Temperature programmed desorption studies of OD coadsorbed with H2 on Pt(111)

Article abstract of DOI:10.1063/1.481327

A molecular beam source of pure hydroxyl radicals has been developed and used to explore the water reaction catalyzed over Pt(111). An electrostatic hexapole selectively focused OD radicals from a supersonic corona discharge source onto a Pt target at a surface temperature of T_S=143 K. Subsequent D2O temperature programmed desorption (TPD) spectra revealed two major features, one near T_S ca. 170 K from desorption of molecular water overlayer and a second near T_S ca. 210 K from the decomposition of an adsorbed OD intermediate. The latter feature was isolated and analysis of TPD spectra revealed that the D2O production reaction was approximately half-order in total oxygen coverage with a pre-exponential factor ranging from υ_d=4+/-1*10¹⁶ to 5+/-2*10¹⁸ molecules^1/2 cm^-1 s^-1 and activation energy E_a=9.7+/-0.1 to 11.5+/-0.1 kcal mol^-1 for initial coverage ranging from θ₀=0.04 to 0.25 ML. Coadsorption studies of OD and H2 revealed that H atoms drive reactions with adsorbed OD at T_S ca. 180 K to form all three water isotopes: D2O, HDO, and H2O. Oxygen (O2) TPD spectra contained three desorption features (T_S=700 K, 735 K, and 790 K). The relative abundance of O2 from these three features was virtually the same in all low temperature (T_S=143 K) TPD experiments. At elevated dosing temperatures (T_S=223 K) the two features at T_S=700 K and 790 K could be selectively titrated from the surface by hydrogen. The presence of hydrogen prior to OD exposure at this elevated temperature prevented the accumulation of oxygen on the surface. The implications of these observations on our mechanistic understanding of the low temperature (T_S<210K) water reaction are discussed.

Full text of DOI:10.1063/1.481327

Temperature programmed desorption studies of OD coadsorbed with H 2 on Pt(111)
Kyle M. Backstrand, Michael A. Weibel, Robert M. Moision, and Thomas J. Curtiss
Citation: The Journal of Chemical Physics 112, 7209 (2000); doi: 10.1063/1.481327
View online: http://dx.doi.org/10.1063/1.481327
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/112/16?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
The structure of the mixed OH + H 2 O overlayer on Pt{111}
J. Chem. Phys. 123, 064711 (2005); 10.1063/1.1996572
An interaction model for OH + H 2 O -mixed and pure H 2 O overlayers adsorbed on Pt(111)
J. Chem. Phys. 122, 194705 (2005); 10.1063/1.1900088
The influence of steps on the dissociation of NO on Pt surfaces: Temperature-programmed desorption studies of
NO adsorption on Pt{211}
J. Chem. Phys. 119, 10844 (2003); 10.1063/1.1619940
The reduction of NO on Pt(100) by H 2 and CO studied with synchrotron x-ray photoelectron spectroscopy
J. Chem. Phys. 119, 6245 (2003); 10.1063/1.1602059
Vibrational and structural properties of OH adsorbed on Pt(111)
J. Chem. Phys. 111, 11147 (1999); 10.1063/1.480472
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.181.251.131 On: Sat, 22 Nov 2014 03:55:10

JOURNAL OF CHEMICAL PHYSICS
VOLUME 112, NUMBER 16
22 APRIL 2000
Temperature programmed desorption studies of OD coadsorbed
with H on Pt„111…
2
Kyle M. Backstrand, Michael A. Weibel, Robert M. Moision, and Thomas J. Curtiss
Department of Chemistry, University of Utah, Salt Lake City, Utah 84112
͑
Received 15 January 1999; accepted 19 January 2000͒
A molecular beam source of pure hydroxyl radicals has been developed and used to explore the
water reaction catalyzed over Pt͑111͒. An electrostatic hexapole selectively focused OD radicals
from a supersonic corona discharge source onto a Pt target at a surface temperature of TS
ϭ143 K. Subsequent D O temperature programmed desorption ͑TPD͒ spectra revealed two major
2
features, one near T ϳ170 K from desorption of molecular water overlayer and a second near TS
S
ϳ210 K from the decomposition of an adsorbed OD intermediate. The latter feature was isolated
and analysis of TPD spectra revealed that the D O production reaction was approximately half-order
2
1
6
in total oxygen coverage with a pre-exponential factor ranging from v ϭ4Ϯ1ϫ10 to 5Ϯ2
d
1
8
1/2
Ϫ1 Ϫ1
Ϫ1
ϫ10 molecules cm
s
and activation energy E ϭ9.7Ϯ0.1 to 11.5Ϯ0.1 kcal mol for initial
a
coverage ranging from ␪ ϭ0.04 to 0.25 ML. Coadsorption studies of OD and H revealed that H
0
2
atoms drive reactions with adsorbed OD at T ϳ180 K to form all three water isotopes: D O, HDO,
S
2
and H O. Oxygen ͑O ͒ TPD spectra contained three desorption features (T ϭ700 K, 735 K, and
2
2
S
7
90 K͒. The relative abundance of O from these three features was virtually the same in all low
2
temperature (T ϭ143 K͒ TPD experiments. At elevated dosing temperatures (T ϭ223 K) the two
S
S
features at T ϭ700 K and 790 K could be selectively titrated from the surface by hydrogen. The
S
presence of hydrogen prior to OD exposure at this elevated temperature prevented the accumulation
of oxygen on the surface. The implications of these observations on our mechanistic understanding
of the low temperature (T Ͻ210 K) water reaction are discussed. © 2000 American Institute of
S
Physics. ͓S0021-9606͑00͒70214-5͔
I. INTRODUCTION
however, key properties are worth highlighting. It is known
9
that both H and O dissociatively adsorb on platinum. Hy-
2
2
The water production reaction on Pt͑111͒ is one of the
most well-studied catalytic reaction systems. Most modern
tools of surface science have been exploited to yield an ex-
tensive database on which to base mechanistic theories.
These studies have led to cycles of contentment and discon-
tent regarding the agreement between the predictions of pro-
posed mechanisms for the reaction and the latest data. In
recent years the level of discontent has been rising, primarily
due to the meticulous efforts of Verheij, Comsa and
10
drogen atoms are believed to be mobile at T Ͼ100 K and
oxygen atoms at T Ͼ300 K.
S
11,12
Oxygen atoms from the ad-
S
sorption of molecular O are known to form ordered p(2
2
13
ϫ2) islands, even at very low coverages. The p(2ϫ2)
layer saturates at a coverage ␪ ϭ0.25 ML relative to the Pt
0
atom density. The sticking probability of H on Pt͑111͒ de-
2
pends on temperature, the oxygen coverage and structure in a
complex way.1
4–16
In general, ordered oxygen at coverages
1
5
1
–3
␪0Ͻ0.25 ML enhance the sticking probability of H2. The
dissociative adsorption of O also has many nuances.
co-workers
to correlate nuances in surface structure with
1
7–20
2
dramatic changes in reactivity at low coverages. Verheij
1
,4–6
Using more aggressive oxidants ͓e.g., NO ͑Ref. 21͒ and O3
2
et al.
have proposed a reaction mechanism, the reactive
͑Ref. 22͔͒ Koel and co-workers have attained saturation cov-
site mechanism, in order to try and consolidate theoretical
results with all experimental results. According to their
model, an intermediate is initially formed ͑OH or H O ) at
erages exceeding the maximum ␪ ϭ0.25 ML coverage at-
0
tained with O . However, to date no water reaction studies
2
2
2
have been undertaken using these highly oxidized platinum
surfaces.
The interactions of water with surfaces have been stud-
ied extensively and reviewed by Thiel and Madey.23 Water
adsorbs molecularly on a clean Pt͑111͒ surface, is mobile at
reactive sites in very low concentrations. At low tempera-
tures where oxygen mobility is low, water is formed by
transport of hydrogen across oxygen islands surrounding the
reactive site to a moving reaction front. This reaction front
has been observed in field ion microscopy ͑FIM͒
7
T Ͼ100 K, and forms hydrogen-bonded islands even at very
experiments. It is within this context that we would like to
S
1
0
low temperatures. The maximum desorption rate for Pt
bonded water occurs near TSϭ180 K and for multilayer ice
near TSϭ165 K.24 We see the same behavior resulting from
D2O TPD measurements. Although water does not dissoci-
ate, studies of the coadsorption of H O and D O indicate
introduce our new experimental results. We have used a
novel hydroxyl radical beam source to prepare a catalytic
surface covered with reaction intermediates in order to probe
the low temperature water production kinetics.
A complete review of the extensive literature on the
water/Pt reaction system is beyond the scope of this paper,
8
2
2
1
0,24
isotopic scrambling of hydrogen is facile.
0
021-9606/2000/112(16)/7209/10/$17.00
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
31.181.251.131 On: Sat, 22 Nov 2014 03:55:10
7209
© 2000 American Institute of Physics
1

7210
J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Backstrand et al.
Reactions between adsorbed hydrogen and oxygen
Fourier components comprising the measured D O wave-
2
2
5
occur at temperatures of T Ͻ130 K.
Gland and co-
S
forms indicated that the D O formation rate was dominated
2
2
4,26–27
workers
studied the water reaction on Pt͑111͒ using
by a process that was linear in coverage of OD͑a͒ with a
smaller contribution from a process that was quadratic in
OD͑a͒ coverage. In this way, Anton and Cadogan believe
they could isolate contributions to the rate from both the
temperature programmed desorption ͑TPD͒, XPS and ultra-
violet photoelectron spectroscopy ͑UPS͒. Gland proposed a
mechanism where water is formed via sequential addition of
H to first O and then OH,
OD͑a͒ϩD͑a͒→D O͑g͒
͑4͒
2
H͑ad͒ϩO͑ad͒→OH͑ad͒,
͑1͒
͑2͒
and
H͑ad͒ϩOH͑ad͒→H O͑ad͒,
2OD͑a͒→D O͑g͒ϩO͑a͒
͑5͒
2
2
and that the kinetics of the reaction are controlled by the
reactions and measured their corresponding Arrhenius pa-
shape of the oxygen islands. Although Sexton10 observed
Ϫ4Ϯ1
Ϫ1
2 Ϫ1
rameters: Aϭ4ϫ10
molecules cm s
,
Eaϭ16
Ϫ1
Ϫ3Ϯ1
Ϫ1
2 Ϫ1
hydroxyl intermediates spectroscopically when H O was
Ϯ2 kcal mol
and Aϭ1ϫ10
molecules cm s ,
2
Ϫ1
coadsorbed onto an O-covered surface, hydroxyl was not ob-
served when hydrogen and oxygen were coadsorbed. This
was explained by the first hydrogen addition, HϩO, being
the rate-limiting step and the second hydrogen addition,
HϩOH, being fast so that no substantial surface concentra-
tion of hydroxyl was seen.
Eaϭ18Ϯ3 kcal mol , respectively. These measurements
were made in the limit of low coverages of all reactants and
intermediates. They also concluded that the OD͑a͒ interme-
diate was formed rapidly and essentially irreversibly.
In order to better understand the low temperature mecha-
nism of the water production reaction on Pt͑111͒, we have
isolated a specific reaction intermediate, OD, and studied its
adsorption alone and with H2, as well as the subsequent re-
action and desorption of water and oxygen.
Hydroxyl-like intermediates have been identified in a
2
8
number of studies. Fisher and co-workers first identified a
water reaction intermediate, OH͑I͒, from the apparent de-
composition of H O adsorbed on a saturation p(2ϫ2) oxy-
gen overlayer. Water production from the decomposition of
2
II. EXPERIMENT
this intermediate occurred at T ϳ215 K. Subsequent studies
S
The OD radical beam source employed in these studies
has been described in detail. Briefly, He carrier gas was
on intermediate OH͑I͒ indicated that the stoichiometry for
forming OH͑I͒ is given by2
8
9,30
bubbled through a D O reservoir held at room temperature.
2
The D O saturated He at a backing pressure of 32 psia was
2
2
H O͑a͒ϩO͑a͒→3OH͑a͒ϩH͑a͒
͑3͒
2
then expanded from a supersonic corona discharge plasma
that the rate law for the disappearance of H O monitored
source with an 80 ␮m diam nozzle.35 Hydroxyl radicals en-
2
with SIMS to form the OH͑I͒ intermediate is first order in
trained in the supersonic flow were focused and rotationally
state selected by an electrostatic hexapole.3
6,37
The hexapole
H O(a) and first order in O͑a͒, and that the activation energy
2
Ϫ1
and pre-exponential factor are E ϭ10.2Ϯ0.8 kcal mol and
Aϭ6.9ϫ10
voltage was fixed at V ϭ12 kV in order to transmit the
a
0
Ϫ1Ϯ1.4
2
Ϫ1 Ϫ1
30
3 3 3
cm molecules
s
,
respectively.
It
͘
͉J⍀M ϭ͉ 2 2 2͘ rotational state through the hexapole exit col-
was believed that the intermediate produced by coadsorption
of water on p(2ϫ2) oxygen, OH͑I͒, was the same interme-
diate formed from adsorbed hydrogen reacting with adsorbed
oxygen. However, using EELS, Mitchell and White sug-
gested that two hydroxyl intermediates are formed, one from
coadsorption of water and oxygen, OH͑I͒, and one from
coadsorption of hydrogen and oxygen. For the purposes of
this discussion, the latter intermediate will be referred to as
limator. A beam-stop was placed at the midpoint of the hexa-
pole assembly to block any oxygen or deuterium atoms or
D O molecules from being transmitted to the Pt crystal. The
2
resulting pure OD radical beam was approximately 1.5 mm
in diameter at the crystal with an approximate flux density of
1
3
Ϫ2 Ϫ1
10 radicals cm
s .
These experiments were conducted in a liquid-nitrogen
trapped, diffusion pumped UHV vacuum chamber with a
3
1
Ϫ10
OH͑II͒. The intermediates OH͑I͒ and OH͑II͒ have distinc-
tive EELS spectra with modes resembling K Pt͑OH͒ and
base pressure of 3ϫ10
Torr. The Pt͑111͒ crystal was pur-
chased from Monocrystals Co., cut and polished to within
0.5 degrees to expose the ͑111͒ surface, as was verified by
Laue x-ray diffraction.
2
6
͓
PtOH͑CH ) ] , respectively. Efforts by Germer and Ho to
3 3 4
reproduce intermediate OH͑II͒ from the reaction of hydrogen
with atomic oxygen presaturated Pt͑111͒ were unsuccessful,
however. Rather, they observed EELS spectra consistent
with OH͑I͒ when they coadsorbed hydrogen and oxygen.32
As the discussion of our results seen below indicates, this
remains a central question.
The experimental investigation of OH reaction kinetics
referenced in the preceding paragraphs have been carried out
at low temperatures, typically below the desorption tempera-
Temperature programmed desorption ͑TPD͒ spectra
were collected using a quadrupole mass spectrometer ͑QMS͒
equipped with a one inch long tube at the end of the electron
impact ionizer. The beam spot on the Pt͑111͒ sample was
centered approximately 1 mm from the 6 mm diameter tube
to help direct the TPD flux into the QMS ionizer. Surface
Ϫ1
temperature ramp rates were typically ␤ϭ4.2 K s for water
Ϫ1
TPD spectra and 10.5 K s for O TPD spectra. Four mass
2
Ϫ
ture for a monolayer of water (T Ͻ180 K). Anton and Ca-
channels were typically monitored: m/e ϭ32, 20, 19, and
18 corresponding to O , D O , HOD , and H O , ͑also
OD from HDO and D O), respectively. Ion signals mea-
S
ϩ
2
ϩ
ϩ
ϩ
dogan have carried out experiments D /O /Pt͑111͒ at higher
2
2
2 2
ϩ
temperatures (373рT р723 K) using molecular beam re-
S
2
3
3,34
laxation spectroscopy ͑MBRS͒.
Investigations of the
sured in the first three channels were directly proportional to
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.181.251.131 On: Sat, 22 Nov 2014 03:55:10

J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Desorption of OD coadsorbed with HD on Pt(111)
7211
the corresponding neutral molecule number density ͑ioniza-
tion cross-sections and QMS sensitivities for oxygen and the
three water isotopes were assumed to be the same͒. The ion
Ϫ
signal in the m/e ϭ18 channel had contributions from the
ionization/fragmentation of HDO and D O in addition to the
2
ionization of H O. In the TPD spectra reported in this paper,
2
the HDO and D O contributions to this channel were sub-
2
tracted by the following procedure. We measured the crack-
ϩ
ing pattern of pure D O in our QMS and found that the OD
2
ϩ
signal was 30% of the D O channel. Therefore, we scaled
2
Ϫ
the m/e ϭ20 TPD signal by a factor of 0.3 and subtracted it
Ϫ
ϩ
from the m/e ϭ18 channel signal. We assumed the OD
signal from HDO would be half that from D O and corrected
2
Ϫ
further the m/e ϭ18 channel accordingly. The H O TPD
2
signals are labeled with an asterisk to signify these correc-
tions.
ϩ
FIG. 1. D O QMS signal from the thermal desorption of D O from Pt͑111͒
2
2
The Pt crystal was spot-welded to two Pt posts and could
as a function of OD exposure at T ϭ143 K. Panel ͑a͒ shows high exposure
data, while panel ͑b͒ shows lower exposure data. The ramp rate was ␤
S
be heated resistively up to T ϭ1400 K. The temperature of
S
Ϫ1
the crystal was measured with a chromel-alumel ͑type K͒
ϭ3.8 K s .
thermocouple spot-welded to the edge of the crystal. The
ϩ
sample was cleaned by sputtering with Ar ions ͑Ar purity
у99.999%͒, then by cycles of heating the crystal in 5
As might be expected, small amounts of H O desorb from
2
Ϫ7
ϫ10 Torr of O (O purity у99.999%͒ followed by an-
the surface. This provides some confidence in the procedures
2
2
Ϫ
nealing to T ϭ1100 K. This procedure has been shown to
we adopted to correct the m/e ϭ18 mass channel for con-
S
1
7
ϩ
produce clean, well-ordered Pt͑111͒ surfaces previously.
tributions from OD . The O spectra are again similar to
2
those reported previously38 and show the same three features
previously labeled A, B, and C.
Standard surface analysis equipment was unavailable for
these experiments, so the surface quality was inferred from
oxygen and CO TPD spectra and He atom scattering, as de-
scribed previously.38
The OD radicals were the dominant source of oxygen
atoms in these experiments. Identical experiments carried out
with the OD beam blocked showed no significant desorption
Calibrating coverages from exposure to our molecular
beam ͑with O , for example͒ have been technically difficult
2
because the beam source was located two meters from the
crystal ͑necessitated by the hexapole͒. By making some as-
sumptions about the origins and behavior of features in the
O TPD curves we have previously made estimates of oxy-
2
3
8
gen coverage for OD exposed Pt͑111͒. In this paper we
have adopted a simpler approach that we believe may be
more accurate. We assume that the initial sticking probability
of OD on clean Pt͑111͒ is S ϵ1. Then, with knowledge of
0
8
our beam flux densities we can estimate coverage ͓see Eq.
͑6͒, below͔.
Hydrogen (H purity у99.9999%͒ doses were achieved
2
by backfilling the UHV chamber through a variable leak
valve.
III. RESULTS
Figure 1 shows TPD spectra of D O as a function of
2
exposure to a pure beam of OD radicals. These results are
similar to those reported previously.38 One sees that there are
two dominant desorption features, one near T ϳ170 K and
S
another near T ϳ210 K. The peak temperature of both fea-
S
tures increased slightly with coverage. The T ϳ210 K fea-
S
ture grows in first, then appears to saturate while the TS
ϳ170 K feature continues to increase with exposure. At very
low exposures, minor desorption features are seen at TS
ϳ230 K and T ϳ250 K. Figure 2 shows TPD spectra of
S
HOD, H O, and O taken simultaneously with the D O spec-
tra shown in Fig. 1. The HOD spectra appear qualitatively
ϩ
ϩ
2
2
2
FIG. 2. QMS signal from O2 (a), H O* (b), and HOD (c) arising from
2
OD doses of 3.8 ML ͑z͒, 2.6 ML ͑y͒, 1.3 ML ͑x͒, 0.51 ML ͑w͒, 0.26 ML
(v), 0.13 ML ͑u͒, and 0.06 ML ͑t͒ at T ϭ143 K. ␤ϭ3.8 Ks .
S
Ϫ1
similar to the D O, but an order of magnitude less intense.
2
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
31.181.251.131 On: Sat, 22 Nov 2014 03:55:10
1

7212
J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Backstrand et al.
tion of D O from abstraction reactions remaining as an over-
2
layer on the oxygen and/or OD(a) covered surface. At high
OD exposures over 50% of the adsorbed oxygen desorbed
from the surface as D O. This is clearly impossible because
2
the total deuterium atom coverage cannot exceed the oxygen
atom coverage. In determining the relative abundances of
O , D O, and HDO desorbing from the surface we assumed
2
2
that the sensitivity of our mass spectrometer in making ions
Ϫ
of m/e ϭ32, 20, and 19, respectively, was the same. The
QMS sensitivity at mass 32 ͑O ͒ was probably less than we
2
assumed. Therefore, the information in Fig. 3͑b͒ should be
considered qualitative.
Previously28 we have speculated that the T ϳ210 K fea-
S
ture is the same feature as that seen when water is coad-
sorbed with oxygen2
4,26,28
and attributed it to the decompo-
sition of adsorbed hydroxyl intermediates. Although this
feature has been isolated in these previous coadsorption stud-
ies, no TPD analysis to extract kinetic parameters has been
undertaken. For us to isolate this feature and probe the reac-
tion kinetics we adsorbed OD for varying exposures and then
FIG. 3. ͑a͒ Oxygen coverage versus OD exposure for several oxygen con-
taining species on the surface along with the fit to Eq. ͑1͒ for each. The data
labeled ‘‘total’’ corresponds to the sum coverage of all oxygen containing
species per oxygen atom. ͑b͒ The fractions ͑in percent͒ of the total number
of oxygen atoms desorbing into three molecular channels (D O, O , HOD)
drove off the T ϳ170 K feature by heating the sample to
S
2
2
TSϳ185 K. We then cooled the sample back down to TS
ϭ143 K and performed our standard TPD experiment at a
are plotted against OD exposure.
Ϫ1
ramp rate of ␤ϭ4.2 K s . The results of these efforts are
shown in Fig. 4. It is clear that we cleanly removed the TS
from any of the four channels being monitored. Previously
we crudely determined that the initial sticking probability of
OD was approximately unity.38 This also makes intuitive
sense. Assuming that the initial OD sticking probability was
unity we calibrated the coverage corresponding to the inte-
grated desorption signal for each channel as a function of OD
exposure. Figure 3͑a͒ shows this information. The relation-
ship between oxygen coverage in each channel and OD ex-
posure can be parameterized with the following function:
ϳ170 K feature and that none of the adsorbed OD interme-
diates that eventually react at T ϳ210 K were converted to
S
adsorbed D O as the sample was cooled. The D O TPD
2
2
spectra in Fig. 4 show more clearly the slight increase in the
peak temperature as the coverage increases.
To analyze these spectra we adopted the differential
methods described by Koel and co-workers.39 It is clear ͓see
Fig. 4͑d͔͒ that the desorption feature peaking at TSϳ230 K
contributes to the trailing edge of the T ϳ210 K feature.
This made a definitive analysis, particularly with regard to
establishing the order of the reaction, difficult. We do not
ϪJS t/␪
S
0
ϱ͒,
␪
ϭ␪ ͑1Ϫe
͑6͒
ϱ
where ␪ϱ is the saturation O atom coverage for forming the
oxygen containing molecule in the channel being monitored,
know the origin of the T ϳ230 K feature and, because we
S
J is the flux density of the incident OD beam ͑in Pt mono-
have limited surface analysis capabilities, we are reluctant to
speculate. However, in order to improve the analysis of our
Ϫ1
layers s ), and S is the initial sticking probability of OD
0
that goes on to form the molecule of interest. For the total
primary interest, the T ϳ210 K feature, we modeled the TS
S
oxygen, S ϵ1.0 and ␪ ϭ3.8 ML. For D O, S ϭ0.49 and
ϳ230 K feature as a first-order process and found a best-fit
0
ϱ
2
0
1
3 Ϫ1
␪ϱϭ3.4. For O , S ϭ0.41 and ␪ ϭ1.03 ͑these compare fa-
pre-exponential factor of v ϭ5ϫ10 s and an activation
2
0
ϱ
d
3
8
Ϫ1
vorably with the parameters determined previously ͒, and
for HDO, S ϭ0.20 and ␪ ϭ0.21. The fact that the sum of
energy of E ϭ14.5 kcal mol . These parameters repro-
a
duced the higher temperature data well. We then could sub-
0
ϱ
initial sticking probabilities for each channel exceeds 1.0 re-
flects uncertainties in the individual fits to the data. It is also
interesting to consider the fraction of the total oxygen that
desorbs into a particular channel as a function of OD expo-
sure. This information is shown in Fig. 3͑b͒. Qualitatively
we see that initially HDO is a dominant channel, but that as
the OD exposure increases, D O and O become dominant
tract from data the contribution from modeled T ϳ230 K
S
feature thereby better isolating the T ϳ210 K feature for
S
analysis. We would like to emphasize that distinguishing
whether a half-order or first-order reaction model best repre-
sents the T ϳ210 K feature depends most heavily on how
S
the respective models capture the high-temperature tail of
this feature. Unfortunately, this is the very data that is
2
2
channels. Background hydrogen (H ) adsorbs during the 5
muddled by the presence of the T ϳ230 K feature.
2
S
minute sample cooling and we believe can react with oxygen
In Fig. 5 we show a plot of ln͓rate͔Ϫn ln ␪ vs (1/T ),
S
to make OH ͑see Discussion͒. At relatively low exposures
corresponding to the highest coverage TPD curve (␪0
1
͑
ϳ0.25 ML͒ the O desorption channel dominates both D O
ϭ0.25 ML) for reaction orders nϭ and 1. We also tried n
2
2
2
and HDO channels. This we attribute to the direct abstraction
of D atoms from adsorbed OD(a) by OD striking the surface.
As the exposure increases, the desorption feature at TS
ϳ170 K begins to dominate. This we attribute to the desorp-
ϭ0 and 2 with clearly unsatisfactory results. The quality of
1
the fit for nϭ 2 was slightly better than for nϭ1 with a
pre-exponential factor of vdϭ5Ϯ2ϫ1018 molecules
1/2
Ϫ1 Ϫ1
cm
s
and an activation energy of E ϭ11.5
a
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.181.251.131 On: Sat, 22 Nov 2014 03:55:10

J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Desorption of OD coadsorbed with HD on Pt(111)
7213
FIG. 5. Results are shown of the least-squares analysis of Koel, plots of
ln͑rate͒Ϫn ln(␪)͔ versus 1/T , corresponding to the ␪ ϭ0.25 ML TPD
͓
S
0
1
data from Fig. 4͑a͒ for nϭ 2 and nϭ1.
Both models capture the data reasonably well up to tempera-
tures just beyond the peak of the T ϳ210 K feature. One
S
additional weakness of this analysis stems from the relative
uncertainty in the initial coverage. Recall we assumed that
the initial OD sticking probability was unity and have relied8
on our estimates of the flux density of the incident OD beam
to calibrate coverage. The uncertainties we quote in our ki-
netic parameters included these contributions determined by
varying the assumed coverage by Ϯ30% and repeating the
analysis.
To help elucidate the mechanism for water formation we
explored the reactivity of OD exposed Pt͑111͒ with hydro-
gen. Figures 6–9 show TPD spectra for D O, HDO, H O,
FIG. 4. D O desorption rate is plotted as a function of surface temperature
2
from layers of the OD hydroxyl intermediate isolated ͑nearly͒ on Pt͑111͒.
Individual panels correspond to oxygen atom coverages desorbing as
D O(␪ ) as indicated. The dashed ͑half-order͒ and solid ͑first-order͒ lines
2
2
and O , respectively, obtained with the following procedure.
2
2
0
We first exposed the surface to 1.3 ML of OD at TS
ϭ143 K. Then, by filling the UHV chamber with varying
correspond to Polyani–Wigner TPD simulations. Analysis provides kinetic
1
6
18
1/2
Ϫ1
parameter estimates of v ϭ4Ϯ1ϫ10 to 5Ϯ2ϫ10 molecules cm
d
Ϫ1
Ϫ1
s
and E ϭ9.7Ϯ0.1 to 11.5Ϯ0.1 kcal mol for half-order and v ϭ3Ϯ1
a
d
1
2
14 Ϫ1
Ϫ1
ϫ10 to 4Ϯ2ϫ10
s
and E ϭ12.2Ϯ0.1 to 14.4Ϯ0.1 kcal mol for
first order ͑see text and Fig. 5 for details͒. ␤ϭ4.2 K s
a
Ϫ1
.
Ϯ0.1 kcal molϪ1 for nϭ , and v ϭ4Ϯ2ϫ10 s and Ea
1
2
14 Ϫ1
d
Ϫ1
ϭ14.4Ϯ0.1 kcal mol for nϭ1. A similar analysis was per-
formed for all the TPD spectra shown in Fig. 4. The pre-
1
6
exponential factor ranges from v ϭ4Ϯ1ϫ10 to 5Ϯ2
d
1
8
1/2
Ϫ1 Ϫ1
ϫ10 molecules cm
s
and E ϭ9.7Ϯ0.1 to 11.5
Ϯ0.1 kcal mol for nϭ 2 as the initial coverage is varied
a
Ϫ1
1
from ␪ ϭ0.04 to 0.25 ML. It is interesting to note that the
0
␪0
ϭ0.25 ML maximum oxygen atom coverage that we see
desorbing as D O at T ϳ210 K was approximately half the
2
S
maximum coverage obtained from the coadsorption of water
4
0
and oxygen.
1
The dashed and solid curves ͑for nϭ 2 and 1, respec-
tively͒ in Fig. 4 indicate simulated TPD curves using the
standard Polyani–Wigner equation with these parameters:
ϩ
FIG. 6. D O TPD spectra from 1 ML OD doses at T ϭ143 K, followed by
2
S
Ϫd␪
ϪE
a
variable H exposures. The solid vertical line shows a peak growing in at
T ϭ178 K. ␤ϭ3.8 K s .
S
2
Rateϭ
ϭv ␪Љ expͩ ͪ
.
͑7͒
d
Ϫ1
dt
kT
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.181.251.131 On: Sat, 22 Nov 2014 03:55:10

7214
J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Backstrand et al.
ϩ
ϩ
FIG. 7. HOD TPD spectra measured concurrently with D O data of Fig.
6
ϩ
2
FIG. 9. O2 TPD spectra arising from 1 ML OD doses at 143 K, followed by
. The solid vertical line is at 178 K.
Ϫ1
variable H exposures. The ramp rate was ␤ϭ10.4 K s
.
2
pressures of H for a fixed time of 3 minutes, we varied the
hydrogen exposure. Figures 6–9 show the resulting TPD
2
Fig. 12 below͒. Figure 8 shows similar behavior in the H O
2
spectrum. The hydrogen appears to be driving the reaction of
adsorbed OD intermediates that decompose in its absence at
T ϳ210 K to form D O, HOD, and H O.
spectra as a function H exposure. The spectra corresponding
2
to ‘‘no H ’’ admitted to the chamber are qualitatively the
2
S
2
2
same as those obtained from comparable OD exposures
It is also interesting to compare the D O, HOD, and H O
2
2
shown in Figs. 1 and 2. Figure 6 shows the D O spectra with
2
TPD spectra corresponding to a 0.3 LH exposure with those
2
the same features at T ϳ170 K and T ϳ210 K discussed
S
S
from ‘‘no H .’’ It’s important to keep in mind that back-
2
previously. As the H2 exposure was increased, the TS
ϳ210 K feature disappeared and a shoulder developed ͑as
ground H provides some unavoidable exposure even for the
2
‘
‘no H ’’ spectra. Two features are worth noting in these
2
highlighted by the solid vertical line at T ϳ178 K) at a tem-
S
spectra. First, the T ϳ210 K feature in the HDO and H O
S
2
perature slightly greater than the peak of the T ϳ170 K fea-
S
channels initially increases in intensity with H exposure.
2
ture. We see submonolayer coverages of D O desorb at the
2
2
4,40
And second, an additional feature seems to appear in all
same temperature, as has been seen previously by others.
water desorption channels near T ϳ195 K after a 0.3 LH2
S
Figure 7 shows the corresponding HOD spectra. Here again
exposure. This feature was reproduced in repetitions of this
experiment.
In Fig. 10, significant information gleaned from the data
in Figs. 6–9 is collectively presented. Figure 10͑a͒ shows the
integrated area under each of the TPD spectra shown in Figs.
we see the small T ϳ210 K feature disappear while a sub-
S
stantial amount of HOD begins to desorb at lower tempera-
tures; later we will argue that this occurs at the same tem-
perature as the shoulder in Fig. 6 and not at T ϳ170 K ͑see
S
6
–9 from T ϭ150 K to 215 K as well as the total. The total
S
oxygen atom coverage from all molecular sources decreases
with increasing H exposure, primarily due to depletions in
2
the O and D O channels. This suggests that chemistry is
2
2
occurring between adsorbed hydrogen and OD that leads to
desorption of some oxygen containing species ͑most prob-
ably water͒ at the low dosing temperature, T ϭ143 K. It can
S
be seen that the depletion of D O is accompanied by slight
2
increases in the amount of HOD and H O desorbing from the
2
surface. Figure 10͑b͒ shows this in greater detail. Here, we
have considered separately the water desorbing between TS
ϭ150 K and 190 K ͑peak 1͒ and that desorbing between
TSϭ190 K and 215 K ͑peak 2͒. We have not tried to dis-
criminate contributions to the D O channel from the shoulder
2
(
TSϳ178 K͒ discussed in the previous paragraph and the
prominent D O feature in Fig. 1 (T ϳ170 K); both are in-
2
S
cluded as peak 1. Figure 10͑b͒ shows clearly that adsorbed
hydroxyl intermediates that, in the absence of hydrogen, de-
ϩ
composed near T ϳ210 K ͑peak 2͒ would, in the presence of
FIG. 8. H O* TPD spectra measured concurrently with D O data of Fig. 6.
S
2
2
The solid vertical line is at 178 K.
hydrogen, react to form all three isotopes of water (D O,
2
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
31.181.251.131 On: Sat, 22 Nov 2014 03:55:10
1

J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Desorption of OD coadsorbed with HD on Pt(111)
7215
FIG. 10. ͑a͒ Oxygen atom coverage ͑ML͒ versus H2 exposure ͑L͒ for all
oxygen containing species on the surface after a 1.3 ML OD exposure at
T ϭ143 K followed by variable H exposures. The data labeled ‘‘total’’
FIG. 11. ͑a͒ Coverage ͑ML͒ versus H exposure ͑L͒ for all oxygen contain-
2
ing species on the surface after variable H exposures at 143 K followed by
2
S
2
a 1.1 ML OD exposure. The data labeled ‘‘total’’ corresponds to the sum
coverage of all oxygen containing species per oxygen atom. ͑b͒ Same as ͑a͒
except the coverage is plotted for two temperature regions: 150 KϽTS
corresponds to the sum coverage of all oxygen containing species per oxy-
gen atom. ͑b͒ Same as ͑a͒ except the coverage is plotted for two temperature
regions: 150 KϽT Ͻ190 K ͑feature 1͒ and 190 KϽT Ͻ215 K ͑feature 2͒
S
S
Ͻ190 K ͑feature 1͒ and 190 KϽT Ͻ215 K ͑feature 2͒ for the three water
for the three water isotopes.
S
channels.
coadsorbing oxygen and water.24,26,28 A striking distinction
HDO, and H O) at lower temperatures ͑peak 1͒.
2
Experiments similar to those characterized in Figs. 6–10
between these methods appears in the O TPD spectra ͓Fig.
2
were carried out except with the OD/H dosing order re-
versed. That is, the Pt͑111͒ sample was exposed first to vary-
2͑a͔͒. When oxygen and water are coadsorbed and excess
oxygen remains on the surface it desorbs as type-C oxygen
or possibly types A and C.41 However, when OD intermedi-
ates react on the surface we see all three oxygen types, types
A, B, and C. In previous studies of the reactivity of adsorbed
2
ing pressures of H and then to 1.1 ML of the OD beam.
2
Figure 11 shows a summary of significant information analo-
gous to that shown in Fig. 10. Qualitatively the behavior was
very similar to that shown in Figs. 6–10. However, some
differences are worth noting. The shoulder from H atom
OD with CO we demonstrated that, at T Ͼ250 K, types A
S
and C oxygen could be selectively titrated from the surface
leaving behind exclusively type-B oxygen ͑see Fig. 5 of Ref.
38͒. Note that this selective reactivity is not evident at tem-
peratures below the OD intermediate decomposition tem-
driven reactions to form D O was better resolved than the
2
shoulder in Fig. 6. Figure 12 shows an example of this. Here
the D O ͑solid line͒ and HDO ͑dashed line͒ TPD spectra are
2
shown resulting from a 1.5 L exposure of H followed by a
perature, T ϳ210 K ͑e.g., see Figs. 9 and 14 from this paper
2
S
1
.1 ML OD exposure. Not only is the T ϳ178 K shoulder
and Fig. 6b from Ref. 38͒. A similar selective reactivity is
seen with hydrogen at T Ͼ210 K. The O TPD spectra
S
feature better resolved in the D O spectrum, but a compari-
2
S
2
son of the two spectra provides convincing evidence that this
shoulder results from the same kinetic process as the low
temperature HDO feature, namely the desorption of sub-
monolayer coverages of water bound to Pt. Another differ-
ence is that the total oxygen coverage is less when hydrogen
is exposed first, particularly at low coverages. Results de-
scribed below ͑Fig. 13͒ suggest that OD may abstract hydro-
gen from the H atom covered surface. After H atoms have
been removed, OD can adsorb as in Fig. 1. The entire sample
shown in Fig. 13 demonstrate this. In Fig. 13͑a͒ a TPD spec-
was exposed to H , but only the center portion to OD. The
2
mobility of hydrogen is high, and ample amounts from areas
surrounding the beam spot exist. This would explain the
similarities when the dosing order was reversed.
ϩ
ϩ
FIG. 12. D O and HOD TPD spectra arising from 1.5 L H doses fol-
2
2
Previously we pointed out the similarity between the
Ϫ1
lowed by 1.1 ML, OD exposures. The ramp rate was ␤ϭ4.2 K s . The
D O TPD spectra shown in Fig. 1 and those obtained from
2
solid vertical line at 178 K shows the new peak produced with H exposure.
2
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
31.181.251.131 On: Sat, 22 Nov 2014 03:55:10
1

7216
J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Backstrand et al.
cept that approximately 10–20 seconds after the OD beam
Ϫ7
was turned on, 1ϫ10 Torr of H was admitted to the UHV
2
chamber for the 5 minute duration of the OD exposure ͑ap-
proximately 30 L of H ). Here we see a cleanly isolated
2
spectrum of type-B oxygen. Finally, we carried out a similar
experiment to that described in Fig. 13͑b͒ except this time
the H was introduced to the chamber approximately 10–20
2
seconds before the OD beam was turned on. Figure 13͑c͒
shows the resulting TPD spectrum with no detectable O2
desorbing from the surface. This we found extremely inter-
esting. Apparently with hydrogen present on the surface OD
radicals will not react and form a stable near-surface oxide.
IV. DISCUSSION
We discuss our results in the context of a reaction model
that can explain the observed isotopic distributions of water
desorbing at various temperatures and the O TPD features.
2
We propose the following. Deuterated hydroxyl radicals
from the gas phase can adsorb on Pt͑111͒ in one of two types
of sites: a threefold hollow site ͑a ␮3 site͒42 and either a
bridging site or an atop site. One site creates adsorbed OD
with anionic, basic chemical properties, the other with acidic
properties. For the moment, let us imagine the acidic site lies
ϩ
FIG. 13. O2 TPD spectra obtained after ͑a͒ dosing the surface with 0.9 ML
OD, ͑b͒ dosing the surface with H beginning just after the start of a 0.9 ML
OD exposure, and ͑c͒ dosing the surface with H before and continuously
after a 0.9 ML, OD exposure.
2
2
in the ␮ site and the basic site at an adjacent atop ͑or bridg-
3
ing͒ site as shown in Fig. 14. An identical arrangement can
be made by coadsorbing an oxygen atom in a threefold hol-
low site with deuterated water that dissociates, donating a
trum is shown from a clean Pt͑111͒ surface exposed to 0.9
ML ͑5 minute exposure time͒ of OD at T ϭ223 K. Although
the type-A oxygen is not well resolved from the type-B oxy-
gen, all three types are evident from the shape of the peak.
See for comparisons Figs. 2 and 4 of Ref. 38. Figure 13͑b͒
shows the TPD spectrum from an identical experiment, ex-
deuterium to the preabsorbed oxygen atom and binding via
S
24,26,28
it’s oxygen atom to the basic site.
However, adsorbing
small amounts of deuterium on an oxygen saturated surface
␪0ϭ0.25 ML) will not produce acid-base pairs at low tem-
(
peratures, but only OD in ␮ sites, a configuration we will
3
identify as the OD͑II͒ intermediate. The acid-base pairs are
strongly correlated, being stabilized by hydrogen bonding,
and we will identify them as the OD͑I͒ intermediate. This
view is supported by the results of White and co-workers
who found the EELS spectra of OH͑II͒ (␮ sites only͒ to
3
resemble ͑Pt͑OH͒Me ͒ , and OH͑I͒ ͑both atop and ␮ sites͒
3
4
3
3
1
to resemble K Pt͑OH͒ . The basic OD groups ͑bridging or
2
6
atop sites͒ are relatively mobile compared with the OD in
acidic sites, and pairs will tend to form islands of OD͑I͒ with
other regions of the surface containing only OD in acidic
sites, OD͑II͒. Reactions modeled as occurring at island edges
1
2
result in -order kinetics, the nominal behavior we see from
the analysis of the data in Fig. 4. In the absence of adsorbed
atomic hydrogen, OD͑I͒ pairs will disproportionate near TS
ϭ210 K producing a deuterated water molecule that imme-
diately desorbs and predominantly what we refer to as type-A
oxygen. It is significant that we have found that a similar
TSϳ210 K TPD feature appears even at low doses of hy-
droxyl radicals ͓Ͻ0.1 ML, see Fig. 1͑b͔͒. This suggests that
the intermediate formed by the adsorption of hydroxyl radi-
FIG. 14. Panel ͑a͒ shows a sketch of a possible structure of the intermediate
layer OD͑I͒. The large white circles represent a Pt͑111͒ surface, the smaller
white circles represent O atoms and the small black circles D atoms. Hy-
droxyls are located in three-fold hollow sites ͑A͒ and atop sites ͑B͒. Panel
cals is probably OH͑I͒. We also see D O TPD features at
2
higher temperatures for low exposure that diminish at higher
exposures. These we attribute to desorption from areas on
the surface with only one type of OD adsorption site popu-
lated, i.e., the OD͑II͒ intermediate. The activation barrier for
͑b͒ shows a cross section of panel ͑a͒ along axis A–B. Hydrogen bonding is
possible in OD͑I͒ intermediate but not in OD͑II͒ intermediate shown in
panels ͑c͒ and ͑d͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
31.181.251.131 On: Sat, 22 Nov 2014 03:55:10
1

J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Desorption of OD coadsorbed with HD on Pt(111)
7217
disproportionation from the less-densely OD͑II͒-configured
isotope distribution. We think the following may be happen-
ing. We believe the cross section for gas-phase OD to ab-
stract a deuterium atom from the OD͑I͒ intermediate is fairly
large. As a consequence, a 1.3 ML OD exposure results in
several surface moieties: OD͑I͒, D-atom depleted OD͑I͒, and
PtO with associated overlayers of D O desorbing at T
surface is higher than for the OD͑I͒ pairs resulting in D O
2
desorption at higher temperatures.
Further exposure of an OD͑I͒ covered surface to gas
phase OD radicals results in D-atom abstraction and an in-
creased fraction of oxygen desorbing as O ͑types A and C͒
2
x
2
S
rather than water. This D O abstraction product could be
ϳ170 K. Surface H atoms can react with OD͑I͒ at low tem-
peratures to make isotopically scrambled water desorbing at
TSϭ178 K. They can also re-hydrogenate D-atom depleted
OD͑I͒ and, with limited H atoms available, this reconstituted
OD/OH͑I͒ intermediate will decompose at TSϳ210 K. That
is why, with background H2 exposures, we always see some
HOD coming off the surface at TSϳ210 K ͓see Fig. 2͑c͔͒, a
TPD feature attributed to disproportionation. Also, the total
oxygen desorbing into all four of the channels we monitor in
our TPD experiments decreases indicating that an additional
2
released into the vacuum resulting in a sticking probability
less than unity, which we see. Continued exposure of the
D-atom depleted OD͑I͒ surface to the strong oxidizing power
of OD radicals will further oxidize the Pt surface to make a
more stable platinum oxide (PtO ) that decomposes in a
x
sharp TPD feature at T ϭ735 K type-B oxygen. Once
S
formed, the PtO oxide is relatively resistant to chemical
x
attack, either by adsorbed hydrogen or carbon monoxide.
Deuterium atoms incorporated at the increasingly PtO -like
x
desorption process takes place at T Ͻ150 K. These are is-
surface by the continuing exposure to OD can be abstracted
to form deuterated water that can remain on the surface.
Desorption from such an environment could well be ex-
pected to be different than desorption from bare Pt(TS
S
sues that we are addressing in ongoing experiments.
When the surface is dosed with OD at temperatures TS
Ͼ210 K ͑specifically, T ϭ275 K), type-B oxygen becomes
S
3
8
the dominant species at high exposures. When the surface
ϳ180 K) or from ice multilayers (T ϳ165 K). This is con-
S
is dosed at temperatures below T Ͻ210 K, all three types of
sistent with the observed T ϳ170 K feature that we attribute
S
S
oxygen appear to be formed. In Fig. 13͑a͒ we see the result
to this process. Exposure of the Pt surface to OD radicals
above the decomposition temperature of the OD͑I͒ interme-
of dosing the crystal with OD at T ϭ225 K. Type-B oxygen
S
appears to dominate just as it did at T ϭ275 K. Then, by
diate (T Ͼ210 K) increases the number of sites available for
S
S
dosing with OD for the same exposure, but adding a static H2
PtO formation and increases the total surface capacity for
x
background just after dosing began, we see the complete loss
oxygen, primarily due to growth of PtO that decomposes as
x
of A-and C-type oxygen ͓Fig. 13͑b͔͒ due to reactions with
hydrogen. We saw the same behavior as a result of reactions
with CO.38 It would appear that B-type oxygen is readily
type-B oxygen.
Now consider the consequences of exposing a OD-dosed
surface to molecular hydrogen ͑H ͒. The dissociative adsorp-
2
formed at a surface temperature of T ϭ225 K and that this
tion of H on bare Pt and oxygen covered Pt has been studied
S
2
‘‘Pt-oxide’’ is relatively resistant to chemical attack by hy-
extensively ͑see Ref. 15, and references therein͒. It is known
that H atoms adsorb in threefold hollow sites and are quite
drogen or CO. However, below T Ͻ210 K the oxygen that
S
eventually forms and desorbs as A, B, and C types are all
diminished equally by chemical attack from hydrogen ͑this
study͒ as well as CO.38 We conclude that, chemically, some-
thing critical may be happening to adsorbed oxygen when
mobile at T Ͼ100 K. Imagine a Pt͑111͒ surface that has
S
been exposed to OD radicals in order to form the OD͑I͒
intermediate described above. We believe that both OD ad-
sorption sites can react with adsorbed H-atoms at relatively
low temperatures resulting in HOD adsorbed at preferred
the OD͑I͒ intermediate decomposes at T ϳ210 K. Further-
S
2
3
more, for OD exposures above T Ͼ210 K, if hydrogen is
S
atop sites. Verheij might argue that this hydrogen-addition
step must occur at a reactive site and that, once incorporated
into the OD͑I͒ intermediate the oxygen bound hydrogen
could easily migrate throughout the intermediate via the hy-
drogen bonding network. Germer and Ho have estimated the
present on the surface no surface oxygen is seen ͓Fig. 13͑c͔͒.
It appears that the presence of hydrogen prevents the growth
of Pt oxides. Similar experiments were not attempted with
CO.
activation energy for forming water from H adatoms reacting
Ϫ1 32
V. CONCLUSION
with adsorbed OH to be E ϭ4.2 kcal mol
.
This seems
a
consistent with our observation that, in the presence of ex-
Finally, we would like to highlight the most significant
observations and conclusions drawn from our studies. The
cess hydrogen, the water production process at T ϳ210 K is
S
sacrificed in favor of a process that generates submonolayer
reaction of OD on Pt͑111͒ is marked by two dominant D O
2
water that desorbs at T ϳ178 K. Complete isotopic mixing
TPD features, at T ϳ170 K (PtO -bound D O desorption͒
S
S
x
2
1
0
occurs readily among coadsorbed H O and D O, so we
and T ϳ210 K ͓reaction limited decomposition of OD͑I͒ in-
2
2
S
would expect a statistical distribution of D O:HDO:H O
termediate͔. The latter appears to be the same as that seen
when coadsorbing water and oxygen. We have explored the
kinetics of this feature for the first time and determined the
2
2
͑1:2:1͒ from hydrogentated OD͑I͒ upon desorption at TS
ϳ178 K, the typical temperature for submonolayer water de-
sorption. We do not see this behavior.
1
8
1/2
reaction to be half-order (v ϭ5Ϯ2ϫ10 molecules
d
Ϫ1 Ϫ1
Ϫ1
Our work on this system is continuing and preliminary
results from experiments involving lower exposures to OD
than the 1.3 ML OD dose corresponding to Figs. 6–9 pro-
vide some insight.41 When a 0.2 ML OD-dosed surface is
exposed to H , we see the H O isotope dominate the water
cm
s
and E ϭ11.5Ϯ0.1 kcal mol for ␪ ϭ0.25 ML) in
a
0
total oxygen coverage. Influenced by the Verheij reactive site
model, we speculate that the half-order rate limiting event
corresponds to the concerted breaking of a deuterium–
oxygen bond somewhere within the OD͑I͒ layer along with
2
2
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.181.251.131 On: Sat, 22 Nov 2014 03:55:10

7218
J. Chem. Phys., Vol. 112, No. 16, 22 April 2000
Backstrand et al.
9
0
P. R. Norton, The Chemical Physics of Solid Surfaces and Heterogenous
Catalysis, edited by D. A. King and D. P. Woodruff ͑Elsevier, Amster-
dam, 1982͒, Vol. 4, Chap. 2, pp. 27–72.
the formation of a Pt–D O moiety at the perimeter of the
2
OD͑I͒ intermediate layer. A kinetically distinct reaction pro-
cess occurs if hydrogen atoms are available to react with the
OD͑I͒ intermediate. In the presence of H atoms the reaction
1
B. A. Sexton, Surf. Sci. 94, 435 ͑1980͒.
11
R. Lewis and R. Gomer, Surf. Sci. 12, 157 ͑1968͒.
J. Wintterlin, R. Schuster, and G. Ertl, Phys. Rev. Lett. 77, 123 ͑1996͒.
H. Steininger, S. Lehwald, and H. Ibach, Surf. Sci. 123, 1 ͑1982͒.
M. B. Hugenschmidt, L. K. Verheij, M. K. Freitag, B. Poelsema, and G.
Comsa, Surf. Sci. Lett. 259, L763 ͑1991͒.
L. K. Verheij and M. B. Hugenschmidt, Surf. Sci. 324, 185 ͑1995͒.
B. Poelsema, L. K. Verheij, and G. Comsa, Surf. Sci. 152Õ153, 496
͑1985͒.
1
1
1
2
3
4
limited production of water (D O, HDO, and H O) from
2
2
OD͑I͒ occurs at T Ͻ178 K. Isotopic product ratios suggest
S
complete isotopic mixing of the hydrogen isotopes, but the
data are not definitive.
15
We proposed a model that two adsorption sites exist for
16
adsorbed OD(a) ,␮ and atop ͑or bridging͒ and that gas phase
3
1
1
7
8
OD radicals abstract D atoms from adsorbed OD(a) . We
speculate that the OD͑I͒ intermediate consists of both OD
sites populated and can be generated by adsorption of OD
C. T. Campbell, G. Ertl, H. Kuipers, and J. Segner, Surf. Sci. 107, 220
͑1981͒.
A. C. Luntz, M. D. Williams, and D. S. Bethune, J. Chem. Phys. 89, 4381
͑
1988͒.
and by coadsorption of D O on oxygen covered Pt. We fur-
2
19
20
A. Cudok, H. Froitzheim, and G. Hess, Surf. Sci. 307–309, 761 ͑1994͒.
A. N. Artsyukhovich, V. A. Ukraintsev, and I. Harrison, Surf. Sci. 347,
ther speculate that adsorbed hydroxyls can also be generated
by the reaction of limited quantities of hydrogen with ad-
303 ͑1996͒.
4
1
21
sorbed oxygen, but that only ␮ sites are populated under
D. H. Parker, M. E. Bartram, and B. E. Koel, Surf. Sci. 217, 489 ͑1989͒.
N. A. Saliba, Y.-L. Tsai, C. Panja, and B. E. Koel ͑unpublished͒.
P. A. Thiel and T. E. Madey, Surf. Sci. Rep. 7, 211 ͑1987͒.
G. B. Fisher and J. L. Gland, Surf. Sci. 94, 446 ͑1980͒.
K. M. Ogle and J. M. White, Surf. Sci. 139, 43 ͑1984͒.
G. B. Fisher, J. L. Gland, and S. J. Schmieg, J. Vac. Sci. Technol. 20, 518
3
2
2
2
2
2
3
4
5
these conditions, OD͑II͒, resulting in disproportionation TPD
features at higher temperatures (T Ͼ210 K). At low expo-
S
sures of OD, disproportionation TPD features were observed
from both OD͑I͒(T ϳ210 K) and OD͑II͒(T Ͼ210 K) areas
26
S
S
of the surface. Ongoing experiments are underway that will
hopefully fortify this interpretation of our data.41
͑
1982͒.
27
J. L. Gland, G. B. Fisher, and E. B. Kollin, J. Catal. 77, 263 ͑1982͒.
G. B. Fisher and B. A. Sexton, Phys. Rev. Lett. 44, 683 ͑1980͒.
G. E. Mitchell, S. Akhter, and J. M. White, Surf. Sci. 166, 283 ͑1986͒.
J. R. Creighton and J. M. White, Surf. Sci. 136, 449 ͑1984͒.
G. E. Mitchell and J. M. White, Chem. Phys. Lett. 135, 84 ͑1987͒.
T. A. Germer and W. Ho, Chem. Phys. Lett. 163, 449 ͑1989͒.
A. B. Anton and D. C. Cadogan, Surf. Sci. 239, L548 ͑1990͒.
A. B. Anton and D. C. Cadogan, J. Vac. Sci. Technol. A 9, 1890 ͑1991͒.
P. C. Engelking, Rev. Sci. Instrum. 57, 2274 ͑1986͒.
2
2
3
3
3
3
3
3
8
9
0
1
2
3
4
5
ACKNOWLEDGMENTS
We gratefully acknowledge funding for this work from
the NSF ͑Grant No. CHE-9727512͒. We also thank Laurens
Verheij for valuable discussions.
1
L. K. Verheij, M. B. Hugenschmidt, B. Poelsema, and G. Comsa, Chem.
36
R. B. Bernstein, Chemical Dynamics via Molecular Beam and Laser
Techniques ͑Oxford University Press, New York, 1982͒.
J. Reuss, in Atomic and Molecular Beam Methods, edited by G. Scoles
Phys. Lett. 174, 449 ͑1990͒.
L. K. Verheij, M. B. Hugenschmidt, L. C o¨ lln, B. Poelsema, and G.
2
37
Comsa, Chem. Phys. Lett. 166, 523 ͑1990͒.
G. E. Gdowski and R. J. Madix, Surf. Sci. 119, 184 ͑1982͒.
L. K. Verheij, M. Freitag, M. B. Hugenschmidt, I. Kempf, B. Poelsema,
͑
Oxford University Press, New York, 1988͒, Vol. I, Chap. XI.
3
38
4
M. A. Weibel, K. M. Backstrand, and T. J. Curtiss, Surf. Sci. 444, 66
2000͒.
͑
and G. Comsa, Surf. Sci. 272, 276 ͑1992͒.
L. K. Verheij, Surf. Sci. 371, 100 ͑1997͒.
L. K. Verheij and M. B. Hugenschmidt, Surf. Sci. 416, 37 ͑1998͒.
B. Sieben, G. Bozdech, and N. Ernst, J. Phys. IV 6, 5 ͑1996͒.
3
4
9
0
5
D. H. Parker, M. E. Jones, and B. E. Koel, Surf. Sci. 233, 65 ͑1990͒.
J. R. Creighton and J. M. White, Chem. Phys. Lett. 92, 435 ͑1982͒.
K. M. Backstrand and T. J. Curtiss ͑unpublished͒.
M. Chen, S. P. Bates, R. A. van Santen, and C. M. Friend, J. Phys. Chem.
B 101, 10051 ͑1997͒.
6
7
41
42
8
T. D. Hain, M. A. Weibel, K. M. Backstrand, and T. J. Curtiss, J. Phys.
Chem. A 101, 7674 ͑1997͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
31.181.251.131 On: Sat, 22 Nov 2014 03:55:10
1