Molecular hydrogen formation from proximal glycol pairs on TiO 2(110)

Article abstract of DOI:10.1021/ja500992b

Understanding hydrogen formation on TiO₂ surfaces is of great importance, as it could provide fundamental insight into water splitting for hydrogen production using solar energy. In this work, hydrogen formation from glycols having different numbers of methyl end-groups has been studied using temperature-programmed desorption on reduced, hydroxylated, and oxidized rutile TiO₂(110) surfaces. The results from OD-labeled glycols demonstrate that gas-phase molecular hydrogen originates exclusively from glycol hydroxyl groups. The yield is controlled by a combination of glycol coverage, steric hindrance, TiO₂(110) order, and the amount of subsurface charge. Combined, these results show that proximal pairs of hydroxyl-aligned glycol molecules and subsurface charge are required to maximize the yield of this redox reaction. These findings highlight the importance of geometric and electronic effects in hydrogen formation from adsorbates on TiO₂(110).

Full text of DOI:10.1021/ja500992b

Communication
pubs.acs.org/JACS
Molecular Hydrogen Formation from Proximal Glycol Pairs on
TiO2(110)
Long Chen, Zhenjun Li,† R. Scott Smith, Bruce D. Kay, and Zdenek Dohnalek*
́
Fundamental and Computational Sciences Directorate and Institute for Integrated Catalysis, Pacific Northwest National Laboratory,
P.O. Box 999, Richland, Washington 99352, United States
S
* Supporting Information
another HOb group.9−13 Upon heating to ∼500 K, the resulting
ABSTRACT: Understanding hydrogen formation on
TiO2 surfaces is of great importance, as it could provide
fundamental insight into water splitting for hydrogen
production using solar energy. In this work, hydrogen
formation from glycols having different numbers of methyl
end-groups has been studied using temperature-pro-
grammed desorption on reduced, hydroxylated, and
oxidized rutile TiO2(110) surfaces. The results from
OD-labeled glycols demonstrate that gas-phase molecular
hydrogen originates exclusively from glycol hydroxyl
groups. The yield is controlled by a combination of glycol
coverage, steric hindrance, TiO2(110) order, and the
amount of subsurface charge. Combined, these results
show that proximal pairs of hydroxyl-aligned glycol
molecules and subsurface charge are required to maximize
the yield of this redox reaction. These findings highlight
the importance of geometric and electronic effects in
hydrogen formation from adsorbates on TiO2(110).
HOb’s recombine with another HOb or an alkoxy to produce a
water or an alcohol molecule, respectively.14−18 For alcohols,
the majority of alkoxy species (except for methanol which
recombinatively desorbs) undergo dehydration to alkenes and
water.15−18 As a consequence, molecular hydrogen desorption
from alcohols and water on TiO2(110) has not been generally
observed.15−21 In line with these observations, theoretical
calculations demonstrate that the kinetic barrier for H2
recombinative desorption from HOb sites on TiO2(110) is
considerably higher than the barrier for water formation.20,21
However, in a recent temperature-programmed desorption
(TPD) study,22 Xu et al. reported that D2 is formed as a minor
product via thermal recombination of the D atoms on Ob sites
from photocatalysis of methanol. Interestingly, hydrogen
desorption between 375 and 500 K was also observed in our
recent study of ethylene glycol (EG) reactions on TiO2(110).23
Molecular hydrogen evolution from EG on TiO2(110) was
first observed in a study of ethanol, n- and 2-propanol, EG, and
diethyl ether via TPD and X-ray photoelectron spectroscopy
(XPS).24 Subsequent detailed quantitative investigations with
different isotopes (i.e., HO(CH2)2OH, DO(CH2)2OD, and
HO(CD2)2OH) demonstrated that the observed hydrogen
originates solely from hydroxyl groups and that high EG
coverages are required for its formation.23 While at low
coverages, ethylene and water were identified as the only
products, at saturation coverage both ethylene + water and
acetaldehyde + hydrogen were observed. Total product yield of
carbon-containing products at saturation coverage was
determined to be ∼0.5 monolayer (ML), which clearly shows
that the reactions proceed not only at VO defect sites but also
on the Ti5c rows.
he technological importance of TiO2 has driven cutting-
Tedge research with the aim of understanding the
elementary steps that underlie catalytic and photocatalytic
reactions on this material.1−5 In this context hydrogen, as a
promising source of clean, renewable, and environmentally
friendly energy, is of great interest.2,6−8 As such, understanding
the factors that control its formation in the thermal and
photocatalytic conversion of water and organics is of critical
importance. While a significant effort has been devoted to this
subject, the mechanism of hydrogen formation on TiO2 is far
from understood.
In this study, we systematically investigate hydrogen
formation from a series of glycols with different steric
constraints on differently prepared TiO2(110) surfaces in
order to further understand hydrogen formation from glycols,
and polyols in general. We find that hydrogen formation is
dramatically attenuated for glycols with methyl end-groups
shielding the hydroxyl functional groups. Further, hydrogen
formation is not observed on disordered or oxidized
TiO2(110). These results suggest that on TiO2(110), pairs of
aligned glycol molecules with hydroxyl groups in close
proximity, together with subsurface charge are required to
carry out this redox reaction.
For catalytic and photocatalytic studies in particular, the most
stable TiO2 surface, rutile TiO2(110), has served as a
prototypical model.4,5 When stoichiometric or slightly reduced,
the surface is unreconstructed and composed of alternating 5-
fold-coordinated Ti4+ (Ti5c) and bridging oxygen (Ob) rows. A
small fraction of bridging oxygen vacancy defects (VO’s) that
form as a consequence of TiO2 bulk reduction during ultrahigh
vacuum (UHV) preparation, play an important role in
determining the surface reactivity of TiO2(110).1,4
Water and alcohols have been thoroughly investigated on
TiO2(110) due to their importance as models in studies of
photocatalytic water splitting and photo-oxidation of organic
contaminants.1−5 Both water and alcohol molecules initially
dissociate at a VO site filling it with a hydroxy (HOb) and/or
alkoxy group, respectively, and splitting hydrogen that forms
Received: January 28, 2014
Published: April 4, 2014
© 2014 American Chemical Society
5559
dx.doi.org/10.1021/ja500992b | J. Am. Chem. Soc. 2014, 136, 5559−5562

Journal of the American Chemical Society
Communication
The experiments were carried out in a UHV molecular beam
surface scattering apparatus (base pressure < 7 × 10−11 Torr)
described previously.16,25 A rutile TiO2(110) crystal (10 × 10 ×
1 mm3, Princeton Scientific) was cleaned by cycles of Ne+
sputtering and annealing to 850−900 K in UHV until
impurities were undetectable in Auger electron spectra. The
surface prepared in this way (referred to as r-TiO2(110))
contained ∼5% VO defects as determined by H2O TPD.14,26
Additionally, hydroxylated (h-TiO2(110)) and oxidized (o-
TiO2(110)) surfaces were prepared by exposing r-TiO2(110) to
D2O at 400 K, and to O2 at 70 K, followed by annealing to 300
K, respectively.27,28 All glycols (ethylene glycol, 1,2- and 1,3-
propylene glycols, 2,3-butylene glycol, 2,3-dimethyl-2,3-buty-
lene glycol) used in this study were obtained from Sigma-
Aldrich and transferred into flasks with baked molecular sieves.
The glycols were purified by pumping at room temperature,
and their temperature during the dose was stabilized in a water
bath. The glycols were dosed onto the various TiO2(110)
surfaces using an effusive molecular beam. TPD measurements
(ramp rate of 1 K/s) were performed using a quadrupole mass
spectrometer (UTI).
To unambiguously determine the origin of hydrogen (O−H
vs C−H), we employ OD-labeled glycols and monitor the D2
signal during the TPD experiments. Since not all OD-labeled
glycols are commercially available we prepared them directly on
TiO2(110) using an H/D exchange procedure as described in
the Supporting Information (SI). The viability of this procedure
has been confirmed by comparing the TPD spectra from the
isotopically exchanged 1,2-propylene glycol (1,2-PG) with
those of commercially available OD-labeled 1,2-PG (Figure S1,
SI). The use of OD-labeled glycols further allows us to
overcome the experimental difficulties associated with high H2
background during TPD due to H2 desorption from a closed-
cycle He-cryostat cooled manipulator.23
Figure 1a shows the coverage-dependent D2 TPD spectra
from OD-labeled ethylene glycol (DO(CH2)2OD), 1,3-
propylene glycol (DO(CH2)3OD), 1,2-propylene glycol
(DOCH2CH(CH3)OD), 2,3-butylene glycol (DOCH(CH3)-
CH(CH3)OD), and 2,3-dimethyl-2,3-butylene glycol (DOC-
(CH3)2C(CH3)2OD) on r-TiO2(110). Other products include
water and a broad range of carbon-containing species. While
the discussion of the carbon-containing products for all
employed glycols is beyond the scope of this report, generally
alkenes are observed at low coverages, and both alkenes and
aldehydes/ketones are seen at high coverages. For EG, all the
products and their coverage dependences were discussed in
detail in our prior publication23 and are also summarized in
Figure S2 in SI. Further, since D2 desorption occurs at lower
temperatures than desorption of any carbon-containing
products (see data for EG23 in Figure S2, SI), we believe that
D2 formation is largely decoupled from the formation of
carbon-containing products at higher temperatures.
The selected sequence of glycol molecules allows us to
explore how the methyl end-groups that surround the hydroxyl
functional groups affect D2 formation. Figure 1a shows that for
each glycol molecule. With the exception of 2,3-dimethyl-2,3-
butylene glycol, the amount of desorbing D2 is strongly
dependent on the coverage. While at low coverages (<0.25
ML), D2 is practically absent, as the coverage increases to 0.5−1
ML, D2 is observed for all glycols except 2,3-dimethyl-2,3-
butylene glycol. As further discussed below, the absence of D2
for 2,3-dimethyl-2,3-butylene glycol throughout the whole
coverage range is a result of the steric hindrance introduced by
Figure 1. (a) Coverage-dependent TPD spectra of D2 (m/z = 4
(D2 )) following ethylene glycol (black), 1,3-propylene glycol (red),
+
1,2-propylene glycol (green), 2,3-butylene glycol (blue), and 2,3-
dimethyl-2,3-butylene glycol (olive) doses listed in the figure. For
comparison, all OD-labeled glycols used in these experiments were
prepared directly on TiO2(110) using an H/D exchange procedure
described in SI. Saturation coverages (defined as 1 monolayer (ML))
correspond to the glycol exposures that saturate the high temperature
TPD peak from the parent glycol molecules (see Figure S3, SI). (b)
Coverage-dependent D2 desorption yields determined from (a). To
obtain normalized glycol coverages (glycol molecules per Ti5c site), we
normalized each coverage in (a) using the corresponding C 1s XPS
peak areas of the C−OH moieties (see Figures S4 and S5 in SI for
details) and employed the known saturation coverage of methanol
(0.77 methanol/Ti5c) as a reference.29 (c) Schematic structures of
studied glycols in the order of increasing D2 yield.
four −CH3 end-groups that surround the hydroxyl functional
groups. The D2 desorption peaks for the remaining four glycols
extend from 375 to 500 K with the peak area increasing and
peak maximum shifting slightly to lower temperature with
increasing initial glycol coverage. The fact that the desorption
temperatures are the same for all the glycols indicates that the
rate limiting step is identical. As the coverage of glycols on Ti5c
rows reaches saturation, D2 desorption saturates as well
(spectra above saturation coverages not shown).
The integrated D2 yields from Figure 1a are displayed as a
function of normalized glycol coverage in Figure 1b. To
compare the relative D2 yield for the same number of adsorbed
glycol molecules, we normalized each glycol coverage from (a)
using the corresponding C 1s XPS peak area for the C−OH
moieties (see Figures S4 and S5 in SI for details) and the
known saturation coverage of methanol (0.77 methanol/
Ti5c).29 As expected, the normalized saturation coverages
decrease with increasing number of methyl end-groups in the
glycols, and the normalized saturation EG coverage of 0.57 EG/
Ti5c is in reasonable agreement with the saturation coverage of
0.43 EG/Ti5c determined previously.23 Further, our previous
EG study has shown that at saturation coverage, one D2 is
obtained on average per every two adsorbed EG molecules.23
This D2 yield is used to calibrate the absolute D2 yields for
other glycols in Figure 1b.
The integrated yields for all the glycols exhibit similar,
superlinear dependences on coverage. In our previous study of
EG we noted that neighboring Ti5c-bound EG molecules may
be required for the D2 formation.23 For spatially random
adsorption on the Ti5c rows, the coverage of EG pairs and
hence D2 yield should scale with the coverage squared. To see
5560
dx.doi.org/10.1021/ja500992b | J. Am. Chem. Soc. 2014, 136, 5559−5562

Journal of the American Chemical Society
Communication
whether this is indeed the case we fit the data for EG to a
parabolic functional form. Generally a good fit is obtained as
illustrated by the solid black line shown in Figure 1b.
The quadratic yield dependence suggests that D2 formation
requires the proximal location of OD from aligned glycols. With
the above picture in mind, the D2 yields from different glycols
can be compared (Figure 1c). The simplest comparison can be
made for ethylene glycol and 1,3-propylene glycol. As both OD
groups are primary in both molecules, a similar yield can be
expected. Indeed, the D2 desorption yields from ethylene glycol
and 1,3-propylene glycols are identical within the uncertainty of
the measurement. Further, the trends in D2 yield can be
evaluated as a function of increasing steric hindrance
introduced by the −CH3 end-groups by following the sequence
of ethylene glycol analogues, i.e., 1,2-propylene glycol (−CH3
at one end), 2,3-butylene glycol (−CH3 on both ends), and 2,3-
dimethyl 2,3-butylene glycol (two −CH3 on both ends). This is
schematically illustrated in Figure 2 for the case of 1,2-PG
Figure 3. Effects of Ne+ sputtering (∼2 ML of Ne+ at 1.5 kV) of clean
TiO2(110) on D2 formation from ethylene (a) and 1,2-propylene (b)
glycols. Saturation coverages of OD-labeled EG and 1,2-PG were
dosed at 80 K. (i) Ordered TiO2(110) before sputtering; (ii) after
sputtering; (iii) sputtered and annealed to 870 K.
Figure 2. Three possible configurations of 1,2-propylene glycol
molecular pairs on TiO2(110). From A to C, the steric hindrance
increases, while D2 formation probability decreases.
Figure 4. TPD spectra of D2 from ethylene (a) and 1,2-propylene (b)
glycols on r-, h-, and o-TiO2(110). Saturation coverages of OD-labeled
EG and 1,2-PG were dosed at 80 K.
where three possible paired configurations can be envisioned.
As suggested by the experimental data (Figure 1), −CH3 end-
groups decrease the probability of an efficient interaction
between the hydroxyl groups of neighboring glycol molecules
and reduce the D2 yield. Accordingly, the saturation D2 yield
decreases from 0.20 ML for ethylene glycol to 0.11, 0.09, and
finally to 0 for 1,2-propylene glycol, 2,3-butylene glycol, and
2,3-dimethyl 2,3-butylene glycol, respectively. The observed
glycol dependences indicate that D2 is formed directly from two
neighboring OD species rather than being a result of
recombination of deuterium diffusing on Ti5c rows.
To further test the hypothesis that D2 is being formed from
pairs of neighboring glycol molecules bound on Ti5c rows, we
examined D2 formation on TiO2(110) disordered by gentle
sputtering (∼2 ML of Ne+). In accord with the steric hindrance
arguments presented above, disordered TiO2(110) is expected
to prevent the formation of proximal pairs of hydroxyl groups
from aligned glycol molecules. Figure 3 shows the effect of Ne+-
sputtering on D2 formation on the surfaces saturated with EG
and 1,2-PG. Consistent with the expectations, D2 is almost
completely suppressed for EG and cannot be detected for 1,2-
PG. However, after annealing to 870 K, surface order is
restored, and the D2 yield is also recovered. These results
provide additional evidence for the importance of the proper
alignment of OD groups of neighboring glycols for D2
formation. Previous studies have also shown that reactions
between two neighboring alcohols are completely suppressed
on Ne+-sputtered TiO2(110).30
(Figure 4b) were dosed at 80 K on three differently prepared
surfaces, i.e., r-TiO2(110), h-TiO2(110), and o-TiO2(110). For
both glycols, D2 TPD peaks from h- and r-TiO2(110) are nearly
identical, suggesting that hydroxylation does not influence D2
formation. Interestingly, on o-TiO2(110), D2 formation is
completely suppressed for both EG and 1,2-PG.
These results can be interpreted in light of what is known
about the effect of surface treatments on the charge state of the
TiO2(110) surface.4 Both photoemission spectroscopy31 and
electron energy loss spectroscopy32 studies have shown that
both r- and h-TiO2(110) exhibit essentially the same defect-
related electronic states. In contrast, the oxidation of
TiO2(110) is known to remove these states.31 Furthermore,
in a recent study27 Petrik et al. demonstrated that the amount
of charge associated with the surface defects determines the
amount of chemisorbed O2, and is equivalent on both r-
TiO2(110) and h-TiO2(110). Similarly, D2 formation from the
glycol OD groups on r- and h-TiO2(110) follows the same
trend.
While for o-TiO2(110), the oxygen adatoms left on Ti5c
rows33,34 are expected to serve as a barrier for the formation of
glycol pairs, such a small amount (∼5%) should not
significantly suppress D2 formation. A more likely origin of
the complete suppression of D2 production stems from the
electron scavenging properties of Oa’s.2,31,35 As a result, on o-
TiO2(110) the defect electrons become unavailable to adsorbed
glycol molecules, preventing the redox reaction leading to D2.
These results clearly show that subsurface charge is required in
D2 formation.
Since D2 formation results from a redox reaction that
requires transfer of electrons from TiO2(110) to the adsorbed
glycol molecules, we have investigated the role the availability
of subsurface charge plays in the reaction. To this end,
saturation coverages of OD-labeled EG (Figure 4a) and 1,2-PG
The chemical insight derived from the glycol studies
presented here should be also applicable to alcohol reactions
on TiO2(110). Surprisingly, hydrogen is generally not
5561
dx.doi.org/10.1021/ja500992b | J. Am. Chem. Soc. 2014, 136, 5559−5562

Journal of the American Chemical Society
Communication
observed15−18 and/or represents only a minority product.22
This can be understood upon closer inspection. From Figure
1a, D2 desorption peaks at ∼400 K. For alcohols, the
desorption from Ti5c sites is usually maximized between 280
and 350 K,29 at least 50 K lower than the temperature of
hydrogen formation. This means that at the temperature of
hydrogen formation, most of the alcohol molecules have
already desorbed, and only alkoxy species formed by alcohol
dissociation on VO sites are present. As such, the probability of
having two neighboring Ti5c-bound alcohol molecules at
temperatures required for H2 formation is negligible. In
contrast, the glycol molecules with two hydroxyl groups bind
more strongly with TiO2(110) and hence desorb at a much
higher temperature than alcohols (see Figure S3 in SI). More
importantly, the fraction of glycols that desorb intact is small
(∼20% for EG) as the majority is being converted to
products.23 Therefore, there is sufficient coverage of glycols
on TiO2(110) at ∼400 K for the formation of two neighboring
glycol species and D2.
In summary, we have shown that molecular hydrogen
formation from glycols on TiO2(110) is dictated by many
factors, that include glycol coverage, steric constraints of glycol
molecules, surface order, and the charge state of TiO2(110).
Hydrogen formation is only observed at high glycol coverages.
Increasing the steric hindrance of glycols is found to inhibit and
eventually eliminate hydrogen formation. Damaging the surface
order or scavenging the available surface charge of TiO2(110)
are also shown to completely suppress hydrogen formation.
These findings provide strong evidence that hydrogen
formation results from the bimolecular reaction between two
hydroxyl groups of neighboring Ti5c-bound glycols, and that the
redox reaction is driven by defect electrons of TiO2(110).
REFERENCES
(1) Diebold, U. Surf. Sci. Rep. 2003, 48, 53.
(2) Henderson, M. A. Surf. Sci. Rep. 2011, 66, 185.
(3) Pang, C. L.; Lindsay, R.; Thornton, G. Chem. Rev. 2013, 113,
3887.
■
(4) Dohnal
85, 161.
́
ek, Z.; Lyubinetsky, I.; Rousseau, R. Prog. Surf. Sci. 2010,
(5) Henderson, M. A.; Lyubinetsky, I. Chem. Rev. 2013, 113, 4428.
(6) Chen, X.; Shen, S.; Guo, L.; Mao, S. S. Chem. Rev. 2010, 110,
6503.
(7) Ye, M.; Gong, J.; Lai, Y.; Lin, C.; Lin, Z. J. Am. Chem. Soc. 2012,
134, 15720.
(8) Khnayzer, R. S.; Thompson, L. B.; Zamkov, M.; Ardo, S.; Meyer,
G. J.; Murphy, C. J.; Castellano, F. N. J. Phys. Chem. C 2011, 116,
1429.
(9) Wendt, S.; Matthiesen, J.; Schaub, R.; Vestergaard, E. K.;
Lægsgaard, E.; Besenbacher, F.; Hammer, B. Phys. Rev. Lett. 2006, 96,
066107.
(10) Bikondoa, O.; Pang, C. L.; Ithnin, R.; Muryn, C. A.; Onishi, H.;
Thornton, G. Nat. Mater. 2006, 5, 189.
(11) Zhang, Z.; Bondarchuk, O.; White, J. M.; Kay, B. D.; Dohnal
Z. J. Am. Chem. Soc. 2006, 128, 4198.
(12) Zhang, Z.; Bondarchuk, O.; Kay, B. D.; White, J. M.; Dohnal
Z. J. Phys. Chem. C 2007, 111, 3021.
́
ek,
́
ek,
́
(13) Zhang, Z.; Rousseau, R.; Gong, J.; Kay, B. D.; Dohnalek, Z. J.
Am. Chem. Soc. 2009, 131, 17926.
(14) Henderson, M. A. Langmuir 1996, 12, 5093.
(15) Henderson, M. A.; Otero-Tapia, S.; Castro, M. E. Faraday
Discuss. 1999, 114, 313.
(16) Kim, Y. K.; Kay, B. D.; White, J. M.; Dohnal
C 2007, 111, 18236.
(17) Kim, Y. K.; Kay, B. D.; White, J. M.; Dohnal
2008, 602, 511.
(18) Gamble, L.; Jung, L. S.; Campbell, C. T. Surf. Sci. 1996, 348, 1.
(19) Yin, X. L.; Calatayud, M.; Qiu, H.; Wang, Y.; Birkner, A.; Minot,
́
ek, Z. J. Phys. Chem.
́
ek, Z. Surf. Sci.
C.; Woll, C. ChemPhysChem 2008, 9, 253.
̈
(20) Kowalski, P. M.; Meyer, B.; Marx, D. Phys. Rev. B 2009, 79,
115410.
ASSOCIATED CONTENT
■
S
* Supporting Information
(21) Du, Y.; Petrik, N. G.; Deskins, N. A.; Wang, Z.; Henderson, M.
A.; Kimmel, G. A.; Lyubinetsky, I. Phys. Chem. Chem. Phys. 2012, 14,
3066.
(22) Xu, C.; Yang, W.; Guo, Q.; Dai, D.; Chen, M.; Yang, X. J. Am.
Chem. Soc. 2013, 135, 10206.
(23) Li, Z.; Kay, B. D.; Dohnal
15, 12180.
(24) Farfan-Arribas, E.; Madix, R. J. J. Phys. Chem. B 2002, 106,
Preparation procedure for OD-labeled glycols; TPD spectra
from TiO2(110) saturated by commercially available and H/D-
exchanged 1,2-PG; TPD spectra of products following 0.06 and
0.57 ML EG dose; TPD spectra of parent glycol molecules; C
1s XPS spectra for glycol saturation coverages; D2 desorption
yields as a function of glycol coverages. This material is
available free of charge via the Internet at http://pubs.acs.org.
́
ek, Z. Phys. Chem. Chem. Phys. 2013,
10680.
́
(25) Dohnalek, Z.; Kim, J.; Bondarchuk, O.; White, J. M.; Kay, B. D.
AUTHOR INFORMATION
J. Phys. Chem. B 2006, 110, 6229.
■
(26) Zehr, R. T.; Henderson, M. A. Surf. Sci. 2008, 602, 1507.
Corresponding Author
zdenek.dohnalek@pnnl.gov
(27) Petrik, N. G.; Zhang, Z.; Du, Y.; Dohnal
Kimmel, G. A. J. Phys. Chem. C 2009, 113, 12407.
(28) Kim, B.; Li, Z.; Kay, B. D.; Dohnalek, Z.; Kim, Y. K. J. Phys.
Chem. C 2012, 116, 1145.
(29) Li, Z.; Smith, R. S.; Kay, B. D.; Dohnal
2011, 115, 22534.
(30) Bondarchuk, O.; Kim, Y. K.; White, J. M.; Kim, J.; Kay, B. D.;
Dohnalek, Z. J. Phys. Chem. C 2007, 111, 11059.
́
ek, Z.; Lyubinetsky, I.;
Present Address
́
†Z.L.: Evans Analytical Group, Sunnyvale, CA
́
ek, Z. J. Phys. Chem. C
Notes
The authors declare no competing financial interest.
́
ACKNOWLEDGMENTS
(31) Wendt, S.; Sprunger, P. T.; Lira, E.; Madsen, G. K. H.; Li, Z.;
Hansen, J. Ø.; Matthiesen, J.; Blekinge-Rasmussen, A.; Lægsgaard, E.;
Hammer, B.; Besenbacher, F. Science 2008, 320, 1755.
(32) Henderson, M. A.; Epling, W. S.; Peden, C. H. F.; Perkins, C. L.
J. Phys. Chem. B 2003, 107, 534.
■
We thank R. Rousseau and N. G. Petrik for numerous
stimulating discussions. This work was supported by the U.S.
Department of Energy, Office of Basic Energy Sciences,
Division of Chemical Sciences, Geosciences & Biosciences,
and performed in EMSL, a national scientific user facility
sponsored by the Department of Energy’s Office of Biological
and Environmental Research and located at Pacific Northwest
National Laboratory (PNNL). PNNL is a multiprogram
national laboratory operated for the DOE by Battelle.
(33) Epling, W. S.; Peden, C. H. F.; Henderson, M. A.; Diebold, U.
Surf. Sci. 1998, 412−413, 333.
(34) Henderson, M. A.; Epling, W. S.; Perkins, C. L.; Peden, C. H. F.;
Diebold, U. J. Phys. Chem. B 1999, 103, 5328.
(35) Zhang, Z.; Cao, K.; Yates, J. T., Jr. J. Phys. Chem. Lett. 2013, 4,
674.
5562
dx.doi.org/10.1021/ja500992b | J. Am. Chem. Soc. 2014, 136, 5559−5562