METAL-INSULATOR TRANSITION IN PYRITE TYPE NiS2-xSex SYSTEM

Specific heat and Raman scattering were measured on sintered samples with x around 0.5, which show a sharp metal-to-insulator transition (MIT) with increasing temperature. Specific heat below MIT can be well expressed by C = yT + P T ~ + 6 ~ ~ In T with parameters comparable to those of TiBez. Spectrum of Raman scattering shows no anomaly through MIT. Since the first suggestion by Mott, the phenomenon of the metal-insulator transition (MIT) has been extensively studied from both experimental and theoretical points of view. MIT's in various compounds were reported so far. However, many of these compounds ehibit simultaneously some crystal distortion through MIT. So, the theoretical interpretation of them is not so simple. NiSz-,Se, system also shows MIT around x = 0.5, and the absence of crystal distortion through MIT was reported from X-ray diffraction measurements [I]. Here, we report specific heat data and Raman scattering data which give the information on the local crystal distortion. The samples used here were prepared by sintering the pressed mixture of Nil S, Se-powders in an evacuFig. Temperature dependence of the specific ated silica tube at about 700 "C for 7 days. The electriheat for N ~ s ~ . ~ ~ s ~ ~ . ~ ~ . ~h~ solid curve is calculated with cal resistivity p and the magnetic susceptibility x were OD= 365 K. also measured. For 0.48 < x < 0.53, as temperature increases through MIT, p increases discontinuously by three orders of magnitude (from to 10-I 0-cm), and x also does but by a few tens percents. Specific heat measurements were made, by means of a heat pulse technique, on two samples with MIT temperature (Tt) of 35 K and 38 K, respectively. Figure 1 shows, by the dots, the temperature dependence of the observed specific heat for x = 0.51 and Tt = 35 K sample. A clear latent heat of 1 J/mol.K was observed at Tt. In figure 1 is also shown, by the solid curve, the calculated curve with Debye temperature (OD) of 365 K. From the observed and the calculated curves, TN is estimated as about 70 K, MIT occuring in the antiferromagnetic phase. The difference between the dots and the solid curve is considered due to the antiferromagnetic ordering and the itinerant electron contribution. Figure 2 shows, by the dots, the observed C/T us. T~ curve at low temperatures below Tt, replotted from figure 1. As seen, C cannot be expressed by Tand terms only. The solid curve in figure 2 is calculated by a function Fig. 2. C/T vs. T~ plot from figure 1. The solid curve is calculated by equation (1) (see Tab. I for the parameters). with y, ,8 and S given in table I. The agreement between the observed and the calculated curves is very good. For x = 0.52 and Tt= 38 K sample, a similar data was obtained with y, P and 6 somewhat larger (by about ten percents) than those for x = 0.51 sample. It is noted that these values are comparable to those of a nearly ferromagnetic metal TiBez (see Tab. I). If 'Fiesearch Institute of Applied Electricity, Hokkaido Univ., Japan. Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1988880 C8 188 JOURNAL DE PHYSIQUE Table I. Experimentally obtained three parameters in equation (1) for several compounds. The units are chosen so that C is measured in mJ/mol.K. (*) Present work. Reference [2]. (b) J. J. M. Fyanse et at., Physica 126B (1984) 116. (C) R. J. Trainor et al., Phys. Rev. Lett. 34 (1975) 1019. (d) S. K. Dhar et al., Phys. Rev. B 36 (1987) 341. Compounds NiSl.rgSeo.s~ (*) TiBe2 (")