Metal-induced progressive alteration of conducting states in memristors for implementing an efficient analog memory: a DFT-supported experimental approach

Abstract

Advancement of the memristor-based artificial synapse (AS) is urgently needed for rapid progress in neuromorphic devices. The precise structural and chemical engineering of metal oxide layers by metal dopants (Ni) is presented as an innovative way to set off a decent performance of the AS. An ON/OFF ratio of 10³ as well as data retention and endurance capabilities of 10⁴ s and 10³ cycles, respectively, are achieved. With these properties, the symmetric alteration in conductance states, short-term plasticity (STP) and long-term plasticity (LTP) are realized within the same device, and compared with the reported values to establish its excellent cognitive behavioural ability. Our combined experimental and the DFT-based first-principles calculation results reveal that the rational designing of AS using metal-cations (Ni) can promote an ultra-low-power of ∼2.55 fJ per pulse (lower than human brain ∼10 fJ per pulse) for STP, promising for next-generation smart memory devices. Here, Ni endorses strong electronic localization, which in turn familiarizes trap states within the forbidden energy gap and improves short-term memory loss. Further, it modifies the local electrostatic barriers to stimulate modulatory action (as commonly observed in the mammalian brain) for LTP. Overall, this work provides a novel pathway to overcome the technological bottleneck.