In order to understand clearly the origins of changes in hydrogenation characteristics resulting from hydrogen absorption and desorption, microstructural changes of $LaNi_5$ have been investigated using analytical electron microscopy (AEM) and high resolution electron microscopy (HREM). TEM specimens of $LaNi_5$ powders can be prepared reproducibly in a straight-forward manner without any damages on the surface layers of particles.
Decomposed layer of 50nm in thickness is found to form evenly on the surfaces of $LaNi_5$ particles which have been hydrogenated to H/M = 1.0 and degassed to $10^{-2}$ torr at 23℃. It is confirmed by indexing electron diffraction pattern and energy dispersive x-ray spectroscopy (EDXS) spectrum that surface decomposition has taken place by hydrogen with final nanocrystalline products of metallic Ni and $LaH_3$. In general, there are some strains at interface between surface decomposition layer and sub-surface and they are accommodated by twins.
Heat and volume expansion accompanying hydride formation in the bulk are not essential conditions for surface decomposition. When the hydrogen concentration of an absorption - desorption cycle has been reduced to H/M = 0.017, the same decomposition layer was still observed. It is considered that this preferential decomposition behavior at the surface of $LaNi_5$ is due to the physical properties of surface.
The nucleation and growth of hydride in $LaNi_5$ has been studied by comparing microstructural changes before and after a cycle of hydrogen absorption and desorption. Prior to hydrogenation reaction, the $LaNi_5$ alloy shows many stacking faults and it has a typical dislocation structure of well-annealed alloys. After a cycle of absorption and desorption with hydrogen concentration up to H/M = 0.017, oval type dislocation loops of 100-200nm in length are observed in the stacking faults and at grain boundaries. They are aligned with a constant separation oriented to the [0001] direction. These loops reveal that hydride's sub-unit were already nucleated. When hydrogen concentration was increased up to H/M = 0.17, dislocation loops were not observed any more. Instead, super-dislocation pairs with line vector U = [0001] were aligned in the form of fence. It explains that there were largely grown hydrides. And the isolated and randomly distributed traces of hydrides show that the assumption of continuous moving boundary is not correct in kinetic study. By comparing the condition of g.b = 0 and line vector U, these dislocations are found to be identical with the loop dislocations. It is suggested that hydride's sub-units are precipitated as a form of thin plate or needle type along the $<2 \bar{1} \bar{1} 0>$ direction and the sub-units grew to a large hydride through coalescence process.
Even one cycle of hydrogen solid solution formation-decomposition induces marked increase in dislocation density. To certain distance from the surface of $LaNi_5$ particle, saw blade-like dislocation arrays are found to form along the specific direction and the leading edge of each blade is confirmed as pure screw component. This morphology of dislocations is considered to reflect the path along which the hydrogen is introduced. The Burgers vector of these dislocations is parallel to one of the diagonal directions of $LaNi_5$ unit cell. These dislocations have high mobility and can cross slip easily.
After a cycle of absorption and desorption with hydrogen concentration up to it's maximum capacity, two kinds of slip bands which are composed of misfit dislocations with Burgers vector $<2 \bar{1} \bar{1} 0>$ and $1/3 <0 \bar{1} \bar{1} 0>$ are aligned with constant sperations oriented to specific directions respectively. The saw blade like dislocations also interact each other to form well-estabilished dislocation walls aligned with a constant separation oriented specific direction. The anisotropic peak broadening of pattern of $LaNi_5$ after activation is originated from these arrangements of dislocations. With increasing the number of absorption-desorption cycles, the overall density of dislocation is not changed but the slip bands are fractured into fine fragments. It can be interpreted as the result of refinement and random orientation of hydride grains due to the saturated defects in the lattice of $LaNi_5$ resulting from previous cycles.
Decomposed layer of more than 400nm in thickness is found to form on the surface of $LaNi_5$ particle which have been hydrogenated under 50 atm. Hydrogen pressure at 120℃ for 144 hrs. And in the bulk of same particle, newly reordered structure is observed. This new structure is modulated three times in a axis and two times in c axis of original unit cell but furture studies are required to characterize it in detail. This different behavior of microstructural change between surface and bulk is supposed to be the main mechanism of intrinsic degradation of $LaNi_5$ alloy.
