Device area invariant conductance linearity in scalable silver nanostructure based neuromorphic devices with threshold activation following nociceptive behavior.

Achieving precise control over conductance linearity and low-power switching across different device geometries remains a significant challenge in neuromorphic computing research aiming to mimic efficient in-memory processing. Despite notable advancements, traditional resistive switching elements often face issues such as stochastic switching and non-linear conductance evolution, limiting precision and scalability in neuromorphic applications. This study presents a self-assembled hierarchical Ag nanostructure active material with interdigitated electrode architecture designed to achieve area-invariant conductance linearity and low-voltage threshold switching (∼0.4 V). Notably, the devices exhibit near-constant switching fields, independent of device area, highlighting an intrinsic field-gradient-driven nanoparticle migration mechanism. We investigate the influence of electrode gap on synaptic behavior, revealing a linear variation in conductance with the number of applied pulses facilitating both short- and long-term potentiation, alongside integrated neuromorphic functions like integrate-and-fire dynamics. These behaviors are attributed to the number and connectivity of filaments, which significantly impact retention dynamics. Furthermore, the devices emulate biological nociceptive responses, demonstrating allodynia and hyperalgesia upon injury, using both conductance and retention as output metrics. Operating efficiently at low current compliances and pulse amplitudes, these devices are well-suited for energy-efficient (∼100 fJ) neuromorphic systems. Thus, by elucidating the role of device geometry in controlling critical neuromorphic functions, this work paves the way for future advancements in the design and optimization of filamentary-based neuromorphic devices, contributing to developing scalable, high-performance, energy-efficient AI platforms for complex computational tasks.