Benzyl alcohol oxidation on carbon-supported Pd nanoparticles: Elucidating the reaction mechanism

Article abstract of DOI:10.1002/cctc.201402552

Experiments were conducted on the liquid-phase oxidation of benzyl alcohol over Pd nanoparticles, with the aim of determining the operative chemical reaction. Experiments were conducted in a batch reactor with para-xylene as the solvent and continuous gas purging of the headspace. The following experimental parameters were varied: the initial benzyl alcohol concentration, the oxygen partial pressure in the headspace, and the reactor temperature. From trends in the concentration profiles and integrated production of each product, it was determined that there are two primary reaction paths: A)an alkoxy pathway leading to toluene, benzaldehyde, and benzyl ether, and B)a carbonyloxyl pathway ("neutral carboxylate") leading to benzoic acid, benzene, and benzyl benzoate. From the mechanism elucidated, it is clear that the coverages of atomic hydrogen, atomic oxygen, and surface hydroxyls must be accounted for to achieve a complete description of the quantitative kinetics. Two key intermediates are involved in the catalytic liquid-phase oxidation of benzyl alcohol. An alkoxy intermediate is proposed for the route to benzyl aldehyde, toluene, and benzyl ether. A carbonyloxyl intermediate is proposed for the route to benzoic acid, benzene, and benzyl benzoate.

Full text of DOI:10.1002/cctc.201402552

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201402552
Benzyl Alcohol Oxidation on Carbon-Supported Pd
Nanoparticles: Elucidating the Reaction Mechanism
[a]
[b]
[b]
[b]
[b]
Aditya Savara,* Carine E. Chan-Thaw, Ilenia Rossetti, Alberto Villa, and Laura Prati
Experiments were conducted on the liquid-phase oxidation of
benzyl alcohol over Pd nanoparticles, with the aim of deter-
mining the operative chemical reaction. Experiments were con-
ducted in a batch reactor with para-xylene as the solvent and
continuous gas purging of the headspace. The following ex-
perimental parameters were varied: the initial benzyl alcohol
concentration, the oxygen partial pressure in the headspace,
and the reactor temperature. From trends in the concentration
profiles and integrated production of each product, it was de-
termined that there are two primary reaction paths: A) an
alkoxy pathway leading to toluene, benzaldehyde, and benzyl
ether, and B) a carbonyloxyl pathway (“neutral carboxylate”)
leading to benzoic acid, benzene, and benzyl benzoate. From
the mechanism elucidated, it is clear that the coverages of
atomic hydrogen, atomic oxygen, and surface hydroxyls must
be accounted for to achieve a complete description of the
quantitative kinetics.
Introduction
[
23–27]
The liquid-phase oxidation of alcohols over supported metal
catalysts, using molecular oxygen as the oxidant, has been ex-
molecules on transition metal surfaces,
with elucidation of
the mechanisms of reactions at the molecular level. Ultrahigh
vacuum studies with small alcohols and aldehydes—less than
[
1–4]
tensively studied in the last decade.
Among the alcohols,
[2–15]
[28–42]
benzyl alcohol is one of the most studied substrates.
De-
four carbons—on Pd and other transition metal surfaces
pending on the reaction conditions (temperature, solvent,
oxygen pressure), many side products including benzene, tolu-
ene, benzoic acid, benzyl benzoate, and benzyl ether have
been reported to be formed in addition to the main product,
have provided insight into the mechanisms of reactions with
[27]
organic oxygenates on transition metal surfaces. Using the
data acquired in this study alongside the information from sur-
face science studies, we elucidated the mechanism of benzyl
alcohol oxidation over Pd nanoparticles supported on activat-
ed carbon.
[
2,6,16,17]
benzaldehyde.
Tentative mechanisms were pro-
including two mechanisms for the formation of
[
11,15,16,18]
posed,
toluene: either by “disproportionation” of benzyl alcohol
or via a geminal diol intermediate formed by reaction with OH
[19–21]
Results and Discussion
[
4,22]
or H O.
However, a detailed and complete mechanism of
2
Pd-catalyzed benzyl alcohol oxidation in organic solvent has
not yet been published, particularly at a level that would
enable microkinetic modeling.
Previous experiments were conducted with cyclohexane as
a solvent, with the reacting liquid in contact with an oxygen
[16]
reservoir. However, in this temperature range the vapor pres-
sures of cyclohexane and several products are small but not in-
significant. The liquid phase for these conditions is on the
order of 1000 times more dense than the gas phase: thus if
even 1% of the liquid phase evaporates in a static batch reac-
tor, such evaporation can “push away” a significant quantity of
oxygen in the absence of a well-mixed system. To keep the
oxygen concentration known and constant, we utilized contin-
uous gas flow through the headspace and switched to a lower
vapor pressure solvent with low reactivity (para-xylene). In this
configuration, some product molecules may still enter the
vapor, but are continuously purged from the system by the
flowing feed gas, preventing an oxygen-deficient stagnation
layer in the gas phase directly above the liquid. Under these
conditions, the gas-phase oxygen concentration is constant.
In Table 1 the studied conditions are listed, and the follow-
ing chemicals were monitored by gas chromatography: ben-
zene, toluene, benzaldehyde, benzyl alcohol, benzoic acid,
benzyl ether, and benzyl benzoate (these products are all
In this study, we have performed experiments in which the
temperature, gas-phase oxygen pressure, and initial concentra-
tion of the benzyl alcohol were varied to elucidate the mecha-
nism. Key insights were also gained from surface science litera-
ture of reactions with alcohols and aldehydes over metal surfa-
ces. There is a long history of studying the reactions of organic
[a] Dr. A. Savara
Chemical Sciences Division
Oak Ridge National Laboratory
1
Bethel Valley Road MS 6201
Oak Ridge, TN 37831 (USA)
E-mail: savaraa@ornl.gov
[b] Dr. C. E. Chan-Thaw, Dr. I. Rossetti, Dr. A. Villa, Prof. L. Prati
Dipartimento di Chimica
Universitꢀ degli Studi di Milano
via Golgi 19
2
0133, Milano (Italy)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201402552.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&1
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
scaled kinetic profiles for each species monitored. In Figure 1,
a typical profile from experiment 2 is shown, and the plot in
Figure 1A.ii is scaled to resolve the profiles of the minor prod-
ucts. As can be seen in Figure 1A.i, there is little change in the
mass balance of the liquid over time, based upon the sum of
the aromatics detected. Under all of the conditions investigat-
ed in experiments, the product distribution followed the order:
benzaldehyde>toluene>minor products. The minor products
detected in the liquid phase consisted of benzoic acid, benzyl
ether, benzyl benzoate, and benzene. The ratios and the con-
centration profiles of the minor products varied as conditions
were changed.
[
a]
Table 1. Experimental conditions investigated.
Experiment
Initial alcohol concentration
vol%]
[O
[bar]
2
]
T
[8C]
[
1
2
3
4
5
6
7
8
9
25
50
75
25
25
25
25
25
25
25
1.0
1.0
1.0
0.0
0.25
0.50
1.0
1.0
1.0
1.0
70
70
70
70
70
70
70
80
90
1
0
100
At the end of experiment, the final concentration of a given
species represents the integrated production during experi-
ment, and is used with varying conditions to gain insight into
the reaction mechanism. Notably, as the final concentration is
an integrated production, we can convert these numbers to
selectivities over the experiment period. In Figure 2–4, it is
shown how the final concentra-
[
a] Experiment 7 is the same run as 1, and was also used as a data point
in the series with varied temperature.
shown in Scheme 1). The full set of experimental data is includ-
ed in the Supporting Information, along with scaled and un-
tions and selectivities vary as
a function of initial alcohol con-
centration, oxygen concentra-
tion, and temperature. These re-
sults will be analyzed and dis-
cussed below.
Alkoxy intermediate for tolu-
ene, benzaldehyde, and benzyl
ether
Alcohols can dissociate homolyt-
ically to produce an alkoxy inter-
mediate: this reaction is known
to occur on noble-metal surfa-
[43]
ces.
Previous surface-science
studies have shown that cleav-
age of an alcohol OH bond on
noble metals is surface-oxygen
[28–29,31,38]
assisted
(as has also
been shown for water OH bond
cleavage). We thus assume that
alkoxy formation by a surface-
oxygen assisted pathway—to
produce OH* and ArꢀCH ꢀO* on
2
the surface—is the first step.
However, a recent computational
study considered a surface hy-
droxyl to facilitate OH bond
cleavage for ethanol on Au(111)
[44]
and Pt(111),
and we cannot
exclude such a mechanism here,
so surface hydroxyl facilitated
OH cleavage will be considered
during microkinetic modeling.
Once an alkoxy is formed, al-
dehydes are often produced by
hydrogen abstraction from the
Scheme 1. Proposed mechanism for benzyl alcohol oxidation.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
2
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
sorbed directly on the surface, consistent with other stud-
[25,47,48]
[49]
ies.
Temperature-programmed desorption
(TPD) of
benzyl alcohol from neat Pd(111) produced toluene but TPD
[50]
of benzaldehyde from neat Pd(111) did not produce tolu-
ene, indicating that the branching occurred prior to benzalde-
[49,50]
hyde formation, consistent with our interpretation. TPD
of
benzyl alcohol and benzaldehyde from oxygen-precovered
[50]
Pd(111) and vibrational spectroscopy monitoring of the sur-
face species are also consistent with our mechanistic assump-
tion that an alkoxy intermediate is involved in benzaldehyde
formation.
The ether molecule could be formed from combination of
an alkyl group and an alkoxy group, or from two alkoxy
groups with an oxygen left behind. In Figure 2–4, the ether
production exhibits a similar trend to that of the toluene pro-
duction as a function of initial concentrations and temperature
(
unlike benzene, benzoic acid, and benzyl benzoate). If toluene
production is proportional to alkyl formation, the correlated
production suggests that the ether is formed from a combina-
tion of one alkoxy group with one alkyl group. A previous
study proved that on Ag(110) diethyl ether was formed from
[41]
a combination of one alkoxy group and one alkyl group, but
a surface oxygen was likely also involved. At present, we will
assume that a surface oxygen is also involved on palladium,
though confirmation would require additional study, and both
options should be considered in future works. An alternative
possibility provided by one of the reviewers of this article is
that the ether may be formed by nucleophilic attack of free
benzyl alcohol on an adsorbed species.
Figure 1. Typical kinetic profiles of A.i) major products and A.ii) minor prod-
ucts, shown for the conditions of experiment 2.
We can write the following equations based on the above
analysis, whereby [site] represents a free active site such as
a palladium atom with no adsorbate:
[34,40]
alkoxy intermediate on Pd and other noble metal surfaces.
The literature suggests that hydrogen abstraction from alkoxy
groups is not oxygen assisted and occurs with a Pd site ab-
ArꢀCH ꢀOH ðlÞ þ OðadsÞ !
2
[
29,35,37,42]
ð1Þ
stracting the hydrogen directly.
Additionally, a surface-
ArꢀCH ꢀO* ðadsÞ þ HO* ðadsÞ
2
science study found that surface oxygen blocked abstraction
of hydrogens from methoxy groups, again suggesting that
ArꢀCH
ꢀO*
ðadsÞ þ ½siteꢁ ! ArꢀCH¼O ðlÞ þ H* ðadsÞ
ð2Þ
ð3Þ
ð4Þ
2
[
40]
empty palladium sites are required.
ArꢀCH ꢀO* ðadsÞ þ ½siteꢁ ! ArꢀCH * ðadsÞ þ O ðadsÞ
2
2
Thus, we assume that the aldehyde comes from the alkoxy
intermediate and an adjacent open Pd site that abstracts a hy-
drogen directly, although it is also possible that a surface hy-
ArꢀCH * ðadsÞ þ H* ðadsÞ ! ArꢀCH ðlÞ
2
3
[
44,45]
droxyl or surface oxygen abstracts the hydrogen.
Next, we
ArꢀCH ꢀO* ðadsÞ þ ArꢀCH * ðadsÞ þ O ðadsÞ !
2
2
ð5Þ
note that the kinetic profiles of the aldehyde and toluene track
with a strong correlation for many of the experiments (see the
scaled plots of experiments 2, 3, 4, 6, 8, and 9 in the Support-
ing Information). From this observation, we infer that the tolu-
ene is produced from an alkyl group that originates from the
same type of alkoxy intermediate as the aldehyde, but only as
a minority product. This interpretation is consistent with litera-
ture data that methyl groups can be produced from methoxy
groups on palladium surfaces (analogous to the alkyl inter-
mediate, which leads to toluene), and almost never as more
ArꢀCH ꢀOꢀCH ꢀAr ðlÞ þ O ðadsÞ
2
2
These reactions are detailed in the upper half of Scheme 1,
and largely correspond to those of a recent publication on eth-
[51]
anol oxidation on Pd(111). The pathways leading to benzyl
aldehyde, toluene, and benzyl ether will be referred to as the
“alkoxy route”. We note that our proposed mechanism is con-
sistent with the data that led to the hypothesis of a “dispropor-
[19–21]
tionation” mechanism.
In our proposed mechanism, tolu-
[
30,32,35,36,39]
than a minority product.
DFT computations also re-
ene production is accompanied by formation of an additional
surface oxygen and aldehyde production is accompanied by
consumption of a surface oxygen (and formation of water pre-
cursors). These reactions result in the same stoichiometry as
a “disproportionation” reaction:
vealed that the activation energy for direct dissociation to pro-
duce an alkyl group is much higher than the barrier to pro-
[
46]
duce an alkoxy one. We assume that CꢀH bond formation
with the alkyl species occurs by addition of a hydrogen ad-
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
3
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 2. A.i, A.ii) Final concentrations and B.i, B.ii) selectivities as a function of the initial alcohol concentration.
2
benzyl alcohol ! toluene þ benzyl aldehyde þ H O
ð6Þ
[53,54]
2
metal oxide surfaces.
In contrast, carbonyloxyl and hydro-
gen atoms (RCOO* and H*) are neutral species that are expect-
ed to be formed on palladium and other metal surfaces, in ac-
However, in Scheme 1, an alkyl intermediate is proposed to
[55]
be involved in toluene formation, which is consistent with our
data and mechanism for ether formation. Further study to re-
solve this issue is merited.
cordance with DFT. Typically, transition-metal surfaces cleave
[56–58]
CH bonds and H homolytically during adsorption,
in ac-
2
[59]
cordance with the Newns–Anderson model. In line with the
above distinctions, we will refer to ArꢀCOO*(ads) as a carbony-
loxyl species, though most of the literature cited uses words
such as “carboxylate”, “formate”, “acetate,” etc. when describ-
ing these types of carbonyloxyl adsorbates on metals.
Carbonyloxyl intermediate for benzene, benzoic acid, and
benzyl benzoate
Herein, we make a nomenclature distinction between a carbon-
yloxyl species and a carboxylate. This distinction is often ne-
glected in the literature and can lead to confusion. The impor-
tance of this distinction can be demonstrated by comparing
heterolytic with homolytic cleavage of an organic acid:
We started with the assumption that benzoic acid is formed
from a dioxy or carbonyloxyl intermediate. Next, we noted that
benzoic acid, benzene, and benzyl benzoate display similar ki-
netic profiles, which track during many experiments (see the
scaled plots of experiments 1, 2, 3, 6, 8, 9, and 10 in the Sup-
porting Information). Moreover, the analyses in Figure 3–6 re-
vealed that the integrated productions for benzene, benzoic
acid, and benzyl benzoate follow similar trends if varying the
initial reactant concentrations and the temperature. These sim-
ilar trends and kinetic profiles suggest a common intermediate,
and in our mechanism we ascribe the role of a common inter-
mediate to a carbonyloxyl: if the carbonyloxyl carbon of the
ꢀ
þ
ROOH ! ROO þ H
ð7Þ
ð8Þ
ROOH ! ROO* þ H*
The heterolytic cleavage produces
a
carboxylate and
a proton—both are ionic species. In contrast, the homolytic
[
52]
cleavage produces a carbonyloxyl radical and a hydrogen
radical. This distinction should be recognized if discussing the
carbonyloxyl breaks off to form CO , then benzene is the co-
2
product. CO has been observed as a coproduct during benzyl
2
ꢀ
[10]
chemistry on surfaces. Carboxylates and protons (RCOO and
alcohol oxidation, and as a product alongside benzene pro-
+
H
are ionic species) are expected to be formed on many
duction during TPD of benzyl alcohol on oxygen-precovered
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
4
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 3. A.i, A.ii) Final concentrations and B.i, B.ii) selectivities as a function of the oxygen partial pressure in the gas flow.
[
60]
Pd(111). In general, reactions in which carbonyloxyls decom-
ArꢀCOO* ðadsÞ þ H* ðadsÞ ! ArꢀCOOH ðlÞ
ArꢀCOO* ðadsÞ þ H* ðadsÞ ! ArꢀH ðlÞ þ CO ðgÞ
ð11Þ
ð12Þ
pose to form CO and a hydrocarbon are known to occur over
2
[
61–62]
2
metals.
Experiments with ethanol are also consistent with
our interpretation: In Table 1 in ref. [38] it is shown that CO2
and ethanoic acid are produced after exposure of ethanol on
palladium at a common temperature during TPD, and the con-
ArꢀCOO* ðadsÞ þ ArꢀCH ꢀO* ðadsÞ !
2
ð13Þ
ð14Þ
ArꢀCOOꢀCꢀAr ðlÞ þ O ðadsÞ
ArꢀCOO* ðadsÞ þ ArꢀCH * ðadsÞ ! ArꢀCOOꢀCꢀAr ðlÞ
comitant production of CO and ethanoic acid was shown to
2
2
[
31]
occur from surface carbonyloxyls. From experimental data, it
was determined that the route to form the acid product re-
These reactions are detailed in the lower half of Scheme 1.
The pathways leading to benzene, benzoic acid, and benzyl
benzoate we refer to as the “carbonyloxyl/dioxy route”. We
write the conversion of an alkoxy to a carbonyloxyl as involv-
ing hydrogen abstraction by empty sites based upon literature
evidence mentioned above, though it is also possible that
a surface hydroxyl abstracts one or both of the hydrogen
[31]
quires a surface hydrogen to react with a carbonyloxyl. The
benzene product also requires a hydrogen source based on
stoichiometry, and the similar concentration dependences sug-
gest that the same hydrogen source is likely utilized, rather
than the acid that is formed directly by reaction of an alkoxy
[
44]
group with a hydroxyl group. Additionally, the formation of
benzyl benzoate from a carbonyloxyl species is a reasonable
working hypothesis, though it is unclear whether the carbony-
loxyl reacts with an alkoxy group or an alkyl species (a ques-
tion that microkinetic modeling may help with). The above
analysis allows us to write the following equations:
[44]
atoms involved. We also note that an aldehyde decarbonyla-
tion route has previously been proposed for aliphatic alde-
[33]
hydes on Pd(111), —the corresponding reaction in our
system would be benzaldehyde conversion to benzene, the
evidence or which we do not see. Perhaps steric hindrance or
site blocking prevents aldehyde decarbonylation in our experi-
ments.
ArꢀCH ꢀO* ðadsÞ þ O ðadsÞ þ ½siteꢁ !
2
ð9Þ
In several individual experiments (2, 3, 9, and 10), aldehyde
exhibits a long-term production that deviates from the long-
term toluene production profile and resembles the carbonylox-
yl/dioxy-route long-term production profile, so there may be
ArꢀCðHÞOO* ðadsÞ þ H* ðadsÞ
ArꢀCðHÞOO* ðadsÞ þ ½siteꢁ ! ArꢀCOO* ðadsÞ þ H* ðadsÞ ð10Þ
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
5
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 4. A.i, A.ii) Final concentrations and B.i, B.ii) selectivities as a function of the reactor temperature for each experiment.
H ðgÞ þ 2½siteꢁ ! 2 H* ðadsÞ
ð15Þ
ð16Þ
ð17Þ
ð18Þ
ð19Þ
ð20Þ
ð21Þ
ð22Þ
ð23Þ
ð24Þ
2
O ðgÞ þ 2½siteꢁ ! 2 O ðadsÞ
2
H O ðlÞ ! HO* ðadsÞ þ H* ðadsÞ
2
H O ðlÞ þ O ðadsÞ ! HO* ðadsÞ þ HO* ðadsÞ
2
HO* ðadsÞ ! H* ðadsÞ þ O ðadsÞ
2
H* ðadsÞ ! H ðgÞ þ 2½siteꢁ
2
2
O ðadsÞ ! O ðgÞ þ 2½siteꢁ
2
HO* ðadsÞ þ H* ðadsÞ ! H O ðlÞ
2
HO* ðadsÞ þ HO* ðadsÞ ! H O ðlÞ þ O ðadsÞ
2
H* ðadsÞ þ O ðadsÞ ! HO* ðadsÞ
Figure 5. Ratio of the carbonyloxyl versus the alkoxy route as a function of
pressure, as determined based on final concentrations.
O
2
Lack of product readsorption
a minor route of carbonyloxyls converting back to aldehyde
precursors, or the existence of multiple site types.
From the above considerations, we consider predicted and ob-
served trends for the proposed reaction mechanism. With the
exception of the starting alcohol, none of our observed species
decreased across time, and no products that we observed
were depleted. This suggests that none of the other species
can displace the alkoxy or oxygen species on the surface,
which effectively prevents dissociative readsorption of the
other products. In the context of our proposed mechanism,
Hydrogen, oxygen, and water reactions
The following reactions are known to occur, and it is assumed
that some gas-phase molecules dissolve in the liquid phase
prior to adsorption or after desorption.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
6
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
as shown in Figure 5. 3) High oxygen pressures suppress for-
mation of toluene and the ether, both of which require empty
sites large enough for alkyl species, and both of which pro-
duce surface oxygen rather than consuming it.
In addition to promotion of alkoxide formation and site
[29,38,40]
blocking to prevent alkoxide dissociation,
oxygen plays
an important role by scavenging hydrogen atoms from the
[60,61]
surface to form OH*(ads) and/or H O(l).
The consequence
2
of this scavenging is that the selectivity of benzaldehyde
versus toluene and benzoic acid versus benzoate should both
decrease as a function of oxygen pressure, as both of those
products require surface hydrogen as an intermediary reactant.
There is literature evidence for acid formation suppression by
[31]
oxygen, and the expected trends are observed in Figure 6A,
B. One topic of interest in the literature is why addition of Au
[63,64]
to Pd nanoparticles increases selectivity to aldehydes:
sta-
bilities of the carbonyloxyl and alkoxy may be an important
factor.
The mechanism in Scheme 1 also explains the anaerobic oxi-
dation of benzyl alcohol: as noted above, each toluene pro-
duced is accompanied by creation of an additional surface
oxygen, whereas benzoate production and ether production
result in no net production/consumption of surface oxygen—
but production of benzaldehyde, benzene, and benzoic acid all
result in oxygen consumption. Thus, under anaerobic condi-
tions the toluene route acts as a source of oxygen that the al-
dehyde route then consumes. In Figure 3B.i and 6 it is shown
that in the absence of oxygen the toluene/benzaldehyde selec-
tivity shifts towards the toluene route, and towards a 1:1
stoichiometry for toluene and aldehyde production.
Figure 6. Ratios of A) toluene versus benzaldehyde production and B) ben-
zoic acid versus benzyl benzoate production and benzene versus benzyl
benzoate production as a function of the O
termined based on final concentrations.
2
gas flow partial pressure, as de-
Initial alcohol concentration dependence
the lack of readsorption of products is not surprising: the sur-
face is likely always covered with a combination of organics,
oxygen, and hydrogen. All of the products would require two
or more open sites to readsorb dissociatively, whereas the al-
cohol only requires one open site adjacent to a surface oxygen
to dissociatively chemisorb. Thus, during the course of experi-
ment the alcohol continued to dissociatively chemisorb where-
as the products formed did not.
In Figure 2, the initial alcohol concentration dependence is
shown. We see several trends that are consistent with our
mechanism. 1) In Figure 2A.i, A.ii it is shown that the alkoxy
route (aldehyde, toluene, benzyl ether) is promoted by addi-
tional alcohol. 2) In Figure 2B.ii it is shown that the selectivity
towards the carbonyloxyl/dioxy route is decreased as the alco-
hol concentration is increased, presumably because the surface
is covered by more alkoxy groups leaving fewer opportunities
for carbonyloxyl/dioxy species to form. 3) The benzyl ether
production displays a different initial alcohol concentration de-
pendence from that of toluene or the aldehyde because the
ether requires two adjacent aromatic groups (i.e., is second
order with respect to the alcohol).
Oxygen pressure dependence
In Figure 3, the oxygen pressure dependence of the integrated
production for the species monitored is shown. Several trends
that are all consistent with our mechanism are observed. 1) In-
creased oxygen results in increased rate of reaction during the
In Figure 2A.i, A.ii, upon changing initial concentration of al-
cohol, all products increase except for the benzoate, which de-
creases. The production of aldehyde indicates that alkoxy spe-
cies are still produced, and the production of acid and ben-
zene indicate that the carbonyloxyl is still produced. However,
the acid production increases by a quantity comparable to the
benzoate decrease, so this phenomenon is probably attributa-
ble to increased surface hydrogen favoring the acid route. In
Figure 2B.i an increase in selectivity towards the toluene rather
than the aldehyde is observed, supporting the interpretation
120 min monitored and higher benzaldehyde production, con-
sistent with the oxygen-promoted alcohol dissociation mecha-
nism 2) The benzoic acid, benzene, and benzyl benzoate inte-
grated production increase as a function of oxygen, consistent
with oxygen promotion of the carbonyloxyl/dioxy route, by in-
creased formation of carbonyloxyls at the expense of alkoxy
groups. The ratio of the carbonyloxyl/dioxy route versus the
alkoxy route as a function of oxygen demonstrates this trend
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
7
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
that there is more surface hydrogen present during the course
of the experiment if the initial alcohol concentration is higher.
The ratio of the carbonyloxyl route to the alkoxy routes
demonstrates that with increasing initial alcohol concentration,
the carbonyloxyl routes are decreased relative to the alkoxy
route. This is not surprising because of the competition be-
tween alkoxy groups and oxygen for surface sites, and similar
mechanistic competition (carbonyloxyl/dioxy route versus
alkoxy route) was seen in experiments with methanol on
cies with a surface oxygen facilitating the reaction, or c) an
alkoxy species and an alkyl species without a surface
oxygen facilitating reaction.
3) Discrimination of whether abstraction of hydrogen from
alkoxy groups occurs by empty sites or by a surface hy-
droxyl as the hydrogen acceptor (in the context of carbony-
loxyl formation and separately in the context of benzalde-
hyde formation).
4) Discrimination between a molecular/alkoxy based dispro-
portionation reaction versus an alkyl pathway for toluene
formation.
[
65]
a PdO(101) thin film.
5
) Confirmation that some of the trends observed from vary-
ing the oxygen pressure are attributable to oxygen scav-
enging surface hydrogen.
Temperature dependence
In Figure 4, we see that the toluene and ether production in-
crease as a function of temperature at the expense of the alde-
hyde production. The activation barrier for CꢀO breaking to
form the alkyl intermediate from the alkoxy is likely higher
than the barrier for aldehyde formation from the alkoxy (see
earlier discussion, alkyl groups from CꢀO breaking in an alkoxy
group on Pd surfaces are almost never formed as a major
product). In this context, if the surface is covered, then the
higher barrier route would be expected to become increasing-
ly important at higher temperatures, as observed. There may
also be a compounded effect if water production from surface
oxygen increases as a function of temperature, such that the
surface oxygen decreases leading to an increase in the tolu-
ene/ether route (similar to moving left in the plots of Fig-
ure 6A).
6) Confirmation that the trends observed from varying tem-
perature are partially attributable to decreased oxygen on
the surface during the course of experiment.
7) It is currently believed that b-hydrogen abstraction is the
[5]
rate limiting step in aldehyde formation, though with the
complicated mechanism presented here there is likely more
than one rate-determining step, depending on the reaction
conditions.
We are currently working towards a microkinetic modeling
follow-up study with the aim of resolving several of the above
issues.
In Figure 4, we also see that the carbonyloxyl/dioxy route
decreases as a function of temperature. This trend is unlikely
to be the result of the activation barriers, because the activa-
tion barrier from alkoxy to carbonyloxyl is probably larger than
the barrier to the aldehyde, based upon the aldehyde appear-
ing at lower temperature than the acid for TPD of ethanol on
Conclusions
By using an experiment designed to keep the oxygen partial
pressure constant in the gas phase, qualitative kinetic analysis
of benzyl alcohol oxidation over carbon-supported Pd nano-
particles has enabled us to elucidate the reaction mechanism.
The elucidated mechanism is shown in Scheme 1 and has the
following main features: alkoxy intermediates are formed,
which lead to benzaldehyde, benzyl ether, and toluene. The
toluene and benzyl ether products share a common alkyl inter-
mediate. A second route occurs if an alkoxy intermediate is
converted to a carbonyloxyl (“neutral carboxylate”), and this
pathway leads to benzene, benzoic acid, and benzyl benzoate.
The proposed mechanism suggests that the selectivity is influ-
enced by not only temperature and the coverage of the reac-
tants, but also by the side reaction of oxygen scavenging sur-
face hydrogen. Inclusion of the coverages of atomic hydrogen,
atomic oxygen, and surface hydroxyls are expected to be re-
quired for a complete description of the quantitative kinetics.
[
31]
Pd(111). Thus, we ascribe the trend of decreasing carbony-
loxyl/dioxy route to a coverage effect, which we speculate as
owing to decreased oxygen coverage during the course of ex-
periment at higher temperatures.
With varying temperature, the final concentration of the al-
cohol (Figure 4A.i) suggests that at low and high temperatures
adsorption of the alcohol is less favored. At low temperatures
the alcohol absorption might be self-blocking owing to
a lower reaction rate, and at high temperatures the equilibrium
for adsorption may be shifted towards fewer alkoxy groups on
the surface.
Remaining questions
Several issues remain to be understood by future studies, likely
to be resolved by microkinetic modeling and possibly DFT cal-
culations.
Experimental Section
Materials
1
2
) Discrimination of whether the dominant mechanism for
alkoxy formation from the alcohol is facilitated by a surface
oxygen or a surface hydroxyl.
Na PdCl , was from Aldrich (99.99% purity) and activated carbon
from Camel (X40S; SA=900–1100 m g ; PV=1.5 mLg ; pH 9–10).
2
4
2
ꢀ1
ꢀ1
NaBH4 of purity> 96% from Fluka, polyvinylalcohol (PVA) (M =
w
) Elucidation of whether the ether molecule is formed from
a) two alkoxy species, b) an alkoxy species and an alkyl spe-
13000–23000, 87–89% hydrolyzed) from Aldrich were used. Gas-
eous oxygen from SIAD was 99.99% pure.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
8
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Catalyst preparation
Acknowledgements
Pd/AC (palladium supported on activated carbon) was prepared by
following the procedure reported in ref. [16]. In brief,
Na PdCl ·2H O (0.043 mmol) salt and freshly prepared 1 wt% PVA
Work by A. Savara was sponsored by the Division of Chemical
Sciences, Geosciences, and Biosciences, Office of Basic Energy Sci-
ences, U.S. Department of Energy.
2
4
2
or PVP solution were added to a 100 mL volume of H O (Pd/PVA
2
ratio 1:1 wt/wt). After 3 min, a NaBH4 0.1m solution (Pd:NaBH4
molar ratio of 1:8) was added within few minutes from its genera-
tion, the suspension was acidified at pH 2 by sulfuric acid and the
support was added under vigorous stirring. The catalyst was fil-
tered and washed several times with distilled water. The samples
were dried at 808C for 2 h. The amount of the support was calcu-
lated to obtain a final metal loading of 1 wt%. The particles had
a size distribution from 2 to 8 nm, centered around 4 nm, and
were characterized by transmission electron microscopy. The sur-
face palladium comprised 31% of the total palladium atoms based
on Table 2.1 of ref. [66], which is within 10% agreement of approxi-
mating the particles as spheres for most of the particles in the size
distribution used here.
Keywords: adsorption · alcohols · oxidation · palladium ·
reaction mechanisms
[
1] M. Besson, P. Gallezot, Catal. Today 2000, 57, 127–141.
[2] T. Mallat, A. Baiker, Chem. Rev. 2004, 104, 3037–3058.
[3] N. Dimitratos, J. A. Lopez-Sanchez, G. J. Hutchings, Chem. Sci. 2012, 3,
20–44.
[16]
[
[
4] S. E. Davis, M. S. Ide, R. J. Davis, Green Chem. 2013, 15, 17–45.
5] Z. Guo, B. Liu, Q. H. Zhang, W. P. Deng, Y. Wang, Y. H. Yang, Chem. Soc.
Rev. 2014, 43, 3480–3524.
[
[
[
6] T. A. G. Silva, R. Landers, L. M. Rossi, Catal. Sci. Technol. 2013, 3, 2993–
2999.
7] S. Sarina, S. Bai, Y. Huang, C. Chen, J. Jia, E. Jaatinen, G. A. Ayoko, Z.
Bao, H. Zhu, Green Chem. 2014, 16, 331–341.
8] Y. T. Chen, H. P. Wang, C. J. Liu, Z. Y. Zeng, H. Zhang, C. M. Zhou, X. L. Jia,
Y. H. Yang, J. Catal. 2012, 289, 105–117.
Catalytic tests
[9] H. T. Tan, Y. T. Chen, C. M. Zhou, X. L. Jia, J. X. Zhu, J. Chen, X. H. Rui,
Q. Y. Yan, Y. H. Yang, Appl. Catal. B 2012, 119, 166–174.
Reactions were performed in a 30 mL glass reactor equipped with
a thermostat and an electronically controlled magnetic stirrer. The
glass reactor was connected by tubing to a mass-flow controller
used to flow gas mixtures, with a second port connected by
tubing to the building exhaust to enable continuous flow of gas
through the reactor such that the gas phase partial pressures re-
mained constant during experiment (with the total pressure always
[
10] D. M. Meier, A. Urakawa, A. Baiker, J. Phys. Chem. C 2009, 113, 21849–
21855.
[11] D. Ferri, C. Mondelli, F. Krumeich, A. Baiker, J. Phys. Chem. B 2006, 110,
22982–22986.
[12] J. D. Grunwaldt, M. Caravati, A. Baiker, J. Phys. Chem. B 2006, 110,
25586–25589.
[
[
[
13] C. Mondelli, D. Ferri, J.-D. Grunwaldt, F. Krumeich, S. Mangold, R. Psaro,
A. Baiker, J. Catal. 2007, 252, 77–87.
14] A. P. G. Kieboom, J. F. De Kreuk, H. Van Bekkum, J. Catal. 1971, 20, 58–
1
bar). Experiments used 60 mg samples of supported catalyst
(
0.6 mg of palladium per sample). Benzyl alcohol and the catalyst
66.
were mixed in p-xylene. The benzyl alcohol/metal ratio was typical-
ly approximately 3000 moles alcohol per mole Pd. Various alcohol/
xylene ratios were used during experiments (benzyl alcohol/xylene
15] Y. T. Chen, H. J. Zheng, Z. Guo, C. M. Zhou, C. Wang, A. Borgna, Y. H.
Yang, J. Catal. 2011, 283, 34–44.
[16] A. Villa, D. Wang, N. Dimitratos, D. S. Su, V. Trevisan, L. Prati, Catal. Today
2
5/75, 50/50, or 75/25 by percent volume) with the total liquid
2010, 150, 8–15.
volume always kept at 10 mL. The reactor temperature was set to
[17] N. Dimitratos, A. Villa, D. Wang, F. Porta, D. S. Su, L. Prati, J. Catal. 2006,
244, 113–121.
ꢀ
1
7
0, 80, 90, or 1008C and the oxygen flow set at 30 mLmin . Heat-
[
18] X. Wang, G. Wu, N. Guan, L. Li, Appl. Catal. B 2012, 115 – 116, 7–15.
19] M. Sankar, E. Nowicka, R. Tiruvalam, Q. He, S. H. Taylor, C. J. Kiely, D. Be-
thell, D. W. Knight, G. J. Hutchings, Chem. Eur. J. 2011, 17, 6524–6532.
20] S. Meenakshisundaram, E. Nowicka, P. J. Miedziak, G. L. Brett, R. L. Jen-
kins, N. Dimitratos, S. H. Taylor, D. W. Knight, D. Bethell, G. J. Hutchings,
Faraday Discuss. 2010, 145, 341–356.
ing to the desired temperature was on the order of minutes, and
reaction time zero was marked by the beginning of stirring. Period-
ic removal of samples from the reactor was performed by a syringe.
Identification and analysis of the products was done by compari-
son with reference samples by gas chromatography using a HP
[
[
7
3
820A gas chromatograph equipped with a capillary column (HP-5
0 mꢁ0.32 mm, 0.25 mm film, made by Agilent Technologies) and
[21] E. Nowicka, J. P. Hofmann, S. F. Parker, M. Sankar, G. M. Lari, S. A. Kon-
drat, D. W. Knight, D. Bethell, B. M. Weckhuysen, G. J. Hutchings, Phys.
Chem. Chem. Phys. 2013, 15, 12147–12155.
a thermal conductivity detector. Quantification of reaction prod-
ucts was done by internal calibrations on the gas chromatograph
using undecane as a standard. A stirring rate of 1250 rpm was
used, and mass-transfer effects were excluded by verifying that the
same the kinetic profiles was obtained with a stirring rate of
[
22] C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Phys. Chem. B 2005, 109,
58–967.
9
[
[
[
[
[
23] R. L. Burwell, J. B. Peri, Annu. Rev. Phys. Chem. 1964, 15, 131.
24] B. E. Bent, Chem. Rev. 1996, 96, 1361–1390.
25] A. Savara, E. Weitz, Annu. Rev. Phys. Chem. 2014, 65, 249–273.
26] F. Zaera, Catal. Lett. 2003, 91, 1–10.
27] J. Kua, F. Faglioni, W. A. Goddard, J. Am. Chem. Soc. 2000, 122, 2309–
2321.
7
50 rpm for one experiment—consistent with another study that
found mass transfer was not rate limiting for stirring speeds of
00 rpm or above for a related catalyst system (supported AuPd
5
[67]
nanoparticles) in a glass stirred reactor. Based on mass balance
of the aromatic rings (i.e., the concentration of aromatics at the
end of experiment versus at the beginning of experiment), little to
no organic evaporated during experiment (<10% of the reactants
and products). Turnover frequencies were on the order of
[28] J. A. Gates, L. L. Kesmodel, J. Catal. 1983, 83, 437–445.
[29] S. W. Jorgensen, R. J. Madix, Surf. Sci. 1987, 183, 27–43.
[
[
[
30] A. K. Bhattacharya, M. A. Chesters, J. Catal. 1988, 109, 314–319.
31] J. L. Davis, M. A. Barteau, Surf. Sci. 1988, 197, 123–152.
32] R. J. Levis, Z. C. Jiang, N. Winograd, J. Am. Chem. Soc. 1988, 110, 4431–
4432.
1
000 molecules per palladium atom per hour (see the Supporting
[33] J. L. Davis, M. A. Barteau, J. Am. Chem. Soc. 1989, 111, 1782–1792.
[34] J. L. Davis, M. A. Barteau, Surf. Sci. 1990, 235, 235–248.
[35] N. Kruse, M. Rebholz, V. Matolin, G. K. Chuah, J. H. Block, Surf. Sci. 1990,
238, L457–L462.
[5,68]
Information), which is in line with literature results.
Synthesis,
experiments, and chemical analysis were performed at Universitꢂ
degli Studi di Milano. Data analysis and mechanistic analysis were
performed primarily at Oak Ridge National Laboratory.
[36] N. Kruse, M. Rebholz, J. H. Block, Chem. Ing. Tech. 1991, 63, 615–618.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&
9
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[
37] M. Rebholz, V. Matolin, R. Prins, N. Kruse, Surf. Sci. 1991, 251, 1117 –
[56] A. Savara, W. Ludwig, S. Schauermann, ChemPhysChem 2013, 14, 1686–
1695.
1122.
[
[
38] J. L. Davis, M. A. Barteau, Surf. Sci. 1992, 268, 11–24.
39] J. J. Chen, Z. C. Jiang, Y. Zhou, B. R. Chakraborty, N. Winograd, Surf. Sci.
[57] G. C. Bond, Metal-Catalysed Reactions of Hydrocarbons, Springer Science,
New York Inc., 2005.
1
995, 328, 248–262.
40] S. M. Francis, J. Corneille, D. W. Goodman, M. Bowker, Surf. Sci. 1996,
64, 30–38.
41] G. S. Jones, M. A. Barteau, J. M. Vohs, J. Phys. Chem. B 1999, 103, 1144–
151.
[58] E. Shustorovich, Metal-surface reaction energetics: theory and applica-
tions to heterogeneous catalysis, chemisorption, and surface diffusion,
VCH, New York, 1991.
[59] I. Chorkendorff, J. W. Niemantsverdriet, Concepts of Modern Catalysis
and Kinetics, Wiley-VCH, Weinheim 2007.
[
[
[
3
1
42] J. Hoffmann, S. Schauermann, V. Johanek, J. Hartmann, J. Libuda, J.
Catal. 2003, 213, 176–190.
43] M. S. Hegde, Proc. Indian Acad. Sci. Chem. Sci. 1984, 93, 373–387.
44] B. N. Zope, D. D. Hibbitts, M. Neurock, R. J. Davis, Science 2010, 330, 74–
[60] R. M. Williams, J. W. Medlin, Surf. Sci. 2014, 619, 30–38.
[61] J. L. Davis, M. A. Barteau, Surf. Sci. 1991, 256, 50–66.
[62] C. M. Friend, X. Xu, Annu. Rev. Phys. Chem. 1991, 42, 251–278.
[63] N. Dimitratos, J. A. Lopez-Sanchez, D. Lennon, F. Porta, L. Prati, A. Villa,
Catal. Lett. 2006, 108, 147–153.
[
[
78.
[45] Z. J. Li, W. T. Tysoe, Surf. Sci. 2010, 604, 1377–1387.
[46] C. J. Zhang, P. Hu, J. Chem. Phys. 2001, 115, 7182–7186.
[47] W. Ludwig, A. Savara, B. Brandt, S. Schauermann, Phys. Chem. Chem.
Phys. 2011, 13, 966–977.
[64] A. Villa, N. Janjic, P. Spontoni, D. Wang, D. S. Su, L. Prati, Appl. Catal. A
2009, 364, 221–228.
[65] C. Hakanoglu, J. A. Hinojosa, J. F. Weaver, J. Phys. Chem. C 2011, 115,
11575–11585.
[
48] A. Savara, W. Ludwig, K. H. Dostert, S. Schauermann, J. Mol. Catal. A
[66] C. P. Poole, F. J. Owens, Introduction to Nanotechnology, Wiley, New York,
2003.
[67] E. H. Cao, M. Sankar, E. Nowicka, Q. He, M. Morad, P. J. Miedziak, S. H.
Taylor, D. W. Knight, D. Bethell, C. J. Kiely, A. Gavriilidis, G. J. Hutchings,
Catal. Today 2013, 203, 146–152.
2
013, 377, 137–142.
49] S. H. Pang, A. M. Roman, J. W. Medlin, J. Phys. Chem. C 2012, 116,
3654–13660.
[
1
[
[
[
50] R. M. Williams, J. W. Medlin, Langmuir 2014, 30, 4642–4653.
51] D. D. Hibbitts, M. Neurock, J. Catal. 2013, 299, 261–271.
52] B. L. J. Poad, B. B. Kirk, P. I. Hettiarachchi, A. J. Trevitt, S. J. Blanksby, T.
Clark, Angew. Chem. 2013, 125, 9471–9474; Angew. Chem. Int. Ed. 2013,
[68] J. Chen, Q. H. Zhang, Y. Wang, H. L. Wan, Adv. Synth. Catal. 2008, 350,
453–464.
52, 9301–9304.
[53] M. A. Barteau, Chem. Rev. 1996, 96, 1413–1430.
[54] J. M. Vohs, Chem. Rev. 2013, 113, 4136–4163.
[55] S. Yanagisawa, T. Tsuneda, K. Hirao, Y. Matsuzaki, J. Mol. Struct. THEO-
CHEM 2005, 716, 45–60.
Received: July 18, 2014
Published online on && &&, 0000
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&10
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
A. Savara,* C. E. Chan-Thaw, I. Rossetti,
A. Villa, L. Prati
&& – &&
Benzyl Alcohol Oxidation on Carbon-
Supported Pd Nanoparticles:
Two key intermediates are involved in
the catalytic liquid-phase oxidation of
benzyl alcohol. An alkoxy intermediate
is proposed for the route to benzyl al-
dehyde, toluene, and benzyl ether. A
carbonyloxyl intermediate is proposed
for the route to benzoic acid, benzene,
and benzyl benzoate.
Elucidating the Reaction Mechanism
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 11
&11
&
ÞÞ
These are not the final page numbers!