Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Multilayer spintronic neural networks with radiofrequency connections

Nature Nanotechnology volume 18, pages 1273–1280 (2023)Cite this article

Subjects

Abstract

Spintronic nano-synapses and nano-neurons perform neural network operations with high accuracy thanks to their rich, reproducible and controllable magnetization dynamics. These dynamical nanodevices could transform artificial intelligence hardware, provided they implement state-of-the-art deep neural networks. However, there is today no scalable way to connect them in multilayers. Here we show that the flagship nano-components of spintronics, magnetic tunnel junctions, can be connected into multilayer neural networks where they implement both synapses and neurons thanks to their magnetization dynamics, and communicate by processing, transmitting and receiving radiofrequency signals. We build a hardware spintronic neural network composed of nine magnetic tunnel junctions connected in two layers, and show that it natively classifies nonlinearly separable radiofrequency inputs with an accuracy of 97.7%. Using physical simulations, we demonstrate that a large network of nanoscale junctions can achieve state-of-the-art identification of drones from their radiofrequency transmissions, without digitization and consuming only a few milliwatts, which constitutes a gain of several orders of magnitude in power consumption compared to currently used techniques. This study lays the foundation for deep, dynamical, spintronic neural networks.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Building blocks of multilayer RF/d.c. spintronic neural networks.
Fig. 2: Experimental demonstration of neurons-to-synapses interconnection.
Fig. 3: Experimental demonstration of the spintronic neural network.
Fig. 4: Drone classification by direct processing of their RF emissions.

Similar content being viewed by others

Data availability

The data supporting the figures of this study are publicly available in the Zenodo repository: https://doi.org/10.5281/zenodo.7956045.

Code availability

The source codes used in this study, in particular to generate the data of Fig. 4, are publicly available in the Github repository: https://github.com/neurophysics-cnrsthales/rf_spintronic_nn.

References

  1. LeCun, Y., Bengio, Y. & Hinton, G. Deep learning. Nature 521, 436–444 (2015).

    Article  CAS  Google Scholar 

  2. Christensen, D. V. et al. 2022 roadmap on neuromorphic computing and engineering. Neuromorphic Comput. Eng. 2, 022501 (2022).

    Google Scholar 

  3. Grollier, J. et al. Neuromorphic spintronics. Nat. Electron. 3, 360–370 (2020).

  4. Worledge, D. C. Spin-Transfer-Torque MRAM: the next revolution in memory. In 2022 IEEE International Memory Workshop (IMW) 1–4 (IEEE, 2022).

  5. Safranski, C., Sun, J. Z. & Kent, A. D. A perspective on electrical generation of spin current for magnetic random access memories. Appl. Phys. Lett. 120, 160502 (2022).

    CAS  Google Scholar 

  6. Lequeux, S. et al. A magnetic synapse: multilevel spin-torque memristor with perpendicular anisotropy. Sci. Rep. 6, 31510 (2016).

    CAS  Google Scholar 

  7. Kurenkov, A. et al. Artificial neuron and synapse realized in an antiferromagnet/ferromagnet heterostructure using dynamics of spin–orbit torque switching. Adv. Mater. 31, 1900636 (2019).

    Google Scholar 

  8. Chen, R. et al. Nanoscale room-temperature multilayer skyrmionic synapse for deep spiking neural networks. Phys. Rev. Appl. 14, 014096 (2020).

    CAS  Google Scholar 

  9. Song, K. M. et al. Skyrmion-based artificial synapses for neuromorphic computing. Nat. Electron. 3, 148–155 (2020).

    Google Scholar 

  10. Mansueto, M. et al. Spintronic memristors for neuromorphic circuits based on the angular variation of tunnel magnetoresistance. Nanoscale 13, 11488–11496 (2021).

    CAS  Google Scholar 

  11. Zhang, X. et al. Spin-torque memristors based on perpendicular magnetic tunnel junctions for neuromorphic computing. Adv. Sci. 8, 2004645 (2021).

    CAS  Google Scholar 

  12. Xu, J.-W. et al. A quantum material spintronic resonator. Sci. Rep. 11, 15082 (2021).

    CAS  Google Scholar 

  13. Kiraly, B., Knol, E. J., van Weerdenburg, W. M. J., Kappen, H. J. & Khajetoorians, A. A. An atomic Boltzmann machine capable of self-adaption. Nat. Nanotechnol. 16, 414–420 (2021).

    CAS  Google Scholar 

  14. Leroux, N. et al. Hardware realization of the multiply and accumulate operation on radio-frequency signals with magnetic tunnel junctions. Neuromorphic Comput. Eng. https://doi.org/10.1088/2634-4386/abfca6 (2021).

  15. Monalisha, P., Kumar, A. P. S., Wang, X. R. & Piramanayagam, S. N. Emulation of synaptic plasticity on a cobalt-based synaptic transistor for neuromorphic computing. ACS Appl. Mater. Interfaces 14, 11864–11872 (2022).

    CAS  Google Scholar 

  16. Torrejon, J. et al. Neuromorphic computing with nanoscale spintronic oscillators. Nature 547, 428–431 (2017).

    CAS  Google Scholar 

  17. Prychynenko, D. et al. Magnetic skyrmion as a nonlinear resistive element: a potential building block for reservoir computing. Phys. Rev. Appl. 9, 014034 (2018).

    CAS  Google Scholar 

  18. Borders, W. A. et al. Integer factorization using stochastic magnetic tunnel junctions. Nature 573, 390–393 (2019).

    CAS  Google Scholar 

  19. Tsunegi, S. et al. Physical reservoir computing based on spin torque oscillator with forced synchronization. Appl. Phys. Lett. 114, 164101 (2019).

    Google Scholar 

  20. Koo, M. et al. Distance computation based on coupled spin-torque oscillators: application to image processing. Phys. Rev. Appl. 14, 034001 (2020).

    CAS  Google Scholar 

  21. Zahedinejad, M. et al. Two-dimensional mutually synchronized spin Hall nano-oscillator arrays for neuromorphic computing. Nat. Nanotechnol. 15, 47–52 (2020).

    CAS  Google Scholar 

  22. Ababei, R. V. et al. Neuromorphic computation with a single magnetic domain wall. Sci. Rep. 11, 15587 (2021).

    CAS  Google Scholar 

  23. Watt, S., Kostylev, M., Ustinov, A. B. & Kalinikos, B. A. Implementing a magnonic reservoir computer model based on time-delay multiplexing. Phys. Rev. Appl. 15, 064060 (2021).

    CAS  Google Scholar 

  24. Mazza, L. et al. Computing with injection-locked spintronic diodes. Phys. Rev. Appl. 17, 014045 (2022).

    CAS  Google Scholar 

  25. Zahedinejad, M. et al. Memristive control of mutual spin Hall nano-oscillator synchronization for neuromorphic computing. Nat. Mater. 21, 81–87 (2022).

    CAS  Google Scholar 

  26. Romera, M. et al. Vowel recognition with four coupled spin-torque nano-oscillators. Nature 563, 230–234 (2018).

    CAS  Google Scholar 

  27. Romera, M. et al. Binding events through the mutual synchronization of spintronic nano-neurons. Nat. Commun. 13, 883 (2022).

    Google Scholar 

  28. Finocchio, G. et al. Perspectives on spintronic diodes. Appl. Phys. Lett. 118, 160502 (2021).

    CAS  Google Scholar 

  29. Siafarikas, D. & Volakis, J. L. Toward direct RF sampling: implications for digital communications. IEEE Microw. Mag. 21, 43–52 (2020).

    Google Scholar 

  30. Farley, B., McGrath, J. & Erdmann, C. An all-programmable 16-nm RFSoC for digital-RF communications. IEEE Micro 38, 61–71 (2018).

    Google Scholar 

  31. Leroux, N. et al. Radio-frequency multiply-and-accumulate operations with spintronic synapses. Phys. Rev. Appl. 15, 034067 (2021).

    CAS  Google Scholar 

  32. Tulapurkar, A. A. et al. Spin-torque diode effect in magnetic tunnel junctions. Nature 438, 339–342 (2005).

    CAS  Google Scholar 

  33. Slavin, A. & Tiberkevich, V. Nonlinear auto-oscillator theory of microwave generation by spin-polarized current. IEEE Trans. Magn. 45, 1875–1918 (2009).

    CAS  Google Scholar 

  34. Wright, L. G. et al. Deep physical neural networks trained with backpropagation. Nature 601, 549–555 (2022).

    CAS  Google Scholar 

  35. Yoon, Y. H., Khan, S., Huh, J. & Ye, J. C. Efficient B-mode ultrasound image reconstruction from sub-sampled RF data using deep learning. IEEE Trans. Med. Imaging 38, 325–336 (2019).

    Google Scholar 

  36. Dai, M., Li, S., Wang, Y., Zhang, Q. & Yu, J. Post-processing radio-frequency signal based on deep learning method for ultrasonic microbubble imaging. Biomed. Eng. Online 18, 95 (2019).

    Google Scholar 

  37. Besler, E., Wang, Y. C. & Sahakian, A. V. Real-time radiofrequency ablation lesion depth estimation using multi-frequency impedance with a deep neural network and tree-based ensembles. IEEE Trans. Biomed. Eng. 67, 1890–1899 (2020).

    Google Scholar 

  38. Merchant, K., Revay, S., Stantchev, G. & Nousain, B. Deep learning for RF device fingerprinting in cognitive communication networks. IEEE J. Sel. Top. Signal Process. 12, 160–167 (2018).

    Google Scholar 

  39. Lien, J. et al. Soli: ubiquitous gesture sensing with millimeter wave radar. ACM Trans. Graph. 35, 142 (2016).

  40. Basak, S., Rajendran, S., Pollin, S. & Scheers, B. Drone classification from RF fingerprints using deep residual nets. In 2021 International Conference on COMmunication Systems & NETworkS (COMSNETS) 548–555 (IEEE, 2021).

  41. USRP X310 High-Performance Software Defined Radio https://www.ettus.com/all-products/x310-kit/ (Ettus Research).

  42. Compare 10 Series Graphics Cards, GeForce https://www.nvidia.com/en-in/geforce/products/10series/compare/ (NVIDIA).

  43. Chao, X., Jamali, M. & Wang, J.-P. Scaling effect of spin-torque nano-oscillators. AIP Adv. 7, 056624 (2017).

    Google Scholar 

  44. Marković, D., Mizrahi, A., Querlioz, D. & Grollier, J. Physics for neuromorphic computing. Nat. Rev. Phys. 2, 499–510 (2020).

  45. Dussaux, A. et al. Large amplitude spin torque vortex oscillations at zero external field using a perpendicular spin polarizer. Appl. Phys. Lett. 105, 022404 (2014).

    Google Scholar 

  46. Tsunegi, S. et al. High emission power and Q factor in spin torque vortex oscillator consisting of FeB free layer. Appl. Phys. Express 7, 063009 (2014).

    CAS  Google Scholar 

  47. Choi, J.-G. et al. Voltage-driven gigahertz frequency tuning of spin Hall nano-oscillators. Nat. Commun. 13, 3783 (2022).

    CAS  Google Scholar 

  48. Martins, L. et al. Non-volatile artificial synapse based on a vortex nano-oscillator. Sci. Rep. 11, 16094 (2021).

    CAS  Google Scholar 

  49. Jiang, S. et al. Reduced spin torque nano-oscillator linewidth using He+ irradiation. Appl. Phys. Lett. 116, 072403 (2020).

    CAS  Google Scholar 

  50. Divinskiy, B., Urazhdin, S., Demokritov, S. O. & Demidov, V. E. Controlled nonlinear magnetic damping in spin-Hall nano-devices. Nat. Commun. 10, 5211 (2019).

    Google Scholar 

  51. Sethi, P. et al. Compensation of anisotropy in spin-Hall devices for neuromorphic applications. Phys. Rev. Appl. 19.6, 064018 (2023).

    Google Scholar 

  52. Jenkins, A. S., Alvarez, L. S. E., Freitas, P. P. & Ferreira, R. Digital and analogue modulation and demodulation scheme using vortex-based spin torque nano-oscillators. Sci. Rep. 10, 11181 (2020).

    CAS  Google Scholar 

  53. Jung, S. et al. A crossbar array of magnetoresistive memory devices for in-memory computing. Nature 601, 211–216 (2022).

    CAS  Google Scholar 

  54. Craven, M. P., Curtis, K. M. & Hayes-Gill, B. R. Frequency division multiplexing in analogue neural network. Electron. Lett. 27, 918–920 (1991).

    Google Scholar 

  55. Leroux, N. et al. Convolutional neural networks with radio-frequency spintronic nano-devices. Neuromorphic Comput. Eng. 2, 034002 (2022).

    Google Scholar 

  56. Noh, S. et al. Spin dynamics in ferromagnetic resonance for nano-sized magnetic dot arrays: metrology and insight into magnetization dynamics. IEEE Trans. Magn. 47, 2387–2390 (2011).

    Google Scholar 

  57. Litvinenko, A. et al. Ultrafast GHz-range swept-tuned spectrum analyzer with 20 ns temporal resolution based on a spin-torque nano-oscillator with a uniformly magnetized ‘free’ layer. Nano Lett. 22, 1874–1879 (2022).

    CAS  Google Scholar 

  58. Kurokawa, Y. et al. Ultra-wide-band millimeter-wave generator using spin torque oscillator with strong interlayer exchange couplings. Sci. Rep. 12, 10849 (2022).

    CAS  Google Scholar 

  59. Bonetti, S., Muduli, P., Mancoff, F. & Åkerman, J. Spin torque oscillator frequency versus magnetic field angle: the prospect of operation beyond 65 GHz. Appl. Phys. Lett. 94, 102507 (2009).

    Google Scholar 

  60. Khymyn, R. et al. Ultra-fast artificial neuron: generation of picosecond-duration spikes in a current-driven antiferromagnetic auto-oscillator. Sci. Rep. 8, 15727 (2018).

    Google Scholar 

  61. Chakravarty, A. et al. Supervised learning of an opto-magnetic neural network with ultrashort laser pulses. Appl. Phys. Lett. 114, 192407 (2019).

    Google Scholar 

  62. Scellier, B. & Bengio, Y. Equilibrium propagation: bridging the gap between energy-based models and backpropagation. Front. Comput. Neurosci. 11, 24 (2017).

    Google Scholar 

  63. Akiba, T., Sano, S., Yanase, T., Ohta, T. & Koyama, M. Optuna: a next-generation hyperparameter optimization framework. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining 2623–2631 (Association for Computing Machinery, 2019).

Download references

Acknowledgements

This work was supported by the European Union’s Horizon 2020 research and innovation programme under grant RadioSpin no. 101017098. The text of the article was partially edited by a large language model (OpenAI ChatGPT).

Author information

Author notes

  1. These authors contributed equally: Andrew Ross, Nathan Leroux.

Authors and Affiliations

  1. Unité Mixte de Physique CNRS/Thales, CNRS, Thales, Université Paris-Saclay, Palaiseau, France

    Andrew Ross, Nathan Leroux, Arnaud De Riz, Danijela Marković, Dédalo Sanz-Hernández, Juan Trastoy, Paolo Bortolotti, Frank Alice Mizrahi & Julie Grollier

  2. Centre de Nanosciences et de Nanotechnologies, Université Paris-Saclay, CNRS, Palaiseau, France

    Damien Querlioz

  3. International Iberian Nanotechnology Laboratory (INL), Braga, Portugal

    Leandro Martins, Luana Benetti, Marcel S. Claro, Pedro Anacleto, Alejandro Schulman, Alex S. Jenkins & Ricardo Ferreira

  4. Laboratoire de l’Intégration du Matériau au Système (IMS; UMR 5218), Univ. Bordeaux, CNRS, Bordeaux INP, Talence, France

    Thierry Taris, Jean-Baptiste Begueret, Sylvain Saïghi & Adrien F. Vincent

Authors
  1. Andrew Ross
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Nathan Leroux
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Arnaud De Riz
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Danijela Marković
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Dédalo Sanz-Hernández
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Juan Trastoy
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Paolo Bortolotti
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Damien Querlioz
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Leandro Martins
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Luana Benetti
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Marcel S. Claro
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Pedro Anacleto
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Alejandro Schulman
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Thierry Taris
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Jean-Baptiste Begueret
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Sylvain Saïghi
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Alex S. Jenkins
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Ricardo Ferreira
    View author publications

    You can also search for this author in PubMed Google Scholar

  19. Adrien F. Vincent
    View author publications

    You can also search for this author in PubMed Google Scholar

  20. Frank Alice Mizrahi
    View author publications

    You can also search for this author in PubMed Google Scholar

  21. Julie Grollier
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

The study was designed and supervised by J.G. and F.A.M.; samples were optimized and fabricated by R.F., A.S.J., L.M., L.B., M.S.C., P.A., A.S. and A.S.J.; experiments and training of the hardware neural network were performed by F.A.M., A.R., N.L., D.M., J.T. and D.S.-H.; simulations of neural networks and drone classification were performed by F.A.M., A.d.R. and N.L.; the estimation of scaled network performances was performed by F.A.M., N.L. and A.F.V.; and F.A.M., P.B., D.Q., T.T., J.-B.B. and S.S. provided insights on applications and circuit design. All authors contributed to the analysis of the results and writing of the paper.

Corresponding authors

Correspondence to Frank Alice Mizrahi or Julie Grollier.

Ethics declarations

Competing interests

The authors declare the following competing interests: patents FR 1800805, 1800806 and 1800807 are held by CNRS and Thales, and J.G. is the inventor. They cover the neural network architecture proposed in this work. The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Nanotechnology thanks Jianhua Yang, Thomas Hayward and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary Figs. 1–10 and Discussion.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ross, A., Leroux, N., De Riz, A. et al. Multilayer spintronic neural networks with radiofrequency connections. Nat. Nanotechnol. 18, 1273–1280 (2023). https://doi.org/10.1038/s41565-023-01452-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41565-023-01452-w

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing