Content uploaded by Rosa Arrigo
Author content
All content in this area was uploaded by Rosa Arrigo
Content may be subject to copyright.
Gold catalyzed liquid phase oxidation of alcohol:
the issue of selectivity
L. Prati,*
a
A. Villa,
a
C. E. Chan-Thaw,
a
R. Arrigo,
b
D. Wang
c
and D. S. Su
b
Received 11th February 2011, Accepted 2nd March 2011
DOI: 10.1039/c1fd00016k
Commercial carbon nanotubes (CNTs) and carbon nanofibers (CNFs) modified
in various ways at the surface have been used as supports for gold nanoparticles
(AuNPs) in order to study their influence on the activity/selectivity of catalysts in
the aqueous oxidation of alcohol. Particularly oxidative treatment was used to
introduce carboxylic functionalities, whereas subsequent treatment with NH
3
at
different temperatures (473 K, 673 K and 873 K) produced N-containing groups
leading to an enhancement of basic properties as the NH
3
treatment temperature
was increased. The nature of the N-containing groups changed as the
temperature increased, leading to an increase in the hydrophobicity of the
support surface. Similar Au particle size and similar textural properties of the
supports allowed the role of chemical surface groups in both the activity and the
selectivity of the reaction of glycerol oxidation to be highlighted. An increase of
basic functionalities produced a consistent increase in the activity of the catalyst,
which was correlated to the promoting effect of the basic support in the
alcoholate formation and the subsequent C–H bond cleavage. The selectivity
towards primary oxidation products (C3 compounds) was the highest for the
catalysts treated with NH
3
at 873 K, which presented the most hydrophobic
surface. The same trend in the catalyst activity has been obtained in the aqueous
benzyl alcohol base-free oxidation. As in the case of glycerol, the increasing of
basicity and/or hydrophobicity increased the consecutive reactions.
Introduction
The success of gold as a catalyst for selective oxidation in the liquid phase is princi-
pally due to the peculiar characteristics of this metal compared to more classical Pd
or Pt based catalysts when O
2
is used as the oxidant. In fact, it has been shown that
a Au catalyst is less prone to deactivation in the presence of O
2
compared to Pd or
Pt.
1–3
Moreover, gold catalysts show high selectivity towards the oxidation of
primary alcohol with respect to secondary alcohol.
1,2
Since the first observation
on ethylene glycol, a lot of papers have dealt with the application of gold catalysts
to this important class of reaction, mainly with two aims. The first aim is to try to
improve the activity and the second aim is to avoid parallel or consecutive reactions,
thus enhancing the selectivity to the desired product. This latter aspect is very impor-
tant from an application point of view and most recent studies focused on it. More-
over, the mechanistic understanding of the factors ruling selectivity become
a priority task for optimizing catalyst design.
a
Dipartimento di Chimica Inorganica Metallorganica e Analitica L. Malatesta, Universit
a degli
Studi di Milano, Via Venezian 21, 20133 Milano, Italy. E-mail: Laura.Prati@unimi.it
b
Fritz Haber Institute of the Max Planck Society, Faradayweg 4-6, D-14195 Berlin, Germany
c
Institut f€
ur Nanotechnologie, Forschungszentrum Karlsruhe in der Helmholtz-Gemeinschaft,
Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany
PAPER www.rsc.org/faraday_d | Faraday Discussions
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
One of the most studied substrates is benzyl alcohol, which can be selectively
oxidized to benzaldehyde. Dehydrogenation on gold-supported nanoparticles to
produce aldehyde can be competitive with the disproportionation of benzyl alcohol
to benzaldehyde and toluene,
4
but dehydrogenation prevails when small gold nano-
particles (< 4 nm) are present. This is due to the increasing of the unsaturated coor-
dination sites, which promotes the bC–H bond cleavage, which represents the rate
determining step. This trend was confirmed also by periodic density functional
theory (DFT) calculations
5
addressing the higher activity of small nanoparticles to
the roughness of the surface. High selectivity to benzaldehyde (> 99%) has been re-
ported depending on reaction conditions, i.e. solvent, temperature, but principally
on the support employed (MgO and hydrotalcite).
4
The main by-products are rep-
resented by benzoic acid, benzyl benzoate (from consecutive reactions) or toluene
and benzene (from parallel reactions). The selectivity of the reaction can be strongly
affected by the presence of a base, whose function is still under investigation. It is
believed that the base can cleave the O–H bond of the alcohol to form an alkoxide
intermediate.
6
In this process, the role of the support (especially at the interface with
the Au nanoparticles) and also the role of coadsorbed oxygen atoms on Au surface
could be relevant.
4,7
Glycerol represents another well-studied substrate, in view of its importance as
a cheap material derived from biomass, on which a new chemical platform can be
based.
8,9
Glycerol presents some fundamental differences compared to benzyl
alcohol: the non-activating nature of the OH groups (pK
a
¼14), the presence of
primary beside secondary OH groups and the chelating nature (also of the oxidation
products), which can strongly affect the catalyst life by providing irreversible
adsorption and/or enhancing the leaching of metal. Moreover, another important
difference lies in the aqueous solubility and the cheap availability of glycerol only
in aqueous solution. Therefore, all the catalytic tests should be carried out in water
instead of in organic solvent (toluene, xylene, cyclohexane) or solventless, as in the
case of benzyl alcohol.
Since the first studies on glycerol oxidation,
1,2
it appeared evident that basic condi-
tions enhanced both activity and selectivity positively. In fact, the presence of a base
enhanced the reaction rate by facilitating the alkoxide formation
10,11
and by favoring
Scheme 1 Reaction scheme for glycerol oxidation under basic conditions.
Faraday Discuss. This journal is ªThe Royal Society of Chemistry 2011
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
the desorption of the highly-chelating hydroxyacid, thus decreasing irreversible
adsorption on metal active sites.
12
In addition, basic conditions also have a beneficial
effect on the selectivity towards the primary alcohol oxidation products (i.e. glycer-
ate) (Scheme 1) through keto-enolic equilibrium and through accelerating the
decomposition rate of H
2
O
2
formed during the reaction, which has been recognized
as responsible for the majority of the C–C bond cleavage products (i.e. glycolate,
oxalate, formate).
10,13
More recently, efforts in oxidizing glycerol in a base-free medium revealed the
specific role of the support, but also showed evidence of the limitation in using
monometallic gold based catalysts.
14
In fact, even under stronger conditions (373
K) than those used when the base was present (323–333 K), the reaction rates
were very low and only the addition of a second metal (typically Pt) yielded a discrete
reaction rate. Selectivity appeared to be determined mainly by the support. Accord-
ing to the model suggested by Davis, on the basis of DFT calculations and labeling
tests,
11
the role of the support becomes fundamental in the second elementary step of
the reaction, i.e. the hydration of aldehyde to acetal with the subsequent formation
of the corresponding carboxylic acid. Thus, in this study, we investigated the role of
the support in the selective glycerol oxidation further, mainly focusing our attention
on the selectivity of the process.
Experimental
Materials
The gold sponge, 99.9999% purity, was purchased from Fluka. Commercial CNFs
PR24-PS from Applied Science (average diameter of 88 30 nm and a specific
surface area of 43 m
2
g
1
) and CNTs Baytubes (average diameter of 10 2nm
and a specific surface area of 288 m
2
g
1
) from Bayer were employed. The function-
alization of the CNFs and the CNTs was performed according to the procedure
report in ref. 15. The oxygen-containing nanocarbons were obtained by treating
the pristine support with HNO
3
according to the following procedure: a solution
of CNTs in HNO
3
concentrate (20 g of CNT per liter of HNO
3
) was kept at 373
K for 2 h under continuous stirring, then rinsed with distilled water, and finally dried
at 343 K for several hours. N-containing CNFs were obtained from the pre-oxidized
CNFs by thermal treatment (10 g for each batch) with NH
3
in the temperature range
473–873 K for 4 h. NaBH
4
of purity > 96% purchased from Fluka and polyvinylal-
cohol (PVA) (M 10,000) from Aldrich were used. NaOH of the highest purity avail-
able was also from Fluka. Gaseous oxygen from SIAD (99.99% pure) was used.
Glycerol (87% wt solution), glyceric acid and all the intermediates were purchased
from Fluka. Benzyl alcohol, benzaldehyde, toluene and all the intermediates were
also from Fluka.
Catalyst preparation
a) PVA-protected gold sol preparation. Solid NaAuCl
4
$2H
2
O (0.043 mmol) and
PVA (2% wt) solution (1.64 mL) were added to 130 mL H
2
O. After 3 min, NaBH
4
(0.1 M) solution (1.3 mL) was added to the yellow solution under vigorous magnetic
stirring. The ruby red Au(0) sol was formed immediately.
16
b) Immobilization step. Within a few minutes of sol generation, the sol was im-
mobilised by adding the support under vigorous stirring. The amount of support
was calculated to have a final gold loading of 1% wt. After 2 h, the slurry was filtered
and the catalyst washed thoroughly with distilled water; it was then used in the wet
form. ICP analyses were performed on the filtrate using a Jobin Yvon JV24 to verify
the gold loading on the support.
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
Catalyst characterisation
a) Titration procedure. Potentiometric titration of the basic sites was carried out
on the samples using a Mettler Toledo Titrator. Approximately 0.2 g of sample was
suspended in 50 mL of KCl 10
3
M and then sonicated and equilibrated for several
hours. Prior to each measurement, the suspension was continuously saturated with
argon to eliminate the influence of CO
2
, until the pH was constant. Volumetric stan-
dards of HCl (0.01M) or NaOH (0.10M) were used as titrant, starting from the
initial pH of the CNFs suspension.
b) Metal loading. The gold content for the sol prepared catalysts was checked by
ICP analysis of the filtrate on a Jobin Yvon JY24. The water content was determined
by drying a sample at 423 K in air for 5 h. A check of Au loading was also performed
directly on the catalysts, confirming the quantitative adsorption of Au NPs of the
sol. The samples were weighed and loaded into a home-made NaCl crucible, then
annealed at 973 K for two hours in air to burn off the carbon support. The residue
was then dissolved in 5 mL of freshly made aqua-regia (3 : 1 hydrochloric/nitric acid
– Omnipure grade – EMD). The sample was diluted to 50 mL with deionized water.
ICP Standards were prepared by serial dilution of an Alfa-Aesar Au ICP standard.
The dissolved NaCl had no effect on the ICP data.
c) Morphology and microstructure of the catalysts were characterized by TEM.
The powder samples of the catalysts were ultrasonically dispersed in ethanol and
mounted onto copper grids covered with carbon film. A Philips CM200 FEG elec-
tron microscope, operating at 200 kV and equipped with a Gatan Tridiem imaging
filter, was used for TEM observation. EDX analysis was performed in the same
microscope using a DX4 analyzer system (EDAX).
Oxidation experiments
The reactions were carried out in a thermostatted glass reactor (30 mL) provided
with an electronically controlled magnetic stirrer connected to a large reservoir
(5000 mL) containing oxygen at 300 kPa.
Glycerol oxidation: 0.3 M Glycerol solution, NaOH (NaOH/glycerol ¼4 mol
mol
1
) and the Au catalyst (glycerol/metal ¼1000 mol mol
1
) were added (total
volume 10 mL). The reactor was pressurised at 300 kPa of O
2
and thermostatted
at 323 K. After an equilibration time of 5 min, the reaction was initiated by stirring
and samples were taken every 15 min and analysed by HPLC on a Varian 9010
HPLC equipped with a Varian 9050 UV (210 nm) and a Waters R.I. detector in
series. An Alltech OA-1000 column (300 mm 6.5 mm) was used with aqueous
H
3
PO
4
0.1% wt/wt M (0.5 mL min
1
) as the eluent. Samples of the reaction mixture
(0.5 mL) were diluted (5 mL) using the eluent. Products were assigned by compar-
ison with authentic samples.
Base-free reactions were carried out under the same conditions except for the
temperature, which was increased to 373 K and glycerol/metal ratio (500 mol mol
1
)
Benzyl alcohol oxidation: 0.3 M benzyl alcohol and the catalyst (substrate/metal
¼500 mol mol
1
) were mixed in distilled water (total volume 10 mL). The reactor
was pressurized at the desired pressure of O
2
and thermostatted at 333 K. The reac-
tion does not start during the heating to temperature, as verified by sampling when
the reaction temperature was reached without stirring (t¼0). The reaction was initi-
ated by stirring. After the end of reaction, the catalyst was filtered off and the
product mixture was extracted with cyclohexane. Recoveries were always 98% 3
with this procedure. For the identification and analysis of the products, a GC-MS
and GC (a Dani 86.10 HT Gas Chromatograph equipped with a capillary column,
BP21 30m 0.53mm, 0.5 mm Film, made by SGE) were used. Comparison with
authentic samples was used. For the quantification of the reactant-products, the cali-
bration method using an external standard was employed.
Faraday Discuss. This journal is ªThe Royal Society of Chemistry 2011
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
Test for hydrogen peroxide degradation
A 1 mM solution of H
2
O
2
(15 mL) was stirred at the appropriate temperature under
N
2
atmosphere. The pH of the solution was adjusted by adding H
2
SO
4
6 M (pH2).
The amount of the catalyst added was the same as in the glycerol oxidation (about
20–30 mg). Hydrogen peroxide was quantified time by time by permanganate titra-
tion. The following procedure has been used: a sample (5 mL) of the filtered reacting
solution was titrated with a 0.01 N KMnO
4
solution at constant pH of 2.0 0.1 (by
adding concentrated H
2
SO
4
). The detection limit of H
2
O
2
was 0.01 mM.
Results and discussion
The selectivity of the reaction, as discussed above, can be strongly affected by the
presence of a base. However, a relationship between particle size and selectivity
was found.
1,2
Larger gold particles (> 10 nm) were more selective towards C
3
-oxida-
tion products than smaller ones, which produced a relevant amount of C–C bond
cleavage products. However, more recent studies revealed that selectivity could be
ruled by a more complex platform of factors.
17–19
Particularly the textural and chem-
ical properties of the supporting material appeared to be fundamental.
17
In fact, it
was reported that selectivity in the glycerol oxidation of Au on MgAl
2
O
4
is tuned
by the Al/Mg ratio at the surface besides the particle size and the exposure of the
metal. Therefore, for studying the pure support effect in more detail, any other
differences such as particle size, surface area and porosity of the supporting mate-
rials should be avoided. The selection of the materials was thus performed, taking
into account the surface area (as high as possible to increase the metal dispersion)
and porosity (avoiding as much as possible micropores for diffusional problems).
Moreover, a surface which is easy to modify should be considered. Carbon nano-
tubes (CNTs) appeared to be good candidates, as they show regular textural prop-
erties with surface groups which can be modified by chemical/physical treatments
without damaging the overall structure.
Two different commercial samples were used as starting materials. The character-
izations have been recently reported:
20
one from Bayer Material Science A.G. (Bay-
tubes, Carbon Nanotubes CNT) and one from Pyrograf Products Inc. (PR24,
Carbon Nanofiber CNF). The physical properties of these two materials are re-
ported in Table 1. An oxidative treatment (HNO
3
65% wt, 2 h, 373 K) increased
the functionalization on the outermost walls, but did not significantly affect either
the grain shape and the size, or the structure. An important effect of the treatment
was, in both cases, to remove residual amorphous carbon from the surface. This
procedure is known to increase the presence of carboxylic acids.
20
The acidic site
titration confirmed an increase of the amount of acidic groups (Table 1). A basic
functionalization was carried out by the subsequent treatment with NH
3
at different
temperatures, following the procedure reported elsewhere.
21
Table 2 reports the
basic site titration data that show the increase of the basic groups with increasing
post-treatment temperature (473 K–673 K–873 K). It should be noted that, even
at the highest temperature (873 K), the overall structure of CNFs was maintained.
21
Table 1 Physical data for CNTs and CNFs
AS (m
2
g
1
) Micropore area (m
2
g
1
) diameter (nm) pH mol acid sites/g
CNT 288
a
40
a
10 210—
CNT ox 321
a
38
a
10 2 4.35 2.6 10
3
CNF 43
a
0
a
88 30 6.5 —
CNF ox 37
a
5
a
88 30 4.52 4.5 10
3
a
From ref. [20,21].
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
CNTs and CNFs presented different surface areas, but the contribution of micro-
pores could be considered negligible in both cases (Table 1).
We prepared thus Au supported nanoparticles (AuNPs) on CNTs and CNFs
through the sol immobilization technique which represents the most suitable method
for obtaining similar particle size distribution, regardless of the supporting material
employed.
16
The Au particle size distributions obtained are reported in Table 3.
During the immobilisation step, we noticed an increase in the size of the nanopar-
ticles, probably due to coalescence that increases with decreasing interaction
strength between the AuNPs and the support surface. As preformed AuNPs are
identically generated in all the cases, we can address the difference in diameter of
supported AuNPs to the different surface chemical groups and in particular to their
density. Indeed, we observed similar particle diameter for AuNPs immobilized on all
the treated supports (about 3–4 nm) with a decreasing of the discrepancy between
diameter in the sol and on the support (i.e. before and after the deposition) by
increasing the functional group density.
The acidic groups, introduced by oxidative treatment, appeared less efficient in
stabilizing the nanoparticle diameter than the basic groups. In fact Au/CNT ox
and Au/CNF ox showed the most remarkable increase in the Au particle diameter
(3.83 and 3.65 nm vs. 2.94 nm for a comparable density of functional groups – Table
3). Pristine CNTs and CNFs lead to poor metallic dispersion, probably also due to
the presence of amorphous carbon, which preferentially adsorbs the AuNPs from
the sol. Some representative TEM pictures for Au/CNF and Au/CNT are shown
(Fig. 1 and 2), demonstrating the increase in metal dispersion obtained after func-
tionalization.
Regarding the activity of the catalysts, there are a lot of examples in the literature
confirming that the activity of the catalysts in the liquid phase is related to the
surface exposure of the active sites that should increase by decreasing the particle
size.
17,18
In the present case, the external functionalization of the support surface,
the negligible presence of micropores and the methodology chosen for the deposition
of the gold nanoparticles should avoid the limitation due to diffusion problems. We
should thus expect a trend of activities which follows the particle sizes. However,
looking at the reaction profiles (Fig. 3 and 4) it can be seen clearly that this expected
trend was not followed. In fact, Au/CNT ox and Au/CNF ox, even showing a very
similar particle size distribution (mean size 3.83 nm vs. 3.65 nm – Table 3), presented
Table 2 Titration data for N-CNFs
Sample mol basic sites/g Generated pH
N-CNF 473 K 4.5*10
4
8.0
N-CNF 673 K 6*10
4
8.2
N-CNF 873 K 1*10
3
8.5
Table 3 Au particle size on CNTs and CNFs
Au
sol
Au/CNT
pristine
Au/
CNT ox
Au/CNF
pristine
Au/
CNF ox
Au/N
CNF
473 K
Au/
N-CNF
673 K
Au/
N-CNF
873 K
Statistical
median (nm)
2.45 4.61 3.83 3.80 3.65 3.15 3.11 2.94
Standard
deviation s
0.27 1.32 0.89 0.95 0.78 0.57 0.64 0.51
Faraday Discuss. This journal is ªThe Royal Society of Chemistry 2011
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
TOFs of 742 h
1
and 186 h
1
, respectively (Table 4 – Fig. 3). Also, Au/N-CNF 473 K
and Au/N-CNF 673 K with almost the same particle size (3.15 nm and 3.11 nm,
respectively) showed TOFs of 112 and 597 h
1
, respectively (Table 4 – Fig. 4).
This definitely supports the idea that the supporting material plays an active role
in determining the rate of glycerol oxidation. This is in agreement with the
Fig. 1 TEM images of a) Au/CNF, b) Au/CNF ox, c) Au/CNT, d) Au/CNT ox.
Fig. 2 TEM images of a) Au/N-CNF 473 K, b) Au/N-CNF 673 K, c) Au/N-CNF 873 K.
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
mechanism depicted by Davis,
11
where it is the support which mediates the second
and rate determining step of the reaction, i.e. the activation of the C–H bond of
the ensuing alkoxide intermediate to form the aldehyde. Considering the NH
3
treated CNFs, it was thus not surprising that by increasing the basic site amount
(Table 2) we observed an increasing of the catalyst activity in the same order
(Fig. 4) even the huge difference between Au/N-CNF 473 K and Au/N-CNF 673
K was not fully justified by simply considering the increase of basic sites [Au/N-
CNF 473 K (TOF 112 h
1
) < Au/N-CNF 673 K (TOF 597 h
1
) < Au/N-CNF 873
K (TOF 853 h
1
)]. The positive effect on the catalytic performance in the liquid
phase of NH
3
treatment has also been observed in the case of Pd supported
NPs.
22
It should be noted, that the different functionalization temperature deter-
mined the presence of different surface chemical groups in the samples: at lower
temperature (473 K) prevalent amide-like groups were observed, which were
Fig. 3 Reaction profile for glycerol oxidation using Au/CNF, Au/CNT, Au/CNF ox and Au/
CNT ox.
Fig. 4 Reaction profile for glycerol oxidation using Au/N-CNF 473 K, Au/N-CNF 673 K,
Au/N-CNF 873 K.
Faraday Discuss. This journal is ªThe Royal Society of Chemistry 2011
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
transformed at higher temperature into pyridine-like groups.
21
In the sample treated
at higher temperature (873 K), microcalorimetric measurements also revealed the
presence of strong oxygen basic sites similar to those observed in hydrotalcite
23
and in g-Al
2
O
3
,
24
along with a majority of weaker basic sites. The amination process
produces a gradual increase of the hydrophobic character of the surface by
increasing the temperature of the treatment.
Concerning the selectivity of the glycerol oxidation reaction (Scheme 1), Table 4
showed some unexpected results. On the one hand, using catalysts with similar Au
particle size, we were expecting similar selectivity.
1,2
This did not occur, clearly indi-
cating that some other factors influence the selectivity of the process. Pristine CNTs
and CNFs showed different selectivity, CNTs being more selective toward C2 prod-
ucts (glycolate and oxalate) than CNFs (Table 4). In both cases, the oxidative treat-
ment increased the selectivity towards C3 products (glycerate and tartronate) and
did not alter the activity of the catalysts very much (Table 4 and Fig. 3). In the
case of CNF, we observed a breakdown of selectivity for NH
3
treated samples,
but only when the temperature of treatment was over 673 K, i.e. when the catalysts
become more active. Indeed, selectivity did not change very much for Au/CNF, Au/
CNF ox and Au/N-CNF 473 K, with 55–65% selectivity to glycerate, none to tartr-
onate, 20–27% selectivity to glycolate and 10–16% selectivity to formate (Table 4—
all the selectivities reported are at 90% conversion). On the contrary, in the case of
Au/N-CNF 673 K and Au/N-CNF 873 K, a decrease of C2–C1 products (glycolate
and formate) to 13% selectivity and 5–6% selectivity, respectively, was detected.
Meanwhile an increase in C3 products (glycerate and tartronate) to 79–82% selec-
tivity was observed (Table 4—all the selectivities reported are at 90% conversion).
This behavior could be possibly explained by the surface properties related to the
presence of different chemical groups. Accepting the H
2
O
2
detected during the reac-
tion as responsible for the C–C bond cleavage
10,13
and accepting that H
2
O
2
derives
from the O
2
reduction by H
2
O adsorbed on the catalyst surface,
11
we could expect
H
2
O
2
formation to be related to the water adsorption capacity of the support, i.e.
its hydrophilicity. The functionalization with HNO
3
and the subsequent treatment
with NH
3
at different temperatures has been shown to increase the hydrophobicity
of the surface progressively by increasing the temperature. Thus in our case, the
increase of the basicity in Au/N-CNF 873 K should be expected to increase the
activity by promoting the alcoholate formation and C–H bond cleavage, whereas
the increase of surface hydrophobicity could lead to an higher selectivity to C3 by
decreasing the H
2
O
2
formation. However, during the tests, no significant differences
in H
2
O
2
concentration have been revealed. This was not surprising as the basic
media could influence the degradation rate of the peroxide. Therefore, to better
Table 4 Glycerol oxidation with Au/CNTs and Au/CNFs under basic conditions
a
Catalyst TOF
b
(h
1
)
Selectivity
c
Glycerate Tartronate Glycolate Oxalate Formate
Au/CNTs 648 26 7 14 53 <1
Au/CNT ox 742 59 11 17 13 <1
Au/CNF 182 55 1 27 0 16
Au/CNF ox 186 65 <1 20 0 10
Au/N-CNF 473 K 112 64 <1 22 0 12
Au/N-CNF 673 K 597 66 13 13 0 6
Au/N-CNF 873 K 853 68 14 13 0 5
a
Reaction conditions: in water, glycerol/metal: 1000 mol mol
1
,T¼323 K, pO
2
¼3 atm., 4eq
NaOH
b
TOF calculated at 30 min on the basis of total metal loaded
c
Selectivity at 90%
conversion
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
clarify this point, we carried out some experiments under base-free conditions, as
H
2
O
2
is more stable at neutral-acidic pH. Tests under base-free conditions (373 K
without any addition of base) showed a higher (even low) activity for Au/N-CNF
873 K, the most basic and hydrophobic catalyst. As under non-basic conditions
we expected H
2
O
2
to degrade more slowly than under basic conditions,
13
we can
use the amount of C–C cleavage as a probe of H
2
O
2
formation. Au/N-CNF 873
K, also under these conditions, produced C–C bond cleavage products (glycolate
and formate) in a smaller amount than Au/CNF 473 K (Table 5). H
2
O
2
quantifica-
tion by titration confirmed a lesser concentration in the case of Au/N-CNF 873 K
(0.3 mmol L
1
for Au/N-CNF 873 K vs 2.0 mmol L
1
for Au/CNFox – Table 5).
Moreover blank experiments highlighted the same degradation rate for H
2
O
2
in
the presence of both catalysts (Fig. 5). These results seem to confirm our hypothesis
that the hydrophobic nature of N-CNF 873 K contributed to the decrease in
peroxide formation i.e. an increase in C3-product selectivity. More accurate kinetic
investigations are being carried out in order to confirm this finding.
Table 5 Base free oxidation of glycerol and H
2
O
2
detected.
a
Catalyst
TOF
b
(h
1
)
Selectivity
c
H
2
O
2d
mmol L
1
Glycerate
Hydroxy
pyruvic Glycolate Oxalate Formate
Au/CNF 0.5 41 17 12 0 30 1.5
Au/CNF ox 0.6 33 26 19 0 22 2.0
Au/N-CNF
473 K
1.3 30 20 22 3 23 2.0
Au/N-CNF
673 K
3.8 66 7 10 0 15 0.5
Au/N-CNF
873 K
4.7 62 8 13 - 16 0.3
a
Reaction conditions: in water, glycerol/metal: 500 mol mol
1
,T¼373 K, pO
2
¼3 atm.
b
TOF
calculated at 30 min on the basis of total metal loaded.
c
Selectivity at 20% conversion.
d
H
2
O
2
from titration at isoconversion (20%).
Fig. 5 H
2
O
2
degradation profile for Au/CNF and Au/CNF 873 K under acidic conditions (pH
2.5).
Faraday Discuss. This journal is ªThe Royal Society of Chemistry 2011
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
Moreover, under base-free experimental conditions, it was shown that secondary
alcohol can also be oxidized and hydroxypyruvic acid formed, highlighting for these
two catalysts a different selectivity towards primary and secondary OH. Au/CNF
473 K appeared the most reactive and, considering that selectivity to glyceric acid
is quite low, we could conclude that the oxidation of secondary alcohol is consecu-
tive and most probably depends on a different adsorption mode.
Additional tests have also been performed using benzyl alcohol. The reaction was
carried out in water without the addition of a base at 333 K. The increase of activity
with the increase of basicity of the support could be expected, as well as the increase
of benzoic acid, the formation of which is known to be favored by a basic environ-
ment
23
(Table 6). It should be noted, however, that, as in the case of glycerol, the
consecutive reaction producing benzoic acid from benzaldehyde is favored by an
increase in hydrophobicity of the support i.e. N-CNF 673 K and N-CNF 873 K
Table 6 Benzyl alcohol oxidation
a
Catalyst TOF
b
(h
1
)
Selectivity
c
Benzaldehyde Benzoic acid Benzyl benzoate Toluene
Au/CNF 7 92* — — 8
Au/N-CNF 473 K 65 93 2 — 5
Au/N-CNF 673 K 88 88 5 3 3
Au/N-CNF 873 K 102 85 8 3 3
a
Reaction conditions: in water, alcohol/metal: 500/1, T¼333 K, pO
2
¼3 atm.
b
TOF
calculated at 30 min on the basis of total metal loaded.
c
Selectivity at 90% conversion.
*selectivity at 40% conversion.
Scheme 2 Reaction scheme for benzyl alcohol oxidation.
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
(Scheme 2 – Table 6). We also observed a different distribution in the minor prod-
ucts of the reaction, namely benzyl benzoate and toluene. Benzyl benzoate could be
considered formed through a consecutive reaction,
7
but the formation of toluene
follows a different pathway, probably through dehydrogenation
5
(Scheme 2).
Conclusions
Alcohol oxidation in the liquid phase has been investigated, paying particular atten-
tion to the selectivity of the process and particularly to the role that the support can
play. Reactions performed in water are of interest, because of the application to
water soluble alcohol and particularly to glycerol, an important renewable feed-
stock. Commercially available CNTs and CNFs have been used as starting mate-
rials, because they did not present a relevant microporosity that could lead to
diffusional limitation during liquid phase oxidation. Functionalization of CNTs
and CNFs has been performed using the procedure set up by Arrigo et al.,
21
which
produced fully characterized supports. Preformed gold sol has been immobilized on
the supports, yielding Au catalysts with almost the same particle size, but showing
different catalytic activity/selectivity as a function of the supporting material. In
the glycerol selective oxidation it has been shown that the activity increased with
the basicity of CNFs, whereas the selectivity appeared most related to the type of
surface groups. Indeed, basic and hydrophobic surfaces enhanced the selectivity to
C3 products, whereas more hydrophilic surfaces increased the C–C bond cleavage
products. This behavior is possibly correlated to a higher concentration of native
H
2
O
2
detected in the presence of hydrophilic support.
Acknowledgements
Fondazione Cariplo is gratefully acknowledged for financial support
References
1 S. Carrettin, P. McMorn, P. Johnston, K. Griffin, C. J. Kiely and G. J. Hutchings, Phys.
Chem. Chem. Phys., 2003, 5, 1329.
2 F. Porta and L. Prati, J. Catal., 2004, 224, 397.
3 S. Demirel-Gulen, M. Lucas and P. Claus, Catal. Today, 2005, 102–103, 166.
4 W. Fang, J. Chen, Q. Zhang, W. Deng and Y. Wang, Chem.–Eur. J., 2011, 17, 1247.
5 M. Boronat, A. Corma, F. Illas, J. Radilla, T. R
odenas and M. J. Sabater, J. Catal., 2011,
278, 50.
6 T. Ishida, M. Nagaoka, T. Akita and M. Haruta, Chem.–Eur. J., 2008, 19, 8456.
7 S. Meenakshisundaram, E. Nowicka, P. J. Miedziak, G. L. Brett, R. L. Jenkins,
N. Dimitratos, S. H. Taylor, D. W. Knight, D. Bethell and G. J. Hutchings, Faraday
Discuss., 2010, 145, 341.
8 C. H. Zhou, J. N. Beltramini, Y. X. Fan and G. Q. Lu, Chem. Soc. Rev., 2008, 37, 527.
9 A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107, 2411.
10 W. C. Ketchie, Y. Fang, M. S. Wong, M. Murayama and R. J. Davis, J. Catal., 2007, 250,
94–101.
11 B. N. Zope, D. D. Hibbits, M. Neurock and R. J. Davis, Science, 2010, 330, 74.
12 N. Dimitratos, A. Villa, D. Wang, F. Porta, D. Su and L. Prati, J. Catal., 2006, 244, 113.
13 L. Prati, P. Spontoni and A. Gaiassi, Top. Catal., 2009, 52, 288.
14 A. Villa, G. M. Veith and L. Prati, Angew. Chem., Int. Ed., 2010, 49, 4499.
15 R. Arrigo, M. H€
avecker, R. Schl€
ogl and D. S. Su, Chem. Commun., 2008, 4891.
16 L. Prati and G. Martra, Gold Bull., 1999, 32, 96.
17 A. Villa, A. Gaiassi, I. Rossetti, C. L. Bianchi, K. van Benthem, G. M. Veith and L. Prati, J.
Catal., 2010, 275, 108.
18 A. Abad, A. Corma and H. Garcia, Chem.–Eur. J., 2008, 14, 212.
19 C. Lemire, R. Meyer, S. Shaikhutdinov and H.-J. Freund, Angew. Chem., Int. Ed., 2004, 43,
118.
20 J. P. Tessonnier, D. Rosenthal, T. W. Hansen, C. Hess, M. E. Schuster, R. Blume,
F. Girgsdies, N. Pfaender, O. Timpe, D. S. Su and R. Schloegl, Carbon, 2009, 47, 1779.
Faraday Discuss. This journal is ªThe Royal Society of Chemistry 2011
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online
21 R. Arrigo, M. Haevecker, S. Wrabetz, R. Blume, M. Lerch, J. McGregor, E. P. J. Parrott,
J. A. Zeitler, L. F. Gladden, A. Knop-Gericke, R. Schloegl and D. Su, J. Am. Chem. Soc.,
2010, 132, 9616.
22 A. Villa, D. Wang, P. Spontoni, R. Arrigo, D. Su and L. Prati, Catal. Today, 2010, 157, 89.
23 W. Winter, X. Xia, B. P. C Hereijgers, J. H Bitter, A. J. van Dillen, M. Muhler and K. P. de
Jong, J. Phys. Chem. B, 2006, 110(18), 9211.
24 J. Shen, R. D. Cortright, Y. Chen and J. A. Dumesic, Catal. Lett., 1994, 26(3–4), 247.
This journal is ªThe Royal Society of Chemistry 2011 Faraday Discuss.
Downloaded by Fritz Haber Institut der Max Planck Gesellschaft on 25 July 2011
Published on 18 July 2011 on http://pubs.rsc.org | doi:10.1039/C1FD00016K
View Online