Abstract
We present a model of the macroscopic anisotropy in spin glasses, based on current ideas distinguishing the ‘‘frozen-in’’ and ‘‘adjustment’’ parts of the energy due to anisotropic interactions. Predictions of the model are tested by computer simulations of Ruderman-Kittel-Kasuya-Yosida spin glasses with Dzyaloshinsky-Moriya (DM) anisotropy. Results from the computer simulations are consistent with the identification of the macroscopic anisotropy measured from ESR or transverse susceptibility with a frozen-in ‘‘first-order’’ anisotropy. The model quantitatively predicts the measured ‘‘adjustment energy’’ due to the anisotropic interactions, using most of the assumptions used to calculate the ‘‘frozen-in’’ anisotropy which would be measured in a macroscopic sample. Calculations of the macroscopic anisotropy for experimental samples using two alternate forms of the microscopic DM interaction, which give anisotropies 20 to 150 times greater than the value derived from hysteresis experiments for CuMnPt depending on the parameters used, are presented in detail.