No abstract

We experimentally investigate and quantitatively analyze the spin Hall magnetoresistance effect in ferromagnetic insulator/platinum and ferromagnetic insulator/nonferromagnetic metal/platinum hybrid structures. For the ferromagnetic insulator, we use either yttrium iron garnet, nickel ferrite, or magnetite and for the nonferromagnet, copper or gold. The spin Hall magnetoresistance effect is theoretically ascribed to the combined action of spin Hall and inverse spin Hall effect in the platinum metal top layer. It therefore should characteristically depend upon the orientation of the magnetization in the adjacent ferromagnet and prevail even if an additional, nonferromagnetic metal layer is inserted between Pt and the ferromagnet. Our experimental data corroborate these theoretical conjectures. Using the spin Hall magnetoresistance theory to analyze our data, we extract the spin Hall angle and the spin diffusion length in platinum. For a spin-mixing conductance of 4 × 10 14 −1 m −2 , we obtain a spin Hall angle of 0.11 ± 0.08 and a spin diffusion length of (1.5 ± 0.5) nm for Pt in our thin-film samples.

We investigate the spin Hall magnetoresistance in thin film bilayer heterostructures of the heavy metal Pt and the antiferromagnetic insulator NiO. While rotating an external magnetic field in the easy plane of NiO, we record the longitudinal and the transverse resistivity of the Pt layer and observe an amplitude modulation consistent with the spin Hall magnetoresistance. In comparison to Pt on collinear ferrimagnets, the modulation is phase shifted by 90 • and its amplitude strongly increases with the magnitude of the magnetic field. We explain the observed magnetic field-dependence of the spin Hall magnetoresistance in a comprehensive model taking into account magnetic field induced modifications of the domain structure in antiferromagnets. With this generic model we are further able to estimate the strength of the magnetoelastic coupling in antiferromagnets. Our detailed study shows that the spin Hall magnetoresistance is a versatile tool to investigate the magnetic spin structure as well as magnetoelastic effects, even in antiferromagnetic multi-domain materials. J c J s σ large ρ J c J s σ FMI HM FMI HM (a) (b) j t n J s stt J s back J c J c J s σ AFI HM AFI HM J s stt J s back FMI/HM bilayer AFI/HM bilayer m m * stephan.gepraegs@wmi.badw.de

Magnons are the elementary excitations of a magnetically ordered system. In ferromagnets, only a single band of low-energy magnons needs to be considered, but in ferrimagnets the situation is more complex owing to different magnetic sublattices involved. In this case, low lying optical modes exist that can affect the dynamical response. Here we show that the spin Seebeck effect (SSE) is sensitive to the complexities of the magnon spectrum. The SSE is caused by thermally excited spin dynamics that are converted to a voltage by the inverse spin Hall effect at the interface to a heavy metal contact. By investigating the temperature dependence of the SSE in the ferrimagnet gadolinium iron garnet, with a magnetic compensation point near room temperature, we demonstrate that higher-energy exchange magnons play a key role in the SSE.

Using X-ray magnetic circular dichroism (XMCD) measurements, we explore the possible existence of induced magnetic moments in thin Pt films deposited onto the ferrimagnetic insulator yttrium iron garnet (Y3Fe5O12). Such a magnetic proximity effect is well established for Pt/ferromagnetic metal heterostructures. Indeed, we observe a clear XMCD signal at the Pt L3 edge in Pt/Fe bilayers, while no such signal can be discerned in XMCD traces of Pt/Y3Fe5O12 bilayers. Integrating the XMCD signals allows to estimate an upper limit for the induced Pt magnetic polarization in Pt/Y3Fe5O12 bilayers.PACS numbers: 78.70. Dm, 75.70.Cn, 75.50.Dd, 81.15.Fg Pure spin currents are a fascinating manifestation of spin physics in the solid state.1-6 Experimentally, the generation or detection of spin currents is often based on the interconversion of spin and charge currents, taking advantage of the spin Hall or the inverse spin Hall effect, respectively. 1,3,4,6 This makes normal metal/ferromagnetic metal (NM/FMM) or normal metal/ferromagnetic insulator (NM/FMI) heterostructures very attractive.7 In so-called spin pumping 4,5,8,9 or spin Seebeck experiments, 6,10-13 the magnetization in the ferromagnetic constituent is driven out of thermal equilibrium, and the ensuing spin current into the normal metal (NM) layer is detected via the corresponding inverse spin Hall electrical current in these heterostructures. Hereby, the paramagnetic NM layer is commonly considered as 'non-magnetic' in the sense that its spin polarization is negligibly small, such that magneto-resistive or magneto-thermo-galvanic effects in the NM layer can be safely ignored.6,14 This assumption appears reasonable for NM/FMM bilayers, in which the unavoidable magneto-thermo-galvanic response of the FMM layer dominates. In contrast, for the case of NM/FMI structures, the absence of an induced spin polarization in the NM layer and thus the complete absence of magnetothermo-galvanic effects in the FMI such as the anomalous Nernst effect are considered as an advantage and even exploited for the interpretation.6,14 Very recently, magnetic proximity effects in NM/FMI structures were inferred from electrical and thermal magnetotransport measurements.15 Thus, a careful investigation of a possible finite induced magnetic polarization in the NM layer in NM/FMI heterostructures is essential for the correct interpretation of spin current related phenomena. In NM/FMM heterostructures and alloys using 3d and 5d elements, the presence of an induced spin polarization in the NM layer in proximity to the interface has been observed by X-ray magnetic circular dichroism (XMCD) experiments.16-20 To our knowledge, no such investigations have been performed in NM/FMI heterostructures.