Optical components

Silicon Photonics Advance PIC Possibilities

Silicon photonics will be the hottest trend in optical components in 2008, according to a new report from Heavy Reading.

The report, Photonic Integration & the Future of Optical Networking, says silicon photonics processes could help optical component and system manufacturers develop photonic integrated circuit (PIC) technology.

Companies attempting to integrate multiple discrete optical components on a single PIC need considerable upfront costs to develop an initial product. According to Heavy Reading analyst Sterling Perrin, the report's author, the greatest barrier to entry is the cost of setting up and running the in-house manufacturing facilities that are necessary for PIC development. (See Optical Vendors Face PIC Challenges.)

"As long as in-house manufacturing remains critical to the PIC sector, the barriers to entry will remain very high, and likely insurmountable, for venture-funded startups," Perrin writes. "And, as long as PICs require exotic materials to produce, such as indium phosphide (InP) today, it is likely that in-house manufacturing facilities will remain critical to the PIC development process."

So far, only one optical vendor, Infinera Corp. (Nasdaq: INFN), has successfully marketed optical systems based on PIC technology. Other vendors, such as JDSU (Nasdaq: JDSU; Toronto: JDU) and NeoPhotonics Corp. (NYSE: NPTN), are developing their own PICs, but they are still in the early stages of the process. (See JDSU Gets Tunable & Pluggable and NeoPhotonics Targets 40-Gig.)

According to Perrin, Infinera has about a four-year jump on the rest of the industry. (See Infinera Unveils PIC Road Map and Infinera Obliterates Q4 Estimates.)

But there is an alternative to indium phosphide-based PICs, as silicon photonics show immediate promise as a low-cost test ground for integrated photonics, as noted in this special LRTV report last month:

Developing PICs using silicon photonics is more cost effective than the indium phosphide-based process because they can be manufactured in existing facilities.

"The benefits are in economies of scale and the ability to use the same commercial facilities in which electronic ICs are produced today," Perrin writes in his report.

As a result, there is a tremendous amount of activity involving silicon photonics, with products already being shipped.

That activity should ramp up in 2008, with the first commercial shipments of silicon photonics-based modules and transponders, according to the report.

Startups Lightwire Inc. , Luxtera Inc. , and Kotura Inc. have emerged as front-runners in the silicon photonics market. All three are building silicon photonics components using complementary metal-oxide-semiconductor (CMOS) technology.

While Lightwire and Luxtera have adopted a fabless semiconductor model, Kotura has built its own CMOS facility.

Perrin writes: "Each of these companies is moving into commercial production in 2008, so the silicon photonics market will emerge during this year, with immediate applications awaiting these products in the data center, as well as in fiber access."

The one current downside to silicon photonics-based components is their reach. For now, applications based on silicon photonics will primarily be in the data center, where they will be used to connect servers, switches, routers, and computers that are within 300 meters of each other. Perrin writes that he is "also seeing silicon photonics-based products moving quickly into telecom access applications in the sub-10km range, with commercial products planned for 2008."

While the reach of silicon photonics-based products could be extended further into the network, it could be some time before they can compete with the capabilities of indium phosphide-based PICs, which are at the heart of Infinera's long-distance optical transport products.

"We expect that silicon photonics will move to longer distance applications in the metro," writes Perrin. "InP PICs, meanwhile, can achieve long distances today but costs aren't right for metro or access. Over time, we believe that InP PIC vendors will move their products downstream," adds the analyst.

— Ryan Lawler, Reporter, Light Reading

ninjaturtle 12/5/2012 | 3:44:49 PM
re: Silicon Photonics Advance PIC Possibilities I would guess based on my experience ~ 2+ years and that's if you are not using an ASIC within your system HW. If you need a Framer and ASIC add another 12 months. INFN uses custom ASICs for there solution. Anything less would have failed or simply not been doable. Broadcom, AMCC etc is not a solution to get desired outcome.
ninjaturtle 12/5/2012 | 3:44:49 PM
re: Silicon Photonics Advance PIC Possibilities These guys talk like they think that Silicon PICs are going to revolutionize the optical space.

PIC are not discrete components that you can sell as a commodity component. In order for the "System" to work you need to apply a tremendous amount of other systems requirements such as hardware and software interfaces that complement the PIC. But it makes for good reading. The applied and commerical acheivements will be the speed bump they will encounter and realize it takes more to make a real PIC then just saying you can integrate circuits on Silicon. They have been doing that for 30 years. Optical components is another WORLD.
[email protected] 12/5/2012 | 3:44:49 PM
re: Silicon Photonics Advance PIC Possibilities Ninja

The reality is that a number of companies have the system s/w written, (the biggest obstacle) and now just require a solution that will span the distances required. Not a trivial task, but also not an insurmounatble one.

Any gurus want to guess time / effort needed to integrate a silicon PIC into an existing, or new system box based on existing s/w / backplane?

Probably showing my ignorance but Hopefully I will learn something.

Sign In