A new type of transistor that can make computers up to 1,000 times faster and more energy-efficient has recently been unveiled by a research team in the US.
The chip, called M3BOLT, uses magnetic and sound waves to process information instead of relying on conventional transistor switching. It was made by scientists at the University of Michigan (U-M) and the University of Wisconsin-Madison.
To develop the device, the research team received funding of up to USD 6 million from the Defense Advanced Research Projects Agency (DARPA). They believe that it could make computers 100 to 1,000 times faster.
According to the researchers, it could also overcome the limitations of traditional transistors and significantly reduce energy consumption. “The ability to harness and process information could scale tremendously,” PhD, an associate professor of materials science at U-M, and project leader, disclosed.
A transistor alternative
A computer processor has billions of transistors. They switch on and off rapidly to process data and carry out instructions. Producers have spent decades shrinking them to boost computer performance and reduce the energy consumption.
But as transistors get smaller, they become less efficient. That’s because electrons can pass through them even when they are switched off, which wastes energy. As per the team, conventional transistors could hit their theoretical energy-efficiency limits by 2037.
To overcome this challenge, the scientists designed M3BOLT using a ferroelectric crystal called lead magnesium niobate-lead titanate (PMN-PT) and a thin layer of magnetic iron-gallium alloy. The crystal changes its internal electrical orientation once an electric field is applied.
This change, in turn, influences the magnetic properties of the iron-gallium layer, and results in waves known as magnons. These can travel through the material, with very little resistance, and reduce the energy lost as heat.
Since the magnetic waves could achieve terahertz frequencies, they could enable computers to process information hundreds of times faster than the consumer gigahertz processors used nowadays.
Processing and storing data
However, the device could also process and store information. To achieve this, the researchers need to change the magnetic orientation of the iron-gallium layer, to switch between magnetic and nonmagnetic sound waves. These two states, could serve as 1 and 0 states, instead of the two voltage levels in current transistors.
Notably, the magnetic orientation would stay unchanged even after the power is switched off. This means the same material could potentially process information and retain stored data without a continuous power supply.
Credit: Marcin Szczepanski, University of Michigan Engineering
This design could save energy by reducing data transfers. “Autonomous or mobile devices could carry better computers without huge power supplies,” Heron stated in a press release.
The University of Wisconsin–Madison team will now use computer simulations to study the chip’s performance and identify potential material defects. Meanwhile, the University of Michigan team will continue to develop the materials, devices, and circuits necessary to make the technology work in real computers.
Wei LU, PhD, an engineering professor at U-M, shared that the team is currently focused on devices for low-power applications. “But imagine that someday we can build supercomputers with them,” Wei concluded. “You might not need a nuclear power plant to run a large data center.”