A while back I wrote an article for H+ on the five most common errors I saw being made in many future predictions, one of which I called linearism, i.e. the assumption that any given technological development requires a linear path to development, proceeding from step A to step B to step C and so on. It’s an easy assumption to make because despite the parallel processing our brain uses, we tend to view things linearly because of our perceptions of time. But this is a bad assumption because technology doesn’t actually get developed in a linear fashion. While Technology B does often follow from Technology A, it also quite frequently takes paths that might seem completely random, and sometimes even seems to proceed backwards before leaping forwards again.
And right now, one of those “leaps backwards” could be an answer to making highly energy efficient nanoscale processors. As a recent EETimes article points out, some researchers are investigating the use of nanoscale relays — mechanical switches of a kind rarely used since the development of transistors — as a means to overcome the problem of “leakage current” i.e. electrical power wasted in an electronic circuit when a transistor is “off” but still transmitting considerable electrical currents due to various “breakdowns” in the materials used to “insulate” the circuit. As transistors grow smaller, this problem has grown to a point where “leakage current” can amount to nearly half of a circuits’ overall energy usage. It’s also one of the current “hurdles” to making fully functional graphene electronics. While there have been massive improvements in the “on/off” modes of electrical flow in graphene, and it has reached a stage where it can potentially be used for making processors, the ultimate goal of any nanoscale logic circuit is to have zero leakage when a switch is “off”. Nanoscale relays could make that possible.
How? By making an actual physical connection necessary for current to flow. Nanoscale relays use an “arm” that can be bent a tiny amount. This arm passes over a “gate” which can use an electrostatic charge to draw the arm down to touch a “drain” in order to make an electrical connection. When no charge is on the gate, the arm straightens back out, and breaks the connection. In essence, this is exactly identical to the electrical forces in a transistor, but because of the actual physical “break”, it has much higher resistance to any form of electrical current flow, and therefore almost no “leakage current.”
Another advantage to such a system is that it can likely be incorporated very easily into current efforts to make graphene and other “film electronics”, and accelerate their development. I could easily see such “relays” being used between two sheets of graphene separated by a layer of boron nitride, or even between who knows how many thousands of layers to make a highly efficient 3d circuit. Additionally, such physical relays are far more resilient to radiation and heat, making them far less vulnerable to EMP or overheating.
There are still many issues that need to be overcome, such as verifying the ability of the arms to remain functional after many billions of “bends” and, of course, proving that such NEMs can integrated into “film” electronics. But it’s looking likely that by stepping back to the dawn of computing, and the enormous computers built entirely of relays like the Harvard Mark 1, we might just be able to make a giant leap forward.