Lecture 22:

Absorption Spectrum of
Semiconductors
Recapping from last class
We have discussed gain and amplification by stimulated emission in previous classes

We learned about the conditions for amplification by stimulated emission and the laser
amplifier

You can see that: Iν(L)>Iν(0) if γ0(ν)>0 Amplification

Iν(L)<Iν(0) if γ0(ν)<0 Attenuation
Intensity versus length

We know also that γ is a

function of frequency v
Absorption and gain
The absorption or gain spectrum refers to the dependence of γ on ν and we have
observed that in the gain spectrum in the previous lecture
Today we will discuss the absorption spectrum of semiconductors

The gain that we have calculated is gain due to stimulated emission, which takes place in
the presence of an incoming photon

Recalling the expression for γ:

Absorption and gain
We first started with a thought experiment at 0K, where fg(ν) could be either 1 or –1

We now write fg(ν) as f0(ν)

Since we are studying attenuation, fg(ν) < 0, and therefore we can write:

so that αa is a positive number which is the attenuation coefficient

1
hv – Eg 2
f0(ν) has a frequency dependence, otherwise ɣ primarily depends on the factor:
𝑣2
Absorption and gain
1
1
The denominator shows and the numerator 𝑥 variation with hv
2
𝑥2

In the last lecture we said that 𝑣 2 variation is very small in the range of interest, which was
the amplification bandwidth and thus could be neglected in the denominator
However, absorption takes place over a wider range and therefore we need to consider that
factor in this case

Absorption
ɣ, absorption coefficient vs Eg
For amplification we considered from Eg to Eg plus 0.1 eV, so v is assumed to be constant
over that range
A photodetector can detect over a wide range of wavelengths, therefore we need to
consider the absorption spectrum over a wide range of energies

1 1
1
*𝑥 2
𝑥2
𝑥2
1
hv – Eg 2
𝑣2
1
𝑥2
Absorption spectrum
Material Bandgap type
Si I
GaAs D
InP D
InSb D
InAs D
Ge I (only just)
GaP I

Do you see a correlation between the graph and table?

Attenuation depth (absorption length = 1/e) GaAs, InP <1μm, Si ~ 10’s of μm
The attenuation length is defined as the depth into the material measured along the direction of the radiation where the
intensity of the radiation falls to 1/e of its value at the surface
Absorption spectrum
For direct band gap materials, absorption rapidly increases as soon as Eλ > Eg, whereas for
indirect band gap materials it rises more slowly
It appears that none of these materials appear to follow our model
There are several reasons for this..

For the theoretical prediction, we considered direct band gap semiconductors, and assumed
only a parabolic approximation close to the band edge
Absorption spectrum
Close to the band edge where Eλ is Eg plus 0.1 eV etc., the parabolic approx. is a valid, but
beyond these values it is not
So this model is only valid close to the band edge, this is point number one

Almost all materials have an indirect band gap, therefore phonon-assisted absorption
needs to be considered – 2nd point, our model only considers DG materials

If you consider the theory of an indirect material and

determine the phonon-assisted absorption, then you have a
variation that looks like this:

K0 is a constant and K1 is temperature dependent

Absorption spectrum
Y-axis values are a little
smaller now compared to
previous plot

α indirect varies with T as K1 is temperature dependent

Arbitrary units used Arbitrary units used

Approx. gives you (there are other smaller components):

What does this mean?
Absorption is a due to direct bandgap absorption plus absorption due to an indirect band
gap (phonon assisted).

This explains why the absorption coefficient doesn’t fall away at higher energy

Why is there a difference in the absorption trend for direct and indirect bandgap
material?

To understand this, we need to return to the E-K diagram

E-K diagram, absorption co-efficient

In the direct case, when hv = Eg In the indirect case at hv = Eg, for absorption
absorption occurs with very high to occurs, k mismatch must be compensated
probability phonon assisted Reduced α
absorption parameter
Consider Ge, Si and GaP
α parameter is increasing at higher photon
energy

However, as energy increases, a less oblique transition is required. At even higher energy,
almost vertical transitions occur, which have high probability.
That is why in Ge, GaP and Si α rapidly increases at higher E, because now vertical transitions
can also take place
How to obtain an absorption spectrum
By observing the absorption spectrum you can determine whether the material is
a direct or indirect bandgap material

If you are given a piece of semiconductor how would you know? Here is a simple
test:

SC Detector

𝑣 Tuneable light source (monochromator) which is

frequency variable
Monochromator
Typically uses two bulbs and is used to select narrow bands of wavelengths scanning over a
much larger window, such as from 300-1100nm by diffraction

Changing the angle of the

diffraction grating controls
the wavelength incident on
your sample
Transitions
Other effects include:

Interband transition: Band-to-band

Intraband transition

Phonon transition: 0.01eV to 0.1eV

Excitonic resonances: excited electron

hole pairs

Absorption comprises several

effects, not just direct band gap
or vertical transitions
Intraband transitions
Intraband transitions are within the band, these are generally low energy transitions

An electron may make a transition within the band, this requires phonon assistance
because there is a change in momentum
The carriers which go up will subsequently thermalize and come down to a local
minima through phonon transitions

A higher K.E. electron is called a hot electron due to a high energy photon,
subsequently it will come down by thermalization, and this is an intraband transition

When a high energy photon leaves a hole in the valence band. Electrons sitting above it
will immediately come down to the hole, and so the hole moves up, within the same
band, these too are intraband transitions
Phonons and excitons
A third type of transitions are phonon transitions, but we know that phonon energies are
typically in the range 0.01 eV to 0.1 eV.
Low energy photons can be directly absorbed by phonons, i.e., low energy photons can
directly generate phonons within the semiconductor.

And a fourth type of absorption which is also possible are excitonic absorptions or excitonic
resonances

An exciton is a bound state of an electron and an electron hole which are attracted to each
other by the electrostatic Coulomb force. It is an electrically neutral quasiparticle that exists
in insulators, semiconductors and some liquids.
Excitons An electron sitting at Ev, k=0 can absorb a photon of
energy which is close to Eg and makes an upward
transition if there is a vacant state
This positively charged hole and the negatively
charged electron build up coulomb attraction

If they are close in real space, a hydrogen atom-like

bond forms, which is held by coulomb interaction
Exciton Bound electron-hole pair where the hole is
localized at a position in the crystal represented
by black dots
The binding is extremely small, typically of the
order of 10 meV or 15 meV

What is the binding energy of hydrogen atom?

13.6eV

So an exciton is not exactly like a hydrogen

atom, but it behaves like one

The excited electron is ‘shielded’ from the hole

by the surrounding covalent bonds
Excition
Normally at room temperature the average kinetic energy is 25 meV, therefore the exciton
gets dissociated because its binding energy is only 10meV

Dissociated means they are no longer held together, they behave like free carriers

At room temperature you would not see this effect

Due to this binding, the energy of this pair is slightly lower than Eg

This leads to states which are close to the band edge, but with energy little less than band
edge

Ɛ = dielectric constant, Si (11.7)

Excition energies
Excition energy (meV)
Si 14.7
Ge 3.8 - 4.1
GaAs 4.2
GaP 3.5
CdS 29
CdSe 15

Ɛ = dielectric constant, Si (11.7), effective mass of

electron also a determining factor

In semiconductors, excitons are typically weakly bound where e-h pair distance is >> lattice
constant ‘a’ (often called Mott or Wannier excitions)
Excitons
So photons of energy slightly less than Eg can also be absorbed and form excitons, these
are exciton states

There is a density of states due to

excitons

When are excitonic transition observed?

Now let us look at the total spectrum

We now see the total spectrum: α vs hν

Starting from 0.01 eV, not 0.1eV as before

Excitons usually observed at low

temperatures in bulk semiconductors
Free carrier absorption
Absorption in a quantum well structure

From previous lecture we know this equals f(E2)−f(E1)

1
ρ(v) ∝ hv – Eg 2 for bulk
In a Q.W. structure we know that ρ(v) is a step
function

The first energy level was Ec + Eq =1

Absorption coefficient step function; QW
For bulk semiconductor absorption should remain almost constant for photon energies from Eg
to a certain value, because the D.O.S remains constant
For a Q.W. absorption behaves like a step function, and then remains almost constant, except
1
for the 2 dependence Derived for K=0,
𝑣
direct bandgap
materials

α
1
2
𝑠𝑡𝑒𝑝 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑓𝑜𝑟 𝑞𝑢𝑎𝑛𝑡𝑢𝑚 𝑤𝑒𝑙𝑙 𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑒
ℎ𝑣

Eg / eV
Density of states
It is the density of states which primarily determines the absorption spectrum or gain
spectrum of a semiconductor.

That is why we spent so much time on the density of states. The step-like variation has
very important applications, which we will discuss in later lectures

This is important because the gain profile and

absorption spectrum is determined by ρ(ν)