Peter J.
Miedziak
a,
Hamed
Alshammari
b,
Simon A.
Kondrat
a,
Tomos J.
Clarke
a,
Thomas E.
Davies
a,
Moataz
Morad
a,
David J.
Morgan
a,
David J.
Willock
a,
David W.
Knight
a,
Stuart H.
Taylor
a and
Graham J.
Hutchings
*a
aCardiff Catalysis Institute, School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff, CF10 3AT, UK. E-mail: hutch@cardiff.ac.uk
bChemistry Department, Faculty of Science, Ha'il University, P. Box 2440, 81451 Ha'il, Saudi Arabia
First published on 10th April 2014
We report the selective oxidation of glucose to gluconic acid under mild conditions and show that if a basic support is used then the reaction can be carried out without the addition of sacrificial base or pH control. The use of sol-immobilisation prepared catalysts supported on magnesium oxide facilitates the use of ambient air as an oxidant source. These mild conditions resulted in an excellent selectivity towards gluconic acid. Different heat treatments result in an improvement in the activity of the catalyst, these improvements are discussed in terms of XRD, DRIFTD and TEM analysis of the catalysts, despite significant particle growth and phase segregation occurring during the thermal treatments.
Early reports, that have used heterogeneous catalysts for this oxidation, focused on the use of platinum on carbon as a catalyst. However, it was found that these catalysts suffered from marked deactivation, which was reduced when the reaction was carried out at high initial pH; but this deactivation could not be eliminated.9 The deactivation of these catalysts was addressed when the pH of the solution was readjusted to pH 9 during the reaction by continuous addition of base. The use of bimetallic catalysts has also been reported for this oxidation Besson et al.10 who used palladium catalysts supported on carbon promoted with bismuth. The selectivity and conversion were close to 100% when the reaction was carried out at pH 9 with bubbling air as the oxidant and the pH was maintained at a constant level throughout the reaction by addition of NaOH. Although the authors demonstrated five reuses of the catalyst they did not rule out deactivation as the reaction is run under mass transfer control. Gallezot11 also reported the oxidation of bismuth palladium supported on carbon, noting that 3–10 nm particles were more active than 1–2.5 nm particles and suggesting the bismuth prevents the over-oxidation of the products.
Wenkin et al.12 carried out an extensive study into the preparation of carbon-supported bismuth-promoted palladium catalysts using various bismuth precursors. Organic based acetate precursors were found to be the most active precursors, however, bismuth leaching into the reaction solution was detected for all the catalysts tested. Gold catalysts have also been reported for this oxidation, Biella et al.4 proposed the use of gold supported on carbon prepared by the sol-immobilisation method for this oxidation, under various pH conditions, the gold catalyst was found to be more active than platinum, palladium–bismuth or platinum–palladium–bismuth trimetallic catalysts. A pH of 9.5 was reported as the optimum value, however, the gold was also reported to maintain activity at pH values other than the optimised pH. Upon recycling the gold catalyst did display a decrease in its activity, thought to be due to metal leaching. In a further study13 “naked” gold nanoparticles were formed by the colloidal method and tested for the oxidation of glucose, the particles of gold were found to undergo growth during the reaction preventing the reaction from reaching completion but the initial turnover frequencies (TOFs) of the oxidation were extremely high demonstrating the potential of gold as a catalyst for this oxidation. Similar nanoparticles of copper, silver, palladium and platinum were tested but found to be inactive. Önal et al.14 studied different sol preparation methods and different carbon supports at pH 7–9.5 and found the best results were obtained at 50 °C and pH 9.5 using Vulcan-type carbons, Comotti et al.7 also reported the oxidation with H2O2 and Prüße and co-workers published a series of papers investigating the long term stability of gold catalysts on various supports.15–18 Gold on alumina was prepared by the deposition precipitation and incipient wetness methods, the relatively low loaded (0.3% Au) catalysts were shown to be active for this reaction. Further work reported 0.45% Au catalysts supported on titania16 prepared by deposition precipitation and sol methodologies, gave reasonable stability of the catalysts demonstrated over seventeen reuses. They further demonstrated the incipient wetness prepared gold on alumina catalysts in a continuous flow reactor17 with high catalyst stability for 110 h. Different deposition precipitation (DP) methods were studied for gold on alumina18 and urea was reported to be the best precipitating agent for this preparation method.
An interesting comparison between a gold catalyst, supported on carbon and an enzymatic reaction was provided by Rossi and co-workers,19 the work demonstrated that if the activity of the gold catalysts was calculated based on theoretical surface gold atoms, the rate of reaction would be of the same order of magnitude as the enzymatic reaction. Ishida et al.20 used the solid grinding method to prepare gold catalysts which they found gave higher activity than the DP catalyst preparation when supported on alumina or zirconia.
Bimetallic catalysts comprising combinations of gold, platinum, palladium and rhodium were studied by Comotti et al.8 and the gold platinum catalysts were the most active and with the exception of gold rhodium the alloying of the metals improved the activity. The fine-tuning of the ratio of gold to platinum led to an optimised ratio of Au:
Pt of 2
:
1 (molar) and with this catalyst at pH 9.5 a TOF of 17
600 h−1 was observed.
Karski et al.21 used thallium alloyed with palladium on silica catalysts and there was an observable increase in the selectivity when the thallium was alloyed. However, high amounts of thallium were required to observe a significant effect and there was significant leaching of the toxic thallium which would be prohibitive for industrial applications. Witońska et al.6 investigated palladium tellurium on silica catalysts and found that tellurium could also be used as a promoter for this reaction, again however deactivation of the catalysts was reported after ten reuses of the catalyst. Recently Zhang and co-workers have prepared homogeneous colloidal nanoparticles of gold–silver2 and gold–platinum–silver catalysts, once again these catalysts have demonstrated the extremely high TOF's with 20090 h−1.
We have previously shown that alloying gold and palladium can lead to a significant increase in the activity of the catalyst when compared to the gold or palladium monometallic equivalents,22 we have also shown that the sol immobilisation method can give catalysts with a narrow particle size distribution that are extremely active for alcohol oxidations.23 Heat treatment of these sol-immobilised catalysts can remove the polyvinyl alcohol (PVA) ligand but simultaneously leads to metal sintering causing deactivation of the catalyst during CO oxidation.24 Recently we have reported the use of magnesium oxide as a support for gold–palladium and gold platinum for the oxidation of glycerol, we have shown that during catalysts preparation the magnesium is predominately hydrated to form Mg(OH)2. The use of these catalysts facilitated the oxidation of glycerol without the addition of sacrificial base and at lower temperatures than those reported when other support materials are used.25 We now show that by combining the use of magnesium oxide/hydroxide support and the sol immobilisation preparation method we can carry out glucose oxidation under green, base-free conditions, using air at ambient pressure as the source of oxygen with excellent selectivity towards gluconic acid. We have shown that various heat treatments on the support and catalyst have a significant effect of the activity of the catalyst and explain these differences using XRD and TEM analysis.
![]() | ||
Fig. 1 The conversion of glucose using Au–Pd/MgO(b): reaction conditions: glucose (5 ml, 0.5 M), catalyst (0.06 g), 24 h at atmospheric pressure, selectivity 100% except at 70 °C which was 98%. |
Catalyst | Conversion (%) | Selectivity (%) |
---|---|---|
a Reaction conditions: glucose (5 ml, 0.5 M), catalyst (0.06 g), 60 °C, 24 h at atmospheric pressure. | ||
No catalyst | 0 | 0 |
Mg(OH)2 | 1.3 | 100 |
0.5% Au/Mg(OH)2(o) | 7.3 | 100 |
Catalyst | Conversion (%) | Selectivity (%) |
---|---|---|
a Reaction conditions: glucose (5 ml, 0.5 M), catalyst (0.06 g), 60 °C, 24 h at atmospheric pressure. | ||
0.5% Au/graphite(z) | 0 | 0 |
0.5% Au/TiO2(z) | 0.8 | 100 |
0.5% Au/Mg(OH)2(z) | 2 | 100 |
Catalyst | Catalyst heat treatment | Conversion (%) | Selectivity (%) | |
---|---|---|---|---|
Under air | Under N2 | |||
a Reaction conditions: glucose (5 ml, 0.5 M), catalyst (0.06 g), 60 °C, 24 h at atmospheric pressure. | ||||
0.5% Au–Pd/Mg(OH)2(z) | X | X | 27 | 100 |
0.5% Au–Pd/MgO2(a) | √ | X | 50 | 100 |
0.5% Au–Pd/MgO2(n) | X | √ | 53 | 100 |
0.5% Au–Pd/MgO2(b) | √ | √ | 62 | 100 |
Treatment of the catalyst under nitrogen also resulted in the dehydration of the Mg(OH)2 to MgO with a crystallite size of 7 nm (Fig. 2). The presence of a broad poorly defined reflection, likely to be associated with a metal phase, was again observed, at 32.29° 2θ. The reflection position is again suggestive of an Au–Pd alloy, though with a unit cell size closer to Pd than that expected for a bulk homogeneous alloy. From the Scherrer equation this phase had an average crystallite size of 10 nm. Interestingly, the sequential heat treatments results in the formation of a metal phase with similar crystallite size to the nitrogen treatment (9 nm) but an apparent unit cell size that is now closer to Au than after the nitrogen treatment but still smaller than the calcined sample. The apparent trend we observe is that calcination, i.e. heating in air, resulted in the presence of an Au–Pd alloy that is possibly Au rich, whereas nitrogen heat treatment a phase that was Pd rich and the combined treatments in a phase that lies between the two single treatments.
TEM images of the unused catalyst are shown in Fig. 4 and associated particle size distributions, based on analysis of at least 200 particles, are shown in Fig. 5. In agreement with previous reports on 1% Au–Pd/MgO for glycerol oxidation25 the sample that is dried only has a narrow size distribution of small nanoparticles (1–10 nm). As we have previously reported for titania-supported catalysts24 the calcination treatment (Fig. 3b and 4b) leads to a growth in the mean particle size, although on magnesium hydroxide this is less evident than when titania is used as the support,24 even at the slightly higher calcination temperature of 450 °C, furthermore the particle size distribution remained positively skewed. The heat treatment under nitrogen at 500 °C also led to a growth in the average particle size although the effect was not as significant as when the calcination regime was applied. Fig. 3d and 4d show the TEM image and associated particle size distribution after both heat treatments have been applied to the catalyst, it is noteworthy that the double heat treated catalyst particle size distribution is most similar in nature to the to the nitrogen heat treated catalyst. The catalytic activity of both the nitrogen heat treated catalyst and the calcined catalyst are very similar suggesting that the particle size is not the single factor that is affecting the activity, we do not know what effect the removal of PVA ligand from the metal particles is having in this case, however it seems apparent that the smallest metal particles are not the most active for this reaction as has been previously suggested by Besson et al.10
![]() | ||
Fig. 5 Time on line data for the oxidation of glucose using the 0.5% Au–Pd/MgO catalyst. Reaction condition: glucose (5 ml, 0.5 M), catalyst (0.06 g), 50 °C at atmospheric pressure. |
Catalyst | Conversion (%) | Selectivity (%) |
---|---|---|
a Reaction conditions: glucose (5 ml, 0.5 M), of 0.5% Au–Pd/MgO(b) (0.06 g), 60 °C, 24 h at atmospheric pressure. | ||
Fresh | 62.0 | 100 |
First reuse | 24.4 | 100 |
First reuse with treatment of catalyst | 33.7 | 100 |
Second reuse | 18.7 | 100 |
Second reuse with treatment of catalyst | 23.2 | 100 |
Catalyst | Gluconic acid | Conversion (%) | Selectivity (%) |
---|---|---|---|
a Reaction conditions: glucose (5 ml, 0.5 M), catalyst (0.06 g), 60 °C, 24 h at atmospheric pressure. | |||
0.5% Au/MgO(b) | — | 62.0 | 100 |
0.5% Au/MgO(b) | 0.1 g | 41.3 | 100 |
Characterisation by DRIFTS, XRD and TEM microscopy of the Au–Pd/MgO(b) catalyst after a 24 h reaction at 50 °C was performed to further investigate catalyst deactivation. XRD analysis (Fig. 6) revealed that the MgO support completely converted to Mg(OH)2 under the aqueous reaction conditions, which could be expected to result in significant agglomeration of the supported nanoparticles. TEM analysis (Fig. 6) did show a modest increase in particle size from a mean of 5.8 nm to 6.5 nm. This small increase in particle size could impact on activity, although it being a significant factor is highly improbable. What is surprising is the minimal change in metal particle size on restructuring of the support. Interestingly DRIFTS of the initial and used catalysts shows, not only the presence of adsorbed organic species, but also the “surface” structure of the catalyst itself (Fig. 7 and Table 7). The spectra of both the initial and used catalysts show strong hydroxyl groups associated with magnesium hydroxide, demonstrating that even the initial Au–Pd/MgO(b) catalyst, shown to be comprised of bulk MgO by XRD, had a hydrated surface. This surface magnesium hydroxide could potentially stabilise the AuPd nanoparticles when the bulk MgO support hydrates to Mg(OH)2 under reaction conditions.
Mg(OH)2 | MgCO3 | Fresh catalyst | Used catalyst | Used at 115 °C | Used at 450 °C | Band assignment |
---|---|---|---|---|---|---|
3698 | — | 3763, 3714, 3701, 3620 | 3699, 3646 | 3699, 3646 | 3697, 3646 | ν(Mg–OH) |
— | — | 3361 | 3400 | 3360 | 3295 | ν(OH) |
— | — | 2930, 2870, 2845, 2740 | 2930, 2870, 2845, 2740 | 2930, 2870, 2845, 2740 | ν(CH) | |
— | — | 1647 | — | δ(H2O) | ||
— | 1635 | 1653 | 1653 | ν(CO) | ||
— | 1450 | 1454 | — | ν 3(CO3) | ||
1396, 1360 | 1405, 1336 | 1420, 1330 | δ(CH)–(OCH) | |||
1232 | 1230 | δ(CH) | ||||
1130 | 1126 | 1126 | ν(CO) | |||
— | 1080 | 1070 | ν 1(CO3) | |||
— | 870 | 830 | ν 4(CO3) | |||
787 | 784 | 785 | ||||
625 | — | 660 |
DRIFTS of the used catalyst did also show distinct bands, relative to the initial catalyst, that correspond to alcohol groups, C–H and CO bands. Heating of the used catalyst in situ under vacuum up to 450 °C resulted in only a reduction in the broad OH band at 3400 cm−1, associated with physisorbed water. This showed that the adsorbed organic species were exceptionally strongly bound to the catalyst surface and supports the idea that product inhibition is the primary cause of the observed deactivation. Identification of the adsorbate from band assignments of the used and used heat treated samples is complicated due to the range of structural isomers of glucose and gluconic acid. However, the presence of the band at 1653 cm−1 in the used and heated samples is suggestive of a carbonyl group. The band position is at the extreme low end of the range observed for carbonyls and is in a similar range to the expected bending mode of physisorbed water. Yet its retention at 450 °C, strongly suggests that the band in question is not associated with physisorbed water. The ν(OH) band remaining in the heated samples, at 3295 cm−1, can be identified as being associated with a carboxylic acid group. It is therefore probable that the adsorbed species was gluconic acid. Clearly product inhibition represents a key area that needs to be addressed in future studies. Perhaps the use of polar solvents could be beneficial but this negates the green chemistry approach of using a solvent-free reaction system however, our observations do show that there may be limitations to the application of aqueous catalysed reactions.
In addition to the previously mentioned post reaction characterisation, MP-AES analysis was carried out on the reaction mixture after the 24 h reaction had been completed, to determine if metal leaching was occurring during the reaction. Analysis of both gold and palladium revealed that neither metal was present in the post reaction sample within the detection limits of the equipment (standards made between 0.6 and 10 ppm). However, we observed that Mg was present (2890 ppm), showing that the Mg(OH)2 support was partially soluble in the reaction medium. Previous use of MgO as a solid base for base free glycerol oxidation resulted in 77 ppm of Mg2+ being leached into solution, though the reaction time was far shorter at 4 h. In view of the extent of Mg2+ that is leached in the current experiments we have to consider the effect of the homogeneous Mg(OH)2 acting as a sacrificial base. The calculated [OH−] was 2.38 × 10−4 mol ml−1, which is equivalent to a [OH] : glucose ratio of 0.048:
1 at the start of the reaction, and a ratio of 0.16
:
1 at 70% conversion. This suggests that while the catalyst itself is unstable, the dissolved Mg(OH)2 does not provide significant amounts of sacrificial homogeneous base to influence the reaction. To test this we carried out a reaction with a stable catalyst which we have previously reported,22 with and without the addition of sodium hydroxide at the same ratio (i.e. [OH−] : glucose = 0.048
:
1). Under our reaction conditions we observed conversions of 3.6 and 3.8% in the absence and presence of base respectively, indicating that the dissolved base is not responsible for the high activity we observe with the catalysts we now report. Though we can determine the efficacy of the work in demonstrating the ability to perform base free oxidation of glucose under mild conditions, the instability of the catalyst support shows that MgO is clearly not fully viable for this application. Further work on other more stable basic supports is now required.
We have demonstrated that the heat treatment of these sol immobilised catalysts can have a marked effect on their activity. XRD and TEM analysis show that there is large particle growth when the catalyst is either calcined, heat treated under nitrogen or undergoes both treatments. There is also apparent phase separation of the metals when both treatments are applied however this is the most active catalyst indicating that only a small amount of alloying is required. Finally we have observed catalyst deactivation with time which we attribute to product inhibition which may be alleviated by using a suitable solvent however support dissolution remains a problem.
This journal is © The Royal Society of Chemistry 2014 |