EUVIMEDEUVIMED entdecken
Crossref

Physics-Aware Co-training to Overcome Device Non-idealities in Canted SOT-MTJ Arrays for In-Memory Computing

Zhihao Zhao, Yifan Zhang, Longchao Liu, Fengjun Dong, Zhipeng Guo, Junbo Yang, Ruihuan Wu, Yiqing Zhou, Kaixue Ma, Xingyu Gao, Guozhong Xing

ACS Applied Electronic Materials · 10.11.2026

DOI: 10.1021/acsaelm.6c01422 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACS Applied Electronic Materials, online erschienen am 10.11.2026. Verfasst von Zhihao Zhao und 10 weiteren Autor:innen. Die Arbeit nennt 58 Literaturangaben. Gefördert von: Chinese Academy of Sciences, National Basic Research Program of China (973 Program), National Natural Science Foundation of China (NSFC).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The physical non-idealities inherent in emerging spintronic devices, such as device-to-device variations, stochastic switching, and nonlinear conductance responses, pose fundamental challenges for their deployment in in-memory computing architectures. Here, we investigate the integration of canted spin–orbit torque magnetic tunnel junctions (SOT-MTJs) for deep learning applications and propose a hardware-aware deep learning framework that embeds into the training process experimentally extracted device characteristics, which are faithfully resolved via a dedicated three-terminal SOT testing scheme that effectively suppresses spin-transfer torque, halves the switching threshold voltage, and eliminates parasitic drifts to minimize measurement-induced fluctuations and reveal intrinsic 1010-cycle endurance, thereby directly facilitating robust hardware–software co-optimization. By tuning a normalized conductance factor gnorm, we achieve a median classification accuracy of 88.9% on the Wine dataset using a 13 × 13 SOT-MTJ array, surpassing baseline software-to-hardware weight mapping. Incorporating hardware-aware co-training further elevates the median accuracy to 94.4%, demonstrating substantial robustness against device imperfections. Overall, our results highlight that explicitly modeling device variability at both the physics and algorithm levels can enable spintronic hardware based on SOT-MTJ arrays to function as compact, energy-efficient neuromorphic platforms, allowing device stochasticity to be incorporated as a learnable constraint, ultimately recovering hardware inference accuracy to within a few percentage points of the ideal software baseline.

Förderung

  • Chinese Academy of Sciences
  • National Basic Research Program of China (973 Program)
  • National Natural Science Foundation of China (NSFC)

Zitieren

APA

Zhao, Z., Zhang, Y., Liu, L., Dong, F., Guo, Z., Yang, J., Wu, R., Zhou, Y., Ma, K., Gao, X., & Xing, G. (2026). Physics-Aware Co-training to Overcome Device Non-idealities in Canted SOT-MTJ Arrays for In-Memory Computing. ACS Applied Electronic Materials. https://doi.org/10.1021/acsaelm.6c01422

Vancouver

Zhao Z, Zhang Y, Liu L, Dong F, Guo Z, Yang J, et al. Physics-Aware Co-training to Overcome Device Non-idealities in Canted SOT-MTJ Arrays for In-Memory Computing. ACS Applied Electronic Materials. 2026. doi:10.1021/acsaelm.6c01422

BibTeX
@article{Zhao2026,
  author = {Zhao, Zhihao and Zhang, Yifan and Liu, Longchao and Dong, Fengjun and Guo, Zhipeng and Yang, Junbo and Wu, Ruihuan and Zhou, Yiqing and Ma, Kaixue and Gao, Xingyu and Xing, Guozhong},
  title = {Physics-Aware Co-training to Overcome Device Non-idealities in Canted SOT-MTJ Arrays for In-Memory Computing},
  journal = {ACS Applied Electronic Materials},
  year = {2026},
  doi = {10.1021/acsaelm.6c01422}
}
Kostenlos starten →