Mechanistic insights into selectivity control for heterogeneous olefin oxidation: Styrene oxidation on Au(111)

Article abstract of DOI:10.1002/asia.200900399

We demonstrate that intermolecular interactions, controlled by both oxygen and styrene coverage, alter reaction selectivity for styrene oxidation on oxygen-covered Au(111). Several partial oxidation products are formed-styrene oxide, acetophenone, benzoic acid, benzeneacetic acid, and phenylketene-in competition with combustion. The maximum ratio of the yields of styrene oxide to the total CO₂ produced is obtained for the maximum styrene coverage for the first two layers (0.28 ML) adsorbed on Au(111) precovered with 0.2 ML of O. Furthermore, our reactivity and infrared studies support a mechanism whereby styrene oxidation proceeds via two oxametallacycle intermediates which, under oxygen-lean conditions, lead to the formation of styrene oxide, acetophenone, and phenylketene. Benzoate, identified on the basis of infrared reflection absorption spectroscopy, is converted into benzoic acid during temperature-programmed reaction. These results demonstrate the ability to tune the epoxidation selectivity using reactant coverages and provide important mechanistic insight into styrene oxidation reactions.

Full text of DOI:10.1002/asia.200900399

FULL PAPERS
DOI: 10.1002/asia.200900399
Mechanistic Insights into Selectivity Control for Heterogeneous Olefin
Oxidation: Styrene Oxidation on Au ACHUNTGRNENUG( 111)
[
b]
[a]
[a, b]
Ryan G. Quiller, Xiaoying Liu, and Cynthia M. Friend*
78
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 78 – 86

Abstract: We demonstrate that inter-
molecular interactions, controlled by
both oxygen and styrene coverage,
alter reaction selectivity for styrene ox-
styrene coverage for the first two
layers (0.28 ML) adsorbed on Au(111)
precovered with 0.2 ML of O. Further-
more, our reactivity and infrared stud-
ies support a mechanism whereby sty-
rene oxidation proceeds via two oxa-
the formation of styrene oxide, aceto-
phenone, and phenylketene. Benzoate,
identified on the basis of infrared re-
flection absorption spectroscopy, is
converted into benzoic acid during
AHCTUNGTRENNUNG
idation on oxygen-covered Au ACHUTNRGNENUG( 111).
Several partial oxidation products are
formed—styrene oxide, acetophenone,
benzoic acid, benzeneacetic acid, and
phenylketene—in competition with
combustion. The maximum ratio of the
yields of styrene oxide to the total CO2
produced is obtained for the maximum
temperature-programmed
reaction.
metallacycle
intermediates
which,
These results demonstrate the ability to
tune the epoxidation selectivity using
reactant coverages and provide impor-
tant mechanistic insight into styrene
oxidation reactions.
under oxygen-lean conditions, lead to
Keywords: gold
·
mechanism
·
oxidation · selectivity · styrene
Introduction
One of the key factors limiting epoxidation selectivity is
the competing combustion pathway. On Ag, atomic
oxygen—a key reactant in epoxidation—acts as a Brønsted
base and reacts with relatively acidic CÀH bonds, leading to
Optimization of reaction selectivity for olefin epoxidation
processes is important industrially because epoxides are val-
uable precursors to chemical products, such as plastics and
combustion. The gas-phase acidity of the protons in styrene
[1]
coatings. By increasing selectivity and lowering the opera-
tion temperature for epoxidation, considerable energy sav-
ings are possible. Heterogeneous ethylene epoxidation is
catalyzed in commercial processes by Ag; however, Au is of
significant interest because Au nanoparticles supported on
titania selectively catalyze olefin epoxidation at low temper-
ature, thereby having the potential to lower energy costs
and ethylene is comparable: DG of deprotonation is 1677Æ
À1
2 and 1604Æ8 kJmol , respectively, for ethylene and sty-
[12]
rene; thus, styrene is a reasonable model for ethylene ep-
oxidation. Styrene epoxidation is also of interest in its own
right, due to its use as a chemical building block.
Model studies of styrene epoxidation on oxygen-covered
[
8]
[13]
[14–17]
[18]
Au
A
H
N
T
E
N
N
(111),
Ag
A
H
U
T
E
N
N
(100),
Ag
A
H
U
G
R
N
U
G
Cu
A
T
N
T
E
N
N
(110),
and
[2]
[19]
and increase reaction selectivity.
The ethylene epoxidation process on Ag has garnered
Cu(111)
ACHTUNGTRENNUNG
have provided insight into the mechanisms for
selective oxidation of styrene. The first step in epoxidation
is widely agreed to be addition of oxygen to one of the
[3–5]
considerable attention,
and the tools of modern surface
[6–8]
chemistry have provided mechanistic insights.
Ethylene
leading to a low surface
carbon atoms in the C=C bond, yielding an oxametallacycle
[9,10]
[7,15,16]
has a low adsorption enthalpy,
intermediate.
The asymmetry of the C=C bond in sty-
lifetime under ultrahigh vacuum (UHV) conditions; hence,
styrene has frequently been used as a model because, similar
to ethylene, it does not contain allylic CÀH bonds. On the
other hand, the phenyl ring will potentially lead to intermo-
lecular interactions and enhanced olefin-surface bonding
that may affect selectivity and activity.
rene leads to two possible oxametallacycles and inequivalent
combustion intermediates, both of which have been pro-
[11]
[16,17,20]
[8,21]
posed on Ag
Herein, we report an in-depth study of the mechanism for
styrene epoxidation on oxygen-covered Au(111) (O/Au-
(111)) prepared by oxidation using ozone at a surface tem-
and Au.
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
perature of 200 K. Oxidation using this method leads to the
formation of O-covered Au nanoparticles. The size distribu-
tion and degree of order in these particles depend on the
temperature used for oxidation and the total O coverage, as
[
a] X. Liu, Prof. C. M. Friend
Department of Chemistry and Chemical Biology
Harvard University
[22]
established previously. Most particles have a diameter of
1
2 Oxford Street, Cambridge, MA 02138 (USA)
ꢀ2 nm for a total O coverage of 0.2 monolayers (ML) and a
Fax : (+1)617-496-8410
E-mail: cfriend@seas.harvard.edu
[22]
surface temperature of 200 K.
At higher O coverage,
more ordered oxidic species are present.
[
b] R. G. Quiller, Prof. C. M. Friend
School of Engineering and Applied Sciences
Harvard University
We previously showed that styrene epoxidation is promot-
[8]
ed by atomic oxygen on Au
ACHTUNGTNERNUNG( 111) below room temperature.
2
7 Oxford Street, Cambridge, MA 02138 (USA)
Recently, both Hughes et al. and Turner et al. reported that
Chem. Asian J. 2010, 5, 78 – 86
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
79

C. M. Friend et al.
FULL PAPERS
gold nanoclusters supported on inert carbon substrates are
[23,24]
also active for the epoxidation of olefins,
leading to the
conclusion that the catalytic activity is a result of the intrin-
sic electronic structure of the gold nanoparticles, and not
due to electronic interaction with the support. For the oxida-
tion of styrene, a conversion of 20% with a selectivity of
[24]
1
4% for styrene epoxidation was reported. Notably, ace-
tophenone and benzaldehyde were also produced from sty-
rene oxidation catalyzed by gold nanoparticles derived from
55-atom gold clusters, which were most active of those in-
vestigated.
In this study, we show that the product distribution for ox-
idation of styrene on O/Au ACHUTNGRENNUG( 111) is dramatically influenced
by the reactant coverages. We also further probe the reac-
tion mechanism by investigating ring-opening of styrene ep-
oxide and reactions of phenylacetaldehyde. The selectivity
for styrene epoxidation displays a local maximum as a func-
tion of oxygen coverage, reflecting the competition between
O addition to form an oxametallacycle and subsequent sec-
ondary oxidation. The selectivity also monotonically increas-
es with styrene coverage up to the saturation coverage
Figure 1. Products formed during oxidation of styrene (0.28 ML) ad-
sorbed on O/Au
consists of Au nanoparticles, 85% of which have a diameter of ꢀ2 nm,
prepared by exposure of Au(111) to ozone at 200 K. The intensities
O
ACTHUNGTRENNUN(G 111) (q =0.2 ML) at 140 K. The O/Au ACHTUNGTRENNNU(G 111) surface
(
0.28 ML). Thus, optimizing selectivity for styrene epoxida-
tion on Au depends on the bonding of oxygen to the surface
and on intermolecular interactions among adsorbed molecu-
lar species. Our results indicate that selectivity can be im-
proved in catalytic reactions by modifying the local environ-
ment of the reacting species.
AHCTUNGTRENNUNG
shown are not corrected for spectrometric sensitivity factors, including
ionization efficiency, transmission gain, electron multiplier gain, and yield
of parent ions.
Results and Discussion
Table 1. The product evolved at 460 K is identified as acetophenone
based on comparison of the measured mass spectrometer fragmentation
patterns of C H O isomers condensed on AuACHTUNGTERNNNUG
8 8
Product Distribution in the Oxidation of Styrene
formed when styrene reacts on AuACHTNUGTRENNUNG
The products of styrene (q=0.28 ML) oxidation on
Isomer of C
8
H
8
O
m/z
20
m/z
119
m/z
m/z
m/z
m/z
O/Au
m/z=120) at 280 K, benzoic acid (parent ion, m/z=122) at
60 K, benzeneacetic acid (parent ion, m/z=136), and com-
bustion products (H O and CO ) at 420 K (Figure 1), in
ACHTUNGTRENNUNG( 111) (q =0.2 ML) are styrene oxide (parent ion,
O
1
105
91
90
77
styrene oxide
phenylacetaldehyde
acetophenone
17
12
14
13
19
0
0
3
0
56
69
100
100
2
36
4
0
23
1
100
100
3
2
2
[8]
agreement with previous studies. Two new products, not
observed previously, were also formed: acetophenone in a
small peak (with an integrated intensity of ca. 35% relative
to styrene oxide) at about 470 K and phenylketene (ca. 70%
relative to styrene oxide) at 435 K. No benzaldehyde pro-
duction was detected. The yields of these products depend
on the coverages of styrene and adsorbed O, as discussed
below.
product at 460 K
0
8
0
All products were identified based on quantitative analy-
sis of their fragmentation patterns, as shown for acetophe-
none in Table 1 and as summarized in our prior work for the
[25,26]
other products,
The only other product detected is H O,
2
which includes a small amount of background water that de-
[27–29]
sorbs at about 170 K.
The production of styrene oxide, acetophenone, and phe-
nylketene can all be explained by the initial formation of
two possible oxametallacycle intermediates, as proposed in
our earlier work (Scheme 1). Recently, the energetics for
the formation and conversion of these two oxametallacycles
Scheme 1. The reaction of branched (1) and linear (2) oxametallacycle in-
termediates for styrene oxidation on Au(111) to form acetophenone, phe-
nylketene, and styrene oxide. Pathways involving combustion, benzoic
acid, and benzeneacetic acid are omitted for simplicity.
AHCTUNGTRENNUNG
on Au ACHTUNGTRENNUNG( 111) were studied using density functional theory
80
www.chemasianj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 78 – 86

Mechanistic Insights into Selectivity Control for Heterogeneous Olefin Oxidation
(
DFT) and both isomers were found to be relatively stable,
the most combustion, and the lowest coverage generated
partial oxidation products including styrene oxide.
with oxametallacycle (1) favored by about 0.03 eV, which is
[21]
nearly the error of the calculations. Acetophenone could
be directly formed by O addition to the secondary carbon
atom in the C=C bond of oxametallacycle (1), followed by
H transfer to the terminal carbon atom (Scheme 1). This
type of 1,2-hydrogen shift was proposed to lead to combus-
tion of ethylene on Ag via acetaldehyde based on DFT stud-
ies.
hydrogen elimination from the other geometric isomer of
the oxametallacycle (2, Scheme 1).
An increase in the oxygen coverage from 0.1 to 0.2 ML
increased the yields of all products, but differentially in-
creased the yield of the epoxide by a factor of 2.6
(Figure 2). The increase in conversion overall is partly at-
tributed to the increase in the amount of styrene which ini-
tially reacts with O, implying that below 0.2 ML of oxygen,
oxygen addition limits the rate of production. Acetophe-
none, phenylketene, and benzeneacetic acid are more signif-
icant for the lowest oxygen coverage, 0.1 ML, suggesting
that at higher oxygen coverages secondary oxidation of
these intermediate products predominates.
[8]
[
7,30]
Phenylketene would most directly form through b-
[8,15,16,25]
The formation of
acetophenone and phenylketene is significant in our work
because it is evidence that both oxametallacycles form
during styrene oxidation on O/Au
A
H
U
G
R
N
N
(111) (Scheme 1).
The yields of products also depend on the coverage of sty-
rene as illustrated for a fixed coverage of O, 0.2 ML
(Figure 3). At low styrene coverage, for example, 0.01 ML,
There is an overall deficiency of oxygen when saturation
coverage of styrene (0.28 ML) is present. All oxygen was
consumed during temperature-programmed reaction, since
no O was detected at about 530 K, the temperature charac-
2
teristic of O recombination. Furthermore, carbon is deposit-
ed during temperature-programmed reaction and is quanti-
fied by measuring the CO yield after exposure of the sur-
2
face to ozone and subsequent heating (data not shown). The
amount of CO2 evolved during post-oxidation was about
5
times larger than that produced during the initial tempera-
ture-programmed reaction shown in Figure 1 for these same
conditions.
Dependence of the Product Yield and Selectivity on
Reactant Coverage
Figure 3. Yields (solid lines) of partial oxidation products as a function of
styrene coverage on O/Au
and benzoic acid (triangles). The same yields relative to the total amount
of CO , including post-oxidation, are shown as dashed lines. The vertical
O
ACHTUNGTNERNUNG( 111) (q =0.2 ML): styrene oxide (squares)
The product distribution and the conversion of styrene
strongly depend on the coverage of oxygen and of styrene
itself. The maximum amount of styrene oxide is produced
for 0.2 ML of oxygen and for saturation coverage of styrene
of 0.28 ML (Figure 2). The relative amount of residual
carbon was measured by post-oxidation to produce CO2.
The yield of styrene oxide was highest at 0.2 ML of oxygen,
among the three oxygen coverages investigated (0.1, 0.2,
and 1.0 ML). The highest oxygen coverage, 1.0 ML, induced
2
axis scales are identical to those in Figure 2.
the oxygen coverage is high compared to the amount of sty-
rene that reacts, resulting in 100% conversion into CO2,
H O, and unreacted oxygen (detected as O at ca. 530 K).
2
2
At higher styrene doses, partial oxidation products are ob-
served. At intermediate styrene coverages, for example,
0
.07 ML, the conversion is about 75%, the yield of styrene
oxide was at maximum, and no residual O was detected.
These results indicate that after addition of O to styrene,
ring-closure and secondary oxidation compete, depending
upon the amount of excess O.
The competition between ring-closure and secondary oxi-
dation results in an initial increase in the total yield for CO2
up to 0.07 ML of styrene, followed by a decrease in combus-
tion up to 0.28 ML of styrene. The ratio of styrene oxide to
the total CO produced is maximized for styrene coverages
2
at or near saturation (0.28 ML, Figure 3).
Increasing the styrene coverage also results in more resid-
ual C-containing species on the surface, which is analogous
to the observations by Klust et al. for styrene oxidation on
Ag ACHUTNRGENNUG( 111)-p ACHTNUGTRENNUN(G 4ꢁ4)-O. The same reaction features were ob-
served for an oxygen coverage of 1.0 ML: decreasing the
coverage of styrene increased the yields of water and CO2,
Figure 2. Yields (solid lines) of partial oxidation products as a function of
O coverage for the saturation styrene coverage of 0.28 ML: styrene oxide
(
squares) and benzoic acid (triangles). Dash lines show the ratio of the
[16]
same yields to the total amount of CO , including post-oxidation. Mass
2
spectrometric factors were corrected for each product in the calculation.
The full scales of each vertical axis are identical to those in Figure 3.
Chem. Asian J. 2010, 5, 78 – 86
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
81

C. M. Friend et al.
FULL PAPERS
and decreased the amount of residual carbon (data not
shown). This is largely attributed to the excess oxygen pres-
ent in sufficient quantities to oxidize carbon-containing by-
products of nonselective styrene oxidation.
phenylacetaldehyde, based on a detailed analysis of the frag-
mentation pattern. The formation of these products and
their peak temperatures closely match those for styrene oxi-
dation, clearly indicating that phenylacetaldehyde reaction
proceeds via the pathway involving the combustion inter-
mediate (Scheme 2) that, in the presence of excess oxygen
atoms, disproportionately converts to phenylketene and ben-
zeneacetic acid. This mechanism is further supported by the
ratio of benzeneacetic acid to phenylketene (1.3:1), which is
in good accord with the ratio obtained in the oxidation of
styrene (ca. 1.1:1).
Oxidation of Phenylacetaldehyde
As a further probe of secondary reactions and of possible
reverse of the 1,2-H shift, we also investigated phenylacetal-
dehyde reactions. Reaction of phenylacetaldehyde on O/Au-
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(111) (q =0.2 ML) yields phenylketene (435 K), benzoic
O
acid (380 K), benzeneacetic acid (430 K), CO , and H O
The absence of styrene oxide and acetophenone forma-
tion is further evidence that they are formed by reaction of
the branched oxametallacycle (1), whereas the combustion
intermediate (Scheme 2) originates from the linear oxame-
tallacycle (2).
2
2
(
Figure 4), consistent with Schemes 1 and 2. Importantly, no
Oxidation of Styrene Oxide
Ring-opening of styrene oxide, the microscopic reverse of
styrene epoxidation, leads to styrene production at 410 K,
along with phenylketene and phenylacetaldehyde (440–
4
2
60 K) on clean Au
82 K competes with reaction. The formation of styrene
ACHTUNGTRENNUNG( 111) (Figure 5). Molecular desorption at
shows that to some extent, the epoxidation process is rever-
sible.
The coincident evolution of phenylketene and phenylace-
taldehyde supports our proposed disproportionation path-
way that is analogous to that shown in Scheme 2, the differ-
ence being that the reaction stops at phenylacetaldehyde in-
stead of benzeneacetic acid on this oxygen-lean surface. It is
conceivable that when excess oxygen becomes available, as
Figure 4. Temperature-programmed reaction spectra for phenylacetalde-
hyde condensed on 0.2 ML O-covered Au(111). The oxygen-covered sur-
AHCTUNGTRENNUNG
face was prepared by exposure to ozone at 200 K.
Scheme 2. Conversion of the linear oxametallacycle (2) to the combus-
tion intermediate, analogous to Ag
acetic acid and phenylketene.
ACHTUNGTRENUNNG( 111), which then converts to benzene-
styrene oxide or acetophenone is produced from phenylace-
taldehyde reaction, so that aldehyde formation is not rever-
sible nor does isomerization to acetophenone readily occur.
On Ag
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(110), phenylacetaldehyde oxidation proceeds via the
Figure 5. Temperature-programmed reaction of styrene oxide on the
clean Au ACHTUNGTRNENUNG( 111) surface. Styrene oxide was dosed onto the surface at
same intermediate associated with styrene oxidation, yield-
2
40 K. In addition to styrene, phenylacetaldehyde and phenylketene are
[25]
ing all of the same products and reaction temperatures.
Similarly, they result in the same secondary oxidation reac-
also observed at higher temperature (440–460 K), indicating the forma-
tion of an oxametallacycle intermediate through ring-opening. The cross-
hatched area for m/z 104 indicates the contribution from styrene oxide.
[17]
tions. The small signal for m/z=120 is due to unreacted
82
www.chemasianj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 78 – 86

Mechanistic Insights into Selectivity Control for Heterogeneous Olefin Oxidation
in the case of styrene reacting with atomic oxygen created
by exposure to ozone, phenylacetaldehyde could be further
oxidized to benzeneacetic acid, or combustion products as
Changes in the C-H stretch region indicate that reaction
of styrene occurs below 240 K (Figure 6c), the temperature
just prior to evolution of styrene oxide (Figure 1). After
heating to 240 K, there are two predominant C-H stretch
[7]
proposed by Barteau et al. The concomitant evolution of
CO at 455 K appears to be consistent with this postulate.
Furthermore, coincident formation of CO , CO, and H O
À1
modes at 3016 and 2990 cm (Figure 6d). The feature at
À1
3016 cm is assigned to be associated with CÀH bonds on
2
2
was observed at 327 K; the source of active oxygen can be
attributed to that released from the deoxygenation of sty-
rene oxide to form styrene at 405 K. These observations
provide strong evidence that styrene oxide ring-opens pref-
erentially at the carbon atom that is bonded to the phenyl
the ring, supported by the observation of this mode for sty-
rene on gold (Figure 6a). The intensity for the peak at
À1
À1
2990 cm increased relative to that at 3016 cm , indicating
changes to the pendant vinyl group due to the interaction
with atomic oxygen on the surface. Furthermore, the simpli-
fication of the spectrum as compared to that for styrene sug-
gests an increase in the symmetry, probably corresponding
to the structure with the phenyl ring nearly parallel to the
surface, consistent with the structures of oxametallacycles as
suggested by recent DFT studies of styrene oxidation on
[20]
ring to form the linear oxametallacycle (2), which is also
one of the intermediate species in the oxidation of styrene.
Infrared Reflection Absorption Spectroscopy
[21]
Infrared spectroscopy was used to probe for stable inter-
mediates present during styrene oxidation. The molecular
complexity of these species and the transition intensities
render identification of species challenging. Previous high-
resolution electron energy loss spectra showed that the
Au AHCTUNGTRENNUN(G 111).
Comparison of the CÀH stretch region for styrene oxide
and the intermediates derived from styrene oxidation is con-
sistent with both forming a common surface species upon
heating to 240 K. Vibrational peaks were observed at 3016,
À1 [31]
À1
phenyl ring is intact, based on peaks at 710 and 950 cm .
2990, and 2910 cm after condensing styrene oxide mono-
This frequency region is not accessible with our infrared
spectrometer; thus we focus on the C-H stretch region.
Four well-defined peaks were detected for multilayers of
layer onto flat, clean Au AHCTUNGTERUNNN(G 111) at 120 K (Figure 6b). The fre-
quency of the mode at 3016 cm indicates that it is associat-
À1
ed with CÀH bonds in the phenyl ring. The peaks at 2990
À1
styrene adsorbed on clean Au
A
H
U
G
R
N
U
G
and 2910 cm can be assigned as stretch of CÀH bonds on
À1
À1
3
016 cm with a shoulder at 2990 cm (Figure 6a). The var-
the epoxide ring, based on comparison with the work on
À1
[32,33]
À1
ious peaks above 3000 cm most certainly arise from CÀH
bonds on the phenyl ring. The detection of several modes in-
dicates a relatively low adsorption symmetry, possibly due
ethylene oxide.
The mode at 2990 cm coincides with
(111) to 240 K,
that developed after heating styrene on O/AuACHTUNGTRENNUNG
indicative of the attack of the vinyl group by surface oxygen
to more than one adsorption geometry. In addition, a peak
to form a CÀO bond in Figure 6d. This peak decreased in
À1
was observed at 1450 cm , which is attributed to the (C-H )
magnitude and shifted to slightly higher frequency with fur-
ther annealing (Figure 6e,f), suggesting the transformation
of the intermediate species formed at 240 K. The feature at
2
bend. Adsorption of styrene onto O-covered Au
A
H
U
G
R
N
N
(111) has
À1
only two well-defined peaks at 3016 and 2990 cm and only
À1
À1
a broad peak at ca. 1500 cm , but no detectable peak at
2910 cm was detected again after annealing to 400 K, prior
À1
1
450 cm (Figure 6c).
to the desorption of phenylketene and benzeneacetic acid,
which could be associated with C-H stretch for the combus-
tion intermediate. The spectrum after heating the surface to
4
40 K shows no detectable features (Figure 6 g).
A small peak associated with the ring (C-O-C) stretch
À1
was observed at 1248 cm for condensed styrene oxide
(
Figure 6b) and assigned based on comparison to both gas
[34]
and liquid phase styrene oxide. An analogous peak assign-
ment was made for ethylene oxide on Ag(111) and support-
Although there are common vibra-
AHCTUNGTRENNUNG
[32]
ed by DFT studies.
tional features with styrene oxide and following the oxida-
tion of styrene, we do not detect the vibrational mode in the
region expected for the (C-O-C) ring stretch, which indi-
cates the absence of the ring for styrene reacting over
O/Au ACHUTNGRNEGUN( 111). This observation suggests that ring closure of
the oxametallacycles is the rate-limiting step in the produc-
tion of styrene oxide, in good agreement with DFT re-
sults.
Infrared spectra provide clear evidence for the formation
of the benzoate intermediate, which is present between 240
and 400 K when 0.07 ML of styrene reacts with 0.2 ML of O
Figure 6. Infrared spectra (IRAS) for a) 0.28 ML of styrene condensed
on Au(111) (bottom trace), b) a single layer of styrene oxide condensed
on Au(111) at 120 K, and 0.07 ML of styrene condensed on O/Au(111)
c) at 120 K followed by annealing to d) 240 K, e) 330 K, f) 400 K, and
g) 440 K. The temperatures shown indicate those that the surface was
heated to before cooling to 120 K to acquire the spectrum. The baselines
have been corrected.
[21]
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
ACHTUNGTRENNUNG
on Au ACHUTNGRNEUNG( 111) (Figure 6d,e, Table 2). Benzoate is identified
Chem. Asian J. 2010, 5, 78 – 86
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
83

C. M. Friend et al.
FULL PAPERS
Table 2. Vibrational frequencies and assignments associated with the in-
frared spectra for the oxidation of 0.07 ML of styrene on O/Au AHCUTNGRNENUG( 111).
volved in the formation of the combustion intermediate. It
can then combust to CO and H O, further oxidize to benze-
2
2
Vibrational frequency Assignment
Refs.
neacetic acid, or undergo CÀC bond scission to form CO2
À1
[
cm
]
and benzoate. Upon further heating to above 500 K, ben-
1
20 K 240 K 330 K
[17]
zoate decomposes to form CO and benzene. It is note-
2
1
1
350
375
1350
1375
phenyl C-H stretch (benzoate)
symmetric O-C-O stretch (benzoate) [35, 36]
[16]
worthy that the population of the combustion intermediate
is highly structure-sensitive; it is more abundant on Ag
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(110)
2
3
990
016
2990
3016
C-H stretch (oxametallacycle)
phenyl ring C-H stretch (styrene,
oxametallacycle, benzoate)
[20]
[8, 37]
[38]
3016
than on Ag AHCTUNTGERNNUN(G 111).
These studies indicate its abundance
may strongly depend on the local bonding and intermolecu-
lar interactions, which is consistent with our observation on
Au ACHUTNGRNENGU( 111) that its derivative products of phenylketene and
benzeneacetic acid only have significant yields within a
narrow range of reactant coverages.
À1
based on two characteristic modes at 1350 cm
1
and
À1
375 cm , corresponding to the phenyl C-H bend and the
symmetric O-C-O stretch, respectively. These modes were
first detected after annealing to 240 K, increase in intensity
up to 330 K, and disappear completely upon heating to
Our results collectively suggest a reaction mechanism that
involves two primary intermediates (Scheme 1). It is possi-
ble that styrene oxide forms via the pathway involving
either oxametallacycle, and this is consistent with the forma-
tion of styrene oxide in all experiments. Likewise, benzoic
acid forms nonselectively and can stem from either reaction
pathway or independent pathways. Its formation can at least
partly be attributed to a benzoate intermediate that was
identified using infrared spectroscopy. Acetophenone forma-
tion was observed under similar conditions of high styrene
concentration and low oxygen coverage that are required
for phenylketene and benzeneacetic acid formation, but its
integrated intensity was independent of other partial oxida-
tion products, suggesting its formation solely stems from the
branched oxametallacycle (1).
4
00 K (Figure 6 f). As in the study of benzoic acid adsorp-
[35]
tion on Au ACHTUNGTRENNUNG( 111) electrodes in acidic media and investiga-
tion of benzoate orientation and stability on Au
[
36]
A
H
U
G
E
N
N
(111), we
À1
attribute the mode at 1375 cm to the formation of a ben-
zoate surface intermediate. The formation and subsequent
disappearance of modes related to benzoate at 400 K is con-
sistent with the evolution temperature of benzoic acid
during temperature-programmed reaction (Figure 1), which
peaks at 360 K.
Mechanistic Framework for Styrene Oxidation
The analysis of the product distribution at different oxygen
and styrene coverages provides new insights into the mecha-
nisms that govern styrene oxidation selectivity on O/Au-
Conclusions
ACHTUNGTRENNUNG( 111). First, the formation of benzeneacetic acid and phenyl-
ketene are linked since they evolve at essentially the same
temperature (Figure 1) and the ratio of the integrated inten-
sities of these two products is about 1:1 in all experiments
where they were observed, within experimental error. This
is strong evidence that they are derived from a common in-
termediate. A likely mechanism for their formation is that
oxametallacycle (2) converts to the combustion intermedi-
ate, which subsequently disproportionates to benzeneace-
tic acid and phenylketene with no net change in hydrogen
content (Scheme 2). It is possible that the linear oxametalla-
cycle (2) has a short surface lifetime due to the relatively
higher stability of the combustion intermediate incurred by
p conjugation. Furthermore, the hydrogen loss in the con-
version of the oxametallacycle to the combustion intermedi-
ate may contribute to the evolution of water at low temper-
ature, ~250 K. This peak qualitatively scales with the rela-
tive yields of the partial oxidation products.
This work establishes important information concerning re-
action pathways and intermediates involved in the partial
oxidation of styrene, and confirms multiple factors that con-
tribute to partial oxidation selectivity combining both tem-
perature-programmed reaction and vibrational spectroscopic
studies. In coverage-dependent temperature-programmed
reaction studies of styrene oxidation on oxygen-covered
[
17]
Au AHCTUNTGERNNUNG( 111), partial oxidation products observed include styrene
oxide, acetophenone, benzoic acid, phenylketene, and ben-
zeneacetic acid. The selectivity toward epoxidation strongly
depends on the surface coverages of styrene and oxygen.
The maximum ratio of styrene oxide to the total CO pro-
2
duced was obtained for 0.28 ML styrene on 0.2 ML O-cov-
ered Au ACHUTNGRENNUG( 111). Our results suggest reaction mechanisms in-
volving oxametallacycles that undergo ring-closure to form
epoxide or secondary oxidation, and a combustion inter-
mediate that is derived from the linear oxametallacycle and
leads to the production of phenylketene and benzeneacetic
acid. Moreover, phenylacetaldehyde oxidation and ring-
opening of styrene oxide were used to confirm the proposed
reaction pathway involving the combustion intermediate.
IRAS data supplements the temperature-programmed reac-
tion studies with spectroscopic evidence for the formation of
oxametallacycles and benzoate intermediate; the latter ulti-
mately reacts and evolves as benzoic acid. These results
Comparison with styrene oxidation on Ag surfaces pro-
vides a framework for the possible formation of the combus-
tion intermediate on Au ACHUTNGERNNUG( 111). In the case of Ag ACHTGNUTRENNUN(G 111), Zhou
et al. determined that the transformation of the oxametalla-
cycle (2) to the combustion intermediate is the kinetic step
[17]
that is critical in controlling the styrene oxide selectivity.
Those authors used isotopic labeling and X-ray photoelec-
tron spectroscopy (XPS) to infer that b-H elimination is in-
84
www.chemasianj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 78 – 86

Mechanistic Insights into Selectivity Control for Heterogeneous Olefin Oxidation
highlight the importance of reactant concentration on opti-
mizing the selectivity and provide further insight into the
mechanistic details of epoxidation reactions.
dividing by the total amount of styrene that reacted. The detailed proce-
[
8]
dure used in the selectivity calculations was previously described. The
amount of CO production in these calculations includes CO formed
2
2
from oxidation of residual C. Factors accounted for include variation in
ionization efficiency, abundances of parent ions, fragmentation of other
products, transmission gain, and electron multiplier gain. Relative errors
in the calculated product yields and selectivities were estimated based on
the sensitivity of the calculations to the multiplier gain, transmission
function, and the standard deviation in four experiments. Relative error
values varied between ca. 10% and ca. 35% depending on the product.
Experimental Section
All experiments were performed in an UHV chamber with a base pres-
À10
sure of less than 2ꢁ10 torr. The system was equipped with an IR spec-
Infrared reflection absorption spectroscopy (IRAS) data were acquired
using a Thermo Nicolet MCT-A semiconductor photodiode detector.
Background spectra were acquired at 120 K after first heating to 600 K
to evolve all products; the baselines of the resulting absorbances were
trometer, a mass spectrometer (Pfeiffer Prisma QMS 200) for tempera-
ture-programmed reaction spectroscopy (TPRS), a cylindrical mirror ana-
lyzer (Physical Electronics model 15–155) for Auger electron spectrosco-
py (AES), and low energy electron diffraction (LEED) optics (Physical
Electronics model 15–180).
corrected. Two sets of 600 scans were co-added for each experiment with
À1
a resolution of 4 cm
.
The Au ACHTUNGTRENNUNG( 111) single crystal (SPL, 8 mm diameter, 2 mm thick) was
mounted by wrapping a tantalum wire (0.75 mm diameter) around its di-
ameter. Temperature was monitored using a type-K, chromel/alumel ther-
mocouple which contacted the sample via a hole in the side of the crystal.
The crystal was radiatively heated using a tungsten filament mounted
about 2 mm behind the sample and conductively cooled using a liquid ni-
trogen reservoir. Temperature-programmed reaction experiments were
Acknowledgements
À1
performed with an average heating rate of ca. 4 Ks in the range of
The authors gratefully acknowledge support of this work by the U.S. De-
partment of Energy, Basic Energy Sciences, Catalysis Science Program
under grant FG02-84-ER13289. R.Q. also acknowledges support from
the National Science Foundation, Chemistry Division, under grant CHE-
0545335.
1
20 K and 600 K. The crystal was biased to À70 V during these experi-
ments to preclude electron-induced reactions. An established method of
cycles of Ar bombardment (1 keV energy, 2 mA current to sample,
+
2
6
0 min duration) and annealing at 900 K for 5 min followed by 700 K for
0 min were used to prepare the crystal.
[
39, 40]
This process was repeated
until satellite spots characteristic of the herringbone reconstructed
[
40,41]
Au
tected using AES.
The O/Au(111) surface was prepared by exposing the crystal to ozone at
00 K using a direct doser, following a method described previously.
The coverage of oxygen was determined by comparing the integrated in-
tensity of the m/z=32 (O ) peak at 530 K for a specific dosing time to
the integrated intensity at saturation oxygen coverage, which is
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(111) surface were obtained in LEED
and no impurities were de-
[
[
[
1] G. A. Trigaux, Modern Plastics 1960, 38, 147–148.
2] T. Hayashi, K. Tanaka, M. Haruta, J. Catal. 1998, 178, 566–575.
3] M.-L. Bocquet, A. Michaelides in Properties of Single Organic Mol-
ecules on Crystal Surfaces (Eds.: P. H. Gruetter, Werner; Rosei, Fed-
erico), Imperial College Press, London, UK, 2006, pp. 389–424.
4] J. Miao, Y.-J. Wang, Gongye Cuihua 2005, 13, 44–47.
ACHTUNGTRENNUNG
[
22,42]
2
2
[
[
5] J. G. Serafin, A. C. Liu, S. R. Seyedmonir, J. Mol. Catal. A 1998, 131,
[
22,42]
ca. 1 ML.
The oxygen coverage varies approximately linearly with
1
57–168.
dosing time. We then used the calibrated dosing time to estimate the ini-
tial oxygen coverage in our reactivity measurements. The integrated in-
tensity of m/z=32 on the clean Au ACHUTNGRENNUG( 111) surface associated with a given
dosing time was checked immediately before and after each series of ex-
periments. The dosing conditions are periodically recalibrated due to
changes in the ozone flux as it is depleted in the source, over a period of
[
6] R. M. Lambert, F. J. Williams, R. L. Cropley, A. Palermo, J. Mol.
Catal. A 2005, 228, 27–33.
7] S. Linic, M. A. Barteau, J. Am. Chem. Soc. 2003, 125, 4034–4035.
8] X. Deng, C. M. Friend, J. Am. Chem. Soc. 2005, 127, 17178–17179.
9] M. A. Barteau, R. J. Madix, Surf. Sci. 1981, 103, L171–L178.
[
[
[
[
[
10] D. Torres, F. Illas, J. Phys. Chem. B 2006, 110, 13310–13313.
11] D. Torres, N. Lopez, F. Illas, R. M. Lambert, J. Am. Chem. Soc.
1
–2 months.
Styrene (Alfa Aesar, 99.5%), styrene oxide (Alfa Aesar, 98%), and phe-
nylacetaldehyde (Alfa Aesar, 95%) were purified using freeze-pump-
thaw cycles. Purity was confirmed by subliming condensed multilayers
from the crystal while monitoring with the mass spectrometer. Exposures
2
005, 127, 10774–10775.
[
12] E. P. Hunter, S. G. Lias in “Proton Affinity Evaluation” in NIST
Chemistry WebBook (Eds: P. J. Linstrom, W. G. Mallard), National
Institute of Standards and Technology (http://webbook.nist.gov),
Gaithersburg MD, 2009.
13] F. J. Williams, D. P. C. Bird, A. Palermo, A. K. Santra, R. M. Lam-
bert, J. Am. Chem. Soc. 2004, 126, 8509–8514.
À9
were performed with a background pressure rise of 1ꢁ10 torr while the
sample temperature was held at 140 K, which is below the desorption
[
[
[
8]
temperature of molecular styrene, styrene oxide, and phenylacetalde-
hyde. The measured enhancement factor of the direct doser was about
14] S. Hawker, C. Mukoid, J. P. S. Badyal, R. M. Lambert, Surf. Sci.
1
0
30. Coverages are given in monolayers (ML), where for styrene,
.12 ML corresponds to the saturation of the broad styrene peak cen-
1
989, 219, L615–L622.
[
[
[
[
15] A. Klust, R. J. Madix, J. Am. Chem. Soc. 2006, 128, 1034–1035.
16] A. Klust, R. J. Madix, Surf. Sci. 2006, 600, 5025–5040.
17] L. Zhou, R. J. Madix, J. Phys. Chem. C 2008, 112, 4725–4734.
18] J. J. Cowell, A. K. Santra, R. Lindsay, R. M. Lambert, A. Baraldi, A.
Goldoni, Surf. Sci. 1999, 437, 1–8.
tered at ca. 280 K and the coverage just before the multilayer peak at
90 K begins to develop. We determined the absolute coverage of styrene
by reacting 0.1 times the amount of styrene needed to saturate the peak
at about 280 K on 0.2 ML O-covered Au(111), which results in 100%
conversion to CO and water with residual oxygen, but no residual
carbon on the surface. Since we know the carbon content of CO and sty-
a carbon balance al-
1
AHCTUNGTRENNUNG
2
[
[
[
19] A. K. Santra, J. J. Cowell, R. M. Lambert, Catal. Lett. 2000, 67, 87–
2
9
1.
[
22,42]
rene and the absolute amount of O that reacts,
20] M. Enever, S. Linic, K. Uffalussy, J. M. Vohs, M. A. Barteau, J.
Phys. Chem. B 2005, 109, 2227–2233.
21] L. Q. Xue, X. Y. Pang, G. C. Wang, J. Comput. Chem. 2009, 30, 438–
lowed us to determine that saturation of the ca. 280 K state corresponds
to 0.12 ML of styrene. Further, the coverage for saturation of both peaks
at 190 and 280 K was determined to be 0.28 ML.
4
46.
Relative yields for each reaction product were calculated using a method
[22] B. K. Min, A. R. Alemozafar, D. Pinnaduwage, X. Deng, C. M.
Friend, J. Phys. Chem. B 2006, 110, 19833–19838.
[
43]
previously described
based on quantitative analysis of mass spectro-
metric data. Selectivity was estimated by accounting for the number of
styrene molecules required to yield one molecule of a given product, and
[23] M. D. Hughes, Y.-J. Xu, P. Jenkins, P. McMorn, P. Landon, D. I.
Enache, A. F. Carley, G. A. Attard, G. J. Hutchings, F. King, E. H.
Chem. Asian J. 2010, 5, 78 – 86
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
85

C. M. Friend et al.
FULL PAPERS
Stitt, P. Johnston, K. Griffin, C. J. Kiely, Nature 2005, 437, 1132–
[34] D. Lin-Vien, N. B. Colthup, W. G. Fateley, J. G. Grasselli, The Hand-
book of Infrared and Raman Characteristic Frequencies of Organic
Molecules, Academic Press, San Diego CA, 1991.
1
135.
[
[
24] M. Turner, V. B. Golovko, O. P. H. Vaughan, P. Abdulkin, A. Beren-
guer-Murcia, M. S. Tikhov, B. F. G. Johnson, R. M. Lambert, Nature
[35] Y. Ikezawa, R. Sekiguchi, T. Kitazume, Electrochim. Acta 2001, 46,
731–736.
[36] H.-Q. Li, S. G. Roscoe, J. Lipkowski, J. Electroanal. Chem. 1999,
2
008, 454, 981–984.
25] X. Y. Liu, A. Klust, R. J. Madix, C. M. Friend, J. Phys. Chem. C
007, 111, 3675–3679.
4
78, 67–75.
2
[
37] M. K. Weldon, M. E. Napier, B. C. Wiegand, C. M. Friend, P. Uvdal,
J. Am. Chem. Soc. 1994, 116, 8328–8334.
[
[
26] R. G. Quiller, L. Benz, C. M. Friend, unpublished work.
27] B. D. Kay, K. R. Lykke, J. R. Creighton, S. J. Ward, J. Chem. Phys.
[
[
38] L. Zhou, R. J. Madix, Surf. Sci. 2009, 603, 1751–1755.
39] M. A. Van Hove, R. J. Koestner, P. C. Stair, J. P. Biberian, L. L. Kes-
model, I. Bartos, G. A. Somorjai, Surf. Sci. 1981, 103, 189–217.
40] B. K. Min, X. Deng, D. Pinnaduwage, R. Schalek, C. M. Friend,
Phys. Rev. B 2005, 72, 121410.
41] M. A. Van Hove, R. J. Koestner, P. C. Stair, J. P. Biberian, L. L. Kes-
model, I. Bartos, G. A. Somorjai, Surf. Sci. 1981, 103, 218–238.
42] N. Saliba, D. H. Parker, B. E. Koel, Surf. Sci. 1998, 410, 270–282.
43] R. G. Quiller, L. Benz, J. Haubrich, M. E. Colling, C. M. Friend, J.
Phys. Chem. C 2009, 113, 2063–2070.
1
989, 91, 5120–5121.
[
[
[
28] R. G. Quiller, T. A. Baker, X. Deng, M. E. Colling, B. K. Min, C. M.
Friend, J. Chem. Phys. 2008, 129, 064702.
[
[
29] J. Wang, M. R. Voss, H. Busse, B. E. Koel, J. Phys. Chem. B 1998,
1
02, 4693–4696.
30] M. L. Bocquet, A. Michaelides, D. Loffreda, P. Sautet, A. Alavi,
D. A. King, J. Am. Chem. Soc. 2003, 125, 5620–5621.
31] X. Y. Deng, Ph. D. Thesis 2006, pp. 33–35.
[
[
[
[
[
32] S. Linic, M. A. Barteau, J. Am. Chem. Soc. 2002, 124, 310–317.
33] G. S. Jones, M. Mavrikakis, M. A. Barteau, J. M. Vohs, J. Am. Chem.
Soc. 1998, 120, 3196–3204.
Received: August 28, 2009
Published online: December 14, 2009
86
www.chemasianj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 78 – 86