New Ce heavy-fermion system: CeCu6

We have discovered a new heavy-fermion system, CeCu6, with a large susceptibility (X=0. 027 emu/moleG at 1. 5 K) and a large, temperature-dependent specific heat y below 10 K that is 840 mJ/moleK at 1. 8 K and extrapolates to 1, 6 J/moleK at T=O in analogy with CeA13. High-field specific-heat measurements agree almost perfectly with a narrow-band picture first proposed for UBe~3.


Materials Science and Technology Division,
Los A/amos National Laboratory, Los Alamos, New Mexico 87545 (Received 24 April 1984) We have discovered a new heavy-fermion system, CeCu6, with a large susceptibility (X=0.027 emu/moleG at 1.5 K) and a large, temperature-dependent specific heat y below 10 K that is 840 mJ/moleK at 1.8 K and extrapolates to 1,6 J/moleK at T=O in analogy with CeA13. High-field specificheat measurements agree almost perfectly with a narrow-band picture first proposed for UBe~3.
Since the discovery' by Steglich et al. in 1979 of bulk superconductivity in CeCu2Si2, an f electron system with highly correlated electrons having effective masses about 200 times the bare-electron mass, a great deal of interest has focused on these so called "heavy-fermion" systems. As a class of materials, the seven heavy-fermion systems known' 7 to date all contain f electrons and have large specific-heat y values (C= y T+/3T3, from which the large effective masses are calculated), a large ()4 A) 4f or Sf atom separation, a high-temperature susceptibility that follows a Curie-Weiss law, a large low-temperature susceptibility, and usual temperature dependence in their resistivity. A possible division of these seven sytems at low temperatures in between those that go superconducting, ' magnetic, or neither6 7 as is the case in CeA13 [y( T = 0) = 1.6 J/mole K2] and Uc»Npa6sBei3 [y(T=O) ) 1.1 J/mole K ]. All but three of these systems have y's that are rapid functions of temperature, varying from under 200 mJ/mole K' at 10 K to over 1000 mJ/mole K2 below 1 K.
We have focused a search for other heavy-fermion systems on Ce and U compounds with large f-atom separations that are not known to order magnetically at low temperatures.
We report here on CeCu6, an orthorhombic structure first solved at Los Alamos with Ce surrounded by a cage of 19 Cu atoms and a Ce-Ce spacing of 4.83 A. We have measured the resistivity from 1.4 to 300 K, dc susceptibility from 1.4 to 100 K, and specific heat from 1.8 to 38 K in zero and 11 T applied field on material prepared by arc melting together the pure constituents and characterized as single phase by x-ray powder diffraction. Measurements of the ac susceptibility showed no evidence of superconductivity down to 0.040 K.
The resistivity data measured by a standard four probe technique are shown in Fig. 1. The data have a minimum at about 29 K and a broad maximum centered at 11 K, with a sharp drop at lower temperatures.
The dc susceptibility, shown in Fig. 2, was measured in a Faraday balance on a 69 mg piece of the arc melted button. Susceptibility data for CeCu6 above 77 K have been reported in the literature and agree with our data in the region of overlap. The magnitude of our measured susceptibility for CeCu6 at 1.5 K, 0.027 emu/moleG is enormousapproximately twice the value found for Uae~3. Also noteworthy in our susceptibility data is the definite feature at about 30 K, which correlates well with the resistivity minimum.
The specific heat in zero field from 1.8 to 38 K, shown in Fig. 3, was measured in a small sample calorimeter on the same piece of material used for the susceptibility measurements. At temperatures below 8 K we see the rapid increase in Cj T that is characteristic of most heavy-fermion sytems. In fact, these low-temperature data are within a few percent of published specific heat results for CeA13. Thus, we expect the y(T = 0) value for CeCu6 to also be about 1.6 J/mole K2. The lowest-temperature point is at 1.85 K. The absolute accuracy of the data is k3% below 20 K, and k7% at 38 K. However, the relative precision of the data at high temperatures is +2%, i.e. , the shoulder observed around 30 K is a real feature. The line drawn through the zero-field data is solely to serve as a guide to the eye.
When plotted as C vs T, these specific heat data for CeCu6 show no evidence for a low-temperature peak, contrary to the results' for CeCu2 Si2 and for' UBei3, but rather a broad shoulder below 4 K.
The higher-temperature data plotted as C/T vs T' in Fig.   3 do show a washed out peak, or shoulder, around 30 K that correlates with the structure seen in the resistivity and susceptibility data. A possible explanation for this anomaly is low-lying crystal field excitations of the Ce + ground state. (The assignment of the ground state as Ce3+ is based on high-temperature susceptibility, Ref. 9.) In order to further understand the nature of the highly correlated electrons in heavy-fermion systems in general, and in CeCu6 in particular, we have measured the specific heat from 1.8 to 17 K in an 11-T applied magnetic field.
These results are shown in Fig. 4. There is a large decrease in C caused by the 11-T field at low temperatures ( -19% at 1.8 K), followed by a crossover (i.e. , zero field effect) at about 3.3 K to an increase in C with field at higher temperatures. This increase is also quite large (+19% at 7.4 K), but falls off at still higher temperatures (+5% at 16.7 K).
These starting results follow almost perfectly what one would expect from a simple narrow-band (30 -40 K wide) model. Such a model cannot only explain the temperaturedependent y in zero field but has also been used" in the case of UBei3 in 11 T to predict, due to band splitting, an increase in C field "above 3.5 K which is a maximum around 6 K falling off at higher temperatures, with a decrease in the specific heat below 3.5 K. " In the case" of UBei3, no net decrease in C with field was found below 3.5 K; merely the observed increase became smaller in this region, while the other predicted features (maximum around 6 K and tapering off of the field-induced increase in C at higher temperatures) were observed. %e believe the almost perfect agreement between a narrow-band model for the specific heat and the measured data, both in zero and applied field, for CeCu6 is strong evidence for the single-particle, narrow-band approach and contradicts the many-body, paramagnon viewat least as a first approximation. This is borne out by recent pointcontact tunneling spectroscopy results on UPt3 (Ref. 12) and CeCu2Si2, ' where sharp features in the density of states are observed of 6 and 1.5 meV (or 70 and 20 K) widths, respectively.