In‐Materia Neuromorphic Properties Induced by Locally Denatured Regions in Sulfonated Polyaniline Networks Incorporating Au Nanoparticles

Abstract Neuromorphic computation has the potential to reduce the energy and resource costs associated with conventional digital computing. Nanomaterials are highly suited for use in neuromorphic computing devices. In this study, locally‐doped sulfonated polyaniline networks incorporating Au‐nanoparticle (SPAN–AuNP networks) are investigated using multiple microscale Au electrodes. Raman spectroscopy results of the SPAN–AuNP networks indicate that the SPAN molecules are partially dedoped in the regions surrounding the AuNPs. These dedoped regions act as bottlenecks for the hopping conduction path and lead to diverse nonlinear current–voltage characteristics that are dependent on the combination of the electrodes used, suggesting percolative conduction. The bottlenecks also act as charge‐trapping sites that impart long‐term hysteresis properties and enable the realization of a sum‐of‐products function. These results indicate that the design‐less SPAN–AuNP networks exhibit neuromorphic behavior similar to that of spiking neurons and thus have potential applications in in‐materia computing.