Self-rectifying TiOx-based memristor with synaptic plasticity

Memristors are a promising option for achieving high-density storage and neuromorphic computing. However, the high-bias electroforming step and the sneak current arises in the crossbar array architecture impeded the further advancement of memristors. To overcome these obstacles, there is a need for in-depth research into the materials and fabrication processes of memristors. In this work, we have chosen the conventional titanium oxide as the functional layer and prepared the memristor with the structure of Pt/TiOx/Al by magnetron sputtering process. The conduction behavior of the Pt/TiOx/Al memristor is attributed to the Schottky emission. The device was an interface-type memristor with electroforming-free switching and self-rectifying effect. The conductance changes with different dynamic synaptic characteristics were demonstrated. The typical long-term potentiation (LTP) and long-term depression (LTD) were implemented by adjusting the pulse width, interval, or amplitude. The non-linearity can be adjusted by pulse parameters, and optimal non-linearity (0.167) were obtained by stepwise pulse stimulus. This study demonstrates that the Pt/TiOx/Al memristor has potential applications in high-density and efficient neuromorphic computing.