Modified Landau-Lifshitz-Gilbert Dynamics Elucidating Asymmetric Spin Transport in NM/FM System

Spin pumping, usually referred as the injection of the spin current pumped by thermal agitation of non-equilibrium magnetization in ferromagnetic (FM) or ferromagnetic insulator (FMI) materials,¹ has been granted trivial without any further discussion on the pumped direction (FM to NM / NM to FM, where NM refers to normal metal). The pumped spin current from FM to NM results from temperature gradient ranged between FM and NM. Over the decade, spin pumping has been investigated thoroughly armed with theories such as linear response theory,¹ and considered consistent with conducted experiments on spin-to-charge conversion. Theoretically, reversing the direction of the temperature gradient should invert the sign of the spin current signal, but its intensity is identical between the two configurations.<br/>On the other hand, repeated experiments on Pt/YIG bilayer structure with temperature gradient up to 20 K (-20 K for opposite direction) shows inconsistency between the experiment and theory, where the experiment kept showing nonlinearities between FM → NM , NM → FM pumped spin current. Herein, we assumed there exists corrections on some parameters when changing the injected direction of the pumped spin current, and carefully suggest modified LLG approach where different magnon lifetime contributes to the asymmetric spin transport across NM/FM interface. However, these kind of minor rectification cannot be solely explained with modified LLG, as additional experimental factors also contributes to the nonlinearities of the spin current signal, such as the change in the average temperature at the Pt/YIG interface depending on the direction of thermal gradient and resultant temperature-dependent spin voltage originated from the temperature dependence of the spin-Seebeck coefficient S(T),² and turned out to be sometimes more dominant than purely rectified spin current originated from interface-dependent magnon lifetime in our model. Nevertheless, the model suggested would provide meaningful applications for pure spin accompanied transport aided by material selection, configuration of spin voltage detecting device.<br/> <br/>(1) Adachi, H.; Uchida, K.; Saitoh, E.; Maekawa, S. Theory of the spin Seebeck effect. <i>Rep. Prog. Phys</i> <b>2013</b>, 76 (3), 036501.<br/>(2) Basso, V.; Sola, A.; Ansalone, P.; Kuepferling, M. Temperature dependence of the mean magnon collision time in a spin Seebeck device. <i>J. Magn. Magn. Mater.</i> <b>2021</b>, 538.