Origin of Strong Two-Magnon Scattering in Heavy-Metal/Ferromagnet/Oxide Heterostructures

Lijun Zhu, Lujun Zhu, D.C. Ralph, and R.A. Buhrman

Phys. Rev. Applied 13, 034038 – Published 16 March 2020

Abstract

We experimentally investigate the origin of two-magnon scattering (TMS) in heavy-metal (HM)/ferromagnet (FM)/oxide heterostructures (FM = $Co$ , ${Ni}_{81} {Fe}_{19}$ , or ${Fe}_{60} {Co}_{20} B_{20}$ ) by varying the materials located above and below the FM layer. We show that strong TMS in HM/FM/oxide systems arises primarily at the HM/FM interface and increases with the strength of the interfacial spin-orbit coupling and magnetic roughness at this interface. TMS at the FM/oxide interface is relatively weak, even in systems where spin-orbit coupling at this interface generates strong interfacial magnetic anisotropy. We also suggest that the spin-current-induced excitation of nonuniform short-wavelength magnons at the HM/FM interface may function as a mechanism of spin memory loss for the spin-orbit torque exerted on the uniform mode.

Received 23 December 2019
Accepted 28 February 2020

DOI:https://doi.org/10.1103/PhysRevApplied.13.034038

Physics Subject Headings (PhySH)

Magnetic anisotropy Magnons Spin torque Spin-orbit coupling Spintronics Surface & interfacial phenomena

Heterostructures Magnetic multilayers

Condensed Matter, Materials & Applied Physics

Authors & Affiliations

Lijun Zhu ^1,*, Lujun Zhu ², D.C. Ralph^1,3, and R.A. Buhrman¹

¹Cornell University, Ithaca, New York 14850, USA
²College of Physics and Information Technology, Shaanxi Normal University, Xi’an 710062, China
³Kavli Institute at Cornell, Ithaca, New York 14853, USA

^*lz442@cornell.edu

Article Text (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 13, Iss. 3 — March 2020

Subject Areas

Reuse & Permissions

Access Options

Article Available via CHORUS

Download Accepted Manuscript

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
Results for $Co$ -based heterostructures. (a) $α$ vs t_Co⁻¹, (b) 4πM_eff vs t_Co⁻¹, (c) $β_{TMS}$ vs (2K_s/M_s)² for magnetic bilayers of ${Si O}_{2} / Co$ $(t_{Co}) / Pt$ (4.5 nm) (series A1), $Pt$ (4.5 nm)/ $Co$ $(t_{Co}) / Mg O$ (A2), $Pt$ (4.5 nm)/ $Hf$ (0.3 nm)/ $Co$ $(t_{Co}) / Mg O$ (A3), and $Ta$ (1 nm)/ $Co$ $(t_{Co}) / Mg O$ (A4). The solid lines represent the best fits of the data to Eq. (4) in (a) and to Eq. (5) in (b). Cross-sectional dark-field transmission-electron-microscopy images of (d) ${Si O}_{2} / Co$ (2.3 nm)/ $Pt$ (4.5 nm) (A1), (e) $Pt$ (4.5 nm)/ $Co$ (2.5 nm)/ $Mg O$ (A2), and (f) $Pt$ (4.5 nm)/ $Hf$ (0.3 nm)/ $Co$ (2.3 nm)/ $Mg O$ (A3). In all cases, both the $Pt$ and the $Co$ layers have a polycrystalline texture, but the roughness is much greater in (d) than in (e) and (f).
Reuse & Permissions
Figure 2
Results for ${Ni}_{81} {Fe}_{19}$ -based heterostructures. (a) α vs t_NiFe⁻¹ and (b) 4πM_eff vs t_NiFe⁻¹ for ${Si O}_{2} / {Ni}_{81} {Fe}_{19}$ (t_NiFe)/ $Pt$ (4.5 nm) (series B1), $Pt$ (4.5 nm)/ ${Ni}_{81} {Fe}_{19}$ (t_NiFe)/ $Mg O$ (B2), $Pt$ (4.5 nm)/ ${Ni}_{81} {Fe}_{19}$ (1.8, 2, 3 nm)/ $Mg O$ (2 nm) (B3), $Mg O$ (2 nm)/ ${Ni}_{81} {Fe}_{19}$ wedge/ $Mg O$ (B4). The lines in (a) represent the best fits of the data to Eq. (4); the straight lines in (b) represent the best linear fits of the data to Eq. (5).
Reuse & Permissions
Figure 3
Results for ${Fe}_{60} {Co}_{20} B_{20}$ -based heterostructures. (a) α vs t_Fe-Co-B⁻¹ and (b) 4πM_eff vs t_Fe-Co-B⁻¹ for $Pt$ (4 nm) $/ {Fe}_{60} {Co}_{20} B_{20} wedge / Mg O$ (series C1), $Ta$ (4 nm) $/ {Fe}_{60} {Co}_{20} B_{20}$ wedge/ $Mg O$ (C2), and $Mg O$ (2 nm) $/ {Fe}_{60} {Co}_{20} B_{20}$ wedge/ $Mg O$ (2 nm) (C3). The lines in (a) represent the best fits of the data to Eq. (4); the straight lines in (b) represent the best linear fits of the data to Eq. (5).
Reuse & Permissions
Figure 4
Effects of TMS in ${Au}_{25} {Pt}_{75} / {Fe}_{60} {Co}_{20} B_{20} / Mg O$ heterostructures. (a) Dependence of the ferromagnetic-resonance linewidth (ΔH) on frequency (f) and (b) f vs FMR resonance field (H_r) for ${Au}_{25} {Pt}_{75}$ (4 nm)/ ${Fe}_{60} {Co}_{20} B_{20}$ (1.6 nm) $/ Mg O$ (sample C4), ${Au}_{25} {Pt}_{75}$ (4 nm)/ $Pt$ (0.5 nm)/ $Hf$ (0.25 nm)/ ${Fe}_{60} {Co}_{20} B_{20}$ (1.6 nm)/ $Mg O$ (C5), and ${Au}_{25} {Pt}_{75}$ (4 nm)/ $Pt$ (0.5 nm)/ $Hf$ (0.25 nm)/ ${Fe}_{60} {Co}_{20} B_{20}$ (1.6 nm)/ $Hf$ (0.1 nm)/ $Mg O$ (C6). (c) and (d) The same quantities after annealing the samples at 450°C for 1 h. All of the data are taken with flip-chip FMR on unpatterned sample pieces. The lines represent the best fits of the data to Eq. (2) in (a) and (c) and to Eq. (3) in (b) and (d).
Reuse & Permissions
Figure 5
(a) Schematic depiction of ST-FMR devices with different orientations (φ) with respect to the gradient of a FM wedge layer. Dependence of (b) magnetic damping and (c) demagnetization field on φ for $Pt$ (5.3 nm)/ $Co$ (2.1 nm) bilayers obtained from a $Co$ -wedge sample, indicating the absence of any anisotropy due to the wedge shape. (d) Dependence of magnetic damping on t_Co⁻¹ for $Pt$ (4.5 nm)/ $Co$ (1.8–6.6 nm)/ $Mg O$ where t_Co is varied either by using a $Co$ wedge (series A2) or by using separate nonwedge-shaped $Co$ layers (A6). All the data are taken using ST-FMR.
Reuse & Permissions
Figure 6
(a) Possible involvement of nonuniform magnons in spin-orbit-torque process. (b) ξ_FMR⁻¹ vs t_Co⁻¹ (ST-FMR measurement), (c) linear dependence of V_a on –V_AH/2(H_k+H_in) (harmonic response, t_Co= 2.3 or 2.5 nm), (d) $ξ_{DL}^{E}$ vs $K_{s}^{ISOC}$ for $Co / Pt$ (sample A1), $Pt / Co$ (sample A2), and $Pt / Hf / Co$ (sample A3). The $Pt$ layer is 4.5 nm thick in all cases, and the $Hf$ layer is 0.3 nm thick. The straight lines in (b)–(d) represent the best linear fits to the data. In (d), $ξ_{DL}^{E}$ for the $Pt / Hf / Co$ samples is the value after correction for the (28 ± 6)% attenuation of spin current in the 0.3-nm-thick $Hf$ spacer layer (see Ref. [48] for more details).
Reuse & Permissions

Physical Review Applied