2023年第70回応用物理学会春季学術講演会

講演情報

一般セッション(口頭講演)

10 スピントロニクス・マグネティクス » 10.2 スピン基盤技術・萌芽的デバイス技術

[16p-D419-1~18] 10.2 スピン基盤技術・萌芽的デバイス技術

2023年3月16日(木) 13:30 〜 18:30 D419 (11号館)

山田 貴大(東工大)、周 偉男(物材機構)、山崎 匠(東北大)

18:15 〜 18:30

[16p-D419-18] Room temperature spin cluster glass mediated spin wave for reservoir computing

〇(P)Kaijie Ma1、Kenyu Terao1、Zhiqiang Liao1、Hitoshi Tabata1 (1.The Univ. of Tokyo)

キーワード:spin wave, spin glass

Physical reservoir computing is adaptable for fast dynamic information processing with low learning cost. Spin glass with nonlinear interaction and short-term memory is suitable to implement as the physical reservoir. To display the magnetization signal of spin glasses, spin waves can be adopted as the information carrier with low energy consumption. However, the conventional spin glass shows frustration in low temperatures and is difficult to generate the spin wave.
Here, a 100-nm-thick Y3Fe4.9Ga0.1O12 (Ga:YIG) thin film was epitaxially grown on YAG (100) substrate by pulsed laser deposition (PLD) at 750 °C and 0.1 Pa, whose high crystallinity was measured by XRD. The temperature dependent magnetization curves in zero-field-cooling (ZFC) and field-cooling (FC) conditions by MPMS demonstrate the typical spin cluster glass with the spin-freezing temperature Tf of 360 K when the magnetic field is 20 Oe. The negative temperature cycling curve for ZFC relaxations indicates the memory effect in Ga:YIG with metastable states. To further confirm the frustrated spin glass effect, in ZFC condition, the magnetization curve with and without applied magnetic field shows the relaxation properties in 300 K. The magnetostatic surface spin wave excited by a microwave technique in Ga:YIG is adopted as the signal carrier to reveal the frustrated magnetization with low energy consumption. The spin wave transmission spectra demonstrate the frequency shift depending on the external magnetic field, which confirms the feasibility of spin wave propagation in spin cluster glass. The spin cluster glass is expected to be applied in physical reservoir computing.