Portal de Periódicos da CAPES

Room-Temperature Nucleophilic Aromatic Fluorination: Experimental and Theoretical Studies

2006; Wiley; Volume: 45; Issue: 17 Linguagem: Inglês

10.1002/anie.200504555

1521-3773

Haoran Sun, Stephen G. DiMagno,

Click Chemistry and Applications

Taming the reagent: The use of anhydrous tetrabutylammonium fluoride (TBAFanh) in nucleophilic aromatic substitution reactions, including variants of the selective halogen-exchange and fluorodenitration processes (see scheme), was investigated. It was shown that TBAFanh permits these reactions to be performed under surprisingly mild conditions if it is used in relatively nonpolar media. Despite their rarity in nature,1 organofluorine compounds play a leading role in the life sciences.2–11 The high electronegativity and small atomic radius of the fluorine atom mean that its incorporation into biologically active compounds induces a relatively small structural perturbation, whereas drastically altering the electronic properties and bioavailability of the compounds.9, 12 Fluorine substitution is a particularly effective strategy for aromatic groups, as the electron-withdrawing nature of the fluorine atom reduces the susceptibility of an aryl ring to cytochrome P-450 catalyzed hydroxylation, a common step in drug catabolism.8 The CF dipole in fluoroaromatic groups can also lead to an increase in molecular recognition through favorable, though generally weak, electrostatic interactions.3, 6, 10, 11, 13 Because of the medicinal importance of fluorinated aromatic compounds, mild and selective methods for their preparation are desirable. We recently reported the preparation of a soluble, highly nucleophilic fluoride-ion source, anhydrous tetrabutylammonium fluoride (TBAFanh).14 TBAFanh can be dissolved in dimethyl sulfoxide (DMSO) and CH3CN (2 M) and fluorinates primary alkyl halides and tosylates rapidly at low temperature (−35 °C, 5 min, THF). The rapid rate observed for SN2 reactions that feature TBAFanh prompted us to investigate nucleophilic aromatic substitution (SNAr) reactions with this reagent. Below we describe variants of the industrially important selective halogen exchange (Halex) and fluorodenitration processes for aromatic fluorination.15–18 We show that TBAFanh permits these reactions to be performed under surprisingly mild conditions. Electron-withdrawing substituents are typically required to activate chloroaromatic substrates toward Halex chemistry. The usual reagents employed to effect halogen exchange are spray-dried KF; a high-boiling-point polar aprotic solvent, such as sulfolane; and a phase-transfer catalyst to improve the solubility of the fluoride ion (Scheme 1). Prolonged and vigorous heating is often required. Such harsh conditions preclude syntheses in which the late introduction of fluorine substituents is desired. In contrast, many aromatic compounds undergo Halex reactions at room temperature in DMSO, CH3CN, or THF upon exposure to TBAFanh. In a typical preparation, TBAFanh (1.3 equiv) and the chloro- or nitroaromatic compound were simply stirred together in DMSO.19 Representative fluorinations are shown in Table 1. A typical Halex reaction.24 DMI=1,3-dimethyl-2-imidazolidinone. Entries Substrate Conditions[a] Product Yield [%][b] 1 TBAFanh (4 equiv) 14 days no reaction 0 2 TBAFanh (4 equiv) 14 days 80 3 TBAFanh (2.5 equiv) 1 h >95 4 TBAFanh (2.5 equiv) 1.5 h >95 5 TBAFanh (1.3 equiv) 20 min >95 6 TBAFanh (1.3 equiv) 30 min >95 7 TBAFanh (1.3 equiv) 1 h >95 8 TBAFanh (2.5 equiv) 30 min >95 9 TBAFanh (1.3 equiv) 30 min >95 10 TBAFanh (1.4 equiv) ≈30 min >90 11 TBAFanh (1.5 equiv) 2 days 80 12 TBAFanh (1.5 equiv) 5 days 2 13 TBAFanh (1.5 equiv) 18 h >95 14 TBAFanh (2.5 equiv) 20 min >95 15 TBAFanh (1.4 equiv) 20 min >85 16 TBAFanh (2.5 equiv) 1 h >90 17 TBAFanh (2.5 equiv) 1.5 h >90 18 TBAFanh (1.3 equiv) 30 min >95 19 TBAFanh (1.3 equiv) 30 min >95 20 TBAFanh (1.3 equiv) 10 min >95 21 TBAFanh (1.3 equiv) 24 h >95 The scope of the room-temperature reactivity of TBAFanh with chloropyridines in DMSO is apparent from inspection of Table 1. If the pyridine ring has no electron-withdrawing groups, the nucleophilic substitution of chloride is sluggish for the ortho position and does not occur at all with m-chloropyridine. Smooth nucleophilic substitution occurs if mildly activating groups are present. A comparison of the mono- and dichloropyridine examples shows that a second electronegative substituent accelerates substitution: fluorination of 2-chloropyridine requires 14 days, whereas 2,6-dichloropyridine is exhaustively fluorinated in 90 minutes under identical conditions. In comparison, typical Halex fluorinations of 2,6-dichloropyridine require heating at 150 °C for ten hours.20–23 Chloropyridine substrates bearing more strongly electron-withdrawing (trifluoromethyl or carbonyl) substituents are also fluorinated rapidly, as are several other heterocyclic derivatives. It is noteworthy that N-benzyl-6-chloropurine is fluorinated quantitatively within 30 minutes. However, 6-chloropurine is deprotonated by TBAFanh if the amine is left unprotected, thus generating an equivalent of bifluoride ion. Fluorination of the 6-chloropurine anion requires excess TBAFanh (4 equiv) and an extended reaction time (14 days, 65 % yield). Heterocyclic substrates are much more reactive than simple arenes in Halex reactions. SNAr reactions were attempted upon the three isomeric chlorobenzonitriles to gauge the reactivity of anhydrous TBAFanh toward substituted benzenes. Suzuki and Kimura synthesized fluorobenzonitriles from chlorobenzonitriles in a pressure reactor using spray-dried KF and tetraphenylphosphonium bromide in hot (290 °C) DMI (Scheme 1).24 TBAFanh fluorinated 2- and 4-chlorobenzonitrile, albeit sluggishly, at room temperature in DMSO. The meta isomer was recovered under these conditions (Table 1). As in the case of pyridine, addition of a second chlorine atom accelerates substitution; 3,4-dichlorobenzonitrile and 2,3-dichlorobenzonitrile yield 3-chloro-4-fluorobenzonitrile and 2-fluoro-3-chlorobenzonitrile, respectively, within 20 minutes. Adams and Clark reviewed aromatic fluorodenitration and reported that is superior to chloride as a leaving group in SNAr;15, 18 thus, we assayed nitroaromatic compounds to expand the scope of room-temperature fluorination with TBAFanh. The three isomeric mononitrobenzonitriles are fluorodenitrated readily with TBAFanh; the nitrite leaving group even facilitates m-fluorination. TBAFanh performs fluorodenitration even on relatively weakly activated arenes; for example, ethyl 4-nitrobenzoate is fluorodenitrated within 30 minutes. Additional examples of fluorodenitration are summarized in Table 2. There are no reported examples of comparable, high-yielding fluorodenitration reactions performed under similarly mild conditions. Entries 6, 9, 10, and 17–19 (Table 2) show that various functional groups, including aryl esters, ethers, aldehydes, ketones, and N-benzyl protecting groups, are compatible with these fluorination conditions. Entries Substrate Conditions[a] Product Yield [%][b] σ− ΔGo [kcal mol−1] (solution) ΔG≠ [kcal mol−1] (solution) 1 TBAFanh (1.3 equiv) 95 1.27 −9.03 24.5 2 TBAFanh (1.3 equiv) 10 min >95[c] 1.00 −8.00 29.8 3 TBAFanh (1.3 equiv) 2 h >90 0.65 −7.96 33.9 4 TBAFanh (2.5 equiv) >5 days 0 −0.03 −4.33 42.6 5 TBAFanh (2.5 equiv) >5 days 0 0 −5.01 42.5 6 TBAFanh (2.5 equiv) >5 days 0 −0.17 −3.97 45.3 While our experimental studies show that electron-withdrawing groups are required for aromatic fluorodenitration with TBAFanh in DMSO, previous work has shown that the gas-phase reaction of nitrobenzene with fluoride ions is fairly favorable (ΔH°=−28.2 kcal mol−1) and has a very low activation barrier (ΔH≠=3.5 kcal mol−1).25 The reason for the attenuated solution-phase reactivity is that fluoride, the most densely charged monoanion, is tightly ion paired and strongly solvated.15, 26 Thus, we conducted computational studies to explore the impact of solvent upon the potential-energy surfaces of fluorodenitration reactions that feature a variety of 4-substituted nitrobenzenes (Scheme 2). The calculated fluorodenitration reactions. Gas-phase structures and energies for all nitrobenzenes, fluorobenzenes, the nitrite ion, and the fluoride ion were calculated using density functional theory (B3LYP/6-311g++(d,p)/6-311g++(d,p)). Frequency calculations for all molecular species were also performed to provide thermal (298 K) and ZPE corrections for the thermochemical analysis. Polarizable continuum (PCM) and self-consistent isodensity polarizable continuum models (SCI-PCM) were employed to model the polar aprotic solvents used in the experiment. In practice, the two models gave comparable potential-energy diagrams and solvation energies. The fluorodenitration reaction energies are shown in Figure 1, in which they are plotted against Hammet σ− parameters. A comparison with the gas-phase experimental data shows that the DFT calculations correctly reproduce the experimentally determined thermodynamic enthalpy for the fluorodenitration of nitrobenzene;25 thus, we deemed the basis set and method to be sufficient for the study of substituted nitrobenzenes. The gas-phase ΔG° and ΔH° values are large and negative for all fluorodenitration reactions of substituted nitrobenzenes, and an excellent linear free-energy relationship is produced. Extrapolation of this linear plot suggests that fluorodenitration is thermodynamically favorable for all para-substituted nitrobenzenes in the gas phase. In contrast, calculated solution-phase ΔG° and ΔH° values are only modestly negative for the free-ion reaction. Inclusion of ion pairing would further lessen the equilibrium constant for fluorodenitration as the fluoride ion is expected to ion pair to a greater extent than the nitrite ion. This suggests that the limitation of fluorodenitration to electron-poor nitroaromatic compounds in polar solvents at room temperature could arise from thermodynamic considerations. A) Calculated thermodynamic free-energy relationships for gas- and solution-phase fluorodenitration reactions. B) Calculated activation parameters for fluorodenitration reactions: a) ΔG≠ in CH3CN, b) ΔH≠ in CH3CN, c) ΔG≠ for gas-phase RM to TS, d) ΔH≠ for gas-phase RM to TS, e) ΔG≠ for gas-phase dissociated reactants to TS, f) ΔH≠ for gas-phase dissociated reactants to TS. For the transition-state (TS) search and activation-barrier calculations, basis-set superposition error (BSSE) was corrected by using the counterpoise method. It was found that BSSE correction resulted in very modest adjustments to the calculated activation barrier (0.7 kcal mol−1 for the gas-phase fluorination of nitrobenzene). In the gas-phase calculations, several stable fluoride–nitrobenzene complexes were found in our extensive search of the TS geometry and reaction-side local minima (RM), in agreement with previous theoretical studies.25 Our focus was limited to the RM which led directly to transition states. Figure 2 shows typical structures of gas-phase RM and TS structures for the fluorodenitration of nitrobenzene and p-cyanonitrobenzene. If the para substituent is H or an electron-donating group, significant structural changes are observed between the RM and the TS. The RM has a weakly bound fluoride ion, little perturbation of the CN bond length, and only a minor deflection of the nitro group from the plane of the aromatic ring, whereas the TS shows a "Meisenheimer-like" structure that indicates significant charge injection into the aromatic ring. In contrast, electron-withdrawing groups lead to Meisenheimer intermediates which require minimal structural reorganization to reach the TS. The calculations for the gas-phase reaction predict an excellent kinetic linear free-energy relationship; all gas-phase fluorodenitration reactions of para-substituted nitrobenzenes should be rapid at room temperature (Figure 1). Optimized structures (top), highest occupied molecular orbitals (HOMO; middle) and electrostatic potential surfaces (ESP; bottom) for nitrobenzene (left) and p-cyanonitrobenzene (right) fluorodenitration reactions. Solution-phase TS structures were located using potential-energy surface scans, in which optimizations were performed as the CF length was stepped through several fixed values. The optimized solution-phase TS structures differ slightly from the gas-phase TS structures; the former exhibit longer CF and CN lengths. Removal of the solvent model and recalculation of the optimized solution-phase TS structures gave energies within 2 kcal mol−1 of the gas-phase TS energy for all of the substituted nitrobenzenes, thereby indicating that the impact of the solvent model upon the fluorodenitration TS geometries and internal energies is quite small. As strong solvation of the fluoride ion results in calculated RMs that feature free ions, activation energies (Table 2) were calculated directly from the energy difference between that of optimized TS structures and the sum of the energies of nitrobenzene and the fluoride ion. Preferential solvation of the fluoride ion is calculated to add 45–55 kcal mol−1 to the gas-phase free energy of activation. Activation energies for the fluorodenitration of nitrobenzene in cyclohexane, THF, and DMSO (ΔG≠=20.5, 37.5, and 42.5 kcal mol−1, respectively) were also obtained using the PCM model; decreasing solvent polarity lowers the activation barrier substantially. Thus, if one can prepare soluble, more weakly ion-paired fluoride salts, it is expected that the scope of room-temperature fluorodenitration reactions can be expanded if nonpolar media are used. A comparison of experimental and theoretical results shows that the calculated activation barriers for fluorodenitration in CH3CN (or DMSO) are much too large. The experimental data indicate that ΔG≠=20.5 and 21.5 kcal mol−1 for the fluorodenitration of p-cyanonitrobenzene and p-trifluoromethylnitrobenzene, respectively, whereas the calculated values are ΔG≠=29.8 and 33.9 kcal mol−1, respectively. Additionally, the Hammett plots (Figure 1) appear to be unrealistically steep (ρ=−13). Three sources of error in the PCM calculation are readily apparent. The calculated solvation energy for the fluoride ion is too large (−90.7 kcal mol−1) relative to the experimental value (−82.6 kcal mol−1), although it is in line with the stated error range of the solvent model.27, 28 Secondly, the overestimated steepness of the Hammett plot suggests that the solvent model cannot account accurately for the localization of negative charge in the less stable (more electron-rich) TS. The inadequacy of PCM to account for the strong electrostatic effects that arise from localized charge–solvent interactions has been highlighted by Truhlar and co-workers.29 Thus, we would expect the experimental Hammett ρ value to be smaller than the PCM calculation indicates. Finally, ion pairing of the TBA cation to the fluoride anion and the TS have not been addressed herein, nor is the nature of the actual reactive species (contact ion pair, solvent-separated ion pair, free ion) known. Thus, this relatively simple fluorodenitration reaction constitutes an exceptionally challenging test case for computational solvent models. In summary, TBAFanh in DMSO is a fluorinating reagent with unprecedented activity in room-temperature Halex and fluorodenitration reactions. The initial results suggest that ion pairing and solvation are the factors that limit the impressive nucleophilicity of the fluoride ion under conditions typically employed for nucleophilic aromatic substitution. Theoretical calculations show that gas-phase fluorodenitration reactions are both thermodynamically favorable and fast, whereas preferential solvation of the fluoride ion leads to the large activation barriers for the solution-phase reactions. It appears that a further expansion in the scope of room-temperature SNAr fluorination is possible, provided one can prepare highly soluble, weakly ion-paired fluoride salts for use in relatively nonpolar media. Supporting information for this article is available on the WWW under http://www.wiley-vch.de/contents/jc_2002/2006/z504555_s.pdf or from the author. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.