Answers are grounded in this paper and cite their sources.
Memristor‐based neuromorphic systems are designed to emulate biological neural processing, however, scaling the functional layer of memristor down to nanometer regime face critical challenges. Such scaling is crucial for enhancing sensitivity and broadening the dynamic response range. Equally important, it helps reduce energy consumption and improve integration density. Herein, we demonstrate a molecular memristor array with a 1.5‐nm‐thick monolayer 2D supramolecular polymer film as the functional medium. This molecular architecture exhibits analog resistive switching characteristics with an exceptional on/off ratio of 10 5 and six orders programmable conductance range. Remarkably, a sub‐100 mV action potential was attained, enabling synaptic events with an estimated energy consumption of 0.67 fJ–an order of magnitude lower than that of biological counterparts. This breakthrough in ultralow activation potential established a new paradigm for energy reduction beyond conventional current/pulse‐width minimization. Intriguingly, voltage‐controlled transition between short‐term plasticity (STP) and long‐term plasticity (LTP) states could mimic bio‐inspired transfer learning through dynamic weight updates based on the stable LTP baselines. Thanks to the excellent dynamic response range and multi‐states, this synapse‐mimicking molecular memristor array achieve high‐precision image classification with an accuracy of 84.9% for the complex Fashion‐MNIST datasets, demonstrating a potential for energy‐efficient neuromorphic computing architectures.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: