Fluorinative ring-opening of cyclopropanes by hypervalent iodine reagents. An efficient method for 1,3-oxyfluorination and 1,3-difluorination

Reaction of 1,1-disubstituted cyclopropanes with hypervalent iodines in the presence of AgBF4 leads to 1,3-difluorination and 1,3-oxyfluorination products.

Fluorinated organic compounds have found broad application in the pharmaceutical, 1 and agrochemical industries 2 as well as in medical diagnostics. 3 The impetus for the application of organouorine compounds in agrochemical and pharmaceutical products is their benecial pharmacokinetic properties, such as high metabolic stability and lipophilicity. 1a-d The useful radionuclear properties of the unnatural isotope 18 F makes 18 F labelled organouoro compounds indispensable for positron emission tomography (PET). 3a The short half-life of 18 F requires development of a rapid late stage introduction of the uorine atom, 3 which is a challenging task in synthetic organic chemistry. 4 In the last decade, many new uorinating reagents have appeared, which in combination of catalysts allowed development of new selective methodologies to access a broad variety of bioactive organouorines. 4a,5 The most efficient methods are even suitable for uorination based difunctionalization reactions. 5b-d The most studied approach involves vicinal difunctionalization reactions, such as 1,2-oxyuorination, 6 1,2-aminouorination, 6a,7 1,2-carbo-uorination 6a,8 and related methods. 9 Recently a number of interesting geminal uorination methods were also reported, such as 1,1-diuorination, 10 1,1-oxyuorination 11 and 1,1-ami-nouorination. 12 The 1,2-difunctionalization methods are usually based on alkene substrates, while the 1,1-difunctionalizations are oen realized using diazo compounds, as substrates. However, the analogue methodology is much less developed for 1,3-difunctionalization based uorination methods. Considering the typical synthetic methodologies for 1,3-difunctionalization reactions, 13 a related uorination reaction can probably be achieved by ring opening of cyclopropane substrates. Recently, we have shown that hypervalent iodine based 14 benziodoxol(on) derivatives are excellent reagents for 1,1-and 1,2-difunctionalization for synthesis of organic tri-uoromethyl and uoro compounds. 6a,9a,10a,11,15 As a part of our concept driven uorine chemistry program, we sought to employ uoro-benziodoxol reagent 1a for a uorinative ring opening of cyclopropane derivatives. To our delight, 1a reacted smoothly with cyclopropane derivative 2a in the presence of AgBF 4 affording 1,3-diuoro substituted compound 4a with 71% yield (Scheme 1).
As we employed 1a and 2a in equimolecular ratio in this reaction, one of the uorine atoms originated from 1a, while the other one is from the BF 4 À counter ion of the Ag-mediator. We have previously reported 10a a similar 1,1-diuorination method of styrenes. Although, several chlorination and bromination methods of cyclopropane are reported in the literature, 16 synthetically useful cyclopropane opening is a very unusual methodology for uorination reactions. As far as we know the above process is the rst 1,3-diuorination reaction. In addition, we have found only a single uorination based 1,3difunctionalization reaction in the literature. Very recently, Lectka and co-workers 17 reported an aminouorination method based on cyclopropane substrates. As mentioned above the 1,3-diuorination of cyclopropane 2a could be carried out selectively and in high yield using 1a and a stoichiometric amount of AgBF 4 (Table 1, entry 1) in CDCl 3 . We used CDCl 3 as the solvent to directly monitor the possible formation of the volatile uorinated (and other) by-products in the reactions. Replacing AgBF 4 with AgPF 6 as a secondary uorine source led to formation of 4a, but the yield dropped to 34% (entry 2). Cu(MeCN) 4 BF 4 can also be used instead of AgBF 4 .
The yield was lowered indicating that silver is a better mediator than copper for this transformation (entry 3). However, simple silver sources such as AgF showed to be inactive in 1,3-diuorination reaction (entry 4). Zinc salts have proved to be efficient activators of benziodoxole reagents. 6a,18 Therefore, we attempted to replace AgBF 4 with Zn(BF 4 ) 2 but the corresponding reaction did not result 4a (entry 4). Other silver salts without transferable uoride in the counter ion, such as AgCN or AgTFA, did not show any activity (entry 5). When a sub-stoichiometric amount (30 mol%) of Ag-salt was used, the yields sharply decreased (entries 6-7). Pd(BF 4 ) 2 (MeCN) 4 (30 mol%) was also inefficient as catalyst (entry 7). Interestingly, Cu(MeCN) 2 BF 4 showed some catalytic activity but the yield was very low (entry 8). Only traces of product 4a (<5%) could be obtained with 30 mol% of AgBF 4 and stoichiometric amount of NaBF 4 (entry 9). The reaction was completely shut down when KF was employed instead of NaBF 4 (entry 10). This indicates that the most efficient secondary uorine source is AgBF 4 . We could not observe any reaction without application of AgBF 4 or 1a (entry 11).
Neither Selectuor nor NFSI could replace uoroiodoxol 1a as the electrophilic uorination reagent (entry 12). When benziodoxole based 1a was replaced by 4-iodotoluene diuoride (Tol-IF 2 ), a related hypervalent iodine reagent, 14a product 4a did not form at all. Unlike 1a, Tol-IF 2 underwent rapid decomposition in the presence of AgBF 4 . When the reaction was performed in the absence of AgBF 4 (3) with Tol-IF 2 a complex reaction mixture was obtained, from which compound 4a could be isolated in 24% yield (entry 13). In general, we found Tol-IF 2 much less bench-stable than 1a and more prone to providing complex product mixtures.
A brief solvent screen has shown that dichloromethane is a less suitable solvent providing the product in 10% yield (entry 14). However formation of product 4a was not observed when chloroform was replaced by acetonitrile or methanol (entry 15).
Subsequently, we investigated the synthetic scope of the silver mediated 1,3-diuorination reaction (Table 2). We found that several substrates required longer reaction times for full conversion relative to 2a (Table 2, entry 1). Under an elongated reaction time 1a underwent partial decomposition (see below). Therefore, in most reactions we employed two equivalents of 1a to obtain a full conversion of 2 and, thus optimal yields of 4. Aliphatic substrate 2b reacted for 4 h at room temperature affording 4b. Dialkyl cyclopropanes such as, 1,1-dibutyl cyclopropane 2c also reacted affording 4c (entry 3). In this case the 30 mol% 3 and 1 equiv. NaBF 4 <5 10 30 mol% 3 and 1 equiv. KF 0 11 Without 3 or 1a 0 12 1 equiv. Selectuor or NFSI instead of 1a 0 13 Without 3, 1 equiv. of Tol-IF 2 instead of 1a 24 14 DCM instead of CDCl 3 10 15 MeCN or MeOH instead of CDCl 3 0 a Reagent 1a (0.1 mmol) cyclopropane 2a (0.1 mmol) and AgBF 4 (3) (0.1 mmol) were mixed in CDCl 3 (0.5 ml). This mixture was stirred at 50 C for 1 h. yield was lower than for diuorination of 2b indicating that the reaction is fairly sensitive to the steric factors of the cyclopropane substituents. We have studied the reactivity of aryl substituted cyclopropanes as well. 1,1-Dipenyl cyclopropane 2d is a particularly challenging substrate. It is sterically hindered and the uorine expected to enter to a dibenzylic position. We found that 2d reacted relatively quickly (2 hours) with 1a resulting in 4d (entry 4) in 47% yield. As expected 4d had a limited stability, which could explain the relatively low yield. A possible reason for the poor stability is the easy dissociation of the uoride from the dibenzylic position. When one of the phenyl groups in 2d was changed to a methyl group, 2e, the reaction required a longer reaction time (6 hours), however the yield of the corresponding product 4e was higher, 59% (entry 5).
Product 4e was also more stable than 4d probably because of the stronger quaternary C-F bond. In the presence of electron donating group in the para position of the aromatic substituent, 2f, we obtained a fast uorination reaction (only 1 hour at room temperature) affording 4f in 55% yield (entry 6). Apparently electron donating groups accelerate the reaction. Naphthyl substituted substrate 2g also reacted smoothly to give 4g in 65% yield (entry 7). The rate of the reaction was much slower in the presence of an electron withdrawing group (e.g. 2h) than for electron donating group (e.g. 2f) in the para position of the aryl substituent. Thus, para-bromo substituted 2h had to be reacted 24 hours to provide 4h (entry 8), while the reaction of para phenyl substituted substrate 2f was compete in 1 hour (entry 6). Similarly to the aliphatic substrates (e.g. 2c) the diuorination reaction can be carried out for longer homologues of the methyl substituents. For example 2i-j reacted with high yields affording diuorinated products 4i-j (entries 9-10). The presented 1,3-diuorination method can be easily scaled up by ve times without signicant change in yield (entry 1). Table 2 shows that the above reaction is suitable for the synthesis of quaternary 1,3-diuoro compounds 4a-j from 1,1disubstituted cyclopropanes 2a-j. However, when we attempted to react 1,2-disubstituted cyclopropanes, we obtained very complex, inseparable mixtures with several uorinated products. The observation that this reaction proceeds faster in the presence of electron donating and/or aryl substituents on the cyclopropane moiety suggests an electrophilic uorinative cyclopropane opening mechanism. As mentioned above (Scheme 1, Table 1) the overall reaction can be regarded as a formal introduction of an F 2 molecule into the cyclopropane substrates. The electrophilic uorine atom (formally F + ) supposedly comes from reagent 1a, while the nucleophilic uorine atom (formally F À ) from the BF 4 À counter ion. 19 Considering this hypothesis, we attempted to introduce uorine and a different functionality to cyclopropanes applying this concept. When we replaced uoroiodoxole 1a with acetoxyiodoxole 1b, the reaction with 2a resulted in 1,3-oxyuorinated product 5a (Table 3, entry 1) in 84% yield. In this reaction, we did not observe formation of diuorinated product 4a. In addition, the regioselectivity was also very high as we could not detect formation of the regioisomer of 5a. Aliphatic and aryl substrates 2b and 2f also reacted with the same chemo-and regioselectivity as 2a (entries 2 and 3). Products 5b-c had a limited stability, and decomposed within a couple of hours at room temperature. Instead of 1b, 1c (PIDA) could also be employed as acetoxy source. In this reaction, we also obtained 5a in good yield (entry 4) without formation of diacetoxy or diuoro (4a) analogues. Interestingly, 1c reacted much faster (20 min) than the iodoxole analogue 1b (4 hours). When benzoyl analogue 1d was used benzoyl product 5d formed instead of 5a (entry 5).
Cyclopropane derivatives 2a and 2b were also reacted with uoroiodoxole 1a in the presence of benzyl alcohol (6) and AgBF 4 . In these reactions the nal products were 1,3-oxy-uorinated species 5e-f (Table 3, entries 6-8) instead of 4a-b (Table 2, entries 1-2), which were formed in the absence of benzyl alcohol. Since in oxyuorination only a single uorine is introduced, we attempted to react 2a and 1a in the presence of benzyl alcohol and sub-stoichiometric amount of AgBF 4 3 (entry 7). However, the yield of the oxyuorinated product 5e substantially decreased (c.f. entries 6 and 7). Apparently, application of stoichiometric amount of AgBF 4 is required, for both as a source for the secondary uorine atom in the diuorination reaction (such as for formation of 4a) and also in the oxyuorination reaction for efficient activation of 1a. In the oxyuorination reactions the activated hypervalent iodine reagents proved to be more stable than in the diuorination reactions. Therefore, in most processes (entries 2-7) one equivalent of the iodine reagent was sufficient to obtain the reported isolated yields.
In order to obtain more insight into the electronic effects of the reactions and the role of the applied hypervalent iodine, we performed a couple of control experiments. When an equimolar ratio of 2e, 2h and 1a reacted in the presence of AgBF 4 , we obtained only 4e, while formation of 4h was not observed (Scheme 2). This competitive reaction indicates that cyclopropane substrates bearing an electron withdrawing group, such as 2h, react much slower than the parent compound 2e.
This conrms the suggestion of the electrophilic mechanism for the opening of the cyclopropane ring. When 2a was reacted with equimolar amounts of uoro-(1a) and acetoxyiodoxoles (1b) products 4a and 5a were formed in 1 : 2 ratio (Scheme 3) indicating that the oxidation power or the electrophilicity of the hypervalent iodine is an important factor for the reaction rate.
Considering the above and the literature data for related reactions, 6a,9a,10a we propose a plausible mechanism for the uorinative opening of cyclopropanes with hypervalent iodines (Scheme 4). Benziodoxole reagents 1a-b are stable 6b under ambient conditions, and usually require activation in the substitution and addition reactions. 6a,9a,10a We suggest that AgBF 4 activates 1a-b by coordination of the oxygen atom of the benziodoxole ring to the silver cation affording intermediate 7.
Similar, Lewis-acid type of activation of benziodoxoles was reported by Togni and co-workers. 18 Unlike, 1a-b, activated benziodoxole 7 is very reactive, and besides the desired uorination reaction it may undergo decomposition (or other side-reactions). This is the reason for application of two equivalents of 1a in some difunctionalization reactions where the substrate has a low reactivity or the rate of decomposition of intermediate is high. We suggest that 7 undergoes side-attack of the cyclopropane ring (8) to give carbocationic intermediate 9 and iodobenzene derivative 10. This mechanism is reminiscent of our proposal for the diuorination of styrenes with 1a. 10a The high regioselectivity of the attack is an interesting feature of the process (Table 3). A possible explanation is that the regioselectivity is controlled by electronic effects, i.e. hyperconjugative stabilization of the tertiary carbocation center. The nal step of the process could be a nucleophilic attack by uorine from the BF 4 À counterion 19 to obtain the nal product (4 or 5).
In case of oxyuorination with benzyl alcohol (Table 3, entries 6-8) 1a was probably reacted with 6 prior to the ring opening providing benzyloxy-benziodoxole (analogue to 1b). In this case intermediate 9 is a benzyl ether (X ¼ OBn). This idea is supported by the control experiment (Scheme 5), in which, we rst performed a diuorination affording 4a, then 6 was added. In this reaction we obtained 5g, which is the regioisomer of 5e (see Table 3, entry 6).
Accordingly, when 2a, 1a, 3 and 6 were mixed at the onset of the reaction (Table 3, entry 6) diuorination product 4a did not form. This reaction lead to the formation of 5e directly (according to the mechanism outlined in Scheme 4). If 4a formed rst in the process, benzyl alcohol (6) would have displaced the tertiary uorine affording 5g (Scheme 5).
Modelling and experimental studies are underway to explore the mechanistic details of the above and related 6a,9a,10a metal mediated reactions of uoro-benziodoxol reagent 1a.
In conclusion, we have shown that the air-and moisture stable uoroiodine reagent 1a is suitable for the silver mediated 1,3-diuorination reaction of 1,1-disubstituted cyclopropanes. The reaction can be extended to 1,3-oxydiuorination by using hypervalent acetoxy and benzoyloxy iodines. The reaction probably proceeds via electrophilic ring opening of cyclopropanes. As the above process is the rst 1,3-diuorination and 1,3-oxydiuorination reaction, it broaden the synthetic scope of the uorination reactions, and the application area of hypervalent uoroiodines.