Cobalt oxide nanoparticles embedded in borate matrix: A conduction mode atomic force microscopy approach to induce nano-memristor switching for neuromorphic applications

•Thin-film of Cobalt Oxide nanoparticles embedded in Borate matrix was prepared.•Synaptic properties via conductive mode atomic force microscopy were tested.•Analog switching is analyzed by device flux, charge, and charge-flux relation.•Optimized pulse stimuli were used to emulate the brain functions i.e. SRDP & STDP.•Learning and forgetting characteristics in the prepared synaptic device. Herein, a cobalt borate (CoBi) based synaptic device (nano-memristor) was fabricated via solution process electrochemical deposition technique, in which equally spaced nanocrystalline cobalt oxide particles were embedded in an amorphous borate (B-O) mesh. The synaptic properties across the fabricated film were investigated with the help of conductive mode atomic force microscopy (CAFM). The structural and chemical analysis of the prepared synaptic device revealed that the presence of ultrathin (≤ 2 nm) interstitial amorphous mesh of B-O is critical to introducing the reproducible analog switching characteristics caused by the gradual formation and dissolution of thermodynamically unstable filament at the confined sub-nanometer scale. The prepared device is analyzed by device flux, device charge, and charge-flux relation, confirming CoBi as an emerging material for neuromorphic computing and emulation of Hebbian learning rules. Hence, the optimized pulse stimuli were used to emulate the brain functions like spike rate-dependent plasticity, spike time-dependent plasticity, and learning and forgetting characteristics in the device. The CoBi synaptic device with the optimized film thickness of 100 nm showed analog switching characteristics with low energy consumption of 42 fj and the current in the range of ∼pA at the applied voltage sweeps of ±3.0 V. From the potentiation and depression characteristics, the nonlinearity factor (NL) for long-term potentiation (LTP) and long-term depression (LTD) are calculated as 3.15 and 3.25, respectively indicating the device's high accuracy performance. This work opens up a new avenue to engineer low-power and cost-effective nanoscale memristors to mimic brain functions. [Display omitted]