The intermediate valence compounds are one of the complicated compounds showing different types of strongly correlated electrons behaviour due to the competition between different contributions to their properties. The strongest valence fluctuations are observed in the Ce and Yb compounds. However, all YbxNiy are known as trivalent. In YbNi₄Si it was expected that the Si addition can slightly modify the valence state of Yb but the magnetic contribution of Ni should be negligible like in the other RNi₄M compounds (R = rare earth, M = B, Al, Cu). The YbNi₄Si compound was prepared by induction melting of stoichiometric amounts of the constituent elements in a water-cooled boat, under an argon atmosphere. The ingots were inverted and melted several times to ensure homogeneity. The crystal structure was established by a powder X-ray diffraction technique, using Cu-Ka radiation. YbNi₄Si compounds crystallize in the hexagonal CaCu₅-type of structure, space group P6/mmm. Yb atoms occupy the (1a) site, Ni(1) the 2c site and Ni(2) and Si are statistically distributed on the 3g positions. The lattice constants are a = 4.820 A and c = 3.996 A for YbNi₄Si. In the Yb-based compounds Yb can exhibit two valence states corresponding to the two almost degenerate 4f configurations: nonmagnetic 4f¹⁴ (Yb²⁺) and magnetic 4f¹³ (Yb³⁺). Therefore, Yb can be treated as a hole counterpart of Ce, which also shows two electronic configurations, i.e., the magnetic 4f ¹(Ce³⁺) and the nonmagnetic 4f ⁰(Ce⁴). Previous studies on YbNi₄Si have revealed its paramagnetic properties with the paramagnetic Curie temperature ?_p = ~0 K and the effective magnetic moment µ_eff= 4.15 mB/f.u. This effective magnetic moment is much lower than the value expected for the free Yb³⁺ ion (4.54 µB). Since magnetic moment of the divalent Yb is zero, the observed reduction of µ_eff can be explained in a natural way assuming a fractional occupation of the excited magnetic state 4f¹³. The Yb²⁺ and Yb³⁺ peaks observed by XPS in the valence band region confirm the domination of the Yb₃₊ valence state. Temperature dependence of the electrical resistivity r(T) in the 4-300 K range is shown in Fig. 1. Below about 60 K the resistivity follows the equation r (T) = r₀ + AT ² with the parameters r₀ = 211 mWcm and A = 0.002 mWcm/K². A relatively large residual resistivity r₀ is probably due to the random distribution of Ni(2) and Si on the 3g site, as has been also observed in CeNi₄Si compound. The quadratic variation of the low temperature resistivity is characteristic of a Fermi liquid. The magnitude of the resistivity is relatively large and the residual resistivity ratio r (300K)/r (4K) = 1.3 is quite small, both findings being probably indicative of considerable atomic disorder. Above 60 K, r(T) was analyzed using the Bloch–Grüneissen–Mott formula. Fig. 2 shows the low temperature part (up to 12 K) of the heat capacity of YbNi₄Si in the applied magnetic fields of up to 9 T. The sharp peak with a maximum at 2.7 K in zero field we ascribe to the transition into a magnetically ordered phase. The applied magnetic field influences strongly this transition. With increasing magnetic field the maximum intensity is decreasing. At 9 T only a small shoulder is present. The small shift of the maximum to the lower temperature at least for the B = 0.1 T could point to the AF character of the transition. The appearance of a phase transition is consistent with the measurements of ac susceptibility where strong maximum is present at this temperature (see Fig. 2.). In addition the applied magnetic field splits this peak and the maximum of the new peak moves to the higher temperatures with increasing field. The inset of Fig. 2 shows magnetization curves for fields up to 8 T and temperatures from 2 K to 30 K. A nonlinear curvature is observed which becomes increasingly pronounced at lowest temperatures with a tendency to the saturation to the value of 2 µ_B. The data can be well described by S = 1/2 Brillouin function even for T = 2 K, which suggests that the curvature in the magnetization curves reflects a normal paramagnetic behaviour.

Figure 1: Temperature dependence of the electrical resistivity (circles) fitted with Eq. (1). Solid line is a fit with the ?_D = 214 K as a free parameter and the dotted line is a fit for fixed value of ?_D = 320 K, as derived from the specific heat analysis. (Inset) Quadratic dependence on temperature in the low temperature range (4–60 K).

Figure 2: Temperature dependence of the heat capacity C(T) of YbNi₄Si.

   We suppose the following scenario. The heat capacity maximum is associated with the antiferromagnetic order. On applying magnetic fields, we expect that T_N is suppressed and the degeneracy of the ground-state doublet is lifted by Zeeman splitting. Accordingly the separation between remaining singlets rises with growing field yielding a shift of the associated Schottky anomaly to higher temperatures. The magnetic field sufficient to suppress the ordering is small of the order about 0.5 T. However, to support our scenario, measurements of the magnetic properties and/or magnetoresistance around the transition temperature are necessary. Taking into account the lowest temperatures of C/T(T²) (for T < 2 K) we obtained ? = 352 mJ/mol.K-2, which is a tendency to a heavy fermion-like behaviour. The thermoelectric power (TEP or S(T)) of Ce- and Yb-based compounds differs considerably from that of the conventional metals and alloys. Instead of a linear temperature dependence of TEP, the heavy fermion systems often show a giant value, one or two orders of magnitude larger than in normal metals. Furthermore, several maxima and minima appear in the thermoelectric power. Such a behavior was explained by Mott based on a relation of the giant TEP to the energy derivative of the enhanced density of states N_D(E) at the Fermi level. Previously, we have performed studies on the CeNi₄Si compound showing a minimum in S(T) at about 8 K. Such a behavior was explained in the framework of the Fischer theory for Kondo systems, where the negative contribution of spin interaction competes with the positive contribution of Kondo effect. In contrast, the TEP of YbNi₄Si below 20 K exhibits rather a linear dependence on temperature (Fig. 3). Besides, as is typical of Ce- and Yb-based compounds TEP is mostly positive for the former and negative for the later. It is a direct consequence of the electronic structure (Yb being a hole counterpart of Ce). The linearity of the S(T) curve of YbNi₄Si for T > 200 K is doubtless related to the diffusion thermopower and the phonon drag certainly plays a minor role.

Figure 3: Temperature dependence of the thermoelectric power of YbNi₄Si – (circles, experiment; line, fit).

M. Reiffers, M. Vasiľová-Zapotoková, M. Timko
A. Kowalczyk, M. Falkowski, T. Toliński (IPM Poznaň, Poland)
J. Šebek, E. Šantavá (Institute of Physics, CAS, Prague, Czech republic)