Low-temperature state of UCu5: Formation of heavy electrons in a magnetically ordered material.

The formation of a heavy-electron state in a magnetically ordered material is established by measurements of the low-temperature specific heat of UCu~ and UAgCu4. In UCus this state undergoes a continuous but hysteretic phase transition which removes parts of the Fermi surface with a high density of electronic states and leads to a resistivity increase of almost an order of magnitude.

The formation of a heavy-electron state in a magnetically ordered material is established by measurements of the low-temperature specific heat of UCu~and UAgCu4. In UCus this state undergoes a continuous but hysteretic phase transition which removes parts of the Fermi surface with a high density of electronic states and leads to a resistivity increase of almost an order of magnitude.
PACS numbers: 65.40.Em,72.15.Eb,75.50.Ee In two previous publications, ' UCuq was identified as ordering antiferromagnetically at 15 K.. This conclusion was based on data from measurements of the magnetic susceptibility and from neutron-diffraction experiments.
Later measurements of the specific heat and the electrical resistivity confirmed the phase transition but also led to the conclusion, in the course of a more general investigation of UNi5 "Cu"com- the abrupt increase of the low-temperature electronic specific heat for x exceeding 4 was a major argument for this conclusion. The electronic specific-heat parameter y was obtained from data in the temperature range between 1.5 and 30 K by extrapolation of a c~/T vs T2 plot to T = 0 K. An anomalous increase of c~/T with decreasing temperature below 4 K was ascribed to the onset of a Schottky-type contribution to the specific heat due to the spontaneous splitting of nuclear levels in the magnetically ordered matrix.
In this Letter we demonstrate that this increase in c~/T is due to many-body effects that are now a familiar feature of heavy-electron materials and that UCu5 undergoes another phase transition around 1 K which is, so far, of unknown origin but, more important, involves the heavy-mass quasiparticles that lead to the enhanced low-temperature specific heat. To our knowledge, this is the first example of such a distinct enhancement effect that occurs in a magnetically ordered material.
Our reasoning is based on data obtained from measurements of the specific heat and the electrical resis-tivity that were made on well-annealed polycrystalline samples of UCuq and UAgCu4. The specific heat was measured between 0.15 and 21 K. Specimens that were cut from the same respective buttons were used for measurements of the electrical resistivity below room temperature, extending to 0.4 K in the case of UCu5 and to 1.2 K for UAgCu4.
In UCu5 the distance between adjacent U atoms of 4.96 A is quite large. In principle, one would therefore expect an integral occupancy of the Sf-electron shell of the U ions. As mentioned above this was first put in question by van Daal et al. who concluded that although U in UNi5, with a U-U distance of 4.80 A, adopts the Sf' configuration, no integral valence can be assumed for U in UCu5. In the work of Schneider et al. , 5 however, photoemission data indicate that the 5f-electron-state occupation barely changes between UNi5 and UCu5 and these authors concluded that in both cases, hybridization effects between U 5 f and Ni or Cu 3d electrons were important. There was also no evidence for two different final-state Sf' multiplets in the valence-band spectrum of UCu~that would indicate valence fluctuations between U + and U + states. The experimental results that we present below also rather indicate itinerant-Sf-electron behavior that is most likely due to hybridization with 3d electrons.
In Fig. 1 we show the results of our measurements of the specific heat cp of UCu5 and UAgCu4 between 0.15 and 21 K. For temperatures above 1.5 K we plot c~/T vs T2 and in the insets we display c~v s T for temperatures below 0.6 K. For UCu5 we confirm the data of van Daal et al. obtained for T ) 1.5 K with the main features of an anomaly induced by magnetic ordering around 15 K and the upturn of c~/T with decreasing temperature below 4 K. Replacement Fig. 2 that for UAgCu4, cp ass over a maxi maximum when the temperature is raised K from . to 0 5 to 2 K. It reaches 340 mJ/mole K at 1.4 K. elec-From these data we conclude that in UCuq the e ectronic subsystem has similar properties to those of well-established heavy-electron materials like UBei3, Upt3, 8 CeA13, 7 9 or CeCU2Si2. " In all of them a pronounced increase of the c~/T ratio is observed with decreasing temperature below about 10 K. Previous experiments have shown that the heavyelectron state itself may be unstable with respect to phase transitions and superconductivity or magnetic order have been observed. ' " What in our opinion is special about UCuq is the formation of a heavyelectron state, as manifested by the enhancement of the specific heat, in a magnetically ordered material, with the subsequent newly discovered phase transition itself showing unusual features. That the enhancement of the specific heat is not simply a precursor to the quoted phase transition is demonstrated by the behavior of UAgCu4. Because it is known that phase transitions in the heavy-electron state can easily be suppressed by impurities, ' ' we deliberately chose this composition to prevent the phase transition but to keep the large electronic specific heat. Previous and our own specific-heat data reveal that the molar entropy associated with the upper magnetic phase transition is only about 0.6R ln2, considerably less than expected for magnetic ordering that induces a spontaneous splitting of an at least doubly degenerate magnetic 5f-electron ground state. van Daal er al. 3 also found a distinct anomaly in the temperature dependence of the electrical resistivity p that is associated with the magnetic ordering. We repeated and extended these measurements and show our results for p(T) of UCu5 between 0.4 and 30 K in Fig. 3. Our data above 30 K up to room temperature confirm the results of Ref. 3 and therefore we concentrate on the low-temperature part. This upper transition clearly influences the electronic structure giving rise to a local maximum in Io ( T) about 2 K below the transition temperature. Such p(T) curves have been observed for itinerant antiferromagnets with Cr as the most prominent example' and they are ascribed to the formation of gaps on the Fermi surface. ' What is of more interest here, however, is the influence of the lower transition on p(T). The inset in Fig.   3 shows the low-temperature part of p(T) on an expanded temperature scale. After a steady decrease with decreasing temperature p(T) passes through a minimum at 1.6 K with a p value that is roughly 5 times less than the preceding maximum and subsequently smoothly increases by a factor of 7 through the transition. Although some increase in p(T) might be expected considering the appreciable loss of Fermi surface as indicated by c~(T), it does not necessarily have to occur as was recently shown for U2Zn&7, ' for example. The behavior of p(T) of UCu5 below the new phase transition certainly adds more questions as to the nature of the low-temperature phase and will be a topic of future investigations. As expected, we find no minimum in p(T) for UAgCu4, where all other features of p(T) of UCu5 below 300 K are retained, however.
Our results imply the following conclusions. The formation of a heavy-electron state characterized by an increasing enhancement of the electronic specific heat with decreasing temperature is also possible in a magnetically ordered material. This observation in a way supports recent discussions' of the mutual influence of Kondo or Ruderman-Kittel-Kasuya-Yoshida interactions. As in other cases, this state in UCu5 undergoes a phase transition, whose origin remains to be established, but which opens gaps in the excitation spectrum of the heavy quasiparticles. Impurities in the sense of replacing part of the Cu by Ag remove the phase transition, and all parts of the electronic spectrum with high densities of states survive down to low temperatures.
New aspects of this phase transition under these circumstances are its thermal hysteresis without its being discontinuous, and the very high electrical resistivity in the state below the transition.
Up to now the general trend of all heavy-electron systems went towards very low values of p(T) for T approaching 0 K, irrespective of the nature of the finally adopted ground state. It is clear that further experiments, especially involving microscopic methods, will have to clarify the situation.
We should like to thank T. M. Rice for helpful discussions and suggestions. The work in Switzerland was supported by the Schweizerische Nationalfonds.