Crossref journal-article
AIP Publishing
Applied Physics Letters (317)
Abstract

A device based on current-induced spin-orbit torque (SOT) that functions as an electronic neuron is proposed in this work. The SOT device implements an artificial neuron's thresholding (transfer) function. In the first step of a two-step switching scheme, a charge current places the magnetization of a nano-magnet along the hard-axis, i.e., an unstable point for the magnet. In the second step, the SOT device (neuron) receives a current (from the synapses) which moves the magnetization from the unstable point to one of the two stable states. The polarity of the synaptic current encodes the excitatory and inhibitory nature of the neuron input and determines the final orientation of the magnetization. A resistive crossbar array, functioning as synapses, generates a bipolar current that is a weighted sum of the inputs. The simulation of a two layer feed-forward artificial neural network based on the SOT electronic neuron shows that it consumes ∼3× lower power than a 45 nm digital CMOS implementation, while reaching ∼80% accuracy in the classification of 100 images of handwritten digits from the MNIST dataset.

Bibliography

Sengupta, A., Choday, S. H., Kim, Y., & Roy, K. (2015). Spin orbit torque based electronic neuron. Applied Physics Letters, 106(14).

Authors 4
  1. Abhronil Sengupta (first)
  2. Sri Harsha Choday (additional)
  3. Yusung Kim (additional)
  4. Kaushik Roy (additional)
References 37 Referenced 79
  1. 10.1109/5.58323 / Proc. IEEE / 30 years of adaptive neural networks: Perceptron, madaline, and backpropagation (1990)
  2. {'key': '2023061800120635000_c2', 'article-title': 'Hierarchical Temporal Memory including HTM cortical learning algorithms'} / Hierarchical Temporal Memory including HTM cortical learning algorithms
  3. 10.1016/j.neunet.2013.05.011 / Neural Networks / Design of silicon brains in the nano-CMOS era: Spiking neurons, learning synapses and neural architecture optimization (2013)
  4. {'first-page': '2850', 'article-title': 'SpiNNaker: Mapping neural networks onto a massively-parallel chip multiprocessor', 'year': '2008', 'key': '2023061800120635000_c4'} / SpiNNaker: Mapping neural networks onto a massively-parallel chip multiprocessor (2008)
  5. {'first-page': '1', 'article-title': 'A digital neurosynaptic core using embedded crossbar memory with 45 pJ per spike in 45\u2009nm', 'year': '2011', 'key': '2023061800120635000_c5'} / A digital neurosynaptic core using embedded crossbar memory with 45 pJ per spike in 45 nm (2011)
  6. {'key': '2023061800120635000_c6', 'article-title': 'Proposal for neuromorphic hardware using spin devices'} / Proposal for neuromorphic hardware using spin devices
  7. 10.1109/TNANO.2012.2202125 / IEEE Trans. Nanotechnol. / Spin-based neuron model with domain wall magnets as synapse (2012)
  8. 10.1088/0957-4484/24/38/384010 / Nanotechnology / Integration of nanoscale memristor synapses in neuromorphic computing architectures (2013)
  9. 10.1109/MCAS.2013.2256271 / IEEE Circuits Syst. Mag. / A proposal for hybrid memristor-CMOS spiking neuromorphic learning systems (2013)
  10. 10.1021/nl904092h / Nano Lett. / Nanoscale memristor device as synapse in neuromorphic systems (2010)
  11. 10.1021/nn501824r / ACS Nano / Ultra-low-energy three-dimensional oxide-based electronic synapses for implementation of robust high-accuracy neuromorphic computation systems (2014)
  12. 10.1103/PhysRevB.39.6995 / Phys. Rev. B / Conductance and exchange coupling of two ferromagnets separated by a tunneling barrier (1989)
  13. 10.1103/PhysRevB.54.9353 / Phys. Rev. B / Emission of spin waves by a magnetic multilayer traversed by a current (1996)
  14. 10.1103/PhysRevLett.83.1834 / Phys. Rev. Lett. / Spin Hall effect (1999)
  15. 10.1038/nmat2613 / Nat. Mater. / Current-driven spin torque induced by the Rashba effect in a ferromagnetic metal layer (2010)
  16. 10.1038/nature10309 / Nature / Perpendicular switching of a single ferromagnetic layer induced by in-plane current injection (2011)
  17. 10.1063/1.3579155 / Appl. Phys. Lett. / Current-induced effective field in perpendicularly magnetized Ta/CoFeB/MgO wire (2011)
  18. 10.1126/science.1218197 / Science / Spin–torque switching with the giant spin Hall effect of tantalum (2012)
  19. 10.1103/PhysRevLett.109.096602 / Phys. Rev. Lett. / Current-induced switching of perpendicularly magnetized magnetic layers using spin torque from the spin Hall effect (2012)
  20. 10.1038/nmat3553 / Nat. Mater. / Domain wall depinning governed by the spin Hall effect (2013)
  21. 10.1038/nmat3522 / Nat. Mater. / Layer thickness dependence of the current induced effective field vector in Ta/CoFeB/MgO (2013)
  22. 10.1126/science.1120506 / Science / Majority logic gate for magnetic quantum-dot cellular automata (2006)
  23. 10.1103/PhysRevLett.107.010604 / Phys. Rev. Lett. / Exploring the thermodynamic limits of computation in integrated systems: Magnetic memory, nanomagnetic logic, and the Landauer limit (2011)
  24. 10.1038/nnano.2013.241 / Nat. Nanotechnol. / Spin Hall effect clocking of nanomagnetic logic without a magnetic field (2014)
  25. 10.1038/nnano.2010.31 / Nat. Nanotechnol. / Proposal for an all-spin logic device with built-in memory (2010)
  26. 10.1063/1.3567772 / Appl. Phys. Lett. / Switching energy delay of all spin logic devices (2011)
  27. 10.1088/0022-3727/38/18/001 / J. Phys. D: Appl. Phys. / Demagnetization factors for elliptic cylinders (2005)
  28. 10.1063/1.4858465 / Appl. Phys. Lett. / Spin-Hall-assisted magnetic random access memory (2014)
  29. {'first-page': '51', 'article-title': 'KNACK: A hybrid spin-charge mixed-mode simulator for evaluating different genres of spin-transfer torque MRAM bit-cells', 'year': '2011', 'key': '2023061800120635000_c29'} / KNACK: A hybrid spin-charge mixed-mode simulator for evaluating different genres of spin-transfer torque MRAM bit-cells (2011)
  30. 10.1109/5.726791 / Proc. IEEE / Gradient-based learning applied to document recognition (1998)
  31. 10.1063/1.2977964 / Appl. Phys. Lett. / An amplifier concept for spintronics (2008)
  32. {'first-page': '15', 'article-title': 'SPINDLE: SPINtronic deep learning engine for large-scale neuromorphic computing', 'year': '2014', 'key': '2023061800120635000_c32'} / SPINDLE: SPINtronic deep learning engine for large-scale neuromorphic computing (2014)
  33. 10.1016/S0129-0657(00)00004-1 / Int. J. Neural Syst. / Neural networks in analog hardware—Design and implementation issues (2000)
  34. 10.1021/nl073225h / Nano Lett. / CMOS compatible nanoscale nonvolatile resistance switching memory (2008)
  35. {'first-page': '88', 'article-title': 'Analog-input analog-weight dot-product operation with Ag/a-Si/Pt memristive devices', 'year': '2012', 'key': '2023061800120635000_c35'} / Analog-input analog-weight dot-product operation with Ag/a-Si/Pt memristive devices (2012)
  36. {'first-page': '230', 'article-title': 'Low-current perpendicular domain wall motion cell for scalable high-speed MRAM', 'year': '2009', 'key': '2023061800120635000_c36'} / Low-current perpendicular domain wall motion cell for scalable high-speed MRAM (2009)
  37. 10.1109/IEDM.2010.5703329 / Tech. Dig. IEEE Int. Electron Devices Meet. / Magnetic tunnel junction for nonvolatile CMOS logic
Dates
Type When
Created 10 years, 4 months ago (April 7, 2015, 1 p.m.)
Deposited 2 years, 2 months ago (June 17, 2023, 8:12 p.m.)
Indexed 3 weeks, 4 days ago (July 30, 2025, 7:02 a.m.)
Issued 10 years, 4 months ago (April 6, 2015)
Published 10 years, 4 months ago (April 6, 2015)
Published Online 10 years, 4 months ago (April 7, 2015)
Published Print 10 years, 4 months ago (April 6, 2015)
Funders 4
  1. Center for Spintronic Materials, Interfaces, and Novel Architecture (C-SPIN) , a MARCO and DARPA sponsored StarNet center
  2. National Security Science and Engineering Faculty Fellowship
  3. National Science Foundation 10.13039/100000001

    Region: Americas

    gov (National government)

    Labels4
    1. U.S. National Science Foundation
    2. NSF
    3. US NSF
    4. USA NSF
  4. Semiconductor Research Corporation 10.13039/100000028

    Region: Americas

    pri (For-profit companies (industry))

    Labels2
    1. Semiconductor Research Corp.
    2. SRC

@article{Sengupta_2015, title={Spin orbit torque based electronic neuron}, volume={106}, ISSN={1077-3118}, url={http://dx.doi.org/10.1063/1.4917011}, DOI={10.1063/1.4917011}, number={14}, journal={Applied Physics Letters}, publisher={AIP Publishing}, author={Sengupta, Abhronil and Choday, Sri Harsha and Kim, Yusung and Roy, Kaushik}, year={2015}, month=apr }