Neuromorphic computing mimics the brain's architecture to create energy-efficient devices. Reconfigurable synapses are crucial for neuromorphic computing, which can be achieved through memory-resistive (memristive) switching. Graphene-based memristors have shown nonvolatile multi-bit resistive switching with desirable endurance. Through first-principles calculations, we study the structural and electronic properties of graphene in contact with an ultra-thin alumina overlayer and demonstrate how one can use charge doping to exert direct control over its interfacial covalency, reversibly switching between states of conductivity and resistivity in the graphene layer. We further show that this proposed mechanism can be stabilized through the p-type doping of graphene, e.g., by naturally occurring defects, the passivation of dangling bonds, or defect engineering.

Relevant reference: 10.1103/PhysRevResearch.5.043147