Hydrogen storage in spherical and platelet palladium nanoparticles

Article abstract of DOI:10.1016/j.jallcom.2004.06.105

The PdH_x system is used to investigate possible benefits of hydrogen storage in nanoparticulate matter. Particles of different morphology (platelets/tabular or spherical) were synthesized with minimum dimension in the 4-10 nm range via reverse and bilayer-micellular techniques. High specific surface area (SSA ～ 40-80 m²/g) was obtained for all Pd samples studied. For most particles, it was found that the SSA can only be maintained if processing or H-adsorption temperatures did not exceed T ～ 50-100 °C. The isothermal hydrogen uptake (to 10 bar) of the nanoparticles was measured gravimetrically at 50 °C and compared with that of bulk powders (～micron grain Pd). It was noticed that the nanoparticulate isotherm plateaus (α + β-phases) were not as flat, or as wide, as in the bulk. Several samples were observed to store 10-20% more than the bulk at 10 bar, suggesting that surface and subsurface sites in nanoparticulate matter provide an additional and significant set of adsorption sites. In fact, using the width of the α + β plateau as a measure of the normal bulk (octahedral, O) site concentration, we can estimate that many of the nanoparticulate samples studied exhibit a larger fraction of subsurface sites than bulk-like O-sites. Post-synthesis hydrazine washing has been observed to be a crucial factor in enhancing the hydrogen uptake performance of the nanomaterials studied - a marked improvement in the washed samples over the unwashed ones suggests a possible removal of some of the disadvantageous organics from the sample surfaces.

Full text of DOI:10.1016/j.jallcom.2004.06.105

Journal of Alloys and Compounds 389 (2005) 234–242
Hydrogen storage in spherical and platelet palladium nanoparticles
S. Kishorea, J.A. Nelsonb, J.H. Adairb, P.C. Eklunda,b,∗
a
Department of Physics, 104 Davey Laboratory, The Pennsylvania State University, University Park, PA 16802, USA
b
Materials Research Institute and Department of Materials Science and Engineering,
The Pennsylvania State University, University Park, PA 16802, USA
Received 6 May 2004; accepted 8 June 2004
Abstract
The PdHx system is used to investigate possible benefits of hydrogen storage in nanoparticulate matter. Particles of different morphology
(platelets/tabular or spherical) were synthesized with minimum dimension in the 4–10 nm range via reverse and bilayer-micellular techniques.
High specific surface area (SSA ∼ 40–80 m2/g) was obtained for all Pd samples studied. For most particles, it was found that the SSA can
only be maintained if processing or H-adsorption temperatures did not exceed T ∼ 50–100 ◦C. The isothermal hydrogen uptake (to 10 bar) of
the nanoparticles was measured gravimetrically at 50 ◦C and compared with that of bulk powders (∼micron grain Pd). It was noticed that the
nanoparticulate isotherm plateaus (␣ + ␤-phases) were not as flat, or as wide, as in the bulk. Several samples were observed to store 10–20%
more than the bulk at 10 bar, suggesting that surface and subsurface sites in nanoparticulate matter provide an additional and significant set
of adsorption sites. In fact, using the width of the ␣ + ␤ plateau as a measure of the normal bulk (octahedral, O) site concentration, we can
estimate that many of the nanoparticulate samples studied exhibit a larger fraction of subsurface sites than bulk-like O-sites. Post-synthesis
hydrazine washing has been observed to be a crucial factor in enhancing the hydrogen uptake performance of the nanomaterials studied—a
marked improvement in the washed samples over the unwashed ones suggests a possible removal of some of the disadvantageous organics
from the sample surfaces.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Palladium; Hydrogen storage; Micellular synthesis; Isotherms; Subsurface sites
1. Introduction
d-bands of the host metal, changing the density of states
at the Fermi surface and causing shifts of the energy bands
[3]. These electronic reactions produce both short-range and
long-range interactions between the electronic species. Be-
sides bringing about changes in local and global electronic
structure of the host metal, hydrogen also induces changes
in the cohesive forces between atoms in the host matrix, as
evidenced by several theoretical investigations [4,5]. Neu-
tron diffraction studies in various metal–hydrogen systems
have shown that hydrogen occupies interstitial sites in the
metal lattice. This process causes the nearest-neighbor metal
atoms to move farther apart from each other, causing an in-
crease in the nearest-neighbor separation between atoms. In
palladium, the lowest energy sites that are occupied by hy-
drogen are the interstitial sites with octahedral symmetry [6].
However, as all the octahedral interstitials are progressively
filled by hydrogen, a ‘spill-over’ occurs as hydrogen occu-
The Palladium–Hydrogen system is one of the most
widely and thoroughly investigated bulk metal–hydrogen
systems studied to date, mainly because of the availability of
pure samples and the fact that no special surface treatments
of the bulk samples are required [1,2]. Palladium belongs to
the second transition metal series, with 10 electrons in the
4d electron shell and exhibits high magnetic susceptibility
and high electronic specific heat, since its electronic density
of states at the Fermi level is the highest of any pure metal.
Experimental and theoretical investigations have proved that
in transition metals like Pd, the electron accompanying the
proton of the dissociated hydrogen atom enters the s- and
∗
Corresponding author. Tel.: +1 814 865 5233; fax: +1 814 865 9851.
E-mail address: pce3@psu.edu (P.C. Eklund).
0925-8388/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2004.06.105

S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
235
pies the next higher energy tetrahedral interstitials, driven by
a higher molecular gas fugacity, which is in turn governed by
the higher gas pressure.
ies that showed that the entire nanocrystalline Pd sample is
converted into the ␤-phase on hydriding. Small angle neu-
tron scattering studies by Sanders et al. [11] have suggested
that this behavior (increase in [H]␣ and decrease in [H]␤)
is attributed to the occupancy of hydrogen in the surface
of the grains contacting the voids. In addition, a significant
decrease in the ‘miscibility gap’ (width of the plateau re-
gion) of the isotherms has been explained based on hydrogen
occupancy at the surface and subsurface sites [14,15]. This
has also been verified through Monte-Carlo simulations of
the pressure-composition isotherms of nanometer-sized pal-
ladium clusters on hydrogen loading [12]. Proton spin-lattice
relaxation times T1 (determined though NMR measurements)
and magnetic susceptibility χ results in nanocrystalline Pd
hydrides indicate an inhomogeneous distribution of hydro-
gen in the lattice with most of the hydrogen appearing to
occupy the grain boundaries. The ␣–␤ phase mixture in the
plateau region of the isotherm is reflected in the anomalous
behavior of the proton spin-lattice relaxation time T1 [13].
From experimental volumetric pressure-composition studies
on nanoscale clusters of Pd, Pundt et al. [14] and Sachs et al.
[15] have postulated that the observed enhancement in hy-
drogen solubility in the solid solution region of the isotherms
is due to the presence of additional subsurface sites for hy-
drogen occupancy. Similarly, recent work on nanocrystalline
magnesium has shown that the kinetics of hydrogen uptake is
significantly faster in nanocrystalline materials, when com-
pared with the bulk [16].
We report here, on the synthesis, characterization and hy-
drogen adsorption studies of nanoparticulates of palladium,
prepared in spherical and platelet morphology. The results
of our hydrogen adsorption studies are also compared with
results on bulk Pd and nanocrystalline Pd. It should be em-
phasized that the investigations carried out here have been
on weakly coupled nanoparticles of Pd, as contrasted with
nanocrystalline materials that have been previously investi-
gated [7,8,10–16].
1.1. Why can ‘nano’ be better for hydrogen storage?
Nanocrystalline materials have already been shown to
have distinct differences in their hydrogen uptake charac-
teristics when compared with their bulk counterparts. Rapid
H-diffusion has been reported to take place along the grain
boundaries in nanocrystalline materials, improving the ki-
netics for H-adsorption and desorption. Nanoparticulates
(or loosely aggregated nanoparticles), on the other hand,
can exhibit characteristic benefits of their own. First, small
nanoparticles have been reported to exhibit dilated lattices
that would result in larger interstitial volumes for hydro-
gen storage (and possibly, lower the H-binding energy).
This could result in better storage characteristics. Second,
nanoparticulates may offer a new set of surface and subsur-
face adsorption sites. Neutron scattering studies [17] have
revealed that hydrogen occupies surface region sites in nano-
materials, in addition to the regular interstitial and grain
boundary sites. Third, the characteristic hydrogen ion dif-
fusion length is automatically reduced to the nanoparticle
radius, or the minimum dimension for tabular particles. This
would, of course, result in faster kinetics for hydrogen ad-
sorption/desorption. Fourth, nanoparticulates are distinctly
more resistant to hydrogen-induced morphological changes
such as hydrogen-induced decrepitation, hydrogen-induced
amorphization[18,19], etc. Fifthandfinally, nanoparticlehet-
erostructures can be designed to optimize hydrogen storage
performance, i.e., core–shell structures and added hydrogen-
permeable passivation coatings, etc. are possible.
2. Experimental details
All chemicals used in this study were reagent grade. Pal-
ladium nitrate hydrate (Aldrich Chemical Co.) was used
as the palladium source. Igepal CO-520 (Rhodia Chemi-
cal), cyclohexane (Aldrich Chemical Co.), octylamine (OA)
(99%, Aldrich Chemical Co.), amylamine (AA) (99%
Aldrich Chemical Co.), ethanol (Pharmco), palladium (pu-
rity 99.9%+, 1–1.5 ␮ Aldrich Chemical Co.), and hydrazine
hydrate (Aldrich Chemical Co.) were all used as received.
De-ionized water was used for all synthesis.
Recent hydrogen adsorption studies on nanocrystalline
palladium [7] have shown that the amount of hydrogen up-
take in nanocrystalline PdHx is higher than that for the bulk
material in the solid solution region of the isotherms (i.e.,
x < 0.1 for the ␣-phase). However, in the hydride phase re-
gion (␤-phase), the maximum hydrogen concentration was
observed to be significantly lower than that observed in bulk
materials. Mutschele and Kirchheim [8,9] proposed an ex-
planation that the increase in the ␣-phase H concentration
[H]␣ was due to an increase in the hydrogen solubility at the
grain boundaries and the decrease in the ␤-phase concentra-
tion [H]␤ was attributed to the fact that the significant grain
boundary regions do not transform to the ␤-phase. However,
Eastman et al. [10] carried out in situ X-ray diffraction stud-
2.1. Reverse micelle synthesis of spherical nanoparticles
Micelles were prepared using a procedure similar to that
of Li et al. [20]. Poly(oxyethylene)nonylphenyl ether (Igepal
CO-520) and cyclohexane are mixed, and the prescribed vol-
umeof0.04 MaqueousPd(NO3)2 isaddedtoformthereverse
micelle phase. The particle size was controlled by varying the
ratio R = [water]/[surfactant]. The solution was rapidly stirred
and equilibrated for 1 h. Hydrazine hydrate was added to re-
duce the Pd2+ to Pd metal, as the solution was stirred rapidly
for 5 min. After 20 h of equilibration, a volume of ethanol,
equal to the volume of cyclohexane was added to the solu-
tion to decompose the reverse micelle. The black product was
allowed to settle and the clear solution was decanted and dis-

236
S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
carded. Ethanol and a small amount of hydrazine hydrate (to
reduce any PdO that might be present) were again added to
ensure that all Igepal CO-520 was removed. This wash pro-
cess was repeated four times and the sample was allowed to
dry in air.
AFM were prepared by placing a drop of sample dispersed
in ethanol onto atomically flat, freshly cleaved mica. Specific
surface area (SSA) measurements (N2-BET) were carried out
using a Micromeritic Gemini. The fractions of PdO, organ-
ics, water, etc., on the nanoparticle surface were determined
by thermogravimetric analysis (TGA) in flowing O2 (TGA:
TA Instruments SDT 2960).
2.2. Lamellar bilayer synthesis of platelets
Lamellar bilayers were prepared using a procedure sim-
ilar to that of Yener et al. [21]. Octylamine, amylamine
and 0.04 M aqueous Pd2+, as the nitride, were used to form
the lamellar bilayer phases. Surfactant octylamine and co-
surfactant amylamine were added in a weight ratio of 25:1,
respectively. It has been shown by Adair et al. [22] that bi-
layers based on octylamine resulted in inorganic (CdS or Ag)
platelets with poorly defined edges. Anisotropically shaped
clusters were also produced. To prevent the formation of an
anisotropically shaped cluster, amylamine was added to in-
crease the amount of hydrophobic functional groups and tail
groups, thereby increasing the amphiphilicity [23] and in-
creasing the thermodynamic phase stability.
To a solution of octylamine and amylamine, the prescribed
volume of aqueous Pd2+ was added. After mechanical agita-
tion for 5 min, the solutions were allowed to equilibrate for
1 h. Hydrazine hydrate was then added and the solution ag-
itated. After equilibrating for 20 h in ethanol, a volume of
ethanol equal to the volume of octylamine was added to de-
compose the bilayer. The resulting black powder was allowed
to settle and the clear solution was decanted. Ethanol was
again added to ensure that all organics were removed. This
wash process was repeated four times and the sample was al-
lowed to dry in air. Morphologies of the resultant particles are
dependent on the ratio R = [water]/[surfactant]. Utilizing the
phase diagram for octylamine and water [24], the synthesis
conditions were chosen to produce the desired morphologies.
In the lamellar bilayer phase region (at 25 ◦C, R < 10.7 but
R > 3.5) platelets were produced, and their thickness and face
size being determined by R. Regions outside of the lamellar
phase region (at 25 ◦C, R < 16 but R > 10.7) gave samples of
mixed morphologies, consisting of both platelets and spheres.
Polarized light microscopy was used to verify that the bire-
fringement bilayer phase was present when desired.
2.4. Hydrogen uptake measurements
Isotherms of hydrogen uptake were measured gravimetri-
cally on 40 mg samples using a high-pressure thermogravi-
metric analyzer (TGA Model #003 Hiden Inc.). Measure-
ments were made primarily at 50 ◦C and in the pressure range
0–10 bar. The weight uptake data were corrected for buoy-
ancy of the gas at all pressures. Ultra-high purity hydrogen
gas (99.999% purity) was used for the uptake studies. Vac-
uum decontamination of the sample reactor and the adjoin-
ing pipe-work was carried out before and after each sam-
ple run to ensure that the system was free from traces of
gaseous impurities. To ensure that the admitted hydrogen gas
was free from oxygen and trace impurities, the hydrogen gas
was passed through an oxygen/moisture trap (Megasorb Gas
Purifier, Supelco Inc.), before being admitted into the sam-
ple reactor. The uptake isotherms for bulk palladium (purity
99.9%+, supplied by Aldrich) were verified using the instru-
ment and the results were found to be in good agreement with
reported results in literature [1].
3. Results and discussion
3.1. Material characterization
Following the synthesis and washing, all palladium
nanoparticles were black, free-flowing powder. Powder X-
ray diffraction (XRD) showed a crystalline Pd phase for all
morphologies (Fig. 1). XRD of the spherical particles synthe-
sized by means of reverse micelle approach initially showed
crystalline PdO as a minor constituent. However, upon wash-
ing with hydrazine hydrate, only pure crystalline Pd was ob-
2.3. Other sample characterization
Powder X-ray patterns were obtained using a Sin-
tage diffractometer using Cu K␣ radiation (of wavelength
˚
1.5418 A) and a Peltier-cooled detector. Samples were placed
onquartzzero-backgroundslides. Electronmicrographswere
obtained using a Phillips 420 transmission electron micro-
scope (TEM) and a JOEL 2010F HRTEM. Energy-dispersive
spectroscopy (EDS) was carried out using the Phillips TEM.
Samples for TEM were dispersed in ethanol and deposited
on Fomvar supported Cu grids. Platelet dimensions were de-
termined by atomic force microscopy (AFM) (MultiMode,
Digital instruments) using the tapping mode. Samples for
Fig. 1. Powder X-ray diffraction of nano-Pd sample #2-65-10. The X-ray
diffraction pattern of the sample matches that of standard Pd h k l’s are given.

S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
237
Table 1
inFig. 2b), boththefacesize(a)andthickness(t)canbedeter-
mined. The aspect ratio (A = a/t, where a is the particle diam-
eter and t is the platelet thickness) in the range 30 ≤ A ≤ 100
was obtained with values of t, 1 ≤ t ≤ 3 nm (Table 1).
TEM images (Fig. 3) of the platelet faces were found to be
in good agreement with the AFM results. Initially, all platelet
samples on the Cu TEM grids were highly agglomerated and
surrounded by organic byproducts. However, when the elec-
tron beam heated the particles, the residual organics appar-
ently volatilized or reacted, breaking up the agglomerates
and nicely dispersing the platelets in the field of view. EDX
of individual platelets was conducted and re-confirmed the
formation of nanoplatelet Pd (Fig. 4).
In some cases, TEM images of samples synthesized by
the bilayer approach exhibited a mix of platelet and spherical
morphologies; these are referred to here as the “hybrid” mor-
phology. N2-BET experiments on the hybrid samples showed
that they exhibit more tolerance to loss of BET surface area
upon heating. Results of these observations will appear else-
where [25]. Particle sizes of the spherical particles produced
by reverse micelle were determined by HRTEM (Fig. 5). The
mean particle size observed from TEM studies is consistent
with results from XRD line broadening. Lattice fringes can
be observed in some of the particles, consistent with their
single crystal nature. From Fig. 5, it can also be observed
that sintering between particles is minimal.
Selected data for nano-Pd samples studied for hydrogen uptake properties
Sample
R
Pd (%) d (nm)b a (nm)c t (nm)d
η
Reverse micelle
2-35 Spherical
2-71 Spherical
2
8
95.34
96.34
4
5
1.05
1.04
Bilayer
2-65-7 Platelet
2-65-10a Mixed 10 87.26
7
94.69
82
1.3
1.5
2.8
2.0
1.06
1.15
109
327
70
3-18 Platelet
8
96.80
1.03
R = [water]/[surfactant] and η = Pd weight correction factor determined by
TGA and used in H-adsorption analysis.
a
Mixture of both spherical and platelet morphology particles, with
platelet particles having dimensions in the range (109 nm × 1.5 nm) to
(327 nm × 2.8 nm).
b
Diameter of spherical particle.
Face size of platelet particle.
Thickness of platelet particle.
c
d
served in the XRD. Particle sizes for spherical particles were
determined from line broadening using Scherrer’s equation
(Table 1), and they were confirmed by TEM studies.
AFM, in the tapping mode, was used to obtain details on
the platelet thickness and face size. The samples were dis-
persed in ethanol and a small drop placed on freshly cleaved
mica. Two-dimensional images of the samples are shown in
Fig. 2. Through an examination of the Z-axis profile (Z-scans
Fig. 2. Atomic force microscopy of sample #2-65-7 primarily containing Pd platelets. Part (a) shows a 2 ␮m × 2 ␮m area image taken in the tapping mode. In
(b), we identify a smaller region of the image in (a) containing ∼8 particles. The solid line in (b) is analyzed in (c) to determine the aspect ratio for a single
platelet. The thickness (1 nm) and the diameter (80 nm) can be determined through examination of the Z-axis.

238
S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
Fig. 3. TEM of platelet palladium particles (sample #3-18) synthesized by lamellar bilayer technique.
The N2-BET SSA versus temperature/time studies were
heating to 100 ◦C and holding for 15 min, the surface area
dropped from 80 to 10 m2/g. Bilayer-synthesized samples
were found to be noticeably more robust against sintering,
e.g., the SSA remains constant for 25 h at 150 ◦C, and then
decreases less rapidly with time on increasing the sample
temperature to 200 ◦C for 2 h. From these studies, we de-
veloped a protocol for activation in hydrogen in the high
pressure TGA, i.e., all samples were heated only to 50 ◦C
for 6 h in vacuum to remove residual organics, as the low
temperature was required to retain the SSA (an exception
done under vacuum (30 m Torr) to determine the characteris-
tic sintering temperature for the Pd nanoparticles, as marked
by a sudden loss in BET SSA. The nanoparticulate samples
were successively heated from room temperature to 200 ◦C
in 50 ◦C increments, and maintained at each temperature for
the various times indicated. The samples were then cooled
and SSA measurements were carried out. For spherical par-
ticles synthesized by the reverse micelle method, the SSA
remains constant at a temperature of 50 ◦C for 32 h. Upon
Fig. 4. Energy-dispersive X-ray spectroscopy of the nano-Pd platelet sample #3-18 shown in the inset. The elements Cu and C are both present in the
carbon-coated copper grid.

S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
239
Fig. 5. HRTEM of spherical particles synthesis by reverse micelle technique. Lattice fringes show that the particles are crystalline in nature. It can also be
observed that sintering between particles is minimal.
was sample #NPd 3-18, which was heated to 150 ◦C for
6 h).
1000 ◦C in O2 (7 ◦C/min). At 200 ◦C, all volatile organic by-
products were removed, and at ∼300 ◦C the Pd is oxidized
to PdO [26], increasing the mass, as shown in the figure. As
the sample temperature reaches 850 ◦C, the PdO is sponta-
neously reduced to Pd. Above 850 ◦C, pyrolyzed carbons are
removed as CO. We conclude that, at 1000 ◦C, the sample
is primarily pure Pd. The TGA data proved to be a valuable
tool to determine the weight fraction of Pd in these high SSA
materials. The ratio between the initial and final weight of
the sample in the TGA was used as a correction factor (η)
in hydrogen uptake studies to adjust the H-uptake data to the
true Pd content (Table 1). The correction factor was typically
∼1.03–1.06.
Thermogravimetric analysis of our nanoparticulate sam-
plesinflowingO2 wascarriedouttodeterminethepercentage
of Pd in the sample (Fig. 6). The materials have large SSA
and therefore, surface oxides and residual organics can be
significant components of the sample mass not involved in
hydrogen storage. Washing with hydrazine resulted in sam-
ples containing 90–96% Pd, the balance being surface con-
taminants. SSA values for these samples range from 50 to
80 m2/g. The results of a typical TGA analysis in flowing O2
are presented in Fig. 6. The samples were heated slowly to
3.2. Hydrogen uptake studies
Since the Pd nanoparticles studied exhibit a loss in surface
area on heat treatment above 100 ◦C (with the exception of
some platelet and hybrid morphology Pd nanoparticulates),
all hydrogen uptake isotherms were recorded at a tempera-
ture of 50 ◦C after outgassing in vacuum at 50 ◦C for 1–2 h.
This degassing was done in situ in the TGA, followed by
isothermal adsorption data collection.
3.2.1. Nanoparticles with spherical morphology
Fig. 7A shows the pressure-composition isotherms of the
spherical and the platelet samples with the pressure axis plot-
ted on a logarithmic scale. Fig. 7B shows the low-pressure
data in the pressure interval 0–800 mbar with the pressure
axis on a linear scale. The hydrogen adsorption isotherms
Fig. 6. TGA of platelet sample #3-33. Initial weight of sample was
9.7621 mg (A). The weight is observed to decrease slowly on heating to
a temperature of ∼200 ◦C, above which the weight is observed to increase,
probably due to the formation of PdO. At ∼800 ◦C, weight loss was again
seen leaving pure Pd 9.219 mg (B) at 1000 ◦C. Sample is 94.4% Pd.

240
S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
with that of bulk Pd. The total weight percent uptake at the
highest pressure of 10 bar for the spherical nanopowders is
less than that for the bulk Pd sample. However, the platelet
morphology particles are observed to store more than the bulk
(see the following). (iii) We note that the pressures P␣ and P␤
in curve (a) marking respectively the roll-over and the upturn
in the isotherm, appear to depend on the particle morphology
(spherical or platelet).
As mentioned earlier, the experimental findings of large
hydrogen solubility in the solid solution region of nanoscale
clusters of Pd [14,15] have been attributed to hydrogen oc-
cupancy in subsurface sites in addition to the occupancy in
interior octahedral sites in the lattice. The considerable in-
crease in surface area of these materials can open up a large
fraction of surface region sites for hydrogen occupancy. For
estimates of the fraction of these sites in our samples, see
Table 2. These sites supplement the regularly available oc-
tahedral sites within the bulk (bulk sites). Besides theoreti-
cal modeling studies, experimental investigations by various
groups [28–32] employing thermal desorption spectroscopy
(TDS) and electron energy loss spectroscopy (EELS) tech-
niques have yielded strong evidence in support of the exis-
tence of surface and subsurface sites, with binding energies
lower than those for the octahedral interstitial sites. These
investigations also indicate that hydrogen has a greater ten-
dency to enter and leave subsurface sites at lower tempera-
tures than surface sites. These results imply that subsurface
sites play a more important role in hydriding–dehydriding
than surface sites. Previous reports have shown that subsur-
face sites have distinctly lower energies than the regular bulk
sites and hence, offer a distinct potential energy minimum
for hydrogen occupancy.
Srivastava and Balasubramaniam [33] have conducted
modeling of hydrogen solubility and metal–hydrogen in-
teractions in Pd clusters using statistical thermodynamics
and a two-site model of hydrogen occupancy. They ne-
glect the interactions between the hydrogen atoms in the
metal. Their calculations have demonstrated the feasibility
of Fermi–Dirac statistics in modeling metal–hydrogen inter-
actions and they have shown that the subsurface site energies
are lower than the regular (interior) octahedral site energies.
Furthermore, they have found that the subsurface sites are
preferentially occupied by hydrogen in clusters when com-
pared with the situation in bulk Pd.
Fig. 7. (A) Hydrogen uptake isotherms for the nanoparticle samples com-
pared with that for bulk Pd sample in the pressure range 0–10 bar (pressure
axis on log scale): (a) bulk Pd, (b) spherical nano-Pd 2-35-2 (4 nm), (c)
spherical nano-Pd 2-71-8 (5 nm), (d) platelet nano-Pd 3-18 (70 nm × 2 nm),
(e) platelet nano-Pd 2-65-7 (90 nm × 1.15 nm) and (f) hybrid nano-Pd 2-65-
10 (109 nm × 1.5 nm). (B) Hydrogen uptake isotherms in the low-pressure
region (0–800 mbar) with pressure axis on a linear scale for the nanoparti-
cle samples compared with that for bulk Pd sample (a). The letters (a)–(f)
identifying the samples have the same meanings as in (A).
recorded using our TGA show excellent reproducibility. The
desorption isotherms for nano-Pd exhibited non-repeatable
hysteresis effects that are not understood. This behavior was
not observed for bulk Pd.
The hydrogen adsorption pressure-composition isotherms
for our nano-Pd materials (Fig. 7) show the general features
expected for bulk palladium [1]. That is, all the isotherms
show three distinct regions, which in the bulk have been
identified with the ␣-phase (indicated by the dash-dotted line
curve (a)—the solid solution phase), the (␣ + ␤)-phase (in-
dicated by the dashed curve (a)—the constant pressure two-
phase region) that marks the miscibility gap, and finally at the
highest pressures, the ␤- or the hydride-phase, indicated by
the dotted curve (a). In particular, the hydrogen adsorption
isotherms for the nanoparticles with spherical morphology
show the following distinct characteristics: (i) The hydrogen
solubility in the ␣-phase region (solid solution region) of the
isotherms appears to be enhanced significantly when com-
pared with that in bulk Pd [27]. (ii) The plateau region of the
isotherm (the constant pressure two-phase field) acquires a
positive slope and is considerably narrowed when compared
In our work, we investigate the hydrogen storage proper-
ties of spherical and platelet Pd nanoparticles. Therefore, the
conclusions from the reported work on clusters [14,15,33]
on preferential occupancy of subsurface sites in nanosized
Pd clusters might be expected to apply to our larger nanopar-
ticles. In fact, we also conclude that the enhanced hydrogen
solubility in the ␣-phase of the isotherms for our nanopar-
ticles is due to preferential occupancy of hydrogen in the
subsurface sites.
In bulk Pd, a transition from the ␣-phase (solid solution
region) to the two-phase plateau field occurs after hydro-
gen occupies all the available potential energy minima sites

S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
241
Table 2
Estimates of the fraction of subsurface sites available in the nano-Pd samples
Pd nanoparticle studied
Estimated fraction of bulk-like sites
[x2 − x1]nano/[x2 − x1]bulk (%)
Estimated fraction of
subsurface sites (%)
Spherical nano-Pd (2-35-2) (∼4 nm)
31
32
55
60
51
69
68
45
40
49
Spherical nano-Pd (2-71-8) (∼5 nm)
Platelet nano-Pd (3-18) (∼70 nm × 2 nm)
Platelet nano-Pd (2-65-7) (∼82 nm × 13 nm)
Hybrid nano-Pd (2-65-10) (∼109 nm × 1.5 nm)
(bulk octahedral sites) to form a solid solution with the metal.
Since the site energy of the subsurface sites available in Pd
nanoparticles is clearly less than that of the bulk sites, it can be
assumed that hydrogen occupying the subsurface sites does
not result in the formation of a hydride phase. Furthermore,
the upper boundaries of the plateau regions in the uptake
isotherms for the Pd nanoparticle and the bulk are located at
similar pressures, indicating that the energies of the bulk-like
interior sites in all the nanoparticles are of comparable mag-
nitude. Based on the above assumptions, a naive calculation
of the fraction of the subsurface sites available for hydrogen
occupancy can be carried out from the knowledge of hydro-
gen concentrations that define the boundaries of the plateau
region (two-phase field), as suggested by Sachs et al. [15] for
Pd clusters.
The positive slope of the nanoparticle isotherm may indi-
cate a distribution in binding energies of the core sites (the
lower energy sites being filled up at lower pressure). From
pressure-composition isotherm measurements and lattice gas
model calculations carried out on Ti–H and Ti94Al6–H sys-
tems, Naito et al. [34] have explained the occurrence of slop-
ingplateausonthebasisofadistributioninthebindingenergy
of the sites occupied by hydrogen atoms in metals. Measure-
ments carried out by Sachs et al. [15] on free clusters of Pd
with a narrow size distribution also show a significant posi-
tive slope, although they do not associate this with the width
of the cluster size distribution.
3.2.2. Particles with platelet and hybrid morphology
If we denote the hydrogen concentration x in PdHx at the
beginning of the plateau region as x1 (the highest concentra-
tion at which the ␣ (solid solution)-phase alone is present),
and the hydrogen concentration at the end of the plateau re-
gion as x2 (the lowest hydrogen concentration at which the
(␣ + ␤)-phase is present), the fraction of the number of sub-
The isotherms for samples with platelet morphology
(Fig. 7) exhibit different behavior when compared with
the spherical particles. One platelet sample [70 nm × 2 nm
(#NPd 3-18)] shows a behavior similar to that of the
spherical nanoparticles. However, another platelet sample
[82 nm × 1.5 nm (sample #NPd 2-65-7)] shows an upshift
of the entire (narrow) miscibility gap to higher concentra-
tions (x). This behavior is also seen in hybrid sample #NPd
2-65-10 (material comprising of both spherical and platelet
nanoparticles). The significant shift in the value of x1(nano)
that is observed (Fig. 7(A)) in hybrid morphology samples
can be attributed to the quenching of the surface sites by
residual organics in the sample. However, further work would
be needed to substantiate this argument. Changes in particle
morphology could result in changes in the fraction of ener-
getically favorable subsurface sites as well as the hydrogen
solubility in these sites due to changes in chemical potential.
Post-synthesis heat treatments of the platelet and hybrid mor-
phology samples revealed a ∼10% increase in surface area
with increasing temperature (not found in spherical morphol-
ogy samples), even when heated from room temperature to
50 ◦C. Of course, an increase in surface area is expected to
bring about an increase in the fractional population of sur-
face and subsurface sites. Based on the above arguments, we
propose that the significant increase in hydrogen solubility
in the ␣-phase and the significant upshift in the narrowed
plateau region and x2(nano) in platelet Pd are probably due to
the combined effects of changes in the particle morphology
as well as surface area of the particles. Similar effects have
been reported from isotherm measurements on 5 nm clusters
of Pd by Sachs et al. [15].
surface sites Nsubsurface to the total number of sites Ntotal
,
available in the nanoparticles can be approximated by the
relation:
Nsubsurface
[x2 − x1]nano
[x2 − x1]bulk
= 1 −
(1)
Ntotal
We use Eq. (1) as a relationship to quantify the changes in
the plateau (two-phase field) brought about by the nanopar-
ticle morphology. From (1), for the spherical nanoparticles,
it is seen that the fraction of subsurface sites is quite high
(∼70%). The fact that the subsurface sites outnumber the
bulk-like sites and that the subsurface sites have significantly
lower site energies when compared with the bulk sites is con-
sistent with the pronounced increase in the ␣-phase hydro-
gen concentrations in the nanoparticles studied. As such, it
is probable that the onset of the two-phase field (nucleation
of the hydride- or ␤-phase) proceeds mostly through the oc-
cupancy of a relatively smaller number of interior bulk-like
sites in the case of the particles. This explains the narrowing
of the miscibility gap (length of the plateau). The filling of
the core bulk sites then occurs once all the subsurface sites
have been saturated. The above arguments also support the
observation that the values of x2(nano) shift towards lower val-
ues in the spherical particles, as compared with the situation
in bulk Pd.

242
S. Kishore et al. / Journal of Alloys and Compounds 389 (2005) 234–242
[11] P.G. Sanders, J.R. Weertman, J.G. Barker, R.W. Siegel, Scr. Metall.
4. Summary and conclusions
29 (1993) 91.
[12] R.J. Wolf, M.W. Lee, J.R. Ray, Phys. Rev. Lett. 73 (1994) 557.
[13] D.S. Sibirtsev, A.V. Skripov, N. Natter, R. Hempelmann, Solid State
Commun. 108 (1998) 583.
[14] A. Pundt, C. Sachs, M. Winter, M.T. Reetz, D. Fritsch, R. Kirchheim,
J. Alloys Compd. 293–295 (1999) 480.
[15] C. Sachs, A. Pundt, R. Kirchheim, M. Winter, M.T. Reetz, D. Fritsch,
Phys. Rev. B 64 (2001) 075408.
[16] A. Zaluska, L. Zaluski, J.O. Strom-Olsen, Appl. Phys. A Mater. Sci.
Process. 72 (2001) 157.
[17] U. Stuhr, H. Wipf, T.J. Udovic, J. Weissmuller, H. Gleiter, J. Phys.
(Condens. Matter) 7 (1995) 219.
[18] R. Djaozandry, J. Huot, S. Boily, J. Goyette, T.K. Bose, R. Schulz,
J. Alloys Compd. 305 (2000) 264.
[19] Z. Dehouche, R. Djaozandry, J. Goyette, T.K. Bose, J. Alloys
Compd. 288 (1999) 312.
[20] T. Li, J. Moon, A.A. Morrone, J.J. Mecholsky, D.R. Talham, J.H.
Adair, Langmuir 15 (1999) 4328.
Crystalline palladium nanoparticles have been synthe-
sized with spherical platelet and mixed sphere platelet hy-
brid morphology. The nanoparticles have been characterized
for their structure through powder X-ray diffraction, for their
surface area behavior with temperature through BET surface
area measurements, and for their morphology through TEM
and AFM techniques. The increase in uptake of hydrogen
by the Pd nanoparticles in the solid solution region of the
isotherms and the decrease in the width of the miscibility
gap, relative to bulk Pd, are attributed to the preferential oc-
cupancy of hydrogen in surface and subsurface sites at lower
pressures. The positive slope of the plateau in the isotherms
has been identified with a distribution of site energies, in the
case of the Pd nanoparticles.
[21] D.O. Yener, J. Sindel, C.A. Randall, J.H. Adair, Langmuir 18 (2002)
8692.
[22] J.H. Adair, T. Li, T. Kido, J. Moon, J. Mecholsky, A. Morrone,
D.R. Talham, M.H. Ludwig, L. Wang, Mater. Sci. Rep. R23 (1998)
139–242.
Acknowledgements
The authors would like to acknowledge Honda R&D,
Japan, for supporting this program of research. They grate-
fully thank J. Fujiwara, H. Goto and T. Furuta for many stim-
ulating discussions during the course of this work.
[23] K.R. Wormuth, E.W. Kaler, J. Phys. Chem. 91 (1987) 611.
[24] J. Sjoblom, P. Stenius, in: Mercel Dekker (Ed.), Surface Science
Series, vol. 23, New York, 1987. p. 398.
[25] J.A. Nelson, S. Kishore, J.H. Adair, P.C. Eklund, Langmuir (submit-
ted).
[26] S.E. Wang, J.P. Dougherty, W. Huebner, J.G. Pepin, J. Am. Ceram.
Soc. 77 (1994) 3051.
[27] It should be noted here that not much of the ␣-phase is observable in
the isotherms in the low-pressure limit of the TGA. A small amount
of the ␣-phase has already been crossed with the onset of the first
pressure point and is therefore, suppressed in the isotherms.
[28] R.J. Behm, K. Christmann, G. Ertl, Surf. Sci. 99 (1990) 320.
[29] R.J. Behm, V. Penka, M.-G. Cattania, K. Christmann, G. Ertl, J.
Chem. Phys. 78 (1983) 7486.
[30] H. Okuyama, W. Siga, N. Takagi, M. Nishijima, T. Aruga, Surf. Sci.
401 (1998) 344.
[31] K.H. Rieder, M. Baumberger, W. Stocker, Phys. Rev. Lett. 51 (1983)
1799.
[32] K. Christmann, Mol. Phys. 66 (1989) 1.
[33] V. Srivastava, R. Balasubramaniam, Mater. Sci. Eng. A 304–306
(2001) 897.
[34] S. Naito, M. Yamamoto, M. Doi, M. Kimura, J. Chem. Soc. Faraday
Trans. 91 (1995) 4143.
References
[1] F.A. Lewis, The Palladium–Hydrogen System, Academic Press, Lon-
don, 1967.
[2] T.B. Flanagan, W.A. Oates, Annu. Rev. Mater. Sci. 21 (1991) 269.
[3] E. Wicke, H. Brodowsky, in: G. Alefeld, J. Volkl (Eds.), Hydrogen
in Metals, vol. II, Springer-Verlag, Berlin, 1978, p. 73.
[4] M.S. Daw, M.L. Baskes, Phys. Rev. B 29 (1984) 6443.
[5] T. McMullen, M.J. Scott, E. Zaremba, Phys. Rev. B 35 (1987) 1076.
[6] Y. Fukai, The Metal–Hydrogen System-Basic Bulk Properties,
Springer-Verlag, Berlin, 1992.
[7] T. Kuji, Y. Matsumura, H. Uchida, T. Aizawa, J. Alloys Compd.
330–332 (2002) 718.
[8] T. Mutschele, R. Kirchheim, Scr. Metall. 21 (1987) 135.
[9] T. Mutschele, R. Kirchheim, Scr. Metall. 21 (1987) 1101.
[10] J.A. Eastman, L.J. Thompson, B.J. Kestel, Phys. Rev. B 48 (1993)
84.