Controlling Atomic Spins Electrically for Next-Gen Energy-Efficient

Computing Power is Booming, But at a Major Energy Cost

As artificial intelligence and other computing needs rapidly expand worldwide, the massive energy demands of running computer systems has become a critical issue. Scientists are racing to develop new energy-efficient computer technologies that can drastically reduce power consumption.

One promising approach uses magnetic materials instead of traditional silicon transistors to represent the 0s and 1s of digital data. Switching the magnetic orientation “up” or “down” can store information just like flipping a transistor on or off.

A new class of ultra-thin “two-dimensional” magnetic materials, composed of atomically thin layers, offers improved scalability and energy efficiency compared to traditional bulk magnetic materials. However, a key challenge has been operating these 2D magnets at room temperature and controlling them with electric fields rather than external magnets.

Now, researchers at MIT have designed a device that solves this problem. Their invention sandwiches an atomically-thin magnetic layer between two other 2D material layers. Applying an electrical pulse causes a spin-polarized current to flow between the layers, switching the magnetic orientation without a magnetic field.

“Our device enables robust magnetization switching without the need for an external magnetic field, opening up unprecedented opportunities for ultra-low power and environmentally sustainable computing technology for big data and AI,” says Professor Deblina Sarkar, the lead researcher. “Moreover, the atomically layered structure of our device provides unique capabilities including improved interface and possibilities of gate voltage tunability, as well as flexible and transparent spintronic technologies.”

The key innovation uses a 2D material with a skewed, asymmetric crystal structure that produces a special spin-polarized current flow. This spin current can flip the magnetic layer’s polarity when combined with the adjacent 2D magnetic layer.

Such energy-efficient magnetic memory and logic devices could drastically reduce the energy footprint of AI, big data processing and other compute-intensive applications as computing demands continue to soar. The MIT team’s breakthrough paves the way for commercializing this transformative “green computing” technology.