N2O adsorption and reaction at Pd(110)

Article abstract of DOI:10.1016/S0039-6028(00)00606-3

The adsorption and decomposition of N₂O on Pd(110) has been studied using molecular beam uptake measurements and reflection-absorption infrared spectroscopy to characterize the reaction products. Nitrous oxide adsorption at 300 K leads to efficient dissociation with an initial sticking probability S₀ = 0.65. The reaction saturates at an uptake of 0.5 ML of adsorbed O, the surface forming an ordered Pd(110)-c(2×4)O overlayer. At higher temperatures the reaction probability drops, consistent with a trapping-dissociation mechanism. Adsorption at temperatures below 100 K leads to efficient dissociation, but the dissociation probability drops rapidly with uptake and thereafter N₂O is adsorbed intact. A secondary maximum in the N₂ product yield is seen for an N₂O uptake of 0.5 ML, with the surface simultaneously ordering to form a (1×2) low-energy electron diffraction pattern that becomes sharp as uptake saturates with a composition of 0.15 ML O and 0.85 ML of N₂O. We suggest that this structure is associated with a row-pairing reconstruction of Pd(110), forming a corrugated Pd surface. The IR spectrum shows bands at 1290 and 2262 cm^-1 due to N₂O adsorbed with a component of its axis perpendicular to the surface. Heating this surface leads to desorption of molecular N₂O at 100 K, followed by two further peaks at 117 and 140 K with partial dissociation. We discuss the origin of the off-normal emission of N₂ during N₂O decomposition and the role of N₂O as an intermediate during the reaction of NO on Pd(110).

Full text of DOI:10.1016/S0039-6028(00)00606-3

Surface Science 463 (2000) 1–10
www.elsevier.nl/locate/susc
N O adsorption and reaction at Pd(110)
2
*
S. Haq, A. Hodgson
Surface Science Research Centre, The University of Liverpool, Liverpool L69 3BX, UK
Received 7 January 2000; accepted for publication 17 May 2000
Abstract
The adsorption and decomposition of N O on Pd(110) has been studied using molecular beam uptake measurements
2
and reflection–absorption infrared spectroscopy to characterise the reaction products. Nitrous oxide adsorption at
300 K leads to efficient dissociation with an initial sticking probability S =0.65. The reaction saturates at an uptake
0
of 0.5 ML of adsorbed O, the surface forming an ordered Pd(110)-c(2×4)O overlayer. At higher temperatures the
reaction probability drops, consistent with a trapping-dissociation mechanism. Adsorption at temperatures below
100 K leads to efficient dissociation, but the dissociation probability drops rapidly with uptake and thereafter N O is
2
adsorbed intact. A secondary maximum in the N product yield is seen for an N O uptake of 0.5 ML, with the surface
2
2
simultaneously ordering to form a (1×2) low-energy electron diffraction pattern that becomes sharp as uptake
saturates with a composition of 0.15 ML O and 0.85 ML of N O. We suggest that this structure is associated with a
2
row-pairing reconstruction of Pd(110), forming a corrugated Pd surface. The IR spectrum shows bands at 1290 and
2262 cm−1 due to N O adsorbed with a component of its axis perpendicular to the surface. Heating this surface leads
2
to desorption of molecular N O at 100 K, followed by two further peaks at 117 and 140 K with partial dissociation.
2
We discuss the origin of the off-normal emission of N during N O decomposition and the role of N O as an
intermediate during the reaction of NO on Pd(110). © 2000 Elsevier Science B.V. All rights reserved.
2
2
2
Keywords: Adsorption kinetics; Low index single crystal surfaces; Molecule–solid reactions; Nitrogen oxides; Palladium; Reflection
spectroscopy
1. Introduction
by reflection–absorption infrared spectroscopy
(RAIRS) [3]. Heating surface previously
a
Nitrous oxide is one of the products formed in
the decomposition of NO at palladium surfaces
exposed to NO at 300 K results in an NO desorp-
tion peak near 495 K [3,4], desorption occurring
concomitantly with reaction to form N and N O
[1,2], along with gaseous N and adsorbed N and
2
2
2
O atoms. NO shows a rich chemistry on Pd(110),
[5], leaving O atoms on the surface that poison
further reaction. In the presence of a fuel, such as
H , CH or CO, the O atoms can be removed,
adsorbing molecularly at room temperature to
reconstruct the metal surface into a complex series
of adsorbate phases which have been characterised
2
4
allowing NO reduction to be run continuously.
Under different reaction conditions there appear
to be at least two different routes for N formation
2
during NO reduction at Pd(110) [5]: one is the
* Corresponding author. Fax: +44-151-708-0662.
E-mail address: andrewh@ssci.liv.ac.uk (A. Hodgson)
recombination of N [6] and the other is the
(a)
0039-6028/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S0039-6028 ( 00 ) 00606-3

2
S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
reaction of N +NO , which may either give
(a) (a)
directly [7] or by N O decomposition [8].
Here we investigate the adsorption and dissoci-
normal along the [001] direction and cos4.2 h in
the [110] direction [4,5]. Sharp angular peaks
result from a preferential repulsive energy release
along the axis of the desorption lobe, usually the
local surface normal [12]. Angular time-of-flight
:
N
2(g)
2
ation of N O on Pd(110) and discuss its possible
2
role as an intermediate in the reduction of NO.
Reaction of NO to form N and N O only
(TOF) measurements confirm that this N is trans-
2
2
2
lationally hot [8], quite unlike the N formed by
2
begins to occur above 420 K on Pd(110), where
NO starts to dissociate slowly or desorb from the
surface. Dissociation of NO is rate limiting [6]
and can be blocked either by the formation of an
recombination at high temperatures [9]. The
source of the lobular distributions was suggested
to be either a local reconstruction forming tilted
O
overlayer or by chemisorbed NO . NO
sites where N is produced, or a preferred collision
(a)
(a)
2
decomposition is catalysed by Pd reconstruction,
occurring preferentially at step or defect sites cre-
ated as Pd reconstructs in response to NO adsorp-
tion or desorption [9]. Steps have also been shown
to be important in promoting reaction on the
Pd(112) face [10,11]. The ratio of N O to N
product increases with the NO coverage [6,9]. This
is consistent with recent isotope experiments by
Ikai and Tanaka [7], who showed that N O was
geometry for the N +NO reaction [7]. A recent
(a)
(a)
study on the stepped Pd(211) surface shows that
N produced during NO desorption is peaked at
2
: :
25° to the surface along the [111] direction, normal
to the (100) steps [13].
Recently, Matsushima and co-workers have
2
2
compared the production of N during TPD of
2
N O and NO from Pd(110) using angular TOF
2
measurements [8]. Decomposition of N O pro-
2
2
duced translationally excited N that desorbed in
2
formed by reaction of NO with preadsorbed
N .
(a)
a lobe at 45° to the surface normal, the distribution
(a)
At high temperature the N is formed entirely
being very similar to that from NO decomposition.
2
by N recombination, since insufficient adsorbed
(a)
Since N O dissociates at temperatures below
2
NO is present to form N O [6,7]. If H is present
200 K, where the Pd(110) surface is immobile [3],
2
2
then O can be removed as water, allowing the
reaction to be run continuously. The N formed
under these conditions desorbs with an angular
distribution that is symmetric about the surface
the off-normal desorption cannot be attributed to
facetting of the surface by bulk displacive recon-
struction of the Pd. It was proposed that the
desorption angle was determined by a balance
(a)
2
normal and approximately thermal (cos h) [7].
between repulsion of the nascent N from the
2
metal surface and from the O atom as the N O
2
State-resolved measurements find that the N has
2
a cold translational and rotational energy distribu-
bond breaks [8]. This implies a transition state
tion but does carry considerable vibrational excita-
tion [9]. Nitrogen recombinative desorption occurs
indirectly via a short-lived molecular state; the
rotational and translational coordinates are cooled
by desorption from this well, but some vibrational
excitation survives trapping into the precursor
state. This picture is consistent with the broad
that lies flat, with the N O bond axis parallel to
the surface and closely aligned along the [001]
2
direction. The similarity of the N angular TOF
2
distributions from NO and N O decomposition
2
suggests that N O is the common intermediate
that leads to N production during NO flash
2
2
desorption [8].
angular distribution seen for N formed during
Here we describe the adsorption behaviour of
N O on Pd(110). We find that partial reaction to
2
the NO+H reaction [5].
2
2
In contrast, the N formed during temperature
form N and adsorbed O atoms occurs even at
2
2
programmed desorption (TPD) of an adsorbed
85 K. The resulting surface shows a (1×2) low-
NO layer shows very different behaviour. Although
energy electron diffraction (LEED) pattern, with
NO and N O desorb with a cos h and a cos2 h
distribution respectively, both centred around the
surface normal, nitrogen showed a lobular distri-
N O adsorbed upright on the surface. RAIR
2
2
spectra for N O adsorption are compared with
2
those obtained for non-dissociative adsorption on
bution, with a sharp maximum at ca 40° to the
Pt(111). We discuss the proposed mechanisms for

S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
3
formation of N during NO or N O dissociation
at high coverage and compare this with NO decom-
position on Pd(110) at higher temperature [9,14].
Pt(111), which was prepared using standard
techniques.
2
2
3. Results
2. Experimental
3.1. N O adsorption at 300 K
2
A thermal (effusive) molecular beam source was
used to monitor the adsorption of N O and NO
on a Pd(110) surface using the direct reflection
A beam of nitrous oxide was scattered from the
2
Pd(110) surface and the uptake of N O and forma-
2
technique to determine sticking probabilities.
Scattered molecules and reaction products were
detected by a quadrupole mass spectrometer
(QMS). The relative sensitivities to different
species (N O, NO and N ) were calibrated from
tion of products monitored using the QMS to
detect molecules scattered into the chamber. As
the beam is admitted to the chamber it reflects
from a flag in front of the sample, giving a constant
signal, proportional to the N O flux (Fig. 1). When
2
2
2
the QMS signals using beams whose relative fluxes
the flag is removed and the beam strikes the
are known for effusive flow conditions. The signals
Pd(110) surface N O adsorption is seen as a
decrease in the mass 44, 30 and 28 signals due to
2
at masses 28 and 30, due to N and NO, were
2
subject to interference from N O cracking. This
N O and its fragment ions. At temperatures above
2
background was removed by measuring the N O
2
2
150 K N O is not stable on Pd(110) and dissociates
2
cracking ratios and subtracting the appropriately
to give molecular N , which desorbs promptly,
and O, which remains on the surface. The mass
2
scaled N O signal, as described later. Absolute gas
2
fluxes were calibrated by determining simulta-
neously the sticking probability S(H) and the
exposure required to form a well-defined overlayer.
All coverages are referenced to the saturation
uptake of CO on an 85 K surface, which has a
coverage of 1 ML and a sharp (2×1)p1g1 LEED
pattern [15]. RAIR spectra were recorded at 85 K
using
a Mattson 6020 FTIR spectrometer.
Adsorption spectra were all obtained for back-
ground dosing, accumulating 200 scans per trace
with 4 cm−1 resolution.
The palladium crystal was polished to within
0.25° of the Pd(110) surface and was cleaned by
Ar+ ion sputtering and O treatment. The crystal
2
was mounted on two heating wires and could be
cooled to 85 K and heated resistively as required.
The surface showed a sharp LEED pattern,
whereas CO adsorption reproduced the
(2×1)p1g1 LEED pattern and the thermal desorp-
tion spectrum reported previously [15]. CO ther-
mal desorption is very sensitive to the presence of
contaminants which inhibit reconstruction of the
surface during desorption, providing a sensitive
Fig. 1. Mass spectrometer trace showing the uptake of N O
(m=44, solid line) and formation of nitrogen (m=28, points)
as the beam is admitted to the chamber and then exposed to
Pd(110) at 300 K. Mass 28 is also a cracking fragment of
2
N O and the traces have been normalised to give the same signal
2
at masses 44 and 28 from N O alone. The beam is admitted to
2
the chamber and reflects from a flag, giving a constant signal
due to the N O flux. When the flag is removed and the beam
2
strikes the Pd surface the N O traps and dissociates to give
2
test of the surface quality. N O uptakes and RAIR
N , which desorbs, and O atoms, which remain on the surface,
2
2
spectra were also recorded for N O adsorption on
poisoning further adsorption.
2

4
S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
44 and 30 signals, which are due solely to N O,
exactly the same coverage observed here for the
2
show an identical decrease as N O is removed by
2
saturation N O uptake at 300 K. An LEED analy-
2
reaction at the surface, followed by a recovery as
sis suggests the c(2×4)-O structure has O atoms
O poisons the surface to further reaction, Fig. 1.
The mass 28 signal shows a different behaviour,
increasing due to nitrogen production before fall-
ing as reaction ceases. Since mass 28 is both a
cracking fragment of N O and the N product
adsorbed in threefold sites on the (111) microfacets
in a zig-zag pattern along the top layer Pd rows,
each O bonding to two Pd atoms from the top
layer and one from the second layer, with O atoms
:
out of phase on neighbouring [110] chains [17].
2
2
mass, this signal is made up of a combination of
This model is consistent with the images observed
in STM studies [23–25]. In contrast to oxygen
a decrease in the cracking signal due to N O and
2
a (larger) increase caused by formation and
adsorption, reaction with N O forms a well-
2
desorption of nitrogen. We find that the N desorp-
ordered O overlayer even at 300 K, but reaction
2
tion signal is a mirror image of the N O removal
2
ceases at 0.5 ML coverage and does not continue
to form the (2×3) structures seen during oxygen
adsorption.
(Fig. 1) and no other products, such as NO,
were formed.
The uptake of N O and the N yield was
2
2
calibrated by using the molecular beam to provide
known gas fluxes and measuring the sticking prob-
ability and relative exposures required to form
well-defined overlayer structures on the Pd surface.
Since the beam operates in the effusive, low-
pressure limit, the flux of molecules depends on
3.2. Temperature dependence of N O reaction
2
As the surface temperature is increased, the
sticking probability for N O decreases, Fig. 2,
2
S(H) decreasing roughly linearly with coverage for
T>473 K, consistent with dissociation on bare
Pd(110) terraces and the surface being blocked to
further adsorption by formation of the c(2×4)-O
overlayer. For temperatures of 300 K or below,
S(H) shows clear evidence of trapping, the sticking
probability falling relatively slowly with coverage,
consistent with an indirect, trapping-dissociation
mechanism with both intrinsic and extrinsic pre-
cursor states. At 85 K the sticking probability is
(mass)−0.5. By running known fluxes of N and
2
N O into the chamber we are able to calibrate the
2
sensitivity of the mass 28 signal to nitrogen formed
by reaction compared with N O cracking in the
2
ion source. This allows us to confirm that N O
2
dissociation forms only N and O, with no N
2
atoms remaining on the surface. The absolute
N O uptake was determined by measuring the
uptake of CO required to form the (2×1)p1g1
2
LEED pattern, which is the saturation overlayer
of CO at 85 K and has a coverage H=1 ML.
Thermal desorption spectra of CO were similar to
those reported previously [15]. Integrating the
uptake curves for CO and N O adsorption, and
2
allowing for the difference in beam fluxes, gave a
saturation N O uptake of 0.50 ML at 300 K.
2
After N O adsorption the surface showed a
2
c(2×4) LEED pattern due to an ordered O over-
layer. Oxygen adsorption on Pd(110) at 400 K
gives rise to c(2×4) and (2×3) LEED structures
[16], which are associated with (1×2) and (1×3)
missing-row reconstructions of the Pd(110) surface
[17]. Only one O species is seen by EELS [18,19]
and the oxygen coverage of the c(2×4)-O struc-
ture was determined to be 0.5 0.05 ML by nuclear
Fig. 2. Coverage dependence of N O sticking as a function of
2
surface temperature. The saturation uptakes of N O at 85 and
2
300 K have been determined from the adsorption curves, while
reaction analysis [20,21] and by O uptake [22],
the uptake remains approximately constant for T≥300 K.
2

S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
5
independent of coverage until very close to satura-
tion, sticking saturating with an uptake of 1.0 ML
of N O, twice the uptake at 300 K. N O traps into
giving an increase in the mass 28 signal, Fig. 3,
but N formation falls off rapidly as uptake pro-
2
ceeds, with a small secondary maximum at a
2
2
a weakly bound precursor (physisorption) state
and either dissociates or desorbs back to the gas
phase, depending on the temperature. As the life-
time of the precursor state drops at higher temper-
atures, the coverage dependence S(H) becomes
linear and only the intrinsic state has a significant
reaction probability. The difference in activation
energy for desorption or dissociation of the intrin-
sic precursor can be estimated if we assume that
the trapping probability is approximately indepen-
coverage of half the saturation N O uptake. At
2
higher doses the mass 28 signal returns to follow
the mass 44 and 30 signals, indicating that N
2
evolution has ceased and N O is adsorbed molecu-
2
larly without significant dissociation
The probability for N O dissociating to
2
O +N
was obtained by using the molecular
(a) 2(g)
beam to calibrate the sensitivity of the QMS, as
described above. Using the relative QMS sensitivi-
ties and the absolute sticking probabilities obtained
from the uptake measurements allows us to obtain
dent of temperature. Plotting ln(1/S −1) against
0
1/T gives an estimate of 14 kJ mol−1 for the
the probability for N O dissociation as a function
2
difference in the activation barrier to desorption
compared with dissociation.
of uptake, Fig. 4. Initially, nitrous oxide adsorp-
tion leads to efficient dissociation on the clean
Below 100 K the adsorption behaviour changes
Pd(110) surface, with around 60% of the N O
dissociating, but this channel drops to just 5%
once half the saturation uptake has been achieved
2
and N O adsorbs intact on the surface. A beam
2
of nitrous oxide hitting the surface at 85 K results
in a decrease in the mass 44 and 30 signals due to
and the remaining N O adsorbs without significant
2
N O uptake, Fig. 3. The sticking probability for
dissociation. Overall, some 15 3% of the N O
2
2
2
N O is 0.75 and is independent of coverage until
which has adsorbed at saturation is dissociated to
adsorption is almost complete, Fig. 2. However,
give adsorbed O atoms and N . N O adsorption
2
2
adsorption is not entirely molecular and a propor-
tion of the N O dissociates to form O and N .
at 85 K produces a surface with 0.15 ML of
adsorbed O atoms and 0.85 ML of undissociated
2
2
The molecular nitrogen desorbs immediately,
N O.
2
The secondary maximum in the N evolution
2
curve for an N O uptake of 0.5 ML is entirely
2
Fig. 3. Mass spectrometer trace showing the time behaviour of
N O adsorption and N evolution at 85 K (cf. Fig. 1). Exposing
2
2
the surface to the N O beam leads to a decrease in the mass 44
2
signal (solid line) while the mass 28 signal ($) increases initially,
Fig. 4. Absolute probability for N formation from N O as a
2
2
due to the formation of N , and then decreases as nitrogen
function of the total N O uptake at 85 K. The calibration of
2
2
production ceases and N O adsorbs intact.
the N O uptake and N yield is discussed in the text.
2
2
2

6
S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
reproducible and is associated with ordering of the
Pd(110) surface. Halting adsorption on the 85 K
surface at an uptake of 0.4 ML results in a very
faint (2×1) LEED pattern, whereas by 0.6 ML
uptake the pattern has changed to a (1×2) pattern,
which becomes intense as adsorption continues up
to 1.0 ML total uptake. The van der Waals radius
O/N O-covered surface following N O adsorption
2
2
below 100 K is attributed to a displacive recon-
struction of Pd(110), resulting in a (1×2) period-
icity for the substrate, although a direct structural
technique such as STM would be required to prove
this conclusively.
˚
of N O (1.55 A) is larger than the atomic radius
3.3. Desorption of N O from Pd(110)
2
2
˚
of the palladium atoms (1.375 A), and the faint
(2×1) pattern at H<0.5 ML is probably due to
Heating a surface saturated with N O at 85 K
resulted in a complex desorption peak, Fig. 5a,
with three desorption features. The first peak, at
101 K, is predominantly due to desorption of
2
ordering of N O along the close-packed direction,
2
just as seen for 0.5 ML NO [3] and CO [26]
coverage, during adsorption at 180 K where
Pd(110) is unable to reconstruct to form a missing-
row structure. Local ordering of O formed by
molecular N O, with little dissociation. Cracking
2
of the N O begins to dominate at higher temper-
2
N O dissociation could also produce this pattern
atures, producing an N desorption peak near
117 K with
2
2
(oxygen adsorption typically forming streaks at
a
further shoulder at 140 K.
the half-order positions even when ordering to
form the missing-row reconstructions is incomplete
along the [001] direction [16,18]), but this is not
consistent with the disappearance of the weak
Desorption of N O from a surface pre-adsorbed
with ca 0.2 ML of O, Fig. 5b, leads to increased
dissociation at 100 K but a 30% lower overall yield
2
of N compared with N O desorption. After
2
2
(2×1) spots at higher N O coverages. The inten-
desorption the surface shows a c(2×4) LEED
2
sity and range of N O coverage over which the
pattern due to adsorbed O atoms. N O adsorption
2
2
(1×2) LEED pattern is observed suggest that this
on a c(2×4)O surface at 85 K gave primarily
ordering is associated with reconstruction of the
substrate, rather than simply due to formation of
an ordered N O overlayer. Since N O is too large
molecular N O desorption at 100 K, with a small
2
amount of residual N O dissociating at temper-
2
atures up to 140 K. The presence of O on the
2
2
:
to pack at full monolayer coverage along [110],
the overlayer must involve some displacement or
Pd(110) surface reduces the dissociation prob-
ability of N O and does not increase the binding
2
2
tilting of the N O (cf. both NO [3] and CO [26]
energy of N O in the same way as observed on
2
adsorption on this surface at high coverage) and
Ru(001) [30], where a large upward shift in the
could not form the simple (1×2) symmetry seen
in LEED. Although adsorption of NO and CO
can also lead to ordered (1×2) LEED patterns,
in this case the reconstruction only proceeds at
temperatures above 240 K where the surface is
sufficiently mobile to allow adequate transport of
Pd atoms to form the missing-row Pd structures
desorption temperature was observed.
3.4. RAIRS of N O adsorption on Pd(110) at 85 K
2
The RAIR spectrum for N O adsorption at
2
85 K is shown in Fig. 6 as a function of exposure.
Since some of the N O dissociates, O atoms are
2
[3,26,27]. Since the Pd(110)–(1×2)N O surface
always co-adsorbed along with the N O, although
2
2
the proportion of O to N O drops rapidly at
2
forms a sharp LEED pattern at 85 K a missing-
row reconstruction can be excluded, leaving a row-
pairing or rumpled-surface reconstruction as the
prime candidates. The Pd(110) surface undergoes
a row-pairing reconstruction for H adsorption
even at 120 K [28,29] and a pairing reconstruction
of Pd has also been proposed for the (2×2)p1g1
structure seen for high coverages of NO at 180 K
[3]. Thus the (1×2) periodicity of the
higher exposures. Initially the spectra show two
bands at 1290 and 2255 cm−1, with a shoulder
near 2240 cm−1. These peaks grow in together
(traces 1–4) until an exposure corresponding to ca
0.5 ML uptake of N O. For higher exposures the
2
surface adsorbs N O without any dissociation
2
(Figs. 3 and 4) and the peaks at 1290 and
2255 cm−1 continue to grow in intensity, with a

S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
7
Fig. 5. (a) Thermal desorption spectrum following uptake of 1 ML of N O at 85 K. (b) Desorption spectrum of a saturation dose
2
of N O from a surface which has 0.2 ML of O already on the surface from a previous dose of N O. In both cases the heating rate
2
2
is 0.9 K s−1 and the N O signal (mass 44) has been scaled to show the cracking signal it contributes at mass 28. The difference signal
2
(dotted lines) represents the formation of N as a function of temperature.
2
slight shift in the latter towards 2262 cm−1 at
saturation (trace 6). These bands correspond to
the fundamental parallel vibrational modes of
chemisorption on Ru(001) [34] and Pt(111) [32].
However, unlike these surfaces, no bands due to
physisorbed N O could be observed for adsorption
2
N O, with corresponding gas-phase frequencies of
on Pd(110) at 85 K and nor did the TPD indicate
2
1285 cm−1 (n ) and 2224 cm−1 (n ) respec-
any state other than the chemisorbed species that
desorbs above 100 K.
(NO)
(NN)
tively [31]. The presence of these bands in the
spectrum indicates that N O is adsorbed upright,
rather than lying down with its molecular axis
The growth of the band at 2240 cm−1 on
ation to form O, only the higher frequency band
2
Pd(110) seems to be associated with N O dissoci-
2
parallel to the surface. The n
slightly shifted with respect to the gas phase value,
band is only
(NO)
growing once N O dissociation ceases at
2
whereas the n =2262 cm−1 band is shifted up
H>0.5 ML, Fig. 6. This suggests that this band
(NN)
in frequency 38 cm−1 compared with the isolated
may be associated with N O binding at sites adja-
2
molecule.
cent to adsorbed O atoms. Annealing the surface
The orientation of N O, whether bound via the
N or the O atom, is not immediately clear. The
briefly up to 103 K (trace 7, Fig. 6) desorbed
2
roughly half the N O from the surface, dissociating
2
shift of the n
cule being bound to the surface via N, interaction
with the surface shifting up the frequency of the
band is consistent with the mole-
some of the N O and increasing the O coverage.
(NN)
2
The high-frequency band loses the sharp
2262 cm−1 peak, leaving a broad band centred
n
mode [32] but not influencing significantly
around 2249 cm−1. On re-adsorbing 0.2 L of
(NN)
the n
band at 1290 cm−1. The absence of any
N O the spectrum now shows two sharp, overlap-
(NO)
2
weakening of the N O bonds during chemisorption
ping peaks at 2260 and 2243 cm−1, suggesting that
2
reflects the non-bonding nature of the filled 7s
the O/N O structure has become more ordered,
2
and 2p orbitals, which are available for e− dona-
tion to the metal from the terminal N, and the
absence of back donation into the antibonding 3p
orbital, which lies rather high in energy [32,33].
sharpening up the low-frequency feature. In con-
trast, Pt(111) does not dissociate N O, as no N
production is observed during beam adsorption
2
2
experiments at either 85 or 300 K. N O adsorption
2
The n
band shifts are similar to those seen for
on Pt(111) was reversible, desorption and
(NN)

8
S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
the formation of a mixed O/N O overlayer with a
2
total adsorbate coverage of 1.0 ML. The adsorp-
tion behaviour is similar to that seen on W(110)
and Ru(001), where N O dissociation at low cov-
2
erages is followed by molecular adsorption, with
N desorbing on Ru(001) but forming adsorbed
2
N on W(110) [35]. Cu(110) also reacts in a similar
way [36], except that in that case dissociation of
N O dominates over molecular adsorption even at
2
85 K. Why Cu(110) should be more reactive to
N O than Pd(110) or Pt(111) is unclear. The high
2
initial dissociation probability on Pd(110) at low
temperature is probably associated with reaction
at step or defect sites. On Ni(755) N O dissoci-
2
ation is associated exclusively with the step sites
[37], whereas NO shows a similar preference for
dissociation at steps on Pd [10,11]. Alternatively,
it is possible that the formation of O atoms on the
surface helps to stabilise chemisorbed N O by
2
providing electron acceptor sites [30], although
judging from the TPD and RAIRS data the stabili-
sation of N O by O seems to be minimal on
2
(a)
Pd(110). At higher temperatures the total N O
2
uptake is reduced (Fig. 2), dissociation of N O
2
Fig. 6. IR absorption spectra of the Pd(110) surface as a func-
saturating with the formation of a c(2×4)-O
tion of exposure to N O. Traces 1–6 show the IR spectrum for
2
LEED pattern, corresponding to a coverage of
0.5 ML of O atoms.
exposure to 0.1, 0.2, 0.3, 0.5, 1.0 and 2.0 L of N O respectively
2
at 85 K. The lower traces show the effect of annealing to 103 K
N O adsorption at 85 K forms an overlayer
with a (1×2) LEED pattern that is ascribed to a
row-pairing or rumpling reconstruction of the
followed by re-exposing the O-covered surface to 0.2 L of
2
N O at 85 K. The features near 1900 cm−1 arise from miss-
2
cancellation of CO adsorbed on the IR windows.
Pd(110) surface. The RAIR spectrum for N O
2
re-exposure simply regenerating the original N O
adsorption at 85 K is characterised by two strong
2
RAIR spectra with no N evolution [32].
bands at 1290 and 2262 cm−1 due to chemisorbed
2
N O. Just as on Pt(111) and Ru(001), the n
2
(NN)
band is blue shifted due to the interaction with the
surface [32]. During TPD the dissociation prob-
ability is sensitive to the amount of O adsorbed
on the surface, a complete c(2×4)O overlayer
leading primarily to molecular desorption, while
4. Discussion
Adsorption of N O proceeds via a precursor
2
mechanism, molecules trapping into a physisorp-
tion well that then leads to dissociation, desorption
or transfer into the chemisorption state, depending
on temperature. At low coverage the desorption
channel has an activation barrier some
intermediate O coverages reduce the N O uptake
and the dissociation probability. However, the
presence of additional O on the surface does not
2
significantly alter the N O binding energy. This
2
14 kJ mol−1 greater than the barrier to N O disso-
contrasts with the behaviour on Ru(001) where O
2
ciation. This trapping–desorption mechanism is
stabilised N O adsorption, shifting the desorption
2
consistent with the thermal angular distributions
peak up in temperature by 40 K [30]. However, at
observed for N O formed during NO TPD [8,13].
low N O coverage, an additional band due to
2
2
N O dissociation occurs even at 85 K, leading to
chemisorbed N O is seen near 2240 cm−1 and
2
2

S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
9
appears to be associated with O coadsorption.
This band becomes sharper and more distinct
distributions formed by N O and NO decomposi-
2
tion suggests that N O may be a common interme-
2
when N O is readsorbed on a surface that has
diate to N formation during thermal desorption
2
2
been annealed to dissociate some of the N O,
Fig. 6.
of NO from a dense overlayer [8]. The angular
2
distribution of N formed during N O flash
2
2
A physisorption state was observed for N O on
desorption has two components, one peaked at
45° to the surface along the [001] direction, while
the second is symmetric about the surface normal.
This distribution has been ascribed to a transition
2
O-covered Ru(001) and also formed at high N O
2
coverages on the clean surface, showing an intense
perpendicular bending mode near 589 cm−1 in
electron energy loss spectroscopy due to N O
state with the N O axis parallel to the surface, the
2
2
physisorbed with its axis parallel to the surface
angle of desorption being controlled by the repul-
[34]. In addition to the upright chemisorbed
sive energy release between N and both the O
and the Pd surface [8]. However, the LEED evi-
2
species, weak n
and n
bands were seen for
(NN)
(NO)
physisorption on Pt(111) at 83 K. Avery [32]
reported a multilayer state that desorbed near
86 K, the strong bending mode again suggesting a
dence that Pd reconstructs following N O adsorp-
2
tion suggests that the metal surface may be
corrugated. N O decomposition does not occur on
2
flat-lying species. The observation of weak n
a flat surface, providing an alternative interpreta-
(NN)
and n
bands indicates that the axis of the
tion of the lobular N product distributions.
(NO)
2
physisorbed species seen here on Pt(111) is tilted
from the surface plane. No evidence was seen for
Decomposition of N O at the (1×2) surface may
2
lead to off-normal desorption peaks simply because
physisorption of N O on Pd(110) at T=85 K,
although the adsorption kinetics indicate such a
state is important during adsorption.
this reflects the local surface normal at the site
2
where N is created. This is indeed the model
2
proposed for N formation by reaction of NO at
2
Adsorption of NO on Ag(111) [38], Cu(111)
[39] and Cu(110) [36] at low temperature pro-
duces N O via the formation of (NO) dimers. On
Pd(211) [13], where the N angular distributions
can be interpreted as a repulsion along the local
2
surface normal from different step and terrace
2
2
Ag(111) the N O formed is stable, whereas on
sites. During NO desorption N O is formed by
2
2
Cu(110) some decomposes to give
adsorbed O atoms. Adsorption of NO on Pd(110)
N
and
the N +NO reaction [7], initiated by NO
2
(a)
(a)
(a)
decomposition which is rate limiting [6]. Some of
between 180 and 230 K also creates (NO) dimers
[3], but in this case no evidence was seen for
the N O formed then dissociates to give O and
2
2 (a)
(a)
N , which promptly desorbs. NO-covered Pd(110)
2
formation of N O. We have extended these meas-
shows a (1×3) missing-row reconstruction for
2
urements down to 85 K to see if N O is formed at
both low and high NO coverage [3], and N
desorbs from this surface in a lobular distribution,
2
2
lower temperatures as the density of NO increases.
The RAIR spectra of NO adsorbed at 85 K were
identical to those reported previously at 180 K [3],
peaked along the (111) microfacets [4,5]. The
Pd(110) surface is very mobile during this reaction
and clearly there is every chance that reaction
occurs on Pd(111) microfacets, irrespective of the
with no evidence for formation of N O. Heating
2
an NO-covered surface reproduced the TPD
behaviour observed previously [3,6], with no
insensitivity of the N angular distribution to
the overall reconstruction of the NO phase [7].
2
N O formation until NO starts to dissociate at
2
temperatures above 420 K [9]. We conclude that
Therefore, although the similar energy release into
N translation certainly suggests that N O may be
there is no evidence for the direct reaction of
adsorbed NO to form N O on Pd(110). N and
2
2
a transient intermediate in both reactions [8], we
propose that the lobular energy release is caused
by the local normal of the reaction site and not
by the balance of repulsion between O, the surface
2
2
N O products are only formed once the surface is
2
sufficiently hot to dissociate NO [6,7], probably at
a
step or defect site as the Pd surface
reconstructs [9].
and the nascent N . This explanation for the off-
normal desorption distributions removes the very
2
The similarity of the N translational energy
2

10
S. Haq, A. Hodgson / Surface Science 463 (2000) 1–10
[7] M. Ikai, K. Tanaka, J. Phys. Chem. B 103 (1999) 8277.
[8] Y. Ohno, K. Kimura, M. Bi, T. Matsushima, J. Chem.
Phys. 110 (1999) 8221.
[9] P. Samson, S. Haq, M.J. Murphy, A. Hodgson, in
preparation.
[10] R.D. Ramsier, Q. Gao, H.N. Waltenburg, K.W. Lee, O.W.
Nooij, L. Lefferts, J.T. Yates, Surf. Sci. 320 (1994) 209.
[11] R.D. Ramsier, Q. Gao, H.N. Waltenburg, J.T. Yates,
J. Chem. Phys. 100 (1994) 6837.
[12] W. van Willigen, Phys. Lett. A 28 (1968) 80.
[13] M. Ikai, K. Tanaka, J. Chem. Phys. 110 (1999) 7031.
[14] M.J. Murphy, P. Samson, J.F. Skelly, A. Hodgson, in: R.
Campargue (Ed.), Advances in Atomic and Molecular
BeamsSpringer, 2000, in press.
[15] J.W. He, P.R. Norton, J. Chem. Phys. 89 (1988) 1170.
[16] G. Ertl, P. Rau, Surf. Sci. 15 (1969) 443.
[17] B. Brena, G. Comelli, L. Ursella, G. Paolucci, Surf. Sci.
375 (1997) 150.
[18] M. Jo, Y. Kuwahara, M. Onchi, M. Nishijima, Chem.
Phys. Lett. 131 (1986) 106.
[19] M. Nishijima, M. Jo, Y. Kuwahara, M. Onchi, Solid State
Commun. 60 (1986) 257.
[20] J.W. He, U. Memmert, K. Griffiths, W.N. Lennard, P.R.
Norton, Surf. Sci. 202 (1988) L555.
specific repulsion between N and the O/Pd(110)
surface proposed by Ohno et al. [8], but is consis-
tent with the vast body of desorption data that
finds the product repulsion aligned along the local
surface normal [12,40].
2
5. Conclusion
N O dissociation proceeds rapidly even at 85 K,
2
the O formed poisoning further reaction and allow-
ing N O to adsorb into a molecular chemisorption
2
state. The O/N O-covered surface shows a (1×2)
2
LEED pattern, suggesting a row-paired or rum-
pling reconstruction of the Pd(110) surface. The
RAIR spectrum shows bands at 1290 and
2262 cm−1 due to N O adsorbed upright on the
2
surface, bound via the N. A shoulder to the n
(NN)
feature near 2240 cm−1 is associated with N O
2
adsorption adjacent to coadsorbed O. No evidence
[21] J.W. He, U. Memmert, K. Griffiths, P.R. Norton, J. Chem.
Phys. 90 (1989) 5082.
[22] J. Goschnick, M. Wolf, M. Grunze, W.N. Unertl, J.H.
Block, J. Lobodacackovic, Surf. Sci. 178 (1986) 831.
[23] H. Tanaka, J. Yoshinobu, M. Kawai, Surf. Sci. 327
(1995) L505.
[24] H. Niehus, C. Achete, Surf. Sci. 369 (1996) 9.
[25] R.A. Bennett, S. Poulston, I.Z. Jones, M. Bowker, Surf.
Sci. 401 (1998) 72.
could be found for N O formation during NO
2
adsorption at low temperature, N O only being
2
produced once NO starts to dissociate at temper-
atures above 420 K. Reconstruction of Pd(110) in
the presence of N O suggests that N O decomposi-
2
2
tion does not occur on a flat surface, providing an
alternative explanation for the off-normal desorp-
tion of the N produced during TPD.
[26] R. Raval, S. Haq, M.A. Harrison, G. Blyholder, D.A.
King, Chem. Phys. Lett. 167 (1990) 391.
2
[27] P. Hu, L.M. Delagarza, R. Raval, D.A. King, Surf. Sci.
249 (1991) 1.
[28] H. Niehus, C. Hiller, G. Comsa, Surf. Sci. 173 (1986) L599.
[29] G. Kleinle, M. Skottke, V. Penka, G. Ertl, R.J. Behm, W.
Moritz, Surf. Sci. 189 (1987) 177.
[30] H.H. Huang, C.S. Seet, Z. Zou, G.Q. Xu, Surf. Sci. 356
(1996) 181.
[31] G. Herzberg, Infrared and Raman Spectra, Van Nostrand,
New York, 1945.
Acknowledgements
We would like to thank Ras Raval for discussing
unpublished STM data and the EPSRC for sup-
port of this work.
[32] N.R. Avery, Surf. Sci. 131 (1983) 501.
[33] S.D. Peyerimhoff, R.J. Buenker, J. Chem. Phys. 49
(1968) 2473.
References
[34] T.E. Madey, N.R. Avery, A.B. Anton, B.H. Toby, W.H.
Weinberg, J. Vac. Sci. Technol. A: 1 (1983) 1220.
[35] E. Umbach, D. Menzel, Chem. Phys. Lett. 84 (1981) 491.
[36] W.A. Brown, R.K. Sharma, D.A. King, S. Haq, J. Phys.
Chem. 100 (1996) 12 559.
[1] S.M. Vesecky, P.J. Chen, X.P. Xu, D.W. Goodman, J. Vac.
Sci. Technol. A 13 (1995) 1539.
[2] S.M. Vesecky, D.R. Rainer, D.W. Goodman, J. Vac. Sci.
Technol. A 14 (1996) 1457.
[3] R. Raval, M.A. Harrison, S. Haq, D.A. King, Surf. Sci.
294 (1993) 10.
[4] M. Ikai, H. He, C.E. Borroni-Bird, H. Hirano, K. Tanaka,
Surf. Sci. 315 (1994) L973.
[5] M. Ikai, K. Tanaka, Surf. Sci. 357 (1996) 781.
[6] R.G. Sharpe, M. Bowker, Surf. Sci. 360 (1996) 21.
[37] H. Orita, H. Kondoh, H. Nozoye, J. Catal. 177 (1998) 217.
[38] W.A. Brown, P. Gardner, D.A. King, J. Phys. Chem. 99
(1995) 7065.
[39] S.K. So, R. Franchy, W. Ho, J. Chem. Phys. 95 (1991)
1385.
[40] A. Hodgson, Prog. Surf. Sci. 63 (2000) 1.