How to Effectively Integrate Silicon Photonics into PCBs?

By Embedded Advisor | Thursday, September 26, 2019

Emerging silicon photonics technology extends a lot of coveted features in computing and communications, including sheer performance, reduced power, and overall increases in bandwidth. However, challenging situations often arise while integrating silicon photonics into PCB Design.

FREMONT, CA: Silicon is transparent at about 1550 nm wavelengths. As a result, Silicon is easily compatible with optical fiber networking systems which run at about 1550 nm. What happens due to this is that there is a noticeable lack of detectors or light sources in silicon photonics made directly from Silicon. The feature of Silicon to be used as the indirect band gap semiconductors is what causes this.

Integrating a detector and light source directly onto a silicon EPIC need bonding of a Ge Layer or an III-V semiconductor directly onto the Silicon. This bonding of III-V materials onto Silicon extends its own set of technical challenges.

If III-V material is utilized as a source of light and detector, then interfacing them with fiber optic requires a conversion between about 2 microns and a wavelength of 1550 nm. For this, a standard 1550 nm transceiver is to be placed on the board. The determining aspect with respect to limiting data rate will be the III-V material, or also this transceiver itself.

Considering a wavelength of about 1550 nm in the EPIC, this would require placing conventional photodetectors besides narrowband infrared LED light sources or laser diodes alongside the EPIC. The prominent challenges, here, include manufacturing and assembly. Further, each of these parts also consumes additional board space for each of the EPICs.

An important aspect to consider is the most imperative strategy for effectively integrating EPICs onto PCBs. A source of light, which is coupled with silicon EPICs, needs to have faster response time for being compatible with the quickest electronics logic approach.