Charge-recombinative triplet sensitization of alkenes for DeMayo-type [2 + 2] cycloaddition

Abstract

Synthetic photochemistry has undergone significant development, largely owing to the development of visible-light-absorbing photocatalysts (PCs). PCs have significantly improved the efficiency and precision of cycloaddition reactions, primarily through energy or electron transfer pathways. Recent research has identified photocatalysis that does not follow energy- or electron-transfer formalisms, indicating the existence of other, undiscovered photoactivation pathways. This study unveils an alternative route: a charge-neutral photocatalytic process called charge-recombinative triplet sensitization (CRTS), a mechanism with limited precedents in synthetic chemistry. Our investigations revealed CRTS occurrence in DeMayo-type [2 + 2] cycloaddition reactions catalyzed by indole-fused organoPCs. Our mechanistic investigations, including steady-state and transient spectroscopic analyses, electrochemical investigations, and quantum chemical calculations, suggest a mechanism involving substrate activation through photoinduced electron transfer, followed by charge recombination, leading to substrate triplet state formation. Our findings provide valuable insights into the underlying photocatalytic reaction mechanisms and pave the way for the systematic design and realization of innovative photochemical processes.

This work revealed the occurrence of charge-recombinative triplet sensitization in DeMayo-type [2 + 2] cycloaddition reactions catalyzed by indole-fused organophotocatalysts.

Introduction

Photochemical cycloaddition belongs to a versatile chemical reaction class that harnesses photon energy to form cyclic compounds, playing a significant role in organic chemistry.^1–8 It enables the synthesis of a wide range of compounds with precise control over regio- and stereoselectivity. Historically, photochemical cycloadditions have been achieved through stoichiometric photon absorption by substrates with certain π conjugations, such as alkenes (Fig. 1a). In the past one and a half decades, synthetic photochemistry has experienced a remarkable resurgence, driven largely by the development of visible-light-absorbing photocatalysts (PCs).^9–13 PC introduction has substantially enhanced the efficiency and enantioselectivity of cycloaddition processes, primarily through charge-neutral energy transfer (EnT) (Fig. 1b).^14–20 Another common approach is photoredox catalytic cycloaddition, where the electron transfer (ET) pathway generates key radical intermediates (Fig. 1c).^21–26 The critical factor for successful EnT and ET processes is the exergonicity determined by the excited-state energies and redox potentials, respectively, of the PC and the organic substrate.^27–30 However, in some cases, PCs have exhibited satisfactory catalytic performance despite noticeable mismatches in energetic or electrochemical conditions.³¹ This observation suggests the existence of alternative, uncharted photoactivating pathways.

Fig. 1. Photoinduced cycloaddition.

In this study, we investigated an alternative route: charge-neutral photocatalytic process via charge-recombinative triplet sensitization (CRTS), a mechanism with limited precedents in synthetic chemistry (Fig. 1d).^32,33 To investigate this uncommon CRTS mechanism, we utilized an indole-based polycyclic organoPC series,^34,35 developed in our laboratory, for a DeMayo-type [2 + 2] cycloaddition^36–44 between methyl-2-(quinolin-2-yl)acetate (1) and various styrene derivatives. Detailed studies involving photophysical and electrochemical measurements and computational analyses provide insights into this unique CRTS pathway.

Results and discussion

Photocatalytic DeMayo-type [2 + 2] cycloaddition reactions

Our investigation began with screening various PCs for the DeMayo-type reaction between model substrates, 1 and styrene (2a) (Table 1). We employed indole-fused polycyclic organoPCs (A–H) (2.5 mol%) in 1,4-dioxane under visible light irradiation, utilizing 18 W blue light-emitting diodes (LEDs) for the formation of 3a.⁴⁵ Our systematic investigations revealed that structural variations in the PCs substantially impacted catalytic activity. Among the tested organoPCs, the pentacyclic organoPC A (CCDC 2225142; see Fig. S1†) and an amino-substituted tetracyclic variant H exhibited reactivity superior to those of the remaining tetracyclic systems.

Photocatalyst screening in DeMayo-type [2 + 2] cycloadditions^a,b.

^aReaction scale: 1 (0.1 mmol), 2a (0.5 mmol) under an Ar atmosphere.

^bThe yield was determined by ¹H NMR spectroscopy using 1,3,5-trimethoxybenzene as an internal standard.

^cYield of 3a: 0% (without PC).

^dYield of 3a: 0% (without light, with A).

^e1 and 2a remain unreacted.

In addition to our indole-fused organoPCs, we also evaluated widely used PCs such as Eosin Y, 4-CzIPN, and [Ir{dF(CF₃)ppy}₂(dtbbpy)]PF₆.⁴¹ However, none of these PCs exhibited superior reactivity compared to A and H. Control experiments showed that in the absence of an organoPC or visible light, the reaction did not proceed at all, confirming its photocatalytic nature.

The quantum yield for the reaction of 1 and 2a, which was determined using the standard ferrioxalate actinometry, is 49% with A and 48% with H. The other photocatalysts show quantum yields for the reaction in the range 4–23% (Table S1†). The values lower than 100% suggest the absence of intermediacy of any chain reaction.

Mechanistic investigations of photocatalytic DeMayo-type [2 + 2] cycloaddition reactions

Subsequently, our focus shifted toward understanding the photocatalytic process mechanism, particularly the origin of the superior reactivity of A and H. In the DeMayo-type [2 + 2] cycloaddition reaction, substrate triplet sensitization is a critical step.^{36,40–42,44} Thus, our investigation emphasized elucidating how indole-based polycyclic organoPCs facilitate photosensitization that leads to the generation of the substrate triplet state.

First, the photoluminescence (fluorescence) spectra were recorded for Ar-saturated 1,4-dioxane containing 1.0 mM A and increasing concentrations of 1 (0–500 mM) under 345 nm photoexcitation. As shown in Fig. 2a, the photoluminescence intensity of A decreased with increasing 1 concentration, suggesting a nonradiative interaction between 1 and excited state A (A*). The quenching interaction hardly required a ground-state association between the two species, as seen from the decreasing photoluminescence lifetime (τ_obs) of A* without a multiphasic transition in the decay trace (Fig. 2b; see Fig. S2† for results for the other organoPCs, B–H). The photoluminescence quenching rate constant (k_obs) is estimated from the relationship: k_obs = 1/τ_obs(1) − 1/τ_obs(0), where τ_obs(1) and τ_obs(0) are τ_obs in the presence and absence, respectively, of 1; the k_obs increases with the molar concentration of 1 ([1]), consistent with bimolecular quenching.

Fig. 2. Excited state interactions. (a) Photoluminescence (fluorescence) spectra of 1.0 mM A containing increasing concentrations of 1 in deaerated 1,4-dioxane. (b) Photoluminescence decay traces of 1.0 mM A in deaerated 1,4-dioxane, recorded after pulsed 345 nm laser photoexcitation at increasing 1 concentrations. (c) Corresponding quenching rate constant, k_obs = 1/τ_obs(1) − 1/τ_obs(0), as a function of 1 concentration, where τ_obs(1) and τ_obs(0) are photoluminescence lifetimes of A in the presence and absence of 1. Values are the rate constants for hetero-bimolecular quenching (k_Qs) of organoPCs by 1. (d) Yield of DeMayo-type [2 + 2] cycloaddition of 1 as a function of k_Q.

Pseudo-first-order kinetic analyses of k_obs yielded the rate constant for hetero-bimolecular quenching (k_Q) of A–H in the range of 1.2–2.7 × 10⁸ M⁻¹ s⁻¹, which approaches the diffusion-limited regime in 1,4-dioxane at 298 K (Fig. 2c and Table 2). Notably, k_Q does not correlate with the yield of the DeMayo-type [2 + 2] cycloaddition reactions of 1 (Fig. 2d), which suggests that the quenching of excited-state organoPCs by the substrate is not a rate-determining step in the overall photocatalytic cycle.

Photophysical and kinetic parameters of PCs and 1.

	λ abs a (nm, ε/104 M−1 cm−1)	E S1 b (eV)	λ em c (nm)	τ obs d (ns)	Φ PL e	k r f (107 s−1)	k nr g (107 s−1)	k ISC h (107 s−1)	J i (1012 M−1 cm−1 nm4)	k Q j (108 M−1 s−1)	k PeT k (108 M−1 s−1)	k CR l (108 M−1 s−1)
A	294 (4.40)	2.74	489	11	0.21	2.0	7.5	1.7	1.5	1.3	1.3	6.9
B	303 (1.29)	2.86	484	8.8	0.20	2.3	9.1	—	6.2	1.8	1.8	0.66
C	290 (1.88)	2.78	491	10	0.25	2.6	7.7	0.63	1.2	1.4	1.4	1.8
D	302 (2.00)	2.93	539	1.2	0.0045	0.37	83	—	—	—	—	0.66
E	290 (1.39)	2.83	479	12	0.47	4.0	4.6	—	2.5	1.5	1.6	1.9
F	290 (1.43)	2.85	478	8.9	0.21	2.3	8.9	10	2.0	1.2	1.3	3.8
G	378 (1.38)	2.81	486	8.5	0.23	2.7	9.1	—	—	—	—	4.1
H	412 (1.57)	2.73	462	3.4	0.39	11	18	1.0	2.7	2.7	2.9	9.3
1	316 (0.42)	2.69	365	—	—	—	—	—	—	—	—	—

^aAbsorption peak wavelength (molar absorbance).

^bThe first singlet state energy was determined from the UV-Vis absorption spectra onset wavelengths.

^cEmission peak wavelength.

^dPhotoluminescence lifetime determined through nonlinear least-squares fitting to a mono-exponential decay model of decay traces after picosecond pulsed 345 nm excitation.

^ePhotoluminescence quantum yield was determined using 9,10-diphenylanthracene as a standard (toluene, Φ_PL = 1.00).

^fRadiative rate constant, k_r = Φ_PL/τ_obs.

^gNonradiative rate constant, k_nr = (1 − Φ_PL)/τ_obs.

^hIntersystem crossing rate.

ⁱSpectral overlap is integral between the absorption spectrum of the substrate and the photocatalyst emission spectrum.

^jThe bimolecular quenching rate constant was determined from pseudo-first-order kinetic analyses of the photocatalyst fluorescence quenching rates with the substrate.

^kThe rate constant for the bimolecular photoinduced electron transfer rate was determined through the relationship k_PeT = −(k_Q × k_diff)/(k_Q − k_diff), where k_diff is the 1,4-dioxane diffusion rate constant at 298 K, estimated from the Stokes–Einstein–Smoluchowski equation.

^lThe rate constant for charge recombination within the radical ion pair [PC˙⁻⋯1˙⁺] to form ³1*.

The quenching of A* by 1 does not result from EnT. We found that k_Q exhibited a poor proportional relationship with the spectral overlap integral (J) between the UV-Vis absorption spectrum of 1 and the photoluminescence spectrum of A (Fig. S3 and S4†). The absence of proportionality indicates that the interaction between A* and 1 does not originate from singlet–singlet EnT that follows the Förster formalism. Moreover, the photoluminescence quenching behaviors are inconsistent with Förster EnT theories (see Fig. S5† for further discussion). In addition, even though the singlet sensitization of 1 to its excited state (¹1*) by the organoPCs would take place, ¹1* is unlikely to be converted to its triplet state (³1*); our quantum calculations, based on the method of Ma et al.,⁴⁶ indicate negligibly low rates (k_ISCs) for intersystem crossing (ISC): 1.4 × 10² s⁻¹ for the benzenoid form and 2.7 × 10³ s⁻¹ for the quinoid form of 1 (Table S2†). These results support the hypothesis that singlet–singlet EnT is not a productive pathway. Furthermore, the formation of an EDA complex and an exciplex can be excluded based on the absence of new bands in the UV-Vis absorption and photoluminescence spectra of a mixture of A and 1 or 2a (Fig. S6 and S7†).

In contrast, our femtosecond and nanosecond transient absorption spectroscopy experiments revealed a relatively fast ISC for the organoPCs, occurring at a rate of k_ISC = 0.63–10 × 10⁷ s⁻¹ (Fig. S8† and Table 2). The ISC of A occurs at a rate (k_ISC = 1.7 × 10⁷ s⁻¹) twice as fast as its k_obs in the presence of 0.050 M 1 (i.e., 6.5 × 10⁶ s⁻¹ = 1.3 × 10⁸ M⁻¹ s⁻¹ × 0.050 M). Based on the ISC behavior, one might consider that 1 undergoes triplet sensitization through consecutive processes involving the ISC of an organoPC, followed by triplet–triplet EnT from the organoPC triplet state to 1. However, quantum chemical calculations at the B3LYP-D3(BJ) level of theory using the TZP basis set, followed by time-dependent w-B97X-D calculations considering solvation effects, contradict this scenario. As shown in Fig. 3, the triplet (T₁) states of the B–H ground-state geometries (1.78–2.01 eV) were located below the T₁ state of the quinoid form of the 1 ground state (2.05 eV). In particular, the significant T₁ state energy difference between H and 1 (0.27 eV) strongly indicates that substrate activation via triplet–triplet EnT is unfavorable, except for A. Notably, H catalyzes the DeMayo reaction of 1, with yields as high as 93%. In the case of A, its T₁ state is isoenergetic with the T₁ state of 1 (2.05 eV). Our quantum chemical calculations for 1 reveal localization of the spin density within the heterocycle and methyne units (Fig. S9†), which suggests an occurrence of the [2 + 2] cycloaddition reaction with 2a, but refutes the possibility for the [4 + 2] cycloaddition.⁴⁷ In addition, the singlet excited (S₁) states of the A–H ground-state geometries (3.02–3.22 eV) are predicted to be lower in energy than the S₁ state of the quinoid form of 1 (3.24 eV). These energetic alignments indicate that EnTs from the excited-state organoPCs to 1 are endoergic, suggesting that triplet sensitization of 1via EnT is not feasible.

Fig. 3. Energy level alignments. Singlet (S₁) and triplet (T₁) state energies of organoPCs A–H and the quinoid form of 1, calculated at the w-B97X-D/TZP//B3LYP-D3(BJ)/TZP level using a solvation method based on the conductor-like screening model parameterized for 1,4-dioxane.

The Dexter theory of electron exchange predicts that the process rate decreases exponentially with increasing distance between the catalyst and the substrate. Good linearity was observed between the logarithm of the photoluminescence quenching rate constant, k_obs, and [1]^−1/3 (Fig. S5b†). This adherence to Dexter formalism implies that an excited-state organoPC is quenched by unidirectional or bidirectional ET—that is, the formation of a radical ion pair or EnT, respectively. Because EnT has been refuted, it is tempting to assert that unidirectional, photoinduced hetero-bimolecular ET forms radical ion pairs of the organoPC and substrate. To examine this possibility, we determined the oxidation (E_ox) and reduction (E_red) potentials of A–H and 1 using cyclic and differential pulse voltammetry (Fig. S10 and S11†). The excited-state redox potentials of organoPCs were subsequently calculated from the relationships = E_ox − E_S1 and = E_red + E_S1, where is the excited-state oxidation potential, E_S1 is the S₁ state energy determined from the onset wavelength of the UV-Vis absorption spectrum, and is the excited-state reduction potential. The ground- and excited-state redox potentials are compiled in Table 3. The (−1.52–1.37 V vs. standard calomel electrode (SCE)) of A–H are found to be more cathodic than the E_red (−1.13 V vs. SCE) of the 1 quinoid form. This electrochemical disposition allows photoinduced ET from organoPCs to 1, forming a geminate radical ion pair consisting of a one-electron oxidized species of organoPC (denoted as PC˙⁺) and a one-electron reduced form of 1 (denoted as 1˙⁻). The driving force for the oxidative quenching of excited-state organoPC (−ΔG^ox_PeT) can be estimated from the relationship −ΔG^ox_PeT = e[(PC) − E_red(1)], where e is the elementary charge; (PC) is the of an organoPC; and E_red(1) is the E_red of 1. The Coulomb term was ignored in this estimation owing to the use of polar 1,4-dioxane. The −ΔG^ox_PeT spans the range 0.24–0.39 eV (Table 3), which indicates that the oxidative quenching of A–H by 1 is exergonic. In addition to oxidative quenching, the reductive quenching of A–H to form a geminate radical ion pair of the organoPC radical anion (PC˙⁻) and the 1 radical cation (1˙⁺) is thermodynamically allowed. The driving force for the reductive quenching (−ΔG^red_PeT), estimated from the relationship −ΔG^red_PeT = e[E_ox(1) − (PC)] where E_ox(1) and (PC) are E_ox of 1 and the excited-state reduction potential of an organoPC, respectively, is in the range 0.31–0.55 eV (Table 3). Taken together, our electrochemical analyses suggest the formation of radical ion pairs of [PC˙⁺⋯1˙⁻] or [PC˙⁻⋯1˙⁺]. To monitor the radical ion pairs directly, we performed nanosecond laser flash photolysis (LFP) experiments on Ar-saturated 1,4-dioxane containing 1.0 mM A and 100 mM 1. The mixture was photo-irradiated under nanosecond pulsed laser excitation at 355 nm. As shown in Fig. 4a, a weak photoinduced absorption (PIA) signal emerged at wavelengths greater than 800 nm, together with significant photoinduced bleaching (PIB) in the shorter wavelength region due to the stimulated emission of A*. The PIA spectrum consisted of bands with approximately 840 and 1400 nm peak wavelengths.

The T₁ state energies, electrochemical potentials, and the driving forces for photoinduced electron transfer and charge recombination between the photocatalyst and 1.

	E T1 a (eV)	E ox (V vs. SCE)	E red (V vs. SCE)	b (V vs. SCE)	c (V vs. SCE)	−ΔGoxPeTd (eV)	−ΔGredPeTe (eV)	−ΔGCRf (eV)	g (eV)
A	2.05	1.37	−1.50	−1.37	1.24	0.24	0.35	2.39	0.34
B	2.01	1.47	−1.42	−1.39	1.44	0.26	0.55	2.31	0.26
C	1.98	1.38	−1.45	−1.40	1.33	0.27	0.44	2.34	0.29
D	1.93	—	—	—	—	—	—	—	—
E	2.01	—	—	—	—	—	—	—	—
F	1.93	1.44	−1.46	−1.41	1.39	0.28	0.50	2.35	0.30
G	1.98	—	—	—	—	—	—	—	—
H	1.78	1.21	−1.53	−1.52	1.20	0.39	0.31	2.42	0.37
1	2.05	0.89	−1.13	−1.80	1.56	—	—	—	—

^aThe T₁ state energies of the photocatalysts were calculated at the w-B97X-D/TZP//B3LYP-D3(BJ)/TZP level with COSMO parameterized for 1,4-dioxane.

^bExcited-state oxidation potential, = E_ox − E_S1.

^cExcited-state reduction potential, = E_red + E_S1. Table 2 presents the E_S1 values of the samples.

^dDriving force for oxidative electron transfer from the excited-state catalyst to 1, −ΔG^ox_PeT = e[(PC) − E_red(1)].

^eDriving force for reductive electron transfer from 1 to the excited-state organoPC, −ΔG^red_PeT = e[E_ox(1) − (PC)].

^fDriving force for the charge recombination reaction, PC˙⁻ + 1˙⁺ → PC + 1, −ΔG_CR = e[E_ox(1) − E_red(PC)].

^gDriving force for the charge recombination reaction, PC˙⁻ + 1˙⁺ → PC + ³1*, = −ΔG_CR − E_T1(1).

Fig. 4. Photoinduced charge separation and charge-recombinative triplet sensitization (CRTS). (a) Nanosecond transient absorption spectra of 1.0 mM A in the presence (top-most panel) and absence (second panel) of 100 mM 1 recorded at 1, 3.5, and 9 μs after pulsed 355 nm laser excitation (temporal resolution = 80 ns; laser power = 580 mW). The negative ΔAbs signals in the range 420–640 nm were due to the stimulated emission of A. The third panel shows UV-Vis-NIR absorption difference spectra of 2.0 mM A (Ar-saturated DMF containing 0.10 M TBAPF₆) recorded with an applied cathodic potential of −2.0 V vs. Ag^+/0. A Pt mesh and Pt wire were used as the working and counter electrodes, respectively. An Ag/AgNO₃ pseudo-reference electrode was used in this study. The area highlighted in yellow indicates absorption due to A˙⁻. The fourth panel shows the simulated UV-Vis-NIR spectrum of A˙⁻ calculated at the unrestricted wB97X-D/TZP//B3LYP-D3(BJ)/TZP level of theory with the COSMO parameterized for 1,4-dioxane. (b) Second-order kinetic analyses of decay traces of A˙⁻ recorded at a wavelength of 1400 nm. (c) Log k_CR as a function of the driving force for the reaction PC˙⁻ + 1˙⁺ → PC + ³1* . The red curve is the theoretical curve predicted by the Jortner formalism, with a reorganization energy of 1.2 eV. (d) Plot of k_CR as a function of yield for DeMayo-type [2 + 2] cycloaddition reactions for 1. (e) Plausible mechanism. These values were determined for A.

On the other hand, the spectrum of the control solution devoid of 1 exhibited negligible PIA bands in the range 1200–1600 nm. The PIA band at 1400 nm coincides with the visible-NIR absorption spectrum of A˙⁻ electrochemically generated at −2.0 V vs. Ag^+/0 (ε at 1400 nm is 285 M⁻¹ cm⁻¹). The simulated UV-Vis-NIR absorption spectrum of A˙⁻ further supports these observations (the fourth panel in Fig. 4a). These results provide compelling evidence for the photoinduced formation of [A˙⁻⋯1˙⁺].

Once formed, the radical ion pair [A˙⁻⋯1˙⁺] rapidly disappears through charge recombination via back-ET from A˙⁻ to 1˙⁺ within the solvent cage. Note that 1˙⁺ or its resonance benzylic radical derivative might react with the alkene; however, we excluded this pathway because it cannot lead to DeMayo-type [2 + 2] cycloaddition, which should occur through biradical intermediates (see Fig. S9 and Table S17†). This charge recombination typically produces charge-neutral ground-state species (that is, A and 1). When the driving force for charge recombination (−ΔG_CR) is greater than the constituent species T₁ state energy, charge recombination sensitizes the triplet state. We have previously reported CRTS in organic light-emitting devices^48,49 and photoredox catalysis.⁵⁰ The −ΔG_CR for [A˙⁻⋯1˙⁺] amounts to 2.39 eV, as estimated by −ΔG_CR = e[E_ox(1) − E_red(A)]. This −ΔG_CR value is greater than the T₁ state energy (2.05 eV) of the quinoid form of 1 (Fig. 3 and Table 3). The corresponding driving force for charge-recombinative sensitization of ³1* is estimated to be 0.34 eV for A from the relationship = −ΔG_CR − E_T1(1). Although we could not directly observe the spectroscopic signatures of ³1*, presumably because of its weak signals, the positive indicates the spontaneity of the CRTS of ³1*. The values for A–H are in the range 0.26–0.37 eV (Table 3).

The question remains as to why A–H produce different yields in DeMayo-type [2 + 2] cycloaddition reactions (Table 1). These differences may originate from their different rates of charge-recombinative substrate triplet sensitization. We determined the charge recombination rate (k_CR) through second-order kinetic analyses of the A˙⁻ PIA decay traces, which were recorded at a wavelength of 1400 nm using the molar absorbance determined spectrophotometrically (ε = 285 M⁻¹ cm⁻¹) (Fig. 4b). k_CR was determined to be as large as 6.9 × 10⁸ M⁻¹ s⁻¹ for A˙⁻ and 1˙⁺. The k_CR values for the other organoPCs, B–H, were also determined (Table 2). As shown in Fig. 4c, k_CR increases with . Based on the Jortner ET formalism,⁴⁹ our analyses suggest that charge recombination occurs in the Marcus normal region of ET with a reorganization energy of 1.2 eV. More importantly, a mild linear relationship was observed between k_CR and the yield for the DeMayo-type [2 + 2] cycloaddition reaction (Fig. 4d). For instance, A and H, which had the largest k_CR values, produced the best catalytic performance, whereas D, which had the smallest k_CR, elicited the lowest reaction yield. Although inconclusive, the linearity strongly suggests that charge recombination, which generally causes quenching of the reaction, is actually a key step in our results. Therefore, it can be concluded that the charge-recombinative triplet-state generation step governs the overall catalytic performance.

Based on these results, we propose a plausible mechanism involving 1 and 2a (Fig. 4e). Because the DeMayo-type [2 + 2] cycloaddition proceeds from the T₁ state, substrate triplet activation is essential. Our proposed mechanism involves a photoinduced cycle of consecutive steps, including (i) initial photon absorption by an organoPC to form an excited-state organoPC (¹PC*) that is intrinsically deactivated at a rate of 2.0 × 10⁷ s⁻¹ in the case of A; (ii) diffusion-controlled formation of an encounter complex between ¹PC* and a substrate [¹PC*⋯substrate];⁵¹ (iii) ET from the substrate to ¹PC* to form a radical ion pair [PC˙⁻⋯substrate˙⁺]; (iv) charge recombination within the radical ion pair to form the substrate T₁ state; (v) dissociation of the triplet substrate and the PC. The triplet substrate subsequently reacts with styrene 2a to generate a biradical adduct, or the biradical intermediate may cyclize readily to form a cyclobutane intermediate. Finally, ring-opening⁵² followed by aromatization produces 3a.

Substrate scope

Next, the reaction scope was investigated using a range of alkene derivatives (2) under optimized reaction conditions (Table 4). Various styrene derivatives underwent regioselective DeMayo-type [2 + 2] cycloadditions to yield the corresponding ring-opening and aromatization products (3a–3t) irrespective of the electron density or position (ortho-, meta-, para-) of the substituents on the aromatic ring. The mild reaction conditions tolerated functional groups, including chloride (3j, 3o) and bromide (3k); medicinally important F (3l, 3m)^53–55 substituents could also be used, and a heteroaryl variant, the pyridine-substituted (3r) derivative, was also suitable for the transformation. Notably, nonaromatic (3p, 3q, 3s) and internal alkenes (3s, 3t) were also suitable coupling partners under the standard conditions. However, simple aliphatic alkenes lacking the ability to form resonance-stabilized radical intermediates were ineffective in this transformation (see Fig. S14†).

Substrate scope^a,b.

^aReactions were performed on a 0.3 mmol scale.

^bIsolated yields are reported.

^cReaction time: 24 h.

^dReaction time: 48 h.

^e2t (1 eq.).

Conclusions

In summary, our investigation explored the uncharted territory of the CRTS process in DeMayo-type [2 + 2] cycloaddition reactions using indole-fused organoPCs developed in our laboratory. The intricate mechanism involves substrate activation through photoinduced ET, followed by charge recombination that leads to substrate triplet state generation. Comprehensive steady-state and transient spectroscopic experiments, electrochemical investigations, and quantum chemical calculations validated this mechanism. Our findings not only shed light on the underlying mechanisms of photocatalytic reactions but also pave the way for realizing systematically designed, novel photochemical processes. Moreover, the remarkable photocatalytic capabilities of our indole-fused polycyclic organoPCs hold significant promise for their use in other valuable photochemical reactions.

Data availability

The data underlying this study are available in the published article and its ESI.† Crystallographic data for A have been deposited at the CCDC2225142.

Author contributions

Y. L., B. H. J., S. W., S. K., and J. B. performed synthetic and mechanistic studies. Y. Y. and E. J. C. coordinated the experiments and analyses. All authors analyzed the experimental data and wrote the manuscript.

Conflicts of interest

The authors declare no competing financial interest.