Chlorination of phenylallene derivatives with 1-chloro-1,2-benziodoxol-3-one: synthesis of vicinal-dichlorides and chlorodienes

Allyl and vinyl chlorides represent important structural motifs in organic chemistry. Herein is described the chemoselective and regioselective reaction of aryl- and α-substituted phenylallenes with the hypervalent iodine (HVI) reagent 1-chloro-1,2-benziodoxol-3-one. The reaction typically results in vicinal dichlorides, except with proton-containing α-alkyl substituents, which instead give chlorinated dienes as the major product. Experimental evidence suggests that a radical mechanism is involved.

Recent reports of reactions between hypervalent iodine reagents and phenylallenes have highlighted the possible product outcomes achievable through ionic and radical reaction path-Scheme 1: Reactions of substituted allenes with HVI reagents.

Results and Discussion
We began our investigation of allene chlorination using p-tolylallene (2a), prepared from 4-methylstyrene through Doering-Moore-Skattebøl reaction [46], and iodane 1a. The reaction was carried out using a slight excess of 1a in acetonitrile, at both room temperature and at reflux, and upon consumption of the allene an inseparable mixture of chlorination products 3a and 3a' were obtained. While the overall yield of the chlorinated products increased when under reflux conditions, very little change in chemoselectivity was observed (Scheme 2) [31]. As these results were consistent with those achievable by other allene chlorination reactions, it was not investigated further.
We next investigated the chlorination of 2a with benziodoxolone 1b [47][48][49], which proved highly regioselective. An initial reaction with 2.2 equiv of 1b in acetonitrile at room temperature failed; however, repeating the reaction under reflux conditions gave 3a as a mixture of E/Z alkenes in 58% yield (Table 1, entries 1 and 2). The reaction was entirely selective for the terminal alkene, with none of 3a' being observed. Toluene, chlorobenzene, DMF and DCE were also tested as reaction solvents, but none were superior to acetonitrile (Table 1, entries 3-6). A small improvement in yield was achieved by adding 2a dropwise over 30 minutes (Table 1, entry 7), and we ultimately found that adding 2a dropwise over one hour was optimal, giving 3a in 90% yield as a E:Z = 1:1.25 mixture (Table 1, entry 8). One final reaction was carried out using the related gem-dimethyl chlorobenziodoxole [49], but the yield of 3a decreased to 45% (Table 1, entry 9). This result is the first example of a selective chlorination reaction of phenylallenes, and as the regiochemical outcome parallels that observed by Liu (Scheme 1a), it is likely that radical pathways are involved [50]. A series of aryl-and allenyl-substituted phenylallenes (2b-v) were then examined in the chlorination reaction. First, phenylallenes with various aryl substituents were investigated, and the p-tolyl and p-biphenyl derivatives gave the 1,2-dichlorides 3a and 3b in excellent yield, favouring the Z-alkene (Scheme 3). The 4-bromo and 2-, 3-or 4-chloro derivatives 2c-f led to 3c-f in only moderate yield, with the mass balance of chlorinated materials being made up by the regioisomeric vicinaldichloroination products 3c'-f' (compare with 3a', Scheme 2) [30]. The p-anisyl derivative 2g was also viable in the reaction, giving 3g in 64% yield, as were the 1-and 2-naphthylallenes (2h and 2i), which gave the desired dichlorides 3h and 3i in 93% and 78% yield. In each case, preference for forming the Z-alkene was observed, with selectivities ranging from 1.2-4.3:1 Z:E. α-Substituents on the allenes were equally viable, as 1,1-diphenylallene gave 3j in 84% yield, and the related mono-methyl and mono-chloro derivatives 2k and 2l gave dichlorides 3k and 3l in 79% and 56% yield, respectively, with little preference observed for formation of either the Z or E alkene. Curiously, with 1,1-di(p-anisyl)allene (2m), only a trace of 3m was observed, and the reaction instead produced iodobenzoate 3m" in 57% yield. Presumably, this anomalous result arose due to the increased stability offered to an electrondeficient radical intermediate by the two methoxy groups, permitting a deviation in reaction outcome.
As with alkoxy substrate 2m, the α-methylated substrates 2u and 2v possessing methoxy group(s) on the arene also deviated from the expected reaction course. These reactions failed to fully consume the starting materials 2u and 2v, even upon prolonged heating, which we discovered to be the result of 1b being also consumed through over-chlorination. 4-Methoxy derivative 2u gave trichloride 5u in 53% yield, with no trace of the expected dichloride 3u or chlorodiene 4u products observable by NMR (Scheme 5). The 3,4-dimethoxy substrate 2v gave trichlorides 5v and 6v in a combined 67% yield, or in 91% yield based on the loading of 1b (Scheme 5). These anomalous outcomes were again rationalized as resulting from the stabi-  lization of radical intermediates gained upon methoxy substitution [51], which permitted further chlorination of either the methyl or arene groups.
To gain insight into the reaction mechanism we carried out two key control experiments. First, to test for rearrangement processes that might not be elucidated through product analysis alone, deuterated biphenylallene [D 2 ]-2b was subjected to the standard reaction conditions, and [D 2 ]-3b (E/Z = 1:2) was obtained in 79% yield (Scheme 6 top, also see Supporting Information File 1). As there was no indication of deuterium scrambling observable by 1 H or 2 H NMR of the product mixture, it appeared that 1,2-phenyl shifts or other rearrangement processes were not involved in the reaction. A further reaction was carried out in the presence of the radical scavenger TEMPO (1.5 equiv), from which only a trace of 3b was recovered, along with 50% of 2b (Scheme 6 bottom). As the chlorination reactivity was suppressed, our hypothesis that these reactions involved radical intermediates was further supported.
When allene chlorination was carried out with 1a, the observed product distributions were consistent with the results previously obtained, suggesting that ionic processes were operative. Furthermore, since no evidence of propargyl chlorides or α-dichloromethylstyrenes were observed, it appears the chlori- nation of allenes with 1a proceeded without interruption of 1,2phenyl shifts or iodane elimination, resulting in a reactivity pattern that differs from the related reagents TolIF 2 , PhI(OH)OTs or PhI(NTs 2 ) 2 (Scheme 1b and c). With 1b, however, the reactions were entirely selective for 2,3-dichlorination of the allene, which was consistent with the regiochemical outcome of reactions involving a trifluoromethyl radical (Scheme 1a). This, coupled with the results of Scheme 6, led us to propose a radical mechanism that was initiated by homolytic cleavage of the I-Cl bond of 1b at elevated temperature ( Figure 1) [50]. Addition of the chlorine atom to the allene central carbon resulted in the highly stabilized radical intermediate E, which then abstracted a chlorine atom from a second equivalent of 1b, giving dichlorides 3. Or, in the case of α-alkyl groups, intermediate E was also subject to a competing hydrogen abstraction pathway, resulting in mixtures of 3 and chlorodienes 4.

Conclusion
In conclusion, we report here an efficient new process for the chlorination of substituted phenylallene derivatives using the hypervalent iodine reagent 1-chloro-1,2-benziodoxol-3-one (1b). The reactions disclosed here represent the first report of a regioselective chlorination of phenylallenes, in which the 2,3allene olefin undergoes selective vicinal dichlorination. Overall, the reactions were mild and operationally-simple, tolerant to a variety of different functional groups, and provided the products in typically good yield. The selectivity of the reaction is presumably derived from it being a radical, not ionic, process, which also enabled the formation of chlorodiene products with α-alkyl substituted allenes. This reaction offers a new strategy for accessing dichlorinated functional group building blocks not readily accessible with other reagents, and our continued work in this area will be disclosed in due course.

Supporting Information
Supporting Information File 1 Experimental and characterization details, and NMR spectra of compounds.