Novel Research of Ferroelectrics Offers Roadmap to Multivalued Logic for Neuromorphic Computing

The language of computers is covered in just two symbols – zeroes and ones, implying no or yes. But a world of excellent possibilities awaits us if we could expand to three or more values so that the same physical switch could encode much more data.

“Important than this, such new logic unit will allow information processing using not just ‘no’ and ‘yes’, but also ‘either no or yes’ or ‘maybe’ operations,” says Valerii Vinokur, a materials researcher and Distinguished Fellow at the U.S. Department of Energy’s Argonne National Laboratory and the corresponding author of the paper, along with Laurent Baudry with the Lille University of Technology and Science and Igor Luckyanchuk with the University of Picardie Jules Verne.

It is the way brains operate, and they are something on the order of a million times more effective than the best computers we have ever managed to build while consuming orders of magnitude less energy. “Our brains process so much more information, but if our synapses were created like or present computers are, the brain would not just boil but evaporate from the total energy they use,” says Vinokur.

Figure: Theoretical map to use Ferroelectric Material

While the benefits of such sort of computing, known as multivalued logic, have long been known, the issue is that we have not discovered a substance system that could execute it. Right now, transistors can only operate as ‘off’ or ‘on’,” so this novel system would have to find a new way to consistently maintain more states, as well as be easy to write and read and ideally to work at room temperature.

Hence Vinokur and the group’s interest in ferroelectrics, a class of materials whose polarization can be regulated with electric fields. As ferroelectrics physically alter shape when the polarization changes, they are highly useful in sensors and other devices, like medical ultrasound machines. Researchers are highly interested in tapping such properties for computer memory and other applications, but the theory behind their behaviours is very much still emerging.

The novel paper lays out a recipe by which we could tap the properties of extremely thin layers of a specific class of ferroelectric material known as perovskites. According to the estimations, perovskite layers could hold two, three, or even four polarization positions that are highly stable, “so they could click into one place, and hence offer a stable platform for encoding data,” says Vinokur.

“When we realize this in a device, it will vastly enhance the efficacy of memory units and processors,” says Vinokur. “This offers a vital step towards the realization of so-called neuromorphic computing, which strives to model the human brain.” According to the Vinokur, the group is working with experimentalists to apply the principles to prepare a working system