Abstract
We demonstrate high-accuracy classification for handwritten digits from the MNIST dataset (∼98.00%) and RGB images from the CIFAR-10 dataset (∼86.80%) by using resistive memories based on a 2D van der Waals semiconductor: hafnium disulfide (HfS2). These memories are fabricated via dry thermal oxidation, forming vertical crossbar HfOxSy/HfS2 devices with a highly ordered oxide-semiconductor structure. Our devices operate without electroforming or current compliance and exhibit multi-state, non-volatile resistive switching, allowing resistance to be precisely tuned using voltage pulse trains. Using low-energy potentiation and depression pulses (0.7–0.995 V, 160–350 ns), we achieve 31 (∼5 bits) stable conductance states with high linearity, symmetry, and low variation over 100 cycles. Key performance metrics—such as weight update, quantization, and retention—are extracted from these experimental devices. These characteristics are then used to simulate neural networks with our resistive memories as weights. Neural networks are trained on state-of-the-art (SOTA) digital hardware (CUDA cores), and a baseline inference accuracy is extracted. IBM’s Analog Hardware Acceleration Kit is used to modify and remap digital weights in the pretrained network based on the characteristics of our devices. Simulations account for factors like conductance linearity, device variation, and converter resolution. In both image recognition tasks, we demonstrate excellent performance, similar to SOTA, with only <0.07% and <1.00% difference in inference accuracy for the MNIST and CIFAR-10 datasets, respectively. The forming-free, compliance-free operation, fast switching, low energy consumption, and high-accuracy classification demonstrate the strong potential of HfOxSy/HfS2-based resistive memories for energy-efficient neural network acceleration and neuromorphic computing.
Funding Information
  • Engineering and Physical Sciences Research Council (EP/T517793/1)

This publication has 23 references indexed in Scilit: