Complementary Digital and Analog Resistive Switching Based on AlOₓ Monolayer Memristors for Mixed-Precision Neuromorphic Computing

Neuromorphic computing is a potential candidate to break the von Neumann bottleneck, in which the trade-off between computational precision and energy consumption remains challenging. In this brief, a complementary memristor cell based on monolayer AlOxfilm is proposed, whose two components exhibit analog (N-Si/AlOx/TiN) and digital (N-Si/AlOx/Cu) resistive switching behaviors. Typical synapse behaviors including spike time-dependent plasticity (STDP) and long-term potentiation/depression (LTP/LTD) are emulated in different working modes. Moreover, with the high/low resistive state (HRS/LRS) in the digital component, the cell shows low/high accuracies with different power consumption, which are ~82% and ~94% in the Modified National Institute of Standards and Technology (MNIST) recognition task. Our findings may provide the potential to develop energy-efficient, feasible integrating, and mixed-precision neuromorphic computing based on AlOxmonolayer memristors.