As the first 20nm parts are announced, how much longer will cmos survive?

1 min read

Part of New Electronics' content during the past year has focused on the future for silicon manufacturing. There's good reason for this; it's becoming harder and harder for companies to design and manufacture chips at the leading edge.

Quite when the physical wall for cmos will be encountered is hard to determine; continuing innovations are pushing that day a bit further into the future. But there is a school of thought which says that 2020 might see the end of cmos scaling. Understandably, there's a lot of interest in what comes after cmos. One of the technology words of the year has been graphene; it's been hard to find a day when some more research on the so called 'wonder material' was not published. Graphene has all kinds of attractive properties that make people think it is the natural successor to silicon. But few, if any, people understand how graphene based electronics devices might be manufactured. In the meantime, the industry continues to be propelled by Moore's Law; 28nm devices are now being manufactured and companies at the leading edge are implementing designs on 20nm lines – IM Flash, for example, is sampling 20nm NAND parts early in 2012. But as the dimensions get ever smaller, the cost rises steeply. Already, foundries like TSMC have concluded the only way to make these parts economically is to move to 450nm wafers. Another school of thought suggests the industry could stop scaling at, say, 14nm, maintaining cmos production and allowing Moore's Law to be satisfied by moving in the Z direction. Is that likely? Probably not; engineers are always intrigued by what's 'out there' and what better way to move the industry forward than by exploring the world of quasiparticles, excitons, superposition and other quantum concepts opened by materials such as graphene?