• • Energy consumption of 0.471–0.218 pJ per synaptic event, approaching biological femtojoule levels; this 2–5× reduction versus typical reported artificial synapses (often >1 pJ) directly addresses the power bottleneck for large-scale neuromorphic arrays, enabling edge deployment in battery-constrained musical classification systems.
• • Hardware-implemented STDP achieves >95% accuracy within 20 epochs on a musical classification task, outperforming software-based STDP and backpropagation after 10 epochs; this demonstrates that low-power analog synapses can match or exceed digital training convergence, reducing reliance on energy-intensive GPU clusters for temporal signal processing.
• • The SiC/SiO2@Ag NWN device operates under both externally biased electric field modulation and zero-bias photoexcitation, emulating UV visual and electrical synaptic plasticity; this dual-mode operation provides design flexibility for sensor-integrated neuromorphic front-ends, eliminating separate photodetector and synapse components in optical preprocessing pipelines.
• • LSPR induced by Ag nanoparticles and one-dimensional confinement effects enable substantial synaptic responses at ultra-low currents; this material-level enhancement allows scaling to nanometer-scale junctions without sacrificing signal-to-noise ratio, critical for dense crossbar arrays where leakage currents otherwise dominate power budgets.