The frequency axes of all NMR spectra presented ͑Figures 1-5, 7͒ were inadvertently reversed; the frequencies labeled as positive are actually negative, and those labeled negative are positive. ͑Note that zero frequency in these figures does not generally correspond to the D 2 O reference frequency.͒ The frequency shifts of deuterium in ZrNiD 3.0 reported in Sec. III A also have the signs reversed. The correct shifts ͑in ppm relative to D 2 O) are: Ϫ10.1 (Zr 3 Ni site͒ and ϩ9.5 (Zr 3 Ni 2 site͒. The correct shift of the ZrNiD 1.0 impurity phase discussed in Sec. III B is ϩ2 ppm ͑not ϩ7 ppm).These corrections only affect the comparison of the ZrNiD 1.0 impurity phase spectrum to that of ZrNiD 3.0 ͑␥ phase͒. The shift of the impurity phase does not correspond to the shift of the stronger ␥ phase resonance as stated in Sec. III B; instead, the impurity phase resonance falls between the two ␥ phase resonances, and agrees within uncertainty with their weighted average. The ␥ phase resonances correspond roughly to ϩ0.35 and Ϫ0.25 on the frequency axis of Fig. 4 ͑not ϩ0.1 and Ϫ0.5). Based on these corrected frequency shifts, we would still have speculated that the impurity phase is most likely ␥ phase. Thus, the corrections detailed above do not affect any of our conclusions.

Both static and magic-angle spinning ͑MAS͒ 2 H nuclear magnetic resonance spectroscopy indicate that two inequivalent sites are occupied in ZrNiD 3.0 , in a 2:1 ratio, in agreement with previous work. The sites occupied in the trihydride phase have been previously identified as Zr 3 Ni and Zr 3 Ni 2 . In ZrNiD 1.0 , two well-resolved lines of equal intensity are observed in the MAS spectrum at temperaturesϽ220 K, indicating that two other inequivalent sites are occupied in a 1:1 ratio, in contrast with previous reports that only one type of site (Zr 4 Ni 2 ) is occupied in the monohydride ͑␤͒ phase at room temperature and above. The temperature dependences of both MAS and static ZrNiD 1.0 spectra indicate that no phase transitions occur over the entire temperature range studied, 160-500 K. The deuterium hopping rate in ZrNiD 1.0 is determined from characteristic changes in the MAS spectra as a function of temperature; the motion is consistent with an activation energy E a Х0.44 eV assuming a rate prefactor 1/ 0 ϳ10 13 s Ϫ1 . In ZrNiD 3.0 , the rate of deuterium motion is determined from signatures of motion in the static and MAS spectra. The motion in ZrNiD 3.0 is consistent with an activation energy E a Х0.62 eV assuming a rate prefactor of ϳ10 13 s Ϫ1 .