantum states as a
function of energy E For a free electron the parameters E and k are related by function of energy E. For a free electron, the parameters E and k are related by
2
2
2mE
k =
(3.65a)
or
1
2 k mE =
(3.65b)
86
2 k mE =
(3.65b)
The differential is
1
2
m
dk dE
E
=
(3.66)
Then, substituting the expressions for k
2
and dk into Equation (3.64), the number
of energy states between E and E +dE is given by
3
3 2
2 1
( )
2
T
L mE m
g E dE dE
E
t
t
 
=

\ .
(3.67)
Since ħ = h/2n, Equation (3.67) becomes
3
4 L t
Equation (3.68) gives the total number of quantum states between the energy E
3/ 2
3
4
( ) (2 )
T
L
g E dE m E dE
h
t
=
(3.68)
87
q ( ) g q gy
and E +dE in the crystal space volume of L
3
.
If we divided by the volume L
3
, then we will obtain the density of quantum
states per unit volume of the crystal. Equation (3.68) then becomes
3/ 2
3
4 (2 )
( )
m
g E E
h
t
=
(3.69)
The density of quantum states is a function of energy E. As the energy of this
free electron becomes small, the number of available quantum states decreases.
This density function is really a double density, in that the units are given in terms
of states per unit energy per unit volume.
88
3.4.2 Extension to Semiconductors 3.4. e s o o Se co duc o s
The parabolic relationship between energy and momentum of a free electron is
given as E = p
2
/2m = ħ
2
k
2
/2m. The E versus k curve near
k = 0 at the bottom of the conduction band can be approximated as a parabola, so
we may write we may write
2 2
*
2
c
k
E E
m
= +
(3.70)
where E
c
is the bottom edge of the conduction band and is electron effective
mass. Equation (3.70) may be rewritten to give
2
n
m
*
n
m
Th l f f h E k l ti f l t i th b tt f
2 2 2 2
*
( . )
2 2
c
n
k k
E E cf E
m m
÷ = =
(3.71)
(3.28)
The general form of the E versus k relation for an electron in the bottom of a
conduction band is the same as the free electron, except the mass is replaced by the
effective mass. We can then think of the electron in the bottom of the conduction
89
band as being a “free” electron with its own particle mass. “free” conduction
electron model
++
The density of allowed electronic energy states in the conduction band
* 3/ 2
3
4 (2 )
( )
n
c c
m
g E E E
h
t
= ÷
(3.72)
is valid for E ≥ E
c
. As the energy of electron on the conduction band decreases, the
number of available quantum states also decreases.
The density of quantum state in the valence band can be obtained by using the same The density of quantum state in the valence band can be obtained by using the same
infinite potential well model, since the hole is also confined in the semiconductor crystal
and can be treated as a free particle.
We may also approximate the E versus k curve near k = 0 by a parabola for a “free”
hole, so that
2 2
k (3 73)
Equation (3.73) may be rewritten to give
*
2
p
k
E E
m
u
= ÷
(3.73)
90
q ( ) y g
2 2
*
2
p
k
E E
m
u
÷ =
(3.74)
Equation (3.74) is of the same form used in the general derivation of the
density of state function. We may the generalize the density of states function
from Equation (3.69) to apply to the valence band, so that o Equat o (3.69) to app y to t e v e ce b d, so t at
* 3/ 2
3
4 (2 )
( )
p
m
g E E E
u u
t
= ÷ (3.75)
is valid for E ≤ E
v
.
3
( ) g
h
u u
( )
Quantum states do not exist within the forbidden energy band, so
g(E) = 0 for E ≤ E ≤ E Figure 3 25 shows the plot of the density of quantum g(E) = 0 for E
v
≤ E ≤ E
c
. Figure 3.25 shows the plot of the density of quantum
states as a function of energy.
91
92
Figure 3.25 The density of energy states in the conduction
band and valence band as a function of energy
3.5 Statistical Mechanics 3.5 S s c ec cs
In dealing with large numbers of particles, we are interested only in the
statistical behavior of the group as a whole rather than in the behavior of
each individual particle.
In a crystal, the electrical characteristics will be determined by the statistical
b h i f l b f l behavior of a large number of electrons.
93
3.5.1 Statistical Laws
In determining the statistical behavior of particles, we must consider the laws that
the particles obey There are three distribution laws determining the distribution the particles obey. There are three distribution laws determining the distribution
of particles among available energy states.
MaxwellBoltzmann probability function
Distinguishable
No limit to the number of particles allowed in each energy state
E ) Gas molec les in a container at fairl lo press re Ex) Gas molecules in a container at fairly low pressure
BoseEinstein function
Indistinguishable
No limit to the number of particles allowed in each quantum state
Ex) Photons or black body radiation
b b l f FermiDirac probability function
Indistinguishable
One particle in one quantum state
94
p q
Ex) Electrons in a crystal
In each case, the particles are assumed to be noninteracting.
A Cubic Lattice in Three Dimensions A Cubic Lattice in Three Dimensions
The energy of electrons in a three dimensional lattice can be written by
2 2 2 2 2 2 2 2 2
( / 2 ) ( / 2 )( ) ( / 2 )(2 / ) E k k k k L
where n
x
, n
y
, and n
z
are integers and n
2
= n
x
2
+ n
y
2
+ n
z
2
.
The energy E and the values of allowed k
2 2 2 2 2 2 2 2 2
( / 2 ) ( / 2 )( ) ( / 2 )(2 / )
x y z
E m k m k k k m L n t = = + + = (3.60)
2 2
0
0, 0 0 0 2, 1 1 0
E = 0, the total number of k
x y z x y z
n n n n n n
n n = =
= 1 1 1 0
2
1 1 0
1, 1 0 0 n = 1 1 0
1 0 0 1 0 1
0 1 0 1 0 1
0 1 0 1 0 1
0 0 1 1 0 1
2 2
1
0 0 1 0 1 1
2
E = / 2 ( ) ; the total number of k = 6 0 1 1 m
L
t
0 1 1
95
0 1 1
0 1 1
2 2
2
2
E = / ( ) ; the total number of k = 12 m
L
t
3.5.2 The FermiDirac Probability Function 3.5.2 The Fermi Dirac Probability Function
A maximum of one particle is allowed in each quantum state by the Pauli
exclusion principle.
The total number of ways of arranging N
i
particles in the i th energy level (where
N
i
≤ g
i
) is N
i
≤ g
i
) is
However, since the particles are indistinguishable, the N
i
! number of
!
( ) ( 1) ( ( 1))
( )!
i
i i i i
i i
g
g g g N
g N
÷ ÷ ÷ =
÷
(3.76)
p g
i
permutations that the particles have among themselves in any given arrangement
do not count as separate arrangements. Therefore, the actual number of
independent ways of realizing a distribution of N particles in the i th level is independent ways of realizing a distribution of N
i
particles in the i th level is
!
W
!( )!
i
i
i i i
g
N g N
=
÷
(3.77)
!( )!
i i i
N g N
96
Eq. (3.77) gives the number of independent ways of realizing a distribution of N
i
particles in the i th level.
The total number of ways of arranging (N
1
, N
2
, N
3
, · · · , N
n
) indistinguishable
particles among n energy levels is the product of all distributions, or
!
n
g
1
!
W
!( )!
n
i
i
i i i
g
N g N
=
=
÷
I (3.78)
The parameter W is the total number of ways in which N electrons can be arranged
in this system, where is the total number electrons in the system.
1
n
i
i
N N
=
=
¯
The maximum Wis found by varying N
i
among the E
i
levels, which varies the
distribution, but at the same time, we will keep the total number of particles
97
, , p p
and total energy constant.
We may write the most probable distribution function as
( ) 1
= ( )
F
N E
f E =
 
(3 79)
( )
( )
1 exp
F
F
f
E E
g E
kT
÷
 
+

\ .
(3.79)
where E
F
is called the Fermi energy.
The number density N(E) is the number of particles per unit volume per unit energy.
(E) i th b f t t t it l it g(E) is the number of quantum states per unit volume per unit energy.
f
F
(E) is called the FermiDirac distribution or probability function.
It gives the probability that a quantum state at the energy E will be occupied It gives the probability that a quantum state at the energy E will be occupied
by an electron.
f
F
(E) is the ratio of filled to total quantum states at any energy E.
98
3.5.3 The Distribution Function and
The Fermi Energy The Fermi Energy
1
( ) f
1
( )
1 exp( )
F
f
f E
E E
kT
=
÷
+
(3.79)
kT
99
100
1 1 1
( )
1 exp(0) 1 1 2
F F
f E E = = = =
+ + 1 exp(0) 1 1 2 + +
Figure 3.31 The Fermi
probability function versus
f diff energy for different
temperatures.
101
Figure 3.32  The probability of a
state being occupied, and the f
F
(E)
probability of a state being empty,
1 f (E) 1  f
F
(E).
Figure 3.33 The FermiDirac
probability function and the
MaxwellBoltzmann
102
approximation.
Consider the case when E – E
F
>> kT, where the exponential term in the
denominator of Equation(3.79) is much greater than unity. We may neglect the 1 in
the denominator, so the FermiDirac distribution function becomes
( )
( ) exp
F
F
E E
f E
kT
÷ ÷
~
(3.80)
Equation (3.80) is known as the Maxwellboltzmann approximation, or simply the
Boltzmann approximation, to the FermiDirac distribution function. Figure
3.33 shows the FermiDirac probability function and the Boltzmann
approximation. This figure gives an indication of the range of energies over which
the approximation is valid. the approximation is valid.
The actual Boltzmann approximation is valid when exp[(EE
F
)/kT]>>1 .
103
3.1 Allowed and Forbidden Energy gy Bands
2
Allowed and Forbidden Energy Bands gy
Formation of Energy Bands The wave functions of the two atom electrons overlap, which means that the two electrons will interact. This Th interaction or perturbation results in the d b l h discrete quantized energy d level splitting into two discrete energy levels. The splitting of the discrete state into two states is consistent with the Pauli t o ith exclusion principle.
Figure 3 1 (a) P b bilit density function f Fi r 3.1 ( ) Probability d it f ti of an isolated hydrogen atom. i l t dh d t (b) Overlapping probability density functions of two adjacent hydrogen atoms. (c) The splitting of the n = 1 state.
3
Formation of Energy Bands gy
If we start with a regular periodic arrangement of hydrogentype atoms that are initially very far apart. l l This effect is shown schematically in Figure 3. and begin pushing the atoms together.
4
. the initial quantized energy level will split into a band of discrete energy levels. where the parameter r0 represents the equilibrium interatomic distance in the crystal.2 The splitting of an energy state into a band of allowed energies. crystal
Figure 3.2.
the h b h discrete energy must split into a band of energies in order that each electron can occupy a distinct quantum state.Formation of Energy Bands gy
At the equilibrium. The energy difference between two energy levels is extremely small. there is a band of allowed energies. the number of allowed quantum states is relatively small.
5
. However since no two electrons can h H i l have the same quantum number. we have a quasicontinuous energy distribution through the allowed energy band. the energies are at discrete levels. At any energy level. but within the allowed band. The Pauli exclusion principle states that the joining of atoms to form a system (crystal) does not alter the total number of quantum states regardless of size. so that for all practical purposes. we must have many y y energy levels within the allowed band. interatomic distance. In order to accommodate all of the electrons in a crystal.
so we have that
E
Solution Substituting the number into this equation.1 p
Objective: To calculate the change in kinetic energy of an electron when the velocity changes by a small value. The increase in kinetic energy is given by i by l f / Th i i ki i i i b
E
Let v2 = v1+∆v. Consider an electron traveling at a velocity of 107 cm/s. Assume the velocity increases b a value of 1 cm/s.111028 J
1 m(2v1v) mv1v 2
which may be converted to units of electron volts as
9.7 109 eV 1.111028 E =5.11103 )(105 )(0.Example 3. we obtain
E (9. Then
1 2 1 2 1 2 mv2 mv1 m(v2 v12 ) 2 2 2
2 2 v2 (v1 v) 2 v2 2v1v (v) 2
But ∆v << v1 .6 1019
6
.01) 9.
so that the discrete energies within an allowed band may be treated as a quasicontinuous distribution.
7
.
Comment A change in velocity of 1 cm/s compared with 107 cm/s results in a change in every of 5. which is orders of magnitude larger than the change i energy of 5 7×1019 eV b h in f 5.7×109 eV.7×10 19 V between energy states i the allowed in h ll d energy band. This example serves to demonstrate that a difference in adjacent energy states of 109 eV is indeed very small.
3. the innermost electrons i the n = 1 l l may i l in h level interact. Finally. then we have bands of allowed energies that the electrons may occupy separated by bands of forbidden energies. the electrons in the n = 2 shell and n = 1 shell may begin to interact and also split into a band of allowed energies.3) If the equilibrium interatomic distance is r0. If these atoms brought closer together. (Fig.The case for containing more than one electron
Suppose the atom contains electrons up through n = 3 energy level. so that this l l may also split i h hi level l li into a band of allowed energies. If the aqtoms are initially very far apart. the outermost electrons in the n = 3 together energy shell will begin to interact initially.
8
. if the atoms become sufficiently close together. If the atoms continue to mover closer together. the electrons in adjacent atoms will apart not interact and will occupy the discrete energy levels. so that this energy level will split into a band of allowed energies.
9
.3 Schematic showing the splitting of three energy states into allowed bands of energies.The case for containing more than one electron
This energyband splitting and the formation of allowed and forbidden bands is the energyband theory of single crystal materials
Figure 3.
4a. The 3s state corresponds to n = 3 and l = 0 and contains two quantum p q states per atom. Ten of the fourteen silicon atom electrons occupy deeplying energy levels close to the nucleus.3. The four remaining valence electrons are relatively weakly bound and are the electrons involved in chemical reactions. reactions Figure 3. since the first two energy shells are completely full and are tightly bound to the nucleus.Band Splitting in Silicon Atom p g
The actual band splitting in a crystal is much more complicated than indicated in Figure 3. This state will contain two electrons at T = 0 K.
10
.4b shows the band splitting of silicon. The 3p state corresponds to n = 3 and l = 1 and contains six quantum states per atom. A schematic representation of an isolated silicon atom is shown in Figure 3. We need only consider n = 3 level for the valence electrons.
As the interatomic distance decreases, the 3s and 3p states interact and overlap. At the equilibrium interatomic distance, the bands have again split, but now four quantum states per atom are in the lower band and four quantum states per atom are i the upper b d in h band. At absolute zero degrees, electrons are in the lowest energy state, so that all states in the lower band (the valence band) will be full and all states in the upper band (the conduction band) will be empty. The bandgap energy Eg between the top of the valence band and the bottom of the conduction band is the width of the forbidden energy band. We have discussed qualitatively how and why bands of allowed and q y y forbidden energies are formed in a crystal. The formation of these energy bands is directly related to the electrical characteristics of the crystal, as we will see later in our discussion.
11
Band Splitting in Silicon Atom p g
Figure 3 4 (a) S h Fi r 3.4 ( ) Schematic of an isolated silicon atom. (b) The splitting of the ti f i l t d ili t Th plitti f th 3s and 3p states of silicon into the allowed and forbidden energy bands.
12
The KronigPenny Model g y
Figure 3.5a – The potential function of a single, noninteracting, oneelectron atom Figure 3.5b – S Fi 3 5b Several atoms are i close l in l and the potential functions overlap. Figure 3.5c – The net potential function for onedimensional singlecrystal material KronigPenny Model : idealized periodic potential representing a onedimensional single crystal
Figure 3.5 Figure 3.6 The onedimensional periodic potential function of the KronigPenny model.
13
take the 3 6 (0 ) i hi h V(x) 0 k h second derivative of Eq.
To obtain the solution to Schroedinger equation.13). The theorem states that all oneelectron wave functions for problems involving periodically varying potential k energy functions.4) (3 4) (3. we make use of a mathematical theorem by Bloch.
d 2u1 ( x) du ( x) 2 jk 1 (k 2 2 )u1 ( x) 0 2 dx dx 2mE 2 2
(3.3)
If we consider regionⅠin Fi id i Ⅰi Figure 3.6 (0<x<a) in which V( ) = 0.2) ( ) (3.5)
14
.(2. and substitute this into the Eq.1).1)
: a constant of motion. wave function must be of the form functions
( x) u ( x)e jkx
(3. u(x): periodic function with period (a+b)
From Chapter 2. (total solution) = (timeindependent solution) ⅹ (time dependent solution)
( x.(3. t ) ( x) (t ) u ( x)e jkx e j ( E / )t u ( x)e j ( kx( E / ) t )
(The travelingwave solution)
(3.
V ( x) V0)
2mV d 2u2 ( x ) du ( x) 2 jk 2 (k 2 2 2 0 )u2 ( x) 0 dx 2 dx
(3.
In region Ⅱ (b x 0.6) can be written as
(3.7)
Eq.8)
The solution to Eq.(3.4).6)
If we define d fi
2mV 2m ( E V0 ) 2 2 0 2 2
d 2u2 ( x ) du ( x) 2 jk 2 (k 2 2 )u2 ( x) 0 2 dx dx
(3.(3. both the ψ(x) and ∂ψ(x)/∂x must be continuous.10)
Since V(x) is everywhere finite.
15
.8). for region Ⅰ
u1 ( x) Ae j ( k ) x Be j ( k ) x (0 x a )
(3.9)
The solution to Eq. for region Ⅱ
u2 ( x) Ce j ( k ) x De j ( k ) x (b x 0)
(3.(3.
15) (3.17)
The periodicity and the continuity condition
u1 (a) u2 (b)
Ae j ( k ) a Be j ( k ) a Ce j ( k ) b De j ( k ) b 0
du1 dx
xa
du2 dx
x b
( k ) Ae j ( k ) a ( k ) Be j ( k ) a ( k )Ce j ( k ) b ( k ) De j ( k ) b 0
(3.16) ( ) (3. the continuity condition
u1 (0) u2 (0) A B C D 0
(3.18) (3.
At the boundary (x = 0).12) (3.13) (3 13)
du1 dx
x 0
du2 dx
x 0
( k ) A ( k ) B ( k )C ( k ) D 0
(3.19)
The result is
( 2 2 ) (sin a )(sin b) (cos a)(cos b) cos k (a b) 2
16
.14) (3.11) (3.
21)
Let the barrier width b → 0 and the barrier height V0→∞ but such that the →∞.
When E < V0.20)
Eq.(3. β is then imaginary.23)
Finally. the electron bound within the crystal. Eq.22)
We may define a parameter P’ as
P' mV0ba 2
(3.(3.19) can be written in terms of γ
2 2 (sin a)(sinh b) (cos a)(cosh b) cos k (a b) 2
(3.21) then reduces to
mV0ba sin a cos a cos ka 2 a
(3. product bV0 remains finite.
j
(3. we have the relation
P'
sin a cos a cos ka a
(3.24)
17
.
18
.7 The parabolic E versus k curve for the free electron.26)
k
Since the potential is zero. for h f l di l f d b
p 2 k 2 2 E 2 m 2m
(3. and is also referred to as a wave number.The kSpace Diagram p g
For the case of V0 = 0.
2m( 1 mv 2 ) p 2mE 2 k 2 2
(3. Therefore parameter k is related to the particle momentum f the free electron. from Eq. the total energy is equal to the kinetic energy.25) (3.(3.28)
Figure 3.27)
P is the particle momentum.24)
cos a cos ka
(3.
(3.8c. We may define the left side of Eq.24) as
f ( a ) P ' sin a cos a a
(3. Eq.The Ek Relation
As the P’ increases. The parameter α is related to the total energy E of the particle.
19
. particle A plot of the energy E of the particle as a function of the wave number can be generated from Figure 3.30) (3 30)
f(αa) must be bounded between +1 and 1. the particle becomes more tightly bound to the potential well or atom.29) (3 29)
Now.24) becomes
f ( a ) cos ka k
(3.(3.
Figure 3. The shaded areas show the allowed values of (αa) corresponding to real values of k.(3. and (c) the entire f(αa) function.(3.8 A plot of (a) the first term in Eq.29).
20
.29). (b) the second term in Eq.
21
.8 The E versus k diagram generated 38 from Figure 3.8. The allowed energy bands and forbidden energy bandgaps are indicated.Figure 3.
as the energy increases. we fi d αa = 2.628.2 p
To determine the lowest allowed energy bandwidth. We see from Figure 3.
22
. Assume that the coefficient P’ = 10 and that the potential width a = 5 Å. we have or
2mE2 a 2 22 2 (1. i l d find 6 d h for ka For αa = π. We see that f k = π.053 0.8c that.407 1019 J 1.(3. p Solution To find the lowest allowed energy bandwidth.447 V Comment . the width of the allowed bands increase from this KronigPenny model.054 1034 ) 2 E2 2. αa = π.628 rad.053 eV.50 1.Example 3.8c).68×1019 J =1. we need to find the difference i αa values as k changes fr m 0 t π (S Figure 3. Eq. we find that E1 =1.50 eV 2ma 2 2(9.111031 )(5 1010 ) 2
a
cos a
For αa = 2. The allowed energy bandwidth is then E E2 E1 1 50 1 053 0 447 eV 1.29) becomes sin a
1 10
By trial and error. For k diff r in l ka h from to (See Fi r 3 8 ) F r ka = 0.
And the entire E versus k pl t is contained within –π/a<k<π/a. cos ka cos(ka 2n ) cos(ka 2n )
(3.10 Th E versus k di Fi 3 10 The diagram showing 2π displacements of several sections of allowed energy bands.
Figure 3.31) (3 31)
Therefore.
23
. r plot i nt in d ithin / <k< /
2
Figure 3. we can displace portions of the curve by 2π.11 The E versus k diagram in the reducedzone representation.
The cosine function is periodic.
We require the wave functions to be periodic in x with period L. the wave function is the standing wave
n ( x) A sin( nx / L)
where n is a positive integer.Ne y ee ec o Nearly Free Electron Model (Kittel.C p
The free particle Schrodinger equation in onedimension is 2 2 ( x) (6) E ( x) 2 2m x If the electrons are confined to a linear solid of length L.
(7)
It is convenient to introduce wave functions that satisfy periodic boundary conditions. Chap 6) ode ( e. Thus
( x) ( x L)
(8)
Wave functions satisfying the freeparticle Schrodinger equations and the periodicity condition are of the form of a traveling plane wave:
( x) exp( jkx)
(9)
24
.
(c) Corresponding probability functions. (b) 26 levels Corresponding wave functions.6 Particle in an infinite potential well: (a) Four lowest discrete energy levels.Figure 2.
25
.
L L
(10)
On substituting (9) in (6) we have the energy Ek of the orbital with wavevector k: (12) 2
Ek 2 k 2m
The magnitude of the wavevector is related to the wavelength
by k = 2π/ .provided that the component of the wavevector k satisfy
k 0. The components of k are the quantum numbers of the problem. where n is a positive or negative integer. for
( x L) exp[ik ( x L)] exp[ j 2n ( x L) / L] exp( j 2n x / L) exp( j 2n ) exp( j 2n x / L) exp( jkx) ( x) (11)
2 4 . We confirm that these values of k satisfy (8).
Any component of k is of the form 2nπ/L.
26
.
the allowed energy values are distributed essentially continuously from zero to infinity. This Thi model answers almost all the qualitative questions about the d l l ll h li i i b h behavior of electrons in metals. We saw in Chapter 6 that
2 2 Ek k 2m
(1)
where.Nearly Free Electron Model (Kittel.
27
. The band structure of a crystal can often be explained by the nearly free electron model for which the band electrons are treated as perturbed only weakly by the periodic potential of the ion cores.
k 0.
The free electron wave functions are of the form
( x) exp( jkx)
(3)
2 4 . for periodic boundary conditions over a linear solid of length L. Chap7)
On the free electron model. L L
(2)
They represent running waves and carry momentum p = ħk.
but with an energy gap at k = ±π/a. The region in kspace between π/a and π/a is the first Brillouin zone of this lattice. 2. (At Bragg reflection.) These energy gaps are of decisive i do i i Fi 2 ) Th fd i i significance in determining whether a solid is an insulator or a conductor.
We know that Bragg reflection is a characteristic feature of wave propagation in crystals. 2. Other energy gaps occur for other values of the integer n. in (a) for entirely free electrons and in (b) for electrons that are nearly free. The low energy portions of the band structure are shown qualitatively in Fig. as in Fig. 28
. The Bragg condition (k+G)² = k² for diffraction of a wave of wavevector k becomes in one dimension 1 n (4) k G 2 a where G = 2πn/a is a reciprocal lattice vector and n is an integer.
We explain physically the origin of energy gaps in the simple problem of a linear solid of lattice constant a. Bragg reflection of electron waves in crystals is the cause of energy gaps. wavelike solutions of the Schrodinger equation d not exist. The first reflections and the first energy gap occur at k = ±π/a.
The zone boundaries of the linear lattice are at k = π/a. The shortest reciprocal lattice l i vectors from the origin are b and –b.
29
.11.Brillouin Zones
A linear lattice in one dimensions in Fig.
Figure 11 Crystal and reciprocal lattices in one dimension. The basis vectors in the reciprocal lattice is b. the perpendicular bi f h i i d b h di l bisectors of f these vectors form the boundaries of the first Brillouin zone. of length equal to 2π/a. The boundaries are at k = +π/a. where a is the primitive axis of the crystal lattice.
() exp( j x / a) exp( j x / a) 2cos( x / a).and l f di i h d leftdirected traveling waves. () exp( j x / a) exp( j x / a) 2 j sin( x / a). We can form two different standing waves from the two traveling waves exp(±jπx/a) namely exp(±jπx/a). The timeindependent state is represented by standing waves. the wave functions are made up of equal parts of waves traveling to the right and to the left. d li
30
. and vice versa. When the Bragg reflection condition k = ±π/a is satisfied by the wave vector. At these special values of k.
(5)
The standing waves are labeled (+) or () according to whether or not they change sign when –x is substituted for x. Both standing waves are composed of equal parts of right. a vector wave traveling to the right is Braggreflected to travel to the left.
The wave functions at k = ±π/a are not the traveling waves exp(jπx/a) or exp(jπx/a) of free electrons. A wave that travels neither to the right nor to the left is a standing wave: it doesn’t go anywhere. Each subsequent Bragg reflection will reverse the direction of travel of the wave.
Figure 2 (a) Plot of energy ε versus wave vector k for a free electron. (b) Plot of energy versus wave vector for an electron in a monatomic linear lattice of lattice constant a. other gaps are gg g p found at ±nπ/a. The energy gap Eg shown is associated with the first Bragg reflection at k = ±π/a. for integral values of n.
.
. attractive
32
. exp( jkx)exp(jkx) 1. 2 . The probability density ρ of a particle is ψ*ψ=ψ². a. for this we have
() () cos 2 x / a
2
The function piles up electrons (negative charge) on the positive ions centered at x = 0. . The potential energy of an electron in the field of a positive ion is negative so that the force between them is attractive. i Fig. charge density is constant. The ion cores bear a net p p positive charge g because the atoms are ionized in the metal. 2a. The charge density is not constant for linear combinations of plane waves. in Fi 3 h th potential n r is l t Figure 3a pictures the variation of the electrostatic potential energy of a conduction electron in the field of the positive ion cores. This is the origin of the energy gap.Origin of The Energy Gap
The two standing waves ψ(+) and ψ() pile up electrons at different regions. with the valence electrons taken off to form the conduction band. negative. and therefore the two waves have different values of the potential energy. Consider the standing wave ψ(+) in (5). 3. For a pure traveling wave exp(jkx) we have ρ = exp(jkx)exp(jkx) = 1 so that the exp(jkx). where the p t nti l energy i lowest.
ψ(+)² ∝ cos² πx/a. The wave function ψ(+) piles up i ² / ² / d for t li Th f ti il electronic charge on the cores of the positive ions. This figure is the key to understanding the origin of the energy gap.
. thereby lowering the potential energy in comparison with the average potential energy seen by a traveling wave. The wave function ψ() ψ( ) piles up charge in the region between the ions. and f a traveling wave. (b) Distribution of probability density ρ in the lattice for ψ()² ∝ sin² πx/a. thereby raising the potential energy in ions comparison with that seen by a traveling wave.Figure 3 (a) Variation of potential energy of a conduction electron in the field of the ion cores of a linear lattice.
ρ( ) ρ( ) Just below the energy gap at points A in Fig. we find that the potential energy of ρ(+) is lower than that of the traveling wave. j g p p ψ( )
34
. whereas the f i l h h f h li h h potential energy of ρ() is higher than the traveling wave. We have an energy gap of width Eg if the energies of ρ() and ρ(+) differ by Eg.
For the other standing wave ψ(). When we calculate the average or expectation values of the potential energy over these three charge distributions. ψ(). the probability density is
() () sin 2 x / a
2
which concentrates electrons away from the ion cores. and for a traveling wave. 3b we show the electron concentration for the standing waves ψ(+). the wave function is ψ(). 2 the wave function is ψ(+).
In Fig. and just above the gap at points B.
normalized over unit length of line. We write the potential energy of an electron in the crystal at point x as
U ( x) U cos 2 x / a
(6)
The firstorder energy difference between the two standing wave states is
Eg d U ( x)[ () () ] dxU
2 2 0
1
2 dxU cos(2 x / a)(cos 2 x / a sin 2 x / a ) U
We see that the gap is equal to the Fourier component of the crystal potential.
35
.Magnitude of The Energy Gap
The wave functions at the Brillouin zone boundary k = π/a are √2 cos πx/a and √2 sin πx/a.
3.2 Electrical Conduction in Solids
36
.
the valence band. is l l The 4N states in the lower band. h d i band. The Th upper energy b d the conduction b d i completely empty. each silicon atom is surrounded by eight valence electrons that are in their lowest energy state and are directly involved in the covalent bonding and all of the valence electrons are in the valence band. electrons
Figure 3.12 Twodimensional representation of the covalent bonding in a semiconductor at T=0K
37
. band. are filled with the valence electrons.The energy band and the bond model
At T = 0 K.
38
. (b) The splitting of the ti f i l t d ili t Th plitti f th 3s and 3p states of silicon into the allowed and forbidden energy bands.Band Splitting in Silicon Atom p g
Figure 3 4 (a) S h Fi r 3.4 ( ) Schematic of an isolated silicon atom.
a few valence band electrons may gain enough thermal energy to break the covalent bond and jump into the conduction band.The energy band and the bond model
As the temperature increases above 0 K.
39
.13 (a) Twodimensional representation of the breaking of a covalent bond.
Figure 3. g ( ) p g (b) Corresponding line representation of the energy band and the generation of a negative and positive charge with the breaking of a covalent bond.
bands Figure 3.14a shows the E versus k diagram of the conduction and valence bands at T = 0 K. We can also relate this bond breaking to the E versus k energy bands. as the negatively y g g y charged electron breaks away from its covalent bonding position. more electrons jump to the conduction b d and more positive “empty l t j t th d ti band. d iti “ t states” are created in the valence band. The energy states in the valence band are completely full gy p y and the states in the conduction band are empty. This means that. more covalent bonds are broken.The energy band and the bond model
The semiconductor is neutrally charged. As the temperature further increases. a positive charged “empty state” is created in the original covalent bonding position in the valence bond.
40
.
Figure 3.14b shows these same bands for T > 0 K. in which some electrons have gained enough energy to jump to the conduction band and have left empty states in the valence band.
41
. We are assuming at this point that no external forces are applied so the electron and “empty state” distribution are symmetrical with k.14 Th E versus k Fi 3 14 The diagram of the conduction and valence bands of a semiconductor at (a) T = 0 K and (b) T > 0 K.
Figure 3.
we considered the individual ion velocities. 3.Drift Current
Current is due to the net flow of charge. If we had a collection of positively charged ions with a volume density N (cm3) and an average drift velocity vd (cm/s).33) is taken y q( ) over a unit volume so that the current density J is still in units of A/cm2.32) (3 32)
If.14a. then we could write the drift current density as
J q vi
i 1 N
(3. t lf is li d
42
.(3. instead of considering the average drift velocity. a net drift of electrons in the conduction band will give rise to a current. is an even function of when no external force i applied. as shown in Fig. then the drift current density is
J qNvd N A/cm A/ 2
(3.33)
where vi is the velocity of the ith ion.
Since electrons are charged particles. The summation in Eq. The electron distribution in the conduction band.
there pp are empty energy states into which the electrons can move.
If an external force is applied to the electrons in the conduction band. it must gain energy.
43
. therefore. and dE is the increase in energy. dx is the differential distance the particle moves. This ff i Thi effect is expressed as d
dE Fdx Fvdt
(3.34)
where F is force. since there are as many electrons with a +k value as there are with a k value. the net drift current density due to these electrons is zero. v is the velocity. This result is certainly expected since there is no externally applied force. If a force is applied to a particle and the particle moves. because of the external force. electrons can gain energy and a net momentum.
Recall that k for a free electron is related to momentum so that.
35) (3 35)
where e is the magnitude of the electronic charge and n is the number of electrons per unit volume in the conduction b d A i the summation i l i l i h d i band.
Figure 3. pp
44
.
The electron distribution in the conduction band may look like that shown in Fig. which implies that the electrons have gained a net momentum. We may write the drift current density due to the motion of electrons as
J e vi
i 1 n
(3.15 The asymmetric g y distribution of electrons in the Ek diagram when an external force is applied. Again.15. 3. h i is taken over a unit volume so the current density is A/cm2.
Electron Effective Mass
The movement of an electron in a lattice will.36)
where Ftotal. be different from that of an electron in free space. is h l i d is h f h i l
Since it is difficult to take into account all of the internal forces. in general. The parameter m* is called the effective mass. acting on a particle in a crystal. and Fint are the total force.37)
where the acceleration a is now directly related to the external force. and the internal forces. We can write
Ftotal Fext Fint ma
(3. takes into account the particle mass and also takes into account the effect of the internal forces. respectively.
45
. In addition to an externally applied force space there are internal force in the crystal due to positively charged ions and negatively charged electrons which will influence the motion of electrons g y g in the lattice. The parameter a i the acceleration and m i the rest mass of the particle. the externally applied force. we will write the equation
Fext m*a
(3. Fext.
38)
(3.(3. 3. If we take the y derivative of Eq.38) can be written as where v is the velocity of the particle.11. (3. such as was shown in Fig. Eq.7. we will be dealing with allowed energy bands that r lm t mpt f l tr n nd th r n r b nd r full f l tr n th t are almost empty of electrons and other energy bands are f ll of electrons. = p2/2m = ħ2k2/2m. In a semiconductor material. (3 28) the energy and momentum are related by E Eq (3.28). we obtain Relating momentum to velocity. where m is the mass of the electron. The first derivative of E with respect to 46 k is related to the velocity of the particle.39)
. Recalling the Eq.
1 dE p v dk m
dE 2 k p dk m m
(3.28) with respect to k. consider the case of a free electron whose E – k curve was shown in Fig. The momentum and wave number k are related by p =ħk. To begin. 3 7 Fig 3.
We can also relate the effective mass of an electron in a crystal to the Ek curves.
so the second derivative function is a constant.
If we now take the second derivative of E with respect to k we have. 3. the mass is a constant.42)
47
d 2 E 2 2 dk m
(3.40) as The second derivative of E with respect to k is inversely proportional to the mass of the particle.40)
1 d 2E 1 2 2 dk m
(3. f h l is l ii i If we apply an electric field to the electron and use Newton’s classical equation of motion we can write motion. We may also note from Fig. For the case of a free electron.
F ma eE
(3.7 that d2E/dk2 is a positive quantity. We may rewrite Eq. (3. which implies h h i li that the mass of the electron i also a positive quantity.41)
.
43) (3 43)
Figure 3.16 (a) The conduction band in reduced k space. charge We may now apply the results to the electron in the bottom of an allowed energy band. Consider the allowed energy band in Fig 3.16a. 48 (b) The valence band in reduced k space. band Fig.
Solving for the acceleration. and the parabolic approximation. and the parabolic approximation.
. we have
a
The motion of the electron is in the opposite direction to the applied electric field because of the negative charge. 3 16a
eE m
(3.
46)
Comparing Eq. 3. the h parameter C1 is a positive quantity. the curvature of the curve in Fig.(3.45)
We may put Eq.41). We may write
49
. Since E > Ec . we may equate ħ2/2C1 to the mass of the particle.(3.44)
The Th energy Ec i the energy at the b is h h bottom of the b d Si f h band. However particle However.
The energy near the bottom of this energy band may be approximated by a parabola.16a will not. Fig 3 16a not general be the same as the curvature of the freeparticle curve. We may write
E Ec C1 (k ) 2
(3.
Taking the second derivative of E with respect to k from Eq.46) with Eq.(3.45) in the form
1 d 2 E 2C1 2 2 2 dk
(3.(3.44). we obtain Eq (3 44)
d 2E 2C1 2 dk
(3. in general. just as that of a free particle.
1 d 2 E 2C1 1 2 * 2 dk 2 m
(3. we may write the acceleration as
a
where m*n is the effective mass of the electron. The effective mass m*n of the electron near the bottom of the conduction band is a constant. If we apply an electric field to the electron in the bottom q w pp y d b of the allowed energy band.48)
50
.47)
The effective mass relates the quantum mechanical results to the classical force equations.
eE * mn
(3.
can also be thought of as a classical particle whose motion can be modeled using Newtonian mechanics.
51
. all valence electrons may gain thermal energy. Figure 3.17 shows the movement of valence electrons in the crystal 3 17 alternately filling one empty state and creating a new empty state.Concept of The Hole p
When a valence electron is elevated into the conduction band. a positively charged “empty state” is created. a motion equivalent to a positive charge moving in the valence band. For T > 0 K. it may hop into the empty state. if a valence electron gains a small amount of thermal energy. as we will see. q p g g This charge carrier is called a hole and. The Th movement of a valence electron i f l l into the empty state i equivalent h is i l to the movement of the positively charged empty state itself.
17 Visualization of the movement of a hole in a semiconductor.
52
.49)
i (filled)
where the summation extends over all filled state. 3. such as shown in Fig. W m r rit Eq (3 49) in th form nt r large number f t t We may rewrite Eq.49) the f rm
J ( e
i ( total)
v ) ( e
i
vi ) e
i (empty)
i (total)
v e
i
vi
(3. This summation is inconvenient since it extends over a nearly full valence band and takes into account a very l r n mb r of states.50)
i (empty)
Figure 3.14b. (3. can be written as J e vi (3.
The drift current density due to electrons in the valence band.
1 dE
. there i a corresponding electron with a velocity v. is zero.51)
We can write the drift current density from Eq. The individual electrons can be thought of as moving with a velocity as given by Eq. all available states are occupied by electrons.39): The band is symmetric in k and each state is occupied so that. then. The net drift current density from a completely full band. l ih l i h is di l ih l i Since the band is full.50) for an almost full J e vi band as (3.52)
i (empty)
where the vi in the summation is the v( E ) ( )( ) associated with the empty dk 53 state.
If we consider a band that is totally full. or
e
i (t t l) (total)
1 dE p 1 dE v v( E ) ( )( ) dk m dk
v
i
0
(3.(3. for every electron with velocity v. (3. the distribution of electrons with respect to k cannot be changed with an externally applied force.
Eq. 3.18a shows the valence band with conventional electronfilled states and empty states while Fig. electron filled states. 3. or neutrally charged.
Figure 3. Fig 3.(3.18 (a) Valence band with conventional electronfilled states and empty states. Fig.
54
. This concept is shown in Fig 3.17. (b) Concept of positive charges occupying the original empty states.18b shows the new concept of positive charges occupying the original empty states. This concept is consistent with the discussion of the positively charged “empty state” in the valence band as shown in Fig.18.52) is entirely equivalent to placing a positively charged particle in the empty states and assuming all other states in the band are empty.
54)
. Since E < Ev for electrons in this band.
The vi in the summation of Eq.
( E Ev ) C2 (k ) 2
(3. 3. then the parameter C2 mist be a positive quantity band quantity. Now consider an electron near the top of the allowed energy band shown in Fig.55)
55
d 2E 2C2 2 dk
(3. (3.53).16b.(3. we obtain We may rearrange this equation so that
1 d 2 E 2C2 2 2 dk 2
(3. Taking the second derivative of E with respect to k from Eq.52) is related to how well this positively charged particle moves in the semiconductor.53) (3 53)
The energy Ev is the energy at the top of the energy band. The energy near the top of the allowed energy band may again be approximated by a parabola so that we may write.
(3.
1 d 2 E 2C2 1 2 * 2 dk 2 m
(3. m* i now a negative quantity.
Comparing Eq.
56
.41). We have argued that C2 is a positive quantity. which now implies that m* is a negative quantity.(3. we will have
(3. so we may write H is i i i eE eE (3.55) with Eq.57) F m*a eE However. we may write where m* is again an effective mass.58) a * * m  m  An l A electron moving near the top of an allowed energy b d moves i the i h f ll d band in h same direction as the applied electric field.56)
If we again consider an electron near the top of an allowed energy band and use Newton’s force equation for an applied electric field.
(3.56) is associated with each state. has a positive effective mass (denoted by m* ) and a p positive electronic charge so it will move in the same direction as an applied field. i d ih h We now can model this band as having particles with a positive electronic charge and a positive effective mass. then. mass The density of these particles in the valence band is the same as the density of empty electronic energy states. provided that a positive electronic charge is associated with each state and that the negative of m* from Eq.
The net motion of electrons in a nearly full band can be described by considering just empty states. The hole.
57
. This new particle is the hole.
(17)
1. But the true wavevector kh of the hole is – ke. If the band is filled all pairs of orbitals k and –k are filled. This result is surprising: the electron is missing from ke and the position of the hole is usually indicated graphically as situated at ke . 7.ke .
The total wavevector of the electrons in a filled band is zero: Σk = 0. which is the wavevector of g the point G if the hole is at E. kh = . The wavevector – ke enters into selection rules for photon absorption. e.Holes (Kittel. it follows that the Brillouin zone of the lattice also has inversion symmetry. the total wavevector of t e a electron s ss g o a o b ta o wavevecto t e tota wavevecto o the system is – ke and is attributed to the hole. The reason is given in five steps in the boxes that follow. 8)
The properties of vacant orbitals in an otherwise filled band are important in semiconductor physics and in solid state electronics. as in Fig. A hole acts in applied electric and magnetic fields as if it has a positive charge +e. and the total wavevector is zero.
58
. Chap. Vacant orbitals in a band are commonly called holes. If an e ect o is missing from an orbital of wavevector ke . This result follows from the geometrical symmetry of the Brillouin zone: every fundamental lattice type has symmetry under the inversion operation r → – r about any lattice point.
ke. 59
. Absorption of a photon of energy ħ and negligible wavevector takes an electron Fi 7 Ab i f h f ħω d li ibl k l from E in the filled valence band to Q in the conduction band. For the entire system the total wavevector after the absorption of the photon is ke + kh = 0 0.ke . and this is the wavevector we must ascribe to the hole if we describe the valence band as occupied by one hole. Thus kh = . it becomes the wavevector of the electron at Q.Figure 7. so that the total wavevector is unchanged by the absorption of the photon and the creation of a free electron and free hole. the wavevector of the hole is the same as the wavevector of the electron which remains at C. If ke was the wavevector of the electron at E. The total wavevector of the valence band after the absorption is .
Eh(kh) = – Ee(ke) . (18)
Let the zero of energy of the valence band be at the top of the band.ke)=Eh (.
60
. and we either say that the hole has wavevector – ke or that the band with one missing electron has total wavevector – ke. The energy of the hole is opposite in sign to the energy of the missing electron. The lower in the band the missing electron lies. Thus if the band is symmetric. We construct in Fig. This hole band is a helpful representation p p because it appears right side up. 2. because it takes more work to remove an electron from a low orbital than from a high orbital. 8 a band scheme to represent the p p p properties of hole.ke)=Eh (kh).The hole is an alternate description of a band with one missing electron. the higher the energy of the system. Ee(ke)=Ee (.
but opposite in sign.Figure 8 Th upper half of the f The h lf f h figure shows the h l b d that simulates the h h hole band h l h dynamics of a hole. The wavevector and energy of the hole are equal.
61
. constructed by inversion of the valence band in the origin. to the wavevector and energy of the empty electron orbital i the valence b d W d d f h l bi l in h l band. We do not show the disposition of the electron removed from the valence band at ke .
↓).k if the spinorbit interaction is neglected. so that mh is positive. but with spin orbit interaction. p valence band.
62
. From Fig. mh = – m e . 9 we see that ∇Eh (kh)=∇Ee(ke) so that vh(kh)= ve(ke) ). ). and for the hole band this has the opposite sign to that for an electron in the p g . vh = ve . (19)
3. 3
The velocity of the hole is equal to the velocity of the missing electron. Without a center of symmetry.
¹Bands are always symmetric under the inversion k→ . 4.↑)=E (k. Even with spinorbit interaction. (20)
We show below that the effective mass is inversely proportional to the curvature d²E/dk². bands are always symmetric if the crystal structure permits the inversion operation. the bands are symmetric if we spinorbit compare subbands for which the spin direction is reversed: E (k. Near the top of the valence band me is negative.
the velocity vx . moving the hole to the state E. Th force on the electrons is in the –kx direction and all electrons make transitions together in the kx directtion.
63
.Figure 9. (c) After a further interval the electrons move farther along in k space and the hole is now at D D. (a) At t = 0 all states are filled except F at the top of the band. is zero at F b t because dE/dkx = 0 (b) An electric field E i applied in th ± di ti The A l t i fi ld is li d i the ±x direction.
dk dk h 1 e(E v h B ) dt c
dk e 1 e(E v e B ) dt c
(21)
This comes from the equation of motion q (CGS) (22)
that applies to the missing electron when we substitute –kh for ke and vh for ve. The equation of motion for a hole is that of a particle of positive charge e. The current is shown in Fig. The positive charge is consistent with the electric current carried by the valence band of Fig. (23)
which is just the current of a positive charge moving with the velocity ascribed to the missing electron at E.5. 10.
64
. 9: the current is carried by the unpaired electron in the orbital G: J = (e)v(G) = (e)[v(E)] = ev (E) .
direction the direction of the electric field.Figure 10. Motion of electrons in the conduction band and holes i the Fi 10 M i f l i h d i b d d h l in h valence band in the electric field E. field
65
. The hole and electron drift velocities are in opposite directions. but their electric currents arc in the same direction.
If an electric field is applied.Metals. so the material can exhibit a very high electrical conductivity. Semiconductor : h relatively f electrons or h l near the b S i d has l i l few l holes h bottom or top of the bands.
66
. Insulator : energy bands are ether completely empty or completely full. The bandgap energy Eg is usually on the order of 3. Metal : partially full band in which there are many electrons available for p y y conduction or conduction and valence bands overlap at the equilibrium interatomic distance. The resistivity can be controlled and varied over may orders. there are very few thermally generated electrons and holes in an insulator. The bandgap energy is usually on the order of 1 eV. Insulators. and Semiconductors
Each crystal has its own energyband structure. the electrons or holes move through the crystal. There are large numbers of electrons as well as large numbers of empty energy states into which the electron can move. so that at room temperature.5 to 6 eV or larger.
Figure 3.Figure 3. and (c) the bandgap energy between the two allowed bands.20 Allowed energy bands showing (a) an almost empty band. 67
.19 Allowed energy bands showing (a) an empty band. (b) a completely full band band. and (c) the bandgap energy between the two allowed bands. (b) a almost full band band.
Figure 3.
68
.21 Two possible energy bands of a metal showing (a) a partially filled band and (b) overlapping allowed energy bands.
3.3 Extension to Three Dimensions
69
.
many details will not be considered. One problem encountered in extending the potential function to a threethree dimensional crystal is that the distance between atoms varies as the direction through the crystal changes.
70
. b d di i l l in f h Ek l bandgap energy. we will extend these concepts to three dimensions and to real crystals. In this section. We will qualitatively consider particular characteristics of the threedimensional crystal i terms of the E k plots. We must emphasize that we will only briefly touch on the basic three threedimensional concepts. and d effective mass.Extension to Three Dimensions
The basic concept of allowed and forbidden energy bands and the basic concept of effective mass have been developed in the last sections. therefore.
l
Figure 3.
71
.22 Two (100) plane of a face centered facecentered cubic crystal showing the [100] and [110] directions.Extension to Three Dimensions
Figure 3.22 shows a facecentered cubic structure with the [100] and [110] directions indicated. Electrons travelling in the different directions encounter different potential patterns and therefore different kspace boundaries. The E versus k diagrams are in general a function of the kspace direction in i a crystal.
k In the case of diamond or zincblende lattices. axis It is normal practice to plot the [100] direction along the normal +k axis and to plot the [111] portion of the diagram so the +k point to the left. the maxima in the valence band energy and minima in the conduction band energy occur at k = 0 or gy gy along one of these two directions. Note that in place of the usual positive and negative k axes.
72
.23 shows an Ek diagram of gallium arsenide and of silicon.The kSpace Diagrams of Si and GaAs Space
Figure 3. The Ek diagram for the onedimensional model was symmetric in k so that no new information is obtained by displaying the negative axis. but do not show many of the details more appropriate for advancedlevel courses. we now show two diff different crystal di i l directions. These simplified diagrams show the basic properties considered in this text.
73
.The kSpace Diagrams of Si and GaAs Space
Figure 3.23 Energy band structures of (a) GaAs and (b) Si.
In GaAs.23a shows the Ek diagram for GaAs. The valence band maximum and the conduction band minimum both occur at k = 0. The electrons in the conduction band tend to settle at the minimum conduction band energy which is at k = 0.
74
. GaAs and other direct bandgap materials are ideally suited for use in semiconductor lasers and other optical devices. holes in the valence band tend to congregate at uppermost valence band b d energy. Transitions between the two allowed bands can take place with no change in p g crystal momentum. the minimum conduction band energy and maximum valence band energy occur at the same k value As semiconductor with value. this property is said to be direct bandgap semiconductor.The kSpace Diagrams of Si and GaAs Space
Figure 3. This direct nature has significant effect on the optical properties of the material. Similarly.
The minimum in the conduction band energy occurs not at k = 0. When electrons make a transition between the conduction and valence bands. A transition in an indirect bandgap material must necessarily include an interaction with the crystal so that crystal momentum is conserved. The maximum in the valence band energy occurs at k = 0 as before. we must invoke the law of conservation of momentum. The difference between the minimum conduction band energy and the maximum valence band energy is still defined as the energy bandgap Eg.The kSpace Diagrams of Si and GaAs Space
The Ek diagram for silicon is shown in Fig.23b. A semiconductor whose maximum valence band energy and minimum conduction b d energy d not occur at the same k value i i d i band do h l is called an indirect bandgap semiconductor. 3.
75
. but along the [100] direction.
is l d h ff i f h l We may note from Fig.Additional Effective Mass Concepts
The curvature of Ek diagrams near the minimum of the conduction band energy i related to the effective mass of the electron.76
. by Eq. For the onedimensional E versus k diagram the effective mass was defined one dimensional diagram.23. 3. the effective mass parameters used in calculations will be a kind of statistical average that is adequate for most device calculations. so the effective mass of silicon an electron in the conduction band of GaAs will be smaller than in silicon.41) as 2
1 1 d E 2 m dk 2
A complication occurs in the effective mass concept in a real crystal.(3. In later sections and chapters. A threedimensional crystal can be described by three k vectors. that the curvature of conduction band at its minimum val e minim m value for GaAs is larger than that of silicon. y y The curvature of Ek diagram at the conduction band minimum may not be the same in the three k directions.
A Linear Lattice in One Dimension e ce O e e so
L = 6a L is the length of a linear lattice and a is the lattice constant of the lattice.
L = 6a
a
k=2 / k=2 / k=2 / k=2 / k=2 /
0 0
2 /6a /3a = 2 /L ψ(k) = expjkx : travelling wave 4 /6a 6 /6a 2 /3a = 4 /L /a = 6 /L
ψ(k) = 2coskx : standing wave 6 /6a /a = 6 /L ψ(k) = 2jsinkx : standing wave k = 2 / 8 /6a 4 /3a = 8 /L
k=2 / k=2 /
10 /6a
5 /3a = 10 /L
12 /6a 2 /a = 12 /L ψ(k) = 2coskx : standing wave
∆k = 2 /3a  /3a = /3a = 2 /6a = 2 /L If L = Na and N = 107, ∆k = 2 /L = 2 /Na is very small. → The k is quasi continuous, The total number of wave vector k in an allowed band is N.
Ek Diagram in The ReducedZone
Representation
.
we indicated that the band of allowed energies was actually made up of discrete energy levels. The number of carriers that can contribute to the conduction process is a p function of the number of available energy or quantum states since. only one electron can occupy a given quantum state.4 Density of States Function y
We wish to describe the current–voltage (JV) characteristics of semiconductor devices. We must determine the density of these allowed energy states as a function of energy i order to calculate the electron and h l f i f in d l l h l d hole concentrations. an important step in the process i to determine th number of electrons and h l th t will be is t d t i the b f l t d holes that ill b available for conduction. Since the current is due to the flow of charge. by the Pauli exclusion principle.3.
81
. state When we discussed the splitting of energy levels into bands of allowed and g g forbidden energies.
z )
elsewhere
where the crystal is assumed to be a cube with length L.3. y . z ) 0
for 0 x L 0 y L 0 zL
(3.1 mathematical Derivation
We will consider a free electron confined to a threedimensional infinite potential well. The potential of the infinite potential well is define as
V ( x.
where nx. ny. y g
Schrodinger’s wave equation in three dimensions can be solved using the separation of variables technique. where the potential well represents the crystal. g
2mE 2 2 2 2 2 k 2 k x2 k y k z2 nx n y nz2 n2 2 L L
2
2
(3.59)
V ( x. y .4.60)
82
. and nz are integers and n2 = nx2+ny2+nz2.
The energy E and the values of allowed k
nx ny 0 nz 0 n 2 0. and nz are integers and n2 = nx 2+ ny 2+ nz 2. 1 1 0 0 0 0 E1 = 2 / 2m ( 0 0 1 1 0 0 0 0 0 0 1 1
2 2 ) . the total num ber of quantum states = 6 L
2 2 ) .60) E ( 2 / 2m)k 2 ( 2 / 2m)(k x2 k y k z2 ) ( 2 / 2m)(2 / L) 2 n 2 (3 60)
nx 1 1 1 1 1 1 1 1 0 0 0 0
ny 1 1 1 1 0 0 0 0 1 1 1 1 1
nz 0 0 0 0 1 1 1 1 1 1 1 1 1
n 2 2.A Cubic Lattice in Three Dimensions
The energy of electrons in a three dimensional lattice can be written by where nx. 0
2 (3.
E 0 = 0. ny. L 83 the total num ber of quantum states = 12 E2 = 2 / m(
. the total num ber of quantum states (k) = 1 n 2 1.
(b) The positive oneeight of the spherical k space
84
.24 (a) A twodimensional array of allowed quantum states in k space.Figure 3.
example by 2 2 2 k x1 k x (nx 1) nx (3.
The distance between two quantum states in the kx direction for example.63)
85
.62)
A differential volume in k space is shown in Figure 3. the volume Vk of a single quantum state is
2 Vk L
3
(3. so the differential density of quantum state in k space can be written as
gT (k )dk 2
4 k 2 dk
2 / L
3
(3.61) (3 61) L L L Generalizing this result to three dimensions.24b and is given by 4 k2dk. is given direction.
We can now determine the density of quantum states as a function of energy E. Equation (3.64)
Equation (3.65b)
86
.64) gives the density of quantum states as a function of momentum.
The factor 4 k2dk i again the diff Th f is i h differential volume and the factor (2 /L)3 i the volume i l l d h f is h l of one quantum state. through the parameter k.65a)
1 k 2mE
(3.
or
2mE k 2
2
(3.63) may be simplified to
k 2 dk 3 gT (k )dk L 3
(3. For a free electron the parameters E and k are related by electron.
Equation (3.67) becomes
4 L3 gT ( E )dE 3 (2m)3/ 2 E dE h
(3.
87
. the number of energy states between E and E +dE is given by
L3 2mE 1 m gT ( E )dE 3 2 dE 2E
(3.68) gives the total number of quantum states between the energy E q ( )g q gy and E +dE in the crystal space volume of L3.68)
Equation (3.64).66)
Then.
The differential is
1 m dk dE 2E
(3. substituting the expressions for k2 and dk into Equation (3.67)
Since ħ = h/2 .
This density function is really a double density. As the energy of this free electron becomes small. in that the units are given in terms of states per unit energy per unit volume.
88
. then we will obtain the density of quantum states per unit volume of the crystal.69)
The density of quantum states is a function of energy E. the number of available quantum states decreases.
If we divided by the volume L3.68) then becomes
4 (2m)3 / 2 g (E) h3
E
(3. Equation (3.
4.71)
The Th general form of the E versus k relation f an electron in the bottom of a lf f h l ti for l t i th b tt f conduction band is the same as the free electron.70)
*
where Ec is the bottom edge of the conduction band and mn is electron effective mass. except the mass is replaced by the effective mass. “free” conduction electron model 89
.3.70) may be rewritten to give 2k 2 E Ec * 2mn
2k 2 (3. so we may write
2 k 2 E Ec * 2mn
(3. Equation (3.4.28) ) (cf . E 2m
(3. 3.2 Extension to Semiconductors e s o o Se co duc o s
The parabolic relationship between energy and momentum of a free electron is given as E = p2/2m = ħ2k2/2m. We can then think of the electron in the bottom of the conduction band as being a “free” electron with its own particle mass. The E versus k curve near k = 0 at the bottom of the conduction band can be approximated as a parabola.
since the hole is also confined in the semiconductor crystal and can be treated as a free particle. We may also approximate the E versus k curve near k = 0 by a parabola for a “free” hole.72)
is valid for E ≥ Ec .73) may be rewritten to give q ( ) y g
2 k 2 E E 2 m* p
(3.
The density of quantum state in the valence band can be obtained by using the same infinite potential well model. the number of available quantum states also decreases.73) (3 73)
Equation (3.+
The density of allowed electronic energy states in the conduction band
* 4 (2mn )3/ 2 gc ( E ) h3
E Ec
(3.74)
90
. so that
2 k 2 E E 2 m* p
(3. As the energy of electron on the conduction band decreases.
4 (2m* )3/ 2 p h
3
E E
( (3. t at
g ( E )
is valid for E ≤ Ev . We may the generalize the density of states function from Equat o (3.25 states as a function of energy. Figure 3 25 shows the plot of the density of quantum 3.
Equation (3. so g(E) = 0 for Ev ≤ E ≤ Ec .69) to apply to the valence band.74) is of the same form used in the general derivation of the density of state function.
91
. so that o Equation app y t e v e ce b d.75) )
Quantum states do not exist within the forbidden energy band.
Figure 3.25 The density of energy states in the conduction band and valence band as a function of energy
92
.
5 Statistical Mechanics S s c ec cs
In dealing with large numbers of particles. the electrical characteristics will be determined by the statistical behavior f large number of electrons.3. In a crystal. b h i of a l b f l
93
. we are interested only in the statistical behavior of the group as a whole rather than in the behavior of each individual particle.
we must consider the laws that the particles obey. MaxwellBoltzmann probability function
Distinguishable No limit to the number of particles allowed in each energy state Ex) E ) Gas molecules in a container at fairly low pressure molec les fairl lo press re Indistinguishable No limit to the number of particles allowed in each quantum state Ex) Photons or black body radiation Indistinguishable One particle in one q p quantum state Ex) Electrons in a crystal
94
BoseEinstein function
FermiDirac probability f b b l function
In each case. There are three distribution laws determining the distribution obey of particles among available energy states. the particles are assumed to be noninteracting.1 Statistical Laws
In determining the statistical behavior of particles.3.5.
.
The energy E and the values of allowed k
nx n 2 0. 0 ny 0 nz 0 n 2 2. the total num ber of k = 12 L
. ny.A Cubic Lattice in Three Dimensions
The energy of electrons in a three dimensional lattice can be written by 2 E ( 2 / 2m)k 2 ( 2 / 2m)(k x2 k y k z2 ) ( 2 / 2m)(2 / L) 2 n 2 (3. nx 1 1 1 n 2 1. and nz are integers and n2 = nx 2+ ny 2+ nz 2.60) where nx. the total num ber of k = 1
2 2 ) . the total num ber of k = 6 L
2 2 ) . 1 1 0 0 0 0 E1 = 2 / 2m ( 0 0 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 0 E2 = 2 / m( ny 1 1 1 1 0 0 0 0 1 1 1 1 nz 0 0 0 0 1 1 1 1 1 1 1 1 1
95
E 0 = 0.
2 The Fermi Dirac Probability Function FermiDirac
A maximum of one particle is allowed in each quantum state by the Pauli exclusion principle. since the particles are indistinguishable. The total number of ways of arranging Ni particles in the i th energy level (where Ni ≤ gi ) is
( gi ) ( gi 1) ( gi ( N i 1))
However.3. Therefore.5. the Ni ! number of p g permutations that the particles have among themselves in any given arrangement do not count as separate arrangements.76)
gi ! (3.77) Wi N i !( gi N i )!
96
. the actual number of independent ways of realizing a distribution of Ni particles in the i th level is
gi ! ( gi N i )!
(3.
i 1
The maximum W is found by varying Ni among the Ei levels. we will keep the total number of p . which varies the distribution.
97
. where N N i is the total number electrons in the system. p particles and total energy constant. · · · . or
W
i 1
n
gi ! N i !( gi N i )!
(3.
Eq.77) gives the number of independent ways of realizing a distribution of Ni particles in the i th level. N2. Nn) indistinguishable particles among n energy levels is the product of all distributions.78)
The parameter W is the totaln number of ways in which N electrons can be arranged in this system. but at the same time. . (3. N3. The total number of ways of arranging (N1.
We may write the most probable distribution function as
N (E) = fF (E) g (E)
1 E EF 1 exp kT
(3. It gives the probability that a quantum state at the energy E will be occupied by an electron.
98
.
The number density N(E) is the number of particles per unit volume per unit energy. g(E) is the number of quantum states per unit volume per unit energy. (E) i th b f t t t it l it fF(E) is called the FermiDirac distribution or probability function. fF(E) is the ratio of filled to total quantum states at any energy E.79) (3 79)
where EF is called the Fermi energy.
3.3 The Distribution Function and The Fermi Energy
1 fF (E) E Ef 1 exp( ) kT
(3.5.79)
99
.
100
.
101
.31 The Fermi probability function versus energy f different for diff temperatures.1 1 1 f F ( E EF ) 1 exp(0) 1 1 2
Figure 3.
102
.fF(E) (E).
Figure 3. and the fF(E) probability of a state being empty.32  The probability of a state being occupied.33 The FermiDirac probability function and the MaxwellBoltzmann approximation. 1 .Figure 3.
Figure 3. so the FermiDirac distribution function becomes
( E EF ) f F ( E ) exp kT
(3.80) is known as the Maxwellboltzmann approximation.79) is much greater than unity. or simply the Boltzmann approximation. We may neglect the 1 in the denominator. The actual Boltzmann approximation is valid when exp[(EEF)/kT]>>1 .
103
.
Consider the case when E – EF >> kT. where the exponential term in the denominator of Equation(3. This figure gives an indication of the range of energies over which the approximation is valid.80)
Equation (3.33 shows the FermiDirac probability function and the Boltzmann approximation. to the FermiDirac distribution function.