シリコン基板上の高性能酸化物スピントロニクスデバイスを開発(Scientists Develop High-performance Oxide-based Spintronic Devices on Silicon Substrates)

ad

2025-04-15 中国科学院(CAS)

シリコン基板上の高性能酸化物スピントロニクスデバイスを開発(Scientists Develop High-performance Oxide-based Spintronic Devices on Silicon Substrates)Heterogeneous integration of single-crystal SrRuO₃ films on silicon for spin-orbit torque devices with low-power consumption (Image by NIMTE)

中国科学院寧波材料技術与工程研究所(NIMTE)の研究チームは、酸化物スピントロニクス材料をシリコン基板に高性能で統合する手法を開発した。この新技術は、転写技術とエピタキシャル成長を組み合わせることで、単結晶SrRuO₃(SRO)膜をシリコン上に形成し、スピン軌道トルク(SOT)デバイスを作製。SROは高いスピンホール伝導率(6.1×10⁴ ħ/2e S·m⁻¹)を示し、低電流密度で磁化反転を実現した。さらに多状態磁化スイッチングにより、生体神経機能の模倣が可能となり、画像認識タスクで88%の精度を達成。この研究は低電力電子機器やニューロモルフィックコンピューティングへの応用が期待される。

<関連情報>

スピントロニクスデバイスのための大きなスピンホール伝導性を持つSrRuO3単結晶薄膜のシリコン上への不均一集積化 Heterogeneous Integration of Single-Crystal SrRuO3 Films with Large Spin Hall Conductivity on Silicon for Spintronic Devices

Zengxing Lu, Xue Bai, Bin Lao, Xuan Zheng, Haoyue Deng, Zhen Fan, Run-Wei Li, Zhiming Wang
Advanced Functional Materials  Published: 13 March 2025
DOI:https://doi.org/10.1002/adfm.202500755

Abstract

Spin-orbit torque (SOT) device has been recognized as a promising candidate for next-generation information devices, owing to its energy-efficient, high-speed and scalable potential. Complex oxides show large spin Hall conductivity (σSH), thus are capable of inducing efficient SOT. However, corresponding SOT device integrated on silicon has rarely been reported due to the technical challenges in film preparation. Here, a hybrid transfer and epitaxy strategy for integrating single-crystalline SrRuO3 on silicon buffered with transferred freestanding SrTiO3 membrane is demonstrated. The integrated SrRuO3 exhibits a large σSH of 6.1 × 104 ħ/2e S m−1, enabling magnetization switching in a CoPt layer with an ultra-low critical current density of 1.3 × 1010 A m−2, being lower than heavy metals by 1–2 orders of magnitude. What’s more, this SOT-induced switching displays multiple intermediate magnetization states, which is able to mimic synapse and neuron behavior. Simulating a two-layer artificial neural network using experimentally extracted device parameters achieve an accuracy of 88% for image recognition. The results showcase the successful integration of single-crystalline oxides on silicon, paving the way for high-performance, silicon-compatible spintronic devices.

0403電子応用
ad
ad
Follow
ad
タイトルとURLをコピーしました