Magnetic proximity effect in magnetic-insulator/heavy-metal heterostructures across the compensation temperature

The magnetic proximity effect in Pt and W thin films grown on ${\mathrm{Dy}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ (DyIG) is examined at temperatures above and below the magnetic compensation temperature of the ferrimagnetic insulator. Polarized neutron reflectometry indicates that the proximity effect is positive in Pt/DyIG both above and below the compensation temperature, and x-ray magnetic circular dichroism shows a weak positive W magnetization below compensation in W/DyIG. This demonstrates a qualitative difference compared to heavy metal/ferrimagnetic rare earth–transition metal alloys, where the proximity-induced magnetism of the heavy metal changes sign at the compensation temperature. Reflectometry, structural, and spin transport measurements show that depositing the heavy metal film on the ferrimagnetic insulator in situ without breaking vacuum avoids the formation of a low density interfacial layer between the Pt and DyIG.

Figure 1

(a)–(c) Characterization of the in situ grown Pt(9 nm)/DyIG(54 nm)/GGG sample. (a) High-resolution XRD of the DyIG/GGG (444) reflection. (b) In-plane room temperature VSM hysteresis loop. (c) Saturation magnetization vs temperature for DyIG/GGG; red dashed lines indicate extrapolation from experimental data. Blue vertical dashed line indicates the bulk value of $T_{\mathrm{M}}$. (d) In-plane room temperature AHE-like SMR on the in situ grown W(5 nm)/DyIG(42 nm)/GGG sample. ADF-STEM images of the (e) Pt/DyIG and (f) DyIG/GGG interfaces of the in situ grown Pt(9 nm)/DyIG(54 nm)/GGG sample.

Figure 2

XAS and XMCD for (a), (b) Fe $L_{2,3}$, (c), (d) Dy $M_{4,5}$, (e), (f) W $L_3$, (g), (h) Pt $L_2$ edges. Panels (a)–(f) were collected on the in situ grown Pt/DyIG/GGG sample and (g) and (h) on the in situ W/DyIG/GGG sample.

Figure 3

(a) Fitted polarized neutron reflectometry curves and (b) spin asymmetry of in situ Pt(9 nm)/DyIG(54 nm)/GGG at 10, 200, 240, and 300 K. Vertically offset for clarity. Error bars are representative of 1σ.

Figure 4

Structural scattering length density profile of (a) ex situ and (b) in situ grown Pt(9 nm)/DyIG(54 nm)/GGG. rho and irho represent the real (density) and imaginary (absorption) components of the structural SLD. (c) Magnetic SLD of in situ sample, representing the magnetization.