Dual-Mode MoO _<i>x</i> /Al ₂ O ₃ Memristors with Ultra-Stable Switching Capability for High-Precision Reservoir Computing

Memristors are an innovative electronic gadget that help solve problems with the von Neumann architecture and make it easier to implement neuromorphic computing systems. Both complementary digital memristors and extremely durable analog memristors are necessary for high-performance neuromorphic computing systems. Here, Ag/MoO_<i>x</i>/Al₂O₃/TiN (A-MA-T) structure-based digital-analog integrated memristors are fabricated. The A-MA-T device's electrical characteristics are examined, exhibiting consistent digital-to-analog conversion capabilities. Then, the conversion mechanism is clarified by contrasting different types of electrodes with distinct functional layers. Interestingly, the manufactured A-MA-T gadget used conductance modulation to create artificial synapse emulation, including excitatory postsynaptic current (EPSC), inhibitory postsynaptic current (IPSC), long-term potentiation (LTP), long-term depression (LTD), paired-pulse facilitation (PPF), and spike-timing-dependent plasticity (STDP). Additionally, the physical model of A-MA-T devices is built and integrated with the RC system, which achieved a recognition rate of 94.67% on the Yale Face Data set. This study offers a framework for achieving a memristor with multifunctional integration.