![ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011
Light sources are needed to generate the signal which travels surrounding layer of SiO2 that is not masked by the hardened
down the optical fiber. These are typically the most costly photoresist can then be etched from the wafer. The hard-
element in optical communications. Most widely used are ened photoresist is then taken off and what is left is the
the Laser Diode (LD) and the Light Emitting Diode (LED). electronic component or circuit. In most cases, this process
These are both semiconductor devices, but as will be later is performed many times over and layer upon layer is built up
explained, not on the miniaturized scale needed for Silicon to form complex components and circuits. The layers them
Photonics. Key to the use of any light source is ensuring it selves can be composed of different materials and some-
is aligned with the optical fiber so that light entering the fiber times the base layer of Si itself is used by altering its chemi-
is propagated down its length with minimal loss. Of particular cal/electrical properties. But in all cases, the process is basi-
impact is loss related to back-reflection at the point of entry. cally the same: masking of some sort, alteration of that which
If the light source is not aligned perfectly, some light will is not masked, then finalized what remains. This process has
catch the edges of the fiber and reflect back into the source, enabled the miniaturization of individual electronic compo-
causing loss in signal strength, as well as creating potential nents initially the size of a dime or a quarter along with elec-
problems for the light source itself in the form of interference tronic circuits that would incorporate countless meters of
and built up of heat. wiring down to something measured in nanometers (nm).
A second key aspect in the use of a light source is the
form of manipulation or modulation. To represent binary IV. NEED FOR SILICON PHOTONICS
data, the light must represent two distinct levels. In optical
Fiber-optic communication is the process of transporting
communications, this is done by turning the light on and off.
data at high speeds on a glass fiber using light. However,
There are two common ways, called modulation techniques,
this technology is an expensive solution. The components
by which this is done. The most obvious is to turn the light
are typically fabricated using exotic materials that are
source on and off by applying or removing voltage to the
expensive to manufacture [7].
source (called direct modulation). While this seems straight
The trouble with multi-core processors is another
forward, it has problems. The two most impacting being the
challenge. Programming multi-core processors is a complex
time required and a distortion in the signal called frequency
process at the same time it is quiet tough to implement. Here
chirp (Herve, Ovadia 2004). A second method, ‘external
the main goal was to develop high-volume, low-cost optical
modulation,’ minimizes these problems. Here the light source
components using standard CMOS processing [4] the same
is run in Continuous-Wave (CW) mode, meaning it is never
manufacturing process used for microprocessors and
shut down, while an external (to the light source) component
semiconductor devices. Moreover, manufacturing silicon
determines when light is allowed to pass into the fiber or not.
components in high volume to the specifications needed by
The final component necessary for optical communication is
optical communications was comparatively inexpensive. Fiber
the light detector that as its name suggests, ‘detects light’ at
is already being used to shuttle data from computers to data
the receiving end. More importantly, it discriminates between
storage devices and from computer to computer. With a
light and no light to reconstruct the patterns of the modulator
potential of terabits in the optical domain and problems
on the transmitting end and convert this back into an electrical
starting in the gigabit range for metal wire circuits, a
signal to be used by the receiving device as digital data.
bottleneck becomes evident.
These are also typically semiconductor-based devices called
Silicon Photonics showed promise as the answer. The
photodiodes, but once again these are of a scale too large to
idea was to build all the components for optical circuits with
integrate onto a microchip with other components.
the CMOS manufacturing processes and eliminate the
bottleneck. Extend the optical communication path inside
III. CMOS MANUFATURING PROCESSES
the computer, inside any electronic devices in the path,
There is an inexorable trend in the electronics industry to perhaps even all the way into the microprocessor and memory
make things smaller, faster and cheaper. To fully understand chips themselves.
the scope of Silicon Photonics requires understanding the Silicon’s key drawback was that it cannot emit laser light.
processes involved in creating a CMOS chip. CMOS chips However, silicon can be used to manipulate the light emitted
are built on a substrate of Silicon (Si). Silicon is used be- by inexpensive lasers so as to provide light that has
cause of its properties as a natural semiconductor – it can characteristics similar to more-expensive devices. This was
function equally well as a conductor or an insulator of elec- just one way in which silicon can lower the cost of photonics
tricity – and it is both inexpensive and abundant (it is made
from sand). A wafer of silicon is produced and a layer of V. PHYSICAL PROPERTIES
Silicon Dioxide (SiO2) is placed on top of the wafer. This is The propagation of light through silicon devices was
then followed up with a layer of chemical called a ‘photore- governed by a range of nonlinear optical phenomena
sist’, so named because when exposed to a certain wave- including the Kerr effect, the Raman effect, two photon
length of light the chemical hardens. Using that wavelength, absorption and interactions between photons and free charge
a pattern is laid out in the layer of photoresist; then the photo carriers. The presence of nonlinearity was of fundamental
resist that was not exposed to light is washed away. The importance, as it enabled light to interact with light, thus
© 2011 ACEEE 27
DOI: 01.IJCOM.02.03. 524](https://image.slidesharecdn.com/524-120911011433-phpapp02/85/Silicon-Photonics-A-Solution-for-Ultra-High-Speed-Data-Transfer-2-320.jpg)
![ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011
permitting applications such as wavelength conversion and C. Two-photon absorption
all-optical signal routing, in addition to the passive Silicon exhibits two-photon absorption (TPA) [9], in which a
transmission of light. pair of photons can act to excite an electron-hole pair.[8] This
process was related to the Kerr effect, and by analogy with
A. Optical guiding and dispersion tailoring
complex refractive index, could have been thought of as the
Silicon is transparent to infrared light with wavelengths imaginary-part of a complex Kerr nonlinearity. At the 1.55
above about 1.1 micrometers .Silicon also has a very high micrometer telecommunication wavelength, this imaginary
refractive index, of about 3.5.The tight optical confinement part was approximately 10% of the real part.
provided by this high index allows for microscopic optical The influence of TPA was highly disruptive, as it both
waveguides, which may have cross-sectional dimensions of wasted light, and generated unwanted heat. It could be
only a few hundred nanometers. This is substantially less mitigated, however, either by switching to longer wavelengths
than the wavelength of the light itself, and is analogous to a (at which the TPA to Kerr ratio drops), or by using slot
sub wavelength-diameter optical fibre. Single mode waveguides (in which the internal nonlinear material has a
propagation can be achieved, thus (like single-mode optical lower TPA to Kerr ratio). Alternatively, the energy lost through
fiber) eliminating the problem of modal dispersion. TPA could be partially recovered (as is described below) by
The strong dielectric boundary effects that result from extracting it from the generated charge carriers.
this tight confinement substantially alter the optical dispersion
relation. By selecting the waveguide geometry, it was D. Free charge carrier interactions
possible to tailor the dispersion to have desired properties, The free charge carriers within silicon can both absorb
which was of crucial importance to applications requiring photons and change its refractive index. This is particularly
ultra-short pulses. In particular, the group velocity dispersion significant at high intensities and for long durations, due to
(that is, the extent to which group velocity varies with the carrier concentration being built up by TPA. The influence
wavelength) can be closely controlled. In bulk silicon at 1.55 of free charge carriers was often (but not always) unwanted,
micrometers, the group velocity dispersion (GVD) is normal and various means had been proposed to remove them. One
in that pulses with longer wavelengths travel with higher such scheme was to implant the silicon with helium in order
group velocity than those with shorter wavelength. By to enhance carrier recombination. A suitable choice of
selecting suitable waveguide geometry, however, it was geometry could also be used to reduce the carrier lifetime.
possible to reverse this, and achieve anomalous GVD, in Rib waveguides (in which the waveguides consist of thicker
which pulses with shorter wavelengths travel faster. regions in a wider layer of silicon) enhanced both the carrier
Anomalous dispersion was significant, as it was a prerequisite recombination at the silica-silicon interface and the diffusion
for modulation instability. of carriers from the waveguide core.
In order for the silicon photonic components to remain A more advanced scheme for carrier removal was to
optically independent from the bulk silicon of the wafer on integrate the waveguide into the intrinsic region of a PIN
which they were fabricated, it was necessary to have a layer diode, which was reverse biased so that the carriers were
of intervening material. This was usually silica, which has a attracted away from the waveguide core. A more sophisticated
much lower refractive index (of about 1.44 in the wavelength scheme still, was to use the diode as part of a circuit in which
region of interest), and thus light at the silicon-silica interface voltage and current are out of phase, thus allowing power to
had (like light at the silicon-air interface) total internal be extracted from the waveguide. The source of this power
reflection, and remained in the silicon. This construct is known was the light lost to two photon absorption, and so by
as silicon on insulator. It was named after the technology of recovering some of it, the net loss (and the rate at which heat
silicon on insulator in electronics, whereby components were is generated) could have been reduced.
built upon a layer of insulator in order to reduce parasitic As is mentioned above, free charge carrier effects could also
capacitance and so improve performance. be used constructively, in order to modulate the light.
B. Kerr nonlinearity E. Raman effect
Silicon has a focusing Kerr nonlinearity, in that the Silicon exhibits the Raman-effect, in which a photon is
refractive index increases with optical intensity. This effect exchanged for a photon with a slightly different energy,
was not especially strong in bulk silicon, but it was greatly corresponding to an excitation or a relaxation of the material.
enhanced by using a silicon waveguide to concentrate light Silicon’s Raman transition was dominated by a single, very
into a very small cross-sectional area. This allowed nonlinear narrow frequency peak, which was problematic for broadband
optical effects to be seen at low powers. The nonlinearity phenomena such as Raman amplification, but was beneficial
could have been enhanced further by using a slot waveguide, for narrowband devices such as Raman lasers. Consequently,
in which the high refractive index of the silicon was used to all-silicon Raman lasers were fabricated.
confine light into a central region filled with a strongly
nonlinear polymer. VI. ESSENTIAL COMPONENTS
In order to”siliconize” photonics, six main building blocks
were used:
© 2011 ACEEE 28
DOI: 01.IJCOM.02.03. 524](https://image.slidesharecdn.com/524-120911011433-phpapp02/85/Silicon-Photonics-A-Solution-for-Ultra-High-Speed-Data-Transfer-3-320.jpg)
![ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011
• An inexpensive light source When the silicon and the indium phosphide based material
• Devices that can route, split, and direct light on the silicon were heated and pressed together, the two oxide layers fused
chip them together [7]. When the voltage was applied, the light
• A modulator to encode or modulate data into the optical generated in the indium phosphide based material passed
signal directly through the glue layer into the silicon waveguide
• A photo detector to convert the optical signal back into below, which acted like the laser cavity to create the hybrid
electrical bits silicon laser. The design of the individual silicon waveguides
• Low-cost, high-volume assembly methods was critical in determining the performance of the hybrid
• Supporting electronics for intelligence and photonics control silicon laser, and would allow future versions to be built that
generate specific wavelengths.
C. Silicon Modulators
One key problem was making high speed modulators,
since silicon was not electro-optic due to its inversion
symmetry [10]. This leaves carrier-optic and thermo-optic
effects as the primary mechanisms for modulation in silicon
and due to the relatively long carrier lifetime in silicon are
slow under carrier injection. The first demonstration of a
Gigabit modulator was in 2004, utilizing an MOS based
structure under depletion. Progress in this area has been
very rapid by many groups around the world. A recent,
exciting advance is the 40 Gbit/s silicon modulator [12]
demonstrated by Intel.
Figure 1.building blocks in silicon photonics
The key challenges that were met in the development of
silicon photonics are:
Light source.
Silicon waveguide.
Silicon modulators.
Photo detector.
A. Light source Figure 2. A 1 Tbit/s transmitter made out of silicon technology
Silicon is an inefficient light emitter because of a D. Photo detector
fundamental limitation called an indirect band-gap. An indirect The use of Ge was important because, unlike Si, it could
band-gap prevents the atoms in silicon from emitting photons efficiently detect light in the near infrared which was the
in large numbers when an electrical charge is applied. Instead, standard for communications. The drawback was that so much
silicon emits heat. Silicon is capable of routing, modulating, stress was developed in pure Ge films deposited on Si that
and detecting light; silicon has needed an external light source defects were introduced near the Ge/Si interface. Careful
to provide the initial light. These external light sources are design and processing were needed to minimize the impact
generally discrete lasers and require careful alignment to the of these defects on the electrical performance of the device.
silicon waveguides [3], [7]. The problem is that accurate This was now a different type of Ge/Si photo detector that
alignment is difficult and expensive to achieve. Even has built-in amplification, which makes it much more useful
submicron misalignment of the laser to the silicon waveguide in instances where very little light falls on the detector. It is
can render the resulting photonic device useless. A long- called an avalanche photo detector. Because an avalanche
standing quest in silicon photonics has been the creation of process occurred inside the device. In order to realize the full
a laser source that can be manufactured directly on the silicon performance potential from this materials, it was needed to
photonic chip, in high volume, and whose emitted light is further reduce the dark current that was coming from the
automatically aligned with the silicon waveguide. defects at the Ge/Si interface, and stop the inter diffusion of
B. Hybrid Silicon Wave Guide Ge and Si that occurred during annealing [7]. This intermixing
was problematic since the Ge caused higher noise than if the
The silicon substrate was the base upon which the other silicon alone was in the multiplication region. If we are
items were placed. On this substrate rests the silicon successful, this work will pave the way for developing low
waveguide [14]. Both the substrate and the waveguide were cost, CMOS-based Ge/Si APD operating at data rates of
manufactured using standard silicon fabrication processes. 40Gbps or higher in the future.
© 2011 ACEEE 29
DOI: 01.IJCOM.02.03. 524](https://image.slidesharecdn.com/524-120911011433-phpapp02/85/Silicon-Photonics-A-Solution-for-Ultra-High-Speed-Data-Transfer-4-320.jpg)
![ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011
VII. APPLICATIONS today’s level of 100 W/cm2 challenges even the most
advanced packaging technologies. The problem of heat
Silicon photonics has its wide range of applications in
dissipation was so severe that it threatens to bring to halt the
Optical communication.
continued advance of the technology, as described by
Data com and telecom applications.
Moore’s law. This fact was highlighted by the recent shift of
VOA’s.
the microprocessor industry away from increasing the clock
ROADM’s.
speed and in favor of multi-core processors. Among photonic
Silicon Triplexer.
components, lasers were the most power hungry photonic
Ring resonators.
devices [1],[3].
Optical shifter /mirror.
The lack of an electrically pumped Si Laser, to date, dictates
Optical multi channel separating filter.
an architecture where the light source remains off-chip. Far
Modulated Raman laser.
from being a compromised solution, this architecture was in
Raman amplifier.
fact preferred as it removed a main source of heat dissipation.
Wavelength converters.
Furthermore, the performance degradation of injection lasers
Splitters and couplers.
at high temperatures will be a major obstacle to their
Attenuators and deflectors.
integration onto the hot VLSI substrate [1].
VIII. CHALLENGES
IX. PROPOSED SOLUTION TO OVERCOME THE CHALLENGES
A. Compatibility-quad
Instead of using silicon wafer if we build it with the gallium
“Compatibility-quad” describes conditions that must be nitride of two inch in size which is almost 1/6th the diameter of
met by silicon photonics to be fully compatible with VLSI standard silicon wafer can be used in producing violet laser
electronics. [4]. At the same time we can have double digit rate of growth,
if the size of gallium nitride is made 100mm or more. It can be
used to conduct heat far better than silicon. By making
substrates from gallium nitride we can provide better
foundation for diodes and transistors [4]. Gallium nitride
families can produce LED’S with outputs ranging from the
ultraviolet to infrared. Using gallium nitride as wafer is one of
the key challenges i.e., heat dissipation can be eliminated
since it won’t succumb below a temperature of 2225 æ%c
Figure 3. Compatibility–quad describing conditions that must be and pressure of 64000 atmospheres. This GaN structure is
met by silicon photonics to be fully compatible with VLSI created through a combined approach of a layer-by-layer
electronics template fabrication technique and selective metal organic
B. Material compatibility and process compatibility chemical vapor deposition (MOCVD). These GaN 3DPC
The vision of silicon photonics has been the introduction exhibit a stacking direction band gap characterized by strong
of photonics into silicon CMOS manufacturing process, thus optical reflectance between 380 and 500 nm. By introducing a
silicon is process compatible. Silicon can guide light through “line-defect” cavity in the fifth (middle) layer of the 3DPC, a
it because it is transparent at the infrared Wavelengths at localized transmission mode with a quality factor of 25–30 is
which optical communication systems operate so; photonics also observed within the photonic band gap
is material compatible [1].
C. Economic compatibility X. CONCLUSION
The external light sources were generally lasers and they This paper has been an attempted overview of the silicon
required careful alignment to the silicon wave guide. The photonics and its application areas, the current state of device
problem here was that accurate alignment was difficult and technology, and the challenges that lie ahead on the path to
expensive to achieve. Even submicron misalignment of the commercial success. Silicon photonics is on the verge of
laser to the silicon waveguide could render the resulting becoming a viable technology for various applications
photonic device useless [1]. This required expensive especially in communication and internet. Already, commercial
processing methods for accurate alignment of lasers to the components such as optical transceivers are available based
silicon waveguides. on the technology. “Economic” and “heat” compatibility with
D. Heat compatibility silicon microelectronics are the main challenges ahead.
Heat compatibility requires that photonic devices must
be able to operate on the hot VLSI substrate and that their
own power dissipation should be negligible. For photonics
to merge with VLSI electronics, it cannot significantly increase
the chip area. Increase in power density of VLSI chips, where
© 2011 ACEEE 30
DOI: 01.IJCOM.02.03. 524](https://image.slidesharecdn.com/524-120911011433-phpapp02/85/Silicon-Photonics-A-Solution-for-Ultra-High-Speed-Data-Transfer-5-320.jpg)
![ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011
REFERENCES [10] J. Spector, M. W. Geis, M. E. Grein, R. T. Schulein, J. U.
Yoon, et al , “High-speed silicon electro-optical modulator that can
[1] Bahram Jalali, and Sasan Fathpour Silicon Photonics, be operated in carrier depletion or carrier injection mode,” 2008
DECEMBER 2006. Conference on Lasers and Electro-Optics (CLEO), (2008).
[2] Mehdi Asghari, Silicon Photonics: A Low Cost Integration [11] G. T. Reed, W. R. Headley, G. Z. Mashanovich, F. Y. Gardes,
Platform for Datacom and Telecom Applications . D. J. Thomson, M. M. Milosevic, “The Evolution of Integration,”
[3] Richard Soref, The Past, Present, and Future of Silicon Photonics. in Silicon Photonics for Telecommunications and Biomedical
IEEE JOURNAL NOVEMBER/DECEMBER 2006. Applications, Eds. Bahram Jalali and Sasan Fathpour, Taylor &
[4] The crystal that will change everything published on 3 August Francis Books, Inc., ISBN 1439806373, 2010
2010 in spectrum.ieee.org. [12] M. J. Shaw, J. Guo, G. A. Vawter, S. Habermehl, C. T. Sullivan,
[5] Benoit reid and Kelvin and Prosyk, Integrated loop mirror “Fabrication techniques for low loss silicon nitride waveguides,”
Published on June 2005. Proc. SPIE
[6] Amir H. Atabaki, Siva Yegnanarayanan, Babak Momeni, Ehsan [12] D. J. Thomson, F. Y. Gardes, Y. Hu, G. Mashanovich, M.
Shah-Hosseini, et al, Implementation of a Coupling- Fournier, P. Grosse, J-M. Fedeli, and G. T. Reed, “High contrast 40
Tunable Resonator for Efficient High-Bandwidth Nonlinear Silicon Gbit/s optical modulation in silicon,” Optics Express, vol. 19, pp.
Photonics Applications. 11507-11516, 2011
[7] V. Raghunathan and B. Jalali, “Stress-induced phase matching [13]D. Thomson, F. Y. Gardes, G. T. Reed, G. Rasigade, M. Ziebell,
in silicon waveguides,” presented at the Conf. Lasers and Electro- D. Marris-Morini,et al, “Silicon optical modulators for high data
Optics (CLEO), Long Beach, CA, May 2006, Paper CMK5 rate applications,” 8th IEEE Group IV Photonics conference,
[ 9] P. Dumon, W. Bogaerts, Fibre-Array Optical Interconnection London, UK, 14-16 September 2011
for Silicon Photonics, Proc. ECOC, Vol. 5, p. 93 (P.2.21), Brussels, [14]N. M. Wright, D. J. Thomson, K. L. Litvinenko, W. R. Headley,
Belgium (2008) A. J. Smith, et al, “Free carrier lifetime modification for silicon
waveguide based devices,” Optics Express, vol. 16, pp. 19779 -
19784, 2008
© 2011 ACEEE 31
DOI: 01.IJCOM.02.03. 524](https://image.slidesharecdn.com/524-120911011433-phpapp02/85/Silicon-Photonics-A-Solution-for-Ultra-High-Speed-Data-Transfer-6-320.jpg)
Silicon Photonics: A Solution for Ultra High Speed Data Transfer
- 1. ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011 Silicon Photonics: A Solution for Ultra High Speed Data Transfer Pratik Ganguly, Rahul .S Research Scientist V R Enterprises Andhra Pradesh, India pratik.vresez@gmail.com Research Scientist V R Enterprises Andhra Pradesh, India rahuls_2u@yahoo.co.in Abstract— Silicon photonics is the integration of integrated so many other electronic devices; make them smaller, faster, optics and photonics IC technologies in silicon. Silicon and cheaper; specifically, to bypass current barriers in opti- photonics has recently attracted a great deal of attention since cal communications by integrating optical computing with it offers an opportunity for low cost solutions for various semiconductor chips. Silicon photonics aims to provide in- applications ranging from telecommunications to chip-chip expensive silicon building blocks that can be integrated to inter connects. Two keys to this advancement are the increased produce optical products that solve real communication prob- speed of communications (now at the speed of light) and the lems for consumers. Silicon is an especially useful material increased amount of data that can be transmitted at once (i.e., bandwidth). Silicon photonics is the study and application of for photonics components because it is transparent at the photonic systems which use silicon as an optical medium. infrared Wavelengths at which optical communication sys- The silicon is usually patterned with sub-micrometer tems operate. precision, into microphotonic components. These operate in Silicon photonic devices can be made using existing the infrared, most commonly at the 1.55 micrometer semiconductor fabrication techniques, and because silicon wavelength used by most fiber optic telecommunication was already used as the substrate for most integrated circuits, systems. The silicon typically lies on top of a layer of silica in it was possible to create hybrid devices in which the optical what (by analogy with a similar construction in and electronic components were integrated onto a single microelectronics) is known as silicon on insulator (SOI). Today microchip. the problems associated with multi-core processors with copper interconnect are Latency, Bandwidth, Power dissipation, Electromagnetic interference and Signal integrity. Micro II. COMPONETS OF OPTICAL COMMUNICATION processor designers use the integration of number of At present optical communication networks consist of transistors that could be squeezed onto each chip to boost three key building blocks: optical fiber, light sources, and computational horsepower. That in turn caused the amount of waste heat that had to be dissipated from each square light detectors. millimeter of silicon to go up. One problem we are facing in Optical fibers are now ubiquitous. They are the long, this effort is that micro processors with large numbers of cores thin glass or Teflon fibers used to propagate a light signal are not yet being manufactured. Fiber optics has a reputation down their length. This signal can be manipulated at both as an expensive solution because of high cost of hardware and ends so as to code and decode the 1s and 0s that comprise Fabrication is done using exotic materials which are costly. the digital data domain. While initial use of optical fiber The methods used in assembly and package of these involved a single stream of data, this was soon improved components are also expensive. A recent break through in upon with Multimode Fibers (MMFs). These fibers can silicon photonics is in the development of a laser modulator support several light sources by using slightly different that encodes optical data at 40 billion bits per second. Finally reached the goal of data transmission at 40 Gbps speed, angles when propagating light down the length of the fiber, matching the fastest devices deployed today with least cost of thus dramatically increasing a single fiber’s bandwidth. With processing and showing the ultimate solutions to the problems such improvements in the ability to manipulate the digital associated with copper interconnects in multi-core processors signal (called modulation), data speeds are now at 10 Gb/s and expensive fiber optics. and could theoretically reach 20 Tb/s. Extrapolate this out to the bundle of fibers that typically run down most residential Index Terms — silicon photonics, Light source, silicon avenues today delivering voice, TV, and data and one starts waveguide, silicon modulator, Photo detector, indirect band to believe in the theory of ‘more than we could ever use’. gap, compatibility-quad, gallium nitride These fibers have made their way beyond the large deployments of the cable and telecommunications providers I. INTRODUCTION and into even small data centers. For instance, it is not at all Silicon is the principal material used in semiconductor uncommon to find mass storage systems attached to high manufacturing today because it is plentiful, inexpensive, well end computing systems via an optical fiber link. Fiber is even understood by the semiconductor industry. Silicon photonics finding its way into ‘the last mile’, being offered by some is a term given to the science of optical communications, a cable providers delivering optic communications all the way science that is now looking to do what has been done with to your home. © 2011 ACEEE 26 DOI: 01.IJCOM.02.03. 524
- 2. ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011 Light sources are needed to generate the signal which travels surrounding layer of SiO2 that is not masked by the hardened down the optical fiber. These are typically the most costly photoresist can then be etched from the wafer. The hard- element in optical communications. Most widely used are ened photoresist is then taken off and what is left is the the Laser Diode (LD) and the Light Emitting Diode (LED). electronic component or circuit. In most cases, this process These are both semiconductor devices, but as will be later is performed many times over and layer upon layer is built up explained, not on the miniaturized scale needed for Silicon to form complex components and circuits. The layers them Photonics. Key to the use of any light source is ensuring it selves can be composed of different materials and some- is aligned with the optical fiber so that light entering the fiber times the base layer of Si itself is used by altering its chemi- is propagated down its length with minimal loss. Of particular cal/electrical properties. But in all cases, the process is basi- impact is loss related to back-reflection at the point of entry. cally the same: masking of some sort, alteration of that which If the light source is not aligned perfectly, some light will is not masked, then finalized what remains. This process has catch the edges of the fiber and reflect back into the source, enabled the miniaturization of individual electronic compo- causing loss in signal strength, as well as creating potential nents initially the size of a dime or a quarter along with elec- problems for the light source itself in the form of interference tronic circuits that would incorporate countless meters of and built up of heat. wiring down to something measured in nanometers (nm). A second key aspect in the use of a light source is the form of manipulation or modulation. To represent binary IV. NEED FOR SILICON PHOTONICS data, the light must represent two distinct levels. In optical Fiber-optic communication is the process of transporting communications, this is done by turning the light on and off. data at high speeds on a glass fiber using light. However, There are two common ways, called modulation techniques, this technology is an expensive solution. The components by which this is done. The most obvious is to turn the light are typically fabricated using exotic materials that are source on and off by applying or removing voltage to the expensive to manufacture [7]. source (called direct modulation). While this seems straight The trouble with multi-core processors is another forward, it has problems. The two most impacting being the challenge. Programming multi-core processors is a complex time required and a distortion in the signal called frequency process at the same time it is quiet tough to implement. Here chirp (Herve, Ovadia 2004). A second method, ‘external the main goal was to develop high-volume, low-cost optical modulation,’ minimizes these problems. Here the light source components using standard CMOS processing [4] the same is run in Continuous-Wave (CW) mode, meaning it is never manufacturing process used for microprocessors and shut down, while an external (to the light source) component semiconductor devices. Moreover, manufacturing silicon determines when light is allowed to pass into the fiber or not. components in high volume to the specifications needed by The final component necessary for optical communication is optical communications was comparatively inexpensive. Fiber the light detector that as its name suggests, ‘detects light’ at is already being used to shuttle data from computers to data the receiving end. More importantly, it discriminates between storage devices and from computer to computer. With a light and no light to reconstruct the patterns of the modulator potential of terabits in the optical domain and problems on the transmitting end and convert this back into an electrical starting in the gigabit range for metal wire circuits, a signal to be used by the receiving device as digital data. bottleneck becomes evident. These are also typically semiconductor-based devices called Silicon Photonics showed promise as the answer. The photodiodes, but once again these are of a scale too large to idea was to build all the components for optical circuits with integrate onto a microchip with other components. the CMOS manufacturing processes and eliminate the bottleneck. Extend the optical communication path inside III. CMOS MANUFATURING PROCESSES the computer, inside any electronic devices in the path, There is an inexorable trend in the electronics industry to perhaps even all the way into the microprocessor and memory make things smaller, faster and cheaper. To fully understand chips themselves. the scope of Silicon Photonics requires understanding the Silicon’s key drawback was that it cannot emit laser light. processes involved in creating a CMOS chip. CMOS chips However, silicon can be used to manipulate the light emitted are built on a substrate of Silicon (Si). Silicon is used be- by inexpensive lasers so as to provide light that has cause of its properties as a natural semiconductor – it can characteristics similar to more-expensive devices. This was function equally well as a conductor or an insulator of elec- just one way in which silicon can lower the cost of photonics tricity – and it is both inexpensive and abundant (it is made from sand). A wafer of silicon is produced and a layer of V. PHYSICAL PROPERTIES Silicon Dioxide (SiO2) is placed on top of the wafer. This is The propagation of light through silicon devices was then followed up with a layer of chemical called a ‘photore- governed by a range of nonlinear optical phenomena sist’, so named because when exposed to a certain wave- including the Kerr effect, the Raman effect, two photon length of light the chemical hardens. Using that wavelength, absorption and interactions between photons and free charge a pattern is laid out in the layer of photoresist; then the photo carriers. The presence of nonlinearity was of fundamental resist that was not exposed to light is washed away. The importance, as it enabled light to interact with light, thus © 2011 ACEEE 27 DOI: 01.IJCOM.02.03. 524
- 3. ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011 permitting applications such as wavelength conversion and C. Two-photon absorption all-optical signal routing, in addition to the passive Silicon exhibits two-photon absorption (TPA) [9], in which a transmission of light. pair of photons can act to excite an electron-hole pair.[8] This process was related to the Kerr effect, and by analogy with A. Optical guiding and dispersion tailoring complex refractive index, could have been thought of as the Silicon is transparent to infrared light with wavelengths imaginary-part of a complex Kerr nonlinearity. At the 1.55 above about 1.1 micrometers .Silicon also has a very high micrometer telecommunication wavelength, this imaginary refractive index, of about 3.5.The tight optical confinement part was approximately 10% of the real part. provided by this high index allows for microscopic optical The influence of TPA was highly disruptive, as it both waveguides, which may have cross-sectional dimensions of wasted light, and generated unwanted heat. It could be only a few hundred nanometers. This is substantially less mitigated, however, either by switching to longer wavelengths than the wavelength of the light itself, and is analogous to a (at which the TPA to Kerr ratio drops), or by using slot sub wavelength-diameter optical fibre. Single mode waveguides (in which the internal nonlinear material has a propagation can be achieved, thus (like single-mode optical lower TPA to Kerr ratio). Alternatively, the energy lost through fiber) eliminating the problem of modal dispersion. TPA could be partially recovered (as is described below) by The strong dielectric boundary effects that result from extracting it from the generated charge carriers. this tight confinement substantially alter the optical dispersion relation. By selecting the waveguide geometry, it was D. Free charge carrier interactions possible to tailor the dispersion to have desired properties, The free charge carriers within silicon can both absorb which was of crucial importance to applications requiring photons and change its refractive index. This is particularly ultra-short pulses. In particular, the group velocity dispersion significant at high intensities and for long durations, due to (that is, the extent to which group velocity varies with the carrier concentration being built up by TPA. The influence wavelength) can be closely controlled. In bulk silicon at 1.55 of free charge carriers was often (but not always) unwanted, micrometers, the group velocity dispersion (GVD) is normal and various means had been proposed to remove them. One in that pulses with longer wavelengths travel with higher such scheme was to implant the silicon with helium in order group velocity than those with shorter wavelength. By to enhance carrier recombination. A suitable choice of selecting suitable waveguide geometry, however, it was geometry could also be used to reduce the carrier lifetime. possible to reverse this, and achieve anomalous GVD, in Rib waveguides (in which the waveguides consist of thicker which pulses with shorter wavelengths travel faster. regions in a wider layer of silicon) enhanced both the carrier Anomalous dispersion was significant, as it was a prerequisite recombination at the silica-silicon interface and the diffusion for modulation instability. of carriers from the waveguide core. In order for the silicon photonic components to remain A more advanced scheme for carrier removal was to optically independent from the bulk silicon of the wafer on integrate the waveguide into the intrinsic region of a PIN which they were fabricated, it was necessary to have a layer diode, which was reverse biased so that the carriers were of intervening material. This was usually silica, which has a attracted away from the waveguide core. A more sophisticated much lower refractive index (of about 1.44 in the wavelength scheme still, was to use the diode as part of a circuit in which region of interest), and thus light at the silicon-silica interface voltage and current are out of phase, thus allowing power to had (like light at the silicon-air interface) total internal be extracted from the waveguide. The source of this power reflection, and remained in the silicon. This construct is known was the light lost to two photon absorption, and so by as silicon on insulator. It was named after the technology of recovering some of it, the net loss (and the rate at which heat silicon on insulator in electronics, whereby components were is generated) could have been reduced. built upon a layer of insulator in order to reduce parasitic As is mentioned above, free charge carrier effects could also capacitance and so improve performance. be used constructively, in order to modulate the light. B. Kerr nonlinearity E. Raman effect Silicon has a focusing Kerr nonlinearity, in that the Silicon exhibits the Raman-effect, in which a photon is refractive index increases with optical intensity. This effect exchanged for a photon with a slightly different energy, was not especially strong in bulk silicon, but it was greatly corresponding to an excitation or a relaxation of the material. enhanced by using a silicon waveguide to concentrate light Silicon’s Raman transition was dominated by a single, very into a very small cross-sectional area. This allowed nonlinear narrow frequency peak, which was problematic for broadband optical effects to be seen at low powers. The nonlinearity phenomena such as Raman amplification, but was beneficial could have been enhanced further by using a slot waveguide, for narrowband devices such as Raman lasers. Consequently, in which the high refractive index of the silicon was used to all-silicon Raman lasers were fabricated. confine light into a central region filled with a strongly nonlinear polymer. VI. ESSENTIAL COMPONENTS In order to”siliconize” photonics, six main building blocks were used: © 2011 ACEEE 28 DOI: 01.IJCOM.02.03. 524
- 4. ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011 • An inexpensive light source When the silicon and the indium phosphide based material • Devices that can route, split, and direct light on the silicon were heated and pressed together, the two oxide layers fused chip them together [7]. When the voltage was applied, the light • A modulator to encode or modulate data into the optical generated in the indium phosphide based material passed signal directly through the glue layer into the silicon waveguide • A photo detector to convert the optical signal back into below, which acted like the laser cavity to create the hybrid electrical bits silicon laser. The design of the individual silicon waveguides • Low-cost, high-volume assembly methods was critical in determining the performance of the hybrid • Supporting electronics for intelligence and photonics control silicon laser, and would allow future versions to be built that generate specific wavelengths. C. Silicon Modulators One key problem was making high speed modulators, since silicon was not electro-optic due to its inversion symmetry [10]. This leaves carrier-optic and thermo-optic effects as the primary mechanisms for modulation in silicon and due to the relatively long carrier lifetime in silicon are slow under carrier injection. The first demonstration of a Gigabit modulator was in 2004, utilizing an MOS based structure under depletion. Progress in this area has been very rapid by many groups around the world. A recent, exciting advance is the 40 Gbit/s silicon modulator [12] demonstrated by Intel. Figure 1.building blocks in silicon photonics The key challenges that were met in the development of silicon photonics are: Light source. Silicon waveguide. Silicon modulators. Photo detector. A. Light source Figure 2. A 1 Tbit/s transmitter made out of silicon technology Silicon is an inefficient light emitter because of a D. Photo detector fundamental limitation called an indirect band-gap. An indirect The use of Ge was important because, unlike Si, it could band-gap prevents the atoms in silicon from emitting photons efficiently detect light in the near infrared which was the in large numbers when an electrical charge is applied. Instead, standard for communications. The drawback was that so much silicon emits heat. Silicon is capable of routing, modulating, stress was developed in pure Ge films deposited on Si that and detecting light; silicon has needed an external light source defects were introduced near the Ge/Si interface. Careful to provide the initial light. These external light sources are design and processing were needed to minimize the impact generally discrete lasers and require careful alignment to the of these defects on the electrical performance of the device. silicon waveguides [3], [7]. The problem is that accurate This was now a different type of Ge/Si photo detector that alignment is difficult and expensive to achieve. Even has built-in amplification, which makes it much more useful submicron misalignment of the laser to the silicon waveguide in instances where very little light falls on the detector. It is can render the resulting photonic device useless. A long- called an avalanche photo detector. Because an avalanche standing quest in silicon photonics has been the creation of process occurred inside the device. In order to realize the full a laser source that can be manufactured directly on the silicon performance potential from this materials, it was needed to photonic chip, in high volume, and whose emitted light is further reduce the dark current that was coming from the automatically aligned with the silicon waveguide. defects at the Ge/Si interface, and stop the inter diffusion of B. Hybrid Silicon Wave Guide Ge and Si that occurred during annealing [7]. This intermixing was problematic since the Ge caused higher noise than if the The silicon substrate was the base upon which the other silicon alone was in the multiplication region. If we are items were placed. On this substrate rests the silicon successful, this work will pave the way for developing low waveguide [14]. Both the substrate and the waveguide were cost, CMOS-based Ge/Si APD operating at data rates of manufactured using standard silicon fabrication processes. 40Gbps or higher in the future. © 2011 ACEEE 29 DOI: 01.IJCOM.02.03. 524
- 5. ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011 VII. APPLICATIONS today’s level of 100 W/cm2 challenges even the most advanced packaging technologies. The problem of heat Silicon photonics has its wide range of applications in dissipation was so severe that it threatens to bring to halt the Optical communication. continued advance of the technology, as described by Data com and telecom applications. Moore’s law. This fact was highlighted by the recent shift of VOA’s. the microprocessor industry away from increasing the clock ROADM’s. speed and in favor of multi-core processors. Among photonic Silicon Triplexer. components, lasers were the most power hungry photonic Ring resonators. devices [1],[3]. Optical shifter /mirror. The lack of an electrically pumped Si Laser, to date, dictates Optical multi channel separating filter. an architecture where the light source remains off-chip. Far Modulated Raman laser. from being a compromised solution, this architecture was in Raman amplifier. fact preferred as it removed a main source of heat dissipation. Wavelength converters. Furthermore, the performance degradation of injection lasers Splitters and couplers. at high temperatures will be a major obstacle to their Attenuators and deflectors. integration onto the hot VLSI substrate [1]. VIII. CHALLENGES IX. PROPOSED SOLUTION TO OVERCOME THE CHALLENGES A. Compatibility-quad Instead of using silicon wafer if we build it with the gallium “Compatibility-quad” describes conditions that must be nitride of two inch in size which is almost 1/6th the diameter of met by silicon photonics to be fully compatible with VLSI standard silicon wafer can be used in producing violet laser electronics. [4]. At the same time we can have double digit rate of growth, if the size of gallium nitride is made 100mm or more. It can be used to conduct heat far better than silicon. By making substrates from gallium nitride we can provide better foundation for diodes and transistors [4]. Gallium nitride families can produce LED’S with outputs ranging from the ultraviolet to infrared. Using gallium nitride as wafer is one of the key challenges i.e., heat dissipation can be eliminated since it won’t succumb below a temperature of 2225 æ%c Figure 3. Compatibility–quad describing conditions that must be and pressure of 64000 atmospheres. This GaN structure is met by silicon photonics to be fully compatible with VLSI created through a combined approach of a layer-by-layer electronics template fabrication technique and selective metal organic B. Material compatibility and process compatibility chemical vapor deposition (MOCVD). These GaN 3DPC The vision of silicon photonics has been the introduction exhibit a stacking direction band gap characterized by strong of photonics into silicon CMOS manufacturing process, thus optical reflectance between 380 and 500 nm. By introducing a silicon is process compatible. Silicon can guide light through “line-defect” cavity in the fifth (middle) layer of the 3DPC, a it because it is transparent at the infrared Wavelengths at localized transmission mode with a quality factor of 25–30 is which optical communication systems operate so; photonics also observed within the photonic band gap is material compatible [1]. C. Economic compatibility X. CONCLUSION The external light sources were generally lasers and they This paper has been an attempted overview of the silicon required careful alignment to the silicon wave guide. The photonics and its application areas, the current state of device problem here was that accurate alignment was difficult and technology, and the challenges that lie ahead on the path to expensive to achieve. Even submicron misalignment of the commercial success. Silicon photonics is on the verge of laser to the silicon waveguide could render the resulting becoming a viable technology for various applications photonic device useless [1]. This required expensive especially in communication and internet. Already, commercial processing methods for accurate alignment of lasers to the components such as optical transceivers are available based silicon waveguides. on the technology. “Economic” and “heat” compatibility with D. Heat compatibility silicon microelectronics are the main challenges ahead. Heat compatibility requires that photonic devices must be able to operate on the hot VLSI substrate and that their own power dissipation should be negligible. For photonics to merge with VLSI electronics, it cannot significantly increase the chip area. Increase in power density of VLSI chips, where © 2011 ACEEE 30 DOI: 01.IJCOM.02.03. 524
- 6. ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011 REFERENCES [10] J. Spector, M. W. Geis, M. E. Grein, R. T. Schulein, J. U. Yoon, et al , “High-speed silicon electro-optical modulator that can [1] Bahram Jalali, and Sasan Fathpour Silicon Photonics, be operated in carrier depletion or carrier injection mode,” 2008 DECEMBER 2006. Conference on Lasers and Electro-Optics (CLEO), (2008). [2] Mehdi Asghari, Silicon Photonics: A Low Cost Integration [11] G. T. Reed, W. R. Headley, G. Z. Mashanovich, F. Y. Gardes, Platform for Datacom and Telecom Applications . D. J. Thomson, M. M. Milosevic, “The Evolution of Integration,” [3] Richard Soref, The Past, Present, and Future of Silicon Photonics. in Silicon Photonics for Telecommunications and Biomedical IEEE JOURNAL NOVEMBER/DECEMBER 2006. Applications, Eds. Bahram Jalali and Sasan Fathpour, Taylor & [4] The crystal that will change everything published on 3 August Francis Books, Inc., ISBN 1439806373, 2010 2010 in spectrum.ieee.org. [12] M. J. Shaw, J. Guo, G. A. Vawter, S. Habermehl, C. T. Sullivan, [5] Benoit reid and Kelvin and Prosyk, Integrated loop mirror “Fabrication techniques for low loss silicon nitride waveguides,” Published on June 2005. Proc. SPIE [6] Amir H. Atabaki, Siva Yegnanarayanan, Babak Momeni, Ehsan [12] D. J. Thomson, F. Y. Gardes, Y. Hu, G. Mashanovich, M. Shah-Hosseini, et al, Implementation of a Coupling- Fournier, P. Grosse, J-M. Fedeli, and G. T. Reed, “High contrast 40 Tunable Resonator for Efficient High-Bandwidth Nonlinear Silicon Gbit/s optical modulation in silicon,” Optics Express, vol. 19, pp. Photonics Applications. 11507-11516, 2011 [7] V. Raghunathan and B. Jalali, “Stress-induced phase matching [13]D. Thomson, F. Y. Gardes, G. T. Reed, G. Rasigade, M. Ziebell, in silicon waveguides,” presented at the Conf. Lasers and Electro- D. Marris-Morini,et al, “Silicon optical modulators for high data Optics (CLEO), Long Beach, CA, May 2006, Paper CMK5 rate applications,” 8th IEEE Group IV Photonics conference, [ 9] P. Dumon, W. Bogaerts, Fibre-Array Optical Interconnection London, UK, 14-16 September 2011 for Silicon Photonics, Proc. ECOC, Vol. 5, p. 93 (P.2.21), Brussels, [14]N. M. Wright, D. J. Thomson, K. L. Litvinenko, W. R. Headley, Belgium (2008) A. J. Smith, et al, “Free carrier lifetime modification for silicon waveguide based devices,” Optics Express, vol. 16, pp. 19779 - 19784, 2008 © 2011 ACEEE 31 DOI: 01.IJCOM.02.03. 524