Synthesis of tertiary alkyl fluorides and chlorides by site-selective nucleophilic ring-opening reaction of α-aryl azetidinium salts

Site-selective nucleophilic ring-opening reactions of 2-arylazetidine-2-carboxylic acid ester-derived tetraalkyl ammonium salts 2 with tetrabutylammonium halides (Bu4NX) to give tertiary alkyl halides are successfully demonstrated. For example, a nucleophilic ring-opening reaction of 2-(o-tolyl) derivative 2a with 1.2 equivalents of tetrabutylammonium fluoride (Bu4NF) in THF at 60 °C preferentially proceeded at a more substituted carbon atom (2-position) compared to a less-substituted carbon atom (4-position) and afforded tert-butyl 4-(dimethylamino)-2-fluoro-2-(o-tolyl)butanoate 3aa in 71% yield as the corresponding tertiary alkyl fluoride. This result was applied to synthesize optically active organofluorine compounds starting from commercially available (R)-1-phenylethylamine.


Introduction
Ring-strained four-membered N-heterocycle azetidines are valuable building blocks in organic synthesis. Although they are chemically stable without any additives, nucleophilic ringopening reactions proceed to give various types of functionalized nitrogen-containing compounds by electrophilic activation of the nitrogen atom by N-quaternization, 1,2 or addition of Brønsted acid (H + ) 3 or Lewis acids 4 (Scheme 1). 5 These transformations are applicable for the synthesis of amino acids, alkaloids, and biologically active drugs.
The initial studies of this ring-opening reaction were mainly performed by Couty's group using tetraalkylazetidinium salts as substrates. 1 One point to consider in this reaction is siteselectivity at the 2-and 4-positions, which reacts with a nucleophile (Nu). In many cases, a less-substituted and/or electron-decient carbon atom is attacked by a nucleophile because of the S N 2 process. For example, a reaction of a substrate with a nucleophile in Scheme 1 proceeded at the 4-position preferentially to afford the corresponding product. However, some nucleophiles do not act according to this tendency, and the reaction occurs at the 2-position, which is a much-substituted carbon atom. Although these phenomena are currently difficult to explain, the site selectivity at the 2-and 4-positions can be determined based on the properties of nucleophiles, substituents at the 2-and 4-positions, and reaction conditions. 1d,g,h Previously, our group reported that the site-selective nucleophilic ring-opening reaction of a-arylazetidine-2carboxylic acid ester-derived tetraalkylammonium salt (S)-2b prepared from 95% ee of (S)-1b (Scheme 2, Our previous work). 6 Cesium acetate (AcOCs) and dimethylamine (Me 2 NH) as nucleophiles reacted at the 4-position. In contrast, sodium azide (NaN 3 ) reacted at the 2-position with inversion of the conguration. This result shows that the S N 2 substitution at the tertiary carbon atom (2-position) proceeded. 7 With the results, our group started to further investigate the scope of this reaction, since some nucleophiles such as uoride (F À ) provide valuable compounds. Furthermore, previous examples of the ring-opening reaction of azetidine derivatives with F À to give organouorine compounds are rare 2a, 8 compared to the reaction of three-membered N-heterocycle aziridine derivatives. 9 Herein, we wish to report the site-selective nucleophilic ring-opening reaction of a-aryl azetidinium salts 2 with halides to afford aaryl-a-halo-carboxylic acid esters 3 (Scheme 2, this work). Further synthetic applications of the resulting products 3, e.g., asymmetric synthesis of organouorine compounds, are also demonstrated.

Results and discussion
We started investigating the nucleophilic ring-opening reaction of 2a with a halide source (Table 1). First, the reaction of 2a with sodium uoride (NaF) as an F À source in DMF at room temperature for 2 h was examined to obtain the corresponding organouorine compounds 3aa and 4aa; however, no products were obtained (entry 1). Although a reaction with potassium uoride (KF) gave the same result (entry 2), the use of cesium uoride (CsF) afforded 3aa in 13% yield (entry 3). We expected that tetrabutylammonium uoride (Bu 4 NF) might be more reactive, and its solubility in organic solvents would improve the yields of 3aa and 4aa. In addition, Ghorai et al. reported the Lewis acid-promoted nucleophilic ring-opening reaction of Ntosylazetidines with tetrabutylammonium chloride (Bu 4 NCl) and bromide (Bu 4 NBr). 10 Thus, we attempted a reaction with a THF solution of Bu 4 NF, and the desired 3aa was obtained in 33% yield with trace amounts of 4aa (<4% yield) (entry 4). The use of THF as a solvent and other F À sources, such as Bu 4 -NF$3H 2 O, did not show any improvements (entries 5 and 6). We found that the yield of 3aa could be improved to 71% with minimization of the formation of 4aa (7% yield) when the reaction was performed at 60 C (entry 7).
Next, we examined the same reaction with other tetrabutylammonium salts (Bu 4 NX) to dene the scope of this siteselective ring-opening reaction. Reactions with Bu 4 NCl in THF, DMF and CH 2 Cl 2 proceeded even at room temperature, and similar yields of 3ab (69-76% yields) and 4ab (14-23% yields) were observed (entries [8][9][10]. At 0 C, the yields of 3ab (34% yield) and 4ab (10% yield) decreased (entry 11). When the reaction was performed at 60 C, the yield of undesired 4ab was slightly improved (27% yield) (entry 12). The use of Bu 4 NBr is also applicable; however, the selectivity between 3ac (61% yield) and 4ac (21% yield) was insufficient (entry 13). Additionally, the resulting isolated bromo products 3ac and 4ac were unstable because of the self-N-quaternization. Therefore, a reaction with tetrabutylammonium iodide (Bu 4 NI) did not give 3ad and 4ad (entry 14). Finally, we applied this reaction for pseudohalogen salts (MCN) to provide a-cyano derivative 3ae with an all-carbon quaternary stereocentre (entries 15 and 16). Unfortunately, both reactions with potassium cyanide (KCN) and tetrabutylammonium cyanide (Bu 4 NCN) gave similar results to provide 3ae and 4ae without selectivities. 11 The ring-opening products 3 and 4 in Table 1 were assigned by NMR analyses, and their representative results are shown in Fig. 1. Fluorine derivatives 3aa and 4aa were clearly identied by the 19 F NMR analysis. Tertiary alkyl uoride 3aa showed a chemical shi of À157 ppm. Primary alkyl uoride 4aa showed a chemical shi of À222 ppm. These values are reasonable for the corresponding alkyl uorides. In contrast, chlorine derivatives 3ab and 4ab did not show clear differences in 1 H and 13 C NMR analyses. Consequently, we assigned these Scheme 2 Nucleophilic ring-opening of a-arylazetidine-2-carboxylic acid ester-derived tetraalkylammonium salts 2. by comparison of 1 H NMR chemical shis of methylene protons. Primary alkyl chloride 4ab had low-eld chemical shis due to an electron-withdrawing effect of chloride. One of the two products (3ab or 4ab) with chemical shis of 3.27 and 2.98 ppm was assigned to 4ab. Another product was assigned to tertiary alkyl chloride 3ab, which showed chemical shis of 2.73-2.33 ppm. Bromine derivatives 3ac and 4ac were assigned by analogy to 3ab and 4ab. Meanwhile, nitrile derivatives 3ae and 4ae could be clearly identied by 13 C NMR analysis. 4ae showed a chemical shi of 12.1 ppm, which is a reasonable value as a primary nitrile. 1d To dene the scope and limitations of this site-selective ringopening reaction to produce tertiary alkyl halides 3, we prepared various azetidinium salts 2b-h and examined their reactions with Bu 4 NF or Bu 4 NCl under identical conditions (Table 2). First, we attempted the reactions of 5-substituted aryl derivatives 2a-e with Bu 4 NF and obtained the corresponding organouorine compounds 3ba-ea in moderate yields (entries [1][2][3][4]. The minor products 4 were not isolated (N.D.), although their formations were observed by TLC analysis. The pure products of these organouorine 4 for spectroscopic characterizations were difficult to isolate because of small amounts (ca. 5% yield). Electron-withdrawing substituents on the a-aryl substituent, such as bromo (2b) and triuoromethyl (2c), might be desirable to yield 3 (entries 1 and 2, approximately 75%). Reactions of methyl (2d) and methoxy (2e) derivatives resulted in lower yields of 3 (entries 3 and 4, approximately 60%). Thus, we next examined the reactions of 4-bromo (2f) and 4-tri-uoromethyl (2g) derivatives and obtained 3fa-ga in approximately 70% yields (entries 5 and 6). However, the reaction of 3bromo derivative 2h was resulted in a 58% yield of 3ha (entry 7).
The use of Bu 4 NCl for the reactions of 2b, 2d, 2f, and 2h provided the corresponding organochlorine compounds 3bbhb (entries 8-11) with a similar tendency to the reaction with Bu 4 NF. In these cases, the minor products 4bb-hb could be isolated as a pure form to perform their spectroscopic characterizations.
We conrmed the chemical stability of products 3 and 4 (Scheme 3) because a transformation between 3 and 4 might proceed via the formation of ammonium salts generated from the alkyl halides and dimethylamino substituents as in the products (self-N-quaternization). A THF solution of tertiary alkyl Scheme 3 Chemical stability of ring-opening products 3aa, 3ab and 4ab.
halides 3aa (X ¼ F) or 3ab (X ¼ Cl) was subjected to the reaction temperature depicted in Table 1. The removal of THF by evaporation and 1 H NMR analysis of the residue did not show any formation of 4aa or 4ab, respectively (eqn (1)). Similarly, a stirring at room temperature of a THF solution of primary alkyl chloride 4ab did not afford 3ab (eqn (2)). The N,N-dimethylamino substituent, as in product 3, is not synthetically valuable because of the impossibility of removing the N-methyl substituents. One N-methyl substituent could be changed into an N-allyl, which would be removable via Rhcatalysed isomerization, by N-quaternization of 1 with allyl tri-ate 12 (Scheme 4). For example, azetidinium salt 5 was prepared from 1b in 70% yield as an 8/2 mixture of diastereomers followed by ring-opening with Bu 4 NF to provide N-allyl derivative 6 in 82% yield. Rh-catalyzed deallylation of 6 gave secondary amine 7 in 84% yield.
To demonstrate the utility of this ring-opening reaction, we attempted further synthetic transformations of organouorine product 3aa. First, Hofmann elimination of 3aa to produce aaryl-a-uoro-a-vinylacetic acid ester 9 was examined (Scheme 5). N-Quaternization with iodomethane (MeI) or methyl tri-uoromethanesulfonate (MeOTf) gave 8-I or 8-OTf in good yields . Treatment of iodide salt 8-I with 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) in reuxing toluene for 1 day gave desired 9 in 36% yield. We expected that the iodide ion in the reaction mixture might cause undesirable side reactions such as nucleophilic substitutions, and the reaction resulted in a low yield. Thus, we examined the same reaction using triate salt 8-OTf. As expected, the yield of 9 was improved to 57%.
Next, the synthesis of optically active tertiary organouorine compounds from chiral (R)-1-phenylethylamine, which is one of the least expensive chiral sources, was examined (Scheme 6). 93% ee of (S)-1a was prepared according to our previous work. 6,13 N-Quaternization of (S)-1a with MeOTf to prepare (S)-2a (quant.) followed by the ring-opening reaction with Bu 4 NF under the conditions in Table 1 afforded (R)-3aa (68% yield). The ee of the obtained 3aa was determined aer conversion into (R)-11 because of the low sensitivity of 3aa towards a UV/vis detector in chiral HPLC analysis. Reduction of (R)-3aa with LiAlH 4 to amino alcohol (R)-10 (73% yield) followed by O-benzoylation gave benzoate (R)-11 (95% yield). The ee of (R)-11 was determined to be 93% ee by the chiral HPLC analysis. No lack of the ee was conrmed during the transformations from (S)-1a into (R)-11. This result indicates that the Bu 4 NF-promoted ringopening reaction of (S)-2a affording (R)-3aa proceeds by inverting the tertiary carbon conguration (S N 2) in the same manner as the reaction of (S)-2b with NaN 3 , which was previously reported by our group. 6 Therefore, the absolute conguration of 3aa was determined to be (R).
To clarify that the a-aryl substituent as in 2 is necessary for this site-selective ring-opening reaction to produce 3, we investigated a reaction a-ethyl derivative 12 with Bu 4 NF (Scheme 7). As expected, the reaction proceeded at 4-position preferentially to give g-uoro product 14 in 62% yield. Identi-able amount of the corresponding a-uoro product 13 was not Scheme 4 Synthesis of N-allyl derivative 5 and 6 and deallylation into 7.
obtained. Instead, a-hydroxy derivative 15, which might be derived from 13, was isolated in 7% yield.
Couty's group described in the previous literature 1d that the nucleophilic ring-opening of a,a-disubstituted azetidinium ions at the quaternary a-carbon (2-position) is intrinsically favoured. Steric repulsions generated by substituents as in the azetidine ring affect the site-selectivity. The highly nucleophilic azide anion (N 3 À ) reacts at 2-position, the less nucleophilic cyanide anion (CN À ) reacts at 2-and 4-positions, and the poor nucleophilic acetate anion (AcO À ) reacts at 4-position. The exact reason of the site-selective ring-opening reaction to produce 3 demonstrated by our group are difficult to explain at present, a size of the nucleophiles might affect the site-selectivity. F À and Cl À are small and enable to react at the quaternary a-carbon (2position) although they are poor nucleophilic anion. Further experimental studies are needed to discuss.

Conclusions
In conclusion, we described that the site-selective nucleophilic ring-opening reaction of 2-arylazetidine-2-carboxylic acid esterderived ammonium salts 2 with Bu 4 NF or Bu 4 NCl proceeded at a much-substituted 2-position preferentially over a lesssubstituted 4-position and produced the corresponding tertiary alkyl uorides and chlorides 3. Our result is a rare successful example of the uoride ion-promoted ring-opening reaction of azetidine derivatives that yields organouorine compounds. Further synthetic transformations of the product 3 were also successfully demonstrated. Our protocol enables the production of optically active organouorine compound (R)-3aa starting from commercially available chiral (R)-1-phenylethylamine, which is an inexpensive chiral compound.

Experimental
General Specic rotations were recorded on a JASCO polarimeter P-1010. Normal phase chiral HPLC analyses were performed using a JASCO HPLC pump (PU-2089) and a UV/vis detector (UV-2075). Infrared spectra (IR) were recorded on a JASCO FT/IR-4600 spectrometer. 1  were performed for representative products. As an internal standard in CDCl 3 , Me 4 Si (d 0 ppm) for 1 H NMR and CDCl 3 (d 77.00 ppm) for 13 C NMR were used. As an internal standard in acetone-d 6 , the residual protons (d 2.05 ppm) for 1 H NMR and acetone-d 6 (d 29.92 ppm) for 13 C NMR were used. In 19 F NMR, hexauorobenzene (C 6 F 6 ) was used as an internal standard (d À162.9 ppm). In 1 H and 13 C NMR, the splitting patterns are denoted as follows: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; and br, broad peak. In 19 F NMR, the splitting patterns are not denoted. High-resolution mass spectra (ESI) were measured on a Thermo Fisher Scientic LC/FT-MS spectrometer. Reactions involving air-or moisture-sensitive compounds were conducted in appropriate round-bottomed asks with a magnetic stirring bar under an argon (Ar) atmosphere. A 1 M tetrabutylammonium uoride (Bu 4 NF) THF solution was purchased from Tokyo Chemical Industry Co., Ltd. (TCI). Anhydrous tetrahydrofuran (THF) was purchased from KANTO Chemical Co., Inc. For the thin layer chromatography (TLC) analysis throughout this work, Silicagel 70 TLC Plate-Wako purchased from FUJIFILM Wako Chemical Corporation was used. The products were puried by column chromatography on silica gel (Wakosil 60, 64-210 mm) purchased from FUJIFILM Wako Chemical Corporation. For strong basic compound such as (S)-10, NH TLC plates and amino-functionalized silica gel (Chromatorex NH-DM1020) purchased from Fuji Silysia Chemical Ltd. (Japan) were used.
Representative procedure for ring-opening of 2a with Bu 4 NF in THF to afford 3aa and 4aa (Table 1,