About This Webinar
Join us for a
FREE webinar!
CMOS-integrated nanophotonics (CINP) has the potential to bring traditional semiconductor industry efficiency and low-cost manufacturing to single-mode, short-reach (<2 km) optical interconnect applications, such as for data centers and advanced computing systems.
Technical solutions available for short-reach optical communications links that incorporate CINP range from a strictly hybrid approach, where the photonic and electronic elements of the transceivers are fabricated in different technologies and bonded together, to monolithically integrated solutions, where the photonic and electronic elements are on a single chip. IBM has a vision of smart partitioning of technology solutions, where the degree of monolithic or hybrid technology used is influenced more by yield, cost and performance considerations, rather than being strictly dictated by the constraints of a photonic-only technology offering. This approach has resulted in the development of a monolithic CINP technology that could nevertheless also be used for hybrid applications.
This webinar will provide an overview of monolithic front-end-integrated electronic/nanophotonic CMOS technology that has been developed at IBM for single-mode optical short-reach transceivers at data rates up to 25 Gb/s.
Presenter
Douglas Gill is a research staff member at IBM's T.J. Watson Research Center in Yorktown Heights, N.Y., where he works on electro-optic systems monolithically integrated within CMOS electronics. Other areas of his research have included high speed LiNbO
3 modulators, integrated optical amplifiers, nonlinear polymers, hybrid integrated transceivers and advanced transmission formats for cost-effective data transport.
Gill has a doctorate from the University of Wisconsin-Madison. Before joining IBM he was a postdoctoral research associate at Northwestern University and a member of the technical staff at Bell Labs, part of Alcatel-Lucent.
He was inducted into the Alcatel-Lucent Technical Academy in 2010, has more than 35 patents and has written 85 refereed journal and conference articles.
Audience questions and the presenter's responses are provided below.
1. Do you or IBM have data, or work with, others types of Modulators, like Quantum wells or Dots? And Data comparing the performance of those.
We are working with MZ and ring modulators. At present we feel that alternative modulators are not mature enough to be used in the application space we are interested.
2. Can you comment on the IBM vision on the integration of light sources on chip, and what is your view of VCSEL technology for on-chip integration and its limitations?
Our laser integration scheme has been outlined in the literature (ECTC 2015). We do not consider a VCSEL solution appropriate for single mode signal transmission at this time.
3. What is the average fiber-coupled transmitter output power for a single wavelength?
The average fiber-coupled output for a single channel from a transceiver chip is very dependent on the specifications for a given application. For example, whether an application has a maximum operation temperature range up to 120 C, versus 80C, will have a big impact on laser efficiency and output power.
4. Will the 4-laser chip with MUX operating in CW mode?
The light sources for the 4-channel transceiver chip will operate in CW mode. The on-chip modulators will then modulate the light before it is coupled into a fiber.
5. Is there a student software available to test and design the modulators?
No. Our effort is focused on commercial applications and the solutions are embedded in the PDK.
6. Is IBM working with other advanced materials, such as 2D materials?
We are not working with alternative 2D materials for our Silicon Photonics technology.
7. How could the technology be useful in data centers and LAN? Is the cost reduction you propose enough to enter those markets? Any cost ranges for the additional cost of Ge + modulators to the CMOS technology?
Our technology aims to hit the cost point given by the industry.
8. Could you talk about the prospect of mode division multiplexing in optical communication?
We do not have plans to introduce mode division multiplexing.
9. Can you comment on the hermeticity of the package and reliability concerns?
There will be a reduction in the number of hermetically sealed elements required using CMOS9WG technology, but the end decision regarding the need for any hermetic packaging is a decision that will be made based on cost and reliability by the vendor that makes a given transceiver solution.
10. (from a crystal-growing company in Germany): Do organic crystals play an important role in CMOS technology, and if yes, which kind? Is there cooperation with companies possible, who developed such CMOS technologies? Who could be possible for further exchange about organic semiconductors in CMOS technology? Could Graphene also possible as a recent or common Material for CMOS Technology?
Collaboration between IBM research and external groups or companies is always a possibility, given common goals can be defined.