Title: A domain wall-magnetic tunnel junction artificial synapse with notched geometry for accurate and efficient training of deep neural networks

Abstract

Inspired by the parallelism and efficiency of the brain, several candidates for artificial synapse devices have been developed for neuromorphic computing, yet a nonlinear and asymmetric synaptic response curve precludes their use for backpropagation, the foundation of modern supervised learning. Spintronic devices—which benefit from high endurance, low power consumption, low latency, and CMOS compatibility—are a promising technology for memory, and domain-wall magnetic tunnel junction (DW-MTJ) devices have been shown to implement synaptic functions such as long-term potentiation and spike-timing dependent plasticity. In this work, we propose a notched DW-MTJ synapse as a candidate for supervised learning. Using micromagnetic simulations at room temperature, we show that notched synapses ensure the non-volatility of the synaptic weight and allow for highly linear, symmetric, and reproducible weight updates using either spin transfer torque (STT) or spin–orbit torque (SOT) mechanisms of DW propagation. We use lookup tables constructed from micromagnetics simulations to model the training of neural networks built with DW-MTJ synapses on both the MNIST and Fashion-MNIST image classification tasks. Accounting for thermal noise and realistic process variations, the DW-MTJ devices achieve classification accuracy close to ideal floating-point updates using both STT and SOT devices at room temperature and at 400 K. Our work establishesmore » the basis for a magnetic artificial synapse that can eventually lead to hardware neural networks with fully spintronic matrix operations implementing machine learning.« less

Authors:

Liu, Samuel  ^[1];

• Search DOE PAGES for author "Liu, Samuel"
• Search DOE PAGES for ORCID "0000-0002-1864-5734"

Xiao, T. Patrick  ^[2];

• Search DOE PAGES for author "Xiao, T. Patrick"
• Search DOE PAGES for ORCID "0000-0001-9066-2961"

Cui, Can  ^[1];

• Search DOE PAGES for author "Cui, Can"
• Search DOE PAGES for ORCID "0000-0002-9277-7428"

Incorvia, Jean C.  ^[1];

• Search DOE PAGES for author "Incorvia, Jean C."
• Search DOE PAGES for ORCID "0000-0002-4805-2112"

Bennett, Christopher H.  ^[3];

• Search DOE PAGES for author "Bennett, Christopher H."
• Search DOE PAGES for ORCID "0000-0002-6989-292X"

Marinella, Matthew J.  ^[3]

• Search DOE PAGES for author "Marinella, Matthew J."
• Search DOE PAGES for ORCID "0000-0002-6537-1836"

---

+ Show Author Affiliations

1. Univ. of Texas, Austin, TX (United States)
2. Sandia National Laboratories, Albuquerque, New Mexico 87123, USA
3. Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)

Publication Date:

Mon May 17 00:00:00 EDT 2021

Research Org.:

Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)

Sponsoring Org.:

USDOE National Nuclear Security Administration (NNSA)

OSTI Identifier:

1784862

Alternate Identifier(s):

OSTI ID: 1784352

Report Number(s):

SAND-2021-6135J
Journal ID: ISSN 0003-6951; 696352; TRN: US2210344

Grant/Contract Number:  

AC04-94AL85000

Resource Type:

Accepted Manuscript

Journal Name:

Applied Physics Letters

Additional Journal Information:

Journal Volume: 118; Journal Issue: 20; Journal ID: ISSN 0003-6951

Publisher:

American Institute of Physics (AIP)

Country of Publication:

United States

Language:

English

Subject:

42 ENGINEERING; Spintronics; Magnetic devices; Artificial neural networks; Spintronic devices; Magnetic memories; Machine learning; Spin-transfer-torque; Memory device; Stochastic processes; Magnetic tunnel junctions