17 Semiconductor Physics

Bloch Electrons and Holes

Electron in the conduction band leaves a hole in the valence

band.
We treat the hole as a particle of charge = +e.
Few holes in the valence band are easier to study than the
many electrons in the band.

kmax

The momentum of a hole

For a completely filled band, the sum of the wave vectors k

of occupied states (i.e., the sum over the entire Brillouin
zone) is zero
(1)
k=0

Now let us remove an electron of wave vector ke from this

completely filled band. The new sum of wave vectors is

kBZ

( k) ke = ke = kh .
kBZ

Instead of summing over all electron states, we only

h = hk
e.
consider a hole of crystal momentum hk

(2)

The energy and effective mass of a hole

When a hole is created, the change in the energy of the

system is the negative of the energy of the removed particle,
therefore the energy of the hole satisfies
Eh (kh ) = Ee (ke )

Close to the top of the valence band, the energy dispersion

for electrons
E = Emax k kmax 2 ,

>0

(4)

2
1
= 2
me
h
According to Eq. 3, the hole has a positive effective mass
mh = me .
The top of the valence band is the lowest energy of the hole.

gives negative effective mass

(3)

The group velocity of an electron

The group velocity in 3D is given by

1


1
vk = k E = (
i+
j+
k) E
ky
kz
h
h kx

(5)

For an electron close to the top of the valence band

E = Emax k kmax 2
E
=
kx

[(kx kx,max )2 + (ky kx,max )2 + (kz kx,max )2 ]

kx
(kx kx,max )

=
(kx kx,max )2
(kx kx,max )
kx
E
= 2(kx kx,max )
= 2(ki ki,max ), i = x, y, z
ki
2
ve = (k kmax )
h
(6)

The group velocity of a hole

Since Eh (kh ) = Ee (ke ), for the hole we have

1
1
1
vh = kh Eh (kh ) = kh Ee (ke ) = ke Ee (ke ) = ve
h
h
h
(7)
= vmissing electron

The semiclassical equation of motion

The force on an electron is given by

Fe =

dpe
= hke = e(E + ve B)
dt

(8)

For the hole, we substitute ke = kh and use ve = vh

Fh =

dph
= hkh = e(E + vh B)
dt

(9)

showing that the hole is a particle with a positive charge.

The current carried by the valence band

No magnetic field, E = Exi

At t = 0, kh = 0 kh = eEx t/h.
The hole carries current j that is parllel to E.
E

Ex
t>0

t=0
s1

kh

s2

The current is carried by the unpaired electron at the

orbital s2 .
j = (e)v(s2 ) = (e)[v(s1 )] = ev(s1 )

(10)

which is the current of a positive charge moving with the

velocity of the missing electron at s1 .

Drude Transport

The main failure of Drude theory. It neglected Pauli

exclusion principle which makes the Fermi energy extremely
high in metals because of the high density of electrons
In semiconductors or band insulators electrons and holes
may have low-densities, and to a very good approximation,
we can ignore Fermi statistics and Drude theory works well.
The semiclassical picture. We can write Drude transport
equation for electrons in the conduction band
m v
dv
= e(E + v B) e ,
dt

and for holes in the valence band

m v
dv
mh
= e(E + v B) h ,
dt

The mobility is defined as

me

v
e
=
E
m

(11)

(12)

(13)

In some materials the main charge carrier is the hole, which

changes the sigh of the Hall coefficient.

Ge

(a)

(b)

(a) A free electron and a hole arising from thermal ionization of an electron originally in a covalent bond,
(b) Motion of free electron (to the right) and hole (to the left) when an electric field is applied as shown.

A Survey of semiconductors
Semiconductors can be grouped in several classes.

Elemental semiconductors. Group IV: C (diamond), Si, Ge

and -Sn (gray tin)
Binary compounds

Group III-V: GaAs, InSb, GaP, InAs, ...

Group II-VI: CdS, ZnS, ...
Group IV-VI: (PbS), PbTe and tin sulfide (SnS)

Oxides. CuO, Cu2 O, ZnO

Layered semiconductors. Lead iodide (PbI2 ), molybdenum

disulfide (MoS2 ) and gallium selenide (GaSe)

Organic semiconductors. Polyacetylene [(CH2 )n ] and

carbon nanotubes

Magnetic semiconductors. Compounds containing magnetic

ions such as europium (Eu) and manganese (Mn).
Examples: EuS and Cd1x Mnx Te

What is a semiconductor

At T = 0 K, the highest occupied energy band, the valence

band (VB), is completely full

At finite temperatures, electrons are excited thermally to

the conduction band (CB) leaving holes at the top of the
valence band. The two band are partially full now

If the energy gap Eg < 2 eV, the number of excited

electrons is appreciable

If the energy gap Eg > 2 eV, the number of excited

electrons is negligible. The substance is insulator

Electrons in CB and holes in VB will contribute to current

if an electric field is applied

Conductivity of semiconductors < that of metal, because of

the small numbers of electrons

Conduction and valence bands of a pure semiconductor

Thermally excited
conduction electrons
Empty conduction band

Ec1
Eg Energy gap

Ev

Filled valence band

Empty valence band states
(holes)

At absolute zero

At temperature T > 0 K

Density of states for an intrinsic semiconductor

1 2 me 3/2
( 2 )
E Eg
2 2 h
1 2 mh 3/2
gh (E) = 2 ( 2 )
E, E > 0
h
2

ge (E) =

f (E) = 1/(1 + e(EEf )/kB T )

Ec1
g(E)
Ev

In the tail region E EF kB T

f (E) = e(EEf )/kB T ,
Maxwell-Boltzmann distribution

Electron concentration: intrinsic semiconductor

ge (E)dE = number of states in the energy range

(E, E + dE) per volume
f (E)ge (E)dE = density of electrons in that range

Note: f (E) gives the probability that an energy state E is

occupied

The concentration of electrons in the CB is

Ec2

n=

Ec1

f (E)ge (E)dE

Ec1 = Eg , energy of the bottom of the CB band

Ec2 = energy of the top of the CB band

(14)

1 2 me 3/2
( 2 )
E Eg dE
2 2 h
Ec1
Ec2

1 2 me 3/2
= 2 ( 2 ) eEf /kB T
eE/kB T E Eg dE
h
2
Ec1
Ec2

n=

e(EEf )/kB T

(15)

The integrand decreases exponentially at high energies, the

choice Ec2 would not cause large error

Change of variables: x = (E Eg )/kB T dE = kB T dx and

E = Ec1 x = 0

1 2 me kB T 3/2 (Ef Eg )/kB T

x
x dx
n= 2(
)
e
e

2
2
h
0


(16)
= /2
= 2(

me kB T 3/2 (Ef Eg )/kB T

2 ) e
2 h

Calculating EF
EF is unknown in Eq. (16). To get EF , we use the fact that for
intrinsic semiconductor
n=p
(17)
We calculate the hole density p as we did for n
0

p=

fh (E)gh (E)dE

(18)

fh (E) = 1 f (E), probability that a hole occupies state E

fh (E) = 1
e

1
(1 + e(EEf )/kB T )

(1 + e(EEf )/kB T )

(19)

(EEf )/kB T

The density of state of hole is

gh (E) =

1 2 mh 3/2
( 2 )
E,
2 2 h

E > 0

(20)

The hole concentration reads

0

p=

e(EEf )/kB T

1 2 mh 3/2
( 2 )
EdE
2 2 h

(21)

Change of variable: E/kB T = x dE = kB T dx and

E = x =

= /2

1 2 mh kB T 3/2 Ef /kB T
p= 2(
)
e
ex xdx

2
h
0
3/2
mh kB T
Ef /kB T
= 2(
2 ) e
2 h

(22)

Since n = p, from (16) and (22) we get

me kB T 3/2 (Ef Eg )/kB T
mh kB T 3/2 Ef /kB T
)
e
+
2
(
2
2 ) e
2 h
2 h
(23)
mh 3/2 Eg /kB T
2Ef /kB T
) e
+e
= (
me

0 = 2 (

Thus,
eEf /kB T = (
and

mh 3/4 Eg /2kB T
) e
,
me

1
3
mh
)
Ef = Eg + kB T ln (
2
4
me

1
Under usual circumstances, kB T Eg Ef = Eg
2

(24)

(25)

Calculating the electron concentration n

From (16) and (24) we get
me kB T 3/2 mh 3/4 Eg /2kB T Eg /kB T
e
2 ) ( me ) e
2 h
kB T 3/2
= 2 ( 2 ) (me mh )3/4 eEg /2kB T
2 h

n = p = 2(

Thus, n and p increase with the temperature T through

T 3/2

eEg /2kB T

The exponential term dominates. Thus as T increases n

increases exponentially. Here the tail of the Fermi function
becomes longer.

(26)

Electron concentration: plot

Numerical example

ln n

Eg = 1 eV
me = mh = m0
T = 300 K
n 1015 electron/cm3

1/T

Impurities in Ge
Ge

As

In

(a) n-type
(a) Free electron arising from ionization of a substitutional arsenic impurity atom;
(b) free hole arising from ionization of a substitutional indium impurity atom.

(b) p-type

Doping
Most applications require one type of carrier. Doping does the
jop
Donors: ntype semiconductors

When Si or Ge atom is replaced by As (pentavalent), the

fifth electron of As cannot enter a bond, so it becomes free
and As becomes an ion As+
As impurities contribute electrons to the CB without
generating holes
As impurities are termed donors

Acceptors: ptype semiconductors

When Si or Ge atom is replaced by a trivalent atom such as

In or Ga, one of the electron bonds remain vacant
If an electrons from another bond fill the vacancy, the
vacant bond moves which creates a free hole
Trivalent impurity is termed an acceptor, because it accepts
an electron to complete its tetrahedral bond

Impurity States
When a free electron is captured by an ionized donor, a
hydrogen-like atom forms. The Coulomb potential reads
V (r) =

e2
4r 0 r

(27)

r = the relative dielectric constant of the medium. Notice,

the hydrogen atom is not in vacuum

r is introduced to take into account the screening due to

the presence of the semiconductor crystal

r = 11.7 in Si, showing the substantial decrease in the

potential energy and hence binding energy

Si

Ground-state binding energy

of the donor using
Bohr model

As

e4 m0
1 me
]
(
)
[
2
2r m0 2 (4 0 h)
e4 m0
[
] = 13.6 eV,
2
2 (4 0 h)
Ed =

H-atom binding energy

radius rd = r a0
Hydrogen-like atom

m0
me

me
0.2, a0 = 0.53
A
m0
Ed 0.01 eV, rd 27
A

r 10,

Donor level is just below CB

At room temperature all donors are ionized

CB
Ed
Donor level
Acceptor level

Ea
VB

kB T = 0.025 eV at T = 300 K
kB T > Ed
Concentration of electrons is nearly
equal to that of donors 1015 cm3
An acceptor ion can attract a hole forming
a hydrogen-like atom with binding energy
Ea 0.01 eV < kB T
The acceptor level is just above VB
All acceptors are ionized at room temperature

In perfect crystal, no electron states exist in the energy gap

However, imperfection such as impurities introduces energy states in the energy gap

Electronic processes in a semiconductor

CB
Electrons in CB can be created by

interband thermal excitation

thermal ionization of the donors
Holes in VB may be generated by

a
VB

interband excitation
excitation of electrons from VB into acceptor level

Electrons may also fall from the donor levels to the acceptor level

Semiconductor statistics: the intrinsic region

The intrinsic concentration ni of carriers is determined by

the interband transitions
kB T 3/2
n = p = ni = 2 ( 2 ) (me mh )3/4 eEg /2kB T
2 h
ni Nd Na is satisfied here

Nd = number of donor atoms

Na = number of acceptor atoms
Note: each donor atom contributes just one electron to CB.
However, Na of these electrons may fall into the acceptors,
leaving Nd Na to the CB

As T increases, ni increases and the intrinsic nature of the

semiconductor is more established

kB T 3
3/2 Eg /kB T
(28)
2 ) (me mh ) e
2 h
This equation is independent of EF , and is generally valid in
intrinsic and extrinsic cases; the law of mass action.
n p = n2i = 4 (

Semiconductor statistics: the extrinsic region

Doping of 1015 cm3 is large enough to exceed the interband
excitation. We have two cases
Nd Na or n p ntype semiconductor

At room temperature, all donors are ionized

n = Nd ni
Hole concentration is small p = n2i /n = n2i /Nd . Here, ni using
Eq. (28)
If n increases because of doping, then p should decrease
according to Eq. (28)

Na Nd or p n ptype semiconductor

At room temperature, all acceptors are ionized

p = Na n i
Electron concentration is small n = n2i /p = n2i /Na . Here, ni
using Eq. (28)
If p increases because of doping, then n should decrease
according to Eq. (28)

Variation of n with T in ntype semiconductor

At very high temperatures, ni will be very large, so the

semiconductor will be intrinsic
At low temperatures, thermal energy is small and electrons
will drop to donor level and the conductivity diminishes;
freeze-out
n

Extrinsic

Freeze-out occurs at

Intrinsic

Nd

kB T Ed
T 100 K

Freeze-out
T

Electrical conductivity and mobility

n-type. The electrical conductivity is

e =

n e2 e
me

(29)

e = lifetime of the electron. e = 1012 s

n = 1021 m3
me = 0.1 m0
e 1 (Ohm-m)1 , for a semiconductor
e 107 (Ohm-m)1 , for a metal
e of semiconductor is small compared to that of a metal,
but still good enough for practical application

Mobility e . It is defined as the ratio ve /E

drift velocity ve =

ee
E , electric field = E
me
e =

ee
me

(30)

The negative sign is disregarded in the definition of

mobility

Mobility is a measure of rapidity or swiftness of the motion

The longer the lifetime of the electron and the smaller its
mass, the higher the mobility
Crystal
Si
ZnS
GaAs
InSb

e cm2 /Vs
1350
120
8500
80000

h cm2 /Vs
475
5
400
750

Using Eqs. (30) and (29), we have

e = n e e
e 1 (Ohm-m)1 , n = 1021 m3 e 100 cm2 /Vs

(31)

p-type. The hole conductivity is

h =

p e2 h
= p e h
mh

Generalization. When both electrons and holes are present,

the total conductivity is
= e + h = n e e + p e h

(32)

(33)

Intrinsic substance. n = p = ni , so
= n e (e + h )
E
Holes

Electrons

(34)

Cyclotron resonance I

In a magnetic field B, electrons and holes execute cyclotron

motion with frequencies
ce =

eB
,
me

ch =

eB
mh

(35)

For electron, the rotation is counterclockwise

For hole, the rotation is clockwise

Some power of an ac signal parallel to B will be absorbed if

the signal frequency is equal to ce or ch . The effective
mass can be determined this way

B
signal

Cyclotron resonance II

Cyclotron resonance can differentiate between electron and

hole if the signal is plane polarized
Plane polarized signal can be resolved into two circularly
polarized waves: counterclockwise and clockwise

The sense of the signal after passing through the sample

will be

Clockwise waves are absorbed by holes

Counterclockwise waves are absorbed by electrons

Clockwise if the sample contains electrons

Counterclockwise if the sample contains holes

Cyclotron resonance is carried out at T 4 K, where

collisions are minimum and is good enough to ensure
c 1. In this case the charge carrier will execute several
circular orbits and absorb enough energy from the signal
before getting scattered

Band Structure of real semiconductor I

Nomenclature

The center point, k = (0, 0, 0), in the Brillouin zone is

denoted
The three high-symmetry directions [100], [110], and [111]
in the Brillouin zone of the fcc lattice are denoted by , ,
and respectively
The points of intersection between the , , and paths
and the edge of the zone are termed X, K, and L
respectively

= E0 , where E0 =
The energy surface is defined as E (k)
constant

Band Structure of real semiconductor II

Sphere. The simple band structure

= Eg +
Ec (k)

2 k2
2 2
h
= h k
, Ev (k)
2 me
2 mh

(36)

has a spherical energy surface with one effective mass. For a

fixed Ec the electron we have
2
Eg = h (k 2 + k 2 + k 2 )
constant = Ec (k)
y
z
2 me x
which the equation of a sphere
kz

ky

E = E0

kx

(37)

Band Structure of real semiconductor III

Prolate ellipsoidal of revolution. It occurs when the charge

carrier has two effective masses ml and mt . For example,
around the primary minimum of the CB in Si the band
structure is
2
k 2 + kz2
=h
2 [ (kx kx0 ) + y
]
Ec (k)
2 ml
2 mt

ky

kx
kz

kx0

(38)

Band Structure of real semiconductor IV

GaAs has a band structure very similar to the ideal band

structure mentioned above, that is
2 2
= Eg + h k
Ec (k)
2 me

CB has minimum at the point k = 0

For small k, the energy E is given by Eq. (39) with
me = 0.072 m0
Electrons most likely populate this region, so a single
effective mass, me = 0.072 m0 , is enough
is no longer quadratic. A single
As k increases, Ec (k)
effective mass is not enough to describe such states
The next (local) minimum in the conduction band is at the
L point, 0.3 eV above the point. The next is along the
[100] direction

Symmetry in Ec (k)

(39)

= Ec (k)

Ec (k)

Band Structure of real semiconductor V

Consequences of symmetry

has a minimum along the [100]

The first symmetry: Ec (k)
direction
has a minimum along the
The second symmetry: Ec (k)
[010], [0
10], [001], [00
1], directions, that is along 100
directions
Similarly for the minima occuring along the 111 directions

Valence Band

GaAs has a cubic symmetry

It splits into three closely spaced subbands due to the

spin-orbit interaction
Subbands have different curvatures leading to different
effective masses for the holes
The holes are termed heavy and light holes

Other III-V semiconductors have band structure similar to

that of GaAs
Si band structure

Band Structure of real semiconductor VI

The primary minimum in the CB occurs along the [100]

direction, about 0.85 the distance from the center to the
edge of the zone
Due to Symmetries, inversion and cubic, there are 6
equivalent primary valleys along the 100 directions
The energy surfaces at these valleys are composed of prolate
ellipsoidal of revolution whose axes of symmetry are along
the 100 directions
The longitudinal mass is ml = 0.97 m0 . The transverse
masses are mt = 0.19 m0
The mass anisotropy ratio is ml /mt 5
Since the primary minimum does occur at k = 0, Si has an
indirect band gap. GaAs has a direct band gap
The VB is represented by a heavy hole, mh = 0.5 m0 , and
two light holes

Ge band structure

CB has its minimum along the [111] direction at the zone

edge

Band Structure of real semiconductor VII

Symmetries indicate that there are 8 equivalent minima at

the edge of the zone
The energy surfaces at these minima are 8 half-prolate
ellipsoids of revolution, or 4 full ellipsoids
The longitudinal mass is ml = 1.6 m0 . The transverse masses
are mt = 0.082 m0
The mass anisotropy ratio is ml /mt 20
The VB is similar to that of Si; it has one heavy hole and
two light holes

Band Structure of real semiconductor VIII

Band structure of bulk Si with no spin-orbit interaction
4

Energy (eV)

2
0

-2
-4
-0.5

0.0
k

0.5

2
A

1.0

Band Structure of real semiconductor IX

Band structure of bulk Si with spin-orbit interaction
4

Energy (eV)

2
0

-2
-4
-0.5

0.0
k

0.5

2
A

1.0

Band Structure of real semiconductor X

kz

ky

kx

Constant energy ellipsoids in Si

Band Structure of real semiconductor XI

Band structure of bulk Ge with no spin-orbit interaction
3

Energy (eV)

2
1
0

-1
-2

-0.5

0.0
k

0.5

2
A

1.0

Band Structure of real semiconductor XII

Band structure of bulk Ge with spin-orbit interaction
3

Energy (eV)

2
1
0

-1
-2

-0.5

0.0
k

0.5

2
A

1.0

Band Structure of real semiconductor XIII

Band structure of bulk GaAs with no spin orbit interaction

Energy (eV)

4
2
0

-2
-4
-6
-0.5

0.0
k

0.5

2
A

1.0