Imec researchers have described a method of creating smaller transistors using materials such as 2D black phosphorus (phosphorene):
Designers can extend Moore's Law scaling beyond the 5-nanometer node by choosing two-dimensional anisotropic (faster with the grain) materials such as monolayers of black phosphorus, according to Imec (Leuven, Belgium). Researchers from the nonprofit semiconductor research institute described their findings at the annual Imec Technology Forum, held in San Francisco on the eve of Semicon West (July 11-13).
Imec's demonstration project focused on field-effect transistors for high-performance logic applications as part of its Core CMOS program. Using co-optimization at the material, device, and circuit levels, Imec and its collaborators proved the concept using 2-D monolayers of anisotropic black phosphorus with a smaller effective mass in the transport direction. The black phosphorus was sandwiched between interfacial layers of low-k dielectric, with stacked dual gates deployed atop high-k dielectrics to control the atomically thin channels.
Imec demonstrated the co-optimization approach at the 10-nm node but says the architecture could function with sub-half volt (<0.5-V) power supplies and an effective oxide thickness of less than 50 angstroms (0.5 nm), allowing its FETs to extend Moore's Law for high-performance logic applications below the 5-nanometer node.
(Score: 2) by kaszz on Saturday July 15 2017, @02:40AM (2 children)
Make the chip design such that failed sections can be disabled or routed around?
Low working temperature might mean that computer owners has to get a better cooling engine and the main processor has to be cooled down before you can make use of it but that will probably make thermal expansion an issue. But the 0.5 V power supply is very close to the bandgap voltage of many semiconductors which can make it hard to generate and then there will be a need for a lot of current, 200 ampere?
They probably solve it. But when considering the conditions needed to make it operate. It might simply cost more than most people are willing to spend. Low yield, cryocooling, exotic power supplies and power bus connections. The answer is likely to find another way to do computing, ie optical or something even better. Maybe slower than optical but more doable.
(Score: 0) by Anonymous Coward on Saturday July 15 2017, @02:52AM
GP doesn't look like a serious post.
(Score: 0) by Anonymous Coward on Saturday July 15 2017, @04:18AM
Put the computer in outer space and beam back the p0rn - problem silved.