There was some mockery of Apple in nanotech circles for branding their latest MP3 player the iPod Nano, merely, it seemed, because it was impressively thin (at least compared to my own much-loved first generation model). Rationalisations that its solid state memory was made with a 65 nm process didn’t seem to cut much ice with the sceptics. Nonetheless, what feels superficially obvious, that microelectronics companies are deeply involved with nanotechnology, both in their current products, and in their plans for the future, really is true.
This was made clear to me yesterday; I was in Newcastle, at a small meeting put together by the regional technology transfer organisation CENAMPS, in which nano academics from some northern UK Universities were pitching their intellectual wares to a delegation from Intel. Discussion ranged from near term materials science to the further reaches of quantum computing and new neuroscience-inspired, adaptive and multiply connected paradigms for computing without software.
The research needs of Intel, and other microelectronics companies, are made pretty clear by the International Semiconductor Technology Roadmap. In the near-term, what seem on the surface to be merely incremental improvements in reducing critical dimensions need to be underwritten by simultaneous improvements in all kinds of unglamorous but vital materials, like dielectrics, resists, and glues. Even to achieve their current performance, these materials are already pretty sophisticated, and to deliver ever-more demanding requirements for properties like dielectric constant and thermal expansivity will rely even more on the nanoscale control of structure of these materials. Much of this activity takes place under the radar of casual observers, because it consists of business-to-business transactions in unglamorous sounding sectors like chemicals and adhesives, but the volumes, values (and margins) are pretty substantial . Meanwhile, as their products shrink, these companies are huge and demanding consumers of nanometrology products.
In the medium term, to keep Moore’s law on track is going to demand that CMOS gets a radical makeover. Carbon nanotube transistors are a serious possibility – they’re now in the road-map – but the obstacles to integrating them in large-scale systems are formidable, and we’re only talking about a window of ten years or so to do this. And then, beyond 2020, we need to go quite beyond CMOS to something quite revolutionary, like molecular electronics or quantum computing. This is a daunting prospect, given that these technologies barely exist in the lab.
And what will be the societal and economic forces driving the development of nano-electronics twenty years out? Now, it’s the need to sell every teenager an MP3 player and a digital camera. Tomorrow, it’s going to be the end of broadcast television, and putting video-on-demand systems into every family home. By 2025, it’s most likely going to be the need to keep the ageing baby boomers out old peoples homes and hospitals and able to live independently. Robotics equipped with something much closer to real intelligence, ubiquitous sensing and continuous medical monitoring look like good bets to me.