Enantioselective Decarboxylative Alkylation Reactions: Catalyst Development, Substrate Scope, and Mechanistic Studies

Abstract

α-Quaternary ketones are accessed through novel enantioselective alkylations of allyl and propargyl electrophiles by unstabilized prochiral enolate nucleophiles in the presence of palladium complexes with various phosphinooxazoline (PHOX) ligands. Excellent yields and high enantiomeric excesses are obtained from three classes of enolate precursors: enol carbonates, enol silanes, and racemic β-ketoesters. Each of these substrate classes functions with nearly identical efficiency in terms of yield and enantioselectivity. Catalyst discovery and development, the optimization of reaction conditions, the exploration of reaction scope, and applications in target-directed synthesis are reported. Experimental observations suggest that these alkylation reactions occur through an unusual inner-sphere mechanism involving binding of the prochiral enolate nucleophile directly to the palladium center.

Introduction

The catalytic asymmetric synthesis of all-carbon quaternary stereocenters stands as a significant challenge in synthetic chemistry.^[1] Despite the demanding sterics of quaternary stereocenter formation, a number of useful catalytic transformations, including Diels–Alder,^[2] Heck,^[3] cyclopropanation,^[4] alkylation,^[5] acylation,^[6] desymmetrization,^[7] and pericyclic^[8] reactions, among others, have been successful in generating these moieties. Although palladium-catalyzed enantioselective allylic alkylation chemistry has been an important tool for organic synthesis, only relatively recently has palladium(II) π-allyl chemistry been used for the formation of quaternary stereocenters.^[9] The vast majority of palladium-catalyzed C–C bond forming allylic alkylations studied by Trost, Helmchen, Pfaltz, and others form tertiary stereocenters by the attack of stabilized (e.g., malonate) anions on prochiral 1,3-disubstituted allyl fragments.^[9] Helmchen has shown that such reactions with palladium phosphinooxazoline (PHOX) complexes typically occur via an outer-sphere malonate attack at the allyl termini.^[10] Although allylic alkylation mediated by other metals, especially copper,^[11] has successfully generated quaternary centers on prochiral allyl groups, to our knowledge this mode of reactivity has not been realized with palladium.^[12]

An alternative, less common strategy in allylic alkylation is the use of prochiral nucleophiles. A quaternary stereocenter may be formed on a prochiral nucleophile when it possesses three distinct substituents. Such reactions require a remote chiral ligand to discriminate between the prochiral faces of an incoming nucleophile, and thus may seem a more challenging strategy. However, Hayashi,^[13] Ito,^[14] and Trost^[15] have reported the asymmetric allylation of prochiral enolates derived from 1,3-dicarbonyl compounds. Trost and coworkers demonstrated that diamine-derived ligands 6 and 7, which were designed to project bulk in front of the allyl fragment due to their large bite angle, are able to favor one face of the in situ generated ketone enolate (Scheme 1).^{[16],[17],[18]} These reactions represented a significant advance in asymmetric allylation technology by forming quaternary stereocenters with excellent yield and good ee. However, the substrate scope of these reactions was restricted to ketone substrates containing either a single acidic site (e.g., ketone 1 and tetralone 2) or two α-sites that have a large difference in acidity (e.g., β-ketoester 4 and α-phenyl ketone 5).^[1d] These limitations prevented direct access to fundamental α-quaternary ketones, such as 2-allyl-2-methylcyclohexanone (9), a cyclohexanone derivative not known as a single enantiomer prior to our work.^[19]

Scheme 1.

Trost’s Allylic Alkylation with Prochiral Nucleophiles.

We were confronted by these limitations during studies toward the synthesis of several natural products.^[20] A survey of the literature revealed that the non-enantioselective alkylation reactions developed by Tsuji in the early 1980s had the proper reactivity and selectivity to form quaternary stereocenters in the presence of less substituted ketone α-sites of similar acidity. For example, allyl enol carbonate and allyl β-ketoester substrates contain both a latent enolate and allyl fragment (Scheme 2, A and B).^[21] Alternatively, enol acetates and silyl enol ethers may serve as enolate precursors in intermolecular reactions with allyl carbonates (Scheme 2, C and D).^[22] These alkylations have the advantage that the reaction occurs under nearly neutral conditions and often at mild temperatures. Despite their advantages, these alkylation reactions had been utilized rarely during the intervening twenty years between 1983 and the time we began research in 2003.^[23],[24]

Scheme 2.

Tsuji’s Allylation Methods.

Our analysis of Tsuji’s alkylation reaction showed it to be an ideal candidate for asymmetric catalysis. These high yielding alkylation methods are a clear case of ligand-accelerated catalysis, occurring only in the presence of phosphine ligands.^[25] Of additional interest was the regiochemical fidelity observed in the alkylation reactions (Scheme 3). Tsuji demonstrated that both the tetrasubstituted allyl enol carbonate isomer 8 and the trisubstituted allyl enol carbonate isomer 11 undergo reaction to give allylated ketones 9 and 12 in ratios essentially unchanged from that of the substrates.^[21a] These reactions were believed to proceed via oxidative addition of palladium to the allyl fragment followed by loss of CO₂ to give [palladium(II)(allyl)] complex 13 and enolate 14. However, the details involving the recombination of the ion pair to give cyclohexanone 9 and a palladium(0) species were unclear at the time of Tsuji’s original reports.^[26] Coupled with the ability to purify the stable enolate precursors, we believed that the regiochemical fidelity afforded in the palladium-catalyzed reaction would provide direct access to enantioenriched α-quaternary ketones if a suitable chiral ligand could be found.^[27]

Scheme 3.

Regiochemical Fidelity in Tsuji’s Palladium-Catalyzed Allylation.

Results and Discussion

Initial Screening of Chiral Ligands

Our initial goal was to show that a chiral ligand could transmit useful levels of asymmetric induction in the reaction while maintaining the important property of enolate regiochemical fidelity found in the non-enantioselective system. We chose allyl enol carbonate 8 as a test substrate to evaluate the effect of various ligands (Table 1). Although allyl enol carbonates are less common than the other enolate precursors explored by Tsuji, they allowed us to add a single, achiral reagent to our catalyst system without extraneous initiators or counterions that might affect enantioselectivity, whereas silyl enol ether, enol acetate, and β-ketoester substrates did not. In accord with Tsuji’s reports, we performed our initial trials in dioxane. Due to the prevalence of bis(phosphine) ligands in asymmetric catalysis, we began with several privileged bis(phosphine) ligands, but found that only Trost’s ligand (6) gave significant ee. However, commercially available (R)-QUINAP (15), a chelating N/P-type ligand,^[28] gave more uniform ee with other ethereal solvents (cf. entries 5 and 6 in dioxane and THF). Encouraged, we quickly found that the phosphinooxazoline (PHOX) class of N/P-type ligands also provided excellent reactivity and promising levels of enantioselectivity. The ready availability of numerous amino acid-derived PHOX ligands^[29] allowed us to rapidly identify that bulkier aliphatic groups on the oxazoline provided higher levels of enantioinduction, with (S)-tBu-PHOX (19) providing α-quaternary ketone (S)-9 in 86% ee when dioxane was used as solvent. Additionally, we observed slightly higher ee in reactions with (R)-iPr-PHOX (17) and (S)-tBu-PHOX (19) when THF was used as solvent (83% and 88% ee, respectively).^[30] Having attained satisfying levels of enantioselectivity and selectivity in forming ketone 9, we began a more thorough investigation of reaction conditions.

Table 1.

Initial Ligand Screening.^[27a]

		dioxane			THF
entry	ligand	time (h)	% yielda	% eeb	time (h)	% yielda	% eeb
1	(R)-BINAP	5	92	5c	5	76	2c
2	(R,R)-Me-DUPHOS	5	61	0	5	66	0
3	(R,R)-DIOP	2	91	2c	2	59	2c
4	(R)-MOP	3	93	18	3	47	13
5	(R,R)-Trost ligand (6)	2	97	46c	5	92	64c
6	(R)-QUINAP (15)	2	98	61	2	97	61
7	(R)-Ph-PHOX (16)	2	95	62c	2	95	65c
8	(S)-Bn-PHOX (17)	3	96	65	5	94	63
9	(R)-i-Pr-PHOX (18)	3	96	82c	2	95	83c
10	(S)-t-Bu-PHOX (19)	2	95	86	2	96	88

^aGC yield of 9 relative to internal standard (tridecane).

^bEnantiomeric excess by chiral GC.

^c(R)-9 produced as the major product.

Optimization of Reaction Parameters

The straightforward experimental procedure for asymmetric alkylation provided several opportunities to optimize the reaction. One important experimental parameter was found to be the complexation time for (S)-tBu-PHOX (19) and tris(dibenzylideneacetone)dipalladium(0) (Pd₂(dba)₃) prior to addition of the substrate.^[31] Mixing the metal and ligand for 30 min before substrate addition resulted in a balance between short complexation times (e.g., 5 min) in which lower yields but complete consumption of the substrate were observed, and longer complexation times (e.g., 1–3 h) in which poor conversion was observed. Storing the catalyst–metal complex for an extended time seemed to be complicated by the introduction of adventitious amounts of O₂ that readily oxidized the ligated PHOX molecule at phosphorus, thereby preventing significant consumption of the starting material.^[32]

We also investigated the effect of concentration on the reaction (Table 2). At higher concentrations, significantly lower yields were observed, as well as slightly decreased ee. No further increase in enantioselectivity was observed below 0.03 M.

Table 2.

Effect of Concentration on Asymmetric Allylation.

entrya	concentration (M)	time (h)	% yieldb	% eec
1	0.500	3	81	82
2	0.250	2	90	84
3	0.125	2	94	84
4d	0.063	2	99	85
5	0.031	2	95	86

^aData reported are the average of three trials.

^bGC yield relative to internal standard (tridecane).

^cEnantiomeric excess measured by chiral GC.

^dData reported are the average of two trials.

Encouraged by our initial discovery that reactions in THF gave better levels of enantioinduction than those in dioxane, we undertook a more thorough study of solvent effects on the reaction (Table 3). Many of the ethereal solvents investigated gave good results. Diethyl ether, tert-butyl methyl ether (TBME), and diisopropyl ether all provided good yields and slightly higher ee than reactions in THF with allyl enol carbonate 8 (entries 3–5). However, the methyl tetralone-derived allyl enol carbonate 20 yielded products with substantially lower ee in diethyl ether and TBME. The poor solubility of the catalyst in these less coordinating ethereal solvents occasionally led to incomplete conversion, a disadvantage that outweighed the slight increase in ee. Interestingly, several non-ethereal solvents also performed well in the transformation. Reactions in benzene and toluene gave similar yield and enantioinduction as those in THF (entries 8 and 9). The reaction tolerated carbonyl-containing solvents: ethyl acetate (entry 10) provided good yields and enantioselectivity in the asymmetric allylation, although acetone (entry 11) afforded inferior yield and ee. Interestingly, triethylamine produced a level of enantioselectivity equal to that observed with the best ethereal solvents, albeit with lower yield (entry 12). Halogenated solvents fared poorly in the reaction, suffering from low conversion and yield (entries 13–15). Overall, we were intrigued by the variety of solvents with vastly different properties that performed equally well.

Table 3.

Effect of Solvent on Asymmetric Allylation.

		9a,b			3a,b
entry	solvent	time (h)	% yieldc	% eed	time (h)	% eee
1	dioxane	2	95	86	1	87
2	tetrahydrofuran	2	96	88	1	88
3	diethyl ether	2	98	89	1	80
4	t-butyl methyl ether	2	98	89	–	–
5	diisopropyl ether	2	95	89	–	–
6	anisole	3	82	81	–	–
7	1,2-dimethoxyethane	2	72	56	1	78
8	benzene	2	99	88	1	89
9	toluene	2	99	88	1	87
10	ethyl acetate	2	97	86	–	–
11	acetone	3	26	60	–	–
12	triethylamine	2	72	89	–	–
13	fluorobenzene	3	58	51	–	–
14	dichloromethane	3	42	13	–	–
15	chloroform	6	0	NA	–	–

^aData reported are the average of three trials.

^bAll reactions proceeded to complete conversion except entries 13–15.

^cGC yield relative to internal standard (tridecane).

^dEnantiomeric excess measured by chiral GC.

^eEnantiomeric excess measured by chiral HPLC.

Fine Tuning of Phosphinooxazoline Ligands

A substantial effort was undertaken to improve the enantioselectivity of the reaction by modifying the PHOX ligand structure (Table 4).^[33] Hoping to continue the trend of increasing enantioselectivity initially noted in moving from iPr- to tBu-PHOX (entries 2 and 4), we undertook the synthesis of numerous PHOX ligands with varied sterics and evaluated these ligands in the reaction of allyl enol carbonates 8 and 20. In general, ligands bearing saturated substituents on the oxazoline fragment performed better than those with aryl groups in terms of ee. Moving the steric bulk away from the oxazoline framework by inserting a methylene group (i.e., ligand 23, entry 6) substantially lowered enantioselectivity. Of particular note are the L-serine-derived ligands 27 and 28 (entries 10 and 11),^[34] which allowed access to the enantiomeric R product series with nearly the same level of enantioselectivity as that observed with (S)-tBu-PHOX (19), but without the need for more costly (R)-tert-leucine.^[35] The known cis-1-amino-indan-2-ol-derived ligand 29 and borneol-derived ligand 30, both with uniquely shaped substituents, gave slightly lower enantioselectivity relative to tBu-PHOX (19).^[36] It is noteworthy that regardless of the shape or type of the substituent on the PHOX ligand, the ketone products contained quaternary stereocenters with a consistent sense of configuration (e.g., (S)-PHOX ligands provide (S)-9).^[37] With tBu-PHOX (19) established as the optimal steric framework, we next considered the electronics of the ligand.^[38]

Table 4.

Effect of Phosphinooxazoline Sterics on Enantioselective Alkylation.

entry	ligand	producta	% eeb	entry	ligand	producta	% eeb
1	16	9	65 (R)	8	25	9	83 (R)
2	17	9	83 (R)	9	26	9	69 (R)
3	18	9	63 (S)	10	27	3	85 (R)
4	19	9	88 (S)	11	28	3	85 (R)
5c	22	3	78 (S)	12	29	3	79 (S)
6	23	3	59 (S)	13	30	3	81 (S)
7	24	9	41 (S)

^aAll reactions proceeded to complete conversion.

^bEnantiomeric excess measured by chiral HPLC or GC.

^cReaction run at 0.2 M in dioxane.

A number of tBu-PHOX derivatives were synthesized to probe the importance of phosphine electronics (Table 5). We investigated ligands ranging from electron rich to electron poor phosphines with allyl enol carbonates 20 and 8 (entries 1–7 and 8–14, respectively). The electronic perturbation had no significant effect on reaction yield, but ligands bearing electron withdrawing groups (especially CF₃ groups) gave a significant increase in the rate of reaction with allyl enol carbonate substrates.^[33a] Further, electron releasing para substituents on the phenyl rings tended to lower the ee of the product relative to that observed with (S)-tBu-PHOX (entries 1 and 8). When tetralone-derived allyl enol carbonate 20 was used as the substrate, a slight increase in enantioselectivity was observed with electron withdrawing substitution at the p-phenyl positions (entries 3–5). However, enantioselectivity decreased significantly with extremely electron poor PHOX ligands (entries 6 and 7). This trend was not apparent in the enantioselectivity with cyclohexyl-derived allyl enol carbonate 8 (entries 8–14). These results suggest a subtle interplay between ligand and substrate electronics.

Table 5.

Effect of Phosphine Electronics on Enantioselective Alkylation.

entry	product	ligand	solvent	% eeb
1	3	31	dioxane	81
2	3	19	dioxane	87
3	3	32	dioxane	88
4	3	33	dioxane	89
5	3	34	dioxane	90
6	3	38	dioxane	83
7	3	39	dioxane	81

entry	product	ligand	solvent	% yieldc	% eed
8	9	35	THF	99	86
9	9	19	THF	96	88
10	9	36	THF	99	89
11	9	32	THF	95	88
12e	9	33	THF	97	87
13e	9	34	THF	89	88
14	9	37	THF	99	87

^aAll reactions proceeded to complete conversion.

^bEnantiomeric excess measured by chiral HPLC.

^cGC yield relative to internal standard (tridecane).

^dEnantiomeric excess measured by chiral GC.

^eData reported are the average of two trials.

As a final perturbation of the PHOX ligand structure, we prepared a number of non-N/P mixed chelates based on the phenyl oxazoline skeleton of the PHOX ligands (Table 6).^[39] The phosphine oxide of tBu-PHOX is inactive as a catalyst.^[32] A sulfur analogue also failed to catalyze the reaction (Table 6, entry 1).^[40] Moving down the periodic table from phosphorus, the arsenic analogue of tBu-PHOX had excellent activity as a catalyst, but gave tetralone 3 in only moderate ee (entry 2). A nitrogen analogue showed little catalytic activity, and the small amount of product generated was nearly racemic (entry 3). As a final derivative that maintains the six-membered chelation, but changes the hybridization of the backbone atoms involved, a known phosphinite ligand, SimplePHOX, was found to give only moderate ee (entry 4).^[41]

Table 6.

Effect of Varied Heteroatom Chelates on Enantioselective Alkylation.

entry	ligand	% conversion		% eeb	entry	ligand	% conversion		% eeb
1			0a	ND	3			<5a	7
2			99a	52	4			87c	56

^aConversion estimated by TLC analysis.

^bEnantiomeric excess measured by chiral HPLC.

^cTrial performed with allyl enol carbonate 8. Conversion measured by GC relative to internal standard (tridecane). Enantiomeric excess measured by chiral GC.

Although some electron deficient derivatives of tBu-PHOX did provide slightly better product ee with certain substrates, ultimately this slight improvement did not typically offset the added difficulty in synthesizing substituted PHOX ligands.^[33] Our studies clearly showed that N/P chelates were particularly effective at inducing high levels of asymmetry in the allylation. As a result, we retained the use of (S)-tBu-PHOX (19) in our optimized conditions for exploring the scope of the reaction.

Asymmetric Allylation of Allyl Enol Carbonates

We have successfully employed a variety of allyl enol carbonates in our asymmetric Tsuji allylation (Table 7).^[42] The allyl group may be substituted at the internal position resulting in slightly higher levels of asymmetric induction (entry 4). The cyclic portion may be unsaturated (entry 5), substituted (entries 6–11), appended (entries 12 and 13), or enlarged (entries 14 and 15). Among these variations, several are particularly noteworthy: the presence of a conjugate acceptor enone (entry 5) does not lead to Michael addition products, highlighting the essentially neutral reaction conditions; quaternary stereocenters may be synthesized vicinal to pre-existing fully substituted carbons (entry 8); the temperature may be lowered in order to improve enantioselectivity when highly reactive substrates are employed (entries 12 and 13), although the reaction time increases significantly.^[43] Overall, we were pleased to find that a wide range of functionalized enantioenriched cycloalkanones were accessible by this method.

Table 7.

Substrate Scope for Asymmetric Allylation of Allyl Enol Carbonates.^[27a]

entry	producta	% yieldb		% eec	entry	producta			% yieldb	% eec
1			85	87	7		R1 = CH2CH3		96	92
2d			85	88 (96)e	8g		R1 = t-Bu		55h	82
3f			90	89	9		R1 = CH2Ph		96	85
					10		R1 = (CH2)3OBn		87	88
4g			89	91	11				94	92
5			91	89	12i			R = H	87	91
					13i			R = OCH3	94	91
6			87	86	14			n = 1	81	87
					15			n = 2	90	79

^aReactions were performed using 1.0 mmol of substrate in THF (0.033 M in substrate) at 25 °C with 2.5 mol% Pd₂(dba)₃ and 6.25 mol% (S)-tBu-PHOX (19), unless otherwise noted. Each reaction was complete in 1–10 h.

^bIsolated yields.

^cMeasured by chiral GC or HPLC.

^dPerformed on 5.1 mmol scale.

^eIn parentheses is the % ee after one recrystallization of the corresponding semicarbazone.

^fReaction performed at 12 °C (GC yield).

^gPerformed with 5 mol% Pd₂(dba)₃ and 12.5 mol% (S)-19.

^hIsolated yield after conversion to the corresponding diketone via Wacker oxidation.

ⁱPerformed at 10 °C.

Our use of allyl enol carbonates enabled direct access to α-quaternary ketones with multiple acidic sites. However, allyl enol carbonates are rarely encountered in the literature, and the synthesis of isomerically pure enol carbonates^[44] often required the synthesis of silyl enol ethers as intermediates. Since Tsuji had used silyl enol ethers in his non-enantioselective allylation, we contemplated adapting our conditions such that silyl enol ethers could be employed directly.

Asymmetric Allylation of Silyl Enol Ethers

Silyl enol ethers are commonly encountered enolate equivalents. Unlike allyl enol carbonates, silyl enol ethers render the alkylation reaction intermolecular, with the enolate precursor and allyl fragment introduced separately. We discovered in our initial studies that silyl enol ethers were not sufficiently nucleophilic to react with the [Pd(II)(allyl)] fragment under our reaction conditions at 25 °C. However, we found that the reaction could be initiated in the presence of tetra-n-butylammonium difluorotriphenylsilicate (TBAT), a commercially available dry fluoride source.

The use of TBAT as an initiator complicated the proposed catalytic cycle (Scheme 4). In the case of allyl enol carbonates, the oxidative addition of allyl carbonate 8 and rapid decarboxylation led directly to the enolate/[Pd(II)(allyl)] ion pair, which could collapse directly to the product (Scheme 3). For the intermolecular reaction, we primarily used diallyl carbonates as allyl precursors, although mixed carbonates were also effective. Oxidative addition of allyl carbonates occurs readily at 25 °C to give [Pd(II)(allyl)] species 13, CO₂, and an alkoxide. The addition of TBAT immediately generates the enolate, which could react with palladium complex 13 by the same enantioselective mechanism observed with allyl enol carbonates. A substoichiometric amount of TBAT was sufficient, as the alkoxide formed in the reaction is also capable of providing the enolate in situ. In principle, it should be possible for the small amount of allyl alkoxide generated from oxidative addition to initiate the reaction.^[45] In practice, we found that 35 mol% TBAT was usually sufficient to ensure complete conversion of the silyl enol ether (40). Having developed an effective means of silyl enol ether activation, we attempted asymmetric alkylation with a range of tetrasubstituted silyl enol ethers.

Scheme 4.

General Catalytic Cycle for Silyl Enol Ether Alkylation.

Gratifyingly, alkylation of the silyl enol ether substrates occurred with levels of enantioinduction similar to those observed for the allyl enol carbonate substrates and over a similar range of substrates (Table 8). Specifically, methyl- and ethyl-substituted quaternary ketones were produced with the same ee observed in the allyl enol carbonate reactions (entries 1 and 2). Tertiary ether stereocenters were accessible from an α-oxygenated silyl enol ether, albeit with moderate ee (entry 3). In addition to diallyl carbonate, dimethallyl carbonate served as a suitable allyl fragment precursor (entries 4 and 5). Interestingly, 2-allyl-2-methallylcyclohexanone, bearing only a remote methyl group to engender chirality, formed in 91% ee. As with the reactions of allyl enol carbonates, substitution about the ring and larger ring sizes were tolerated as well (entries 6–8).

Table 8.

Substrate Scope for Asymmetric Allylation of Silyl Enol Ethers.^[27a]

entry	producta		% yieldb	% eec	entry	producta		% yieldb	% eec
1		R1 = CH3	95	87	6			99	81
2		R1 = CH2CH3	96	92
3		R1 = OBn	83	59
4d		R1 = CH3	79	91	7		n = 1	94	86
5d		R1 = allyl	82	91	8		n = 2	96	79

^aReactions were performed using 1.0 mmol of substrate in THF (0.033 M in substrate) at 25 °C with 2.5 mol% Pd₂(dba)₃, 6.25 mol% (S)-tBu-PHOX (19), diallyl carbonate (1.05 equiv), and TBAT (35 mol%) unless otherwise noted. Each reaction was complete in 2–5 h.

^bIsolated yields.

^cMeasured by chiral GC or HPLC.

^dReaction performed with dimethallyl carbonate (1.05 equiv).

In addition to the flexibility afforded by the intermolecular reaction of silyl enol ethers and diallyl carbonates, the use of silyl enol ethers as a means to generate enolates independent of the allyl fragment allowed the catalytic cycle to commence at the stage of a [Pd(II)(allyl)] complex 41 (Scheme 5). Upon mixing (S)-tBu-PHOX, [Pd(allyl)Cl]₂, and NH₄PF₆ in ethanol,^[46] [Pd(II)(allyl)PHOX]⁺[PF₆]⁻ salt 41 readily precipitated and this solid could be recrystallized and characterized crystallographically as a partial EtOH adduct.^[47] The complex (41) serves as an active catalyst in the enol silane variant of our enantioselective alkylation reaction, giving good yield and nearly identical product ee compared to that observed with the in situ-generated catalyst. The Pd(II)PF₆ salt 41 has several practical advantages: it is an air stable, nonhygroscopic solid that may be stored indefinitely. Moreover, using the preformed [Pd(allyl)PHOX] catalyst obviated the introduction of dibenzylideneacetone (dba), which often complicated the purification of the α-quaternary ketone products.^[48]

Scheme 5.

Allylation with Pd(II)PF₆ Salt 41.

Enantioselective Alkylation of Dioxanone Substrates for the Synthesis of Tertiary Ethers and Alcohols

Given our success in preparing α-quaternary cycloalkanones, we wished to explore the direct synthesis of enantioenriched α-oxygenated ketones, which are a particularly challenging structure and appear in a number of biologically active compounds.^[49] Our preliminary entries into these systems included a Baeyer–Villiger oxidation of an α-quaternary ketone (Scheme 9a, vide infra) and a single example of an oxygenated enol silane that provided a tertiary ether in only moderate ee (Table 8, entry 3). The latter result indicated that an exocyclic heteroatom directly attached to the putative enolate intermediate caused significant disruption to the stereocontrol elements of the catalyst. Therefore, we chose to explore a class of substrates with the heteroatom contained within the ring where it would be less likely to interact with the metal center.^[27c] Dioxanone substrates have been employed in chiral auxiliary-controlled enolate alkylations, and we hypothesized that our catalytic enolate alkylation procedure might provide a more efficient entry into these oxygenated products.^[50],[51]

Scheme 9.

Useful Derivatives of Enantioenriched Cyclic Ketones.^[27a,c]

We prepared enol silane 42 (Table 9) as a representative substrate of the dioxanone class.^[52] We found that solvent had a relatively larger effect on the yield and ee of the product with this class of substrates (Table 9) and that a full equivalent of TBAT was required for efficient conversion of the TES enol ethers to product. In the optimal case, exposure of enol ether 42 to the complex derived from [Pd(dmdba)₂]^[48] and (S)-19, diallyl carbonate, and TBAT (1 equiv) in toluene at 25 °C led to dioxanone product 43 in 74% yield and 88% ee. We explored the scope of this transformation and found that substitutions at the 2-position and on the allyl group were possible while maintaining high yields and product ee (Table 10). These oxygenated products were amenable to a number of subsequent transformations that enabled the enantioselective synthesis of several challenging structural motifs (Scheme 9d, vide infra).

Table 9.

Effect of Solvent in Dioxanone Alkylation.

entry	solvent	% yielda	% eeb
1	THF	40	81
2c	Et2O	59	89
3	1,4-dioxane	65	67
4	benzene	67	84
5	toluene	74	88

^aIsolated yield from reaction using 0.1 mmol of substrate in solvent (0.033 M in substrate) at 25 °C with 5 mol% Pd(dmdba)₂, 5.5 mol% (S)-tBu-PHOX (19), diallyl carbonate (1.05 equiv), and TBAT (1 equiv) unless otherwise noted. Each reaction was complete in 5–7 h.

^bMeasured by chiral HPLC of a derivative.^[53]

^cReaction performed at 30 °C.

Table 10.

Enantioselective Alkylation Using Dioxanone Derived Enol Silanes.^[27c]

entry	producta		% yieldb	% eec	entry	producta	% yieldb	% eec
1		R1 = H	86	87
2d,e		R1 = CH3	59	89	7		88	85
3f		R1 = Cl	59	92	8e		93	88
4g		R1 = Ph	73	94
5		R = CH3	79	93	9		83	92
6		R = Ph	85	86

^aReactions were performed using 0.5 mmol of substrate in PhMe (0.033 M in substrate) at 25 °C with 5 mol% Pd(dmdba)₂, 5.5 mol% (S)-tBu-PHOX (19), diallyl carbonate (1.05 equiv), and TBAT (1 equiv) unless otherwise noted. Each reaction was complete in 5–10 h.

^bIsolated yields.

^cMeasured by chiral GC or HPLC, in some cases as a derivative.^[53]

^dReaction performed with trimethyl silyl enol ether with 35 mol% TBAT in Et₂O (0.0167 M in substrate).

^eReaction performed with dimethallyl carbonate (1.05 equiv). The major product is enantiomeric to the structure shown.

^fReaction performed with dichloroallyl carbonate (1.05 equiv) at 35 °C.

^gReaction performed with diphenylallyl carbonate (1.05 equiv).

Our work with silyl enol ethers demonstrated that our enantioselective process is robust enough to permit introduction of the enolate and allyl fragments separately and to tolerate the presence of other ions in solution. Silyl enol ethers are a more familiar substrate class and greatly increase the practicality of the reaction. However, we rely on thermodynamically driven silyl enol ether syntheses, which typically produce a 10:1 ratio of isomers that requires a somewhat tedious purification to obtain isomerically pure substrates.^[44] A direct method for the synthesis of isomerically pure substrates would further increase the practicality of the catalyst system.^[26] Nonetheless, the enantioselective alkylation reactions employing either allyl enol carbonate or enol silane substrates provide broad access to enantioenriched α-quaternary cycloalkanones.

Asymmetric Decarboxylative Allylation of β-Ketoesters

Encouraged by our previous results, we sought to exploit this technology for the preparation of a variety of interesting synthetic targets. In one such effort, we encountered difficulty in the preparation of the requisite enol carbonate or silyl enol ether (Scheme 6). A variety of conditions for enolate formation and trapping led to inseparable mixtures of enol isomers. As a consequence of the high degree of regiochemical fidelity in the allylation reaction,^[21,22] corresponding mixtures of allylated products were obtained. To address this problem, we sought a different class of substrates for our catalytic reaction that would rely on an alternative method of enolate generation that was A) specific for the production of a desired tetrasubstituted enolate, B) did not rely on a thermodynamic equilibration under basic conditions, and C) maintained high enantioselectivity in the overall transformation.

Scheme 6.

Non-selective Enolization Leads to Product Mixtures.

Results and Discussion

Enantioselective Alkylation with Racemic Allyl β-Ketoesters

β-Ketoesters represent a classical solution to the problem of regioselective ketone alkylation (e.g., the acetoacetic ester synthesis).^[54] In 1980, Tsuji and Saegusa reported allyl β-ketoesters as substrates in non-enantioselective Pd-catalyzed enolate alkylations.^[21] We reasoned that an allyl β-ketoester might solve the problem of selective enolate formation in the course of an enantioselective allylation reaction. However, as stereogenic racemic substrates for a catalytic asymmetric reaction, allyl β-ketoesters presented some unique challenges: the catalyst must deallylate the substrate nonselectively to avoid kinetic resolution of the substrate, the C–C bond must break at a reasonable temperature in the decarboxylation step to prevent loss of selectivity in the subsequent bond-forming step, and the stereogenic intermediates must not experience diastereomeric transition states that are detrimental to the selectivity of the process.^[55] Nonetheless, we prepared the appropriate substrate (44, Scheme 7) using standard methods^[53] and exposed racemate (±)-44 to our standard reaction conditions (Pd₂(dba)₃, (S)-tBu-PHOX, THF, 25 °C). Unfortunately, no conversion was observed (by TLC) under these conditions after 24 h. However, we were delighted to find that elevating the temperature to 30 °C and carrying out the reaction in Et₂O restored reactivity, providing enone (−)-45 in 73% yield (>99% conversion) and 86% ee after 9 h. This experiment provided a key proof-of-concept for an enantioconvergent reaction using a racemic allyl β-ketoester.

Scheme 7.

Enantioselective Decarboxylative Alkylation with an Allyl β-Ketoester.^[27b].

When further exploring this pivotal reaction, we noted that no significant kinetic resolution of the allyl β-ketoesters was observed.^[56] In fact, this facet of the reaction was critical in order to obtain high yield in a reasonable reaction time. The similar levels of enantioselectivity observed with the allyl β-ketoester substrates and the other substrate classes (cf. Tables 7 and 8 with Tables 12 and 13) suggest that the enantiodetermining transition state of the reaction remains unchanged. Although the extent to which decarboxylation slowed at ambient temperature varied from substrate to substrate, β-ketoesters were uniformly more sluggish in decarboxylation than allyl enol carbonates.^[57] However, the rate of reaction increased greatly at slightly higher temperatures, and thus reasonable reaction times (~2 h) could be achieved by increasing the reaction temperature by ca. 5 °C, which had a negligible effect on product ee (Table 11).

Table 12.

Enantioconvergent Decarboxylative Allylation of α-Substituted 2-Carboxyallylcyclohexanones.^[27b]

entry	R	time (h)	yielda	eeb	entry	R	time (h)	yielda	eeb
1	CH3	7.5	85	88	6	CH2C6H5	0.5	99	85
2c	CH3	4.75	89	88	7	CH2(4-CH3OC6H4)	10	80	86
3c,d	prenyl	6	97	91	8	CH2(4-CF3C6H4)	0.5	99	82
4c	CH2CH2CN	6.5	97	88	9e	CH2OTBDPS	5	86	81
5c,e	CH2CH2CO2Et	6	96	90	10c,d	F	3.5	80	91

^aIsolated yield (%) from reaction of 1.0 mmol substrate with 2.5 mol% Pd₂(dba)₃ and 6.25 mol% (S)-tBu-PHOX (19) at 0.033 M in THF at 25 °C, unless otherwise noted.

^bEnantiomeric excess (%) determined by chiral GC or HPLC.

^cPerformed in Et₂O solvent.

^dPerformed at 30 °C.

^e4 mol% Pd₂(dba)₃, 10 mol% (S)-19, 0.021 M.

Table 13.

Enantioconvergent Decarboxylative Allylation of β-Ketoesters with Substituted Carbocycles and Allyl Fragments.^[24e,f,27a]

entry	product	% yielda		% eeb	entry	product		% yielda	% eeb
1			94	85	8f			98	90
2c			94	86
3			89	90	9d,e		R = CH3	87	92
					10d,e		R = Cl	87	91
4			90	85	11g			96	94
5d,e			77	90	12			91	92
6d			97	92	13h		R = i-BuO	86	75
					14i		R = PhS	85	92
7			83	87	15d,j		R = H	68	83
					16d,j		R = Cl	97	91

^aIsolated yield from reaction of 1.0 mmol substrate, 2.5 mol% Pd₂(dba)₃ and 6.25 mol% (S)-tBu-PHOX (19) at 0.033 M in THF at 25–35 °C, unless otherwise noted. Each reaction was complete in 1.5–12 h.

^bDetermined by chiral GC or HPLC.

^c25 mmol substrate, 1.5 mol% Pd₂(dba)₃, and 3.75 mol% (S)-19, 24 h reaction time.

^dPerformed in Et₂O.

^e4 mol% Pd₂(dba)₃, and 10 mol% (S)-19 at 0.021 M.

^fReaction performed with 3.17 mmol substrate, 3 mol% Pd(dmdba)₂, and 3.8 mol% (S)-19 at 0.060 M.

^g2.5 mol% Pd(dmdba)₂, 2.5 mol% (S)-19.

^hPerformed with 2.5 mol% Pd₂(pmdba)₃ and 6.25 mol% (S)-19 in PhMe at 80 °C.

ⁱPerformed with 2.5 mol% Pd₂(pmdba)₃ and 6.25 mol% (R)-19 in PhMe at 50 °C; major product was enantiomeric to the shown structure.

^j5 mol% Pd(dmdba)₂, 6.25 mol% (S)-19.

Table 11.

Temperature Effects in Decarboxylative Allylation of Allyl β-Ketoesters.

entry	temperature (°C)	time (h)	% yielda	% eeb
1	18	48	0	ND
2	25	7.5	85	88
3	30	2.25	82	87
4	35	1.25	85	86
5	40	0.67	86	85
6	60	0.15	82	83

^aIsolated yield from reaction of 1.0 mmol substrate at 0.033 M.

^bDetermined by chiral GC.

This enantioconvergent reaction is unusual due to the use of quaternary β-ketoesters (Scheme 8). Typical stereomutative enantioconvergent processes (e.g., dynamic kinetic resolution) involve a pre-equilibrium epimerization of the starting material, A, followed by enantioselective conversion to product B (Pathway I).^[58] However, quaternary stereocenters are not typically epimerizable, and we believe that both enantiomers of the starting material, A, convert irreversibly to a prochiral intermediate, C,^[59] which preferentially forms one enantiomer of the product, B, under the influence of the chiral catalyst (Pathway II). We have termed such transformations as stereoablative enantioconvergent processes due to a lack of an existing term to adequately describe this interesting reaction pathway.^[60],[61]

Scheme 8.

Stereoablative Enantioconvergent Catalysis.

As a result of the facile synthesis of racemic quaternary β-ketoester substrates, we have been able to expand the substrate scope of the asymmetric alkylation greatly.^[53] A number of α-substituted 2-carboxyallylcyclohexanones were readily prepared by the above methods and successfully underwent enantioconvergent decarboxylative allylation (Table 12). The system was permissive of an array of functionality and substitution, and formed products in excellent yield and high ee. Remarkably, the presumed enolate intermediate tolerated both enolizable (entries 4 and 5) and β-heteroatom (entry 9) substituents without side products corresponding to enolate scrambling or β-elimination.^[62] Our conditions also allowed for incorporation of a fluorine atom for the high-yielding synthesis of a stereodefined tertiary fluoride (entry 10). Subsequent to our initial disclosure, others have utilized this chemistry to prepare a range of tertiary fluoride-containing compounds that could be of interest as non-epimerizable pharmaceutical analogues.^{[42c],[42e],[42h],[42j],[63]}

In addition to modifications at the α-position of the substrate, the decarboxylative asymmetric allylation of β-ketoesters was also amenable to a wide variety of modification to the carbocycle and allyl fragment (Table 13). In particular, the reaction was exceptionally tolerant to the steric demands of substitution at the 3-, 4-, 5-, and 6-positions of the cyclohexane ring; each position can be fully substituted without significantly affecting yield or enantioselectivity (entries 1–4). Important to multi-step synthetic efforts, we found the protocol to be scalable with no detriment to the results (entry 2).^[64] Unsaturated substrates (entries 5 and 6) as well as seven-membered ring containing substrates (entries 7 and 8) performed well in the reaction. As with the previous substrate classes (i.e., allyl enol carbonates and silyl enol ethers), substitution at the central position of the allyl fragment had a beneficial effect on the enantioselectivity of the reaction (entries 9–11).^[65] The incorporation of a chlorine atom on the allyl fragment (entry 10) provides both another functional group handle for further manipulation and a higher oxidation state. Inclusion of a phenyl group on the allyl fragment (entry 11) produced the highest ee product we have observed to date (94% ee). Heterocyclic compounds were also accessible by this method (entry 12). Vinylogous esters were competent in the reaction, but elevated temperatures (up to 80 °C) were required in some cases to achieve high conversion and the ee was somewhat decreased (entry 13). Employing a vinylogous thioester lowered the required temperature and permitted the isolation of highly enantioenriched product (entry 14).^[66] This particular vinylogous thioester was employed in the enantioselective syntheses of (+)-carissone^[24e] and (+)-cassiol^[24f] (see Scheme 10e). We have demonstrated that dioxanones are viable substrates as well (entries 14 and 15). The tetrasubstituted α-alkoxycarbonyl products were furnished in good yields and enantiomeric excesses, and the products could be readily derivatized to access a number of enantioenriched α-hydroxycarbonyl compounds.^[27c]

Scheme 10.

Enantioenriched Cyclic Ketones in the Synthesis of Natural Products.^[24],[27c]

Challenging Substrate Classes

Synthesis of Tertiary Stereocenters from Acyclic Enolate Precursors

Although we have principally employed our asymmetric allylation for the synthesis of fully-substituted stereocenters, the mild and nearly neutral conditions of the reaction are well suited for the synthesis of tertiary stereocenters α to carbonyls (Table 14).^[67] Such stereocenters are prone to epimerization and over-alkylation under the strongly basic conditions traditionally used for enolate generation.^[68] We found that the allyl enol carbonate underwent allylation to form phenyl ketone (R)-46 in 67% ee (entry 1). In analogy to the results of Hou and coworkers,^[18b] we found that the addition of AgBr gave a significant increase in the ee of the product (entry 2). Both TMS^[69] and TBS^[70] silyl enol ethers were competent enolate precursors when TBAT was used as an initiator (entries 3 and 4). Unlike the tetrasubstituted silyl enol ethers, CsF proved to be the optimal fluoride source for the less substituted silyl enol ethers, resulting in noticeably higher ee (entry 5). Unfortunately, the use of AgBr in the presence of CsF greatly reduced reactivity and only increased the product ee slightly (entry 6). Similarly, the use of ethyl ether as a solvent increased enantioselectivity at the expense of yield (entry 7). Although our allylation methods have provided tertiary stereocenters in only moderate ee, we have recently disclosed a complementary decarboxylative protonation of quaternary allyl β-ketoesters based on a similar catalyst system that provides access to tertiary stereocenters α to ketones in excellent ee.^[71]

Table 14.

Enantioselective Allylation of Allyl Enol Carbonates and Silyl Enol Ethers Not Contained in a Ring.

entry	R	additive	% yielda	% eeb	entry	R	additive	% yielda	% eeb
1	CO2allyl	none	79	67	5d	TMS	CsFf	79	77
2	CO2allyl	AgBrc	75	79	6d	TMS	CsFf	20	79
3d	TMS	TBATe	60	62			AgBrc
4d	TBS	TBATe	82	73	7d,g	TMS	CsFf	38	82

^aIsolated yield from reactions with 0.092 mmol of substrate in THF (0.092 M) with 2.5 mol% Pd₂(dba)₃ and 6.25 mol% (S)-tBu-PHOX (19), unless otherwise noted.

^bMeasured by chiral HPLC.

^c40 mol% additive.

^dReaction performed with diallyl carbonate (1.05 equiv).

^e35 mol% additive.

^f48 mol% additive.

^gPerformed in Et₂O as solvent.

Application of the Palladium•PHOX Catalyst System to Propargylation

In addition to allylation, we also explored propargylation of enol carbonates with the Pd•PHOX catalyst system (Table 15).^[72] Our preliminary studies showed that propargylation of enol carbonate 47 required significantly higher temperatures than are needed for allylation.^[73] Additionally, the optimal structure of the PHOX ligand is significantly different for propargylation than for allylation. Moving the bulk of the tBu group away from the oxazoline by insertion of a methylene group gave (S)-48 with higher ee (entries 1 and 2). Unlike allylation (cf. Table 4), PHOX ligands prepared from phenylalanine derivatives gave higher product ee than those prepared from saturated amino acids, with the 9-anthracenylalanine derivative (51) giving the highest level of enantioselectivity (entry 7, Table 15). Although still preliminary, these studies suggest that the Pd•PHOX catalyst system may find use with electrophiles other allyl groups.^[27e],[71]

Table 15.

Enantioselective Propargylation.

entry	ligand	% yielda		% eeb	entry	ligand	% yielda		% eeb
1	19, R = t-Bu		84	12	4	24, R = CH2-1-Naphthyl		57	37
2	23, R = CH2-t-Bu		83	32	5	49, R = CH2-2-Naphthyl		71	26
					6	50, R = CH2-(3,5-di-t-Bu-Ph)		94	25
3	18, R = CH2Ph		54	26	7	51, R = CH2-9-Anthracenyl		80	44

^aGC yield relative to internal standard (tridecane).

^bEnantiomeric excess measured by chiral GC. In all cases, (S)-48 was the major enantiomer.

Substrates Proceeding via Weakly Basic Enolates

The alkylation of substrates derived from ketones of unusually low pK_a (i.e., stabilized enolates) as a group gave by far the lowest levels of enantioselectivity we have observed with the Pd•PHOX catalyst system (Table 16). Despite the low selectivity, these substrates gave consistently excellent chemical yields of the allylated products. It is noteworthy that the β-ketoester (entries 1 and 2) and α-aryl ketone (entries 3 and 4) derived enolates, which gave excellent enantioselectivity in Trost’s earlier asymmetric allylation reaction^[15],[16b] failed to give useful levels of enantioinduction under our conditions.^[74] An oxazole substrate, designed with the intention of executing an enantioselective synthesis of α,α-disubstituted amino acids, also underwent allylation with little enantioselectivity, presumably due to the stability of the intermediate aromatic enolate (entry 5). The orthogonality of the substrate scope, in terms of asymmetric induction, between our method and Trost’s early reports may be indicative of fundamentally different mechanisms underlying the two allylation reactions and is an important piece of evidence in our mechanistic hypothesis.^[26],[75]

Table 16.

Asymmetric Alkylation via Stabilized Allyl Enol Carbonates.^[75]

entry	substrate	product	% yielda		% eeb	entry	substrate	product	% yielda	% eeb
1				89	24c	4			93	0
2				87	2
3				99	11c	5			89	2

^aIsolated yield from reactions with 1.0 mmol of substrate in THF (0.033 M) at 25 °C with 2.5 mol% Pd₂(dba)₃ and 6.25 mol% (S)-tBu-PHOX (19). Each reaction was complete in 2 h.

^bMeasured by chiral GC or HPLC.

^cAbsolute stereochemistry of products assigned by analogy.

Substrates Containing Five-Membered Rings

Although enolate precursors contained in six-membered rings composed the bulk of our substrates, we have demonstrated that seven- and eight-membered rings are tolerated with only a slight decrease in product ee (see Tables 7, 8, and 13). Allyl β-ketoesters constructed on five-membered rings provided modest to useful levels of enantioselectivity (Table 17). These substrates generally produced α-quaternary ketone products in good yields with enantiomeric excesses about 10% lower than that observed with the cyclohexanone analogue. Ethyl substituted cyclopentanone was formed in 86% ee, only 6% lower than observed with the corresponding cyclohexanone (entry 1). Indanones were produced in good yield and with useful ee (entries 4 and 5). Benzyl appended cyclopentanone substrates gave consistent yields, but electron deficient aromatic rings decreased product ee more significantly than in the reactions of six-membered β-ketoesters (entries 6–8).

Table 17.

Enantioconvergent Decarboxylative Allylation of β-Ketoesters Containing Five-Membered Rings.

entry	product			yielda	eeb	entry	product		yielda	eeb
1		R = CH2CH3		82	86
2		R = CH(CH3)2		77	84	6		R = OMe	84	73
3		R = CH2NPhth		67	48	7		R = Me	84	73
4c			R = Me	82	80	8		R = CF3	83	60
5			R = Bn	93	71

^aIsolated yield (%) from reaction of 1.0 mmol of substrate in THF (0.033 M) with 2.5 mol% Pd₂(dba)₃ and 6.25 mol% (S)-tBu-PHOX (19), unless otherwise noted. Each reaction was complete in 1–3 h.

^bEnantiomeric excess (%) measured by chiral GC or HPLC.

^cPerformed on 0.1 mmol scale with 5 mol% Pd₂(dba)₃ and 12.5 mol% (S)-19.

Synthetic Applications

The α-quaternary cycloalkanones produced in the asymmetric Tsuji alkylation are highly useful chiral building blocks. Each substrate contains at least two functional groups, a ketone and an olefin, for further manipulation. Moreover, the preceding section has demonstrated these allylation reactions to be highly functional group tolerant. The application of this suite of allylation reactions to the catalytic asymmetric synthesis of natural products is an ongoing topic of research in our laboratories.^[24]

To further demonstrate the utility of these products, we transformed ketone (S)-9 into several familiar cyclic frameworks (Scheme 9a). Wacker oxidation of (S)-9 followed by aldol condensation gave enone 52 in good yield. Another functionalized [6-5] skeleton was formed in a three-step sequence by olefin cross metathesis with methyl vinyl ketone catalyzed by Grubbs’ second generation Ru-complex 53,^[76] olefin hydrogenation, and aldol condensation under basic conditions to afford exocyclic enone 54. Carbocyclic [6-6] ring systems were accessible as well: multi-step elaboration of the allyl group afforded an intermediate diketone, which underwent aldol condensation to enone 55 in 42% overall yield. Enone 55, which has been classically produced by Robinson annulation, has been used extensively in synthesis.^[77] As a final transformation of (S)-9, we executed a Baeyer–Villiger oxidation with peracetic acid to give caprolactone 56. This transformation demonstrates the conversion of our enantioenriched quaternary stereocenter into a latent tertiary alcohol with defined absolute stereochemistry and entry into acyclic systems.

Carbocyclic spiro compounds represent a particularly challenging subclass of quaternary stereocenters. To complement the syntheses of fused cyclic skeletons above, allyl methallyl ketone 57 was used as an entry into the synthesis of ring systems containing spiro quaternary stereocenters. Ketone 57 could be treated with Grubbs’ second generation catalyst (53)^[76] in dichloromethane to give a good yield of the spiro[4.5]ketone 58 (Scheme 9b). Spiro[5.5]enone 59 was produced in modest yield by treatment of ketone 57 with standard ketal protection conditions, followed by ozonolysis, and exposure to base (Scheme 9c).

The dioxanone class of substrates provided entries into other important derivatives (Scheme 9d). Cleavage of the acetal with p-toluenesulfonic acid generated dihydroxy ketones (e.g., 60 → 61). These dihydroxy ketones underwent selective oxidative cleavage upon exposure to periodic acid and after esterification the corresponding hydroxyesters were isolated (e.g., 62). Overall yields for the transformation from dioxanone to hydroxyester (60 → 62) were 41–81% for three steps. In the case of chlorinated compound 62, subsequent ozonolysis formed citramalic acid dimethyl ester (63) that was correlated to literature data to establish the absolute configuration of dioxanone (−)-60 as S.^[78]

Similar tactics have enabled the use of the enantioselective Tsuji alkylation in a number of synthetic efforts.^[24] Ketone (S)-64 (Scheme 10a) was readily converted to enone 65, which could be recrystallized via the semicarbazone derivative to 97% ee. Enantioenriched enone 65 was then elaborated to the antipode of the natural product dichroanone (66) in seven additional steps.^[24a] A strategy comprising enantioselective allylation and ring-closing metathesis was employed to synthesize the spirocyclic natural products laurencenone B (68) and elatol (69) from vinylogous ester 67 (Scheme 10b).^[24c] An ongoing effort toward the preparation of the marine alkaloid zoanthenol (72) features α-quaternary ketone 70 as a key synthon for the preparation of the ABC tricyclic substructure (71) of the target molecule (Scheme 10c).^[24b] Dioxanone 73 was converted to cyclohexene 74 in three steps and 50% yield (Scheme 10d). Hydroxy acid 74 has previously been employed in the synthesis of quinic acid (75),^[79] a common chiral pool starting material used in many synthetic efforts,^[80] including the total synthesis of dragmacidin F reported by our laboratory.^[81] Vinylogous ester 76 is a versatile precursor and can be elaborated to the natural products carissone (77)^[24e] and cassiol (78).^[24f]

In an effort to construct more than one quaternary stereocenter in a single transformation, we designed substrate 79 (Scheme 11a), which contained a latent allyl β-ketoester moiety to be revealed by the reaction of the allyl enol carbonate portion of the molecule. To our delight, cascade allylation occurred to afford C₂-symmetric ketone 80 as the predominant product in 92% ee with 4:1 dr. Similarly, bis(β-ketoester) substrates may be employed in a stereoconvergent process where each of the three stereoisomeric starting materials (i.e., two C₂ symmetric enantiomers and one meso diastereomer) are converted to enantioenriched products with excellent stereocontrol (e.g., 81 → 80, Scheme 11b and 82 → 83, Scheme 11c). We have successfully employed the double enantioselective decarboxylative allylation strategy in the total synthesis of the cyathin diterpenoid cyanthiwigin F (84, Scheme 11c).^[24d]

Scheme 11.

Enantioselective Cascade Allylations Generating Two Quaternary Centers.^[27b,24d]

Mechanistic Insights

Although we have not fully elucidated the fine mechanistic details of enantiodiscrimination with the Pd•PHOX catalyst, an intriguing picture of the reaction’s general mechanism has emerged from our experimental studies. A number of experimental observations suggest that the mechanism of our allylation of prochiral nucleophiles differs substantially from that of Pd•PHOX-catalyzed malonate alkylation of prochiral allyl fragments, which has been studied in detail by Helmchen^[82a–d] and Pfaltz.^[82e]

Helmchen’s model for the asymmetric allylic alkylation of prochiral allyl electrophiles (R ≠ H, Scheme 12) with Pd•PHOX catalysts involves attack at one allyl terminus of a (PHOX)Pd-π-allyl complex by an outer-sphere malonate anion. The allyl group isomers 85 and 86 are in rapid equilibrium such that the nucleophile’s preferred attack (86), from the open quadrant at the allyl terminus trans to phosphorus, is the nearly exclusive reaction pathway. The resulting Pd(0)•olefin complexes (e.g., 87) have been observed by low temperature NMR.^[82a,b] Notably, stereoinduction in such alkylation reactions suffered when the steric bulk of the allyl substituents was reduced (e.g., 96% ee when R = i-Pr, 71% ee when R = Me).^[83] It was difficult to rationalize the high levels of stereoinduction observed in our allylation by this outer-sphere mechanism in that our substrates typically have unsubstituted allyl termini. Additionally, formation of the new stereocenter on the enolate via an outer-sphere mechanism would require the chiral Pd complex to differentiate the prochiral faces of an unassociated enolate, where steric and electronic interactions between the nucleophile and the distant chiral backbone would be minimal.^[84] Moreover, a gearing effect through the allyl ligand is unlikely given the lack of sterically large groups (R = H).

Scheme 12.

Helmchen’s Outer-Sphere Mechanism for Asymmetric Allylation of Prochiral Allyl Electrophiles.

In addition to these hypothetical arguments, we have made several experimental observations that do not correlate well to an outer-sphere enolate attack mechanism. The high enantioselectivity under our conditions appears to correspond with circumstances that would keep ion pairs tightly associated. The range of effective solvents found during our optimization studies demonstrated this trend: ethereal solvents (e.g., THF), aromatic solvents (e.g., benzene), ethyl acetate, and triethylamine share few properties other than having low dielectric constants in the range of 2 to 8, but each provided products in high ee (see Table 3). In such low dielectric media, dissociative solvation of ion pairs is difficult.^[85] In conjunction with the lack of other counterions in the reaction, the low dielectric value would tend to enforce an inner-sphere mechanism wherein close contact of the enolate and the chiral environment would facilitate the discrimination between the prochiral enolate faces.^[86]

Based on our initial reasoning and empirical observations, we developed a working model for the course of the reaction via an inner-sphere mechanism (Scheme 13). Oxidative addition of Pd(0)•PHOX complex 88 to substrate 8 or 10 leads to [Pd(allyl)PHOX] complex 41 and carboxylate 89 or mixed carbonate 90. We believe that an equilibrium exists between the charge-separated form and the coordination complex bearing an η¹-allyl group (e.g., 91). We have isolated and characterized the β-ketocarboxylate analog (91) by X-ray crystallography and found that it can be decomposed to deliver enantioenriched ketone 9, and therefore it appears to be an important intermediate on the reaction pathway.^[87] Subsequent loss of CO₂, which is most likely Pd-assisted, leads to enolate 14 and the Pd-complex 41. Association of these two fragments leads to 5-coordinate intermediate 92, a compound resembling a transition state for associative substitution with the enolate bound to the sterically less-encumbered apical position of the square-planar metal. By slightly different means, allyl enol carbonate and silyl enol ether substrates may intercept the key structure 92. Shifting of the allyl group from an η³- to an η¹-binding mode allows the enolate fragment to move into the square plane and form intermediate 93, thus completing the associative ligand substitution. From this complex, a standard 3-centered reductive elimination would require a shift from O- to C-bound Pd-enolate.^[88] Although possible, we reasoned that this isomerization would be unfavorable due to the steric demands of the resulting tertiary palladium species. As an alternative, we postulated an extended reductive elimination pathway proceeding through a cyclic 7-membered pericyclic transition state (94-boat or 94-chair) similar to a sigmatropic rearrangement.^[89],[90] Importantly, situating the carbocyclic ring of the enolate fragment away from the bulky tBu group of the oxazoline provides a plausible predictive model for the observed absolute sense of enantiofacial selectivity.^[37] Liberation of the product ketone (9) from complex 95 regenerates the catalytic Pd(0) species 88.

Scheme 13.

Proposed Inner-Sphere Mechanism for Asymmetric Alkylation of Prochiral Enolate Nucleophiles.

Typically, inner- and outer-sphere processes may be differentiated by stereochemically labeling the allyl fragment and evaluating the relative stereochemistry of the products.^[91] For example, this type study was carried out by Trost and coworkers on their decarboxylative allylic alkylation system, and the results suggest an outer-sphere mechanism.^[42l] However, in the present case, these experiments could not be performed because the necessary cyclic carbonates suffered from very poor reactivity and enantioselectivity. In fact, even simple substitution of the termini of acyclic allyl groups (e.g., crotyl carbonates) severely impacted reactivity and perhaps suggests that for such substrates a different mechanism is operative.^[92] Given these substrate limitations, we were obliged to seek out other forms of mechanistic evidence to evaluate our proposal.

Since it is conceivable that multiple ion pairs could be involved in the transition state (e.g., an enolate bound to one [Pd(allyl)PHOX] fragment could attack the Pd-bound allyl fragment associated with another enolate), we conducted simple kinetics experiments that support a pathway involving a single metal center (Figure 1 through Figure 4).^[53] Initial attempts to examine the kinetics of the reaction with respect to the Pd•PHOX complex did not give clear results. We suspect that the precision of the calculated initial rates may be complicated by inconsistent aerobic oxidation of the phosphine ligand^[32] leading to some uncertainty in the effective catalyst concentration. A fit to a rate law first-order in Pd•PHOX concentration is most consistent with the experimental observations, tentatively supporting the notion that a single metal center is involved in the mechanism (Figure 1). Additionally, the lack of an observed non-linear effect between product and ligand ee is also consistent with the action of a single metal center in the enantiodetermining step (Figure 2).^[93] Interestingly, kinetic studies also found a zero-order dependence on substrate concentration in the allyl enol carbonate (Figure 3) and β-ketoester (Figure 4) cases.^[94] Based on these results, we suspect that the rate-limiting step is either loss of CO₂ from a Pd-carboxylate intermediate (e.g., 91 → 41 + 14, Scheme 13), or C–C bond formation (e.g., 93 → 95, Scheme 13). The isolation of the relatively stable Pd-carboxylate complex 91 is indicative of a slow decarboxylation step. The longer reaction times observed with β-ketoester substrates relative to the corresponding allyl enol carbonates is also more consistent with decarboxylation as the rate determining step, as the rate difference would be attributed to the difference in breaking a C–C bond (β-ketoester) compared to a C–O bond (enol carbonate). However, we cannot rule out the possibility that different substrate classes may have different rate-determining steps.

Figure 1.

Plot of k_obs vs. Concentration of Pd(PHOX) Complex for an Allyl Enol Carbonate.

Figure 4.

Plot of Allyl β-Ketoester ¹H NMR Integral vs. Time (s).

Figure 2.

Plot of iPr-PHOX ee vs. Product ee.^[95]

Figure 3.

Plot of Allyl Enol Carbonate ¹H NMR Integral vs. Time (s).

In addition to these studies, we designed other experiments to differentiate our proposed mechanism from that elucidated by Helmchen.^[82] Enolates where charge is delocalized (e.g., those derived from β-ketoesters or α-aryl ketones), and should therefore tend to form weak ion pairs, give high yields but extremely low levels of enantioselectivity (see Table 16).^[96] This result suggests that, in such cases, allylation proceeds primarily via the more conventional, although in this case poorly selective, outer-sphere attack. The reactivity of sterically demanding substrates in our reaction is also inconsistent with an outer-sphere, intermolecular nucleophilic enolate attack. In a nucleophilic bimolecular reaction, such as the Helmchen allylation mechanism (Scheme 12), steric bulk near the site of bond formation typically impedes the rate of the reaction.^[97] However, in our case there is little difference in the reaction time, yield, or enantioselectivity when comparing the formation of ketone 9 with more sterically demanding ketones (Figure 5). This observation is more consistent with an intramolecular reaction mechanism.

Figure 5.

Comparison of Sterically Varied Products Formed at 25 °C in THF.

Another observation at odds with an external attack mechanism is the unusual tolerance to water (Table 18). Multiple equivalents of water introduced into the reaction have only a moderate effect on the yield of the reaction. This contrasts typical enolates, which are rapidly quenched by water even at low temperatures, and suggests that once decarboxylation occurs the intermediate enolate formed in the reaction is tightly associated with or covalently bound to its Pd counterion for most of its lifetime.^[98]

Table 18.

Effect of Water on Asymmetric Alkylation.

entrya	H2O added (equiv)b	% yieldc	% eed
1	0	95	86
2	0.55	99	87
3	1.64	88	84
4	8.25	70	61
5	16.5	67	49
6	33.3	63	40

^aData reported are the average of three trials.

^bH₂O added after Pd/PHOX complexation, but before substrate.

^cGC yield relative to internal standard (tridecane).

^dEnantiomeric excess measured by chiral GC.

In an effort to trace the fate of the allyl and enolate fragments in the course of the reaction, we performed crossover experiments with a 1:1 mixture of deuterated allyl enol carbonates 97 and 98 in THF, dioxane, and benzene (Scheme 14).^[99] Analysis of the products by high-resolution mass spectrometry revealed the presence of all four possible product masses in nearly equal amounts.^[100] In conjunction with the water addition experiments, this observation suggests that a palladium enol carbonate species (41 + 90 or 41•90) may be a long-lived intermediate. Such an intermediate would not be readily protonated by water,^[101] and this delocalized anion may facilitate crossover by dissociation from the metal center. In the case of β-ketoester substrates, a β-ketocarboxylate intermediate (e.g., 91) may play an analogous role.^[102]

Scheme 14.

Crossover Experiments with Deuterated Allyl Enol Carbonates.^[27b]

In order to validate our hypothesis regarding the persistence of an intermediate carboxylate or mono-alkyl carbonate (89 or 90, respectively), we collected ³¹P NMR data during the reaction.^[27d],[53] Non-ligated tBu-PHOX (19) exhibited a ³¹P resonance at −5.9 ppm. When Pd₂(dba)₃ and tBu-PHOX were complexed, a new resonance was observed at 18.8 ppm along with excess ligand (Figure 6, spectrum A). When substrate (±)-10 was added, the resonance at 18.8 ppm disappeared immediately and the only observed intermediate was found at 30.9 ppm, which persisted until the end of the reaction (Figure 6, spectrum B). When the substrate was completely consumed, the initial Pd•PHOX complex returned at 18.8 ppm (Figure 6, spectrum C).^[103] Interestingly, similar behavior was observed in reactions with allyl carbonate reactions, although the chemical shift of the intermediate species differed slightly.

Figure 6.

³¹P NMR collected during the enantioselective allylation reaction: Spectrum A: (PHOX)Pd(dba) complex (18.8 ppm)^[27d] and excess PHOX ligand (S)-19 (−5.9 ppm) prior to addition of substrate; Spectrum B: After addition of substrate (±)-10, during reaction. Carboxylate 91 is observed at 30.9 ppm; Spectrum C: After complete consumption of substrate.

The separately prepared [Pd(II)(allyl)(PHOX)]⁺[PF₆]⁻ complex (41, Scheme 5) exhibited two ³¹P resonances (other than those due to PF₆⁻) at 23.5 and 22.5 ppm in a 2:1 ratio (in THF), due to the mixture of endo- an exo-allyl isomers.^[82] The different chemical shifts for this Pd(II) complex relative to that observed during the reaction suggest a different ligation environment about the Pd center. To correlate these two different species, nBu₄NOAc was added to a solution of [Pd(II)(allyl)(PHOX)]⁺[PF6]⁻, and a new species was observed at 30.9 ppm, which matches independently synthesized [Pd(II)(allyl)(PHOX)(OAc)].^[27d] Based on these experiments, we believe that the observed catalyst resting state in solution is likely the [Pd(II)(allyl)(PHOX)]⁺/[carboxylate]⁻ ion pair (41 + 89, Scheme 13) or the coordination complex 91. Efforts to fully characterize additional intermediate structures are ongoing.

The solid-state structure of independently prepared [Pd(II)(allyl)PHOX]⁺[PF₆]⁻ salt 41 lends credence to the possibility of an apically bound palladium enolate. As shown in Scheme 5, after recrystallization the complex partially co-crystallizes with a molecule of ethanol in its unoccupied quadrant.^[47] We envision this as the likely site of enolate coordination (e.g., apically bound enolate complex 92, Scheme 13).

Given the body of evidence accrued in these experimental studies and our inability to probe the mechanism through classical stereochemical labeling experiments,^[91] we initiated a collaboration with the Goddard group at Caltech in order to investigate the details of the mechanism via computational modeling.^[75] The computational studies carried out have corroborated our initial hypothesis regarding the inner-sphere mechanism, the difficult isomerization of O- to C-bound Pd enolate, and the unusual cyclic 7-membered reductive elimination pathway (Scheme 13). Notably, these calculations have found that the traditional outer-sphere mechanism (Scheme 12) is significantly higher in energy than the inner-sphere pathway. The simulations have not yet elucidated the fine details of the origin of enantioselectivity. Experiments and computational studies to better characterize the reaction mechanism are under way with the hope of improving the enantioselectivity and scope of the asymmetric alkylation.

Conclusion

In order to address a significant deficiency in the asymmetric alkylation literature, we have developed a strategy to prepare enantioenriched cycloalkanones in high yield. This work provides direct access to valuable enantioenriched α-quaternary ketones from readily accessible enolate precursors. Critical to our success was the use of enolate precursors, which can be converted to enolates under mild conditions, and the use of a chiral catalyst that exhibited the enolate isomeric fidelity found in Tsuji’s non-enantioselective system. The reactivity and enantioselectivity of the alkylation reactions have proven to be quite general with respect to substrate steric bulk, ring size, unsaturation, and diverse functional groups. We have demonstrated the relevance of the α-quaternary ketones produced in the reaction by their conversion to a number of carbocyclic chiral building blocks, including several spiro quaternary motifs, as well as in the asymmetric syntheses of several natural products. Studies directed toward further development of the scope of this reaction and catalyst system are ongoing. Additionally, the continued application of our asymmetric alkylation as a key enantiodetermining step in natural products synthesis will be reported in due course.

Experimental Section

Representative procedure for the enantioselective decarboxylative alkylation of allyl enol carbonates

A 50 mL round-bottom flask equipped with a magnetic stir bar was flame dried under vacuum. After cooling under dry argon, Pd₂(dba)₃ (22.9 mg, 0.025 mmol, 0.025 equiv) and (S)-tBu-PHOX (19, 24.2 mg, 0.0625 mmol, 0.0625 equiv) were added. After the flask was flushed with argon, THF (30 mL) was added, and the contents were stirred at 25 °C for 30 min, at which time allyl enol carbonate 8 (196.2 mg, 1.0 mmol, 1.0 equiv) was added by syringe in one portion. When the reaction was complete by TLC (2 h), the reaction mixture was evaporated under reduced pressure and the residue chromatographed (2→3% Et₂O in pentane on SiO₂) to afford (S)-2-allyl-2-methylcyclohexanone ((S)-9) (129.6 mg, 85.1% yield, 87% ee) as a colorless oil; ¹H NMR (300 MHz, CDCl₃) δ 5.75-5.61 (m, 1H), 5.05 (s, 1H), 5.01 (m, 1H), 2.40-2.31 (comp. m, 3H), 2.21 (dd, J = 13.8, 7.5 Hz, 1H), 1.78 (comp. m, 5H), 1.56 (m, 1H), 1.06 (s, 3H); ¹³C NMR (75 MHz, CDCl₃) δ 215.4, 133.7, 117.9, 48.4, 41.9, 38.8, 38.5, 27.4, 22.6, 21.0; IR (Neat Film NaCl) 2934, 2865, 1707, 1451, 912 cm⁻¹; HRMS (EI) m/z calc'd for C₁₀H₁₆O [M]+: 152.1201, found 152.1204; [α]_D²¹ −49.64 (c 2.38, hexane, 98% ee).

Representative procedure for the enantioselective decarboxylative alkylation of silyl enol ethers

A 50 mL round-bottom flask equipped with a magnetic stir bar was flame dried under vacuum. After cooling under dry argon, Pd₂(dba)₃ (22.9 mg, 0.025 mmol, 0.025 equiv), (S)-tBu-PHOX (19, 24.2 mg, 0.0625 mmol, 0.0625 equiv), and TBAT (189 mg, 0.35 mmol, 0.35 equiv) were added. After the flask was flushed with argon, THF (30 mL) was added, and the contents were stirred at 25 °C for 30 min, at which time diallyl carbonate (150.6 µL, 1.05 mmol, 1.05 equiv) and silyl enol ether SI23 (184.35 mg, 1.0 mmol, 1.0 equiv) were added sequentially by syringe in single portions. When the reaction was complete by TLC (2 h), the reaction mixture was evaporated under reduced pressure and the residue chromatographed (2→3% Et₂O in pentane on SiO₂) to afford ketone (S)-9 (144.3 mg, 94.8% yield, 87% ee).

Representative procedure for the enantioselective decarboxylative alkylation of β-ketoesters

A 100 mL rb flask was equipped with a magnetic stir bar and flame dried under vacuum. After cooling under dry nitrogen, Pd₂(dba)₃ (22.9 mg, 0.025 mmol, 0.025 equiv) and (S)-tBu-PHOX (19, 24.2 mg, 0.0625 mmol, 0.0625 equiv) were added. The flask containing the solids was evacuated for 15 min and then refilled with dry nitrogen. Dry THF (30 mL) was then added and the resulting solution stirred at 25 °C for 30 min. At this point, allyl 1-methyl-2-oxocyclohexanecarboxylate ((±)-10) was added via syringe in one portion. When the reaction was complete by TLC (7.5 h), the reaction mixture was evaporated under reduced pressure and the residue purified by column chromatography (SiO₂, 1.5→2.5% Et₂O in pentane) to afford ketone (S)-9 (129.6 mg, 85% yield, 88% ee).