Solution‐Processed Thin‐Film Transistors With Tunable Temporal Dynamics for Neuromorphic Computing

Flexible and large‐area sensors promise ubiquitous, low‐cost data acquisition. Their widespread deployment, however, is limited by the power and communication demands of centralized processing. Neuromorphic electronics offer a route to on‐sensor computation, yet the role of device‐level temporal dynamics in determining system‐level performance remains poorly understood. Here, we demonstrate that the short‐ and long‐term plasticity of ion‐gated transistors can be systematically tuned through materials design in solution‐processed hybrid semiconducting films. Through combining carbon nanotube networks with poly(3‐hexylthiophene) using controlled crystallization pathways, we engineer synaptic transistors whose functional timescales are directly impacted by the film composition and morphology. Further, we show that these materials‐defined temporal characteristics directly determine neuromorphic function. For example, relatively fast CNT‐based devices are optimal for high‐frequency spiking operations, whereas hybrid devices with long ionic memory enable efficient reservoir computing and data compression, specifically for less frequent input data. Device‐informed simulations of spiking neural networks and reservoir architectures reveal a clear crossover in performance as a function of input timescale. More broadly, this work establishes materials‐controlled temporal dynamics as a foundational design parameter for neuromorphic electronics, enabling adaptive, on‐sensor intelligence without increasing circuit or algorithmic complexity.