Extracting Free Energy Landscape Of Agmn Spin Glass Thin Films With 1/F Resistance Fluctuations

Abstract

1/f noise is used to probe the spin glass dynamics in AgMn (12 at.% Mn) mesoscopic thin films with the thickness ranging from 15nm to 80nm. The measurements are made by quenching above the freezing temperature, T f , to a measurement temperature, T < T g . Using the analysis method developed by the previous noise measurement in mesoscopic CuMn samples, the barrier distribution of the spin glass free energy landscape can be extracted. Our measurements show strong temperature dependence in the barrier distribution contradicting to the recent magnetic measurements of the spin glass thin films showing a temperature-independent distribution. When the film thickness increases, the stronger temperature dependence of the barrier distribution is observed and consistent with the previous noise measurements in CuMn thin films. In addition, the observation of the mismatch of the freezing temperature and the noise onset temperature, Ton, where the noise increases more than one order magnitude, in the thicker samples is discussed and was not addressed in the previous mesoscopic spin glass measurements. The universal conductance fluctuations (UCF) couple the resistance fluctuations to the magnetic fluctuations and lead to a huge noise increase at low temperatures. To observe the UCF, the condition where the inelastic scattering length, L i , must be larger than the elastic mean free path, l e , needs to be satisfied and sets the corresponding temperature, T UCF . For observing the strong noise increase from the spin glass transition, both conditions, T < T f and T < T UCF , must be satisfied and we conclude T on = min(T f , T UCF ). Because T UCF is determined by the intrinsic properties of the spin glass materials, the noise sensitivity is limited above a certain film thickness as T f > T UCF . The T UCF = 20K is observed in our AgMn material and it is lower than most of T f measured in our samples. Therefore, the weaker thickness dependence of the related magnetic properties extracted from the noise data can be explained by this temperature limit.