Asymmetric nucleophilic fluorination under hydrogen bonding phase-transfer catalysis

See allHide authors and affiliations

Science  11 May 2018:
Vol. 360, Issue 6389, pp. 638-642
DOI: 10.1126/science.aar7941

H-bond to deliver fluoride

Simple fluoride salts are theoretically convenient reagents for carbon-fluorine bond formation. In practice, they are often insoluble in the solvents that dissolve their reaction partners. Pupo et al. developed urea-based catalysts that make fluoride soluble through hydrogen bonding. Moreover, their chiral substituents bias the reaction toward one of two mirror image products of C–F bond formation. This strategy should be applicable to the asymmetric addition of other salts, too.

Science, this issue p. 638


Common anionic nucleophiles such as those derived from inorganic salts have not been used for enantioselective catalysis because of their insolubility. Here, we report that merging hydrogen bonding and phase-transfer catalysis provides an effective mode of activation for nucleophiles that are insoluble in organic solvents. This catalytic manifold relies on hydrogen bonding complexation to render nucleophiles soluble and reactive, while simultaneously inducing asymmetry in the ensuing transformation. We demonstrate the concept using a chiral bis-urea catalyst to form a tridentate hydrogen bonding complex with fluoride from its cesium salt, thereby enabling highly efficient enantioselective ring opening of episulfonium ion. This fluorination method is synthetically valuable considering the scarcity of alternative protocols and points the way to wider application of the catalytic approach with diverse anionic nucleophiles.

Phase-transfer catalysis (PTC) (1) has progressed enormously with the appearance of powerful asymmetric methods based on lipophilic chiral cationic (2) or anionic salts (3) as catalysts. Despite these advances, an outstanding challenge is asymmetric synthesis where the solid-phase reagent is a simple inorganic salt. Here, we propose a solution, based on anion recognition by hydrogen bonding, to enable asymmetric PTC in organic media with poorly soluble inorganic nucleophiles. For proof of concept, we focused on the activation of alkali-metal fluorides for the enantioselective installation of sp3 CF, a highly important functional group for applications in pharmaceutical sciences (46).

Catalytic enantioselective fluorination is dominated by methods employing electrophilic fluorinating reagents (7). Although alkali-metal fluorides are abundant and inexpensive, their low solubility and high Brønsted basicity have hindered their application to asymmetric nucleophilic fluorination (812). Both cationic (13) and anionic (3) PTC have been successfully applied to asymmetric electrophilic fluorinations (Fig. 1A). Phase-transfer agents enabling nucleophilic rather than electrophilic fluorination have been extensively investigated, and the most common approach to enhance metal fluorides’ reactivity in organic solvents entails metal encapsulation with a crown ether (14, 15). Despite these advances, enantioselective fluorination with metal fluorides under PTC remains an unsolved problem. This state of play prompted us to formulate a strategy that embraces the poor solubility of metal fluorides in organic solvents and the capacity of fluoride to engage in H-bonding interactions. Fluoride abstraction, and more generally anion abstraction, with a hydrogen bond donor (HBD) catalyst (1621) has been explored to activate electrophiles via chiral ion pairs that react with an external nucleophile (Fig. 1B). Here, we introduce an alternative scenario in which the H-bonded fluoride complex itself is the nucleophile for fluorination. We envisioned that a chiral HBD could act as a solid-liquid phase-transfer catalyst enabling enantioselective nucleophilic fluorination with a metal fluoride insoluble in organic media. The in situ–formed H-bonded fluoride complex would become soluble and capable of mediating fluorination of an organic substrate with release of the HBD catalyst (Fig. 1C). Complexation of fluoride with HBDs is well documented but has not been explored to access enantioenriched alkyl fluorides (2224). We reported that urea-fluoride complexes are suitable reagents for the nucleophilic substitution of alkyl bromides, the reactive species being a 1:1 urea-fluoride complex (25). We therefore selected ureas as catalysts capable of bringing insoluble alkali metal fluorides (e.g., KF or CsF) into nonpolar solution. For substrate choice, we sought inspiration from the fluorinase enzyme. The only known fluorination reaction in nature involves S-adenosyl-l-methionine with a sulfonium leaving group and proceeds via an SN2 pathway, leading to a primary alkyl fluoride product (Fig. 1D). We focused on β-bromosulfides as model substrates because (i) they form highly electrophilic episulfonium ions that readily undergo diastereospecific bond-forming reactions with nucleophiles (21); (ii) the positively charged sulfur of the episulfonium ion can interact with the in situ–formed urea-fluoride complex in a manner reminiscent of the HB-fluoride-sulfonium prereaction complex characteristic of the enzyme (2629); (iii) meso-episulfonium ions are suitable for enantioselective desymmetrization with a chiral urea catalyst; and (iv) the fluorination products contain both fluorine and sulfur, which are important elements in drug design, thus underlining the value of the products for medicinal chemistry (46, 30).

Fig. 1 Catalytic reaction design.

(A) Enantioselective electrophilic fluorination under PTC. (B) Electrophile activation under hydrogen bonding catalysis. (C) Proposed enantioselective nucleophilic fluorination under hydrogen bonding PTC; urea-catalyzed fluorination of an episulfonium ion precursor with an alkali metal fluoride (M+F-). (D) Hydrogen-bonded fluoride complex for nucleophilic fluorination of S-adenosyl-l-methionine catalyzed by the fluorinase enzyme.

Racemic β-bromosulfides were conveniently prepared from the corresponding cis-epoxides via a two-step sequence involving epoxide ring opening followed by bromination, or via a one-pot protocol from the corresponding cis-alkene (figs. S5 to S7). Preliminary experiments (fig. S1) revealed that 1a performed particularly well under the designed catalytic conditions (Fig. 2A). No fluorination occurred at room temperature when 1a was treated with 1.2 equivalents of CsF in toluene, acetonitrile, or dichloromethane (0.25 M); in contrast, C–F bond formation took place in the presence of 10 mol % of urea 2a (31), affording the desired product 3a in yields up to 80% in CH2Cl2. Retention of configuration occurred, an observation consistent with fluoride attack on an in situ–formed episulfonium ion. The reaction also proceeded with KF under urea catalysis, but this reagent required longer time (table S1). This solid-liquid phase-transfer manifold has advantages over the use of soluble fluoride sources such as AgF or n-Bu4NF·3H2O, not least because it generates reactive fluoride at controllable low concentration and offers the possibility to induce asymmetry with a chiral nonracemic urea catalyst. Excision of the catalyst’s HBD groups, or reduction of acidity through substituent effects, led to no reaction or a decrease in reactivity (table S1). No product was formed with the corresponding thiourea (fig. S3), prompting the use of urea catalysts for subsequent studies. These data gave insight into the parameters enabling catalysis and suggested that urea–fluoride HB is a key interaction to enable C–F bond formation.

Fig. 2 Reaction development.

(A) Urea-catalyzed fluorination of rac-1a. (B) Computed Gibbs energy profile (ωB97X-D3/(ma)-def2-TZVPP/COSMO(CH2Cl2)//M06-2X/def2-SVP(TZVPPD)/ CPCM(CH2Cl2) at 298.15 K in kJ/mol) for urea-catalyzed (2a) fluorination of rac-1k, compared with the uncatalyzed reaction. (C) Enantioselective nucleophilic fluorination of rac-1a catalyzed by (S)-4a–h; rt, room temperature.

Computational analysis of the reaction catalyzed by urea 2a and the corresponding uncatalyzed pathway was undertaken to gain more insight (Fig. 2B). The thermodynamics of phase transfer were evaluated by combining experimental CsX formation free energies with density functional theory (DFT)–computed data in an anion-exchange process (fig. S9). The reaction is initiated by ionization of rac-1k to form the tight episulfonium-bromide ion pair ii, with a barrier to auto-ionization of 90 kJ/mol (i). Following ion-pair dissociation, uncatalyzed halide anion exchange (iii and iv) of free bromide for free fluoride in solution is unfavorable by 34 kJ/mol, the much higher lattice energy of CsF relative to CsBr playing a crucial role. Unfavorable ion transport is therefore responsible for the prohibitively high energetic span of the uncatalyzed pathway (122 kJ/mol). The key catalytic role of the urea is in promoting anion-exchange through preferential stabilization of fluoride in solution. With urea 2a, anion exchange becomes favorable by 17 kJ/mol, owing to the stronger H-bonding of F over Br to the catalyst. The lower reactivity of fluoride sources such as KF arises from larger differences in lattice energies between the metal fluoride and bromide salts, which disfavors ion transport (table S9). Fluoride delivery [via transition structure (TS) vi] to form the product is irreversible, even in the presence of catalyst, with a barrier to the reverse reaction of 135 kJ/mol, indicating that the product is not susceptible to racemization and that this step is enantio-determining with a chiral urea catalyst. In the fluoride delivery TS, the phenyl ring on the α-carbon stabilizes the TS by aligning its π system with the forming/breaking bonds (table S10).

Activation of otherwise insoluble fluoride by a urea ligand provides a platform for asymmetric catalysis (Fig. 2C). We considered axially chiral C2-symmetric ureas derived from BINAM ([1,1'-binaphthalene]-2,2'-diamine) (32) because these systems can be readily tuned through structural modification. After initial optimization, 1a underwent fluorination with (S)-4a (10 mol %) (32) in 1,2-difluorobenzene at room temperature, affording (+)-3a after 1.5 hours in >95% yield and an enantiomeric ratio (e.r.) of 86:14 (table S2). Catalysts (S)-4b–e modified at the binaphthyl core or presenting a phenyl instead of the 3,5-(trifluoromethyl)phenyl group were less effective (fig. S2). The presence of two urea motifs within catalyst 4a prompted us to interrogate which H-bond interactions are necessary for reactivity and enantioselectivity, experimentally and computationally.

Molecular dynamics (MD) simulations of the solution-phase conformation of 4a binding cesium fluoride indicated that isomerism of one urea proximal to the binaphthyl core from anti-anti to anti-syn was likely, and DFT calculations reinforced the energetic preference for the resulting tridentate H-bonding mode (tables S14 and S15). Experimentally, the fluorination of rac-1a was carried out with the corresponding mono- and dimethylated catalysts to probe the requirements for effective H-bonding (fig. S2). These catalysts were less effective or ineffective with the exception of 4f, methylated at the N-H that is predicted computationally not to interact with fluoride. This catalyst is as active as 4a and enhances enantiocontrol (>95% yield, 88:12 e.r.). The replacement of the N-methyl group with a larger alkyl group (4g–h), and further optimization (−30°C), afforded (+)-3a in 90% yield (95.5:4.5 e.r.).

On the basis of these data, we selected catalyst (S)-4h to study the scope of this process (Fig. 3A). Variation of the substituents on the carbon backbone of the electrophile revealed that aryl groups with meta- and para-positioned electron-donating and electron-withdrawing functionalities are compatible, affording the desired products (S,S)-3a-l in high yields and enantioselectivities. The sulfur substituent can also be modified with phenylethyl, affording the products with the highest enantioselectivity. By this route, a gram quantity of 1d underwent fluorination affording (S,S)-3d in 51% overall yield as a single enantiomer (>99.9:0.1 e.r.) after one recrystallization. Despite recent interest in the preparation and properties of molecules containing the FCCS(O)n (n = 0, 1 and 2) motif (30, 33), no alternative method to access these scaffolds in enantiopure form has been developed to date.

Fig. 3 Scope and mechanistic insights.

(A) Substrate scope. Absolute configuration was assigned by x-ray diffraction analysis of 3d. (B) Single-crystal structure of TBAF·4h overlaid with the DFT-predicted solvated structure of CsF·4h (excluding cations). (C) Competing major and minor transition states (DFT) leading to either enantiomer.

The linear relationship between the enantiopurity of catalyst and product indicates that one chiral urea is involved in the enantio-determining step (fig. S4). After extensive MD simulation of the reactive ion pair, more than 30 DFT-calculated TSs were optimized for catalysts (S)-4f and (S)-4h with substrate 1k (figs. S24 and S27). A Boltzmann ensemble of competing (S)-4h TSs predicted (S,S)-product formation (supported by single-crystal x-ray diffraction) (fig. S33) in 96.5:3.5 e.r. (at 243.15 K), a result that aligns with the experimental value (91:9 e.r.). The alkylated urea adopts an anti-syn conformation in both the fluoride complex and the populated TSs, which were subsequently found to overlay well with the x-ray structure of tetrabutylammonium fluoride·4h (TBAF·4h) (fig. S32), in terms of catalyst conformation, fluoride binding mode, and position of the cation (Fig. 3B) (fig. S28). In the lowest-energy TSs (Fig. 3C), the catalyst interacts favorably with the substrate through cation-π and CH-π noncovalent interactions (fig. S30). Catalyst:substrate noncovalent interactions are similar in competing TSs; however, shorter distances in the major TS are consistent with preferential binding. Substrate conformation also contributes to the sense of selectivity. Phenyl ring rotation adjacent to the site of nucleophilic substitution is unfavorable, owing to loss of conjugation, and is more pronounced in the less favorable (i.e., minor) pathway contributing half of ΔΔG (fig. S29). Only in the major TS can the substrate dock into the catalyst, with a full complement of noncovalent interactions, while maintaining phenyl conjugation.

Because of the early TS position along the intrinsic reaction coordinate (IRC) pathway, all three H–F interactions remain bonded in the TS (<1.9 Å), but lengthen over the IRC pathway as a result of charge neutralization. This effect is strongest for H-bond 3, which lengthens at 1.8 to 3.5 times the rate of the other H-bonds, resulting in bidentate binding. In forming the major enantiomer, lengthening of H-bond 2 also occurs along the IRC pathway once the C–F bond is fully formed, resulting in one dominant H-bond with the product (fig. S31). The evolution of H-bonds closely mimics the fluorinase enzymatic mechanism, with the bidentate urea (H-bonds 1 and 2) mimicking Ser158 and the alkylated urea (H-bond 3) mimicking the role of Thr80 (28).

We have introduced hydrogen bonding phase-transfer catalysis (HB PTC) and have applied this approach to asymmetric nucleophilic fluorination with a metal fluoride. The protocol employs a safe fluoride source and a readily accessible urea catalyst, avoiding transition metals and the need to exclude air and moisture. Beyond fluorination, we anticipate that many inexpensive nucleophiles insoluble in organic solvents can be productively applied to enantioselective catalysis with this approach. More generally, this research opens new opportunities in the design of chiral catalysts for enantioselective catalysis.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S33

Tables S1 to S31

References (34103)

Data File S1

References and Notes

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 97.
  97. 98.
  98. 99.
  99. 100.
  100. 101.
  101. 102.
  102. 103.
Acknowledgments: We thank R. Surgenor for selected experiments. Funding: This work was supported by the EU Horizon 2020 Research and Innovation Programme (Marie Skłodowska-Curie agreements 675071 and 316882), EPSRC (EP/R010064, SBM-CDT EP/L015838/1), and Lilly Research Award Program. We acknowledge the University of Oxford Advanced Research Computing facility ( and the Extreme Science and Engineering Discovery Environment (XSEDE) through allocation TG-CHE180006. Author contributions: G.P. and F.I. performed the experimental work in collaboration with A.C.V. Preliminary studies were performed by P.R. and L.P. The computational studies were performed by D.M.H.A. and R.S.P., and K.E.C. acquired the x-ray data. All authors contributed to the design of the experimental and computational work and to data analysis, discussed the results, and commented on the manuscript. R.S.P. and V.G. wrote the manuscript. V.G. conceived and supervised the project. Competing interests: All authors declare no conflicting interests. Data and materials availability: Crystallographic data are available free of charge from the Cambridge Crystallographic Data Centre under references CCDC 1812187 and CCDC 1812188. Additional optimization and mechanistic data are provided in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article