By Hamish Johnston at the AAAS Annual Meeting, Vancouver, Canada
Depending on how you express it, Moore’s law has held up remarkably well over the past 40 years. In particular, chipmakers have been able to double the number of transistors that can be squeezed onto a chip every two years. This explains why the mobile phone in your pocket is more powerful than the most advanced “supercomputers” of the early 1970s.
Round about 2004 however, one aspect of this exponential growth hit the buffers – it became very difficult to remove the vast amount of heat produced by all these tightly packed circuits. The result being that it is not possible to use all circuits to their full potential. This problem has the ominous moniker of “dark silicon”.
Looking into the not-so-distant future, another problem is expected to crop up when the size of insulating structures in circuits drops below about 4 nm. At this point, electron tunnelling between circuits is expected to put an end to Moore’s law.
One way round this, according to Ralf Cavin of the Semiconductor Research Corporation, is to make the electrons heavier and therefore less likely to tunnel. While this might sound crazy, an electron in a solid has an effective mass that is often greater than its mass in free space. The thing that I don’t understand about this solution is that the speed at which a transistor operates is related to the effective mass of the electrons. The heavier the electron, the slower the speed; so it seems this is at odds with sustaining Moore’s law.
Ultimately, engineers will have to look beyond Moore’s law, and that was the topic of a session where Cavin spoke at the American Association for the Advancement of Science (AAAS) meeting here in Vancouver.
Cavin is keen on in carbo electronics – devices that are based on the remarkably efficient information processing done by living organisms. The benefits, according to Cavin, are many. For one thing, biological molecules such as DNA can store data at much higher densities than the ultimate upper limit of semiconductor devices. Living systems are also highly parallel and extremely energy efficient. On the downside, living circuits are much slower than silicon.
I’m not sure when your mobile phone will contain in carbo devices, if ever, but work has begun in that direction.