You are on page 1of 18

Semiconductors

Semiconductors are materials with a (relatively) small band


gap (typically 1eV) between a filled valence band and an
empty conduction band.
Chemical potential μ (often called Fermi energy) lies in the
band gap.
Insulators at T=0, with a small density of electrons excited
at finite temperatures.
Typical semiconductors are Silicon and Germanium
or III-V compounds such as GaAs
2 atoms in primitive basis have 4 electrons each (or 3 + 5); 8
electrons fill 4 bands made of s and p orbitals

Band Structure of Semiconductors


Plot Energy versus k. EF separates Electrons are only
filled and empty states excited in a very
limited region of
energy

EF

RJ Nicholas – Semicond. HT10 1


Energy levels of electrons and holes

Close to the band edge minima and maxima we can write:

= 2k 2
conduction band ε = EG +
2me
= 2k 2
valence band ε = −
2mh

Band gap determines the optical


properties - strong absorption
when hν > EG

Optical absorption
Excitation promotes an
electron from the valence
band to conduction band.
An empty state left in valence
band is known as a hole.
= 2k 2 ⎛ = 2k 2 ⎞
Δε = EG + − ⎜⎜ − ⎟⎟
2me ⎝ 2mh ⎠
Absorption above
⎛ = 2k 2 ⎞ = 2k 2
Δε = ⎜⎜ EG + ⎟⎟ + the band gap is
⎝ 2me ⎠ 2mh
strong ~ 106 m-1

Electron energy Hole energy

RJ Nicholas – Semicond. HT10 2


If band minima and
maxima are at different
points then we have an
indirect semiconductor.

The classic example is


silicon.

This affects the optical


properties such as
absorption where Δk ≈ 0

Hole picture
Remove one electron from a filled band and electricity can be
conducted by the movement of all of the electrons present.
The sum of this motion is equivalent to one positively
charged particle: a hole

filled band Electron state part-filled hole


removed band

crystal 0 ke -ke kh = -ke


momentum

Energy 0 εe(ke) -εe(ke) εh(kh) = - εe(ke)

RJ Nicholas – Semicond. HT10 3


Holes and their properties
1 ∂ε
velocity vh = = ve
= ∂k
Work done ∂ε F ∂ε
= F vg =
by a force F ∂t = ∂k
∂ε ∂k
=
∂k ∂t
∂k ∂k
∴ = e = F = −= h
∂t ∂t
Rate of change of momentum is opposite sign for holes
∂ke ∂k h
= = F = − e (E + v × B ) = = + e (E + v × B )
∂t ∂t

Effective masses
Force accelerates the electron or hole:

∂v g 1 ∂ 2ε 1 ∂ 2ε ∂k ∂k
= = but F = =
∂t = ∂k ∂t = ∂k 2 ∂t ∂t
∂v g 1 ∂ 2ε F
= 2 2 F =
∂t = ∂k m*

Effective mass:
1 1 ∂ 2ε
Negative for holes unless = 2
we say charge is positive
m* = ∂k 2

RJ Nicholas – Semicond. HT10 4


Meaning of the effective mass
−1
2⎛ ∂ ε ⎞
2
m * = = ⎜⎜ 2 ⎟⎟
⎝ ∂k ⎠

Effective mass changes as we


move through a band in k-space
positive (electron-like) at the
bottom, becoming negative
(hole-like) at the top

Typical values in semiconductors are in the range 0.01 to 0.5 me

Concentrations of Electrons and Holes


Calculate carrier density from density of states and
distribution function:

= 2k 2
n = ∫ g (ε ) f (ε ) dε , where ε = EG +
e
F −D
EG
2me
For most semiconductors the chemical potential, μ, (often
also called the Fermi Energy (EF)), lies in the band gap so:
μ −ε
3/ 2
1 ⎛ 2me ⎞
f e
F −D ≈ exp ( ) and g (ε ) = ⎜ ⎟ (ε − EG )1/ 2
k BT 2π 2 ⎝ = 2 ⎠
giving :
3/ 2 ⎛ μ ⎞ ∞ ε
⎜⎜ ⎟⎟
1 ⎛ 2me ⎞ −
n = ⎜ ⎟ ⎝ k BT


(ε − EG ) dε
1/ 2 k BT
e e
2π 2 ⎝ = 2 ⎠ EG

RJ Nicholas – Semicond. HT10 5


3/ 2 ⎛ μ − EG ⎞ ∞
1 ⎛ 2me k BT ⎞ ⎜⎜ ⎟⎟
ε − EG
n = ⎜
2π 2 ⎝ = 2 ⎠
⎟ e ⎝ k BT

0
x1/ 2 e − x dx , with x =
k BT

3/ 2 ⎛ μ − EG ⎞ ⎛ μ − EG ⎞
⎛m k T ⎞ ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟
n = 2 ⎜⎜ e B 2 ⎟⎟ e ⎝ k BT ⎠
= NC e ⎝ k BT ⎠

⎝ 2π = ⎠
Effective number of conduction band states at ε = Eg

Now use same procedure for holes


0
= 2k 2
p = ∫ g (ε ) f (ε ) dε , where ε = −
h
F −D
−∞
2mh
1 1
f Fh− D = 1 − f Fe− D = 1 − (ε − μ ) / k BT = ( μ − ε ) / k BT ≈ e (ε − μ ) / k BT
e +1 e +1

3/ 2 ⎛ −μ ⎞ 0 ε
⎜⎜ ⎟⎟
1 ⎛ 2mh ⎞
p = ⎜ ⎟ ⎝ k BT ⎠
∫ (−ε ) dε
1/ 2 k BT
e e
2π 2 ⎝ = 2 ⎠ −∞

⎛ −μ ⎞ ∞
− εe ε
3/ 2 ⎜⎜ ⎟⎟
1 ⎛ 2mh k BT ⎞
p = ⎜
2π 2 ⎝ = 2 ⎠
⎟ e ⎝ k BT

0
x1/ 2 e − x dx , with x =
k BT
= h
k BT

3/ 2 ⎛ −μ ⎞ ⎛ −μ ⎞
⎛m k T ⎞ ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟
p = 2 ⎜⎜ h B 2 ⎟⎟ e ⎝ k BT ⎠
= NV e ⎝ k BT ⎠

⎝ 2π = ⎠

Effective number of valence band states at ε = 0

RJ Nicholas – Semicond. HT10 6


Law of Mass Action
⎛ μ − EG ⎞ ⎛ −μ ⎞
Densities of electrons ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟
n = N C e⎝ ⎠
p = NV e ⎝ k BT ⎠
k BT
,
and holes depend on μ

μ is determined by density of charge introduced by impurities


(doping)

Product np is independent of μ: Law of Mass Action


3 ⎛ − EG ⎞ ⎛ − EG ⎞
⎛ k BT ⎞ ⎜
3 / 2 ⎜⎝
⎟⎟ ⎜⎜ ⎟⎟
np = 4 ⎜⎜ ⎟ (me mh ) e
2 ⎟
k BT ⎠
= N C NV e ⎝ k BT ⎠

⎝ 2π = ⎠

Intrinsic Semiconductors
A semiconductor is said to be intrinsic if it is undoped, and
the only source of electrons and holes is by thermal
excitation from the valence band to the conduction band.
In this case:
⎛ − EG ⎞ 3/ 2 ⎛ − EG ⎞
⎜⎜ ⎟⎟
⎛ k T ⎞ ⎜⎜ ⎟⎟
n = p = ni = N C NV e ⎝ 2 k BT ⎠
= 2 ⎜⎜ B 2 ⎟⎟ (me mh )
3/ 4
e ⎝ 2 k BT ⎠

⎝ 2π = ⎠

We can use this relation to measure the Band Gap, by


measuring the carrier densities from the Hall Effect
At low temperatures ni → 0 and impurities are important

RJ Nicholas – Semicond. HT10 7


Doping Semiconductors
We can control the numbers of electrons and holes in a
semiconductor by adding impurities which dope the
material.
Donors donate an electron to make the material more n-type.
A typical example is by adding a group V element (such as
As or P) to a group IV semiconductor such as silicon.
Four of the electrons participate in the sp3 bonds as if they
were from silicon, but the fifth electron is left over. Extra
charge on P nucleus creates a +ive core, and the fifth
electron can be bound to this, but the binding is weak.

Shallow Donors
Impurity binding looks like a hydrogen atom
Binding energy e 4 me* me* / me
R =
*
= R0
2(4πε r ε 0= ) 2 εr2

Binding energy is very small because:


(i) the effective mass is small (typically m* = 0.1 me)
(ii) The wavefunction is large (much more than the crystal
unit cell), so we must include the relative dielectric constant
of the medium εr, - typically ≈10.

∴ R * ≈ 10 −3 R0 = 13.6 meV (155 K )

RJ Nicholas – Semicond. HT10 8


Acceptors
Dope semiconductors with holes by adding group III elements to
a group IV material. e.g. put Ga into silicon.
One valence electron is missing. This creates a vacant state, a
hole, which binds to the ion core of the Ga which is negatively
charged.
e 4 mh* mh* / me
Binding energy is: R =
*
= R0
2(4πεε 0 = ) 2 ε2
Where is the energy level?
Ionized Acceptor or Donor is a free hole or electron at the
top or the bottom of the band
∴ Acceptor is R* above the valence band edge
Donor is R* below the conduction band edge

Extrinsic Carrier Densities


Density of impurities (e.g. Donors) usually much less than NC,
NV. Impurities can be ionized Nd+, or neutral Nd0 so:
Nd = Nd+ + Nd0
Using charge neutrality we have:
n = p + Nd+
Simple argument:
At high (e.g. Room) Temperature kT > R* therefore all
donors will be ionized but the density of holes created by
excitation across the band gap is still small
∴ n ≈ Nd+ ≈ Nd

RJ Nicholas – Semicond. HT10 9


Chemical Potential μ
At high (room) temperature most impurities are ionized
∴ n ≈ Nd+ ≈ Nd
μ lies below the Donor level so that most impurities are
empty (ionized), but is still close to conduction band.
Minority Carriers
By Law of Mass action the (very small) density of holes is:
p = ni(T)2/Nd

Temperature
dependent density and
chemical potential

Density is constant in
region around room
temperature

High temperature gives


intrinsic behaviour

RJ Nicholas – Semicond. HT10 10


Conductivity of Semiconductors

σ = neμ e + peμ h , μ =
m*
μ is the mobility, which is defined by v = μE, the dift
velocity per unit electric field.

Conductivity is dominated by variation in densities.

Mobility is determined by scattering rate:


Low T: impurity scattering gives μ ~ T3/2
High T: phonon scattering gives μ ~ T-3/2

p-n junction
• Unbiased junction • Forward bias of V

n ∼ Ncexp -(Eg/kT) n ∼ Ncexp -(Eg/kT)


------- n ∼ Ncexp -(Eg - eV/kT) n ∼ Ncexp -(Eg /kT)
-------

n ∼ Nc n ∼ Nc

eV
p ∼ Nv

RJ Nicholas – Semicond. HT10 11


p-n junctions
What is current flow across junction? j = nev
What is v? - due to diffusion of carriers with
diffusion coefficient D and lifetime τ.
L D
Diffusion length L = Dτ so v = =
τ τ
Electron current/area from: E
D − G

(i) p → n j p→n = e
e
NC e k T B

τ
EG
D −
(ii) n → p j e
n→ p = e NC e k BT

τ
EG
D −
(iii) Total je
= e NC e k BT
(1 − 1) = 0
τ
Total

Add bias voltage V - This only affects electron current


from n → p
Electron current/area from:
EG
D −
(i) p → n j e
p →n = e NC e k BT

τ
− EG + eV
D
(ii) n → p j e
n→ p = e NC e k BT

τ
EG eV eV
D −
(iii) Total je
= e NC e k BT
(e k BT
− 1) = I 0 (e k BT
− 1)
τ
Total

An equivalent pair of currents flows for holes with the


same formulae, except for masses and lifetimes of holes
(D = μk BT / e, the Einstein relation )

RJ Nicholas – Semicond. HT10 12


Low Dimensional Structures and Materials

• Anisotropic Materials
• Artificial layered structures -
Quantum Wells and
Superlattices

Layers may be only a few


• Electric or Magnetic Fields atoms thick
applied in one direction.

Heterojunctions
Energy levels for 2 different Energy line up at junction of
semiconductors two (undoped) materials

RJ Nicholas – Semicond. HT10 13


Reduced Dimensionality
Quantum Well removes 1 Dimension by quantization

Electron is bound in well and can only move in plane

2-D system - motion in x, y plane

Quantum Well - Type I

Typical Materials: 1: GaAs 2: (Al0.35Ga0.65)As


(Eg = 1.5 eV) (Eg = 2.0 eV)

Energy levels are quantized in


z-direction with values En for
both electrons and holes ∴

E = En + =2k⊥2/2m*
↑ ↑
1-D 2-D

RJ Nicholas – Semicond. HT10 14


Infinite well - Particle in a box
L
• 1-D Motion in z-direction
n=3
= 2 d 2ψ nπz
− =Eψ ψ n = A sin
2m * dz 2 L n=2

2
= 2 ⎛ nπ ⎞ = 2 k ⊥2 n=1
En = ⎜ ⎟ +
2m *⎝ L ⎠ 2m *

Typical values L = 10 nm, me* = 0.07 me → En = 54 n2 meV

∴System is Two-Dimensional when:


E2 - E1 > kT
162 meV 25 meV at 300 K

Density of States g(k)dk g(ε)dε


3-D
4πk 2 dk
3/ 2
Travelling waves V ⎛ 2m * ⎞ 1/ 2
⎜ ⎟ ε dε
eikx (eik.r) (2π / L )3 (2π )2 ⎝ = 2 ⎠

Periodic boundary condit.


ψ(x) = ψ(x + L)
2-D
2πk dk A ⎛ 2m * ⎞
⎜ ⎟ dε
(2π / L )2 4π ⎝ = 2 ⎠
∴ eikL = 1 → k = ±2nπ/L
→ δk = 2π/L
1-D
ε = =2k2/2m*, 2dk
1/ 2
L ⎛ 2m * ⎞ −1/ 2
⎜ ⎟ ε dε
dε = (=2/2m*) 2k dk 2π / L 2π ⎝ = 2 ⎠

RJ Nicholas – Semicond. HT10 15


Optical Properties
3-D
Absorption coefficient is
proportional to the
density of states:

∴ α ~ ε1/2

Modified close to the band


gap due to ‘excitons’

2-D - Big Changes

Multiple Band gaps -


Band gap shift -
Sharper edge

For wide wells the sum of


many 2-D absorptions becomes
equivalent to the 3-D
absorption shape (ε1/2)
Correspondence principle.

RJ Nicholas – Semicond. HT10 16


GaAs/Al0.35Ga0.65As
Quantum Well
absorption

• Sharp peaks due to


excitons
• peaks doubled due to
heavy and light holes

Semiconductor
lasers
Forward biased p-n
junction

Quantum Well laser


Fibre Optic Communications,
CD players, laser pointers

RJ Nicholas – Semicond. HT10 17


How do we achieve low dimensionality?

Naturally anisotropic crystals


Controlled growth of layers and/or apply external potential
Deposit thin layers of single crystals to create
‘heterostructures’

Two Main techniques:

I) Molecular Beam Epitaxy (MBE)

II) Metal Organic Vapour Phase Epitaxy (MOVPE)

Molecular Beam Epitaxy


(MBE)

• Ultra High Vacuum


evaporation of molecular
species of elements
(Molecular Beam)

• Epitaxy - maintaining
crystal structure of the
‘substrate’ - which is a
single crystal

RJ Nicholas – Semicond. HT10 18

You might also like