Volume 15, Issue 1 e202201072
Research Article
Open Access

A Common Active Intermediate in the Oxidation of Alkenes, Alcohols and Alkanes with H2O2 and a Mn(II)/Pyridin-2-Carboxylato Catalyst

Dr. Johann B. Kasper

Dr. Johann B. Kasper

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Search for more papers by this author
Dr. Pattama Saisaha

Dr. Pattama Saisaha

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Search for more papers by this author
Maurits de Roo

Maurits de Roo

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Search for more papers by this author
Mitchell J. Groen

Mitchell J. Groen

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Search for more papers by this author
Laia Vicens

Laia Vicens

Institut de Química Computacional i Catàlisi (IQCC) and Departament de Química, Universitat de Girona, Campus Montilivi, Girona, E-17071, Catalonia Spain

Search for more papers by this author
Dr. Margarida Borrell

Dr. Margarida Borrell

Institut de Química Computacional i Catàlisi (IQCC) and Departament de Química, Universitat de Girona, Campus Montilivi, Girona, E-17071, Catalonia Spain

Search for more papers by this author
Dr. Johannes W. de Boer

Dr. Johannes W. de Boer

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Search for more papers by this author
Prof. Ronald Hage

Prof. Ronald Hage

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Catexel B.V., BioPartner Center Leiden, Galileiweg 8, 2333BD Leiden, The Netherlands

Search for more papers by this author
Prof. Miquel Costas

Prof. Miquel Costas

Institut de Química Computacional i Catàlisi (IQCC) and Departament de Química, Universitat de Girona, Campus Montilivi, Girona, E-17071, Catalonia Spain

Search for more papers by this author
Wesley R. Browne

Corresponding Author

Wesley R. Browne

Molecular Inorganic Chemistry, Stratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, 9747AG Groningen, The Netherlands

Search for more papers by this author
First published: 07 November 2022

Graphical Abstract

Manganese oxidation chemistry: In line solution and head space Raman spectroscopy combined with 18O atom tracking reveal a common reactive species in the manganese-pyridine carboxylic acid catalysed oxidation of alkenes, alkanes and alcohols and unexpected disproportionation of H2O2.

Abstract

The mechanism and the reactive species involved in the oxidation of alkenes, and alcohols with H2O2, catalysed by an in situ prepared mixture of a MnII salt, pyridine-2-carboxylic acid and a ketone is elucidated using substrate competition experiments, kinetic isotope effect (KIE) measurements, and atom tracking with 18O labelling. The data indicate that a single reactive species engages in the oxidation of both alkenes and alcohols. The primary KIE in the oxidation of benzyl alcohols is ca. 3.5 and shows the reactive species to be selective despite a zero order dependence on substrate concentration, and the high turnover frequencies (up to 30 s−1) observed. Selective 18O labelling identifies the origin of the oxygen atoms transferred to the substrate during oxidation, and is consistent with a highly reactive, e. g., [MnV(O)(OH)] or [MnV(O)2], species rather than an alkylperoxy or hydroperoxy species.

Introduction

The activation of small molecule oxidants such as O2 and H2O2 by transition metal-based catalysts to form reactive metal-based oxidants is important in the industrial production of fine and bulk chemicals as well as in the transition towards green and sustainable synthetic methods.1, 2 Optimizing the reactivity of transition metal catalysts through fine tuning of the coordination sphere (ligand design) is key in increasing selectivity, especially in enantioselective reactions, where the first and, more recently, second coordination sphere play a key role.3-7 However, the cost of ligand synthesis can be prohibitive for large scale processes and hence methods based on simple metal salts and readily available reaction components are favoured, e. g., those based on manganese salts and tungsten oxide based epoxidation catalysts.8-10 Despite the operational simplicity these systems provide through in situ formation of the catalyst, understanding the mechanisms involved is essential for fine tuning reactivity. However, even when the catalyst is used in relatively high concentrations, the actual reactive species involved can be just a small fraction of the total metal content.11

An in situ prepared catalyst (Scheme 1) for epoxidation and syn-dihydroxylation of alkenes, comprising of a manganese(II) salt, pyridine-2-carboxylic acid (PCA) and a base (e. g., NaOAc, NaOH etc.), was reported over a decade ago for the oxidation of alkenes affording both syn-diol and epoxide products in good yields and with good efficiency in regard to the terminal oxidant, H2O2.12, 13 Later, this catalyst was applied in the oxidation of secondary alcohols and hydroxylation of alkanes.14 In many cases, turnover frequencies (TOF) of up to 30 s−1 and turnover numbers (TON) >300,000 were achieved at ambient temperatures with a wide substrate scope.15, 16

Details are in the caption following the image

Oxidation of alkenes and alcohols with the MnII pyridine-2-carboxylic acid base catalyst prepared in situ.

The ketone is a key component in regard to catalytic activity by forming adducts with H2O2,15 which are the effective terminal oxidant in the system.15 Although the solvent scope of the reaction was initially limited to acetone and 2-butanone for this reason, ketones, such as 1,1,1-trifluoroacetone and butanedione, that could be used substoichiometrically were identified later,13 broadening the solvent scope to include even the apolar solvent mixtures needed for polyolefin oxidation.15, 17

Mechanistic studies have established a broad understanding of the various roles played by the various reaction components (Scheme 2).15, 16, 18 A non-ordinal dependence on the concentration of manganese and pyridine-2-carboxylic acid was noted, with the latter only affecting the rate of reaction when less than 3 equiv. is used with respect to manganese, irrespective of the latter's concentration.16 Of importance for mechanistic studies discussed below, the reaction shows a zero order dependence on substrate for all transformations due to the macroscopic rate determining step being the interaction of the catalyst with the hydroperoxy-hydroxy adduct to form the reactive species (Scheme 2).13, 16 Despite this, in multifunctional substrates, selective oxidation of alkenes is observed over oxidation of alcohols and similarly alcohol oxidation is observed predominately even when C−H oxidation is possible.15

Details are in the caption following the image

Overall mechanism of oxidation of substrates with the catalyst prepared in situ from MnII and pyridine-2-carboxylic acid. The structures of the resting state and active species are unknown.

These studies have provided a useful working model, which enables prediction of the behaviour of the reaction, when mixtures of oxidisable functional groups are present, and of reaction rates. However, insight into the nature of the reactive species itself is limited primarily due to the low concentrations of the catalyst used (<0.1 mM) under reaction conditions, which make the detection of catalytically active species spectroscopically unfeasible. Increasing catalyst concentration above 0.5 mM to overcome this limitation has a negative impact on catalyst performance.16 Furthermore, it is not certain that the same species is responsible for all oxidative transformations observed, i. e. alkene epoxidation and syn-dihydroxylation, C−H oxidation of alkanes and alcohol and ketone oxidation. A Mn(III)/Mn(V) cycle seems plausible, with a Mn(III) resting state given the absence of X-band EPR signals and visible absorption.16 Indeed such a cycle was proposed by Che and coworkers in cis-dihydroxylation of alkenes with Oxone and [MnII(S,S-BQCN)Cl2], where BQCN is N,N‘-dimethyl-N,N‘-bis(8-quinolyl)cyclohexane-diamine. In that study a cis-dioxomanganese(V) (Mn(V)O2) intermediate was identified by ESI-MS.19

In the present report, the nature of the reactive species, in the manganese pyridine-2-carboxylic acid system (Scheme 1), is investigated through a combination of reaction progress monitoring and isotope labelling to track oxygen from various sources in the system. The data indicate that a single reactive manganese species engages in the oxidation of all substrates and the origin of the oxygen atoms transferred to the substrate during oxidation, is consistent with a highly reactive but nevertheless selective, e. g., [MnV(O)2], species rather than an alkylperoxy or hydroperoxy species.

Results

Reactions were carried out using conditions described earlier16 and at ambient temperatures (Scheme 1). Specifically, a minor excess of H2O2 (1.5 equiv.) was used with respect to the substrate to ensure full conversion was reached and a catalyst loading (0.1 mol % Mn salt and 0.5 mol % pyridine-2-carboxylic acid) which allowed for high turnover numbers (1000) within a few minutes. Reaction progress was monitored in-line by Raman spectroscopy to monitor substrate conversion and consumption of oxidant and ketone and by headspace Raman spectroscopy to determine O2 evolution.20 Headspace FTIR absorption spectroscopy was employed for the detection of CO2 formed in the reaction. 18O labelling of water, butanedione and H2O2 allowed for atom tracking with off-line analysis by GC-MS.

Competition reactions with alkenes and alcohols

The oxidation of alkenes, alcohols and alkanes with the MnII/PCA catalyst proceeds in all cases with a zero order dependence on substrate concentration,16 and the observed rate of oxidation is the same for all substrates, as shown for example in the oxidation of styrene and of 1-phenyl ethanol (Figure S3). Despite this, the oxidation is highly chemoselective with the oxidation of alkenes proceeding cleanly in the presence of alcohols and benzylic motifs.15

The selectivity is exemplified in the oxidation of 1 : 1 mixtures of styrene and 1-phenyl ethanol (Figure 1). Conversion of styrene proceeds at the same rate with and without 1-phenyl ethanol (Figure 1b and S3). In contrast, the oxidation of 1-phenyl ethanol is slow until close to when styrene is nearly full consumed (Figure 1b), however thereafter the oxidation of 1-phenyl ethanol proceeds at the same rate as observed (Figure S3) in the absence of styrene. These data show that the reactive intermediate is selective towards alkene oxidation over alcohol oxidation (as seen earlier with bifunctional substrates such as 2-decen-1-ol)15 and indicate strongly that the oxidation of the two distinct functional groups is due to the same reactive intermediate.

Details are in the caption following the image

(a) Oxidation of a mixture of styrene (0.25 M) and 1-phenyl ethanol (0.25 M), see Scheme 1 for conditions, (b) concentration of products over time, (c) Raman spectra at selected times showing bands at 1255 and 1690 cm−1 of the epoxide and ketone products. (black) initial spectrum, (red) 3.5 min (blue) 7 min (grey) 22 min after addition of H2O2.

The ratio between epoxidation and syn-dihydroxylation is dependent on substrate. Electron-rich alkenes, such as styrene and cyclooctene, give predominantly epoxide and electron-poor alkenes, such as diethyl fumarate, give only diol, regardless of reaction conditions.15 Despite the differences in chemoselectivity, the same rate of conversion of styrene, cyclooctene and diethyl fumarate is observed under the same conditions, consistent with the zero-order dependence of the reaction on substrate concentration. In contrast with competition experiments with 1-phenylethanol, the rates of conversion in competition experiments, between 1 : 1 mixtures of styrene and cyclooctene, and of styrene and diethyl fumarate (Figure S4), are the same. Similarly, in the oxidation of a 1 : 1 mixture of H8-styrene and D8-styrene, the rate of conversion of both substrates is identical (i. e. a 2e KIE of 1, Figure S5).

Disproportionation of H2O2 and evolution of CO2

A slight excess of H2O2 is required to reach full substrate conversion under standard conditions, which is in part expected due to the concomitant oxidation of the butanedione to acetic acid noted earlier.16 However, although excessive disproportionation of H2O2 was not observed under reaction conditions, analysis of the headspace above the reaction mixture by Raman spectroscopy (λexc 785 nm) shows that O2 was evolved to a certain extent prompting further analysis (Figure 2). In the presence of substrate (styrene) a small amount of O2 was evolved and the time dependence of its evolution indicates that it becomes significant process only when near full conversion of styrene has occurred. In contrast, in the absence of styrene, O2 evolution begins immediately and to a much greater extent with almost all H2O2 converting to O2 within 30 s, indicating that despite the relatively high efficiency of the reaction, H2O2 disproportionation is a significant competing pathway. It should be noted that the butanedione present is also a substrate, however, its concentration is low in the presence of excess H2O2 due to formation of its hydroperoxy adduct.15, 21 Hence, the catalyst reacts preferentially with alkenes and then alcohols and alkanes and oxidation of H2O2 and butanedione occurs only where other substrates are not available.

Details are in the caption following the image

(a) O2 formation detected by headspace Raman spectroscopy (λexc 785 nm) before (start) addition of H2O2 (at 25 s), during (and at the end (end) of the reaction, (b) ratio of O2 and N2 bands over time after addition of H2O2 (at 25 s) in the presence (black) and absence (blue) of styrene under standard conditions (scheme 1). The spectra were normalised to the N2 Raman band. Note that the Mn(II) concentration used was 0.5 mM to increase reaction rate. See Figure 3 for conditions.

With H216O2, only 16O-16O was observed even with H218O and 18O labelled butanedione present (vide infra); i. e. Raman bands from 16O-18O or 18O-18O were not observed. It is of note from a mechanistic perspective that the intensity of the Raman bands of 18O-butanedione decrease upon addition of H216O2, and only the signal for 16O-butanedione recovers afterwards showing that although the 18O from butanedione is exchanging during catalysis (vide infra), its oxygen atoms are not incorporated into the O2 evolved (Figure 3).

Details are in the caption following the image

Concentrations of 18O-butanedione (red), 16O-butanedione (black) monitored by Raman spectroscopy (λexc 785 nm) and amount of 16O2 (blue) formed after 2 additions of H2O2 (at ca. 1 and 6 min) monitored by headspace Raman spectroscopy (λexc 785 nm) in a reaction mixture without added substrate (0.5 mM MnII-salt, 2.5 mM PCA, 5 mM NaOAc, 0.1 M AcOH, 0.25 M butanedione) and with initially 18O-(75 %)labelled butanedione.

Butanedione is partially consumed during the reaction forming acetic acid.15, 16 However, while a simple cleavage of the 2–3 carbon-carbon bond of butanedione can be envisaged, headspace FT-IR spectroscopy (Figure S6) indicates that decarboxylation to form CO2 occurs to a modest extent. Notably with 18O-labelled butanedione, the CO2 evolved was predominantly 16O=C=18O initially and towards the end of the reaction the proportion of 16O=C=16O present increases, consistent with exchange of oxygen atoms of butanedione during reaction of the hydroperoxy adduct with the manganese catalyst (Figure S7). Hence, in 18O labelling studies described below, all reactions are performed with sub-stoichiometric amounts of H2O2 with respect to butanedione.

Primary kinetic isotope effects with benzyl alcohols

Competition experiments between an alkene and alcohol confirm that the active species is less reactive towards alcohols. Since the oxidation of alcohols involves the cleavage of a C−H bond, determining the primary kinetic isotope effect (KIE) is of interest.22, 23 KIEs were determined with mixtures of 1-phenyl-ethanol/1-D1-1-phenyl-ethanol, and of benzyl-alcohol/D2-benzyl-alcohol (Scheme 3). A disadvantage of the intermolecular approach to determining KIEs is that the concentration ratio between the two substrates in each case will change over time. This complication is avoided in the oxidation of D1-benzyl-alcohol as here a single substrate allows for kH/kD to be determined from the product ratio of D-aldehyde and H-aldehyde formed. A further consideration in determining KIE from product ratios benzyl alcohol, is the further oxidation of the aldehyde products to the corresponding carboxylic acid, which is around 15–30 % under the conditions employed. However, in all cases the KIE determined was between 3 and 4.5, which is consistent with C−H bond breaking by a selective reactive intermediate.

Details are in the caption following the image

Substrates and conditions used to determine primary KIE in alcohol oxidation.

Atom tracking with 18O labelling

Since oxygen is incorporated into the product in the case of alkene oxidation, it is of interest to determine the source of these oxygen atoms. H2O2 is an obvious source, however, the substrate selectivity and KIE data indicate that a manganese species is responsible for oxygen atom transfer and therefore high valent Mn−OH and Mn=O species may play a role. Hence, exchange with H2O and thereby incorporation of oxygen from water added together with the H2O2 should be considered also.24 Less obvious sources of oxygen are the oxygen atoms in butanedione and the acetic acid, with the latter either added to the reaction mixture or formed in situ. Each of these sources of oxygen atoms are considered in turn with 18O labelling.

The formation of 3-hydroperoxy-3-hydroxybutanone from H2O2 and butanedione is reversible (scheme 2), which together with the potential exchange with water, presents challenges in 18O isotope labelling studies. 16O/18O exchange between water and butanedione proceeds relatively slowly even under neat conditions (vide infra) and is slow under the more dilute conditions of the reaction. Hence, exchange between 18O and 16O during the reaction was avoided by applying conditions in which the reaction is complete within 5 min, i. e. 0.1 mol % MnII (0.5 mM, Scheme 4). Furthermore, for tracking 18O incorporation in the epoxide and syn-diol products, it is essential to preclude multiple cycles of formation of the hydroperoxy-hydroxy adduct. Therefore, butanedione was present in two fold excess with respect to H2O2.

Details are in the caption following the image

Conditions used for 18O labelling studies.

As expected, 18O-labelled H2O2 shows incorporation of 18O into the epoxide and syn-diol products (Figure S8). Remarkably, only ca. 50 % of the epoxide formed contained 18O. On the other hand, irrespective of the substrate, the corresponding diol product showed only one 18O atom incorporated, with only minor variations in the specific percent incorporation between substrates (see Supporting information section 6, Tables S1 and S2 for details). The data confirm that there are two sources of oxygen atoms available for transfer from the catalyst. It is also evident from the epoxidation labelling data that the species responsible for oxygen atom transfer bears two oxygen atoms that are equivalent.

Although 50 wt % H2O2 is used and hence 0.2 M H216O is present in the reaction mixture by default, the base (NaOAc) is added as a solution in water allowing H218O to be introduced up to a concentration of 1.5 M. Despite the high H218O content, incorporation of 18O from H218O is not observed in either epoxide or syn-diol products (Figure S9). These data confirm that water is not a source of the oxygen transferred to the epoxide and diol products and hence the reactive catalytic species responsible for oxygen atom transfer does not readily exchange oxygen atoms with water. These data are reminiscent of the observations of Che et al. in stoichiometric syn-dihydroxylation with a Ru(IV) complex.25

Since the only other significant source of oxygen in the reaction mixture available are the oxygen atoms of butanedione present (0.25–0.5 M), the question arises as to its involvement and the rate of exchange of its oxygen atoms with water.

Incorporation of 18O in both the syn-diol and the epoxide products is observed with 18O labelled butanedione (Figure S10 and S11). After correction of the data for the percentage of 18O-labelling of butanedione (vide infra), it is apparent that oxygen from butanedione accounts for ca. 50 % of the oxygen incorporation in the epoxide product and one of the oxygen atoms present in the syn-diol product (Scheme 5). With cyclooctene the hydroxyketone (a product of oxidation of cyclooctan-1,2-diol) is observed and shows similar labelling ratios as the diol product (Table S3).

Details are in the caption following the image

Average 18O incorporation into 5 substrates using (top) H218O (middle) H218O2, and (bottom) 18O labelled butanedione (see Tables S1 and S2, and Figure S8-S11 for details) corrected for isotope composition of reagents.

The incorporation of 18O from butanedione and not from water is counterintuitive since 18O labelled butanedione is prepared simply by shaking water (H218O) with neat butanedione (vide infra). The lack of incorporation of 18O from water and the essentially stoichiometric incorporation from butanedione presents a dichotomy that prompted us to study the exchange between water and butanedione under reaction conditions in detail also.

Exchange of oxygen between water and butanedione

In the present system, the formation of 3-hydroperoxy-3-hydroxybutanone is a prior step in the formation of the reactive form of the catalyst (Scheme 2). Oxygen from butanedione is incorporated into the product(s) and hence originates from its hydroperoxide adduct. However, small variations in 18O incorporation are observed, which necessitates that the extent and rate of exchange of oxygen atoms between water and carbonyls of butanedione are determined also.

Oxygen exchange between carbonyls and water is well established,26, 27 however, the rate of exchange is highly dependent on conditions. In neat mixtures of water and butanedione exchange is slow (up to 20 min to reach equilibrium – see Figure S12–S14 and SI for discussion and additional data). The rate of exchange under reaction conditions is of particular concern in interpretation of the data obtained with 18O labelled butanedione. Butanedione was mixed with H218O (4 mol equiv.) with all other major components present at the same concentrations as analogous experiments with unlabelled water. The exchange of the carbonyl oxygen atoms was monitored over time by Raman spectroscopy (Figure S15). Equilibration was not reached until after 5 h. Given that the reactions described here proceed within several minutes, on this time scale significant 16/18O exchange between water and butanedione does not occur. Reactions in which H218O was added do not show significant incorporation of 18O in the products further supporting this conclusion (vide supra).

Exchange of H2O2 between butanedione and 3-hydroperoxy-3-hydroxybutanone

As a final point, 3-hydroperoxy-3-hydroxybutanone forms by reaction of H2O2 with butanedione almost immediately under the conditions employed (Scheme 6). However, it is possible that after one cycle (reaction of 3-hydroperoxy-3-hydroxybutanone with the manganese catalyst to form the reactive species) the butanedione recovered has one oxygen atom from H2O2. So for example if 18O18O-labelled butanedione is used with H216O2 then after one cycle 18O16O butanedione is present in solution. That this occurs is evident from the change in isotopic composition of CO2 evolved during the reaction with excess H2O2 (vide supra).

Details are in the caption following the image

Equilibria between butanedione and H2O2 and exchange of H2O2 between butanedione molecules.

However, when the amount of H2O2 used is less than that of butanedione (as is the case in most experiments described here) exchange of H2O2 bound to butanedione with other butanedione molecules in solution is negligible on the time scale of the reaction (see SI Figure S16 and S17 for details). Hence even though 18O16O-butanedione is recovered after each catalysed turnover this recovered butanedione does not engage further in the reactions.

Discussion

Although the concentration of the catalyst is below that amenable to spectroscopic detection of intermediates, a combination of competition experiments, headspace analysis and labelling studies provides considerable mechanistic insight into the catalytic system. Earlier studies have shown that the rate of reaction over a wide range of conditions16 has a zero order dependence on the concentrations of H2O2 and substrates, i. e. the observed reaction rate is the same for alkene substrates as for alcohols. Nevertheless, in multifunctional substrates, selective oxidation of alkenes is observed over oxidation of alcohols and similarly alcohol oxidation is observed predominately even when C−H oxidation is possible.15 The involvement of hydroxyl radicals or other O-based radicals (e. g., HOO⋅ or ROO⋅) can be excluded based on substrates sensitive to such species and that addition of (10 vol %) of CH2Br2 affects neither the oxidation of diethyl fumarate nor tetraline.12 Furthermore, in the syn-dihydroxylation of alkenes, the configuration of the double bond is retained fully in the syn-diol products but this is not the case in epoxide products of the same substrates.15 Hence, while radical intermediates cannot be excluded on the basis of retention of configuration of alkenes, a cationic intermediate in epoxidation should be considered.15, 16

The experimentally observed rate limiting step is the reaction of 3-hydroperoxy-3-hydroxybutanone with the catalyst and hence the oxidising species can react with substrate much more rapidly than the rate of its formation. Indeed, competition experiments between electronically distinct alkenes shows no difference in the rate of conversion, e. g., mixtures of styrene and diethyl fumarate, and the intermolecular secondary KIE is 1 for a mixture of H8-styrene and D8-styrene. Nevertheless, while the observed rate of reaction with styrene is the same as with 1-phenylethanol, competition experiments between them show preference for oxidation of styrene but that once styrene is near fully consumed the oxidation of the alcohol proceeds at the expected rate. These data strongly suggest also that it is the same reactive species that engages with both substrates during oxidation. A primary KIE of ca. 4 is observed in the oxidation of benzyl alcohols, consistent with selectivity, in regard to the substrate/functional group, occurring after the (macroscopic) rate determining step.

Under standard conditions the reaction shows generally good efficiency in the use of oxidant. Depending on substrate full conversion can be achieved with close to stoichiometric H2O2 with some losses due to oxidation of butanedione to acetic acid/CO2 also. However, when near full conversion of substrate is reached the rate of O2 evolution increases. In the absence of substrate O2 evolution due to H2O2 disproportionation as well as oxidation of butanedione is observed immediately and at the same rate as the oxidation of, e. g., alkenes, occurs. These data emphasize that H2O2 serves as a substrate for the reactive manganese species also and its disproportionation, following an initial hydrogen atom abstraction to form superoxide for example, is in competition with substrate conversion rather than due to additional reaction pathways. It is of note that the results of the labelling studies are reminiscent of the labelling studies with DMDO (dimethyl dioxirane) given that one oxygen atom from the ketone is incorporated in the dioxirane moiety, in which both oxygen atoms are then equal.28, 29 Decomposition of DMDO leads to release of O2 and the recovery of the ketone.30 However, in the present study the lack of 18O incorporation in the released O2 and the lack of retention of configuration of cis/trans alkenes (in contrast to that obtained with DMDO), point towards a metal based rather than a ketone based reactive species.

The release of singly labelled CO2 (16O=C=18O) from 18O-labelled butanedione into the headspace confirms that its oxidation involves decarboxylation of intermediates and not simply cleavage of the central carbon-carbon bond to form two equivalents of acetic acid. Again, the same catalytically active species is responsible for its oxidation as for the oxidation of alkenes and alcohols.

18O2 labelling provides indirect insight into the structure of the reactive species. With H218O2 only ca. 50 % of the epoxide product contains 18O and the syn-diol product is exclusively singly labelled (18O−16O). Therefore 50 % of the oxygen atoms in the products originates from another source, namely, butanedione, which provides ca. 50 % of oxygen atoms incorporated into the epoxide product and one oxygen atom to the syn-diol product. Fortuitously, in the present study, although the oxygen atoms of butanedione can exchange with those of water, the rate is sufficiently slow that it can be largely disregarded on the timescale of the reaction. Indeed, addition of H218O to the reaction mixture does not result in incorporation into either epoxide or syn-diol products.

These data together indicate the formation of an active species with two essentially equal oxygen atoms available for transfer to the substrate and is directly formed from 3-hydroperoxy-3-hydroxybutanone and the manganese catalyst.

Summary of mechanism

The data available here and in earlier studies16, 31 supports the mechanistic model shown in Scheme 7. Spectroscopic data (Raman and UV.Vis absorption) shows that the formation of the hydroperoxyadduct of the ketone, in this case butanedione, is rapid and the equilibrium lies heavily on the side of the adduct (Scheme 6). The observed rate of the reaction is dependent on the concentration of this species and not of the H2O2 (when in excess of butanedione and vice versa). The presence of acetic acid, although its role is not clear, is also necessary for the reaction to proceed. The precise structure of the catalyst remains elusive, however, the reaction rate is dependent on the concentration of manganese ions and curiously independent of the concentration of pyridine-2-carboxylic acid provided it is in at least 3-fold excess of the manganese ion concentration. The absence of EPR signals gives an indication that mononuclear solvated Mn(II) ions are not present under reaction conditions and it is inferred that the resting oxidation state is Mn(III), stabilised by the acidic conditions. Considering the nature of the reactive species, the absence of direct spectroscopic/structural evidence forces us to consider indirect evidence and in particular competition experiments and atom tracking by isotope labelling. The most notable observation is that oxygen from water present is not incorporated in the products of any of the reactions observed and hence any mechanisms or reactive species which can allow for exchange with water prior to oxygen atom transfer can be excluded. The fate of oxygen atoms in all sources in the reaction mixture are indicated in Scheme 7. Notably the 18O labelling data for the epoxide products indicate that both an oxygen atom from the butanedione as well as the H2O2 become chemically equivalent in the reactive species. All data considered, we can speculate that a [MnV(O)(OH)] or [MnV(O)2] intermediate, forms from a [MnIII-(3-hydroperoxy-3-hydroxybutanone)] species. This high valent manganese species is directly responsible for the oxidation of alkenes, alcohols, alkanes, butanedione and the net disproportionation of H2O2. The absence of oxygen originating from water also precludes exchange of water with the reactive species, which is understandable given the relatively slow exchange of oxido/hydroxide/aquo ligands in high-valent manganese species.32, 33

Details are in the caption following the image

Proposed mechanism including oxygen atom tracking with oxygen in H2O2 (red) and oxygen in butanedione (blue).

Conclusion

Elucidating the mechanism of catalytic reactions in which the catalyst is present at concentrations below the limit of detection for spectroscopic techniques presents a considerable challenge in establishing the nature of the species responsible for the reactivity observed. In this contribution, a combination of competition experiments and isotope labelling studies establish that it is highly likely that a single reactive species is responsible for all the oxidative transformations observed including the decomposition of H2O2. The confirmation that H2O2 itself as well as butanedione can be substrates for the catalyst, albeit that they react less readily than, e. g. alkenes, helps to understand how the reaction can be made more efficient in oxidant by maintaining an excess of substrate. The reactive species is likely to be a high valent oxido species that presents two equivalent oxygens for transfer to substrate or for hydrogen atom abstraction in the case of alkane and alcohol oxidation. Despite the relative simplicity of the catalyst in terms of ligand and that the concentrations at which catalysis is carried out, indirect methods can provide deeper insight into the nature of the reactive species and allow for comparison with less active but structurally more clearly defined catalysts based on more complex ligands.

Experimental Section

All reagents and solvents were obtained from commercial sources and used as received unless state otherwise. H218O2 (97 % 18O) was obtained from GiOxCat (Spain) as a solution in H216O. H218O (98 % 18O) was purchased from Rotem Industries (Israel), D2-benzyl alcohol was purchased from Aldrich. D1-benzyl alcohol and D1-1-phenylethanol were prepared by reduction of benzaldehyde and acetophenone with NaBD4 using standard procedures (SI). 18O labelled butanedione was prepared by mixing H218O with butanedione by shaking followed by decanting the water, see supporting information for details. The extent of exchange was determined by Raman spectroscopy. Raman spectra were recorded at 785 nm with a laser (500 mW, Cobolt lasers) fibre coupled to a Raman probe (Avantes). Raman scattering was detected with a fibre coupled shamrock163i spectrograph (Andor Technology) equipped with a 600 l/mm ruled grating blazed at 830 nm and a idus-420-OE CCD detector (Andor Technology). FTIR spectra were recorded using a FP4600 FT-NIR spectrometer (JASCO). UV/vis absorption spectra were recorded using a Specord600 UV/vis spectrometer (AnalyticJena). Samples were held in 1 cm pathlength quartz cuvettes with a cap sealed with a septum. 1H NMR spectra were recorded on a Bruker Avance (400 MHz). For 18O labelling studies, GC-MS spectral analyses were performed on an Agilent 7890 A gas chromatograph (HP-5MS column, 30 m×0.25 mm, 0.25 μm, Agilent J&W) interfaced with an Agilent 5975X mass spectrometer. NH3 was used as the ionization gas. Oxidation products were identified by comparison of their GC retention times with those of authentic compounds.

Acknowledgments

Financial support was provided by The Ministry of Education, Culture and Science of the Netherlands (Gravity Program 024.001.035 to W.R.B.), and the European Research Council (ERC 279549, W.R.B.). Support by the Spanish Ministry of Science, Innovation and Universities; PhD grant BES-2016-076349 to M.B. and PhD grant FPU16/04231 to L.V) and Generalitat de Catalunya (ICREA Academia Award to M.C. and 2017 SGR 00264) is acknowledged. STR of UdG are acknowledged for technical support.

    Conflict of interest

    The authors declare no conflict of interest.

    Data Availability Statement

    The data that support the findings of this study are available from the corresponding author upon reasonable request.