Summary of New Results

The primary focus of this thesis has been to understand superconductivity in the KHg-GIC's. There are several aspects to the problem. One is to understand the stage-dependence of T_c. As discussed in Section and , the increase of T_c with increasing stage can be viewed as evidence of an important role for graphitic electrons in GIC superconductivity. If graphitic electrons also participate in superconductivity, then the apparent increase of T_c with increasing stage in the KHg-GIC's is not so surprising.

The most unusual aspect of GIC superconductivity is the anisotropy of the upper critical field, H_c2. This anisotropy can also be understood in terms of participation by graphitic electrons in the superconducting state.[4] H_c2(theta, T) data for C₄KHg have been interpreted in terms of the anisotropic Ginzburg-Landau model (AGL).[127,175,244] Quantitative agreement was found for gold specimens with T_c about 0.8 K, in accord with the reports of Iye and Tanuma.[120]

For pink specimens with T_c about 1.5 K, corrections to the AGL model had to be included in order to achieve a good fit. The AGL H_c2(theta) formula agreed well with the pink C₄KHg data once allowance was made for type I superconductivity. Type I superconductivity occurs in a small temperature-dependent range of applied-field orientations near vecH || ^c. The occurrence of type I superconductivity in pink C₄KHg is in accord with calculations based on the specific heat data of Alexander et al.[8] The values of the thermodynamic critical field H_c obtained from fits to the H_c2(theta) data imply a smaller density of states at E_F, N(0), than that measured by Alexander.[8] This discrepancy concerning the magnitude of N(0) could simply be due to demagnetization effects on the H_c2(theta) measurements, which could reduce the apparent value of H_c. On the other hand, if the magnitude of N(0) obtained from the angular dependence data is correct, this smaller value could help to explain why T_c is lower in C₄KHg than in C₈KHg.

Extended linearity of H_c2(T) was also observed in C₄KHg. The temperature dependence of H_c2 in C₄KHg (and in other superconducting GIC's[141,120]) is reminiscent of critical field data on the layered TMDC's.[114,51] Because of the evidence that both intercalant and graphitic electrons participate in GIC superconductivity, multiband models of H_c2 would appear to be appropriate.[5,80] The nearly cylindrical shape of the graphitic piece of a GIC's Fermi surface[67] suggests that anisotropic Fermi-surface models might also be good candidates to describe H_c2 of superconducting GIC's.[33,274] More quantitative information about the Fermi surface of the GIC's is necessary before these models can be tested, although some calculations are available.[112]

Pink and gold C₄KHg specimens were found to differ in their superconducting transition temperature and critical field behavior, as mentioned above. The only detectable distinction in the normal-state properties of these compounds is that the gold specimens often (and perhaps always) contain some of the higher- I_c beta phase. The pink samples, on the other hand, are composed solely of the lower- I_c alpha phase. The relative amounts of the two phases can be controlled, within limits, by fine-tuning the intercalation conditions.

The depression of T_c by a small fraction of the minority beta phase is unusual. In general, only magnetic impurities are so potent in lowering T_c.[252] Surprisingly, adding a small amount of hydrogen or applying a minute hydrostatic pressure[55] restores T_c to about 1.5 K. A T_c enhancement by hydrogenation and applied pressure is also seen in several of the transition metal dichalcogenides.[90] These TMDC's support charge-density waves,[265] so the increase of T_c with the application of a small perturbation is attributed to the suppression of the CDW state.[90] The suppression of a CDW by hydrogen and applied pressure also seems reasonable in C₄KHg.[55] The similarity of the Fermi surfaces of the TMDC's and superconducting GIC's lends additional credence to the CDW hypothesis.[115]

The highly variable range of T_c in C₄KHg is matched by the CsBi-GIC's. For C₄CsBi_0.5, T_c has been reported to be as high as 4 K,[168] while other investigators[36,223] have found no superconductivity down to 50 mK. Extensive efforts to reproduce the higher T_c value have been unsuccessful.[144]

The apparent lack of superconductivity in the CsBi-GIC's is puzzling considering that the alloy CsBi₂ is superconducting with T_c = 4.75 K.[168] The high resistivity anisotropy of the MBi-GIC's (M = a heavy alkali metal) may offer an important clue. McRae et al.[166] note that rho_c / rho_a is much higher in the MBi-GIC's than in other donor compounds. They interpret their MBi-GIC c-axis resistivity data in terms of the hopping conduction mechanism proposed by Sugihara.[229,230] Perhaps c-axis band conduction in s-like intercalant states is necessary for the existence of superconductivity in GIC's. Similar ideas have been discussed by Al-Jishi.[4] In light of the hypothetical CDW state in C₄KHg, the possibility of a CDW should also be taken seriously for the MBi-GIC's.

These considerations may also apply to the acceptor GIC's, which also tend to have hopping conduction along the c-axis.[166] The acceptor GIC's are discussed in more detail below.