Integrated Photonics
Integrated Photonics
Integrated Photonics
Integrated Photonics
Integrated Photonics
© 2016 University of Central Florida
This text was developed by the National Center for Optics and Photonics Education (OP-TEC),
University of Central Florida, under NSF ATE grant 1303732. Any opinions, findings, and
conclusions or recommendations expressed in this material are those of the author(s) and do not
necessarily reflect the views of the National Science Foundation.
ISBN 978-0-9858006-6-6
PREFACE
Silicon photonics (SiP) is an emerging field in opto-electronics. In silicon photonics,
complementary metal-oxide semiconductor (CMOS) electronics foundries are used to
manufacture optical chips, known as photonic integrated circuits (PICS). Revolutionary new SiP
systems are being developed for applications such as optical and wireless communications,
bioenvironmental sensing, and computing. Silicon photonics is currently in the early stage of
expansion, as electronics was in the 1970–80s, but with a major advantage for chip fabrication:
existing silicon foundries.
The rapidly emerging field of integrated photonics, sometimes called silicon photonics (SiP)
requires engineering technicians who work in microchip or nanochip fabrication facilities and in
materials R&D fabrication laboratories. These technicians need to understand not only the
technology, hardware, and procedures of clean rooms and high-vacuum deposition, but also
geometric and wave optics applications and diagnostic equipment, as well as vertical cavity,
surface-emitting laser diodes (VCSELs) and nanofabrication equipment, procedures, and
inspection techniques. Technicians currently working in chip fabrication facilities require
additional education and training in the technologies that support SiP.
Integrated Photonics is a five-module course that provides an overview of the technology,
device characteristics, fabrication techniques and equipment, and applications in high-speed
computing, telecommunications, biomedicine, and other fields. The contents are:
Module 1. Photonic Integrated Circuits: Materials and Fabrication Technologies
Module 2. Silicon Photonic Integrated Circuits and Devices
Module 3. III-V and Compound Semiconductor Devices
Module 4. Dielectric and Polymer Waveguides and Waveguide Devices
Module 5. Integrated Photonic Circuits and Systems
Prerequisites for this course include knowledge of technical mathematics and solid-state
materials, as well as Fundamentals of Light and Lasers (OP-TEC Course 1).
Sections in the back of the book after Module 5, contain a glossary of terms and an index.
OP-TEC would appreciate feedback and comments on your use of this material.
Acknowledgments
The author of these materials is Dr. Anca Sala, Dean of Engineering and Computer Technology
at Baker College in Michigan. Dr. Sala is an experienced engineer in integrated photonics.
iii
CONTENTS
Module 1: Photonic Integrated Circuits: Materials and Fabrication Technologies
Module 2: Silicon Photonic Integrated Circuits and Devices
Module 3: III-V and Compound Semiconductor Devices
Module 4: Dielectric and Polymer Waveguides and Waveguide Devices
Module 5: Integrated Photonic Circuits and Systems
v
Photonic Integrated
Circuits
Materials and Fabrication
Technologies
Module 1
of
Integrated Photonics
ISBN 978-0-9903125-5-0
CONTENTS OF MODULE 1
Introduction .................................................................................................................................... 1
Prerequisites ................................................................................................................................... 3
Objectives ....................................................................................................................................... 3
Scenario .......................................................................................................................................... 4
Basic Concepts ............................................................................................................................... 5
What are Photonic Integrated Circuits? ..................................................................................... 5
Optical Waveguides ............................................................................................................... 6
Optical Waveguide Modes ..................................................................................................... 9
Polarization of Light Waves Propagating in Optical Waveguides ....................................... 11
Mode Field Distributions in the Waveguide Cross Section ................................................. 12
Planar Optical Waveguides .................................................................................................. 14
Characteristics of Planar Optical Waveguides ..................................................................... 16
Materials Used to Fabricate PICs ......................................................................................... 19
Photonic Integrated Circuit Fabrication ................................................................................... 21
Deposition ............................................................................................................................ 21
Photolithography and Etching .............................................................................................. 22
Second Deposition................................................................................................................ 23
Passivation ............................................................................................................................ 23
Metallization ......................................................................................................................... 23
Direct Laser Writing ............................................................................................................. 24
Equipment used in Fabrication of PICs.................................................................................... 24
Deposition Equipment .......................................................................................................... 24
Photolithography Equipment ................................................................................................ 25
Etching Equipment ............................................................................................................... 26
Summary ...................................................................................................................................... 27
Problem Exercises and Questions ................................................................................................ 28
References .................................................................................................................................... 29
Module 1
Photonic Integrated Circuits
INTRODUCTION
Photonics encompasses the science of light and the technology of using light to generate and
control energy and to transmit and detect information. The quantum unit of light is the photon. It
is expected that the twenty-first century will depend as much on photonics as the twentieth
century depended on electronics. During the last century, tremendous technological advances
were made possible by the invention of the electronic integrated circuit (EIC) and its further
development.
The rapid pace of advance of EICs is captured by Moore’s law, which states, “The number of
transistors in an integrated circuit doubles every two years.” This has held true for four decades,
starting in the 1970s, and has resulted in the now ubiquitous and affordable personal computers,
cell phones, tablets, and the myriad other consumer electronic products we are all using today.
Figure 1-1 shows the growth in the number of transistors in a microprocessor chip between 1970
and 2010. Today it appears that the expansion predicted by Moore’s law is starting to level off.
Some limiting factors that come into play as the number of processors per chip increases are
power dissipation, packaging, and communication bottlenecks between chips. At the same time,
the younger field of integrated photonics is now poised for rapid growth and expansion, and it is
expected that this field’s growth will help sustain the pace of progress we have enjoyed so far.
Using photonic integrated circuits (PIC), for example for communication between chip
processors and from one chip to another, can mitigate the limitations above.
A photonic integrated circuit is a device that integrates multiple photonic functions, similar to an
electronic integrated circuit. Another term that describes such devices is the optoelectronic
integrated circuit (OEIC). For example, a photonic interconnection network can route
information between the processor and memory chips in a computer. PICs offer many
advantages: when replacing a system constructed from bulk or discrete optical components, they
allow the system to be much more compact and efficient and to operate at higher speeds.
1
Figure 1-1 Illustration of Moore’s law, which predicts that the number of transistors in a
microprocessor chip doubles every two year
Electronic and photonic integrated circuits are not just analogous terms used to describe
electronic vs. optical integrated devices. EICs and PICs are also related in other ways. They are
in some cases made from the same materials and using similar fabrication processes. Being able
to leverage the very mature technology of fabrication and advanced equipment from the
semiconductor industry to produce PICs is very attractive. However, while the EICs today are
predominantly made out of silicon and fabricated using the technology of the complementary
metal-oxide semiconductor (CMOS), PICs are made from a variety of materials, depending on
their functionality. This is due to the fact that no one material platform and technology is
currently capable of producing all the required photonic components at high volume, low cost,
and the desired level of performance. Certain materials are better suited for the fabrication of
semiconductor lasers, others are best for photodetectors, and still others are appropriate for the
variety of optical components used in telecommunications and photonic sensors.
The ultimate goal of integrated photonics is to combine together the optical sources, processing,
and detection components in one circuit, leading to minimum cost and footprint and avoiding
complicated assembly operations. Integrating all devices on the same platform is called
monolithic integration. Moreover, photonic and electronic circuits could be integrated together
in one circuit, resulting in the best performance and cost functionality. These technological
advances will fuel the growth and integration of vast, low-power, reliable networks of connected
devices that collect, process, communicate, and exchange data. Integrated optical and photonic
systems will enable new and emerging networking technologies, sensors, high bandwidth
connectivity, and processing and storage technologies. These technologies include athletic
wearables with integrated sensors that collect, process, and communicate biometric information;
medical devices that use integrated labs-on-chips without drawing blood, thus eliminating
needles; and advanced automotive driver-assistance systems, including surrounding sensors,
intelligent headlamps, optical car-to-car communications, and heads-up displays, along with
self-driving features. While this level of integration, information collection, processing, and
communications does not exist today, groups of scientists and engineers around the world are
working to create faster, more compact, and lower-cost components to bring these possibilities
closer to reality.
PREREQUISITES
OP-TEC’s Fundamentals of Light and Lasers Course 1
OBJECTIVES
Upon completion of this module, the student should be able to:
Define photonic integrated circuits (PIC)
Give examples of PIC and their applications
Explain monolithic integration and its advantages
Describe the principle of operation of optical waveguides
Explain and calculate the refractive index contrast, numerical aperture, and maximum
half-acceptance angle of an optical waveguide
Define modes of optical waveguides
Explain the electromagnetic field intensity distribution in the cross section of the
waveguide
Describe planar optical waveguides and their geometries
Determine whether an optical fiber and a slab waveguide are single mode or multimode
Describe basic characteristics of planar optical waveguides, including dimensions, bend
radius, propagation loss, and thermo-optic effect
Identify the main groups of materials used for PIC and their properties
Describe the fabrication technologies used to make PIC
Specify equipment used to fabricate PIC in a clean-room environment, such as
deposition, photolithography and etching systems, and direct laser writing systems
SCENARIO
Antonio works as a wafer fab equipment technician with a company that designs and
manufactures photonic integrated circuit devices. His duties are in maintenance,
troubleshooting, upgrades, and installations of clean-room equipment. Antonio starts his day at
his computer, where he reads the reports from the previous shift to see what tools are down and
reviews charts and maintenance needs. After reviewing the equipment status, he puts on the
clean-room bunny suit and enters the clean room. He works on equipment that needs
maintenance, calibrates tools so they can be put back in production, and addresses urgent
problems that show up. When a tool does not perform at the required level or stops working,
Antonio gathers information from the process team to help him troubleshoot the problem. He
discusses issues with engineers to help him decide if the equipment can be repaired with
existing parts or if new parts must be ordered.
Throughout the day, Antonio communicates with other team members through e-mail, page, and
text messages. He often needs to multitask and resolve multiple priority issues quickly and
efficiently to improve the overall fab performance. He loves solving problems and working to
improve the tools and process performance. As new equipment is added to the fab, there is
always something new to learn. Antonio’s job is challenging and rewarding, and he brings an
important contribution to the success of his company.
BASIC CONCEPTS
Figure 1-2 Illustration of distributed Bragg reflector semiconductor laser. The active section and the
grating sections serving as mirrors are integrated in the same structure.
In the telecommunications field, a variety of PICs are used to process the optical signals carried
by optical fibers, including devices such as couplers, switches, modulators, and filters. While
couplers and switches distribute and route the optical signals from various input ports to various
output ports, modulators operate on the phase and power level of the signal, and filters operate
on the signal wavelengths. A PIC, which integrates optical sources with a circuit to route the
optical signals to various outputs and with photodetectors to convert optical to electrical signals,
is shown in Figure 1-3.
Figure 1-3 PIC integrating optical sources, a routing circuit, and photodetectors
PICs are expected to play a growing role in the further development of computers and data
servers. They are capable of connecting electronic chips to one another at very high speeds,
using low power. Other applications of PICs are in photonic sensors applied in biological and
biomedical fields. PIC-based sensors, capable of optically measuring human blood-glucose
levels without the need for a blood sample, have been demonstrated. Later modules will
describe in detail photonic integrated circuits used in applications such as the ones mentioned
above.
Optical Waveguides
The basic element of all PICs is the optical waveguide. An optical waveguide is a physical
structure that guides and confines light waves. Optical fibers are a familiar type of optical
waveguide that consist of a cylindrical core surrounded by a cladding. Because the core has a
higher index of refraction than the cladding, total internal reflection (TIR) at the interface
between core and cladding confines and guides the light traveling down the fiber.
There are two conditions that must be met for total internal reflection to occur. One condition is
that the index of refraction of the medium in which the light beam is traveling (in this case, the
core) must be greater than the index of refraction of the medium that the beam is moving toward
(the cladding). The second condition is that the angle of incidence of the light beam at the
interface between core and cladding must be greater than the critical angle, which is given by
the following equation:
Where: nclad is the index of refraction for the cladding material and
ncore is the index of refraction for the core material
The optical fiber, shown in Figure 1-4, consists of a central glass core of radius “a” surrounded
by an outer cladding made of glass with a slightly lower refractive index. The corresponding
refractive index distribution in the transverse direction is given by
Figure 1-4 shows a light ray incident on the air-core left interface at an angle i. The ray refracts
at angle in accordance with Snell’s law (the law of refraction) and then strikes the core-
cladding interface at angle . In the drawing shown in Figure 1-4b, the angle is greater than
the critical angle c defined in Equation 1-1, thus leading to total internal reflection at A. The
reflected ray is internally reflected again at C and B and so on, remaining trapped in the fiber as
it propagates along the core axis.
Example 1
An optical fiber has a core index equal to 1.465 and a cladding index equal to 1.45.
Calculate the critical angle for total internal reflection at the interface between core and
cladding.
-1 -1
Solution: Using Equation 1-1 we obtain ϕc = sin (nclad/ncore) = sin (1.45/1.465) = 81.8º.
In this case, light inside the waveguide must strike the core-cladding interface at angles
greater than 81.8 º to remain confined to the waveguide core.
An important parameter for optical waveguides is the numerical aperture (NA). The numerical
aperture of a waveguide is a measure of its light-gathering ability and is defined by
where (a)max is the maximum half-acceptance angle of the waveguide, as shown in Figure 1-5.
The larger the numerical aperture is, the greater the light-gathering ability of the fiber is. Typical
values for NA are 0.1 to 0.2 for single-mode fiber. The numerical aperture is related to the index
of refraction of the core and cladding by the following equation:
As can be seen, the larger the difference between the core and cladding indexes, the larger the
numerical aperture.
The NA may also be expressed in terms of the relative refractive index difference (), where
For the case of ncore nearly equal to nclad the equation simplifies to
Another name for the relative refractive index difference is refractive index contrast. Combining
Equations 1-4 and 1-6, we obtain a useful relation, Equation 1-7.
Equation 1-7 shows that a large refractive index difference between core and cladding results in
a large acceptance angle and large numerical aperture. Depending on the application for the
optical fiber, values for the refractive index contrast are in the range from 0.003 to 0.03 (or 0.3%
to 3%).
Example 2
Find the refractive index contrast, numerical aperture, and maximum half-acceptance angle
for an optical fiber with core index equal to 1.465 and cladding index equal to 1.45.
Solution: Using Equation 1-5,
Δ = (ncore2 – nclad2)/(2 ncore2) = (1.4652 – 1.452)/(2 x 1.4652) = 0.01 = 1%
The numerical aperture is found using Equation 1-7:
NA = ncore (2Δ)1/2 = 1.465 (2 x 0.01)1/2 = 0.21
The maximum half-acceptance angle can also be calculated from Equation 1-7:
(a)max = sin-1(NA) = 12.1º.
Optical fibers are made of glass, which is silicon dioxide (also called silica). The cladding is
usually pure silica, while the core is silica doped with germanium, which increases the refractive
index slightly from nclad to ncore.
Here, Δϕ is the phase difference between the two interfering waves, and m is an integer. The
equation states that constructive interference takes place when the phase difference is an integer
multiple of 2π.
When Equation 1-8 is applied to the use of waveguides to determine the specific angles of
incidence that allow for the propagation of light, complicated mathematical equations result.
Without going into details, the solutions of those equations give us the specific angles of
incidence that work for light propagation. The light waves corresponding to these angles are
called the modes of the waveguide.
The modes are labeled by integer values, denoted by the letter m, starting with 0 for the first
mode and continuing with 1, 2, and so on. Negative values of m are not used in waveguides.
The mode with m = 0 is called the fundamental mode of the waveguide. It corresponds to the
greatest angle of incidence for light traveling in the waveguide. The modes labeled 1, 2, etc. are
called higher-order modes. The angle of incidence gets progressively smaller as m gets larger,
but remains higher than the critical value, ϕc. (If the angle drops below ϕc the mode will not be
guided any longer and will leak out of the core.)
Figure 1-6 shows the direction of the light beams for modes 0 to 3. The fundamental mode
travels almost along the axis of the waveguide, while the higher-order modes often strike the
core-cladding interface.
Figure 1-6 Illustration of the light direction of propagation for fundamental and higher-order
modes of a waveguide
The number of allowable modes to propagate in a waveguide depends on the wavelength and
the polarization of the light, the indexes of refraction of the core and cladding materials, and the
dimensions of the core. Generally, a large refractive index difference between core and cladding
and/or a large core allow for many modes to propagate. For a given wavelength, once the core
and cladding indexes have been chosen, the size of the core determines the number of allowable
modes. In some applications, only one mode is allowed to propagate. An example is in long-
haul optical fiber transmission systems, where light travels for hundreds and even thousands of
miles in the optical fiber. Because the different modes travel with slightly different speeds, they
do not arrive at a destination together, which introduces distortions in the signal or transmitted
data. Using only one mode for light propagation avoids this issue.
A waveguide that only allows the fundamental mode (m = 0) to propagate is called a single-
mode (SM) waveguide. If more modes propagate in a waveguide, the waveguide is called
multimode. For a given wavelength and indexes of refraction of the core and cladding, there will
be a maximum radius for the core of the fiber that allows the optical fiber waveguide to be
single mode. A typical single-mode optical fiber has a refractive index contrast of 0.3% to 0.8%
and a core diameter of 8–10 µm.
The wavenumber of the light is expressed by the following equation:
k = 2π/λ (1-9)
The so-called V parameter of the fiber allows us to determine whether a fiber is single or
multimode. The V parameter can be calculated with the following equation, where a is the fiber
core radius:
Example 3
Determine whether an optical fiber with core radius equal to 9 µm, a cladding index of
1.445, and a core index of 1.455, operating at a wavelength of 1550 nm, is single mode.
Solution: To obtain the correct answer, use Equation 1-10 to calculate the V parameter of
the fiber, being careful to express the wavelength and the core radius in the same units.
Convert the wavelength to microns (λ = 1550 nm = 1.55 µm). Obtain the fiber radius, a =
diameter/2 = 9/2 µm = 4.5 µm.
We are now ready to calculate the V parameter.
V = (2π/1.55 µm) (4.5 µm) (1.4552 – 1.4452)1/2 = 3.11.
Because V > 2.405, the fiber is multimode. The fiber can become single mode if the core
index or the core diameter are reduced.
Other polarizations for electromagnetic waves are Transverse Electric (TE) and Transverse
Magnetic (TM). Let’s assume the direction of propagation is along the z axis in a Cartesian
system of coordinates. In the case of TE waves, the electric field is perpendicular to the
direction of propagation, so the vector is aligned to a direction in the x-y plane, for example the
x-axis. For TM waves, the magnetic field is perpendicular to the direction of propagation.
In the case of waveguides, the waves can be TE waves or TM waves, or they can have more
complicated types of polarizations. In optical fibers, because of their cylindrical symmetry, the
polarization of the fundamental mode propagating in the fiber can be randomly oriented in the
x-y plane.
Higher order modes have different patterns of variation over the waveguide cross section and in
general extend more outside the waveguide core. Figure 1-9 shows the pattern of variation for
the fundamental mode and two higher-order modes along one direction, passing through the
center of the waveguide core. The waves are TE polarized. Once again the fundamental mode
has a maximum in the center and decreases in intensity with distance from the center. The
evanescent field extends slightly outside the core. The first higher-order mode has a minimum in
the center and two intensity peaks inside the core. It extends a little more outside the core than
the fundamental mode. The second higher-order mode has three intensity peaks inside the core.
Figure 1-9 Intensity distribution for fundamental mode and two higher-order modes along a line
passing through the center of the waveguide
It is desirable to have most of the optical power carried by the wave contained in the core of the
waveguide. The fundamental mode satisfies this condition best. As mentioned previously, many
applications require single-mode waveguides, which only allow for the propagation of the
fundamental mode. Working with a core size close to the maximum allowed by the single-mode
condition also helps confine most of the light in the waveguide core.
Similar to the case of optical fibers, the core of a planar waveguide has a higher index of
refraction than the surrounding layers. For the waveguide in Figure 1-10, n1 > ns, and n1 > n0.
This allows the waveguide to confine and guide the light through total internal reflection the
same way it does in optical fibers.
Several names are used for the layers surrounding the core; sometimes these layers are called
lower cladding and upper cladding. The index of refraction of the upper and lower cladding is
the same in most cases. The relevant parameters that describe a planar waveguide are the
dimensions of the core and the indexes of refraction of the core and cladding materials. As in
the case of optical fibers, light propagates in planar waveguides in the form of modes, once
again based on the constructive interference condition. The modes are described by the integer
m, with values from m = 0 for the fundamental mode to m = 1, 2, . . . for higher-order modes.
There are several types of planar waveguides, each with its associated geometry. A planar
waveguide with a very thin core, having a much wider width than height, such as the one
illustrated in Figure 1-10, is called a slab waveguide. The only dimensional parameter that
describe such a waveguide is the thickness, labeled “2a” in the figure. Slab waveguides confine
light through total internal reflection only in a vertical direction, labeled “x” in the figure. The
electromagnetic fields describing light propagating in the waveguide do not change in the y
direction.
The V parameter used to describe single vs. multimode optical fibers can also be used in the
case of slab waveguides that have the same upper and lower cladding indexes, nclad. The V
parameter is defined by Equation 1-10, where “a” is only half the core thickness shown in
Figure 1-10.
The condition for single-mode slab waveguides is, however, different than for optical fibers. If
V is less than π/2, the slab waveguide is single mode.
Example 4
Determine whether a slab waveguide with thickness equal to 6 µm, upper and lower
cladding index of 1.445, and core index of 1.449, operating at a wavelength of 1310 nm, is
single mode. Calculate also the refractive index contrast between core and cladding.
Solution: Using Equation 1-10, we calculate the V parameter of the slab waveguide. (Once
again, we need to be careful to express the wavelength and the core thickness in the same
units to obtain the correct answer.)
Convert the wavelength to microns first: λ = 1310 nm = 1.31 µm
Obtain the half thickness: a = 6/2 µm = 3 µm
We are now ready to calculate the V parameter.
V = (2π/1.31 µm) (3 µm) (1.4492 – 1.4452)1/2 = 1.55.
Because V < π/2 = 1.57, the slab waveguide is single mode.
The refractive index contrast is calculated using Equation 1-5:
Δ = (ncore2 – nclad2)/(2 ncore2) = (1.4492 – 1.4452)/(2 x 1.4492) = 0.003 = 0.3%
Slab waveguides are a useful model to understand how light propagates in planar optical
waveguides and are relatively easy to solve mathematically.
The most common geometry for planar optical waveguides used in PIC is the rectangular or
channel waveguide. The core of these waveguides has comparable width and height, as shown
in Figure 1-11a, illustrating the waveguide cross section. To describe these waveguides, we
need both the width, W, and the height, H, in addition to the refractive indexes of the core and
cladding. Another possible geometry is rib or ridge waveguides, illustrated in Figure 1-11b.
Ridge waveguides are described by width W and height H, as well as thickness of the ridge
layer (r.H), in Figure 1-11.
Channel and ridge waveguides are also described by modes and can be single-mode or
multimode. The mathematical equations describing these waveguides are more complicated than
the slab waveguide equations and depend on all three spatial (x, y, and z) variables. The
equations can be solved either using various approximations or numerically, using specialty
software programs. It is not possible to determine whether the waveguide is single mode using
Equation 1-10 for these waveguides. However, the behavior is similar, and waveguides with a
large refractive index difference between core and cladding and/or a large core allow
propagation of more modes.
The materials composing the channel and ridge waveguides, together with the wavelength,
determine the indexes of refraction of the core and cladding. The thickness chosen for the core
layer is the same across the entire PIC. This comes from the fabrication process, which will be
explained later in the module. The only parameter that can and does vary throughout the PIC is
the waveguide width. With all other parameters fixed, the waveguide width must not exceed a
maximum value to keep the waveguide single mode, if required by the application. The
maximum width for SM waveguides can be determined using software programs.
Dimensions
Compared with the cores of optical fibers, the cores of planar waveguides have smaller
dimensions. For optical fibers, it is advantageous to have large cores. This helps when directing
light into the fiber and also makes it easier to handle the fibers. Single-mode fibers typically
have a core diameter of 9 µm and a cladding diameter of 125 µm. A coating with a diameter of
250 µm and an external buffer of diameter 900 µm protect the cladding. Multimode fibers have
core diameters of 50 µm or 62.5 µm and can have the same cladding, coating, and buffer as SM
fibers.
By contrast, in the case of PIC devices, small waveguides are beneficial for creating highly
dense circuits that occupy a small footprint. As we discussed previously, waveguides tend to be
multimode when the refractive index difference between core and cladding is large and/or the
dimensions of the core are large. If single-mode waveguides are required, the higher the
refractive index difference is, the smaller the dimensions of the core will be.
Channel waveguides fabricated from silica and doped silica, the same materials optical fibers
are made of, typically have square cores with sides of 6 µm. For silica and doped silica, the
refractive index contrast is small. It generally varies from 0.8% to 2% for planar waveguides.
For other combinations of materials, such as silica and silicon, the refractive index contrast is
much higher. Silicon has an index of refraction equal to 3.5, which is much higher than silica’s
index of 1.45. Because of the higher index difference, single-mode waveguides based on these
materials are much smaller. Typical silicon-on-insulator channel waveguides have dimensions
of 0.5 µm x 0.2 µm. These tiny waveguides hold the promise of a very high level of integration,
which will be needed as PIC devices become more complex.
Example 5
Calculate the refractive index contrast of silicon (index of refraction 3.5) and silica (index
of refraction 1.45). Compare this with the typical contrast of silica and doped silica,
typically 1%.
Solution: Using Equation 1-5 for the index contrast, we obtain:
Δ = (3.52 – 1.452)/(2 x 3.52) = 0.41 or 41%.
This is much higher than the contrast between silica and doped silica.
Bend Radius
Another advantage of strong light confinement in a small waveguide core is the ability to have
bends with small radii of curvature in the waveguide. Bends are necessary to route and connect
different portions of the waveguides. If the bend radius of curvature is too small, the light
striking the interface between core and cladding may no longer satisfy the condition of total
internal reflection, which would allow light to leak out of the cladding and create losses that
affect the performance of the waveguide.
This constraint also holds true for optical fibers, which cannot be bent with a radius of curvature
smaller than a known minimum. Figure 1-12 shows how some of the light traveling in a bend
waveguide can escape and be lost if the radius of curvature is too small. The red line depicts the
light-intensity distribution inside the waveguide, which extends a bit into the cladding, as we
have seen before. The angle of incidence at the interface between core and cladding is labeled
“θ”. In the straight waveguide, the angle is higher than the critical angle, which satisfies the
condition for total internal reflection and allows light to propagate inside the waveguide. In the
bend waveguide shown on the right, the geometry of the waveguide is such that after two
reflections, the angle of incidence becomes smaller than the critical angle. This violates the total
internal reflection condition and results in some of the light entering the cladding. The cladding
light then escapes the waveguide and is lost. This is also seen in the distribution of the field,
which extends more into the cladding and reaches out of the waveguide completely. To avoid
such losses, the radius of curvature cannot be less than a minimum value that is specific to each
type of waveguide. The larger this minimum bend radius is, the more space the bends will take
up, increasing the footprint of the device.
Figure 1-12 Light-intensity distribution for straight and bend waveguides. A small amount of light is
lost in the bend waveguide due to the violation of the total internal reflection condition.
For comparison, the minimum bend radius for silica and doped silica channel waveguides is 4–5
mm, while the corresponding radius for silicon-on-insulator channel waveguides is about 10
µm. This shows the advantage of using waveguides with a high refractive index contrast. The
disadvantage is the difficulty of coupling light into these waveguides.
Propagation Loss
Several other important characteristics of planar optical waveguides depend on the material used
to fabricate the waveguides and the technology of fabrication. One of these is the propagation
loss that light experiences as it travels through the waveguide, with typical units of dB/cm. For
long waveguides, it is important to have very little propagation loss, because the loss
accumulates with distance traveled. For example, optical fibers used for long-haul transmission
(>100 meters) have a loss of only 0.2 dB/km. Planar waveguides have not reached such a high
level of performance, due to a different fabrication process, but they are much shorter. An
important loss mechanism in planar optical waveguides is scattering loss, which can occur due
to volume scattering, surface scattering, and sidewall loss. Volume scattering is caused by
imperfections in the waveguide material, such as impurities, defects, and voids, among others.
Surface scattering takes place at the core-cladding interfaces and is primarily due to surface
roughness (the same as sidewall loss.) These losses depend on the process used to fabricate the
waveguides. In addition to scattering, absorption losses play an important role. These losses
depend on the wavelength and are minimized by operating at specific wavelengths. In
telecommunications, where light is transmitted for long distances in optical fibers, several
windows of operation have been defined that correspond to wavelengths for which glass
absorption is minimal. These are shown in Table 1.
Thermo-Optic Effect
Another material characteristic is the strength of the thermo-optic effect, expressed by the
variation of the index of refraction, n, with temperature, through the coefficient of variation
dn/dt in units of 1/°C or 1/K. The index at a temperature, T, can be calculated with the following
equation, where n(T0) is the index at a reference temperature:
Example 6
Silica’s index of refraction has a coefficient of variation with temperature equal to 1.0 x
10-5 /K. If the index of refraction at room temperature (21 ºC) is 1.45, what is the index of
refraction at a temperature of 75 ºC?
Solution: Using Equation 1-11, we obtain:
n(75) = n(21) + 1.0 x 10-5 x (75 – 21) = 1.45 + 1.0 x 10-5 x (75 – 21) = 1.4505.
The index of refraction of silica changes very little as the temperature goes up to 75 ºC from
the room temperature of 21 ºC. This is because the coefficient of variation dn/dt for silica
has a very low value.
The thermo-optic effect is used in devices, such as switches and modulators, where the behavior
of the device is controlled by changing the temperature of the waveguide. For these devices, a
large coefficient of variation dn/dt is advantageous. On the other hand, the same effect can
negatively affect the performance of certain devices. In resonant devices that need to be
precisely tuned to certain wavelengths, small changes in the refractive index with temperature
will result in the devices becoming detuned. Such devices require temperature stabilization to
operate properly.
compounds. These materials are also used to fabricate semiconductor optical amplifiers (SOA).
Optical signal processing devices used in telecommunications can be fabricated from several
material platforms, including silicon-on-insulator (SOI), silicon dioxide (glass), lithium niobate
(LiNbO3), and various polymers. Of these materials, the SOI platform has emerged in recent
years as the material of choice because it offers the potential to fabricate highly integrated and
highly performing PIC devices capable of satisfying the ever increasing bandwidth and capacity
requirements of data storage, data transmission, and cloud computing, to name just a few
applications. Silicon is also a material of choice for many photodetectors, devices that detect
light and convert it to electrical signals.
To date, no single platform or technology is capable of producing the entire array of photonic
devices needed in various applications and fields and integrating them in the same circuit. For
example, using silicon to build lasers is not feasible, because silicon is an indirect gap material.
This makes silicon-based lasers inefficient sources of light. One option to integrate an optical
source with processing devices based on silicon is to grow gallium arsenide lasers on a silicon
substrate. However, this is difficult, due to large lattice mismatch and large difference between
the thermal expansion coefficients of the two materials. (The lattice refers to the periodic
structure that characterizes crystalline materials.) However, the interest in having such lasers
available is huge, because they would allow for monolithic integration of optical sources,
processing, and detection devices in the same chip. Research groups around the world are
working on creating efficient silicon-based lasers, and there has been progress toward obtaining
such a device. Silicon also allows for integration between photonic and electronic integrated
circuits (the ultimate goal of the technology), with all the economic and performance advantages
that this integration would bring.
Table 2 summarizes the characteristics of waveguides and PICs based on the material platforms
and technologies described above. As you can see, each of the platforms has advantages and
disadvantages. The III-V semiconductors dominate laser, optical amplifier, and detector
applications. Silicon refers to waveguides having a silicon core. The most used platform for
such waveguides is silicon-on-insulator or SOI. In this platform, the silicon core sits on a silicon
dioxide layer, which is an insulator. The upper cladding can be either silicon dioxide or air.
“Silica on silicon” refers to devices fabricated from silica and doped silica, created on a silicon
wafer. This platform has been successfully used to fabricate many telecommunications devices,
such as filters, modulators, switches, and splitters. It allows for very reliable devices with very
low propagation loss. However, due to the low index contrast, the devices are relatively large
and so occupy a big footprint. Polymers have been used for some applications, but these
materials do not allow operation at high temperatures and are less reliable.
The amount of integration between devices on the same platform has not yet been very large.
Most devices are made separately and assembled together in modules using fiber connections.
As stated previously, silicon has emerged as holding the biggest promise for monolithic
integration of PIC and, ultimately, integration of PIC and EIC together for best cost and
performance.
Table 1-2. Comparison of material and waveguide characteristics for main material
platforms used for Photonic Integrated Circuits Wavelengths
Material Devices Refractive Propagation Thermo-optic Compatibility Reliability
index loss coefficient with CMOS
contrast technology
III-V Lasers, optical
Semicon- amplifiers,
Low Relatively high High No High
ductors modulators,
detectors
Silicon Filters,
modulators, High Relatively high High Yes High
switches
Silica on Filters,
silicon modulators,
Low Very low Low Yes High
switches,
splitters
Polymer Modulators,
Low Low High Yes Low
attenuators
Deposition
Deposition is the process through which layers of materials are added to a silicon wafer.
Initially, a substrate layer and a core layer with an index of refraction higher than the substrate
are formed on the wafer. The core layer is where the optical waveguide will be formed, while
the lower-index layer will serve as the lower cladding. The layers can be formed through a
variety of methods: (a) oxidation or thermal growth, by which a layer of silicon dioxide (silica)
forms on top of silicon, (b) implantation, (c) physical vapor deposition (PVD), (d) chemical
vapor deposition (CVD), and (e) spin-on. The thickness of the layers varies from several
hundred nanometers to a few microns. PVD is used for materials such as gold, aluminum, and
titanium nitride. CVD is the most common method of deposition and is used for layers of silicon
dioxide, silicon nitride, and tungsten. Spin-on is used to deposit photoresist and polymers.
Figure 1-14 Photolithography and etching of silicon waveguides with silica lower cladding
Second Deposition
A new deposition step is used to cover the patterned core layer with an upper cladding if
necessary. In some cases, the upper cladding for the waveguide can simply be air.
Passivation
Depositing a protective layer over the surface of the wafer (if necessary), a process called
passivation, comes next. Silicon nitride (Si3N4) is a material commonly used for this purpose.
Metallization
Some PICs make use of the thermo-optic effect to modulate the amount of light at the output.
All materials used to fabricate PICs have an index of refraction dependent on temperature, but
the index change with temperature is more pronounced in certain materials than in others. For
example, silica (silicon dioxide) has an index coefficient of variation with temperature of
10-5/degree, while silicon has a higher coefficient of about 10-4/degree. To take advantage of this
effect, thin film heaters can be formed above certain regions of the optical waveguides. When
electric current runs through the thin film heaters, the heat generated over the waveguides
changes their index of refraction. The heaters are formed after completing the steps described
above and require another round of deposition and photolithography using an additional
photomask that contains the heater patterns.
The process described above corresponds to the process used to fabricate electronic integrated
circuits in the semiconductor industry. There are other methods used to fabricate PICs, and these
methods are specific to certain materials. In the case of lithium niobate, the LiNbO3 crystal is
first grown using a method called the Czochralski technique. A waveguide can then be formed
in this material by diffusing certain ions into it. One possibility is to diffuse titanium ions,
resulting in an area with higher refractive index that will act as the core of the waveguide. The
most popular method, however, is the proton exchange method, in which lithium ions and
protons (hydrogen ions) exchange places in a superficial layer on the surface of the lithium-
Figure 1-15 Direct laser writing of waveguides inside a substrate using a femtosecond (fs) laser
Deposition Equipment
The most common process used for deposition of thin films is chemical vapor deposition
(CVD). In this process, reactant gases are introduced in a reaction chamber. The gas molecules
move to the wafer’s surface and adhere to it in a process called adsorption. A chemical reaction
takes place on the surface of the wafer in the presence of heat; this reaction forms the thin film
layer. The gaseous by-products are then removed from the wafer’s surface and vented out of the
reaction chamber.
Several parameters influence the CVD process. The pressure in the reaction chamber must be
less than the atmospheric pressure but greater than 1 millitorr (medium vacuum range) for low-
pressure chemical vapor deposition (LPCVD). This requires vacuum pumps to maintain the
desired pressure. Atmospheric pressure chemical vapor deposition (APCVD) systems do not
require vacuum pumps but need complex exhaust systems to contain the toxic gases in the
deposition zone and prevent their release into the environment. LPCVD is used for high-purity,
thin films, while APCVD results in a high deposition rate. The second important parameter is
the temperature in the reaction chamber. In hot-wall systems, the wafer and chamber are at
similar high temperatures, and deposition takes place not only on the heated wafer but also on
the chamber walls, which must often be cleaned. In cold-wall systems, the wafer is the hottest
element in the system, and deposition takes place primarily on it. Finally, energy must be
provided to the reactant gases in the chamber. This can be done through resistive heating,
infrared lamp heating, or RF induction heating. Plasma enhanced CVD (PECVD) is also used;
in this method, the reactant gases are ionized to increase their reactivity and reduce the need for
very high temperatures in the chamber.
CVD systems can process wafers in batch, one at a time (single-wafer), or continuous wafer
motion. Systems that can process 100 or more wafers at a time have very high throughput but
are becoming less practical as the size of the wafers increases. Single-wafer systems have much
better film thickness uniformity than batch systems. Continuous-motion systems process wafers
in a continuous mode and have good throughput with good uniformity. Examples of CVD
equipment include Novellus’s Continuous Motion LPVCD, Genus’s Cold Wall PECVD,
Applied Materials’ Radiantly Heated Barrel Reactor, and Pacific Western Systems’ Horizontal
Flow PECVD Reactor.
Photolithography Equipment
There are two important steps in the photolithography process; each requires specific
equipment.
First, a layer of photoresist must be applied to the wafer using a coater. The goal in this step
is to obtain a uniform layer of resist on the entire wafer. The resist is first dispensed in the
center of the wafer in a static or dynamic process and then distributed to the entire spinning
wafer. The wafer is mounted onto a vacuum chuck and spun at high velocities to uniformly
spread the resist on its surface. An exhaust system extracts and eliminates fumes from the
photoresist. Fully automated coaters allow for programmed recipes to control the selection
and dispensing of photoresist, spin speeds, and event duration times; they also allow the
wafers to move through the entire process without manual intervention. The operator control
panel handles all activities associated with recipe entry and selection, error reporting, and
process and equipment monitoring.
The second step in the photolithography process is exposure of the coated wafer to light
through a photomask to allow the transfer of the mask patterns to the wafer. Methods used
for this purpose include contact printing, proximity printing, projection printing, reduction
stepping, and reduction scanning. In contact printing, the mask is placed on top of the
photoresist, which allows for very good accuracy of transferring the pattern onto the wafer.
However, the integrity of the mask cannot be maintained, and the feature sizes and
resolutions are limited. Proximity printing keeps the mask at a small distance from the resist,
but, once again, small feature sizes are hard to reproduce. In projection printing, additional
optics are placed between the mask and wafer, bringing the feature size to less than one
micron using a projection aligner. The next step, to obtain very high-resolution images, is to
use optics with refractive lens elements, but these will limit the size of the exposure area. In
reduction systems, only a small area is exposed at a time; this area is called the reticle. The
exposure is then repeated over the entire surface of the wafer in a process of “step and
repeat.” This creates many, identical, high-resolution dies across the wafer. The technology
in use today is based on the reduction scanning method. In this method, the reticle has
features 4 to 10 times bigger than the end product, which is then reduced using a reduction
lens positioned between the reticle and the wafer. By continuously moving or scanning the
reticle vs. the wafer, higher throughput is achieved together with very high feature
resolution. Figure 1-16 shows a schematic representation of an exposure system.
The light source used in the exposure process is either a mercury lamp or an excimer laser. The
latter is now used for state-of-the-art feature sizes, which are now well below 0.25 micron.
Feature sizes of 45 nm have been achieved using excimer lasers operating at the deep UV
wavelengths of 248 nm, 193 nm, and 157 nm. The smaller the source wavelength is, the smaller
will be the minimum feature size that can be achieved.
Companies that manufacture photolithography systems include Canon, Nikon, ASML,
Ultratech, and others.
Etching Equipment
Reactive-ion etching (RIE) is the current method of choice for the dry etching process used in
the fabrication of PICs. The method is based on the use of chemically reactive plasma generated
under low pressure by an electromagnetic field to remove material deposited on wafers. The
RIE equipment consists of a cylindrical vacuum chamber with a wafer platter at the bottom.
Gases are introduced in the chamber at a pressure from a few millitorr to a few hundred
millitorr. Plasma is formed when a strong radio frequency (RF) electromagnetic field is applied
to the wafer platter. The electromagnetic field ionizes the gas molecules by stripping them of
electrons, thus creating the plasma. The electrons hitting the wafer platter build up a negative
charge on it, and this charge attracts the positive ions. The ions collide with the material to be
etched, and a chemical reaction takes place together with a physical etch, in which material is
removed when the ions transfer their kinetic energy to the wafer. The process is controlled
through the RF power, pressure, time, and gas selection. Figure 1-17 shows commercial RIE
equipment. Manufacturers of such systems include Hitachi, Canon, Lam Research, Oxford
Instruments, Applied Materials, and others.
SUMMARY
This module presented an introduction to photonic integrated circuits, their applications, and
their role as a key contributor to the sustained growth of fields such as communications and
computing. The module discussed planar optical waveguides, materials for PIC, and PIC
fabrication steps and equipment. Subsequent modules will present PIC devices based on various
materials and their configurations, functions, and applications. Systems of PIC devices and basic
design of PICs will also be presented.
REFERENCES
Bergman, K., L. P. Carloni, A. Biberman, J. Chan, and G. Hendry. 2014. Photonic Network-on-
Chip Design. New York: Springer-Verlag.
Deen, M. J., and P. K. Basu. 2012. Silicon Photonics, Fundamentals and Devices. Chichester,
UK: Wiley.
Lifante, G. 2003. Integrated Photonics: Fundamentals. Chichester, UK: Wiley.
Maricopa Advanced Technology Education Center. 2013. MATEC Module Library Tempe:
Maricopa Advanced Technology Education Center. http://matec.org/ps/library3/
The National Center for Optics and Photonics Education. 2013. Fundamentals of Light and
Lasers. 2nd ed. Waco, TX: OP-TEC.
Okamoto, K. 2006. Fundamentals of Optical Waveguides. Burlington, MA: Academic Press.
Chen, C.L. 2007. Foundations for Guided Wave Optics. Hoboken, NJ. John Wiley & Sons.
Silicon Photonic
Integrated Circuits
and Devices
Module 2
of
Integrated Photonics
© 2016 University of Central Florida
This material was created under Grant # 1303732 from the Advanced Technological Education
division of the National Science Foundation. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the author(s) and do not necessarily
reflect the views of the National Science Foundation.
ISBN 978-0-9903125-6-7
CONTENTS OF MODULE 2
Introduction ..................................................................................................................................... 1
Prerequisites .................................................................................................................................... 1
Objectives ........................................................................................................................................ 1
Scenario ........................................................................................................................................... 3
Basic Concepts ................................................................................................................................ 4
Material Properties of Silicon ...................................................................................................... 4
Active vs. Passive Devices .......................................................................................................... 6
Optical Fiber Communication Systems ....................................................................................... 8
Silicon Photonic Integrated Circuits .......................................................................................... 13
Passive Silicon PIC Devices ...................................................................................................... 14
Silicon-on-Insulator Optical Waveguide ............................................................................... 14
Bend Optical Waveguide ....................................................................................................... 17
Input/Output Coupling to Silicon PIC Devices ..................................................................... 18
Directional Coupler ................................................................................................................ 20
Y-Branch ................................................................................................................................ 22
Mach-Zehnder Interferometer ................................................................................................ 23
Ring Resonator....................................................................................................................... 26
Bragg Grating......................................................................................................................... 30
Active Silicon PIC Devices ....................................................................................................... 32
Lasers ..................................................................................................................................... 32
Modulators ............................................................................................................................. 34
Photodetectors ........................................................................................................................ 36
Fabrication of Silicon PICs........................................................................................................ 37
Testing.................................................................................................................................... 39
Assembly and Packaging ....................................................................................................... 41
Summary ....................................................................................................................................... 45
Problem Exercises and Questions ................................................................................................. 46
References ..................................................................................................................................... 48
I NTEGRATED P HOTONICS C OURSE FOR T ECHNICIANS
Module 2
Silicon Photonic Integrated
Circuits and Devices
INTRODUCTION
Photonic Integrated Circuit (PIC) devices are fabricated from a variety of materials, including
silicon, indium phosphide, silicon dioxide, lithium niobate, polymers, and others. Out of all of
these, silicon has emerged as the material of choice for integrating multiple components in a
single chip. Monolithic integration describes integrating light sources, signal processing devices,
and photodetectors in a single chip. Monolithic integration reduces the footprint, the bill of
materials, and the cost, and it also improves the system’s performance. Moreover, PIC devices
based on silicon can be integrated with electronic integrated circuit (EIC) devices in the same
chip, further enhancing the functionality of the products. Silicon PICs are fabricated following
the same steps used to fabricate EICs by CMOS technology, thus leveraging the advanced
technology and equipment of the microelectronics industry.
While silicon is the only material that holds the promise of ultimate integration of both photonic
and electronic devices, it is currently not possible to fabricate all photonic devices in silicon with
the required level of performance. Today, devices are fabricated separately using the optimal
material for performance. They are then assembled together in modules and connected using
optical fibers. The assembly operations required to build the modules are complex and
expensive, and the modules take up valuable space. This module will describe silicon PIC
devices, their current characteristics, and directions leading to future enhanced integration and
performance.
PREREQUISITES
OP-TEC’s Fundamentals of Light and Lasers Course 1
OP-TEC’s Integrated Photonics: Module 1
OBJECTIVES
Upon completion of this module, the student should be able to:
Explain the material properties of silicon that are important for silicon PICs
1
Classify passive and active photonic devices and calculate insertion loss in dB
Describe the components of a basic optical fiber transmission system
Explain attenuation and dispersion in optical fibers and how to mitigate their effects
Describe characteristics of the following passive silicon PIC devices:
o SOI waveguide
o Bend waveguide
o Coupling between optical fiber and SOI waveguide
o Directional coupler
o Y-branch
o Mach-Zehnder Interferometer
o Ring resonator
o Bragg grating
Describe characteristics of active PIC devices, including laser, modulator, and
photodetector
Describe challenges of integrating active devices in a silicon platform
Discuss back-end fabrication processes such as testing, assembly, and packaging
Describe system-in-package technologies and compare wire bond and flip-chip
approaches
Explain advantages and implementation of silicon optical interconnects
The absorption properties of a material are important for the application of photodetectors.
Photodetectors are devices that convert a light input into an electrical output. They are essential
elements in any optical power measurement system. They are also used in optical fiber
communication systems as part of the receiver modules. Important parameters that characterize
photodetectors are responsivity, sensitivity to wavelength, temporal response, and noise level.
Silicon PIN photodiodes are fabricated from a p-n junction with an intrinsic, i, layer inserted
between the p- and n- regions; these photodiodes have excellent characteristics.
For wavelengths higher than 1.1 μm, silicon becomes transparent, which is a useful property for
PIC devices that transmit rather than absorb light. Numerous such devices are used in optical
fiber telecommunication systems, which operate in the wavelength range between 1200 nm and
1700 nm, as described in Module 1. The most used transmission band is the C (Conventional)
band, centered at 1550 nm. For transmission based devices, the index of refraction is an
important parameter. Figure 2-2 shows the index of refraction of silicon for the wavelength range
1100–1700 nm. Compared with silicon dioxide (glass), which is the material used to fabricate
optical fibers, silicon has a much higher index of refraction. The index of silicon is about 3.5, as
the figure below shows, while silicon dioxide has an index about 1.45. This large index
difference allows for compact devices and a high degree of integration, making silicon an
attractive PIC platform.
Figure 2-2 also shows that the index of refraction for silicon decreases with increasing
wavelength. The phenomenon of index variation with wavelength is called dispersion. For
wavelengths between 1100 nm and 1700 nm, the silicon index’s dependence on λ can be
approximated by a straight line with a decreasing slope equal to -1 x 10-4 nm-1.
Silicon’s index of refraction also varies with temperature, which is beneficial in certain devices
but undesirable in others. Devices such as modulators exploit the index’s change with
temperature to effect changes in the device’s output of optical power. On the other hand, filters
and other resonant devices are based on a constant index of refraction and would become
detuned if the index were to change as the ambient temperature changed. These latter devices
require temperature stabilization to operate properly.
The change in index of refraction with temperature is expressed by the coefficient of variation
with temperature, dn/dt. It is equal to the change in index corresponding to a one-degree change
in temperature. Here the temperature change is expressed in units of Kelvin (absolute
temperature). In the case of silicon, the coefficient of variation = dn/dt = 1.87 x 10-4 K-1. As with
the index of refraction, this parameter is higher for silicon than for silicon dioxide, whose
coefficient is about 1.1 x 10-5 K-1.
A characteristic parameter for lasers is their efficiency, which equals the ratio of the total optical
output power to the pump power. This parameter describes how efficient these devices are in
converting the energy of a pump source into output light energy. In the case of semiconductor
lasers, the efficiency is the ratio of the output power to the electrical power used as a pump.
Semiconductor-based lasers fabricated from III-V combinations of materials have a high
efficiency of 60% and above.
η = Pout/Ppump (2-2)
Passive devices perform operations on an incoming optical signal. Examples of passive devices
include splitters, couplers, and filters. They are not intended to increase the optical power level
and so do not require an energy source to operate. Ideally, the optical power at the output of a
passive device should equal the input power. In reality, a small amount of power will be lost as
the device operates on the light signal. The parameter that characterizes this is the insertion loss
(IL), defined as the ratio of the total power at the output of the device to the input power.
IL = 10 x log10(Pout/Pin) (2-3)
The unit used for IL is the decibel, dB. The advantage of using dB as unit is that the total loss in
the optical signal after it travels through a series of devices can be obtained simply by adding the
ILs of all the devices.
Example 1
Calculating the insertion loss in dB. A light beam with 1 mW of optical power is incident on
a passive PIC device. A power meter indicates that the output optical power is 0.95 mW.
Find the insertion loss of the PIC device in dB. P P
in out
Solution: IL = 10 x log10(Pout/Pin) = 10 x log10(0.95/1.0) = -0.22 dB PIC
The negative sign in the IL above signifies that the output power is less than the input power.
This correctly describes a loss of power. If Equation 2-3 were used to express the amplifier gain
of an optical amplifier in dB, it would result in a positive value. Manufacturers of passive
devices most often provide the IL values without the negative sign. Even so, remember that IL
Example 2
Insertion loss of an optical power coupler/splitter. An optical power coupler/splitter is a
passive device that divides an input light beam into two output beams. Assume that the input
light beam has a power of 1 mW and the two output beams have powers of 0.43 mW each.
Calculate the insertion loss of the coupler.
Solution: The total output power is given by the sum of the powers of each of the two
outputs.
Pout = 0.43 mW + 0.43 mW = 0.86 mW.
IL = 10 x log10(Pout/Pin) = 10 x log10(0.86/1.0) = -0.66 dB
As previously mentioned, total loss from cascaded devices can easily be obtained by summing
the losses from each device in the series.
Example 3
Calculate the total loss of an optical signal after it has propagated through two cascaded PIC
devices with IL of -0.4 dB and -0.55 dB respectively.
Solution: IL1 = -0.4 dB, IL2 = -0.55 dB.
Total loss IL = IL1 + IL2 = -0.4 + (-0.55) = -0.95 dB.
Important applications of PIC devices at this time are in optical fiber communication systems
and fiber optics networks. The following section summarizes the main characteristics of these
systems and provides a background about the role that PIC devices play in them.
Optical fiber transmission systems can be classified as long haul, regional, or metropolitan,
depending on the distance covered. Long haul systems extend over thousands of miles to connect
states, countries, and continents. Regional and metropolitan systems cover shorter distances and
do not require the very high level of performance that long haul systems do. Figure 2-4 illustrates
the different types of systems and their interconnections.
Optical fiber transmission systems, and long haul systems in particular, need to address two
issues associated with optical fibers: attenuation and dispersion. Attenuation is the gradual loss of
optical power due to absorption and scattering of the light in the fiber material. Attenuation can
be described by an attenuation coefficient expressed in units of inverse centimeters, similar to the
absorption coefficient. (For example, from Figure 2-1 we can find that the absorption coefficient
of crystalline silicon at a wavelength of 800 nm is about 103 cm-1.) Alternately, we can look at
the loss of power per unit length of optical fiber expressed in units of dB/km. Figure 2-5 shows
fiber loss in relation to wavelength. The figure indicates the attenuation maxima due to light
absorption by water molecules, as well as the regions with minimum attenuation. As illustrated,
the minimum loss occurs at the wavelengths of 1310 nm and 1550 nm (1.31 µm and 1.55 µm).
For this reason, the wavelengths around these values have been chosen as transmission bands for
the light signals in fiber optic communication systems.
Single-mode fibers used in long haul transmission have a very small loss coefficient, lower than
0.2 dB/km. Due to the very large distances covered, even this small coefficient of loss results in a
reduction of the optical power to levels that detectors cannot sense.
Example 4
Light is transmitted through a 500 km-long fiber with a loss coefficient of 0.2 dB/km. Find
the loss of optical power at the end of the fiber.
Solution: Total loss of power = 0.2 dB/km x 500 km = 100 dB
In addition to the power loss caused by attenuation, light pulses are spread out due to dispersion
as previously discussed. One solution to both problems is to regenerate the signal every hundred
kilometers or so. This is achieved by converting the optical signal to an electrical signal,
recreating it according to the original using electronic circuits, converting it back into an optical
Figure 2-7 An optical communication system that regenerates the light signal along the transmission
path
Another solution—one based on optical amplifiers—has been introduced in an effort to bring the
signal power back up. Using optical amplifiers is advantageous because they operate directly on
the optical signal, thus avoiding the expensive conversions between electrical and optical
domains. Figure 2-8 shows such a system. The optical receiver here includes a transimpedance
amplifier (TIA), which amplifies the current created by the photodiode.
Figure 2-8 An optical communication system that amplifies the light signal along the transmission path
Figure 2-10 a) SOI material structure Figure 2-10 b) Channel (or strip)
SOI waveguide
SOI waveguides have a higher propagation loss than waveguides realized in other material
platforms. One component of the loss, due to sidewall roughness, is 2-3 dB/cm. However, SOI
waveguides are more compact, and the short length of the devices compensates for some of the
higher loss per unit of length.
Example 5
An SOI waveguide 500 μm long has a propagation loss of 2.5 dB/cm. A silicon dioxide
(silica) waveguide has a 0.1 dB/cm propagation loss and a length of 2 mm. Find the loss at
the end of each waveguide.
Solution: Total loss of power for SOI waveguide = 2.5 dB/cm x 0.05 cm = 0.125 dB
Total loss of power for silica waveguide = 0.1 dB/cm x 0.2 cm = 0.02 dB
Example 6
Finding the index of refraction of silicon and silica at a temperature of 75 °C. Assume that
the indices of refraction at room temperature (21°C) are the ones provided above. Calculate
the indices of refraction at 75 °C.
Solution:
For silicon n(T2) = n(T1) + dn/dt *(T2 – T1) = 3.473 + 1.87 x 10-4 *(75 – 21) = 3.483
For silica n(T2) = n(T1) + dn/dt *(T2 – T1) = 1.444 + 1.1 x 10-5 *(75 – 21) = 1.445
Using the solution from Example 6, we can see that with this temperature change, the change in
index of refraction for silicon is 0.01, while the change for silica is only 0.001. Because silica’s
index of refraction changed so little compared with silicon’s, when we design devices based on
the thermo-optic effect, we can ignore the change in the oxide index with temperature and take
into account only the change in index of silicon.
In the case of a wave confined inside a waveguide with effective index of refraction neff, the
phase is again calculated using the formula above, in this case with the effective index replacing
the general index n.
Φ = (2π/λ) neff L (2-6)
When determining the effective index of a waveguide, there will generally be two solutions, even
in the case of SM waveguides. The two solutions will correspond to two possible polarizations
for light traveling inside the waveguide, TE and TM. Recall that TE and TM stand for transverse
electric and transverse magnetic and reflect the light’s property as a transverse wave. In the case
of TE, the electric field vector is perpendicular to the direction of propagation of the wave, while
in the TM case, the magnetic field vector is perpendicular to the direction of propagation of the
wave.
Because the two effective indices for TE and TM polarizations are different, the device can
experience some performance degradations due to the different speeds of travel of TE and TM
components inside the waveguides. The difference between the two effective indices is called
waveguide birefringence, Δneff.
Δneff = neff TM - neff TE (2-7)
SOI waveguides have high birefringence, a consequence of the high index contrast between core
and cladding. Various methods can be applied to counteract the negative effects of waveguide
birefringence: operating with polarized light (either TE or TM), using a special waveguide
design to bring the neff TE and neff TM close to each other, inserting components that compensate
for the effects of birefringence, or separating the two polarizations and balancing their pathways
through the waveguides (polarization diversity). Each of these methods has specific advantages
and disadvantages that we discuss in a later module.
Figure 2-11 Directional coupler device containing four S-bend waveguides to bring waveguides close
to each other and then separate them
The parameter that characterizes bend waveguides is the radius of curvature, R. As discussed in
Module 1, if the radius of curvature is too small, light leaks out of the waveguide as the total
internal reflection condition is violated. This results in loss of optical power through radiation.
To keep the bend losses negligible, the radius of curvature must be kept above a minimum value,
which introduces a trade-off between device loss and footprint. The designer of the device will
use the minimum value for bend radius while creating a layout that minimizes the device’s
footprint.
Due to the mode mismatch, an optical power loss in excess of 10 dB takes place when the SM
fiber is directly attached to the SOI waveguide (a method called butt coupling). To avoid this
large loss, alternate coupling methods have been found. They are briefly described below and
illustrated in Figure 2-13.
Grating coupling
A grating is a periodic structure that can diffract light and change its direction of propagation. An
SOI waveguide grating coupler can be created by introducing a periodic corrugation at the top of
the SOI layer. Light from the optical fiber is incident at an oblique angle onto the SOI
waveguide. When the phase-matching condition (constructive interference) is met, the SOI
waveguide captures a large fraction of the light carried by the optical fiber. The period of the
grating (the distance between two successive corrugations) is the most important parameter that
determines the coupling efficiency from fiber to waveguide. Efficiencies of 50–60% have been
achieved with grating periods between 600 nm and 700 nm. The primary limitation of this
method is that for a given grating period only a relatively narrow band of input wavelengths can
be efficiently coupled.
R = [(n1-n2)/(n1+n2)]2 (2-8)
For example, the percentage of optical power reflected and thus lost at the interface between
glass and air at normal incidence is about 4%. In many cases, SOI waveguides use antireflection
coatings to reduce losses due to reflection, since the index difference between optical fiber and
silicon is close to 2.
Directional Coupler
A directional coupler is another fundamental building block of PIC devices. It is used both as a
stand-alone device and as part of other components. The function of a directional coupler is to
split or combine light. It consists of two parallel waveguides that are initially well separated, then
brought close together in the coupling region, and then separated again, as shown in Figure 2-14.
In the coupling region, due to the close proximity, the waveguides interact with one another, in
the sense that light starts to transfer from one waveguide to the other. The distribution of light in
the fundamental mode of an SOI waveguide was illustrated on the right side of Figure 2-12,
together with the waveguide core edges in white. As that figure shows, the electromagnetic field
extends outside the waveguide core and decreases in intensity as it moves away from the core.
The field outside the core is called the evanescent field. When two parallel waveguides are
brought close together, the evanescent field of the waveguide carrying the light extends over the
second waveguide, and this is enough to excite or initiate a guided mode in the second
waveguide.
If the coupling region is long enough, 100% of the light will transfer to waveguide 2, and then it
will gradually transfer back to waveguide 1. The directional coupler can be designed to achieve
any split of the optical power between the two waveguides by adjusting the length of the
coupling region or the gap between the waveguides. A common use of the directional coupler is
to split the light equally between the two waveguides, in which case the device is called a 3dB
directional coupler. Figure 2-14 shows the main parameters that determine the power split
between two waveguides: the gap between the waveguides in the coupling region, labeled S, and
the length of the coupling region, L. With SOI waveguides, the gap can be as small as 200 nm.
The coupling length, denoted by Lc and defined as the length at which 100% of the light transfers
to the second waveguide, is around 40 µm.
Directional couplers can be described by a coupling coefficient, K. The optical power in each of
the two output waveguides is given by the following set of equations, where P0 is the initial
power carried by waveguide 1 and L is the length of the coupling region:
To find the coupling length, Lc, that is, the length at which 100% of the light transfers to the
second waveguide, we need to set P1(Lc) = 0 and P2(Lc) = P0. Using Equations 2-9a and b, we
find: cos2(K Lc) = 0, and sin2(K Lc) = 1. Solving for Lc, we find:
Lc = π/(2K) (2-10)
Similarly, for a 3dB coupler, we want an equal split of the power between waveguides 1 and 2.
That means P1(L) = P2(L) = P0/2. Using Equations 2-9a and b, we find: cos2(KL) = sin2(KL), or
tan2(KL) = 1. Solving for L, we find:
Directional couplers can also be used as combiners by reversing the direction that light travels.
For example, in Figure 2-14, if light enters the device at P1 and P2 from the waveguides on the
right and is combined into one output, P0, in the lower waveguide on the left, then the two waves
are coherent.
In general, directional couplers are very sensitive to fabrication process variations. One of the
challenges of the fabrication process is to maintain a high level of uniformity across the entire
silicon wafer. Variations in waveguide width happen, and these also affect the gap between
Y-Branch
A Y-branch is a device that performs the same function as a directional coupler: power splitting
or combining. The device consists of a waveguide that branches into two identical waveguides,
as shown at the top of Figure 2-15. When used as a splitter, the Y-branch divides the power
evenly between the two outputs. In this sense, it is a simpler device than a directional coupler. As
is the case with all devices, the Y-branch has a nonzero insertion loss (IL), meaning that the sum
of the two output powers is slightly less than the input power. Design optimized Y-branches have
IL < 0.5 dB. When used as a combiner, if the waves traveling in the two waveguides are in
phase, the optical powers combine to double the power in the output waveguide (minus the IL).
However, if the waves are out of phase, destructive interference will occur and result in zero
power at the output.
Figure 2-15 Top: Single Y-branch splitting the incident power into equal powers in the two output
waveguides. Bottom: Cascaded Y-branches.
Y-branches can be used by themselves or as part of more complex devices. They can be
cascaded to obtain various power fractions in the output waveguides. The bottom image in
Figure 2-15 illustrates a device with five outputs with decreasing amounts of power in
Mach-Zehnder Interferometer
Interferometers are devices based on the phenomenon of wave interference. Interferometers are
configured in several ways and are used in a large number of optics and photonics applications.
The Mach-Zehnder interferometer (MZI) uses one source of light split into two waves that travel
through different pathways before being brought back together and allowed to interfere.
Depending on the accumulated path difference between the two waves, the resulting optical
power can vary from zero (destructive interference) to a maximum power (constructive
interference).
In PIC devices, MZIs are constructed from two power splitting/combining elements connected
by two interferometer arms, as shown in Figure 2-17. For power splitting and combining, either
3dB directional couplers or Y-branches can be used.
The two interferometer arms have different optical path lengths; in the figure the top arm is the
reference arm. Consequently, the waves traveling in the two arms accumulate a phase difference
of Φ. The two waves interfere in the combiner and, depending on how large Φ is, produce
various power outputs in the output waveguides A and B.
PIC MZI devices are much more useful as active devices, which allow the amount of power in
waveguides A and B to be controlled by changing the phase Φ. One way to achieve this is by
changing the effective index (neff) of the bottom waveguide. It is possible to use the thermo-optic
effect by depositing a thin film heater directly above the bottom waveguide. The thin film heater
is a resistor that radiates heat when electric current runs through it. In this case, even though the
geometrical length can be kept the same, the effective indices of the top and bottom waveguides
will not be the same due to the increased heat experienced by the bottom waveguide (recall
Equation 2-4 and the variation of index with temperature). The phase difference is now obtained
using this equation:
By varying the electric current through the thin film heater, it is possible to change the effective
index neff bottom and thus change the phase difference Φ, while neff top stays constant.
The dependence of the output powers A and B on the phase difference is given by
Equations 2-14a and b. P0 denotes the input power. To simplify the calculations, the two
equations assume zero insertion loss.
PA = P0 sin2(Φ/2) (2-14 a)
PB = P0 cos2(Φ/2) (2-14 b)
Figure 2-18 shows a graph of the two output powers as a function of Φ. Part a shows the ratio of
output powers to input power, while part b shows these ratios expressed in dB, the customary
unit for PIC devices’ power output.
From the two equations and the graph, we can see that for a phase difference equal to zero, the
input power emerges completely from output B (the lower arm), and power in output A is zero.
The situation changes completely when the phase difference equals π, or 3.14. In this case,
output power B becomes zero, and the entire input power exits from output A. This behavior is
exploited in devices called switches; these devices are capable of switching light between outputs
A and B under user control.
To switch the output from B to A, the phase difference Φ needs to be changed from 0 to π.
Assuming a typical length of 50 µm for the arms and a wavelength of 1.55 µm, we can use
formula 2-13 to get the necessary index change between the bottom and top waveguides of a
switch device, as well as the temperature change necessary to create this index change.
Example 7
Find the effective index change between the two interferometer arms equivalent to a phase
difference equal to π.
bottom
Solution: Use formula 2-13 to solve for (neff - neff top) as follows:
(neff bottom - neff top) = (Φ λ)/(2 π L)
(neff bottom - neff top) = (π x 1.55µm)/(2 x π x 50µm) = 0.0155
Recall that the silicon index of refraction varies with temperature by dn/dt = 1.87 x 10-4 K-1.
Rewriting Equation 2-4:
dn/dt = (neffbottom – nefftop)/(T2 – T1)
(1.87 x 10-4 K-1) = 0.0155/(T2 – T1)
Temperature change = (T2 – T1) = 0.0155/(1.87 x 10-4 K-1) = 83 K or 83 ºC
Another important application of PIC MZI devices is in modulators. In the case of modulators,
we are interested in controlling an output optical power through the entire range of values
between minimum and maximum. This range of values can be described as an analog output,
while a switch device output can be thought of as a digital value that returns either a 0 or a 1. For
example, we can choose output B of the previous MZI. Figure 2-18 a) shows that any value for
PB/P0 can be obtained by choosing the corresponding value of the phase Φ between 0 and π. This
can be achieved by controlling the current running through the thin film heater. An MZI
modulator can use Y-branches in the place of 3dB directional couplers. In this case, there is only
one output, which is all that is needed for a modulator. A variable optical attenuator (VOA) is
another application of MZIs. These devices are used to control the overall power levels in PICs.
An entirely different class of applications for PIC MZI devices exists outside fiber optics
transmission systems. These are sensor applications where the MZI is used to detect certain
substances and their properties. The principle of operation of such sensors is illustrated in the
figure below, which shows a Y-branch-based MZI. In this configuration, the reference arm is on
the bottom and the arm under test is at the top. A trench is created in the top waveguide, where
the biochemical substance under test will be placed. The substance has a different refractive
index than the waveguide, which will create a phase difference between the two arms. By
working backward from the detected output power to the phase difference that produced it, it is
possible to determine the substance under test.
Ring Resonator
A ring resonator is a filter type of device that performs operations based on the wavelength of the
input light wave. Resonators, in general, oscillate with high amplitude at certain frequencies
λ = v/f (2-15)
In the simplest case, a PIC ring resonator is formed by bringing a straight waveguide, sometimes
called the bus, and a ring waveguide close together, as shown in Figure 2-20.
An input light wave traveling from the left in the straight waveguide will transfer to the ring
through the evanescent field of the bus waveguide. Recall that light behaves the same way in
directional couplers that bring two parallel waveguides close to each other in the coupling
region. After completing a round trip around the ring, the light wave will interfere with other
waves at the same location. Depending on the phase difference accumulated around the ring, the
interference can be constructive or destructive. This phenomenon is analogous to light moving
back and forth in a laser cavity. The ring has a different geometrical structure, but the basic
behavior is similar. The phase difference accumulated in one round trip around a ring of radius r
is given by:
The resonance condition takes place when the phase difference equals an integer multiple of 2π.
The resonant wavelengths are almost equally spaced at an interval called the free spectral range
or FSR, given by the equation below:
Figure 2-21 Notch filter transmission vs. wavelength. Resonant wavelengths appear at approximately
1531, 1541, and 1551 nm, with an FSR of about 10 nm.
Another type of filter is an add-drop filter, which can be constructed from a ring resonator
placed between two straight waveguides. A device like this has four ports, labeled Input,
Through, Add, and Drop, as shown in Figure 2-22. As in the case of the all-pass ring resonator
filter described above, light at a resonant wavelength will couple into the ring. This time,
however, these wavelengths will exit through the Drop port at the top left in the figure. By
symmetry, light at a resonant wavelength sent in through the Add port will be coupled into the
ring and exit at the Through port. A device of this type is used in nodes of optical fiber networks
where some signals need to be routed to certain locations and other signals need to be introduced
in the optical fiber.
Figure 2-23 shows the transmissions in the Through and Drop ports of the add-drop filter, with
transmission in units of dB.
Figure 2-23 Add-drop filter transmission for the Through and Drop ports vs. wavelength. Resonant
wavelengths once again appear at approximately 1531, 1541, and 1551 nm.
An alternate configuration for the add-drop ring resonator filter is the racetrack configuration.
Here the ring is divided in two 180 degree sections, and straight waveguides are inserted in
between. The figure below shows a microscope image of such a configuration. The racetrack
configuration is easier to realize in practice because it allows for a larger gap between the two
bus waveguides and the ring. Because the larger gap reduces coupling between the bus and ring
waveguides, the straight portions of waveguide have been added to increase the length of the
coupling region.
Silicon-on-insulator (SOI) waveguides are very well suited for creating ring resonator filters. The
strong confinement of the electromagnetic field in the waveguides allows for small bend radii, as
low as 5 µm. This makes the entire ring resonator device very compact. A small ring radius will
also result in a large FSR (as you can tell from Equation 2-19, where the radius appears in the
denominator), and a large FSR is a desirable filter characteristic.
Figures 2-21 and 2-23 correspond to ring resonators of radius 10 µm created from SOI rib
waveguides with widths of 500 nm, SOI thicknesses of 220 nm, and rib thicknesses of 90 nm.
The SOI waveguides are covered with oxide. Silicon ring resonators can also be used as
biosensors, in which case the upper cladding material is air. The substance under test can be
placed directly on top of the ring silicon waveguide. The unknown substance will change the
effective refractive index of the waveguide, and this will change the resonant wavelengths of the
ring. The shift in one of the resonant wavelengths is measured and used to determine the index of
refraction of the substance under test, allowing for its identification.
Bragg Grating
A waveguide Bragg grating is a structure consisting of a periodic variation of the effective
refractive index along the waveguide in the direction of propagation of the light. The structure
can be created several different ways, depending on the type of waveguide. In an optical fiber,
the structure is called fiber Bragg grating (FBG) and is created by periodically modulating the
index of refraction inside the core. This is done by “writing” the grating using an intense
ultraviolet (UV) source of light. Because the germanium-doped fiber core is photosensitive,
exposing the core to UV light through a mask induces a permanent index change in the exposed
regions. In the case of planar waveguides, the fabrication process allows for changing the
physical dimensions of the core to create the grating. Either the height or width of the core can
be changed, as in cases a and b in Figure 2-25. The period of the grating is denoted by Λ.
Bragg gratings are resonant structures with a filter response. Recall that when light is incident on
an interface between two media with different indices of refraction, some of the light is reflected
back to the first medium, and the rest is transmitted in the second medium. In a Bragg grating,
successive reflections take place at each location where the index changes. The reflected waves
interfere with one another. When the wavelength of the light is such that the interference
between all reflected waves is constructive, the reflected wave builds up. This happens in a
wavelength region centered on the so-called Bragg wavelength, λB.
A typical Bragg grating totally reflects light at wavelengths within a bandwidth centered around
the Bragg wavelength. Light outside this bandwidth is transmitted instead of reflected. The
Bragg wavelength is given by the following equation, where Λ is the period of the grating and
neff is the effective refractive index of the waveguide:
λB = 2Λ neff (2-20)
In SOI waveguides, typical Bragg gratings have 200–300 periods, and the period is around 300
nm. The change in waveguide width for a sidewall grating can be tens of nm up to 200 nm.
Example 8
Find the grating period corresponding to a desired reflected wavelength. We want to design
a Bragg grating to reflect light around the wavelength of 1535 nm. The grating is created in
an SOI waveguide with an effective refractive index equal to 2.35. Find the grating period.
Solution: Use Equation 2-20 to solve for Λ.
Λ = λB/(2 neff) = 1535 nm/(2 x 2.35) = 327 nm = 0.327 µm
The example illustrates why SOI waveguides are very appropriate for Bragg gratings. The
grating periods corresponding to wavelengths in the communications windows, 1.31 µm and
1.55 µm, have values around 300 nm, a dimension well less than a micron. It is difficult to
obtain such small features in some other platforms.
Figure 2-26 Optical power reflected from a Bragg grating vs. wavelength
Waveguide Bragg gratings are used as integrated mirrors at the ends of the active medium in
diode lasers. They are also used to fabricate narrow-band optical filters and evanescent field
sensors. In more complicated configurations, they can also be used to construct add-drop
multiplexer devices.
Lasers
As previously mentioned, a big barrier to achieving monolithic integration on a silicon platform
is the lack of a silicon diode laser. Silicon is an indirect band gap material. In direct band gap
materials, an electron can transition between the conduction band and the valence band, emitting
a photon in the process. By contrast, in an indirect band gap material, an electron must pass
through an intermediate state and emit a particle called phonon before a photon can be emitted.
This makes the emission process inefficient for these materials, which include silicon and
germanium.
Commercial diode lasers are instead fabricated from direct band gap materials, using
combinations of elements from groups III and V in the periodic table, such as gallium (Ga),
aluminum (Al), arsenic (As), indium (In), and phosphorus (P). These diode lasers have high
efficiencies and are capable of high power outputs. GaAs laser diodes emit light with a
wavelength of 870 nm. Communications applications use InGaAsP laser diodes emitting in the
range 1150-1650 nm.
Diode lasers such as these are used as the light sources in optical fiber transmission systems. The
lasers are used as external components and optical fibers typically connect them to the PIC
devices. First, light emitted by the diode laser has to be coupled into the optical fiber. The modes
of the laser and optical fiber are not well matched because they are made of different materials
Modulators
Modulators are devices capable of adjusting the power level of an optical signal to any value
between a minimum value and a maximum value. A controlling signal, usually an electric
current, is used to achieve the desired output optical power. The physical effects modulators are
based on are electroabsorption and electrorefraction. In electroabsorption, the controlling signal
produces a change in the absorption coefficient, while in electrorefraction, the controlling signal
changes the material’s index of refraction. Electrorefraction can be achieved by either thermo-
optic or electro-optic effects. As the name indicates, a thermo-optic effect is a change in the
refractive index when heat is applied to the material. The strength of the thermo-optic effect in
silicon is described by the index coefficient of variation with temperature,
dn/dt = 1.87 x 10-4 K-1. The electro-optic effect is a change in the refractive index when an
electric field is applied. The thermo-optic effect is typically slower than the electro-optic effect,
because heat transfer through the material is slower than applying an electric field.
In the case of silicon, changes in both the refractive index and the absorption coefficient can be
achieved by changing the concentration of charge carriers in the material. This effect is called the
plasma dispersion effect. Carriers can be either injected into the device or removed from it, using
either a PIN junction or a p-n junction.
Modulators based on SOI devices can be created in a Mach-Zehnder interferometer (MZI)
configuration or a ring resonator configuration. In the case of MZI devices, heat can be applied
by thin film heaters deposited over one of the interferometer arms to achieve the thermo-optic
effect. We can pick either one of the two outputs to give the desired level of output power. For
example, referring to Figure 2-18, by controlling the phase difference between the values of 0
and π, any level of output power can be achieved in output B, from maximum transmission equal
to 1 to minimum transmission equal to 0. In practice the device will have some insertion loss,
which means that the maximum power in output B will be slightly less than the input power.
Also the minimum transmission in B cannot be exactly zero, but it can be a very small value. The
plasma dispersion effect can also be used with an MZI configuration. In this case, a PIN junction
is inserted in one of the MZI arms to allow for charge carrier injection through the application of
voltage.
A ring resonator, in either the all-pass or add-drop configuration, can also be used to create a
modulator. Referring to Figure 2-21, by working in a wavelength range near a resonant
Figure 2-27 An SOI micro-ring modulator based on the plasma dispersion effect
Micro-ring modulators are much more compact than MZI devices. The latter need long arms to
accumulate the required phase difference, while ring resonators take advantage of the light
traveling many times around the same path in the ring. MZI devices are, however, easier to
fabricate by not requiring the very tight tolerances ring resonators do.
All modulators are characterized by several important parameters including modulation depth,
speed of response or bandwidth, and power consumption. Modulation depth is defined as
follows:
Here Pout max is the maximum output power and Pout min is the minimum output power of the
modulator. Ideally MD = 1, which requires Pout min = 0. In practice MD values between 70-90%
are common.
With SOI modulators, response times are tens to hundreds of picoseconds (1 ps = 10-12 s) and
bandwidth values are on the order of GHz. The power consumption is on the order of mW for
silicon modulators based on the plasma dispersion effect. These modulators perform excellently
and are very compact compared with modulators created in other platforms.
R = I/P (2-22)
Here, I is the electric current generated, and P is the optical power incident on the photodiode.
Example 9
Find the responsivity of a photodiode that generates 800 µA of electric current in response
to an input optical power of 1.2 mW.
Solution: Use Equation 2-22 to calculate the responsivity. The units for current and power
must be converted to A and W.
R = 800 x 10-6 A/(1.2 x 10-3 W) = 0.67 A/W
The wavelengths of light used in optical fiber transmission systems center on 1310 nm and 1550
nm. Silicon is not a good absorber of the light at these wavelengths, and this makes silicon-based
photodetectors inefficient in this case. Figure 2-1 at the beginning of the module shows that
germanium, also a group IV element, absorbs light more efficiently than silicon at these
communications wavelengths. Due to this property, and its compatibility with the CMOS
process, germanium grown on silicon has emerged as the material of choice for photodetectors
used in optical fiber transmission systems. Some fabrication challenges still exist with such
photodetectors due to a 4% mismatch between the lattice constants of silicon and germanium.
However, Ge-on-Si detectors with good performance have been obtained.
Similar to the case of silicon photodetectors, the PIN structure is the most used structure for Ge
on Si detectors. High responsivity values of 0.9 to 1.0 A/W have been achieved. The bandwidth
is typically 20–40 GHz, and higher values up to 120 GHz have been demonstrated. The dark
current can be as small as 1–10 µA, with higher values for detectors that have a large area.
The PIC fabrication process does not end with the creation of the devices on the wafer. Further
processing includes inspection, dicing, testing, assembly, and packaging. These usually take
place in environments with less stringent requirements than a clean room. These processes are
referred to as the “back-end.”
The processed wafer is first inspected to find physical abnormalities that indicate if any damage
is present. Even though this can be done visually by a human operator, in high-volume
production, wafer inspection is performed automatically using specialized equipment. Vendors
of patterned wafer inspection equipment include Applied Materials, KLA-Tencor, and Nikon.
Dicing is the process of cutting the wafer to separate the individual devices. A circular dicing
blade can be used in this process, as shown in Figure 2-30. An individual device separated from
the wafer is known as a die. Dicing allows the individual devices on a wafer to be tested for
performance and compliance before being packaged.
Figure 2-30 Dicing of wafer into individual devices. Courtesy of Advanced Motion Controls.
Assembly and packaging are the processes through which a die is bonded to other dies and
joined to the package and permanent electrical and optical connections are made to the dies. To
have a good yield, the dies need to be tested before they are sent for assembly and packaging. In
the electronics industry, this is called using “known good dies.”
Figure 2-32 a) Fiber arrays, courtesy of AiDi Figure 2-32 b) Fiber array, courtesy of
Hantech. This fiber array is angle polished to
avoid back reflections of the optical signal.
Electrical signals are sent to the die through thin metal wires made of gold, copper, or aluminum.
The process of attaching electrical wires to the die is called wire bonding.
Testing individual dies one at a time is a time-consuming process. To speed it up, the wafer
might initially be diced into strips rather than individual dies. Each strip is mounted on the test
station, and optical fibers are brought into contact with the first die on the input and output sides.
In this case, light is coupled into and out of the PIC by edge coupling. Probe conductive needles
can be positioned on the surface of the die for electrical signals. The test runs automatically, in
conjunction with moving the strip vs. optical fibers and probe needles from die to die until the
entire strip has been tested. The strips are then diced to obtain the individual dies, which are
dispositioned according to the product specifications and test results.
Currently, state-of-the-art testing of dies is done at the wafer level, without the need to dice the
wafer upfront. Both electrical and optical parameters can be tested this way. Light is coupled to
and from the die through grating couplers, as described above. Fiber arrays mounted at an angle
to the surface of the die are used in many cases. Figure 2-33 shows two automated test stations
performing wafer level testing. Fiber arrays are visible in the left picture.
Figure 2-34 a) Typical diode laser package Figure 2-34 b) Typical packaged device
Because of the continued demand for more processing power and higher speed, the electronics
industry has recently been evolving to include several chips in a single package, using what is
called system-in-package (SiP) technology. This approach increases speed and reduces the size
of the package. To achieve this, several dies are stacked on top of one another, and wire bonding
connects the dies to one another or to the package. This method is known as 3D packaging.
Often the die at the bottom of the stack is turned upside down and electrically connected to the
package through solder material, eliminating the need for wire bonds between the package and
chips. This technology is called flip-chip. The advantage of the flip-chip technology is that it
enables a shorter distance between electrical connections than wire bonds do, and the shorter
distance translates to better electrical performance. Figure 2-36 illustrates the wire bond and flip-
chip approaches.
A solution to the problem of increased wire bond density per package is to use a different
method for electrical connections. This method consists of electrically connecting the dies in the
stack to one another and the package using through-silicon vias or TSVs. These are vertical
openings in the silicon substrate that reach all the way down through the die and wafer. The
openings are filled with materials such as copper, tungsten, solder material, and conductive
adhesive. Many TSVs serve as electrical connections, but some are thermal TSVs (TTSVs) that
allow for heat dissipation and thermal management. Compared with wire bonds, TSVs have the
advantages of enabling a higher density of electrical connections and allowing for shorter
distances, which improve performance.
At this time, the full 3D packaging with TSVs solution is not yet fully implemented due to its
very high level of complexity. An intermediate solution denoted as 2.5D packaging technology is
currently used. In this technology, a layer of silicon known as a silicon interposer is placed
between the dies and the substrate or package. The silicon interposer is where TSVs are formed
to connect the dies to one another. Figure 2-38 illustrates the 2.5D technology.
Figure 2-38 A 2.5D system packaged using a silicon interposer and TSVs
Figure 2-40 Optical interconnect between two processor chips. Courtesy of APIC Corporation.
© 2016 University of Central Florida
This material was created under Grant # 1303732 from the Advanced Technological Education
division of the National Science Foundation. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the author(s) and do not necessarily
reflect the views of the National Science Foundation.
ISBN 978-0-9903125-7-4
CONTENTS OF MODULE 3
Introduction ..................................................................................................................................... 1
Prerequisites .................................................................................................................................... 2
Objectives ........................................................................................................................................ 2
Basic Concepts ................................................................................................................................ 2
Material Properties of III-V Semiconductors .............................................................................. 2
III-V Semiconductor Photonic Integrated Circuits ..................................................................... 5
Passive III-V Semiconductor PIC Devices ................................................................................. 5
III-V Passive Optical Waveguide ............................................................................................ 5
Input/Output Coupling to III-V Semiconductor PIC Devices ................................................. 8
Other Passive Devices ............................................................................................................. 9
Multimode Interference (MMI) Coupler ................................................................................. 9
Arrayed Waveguide Grating ................................................................................................. 12
International Telecommunications Union (ITU) Grid .......................................................... 15
AWG Spectral Response ....................................................................................................... 17
Characteristic Parameters of AWG ....................................................................................... 18
Performance of III-V Semiconductor AWG Devices ........................................................... 20
Active III-V Semiconductor PIC Devices ................................................................................. 21
Diode Lasers .......................................................................................................................... 21
Advances in Laser Junction Structure ................................................................................... 26
Semiconductor Optical Amplifier (SOA).............................................................................. 29
Modulators............................................................................................................................. 32
Photodetectors ....................................................................................................................... 35
Fabrication of III-V Semiconductor PICs ................................................................................. 36
Summary ....................................................................................................................................... 37
Problem Exercises and Questions ................................................................................................. 38
References ..................................................................................................................................... 40
I NTEGRATED P HOTONICS C OURSE FOR T ECHNICIANS
Module 3
III-V Semiconductor Devices
INTRODUCTION
Previous modules have discussed the importance of semiconductors in photonic integrated
circuits and devices. III-V semiconductors are one of the most common types used in PICs. III-
V semiconductors are obtained by combining an element from group III in the periodic table
with an element from group V. These combinations are also called binary compound
semiconductors. II-VI and IV-VI binary compound semiconductors are also possible. In
addition, ternary (three-element) and quaternary (four-element) alloys from elements in groups
III and V have been created.
III-V semiconductors have attractive electrical and optical properties with important
applications in both electronic and photonic devices. Electronic transistors based on gallium
arsenide (GaAs) and indium phosphide (InP) perform faster than their silicon counterparts and
are used in high-power and high-speed electronic integrated circuits. In regard to optical
properties, GaAs and InP are good light emitters due to their direct energy bandgap. This
property is exploited in the fabrication of highly efficient semiconductor diode lasers. By
contrast, silicon and germanium are inefficient light emitters due to their indirect bandgap, as
discussed in the previous module. Using ternary and quaternary alloys such as aluminum
gallium arsenide, indium gallium arsenide phosphide, and others has expanded the range of
wavelengths emitted by diode lasers from 600 nm to 1650 nm.
Numerous photonic devices based on III-V semiconductors, both active and passive, have been
developed and are widely used today. Active devices include lasers, modulators, and
photodetectors. Passive devices include couplers, gratings, multiplexers/demultiplexers,
interferometers, and others. Semiconductor diode lasers are currently the preferred light source
for virtually all photonic applications. This includes applications of silicon based photonic
integrated circuit (PIC) devices, that lack a practical light source based on silicon. An important
advantage of PIC devices based on III-V semiconductors is the possibility of monolithic
integration of all devices including light sources, modulators, passives and photodetectors in one
chip. Recently a very high degree of integration of 600 optical functions on two indium
phosphide chips has been achieved in commercially available devices.
At the same time, there are some disadvantages associated with III-V semiconductor PIC
devices. The raw materials from which these are made are not as widely available as silicon, and
are consequently more expensive. The processes used to build these devices are not fully
compatible with complementary metal-oxide semiconductor (CMOS) technology. The GaAs
and InP substrate wafers the devices are built on are more brittle, resulting in smaller wafers. A
common diameter for these is wafers is 4 inches, although 6-inch-diameter wafers are now
possible. Nevertheless, III-V semiconductor devices dominate in certain applications such as
1
diode lasers and devices used in long-haul optical fiber transmission systems. This module
discusses basic properties of III-V semiconductors and how these are applied in active and
passive PIC devices.
PREREQUISITES
OP-TEC’s Fundamentals of Light and Lasers Course 1
OP-TEC’s Integrated Photonics: Modules 1 and 2
OBJECTIVES
Upon completion of this module, the student should be able to:
Explain material properties of III-V semiconductors important for PICs
Describe characteristics of the following passive III-V semiconductor PIC devices:
o Straight and bend waveguide
o Coupling between optical fiber and III-V waveguide
o Multimode interference coupler
o Arrayed waveguide grating
Explain the ITU grid used in wavelength division multiplexing and perform calculations
involving ITU frequencies and wavelengths
Describe characteristics of the following active III-V semiconductor PIC devices:
o Diode laser
o Semiconductor optical amplifier (SOA)
o Modulator
o Photodetector
Provide examples of monolithically integrated III-V PIC devices and discuss their
advantages over discrete devices
Describe the MOCVD deposition process used to obtain III-V semiconductor thin films
BASIC CONCEPTS
Figure 3-2 shows the variation of the index of refraction of gallium arsenide and indium
phosphide with wavelength, based on experimental measurements. At the communication
wavelength of 1.55 µm, the index of refraction of GaAs is about 3.37, while the index of
refraction of InP is about 3.17. Similar to the indices of refraction of glass and silicon, the index
of refraction of the two III-V semiconductors decreases with increasing wavelength.
n GaAs n InP
3.6
3.55
3.5
Index of refraction n
3.45
3.4
3.35
3.3
3.25
3.2
3.15
3.1
1000 1100 1200 1300 1400 1500 1600 1700 1800
Wavelength (nm)
Figure 3-2 Index of refraction of gallium arsenide and indium phosphide vs. wavelength
Typical dimensions for the core of the channel waveguide in Figure 3-3 a) are 1 to 5 µm width,
and 0.5 to 2 µm thickness or height. The buried channel is surrounded by InP on all sides. For
the buried rib waveguide in Figure 3-3 b) the height of the rib, HR, is typically 1 to 2 µm, and
the width is 3 to 7 µm. The buried rib waveguide has the advantage of remaining single-mode
even for the wider widths of 6 or 7 µm.
The raised strip and deep ridge waveguides in Figure 3-3 c) and d) have air as cladding on three
or two sides. This increases the index contrast in those directions and results in better
confinement of the light inside the core. The deep ridge can have a very thin core, 0.3 - 0.5 μm,
which makes it easier to fabricate.
Example 1
Lateral refractive index contrast for deep ridge waveguide. Calculate the lateral refractive
index contrast of waveguides having InGaAsP cores and air as lateral cladding.
Solution: The refractive index contrast is calculated using Equation 1-5 in Module 1
Δ = (ncore2 – nclad2)/(2 ncore2) = (3.292 – 12)/(2 x 3.292) = 0.45 = 45%
Historically, the propagation loss of III-V semiconductor waveguides has been high, on the
order of 10 dB/cm or more. Advances in the design and fabrication process have brought
Figure 3-4 Effective indices of TE and TM modes of a deep ridge waveguide as a function of the
waveguide core width
Table 3-1 shows the complete waveguide parameters for a single-mode buried channel
waveguide.
A vertical taper works similarly to the lateral taper, but this time the thickness of the waveguide
is tapered down. Even though the principle of operation is the same, the fabrication process for
such a taper is quite different from the one used to change the waveguide width. The waveguide
width is determined during the photolithography process. A thickness change is more
complicated; one possible solution is to use several etching steps.
Combined lateral and vertical tapers are also possible. SSCs are capable of reducing the coupling
loss between III-V semiconductor waveguides and single-mode optical fiber to values less than
1dB, which is considered an acceptable loss.
The principle of operation of an MMI device is based on interference between the waveguide
modes. Light enters the device through one of the single-mode input access waveguides. When it
arrives at the entrance to the wide section, it excites the modes available in the wide section and
creates a certain distribution of the electric field. The different optical modes each travel with a
slightly different speed inside the wide waveguide. Consequently, each mode will accumulate its
own phase difference after traveling the distance L. If the phase differences between all modes at
the end of the wide waveguide are multiples of 2π, the modes will interfere constructively, and
the final distribution of the electric field will reproduce the initial distribution at the start of the
wide waveguide. This phenomenon is denoted as self-imaging. Specifically, when phase
differences are multiples of 2π we obtain a single image that exactly reproduces the input light,
that is, the output light exits from the waveguide directly across the input waveguide. This is
called a direct single image. MMI devices are often designed to achieve self-imaging.
A mirrored single image is also possible with an MMI coupler; this happens when light exits an
output waveguide in a mirror position from the input waveguide. In addition to single images, we
can also create multiple images by extending the distribution of the electric field over several
output waveguides. In this case, the input power is split between the output waveguides.
The width W of the wide waveguide determines how many modes can propagate through the
device. This is a critical parameter for the MMI coupler imposing tight fabrication tolerances. In
general, MMI couplers are designed to allow from 3 to 8 modes in the wide region. The length,
L, is the parameter that determines what kind of image forms at the exit from the wide region. Of
interest are direct single, mirrored single, and certain multiple images.
Recall that each mode has its own effective refractive index, neff, that determines the phase
accumulated by the light traveling in the waveguide. The equation for the phase was given in
Module 2 and is repeated below.
Φ = (2π/λ) neff L (3-1)
The different modes of the waveguide are labeled 0, 1, 2, …, with mode 0 being the fundamental
mode. The corresponding effective refractive indices are neff0, neff1, neff2, … We define a length
Lπ to be equal to:
Lπ = λ/[2 (neff0 – neff1)] (3-2)
Due to the properties described above, MMI devices are used in coupler and splitter applications.
Module 2 described directional couplers and Y-branches that achieve these functions. Recall that
a directional coupler has two inputs and two outputs and that we can create any power splitting
ratio between the two outputs by controlling the length of the coupling region. The most used
application of a directional coupler is equal power splitting between the two outputs, in which
case the device is known as a 3dB coupler. 3dB directional couplers are used by themselves or as
part of Mach-Zehnder Interferometer devices. A Y-branch has one input and two outputs that
carry equal power.
The MMI coupler in Figure 3-6 has the same functionality as a directional coupler. Depending
on the length L, power can exit entirely from the output waveguide, which is located directly
across the input waveguide; this is called the bar state of the coupler. Power can also exit from
the output located diagonally across from the input, in which case the coupler is said to be in
cross state. Power can also be split equally between the two outputs.
Example 2
Find the shortest length L of an MMI coupler with two inputs and two outputs that results in
a cross state for the coupler, and in a 3dB coupler. Assume that the effective refractive
indices of the fundamental and first order modes of the wide waveguide are 3.2380 and
3.2343 respectively. The wavelength is 1.55 µm.
Solution: First calculate length Lπ using equation 3-2.
Lπ = λ/[2 (neff0 – neff1)] = 1.55 µm/[2 (3.2380 - 3.2343)] = 209.5 µm
The cross state for the coupler corresponds to a mirrored image. Using equation 3-3b the
first mirrored image is obtained when L = 3 Lπ = 628 µm.
The 3dB coupler corresponds to a multiple image. Using equation 3-3c the first multiple
image is obtained when L = (1/2) (3 Lπ) = 314 µm.
The following phenomena take place as we follow the light signal along the AWG, resulting in
the separation of the wavelengths at the exit. The input light enters the first planar region, also
called the input slab. Recall from Module 1 that a slab waveguide confines light in only one
direction, the vertical direction in this case. In the horizontal direction, the large width of the slab
allows for the broadening shown in the figure. This is needed to allow light to be coupled to all
the waveguides in the arrayed region.
When the light reaches the end of the planar region, it enters the arrayed waveguide grating
region. This region is made up of waveguides of varying lengths, with shorter waveguides on the
bottom and progressively longer waveguides as we move toward the top. The length difference
between any two adjacent waveguides is the same and is equal to an integer multiple of the
central wavelength of the demultiplexer. We denote the central wavelength by λc and the length
difference between waveguides by ΔL.
The individual light waves carried by the arrayed waveguides will interfere in the second planar
region (or output slab). Light traveling in each of the arrayed waveguides will accumulate a
certain phase, whose value depends on the wavelength. For a wavelength equal to λc the phases
at the end of the arrayed waveguides will differ by multiples of 2π. This will result in
constructive interference at the end of the second planar region, which will have the effect of
recreating the input field at the output. The green arrow in Figure 3-8 illustrates this effect.
For wavelengths different from λc, the phases at the end of the arrayed waveguides differ by
quantities other than 2π. Because of this, the input field will now be recreated in a different
output location than across from the input. The red arrow, corresponding to a higher wavelength
than the green one, illustrates wavelengths larger than λc exiting from the bottom half of the chip.
Shorter wavelengths (represented by the blue arrow) will exit from the top half of the chip.
If the direction of propagation of the light in Figure 3-8 is reversed, the AWG functions as a
multiplexer. In this case, multiple beams of light, each with a different wavelength, enter the
device from the right. The exit to the left is through one output carrying all the different
wavelengths in the same waveguide.
Example 3
Find the ΔL of an AWG with central wavelength equal to 1550 nm, based on arrayed
waveguides with effective index equal to 3.21. Assume that the grating order m is equal to
64.
Solution: Using equation 3-4, we obtain ΔL = mλc/neff = (64)(1.55µm)/3.21 = 30.9 µm.
Figure 3-9 shows more details about the geometry of the AWG’s elements. In this figure, the
planar regions are called free propagation regions, or FPR. As the figure shows, the first FPR
ends in a circular arc, centered at the point where light enters the slab. The radius of the arc is
denoted by R. The second planar region which is symmetric to the first one, has the same radius
of curvature, R. Other parameters that determine the separation between the wavelengths at the
output of the AWG are: the effective index of the slab, ns; the distance between two adjacent
arrayed waveguides at the end of the first planar region, also called the pitch of the arrayed
waveguide, d; and the distance between two adjacent output waveguides at the exit from the
second planar region, D.
The array pitch, d, is a critical element for an AWG: it directly influences the device’s insertion
loss. It is desirable for the pitch to be very small so that the field exiting from the first slab can be
coupled in its entirety to the arrayed waveguides. In practice, there will always be a small
amount of coupling loss between slab and waveguides, because the fabrication process cannot
reduce gaps between waveguides below certain values. With III-V semiconductors, we can
create waveguide gaps of about 0.5 μm, which results in very little coupling loss. We need to
keep in mind, though, that this loss is multiplied by two because there are two slabs.
Example 4
For the AWG in Example 3, find the FSR and the maximum number of channels. Assume a
wavelength separation between channels equal to 0.8 nm.
Solution: The AWG in Example 3 has a central wavelength λc equal to 1.55 µm and a
grating order m = 64. Using Equation 3-5, we obtain FSR = 1550 nm/64 = 24.2 nm.
The maximum number of channels can be obtained using equation 3-6:
N = 24.2 nm/0.8 nm = 30.27. By rounding the result to an integer we obtain N = 30.
Note that the ITU frequencies are represented by whole numbers, and are equally spaced. By
contrast, the corresponding ITU wavelengths are not whole numbers, and are not equally spaced.
We can determine the wavelength spacing Δλ corresponding to a known frequency spacing Δf by
using the following equation:
Δλ = λ2 Δf/c (3-8)
Example 6
Find the wavelength spacing corresponding to a 100 GHz frequency spacing for a
wavelength equal to 1530 nm and a wavelength equal to 1560 nm.
Solution: For light with wavelength equal to 1530 nm, the wavelength spacing can be
calculated from equation 3-8.
Δλ = λ2 Δf/c = (1530 x 10-9 m)2 x (100 x 109s-1)/(299,792,458 m/s) = 7.81 x 10-10m
= 0.781 nm
Applying the same equation for the wavelength of 1560 nm, we obtain:
Δλ = λ2 Δf/c = (1560 x 10-9 m)2 x (100 x 109 s-1)/(299,792,458 m/s) = 8.12 x 10-10m
= 0.812 nm
For wavelength division multiplexing (WDM) the most common values for channel spacing are
200 GHz, 100 GHz, and 50 GHz. The corresponding approximate wavelength spacings are 1.6
nm, 0.8 nm, and 0.4 nm. When a spacing of 200 GHz is used, we can use either the even
numbered channels or the odd numbered channels described above. When 50 GHz is used,
channels placed at the mid points of the ones described above must be obtained.
Figure 3-10 Transmission vs. wavelength for one output of an arrayed waveguide grating, for TE and
TM polarizations
Figure 3-11 shows the superimposed transmission curves for all outputs of an AWG. Each color
represents a different output. The device illustrated has 8 outputs, with a channel spacing of 200
GHz, or about 1.6 nm. The periodic response of the AWG is illustrated by the repetition of the
eight channels every 30 nm. Three such repetitions are visible in the figure.
The FSR (nm or GHz) is the minimum wavelength (or frequency) range corresponding to two
input signals that exit from the same AWG output.
The insertion loss (dB) can be found using the peak of the transmission curve in each channel;
the minimum peak loss across all channels is the device’s insertion loss. The insertion loss
should be minimized to avoid losing optical power as light travels through the AWG. Values
below -3dB are desirable. Recall that insertion loss is typically expressed as a negative number.
Insertion loss non-uniformity (dB) is the maximum spread between the peak losses of all
channels. This non-uniformity should be as close to zero as possible, desirable values being less
than 1.0 dB. There is a typical insertion loss roll-off as we move from an AWG’s center channel
to an edge channel (shown in Figure 3-12). This roll-off can be 3dB or even higher. Edge
channels tend to have the most loss non-uniformity, while channels closer to the center are closer
in insertion loss.
One way to reduce the non-uniformity is to design the AWG with a higher FSR than necessary,
and not use the edge channels. For example, we might use only the 16 outputs of an AWG that
has been designed with an FSR that allows for 24 channels. In this case, empty slots will appear
in the AWG’s transmission. For example, the AWG in Figure 3-11 was designed to allow for 10
channels, but only 8 channels are used. Using this method, we can acheive non-uniformity of
values less than 1 dB, at the expense of increasing the size of the device (bigger FSR requires
bigger radius for the free propagation regions).
Polarization dependent loss (PDL) (dB) is the difference between the insertion loss
corresponding to the TE and TM polarizations. This parameter was illustrated in Figure 3-10,
where the PDL was about 1 dB. This parameter should be less than 0.5 dB.
3dB Bandwidth (nm or GHz) is the wavelength (or frequency) range for two points on the
transmission curve of one output for which the transmission is 3dB down from the peak. The
3dB bandwidth should be as large as possible to accommodate drifts, particularly drifts in the
center wavelengths of the lasers used as light sources for the system. 3dB bandwidth values
Diode Lasers
III-V semiconductors have excellent light emitting properties, which are exploited in the
fabrication of diode lasers. Diode lasers are described below starting with the first such laser
realized in gallium arsenide (GaAs), followed by double heterostructure and other types of diode
lasers that are used in photonic integrated circuit devices.
Figure 3-13a shows the relative populations of the energy bands on both sides of a p-n junction
with no external voltage applied to the diode. The conduction band of the n-type material
contains electrons that act as current carriers, whereas the valence band in the p-type material has
holes that act as current carriers. When a forward voltage is applied to a diode, the energy levels
across the junction region of the diode shift, as shown in Figure 3-13b. This shift allows
electrons in the n-type material and holes in the p-type material to diffuse more readily into the
junction region to establish a current through the diode. While diffusing through the junction,
some electrons will combine with holes. Since the electrons in the conduction band have
energies greater than those in the valence band, the electrons that combine with holes lose an
amount of energy equal to the semiconductor’s band gap energy. These electrons’ energy loss
appears as photons of light with a wavelength λ given by Equation 3-9.
ΔE = hc/λ (3-9)
Where ΔE is the bandgap energy, h is Planck’s constant, c is the speed of light, and λ is the
wavelength of the emitted photon.
As the forward voltage is increased across the diode, the current in the diode increases, and more
and more electrons and holes enter the junction region and combine with one another.
Each time a hole and an electron combine, the electron drops from the conduction band to the
valence band and generates a photon. At this point, the diode is operating as a light emitting
diode (LED). As the forward voltage is further increased, the number of photons generated in the
junction region becomes large enough to initiate the process of stimulated emission. This means
that photons produced from the combining of electrons and holes stimulate other conduction
band electrons in the junction region to drop to the valence band and release photons that are
identical to the stimulating photon. The diode current at which stimulated emission begins is
called the threshold current. At the threshold current, the diode begins operating as a laser.
Usually, the mirrors at the ends of the laser cavity are formed from the cleaved ends of the laser
diode, with no further coating. The reflectivity at the interface between gallium arsenide and air
is approximately 36%. If output is desired from only one end of the device, or if mirrors of
higher reflectivity are desired to reduce the threshold for laser operation, the reflectivity may be
increased by coating the surfaces with metal films. Optical standing waves may exist between
any two of the parallel surfaces of the diode. Two sides are purposely roughened to reduce
reflection and prevent lasing “across” the diode cavity.
The output power available from this laser is limited by the gain of the active medium, which is
dependent on the current density through the junction. Higher currents produce greater power,
but higher currents also increase heating effects that can damage the device.
Losses in the laser cavity have two primary causes. The first is diffraction loss. The active region
has a width of only about one micrometer. Thus, light quickly diverges out of the active region.
This loss may be reduced by making the junction wider and by better confining the light to the
Figure 3-15 Variation of the wavelength of a commercial semiconductor laser with temperature
Figure 3-16 illustrates the beam divergence of a typical semiconductor laser. The emission from
a semiconductor laser tends to be an elliptical beam with a full angle divergence around 20° in
the direction perpendicular to the junction and around 5° in the direction parallel to the junction.
These angles may vary considerably between different lasers. The elliptical beam may be
rounded by using lenses that have different focal lengths in the two directions perpendicular to
the direction of propagation. This will produce a beam with a more circular profile.
Figure 3-16 Beam profile from a stripe geometry heterojunction semiconductor laser
The wavelength of the semiconductor laser emission may be varied over a large range by varying
the composition of the material. The most widely used semiconductor materials are the so-called
III-V compounds, based on materials from columns III and V of the periodic table. Gallium
arsenide is an example. But many more material compositions have been used, including ternary
Of the materials in the table, the most widely used ones have been indium gallium arsenide
phosphide (for communication applications), aluminum gallium arsenide (for data-storage
applications, such as compact disc players) and aluminum gallium indium phosphide for shorter
wavelength applications (such as replacements for helium-neon lasers).
The materials discussed so far emit light in the red and infrared portions of the spectrum.
Applications requiring shorter wavelengths, in the blue or near-ultraviolet regions, use materials
based on gallium nitride, such as aluminum gallium nitride (AlxGa1-xN) or indium gallium nitride
(InxGa1-xN). The wavelengths available from these devices cover the range from 375 nm to 640
nm. They have become important in data-storage devices requiring high packing density.
This structure improves the electrical and optical confinement of semiconductor lasers and
reduces the threshold currents required to operate them, thus increasing the output laser power.
This is only one example of the many variations in semiconductor laser structures that have been
developed, but it shows how changes in laser structure can improve performance.
In many modern lasers, feedback is caused by reflection from a structure of finite length. These
structures are said to cause distributed feedback. There are a number of types of distributed-
feedback structures. Here, we describe two of them: distributed-feedback lasers and distributed
Bragg reflectors. Both rely on a concept called Bragg reflection, which uses the periodic
variation in some property of a material, such as thickness, to produce reflection. Bragg
reflection occurs when the light passes through a structure consisting of multiple layers of
alternating materials with different indices of refraction. It can also be caused by a periodic
variation of some property, such as height, in a waveguide. At each boundary between
successive layers with different properties, there is a partial reflection of the light. If the optical
thickness of the layers is close to one-fourth of the wavelength of the light, the partial reflections
can interfere constructively. The structure can serve as a high-quality reflector for a specific
wavelength. If the laser’s output has a broad spectrum, the reflector can select the wavelength
that is four times the layer thickness and reflect that wavelength, thus acting like a narrow band
filter.
Figure 3-18 Diagram of the side view of: a) a distributed-feedback semiconductor laser and
b) a distributed Bragg reflector
In the DFB laser, the entire laser cavity has a periodic structure that acts as a reflector.
A distributed Bragg reflector (DBR) is similar to the DFB device, both in construction and
operation; see Figure 3-18 b). But the periodic structure is a grating that is outside the laser’s
active medium. This makes the device easier to fabricate. The gratings are at each end of the
active medium and act as reflectors.
Both the DFB and DBR devices act as wavelength-selecting elements. The spacing of the
periodic variations is chosen to select only a narrow band of wavelengths. Thus, they can reduce
the spectral width of the laser output substantially, as compared to what was described earlier.
Additionally, these devices allow the selection of a longitudinal mode from the spectrum. If the
temperature of the device changes, thermal expansion will change the spacing of the periodic
variations, and the wavelength of the laser will change. Thus, if a stable wavelength is needed,
the temperature of the laser must be controlled.
Figure 3-19 A semiconductor laser using a multiple quantum well structure in the active region
Example 7
Find the gain in dB of an optical amplifier that takes in 15 µW of power and produces an
output power equal to 2.5 mW.
Amplification corresponds to a gain, G, greater than one. To create a gain in the optical power,
the amplifier needs a pump or an energy source. In the case of the semiconductor optical
amplifier (SOA), the gain is created by the active region of the amplifier under the action of an
electric current running through it which acts as the pump. In other words, the SOA is similar to
a semiconductor laser except that it does not have feedback, which was provided by the cavity
mirrors in lasers. For the amplifier, reflection of the light at the ends of the active region must
actually be prevented for the device not to start lasing. This can be achieved by depositing
antireflection coatings at the two ends.
Figure 3-20 shows a schematic of an SOA. The pump current runs vertically between the two
metal electrodes. The active or gain region is the layer of intrinsic InGaAsP semiconductor. This
layer also acts as the core of the optical waveguide that guides the light through the structure.
Like other optical amplifiers, the SOA exhibits certain characteristics. One of these is gain
nonlinearity or saturation. This happens when the output optical power reaches high levels, at
which point the gain starts to level off. For an SOA the maximum output power is about 63 mW.
SOAs also have some amount of noise, meaning that there is a background to the output power
that is coming from other sources than the input light. This background is called amplified
spontaneous emission (ASE) and is made out of photons spontaneously emitted in the active
medium. By contrast, it is the stimulated emission photons that are responsible for the
amplification of the input light. To reduce the effect of the ASE noise, filters can be used at the
exit from the amplifier. Table 3-3 summarizes some of the main properties of SOAs.
Important advantages of SOAs include their electrical pumping, small size, low cost, and
potential to be integrated with other semiconductor PICs. One example of such monolithic
integration resulting in a high performance PIC device is the multiwavelength laser.
Figure 3-21 Integrated multiwavelength laser based on SOA array and AWG multiplexer
A recent advanced device based on this principle was demonstrated in the form of a four-
wavelength, short-pulse laser integrated on a single chip. Figure 3-22 illustrates the chip, which
is based on InP materials, including multiple quantum well structures in the active devices. The
The integrated multiwavelength laser is emitting 4 channels separated by 400 GHz, each
producing short pulses of light with a repetition frequency of 2.53 GHz. This is the first
demonstration of a monolithically integrated short pulse multiwavelength laser. Previous such
devices were based on discrete (bulk) optical elements, making the device much bigger in size,
more expensive, and harder to create in high volume.
Modulators
Modulators are devices capable of adjusting the power level of an optical signal to any value
between a minimum value and a maximum value. A controlling signal, usually an electric
current, is used to achieve the desired output optical power.
Modulators fabricated from III-V semiconductors are based on the phenomenon of
electroabsorption: the absorption of the material changes when an electric field is applied to it.
When the amount of light absorbed by the material increases, the optical power transmitted at the
output decreases. The increase in absorption is achieved by applying a reverse voltage. These
modulators are called electroabsorption modulators (EAM).
The voltage applied to the modulator consists of short radio frequency (RF) square pulses. Under
the action of the applied voltage, the modulator switches the optical power at the output on and
off. The extinction ratio (ER) of the modulator, expressed in dB, is defined as the ratio between
the output power in the ON state and the minimum output power that can be achieved in the OFF
state (Equation 3-11). This minimum cannot be made exactly zero, but it can be a small value.
EAMs can be realized in several configurations. The simplest one is based on a single straight
waveguide as shown in Figure 3-23a. The applied square voltage pulses together with the output
optical power are shown in Figure 3-23b. The waveguide can have a double heterostructure and
can also contain quantum well structures. The ER of these modulators can be increased by
increasing the length of the waveguide, but that increases the propagation loss. The typical ER
for such modulators is 15 dB. The applied voltage has an amplitude of 2-3 volts. The speed of
response is high, which is characteristic of semiconductor materials. Rates of 10 Gb/s can be
achieved. The chip temperature needs to be kept constant due to the material’s sensitivity to
temperature changes; this is often done using a thermo-electric cooler.
Another configuration for EAMs is based on the Mach-Zehnder interferometer (MZI), described
in Module 2. An MZI consists of two 3dB couplers with two parallel waveguide arms running
between them. The 3dB couplers can be realized in different ways. In Module 2 these were either
based on directional couplers or Y-splitters. Semiconductor modulators based on MZI are called
Mach-Zehnder modulators or MZM. Many times the 3dB couplers in MZM are implemented as
multimode interference couplers (MMI), which were introduced earlier in this module. The two
arms of the MZM can be equal in length or can have a length difference to bias the modulator.
To modulate the output power a reverse voltage is applied to one of the arms.
Figure 3-24 Semiconductor Mach-Zehnder modulator configuration. The bottom half of the figure
shows the waveguide structure and the traveling-wave electrode structure.
Semiconductor MZMs have better performances than waveguide based EAM. They require
smaller voltages to operate, typically 1.5 V. They have better ERs, around 25 dB, and are also
faster being capable of operating at rates of 30 Gb/s and even 40 Gb/s.
Like SOAs, semiconductor modulators have the big advantage of being able to be integrated
with other PIC devices in the same chip. Modulators are commonly integrated with
semiconductor lasers for example.
Photodetectors
Module 2 presented photodetectors (PD) and their characteristic parameters. For silicon-based
PICs, germanium-based photodetectors are preferable, since germanium can be grown on silicon
substrates. With III-V semiconductors, indium gallium arsenide (InGaAs) is a material
combination that works for a wide range of wavelengths, including the telecommunications
wavelengths of 1.31µm and 1.55 µm, as shown in Figure 3-1 at the beginning of this module.
The general principle of operation for semiconductor photodetectors is the creation of charge
carriers (electron-hole pairs) under the action of light. When a semiconductor material is
illuminated by photons of an energy greater than or equal to its bandgap, the absorbed photons
promote electrons from the valence band into the conduction band. Electrons in the conduction
band behave like free electrons able to travel long distances across the crystal structure under the
influence of the applied electric field. In addition, the positively-charged holes left in the valence
band contribute to electrical conduction by moving from one atomic site to another under the
effect of the electric field. In this way the separation of electron-hole pairs generated by the
absorption of light gives rise to a photocurrent.
The most common structure of a semiconductor photodetector is either the p-n junction or the p-
i-n junction. P-i-n photodetectors are faster than p-n ones. The p-i-n structure is also used to
build avalanche photodetectors which have larger responsivity than regular photodetectors.
InGaAs photodetectors are used in a variety of applications including optical power meter,
optical communications, laser monitors, and others. The InGaAs photodetectors used in the
wavelength range 1.0 to 1.7 µm have excellent characteristics. These include high responsivities
around 1A/W, low dark current on the order of nA, and high cutoff frequencies of hundreds of
MHz to a few GHz.
Like the other active III-V semiconductor devices, the photodetectors can be integrated on the
same chip with other PIC devices, leading to space and cost savings, and improved performance.
For example, an AWG demultiplexer device can be integrated with photodetectors for each
channel. The applications for this combined device are in receiver circuits at the end of the
transmission, as well as in power monitoring devices in wavelength division multiplexing optical
communication systems.
Figure 3-26 Semiconductor AWG demultiplexer monolithically integrated with photodetectors
SUMMARY
This module presented an overview of III-V semiconductor photonic integrated circuits. The
material properties of III-V semiconductor were presented first, followed by descriptions of
passive and active building-block devices. New passive devices introduced in this module
included multimode interference (MMI) couplers and arrayed waveguide gratings (AWG).
Active devices presented included diode lasers, semiconductor optical amplifiers (SOA),
modulators, and photodetectors. Monolithically integrated devices combining passive and active
PICs were discussed together with their advantages over discrete devices.
ISBN 978-0-9903125-8-1
CONTENTS OF MODULE 4
Introduction ..................................................................................................................................... 1
Prerequisites .................................................................................................................................... 2
Objectives ........................................................................................................................................ 2
Basic Concepts ................................................................................................................................ 3
Material Properties of Silica, Lithium Niobate, and Polymers ................................................... 3
Silica ........................................................................................................................................ 3
Lithium Niobate ...................................................................................................................... 5
Polymers .................................................................................................................................. 7
Silica-Based Photonic Integrated Circuits .................................................................................. 8
Silica-on-Silicon Straight and Bend Waveguides ................................................................... 8
Input/Output Coupling to Silica PIC Devices ....................................................................... 10
Silica Passive Devices ........................................................................................................... 11
Lithium Niobate Photonic Integrated Circuits .......................................................................... 18
Lithium Niobate Waveguides ................................................................................................ 18
Lithium Niobate Modulators and Switches ........................................................................... 19
Polymer-Based Photonic Integrated Circuits ............................................................................ 20
Polymer Waveguides ............................................................................................................. 20
Waveguide Interconnects ...................................................................................................... 21
Variable Optical Attenuator .................................................................................................. 22
Digital Optical Switch ........................................................................................................... 23
Tunable Bragg Grating .......................................................................................................... 24
Tunable Wavelength Laser .................................................................................................... 25
Strain Sensor with Tunable Bragg Grating ........................................................................... 26
Comparison of Waveguide Parameters across Materials Platforms ......................................... 27
Summary ....................................................................................................................................... 28
Problem Exercises and Questions ................................................................................................. 29
References ..................................................................................................................................... 30
I NTEGRATED P HOTONICS C OURSE FOR T ECHNICIANS
Module 4
Dielectric and Polymer Waveguides
and Waveguide Devices
INTRODUCTION
The term “integrated optics” was proposed by S. E. Miller in 1969. It encompasses devices
based on optical waveguides, particularly planar optical waveguides. The motivation for
developing waveguide devices was provided by the emerging field of optical communications.
Diode lasers and optical fibers demonstrated in the sixties were envisioned as acting as optical
sources and as transmission media for optical sources. However, optical communication systems
also needed optical components such as transmitters, receivers, modulators and others. Several
categories of material are used to create both waveguides and some optical devices; silicon-
based PICs and III-V semiconductor PICs were introduced and discussed in depth in previous
modules.
Silica, also called silicone dioxide, is a natural choice of material for planar waveguide devices,
because it was also the material that optical fibers were (and continue to be) made of. Efficient
coupling of light between optical fibers and silica-based planar waveguides was initially
expected, and strategies used to minimize insertion loss with silica waveguides are discussed
later in the module. Silica does not have light-emitting properties. Consequently, it is used for
passive PICs. Its advantages are low cost, excellent transparency, high threshold of optical
damage, high reliability, and mechanical rigidity. Silica PIC devices can be fabricated in high
volume using the same advanced technology used to fabricate electronic integrated circuits.
Many high-performance, passive PIC devices based on silica have been fabricated commercially
and deployed in optical transmission systems since they were envisioned five decades ago.
Another material that was adopted for early use in integrated optics devices is lithium niobate
(LiNbO3). Lithium niobate is a crystalline material with electro-optic properties; that is, its
index of refraction can be changed by applying an electric field. This material shares many of
the properties of silica: it is highly transparent in the near-infrared region used in optical
communications; it is thermally, chemically, and mechanically stable; and it is compatible with
integrated circuit fabrication technology. Because of its electro-optic properties, lithium niobate
has been used extensively to fabricate optical modulators. These devices are often used as
external modulators in combination with diode lasers.
A different class of materials used to fabricate a variety of PICs is polymers. Polymers have
several advantages, including low cost, simpler fabrication processes, and compatibility with
various substrates. Specific properties that make them attractive for PICs are their large thermo-
optic and electro-optic coefficients, photosensitivity, and tunability of the refractive index. A
1
new class of PICs, one that takes advantage of the flexibility of some polymer materials, has
been proposed. The new class is called flexible integrated optic devices. Polymer materials also
have disadvantages, the most important of which are their low thermal and chemical stability.
Many polymers cannot withstand very high temperatures, for example. This imposes more
stringent environmental requirements for polymer-based devices than for devices made of
materials like lithium niobate or silica.
For the different reasons just listed, silica, lithium niobate, and polymers are all attractive
materials for the fabrication of PICs. Their general disadvantage is that none can be a platform
for monolithic integration, as opposed to silicon and III-V semiconductors. The devices must be
used in conjunction with optical sources and photodetectors made of different materials.
Connections between them are made using optical fibers, which adds cost and complexity to the
system. This module presents properties of silica, lithium niobate, and polymer materials and
discusses several PIC devices based on them.
PREREQUISITES
OP-TEC’s Fundamentals of Light and Lasers Course 1
OP-TEC’s Integrated Photonics: Modules 1-3
OBJECTIVES
Upon completion of this module, the student should be able to:
Explain material properties of the following materials that are relevant to the fabrication
of PICs:
o silica
o lithium niobate
o polymers
Describe characteristics of the following passive silica PIC devices:
o straight and bend waveguide
o coupling between optical fiber and silica waveguide
o arrayed waveguide grating including athermal AWG
o AWG integrated with other PICs
Describe characteristics of lithium niobate waveguides and Mach-Zehnder
Interferometer electro-optic modulators
Describe characteristics of the following polymer PIC devices
o straight waveguide
o waveguide interconnects
o Variable optical attenuator
o digital optical switch
o tunable Bragg grating
o tunable wavelength laser
o strain sensor based on Bragg grating
BASIC CONCEPTS
Silica
Silica, SiO2, is abundant in the natural world. About 10% of the earth’s crust is made out of
quartz, which is silica. The material is also produced synthetically. Silica is the main component
of silicate glass, the most common type of glass. While quartz is a crystalline material, glass is
amorphous, lacking the ordered atomic or molecular structure that crystals possess. Optical
fibers and planar optical waveguides are based on glass.
A very thin layer of silica, about 1 nm thick, forms naturally on top of silicon wafers through
thermal oxidation of the silicon. The oxide formed this way is called native oxide. PIC devices
require much thicker layers of silica; these are obtained through controlled oxidation of silicon
at high temperature, between 600ºC and 1200ºC.
Silica is an electrical insulator—a material that does not conduct electricity. One of its
applications is to electrically isolate various parts of electronic and photonic integrated circuit
devices. On the other hand, glass has excellent optical properties. It is transparent for a wide
range of wavelengths in the visible and infrared portions of the spectrum. Figure 4-1 shows a
representative percent transmission vs. wavelength variation for fused silica. Transmission
above 90% is observed for the visible range as well as the optical communications windows
around the wavelengths of 1.31 µm (1310 nm) and 1.55 µm (1550 nm).
Figure 4-2 shows the variation of the index of refraction of silica with wavelength, based on
experimental measurements. At the communication wavelength of 1.55 µm, the index of
1.55
Index of refraction, n 1.5
1.45
1.4
1.35
0 0.5 1 1.5 2 2.5 3 3.5 4
Wavelength (µm)
Silica optical waveguides use pure silica for the cladding and doped silica for the core. The core
needs a higher refractive index than the cladding (recall the conditions necessary for total
internal reflection, discussed in Module 1). This can be obtained by introducing small amounts
of certain materials (dopants) in the silica in a process called doping. Materials that can be used
as dopants to increase the refractive index of silica are germanium dioxide (GeO2), titanium
dioxide (TiO2) and phosphorus pentoxide (P2O5). It is also possible to decrease the index of
silica by adding boron oxide (B2O3) or fluorine (F). Figure 4-3 shows representative variations
of the index of refraction vs. wavelength for silica and doped silica materials. The
concentrations of the dopants are shown in the legend. For each material, the remainder to 100%
is SiO2. By varying the concentration of the dopants, the refractive index can be adjusted
throughout a range of values.
1.6
Silica
13.5 % GeO2
1.55 9.1% P2O5
Index of refraction, n
13.3% B2O3
1% F
1.5
1.45
1.4
1.35
0 0.5 1 1.5 2 2.5 3 3.5 4
Wavelength (µm)
Figure 4-3 Index of refraction of silica and doped silica vs. wavelength
Lithium Niobate
Lithium niobate is a synthetic material with applications in a wide range of fields including
radar, microwave communications, optical communications, and others. Its chemical formula is
LiNbO3. The material is transparent for a wide range of wavelengths, from 350 nm to 5200 nm.
Figure 4-4 shows the percent of external transmission vs. wavelength of LiNbO3.
Lithium niobate is a crystal; its constituent atoms are arranged in an ordered, periodic structure.
The crystal structure, shown in Figure 4-5, is not symmetric. Like other crystals lacking
symmetry, lithium niobate is birefringent. This means that two different indices of refraction are
associated with it. The two indices are called the ordinary index, no, and the extraordinary
index, ne.
2.8
2.7
no ne
Index of refraction, n
2.6
2.5
2.4
2.3
2.2
2.1
2
0 0.5 1 1.5 2 2.5 3 3.5 4
Wavelength (µm)
Figure 4-6 Ordinary and extraordinary indices of refraction of lithium niobate vs. wavelength
Two main methods are used to create an optical waveguide using lithium niobate. In the first
method, a stripe of titanium is deposited on the lithium niobate substrate. The materials are
heated at high temperature (900ºC), which facilitates the diffusion of titanium ions into the
lithium niobate. The addition of titanium raises the index of refraction in the area with diffused
ions. This area will act as the core of the waveguide, and the undoped lithium niobate will act as
the cladding. Using this method, both the ordinary and the extraordinary indices of refraction
increase by the addition of the titanium ions.
The second method is called the annealed proton exchange (APE) process. This process can
take place at lower temperatures than titanium diffusion, typically 120ºC–250ºC. The material is
immersed in an acid bath in which protons from the bath exchange places with lithium ions
from the LiNbO3. The exchange is followed by annealing at 250ºC–400ºC to stabilize the index
of refraction. In this method, only the extraordinary index of refraction increases in the area
where the exchange has taken place. The ordinary index of refraction remains the same. The
waveguides obtained this way only work with one polarization of the light, TM. Both methods
produce moderate refractive index increases in the core region, typically less than 0.1 at optical
communication wavelengths.
Figure 4-7 Percent transmission and index of refraction of SU-8 vs. wavelength
Polymers have an interesting behavior relative to temperature variation. The index of refraction
of polymers depends primarily on packing density. When temperature increases, the density
goes down, and this results in a decrease in the polymer’s index of refraction. The coefficient of
variation, dn/dt, will thus be negative. Typical values of dn/dt for polymers range between
–1 x 10-4 K-1 to –3 x 10-4 K-1. The materials described previously, including silicon, silica, and
III-V semiconductors, all had positive dn/dt values. In addition to being negative, the dn/dt of
polymers is also large in absolute value. This allows for the construction of specific PICs that
cannot easily be realized in the other materials.
The waveguides can be characterized by the refractive index contrast (Δ) between core and
cladding, given by equations 4-1 and 4-2. Initially, waveguides were constructed with small
delta (Δ), similar to the delta of optical fibers. Those waveguides had very large cross sections
Example 1
Calculate the core refractive index of a high delta silica waveguide. Assume the wavelength
is 1.55 µm and the cladding index is 1.444.
Solution: High delta silica waveguides have Δ = 0.75%. Use equation 4-2 to solve for
ncore.
Δ = (ncore – nclad)/ncore
ncore = nclad/(1 - Δ) = 1.444/(1 – 0.0075) = 1.455
In Example 1, we arrived at a useful equation for solving for the core index when the cladding
index and refractive index contrast are known. This equation can only be used when Δ has small
values, about 1% or less.
ncore = nclad/(1 - Δ) (4-3)
The high delta waveguides have a 6 µm x 6 µm cross section. This is much larger than the cross
section of both silicon-on-insulator (SOI) waveguides and III-V semiconductor waveguides. A
large cross section is advantageous for butt-coupling planar waveguides to single-mode optical
fibers. However, the drawback of Δ = 0.75% silica waveguides is that they require a minimum
bend radius of 5 mm. This is about three orders of magnitude larger than the bend radius of SOI
waveguides (10 µm) and one order of magnitude larger than the bend radius of III-V
semiconductor waveguides (300 µm). The consequence of the larger bend radius of silica
waveguides is larger PIC devices.
Super-high delta waveguides (Δ = 1.5%) were introduced as a way to decrease the size of PICs
using silica-on-silicon waveguides. The refractive index contrast of a super-high delta
waveguide is double that of a high delta waveguide. The super-high delta waveguides have
smaller cross sections, about 4.5 µm x 4.5 µm, and the bend radius is reduced to 2 mm. This is
still very large compared with SOI and III-V waveguides, but it does allow for a good reduction
in the size of PIC devices compared to PIC devices with high-delta waveguides (Δ = 0.75%).
Table 4-1. Waveguide Parameters for a Single-mode, Buried Channel Silica Waveguide
Parameter Value
Thickness, H 6 µm
Width, W 6 µm
Index of refraction of GeO2 doped silica at 1550 nm and 1.455
room temperature, ncore
Index of refraction of silica at 1550 nm, and room 1.444
temperature, nclad
Effective index TE mode, neff TE 1.450
Effective index TM mode, neff TM 1.450
Waveguide birefringence negligible
The transmission vs. wavelength of all 40 channels for the two AWGs is shown in Figure 4-10.
Each channel is represented by a different color. Both figures illustrate the excellent uniformity
across channels and the low level of background noise characteristic of a well-controlled
fabrication process.
Example 2
Calculate the change in the central wavelength of an AWG when the temperature rises from
room temperature, 21˚C, to 55˚C. Is this change acceptable in light of the required
wavelength accuracy given in Table 4-2? The effective index of the silica waveguides at
room temperature is 1.45, the grating order of the AWG is 45, and the length difference ΔL
is 48.1 μm.
Solution: Using Equation 4-4, calculate the central wavelength at room temperature.
λc (21˚C) = neff ΔL/m = 1.45 x 48.1 μm/45 = 1.55 μm
At higher temperature, the effective index increases according to equation 2-4 in Module 2.
n(T2) = n(T1) + dn/dt *(T2 – T1) = 1.45 + 1.1 x 10-4 x (55 - 21) = 1.454
Using equation 4-4 one more time, calculate the central wavelength at the high temperature.
λc (55˚C) = neff ΔL/m = 1.454 x 48.1 μm/45 = 1.554 μm
The change in wavelength is given by 1.554 – 1.55 = 0.004 μm = 4 nm.
This change is much higher than the required accuracy of ±0.02 nm.
Example 4-2 shows the effect of temperature variation on the waveguide effective index, which
results in a change in the central wavelength of the AWG. Other thermal effects include the
expansion of the substrate and silica layers with increasing temperature. This changes the
waveguide dimensions, resulting in a change in ΔL. It also changes the stress state of the
waveguide, due to the different coefficients of thermal expansion of silicon and silica, as
explained previously. Due to the influence of these effects on the central wavelength, the
temperature of the AWG must be kept constant using an external device, an RTD (resistive
temperature-sensing device) or a thermistor. The drawback of these elements is that they
increase the device’s power consumption. An elegant solution that does not require additional
components to keep the temperature constant is the athermal AWG, described below.
Athermal AWG
An athermal AWG is a device capable of maintaining an almost constant central wavelength
(within some very tight limits) when the ambient temperature changes without the use of
external elements. Such AWGs are commercially available and use one of two primary
approaches to maintain wavelength. One approach is through mechanical control, and the other
is through control of the refractive index. The mechanical control method, illustrated in Figure
4-11, relies on inserting a thin copper plate in the first free propagation region (FPR). When the
copper plate heats, it expands and produces a movement of the output waveguides coming out
of the second FPR. When correctly designed, the center output waveguide will move to the
In the second approach, trenches are formed in the first FPR that will be filled with a resin.
Alternatively, trenches or grooves filled with resin are placed in the arrayed waveguide region.
The two options are shown in Figure 4-12. The resin (a polymer) has a refractive index that
decreases as temperature increases. Thus the effect of the polymer is to compensate for the
increase in effective index in the doped silica portion of the waveguides. Moreover, the
coefficient of variation with temperature of the polymer index of refraction is much higher than
the silica coefficient. So only a short length of resin will be needed to compensate for the
variation of the index of silica with temperature.
The methods presented above to athermalize the AWG require additional fabrication steps.
However, these methods result in very good stability of the central wavelength, ±0.04 nm or
better, for a wide temperature range from –5 ˚C to 65 ˚C.
AWGs can be further integrated with other PICs to form devices with increased functionality.
Ideally, all PICs are integrated in the same chip, which eliminates the need to use optical fibers
VOA/Multiplexer
When AWGs are used as multiplexers (MUXs), they combine multiple wavelengths and send
them into an optical fiber. It is often desirable to control the power level of each channel to
make it possible to equalize all powers or for other purposes. This can be achieved by inserting
a variable optical attenuator (VOA) in each channel. For a 40 channel AWG, an array of 40
VOAs is used either as a separate chip or in the same chip with the AWG.
The VOA’s function is to adjust the power of the light wave going through it. Power modulation
can be obtained with a Mach-Zehnder interferometer device, as discussed in Module 2. The
device is based on the thermo-optic effect, which is the only one available in silica, since the
electro-optic effect is absent. One arm of the interferometer will have a thin film heater
deposited on top of it to heat the waveguide below. This will change the index of refraction and
create a phase difference compared with the reference arm. The thermo-optic effect works well
for VOA function, but it has the drawback of being slow. It takes a couple of milliseconds for
the heat to propagate to the waveguide and achieve the desired phase difference. By contrast, the
electro-optic effect in semiconductors is very fast, on the order of nanoseconds and below. The
latter effect allows for the construction of electro-optic modulators used for the modulation of
high data rate signals.
The main parameters of VOA are insertion loss, dynamic range (or maximum attenuation), and
polarization dependent loss (PDL). Silica VOAs have insertion loss around 2 dB, dynamic range
of 20 dB, and PDL around 0.5 dB.
The VOA/MUX combination, called a VMUX, can be further enhanced with the option of
power monitoring in each channel. To measure the optical power in a channel, a small fraction
of the light wave is deviated into a secondary waveguide and then coupled to a photodetector.
This is achieved using a tap, which is usually a directional coupler with a 95%–5% splitting
ratio. A tap and photodetector are placed after the VOA in each channel. In this case, 5% of the
power is tapped and measured by the photodetector, while the rest continues to the multiplexer.
Another tap and photodetector can be placed at the output of the AWG multiplexer to monitor
the total power. In this case, the amount of power used for the tap can be even lower, for
example 2%.
Example 3
Calculate the length of the coupling region, L, for a directional coupler that splits the
incoming light in a ratio of 95%–5% between output waveguides 1 and 2. Assume the
coupling coefficient k is known.
Solution: The equations for the powers in outputs 1 and 2 of a direction coupler are:
P1(L) = P0 cos2(kL) and P2(L) = P0 sin2(kL).
The VMUX with optical power monitoring allows for very precise control of the power in each
channel. The measurement of the actual power in each channel provides a feedback mechanism
and transforms the system into a closed-loop control system. By contrast, the VMUX without
power monitoring is an open-loop system, where the desired power level is set by the VOA but
there is no mechanism to check if the actual power is equal to the desired one.
Figure 4-13 shows a schematic representation of the open-loop and closed-loop VMUX devices.
Figure 4-13 a) VMUX device Figure 4-13 b) VMUX with optical power
monitoring
Polymer Waveguides
The process of creating a polymer waveguide is very cost effective. A layer of polymer that will
act as the waveguide lower cladding is spin coated on the substrate, followed by baking and UV
curing. The core layer is then spin coated, baked, and UV cured. Patterning through
photolithography will define the circuit paths. Direct laser writing can also be used for this
purpose. The upper cladding layer is then added through the same process as the lower cladding.
Polymer waveguides can be obtained on various substrates, including glass, quartz, silicon, and
others. A typical polymer waveguide on silicon substrate is a buried channel waveguide with
polymer cladding and a polymer core with higher refractive index than the cladding, as shown in
Figure 4-17. To obtain the higher index for the core, polymers with different indices can be
blended together. This versatile method can produce a wide range of well controlled values of
the refractive index.
Similar to silica-on-silicon waveguides, polymer waveguides can be obtained with the typical
values of 0.75% and 1.5% for the refractive index contrast. For these values, the waveguide
cores have cross sections of 6 µm x 6 µm and 4.5 µm x 4.5 µm. The birefringence of the
waveguides is very low, and the propagation loss can be as low as 0.1 dB/cm. Bend radii are
between 2 mm and 5 mm. The advantage of some polymer devices is that they can be
constructed with very few bends, thus occupying less space than devices with the same functions
made of other materials.
Waveguide Interconnects
Polymer waveguides are an efficient and cost-effective solution for high bit rate interconnections
of electronic and optical components and modules. High-speed interconnections based on copper
suffer from high losses and electromagnetic interference. Optical interconnections using fibers
require complex assembly operations and occupy large volumes due to the fiber requirements for
minimum bend radius. Polymer waveguides can eliminate these issues and are less expensive
than other types of waveguides. Using a polymer for interconnection allows for close integration
of electrical and optical functions. An example of polymer waveguide interconnections is shown
in Figure 4-18.
A polymer used in waveguide interconnections is silicone, a cross between glasses and organic
polymers. Silicone (not to be confused with the chemical element silicon) shares the thermal
stability and transparency of glass, the ease of fabrication and tailoring of polymers, and the
optical properties of organic polymers. This makes it attractive for optical waveguide
applications. Silicone can be deposited on a flexible substrate, as shown in Figure 4-19, and
waveguide sheets can be stacked on top of one another. The structure shown consists of eight
stacked sheets.
Figure 4-19 Silicone waveguides for optical interconnects. Courtesy of Dow Corning.
Figure 4-20b shows the VOA transmission vs. the electrical power applied to the heater at the
two optical communications wavelengths 1.31 µm and 1.55 µm. Very high values of attenuation,
up to 40 dB, can be achieved. The electrical power used by such a VOA is considerably smaller
than the power needed by silica-on-silicon MZI-based VOAs. The latter devices require about
250 mW of electrical power to switch off the optical power. The excellent performance of the
simpler polymer VOA is made possible by the large index change with temperature of these
materials. For example, a large change of 0.01 in the refractive index is possible with a polymer
having dn/dt = -2 x 10-4 K-1 by a change in temperature of 50ºC.
A drawback of the DOS is that electrical power has to be applied at all times, on one branch of
the Y-splitter or the other, to have light exit through one of the outputs. Different configurations
have been designed to allow one path to be achieved with zero electric power applied to the
heaters, with the other path requiring the application of electric power. Such a configuration is
presented in Figure 4-22. In this configuration, the bar states, I-1 to O-1 and I-2 to O-2, do not
require electric power to the heaters. Electric power is applied only when the paths I-1 to O-2 or
I-2 to O-1 are desired.
Figure 4-23 a) Laser configuration of a Figure 4-23 b) Output power emitted by the
tunable wavelength laser with a polymer Bragg tunable wavelength laser with a polymer Bragg
grating grating
Tunable lasers such as the one described above are very attractive for optical communication
systems using wavelength division multiplexing. A system with 40 channels that uses diode
lasers emitting at fixed wavelengths requires 40 different lasers. If tunable lasers are used, the
same laser can be used throughout the system and the wavelength can be easily adjusted by
applying power to the Bragg grating.
This material was created under Grant # 1303732 from the Advanced Technological Education
division of the National Science Foundation. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the author(s) and do not necessarily
reflect the views of the National Science Foundation.
Dan Hull
PI, Executive Director, OP-TEC
316 Kelly Drive
Waco, TX 76710
(254) 751-9000
hull@op-tec.org
Taylor Jeffrey
Curriculum Development Engineer
316 Kelly Drive
Waco, TX 76710
(254) 751-9000
tjeffrey@op-tec.org
ISBN 978-0-9903125-9-8
CONTENTS OF MODULE 5
Introduction .................................................................................................................................... 1
Prerequisites ................................................................................................................................... 1
Objectives ....................................................................................................................................... 1
Basic Concepts ............................................................................................................................... 3
PIC Subsystems Used in Telecommunications .......................................................................... 3
Wavelength Division Multiplexing Transmitter and Receiver Based on III-V
Semiconductors ...................................................................................................................... 3
Reconfigurable Optical Add/Drop Multiplexer (ROADM) Based on Silica-on-Silicon and
Polymer Waveguides .............................................................................................................. 7
PIC Sub-Systems Used in Data Communications.................................................................... 14
Optical Transceiver Based on Silicon Photonics (SiP) ........................................................ 14
PICs Used in Sensors ............................................................................................................... 16
Waveguide-Based Photonic Biosensors and Lab-on-a-Chip Devices ................................. 16
Micro-Opto-Electro-Mechanical Systems (MOEMS) ............................................................. 20
MEMS and MOEMS ............................................................................................................ 20
Summary ...................................................................................................................................... 23
Problem Exercises and Questions ................................................................................................ 25
References .................................................................................................................................... 26
I NTEGRATED P HOTONICS C OURSE FOR T ECHNICIANS
Module 5
Integrated Photonics Circuits
and Systems
INTRODUCTION
Previous modules introduced planar optical waveguides and fundamental PIC devices based on
the material platforms silicon-on-insulator, III-V semiconductors, silica-on-silicon, lithium
niobate, and polymers. This final module presents subsystems obtained by combining several
PIC devices. These subsystems have applications in telecommunications, data center processing,
sensing and other areas. The level of integration of multiple functions in the same chip in these
subsystems varies depending on the material platform used and the number of functions. Some
are monolithically integrated, while others are based on separate chips connected by optical
fibers or directly attached to each other. PIC systems are expected to continue grow at a very
high rate because they offer significant improvements in size, power consumption, reliability,
and cost.
PREREQUISITES
OP-TEC’s Fundamentals of Light and Lasers Course
OP-TEC’s Integrated Photonics: Modules 1–4
OBJECTIVES
Upon completion of this module, the student should be able to:
Describe subsystems of integrated PIC devices used in telecommunications and data
communications:
o Wavelength division multiplexing transmitter and receiver
o Reconfigurable optical add/drop multiplexer (ROADM)
o Transceiver
Describe PICs used for sensing applications:
o Biosensors
o Lab-on-a-chip applications
1
Discuss micro-electro-mechanical systems (MEMS), micro-opto-electro-mechanical
systems (MOEMS), and their applications
Discuss advantages offered by PIC devices and systems
Current optical transmission systems send multiple data streams, each modulating a different
wavelength of light, through the same optical fiber. This method is called wavelength division
multiplexing (WDM), with a subset called dense wavelength division multiplexing (DWDM) in
which the frequency separation between channels is on the order of 100 GHz. With this
approach, a much larger amount of information can be transmitted simultaneously. For such a
system, the transmitter contains multiple light sources, each controlled by its own driver, as well
as a multiplexing device that combines the different channels together before launching them in
the optical fiber.
While the multiplexer and the light sources can be separate chips connected to each other by
optical fibers, it is more advantageous to integrate all the transmitter functions for DWDM
systems onto a single PIC. Monolithic integration of the transmitter functions can only be
achieved with III-V semiconductors, because these materials are the only ones that have
efficient light emitting properties. One example of an advanced transmitter PIC is the indium
phosphide based TX LS-PIC (Transmitter Large-Scale Photonic Integrated Circuit),
commercialized by Infinera Corporation. The device aggregates 10 channels, each operating at
10 Gb/s, thus achieving a total transmission rate of 100 Gb/s. A schematic of the chip
architecture is shown in Figure 5-2.
Five different types of fundamental devices are used in the TX LS-PIC: tunable distributed
feedback (DFB) diode lasers, electro-absorption modulators (EAM), variable optical attenuators
(VOA), optical power monitors (OPM), and arrayed waveguide grating (AWG) multiplexer.
The principles of operation of these devices were described in previous modules. The TX PIC
thus integrates 50 optical functions on the same chip.
The DFBs emit a narrow spectrum of light around a central wavelength that can be tuned over a
300 GHz range. The electro-absorption modulators achieve the high-speed modulation of the
light emitted by the diode lasers needed to encode the 10 Gb/s data signals onto the light
carriers. The AWG combines the ten signals into one optical output that is sent through the
optical fiber. Both the DFBs and the AWG are thermally tuned so that their channels are aligned
to each other and to the ITU grid. The channel spacing of the AWG is 200 GHz. Figure 5-3
shows the spectrum of the output light from the transmitter chip.
Complementing the essential light emission, modulation, and multiplexing functions of a WDM
transmitter, two additional functions enhance its operation. These are power attenuation and
power monitoring. The attenuation provided by the VOAs allows control over the output power
profile across all channels. It is important for the performance of the entire optical transmission
system to have the same optical power in all ten channels. The VOAs can compensate for small
power differences between the output powers of the DFB lasers. Figure 5-4 shows the output
spectrum of a transmitter PIC in which the VOAs were used to equalize the power level in the
ten channels. The output power of the lasers is monitored using the OPMs, which are
implemented by monolithically integrating photodiodes into the back of each channel. The
OPMs perform optical power monitoring of the diode lasers over the lifetime of the chip.
Figure 5-4 Spectrum of the output light from transmitter chip with equal power in
the ten channels. Courtesy of Infinera
The TX LS-PIC is capable of transmitting data at the rate of 100 Gb/s with excellent
performance over a total 375 km of optical fiber. This distance was covered in five spans of
75 km each, with an erbium doped fiber amplifier (EDFA) inserted after each span. The
performance described above requires a tightly controlled PIC manufacturing capable of
producing waveguides with precise dimensions (thickness, width and waveguide spacing) and
low propagation loss. Such manufacturing processes are available today for indium phosphide
PICs.
At the other end of the transmission system, a receiver converts the optical signal into an
electrical signal that is further processed to recover the transmitted information. A basic receiver
configuration for a receiver working with one optical wavelength is shown in Figure 5-5. The
electric current at the output of the photodiode passes through the transimpedance amplifier
(TIA) and the post amplifier for noise filtering and amplification.
In WDM systems, the multiple light waves of different wavelengths traveling simultaneously
through the optical fiber are first separated using a demultiplexer and then converted by
Figure 5-6 Spectrum of the receiver chip output showing the ten channels separated by 200 GHz.
Courtesy of Infinera
When light reaches the end of transmission, its state of polarization is unknown. For this reason,
the AWG demultiplexer in the receiver must be polarization independent. The polarization
dependence of an AWG is described by the polarization dependent loss (PDL) parameter
described in Module 3. The AWG used in the receiver chip of our example device has a PDL
less than 0.4 dB in all ten channels. The adjacent channel crosstalk is better than -25 dB for all
channels. The high-speed PIN photodiodes have low dark current values, below 10 nA. As in
the case of the transmitter chip, this level of performance for the receiver is possible due to the
excellent control of the manufacturing process of active and passive indium phosphide
waveguides.
Since the transmitter and receiver subsystems described above became available commercially
several years ago, Infinera has continued to integrate more functions on a single indium
phosphide chip, recently reaching 600 total functions.
The advantages of such a ROADM are the capability to drop or add all N channels, its
simplicity, and its low cost. However, there are also some disadvantages associated with such a
configuration. Primarily, each Add and Drop port corresponds to a fixed wavelength, dictated
by its position relative to the AWG. Manual operations are required to connect a certain
wavelength to its specific position.
A low-power-consumption, compact, type I PLC 40 channel ROADM is produced
commercially by the company Enablence. To optimize the power consumption, the subsystem
uses silica-on-silicon, athermal AWGs that do not require temperature stabilization devices and
polymer-based thermo-optic switches and VOAs. The athermal AWGs are flat-top, wide-
bandwidth devices for which the center wavelength is maintained within a range of ±0.3pm/°C
when temperature varies from -30°C to 70°C. The 1 dB bandwidth of the AWGs is 0.48 nm, and
the 3dB bandwidth is 0.65 nm. Figure 5-8 shows the transmission for one of the AWG channels
for three different temperatures. No changes are seen in the center wavelength or the filter shape
as the temperature changes from the minimum to the maximum value.
The second low-power technology used to construct the ROADM is based on the use of
polymer materials for the thermo-optic switches and VOAs. A basic device configuration used
for thermo-optic switches and VOAs is the Mach-Zehnder interferometer (MZI). MZIs can be
constructed from all materials used for PICs, including silicon, III-V semiconductors, silica,
polymers, and lithium niobate. Due to the high thermo-optic coefficient of polymer materials, it
is possible to obtain a very low-power MZI VOA using these materials. The Enablence VOAs
use only 1.4 mW to achieve an attenuation of 30 dB and have excellent polarization dependent
loss (PDL) performance of less than 0.2 dB for the entire attenuation range. For switches,
Enablence is using a different configuration than MZI. Digital optical 1 x 2 switches were
described in Module 4; these devices have a power consumption of about 3 mW, with an
isolation in the off state of -40 dB. Using these technologies, the Enablence ROADM achieves a
worst-case low power consumption of 5W for both the optical and electronic components.
The arrays of polymer switches and VOAs are integrated together in one small chip. Because
the polymers have a high thermo-optic coefficient as well as low thermal conductivity, it is
possible to have dense arrays of switches and VOAs. The components can be placed close to
each other, resulting in a very compact chip. Photodiodes are attached to this chip through flip-
chip bonding, avoiding the used of optical fibers in between. For the light to be redirected at 90°
toward the photodiodes mounted on top of the chip, 45° mirrors are created in the polymer
material by excimer laser ablation followed by metallization of the mirror surface.
The different polymer and silica AWG chips are attached to each other directly through chip-to-
chip bonding. This method is similar to the one used to attach a fiber array to a silica-on-silicon
chip. The advantages of chip-to-chip bonding are cost reduction, achieved by eliminating the
fiber arrays to connect different chips; space reduction, since the fibers require a large minimum
bend radius; reduced optical power loss at the interface between fibers and planar waveguides;
elimination of the time and expense associated with active alignment of fibers to chip; and
increased reliability.
Figure 5-9 shows a subassembly of the Type I ROADM consisting of the chip-to-chip assembly
of a silica AWG chip and polymer switch/VOA array.
The ROADM described above has very good performance, with 7 dB total loss for the Express
path, VOA dynamic range of 20 dB, channel-to-channel crosstalk of 50 dB, and switch
extinction of 50 dB. Another important specification for ROADMs is related to the cascadability
of these devices. With each pass through an optical filter such as an AWG, the bandwidth
becomes a little bit smaller. If the bandwidth gets reduced too much, then the signal will be
distorted at the end of the transmission. The ROADM described above allows for cascading 16
subsystems in a ring network before performance degradation is observed.
Type I ROADMs are successfully used in ring networks. However other network
configurations, such as mesh networks, which have multiple connections between nodes, require
higher functionality for ROADMs. Multi-degree ROADMs are needed to provide the capability
to switch between multiple Express paths as well as Add/Drop. In addition, the add/drop
function should be colorless, directionless, and contentionless. Colorless means that the
ROADM should be capable of adding or dropping any wavelength to any port, not just the fixed
ones that are possible with the Type I sub-system. Directionless means that the wavelength
could come from any direction, while contentionless means that the same wavelength, coming
from multiple directions, can be dropped at one port.
These requirements can be described by an M x N ROADM that has M inputs and N outputs or,
in other words, M directions and N drop (or add) ports. A much more complex configuration is
needed to realize such a ROADM. While different architectures exist to implement it, we again
present a solution offered by Enablence based on silica-on-silicon AWGs and polymer switches,
VOAs, and tap arrays. The configuration of an M x N ROADM operating with X wavelengths is
illustrated in Figure 5-10. The number of components is much higher for this configuration, and
some new components are included. The configuration requires M demultiplexer AWGs at the
input and N multiplexer AWGs at the output. An M x N switch is required for each of the X
wavelengths. The X switches are represented by the block of rectangles in the middle of the
figure. Finally, two waveguide shuffling components are inserted between the AWGs and
switches. The first waveguide shuffling component reorders the M x X signals such that
wavelength λ1 from all the M demultiplexers go to switch number 1, wavelength λ2 from all the
M demultiplexers go to switch number 2, and so on. On the output side, the second waveguide
Figure 5-11 Super high delta silica-on-silicon chip integrating eight AWGs
The forty 8 x 8 switches are implemented using polymer waveguides, as described in our
discussion of the Type I ROADM subsystem. The chip containing the switches also has VOAs
for power balancing and taps and photodetectors for power monitoring for each output. The
8 x 8 switches are realized by combining 112 1 x 2 switches. Each 1 x 2 switch is a Y-branch
digital optical switch with heaters on both arms that uses only 7.5 mW of electrical power. The
polymer waveguides also have a refractive index contrast of 1.5%, making the chip compact and
the interfacing with the silica-on-silicon waveguides efficient. The area occupied by the
Figure 5-12 Super-high delta polymer chip integrating switches, VOAs, and taps for photodetectors
The third component is the waveguide shuffling chip. This chip consists of polymer waveguides
that connect a specific input to a specific output. The simplest solution is to use bends to route
the waveguides in the desired fashion. This solution, however, results in a large number of
waveguide crossings. When two optical waveguides cross each other at a large angle (close to
90°), there is very little crosstalk, meaning that a negligible amount of light gets transferred
from one waveguide to the other. The crosstalk is equivalent to a small loss of light in each
waveguide. This small loss can be considered negligible when there are only a few crossings in
the path of a waveguide. However, in the case of the 8 x 8, 40 channel ROADM, there are 320
waveguides that need to be shuffled (8 inputs x 40 channels). The number of crossings in this
case is very large and creates a non-negligible loss of optical power. An alternate solution is to
use an interposer polymer chip, which sits on top of and is flip-chip bonded to the main
shuffling chip. Light is directed to the interposer and back to the main chip using 45° mirrors.
The interposer allows all the waveguides to be shuffled without any waveguide crossing. The
length of the interposer is less than 1 cm.
Once again, the silica and polymer chips are chip-to-chip bonded together. Together, the chips
occupy a space of only 9.8 cm x 9.0 cm. The overall power consumption is 14.7 W, and the total
insertion loss is 18.6 dB, both of which are comparable to those of ROADMs based on other
technologies. The small size, low power consumption, very good optical performance, and high
reliability of PLC components result in a ROADM subsystem that competes very well with
those fabricated using other technologies.
The optical path difference between the arms of each MZI is critical in achieving a low level of
crosstalk between channels, as in the case of AWGs. Variations in the manufacturing process
can result in effective index differences between the arms that affects the path difference of the
interferometers. Temperature variations also influence the path length difference, due to
silicon’s large thermo-optic coefficient (about 2 x 10-4 K-1). These unwanted effects are
compensated by introducing a resistor in one arm of each interferometer. Such resistors are
implanted in the waveguides. When heated, the effective index of the waveguide is controlled to
achieve the necessary path length difference between the arms.
The measured transmission of a tuned interleaver is shown in Figure 5-14. Crosstalk values
better than 20 dB are obtained for each of the four channels.
The transmission path ends with the output of the multiplexer being coupled out of the chip
using another grating coupler. The receiver portion of the chip starts with light with the four
multiplexed wavelengths being coupled from the receiving fiber to the chip by a grating coupler.
Light is guided to an interleaver similar to the one above but working as a demultiplexer in this
case. The four channels are detected by high-speed germanium photodiodes that are
Figure 5-15 Microscope view of monolithically integrated transceiver chip. TX indicates transmitters;
RX indicates receivers. Courtesy of Luxtera.
The propagation of light through the multimode waveguide is based on mode interference. In
Module 3, multimode interference (MMI) couplers were described. They split the optical power
between two or more output waveguides. In the current application, the interference of the two
modes produces a specific pattern for the distribution of the electric field at the end of the
bimodal waveguide, as shown in the figure. This pattern is related to the change in the refractive
index that takes place when the substance under test flows above the waveguide. A photodiode
with two sections is used to detect the pattern. The TSP produces two outputs, Vup and Vdown.
From these two voltages, a combined output is calculated using the following equation:
The value of the parameter S shows whether there is more power in the top half or bottom half
of a photodiode. If there is equal power in both, S will be 0. In experimental measurements with
a 1 cm long bimodal waveguide with sensing area 3 mm long, the parameter S varied between 0
and 0.96. The ideal range for S is 0 to 1.
The configuration of the entire sensor is shown on the left side of Figure 5-18. The image on th
e right shows a chip with five sensors and grating couplers coupling the light to the chip. The
gratings are written directly onto the waveguides. They have a depth of 20–40 nm, periods
between 300 nm and 500 nm, and length up to 100 µm.
Figure 5-18 Left: bimodal waveguide based photonic biosensor. Right: five-sensor chip with light
coupled to the chip through gratings.
As mentioned above, the substance under test must flow above the waveguide in order to
interact with the evanescent field of the waveguide. Microfluidic channels have to be created in
the chip for this purpose, and their method of fabrication must be compatible with the
fabrication process of the waveguide sensor. The group that developed the bimodal waveguide
sensor created channels with height of 20–100 µm and width of 50–150 µm in a biocompatible
polymer (SU-8) layer above the waveguides. The microfluidic channels were perfectly sealed.
The sensor was capable of detecting index changes as low as 3.3 x 10-7 m and is configured
much more simply than the MZI based sensor. The group chose this configuration to continue
developing a complete lab-on-a-chip device.
A true LOC device needs to integrate the following components into one chip: photonic sensors,
flow cells and flow delivery system, light sources and photodetector array or miniaturized CCD
camera, processing electronics, and the final package with required firmware and software. This
ultimate integration is illustrated in Figure 5-19. While some of these elements have been
integrated in the same chip, others, such as sources, photodetectors, and electronics, have not yet
been integrated. Silicon photonics has been the key to integration in the past, and it may also be
the key to achieving the ultimate integration for a lab-on-a-chip device.
As with the thin membrane, we can determine the applied force by measuring the beam’s
deflection. The device acts as a sensor in this case. If the cantilever beam is deliberately moved
under the action of the force, the element acts as an actuator.
A torsion plate is a thin plate of semiconductor material supported by small beams at two points.
The beams supporting the plate form torsional springs that allow the plate to rotate through an
angle without snapping off. The plate can be made to rotate by application of an electric or
magnetic field.
Optical elements can be integrated with micromechanical and electronic elements in the same
chip through a similar fabrication process as the one used for PICs. In this case, we have
MOEMS or micro-opto-electro-mechanical systems. The acronym MEMS often includes
MOEMS, since MEMS is easier to use. Examples of optical elements constructed this way are
microlenses, micromirrors, couplers, filters, and beam splitters. Microlenses with a cylindrical
shape have been fabricated directly on the emitting surface of a diode laser. The microlens
converts the elliptical shape of the beam emitted by the laser into an almost circular shape that
matches the mode of optical fibers much better.
The pressure sensor mentioned above uses a thin circular silicon membrane to sense the force or
pressure applied to the sensor. The sensor is fabricated by etching a shallow cavity of depth 53
μm and diameter 600 μm in a glass substrate. A layer of silicon is bonded on top of the cavity.
The thin membrane is formed by etching the silicon layer down to a thickness of 26 μm. The
deflection of the membrane can be sensed using an interferometer with an optical fiber carrying
light of wavelength λ = 850 nm acting as the light source. The sensor measures pressure in the
range 0 to 30 psi, with a maximum deflection of λ/8.
Silicon membranes can also be used to make deformable mirrors (DM). These are devices
capable of correcting aberrations in an optical beam. To achieve this, the shape of the membrane
must change from a flat surface to a surface with peaks and valleys in various locations. The
peaks and valleys are created by a controllable array of electrostatic parallel-plate actuators
placed under the silicon membrane. The device is illustrated in Figure 5-21.
A related application is the steerable micromirror array, with applications in digital display
technology and other imaging applications. The number of mirrors in the array can be as high as
2048 x 512, with each mirror independently controlled. An example of a 12-element
micromirror array is shown in Figure 5-22.
Cantilever beam-based MEMS can be used for optical switching applications. Numerous
configurations exist for such optical switches. Figure 5-23 illustrates one simple configuration
for an on-off switch.
A light wave of wavelength 633 nm emitted by a helium neon (HeNe) laser propagates in a
silicon oxynitride (SiOxNy) waveguide formed on a 2.5 cm silicon substrate. The refractive
index of SiOxNy depends on the relative concentration of oxygen and nitrogen in the chemical
formula. This material is used for both the core and cladding layers of the waveguide. The
indices of the core and cladding are 1.98 and 1.61 respectively, and the core dimensions are
approximately 2 μm x 2 μm. A thin film resistor is deposited on top of the cantilever beam.
When the heater is off, the cantilever is horizontal and the two waveguide segments are aligned,
allowing light to pass. The switch is in the ON state. When current is sent through the resistor,
the heated cantilever moves down due to the difference in thermal expansion of the materials. In
this case, the light path is blocked and the switch is OFF. The free end of the cantilever was
displaced a maximum distance of 30 μm when a 25 mA current was applied to the heater.
The cantilever beam is also the structure that accelerometers are based on. Most commonly, the
acceleration is measured by the difference between two electrical capacitances induced by the
motion of the cantilever arms. The acceleration can also be measured through optical means,
where the motion of the cantilever changes the distribution of optical power between two optical
waveguides. MEMS and MOEMS are versatile and reliable devices with numerous applications.
New devices with higher levels of integration and performance are expected in the future.
SUMMARY
This final module of Integrated Photonics built on information about the foundation PIC
devices previously introduced to present complex subsystems with applications in
telecommunications, data communications, and sensors. This module included almost all
material platforms from previous modules, including silicon-on-insulator, III-V semiconductors,
silica-on-silicon, and polymers. The high performance, highly integrated subsystems are made
possible by the continuous advances in PIC fabrication processes and technologies.
The five modules in this course are intended as an introduction to photonic integrated circuits
and their applications. While fundamental devices and systems were presented, the course
material is not exhaustive; many other devices and systems exist that could not be included due
to size and prerequisite knowledge limitations.
Constructive interference – when the result of wave interference and superposition is a wave
with the maximum possible amplitude. It occurs when the two waves are completely in phase
with each other. This is also called the phase-matching condition.
Core – a transparent medium through which light can propagate, usually surrounded by
cladding.
Coupling – a method of interconnecting two devices to transfer an optical signal, using light
waves.
Coupling coefficient – a non-dimensional, variable characteristic of couplers.
Coupling length – the length of the coupling region for which 100% of light transfers to a
second waveguide.
Critical angle – the largest angle of incidence for which refraction can still occur.
Cross state – when power input and power output occur in waveguides diagonally across from
each other in an MMI coupler.
Crosstalk – unwanted transfer of light between parallel waveguides.
Demultiplexing – a process that separates signals which have been combined using wavelength
division multiplexing (WDM).
Deposition – the process through which layers of materials are added to a silicon wafer.
Destructive interference – when the result of wave interference and superposition is a wave
with zero amplitude, which occurs when the two waves are completely out of phase with each
other.
Detector – a device that processes an incoming optical signal and converts it to an electrical
signal.
Die bonding – attaching the die to the frame of the package.
Diode laser – a laser that uses a semiconductor diode as the active medium.
Direct laser writing – a method of fabricating PICs by using a laser to create 2D or 3D patterns
inside a host material.
Directional coupler – a passive device with four ports and two parallel waveguides, used to
split or combine light.
Dispersion – pulse spreading during transmission, caused by the variation in index of refraction
with wavelength of the transmission medium.
Distributed Bragg reflectors – a light reflecting device (mirror) based on Bragg reflection.
Distributed feedback (DFB) lasers – a type of diode laser where the active region has periodic
sections of diffraction grating.
Doping – adding other materials to the core to change properties like refractive index.
Double heterostructure configuration – where a laser diode has a layer of low bandgap
material sandwiched between two layers of high bandgap material.
Effective refractive index – a number quantifying the phase delay per unit length in a
waveguide, relative to the phase delay in vacuum.
Efficiency – the ratio of total optical (or laser) output power to the pump power.
Electroabsorption – when change in a material’s absorption coefficient is caused by a
controlling signal, usually electric current.
Electro-absorption modulators (EAM) – modulators that change the absorption of a material
when an electric field is applied to it.
Electro-optic effect – a change in refractive index due to an electric field applied to a material.
Electrorefraction – when change in a material’s index of refraction is caused by a controlling
signal, usually electric current.
Etching – a chemical process used to remove portions of the core layer outside the waveguide
patterns.
Evanescent field –an oscillating electric and/or magnetic field which does not propagate as an
electromagnetic wave but whose energy is spatially concentrated in the vicinity of the source.
Extinction ratio (ER) – the ratio between the maximum output power (when modulator is on)
and the minimum output power (when modulator is off).
Fiber array – a line or array of fibers that have been placed in a grooved substrate and are used
for input and/or output to PIC dies.
Flip-chip – when the die at the bottom of a stack is turned upside down and connected to the
package using solder, eliminating the need for a wire bond from that die to the package.
Free propagation regions (FPR) – areas of the AWG with planar geometry where light enters
the AWG or exits the AWG.
Free spectral range (FSR) – the maximum wavelength range that can pass through an AWG
without repeating output waveguides.
Front-end process – the creation of PIC wafers that takes place entirely in a clean room
environment.
Fundamental mode – the mode corresponding with the greatest angle of incidence for light
traveling in the waveguide.
Gain parameter – the ratio of output power to input power, often used to describe
amplification.
III-V semiconductor materials – semiconductor compounds made of elements from group III
and group V on the periodic table.
III-V semiconductors – binary compound (two element) semiconductors made by combining
elements from Group III and Group V on the periodic table.
Indirect bandgap material – a material where photons cannot be directly emitted by electrons
due to the alternate momentum of electrons and holes in the conduction band and valence band.
Insertion loss – the ratio of the total power at the output of the device to the input power.
Interference – when two waves interact and their superposition forms a resultant wave that
varies between a minimum and maximum value.
ITU grid – wavelengths and frequencies used in wave division multiplexing that have been
standardized by the International Telecommunication Union (ITU).
Known good dies – dies that have been tested and proved functional.
Lateral taper – a taper which changes the waveguide width.
Lower cladding – the cladding layer below the core, also called the substrate.
Mach-Zehnder interferometer (MZI) – a device that uses one source of light split into two
waves that travel through different pathways before being brought back together and allowed to
interfere.
Metallization – the application of thin film heaters over certain regions of an optical waveguide
to modulate its refractive index.
Modes – the light waves corresponding to solutions of Maxwell’s equations, at which angles of
incidence light can propagate. Waveguides have a finite number of modes.
Modulation depth (MD) – a parameter that helps characterize modulators. It is the relationship
between minimum and maximum power out, and is equal to 1 in an ideal setting.
MD = (Poutmax-Poutmin)/Poutmax
Modulators – devices used to control beam power in single or multiple device outputs.
Molecular beam epitaxy – a method of thin-film deposition.
Molecular bonding – when two materials bond at the molecular level, due to sharing of
electrons in their outer shell.
Monolithic integration – the integration of all components and devices into a single chip or
base semiconductor.
Multimode Waveguide – allows for propagation of more than one mode.
Multimode Interference (MMI) coupler – a passive device that is a wide waveguide which
allows the propagation of several modes, and is based on interference between the waveguide
modes for coupling.
Multiple quantum well (MQW) laser – a laser where the active region includes thin layers of
semiconductor materials with alternating bandgaps, smaller than the bandgaps of the
surrounding material. MQWs are used to control the emitted wavelength of the laser.
Multiplexing – combining multiple signals.
N-type semiconductor – a material that has an excess of electron charge carriers, which are
free negative charge carriers.
Numerical aperture – a measure of the light-gathering ability of an optical waveguide. It is a
function of the refractive indices of the core and cladding.
Optical amplifier – a device that boosts the power of optical signals.
Optical fibers – a type of optical waveguide consisting of a cylindrical core surrounded by
cladding.
Optical Switches – devices used to control light going to multiple outputs, capable of being
“on” or “off”, with an output of either full power or zero power, with no values in between.
Optical waveguide – a physical structure that guides and confines light waves.
Passivation – a process that deposits a protective layer over the surface of a wafer.
Passive optical device – an optical component requiring no external power source, like a fiber
or lens that can transmit and/or alter an optical beam (signal).
Phonon – a quantum (or quasiparticle) of vibrational energy.
Photodetectors – devices that detect light.
Photolithography – a process used in microfabrication to pattern parts of a thin film of a
substrate, which can define the waveguide core geometry using light to transfer the pattern from
a photomask onto the substrate.
Photomask – a geometric pattern placed over a layer of photoresist in photolithography. Light
shines through the mask and transfers the geometric pattern to the photoresist layer.
Photonic integrated circuit (PIC) – a device that integrates multiple photonic or optical
components.
Photoresist – a material sensitive to light and used to form patterned coatings on a surface.
Planar optical waveguides – waveguides based on a rectangular geometry, with the core
having a square or rectangular shape.
Plasma dispersion effect – when the concentration of charge-carriers in a material changes,
causing changes in both refractive index and absorption coefficient.
PN junctions – when p-type and n-type semiconductors are joined to make a diode, where the
electrons jump from the n-material to the p-material and holes jump from the p-material to the
n-material.
Polarization maintaining fiber – a special type of fiber that allows only one polarization
(either TM or TE) to propagate.
Propagation loss – loss that light experiences as it travels through the waveguide, typically in
units of dB/cm.
P-type semiconductor – a material that has an excess of “hole” charge carriers, which are free
charge carriers, the same magnitude as an electron’s charge, but with a positive sign.
Receiver – a device that processes an incoming electrical or optical signal.
Refractive index contrast – also called the relative refractive index difference. It is the measure
of the relative difference in refractive index of the core and the cladding.
Regenerator – a device that converts an optical signal to an electrical signal and then back (also
called repeater).
Resonance – when one object (or wave) vibrating causes another object (or wave) to vibrate
(oscillate) at greater amplitudes for a predetermined frequency.
Resonant frequency – the natural frequency of an object at which resonance occurs. This
frequency is determined by the physical characteristics of the object.
Resonant wavelengths – wavelengths that satisfy the resonant condition and will be amplified
during resonance.
Ridge waveguide – a type of planar waveguide with comparable width and height, described by
the width, height, and thickness of the ridge layer of the core.
Ring resonator – a filter device that performs operations based on the wavelength of the input
light. The simplest set-ups utilize a ring waveguide and a straight waveguide.
Scattering loss – loss that can occur due to volume scattering, surface scattering, and sidewall
loss.
Self-imaging – In an MMI coupler, if the phase differences between all modes at the end of the
wide waveguide are multiples of 2π, the modes will interfere constructively and the final
distribution of the electric field will reproduce the initial distribution at the start of the wide
waveguide.
Semiconductor Optical Amplifier (SOA) – a device that creates gain in the active region of
the amplifier.
Silica on silicon – refers to devices fabricated from silica and doped silica, created on a silicon
wafer.
Silicon-on-insulator (SOI) – a layered silicon and insulator substrate.
Single-mode – a single-mode waveguide allows for propagation of only the fundamental mode.
Slab waveguide – a type of planar waveguide with a very thin core and a much larger width
than height, described by the core thickness.
System-in-Package (SiP) – when multiple chips are included in a single module (package).
Through Silicon via (TSV) – a vertical electrical connection (via) passing completely through a
silicon wafer or die. TSV’s are high performance interconnect techniques to create 3D ICs with
higher density.
Thermal TSVs (TTSVs) – TSVs that are used for heat dissipation and thermal management
rather than electrical connection.
Thermo-optic effect – a change in refractive index when heat is applied to a material.
Threshold current – the current in a laser diode at which stimulated emission begins.
Total internal reflection (TIR) – a phenomenon where light strikes a medium boundary at an
angle larger than the critical angle, and where the index of refraction of the traveling medium is
greater than the index refraction of the adjacent medium.
Transceivers – a device capable of transmitting and receiving electrical signals.
Transmission medium (optical) – the material used to transmit data, in the form of optical
signals, over a distance.
Transmitter – a device that impresses data in the form of an electrical signal onto a light
source.
Transverse electromagnetic wave (TEM) – a traveling light wave in which the electric field
and the magnetic field are perpendicular to each other and to the direction of propagation.
Upper cladding – the cladding layer above the core in an optical waveguide.
V parameter – a normalized frequency parameter used to determine if a fiber is single-mode or
multimode. Single-mode fibers have a V parameter less than 2.405.
Variable optical attenuator (VOA) – another name for a PIC modulator.
Vertical taper – a taper which changes the waveguide thickness.
Wavelength division multiplexing (WDM) – a process that combines multiple signals of
slightly different wavelengths and launches them in the same optical fiber to increase the
amount of information transmitted in an optical communication system. AWGs are an important
component for WDM.
Wavenumber – the spatial frequency of a wave, often expressed in radians per unit distance.
Wire bonding – a method using thin metal wires to make electrical connections to the die.
Y-branch coupler – splits an optical signal in one waveguide into two copies in different
waveguides.
Index
C
#
Cantilever Beam-Based MEM, Module 5 p.
2.5D packaging, Module 2 p. 43
22
3D packaging, Module 2 p. 42
Cantilever Beam, Module 5 pp. 19, 20, 21
A Channel waveguide, Module 1 pr. 15, 16
Coefficient of variation, Module 2 p. 6
Absorption coefficient, Module 3 pp. 3, 4 Colorless Add/Drop, Module 5 p. 9
Absorption loss, Module 1 p. 18 Constructive interference, Module 1 p. 9
Absorption, Module 2 p. 5 Contact printing, Module 1 p. 25
Active alignment, Module 2 p. 41 Contentionless Add/Drop, Module 5 p. 9
Active devices, Module 2 pp. 6, 13, 24 Coupling length, Module 2 p. 21
Active devices, Module 3 pp. 31-32 Coupling, Module 2 pp. 18-22
Add/Drop path, Module 5 p. 7 Coupling, Module 3 pp. 8, 11
Amplified spontaneous emission, Module 3 Coupling, Module 4 pp. 10, 11
p. (ASE) 30 Critical angle, Module 1 pp. 6, 9, 17
Amplifier, Module 2 p. 13 Crosstalk, Module 2 p. 22
Annealed proton exchange (APE), Module Crosstalk, Module 5 pp. 5, 9, 11, 13
4 pp. 6, 18
Array pitch, Module 3 p. 14 D
Arrayed Waveguide Grating (AWG)
Deformable Mirror (DM), Module 5 p. 20
Multiplexer, Module 5 p. 3
Demultiplexing, Module 2 p. 9
Arrayed waveguide grating (AWG),
Demultiplexer, Module 3 p. 12
Module 3 pp. 12-14, 17-20
Demultiplexer, Module 5 pp5, 6, 10, 12, 14
Arrayed waveguide grating (AWG),
Dense Wavelength Division Multiplexing
Module 4 pp. 12-19
(WDM), Module 5 pp. 2, 6, 12
Athermal AWG, Module 4 pp. 14, 15, 18
Deposition, Module 1 pp. 21, 23, 24-25
Attenuation, Module 2 p. 10
Destructive interference, Module 1 p. 9
AWG parameters, Module 3 pp. 18-20
Die bonding, Module 2 p. 41
B Die testing, Module 2 p. 40
Digital optical switch (DOS), Module 4 pp.
Bandgap energy, Module 3 pp. 22, 24 23-24
Bend radius, Module 1 p. 17 Diode Laser, Module 5 pp. 2, 3, 4, 12, 19
Bend waveguides, Module 2 p. 17 Diode lasers, Module 3 pp. 21-26
Binary compound, Module 3 p. 1 Direct band gap, Module 2 p. 32
Biosensors, Module 5 p. 15 Direct laser writing, Module 1 p. 24
Birefringence, Module 2 p. 16 Directional coupler, Module 2 p. 17
Birefringence, Module 3 p. 7 Directional coupler, Module 2 pp. 20-21
Birefringence, Module 4 pp. 10, 21 Directionless Add/Drop, Module 5 p. 9
Bragg grating, Module 2 pp. 30-32 Dispersion, Module 2 p. 11
Bragg reflectors, Module 3 pp. 27-28
Bunny suit, Module 1 p. 4
Distributed Feedback (DFB) Diode Lasers, H
Module 5 p. 3 Heat sinking, Module 3 p. 24
Distributed feedback, Module 3 p. 27 Heterostructure, Module 3 p. 5
Doping, Module 2 p. 4 High delta waveguide, Module 4 pp. 9-11
Doping, Module 4 p. 4 Higher-order modes, Module 1 pp. 10, 13
Dry etch, Module 1 p. 22
I
E
Index of refraction, Module 1 pp. 6, 11, 16,
Effective index, Module 2 p. 16 19, 23
Effective index, Module 3 p. 7 Index of refraction , Module 2 p. 15
Effective index, Module 4 p. 10, 13 Index of refraction, Module 3 p. 4
Efficiency, Module 2 p. 7 Index of refraction, Module 4 pp. 4, 6, 7,
Electro-Absorption Modulator (EAM), 10, 15, 16, 19, 23
Module 5 p. 3 Indirect band gap, Module 2 p. 32
Electro-optic effect, Module 4 pp. 16, 19, Indirect gap material, Module 1 p. 20
20 Insertion loss, Module 2 p. 7
Electroabsorption modulators (EAMs), Insertion loss, Module 3 pp. 19, 20
Module 3 pp. 32, 34 Insertion loss, Module 4 pp. 1, 10, 12, 16
Erbium Doped Fiber Amplifier (EDFA), Integrated multiwavelength laser, Module 3
Module 5 p. 4 p. 31
Etching, Module 1 pp. 22, 26 Integrated optics, Module 4 pp. 1, 20, 27
Evanescent field, Module 1 p. 12 Interconnections, Module 4 pp. 21-22
Evanescent field, Module 2 p. 20 International Telecommunication Union
Evanescent field, Module 3 p. 36 (ITU) grid, Module 3 pp. 15-16
Express path, Module 5 pp. 6, 7, 9
Extinction ratio (ER), Module 3 p. 32 L
F Lab-On-a-Chip (LOC), Module 5 pp. 15, 17
Laser junction structure, Module 3 pp. 26-
Fabrication, Module 2 pp. 37-38 27
Fabrication, Module 3 p. 36 Lithium niobate on insulator (LNOI),
Fiber array, Module 2 p. 39 Module 4 p. 19
Filter 26, Module 2 p. 28 Lithium niobate, Module 4 pp. 6-7, 19-21
Flexible integrated optic devices, Module 4 Lower cladding, Module 1 p. 14
p. 2
Flip-chip, Module 2 p. 42 M
Free propagation region (FPR), Module 4 p.
14 Mach-Zender interferometer (MZI), Module
Free propagation regions (FPR), Module 3 2 pp. 23-26
pp. 14, 19 Mach-Zender interferometer (MZI), Module
Fundamental mode, Module 1 pp. 10-14 4 pp. 19, 20, 22, 23
Mach-Zehnder interferometer (MZI),
G Module 5 pp. 8, 13, 16, 17
Mach-Zender modulator (MZM), Module 3
Gain parameter, Module 2 p. 7 pp. 33-34
Gain, Module 3 pp. 29, 30 Mask, Module 1 pp. 22, 25
Grating order, Module 4 pp. 13, 14 Materials used, Module 1 p. 19
Mesh Network, Module 5 p. 9
Metallization, Module 1 p. 23 Phase-matching condition (see constructive
Micro-Electro-Mechanical System (MEM), interference), Module 1 p. 9
Module 5 pp. 20, 21, 22, 23 Phase-matching condition, Module 2 p. 19
Micro-Electro-Mechanical System Photodetectors, Module 2 p. 36
(MEMS), Module 5 pp. 18, 19, 20, 21 Photodetectors, Module 3 pp. 35-36
Micro-Opto-Electro-Mechanical System Photolithography, Module 1 pp. 22, 25
(MOEMS), Module 5 pp. 18, 19, 21 Photonic Biosensors, Module 5 pp. 14, 15
Modulation depth, Module 2 p. 35 Photoresist, Module 1 p. 22
Modulators, Module 2 pp. 26, 34 Planar optical waveguides, Module 1pp. 14,
Modulators, Module 3 pp. 32-35 21
Modulators, Module 4 p. 20 Polarization, Module 2 pp. 16, 33
Monolithic integration, Module 1 p. 2 Polarization, Module 3 p. 17
Monolithic integration, Module 2 pp. 33-34 Polymers, Module 4 pp. 7, 20-27
Multimode interference (MMI) coupler, Power monitoring, Module 4 pp. 16, 17
Module 3 pp. 9-12 Power splitting, Module 2 pp. 22-24
Multimode Interference (MMI), Module 5 Power splitting, Module 3 p. 11
p. 16 Projection printing, Module 1 p. 25
Multimode, Module 1 pp. 11, 16 Propagation loss, Module 1 p. 18
Multiple quantum well (MQW) lasers, Propagation loss, Module 2 p. 14
Module 3 pp. 28-29 Propagation loss, Module 3 p. 20
Multiplexer (MUX), Module 4 p. 16 Proton exchange method, Module 1 p. 23
Multiplexer, Module 3 pp. 12-13 Proximity printing, Module 1 p. 25
N R
Native oxide, Module 4 p. 3 Radio frequency (RF), Module 3 p. 32
Negative uniaxial crystal, Module 4 p. 6 Radius of curvature, Module 2 p. 17
Noise level, Module 2 p. 36 Reactive-ion etching (RIE), Module 1 p. 22
Numerical aperture, Module 1 p. 7 Reconfigurable Optical Add/Drop
Multiplexer (ROADM), Module 5 pp. 6-12
O Rectangular waveguide (see channel
Optical fiber, Module 1 p. 6 waveguide), Module 1 p. 15
Optical fiber transmission system, Module 2 Reduction scanning method, Module 1 p. 26
pp. 9, 10 Reduction systems, Module 1 p. 25
Optical Power Monitor (OPM), Module 5 p. Reflection loss, Module 2 p. 20
3 Refractive index contrast, Module 1 pp. 8,
Optical Receiver, Module 5 p. 5 9, 10, 17
Optical Sensors, Module 5 pp. 14, 22 Refractive index contrast, Module 4 pp. 7,
Optical Transmitter, Module 5 p. 2 8, 9, 19, 21, 23, 25, 27
Optical waveguide, Module 1 p. 6 Regenerator, Module 2 p. 12
Optical waveguide, Module 2 pp. 14-18 Relative refractive index difference (see
refractive index contrast), Module 1 p. 8
P Repeater, Module 2 p. see Regenerator, 12
Resonance, Module 2 p. 27
P-n junction, Module 3 p. 21 Responsivity, Module 2 p. 36
Passivation, Module 1 p. 23 Reticle, Module 1 p. 25
Passive devices, Module 2 pp. 6, 13, 14 Rib waveguide (see ridge waveguide),
Passive devices, Module 3 pp. 5-21 Module 1 p. 15
Percent transmission, Module 4 pp. 3, 5, 7
Ridge waveguide, Module 1 p. 15 Tunable wavelength laser, Module 4 p. 25
Ring Network, Module 5 p. 9 Tuned Interleaver, Module 5 p. 13
Ring resonator, Module 2 pp. 26-30 Two Sectional Photodetector (TSP),
Module 5 p. 16
S
Scattering loss, Module 1 p. 18 U
Self-imaging, Module 3 p. 10 Upper cladding, Module 1 pp. 14, 23
Semiconductor lasers, Module 3 p. 25
Semiconductor optical amplifier (SOA), V
Module 3 pp. 29-32 V parameter, Module 1 p. 11
Sensitivity to wavelength, Module 2 p. 36 Variable optical attenuator (VOA), Module
Silica-on-silicon, Module 4 pp. 8-12, 21, 23 2 p. 26
Silica, Module 4 pp. 1-2, 3-5, 8-18 Variable optical attenuator (VOA), Module
Silicon-On-Insulator (SOI) Ring Resonator, 4 pp. 16-17, 22-23
Module 5 p. 15 Variable Optical Attenuator (VOA),
Silicon-on-insulator (SOI), Module 1 p. 16 Module 5 pp. 3, 4, 7, 8, 9, 10, 11
Silicon-on-insulator (SOI), Module 2 pp.
14-17, 30 W
Single-mode, Module 1 pp. 11, 15, 16
Slab waveguide, Module 1 p. 14 Waveguide parameters, Module 3 p. 7
Spectral response, Module 3 p. 17 Waveguide parameters, Module 4 p. 10
Steerable Micromirror Array, Module 5 p. Waveguide Shuffling Chip, Module 5 p. 11
20 Wavelength accuracy, Module 4 p. 11
Super-high delta waveguide, Module 4 pp. Wavelength Division Multiplexing (WDM),
9-10 Module 2 p. 9
Switch, Module 2 p. 25 Wavelength Division Multiplexing, Module
System-in-package, Module 2 p. (SiP) 42 3 p. 17
Wavelength Division Multiplexing (WDM),
T Module 5 pp. 2, 3, 4, 5
Wavelength spacing, Module 3 p. 16
Tapers, Module 3 p. 8 Wavenumber, Module 1 p. 11
Tapers, Module 4 pp. 11, 22 Wet etch, Module 1 p. 23
Temporal response, Module 2 p. 36 Wire bonding, Module 2 p. 41
Thermo-Optic effect, Module 1 p. 19
Thermo-Optic Switch, Module 5 pp. 7, 8 Y
Threshold current, Module 3 p. 22
Through-silicon vias (TSVs), Module 2 p. Y-branch, Module 2 p. 22
43
Titanium diffusion, Module 4 p. 18
Torsion Plate, Module 5 pp. 18, 19
Total internal reflection (TIR), Module 1
pp. 6, 7, 17
Transceiver, Module 5 pp. 12, 14
Transverse electric (TE), Module 1 p. 12
Transverse electromagnetic wave, Module 1
p. 11
Transverse magnetic (TM), Module 1 p. 12
Tunable Bragg grating, Module 4 pp. 24-26