Dual functionality of threshold and multilevel resistive switching characteristics in nanoscale HfO2-based RRAM devices for artificial neuron and synapse elements

We demonstrate the dependency of dual functionality on the operating current with threshold and multilevel switching behaviors in HfO2-based resistive memory (RRAM) devices. These devices can be used to produce electronic neurons and synapses for neuromorphic computing applications. The control of the formation and rupture of a conductive filament (CF) driven by the movement of oxygen vacancies (V-0) in a high-current (100 mu A) operated RRAM acting as synapse enables multilevel conductance states to be achieved. On the other hand, operation of the device in the low-current regime (<= 10 mu A) leads to a transition from-memory to threshold switching, which is activated only by applying voltage. This behavior is described by a weak CF composed of a few V0 created by using a Poole-Frenkel based analytical model. Thus, threshold switching in RRAM operated at a low current plays a role in generating output spikes as neurons when the accumulated inputs exceed the critical value. (C) 2017 Elsevier B.V. All rights.reserved.