Electrochemical RAM insight could aid AI computation

Technology News | May 2, 2025

By Peter Clarke

South Korean and US researchers have discovered an operational mechanism within a type of non-volatile memory – the Electrochemical RAM or ECRAM – that could be used to make a faster and more efficient in-memory computing for AI.

The researchers used a tungsten-oxide material system and added terminals to three-terminal devices allow a parallel dipole line Hall-Effect measurement system. This revealed that oxygen vacancies inside the ECRAM create shallow donor states that bridge to allow electrons to move freely. The authors ascribe this to a Mott variable range hopping mechanism. This mechanism remained stable at temperatures down to 50K, the researchers reported.

ECRAM is typically constructed as a three-terminal device with the resistance of the channel modulated by ionic exchange at the interface between the channel and an electrolytic reservoir. The ECRAM has been the subject of research over a couple of decades because of its ability to support analog storage or multiple distinct levels and act as an artificial synaptic weight. However, while the use of some material systems, such as lithium, have supported research they have not provided an easy path to development and commercialization.

Now, Professor Seyoung Kim and Hyunjeong Kwak at Pohang University of Science and Technology (Postech) in South Korea and Oki Gunawan from the IBM T.J. Watson Research Center have published a paper Unveiling ECRAM switching mechanisms using variable temperature Hall measurements for accelerated AI computation in Nature Communications. In this they discuss switching mechanisms within a tungsten oxide channel with hafnium-oxide as the electrolyte layer and tungsten as the gate electrode.

ECRAM devices store and process information using ionic movements, allowing for continuous analog-type data storage. As research has moved towards transition-metal oxides to facilitate manufacture of devices in foundries, they are becoming similar to so-called Resistive RAMs. However, for both ECRAM and ReRAM understanding the behaviour of high-resistive oxide materials has remained challenging, significantly hindering commercialization.

In a statement Professor Kim said: “This research is significant as it experimentally clarified the switching mechanism of ECRAM across various temperatures. Commercializing this technology could lead to faster AI performance and extended battery life in devices such as smartphones, tablets, and laptops.”