New materials platforms have long been an enabling technology for communications. Gallium arsenide was a prime mover in RF and high-speed processing. Lithium niobate improved optical modulators and amplifiers. Indium phosphide was to lead to integrated optoelectronics before the optical industry bust killed the dream. Not surprising then, with all the talk about nanotechnology, many in the optical communications business wonder if nanotech can improve their fortunes. Startups have sprung up in the past few years claiming to be commercializing “nanophotonics,” and the European Union has just launched a large R&D program focused on exactly this area.
Nano-engineering has been around for a while but has gained from a new kit of manufacturing tools, such as atomic force microscopes, dip-pen nanolithography, and nano-imprint lithography that make it easier to create nano-scale features. They enable developers to build devices that are smaller and cheaper and have higher-performance characteristics than those that went before. That is, for example, essentially the key selling proposition of NanoOpto, which makes a range of nano-engineered polarizers, splitters, and “waveplates” using a nano-engineering process.
NanoOpto has attracted more investment than most optical (or nanotechnology) startups and has sold into both the networking and consumer electronics industry. But not all nanotechnology applications in optical networking have lived up to their initial promise. Early hopes for quantum dot lasers appear to have evaporated because there are cheaper ways of building uncooled lasers for telecommunications.
Nonetheless, quantum dot lasers show real commercial potential in other areas. Quantum dot encryption requires them, and Toshiba claimed a breakthrough when it recently announced quantum dot light sources that could transmit single photons for this application. Quantum encryption is already being used by financial institutions, government, and defense contractors; given the patterns of the past, large decreases in price are to be expected in this sector. Even so, no one is expecting quantum encryption to be a mass market for perhaps a decade. Still less so is the closely related area of quantum computing.
It seems more likely that the first big business for nano-engineered lasers will come from chip interconnection. Until recently, the speed bottleneck in computing and telecom was the speed of the processors. In the last few years, however, the processor speed has reached a point at which the interconnections are now the limiting factor.
In response, semiconductor manufacturers have moved from aluminum interconnects to copper interconnects and are now experimenting with optical interconnection as well as exotic lower-k materials and carbon nanotubes. Optical interconnection could supply more than enough bandwidth to suck up and supply data to even the fastest processors. However, the requirements for lasers to support such an application would be very demanding in terms of size and cost. At the board level, some kind of VCSEL could be deployed, and costs could be lowered by using a polymer fiber. At the chip level, only nano-engineering of both lasers and the waveguides would work.
Another challenge is how to integrate the photonic structures with the electronics. That has inspired the budding area of silicon photonics. If silicon can be used to create both photonics and electronics, then integration is no problem. The huge knowledge base for silicon manufacturing could be leveraged into rapid market entry for optical interconnection products as well. This field has attracted the attention of the very biggest names in the semiconductor industry. Intel has built silicon modulators and lasers, and IBM has just announced a way to build lightwave circuits on a silicon chip. Smaller firms are also involved; Luxtera is building silicon modulators, for example.
Just how big is the market for nanophotonic interconnection? It’s hard to say. For on-chip applications, the lasers would have to be embedded and their value would be subsumed by that of the entire chip. But an examination of the on-board market suggests that addressable markets for nano-engineered interconnects could be huge. Consider a board with 10 devices on it. If these devices were fully interconnected, 90 lasers would be required. Given that hundreds of millions of boards like this are sold every year, we are talking about a lot of lasers here.
Meanwhile, the pervasive computing model, which firms like Intel, Motorola, and IBM tell us is the next big thing for networking, may have a role for optical nanosensor networks. This application has some legs to it commercially because nanosensors are a major beneficiary of government R&D grants, given their applications in homeland security and the military.
Nanotechnology may also ultimately help bring down the cost of 10- and 40-Gbit/sec networks. And with the FTTX sector growing fast, there will certainly be a need for low-cost components. But although some of the emergent nanophotonics firms are fussing about lowering costs for existing networking technologies, it is not clear that nano-engineering need play a role here. In the real world, equipment manufacturers will not take a leap into the dark with entirely new technologies, unless they offer overwhelmingly compelling economics. The example of quantum dot lasers given above shows how such economics are not always characteristic of nanotechnology.
The real impact of nanotechnology on optical networking will be to take it in new directions rather than just supplement existing applications. That may not be exactly what some readers want to hear, but isn’t a new start exactly what the optical networking business needs?