NARROW-GAP SIGNATURE OF FEXCO1-XSI SINGLE-CRYSTALS

Abstract Heat capacity, resistance, and magnetic susceptibility have been measured on FeSi, Fe0.90Co0.10Si, Fe0.10Co0.10Si, and CoSi single crystals. From resistance experiments, the activation energy (Δ) shows a small variation from Δ≈310 K in FeSi to Δ≈325 K for the Fe0.90Co0.10Si compound. For the 90% Co compound the gap signature disappears completely. Heat capacity experiments performed in the temperature range 1.5 K


Introduction
Many years have elapsed since the discovery of anomalous magnetic behavior in the small gap semiconductor FiSi (see ref. [1] and references therein). This intermetallic compound presents an unusual behavior of the magnetic susceptibility x(T) which exhibits a pronounced maximum in the vicinity of 500 K. Furthermore, neutron diffraction studies [2] show no ordering at low temperatures. A semiconducting band model [1] has been proposed , with a narrow gap, to describe the temperature dependence of x(T) and heat capacity. However, the gap model is not completely accepted to describe the thermal properties of FeSi. The possibility of having a temperature-induced paramagnetic moment has been pointed out theoretically [3], and experimentally [4] on the basis of neutron-scattering measurements. In the case of Fe x-COl_xSi alloys, weak itinerant ferromagnetism, with a helical spin structure (long period >300/~), has been reported [5] for 0.3 < x < 0.90.
In view of the controversy surrounding an interpretation of the physical properties of FeSi, we have investigated the gap formation as a function of Co content. Here we present an account of thermodynamic and transport measurements in a wide temperature range for FexCol_xSi single crystals.

Materials and experimental techniques
Single crystals were grown via chemical transport. Approximately 1 g of starting material and 60 mg of iodine were sealed in a quartz tube. The tube was placed in a gradient furnace with starting material at the hot end (900°C) and samples growing a the cold (750°C) end. Samples of a few mm 3 grew over the course of one week. Resistance was measured using a conventional AC four-probe method in the temperature range 4.2 K < T < 300 K. Magnetic susceptibility measurements were performed on a SQUID magnetometer between 2 and 350K. Finally, heat capacity experiments were carried out using a quasi-adiabatic thermal relaxation technique in the temperature range 1.5K< T<20K. Figure 1 shows the temperature dependence of the resistance of FeSi, Feo.9oCoo.10Si , Feo.mCoo.9oSi and CoSi between 4.2 and 300 K. The resistance of FeSi and the 10% Co alloy increase as the temperature decreases from room temperature to the lowest temperature investigated. The rapid rise in the resistance for these two compounds can be interpreted as a gap formation in the electronic density of states. The activation energy A can be estimated by plotting In R as a function of 1/T; a linear regime is observed from 0921-4526/93/$06.00 © 1993-Elsevier Science Publishers B.V. All rights reserved holder. Increasing the Co content moves the minimum in x(T) towards higher temperatures.

Results and discussion
Heat capacity measurements have been performed in a wide temperature range. In fig. 3 we plot C~ T as a function of temperature for FeSi, Fe090Co010Si, Feo.~oCoo 90Si and CoSi. The FeSi compound presents the lowest heat capacity compared with all alloys and the CoSi compound. By increasing the Co content an upturn in C/T at low temperature appears. Correspondingly an upturn at low temperatures was also observed in the magnetic susceptibility. The C/T rise with decreasing temperatures is drastically suppressed for the CoSi compound. Using a Debye and electronic (C : TT + ]3T 3) fit for the FiSi and CoSi data gives 3' = 1 mJ mol ' K 2, /3 = 0.035 mJ moli K 4 (~D = 33 to 167K for FeSi and from 33 to 167K for the Fe09oCo0.1oSi compound. The gap energy (2A) estimated from the linear regime is ~620 and 650 K for FeSi and Feo.9oCoo 10Si, respectively. The lowtemperature feature is attributed to impurities incorporated during sample growth. In fig. l(b) we show the resistance of CoSi and Feo.10Coo.9oSi (inset); metallic behavior is observed for both in the temperature range investigated.
We show in fig. 2 the magnetic susceptibility as a function of temperature for FeSi, CoSi and Feo 9oCoo ,~Si. A very pronunced rise in x(T) is observed for the 10% Co compound at temperatures below 57 K. This rise is consistent with moment formation with increasing Co content. For FeSi a considerable background signal is observed in our susceptibility experiments. This appears to be due to the low sample contribution compared with that of the sample CoSi 480K) and 3,=2mJmol-~K -2 and /3=0.056 mJ mol ~ K -4 (¢9 D = 410 K) for the two compounds. In our temperature range, a large variation is observed in the heat capacity of both dilute and highly concentrated Co materials as compared to 'pure' CoSi and FeSi. It is possible that magnon excitations in a helimagnetic phase could be present, since we are close to the known phase boundary for magnetic order [5]. However, we see no evidence for this is our magnetic susceptibility measurements.
In addition to substitutions of Fe and Co, we have substituted P for Si. The advantage of this substitution is that there is no disruption of the Fe sublattice, and phosphorous has no magnetic moment. Figure 4 shows the temperature-dependent electrical resistance and magnetic susceptibility of FeSi:_xp ~ with nominal x = 0.05. These samples were grown using a flux (Sb) growth technique [6]. The magnetic susceptibility is qualitatively similar to that of Feo.9oCoo.,oSi. This supports the hypothesis that the low-temperature Curie tail in this compound is not due to the Co moment, but associated with the disrupted FeSi lattice itself. The resistance, on the other hand, is very different from the 10% Co substitution. The phos- phorus substitution (x = 0.05) leads to a virtual metalization of the temperature-dependent electrical resistance, although a maximum is observed around 100 K. FeSi shows striking similar properties to the group of rare earth and actinide compounds known as hybridization gap materials, in which the gap is roughly one order of magnitude smaller. The similarities extend to details of the x(q, oJ) as measured via neutron scattering. Dilution studies we have reported [7] on the Ce3Bi4Pt 3 material in this class show quite similar behavior with doping in both rare earth and non-rare earth sites. The comparison of these results further supports our contention that the underlying physics in the rare earth on the transition metal materials are closely similar.
Work at Los Alamos was performed under the auspices of the US Department of Energy. Part of the research of one of us (MBM) was supported by National Science Foundation, grant No. DMR87-21455.