Molybdenum-catalyzed enantioselective sulfoxidation controlled by a nonclassical hydrogen bond between coordinated chiral imidazolium-based dicarboxylate and peroxido ligands

molecules Article Molybdenum-Catalyzed Enantioselective Sulfoxidation Controlled by a Nonclassical Hydrogen Bond between Coordinated Chiral Imidazolium-Based Dicarboxylate and Peroxido Ligands Carlos J. Carrasco ID , Francisco Montilla * ID and Agustín Galindo * ID Departamento de Química Inorgánica, Universidad de Sevilla, Aptdo. 1203, 41071 Sevilla, Spain; [email protected] *Correspondence: [email protected] (F.M.); [email protected] (A.G.) Received: 22 May 2018; Accepted: 28 June 2018; Published: 30 June 2018   Abstract: Chiral alkyl aryl sulfoxides were obtained by molybdenum-catalyzed oxidation of alkyl aryl sulfides with hydrogen peroxide as oxidant in mild conditions with high yields and moderate enantioselectivities. The asymmetry is generated by the use of imidazolium-based dicarboxylic compounds, HL R . The in-situ-generated catalyst, a mixture of aqueous [Mo(O)(O 2 ) 2 (H 2 O) n ] with HL R as chirality inductors, in the presence of [PPh 4 ]Br, was identified as the anionic binuclear complex [PPh 4 ]{[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L R )}, according to spectroscopic data and Density Functional Theory (DFT) calculations. A nonclassical hydrogen bond between one C–H bond of the alkyl R group of coordinated (L R ) − and one oxygen atom of the peroxido ligand was identified as the interaction responsible for the asymmetry in the process. Additionally, the step that governs the enantioselectivity was theoretically analyzed by locating the transition states of the oxido-transfer to PhMeS of model complexes [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L R )] − (R = H, i Pr). The ∆∆ G 6= is ca. 0 kcal · mol −1 for R = H, racemic sulfoxide, meanwhile for chiral species the ∆∆ G 6= of ca. 2 kcal · mol −1 favors the formation of (R)-sulfoxide. Keywords: sulfoxidation; asymmetric catalysis; molybdenum; hydrogen peroxide; Density Functional Theory 1. Introduction The synthesis and use of enantiopure sulfoxides is a topic of extraordinary interest in asymmetric synthesis, asymmetric catalysis and in the pharmaceutical industry [ 1 – 9 ]. The enantioselective sulfoxidation of prochiral sulfides is one of the most challenging approaches to chiral sulfoxides, and catalyzed processes based on metal complexes [ 3 , 4 , 10 – 16 ] and metal-free systems [ 17 ] have been described in the literature. Molybdenum-catalyzed enantioselective sulfoxidations have been investigated [ 18 – 25 ] and, in general, the Mo catalysts provided results that are somewhat lower than those of other metals, as for example titanium [ 26 – 32 ] or vanadium [ 33 – 41 ] complexes. However, we have recently demonstrated that the use of the imidazolium-based dicarboxylic compound (S,S)-1-(1-carboxy-2-methylpropyl)-3-(1-carboxylate-2-methylpropyl)imidazolium (HL iPr in Scheme 1), as inductor of chirality, in combination with oxidoperoxidomolybdenum complexes, afforded a system capable to achieve by kinetic resolution a value of 83% ee in the sulfoxidation of alkyl aryl sulfides [ 42 ]. This system is easily accessible, simple, environmentally friendly, and compatible with a green oxidant as aqueous hydrogen peroxide. In some cases, these advantages are not compatible with Molecules 2018,23, 1595; doi:10.3390/molecules23071595 www.mdpi.com/journal/molecules Molecules 2018,23, 1595 2 of 12 Ti or V catalysts. Following our recent research on Mo-catalyzed sulfoxidations [ 42 – 44 ], we report here the extension of our system [ 42 ] to other imidazolium-based dicarboxylic compounds, HL R , in order to improve its efficiency in the catalytic asymmetric oxidation of prochiral sulfides using aqueous hydrogen peroxide (Scheme 1). Moreover, spectroscopic data and Density Functional Theory (DFT) calculations have allowed identification of the nature of the molybdenum catalytic species, {[Mo(O)(O2)2(H2O)]2(µ-LR)}−, and the origin of the asymmetry in the sulfoxidation process. Molecules 2018, 23, x FOR PEER REVIEW 2 of 12 improve its efficiency in the catalytic asymmetric oxidation of prochiral sulfides using aqueous hydrogen peroxide (Scheme 1). Moreover, spectroscopic data and Density Functional Theory (DFT) calculations have allowed identification of the nature of the molybdenum catalytic species, {[Mo(O)(O2)2(H2O)]2(μ-LR)}−, and the origin of the asymmetry in the sulfoxidation process. Scheme 1. Enantioselective sulfoxidation with hydrogen peroxide in the presence of chiral inductors HLR. 2. Results and Discussion 2.1. Enantioselective Oxidation of Different Sulfides with Aqueous Hydrogen Peroxide Catalyzed by the System [Mo(O)(O2)2(H2O)n]/HLR/[PPh4]Br The optimization of the reaction conditions were performed with the (S,S)-HLiPr compound, 1c, and methyl phenyl sulfide, and were previously communicated [42]. Chloroform was used as solvent (1 mL) with a 1:1:0.025:2 ratio of methyl phenyl sulfide:H2O2:Mo-complex:[PPh4]Br. Reactions were carried out in a micro-reactor, at 0 °C during 1 h, on 1 mmol scale. A solution of MoO3 (2.5% mmol) in aqueous hydrogen peroxide, namely [Mo(O)(O2)2(H2O)n] (see Materials and Methods), in conjunction with 1c and tetraphenylphosphonium bromide was employed to in-situ generate the catalyst. In these conditions, a 94% of conversion with high selectivity to sulfoxide (95%) and 40% ee to the (R)-sulfoxide was obtained [42]. A number of additional imidazolium-based zwitterionic dicarboxylic acids were also tested as chiral inductors in the enantioselective oxidation of methyl phenyl sulfide (Table 1). They are derived both from natural α-amino acids of general formula (S,S)- HLR (R = Me, 1b; CH2Ph, 1d; iBu, 1e, (S)-sec-Bu, 1f) or non-natural α-amino acids, such as (R,R)-HLiPr (1c’) and (S,S)-HLtBu (1g) (Scheme 1). They were prepared by condensation of 2 equiv. of the corresponding amino acid with glyoxal and p-formaldehyde in water at 90 °C for one hour, following the procedures described in the literature [45,46]. Additionally, the new compounds (S,S)-HLtBu (1g) and (R,R)-HLiPr (1c’) were also straightforwardly obtained in an enantiopure form by the same procedure using the corresponding nonproteinogenic amino acids. Compounds 1g and 1c’ were characterized by IR, NMR (1H and 13C{1H}) and mass spectra (see Materials and Methods and Figures S1–S5 in Supplementary Materials). These compounds were employed to investigate the influence of (i) absence of chirality in the ligand (HLH, 1a); (ii) size and branching of alkyl substituents (1c–e,g); (iii) an additional chiral center (1f); and (iv) a chiral center with opposed sense of chirality (1c vs 1c’). Scheme 1. Enantioselective sulfoxidation with hydrogen peroxide in the presence of chiral inductors HL R . 2. Results and Discussion 2.1. Enantioselective Oxidation of Different Sulfides with Aqueous Hydrogen Peroxide Catalyzed by the System [Mo(O)(O2)2(H2O)n]/HLR/[PPh4]Br The optimization of the reaction conditions were performed with the (S,S)-HL iPr compound, 1c , and methyl phenyl sulfide, and were previously communicated [ 42 ]. Chloroform was used as solvent (1 mL) with a 1:1:0.025:2 ratio of methyl phenyl sulfide:H 2 O 2 :Mo-complex:[PPh 4 ]Br. Reactions were carried out in a micro-reactor, at 0 ◦ C during 1 h, on 1 mmol scale. A solution of MoO 3 (2.5% mmol) in aqueous hydrogen peroxide, namely [Mo(O)(O 2 ) 2 (H 2 O) n ] (see Materials and Methods), in conjunction with 1c and tetraphenylphosphonium bromide was employed to in-situ generate the catalyst. In these conditions, a 94% of conversion with high selectivity to sulfoxide (95%) and 40% ee to the (R)-sulfoxide was obtained [ 42 ]. A number of additional imidazolium-based zwitterionic dicarboxylic acids were also tested as chiral inductors in the enantioselective oxidation of methyl phenyl sulfide (Table 1). They are derived both from natural α-amino acids of general formula (S,S)-HLR(R = Me, 1b; CH2Ph, 1d ; i Bu, 1e , (S)-sec-Bu, 1f ) or non-natural α -amino acids, such as (R,R)-HL iPr ( 1c ’) and (S,S)-HL tBu ( 1g ) (Scheme 1). They were prepared by condensation of 2 equiv. of the corresponding amino acid with glyoxal and p-formaldehyde in water at 90 ◦ C for one hour, following the procedures described in the literature [ 45 , 46 ] . Additionally, the new compounds (S,S)-HL tBu ( 1g ) and (R,R)-HL iPr ( 1c ’) were also straightforwardly obtained in an enantiopure form by the same procedure using the corresponding nonproteinogenic amino acids. Compounds 1g and 1c ’ were characterized by IR, NMR ( 1 H and 13 C{ 1 H}) and mass spectra (see Materials and Methods and Figures S1–S5 in Supplementary Materials). These compounds were employed to investigate the influence of (i) absence of chirality in the ligand (HL H , 1a ); (ii) size and branching of alkyl substituents ( 1c – e , g ); (iii) an additional chiral center ( 1f ); and (iv) a chiral center with opposed sense of chirality (1c vs 1c’). Molecules 2018,23, 1595 3 of 12 Table 1. Enantioselective oxidation of different sulfides with the system [MoO(O 2 ) 2 (H 2 O) n ]/H 2 O 2 / HLR/[PPh4]Br a. Entry HLR Sulfide Conversion (%) b Selectivity to Sulfoxide (%) b Selectivity to Sulfone (%) b Sulfoxide Yield (%) Sulfoxide ee (%) and Configuration c 1HLH,1a PhMeS 93 95 5 88 Racemic 2(S,S)-HLMe,1b PhMeS 93 95 5 88 2 (R) 3(S,S)-HLiPr,1c PhMeS 94 95 5 89 40 (R) 4(R,R)-HLiPr,1c’PhMeS 95 95 5 90 42 (S) 5(S,S)-HLCH2Ph,1d PhMeS 67 100 0 67 5 (R) 6(S,S)-HLiBu,1e PhMeS 88 96 4 85 14 (R) 7(S,S)-HLsec-Bu,1f PhMeS 95 95 5 90 47 (R) 8(S,S)-HLtBu,1g PhMeS 92 96 4 88 32 (R) 9(S,S)-HLsec-Bu,1f (p-Me-C 6 H 4 )MeS 90 91 9 82 55 (R) 10 (S,S)-HLsec-Bu,1f (p-Cl-C6H4)MeS 89 96 4 85 44 (R) 11 (S,S)-HLsec-Bu,1f (p-Br-C6H4)MeS 91 87 13 79 51 (R) 12 (S,S)-HLsec-Bu,1f Ph(PhCH2)S 90 64 36 58 53 (R) 13 (S,S)-HLsec-Bu,1f Ph(HOCH 2 CH 2 )S 81 36 0 29 43 (S) a Reaction conditions: catalyst [MoO(O 2 ) 2 (H 2 O) n ] 0.025 mmol, HL R 0.0125 mmol, [PPh 4 ]Br 0.05 mmol, sulfide 1.0 mmol, solvent: Cl 3 CH 1.0 mL, oxidant: H 2 O 2 (30% aq.), oxidant:sulfide ratio 1:1, 1 h, T = 0 ◦ C. b Determined by Gas Chromatography (50 µ L of dodecane as the internal standard). c Determined by High-Performance Liquid Chromatography (HPLC, see details in Supplementary Materials). As expected, the [Mo(O)(O 2 ) 2 (H 2 O) n ]/HL R /[PPh 4 ]Br system was effective for the sulfoxidation of methyl phenyl sulfide with conversions ranging from 67%, for 1d (entry 5), to 93–95% for reagents 1a – c , 1f and 1g . In all cases, reactions proceeded with chemoselectivity with nearly quantitative sulfoxide yields. The nature of the chiral inductor HL R clearly controls enantioselectivity. The use of the achiral reagent 1a gave the expected racemic mixture (entry 1). When reagents 1b – g were employed, it was observed that an increase in the branching at the C α atom of the R group of the ligand seemed to have a beneficial effect on the enantioselectivity. Specifically, reactions performed with chiral ligands with unbranched alkyl groups, such as 1b and 1d (entries 2 and 5, respectively), gave rise to low ee values of 2% and 5%, respectively. Conversely, the use of ligands with branched alkyl groups, such as 1c , 1f and 1g (entries 3, 7 and 8, respectively), produced ee values higher than 30%. The highest ee was observed with the reagent 1f (47% ee) in which the additional chiral center could have a positive effect in the enantioselectivity. Importantly, the reaction performed with (R,R)-HL iPr ( 1c ’) (entry 4) gave an ee result comparable to that of its (S,S)-enantiomer, 1c , only with opposed sense of sulfoxide chirality. The adequate selection of the HL R inductor chirality controls the production of the sulfoxide enantiomer. Finally, the activity of the system was tested with other sulfide substrates using compound 1f as chiral inductor (entries 9–15). In general, good conversions and enantioselectivities close to 50% for the corresponding (R)-sulfoxide were found, with the exception of the sulfide Ph(HOCH 2 CH 2 )S, which showed lower values (29% sulfoxide yield and 43% ee for the (S) enantiomer, entry 13). Conversions obtained with 1f were similar to those found with 1c [ 42 ], but enantioselectivity values were slightly superior using 1f than 1c for the same substrates [42]. One equivalent of hydrogen peroxide per substrate was used in all experiments because formation of the corresponding sulfone was observed when two or more equivalents of the oxidant were employed [ 42 , 43 ]. As we previously communicated, the ee can be increased by kinetic resolution and the (R)-sulfoxide PhMeSO was obtained in 83% ee with a 1.6-fold excess of the oxidant [ 42 ]. To probe the kinetic resolution process in more detail, we performed the oxidation of racemic PhMeSO sulfoxide, under the same reaction conditions, varying the oxidant-to-substrate ratio (Figure 1). From the analysis of the variation of the enantiomeric excess with respect to the conversion of sulfoxide, it was possible to determinate a stereoselectivity factor Eof 2.8 (E= kS’/kR’, see Supplementary Materials for details) [ 47 ]. Therefore, one may conclude that the enantiomeric excess of the sulfoxide can be controlled by adjusting the degree of conversion (at the expense of the sulfoxide yield). Molecules 2018,23, 1595 4 of 12 Molecules 2018, 23, x FOR PEER REVIEW 4 of 12 the sulfide Ph(HOCH2CH2)S, which showed lower values (29% sulfoxide yield and 43% ee for the (S) enantiomer, entry 13). Conversions obtained with 1f were similar to those found with 1c [42], but enantioselectivity values were slightly superior using 1f than 1c for the same substrates [42]. One equivalent of hydrogen peroxide per substrate was used in all experiments because formation of the corresponding sulfone was observed when two or more equivalents of the oxidant were employed [42,43]. As we previously communicated, the ee can be increased by kinetic resolution and the (R)-sulfoxide PhMeSO was obtained in 83% ee with a 1.6-fold excess of the oxidant [42]. To probe the kinetic resolution process in more detail, we performed the oxidation of racemic PhMeSO sulfoxide, under the same reaction conditions, varying the oxidant-to-substrate ratio (Figure 1). From the analysis of the variation of the enantiomeric excess with respect to the conversion of sulfoxide, it was possible to determinate a stereoselectivity factor E of 2.8 (E = kS’/kR’, see Supplementary Materials for details) [47]. Therefore, one may conclude that the enantiomeric excess of the sulfoxide can be controlled by adjusting the degree of conversion (at the expense of the sulfoxide yield). Figure 1. Kinetic resolution of racemic PhMeSO with catalyst [MoO(O2)2(H2O)n]/1c/[PPh4]Br (CHCl3, 0 °C, sulfoxide:Mo ratio of 100:2.5): sulfoxide and sulfone yields and the ee of the (R)-sulfoxide versus the oxidant:substrate ratio. 2.2. Nature of the Molybdenum Catalyst and Origin of the Enantioselectivity With the purpose of gaining evidence about the nature of the molybdenum catalyst, the reaction of [Mo(O)(O2)2(H2O)n] with 2 equiv. of the sodium salt of (S,S)-HLiPr was carried out. On the basis of Infrared (IR), Nuclear Magnetic Resonance (NMR) and Mass Spectrometry (MS) data (see Experimental and Supplementary Materials), the binuclear formulation Na{[Mo(O)(O2)2(H2O)]2(μLiPr)} was proposed for the isolated yellow powder. Further confirmation came from DFT calculations, which were carried out at the B3LYP level of theory for the anion {[Mo(O)(O2)2(H2O)]2(μLiPr)}−, 2c (optimized structure shown in Figure 2). The computed IR spectrum of this anion fits well with the experimental one of complex Na{[Mo(O)(O2)2(H2O)]2(μ-LiPr)} (Figure S7, Supplementary Materials), thus supporting the proposed formulation. This allowed us to assign several IR absorptions of compound Na{[Mo(O)(O2)2(H2O)]2(μ-LiPr)}, as for instance the asymmetric and symmetric ν(COO) bands at 1611 and 1391 cm−1, respectively. This attribution gave a Δ(νCOOasym − νCOOsym) value of ca. 220 cm−1, which is compatible with the monodentate κ1-O coordination of the carboxylate group observed in the optimized structure. Besides the carboxylate absorptions, the oxido group generates a characteristic ν(Mo=O) band at 962 cm−1, while the peroxide ligands display distinctive ν(OO), νas[Mo(OO)] and νs[Mo(OO)] absorptions at 861, 643 and 582 cm−1, respectively, in the expected ranges for this ligand [48]. Figure 1. Kinetic resolution of racemic PhMeSO with catalyst [MoO(O 2 ) 2 (H 2 O) n ]/ 1c /[PPh 4 ]Br (CHCl 3 , 0 ◦ C, sulfoxide:Mo ratio of 100:2.5): sulfoxide and sulfone yields and the ee of the (R)-sulfoxide versus the oxidant:substrate ratio. 2.2. Nature of the Molybdenum Catalyst and Origin of the Enantioselectivity With the purpose of gaining evidence about the nature of the molybdenum catalyst, the reaction of [Mo(O)(O 2 ) 2 (H 2 O) n ] with 2 equiv. of the sodium salt of (S,S)-HL iPr was carried out. On the basis of Infrared (IR), Nuclear Magnetic Resonance (NMR) and Mass Spectrometry (MS) data (see Experimental and Supplementary Materials), the binuclear formulation Na{[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} was proposed for the isolated yellow powder. Further confirmation came from DFT calculations, which were carried out at the B3LYP level of theory for the anion {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} − , 2c (optimized structure shown in Figure 2). The computed IR spectrum of this anion fits well with the experimental one of complex Na{[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} (Figure S7, Supplementary Materials), thus supporting the proposed formulation. This allowed us to assign several IR absorptions of compound Na{[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )}, as for instance the asymmetric and symmetric ν (COO) bands at 1611 and 1391 cm −1 , respectively. This attribution gave a ∆ ( ν COO asym −ν COO sym ) value of ca. 220 cm −1 , which is compatible with the monodentate κ1 -O coordination of the carboxylate group observed in the optimized structure. Besides the carboxylate absorptions, the oxido group generates a characteristic ν (Mo=O) band at 962 cm −1 , while the peroxide ligands display distinctive ν (OO), νas [Mo(OO)] and νs [Mo(OO)] absorptions at 861, 643 and 582 cm −1 , respectively, in the expected ranges for this ligand [48]. In order to support the formulation of the Mo catalyst, the activity of the isolated complex Na{[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} was tested in the sulfoxidation reaction of methyl phenyl sulfide, under the optimized reaction conditions. The conversion (93%) and ee (42%) values achieved were completely similar to those observed when the catalytic species was in-situ formed [ 42 ], thus proving the nature of the catalyst as a binuclear {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L R )} − oxidodiperoxidomolybdenum(VI) species. Molecules 2018,23, 1595 5 of 12 Molecules 2018, 23, x FOR PEER REVIEW 5 of 12 In order to support the formulation of the Mo catalyst, the activity of the isolated complex Na{[Mo(O)(O2)2(H2O)]2(μ-LiPr)} was tested in the sulfoxidation reaction of methyl phenyl sulfide, under the optimized reaction conditions. The conversion (93%) and ee (42%) values achieved were completely similar to those observed when the catalytic species was in-situ formed [42], thus proving the nature of the catalyst as a binuclear {[Mo(O)(O2)2(H2O)]2(μ-LR)}− oxidodiperoxidomolybdenum(VI) species. Figure 2. Optimized structure of the {[Mo(O)(O2)2(H2O)]2(μ-LiPr)}− anion, 2c, and proposed formulation of the Mo catalyst species. Once the catalyst structure is known, the origin of the enantioselectivity was theoretically investigated. In principle, the simple κ1-O-carboxylate coordination of the chiral ligand (LiPr)− can be in contradiction with the experimentally observed asymmetric process because chiral inductors are usually bior polydentate ligands. However, the analysis of the optimized structure {[Mo(O)(O2)2(H2O)]2(μ-LiPr)}−, 2c, reveals two additional interactions. One is a hydrogen bond between the O–H from the water ligand and the noncoordinated oxygen atom of the carboxylate group of (LiPr)− (explicitly shown in Figure 2, O–H···O distance of 2.75 Å ). The second one is subtler and consists of a nonclassical hydrogen bond [49–51] between one C–H bond of the isopropyl group and one oxygen atom of one of the peroxido ligands (C–H···O distance of 2.51 Å ). This interaction is also present in other optimized complexes {[Mo(O)(O2)2(H2O)]2(μ-LR)}− (R = iBu, 2e; secBu, 2f; and tBu, 2g), while is weaker (for R = CH2Ph, 2d) or absent in complexes containing R = H, 2a, and Me, 2b, substituents (see Table 2 and optimized structures in Figure S10 in Supplementary Materials). Compounds {[Mo(O)(O2)2(H2O)]2(μ-LR)}− (R = iPr, 2c; iBu, 2e; secBu, 2f; and tBu, 2g) display C–H···O distances within the range 2.50–2.66 Å and C–H···O angles higher than 160° (Table 2), which are typical parameters of nonclassical C–H···O hydrogen bonds (cut-off values of distances <2.8 Å and angles >90°) [52–54]. Interestingly, these compounds are those in which an asymmetric process is observed (inductors 1c,e–g in Table 1), while for compounds without the C–H···O interaction, low or null activity is found (inductors 1a,b,d in Table 1). Table 2. Selected structural data for classical and nonclassical hydrogen bonds in optimized structures {[Mo(O)(O2)2(H2O)]2(μ-LR)}− 2. R (μ–LR)− Distances, Å Angles, ° C–H···O O–H···O C–H···O O–H···O H 2a - 1.814, 1.822 - 158 Me 2b >4 1.814, 1.831 - 158, 159 iPr 2c 2.509, 2.521 1.807, 1.817 168 159, 160 CH2Ph 2d 2.352 (C–Harom.), 3.570 1.803, 1.815 138 (C–Harom.), 111 160 iBu 2e 2.552, 2.558 1.796, 1.811 170, 171 160 secBu 2f 2.621, 2.656 1.793, 1.803 160 160, 161 tBu 2g 2.509, 2.521 1.807, 1.817 166 160 Figure 2. Optimized structure of the {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} − anion, 2c , and proposed formulation of the Mo catalyst species. Once the catalyst structure is known, the origin of the enantioselectivity was theoretically investigated. In principle, the simple κ1 -O-carboxylate coordination of the chiral ligand (L iPr ) − can be in contradiction with the experimentally observed asymmetric process because chiral inductors are usually bior polydentate ligands. However, the analysis of the optimized structure {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} − , 2c , reveals two additional interactions. One is a hydrogen bond between the O–H from the water ligand and the noncoordinated oxygen atom of the carboxylate group of (L iPr ) − (explicitly shown in Figure 2, O–H ··· O distance of 2.75 Å). The second one is subtler and consists of a nonclassical hydrogen bond [ 49 – 51 ] between one C–H bond of the isopropyl group and one oxygen atom of one of the peroxido ligands (C–H ··· O distance of 2.51 Å). This interaction is also present in other optimized complexes {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L R )} − (R = i Bu, 2e ; sec Bu, 2f ; and t Bu, 2g ), while is weaker (for R = CH 2 Ph, 2d ) or absent in complexes containing R = H, 2a , and Me, 2b , substituents (see Table 2and optimized structures in Figure S10 in Supplementary Materials). Compounds {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L R )} − (R = i Pr, 2c ; i Bu, 2e ; sec Bu, 2f ; and t Bu, 2g ) display C–H ··· O distances within the range 2.50–2.66 Å and C–H ··· O angles higher than 160 ◦ (Table 2), which are typical parameters of nonclassical C–H ··· O hydrogen bonds (cut-off values of distances <2.8 Å and angles >90 ◦ ) [ 52 – 54 ]. Interestingly, these compounds are those in which an asymmetric process is observed (inductors 1c , e – g in Table 1), while for compounds without the C–H ··· O interaction, low or null activity is found (inductors 1a,b,din Table 1). Table 2. Selected structural data for classical and nonclassical hydrogen bonds in optimized structures {[Mo(O)(O2)2(H2O)]2(µ-LR)}−2. R (µ–LR)−Distances, Å Angles, ◦ C–H···O O–H···O C–H···O O–H···O H2a - 1.814, 1.822 - 158 Me 2b >4 1.814, 1.831 - 158, 159 iPr 2c 2.509, 2.521 1.807, 1.817 168 159, 160 CH2Ph 2d 2.352 (C–Harom.), 3.570 1.803, 1.815 138 (C–H arom. ), 111 160 iBu 2e 2.552, 2.558 1.796, 1.811 170, 171 160 secBu 2f 2.621, 2.656 1.793, 1.803 160 160, 161 tBu 2g 2.509, 2.521 1.807, 1.817 166 160 With the aim of confirming that these interactions are responsible of the asymmetry, we have selected the model complexes [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L R )] − (R = H, 3a , and i Pr, 3c ), containing Molecules 2018,23, 1595 6 of 12 for simplicity only one molybdenum atom, and studied the step that controls the enantioselectivity. This is the oxido-transfer step, which follows a Sharpless-type outer-sphere concerted mechanism according to previous studies [ 43 ]. The oxygen atom transfer is produced by the nucleophilic attack of sulfide onto the peroxide ligand that cleaves the O–O bond with sulfoxide formation. This transition state (TS) reflects the interaction between the HOMO (Highest Occupied Molecular Orbital) of the sulfide substrate and the σ *(O–O) LUMO (Lowest Unoccupied Molecular Orbital) of peroxide and it is characterized by the approaching of sulfide reagent with associated elongation of the O–O linkage. Taking into account the presence of two peroxide ligands and two prochiral faces of the sulfide, four transition states have been located for the oxido-transfer. Figure 3shows two of the calculated TSs for [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L R )] − (R = H and iPr), while the other calculated TSs are shown in Figure S11 (Supplementary Materials). The four transition states optimized for the nonchiral [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L H )] − species, 3a , have the same Gibbs free energy ( ± 0.2 kcal · mol −1 ) with a barrier for the oxido-transfer step of ca. 35 kcal · mol −1 . The ∆∆ G 6= is ca. 0 kcal · mol −1 , which is compatible with the formation of racemic sulfoxide using 1a (entry 1, Table 1). By contrast, for chiral [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L iPR )] − species, 3c , there are two transition states, those that yield the R sulfoxide TS_c1 and TS_c4 , showing lower energies than TS_c2 and TS_c3 that afford the Ssulfoxide. The calculated ∆∆ G 6= of ca. 2 kcal · mol −1 is well suited for the asymmetric process observed using 1c (entry 3, Table 1). Molecules 2018, 23, x FOR PEER REVIEW 6 of 12 With the aim of confirming that these interactions are responsible of the asymmetry, we have selected the model complexes [Mo(O)(O2)2(H2O)(κ1-O-LR)]− (R = H, 3a, and iPr, 3c), containing for simplicity only one molybdenum atom, and studied the step that controls the enantioselectivity. This is the oxido-transfer step, which follows a Sharpless-type outer-sphere concerted mechanism according to previous studies [43]. The oxygen atom transfer is produced by the nucleophilic attack of sulfide onto the peroxide ligand that cleaves the O–O bond with sulfoxide formation. This transition state (TS) reflects the interaction between the HOMO (Highest Occupied Molecular Orbital) of the sulfide substrate and the σ*(O–O) LUMO (Lowest Unoccupied Molecular Orbital) of peroxide and it is characterized by the approaching of sulfide reagent with associated elongation of the O–O linkage. Taking into account the presence of two peroxide ligands and two prochiral faces of the sulfide, four transition states have been located for the oxido-transfer. Figure 3 shows two of the calculated TSs for [Mo(O)(O2)2(H2O)(κ1-O-LR)]− (R = H and iPr), while the other calculated TSs are shown in Figure S11 (Supplementary Materials). The four transition states optimized for the nonchiral [Mo(O)(O2)2(H2O)(κ1-O-LH)]− species, 3a, have the same Gibbs free energy (± 0.2 kcal∙mol−1) with a barrier for the oxido-transfer step of ca. 35 kcal∙mol−1. The ∆∆G≠ is ca. 0 kcal∙mol−1, which is compatible with the formation of racemic sulfoxide using 1a (entry 1, Table 1). By contrast, for chiral [Mo(O)(O2)2(H2O)(κ1-O-LiPR)]− species, 3c, there are two transition states, those that yield the R sulfoxide TS_c1 and TS_c4, showing lower energies than TS_c2 and TS_c3 that afford the S sulfoxide. The calculated ∆∆G≠ of ca. 2 kcal∙mol−1 is well suited for the asymmetric process observed using 1c (entry 3, Table 1). Figure 3. Calculated transition states for the oxido-transfer step from [Mo(O)(O2)2(H2O)(κ1-O-LR)]− complexes to PhMeS (R = H, TS_a1, top; and iPr, TS_c1, bottom). 3. Materials and Methods 3.1. General Synthetic reactions were carried out under aerobic conditions. Chemicals were obtained from commercial sources and used as supplied, while solvents were appropriately purified using standard procedures. Infrared spectra were recorded on a Perkin-Elmer FT–IR Spectrum Two spectrophotometer (pressed KBr pellets). NMR spectra were recorded at the Centro de Investigaciones, Tecnología e Innovación (CITIUS) of the University of Sevilla by using Bruker AMX300 or Avance III spectrometers with 13C{1H} and 1H shifts referenced to the residual solvent signals. All data are reported in ppm downfield from Si(CH3)4. The gas chromatograms (GC) were obtained using a Varian Chromatogram CP-3800 with nitrogen as the carrier gas. The chromatogram used a Figure 3. Calculated transition states for the oxido-transfer step from [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L R )] − complexes to PhMeS (R = H, TS_a1, top; and iPr, TS_c1, bottom). 3. Materials and Methods 3.1. General Synthetic reactions were carried out under aerobic conditions. Chemicals were obtained from commercial sources and used as supplied, while solvents were appropriately purified using standard procedures. Infrared spectra were recorded on a Perkin-Elmer FT–IR Spectrum Two spectrophotometer (pressed KBr pellets). NMR spectra were recorded at the Centro de Investigaciones, Tecnología e Innovación (CITIUS) of the University of Sevilla by using Bruker AMX-300 or Avance III spectrometers with 13 C{ 1 H} and 1 H shifts referenced to the residual solvent signals. All data are reported in ppm downfield from Si(CH 3 ) 4 . The gas chromatograms (GC) were obtained using a Varian Chromatogram CP-3800 with nitrogen as the carrier gas. The chromatogram used a Varian automatic injector, Molecules 2018,23, 1595 7 of 12 model CP-8410, flame ionization detector (FID), and an Agilent column, model CP-7502. The HPLC chromatograms were performed on an Agilent 1260 Infinity instrument with a Chiralpak IA column at a flow rate of 1.0 mL/min with AcOEt/heptane = 6/4 (v/v) and using a UV detector at 254 nm. For Ph(HOCH 2 CH 2 )SO sulfoxide, a flow rate of 0.5 mL/min with heptane/ i PrOH = 9/1 (v/v) was employed. The absolute configuration (reported in Table 1) was determined by comparing HPLC elution orders and the sign of the specific rotations with the literature data [ 14 , 15 ]. Polarimetry was carried out using a JASCO P-2000 Digital Polarimeter and the measurements were made at ca. 25 ◦ C (concentration of ca. 10 mg/mL). High-resolution mass spectra (HRMS) were carried out by using a Q-Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer from Thermo Scientific at the CITIUS of the University of Sevilla. 3.2. Synthesis of Chiral Imidazolium-Based Zwitterionic Dicarboxylic Acids HLR The syntheses of compounds (S,S)-HL R ( 1a – f ) have been previously described [ 45 , 46 ] and they were identified by comparison of their IR, NMR ( 1 H and 13 C{ 1 H}) and mass spectra with those previously reported (see Figure S6, Supplementary materials). (R,R)-1-(1-carboxy-2-methylpropyl)-3-(1-carboxylate-2-methylpropyl)imidazolium, (R,R)-HL iPr ( 1c ’). A solution of D-valine (10 g, 84 mmol) in water (25 mL) was reacted with glyoxal (4.80 mL, 40% w/w solution in water, 42 mmol) and formaldehyde (3.13 mL, 37% w/wsolution in water, 42 mmol) at 95 ◦ C for 2 h. Compound (R,R)-HLiPr, 1c ’, was obtained by removing the solvent under reduced pressure. Recrystallization from water yields 5.18 g (46%) of the product as light-brown solid. IR (KBr, cm −1 ): 3464 (br), 3166 (w), 3114 (m), 3046 (m), 2970 (s), 2935 (w), 2878 (m), 1686 (vs,br), 1548 (s), 1473 (m), 1392 (m), 1375 (m), 1344 (w), 1295 (w), 1265 (m), 1162 (s), 1120 (m), 1096 (m), 1015 (w), 977 (w), 912 (w), 871 (w), 838 (w), 760 (w), 712 (w), 652 (w). 1 H NMR (300 MHz, D 2 O): δ 0.91, 1.00 (d, 3 J HH = 6.6 Hz, 6H, CH(CH 3 ) 2 ), 2.55 (m, 2H, CH(CH 3 ) 2 ), 4.84 (d, 3 J HH = 7.8 Hz, 2H, CH i Pr), 7.68 (s, 2H, C 4 H/C 5 H), 9.13 (s, 1H, C 2 H). 13 C{ 1 H} NMR (75 MHz, D 2 O): δ 17.3, 18.4 (s, CH(CH 3 ) 2 ), 31.2 (s, CH(CH 3 ) 2 ), 69.8 (s, CH i Pr), 122.3 (s, C 4 H/C 5 H), 136.2 (s, C 2 H), 172.3 (s, CO). [α]25 D=− 106.5 (H 2 O). HRMS for C13H20N2O4: [M + 1]+requires m/z269.15, found m/z269.1492. (S,S)-1-(1-carboxy-2,2-dimethylpropyl)-3-(1-carboxylate-2,2-dimethylpropyl) imidazolium, (S,S)-HL tBu ( 1g ). A solution of L-tert-leucine (2 g, 15 mmol) in water (20 mL) was reacted with glyoxal (866 µ L, 40% w/wsolution in water, 8 mmol) and formaldehyde (566 µ L, 37% w/wsolution in water, 8 mmol) at 95 ◦ C for 4 h. Compound (S,S)-HL tBu , 1g , was obtained by removing the solvent under reduced pressure. Recrystallisation from water yields 1.83 g (82%) of the product as light-brown solid. IR (KBr, cm −1 ): 3452 (br), 3187 (m), 3160 (m), 3108 (m), 3038 (m), 2965 (s), 2915 (w), 2878 (w), 1686 (vs,br), 1553 (s), 1482 (s), 1447 (w), 1403 (m), 1375 (s), 1369 (m), 1353 (m), 1315 (m), 1268 (m), 1215 (m), 1159 (s), 1101 (m), 1051 (m), 1029 (w), 938 (m), 892 (m), 855 (m), 820 (w), 801 (w), 789 (m), 769 (m), 731 (s), 699 (m), 681 (m), 657 (m), 643 (m). 1 H NMR (300 MHz, CD 3 OD): δ 1.10 (s, 18H, C(CH 3 ) 3 ), 4.87 (s, 2H, CH t Bu), 7.75 (d, 4 J HH = 1.5 Hz, 2H, C 4 H/C 5 H), 9.49 (s, 1H, C 2 H). 13 C{ 1 H} NMR (75 MHz, CD 3 OD): δ 26.0 (s, C(CH 3 ) 3 ), 34.7 (s, C(CH 3 ) 3 ), 72.6 (s, CH t Bu), 122.3 (s, C 4 H/C 5 H), 137.3 (s, C 2 H), 169.6 (s, CO). [α]25 D= +144.4 (H2O). HRMS for C15H24N2O4: [M + 1]+requires m/z297.18, found m/z297.1804. 3.3. Preparation and Titration of [Mo(O)(O2)2(H2O)n] Solution Solutions of the aqua complex of oxidodiperoxidomolybdenum in aqueous hydrogen peroxide were prepared as previously described [ 55 ]. For the purpose of simplicity the solution is referred to in this work simply as aqueous [Mo(O)(O2)2(H2O)n]. The resulting aqueous solution of molybdenum complex has an excess of hydrogen peroxide. The addition of the 0.025 mmol of molybdenum species in the catalytic essays includes a supplementary amount of oxidant. In order to avoid the formation of sulfone product, one equivalent of 30% hydrogen peroxide per sulfide substrate should be used. Thus, freshly prepared [Mo(O)(O 2 ) 2 (H 2 O) n ] 0.25 M solutions were employed, which were conveniently titrated before each catalytic test. The titration was Molecules 2018,23, 1595 8 of 12 carried out as follows. To 10 mL of [Mo(O)(O 2 ) 2 (H 2 O) n ] solution was added H 2 SO 4 6M (10 mL) and the mixture diluted with water (25 mL). This solution was titrated with KMnO 4 ca. 0.2 M (previously standardized with Na 2 C 2 O 4 ). The mean of five titrations afforded a typical value of ca. 0.2 mmol of hydrogen peroxide per 100 µ L of solution. On this basis, the exact amount of hydrogen peroxide 30% employed in the catalytic test can be easily calculated. 3.4. Synthesis of Complex Na{[Mo(O)(O2)2(H2O)]2(µ-LiPr)} Over a solution of (S,S)-HL iPr (0.504 g, 1.88 mmol) in water (15 mL) was added dropwise a solution of NaHCO 3 (0.158 g, 1.88 mmol) in water (5 mL) and the mixture was stirred at room temperature until the evolution of CO 2 ceased (5–10 min). Over this solution was added [Mo(O)(O 2 ) 2 (H 2 O) n ] (15.03 mL, 0.25 M aqueous solution, 3.76 mmol) and the mixture was stirred at room temperature for 1 h. The resulting solution was evaporated to dryness affording a yellow powder identified as Na{[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L iPr )} (1.216 g, 96%). IR (KBr, cm −1 ): 3440 (br vs), 3136 (m), 2966 (s), 2935 (m), 2877 (m), 1611 (vs), 1563 (m), 1548 (m), 1468 (m), 1423 (m), 1391 (s), 1344 (m), 1258 (m), 1234 (w), 1155 (s), 1120 (m), 1025 (w), 962 (s), 861 (s), 750 (m), 712 (w), 643 (m), 582 (m), 537 (m). 1 H NMR (D 2 O, 300 MHz): δ 0.80 (br d, 6H, 2CH(CH 3 ) 2 ), 0.91 (br d, 6H, 2 × CH(CH 3 ) 2 ), 2.44 (br m, 2H, 2 × CH(CH 3 ) 2 ), 4.58 (br m, 2H, 2 × CH i Pr), 7.53 (s, 2H, 2 × CH, H 4 /H 5 ), 8.92 (s, 1H, CH, H 2 ). 13 C{ 1 H} NMR (D 2 O, 75 MHz): δ 17.5 (s, 2 × CH(CH 3 ) 2 ), 18.6 (s, 2 × CH(CH 3 ) 2 ), 31.1 (s, 2 × CH(CH 3 ) 2 ), 71.4 (s, 2 × CH i Pr), 122.0 (s, 2 × CH, C 4 /C 5 ), 135.8 (s, CH), 173.6 (s, 2 × COO). Electrospray ionization-MS: positive mode, found m/z269.15 (for HL iPr + 1, C 13 H 20 N 2 O 4 , 268.14) and 291.13 (for NaL iPr + 1, NaC 13 H 19 N 2 O 4 , 290.12). ESI-MS: negative mode, found m/z267.13 (for HL iPr -1, C 13 H 20 N 2 O 4 , 268.14), 445.01 (for MoO5LiPr-1, C13H20MoN2O9, 446.02). 3.5. General Procedure for Enantioselective Mo-Catalyzed Oxidation of Sulfides in the Presence of HLR The reactor (a 50 mL vial equipped with a Young valve and containing a stirrer flea) was charged with [Mo(O)(O 2 ) 2 (H 2 O) n ] (100 µ L, 0.25 M aqueous solution, 0.025 mmol), HL R (0.0125 mmol), [PPh 4 ]Br as specified (typically 0.05 mmol), the reaction solvent (1 mL), the oxidant (30% aqueous H 2 O 2 ; 1 mmol per sulfide substrate, see details above) and the sulfide substrate (1 mmol), in the aforementioned order. The reactor was sealed and maintained at the working temperature, with constant stirring (600 rpm) in a thermostatted bath for the duration of the reaction. Upon completion, the reaction mixture was treated with diethyl ether (10 mL) and then filtered with 0.45 µ m nylon syringe filter. The resulting solution was analyzed by GC (by adding 50 µ L of dodecane as the internal standard). Afterwards the solution was evaporated to dryness by using a rotavap. The resulting residue was then analyzed by HPLC (by adding 20 mL of ethyl acetate). 3.6. Computational Details The electronic structure and geometries of the model compounds [Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L R )] − ( R=H , 2a ; Me, 2b ; i Pr, 2c ; CH 2 Ph, 2d ; i Bu, 2e ; sec Bu, 2f ; and t Bu, 2g ) and [Mo(O)(O 2 ) 2 (H 2 O)( κ1 -O-L R )] − (R = H, 3a , and i Pr, 3c ) were computed using density functional theory at the B3LYP level [ 56 , 57 ]. The Mo atom was described with the LANL2DZ basis set [ 58 , 59 ] while the 6-31G(d,p) basis set was used for the C, N, O, S and H atoms. The transition states of the interaction of PhMeS with 3a and 3c , namely TSa1 – 4 and TSc1 – 4 , were located at the same level of theory. Geometries of all model complexes were optimized without symmetry constraints. Frequency calculations were carried out at the same level of theory to identify all of the stationary points as transition states (one imaginary frequency) or as minima (zero imaginary frequencies) and to provide the thermal correction to free energies at 298.15 K and 1 atm. The DFT calculations were performed using the Gaussian 09 suite of programs [ 60 ]. Coordinates of the optimized compounds are collected in Table S4 (Supplementary Materials). Molecules 2018,23, 1595 9 of 12 4. Conclusions A simple process for the enantioselective Mo-catalyzed sulfoxidation with aqueous hydrogen peroxide, by using imidazolium-based dicarboxylate compounds HL R , 1b – g , as chiral inductors, has been developed. The advantages of this system are: (i) better reaction times (1 h); (ii) commercial and cheap molybdenum starting material, MoO 3 ; and (iii) straightforward synthesis of the (S,S)- or (R,R)-HL R inductors, in comparison with the elaborated chiral ligands reported in the bibliography. By combination of spectroscopic data and DFT calculations, the binuclear anion {[Mo(O)(O 2 ) 2 (H 2 O)] 2 ( µ -L R )} − , 2 , has been proposed as the chiral catalytic species. A nonclassical hydrogen bond between one C–H bond of the alkyl R group and one oxygen atom of one of the peroxido ligands controls the enantioselectivity of the sulfoxidation. This subtle interaction is only present in optimized complexes 2c , e , f , g , those that showed an acceptable ee value. This has been additionally demonstrated by analysing the transition states of the oxido-transfer of model complexes [Mo(O)(O2)2(H2O)(κ1-O-LR)]−(R = H, 3a, and iPr, 3c) to PhMeS sulfide. Supplementary Materials: The following are available online, Figures S1–S6: NMR and MS spectra of 1 compounds, Figure S7: calculated IR spectrum of 2c , Figures S7 and S8: determination of the stereoselectivity factor, Figures S10 and S11: optimized structures of transition states and compounds 2 , Figure S12: selected chiral HPLC diagrams of optical active sulfoxides, Table S1: energies of the transition states for the oxido-transfer, and Table S2: Coordinates of the optimized structures. Author Contributions: C.J.C., F.M., and A.G. designed the experiments; C.J.C. and F.M. performed the experiments; A.G. designed the theoretical analysis and performed the theoretical calculations; C.J.C., F.M., and A.G. wrote the manuscript. Funding: This research was funded by Junta de Andalucía (Proyecto de Excelencia FQM-7079) and Universidad de Sevilla (VI Plan Propio). Acknowledgments: Financial support is gratefully acknowledged. We thank to the Centro de Servicios de Informática y Redes de Comunicaciones (CSIRC), Universidad de Granada, for providing the computing time. Conflicts of Interest: The authors declare no conflict of interest. References 1. Patai, S.; Rappoport, Z. (Eds.) Syntheses of Sulphones, Sulphoxides and Cyclic Sulphides; John Wiley & Sons, Ltd.: Chichester, UK, 1995; ISBN 9780470666357. 2. Wojaczy´nska, E.; Wojaczy´nski, J.; Wojaczy´nska, E. Enantioselective Synthesis of Sulfoxides: 2000–2009. Chem. Rev. 2010,110, 4303–4356. [CrossRef] [PubMed] 3. O’Mahony, G.E.; Ford, A.; Maguire, A.R. Asymmetric oxidation of sulfides. J. Sulfur Chem. 2012 ,34, 301–341. [CrossRef] 4. Kagan, H.B. Asymmetric Synthesis of Chiral Sulfoxides. In Organosulfur Chemistry in Asymmetric Synthesis; Toru, T., Bolm, C., Eds.; Wiley: Weinheim, Germany, 2008; pp. 1–30. 5. Bentley, R. Role of sulfur chirality in the chemical processes of biology. Chem. Soc. Rev. 2005 ,34, 609–624. [CrossRef] [PubMed] 6. O’Mahony, G.E.; Kelly, P.; Lawrence, S.E.; Maguire, A.R. Synthesis of enantioenriched sulfoxides. Arkivoc 2011,2011, 1–110. [CrossRef] 7. Fernández, I.; Khiar, N. Recent developments in the synthesis and utilization of chiral sulfoxides. Chem. Rev. 2003,103, 3651–3705. [CrossRef] [PubMed] 8. Otocka, S.; Kwiatkowska, M.; Madali´nska, L.; Kiełbasi´nski, P. Chiral Organosulfur Ligands/Catalysts with a Stereogenic Sulfur Atom: Applications in Asymmetric Synthesis. Chem. Rev. 2017 ,117, 4147–4181. [CrossRef] [PubMed] 9. Han, J.; Soloshonok, V.A.; Klika, K.D.; Drabowicz, J.; Wzorek, A. Chiral sulfoxides: Advances in asymmetric synthesis and problems with the accurate determination of the stereochemical outcome. Chem. Soc. Rev. 2018,47, 1307–1350. [CrossRef] [PubMed] 10. Srour, H.; Le Maux, P.; Chevance, S.; Simonneaux, G. Metal-catalyzed asymmetric sulfoxidation, epoxidation and hydroxylation by hydrogen peroxide. Coord. Chem. Rev. 2013,257, 3030–3050. [CrossRef]