A Photocatalytic System Composed of Benzimidazolium Aryloxide and Tetramethylpiperidine 1-Oxyl to Promote Desulfonylative α-Oxyamination Reactions of α-Sulfonylketones.

A new photocatalytic system was developed for carrying out desulfonylative α-oxyamination reactions of α-sulfonylketones in which α-ketoalkyl radicals are generated. The catalytic system is composed of benzimidazolium aryloxide betaines (BI+-ArO-), serving as visible light-absorbing electron donor photocatalysts, and 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO), playing dual roles as an electron donor for catalyst recycling and a reagent to capture the generated radical intermediates. Information about the detailed nature of BI+-ArO- and the photocatalytic processes with TEMPO was gained using absorption spectroscopy, electrochemical measurements, and density functional theory calculations.

Introduction

Aminoxyl radicals are compounds possessing the structure R2N–O• and are well-known to serve as effective radical traps.[1] 2,2,6,6-Tetramethylpiperidine 1-oxyl (TEMPO), the best-known aminoxyl radical,[2] has been employed in various fields of chemistry owing to its high stability and favorable chemical properties including good radical trapping ability and reversible redox reactivity.[3,4] Nearly a decade ago, photoinduced electron transfer (PET) processes promoted by visible light-absorbing transition metal redox catalysts (often called photoredox catalysts) were demonstrated to have potential applicability in synthetic organic chemistry.[5] Since then, numerous synthetically useful photoredox reactions catalyzed by metal complexes of this type as well as non-metal catalysts have been uncovered.[6] In these photoredox-catalyzed processes, TEMPO has been frequently employed as a redox mediator as well as a radical trapping agent.[3f,4d,4e,4h,4i,4k,4m−4o]

Coupling of α-carbonyl alkyl radicals and TEMPO is the way to achieve α-oxyamination of carbonyl compounds. In principle, these radical intermediates can be generated through single electron transfer (SET) of appropriate precursors and then be trapped by TEMPO. Two examples of these approaches are oxidative fragmentation of enolates and reductive cleavage of carbon–heteroatom bonds adjacent to carbonyls (Scheme ).[7] Indeed, various oxidative processes that utilize enolates or their equivalents (enols and enamines) and TEMPO, along with appropriate oxidants, have been investigated (above in Figure ).[8] Furthermore, coupling reactions in which enolates react with the oxidized TEMPO (N-oxoammonium, TEMPO+), added as a salt[9] or in situ generated,[8e,8l] have been described. Although these oxidative protocols are useful, they have some limitations. For example, enolate-stabilizing substituents are usually required in the carbonyl substrates and, if not, strong bases and low temperatures are needed. In addition, enamine substrates for these reactions are restricted to aldehydes mainly. In contrast, the applicability of reductive fragmentation reactions of α-heteroatom-substituted carbonyl substances to oxyamination reactions promoted by TEMPO (below in Figure ) has been explored to a lesser extent.[10]

Scheme 1

Generation of α-Carbonyl Alkyl Radicals by Single Electron Transfer Reactions of the Precursor Enolates and α-Heterosubstituted Carbonyl Precursors

Figure 1

α-Oxyamination of carbonyl compounds using TEMPO.

In earlier efforts, we investigated the reductive reactions of various organic substrates that are promoted by PET using aryl- and hydroxylaryl-substituted 1,3-dimethylbenzimidazolines (BIH–Ar and BIH–ArOH) as electron and hydrogen atom donor reagents.[11] Recently, we found that an α-TEMPO adduct was formed in the PET reaction of a α-sulfonylketone induced by 2-hydroxynaphthyl-1,3-dimethylbenzimidazoline (BIH–NapOH), which is utilized as a stoichiometric photo-reductant in the presence of TEMPO.[12] In addition, we demonstrated that benzimidazolium aryloxides (BI+–ArO–), which are oxidized forms of BIH–ArOH, serve as unprecedented betaine photocatalysts for reductive organic transformations.[13] Observations made in these studies motivated us to explore the possibility that these betaines can be utilized as photocatalysts in the presence of TEMPO to promote α-oxyamination reactions of carbonyl compounds possessing anionic leaving groups (LG) at α carbons (Scheme ). Because TEMPO has a relatively strong electron-donating ability in organic solvents (Eox = +0.63 to +0.68 V vs SCE),[14] it is expected to act not only as a radical trap but also as an electron donor to return the PET-oxidized form of the catalyst (BI+–ArO•) back to the starting betaine (BI+–ArO–) in the photocatalytic system. To assess the validity of this hypothesis, α-sulfonylketones were selected as probe substrates because they have been frequently utilized in PET-promoted reduction reactions.[12,13,15]

Scheme 2

Expected Photocatalytic Cycle for Oxyamination Reactions of α-Substituted Carbonyls Using Betaine Catalysts (BI+–ArO–) and TEMPO

In the study described below, we demonstrated that the cooperative photocatalytic system composed of benzimidazolium aryloxides (BI+–ArO–) 1 and TEMPO effectively photocatalyzes the α-oxyamination reactions of α-sulfonylketones 2 (Figure ).

Figure 2

Photocatalytic α-oxyamination reactions of α-sulfonylketones 2.

Results and Discussion

To assess their use as electron donor photocatalysts, benzimidazolium aryloxides 1a–e were subjected to absorption spectroscopic analysis, which demonstrated that there are absorption bands in the visible region except for 1e (Figure S1 and Table S1). In addition, density functional theory (DFT) calculations were conducted to gain information about their structures and frontier orbital energies and distributions (Figures S2 and S3). As has been discussed previously in the context of benzimidazolium naphthoxide (BI+–NapO–) 1a,[13a] the phenoxide oxygen in the related benzimidazolium o-phenoxide 1b is located sufficiently close to the positively charged benzimidazolium ring to enable a strong intramolecular electrostatic interaction (Figure ). The respective dihedral angles (DA) of benzimidazolium and aryloxide moieties of 1a and 1b of 62° and 54° are also similar (Figure S2). In contrast, benzimidazolium p-phenoxides 1c and 1d have smaller DAs (41–42°) and shorter lengths for the carbon–carbon bonds connecting the benzimidazolium and phenoxide moieties than do 1a and 1b (Figure S2). Thus, quinoid forms contribute to the structures of 1c and 1d, although complete coplanar alignment of the benzimidazolium and phenoxide rings is difficult to achieve because of the steric repulsion between N-methyl and phenoxide m-hydrogens (Figure ). On the other hand, such interactions between the benzimidazolium and phenoxide moieties are not possible in the m-phenoxide analog 1e.

Figure 3

Contributing structures of 1a, 1b, 1c, and 1d.

The results of DFT calculations show that the frontier orbitals of 1 are distributed over the entire molecule and that the LUMO and HOMO coefficients are slightly larger on the respective imidazolium and aryloxide moieties (Figure S3). It is notable that the HOMO coefficients are more greatly distributed over the aryloxide moieties in 1a and 1b as compared to those in 1c and 1d. On the other hand, the HOMO distributions of 1c and 1d are also consistent with the contribution of quinoid structures. To gain information about the nature of the electronic transitions of the betaines, TD-DFT calculations were performed (Figure S4, also Figures S10–S14 and Tables S6–S10). The results show that the calculated absorption spectra of 1a–1d match the experimental spectra with respect to the molar extinction coefficient (ε) and absorption maximum trends (compare Figure S4 to Figure S1), although the ε value of 1d is somewhat underestimated in the calculation results. The characteristic absorption at ca. 400 nm for each of the BI+–ArO– derivatives is dominated by a HOMO to LUMO transition (Figures S10–14 and Tables S6–S10). These findings are in accord with the observation that the intensities of the absorption bands of 1c and 1d are significantly greater than those of 1a and 1b (Figure S1) owing to the more effective HOMO–LUMO overlap in the former betaines. The observation that the HOMO–LUMO overlap is minimal for 1e is consistent with it, displaying no significant light absorption in the visible region.

The results described above suggest that the phenoxide moiety of 1b is stabilized by an electrostatic interaction with the positively charged benzimidazolium ring in a manner similar to the naphthoxide moiety in 1a.[13a] Indeed, the oxidation potential of 1b (E1/2ox = +0.61 V vs SCE), which is similar to that of 1a (E1/2ox = +0.64 V vs SCE), is much larger than that of an isolated phenoxide (e.g., 2-methylphenoxy anion, E1/2ox = +0.14 V vs SCE) (Figure S5). Also, because the extended conjugation over the whole molecule in the quinoid form stabilizes the HOMO, 1c and 1d have relatively larger oxidation potentials (E1/2ox = +0.56 V vs SCE for 1c and +0.36 V vs SCE for 1d) than that of simple phenoxide ions.[16] The oxidation potentials of the electronic-excited states of 1, estimated by using their ground state oxidation potentials and excitation energies, are calculated to be −2.08 to −2.65 V (Table S2).[17]

Although TEMPO, as a red-orange solid, absorbs light in the visible region (360–560 nm), its absorptivity is significantly lower (λmax = 461.5 nm, log ε = 1.08) than that of 1a (λmax = 415.0 nm, log ε = 3.92) (Figure S6). Thus, TEMPO should not significantly interfere with the visible light-promoted excitation of 1a. Previous investigations of the reactions photocatalyzed by 1a suggest that dimethyl sulfoxide (DMSO) is a suitable solvent to activate 1a by specific solvation.[13a] Based on these preliminary considerations, we explored the photocatalytic reactions of α-arylsulfonyl ketones 2 (Figures S7 and S8 and Table S3) promoted by benzimidazolium aryloxides 1 and TEMPO in DMSO. First, the photoreactions of α-phenylsulfonyl isobutyrophenone (2a) (E1/2red = −1.66 V vs SCE) using a catalytic quantity of 1a (10 mol %) and TEMPO in DMSO were performed (Table ). Irradiation (4 h, Xe lamp) of a solution of 2a and 1a in the presence of molar equivalent quantities of TEMPO and 2a resulted in 65% consumption of 2a and 50% formation of the expected α-TEMPO adduct 3a (entry 1). Prolonged irradiation for 14 h did not influence the reaction progress and the yield of adduct formation (entry 2). On the other hand, the presence of one more equivalent of TEMPO in the solution irradiated for 4 h led to a slightly increased yield of 3a based on the conversion of 2a (entry 3). Notably, 14 h irradiation of the same solution resulted in complete consumption of 2a and excellent production of 3a (entry 4). In the absence of irradiation or 1a, 2a was only slightly consumed and 3a was not generated (entries 5 and 6). Finally, the reaction proceeds in other solvents such as N,N-dimethylformamide (DMF), MeCN, and CH2Cl2 while DMSO and DMF are most suitable (Table S4).

Table 1

Desulfonylative α-Oxyamination Reaction of α-Sulfonylketone 2a with TEMPO Photo-Catalyzed by Benzimidazolium Naphthoxide 1a in DMSOa

entry	TEMPO (equiv)	irradiation time (h)	conv of 2a (%)b	yield of 3a (%)b,c
1	1.0	4	65	50 (77)
2	1.0	14	65	49 (75)
3	2.0	4	61	51 (84)
4	2.0	14	100	92
5d	2.0	14	6	0
6e	2.0	14	4	0

2a (0.10 mmol), 1a (10 mol %), DMSO (1.0 mL); 500 W Xe lamp (λ > 390 nm).

Determined by using 1H NMR.

Yields in parentheses are based on conversions.

Stirred in the dark.

In the absence of 1a.

It is interesting to note for comparison purposes that α-bromo isobutyrophenone (4) (E1/2red = −1.15 V vs SCE)[18] reacts with a significantly lower efficiency under the ideal conditions used for its sulfonyl analog 2a (entry 4 of Table ) and produces 3a in a much lower yield although 4 has greater electron-accepting ability than 2a (eq ).

The photo-desulfonylative α-oxyamination reactions of β-phenyl-α-phenylsulfonyl propiophenone (2b) (E1/2red = −1.48 V vs SCE),[13a] which possesses a greater electron-accepting ability than 2a, were explored using various photocatalysts and light sources (Table ). Under the ideal irradiation conditions given in entry 4 of Table , 2b was also an effective substrate in that 14 h irradiation with a Xe lamp led to 100% conversion and a 94% yield of the corresponding oxyamination product 3b (entry 1, Table ). Similar to the reaction of 2a in the presence of 1 equiv of TEMPO, 60% conversion of 2b and 52% yield of 3b (high yielding based on the conversion of 2b) were observed (entry 2). Household light-emitting diodes (LEDs) can also be used as light sources although longer irradiation times are required (entries 3–7). Photocatalysts 1b, 1c, and 1d also promote the reaction of 2b although less effectively than does 1a (compare entry 1 to entries 8, 9, and 10). On the other hand, 1e was not effective at all (entry 11). Neither were eosin Y[6e] nor Ru(bpy)22+ effective,[19] which are a well-known organic photocatalyst and transition metal catalyst, respectively. The non-compatibility of these substances with the TEMPO-based photocatalytic protocol is likely a consequence of the inability of the expected reductant species derived from these photocatalysts to efficiently reduce 2b (entries 12 and 13).[20]

Table 2

Desulfonylative α-Oxyamination Reactions of α-Sulfonylketone 2b with TEMPO Photo-Catalyzed by Various Photo-Catalysts (PC) in DMSOa

entry	photocatalyst (PC)	light source	irradiation time (h)	conv of 2b (%)b	yield of 3b (%)b,c
1	1a	Xe	14	100	96
2d	1a	Xe	14	60	52 (87)
3	1a	2 × LED-1	20	100	95 [92]e
4	1a	LED-1	24	79	76 (96)
5	1a	LED-1	48	100	96
6	1a	LED-2	48	84	83 (99)
7	1a	LED-2	72	100	95
8	1b	Xe	14	69	68 (99)
9	1c	Xe	14	22	16 (73)
10	1d	Xe	14	22	15 (68)
11	1e	Xe	14	0	0
12	eosin Y	Xe	14	5	0
13	Ru(bpy)3Cl2	Xe	14	10	trace

2b (0.10 mmol), PC: 1 (10 mol %), eosin Y (10 mol %), Ru(bpy)3Cl2 (1 mol %), TEMPO (2.0 equiv). DMSO (1.0 mL); 500 W Xe lamp (λ > 390 nm) (Xe), 10.8 W white LED (LED-1), 7.3 W white LED (LED-2).

Determined by using 1H NMR.

Yields in parentheses are based on conversions.

TEMPO (1.0 equiv vs 2b).

Isolated by using column chromatography.

Recently, we developed a new photocatalytic protocol in which a catalytic quantity of the reduced form of 1a (1a-HH) is used instead of 1a along with a stoichiometric quantity of cooperate materials.[13b] We explored the use of this protocol to promote the desulfonylative TEMPO addition reaction (Table ). When a solution of 2b containing 1a-HH and TEMPO was irradiated using the conditions shown in entry 1 of Table , 3b was formed with a good mass balance although a considerable amount of 2b was recovered (entry 1). The presence of t-BuOK was found to significantly enhance the reaction progress (entry 2). The observation resembles those made previously in the studies of reactions using 1a-HH as the photoreductant[11e] and 1a as the photocatalyst[13b] along with t-BuOK. Under the new conditions, prolonged irradiation led to complete consumption of 2b and highly efficient (97%) production of 3b (entry 3).

Table 3

Desulfonylative α-Oxyamination of α-Sulfonylketone 2b with TEMPO Photo-Catalyzed by Hydroxynaphthyl Benzimidazoline 1a-HH in DMSOa

entry	t-BuOK (equiv)	irradiation time (h)	conv of 2b (%)b	yield of 3b (%)b,c
1		14	30	27 (90)
2	0.5	14	68	57 (84)
3	0.5	24	100	97

2b (0.10 mmol), 1a-HH (10 mol %), TEMPO (2.0 equiv), DMSO (1.0 mL); 500 W Xe lamp (λ > 390 nm).

Determined by using 1H NMR.

Yields in parentheses are based on conversions.

To obtain information about the mechanistic pathway of the photocatalytic reaction between 1a and 2b, the process was first carried out in the absence of TEMPO (Table ). Although 2b was consumed to some extent, no identifiable products were produced (entry 1). On the other hand, when 1,4-cyclohexadiene (CHD) was present, the photoreaction generated desulfonylated ketone 5b in excellent yield based on 56% conversion of 2b (entry 2). This observation suggests that CHD reacts as a hydrogen atom donor with the α-ketoalkyl radical intermediate derived by the loss of a phenylsulfinate anion (PhSO2–) from the radical anion of 2b.[12] In addition, PhSO2– has a strong electron-donating ability based on the oxidation potential of PhSO2Na·2H2O (E1/2ox = +0.57 V vs SCE, see Figure S9 and Table S5).[21] Indeed, the presence of PhSO2Na·2H2O in the irradiated solution containing CHD caused the complete consumption of 2b along with a good yield of 5b (entry 3).

Table 4

Desulfonylation Reaction of α-Sulfonylketone 2b Photocatalyzed by Benzimidazolium Naphthoxide 1a with Additives in the Absence of TEMPO in DMSOa

entry	additive (equiv)		conv of 2b (%)b	yield of 5b (%)b,c
1			46	0
2	CHD (3.0)		56	53 (93)
3	CHD (3.0)	PhSO2Na (2.0)	100	71

2b (0.10 mmol), 1a (10 mol %), DMSO (1.0 mL), 1,4-cyclohexadine (CHD), PhSO2Na·2H2O; 500 W Xe lamp (λ > 390 nm).

Determined by using 1H NMR.

Yield in parentheses is based on conversion.

The observations described above and the results of previous studies[13a] suggest that the mechanistic pathway shown in Scheme is plausible for oxyamination reactions of sulfonylketones 2 photocatalyzed using the 1a and TEMPO couple. In the process, photoexcited 1a, [1a]*, donates a single electron to 2 to give a radical cation 1a and radical anion 2. The radical ions generated in this way follow the respective single electron reduction and fragmentation (−PhSO2–) pathways. As described above, more than a stoichiometric quantity of TEMPO is required to bring about complete conversion of 2 (see Tables and 2) as a result of the fact that it acts both as a radical trapping agent in the formation of 3 from α-ketoalkyl radicals 6 and as an oxidant (E° = +0.68 V vs SCE, see Figure S9 and Table S5) to reduce 1a to reform 1a although this process would be slightly endergonic. Oxidation potential considerations and the observed acceleration effect of PhSO2Na (see Table ) suggest the possibility that PhSO2– liberated from 2 also serves as a reductant for 1a. However, the concentration of the liberated PhSO2– would be lower than that of TEMPO, particularly at the early stages of the reaction. Thus, TEMPO is a more effective cooperative reductant than PhSO2– in the photocatalytic cycle. It is noteworthy that the bromide ion liberated by fragmentation of the anion radical of bromoketone 4 (see eq ) cannot serve the same role because its oxidation potential (E°ox = +1.11 V vs SCE) is excessively high.[22] Indeed, 4 is less reactive than 2 under the same photocatalytic conditions although it has a greater electron-accepting ability. Finally, the catalyst ability of 1a is compared with those of 1b, 1c, 1d, and 1e. The reducing power of photocatalysts 1b–1e estimated by their excited state oxidation potentials is greater than that of 1a (Table S2). The absorptivities of 1c and 1d are significantly large, while that of 1e is extremely small (Figure S1). On the other hand, the oxidation potentials of these catalysts are lower than that of 1a (Table S2), which suggests that the radical cations of 1b–1e less efficiently return to their neutral forms by SET with TEMPO. Consequently, the catalytic efficiencies of 1b–1e would be lower than that of 1a, which was indeed witnessed (see Table ).

Scheme 3

Plausible Mechanism of Photo-Desulfonylative α-Oxyamination Reactions of α-Sulfonylketones 2 with TEMPO Catalyzed by Benzimidazolium Naphthoxide 1a

The scope of the developed photocatalytic oxyamination process was briefly explored by utilizing other α-sulfonylketones 2 (see Figure S8 and Table S3) and LED irradiation (Table ). Isobutyrophenone derivatives 2c (E1/2red = −1.68 V vs SCE) and 2e–2g (E1/2red = −1.54 to −1.78 V vs SCE) and propiophenone derivatives 2h–2j (E1/2red = −1.38 to −1.53 V vs SCE) were found to be effectively converted to the corresponding TEMPO-adducts 3 in high yields (also see entry 3 of Table : 3b from 2b). As previously discussed,[12] desulfonylation of the radical anion of 2d (E1/2red = −1.74 V vs SCE) proceeds more slowly than that of 2c because the p-methoxyphenylsulfonyl anion is a less effective leaving group than the p-tosyl anion. This fact is consistent with the observation that the formation of 3a from 2c is more efficient than it is from 2d. Unexpectedly, DMSO was found not to be a suitable solvent for the reactions of indanone 2k (E1/2red = −1.52 V vs SCE) and tetralone 2l (E1/2red = −1.50 V vs SCE), and thus it was replaced by DMF to carry the processes that produced 3k and 3l, respectively, in modest to good yields. In the reaction of acetophenone derivative 2m (E1/2red = −1.26 V vs SCE), the formation of the expected TEMPO-adduct 3m was accompanied by a small amount of the bis-α-TEMPO adduct 7. Moreover, under typical conditions, 4-phenyl-3-phenylsulfonyl-2-butanone 8 (E1/2red = −1.89 V vs SCE) possessing less electron-accepting property was inert, and it did not produce an expected TEMPO adduct 9. Addition of t-BuOK as an activator (as described above) did not assist the reaction of 8 at all.

Table 5

Desulfonylative α-Oxyamination Reactions of α-Sulfonylketones 2 and 8 with TEMPO Photocatalyzed by Benzimidazolium Naphthoxide 1a in DMSOa

2 or 8 (0.10 mmol), 1a (10 mol %), TEMPO (2.0 equiv), DMSO (1.0 mL); 10.8 W white LED × 2 for 20 h or for 40 h. Yields of 3, 7, and 9 were determined by using 1H NMR.

Isolated by using column chromatography.

DMF (1.0 mL) was used instead of DMSO.

Protonation is among the most general reaction pathways of carbonyl radical anions (ketyl radicals) in redox reactions.[7b,23] Because intramolecular hydrogen bonding exists between the carbonyl oxygen and phenolic hydroxy substituent of 2j (1H NMR δ = 11.78), the electron-accepting ability of carbonyl should be increased. As expected, the reduction potential (E1/2red = −1.38 V) of 2j was found to be smaller than those of 2h (−1.53 V) and 2i (−1.49 V) (see Figure S8 and Table S3). Thus, a mechanism involving sequential SET and intramolecular proton transfer (intra-PT) from the phenolic hydroxy to the formed ketyl radical intermediate is hypothesized for the reaction of 2j (Scheme ). Another possible reaction of 2j to take place would be the proton-coupled electron transfer.[24] These pathways generate the rearranged ketyl radical of 2j (rear-2j). If rear-2j releases the sulfonyl radical, the formed anion 10 reacts with the TEMPO cation derived from TEMPO (also see Scheme ) to produce the TEMPO adduct 11 (path a). On the other hand, if rear-2j undergoes elimination of the sulfinate anion, the resulting radical 12 is subsequently captured by TEMPO to give 11 (path b, also see Scheme ). Tautomerization of 11 gives the observed 3j. If path a is operated, 1 equiv of TEMPO vs 2j should be enough to complete the reaction. However, when the 1a-photocatalyzed reaction of 2j using an equimolar TEMPO was performed, 50% formation of 3j was observed at 53% conversion of 2j. This observation is similar to those of 2a (entry 2 of Table ) and 2b (entry 2 of Table ) using 1 equiv of TEMPO. Consequently, path b is considered to be more plausible than path a for the reaction of 2j.

Scheme 4

Hypothetical Reaction Pathways for Desulfonylative Fragmentation of Radical Anion of o-Hydroxyphenyl α-Sulfonyl Ketone 2j Assisted by Intramolecular Proton Transfer

Studer et al. reported that bis-TEMPO adduct 7 was formed in the α-oxyamination reaction of acetophenone using TEMPO and chlorocatecholborane.[8g] Thus, we briefly examined the source of 7 in 1a-photocatalyzed reaction of 2m. Under the photocatalytic conditions using 3 h irradiation with a Xe lamp (Table ), 2m reacts to form 3m (33%) along with 7 (7%) at 42% conversion of 2m (entry 1). Stirring the irradiated solution for 24 h in the dark did not significantly alter the conversion and product distribution (entry 2), which implies that radical chain reactions do not take place. Moreover, prolonged irradiation (14 h) of the 2m-containing solutions led to increased yields of 7 and decreased yields of 3m based on the increased conversions of 2m (entries 3 and 4). These observations suggest that 7 is produced via the secondary photocatalyzed reaction of 3m. For comparison, the reaction of α-acetoxy acetophenone 13 was also carried out. In contrast, the reaction of 13 did not produce 7 at all with the excellent mass balance of 3m formation (entry 5). On the basis of these observations, the plausible reaction pathways to produce 7 from 3m are proposed as shown in Scheme . The in situ generated phenylsulfonyl radical from the sulfinate anion, although the concentration could be low, abstracts the α-hydrogen atom of 3m to give the α-ketoalkyl radical 14. Then, 14 is captured by TEMPO to produce 7. Similar radical generation from the acetate anion is unlikely due to its weaker electron-donating ability (Epox = +1.24 V).[25]

Table 6

Desulfonylative and Deacetoxylative α-Oxyamination Reactions of α-Sulfonylacetophenone 2m and α-Acetoxyacetophenone 13 with TEMPO Photocatalyzed by Benzimidazolium Naphthoxide 1a in DMSOa

					yield (%)b,c
entry	substrate	Z	irradiation time (h)	conv of 2m or 13 (%)b	3m	7
1	2m	PhSO2	3	42	33 (79)	7 (19)
2d	2m	PhSO2	3 + 24	45	29 (64)	5 (11)
3	2m	PhSO2	14	75	34 (45)	19 (25)
4e	2m	PhSO2	14	70e	30 (43)e	15 (21)e
5	13	CH3CO2	14	31	30 (97)	0

2m or 13 (0.10 mmol), 1a (10 mol %), TEMPO (2.0 equiv), DMSO (1.0 mL); 500 W Xe lamp (λ > 390 nm).

Determined by using 1H NMR.

Yields in parentheses are based on conversions.

After the operation (entry 1), the solution was stirred for 24 h without irradiation.

Preparative experiment to isolate 2m, 3m, and 7.

Scheme 5

Plausible Reaction Pathways from TEMPO Adduct 3m to bis-α-TEMPO Adduct 7 in the Photocatalyzed α-Oxyamination Reaction of α-Sulfonylacetophenone 2m

Conclusions

In the investigation described above, we demonstrated that benzimidazolium aryloxides (BI+–ArO–) serve as visible light-absorbing photocatalysts in cooperation with 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO) for efficient oxyamination reactions of α-sulfonylketones that generate α-TEMPO-substituted ketones. DFT calculations as well as electrochemical measurements were performed to gain information about the structures and the redox properties of BI+–ArO–. The results suggest that photoexcitation of BI+–ArO– produces a strong single electron donor that donates an electron to the α-sulfonylketone substrate. The radical anion formed in this manner losses a sulfinate anion to produce an α-ketoalkyl radical that is trapped by TEMPO to generate the oxyamination product. A characteristic feature of the proposed photocatalytic cycle is that TEMPO not only serves as a trap of the α-ketoalkyl radical intermediate but also as a SET reductant to regenerate BI+–ArO– from its radical cation. Although these properties of TEMPO have been well documented,[3f,4d,4e,4h,4i,4k,4m−4o] a photocatalytic system in which TEMPO acts in this type of dual cooperative role is unique because such two roles are not usually played together in a TEMPO involving photocatalytic cycle. To gain more mechanistic information and expand the scope of the substrate for the developed new photocatalytic protocol, further investigations need to be conducted.

Experimental Section

General Methods

1H and 13C{1H} NMR spectra were recorded on CDCl3 with tetramethylsilane (Me4Si) as an internal standard at 400 MHz for 1H NMR and 100 MHz for 13C NMR. Proton-decoupled 13C NMR data are reported. High-resolution mass spectra (HRMS) were recorded on an electrospray ionization (ESI) Orbitrap spectrometer. Uncorrected melting points are reported. According to a previously described procedure,[26] oxidation and reduction potentials in MeCN and DMF were measured using cyclic voltammetry. Conversion of the obtained potentials (V vs Ag/AgNO3) to the reported potentials (V vs SCE) was performed using the measured formal potential of ferrocene/ferrocenium (Fc/Fc+) couple (V vs Ag/AgNO3) and the difference of potentials between Fc/Fc+ and SCE (0.439 V). Half-wave oxidation and reduction potentials (E1/2ox and E1/2red) reported in the manuscript were obtained from peak potentials by subtracting or adding 0.029 V, respectively. The light sources for photoreactions were a 500 W Xe lamp with a glass filter L-42 (λ > 390 nm) and bulb-type white LED lamps (7.3 W Toshiba LDA7N-G-K and 10.8 W Toshiba LDA11D–G). Column chromatography was performed with silica gel. Anhydrous solvents for photoreactions were obtained as follows. Anhydrous DMSO and DMF were purchased and used without distillation. MeCN was distilled over P2O5 and subsequently distilled with K2CO3. CH2Cl2 was purified by the treatment with H2SO4, water, 5% NaOH, water, and CaCl2 and then distilled over CaH2. Other reagents and solvents were purchased and used without further purification.

Preparation of Benzimidazolium Aryloxides 1 and Hydroxynaphtyl Benzimidazoline 1a-HH

1,3-Dimethyl-2-(2-naphthox-1-yl)-benzimidazolium 1a,[13a] benzimidazolium phenoxides 1b,[13b]1c,[13b]1d,[13b] and 1e,[13b] and 1,3-dimethyl-2-(2-hydroxy-1-naphthyl)benzimidazoline (1a-HH)[11c] are known compounds and were prepared by using the previously reported procedures.

Preparation of Substrates

New sulfonylketones 2g and 2j were synthesized and characterized as described below. Other sulfonylketones 2a,[27]2b,[28]2c,[29]2d,[12]2e,[12]2f,[12]2h,[30]2i,[31]2k,[32]2l,[15c]2m,[27] and 8(15a) are known compounds. Thus, these substrates were prepared using the procedures described in Figure S7. Bromoketone 4(33) and acethoxyketone 13(34) are also known.

Synthesis of 1-(4-Bromophenyl)-2-methyl-2-phenylsulfonyl-1-propanone (2g)

A DMF (3.0 mL) solution containing 2-bromo-1-(4-bromophenyl)-2-methyl-1-propanone[35] (195 mg, 0.64 mmol) and PhSO2Na·2H2O (260 mg, 1.30 mmol) was purged with N2 and then stirred at room temperature for 21 h. The mixture was diluted with water (40 mL) and extracted with EtOAc/n-C6H14 (1/4, 40 mL × 3). The combined extracts were washed with water (30 mL × 2) and brine (40 mL), dried over anhydrous MgSO4, and concentrated in vacuo to give a white solid. The solid was crystallized from EtOH to give 2g (185 mg, 0.50 mmol, 78%); mp 129–131 °C; 1H NMR (400 MHz, CDCl3) δ 7.87 (dt, J = 8.8, 2.4 Hz, 2H), 7.79–7.76 (m, 2H), 7.71–7.67 (m, 1H), 7.60 (dt, J = 8.8, 2.4 Hz, 2H), 7.56–7.52 (m, 2H), 1.69 (s, 6H). 13C{1H} NMR (100 MHz, CDCl3) δ 198.1, 136.3, 135.1, 134.4, 131.7, 130.9, 130.5, 128.9, 127.5, 73.5, 22.7. HRMS (ESI) m/z: calcd for C16H1579BrO3S 366.9998, found 366.9998 [M + H]+, calcd for C16H1581BrO3S 368.9978, found 368.9976 [M + H]+.

Synthesis of 1-(2-Hydroxyphenyl)-2-phenylsulfonyl-1-propanone (2j)

A EtOAc (10.0 mL) solution containing 1-(2-hydroxyphenyl)-1-propanone (32.3 μL, 3.26 mmol) and CuBr2 (870 mg, 3.91 mmol) was stirred under reflux for 8 h. The mixture was filtered with EtOAc (10 mL), and the filtrate was diluted with water (40 mL) and extracted with EtOAc (40 mL × 3). The combined extracts were washed with water (30 mL × 2) and brine (40 mL), dried over anhydrous MgSO4, and concentrated in vacuo. To the residue was added PhSO2Na·2H2O (1.33 g, 6.65 mmol) and DMF (13 mL), and the resulting mixture was stirred under N2 at room temperature for 38 h. Then, the mixture was diluted with water (40 mL) and extracted with EtOAc/n-C6H14 (1/4, 40 mL × 3). The combined extracts were washed with water (30 mL × 2) and brine (40 mL), dried over anhydrous MgSO4, and concentrated in vacuo to give a yellow solid. The solid was crystallized from CH2Cl2/n-C6H14 to give 2j (438 mg, 1.51 mmol, 46%); mp 92–93 °C; 1H NMR (400 MHz, CDCl3) δ 11.78 (s, 1H), 7.82–7.79 (m, 3H), 7.67 (tt, J = 7.4, 1.2 Hz, 1H), 7.56–7.50 (m, 3H), 6.99 (dd, 8.4, 1.2 Hz, 1H), 6.95–6.91 (m, 1H), 5.18 (q, 6.8 Hz, 1H), 1.61 (d, 6.8 Hz, 3H). 13C{1H} NMR (100 MHz, CDCl3) δ 195.5, 163.3, 137.7, 135.9, 134.4, 130.0, 129.7, 129.0, 119.4, 118.7, 64.6, 13.0. HRMS (ESI) m/z calcd for C15H14O4S [M + H]+, 291.0686; found 291.0682.

Photoreaction Procedure

A general procedure for the photocatalytic TEMPO addition reactions utilizing catalyst 1 and TEMPO (Tables , 2, 5, and 6) is described below. A solution of substrate (2, 4, 8, or 13 (0.10 mmol)), 1 (0.01 mmol, 10 mol %), and TEMPO (0.10–0.20 mmol, 1.0–2.0 equiv) in the solvent (1.0 mL) in a pre-dried Pyrex test tube (1.4 cm diameter) was purged with N2 for 10 min and then irradiated with a Xe lamp or LEDs at room temperature for an appropriate time. The test tube was immersed in a Pyrex-water bath for the reactions using a Xe lamp. The distances between the test tube and the lamps are approximately 8.0 cm for Xe and 2.0 cm for LED. The temperature changes of DMSO in the test tube after irradiation were measured (Xe lamp, 14 h, 23–25 °C; LEDs, 20 h, 23–26 °C). The photolysate was diluted with water (30 mL) and extracted with EtOAc (20 mL × 3). The combined extract was washed with water (30 mL × 2) and brine (30 mL), dried over anhydrous MgSO4, and concentrated in vacuo to give a residue. The conversion of 2, 4, 8, or 13 and the yield of products 3, 7, or 9 were determined using 1H NMR analysis of the residue with triphenylmethane as an internal reference in CDCl3. The 1H NMR charts of the reaction mixtures of the selected experiments using 1 and 2 are presented in the Supporting Information (see Figure S15, 16, and S17). The preparative photoreactions of the selected substrates are described below. TEMPO adducts 3a,[8g]3b,[36]3h,[8g]3i,[8g]3k,[8g]3l,[8g] and 3m(8g) are known compounds. 5b is a commercial material. In the reaction of 8, formation of 9 was not confirmed. Although the generation of 7 is reported, no data is available in the literature.[8g] Thus, isolation of 7 was attempted (described below).

Preparative Photoreaction

Preparative reactions to isolate known 3b, 3i, and new TEMPO adducts 3e, 3f, 3g, and 3j were performed and are described below.

Preparative Photoreaction of 2b and TEMPO with 1a (Entry 3 of Table , Also See Figure S18)

Irradiation of 2b (35.0 mg, 0.10 mmol), 1a (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) in DMSO (1.0 mL) was carried out using two LEDs (10.8 W) under N2. The reaction mixture obtained after the same work-up procedure described above was subjected to column chromatography using EtOAc and n-C6H14 (2/1) to give 3b (33.7 mg, 0.092 mmol, 92%). White solid; mp 67–69 °C; 1H NMR (400 MHz, CDCl3) δ 7.90–7.77 (m, 2H), 7.48 (tt, J = 7.4, 1.4 Hz, 1H), 7.38–7.34 (m, 2H), 7.14–7.02 (m, 5H), 5.18 (dd, J = 10.2, 5.2 Hz, 1H), 3.45 (dd, J = 13.2, 5.2 Hz, 1H), 3.14 (dd, J = 13.2, 10.2 Hz, 1H), 1.52–0.85 (m, 18H).

Preparative Photoreaction of 2e and TEMPO with 1a (Table )

Irradiation of 2e (30.2 mg, 0.10 mmol), 1a (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) in DMSO (1.0 mL) was carried out using two LEDs (10.8 W) under N2. The reaction mixture obtained after the same work-up procedure described above was subjected to column chromatography using EtOAc and n-C6H14 (2/1) to give 3e (31.1 mg, 0.098 mmol, 98%). White solid; mp 78–80 °C; 1H NMR (400 MHz, CDCl3) δ 8.22 (d, J = 8.0 Hz,2H), 7.25–7.23 (m, 2H), 2.41 (s, 3H) 1.60 (s, 6H), 1.46–1.43 (m, 6H), 1.00–0.99 (m, 12H). 13C{1H} NMR (100 MHz, CDCl3) δ 201.9, 143.0, 132.5, 131.4, 128.8, 87.5, 59.6, 40.6, 33.6, 25.9, 21.8, 21.4, 17.3. HRMS (ESI) m/z calcd for C20H31NO2 [M + H]+, 318.2428; found 318.2429.

Preparative Photoreaction of 2f and TEMPO with 1 (Table )

Irradiation of 2f (32.3 mg, 0.10 mmol), 1 (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) in DMSO (1.0 mL) was carried out using two LEDs (10.8 W) under N2. The reaction mixture obtained after the same work-up procedure described above was subjected to column chromatography using EtOAc and n-C6H14 (2/1) to give 3f (31.5 mg, 0.093 mmol, 93%). White solid; mp 88–90 °C; 1H NMR (400 MHz, CDCl3) δ 8.28–8.25 (m, 2H), 7.43–7.41 (m, 2H), 1.60 (s, 1H), 1.45–1.43 (m, 6H), 0.99–0.98 (m, 12H). 13C NMR (100 MHz, CDCl3) δ 201.0, 138.8, 133.4, 132.7, 128.4, 87.5, 59.7, 40.6, 33.6, 25.8, 21.4, 17.2. HRMS (ESI) m/z calcd for C19H2834ClNO2 338.1881; found 338.1879 [M + H]+, calcd for C19H2836ClNO2 340.1852; found 340.1850 [M + H]+.

Preparative Photoreaction of 2g and TEMPO with 1a (Table )

Irradiation of 2g (36.7 mg, 0.10 mmol), 1a (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) in DMSO (1.0 mL) was carried out using two LEDs (10.8 W) under N2. The reaction mixture obtained after the same work-up procedure described above was subjected to column chromatography using EtOAc and n-C6H14 (2/1) to give 3g (31.5 mg, 0.082 mmol, 82%). White solid; mp 85–87 °C; 1H NMR (400 MHz, CDCl3) δ 8.18–8.15 (m, 2H), 7.95–7.56 (m, 2H), 1.58 (s, 6H), 1.44–1.22 (m, 6H), 0.98–0.97 (m, 12H) 13C NMR (100 MHz, CD3Cl) δ 201.8, 133.8, 132.8, 131.4, 127.7, 87.2, 59.9, 40.5, 33.5, 25.8, 21.4, 17.2. HRMS (ESI) m/z calcd for C19H2879BrNO2 382.1376; found 382.1375 [M + H]+, calcd for C19H2881BrNO2 384.1356; found 384.1355 [M + H]+.

Preparative Photoreaction of 2i and TEMPO with 1a (Table , Also See Figure S19)

Irradiation of 2i (30.9 mg, 0.10 mmol), 1a (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) in DMSO (1.0 mL) was carried out using two LEDs (10.8 W) under N2. The mixture obtained after the same work-up procedure described above was subjected to column chromatography using EtOAc and n-C6H14 (2/1) to give 3i (26.1 mg, 0.0806 mmol, 81%). Colorless oil; 1H NMR (400 MHz, CDCl3) δ 8.06 (d, J = 8.4 Hz,2H), 7.44 (d, J = 8.8 Hz, 2H), 4.92 (q, J = 6.8 Hz,1H), 1.51–0.84 (m, 21H).

Preparative Photoreaction of 2j and TEMPO with 1a (Table )

Irradiation of 2j (29.0 mg, 0.10 mmol), 1a (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) in DMSO (1.0 mL) was carried out using two LEDs (10.8 W) under N2. The mixture obtained after the same work-up procedure described above was subjected to column chromatography using EtOAc and n-C6H14 (2/1) to give 3j (26.3 mg, 0.086 mmol, 86%). Colorless oil; 1H NMR (400 MHz, CDCl3) δ 12.4 (s, 1H), 8.02 (d, J = 8.8 Hz, 1H), 7.48 (t, J = 8.0 Hz, 1H), 7.00 (d, J = 8.8 Hz, 1H), 6.91 (t, J = 8.0 Hz, 1H), 5.02 (dd, J = 8.8 Hz, 1H), 1.50–0.88 (m, 21H) 13C NMR (100 MHz, CDCl3) δ 208.0, 163.4, 136.5, 131.0, 118.9, 118.8, 117.9, 86.0, 60.0, 59.9, 40.3, 34.1, 33.6, 20.5, 19.7, 17.2. HRMS (ESI) m/z calcd for C18H27NO3 [M + Na]+, 328.1883; found 328.1887.

Photoreaction of 2b and TEMPO with Hydroxynaphthyl Benzimidazoline 1a-HH

A DMSO (1.0 mL) solution of 2b (35.0 mg, 0.10 mmol), 1a-HH (2.9 mg, 0.01 mmol), and TEMPO (31.3 mg, 0.20 mmol) with or without t-BuOK (5.6 mg, 0.05 mmol) was irradiated with a Xe lamp with a glass filter under N2. The mixture obtained after the same work-up procedure described above was subjected to 1H NMR analysis to determine the conversion of 2b and the yield of 3b (see Table ).

Photoreaction of 2b with 1a in the Absence or Presence of Additives

A DMSO (1.0 mL) solution of 2b (35.0 mg, 0.10 mmol) and 1a (2.9 mg, 0.01 mmol) in the absence or presence of 1,4-cyclohexadiene (28.4 μL, 0.30 mmol) and/or PhSO2Na·2H2O (40.9 mg, 0.2 mmol) was irradiated with a Xe lamp with a glass filter under N2. The mixture obtained after the same work-up procedure described above was subjected to 1H NMR analysis to determine the conversion of 2b and the yield of 5b (see Table ).

Photoreactions of α-Sulfonylacetophenone 2m or α-Acetoxyacetophenone 13 and TEMPO with 1a

A DMSO (1.0 mL) solution of 2m (26.0 mg, 0.10 mmol) or 13 (17.8 mg, 0.10 mmol) with 1a (2.9 mg, 0.01 mmol) and TEMPO (31.3 mg, 0.20 mmol) was irradiated with a Xe lamp with a glass filter under N2. The mixture obtained after the same work-up procedure described above was subjected to 1H NMR analysis to determine the conversion of 2m and 13 and the yield of 3m and 7 (see Table ). Separation of the mixture using column chromatography with EtOAc and n-C6H14 (1/20) gave 2m (7.8 mg, 0.030 mmol, 30%), 3m (8.2 mg, 0.030 mmol, 30%), and 7 (6.3 mg, 0.0146 mmol, 15%) (entry 4). 7 as white solid; mp 58–60 °C; 1H NMR (400 MHz, CDCl3) δ 8.25–8.23 (m, 2H), 7.56–7.51 (m, 1H), 7.47–7.43 (m, 2H), 5.79 (s, 1H), 1.48–0.96 (m, 36H) 13C NMR (100 MHz, CDCl3) 193.87, 132.86, 130.29, 128.13, 60.94, 60.92, 40.81, 40.78, 40.76, 17.13. [M + H]+, 431.3268; found 431.3262.

Density Functional Theory (DFT) and Time-Dependent DFT (TD-DFT) Calculations

DFT and TD-DFT calculations were performed using the Gaussian 09 package.[37] DFT calculations were performed on the ground state structures using the restricted B3LYP functional with the 6-31+G(d) basis set. The polarizable continuum model with the integral equation formalism (IEFPCM) was used for the solvent effect of DMSO. In addition, frequency calculations were performed on all the optimized structures to confirm that no imaginary frequencies exist. TD-DFT calculations were performed on the optimized structures using the unrestricted B3LYP functional with the 6-31+G(d) basis set and IEFPCM (DMSO). The optimized structures, molecular orbitals, and simulated UV–vis absorption spectra were visualized using GaussView 5.0.9.