The comment from Butch et al. gives the view of physicists who have carefully studied the effects of hydrostatic pressure (P) conditions on a specific crystal. In our article, we do not discuss this problem but we concentrate on the single crystal properties. The main reason is that it is a delicate point as pressure experiments are usually conducted with a constant volume in a clamp cell. For example, the combined effects of the solidification of the pressure medium and even feedback at the phase transition [for example here, hidden order (HO)] when the material may expand or contract in different directions can cause a departure from ideal hydrostatic conditions. As an observed experimental rule, the best approach to hydrostaticity is given when the crystal size is small in compared with the volume of the pressure cell, a condition which is never realized in pressure inelastic neutron scattering experiments. A key point in a detailed study of the pressure conditions to succeed is to realize comparative measurements in the same pressure cell as used in neutron scattering, as well as in situ pressure measurements. Such condition was realized in neutron scattering experiments, where a strain gauge was glued on the crystal face to trace where macroscopic phase transition occurs, and even to estimate pressure inhomogeneity. The work presented in our article was realized to characterize the sample quality at most and to be able to select samples for quantum oscillation experiments (Shubnikov–de Haas (SdH) and de Haas–van Alphen effect) to precisely determine the feedback between HO and antiferromagnetic (AF) properties on the Fermi Surface. The residual resistivity ratio (RRR) for URu2Si2 is a simple and an effective criterion for selecting the best material to enhance the quantum oscillation signal, even if RRR is estimated from the value at 2K and validity of Matthiesen’s rule can be discussed. In our experience, for example, no SdH signal was detected for samples with low RRR of below 50. It should be noted that complementary information is given by specific heat measurement. Because of the large extinction observed in the neutron diffraction pattern of crystals, the determination of an absolute value of the extrinsic tiny sublattice magnetization via the intensity of ‘‘residual’’ AF reflections is a difficult and delicate task. Electron microscopy up to now has not been sensitive enough in the characterization of strongly correlated uranium intermetallic compounds, as shown by contradictory results on UPt3 for example. 7,8) As suggested in our paper, electronic correlation may generate a new type of extended defects. In the article by Butch et al., the neutron diffraction experiments were realized under isobar conditions where the pressure in the pressure cell was estimated from the pressure applied through a capillary. At least for us, it is not obvious whether the isobar conditions are realized on the crystal as no direct measurement of the pressure is achieved on the crystal’s location in the pressure cell at low temperatures. Furthermore, the resistivity measurement was realized independently in a classical piston cylinder cell in a commercial cryostat with a conventional medium of a 1 : 1 volume mixture of n-pentane/isoamyl alcohol. The interesting point is that the disappearance of HO and of superconductivity occurs at Px, as already obtained from ac susceptibility measurements. Our experimental observation of P measurements on different crystals indicates that the collapse of the SC transition never coincides with Px obtained through resistivity measurements and varies from crystal to crystal with no simple relation with the RRR value. This is quite different from the data obtained by Butch et al. which shows a coincidence with Px. Figure 1 shows the resistivity data of (a) the crystal examined by the Butch et al. and (b) a crystal with RRR 1⁄4 270. As can be seen, a nice anomaly at HO is detected in (a), but quite different behavior is detected between the single crystals (a) and (b) at low temperatures T < 10K. It must be noted that in our article the RRR domain near 30 is the range in which the power law dependence seems to reach a minimum, suggesting a correlation between the electronic mean free path and the 1 10 10
Read full abstract