Li‐Well ZnO Memtransistors: High Reliability for Neuromorphic Applications

K Ki‐Hoon Son (Department of Materials Science & Engineering Kyung Hee University Yongin 17104 Republic of Korea) H Hyun‐Sik Kim (Department of Materials Science & Engineering Kyung Hee University Yongin 17104 Republic of Korea) D Dae‐Hee Han (Department of Materials Science & Engineering Kyung Hee University Yongin 17104 Republic of Korea) H Hyung‐Kyu Lim (Department of Chemical Engineering Interdisciplinary Program in Advanced Functional Materials and Devices Development Kangwon National University 1 Kangwondaehak‐gil Chuncheon Gangwon 24341 Republic of Korea) H Hong‐Sub Lee (Department of Materials Science & Engineering Kyung Hee University Yongin South Korea)

Abstract

Abstract Memtransistors are active analog memory devices utilizing ionic memristive materials as channel layers. Since their introduction, the term “memtransistor” has widely been adopted for transistors exhibiting nonvolatile memory characteristics. Currently, memtransistor devices possessing both transistor on/off functionality and nonvolatile memory characteristics include ferroelectric field‐effect transistors (FeFETs) and charge‐trap flash (floating gate), yet ionic memtransistors have not matched their performance. Here a facile and extendable lithium (Li)‐well oxide memtransistor (LWOM) is reported as a promising candidate. Forming a Li well, analogous to an n + well beneath electrodes in n‐metal‐oxide‐semiconductor field‐effect transistor (MOSFET) processes, induces Li⁺‐ion migration via write V DS , achieving analog memory characteristics through Schottky barrier modulation. LWOM enables low‐voltage weight updates and precise gate‐controlled weight update characteristics. Analysis via 3D secondary ion mass spectrometry (SIMS) confirms Li‐ion redistribution and the resistance‐switching mechanism. A 21 × 21 crossbar array demonstrates 99.31% operational yield and successful weight updates to target conductance values. Fabricated using mature oxide semiconductor technology with a 230 °C thermal budget and a simple process, LWOM stands as a strong contender for next‐generation nonvolatile memory and artificial neural network (ANN) acceleration hardware.

Article Details

Volume / Issue Vol. 38, Issue 1
Published January 01, 2026
ISSN 0935-9648
Publisher Unknown Publisher

Journal Info

Advanced Materials

Unknown Publisher

ISSN: 0935-9648 Physical Sciences

Authors (5)

K

Ki‐Hoon Son

Department of Materials Science & Engineering Kyung Hee University Yongin 17104 Republic of Korea

H

Hyun‐Sik Kim

Department of Materials Science & Engineering Kyung Hee University Yongin 17104 Republic of Korea

D

Dae‐Hee Han

Department of Materials Science & Engineering Kyung Hee University Yongin 17104 Republic of Korea

H

Hyung‐Kyu Lim

Department of Chemical Engineering Interdisciplinary Program in Advanced Functional Materials and Devices Development Kangwon National University 1 Kangwondaehak‐gil Chuncheon Gangwon 24341 Republic of Korea

H

Hong‐Sub Lee

Department of Materials Science & Engineering Kyung Hee University Yongin South Korea