Results

The unanticipated outcome of the T_c measurements was that no superconductivity could be detected in either alpha-phase or mixed alpha + beta-phase samples.[36] By using Equation , it was determined that the superconducting effective area that would give a signal-to-noise ratio of one corresponded to about 2% of the area of the CsBi polycrystals. Therefore, from the inductive measurements it can be said with some confidence that no more than 2% of these samples is superconducting within the temperature range from 4.2 to 0.5 K.

The temperature region above liquid helium temperature was only searched briefly for a transition at a T_c higher than that reported by Lagrange et al. It is safe to conclude that a higher-temperature transition is not present, though, since an applied magnetic field of 23 Tesla did not produce any kind of superconducting transition. (The highest critical field known is well under 1 T.[120])

In addition to the null result from the inductive measurements, no interesting features in the resistivity were detected at low temperature. The compounds appeared to exhibit typical metallic behavior ( drho/dT ;SPM_gt; 0). The measured resistivity at 4.2 K was (8 ± 2) muOmega-cm in one stage 1 alpha-phase sample, and (30 ± 7) muOmega-cm in a stage 1u alpha + beta sample. These values are somewhat higher than the 5 K resistivity values quoted by Lagrange et al., who measured their resisitivity values inductively.[168] An applied magnetic field produced only traces consistent with magnetoresistive behavior, nothing like a superconducting-normal transition. [See discussion below.]

The MIT superconductivity measurements are obviously in conflict with those described previously[146], as are those of several other groups who also failed to find superconductivity.[270,223] In the most extensive set of measurements to date, Stang and coworkers in Berlin failed to find superconductivity in stage 1 alpha-phase CsBi-GIC's down to 50 mK even though they had sensitivity to a transition in about 0.1% of the sample volume.[223] Even the Nancy group has had difficulty in reproducing the T_c values they reported.[144] A summary of all T_c measurements on the MBi-GIC's to date is reported in Table .

  
Table: Superconducting transition temperatures of the MBi-GIC's and related alloys. For Ref.[36], the volume fraction column actually contains the effective areal fraction of superconductivity, as explained in Section . NR=Not reported. NA=Not applicable.

The irreproducibility of the T_c measurements in the CsBi-GIC's would be more surprising were it not for the wide range of transition temperatures found for other superconducting GIC's.[64] In C₄KHg, as was discussed previously, fairly subtle variations in intercalation conditions have a large effect on superconductivity. The wide range of T_c values reported for C₄KHg is thought to be due to the presence of the beta phase [see Section ] in low- T_c samples. One possible explanation for the lack of superconductivity in the non-superconducting CsBi-GIC's is therefore that they possess minority phases which suppress superconductivity. Alternatively, as Lagrange has suggested,[145] it seems plausible that the differences found in the superconducting behavior in the CsBi experiments could be attributable to small differences in in-plane ordering or in stoichiometry. This type of sensitivity to ordering and/or stoichiometry has also been considered as an explanation for the wide range of T_c values reported for C₄KHg.[206]

On the other hand, there is another simple line of reasoning which explains the discrepancy among the various sets of experiments. The alternative explanation, also discussed by Stang and coworkers,[223] is that the superconducting transitions reported by Lagrange et al. were actually transitions of inclusions of the alloy CsBi₂. This possibility is suggested by the presence of inclusions in the CsBi-GIC's which could not be completely removed by repeated cleaving. The inclusions were imaged during TEM studies which showed that they appeared throughout the samples prepared both here at MIT and at the University of Kentucky.[222] An electron micrograph of an inclusion is shown in Figure .

  
Figure: An electron micrograph showing intercalant inclusions (bright regions) in a C₄CsBi_x alpha + beta-phase polycrystal grown here at MIT. The magnification for this micrograph is indicated by the 100 nm scale bar. [Micrograph prepared by J. Speck, MIT.]

In the case of the MIT samples, the inclusions presumably had the approximate composition of the starting alloy, Cs₅Bi₄, a composition which is not superconducting. As would be expected, therefore, no inclusion-derived superconducting signal was seen in the MIT CsBi-GIC's. However, as shown by the phase diagram in Figure , Cs₅Bi₄ is adjacent to CsBi₂, a phase which is superconducting. The implication is that a starting alloy with a composition between Cs₅Bi₄ and CsBi₂ (such as CsBi) should be a mixture of superconducting and non-superconducting material. One would therefore expect that a GIC made from such a starting alloy would have some superconducting inclusions. Inclusion-derived superconductivity was apparently seen by Stang et al..[223] They observed that superconductivity was present only in small volume fractions in the CsBi-GIC's, consistent with inclusion-derived transitions. Further, Stang and coworkers found that T_c was about 4.7 K almost independent of stage and phase. [See Table .] The only processing variable which seemed to impact T_c was the starting alloy composition, which presumably determines what fraction of the inclusions are superconducting. A plot of T_c versus starting alloy composition is displayed in Figure .

  
Figure: A plot of superconducting transition temperature T_c for C₄CsBi_x versus starting alloy Bi/Cs ratio. bigotimes, MIT data from Ref. [36]; bigtriangleup, University of Kentucky data from Ref. [270]; and bigcirc, Freie Universität Berlin data from Ref. [223]. The X are alloy (not GIC) data from the CRC Handbook. The presence of downarrow means that the nearest point represents an upper bound on T_c. Data from the University of Nancy is not included because precise starting alloy compositions are not given for their samples.

The implication of the MIT and Berlin work[223] is that superconductivity in contact-intercalated GIC's can be mimicked by inclusions of a superconducting intercalant phase. In this interpretation, the superconductivity observed by Lagrange and colleagues in resistivity measurements was due to a network of CsBi₂ through the sample, either adsorbed on the surface or interconnected through the bulk. (Bulk interconnections could perhaps be provided by Daumas-Herold domains.) This superconducting alloy would completely short out any normal-state transport due to the GIC itself. In this interpretation, the University of Nancy low-temperature resistivity measurements do not measure a property intrinsic to the GIC, but to the adsorbed alloy. Then the reason that the MIT measurements give a higher value of rho_a at 4.2 K than the Nancy measurements is that the MIT measurements give rho_{a, GIC} while the Nancy ones give rho_{a, alloy}. Interpretation of these results would be more certain if it were known exactly what starting alloy compositions the French used and whether they cleaved their samples before performing the transport measurements.

In the end, one cannot make a firm statement as to whether the CsBi-GIC's are superconducting or not. Circumstantial evidence strongly indicates that the answer is no, but considering the long history of variability in the T_c's of GIC's, it is wise to be cautious. After the discovery of superconductivity in C₈K at 0.55 K in 1965, a transition was not observed again until 1978 at a much lower temperature of 0.139 K,[140] with several unsuccessful attempts to reproduce the original experiment in intervening years.[196] Considering that it is not possible to tell whether the non-superconducting samples made outside Nancy are similar in every detail to those the superconductors synthesized there, an authoritative statement ruling out superconductivity in MBi-GIC's cannot be made. This is especially true for the KBi-and RbBi-GIC's, which have not yet been intensively studied outside Nancy. The best one can do at this time is say that available evidence does not support the identification of the MBi-GIC's as superconductors. A possible explanation why these compounds are not superconducting is discussed in Section .