Magnetic susceptibility of a two-component Kondo system: Proposed experiment to distinguish between various models for thec ² dependence of χ

Abstract

Recent experiments on the low-temperature magnetic susceptibility χ of very dilute concentrations of iron impurities in copper show a significant contribution proportional to the square of the iron impurity concentrationc. The two explanations that have been proposed for thisc ² dependence are: (a) a qualitative explanation in which thec ² term in χ arises from the “clustering” of pairs of Fe impurities in the Cu-Fe system; (b) a model in which thec ² dependence arises from the suppression of the spin-compensated state by the many-impurity interaction. In order to study mechanism (b) further, we obtain the magnetic susceptibility for a two-component Kondo system having different Kondo temperatures. We find that thec ²-dependent contributions to χ for the two-component system will be qualitatively different for explanations (a) and (b). We propose an experiment which could possibly distinguish between them. We also find that the suppression of the Kondo temperature by the impurity-impurity interaction yields an effective Kondo temperatureT ^E_K which is concentration-dependent. Thus there may be a large discrepancy between the effective concentration-dependent Kondo temperatureT _K (c) obtained from the magnetic susceptibility χ ∝ [T+T _K (c)]⁻¹ andT _K, the Kondo temperature at infinity dilution. Our results indicate that for Cu-Mn one would have to go to a dilution of less than 1 ppm in order forT _K (c) to approachT _K in the expression for the magnetic susceptibility.