Answers are grounded in this paper and cite their sources.
Nitrogen-doped vanadium oxide (N:VO x ) based memristors are promising candidates for neuromorphic applications by combining volatile threshold switching for neuron-like dynamics with non-volatile switching for synaptic weight modulation, providing a simple and efficient route towards neuromorphic hardware implementation . Here, we report the dual switching behavior of Pt/Ti/N:VO x /p + -Si devices and demonstrate their potential application in neuromorphic computing. In this work, a VO 2 phase-dominant N:VO x thin film was fabricated by converting a sputtered vanadium nitride (VN) thin film through oxygen annealing using rapid thermal annealing (RTA). Structural and chemical analyses using X-ray diffraction and X-ray photoelectron spectroscopy confirm the transformation of metallic VN into a VO 2 -dominant VO x phase with residual nitrogen. The film thickness increased from 53.38 nm to 194.06 nm due to oxidation-induced lattice expansion, which was confirmed by atomic force microscopy. In addition, density functional theory (DFT) calculations were performed to theoretically investigate the structural changes in the material during the conversion process from VN to N:VO x . The I-V measurements exhibit volatile and non-volatile switching with changes in compliance current (CC). Under a low CC of 1 mA, the device exhibits volatile threshold switching, and temperature-dependent measurements from 30 °C to 120 °C shows the insulator-to-metal transition at approximately 57.5 °C. This volatile switching behavior was then employed to experimentally demonstrate a leaky integrate-and-fire (LIF) neuron circuit, with the observed spiking response supported by LTspice simulation. In contrast, at a higher CC of 50 mA, the device shows non-volatile bipolar resistive switching with gradual conductance modulation suitable for synaptic weight modulation, which was implemented in a CNN-LeNet model achieving 95% MNIST recognition accuracy. These results highlight the potential of the N:VOₓ device for multifunctional neuromorphic computing applications.
No takes yet. Share an insight, caveat, or question.
R et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: