Light-driven activation of carbon-halogen bonds (C–X) has become a powerful approach for constructing new chemical bonds in valuable pharmaceutical and industrial intermediates, as well as for the elimination of polyhalogenated aromatic pollutants [1-8]. Essentially, the process relies on high-energy excited states/intermediates that trigger electron transfer to cleave carbon-halogen bonds (C–X + e^- → C· + X^-, dissociative electron transfer), affording carbon radicals that can engage in diverse reactions, including hydrogenation and cross coupling [9-18]. For a thermodynamically favored process, energy levels of the photogenerated species should locate above the anti-bonding σ* or π* orbital of halogenated substrates. Promising results have been reported with classical photosensitizers, Ru(bpy)₃²⁺ and fac-Ir(ppy)₃ (bpy is 2, 2'-bipyridine and ppy is 2-phenylpyridine), where their triplet excited states (-0.8 and -1.7 V vs. SCE) [1, 19], or the reduced complexes (-1.3 and -2.1 V vs. SCE) enable dissociative electron transfer. Recently, hexachlorocerate(Ⅲ) (Ce^ⅢCl₆^3-), with a strongly reducing excited state (-3.45 V vs. Fc/Fc⁺), has been developed as a sensitizer for the reduction of aryl chlorides under UVA irradiation [20, 21]. Copper complexes are likewise involved in conversions of carbon-halogen bonds [22, 23], e. g., C-F functionalization under pulsed light-emitting diodes (LEDs) [24]. Organic sensitizers, such as diaryl dihydrophenazines [25], phenylphenothiazine [26, 27], perylene diimide[28] and pyrene [2], are also efficient photosensitizers, producing active reducing species (-2.06 to -2.5 V vs. SCE) via photon upconversion or consecutive two-photon absorption. These organic molecules are more cost effective and less prone to deactivation (e.g., halogen coordination in the case of transition metal catalysts) [3, 26, 27]. Halogen bonds or π-π interaction between catalyst and substrate may further facilitate bond activation via an inner sphere electron transfer pathway, however synthetic efforts are required in order to modulate the intermolecular interactions [15, 29].

Herein, we establish a metal- and additive-free photocatalytic system based on a N, N, N', N'-tetramethyl-p-phenylene- diamine (TMPD), or N, N, N', N'-tetramethylbenzidine (TMB) sensitizer featuring strongly reducing excited singlet states and diisopropylethylamine (DIPEA) as an electron/proton donor, for hydrodehalogenation of a variety of haloarenes under UV irradiation (λ > 360 nm). In this approach, photogenerated excited states of the amine sensitizers enable diffusion-controlled electron transfer to halogenated substrates, followed by cleavage of the C–X to give an aryl radical; subsequent radical hydrogenation affords the dehalogenated product (Scheme 1).

TMPD displays S₀-S₁ absorption extended to 380 nm (Fig. 1(a)). In the presence of selected aryl halides (4'-bromoacetophenone, hexafluorobenzene and chlorobenzene), TMPD absorption variations were negligible, excluding the interaction of the TMPD ground state with these compounds. The viability of TMPD in photo-induced activation of carbon-halogen bonds was initially evaluated by steady-state and time resolved emission spectroscopies. The emission intensity of TMPD (0.1 mM, 360 nm excitation) decreased upon introduction of all three substrates (2-10 mM, Fig. 1(b) and Fig. S1), which is consistent with an electron transfer from the singlet excited states to these compounds. The emission lifetime measured at 405 nm likewise decreased (Fig. 1(c)). The parallel Stern-Volmer plot of emission intensity and lifetime (Fig. 1(d)) indicates an outer sphere electron transfer to hexafluorobenzene, with negligible contribution of static quenching via a pre-associated ground state complex, in agreement with the unchanged absorption spectra [30-32]. The rate constants calculated by Stern-Volmer analysis (Fig. S1) were 5.8 × 10¹⁰ M^-1 s^-1 for 4'-bromoacetophenone, 4.4 × 10¹⁰ M^-1 s^-1 for hexafluorobenzene and 2.6 × 10¹⁰ M^-1 s^-1 for chlorobenzene (Fig. S1). Notably, the observed electron transfer is diffusion-limited, even in the case of exceptionally inert chlorobenzene [20]. For TMB, the electron transfer to hexafluorobenzene was also likewise diffusion-limited (3.6 × 10¹⁰ M^–1 s^–1), but became activation-controlled in the case of chlorobenzene (6.2 × 10⁹ M^–1 s^–1, Fig. S1). The slower electron transfer observed for TMB could be attributed to the fact that its excited state energy is lower than that of TMPD (–3.48 and –3.56 vs. Fc/Fc⁺, respectively) [33]. These results established a kinetically feasible electron transfer, the initial step for hydrodehalogenation [6].

In the initial test, where 4′-bromoacetophenone (1 mM) was used as the model substrate with a stoichiometric amount of TMPD (2:1 according to the 2e/1H⁺ process) as the photoreductant, a conversion of 86% with a selectivity of 90% for hydrodebromination was observed after 6 h of irradiation (Table 1, entry 1). The quantitative production of bromide was additionally confirmed by ionic chromatography (Fig. S2). Involvement of aryl radical intermediates via the initial electron transfer was supported by the generation of C–C coupled products in the presence of N-methyl pyrrole (Fig. S3) [3, 34]. In situ electron paramagnetic resonance (EPR) experiments and the simulated EPR spectra also suggested the generation of TMPD^+· and an acetophenone radical (via C–Br bond cleavage of the 4′-bromoacetophenone radical anion) (Fig. S4 and Table S1). The aryl radical then abstracts a hydrogen atom from TMPD/TMPD^+· to give the hydrogenated product. TMB exhibited comparable activity under these conditions, with a yield of 88% (Table S2). Control experiments confirmed that photoirradiation and amines were essential for efficient hydrodehalogenation. The highest hydrodebromination yield was obtained in acetonitrile, while significantly inferior yields were observed with dimethylformamide (27%), methanol (20%), dichloromethane (17%), and n-hexane (10%) (Table S2). For hexafluorobenzene (Table 1, entry 2), both mono- and di-defluorination products were observed after 5 h of reaction time, albeit in relatively low yields (pentafluorobenzene, 26%, 1, 2, 4, 5-tetrafluorobenzene, 17%). It is possible that the fluorinated aryl radical engages in other processes due to the lack of efficient H donors. Chlorobenzene also demonstrated satisfactory conversion (63% at 5 h) and selectivity (80%, Table 1, entry 3).

Next, we conducted the hydrodehalogenation under catalytic conditions with 5% TMPD, using DIPEA (15 equivalents) as the electron/hydrogen donor for radical intermediates. DIPEA also reduces TMPD^+· to recover the photosensitizer (Fig. S5). The results are summarized in Table 2 and Fig. S6. Gratifyingly, the conversion of 4′-bromoacetophenone occurred smoothly, with a hydrodebromination yield of 90% (entry 1), corresponding to a turnover number (TON) of 18. The hydrodebromination yield of bromobenzene was 37% after 6 h of irradiation (entry 2). Hydrodefluorination yields above 70% were obtained for hexafluorobenzene and pentafluoropyridine (enter 3 and 4). Notably, the yields of pentafluorobenzene (25%) and the di-hydrodefluorination product, 1, 2, 4, 5-tetrafluorobenzene (50%, Fig. S7) were significantly higher than those obtained under stoichiometric conditions, presumably due to a more efficient hydrogenation by DIPEA/DIPEA^+·. In the case of chlorobenzene, the hydrodechlorination yield approached 40% after 2 h of irradiation, but did not increase with extended reaction time (entry 5). The dechlorination could be resumed by the addition of TMPD (0.1 equivalent, Fig. S6). The hydrogenation yield of methyl 4-chlorobenzoate reached 72% after 7.5 h of irradiation (entry 6). In addition, the catalytic system demonstrated compatibility with halogenated heteroarenes (entries 7 and 8), affording moderate to good product yields. The TON of these substrates with TMPD ranged from 7.4 to 14.4. When TMB was employed as the photocatalyst, a low dechlorinated product yield of 15% was observed under the same conditions, despite extended reaction times of 20 h (entry 9). This result could be attributed to the weaker electron transfer driving force from TMB excited states to the halogenated substrates, compared to TMPD, as indicated by emission quenching dynamics. 4, 4', 4''-(1, 3, 5-tri- azine-2, 4, 6-triyl)-tri-aniline (TTA), which demonstrates similar absorption to TMPD, was also tested for its photocatalytic ability in the hydrodehalogenation of 4′-bromoacetophenone (Fig. S8). The product yield after 4 h of irradiation reached 90% (entry 10). These results demonstrate that amine-based sensitizers are promising photocatalysts in light-driven organic transformations.

We further investigated the performance of TMPD as a photocatalyst in the debromination of 2, 2', 4, 4'-tetrabromo- diphenyl ether (BDE47), a typical persistent organic pollutant that is resistant to conventional reductive treatments [10, 35, 36]. The removal of bromines in polybrominated diphenyl ethers with low bromine substitution is challenging, as debromination becomes less thermodynamically favorable with decreasing bromine atom numbers. Under photocatalytic conditions, Pd cocatalysts are usually required to provide high degrees of debromiantion [10, 37, 38]. It was impressive to find that with 1 equivalent of catalyst and 2.5 v% DIPEA, BDE47 underwent facial debromination and 83% debromination efficiency was realized after 30 h, producing the completely debrominated product, diphenyl ether in 40% yield (Fig. S9). The distribution of debrominated intermediates, e.g., the 4:1 ratio of 2, 4, 4′-tribromodiphenyl ether (BDE28) to 2, 2′, 4-tribromodiphenyl ether (BDE17), indicated a debromination preference for the ortho position, which is in accordance with an electron transfer-triggered debromination [39, 40].

Emission quenching kinetics and the results of photochemical/photocatalytic hydrodehalogenation are consistent with the cleavage of carbon halogen bonds via an electron transfer from singlet excited states of TMPD or TMB. Control experiments indicated that DIPEA had a negligible effect on the emission of TMPD (Fig. S10), which is different from the activation of carbon-halogen bonds by the pre-photoreduced sensitizer in pyrene derivatives, and fac-Ir(ppy)₃ catalyzed hydrodehalogenation [1, 16]. Analogous pathways have been reported for phenylphenothiazine and Ce^ⅢCl₆^3– under near-UV or UVA irradiation [20]. The aromatic amines feature strongly reducing excited states, and their electron transfer kinetics are one order of magnitude faster than the activation-limited transfer from Ce^ⅢCl₆* to 4-F-C₆H₄X (X = Br, Cl, F, 0.9–3.0 × 10⁹ M^–1 s^–1). Notably, the overall conversion is additionally limited by the competition of carbon-halogen bond cleavage of the aryl halide radical anion intermediates, and the non-productive thermal back electron transfer [6]. Nonetheless, the activation of C–Cl and C–F bonds with readily available amines is an attractive alternative for selective hydrodehalogenation under mild conditions, and in a sustainable manner. The development of related compounds with suitable excited state energies and HOMO-LUMO levels are expected to achieve visible light activity, rapid catalyst turnover and enhanced activity. These amine sensitizers, in combination with a sencond light absorber could be integrated into tandem photocatalytic systems, in order to inhibit the back electron transfer, thus favoring the cleavage of carbon-halogen bonds.

In summary, we demonstrated that excited states of amine sensitizers enable diffusion-controlled electron transfer to effectively cleave a number of highly challenging carbon-halogen bonds. Successful light-driven hydrodehalogenation of various aromatic halides and organic pollutants was achieved in the absence of hazardous reagents. This method is a practical and cost-effective strategy, enabling exceptionally high activity.