S. Doherty*a,
J. G. Knight*a,
M. A. Carrolla,
A. R. Clemmeta,
J. R. Ellisona,
T. Backhousea,
N. Holmesb,
L. A. Thompsonb and
R. A. Bourneb
aNUCAT, School of Chemistry, Newcastle University, Bedson Building, Newcastle upon Tyne, NE1 7RU, UK. E-mail: simon.doherty@ncl.ac.uk; Tel: +44 (0)191 208 6537
bInstitute of Process Research & Development, School of Chemistry, University of Leeds, Woodhouse Lane, Leeds LS2 9JT, UK. E-mail: cmlt@leeds.ac.uk
First published on 22nd July 2016
Styrene-based peroxotungstate-modified polymer immobilized ionic liquid phase catalysts [PO4{WO(O2)2}4]@ImPIILP (Im = imidazolium) are remarkably efficient systems for the selective oxidation of sulfides under mild conditions both in batch and as a segmented or continuous flow process using ethanol as the solvent and mobile phase, respectively. The performance of these styrene-based systems has been compared against their ring opening metathesis polymerisation derived counterparts to assess their relative merits. A comparative survey revealed the catalyst supported on N-benzyl imidazolium decorated polymer immobilised ionic liquid to be the most efficient and a cartridge packed with a mixture of [PO4{WO(O2)2}4]@ImPIILP and silica operated as a segmented or continuous flow system giving good conversions and high selectivity for sulfoxide. The immobilised catalyst remained highly active for the sulfoxidation of thioanisole in ethanol with a stable conversion–selectivity profile for up to 8 h under continuous flow operation; for comparison conversions with a mixture of [NBu4]3[PO4{WO(O2)2}4] and silica dropped dramatically after only 15 min as a result of rapid leaching while [PO4{WO(O2)2}4]@ImPIILP prepared from commercially available Merrifield resin also gave consistently lower conversions; these benchmark comparisons serve to underpin the potential benefits of preparing the polymer immobilized ionic liquid supports.
Ionic liquids are an intriguing class of solvent that has been widely utilized for immobilisation of catalysts under homogeneous, liquid–liquid biphasic and liquid–solid (SILP) biphasic conditions, in some cases with remarkable success.29 Recent endeavours in this area include highly selective sulfoxidations catalysed by a SILP system based on imidazolium modified SBA-15 and [MoO(O2)2(H2O)n],30 a magnetically recoverable sulfoxidation catalyst based on magnetic nanoparticles entrapped in a tungstate-functionalised polyionic liquid,31 an eco-friendly protocol for the oxidation of sulfides to sulfones catalysed by V2O5 in [C12mim][HSO4]32 and efficient and selective sulfoxidation catalysed by peroxotungstates immobilised on multilayer ionic liquid brushes-modified silica.33 Other recent developments include selective oxidation of sulfides with H2O2 catalysed by heterogeneous ionic liquid-based polyoxometalates,34 selective oxidation of sulfides with a sulfoacid-hexafluorotitanate(IV) bifunctional ionic liquid,35 ionic liquid-mediated oxidation of sulfides to sulfoxides,36 efficient eco-friendly selective oxidation of sulfides to sulfoxides with molecular oxygen catalysed by Mn(OAc)2 in [C12mim][NO3],37 rapid oxidation of sulfides by thermoregulated polyoxometalate based ionic liquids,20b,38 selective and efficient desulfurization by amphiphilic polyoxometalate-based ionic liquid supported silica,39 and heterogeneous selective sulfoxidation with polymeric ionic liquid nanogel-immobilised tungstate anions.40
We have recently applied the concept of SILP-based technology to develop peroxotungstate-based Polymer Immobilised Ionic Liquid Phase (PIILP) oxidation catalysts in order to combine the favourable and tuneable properties of ionic liquids with the advantages of a solid porous support.41 Ring opening metathesis derived ionic liquid polymers were used to prepare the corresponding peroxotungstate-based PIILP catalyst, [PO4{WO(O)2}4]@PIILP on the basis that the well-behaved functional group tolerant nature of ruthenium-catalysed living polymerisation would enable surface properties, ionic microenvironment, porosity and hydrophilicity to be modified and thereby catalyst–surface interactions, substrate accessibility and catalyst efficacy to be optimised in a rational and systematic manner. Gratifyingly, our initial foray in this area demonstrated that peroxotungstate immobilised on pyrrolidinium-decorated norbornene/cyclooctene copolymer was a remarkably efficient system for the selective oxidation of sulfides in batch and continuous flow. This was the first report of continuous flow sulfoxidation and despite the potential importance of this technology there are still relatively few examples in the academic literature. In this regard, following our initial disclosure Alemán and co-workers developed a Pt(II)-based visible light photocatalyst for the oxidation of sulfides both in batch and flow; the system gave complete chemoselectivity for sulfoxide but required long reaction times (10 h) to reach good conversions.42 We have now undertaken a comparison of the efficiency of our original system against a range of polystyrene-based polymer immobilised ionic liquid supported peroxotungstates in order to assess the relative merits of both, reasoning that styrene-based monomers are easy to prepare and the corresponding polymers would be more cost effective and have good thermal and mechanical integrity. Herein we report the results of this comparison which demonstrates that styrene-based polymer immobilised ionic liquid phase supported peroxotungstates give high conversions and excellent sulfoxide selectivity under mild conditions, both in batch and under continuous flow operation using ethanol as the solvent or mobile phase, and that the most efficient system outperforms its ROMP-derived counterpart. Moreover, the remarkable stability of the performance–time profile allowed continuous flow operation to be maintained over extended periods of time with only a minor reduction in performance. As continuous flow processing of sulfoxidation has not been thoroughly investigated this study will provide a valuable benchmark and platform for future developments in this key area.
The molecular weight (Mw) of 2a–c determined by gel permeation chromatography was measured to be 31600 (2a), 26100 (2b) 27800 (2c) relative to polystyrene standards and the polydispersities of 1.32, 1.19, and 1.17, respectively, are consistent with relatively narrow monomodal molecular weight distributions. The ratio of imidazolium monomer to styrene incorporated into the polymer was determined to be ca. 0.5 which corresponds to m and n values of 32 and 16, respectively, based on the average molecular weights determined by GPC. The thermal stability of co-polymers 2a–c was investigated by thermogravimetric analysis and differential scanning calorimetry. The TGA of 2a–c showed an initial weight loss at ca. 100 °C, due to removal of physisorbed water and ethanol, followed by two main degradation pathways, indicating that the polymers are thermally stable up to 300 °C; this is well above the reaction temperature required for liquid phase catalysis. Solution and solid state NMR spectra of 2a–c confirm that the samples do not contain any imidazolium or styrene monomer as evidenced by the absence of signals at δ 5.2 and 5.8 ppm characteristic of vinylic protons. A reliable assignment of the signals in the solid state 13C NMR spectrum of 2a–c was obtained by conducting pairs of measurements, one with full cross-polarisation (dipolar dephasing with 0 μs delay) and one with a 50 μs dephasing delay to remove the CH and CH2 signals; this enabled the quaternary and CH3 signals to be identified.
Peroxotungstate-based PIILPs 3a–c were prepared by stoichiometric exchange of the halide anion in 2a–c with [PO4{WO(O2)2}4]3−, generated by hydrogen peroxide-mediated decomposition of the heteropolyacid H3PW12O40 (Fig. 1b).43 The desired product typically precipitated as an amorphous white solid and was characterised by a variety of techniques including solid state NMR spectroscopy, IR spectroscopy, TGA, SEM, XPS and elemental analysis. Decomposition of H3PW12O40 into [PO4{WO(O2)2}4]3− was confirmed by a signal at δ 2.9 ppm in the solid state 31P NMR spectrum; in the case of 3c the spectrum also showed the presence of minor phosphorus-containing species previously identified by Hill and co-workers during their early studies on the formation, reactivity and stability of [PO4{WO(O2)2}4]3−.44 Surprisingly, TGA analysis revealed that thermal decomposition of 3a–c occurred between 250 and 300 °C which is slightly lower than for 2a–c; this may be associated with a reduction in the binding affinity due to the large size of the peroxotungstate anion compared with halide and/or initial loss of coordinated peroxide which occurs below 200 °C. A similar effect has recently been reported for a polymer ionic liquid nanogel-anchored tungstate which was less thermally stable than the corresponding parent polymeric ionic liquid nanogel.40 Scanning electron microscopy revealed a stark difference in surface morphology of the polymers after loading of the peroxometalate (ESI‡). Specifically, the surface of polyoxotungstate loaded 3a–c exhibit a rough granular texture compared with the smooth flat surface of polymers 2a–c. The X-ray photoelectron spectra of 3a–c each contain characteristic W 4f7/2 and 4f5/2 doublets with binding energies of 37.1 and 39.1 eV, respectively, in good agreement with available literature data for tungsten ions in the +6 oxidation state.44 The tungsten loadings of 32.0–35.0 wt% for 3a–c were determined from elemental analytical data and are consistent with complete exchange of the bromide in 2a–c. With the aim of comparing and evaluating the efficacy of in-house synthesised polymer immobilised ionic liquid supports 2a–c against commercially available systems, [PO4{WO(O2)2}4]3− was also supported on macroreticular resin 2d and imidazolium-modified Merrifield resin 2e (Fig. 1c and d).
A series of catalytic reactions were first conducted under batch conditions to establish optimum conditions for comparative catalyst evaluation, substrate screening and recycle experiments as well as to identify potential systems for use in developing a continuous flow process,45 full details are presented in Table 1. Our initial optimisation focused on the sulfoxidation of thioanisole as the benchmark reaction as this oxidation has recently been catalysed by peroxometalate-based systems hosted in layered double hydroxides with enhanced activity and sulfoxide selectivity,28 polyoxometalate-calix[4]arene hybrids,46 thermoregulated Keggin-type polyoxometalate-based ionic liquids,20b,38 polymeric ionic liquids nanogels,40 composite polyoxometalates supported on Fe2O3,47 poly(ionic) liquid entrapped magnetic nanoparticles,31 and peroxometalates immobilised on the surface of ionic liquid modified silica.33,39 Gratifyingly, good conversions and high sulfoxide selectivity were obtained in methanol and ethanol after 15 min using a 0.5 mol% loading of 3a at room temperature and a H2O2:S mole ratio of 2.5 (entries 1 and 2). High selectivities were also achieved in propan-2-ol and ethylene glycol under the same conditions and even though reactions in the latter solvent were slower comparable conversions could be achieved at elevated temperatures (entries 3 and 4). Although slightly lower conversions were obtained in acetonitrile and 2-Me-THF, sulfoxide selectivity remained high (entries 5 and 6). For comparison the corresponding ROMP-based POM@PIILP system gave a slightly lower sulfoxide selectivity of 84% in acetonitrile, under the same conditions and at a similar conversion. In this regard, higher sulfoxide selectivity is generally obtained in protic solvents such as methanol and ethanol, which has been attributed to their high hydrogen-bonding capacity,27d,g,48 however, while alcohols are often the solvent of choice to achieve high sulfoxide selectivity, there have been recent reports in which acetonitrile has been identified as the optimum solvent.49 The minor decrease in conversion with increasing alcohol carbon number (entries 1–3) may be associated with the different polymer swelling capacity of these solvents which could affect access of the substrate to the active site, however, the differences in conversion are relatively minor and any interpretation should be treated with caution. The high selectivity and conversion obtained in ethanol coupled with its green and sustainable credentials prompted us to use this solvent for the remaining optimisation studies.
Entry | Solvent | Catalyst | H2O2 equiv. x | Conversionb | % sulfoxideb | % sulfoneb | Sulfoxide selectivityb,c | TOFd |
---|---|---|---|---|---|---|---|---|
a Reaction conditions: 0.56–0.58 mol% 3a–e, 1 mmol thioanisole, 1.0–3.0 mmol 35% H2O2, 3 mL solvent, 25 °C, 15 min.b Determined by 1H NMR spectroscopy.c Sulfoxide selectivity = [% sulfoxide/(% sulfoxide + % sulfone)] × 100%.d TOF = moles sulfide consumed per mole catalyst per hour.e Reaction conducted without catalyst in the presence of 0.5 mol% 2a. | ||||||||
1 | MeOH | 3a | 2.5 | 99 | 95 | 4 | 96 | 689 |
2 | EtOH | 3a | 2.5 | 94 | 91 | 3 | 96 | 654 |
3 | i-PrOH | 3a | 2.5 | 92 | 88 | 4 | 96 | 640 |
4 | EG | 3a | 2.5 | 44 | 43 | 1 | 98 | 334 |
5 | MeCN | 3a | 2.5 | 81 | 78 | 3 | 97 | 564 |
6 | 2-Me-THF | 3a | 2.5 | 54 | 44 | 2 | 96 | 376 |
7 | EtOH | 3a | 2.0 | 76 | 74 | 3 | 98 | 528 |
8 | EtOH | 3a | 3.0 | 95 | 91 | 3 | 96 | 661 |
9 | EtOH | 3a | 4.0 | 100 | 91 | 9 | 91 | 696 |
10 | EtOH | 3a | 5.0 | 100 | 83 | 17 | 83 | 696 |
11 | EtOHe | — | 2.5 | 0 | — | — | — | — |
12 | EtOH | 3b | 2.5 | 25 | 25 | 0 | 100 | 173 |
13 | MeCN | 3b | 2.5 | 49 | 48 | 1 | 98 | 336 |
14 | EtOH | 3c | 2.5 | 36 | 35 | 1 | 99 | 234 |
15 | MeCN | 3c | 2.5 | 53 | 52 | 1 | 99 | 359 |
16 | EtOH | 3d | 2.5 | 5 | 5 | 0 | 100 | 39 |
17 | MeCN | 3d | 2.5 | 18 | 17 | 1 | 94 | 125 |
18 | EtOH | 3e | 2.5 | 57 | 56 | 1 | 99 | 403 |
19 | MeCN | 3e | 2.5 | 42 | 41 | 1 | 99 | 297 |
20 | EtOH | 2a/H3PW12O40 | 2.5 | 2 | 2 | 0 | 100 | 19 |
Systematic variation of the H2O2:substrate mole ratio revealed that the best compromise between conversion and sulfoxide selectivity was obtained for a peroxide to substrate ratio of 2.5; below this ratio conversions were markedly lower (entry 7) while higher ratios gave complete consumption of sulfide but at the expense of selectivity which was markedly lower (entries 8–10). As sulfones are a useful class of compound the conversion–selectivity profile was also monitored as a function of temperature, with a peroxide to substrate ratio of 2.5, in order to identify conditions for the selective formation of methyl phenyl sulfone. Fig. 2 shows that sulfoxide selectivity drops dramatically with increased temperature such that sulfone was obtained as the major product in 93% selectivity after 15 min at 328 K. A control reaction for the oxidation of thioanisole conducted in ethanol in the absence of peroxotungstate but with 0.5 mol% 2a and 2.5 equivalents of H2O2 gave no conversion, which confirmed the active role of the catalyst (entry 11).
In order to explore the effect of the imidazolium cation on catalyst performance the efficiency of 3a–c for the sulfoxidation of thioanisole in ethanol and acetonitrile was investigated under the optimum conditions identified above and compared with the corresponding systems prepared from commercially available resin 3d–e, details of which are also summarised in Table 1 (entries 12–19). While 3a–c all gave high sulfoxide selectivities at room temperature in ethanol under optimum conditions, 3a is the most active with a TOF of 654 h−1 compared with 173 h−1 and 234 h−1 for 3b and 3c, respectively (entries 2, 12 and 14).
The data in Table 1 also highlights the merits of using catalyst prepared with in-house synthesised polymer immobilised ionic liquids as 3d and 3e only reached 5% and 57% conversion, respectively, in ethanol which correspond to TOF's of 39 h−1 and 403 h−1, respectively, both of which are significantly lower than that of 654 h−1 obtained with 3a (entries 16 and 18). In contrast, even though 3a was also more active than either 3b or 3c in acetonitrile, the difference in performance was not as marked as in ethanol, as evidenced by the TOF of 564 h−1 for 3a compared with 336 h−1 and 359 h−1 for 3b and 3c, respectively (entries 5, 13 and 15). Gratifyingly, 3a–c all outperformed 3d by a considerable margin, even though the TOF of 125 h−1 obtained in acetonitrile was a marked improvement on that in ethanol (entry 17). With the aim of investigating the possibility of generating [PO4{WO(O2)2}4]@ImPIILP in situ immediately prior to catalysis, in order to avoid the need to prepare, isolate and store the catalyst, H3PW12O40 was supported on 2c by wet impregnation from ethanol–water. Unfortunately, catalyst generated by treatment of the resulting H3PW12O40/2a with hydrogen peroxide was essentially inactive for sulfoxidation of thioanisole in ethanol and only achieved 2% conversion under the same conditions in the same time (entry 20).
A comparative study of the variation in conversion against sulfoxide and sulfone as a function of time for the sulfoxidation of 4-nitrothioanisole catalysed by 3a in ethanol and acetonitrile at room temperature shows that the composition–time profiles are qualitatively similar but that oxidation to sulfone is more rapid in acetonitrile than in ethanol (Fig. 3). Approximate rate constants for the formation of methyl phenyl sulfoxide (ka) and methyl phenyl sulfone (kb) in ethanol and acetonitrile were extracted by fitting the concentration–time profile for the consumption of sulfide and the formation of product using pseudo steady state analysis. It should be noted that 2 equivalents of H2O2 are consumed during the reaction and as such the derived rate constants will only be meaningful for this comparison, even though the data fit is visually very good. The data confirms that the solvent has a more significant effect on the second oxidation compared with the first; this may be associated with the increased hydrogen bond capacity of ethanol which could solvate the H2O2 effectively and thereby reduce its availability at the catalyst as it becomes depleted and/or solvate the sulfoxide and thereby stabilise it with respect to further oxidation. However, catalyst solvation may also be responsible for the solvent dependent difference in kb as it would be reasonable to expect solvation by ethanol to impede access of sulfoxide to the active centre to a greater extent than acetonitrile (Table 2).
H2O2 | MeCN | EtOH | ||
---|---|---|---|---|
ka | kb | ka | kb | |
a Data obtained using 4 mmol thioanisole, 12.2 mg 3a, 12 mL solvent, 10 mmol H2O2 and monitored by analysing 0.2 mL aliquots over 250 min. | ||||
2.5 | 0.06 | 0.009 | 0.068 | 0.006 |
Encouraged by the efficacy of 3a–c for the selective oxidation of thioanisole, catalyst testing was extended to explore their performance across a range of substrates under the optimum conditions identified above, full details of which are summarised in Table 3. The tabulated data clearly shows that 3a outperforms both 3b and 3c across the entire range of substrates examined, in both ethanol and acetonitrile, as evidenced from the consistently higher conversions, however, it is more difficult to use selectivity as a parameter to compare performance as 3a–c are all highly selective for sulfoxide within a relatively narrow range between 95 and 100%, albeit in some cases at low conversion. Interestingly, 3a gave higher TOFs for sulfoxidation in ethanol compared with acetonitrile for all but one substrate; in contrast, 3b and 3c gave higher TOF's in acetonitrile than in ethanol for all substrates tested. Moreover, the performance of 3b and 3c is highly substrate specific with some quite marked differences in TOF. Interestingly, the difference in performance between 3a and 3b–c is most clearly manifested in ethanol as evidenced by the greater disparity in TOF's. The contrasting, disparate and solvent dependent conversions obtained even within this closely related series of catalysts highlights the complex nature of these PIILP systems, and, while it is not possible to identify a support-catalyst performance relationship at this stage, the data in Table 3 suggests that it may well be possible to tailor the ionic environment on the support to modify and optimise catalyst efficiency and enhance stability and longevity.
Substrate | Catalyst | Solvent | % conversionb | % sulfoxideb | % sulfoneb | % sulfoxide selectivityb,c | TOFd |
---|---|---|---|---|---|---|---|
a Reaction conditions: 0.56–0.58 mol% 3a–c, 1 mmol substrate, 2.5 mmol 35% H2O2, 3 mL solvent, 25 °C, 15 min.b Determined by 1H NMR spectroscopy using 1,3-dinitrobenzene as internal standard.c Sulfoxide selectivity = [% sulfoxide/(% sulfoxide + % sulfone)] × 100%.d TOF = moles sulfide consumed per mole of catalyst per hour. Average of 3 runs.e Determined by 13C NMR spectroscopy using 1,3-dinitrobenzene as internal standard.f Reaction conducted at 25 °C for 30 min. | |||||||
3a | EtOH | 94 | 91 | 3 | 97 | 654 | |
3b | EtOH | 25 | 25 | 0 | 100 | 173 | |
3c | EtOH | 34 | 34 | 1 | 99 | 58 | |
3a | MeCN | 76 | 74 | 2 | 97 | 532 | |
3b | MeCN | 49 | 48 | 1 | 98 | 337 | |
3c | MeCN | 52 | 51 | 1 | 98 | 89 | |
3a | EtOH | 85 | 82 | 3 | 96 | 594 | |
3b | EtOH | 19.5 | 19 | 0.5 | 97 | 76 | |
3c | EtOH | 27 | 26 | 1 | 97 | 182 | |
3a | MeCN | 77 | 73 | 4 | 100 | 539 | |
3b | MeCN | 49 | 48 | 1 | 99 | 260 | |
3c | MeCN | 67 | 61 | 3 | 96 | 436 | |
3ae | EtOH | 75 | 73 | 2 | 98 | 525 | |
3be | EtOH | 11 | 11 | 0 | 100 | 260 | |
3ce | EtOH | 15 | 15 | 0 | 100 | 102 | |
3ae | MeCN | 69 | 67 | 2 | 97 | 482 | |
3be | MeCN | 38 | 36 | 2 | 96 | 76 | |
3ce | MeCN | 40 | 38 | 2 | 96 | 271 | |
3a | EtOH | 37 | 36 | 1 | 96 | 258 | |
3b | EtOH | 5 | 5 | 0 | 100 | 36 | |
3c | EtOH | 7 | 7 | 0 | 100 | 52 | |
3a | MeCN | 53 | 50 | 3 | 94 | 376 | |
3b | MeCN | 23.5 | 23 | 0.5 | 98 | 167 | |
3c | MeCN | 36 | 35 | 1 | 97 | 247 | |
3a | EtOH | 65.5 | 64 | 1.5 | 98 | 459 | |
3b | EtOH | 13.5 | 13 | 0.5 | 96 | 91 | |
3c | EtOH | 16.5 | 16 | 0.5 | 97 | 111 | |
3a | MeCN | 72 | 69 | 3 | 96 | 499 | |
3b | MeCN | 64.5 | 62 | 2.5 | 96 | 449 | |
3c | MeCN | 45.5 | 44 | 1.5 | 97 | 276 | |
3a | EtOH | 59 | 57 | 2 | 97 | 474 | |
3b | EtOH | 11 | 11 | 0 | 100 | 76 | |
3c | EtOH | 15 | 15 | 0 | 100 | 108 | |
3a | MeCN | 62 | 60 | 2 | 96 | 436 | |
3b | MeCN | 48.5 | 47 | 1.5 | 97 | 336 | |
3c | MeCN | 44 | 43 | 1 | 99 | 222 | |
3a | EtOH | 100 | 95 | 5 | 95 | 697 | |
3b | EtOH | 54.5 | 54 | 0.5 | 100 | 380 | |
3c | EtOH | 69.5 | 69 | 0.5 | 100 | 473 | |
3a | MeCN | 96 | 94 | 2 | 97 | 675 | |
3b | MeCN | 89 | 87 | 2 | 98 | 618 | |
3c | MeCN | 75 | 74 | 1 | 99 | 512 | |
3af | MeCN | 41 | 32 | 9 | 79 | 143 | |
3bf | MeCN | 3 | 3 | 0 | 100 | 12 | |
3cf | MeCN | 18 | 13 | 5 | 71 | 63 |
Not surprisingly, high TOFs were obtained for the sulfoxidation of n-decyl methyl sulfide in ethanol and acetonitrile with each of the catalysts tested as this substrate is electron-rich and consequently easy to oxidise; as such it is not a reliable candidate for differentiating catalyst performance. The moderate to low conversions obtained for the [PO4{WO(O2)2}4]@ImPIILP-catalysed sulfoxidation of dibenzothiophene at room temperature in acetonitrile are consistent with the widely accepted electrophilic pathway and the lower nucleophilicity of this substrate; a recent computational study also supports this pathway50 as do numerous reports of increasing rates of oxidation with increasing nucleophilicity of the sulfide.20b,27d,48c,49 The TOF of 143 mol product per (mol cat) per h obtained with 3a at room temperature is a significant improvement on that of 9.6 mol product per (mol cat) per h for a Merrifield resin supported peroxomolybdenum(VI) catalyst at 78 °C,27d 25 mol product per (mol cat) per h for oxodiperoxomolybdenum(VI) immobilised onto ionic liquid modified SBA-15,30 4 mol product per (mol cat) per h for V2O5 in [C12mim][HSO4] at 45 °C (ref. 32) and 40 mol product per (mol cat) per h for a titanium cyclopentadienyl-silsesequioxane.17e Unfortunately, it was not possible to obtain reliable data for the sulfoxidation of dibenzothiophene in ethanol due to its low solubility in this solvent. Oxidation of allylphenyl sulfide and homoallylphenyl sulfide occurred with complete chemoselectivity for sulfoxide and sulfone with no evidence for epoxidation of the double bond; this is most likely due to the mild conditions and short reaction times.19b,20c,27b,d
Reassuringly, the optimum selectivities and TOFs in Table 3 either compete with or are an improvement on those of other immobilised polyoxo- or peroxometalate-based systems such as modified Merrifield resin supported peroxomolybdenum(VI),27g modified SBA-15-based tungstates,27a polyoxometalates hosted in layered double hydroxides,28 polymeric ionic liquid nanogel-anchored tungstates,40 a divanadium-substituted phosphotungstate supported on Fe2O3,48 poly(acrylonitrile)-immobilised peroxotungstate,27d tungstate-based poly(ionic liquid) entrapped magnetic nanoparticles31 and peroxotungstates immobilised on multilayer ionic liquid brushes-modified silica.27c We believe that catalysts 3a–c most likely operate via a three-step mechanism involving (i) rate determining attack of sulfide at polymer immobilised ionic liquid supported peroxotungstate (I) to afford (II), (ii) sulfoxide dissociation to generate tungsten-oxo (III) and (iii) catalyst regeneration (Fig. 4). As such it should therefore be possible to control factors that influence catalyst efficacy such as the accessibility of the active site, the electrophilicity of the active peroxotungstate and catalyst stability by modifying the ionic microenvironment of the polymer immobilised ionic liquid support or introducing additional functional groups and cross linking.
Fig. 4 Proposed mechanism for the peroxotungstate catalysed oxidation of sulfides with hydrogen peroxide. |
Fig. 5 Recycle study for the sulfoxidation of thioanisole in ethanol catalysed by [PO4{WO(O2)2}4]@ImPIILP (3a). |
Analysis of solvent collected during recovery of the catalyst from the first five runs revealed that the tungsten content was too low to be detected by ICP-OES (i.e. <1 ppm), a strong indication that the peroxotungstate was efficiently retained by the polymer immobilised ionic liquid. Moreover, analysis of catalyst recovered after the fifth run gave a tungsten content of 30.6% which is similar to that of the unused catalyst, a further indication that leaching was negligible. The IR spectrum of 3a contains bands at 1078 cm−1, ν(P–O), 941 cm−1, ν(WO), 588 cm−1, νasym(W–O2) and 529 cm−1, νasym(W–O2), which is a close match to those reported for related systems.51 A sample of catalyst recovered after run five contained IR bands that were essentially superimposable on those of fresh catalyst and an SEM image of the sample showed no significant morphological changes, indicating that the peroxotungstate is stable and remains intact under the reaction conditions; a copy of these IR spectra and the SEM image are provided in the ESI.‡ The gradual erosion in conversion on successive recycles is thought to be due to attrition during the filtration and catalyst recovery procedure rather than deactivation as the mass of catalyst recovered after the 12th run (0.011 g) is significantly less than the initial mass of catalyst (0.026 g) used in the first run. To this end, the turnover frequency of 619 calculated using the mass of catalyst recovered after run 12 is close to that of 654 obtained in the first run.
The configuration of the flow system is shown in Fig. 6 and is a based on a Uniqsis FlowSyn reactor. Preliminary optimisation studies were conducted using a segmented flow set-up in which 1 mL aliquots of thioanisole (0.2 M) in ethanol and 30% hydrogen peroxide (0.2–0.6 M) were simultaneously pumped through a reactor cartridge packed with 2.0 g of silica (Geduran® Si60 43–60 μm) mixed with 0.1 g of [PO4{WO(O2)2}4]@ImPIILP (3a) and using ethanol as the mobile phase; flow rates were varied with precise control between 0.146 and 8.8 mL min−1, which correspond to space velocities between 0.033 and 2.0 min−1, respectively, and residence times between 30 and 0.5 min, respectively. The exiting product stream was collected in triplicate as 2 mL aliquots, subjected to an aqueous work-up and analysed by either 1H or 13C NMR spectroscopy to determine the conversion and selectivity.
Fig. 6 Schematic representation of the reactor configuration for segmented and continuous flow sulfoxidation catalysed by [PO4{WO(O2)2}4]@ImPIILP (3a). |
A survey of the effect of space velocity (sv) on selectivity and conversion as a function of the H2O2:thioanisole ratio revealed that the optimum conversion–selectivity profile for the sulfoxidation of thioanisole at 25 °C with ethanol as the mobile phase was obtained with 1.5 equiv. of H2O2 at a space velocity of 0.1 min−1, details of which are shown in Fig. 7. Under these conditions, conversions increased gradually with decreasing space velocity from 8% for a space velocity of 2 min−1 to 88% when this was decreased to 0.1 min−1 while sulfoxide selectivity decreased slightly from 99% to 94% over the same time. Not surprisingly, when the reactor column was cooled to 0 °C the space velocity had to be reduced (sv > 0.017 min−1 corresponding to a residence time ca. 60 min) to reach acceptable conversions, albeit with no improvement in selectivity which remained at 94%. Although good conversions were obtained at shorter residence times when the column was heated to 50 °C this was at the expense of sulfoxide selectivity which dropped below 90%; full details of the effect of H2O2:substrate ratio and temperature on the conversion–selectivity profile are provided in the ESI.‡ Gratifyingly, the optimum conversion and sulfoxide selectivity compared favourably with that of 94% and 96% obtained in batch but with the advantage that a much lower H2O2:substrate ratio is required. Moreover, the catalyst cartridge could be stored overnight and reused with only a minor reduction in performance indicating that the system may be stable and suitable for use in continuous flow (vide infra).
The high selectivity and conversion obtained for the sulfoxidation of dibenzothiophene in acetonitrile under batch conditions prompted us to explore the potential for developing a continuous flow process for oxidative desulfurization of crude oil as the overwhelming majority of studies involving ionic liquids have focused on batch extraction based protocols.52 A survey of the conversion and selectivity as a function of residence time at 90 °C with acetonitrile as the mobile phase revealed that the concentration of sulfoxide peaked at a space velocity of 0.5 min−1, after which sulfone selectivity increased rapidly with increasing conversion, ultimately reaching 96% at a space velocity of 0.07 min−1 as shown in Fig. 8. Not surprisingly, much lower conversions were obtained at room temperature across the range of space velocities examined (Fig. S91‡). While this is most likely due to a temperature-rate affect we cannot rule out temperature dependent changes in the structure of the polymer affecting access of the substrate to the active site.
Encouraged by the promising conversion–selectivity profile achieved under segmented flow, a comparative continuous flow study was conducted using ethanol as the mobile phase; parallel reactions were also conducted with freshly prepared [NEt4]3[PO4{WO(O2)2}4] and Merrifield resin-derived 3e supported on silica as benchmarks. The continuous flow sulfoxidation of thioanisole was conducted by purging a catalyst column packed with a mixture of 3a and silica with a 0.2 M solution of thioanisole in ethanol and a 0.3 M solution of peroxide at a rate of 0.44 mL min−1 (sv = 0.1 min−1) at 25 °C and monitored over an 8 hour period by sampling 5 mL aliquots in triplicate. The resulting performance–time profile in Fig. 9a shows a slight decrease in conversion with time-on-stream from 87% to 76% while the sulfoxide selectivity remained relatively stable and constant at 92–94%. Interestingly, this conversion–selectivity profile is markedly more stable than its ROMP-derived counterpart in methanol which experienced a 30% drop in conversion and a concomitant reduction in sulfoxide selectivity from 77% to 53% after 8 h of continuous operation.41b A comparative life-time study conducted using a reactor cartridge packed with [NEt4]3[PO4{WO(O2)2}4] in silica was also undertaken to further assess the performance of our optimum POM@PIILP system. Under the same conditions [NEt4]3[PO4{WO(O2)2}4]/SiO2 was highly active for the sulfoxidation of thioanisole in ethanol during the first hour after which conversions dropped quite dramatically with time such that the system was completely inactive after 3 h; this was associated with efficient leaching of the peroxotungstate as quantified by ICP analysis (Fig. 9b). Having demonstrated that catalyst generated from in-house synthesised polymer immobilised ionic liquid outperformed that prepared from commercially available Merrifield resin modified with imidazolium ions for sulfoxidations conducted in batch, a performance–time profile was obtained under continuous flow operation in order to compare the efficiency of this system. Under the same conditions, a reactor column packed with Merrifield resin-derived 3e on silica showed a steady decrease in conversion from 65% to 47% together a minor decrease in selectivity from 94% to 91% (Fig. 9c). Although the drop in selectivity was relatively minor, the conversions are markedly lower than those obtained for 3a under the same conditions which highlights the advantages of developing polymer immobilised ionic liquid supports in house.
Finally, the reusability of the catalyst cartridge and the stable conversion–selectivity profile obtained under continuous flow prompted us to conduct a semi-quantitative scale-up and isolation experiment using ethanol as the mobile phase. Under optimum conditions 2.5 g of thioanisole was processed in 8 hours with a conversion of 82%, a sulfoxide selectivity of 92% and a total turnover number (TON) of 12040; this is a marked and significant improvement on the 52% conversion obtained with ROMP-derived peroxotungstate-based [PO4{WO(O2)2}4]@PIILP under similar conditions and in the same time.
Footnotes |
† This paper is dedicated to the memory of Professor Malcolm H. Chisholm (FRS). |
‡ Electronic supplementary information (ESI) available: Synthesis and characterisation of imidazolium-based monomers 1a–c, co-polymers 2a–c, polymer immobilized peroxotungstates 3a–e, TGA and DSC curves for 2a–c and 3a–c, SEM images, FTIR traces and X-ray photoelectron spectra for 3a–c, characterisation data for sulfoxides and sulfones, details of catalysis, recycle experiments and graphs showing conversion–selectivity profiles as a function of residence time for [PO4{WO(O2)2}4]@PIILP-catalysed sulfoxidations under segmented and continuous flow. See DOI: 10.1039/c6ra11157b |
This journal is © The Royal Society of Chemistry 2016 |