Tag Archives: Semiconductors

Semiconductor Growth Technology: Molecular Beam Epitaxy and MOCVD

The development of advanced semiconductor technologies presents one important challenge: fabrication. Two methods of fabrication that are being used to in bandgap engineering are Molecular Beam Epitaxy (MBE) and Metal organic chemical vapour deposition (MOCVD).

Molecular Beam Epitaxy uses high-intensity vacuums to fabricate compound semiconductor materials and compounds. Atoms or molecules containing the desired atoms are directed to a heated substrate. Molecular Beam Epitaxy is highly sensitive. The vacuums used make use of diffusion pumps or cryo-pumps; diffusion pumps for gas source MBE and cryo-pumps for solid source MBE. Effusion cells are found in MBE and allow the flow of molecules through small holes without collusion. The RHEED source in MBE stands for Reflection Hish Energy Electron Diffraction and allows for information regarding the epitaxial growth structure such as surface smoothness and growth rate to be registered by reflecting high energy electrons. The growth chamber, heated to 200 degrees Celsius, while the substrate temperatures are kept in the range of 400-700 degrees Celsius.

MBE is not suitable for large scale production due to the slow growth rate and higher cost of production. However, it is highly accurate, making it highly desired for research and highly complex structures.



MOCVD is a more popular method for growing layers to a semiconductor wafer. MOCVD is primarily chemical, where elements are deposited as complex chemical compounds containing the desired chemical elements and the remains are evaporated. The MOCVD does not use a high-intensity vacuum. This process (MOCVD) can be used for a large number of optoelectronic devices with specific properties, including quantum wells. High quality semiconductor layers in the micrometer level are developed using this process. MOCVD produces a number of toxic elements including AsH3 and PH3.

MOCVD is recommended for simpler devices and for mass production.



Heterostructures & Carrier Recombination

Heterojunction is the term for a region where two different materials interact. A Heterostructure is a combination of two or more materials. Here, we will explore several interesting cases.


The AlGaAs-InGaAs interaction is interesting due to the difference in energy bandgap levels. It was found that AlGaAs has a higher bandgap level, while InGaAs has a lower bandgap. By layering these two materials together with a stark difference in bandgap levels, the two materials make for an interesting demonstration of a heterostructure.

The layering of a smaller bandgap material between a wider bandgap material has an effect of trapping both electrons and holes. As shown on the right side of the below picture, the center region, made of AlGaAs exibits high concentrations of both electrons and holes. This leads to a higher rate of carrier recombination, which can generate photons.


Here, the lasing profile of the material under bias:






A commonly used group of materials is InGaAsP, InGaAs and InP. Unlike the above arrangements, these materials may be lattice-matched. Lattice-matching may be explored in depth later on.Simulations suggest low or non-existent recombination rates. Although this is a heterostructure, one can see that there are no jagged or sudden drastic movements in the conduction and valence band layers with respect to each other to create a discontinuity that may result in a high recombination rate.



Conduction & Valence Band Energies under Biasing (PN & PIN Junctions)

Previously, we discussed the effect of doping concentrations on the energy band gap. The conclusion of this process was that the doping concentration alone does not alter the band gap. The band gap is the difference between the conduction band and valence bands. Under biasing, the conduction and valence bands are in fact affected by doping concentration.

One method to explain how the doping level will influence the conduction band and valence band under bias is by demonstrating the difference between the energy bands of a PN Junction versus that of a PIN Junction. Simulations of both are presented below. The intermediate section found between the p-doped and n-doped regions of the PIN junction diode offer a more gradual transition between the two levels. A PN junction offers a sharper transition at the conduction and valence band levels simulatenously. A heterostructure, which is made of more than one material (which will have different band gaps) may produce even greater discontinuities. Depending on the application, a discontinuity may be sought (think, Quantum well), while in other situations, it may be necessary to smooth the transition between band levels for a desired result.

The conduction and valence bands are of great importance for determining the carrier concentrations and carrier mobilities in a semiconductor structure. These will be discussed soon.

PN Junction under biasing (conduction and valence band energies):


Code Used (PN Junction):

#TOP TO BOTTOM – Structure Specification
region num=1 bottom thick = 0.5 material = GaAs NY = 20 acceptor = 1e18
region num=2 bottom thick = 0.5 material = GaAs NY = 20 donor = 1e18


PIN Junction Biased:


PIN Junction Unbiased:


Code Used (PIN Junction):

#TOP TO BOTTOM – Structure Specification
region num=1 bottom thick = 0.5 material = GaAs NY = 20 acceptor = 1e18
region num=3 bottom thick = 0.2 material = GaAs NY = 10
region num=2 bottom thick = 0.5 material = GaAs NY = 20 donor = 1e18

Here, the carrier concentrations are plotted:


Energy Bandgaps

Previously, a PN Junction Simulator in ATLAS program was posted. Now, we will use and modify this program to explore more theory in respect to semiconductor materials, high speed electronics and optoelectronics.

The bandgap, as mentioned previously is the difference between the conduction band energy and valence band energy. The materials GaAs, InP, AlGaAs, InGaAs and InGaAsP are simulated and the bandgap values for each are estimated (just don’t use these values for anything important).

  • GaAs: ~ 1.2 eV
  • InP: ~ 1.35 eV
  • AlGaAs: ~ 1.8 eV
  • InGaAs: ~0.75 eV
  • InGaAsP: 1.1 eV


Here the conduction band and valence band are shown.


The structure used in the PN Junction Simulator is found below:

#TOP TO BOTTOM – Structure Specification
region num=1 bottom thick = 0.5 material = GaAs NY = 20 acceptor = 1e17
region num=3 bottom thick = 0.001 material = InP NY = 10
region num=4 bottom thick = 0.001 material = GaAs NY = 10
region num=5 bottom thick = 0.001 material = AlGaAs NY = 10 x.composition=0.3 grad.3=0.002
region num=6 bottom thick = 0.001 material = GaAs NY = 10
region num=7 bottom thick = 0.001 material = InGaAs NY = 10 x.comp=0.468
region num=8 bottom thick = 0.001 material = GaAs NY = 10
region num=9 bottom thick = 0.001 material = InGaAsP NY = 10 x.comp=0.145 y.comp = 0.317
region num=2 bottom thick = 0.5 material = GaAs NY = 20 donor = 1e17

Is the bandgap affected by doping the concentration level?

A quick simulation (below) will tell us that the answer is no. What might influence the bandgap however? And what could the concentration level change?


This (above) is a simulation of GaAs with layers at different doping concentration levels. The top is a contour of the bandgap, which is constant, as expected. The top right is a cross section of this GaAs structure (technically still a pn junction diode); the bandgap is still constant. The bottom two images are the donor and acceptor concentrations.

The bandgap energy E_g is the amount of energy needed for a valence electron to move to the conduction band. The short answer to the question of how the bandgap may be altered is that the bandgap energy is mostly fixed for a single material. In praxis however, Bandgap Engineering employs thin epitaxial layers, quantum dots and blends of materials to form a different bandgap. Bandgap smoothing is employed, as are concentrations of specific elements in ternary and quarternary compounds. However, the bandgap cannot be altered by changing the doping level of the material.

PN Junction Simulator in ATLAS

This post will outline a program for ATLAS that can simulate a pn junction. The mesh definition and structure between the anode and cathode will be defined by the user. The simulator plots both an unbiased and biased pn junction.

go atlas


#Define the mesh

mesh auto
x.m l = -2 Spac=0.1
x.m l = -1 Spac=0.05
x.m l = 1 Spac=0.05
x.m l = 2 Spac =0.1

#TOP TO BOTTOM – Structure Specification
region num=1 bottom thick = 0.5 material = GaAs NY = 20 acceptor = 1e17
region num=2 bottom thick = 0.5 material = GaAs NY = 20 donor = 1e17

#Electrode specification
elec num=1 name=anode x.min=-1.0 x.max=1.0 top
elec num=2 name=cathode x.min=-1.0 x.max=1.0 bottom
#Gate Metal Work Function
contact num=2 work=4.77
models region=1 print conmob fldmob srh optr
models region=2 srh optr
material region=2

solve init outf=diode_mb1.str master
output con.band val.band
tonyplot diode_mb1.str

method newton autonr trap maxtrap=6 climit=1e-6
solve vanode = 2.5 name=anode
save outfile=diode_mb2.str
tonyplot diode_mb2.str

This program may also be useful for understanding how different materials interact between a PN junction. This simulation below is for a simple GaAs pn junction.

The first image shows four contour plots for the pn junction with an applied 2.5 volts. With an applied voltage of 2.5, the recombination rate is high at the PN junction, while there is low recombination throughout the unbiased pn junction. The hole and electron currents are plotted on the bottom left and right respectively.


Here is the pn junction with no biasing.


The beam profile can also be obtained:


Parameter Analysis of the MESFET, Channel Width Calculation

Engineering design regularly involves an analysis of the formulae behind the various parameters of a system one is trying to build or improve. Some parameters are static, such a particular qualities of the materials being used. Perhaps there is a constraint made on the system or a goal, such as achieving function at a certain frequency or to reduce the size as much as possible. Today, many programs exist that can perform complicated calculations for the engineer. To construct a problem or calculation that produces the desired result may need more attention.

The MESFET uses a contact between n-doped semiconductor material with highly n-doped semiconductor material to form a junction field effect transistor. The great advantage of not using a p-doped semiconductor material is that the transistor can be built without using hole transfer. Since hole transfer is much slower than electron transfer, the MESFET can function much faster than other types of transistors.

For the MESFET, it may not be possible to examine all parameters. Consider first the following:


Potential variation along the channel (notice the similarity of the following to Ohm’s law, V=IR):


Where the resistance along the channel is:


Depletion Width (also referenced in the above formula) under the gate:


Pinch-off Voltage:


Threshold Voltage:


Built-in Potential:


The above formulas alone would be enough to put to use. While constructing a MESFET, it was found that the doping concentration of donor electrons in the channel played an important role. N_D, the donor doping concentration is found in most of the above formulas. The doping concentration is of particular importance, since it can be directly manipulated. The pinch-off voltage and the donor concentration are directly proportional. By achieving an estimate (or of the values are known) for other parameters, it would be possible to perform a parameter sweep for the MESFET system for doping concentration. This method may become critical for optimizing semiconductor device designs.


MESFET Design Problem

Let’s say we want to calculate the channel width of an n-channel GaAs MESFET with a gold Schottky barrier contact. The barrier height (φ_bn) is 0.89 V. The temperature is 300 K. The n-channel doping N_d is 2*10^15 cm^(-3). Design the channel thickness such that V_T = +0.25V.


GaAs MESFET Designs

A GaAs MESFET structure was built using Silvaco TCAD:

• Channel Donor Electrons: 2e17
• Channel thicknes s : 0.1 microns
• Bottom layer: p doped GaAs (5 micron thick, 1e15p doping)
• Gate length: 0.3 micron
• Gate metal work function: 4.77eV
•Separation between the source and drain electrode: 1 micron


The IV curve is as follows. Of primary importance are the two bottom curves, which are for a gate voltage of -0.2V and -0.5V. The top curve is 0V, over which would be undesirable for the MESFET operation.


Now, in terms of designing a MESFET, there is a large amount of theory that one may need to grasp to build one from scratch – you would probably first start by building one similar to a more common iteration. That said, there are a number of parameters that one may wish to tweak and to achieve, to name a few: saturation current, threshold voltage, transit frequency, maximum frequency, pinch-off voltage.

The iteration above does not show a highly doped region under the source and drain contacts. The separation between source and drain may also be increased and the size of the gate decreased.


Channel doping level was found to make a significant difference in overall function. The channel must be doped to a certain level, otherwise the structure may not behave properly as a transistor.

go atlas


# Define the mesh

mesh auto
x.m loc = 0 Spac=0.1
x.m loc = 1 Spac=0.05
x.m loc = 3 Spac=0.05
x.m loc = 4 Spac =0.1

# n region

region num=1 bottom thick = 0.1 material = GaAs NY = 10 donor = 2e17

# p region

region num=2 bottom thick = 5 material = GaAs NY = 4 acceptor = 1e15

# Electrode specification
elec num=1 name=source x.min=0.0 x.max=1.0 top
elec num=2 name=gate x.min=1.95 x.max=2.05 top
elec num=3 name=drain x.min=3.0 x.max=4 top

doping uniform conc=5.e18 n.type x.left=0. x.right=1 y.min=0 y.max=0.05
doping uniform conc=5.e18 n.type x.left=3 x.right=4 y.min=0 y.max=0.05

#Gate Metal Work Function
models fldmob srh optr fermidirac conmob print EVSATMOD=1
contact num=2 work=4.77

# specify lifetimes in GaAs and models
material material=GaAS taun0=1.e-8 taup0=1.e-8
method newton

solve vdrain=0.5
LOG outf=proj2mesfet500mVm.log
solve vgate=-2 vstep=0.25 vfinal=0 name=gate
save outf=proj2mesft.str
output band.param photogen opt.intens con.band val.band

tonyplot proj2mesft.str
tonyplot proj2mesfet500mVm.log

III-V Semiconductor Materials & Compounds

The Bandgap Engineer’s Periodic Table

In contrast with an elemental semiconductor such as Silicon, III-V Semiconductor compounds do not occur in nature and are instead combinations of materials from the III and V category groups on the periodic table. Silicon, although a proven as a functional semiconductor for electronic applications at lower frequencies is unable to perform a number of roles that III-V semiconductors are able to. This is in large part due to the indirect bandgap quality of Silicon. III-V semiconductor materials under a number of applications and combinations are direct bandgap semiconducting materials. This allows for operation at much higher speeds. Indirect bandgap materials will be unable to produce light.


Ternary and Quaternary III-V

The following list introduces the main III-V semiconductor material compounds used today. In a follow-up discussion, ternary and quarternary III-V semiconductors will be discussed in greater depth. To begin however, these may be understood as a process of mixing, varying or transitioning between two or more material types. For instance, a transition region between GaAs and GaP is described as GaAsxP1-x. This is the compound GaAsP, a blend of both GaAs and GaP, but at end of the material region, it is GaAs and at the other end it is equal to GaP.


GaAs was the first III-V material to play a major role in photonics. The first LED was fabricated using this material in 1961. GaAs is frequently used in microwave frequency devices and monolithic microwave integrated circuits. GaAs is used in a number of optical and optoelectronic near-infra-red range devices. The bandgap wavelength is λg = 0.873 μm.

Not long after GaAs was used, other III-V semiconductor materials were grown, such as GaSb. The bandgap wavelength of GaSb λg = 1.70 μm, making it useful for operation in the Infra-red band. GaSb can be used for infrared detectors, LEDs, lasers and transistors.

Similar to GaAs, Indium Phosphide is used in high-frequency electronics, photonic integrated circuits and optoelectronics. InP is widely used in the optical telecommunications industry for wavelength-division multiplexing applications. It is also used in photovoltaics.

An alloy of GaAs and GaP, Gallium Arsenide Phosphide is used for the manufacture of red, orange and yellow LEDs.

Indium Gallium Arsenide is used in high-speed and high sensitivity photodetectors and see common use in optical fiber telecommunications. InGaAs is an alloy often written as GaxIn1-xAs when defining compositions. The bandgap energy is approximately 0.75 eV, which is convenient for longer wavelength optical domain detection and transmission.

Indium Gallium Arsenide Phosphide is commonly used to create quantum wells, waveguides and other photonic structures. InGaAsP can be lattice-matched to InP well, which is the most common substrate material for photonic integrated circuits.

Indium Gallium Arsenide Antimonide has a narrow bandgap (0.5 eV to 0.6 eV), making it useful for the absorption of longer wavelengths. InGaAsSb faces a number of difficulties in manufacture and can be expensive to make, although when these difficulties are avoided, devices (such as photovoltaics) that use it may achieve high quantum efficiency (~90%).

Aluminum Gallium Aresinide has nearly the same lattice constant as GaAs, but with a larger bandgap, between 1.42 eV and 2.16 eV. AlGaAs may be used as part of a border region of a quantum well with GaAs as the inner section.

AlInGaP sees wide use in the construction of diode lasers and LEDs from deep ultraviolet to infrared ranges.

GaN has a wide bandgap of 3.4 eV and sees use in high frequency high power devices and optoelectronics. GaN transistors operate at higher voltages than the GaAs microwave transistors and sees possible use in THz devices.

InxGa1−xN is another ternary III-V semiconductor that can be tuned for use in optoelectronics from the ultraviolet (see GaN) to infrared (see InN) wavelengths.

AlxGa1−xN is another compound that sees use in LEDs for blue to ultraviolet wavelengths.

Although AlInGaN is not used much independently, it sees wide use in lattice matching the compounds GaN and AlGaN.
Indium Antimonide is an interesting compound, given that it has a very narrow bandgap of 0.17 eV and the highest electron mobility of any known semiconductor. InSb can be used in quantum wells and bipolar transistors operating up to 85 GHz and field-effect transistors operating at higher frequencies. It can also be used as a terrahertz radiation source.

High Speed Waveguide UTC Photodetector I-V Curve (ATLAS Simulation)

The following project uses Silvaco TCAD semiconductor software to build and plot the I-V curve of a waveguide UTC photodetector. The design specifications including material layers are outlined below.


Simulation results

The structure is shown below:



Forward Bias Curve:



Negative Bias Curve:



Current Density Plot:



Acceptor and Donor Concentration Plot:



Bandgap, Conduction Band and Valence Band Plots:




Construct an Atlas model for a waveguide UTC photodetector. The P contact is on top of layer R5, and N contact is on layer 16. The PIN diode’s ridge width is 3 microns. Please find: The IV curve of the photodetector (both reverse biased and forward bias).

The material layers and ATLAS code is shown in the following PDF: ece530proj1_mbenker



Semiconductor Distribution of Electrons and Holes

Charge Flow in Semiconductors

Charge flow in a semiconductor is characterized by the movement of electrons and holes. Considering that the density and availability of electrons and holes in a material is determined by the valence and conduction bands of that material, it follows that for different materials, there will be different densities of electrons and holes. The electron and hole density will determine the current throughput in the semiconductor, which makes it useful to map out the density of holes and electrons in a semiconductor.


Density of States

The density of electrons and holes is related to the density of states function and the Fermi distribution function. States are the formations of electrons and holes that can be formed in a semiconductor. A density of states is the amount of possible formations that can exist in a semiconductor. The Fermi-Dirac probability function is used for determining the the density of quantum states. The following formula determines the most probable formation distribution or state. By varying Ni (number of particles) along energy levels, the most probable state can be found, while gi refers to remaining particle positions in the distribution.


Density of States Calculation using ATLAS

By integration of Fermi-Dirac statistics for the density of states in the conduction and valence bands arises the formulae for electron and hole concentration in a semiconductor:


where Nc and Nv are the effective density of states for the conduction bands and valence bands, which are characteristics of a chosen material. If using a program such as ATLAS, the material selection will contain parameters NC300 and NV300.



Charge Carrier Density

Charge carriers simply refer to electrons and holes, which both contribute to the flow of charge in a semiconductor. The electron distribution in the conduction band is given by the density of quantum states multiplied by the probability (Fermi-Dirac probability function) that a state is occupied by an electron.

Conduction Band Electron Distribution:


The distribution of holes in the valence band is the density of quantum states in the valence band multiplied by the probability that a state is not occupied by an electron:



Intrinsic Semiconductor

An intrinsic semiconductor maintains the same concentration of electrons in the conduction band as holes in the valence band. Where n is the electron concentration and p is the hole concentration, the following formulae apply:


The overall intrinsic carrier concentration is:


Eg is the band gap energy, which is equal to the difference of the energy is the conduction band and the energy in the valence band. Eg = Ec – Ev.

Electron and Hole concentrations expressed in terms of the intrinsic carrier concentration, where Ψ is the intrinsic potential and φ is the potential corresponding to the Fermi level (Ef = qφ):



Donor Atoms Effect on Distribution of Electrons and Holes (Extrinsic Semiconductor)

Adding donor or acceptor impurity atoms to a semiconductor will change the distribution of electrons and holes in the material. The Fermi energy will change as dopant atoms are added. If the density of holes is greater than the density of electrons, the semiconductor is a p-type and when the density of electrons is greater than the density of holes, the semiconductor is n-type (see Density of States formulas above).

[8], [10]


Thermoelectric Effect, Thermoelectric current and the Seebeck Effect

There are three types of current flow in a semiconductor: Drift, diffusion, and thermoelectric. Drift current is very familiar as the study of conductors leads us to know that when a potential gradient (voltage) is established, electrons will flow in a conductor to balance this out. The same effect happens in semiconductors. However, there are two types of charge carriers in semiconductors: electrons AND holes. This leads to diffusion current, which is caused by a concentration gradient rather than a potential gradient.

The third kind of current within a semiconductor is called thermoelectric current. which involves the conversion of a temperature gradient to a voltage. A thermocouple is a device which measures the difference in potential across two dissimilar materials where one end is heated and the other is cold. It was found that the temperature difference was proportional to the potential difference. Although Alessandro Voltage first discovered this effect, it was later rediscovered by Thomas Seebeck. The combination of potential differences leads to the full definition of current density.



S is called as the “thermopower” or “Seebeck coefficient” which is units of Volts/Kelvin. The two equations of Ohm’s law (point form) and E_emf look remarkably similar.


The Seebeck coefficient is negative for negative charge carriers and positive for positive charge carriers, leading to a difference in the Seebeck Coeffecient between the P and N side of the PN junction above. This leads to the above circuit being used as a thermoelectric generator. If a voltage source replaces the resistor, the circuit becomes a thermal sensor. These (thermoelectric generators) are often employed by power plants to convert wasted heat energy into additional electric power. They are also used in car engine engines for the same reason (fuel efficiency). Solid state devices have a huge advantage in the sense that they require no moving parts or fluids which eliminates much of the need for maintenance. They also reduce environmental impact by converting waste heat into electrical energy.

Photovoltaic Effect and Theory of Solar Cells

Just as plants receive energy from the sun and use it to produce glucose, a photovoltaic cell receive energy from the sun and generates an electrical current. The working principle is based on the PN junction, which will be revisited here.

Silicon can be subdivided into several discrete energy levels called “bands”. The major bands of concern are the valence and conduction bands. The bottom bands are fully occupied and don’t change.


For silicon, the bandgap energy is 1.1eV. For an intrinsic semiconductor, the Fermi level is directly between the conduction and valence band. This is because there is an equal number of holes in the valence band as electrons in the conduction band. This means the probability of occupation of energy levels in both bands are equal. The Fermi level rises in the case of an n-type semiconuctor (doped with Phosphorous) and declines towards the valence band in a p-type (doped with Boron).

The following illustrates an energy band diagram for a semiconductor with no bias across it. Photodiodes (light sensors) operate in this manner.


The Fermi energy is shown to be constant. On the far right hand side away from the depletion region, the PN junction appears to be only P-type (hence the low Fermi level with respect to the conduction band). Likewise, on the left the Fermi level is high with respect to the conduction band. The slope of the junction is proportional to the electric field. A strong electric field in the depletion region makes it harder for holes and electrons to move away from the region. When a forward bias is applied, the barrier decreases and current begins to flow (assuming the applied voltage is higher than the turn on voltage of 0.7V). Current flows whenever recombination occurs. This is because every time an electron recombines on the P side, an electron is pushed out of the N side and beings to flow in an external circuit. The device wants to stay in equilibrium and balance out. This is why solar cells (as opposed to photodiodes) are designed to operate in a forward bias mode.

The sunlight produces solar energy in the frequency bands of Ultraviolet, infrared and visible light. In order to harness this energy, silicon is employed (made from sand and carbon). Silicon wafers are employed in solar cells. The top layer of the silicon is a very thin layer doped with phosphorous (n-type). The bottom is doped with P-type (doped with Boron). This forms the familiar PN junction. The top layer has thin metal strips and the bottom is conductive as well (usually aluminum). Only frequencies around the visible light spectrum are absorbed into the middle region of the solar cell. The photon energy from the sun knocks electrons loose in the depletion region which causes a current to flow. The output power of a single solar cell is only a few watts. To increase power, solar cells are wired in series and parallel to increase the voltage and current. Because the output of the solar cells is DC, the output is run through an inverter, a high power oscillator that converts the DC current to an 240V AC current compatible with household appliances.


Direct-Bandgap & Indirect-Bandgap Semiconductors

Direct Semiconductors

When light reaches a semiconductor, the light is absorbed if the photon energy is greater than or equal to the band gap, creating electron-hole pairs. In a direct semiconductor, the minimum of the conduction band is aligned with the maximum of the valence band.



One example of a direct semiconductor is GaAs. The band diagram for GaAs is shown to

the right. As the gap between the valence band and conduction band is 1.42eV, if a

photon of same or greater energy is applied to the semiconductor, a hole-electron pair is created for each photon. This is termed the photo-excitation of semiconductors. The photon is thereby absorbed into the semiconductor.




Indirect Semiconductors and Phonons

indiresemicFor an indirect semiconductor to absorb a photon, the process must be mediated by phonons, which are quanta of sound and in this case refer to the acoustic vibration of crystal lattice. A phonon is also used to provide energy for radiative recombination. When understanding the essence of a phonon, one should recall that sound is not necessarily within hearing range (20 – 20kHz). In fact, the sound vibrations in a semiconductor may well be in the Terrahertz range. The diagram to the right shows how an indirect semiconductor band would appear and also the use of phonon energy to mediate the process of allowing the indirect semiconductor to behave as a semiconductor.



Excitons are bound electron-hole pairs that are created in pure semiconductors when a photon with bandgap energy or larger is absorbed. In bulk semiconductors, these excitons will dissipate rapidly. In quantum wells however, the excitons may remain, even at room temperature. The effect of the quantum well is to force an electron and hole to be very close to each other. This allows for a strong bonding effect to take place and allows the quantum well the ability to generate light as a semiconductor laser.



The band structure of a semiconductor is given by:


Where mc = 0.2 * m0 and mv = 0.8 * m0 and Eg = 1.6 eV. Sketch the E-k Diagram.


Optoelectronic Integrated Circuit Substrate Materials

The substrate material used on an optical integrated circuit (OIC) is dependent primarily on the function performed by the circuit. An optical integrated circuit may consist of sources, modulators, detectors, etc and no one substrate will be optimal for all components, which means that a compromise is needed when building an integrated circuit. There are two main approaches that taken to deciding on a solution to this compromise: hybrid and monolithic approaches.


Hybrid Approach

The hybrid approach attempts to bond more than one substrate together to obtain an optimization for each device in the integrated circuit. This approach allows for a more optimized design for each component in theory, however the process of bolding the various elements together is prone to misalignment and damage from vibration and thermal expansion. For this reason, although the hybrid approach is a theoretically more otpimized approach, it is more common to use the monolithic approach for OIC.


Monolithic Approach

The monolithic OIC uses a single substrate for all devices. There is one complication in this approach which is that most OIC will require a light source, which can only be fabricated in optically active materials, such as a semiconductor. Passive materials, such as Quartz and Lithium Niobate are effective as substrates, however an external light source would need to be coupled to the substrate to use it.


Optically Passive and Active Materials

Optically active materials are capable of light generation. The following are examples of optically passive materials:

  • Quartz
  • Lithium Niobate
  • Lithium Tantalate
  • Tantalum Pentoxide
  • Niobium Pentoxide
  • Silicon
  • Polymers

The following are optically active materials:

  • Gallium Arsenide
  • Gallium Aluminum Arsenide
  • Gallium Arsenide Phosphide
  • Gallium Indium Arsenide
  • Other III-V and II-VI semiconductors


Losses in Substrate due to Absorption

Monolithic OICs are generally limited to the active substrates above. Semiconductors emit light at a wavelength corresponding to their bandgap energy. They also absorb light at a wavelength equal to or less than their bandgap wavelength. It follows then, for example, if a light emitter, a waveguide and a detector are all fabricated in a single semiconductor, there is a considerable issue of light being absorbed into the substrate, meaning that not enough light will be present for the detector. Thus, reducing losses due to absorbtion is one of the main concerns in substrate materials.


Gas Laser and Semiconductor Lasers


The Gas Laser

In laboratory settings, gas lasers (shown right) are often used to eveluate waveguides and other interated optical devices. Essentially, an electric charge is pumped through a gas in a tube as shown to produce a laser output. Gasses used will determine the wavelength and efficiency of the laser. Common choices include Helium, Neon, Argon ion, carbon dioxide, carbon monoxide, Excimer, Nitrogen and Hydrogen. The gas laser was first invented in 1960. Although gas lasers are still frequently used in lab setting sfor testing, they are not practical choices to encorperate into optical integrated circuits. The only practical light sources for optical integrated circuits are semiconductor lasers and light-emitting diodes.


The Laser Diode


The p-n junction laser diode is a strong choice for optical integrated circuits and in fiber-optic communications due to it’s small size, high reliability nd ease of construction. The laser diode is made of a p-type epitaxial growth layer on an n-type substrate. Parallel end faces may functions as mirrors to provide the system with optical feedback.


The Tunnel-Injection Laser

The tunnel-injection laser enjoys many of the best features of the p-n junction laser in it’s size, simplicity and low voltage supply. The tunnel-injection laser however does not make use of a junction and is instead made in a single crystal of uniformly-doped semiconductor material. The hole-electron pairs instead are injected into the semiconductor by tunneling and diffusion. If a p-type semiconductor is used, electrons are injected through the insulator by tunneling and if the semiconductor is n-type, then holes are tunneled through the insulator.