Measurements are reported of the temperature dependence of the proton spin-lattice and spin-spin relaxation times $T_1$ and $T_2$ in yttrium and lanthanum dihydrides containing controlled levels of gadolinium as low as 50 ppm. The results demonstrate unambiguously that paramagnetic ions in concentrations so low as to have heretofore been regarded as insignificant have marked effects on the magnitude, frequency dependence, and temperature dependence of $T_1$ and to a lesser extent on $T_2$, and on the electronic structure and hydrogen diffusion parameters derived therefrom. The ${\mathrm{Gd}}^{3+}$ ion contributes an additional spin-lattice relaxation rate $T_{1p}$, which in these hydrides arises entirely from the dipolar coupling between impurity and proton moments. Proton magnetization is transported to the relaxation centers by spin diffusion at low temperatures and by hydrogen-atom diffusion at intermediate and high temperatures. The rate $R_{1p}$ is directly proportional to Gd-ion concentration at both low and high temperatures, but in the atom diffusion regime $R_{1p}$ is 20-25 times greater than for spin diffusion. The impurity-induced relaxation is shown to have profound effects on the apparent nuclear-nuclear dipolar relaxation rate $R_{1d}$ associated with hydrogen diffusion. At impurity levels as low as 10 ppm Gd, a secondary minimum appears in the temperature dependence of $T_1$ which may be readily misinterpreted in terms of a second motional process with lower activation energy. Even lower impurity levels yield a characteristic "slope-change" effect, which may be construed as indicating a change in the activation energy for hydrogen diffusion. At low temperatures $R_{1p}$ interferes with the determination of the conduction-electron contribution $R_{1e}$ and the Korringa product $T_{1e}T$. Separation of $R_{1e}$ and $R_{1p}$ is complicated by the fact the $R_{1p}$ is not temperature independent as has typically been assumed. Methods of achieving this separation are discussed, and it is shown experimentally that this difficulty can be circumvented by replacing the major part of the hydrogen with deuterium, thereby inhibiting spin diffusion. Measurement of $T_1$ as a function of resonance frequency and of $T_2$ can also be of value in separating the various sources of relaxation.

• Received 6 June 1983

DOI:https://doi.org/10.1103/PhysRevB.28.6227