A cluster-nanozyme-coenzyme system mimicking natural photosynthesis for CO₂ reduction under intermittent light irradiation

Abstract

Natural photosynthesis utilizes solar energy to convert water and atmospheric CO₂ into carbohydrates through all-weather light/dark reactions based on molecule-based enzymes and coenzymes, inspiring extensive development of artificial photosynthesis. However, development of efficient artificial photosynthetic systems free of noble metals, as well as rational integration of functional units into a single system at the molecular level, remain challenging. Here we report an artificial system, the assembly system of Cu₆ cluster and cobalt terpyridine complex, that mimics natural photosynthesis through precise integration of nanozyme complexes and ubiquinone (coenzyme Q) on Cu₆ clusters. This biomimetic system efficiently reduces CO₂ to CO in light reaction, achieving a production rate of 740.7 μmol·g⁻¹·h⁻¹ with high durability for at least 188 hours. Notably, our system realizes the decoupling of light and dark reactions, utilizing the phenol-evolutive coenzyme Q acting as an electron reservoir. By regulating the stabilizer of coenzyme Q, the dark reaction time can be extended up to 8.5 hours, which fully meets the natural day/night cycle requirements. Our findings advance the molecular design of artificial systems that replicate the comprehensive functions of natural photosynthesis.

Introduction

Nature has evolved over billions of years to carefully select specific functional molecules or molecular complexes, known as enzymes or coenzymes, to assemble customized biotransformation systems, in which these simple functional molecules can perform unexpected roles through elaborate integration^1,2,3,4,5,6. For example, the natural photosynthetic system integrates light-harvesting complexes (LHCs), catalytic enzymes, and multiple coenzymes, including quinones, adenosine triphosphate (ATP), and nicotinamide adenine dinucleotide phosphate (NADPH), to efficiently convert water and CO₂ into energy-rich carbohydrates through all-weather light/dark reactions^7,8,9,10,11. This efficiency has inspired the emerging field of artificial photosynthesis though it faces significant challenges^{6,12,13,14,15,16}. The primary bottleneck is the lack of suitable functional materials and the difficulty in synergistically integrating various functional units into a single system without compromising their efficiency. In natural photosynthesis (Fig. 1a), numerous LHCs (chlorophyll and/or carotenoid pigments) harvest solar light to drive water oxidation, producing protons and electrons¹⁷. Electrons are then stored and transferred efficiently with the help of coenzyme factors¹⁸, allowing dark reactions for CO₂ reduction by specific enzyme catalysts to occur¹⁹. This close cooperation between units results in a highly efficient system, where the ingenious separation of light and dark reactions enables plants to survive under intermittent sunlight²⁰. The key to replicating such a delicate system, therefore, lies in the development of corresponding functional materials and their tactful integration.

Fig. 1: Design of artificial photosynthetic system by mimicking natural photosynthesis.

a, b Schematic illustration of natural photosynthesis (a) and artificial photosynthesis in this work (b).

Notably, each component of the natural photosynthetic system is constructed from molecule-based units, enabling efficient electron transfer and reactant activation through direct electronic and chemical bond coupling. This precise molecular arrangement likely contributes to the high efficiency of the system². Recognizing this principle, designing and assembling systems at the molecular level presents a promising solution to the current challenges in artificial photosynthesis. Molecule-based artificial photosynthetic systems, which typically employ two different soluble metal complexes as LHCs and catalysts, have shown success in homogeneous photocatalysis²¹. However, these systems often suffer from low photostability and a heavy reliance on noble metals^21,22. There is an urgent need to develop alternative materials that can mimic natural biotransformation systems more effectively. Atomically precise metal clusters with their well-defined structures, consisting of a metal core and protected ligands, inherit many benefits of molecular complexes, such as tunable bandgaps and enhanced stability^23,24,25. These clusters, particularly advantageous for their atomically engineered structures and externally functionalized ligands, provide numerous open sites for fine-tuning the microenvironment at active sites, making them ideal candidates for LHCs and potentially offering a robust platform for enzyme-mimicking activities²⁵.

Another critical aspect of replicating the natural photosynthetic system is the long-term storage of electrons for dark reactions, which is essential to maintain the continuous operation of all solar-driven reactions. Nature utilizes simple redox-active quinone compounds, such as ubiquinones (also known as coenzyme Q) and plastoquinone, capable of uptaking and releasing protons/electrons (Fig. 1a), to create an electron reservoir (Q pool)^26,27. Despite their efficacy, these potent coenzyme molecules are often overlooked in the design of artificial photosynthetic systems, possibly due to the complexity of designing specific catalytic sites that can drive the Q cycle²⁷. Additionally, these redox-active compounds can induce undesirable side reactions, such as competing with the light reaction²⁸. Therefore, designing appropriate active sites and utilizing these functional factors effectively could provide a viable approach to adapt them for use in artificial systems.

To this end, herein, we employed an atomically precise Cu₆ cluster—an earth-abundant transition metal—as a model LHC, constructing peripheral metal complexes as catalytic centers (nanozymes) and coupled phenol to create a Q pool, thereby duplicating the natural photosynthetic system (Fig. 1b). In our system, LHCs and catalytic centers are connected directly via coordination bonds, which can be expected to enhance electron transfer efficiency and reduce reliance on random collisions. The open sites of the cluster allow us to regulate their chemical environment, facilitating the activation and stabilization of coenzyme intermediates. Crucially, the phenol-evolutionary Q cycle drives the dark reaction without compromising the efficiency of the light reaction. As a proof of concept, we assembled terpyridyl Co complexes (Cotpy) on the Cu₆ cluster to form a crystalline Cu₆-Cotpy nanocluster for CO₂ photoreduction. This system achieves a rate of 740.7 μmol·g⁻¹·h⁻¹, which is ca. 93 times greater than the bare Cu₆ cluster, with a selectivity up to 99% for CO production and maintains the activity for at least 188 h. Additionally, we realized a decoupled dark reaction driven by semiquinone species, with the reaction time extendable up to 8.5 hours by adjusting the stabilizer of the semiquinone radical, aligning with natural day/night cycles. This is an example of using bioinspired coenzyme Q to construct an electron reservoir that decouples the light and dark reactions, paving the way for the development of all-weather artificial photosynthetic systems. The relationship between the structure and activity of catalytic centers was systematically explored through in situ spectroscopic characterization and theoretical calculations.

Results and discussion

Construction and characterization of artificial photosynthetic system

We chose Cu₆ cluster as a model light harvester, which captured our attention due to the following advantages: (i) copper is an earth-abundant transition metal; (ii) it has a narrow bandgap (1.86 eV)²⁹ for promising visible-light response (>600 nm) and sufficiently negative energy level for catalytic CO₂ reaction (Supplementary Fig. 1); (iii) its six exoteric carboxylate groups can serve as peripheral coordination sites to construct functional species³⁰; (iv) it remains stable and soluble in most polar solvents, including water³¹. However, we also recognized the concerning fact that the majority of the metal sites on bare metal clusters are covered by ligands, leading to poor catalytic activity³².

Based on these recognitions, we assembled the peripheral metal sites on the Cu₆ cluster through a stepwise methodology³⁰ (see detailed methods in experimental information). In our selected Cu₆(Hmna)₆ cluster (hereafter, abbreviated as Cu₆, H₂mma = 2-mercaptonicotinic acid), the Cu(I) cation, being a soft acid, has a strong coordination ability with soft bases or borderline bases (RS⁻ and N in mna ligand). As a result, the RCOO⁻ in mna ligand, being a hard base, remains uncoordinated and can be used as an anchor site for assembly with hard acids and borderline acids. Primitively, we selected single metal sites as active centers of the nanozyme to assemble on Cu₆, as reported in our previous study³². As illustrated in Fig. 2a and Supplementary Fig. 2, when naked Co^II salts were added to the photosynthetic system of the Cu₆ cluster, a compound of (Cu₆(mna)₆[Co(H₂O)₄]₃·8H₂O (hereafter, abbreviated as Cu₆-Co) was obtained. Six carboxylate sites in [Cu₆(mna)₆]⁶⁻ unit are coordinated with Co^II ions, forming a two-dimensional (2D) framework that shows an octahedral coordination environment with four coordinated water molecules. Taking advantage of the open sites of the Cu₆ cluster, we can readily assemble different metal cations on the cluster. By substituting Ni^II salts for Co^II salts, a compound of Cu₆(mna)₄(Hmna)₂ [Ni(H₂O)₄]₂Ni(H₂O)₆·12H₂O (hereafter, abbreviated as Cu₆-Ni) was formed. Compared to the Cu₆-Co complex, only two Ni^II ions are coordinated with Cu₆(mna)₄(Hmna)₂ on different sides, forming a one-dimensional (1D) structure (Fig. 1b and Supplementary Fig. 3). However, the coordination environment of each assembled Ni atom is the same as that of the Co in Cu₆-Co, in which two Ni^II centers are coordinated with four water molecules and two O atoms in mna ligands, respectively. To vary the coordination environment of catalytic centers, we introduced bidentate chelating ligand 2,2’-bipyridine (bpy) as a modifier, taking its various coordination modes and the good photoreduction activity of polypyridine complexes into account³³. When bpy was introduced into the system of Cu₆-Co, a 1D compound with the formula as Cu₆(mna)₆[Co(bpy)(H₂O)₂]₂Co(bpy)(H₂O)₄·4H₂O (donated as Cu₆-Cobpy) was obtained (Fig. 2c and Supplementary Fig. 4). Compared to the compound of Cu₆-Co, two coordinated water molecules with the Co^II center are replaced by one bpy ligand, which also serves as a terminal ligand to prevent the structure from extending in some extent. There are two types of Co^II ions in the compound of Cu₆-Cobpy, both of which adopt distorted octahedral coordination environments. One type is formed through the coordination with two water molecules, two O atoms in mna linkers, and two N atoms in the bpy ligand. The other type, via coordination with four water molecules and two N atoms in the bpy ligand, acts as counterions. Interestingly, when replacing Co^II ions with Mn^II ions, we obtained an isostructural compound of Cu₆(mna)₆[Mn(bpy)(H₂O)₂]₂Mn(bpy)(H₂O)₄·4H₂O (donated as Cu₆-Mnbpy). The Mn^II ions exhibit identical coordinated environments to those of Co^II ions in Cu₆-Cobpy. To further tune the coordination environment of the nanozyme, a tridentate chelating ligand 2,2’:6’,2”-terpyridine (tpy) was implemented. As a result, a 0D compound of Cu₆(mna)₆[Co(tpy)H₂O]₂[Co(H₂O)₆]·14H₂O (donated as Cu₆-Cotpy) was obtained (Fig. 2d and Supplementary Fig. 5). Compared to Cu₆-Co, three coordinated water molecules with Co²⁺ are replaced by one tpy ligand, which acts as terminal ligand and further prevents the structure from extending. Specifically, the Co^II ions in Cu₆-Cotpy still adopt distorted octahedral coordination environments, like those in Cu₆-Cobpy, and are coordinated with one coordinated water molecule, one O atom, one S atom in mna linker, and three N atoms in tpy ligand. It is worth noting that one S atom of the Cu₆ clusters is directly bonded to the Co center in Cu₆-Cotpy, which is markedly different from the above compounds.

Fig. 2: Synthesis and structure illustrations of the proposed artificial photosynthetic system.

a–d Crystal structures of Cu₆-Co (a), Cu₆-Ni (b), Cu₆-Mnbpy or Cu₆-Cobpy (c), and Cu₆-Cotpy (d).

The phase purity of the five compounds was proven by powder X-ray diffraction (PXRD) patterns and corresponding crystal photographs (Supplementary Fig. 6 and Supplementary Tables 1−7), and the presence of exotic metals in these compounds was confirmed by energy dispersive spectroscopy (EDS) (Supplementary Fig. 7). X-ray photoelectron spectroscopy (XPS) reveals that the Cu element in central Cu₆ clusters is monovalent Cu^I and the assembled metal ions of Mn, Co, and Ni are divalent (Supplementary Fig. 8), in agreement with the results of bond valence sums (BVS) analysis (Supplementary Tables 8−12). No noticeable changes are found in the Cu 2p and CuLMM XPS spectra of the five compounds in comparison with the bare Cu₆ cluster (Supplementary Fig. 8a, b), indicating that the intrinsic properties of the Cu₆ cluster are well preserved during these assembly processes. S 2p XPS spectra also show no significant changes, except that a weak peak is found at 166.0 eV in Cu₆-Cotpy (Supplementary Fig. 8c), which likely derives from the Co−S bond, confirming its special bonding mode. The most pronounced change occurs in the O1s XPS spectra (Supplementary Fig. 8d), where the fraction of hydroxide O at 532.8 eV is significantly weakened and shifts to carbonate O at 531.1 eV after metal site assembly, demonstrating that the carboxyl groups of Cu₆ are involved in anchoring metal ions in all the five compounds. This conclusion was further confirmed by the Fourier-transform infrared (FT−IR) spectroscopy (Supplementary Fig. 9), which shows that the asymmetric stretching vibrations of the carboxyl group around 1620 cm⁻¹ have shifted to lower wavenumber.

CO₂ photoreduction performance

According to the compositional and structural information, we evaluated their light-harvesting capabilities through ultraviolet-visible (UV−Vis) absorption spectroscopy. As shown in Fig. 3a, the bare Cu₆ cluster exhibits considerable visible-light absorption up to ca. 700 nm, together with its more negative lowest unoccupied molecular orbital (LUMO) level (−0.60 V vs. NHE, pH = 7) than the reduction potentials of CO₂ (Supplementary Fig. 1), indicating that the as-selected Cu₆ cluster is a competent light harvester for CO₂ photoreduction. The other five synthesized compounds also show good visible-light response, which can be attributed to the broad light harvesting of the central Cu₆ cluster. The distinct d-d excitation at ca. 1100 nm on Cu₆-Co and Cu₆-Ni demonstrates their characteristics of single metal sites. The broadest light harvesting is realized on Cu₆-Cotpy, which covers almost the entire solar spectrum, implying that more efficient photocatalytic performance can be expected.

Fig. 3: Photoresponse and photocatalytic performance.

a UV-Vis diffuse reflectance spectra of Cu₆, Cu₆-Co, Cu₆-Ni, Cu₆-Cobpy, Cu₆-Mnbpy and Cu₆-Cotpy. b Average CO and CH₄ production rates in photoreduction CO₂ on Cu₆-Co, Cu₆-Ni, Cu₆-Cobpy, Cu₆-Mnbpy and Cu₆-Cotpy with visible-light irradiation (λ ≥ 420 nm) on a CO₂-saturated ACN solution containing 10 mg of photocatalysts, 0.05 M BIH (electron donor), and 0.03 M PhOH (proton source), in comparison with pristine Cu₆ cluster and other control experiments under the same conditions. c GC-MS analysis of ¹³CO (m/z = 29) produced by the photoreduction of ¹³CO₂ over Cu₆-Cotpy. d An action spectrum for CO production by Cu₆-Cotpy along with the absorption spectra of Cu₆-Cotpy and Cu₆. e Light-driven catalytic durability over Cu₆-Cotpy. Each cycle takes 47 h. f Comparison of the photoreduction CO₂ performance of Cu₆-Cotpy with noble-metal-free non-molecular systems reported in the recent literature listed in Supplementary Table 15. The error bars represent the standard deviation of the experiments.

Given the promising light-harvesting capabilities and the presence of open metal sites in the nanozyme, we evaluated the performance of visible-light-driven (λ ≥ 420 nm) CO₂ reduction without additional photosensitizers and cocatalysts. 10 mg of photocatalysts were ultrasonically dispersed in 100 mL of CO₂-saturated ACN solution, containing 0.05 M BIH and 0.03 M PhOH as an electron donor and a proton source, respectively. As shown in Fig. 3b and Supplementary Fig. 10, the pristine Cu₆ cluster exhibits only negligible activity for CO (8.0 μmol·g⁻¹·h⁻¹) and CH₄ (0.5 μmol·g⁻¹·h⁻¹) production, indicating that the bare Cu₆ cluster is almost inert for CO activation. After decoration of the cluster with the hydrated single metal sites of Co (Cu₆-Co), the production rate is slightly improved to CO (10.4 μmol·g⁻¹·h⁻¹) and CH₄ (1.0 μmol·g⁻¹·h⁻¹). The same situation occurs when the metal sites of Co are replaced with Ni (Cu₆-Ni), indicating that the hydrated single metal sites are ineffective for promoting CO₂ activation. Inspired by the fact that the activity of metalloenzymes in biosystems is highly dependent on the electronic and/or steric nature of metal sites, we introduced polypyridine ligands to regulate the coordination environment of active centers, as their affiliative coordination and rigid backbone easily form stabilized complexes³⁴. More importantly, its delocalized π system allows reserving the reduced equivalents to promote charge separation and mediate the transfer of the electrons to CO₂ molecules³³. As displayed in Fig. 3b, the activity of Cu₆-Cobpy for CO production is enhanced to 29.1 μmol·g⁻¹·h⁻¹, which is 3.6 times that of pristine Cu₆ cluster, and its activity can be further improved by the carbonylation which replaces the coordinated H₂O (Supplementary Fig. 11). Additionally, when the metal center is replaced by Mn^II, the production rate for CO reaches up to 72.5 μmol·g⁻¹·h⁻¹ with Cu₆-Mnbpy, demonstrating that the strategy of modulating the coordination environment of active centers is pronounced for improving catalytic activity. Encouraged by this improvement, we further changed the polypyridine ligand to terpyridyl. Remarkably, a robust enhancement in activity was achieved on Cu₆-Cotpy with a production rate of 740.7 μmol·g⁻¹·h⁻¹ for CO production, which is ca. 93-fold enhancement over bare Cu₆ cluster. It is worth mentioning that Cu₆-Cotpy requires a long pre-activation time (5 to 40 h, during which negligible CO can be detected) before its robust performance. This feature may be attributed to the potential reconstruction of the local microenvironment around active sites.

To verify the origin of carbonous products, we carried out control experiments by excluding catalysts, CO₂, or light under the same conditions (Fig. 3b). It turns out that negligible or no CO/CH₄ products can be detected, suggesting that the carbonous products originate from the photocatalytic conversion of CO₂. On the other hand, the control experiments without 1,3-dimethyl-2-phenyl-2,3-dihydro-1Hbenzo[d]imidazole (BIH) or phenol (PhOH) exhibit very low activity for CO₂ conversion, indicating that both the electron donor and proton source are essential for this system. In addition, the control experiment with the physical mixture of Cu₆ and nanozyme complex (Cotpy-mna) exhibits significantly lower performance (48.68 μmol·g⁻¹·h⁻¹) than that of the assembled Cu₆-Cotpy (Supplementary Fig. 12), demonstrating that the direct bonding of Cu₆ with nanozyme via coordination is a requirement for efficient CO₂ conversion. In the meantime, isotope-labeling experiments were performed in a ¹³C-labeled CO₂ (¹³CO₂) atmosphere, and an obvious signal peak at m/z = 29 (¹³CO) was detected (Fig. 3c), confirming that the generated CO originates from photoreduction CO₂. Additionally, the apparent quantum yield (AQY) of Cu₆-Cotpy under different wavelengths of incident light was evaluated by $\frac{N({{\mathrm{electrons}}})}{N({{\mathrm{photons}}})}\times 100\%$ (Fig. 3d and Supplementary Table 13). The action spectrum indicates that the AQY trend matches well with the absorption spectrum of Cu₆ rather than Cu₆-Cotpy, suggesting that the reaction is indeed driven by the photogenerated electrons of Cu₆. As such, the Cu₆ clusters serve as light harvesters like the LHCs in natural photosynthesis.

We also examined the durability of Cu₆-Cotpy in four successive cycles, each of which took 47 h. Owing to the slight solubility of Cu₆-Cotpy in acetonitrile (ACN), each new cycle was tested by adding fresh BIH and renewing the reaction atmosphere with pure CO₂. As shown in Fig. 3e, the catalytic CO₂ conversion over Cu₆-Cotpy can keep active for at least 188 h, indicating its good photostability. This durability was also proven by comparing the XRD patterns and FT−IR spectra of the Cu₆-Cotpy before and after the reaction (Supplementary Fig. 13). Likewise, Cu₆-Cobpy also shows high stability, maintaining activity for at least 358 h (Supplementary Fig. 14). Such catalytic activity and durability of Cu₆-Cotpy are much higher than, or comparable to, those of most noble-metal-free non-molecular photocatalysts reported in recent years, as illustrated in Fig. 3f and Supplementary Table 15.

Photogenerated charge dynamics and photocatalytic mechanism

Given that the performance measurements have demonstrated that the Cu₆-Cotpy cluster can function as an efficient system for CO₂ photoreduction, it is imperative to elucidate the underlying mechanism why it can stand out from other counterparts. Considering that all five compounds have the same light harvester of the Cu₆ cluster, the superior performance of Cu₆-Cotpy may be attributed to the efficiency of charge separation, and/or subsequent catalytic activity of the nanozyme. We first focus on the charge transfer performance between the Cu₆ cluster (light harvester) and assembled catalytic centers (nanozyme). As shown by the transient photocurrent responses in Fig. 4a, Cu₆-Cotpy exhibits a substantially higher response than the other five counterparts, indicating its marked ability for photogenerated charge separation. This result was confirmed by electrochemical impedance and steady-state fluorescence tests (Supplementary Fig. 15). From the structure of Cu₆-Cotpy, this superiority should be enabled by two features: (i) the Co^II ions are bonded directly to the S of Cu₆ cluster, forming a shorter charge transfer channel from Cu₆ to Co centers; (ii) the large π conjugation of the tpy ligands serves as a reservoir for storing photogenerated electrons, thereby preventing the electrons on Cotpy from flowing back to Cu₆²⁷.

Fig. 4: Photogenerated charge dynamics.

a Photocurrents of Cu₆, Cu₆-Co, Cu₆-Ni, Cu₆-Mnbpy, Cu₆-Cobpy, and Cu₆-Cotpy on electrodes in 0.1 M tetrabutylammonium hexafluorophosphate ACN solution under visible-light irradiation. b, c The fs-transient absorption (TA) pseudo-color maps Cu₆ (b) and Cu₆-Cotpy (c) in CO₂-saturated ACN solution containing 0.05 M BIH and 0.03 M PhOH. d ESR spectra of Cu₆-Cotpy in the dark or under different irradiation time in ACN solution containing 0.05 M BIH and 0.03 M PhOH at room temperature. Inset in (d) is the photograph of the suspension of Cu₆-Cotpy before (left) and after (middle) photocatalytic reaction and recovered (right) by exposure to air. e, f High-resolution in situ XPS spectra of Co2p (e) and CuLMM (f) for Cu₆-Cotpy before and after light irradiation.

To confirm these features, time-resolved optical transient absorption (TA) spectroscopy experiments were performed to examine the photogenerated charge dynamics. The intrinsic charge transfer dynamics of the catalysts were first studied under an Ar atmosphere. A 355 nm pump pulse was employed to excite electrons from the highest occupied molecular orbital (HOMO) of the Cu₆ cluster to its lowest unoccupied molecular orbital (LUMO) level, and white light continuum probe pulses were used to monitor its spectral evolution of the ¹∑ exciton band in the range of 400 to 680 nm³⁵. The results indicate that the pristine Cu₆ cluster has a long-lifespan triplet to sustain excited electrons, suggesting that it can serve as a good electron donor (Supplementary Fig. 16). Upon the attachment of the nanozymes, photogenerated electrons from the Cu₆ cluster were transferred fast (within about 1 ps) to the assembled metal centers, but then rapidly relaxed to the HOMO for recombination with a lifetime of ca. 20 ps. This process occurred almost equivalently on the five as-prepared cluster−nanozyme complexes (Supplementary Figs. 16, 17). However, the characteristic charge-transfer feature (ca. 495 nm) became unobservable when BIH was introduced, probably due to the filling of electrons from BIH into the HOMO of Cu₆ cluster with BIH [BIH⁺-Cu₆*] (Supplementary Fig. 18 and Supplementary Table 17). It is worth noting that this change occurred from the beginning (less than 1 ps), suggesting that the BIH should be pre-adsorbed on the surface of the Cu₆ cluster or form a binary adduct for efficient charge transfer since charge transfer by solution diffusion takes at least 10⁻¹⁰ s (k_diff = 10¹⁰ M⁻¹·S⁻¹)³⁶. When the PhOH was introduced, the charge dynamic behavior of Cu₆ was obviously different from that of Cu₆ alone and in the presence of BIH in the ultrafast timescale, implying that PhOH was also pre-combined onto the surface of the Cu₆ cluster. More importantly, the difference between the five compounds became profoundly evident in the presence of PhOH and CO₂. As shown in the 2D pseudo-color maps of the TA spectra in Fig. 4b, c and Supplementary Fig. 19, the negative ΔOD bleaching signal ascribed to the reduced state of Cu₆ (Cu₆⁻) at ~590 nm was only obviously observed in the pristine Cu₆ cluster or Cu₆-Cotpy, but was almost negligible in those with 1D and 2D structures (Cu₆-Co, Cu₆-Ni, Cu₆-Cobpy and Cu₆-Mnbpy), indicating that the clusters with open sites (0D structures) are more favorable to oxidize BIH to produce Cu₆⁻ (Supplementary Figs. 20, 21 and Supplementary Table 18). Additionally, global fitting results reveal that the formation of the reduced state is faster for Cu₆-Cotpy (ca. 52 ps) than Cu₆ (ca. 300 ps), and their lifetimes are prolonged (more than 6 ns). It should be noted that the photogenerated electrons in the reduced state of pristine Cu₆ can only reside in the copper center with little catalytic activity. In stark contrast, in the case of Cu₆-Cotpy, once the reduced state is formed, its LUMO is fully occupied, and the photogenerated electrons previously transferred onto the catalytic Cotpy moiety have to stay where they are, forming a stable charge-transfer state. Obviously, the above TA results verify the simultaneous formation of the effective charge separation and long-lived reduced state in the Cu₆-Cotpy system, which likely accounts for its superior catalytic performance.

Based on the information for the charge transfer dynamics from photoexcited Cu₆ to assembled Cotpy, we further devoted to determining the specific sites of electron acceptors on the Cotpy. In situ electron spin resonance (ESR) spectroscopy was first employed to resolve the variations in the valence states of the metal sites by examining their spin state before and after trapping electrons. As shown in Fig. 4d, before light irradiation, a broadband signal was observed at g = 2.049, which can be attributed to a high-spin state of Co^II 37, given that the metal of Cu^I in this system is nonmagnetic. After visible-light illumination, the signal intensity was greatly attenuated, clearly indicating a light-induced valence transformation from Co^II in the high-spin state to Co^I in the low spin state³². This argument is also supported by an apparent photochromism that the orange suspension of Cu₆-Cotpy turned brown during the photocatalytic reaction and then recovered when exposed to air, which can be attributed to the photoreduction of Co species and its reoxidation by oxygen in air (Fig. 4d, inset). These results fully demonstrate that the accepted electrons from Cu₆ are trapped at the Co sites, which should, in turn, serve as active sites for subsequent CO₂ reduction. This conclusion was further testified by in situ XPS spectroscopy. As shown in high-resolution XPS spectra (Fig. 4e), with increasing irradiation time, the peaks of CuLMM were gradually shifted toward higher binding energy, indicating that the electron density of Cu has been reduced under photoexcitation. By contrast, the peak of Co2p (780.7 eV) was shifted toward the opposite direction by about 0.4 eV and 0.9 eV after 5 min and 10 min of irradiation (Fig. 4f), respectively, manifesting that the electron density of the Co atoms rises by prolonging irradiation time. These results conclusively demonstrate that the photogenerated electrons of the Cu₆ clusters are transferred to the Co sites of Cotpy complexes.

Upon acquiring the charge transfer behavior between the light harvester and active sites, we turned our attention to the catalytic mechanism over the nanozyme. In situ far-infrared diffuse reflectance Fourier-transform spectroscopy (DRIFTS) was performed to investigate the dynamic evolution of the coordination environment of Co sites under stimulated photocatalytic reaction conditions. As displayed in Fig. 5a, with respect to the unilluminated sample, peaks centered at 668, 570, and 555 cm⁻¹ appeared after 15 min of irradiation, which can be ascribed to the ν_Co-O vibration absorption of Cotpy complex and/or [Co(H₂O)₆]²⁺ counterion³⁸. The appearance of these peaks indicates that the corresponding coordination bonds varied after light irradiation, which were elongated or weakened after the Co^II center with high coordination was reduced to a low-coordination Co^I site by photogenerated electrons and/or counterions coordinated with oxygenated species such as phenol derivatives. With continuous irradiation, the intensity of the characteristic peak at 555 cm⁻¹ was increased while the peak position was red-shifted to lower wavenumbers, indicating that the Co-O bonds have been further weakened. When the irradiation time reached 55 min, a new vibrational absorption of ν_Co-N at 542 cm⁻¹ was observed^39,40, which can be attributed to the coordination of solvent ACN molecules to the Co center of the Cotpy complex. With a further increase in irradiation time, the peak intensity of ν_Co-N vibration absorption was gradually promoted, while the intensity of ν_Co-O was reduced, suggesting that Co sites, which were originally coordinated to oxygen, have switched to coordination with nitrogen. Even after 120 min, the peak at 555 cm⁻¹ disappeared, which can be attributed to the fact that the coordinated H₂O molecules were completely replaced by ACN or the Co−O bond between Co center and carboxyl was broken. This long process should be the pre-activation time for the catalysts during the initial stage of the photocatalytic reaction.

Fig. 5: Photocatalytic mechanism of the nanozyme.

a, b In situ DRIFTS of the Cu₆-Cotpy during the CO₂ photoreduction with different irradiation times in the mid-far-infrared (a) and near-infrared (b) regions. c DFT-calculated free energy profiles for CO₂ reduction by Cu₆-Cobpy (orange lines) and Cu₆-Cotpy (cyan lines). d Schematic mechanistic pathway for the CO₂ reduction reaction catalyzed by Cu₆-Cotpy.

To decode the CO₂ activation mechanism on the dynamic nanozyme sites, NIR in situ DRIFTS measurements dependent on irradiation times were also carried out. As shown in Fig. 5b, a distinct absorption band at 1408 cm⁻¹, ascribed to the symmetric stretching of *HCO₃⁻, was first detected after 3 min of irradiation. This *HCO₃⁻ likely was formed by the hydration of CO₂ at the Co site of Cotpy⁴¹. With increasing irradiation time, this characteristic peak of *HCO₃⁻ was gradually blue-shifted toward higher wavenumbers, and other intermediate species of monodentate carbonate (m-CO₃²⁻, at 1508 and 1479 cm⁻¹) and bidentate carbonate (b-CO₃²⁻, 1543, and 1317 cm⁻¹) were observed⁴¹, indicating that the absorption strength of *HCO₃⁻ on Co sites was enhanced and the *HCO₃⁻ further evolved into adsorbed carbanions by deprotonation. This is evidenced by the receded absorption band of *HCO₃⁻ and the increased intensity for carbanions. Meanwhile, the crucial intermediate of *COOH for the reduction of CO₂ to CO appeared at 1602 cm⁻¹ after 4 min of irradiation^41,42. The directly adsorbed *CO₂ species was also found at 1686 cm⁻¹, which may be attributed to rapid CO₂ adsorption and scarce dissociated hydroxyl group. These characteristic peaks were confirmed by the control experiments with isotopically labeled ¹³CO₂ and Ar (Supplementary Fig. 22). Overall, based on the above in situ DRIFTS results, a tentative mechanism for CO₂ activation on the nanozyme can be proposed. CO₂ is initially co-adsorbed at the active site with H₂O to form *HCO₃⁻ intermediate, and then evolves into *CO₃²⁻ by deprotonation. In the following step, *CO₃²⁻ accepts photogenerated electrons and is coupled with protons to form *COOH during irradiation. Finally, the *COOH intermediate is deprotonated to form CO molecules.

Based on the information on charge dynamics, coordination environment evolution, and intermediate species, we further depicted the reaction pathway on the nanozyme by density functional theory (DFT) calculations, taking advantage of their well-defined structures for theoretical modeling. CO₂ molecule was found to be difficult to activate on pristine Cu₆-Cotpy, until two electrons and one ACN molecule were introduced into the system (Supplementary Table 19). Specifically, the adsorption energy of CO₂ on pristine Cu₆-Cotpy, [Cu₆-Cotpy]⁻, [Cu₆-Cotpy]²⁻, and [Cu₆-Cotpy]²⁻-ACN was determined to be −0.06, 0.38, 0.37, and −0.18 eV, respectively. Correspondingly, the bond angle of attached CO₂ (∠OCO) at the sites was calculated to be 180.0°, 175.2°, 140.7°, and 143.5°, respectively. Only on [Cu₆-Cotpy]²⁻-ACN, both the negative adsorption energy and the twisted bond angle can be obtained. This suggests that the pristine Cu₆-Cotpy is catalytically inert, and the coordinated ACN benefits CO₂ activation, which is consistent with our previous findings on the [Ni(tpy)₂]²⁺ complex⁴³. Energy optimization calculations indicate that the coordination of ACN to Co center promotes the desorption of coordinated H₂O and the cleavage of the bond between Co and carboxyl (Supplementary Fig. 23), which is consistent with the DRIFTS results in the far-infrared region. The first step that the Co sites accept two electrons was found to be the rate-determining step (RDS) for CO₂-to-CO reduction reaction on Cu₆-Cobpy and Cu₆-Cotpy (Fig. 5c and Supplementary Figs. 24,25), confirming that the pre-activation process is crucial for the subsequent catalytic reaction. The significantly lower formation energies in most of the reaction steps on Cu₆-Cotpy than those on Cu₆-Cobpy account for its superior photocatalytic performance.

Taken together, the coordination structure, charge dynamics, in situ, spectroscopic characterization, and theoretical calculations provide a clear reaction pathway. As illustrated in Fig. 5d, the nanozyme of Cotpy complex (1) first accepts two photogenerated electrons from Cu₆ and loses a coordinated H₂O molecule to form [Cu₆-Cotpy]²⁻ (2). [Cu₆-Cotpy]²⁻ is subsequently coordinated with ACN in the solvent (3), in which the Co−O bond between the Co center and carboxyl is broken or stretched. This pre-activated site is favorable to adsorb CO₂ to form the [Cu₆-Cotpy]⁻-ACN-*COO⁻ (4) species, which is then coupled with the proton of phenol to form the key intermediate adduct [Cu₆-Cotpy]⁻-ACN-*COOH (5). The next step is the dehydroxylation of the *COOH intermediate by a one-proton coupled one-electron transfer process to produce the intermediate of [Cu₆-Cotpy]⁻-ACN-*CO (6). Finally, CO and ACN molecules are orderly desorbed from the Co site through the complex (7) and back to (2), completing one cycle. This catalytic cycle is functionally consistent with the metalloenzymes in biotransformation systems, and the flexible and recyclable catalytic sites most likely are the underlying reason for its high durability.

Dark reaction performance and mechanism

Upon identifying the photocatalytic features of Cu₆-Cotpy, we are now in a position to explore another important function in natural photosynthesis: the storage of photogenerated electrons for dark reaction to address the challenge of discontinuous incident light or light excess²⁰. Inspired by the role of quinones in natural photosynthesis for electron shuttling and storage²⁷, we chose phenol to mimic this function. To prove this concept, we evaluated the catalytic performance of Cu₆-Cotpy in the dark after 2 h of visible-light illumination in the presence of PhOH as a precursor of benzoquinone (BQ). As demonstrated in Fig. 6a, the CO evolution reaction can continue for at least 3 h after turning off the incident light. The evolution rate has a first-order kinetic relationship with the number of stored electrons (Supplementary Fig. 26), indicating that the number of stored electrons is the direct driving force for the dark reaction. Moreover, with further light illumination, the Cu₆-Cotpy catalyst can be recycled for dark reaction (Supplementary Fig. 27), demonstrating that we have successfully replicated the function of decoupling light reaction and dark reaction. To prove that PhOH acts as the coenzyme precursor to drive the dark reaction, we performed a series of control experiments. It turned out that the system cannot convert CO₂ in the dark reaction without pre-illumination, catalysts, BIH, or PhOH (Supplementary Fig. 28), implying that the light reaction is indispensable for energy storage. When the reaction atmosphere of CO₂ was replaced by Ar after the light reaction, no product was detected, demonstrating that the produced CO in the dark reaction originates from the reduction of CO₂ by storage electrons. Additionally, we examined the relationship between irradiation time and CO yield. As shown in Fig. 6c, when the dark reaction time was prolonged from 100 min to 180 min, the absolute yield of the CO was enhanced from 121.4 to 188.9 μmol·g⁻¹ by extending the illumination time from 1 h to 2 h, suggesting that the amount of storage electrons depends on the pre-irradiation time (Supplementary Fig. 29). However, when the irradiation time was extended beyond 2 h, the increase in CO yield stagnated, indicating that storage saturation is reached after 2 h of illumination. This decoupled phenomenon was also found on the Cu₆-Cobpy (Supplementary Fig. 30). However, it does not work on the Cu₆-Co, which likely is limited by its lagging performance in the light reaction. As such, we can conclusively infer that the high activity of the light reaction is the prerequisite for the dark reaction to proceed.

Fig. 6: Dark reaction performance and mechanism.

a Average production rates of CO on Cu₆-Cotpy in the dark after visible-light illumination for 2 h. b Light-induced in situ ESR spectra on Cu₆-Cotpy at room temperature (CO₂-saturated ACN solution containing 0.05 M BIH and 0.03 M PhOH). c Proposed mechanism for the formation of coenzyme Q that dominates electron storage and release. d Average CO production rates in the dark reaction on the Cu₆-Cotpy system with the addition of different amounts of Co(NO₃)₂·6H₂O after visible-light illumination for 2 h. Inset in (a) is the local amplification of the coordinates inside the dotted box.

This conclusion guides us to investigate the system after the reaction, where some substrates may have been generated to store photogenerated electrons during the light reaction. As in our designed blueprints, BQ was detected in the solution after light reaction by ¹H NMR spectrum (Supplementary Fig. 31), which should evolve from PhOH oxidation by photogenerated holes in Cu₆ clusters⁴⁴. In the living system, BQ can store electrons and protons by shuttling between its three consecutive oxidation states of quinone, semiquinone (SQ), and hydroquinone (HQ)²⁷. However, in our system, the dark reaction did not occur after direct addition of HQ, suggesting that the dark reaction was driven by the first half of the shuttle between BQ and SQ. In situ ESR spectra evidenced this inference by detecting the characteristic quintuplet peaks of the SQ free radical with an intensity ratio of 1:4:6:4:1 and a g-value of 2.002 after light irradiation (Fig. 6b)⁴⁵. Their peak intensities increased with prolonged irradiation time, indicating that the amount of SQ radicals can be accumulated in this system. After switching off the light, the peak intensities were reduced due to the consumption of SQ species by the dark reaction. It is worth noting that no SQ species were detected in the system containing only PhOH (Supplementary Fig. 32a), suggesting that BIH is required to supply sufficient electrons to reduce the BQ. In contrast, some faint multiplet peaks were found in the system containing only BIH (Supplementary Fig. 32b), which can be ascribed to some heterocyclic radicals by oxidation BIH. The lifetime of these species is quite short, so their peak intensity cannot increase with extending irradiation time. Furthermore, although decoupled reaction can be achieved by direct addition of BQ without PhOH, the absolute yield of CO is only 1/8 that of the PhOH-mediated reaction (Supplementary Fig. 33). Other proton donors, such as CF₃CH₂OH and water, were found to be unable to replace the role of PhOH (Supplementary Fig. 34). These results fully demonstrate the importance of PhOH for regulating the decoupled reaction.

Taken together, catalytic performance and spectroscopic characterization have elucidated the mechanism of the dark reaction. As illustrated in Fig. 6c, in the light reaction, PhOH is deprotonated to form phenol anion (^II), which is then oxidized to phenol radical (III) by photogenerated holes. During this process, PhOH acts as a dual-functional electron and proton donor, which benefits the light reaction rather than directly adding quinones as electron reservoirs. Subsequently, as the phenol radicals are very active, they will capture oxygen from oxygen-rich materials, such as water, to form BQ (IV)⁴⁴. BQ is then reduced by excess photogenerated electrons to form SQ radicals (V). The SQ radicals, in fact, cannot be isolated due to their short lifetime but can be stabilized by coordination with one or more metal centers⁴⁶. In our system, the SQ radicals most likely are coordinated to counterions of Co^II, which is supported by the fact that the total moles (3.77 μmol) of the saturated electrons for CO production are equivalent to the moles of the catalysts involved (3.89 μmol) rather than the amount of added PhOH. This feature is also confirmed by the results that the addition of extra Co^II ions to the light reaction system can further prolong the dark reaction time compared to the normal reaction system of Cu₆-Cotpy (Fig. 6d). When the addition of Co^II ions reaches 20 μmol, the dark reaction time is prolonged to 8.5 h, which is close to the length of summer nights (ca. 9 h) in most parts of China. Beyond the 20 μmol of Co^II ions, the dark reaction time cannot be further prolonged. Note that the addition of Co^II ions has no obvious effect on the performance of the light reaction (Supplementary Fig. 35). Finally, the SQ radicals release electrons to the active sites of the nanozyme to drive the dark reaction. Overall, the PhOH and its derivatives in this system serve as smart electron regulators to address the issues of light discontinuity and excess. It is worth pointing out that to achieve the Q cycle, the material should provide abundant sites to stabilize the SQ radicals in addition to the ability of generating sufficient photoexcited electrons. In this case, P25 TiO₂ can only offer low capability for the dark reaction while the dark reaction cannot be achieved on g-C₃N₄ (Supplementary Fig. 36).

In summary, we have constructed an artificial photosynthetic system, Cu₆-Cotpy, by assembling nanozyme complexes and coenzyme Q on an atomically precise Cu₆ cluster to mimic the light/dark reaction processes of natural photosynthesis. In our design, the coordination environment of the nanozymes is meticulously tuned through the optimization of their metal centers and peripheral ligands. As a result, efficient CO₂ photoreduction was achieved with terpyridyl cobalt complex (Cotpy) as the nanozyme. The Cu₆-Cotpy offers a photoreduction rate of 740.7 μmol·g⁻¹·h⁻¹ and a selectivity of up to 99% for CO production, and a high durability for at least 188 h. This makes it the most efficient and durable cluster-based photocatalyst reported to date. Additionally, this system features open coordination sites that enable Q cycle, shuttling between quinone and semiquinone radicals to act as an electron reservoir, thereby decoupling light and dark reactions. Our findings offer a versatile strategy for integrating various functional units within a single system at the molecular level, effectively replicating the comprehensive functions of natural photosynthesis under intermittent light irradiation.

Methods

Material synthesis

Synthesis of Cu₆ cluster

Cu₆ cluster was synthesized by modifying a liquid/liquid diffusion method reported by ref. ³¹ First, H₂mna (0.062 g, 0.4 mmol) was dissolved in 4 mL DMF to form a solution A, and solution B was prepared by dissolving Cu(OAc)₂·H₂O (0.0798 g, 0.4 mmol) in 4 mL ACN. After that, solution A was put on the bottom of the test tube (20 mL with a diameter of 13 mm), and then the buffer of 4 mL DMF/ACN mixed solution (V:V = 1:1) was slowly added onto the top of solution A. Finally, solution B was carefully layered onto the buffer layer. This solution was allowed to stand at room temperature for about 5 days, and yellow massive crystals can be obtained. The crystals were filtered off, washed with ethanol, and dried in air. Yield: Ca. 85% based on Cu.

Synthesis of Cu₆-Co and Cu₆-Ni

Peripheral metal sites were assembled onto the Cu₆ cluster through a stepwise diffusion approach. Step 1: the precursor solution A of the Cu₆ cluster was prepared by mixing CuCl (0.02 g, 0.2 mmol) and H₂mna (0.032 g, 0.2 mmol) into a beaker with 6 mL H₂O to form yellow precipitates, and then two drops of concentrated NH₃ aqueous solution (25%) were added to give a transparent yellow solution. Step 2: peripheral metal solution B was prepared by dissolving Co(OAc)₂·4H₂O (0.04 g, 0.16 mmol) to the mixed solution of water and ethanol (6 mL, V:V = 2:1) via ultrasonic treatment for 20 min. Finally, solution B was carefully layered onto solution A in a 20 mL test tube (diameter 13 mm), which was then stood at room temperature for 3 days to obtain yellowish-brown flaky crystals. The obtained crystals were filtered off and washed with ethanol and dried in air. Yield: Ca. 75% based on Co.

The synthetic process of Cu₆-Ni is similar to that of Cu₆-Co except that Co(OAc)₂·4H₂O in the Cu₆-Co was replaced by Ni(OAc)₂·4H₂O (0.04 g, 0.16 mmol). After the final mixture solution was stood at room temperature for 5 days, yellowish-green crystals were obtained with a yield ca. 75% based on Ni.

Synthesis of Cu₆-Mnbpy and Cu₆-Cobpy

The synthesis method of Cu₆-Mnbpy is similar to the Cu₆-Co except that 2,2’-bipyridine was introduced into solution B. Typically, the precursor solution A of Cu₆ cluster was prepared by mixing CuCl (0.1 g, 1 mmol) and H₂mna (0.16 g, 1 mmol) into a beaker with 8 mL H₂O to form yellow precipitates, and then four drops of concentrated NH₃ aqueous solution (25%) were added to give a transparent yellow solution. Subsequently, solution B was prepared by dissolving Mn(OAc)₂·4H₂O (0.049 g, 0.2 mmol) and 2,2’-bipyridine (0.0312 g, 0.2 mmol) into the mixture solution of water and ethanol (6 mL, V:V = 2:1) via ultrasonic treatment for 20 min. Finally, solution B was carefully layered onto solution A in a 20 mL test tube (diameter 13 mm), which was then stood at room temperature for 3 days to obtain yellowish-green flaky crystals. The obtained crystals were filtered off, washed with ethanol, and dried in air. Yield: Ca. 75% based on Mn.

When solution B of Cu₆-Mnbpy was replaced by Co(OAc)₂·4H₂O (0.199 g, 0.8 mmol) and 2,2’-bipyridine (0.125 g, 0.8 mmol), and the final mixture solution was stood at room temperature for 3 days, yellowish-green crystals were obtained with a yield ca. 70% based on Co. It is denoted as Cu₆-Cobpy.

Synthesis of Cu₆-Cotpy

Cu₆-Cotpy was synthesized through a similar stepwise methodology except that 2,2’:6’,2”-terpyridine was introduced into solution B. Typically, the precursor solution A of Cu₆(Hmna)₆ clusters was prepared by mixing CuCl (0.01 g, 0.1 mmol) and H₂mna (0.016 g, 0.1 mmol) into a beaker with 3 mL H₂O to form yellow precipitates, and then one drop of concentrated NH₃ aqueous solution (25%) was added to give a transparent yellow solution. Subsequently, solution B was prepared by dissolving Co(OAc)₂·4H₂O (0.0249 g, 0.1 mmol) and 2,2’:6’,2”-terpyridine (0.023 g, 0.1 mmol) into the mixture solution of water and ethanol (3 ml, V:V = 2:1) via ultrasonic treatment for 20 min. Finally, solution B was carefully layered onto solution A in a 15 mL test tube (diameter 10 mm), which was then stood at room temperature for 2 days to obtain brownish-red block crystals. The obtained crystals were filtered off, washed with ethanol, and dried in air. Yield: Ca. 62% based on Co.

Synthesis of Cotpy-mna complex

The synthesis process of Cotpy-mna complex was the same as that of Cu₆-Cotpy except that CuCl was not added in the solution A. Acetone was used as a precipitant to separate the Cotpy-mna complex through centrifugation, and a gray powder sample was obtained.

Photocatalytic CO₂ reduction measurement

The photocatalytic performance was evaluated using an automatic online gas analysis system (Perfect Light LabSolar 6 A, China) coupled with a gas chromatograph (GC)⁴⁷. About 10 mg of photocatalysts were dispersed in 100 mL of ACN solution containing 0.05 M BIH and 0.1 M PhOH. The entire reaction setup was first vacuum-degassed and then filled with high-purity CO₂ gas to a pressure of 0.8 bar. A 300 W xenon lamp (PLS-SXE300+, China) with a 420 nm cut-off filter (i.e., λ > 420 nm) was used as the visible-light source. The intensity of incident light was measured to be 250 mW·cm⁻². A GC (FULI, GC-9790II, China) equipped with a TDX-01 packed column and a 5 Å molecular sieve column was used to determine the amounts of produced CO and H₂, respectively. Ar was used as the carrier gas. CO was converted to CH₄ by a methanation reactor and then analyzed by a flame ionization detector (FID), and H₂ was detected by a thermal conductivity detector (TCD). For the durability testing, equal amounts of fresh BIH solution were added to each cycle. Isotope-labeling experiments were performed using ¹³CO₂ instead of ¹²CO₂, and the products were analyzed using gas chromatography–mass spectrometry (GC−MS, 7890A and 5975C, Agilent).

Determination of apparent quantum efficiency

The apparent quantum yield (AQY) was obtained using 10 mg of Cu₆-Cotpy as a photocatalyst and a 300 W Xe lamp with different band-pass filters (centered at 365, 450, 500, 550, 500, 600, and 650 nm) as a monochromatic light source under the same reaction conditions. Depending on the amounts of CO produced by the photocatalytic reaction in an average of two hours, the AQY was calculated by the following equation:

$${{{\mathrm{AQY}}}}=\frac{{2n}_{{CO}}\times {N}_{A}\times h\times c}{I\times S\times t\times \lambda }\times 100\%$$

where N_A is Avogadro constant (6.022 × 10²³ mol⁻¹), h is Planck constant (6.626 × 10⁻³⁴ J·s), c is the speed of light (3 × 10⁸ m·s⁻¹), I is the average illumination intensity, S is the irradiation area (28.26 cm²), t is the illumination time (7200 s), and λ is the wavelength of the monochromatic light.