The oxidative cleavage of trans-1,2-cyclohexanediol with O2: Catalysis by supported Au nanoparticles

Article abstract of DOI:10.1016/j.apcata.2018.03.019

This paper reports on the catalytic oxidative cleavage of trans-1,2-cyclohexanediol with air, catalysed by supported Au NPs, as one of the steps of a new adipic acid synthesis process. Catalysts proved to be active, with a moderate cyclohexanediol conversion and selectivity to adipic acid close to 70%. The reaction network included several steps in sequence, amongst which the key one is the oxidation of the diol into 2-hydroxycylohexanone, which is then oxidised by air – even in the absence of a catalyst – to adipic acid and lesser amounts of lighter acids, i.e. glutaric and succinic acids. The oxidation of the second hydroxyl moiety in the diol would lead to the formation of 1,2-cyclohexanedione. The latter, however, is rapidly transformed into several by-products, especially glutaric acid, under the basic conditions which are necessary for allowing the reaction to occur at an acceptable rate. With Au-based catalysts, this undesired reaction occurs much more slowly than with the previously investigated Ru hydroxide catalysts. The nature of the support, either TiO₂ or MgO, also affected catalytic performance; the best performance was shown by the Au/MgO catalyst which, however, suffered from a remarkable deactivation, found to be due to both the increase in NPs size and the formation of carbonaceous residua on the catalyst surface.

Full text of DOI:10.1016/j.apcata.2018.03.019

Accepted Manuscript
Title: The oxidative cleavage of trans-1,2-cyclohexanediol
with O2: Catalysis by supported Au nanoparticles
Authors: Stefania Solmi, Elena Rozhko, Andrea Malmusi,
Tommaso Tabanelli, Stefania Albonetti, Francesco Basile,
Stefano Agnoli, Fabrizio Cavani
PII:
S0926-860X(18)30138-8
DOI:
Reference:
https://doi.org/10.1016/j.apcata.2018.03.019
APCATA 16593
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
6-2-2018
11-3-2018
20-3-2018
Please cite this article as: Solmi S, Rozhko E, Malmusi A, Tabanelli T, Albonetti S,
Basile F, Agnoli S, Cavani F, The oxidative cleavage of trans-1,2-cyclohexanediol with
O2: Catalysis by supported Au nanoparticles, Applied Catalysis A, General (2010),
https://doi.org/10.1016/j.apcata.2018.03.019
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

The oxidative cleavage of trans-1,2-cyclohexanediol with O2: catalysis by
supported Au nanoparticles
Stefania Solmi,1 Elena Rozhko,1 Andrea Malmusi,1,2 Tommaso Tabanelli,1,2 Stefania
Albonetti1, Francesco Basile1, Stefano Agnoli,3 Fabrizio Cavani1,2
1 Department of Industrial Chemistry “Toso Montanari”, Viale Risorgimento 4, 40136
Bologna, Italy
2 Consorzio INSTM, Research Unit of Bologna, Florence, Italy
3 Department of Chemical Science, University of Padua, Via Marzolo 1, 35131 Padua, Italy
Graphical abstract
1

Highlights
 The oxidation of 1,2-cyclohexanediol can be used to synthesize adipic acid
 The reaction can be carried out using O2 as the oxidant and supported Au NPs
 The selectivity to adipic acid achieved is close to 80%
 The reaction mechanism goes through the formation of 2-hydroxycyclohexanone
 Catalyst deactivation was due to the increase of Au NPs size
Abstract
This paper reports on the catalytic oxidative cleavage of trans-1,2-cyclohexanediol with air,
catalysed by supported Au NPs, as one of the steps of a new adipic acid synthesis process.
Catalysts proved to be active, with a moderate cyclohexanediol conversion and selectivity to
adipic acid close to 70%. The reaction network included several steps in sequence, amongst
which the key one is the oxidation of the diol into 2-hydroxycylohexanone, which is then
oxidised by air – even in the absence of a catalyst – to adipic acid and lesser amounts of
lighter acids, i.e. glutaric and succinic acids. The oxidation of the second hydroxyl moiety in
the diol would lead to the formation of 1,2-cyclohexanedione. The latter, however, is rapidly
transformed into several by-products, especially glutaric acid, under the basic conditions
which are necessary for allowing the reaction to occur at an acceptable rate. With Au-based
catalysts, this undesired reaction occurs much more slowly than with the previously
investigated Ru hydroxide catalysts. The nature of the support, either TiO2 or MgO, also
affected catalytic performance; the best performance was shown by the Au/MgO catalyst
which, however, suffered from a remarkable deactivation, found to be due to both the
increase in NPs size and the formation of carbonaceous residua on the catalyst surface.
2

Keywords: Adipic acid; 1,2-Cyclohexandiol; Oxidative cleavage; Au nanoparticles;
Selective oxidation
Introduction
The oxidative cleavage of vicinal diols is an important reaction for the synthesis of carboxylic
acids (dicarboxylic acids when starting from cyclic aliphatic diols), a reaction of practical
interest in the chemical industry. Although this reaction can be carried out using conventional
stoichiometric oxidants such as periodates, in recent years the attention of researchers has
been focused on catalytic approaches and the use of greener oxidants, such as oxygen and
hydrogen peroxide (HP) [1,2].
One of the most investigated substrates is trans-1,2-cyclohexanediol (CHD), which can be
used as a starting reagent for the synthesis of adipic acid (AA) [3–6]. This reaction can be
carried out with HP starting from either CHD or even directly from cyclohexene in a one-pot
approach, by using different types of homogeneous and heterogeneous catalysts, such as
tungstate anion [7,8], Ti-Y [9], Ti-MMM-2, Ce-SBA-15, TAPO-5 and Ti-AlSBA-15 [10–12],
surfactant-type polyoxometalates or quaternary ammonium salts of tungstophosphates [13–
16], and Ti-containing sandwich-type As/W polyoxometalates [17].
Already several years ago, Brégeault and co-workers demonstrated that the aerobic catalytic
oxidative cleavage of ketones, -ketols, and -diols can be catalysed by several transition
metal ions: V, Cu, Fe, and Mn [18–22], while even more active are P/Mo/V polyoxometalates
[23–31], which may be selective catalysts especially if the reaction is carried out in the
presence of an alcohol for the in-situ esterification of the acid. Other catalysts which are
active and selective in the oxidative cleavage of diols and ketones with O2 are based on Ru in
3

the ionic form [6]. Vennat and Brégeault [32] reported that homogeneous systems with either
Ru2+/Ru3+ acetate complexes or [Ru(H2O)6]-(tosylate)2 catalyse the aerobic oxidative
cleavage of -substituted cycloalkanones into oxo-acids. RuCl3 can catalyse the oxidative
cleavage of 1-methylcyclohexene into 6-oxoheptanoic acid in acetonitrile/water medium,
under the conditions at which RuO4 is generated in situ [33,34]. Under alkaline conditions,
the oxidative cleavage of vicinal diols into the dicarboxylic acid is catalysed by Ru-based
pyrochlore oxides, already at room temperature [35,36].
The last results on CHD aerobic oxidative cleavage are reported by Obara et al., who used a
Pt/C and V2O5 combined catalyst: they were able to achieve a 90% AA yield thanks to the
oxidative dehydrogenation of CHD to 2-hydroxyclohexanone (HCHN), which was catalysed
by Pt, and to the oxidative cleavage of the intermediate performed by V [37].
In previous works, we investigated the dihydroxylation of cyclohexene to trans-1,2-
cyclohexanediol with HP [38], and the oxidative cleavage of the diol with O2, catalysed by
either Ru(OH)3 supported on alumina or Keggin type P/Mo/V polyoxometalates (POMs) [6].
Ruthenium catalysts were found to be active in a basic aqueous medium, but the reaction
network was very complex, with many side reactions and a low selectivity to AA. In these
conditions, the oxidehydrogenation of CHD generates 1,2-cyclohexanedione (CHDO), which
is mainly transformed into the undesired 1-hydroxycyclopentanecarboxylic acid (HCPA).
The dione is the key intermediate under basic conditions and its formation is due to the ability
of Ru(IV) oxo complexes to coordinate both hydroxyl groups of CHD while transforming
them into carbonyl groups at the same time [39,40]. Homogeneous POMs were more
selective in CHD oxidation to AA, but under the acidic conditions used, AA reacted with
unconverted CHD, producing the corresponding ester; therefore an additional hydrolysis step
is required to produce the acid. Conversely, it was found that if only one C-OH moiety in
CHD is oxidised to the carbonyl, the HCHN obtained can be readily oxidised to AA in a
4

basic aqueous medium, even in the absence of the catalyst [6]. Hence, a catalyst which is
capable of oxidising only one of the two C-OH groups in the diol – the second one being
oxidised later in the sequence, and not at the same time – is desirable for obtaining better
selectivity to AA in a basic medium.
According to literature, catalysts based on Au NPs can oxidise alcohols, diols, and polyols
[41]; for instance, supported Au catalysts were recently reported to be active in the oxidative
cleavage of 9,10-dihydroxystearic acid (obtained by dihydroxylation of oleic acid) to azelaic
and pelargonic acids [42–44]. One peculiarity of Au-based catalysts is that they coordinate
only one hydroxyl group in vicinal glycols. For example, depending on the conditions, Au
NPs can either oxidise the primary OH group of 1,2-propandiol to lactic acid or oxidise them
in sequence, starting from the secondary hydroxyl group and producing pyruvic acid, which
is then oxidised to acetic acid and CO2 [45].
Based on these premises, we decided to investigate the reactivity of Au NPs in CHD
oxidative cleavage to AA, placing them on two different supports, TiO2 and MgO.
Experimental
Catalyst preparation
The catalysts were prepared by the incipient wetness impregnation of a suspension of Au NPs
on supports. The suspension was prepared according to the procedure described in literature
[46]. The synthesis was conducted in water as the solvent and consisted of the reduction of
Au(III) by glucose in a basic medium of NaOH using PVP (polyvinylpyrrolidone) as the
stabilizer. The molar ratio used for the synthesis was the following: Au : glucose : NaOH :
PVP = 1 : 2 : 8 : 2.75. The concentration of the final suspension was 5∙10-3 M with a volume
of 100 ml. The synthesis started with the dissolution of the proper amount of HAuCl4 (Sigma-
Aldrich) in 10 ml of distilled water. PVP and sodium hydroxide were dissolved in 90 ml of
5

distilled water in a three-necked round-bottom flask provided with a magnetic stirrer (600
rpm), thermometer, and vapour condenser, and heated to 95°C in a glycol bath. When the
desired temperature was reached, glucose was added to the solution and the mixture was
warmed up for no more than 30 seconds to avoid excessive sugar degradation. Then the Au
solution was added, and after 2.5 minutes the reaction was stopped by fast cooling the flask in
ice and water. The final Au suspension has a dark red colour. In order to check the quality of
Au NPs, their hydrodynamic diameter was measured by means of DLS. Based on the
required weight percentage of Au on the support, the precise amount of suspension was
weighted and concentrated by centrifugation in tubes fitted with filters of cellulose (50 kDa
Amicon Ultra Filters, Millipore), at 1700 rpm for no more than 25 minutes. After the
incipient wetness impregnation of the concentrated suspension, the catalyst was dried for one
night at 120°C, then it was washed with boiling water for 30 minutes, dried again overnight
at 120°C, and finally calcined in static air at 300°C for 3 hours. Following this method,
Au/TiO2 and Au/MgO (both with Au loading 1.5 wt%) were prepared.
The TiO2 support was a commercial Cristal Active DT-51 (Millenium Chemicals) sample,
100% anatase (see XRD pattern in Figure S1, Supporting Information). BET surface area was
903 m2/g.
MgO was synthesized by means of a precipitation method in which Mg(NO3)2*6H2O
dissolved in water was used as metal precursor (1 mol/L in distilled water), and the solution
was then added dropwise to a second solution of Na2CO3*10H2O (1 mol/L in distilled water),
under magnetic stirring, at 60°C. The pH was maintained between 9.8 and 10.2 and
monitored with an Amel Instrument 380, calibrated with two buffers at pH 7 and 13. The
obtained magnesium hydroxide was filtered, washed with distilled water and dried overnight
at 110°C. The material was then crushed and calcined at 450°C for 5 hours. The crystalline
6

phase of MgO was verified by XRD (see Figure S1 in Supporting Information). BET surface
area was 1645 m2/g.
Catalyst characterisation
Dynamic Light Scattering (DLS)
The hydrodynamic diameter of the particles in the suspension was measured by DLS analysis
(Dynamic Light Scattering), using a Zetasizer Nanoseries (Malvern Instruments) to check the
quality of the suspension synthesis. The hydrodynamic diameter is the diameter of the
particles in motion in a solution that includes the coordination sphere and any adsorbed
species on its surface (e.g. the stabilizer). Before the analysis, 10 drops of samples were
diluted in 10 ml of water to avoid interactions amongst the particles that could change the rate
of diffusion and therefore the size estimation.
Transmission Electron Microscopy
TEM analyses were performed with a TEM/STEM FEI TECNAI F20 microscope equipped
with a Schottky emitter and operating at 200 KeV. The instrument is fitted with an Energy
Dispersive X-Ray Spectrometer (EDX). Samples were suspended in ethanol and treated with
ultrasounds for 15 minutes. A drop of the suspension was deposited on a multifoil holey-
carbon film supported by a copper grid. The sample was dried at 100°C. TEM images were
used to calculate the particle size; values were averaged over approximately 100 particles.
Metal dispersion was calculated using the equation [47]:
⁄
(푣푚 푎푚)
퐷 = 6 ∙
푑푉퐴
where:
D: dispersion
7

vm: volume occupied by an atom m in the bulk of metal;
am: surface occupied by an atom m on a polycrystalline surface;
dVA: mean particle size (obtained by TEM analysis);
Value of vm and am for Au atoms were obtained from ref [47, pag.741].
BET analysis and porosimetry
The porosimetry analysis was performed with an ASAP 2020 Micromeritics apparatus
measuring N2 adsorption/desorption isotherms at -196°C. Adsorption/desorption isotherms
for Au/TiO2 and Au/MgO catalysts are shown in Figure S2, in Supporting Information.
Metal loading analysis
Catalyst metal loading was measured by atomic absorption spectroscopy using a Thermo
Scientific iCE 3000 Series, AA Spectrometer. Samples were treated with boiling aqua regia
to dissolve gold nanoparticles. Analyses of Au loading were performed by measuring
absorbance at 242.8 nm, with an air-acetylene flame.
Catalyst metal loading was also determined with a wavelength dispersive X-Ray
Fluorescence Spectrometer (Panalytical Axios Advanced), equipped with a Rh anode as the
X-Ray source. Measurements were carried out under a vacuum on samples in the form of
pellets prepared by adding 100 mg of an organic wax to 300 mg of catalyst.
X-Ray Photoelectron Spectroscopy (XPS)
X-ray photoemission spectra have been acquired using a VG Escalab Mk II energy analyser
mounted on a custom designed ultra-high vacuum systems operating at a base pressure of 10-9
mbar. Thin films of catalyst powders have been prepared by drop casting water suspensions
on electropolished copper supports. All measurements have been taken at room temperature
8

after degassing the samples overnight using a non monochromatized Mg K source (1253.6
eV) and setting the analyser pass energy to 20 eV.
Catalytic tests
Reactants (trans-1,2-cyclohexanediol, 2,hydroxycyclohexanone, and 1,2-cyclohexanedione),
main products (adipic acid, glutaric acid and succinic acid), and minor by-products (see Table
2) were purchased from Sigma-Aldrich. Reactions were conducted in a glass semi-continuous
reactor (volume 100 ml, model Büchi Miniclave) supported on a rigid metal grid, closed with
a steel cap and equipped with gas lines. CHD (0.5 g), the catalyst, the base, and the solvent
(water, 25 ml) were loaded into the reactor and heated. When the desired temperature was
reached, the oxygen valve was opened and the reaction started (oxygen flow 100 ml/min,
pressure 4 bar). At the end of the reaction, the oxygen valve was closed and the reactor was
cooled down. The catalyst was separated by centrifugation, the supernatant was acidified with
85% H3PO4 to pH between 2 and 3, and the final volume was measured. Quantitative
analyses were conducted by means of a HPLC Agilent 1260 Infinity Series Quaternary LC
provided with Agilent Technologies Poroshell 120 EC-C18 4.0 μm (4.6 x 250 mm) operating
at 25°C, a manual injector (loop 20 μl), and a Diode-Array Detector set at 254, 210, and 192
nm. We used the DAD detector because all the main chemicals involved in the reaction
(substrate and products) are UV-active; this detector allows a faster quantification compared
to the Refractive Index Detector (RID), which typically requires longer time for stabilization.
Elution was performed in isocratic conditions, with a mobile phase composed of H3PO4 0,01
M : CH3CN 95:5 v/v ratio, flow rate 0,800 ml/min. An example of the HPLC chromatogram
is shown in the Supplementary Information (Figure S3).
In some cases, qualitative analyses were also performed by means of ESI-MS spectroscopy.
9

Conversion of reactants, XA (A= CHD,
2-hydroxycyclohexanone HCHN, 1,2-
cyclohexanedione CHDN) has been calculated by means of the formula:
푛퐴,푖 − 푛퐴,푓
푋퐴 =
∙ 100
푛퐴,푖
where na,i is the initial number of moles of A, nA,f is the number of A moles at the end of the
reaction.
Yield to products YR has been calculated using the formula:
푛푅,푓
푌 =
푅
∙ 100
푛퐴,푖
where nR,f is the final number of moles for product R.
Results and discussion
Au/TiO2 catalyst
The catalyst was characterised by TEM microscopy in order to determine nanoparticle size
distribution and dispersion (Figures 1 and 2).
20
16
12
8
4
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Particle diameter (nm)
Figure 1. Size distribution of Au NPs calculated from TEM analysis.
10

Figure 2. TEM pictures of Au/TiO2 catalyst (at increasing magnification)
The average nanoparticle size was 7.4 nm, while dispersion was 16%. AAS gave an effective
metal loading of 1.20.1 wt%. Porosimetry analysis yielded the following results: BET
surface area 86 m2/g, pore size 14.7 nm, total volume of pores: 0.18 cm3/g. Table S1
summarizes main features of fresh catalyst.
Figure 3 shows the effect of reaction time on the catalytic behavior of the Au/TiO2 catalyst;
the conditions chosen were: CHD : Au : NaOH = 1000 : 1 : 1000 (molar ratios); temperature
90°C, oxygen pressure 4 bar, oxygen flow 100 ml/min (these conditions were chosen after
the results obtained by changing reaction parameters, see discussion below). In fact, the
direct oxidation of alcohols into acids is usually performed under basic conditions [48–57].
11

50
40
30
20
10
0
CHD
AA
GA
SA
0
5
10
15
20
25
Time (h)
Figure 3. CHD conversion (♦) and yields to AA (), GA (), SA (X) based on reaction time
with Au/TiO2. Conditions: CHD : Au : NaOH = 1000 : 1 : 1000 (molar ratios); temperature
90°C, oxygen pressure 4 bar, oxygen flow 100 ml/min. Starting pH was 12.5.
The products of reaction were adipic acid (AA), glutaric acid (GA) and succinic acid (SA).
The formation of 2-hydroxycylohexanone (HCHN) and 1,2-cyclohexanedione (CHDO) was
observed also, with yields lower than 1% in both cases; these two latter compounds were
found to be the key reaction intermediates in CHD oxidation with Ru(OH)x/Al2O3 catalysts
(under similar reaction conditions as those used with supported Au NPs), and P/Mo
polyoxometalates [6]. Other by-products were shown from the HPLC plot, falling within the
region of lighter dicarboxylic acids (propanedioic acid, oxalic acid and carbonic acid), but
were not individually quantified; by taking into account the good C balance registered and the
error in conversion calculation, it is possible to assume that the overall yield to these lighter
compounds was increasing slightly over time, but was never higher than 5%.
CHD conversion and AA yield increased when the reaction time was increased, but then they
slowed down after 3 hours, as indicated by the TOF listed in Table S2 (in Supporting
Information). This could be due to some deactivation phenomena of the catalyst as shown by
recyclability tests (see results below). It is likely that during the reaction, some products were
12

adsorbed on the surface, thus blocking active sites; on the other hand, during recyclability
experiments catalysts were washed before each reloading, thus setting Au NPs almost free
from weakly interacting adsorbed species.
This hypothesis was confirmed by performing ATR analysis on the support and the fresh,
used, and used + washed catalyst. As can be seen in Figure S4 (in Supporting Information),
new bands appeared on the used catalyst, whereas either their intensity was reduced or they
disappeared after the hot water washing. Looking at the wavenumber, these signals are
compatible with reaction products with –C=O and –COOH groups.
HCHN and CHDO were the only products identified at the lowest reaction time (0.1 h); this
confirms a possible role of these compounds as reaction intermediates. On the other hand, the
formation of 1-hydroxycyclopentanecarboxylic acid (HCPA) was not detected, while it was
one of the prevailing products in CHD oxidation with Ru(OH)x-based catalysts [6]. This
compound is very quickly formed under basic conditions by the rearrangement of CHDO,
which is the main intermediate with supported Ru(OH)x catalysts. This suggests that: (a) with
Au/TiO2, the oxidative cleavage indeed occurs on HCHN, leading to the formation of
dicarboxylic acids, and (b) the side oxidation leading to CHDO, which may also undergo
oxidative cleavage, gives a marginal contribution to the formation of final products.
Dedicated experiments carried out by directly reacting HCHN and CHDO confirm this
hypothesis (see below).
GA and SA yields were very low; even though their formation was not detected at the lowest
reaction time examined (0.1 h), the trend of the yields was not as expected when based on a
consecutive reaction network; in fact, GA yield initially increased but then reached a stable
value close to 3%, while SA yield reached 2% and then descended to 1%. Furthermore, AA
yield increased continuously and apparently did not undergo any consecutive oxidative
degradation during the experiment time. This was confirmed by performing a reaction
13

starting from AA, in the presence of a catalyst under the same conditions chosen for the
reaction from CHD (CHD : Au : NaOH = 1000 : 1 : 1000 (molar ratios); temperature 90°C,
oxygen pressure 4 bar, oxygen flow 100 ml/min), which showed nil conversion and nil
formation of GA and SA. Thus it seems that the formation of the three main dicarboxylic
acids followed three kinetically independent, parallel reaction pathways; the hypothesized
intermediates – HCHN, CHDO, or both – were so reactive (as it is shown by the dedicated
experiments, see below), that they were quickly transformed into successive products;
therefore, dicarboxylic acid yields showed a dependence upon time which is that typically
observed for kinetically primary products.
Experiments were then conducted with the aim of verifying the effective heterogeneity of the
Au active phase. A leaching test was performed by first carrying out a 3 h reaction (molar
ratio CHD : Au : NaOH = 1000 : 1 : 1000; T = 90°C); then the catalyst was removed by hot
filtration and the mixture was reloaded into the vessel for another 3 h of reaction without the
catalyst. The absence of Au in the filtered solution was verified by means of X-Ray
Fluorescence. No important change in CHD conversion (from 25  3 to 27  3%) or yields of
AA (from 15 to 16%) or other products was seen; therefore, it is possible to rule out any
contribution from homogeneous Au deriving from the loss of active phase during the
reaction.
Using the same operating conditions, recyclability tests were performed in order to
investigate if the catalyst maintained its initial activity. After a 3 h reaction, the catalyst was
separated from the reaction mixture, washed with water, dried at 120°C overnight, and
reloaded with fresh reagents for another 3 h. This procedure was repeated four times, and
results are shown in Figure 4.
14

35
30
25
20
15
10
5
70
60
50
40
30
20
10
0
CHD
AA
GA
SA
S AA
0
1
2
3
4
5
Number of uses
Figure 4. CHD conversion, yield to AA, GA, and SA based on the number of uses.
Conditions: CHD : Au : NaOH = 1000 : 1 : 1000 (molar ratios); temperature 90°C, oxygen
pressure 4 bar, oxygen flow 100 ml/min, reaction time 3 h.
It was shown that both CHD conversion and AA yields decreased with the number of uses,
even though the selectivity to AA was almost constant. This loss of activity could be due to
the average nanoparticle size growing with the number of uses, as shown by TEM analysis
(Figure S5); in fact, fresh catalyst average dimension was 7.4 nm, while after the first use it
was 7.8 nm, and after the fifth 9.1 nm.
XPS characterisation of used (after 5 uses) catalyst also revealed that under the conditions
employed no important changes in the Au oxidation state had occurred, compared to the fresh
one (Figure S6). In fact, in both fresh and used catalysts Au appeared to be completely
reduced to the metallic state, with no presence of Auδ+ species. It is worth noting that also in
the case of glucose oxidation to glucaric acid with a catalyst based on Au/AC [58], under
conditions similar to those used in the present experiments, Au appeared to be completely
reduced in the used catalyst.
15

Table 1 sums up the results obtained by changing certain reaction parameters, such as the
ratio between substrate and catalyst (entries 1-3), the ratio between substrate and NaOH
(entries 4-8), temperature (entries 9-11), and oxygen pressure (entries 12-14). As expected, it
was seen that an increase of the CHD : Au ratio led to a decline in conversion and yields,
passing from a CHD : Au molar ratio of 1000:1 to 5000:1, however, the change from 500:1 to
1000:1 did not cause any further modification of the performance. The selectivity to AA,
however, was not much affected by the conversion degree (taking into account the
experimental error in the calculation of CHD conversion), thus further confirming that the
reaction network consists of parallel reactions for the formation of the main dicarboxylic
acids. The ratio CHD : Au = 1000 : 1 was chosen for further experiments (as in tests reported
in Figures 3 and 4). With regard to the effect of NaOH, it was shown that in the absence of
the base, CHD conversion was negligible, and no formation of dicarboxylic acids was
observed; the only products formed in very low yields were HCHN (yield 1.1%) and CHDO
(yield 0.1%). The presence of the base led to an increased CHD conversion up to a maximum
value shown for a CHD : NaOH ratio equal to 1:1. A further increase led to a decline of both
CHD conversion and AA yield, but also to an increase in GA and SA yields. The optimal
ratio CHD : NaOH 1:1 (also used for experiments in Figures 3 and 4) was chosen for
additional reactivity tests. Experiments carried out with changes in temperature (entries 9-11)
showed the expected increased conversion and yields along with the temperature raise.
Finally, the increased oxygen pressure (entries 12-14) gave 4 bar as the optimal value (also
used for experiments in Figures 3 and 4) producing the best condition for obtaining the
highest AA selectivity and yield.
To confirm the role of the two possible intermediates, HCHN and CHDO, we conducted
experiments using these compounds as reactants. Tests were carried out at 90°C for 3 h,
under an oxygen flow of 100 ml/min and 4 bar pressure (in one case without oxygen), and in
16

the presence or absence of the Au catalyst. Moreover, since the pH of the solution is an
important parameter in determining the reactivity of these compounds, tests were also carried
out by changing the amount of NaOH added. Results are shown in Table 2.
Table 1. Results of experiments conducted based on reaction parameters. For all experiments,
reaction time was 3 h. X: conversion, Y: yield, S: selectivity.
T
(°C)
90
O2 pressure CHD:Au:NaOH CHD X AA Y GA Y SA Y AA S
Entry
(bar)
4
(molar ratios)
500:1:500
(%)
293
283
92
(%)
17
18
6
(%)
4
(%)
2
(%)
59
64
67
0
1
2
90
4
1000:1:1000
5000:1:5000
1000:1:0
3
1
3
90
4
1
1
4
90
90
90
90
90
60
80
90
90
90
90
4
0
0
0
32
5
4
1000:1:500
1000:1:1000
1000:1:1500
1000:1:2000
1000:1:1000
1000:1:1000
1000:1:1000
1000:1:1000
1000:1:1000
1000:1:1000
132
283
273
233
72
10
18
17
13
3
0.4
3
1
77
64
63
57
43
73
64
57
64
52
6
4
1
7
4
5
2
8
4
6
3
9
4
0.1
1
0.2
0.8
1
10
11
12
13
14
4
152
283
302
283
253
11
18
17
18
13
4
3
2
2
1
4
3
1
6
3
3
Table 2. Reactivity of intermediates HCHN and CHDO. Substrate: Au 1000:1; temperature
90°C, reaction time 3 hours.
Other
products
HCPA AA Y GA Y SA Y
Entry Reactant
pH
Catalyst X (%)
from
Y (%)
(%)
(%)
(%)
HPLC and
ESI-MS
17

CHDO
dil
Na
Na
Na
Na
1
2
> 13
> 13
>13
12.5
7
No
No1
Yes2
Yes
No
100
100
99
38
61
0
0
0
28
0
0
0
CHDO
dil
CHDO
dil
3
14
6
36
19
21
24
47
15
5
5
CHDO
dil
4
99
0
2
CHDO
dil
CP, HCL,
AAN
5
76
0
0
0
CHDO
dil
Na
6
7
Yes
Yes2
No
76
0
8
1
CHDO
dil
Na
7
7
64
0
9
1
HCHN
dil
CHDO,
HCL
8
>13
>13
>13
100
100
99
52
0
20
60
73
11
4
HCHN
dil
HCL
9
Yes
Yes3
HCHN
dil
Na
10
0
7
3
CP, HCL,
AAN,
HCHN
dil
11
12
13
12.5
12.5
12
No
No
100
99.8
99.8
0
64
18
14
2
5
5
5
9
2
CHDO
HCL,
HCHN
conc
51
56
CHDO
HCL,
HCHN
conc
Yes
AAN,
CHDO, CP
HCL,
HCHN
conc
14
15
16
17
11.5
11.5
7
No
Yes
No
87
85
34
40
18
20
0
24
28
2
0
0
0
1
0
0
0
3
CHDO, CP
HCL,
HCHN
conc
CHDO, CP
CHDO
HCHN
conc
HCHN
conc
HCL,
7
Yes
0
7
CHDO
18

Legend: dil: 0.1 g of reactant in 25 ml H2P; conc: 0.5 g of reactant in 25 ml H2O, 1: No
2
oxygen flow and Patm; : CHDO: Au 100:1; 3: HCHN:Au200:1; CP: condensation products;
HCL: 6-hydroxycaprolactone; AAN: adipic anhydride; Na=not available.
Both HCHN and CHDO were much more reactive than CHD, which confirms the hypothesis
as to the reaction network and their role as possible intermediates. In diluted conditions, in
the absence of the catalyst, and under strongly basic pH (pH > 13.5, entry 1), CHDO was
converted into HCPA, with 38% yield; this compound was never observed in usual reactivity
experiments from CHD. A considerable yield to GA was also shown (28%), but did not form
in the absence of O2 (entry 2). At lower pH (12.5, entry 4), at conditions closer to those used
for experiments with CHD, and in the presence of the catalyst, no HCPA formed, while AA
was detected; however, GA was still the prevailing product. The same occurred at neutral pH
(entries 5-7); GA yield always prevailed over AA. This demonstrates that CHDO cannot be
the key reaction intermediate in CHD oxidation with the Au/TiO2 catalyst; in fact, the
cleavage of CHDO leads mainly to GA, whereas from CHD the latter is only a minor by-
product. Worthy of note, with supported Ru(OH)x – a much less selective catalyst than the
Au-based ones reported here – experimental evidence was in favour of CHDO as the main
reaction intermediate [6].
When the reaction was carried out starting from HCHN in diluted conditions, at strongly
basic pH and without a catalyst (entry 8), the formation of HCPA (a compound which forms
from CHDO) was seen again, thus indicating that the transformation of HCHN to CHDO was
very rapid. Unlike what was seen from CHDO, however, significant amounts of AA, GA and
SA were also formed. This means that even though CHDO undergoes further cleavage to GA,
the direct cleavage of HCHN (which also occurs in the absence of the catalyst) leads mainly
to AA. In the presence of the catalyst (entry 9), the cleavage of HCHN to AA (with minor
formation of GA and SA) was greatly accelerated compared to the competitive reaction of
19

HCHN oxidation to CHDO, and a high yield to AA was seen. The key role of the catalyst is
also demonstrated by the fact that the test carried out with a higher amount of catalyst led to a
73% yield to AA (entry 10). When a lower pH was used, under conditions at which the
formation of CHDO and HPCA was less probable (entry 11), a very high yield to AA was
still seen, even without a catalyst. Results showed us that AA formation occurs selectively
from HCHN also in the absence of the catalyst, but a very basic pH should be avoided in
order to limit the competitive oxidation of HCHN to CHDO, which is a precursor for GA
formation. However, the presence of the Au catalyst accelerates HCHN cleavage compared to
its oxidation to CHDO.
Similar reactions starting from HCHN, but in more concentrated initial conditions, with and
without a catalyst, were carried out varying the pH (entries 12-17). Compared to the diluted
conditions, HCHN conversion and AA yield were lower, producing results more similar to
those obtained starting from CHD and to those published by Obara et al. [37] as well. This
indicates that: (a) HCHN formation was very rapid in the reaction medium and a relatively
high concentration of it was achieved, but also (b) its transformation to subsequent products
was fast since HCHN was detected in a yield lower than 0.1%, while no HCPA was produced
in the reaction starting from CHD. Nevertheless, a lower HCHN concentration can lead to
higher AA selectivity and avoid HCPA formation in the presence of the catalyst if the
basicity of the reaction medium is not excessive.
It is possible to conclude that HCHN is the key intermediate compound in CHD aerobic
oxidation with the Au-based catalyst. CHDO is a reaction intermediate which minimally
contributes to AA formation, leading to several side reactions; its formation, however, is
limited because of the less basic conditions necessary for the oxidation with Au NPs
compared to Ru(OH)x.
20

In regard to the mechanism through which HCHN is cleaved to AA, two different hypotheses
may be formulated. One mechanism involves the hydration of the carbonyl to yield the
geminal glycol, which then rearranges under oxidising conditions by losing two H atoms with
the breaking of the C-C bond and the formation of 6-oxohexanoic acid (precursor of AA) and
H2O (Figure 5). This mechanism, similar to that recently proposed for oleic acid oxidative
scission [51], might explain the parallel formation of AA, GA, and SA by considering that the
cleavage may indeed occur in different positions. If it occurs at the C(OH)-C(O) bond, it
gives rise to the formation of AA. However, a concomitant cleavage at both the latter bond
and one of the two vicinal C-C bonds (either C-C(O) or C-C(OH)) might lead to the
formation of GA with the loss of one C atom in the form of CO2, whereas the concomitant
cleavage of C-C(O) and C-C(OH) bonds would lead to the direct formation of SA with the
loss of two CO2 molecules. The lower probability associated with the concomitant cleavage
of two C-C bonds instead of one only would explain the very low selectivity to GA and SA
which was experimentally observed. As for the oxidation of CHDO, which leads selectively
to GA, this might be attributed to the fact that because of the high reactivity of the diketone,
the concomitant cleavage of two vicinal C-C bonds (C(O)-C(O) and C-C(O)), with release of
CO2, occurs faster than the cleavage of the C(O)-C(O) bond alone.
21

Main pathway
Figure 5. First hypothesis for the mechanism of CHD and HCHN oxidation to AA with
Au/TiO2 catalyst.
In an alternative hypothesis (Figure 6), the mechanism might involve the reaction of a radical
at the C atom holding the -OH moiety in the intermediate HCHN (whose formation would be
facilitated by the delocalisation over the vicinal carbonyl) with O2, to yield an hydroxo-
peroxo species and then a hydroxo-hydroperoxo species by a H abstraction (resembling the
Criegee intermediate obtained by hydroperoxidation of carbonyls), which would then
rearrange into adipic anhydride; the latter would be finally hydrolysed to yield AA. The GA
formation would take place starting from CHDO (confirming experimental results), since an
intramolecular rearrangement in basic conditions would lead to 6-hydroxycaprolactone
(HCN). The latter might either oxidehydrogenate to produce adipic anhydride, or develop a
radical species whose reaction with O2 would again generate a hydroxo-hydroperoxo species.
The further rearrangement of this latter compound and the subsequent hydrolysis might lead
to the release of CO2 and to the formation of 5-hydroxy-pentanoic acid, a precursor of GA.
22

Figure 6. Second hypothesis for the mechanism of CHD and HCHN oxidation to AA with
Au/TiO2 catalyst.
In addition to the results obtained by reacting the possible intermediates (HCHN and CHDO),
another piece of experimental evidence in favour of the second mechanism is the finding that
both 6-hydroxycaprolactone (HCL) and adipic anhydride (AAN) were detected amongst
reaction products (but not quantified) when experiments were conducted starting from HCHN
and CHDO (see Table 2).
Moreover, a possible role of in-situ generated HP cannot be excluded [59], which might
generate the Criegee-type intermediate by reaction with the carbonyl bond in HCHN [60].
Au/MgO catalyst
Experiments with Au/TiO2 showed that a controlled basicity in the reaction media is
necessary in order to activate one of the hydroxyl groups of the substrate and avoid side
reactions. Therefore, we decided to support Au NPs on MgO in order to verify if the basicity
of the oxide was sufficient for CHD activation.
23

TEM analysis (Figure 7 and 8) of the Au/MgO gave an average nanoparticle diameter of 7.9
nm and a dispersion equal to 15%.
20
18
16
14
12
10
8
6
4
2
0
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20
Particle diameter (nm)
Figure 7. Nanoparticle size distribution of Au/MgO catalyst.
24

Figure 8. TEM pictures of Au nanoparticles of Au/MgO catalyst (at increasing
magnification).
AAS analysis gave an effective metal loading of 1.50.1 wt%. Porosimetry analysis produced
the following results: BET surface area 71 m2/g, pore size 14.9 nm, total volume of pores:
0.09 cm3/g. Table S1 summarizes main catalyst features.
Starting from the optimal reaction conditions determined with the Au/TiO2 catalyst, the effect
of the CHD : NaOH molar ratio was investigated. Without the addition of a base, no AA
formation was seen: a clear indication of the fact that MgO basicity is not sufficient to
activate the alcohol. After the addition of NaOH to the reaction mixture, AA formed, and its
amount increased as the basicity increased (CHD conversion 31%, AA yield 18%), so
additional experiments were carried out with CHD : NaOH molar ratio equal to 1:1 (pH
11.5), the same used for the Au/TiO2 catalyst also.
The catalytic performance of Au/MgO was studied based on the reaction time in the same
conditions as those used for Au/TiO2, i.e. molar ratio CHD : Au : NaOH = 1000 : 1 : 1000,
90°C, oxygen pressure 4 bar, oxygen flow 100 ml/min. Results are shown in Figure 9. Trends
for conversion of CHD and product yields are similar to those obtained with the Au/TiO2
catalyst, so it is possible to conclude that the reaction pathway is the same for both catalysts.
Au/MgO, however, was slightly more active and more selective than Au/TiO2: after 24 h for
Au/MgO, CHD conversion, AA yield, and AA selectivity were 45%, 39%, and 87%
respectively, while for Au/TiO2 the corresponding values were 45%, 33%, and 74%.
Also in the case of MgO, TOF decreased from 101 h-1 down to 16 h-1 after 24 h of reaction.
Thus experiments aiming at checking the stability of the catalyst with leaching and
recyclability tests were performed, in order to better understand the observed phenomenon.
Dedicated experiments did not show any active phase dissolution since the conversion of
25

CHD and AA yield did not increase during the reload of the reaction mixture after catalyst
filtration (CHD conversion constant at 27% and AA yield from 18 to 19%). For the reuse test,
the catalyst was made to react three times for 3 hours each, and after every use it was washed
with cold water and dried at 120°C for one night. CHD conversion and AA yield decreased
after every use, indicating the presence of a deactivation phenomenon (Figure 10), while GA
and SA yields did not seem to be affected. A similar phenomenon, but much less significant,
was also seen with Au/TiO2 (Figure 4).
50
40
30
CHD
AA
20
GA
SA
10
0
0
5
10
15
20
25
Time (h)
Figure 9. CHD conversion (♦) and yields to AA (), GA (), SA () based on reaction time
for Au/MgO. Conditions: CHD : Au : NaOH = 1000 : 1 : 1000 (molar ratios); temperature
90°C, oxygen pressure 4 bar, oxygen flow 100 ml/min.
26

35
30
25
20
15
10
5
100
90
80
70
60
50
40
30
20
10
0
CHD
AA
GA
SA
S AA
0
1
2
3
After
calcination
Number of uses
Figure 10. CHD conversion, yield to AA, GA and SA based on the number of uses of
Au/MgO. Conditions: CHD : Au : NaOH = 1000 : 1 : 1000 (molar ratios); temperature 90°C,
oxygen pressure 4 bar, oxygen flow 100 ml/min, reaction time 3 h.
TEM analysis of the used catalyst (Figure 11) revealed, on the surface, the presence of
filaments of carbonaceous material that could not be removed by simple cold water washing,
but could be burnt after a thermal treatment in the air at 300°C for 3h. Moreover, nanoparticle
size increased already after the first use, from 7.9 to 9.3 nm. Also after calcination at 300°C,
the catalyst did not recover its initial reactivity: after another 3 hours of reaction, CHD
conversion and AA yield were 10% and 8.5% respectively (Figure 10). In fact, the thermal
treatment caused an increase of Au NPs dimensions up to an average value of 10.3 nm.
It may be concluded that even though Au/MgO offers better selectivity to AA when
compared to Au/TiO2, the former deactivates more rapidly than the latter, due not only to the
increased average particle size, but also to the formation of carbonaceous residua. The latter
may form from compounds which interact strongly with the MgO support, and are finally
transformed into residues, despite the oxidising conditions.
27

Figure 11. TEM pictures of Au/MgO catalyst used 3 times.
Conclusions
Previous studies with Ru-based catalysts and POMs for the oxidative cleavage of CHD
showed that, in order to carry out the reaction with good selectivity, it is necessary to use a
catalyst capable of oxidising hydroxyl groups in sequence, and not at the same time. The aim
is to obtain 2-hydroxycyclohexanone (HCHN) which can be oxidised to AA with high
selectivity, even in the absence of a catalyst, in the basic medium, a slow oxidation of HCHN
into 1,2-cycohexanedione (CHDO) being a requisite to achieve high selectivity to AA.
Au/TiO2 and Au/MgO catalysts turned out to be able to oxidise CHD to HCHN, obtaining
AA with a high selectivity and working in less basic conditions than Ru(OH)x/Al2O3. GA
and SA were the minor products of the reaction formed by parallel reaction pathways, and
28

both catalysts were effectively heterogeneous. The basic character of MgO was not sufficient
to provide the basicity needed, but with Au/MgO, AA formed with a higher selectivity than
Au/TiO2 (87 and 74% respectively), at a moderate CHD conversion. This characteristic may
be due to a different interaction between the catalyst and CHD; this aspect is an interesting
subject for further investigation.
Both catalysts showed deactivation during the reaction, while AA selectivity remained almost
constant. TEM analysis of used catalysts revealed an increase in the nanoparticle sizes
compared to the fresh catalyst; moreover, in the case of MgO, carbonaceous filaments were
deposited on the support surface and could be removed only after calcination at 300°C.
However, the thermal treatment also led to an increased Au NPs size, which was the reason
for the poor performance of the recovered catalyst.
Two different hypothesis for the reaction mechanism have been proposed, both involving
HCHN as the main reaction intermediate. HCHN can be either transformed to adipic
anhydride, in the same way as recently proposed for CHD oxidation with supported Ru(OH)x
catalyst [6], or undergo ring-opening and further oxidation to AA. Experiments carried out
starting from HCHN support the first mechanism as the prevailing one.
Acknowledgements
Consorzio INSTM is gratefully acknowledged for financing AM’s PhD grant.
29