If they remade The Graduate today, the buzzword tossed at Benjamin Braddock wouldn’t be “plastics” but rather “photonics” - the use of light, rather than electricity, for transmission of data. The use of photons over electrons is what lets fibre optics handle high-bandwidth communications over long distances, but expense and complexity limit their use. And as researchers put greater effort into cramming more information through copper wires, photonics’ day in the sun, as it were, was always pushed further out. For mainstream applications, copper wire continues to dominate as the communications medium of choice.
But what if photonics could provide the viability of copper at the same cost? Many companies, including Intel, are researching future optical possibilities. (Ed note: In Australia, CUDOS is conducting an ambitious photonics research program, see next article.)
One startup has recently demonstrated a possible technology breakthrough that could accelerate the transition from copper to optical links by providing a silicon solution as cost-effective as copper, starting in 2006.
The battle between electrons and photons is starting in a logical place: in networking applications targeting 10-Gbit/s transmission. The end result is easy to predict: Photons will eventually win.
Photons can travel farther than electrons (in the proper medium), use less power and generate less heat. Unlike electricity, light beams can pass through other light beams without interacting, so there’s no interference. The fundamental laws of physics dictate that photonic communications should eventually replace electronics, once certain obstacles are overcome. Certainly, though, photonic components today are not as inexpensive, or as readily manufacturable, as electronic components.
Lighting the fibre
Once the solution of price and manufacturing issues reduces optical costs further, though, we should see additional fibre in the data centre - first in rack-to-rack connections, then in board-to-board and eventually in chip-to-chip. Even with reductions in the cost of 10-Gbit/s modulation and fibre connections to silicon die, the industry will need to develop standards for optical backplanes as well as high-volume manufacturing techniques for optical routing on PCBs before the migration can proceed beyond networking.
The main advantages of copper are that it’s already installed and well known and that it’s easy to work with (being a malleable metal that’s easy to cut and splice). The cost of the fibre infrastructure has always been an issue, but we may be at the point where fibre optic component costs will drop significantly. Between the limitations of the copper solutions at 10 Gbit/s and the need for future bandwidth expansion, companies need to start thinking about making the jump to optical’s light speed.
The cost problem of the fibre (versus copper wires) is not a long-term problem, because optical fibres are based on a fundamentally cheaper base material (silicon-based glass, versus copper metal). The volume of single-mode fibre produced today is already fairly high, because of its use in very long-haul applications. The expensive aspect of using fibre is attributable largely to the costs of connectors and laser drivers.
The challenges of optical communication links have been the difficulty of modulating optical lasers and the extremely precise mechanical designs required to mate the chip components to the optical fibres. One company, Luxtera, claims to have found a very cost-effective way to mate the laser to a modulation chip and the modulator to the fibre. Luxtera’s breakthrough involves mating traditional silicon-wafer technology with optical components and modulating the laser light by using silicon rather than more-exotic materials. Although silicon looks opaque to visible light, the material is amazingly transparent to infrared light. Silicon structures can be constructed that can guide and modify infrared laser light.
Silicon laser
Intel is also exploring techniques to modulate optical transmissions in its photonics lab, guide light in silicon structures and mechanically mate fibre to silicon. Although silicon’s properties are not suitable for traditional laser generation, Intel was able to create the first silicon laser, in February 2005. (See Electronics News March 2005 front cover.) So far, though, Intel’s efforts are still lab experiments, and the company has not gone on record to say when it would ship production-level solutions. In addition, the laser structure is still enormous compared to traditional logic structures.
Luxtera’s first product allows the company to drop optical modules into an existing chip infrastructure at about one tenth the cost of previous optical solutions. The company’s plans beyond these modules are unknown, but Luxtera may continue to reserve the technology for its own silicon-based physical-layer networking components.
If the company’s ultimate goal is ubiquitous optical networking down to the system-chip level, Luxtera will need to license the technology to mainstream silicon manufacturers for incorporation into high-performance microprocessors and system components.
The future really does look bright for photonic, but expect that engineers will continue to push copper technology to at least 10 Gbit/s. Scaling network data beyond 10 Gbit/s will require optical technology, because there is little hope that copper cables can support 100-Gbit/s speeds. The transition will take place at 10 Gbit/s. Networking and technology such as Luxtera’s may hasten the transition, starting in 2006.