Semiconductor Optoelectronics

Prof. M.R. Shenoy

Department of Physics
Indian Institute of Technology, Delhi

Lecture - 11
Semiconductor Heterostructures - Lattice-Matched Layers

We continue with our discussion of semiconductor materials. Today we will discuss about
Semiconductor Heterostructures and Lattice-matched layers. We will talk about what Lattice-
matched layers are, and a continuation of that would be strained layer epitaxy, which we will
take up in the next class. Before I start, I briefly discuss the quiz we had in the last class.

(Refer Slide Time: 00:58)

Recall the quiz:

At room temperature, the energies corresponding to the band edges and the Fermi level are
given. Identify whether it is a p-type or n-type semiconductor. Give reasons for your answer.

The answer is straight forward. You just observe that the Fermi level is below the middle of
the band gap, and because m c = mv, if it is below the midpoint, then it is definitely a p-type
semiconductor.
(Refer Slide Time: 01:36)

So, the expected answer is shown above. You just draw the band diagram and show that E f is
closer to the valence band. You say that because the Fermi level is closer to the valence band,
it is a p-type semiconductor.

(Refer Slide Time: 02:16)

In part b of the question, you are asked to calculate the probability of electrons, so this is an
exercise to check whether you are familiar with the numbers, i.e., the numerical values of the
band energy levels. To calculate the probability of occupation of electrons, find the value of
Fermi function at E = Ec. Simply substitute E with the value of E c, and you get the answer as
approximately e-40.

In a quiz question, normally approximations are allowed, so you don’t need to use the
calculator. If you calculate, kT actually comes out to be around 0.0258, but you can use kT =
0.025 in simple calculations, because it easily cancels and you get simpler numbers.

(Refer Slide Time: 03:37)

Please note that the given numbers are negative. Why? This is because all the states which we
are considering here are bound states. Bound states have negative potential energy, so you
have to supply energy to free the electrons from the bound state. Potential energy is 0 for a
free particle.

Many of you may be knowing that the work function of materials is of the order of 4 eV, 5
eV, 3.5 eV. What is work function? It is the minimum energy required to free an electron from
the surface of the material. We deal with similar kind of numbers when we talk of bands in
the semiconductor.

Now, you find that the calculated probability of occupation f(E) is extremely small, of the
order of e-40. Although it is extremely small, it is not 0. If it were 0, there would be be no
electron, as there is no probability of having an electron in the conduction band. Yes, it is
extremely small, but remember that this will be multiplied by N c or Nv, which is very large:
1017 or 1018, therefore there are electrons present in the conduction band. This is the primary
reason that we have to put some numbers, to get a qualitative feel of what’s happening.

(Refer Slide Time: 06:12)

If you recall, in the last class we discussed about semiconductor materials, and in particular
we talked about ternary and quaternary compounds. They are also called alloy
semiconductors. The question arises, how are these alloys different from a doped
semiconductor? In other words, we’re asking: what are the differences between doping and
alloying?

(Refer Slide Time: 07:16)

See the above table carefully. When you want make Silicon p-type, you dope with Boron to
make it p-type. If you want to make it n-type, then you dope it with Arsenic or Phosphorous.
So, we see that for doping the group IV semiconductors, you add an impurity which is from
the adjacent groups, i.e., group III (for p-type) and V (for n-type).

(Refer Slide Time: 08:10)

What about in the case of alloying? The added element is from the same group. In Aluminum
Gallium Arsenide (AlxGa1-xAs), x fraction of Gallium is replaced by Aluminum, which is
from the same group. So, the first difference is—the substitutional element in alloying is from
the same group.

Next, why do you dope? This is done to change the conductivity, as conductivity is increased
by doping. Whether you dope to make it p-type or n-type, in either case the carrier
concentration increases, and hence conductivity increases. And what do we achieve by
alloying? We get bandgap modification. So the second point is—doping is for changing
conductivity, and alloying is for bandgap modification. There is very little changes in terms
of conductivity, but we have bandgap modification.
What about the doping concentration? Doping concentrations are typically 10 12 to 1019 per cc.
Therefore, the number of atoms with which you dope is much smaller compared to the total
number of atoms, i.e., the dopants are very small in number. In the case of alloying, you are
replacing x fraction of one element, which is say 0.3, which means 30 percent of the atoms
are substituted. Typically, semiconductors have 10 22 atoms per cc, and the concentration of
the substitutional atoms is also of the same order, 10 22. So, the third point is—the
concentration of the substitutional atoms is much higher in alloying than the dopant
concentration in doping.

The above discussion states the three basic differences between alloying and doping.

In the previous lecture I was answering the question, why do we go for alloy semiconductors?
Why do we need ternary and quaternary compound? I had given two primary reasons.
Actually there are many other reasons. For example, from fabrication point of view, there are
certain design flexibilities which are provided by using the additional degrees of freedom that
you get. But I am discussing the two most important points (reasons), the first of which is
bandgap modification.
(Refer Slide Time: 12:34)

We discussed bandgap modification in the last class. The second point I had written was that
it enables lattice-matched growth of Heterostructures. Today we discuss this second point.

What are Heterostructures? Heterostructures are structures where you grow a semiconductor
of a different type on a substrate. If you look at the fabrication of devices, generally
optoelectronic devices are much bigger in size compared to electronic devices. We are
familiar with IC, which are integrated circuits in electronics. Exactly like that, we have
optoelectronic integrated circuits, or photonic integrated circuits (PICS). Earlier, they were
also called OEICs.

In integrated circuits with very large scale integration, on a small chip you can have a million
electronic components, which is a very large number of components on a single chip. But
optoelectronic devices such as laser diodes (LDs) and photodetectors are much larger in size,
so the number of components on a chip are very few, like 3, 4, 5, 6, 8, etc. They are small
numbers. This is because the dimension of each component is typically tens of microns to
hundreds of microns, whereas in microelectronics we have components which go to sub-
micron size, so the size of optoelectronic components is very large.
(Refer Slide Time: 16:35)

What is the reason for this? Normally, in fabrication of components, you start with a
substrate, say a Silicon or Gallium Arsenide substrate, which is typically about 60 to 100 μm
thick. On this, you grow a large number of components. Most of the electronic components
involve p-n junction; you can deposit the required p-type or n-type dopant and diffuse. You
can have any number of layers on the surface of the substrate; you just deposit and diffuse.
Once you have a p-type or n-type, you can make npn transistors and other components.
Similarly, for MOS devices, you have to deposit oxide layers. This way, the final chip which
comes has passed through several stages of diffusion and deposition.

These layers which are deposited on top are called epitaxial layers. Basically, there are two
processes: one is diffusion and the other is epitaxy, growth on top. Most of the electronic
chips use both these processes, and finally you come up with a very large number of devices.
Most of the optoelectronic devices are fabricated by epitaxy, which means you start with a
substrate and then deposit layer by layer.

There are three commonly used processes for epitaxy. You can use Liquid Phase Epitaxy
(LPE), Vapor Phase Epitaxy (VPE), and Molecular Beam Epitaxy (MBE). We will not go into
the discussion of fabrication here. CVD (Chemical Vapor Deposition) and MOCVD are
variants of VPE.

So, epitaxy means layer by layer growth. You can deposit p-doped layer, n-doped layer and so
on. You can also deposit materials with different composition. For example, you start with n-
Gallium Arsenide, and the first layer could be n + Gallium Arsenide, the second layer could be
p-type Aluminum Gallium Arsenide, AlxGa1-xAs, and so on. So, a structure wherein you grow
layers of different composition and bandgaps is called Heterostructure.

(Refer Slide Time: 22:27)

Say you start with n Silicon, and then you deposit p plus Silicon, forming a p-n junction. It is
a junction between p-type and n-type of the same material, and that is called a homojunction.
Notice that p-type Silicon and n-type Silicon have the same bandgaps. On the other hand, you
can use layers of different material and different bandgaps. For example, I start with n-
Gallium Arsenide, and on top of that I deposit n Aluminum Gallium Arsenide, and then I
again deposit p or p plus Gallium Arsenide, and then deposit another layer of p plus Gallium
Arsenide.

Then I have a p-n junction between Gallium Arsenide and Aluminum Gallium Arsenide,
whose bandgaps are different. So these are heterojunction. There are two heterojunctions in
the above example, and this structure is called a double heterostructure. You will see that
almost all the laser diodes are double heterostructures. In fact, they are also called double
heterostructure lasers.

So, what is a double heterostructure? It is a heterostructure which has two heterojunctions.

We will discuss why such a structure is required, and what are the benefits it has. The
structure was proposed by Alferov in 1970, who got Nobel Prize later. You will see the
advantages of a double heterostructure. We will discuss these in detail.
(Refer Slide Time: 26:51)

If the heterojunction is to be free of defect, then they have to be lattice-matched. Lattice-

matched means the lattice constant of the layer which is being grown must be equal to the
lattice constant of the layer on which growth is taking place. If the lattice constant of the two
layers are equal, then it is lattice-matched, and that minimizes the possibility of defects at the
junction. Defects are very very detrimental to optoelectronic devices. So, how do we have
lattice-matched growth, and why it is important? We see a simple picture:

(Refer Slide Time: 28:13)

I draw the two dimensional lattice of a material which has some lattice constant. The lines are
bonds, and the circles are atoms. I have shown identical atoms, but if it is Gallium Arsenide,
then Gallium and Arsenide are placed alternately. So I have shown a simple picture. On this
material you want to grow a layer of another material which has a larger lattice constant
(say). So when the second material is grown, the bonds get tilted. You find that there are
certain atoms whose bond is not complete. Such bonds are called dangling bonds (shown in
the above figure).

The electrons which correspond to these dangling bonds have energy different from the
electrons which have formed bonds. As two shared electrons form a bond, the electron energy
is different from a free electron which has not been able to form a bond. When you see the
band diagram, these states have different energy. The bonded electrons have energy within
the band, but the states of the dangling bonds may have energy somewhere within the
bandgap, and we call these as trap states. Trap states can occur due to various reasons other
than dangling bonds. For example, it could be due to impurity dopants. Substitutional
impurities and interstitial impurities also lead to trap states.

So, there can various reasons for trap states, but even if the two materials are very pure, with
no impurity, but if the lattice constants do not match, then there may be trap states due to
dangling bonds. Trap states increase non-radiative recombinations. What does that mean?

There are radiative and non-radiative transitions. In optoelectronics, we are always interested
in radiative transitions. What is radiative transition? It is a transition which involves emission
or absorption of a photon. An electron from the conduction band can come down to the
valance band giving out a photon, and an electron can also make a transition to the
intermediate trap state, and come down through phonon transitions, without emission of a
photon.

In optoelectronic, you pass a current. You would like that every electron recombining with a
hole ideally emits photon. This is because our interest is to make an optical source. But if
electrons start coming through these trap states, then the probability of emission of a photon
decreases, or the rate of radiative transitions decreases, which is detrimental to the
performance of a device. So it is very important that non-radiative transitions are minimized.

Transition means it could be emission or absorption. ‘Radiative’ means a transition which

involves emission or absorption of a photon. Non-radiative transition, as the name indicates,
does not involve emission or absorption of a photon, and generally these transitions take
place by phonons. The phonons carry away the difference in energy. As traps and the defects
in a material reduce the radiative transition probability, we would like to keep them as
minimum as possible.

There are many electronic devices where you deliberately introduce trap states, but in
optoelectronics it is very rare, and you would not like any trap states to be there, which means
you would like to have growth where the lattice constant of the second material is the same as
lattice constant of the first material, so that such a thing does not happen. If I am growing a
heterostructure, which means say on Gallium Arsenide I am growing a layer of Aluminum
Gallium Arsenide, then we would like the lattice constants of the two materials to be same.
Then every atom finds a corresponding atom to form all completed bonds.

So we want materials with different bandgaps, because we would like to have different
emission wavelengths, which are required for different applications. But when we take
materials of different band gap, what is the substrate that we should choose, and what is the
composition that we should choose? Is there lattice matching or not? This is the concern
when you talk of lattice-matched growth of layers. I had given a list of binary
semiconductors. You see that although the band gaps are different for different materials, the
lattice constants are also different. And therefore, we generally cannot grow a binary
compound semiconductor on another binary semiconductor, because of lattice mismatch.

(Refer Slide Time: 37:23)

Now, how to ensure that a particular composition is lattice-matched to a substrate? There is
an important graph, which is lattice constant versus energy. You will find this in almost all
books which discuss heterostructures and growth of heterostructures. This is an important
graph for engineers to determine which materials are lattice-matched. First, let me show you
a list of some common optoelectronic materials (below), particularly with emphasis on
materials which are very relevant to optical communication.

(Refer Slide Time: 38:15)

In the list, you can see Aluminum Arsenide, Gallium Arsenide, Indium Phosphide, Aluminum
Phosphide, Indium Arsenide, Gallium Phosphide, Germanium and Silicon. I have also given
the band gap energy and the lattice constant of the materials. The last column indicates
whether it is direct band gap or indirect band gap type.
(Refer Slide Time: 38:54)

You draw a graph where you plot bandgap energy E g in eV versus lattice constant a in
angstroms. There are 6 materials here. You pick them from the list one by one, and mark their
positions in the graph. Shown below is the plot which I have made. This is not a theoretical
plot; these are experimental plots.

(Refer Slide Time: 39:37)

You simply plot the points first, and then they are linked. For example, Gallium Arsenide is
linked to Indium Arsenide by a curve. This curve represents Indium Gallium Arsenide. Why?
If I write Indium Gallium Arsenide as InxGa1–xAs, at one end of the curve is Gallium
Arsenide, for x = 0, and on the other end is Indium Arsenide, for x = 1. So the points on this
curve represent InxGa1–xAs for different values of the fraction x. The entire ‘line’ represents
various compositions of Indium Gallium Arsenide. You can understand similarly for the other
curves.

Notice the point at Indium0.53Ga0.47As. All materials on a vertical line have the same lattice
constant. So Indium0.53Ga0.47As is lattice-matched to Indium Phosphide. That means if I start
with the substrate as Indium Phosphide and if I grow a layer of Indium 0.53Ga0.47As, then they
are lattice-matched. Indium Phosphide has a band gap 1.35 eV and Indium Gallium Arsenide
has a band gap 0.74 eV.

So, the bandgaps are different, but lattice constant is the same, which means it is a lattice-
matched hetero junction. Indium Gallium Arsenide is used for photodetectors in optical
communication: very high speed and extremely good detectors are Indium Gallium Arsenide
detectors.

In the last lecture I mentioned about Aluminum Al xGa1–xAs, Aluminum Gallium Arsenide. I
mentioned that most of the commercial devices use it. For x = 0, the composition is Gallium
Arsenide, and for x = 1, it is Aluminum Arsenide. It so happens that the lattice constant of
Aluminum Arsenide is almost same as lattice constant of Gallium Arsenide. You can see this
from the plot. Although the correct values are 5.65 Å for Gallium Arsenide and 5.66 Å for
AlAs, the line representing Aluminum Gallium Arsenide is almost vertical, showing nearly
lattice-matching. As you move along the vertical line, E g is changing, but lattice constant is
not changing. So by choosing different values of x, you can have lasers or LEDs which emit
at different wavelengths.

You see some dotted lines and solid lines in the plot. Solid lines indicate the region where the
material is direct band gap, and dotted lines indicate the region where it is indirect band gap.
In the last lecture, we discussed that up to x = 0.4, Aluminum Gallium Arsenide is direct band
gap, and beyond that it is indirect band gap. So, it is an important curve. This graph has a
small portion of lattice constants. We have not shown Gallium Nitride and Aluminum Nitride,
which have lattice constants small, like 3.3, 3.4, and 3.1 Angstroms. You can go to other
portions for different materials which are relevant to UV or infrared. I have just expanded the
graph; you can see that the entire x axis range is only 5.4 to 6.2 Å. I have taken only this
small portion because our interest is materials which are used in optoelectronics and optical
communication.

When you move along the line joining Indium Gallium Arsenide, Indium 0.53Ga0.47As and
Indium Phosphide, then the material that you have is Indium Gallium Arsenide Phosphide,
InxGa1–xAs1–yPy. The points inside the polygons correspond to quaternary compounds, because
the lines are ternary compounds. The corners are binary compounds. For example, in this case
the corners are Aluminum Arsenide, Indium Phosphide, Aluminum Arsenide and Gallium
Arsenide.

If I write In1–xGaxAsyP1–y, it means that Indium Phosphide is the substrate, which is a binary
compound. In optoelectronics you start with a substrate which is usually binary compound. In
electronics, you normally start with Silicon. So, you start with a binary compound, like
Gallium Arsenide or Indium Phosphide, and deposit these alloy semiconductors on top to
make heterostructures. For example, Indium Gallium Arsenide Phosphide gives you the laser
diodes for optical communication. All lasers in the 1.3 μm to 1.55 μm range can be made by
different compositions of x and y, and it is lattice-matched to Indium Phosphide. We stop this
lecture here.