The interaction of hydrogen with internally oxidized Pd alloys as illustrated by Pd-Fe alloys

Article abstract of DOI:10.1016/S0925-8388(03)00164-6

Hydrogen interactions in internally oxidized Pd alloys have been illustrated using Pd-Fe alloys. Internal oxidation of Pd-Fe alloys results in composites consisting of fine dispersions of nano-sized Fe₂O₃ precipitates in a matrix of pure Pd. Hydrogen solubility measurements can be used as a probe to characterize the reversible and irreversible trapping at the metal-oxide interfacial regions. The effect of hydriding and dehydriding (cycling) on solubilities has been characterized. Hydrogen isotherms can be used to characterize the extent of internal oxidation, and understand the effect of the dispersed internal oxides on hysteresis and H capacity.

Full text of DOI:10.1016/S0925-8388(03)00164-6

Journal of Alloys and Compounds 356–357 (2003) 3–8
L
www.elsevier.com/locate/jallcom
The interaction of hydrogen with internally oxidized Pd alloys as
illustrated by Pd–Fe alloys
D. Wanga, T.B. Flanagana, R. Balasubramaniamb ,
*
aUniversity of Vermont, Department of Chemistry and Materials Science Program, Burlington, VT 05405, USA
bDepartment of Materials and Metallurgical Engineering, Indian Institute of Technology, Kanpur 208016, India
Received 2 July 2002; accepted 20 November 2002
Abstract
Hydrogen interactions in internally oxidized Pd alloys have been illustrated using Pd–Fe alloys. Internal oxidation of Pd–Fe alloys
results in composites consisting of fine dispersions of nano-sized Fe2O3 precipitates in a matrix of pure Pd. Hydrogen solubility
measurements can be used as a probe to characterize the reversible and irreversible trapping at the metal–oxide interfacial regions. The
effect of hydriding and dehydriding (cycling) on solubilities has been characterized. Hydrogen isotherms can be used to characterize the
extent of internal oxidation, and understand the effect of the dispersed internal oxides on hysteresis and H capacity.
2003 Elsevier B.V. All rights reserved.
Keywords: Hydrogen; Hydriding; Pd alloys; Internal oxidation; Characterization
1. Introduction
2. Interaction of hydrogen with internally oxidized Pd
alloys
Upon high temperature exposure ($700 8C) of a Pd–M
alloy to an oxidizing environment, oxygen dissociates at
the surface and O atoms dissolve and diffuse to internally
oxidize M, where M is a solute more easily oxidizable than
Pd, to produce oxide precipitates in a matrix of pure Pd.
This is an example of internal oxidation (IO) [1]. As these
precipitates are generally nano-sized, the material pro-
duced after IO can be termed as nano-oxide dispersed
composites. Hydrogen can be used as a probe to help
characterize these composites [2–4]. Hydrogen behavior in
internally oxidized systems can be monitored either by
electrochemical [2] or gas phase [5] methods. Each of
these has its advantages. While the electrochemical tech-
nique is sensitive to hydrogen chemical potential changes
at very low H contents, the gas phase charging method
additionally provides information about the results of
hydriding and dehydriding (cycling) of the Pd matrix.
Kirchheim and co-workers utilized the electrochemical
technique to charge hydrogen and study hydrogen trap-
ping, hydrogen solubilities and diffusion in internally
oxidized alloys like Pd–Al, Pd–Mg, Pd–Zr and Pd–Zn
[2–4].
Two important variables in IO (temperature and time)
control the extent of internal oxidation. IO progresses by
the movement of the IO front from the surface into the
material. This movement is limited by the permeation of
oxygen from the surface to the IO zone front. In case the
IO front has not fully penetrated the material, partial
internal oxidation results, wherein the region from the
surface inward is composed of nano-sized oxides in a
matrix of pure metal, e.g. Pd, while the inner region will
be the unoxidized alloy. The hydrogen solubilities in these
regions differ and therefore H solubilities can be used to
probe these differences. Noh et al. [5] have shown that
after complete IO of Pd–Al alloys, the hydrogen isotherm
corresponds to one of pure Pd, while after partial IO, the
H2 isotherm consists of two parts: one corresponding to
pure Pd–H and the other to the unoxidized Pd–Al alloy.
Therefore hydrogen isotherms can be used to determine the
fraction of IO. Hydrogen isotherms following the IO of
ternary Pd alloys revealed several additional interesting
features [6].
The thermodynamic stability of the internal oxide and its
interaction with hydrogen is also of interest. The oxide that
forms in Pd–Al alloys is alumina [7,8], in Pd–Y alloys it is
yttria [9], and a mixed oxide PdRhO2 forms in Pd–Rh
*
Corresponding author.
E-mail address: bala@iitk.ac.in (R. Balasubramaniam).
0925-8388/03/$ – see front matter
doi:10.1016/S0925-8388(03)00164-6
2003 Elsevier B.V. All rights reserved.

4
D. Wang et al. / Journal of Alloys and Compounds 356–357 (2003) 3–8
alloys [10,11]. A thermodynamically more stable oxide
results in smaller precipitate sizes [1]. SANS (small angle
neutron scattering) measurements revealed a smaller mean
particle size in internally oxidized Pd–Al alloys [12] when
compared to Pd–Rh alloys [11]. Moreover, a very stable
oxide like Al2O3 will not be reduced by the dissolved
hydrogen whereas a less stable oxide, such as PdRhO2,
will be reduced with H2 (573–623 K) forming PdRh
precipitates within the Pd matrix [10,11]. Therefore,
hydrogen reduction of internal oxides is a technique for
segregating components of a substitutional solid solution
binary alloy [10,11].
In internally oxidized Pd–M alloys, a large interfacial
area between the internal oxides and Pd matrix is created,
i.e. several m2 per cm3 composite, which will affect
hydrogen solubilities. For example, in the very low
hydrogen content region, hydrogen is trapped both irrever-
sibly and reversibly by the interface [2–4]. The irrever-
sibly held hydrogen in Pd–Al alloys was believed to be
bonded to unsaturated oxygen atoms at the interface
forming an interfacial monolayer, which could be removed
by evacuation at elevated temperature, $573 K [3]. There
was a solubility enhancement relative to Pd in the dilute
phase region; this H is readily removed by evacuation at
323 K [13]. This was shown to originate from the stress
fields of the precipitates [8].
Metallurgical aspects of dislocation formation and
stabilization in composite materials can be probed by
hydrogen. Hydrogen dissolved in metals segregates to the
tensile stress field of dislocations and this segregation leads
to readily measurable H2 solubility enhancements in
deformed Pd [12]. H2 solubility enhancements serve as a
convenient way to monitor relative dislocation densities.
Larger dislocation densities were observed using dilute
phase H2 solubilities in Pd/Al2O3 composites compared to
Pd after cycling through the hydride phase and after
subsequent annealing [14]. Moreover, significantly higher
temperatures are needed to remove the dislocations intro-
duced by cycling a Pd0.97Al0.03 alloy after IO at 1073 K
than pure Pd. The annealing temperature needed to remove
dislocations in the alloy after IO at 1273 K was compar-
able to Pd. Therefore, H2 solubility measurements reveal
that closely spaced precipitates resulting from IO at 1073
K stabilize the dislocation array produced by cycling,
whereas the larger and further spaced precipitates formed
after IO at 1273 K do not significantly stabilize the
dislocations [13]. This is in agreement with what is known
about dispersion hardening by internal oxidation [1].
constant of dissolved oxygen in Pd [15]. The slow step in
IO is believed to be the diffusion of dissolved oxygen and
therefore the movement of the IO front allows determi-
nation of the oxygen diffusion constant. The Pd–Fe alloys
met this criterion of a well-defined internal oxidation front
with the diffusion constant of Fe quite small in comparison
to that of oxygen [15].
4. Experimental
The Pd–Fe alloys were prepared by arc-melting the pure
elements under argon. The buttons were flipped and re-
melted several times. They were then annealed in vacuo
for 3 days at 1133 K, rolled into foils and then re-annealed
for 2 days at 1133 K. The alloys were then oxidized in air
at several different temperatures from 973 to 1273 K for
various lengths of time. They were quenched after internal
oxidation in order to avoid the oxidation of Pd.
5. Results and discussion
5.1. Material characterization
The microstructure after IO and hydrogen interactions
can be characterized by a wide variety of techniques. After
internal oxidation at 1073 or 1098 K, weight gain measure-
ments indicate that the internal oxide possesses a stoi-
chiometry close to Fe2O3, as reported by Gegner [15] after
internal oxidation at 1273 K. In the case of Pd–Fe alloys,
therefore, the internal oxide is independent of the internal
oxidation temperature.
Scanning electron microscopy (SEM) shows that pure
Pd nodules form on the surface after IO, resulting from the
compressive stresses that are generated within the Pd
matrix during internal oxide formation [16,17]. Upon
hydrogen cycling,
a considerable number of grain
boundaries were cracked in the case of Pd–Al alloys [18],
while such cracks were not observed for Pd–Fe alloys. The
cracked grain boundaries in internally oxidized Pd–Al
alloys exhibit a honeycomb-like appearance and there is an
increase in surface area after IO and cycling. This results
in enhanced rates of hydrogen absorption [18].
Transmission electron microscopy (TEM) reveals the
shape of the internal oxides, their distribution and disloca-
tion densities. Pd0.99Fe0.01 alloy samples, internally oxi-
dized at 1063 and 1273 K, were examined in a TEM. Both
specimens showed large dislocation densities with the
lower oxidation temperature having a greater density. This
differs from the TEM of internally oxidized Pd–Al alloys
where few dislocations were observed [7,8]. Dislocations
are created, after IO, due to relaxation of some of the
thermal residual stresses that are generated on cooling the
material after internal oxidation. The elastic constants
(elastic modulus and Poisson’s ratio) for iron oxides are
lower than that of alumina [19]. The differences in elastic
3. Pd/Fe2O3 composites
In the present communication, the interaction of hydro-
gen in the internally oxidized Pd alloys would be illus-
trated with Pd–Fe alloys. Internal oxidation of Pd–Fe
alloys has been recently utilized to determine the diffusion

D. Wang et al. / Journal of Alloys and Compounds 356–357 (2003) 3–8
5
properties of the oxides are reflected in the relaxation of
part of the thermal stresses in internally oxidized Pd–Fe
alloys by dislocation punching, as evidenced by the higher
dislocation densities in the internally oxidized Pd–Fe
alloys compared to Pd–Al alloys. The magnitude of the
thermal stresses is determined by the size, shape and
distribution of the internal oxides formed, and the differ-
ence in thermal expansion coefficients of the oxides and
the Pd matrix [8]. Electron diffraction in the TEM provides
the structure of the internal oxide. In Pd–Fe alloys, the
crystalline precipitates were identified as Fe2O3. Only very
few studies have addressed microstructural features of the
Pd matrix after cycling, which leads to an increase in the
dislocation densities [8].
measurements for an internally oxidized Pd0.97Al0.03 alloy
provided similar numbers [13] while in the case of Pd–Rh
alloys, the sizes were larger [11]. The estimated volume
fractions (2.9% for Pd0.981Fe0.019 and 5.2% for
Pd0.963Fe0.037) agreed well with the fraction expected if all
of the Fe had been oxidized. The trapping sites for H at the
interface can be calculated from the Sv values. This gives
5.231014 and 7.431014 sites/cm2 for Pd0.981Fe0.019 and
Pd0.963Fe0.037, respectively, which are reasonable.
5.2. Dilute phase H2 solubilities
Dilute phase H2 solubilities reveal irreversible as well as
reversible trapping tendencies. As an example, the H2
SANS measurements were performed for the
Pd0.981Fe0.019 and Pd0.963Fe0.037 alloys after internal oxida-
tion at 1093 K. The alloys showed increased scattering
after internal oxidation with the Pd0.963Fe0.037 alloy having
a greater scattering than the Pd0.981Fe0.019 alloy, as ex-
pected. Scattering data in absolute units was collected over
solubilities in internally oxidized Pd0.963Fe0.037 are plotted
1 / 2
as p
against r in Fig. 1. The initial and repeat dilute
H2
phase solubilities were determined before formation of the
hydride phase. The initial H2 solubilities after internal
oxidation have positive intercepts along the r5H/M axis
indicating irreversible trapping of H at/or near the Pd/
oxide interface. In internally oxidized Pd–Fe alloys, the
intercepts are nearly directly proportional to the Fe con-
tent. After evacuation at 323 K for 2 h, the repeat dilute
phase solubility passed through the origin but, otherwise,
had a similar shape as the initial solubility. The irrever-
sibly trapped H is not removed by this evacuation (323 K)
because, otherwise, the solubility relation would not have
passed through the origin.
The solubility enhancement observed in the dilute phase
in the internally oxidized alloy indicates reversible hydro-
gen trapping (Fig. 1) and is related either to thermal
residual stress or to dislocation densities from IO, but the
former is more likely because the dislocation densities in
the momentum transfer (q) range of 0.002 to 0.3 A21. The
˚
large Q data was fitted to Porod’s asymptotic relation:
Cp 5 limq→` q4I(q)
(1)
where Cp is the Porod scattering constant and I(q) is the
scattering in absolute units. The total surface area of the
Fe2O3 precipitates per unit volume of sample, Sv, can be
calculated from the Porod constant by:
Cp
pDr2
]
Sv 5
(2)
where Dr is the difference in scattering length density
between the Fe2O3 precipitates and the Pd matrix. The
estimated Sv values are 1.663105 cm21 for Pd0.981Fe0.019
and 3.693105 cm21 for Pd0.963Fe0.037. For all two phase
systems having uniform scattering length densities in each
phase, the volume fraction, f, can be determined from the
scattering invariant:
Qt
2p2Dr2
]
f 1 2 f 5
(3)
s
d
where the scattering invariant Qt is determined from the
integration of all the scattering by:
`
Qt 5
E
q2 dS/dV ? q dq
(4)
s
d
s d
0
Assuming the scattering is from a size distribution of
spherical particles, the volume weighted mean particle
radius is then Rv 5 3f/Sv. The mean particle radii were
similar for the two alloys, with the higher Fe content alloy
Fig. 1. Dilute phase solubilities (323 K) for the Pd0.963Fe0.037 alloy after
internal oxidation at 1098 K, 72 h. Dashed line is solubility for pure Pd.
The solubility measurements were performed in the following sequence:
(n) initial solubility; (s) repeat solubility after evacuation at 323 K for 2
h; (d) solubility after cycling the alloy.
exhibiting
a slightly smaller mean (10.4 nm for
Pd0.981Fe0.019 and 8.4 nm for Pd0.963Fe0.037). The current
q-range cannot probe particles larger than 50 nm. SANS

6
D. Wang et al. / Journal of Alloys and Compounds 356–357 (2003) 3–8
the internally oxidized Pd–Fe alloys before cycling are not
that large. For comparable IO conditions, the solubility
enhancements are higher for the internally oxidized Pd–Al
alloys compared to Pd–Fe alloys. The higher solubility
enhancements for the Pd–Al system must be related to the
higher residual thermal stresses present in the Pd/Al2O3
composites after internal oxidation, which is reasonable
because the thermal expansion coefficients of iron oxides
are nearly twice that of alumina [19] and hence much
closer to that of pure Pd.
The effect of the temperature of IO on H solubilities is
interesting (Fig. 2). The Pd0.926Fe0.074 alloy internally
oxidized at 973 K has a greater dilute phase solubility
enhancement relative to Pd than the alloy internally
oxidized at 1098 K, which was in turn greater than for the
alloy internally oxidized at 1273 K. Lower oxidation
temperatures result in smaller precipitates, around which
the thermal residual stresses are higher [8]. Moreover, the
dislocation densities are also higher after lower oxidation
temperatures for internally oxidized Pd–Fe alloys. A
consequence of the higher residual stresses and dislocation
Pd–Al alloys where all of the trapped H is removed (323
K) [13]. There is a nearly linear correlation between the
amount of H removed by cycling/evacuation and XFe. For
internally oxidized Pd–Al alloys the removal of the
trapped H by cycling/evacuation was suggested to be due
to dislocation movement during cycling and interaction
with the obstacles, the precipitates; this alters the traps
sufficiently to allow removal of the trapped H [13]. This
would suggest that in Pd–Fe alloys the more strongly held
H is situated within subsurface layers of the Fe2O3
precipitates and cannot be removed by dislocation motion.
There is a solubility enhancement after cycling for the
internally oxidized alloys with respect to annealed Pd,
which increases with XFe. This solubility enhancement is
due to dislocation formation by cycling the internally
oxidized alloys. However, relative to the un-cycled, inter-
nally oxidized alloys this is small and nearly independent
of XFe.
5.4. Complete H2 isotherms for internally oxidized alloys
densities
with
decreasing
temperature
is
the
The extent of IO can be followed by complete hydrogen
isotherms. An isotherm for a partially internally oxidized
Pd0.963Fe0.037 alloy is shown in Fig. 3 where two plateaux
are shown; the lower one corresponds to the internally
dilute→hydride transition is very gradual for the alloys
internally oxidized at the lower temperatures (Fig. 2).
5.3. Effect of cycling on the intercepts
The internally oxidized alloys were completely hydrided
(323 K) and evacuated (cycled). The re-measured dilute
phase solubilities are also shown in Fig. 1. The intercepts
are finite but smaller than the initial ones. This indicates
that some H is removed during cycling and evacuation
(323 K) but not all, in contrast to internally oxidized
Fig. 3. Isotherms (323 K) for partially internally oxidized Pd0.963Fe0.037
alloy (1073 K, 24 h). The open symbols are for absorption and the closed
symbols for desorption. The plateau pressures for pure Pd and unoxidized
Pd0.963Fe0.037 alloy are indicated by the dashed lines.
Fig. 2. Dilute phase solubilities (323 K) for the internally oxidized
Pd0.926Fe0.074 alloy after internal oxidation at (h) 973 K, (,) 1098 K and
(s) 1273 K.

D. Wang et al. / Journal of Alloys and Compounds 356–357 (2003) 3–8
7
oxidized portion (Pd–H) and the other to the unoxidized
6. Conclusions
portion of the Pd0.926Fe0.074 alloy.
For completely internally oxidized Pd–M alloys, iso-
therms reveal the effect of internal oxides on hysteresis
and capacity. The capacities are generally not affected
[5,9]. Complete H2 isotherms for the internally oxidized
(1098 K) Pd0.926Fe0.074 alloy are shown in Fig. 4. The
initial absorption isotherm is similar to that for the second
cycle of Pd but the second and third plateau pressures are
higher than the first. After the initial cycle where the
trapping occurs, the H capacities are the same as for Pd–H.
The desorption plateau is lower than that for Pd–H; a
lowering of desorption plateau pressure was found for all
of the internally oxidized Pd–Fe alloys with its extent
nearly directly proportional to XFe. Since the hydride
formation plateaux for cycles after the initial one are
higher than for Pd and the decomposition plateau lower,
hysteresis is slightly greater for the internally oxidized
Pd–Fe alloys. Interesting differences are found after IO for
Pd–Rh alloys, where the absorption plateau is greater than
for Pd [11]. After IO of Pd–Al alloys, the initial isotherms
are the same as Pd, except that differences appear after
cycling [13] and at high temperature, isotherms revealed
significant differences in the hysteresis behavior [20].
The interaction of hydrogen with nano-oxide dispersed
Pd composites, prepared by internal oxidation of Pd–M
alloys, has been reviewed. Internally oxidized Pd–Fe
alloys have been used as an example. Hydrogen can be
used as probe to characterize the reversible and irreversible
trapping at the metal–oxide interfacial regions by hydro-
gen solubility measurements. The effect of hydrogen
cycling on solubilities has been discussed. Hydrogen
isotherms can be used to determine the extent of internal
oxidation. They can also be used to help in the understand-
ing of the effect of the internal oxides on hysteresis and
hydrogen capacity.
Acknowledgements
T.B.F. wishes to acknowledge financial support from the
NSF and R.B. the DST, India and to the University of
Vermont for its hospitality while this research was carried
out. We acknowledge support of the National Institute of
Science and Technology, US Department of Commerce for
providing the neutron facilities. Dr. John Barker is thanked
for carrying out the SANS studies.
References
[1] J. Meijering, in: H. Herman (Ed.), Advances in Materials Research,
Vol. 5, Wiley, New York, 1971, p. 1.
[2] X. Huang, W. Mader, J. Eastman, R. Kirchheim, Scripta Metall. 22
(1988) 1109.
[3] X. Huang, W. Mader, R. Kirchheim, Acta Metall. Mater. 39 (1991)
893.
¨
[4] R. Kirchheim, X. Huang, T. Mutschele, in: N. Moody, A. Thompson
(Eds.), Hydrogen Effects on Material Behavior, The Minerals,
Metals and Materials Society, 1990, p. 85.
[5] H. Noh, T. Flanagan, R. Balasubramaniam, J. Eastman, Scripta
Mater. 34 (1996) 863.
[6] D. Wang, T.B. Flanagan, R. Balasubramaniam, Scripta Mater. 42
(2000) 911.
¨
[7] J. Eastman, M. Ruhle, Ceram. Eng. Sci. Proc. 10 (1989) 1515.
[8] R. Balasubramaniam, H. Noh, T. Flanagan, Y. Sakamoto, Acta
Mater. 45 (1997) 1725.
[9] W. Zhang, S. Luo, D. Wang, T.B. Flanagan, R. Balasubramaniam, J.
Alloys Comp. 330–332 (2002) 607.
[10] D. Wang, T. Flanagan, R. Balasubramaniam, Y. Sakamoto, Scripta
Mater. 43 (2000) 685.
[11] D. Wang, J.D. Clewley, T.B. Flanagan, R. Balasubramaniam, K.L.
Shanahan, Acta Mater. 50 (2002) 259.
[12] R. Kirchheim, Prog. Mater. Sci. 32 (1988) 261.
[13] D. Wang, H. Noh, S. Luo, T. Flanagan, J. Clewley, R. Balasub-
ramaniam, J. Alloys Comp. 339 (2002) 76.
[14] D. Wang, T.B. Flanagan, R. Balasubramaniam, Scripta Mater. 41
(1999) 517.
[15] J. Gegner, Segregation von Sauerstoff an inner Metall/Oxid-Phasen-
Fig. 4. Isotherms (323 K) for the internally oxidized Pd0.926Fe0.074 alloy
(1098 K, 72 h): (s) first cycle; (h) second cycle; (n) third cycle; (—)
pure Pd second cycle. The open symbols are for absorption and the closed
symbols for desorption.
¨
grenzen, Ph.D. Thesis, Universitat Stuttgart, Stuttgart, 1995.
[16] J. Mackert, R. Ringle, C. Fairhurst, J. Dent. Res. 62 (1983) 1229.

8
D. Wang et al. / Journal of Alloys and Compounds 356–357 (2003) 3–8
¨
[17] R. Balasubramaniam, R. Kirchheim, D. Wang, T.B. Flanagan, J.
Alloys Comp. 293–295 (1999) 306.
[19] M. Shutze, in: Protective Oxide Scales and their Breakdown, Wiley,
New York, 1997, pp. 17–65.
[18] D. Wang, J.D. Clewley, T.B. Flanagan, R. Balasubramaniam, J.
Alloys Comp. 298 (1999) 261.
[20] D. Wang, H. Noh, T.B. Flanagan, R. Balasubramaniam, J. Alloys
Comp. 348 (2003) 119.