A new magnetic transistor switches current ten times more strongly than silicon chips while operating at lower energy, and also integrates memory.
As an alternative to silicon chips, MIT engineers have developed a transistor which is 10x stronger at transistor switching and built-in memory, using Chromium Sulfur Bromide, a 2D magnetic semiconductor.
This device controls electron flow by manipulating magnetic states, and can operate at lower voltages than traditional silicon transistors thus reducing energy consumption.
It operates using an external magnetic field to alter states, though research aims for fully electrical control. It also provides inherent memory functionality, allowing data retention without continuous power, which could reduce circuit complexity.
The transistor addresses silicon’s physical constraints, including the minimum voltage needed for switching. By utilising quantum tunneling and electron spin properties, it achieves current amplification ten times and switches between high and low resistance states with reduced power requirements. The material, chromium sulfur bromide, is air-stable and integrates with silicon substrates, with layers measuring a few tens of nanometers thick.
The fabrication process involves patterning electrodes on a silicon base and transferring the magnetic material via mechanical exfoliation. This reduces contamination by avoiding solvents to maintain a clean interface.
Early experiments used external magnetic fields to toggle the state, requiring less energy than silicon transistors. Later tests showed electrical currents could also control the state, which is necessary for integration into chips without relying on external magnets.
The magnetic transistor’s ability to retain information means that logic and memory functions can be combined. Conventional designs require separate transistors and memory cells. A single component capable of both tasks could reduce system complexity and accelerate data processing in future electronic designs.
Current prototypes are limited to lab scale and require external fields for some functions.
