Enantioselective Construction of Tertiary Fluoride Stereocenters by Organocatalytic Fluorocyclization

1,1-Disubstituted styrenes with internal oxygen and nitrogen nucleophiles undergo oxidative fluorocyclization reactions with in situ generated chiral iodine(III)-catalysts. The resulting fluorinated tetrahydrofurans and pyrrolidines contain a tertiary carbon–fluorine stereocenter. Application of a new 1-naphthyllactic acid-based iodine(III)-catalyst allows the control of tertiary carbon–fluorine stereocenters with up to 96% ee. Density functional theory calculations are performed to investigate the details of the mechanism and the factors governing the stereoselectivity of the reaction.


■ INTRODUCTION
Carbon−fluorine bonds frequently occur in drug substances, agrochemicals, and substances used for medical diagnostics. 1−3 The predominant structures are aryl fluorides and trifluoromethylated arenes. 4,5 In contrast, alkyl fluoride motifs are far less abundant in small molecules relevant for life sciences. Nevertheless, recent studies have highlighted the beneficial properties of alkyl fluorides for the modification of molecular conformation, polarity, acid−base properties, and electronic interactions based on gauche/anomeric effects. 6−14 There are a few examples of marketed alkyl fluoride drugs, such as macrolide antibiotics solithromycin, fluticasone, and sofosbuvir, which are used for treatment of asthma and hepatitis C (Figure 1a). 4 One of the main reasons for the relatively low abundance of fluoroalkyl motifs in pharmaceutical compounds is attributed to methodological limitations for the selective synthesis of alkyl fluorides, especially with tertiary C−F stereocenters (Figure 1a). 15 On the basis of our previous results on fluorocyclization of alkenes with fluorobenziodoxol (a hypervalent iodine reagent), 16 we decided to develop an asymmetric version of this reaction to access heterocyclic compounds with endocyclic tertiary fluoride motifs. Chiral hypervalent iodines have become attractive catalysts and mediators for asymmetric oxidation reactions. 17−20 These species have been successfully employed for a wide range of enantioselective C−C, 21−23 C−O, 24−28 and C−N 29,30 bond formation reactions using alkene derivatives as substrates. The groups of Kitamura, 31 Jacobsen, 32 and Gilmour 33 reported pioneering methods for catalytic fluorination of olefins with in situ generated iodine(III) fluorides. Chiral iodine(III) fluorides were first applied by Nevado and co-workers for the asymmetric synthesis of fluorinated compounds, such as piperidines and azepanes. 34 This seminal work has been followed by several catalytic asymmetric protocols. 35−40 As mentioned above, enantioselective construction of tertiary fluorides is considered to be particularly challenging but relevant for drug design ( Figure 1a) and for modern synthetic methodology development. 15 Although many excellent techniques were reported for the synthesis of these species, 41−51 relatively few methods are based on hypervalent iodine reagents. 32,37,38,52−54 Jacobsen and co-workers presented studies on asymmetric aziridination 38 and 1,2difluorination reactions 32,37 (Figure 1b) leading to tertiary fluorides. The groups of Rueping 52 and Lu/Zheng 53 described a method for the synthesis of chiral tertiary fluoro β-ketoesters using hypervalent iodine based catalysts. Fluorocyclization with hypervalent iodides 16,55−59 is a challenging but very efficient approach for accessing pharmaceutically important heterocycles, such as tetrahydrofurans and pyrrolidines. However, fluorocyclization for the synthesis of endocyclic chiral tertiary fluorides with hypervalent iodides (Figure 1c) has not been reported so far. In addition, the mechanistic features governing the reactivity and selectivity in these types of fluorocyclization reactions have been unexplored.

■ RESULTS AND DISCUSSION
In the initial fluorocyclization reactions we employed 1,1disubstituted styrenes with tethered hydroxy groups as substrates in the presence of C 2 -symmetric aryl iodide catalysts ( Figure 2) with HF-pyridine as fluorine source and mCPBA as oxidant (Table 1). According to previous reports, catalysts (R,R)-1a−c are particularly suitable for asymmetric fluorination reactions of alkenes ( Figure 2). 34,38,56,60 Optimization of the Reaction Conditions. We selected p-nitrostyryl alcohol 2a as a model compound for the optimization of the reaction conditions. When the reaction was carried out under the conditions given in Table 1, using lactate-derived catalyst (R,R)-1a, 38 fluorotetrahydrofuran derivative 3a was obtained with encouraging levels of ee (72%) and yield (41%) ( Table 1, entry 1). The absolute configuration of the tertiary carbon−fluorine stereocenter in 3a is (S) based on X-ray diffraction studies. When (R,R)-1a was replaced by phenyllactate catalyst (R,R)-1b, 38 both the yield (50%) and the enantioselectivity (88%) increased, probably because of the increased steric demand of the α-substituent in the so-called "side arm" of the ester group (entry 2). Possibilities to improve the stereoselectivity were further studied by increasing the size of the side arm substituent. Indeed, the use of mesityl lactate derivative (R,R)-1c 60 increased the enantioselectivity to 90% (entry 3).
This selectivity enhancement prompted us to study the subtle steric effect of the ortho substitution of the side arms. Therefore, we prepared two new catalysts, with naphthyl (R,R)-1d and phenanthryl (R,R)-1e groups in the side arms. To our delight, by use of the newly developed 1naphthyllactate catalyst (R,R)-1d, the ee could be increased to 93%, affording 3a with a high yield of 83% (entry 4). The naphthyl substituents of (R,R)-1d probably exert the ideal substituent effects in the stereoinduction step, as the ee of the reaction declined to 88% (entry 5) with the bulkier phenanthryllactate catalyst (R,R)-1e. An in-depth analysis on the substituent effects of the side-arms on the stereoinduction is given in the DFT/stereoinduction section below.
We briefly studied the solvent effects on the reaction. By use of catalyst (R,R)-1c, chloroform was replaced by methylene chloride and toluene. In both cases the enantioselectivities   decreased (cf. entry 2 with entries 6 and 7). Conducting the reaction at 0°C did not influence the yield, but the enantioselectivity was lower (entry 8). Dilution (entry 9) and decrease of the amount of HF-pyridine (entry 10) led to a slight decrease of the ee and the same or improved yield. Decreasing of the amount of catalyst from 10% to 5% (entry 11) decreased the ee to 73%. The decrease of the ee was surprising as in the absence of catalyst formation of 3a was not observed (entry 12). This indicates that there is no racemic background reaction in the absence of iodoaryl catalyst. We found (see Supporting Information for experimental studies, page S65) that the catalyst did not undergo racemization under the applied reaction conditions. This control experiment (Supporting Information, page S65) suggests that the catalyst underwent transformation/degradation under the applied oxidative conditions and perhaps these products also catalyzed the fluorocyclization, albeit with lower selectivity than (R,R)-1c,d. Unfortunately, we were unable to isolate these degradation products of the catalyst. In a very recent publication, Sigman, Jacobsen, and co-workers 54 reported the isolation of chiral iodoresorcinol derivatives, which were formed under the oxidative conditions and still catalyzed the difluorination reactions. Tetrahydrofurans by Asymmetric Fluorocyclization. Since both catalysts (R,R)-1c and (R,R)-1d induced high levels of enantioselectivities and good yields, we decided to explore the substrate scope of the reaction with both catalysts (Scheme 1). We studied the effects of the aromatic substituents on the reactivity of the styrene substrates and on the enantioselectivity of the fluorocyclization (Scheme 1).
Somewhat lower enantioselectivities (73−88% ee) were observed for the formation of nitrile-, keto-, and estersubstituted products 3d−f. Interestingly, benzamide 3g and tertiary benzamide 3h were formed with high levels of enantioselectivity (up to 92% ee). However, the yield of the primary amide 3g was only 27%/41% (with (R,R)-1d/(R,R)-1c) probably because of poor solubility of the corresponding substrate (2g) in chloroform. As mentioned above, metasubstituted products 3i−k were formed with lower selectivity. While para-nitro derivative 3a was formed with 93% ee, its meta-nitro analog 3i was obtained with 86% ee using catalyst (R,R)-1d. Similarly, meta-methylcarbonyl compound 3j was obtained with 82% ee, while the para-substituted isomer 3e was formed with 86% ee. In the case of meta-benzyl ether substitution (3k) the fluorocyclization proceeded with relatively low ee (61%) and poor yield (36%). Fluorocyclization of ortho-cyano-substituted styrene 2l with catalyst (R,R)-1d resulted in formation of 3l with 39% ee. Surprisingly, the reaction with catalyst (R,R)-1c gave the opposite enantiomer ent-3l with 33% ee. In the case of an ortho-chloro substituent, the major enantiomer formed with the same configuration using either (R,R)-1c or (R,R)-1d, albeit with low ee of 16% or 30%, respectively. While para-and meta-substituted nitro aryl Journal of the American Chemical Society pubs.acs.org/JACS Article derivatives 3a and 3i formed with high ee's and yields, formation of the ortho-substituted analog 3n was not observed. The effects of the ortho-nitro substitution on fluorination and fluorocyclization reactions have been previously reported. 39,60 As shown in Table 1, the reaction of 2a gave a higher enantioselectivity at a lower temperature (cf. Table 1, entries 3 and 8). Therefore, we carried out the reaction of some substrates at −50°C using (R,R)-1d as catalyst, to see whether the enantioselectivity could be further increased. We found that the para-substituted substrates 2a, 2c, 2e and metasubstituted substrate 2i reacted with higher enantioselectivities (94−96%) at −50°C than at −35°C (86−93%). However, the yields were lower at −50°C than at −35°C. The lower yields at −50°C have been a consequence of the lower reactivity, which may be caused by solubility issues. The reactions can be easily scaled up without significant change of the selectivity or yield. For example, under the standard optimized conditions 3a was obtained with 89% ee (0.5 mmol scale) and 90% ee (1 mmol scale) using catalysts (R,R)-1c and (R,R)-1d, respectively. We have also studied the effects of the alteration of the aliphatic tether of the alcohol nucleophile (Scheme 2a−c). When the carbon chain was elongated by a single methylene unit, the corresponding tetrahydropyran 3o was formed in lower yield (24%/39%) and selectivity (57%/65% ee) than 3a using (R,R)-1c or (R,R)-1d as catalyst (Scheme 2a). We did not observe formation of tetrahydrofuran products. The lack of five membered ring (tetrahydrofuran) products indicates that the elongation of the tether did not lead to change of the mechanism of the fluorocyclization involving a phenonium ion intermediate. 36,57,59,61−63 Possible reasons to explain the drop of the ee on elongation of the tether are given below in the DFT/stereoinduction section. We have also attempted to study the Thorpe−Ingold effect 64,65 on the reactivity of the substrates by dimethyl substitution of the tether (Scheme 2b,c). However, formation of fluorocyclization products 3p,q was not observed using para-iodotoluene as catalyst. In the case of the attempted fluorocyclization of 2p, allyl fluoride 4 formed (Scheme 2b). The mechanism of this reaction involving a formal extrusion of a CH 2 O molecule is unclear. When fluorocyclization of 2q was attempted, a complex reaction mixture of various fluorinated products was obtained (Scheme 2c). In addition, we also studied the effects of protection of the alcohol nucleophile (Scheme 2d). In the reaction of benzyl protected substrate 2r, the same fluorocyclization product 3a was observed as with the parent alcohol (2a), albeit with lower yield (38%) and selectivity (73% ee). Formation of the identical major enantiomer with 2a and 2r under the same reaction conditions indicates that the enantioselectivity is not controlled by the nucleophilicity of the oxygen, which means that the cyclization process is not involved in the enantiodetermining step of the oxyfluorination. This suggests that the enantioselectivity is determined in the fluorination process. 60,62 Extension to Pyrrolidines. The asymmetric fluorocyclization reaction of alcohol nucleophiles 2 can be extended to sulfonate-protected nitrogen nucleophiles 5 ( Table 2). We have found that the selectivity of the aminofluorination reaction is dependent on the sulfonyl protecting group. The fluorocyclization of sulfonamide analogs (6a−c) of nitro compound 3a was studied (Table 2) under the optimal conditions of the oxyfluorination (Table 1, entry 4). Formation of pyrrolidine derivatives 6a and 6c (70%/72% ee) with nosyl and mesyl protecting groups ( Table 2, entries 1 and 3) occurred with higher selectivity than with tosyl-substituted (entry 2) analog 6b (56% ee). Similar to the results obtained for oxyfluorination reactions, application of our newly developed 1-naphthyllactate catalyst (R,R)-1d gave higher selectivity than the mesityl derivative (R,R)-1c (cf. entries 1 and 4). A slight change of the applied amount of pyr•9HF reagent and dilution of the reaction mixture led to an increase of the ee from 70% to 75% (cf. entries 1 and 5). The difference

Scheme 2. Control Experiments
Journal of the American Chemical Society pubs.acs.org/JACS Article in selectivity for the nosyl-(6a) and mesyl-substituted (6c) products was relatively small. As the nosyl protection group is easier to remove than the mesyl group, we selected nosyl protection for the study of the synthetic scope of the reaction. By use of the optimal conditions (Table 2, entry 5), pyrrolidine derivatives with tertiary C−F bonds were obtained (Scheme 3) with high selectivity (up to 94% ee, 6e). The enantioselectivity of the aminocyclization reactions (Scheme 3) was slightly lower, but the yields were somewhat higher than for the oxyfluorination reactions (Scheme 2). Pyrrolidine derivatives with trifluoromethyl-(6d), cyano-(6f), and methyl ester-(6h) substitution formed with the highest selectivity (86−92% ee). Nitro-(6a), methyl sulfonyl-(6e), methyl carbonyl-(6g), and dimethylamide (6i) products were obtained with lower enantioselectivity (72−82% ee) than the corresponding tetrahydrofuran analogs (86−93% ee). The absolute configuration of 6e is (S) according to the X-ray crystallography data. The fact that the absolute configuration of tetrahydrofuran 3a and pyrrolidine 6e formed under similar reaction conditions with the same (S) absolute configuration and with similar levels of enantioselectivities indicates that the mechanism and stereoinduction in the asymmetric oxyfluorination and aminofluorination reactions are closely similar.
Similar to the oxyfluorination processes, the aminofluorination reactions of meta-substituted substrates (5j,k) proceeded with similar or lower selectivity (75%/68% ee) than the parasubstituted analogs (75%/82% ee for 6a/6g). Alkyl-substituted alkene derivative 5l also underwent fluorocyclization reaction affording 6l in 56% yield. However, the enantioselectivity (39% ee) was much lower than for the aromatic analogs indicating the important role of the aromatic substituent in the stereoselection process. As expected, some of the aminofluorination reactions proceeded with higher selectivity at −50°C than at −35°C. Using (R,R)-1d as catalyst, we also carried out the fluorocyclization reactions of para-substituted compounds 5a, 5e, 5g and meta-substituted compounds 5j and 5k at −50°C. The enantioselectivity increased in all reactions, but the yield decreased. A substantial improvement of the selectivity was observed for the formation of sulfonesubstituted fluoropyrrolidine 6e, which was obtained with excellent ee (94%) and in good yield (56%) at −50°C. Similar to the oxyfluorination reactions, several other substrates reacted with very low yield at −50°C. The nosyl group can be easily removed from the pyrrolidine ring without affecting the enantiopurity of the products (Scheme 4). For example, treatment of 6d with 2-mercaptobenzoic acid in the presence of K 2 CO 3 leads to the free amine product 7. We were not able to determine the enantiopurity of 7, and therefore it was protected with a tosyl group (8). The ee of the tosylated product 8 indicated that the deprotection of the nosyl derivative 6d did not lead to decrease of the enantiopurity.
DFT Modeling Studies. In order to gain mechanistic insights into the fluorocyclization reactions, we performed density functional theory (DFT) calculations, using nitrophenyl styrene 2a as a model substrate, with (R,R)-1c as catalyst (Table 1, entry 3). The calculations were carried out using the B3LYP-D3 functional, 66−69 and implicit solvation using the SMD 70 model with the parameters for chloroform

Scheme 4. Deprotection of Product 6d
Journal of the American Chemical Society pubs.acs.org/JACS Article was included in the geometry optimizations (see Supporting Information for computational details). Catalytic Cycle and Reaction Profile. The reaction mechanism for the fluorocyclization of 2a obtained from the calculations is shown in Scheme 5. The associated free energy profile is displayed in Figure 3, and the optimized geometries of the intermediates and transition states are given in the Supporting Information for the computational part. The catalytic cycle starts with the formation of (R,R)-1c-F 2 by oxidation and deoxyfluorination of iodoarene (R,R)-1c. This step was not considered explicitly by the calculations, but the formation of (R,R)-1c-F 2 is supported by experiments that confirm the oxidation of iodoarene by mCPBA to form the iodosylarene (ArIO), which is readily converted to the difluorinated species by reaction with HF. 35,71−73 The next step of the cycle is the loss of a fluoride from (R,R)-1c-F 2 to yield the cationic fluoroiodonium active catalytic species Int1. Similar to the mechanism proposed by Jacobsen, Xue, and Houk for the aryl iodine-catalyzed asymmetric difluorination of β-substituted styrenes, 62 we employed two molecules of HF as an activation model for the iodoarene difluoride. Coordination of the two HF molecules results in the formation of the hydrogen-bonded difluoride complex (R,R)-1c-F 2 -2HF, which is 0.6 kcal/mol lower than (R,R)-1c-F 2 . Then, abstraction of fluoride occurs through transition state TS1, which is 18.7 kcal/mol higher than (R,R)-1c-F 2 -2HF, resulting in intermediate Int1.
In line with the results on the aryl iodine-catalyzed asymmetric difluorination of β-substituted styrenes, 62 the transformation of (R,R)-1c-F 2 to Int1 is assisted by one of the carbonyl groups of the side chain of the catalyst through an I + ···O interaction that also stabilizes the cationic species Int1. Note that Int1 here is modeled as an ion-pair, consisting of a cationic catalyst species and an (HF) 2 F − counterion. Many (>20) initial geometries of this ion-pair complex were considered in order to make sure that the lowest-energy conformation is located (the same applies to Int2, Int4, TS3, and Int5 below). We have also calculated the case in which the ions (i.e., (HF) 2 F − and the cationic part) are separated. The two different approaches led to some significant changes in the energy profiles but not to the mechanistic conclusions (see comparison in the Supporting Information for computational details on page S3). In the next step, substrate 2a coordinates to Int1 to give iodonium ion intermediate Int2(S), which is calculated to be 12.2 kcal/mol lower in energy than Int1. The coordination can take place with either the Si or Re face of the double bond. In this case it was found that the coordination to Re-face was lower in energy, which will have implications on the enantioselectivity of the reaction (see "stereoinduction" section below). We made many attempts to locate the transition state for the coordination of the olefin but without success. The optimizations led always to Int2(S).
The olefin then undergoes a nucleophilic attack by the (HF) 2 F − counterion at the most substituted carbon (via TS2(S)), generating Int3, in which two new σ-bonds are formed (C−I at 2.21 Å and C−F at 1.44 Å) and the double bond of the substrate is converted into a single bond. The barrier for this step is calculated to be 7.4 kcal/mol relative to Int2(S). We also considered the possibility of intramolecular attack by the hydroxyl group at the iodine(III)-activated alkene (see Supporting Information for computational details, S5). However, this mechanistic scenario was found to be associated with a considerably higher barrier than TS2(S). These results are also in line with our control experiment (Scheme 2d) indicating that the enantioselectivity is determined in the fluorination process prior to the cyclization step. The formation of the C−I bond in Int3 weakens the I−F bond (2.15 Å in Int3 vs 2.01 Å in Int2(S)), which makes the dissociation of fluoride easier. According to the calculations, two HF molecules can abstract the fluoride to yield ion-pair intermediate Int4, which is 21.1 kcal/mol lower than the neutral iodoarene intermediate Int3. Next, an intramolecular nucleophilic attack by the oxygen atom of 2a displaces the aryliodonium moiety, which is an excellent leaving group. The step occurs via TS3, with a barrier of 11.6 kcal/mol, and the resulting Int5 that contains the protonated form of the cyclic product 3a is 15.3 kcal/mol lower than Int4 (Figure 3). The last step of the cycle is the deprotonation of the protonated cyclic compound to give the final product 3a and regenerate the iodoarene catalyst (R,R)-1c. The (HF) 2 F − counterion can achieve this step, which is calculated to be exergonic by 2.6 kcal/mol.
Factors Determining the Stereoselection. The overall energy profile obtained for the mechanism (Figure 3) shows that the coordination of substrate 2a to Int1 to form Int2(S) is an irreversible process. The coordination of the substrate can take place with either the Re or Si faces of the double bond, resulting in Int2(S) or Int2(R), respectively, indicating that this is the selectivity-determining step of the reaction. Int2(S) leads ultimately to the S-enantiomer (major enantiomer according to the experimental studies) of the product, while Int2(R) leads to the R-enantiomer. As mentioned above, it was not possible to locate the transition state for this coordination step. Instead, we optimized the geometry of Int2(R), which was found to be 1.7 kcal/mol higher in energy than Int2(S). Subsequently, the potential energy surface was calculated backward from these intermediates. Thus, we performed constrained optimizations starting from Int2(S) and Int2(R), in which the distance between the iodine center and the double bond of the substrate was increased gradually (see S6 in the Supporting Information for computational studies). This procedure shows that the energy difference between the intermediates is maintained along the reaction coordinate of the coordination, even at long distances that resemble the TS structures. The calculated energy difference is in good agreement with the experimentally observed ee of 90% in favor of the S-product (Table 1, entry 3), indicating that the factors governing the enantioselectivity can be deduced by analyzing the geometries of Int2(S) and Int2(R), as shown in Figure 4. Scrutiny of these geometries shows that the substrate in Int2(S) fits better into the chiral pocket of the catalyst as compared to the Int2(R) (Figure 4). Namely, Int2(R) is destabilized by a steric repulsion between one of the orthomethyl substituents of the aryl group of the catalyst side arms and one of the methylene carbons of substrate 2a (Figure 4b).
The ortho-substituents in the side arm of the catalyst thus play an important role in the stereoinduction, explaining why the enantioselectivity is affected by the variation of these substituents in the experiments (Figure 2 and Table 1). On the basis of this stereoselection model, we also conclude that elongation of the tether, such as in 2o, leads to more repulsive interactions in both homologs of Int2(S) and Int2(R), leading to lower selectivity in fluorocyclization of 2o (57% ee) than for 2a (90% ee). In addition, this stereochemical model suggests that the presence of ortho-substituents in the substrates (such as in 3l−n) generates major clashes with the iodoresorcinol catalysts in the intermediates corresponding to Int2(S) and Journal of the American Chemical Society pubs.acs.org/JACS Article Int2(R), which leads to a major drop (3m) or even inversion (3l) of the stereoselectivity.

■ CONCLUSIONS
In summary, we have developed a new method for the synthesis of chiral tetrahydrofurans and pyrrolidines with endocyclic tertiary C−F stereocenters. The fluorocyclization reactions of various 1,1-disubstituted styrene derivatives were catalyzed by in situ generated hypervalent iodines. The best selectivities (up to 96% ee) were achieved using a newly developed catalyst, (R,R)-1d, with 1-naphthyl substituents in the side arm. Our experimental findings suggest that the selectivity of the reaction is dependent on the effects of substituents attached to the aromatic ring in the styrene substrates. DFT modeling was conducted to rationalize the factors governing the enantioselectivity in the fluorocyclization reaction resulting the tetrahydrofuran product. We present a new stereoselection model for formation of a tertiary C*−F bond using hypervalent iodine catalyst in an internal oxyfluorination reaction. The calculations show that the major factor determining the selectivity is a steric repulsion between the side arm substituent of the chiral catalyst and one of the methylene groups in the tether of the oxygen nucleophile. Our stereoselection model also accounts for the drop of the stereoselectivity on elongation of the tether of the nucleophile, as well as the adverse effects of ortho substituents in the aromatic ring of the styrene substrates. The above study forwards the catalytic asymmetric syntheses of tertiary fluorides, which are considered to be one of the most challenging methodological problems in organic synthesis. 15 Extension of the toolbox of asymmetric catalysis to access tertiary fluorides is important to increase the molecular diversity of complex organofluorine species used in lifesciences, such as in drug development (Figure 1a). 4,15 ■ ASSOCIATED CONTENT
Experimental procedures, characterization data, and NMR spectra of compounds (PDF) Computational details and additional computational results discussed in the text, absolute energies and energy corrections, and Cartesian coordinates (PDF) X-ray crystal structure file for compound 3a (CIF) X-ray crystal structure file for compound 6e (CIF)

■ AUTHOR INFORMATION
Corresponding Authors