HBF4- and AgBF4-Catalyzed ortho-Alkylation of Diarylamines and Phenols.

A silver-tetrafluoroborate- or HBF4-catalyzed ortho-alkylation reaction of phenols and diarylamines with styrenes has been explored. A broad substrate scope is presented as well as mechanistic experiments and discussion.

Modern hydroarylation methods are increasingly popular for the construction of C–C bonds. Indeed, some elegant strategies have recently appeared that allow excellent Markovnikov or anti-Markovnikov regioselectivity and broad functional group tolerance.[1] In 1999, Beller et al. reported the case of a Rh(I)/HBF4 cocatalyzed system for the ortho-alkylation of primary electron-rich anilines with styrene. For the most electron-rich anilines (pKa of the corresponding ammonium >5), it was even found that the reaction could proceed without the Rh catalyst (Scheme , eq 1).[2] This brought us to wonder what it would take to bring this very simple HBF4-catalyzed hydroarylation system to both lower reaction temperatures and especially to broader and less reactive substrate classes (lower basicity of the substrate; pKa of the corresponding ammonium <2). With phenols, for example,[3a−3c] elegant methods were very recently reported by Caputo[3a] and independently by Li,[3b] which demonstrate the use of a powerful and increasingly popular Lewis acid catalyst, tris(perfluorophenyl)borane (Scheme , eq 2). We therefore contemplated whether a redox approach might provide a superior strategy, in particular, in terms of the ortho selectivity, a persistent problem. We thus turned our attention to Ag(I) salts as prospective catalysts.[4] We considered, in particular, AgBF4[4] for poorly O- or N-basic phenol and diarylamine substrates. Indeed, we anticipated that radical mechanisms[5] might improve the reactivity and regioselectivity while providing a cheaper and operationally simpler synthetic method compared with perfluoro organo-boron Lewis acidic catalysts (Scheme , eq 3). In parallel, we also re-explored Beller’s control HBF4-catalyzed approach, without the rhodium catalyst, to evaluate the impact of the redox-active Ag(I) component. To our surprise, and in contrast with the literature,[2a] we found that the considerably cheaper HBF4 catalyst (Scheme , eq 3) also performs admirably well in the catalytic alkylation of anilines and phenols, with only small differences. This study is therefore focused on both AgBF4 and HBF4 catalysts and on related mechanistic considerations.

Scheme 1

Introduction

Phenothiazine was selected as a first convenient nonbasic diarylamine test substrate, a compound known to easily undergo radical oxidation.[6] Phenothiazines are, moreover, interesting scaffolds in some fields of organic materials[7] as well as essential bioactive compounds.[8] Some optimization elements are shown in Table . (See the SI for other parameters such as solvent and temperature.) Importantly, it was found that the reaction proceeds well in a number of very diverse conditions, whether potentially radical (Table , entry 1), Brønsted-acid-catalyzed (entry 23), or Lewis-acid-catalyzed (entry 24). For the phenothiazine test substrate, the AgBF4 catalyst (entry 1) delivered the highest yield of desired (monoalkylated) product. We moreover screened numerous counterions (Table , entries 4–14), thereby demonstrating the clear superiority of the tetrafluoroborate anion.

Table 1

Reaction Optimizationa

	catalyst	loading	1a/2a (mmol)	yield (%)a
1b	AgBF4	10 mol %	0.5/0.75	90 (84)
2	AgBF4	5 mol %	0.5/0.75	77
3	NaBF4	10 mol %	0.5/0.75	0
4	Ag2O	10 mol %	0.5/0.75	0
5	AgNO3	10 mol %	0.5/0.75	trace
6	AgOAc	10 mol %	0.5/0.75	0
7	AgF	10 mol %	0.5/0.75	trace
8	AgCl	10 mol %	0.5/0.75	0
9	AgBr	10 mol %	0.5/0.75	0
10	AgI	10 mol %	0.5/0.75	0
11	AgOTf	10 mol %	0.5/0.75	54
12	AgSbF6	10 mol %	0.5/0.75	64 (63)
13	AgSbF6	5 mol %	0.5/0.75	57
14	AgSbF6	10 mol %	0.5/1.00	54
15	CuCl2	10 mol %	0.5/1.00	0
16	AuCl3	10 mol %	0.5/1.00	8
17	PPh3AuCl	10 mol %	0.5/1.00	0
18	AgBF4	10 mol %	0.5/0.5	46
19	AgBF4	10 mol %	0.5/1	70
20	AgBF4	10 mol %	0.75/0.5	55
21	AgBF4	10 mol %	1/0.5	80
22	AgBF4	10 mol %	3/0.5	82
23c	HBF4Et2O	20 mol %	0.5/0.75	65
24d	PPh3AuX	10 mol %	0.5/0.75	48

Yields were determined by GC using n-dodecane as the standard (isolated yield in parentheses).

+15% of a mixture of bis-alkylated products.

+31% of a mixture of bis-alkylated products.

X = [N(CF3SO2)2].

With the AgBF4-catalyzed optimized conditions in hand (Table , entry 1), we then screened a number of phenothiazines and styrenes (Scheme ). Interestingly, the branched (Markovnikov) ortho (C1) alkylated product is typically by far the major product. In some cases, small amounts of bis-alkylated products can be observed (i.e., Table , entry 1); however, the first alkylation step seems to consistently occur in the ortho position to the X–H functional group (Scheme ). This is moreover a synthetically interesting regioselectivity outcome in light of the usual preference of phenothiazine for C3-(para-) electrophilic aromatic substitution.[9] This strong preference for the ortho-branched alkylated product is in good agreement with the concerted mechanism of Scheme . Even 1,1-and 1,2-disubstituted styrenes were found to be competent hydroarylation substrates, albeit in lower yields (3i, 43%; 3j, 38%). Acrylates, however, or heterocyclic olefins such as vinylpyridines, did not afford any hydroarylation product (Scheme ).

Scheme 2

Phenothiazine Scope, Isolated Yields

With this first set of phenothiazine examples in hand, we wondered whether noncyclic diarylamines and phenols (all with lower basicity than the primary anilines of Beller)[2] would also be applicable. Diarylamines and phenols are less easily protonated or oxidized than phenothiazines, however, necessarily implying higher activation energies and potentially shorter-lived radical intermediates. Fortunately, simply increasing the reaction temperature to, respectively, 80 and 100 °C allowed the hydroarylation reaction to proceed under otherwise altered starting material ratios. Elements of the substrate scope are presented in Schemes and 4, again with very high ortho-alkylation selectivity.

Scheme 3

Diarylamine Scope, Isolated Yields

Numbers in black are the yields with 10 mol % AgBF4; numbers in red are the yields with 20 mol % HBF4.

Scheme 4

Phenol Scope, Isolated Yields

Numbers in black are the yields with 10 mol % AgBF4; numbers in red are the yields with 20 mol % HBF4.

There, too, we could not find or identify any para-monoalkylated byproducts. In the case of product 5a, much of the excess of the diarylamine substrate 4a could be recovered and reisolated (1.97 mmol; see the SI), which seems to be a general trend when examining the various crude products presented herein. In contrast, none of the limiting coupling partners is ever reisolated, indicating the full conversion and probable decomposition of the missing mass balance. Importantly, we noted a superior isolated yield with the simple Brønsted HBF4 catalyst in almost all diarylamine cases (Scheme , red yields in parentheses).

We then performed a series of mechanistic experiments to probe some of the possible scenarios, in particular, with the ambiguous AgBF4 catalyst. First, N-methyl-phenothiazine does not provide any hydroarylated product (Scheme , eq 4), thus confirming the requirement for a heteroproton ortho to the functionalized C–H bond. This is strong evidence that the concerted protonation/C–C bond-formation hypothesis postulated by Beller (Scheme ) is probably also important with the AgBF4 catalyst. Second, the presence of TEMPO, a typical radical scavenger, does not allow the reaction to proceed (eq 5). TEMPO might either inhibit radical chains or alternatively reduce the Ag(I) catalyst toward the piperidinium-2,2,6,6-tetramethyl-1-oxo-tetrafluoroborate salt, which would, in turn, no longer be a competent oxidant for initiating the catalytic cycle. Furthermore, labeled phenol-d6 was engaged in the hydroarylation reaction, yielding a 25% D-enriched branched methyl group in the coupling product (eq 6). This corresponds to a 76% deuteron transfer efficiency and therefore also supports the ortho-concerted mechanism of Scheme . It could be noted that the deviation from the theoretical 33% deuterium content at the methyl group (full deuteron transfer efficiency) may come from either the integration approximation of the corresponding 1H NMR experimental profile or traces of water contamination in some of the components, which might lead to rapid OD/OH scrambling. We then compared the initial reaction rate between labeled phenol-d6 and natural abundance phenol in a competition experiment, yielding an initial kinetic isotope effect (KIE) of 1.4 (eq 7). This may indicate that C–H bond cleavage is not rate-limiting, in contrast with the prior concerted C–C bond-formation step. Moreover, interestingly, when measuring the initial KIE between phenol and phenol-d6 in two parallel reactions, a somewhat higher KIE of 2.4 was observed under otherwise identical conditions. This suggests that the cyclic concerted C–C bond-forming step and proton/deuteron oxygen-to-carbon transfer may indeed be rate-significant. Finally, to probe the suspected radical character of the AgBF4-catalyzed reaction, we performed a final control experiment in which the speculated catalytic electron hole is generated by a nonmetallic single electron oxidant (eq 8). For this purpose, we selected the NOBF4 salt as the nonmetallic catalytic electron hole generator because it possesses the same counterion as our AgBF4 precatalyst and because it is reputed to possess a similar (slightly superior) redox potential as well.[10]

Scheme 5

Mechanistic Experiments, Isolated Yields

To our surprise, when we indeed replaced the catalytic AgBF4 salt with the same catalytic amount of NOBF4 salt (10 mol %) in the alkylation of diphenylamine under otherwise unaltered reaction conditions (Scheme ), we isolated almost exactly the same amount of hydroarylated product 5a (65 vs 66%, respectively, eq 8). This result, in combination with the TEMPO poisoning experiment of eq 5, indicates that an electron-hole-catalyzed pathway is possible in the case of AgBF4. This is moreover in line with the usual observation of shiny Ag0 particles in suspension in the crude product mixtures. The fact that HBF4 and a cationic gold species are also competent catalysts (Table , entries 23 and 24) nevertheless suggests that the various mechanistic scenarios considered herein are not necessarily mutually exclusive,[11] especially if partial in situ hydrolysis of the AgBF4 would take place to generate active HBF4. These scenarios are summarized in Scheme .

Scheme 6

Possible ortho-Selective Transition States

Finally, to demonstrate the utility of the reaction with the cheapest herein studied catalyst, HBF4, we scaled up the synthesis of new compound 5a on a multigram level. We were satisfied to obtain 3.03 g of product 5a in a single batch (74%, Scheme ).

Scheme 7

Scale-Up of 5a, Isolated Yield

In conclusion, we have developed a AgBF4- and HBF4-catalyzed alkylation method of phenothiazines, diarylamines, and phenols. These methods allow the alkylation of considerably less basic anilines and phenols compared with previous methods,[2] with moreover excellent ortho-selectivity. Several mechanistic pathways were identified depending on the reaction conditions: Brønsted acid catalysis, Lewis acid catalysis, and also electron hole catalysis. The proximal XH functional group was found to be essential for reactivity and ortho regioselectivity through a characteristic concerted protonation/C–C bond-formation pathway. The herein presented reactivity elements are expected to complement the hydroarylation/alkylation toolbox.