The hydrogen-doped stage 1 KHg-GIC's described here were prepared in a two-step process, analogous with the preparation of C8KHx.[10] First, three types of stage 1 KHg-GIC's were prepared by the usual two-zone method. Two batches were prepared isothermally at temperatures of 200 and 260°C, respectively, and one was prepared at 200°C with a temperature difference of 4°C. As was described previously,[7] the two batches of GIC's which were prepared at 200°C showed one repeat distance, Ic, of (10.24 ± 0.03)Å in (00l) x-ray scans, while those prepared at 260°C showed two sets of Bragg peaks, one corresponding to Ic= (10.22 ± 0.03)Å, and one to Ic= (10.83 ± 0.03)Å.
In order to perform the hydrogen doping, the same samples that were characterized without hydrogen were transferred under vacuum to an ampoule containing 200 mbar of highly purified hydrogen gas. The two isothermally prepared GIC's, which were initially pink, became blue after about five minutes of exposure, and then turned a dark purple. The GIC which was prepared with a temperature difference was initially gold, but remained blue-violet indefinitely afterward. The superconducting transitions of all three types of samples were measured inductively (see Fig. 1), while the temperature was monitored by observation of the 3He vapor pressure. Tc was determined from the intersection of a tangent drawn to the transition curve with the level upper portion of the trace, and delta Tc was determined by measuring the difference in temperature between the points where the tangent intersected the upper and lower levels of the traces. Subsequent (00l) scans showed that the samples' repeat distances were unchanged upon hydrogen doping, but the effect on the superconductivity was dramatic.
The superconducting transitions of the three types of samples before and after hydrogen doping are shown in Fig. 1. The addition of hydrogen is seen to have produced two effects. One is a marked narrowing of the transition width delta Tc which occurred in all samples measured, especially in the one intercalated at 260°C in which delta Tc went from 0.25 K to 0.03 K (Fig. 1c). The other notable effect was the increase of the Tc of all the samples to 1.5 K, except for the GIC prepared at 200°C, whose initial Tc of 1.53 K was unchanged by hydrogen uptake (Fig. 1b). Tc increased eighty percent in the compound prepared with a temperature difference, rising 0.66 K (Fig. 1a). The most significant result overall, though, is that the transitions of all three samples after hydrogen addition strongly resemble one another, whereas before they were quite distinct.
Previous work has correlated the existence of a narrow transition in stage 1 KHg-GIC samples with the presence of a well-ordered in-plane structure, probably a (2 × 2) R0° phase.[6,7] The initially broad transition of the GIC with two repeat distances is then connected with the presence of a second in-plane phase, possibly a (cube-root × cube-root) R30° or a (cube-root × 2) R(30°,0°) structure,[7] or a disordered structure. Experiments with the application of external pressure to KHg-GIC's have produced a sharpening of the superconducting transition[5] apparently similar to that produced by hydrogen doping. The effect of pressure seems to be the formation of a more homogeneous intercalate structure. The transition-narrowing seen in association with hydrogen exposure must also be due to greater uniformity, probably caused by hydrogen diffusion into mercury vacancies associated with disorder. Rapid diffusion of hydrogen atoms is very plausible because of their small size, and is well known to occur in other metallic systems.[12] The total hydrogen uptake may then be limited by the number of available vacant Hg sites, or it may be determined by electronic effects, as in KHx-GIC's, where the maximum possible value of x=0.8 is limited by the finite number of electrons available for transfer to the hydrogen.[13]
Charge transfer to hydrogen is also to be expected in doped stage 1 KHg-GIC samples because of hydrogen's high electron affinity. The filling of low-lying hydrogen states must remove electrons from intercalate and graphite bands and thus lower the Fermi energy. In the KH-GIC's, the transfer of electrons from potassium to hydrogenic states has already been directly observed using ESR[14] and 13C NMR.[15] Moreover, recent Shubnikov-de Haas measurements on these compounds show them to have a lower Fermi energy than the parent compound, C8K.[13] To see how the change in electronic occupation might affect the transition temperature of the KHg-GIC's, we can refer to the BCS theory of superconductivity, in which the superconducting transition temperature increases with increasing density of states at the Fermi level according to Tc =1.14 thetaD exp [- 1/N(Ef)V]. Here thetaD is the Debye temperature, N(Ef) is the density of states at the Fermi level, and V is the electron-phonon coupling matrix element. We see that if the rise in the transition temperature with the uptake of hydrogen is due to a change in the density of states, then the experimentally observed decrease in Ef must lead to an increase in N(Ef). The effect of hydrogen on the three types of samples can then be explained using the schematic density of states curves for the normal state compound shown in Fig. 2.
Figure 2a displays the form proposed for the density of
states of a KHg-GIC with Tc less than 1.5 K. There
are small contributions from the graphite bonding pi and
antibonding pi* bands, but the density of states
is dominated by s-like and p-like bands due to potassium and
mercury. Note that the Fermi level is above a local maximum
in the schematic density of states of the intercalate-derived
bands. The basic shape of the density of states is consistent
with recent EELS experiments which found strong evidence for
the presence of mercury-like and potassium-like contributions
to N(Ef).[16] Figure
2b shows the final configuration for a hydrogenated stage 1
KHg-GIC. Now the Fermi level is lower because of electron
transfer from the intercalate and graphite bands into the
hydrogen states, but the density of states at the Fermi level
is higher. A GIC whose Tc was unchanged from 1.5 K
by hydrogenation must have had its N(Ef) near that
indicated in Fig. 2b before doping. Therefore one concludes
that either its hydrogen absorption and associated Fermi
level shift must have been small, or that its Fermi level
lies near a broad local maximum in the density of states. Low
hydrogen uptake for samples with an undoped Tc of
1.5 K is consistent with the observation that their color
change after hydrogenation was less noticeable than that of
the GIC's with a lower undoped Tc.