Deoxygenative radical addition of aryl carboxylic acids to alkenes with phosphine in a two-molecule photoredox system

Abstract

We report a two-molecule photoredox system that enables deoxygenative radical addition of aryl carboxylic acids to electron-deficient alkenes under mild conditions. The cooperative action of an organic photocatalyst and a phosphine facilitates acyl radical generation and efficient carbon–carbon bond formation without metal reagents.

---

Deoxygenative transformations of aryl carboxylic acids have recently attracted attention because they provide a direct route to acyl radicals,^1–6 which are valuable intermediates for constructing aryl ketones, essential motifs in natural products and pharmaceuticals.^7,8 Although several photoredox-based deoxygenative methods have been reported, including the excellent Ir/PPh₃ protocol by Zhu and Xie (Scheme 1a),⁹ many existing approaches rely on precious metal catalysts or specially modified photocatalysts.^10–14

Scheme 1 General concept of phosphine-mediated modulation of a two-molecule photoredox system enabling deoxygenative acyl radical generation. (a) Previous deoxygenative acyl radical generation by Xie and Zhu (2018). (b) Mechanism of the two-molecule photoredox system. (c) This work.

Recently, we reported a two-molecule photoredox system, which is a catalytic system that enables radical generation through the cooperative activation of an organic electron donor (ED), such as dibenzo[g,p]chrysene (DBC), and an electron acceptor (EA), such as 1,4-dicyanobenzene (1,4-DCB).¹⁵ This system offers several advantages, such as distinct reactivity compared to one-molecule photoredox systems using Ir and Fukuzumi catalysts, because of the suppression of back electron transfer (BET) and facile replacement of these photocatalysts with other photocatalysts for modulating the oxidation/reduction potential of the substrate.^16,17 As shown in Scheme 1b, the photoexcitation of either ED or EA via photoinduced electron transfer (PET) produces ED˙⁺ and EA˙⁻, which oxidize the carboxylate ions to generate carboxy radicals. These carboxy radicals undergo rapid decarboxylation to generate alkyl or aryl radicals that add to electron-deficient alkenes, subsequently furnishing the corresponding adducts through BET and protonation.^18–22 For example, the use of DBC (E_ox = +1.17 V vs. SCE) as the electron donor enables the generation of alkyl radicals from aliphatic carboxylic acids (E_ox ≈ +1.16 V vs. SCE),¹⁵ whereas it is insufficient to induce decarboxylation of aryl carboxylate ions (E_ox = +1.40 V vs. SCE). In contrast, employing a more strongly oxidizing electron donor such as biphenyl (BP, E_ox = +1.95 V vs. SCE) allows efficient generation of aryl radicals from aryl carboxylic acids.^18,22 The redox properties of our system can be easily tuned by changing the ED/EA combination,^16,17,23 which enables the activation of aryl carboxylic acids and demonstrates the broad tunability of the redox properties within the system. In particular, the low efficiency of BET in this system leads to the successful direct photoinduced decarboxylation of aryl carboxylic acids, even though the one-molecule photoredox system failed.^18,22 The tunability and expansibility of our two-molecule photoredox system encouraged us to investigate the photoinduced deoxygenation of aryl carboxylic acids with phosphine through acyl radical generation (Scheme 1c). The generated acyl radicals were added to electron-deficient alkenes to provide a variety of aryl ketones from commercially available aryl carboxylic acids. This study expands the reactivity scope accessible through our two-molecule photoredox system and provides a new tunable, metal-free, and mild methodology for the generation of acyl radicals from aryl carboxylic acids.

Initially, the phosphine was evaluated in the two-molecule photoredox system. The aqueous acetonitrile solution (CH₃CN : H₂O = 49 : 1, v/v) of 4-methoxybenzoic acid 1a (30 mM), 3 equiv. of acrylonitrile 2A (90 mM), DBC (2 mM, 6.7 mol%), 1,4-DCB (10 mM, 34 mol%), 1.3 equiv. of tris(4-methoxyphenyl)phosphine (I, 40 mM, E_ox = +0.63 V vs. SCE) and 1 equiv. of K₂HPO₄ (30 mM) with a blue LED (18 W, 405 nm) under an argon atmosphere for 3 h at room temperature furnished adduct 3aA in a 96% yield (entry 1, Table 1). The use of triphenylphosphine II, which has a higher oxidation potential (E_ox = +0.98 V vs. SCE), decreased the efficiency and yield of 3aA (78%, entry 2). Replacing the phenyl group with a cyclohexyl group in the phosphine lowered the oxidation potentials, and an improved yield of 3aA (94%, entry 3) was observed using III (E_ox = +0.93 V vs. SCE). However, the use of dicyclohexylphenylphosphine IV (E_ox = +0.76 V vs. SCE) decreased the yield of 3aA (72%, entry 4) because of the poor solubility and lower stability of the corresponding radical cation of phosphine compared to that of I. In fact, the photoreaction with tricyclohexylphosphine V (E_ox = +0.54 V vs. SCE), which has poor solubility and lower stability of the radical cation, did not lead to the formation of 3aA (entry 5). No product was obtained using electron-deficient tris(pentafluorophenyl)phosphine (VIE_ox > +3.0 V vs. SCE, entry 6). The absence of phosphine or light resulted in no product formation, indicating the essential roles of phosphine and light in generating acyl radicals. When K₂HPO₄ was replaced with Na₂HPO₄, KH₂PO₄, K₂CO₃, or KOH as the base, lower yields of 3aA were observed (for details, see Table S1 in the SI), and in the absence of the base, the yield of 3aA decreased slightly (entry 7). Thus, the use of phosphine I and K₂HPO₄ (entry 1) was proven to be a suitable phosphine and base in the photoreaction of 1a.

Table 1 Optimization of the phosphine and base for photoinduced deoxygenative radical addition of 1a in the two-molecule photoredox system

Entry	Phosphine	Base	Yield of 3aA^a (%)	Recovery of 1a^b (%)	Recovery of phosphine^b (%)
a Isolated yields. b The yield was determined by ^1H NMR spectroscopy using 2,4,6-trimethoxybenzene as an internal standard.
1	I	K2HPO4	96	0	21
2	II	K2HPO4	78	0	2
3	III	K2HPO4	94	0	0
4	IV	K2HPO4	72	20	24
5	V	K2HPO4	0	83	93
6	VI	K2HPO4	0	Quant.	0
7	I	None	90	0	18

---

Next, the effects of ED and EA were investigated (Table 2). In the screening of ED, DBC yielded the highest results among the diverse EDs assessed (for details, see Table S2 in the SI). The effect of EA concentration has revealed that increasing the EA concentration leads to a higher concentration of the EA˙⁻ in the reaction system, suppressing oligomer formation, as reported by us.^20,21 Consequently, a higher concentration of 1,4-DCB afforded adduct 3aA in a higher yield (for details, see Table S3 in the SI), demonstrating the flexible tunability of the reaction efficiency by the EA concentration in this two molecule photoredox system. In the investigation of the EA types, variations in their acceptor capabilities resulted in differences in the yield (entries 1–4). Notably, in our system, combinations of BP and DCA, which typically facilitate the decarboxylation of aryl carboxylic acids,^18,22 favored the deoxygenation reaction over decarboxylation in the presence of phosphine (entry 5). Even in the absence of EA, 3aA was obtained (entry 6), indicating that direct PET between DBC and phosphine I occurred. As will be elucidated in the mechanism section, the rate of PET between DBC and 1,4-DCB is faster than that between DBC and phosphine I. Thus, the reaction proceeded more efficiently in the presence of both ED and EA. Furthermore, the application of Ir and Fukuzumi catalysts under these conditions resulted in low yields or no reaction (entries 7 and 8). Thus, the two-molecule photoredox system is suitable for efficient adduct formation by tuning the EA to enable efficient PET.

Table 2 Visible-light-induced deoxygenative radical addition from 1a and 2A to yield 3aA using various photoredox catalysts

Entry	ED (mM)	EA (mM)	Yield of 3aA^a (%)	Recovery of 1a^b (%)
a Isolated yields. b The yield was determined by ^1H NMR spectroscopy using 2,4,6-trimethoxybenzene as an internal standard.
1	DBC (2)	1,2-DCB (10)	80	5
2	DBC (2)	1,3-DCB (10)	27	62
3	DBC (2)	1,4-DCN (10)	72	6
4	DBC (2)	1,4-DMCB (10)	64	26
5	BP (2)	9,10-DCA (2)	54	25
6	DBC (2)	None	24	63
7	Ir[dF(CF3)ppy]2(dtbbpy)(PF6) (2)	–	27	3
8	Fukuzumi catalyst (2)	–	0	42

---

The scopes of the substrates and alkenes were examined under the optimized conditions (Table 3). The use of benzoic acid 1b, which has a lower donor ability, resulted in a moderate yield of adduct 3bA. Aryl carboxylic acids 1c–1g with electron-donating groups provided the corresponding adducts 3cA–3gA in high yields. When halogen-substituted aryl carboxylic acids 1h and 1i were used, adducts 3hA and 3iA were obtained in moderate yields. In the case of electron-withdrawing aryl carboxylic acids 1j, adduct 3jA was obtained in a low yield, and when 4-cyanobenzoic acid was used, the adduct was not obtained. For naphthoic acids 1k and 1l and 4-biphenylcarboxylic acid 1m, the adducts were obtained in moderate yields (3kA–3mA). In the alkene scope, it was found that decreasing the electron-withdrawing ability relative to 2A led to the corresponding adducts being obtained in low yields (3aB–3aG). The use of styrene 2H led to a low yield of adduct 3aH. This is attributed to the slower BET, which results in the formation of byproducts, such as oligomers. When an aliphatic primary carboxylic acid 1n was employed instead of an aryl carboxylic acid, a similar photoinduced deoxygenative adduct was not obtained at all, and decarboxylative adduct 4nA was observed (Scheme 2). In particular, the use of phosphine IV instead of I and II improved the yield of 4nA. This result is attributable to rapid decarbonylation following the formation of a primary acyl radical, leading to the generation of an alkyl radical.

Table 3 Aryl carboxylic acid and alkene scope using the two-molecule photoredox system with phosphine I

a The irradiation time was 12 h. b The concentration of 2G was 30 nM.

---

Scheme 2 Phosphines I, II and VI enabled the decarbonylative radical addition of aliphatic carboxylic acid 1n.

The fluorescence quenching Stern–Volmer plots were used to compare PET between DBC and phosphine I or 1,4-DCB. PET between DBC and 1,4-DCB occurred more rapidly than with phosphine I, demonstrating that efficient acyl radical formation occurred in the two-molecule photoredox system (Fig. 1).

Fig. 1 Fluorescence quenching of DBC (1 × 10⁻⁵ M) with phosphine I or 1,4-DCB excited at 380 nm in an aqueous acetonitrile solution (CH₃CN: H₂O = 49 : 1, v/v).

A TEMPO trapping experiment prevented this photoreaction, and the adduct with TEMPO and the corresponding acyl radical were detected by GC–MS. Based on these findings, it is considered that acyl radicals are generated, and adducts are obtained via radical reactions.

Based on these results, the proposed reaction mechanism is illustrated in Scheme 3. When the DBC absorbs light and reaches an excited state, PET proceeds from the DBC to the EA, generating DBC˙⁺ and EA˙⁻. The DBC˙⁺ returns to its neutral form by oxidizing phosphine , producing a phosphine radical cation . Notably, DBC˙⁺ oxidizes phosphines (E_ox = +0.63 to +0.98 V vs. SCE) preferentially and more rapidly than aryl carboxylate ions (E_ox = +1.40 V vs. SCE), leading to the selective formation of the phosphine radical cation. The carboxylate ion is then added to this radical cation, followed by rapid deoxygenation to generate an acyl radical. For aryl carboxylic acids, the acyl radical undergoes relatively slow decarbonylation, allowing direct addition of the acyl radical to the alkene to proceed. In contrast, aliphatic acyl radicals undergo rapid decarbonylation to generate alkyl radicals, which subsequently add to the alkene to afford the corresponding adducts via a radical pathway. The resulting radical intermediate is reduced via BET with EA˙⁻ to give an anion, and subsequent protonation affords the final adduct 3.

Scheme 3 Proposed mechanism.

In conclusion, this study demonstrates that cooperative use of a phosphine and an organic photocatalyst enables controlled access to acyl radical reactivity from aryl carboxylic acids under mild conditions. The present findings underscore the versatility of two-molecule photoredox systems for deoxygenative radical transformations using readily available feedstocks. Beyond the specific transformation reported here, this work provides insight into phosphine-assisted photochemical processes and is expected to inform the development of related reactions and expanded substrate classes in future studies.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support the findings of this study are available in the supplementary information (SI) of this article. Supplementary information is available. See DOI: https://doi.org/10.1039/d6cc00700g.

Acknowledgements

This work was partially supported by the Japan Society for the Promotion of Science (JSPS), Grant-in-Aid no. 22K18199 (M. Y.) for scientific research, the Iketani Science and Technology Foundation (M. Y.) and the TAKEUCHI Scholarship Foundation Grant No. Takeuchi 2023–J–05 (M. Y.). In particular, Kurogane Kasei Co., Ltd. provided us with dibenzo[g,p]chrysene.

References

1. J. J. Brunet and R. Chauvin, Chem. Soc. Rev., 1995, 24, 89–95.
2. X. Q. Hu, Z. K. Liu, Y. X. Hou and Y. Gao, iScience, 2020, 23, 101266.
3. M. Roy, B. Sardar, I. Mallick and D. Srimani, Beilstein J. Org. Chem., 2024, 20, 1348–1375.
4. S. Cao, H. Guo, Z. Zhong, D. Chen, W. Song, Y. Liu and J.-P. Wan, Sci. Adv., 2025, 11, eaea4120.
5. W.-C. Gao, Y. Teng, J. Yang, W.-D. Li, W.-G. Li, K.-X. Huang and T. Li, Chem. Commun., 2025, 61, 8743–8746.
6. W.-H. Tang, L.-Y. Wu, Q.-Q. Zhou and J.-P. Wan, Org. Chem. Front., 2025, 12, 2340–2345.
7. D. Felipe-Blanco and J. C. Gonzalez-Gomez, Adv. Synth. Catal., 2018, 360, 2773–2778.
8. S. Bhagat, N. Goel, P. Kumari, G. Gunjan and A. Khullar, Org. Biomol. Chem., 2025, 23, 5273–5292.
9. M. Zhang, J. Xie and C. Zhu, Nat. Commun., 2018, 9, 3517.
10. A. R. Norman, A. Elmarrouni and K. V. Chuang, Org. Biomol. Chem., 2018, 16, 1050–1054.
11. J. I. A. Martínez, A. B. Ertel, A. Stegner, E. E. Stache and A. G. Doyle, Org. Lett., 2019, 21, 9940–9944.
12. X. Q. Hu, J.-R. Chen and W.-J. Xiao, iScience, 2020, 23, 101098.
13. L. Zhang, S. Chen, H. He, W. Li, C. Zhu and J. Xie, Chem. Commun., 2021, 57, 9064–9067.
14. A. Nagaraju, T. Saiaede, N. Eghbarieh and A. Masarwa, Chem. – Eur. J., 2023, 29, e202202646.
15. M. Yamawaki, R. Hashimoto, Y. Kawabata, M. Ichihashi, Y. Nachi, R. Inari, C. Sakamoto, T. Morita and Y. Yoshimi, Eur. J. Org. Chem., 2022, e202201225.
16. Y. Yoshimi, J. Photochem. Photobiol., A, 2017, 342, 116–130.
17. Y. Yoshimi, Chem. Rec., 2024, 24, e202300326.
18. S. Kubosaku, H. Takeuchi, Y. Iwata, Y. Tanaka, K. Osaka, M. Yamawaki, T. Morita and Y. Yoshimi, J. Org. Chem., 2020, 85, 5362–5369.
19. Y. Shinkawa, T. Furutani, T. Ikeda, M. Yamawaki, T. Morita and Y. Yoshimi, J. Org. Chem., 2022, 87, 11816–11825.
20. M. Yamawaki, K. Matsumoto, T. Furutani, S. Sugihara and Y. Yoshimi, Polymer, 2024, 308, 127336.
21. T. Furutani, Y. Matsui, R. Hashimoto, H. Ikeda, T. Ogaki, M. Yamawaki, H. Suzuki and Y. Yoshimi, J. Org. Chem., 2025, 90, 4028–4036.
22. M. Kuwano, A. Asano, R. Hashimoto, M. Yamawaki, T. Furutani, H. Suzuki and Y. Yoshimi, Org. Lett., 2025, 27, 9347–9351.
23. R. Hashimoto, T. Furutani, H. Suzuki and Y. Yoshimi, Synlett, 2024, 357–361.

---