Cluster size effects on hydrazine decomposition on Irn/Al 2O3/NiAl(1 1 0)

Article abstract of DOI:10.1016/j.susc.2005.10.054

A series of planar model catalysts were prepared by deposition of size-selected Irn+ on Al₂O₃/NiAl(1 1 0), and hydrazine decomposition chemistry was used to probe their size-dependent chemical properties. Small Ir_n (n ≤ 15) on Al₂O ₃/NiAl(1 1 0) are able to induce hydrazine decomposition at temperatures well below room temperature, with significant activity first appearing at Ir₇. Both activity and product branching are strongly dependent on deposited cluster size, with these small clusters supporting only the simplest decomposition mechanism: dehydrogenation and N₂ desorption at low temperatures, followed by H₂ recombinative desorption at temperatures above 300 K. For Ir₁₅, we begin to see ammonia production, signaling the onset of a transition to clusters able to support more complex chemistry.

Full text of DOI:10.1016/j.susc.2005.10.054

Surface Science 600 (2006) 461–467
www.elsevier.com/locate/susc
Cluster size effects on hydrazine decomposition on Ir /Al O /NiAl(110)
n
2 3
Chaoyang Fan, Tianpin Wu, William E. Kaden, Scott L. Anderson *
Department of Chemistry, University of Utah, 315 S. 1400 E. Rm 2020, Salt Lake City, UT 84112-0850, United States
Received 18 July 2005; accepted for publication 24 October 2005
Available online 7 December 2005
Abstract
þ
A series of planar model catalysts were prepared by deposition of size-selected Ir on Al O /NiAl(110), and hydrazine decomposition
n
2
3
chemistry was used to probe their size-dependent chemical properties. Small Ir
zine decomposition at temperatures well below room temperature, with significant activity first appearing at Ir
uct branching are strongly dependent on deposited cluster size, with these small clusters supporting only the simplest decomposition
mechanism: dehydrogenation and N desorption at low temperatures, followed by H recombinative desorption at temperatures above
00 K. For Ir15, we begin to see ammonia production, signaling the onset of a transition to clusters able to support more complex
chemistry.
2005 Elsevier B.V. All rights reserved.
n
(n 6 15) on Al
2
O
3
/NiAl(110) are able to induce hydra-
7
. Both activity and prod-
2
2
3
ꢀ
Keywords: Iridium; Catalysis; Hydrazine; Surface chemical reaction; Nanocluster
1
. Introduction
results under temperature-programmed desorption (TPD)
conditions, for Ir , n < 15.
n
Decomposition of hydrazine on metal surfaces and sup-
There are a number of studies in the literature that pro-
vide interesting points of comparison with this work.
Hydrazine decomposition has been studied under UHV
conditions for metal surfaces such as Ir(111) [1],
Ru(0001) [2], Pd(100) [3], Pt(111) [4], and Ni(111) [5].
There are also a number of temperature-programmed
desorption studies of NH and H on various bulk metal
ported metal particles is important in monopropellant
thrusters and gas generators, and potentially interesting
as a source of pure hydrogen for fuel cells. Catalysts for
thruster and gas generator applications must be stable at
high temperatures, and are typically refractory transition
metals dispersed on a refractory oxide support. Iridium dis-
persed on a high surface area aluminum oxide support is
the benchmark commercial catalyst (Shell 405). Issues of
importance include particle sintering, support degradation,
and activity over a wide temperature range. Even though
the thrusters and gas generators operate at high tempera-
tures, it is important to have good low temperature activity
to insure smooth cold starts in intermittent operation in
spacecraft environments. As part of a program to explore
fundamental properties of such catalysts, we are studying
size-selected Ir particles supported on planar epitaxial
Al O supports. Here we report hydrazine decomposition
3
2
surfaces that provide important insights for interpretation
of the hydrazine results [6–9]. It would be nice to compare
our results for Ir /Al O /NiAl with results for real high
n
2
3
surface area catalysts. The only detailed work on hydra-
zine/Ir/Al O high surface area catalysts is that of Falconer
2
3
and Wise [10], who studied TPD and reactions under stea-
dy flow conditions, but only above room temperature and
for hydrazine exposures much higher than in our experi-
ments. We previously reported a study of TPD of hydra-
zine on a model catalyst prepared by depositing enough
Ir on a planar Al O /NiAl(110) support to grow some dis-
2
3
2
3
tribution of non-size-selected Ir clusters [11], and found
chemistry qualitatively similar to that on bulk metal sur-
faces. We also reported preliminary results for reaction of
*
Corresponding author. Tel.: +1 801 585 7289; fax: +1 801 581 8433.
E-mail address: anderson@chem.utah.edu (S.L. Anderson).
0
039-6028/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.10.054

462
C. Fan et al. / Surface Science 600 (2006) 461–467
a few size-selected Ir under steady state conditions. The
The main UHV chamber has a mass spectrometer to
monitor gas composition, and a second, differentially
pumped mass spectrometer is used to monitor species
desorbing from the sample. This spectrometer views the
sample through a 3 mm aperture at the end of a skimmer
cone. We have found [18] that species desorbing from the
cluster-containing spot are detected substantially more effi-
ciently than those from the surrounding substrate area,
probably because of additional collimation from the ion-
izer geometry. One serious problem is that hydrazine
decomposes on internal surfaces of the mass spectrometer,
creating background at the product masses. To minimize
this background, the differential pumping skimmer cone,
including all wall surfaces in the vicinity of the ionizer, is
cooled to ꢀ87 K by flowing liquid nitrogen. In addition,
hydrazine cracks in the mass spectrometer ionizer yielding
ions at the masses of interest for the decomposition reac-
tion on the surface, thus it is critical to have an accurate
method for subtracting this background.
n
chemistry observed here for TPD from small Ir is quite
n
different from that seen for bulk metals and the high cover-
age model catalyst, indicating that both activity and prod-
uct branching are strongly affected by cluster size in this
system.
2
. Experimental methodology
The experiments were carried out in a UHV chamber
ꢁ
10
(
base pressure ꢀ2 · 10
Torr) that has been described
previously [11,12]. The chamber is attached to the end of
a mass-selected, ion deposition beam line that is used to de-
þ
þ
posit Irn on the sample surface. Irn is generated by laser
vaporization of a rastering Ir target, with the resulting Ir
vapor confined and cooled by a pulsed helium flow. Cat-
ions exiting the source are transported and mass-selected
by a series of quadrupole ion guides, and deposited at an
energy of 1 eV/atom through a 2 mm exposure mask posi-
tioned just in front of the sample surface. The system con-
tains facilities for in situ sample preparation, and
characterization by X-ray photoelectron spectroscopy
In this work, hydrazine was introduced through an in-
ert, pulsed inlet system [19], to minimize decomposition
of hydrazine in the inlet system. At the beginning of each
day, the liquid hydrazine sample (N H , 98.5% Alpha, or
(
XPS), Auger electron spectroscopy (AES), and ion scatter-
2
4
ing spectroscopy (ISS). The XPS measurements used an
AlKa source and XPS binding energies were calibrated
using the O and Al peaks of the Al O /NiAl substrate.
N D , 95%, C/D/N, Isotopes Inc.) was purified with sev-
2 4
eral freeze–pump–thaw cycles using both liquid nitrogen
vapor and dry ice/acetone baths to freeze N H (m.p. =
2
3
2
4
The sample substrate is a 7 · 7 mm NiAl(110) single
crystal (Surface Preparation Laboratory), spot welded to
a pair of tantalum heating wires, and suspended by tung-
sten rods from a liquid nitrogen cryostat. The sample tem-
perature can be controlled in the range from ꢀ90 K to
275 K) while maintaining substantial vapor pressure for
the NH decomposition product (b.p. = 240 K). During
3
use, the hydrazine container is kept in an ice–water bath
to maintain constant vapor pressure of ꢀ4.5 Torr [20].
Even though the inlet system is constructed entirely of
glass, teflon, and perfluoroalkoxy materials, there is still
some hydrazine decomposition. Before each sample dosing
operation, the hydrazine is additionally purified by repeat-
edly evacuating the head space of the hydrazine container
to preferentially pump away the higher vapor pressure
NH , N , and H decomposition products. With these
>
1300 K. Sample temperatures were measured by a K-type
thermocouple spot welded to the back of the NiAl(110)
crystal. The substrate for these experiments was an alumina
film grown epitaxially on the NiAl(110) crystal using a
procedure described by Kulawik et al [13]. The method
˚
has been shown to give alumina films ꢀ5 A thick, with
3
2
2
good local order and a structure similar to that of c-
precautions, the purity of the delivered hydrazine is excel-
lent [19].
Al O [14–17]. In brief, the NiAl(110) single crystal was
2
3
+
initially cleaned by repeated cycles of Ar bombardment
and 1270 K annealing. Before each experiment, the NiAl
crystal was Ar sputtered to remove deposited Ir, then an-
3. Results and discussion
+
nealed to 1270 K for 5 min in UHV. The Al O film was
Before analyzing the decomposition of hydrazine on Irn/
alumina, it is useful to review the TPD behavior of hydra-
zine from an inert surface. The inset to Fig. 1 shows a series
of hydrazine TPD spectra from a clean (Ir-free) alumina/
NiAl(110) surface. For unit sticking probability, ꢀ1.8 L
exposure results in deposition of a monolayer [11]. The first
monolayer desorbs in a broad peak around 181 K. For
coverages between 1 and ꢀ3 ML, the monolayer desorp-
tion peak is preceded by a sharp peak (153 K) assigned
to desorption of multilayer (i.e., mostly 2nd layer) hydra-
zine. For thicker films, a third peak at 166 K appears, be-
tween the multilayer and monolayer peaks. The smooth
0th order desorption behavior expected for evaporation
of a structureless multilayer film is observed only for thick
films (>10 ML). The origin of the 166 K desorption feature
2
3
grown by exposure to 1200 L of O at a sample tempera-
2
ture of 550 K, followed by annealing at 1070 K for 5 min.
To insure that there were no open patches in the Al O
2
3
layer, the oxidation treatment was repeated. The as-pre-
pared Al O thin film was characterized with XPS and
2
3
ISS, the latter showing that the film is continuous, with
no exposed Ni atoms. (To avoid sputter damage, the ISS
analysis was not done on samples used in the deposition
þ
experiments.) After the Al O thin film was prepared, Ir
2
3
n
clusters were deposited on the thin film at room tempera-
ture, to a density equivalent to 10% of a close packed Ir
1
4
2
monolayer (1.6 · 10 atoms/cm ). The sample was then
cooled to 100 K for dosing hydrazine prior to study by
temperature-programmed desorption.

C. Fan et al. / Surface Science 600 (2006) 461–467
463
coverage. This is to be expected, because our Ir coverage
is only 0.1 ML, so that even in the monolayer, most hydra-
zine is adsorbed on Ir-free regions of the alumina substrate.
For the decomposition products, there are contributions
from both desorption of decomposition products, and
from ionizer cracking of intact hydrazine desorbing from
the surface. If no decomposition occurs on the surface,
then the TPD spectral shape for the products should simply
track that of the hydrazine parent. Consider the NH prod-
3
uct. There is a sharp peak at 153 K, attributed entirely to
ionizer cracking of intact multilayer hydrazine desorbing
from the surface. If we scale the hydrazine spectrum to
match the intensity of this multilayer peak, then it can be
seen that the entire NH spectrum (data points) simply
3
tracks the scaled hydrazine signal (solid curve), indicating
that hydrazine ionizer cracking is the only significant
source of mass 17 signal.
In contrast, for N , the shape and intensity of the mono-
2
layer desorption component of the spectrum (data points)
is substantially different than what would be expected if
ionizer cracking were the only source of mass 28 signal
(
solid curve). Clearly, there is significant N desorption
2
from the sample, for temperatures ranging from just after
the multilayer desorption peak to ꢀ400 K. Finally, for
H , the two signal components observed below room tem-
2
perature can be mostly attributed to ionizer cracking of
desorbing N H , but there is a broad H signal component
2
4
2
from ꢀ300–550 K that must result from H desorption
2
from the sample. In this figure the intensities have not been
corrected for the variation in electron impact ionization
efficiency with product mass. For the ionizer conditions
Fig. 1. Inset: Coverage-dependent TPD spectra of hydrazine from clean
Al /NiAl(110). Main figure: TPD of hydrazine from Ir /Al
NiAl(110). Data points show the intensities for masses corresponding to
, N , NH , and H . Solid traces through the N , NH , and H spectra
show the ionizer cracking contribution to each mass.
2
O
3
7
2 3
O /
N
2
H
4
2
3
2
2
3
2
used, the H intensity should be multiplied by ꢀ2.3 to bring
2
it into proper scale with respect to the N product, and this
2
gives approximately the 1:2 N :H stoichiometry expected
2
2
is not entirely clear, but it may indicate either an phase
transition within the film during the heating process [21],
or possibly some layer structure in the multilayer hydrazine
film [22]. For our purposes, the important point is that
multilayer hydrazine desorbs intact, even from Ir/alumina
for complete decomposition of N H . Note that the higher
2 4
noise levels for N and H , relative to N H and NH , are a
2
2
2
4
3
consequence of the relatively high background at these
masses (i.e. CO and H ) typical of UHV systems. For some
2
experiments, we also used N D to better observe the
2
4
[
11], while monolayer hydrazine may decompose. The mul-
hydrogen product (i.e., D2).
tilayer hydrazine, thus, provides an internal standard
needed to correct for background at product masses origi-
nating from cracking of hydrazine in the mass spectrometer
ionizer.
It is interesting to compare these results with what is ob-
served in analogous experiments with a ‘‘high coverage’’
model catalyst prepared by deposition of 0.5 ML of Ir
+
on Al O /NiAl(110), where some distribution of larger
2
3
clusters was assumed to form by agglomeration [11]. In
those experiments, all three products (N , H , and NH )
3
.1. Decomposition during TPD
2
2
3
are observed, and each product has both a relatively sharp
low temperature feature (<250 K), and a broad high tem-
perature component (>350 K). This behavior is similar to
what is seen in hydrazine TPD from bulk Ir [1], Ru [2],
Pd [3], and Rh [23] surfaces. From comparison with these
TPD results for hydrazine, and TPD of NH and H on
various bulk metal surfaces [6–9], the sharp low tempera-
ture desorption features were attributed to primary hydra-
zine decomposition reactions, while the high temperature
features are clearly the result of recombinative desorption
of Hads, Nads, and NHx,ads species left on the surface by
The main figure shows desorption of hydrazine and its
decomposition products from a sample prepared by depos-
þ
iting 0.1 ML equivalent of iridium in the form of Ir .
7
Hydrazine was dosed at 100 K sample temperature, and
the exposure was set to produce approximately 2 ML
hydrazine coverage. The sample temperature was then
ramped at 3 K/s, while monitoring masses corresponding
to hydrazine, NH , N , and H . The top trace shows
3
2
3
2
2
N H signal vs. temperature, and is quite similar to that
2
4
for desorption from clean alumina for similar hydrazine

464
C. Fan et al. / Surface Science 600 (2006) 461–467
the primary decomposition chemistry. In contrast, the
hydrazine chemistry for the Ir /Al O (110) sample is much
7
2
3
simpler. After accounting for ionizer cracking, no NH3
desorption is observed, N is observed only at low temper-
2
atures, and H is observed mostly at high temperatures.
2
These results suggest a quite simple decomposition mecha-
nism for N H on Ir /Al O :
2
4
7
2
3
LowT : N H
! N " þ4H
2
4;ads
2
ads
HighT : 2Hads ! H "
2
The low temperature N feature for Ir is quite broad,
2
7
however, with a tail extending to ꢀ300 K, suggesting that
the low temperature N production mechanism may in-
2
volve several processes. For example, the dehydrogenation
reaction ultimately generating N + 4H
may pass
2
ads
through various N H intermediates, and there may be
2
x
Fig. 2. N
2
desorption signal corrected for ionizer cracking background.
some fission to NH with recombination as well (though
x
Vertical lines indicate temperatures of multilayer (153 K) and monolayer
none resulting in NH desorption). The fact that N contin-
3
2
(
181 K) hydrazine desorption features.
ues to evolve well above the temperature where N H
2
4
desorption stops (300 K), indicates that at least some of
the N is produced by secondary reactions of species left
þ
2
spectrum from the N spectrum. For comparison, we also
2
on the surface after all the intact N H has desorbed.
2
4
include an analogously corrected N TPD spectrum for
hydrazine on the clean (Ir-free) Al O /NiAl(110) sub-
strate. The data shown for Ir10 and Ir15 were taken using
N D . The vertical lines in the figure indicate the positions
of the multilayer and monolayer hydrazine desorption
features.
Note that there is a small amount of N production on
the Al O substrate, i.e., the N :N H intensity ratio is
slightly higher during desorption of monolayer N H , rela-
tive to the multilayer. This observation indicates that the
Al O /NiAl(110) film is not completely inert, however
2
The absence of a high temperature N2 recombinative
2
3
desorption feature for the Ir /alumina sample indicates
7
that there are no Nads or NHx,ads species left on the surface
at temperatures above 300 K, unlike the situation for the
high coverage catalyst or for bulk surfaces.
2
4
For H , in contrast, it is the low temperature feature
2
2
that is missing in the Ir /alumina results (or at least
þ
þ
7
2
3
2
2
4
strongly attenuated compared to the 0.5 ML Ir/Al O /
2
3
2
4
NiAl(110) sample). This observation indicates that the pri-
mary N H decomposition chemistry generates H that is
2
4
ads
2
3
mostly stable with respect to recombination until tempera-
tures >300 K are reached.
Low temperature ammonia production on the model
high coverage catalyst [11] occurred in a sharp feature at
the relatively low N intensity compared to those for the
2
0
.1 ML Ir /Al O /NiAl(110) samples, suggests that the
decomposition is probably limited to reaction at a low den-
sity of defects inevitably present in such epitaxial films (e.g.
domain boundaries [13,24]). The N intensities for samples
prepared with Ir , Ir , and Ir are only slightly greater than
n 2 3
2
15 K, similar to what is observed in hydrazine TPD on
2
bulk metal surfaces [1,23]. Several reactions generating
1
3
5
NH can be proposed:
3
that for the Ir-free Al O film, indicating that the Ir in these
2
3
samples is rather unreactive. In contrast, for samples pre-
pared with Ir , Ir , and Ir , the N production activity
N H
2
+ 2Hads ! 2NH ", or NH
+ Hads ! NH "
4;ads
3
2;ads
3
7
10
15
2
where Hads and NH2,ads are generated by decomposition of
N H . The absence of ammonia desorption for the Ir /alu-
mina sample indicates that for small clusters, such reac-
tions are not possible. The absence of a high temperature
is substantially higher than on the alumina film, indicating
that these cluster sizes have significant activity for hydra-
zine decomposition. In addition to indicating an activity
onset between Ir and Ir , this pattern also shows that
2
4
7
5
7
N recombinative desorption feature is also consistent with
no NHx,ads being generated in the decomposition mech-
anism.
whatever diffusion/agglomeration processes occur in these
samples is not sufficient to wipe out the dependence on
deposited cluster size, at least for temperatures up to
ꢀ300 K, where most of the hydrazine decomposition and
2
Fig. 2 shows the N signal observed for a series of
2
analogous experiments performed on samples prepared
N desorption has occurred.
2
þ
by deposition of Ir ; n ¼ 1; 3; 5; 7; 10; 15, on Al O /
It should be noted, however, that the heating used in
TPD does alter the samples significantly. This effect is most
obviously seen by simply running sequential TPD experi-
n
2
3
NiAl(110). In this figure, the contribution from ionizer
cracking has been subtracted by scaling the N
þ
2
H signal re-
4
corded during each TPD run such that the multilayer
ments on the same sample. Take the Ir results in Fig. 1
as an example. In the second TPD run on this sample,
7
þ
desorption peak intensity matches the N signal from the
2
þ
multilayer desorption, then subtracting the scaled N
2
H
the N production is reduced by a factor of ꢀ2.5. Similar
4
2

C. Fan et al. / Surface Science 600 (2006) 461–467
465
decreases are seen for all cluster sizes. The most likely
explanation is that the initially highly dispersed Ir clusters
are sintering at the high temperatures reached in the first
TPD run, resulting in large Ir particles. Samples containing
only large particles would have a substantially lower den-
sity of exposed Ir sites, and thus activity is reduced sub-
stantially. It is also possibly that some Ir is diffusing
subsurface, however, Libuda et al. have reported [25] that
this process is significant only for temperatures above our
maximum TPD temperature (600 K), at least for Pt. It is
interesting to compare the sintering behavior of Ir/Al O /
temperatures (Fig. 2), and evidently the recombinative
desorption of H2 occurs at lower temperatures on Irn/
Al O , compared to Ir-free Al O . Interpretation of this
high temperature desorption behavior is complicated by
the fact that the Ir clusters are probably sintering during
heating through this temperature range, as discussed
above.
2
3
2
3
3.2. Reactions with the alumina support
To further probe the effect of deposited Ir size on sam-
2
3
n
NiAl(110) with sintering experiments for Au /Al O /
ple properties, X-ray photoelectron spectra (XPS) were re-
corded for each sample. Unfortunately, the Ir XPS is
difficult to analyze quantitatively because the Ir coverage
is low, and the most intense Ir transitions (4d and 4f) have
substantial interference from Al and Ni transitions of the
substrate. About all we can say is that the apparent posi-
tion of the Ir4d5/2 feature is at ꢀ298 eV—ꢀ1 eV higher
than the 297 eV binding energy reported for bulk Ir [27].
A shift to higher binding energy is expected for small Ir
clusters, because of reduced final state screening in smaller
clusters. Unfortunately the combination of low Ir signal
and interference from an Al2p MgKa ghost line makes it
difficult to tell if there is any shift with deposited cluster size
for these samples.
n
2
3
NiAl(110) [26]. For Au, significant sintering was observed
to occur during deposition at room temperature, whereas
for Ir, the size-dependent TPD results suggest that sintering
is significant only when the sample is heated. Given the
much higher melting point of Ir (2683 K vs. 1338 K for
Au), it is perhaps not surprising that Ir requires higher tem-
peratures to activate diffusion and agglomeration.
The shapes of the corrected desorption features in Fig. 2
provide some additional insight into the decomposition
mechanism. For all samples, and particularly for those
with active size Ir (n = 7, 10, 15), there is a sharp onset
n
of N desorption at a temperature just above that where
2
desorption of the multilayer hydrazine film is complete.
(
The derivative line shape seen for some spectra at the po-
The O1s region of the XPS taken after TPD is plotted
for all samples in Fig. 3. Prior to TPD, the O1s binding en-
ergy is independent of deposited cluster size. Note that the
peaks observed for the samples prepared by deposition of
Ir , Ir , and Ir are consistently shifted ꢀ0.25 eV to higher
sition of the multilayer desorption peak (153 K) results
from error in subtracting this sharp feature in the ionizer
cracking background.) The sharp onset indicates either
that hydrazine decomposition in the monolayer film is trig-
gered by desorption of the multilayer, or that some hydra-
zine decomposition has occurred at lower temperatures,
1
3
5
binding energy compared to the samples prepared with Ir ,
7
Ir , and Ir . In other words, the final average O chemical
1
0
15
with N desorbing as soon as the overlayer is gone. In
either case, it is clear that these small clusters are active
environment is slightly different for samples prepared with
clusters that are active or inactive for hydrazine decompo-
sition, presumably because the more active clusters induce
production of some hydrazine decomposition product that
reacts with the alumina film. The two obvious possibilities
2
for hydrazine decomposition/N desorption at tempera-
2
tures below 200 K. In contrast, for the model catalyst pre-
pared with 0.5 ML Ir deposition [11], the low temperature
N desorption peaks sharply at 215 K, although there is
2
some N desorption at lower temperatures. N evolution
2
2
continues up to ꢀ300 K—well after intact desorption of
both multilayer and monolayer hydrazine is complete
(
ꢀ220 K), indicating that some N H species remains on
x
y
the surface, continuing to generate N .
2
The TPD behavior for the H product is more problem-
2
atic to interpret. For all Ir /Al O samples, H desorption
n
2
3
2
is observed in a broad feature running from ꢀ300 to 550 K
(
Fig. 1). Closer examination suggests that this feature has
two components—one peaking around 350 K, and one
peaking around 475–500 K. In hydrazine TPD from the
Ir-free alumina film, we do not observe H in the tempera-
2
ture range below 400 K, but there is H desorption at high-
2
er temperatures with intensity comparable to that observed
for T > 400 K for the Ir /Al O samples. We conclude that
n
2
3
the H desorption from 300–400 K results from hydrazine
2
decomposition on Ir , but the desorption at higher temper-
n
atures is from hydrazine decomposition on the alumina
film. The actual hydrazine decomposition occurs at low
Fig. 3. O1s XPS after TPD for all samples.

466
C. Fan et al. / Surface Science 600 (2006) 461–467
are reaction with hydrogen to partially hydroxylate the sur-
face, or reaction with nitrogen to generate some Al O N
z
either reaction of one hydrazine molecule with Hads gener-
ated in decomposition of a neighboring molecule, or
recombination of NH2 with Hads. The first mechanism
can only occur on clusters with enough binding sites to sta-
bilize at least two hydrazine molecules and their decompo-
sition products. The second pathway might occur with only
a single hydrazine parent, but still requires enough binding
x
y
species. McCafferty and Wightman report [28] that the
O1s binding energy for OH groups on alumina films are
shifted ꢀ1.2 eV to higher binding energy relative to the alu-
2
ꢁ
mina O centers. In our earlier work on TPD of hydrazine
from the 0.5 ML Ir/Al O /NiAl(110) model catalyst, N1s
2
3
XPS showed a small amount of Ir-induced nitrogen uptake
by the substrate, forming a nitride-like compound. In the
present work, no distinct N1s signal is observed in the
post-TPD analysis, but this is not unexpected, because
the Ir coverage is five time smaller. It is not clear what effect
nitrogen incorporation into the alumina film would have
on the O1s binding energy, however, the very low
sites to accommodate NH , Hads, and also the NHads or
2
Nads + Hads produced in concert with the Hads reactant.
Evidently, the small clusters simply cannot accommodate
this chemistry.
As cluster size increases, at some point we should begin
to see more complex chemistry, and the most sensitive
probe is probably ammonia production, because of low
background in the mass spectrometer. Fig. 4 shows that,
indeed, a small amount of ammonia is generated for the
largest cluster size studied here: Ir . The figure compares
(
398 eV) N1s binding energy indicates that the nitrogen is
electron rich, and this tendency to withdraw electron den-
sity should raise the binding energies of other elements in
the film. Therefore, the slight O1s binding energy shift
could result from incorporation of either H or N into the
film in small concentrations.
1
5
ND production from N D TPD from Ir /Al O and
3
2
4
10
2
3
þ
Ir /Al O samples. The raw ND3 signal is shown for
Ir , and the two upper traces show the ND signals for
1
5
2
3
þ
1
0
3
both Ir10 and Ir15 after subtracting the background from
ionizer cracking of desorbing N D . Clearly for Ir , there
3
.3. Cluster size effects
2
4
10
is no significant ammonia production, however, for Ir15
Two different types of cluster size effects are observed in
there is a small ND desorption feature visible between
3
this study. There is a significant increase in hydrazine
decomposition activity as the deposited cluster size is in-
creased from Ir to Ir , as shown by the N TPD results
ꢀ250 and 350 K. We conclude that Ir is just at the limit
1
5
where the complex chemistry observed on large Ir particles
and bulk surfaces begins to be possible.
5
7
2
in Fig. 2. A more dramatic size effect can be inferred by
comparing the results for the present Ir /Al O samples
We previously reported a study where Ir /Al O /
NiAl(110), n = 1, 5, 7, 10, model catalysts were used to
n
2
3
n
2
3
(
0
e.g. Fig. 1) with analogous results for the high coverage
.5 ML Ir/Al O sample [11], or for bulk Ir surfaces [1].
decompose hydrazine under steady state conditions, i.e.,
ꢁ
8
constant temperature and 5 · 10 Torr of hydrazine in
the chamber background. These conditions resulted in high
and rising background of hydrazine decomposition prod-
ucts from wall reactions, but by moving the sample so that
2
3
As outlined above, the decomposition chemistry is much
simpler for the small, size-selected clusters, with no NH3
production, no N recombinative desorption at high tem-
2
perature, and little if any H2 production below room
temperature.
There are several ways of thinking about the size depen-
dence of both activity and the decomposition mechanism.
The onset of activity may relate to changes in the Ir elec-
n
tronic properties with increasing cluster size, and it would
be interesting to examine such effects by valence photo-
emission and/or theory. It is also likely that both activity
and mechanism are affected by the number and type of sites
available for binding both hydrazine and various reaction
intermediates and products. Consider N production. For
2
the recombinative desorption mechanism to occur, the ini-
tial hydrazine decomposition must break the N–N bond to
generate separated NH species that subsequently dehydro-
x
genate and recombine to generate both N and H . The ab-
2
2
sence of this chemistry for samples with small Ir may
n
simply result from the clusters being too small to bind sep-
arated NH fragments. Instead, the nitrogen is lost as N at
x
2
low temperatures, and only Hads is left on the surface to
recombine at high temperatures.
A similar argument may account for absence of signifi-
cant ammonia production at any temperature. As outlined
3 2 4
Fig. 4. TPD of ammonia (ND ) following N D exposure to samples
prepared with Ir10 and Ir15, showing raw data and signal remaining after
subtraction of ionizer cracking contribution.
above, any mechanism for NH production will require
3

C. Fan et al. / Surface Science 600 (2006) 461–467
467
the mass spectrometer alternately viewed the Ir-containing
References
spot, and an adjacent region of Ir-free alumina, it was pos-
sible to observe signal attributable to Ir-catalyzed chemis-
try. It is interesting that under these conditions, ammonia
was observed at Tsurface = 300 K and 400 K. There are sev-
eral reasons why the product branching might be different
for these experiments, compared to the present TPD re-
[
[
1] H.H. Sawin, R.P. Merrill, J. Chem. Phys. 73 (1980) 996.
2] H. Rauscher, K.L. Kostov, D. Menzel, Chem. Phys. 177 (1993)
73.
3] R. Dopheide, L. Schr o¨ ter, H. Zacharias, Surf. Sci. 257 (1991) 86.
4] D.J. Alberas, J. Kiss, Z.-M. Liu, J.M. White, Surf. Sci. 278 (1992) 51.
4
[
[
[5] J.L. Gland, G.B. Fisher, G.E. Mitchell, Chem. Phys. Lett. 119 (1985)
9.
[
[
8
sults. The NH production probably resulted from reaction
3
6] S.A.C. Carabineiro, B.E. Nieuwenhuys, Surf. Sci. 505 (2002) 163.
7] A.K. Santra, B.K. Min, C.W. Yi, K. Luo, T.V. Choudhary, D.W.
Goodman, J. Phys. Chem. B 106 (2002) 340.
of impinging N H with H (or NHx,ads) already present
2
4
ads
on the surface from previous decomposition reactions.
Note that the temperatures studied were low enough that
reaction intermediates like Hads, NHx,ads, etc. would be sta-
ble on the surface to react with impinging hydrazine. It is
also possible that the clusters sintered to large clusters
during the ꢀ60-min reaction period, although the present
results suggest that sintering is not facile at room tempera-
ture, at least in the absence of adsorbates. We plan to study
these effects with our new inlet system, using pulsed hydra-
zine dosing to improve the signal/background ratio.
[
8] J.C.L. Cornish, N.R. Avery, Surf. Sci. 235 (1990) 209.
[9] D.E. Ibbotson, T.S. Wittrig, W.H. Weinberg, J. Chem. Phys. 72
1980) 4885.
10] J.L. Falconer, H. Wise, J. Catal. 43 (1976) 220.
(
[
[
11] S. Lee, C. Fan, T. Wu, S.L. Anderson, J. Phys. Chem. 109 (2005)
3
81.
[
12] S. Lee, C. Fan, T. Wu, S.L. Anderson, Surf. Sci. 578 (2005) 5.
[13] M. Kulawik, N. Nilius, H.-P. Rust, H.-J. Freund, Phys. Rev. Lett. 91
2003) 256101-1.
(
[
[
[
[
[
14] R.M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wuttig, W. Hoff-
mann, R. Franchy, H. Ibach, Surf. Sci. 259 (1991) 235.
15] J. Libuda, F. Winkelmann, M. B a¨ umer, H.-J. Freund, T. Bertrams,
H. Neddermeyer, K. M u¨ ller, Surf. Sci. 318 (1994) 61.
16] G. Ceballos, Z. Song, J.I. Pascual, H.-P. Rust, H. Conrad, M.
B a¨ umer, H.-J. Freund, Chem. Phys. Lett. 359 (2002) 41.
4
. Conclusions
We have shown that small Irn (n 6 15) on Al O /
2
3
17] T.T. Lay, M. Yoshitake, B. Mebarki, J. Vac. Sci. Technol. A 20
NiAl(110) are able to induce hydrazine decomposition at
temperatures well below room temperature. Both activity
and product branching are strongly dependent on depos-
ited cluster size, with the small clusters supporting only
the simplest decomposition mechanism: dehydrogenation
and N desorption at low temperatures, followed by H
(
2002) 2027.
18] S. Lee, C. Fan, T. Wu, S.L. Anderson, J. Chem. Phys. 123 (2005)
24710.
[19] C. Fan, T. Wu, S.L. Anderson, J. Vac. Sci. Technol. (submitted for
publication).
1
[
20] E.W. Schmidt, Hydrazine and Its Derivatives, Wiley-Interscience,
2
2
New York, 2001.
recombinative desorption at temperatures above 300 K.
For Ir , we begin to see ammonia production, signaling
[
21] R.S. Smith, C. Huang, E.K.L. Wong, B.D. Kay, Surf. Sci 367 (1996)
L13.
1
5
the onset of a transition to clusters able to support more
complex chemistry.
In future work we plan to examine the hydrogen chem-
istry in greater detail, and to try to passivate defects in the
[22] M.C. Yang, T.J. Rockey, D. Pursell, H.L. Dai, J. Phys. Chem. B 105
2001) 11945.
(
[
[
23] J. Prasad, J.L. Gland, Langmuir 7 (1991) 722.
24] A. Stierle, F. Renner, R. Streitel, H. Dosch, W. Drube, B.C. Cowie,
Science 303 (2004) 1652.
Al O /NiAl(110) film apparently responsible for spurious
H desorption features at high temperatures.
2
2
3
[25] J. Libuda, M. Baeumer, H.J. Freund, J. Vac. Sci. Technol. A 12
(1994) 2259.
[
26] S. Lee, C. Fan, T. Wu, S.L. Anderson, J. Phys. Chem. B 109 (2005)
1340.
1
Acknowledgment
[
27] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, J.J. Chastain,
R.C. King (Eds.), Handbook of X-ray Photoelectron Spectroscopy,
Physical Electronics, Eden Prarie, MN, 1995.
This work was supported by the US Air Force Office of
Scientific Research, under Grant F49620-03-1-0062.
[28] E. McCafferty, J.P. Wightman, Surf. Interface Anal. 26 (1998) 549.