Amino-substituted triazatriangulenium photosensitizers for CO₂ capture and aminocarbonylation to amides

Abstract

Efficient photoreduction of CO₂ to CO using noble-metal-free systems remains a significant challenge in artificial photosynthesis. Developing low-cost photosensitizers capable of capturing CO₂ and facilitating electron transfer is therefore essential. Here, we report an amino-substituted triazatriangulenium photosensitizer (A-TATA) that enables light harvesting and CO₂ capture in a photocatalytic system. Systematic studies show that A-TATA, functionalized with free amino groups, captures CO₂ as carbamic acid, serving as a local CO₂ reservoir. This increased local concentration of CO₂ lowers the onset potential of the cobalt catalyst. Notably, the system achieves a turnover number of 33,976 with 98% selectivity and an optimized quantum yield of 51% for CO−among the highest reported for molecular photocatalysis. Furthermore, the generated CO is converted into amides via aminocarbonylation, achieving 85% atomic efficiency and operating effectively even in the absence of solvent. These findings offer a strategy for designing versatile organic photosensitizers for sustainable CO₂ capture and conversion.

A noble-metal-free triazatriangulenium photosensitizer captures CO₂ as carbamic acid to boost local concentration to increase reduction efficiency. It achieves 51% quantum yield for CO, and forms amides with 85% atom efficiency even in the absence of solvent.

Introduction

A cost-effective artificial photosynthesis system that utilizes solar energy to convert CO₂ into fuels and valuable chemicals has gained significant attention as a promising strategy for CO₂ recycling^1–4. However, the inherently low reactivity and high kinetic stability of CO₂ present significant challenges to achieving the efficiency needed for practical applications^5,6. Prodigious efforts have focused on developing efficient photocatalytic systems capable of reducing CO₂ to C₁ products like CO or HCOOH^1–4. Despite these advancements, the subsequent utilization of photogenerated CO for carbonylation reactions remains a critical challenge^7–9. Moreover, photocatalytic CO₂ reduction is hindered by the competitive reduction of water to H₂, resulting in gaseous products often mixed with unconverted CO₂ and varying amounts of CO and H₂, which necessitates complex and costly separation processes before further use⁷. Hence, developing a highly selective photocatalytic system that facilitates the conversion of C₁ products into multi-carbon products (C₂₊) represents a promising direction for CO₂ reduction research.

In the field of artificial photosynthesis, homogeneous molecular photocatalytic systems composed of a photosensitizer (PS) and a catalyst (Cat), are valued for their well-defined and tunable molecular structures^10–12. The PS plays a critical role in photocatalysis, not only absorbing light but also facilitating electron transfer to the catalytic center^10,11. Despite numerous studies exploring various metal-complex PS and Cat combinations, many systems still rely on noble or rare metals. For example, complexes of Ru(II)^13,14, Ir(III)^15,16, and Re(I)^17,18, and their combinations^16,19, have demonstrated effective CO₂ photoreduction, achieving quantum yields of up to 82% and turnover number (TON) exceeding 1000. Recently, there has been growing interest in CO₂ photoreduction systems based on earth-abundant elements, with Fe(II)^20–24, Co(II)^25–28, Ni(II)^29–32, and Mn(II)^33,34 complexes being extensively studied as catalysts. PSs based on first-row transition metals, like Cu^33,35–39 and Zn^18,40, as well as purely organic chromophores^27,41,42, have gained considerable attention for light-driven CO₂ reduction due to their impressive photocatalytic activity. Consequently, the development of efficient noble-metal-free PSs has become a key focus in the quest to build effective CO₂ photoreduction systems. For example, Ishitani et al.^33,35, Beller et al.^36,37, and other researchers³⁹ have recognized and predicted the significance of high-performance Cu(I)-based PSs, highlighting their potential when paired with diverse photocatalysts for CO₂ reduction applications. Robert and Lau demonstrated that using purpurin as a PS with iron Cat could achieve a TON of 1365 with 92% selectivity (S_CO), though the QY was only 0.8%²⁰. Similarly, a triazatriangulenium salt complex paired with Co(II)-quaterpyridine system achieved a TON of 19,000 and a selectivity of 93%, but with the QY was of just 0.8%²⁷. In contrast, aminoanthraquinone dyes as PSs paired with an Fe porphyrin catalyst yield a TON of 21,616 and an S_CO > 99.9%, with a notably higher QY of 11.1%⁴¹. Furthermore, a donor–acceptor-type organic dye functionalized with four phosphonic acid groups, used as a PS in aqueous solution and paired with cobalt porphyrin catalyst, achieved a QY of 17.4%, although the TON was only 2700 with an S_CO of 93%²⁸. Additionally, phenoxazine⁴³, Eosin Y³⁰, 2,4,5,6-tetrakis(carbazole-9-yl)−1,3-dicyanobenzene (4CzIPN)⁴⁴, and Bodipy-anthryl¹² have also been utilized as PSs in CO₂ photoreduction systems with various non-noble metal molecular catalysts. While previous research has focused primarily on improving photocatalyst efficiency, cost-effectiveness, and selectivity. However, little attention has been given to developing noble-metal-free systems capable of capturing CO₂ and converting it into multi-carbon compounds.

Typical molecular photocatalytic systems designed for CO₂ capture involve CO₂ insertion into M-X bonds, forming CO₂-TEOA (triethanolamine) complexes, and modifying the secondary coordination sphere to stabilize catalysts via Lewis acid acceptors (Fig. 1a). CO₂ molecules can insert into M−H^45,46, M−OH^47,48, M−C^49,50, M−N^49,51, and M−OR (R = alkyl or aryl groups) bonds^52,53, which enhance equilibrium and rate constants for CO₂ binding. However, further applications of these photocatalysts have not been fully explored (Fig. 1a, I). Re(I)^54–56 photocatalysts have demonstrated the capability to capture CO₂ in the presence of TEOA, forming a long-lived carboxylate–ester complex that exhibits high photoreduction efficiency for CO₂. (Fig. 1a, II). Notably, the incorporation of pyridine pendants or urea functionalities to modify the secondary coordination sphere has been shown to enhance the stability and performance of CO₂-bound photocatalysts (Fig. 1a, III)^31,57,58. Molecular photosensitizers capable of capturing CO₂ are particularly intriguing, as their structures can be tailored to fine-tune CO₂ binding properties. These examples highlight the significance of CO₂ capture in photocatalytic systems. However, to the best of our knowledge, the role of molecular photosensitizers in promoting CO₂ capture for photoreduction has not been thoroughly examined.

Fig. 1. Compounds investigated in this study.

a Molecular photocatalytic systems with CO₂ capture abilities. b The photosensitizers include N,N′,N″-tris-(aminoethyl)-triazatriangulenium hexafluorophosphate (A-TATA), N,N′,N″-tris(2-hydroxyethyl)-triazatriangulenium hexafluorophosphate (HEA-TATA) and N,N′,N″-tris(n-propyl)-triazatriangulenium hexafluorophosphate (ⁿPr-TATA). Additionally, the study examines catalyst C-1 and the electron donor BIH.

Herein, we report a triazatriangulenium carbocationic (TATA⁺) complex decorated with free amines (Fig. 1a, IV and b) that functions as an efficient PS for CO₂ capture. The photoreduction of CO₂ was evaluated in combination with [Co^II(qpy)(OH₂)₂]²⁺ (C-1) (qpy = 2,2′:6′,2′′:6′′,2′′′-quaterpyridine) and 1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzo[d]imidazole (BIH). Notably, under optimized conditions, CO₂ was efficiently reduced to CO using A-TATA, achieving a yield of n_CO = 397 μmol, TON = 9925, and S_CO = 99%. The system continued to produce CO until it approached the theoretical maximum yield based on the electron donor BIH. This remarkable photocatalytic performance inspired us to explore the conversion of CO into valuable multi-carbon compounds using the A-TATA system. Subsequently, the generated CO was successfully transformed into pharmaceutical intermediates via a tandem palladium-catalyzed carbonylation reaction, with an atomic utilization efficiency of up to 85%. These findings highlight the potential for designing CO₂-capturing PSs to develop noble-metal-free photoreduction systems and drive continuous innovations in photocatalytic technologies for solar-to-fuel conversion.

Results and discussion

Molecular design and synthesis of the TATAs

TATA⁺ was selected as the primary molecular scaffold for the PS due to its planar structure, three electron-donating nitrogen atoms that facilitate the stabilization of TATA^• through radical delocalization, high molar extinction coefficient (~14,000 M⁻¹cm⁻¹, at ~530 nm) in the visible region, and suitable redox potential, ensuring effective sensitizing ability for the photocatalytic process^59,60. The ligand L-3 (tris(2,6-dimethoxyphenyl)carbenium ion) was synthesized according to a published procedure, yielding 72% from commercially available materials. Target molecules were synthesized in a one-pot procedure by direct introduction of the alkyl side chains through triple aromatic nucleophilic substitution (S_NAr) and ring closing reactions with primary amines on ligand L-3⁵⁹. To enhance their solubility, the complexes were isolated as hexafluorophosphate (PF₆⁻) salts. Detailed synthetic procedures and chemical structures are provided in the Supporting Information and illustrated in Supplementary Fig. 1. All chemical structures were fully confirmed by ¹H and ¹³C NMR and ESI-MS analysis (Supplementary Figs. 2–16, Supplementary Table 1).

As indicated in Fig. 1, we synthesized TATAs decorated with ligands terminated with amines (A-TATA), and for comparison, with hydroxy (HEA-TATA) and methyl (ⁿPr-TATA) groups or for comparision to the best performing system ([TATA]PF₆ (TATA⁺ = 4,8,12-tri-(n-butyl)-triazatriagulenium)²⁷. The following section discusses the TATA⁺ framework containing the amino-functionalized A-TATA. In short, various TATAs were evaluated for CO₂ photoreduction with C-1 catalysis, but only A-TATA captured the CO₂ as carbamic acid, thereby enhancing the catalytic activity of C-1 and minimizing the competitive reduction of water to hydrogen.

Performance of the TATA-C-1 systems in visible-light-driven CO₂ reduction to CO

The photocatalytic performance of CO₂ reduction using PSs A-TATA/HEA-TATA/ⁿPr-TATA was investigated in CO₂-saturated CH₃CN/H₂O solution mixtures under irradiation with a xenon lamp (λ > 400 nm). C-1 served as the catalyst for CO₂ reduction, selected for its previously demonstrated high activity in photocatalytic systems²⁰. Additionally, 1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzo[d]imidazole (BIH) and triethanolamine (TEOA) acted as sacrificial electron donors. NaHCO₃ was also added to the reaction system to potentially buffer the solution and enhance CO₂ solubility. In a typical experiment, photocatalytic CO₂ reduction took place in a self-made quartz cell (Supplementary Fig. 17), the reaction mixture comprising 4 mL of CO₂-saturated CH₃CN/H₂O (2:1, v/v), 120 μM TATAs, 10 μM C-1, 100 mM BIH, 15% TEOA, 0.2 M NaHCO₃, was then irradiated with a xenon lamp (λ > 400 nm) at room temperature (Fig. 2 and Table 1). Gas products were quantitatively monitored in real-time using headspace gas chromatography (GC) equipped with flame ionization detectors (FID) and thermal conductivity detectors (TCD). Further details of the photocatalytic settings are provided in the Supporting Information.

Fig. 2. Photocatalytic CO₂ reduction reaction.

a Time-dependent CO production using 0.12 mM of various TATAs. b CO yield as a function of initial [A-TATA] (0.03, 0.12, 0.14 mM). c TON for systems containing 1 μM C-1 catalyst. d Dual turnover activity: CO generation from both PS (12 μM A-TATA) and catalyst (8 μM C-1). e CO production in CO₂-saturated aqueous solution (120 μM TATAs, 1 μM C-1, 15% TEA, 0.2 M NaHCO₃). f GC-MS analysis of ¹³CO under a ¹³CO₂ atmosphere, confirming CO origin. Standard conditions (a–c, e, f): CO₂-saturated CH₃CN/H₂O (2:1 v/v), 0.12 mM PS, 10 μM C-1, 100 mM BIH, 15% TEOA, 0.2 M NaHCO₃, 4 mL (a, b) or 2 mL (c) reaction volume. Condition for d: 12 μM A-TATA, 8 μM C-1, other parameters as above. Light source: Xe lamp (λ > 400 nm, 200 mW/cm²). Error bars represent the standard deviations from at least three independent experiments. Source data are provided as a Source Data file.

Table 1.

Photocatalytic performance for various TATAs and conditions^a

	PS	Reaction conditions	t (h)	CO (μM)	H2 (μM)	SCO (%)	TONCat	TOFCO (h-1)	TONPSb	QYCOc (%)
1	A-TATA	TEOA	10	7.1	0.5	94	176	18	163	1.01
2	A-TATA	BIH	10	0.46	0.06	89	12	1	10	0.07
3	A-TATA	TEOA + BIH	20	386.4	5.2	99	9659	483	6605	42.74
4	A-TATA	TEOA + BIH+NaHCO3	13	397.0	2.2	99	9925	763	7344	51.03
5	HEA-TATA	TEOA + BIH+NaHCO3	13	257.9	3.2	99	6446	496	3904	26.75
6	nPr-TATA	TEOA + BIH+NaHCO3	13	68.1	3.5	95	1703	131	683	7.06
7	A-TATA	TEOA + BIH+NaHCO3	5	67.95	1.5	98	33976	6795	1123	10.67

^aPhotocatalysis experiments containing PSs (120 μM), C-1 (10 μM (entries 1–6), 1 μM (entry 7)), BIH (100 mM), TEOA (15%, vol%), and NaHCO₃ (0.2 M) were irradiated with a xenon lamp (λ > 400 nm, 200 mW/cm²) in either 4 mL (entries 1–6) or 2 mL (entry 7) of CO₂-saturated CH₃CN/H₂O (2:1, v/v) solution. ^bThe concentration of the mixed catalyst was adjusted to PSs (10 μM), C-1 (12 μM). ^cIrradiation was carried out using a 450 nm LED (75 mW/cm⁻²). Source data are provided as a Source Data file.

Prior to optimizing reaction conditions, preliminary screening of sacrificial reductants was conducted (Fig. 2). Initial trials using TEOA as the sole reductant achieved a CO TON of 176 with 93.6% selectivity (Table 1, entry 1). In contrast, BIH alone yielded a lower TON of 12 and 89% selectivity (Table 1, entry 2). Remarkably, combining BIH (100 mM) with TEOA synergistically enhanced reaction kinetics, achieving TON = 9659 and 98.7% CO selectivity a 805-fold improvement over BIH alone (Table 1, entry 3). Further optimization focused on selectivity by introducing NaHCO₃ to buffer the solution and enhance CO₂ solubility. Following CO₂ saturation, the optimized reaction system maintained a stable pH of 8.50 ± 0.14 through the buffering effect of NaHCO₃, with catalytic activity demonstrating pH independence within this narrow range (Supplementary Table 2). Under identical irradiation conditions, the optimized systems employing A-, HEA-, and ⁿPr-TATA catalysts exhibited CO selectivities of 99%, 99%, and 95%, with TON values of 9925, 6446, and 1703, respectively (Table 1, entries 4–6; Fig. 2a). Notably, alongside the improvement in TON from 176 to 9925, and the TOF also increased from 18 to 763 h^-1, indicating the consumption of the total amount of the sacrificial electron donor (n(BIH) = 400 μmol) initially present in the reaction mixture. However, at fixed concentrations of C-1 (10 μmol) and BIH (100 mM), increasing the concentrations of A-TATA from 0.12 to 0.14 mM led to a decrease in the catalytic reaction rate due to the rapid consumption of BIH (Fig. 2b). Despite this limitation, the systems continued to produce CO until approaching a level near the theoretical maximum yield of the sacrificial reductant BIH.

To confirm that BIH consumption is the primary limiting factor in the A-TATA−C-1 system, we conducted a series of controlled experiments (Supplementary Figs. 18 and 19). Furthermore, the addition of 200 μmol or 400 μmol of BIH resulted in quasi-stoichiometric CO production, confirming BIH as the limiting reagent (Supplementary Fig. 19a). However, increasing the BIH amount to 800 μmol did not further enhance CO production, due to solubility limitations in the 4 mL reaction volume. To address this, we expanded the reaction volume to 20 mL, which restored the quasi-stoichiometric relationship between BIH and CO production for additions of 200 μmol, 400 μmol, and 800 μmol of BIH (Supplementary Fig. 19b). These results confirm that BIH is the primary limiting reagent in the reaction. In the A-TATA photocatalytic system, depicted in Fig. 2b, the rate of product formation reached a plateau after 13 h. Subsequently, we replenished the solution with A-TATA, C-1, TEOA, or BIH to restore catalytic activity (Supplementary Fig. 20). Significant catalytic activity was observed upon the addition of BIH, with CO production increasing by ~60% (Supplementary Fig. 20d). The decline in catalytic activity is primarily attributed to the decomposition of the photocatalyst and photosensitizer, as well as the consumption of BIH.

In order to explore the stability of the highly efficient A-TATA in CO₂ photoreduction, we examined the performance of A-TATA, HEA-TATA, and noble-metal-based PSs (Ru-1 and Ir-1) by replenishing the catalyst and electron donor every 10 h (Supplementary Fig. 21). Notably, A-TATA and HEA-TATA maintained stable catalytic activity over three cycles (Supplementary Fig. 21a, b), indicating that A-TATA is a robust molecular photosensitizer for CO₂ reduction. In contrast, the catalytic performance of Ru-1 significantly declined, with the TON dropping from 4855 in the first cycle to 1070 in the third cycle (Supplementary Fig. 21c). This corresponds to a 78.1% reduction in activity after three cycles, suggesting substantial decomposition of Ru-1. Similarly, the TON for Ir-1 decreased from 6496 in the first cycle to 2177 in the third cycle (Supplementary Fig. 21d), reflecting a 66.5% reduction in activity, indicative of significant decomposition of Ir-1.

To further optimize the photocatalytic system, we systematically adjusted the concentrations of [PS] and [Cat]⁶¹. At a fixed A-TATA (0.12 mM) and BIH (100 mM) (Supplementary Fig. 22), the system exhibited the highest TON-based activity at lower catalyst concentrations. CO production increased with [C-1] up to 10 μM, indicating that electron transfer to the catalyst limits system performance. The TON increased notably when the C-1 concentration was reduced to 1 μmol (Fig. 2c and Table 1, entry 7), resulting in unprecedented TONs and TOFs of 33976 and 6795 h⁻¹ after irradiation for 5 h (Supplementary Fig. 23), respectively. This phenomenon of TON increases as catalyst concentration decreases has also been observed in other systems^29,38,62,63. It likely arises because only a portion of the catalyst molecules actively participate in CO₂ reduction, while the remainder acts as a reservoir, gradually deactivating over time. Quantum yields for CO production (Φ_CO) were determined using ferrioxalate actinometry. When A-TATA was employed as the photosensitizer, the quantum yields for the overall photocatalytic reduction of CO₂ to CO reached 51% after 8 h of irradiation at 456 nm (see the Experimental Section and Supplementary Fig. 24). To the best of our knowledge, this represents the highest quantum yield for CO₂ reduction achieved with noble-metal-free systems utilizing organic photosensitizers. Remarkably, under optimized conditions with a mixed catalyst concentration of 12 μM A-TATA and 8 μM C-1, the system achieved remarkable TON for both the photosensitizer and catalyst within a single photocatalytic setup, yielding TON_PS = 6716 and TON_C−1 = 10073 (Fig. 2d and Supplementary Table 3). This dual high activity underscores the efficiency of our integrated photocatalytic design.

The optimized QY of our A-TATA system (QY = 51%) is a factor of 60 times higher than the optimized value in the benchmark report, which uses a combination of a TATA⁺ sensitizer and the C-1 catalyst²⁷. The optimized TON of our A-TATA system (TON = 33967 with S_CO = 98%) is a factor of 1.8 fold improvement than the optimized value in the benchmark report (TON = 19000, S_CO = 93%) (Fig. 2c and Table 1), even though in fully aqueous solution the optimized of our A-TATA system achieving a remarkable TON of 6738, and S_CO = 95.4%, QY_max = 11.5% than the optimized value in the benchmark report (TON = 2600, S_CO = 94%, QY = 0.2% (20% H₂O))²⁷ (Fig. 2e and Supplementary Table 4).

Under optimized photocatalytic conditions, only CO and H₂ were observed as gaseous products by gas chromatography (GC). Analysis of the liquid phase by ¹H NMR and ion chromatography (IC) revealed no detectable liquid phase products following the photocatalytic reaction (Supplementary Figs. 25, 26). Control experiments also demonstrate that the absence of light, PS, Cat or BIH results in no production of CO or H₂ (Supplementary Tables 3 and 5), indicating that all of the above factors are necessary for efficient CO₂ reduction. Furthermore, dynamic light scattering (DLS) measurements did not detect the presence of any particles in the solution after 25 h of irradiation (Supplementary Fig. 27). The addition of excessive metallic Hg(0) to the photocatalytic system does not induce significant changes in photocatalytic activity (Supplementary Fig. 28), demonstrating the homogeneity of the reaction system. Additionally, GC-MS analysis of the product revealed that no CO evolved under argon, and ¹³CO was detected as the product under an atmosphere of ¹³CO₂ in isotope labeling experiments (Fig. 2f), unambiguous confirming that CO originates from CO₂ reduction. Given that TATA-C-1 exhibits excellent photocatalytic properties for CO₂ reduction and outperforms widely used benchmark PSs such as Ru(bpy)₃²⁺ (Ru-1), Ir(bpy)(ppy)₂⁺ (Ir-1) (ppy = 2-phenylpyridine) and EET-TATA this represents significant progress in replacing noble metal PSs with superior TATA-based organic dye alternatives (Supplementary Tables 6 and 7). Detailed studies using steady and transient spectroscopy and electrochemical analysis were conducted to elucidate these structure-activity relationships.

Electrochemical properties

Electrochemical measurements were conducted to evaluate the thermodynamic feasibility of the electron-transfer processes within the three components (PS, C-1, and BIH) of the photocatalytic system (Supplementary Fig. 29 and Supplementary Table 8). The cyclic voltammograms (CVs) of TATAs measured in degassed acetonitrile showed similar potential ranges of 1.22 to 1.25 V and −0.65 to −0.77 V vs saturated calomel electrode (SCE), respectively. These measurements were subsequently used to calculate their excited-state redox potentials, as molecules in the excited state are stronger oxidizing and reducing agents compared to their ground state. To our delight, the oxidation potentials of BIH were notably more negative than the excited-state reduction potentials of TATA PSs (TATA⁺*), calculated to fall between 1.58 and 1.6 V vs. SCE ($E_{red}^{*}=E_{red}+E_{00}$). This observation suggests a sufficiently favorable driving force for electron transfer from BIH to the excited TATA PSs. Furthermore, the excited-state oxidative potential is estimated to be in the range of −1.0 to −1.02 V vs. SCE ($E_{ox}^{*}=E_{ox}-E_{00}$) indicates that TATA⁺* is also susceptible to oxidation quenching by C-1. The subsequent reduction of the cobalt catalyst first generates [Co^I(qpy)]⁺ at −0.59 V vs. SCE, followed by a two-electron reduction to produce [Co^I(qpy^•−)] at −0.81 V vs. SCE⁶⁴. To further elucidate the reaction pathway, luminescence quenching experiments and transient absorption spectroscopy were conducted on various components of the photocatalytic system. These studies aimed to uncover the underlying factors contributing to the high photocatalytic activity of the TATAs−C-1 system.

Steady-state absorption and fluorescence spectroscopy

The UV-vis absorption and emission spectra of the TATAs were investigated to determine their photophysical properties (Supplementary Figs. 30 and 31 and Supplementary Table 1). The photophysical properties of compounds A-TATA, HEA-TATA and ⁿPr-TATA both ethyl derivatives of triazatriangulenium with equal chain lengths, are very similar. These compounds exhibit moderately strong absorption in the visible region with maxima around 530 nm and molar absorption coefficients ranging from 1.38 to 1.44 × 10⁴ M⁻¹cm⁻¹, distinctly separated from the more intense UV transitions observed below 370 nm. Upon irradiation with 500 nm light, the emission maxima of the TATAs shifted only a few nanometers, with lifetimes (τ₀) ranging from 10.31 to 12.26 nanoseconds.

Moreover, the absorption and time-resolved photoluminescence (TRPL) spectra of TATAs remained unchanged upon addition of C-1 or BIH, indicating no ground-state electronic interaction between TATAs and C-1/BIH (Supplementary Figs. 32 and 33). To gain further insight into the photocatalytic mechanism and electron transfer process, we performed photoluminescence quenching experiments on two TATAs using BIH and C-1 quenchers. The rate constants for reductive quenching (electron transfer from BIH to PS*) and oxidative quenching (electron transfer from PS* to Cat) were determined to be near the diffusion-controlled limit (Supplementary Figs. 34 and 35). To address this issue, we have re-evaluated the quenching constants (K_q) values by measuring the change in emission lifetime rather than fluorescence intensity (Supplementary Figs. 36 and 37). As a result, all K_q values are now below the diffusion-controlled limit. These results, combined with the fact that the concentration of BIH is over 10,000 times higher than that of C-1 in photo-catalytic CO₂ reduction, suggest that the system operates via a reductive quenching pathway and was also supported by transient absorption spectra.

Operando UV-vis monitoring of the photocatalytic system (TATAs/C-1/BIH/TEOA in CO₂-saturated CH₃CN) revealed distinct photostability trends (Supplementary Figs. 38–40). For A-TATA, the 450–600 nm absorption band diminished gradually with a concurrent 395 nm peak emergence over 25 min, followed by slow attenuation of this feature (depletion rate: 0.04 h⁻¹). In stark contrast, HEA-TATA and ⁿPr-TATA exhibited rapid spectral changes, complete 450–600 nm band quenching and 395 nm peak formation within 2 min, followed by accelerated signal decay (depletion rates: 1.74 h⁻¹ and 1.82 h⁻¹, respectively). This 44-fold difference in reduced photosensitizer (PS) depletion kinetics, observed under identical conditions except for PS structure, demonstrates A-TATA’s superior robustness (Supplementary Fig. 41). The gradual PS* consumption in A-TATA systems aligns with its sustained catalytic activity, whereas rapid PS degradation in HEA/ⁿPr-TATA correlates with their diminished performance (Supplementary Fig. 21).

Transient absorption spectroscopy

Nanosecond transient absorption spectra of TATA PSs with BIH or C-1 were further analyzed to investigate the excited-state properties of TATAs and their photocatalytic processes (Fig. 3 and Supplementary Fig. 42). When photoexcited with a 527 nm pulse, both A-TATA and HEA-TATA exhibited a strong ground-state bleach (GSB) in the region of ~460–550 nm, corresponding to their ground-state absorption. Additionally, two excited state absorption (ESA) bands appeared on either side of the GSB (Fig. 3a). A-TATA showed a triplet state lifetime of 41.8 μs, which facilitates intermolecular electron transfer (Fig. 3e). Adding BIH to the A-TATA solution resulted in the disappearance of the two positive peaks of A-TATA. Moreover, the GSB of A-TATA with BIH did not decay significantly, persisting for over 1600 μs in the nanosecond transient absorption spectra (Fig. 3b). This suggests that BIH donates electrons to A-TATA, forming an ultra-long-lived reduced state (Fig. 3f). Quenching experiments and transient spectra analysis confirmed the efficient electron transfer from the reduced state of A-TATA to C-1 (Fig. 3c). As shown in Fig. 3g, C-1 efficiently quenched the reduced state, reducing its lifetime to 1.2 μs in the presence of 193.0 μM C-1. The positive peaks reappeared in the transient spectrum of reduced A-TATA with C-1, indicating that the reduced A-TATA efficiently transfers electrons to C-1, restoring the triplet state. Furthermore, the triplet lifetime of A-TATA decreased with increasing concentrations of C-1, although the overall spectrum remained consistent before and after adding C-1 (Fig. 3d, h). These results indicate that the initial electron transfer occurs from BIH to³(A-TATA)*, subsequently generating (A-TATA)^•–, which then efficiently transfers electrons to C-1, driving the photoreduction of CO₂. A similar electron transfer pathway was observed in the HEA-TATA system (Supplementary Fig. 42). Consistent with the PL quenching experiment, this electron transfer is primarily driven by a reductive mechanism. Consequently, we further explored the impact of the terminal amine groups of A-TATA on the system’s photocatalytic performance.

Fig. 3. Time-resolved absorption spectroscopy and excited state dynamic of A-TATA.

a Nanosecond transient absorption spectra of A-TATA. b A-TATA in the presence of BIH (10 mM). c A-TATA in the presence of BIH (10 mM) and C-1 (183.0 μM). d A-TATA in the presence of C-1 (106.0 μM). e The decay of A-TATA at 535 nm. f Kinetic traces of the reduced A-TATA followed at 535 nm. g Kinetic traces of A-TATA with different concentration of C-1 in the presence of BIH (10 mM) followed at 535 nm. h Kinetic traces of A-TATA with different concentration of C-1 followed at 520 nm. These spectra were recorded in CH₃CN after pulsed excitation at 527 nm under Ar. C_ps = 5.0 μM. Source data are provided as a Source Data file.

Key factors influencing the photocatalytic performance of the TATA-C-1 system in CO₂ reduction to CO

Reversible CO₂ capture by terminal amines of A-TATA forming carbamic acid precursors

We have recently demonstrated that TATAs with terminal amine groups exhibit excellent photocatalytic performance in converting CO₂ to CO. To further investigate this, we bubbled ¹³CO₂ into the A-TATA solution and characterized it using ¹³C NMR spectroscopy. The spectrum displayed a peak at 160.8 ppm, consistent with the formation of carbamic acid via CO₂ complexation with amines, as previously studied^65,66. In contrast, the NMR spectrum of the NHE-TATA solution only displayed a peak at 125.3 ppm, corresponding to ¹³CO₂ (Fig. 4a, b). Additionally, it has been demonstrated that the amines on A-TATA can capture and react with CO₂ under our experimental conditions. Further use of argon gas to sweep the A-TATA aqueous solution containing ¹³CO₂ resulted in the disappearance of the peak at 160.8 ppm. These isotope-labeled NMR experiments confirm the lability of the N-C bond in carbamic acid. Weakly basic amines are known to interact with CO₂ to form carbamic acid, which is characterized by a weak N-C bond, but they are not basic enough to form carbamates, which have a strong N-C bond, under a pH of 8.53 in our photocatalytic conditions⁶⁵. As a result, the terminal amines of A-TATA can trap CO₂, but the resulting bonds are unstable. Furthermore, the exclusive formation of the carbamic acid intermediate was corroborated by ESI mass spectrometry (Supplementary Fig. 43). The ESI mass spectrum of the reaction mixture under CO₂-saturated CH₃CN conditions revealed a characteristic ion peak at m/z = 455.22, which is assignable to the A-TATA–CO₂ adduct, thereby providing direct spectroscopic evidence for carbamic acid generation at the terminal amine site. To further substantiate the CO₂–trapping role of A-TATA, we performed GC–TCD analysis (Supplementary Figs. 44 and 45), which showed markedly higher CO₂ uptake in the presence of A-TATA versus control. FTIR spectra (Supplementary Fig. 46) revealed the characteristic CO₂ asymmetric stretch at 2345 cm⁻¹ and an additional C=O band at 1638 cm⁻¹, confirming carbamate formation. Complementary DFT calculations predict a thermodynamically favorable carbamic acid adduct ([A-TATA–CO₂]^•−, ΔG = 1.51 kcal/mol⁻¹) with an insertion barrier of 31.6 kcal·mol⁻¹ (Supplementary Fig. 47). Together, these data demonstrate that A-TATA’s terminal amines chemically capture CO₂ as a carbamic acid intermediate under mild conditions, thereby enriching local substrate concentration and enhancing photoreduction efficiency.

Fig. 4. ¹³C NMR spectrum experiments and electrochemistry.

¹³C NMR spectrum of (a) A-TATA and (b) HEA-TATA in H₂O/D₂O (9:1 v/v) mixture after exposure to ¹³CO₂. The peak at 160.8 ppm is assigned to carbamic acid, while the peak at 125.3 ppm is attributed to ¹³CO₂. Cyclic voltammograms were recorded for (c) A-TATA and (d) HEA-TATA, both in the presence (red) and absence (blue) of 1 mM C-1, 0.5 mM PS, ferrocene, and 0.1 M Bu₄NPF₆ as the supporting electrolyte, scan rate was 0.1 V/s^-1 under CO₂-saturated conditions. The vertical dashed lines represent the onset potentials for CO₂ reduction, determined by the intersection of the tangents between the baseline and the signal current. Source data are provided as a Source Data file.

The photocatalytic ability of the A-TATA−C-1 system for CO₂ reduction was investigated using a gas-flow cell with Ar based gas containing various concentrations of CO₂ (Supplementary Fig. 48 and Supplementary Table 2). Remarkably, the A-TATA−C-1 system demonstrated exceptional adaptability to dilute CO₂ streams, achieving near-identical CO production under 50% CO₂ (TON_CO = 9622) compared to pure CO₂ conditions (Supplementary Fig. 48a). Even under ultra-dilute 10% CO₂, A-TATA retained 21% efficiency (TON_CO = 2080), outperforming ⁿPr-TATA (TON_CO = 105, 1.5% efficiency) by two orders of magnitude. Temporal stability analyses further differentiated the systems, showing 4 h activity losses of 19% (A-TATA), 65% (HEA-TATA), and 73% (ⁿPr-TATA) at 20% CO₂ (Supplementary Fig. 48b). These results support the proposed mechanism, wherein A-TATA’s terminal amines facilitate reversible CO₂ capture via dynamic carbamic acid formation-an advantage absent in analogs lacking accessible amine functionalities.

A-TATA significant reduction in onset potential for photocatalytic CO₂ reduction

Electrochemical studies have demonstrated that in the presence of A-TATA with terminal amines, CO₂ photoreduction is facilitated through the sequestration of CO₂ as carbamic acid, as evidenced by the electrochemical behavior of the C-1 photocatalyst (Fig. 4c, d). As shown in Fig. 4c, the onset potential for CO₂ photocatalytic reduction shifts anodically by 40 mV, from −0.61 V for A-TATA to −0.57 V vs. SCE for A-TATA−C-1. In contrast, no significant changes in this potential are observed with the addition of HEA-TATA (Fig. 4d). This finding confirms that the reduction in onset potential for catalysis is due to the presence of TATA-bound carbamic acid, formed when A-TATA is bubbled with Ar (Supplementary Fig. 49). Based on these experiments, A-TATA demonstrates a significant advantage over HEA-TATA and Ru-1 systems in the photoreduction of CO₂ to CO by leveraging the sequestration of CO₂ as carbamic acid. The dynamic equilibrium between carbamic acid and both dissolved and headspace CO₂ enables the carbamic acid to act as a regenerable intermediate. This creates a catalytic reservoir for CO₂ and serves as a precursor, either directly or indirectly, to the CO₂-bound photocatalytic. Several studies have demonstrated “second-sphere” effects on CO₂ trapping and activation. For instance, Savéant and colleagues achieved this by introducing phenolic groups into iron tetraphenylporphyrin^67,68. Similarly, Weiss and co-workers incorporated amines onto quantum dots²⁶, while Aukauloo and his team introduced urea arms on iron tetraphenylporphyrins to achieve comparable effects^21,57.

Exploration of photocatalytic carbonylation scope

Although CO has proven to be a versatile reagent in organic synthesis, the CO generated from the photoreduction of CO₂ has not shown broad utility as a reactive intermediate in such processes⁹. The primary challenge in using CO for tandem transformations lies in its high flammability and toxicity; as a colorless and odorless gas, CO poses significant safety risks that require stringent handling precautions⁶⁹. However, if CO can be generated through photocatalytic CO₂ reduction and subsequently utilized as a reactant in palladium-catalyzed carbonylation reactions, a cascade catalytic process could be achieved. This approach would significantly mitigate the associated risks while converting CO into valuable carbonyl products. Moreover, there is increasing interest in finding practical applications for CO₂ as a feedstock for chemical production. By leveraging the advantages of a two-chamber reactor⁷⁰, we envisioned that the CO produced from a noble-metal-free photocatalytic CO₂ reduction system could be effectively utilized in a palladium-catalyzed carbonylation reaction, thereby establishing a foundation for innovative chemical transformations. For instance, Aukauloo et al. employed Ru-1 as a PS and a Re-bipyridyl complex as the Cat to reduce CO₂ to CO, which was then simultaneously used in a palladium-catalyzed aminocarbonylation reaction⁷¹. Similarly, Lan and co-workers integrated CO₂ photocatalytic reaction with carbonylation reactions in a two-chamber reactor, employing Ru-1 as the PS, Ni-based cluster assemblies as the Cat, with CO serving as the reactant to achieve a high yield of aminocarbonylation reactions^7,8. However, to the best of our knowledge, the generation of CO from a molecular noble-metal-free photocatalytic system and its efficient incorporation into palladium-catalyzed carbonylation reactions achieved through a simple sealed two-chamber configuration remain rare in photocatalytic CO₂ reduction.

Encouraged by the exceptional performance of the A-TATA−C-1 noble-metal-free photocatalytic system in converting CO₂ to CO, we subsequently utilized the generated CO as a reactant to explore its potential for in situ transformation into valuable fine chemicals through a tandem carbonylation reaction (Supplementary Fig. 50, Supplementary Note 1). The synthesis of these compounds was carried out using a homemade two-chamber system (A-B). In chamber A, the light source was controlled to regulate the on-off cycles for CO production, while the connection to chamber B was carefully managed. This setup ensured both the reliability and reproducibility of the tandem reactions under tightly controlled conditions. The goal was to perform the aminocarbonylation in chamber B as the final step, maximizing the utilization of CO generated from the photocatalytic reduction of CO₂ in chamber A. All the aryl iodides and aryl bromides were aminocarbonylated using the Pd(PPh₃)₄ catalytic system in 1,4-dioxane. Thus, a small optimization of both the photocatalytic CO₂ reduction and the CO-consuming aminocarbonylation reaction was successfully performed in both chambers. This resulted in the synthesis of insecticide (1) in chamber B, achieving a high isolated yield of 82.5% and an impressive atomic efficiency of nearly 85.7% for the entire tandem system. Under the optimized conditions, sixteen different Pd-catalyzed aminocarbonylation couplings of various substrates with different functional groups were performed using the two-chamber system (Fig. 5, Supplementary Table 9). All carbonyl products are synthesized from CO precursor produced by photocatalytic reduction CO₂ in chamber A at room temperature.

Fig. 5. Design plan for carbonylative transformations via CO₂ photoreduction.

a Two-Chamber system for carbonylative couplings using CO from photocatalytic CO₂ reduction. b Explore the substrate scope for CO₂ photoreduction in carbonylation reactions. c ¹³C isotope labeling experiments. All yields are of isolated products. See Supporting Information for details. Source data are provided as a Source Data file.

Variations in the substitution patterns of aryl iodide derivatives, including methyl, methoxy, chlorine, phenoxy, and pyridyl groups, result in isolated compounds (2–13) with yields ranging from 61% to 87%. However, when aryl bromide derivatives were used as reactants under the same conditions, their yields dropped to approximately 40% for the isolated products (14–16), with only 20% of the CO being converted. This suggests that the reaction is susceptible to CO concentration, significantly decreasing activity at low CO levels. Further studies were conducted to validate the synthesis of ¹³C-carbon isotope-labeled diethyltoluamide and its corresponding amide compounds using a ¹³CO₂ isotope tracing experiment (Fig. 5c, Supplementary Figs. 51 and 52). This tandem reaction successfully validated our initial goal of converting CO₂ to CO through the visible-light-driven catalytic photoreduction, followed by the in-situ transformation of CO into valuable, easily isolated fine organic chemicals. Moreover, using ¹³CO₂ as a source of isotopically labeled amide drug molecules demonstrates considerable potential for applying tandem reactions in metabolic.

Next, we carried out scale-up experiments to demonstrate the practical applicability of the photocatalytic tandem reaction (Fig. 6, Supplementary Table 10). Under tentatively optimized conditions (0.24 mM A-TATA, 20 μM C-1, 20% TEOA, 0.2 M BIH, and 0.4 M NaHCO₃), a 20 mL reaction mixture in chamber A delivered a high CO output of 2546 μmol with a TON of 6365 and nearly 100% selectivity after 20 h of illumination. This represents a 6.4-fold enhancement compared to the benchmark 4 mL system (397 μmol CO output). Subsequent aminocarbonylation in chamber B demonstrated remarkable scalability, the amount of aryl iodide was increased to 3.4 mmol over 17 times higher than depicted in Fig. 6 and conducted the reaction without a solvent (Supplementary Fig. 53). In this experiment, 3-iodotoluene and 3-iodopyridine were used as reactants, paired separately with n-butylamine and diethylamine as coupling partners. Despite the challenging conditions, we successfully achieved excellent yields of amides 2 and the drug molecule nikethamide (7), with CO conversion rates of 83% and 80%, respectively (Supplementary Fig. 54).

Fig. 6. Highlighting the practicality of the aminocarbonylation synthesis.

a, b the carbonylative protocol demonstrates high efficiency in the absence of reaction solvent. All yields are of isolated products. See Supporting Information for details. Source data are provided as a Source Data file.

Importantly, we constructed a modular mobile-phase setup using available lab infrastructure to simulate continuous-flow operation (Supplementary Fig. 55). Integration into this flow-based system resulted in a substantial improvement in performance: the product yield increased to 80%, and the overall TON rose from 6365 to 9868, consistent with full consumption of the initial 4000 μmol of BIH. These findings confirm the efficient coupling between photocatalytic CO generation and its downstream utilization, highlighting the system’s adaptability to flow-mode operation and its promise for future industrial applications. Notably, the entire process operates under green chemistry principles featuring solvent-free synthesis and selective CO₂ reduction without byproduct formation⁹. A preliminary techno-economic analysis suggests that the tandem setup reduces processing costs while enhancing profitability, underscoring the versatility and practical potential of this molecular-level innovation (Supplementary Note 2). Systematic reactor-scale optimization studies are underway and will be reported in due course.

Mechanistic considerations

BIH serves as a donor of two electrons and one proton in CO₂ reduction^72,73. To better understand its rols, we conducted electrochemical analyses and mechanistic studies. Cyclic voltammetry (CV) was performed in acetonitrile with 5 mM BIH, both in the absence and presence of 0.2 M TEOA (Supplementary Fig. 56). The inclusion of TEOA led to the appearance of a distinct reduction wave at −1.66 V vs. SCE, indicative of BI^• radical formation⁷⁴, whereas no such feature was observed in the absence of TEOA. Photocatalytic CO₂ reduction experiments revealed a stark contrast in efficiency between systems containing TEOA and those without. In TEOA-free conditions, CO production decreased by 805-fold (TON_CO = 12 vs. optimized conditions), confirming that TEOA functions as a base for deprotonation of the one-electron oxidized form of BIH^{+• 72}, playing a critical role in facilitating electron-proton transfer. Additionally, quantitative analysis of the reaction mixture indicated near-complete BIH consumption (397 μmol products from 400 μmol BIH), consistent with BIH acting as a two-electron donor. The absence of formate byproducts in all conditions further supports the proposed mechanism wherein BIH participates in sequential single-electron transfers rather than hydride donation.

A photocatalytic mechanism consistent with our experimental results is depicted in Fig. 7, with Supplementary Fig. 57 providing the hypothesized role of a terminal-amine functionalized PS in CO₂ capture. In this model, the PS first binds CO₂ as a carbamic acid (Supplementary Fig. 44), thereby locally enriching the CO₂ concentration at the catalyst interface and priming the system for efficient reduction. Upon visible-light irradiation, A-TATA is promoted to its singlet excited state (¹A-TATA*), which undergoes intersystem crossing (ISC) to the triplet manifold (³A-TATA*). The long-lived ³A-TATA* then accepts an electron from BIH, generating the reduced radical anion A-TATA^•–. This electron‐rich species transfers its charge to the cobalt-quaterpyridine catalyst C-1, driving the multistep CO₂‐to‐CO conversion (for mechanistic details of C-1, see Robert et al.^20,75).

Fig. 7. Proposed mechanism for the tandem reaction.

The reaction pathway illustrates the photocatalytic reduction of CO₂ to CO, followed by a value-added aminocarbonylation step. ISC refers to intersystem crossing.

The CO generated by this molecular photocatalytic system can be further transformed into a valuable and easily separable fine chemical via a carbonylation reaction, which takes place in a two-chamber tandem reactor. In the aminocarbonylation pathway, the process typically begins with the oxidative addition of a palladium(0) center to Ar-X, followed by CO insertion into the organopalladium complex to form the corresponding carbonyl-palladium intermediate, and then undergoes transmetallation and reductive elimination, leading to the formation of the amide product^7,8. The integration of visible-light-driven CO₂ reduction for CO generation with a tandem reactor system offers precise control over the CO delivery to the carbonylation reaction chamber, which can be finely tuned by adjusting the light intensity and irradiation duration. This approach, which combines photocatalytic CO₂ reduction and carbonylation, provides substantial value for synthetic chemists. It offers a safe and storable source of CO gas and provides a practical method for applications in CO-related chemistry, including isotope labeling studies.

Discussion

In summary, we have developed a highly efficient A-TATA photosensitizer for the photoreduction of CO₂ to CO using visible light without any noble metals. This system demonstrates exceptional performance, achieving a turnover number of 33976 and an optimized quantum yield of 51%, with 98% selectivity−one of the highest values reported in molecular photocatalysis. Notably, even under 20% CO₂ gas, this system can reduce CO₂ selectively to CO with 46% the efficiency of a system using pure CO₂ atmosphere.

Notably, in fully aqueous solutions, the system achieves a groundbreaking TON_CO of 6738-2.5× higher than prior benchmarks (TON_CO = 2700)²⁸, with 95.4% selectivity and QY = 11.5%, outperforming all reported noble-metal-free aqueous systems. Unlike conventional designs where catalyst or photosensitizer activity limits overall performance, our system simultaneously delivers high dual turnover: TON_PS = 6716 for the photosensitizer and TON_Cat = 10073 for the catalyst. This underscores the exceptional photocatalytic performance of the A-TATA−C-1 system.

Mechanistic studies reveal that terminal amines in the A-TATA functionality are crucial for CO₂ binding and activation, facilitating faster photocatalytic conversion by (i) lowering the onset potential for CO₂ photoreduction, and (ii) establishing a dynamic equilibrium between CO₂ and carbamic acid, which increases the local concentration of available CO₂. Moreover, the optimized photocatalytic system using CO₂ as a CO precursor in a two-chamber setup, not only provides a safe and storable source of CO gas but also promotes environmentally friendly and economical applications in CO-related pharmaceutical research. The aminocarbonylation process within this system achieves up to 85% atomic efficiency and operates efficiently even without a solvent. To the best of our knowledge, the systematic modification of multi-functional photosensitizers with free amines for CO₂ photoreduction to CO, followed by ex-situ carbonylation in a noble-metal-free system, has never been carefully examined. This work offers avenues for designing low-cost, efficient photosensitizers and provides insights into using CO₂ as a valuable synthon.

Methods

Materials

All solvents and chemicals were of commercial reagent quality and used without further purification unless indicated otherwise. All synthetic reactions were conducted under argon atmosphere and protected from light. n-BuLi (1.6 M in hexane), ferrocene, KPF₆, HPF₆, Pd(PPh₃)₄, and Tris(2,2′-bipyridine)dichlororuthenium(II) hexahydrate, ([Ru(bpy)₃]Cl₂·6H₂O, 98%), were obtained from Sigma-Aldrich. N,N,N′,N′-tetramethylethylenediamine (TMEDA), diethylcarbonate, ethylenediamine, n-propylamine were procured from Adamas. Carbon-¹³C dioxide (99%) was supplied by Wuhan Newradar Special Gas Co., Ltd. 1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzo[d]imidazole (BIH) was synthesized following published procedures⁶³. The synthetic route for TATAs is detailed in Supplementary Fig. 1. The synthetic intermediates and final complexes were characterized using NMR and mass spectroscopy.

Instruments

¹H NMR and ¹³C NMR spectra were recorded on a Bruker Avance–500 (500 MHz for ¹H and 126 MHz for ¹³C) at room temperature using CDCl₃ and DMSO-d₆ with TMS as an internal standard. High-resolution mass spectra were obtained in a Q-TOF LC-MS in ESI mode. UV-vis-NIR absorption spectra were recorded on a Hitachi 3900H spectrophotometer, while fluorescence spectra were captured using a Hitachi F-7000 spectrofluorometer. Time-resolved photoluminescence (TRPL) spectra were measured with an FLS1000 fluorescence spectrometer (Edinburg, UK). Transient absorption spectra were obtained using the LP980 laser flash photolysis instrument (Edinburgh, UK). Dynamic light scattering experiments were conducted with a Brookhaven Elite Sizer zata-potential and particle size analyzer. Electrochemical measurements were performed on a CHI 660E electrochemical workstation at room temperature, with ferrocene added as an internal standard and all potentials referenced to the Fc⁺/Fc pairs. A 300 W Xenon lamp (CEL-HXF300-T3, CEAULICHT) with 400 nm filter served as the visible light source.

Photocatalytic CO₂ reduction

The reaction was conducted in a home-made quartz cuvette cell with an 18 mL headspace containing PS (0.12 mM), C-1 (10 μM), BIH (100 mM), TEOA (0.6 mL, 15% V/V, 1.13 mM) and NaHCO₃ (67.2 mg, 0.2 M) in 4 mL of a 2:1 mixture of CH₃CN and H₂O. A similar reaction was performed in a 95 mL reactor with a larger containing the PS (0.24 mM), C-1 (20 μM), BIH (200 mM), TEOA (6.0 mL, 15% v/v, 11.3 mM) and NaHCO₃ (672 mg, 0.4 M) in 20 mL of the same solvent mixture. Before irradiation, the reaction solutions were bubbled with CO₂ for 20 min. A xenon lamp equipped with a 400 nm cut-off-filter as light source for irradiation. After each experiment, the headspace of the photochemical reactor was analyzed for possible products using gas chromatography (Shimadzu GC-2014), fitted with serial Molecular sieve-13X (3.0 m × 3.2 mm) and Porapak-N (2.0 m × 3.2 mm) packed columns and equipped with thermal conductivity detectors (TCD) for H₂ detection, and flame ionization detectors (FID) for CO and other hydrocarbons detection. Potential products in the solution were measured using ion chromatography (930 Compact IC Flex, Metrohm). ¹³C-labeled CO₂ experiments were conducted following the same procedure except ¹³CO₂ was used and the gas products were detected by GC-MS (Agilent 5977B MSD).

Nanosecond transient absorption spectroscopy

Nanosecond transient absorption spectra were recorded using an LP980 laser flash photolysis spectrometer (Edinburgh Instruments, UK) with a Tektronix TDS 3012B oscilloscope. Sample solutions were excited by an OpoletteTM 355II + UV nanosecond pulsed laser (OPOTEK, USA), tunable between 200 and 2200 nm, with a typical pulse energy of 5 mJ. Signals were digitized on a Tektronix MDO 3022 oscilloscope. Prior to measurement, all samples were degassed with Ar for approximately 15 min to eliminate dissolved oxygen. Kinetic decay traces and data fitting were processed using L900 software.

Electrochemical measurements

Cyclic voltammetry (CV) was performed at room temperature using a CHI660 workstation. The measurements were conducted in a three-electrode electrochemical cell containing 0.1 M tetrabutylammonium hexafluorophosphate (ⁿBu₄NPF₆) as the electrolyte under Ar or CO₂. A glassy carbon electrode served as the working electrode, a platinum wire as the counter electrode, and a Ag/AgNO₃ electrode (in CH₃CN solution with 0.10 M ⁿBu₄NPF₆ and 0.01 M AgNO₃) as the reference electrode. Before measurement, the glassy carbon electrode was sonicated in deionized water, and the solution was purged with argon for 20 min. The potentials were calibrated against the standard ferrocene/ferrocenium (Fc/Fc⁺) couple. The potential ofthe saturated calomel electrode (SCE) was calculated using the equation SCE = Fc/Fc⁺ + 0.4 V.

Fluorescence quenching

A solution of TATAs PS was degassed with Ar for 15 min in a sealed quartz cuvette with a septum cap, BIH was then added to the mixture. The quenching rate constant (k_q) was calculated using the Stern-Volmer Eq. 1:

$$\frac{I_0}{I}or\frac{{\tau }_0}{\tau }=1+K_q{\tau }_0\left[Q\right]$$ 1

In this equation, I₀ and I represent the fluorescence intensity of the photosensitizer in the absence and presence of the quencher, k_q is the quenching rate constant, τ₀ is the lifetime of photosensitizer, and [Q] is the concentration of quencher.

Quantum yield determination

The quantum efficiency (Φ) of photocatalytic CO₂ reduction was calculated using the following Eq. 2:

$$\Phi \left(\%\right)=\frac{2\times R}{I}\times 100\%$$ 2

where R (mol s^-1) represents the rate of CO evolution and I_abs (einstein s^-1) is the photon flux absorbed by the complex (I_abs = I_inc – I_trans). Therefore, 2 mL (V₁) of an aqueous K₃[Fe^III(C₂O₄)₃] solution (10 mM) was irradiated with monochromatized light (λ = 456 nm) in the same reactor used for the photocatalytic reactions. After irradiation, 0.5 mL (V₂) of this solution was mixed with 10 mL (V₃) of a phenanthroline buffer solution and kept in the dark for 1 h. The absorbance of this solution at 510 nm was then measured (Fig. S20) and the slope of absorbance versus irradiation time was recorded. The incident photon rate (I) was calculated using 3:

$$I\left(\mathrm{einstein}{\mathrm{s}}^{-1}\right)=\frac{\mathrm{\Delta }A\cdot V_1\cdot V_3}{\Phi \cdot \mathrm{\lambda }\cdot {\epsilon }_{510}\cdot V_2\cdot l\cdot t}$$ 3

V₁: volume of potassium ferrioxalate solution irradiated;V₂: volume of aliquot used for analysis;V₃: diluted the final volume of aliquot V₂; ∆A: measured absorbance at 510 nm; ε₅₁₀: molar extinction coefficient of the Fe²⁺ complex (11590 L mol^‒1 cm^‒1); l: length of cuvette (1 cm); Φ_λ: quantum yield for Fe²⁺ complex (1.12); t: irradiation time (s).

A 4 mL solution of CH₃CN/H₂O (2:1, v/v) containing 0.12 mM A-TATA, 100 mM BIH, 15% TEOA, and 0.2 M NaHCO₃ was irradiated under a CO₂ atmosphere. The incident light intensity (I_inc) and transmitted light intensity (I_trans) were measured as 9.056 × 10^-8 and 4.721 × 10^-8 Einstein s^‒1, respectively. The same mixture in a CO₂-saturated CH₃CN/H₂O (2:1, v/v) (4 mL) was irradiated with the monochromic light of 456 nm, and the generated products were determined using a calibrated GC. After 8 h, 318.54 μmol of CO was generated, with K₃[Fe^III(C₂O₄)₃] serving as the actinometer to determine I_abs = 4.335 × 10⁻⁸ einstein s^‒1. Under these conditions, the calculated quantum yield Φ_CO = 51.03%.

General procedure for photocatalytic tandem reaction experiments

Tandem reaction experiments were meticulously conducted using a two-chamber system. Chamber A, with an 18 mL capacity was designated for the photocatalytic CO₂ reduction reaction, utilizing A-TATA (0.12 mM), C-1 (10 μM), BIH (100 mM), TEOA (0.6 mL, 15% v/v, 1.13 mM) and NaHCO₃ (67.2 mg, 0.2 M) dissolved in a 4 mL solution of CH₃CN/H₂O (2:1 v/v). The reaction mixture in chamber A was purged with CO₂ for 20 min to ensure adequate saturation, followed by 15 h of irradiation with a xenon lamp. Chamber B, with a 10 mL capacity was reserved for carbonylative coupling reaction contained aryl iodide or aryl bromide (0.2 mmol), Pd(PPh₃)₄ (0.015 mmol), Et₃N (83.2 μL, 0.6 mmol), and the corresponding amine (0.6 mmol),with 1,4-dioxane (5 mL) as the solvent. The two chambers were connected via a short linker, and all joints were initially sealed. Chamber B was then flash-frozen at 77 K using a liquid nitrogen bath and subjected to three freeze-pump-thaw cycles, reducing the internal pressure to approximately 100 mTorr. Upon warming chamber B to room temperature, chamber A underwent photocatalysis. After the 15-hour photocatalytic reaction in chamber A, the joint to chamber B was carefully opened. Chamber B was then stirred in an oil bath at 65 °C for 12 h. The reaction products were detected and quantified using ESI-MS and NMR analysis following purification through column chromatography on silica gel. This protocol ensured both the reliability and reproducibility of the tandem reactions under controlled experimental conditions.

DFT calculations

The calculations in this study were conducted using the Gaussian 09 program package, based on density functional theory (DFT). For the calculations using the Gaussian 09 program package, B3LYP/6-31 g (d,p) was used to optimize the molecular structure for a stable configuration^76,77.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

K.-K.C. conceived the project. K.-K.C., Z.W., and S.H. designed the catalytic reactions. S.H. synthesized the compounds and carried out the CO₂ reduction and tandem catalytic reaction experiments with the help of P.L. and Q.H. L.Z. and Y.W. participated in the spectroscopy experiments. K.-K.C., Z.W., and S.H. prepared the manuscript. All the authors reviewed and contributed to this paper.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are available in the main text or the supplementary materials. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-61229-8.