Researchers have developed hybrid phase-change memristors that could provide fast, low-power and high-density computational memory. By strategically straining a material as thin as a single layer of atoms, University of Rochester scientists have developed a new type of computational memory that is fast, dense and uses low power. The researchers outline their new hybrid resistive switch in a study published in Nature Electronics.

Artist's rendering of a two-dimensional material strategically strained to lie between two different crystalline phases. Stephen Wu, an assistant professor at the University of Rochester, is using this material to create hybrid phase-change memristors to provide fast, low-power and high-density computational memory. Image credit: University of Rochester Illustration/MichaelOsadciw

Hybrid Resistor Switch

The approach, developed in the lab of Stephen M. Wu, assistant professor of electrical and computer engineering and physics, combines the best qualities of two existing forms of resistive switching for memory: memristors and phase-change materials. Both have advantages but also disadvantages compared to today's most common forms of memory, including dynamic random access memory (DRAM) and flash memory.

Memristors, which work by applying a voltage to a thin wire between two electrodes, tend to lack reliability compared with other forms of memory, Wu said. At the same time, phase change materials need to selectively melt a material into an amorphous or crystalline state, which requires excessive power consumption.

Breakthroughs in memory technology

The researchers combined ideas from memristors and phase-change devices to transcend the limitations of both devices. "We are making a two-terminal memristor device that drives one crystal phase into another crystal phase. The two crystal phases have different resistances, and then you can store that as a memory," Wu said.

The key lies in using a two-dimensional material that can be stretched to the critical point between two different crystal phases and pushed in either direction with relatively little force.

Engineering and collaboration

"Our engineering design essentially just stretches the material in one direction and compresses it in the other direction," Wu said. "By doing this, performance can be improved by several orders of magnitude. It seems to me that this material could eventually be used in home computers as a form of ultra-fast, ultra-efficient memory. This could have a significant impact on computing as a whole."

Wu and his team of graduate students conducted experimental work and collaborated with researchers in Rochester's Department of Mechanical Engineering, including assistant professors Hessam Askari and Sobitt Singh, to determine where and how to apply strain to the material. The biggest hurdle in making phase-change memristors is continuing to improve their overall reliability, but he's encouraged by the progress his team has made so far.

Reference "Strain engineering of vertical molybdenum ditelluride phase change memristor", author: Hou Wenhui, Ahmad Azizimanesh, Aditya Dey, Yang Yufeng, Wang Wuxiu, Shao Chen, Wu Hui, Hesam Askari, Sobhit Singh and Stephen M. Wu, November 23, 2023, "Nature - Electronics".

DOI:10.1038/s41928-023-01071-2

Compiled source: ScitechDaily