Memristors with programmable conductance are thought of promising for energy-efficient analog reminiscence and neuromorphic computing in edge AI programs. To enhance reminiscence density and computational effectivity, attaining a number of secure conductance states inside a single gadget is especially vital. On this work, we reveal multilevel conductance tuning in few-layer tin hexathiophosphate (SnP2S6, SPS) memristors, attaining 325 secure states by a pulse-based programming scheme. By analyzing conductive filament evolution, we devised a voltage-pulse method that successfully suppresses present noise, thereby maximizing the variety of distinguishable states throughout the gadget ON/OFF ratio. Moreover, we experimentally emulated synaptic plasticity behaviors together with long-term potentiation and despair, and validated their efficiency by synthetic neural community simulations on digit classification. These outcomes spotlight the potential of SPS memristors as high-resolution analog reminiscence and as constructing blocks for neuromorphic computing, providing a pathway towards compact and environment friendly architectures for next-generation edge intelligence.
