Selective oxidation of cyclohexanol and 2-cyclohexen-1-ol on O/Au(111): The effect of molecular structure

Article abstract of DOI:10.1021/la1015302

We combine reactivity studies with infrared reflection absorption spectroscopy to provide molecular-scale insights into the oxidation of two cyclic alcohols, cyclohexanol and 2-cyclohexen-1-ol, by atomic oxygen adsorbed on Au(111). The two alcohols share common features in their reaction pathways: they are both activated by Bronsted acid-base reactions with adsorbed oxygen. Cyclic ketones, cyclohexanone and 2-cyclohexen-1-one, are the major products, formed from cyclohexanol and 2-cyclohexen-1-ol, respectively. These ketones also undergo secondary ring C-H bond activation. The product distributions reflect a substantial difference in the secondary reactions for these two ketones, which correlate with their gas-phase acidity. The allylic alcohol (2-cylohexen-1-ol) has a greater degree of ring C-H activation, yielding the diketone (2-cyclohexene-1,4-dione) and phenol. Our results provide clear evidence for the importance of C=C functionalities in determining the reactivity of molecules in heterogeneous oxidative transformations promoted on Au-based materials.

Full text of DOI:10.1021/la1015302

pubs.acs.org/Langmuir
© 2010 American Chemical Society
Selective Oxidation of Cyclohexanol and 2-Cyclohexen-1-ol
on O/Au(111): The Effect of Molecular Structure†
Xiaoying Liu‡ and Cynthia M. Friend*,‡,§
‡Department of Chemistry and Chemical Biology, §School of Engineering and Applied Sciences,
Harvard University, 12 Oxford Street, Cambridge, Massachusetts 02138
Received April 17, 2010. Revised Manuscript Received June 29, 2010
We combine reactivity studies with infrared reflection absorption spectroscopy to provide molecular-scale insights into
the oxidation of two cyclic alcohols, cyclohexanol and 2-cyclohexen-1-ol, by atomic oxygen adsorbed on Au(111). The two
alcohols share common features in their reaction pathways: they are both activated by Brønsted acid-base reactions with
adsorbed oxygen. Cyclic ketones, cyclohexanone and 2-cyclohexen-1-one, are the major products, formed from
cyclohexanol and 2-cyclohexen-1-ol, respectively. These ketones also undergo secondary ring C-H bond activation. The
product distributions reflect a substantial difference in the secondary reactions for these two ketones, which correlate with
their gas-phase acidity. The allylic alcohol (2-cylohexen-1-ol) has a greater degree of ring C-H activation, yielding the
diketone(2-cyclohexene-1,4-dione) and phenol. Ourresultsprovideclear evidence for the importance of CdC functionalities
in determining the reactivity of molecules in heterogeneous oxidative transformations promoted on Au-based materials.
Introduction
nol on oxidized Au(111) at low pressure with catalytic processes,
including ethanol oxidation over Au nanoparticles supported on
metal oxides such as TiO2 under high-pressure, aqueous phase
conditions14,15 and vapor-phase methanol oxidative coupling by
nanoporous Au.16 In parallel studies of the catalytic oxidative
coupling of methanol using nanoporous Au at atmospheric
pressure and using O2 as an oxidant, the mechanisms established
at low pressures on O/Au(111) successfully predicted the catalytic
selectivity as a function of oxygen partial pressure. These studies
illustrate the value of our mechanistic studies as a means of identi-
fying key elementary steps in catalytic processes on Au-based
materials and, in particular, the effect of oxygen coverage. Even
though the steady-state oxygen coverage is low in catalytic
processes because of the low rate of O2 dissociation and the high
reactivity of atomic O, it is important to identify pathways that
are opened by excess local concentrations of O on the surface.
In this work, we present a mechanistic study of the oxidation of
two cyclic alcohols, cyclohexanol and 2-cyclohexen-1-ol, over
O/Au(111) in order to examine the effect of the acidity of ring
hydrogens on the reaction pattern. By analogy with ethanol and
methanol, dissociation of the alcoholic O-H bond induced by
adsorbed oxygen atoms, in a Brønsted acid-base reaction, is
anticipated to be the first step. Our infrared spectroscopic studies
indicate that subsequent β-H elimination is the rate-limiting step
in the formation of corresponding ketones of the cyclic, secondary
alcohols under study here.
Selective oxidation of alcohols represents one of the most
important organic transformations in chemical synthesis. Gold-
based catalytic transformations of alcohols have been intensely
investigated, due to the promise for selective and energy-efficient
oxidation processes to replace current technologies that use heavy
metal-based oxidizing reagents. The conversion of primary1-3
and secondary alcohols3,4 as well as polyols such as glucose5,6
have been previously reported using supported gold nanoparti-
cles. A recent study by Miyamura et al. showed that aerobic
oxidation of secondary alcohols was catalyzed at room tempera-
ture by gold nanoclusters supported on polymers.7
Molecular-level understanding of the alcohol conversion is
crucial to guide the design of processes and materials and to
control selectivity. To this end, oxidation of alcohols has been
studied over gold surfaces precovered with atomic oxygen.8-12
Our recent studies of ethanol and methanol oxidation over oxi-
dized Au(111) (O/Au(111))9,10 and prior studies13 have provided
the identity of reaction intermediates and elementary steps. There
are strong parallels between the oxidation of ethanol and metha-
† Part of the Molecular Surface Chemistry and Its Applications special
issue.
*To whom correspondence should be addressed. E-mail: cfriend@
seas.harvard.edu.
(1) Abad, A.; Concepcion, P.; Corma, A.; Garcia, H. Angew. Chem., Int. Ed.
2005, 44, 4066.
Product distributions were measured at two different initial O
coverages, 0.2 and 0.5 ML. Under the conditions used to create
these O coverages, rough surfaces containing Au particles covered
with O are created. The majority of the particles are 2 nm in
diameter for 0.2 ML of O, whereas larger particles, on the order of
10-20 nm in diameter and with a greater degree of order, are
formed at an O coverage of 0.5 ML.17
(2) Enache, D. I.; Knight, D. W.; Hutchings, G. J. Catal. Lett. 2005, 103, 43.
(3) Biella, S.; Rossi, M. Chem. Commun. 2003, 378.
(4) Abad, A.; Almela, C.; Corma, A.; Garcia, H. Tetrahedron 2006, 62, 6666.
(5) Biella, S.; Prati, L.; Rossi, M. J. Catal. 2002, 206, 242.
(6) Prati, L.; Porta, F. Appl. Catal., A 2005, 291, 199.
(7) Miyamura, H.; Matsubara, R.; Miyazaki, Y.; Kobayashi, S. Angew. Chem.,
Int. Ed. 2007, 46, 4151.
(8) Gong, J. L.; Mullins, C. B. J. Am. Chem. Soc. 2008, 130, 16458.
(9) Liu, X. Y.; Xu, B. J.; Haubrich, J.; Madix, R. J.; Friend, C. M. J. Am. Chem.
Soc. 2009, 131, 5757.
(10) Xu, B. J.; Liu, X. Y.; Haubrich, J.; Madix, R. J.; Friend, C. M. Angew.
Chem., Int. Ed. 2009, 48, 4206.
(11) Gong, J. L.; Flaherty, D. W.; Ojifinni, R. A.; White, J. M.; Mullins, C. B.
J. Phys. Chem. C 2008, 112, 5501.
(14) Abad, A.; Corma, A.; Garcia, H. Chem.;Eur. J. 2008, 14, 212.
(15) Liu, X. Y.; Madix, R. J.; Friend, C. M. Chem. Soc. Rev. 2008, 37, 2243.
(16) Wittstock, A.; Zielasek, V.; Biener, J.; Friend, C. M.; Baeumer, M. Science
2010, 327, 319.
(12) Gong, J. L.; Flaherty, D. W.; Yan, T.; Mullins, C. B. ChemPhysChem 2008,
9, 2461.
(17) Min, B. K.; Alemozafar, A. R.; Pinnaduwage, D.; Deng, X.; Friend, C. M.
J. Phys. Chem. B 2006, 110, 19833.
(13) Outka, D. A.; Madix, R. J. J. Am. Chem. Soc. 1987, 109, 1708.
16552 DOI: 10.1021/la1015302
Published on Web 07/21/2010
Langmuir 2010, 26(21), 16552–16557

Liu and Friend
Article
Our studies presented herein strongly parallel prior work
describing the oxidation of secondary alcohols catalyzed by
supported gold nanoparticles, where ketones are formed with
near 100% selectivity.3,4,7 In addition, we show that the activation
of the ring C-H bonds leads to the modification to the ring
skeleton, which opens new reaction pathways, namely, O inser-
tion into an allylic C-H bond on the ring to yield the diketone,
and dehydrogenation around the ring to afford phenol. The
stability of the aromatic phenol product also reduces the amount
of combustion for 2-cyclohexen-1-ol relative to cyclohexanol.
Comparison of the product distributions for these two alcohols
establishes that the CdC bond functionality on the ring plays a
role in determining the reactivity by inducing slightly different
acidities for the C-H bonds. These results suggest that the effect
of other functional groups on molecular bonding is important to
consider in directing the oxidation of alcohols.
Experimental Section
All experiments were performedin an ultrahigh vacuum (UHV)
chamber with a base pressure below 2 Â 10-10 Torr. The system
was equipped with an IR spectrometer (Thermo Nicolet 670), a
mass spectrometer (Pfeiffer Prisma QMS 200) for temperature
programmed reaction spectroscopy (TPRS), a cylindrical mirror
analyzer (Physical Electronics model 15-155) for Auger electron
spectroscopy (AES), and low energy electron diffraction (LEED)
optics (Physical Electronics model 15-180).
Gold precovered with atomic oxygen (O/Au(111)) was pre-
pared by exposure of clean Au(111) to ozone at 200 K. Under
these conditions, O-covered Au particles, 85% of which are ∼2 nm
in diameter, are present. Silica gel was used to trap ozone and
maintained at -77 °C using a mixture of ethanol and dry ice. The
oxygen coverage was calibrated by comparing the amount of O2
evolution in temperature programmed reactions (TPR) to that for
a saturation coverage of oxygen atoms, which is defined as 1 ML.
Multiple layers of both alcohols were condensed on O/Au(111) at
170 K. The total exposure was fixed for the two oxygen coverages,
0.2 and 0.5 ML. Temperature programmed reaction studies were
conducted at an average heating rate of ∼5 K/s.
Figure 1. Temperature programmed reaction of cyclohexanol
(m/z 100) on atomic oxygen precovered Au(111) with an initial O
coverage of 0.2 ML. The spectra show the formation of cyclohexa-
none (m/z 98), 2-cyclohexen-1-one (m/z 96), CO2 (m/z 44), and
H2O (m/z 18), in addition to desorption of cyclohexanol. No other
products were observed; in particular, there was no peak at m/z
110. The oxidized Au surfaces were prepared by exposure to ozone
at 200 K.
Table 1. Relative Yields of Products of Cyclohexanol Oxidation on
O/Au(111)a Using Cyclohexanone As a Reference
relative yieldb
0.2 ML 0.5 ML
selectivityc
product
0.2 ML
0.5 ML
cyclohexanone
2-cyclohexen-1-one
CO2
1
0.01
0.33
1
0.01
0.87
94%
1%
5%
87%
1%
12%
The reaction products, cyclohexanone, 2-cyclohexen-1-one,
2-cyclohexene-1,4-dione, and phenol, were identified by quanti-
tative comparison of their fragmentation patterns to those obtai-
ned for multilayers of authentic samples sublimed from clean
Au(111) obtainedinour mass spectrometer (SupportingInforma-
tion, Table S1). 2-Cyclohexene-1,4-dione was identified by com-
paring the mass fragmentation patterns for the product in
temperature programmed reaction with those from the NIST
database.
Infrared reflection absorption spectroscopy (IRAS) experi-
ments were conducted on a Thermo Nicolet 670 spectrometer
using the MCT-A semiconductor photodiode detector. All spec-
tra were acquired at 130 K. Background spectra were collected at
130 K after heating to 500 K; the baselines of the resulting
absorbances were corrected. Two sets of 300 scans were coadded
for each experiment with a resolution of 4 cm-1. The lower bound
of frequencies that are probed in our experiment is ∼1000 cm-1
because we use CaF2 windows on our chamber.
a The Au(111) surface was oxidized via exposure to ozone at 200 K.
b Ionization efficiency, quadrupole transmission, multiplier gain, and
yield of the parent ion have been corrected. c The production of 6 CO2
molecules per cyclohexanol has also been taken into account.
well-defined water peak at 240 K, above the temperature for the
molecular desorption which is known to be 170 and 190 K on
O/Au(111),18 indicating that the overall kinetics for the most
prominent water peak at 240 K is limited by reaction. Further-
more, the coincident evolution of H2O and cyclohexanone is
strong evidence that water formation is associated with dehydro-
genation to produce the ketone. No other species were detected
other than cyclohexanol desorption at 190 K, attributed to
sublimation of the condensed alcohol layers. There is no residual
oxygen in the reaction because no O2 is detected at the character-
istic temperature of 535 K (data not shown). There is also no
residual carbon, since no CO2 is produced during subsequent
temperature programmed oxidation using ozone. Thus, the
selectivity for cyclohexanone formation is high.
Results and Discussion
Oxidation of Cyclohexanol. The predominant product of
cyclohexanol reaction with atomic oxygen adsorbed on Au(111)
(θO = 0.2 ML) is cyclohexanone, formed in a peak at 240 K
during temperature programmed reaction (Figure 1). There is
also a trace amount of 2-cyclohexen-1-one at 350 K, trailing the
cyclohexanone formation, and a small amount of combustion to
yield CO2 in a broad feature in the region of 210-460 K. The
evolution of H2O commences at ∼160 K, indicating that stripping
of hydrogen occurs near the adsorption temperature. There is a
The yields of all products increaseas the initial oxygencoverage
is raised to 0.5 ML (Table 1). For an initial oxygen coverage of
0.2 ML, there is 94% selectivity for cyclohexanone production.
2-Cyclohexen-1-one and CO2 account for only 1% and 5% of the
products, respectively. The yield of cyclohexanone increases by a
(18) Quiller, R. G.; Baker, T. A.; Deng, X. Y.; Colling, M. E.; Min, B. K.;
Friend, C. M. J. Chem. Phys. 2008, 129, 064702.
Langmuir 2010, 26(21), 16552–16557
DOI: 10.1021/la1015302 16553

Article
Liu and Friend
the C-O stretch peak at 1075 cm-1 by a factor of ∼2 (Figure 2).
The other vibrations observed are the C-H stretch modes at 2936
and 2858 cm-1, C-H deformations at 1453 and 1363, and a C-H
bend at 1259 cm-1 (Figure 2), which are assigned by comparison
to gas-phase cyclohexanol.23 The C-H bend mode is also
increased in intensity for the O/Au(111) surface. We interpret
the significant increase in the intensity of the ν(C-O) mode at
1075 cm-1 on O/Au(111) relative to that for cyclohexanol on
clean Au(111) as an indication that the alkoxy is formed. Gene-
rally, the C-O stretch intensity is significantly higher for the
alkoxy comparedto the alcohol, most likelydue to chargeredistri-
bution and stronger coupling to metal electrons upon O-H bond
dissociation.24 Furthermore, there is precedent for the formation
of alkoxy from O-assisted O-H bond dissociation of methanol
and ethanol on the same O/Au(111) surface. Another test for
alkoxy formation is the disappearance of the O-H bend mode;
however, we are not able toobserve the OH bendmodebecause its
frequency is ∼750 cm-1, below our detection cutoff for our
spectrometer (∼1000 cm-1). Unfortunately, the infrared data
do not provide unambiguous evidence for the presence of the
cyclohexoxy species.
Figure 2. Infrared reflection absorption spectra (IRAS) for cyclo-
hexanol adsorbed on the O/Au(111) surface (upper traces) and on
thecleanAu(111)surface (middletraces) andfor cyclohexanoneon
the clean Au(111) surface (bottom traces). Both molecules were
dosed onto the surface at 170 K. All spectra were collected after the
sample was cooled down to 130 K.
Scheme 1. Schematic of the Oxidation of Cyclohexanol to Form
Cyclohexanone and H2O via O-H Activation and β-H Elimination
Promoted by Adsorbed O
The O-H stretch also does not unequivocally resolve the ques-
tion about whether an alkoxide forms at 170 K. There is a broad
feature centered at 3288 cm-1, which is characteristic of an O-H
stretch, visible after cyclohexanol adsorption on O/Au(111)
(Figure 2). The width of this vibration indicates that there is
substantial hydrogen bonding. Notably, there is very little inten-
sity in this region for cyclohexanol on clean Au(111), making it
difficult to assign unambiguously. This peak could be due to
water, which could either be the result of O-H bond dissociation
associated with alkoxy formation or from ice building up in the
MCT detector.25 Hydroxyl is not stable with respect to dispro-
portionation on Au(111); instead, hydrogen-bonded water com-
plexes are formed on O/Au(111), which have desorption peaks at
170 and 190 K.18,26 On clean Au(111), water desorbs intact in a
peak at 170 K, so it is not expected to accumulate in the
cyclohexanol experiment shown for clean Au. Another possibility
for the peak at 3288 cm-1 is the O-H stretch of hydrogen bonded
alcohol molecules, due to the presence of surface oxygen (Figure 2).
Although the infrared data do not provide conclusive evidence
for the formation of cyclohexoxy, we argue by analogy to
previous spectroscopic studies (high-resolution electron energy
loss) of methanol10 and ethanol9 on the same oxidized Au surface,
in which dissociation of the O-H bond is the first reaction step at
170 K and the analogous process occurs along the path to
formation of the ketone, cyclohexanone. The deprotonation of
the alcoholic O-H has been widely established to occur on silver,
gold, and copper surfaces through the Brønsted acid-base
reaction with surface oxygen.13,19-22
factor of 2.5 for the oxygen coverage of 0.5 ML, based on the
increase in the integrated area of the corresponding peaks in the
temperature programmed reaction data (Table 1). These data
suggest that the coverage of O determines the amount of reaction
overall and that there is not a strong sensitivity to the surface
order or Au particle size. The amount of H2O likewise increases
to 2.5 times that for the oxygen coverage of 0.2 ML, providing
evidence for the link between water and cyclohexanone formation.
A common pathway for the formation of cyclohexanone and
H2O would be initial O-H bond dissociation in cyclohexanol by
atomic oxygen, followed by elimination of the H at the carbon
adjacent to O in the alkoxy (Scheme 1). Analogous production of
cyclohexanone via a similar pathway for cyclohexanol was pre-
viously reported for O-covered Ag(110).20 In previous studies on
coinage metals, alkoxy formation was generally reported and the
ensuing β-H elimination of the alkoxy leads to the formation of
ketones or aldehydes,13,19-22 in good agreement with our results.
Vibrational (infrared reflection absorption) spectroscopy pro-
vides supporting evidence that the rate of cyclohexanone evolu-
tion in the oxidation of cyclohexanol is determined by β-H
elimination. First, no cyclohexanone is detected using infrared
spectroscopy after adsorption of cyclohexanol on O/Au(111)
(θO = 0.2 ML) at 170 K (Figure 2). Specifically, there is no inten-
sity in the region expected for the CdO bond stretch, which was
measured to be 1714 and 1683 cm-1 for the ketone itself (Figure 2).
The infrared data obtained are consistent with the coexistence
of intact cyclohexanol and the alkoxy, cyclohexoxy. The spectrum
obtained for cyclohexanol on clean Au(111) is used as a reference
for the intact alcohol, since it reversibly desorbs from the surface
without reaction. The most significant difference in the spectrum
for cyclohexanol on O/Au(111) is the increase in the intensity of
The formation of the ketone with a CdC bond, 2-cyclohexen-
1-one, is a pathway for which there is no precedent in the previous
studies of linear alcohols. This pathway is attributed to secondary
reaction of cyclohexanone. In separate studies of cyclohexanone
on O/Au(111), 2-cyclohexen-1-one is also produced at 355 K
(Figure 3), a temperature very similar to that observed in the
oxidation of cyclohexanol (Figure 1). Therefore, the formation of
2-cyclohexen-1-one is attributed to the dehydrogenation of the
predominant product, cyclohexanone, which has acidic C-H
(23) Linstrom, P. J.; Mallard, W. G. NIST Chemistry WebBook; National
Institute of Standards and Technology: Gaithersburg, MD (http://webbook.nist.gov).
(24) Weldon, M. K.; Friend, C. M. Chem. Rev. 1996, 96, 1391.
(25) Camplin, J. P.; McCash, E. M. Surf. Sci. 1996, 360, 229.
(26) Ojifinni, R. A.; Froemming, N. S.; Gong, J. L.; Pan, M.; Kim, T. S.; White,
(19) Wachs, I. E.; Madix, R. J. Surf. Sci. 1978, 76, 531.
(20) Wachs, I. E.; Madix, R. J. Surf. Sci. 1978, 1, 303.
(21) Merrill, P. B.; Madix, R. J. Langmuir 1991, 7, 3034.
(22) Madix, R. J.; Telford, S. G. Surf. Sci. 1995, 328, L576.
J. M.; Henkelman, G.; Mullins, C. B. J. Am. Chem. Soc. 2008, 130, 6801.
16554 DOI: 10.1021/la1015302
Langmuir 2010, 26(21), 16552–16557

Liu and Friend
Article
temperature at which cyclohexanone is produced from cyclohex-
anol(Figure1). The selective formation of the ketoneis analogous
to cyclohexanone formation from cyclohexanol and to the
production of aldehydes and ketones from primary8-10,12,16 and
secondary27 alcohols, respectively.
There are also two new dehydrogenation mechanisms observed
for the oxidation of 2-cyclohexen-1-ol: formation of phenol and
2-cyclohexene-1,4-dione (Figure 4a). These products would form
via dehydrogenation around the ring and O insertion into the
allylic C-H bond at the 4 position, respectively (Scheme 2). The
selectivity for phenol and 2-cyclohexene-1,4-dione formation is
16% and 6%, respectively, under these conditions (Table 2). The
availability of these pathways decreases the selectivity to ketone
formation to 77% for the 2-cyclohexen-1-ol, as compared to 94%
for cyclohexanol under the same conditions (Table 1).
Phenol and 2-cyclohexene-1,4-dione both form from further
oxidation of2-cyclohexen-1-one, as demonstrated bytemperature
programmed reaction studies of 2-cyclohexen-1-one on the
O/Au(111) surface (Figure 4b). The temperatures for product
formation are nearly identical for the production of phenol and
2-cyclohexene-1,4-dione from both 2-cyclohexen-1-ol and 2-cy-
clohexen-1-one. The secondary dehydrogenation of the ring is
reminiscent of that found for its saturated counterpart, cyclohex-
anone (Figure 3).
The predominant reaction pathways for both cyclohexanoland
2-cyclohexen-1-ol are those anticipated from our general mechan-
istic studies of alcohols on O/Au(111) (Scheme 2, pathway A).
The ketones are produced from β-H elimination from an alkoxy
formed via O-H bond dissociation. The pattern of reactivity is
consistent with O adsorbed on Au acting as a Brønsted base. In
contrast to primary alcohols, such as ethanol and methanol, there
is no pathway for coupling reactions to form esters, as this would
require ring-opening. Two factors contribute to the selective
formation of the ketone: (1) the elimination of the β-C-H bond
is more facile because of the lower C-H bond strength of
secondary alcohols, and (2) coupling is prohibited because it
would require C-C bond breaking and ring-opening to form a
stable product. Mullins et al. also reported the ketone formation
with near 100% selectivity in the reaction of 2-butanol with
atomic oxygen on gold,27 which is consistent with our results
herein. The reactions observed inour experiments are also directly
analogous to those reported for Au nanoparticles embedded in a
polymer matrix.7 Room-temperature, aerobic oxidation of second-
ary alcohols, such as 1-phenylethanol, yielded ketones with
selectivity up to >99%.
Figure 3. Temperature programmed reaction of cyclohexanone
on the O/Au(111) surface. The oxidized Au surface was prepared
by exposure to ozone at 200 K with an initial oxygen coverage of
0.2 ML. Cyclohexanone was introduced onto the O/Au surface at
170 K. The products include 2-cyclohexen-1-one, CO2, and H2O.
No other products were observed; in particular, m/z 110 signal was
not detected.
bonds that are adjacent to the CdO group (ΔG = 358.7 kcal/
mol).23 Since this reaction only occurs when O is adsorbed on
Au(111), the dehydrogenation of the ring is generally consistent
with the pattern of Brønsted acid-base reactions.
There is also competing desorption of cyclohexanone and
combustion to CO2 and H2O in the temperature programmed
reaction following adsorption of cyclohexanone on O/Au(111)
(θO = 0.2 ML) (Figure 3). The peak temperature for cyclohexa-
none desorption is ∼5 K higher than the peak temperature for its
formation from cyclohexanol (Figure 1). Furthermore, there are
significant differences in their line shapes: a sharp onset with an
asymmetric peak is observed for cyclohexanone desorption
(Figure 3) versus a broad, somewhat more symmetric peak for
the reaction of cyclohexanol (Figure 1). One potential explana-
tion for these differences is that the desorption temperature
depends on the O coverage, which is decreased by production
of cyclohexanone from cyclohexanol since one oxygen per reac-
tion event is used by stoichiometry. Molecular desorption does
not require loss of O. Alternatively, reactive events leading to
combustion products (CO2 and H2O) occur at low temperature in
competition with cyclohexanone desorption, that could lead to an
apparent shift in the peak temperature and the unusual, asym-
metric line shape. In either case, the main point is illustrated:
secondary oxidation of cyclohexanone, the main product of
cyclohexanol reaction, yields 2-cyclohexen-1-one.
The introduction of the ring and, specifically, of an acidic C-H
bond in the unsaturated 2-cyclohexen-1-ol opens new reaction
pathways that were not reported for 2-butanol or other secondary
alcohols. It is widely known that acidic C-H bonds react with O
on coinage metals, often leading to combustion to CO2.15
Secondary oxidation of 2-cyclohexen-1-one formed from 2-cyclo-
hexen-1-ol opens the reactive pathway that involves dissociation
of the acidic (allylic) C-H bonds, similar to reactions of allylic
olefins.28,29
Oxidation of 2-Cyclohexen-1-ol. The reactions of 2-cyclo-
hexen-1-ol were investigated to further probe for possible ring
dehydrogenation via acid-base reactions. The allylic C-H bonds
on the ring of 2-cyclohexen-1-ol are slightly more acidic, with a
ΔG for proton abstraction from the ring of 349.8 kcal/mol,
compared to 358.7 for cylcohexanol.23
The ring structure leads to an important difference compared
to straight-chain allylic groups: there is considerably less combus-
tion because of the formation of phenol, which has a resonance-
stabilized structure. The mechanism for the formation of phenol
will be discussed in more detail elsewhere.30 The other pathway
2-Cyclohexen-1-one is the predominant product of 2-cyclohex-
en-1-ol reaction on O/Au(111) at the lower O coverage, 0.2 ML,
accounting for 77% of the product (Figure 4a and Table 2).
2-Cyclohexen-1-one evolves at 240 K, which is similar to the
(27) Yan, T.; Gong, J. L.; Mullins, C. B. J. Am. Chem. Soc. 2009, 131, 16189.
(28) Deng, X. Y.; Min, B. K.; Liu, X. Y.; Friend, C. M. J. Phys. Chem. B 2006,
110, 15982.
(29) Liu, X. Y.; Friend, C. M. J. Phys. Chem. C 2010, 114, 5141.
(30) Liu, X. Y.; Baker, T. A.; Friend, C. M. Unpublished work.
Langmuir 2010, 26(21), 16552–16557
DOI: 10.1021/la1015302 16555

Article
Liu and Friend
Figure 4. Temperatureprogrammedreactionof(a)2-cyclohexen-1-ol(m/z98) and(b)2-cyclohexen-1-one(m/z96) ontheoxygenprecovered
Au(111) with an initial O coverage of 0.2 ML. The formation of phenol (m/z 94) and 2-cyclohexene-1,4-dione (m/z 110) are observed in both
experiments.
Table 2. Relative Yields of Different Products in the Reaction of
2-Cyclohexen-1-ol on the Oxidized Au(111) Surface (θO = 0.2 ML)
(the Yield of 2-Cyclohexen-1-one Was Used as the Reference)
ring hydrogens: the CdC bond imparts a higher acidity, opening
this pathway.
The relative yields for products involving the activation of ring
C-H bonds are also very different for cyclohexanol and 2-cyclo-
hexen-1-ol. For an oxygen coverage of 0.2 ML, the relative yield
of phenol to 2-cyclohexen-1-one is 20% (Table 2), in sharp
contrast with 1% for the ratio of 2-cyclohexen-1-one to cyclohexa-
none as observed in the oxidation of cyclohexanol (Table 1).
This substantial difference for the dehydrogenation efficiency of
the two molecules, 2-cyclohexen-1-one versus cyclohexanone, is
attributed to the different gas-phase acidity of the ring C-H
bonds. Indeed, 2-cyclohexen-1-one is more acidic (ΔG = 349.8
versus 358.7 kcal/mol for cyclohexanone23), clearly due to the
presence of allylic C-H bonds at C-4. Therefore, it is more prone
to the selective attack by excess oxygen on the surface, leading to
more efficient secondary transformations (Scheme 2). Excess O
available in the oxidation of cyclohexanol instead leads to
combustion, as indicated by the higher relative yield of CO2 in
the reaction of cyclohexanol (Table 1) compared to 2-cyclohexen-
1-ol (Table 2). Furthermore, the Brønsted acidity of 2-cyclohexen-
1-one also paves the way for another pathway, namely, inser-
tion of oxygen into one of the allylic C-H bonds, which leads
to the production of 2-cyclohexene-1,4-dione. The analogous
reaction is not observed for the less acidic cyclohexanone, clearly
indicating that a small change in the chemical properties of
the reactants can result in substantially different reactivity
patterns.
product
relative yielda
selectivityb
2-cyclohexen-1-one
phenol
2-cyclohexene-1,4-dione
CO2
1
77%
16%
6%
0.20
0.07
0.15
1%
a Ionization efficiency, quadrupole transmission, multiplier gain, and
yield of the parent ion have been corrected. b The production of 6 CO2
molecules per combustion of 1 molecule of 2-ohexen-1-ol has been taken
into account.
Scheme 2. Schematics for Reaction of 2-Cyclohexen-1-ol through
(A) Brønsted Acid-Base Reaction to Form 2-Cyclohexen-1-one,
Which Further Reacts via (B) Ring-Dehydrogenation and
(C) Oxygen-Insertion to Ultimately Form Phenol and 2-Cyclohexene-
1,4-dione, Respectively
While the presence of 2-cyclohexen-1-oxy is inferred by ana-
logy to other alcohols (Scheme 2), we were not able to clearly
identify it on the surface using infrared spectroscopy (Figure 5).
We never observed an intense C-O stretch for reaction of
2-cyclohexen-1-ol on O/Au(111). Nevertheless, analogy with the
reactivity pattern for other alcohols suggests that the first step in
activating 2-cyclohexen-1-ol is formation of the alkoxy via
O-assisted O-H dissociation; however, we do not have firm
evidence for this intermediate. Since we do not detect significant
intensity in the O-H stretch region above 3000 cm-1 for either
2-cyclohexen-1-ol or phenol on either clean Au(111) or O/Au-
(111), the absence of this mode cannot be used to argue for alkoxy
formation.
for the transformation of 2-cyclohexen-1-one proceeds through
O-insertion into the allylic C-H bond (Scheme 2, pathway C) and
subsequent β-H elimination to form 2-cyclohexene-1,4-dione.
This mechanism is similar to that observed for the oxidation of
olefins with an allylic C-H bond, previously established for
propene oxidation to acrolein and acrylic acid28 and oxidation
of trans-β-methylstyrene to cinnamic acid.29 This pathway was
not observed for cyclohexanone formed in the oxidation of
cyclohexanol, which is consistent with the relative acidity of the
16556 DOI: 10.1021/la1015302
Langmuir 2010, 26(21), 16552–16557

Liu and Friend
Article
Figure 5. Infrared spectra for (a) 2-cyclohexen-1-one adsorbed on the O/Au(111) surface, (b) phenol on O/Au(111), (c) 2-cyclohexen-1-ol on
the clean Au(111) surface, and (d) 2-cyclohexen-1-ol on O/Au(111). All molecules were introduced onto the surface at 170 K. Spectrum (e) is
for the same preparation in (d) after annealing the surface to 315 K. All spectra were collected after the samples were cooled down to 130 K.
Our infrared data do provide clear evidence that neither
2-cyclohexen-1-one nor phenoxy accumulates on the surface as
it is heated (Figure 5). If 2-cyclohexen-1-one were present, there
would be a strong CdO stretch near 1686 cm-1 (Figure 5a).
However, there is no measurable difference in this spectral region
for 2-cyclohexen-1-ol on clean versus O-covered Au(111) at 170 K
(Figure 5c and d). If phenoxy were formed on the surface prior to
its evolution, there would be an intense C-O stretch peak at 1170
cm-1, based on the spectrum obtained for reaction of phenol on
O/Au(111) (Figure 5b) and in the C-H stretch region at 2957
cm-1. Again, there is no significant intensity in this region for
2-cyclohexen-1-ol on O/Au(111) at either 170 or 315 K (Figure 5d
and e).
Our results show the rich chemistry of the oxidation of allylic
alcohols and the effect of molecular structure on selectivity. We
clearly establish that the reactions of cyclic alcohols fit into the
general framework of Brønsted acid-base chemistry on Au,
similar to other alcohols.9,10,13 The presence of CdC bonds in
the alkyl group of the alcohols opens pathways for abstraction of
H from the adjacent allylic positions, which significantly de-
creases the selectivity for the production of ketones. The selec-
tivity for cyclohexanol and 2-cyclohexen-1-ol is 94% and 77%,
respectively. The same trend was also observed in aerobic oxida-
tion of secondary alcohols in solution, which establishes the
generality of this mechanism. In a previous study by Kaneda et
al., 2-octanol showed a selectivity of 90% toward ketone forma-
tion over supported gold catalyst with atmospheric O2, while the
selectivity decreased to 83% for the oxidation of 3-octen-2-ol,31
which suggests that CdC bonds facilitate dehydrogenation dur-
ing alcohol oxidation.
product yield from the saturated alcohol, cyclohexanol (Table 1),
whereas it is only 1% of the products from the oxidation of
2-cyclohexen-1-ol (Table 2) under the same conditions. We attri-
bute the reduction in the amount of combustion to the availability
of the pathway to form phenol, which is stabilized by resonance in
the ring. Our work suggests that molecular structure can be used
to manipulate the selectivity for oxidation of species containing
allylic protons.
Conclusions
The reaction patterns for cyclohexanol and 2-cyclohexen-1-ol
are examined using temperature programmed reactions and
infrared spectroscopy. Both O-H and ring C-H activation are
observed for these two secondary alcohols. Brønsted acid-base
reaction between O-H and surface oxygen followed by β-H
elimination leads to the formation of cyclic ketones, the most
predominant product in each reaction. Their selectivity is 94%
and 77% for cyclohexanol and 2-cyclohexen-1-ol, respectively.
Clearly, the efficiency of ring modification is governed by their
gas-phase acidity, with the allylic alcohol showing more C-H
bond activation. Our results provide mechanistic understanding
on the pathways that lead to side products and show, in general,
the importance of other functionalities in determining the reactive
patterns of alcohol oxidation over gold-based catalysts. We
further show that molecular structure can lead to thermodyna-
mically favorable products that have the effect of suppressing
combustion.
Acknowledgment. We gratefully acknowledge the support of
this work by the U.S. Department of Energy, Basic Energy
Sciences, under Grant No. FG02-84-ER13289.
The ring structure also has an effect on selectivity and product
distribution. While combustion is a major pathway that is
generally open when there are acidic, allylic protons in a molecule
reacting on O/Au, combustion is actually reduced when the allylic
proton is introduced into the C6 ring. CO2 accounts for 5% of the
Supporting Information Available: Identification of reac-
tion products based on the comparison of their mass frag-
mentation patterns with those of authentic samples conden-
sed on the clean Au(111) surface. This material is available
free of charge via the Internet at http://pubs.acs.org.
(31) Mitsudome, T.; Noujima, A.; Mizugaki, T.; Jitsukawa, K.; Kaneda, K.
Adv. Synth. Catal. 2009, 351, 1890.
Langmuir 2010, 26(21), 16552–16557
DOI: 10.1021/la1015302 16557