Distributions of hydrogen and strains in LaNi5 and LaNi4.75Sn0.25

Article abstract of DOI:10.1016/S0925-8388(01)01857-6

Hydrogen distributions and internal strains that accompany hydriding of binary LaNi₅ were compared to those of the ternary alloy LaNi_4.75Sn_0.25, which is known to have a cycle life superior to that of LaNi₅ in electrochemical cells and in gas storage applications. X-ray diffractometry shows that the unit cell volume of the hydride phase changes more continuously with hydrogen concentration in LaNi_4.75Sn_0.25 than in binary LaNi₅. Gas-phase isotherms show that the Sn atoms make significant changes to the local chemical potential of hydrogen atoms. Using generic hydrogen-solute interactions in Monte Carlo simulations and physical arguments, it is shown that normal coarsening of hydride zones will be altered, or even arrested, by hydrogen-solute interactions. Small-angle neutron scattering shows that the distribution of deuterium in partially-deuterated LaNi_4.75Sn_0.25 is more homogeneous than in partially-deuterated LaNi₅, at least on the spatial scales around 100 ?. It is suggested that the more homogeneous deuterium distribution in LaNi_4.75Sn_0.25 suppresses the strain gradients that cause decrepitation of the metal hydride.

Full text of DOI:10.1016/S0925-8388(01)01857-6

Journal of Alloys and Compounds 335 (2002) 165–175
L
www.elsevier.com/locate/jallcom
Distributions of hydrogen and strains in LaNi5 and LaNi4.75Sn0.25
a
b
B. Fultza , , C.K. Witham , T.J. Udovic
*
aKeck Laboratory, Engineering and Applied Science, MS 138-78, California Institute of Technology, Pasadena, CA 91125, USA
bNIST Center for Neutron Research, National Institute of Standards and Technology, 100 Bureau Dr., MS 8562, Gaithersburg, MD 20899-8562,
USA
Received 6 July 2001; accepted 31 August 2001
Abstract
Hydrogen distributions and internal strains that accompany hydriding of binary LaNi5 were compared to those of the ternary alloy
LaNi4.75Sn0.25, which is known to have a cycle life superior to that of LaNi5 in electrochemical cells and in gas storage applications.
X-ray diffractometry shows that the unit cell volume of the hydride phase changes more continuously with hydrogen concentration in
LaNi4.75Sn0.25 than in binary LaNi5. Gas-phase isotherms show that the Sn atoms make significant changes to the local chemical potential
of hydrogen atoms. Using generic hydrogen–solute interactions in Monte Carlo simulations and physical arguments, it is shown that
normal coarsening of hydride zones will be altered, or even arrested, by hydrogen–solute interactions. Small-angle neutron scattering
shows that the distribution of deuterium in partially-deuterated LaNi4.75Sn0.25 is more homogeneous than in partially-deuterated LaNi5, at
˚
least on the spatial scales around 100 A. It is suggested that the more homogeneous deuterium distribution in LaNi4.75Sn0.25 suppresses
the strain gradients that cause decrepitation of the metal hydride.
2002 Elsevier Science B.V. All rights reserved.
Keywords: Hydriding; Hydrogen–solute interactions; Nickel–metal hydride
1. Introduction
is unstable in the presence of hydrogen. A disproportiona-
tion reaction:
´
Compared to lead/acid batteries, developed by Plante in
LaNi5H6 → LaH2 1 5Ni 1 2H2
1860, and nickel–cadmium batteries, developed by Edison
and by Junger in 1901, nickel–metal hydride batteries are
new, being proposed around 1970 but becoming practical
only since the 1980s. The market for nickel–metal hydride
batteries now exceeds that of nickel–cadmium, and con-
tinues to grow. The commercial success of nickel–metal
hydride battery systems has occurred so quickly that some
important issues affecting their performance are poorly
understood, in particular the issue of capacity fade, or the
gradual deterioration of the energy that can be delivered by
a cell after it has undergone a few hundred charge/
discharge cycles. It is well known that capacity fade
originates with corrosion of the metal hydride electrode,
LaNi5:
is driven by a free energy of 26 kcal/mol [3–5]. Alloying
the LaNi5 with small amounts of solutes can reduce its free
energy, helping to stabilize it against the corrosion and
disproportionation reactions. Unfortunately, the free ener-
gies of these two reactions are far larger than any such
stabilization by alloying. For example, alloying with 4%
Sn increases the stability of the LaNi5 by about 4 kcal/mol
(using the approach of Miedema [6]). Microstructural
modifications to the alloy, such as control of grain size or
defect density, have even smaller effects on the free
energy. It is therefore impossible to develop a LaNi5-based
electrode that is thermodynamically stable in an alkaline
cell.
The capacity fade of LaNi5-based electrodes is con-
trolled by kinetics, and these kinetics can be varied greatly
by solute substitution for La or Ni. Work by the groups of
Sakai [2,7–10], Willems [1,11], Srinivasan and McBreen
2LaNi5 1 6H2O → 2La(OH)3 1 10Ni 1 3H2
with a free energy change much greater than 100 kcal/
(mol LaNi5) [1,2]. It should also be pointed out that LaNi5
´
[12–14], Percheron-Guegan [15], Schlapbach [16–18], and
Fultz [19–22] have shown clear effects of solute additions
on the cycle life of LaNi5, or its misch-metal counterpart,
MmNi5. For example, some commercial formulations of
*
Corresponding author.
E-mail address: btf@hyperfine.caltech.edu (B. Fultz).
0925-8388/02/$ – see front matter
PII: S0925-8388(01)01857-6
2002 Elsevier Science B.V. All rights reserved.

166
B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
metal hydride electrode materials with good cycle life have
compositions close to Mm1Ni3.5Co0.8Mn0.4Al0.3. The
mechanism by which these solute substitutions affect
corrosion kinetics are, unfortunately, poorly understood.
There is a broad consensus that high surface areas are
detrimental to corrosion resistance [1,2,7,10,23–25]. For
example, Notten et al. [23,26] have demonstrated that the
cycle life of metal hydride electrodes deteriorates when the
metal hydride alloy comprises smaller particles. (They
were careful to use particle sizes smaller than would occur
after hydrogen absorption by a fresh ingot.) Others have
shown a correlation between fracture or oxidation of the
metal alloy and poor cycle life [27–29]. The breakup of
the LaNi5 into smaller particles will provide a larger
surface-to-volume ratio, offering more reaction sites for
corrosion. This breakup or ‘decrepitation’ is often attribu-
ted to the strains in LaNi5 during hydriding, which are
enormous, being 5–7%. The elastic limit of brittle LaNi5 is
well below 1%, and with abundant flaws in the material it
seems intuitive that cracking should occur. Nevertheless,
the ‘beneficial solute additions’ do not suppress the volume
dilatations of hydriding anywhere near what is required to
reduce the hydriding strains below the elastic limit of
LaNi5. Other mechanical properties such as fracture tough-
ness are also altered by alloying [1,2,7,10,30]. The rela-
tionship between hydriding-induced strains and cycle life
is not obvious, and motivated the present study.
The volume dilatation of hydriding cannot affect directly
the decrepitation of the metal hydride because uniform
hydrostatic pressure does not drive crack propagation. If
the volume dilatation of hydriding is large, but occurs
uniformly throughout the material, there can be no release
of elastic energy by crack propagation. There is evidence
that large, but homogeneous, dilatations allow good cycle
life. In a series of experiments, the Philips group has
studied hydrogen absorption in La–Ni–Cu alloys by in-situ
X-ray diffractometry [24,31,32]. They varied the Ni con-
centration to deviate from the AB5 stoichiometry. The
off-stoichiometric alloys were shown to have: (1) sloping
plateaus in absorption isotherms; (2) gradual shifts in the
positions of the diffraction peaks from the a-phase and the
b-phase with hydrogen concentration, and (3) much im-
proved cycle lives when used in battery electrodes. The
Philips group argued that the continuous nature of the
hydriding transformation for the alloy AB5.4 caused the
interface between the a-phase and the b-phase to broaden,
thus reducing the stress intensities at the a/b interface.
The present study was designed to measure both strains
and hydrogen distributions in the alloys LaNi5 and
LaNi4.75Sn0.25. The LaNi4.75Sn0.25 alloy has been shown to
have a superior cycle life to LaNi5 both in gas-phase
reactions [33] and in electrochemical cells [13,19–22,34–
37]. Our X-ray diffractometry of hydrided materials
showed that the overall volume expansions of the two
alloys were approximately the same. The two materials
LaNi5 and LaNi4.75Sn0.25 had differences in unit cell
volumes between fully-hydrided and fully-dehydrided
states that were 23 and 21%, respectively. More signifi-
cant, however, was the observation of how the unit cell of
the LaNi4.75Sn0.25 shrinks more continuously with de-
hydriding than that of LaNi5. Small angle neutron scatter-
ing (SANS) measurements were performed on deuterated
samples of both LaNi5 and LaNi4.75Sn0.25, and showed
that the deuterium distribution was more heterogeneous in
partially-deuterated LaNi5. By generic Monte Carlo simu-
lations and physical arguments, we show that interactions
between Sn and deuterium atoms could be responsible for
homogenizing the deuterium distribution on a mesoscopic
scale. Small angle neutron scattering shows that the
distribution of deuterium in partially-deuterated
LaNi4.75Sn0.25 is more homogeneous than in partially-
˚
deuterated LaNi5, at least on the spatial scale of 40–150 A.
We suggest this more homogeneous deuterium distribution
suppresses the strain gradients that cause defect formation
and fracture during hydriding.
2. Experimental
2.1. Materials preparation
Several different alloys were prepared for this work.
X-ray diffractometry (XRD) and isotherm measurements
were obtained on all materials to ensure they were similar.
Alloys were prepared at Caltech by induction melting
stoichiometric amounts of La and Ni (99.99%) and anneal-
ing in argon at 9508C for 72 h. The Sn-substituted alloys
used to measure pressure–composition isotherms and XRD
lattice parameters were prepared in the same way by
Hydrogen Consultants, Inc. The binary alloy used for
SANS measurements was obtained from Alfa Aesar,
certified at 99.9% purity and packed under argon. The
Sn-substituted sample used for SANS analysis was pre-
pared at Ames Lab from Belmont Sn, zone-refined Ni, and
La purified to 99.99%. Ingots were prepared by arc
melting, followed by annealing in argon at 9508C for 120
h. All handling after annealing was performed in an argon
glove box. All materials were stored in argon-filled vials
after preparation.
2.2. Sievert’s apparatus
Isotherm measurements on the binary alloy were per-
formed with an automated Sievert’s apparatus [33], and
those of the Sn-substituted alloys are from the work of Luo
et al. [38,39]. Hydrogen gas of ULSI grade (99.9999%)
from Matheson Gas Products was used for the isotherm
studies. All gas handling equipment was constructed using
316 stainless steel components, and the MH alloys were
contained in a double-walled copper reactor designed to
enhance thermal conductivity. The ingots were activated

B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
167
before isotherm measurement by hydrogen absorption
followed by several thermally-driven absorption/desorp-
tion cycles to ensure that the surface areas of the samples
were approximately steady state.
were therefore designed to measure differences in I(Q)
from samples with varying amounts of deuterium. If t→he
object of interest were large and homogeneous, G(R)
would extend over large R, so the Fourier transform of Eq.
(1) would be a point at Q50. The SANS profile would be
indistinguishable from the f→orward beam. On the other
hand, small objects have a G(R) that has a shape somewhat
larger than the object itself, and have intensity at larger Q.
The scattering measurements were performed on the
8-m SANS instrument at the NIST Center for Neutron
2.3. XRD procedures
X-ray diffractometry with an Inel CPS-120 diffractome-
˚
ter using Co Ka radiation (la 51.7902 A) was performed
on all alloys to check phase composition, and to measure
alloy lattice parameters. To measure the lattice expansion
induced by hydrogen absorption, the method of Johnson
and Reilly was used to ‘poison’ the surface layers of
hydrided alloys with CO [40]. Surface poisoning of the
alloys was performed after hydrogen activation in the
Sievert’s apparatus. The sample chamber was then sealed
at high pressure, removed from the Sievert’s apparatus,
and relocated to the poisoning stand, where it was im-
mersed in liquid nitrogen. After cooling, the sample was
exposed to CO at 5 atm. After the CO had condensed on
the alloy surface, the liquid nitrogen was allowed to boil
off, slowly bringing the sample back to room temperature.
When the sample had reached room temperature, the
pressure of CO in the chamber was 40–45 atm. This
procedure creates a lanthanum carbonate film on the
powder surface, allowing the alloys to retain their hydro-
gen composition when exposed to air. After the poisoning
treatment, the alloys lose hydrogen slowly so that diffrac-
tion patterns could be recorded over intervals of several
days, each at different hydrogen compositions. Lattice
parameters were determined by Rietveld analysis using the
program RIETAN-2000 [41].
˚
Research. Incident neutron wavelengths of both 5 and 15 A
(Dl/l50.25) were used, providing a usable Q range of
0.007–0.4 A21. All measurements were made at 295 K.
˚
The scattering data were corrected for background, empty-
cell scattering, and pixel-by-pixel detector sensitivity. All
scattering patterns had cylindrical symmetry about the
forward beam, so the data were averaged over rings. The
reported intensities are in absolute units based on direct
beam flux measurements.
Approximately 5 g of LaNi5 and 6 g of LaNi4.75Sn0.25
were precycled once each in the following manner. The
virgin alloy was coarsely ground in a He-filled glovebox to
enhance the rate of D2 absorption, then loaded into a
stainless steel tube sealed by a Cu gasket and equipped
with a gas valve. The alloy was vacuum annealed at 473 K
for 1 h followed by the introduction of ¯0.7 MPa of D2
gas (Cambridge Isotope Laboratories1, 99.9% isotopic
purity). The alloy was subsequently cooled to 293 K in the
D2 atmosphere and equilibrated there for ¯20 h. On
completion of this first deuterium loading, the number of
deuterium atoms absorbed per formula unit as determined
by volumetric uptake measurements were 6.4 and 6.2 for
LaNi5 and LaNi4.75Sn0.25, respectively. The deuterium was
finally removed by evacuating at 373 K for 3 h.
2.4. SANS procedures
A 1-g portion of each precycled alloy was reloaded in
the glovebox into its own stainless steel SANS cell
equipped with a valve and two Viton-o-ring-sealed, quartz
windows of 3.2-mm-thickness. The sample itself was a
disk with a diameter of 25.4 mm and thickness of 1 mm.
The alloys were fully reloaded with D by exposure to D2
at a pressure of approximately 0.4 MPa and a temperature
of 293 K. After SANS measurements, partial removal of D
was accomplished by controlled evacuation of the desorb-
ing D2 gas at 293 K. After each such step, the cell valve
was sealed, the sample quickly equilibrated, and another
SANS measurement was performed. To ensure complete D
removal for the last step of each complete desorption cycle
(i.e. the D-free alloy), the sample was evacuated at
approximately 350 K for up to 1 h. Table 1 indicates the
treatment history after precycling, starting with the fully-
deuterided samples used in the first SANS measurements.
Small angle neutron scattering measures an intensity, I,
versus scattering angle, 2u, near the forward beam. The
reciprocal space distance is Q54p sin(u )/l, where l is
the neutron wavelength (at small angles Q54pu/l). The
intensity I(Q) can be shown to be proportional to:
`
→
→
→
→
I(Q)~
E
exp(2iQ ?R)G(R)d3→R
(1)
2`
where
`
→
→
G(R)~
E
Dr(→r )Dr(→r 1R)d3→R
(2)
2`
→
The SANS intensity I(Q) reflects heterogeneities in a
density, Dr(→r ). This is the density of scattering lengths, b,
→
of the nuclei in the object. For an isolated object, I(Q) is
proportional to ubu2. In the present work, we seek the
density variation of deuterium nuclei. Unfortunately, the
metallic alloy itself also has density variations that contrib-
ute significantly to the SANS intensity. The measurements
1Manufacturers are identified in order to provide complete identifica-
tion of experimental conditions, and such identification is not intended as
a recommendation or endorsement by NIST.

168
B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
Table 1
3.2. XRD
Deuteriding and hydriding sequence for SANS measurements
First SANS desorption cycle (samples initially deuterided)
Fig. 2a shows that the lattice parameters of the binary
LaNi5 change discontinuously with the state of hydro-
genation. The set of peaks from b-phase (the hydride) shift
position to higher 2u-angles once the a-phase has begun to
form, but the diffraction peaks of neither the a-LaNi5 nor
the b-phase change position once the a-phase has begun to
form. With increasing time and loss of hydrogen from the
sample, the intensities of diffractions from the a-phase
continue to grow at the expense of those of the b-phase,
but the lattice parameters of each do not change. Fig. 2b
for LaNi4.75Sn0.25 shows a different behavior. With in-
creasing time, the diffraction peaks from the b-phase
gradually undergo a large shift in position as hydrogen
leaves the sample. The sample may not lose hydrogen
homogeneously, perhaps owing to variations in the surface
permeability, and some of the peaks are broadened or show
splitting. Nevertheless, a Voigt function was fit to all peaks
of the b-phase, and peak centers were used to obtain lattice
parameters from pattern refinement calculations. Results
LaNi5
LaNi4.75Sn0.25
6.4 D
6.2 D
4.0 D
4.0 D
1.6 D
1.9 D
0 D
0 D
Second SANS desorption cycle (samples initially deuterided)
LaNi5
LaNi4.75Sn0.25
6.3 D
6.1 D
3.9 D
4.1 D
1.6 D
1.9 D
Hydrogen or deuterium concentration, x, in formula unit: LaNi5Dx.
3. Results
3.1. Isotherms
The isotherms of the various LaNi52ySny alloys are
shown in Fig. 1, including data measured by Luo et al.
[38,39]. Three effects can be seen with increasing com-
position of tin. First, the plateau pressures decrease as a
result of the larger alloy lattice parameter. Second, the
width (in hydrogen composition) of the isotherm de-
creases, a result of hydrogen blocking by tin atoms.
Finally, the edges of the plateau regions, where the second
phase is nucleating, become less sharp.
Because the isotherm of LaNi5 is extremely sensitive to
composition, annealing conditions, and associated atomic
disorder, it can be used to test the preparation of the alloy.
Low-resolution isotherms were used to confirm that the
LaNi5 –X alloys of the same composition were equivalent.
Additionally, isotherms of both Sn0.25 alloys measured by
Luo et al. [39] also confirm that these alloys are almost
identical in composition and atomic order.
Fig. 2. X-ray diffraction patterns of: (a) binary LaNi5 and (b) ternary
LaNi4.75Sn0.25 containing various amounts of hydrogen. In these figures,
the diffraction pattern from the fully-hydrided material is at the top, and
the lower patterns were measured after increasing amounts of hydrogen
loss. Notice that only the b-phase of the ternary LaNi4.75Sn0.25 alloy
shows any continuous variation with hydrogen desorption.
Fig. 1. Absorption and desorption isotherms of LaNi52ySny alloys, with y
varying from 0 to 0.5. Data for y.0 are from Ref. [38].

B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
169
are presented in Fig. 3. There is no linear correlation
between the horizontal axis (time scale) and the hydrogen
concentration, except that there is less hydrogen in the
samples at later times. Fig. 3 shows that the binary alloy
has very small regions where the lattice expands continu-
ously, consistent with results from Notten et al. [32].
However, the lattice parameters of the Sn-substituted alloy
initially exhibit a gradual reduction until the a-phase
nucleates.
as expected owing to the additional scattering of
deuterium. The shapes of the curves do not change
significantly with deuterium concentration. A second
deuterium cycle showed the same trend, but the changes in
the curves with deuterium concentration were somewhat
larger.
Fig. 5 compares SANS curves for fully deuterated LaNi5
and fully deuterated LaNi4.75Sn0.25. The curves are nearly
identical. The other SANS curves for LaNi5 did not show a
simple shift upwards with deuterium concentration as for
LaNi4.75Sn0.25 in Fig. 4. Fig. 6 shows the SANS data for
the two materials, LaNi4.75Sn0.25 and LaNi5, deuterated to
1/3 of their capacities. Although these curves are similar at
small Q, they diverge at larger values of Q. The sample of
LaNi5 shows evidence of larger heterogeneities in
deuterium concentration at small spatial scales.
3.3. SANS
Fig. 4 shows the SANS curves from the sample of
LaNi4.75Sn0.25 during its first discharge with deuterium.
˚
The data sets acquired with 5 and 15 A neutrons cover
different ranges of Q, but overlap well from 0.03,Q,0.1
˚
A where both sets are reliable. The data show a systematic
Fig. 7 shows the difference when the SANS intensity
from LaNi4.75Sn0.25 was subtracted from that of LaNi5 at
increase in intensity as deuterium is added to the material,
Fig. 3. X-ray lattice parameters and unit cell volume of LaNi52ySny versus time for hydrogen desorption; (a) a-parameter; (b) c-parameter; (c) unit cell
volume; (d) fractional difference between a- and b-phase volumes.

170
B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
Fig. 4. Porod (log–log) plot of the measured SANS intensity from the
first deuterium desorption cycle of sample of LaNi4.75Sn0.25
Fig. 6. Porod plot of the measured SANS intensity from the first full
.
deuterium desorption to 33% absorption (H/unit cell52.0) for LaNi5 and
LaNi4.75Sn0.25
.
the same state of deuteration. The difference is representa-
tive of the larger heterogeneity of the deuterium con-
centration in LaNi5 –D. This difference is small for the
fully-deuterated samples — their curves in Fig. 5 are
nearly the same. The difference between the curves in Fig.
6 for samples deuterated to 1/3 their capacity is about five
times larger, at least at large Q. Furthermore, the excess
scattering intensity of the partially-deuterated samples of
LaNi5 has
LaNi4.75Sn0.25. Similar plots were obtained by subtracting
the spectra of the deuterated LaNi5 from the undeuterated
samples of LaNi5. Lines of slope 23 and24 are drawn on
a
different shape from that of the
Fig. 5. Porod plot of the measured SANS intensity from the first full
Fig. 7. Porod plot of the difference in measured SANS intensity from
LaNi5 and LaNi4.75Sn0.25 at different states of deuteration.
deuterium charging of LaNi5 and LaNi4.75Sn0.25
.

B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
171
with a minority solute species could affect distributions of
hydrogen atoms in alloys. The simulations were performed
with software codes used in our previous studies on atom
diffusivities in binary and ternary metallic alloys [42–44],
but modified to include a sublattice for interstitial hydro-
gen atoms. Only hydrogen atoms were allowed to move by
jumping into adjacent empty interstitial sites, using the
algorithm of Metropolis et al. [45]. Interactions between
hydrogen atoms were attractive over first-neighbor dis-
tances. The hydrogen atoms were either attracted or
repelled by the minority solute atoms. Initial distributions
of the hydrogen atoms over the interstitial sites were either
random or clustered, and the evolution of the hydrogen
atom distribution was monitored in time and in Monte
Carlo steps [46].
An enlarged region of metal atoms plus hydrogen atoms
is shown in Fig. 9a, whereas Fig. 9b,c shows only the
Fig. 8. Guinier plot of difference in measured SANS intensity from
LaNi5 minus that of LaNi4.75Sn0.25 at different states of deuteration.
Fig. 7, showing that the excess scattering from the sample
of 1/3 or 2/3 deuterated LaNi5 has a slope closer to 23
than 24.
The data of Fig. 7 are plotted in Guinier form in Fig. 8.
Approximate Guinier radii, rG, were obtained from the
slopes, m, of the low-Q and high-Q parts of the curves in
Fig. 8 (below and above Q250.008), as rG2 53 ln(m).
˚
These radii of gyration were 140 A for the low-Q part and
˚
44 A for the high-Q part. These values are not accurate for
several reasons, especially since QrG .1 over at least some
of these ranges, so the Guinier approximation is not
reliable. Nevertheless, it seems that the excess SANS
intensity from the partially-deuterated LaNi5 arises from
heterogeneities of the deuterium distribution on the order
˚
of 40–150 A.
The SANS intensities at the minimum and maximum
values of
Q do not provide information on the
heterogeneities of deuterium concentration. At Q.0.1, the
SANS curves are dominated by instrument background and
incoherent scattering from hydrogen impurities or the gas.
The flat part of the intensity at Q.0.3 was reasonably
linear with the amount of deuterium in the sample, as
expected if there was a residual hydrogen concentration in
the deuterium gas. At small Q the data decrease, owing to
effects of the beamstop. This occurs at approximately 0.01
21
˚
and 0.03 A for neutrons with incident wavelengths of 15
and 5 A, respectively.
˚
Fig. 9. (a) Typical region from simulation of part c, showing large metal
atoms and small hydrogen atoms. Note preference of hydrogen atoms for
the solute B-atoms (largest circles); (b) Hydrogen distributions in random
alloy, cB 50.08, cH 50.5, VA–H 50, VB–H 522 over long-range VH–H 52
1.7; (c) Hydrogen distributions in random alloy, cB 50.08, cH 50.5,
4. Monte Carlo simulations
V
A–H 50, VB–H 521.2, VH–H 521.7. Interatomic potential between hydro-
Two-dimensional Monte Carlo simulations were helpful
for understanding how the interactions of hydrogen atoms
gen (H), metal (A), and minority solute (B) atoms are in units of
kT/atom.

172
B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
hydrogen atoms. The underlying metal atom lattice con-
tained 8% minority solute atoms for all simulations, and
the random metal atom arrangement was kept fixed
throughout all simulations. Fig. 9b shows the hydrogen
distribution (after an intermediate time) for the case where
all metal atoms had no interactions with hydrogen atoms.
The clustering of hydrogen for this case proceeded with
normal kinetics, where the length of the characteristic
hydrogen clusters grew approximately as t1 / 3. Hydrogen
clustering occurred initially over short correlation lengths,
and for Fig. 9b the correlation length has grown to about
10 lattice parameters. When there were no interactions
between hydrogen atoms and the minority metal atoms, a
sharp interface was always observed around the clusters of
hydrogen atoms.
Fig. 9c is a snapshot obtained after long times for a
simulation with an attractive hydrogen–solute interaction.
For the earliest times, clustering of hydrogen atoms
occurred normally. When the correlation length for the
hydrogen atom clusters became slightly larger than the
mean separation between the minority solute atoms, the
coarsening stopped. We believe the state of Fig. 9c is an
equilibrium state because similar states were reached from
initial distributions of hydrogen atoms that were either
clustered or random. The reason for the arrest of coarsen-
ing is that energetically favorable metal–hydrogen bonds
would have to be sacrificed for the hydrogen distribution to
coarsen further.
The simple picture of an attractive interaction between a
hydrogen atom and a neighboring solute atom is consistent
with a lowering of the plateau pressure when solute is
added to the alloy. However, a lowering of plateau
pressure could also occur when the hydrogen–solute first-
neighbor interaction is repulsive, but the hydrogen–matrix
interaction becomes more attractive with solute concen-
tration. Simulations of this case also evolve hydrogen
distributions similar to those of Fig. 9c, and also show an
arrest of coarsening.
These Monte Carlo simulations are inadequate for
predicting precise hydrogen distributions or kinetics. For
example, the simulations are two-dimensional and do not
include elastic interactions. Nevertheless, the simulations
do show some general trends of how local variations in the
density of solute atoms can alter the distribution of
hydrogen atoms. The arrest of coarsening caused by
hydrogen–solute interactions is particularly interesting. For
a fixed concentration of hydrogen, clustered into three-
dimensional hydride (b-phase) zones, the surface-to-vol-
ume ratio is larger for smaller clusters. The surface energy
therefore decreases as the clusters grow larger, and this
provides the well-known driving force for coarsening. The
driving force diminishes as the clusters grow larger. The
hydrogen–solute interactions alter this picture of coarsen-
ing. At low concentrations the Sn solute atoms are
distributed randomly over the 3g sites of the LaNi5 crystal.
A random distribution is not a uniform one, so there will
be statistical fluctuations in the local concentration of
solute2. When a solute concentration, c, is averaged over a
number of sites, N, the fluctuations in c are approximately
(cN)21 / 2. A cluster of b-phase will seek out the favorable
fluctuations in concentration, but these become less favor-
able as the cluster grows larger. This solute–hydrogen
interaction energy therefore favors small clusters, or
perhaps ramified clusters. The balance between these two
effects results in the arrest of coarsening at a particular
cluster size.
5. Discussion
The isotherms in Fig. 1 show that the width of the
plateau decreases with an increase in concentration, y, of
Sn in the composition LaNi12ySny. For example, approxi-
mately 60% of the low-energy hydrogen sites in the b-
phase are available when y50.5. When y50.5, the Sn
atoms occupy approximately 17% of the 3g sites, the other
3g sites being occupied by Ni atoms. Evidently, therefore,
the addition of each Sn atom to a 3g site blocks approxi-
mately two sites from low-energy hydrogen adsorption in
the b-phase. These two higher-energy sites could be those
adjacent to Sn atoms, but this is beyond the level of detail
that can be deduced from knowledge only of isotherms and
hydrogen site occupancies in LaNi5 [47,48]. The Sn
reduces the energy of hydrogen absorption of other low-
energy sites in the alloy, since Fig. 1 shows that the
plateau pressure decreases upon alloying with Sn. From
the change in equilibrium pressure by about a factor of 40
between binary LaNi5 and LaNi4.5Sn0.5, we deduce that
the chemical potential for a hydrogen atom is reduced by
1/2 kBT ln(40)51/2 25 (3.7)546 meV in the Sn-con-
taining alloy. (Standard tables provide 40 meV [49]).
These low-energy sites are presumably located somewhere
in the vicinity of the Sn atoms, but again we cannot deduce
the crystallographic sites from information from the iso-
therms. We can say, however, that the addition of Sn atoms
causes the sites occupied by hydrogen in binary LaNi5 to
have large differences in the chemical potential for hydro-
gen atoms.
The origin of the reduction in absorption energy of the
low-energy sites in LaNi5 –Sn alloys has been attributed to
an increase in the size of the interstitial sites where
absorbed hydrogen is located, which is correlated with the
expansion of the lattice parameter as Sn is substituted for
Ni [39]. Our measurements show that the lattice parameter
increases with Sn concentration [50] (cf. Fig. 3a–c). It
seems reasonable that the energy penalty for inserting a
hydrogen atom in the lattice is reduced when the interstitial
2The concentration itself depends on the volume of sampling. Consider
the extreme cases — over the full crystal there will be a uniform
concentration, whereas over an individual 3g site the solute concentration
will be 0.0 or 1.0.

B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
173
sites are larger, but the actual energy penalty depends on
the amount of expansion caused by the hydrogen atom.
This is obtained from our X-ray measurements on
poisoned alloys. The expansion of unit cell volume after
fully hydriding is found to be 24% for LaNi5, with 21% of
this occurring discontinuously between the a- and b-
phases. The expansions for LaNi4.75Sn0.25 is 21% for
fully-hydrided material. With hydrogen depletion, how-
ever, Fig. 3d shows that the difference in specific volume
between the a- and b-phases becomes as low as 12%.
The SANS intensity profiles have the same shape for all
states of deuteration of the LaNi4.75Sn0.25. The log–log
curves of Fig. 4 are nearly coincident when they are shifted
vertically, meaning that they all have the same mathemati-
cal form, and differ only by a multiplicative constant.
Because the additional scattering from the deuterium has
the same shape as that from the alloy without deuterium,
the density distribution of deuterium is the same as the
density distribution of the alloy. The deuterium is homoge-
neous in the alloy, at least over the spatial range of SANS
sensitivity. This homogeneity is also found for the fully-
deuterated LaNi5 alloy. The curves in Fig. 5 are nearly
coincident, even without vertical shifts. This is as ex-
pected, since the only difference between the two samples
is a small amount of Sn, which has a coherent scattering
length intermediate between La and Ni. For fully-deuter-
ated samples of LaNi4.75Sn0.25 and LaNi5, the SANS
profiles are nearly the same, and have the same shape as
for the undeuterated alloy itself. This suggests that all
available sites for deuterium are filled in the fully-deuter-
ated samples, and the deuterium concentration is homoge-
neous over the spatial range of the SANS sensitivity.
In partially-deuterated samples of LaNi5, however, there
is a distinct difference in the shape of the SANS intensity
(Fig. 6). The deuterium in these materials does not follow
the density of the alloy, but has its own heterogeneities.
These variations in deuterium density are on the order of
These are, presumably, sites somewhere in the vicinity of
Sn atoms. Over the larger spatial scale where SANS is
sensitive, the following energy argument suggests that a
more homogeneous distribution of hydrogen atoms is
expected. Assume again that the Sn atoms are distributed
at random on the 3g sites. As described in Section 4, for a
Sn concentration of c, the fluctuations in c are approxi-
mately (cN)21 / 2 when taken over a number N of 3g sites.
3
˚
With 23 A associated with each 3g site, the number of 3g
˚
sites in a sphere of diameter 50 A is approximately 3000,
so the fluctuations in this number are (0.18 3000)21 / 2, or
about 4%. Using a variation in chemical potential of
hydrogen of 40 meV, these fluctuations in hydrogen
concentration correspond to fluctuations in chemical po-
tential of only about 2 meV per hydrogen atom in the
cluster, which is small compared to room temperature. We
therefore expect the distribution of hydrogen in
LaNi4.75Sn0.25 to tend to be uniform on spatial scales of 50
˚
A or larger, in spite of the local variations in chemical
potential in hydrogen sites near Sn atoms.
On the other hand, without solute atoms to cause local
variations in the chemical potential of hydrogen, the
distribution of hydrogen in binary LaNi5 should be affect-
ed primarily by the surface energy of the hydride/matrix
interface, modified by elastic interactions between the
hydride zones. Our SANS curves of partially-deuterated
LaNi5 suggest the hydride zones are two-dimensional. A
plate is a common morphology for an elastic precipitate in
an anisotropic crystal. Previous SANS work has shown
similar evidence for Pd hydride [52]. Planar defects have
been reported by TEM observations on LaNi5 [53].
The spatial scales of these hydrided zones, or rather the
correlation length for the hydrogen density fluctuations in
˚
partially-hydrided binary LaNi5, are of order 40–150 A.
With a linear strain of 5–7% from full hydriding, the
variation of hydrogen concentration from 0 to 100% over
˚
distances from 40 to 150 A results in a large strain gradient
˚
40–150 A. Furthermore, the decrease in the slope of the
of 0.2–1% across a unit cell. It is not surprising that this
large strain gradient could lead to the generation of
dislocations or other crystalline defects that could initiate
fracture. We propose that the effect of Sn atoms on
smoothing the mesoscopic hydrogen concentration profile
is why LaNi–Sn sustains less microstructural damage than
binary LaNi5. Perhaps this is responsible for the improved
cycle life of solute-modified LaNi5 for application as both
gas storage and electrochemical cell materials.
Porod plot of Fig. 7 from –4 to –3 could indicate that
these hydrogen-rich regions in the partially-deuterated
samples of LaNi5 have two-dimensional features [51]. The
distribution of deuterium in the LaNi5 is not homogeneous
˚
on the spatial scale of 40–150 A.
Our SANS results show, to the best of our knowledge
for the first time, differences in the distribution of hydro-
gen (deuterium) in LaNi5 and LaNi–Sn. At spatial scales
of tens of unit cells, the distribution of hydrogen in
partially-hydrided LaNi–Sn is more homogeneous than in
partially-hydrided LaNi5. Based on the measured iso-
therms, we argued that a Sn atom affects the chemical
potential of an interstitial hydrogen atom by up to 40 meV.
This is large compared to room temperature, and is
arguably comparable to the VB–H 521.2 kBT used in the
two-dimensional Monte Carlo simulations of Section 4. We
therefore expect that as hydrogen is added to the material,
the hydrogen atoms will first occupy the low-energy sites.
6. Conclusion
We performed a comparative study of hydrogen dis-
tributions and strains in LaNi5 and the alloy
LaNi4.75Sn0.25, which has been shown to have a superior
cycle life to binary LaNi5 for hydrogen storage and in
electrochemical cells. X-ray diffractometry showed that the

174
B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
[12] A. Anani, A. Visintin, K. Petrov, S. Srinivasan, J.J. Reilly, J.R.
Johnson, R.B. Schwarz, P.B. Desch, J. Power Sources 47 (1994)
261.
[13] M.P. Sridhar Kumar, W. Zhang, K. Petrov, A.A. Rostami, S.
Srinivasan, G.D. Adzic, J.R. Johnson, J.J. Reilly, H.S. Lim, J.
Electrochem. Soc. 142 (1995) 3424.
[14] G.D. Adzic, J.R. Johnson, J.J. Reilly, J. McBreen, S. Mukerjee, M.P.
Sridhar Kumar, W. Zhang, S. Srinivasan, J. Electrochem. Soc. 142
(1995) 3429.
overall volume dilatations during full hydriding were
approximately the same for the two alloys, but the unit cell
volume of the b-phase (the hydride) in LaNi4.75Sn0.25
became closer to that of the matrix a-phase during loss of
hydrogen. Gas-phase isotherms show that the Sn atoms
cause large changes in the chemical potential at sites of
hydrogen occupancy, with some sites having low and other
sites having high chemical potentials. We used generic
hydrogen–solute interactions in Monte Carlo simulations
and physical arguments to show why the normal coarsen-
ing behavior of hydride zones will be altered, and even
arrested, by hydrogen–solute interactions. Interactions
between Sn and deuterium atoms provide a mechanism for
homogenizing the deuterium distribution over mesoscopic
spatial scales somewhat larger than a unit cell.
´
[15] A. Percheron-Guegan, M. Latroche, J.C. Achard, Y. Charbre, J.
Bouet, Electrochem. Soc. Proc. 94–27 (1994) 196.
¨
[16] F. Meli, A. Zuttel, L. Schlapbach, J. Alloys Comp. 202 (1993) 81.
¨
[17] F. Meli, A. Zuttel, L. Schlapbach, Z. Phys. Chem. 183 (1994) 371.
¨
[18] F. Meli, A. Zuttel, L. Schlapbach, J. Alloys Comp. 231 (1995) 639.
[19] B.V. Ratnakumar, C.K. Witham, G. Halpert, B. Fultz, J. Electrochem.
Soc. 141 (1994) L89.
[20] B.V. Ratnakumar, C.K. Witham, R.C. Bowman, Jr, A. Hightower, B.
Fultz, J. Electrochem. Soc. 143 (1996) 2578.
[21] B.V. Ratnakumar, S. Surampudi, B. Fultz, C. Witham, R.C. Bowman,
Jr, A. Hightower, NASA Tech Briefs, August 1998, pp. 60–61.
[22] C.K. Witham, B. Fultz, B.V. Ratnakumar, R.C. Bowman, Jr, in: P.D.
Bennett, T. Sakai (Eds.), Proc. Symp. on Hydrogen and Metal
Hydride Batteries, The Electrochemical Society, Inc., 1995, 94-27.
ISBN: 1-56677-086-6, p. 68.
Small angle neutron scattering shows that the distribu-
tion of deuterium in partially-deuterated LaNi4.75Sn0.25 is
more homogeneous than in partially-deuterated LaNi5, at
˚
least on the spatial scale of 40–150 A. The deuterium
distribution is found to be the same in fully-deuterated
LaNi5 and LaNi4.75Sn0.25, and this distribution matches
well the density distribution of the sample itself (i.e. the
deuterium distribution is homogeneous). We suggest that a
more homogeneous deuterium distribution could suppress
the internal stresses upon hydriding that cause defect
formation and fracture.
[23] P.H.L. Notten, R.E.F. Einerhand, J.L.C. Daams, J. Alloys Comp.
231 (1995) 604.
[24] P.H.L. Notten, in: F. Grandjean, G. Long, K.H.J. Buschow (Eds.),
Interstitial Metallic Alloys, Kluwer, London, 1995, p. 151.
[25] O.A. Petrii, S.Y. Vasina, I.I. Korobov, Russ. Chem. Rev. 65 (1996)
181.
[26] P.H.L. Notten, R.E.F. Einerhand, J.L.C. Daams, J. Alloys Comp.
210 (1994) 221.
[27] T. van Waldkirch, P. Zurcher, Appl. Phys. Lett. 33 (1978) 689.
[28] A.H. Boonstra, T.N.M. Bernards, J. Less-Common Met. 161 (1990)
245.
Acknowledgements
[29] A.H. Boonstra, T.N.M. Bernards, G.J.M. Lippits, J. Less-Common
Met. 159 (1990) 327.
¨
[30] D. Chartouni, F. Meli, A. Zuttel, K. Gross, L. Schlapbach, J. Alloys
The authors thank Dr R.C. Bowman, Jr. for discussions
and advice, and Dr J.G. Barker for assistance with the
SANS measurements. This work was supported by DOE
through Basic Energy Sciences Grant DE-FG03-
00ER15035.
Comp. 241 (1996) 160.
[31] P.H.L. Notten, J.L.C. Daams, R.E.F. Einerhand, J. Alloys Comp.
210 (1994) 233.
[32] P.H.L. Notten, J.L.C. Daams, A.E.M. De Veirman, A.A. Staals, J.
Alloys Comp. 209 (1994) 85.
[33] R.C. Bowman, C. Luo, C.C. Ahn, C.K. Witham, B. Fultz, J. Alloys
Comp. 217 (1995) 185.
[34] S. Mukerjee, J. McBreen, J.J. Reilly, J.R. Johnson, G. Adzic, K.
Petrov, M.P.S. Kumar, W. Zhang, S. Srinivasan, J. Electrochem. Soc.
142 (1995) 2278.
References
[1] J.G.G. Willems, K.H.J. Buschow, J. Less-Common Met. 129 (1987)
13.
[2] T. Sakai, M. Matsuoka, C. Iwakura, in: K. Gschneidner, Jr, L.
Eyring (Eds.), Handbook on Physics and Chemistry of Rare Earths,
Vol. 21, Elsevier, Amsterdam, 1994, p. 133.
[3] R.L. Cohen, K.W. West, J.H. Wernick, J. Less-Common Met. 70
(1980) 229.
[4] R.L. Cohen, J.H. Wernick, Science 214 (1981) 1081.
[5] R.L. Cohen, K.W. West, J. Less-Common Met. 95 (1983) 15.
[6] A.R. Miedema, P.F. de Chatel, F.R. de Boer, Physica 100B (1980)
1.
[35] M.L. Wasz, P.B. Desch, R.B. Schwarz, Phil. Mag. 74 (1996) 15.
[36] M.L. Wasz, R.B. Schwarz, Mater. Sci. Forum 225–227 (1996) 859.
[37] M.L. Wasz, R.B. Schwarz, S. Srinivasan, M.P.S. Kumar, Mater. Res.
Soc. Symp. Proc. 393 (1995) 237.
[38] S. Luo, W. Luo, J.D. Clewley, T.B. Flanagan, L.A. Wade, J. Alloys
Comp. 231 (1995) 467.
[39] S. Luo, J.D. Clewley, T.B. Flanagan, R.C. Bowman, L.A. Wade, J.
Alloys Comp. 267 (1998) 171.
[40] J.R. Johnson, J.J. Reilly, Inorg. Chem. 17 (1978) 3103.
[41] F. Izumi, T. Ikeda, Mater. Sci. Forum 198 (2000) 321.
[42] B. Fultz, J. Chem. Phys. 87 (1987) 1604.
[7] T. Sakai, K. Oguru, H. Miyamura, N. Kuriyama, A. Kato, H.
Ishikawa, C. Iwakura, J. Less-Common Met. 161 (1990) 193.
[8] T. Sakai, T. Hagama, H. Miyamura, N. Kuriyama, A. Kato, H.
Ishikawa, J. Less-Common Met. 172–174 (1991) 1175.
[9] T. Sakai, H. Yoshinaga, H. Miyamura, H. Ishikawa, J. Alloys Comp.
180 (1992) 37.
[43] H. Ouyang, B. Fultz, J. Appl. Phys. 66 (1989) 4752.
[44] T.F. Lindsey, B. Fultz, in: B. Fultz, R.W. Cahn, D. Gupta (Eds.),
Diffusion in Ordered Alloys. TMS Monograph Series, Warrendale,
1993, pp. 91–106.
[45] N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller, E.
Teller, J. Chem. Phys. 21 (1953) 1087.
[10] T. Sakai, H. Miyamura, N. Kuriyama, H. Ishikawa, I. Uehara, Z.
Phys. Chem. 183 (1994) 333.
[46] Y. Limoge, J.L. Bocquet, Acta Metall. 36 (1988) 1717.
´
[47] C. Lartigue, A. Percheron-Guegan, J.C. Archard, J.L. Soubeyroux, J.
[11] J.G.G. Willems, Philips J. Res. 39 (Suppl. 1) (1984) 1.
Less-Common Met. 113 (1985) 127.

B. Fultz et al. / Journal of Alloys and Compounds 335 (2002) 165–175
175
[48] P. Thompson, J.J. Reilly, L.M. Corlis, J.M. Hastings, R. Hem-
plemann, J. Phys. F Met. Phys. 16 (1986) 675.
[50] C.K. Witham, Ph.D. Thesis, Materials Science, California Institute
of Technology, 1999.
[49] R.D. McCarty, J. Hord, H.M. Roder, in: Selected Properties of
Hydrogen, NBS Monograph 168, US Government Printing Office,
Washington, DC, 1981, pp. 6-130–6-269. W.E. Forsythe (Ed.),
Smithsonian Physical Tables, 9th ed., Smithsonian Institute,
Washington, 1964.
[51] B. Fultz, J.M. Howe, Transmission Electron Microscopy and
Diffractometry of Materials, Springer, Heidelberg, 2001, Chapter 8.
[52] B.J. Heuser, J.S. King, W.C. Chen, J. Alloys Comp. 292 (1999) 134.
[53] A.E.M. De Veirman, A.A. Staals, P.H.L. Notten, Phil. Mag. A 70
(1994) 837.