Adsorption and reaction of sulfur dioxide with Cu(110) and Cu(110)-p(2×1)-O

Article abstract of DOI:10.1063/1.1450545

Adsorption and reaction of sulfur dioxide with Cu(110)-p(2×1)-O were discussed. On the Cu(110)-p(2×1)-O surface, the nucleation and growth of the sulfite structures at p(2×1-O island boundaries was apparently accompanied by a structural relaxation. The fo

Full text of DOI:10.1063/1.1450545

Adsorption and reaction of sulfur dioxide with Cu(110) and Cu (110)-p(2×1)- O
Ali R. Alemozafar, Xing-Cai Guo, and Robert J. Madix
Citation: The Journal of Chemical Physics 116, 4698 (2002); doi: 10.1063/1.1450545
View online: http://dx.doi.org/10.1063/1.1450545
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/116/11?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Oxygen adsorption-induced nanostructures and island formation on Cu{100}: Bridging the gap between the
formation of surface confined oxygen chemisorption layer and oxide formation
J. Chem. Phys. 129, 124703 (2008); 10.1063/1.2980347
Adsorption of α-pyridone on Cu(110)
J. Chem. Phys. 116, 8988 (2002); 10.1063/1.1471244
Adsorption, decomposition, and stabilization of 1,2-dibromoethane on Cu(111)
J. Vac. Sci. Technol. A 19, 1474 (2001); 10.1116/1.1376702
Adsorption of hydrogen on clean and potassium modified low index copper surfaces: Cu(100) and Cu(110)
J. Vac. Sci. Technol. A 19, 1988 (2001); 10.1116/1.1359552
Elevated temperature scanning tunneling microscopy study of formic acid adsorption and reaction on oxygen
(2×1) covered Cu(110)
J. Chem. Phys. 108, 6916 (1998); 10.1063/1.476106
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

JOURNAL OF CHEMICAL PHYSICS
VOLUME 116, NUMBER 11
15 MARCH 2002
Adsorption and reaction of sulfur dioxide
with Cu„110… and Cu„110…-p„2Ã1…-O
Ali R. Alemozafar, Xing-Cai Guo, and Robert J. Madixa)
Department of Chemical Engineering, Stanford University, Stanford, California 94305-5025
͑Received 24 September 2001; accepted 20 December 2001͒
On Cu(110)-p(2ϫ1)-O at 300 K SO2(g) reacts stoichiometrically with O͑a͒ to form a surface
covered with both c(4ϫ2)-SO3 and p(2ϫ2)-SO3 structures. With heating SO2(g) evolves from
the surface in distinct reaction-limited states at 384 K, 425 K, and 470 K, and the surface reverts to
its initially oxidized state. On Cu͑110͒, SO2(g) adsorbs molecularly below 300 K; upon annealing
to 300 K, the sulfur dioxide disproportionates according to 3SO2(a)→S(a)ϩ2SO3(a) with
concomitant desorption of excess SO2(a). The surface formed in this manner exhibits large c(2
ϫ2)-S domains which encompass scattered c(4ϫ2)-SO3 and p(2ϫ2)-SO3 structures in a 1:2
coverage ratio. After being annealed to 400 K, the surface exhibits large p(2ϫ2)-SO3 domains
surrounding smaller c(4ϫ2)-SO3 and c(2ϫ2)-S islands. Continued heating past 400 K results in
decomposition of sulfite according to SO3(a)→SO2(g)ϩO(a), evolving sulfur dioxide at 470 K
and leaving the surface covered with atomic sulfur and oxygen. Real-time STM images show the
mobility of oxygen at island boundaries and the mobility of sulfite amid the p(2ϫ1)-O structures.
STM measurements suggest that the sulfite occupy four-fold hollow sites. © 2002 American
Institute of Physics. ͓DOI: 10.1063/1.1450545͔
I. INTRODUCTION
spontaneous and dissociative, while molecular adsorption
dominates on Ag.3,4
Sulfur dioxide (SO2) is infamous for its role as an envi-
ronmental pollutant. Understanding its interactions with
metal surfaces may provide fundamental knowledge useful
for the development of catalysts for the reduction of air pol-
lution in industrial and urban settings. Emissions in the
United States alone are 1ϫ1010 Kg/year, though this figure
is down nearly 29% from what it was in the early 90’s due to
strict emissions regulations.1 Sulfur dioxide is a major con-
stituent of the flue gas from coal-fired power plants and is
readily oxidized by water in the atmosphere into sulfuric
acid, which produces acid rain.
In addition to its deleterious environmental effects, by
taking up available active sites on metal surfaces, SO2 , and
possibly its decomposition product sulfur, is a well known
catalyst poison. Therefore, further advances in our under-
standing of this interesting molecule and its bonding and
reactions with different metals may also aid the production
of better flue gas desulfurization units2 and help avoid pit-
falls in the design of efficient catalysts.
Over the past 20 years the adsorption and reaction of
SO2(g) has been investigated on a number of single crystal
metal surfaces with a number of powerful surface analytical
techniques, including x-ray photoelectron spectroscopy
͑XPS͒, near-edge x-ray absorption fine structure spectros-
copy ͑NEXAFS͒, temperature-programmed reaction spec-
troscopy ͑TPRS͒, infrared reflection absorption spectroscopy
͑IRAS͒, and, more recently, scanning tunneling microscopy
͑STM͒. In general, results suggest that SO2(g) adsorption on
metal surfaces like Fe, Rh, W, Ni, Pd, Pt, Cu, and Zn is
Recently, the interaction of SO2(g) with the Cu͑100͒
and Cu͑110͒ surfaces has been examined in some detail us-
ing XPS, TPRS, and STM. An XPS and TPRS investigation
by Polcik et al. has indicated that SO2(g) dissociates on
Cu͑100͒ to yield SO(a) and O(a), which are stable up to
300 K, at which temperature SO(a) and O(a) recombine to
yield SO2(g).5 In subsequent XPS studies with Cu͑100͒
Ohta et al. suggested that SO2(g) disproportionates into
sulfite (SO3) and sulfur. NEXAFS studies indicate that the
SO3(a) coordinates via the S at four-fold hollow sites. From
STM results they suggested local distributions of the SO3(a)
and S(a) on the surface.6 Recently Woodruff et al. in their
XPS and NEXAFS investigation of the interactions of SO2
with Cu͑111͒ suggested that SO2(g) disproportionates on the
surface into sulfur and SO3(a). It was further suggested that
the sulfite occupies atop sites, oriented with the oxygen at-
oms pointing toward the surface.7
Despite the relatively extensive work on Cu͑100͒ and
Cu͑111͒, little has been done to understand how SO2(g) in-
teracts with the Cu͑110͒ surface. Recently Pradier et al.
found using IRAS that in the presence of oxygen SO2(g)
͑600–5400 langmuirs exposure͒ forms SO3(a) and sulfate
on the Cu͑110͒ surface at room temperature; on the surface
exposed to 200 langmuirs of background oxygen, presumed
to form a c(6ϫ2) oxygen adlayer, both sulfate and sulfite
were observed, the sulfate being the dominant species; on the
surface exposed to sufficient oxygen to form a Cu2O film,
sulfite resulted from the reaction with SO2(g).8 Moreover, it
was suggested that in the absence of adsorbed oxygen
SO2(g) undergoes complete decomposition, presumably into
S(a) and O(a).
a͒
Electronic mail: rjm@chemeng.stanford.edu
0021-9606/2002/116(11)/4698/9/$19.00
4698
© 2002 American Institute of Physics
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Adsorption and reaction of sulfur dioxide
4699
Despite the conclusions reached by the latter investiga-
tion, further work is needed to determine the local distribu-
tion of surface species following the interaction of SO2(g)
with the Cu͑110͒ and Cu(110)-p(2ϫ1)-O surfaces. In this
paper we report the results of a coordinated study of the
reaction of SO2(g) with clean and oxygen-covered Cu͑110͒
by STM, XPS, and TPRS. The mechanism of the reaction
and the distribution of structures produced by the reaction
products are clarified.
of the foil. The temperature was monitored by a Chromel-
Alumel thermocouple spot welded to the Ta foil at a location
directly behind the crystal. The surface was cleaned by three
sputter-anneal cycles, with the first anneal done in an O2
atmosphere at 10Ϫ7 Torr. Surface cleanliness and composi-
tion were probed with XPS using nonmonochromatic MgK␣
x-rays. The photoelectrons were collected normal to the sur-
face by the energy analyzer utilizing a 25 eV pass energy.
Binding energies were calibrated with respect to the Au 4f
peak ͑84.00 eV͒ and referenced to the Cu(2p3/2) peak.
The purity of the SO2(g) ͑Praxair, 99.98%͒, O2(g)
͑16O2 , Praxair, 99.999%͒, 18O2(g) ͑MSD Isotopes, 97.7%͒,
and D2S(g) ͑Cambridge Isotope Laboratories, 98%͒ were
monitored with the QMS during dosing; all gases were dosed
from the background. Exposures are reported in units of
Langmuir (1 Lϭ10Ϫ6 Torr s). Typical SO2(g) dosing pres-
sures were on the order 1ϫ10Ϫ8 Torr. The p(2ϫ1) oxygen
overlayer at 0.24 monolayer ͑ML͒ coverage was prepared by
dosing 1.5 L (1ϫ10Ϫ8 Torr, 2.5 min͒ of O2(g) at 450 K,
yielding p(2ϫ1) oxygen islands separated by the uncovered
Cu surface.
II. EXPERIMENT
Most experiments were performed in an ultrahigh
vacuum chamber equipped with STM, low energy electron
diffraction ͑LEED͒, Auger electron spectroscopy ͑AES͒, and
temperature-programmed reaction spectroscopy ͑TPRS͒. The
chamber was equipped with a sputter ion gun and stainless
steel gas dosers. The system exhibited a base pressure of 2
ϫ10Ϫ10 Torr following cleaning, which rose to approxi-
mately 5ϫ10Ϫ10 Torr during experiments.
The homemade ‘‘Johnnie Walker’’ type STM employed
utilizes RHK STM 100 control electronics and a Pt/Ir tip.
The tip was cleaned via induced field evaporation onto a
gold surface ͑ϳ4 ␮A, 15 min͒ prior to imaging.
III. RESULTS AND DISCUSSION
Scan dimensions were calibrated using the
Cu(110)-p(2ϫ1)-O structure.9 All images were processed
with x-offset and x-slope subtraction.
A. Surface characterization
1. XPS
The Cu͑110͒ crystal used was aligned to within 0.5° of
the ͑110͒ plane using Laue backscattering and was mechani-
cally polished down to 0.3 ␮m alumina paste. The crystal
was cleaned in vacuum by three Ar ion sputter ͑2 ␮A, 500
eV, 15 min at 600 K͒ and anneal ͑800 K, 10 min͒ cycles,
with the first anneal done in an oxygen atmosphere (1
ϫ10Ϫ7 Torr) to cleanse the surface of impurities observed in
STM images. Both sharp p(1ϫ1) LEED patterns and sub-
sequent high-resolution STM images were used to assess the
degree of surface cleanliness and order. The crystal could be
cooled to 120 K with liquid nitrogen and heated to 1100 K
by electron bombardment to the back of the crystal. The
temperature was monitored by a Chromel-Alumel thermo-
couple spot welded to a Ta foil in direct contact with the
back of the crystal. The STM ramp housing the crystal and
the STM scan head were allowed to thermally equilibrate
prior to STM measurements.
Separate XPS and TPRS measurements were made in a
second UHV system consisting of interconnected preparation
and analysis chambers. The analysis chamber exhibited a
base pressure of 3ϫ10Ϫ10 Torr and was equipped with
LEED optics, a Perkin-Elmer 04-548 dual anode x-ray
source, an EA-10-plus hemispherical energy analyzer from
SPECS, and a UTI 100c quadrupole mass spectrometer
͑QMS͒ used for TPRS measurements. The ionizer of the
QMS was enclosed in a glass cap with a small hole facing
the crystal surface. A computer coupled to the QMS was
used to record TPR-spectra. The preparation chamber
reached a base pressure of 6ϫ10Ϫ10 Torr and was equipped
with a sputter ion gun and stainless steel gas dosers. The two
chambers were isolated from each other during experiments.
In this system the crystal was supported on a Ta foil and
heated resistively via two Ta wires spot-welded to the back
XPS measurements following adsorption and reaction of
SO2(g) with the Cu͑110͒ surface are shown in Fig. 1͑a͒ and
summarized in Table I.10–17 Except for adsorption and mea-
surement at 240 K, all spectra were taken at room tempera-
ture following anneal to the specified temperature. An O(1s)
spectrum obtained for a saturated p(2ϫ1)-O overlayer at
0.5 ML coverage and a S(2p) spectrum obtained for S(a)
following decomposition of D2S are shown at the top of the
figure for reference. The position of the sulfur S(2p) XPS
signal is consistent with that observed by Roberts and co-
workers in their investigation of H2S(g) reaction on
Cu͑110͒.
Due to the absence of a feature for atomic sulfur, the
S(2p) doublet observed following adsorption at 240 K is
attributed to molecularly adsorbed SO2 . The binding ener-
gies observed are very close to those previously reported for
weak molecular adsorption of SO2 on Ag͑110͒, but are some-
what different from the measured values for Ni, for which
stronger interactions would be expected ͑Table I͒. From the
molecularly adsorbed state the ratio of the photoionization
cross-sections of O(1s) and S(2p) was determined to be
1.60, which is within 3% of the theoretical value of 1.64
determined by Scofield.18 The x-ray photoelectron spectrum
of SO2(a) does not change significantly with temperature
below 300 K.
After the surface is annealed to 324 K the x-ray photo-
electron spectrum is marked by the presence of S(2p) dou-
blets at 161.4 eV and 166.2 eV. The S(2p) peak at 161.4 eV
coincides with that for atomic sulfur, which is evident by
comparison to the reference spectrum for S(a) ͓Fig. 1͑a͔͒.
The O(1s) peak observed after heating to 324 K is shifted by
0.8 eV from that for a p(2ϫ1) oxygen overlayer ͓Fig. 1͑a͔͒.
We attribute this O(1s) peak and the S(2p) doublet at 166.2
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

4700
J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Alemozafar, Guo, and Madix
as shown by the 750 K scan. The binding energy of the
O(1s) peak is shifted by 0.2 eV, possibly due to interaction
with the sulfur at island boundaries. From the p(2ϫ1)-O
reference spectrum the coverage of atomic oxygen at 750 K
was determined to be 0.23 ML.
X-ray photoelectron spectra for the adsorption and reac-
tion of SO2(g) with Cu(110)-p(2ϫ1)-O are shown in Fig.
1͑b͒ and summarized in Table I. A p(2ϫ1)-O overlayer with
0.24Ϯ0.06 ML oxygen coverage ͑as determined by STM im-
ages͒ was prepared by delivery of 1.5 L O2(g) at 450 K. A 1
L dose of SO2(g) to this surface at 300 K yielded a spectrum
with S2p and O(1s) peak positions nearly identical to those
observed for the sulfite formed on the clean surface ͓Fig.
1͑a͔͒. This exposure was sufficient to react with all the
preadsorbed oxygen ͑see below͒, and at room temperature
any excess molecular SO2 desorbed. The oxygen to sulfur
ratio was determined to be 3.03, as expected for SO3(a).
Therefore we conclude that at 300 K on Cu(110)-p(2
ϫ1)-O sulfur dioxide reacts stoichiometrically with ad-
sorbed oxygen to form sulfite according to
SO g͒ϩO a͒→SO a͒.
͑
͑
͑
2
3
Annealing this surface beyond 530 K leads to the reversible
loss of sulfite, leaving the initial 0.24 ML of atomic oxygen
on the surface, as shown by the x-ray photoelectron spectrum
taken after the anneal to 750 K ͓Fig. 1͑b͔͒.
2. Temperature-programmed reaction spectroscopy
Following adsorption of SO2(g) on Cu͑110͒, SO2(g)
evolves at 333 K and 470 K upon heating ͓Fig. 2͑a͔͒. For the
temperature-programmed reaction spectrum shown in Fig.
2͑a͒ the surface was exposed to 0.05 L SO2(g) at 250 K and
heated to 750 K. No increase in area of either peak was
observed following an exposure of SO2(g) twice this
amount. The TPRS are interpreted as the molecular desorp-
tion of SO2(g) centered at 333 K followed by a reaction-
limited evolution centered at 470 K. This interpretation is
supported by the XPS results discussed above. Assuming a
FIG. 1. XPS spectra for ͑a͒ SO2(g) on Cu͑110͒ and ͑b͒ SO2(g) on
Cu(110)-p(2ϫ1)-O at the indicated temperatures. All S(2p) curves are
doublets consisting of S(2p3/2) and S(2p1/2) components approximately 1
eV apart, with the former at lower binding energy. A deconvolution of the
SO3(a) S(2p) peak ͑dashed box͒ is shown in the inset to ͑b͒. Binding
energies at peak maxima ͑vertical tick marks͒ corresponding to S(2p3/2) and
O(1s) peaks have been indicated. Atomic sulfur and oxygen reference spec-
tra for c(2ϫ2)-S and p(2ϫ1)-O covered surfaces are shown in ͑a͒.
first order pre-exponential factor of 1ϫ1013 Ϫ1, we esti-
s
mate the activation energy for desorption at 470 K to be 31
kcal/mol. No other species were observed to desorb.
Temperature-programmed reaction spectra following the
adsorption of SO2(g) on Cu(110)-p(2ϫ1)-O reveal the
molecular desorption peak centered at 333 K and peaks cen-
tered at 384 K, 425 K, and 470 K ͓Fig. 2͑b͔͒. Using a pre-
eV to an SOx(a) species. Using the predetermined calibra-
tion factor for the oxygen to sulfur ratio, we calculate x to be
2.98; the species is SO3(a). The S(2p) and O(1s) binding
energies observed for sulfite agree reasonably well with
those reported for SO3(a) on Ag͑110͒ ͑Table I͒.
exponential factor of 1ϫ1013 Ϫ1, we estimate the activation
s
energy for the formation of each of these reaction-limited
states to be 25 kcal/mol ͑384 K͒, 28 kcal/mol ͑425 K͒, and
31 kcal/mol ͑470 K͒. With increasing exposure, states were
seen to populate and saturate sequentially beginning with the
peak at the highest temperature.
From the S(2p) peak areas the ratio of atomic sulfur to
SO3(a) is determined to be 1:2, indicating that near 300 K,
3SO a͒→S a͒ϩ2SO a͒.
͑
͑
͑
2
3
Annealing the surface to 430 K does not result in any sig-
nificant change in binding energies, though there is a de-
crease in the area of the S(2p) feature associated with sulfite
and a broadening of the O(1s) peak, which, to be discussed
shortly, is consistent with the decomposition of SO3(a) into
O(a) and SO2(g) in this temperature range. Continued heat-
ing to 530 K results in the disappearance of all the sulfite
peaks, leaving only atomic sulfur and oxygen on the surface,
Experiments using isotopically labeled oxygen suggest
that the sulfite species has monondate coordination to the
surface ͓Fig. 2͑c͔͒. Initially 18O2 was used to prepare the
Cu(110)-p(2ϫ1)-18O surface, and SO2(g) was introduced
at 320 K to yield the sulfite-covered surface. Of the possible
isotopes for sulfur dioxide which could evolve upon heating,
only mass-to-charge ratios of 64 (SO2) and 66 (SO18O) were
detected; S18O18O could not be detected. Previous investiga-
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Adsorption and reaction of sulfur dioxide
4701
TABLE I. Summary of the XPS binding energies ͑eV͒ observed in this work ͑row 2, first entry͒. Those reported
in other ͑relevant͒ investigations have been shown for comparison. S(2p) binding energies correspond to the
S(2p3/2) signal. Unavailability of experimental data is indicated by a horizontal dash.
SO2
SO3
SO4
S
O
S(2p)
O(1s)
O(1s)
539.8
530.9
S(2p)
174.8
165.3
O(1s)
540.7
530.7
S(2p)
176.7
166.1
O(1s)
S(2p)
Gas ͑Ref. 11͒
-
-
-
-
-
-
Cu͑110͒
Cu͑110͒ ͑Ref. 10͒
161.4
161.3
529.9
Cu2S ͑Ref. 12͒
CuO ͑Ref. 12͒
CuSO4 ͑Ref. 13͒
Ag͑110͒ ͑Ref. 14͒
Ag͑111͒ ͑Ref. 15͒
Ag2S ͑Ref. 16͒
Ag foil ͑Ref. 17͒
Ni ͑Ref. 12͒
161.3
-
-
-
-
-
-
-
-
529.7
-
-
-
-
-
-
-
-
-
-
-
-
532.2
169.5
-
528.4
530.6
165.4
530.2
166.1
530.6
167.9
161.8
-
-
-
-
-
-
-
-
-
-
-
531.3
168.4
161.6
-
528.3
-
-
-
-
-
-
-
-
-
-
-
-
-
531.2
165.8
tors have suggested that sulfite coordinates to Ag͑110͒
through either one ͑monodentate͒ or two oxygen atoms
to be intraconvertible among the other S–O bonds, the latter
two species would yield peak area ratios SOO to SO18O of
one to two, respectively. With the same assumptions the
monodentate sulfite formed by the addition of SO2(g) to the
surface containing preadsorbed 18O would yield an infinite
ratio of SOO to SO18O, as it is based solely on the probabil-
ity of breaking and making inconvertible O–Cu bonds. The
observed ratio of the peak areas of approximately ten to one
is indicative of the dominance of a monodentate sulfite
which decomposes primarily by cleavage of the S–O bond.
19
͑bidentate͒ and to Cu͑100͒ or Cu͑111͒ by three oxygen
atoms.6,7 Assuming the O-metal bonds to be equivalent and
3. Reaction summary
Below room temperature SO2(g) adsorbs molecularly
͑Fig. 1͒ without any significant decomposition, i.e., SO2(g)
→SO2(a). Upon heating SO2(a) disproportionates into sul-
fur and sulfite ͓Fig. 1͑a͔͒ and excess SO2 desorbs ͓Fig. 2͑a͔͒
3SO a͒→S a͒ϩ2SO a͒.
͑1͒
͑
͑
͑
2
3
Because SO2 reacts stoichiometrically with preadsorbed oxy-
gen, we propose that a fraction of the SO2(a) population
undergoes complete decomposition through a series of el-
ementary surface reactions, and the oxygen which is liber-
ated reacts with SO2(a) to form SO3(a):
SO a͒→SO a͒ϩO a͒,
͑2a͒
͑2b͒
͑3͒
͑
͑
͑
2
SO a͒→S a͒ϩO a͒,
͑
͑
͑
SO a͒ϩO a͒→SO a͒.
͑
͑
͑
3
2
This reaction sequence gives the overall stoichiometry ob-
served. The fact that SO(a) was not observed with XPS
suggests that the rate of SO(a) decomposition ͓Eq. ͑2b͔͒ is
faster than that for the decomposition of SO2(a) ͓Eq. ͑2a͔͒.
The sulfite formed on the clean surface is stable up to
400 K, at which temperature its decomposition is marked by
a broad SO2(g) peak in TPRS centered at 470 K ͓Fig. ͑2a͔͒:
FIG. 2. TPRS spectra for ͑a͒ SO2(g) on Cu͑110͒, ͑b͒ SO2(g) on
Cu(110)-p(2ϫ1)-O, and ͑c͒ SO2(g) on Cu(110)-p(2ϫ1)-18O. The aver-
age heating rate was 0.7 K/s. The vertical axis corresponds to the QMS
signal for the species with the indicated mass to charge (m/q) ratio.
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

4702
J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Alemozafar, Guo, and Madix
smaller islands. This STM result is consistent with the
streaky c(4ϫ2) LEED pattern observed. The species com-
prising the c(4ϫ2) and p(2ϫ2) structures have identical
corrugation. Since this surface species was identified as
sulfite by XPS ͓Fig. 1͑b͔͒, we interpret the c(4ϫ2) and
p(2ϫ2) structures as SO3(a) domains.
The formation of the c(4ϫ2) structure may be driven
by repulsive lateral interactions within the sulfite overlayer.
This c(4ϫ2)-SO3 structure can be formed by shifting alter-
nate rows of the p(2ϫ2) structure by one lattice unit along
the ͓001͔ azimuth ͑Fig. 7͒. The distance between sulfite near-
est neighbors in the c(4ϫ2) structure is 6.24 Å, compared to
the nearest neighbor distance of 5.1 Å in the p(2ϫ2). The
local coverage of sulfite in both the c(4ϫ2) and p(2ϫ2)
structures is 0.25 ML. This coverage is close to the initial
0.24 ML oxygen coverage, and it would be expected to result
from the stoichiometric reaction of sulfur dioxide and pread-
sorbed oxygen. A two-dimensional model with dimensions
consistent with STM measurements is shown ͑Fig. 3͒.
For the surface prepared by dosing SO2(g) onto the par-
tially oxidized surface at 400 K, the p(2ϫ2)-SO3 structure
dominates the surface ͓Fig. 3͑c͔͒. This fact is reflected by the
streaky p(2ϫ2) LEED pattern observed. A higher resolution
scan of the surface ͑inset͒ shows p(2ϫ2) and ‘‘shifted’’
p(2ϫ2) structures, similar to the p(3ϫ2) and ‘‘shifted’’
p(3ϫ2)-CO2 structures observed on Ag͑110͒.21 The shifted
structures, which require alternate rows of sulfite to shift
FIG. 3. STM images of ͑a͒ a p(2ϫ1)-O covered surface at 0.24 ML oxy-
gen coverage, with the inset showing the surface at higher resolution; tun-
neling conditions: 97.4 mV, 0.30 nA and inset, Ϫ405 mV, 0.68 nA ͑constant
height͒. ͑b͒ SO3(a) covered surface following SO2(g) dose onto the surface
in ͑a͒ at 300 K, with c(4ϫ2) and p(2ϫ2) structures indicated in the inset;
tunneling conditions: Ϫ251 mV, 0.57 nA ͑constant height͒ and inset, Ϫ111
mV, 1.31 nA ͑constant height͒. ͑c͒ SO3(a) covered surface following
SO2(g) dose onto the surface in ͑a͒ at 400 K, with p(2ϫ2) and ‘‘shifted’’
p(2ϫ2) structures indicated in the inset; tunneling conditions: Ϫ0.56 V,
0.55 nA and inset, Ϫ453 mV, 0.53 nA ͑constant height͒. The accompanying
model, with dimensions consistent with STM measurements, depicts the
p(2ϫ2) and c(4ϫ2)-SO3 structures appearing in ͑b͒ as well as the
‘‘shifted’’ p(2ϫ2)-SO3 structure in ͑c͒.
¯
along the 110 direction by one lattice unit, may arise from
͓
͔
¯
mobility of the sulfite species along the 110 direction.
͓
͔
They appear as defects in the large domains of the p(2
ϫ2) structure.
2. SO2„g… on Cu(110)
SO a͒→SO g͒ϩO a͒.
͑4͒
͑
͑
͑
3
2
STM reveals three local structures that result from the
exposure of the clean Cu͑110͒ surface to SO2(g): c(2ϫ2),
c(4ϫ2), and p(2ϫ2) ͓Fig. 4͑a͔͒. The c(2ϫ2) structures
form large islands, while the c(4ϫ2) and p(2ϫ2) struc-
tures are dispersed randomly throughout the scan area into
much smaller domains, approximately three to four unit cells
The p(2ϫ1)-O overlayer shifts the surface chemistry
away from the decomposition pathway. At 300 K SO2(g)
reacts stoichiometrically with atomic oxygen to yield sulfite
͓Fig. 1͑b͔͒:
SO g͒ϩO a͒→SO a͒.
͑5͒
͑
͑
͑
2
3
¯
wide along the 110 . The LEED pattern observed for this
͓
͔
This sulfite species is stable up to 350 K, at which tempera-
ture the reaction limited evolution of SO2(g) from states
with peaks centered at 384 K, 425 K, and 470 K ͓Fig. 2͑b͔͒
results in the conversion of the surface back to its initially
oxidized state ͓cf. Eq. ͑4͔͒.
surface is a c(2ϫ2), in agreement with the STM images, as
islands of this structure dominate the scan area. The corru-
gation of the species comprising the c(2ϫ2) domains is
different from that of the c(4ϫ2) and p(2ϫ2) structures.
As discussed above ͑Sec. III B 1͒, the c(4ϫ2) and p(2
ϫ2) structures are attributed to sulfite.
B. Local structure
STM and LEED results for the decomposition of D2S(g)
on Cu͑110͒ suggest that the c(2ϫ2) structures observed in
Fig. 4͑a͒ are due to sulfur ͓Fig. 4͑b͔͒. D2S(g) is expected to
decompose into D2(g) and c(2ϫ2)-S,14 and the corrugation
of the c(2ϫ2)-S structure is identical to that observed for
the c(2ϫ2) structure formed in the disproportionation of
SO2(g) ͓Eq. ͑1͔͒. The c(2ϫ2)-S domains in Fig. 4͑b͒ are
separated by streaky anti-phase domain boundaries, where
the sulfur adatoms have a mobility too large to be imaged by
the STM at the scan speed used. An area calculation ͓Fig.
4͑c͒, bottom͔ for a typical scan area ͓Fig. 4͑c͒ top͔ shows
that the coverage of c(2ϫ2͒-S ͑dark patches͒ is 18.0%. Av-
1. SO2„g… on Cu„110…-p„2Ã1…-O
The initial state of this surface consisted of islands of
Cu–O chains running along the ͓001͔ direction, each row
¯
spaced by two lattice distances along the 110 direction,
͓
͔
the islands being separated by regions of uncovered surface
͓Fig. 3͑a͔͒.20 The coverage of oxygen on this surface was
approximately 0.24 ML.
SO2(g) reacts with this p(2ϫ1)-O overlayer to yield a
surface covered with c(4ϫ2) and p(2ϫ2) structures ͓Fig.
3͑b͔͒. All oxygen rows react away. The c(4ϫ2) structure
forms large domains, while the p(2ϫ2) is scattered in much
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Adsorption and reaction of sulfur dioxide
4703
FIG. 4. STM images of ͑a͒ SO2(g) on Cu͑110͒, showing large c(2ϫ2)
domains and scattered p(2ϫ2)-SO3 and c(4ϫ2)-SO3 structures; tunneling
conditions: Ϫ1.52 V, 0.52 nA. ͑b͒ D2S(g)/Cu(110), showing c(2ϫ2)-S
structures separated by streaky anti-phase domain boundaries; tunneling
conditions: Ϫ0.50 V, 1.23 nA ͑constant height͒. From this image it is deter-
mined that the c(2ϫ2) structures following SO2(g) adsorption ͑a͒ are due
to sulfur. ͑c͒ The top image is an STM scan of the Cu͑110͒ surface following
disproportionation of SO2(g) into sulfite and sulfur structures at 300 K,
where the sulfite domains surround sulfur islands of lower corrugation; tun-
neling conditions: Ϫ343 mV, 0.49 nA ͑constant height͒. The bottom image
corresponds to the STM scan above, where it is calculated that the sulfite
structures ͑white area͒ cover 82% of the surface and the sulfur domains
͑black patches͒ the remaining 18%. An average coverage weighted sulfite to
sulfur area ratio of 2:1 is obtained from similar images, which is consistent
with the 2:1 sulfite to sulfur ratio determined by XPS. ͑d͒ SO2(g) on
Cu͑110͒ at 400 K, showing large p(2ϫ2)-SO3 islands surrounding scat-
tered c(4ϫ2)-SO3 and c(2ϫ2)-S domains; tunneling conditions: 0.56 V,
0.66 nA.
FIG. 5. A selection of STM images obtained at 2–3 s/frame during SO2(g)
dosing onto a Cu(110)-p(2ϫ1)-O surface at 0.24 ML oxygen coverage;
tunneling conditions: Ϫ135 mV, 0.54 nA ͑constant height͒. The time of each
frame ͑lower right͒ is in reference to that at zero seconds ͑s͒ ͑frame a͒. Rows
orthogonal to the oxygen structures appear and grow in number with time
͑frames a–o͒. They display two-fold spacing in the ͓001͔ ͑frame l, arrows 7
¯
and 8͒ and 110 directions ͑frame i, inset, arrow 4͒ and are interpreted as
͓
͔
sulfite structures, identical to those comprising the p(2ϫ2)-SO3 structures
͓Figs. 3͑c͒ and 4͑d͔͒. The inset to frame i is an enlargement of the boxed
area shown at the upper left hand corner of the frame. The arrows corre-
spond to features described in the text. See http://www.stanford.edu/group/
madix for a video clip.
mains of lower corrugation, which we interpret as c(2
ϫ2)-S structures. There is apparently a significant rear-
rangement of the surface structure from that at 300 K ͓Fig.
4͑a͔͒, since the large c(2ϫ2)-S domains have dissipated. A
c(4ϫ2)-SO3 structure has been indicated in the figure and is
seen to sit in-between p(2ϫ2) domains.
eraging over 15 scan regions approximately the same size as
that in Fig. 4͑c͒ ͑top͒ gives 20.53%Ϯ1.53%, giving a area
weighted coverage of S of 0.1 ML. Similarly, the sulfite cov-
erage can be determined to be 0.2 ML. Thus, the sulfite to
sulfur ratio determined by STM is 2:1, in accordance with
the XPS findings ͓Eq. ͑1͔͒.
Introducing SO2(g) at 400 K, a temperature at which
some decomposition of sulfite can occur, produces a surface
with a dominant p(2ϫ2) structure with a few scattered
c(2ϫ2) and c(4ϫ2) defects ͓Fig. 4͑d͔͒. The prevalence of
the p(2ϫ2) structure is similar to what was observed for
sulfite on the oxygen covered surface at 400 K ͓Fig. 3͑c͔͒.
The p(2ϫ2) LEED pattern for this surface is consistent with
this being the dominant structure on the surface. The p(2
ϫ2)-SO3 areas appear to be separated by small ͑dim͒ do-
3. Real-time imaging: SO2„g… on 0.24 ML p„2Ã1…-O
Real-time STM images show the mobility of sulfite and
oxygen rows during reaction of SO2(g) with preadsorbed
oxygen at 300 K ͑Figs. 5 and 6͒. The Cu͑110͒ surface was
initially prepared with a p(2ϫ1)-O overlayer at 0.24 ML
coverage ͓Fig. 3͑a͔͒, and sulfur dioxide was dosed at a pres-
sure of 1ϫ10Ϫ9 Torr. A specific area with well-defined fea-
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

4704
J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Alemozafar, Guo, and Madix
As reaction proceeds, increasing numbers of short,
¯
bright rows, oriented along the 110 direction appear, span-
͓
͔
ning the space between the p(2ϫ1)-O islands. These fea-
tures have the same corrugation as the features in the p(2
ϫ2)-SO3 structures discussed above ͓Figs. 3͑c͒ and 4͑d͔͒. In
Fig. 5͑i͒ ͑arrow 4, inset͒ these chains can be seen to consist
¯
of units with two-fold spacing along the 110 direction.
͓
͔
Furthermore, at closest approach the rows exhibit a two-fold
spacing along the ͓001͔ direction ͑see below͒. Thus they ap-
pear to form a local p(2ϫ2) configuration like those of
sulfite discussed above. Hence we interpret these rows as
sulfite structures. Lateral interactions prevent these rows
from approaching one another more closely than 7.2 Å,
which corresponds to two unit cell vectors along the ͓001͔
direction of the underlying copper surface.
The mobility of sulfite is appreciable along the ͓001͔
azimuth. When viewed continuously, the motion of the
sulfite rows up and down the channels separating the oxygen
islands appears random, suggesting little interference by the
STM. However, they do migrate together to form transient,
rudimentary p(2ϫ2) structures ͓Figs. 5͑k͒ and 5͑i͒, arrows
¯
7 and 8͔. Attractive interactions ͑along the 110 direction͒
͓
͔
keep these rows intact, but random migration leads to sepa-
ration of the rows forming the p(2ϫ2) structure.
Fragmentation of an SO3(a) row accompanied by a shift
of one lattice unit in the ͓001͔ direction would lead to a
rudimentary c(4ϫ2) structure, and there is indication of
such behavior in Figs. 5͑f͒ and 5͑g͒ ͑arrow 4͒ and Fig. 5͑h͒
͑arrow 10͒. The two portions of the fragmented rows appear
to interact with the adjoining oxygen islands, resulting in a
‘‘snakelike’’ migration of the sulfite structure, through this
intermediate structure along the ͓001͔ as shown by arrow 5
in Fig. 6͑a͒. Their translation is seen to take place in steps in
that one portion of the row moves first and is subsequently
followed by the remainder, similar to the mechanism by
which rows of oxygen translate from one p(2ϫ1)-O island
to another.
There are also numerous examples of the fracture of
sulfite rows caused by the collision of a migrating sulfite row
with a narrowing of the space between the oxygen islands.
This is shown in Figs. 5͑d͒–5͑h͒ ͑arrow 4͒ and further dem-
onstrated in Fig. 6͑a͒ ͑arrow 5͒, in which the sulfite row
͑arrow 5͒ is split into two segments following collision with
an oxygen row ͑arrow 1͒.
FIG. 6. A selection of STM images obtained at 2–3 s/frame during SO2(g)
dosing onto a Cu(110)-p(2ϫ1)-O surface at 0.24 ML oxygen coverage;
tunneling conditions: Ϫ135 mV, 0.54 nA ͑constant height͒. The time of each
frame ͑lower right͒ is in reference to that at zero seconds ͑s͒ in Fig. 5͑a͒. ͑a͒
Sequence of images accompanying Fig. 5͑h͒, demonstrating the snakelike
migration of a sulfite row ͑arrow 5͒ and subsequent collision with an oxygen
row ͑arrow 1͒. After collision at 560 s, a portion of the sulfite row is left
behind. ͑b͒ Sequence of STM images shortly following Fig. 5͑l͒, showing
the snakelike migration of an oxygen row ͑arrow 9͒ from one p(2ϫ1)-O
island to another, apparently involving reptilation ͑see text͒.
tures was tracked at a scan rate between 2–3 s/frame. Cu–O
rows in the p(2ϫ1) islands, separated by regions of clean
surface are clearly visible in Fig. 5͑a͒.
This time sequence reveals several interesting features of
the reaction. Rows of oxygen at the edges of the p(2
ϫ1)-O islands are mobile, consistent with the heightened
reactivity of oxygen at island edges.22 This behavior is dem-
onstrated by row 1 in Fig. 5͑b͒ and 5͑o͒ and row 6 in Figs.
5͑h͒–5͑j͒, which show the oxygen rows moving from one
p(2ϫ1)-O island to another. Further inspection reveals that
the rows of oxygen translate from one island to another in a
‘‘snakelike’’ fashion. This behavior can be seen in Figs. 5͑b͒
and 5͑c͒ ͑arrows 2 and 3͒ and Fig. 6͑b͒ ͑arrow 9͒. At first a
small segment of the chain moves laterally, and the remain-
der follows shortly thereafter, apparently involving reptila-
tion. The migration of p(2ϫ1)-O segments has been ob-
served previously by Besenbacher and co-workers, who
4. Binding sites and two-dimensional model
A partial coverage of sulfite on the p(2ϫ1)-O surface
was created in order to determine the binding site of
SO3(a).23 The surface was initially prepared with a p(2
ϫ1)-O overlayer at 0.24 ML coverage. SO2(g) and O2(g)
were subsequently co-dosed to produce the surface shown in
Figs. 7͑a͒ and 7͑b͒. Figures 7͑a͒ and 7͑b͒ show atomically
resolved Cu–O rows in their p(2ϫ1) arrangement and
sulfite moieties between the islands appear as bright circular
features. The sulfites are arranged in both c(4ϫ2) and p(2
ϫ2) structures, though defects ͑d͒ are evident. These defects
may result from the inability of the sulfite rows to shift to the
c(4ϫ2) structure due to lateral interactions with adjacent
calculated
an
effective
diffusion
coefficient
D
у10Ϫ13 cm2/s.20
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Adsorption and reaction of sulfur dioxide
4705
brighter p(2ϫ1) features in the oxide island, as expected if
copper atoms are imaged in the oxide islands. We thus con-
clude that we have imaged the copper atoms in the oxygen
islands. This result agrees well with previous work.24
With the location of the copper in the p(2ϫ1) image of
the oxide island known, the binding site of sulfite was
straightforwardly determined to be the four-fold hollow us-
ing the line scan in Fig. 7͑b͒; it is consistent with the registry
of the sulfites ͑arrows 1 and 2͒ relative to adjacent p(2
ϫ1) rows ͓Fig. 7͑b͔͒. The edge of the sulfite domain begins
with the four-fold hollow site next nearest to the oxygen
¯
island edge ͑along the 110 direction͒, not the adjacent
͓
͔
four-fold hollow site. A four-fold hollow binding site con-
figuration does not preclude the possibility, however, of the
incorporation of an ‘‘added’’ Cu atom into the resulting
sulfite structure following reaction between SO2(g) and
O(a) ͓Eq. ͑5͔͒. From the sulfite binding site it follows that
the sulfur atoms observed on the clean surface ͓Fig. 4͑a͔͒
formed by the disproportionation of SO2(a) ͓Eq. ͑1͔͒ also
occupy four-fold hollows. This is in agreement with the
binding site as suggested by King and co-workers.25
5. Nucleation and growth of SO3„a… structures
The pattern of growth of sulfide and sulfite following the
exposure of Cu͑110͒ to SO2(g) suggests certain qualitative
features of the growth process. The basic observation is that
sulfite forms c(4ϫ2) and p(2ϫ2) patches around c(2
ϫ2)-S islands. Following decomposition of SO2(a) into
sulfur and oxygen ͓Eq. ͑2͔͒, SO2(g) reacts rapidly with the
oxygen adatoms to form sulfite. During this process sulfur
and sulfite segregate into patches of c(2ϫ2), p(2ϫ2), and
c(4ϫ2) structures ͓Figs. 4͑a͒ and 4͑c͔͒. The c(2ϫ2)-S do-
mains apparently prevent sulfite from relaxing into the c(4
ϫ2) configuration.
FIG. 7. STM images of ͑a͒ and ͑b͒ SO2(g)ϩO2(g) co-dosed onto a
Cu(110)-p(2ϫ1)-O surface at 0.24 ML coverage; tunneling conditions: ͑a͒
Ϫ415 mV, 0.25 nA ͑constant height͒ and ͑b͒ Ϫ416 mV, 0.60 nA ͑constant
height͒. ͑a͒ and ͑b͒ are of the same general area, the solid lines connect
corresponding p(2ϫ1)-O and sulfite rows. The solid line in ͑b͒ accompa-
nies the line scan shown in the inset; arrows 1 and 2 correspond to the sulfite
features indicated in the line scan. ͑c͒ A surface with p(2ϫ1)-O rows amid
the Cu(110)-p(1ϫ1) lattice; tunneling conditions: Ϫ98.4 mV, 0.60 nA
͑constant height͒. The dashed lines, corresponding to rows of Cu atoms
¯
oriented along the 110 , are seen to sit in-between bright features com-
͓
͔
On the Cu(110)-p(2ϫ1)-O surface the nucleation and
growth of the c(4ϫ2) and p(2ϫ2) sulfite structures at
p(2ϫ1)-O island boundaries is apparently accompanied by
a p(2ϫ2)→c(4ϫ2) structural relaxation. A minor amount
of the p(2ϫ2) structure can result from a one lattice unit
mismatch of nucleation sites of the c(4ϫ2) structure along
the ͓001͔ direction at two adjacent oxygen structures, as in-
dicated by the boxed area in Fig. 7͑a͒. Shown, for example,
is a full c(4ϫ2) unit cell with an adjacent half unit cell at
opposite sides of a region separating two oxygen islands; a
domain boundary between these two structures forms a chain
of p(2ϫ2) cells running along the ͓001͔ direction. Lateral
sulfite–sulfite and sulfite–oxygen forces appear to pin the
p(2ϫ2) structure and prevent its relaxation into the c(4
ϫ2). This effect leads to a mixture of sulfite moieties amid
the p(2ϫ1)-O structures ͓Figs. 7͑a͒ and 7͑b͔͒.
prising the p(2ϫ1) structure, suggesting that the species imaged by the
STM in the p(2ϫ1)-O overlayer are the ‘‘added’’ Cu atoms. ͑d͒ A struc-
tural model consistent with line-scan measurements in ͑a͒ and ͑b͒, where it
is proposed that the sulfites comprising the p(2ϫ2) and c(4ϫ2) structures
occupy four-fold hollow sites.
sulfite and oxygen structures. The line scan in Fig. 7͑b͒ ͑in-
set͒ shows the corrugation of the sulfites ͑arrows 1 and 2͒
relative to the species comprising the p(2ϫ1) structure.
In order to determine the binding site of sulfite the entity
imaged in the Cu–O islands must be first determined so that
it can be used as a reference point for the sulfite. The STM
scan shown in Fig. 7͑c͒ suggests that the species imaged in
the p(2ϫ1) structures are the ‘‘added’’ Cu atoms. Indicated
in the figure are p(1ϫ1) and p(2ϫ1) unit cells. The p(1
ϫ1) is due to the Cu͑110͒ rectangular lattice and we assume
that it is due to the imaging of Cu atoms.
Results of previous studies suggest that the p(2ϫ1)-O
structure results from the sequential addition of oxygen at-
oms to long bridge sites and Cu atoms to four-fold hollow
sites along the ͓001͔ azimuth.9,20 According to Fig. 7͑c͒,
lines extended through the close-packed rows of Cu atoms
IV. SUMMARY
͑1͒ On the Cu͑110͒ surface SO2(g) adsorbs molecularly be-
low 300 K. Heating the surface to 300 K results in the
disproportionation of a fraction of the SO2(a) popula-
tion into SO3(a) and S(a):
¯
͑dashed lines͒ oriented along the 110 direction straddle the
3SO a͒→S a͒ϩ2SO a͒
͓
͔
͑
͑
͑
2
3
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42

4706
J. Chem. Phys., Vol. 116, No. 11, 15 March 2002
Alemozafar, Guo, and Madix
1 EPA, http://www.epa.gov ͑2000͒.
The surface is covered by large c(2ϫ2)-S and scattered
c(4ϫ2)-SO3 and p(2ϫ2)-SO3 structures.
2 C. D. Cooper and F. C. Alley, Air Pollution Control: A Design Approach,
2nd ed. ͑Waveland, Prospect Heights, IL, 1994͒.
͑2͒ Disproportionation at 300 K is concomitant with the de-
sorption of excess SO2(a). Continued heating to 400 K
yields a surface covered with p(2ϫ2)-SO3 and scat-
tered c(2ϫ2)-S and c(4ϫ2)-SO3 structures. Further
heating leads to the decomposition of sulfite at 470 K
according to SO3(a)→SO2(g)ϩO(a), which is com-
plete by 530 K, yielding a surface covered with atomic
sulfur and oxygen.
͑3͒ On the Cu(110)-p(2ϫ1)-O surface SO2(g) reacts ac-
cording to SO2(g)ϩO(a)→SO3(a)at 300 K to yield a
surface covered with c(4ϫ2)-SO3 and scattered p(2
ϫ2)-SO3 structures.
͑4͒ Real-time image sequences show the mobility of oxygen
at p(2ϫ1)-O island boundaries and the mobility of
sulfite amid the oxygen structures.
͑5͒ Sulfite and sulfur occupy four-fold hollow binding sites.
TPRS measurements with isotopically labeled oxygen
indicate that sulfite coordinates as a monodentate spe-
cies.
͑6͒ On the oxygen covered surface, SO2(g) is oxidized at
p(2ϫ1)-O island boundaries into p(2ϫ2)-SO3 and
c(4ϫ2)-SO3 structures. A one lattice unit mismatch
͑along the ͓001͔͒ of nucleation sites at opposing p(2
ϫ1)-O islands leads to the p(2ϫ2) structure at c(4
ϫ2) domain boundaries. Once all of the oxygen has
been reacted away, relaxation of the sulfites yields a
c(4ϫ2)-SO3 covered surface.
3 J. Haase, J. Phys.: Condens. Matter 9, 3647 ͑1997͒.
4 H. W. Wassmuth, J. Ahner, M. Hofer, and H. Stolz, Prog. Surf. Sci. 42,
257 ͑1993͒.
5 M. Polcik, L. Wilde, J. Haase, B. Brena, D. Cocco, G. Comelli, and G.
Paolucci, Phys. Rev. B 53, 13720 ͑1996͒.
6 T. Nakahashi, S. Terada, T. Yokoyama, H. Hamamatsu, Y. Kitajima, M.
Sakano, F. Matsui, and T. Ohta, Surf. Sci. 373, 1 ͑1997͒.
7 G. J. Jackson, S. M. Driver, D. P. Woodruff, N. Abrams, R. G. Jones, M.
T. Butterfield, M. D. Crapper, B. C. C. Cowie, and V. Formoso, Surf. Sci.
459, 231 ͑2000͒.
8 C. M. Pradier and P. Dubot, J. Phys. Chem. B 102, 5135 ͑1998͒.
9 D. J. Coulman, J. Wintterlin, R. J. Behm, and G. Ertl, Phys. Rev. Lett. 64,
1761 ͑1990͒.
10 A. F. Carley, P. R. Davies, R. V. Jones, K. R. Harikumar, G. U. Kulkarni,
and M. W. Roberts, Surf. Sci. 447, 39 ͑2000͒.
11 A. Bakke, H. Chen, and W. Jolly, J. Electron Spectrosc. Relat. Phenom.
20, 333 ͑1980͒.
12 C. R. Brundle and A. F. Carley, Faraday Discuss. Chem. Soc. 60, 51
͑1975͒.
13 B. R. Strohmeier, D. E. Leyden, R. S. Field, and D. M. Hercules, J. Catal.
94, 514 ͑1985͒.
14 D. A. Outka and R. J. Madix, Surf. Sci. 137, 242 ͑1984͒.
15 G. Tibbetts, J. Vac. Sci. Technol. 15, 497 ͑1978͒.
16 D. Lichtman, J. H. Craig, V. Slater, and M. Drinkwine, Appl. Surf. Sci. 7,
325 ͑1981͒.
17 R. W. Joyner and M. W. Roberts, Chem. Phys. Lett. 60, 459 ͑1979͒.
18 J. H. Scofield, J. Electron Spectrosc. Relat. Phenom. 20, 333 ͑1976͒.
19 D. A. Outka, R. J. Madix, G. B. Fisher, and C. DiMaggio, J. Phys. Chem.
B 90, 4051 ͑1986͒.
20 F. Jensen, F. Besenbacher, E. Laegsgaard, and I. Stensgaard, Phys. Rev. B
41, 10233 ͑1990͒.
21 X.-C. Guo and R. J. Madix, J. Phys. Chem. B 105, 3878 ͑2001͒.
22 X.-C. Guo and R. J. Madix, Surf. Sci. 367, L95 ͑1996͒.
23 F. M. Leibsle, Surf. Sci. 311, 45 ͑1994͒.
24 F. Besenbacher and I. Stensgaard, in The Chemical Physics of Solid Sur-
faces, Vol. 6, edited by D. A. King and D. P. Woodruff ͑Elsevier Science,
New York, 1993͒, Chap. 15.
ACKNOWLEDGMENT
We gratefully acknowledge the National Science Foun-
dation ͑NSF CHE 9820703͒ for support of this work.
25 A. Atrei, A. L. Johnson, and D. A. King, Surf. Sci. 254, 65 ͑1991͒.
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:
152.23.0.100 On: Mon, 24 Nov 2014 21:04:42