Mechanism of the Enantioselective Oxidation of Racemic Secondary Alcohols Catalyzed by Chiral Mn(III)–Salen Complexes

Abstract

The experiments described here clarify the mechanism and origin of the enantioselectivity of the oxidation of racemic secondary alcohols catalyzed by chiral Mn(III)–salen complexes using HOBr, Br₂/H₂O/KOAc or PhI(OAc)₂/H₂O/KBr as a stoichiometric oxidant. Key points of the proposed pathway include (1) the formation of a Mn(V)–salen dibromide, (2) its subsequent reaction with the alcohol to give an alkoxy-Mn(V) species, and (3) carbonyl-forming elimination to produce the ketone via a highly organized transition state with intramolecular transfer of hydrogen from carbon to an oxygen of the salen ligand.

Introduction

Chungu Xia and his group showed that chiral Mn(III)–salen complexes can catalyze the oxidation of racemic secondary alcohols of the type R₁R₂CHOH at partial conversion to a mixture of the corresponding ketone and chiral R₁R₂CHOH with excellent enantioselection under the optimized conditions.^1-4 It was reported that the addition of substoichiometric amounts of a bromide salt to a mixture of catalyst 2a, PhI(OAc)₂, and a biphasic CH₂Cl₂/H₂O medium is important for high enantioselectivity and that k_rel in favorable cases can be as high as 450 (Scheme 1).^1b Kita and co-workers⁵ had earlier noted that KBr accelerated the oxidation of secondary alcohols by PhIO. The Xia group proposed that the active oxidant for the oxidation of secondary alcohols may be a species having composition salen–Mn(V)–OIPh but provided no explanation for the effect of bromide on the enantioselectivity or for the absolute stereochemical course of the reaction.^1b Representative examples of the enantioselective oxidation of racemic secondary alcohols are displayed in Scheme 1.⁶ There are limitations of the Xia method with R₁R₂CHOH substrates in which R₁ and R₂ have similar steric bulk. Alcohols such as 1-phenylpropanol (11), for example, are oxidized with only poor selectivity.^1d

Scheme 1.

Enantioselective Oxidation of Racemic Secondary Alcohols as Reported by Xia and Coworkers^a

^aFor further details, see ref 1 and the Supporting Information.

Our group recently proposed a logical mechanistic explanation for the absolute stereochemical course of the Mn(III)–salen-catalyzed epoxidation of olefins (Jacobsen epoxidation)⁷ that is in cases highly enantioselective.⁸ The key features of this pathway are as follows: (1) the epoxidation involves electrophilic attack by the oxygen of a salen–Mn(V)–oxo complex on one of the olefinic carbons to form the more stable carbocationic intermediate; (2) the relative orientation of the olefinic substrate and the Mn(V)–oxo complex is determined by the electrostatic attraction between the cationic carbon and one of the oxygens of the salen ligand; and (3) the olefinic π-face selectivity arises from differences in steric repulsion for the two diastereomeric complexes. This model leads, for example, to the pre-transition-state assembly shown in Figure 1 for the epoxidation of indene by Mn–salen complex 2a.

Figure 1.

An explanation for the origin of enantioselectivity in the Jacobsen epoxidation.

We believe that some features of this model for the Jacobsen epoxidation of olefins carry over to the enantioselective oxidation of secondary alcohols developed by Xia.¹ This line of analysis based on our earlier work⁸ together with the experiments described herein have led to a new and rational explanation for the Mn–salen-catalyzed enantioselective oxidation of secondary alcohols. We have followed a heuristic that has been extremely useful in many of our previous research efforts on the design and understanding of enantioselective chemical reactions.⁹ In its simplest form, that heuristic is as follows: (1) for cases in which a chemical reaction is highly enantioselective and also wide in scope, the pre-transition-state assembly is likely to be highly organized in the sense that there is one preferred three-dimensional assembly of the atoms involved; (2) that condition very sharply limits the number of mechanistic pathways by which the reaction can occur; and (3) the favored pathway can often be derived from three-dimensional modeling and an analysis of the steric, electrostatic, and stereoelectronic effects that determine the reaction energetics and rates.

Results and Discussion

Role of the Additive Bromide Ion

We start by examining the role of the additive bromide ion on enantioselectivity in the oxidation of racemic secondary alcohols since Xia et al.¹ reported that potassium bromide is absolutely crucial for fast and highly enantioselective oxidation. Several points regarding the data are noteworthy (see Table 1). Reactions in which potassium bromide was replaced by potassium chloride afforded only low conversion of (±)-1-indanol (1) (Table 1, entry 2). Reactions with potassium iodide gave poorer conversion than reactions with potassium bromide (Table 1, entry 3).¹⁰ Reactions carried out in the absence of any halide ion or with added chloride ion were inefficient (Table 1, entry 4). Water is a necessary component for high enantioselectivity (Table 1, entry 5). There is a significant Mn-independent background reaction, since the oxidation of 1 with PhI(OAc)₂ and potassium bromide without any Mn complex 2a was nearly complete within 30 min (Table 1, entry 6). The enantioselective oxidation of 1 in the presence of Mn complex 2a and KBr is much faster and was complete in less than 10 min. This acceleration of the Mn–salen oxidation by bromide ion is clearly large enough to dominate over the competing, nonenantioselective background process (without any Mn–salen) and thus leads to enantioselectivity.

Table 1.

Survey of Reaction Conditions for the Biphasic Oxidation of 1

entry	solvent	additive	conv (%)a	ee (%)b	krelc
1	CH2Cl2/H2O (1:2)	8 mol % KBr	68	>99	>13
2	CH2Cl2/H2O (1:2)	8 mol % KCl	12	<2	1.0
3	CH2Cl2/H2O (1:2)	8 mol % KI	29	22	4.1
4	CH2Cl2/H2O (1:2)	–	13	<2	1.0
5	CH2Cl2	8 mol % Et4NBr	63	55	3.2
6d	CH2Cl2/H2O (1:2)	8 mol % KBr	49	–	–

^aDetermined by ¹H NMR (500 MHz) analysis of the unpurified reaction mixtures.

^bDetermined by HPLC analysis with a chiral column.

^ck_rel = ln[(1 − conv)(1 − ee)]/ln[(1 − conv)(1 + ee)].

^dThe reaction was carried out in the absence of catalyst 2a.

Because one possible role of bromide might be the conversion of the salen–Mn–chloro complex 2a into the corresponding salen–Mn–Br complex 12, we prepared complex 12. As expected and as indicated in Table 2, oxidation of 1 catalyzed by 12 under the optimized conditions gave results identical to those for 2a (compare Table 1, entry 1 with Table 2, entry 1). However, the use of additional potassium bromide with salen–Mn–Br complex 12 is essential for achieving optimum stereoselection in the oxidation of 1 (Table 2, entry 2), as it is for the corresponding salen–Mn–chloro complex.¹ The bromide attached to Mn in 12 provides only a modest improvement in oxidative enantioselection relative to catalyst 2a in the absence of additional potassium bromide (compare Table 1, entry 4 with Table 2, entry 2).¹¹

Table 2.

Biphasic Oxidation of 1 with salen–Mn(III)–Br Catalyst 12

entry	solvent	additive	oxidant	conv (%)a	ee (%)b	krelc
1	CH2Cl2/H2O (1:2)	8 mol % KBr	PhI(OAc)2	65	>99	>16
2	CH2Cl2/H2O (1:2)	–	PhI(OAc)2	64	72	4.8
3d	CH2Cl2/H2O (1:2)	2.0 equiv of KOAc	Br2e	65	98	14
4d	CH2Cl2	2.0 equiv of n-Bu4NOAc	Br2e	27	<5	1.3
5d	CH2Cl2/H2O (1:2)	1.3 equiv of KOAc	HOBrf	69	97	9.5
6d	CH2Cl2/H2O (1:2)	1.0 equiv of KOAc	t-BuOBrg	66	98	13
7d	CH2Cl2/H2O (1:2)	1.0 equiv of NaHCO3	t-BuOBrg	66	89	7.4

^aDetermined by ¹H NMR (500 MHz) analysis of the unpurified reaction mixtures.

^bDetermined by HPLC analysis with a chiral column.

^ck_rel = ln[(1 − conv)(1 − ee)]/ln[(1 − conv)(1 + ee)].

^dThe reaction was carried out with 10 mol % 12 and oxidant added over 30 min. See the Supporting Information for details.

^eUsing 0.75 equiv of Br₂.

^fUsing 0.90 equiv of HOBr.

^gUsing 0.80 equiv of t-BuOBr.

We determined experimentally that bromide ion is oxidized by PhI(OAc)₂ to Br₂ (and therefore to HOBr, which in the presence of H₂O is known to be in equilibrium with Br₂) according to reactions R1 and R2:¹²

$$\mathrm{PhI}{\left(\mathrm{OAc}\right)}_2+2\mathrm{KBr}\to \mathrm{PhI}+{\mathrm{Br}}_2+2\mathrm{KOAc}$$ (R1)

$$\mathrm{PhI}{\left(\mathrm{OAc}\right)}_2+\mathrm{KBr}+{\mathrm{H}}_2\mathrm{O}\to \mathrm{PhI}+\mathrm{HOBr}+\mathrm{HOAc}+\mathrm{KOAc}$$ (R2)

The fact that a positive Br species is formed under the Xia conditions suggests that PhI(OAc)₂ can be replaced by Br₂ or HOBr as the stoichiometric oxidant in the enantioselective oxidation. Indeed, highly enantioselective oxidations can be carried out with Br₂ or HOBr as the stoichiometric oxidant [no PhI(OAc)₂ added] provided that sufficient potassium acetate is used to neutralize the HBr generated during the oxidation and buffer the reaction mixture (Table 2, entries 3 and 5). Sodium bicarbonate can also be used to neutralize HBr in the enantioselective oxidation of 1 (Table 2, entry 7). It is important that the oxidant be added slowly enough to prevent the detrimental build up of HBr. Reactions carried out with Br₂ or HOBr as the stoichiometric oxidant in the absence of potassium acetate were nonselective because of the adverse effect of HBr, and lower selectivities were observed (k_rel ≈ 4–9) when the oxidant was added rapidly.¹² As evidenced by the experiment summarized in Table 2, entry 4, the presence of water in the biphasic reaction mixture is necessary for the highly enantioselective reaction to occur. Our best procedure involved the slow addition of Br₂ or HOBr (0.75–0.90 equiv; see Table 2) to a mixture of secondary alcohol (1.0 equiv), potassium acetate (2.0 equiv for Br₂ or 1.0 equiv for HOBr), and salen–Mn–Br complex 12 (10 mol %) in 1:2 CH₂Cl₂/H₂O with stirring. Under the optimized reaction conditions, t-BuOBr afforded results that were almost identical to those for Br₂ or HOBr (Table 2, entry 6).¹³

Our experimental results indicate that either Br₂ or HOBr alone can convert the chiral (S,S)-Mn(III)–salen complex 12 into an oxidizing species that catalyzes the transformation of (±)-1-indanol into a mixture of 1-indanone and (S)-1-indanol with essentially the same enantioselectivity as that achieved under the most effective Xia procedure, with comparable reaction conditions (1:2 CH₂Cl₂/H₂O, 23 °C). The role of PhI(OAc)₂ in the presence of bromide ion in the Xia procedure is to generate HOBr and Br₂. There is no obvious reason why it needs to be invoked as an integral component of the transition-state assembly for enantioselective oxidation.¹

It should be emphasized that the Xia process for the enantioselective oxidation of secondary alcohols is biphasic with aCH₂Cl₂ layer and a water layer. The Mn(V)–salen catalyst and the PhI(III) oxidant are concentrated in the CH₂Cl₂ phase, as are 1-indanol and the product 1-indanone. Any bromide ion is concentrated in the aqueous phase, whereas Br₂ and HOBr are probably available in both the CH₂Cl₂ and water layers. Essentially all of the acetate ion buffer is in the water layer,¹² and for this reason among others, this species is unlikely to be involved in the oxidation itself. Furthermore, under the Xia conditions, acetate ion is not present and therefore cannot be involved in the transition state for secondary alcohol oxidation.

At the outset of this study, perhaps the most obvious pathway for enantioselective oxidation of secondary alcohols by the reagents Mn(III)–salen, KBr, and PhI(OAc)₂ under the conditions of Xia et al.¹ that would follow from our earlier mechanistic proposal for the Jacobsen epoxidation (Figure 1)⁸ seemed to be the one summarized in Figure 2. Such a pathway involving abstraction of the carbinol C–H by the oxygen of a salen-Mn(V)–oxo complex with H bonding of the hydroxyl proton and one of the oxygens of the salen ligand can definitely be ruled out, as this process would lead to the wrong enantiopreference. As illustrated in Figure 2, such a model incorrectly predicts that (S)-1-indanol would oxidize in preference to (R)-1-indanol with (S,S)-Mn(III)–salen complexes, which is contrary to fact.

Figure 2.

Pre-transition-state assembly with the Mn(V)–oxo complex predicts an incorrect absolute stereochemistry for the recovered alcohol.

We have obtained strong evidence that any mechanism involving direct C–H abstraction by a Mn(V)–oxo complex is unlikely on the basis of the following observations: (1) under conditions that are conducive to formation of Mn(V)-oxo complexes (and epoxidation of olefins), enantioselective oxidation of secondary alcohols is inefficient and nonselective, since treatment of Mn complex 2a (1.1 equiv) with m-CPBA¹⁴ or PhIO¹⁵ (1.0 equiv) in 1:2 CH₂Cl₂/H₂O at 23 °C followed by addition of 1 (1.0 equiv) led to poor conversion and enantioselectivity (<25% conv, <10% ee);¹²(2) treatment of salen–Mn–Br complex 12 (1.1 equiv) with t-BuOBr (1.0 equiv) and KOAc (1.0 equiv) in 1:2 CH₂Cl₂/H₂O (optimal conditions for enantioselective oxidation of 1) converted indene to dibromide 13 rather than to the epoxide (<2% observed), as would be expected if a Mn–oxo complex were generated (Scheme 2).¹⁶

Scheme 2.

Stoichiometric Reactions with salen–Mn(III)–Br Complex 12

Further evidence against direct catalytic involvement of a salen–Mn(V)–oxo complex in the enantioselective oxidation of alcohols was obtained from an experiment showing that bromide ion is rapidly oxidized to Br₂ by the salen–Mn(V)–oxo species. Experimentally, the salen–Mn(V)–oxo complex was generated by reaction of the corresponding Mn(III) complex with 1.0 equiv of m-CPBA and then treated with 2.0 equiv of aqueous KBr. Addition of indene and isolation of indene dibromide as outlined in eq 1 established that bromide ion is indeed oxidized to Br₂ by the Mn(V)-oxo salen complex.

(1)

Pathway for Enantioselective Mn-salen-Catalyzed Oxidation of Secondary Alcohols

All of our data are in accord with a pathway in which HOBr or Br₂/H₂O converts the Mn(III)–salen complex (I) into a brominated Mn(V)–salen species, such as [salen–Mn(V)–Br₂]⁺(II), which then reacts reversibly with the secondary alcohol to form the alkoxy complex [salen–Mn(V)–Br–OR]⁺ (III) as the key reaction intermediate and precursor of the ketonic product, as shown in Scheme 3. We surmise further that the Mn(V)–alkoxide complex III is converted to the Mn(III) species V by intramolecular hydrogen transfer from the carbinol C–H to the neighboring phenolic oxygen along a five-membered cyclic pathway via IV. This cyclic pathway is reasonable because the electron lone pair density on one of the metal-bound phenoxy oxygens is properly oriented to facilitate the carbonyl-forming elimination process in a fashion analogous to the electrostatic effect of that oxygen in organizing the transition state and stabilizing the positive charge during the Jacobsen epoxidation of olefins, as pictured in Figure 1.⁸ There is also an obvious analogy with the Westheimer chromate ester pathway for oxidation of secondary alcohols by Cr(VI) reagents.¹⁷ The mechanism outlined in Scheme 3 is attractive because it provides a uniquely simple explanation for the absolute stereocourse of the Mn–salen-promoted enantioselective oxidation. Intermolecular processes for proton abstraction from the Mn(V)-alkoxide complex III (e.g., deprotonation by any H₂O present in the CH₂Cl₂ layer) seems a highly unlikely alternative, as that would lead to a large number of geometrical paths for oxidation and provide no rational basis for the enantioselectivity. We do not rule out the possibility that a water molecule may be involved in the formation of ketone from IV.

Scheme 3.

Plausible Catalytic Cycle

We have obtained evidence that the rate-limiting step of the oxidation involves C–H cleavage (III → V), as might be anticipated, and not alcohol–Mn(V) complexation (II → III). Measurement of the relative rates of oxidation of 1-deuteriolabeled 1-indanol and 1-indanol indicated a small but definite primary kinetic isotope effect, with k_H/k_D = 2.7 ± 0.6.¹² Although this is below the values of k_H/k_D generally observed for the Cr(VI)-mediated oxidation of secondary alcohols,¹⁷ we believe that it is entirely consistent with expectations for a more exothermic process for the Mn(V) oxidation and an earlier, and bent, transition state.^18,19

The above pathway involves the same canted (i.e., nonplanar) geometry for the Mn(V)–salen complex III as proposed recently for the Jacobsen epoxidation.^7,8 That canted conformation places one of the ortho tert-butyl groups above and one below the N₂O₂ ligand plane, with the further assumption that the hydrogen is transferred to that phenoxy oxygen with the proper orientation of the one electron pair acceptor.²⁰ Two possible stereoelectronically favorable pre-transition-state structures emerge, one for the R enantiomer and one for the S enantiomer (VI and VII, respectively; Figure 3). This model leads to the expectation that the oxidation of the S enantiomer (via VII) should be slower than the oxidation of the R enantiomer (via VI) because of the serious repulsive steric clash of the indanol aryl subunit in VII with the tert-butyl group next to the chelate ring.

Figure 3.

Plausible pre-transition-state assemblies for the (a) favored and (b) disfavored modes of oxidation.

In accord with this proposal was the finding that catalysts 2b and 2c, in which the proximate tert-butyl group (at C3) is replaced by CH₃ and H, respectively (see Table 3, entries 2 and 3), led to much-diminished enantioselectivity (k_rel = 1.9–2.6). Further support for the mechanistic model came from the study of catalyst 2d (Table 3) having nitro in place of t-Bu as a substituent at C5 of the benzenoid ring. As indicated in Table 3, entry 4, the effect of the change was greatly diminished enantioselectivity (k_rel = 1.6) for catalyst 2d relative to the standard catalyst 2a (Table 3, entry 1).^8,21 This result is readily understandable on the basis of the internal proton transfer from carbon to oxygen that is shown in structure VI. Since the nitro substituent at C5 of the benzenoid ring of catalyst 2d is para to the phenolic oxygen serving as the proton acceptor in VI, it markedly decreases the electron density at that oxygen and hence reduces the rate of the enantioselective pathway. Dramatic confirmation of this rate retardation was obtained in a competition experiment involving a 1:1 mixture of the S,S enantiomer of 2a and the R,R enantiomer of catalyst 2d (eq 2). As indicated in eq 2, the presence of an equimolar amount of the nitrosubstituted catalyst 2d had virtually no effect on the enantioselective oxidation of 1 by catalyst 2a. Thus, it can be concluded that the effect of the nitro substituent in 2d is to decrease the rate of oxidation of 1 as well as to negate the enantioselective pathway.²²

(2)

Table 3.

Effect of Catalyst Structure on the Efficiency of Enantioselective Oxidation

entry	R1	R2	catalyst	conv (%)a	ee (%)b	krelc
1	t-Bu	t-Bu	2a	68	>99	>13
2	t-Bu	CH3	2b	65	34	1.9
3	t-Bu	H	2c	67	50	2.6
4	NO2	t-Bu	2d	68	26	1.6

^aDetermined by ¹H NMR (500 MHz) analysis of the unpurified reaction mixtures.

^bDetermined by HPLC analysis with a chiral column.

^ck_rel = ln[(1 − conv)(1 − ee)]/ln[(1 − conv)(1 + ee)].

Although our mechanistic model is not based on the X-ray crystal structure of the Mn(III)–alkoxide complex 14 (measured in this work²³) or that of the Mn(IV)-alkoxide complex 15 (previously reported²⁴), which appear in Figures 4 and 5, respectively, it is interesting to note that in each structure the alkoxy ligand is oriented with the carbinol C–H aligned over a phenolic oxygen, similar to the arrangement that we have proposed for the pre-transition-state assembly VI in Figure 3. This fact may be entirely coincidental but is nonetheless worth mentioning.

Figure 4.

X-ray crystal structure of Mn(III) complex 14.²³Selected distances (Å): Mn₁–O₁, 2.25; H₁–O₂, 2.59.

Figure 5.

X-ray crystal structure of Mn(IV) complex 15 (structure originally reported by Fujii and co-workers²⁴). Selected distances (Å): Mn₁–O₁, 1.88; H₁–O₂, 2.22.

Although a priori it is not unreasonable that the enantioselective Mn–salen-mediated oxidation of secondary alcohols might involve a Mn(IV)–salen complex²⁵ and a Mn(IV) ⇄ Mn(II) redox cycle (as in the MnO₂ oxidation of secondary alcohols to ketones), experimental evidence argues against this possibility. In a typical experiment, it was found that the stoichiometric reaction of the Mn(II)–salen complex (0.50 equiv) and HOBr (0.50 equiv) in 1:2 CH₂Cl₂/H₂O at 23 °C with 1 (1.0 equiv) was ineffective, as there was very little reaction and the recovered 1-indanol was racemic.¹²

Conclusions

In conclusion, we have presented evidence to explain the mechanistic and stereochemical pathway for the enantioselective Mn–salen-catalyzed oxidation of racemic secondary alcohols reported by Xia et al.¹ The salient features of the proposed mechanism are as follows: (1) A positive bromine species is generated under the reaction conditions by oxidation of bromide ion with PhI(OAc)₂. (2) It is the positive bromine species that oxidizes the Mn(III)–salen complex to a Mn(V)–salen complex. (3) The most likely composition of the Mn(V)-salen complex is that of a dibromo-Mn(V) species (II, Scheme 3). (4) An additional effect of added bromide ion arises from its ability to reduce the salen–Mn(V)–oxo complex with formation of Br₂. (5) The dibromo–Mn(V) species II undergoes reversible ligand exchange with (±)-1-indanol as the substrate to generate a Mn(V)–alkoxy complex (III, Scheme 3). (6) Oxidation of the secondary alcohol occurs by intramolecular hydrogen transfer from carbon to a phenoxy oxygen of the salen ligand. (7) The stereochemical course of the enantioselective oxidation is easily rationalized by comparing the degree of steric repulsion in the structures VI and VII that are shown in Figure 3.