The Elliott-Yafet spin relaxation mechanism posits linear proportionality between spin and momentum lifetimes in low spin-orbit coupling nonmagnetic metals, and is widely accepted in spintronics. Accurate experimental determination of the Elliott-Yafet proportionality constants (βi) between the spin and momentum relaxation times for individual scattering sources is challenging, however. This is apparent from the literature on nonlocal spin transport in Cu, for example, where reported phonon (βph) and defect (βdef) Elliott-Yafet constants vary by an order of magnitude. In recent work we discovered that even part-per-million-level magnetic impurity concentrations can substantially influence spin relaxation in Cu, via a spin transport analog of the Kondo effect. To clarify whether this could explain the reported variability in βi, here we report on a comprehensive study of spin transport in Cu-based lateral nonlocal spin valves, varying the ferromagnetic contact material, interface structure, Cu thickness, and post-fabrication annealing conditions, resulting in widely varied microstructures and magnetic impurity concentrations. Quantifying the effects of magnetic impurities on charge and spin transport we demonstrate the dramatic, even dominant, effect these can have on spin relaxation rates, and thus extracted βi. Minimization of magnetic impurity effects is achieved via Al interlayer insertion or moderate annealing, restoring the expected temperature dependence for phonon-mediated spin relaxation, and enabling more reliable determination of βi for phonons (740 ± 200), and nonmagnetic defects (240 ± 50). The latter contribution is shown to be dominated by grain boundaries in these polycrystalline Cu films. Cross-sectional transmission electron microscopy measurement of grain sizes in actual nonlocal spin valve devices then establishes a useful empirical relationship between average grain size and spin diffusion length. These measurements highlight the importance of magnetic impurities in metallic spin transport, explain the wide variability in reported βph and βdef in Cu, and elucidate the relationship between metallic spin transport and microstructure.
How much support was provided by MRSEC?
Reporting period for MRSEC
- Period 6