A nuclear magnetic resonance probe of group IV clathrates

Abstract

The clathrates feature large cages of silicon, germanium, or tin, with guest atoms in the cage centers. The group IV clathrates are interesting because of their thermoelectric efficiency, and their glasslike thermal conductivity at low temperatures. Clathrates show a variety of properties, and the motion of cage center atoms is not well understood. In Sr8Ga16Ge30, we found that the slow atomic motion in the order 10-5 s is present in this system, which is much slower than what would be expected for standard atomic dynamics. NMR studies of Sr8Ga16Ge30 showed that Knight shift and T1 results are consistent with low density metallic behavior. The lineshapes exhibit changes consistent with motional narrowing at low temperatures, and this indicates unusually slow hopping rates. To further investigate this behavior, we made a series of measurements using the Carr-Purcell-Meiboom-Gill NMR sequence. Fitting the results to a hopping model yielded an activation energy of 4.6 K. We can understand all of our observations in terms of non-resonant atomic tunneling between asymmetric sites within the cages, in the presence of disorder. For Ba8Ga16Ge30, the relaxation behavior (T1) deviates from the Korringa relation, and the Knight shift and linewidth change with temperature. Those results could be explained by carrier freezout, and the development of a dilute set of magnetic moments due to these localized carriers. For Ba8Ga16Ge30 samples made from Ga flux, we observed different T1 and Knight shift behavior as compared to n type material. This is due to the differences in carrier type among these different samples. The p type sample has a smaller Knight shift and a slower relaxation rate than n type samples made with the stoichiometric ratio, which is consistent with a change in orbital symmetry between the conduction and valence bands. WDS study for Ba8Al10Ge36 showed the existence of vacancies in the Al-deficient samples, which results in some degree of ordering of Al occupation on the framework sites. In Al NMR measurements on Ba8AlxGe40-x with x = 12 to 16, we found that T1 of all Al samples follows the Korringa relation. The broadening of the single NMR central peak of Ba8Al16Ge30 is due to the inhomogeneous Knight shifts for occupation of different framework sites. For Ba8Al12Ge34 and Ba8Al13Ge33, we observed two peaks, and NMR results show that they are from distinct Al sites, while for each peak, the inhomogeneous broadening is much smaller. The difference in lineshapes we attributed to the existence of vacancies which we detected in the Al-deficient materials, and we assign one of the two Al peaks to Al adjacent to a vacancy.