Chinese researchers achieve data storage with a single electron

Researchers at Fudan University in Shanghai have developed a two-dimensional flash memory chip utilizing graphene, capable of storing data at room temperature with a single electron. This innovation significantly reduces the energy required for data processing, potentially enabling artificial intelligence models to operate on mobile devices.

Current random access memory (RAM) devices are designed to store binary data and prevent it from degrading into noise, requiring substantial energy in the form of electrons. Typically, about 200,000 electrons are needed to reliably store a single bit of information. Reducing this energy consumption could facilitate more local data processing, decreasing reliance on large data centers.

The challenge of detecting a single electron in memory storage is comparable to identifying ripples from a single drop in a vast reservoir. Led by microelectronics professor Zhou Peng, the team addressed this by designing a two-dimensional flash chip using graphene, a super-thin material. This design prevents electron escape and amplifies the signal of a single electron to about 0.5 volts, compared to tens of millivolts in other systems.

Inspired by a Chinese Buddhist phrase, the design is referred to as Guiyi, meaning 'return to one.' This technique not only demonstrates single-electron data storage but also verifies programming voltage quantization and self-limiting programming. These quantum behaviors could enhance the precision of reading and writing quantum states, increasing data storage capacity on a single chip.

Zhou plans to commercialize this technology within five years, aiming to enter mass production and potentially transform the data storage industry. In addition to the 2D chip, Zhou's team previously introduced PoX, the fastest non-volatile storage, and Changying, an atomic chip integration framework. Changying enables PoX to integrate with silicon-based hardware, creating the first 2D silicon hybrid flash memory chip.

These developments pave the way for future AI models to run on mobile devices with minimal power consumption.