Daniil S.
Ovoshchnikov‡
a,
Baira G.
Donoeva‡
ab,
Bryce E.
Williamson
a and
Vladimir B.
Golovko
*ab
aDepartment of Chemistry, University of Canterbury, Christchurch 8140, New Zealand. E-mail: vladimir.golovko@canterbury.ac.nz; Fax: +64 3 364 2110; Tel: +64 3 364 2442
bThe MacDiarmid Institute for Advanced Materials and Nanotechnology, New Zealand
First published on 17th December 2013
The selectivity of supported gold catalysts in aerobic oxidation of cyclohexene under solvent-free conditions without addition of a radical initiator was tuned by either WO3 or the metal–organic framework MIL-101, used as a support/co-catalyst. WO3 was found to promote the formation of cyclohexene oxide via reaction of cyclohexenyl hydroperoxide with cyclohexene, while MIL-101 catalysed conversion of cyclohexenyl hydroperoxide to 2-cyclohexen-1-one.
Since the discovery of the exceptional catalytic activity of gold in low-temperature oxidation of CO by Haruta6 and in hydrochlorination of ethylene by Hutchings,7 gold catalysts have proven to be efficient in a number of different reactions: cyclizations, rearrangements, selective hydrogenation, C–C coupling reactions, and selective oxidation of alcohols and olefins.8–10 Among other reactions, aerobic oxidation of cyclohexene was shown to be catalysed by supported nanosized gold.11–14 However, the reaction required the use of t-butyl hydroperoxide as a radical initiator,13 and the high yield of cyclohexene oxide was achieved only by conducting the reaction in a specific solvent.11 Under solvent- and radical initiator-free conditions, mainly products of allylic oxidation are formed.14,15
One of the approaches to improve the selectivity of the catalysed reaction towards the desired product is to utilize a bifunctional catalyst.16 An elegant example of such an approach is the phenol hydrogenation process catalysed by palladium deposited on Al2O3: while the palladium nanoparticles activate the hydrogen, the support, being a Lewis acid, activates the substrate and stabilizes cyclohexanone, preventing its over-hydrogenation to cyclohexanol.17
Recently, we showed that metallic gold nanoparticles formed from triphenylphosphine stabilized Au clusters catalyse formation of cyclohexenyl hydroperoxide in the aerobic oxidation of cyclohexene.15 In the present study, we investigated the effects of the supports and co-catalysts on the activity and selectivity of the Au-based catalysts for aerobic oxidation of cyclohexene under solvent-free conditions, without the addition of a radical initiator. We showed how the selectivity of this reaction can be tuned towards formation of either cyclohexene oxide or 2-cyclohexen-1-one as major products.
As-made gold clusters were deposited onto the oxide supports from CH2Cl2 solution. Typically, a calculated amount of gold cluster dissolved in CH2Cl2 (10 mL, 1–2 mg mL−1) was added dropwise to a vigorously stirred slurry of SiO2, TiO2 or WO3 (500 mg) in CH2Cl2 (15 mL). The mixture was stirred for 30 min and the solid was collected by centrifugation. A colourless supernatant solution confirmed complete cluster deposition. The catalysts were washed with CH2Cl2 (20 mL) and dried under vacuum at room temperature.
In the case of deposition of Au9 on WO3, for loadings greater than 0.1 wt%, n-hexane (50 mL) was slowly added to the slurry to ensure cluster deposition. Solids were collected by centrifugation, washed with n-hexane and dried under vacuum at room temperature.
The metal–organic framework MIL-101 was synthesized following a previously described method20 and characterized using PXRD, TGA and surface area measurements (see the ESI†).
The liquid samples were analysed via gas chromatography (GC) using a Shimadzu GC-2010 equipped with an Rxi-5SilMS capillary column (30 m × 0.25 mm × 0.25 μm) and a flame ionization detector (FID). Products were identified by gas chromatography-mass spectrometry (GC-MS) using a Shimadzu GCMS-QP2010. Quantitative analysis of reaction mixtures was performed via GC-FID using calibration solutions of commercially available products. The concentration of cyclohexenyl hydroperoxide was determined using iodometric titration in 80% acetic acid.21 Pre-treatment of the reaction mixture with PPh3 (ref. 22) did not change the concentration of 2-cyclohexen-1-one compared to an untreated mixture, indicating that cyclohexenyl hydroperoxide was not decomposing during GC analysis and therefore allowing us to plot a calibration curve for cyclohexenyl hydroperoxide on the basis of comparison of GC data and iodometric titration results.
Two series of catalysts with target gold loadings ranging from 0.1 wt% to 0.5 wt% were prepared. The actual loadings, established using AAS, are shown in Table 1, where the two-digit prefixes in the sample designations are indicative of the approximate loadings in wt%. It was found that clusters readily adhere from CH2Cl2 solution to SiO2 and TiO2 at all target loadings.
Catalyst | Gold loading, AAS (wt%) | Nanoparticle diameter, TEM (nm) | |
---|---|---|---|
Mean ± s.e.f | s.d.f | ||
a Recovered after the 1st catalytic cycle. b Recovered after the 2nd catalytic cycle. c Recovered after the 3rd catalytic cycle. d No gold particles were detected by bright-field TEM. e Ref. 15. f s.e. – standard error of the mean, s.d. – standard deviation of the distribution. | |||
0.5Au101/WO3 | 0.52 | 2.24 ± 0.08 | 0.7 |
0.5Au101/WO3a | 0.47 | 5.08 ± 0.27 | 2.1 |
0.5Au101/WO3b | 0.47 | 5.39 ± 0.17 | 1.8 |
0.1Au101/WO3 | 0.097 | 2.25 ± 0.05 | 0.5 |
0.1Au101/WO3a | 0.070 | 4.40 ± 0.17 | 1.6 |
0.3Au9/WO3 | 0.31 | <1d | — |
0.3Au9/WO3a | 0.28 | 5.13 ± 0.11 | 1.5 |
0.3Au9/WO3b | 0.28 | 5.79 ± 0.25 | 2.1 |
0.3Au9/WO3c | 0.28 | 6.13 ± 0.21 | 2.0 |
0.1Au9/WO3 | 0.093 | <1d | — |
0.1Au9/WO3a | 0.087 | 7.93 ± 0.27 | 2.2 |
0.5Au101/SiO2 | 0.44 | 1.97 ± 0.04 | 0.6 |
0.5Au101/SiO2a | 0.29 | 5.06 ± 0.18 | 2.0 |
0.1Au101/SiO2e | 0.12 | 1.59 ± 0.04 | 0.4 |
0.1Au101/SiO2a,e | 0.068 | 4.94 ± 0.15 | 1.5 |
0.3Au9/SiO2 | 0.27 | <1d | — |
0.3Au9/SiO2a | 0.20 | 6.27 ± 0.17 | 1.6 |
0.3Au9/SiO2b | 0.20 | 6.85 ± 0.23 | 2.3 |
0.1Au9/SiO2 | 0.098 | <1d | — |
0.1Au9/SiO2a,e | 0.073 | 9.6 ± 0.6 | 3.9 |
In the case of WO3, for target loadings greater than 0.1 wt%, successful deposition of Au9 required addition of n-hexane, and maximum actual loadings of only ca. 0.3 wt% could be achieved. Deposition of Au101 on WO3 was readily achieved without the use of n-hexane for all target loadings.
After deposition, the mean diameter of Au101 clusters, as determined by TEM, slightly increases (Table 1). We were unable to detect as-deposited Au9 clusters using bright-field TEM, which indicates that the size of metal core remains below 1 nm. However, during the course of catalytic cyclohexene oxidation, both types of clusters sinter to form particles with mean diameters ranging from ca. 4 to 10 nm (Table 1 and Fig. 1). The mean diameter of the nanoparticles formed from Au101 during the catalytic cycle is decreasing with lower gold loading. In contrast, Au9 clusters form bigger nanoparticles at lower loading. This could suggest different sintering mechanisms for two clusters – Au9 acts as feedstock for nanoparticle growth, while Au101 clusters collide and agglomerate through surface diffusion. We suggest that agglomeration of Au9 clusters occurs similarly to the process of crystal formation in which lower concentration of precursor leads to the creation of smaller amount of nucleation sites and thus to bigger crystals. Thus, at high gold loadings, the majority of Au9 clusters rapidly sinter into stable nanoparticles with mean diameters of 4–6 nm, while at lower surface concentrations, clusters initially form fewer nanoparticles that act as nucleation sites that keep growing, consuming the remaining Au9 clusters to eventually form 8–10 nm particles. As agglomeration of Au101 proceeds via cluster collision, the degree of sintering should decrease with the lower surface concentration of clusters.
Catalyst | Conversion, % | Selectivity, % | |||
---|---|---|---|---|---|
Cy-oxide | Cy-ol | Cy-one | CyOOH | ||
a Reaction conditions: cyclohexene (5 mL), n-decane (0.2 M) as an internal standard, catalyst (0.05 g), O2 (~1 atm), 65 °C, 16 h, glass reactor. b Ref. 15. | |||||
Blank | 2 | — | — | — | — |
TiO2 | 2 | — | — | — | — |
SiO2 | 2 | — | — | — | — |
WO3 | 9 | 33 | 30 | 4 | 32 |
0.5Au9/TiO2 | 2 | — | — | — | — |
0.5Au101/TiO2 | 2 | — | — | — | — |
0.5Au101/WO3 | 50 | 26 | 18 | 17 | 19 |
0.1Au101/WO3 | 36 | 35 | 23 | 12 | 18 |
0.3Au9/WO3 | 50 | 27 | 20 | 16 | 17 |
0.1Au9/WO3 | 33 | 34 | 24 | 11 | 21 |
0.5Au101/SiO2b | 48 | 6 | 15 | 24 | 38 |
0.1Au101/SiO2b | 39 | 7 | 11 | 17 | 54 |
0.3Au9/SiO2 | 43 | 7 | 12 | 19 | 51 |
0.1Au9/SiO2b | 25 | 5 | 7 | 12 | 68 |
Cyclohexene conversion and product distribution depend strongly on the nature of the support. TiO2-based catalysts are inactive, showing conversions comparable to that of the reaction in the absence of a catalyst (blank). Clusters deposited on SiO2 show high conversions of cyclohexene, with cyclohexenyl hydroperoxide (CyOOH) being the main product. Other products were 2-cyclohexen-1-one (Cy-one), 2-cyclohexen-1-ol (Cy-ol) and cyclohexene oxide (Cy-oxide). WO3-supported clusters have comparable activity, but the main product is cyclohexene oxide.
Unsupported gold nanoparticles have previously been shown to have high catalytic activity in some liquid-phase reactions;23 thus, it is important to investigate whether the observed catalytic activity should be attributed to supported or leached gold species. As seen from the AAS studies (Table 1), gold does leach into solution from as-prepared samples during the first catalytic cycle.
However, no such leaching was detected during subsequent recyclability tests. We suggest that as clusters agglomerate into bigger particles, leaching of gold species stops. This suggestion correlates with the fact that the degree of leaching during the first cycle decreases with higher loading. A higher surface density of gold clusters accelerates agglomeration and hence fewer non-sintered clusters have time to leach into solution. Because the catalyst retains ca. 90% of its catalytic activity during catalytic cycles 2–5 (Fig. 2), while leaching becomes undetectable on these stages, we conclude that the activity is very predominantly attributable to the supported, agglomerated particles and that gold species leached during the first cycle are essentially catalytically inactive. The methodology of hot filtering is typically used to determine whether the catalyst is homogeneous or heterogeneous.24 In our case, the reaction did not slow down upon Au/WO3 removal using a 0.2 μm filter after 6 h, which, according to the method, should indicate the homogeneous nature of the catalysis. However, it is known that cyclohexenyl hydroperoxide can catalyse the autoxidation of cyclohexene.25 Therefore, the hot filtering test is not suitable for distinguishing between heterogeneous and homogeneous catalysis for cyclohexene oxidation when cyclohexenyl hydroperoxide is formed in a sufficient amount.
![]() | ||
Fig. 2 Recyclability of 0.3Au9/WO3. Conversion of cyclohexene (black) and selectivity towards cyclohexene oxide (grey). |
To provide further evidence that leached gold species are inactive in cyclohexene oxidation, we subjected the catalysts to a pair of tests under reaction-like conditions in which gold leaching would occur but cyclohexenyl hydroperoxide formation would be suppressed. In the first test, an initial reaction was performed at 65 °C for 6 h using as-made catalyst, but the reactor was filled with argon instead of oxygen. The solid catalyst was then removed by hot filtration and the liquid reaction mixture was subjected to typical reaction conditions (65 °C or 16 h) with the reactor now filled with ca. 1 atm oxygen. The existence of gold leachate in the reaction mixture was confirmed using ICP-MS, but no detectable cyclohexene conversion was observed. In the second test, cyclohexenyl hydroperoxide formation was suppressed by addition of n-hexane. An experiment was conducted under oxygen atmosphere (ca. 1 atm) using the mixture of cyclohexene (50 μL) and n-hexane (2 mL). After 6 h, the liquid phase was separated by hot filtering and added to 5 mL of cyclohexene. When subjected to typical reaction conditions, the mixture showed no cyclohexene conversion over 16 h. In contrast, a solid catalyst, isolated on the first stage of the second test, catalysed the oxidation of cyclohexene in the mixture of n-hexane (2 mL) and cyclohexene (5 mL), giving 11% conversion.
![]() | ||
Fig. 3 DR UV-Vis spectra of the catalysts. (a) 0.3Au9/WO3 as deposited. Catalysts recovered after the 1st catalytic cycle: (b) 0.3Au9/SiO2, (c) 0.3Au9/WO3 and (d) 0.5Au101/WO3. |
The product evolution profiles for both types of catalysts have similar cyclohexenyl hydroperoxide accumulation stages. Based on these data, we suggest that gold particles catalyse cyclohexene conversion to cyclohexenyl hydroperoxide (reaction I, Scheme 1), while WO3 catalyses reaction of cyclohexenyl hydroperoxide with cyclohexene, producing cyclohexene oxide and 2-cyclohexen-1-ol (reaction II, Scheme 1).
To support this hypothesis, we conducted a series of experiments in which pure WO3 powder was mixed with 03Au9/SiO2 at different ratios (Fig. 5A). Interestingly, the addition of just 2 wt% of WO3 (1 mg) changes the selectivity of the reaction, with cyclohexene oxide becoming the main product. With 10 wt% of WO3 (5 mg), the distribution of products is almost identical to that for 03Au9/WO3. It is possible, however, that in this series of experiments, some gold species leach from silica-based catalyst and adsorb on tungsten oxide, thus forming a catalytic system with high selectivity towards cyclohexene oxide.
![]() | ||
Fig. 5 Effect of the co-catalyst on the selectivity in cyclohexene oxidation. Co-catalyst: WO3 (A) or MIL-101 (B). |
To exclude this possibility, we performed a reaction with the silica-based catalyst and hot filtered the reaction mixture into a vial containing tungsten oxide (5 mg), which was then collected by centrifugation. Inasmuch as potentially impregnated WO3 was inactive in the conversion of the fresh cyclohexene, we conclude that the observed change of selectivity upon addition of WO3 to the Au/SiO2 catalyst should be attributed to properties of pure WO3 acting as a co-catalyst rather than to a synergistic effect between support and Au nanoparticles. The discovered ability of WO3 to activate cyclohexenyl hydroperoxide, promoting its reaction with cyclohexene, is consistent with reports on activation of H2O229 and alkyl hydroperoxides30 by WO3 through the formation of peroxo complexes of tungsten which are highly active and selective in the epoxidation of olefins.31
We also found that the use of a different co-catalyst can shift selectivity of Au/SiO2 towards allylic oxidation products. For such a co-catalyst, we have chosen the metal–organic framework MIL-101, which was recently reported as a catalyst for allylic oxidation of cyclohexene with molecular oxygen, with 2-cyclohexen-1-one being the main product.32,33 Results of catalytic testing of the mixture consisting of pure MOF and 0.3Au9/SiO2 in different ratios show the effect of altering the selectivity of reaction similar to the one found for WO3, but with 2-cyclohexen-1-one becoming the main product (Fig. 5B). Reaction catalysed by the mixture of MIL-101:
0.3Au9/SiO2 (5
:
45, mg
:
mg) gives the maximum 2-cyclohexen-1-one yield of 16%, which is twice the yield we have been able to achieve with the pure MOF.
We found that cyclohexenyl hydroperoxide, the formation of which is catalysed by Au nanoparticles, can be converted to other products in the presence of different heterogeneous co-catalysts. Cyclohexene oxide is formed via reaction of cyclohexenyl hydroperoxide with cyclohexene catalysed by WO3, either present as a support or introduced to the reaction as a co-catalyst physically admixed with silica-supported gold catalyst. Selectivity towards formation of 2-cyclohexen-1-one is shifted by using MIL-101 as a co-catalyst for gold supported on SiO2.
We have shown that careful choice of support or co-catalyst for supported gold nanoparticles can tune the selectivity of cyclohexene oxidation towards cyclohexene oxide or 2-cyclohexen-1-one under solvent-free conditions without addition of a radical initiator and using oxygen as the only oxidant.
Footnotes |
† Electronic supplementary information (ESI) available: TGA of gold clusters. Characterization of MIL-101, including PXRD patterns, TGA and nitrogen physisorption analysis data. See DOI: 10.1039/c3cy01011b |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2014 |