A down-and-up year for silicon photonics

A delayed introduction from Intel was followed by a boost in federal funding for the technology.

1509cim Datacenterf

Silicon photonics represent a technology of promise that researchers look upon favorably as enabling highly efficient computing. In simplistic terms, silicon photonics deploys optical (photonic) communication at the chip level, eliminating the presence of electrical, copper-based connectivity in these computing systems. The technology made headlines in the cabling industry approximately two years ago when Corning announced the Pretium Edge SiPh Cable Assembly, specifically for Intel Corporation, to support Intel's silicon photonics initiatives. When announcing the multimode fiber-optic assembly in September 2013, Corning stated, "The combination of Intel and Corning technologies is expected to help send massive amounts of data - up to 1.6 Terabits per second - at lengths up to 300 meters into and around servers in data centers."

The assembly incorporates the MXC Connector, which provides up to 64-fiber connectivity. With each fiber carrying 25 Gbits/sec of data, the entire assembly could support the aforementioned 1.6-Tbit/sec capacity. The MXC connector uses lenses to expand the light emitting from fiber endfaces, Corning explained at that time. "This light beam is then transmitted through an air gap to the mating fiber. In contrast, traditional MPO connectors rely on physical contact of mating fiber endfaces." The company added that the expanded-beam lenses within the MXC make the connector "less susceptible to optical failure due to debris on the fiber endface. This can provide critical reliability, particularly in areas of the network where connectors are moved and changed frequently, such as the electronics."

A step back

As a technology, silicon photonics has seen highs and lows over the past couple years, and in 2015 in particular. In February Intel announced it would delay the commercial introduction of its silicon photonics components. When that happened, analyst firm 451 Research commented, "Intel's big bet is that silicon photonic communications will rise in value and importance rapidly as we literally reach the limits of practical copper communications, and will play a key role in ensuring that system designers are able to exploit the continuing improvements in Intel CPUs. Intel's efforts are unique because of its leadership in silicon integrated-circuit technology and manufacturing, and its prominent position within the systems ecology. Its silicon capabilities and silicon photonics investments enable lower-cost and higher-volume parts than would otherwise be possible. Intel's current painful decision to delay its commercial introduction should ensure that the use of photonic parts in volume won't be constrained by manufacturing, cost or environmental concerns."

451 further noted, "Because Intel's focus is forward-leaning - create technology that can be used beyond the possibility of copper, and drive the cost down so the parts can be used broadly - most of the impact of [the decision to delay commercial introduction] will be measured in the years to come, not in the short term. For example, in the short term (2015), Intel won't be selling components that implement the emerging CLR4 standard (the 100G data center optical standard that Arista and Intel are driving), and the available CLR4 parts will be more costly to produce than if Intel made them, and presumably sold at a higher price as a result, in turn slowing the adoption of CLR4 technology. However, market adoption and volume ramp for 100G is expected to come in 2016, so the impact here should be minimal.

"The competitive impact of Intel parts will be more evident in uses like rack-scale systems integration, where optical links are a necessary part of the new systems. Intel's process changes won't impact the long-term trajectory of silicon photonics much, but they do push the commercial implications further out into 2016."

Public-private boost

That "down" in silicon photonics news earlier this year was followed by an announcement from the United States federal government in July of the Manufacturing Innovation Institute for Integrated Photonics. The institute combines $110 million in federal funds and $510 million in non-federal contributions to a consortium comprising 124 companies, nonprofits, and universities led by the Research Foundation for the State University of New York. When Vice President Joe Biden made the announcement on July 27, he touted it as an investment in U.S. manufacturing, among other characteristics. It also has evident funding benefits to the development of silicon photonics technologies.

The announcement stated, "Just as integrated electronic circuits allowed for advanced processing in computers and cell phones, integrated photonic components can pack even more processing power into a single chip, creating new possibilities for computing and telecommunications. An emerging technology for carrying lightwaves, integrated photonics has the potential to revolutionize entire industries - from increasing the carrying capacity of broadband communications tenfold, to creating needle-free tests for common conditions like diabetes, and to improving imaging capabilities in defense operations." Among the 55 companies listed on the institute's roster are Corning and TE Connectivity.

On the web page manufacturing.gov/ip-imi.html, the U.S. federal government lists a frequently-asked-questions document concerning the initiative. "Why invest in integrated photonics now?" is one question, which is answered in part: "In contrast to today's prevailing technology, integrated photonics manufacturing has the potential to revolutionize the speed of internet networks at much lower costs than can be attained today. Perhaps equally important is that this technology offers a path to achieve these feats while consuming less power - an essential consideration for today's power-hungry data centers. Beyond the Internet of telecommunications, integrated photonics can revolutionize areas such as medical technology and defense."

The page also asks, "How will advances in integrated photonics help data centers?" and answers: "Data center are estimated to consume 3 percent of the total power generated in the United States. 10-Gbit/sec Ethernet optical transceivers currently consume only one watt of power, compared with 10 watts for electrical transceivers … If data centers employed optical transceivers on a more widespread scale, they could achieve … dramatic energy savings. The path to dramatic energy savings is to use photonic integration and on-chip WDM [wave division multiplexing] to scale to 100 Gbits/sec and higher. In the future, solutions based on increased photonic integration, especially silicon photonics, will implement 100 Gbits/sec with less power per bit than current 10-Gbit/sec solutions. The size, power consumption, and performance of integrated photonics will make their use ideal for data center applications which require transporting data over distances less than a meter to 10 km."

As an emerging technology, silicon photonics is an example of a concept that holds significant promise while at the same time is subject to the challenges of being brought to the commercial marketplace. Through a down-and-up 2015, the promises and challenges remain.

Patrick McLaughlin is our chief editor.

More in Connectivity