Emulating Short-Term Plasticity with High Performance Self-Rectifying Volatile Memristor

Memristors have demonstrated significant potential in emulating synaptic plasticity for neuromorphic computation. However, the sneak current prevents the large-scale integration of memristors. In this study, a high-performance memristor with $\mathrm{Pt} / \mathrm{HfO}_{2} / \mathrm{Ta}_{2} \mathrm{O}_{5} / \mathrm{ZnO} / \mathrm{ITO}$ stacked structure was fabricated. The memristor exhibits volatile, gradual resistance switching behavior and self-rectifying characteristics ($10^{6}$ rectifying ratio and $10^{2} \mathbf{R}_{\text {OFF }} \boldsymbol{/} \mathbf{R}_{\text {ON }} \mathbf{r a t i o}$), effectively suppressing sneak current in crossbar arrays. In addition, the memristor was used to emulate some essential synaptic behaviors. The result shows that the device can decay back to its initial state within a few tens of milliseconds after the stimulation, similar to the short-term plasticity of biological synapses. The various short-term plasticity behaviors of the memristors such as paired-pulse facilitation (PPF), spike-width dependent plasticity (SWDP), spike-rate dependent plasticity (SRDP), have also been demonstrated.