Photonics and Optical Communication, Spring 2007, Dr. D.

Knipp

Photonics and Optical Communication

(Course Number 300352)
Spring 2007
Optical Source
Dr. Dietmar Knipp
Assistant Professor of Electrical Engineering

http://www.faculty.iu-bremen.de/dknipp/

Optical Sources 1
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

Photonics and Optical Communication

5 Optical Sources
5.1 Introduction
5.2 Absorption and Emission of light
5.2.1 Spontaneous Emission
5.2.2 Stimulated Emission
5.3 Light emitting diodes versus laser diodes
5.4 Introduction to semiconductors
5.4.1 Structural Properties of Semiconductors
5.4.2 Energy Bands in Semiconductors
5.4.3 The pn-junction
5.4.4 Diodes under forward bias
5.5 Light emitting diodes (LEDs)
5.5.1 Direct and indirect Semiconductors
5.5.2 Device structures
5.5.3 Application of Light emitting Diodes
5.6 Lasers
5.6.1 Spontaneous Emission
5.6.2 Population inversion
5.6.3 Three and four energy level systems
5.6.4 Optical feedback and laser resonators
Optical Sources 2
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.5 Threshold condition for laser oscillation

5.6.6 Requirements for lasing
5.7 Semiconductor Lasers
5.7.1 Stimulated emission and lasing in Semiconductors
5.7.2 Semiconductor Materials for lasing applications
5.7.3 Efficiency of LEDs and laser diodes
5.7.4 Laser Diode structures
5.7.4.1 Fabry Perot Homojunction laser diode
5.7.4.2 Double heterostructure laser diode
5.7.4.3 Quantum well lasers
5.7.4.4 Distributed Feedback (DFB) Lasers
5.7.4.5 Vertical Cavity Surface Emitting Lasers (VCSELs)

References

Optical Sources 3
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.1 Introduction
The success of optical communication technology is stimulated by the
development of optical fibers and optical fiber technology on one side and the
invention of solid state lasers and laser diodes on the other side. Solid state
lasers are compact, reliable and inexpensive. Optical communication systems
with very high bandwidth-distance products can only be implemented by using
lasers or laser diodes.

Laser diode package and micrograph of

inside of a laser package. Ref.: Infineon

Optical Sources 4
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.1 Introduction
In general the generation of light is caused by the transition of an electron form
an energetically higher energy state to a lower energy state. The energy
difference due to the transition of the electron leads to a radiative or a non-
radiative process. We are of course interested in radiative processes as we like
to “build” an optical source. The non-radiative processes typically lead to the
creating of heat. The energy is simply dissipated by heat. In the case of a
radiative process photons are emitted. The emission of light, can take place
either spontaneously or it can be stimulated by the presence of another photon
of the “right” energy.
In order to understand the processes of light-generation, it is necessary to
consider fundamental processes like structural and optical properties and energy
levels in materials and the electronic device concepts. An understanding of the
structural and optical properties is needed to actually understand the process of
light generation and an understanding of the devices is needed to make use of
such an effect.

Optical Sources 5
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.2 Absorption and Emission of light

The interaction of light and matter in the form of absorption and emission
requires a transition from one discrete energy level to another energy level. The
frequency and the wavelength of the emitted or absorbed photon is related to
the difference in energy E, between the two energetic states, where h is the
Planck constant h=6.626 x 10-34J, f is the frequency and λ is wavelength of the
absorbed or emitted light.

hc
E = E 2 − E1 = hf = Photon energy
λ

Optical Sources 6
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.2 Absorption and Emission of light

The figure illustrates transitions

between two energy states. When
a photon with the energy (E2-E1) is
incident on the material an electron
may be excited from the energy
state E1 into an higher energy state
E2 through the absorption of the
photon.
Alternatively, when the electron is
initially on a higher energy level it
can make a transition to a
energetically lower state and the
provided energy loss leads to the
emission of a photon. Here the
transition is assumed to be a
radiative transition.
Energy state diagram showing (a) absorption, (b) spontaneous emission,
(c) stimulated emission. Ref.: J.M. Senior, Optical Fiber Communications
Optical Sources 7
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.2 Absorption and Emission of light

We have to distinguish between radiative and non-radiative processes. In the
case of a non-radiative process the energy is dissipated as heat. The question
whether a transition is non-radiative or radiative depends on the involved
species of carriers, the material itself, the level of impurities in the material, the
temperature and the device structure.
In the case of radiative emission we can than distinguish between
spontaneous and stimulated emission.

Optical Sources 8
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.2.1 Spontaneous Emission

For most of the light sources the photons are emitted spontaneously (sun light,
light bulb, halogen lamp). In a first step an electron is elevated to an
energetically higher state which is usually unstable. In the second step the
electron will spontaneously return to an energetically more stable state (which
is typically the energetically lower state). This process is a statistical process
which can happen very fast. As a consequence the spontaneously (or
randomly) emitted photons are incoherent (very short coherence time) and the
emitted spectrum has broad spectral width.

Energy state diagram for spontaneous emission of a photon.

Ref.: J.M. Senior, Optical Fiber Communications

Optical Sources 9
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.2.2 Stimulated Emission

The operating principle of a laser is based on stimulated emission. We speak
about stimulated emission if the electron which enters an energetically higher
state (excited state) remains in this state until it is “stimulated” by the presence
of a photon to leave this higher energetically state and return to the more stable
lower energetically state (ground state). One of the requirements for stimulated
emission is that the electron can stay in its excited state a relatively long period
of time (a few microseconds) before it changes its state spontaneously. In the
case of spontaneous emission the electron stays in this excited state usually for
a shorter period of time (picoseconds). In the case of stimulated emission the
electron can be “stimulated” by the presence of a photon to emit its energy in
the form of another photon.

Energy state diagram for stimulated emission. Ref.: J.M. Senior, Optical Fiber
Communications
Optical Sources 10
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.2.2 Stimulated Emission

In this case the energy of the incident photon has to be very close to the energy
of the excited electron.
Stimulated emission takes place when the emitted photon has the same energy
(the same wavelength), phase and direction as that of the photon which
stimulated it!
Stimulated emission is the inverse process of absorption!

5.3 Light emitting diodes versus laser diodes

In order to observe spontaneous or stimulated emission we have to excite the
electrons first before they can return to a lower energetic states. Of course
energy has to be provided to excite the electron. The energy can be provided
by heat, absorption of photons (photoluminescence) or electrical current
(electroluminescent). We are interested in the later case, where the energy is
provided by an electrical current.

Optical Sources 11
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.3 Light emitting diodes versus laser diodes

In both types of devices the recombination of carrier is used to provide a photon
flux. However, the emission of light in a light emitting diode is a spontaneous
process, whereas it is a stimulated process in a laser diode. Therefore, the
description of an LED (light emitting diode) is different from the description of a
laser diode. The description of an LED is by far simpler than the description of a
laser diode.
In both cases a semiconductor diode is used, which operates under forward
bias conditions. Furthermore, the same structure can be used to build an
optical amplifier.

Forward biased pn-diode

operating as (a) LED, (b)
semiconductor amplifier, (c)
semiconductor injection laser.
Ref.: Saleh & Teich,
Fundamentals of Photonics

Optical Sources 12
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4 Introduction to semiconductors

In order to get an understanding of semiconductor based optical light sources
we have to review some of the basic semiconductor properties. We will
concentrate in this lecture on the description of the basic operating principle of
a pn-junction (pn diode) as the light emitting diode (LED) and the laser diode
are based on such structure.

5.4.1 Structural Properties of Semiconductors

First the structural properties of semiconductors will be discussed. The
structural properties have a strong effect on the electronic and the optical
properties of the material.
In general we can distinguish semiconductors in terms of their structural
properties. Semiconductors exist as crystalline or amorphous materials.
Crystalline material exhibit a high degree of structural order, whereas
amorphous materials are characterized by a random or partly random
distribution of the atoms or molecules in the solid. We concentrate here only on
crystalline materials.

Optical Sources 13
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.1 Structural Properties of Semiconductors

We assume that the material has a perfect crystal structure, which means that
the material has no structural defects (structural disorder) and they do no
contain impurities. Based on the Bohr’s atom model we can derive the energy
band structure of a semiconductor.

Different kinds of materials in terms of structural properties.

Ref.: Saleh & Teich, Fundamentals of Photonics

Optical Sources 14
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.2 Energy Bands in Semiconductors

The band structure of a single atom is characterized by allowed discrete energy
levels. Electrons can stay on these energy levels. When moving from a single
atom to a solid several of these energy levels overlap and discrete energy
bands are formed.
The conduction and the valence band are the two highest energy levels. The
band structure of semiconductor is characterized by a valence band and a
conduction band. The two bands are separated by the bandgap. The
energetically states in the bandgap are forbidden due to Bohr’s atom model.

Energy bands in two

semiconductors (a) silicon,
(b) Gallium Arsenide
Ref.: Saleh & Teich,
Fundamentals of
Photonics

Optical Sources 15
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.3 The pn-junction

A band structure for an intrinsic semiconductor is shown in the figure. Intrinsic
means that the semiconductor is not doped. Doping is the incorporation of
impurities in the crystalline semiconductor. The intrinsic semiconductor is
therefore considered to be a highly pure semiconductor.
The fact whether the conduction band is occupied by an electron is described
by the Fermi-Dirac statistic. The Fermi Dirac function is a probability function.

Energy bad structure of an intrinsic

semiconductor at a temperature
above absolute zero, and the
Fermi-Dirac distribution for holes.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 16
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.3 The pn-junction

The Fermi-Dirac statistic is closely related to the Fermi energy.

1
P (E ) =
1 + exp[(E − E F ) kT ] Fermi-Dirac Distribution

where k is the Boltzmann factor and EF is the Fermi energy. The Fermi-Dirac
function P(E)=0.5 for E=EF.
The Fermi-Dirac distribution is shown (on the previous slide) for a
temperature above absolute zero. The excitation of an electron from the
valance band to the conduction band leaves an empty state in the valence
behind. A missing electron in the valence band is called a hole.
Electrons can move in the conduction band and holes move in the valence
band.

Optical Sources 17
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.3 The pn-junction

The Fermi level is a measure of the distribution of carriers in the
semiconductor. The Fermi level is almost in the middle of the bandgap for an
intrinsic semiconductor. The Fermi level can be seen as a probability function,
which describes whether the bands are occupied or not (by a carrier). As we
are speaking about electrons and hole we can define a Fermi-Dirac distribution
for electrons and holes. The probability that a state is occupied by an electron
or hole is exact 50% for the Fermi energy.

1 Fermi-Dirac Distribution
Pe (E ) =
1 + exp[(E − E F ) kT ] for Electrons

1
Ph (E ) = Fermi-Dirac Distribution
1 + exp[(E F − E ) kT ] for Holes

Pe (E ) + Ph (E ) = 1 Fermi-Dirac Distribution

Optical Sources 18
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.3 The pn-junction

To create an extrinsic semiconductor the material is doped with impurities
which either create free electrons (donors) or free holes (acceptors).
When donor impurities are added to the material, the thermally excited
electrons from the donor level are raised in the conduction band to create free
carriers so to say electrons. Due to the increase of the electron concentration in
the conduction band the Fermi level shifts closer to the conduction band.
Is the semiconductor material doped with acceptors the number of free carriers
in the valence band is increased which are in this case holes. Under such
conditions the Fermi level shifts closer to the valence band.

Energy band
diagram for an n-type
and an p-type
semiconductor
Ref.: J.M. Senior,
Optical Fiber
Communications

Optical Sources 19
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.3 The pn-junction

A pn-junction is realized by the
formations of a semiconductor with
adjacent p- and n-regions. As a
consequence a depletion layer is
formed by the recombination of free
carriers in the vicinity of the pn-
junction. Only the localized states are
left behind which leads to the formation
of a high electric field in the depletion
region. The electric field distribution
(potential barrier) leads to the limitation
of the interdiffusion of the free carriers.
In the absence of an applied bias
voltage no net current is flowing Cross section and band diagram of a
through the diode. The width of the pn-junction under thermal equilibrium.
depletion layer and the magnitude of
the electric field (potential barrier) Ref.: J.M. Senior, Optical Fiber
depends on the doping level in the p- Communications
and n-region.
Optical Sources 20
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.3 The pn-junction

If now a voltage is applied to the pn-junction the potential barrier is either
increased or lowered. For positive applied voltages the diode is forward biased
and the potential barrier is lowered, whereas for negative voltages the diode is
in reverse biased and the potential barrier is increased. As a consequence the
electric field in the depletion layer is increased or decreased.

Optical Sources 21
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.4 Diodes under forward bias

Under forward bias conditions electrons and holes are injected in the
semiconductor via the contacts. Electrons are injected via the n-type region
and holes are injected via the p-tpye region. The schematic operation
behavior of a diode under forward bias conditions is shown in the figure. Due
to injection of carriers the concentration of minority carriers will increase in
the vicinity of the depletion region.

Pn-junction under forward bias. The

recombination of electron hole pair
leads to the spontaneous emission of
photons.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 22
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.4.4 Diodes under forward bias

Electrons are minority carriers in an p-type semiconductor and holes are
minority carriers in an n-type semiconductor. The opposite of minority carriers
are majority carriers. Therefore, electrons are majority carriers in an n-type
semiconductor and holes are majority carriers in an p-type semiconductor.
If we now increase the concentration of minority carriers in a semiconductor a
lot of the carriers will recombine via the majority carriers in this material. For
example the concentration of electrons in the p-region of a pn-diode can be
increased by a applying forward bias to the diode. The concentration of
minority carriers in the vicinity of the depletion region is increased due to the
diffusion of carriers.
We already discussed that the recombination can be radiative or non-radiative.
In the case of non-radiative recombination the energy is dissipated by lattice
vibrations (heating up of the sample).

Optical Sources 23
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5 Light emitting diodes (LEDs)

In the case of radiative recombination the wavelength of the emitted photons is
given by,

c hc 1.24
λ= = = µm Wavelength
f E E (eV )

where c is the speed of light and h is the Planck constant. If a material

(semiconductor) emits light as a consequence of the injection of charges
(current) we speak about electroluminescence. The name implies that the
emission is stimulated by electron injection. The alternative would be
photoluminescene, where we shine light on a sample and the sample emits
light (at a lower energy / higher wavelength). In both cases the emission is
caused by the recombination of electron-hole pairs.

Optical Sources 24
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5 Light emitting diodes (LEDs)

The efficiency of the recombination process, which leads to radiative
recombination (light emission), depends on the semiconducting material itself,
the level of impurities, the structural properties, the temperature and the device
fabrication.
Out of the list of important requirements for an efficient light emission process
the material itself is the most important factor. The most important materials for
the manufacturing of LEDs and laser diodes are out of the class of direct
semiconductors, whereas indirect semiconductors usually exhibit very low
efficiencies in converting electrical energy in light.

Schematic sketch of carrier

recombination in a pn-
junction. The recombination
leads to the emission of
photons (radiative process).
Ref.: J.M. Senior, Optical
Fiber Communications

Optical Sources 25
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.1 Direct and indirect Semiconductors

Typical examples of direct semiconductors are Gallium Arsenide (GaAs),
indium phosphide (InP) or Gallium Nitride (GaN). Typical materials for indirect
semiconductors are silicon (Si) or germanium (Ge). Therefore, the most
important semiconductor (silicon) can not be used as an efficient light emitter.
In order to understand the difference between a direct and an indirect
semiconductor we have to look a little bit closer in the electronic structure of
semiconductor materials. We already discussed that the bands (valence band
and conduction bands) are formed by the superposition of the orbitals for each
atom.

Energy momentum diagram

showing the emission of a
photon for an direct (a) and
an indirect (b) semiconductor.
Ref.: J.M. Senior, Optical
Fiber Communications

Optical Sources 26
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.1 Direct and indirect Semiconductors

However, the band diagram we are using so far is a drastically
simplification of the real picture. If we consider the location of the individual
atoms in a lattice we will get a more precise picture of the electronic
structure. The 3 dimensional nature of a lattice is considered in the
momentum energy diagram for a specific crystal orientation of the material.
We speak about a direct semiconductor if the minimum of the conduction
band and the maximum of the valence band are adjacent to each other and
electron and the hole pairs can directly recombine. In the case of an
indirect semiconductor an additional momentum of the electron is needed
to recombine with a hole in the valence band.

Optical Sources 27
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.2 Device structures

LEDs and laser diodes can be constructed based on surface emitting or edge-
emitting device structures. The surface-emitting devices emit light from a face of
the device, whereas the edge-emitting structure emits light from the edge.
The edge emitting structure is usually applied for solid-state laser diodes. Most
of the LEDs are based on a surface emitting structure. The different structures
have clear advantages and disadvantages. The surface emitting structure is
very well suited for lighting applications. Furthermore, it is possible to test
devices on the wafer level. In the case of a surface emitting laser structure the
cavity of the laser diode is short so that the reflectivity of the mirrors has to be
high, but still the laser diodes can be tested on the wafer level.

Implementation of LED and

laser structures, (a) Surface
emitting Device, (b) Edge-
emitting Device.
Ref.: Saleh & Teich,
Fundamentals of Photonics

Optical Sources 28
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.2 Device structures

The photons are created by the electron-hole recombination in the vicinity of the
depletion region. Therefore, the emission light occurs in this region. The
efficiency of such a structure is relatively low. Such a structure is called a
homojunction. The semiconductors in both regions of the pn diode have the
same bandenergies so that not no discontinuity occurs in the bandgap. To
improve the device performance and therefore the efficiency we have find a way
of producing light in a more localized area, with a greater intensity. The situation
can be improved by introducing a heterostructure.

Homejunction and double Heterojunction LED.

Ref.: H. J.R. Dutton, Understanding optical communications

Optical Sources 29
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.2 Device structures

A heterojunction is different from an ordinary homojunction. In such case a
discontinuity in the band diagram is observed. In such a configuration the
charge carriers (electrons or holes) are attracted over the barrier from the
material of higher bandgap energy to the one of lower bandgap energy. As a
consequence most of the recombination occurs in the region of lower optical
bandgap. The situation can even be improved by using a double
heterostructure.

Optical Sources 30
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.2 Device structures

The double heterostructure is realized by introducing a layer of lower bandgap
material in between two layers of higher bandgaps. A double heterojunction
consists of two heterojunctions. Again the recombination of carriers is restricted
to the low bandgap region which is called „the active region“ of the diode.
An example of an energy diagram of a double heterojunction based on indium
phosphide is shown in the figure.

Double Heterojunction
LED.
Ref.: H. J.R. Dutton,
Understanding optical
communications

Optical Sources 31
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.3 Application of Light emitting Diodes

The manufacturing cost of LEDs is low in comparison to lasers diodes.
However, LEDs are only used for very short range optical communication
systems. The main advantage of LEDs based optical communication systems is
the cost in implementing such systems.
There are two reasons which limit the performance of LED based fiber
communication systems

Incoherent Light: The light produced by a LED is incoherent. Furthermore, the

light is not guided inside of the LED structure, therefore, the light is emitted in all
directions. As a consequence it is difficult to couple light in a fiber. It is almost
impossible to couple the light in a single mode fiber.
Spectral Width: LEDs do not produce a single wavelength but rather a band of
wavelengths. The range (or band) of wavelengths emitted is defined by the
“spectral width”. Depending on the material system and the emitted wavelength
the spectral width is typically in the range of 20-80nm. The spectral width
increases with the square of the wavelength.

Optical Sources 32
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.3 Application of Light emitting Diodes

Optical Output Power
The optical output power of a LED is typically much lower than the output power
of a laser. If we compare semiconductor lasers with LEDs the optical output is
comparable. In particular the output power of LEDs has been drastically
increased due to the development of high brightness LEDs for lighting
applications.

Applications of High
Brightness LEDs in
2002.
Ref.:
www.lumileds.com

Optical Sources 33
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.3 Application of Light emitting Diodes

The light extraction of a LED (high brightness LED) can be improved by the
geometric design of the device structure.

Historical development of LED device design for high brightness LEDs.

Ref.: www.lumileds.com

Optical Sources 34
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.5.3 Application of Light emitting Diodes

Most of the LEDs (high brightness LEDs) are used for applications in the areas
of signs, automotive and mobile appliances.

Applications of High Brightness

LEDs in 2002.
Ref.: www.lumileds.com

Optical Sources 35
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6 Lasers
Laser stands for “Light Amplification by the Stimulated Emission of Radiation”.
Optical Communication Systems with high Bandwidth-Distance products as high
as 10Tbit/s wouldn’t be possible without the invention of the laser in the 1960’s
and the semiconductor laser diode in the 1970’s.

Advantages of Lasers:
• Extreme narrow spectral width (Linewidth)
• Source of coherent light
• Laser can be modulated very fast
• Compact and inexpensive (laser diodes)

Disadvantages of lasers:
• Lasers need temperature and output power control (high cost).
• For optical communication systems a cooler “peltier effect” is needed.
• Analog modulation is difficult due to the non linearity of the output power
(e.g. current threshold)
Optical Sources 36
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.1 Stimulated Emission

The operating principle of a laser is based on stimulated emission. We speak
about stimulated emission if the electron which enters an energetically higher
state (excited state) remains in this state until it is “stimulated” by the presence
of a photon to leave this higher energetically state and return to the more stable
lower energetically state (ground state). One of the requirements for stimulated
emission is that the electron can stay in its excited state a relatively long period
of time (a few microseconds) before it changes its state spontaneously. In the
case of spontaneous emission the electron stays in this excited state usually for
a shorter period of time (picoseconds). In the case of stimulated emission the
electron can be “stimulated” by the presence of a photon to emit its energy in
the form of another photon.

Energy state diagram for

stimulated emission.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 37
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.1 Stimulated Emission

Let‘s assume that we start our discussion on stimulated emission by

introducing a two level atomic system. In general three different transition
processes exist in a two level atomic system: Absorption, the spontaneous
emission and the stimulated emission. In 1927 Albert Einstein demonstrated
that these processes are mathematically related. We will not discuss a
mathematical description of these transition processes here. In general, due to
thermal equilibrium the rate of upwards transitions has to be equal to the rate of
downwards transitions.
Rate equations can be defined, which describe the probability of a transition
from level 1 (lower energetic state or ground state) to level 2 (higher energetic
state) and vice versa. For the level 1 rate equation we have to considered only
the absorption, whereas for the level 2 rate equation we have to consider
spontaneous and stimulated emission. This leads to the following relationship:

Spon tan eous emission rate 1

=
Stimulated emission rate  hf 
exp  −1
 kT 
Optical Sources 38
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.1 Stimulated Emission

For most of the systems in thermal equilibrium spontaneous emission is by far
the dominate process. Therefore, the emission of an ordinary light source in the
visible spectrums occurs in a random fashion, which as a consequence leads
to incoherent light.
The spontaneous emission term of the rate equation of level 2 is much larger
than the term of stimulated emission. In order to produce coherent light the
stimulated term has to be drastically increased.

5.6.2 Population inversion

In the case of a two level system in thermal equilibrium, which can be
described by a Boltzmann distribution, the lower energy level E1 contains more
atoms than the upper energy level. This situation is normal for structures at
room temperature.
To achieve optical amplification (stimulated emission) it is necessary to create
a non-equilibrium distribution of atoms. The population of atoms in the higher
energetic state has to be (significantly) higher than the population of atoms in
the lower energetic state.
Optical Sources 39
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.2 Population inversion

This situation is called population inversion.

Population in two energy level system:

(a) Boltzmann distribution for a system in thermal equilibrium,
(b) Non-equilibrium distribution indicating the population inversion.
Ref.: J.M. Senior, Optical Fiber Communications

Optical Sources 40
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.2 Population inversion

In order to achieve population inversion it is necessary to excite atoms into the

higher energetic state and hence achieve population inversion. This process is
usually called “pumping”. We can distinguish between electrical and optical
pumping. The medium is optically pumped for gas lasers (e.g. HeNe lasers) or
crystal lasers (e.g. Ruby laser) where the atoms are excited by an external
optical source. The systems are usually pumped by an intensive radiation
source like a flash lamp. The intense radiation leads to the transition of the
atoms from the lower to the energetically higher states which we would
normally be called absorption, but we can even call it “stimulated absorption”
The term “stimulated emission” indicate the similarities between the absorption
and the stimulated emission. (The stimulated emission is the inversion of the
(stimulated) absorption.)
However, so far we discussed the behavior of two level systems which are not
suitable for population inversion. The probability of absorption and emission
can be in the best case to be equal.
Population inversion however, can be observed in certain three or four energy
level systems.

Optical Sources 41
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.3 Three and four energy level systems

The principle transition for a three level system is illustrated in the figure. The
three level system consists of a ground level E0, a stable level E1 and a third
level E2 which is again metastable. Initially (before pumping) the system is in
thermal equilibrium and the atomic distribution can be described by a
Boltzmann distribution.

Energy diagrams showing population

inversion for a three level system (ruby
crystal).
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 42
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.3 Three and four energy level systems

Due to pumping electrons get excited from the ground level to the level 2. The
level 2 is considered to be a “normal” energy level so that the electrons rapidly
decay in the lower energetic state 1. Of course a certain number of electrons
will directly go back to the energy level 1, but most of the electrons will end up
on level 1. As a consequence empty states will be always available on level 2.
The lifetime of the electrons on level 1 is however much longer than the lifetime
of the carriers on level 2, so that a large number of electrons is accumulated on
level 1. The long lifetime of the carriers on level 1 leads to the population
inversion between level 1 and the ground level. The stimulation of an electron
on level 1 leads now to lasing.
The disadvantage of a three level system is that it needs very high levels of
pumping power. More than half the electrons in the ground state have to
pumped in order to achieve population inversion.
A more efficient system is a four level system as it was shown for a HeNe laser
system. Here much lower pumping power is needed. In this case an atom is
pumped from the ground state into the highest energetic state on level 3.

Optical Sources 43
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.3 Three and four energy level systems

Again the transition from level 3 to level 2 is a rapid decay, because the lifetime
of the carriers on level 3 is short. The lifetime of carriers on level 2 however is
much longer (several orders of magnitude) so that population inversion is
observed for the level 2. As a consequence lasing is detected between the
levels 3 and 2.

Energy diagram
showing population
inversion for a four level
system.
Ref.: H. J.R. Dutton,
Understanding optical
communications

Optical Sources 44
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.4 Optical feedback and laser resonators

Stimulated emission and therefore lasing occurs when a photon stimulates the
emission of another photon and then both photons continue to stimulate the
release of further photons. In order to generate coherent light the stimulation
process has to be continuous and effective. An amplification due to the
stimulated multiplication of photon is necessary to accomplish coherent
emission. To ensure that the emitted light is coherent the laser medium is
placed in a laser resonator which can be formed by two mirrors. The laser
medium and the two mirrors form a laser cavity. Hence the emitted light is
reflected at the mirrors and fed back into the resonator to simulate more
photons. Such a structure is called a Fabry Perot resonator.

Schematic cross section of

a laser structure based on a
laser medium and two
partially transmissive
mirrors.
Ref.: J.M. Senior, Optical
Fiber Communications

Optical Sources 45
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.4 Optical feedback and laser resonators

The amplification of the signal after one roundtrip is usually relatively small, but
after several roundtrips the net amplification of the structure can be large.
Typically the mirrors of such a Fabry Perot resonator are not perfect. A small
fraction of the light is transmitted through the mirrors. As the mirrors typically
consist of a semiconductor layer stack rather than a metal mirror the
transmission of the mirror can be controlled very precisely by the manufacturing
process. Therefore, the mirrors are partially transmissive.
A continuous and stable optical output signal can be achieved when the gain of
the structure is higher than the loss of the structure. Loss is caused by the
absorption and the scattering of light in the cavity and absorption losses in the
partially transmissive mirrors. Furthermore, the light has to be guided inside
the structure which leads to further losses. The wavelength of the emitted light
depends on the radiative transitions in the laser medium. However, it is clear
that the emitted light is not perfectly monochromatic, moreover, a narrow
spectrum of wavelengths is emitted. The emission spectrum follows the
frequency range determined by the gain curve of the structure.

Optical Sources 46
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.4 Optical feedback and laser resonators

Due to the formation of the laser cavity a sufficient population inversion exists
in the laser medium. The superposition of the back and forth propagating
waves leads to the formation of a standing wave in the resonator. The standing
wave exists only for wavelengths for which the distance of the mirrors is an
integral number of the half of the wavelength.

λ Resonance condition
L=k⋅
2n

Relative amplification in a
laser cavity showing the
broadening of the laser gain.
Ref.: J.M. Senior, Optical
Fiber Communications

Optical Sources 47
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.4 Optical feedback and laser resonators

where n is the refractive index of the laser medium, k is an integer and λ is the
wavelength of the emitted light.
Based on the resonance condition the spectral width of the amplified signal can
be determined to be
λ2
∆λ = Spectral width
2nL
Based on the equation it can be seen that the spectral width can be reduced by
increasing the length of the cavity. However, at the same time the number of
modes which can propagate in the cavity is increased. The number of mode
can be calculated by:

2nL
k= Propagating modes
λ

Optical Sources 48
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.4 Optical feedback and laser resonators

The propagation of modes in the laser resonator is illustrated in the figure. The
spectral output of the laser is defined by the gain curve.

Propagating modes in a laser

cavity and modes in the laser
output.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 49
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.5 Threshold condition for laser oscillation

Lasing only occurs when the gain in the amplification medium equals the overall
losses of the laser structure.
Therefore, population inversion alone is not sufficient to achieve laser operation.
A minimum or a threshold gain has to be overcome to initiate and sustain laser
emission. In the following we will derive a simple expression for the required
threshold gain. In the first step we assume that the overall losses of the
structure can be described by a loss (or absorption) coefficient per unit length.

α Loss coefficient per unit length

The length of the cavity is given by L and the reflectivity of the mirrors is
provided by R1 and R2. Based on the given parameters we can calculate the
loss for a round trip:

(
Fractional loss = R1R 2 exp − 2α ⋅ L ) Fractional loss

Optical Sources 50
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.5 Threshold condition for laser oscillation

Not only the absorption but also the gain of a round trip can be described by a
power law,
(
Fractional gain = exp 2g ⋅ L ) Fractional gain

where the gain coefficient per unit length is given by:

g Gain coefficient per unit length

In the case of threshold the fractional losses have to be equal to the fractional
gain so that we can define a threshold gain by
1  1 
g th =α + ⋅ ln  Threshold gain
2L  R1R 2 
Hence, the threshold gain corresponds to the absorption losses plus the light
that is transmitted through the partially transmissive mirrors.

Optical Sources 51
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.6.6 Requirements for lasing

In summary, the following requirements have to be fulfilled to realize a laser.

1. Population inversion is required to achieve stimulated emission.

2. A material system which exhibits several energy states of which a high
energy state has to be metastable. The wavelength of the emitted light will
be than defined by the energy difference between the higher metastable
energy state and the next lower energy state.
3. The material (system) has to be transparent for the emitted light.
Otherwise the light can not leave the material.
4. Confinement of the material and the light by a resonator.
5. The threshold gain has to be larger than the fractional losses plus the
emitted light.

Optical Sources 52
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7 Semiconductor Lasers

So far we discuss the concept of lasing on a generic level. For example
population inversion is necessary for all kinds of lasers. It does not matter
whether the laser structure is a gas, a crystal laser or a semiconductor laser
which is electrically or optically pumped.
The same is true for the resonator structure. The gain of the laser structure has
to be higher than the overall losses.

5.7.1 Stimulated emission and lasing in Semiconductors

Now we will concentrate on semiconductor laser diodes. We will extend the
discussion we already started as part of the LED presentation.
A major requirement for the stimulated emission is the carrier population
inversion which is achieved in an intrinsic (undoped) semiconductor by the
injections of electrons in the conduction band and holes in the valence band.
However, under “normal” conditions the number of injected carriers is not high
enough to get stimulated emission. The recombination of carriers will lead to
spontaneous emission as we already discussed for the LED structure.

Optical Sources 53
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.1 Stimulated emission and lasing in Semiconductors

In order to achieve population inversion the concentration of carriers in the
bands has to be significantly increased.
One way to increase the concentration of carriers in the bands is doping. So far
we assumed that the p- and the n-regions of the laser diode are “normally”
doped, which leads to a shift of the Fermi level.

Filled electron states in a direct

bandgap semiconductor in
equilibrium and at high carrier
injection.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 54
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.1 Stimulated emission and lasing in Semiconductors

In the case of n-type doping the Fermi level shift towards the conduction band
and in the case of p-type doping the Fermi level shift towards the valence band.
The higher the doping concentration the closer the Fermi energy will be to the
bands. Population inversion in a pn-junction can only be reached if the p- and
the n-type material is very heavily doped. Such heavy doping is called
degenerated doping. The doping level is so high that the Fermi levels shift into
the conduction and the valence band.

Degenerated pn-junction in
thermal equilibrium.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 55
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.1 Stimulated emission and lasing in Semiconductors

An incident photon with an energy of higher than Eg but less than Eq cannot be
absorbed, because the necessary states in the conduction band are already
occupied by electrons. However, this photon can stimulate a downward
transition of an electron from the filled conduction band states to the empty
valence band state.
The basic condition for stimulated emission is therefore given by the following
relationship:

E FC − E FV = E q > hf > E g Condition for stimulated emission

Instead of using the occupation of states in the conduction and valence band we
can describe stimulated emission in terms of quasi Fermi levels. So far we used
only the term “Fermi level”. The term Fermi level, however, describes the
probability of states being occupied by electrons or holes in thermal equilibrium.

Optical Sources 56
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.1 Stimulated emission and lasing in Semiconductors

In the case of non thermal equilibrium which occurs under forward or reverse
bias conditions the Fermi level splits up so that we can define separate Fermi
levels for electrons and holes.
Population inversion is observed if the quasi Fermi levels shift into the band so
that an incident photon cannot excite an electron form the valance band to the
conduction band because the states in the conduction band are already
occupied. As a consequence an downward transition will be observed.

Degenerated pn-junction under

strong forward bias so that the
separation of the quasi Fermi
level in the active region is
higher than the energy of the
emitted photons. Therefore,
stimulated emission is obtained.
Ref.: J.M. Senior, Optical Fiber
Communications
Optical Sources 57
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.1 Stimulated emission and lasing in Semiconductors

In order to obtain stimulated emission in the active region of the laser diode a
high forward bias has to be applied so that the separation of the quasi Fermi
level in the active region is higher than the energy of the emitted photons. As
consequence the incident photon stimulates emission rather than excites an
electron form the conduction band to the valence band (absorption).
Only if all these requirements are satisfied
stimulated emission can be observed for a
semiconductor laser diode.
For current levels below the threshold the
emission of the laser diode is spontaneous
emission. Only for current levels higher than
the current threshold the emission is
stimulated.

Ideal light output curve of a laser diode against the injected current.
Ref.: J.M. Senior, Optical Fiber Communications
Optical Sources 58
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.2 Semiconductor Materials for lasing applications

As we already learned the materials applied for the manufacturing of laser
diodes have to fulfill several requirements. The list of materials suitable for the
realization of laser diodes is almost identical with the list of materials used for
LEDs. In general only direct semiconductors are used for the realization of laser
diodes. A detailed list of the materials can be found in literature.
The requirements in terms of the material quality and the mirror are much higher
for a laser than for a LED structure (A LED usually does not even have mirrors).
The fact that the materials are sufficiently good for an LED structure does not
mean it is sufficiently good enough for a laser diode.

Optical Sources 59
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.3 Efficiency of LEDs and laser diodes

The efficiency of a LED or a laser diode can be defined by different ways. We will
start with the internal efficiency which can be calculated by:

number of emitted photons Internal efficiency

ηi =
number of injected electrons

The internal efficiency can be very high. Depending on the material quality and
the devices structure the efficiency can be close to 100%. The external efficiency
can be determined by,
number of output photons Pop
η ex = = External efficiency
number of injected electrons I ⋅ E g

where Pop is the emitted optical power which changes linearly with the injected
current I greater than the threshold current Ith.

Optical Sources 60
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.3 Efficiency of LEDs and laser diodes

The external efficiency can be calculated by

 
 
 1   I th  External efficiency
η ex = ηi   ⋅ 1 − 
 1 +  2α ⋅ L    I 
  ln(1 R1R 2 )  
  

The external optical power efficiency can be determined by:

Pop  Eg 
η op = = η ex ⋅  

External optical power
Pe  eV  efficiency

Optical Sources 61
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.4 Laser Diode structures

5.7.4.1 Fabry Perot Homojunction laser diode
A schematic diagram of an early GaAs laser diode is shown on this slide. The
cavity of the laser is formed by a Fabry Perot structure. The cleaved edges of the
structure act as partially transmissive mirrors. The device structure itself consists
of a homojunction GaAs pn-junction. The threshold current of such a structure is
very high, because the reflection of the partially transmissive mirrors is relatively
low and the standing wave inside of the cavity is not confined. The confinement
problem could be solved by using a double heterostructures.

Schematic diagram of a GaAs

homojunction laser diode based
on a Fabry Perot cavity.
Ref.: J.M. Senior, Optical Fiber
Communications

Optical Sources 62
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.4.2 Double heterostructure laser diode

The double heterostructure provides
an electrical and an optical
confinement within the laser diodes
which leads to a drastically increase of
the efficiency. However, the structure
is only confined in the vertical
direction, but lasing occurs along the
entire width of the structure. To solve
this problem the active region has to
be confined along the active region.
This problem is usually solved by
guiding the light parallel to the lasing
direction.

Directing and guiding light in a Fabry

Perot laser structure.
Ref.: H.J.R. Dutton, Understanding
Optical Communications

Optical Sources 63
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.4.2 Double heterostructure laser diode

The easiest way to confine the light in lateral direction is a gain guided structure.
In this case the laser structure is electrically confined. A region of high gain is
created by the patterning or definition of the electrodes or contacts of the laser
diode. As a consequence carriers are only injected in certain parts of the device
so that the light is defined by the region of high gain.
A more efficient structure is an index guided structure, where the light is guided
optically. Similar to a wave structure regions of high and low refractive index are
created which guide the light. If the optical guidance is combined with an
electrical confinement of the laser cavity an efficient laser diode can be formed.

Optical Sources 64
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.4.3 Quantum well lasers

The fabrication technique of lasers has been
improved drastically throughout the last 20-30
years so that the layer thickness can now be
controlled down to the sub nanometer range.
This offers the possibility of manufacturing
double heterostructure lasers with very thin
active regions of around 10nm. If the
dimension of the active region is reduced to
that range the carrier motion normal to the
active region of the laser gets restricted.
Instead of having well defined electronic bands
we have well defined discrete energy level like
it is predicted by quantum mechanics. We get
quantized energy levels.

Energy band diagrams showing various types of quantum well structures,

(a) single quantum-well structure, (b) multiple quantum well structure (c) modified
quantum well structure. Ref.: J.M. Senior, Optical Fiber Communications
Optical Sources 65
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.4.3 Quantum well lasers

In such a case we speak about a quantum well like it is shown in figure a on the
previous slide. In figure a only single quantum well is shown therefore the
structure is called single quantum well structure.
This effect can be used for lasers, because these quantum well structures
exhibit the fundamental advantages that high gain can be accomplished at low
carrier densities and therefore a significant lower threshold currents.
Instead of using a single quantum well a series of quantum well can be used
and we use the general term of a quantum well structure. Nowadays this
process has to been optimized leading to modified quantum well structures.

5.7.4.4 Distributed Feedback (DFB) Lasers

Fabry Perot lasers exhibit significant problems if they are used for long distance
communication. Wavelength Division Multiplexed (WDM) systems require the
transmission of several channel in a single fiber. To do this it is important for
each signal/channel to have a narrow spectral width. The spectral width of a
Fabry Perot lasers however is too high for such applications.

Optical Sources 66
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

5.7.4.4 Distributed Feedback (DFB) Lasers

Distributed Feedback (DFB) lasers are much more suitable for long distance
communication. The idea is that a Bragg grating is put into the laser cavity of an
index-guided Fabry Perot laser. The operation and the device structure will be
presented later on (Student Presentation).

5.7.4.5 Vertical Cavity Surface Emitting Lasers (VCSELs)

Vertical Cavity Surface Emitting Lasers (VCSELs) got very popular during the
last 5-10 years. VCSELs are very attractive light sources for metropolitan
networks, where the requirements in terms of stability, output power, cost and
mode propagation are different from wide area networks. However, the
advantages of a VCSEL structure are clear. The lasers exhibit low threshold
current and the structures can be tested on the wafer level. The operation and
the device structure will be presented later on (Student Presentation).

Optical Sources 67
 Photonics and Optical Communication, Spring 2007, Dr. D. Knipp

References:
John M. Senior, Optical Fiber Communications, Prentice Hall Series in
Optoelectonics, 2nd edition, 1992.
Bahaa E.A. Saleh, Malvin Carl Teich, Fundamentals of Photonics,
Wiley-Interscience (1991)
Harry J. R. Dutton, Understanding Optical Communications,
Prentice Hall Series in Networking, 1998. (Formerly freely available as a red
book on the IBM red book server.
Stamatios V. Kartalopoulos, DWDM, Networks, Devices and Technology,
IEEE press and Wiley Interscience, 2003.
Joseph C. Palais, Fiber Optic Communications,
Prentice Hall Series, 1998. 4th edition.

Optical Sources 68