Despite changes afoot at their troubled semiconductor division, researchers at Dutch multinational Philips Electronics continue to develop technology that will enable the next generation of chips. They are among a handful of companies using nanoscale components to make transistors ever smaller.
“Historically the semiconductor roadmap has always been about more Moore,” said Ronald Wolf, group leader at Philips Research, referring to Intel co-founder Gordon Moore. Moore predicted that chips would need to continually shrink to meet consumer demand. “Today, instead, we’re looking at more than Moore, trying to equip silicon with added functionality.”
At the base of this development are nanowires. Nanowires appear to be a frontrunner for replacing conventional CMOS technology, said Leo Kouwenhoven, a professor at the Kavli Institute of Nanoscience at Delft University of Technology and a Philips research partner. Philips Electronics is not alone it its quest: Multinationals such as Intel, IBM and Infineon also have nano initiatives.
Philips researchers deposit gold particles onto a silicon substrate that act as seed particles out of which nanowires made of semiconductor material like gallium arsenide or indium phosphide are grown. “The beauty of our process,” said Philips Research’s Erik Bakker, “is that it finally enables us to integrate the superior electrical and optical properties of these expensive non-silicon semiconductors with mature and cheap silicon technology.”
Wolf said that multifunctional capability may allow Philips to make smaller and faster chips as well as new types of biosensors and more efficient lighting – the company’s sole focus when it was founded in 1891. Light-emitting wires on a chip could distribute the clock frequency optically instead of electrically, enabling much faster processors. They demonstrated in a Science article published last year that they could make superconducting transistors as well. These accomplishments are feasible on an industrial scale, Wolf and Kouwenhoven said, but not for another 10 years.
Philips claims that its technique can yield more than 4 billion nanowires in a single step. Image courtesy of Philips Electronics. |
Charles Harris, chief executive officer of the publicly traded venture capital firm Harris & Harris Group, pointed out that not every application is a decade away. Harris and Harris invests solely in micro and nanotechnology companies. “You have, for example carbon nanotubes in the bumpers of General Motors’ cars,” he said. “And there is Nanomix, a company we have an investment in, that is employing nanotubes in a sensor device that you can already purchase.”
Nanosys, another company backed by Harris & Harris, has been working on commercial applications for nanostructures since its founding in 2001. Nanosys co-founder Stephen Empedocles said, “We don’t look at anything that far (10 years) out. We have products that we are projecting for market introduction in the next 12 to 18 months.”
Harris said the differing timeframes are typical in industry. “Big multinationals are thinking about nanotechnology first and foremost from a defensive point of view, as they should. They have their leading market position to lose, while startup companies have nothing to lose, everything to win.”
Nanosys targets similar applications as Philips, as well as solar cells, memory chips and fuel cells. Instead of looking at single-wire devices, Nanosys produces ink consisting of billions of nanowires that its engineers coat onto a surface by a low-cost roll-to-roll process. “In a way, it’s like cheap coatable silicon,” Empedocles said. “Our technology allows you to create thin-film transistors with electronic characteristics that you would normally only find in crystalline silicon wafers.”
Empedocles said single-wire electronics have limitations. “How (do you) place the hundreds of billions of wires right where you need them to form an electronic circuit?” he said. “In our case, we don’t really worry where those transistors are because they’re everywhere and the circuit is where you apply the electrodes.”
Ideally, a single-wire circuit would assemble itself, but current self-assembly processes can only be used for simple structures such as thin films, Kouwenhoven said. That is one reason for Philips’ 10-year development projection.
Philips researchers may not have the luxury of time, though. Last September, the company announced that weak demand had prompted it to consider restructuring its semiconductor division. In mid-December, Philips announced that it planned to break out the semiconductor divisio
– Colin Ashruf