Field-free programmable bipolar magnetic heterostructures for neuromorphic computing

Abstract

Neuromorphic computing systems inspired by biological principles have garnered significant attention for their superior energy efficiency and computational capabilities, with memristors serving as fundamental building blocks. However, realizing biologically plausible signal processing—particularly in applications like biological vision—requires multi-state bidirectional (excitatory/inhibitory) responses to emulate synaptic modulation, a functionality unattainable with conventional unidirectional resistive memristors. Here, we demonstrate a field-free spin–orbit torque magnetic memristor based on Fe₃GeTe₂/WTe₂ heterostructures, operating within the 110–150 K temperature range, that overcomes this limitation. The thickness-dependent ferromagnetic properties of Fe₃GeTe₂ enable multi-state Hall resistance modulation, while WTe₂'s localized spin injection precisely controls magnetic domain reversals in Fe₃GeTe₂—without an external magnetic field—ensuring the compatibility with CMOS platform. This synergistic approach achieves eight distinct resistance states (3-bit precision) spanning both polarities. We demonstrate the technology's neuromorphic potential through two benchmarks: high-fidelity image feature extraction (matching software-implemented algorithms) and competitive neural network clustering, both achieved with minimal hardware overhead. Our work provides a device-level solution for implementing signed weight updates in neuromorphic hardware, opening new avenues for energy-efficient computing systems capable of complex signal processing.

References

[1] Grollier J, Querlioz D, Camsari KY, Everschor-Sitte K, Fukami S, Stiles MD. Neuromorphic spintronics. Nat Electron. 2020; 3(7): 360-370.

[2] Mehonic A, Kenyon AJ. Brain-inspired computing needs a master plan. Nature. 2022; 604(7905): 255-260.

[3] Joshi V, Le Gallo M, Haefeli S, et al. Accurate deep neural network inference using computational phase-change memory. Nat Commun. 2020; 11(1): 2473.

[4] Wang Z, Joshi S, Savel'ev S, et al. Fully memristive neural networks for pattern classification with unsupervised learning. Nat Electron. 2018; 1(2): 137-145.

[5] Chanthbouala A, Garcia V, Cherifi RO, et al. A ferroelectric memristor. Nat Mater. 2012; 11(10): 860-864.

[6] Gupta R, Bouard C, Kammerbauer F, et al. Harnessing orbital Hall effect in spin-orbit torque MRAM. Nat Commun. 2025; 16(1): 130.

[7] Li R, Song M, Guo Z, et al. In-memory mathematical operations with spin-orbit torque devices. Adv Sci. 2022; 9(25):2202478.

[8] Stoner EC, Wohlfarth EP. A mechanism of magnetic hysteresis in heterogeneous alloys. IEEE Trans Magn. 1991; 27(4): 3475-3518.

[9] Slaughter JM, Dave RW, Durlam M, et al. IEEE International Electron Devices Meeting 2005: IEDM Technical Digest. IEEE; 2005: 873-876.

[10] Ikeda S, Miura K, Yamamoto H, et al. A perpendicular-anisotropy CoFeB–MgO magnetic tunnel junction. Nat Mater. 2010; 9(9): 721-724.

[11] Myers EB, Ralph DC, Katine JA, Louie RN, Buhrman RA. Current-induced switching of domains in magnetic multilayer devices. Science. 1999; 285(5429): 867-870.

[12] Miron IM, Garello K, Gaudin G, et al. Perpendicular switching of a single ferromagnetic layer induced by in-plane current injection. Nature. 2011; 476(7359): 189-193.

[13] Shao Q, Li P, Liu L, et al. Roadmap of spin-orbit torques. IEEE Trans Magn. 2021; 57(7): 1-39.

[14] Fei Z, Huang B, Malinowski P, et al. Two-dimensional itinerant ferromagnetism in atomically thin Fe₃GeTe₂. Nat Mater. 2018; 17(9): 778-782.

[15] Tan C, Lee J, Jung S-G, et al. Hard magnetic properties in nanoflake van der Waals Fe₃GeTe₂. Nat Commun. 2018; 9(1): 1554.

[16] Huang B, Clark G, Navarro-Moratalla E, et al. Layer-dependent ferromagnetism in a van der Waals crystal down to the monolayer limit. Nature. 2017; 546(7657): 270-273.

[17] Gong C, Li L, Li Z, et al. Discovery of intrinsic ferromagnetism in two-dimensional van der Waals crystals. Nature. 2017; 546(7657): 265-269.

[18] Huang B, Clark G, Klein DR, et al. Electrical control of 2D magnetism in bilayer CrI₃. Nat Nanotechnol. 2018; 13(7): 544-548.

[19] Park SY, Kim DS, Liu Y, et al. Controlling the magnetic anisotropy of the van der Waals ferromagnet Fe₃GeTe₂ through hole doping. Nano Lett. 2020; 20(1): 95-100.

[20] Liu C, Zhang H, Zhang S, et al. Emergent, non-aging, extendable, and rechargeable exchange bias in 2D Fe₃GeTe₂ homostructures induced by moderate pressuring. Adv Mater. 2023; 35(1):2203411.

[21] Deng Y, Yu Y, Song Y, et al. Gate-tunable room-temperature ferromagnetism in two-dimensional Fe₃GeTe₂. Nature. 2018; 563(7729): 94-99.

[22] Liu B, Liu S, Yang L, et al. Light-tunable ferromagnetism in atomically thin Fe₃GeTe₂ driven by femtosecond laser pulse. Phys Rev Lett. 2020; 125(26):267205.

[23] Zhang G, Wu H, Yang L, et al. Room-temperature highly-tunable coercivity and highly-efficient multi-states magnetization switching by small current in single 2D ferromagnet Fe₃GaTe. ACS Mater Lett. 2024; 6(2): 482-488.

[24] Yang C, Huang Y, Pei K, et al. Current-controllable and reversible multi-resistance-state based on domain wall number transition in 2D ferromagnet Fe₃GeTe₂. Adv Mater. 2024; 36(18):2311831.

[25] Peng L, Yuan Y, Li G, et al. Observation of topological states residing at step edges of WTe₂. Nat Commun. 2017; 8(1): 659.

[26] Kang K, Li T, Sohn E, Shan J, Mak KF. Nonlinear anomalous Hall effect in few-layer WTe₂. Nat Mater. 2019; 18(4): 324-328.

[27] Kao I-H, Muzzio R, Zhang H, et al. Deterministic switching of a perpendicularly polarized magnet using unconventional spin–orbit torques in WTe₂. Nat Mater. 2022; 21(9): 1029-1034.

[28] Wang L, Xiong J, Cheng B, et al. Cascadable in-memory computing based on symmetric writing and readout. Sci Adv. 2022; 8(49):eabq6833.

[29] Euler T, Haverkamp S, Schubert T, Baden T. Retinal bipolar cells: elementary building blocks of vision. Nat Rev Neurosci. 2014; 15(8): 507-519.

[30] Zheng G, Xie W-Q, Albarakati S, et al. Gate-tuned interlayer coupling in van der Waals ferromagnet Fe₃GeTe₂ nanoflakes. Phys Rev Lett. 2020; 125(4):047202.

[31] Chiu L-C, Chang T-S, Chen J-Y, Chang NY-C. Fast SIFT design for real-time visual feature extraction. IEEE Trans Image Process. 2013; 22(8): 3158-3167.

[32] Canny J. A computational approach to edge detection. IEEE Trans Pattern Anal Mach Intell. 1986; PAMI-8(6): 679.

[33] Grossberg S. Competitive learning: from interactive activation to adaptive resonance. Cognit Sci. 1987; 11(1): 23-63.

[34] Choi J, Wang Z, Venkataramani S, Chuang PI-J, Srinivasan V, Gopalakrishnan K. ArXiv Prepr ArXiv180506085. 2018.

[35] Hubara I, Courbariaux M, Soudry D, El-Yaniv R, Bengio Y. Quantized neural networks: Training neural networks with low precision weights and activations. J Mach Learn Res. 2018; 18(187): 1-30.

[36] Diware S, Singh A, Gebregiorgis A, Joshi RV, Hamdioui S, Bishnoi R. Accurate and energy-efficient bit-slicing for RRAM-based neural networks. IEEE Trans Emerg Top Comput Intell. 2022; 7(1): 164-177.

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

[38] Shi S, Liang S, Zhu Z, et al. All-electric magnetization switching and Dzyaloshinskii–Moriya interaction in WTe₂/ferromagnet heterostructures. Nat Nanotechnol. 2019; 14(10): 945-949.

[39] Wang H, Wu H, Zhang J, et al. Room temperature energy-efficient spin-orbit torque switching in two-dimensional van der Waals Fe₃GeTe₂ induced by topological insulators. Nat Commun. 2023; 14(1):5173.

[40] Liu S, Yuan X, Zou Y, et al. Wafer-scale two-dimensional ferromagnetic Fe₃GeTe₂ thin films grown by molecular beam epitaxy. Npj 2D Mater Appl. 2017; 1(1):30.

[41] Kim S, Kwak J, Ciobanu CV, Kwon S. Recent developments in controlled vapor-phase growth of 2D group 6 transition metal dichalcogenides. Adv Mater. 2019; 31(20):1804939.

[42] Watson AJ, Lu W, Guimarães MH, Stöhr M. Transfer of large-scale two-dimensional semiconductors: challenges and developments. 2D Mater. 2021; 8(3):032001.

[43] Castellanos-Gomez A, Buscema M, Molenaar R, et al. Deterministic transfer oftwo-dimensional materials by all-dry viscoelastic stamping. Lett. 2014; 1(1):011002.

[44] Sahoo PK, Memaran S, Xin Y, Balicas L, Gutiérrez HR. One-pot growth of two-dimensional lateral heterostructures via sequential edge-epitaxy. Nature. 2018; 553(7686): 63-67.

[45] Garello K, Yasin F, Couet S, et al. SOT-MRAM 300 mm integration for low power and ultrafast embedded memories. 2018 IEEE Symposium on VLSI Circuits; 2018: 81-82.

[46] Dieny B, Prejbeanu IL, Garello K, et al. Opportunities and challenges for spintronics in the microelectronics industry. Nat Electron. 2020; 3(8): 446-459.

RIGHTS & PERMISSIONS

2026 The Author(s). InfoMat published by UESTC and John Wiley & Sons Australia, Ltd.