Ligand-Dependent Regiodivergent Enantioselective Allylic Alkylations of α-Triuoromethylated Ketones

The asymmetric introduction of CF 3 unit is a powerful tool for modifying pharmacokinetic properties and slowing metabolic degradation in medicinal chemistry. A catalytic and enantioselective addition of α-CF 3 enolates allows for expeditious access to functionalized chiral building blocks with CF 3 -containing stereogenicity. The computational studies reveal that the choice of ligand in a designed palladium-complex system regulates the regioselectivity and stereoselectivity of the asymmetric allylic alkyation (AAA) of α-CF 3 ketones and Morita–Baylis–Hillman (MBH) adducts. Multiple C-H···F interactions are involved in the transition states.


Introduction
The wide application of uorinated compounds in agrochemicals, pharmaceuticals and materials science has triggered every endeavor to develop e cient methods for selective incorporation of a tri uoromethyl group into organic molecules. [1][2][3][4][5][6][7][8][9][10][11][12] The reliable methodology to access CF 3 -containing stereogenicity is still underdeveloped in the context of matured asymmetric synthesis. 13 The electrophilic tri uoromethylation of ketones has shown low reactivity and enantioselectivity, even with the aid of chiral auxiliary (Fig. 1a). 14 Meanwhile, the exploitation of α-CF 3 enolate as an active nucleophile for enantioselective C − C bond-forming reactions is a viable strategy for pursuing this end, allowing rapid access to densely functionalized chiral building blocks. 15,16 Despite the signi cant advances in enolatebased chemistry over the past decades, [17][18][19] α-CF 3 enolates have only been scarcely explored because of the M-F elimination of their metal enolates. 20 Electrophilic alkylation of prefunctionalized α-CF 3 ketones provided remarkable outcomes by the use of chiral auxiliaries or directing groups (Fig. 1b). [21][22][23][24] In contrast, the direct asymmetric alkylation of naked α-CF 3 ketones represents unmet challenges in terms of reactivity and selectivity that has not been addressed. 25,26 Despite the broad application of Morita-Baylis-Hillman (MBH) adducts as functionalized allylic synthons, good regio-and enantiocontrol of metal-catalyzed C1-selective adducts has not yet been realized. [27][28][29][30][31][32] The compherehsive studies on palladium-catalyzed AAA reactions by Trost and co-workers revealed that the regioselectivity can be modulated by both strics and electronics of the ligand. [33][34][35][36] The nucleophilic attack of α-CF 3 enolate from Re/Si faces to either terminal of the π-allyl-metal complexes would result in a number of stereoisomers. To overcome the above issues, we have designed a regiodivergent enantioselective allylic allylation of auxiliary-free α-CF 3 ketones with MBH adducts. By only switching the chiral ligands of the palladium complexes, excellent regio-and stereocontrol can be achieved on both C1 and C3 adducts for the construction of CF 3 -bearing quaternary centers (Fig. 1c).
Further optimization with L8 showed that using dichloromethane as the solvent did not improve the reaction yield (entry 9). With toluene, the yield increased to 92% (entry 10). By adding 2 equivalents of Et 3 N, both reaction yields and selectivities were simultaneously (entry 11). The best results were obtained with DIPEA (entry 12, 96%, 95:5 for 3a, >20:1 d.r., 99% ee). Thus, the optimized conditions were selected for further investigation of the C1-selective AAA reaction.
Substrate scope of C3-alkylation. The reaction scope of asymmetric allylic alkylation is further extended to a range of MBH esters with L5 to generate C3-selective adducts (Scheme 2). MBH esters 2 with -OAc leaving group and phenyls bearing both electron-withdrawing and electron-donating groups could readily furnish the CF 3 -ketones in good yields with high d.r. and ee's (4a-4j). Napthalene-derived MBH esters were also tolerated (4k). Using CF 3 -substituted tetralones, the corresponding adducts were also formed with equally high distereoselectivity (4l-4o).
Theoretical calculation study. To gain insight into the regioselectivity, DFT calculation was carried out with M06L/6-311++G(2d,p)-SDD-SMD(THF)//B3LYP/6-31G(d)-LANL2DZ-SMD (THF). For the Pd/L5 system, 39 calculation suggests an outer-sphere S N 2 type attack is 3.6 kcal/mol lower than that for C1 attack, consistent with the experimental regioselectivity (Fig. 2a). Interestingly, the calculated ΔΔG ‡ value of the nucleophilic addition TS parallels the calculated ΔΔG ‡ of their corresponding π-allyl-Pd precursor (Fig. 2b) with 2.4 kcal/mol energy difference. Thus, the relative stability of the π-allyl-Pd complexes preserved in thesubsequent nucleophilic addition TSs and thus dictates the regioselectivity of the Pd/L5 system. A closer look at Pd-allyl complexes reveals longer C-Pd distances in C1-attack precursor, indicating a looser Pd-allyl binding. This is likely the result of the trans-in uence of phosphine on the PHOX ligand as well as the delocalization of positive charges on C1 by the conjugated phenyl group. As shown in Fig. 2c, the back-donation interaction involving d orbital of Pd-center and n-π orbital of allyl moiety favors C3-attack precursor (-5.90 eV vs -5.79 eV on HOMOs).In order to disclose the regio-and stereo-effects of the CF 3 group for the AAA reaction, further calculations on the tri uoromethyl and the methyl ester analogue have been performed. The free energy difference between C1 and C3 attack for the -CO 2 Me substrate is only 1.3 kcal, much smaller than that of CF 3 substrate (3.6 kcal). Thus, poor regioselectivity is expected. The bond lengths between Pd center and allyl group in each transition state for CF 3 and CO 2 Me substituted ketones remain no change. For the three centers involved in S N 2 type reactions, there are obvious differences between the breaking Pd-C bond length and the forming C-C bond length. This is because tri uoromethyl group is close to a spherical structure compared with the planar structure of methyl ester. Therefore, the repulsion of allylic spherical hindrance and repulsion is greater than that of CO 2 Me group, which makes the transition state C-C bond longer than that of CO 2 Me in the reaction intermediate. Meanwhile, in the transition state of C3 attack, it was found that two substituents have obvious differences in the weak interaction between allylic group and ligands. For spherical tri uoromethyl moiety, multiple C-H···F interactions can be observed. This weak interaction can stabilize the transition state of C3 attack and reduce the energy of transition state, which enhances the reaction regioselectivity. While in the un uorinated system, this effect was not identi ed (Fig. 3).
The plausible reaction pathways that based on previous reports [40][41][42] and computational studies are illustrated in Fig. 4. The AAA process was initiated by the coordination of Pd-L* complex to the MBH ester followed by oxidative addition to generate Pd-πallyl species. Subsequent nucleophilic addition of α-CF 3 enolate to Pd-π allyl species at C1 or C3-position afforded tri uoromethylated ketones depending on the ligand-regulated process. The nal decomlexation releases the corresponding product 3/4 and regenerate palladium catalysts. The key selectivity deviation is originated from each catalytic pathway using bidentate or monodentate ligand. For the bidentate Phoxphos L5, complextion of A with MBH ester and oxidative addition generated Pd-π allyl specie B. Because B is stable enough and ligand exchange with L5 is not likely to occur. Hence, nucleophilic addition of CF 3 -enolate to B from outer-sphere affords C3selective intermediate D. For the monodentate SIPHOS ligand L8, only one phosphoramidite ligand can coordinate to the metal center of the allylpalladium complexes. [43][44][45] Thus, similar oxidative addition process occurs rst. The following decarboxylation of the Boc group releases t BuOand the Pd-t BuOcomplex F is obtained. 46 Here, an equilibrium of ligand exchange between the CF 3 -enolate and t BuOcontrols the regioselectivity of the product. Con guration G with less steric hindrance against the Ar group of MBH ester is more favorable than H, which explains the C1 selectivity for SIPHOS L8.

Discussion
In summary, we have developed a highly-tunable ligand-regulated regiodivergent asymmetric allylic alkylation of uorinated ketones with MBH adducts. The choices of ligand in the palladium catalytic systems turn out to be critical for both reactivity and selectivity for the construction of the CF 3 -containing quaternary stereocenters. This protocol could access to a variety of uorine-bearing motifs with high e ciency and selectivity.

Methods
General procedure for for C1-selective asymmetric allylic alkylation. (Conditions A) To a reaction tube with magnetic stirring bar were added the chiral ligand L8 (0.06 mmol), Pd 2 (dba) 3 (0.0025 mmol) and Toluene (0.4 mL) under argon. The mixture was stirred for 30 minutes at room temperature. Then the mixture was sequentially added tri uoromethylated ketones (0.05 mmol) and MBH adducts (0.06 mmol) and DIPEA (0.1 mmol). The resulting mixture was stirred for 12 h at room temperature. Then the solvent was removed in vacuo. The ratio of C1-selective/C3-selective product and the distereomeric ratios of the C1-selective product were determined by 19 F NMR analysis of the crude mixture. The crude product was puri ed by ash chromatography on silica gel with ethyl acetate/petroleum ether (1/10) and then dichloromethane/petroleum ether (1:1) as the eluent to afford the corresponding product. The ee value of the C1-selective product was determined by HPLC analysis using a Chiralcel IG-3 column.
General procedure for for C3-selective asymmetric allylic alkylation. (Conditions B) To a 10-mL schlenk tube equipped with magnetic stirring bar were added the chiral ligand L5 (0.012 mmol), Pd 2 (dba) 3 (0.005 mmol) and THF (0.4 mL) under argon. The mixture was stirred for 20 minutes at room temperature and another 20 minutes at -30 ℃. Then the mixture was sequentially added tri uoromethylated ketones (0.05 mmol), Na 3 PO 4 (0.15 mmol) and MBH adducts (0.06 mmol in 0.6 mL THF). The resulting mixture was stirred for 48 h at -30 ℃. Then the reaction was quenched by addition of saturated ammonium chloride solution (2 mL) and water (10 mL), and the mixture was extracted with ethyl acetate (2 × 10 mL). The combined organic phases were dried over Na 2 SO 4 , ltered, and the solvent was removed in vacuo. The ratio of C3-selective/C1-selective product and the E/Z value of C3-selective product were determined by 19 F NMR analysis of the crude mixture. The crude product was puri ed by ash chromatography on silica gel with ethyl acetate/petroleum ether (1/10) as the eluent to afford the corresponding compound. The ee value of C3-selective product was determined by HPLC analysis using a Chiralcel IG-3 column.

Data availability
The authors declare that the main data supporting the ndings of this study, including experimental procedures and compound characterization, are available within the article and its Supplementary Information les. X-ray structural data of compound 3i and 4n are available free of charge from the Cambridge Crystallographic Data Center under the deposition number CCDC 1911158 and 2009810. These data can be obtained free of charge from The Cambridge Crystallographic Data Center via www.ccdc.cam.ac.uk/data_request/cif. Figure 1 Overview of α-CF3 enolate chemistry.   Weak interactions in C3 attack transition states of CF3 and COOMe substituents. Bond lengths are denoted in Å.