Left to right, Professor Jin Sung-hoon Ajitkumar, Ph.D., Researcher Kim Jin-woo

A joint research team led by Professor Jin Sung-hoon of Incheon National University revealed that the resistance-changing memory (RRAM) device based on Cesium tin Cs2SnI6 (CSI), which is attracting attention as a next-generation memory technology, shows excellent potential in the application of binary neural networks (BNNs).

From computers and smartphones to ChatGPT, an interactive artificial intelligence (AI) service, advances in modern technologies are increasing the demand for sustainability with the increase in computing power. Against this backdrop, one of the perovskite semiconductors, Cs2SnI6 (CSI)-based Resistance Change Memory (RRAM) devices, was attracting attention as a next-generation memory technology.

CSI RRAM said that existing memory technologies have shown the potential to solve several challenges that have had to be solved: energy efficiency, speed, and environmental stability.

The key to this technique is to store data using the transition between high resistance state (HRS) and low resistance state (LRS). In particular, the low variability in LRS of CSI-resistant memory devices improves the reliability and accuracy of memory, which directly contributes to the performance improvement of Binary Neural Network (BNN).

This suggests its applicability, especially in AI applications and Internet of Things (IoT) devices.  CSI RRAM also does not use harmful substances such as lead, and can maintain stable performance in the air, which can contribute to increasing environmental sustainability. These characteristics will make it possible to apply memory devices to more flexible and diverse environments, and will be of great help in developing new types of electronic devices such as wearable technology in the future, the researchers predicted. 

Professor Jin said, "This technology can provide a new turning point in the development of AI technology for a sustainable future, beyond just fast and efficient memory solutions."

The results of this study were co-authored by Ajit Kumar, Mokurala Krishnia, and Park Jin-woo of the joint research team, and professors Kim Hyung-jin and Jin Sung-hoon participated as corresponding authors, and were published in the April issue of Advanced Functional Materials (IF=19) to be recognized for their academic excellence.

Application of artificial intelligence, accuracy of BNN according to LRS and HRS deviation, and schematic diagram of the synthesis process of cesium tin iodine semiconductors