Spinel CuGa2O4-based memristor enabling synaptic plasticity and associative learning for next-generation memory devices

Abstract

As conventional memory technologies face limitations in scalability, volatility, and energy efficiency, resistive random-access memory (RRAM) has emerged as a promising candidate for next-generation memory and neuromorphic computing. Among various oxide materials, copper oxide (CuO) has been studied for RRAM applications due to its simple binary structure, ease of synthesis, and inherent defect-mediated switching behavior. However, despite these advantages, CuO-based devices often suffer from limited endurance and poor retention. To overcome these drawbacks, we explore CuGa2O4, a complex spinel oxide, as a more stable and tunable alternative. The multinary composition and the spinel structure of CuGa2O4 offer enhanced control over defect chemistry and switching dynamics, leading to improved reliability and multifunctionality. The device demonstrates analog switching characteristics, including transition from short to long-term plasticity under repeated stimulation, mimicking biological synapses. Furthermore, associative learning behavior, reminiscent of the Pavlovian conditioning model, is observed, demonstrating the device's potential in neuromorphic systems.