Oxidation of difluorocarbene and subsequent trifluoromethoxylation

As a versatile intermediate, difluorocarbene is an electron-deficient transient species, meaning that its oxidation would be challenging. Herein we show that the oxidation of difluorocarbene could occur smoothly to generate carbonyl fluoride. The oxidation process is confirmed by successful trifluoromethoxylation, 18O-trifluoromethoxylation, the observation of AgOCF3 species, and DFT calculations.

D ue to the unique properties of fluorine element such as strong electronegativity and small atomic radius, the incorporation of fluorine atom(s) into organic molecules could usually lead to profound changes of the latter's physical, chemical, and biological properties 1 . Therefore, significant efforts have been directed towards the development of efficient methods for introducing fluorine or fluorinated moieties into organic compounds 2,3 . Difluorocarbene (:CF 2 ) has served as a versatile intermediate and the transformations of difluorocarbene has proved to be quite efficient for fluorine incorporation 4,5 . Typical difluorocarbene conversions, including insertions into X-H bonds (X=O, N, S, etc.) 4,6,7 , [2 + 1] cycloadditions with multi-bonds 8,9 , and coupling with other carbenes [10][11][12] , can conveniently construct various fluorinated functionalities, such as difluoromethyl, gem-difluorocyclopropyl and gem-difluoroalkenyl groups. However, these typical reactions are limited to the incorporation of a -CF 2 -moiety. We have previously found that difluorocarbene is so reactive that it can be readily trapped by a suitable sulfur [13][14][15] , selenium 16 , or nitrogen source 17 to generate thiocarbonyl fluoride (CF 2 =S), selenocarbonyl fluoride (CF 2 =Se), and cyanide anion (CN − ), respectively (Fig. 1a-c). On the basis of these findings, which offers more possibilities for difluorocarbene chemistry, it is reasonable to conceive that the oxidation of difluorocarbene with a suitable oxygen source may proceed to afford carbonyl fluoride (CF 2 =O) (Fig. 1d). Usually, oxidation reactions could proceed smoothly to oxidize electron-rich substrates, but not to electrondeficient substrates 18,19 . Since difluorocarbene is an electrondeficient transient intermediate 20 , its oxidation would be a challenging task. Furthermore, because CF 2 =O is a highly reactive gas and thus hard to detect, it cannot be determined simply by spectroscopic monitoring of the reaction whether the oxidation process occurs or not.
Herein we describe the oxidation of difluorocarbene by using diphenyl sulfoxide (Ph 2 S=O) as the oxidant to provide carbonyl fluoride, a process which is confirmed by successful trifluoromethoxylation and 18 O-trifluoromethoxylation reactions, the observation of AgOCF 3 species, and DFT calculations. A latestage trifluoromethoxylation for the synthesis of a Trioxsalen derivative is shown to further demonstrate the synthetic utility of this trifluoromethoxylation protocol.

Results
Optimization of the trifluoromethoxylation conditions. Ph 3 P + CF 2 CO 2 − , developed by us recently 21 , and AgF were used as a difluorocarbene reagent and the fluoride source, respectively, in our efforts to ascertain the oxidation process via the trifluoromethoxylation of benzyl bromide 1-1 (Table 1). AgF was used to convert CF 2 =O into AgOCF 3 , which may be experimentally observed 22 to support the oxidation process. The oxidants were initially screened, but no desired trifluoromethoxylation product was detected in most cases (    19 F NMR spectroscopy c 2,2′-Bipyridine (1 equiv) was used as a ligand d 2,3,11,12-Dibenzo-18-crown-6 (1 equiv) was used as a ligand e 2,2′-Bipyridine (1.5 equiv) and 2,3,11,12-dibenzo-18-crown-6 (0.5 equiv) were used f THF (1.5 mL) was used g 2,2′-Bipyridine (1.5 equiv) was used without the crown ether. h 2,3,11,12-Dibenzo-18-crown-6 (0.5 equiv) was used without 2,2′-bipyridine Mechanistic investigations. Further experimental evidence was collected to support the difluorocarbene oxidation process. The use of other difluorocarbene reagents such as FSO 2 CF 2 CO 2 TMS 23 and TMSCF 2 Br 8 could also give the desired trifluoromethoxylation product, albeit in a low yield, suggesting that difluorocarbene is a key intermediate (Fig. 2a). CF 2 =O could not be detected in the reaction mixtures, because it is a highly electrophilic species and would be rapidly attacked by AgF to provide AgOCF 3 . Even stirring the mixture of Ph 3 P + CF 2 CO 2 − and Ph 2 S=O alone could not lead to the observation of CF 2 =O, because CF 2 =O would easily react with the nucleophile, Ph 3 P generated from Ph 3 P + CF 2 CO 2 −9 . Ph 2 S=O should be the oxygen source to oxidize difluorocarbene to generate CF 2 =O, since 18 O-labeled diphenyl sulfoxide afforded the CF 3 18 O product (Fig. 2b), and diphenyl sulfoxide underwent deoxygenation to afford diphenyl sulfide (Ph 2 S) in a high yield based on Ph 2 S=O consumed (39% of Ph 2 S=O was recovered) (Fig. 2c) (Supplementary Methods). A stepwise reaction was performed to confirm the generation of the AgOCF 3 complex (Fig. 2d). Without the presence of a substrate, heating a mixture of Ph 3 P + CF 2 CO 2 − /Ph 2 S=O/AgF with ligands at 60°C for 0.5 h led to the formation of a number of unkonwn species, as detected by 19 F NMR spectroscopy ( Supplementary  Fig. 2). Two broad signals, appearing at −21.66 and −21.94 ppm in the 19 F NMR spectrum, respectively, may correspond to two different ligand-coordinated AgOCF 3 complexes 22 . Subsequent addition of substrate 1-1 afforded the desired trifluoromethoxylation product, further supporting that AgOCF 3 was generated from the Ph 3 P + CF 2 CO 2 − /Ph 2 S=O/AgF system (Fig. 2d). In addition to the trifluoromethoxylation product, a fluorination byproduct was observed (Fig. 2d). However, almost no fluorination byproduct was observed under the optimal conditions (Table 1, entry 21), which suggests that AgOCF 3 was too reactive and decomposed easily.
DFT calculations at the M062X//6-31 + + G(d,p)/LANL2DZ level provided insights into the mechanism of the oxidation of difluorocarbene and the subsequent trifluoromethoxylation. We have previously demonstrated that Ph 3 P + CF 2 CO 2 − is an efficient difluorocarbene precursor, and has proposed that difluorocarbene is generated via a decarboxylation process, i.e., Ph 3 P + CF 2 CO 2 − → Ph 3 P + CF 2 − → :CF 2 14,15,24 . Calculations revealed that the activation energy for this process is quite low (10.12 kcal mol −1 ) (Supplementary Fig. 3 and Supplementary Data 1), supporting the mechanistic proposal. As an electron-deficient species, difluorocarbene can be attacked by Ph 2 S=O to form an O-CF 2 bond ( Complete cleavage of the S-O bond releases Ph 2 S and carbonyl fluoride (CF 2 =O), a process which is thermodynamically favored. CF 2 =O is electrophilic and is therefore trapped by AgF to generate AgOCF 3 , which can readily convert the substrates to the final products. The Ag ion can activate the substrates by precipitating the AgBr salt. Identification of transition state TS-2 enabled us to calculate the overall activation energy, i.e., 17.60 kcal mol −1 ; this value is low and in agreement with the rapid process.
The introduction of CF 3 O installation. The above results revealed that difluorocarbene could indeed be oxidized to give carbonyl fluoride. The oxidation of difluorocarbene and the subsequent trifluoromethoxylation provides an efficient protocol for CF 3 O incorporation. CF 3 O incorporation has received increasing attention because the CF 3 O group is a common structural motif in pharmaceuticals 25,26 , agrochemicals 27,28 , and functional materials 29
The substrate scope of trifluoromethoxylation. Since difluorocarbene could be oxidized and the subsequent trifluoromethoxylation proceeded smoothly (Table 1, entry 21), we then investigated the substrate scope of trifluoromethoxylation. Figure 4 shows that electron-deficient, -neutral, and -rich benzyl bromides were all converted to the desired products in moderate to good yields (5-1~5-17). Various functional groups were tolerated, e.g., halide, ketone, ester, alkene, cyano, nitro, ether, and various heterocycles. Heterocycles usually have interesting physicochemical properties, and therefore the easy access to CF 3 O-containing heterocycles could be useful in the life sciences (5-15~5-17). Transformation of secondary benzyl bromides gave moderate yields (5-18~5-22). The diphenyl substituted product (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) was unstable, and a heterolytic cleavage of the C-OCF 3 bond readily occurred to form a diphenyl-stabilized methyl cation, hydrolysis of which led to an alcohol by product (Ph 2 CH-OH) in 35% isolated yield. In addition to benzyl bromides, allyl bromides were also converted under these conditions (5-23~5-28). The reactivity of alkyl bromide  was much lower than that of benzyl bromides. Alkyl iodides (5-30~5-33) underwent the desired reaction smoothly to give the expected products in moderate yields. A method for achieving direct access to a flavone derivative was developed (5-34) and a moderate yield was obtained for a large-scale reaction , demonstrating the synthetic utility of this trifluoromethoxylation protocol. Trioxsalen, a furanocoumarin and a psoralen derivative obtained from plants, can be used for phototherapy treatment of vitiligo and hand eczema 51 . A convenient route to the CF 3 Ocontaining Trioxsalen derivative (8) was developed to further show the synthetic utility of this trifluoromethoxylation strategy. The trifluoromethoxylation of the precursor (7), prepared from the commercially available m-benzenediol by a reported procedure ( Supplementary Fig. 1) 52,53 , occurred smoothly to give the Trioxsalen derivative in a moderate yield (Fig. 5). 18 O-Trifluoromethoxylation. 18 O-Labeling trifluoromethoxylation is challenging, because all reported trifluoromethoxylation methods have to use a CF 3 O-containing reagent and the corresponding CF 3 18 O-reagents are difficult to prepare. Recently, Tang used an 18 O-labeled reagent, ArSO 2 -18 OCF 3 , to explore and elucidate the mechanism of the trifluoromethoxylation reaction; only a 33% 18 O content was obtained in the desired product 37 . They proposed that the low 18 O-content was because of the 16 O-18 O exchange in the SO 2 -18 OCF 3 moiety from the reagent. We employed 18 O-labeled diphenyl sulfoxide (Ph 2 S= 18 O, 18 O content: 89%) as the oxygen source in this difluorocarbene-oxidationbased trifluoromethoxylation reaction. Since the reagent, Ph 2 S= 18 O, did not contain any 16 O atom, no 16 O-18 O exchange would occur and therefore the expected products were obtained with high 18 O contents (Fig. 6).