The magnetic scattering of Fe atoms in Au is measured as a function of temperature in the range between 0.08 and 4 K by means of weak localization. We find a strong temperature dependence with maximum scattering at the characteristic temperature TK ~ 1 K of this well-known Kondo system. This is a new and direct experimental approach to observation of the theoretically predicted Kondo maximum and is analogous to pair-breaking experiments in superconductivity.PACS numbers: 75.20.Hr, 71.55Jv During the last few years weak localization has been developed into a powerful new method in solid-state physics (see, for example, Bergmann 1 and Lee and Ramakrishnan 2 ). Magnetoresistance measurements on disordered thin films correspond to a time-of-flight experiment with conduction electrons, yielding their inelastic lifetime, and spin-orbit and magnetic scattering times. In particular, the opportunity of directly observing magnetic scattering rates promises new insights into the Kondo problem which is recently again receiving considerable interest. 3 " 5The scattering of conduction electrons by Kondo impurities can be separated into two contributions, the spin-nonflip and the spin-flip parts. Only the latter causes dephasing in superconductivity and in weak localization. This has been intensively studied in superconducting alloys where it is possible to obtain this rate from the T c depression at different concentrations 6 and the dependence of H c i on temperature. 7 These measurements show an increase of the spin-flip rate with decreasing temperature 8 in the temperature range above the Kondo temperature T&. The predicted maximum of the rate at T& and its decrease for T < T& could, however, not be verified by this method. Therefore, the theoretical prediction of a maximum in the spin-flip scattering waited almost 20 yr for an experimental proof. Using the method of weak localization we find for the first time a maximum of the spin-flip rate in the Kondo system AuFe and a clear decrease of this rate below T&.Weak localization as a method to measure the spin-flip rate has a number of advantages over the superconducting experiments. It does not require a superconducting host, nor does it need different samples for every point in the T dependence. It can be applied to convenient values of 7"K and small concentrations (to stay within the single-impurity limit) and measures the temperature dependence of the spin-flip scattering rate in a wide temperature range.When weak localization is employed it is favorable to use disordered thin films. l To prepare very homogeneous and highly disordered films, they are quench condensed inside a dilution refrigerator while the substrate is maintained at 4 K. This has the very desirable feature that one can produce multilayers step by step and analyze them after each evaporation step. Therefore, it is possible to separate the additional scattering caused by a small fraction of an atomic layer of Fe atoms from the total phase-breaking scattering. It is not necessary to obtain the informatio...

No abstract

No abstract

No abstract

No abstract