´ ´
L. Guenee et al. / Journal of Alloys and Compounds 348 (2003) 129–137
130
˚
losses magnesium was first pre-reacted with nickel to the
binary metal compound Mg2Ni, and pellets of powder
mixtures of La, Mg2Ni and Ni were then melted at various
nominal compositions. The resulting weight losses did not
exceed 2%. As-cast samples of nominal composition
La1.08MgNi4 consisted of a majority phase whose cubic
parameters a 5 7.18 and 7.17 A. However, the quality of
the data was insufficient to derive reliable Mg/Ni ratios for
either one of the phases and thus prove their different
chemical compositions. For the neodymium sample,
˚
synchrotron data were collected at l 5 0.50012 A within
the angular range 2u 5 3.028–33.268 in steps of 0.0048.
The sample contained two phases of which the principal
one could be attributed to a NdNi4Mg phase crystallising
with the MgCu4Sn type structure and the minor one to a
NdNi3 phase crystallising with the PuNi3 type structure.
The structures were refined by using the program Fullprof
˚
cell parameter (a 5 7.17 A) was within the previously
reported range for the C15 type solid solution
La1 2 xMgxNi2. However, the presence of 200, 420, 600
and 640 diffraction planes in the X-ray patterns were not
¯
consistent with this structure type (space group Fd3m,
h00: h 5 4n and hk0: h 1 k 5 4n) and rather indicated a
MgCu4Sn type structure (space group F43m). Further-
more, the samples contained two minority phases, one
[7] and by varying 21 parameters (NdNi4Mg: 6
¯
structural15 profile1scale factor; PuNi3: 9 structural, 1
scale). For NdNi4Mg a small substitution of Ni by Mg (site
4c) allowed to improve the agreement indices, in particular
RBragg. The patterns are represented in Fig. 1 and refine-
ment results are summarised in Table 1.
having the hexagonal La2Ni7 type structure (a55.038
˚
c524.35 A) and an unidentified phase that could be
indexed on an orthorhombic C-centred lattice with cell
˚
parameters a54.21, b510.27 and c58.34 A. In order to
detect a possible deviation of the majority phase from the
ideal metal ratio Ni/Mg54 samples of composition
LaNi3Mg2 (Ni/Mg,4) and LaNi4.5Mg0.5 (Ni/Mg.4)
were prepared. The former was found to consist mainly of
the cubic LaNi4Mg phase and showed diffraction line
broadening and splitting. The concentration of the non-
identified orthorhombic phase was higher than in the
La1.08MgNi4 sample. The LaNi4.5Mg0.5 sample contained
LaNi4Mg, LaNi5 (CaCu5 type structure) and elemental Ni.
None of the lanthanum alloys were annealed. For the
neodymium compound compressed powder mixtures of Nd
2.3. Hydrogen absorption and desorption
Due to the relatively large concentration of secondary
phases in the lanthanum samples the hydrogen absorption
properties were mainly studied on the neodymium sam-
ples. One of the latter ([1, nominal composition NdNi4Mg
annealed at 800 8C for 24 h) was powdered in a glove box
and introduced into a microbalance (Hiden Analytical) that
was evacuated for a few hours under preliminary vacuum
(1023 bar). Hydrogenation was carried out under a pres-
sure of 7 bars at 50 8C for 24 h. After a few minutes the
sample started to absorb slowly but steadily as shown in
Fig. 2. After about 3 h the weight increase levelled off at a
value corresponding to about 3.6 H/f.u. (value corrected
for the presence of |26 wt.% hydrogen inert NdNi3 phase,
see ch. 2.4). A pressure–composition-isotherm ( p–c–T)
obtained under the same conditions on part of another
sample ([3, nominal composition Nd1.038Mg1.087Ni4, an-
nealed at 780 8C for 28 h, 18% secondary NdNi3 phase) is
shown in Fig. 3. The NdNi4Mg phase absorbs up to 4
H/f.u. (value corrected for the presence of hydrogen inert
NdNi3 phase). One notices a relatively narrow (|2 H/f.u)
and ill-defined plateau region at p(H2)|1bar, followed by
a relatively wide (2H/u.f.) single-phase domain at higher
pressures. Thus the hydrogenation properties of NdNi4Mg
are rather different from those of LaNi5 type hydrides for
which long plateaus and narrow single phase domains were
found [8]. The absorption/desorption kinetics is relatively
slow, thus precluding p–c–T measurements. After the
experiment, the sample was investigated by X-ray diffrac-
tion in air, and then replaced into the thermobalance and
evacuated at 80 8C for a few hours. The patterns indicated
almost complete loss of hydrogen and recovery of the
initial alloy structure. Interestingly, after the hydrogenated
samples were exposed to air in a glass container, water
droplets appeared after a few hours on the surface of the
container walls. No such effects occurred if the hydride
was kept in the glove box. Finally, one sample ([2,
¨
(Riedel–de Haen, .99.9%), Mg2Ni and Ni of various
nominal compositions were melted in an induction furnace
and annealed for up to 33 h at 780–800 8C in quartz tubes
filled by 0.3 bar argon (6N). The weight losses did not
exceed 3%. All samples contained a majority phase having
the MgCu4Sn type structure and a minority phase (up to
25%) of binary NdNi3 having the PuNi3 type structure [6].
Samples of nominal composition NdNi4Mg ([1),
Nd1.008Ni4Mg ([2) and Nd1.038Ni4Mg1.087 ([3) were
used for hydrogen absorption, synchrotron diffraction and
pressure-composition–isotherm/neutron diffraction mea-
surements, respectively.
2.2. Structure analysis of alloys
The alloy samples LaNi3Mg2 and NdNi4Mg ([2) were
investigated by synchrotron powder diffraction (ESRF,
Grenoble, SNBL, capillaries of 0.3 mm diameter). For the
˚
lanthanum sample, the pattern (l 5 0.85022 A) confirmed
the presence of two phases, one having the MgCu4Sn type
structure and thus the likely composition LaNi4Mg (La, Ni
and Mg on Mg, Cu and Sn sites, respectively), and the
other a hitherto unknown orthorhombic C-centred struc-
ture. The diffraction lines of the LaNi4Mg phase were split
and had to be modelled during structure refinement by the
assumption of two MgCu4Sn type phases having the cell