|Home | About | Journals | Submit | Contact Us | Français|
The mechanistic details of Mn(OAc)3-based oxidative free-radical additions and cyclizations are reviewed. The mechanisms of electron transfer to generate radicals, electron transfer to convert the radicals to oxidized products, and further oxidation of the products are covered.
The oxidative addition of acetic acid to alkenes by two equivalents of Mn(OAc)3 in AcOH at reflux to give γ-lactones was first reported in 1968 by Heiba and Dessau,1a and Bush and Finkbeiner.1b Over the past 40 years the use of Mn(OAc)3 to oxidatively initiate free radical reactions of mono and 1,3-dicarbonyl compounds has been extensively developed and widely reviewed with an emphasis on synthetic applications.2 These reactions are mechanistically complex with several competing pathways. Slight changes in the substrate or reaction conditions can produce major changes in the reaction pathway that can have a major impact on the products formed. The mechanistic details needed to predictably use these reactions are widely scattered. In this brief review, the various mechanisms of these reactions are presented with a focus on the one-electron oxidation initiation step that generates a free radical and the oxidation step that converts a free radical to the final product.
Fristad and Peterson extensively studied the mechanism of the oxidative addition of acetic acid to alkenes to form lactones.3 The hydrated form of Mn(OAc)3 that is usually used, Mn(OAc)3•2H2O, is an oxo-centered trimer of MnIII with bridging acetates.3 The rate determining step is the loss of a proton from a complexed acetate such as 1 to give a bis MnIII-enolate such as 2. This is followed by a rapid electron transfer with loss of MnII to form the complexed free-radical 3. The carbon-carbon bond forming step involves the addition of radical 3 to an alkene to give radical 4. The alkene is not involved in the rate determining step, which requires that a slow step occurs before formation of the carbon-carbon bond. Enolization is likely to be the rate determining step because the log of the rate of oxidation relative to that of acetic acid equals 0.344(ΔpKa) for five monosubstituted acetic acids covering a broad acidity range. Enolization appears to be irreversible since deuterium is not incorporated into 1 when the reaction is run in deuterated acetic acid.
The conversion of radical 4 to the final product lactone 8 also involves an oxidative electron transfer. The detailed mechanism of this oxidation is not known, but it is well-known that MnIII will not oxidize isolated secondary radicals to cations. Therefore the carboxylate group must be intimately involved in the oxidation step. One possibility is the oxidation of the radical by MnIII bound to the carboxylate to give cation 5 that cyclizes to 8. A second possibility involves addition of the radical to the oxygen of the carbonyl group to give radical 6, which should undergo electron transfer with loss of MnII to give lactone 8. A final possibility involves bonding of the MnIII to the radical to give metallocycle 7, which undergoes reductive elimination with loss of MnII to give lactone 8. Numerous variations of these mechanisms can also be considered.
Our studies of the oxidation of α-alkyl β-keto esters such as 9 with Mn(OAc)3 indicated that enolization is also the rate determining step (see equation 1).4 Formation of 10 by enolization is slow and electron transfer with loss of MnII to give radical 11 is rapid. The rate of reaction is therefore independent of alkene concentration or the nature of the tether in cyclizations. The geometry of radical 11 with the carbonyl group anti to the ester is inferred by analysis of the stereochemistry of the cyclic products formed. In collaboration with Dennis Curran, we carried out a comparison series of reactions in which radical 13 was obtained by oxidation of a series of β-keto esters 12 with Mn(OAc)3 or by atom transfer reaction from α-halo-β-keto esters with hexamethylditin (see equation 2).5 Comparable regio- and stereochemical results were obtained in all cases strongly suggesting that free radical 13, which is no longer complexed to manganese, is involved in the Mn(OAc)3-mediated oxidative cyclizations. Some differences in regiochemistry or stereochemistry between oxidative cyclizations and atom-transfer cyclizations would be expected if a metal-complexed radical were involved.
To our initial surprise, replacement of the methyl group of 9 with the second α-hydrogen atom of 15 changes the rate determining step in the mechanism. Enolization of 15 to give 16 is fast and reversible, and electron transfer to give the radical 18 is very slow and probably not relevant to product formation (see equation 3).4 The rate depends on alkene concentration and the rate-determining step is presumably the reaction of MnIII enolate 16 with the alkene to give radical 17 with loss of MnII. If addition of the alkene to the MnIII enolate is the rate-determining step, the length of the tether should, and does, affect the rate of oxidative cyclization of unsaturated α-unsubstituted β-keto esters.
Corey studied the oxidative free radical cyclization of a series of unsaturated β-keto acids 19 with Mn(OAc)3.6 The nature of the tether also affects the rate of oxidative cyclization of unsaturated β-keto acids. The time required for complete reaction ranges from 20 minutes to 24 hours as a function of the nature of the tether. Therefore, the mechanism for these reactions also involves a fast and reversible formation of enolate 20, followed by a slow, rate determining addition of the alkene to the manganese enolate to give radical 21 with loss of MnII. The oxidation of radical 21 to give lactone 22 is analogous to the oxidation of radical 4 to give lactone 8.
Why does the presence of the α-alkyl group have such a profound effect on the mechanism of the reaction? The rate of enolization will depend on the both the thermodynamic and kinetic acidity of the α-proton(s). Introduction of an electron-donating alkyl group decreases the thermodynamic acidity of the α-proton by 1-2 orders of magnitude. All α-substituents, regardless of their electronic character, will sterically retard the enolization. On the other hand, the α-alkyl group should facilitate the oxidation of 18 to 19. Electrochemical data for the oxidation of enolates of β-dicarbonyl compounds to the radical in DMSO indicates that an α-methyl group facilitates the oxidation by 0.25−0.4 V.7
The introduction of an α-alkyl substituent increases the pKa which decrease the rate of enolate formation and accelerates the oxidation by making the enolate easier to oxidize. This mechanism with rate determining enolate formation holds for monocarbonyl and less acidic 1,3-dicarbonyl compounds. For more acidic compounds such as α-unsubstituted β-keto esters and acids and β-diketones, enolization occurs readily and oxidation is slow. Mn(AcAc)3 is a stable commercially available MnIII enolate! For these compounds, the rate determining step appears to be the interaction of the MnIII enolate with an alkene to form a radical such as 17 or 20 with loss of MnII. Unfortunately, the addition of a MnIII enolate to an alkene to give a radical and MnII does not fit well with our arrow pushing model of organic chemistry making it hard to conceptualize this mechanism, which is required by the observation of rate dependence on alkene concentration or tether length. Mechanisms are still routinely drawn with intermediates such as 18 that are not consistent with experimental data.
The different mechanisms of these reactions can have profound effects on the yields of these reactions. For instance, treatment of 23 with Mn(OAc)3 and Cu(OAc)2 in AcOH at 25 °C for 2 hours affords 26 in 72% yield as a mixture of stereoisomers (see Scheme 2).8 A similar reaction with 27 affords a complex polymeric mixture. Reaction of 23 with MnIII should give enolate 24 rapidly. Loss of MnII from 24 to give the radical will be very slow because there is a hydrogen atom at the α-position of the enolate. The slow rate determining step is cyclization of the alkene onto the MnIII enolate with loss of MnII to give radical 25, which is oxidized by Cu(OAc)2 to give 26 as discussed below. On the other hand, 27 reacts with MnIII in a slower rate determining step to give α-methyl enolate 28. Rapid loss of MnII affords radical 29, which polymerizes rather than cyclizing through the higher energy conformation shown with the alkene in close proximity to the free radical.
In a study of Mn(OAc)3-based oxidative cyclization of α-arylalkylmalonates, Citterio suggested a variation of this mechanism.9 The reaction of malonate 30 to give 33 was shown to be first order in both 30 and MnIII as expected for the rate determining formation of enolate 31 from an α-substituted substrate (see Scheme 3). Loss of MnII from 31 to give radical 32 and cyclization of radical 32 to give 33 are rapid. However, the oxidative cyclization of 34a afforded 37a in 44% yield after 3 hours with 93% conversion of 34a whereas the oxidative cyclization of 34b afforded 37b in only 30% yield after 24 hours with only 60% conversion of 34b. This was explained by suggesting that the loss of MnII from enolate 35 to give radical 36 is reversible. Cyclization of 36 to form an indane is slow and the 4-methoxy group further retards this cyclization so that substrate dependence is observed.
The different mechanisms and their effect on the products formed show up very clearly in a comparison of the reactions of dimethyl 4-pentenylmalonate (38) and 4-pentenyl Meldrum's acid (47) (see Schemes 4 and and5).5). Enolization of dimethyl 4-pentenylmalonate (38) to give manganese enolate 39 is the slow, rate-determining step.10 Rapid loss of MnII from 39 generates radical 40, which cyclizes to give a ~9:1 mixture of cyclopentanemethyl radical 42 and cyclohexyl radical 43. Reaction of 42 with Cu(OAc)2 gives CuIII intermediate 41 which undergoes oxidative elimination to give methylenecyclopentane 44 and ligand transfer to give lactone 45. A similar oxidation converts 43 to cyclohexene 46. Reaction in AcOH (55 °C, 28 h) affords a 2.5:1 mixture of 45 and 44. Reaction in EtOH (60 °C, 156 h) is much slower, but yields a 2:3 mixture of 45 and 44, whereas reaction in DMSO (75 °C, 68 h) forms a 1:10 mixture of 45 and 44.10
Oxidative cyclization of 47 in EtOH with Mn(OAc)3 and 1 equivalent of Cu(OAc)2 for 10 minutes at 25 °C provides 8% of methylenecyclopentane 50, 65% of cyclohexene 53, and 4% of cyclohexene 54.11 Oxidation of cyclohexyl radical 52 with CuII selectively removes the least hindered proton to form mainly cyclohexene 53. As expected, oxidative cyclization of 47, with a half life in EtOH of < 5 minutes at 25 °C and 2 hours at −30 °C, is much faster than that of the less acidic dimethyl ester 38, which requires 156 hours at 60 °C for complete reaction. More surprisingly, 47 provides predominantly products derived from cyclohexyl radical 52, whereas 38 yields mainly products derived from cyclopentanemethyl radical 42 suggesting that these two reactions are mechanistically distinct. Furthermore, oxidative cyclization of the analogue of 47 with a hexyl group on the distal end of the double bond proceeds with a half-life of 2 hours at 25 °C. The change in half life from < 5 minutes with 47 to 2 hours with the addition of an alkyl substituent establishes that the double bond participates in the rate-determining step, which is cyclization of 48 to give 51 and 52. Loss of a proton from 47 to give MnIII enolate 48 should be rapid and reversible. Loss of MnII from enolate 48 to form radical 49 cannot be occurring because this reaction would proceed at the same rate regardless of substituents on the double bond.
Bausch's studies of the oxidation potential of the enolates of dimethyl malonate and Meldrum's acid in DMSO provide data that helps explain the differing behavior of malonate 38 and Meldrum's acid 47.12 The pKas of Meldrum's acid and dimethyl malonate in DMSO are 7.3 and 15.9, respectively, indicating that enolization of Meldrum's acid is favored by 11.8 kcal/mol (equations 5 and 6). On the other hand the oxidation potentials of the enolates are 1.16 and 0.77 V, respectively, indicating that it is 9 kcal/mol easier to oxidize the enolate of dimethyl malonate. Loss of a proton from 38 to give enolate 39 should be slow, and of MnII from 39 to give acyclic radical 40 should be fast On the other hand, loss of a proton from 47 to give enolate 48 should be fast. However, loss of MnII to give acyclic radical 49 will be slow because the oxidation potential of 48 is large, so that cyclization of 48 to 51 and 52 is the rate-determining step. Since these two cyclizations are mechanistically distinct, there is no contradiction in the preferential formation of cyclopentanemethyl radical 42 from 38 and cyclohexyl radical 52 from 47.
Demir and Watt have developed a widely used and versatile α'-acetoxylation of α,β-unsaturated ketones with anhydrous Mn(OAc)3 in benzene (see Scheme 6).2h,13 Initially this reaction was proposed to proceed by conversion of the enone 55 to the manganese enolate 56 followed by ligand transfer to give the observed product 58. However, α-keto radical 57 can be formed by oxidation of 2-cyclohexenone (55) and is probably also an intermediate in the formation of 58. For instance, oxidative cyclization of 59 afforded 61 in 61% yield via the intermediacy of radical 6014 and oxidation of cyclohexenone (55) in the presence of methylenecyclohexane afforded dihydrobenzofuran 64 (42%) and enone 63 (18%) resulting from formation of α-keto radical 57, which adds to the alkene to give 62 which then reacts further to give 63 and 64.15 Saturated ketones are also converted to α-keto radicals.16
Narasaka introduced manganese picolinate [Mn(pic)3].17 Mn(pic)3 has an octahedral manganese, with three picolinates bound to a single MnIII, while Mn(OAc)3 is an oxo-centered trimer. Not surprisingly, there are important differences in the reactivity of Mn(OAc)3 and Mn(pic)3, which have been fully addressed.10c,17 For instance, Mn(pic)3, or Mn(pic)2, which is produced as a byproduct of oxidative radical formation, can suppress the oxidation of some radicals by Cu(OAc)2.17
Ceric Ammonium nitrate (CAN) and the non-polar variant ceric tetrabutylammonium nitrate (CTAN) have been widely used to generate radicals oxidatively.18 These reactions were initially studied mechanistically by Baciocchi and Ruzziconi19 and more recently by Flowers.20 CAN oxidations are quite different from Mn(OAc)3 oxidations in that oxidation can occur without initial deprotonation of the dicarbonyl compound. The enol tautomer is oxidized to the radical cation, which may lose a proton to give a radical.
A more significant difference involves the conversion of the radical intermediates to products. Mn(OAc)3 is often used with Cu(OAc)2, which oxidizes radicals to alkenes selectively as discussed below. CAN is much cheaper than Mn(OAc)3 and can be used in a wide variety of solvents, whereas AcOH is usually the best solvent for Mn(OAc)3 reactions. However, Cu(OAc)2 cannot be used with CAN so products such as 26 and 61 cannot be obtained from CAN oxidations and α-acetoxy enones such 58 cannot be obtained with CAN. Furthermore, nitrate esters are often formed with CAN.21 Both CAN and Mn(OAc)3 can be used to generate radicals, but the nature of the oxidation steps that lead to products are quite different as is seen in the detailed study of the oxidation of 4-pentenyl malonate esters with Mn(OAc)310 and CAN21 so that Mn(OAc)3 is the preferred reagent for many applications.
MnIII will oxidize γ-carboxy radicals such as 4 that are formed by oxidation addition of acetate to alkenes to γ-lactone 8 regardless of whether the radical is secondary or tertiary. The addition of acetic acid and substituted acetic acids to alkenes to give γ-lactones is general for alkenes. However, isolated primary or secondary radicals abstract a hydrogen atom from the solvent more rapidly than they are oxidized by MnIII, so the oxidation of 4 must involve participation by the carboxylate. This might involve intramolecular transfer by a bound MnIII to give 5, addition of the radical to the carboxylate to give 6, which would be readily oxidized to 8, or by formation of 7 followed by reductive elimination of MnII to yield 8. MnIII will oxidize tertiary radicals to give tertiary cations that undergo normal E1 and SN1 reactions.
Addition of 1,3-dicarbonyl compounds to alkenes affords isolated radicals such as 17 and 25 that do not contain a proximal manganese carboxylate. If these radicals are tertiary, they will be oxidized by MnIII to cations that can lose a proton to give an alkene or react with solvent to give a tertiary acetate. MnIII will also oxidize allylic radicals to allylic acetates and cyclohexadienyl radicals, which are formed by the addition of radicals to aromatic rings, to cyclohexadienyl cations, which lose a proton to give substituted aromatic rings as in the formation of 33 and 37. Hydrogen atom abstraction from solvent is the major pathway if these radicals are primary or secondary, e. g. 17 and 25, unless Cu(OAc)3 or another co-oxidant is used.
Kochi and co-workers demonstrated that CuII reacts rapidly (~106 s1 M−1) with radicals such as 65 to give alkylcopperIII intermediates such as 66 (see Scheme 7).22 These can react further with loss of CuI to form either an alkene 68 by oxidative elimination, to transfer a ligand to give 70, or to form carbocation 67 Formation of alkene 68 by oxidative elimination is the major pathway from the reaction of Cu(OAc)2 with primary and secondary radicals. Tertiary, allylic, and other easily oxidized radicals give cations with copperII carboxylates. Other CuII salts give cations and ligand transfer products with all types of radicals. Heiba and Dessau found that Cu(OAc)2 is compatible with Mn(OAc)3 and that CuII oxidizes secondary radicals to alkenes 350 times faster than MnIII does.23 The CuI that is produced is rapidly reoxidized to CuII by MnIII so that only a catalytic amount of Cu(OAc)2 is needed and 2 equivalents of Mn(OAc)3 are still required. During the course of our studies we observed that, contrary to earlier indications,24 Cu(OAc)2 oxidizes secondary radicals to give primarily (E)-alkenes and the less substituted double bond (Hofmann elimination product).25 This selectivity is synthetically valuable since CuII oxidation of primary and secondary radicals formed in oxidative cyclizations often gives primarily or exclusively a single regio- and stereoisomer as detailed below.
The elimination probably occurs by formation of an alkyl CuIII intermediate with two bound acetates. The acetate abstracts a proton with formation of an alkene, acetic acid and Cu(OAc) in a syn elimination (see 69). Preference for the formation of the less substituted double bond (Hofmann product) and (E)-alkene is usually observed in syn eliminations. This mechanism also explains the observation that elimination occurs selectively only with copper carboxylates. Ligand transfer or oxidation to the cation occurs with copper halides, triflate, or sulfate.
Secondary radicals are almost always oxidized to alkenes by Cu(OAc). The organocopperIII intermediate formed from primary radicals can interact with adjacent functionality to give lactones as in the conversion of radical 42 to lactone 45 and to give cyclopropanes as in the conversion of 71 to cyclopropane 72 (see Scheme 8).
Oxidative cyclization of δ-hydroxy β-keto ester 73 affords epoxide 75 as the major product in 50–60% yield.26 β-Hydroxy radical 74 is oxidized to the epoxide by either MnIII or CuII in a process that is analogous to the oxidation of radical 4 to lactone 8. Epoxides are also formed from β-hydroxy radicals generated by Pb(OAc)4 oxidative decarboxylation of β-hydroxy acids by either PbIV or CuII, which suggest that oxidation of β-hydroxy radicals to epoxides is general.25
Addition of other anions to the reaction mixture affects the nature of the electron transfer process. For instance, Vinogradov and Nikishin reported that oxidation of ethyl acetoacetate (76) with 4 equiv of Mn(OAc)3 and excess LiCl in the presence of 1-hexene results in the formation of dichloride 78 (see Scheme 9).27,28 Monochloride 77 is formed initially. The use of chloride ion is not compatible with CuII; only α,α-dichlorination to give 79 is observed. The combination of Mn(OAc)3 and LiCl has seen very limited synthetic applications because the chlorides that are formed are of little utility. However, oxidation of 80 with Mn(OAc)3 and excess LiCl forms 81, which reacts further to give 82 as the major product.4,26
The radicals formed in the MnIII-based oxidative free-radical cyclizations of β-keto esters and malonate esters can be trapped oxidatively with Mn(OAc)3 and sodium azide to provide cyclic and bicyclic azides such as 84 and 85 in 30-80% yield (see Scheme 10). Reduction of the azides affords bicyclic and tricyclic lactams.29
These reactions can also be terminated by the addition of the radical to a nitrile or carbon monoxide, or by hydrogen atom abstraction from the solvent. This is particularly useful in converting vinyl radicals (obtained from addition to alkynes) to alkenes, since vinyl radicals are not oxidized to vinyl cations. The hydrogen atom can come from the solvent or from the α-hydrogen atom of another molecule of the β-dicarbonyl compound. Ethanol is the preferred solvent for these reactions, since it is a better hydrogen atom donor than acetic acid.10a,30 These reactions have been reviewed previously2 and are not discussed in detail here because they don't involve electron transfer.
Oxidative cyclization of unsaturated β-dicarbonyl compounds that have two α-hydrogen atoms will give products that still have one α-hydrogen atom and can be oxidized further. If the product is oxidized more slowly than the starting material, the cyclization product can be isolated in good yield. Reaction of 86 with Mn(OAc)3 affords enolate 87, which cyclizes to 88 in the slow rate-determining step (see Scheme 11). Oxidation by Cu(OAc)2 provides ketone 89, which tautomerizes to give a 1.3:1 equilibrium mixture of enol 90 and ketone 89 in 71% yield.31,32 Further oxidation of 89 or 90 to give 91 does not occur.
In other cases, the product is oxidized at a rate competitive with that of the starting material so that mixtures of products are obtained. For instance, oxidative cyclization of 92 affords 36% of 93 and 10% of dienone 95 formed by further oxidation of 93 to give radical 94, which is further oxidized to give 95 (see Scheme 12). Competitive oxidation of the product is usually not a problem in intermolecular addition reactions because a vast excess of the oxidizable substrate, such as acetone or acetic acid, is usually used as solvent. Use of excess substrate is not possible in oxidative cyclizations.
The rate-determining step in the cyclization of α-unsubstituted β-keto esters is the addition of the double bond to the manganese enolate. Oxidative cyclization of 86 is faster than oxidative cyclization of 92 since the double bond is better able to participate in the rate-determining step with a longer tether. Furthermore, oxidation of 93 to 95 (50%, 1 day) is much faster than the oxidation of 89 and 90 (0%, 6 days). We cannot explain this difference, but note that 93 is ketonic, whereas 90 is enolic. In other cases we have also observed that enolic 1,3-dicarbonylcompounds are oxidized slowly by Mn(OAc)3.
In the third category, the product is oxidized much more readily than the starting material so that none of the initial product is isolated. These reactions may still be synthetically useful if the products of further oxidation are monomeric. For instance, oxidative cyclization of 96 provides 78% of methyl salicylate (99) (see Scheme 13).26 Oxidative cyclization gives radical 97; oxidation of 97 by CuII gives 98, probably as a mixture of double-bond positional isomers. The unsaturated cyclic β-keto ester 98 is more acidic than 96 and is rapidly oxidized further by 2 equiv of MnIII to give a cyclohexadienone that tautomerizes to phenol 99. The overall reaction consumes 4 equiv of Mn(OAc)3. Oxidation of simple ketones will often give products that can be oxidized further. The oxidative cyclization of alkenyl substituted cyclic ketones works well because enolization (and further oxidation) of the bicyclic ketone product is prevented by Bredt's rule.16
The oxidative cyclization of crotyl malonate esters also falls into the third category. Oxidative cyclization of 100 affords 101, which is rapidly oxidized to radical 102. Radical 102 gives intractable material if R = Me, but affords 66% of 103 if R = crotyl.33 The lactone group makes the α-hydrogen of 101 much more acidic34 than those of 100 so that product lactone 101 is oxidized more rapidly than diester 100.
Further oxidation cannot occur if there are no acidic α-hydrogens in the product. α-Alkyl groups prevent further oxidation, but cannot then be removed. α-Chloro substituents serve as protecting groups preventing further oxidation of the product.4,35-37 Oxidative cyclization of 104 affords 82% of a 3.1:1 mixture of 105 and 106, which was elaborated to avenaciolide (108). Alternatively, reduction of the mixture with Zn afforded 95% of lactone 107, which is not accessible by oxidative cyclization of the unchlorinated malonate because 107 is oxidized further analogously to 101.35
Mn(OAc)3-based oxidative free-radical additions and cyclizations proceed by at least two different mechanisms. For compounds that are not acidic and form easily oxidized enolates, the slow rate determining step involves formation of the MnIII enolate followed by rapid MnII loss to generate a manganese-free free-radical. For more acidic compounds that form harder to oxidize enolates, the slow rate determining step involves addition of the alkene to the manganese enolate with loss of MnII and formation of a carbon-carbon bond in the same steps. The product radical can be oxidized to alkenes by Cu(OAc)2 through CuIII intermediate 69 by a Hofmann type elimination to give selectively the less substituted (E)-alkene.
I thank the National Institute of General Medical Sciences and National Science Foundation for financial support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.