Abstract

The placement of a transition metal impurity in a noble metal host induces a collective response in the electron gas which attempts to screen the magnetic moment of the impurity. At low impurity concentration, enhanced scattering of electrons from the screened impurity leads to a characteristic logarithmic increase of the resistance with decreasing temperature, the well-known Kondo effect [1]. If the concentration of impurities is high, the interaction of the screening electrons around one impurity with those around other impurities leads to an effective impurity-impurity interaction, the RKKY (Ruderman-Kittel-Kasuya-Yosida) interaction [2]. At high temperatures, the thermal energy of the impurity spins is sufficient to overcome the RKKY interaction, and each individual spin is free to rotate independently. As the temperature is reduced, however, the impurity spins are increasingly fixed in random orientations by the RKKY interaction. The onset of this spin-glass order is signaled by a drop in the resistance of the host metal due to the reduced magnetic scattering of the conduction electrons. In combination with the Kondo effect at higher temperatures, this gives rise to a maximum in the resistance as a function of temperature which is characteristic of spin glasses [2]. Earlier work on magnetic impurities in metals concentrated on the properties of bulk materials; recently, with the opportunities presented by nanolithography, interest has focused on the properties of samples whose dimensions are comparable to relevant microscopic length scales. A number of such microscopic length scales have been proposed for both the Kondo effect [3] and spin glasses [4]. The hope is that measurements on mesoscopic samples would allow one to verify directly the existence of these microscopic length scales. However, the experimental evidence in both Kondo systems and spin glasses has so far been inconclusive. For example, measurements by some groups [5,6] of the Kondo effect in thin films, wires, and small point contacts defined by break junctions show a definite size dependence, but on vastly different length scales, while measurements by other groups [7] on AuFe wires found no size dependence on the Kondo effect. The situation is similar for samples in the spin-glass regime [8 ‐ 11]. Thus the issue of the existence of fundamental length scales in both the spin-glass and the Kondo regimes remains open. In this Letter, we report on measurements of the low temperature differential resistance RsId › dV ydI of AuFe wires as a function of dc current bias I .A s reported previously by other groups in measurements on CuCr wires [9] and point contact break-junction devices [10], we find that the shape of RsId reflects the behavior of the temperature dependent resistance RsT d, in that it has a maximum at a particular current Im. In addition, however, we find that RsId is asymmetric in I, even in zero magnetic field. The asymmetry is small at high temperatures, but grows by more than an order of magnitude as the temperature is lowered, indicating that it is associated with enhanced spin scattering at low temperatures. The degree of asymmetry is sample specific, being larger in some samples and smaller in other nominally identical samples. Furthermore, we find that the asymmetric component of RsId is sensitive to the particular configuration of contact leads used in a four terminal measurement. We also find that the asymmetry depends on the size of the wire, being generally larger for shorter and narrower wires and disappearing entirely for very long samples. These observations are indicative of the mesoscopic nature of this phenomenon. The samples in this experiment were patterned onto oxidized silicon substrates by conventional e-beam lithography techniques. After thermal deposition of 99.999% Au, the samples were implanted with Fe ions at energies and dosages calculated to give impurity concentrations of 0.2 and 0.4 at. % [12]. All samples of one concentration were fabricated and ion-implanted at the same time to ensure uniform film properties. The inset of Fig. 1(a) shows a schematic of one of the samples. The thickness of the films was 33 nm, and the sheet resistance Rh was ,1 V at 4.2 K after ion implantation. The samples were measured in 4 He and 3 He cryostats (for the 0.4 at. % samples),

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call