Organo-Redox-Catalysis for the difunctionalization of alkenes and oxidative Ritter reactions by C-H functionalization

Transition metals are the dominant catalysts for redox-reactions between peroxides and organic substrates. Here, we show that triarylamines can act as organic redox-catalysts, enabling oxidative difunctionalization reactions of alkenes and oxidative Ritter-reactions. Styrene derivatives can be functionalized with alkyl radicals, generated from plain and halogenated hydrocarbons, and with nucleophiles, including nitriles, acetic acid, alcohols and fluoride. An oxidative Ritter reaction can be conducted between allylic C-H bonds as well as fluorene and acetonitrile. Benzoyl peroxide is the oxidant in both reactions. Mechanistic studies suggest that the triarylamines are catalysts and not initiators, mediating the reaction by electron transfer to the peroxide, forming benzoyloxyl radicals, and from C-radical intermediates, forming carbocations.


Introduction
The difunctionalization of alkenes is a powerful method for the construction of C-C and C-X bonds. [1][2][3][4] A very interesting type amongst those is the successional addition of a radical and a nucleophile. 1,3 This method enables functionalizing olefins with a wide variety of reagents in a selective manner, given that radicals and nucleophiles generally react complementarily. However, these reactions are still lacking a truly broad substrate scope, spurring our efforts of further research. A widely used strategy for generating radicals from simple substrates is C-H bond cleavage by hydrogen atom transfer (HAT) to oxyl radicals, which are readily generated from peroxides. 5 The subsequent addition of a nucleophile requires an electron transfer (ET) step to convert the radical intermediate into a carbocation. Thus, transition metals are widely used as redox-catalysts in such reactions, as they can mediate peroxide O-O bond cleavage and subsequent ET (Scheme 1a). 1,3,[6][7][8][9] Alternative methods for the consecutive addition of radicals and nucleophiles utilize organic photocatalysts, 10,11 hypervalent iodine reagents or iodide as catalyst [12][13][14] and electrochemistry. 15, 16 We are not aware, however, of the use of an organo-redox catalyst independent of irradiation in such reactions. Here, we report the use of triarylamines as catalysts in the activation of peroxides for synthetic radical reactions (Scheme 1b). Triarylamines can form stable ammoniumyl radical cation salts by ET, and variation of the aryl-substituents allows for finetuning of their properties. [17][18][19][20][21] Both the amines and the radical cations are widely applied in electro-optical materials. [18][19][20]22 The radical cations can be used as stoichiometric single-electron oxidants in chemical reactions, 23 or in substoichiometric amounts as initiators of radical chain reactions. 17,24 Both the amines and the radical cations are also utilized as redoxcatalysts in electrochemical 25,26 or photochemical reactions, 27,28 as well as in aerobic oxidations. 29,30 Despite this plethora of applications, we are not aware of amine-based redox catalysis in the activation of peroxides, which would open many opportunities for synthetic applications.
N,N-Dialkylanilines like 1 are well-known to generate radicals from diacylperoxides, especially benzoyl peroxide (BPO, Scheme 2a). 31,32 The reaction is irreversible due to the reactivity of the ammoniumyl radical cation 2, which readily forms a Cradical 3, an iminium ion 4 and other products derived thereof. 32,33 In contrast, we assumed that triarylamines A could be suitable candidates for catalysis, as the stable radical cation salts A+ could be regenerated by ET (Scheme 2b). We had previously utilized BPO in the addition of thioxanthene and similarly facile radical precursors together with nucleophiles to styrenes, which was rationalized as a radicalchain reaction. 34 The addition of hexafluorophosphoric acid (HPF 6 ) was found to modulate the redox potential of BPO, and the addition of N,N-dimethylanilines as initiators allowed difunctionalization with acetonitrile with moderate success. We kept working on finding a more efficient method for a broad substrate scope that would also avoid the use of a strong acid.

Results and discussion
We found that the addition of cyclohexane and acetonitrile to styrene (5a) took place in the presence of catalytic amounts of some triarylamines and NaPF 6 as an additive, forming the desired product 6a in good yields ( Table 1, entries 1-3). The most effective amine was 4-iodo-N, N-diphenylaniline (A1), closely followed by tris(4-methylphenyl)amine (A2) and tris(4bromophenyl)amine (A3), other amines were much less efficient. Without triarylamine, product 6a was not formed (entry 4) and the addition of NaPF 6 is indispensable (entry 5). A reduction in the product's yield was also seen with other additives and oxidants (see the Supporting Information for a detailed investigation). The product's structure supported the subsequent addition of a cyclohexyl radical and acetonitrile as a nucleophile in a Ritter reaction. With these reaction conditions, we investigated the product scope by testing other substrates. Using cyclopentane, cyclohexane, cycloheptane, and cyclooctane as radical precursors with acetonitrile as nucleophile afforded the products 6a-6d in good yields of 80-86% (Scheme 3).
Methylcyclohexane gave a mixture of regioisomers from which we could isolate 6e, the major one, in 34% yield. With n-hexane, a mixture of the isomeric products 6f-6h was isolated in 80% overall yield, from which we could isolate the isomer 6f in 4% yield as a pure compound by column chromatography. With 2methylpentane, the selectivity for the tertiary C-H bond was relatively high, allowing for isolation of the major isomer 6i in 41% yield. With 2,2,4,4-tetramethylpentane, only isomer 6j was isolated, apparently because the methylene group is sterically shielded, resulting in HAT from a primary C-H bond. The haloalkanes dichloromethane, chloroform, dibromomethane and bromoform could also be employed successfully in this reaction, producing the products 6k-6m. When using bromoform, not only 6n was formed by HAT, but also 6m by bromine atom transfer. 35,36 When only acetonitrile was used as solvent, 6o was isolated in 86%. Styrenes with various substituents on the aromatic ring afforded the desired products in generally good yields (Scheme 4). There is no clear electronic substituent effect on the product yields, also substitution in the ortho position was not detrimental (7l-7n). Only in the case of p-methoxystyrene, the desired product was only observed in traces (7b). 2-Methylstyrene and stilbene could also be employed, giving the expected products 7o and 7p in medium yields and as mixtures of diastereomers. Next, the scope with respect to nucleophiles was explored. As shown in Scheme 5a, different nitrile solvents and cyclohexane delivered the desired amides 8a-8e in good yields of 48%-76%. Similarly, when acetic acid was used, the corresponding acetate 8f was formed in 63% yield. With 1,1-diphenylethylene, difunctionalization with the α-cyanoalkyl radical from acetonitrile and alcohols was possible. With methanol and ethanol, the products 8g and 8h were isolated in 51% and 41% yields, respectively, however, long-chain alcohols showed a low reactivity. Very similar tertiary alcohols had recently been synthesized by copper-catalysis at higher temperature. 37 We also isolated the dimer 9 from unsuccessful tests of other nucleophiles, supporting the occurrence of radical intermediate 15. 38,39 In nitromethane as solvent, phenyl groups were incorporated into the products (10a-10c, Scheme 5b). These likely originated from phenyl radicals, formed by decarboxylation of the benzoyloxyl radicals. Acetonitrile and methanol could be used as nucleophiles, and with triethylamine hydrofluoride even fluoride, albeit in low yield (10c). The organo-redox system also proved to catalyze the oxidative Ritter reaction of allylic and benzylic C-H bonds; for these reactions, amine A3 was found to be superior to A1 (see the Supporting Information for details). As shown in Scheme 6a, cyclohexene and bicyclo[3.2.1]oct-2-ene delivered the amides 11a and 11b in 71% and 72% yield, respectively. When propionitrile was used, the corresponding product 11c was isolated in 33% yield. With (E)-4-octene, 11d and 11e were isolated in 89% yield as a mixture of two isomers with a ratio of 1 : 0.8. When (Z)-4-octene was used, the same products were isolated with a ratio of 1 : 3. (E)-5-Decene gave the E-isomers 11f and 11g in equimolar amounts. When we used 9H-fluorene, the benzylic methylene group was functionalized, providing the amide 13 in 70% yield (Scheme 6b). Are the triarylamines efficient initiators of a radical chain reaction, reacting irreversibly, or are they catalysts, achieving turnover? 40 As mentioned above, N,N-dialkylanilines are wellknown activators of benzoyl peroxide in radical chain reactions; 31 however, they proved to be inefficient for the reactions presented here (Scheme 7a). Also, the reaction did not proceed when employing the well know radical initiator azobisisobutyronitrile (AIBN), suggesting that a radical chain mechanism is not operating. We used two radical cation salts, [(A1+)SbF 6 − ] and [(A3+)SbCl 6 − ], in place of the corresponding triarylamines under otherwise unchanged reaction conditions (Scheme 7b). In both cases, comparable yields of 6a were achieved (89% and 84%, resp.). When A3+ was used, its dark blue colour disappeared and 10% of the reduced form A3 could be isolated, ruling out that the triarylamines are irreversibly consumed and supporting that both the amine and the oxidized radical cation salts are involved in a catalytic reaction. Frontiers   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  After completion of the reaction, the system remains active. When we added another batch of BPO and styrene to a reaction mixture forming 6a that had gone to completion, a further 85% of these added substrates were converted to 6a after another 30 hours. Also, a third batch of substrates could be converted to product, which further supports that the reaction is catalytic in nature (see the Supporting Information for details).

Page 4 of 8 Organic Chemistry
We also investigated the system by 1 H-NMR at the reaction temperature of 70°C ( Figure 1). Ca. 40% of BPO on its own had decomposed after 10 h, in line with the reported 10 hour halflife temperature of 73 °C (red bottom line). 41 Benzene and benzoic acid were formed in roughly equal amounts, indicating that half of the benzoyloxyl radicals had decarboxylated before they were quenched by HAT. Adding catalytic amounts (10 mol%) of A1 accelerated the decomposition of BPO during the first 2 hours and resulted in forming predominantly benzoic acid (blue upper line). These results indicate that the amine catalyses the decomposition of BPO by an effective ET reaction (see Scheme 7b below and its discussion for details), forming one molecule each of benzoate -observed as benzoic acid by NMR -and benzoyloxyl radical. The latter can decarboxylate to a phenyl radical and both radicals can react by HAT reactions from the medium to form benzoic acid and benzene, respectively. This explains the appearance of significantly more benzoic acid than benzene, compared to the thermal decomposition of BPO. During the difunctionalization reaction, a white precipitate formed, which we found to be sodium benzoate, in line with the suggested cleavage of BPO by ET (see the Supporting Information). All these results enable us to propose a mechanism for the formation of difunctionalization products like 6a (Scheme 8a) The peroxide bond of BPO is cleaved by an ET reaction from the triarylamine catalyst A, forming the radical cation A+, a benzoate anion and a benzoyloxyl radical. The presence of NaPF 6 likely helps stabilizing the radical cation A+ in the form of the salt [A1+]PF 6 − . The benzoyloxyl radical can engage with cyclohexane or other substrates in a HAT reaction, forming a carbon radical 14, which then adds to styrene, forming the benzylic radical 15. Oxidation by the ammoniumyl radical cation A+ regenerates the triarylamine A and forms the carbocation 16, which is attacked by nucleophiles to provide the final product. To some extent, the benzoyloxyl radicals decarboxylate, generating phenyl radicals which can either participate in HAT reactions, too, or add to styrene, as was shown in Scheme 5b above. Whether the reaction between the triarylamine A and BPO proceeds by an outer-sphere ET, directly forming benzoate and two radicals, or by an S N 2-reaction via N-benzoyloxylammonium salt 17, which in a second step decomposes homolytically into the same products (Scheme 8b), is at present unclear. Both pathways have been suggested for reactions between diacylperoxides and amines, 32,42-44 but the combination of BPO with the triarylamines used in this study has not been investigated yet.
Additionally, we studied the redox potentials of several triarylamines and the radical cation A1+ as well as BPO by cylic voltammetry (see the Supporting Information). The reduction potential of A1 was indeed lower than that of BPO, supporting that A1 can reduce BPO by ET. However, BPO has the higher oxidation potential of the two, suggesting that it is the better electron acceptor. However, we consider these results of separate measurements as not fully conclusive for the interpretation of the mechanism, since they are not in agreement with the aforementioned results supporting catalysis by the amine. Furthermore, they do not take potential interactions in the reaction mixture into account. For example, the addition of NaPF 6 is crucial for the reaction to occur, which indicates an ionic interaction that might shift the redox potential of the ammoniumyl radical cation. 34 A species like 17 could be involved as electron acceptor, or the radical cation A1+ could be transformed into the actual catalyst in situ by attack at its free para-positions. 21 Thus, we acknowledge that not all details of the present reactions are understood and we are therefore planning more detailed investigations.

Conclusions
In summary, we have established triarylamines as organo-redox catalysts for oxidative C-H functionalization reactions, with piodophenyl diphenylamine as the catalyst of choice in the newly developed method. By using benzoyl peroxide as oxidant, the difunctionalization of styrenes could be accomplished with radicals generated from hydrocarbons by hydrogen atom transfer and with nucleophiles, including nitriles, alcohols, acetic acid and fluoride. Besides, the amination of allylic and benzylic C-H bonds is also achieved under the same reaction conditions. The method does not require irradiation, electrolysis, transition metals nor significantly elevated temperatures. This application of a relatively simple amine might pave the way for further developments of organo-redoxcatalysts, which may thus become another established class amongst organocatalysts. 45