Gradual volatile memristor–based artificial neurons with high uniformity for neuromorphic computing

Abstract

Artificial neurons are pivotal for neuromorphic hardware, but the development of compact and uniform devices remains challenging. Conventional volatile memristors suffer from abrupt switching, which hinders spatiotemporal consistency. In this study, we developed a two-terminal artificial neuron with intrinsic leaky integrate-and-fire (LIF) dynamics, eliminating the need for bulky capacitors or additional reset circuits and enabling exceptional compactness. Crucially, the device exhibited superior spatiotemporal uniformity across arrays compared to typical volatile memristors—which show abrupt transitions—achieved through gradual volatile switching. Combined theoretical and experimental analyses revealed that this behavior resulted from the controlled formation and self-rupture of pure oxygen vacancy–based conductive filaments, which were modulated by electric field and Joule heating. Neuronal dynamics, including the firing threshold and relaxation, were tuned by adjusting the input amplitude and frequency. To validate functionality, a two-layer spiking neural network leveraging these neurons was developed, which achieved 97.4% accuracy on MNIST classification, rivaling ideal LIF models even under noisy conditions. This highlights the remarkable noise tolerance of the device, which is crucial for real-world applications. This study elucidates filament-driven volatility mechanisms and establishes a scalable approach to energy-efficient neuromorphic systems, advancing the development of bio-inspired computing hardware.

Keywords

artificial neuron / compact gradual volatility / high uniformity / memristor

References

[1] LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015; 521(7553): 436-444.

[2] Furber S. Large-scale neuromorphic computing systems. J Neural Eng. 2016; 13(5): 051001.

[3] Han J, Yun S, Lee S, Yu J, Choi Y. A review of artificial spiking neuron devices for neural processing and sensing. Adv Funct Mater. 2022; 32(33): 2204102.

[4] Kwon D, Woo S-Y, Lee J-H. Review of analog neuron devices for hardware-based spiking neural networks. JSTS. 2022; 22(2): 115-131.

[5] Zhu J, Zhang T, Yang Y, Huang R. A comprehensive review on emerging artificial neuromorphic devices. Appl Phys Rev. 2020; 7(1): 011312.

[6] Duan Q, Jing Z, Zou X, et al. Spiking neurons with spatiotemporal dynamics and gain modulation for monolithically integrated memristive neural networks. Nat Commun. 2020; 11(1): 3399.

[7] Wan Q, Sharbati MT, Erickson JR, Du Y, Xiong F. Emerging artificial synaptic devices for neuromorphic computing. Adv Mater Technol. 2019; 4: 1900037.

[8] Cui Y, Yao H, Zhang J, et al. Over 16% efficiency organic photovoltaic cells enabled by a chlorinated acceptor with increased open-circuit voltages. Nat Commun. 2019; 10(1): 2515.

[9] Kim S, Konar A, Hwang W-S, et al. High-mobility and low-power thin-film transistors based on multilayer MoS₂ crystals. Nat Commun. 2012; 3(1): 1011.

[10] Jain S, Ankit A, Chakraborty I, et al. Neural network accelerator design with resistive crossbars: opportunities and challenges. IBM J Res Dev. 2019; 63(10): 1-10.

[11] Choi SH, Park S-O, Seo S, Choi S. Reliable multilevel memristive neuromorphic devices based on amorphous matrix via quasi-1D filament confinement and buffer layer. Sci Adv. 2022; 8(3): eabj7866.

[12] Ambrogio S, Narayanan P, Tsai H, et al. Equivalent-accuracy accelerated neural-network training using analogue memory. Nature. 2018; 558(7708): 60-67.

[13] Wang Z, Joshi S, Savel’ev S, et al. Fully memristive neural networks for pattern classification with unsupervised learning. Nat Electron. 2018; 1: 137.

[14] Wang R, Yang J-Q, Mao J-Y, et al. Recent advances of volatile memristors: devices, mechanisms, and applications. Adv Intell Syst. 2020; 2(9): 2000055.

[15] Cho S. Volatile and nonvolatile memory devices for neuromorphic and processing-in-memory applications. JSTS. 2022; 22(1): 30-46.

[16] Chen FT, Lee H, Chen Y, et al. Resistance switching for RRAM applications. Sci China Inf Sci. 2011; 54(5): 1073-1086.

[17] Zhao J, Tong L, Niu J, et al. A bidirectional thermal sensory leaky integrate-and-fire (LIF) neuron model based on bipolar NbO_x volatile threshold devices with ultra-low operating current. Nanoscale. 2023; 15(43): 17599-17608.

[18] Yang J-H, Mao S-C, Chen K-T, Chen J-S. Emulating nociceptive receptor and LIF neuron behavior via ZrO_x-based threshold switching memristor. Adv Electron Mater. 2023; 9(3): 2201006.

[19] Qian F, Chen R-S, Wang R, et al. A leaky integrate-and-fire neuron based on hexagonal boron nitride (h-BN) monocrystalline memristor. IEEE Trans Electron Dev. 2022; 69(11): 6049-6056.

[20] Ke S, Pan Y, Jin Y, et al. Efficient spiking neural networks with biologically similar lithium-ion memristor neurons. ACS Appl Mater Interfaces. 2024; 16(11): 13989-13996.

[21] Cao R, Zhang X, Liu S, et al. Compact artificial neuron based on anti-ferroelectric transistor. Nat Commun. 2022; 13(1): 7018.

[22] Park S-O, Jeong H, Park J, Bae J, Choi S. Experimental demonstration of highly reliable dynamic memristor for artificial neuron and neuromorphic computing. Nat Commun. 2022; 13(1): 2888.

[23] Zahoor F, Azni Zulkifli TZ, Khanday FA. Resistive random access memory (RRAM): an overview of materials, switching mechanism, performance, multilevel cell (MLC) storage, modeling, and applications. Nanoscale Res Lett. 2020; 15(1): 90.

[24] Fan X, Chen A, Li Z, et al. Metaplasticity-enabled graphene quantum dot devices for mitigating catastrophic forgetting in artificial neural networks. Adv Mater. 2025; 37(6): 2411237.

[25] Peleg M, Normand MD, Corradini MG. The Arrhenius equation revisited. Crit Rev Food Sci Nutr. 2012; 52(9): 830-851.

[26] Fang W, Chen Y, Ding J, et al. SpikingJelly: an open-source machine learning infrastructure platform for spike-based intelligence. Sci Adv. 2023; 9(40): eadi1480.

[27] Guo Z, Liu G, Sun Y, et al. High-performance neuromorphic computing and logic operation based on a self-assembled vertically aligned nanocomposite SrTiO₃:MgO film memristor. ACS Nano. 2023; 17(21): 21518-21530.

[28] Hohenberg P, Kohn W. Inhomogeneous electron gas. Phys Rev. 1964; 136(3B): B864-B871.

[29] Kohn W, Sham LJ. Self-consistent equations including exchange and correlation effects. Phys Rev. 1965; 140(4A): A1133-A1138.

[30] Kresse G, Hafner J. Ab initio molecular dynamics for liquid metals. Phys Rev B. 1993; 47(1): 558-561.

[31] Kresse G, Furthmüller J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput Mater Sci. 1996; 6(1): 15-50.

[32] Perdew JP, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Phys Rev Lett. 1996; 77(18): 3865-3868.

[33] Langreth DC, Mehl MJ. Beyond the local-density approximation in calculations of ground-state electronic properties. Phys Rev B. 1983; 28(4): 1809-1834.

[34] Becke AD. Density-functional exchange-energy approximation with correct asymptotic behavior. Phys Rev A. 1988; 38(6): 3098-3100.

[35] Rumelhart DE, Durbin R, Golden R, Chauvin Y. Backpropagation: the basic theory. In: Backpropagation. Psychology Press; 1995.

[36] Ruder S. An overview of gradient descent optimization algorithms. arXiv. 2017.

Rights & permissions

2025 The Author(s). InfoScience published by UESTC and John Wiley & Sons Australia, Ltd.