You are on page 1of 3

Silicon Photonics and their Applications

The main question to start with while thinking about photonics is What is Light?. When
we think about light, its color but, what is color?, color is the wavelength of an
Electromagnetic wave, some of it we can see and some of it we cant. Electromagnetic waves are
nothing but oscillation electric and magnetic fields. These fields are made up with photons which
are little bundles of energy and those photons each have discrete energy shown by the equation
which relates to the wavelength of the photon (E = hc/l).
The concept of Photonics and photonics signal processing has been around for a long
time, a several decades but, it hasnt really gained a lot of ground for a couple of reasons.
Traditionally, photonics processing required a lot of exotic processes. Silicon Photonics is
making photonic integrated circuits on silicon using CMOS process technology in a CMOS fab.

Why Silicon Photonics?


o Integrate photonics with Electronics
Same wafer
Bump bonding of silicon PIC with silicon IC
1. Same coefficient of thermal expansion
3D Stacking.
o Reduce cost by going into lager diameter wafers (300 mm)
o Reduce cost by sharing VLSI facility with electronics
o Improve yield by taking advantage of silicon process development
o Volume driver: Solve IC interconnect bottleneck (from 4 Tbps to 1 Pbps).

The above graph from Oracle shows from the last 10-15 years the technology has gone
from 100gigabits to 10 terabits per second and will continue expanding in fact google also made
a comment that they could make chips with much faster switching capacity but they limited by
the IO capacity. The graph also shows the number of pins per package, by using silicon
photonics it could be further increased and improve the performance

Applications

Communications: Over the past four decades, Moores law has driven the development of the
microelectronics industry by continuing to advance device performance while lowering their
cost. With continued advancement of Moores law and particularly with the introduction of
multi- and many-core architectures, future microprocessors will become faster and more
powerful. With this comes the ever increasing demand and constraints on the electrical
interconnects which feed data into these future microprocessors.
One of the possible solutions to delivering high bandwidth is by moving to an optical
interconnect with its enormous bandwidth improvement over copper without the constraint of
distance. Although optical technology has been successfully employed for long-haul fiber
communications, its application to data center, rack-to-rack or board-to-board communications is
hampered due to the relatively high cost of this technology. Developing a cost effective
technology is essential for the adoption of optical technology into the computer industry. As
shown in Fig, the electrical-to-optical interconnect transition is expected to be a gradual process,
dependent upon specific applications and cost performance trade off with copper based
technologies. Copper signaling over short distances will continue to improve with data rates
today of 20Gb/s demonstrated over 6 inches. This will result in silicon based optoelectronics first
penetrating in the enterprise and then derivatives of this technology could increasingly penetrate
to shorter distances as the technology matures.

Benefits of silicon photonics

The principal benefit of the hybrid silicon laser is that silicon photonics components no
longer need to rely on aligning and attaching discrete lasers to generate light into a silicon
photonic chip. In addition, dozens and maybe even hundreds of lasers can be created with a
single bonding step. This has several advantages:
The laser is compact so it allows many lasers to be integrated on a single chip. The first
demonstration hybrid silicon laser is only ~800 microns long. Future generations will be
significantly smaller.
Each of these lasers can have a different output wavelength by simply modifying the silicon
waveguide properties without having to modify the Indium phosphide based
material.
The materials are bonded with no alignment and are manufactured using high volume, low cost
manufacturing processes.

Conclusion

As Moores Law continues to push microprocessor performance, and as increasing


volumes of data are sent across the Internet, the demands placed on network infrastructure will
increase significantly. Optical communications and silicon photonic technology will allow
enterprises to scale bandwidth availability to meet this demand. In addition, due to the low cost
of silicon solutions, servers and high end PCs might one day come standard with an optical port
for high bandwidth communication. Likewise, other devices will be able to share in the
bandwidth explosion provided by the optical building blocks of silicon photonics. By creating
the PIN device to sweep away free electrons in silicon waveguides, Intel delivered a significant
breakthrough: a silicon component that can create continuous-beam Raman lasers and optical
amplifiers.
Research into silicon photonics is an end-to-end program that pushes Moores Law into
new areas. It brings the benefits of CMOS and Intels volume manufacturing expertise to fiber-
optic communications. The goal is not only achieving high performance in silicon photonics, but
doing so at a price point that makes the technology a natural fit even an automatic feature for
all devices that consume bandwidth. Intels breakthrough continuous silicon Raman laser will
undoubtedly contribute to the reality of this vision.

You might also like