New 'phase-change' devices can make computers 1,000 times faster

First developed in the 1960s, ‘phase-change materials' (PCMs) can switch between two types of structures to achieve very different electrical states. This means they can start off in a crystalline structure that conducts electricity and allows a current to flow freely through them, before switching to a glassy structure that acts as an insulator instead, stopping the electrical current in its tracks. These materials can achieve this switch between conductor to insulator in mere billionths of a second.

With this in mind, an international team of scientists from the University of Cambridge in the UK, the Singapore A*STAR Data-Storage Institute and the Singapore University of Technology and Design has developed PCM-based processors for computers that are capable of processing information 500 and 1,000 times faster than today's average laptop computer. And they do this while using less energy than silicon-based devices.

Today's computers, smartphones and tablets use silicon-based 'logic devices' to perform the calculations they need to run. Once the calculations are performed, their results are stored in a silicon-based memory device. Until now, if engineers wanted to make a faster computer, they would need to increase the number of logic devices that could fit in the machine, which means making each logic device tinier and tinier. But we’re now at the point where silicon-based logic devices can’t get much smaller. 

“As demand for faster computers continues to increase, we are rapidly reaching the limits of silicon's capabilities,” said one of the team Stephen Elliott from the University of Cambridge's Department of Chemistry, in a press release.

Right now, the smallest silcon-based logic and memory devices are about 20 nanometres in size - so about 4,000 times thinner than a human hair - and are constructed in layers. "As the devices are made ever smaller in order to increase their numbers on a chip, eventually the gaps between the layers will get so small that electrons which are stored in certain regions of flash non-volatile memory devices will be able to tunnel out of the device, resulting in data loss,” the team reports. "PCM devices can overcome this size-scaling limit since they have been shown to function down to about two nanometres."

If you can't make your logic and memory devices any smaller, another way of speeding up computers is increasing the amount of calculations each one can complete and store. But just as silicon devices can't get any smaller, they also can't be taught to calculate any faster than they already do. So the team from Cambridge and Singapore showed that unlike with silicon, this is possible for PCM-based logic and memory devices. And because they don't suffle information around inefficiently like silicon-based devices, PCM devices will cost less time and energy to run. 

"Ideally, we'd like information to be both generated and stored in the same place [in PCM devices],” said lead author Desmond Loke of the Singapore University of Technology and Design, in the press release. "Silicon is transient: the information is generated, passes through and has to be stored somewhere else. But using PCM logic devices, the information stays in the place where it is generated.”

The team now has to figure out how to integrate their new devices into a computer, but once they do that, we can expect faster computers that take up less space and energy than the computers we have right now.

The results of the study are published in the journal Proceedings of the National Academy of Sciences.