As IT professionals know, enterprise server farms are huge energy consumers, and the larger they are, the more voracious their appetite.
Quantum computing could help because it is not only supposed to solve complex problems exponentially faster than classical computing, it’s also supposed to do so while consuming less energy. However major barriers—such as creating the extremely low temperatures required to enable superconductivity that is used in quantum-computing components—stand in the way.
Now researchers at MIT are working on a tiny device that could help enable quantum computing and drastically reduce server-farm costs. The roots of this device—made from a superconducting nanowire—stretch back to a similar concept developed in the mid-1950s by an MIT electrical engineer who died tragically young before his vision could become reality.
Let’s start with the present. A group of researchers in MIT’s Department of Electrical Engineering and Computer Science led by electrical engineering professor Karl Berggren have been tackling the problem of superconductor efficiency.
Superconductors are typically metals that conduct electricity without resistance at extremely low temperatures and are used in MRI scanners, particle accelerators, and quantum computers. Two challenges with superconductors, though, are that they are vulnerable to the effects of environmental noise, and they cost a lot to produce.
One of the reasons quantum computers are expensive is they rely on a computer switch called a “Josephson junction,” which basically is two layers of superconducting material with an extremely thin layer of non-superconducting material in between. Berggren tells MIT News that the Josephson junction “is fundamentally quite a delicate object” and thus difficult and costly to manufacture.
Berggren says he and his colleagues are working on a tiny, elegant device that could make the Josephson junction unnecessary. Their inspiration goes back 65 years.
“In 1956, MIT electrical engineer Dudley Buck published a description of a superconducting computer switch called the cryotron,” writes MIT News. “The device was little more than two superconducting wires: One was straight, and the other was coiled around it. The cryotron acts as a switch, because when current flows through the coiled wire, its magnetic field reduces the current flowing through the straight wire.” (Buck also helped invent RAM memory and an early version of the flash drive before his death in 1962.)
Buck died before his cryotron could be developed, but Berggren has built upon Buck’s ideas regarding superconductive computer switches. “The devices we’re making are very much like cryotrons in that they don’t require Josephson junctions,” he tells MIT News.
These new devices, which Berggren has dubbed nano-cryotrons in tribute to Buck, can serve as an interface between superconductors and classical, transistor-based electronics. The simplicity of Berggren’s device (“at the end of the day it’s just a wire”) should make it cheaper and easier to manufacture than Josephson junctions, and less susceptible to environmental noise.
Berggren envisions nano-cryotrons eventually being used in both superconducting quantum computers and classical computers, which could result in considerably lower energy costs.
“It’s probably not going to replace the transistors in your phone,” he tells MIT News. “But if it could replace the transistor in a server farm or data center? That would be a huge impact.”
Copyright © 2021 IDG Communications, Inc.
