Enantioselective α-Allylation of Aryl Acetic Acid Esters via C1-Ammonium Enolate Nucleophiles: Identification of a Broadly Effective Palladium Catalyst for Electron-Deficient Electrophiles

Abstract

We have identified a generally effective Pd catalyst for the highly enantioselective cooperative Lewis base/Pd-catalyzed α-allylation of aryl acetic esters using electron-deficient electrophiles. Changing between aldehyde, ketone, ester, and amide substituents at the terminus of intermediate cationic π-(allyl)Pd species affects both the efficiency of the reaction and, in the case of amides, control over the stereochemistry of the product alkene, as a function of the ligand. Tris[tri(2-thienyl)phosphino]Pd(0) serves as a broadly effective catalyst and overcomes these challenges to provide a general, high-yielding, and operationally simple C(sp³)–C(sp³) bond-forming method that gives products with high levels of enantioselectivity.

Graphical Abstract

1. INTRODUCTION

Enantioselective Pd-catalyzed allylic alkylation has emerged as one of the most robust and versatile methods for the construction of C(sp³)−C(sp³) bonds.¹ During such reactions, a range of carbon-based nucleophiles engage intermediary cationic π-(allyl)Pd(II) electrophiles and, by effectively tailoring the steric and electronic parameters of the supporting chiral nonracemic P(III) ligands on palladium,² excellent control over both reactivity and enantiofacial selectivity is possible.

Our laboratory has a long-standing interest in enantioselective allylic alkylation, and we have recently described a cooperative Lewis base/Pd dual catalysis platform that efficiently addresses the stereocontrol challenges engendered by acyclic ester pronucleophiles.³ In this regime, the Lewis base serves to control and direct intermediate enolate geometry, as well as enantiofacial selectivity, whereas the Pd catalyst is primarily responsible for regulation of electrophile reactivity (see Figure 1a and Scheme 1 (left)). This process has proven to be general, as a consequence of the disparate roles played by these two catalysts, which permits the reactivity of the Pd center to be tuned via the ancillary ligands without compromising the nature or efficiency of enantiocontrol, which is governed by the Lewis base. We have exploited this feature to overcome unexpected reactivity and stereocontrol challenges posed by differentially functionalized electrophiles.³ Herein, we further demonstrate the utility of this by identifying a broadly effective Pd(0) catalyst that allows for the enantioselective α-alkylation of aryl acetic acid esters by a wide range of carbonyl-substituted electrophiles.

Figure 1.

(a) Cooperative Lewis base/Pd-catalyzed enantioselective allylic alkylation using acyclic ester pro-nucleophiles. (b) Can a Pd catalyst be identified to overcome the varying influence of common electron-withdrawing groups on reaction yield, E/Z selectivity, and enantioselectivity?

Scheme 1.

Postulated Mechanism (Left); Challenges for Pd(0) Catalysis (Right)

The efficiency of Pd-catalyzed allylic alkylation is significantly affected by the precise electronic and steric composition of intermediate π-(allyl)Pd(II) species.¹ Although their reactivity can be fine-tuned by the supporting ligands, it is nonetheless challenging to identify single-catalyst systems that are broadly effective. For example, the presence of strongly electron-withdrawing carbonyl functional groups at the termini of the organopalladium intermediates directs attack by the incoming nucleophile to the distal carbon by virtue of orbital control,⁴ thus reinforcing the regiochemical preference in favor of linear products as is common in Pd-catalyzed allylic alkylation. However, the degree of electron-withdrawing character, dipole, and steric demand varies significantly across the carbonyl series⁵ and can drastically affect the behavior of the electrophilic intermediates. For example, isomeric syn and anti π-(allyl)Pd(II) complexes can differ greatly in their relative electrophilicities and, as such, their ratio and the rate of their interconversion via π−σ−π isomerization can affect both the product alkene stereochemistry and the enantioselectivity of bond formation (see Scheme 1 (right)).^6,7 Thus, identification of a broadly effective and general Pd/ligand system is far less trivial than it might first appear (see Figure 1b and Scheme 1 (right)).

2. RESULTS AND DISCUSSION

2.1. Initial Substituent Assessment.

Beginning with our previously developed conditions employing Birman’s benzotetramisole (BTM),^8,9 in conjunction with Buchwald’s thirdgeneration Xantphos-ligated Pd precatalyst,¹⁰ we evaluated the influence of four common carbonyl-containing functional groups on the reactivity, alkene stereoselectivity, and enantioselectivity of our cooperative catalysis process (see Scheme 2). Thus, the direct α-allylation of phenyl acetic acid pentafluorophenyl ester with allylic tosylates¹¹ 1a–d revealed stark differences in the challenges facing the Pd catalyst and control elements that are inherent to the putative cationic π-(allyl)Pd(II) intermediates. The ester-substituted electrophile 1a provided the expected product in good yield, as a single alkene isomer and with good levels of enantiocontrol.¹² However, decreasing the carbonyl oxidation level in electrophiles 1c and 1d provided lower yields of products, which also exhibited slightly lower levels of enantiocontrol, albeit as a single alkene E-isomer. Finally, most challenging was the amide-appended electrophile 1b, which also gave lower chemical yields as well as poor levels of control over alkene stereochemistry. Noteworthy is the high enantioselectivity of the major E-isomer. While these results demonstrate the flexibility of this cooperative catalysis protocol and further validate the broad capability of the Lewis base catalyst to impart useful levels of enantioselectivity, they also reveal disparate challenges facing the Pd catalyst that must be overcome in order to develop a fully general protocol for the direct asymmetric α-allylation of aryl acetic acid esters by electron-deficient electrophiles.

Scheme 2.

Influence of Commonly Encountered Electron-Withdrawing Groups on Reaction Yield, E/Z Selectivity, and Enantioselectivity

We have previously described the inherent modularity of this mechanistic construct, where the supporting ligands on Pd can be tuned to (i) overcome poor steric-derived reactivity, (ii) control chemoselectivity, and (iii) control alkene stereochemistry, in a manner that is independent of enantioselectivity, which continues to be administered by the Lewis base catalyst.³

In considering these results, and other notable contributions by the groups of Smith,¹³ Hartwig,¹⁴ and Gong,¹⁵ we expected that a Pd catalyst could be identified that would address the distinct challenges posed by each of the functional groups presented in Figure 1. As the amide-substituted electrophile 1b possessed the added complication of control over the alkene stereochemistry, we began our investigations with this subclass.

2.2. Amide-Substituted Electrophiles.

Informed by our earlier studies, we surveyed a representative range of bidentate and monodentate phosphine ligands and assessed their capacity to influence both chemical yield and alkene stereoselectivity (see Table 1). Comparison of Xantphos-ligated Pd(0) derived from Buchwald’s G3 precatalyst,¹⁰ or formed in situ by stirring with Pd₂dba₃ gave similar levels of E/ Z selectivity and enantioselectivity, although product yield using the latter method was slightly improved (Table 1, entries 1 and 2). Using in situ catalyst preparation, other bidentate phosphines with varying bite angle and backbone flexibility offered no improvement (Table 1, entries 3–5). Standard monophosphines (Table 1, entries 6–10) exhibited a range of activity and E/Z selectivity, with PCy₃ and P(o-tolyl)₃ providing no reaction (Table 1, entries 6 and 7) and P(4-OMePh)₃ affording comparable levels of reactivity and selectivity to the parent PdXantphos G3 system (see Table 1, entry 8 vs entry 1). We have previously described the enhanced reactivity engendered by P(2-furyl)₃ and P(2-thienyl)₃ ligands in related cooperative catalysis processes via C1-ammonium enolate nucleophiles.^3c,d In the case of the former, the product was obtained exclusively as the E-isomer, albeit in prohibitively low yield and with a slight reduction in enantioselectivity (Table 1, entry 9). The latter provided the product in useful 6.6:1 E/Z ratio, with excellent levels of enantioselectivity and more useful chemical yield (entry 10). A brief assessment of Pd:ligand stoichiometry (entries 11 and 12) resulted in further yield enhancement (90%) and E/Z stereoselectivity (7.9:1) without compromising the level of enantiocontrol (96:4 er). At this juncture, we considered our optimization complete and moved forward to assess the scope of the appended amide and performed a direct comparison between our previous conditions employing PdXantphos G3 and our newly identified in situ Pd₂dba₃/P(2-thienyl)₃ conditions (see Scheme 3, columns 1 and 2).

Table 1.

Optimization of 4-Me-Anilide-Substituted Electrophile: Effect of Ligand

Entrya	Pd(mol%)	PR3(mol%)	Yield[%]b	E:Zc	erd
1	PdXantphos G3	--	40	2.3:1	96.4
2	Pd2dba3(5)	Xantphos (10)	56	2.5:1	95.5
3	Pd2dba3(5)	DPEphos (10)	40	3:1	96.4
4	Pd2dba3(5)	dppf (10)	37	2.1:1	95.5
5	Pd2dba3(5)	dppe (10)	26	1.9:1	96.4
6	Pd2dba3(5)	PCy3 (20)	0	--	--
7	Pd2dba3(5)	P(o-tolyl)3 (20)	0	--	--
8	Pd2dba3(5)	P(4-OMePh)3 (20)	28	2.5:1	94.6
9	Pd2dba3(5)	P(2-furyl)3 (20)	15	100:1	92.8
10	Pd2dba3(5)	P(2-thienyl)3 (20)	43	6.6:1	96.4
11	Pd2dba3(5)	P(2-thienyl)3 (10)	85	7.2:1	94.6
12	Pd2dba3 (10)	P(2-thienyl)3 (25)	90	7.9:1	96.4

^aReactions performed on a 0.1 mmol scale.

^bYields determined by ¹H NMR by comparison with an internal standard (1,2,4,5-tetramethylbenzene).

^cE/Z ratio calculated from crude ¹H NMR.

^dDetermined by chiral HPLC analysis.

Scheme 3.

Scope and Comparison of Ligands/Conditions for Amide-Substituted Electrophiles

^aYields determined by ¹H NMR by comparison with an internal standard (1,2,4,5-tetramethylbenzene). Isolated yields in parentheses. Enantiomeric ratio determined by chiral HPLC in comparison to the racemate. P(2-Th)₃ = tri(2-thienyl)phosphine.

Comparison of secondary and tertiary anilides (2 and 3), branched secondary amide (4), acyclic secondary amide (5), and Weinreb amide (6) revealed the broad superiority of the Pd₂dba₃/P(2-thienyl)₃ protocol in terms of yield and E/Z selectivity; however, enantioselectivity was largely variable and was attributed to product epimerization during the prolonged reactions times.¹⁶ Accordingly, and because of the induction time necessary for in situ catalyst formation, we expected that a preformed Pd(0) catalyst would exhibit greater reactivity and enable a reduction in reaction time (see Supporting Information for full time study results). Employing Pd[P(2-thienyl)₃]₃ revealed its generally superior control over the alkene E/Z ratio and preservation of the high levels of enantioselectivity (see Scheme 3, column 3).¹⁷

At this juncture, it is appropriate to offer some comments concerning the alkene isomer ratio. In Pd-catalyzed allylic alkylation the E/Z selectivity originates from the relative stability (and reactivity) of syn and anti π-(allyl)Pd(II) intermediates,⁶ which, in turn, can derive from electronic and/or steric bias imparted by both the substrate and the supporting ligands. On a case -by-case basis, some combination of these influence not only the relative energies of syn and anti but also the relative facility of π–σ−π isomerization en route.¹⁸ The data presented in Table 1 and Scheme 3 indicate clear ligand dependence, with respect to the obtained alkene stereochemistry and led us to consider those factors responsible. Although neither in situ spectroscopic interrogation nor single-crystal analysis of species relevant to catalysis have offered any insight, available Xantphos ligand analogues provide a toolbox for potential qualitative assessment via modulation of the steric demand or electronic parameters of the P(III) donor atom substituents (see Scheme 4, Question).¹⁹ Consequently, formation of the parent α,β-unsaturated anilide 2 (E/Z 2.3:1) was re-evaluated using commercially available Xantphos ligand analogues 9–12 that present an indicative range of steric and electronic features. The gem-dimethyl → NH ligand backbone modification in NiXantphos 9 constitutes a remote electronic modification that does not significantly affect the donor properties of the P(III) donor atom.²⁰ Consistent with this, 9 gave the product in almost-identical alkene ratio (2.5:1 E/Z) to the parent ligand (see Scheme 3, column 1) suggesting little substantive influence over the reactivity of the associated π-(allyl)Pd(II) fragment. More direct influence can be expected by altering the substituents on the P(III) donor atom, and the corresponding bis-phosphonous diamide 10 or Reetz’s (R)-BINOL-derived diphosphonite ligand 11 only further favored the E alkene isomer relative to Xantphos itself. Given the enhanced π-accepting character of these substituents via the classical backdonation from the appropriate metal d-orbital to the lowlying σ*_P–N and σ*_P–O orbitals,²¹ respectfully, this might well indicate why 2-(thienyl)phosphine, which is a ligand with low steric demand and strong π-accepting character,²² is so generally effective. Finally, poorer acceptor properties coupled with increased steric demand via di-t-butyl substitution (12) shut down the reaction completely. While this last result might suggest inhibition of catalysis due to deleterious steric effects, the poor reactivity of PCy₃ (see Table 1, entry 6) more likely suggest that electron-rich alkyl-substituted phosphines lack the electronic properties necessary for catalysis at the Pd(0) center. While elucidation of the factors responsible for E/Z control must await further study, these data further illustrate the unexpected breadth of effect that relatively minor modifications to a given ligand scaffold can have on product selectivity.

Scheme 4.

Comparison of Xantphos Ligands, Showing the Effect of Electronic and Steric Modification on E/Z Selectivity^a

^aStructural variations from Xantphos are highlighted in blue.

2.3. Ketone-Substituted Electrophiles.

As depicted in Scheme 2, using the PdXantphos G3 catalyst,¹⁰ ketone, and aldehyde-substituted electrophiles provided products in low yield, although they did not suffer from alkene isomer distributions. As indicated by the results of our amide assessment, we reasonably expected that the chemical yield could be improved upon identification of an appropriate Pd catalyst, and that any decrease in enantioselectivity might also be addressed through the ligand selection or by limiting in situ product racemization. Assessment of the same ligands/ conditions again revealed the critical influence of the supporting phosphine ligand (see Table 2, entries 2–10), and again revealed P(2-thienyl)₃ to be superior (Table 2, entries 10–12). Finally, the preformed zerovalent Pd[P(2-thienyl₃)]₃ again proved optimal (Table 2, entry 13).

Table 2.

Optimization of t-Butylketone-Substituted Electrophile: Effect of Ligand

Entrya	Pd (mol%)	PR3 (mol%)	Yield [%]b	E:Zc	erd
1	PdXantphos G3	--	45	E only	85:15
2	Pd2dba3(5)	Xantphos (10)	41	E only	--
3	Pd2dba3(5)	DPEphos (10)	0	--	--
4	Pd2dba3(5)	dppf (10)	36	E only	-
5	Pd2dba3(5)	dppe (10)	0	--	-
6	Pd2dba3(5)	PCy3 (20)	NA	NA	NA
7	Pd2dba3(5)	P(o-tolyl)3 (20)	NA	NA	NA
8	Pd2dba3(5)	P(4-OMePh)3 (20)	0	--	-
9	Pd2dba3(5)	P(2-furyl)3 (20)	50	E only	-
10	Pd2dba3(5)	P(2-thienyl)3 (20)	70	E only	86:14
11	Pd2dba3(5)	P(2-thienyl)3(10)	70	E only	93:7
12	Pd2dba3 (10)	P(2-thienyl)3 (25)	70	E only	93:7
13	Pd[P(2-thienyl)3]3 (5)	--	87	E only	96:4

^aReactions performed on a 0.1 mmol scale.

^bYields determined by ¹H NMR by comparison with an internal standard (1,2,4,5-tetramethylbenzene).

^cE/Z ratio calculated from crude ¹H NMR.

^dDetermined by chiral HPLC analysis

Using these optimal conditions, we assessed the steric profile of aliphatic ketones (Scheme 5, 13–16), which provided the products in good isolated yields as single E-isomers and with consistently high levels of enantioenrichment. Cyclopropyl ketone 17 and phenyl ketone 18 were similarly effective.

Scheme 5.

Scope of Ketone-Substituted Electrophiles

^aIsolated yields. Enantiomeric ratio determined by chiral HPLC in comparison to the racemate. P(2-Th)₃ = tri(2-thienyl)phosphine.

2.4. Ester-Substituted and Related Electrophiles.

The identification of Pd[P(2-thienyl)₃]₃ as an effective Pd catalyst offered some prospect of a general Lewis base/Pd protocol independent of the carbonyl substituent on the electrophile. Therefore, we assessed its competence with electrophiles bearing ester (19–23), thioester (24), and nitrile (25) moieties (see Scheme 6). All products were formed in high chemical yield, with high levels of enantioenrichment and with exclusive control over the alkene stereochemistry. The nitrile product 25 displayed only minimal deviation from this (9:1 E/Z) and slightly lower levels of enantioselectivity. Noteworthy is the performance of the aldehyde-substituted electrophile, which is sensitive and must be used immediately. The product 26 was obtained in 50% yield and with enhanced levels of enantioenrichment (cf Scheme 2).

Scheme 6.

Scope of Ester-Substituted Electrophiles*

*Isolated yields. Enantiomeric ratio determined by chiral HPLC in comparison to the racemate. ^aReaction performed at 0 °C. P(2-Th)₃ = tri(2-thienyl)phosphine.

Finally, we sought to evaluate control over alkene stereochemistry in trisubstituted α,β-unsaturated esters. In the event, α-Me and β-Me substituents (27 and 28) displayed marked preference for the E-stereoisomer while retaining the levels of enantioselectivity (see Scheme 6, bottom).

2.5. Nucleophile Scope.

Our systematic investigation of carbonyl-based substituent effects on the stereochemical outcome of our direct asymmetric α-allylation of prochiral aryl acetic acid ester nucleophiles²³ has led to the identification of a general and effective cooperative Lewis base/Pd-catalyzed protocol. In concert with this broad tolerance of electrophile structure, we evaluated a range of aryl acetic acid ester nucleophiles and further confirmed the effectiveness of this catalytic system (see Scheme 7). We began by assessing the reaction of the ethyl ester-bearing electrophile with three standard nucleophiles, all of which performed as expected (29–31) and provided the products in good yields as single E-isomers and with high levels of enantioselectivity. From here, we explored the efficiency of this catalyst via different nucleophile–electrophile combinations (32–43). Again, these all performed as expected. Noteworthy is the ability to incorporate aryl halides in conjunction with different α,β-unsaturated carbonyl systems. In combination with electro-philic Pfp esters,²⁴ these moieties provide ample opportunity for further orthogonal diversification.

Scheme 7.

Demonstration of Generality–Nucleophile Scope, in Combination with Different Electrophiles*

*Isolated yields. Enantiomeric ratio determined by chiral HPLC in comparison to the racemate. ^aReaction time = 2 h. ^bReaction performed in 1,4-dioxane. P(2-Th)₃ = tri(2-thienyl)phosphine.

3. CONCLUSIONS

In summary, we have identified the compatibility and broadly effective cooperative catalytic effect of BTM/Pd[P(2-thienyl₃)]₃ for the direct enantioselective α-allylation of prochiral esters. Products are obtained exclusively as the linear regioisomer, with high levels of E-selectivity (or exclusively E) and with high levels of enantiocontrol. Within the context of our own efforts, this cements BTM/Pd as a general reactivity platform for the direct enantioselective alkylation of acyclic prochiral esters. Moreover, this study furthers the notion that Lewis base/transition-metal cooperative catalysis provides unique opportunities for enantioselective reaction design as the properties of the metal can be tuned and modulated independently of the source of enantioselectivity, which resides on the Lewis base catalyst. This can be considered complementary to traditional ligand-centered asymmetric induction.