Reconfigurable Floating Gate Memristors for High-Accuracy Neuromorphic Computing.

Feb 11, 2026 by Decheng Wang, Zihuan Jiao, Linjun Li (ACS Applied Materials and Interfaces)

DOI 10.1021/acsami.5c25387

We built a MoS2/hBN/graphene floating gate memristor that not only stores 66 linearly tunable conductance states with robust retention but, by biasing the back gate below Vth, can be reconfigured on the fly to act as native ReLU or Leaky ReLU activations—so one compact device serves as both tunable synapse and nonlinear neuron, boosting LeNet/AlexNet inference on FashionMNIST and traffic-sign data and paving a path to fully hardware native neuromorphic systems.

source S2, crossref