Researchers at Northwestern University in the US appear to have progressed brain-like’ computing further than ever before. The human brain – thought of as possibly one of the most efficient computers on Earth – was crafted through millions of years of evolution and has served as a benchmark for computing technology. The brain is super-fast, has almost limitless memory, starts up without ‘booting up’ and uses very little energy to run. Both academic and industrial laboratories have been working to find a way to make a computer that functions similarly. “Neurons [may] achieve very [complex] computation with [minimal] power consumption compared to a digital computer,” according to lead researcher Mark Hersam. A computer that does this might revolutionise technology.
The research is described in the journal Nature Nanotechnology. Hersam (the Bette and Neison Harris Chair in Teaching Excellence at Northwestern University’s McCormick School of Engineering) and his team believe one key to unlocking a ‘brain-like’ level of computing might lie in the use of a memory resistor, or ‘memristor’. Memristors are capable of remembering how much current has flown through them. Therefore, unlike like RAM and flash memory, memristor technology aims to be much faster and more reliable. “Computers are very impressive in many ways, however they’re [unequal in comparison] to the mind,” said Hersam.
The typical computer relies on RAM, which works very fast. However, unsaved data vanishes if power to the system is removed. Flash memory, even though it stores information when the machine is unpowered, is much slower than RAM. Memristors are thought to provide the best of both worlds – both faster and more reliable than conventional devices. “Memristors [may] be used as a memory element in an integrated circuit or computer,” Hersam justified.
Hersam explains that, “Unlike other memories that exist today in modern electronics, memristors are stable and remember their state even [without] power.” Memristors were initially proposed back in 1971. Until now, scientists have only managed to create two-terminal devices to direct one voltage channel, which is incompatible with current computing technology. The team added an extra electrode to transform the memristor into a three-terminal device, meaning it may be used in more advanced electronic circuits and systems. “With a memristor that [may] be tuned with a third electrode, we have the possibility to realise a function previously [inaccessible],” Hersam described. “A three-terminal memristor has been proposed as a means of realising brain-like computing. We are now actively exploring this possibility in the laboratory.”
The team were able to achieve this by using the single-atom-thick nanomaterial semiconductor molybdenum disulphide (MoS2). By arranging the MoS2 in different directions to the memristors, similar to the grains in wood, researchers were then able to manipulate the flow of electrical current. “Because the atoms are in [different] orientations, there are unsatisfied chemical bonds at that interface,” Hersam clarified, “These grain boundaries influence the flow of current, so they [may] serve as a means of tuning resistance.” With this unique approach, the team successfully produced a technology that might open up an entirely new realm of control and complexity for computing. When a large electric field is applied, the ‘grain’ boundaries move, creating a change in resistance. By using MoS2 instead of the conventionally used metal-oxide-metal memristor structure, the team were able to produce a tuneable, three-terminal memristive device for the first time. The technology appears to have the potential to achieve advances in technology, possibly even a ‘brain-like’ level of computing.
Which other biological organs or systems might spark technological innovation?