Tridentate Coordination‐Driven Covalent Organic Frameworks Incorporating Recycled Cobalt(II) for Enhanced Photocatalytic CO₂ Reduction

ABSTRACT

Photocatalytic CO₂ reduction represents a promising route toward carbon neutrality and renewable fuel production. Covalent organic frameworks (COFs) are attractive photocatalytic platforms because their tunable structures can stabilize isolated metal sites and regulate the local environment of catalytic centers. Herein, a series of COFs featuring N₁O₂‐tridentate coordination sites were synthesized by condensing 2,4,6‐trihydroxybenzene‐1,3,5‐tricarbaldehyde with hydrazide‐ or amine‐containing building blocks. Combined experimental and theoretical results show that the hydrazone‐linked N₁O₂‐tridentate environment enables stronger Co anchoring and a more favorable local Co‐site environment, which together promote charge transfer and lower the free‐energy barrier for key CO₂RR intermediates. Among the three Co‐loaded COFs, Co‐AH‐COF exhibits the best photocatalytic CO₂RR performance, affording a CO production rate of 34.2 mmol g⁻¹ h⁻¹. AH‐COF was further used to capture cobalt species from spent lithium‐ion battery leachate under mild conditions to prepare Co_R‐AH‐COF, which delivers a CO production rate of 33.6 mmol g⁻¹ h⁻¹ under identical conditions. These results demonstrate the feasibility of employing waste‐battery‐derived cobalt species as a precursor for efficient COF‐based photocatalysts and provide a route for integrating CO₂ conversion with waste‐metal reuse.

A covalent organic framework (AH‐COF) with novel N₁O₂‐tridentate structures stabilizes Co ions and tunes their electronic state, lowering the energy barrier for photocatalytic CO₂ reduction. Applied to recover cobalt from spent batteries, the resulting Co_R‐AH‐COF catalyst achieves a high CO yield of 33.6 mmol g⁻¹ h⁻¹, presenting a strategy for synergistic waste metal and CO₂ utilization.

1. Introduction

The escalating global greenhouse effect [1] and relentless consumption of fossil fuels have positioned CO₂ capture and conversion as a central research focus in the energy and environmental fields [2]. Photocatalytic CO₂ reduction reaction (CO₂RR), which utilizes solar energy to not only decrease atmospheric CO₂ levels but also convert it into value‐added chemicals and fuels such as CO [3, 4], CH₄ [5, 6], and C₂H₄ [7], represents a promising strategy for achieving carbon neutrality. Among these products, CO stands out as a fundamental feedstock in the chemical and metallurgical industries, where its strong reducibility and high reactivity make it indispensable for synthesizing key chemical products, serving as a primary reducing agent, and acting as a specialized fuel or energy carrier [8].

Covalent Organic Frameworks (COFs) are considered premier photocatalyst candidates due to their structural regularity, porous skeletons, robust stability, and synthetic tailorability, which collectively contribute to superior light absorption and the precise engineering of active sites [9, 10, 11, 12, 13, 14]. COFs not only facilitate the separation and migration of photogenerated electron‐hole pairs but also enable precise anchoring of metal species through their abundant functional sites, providing an ideal platform for efficient photocatalysis [15, 16, 17, 18, 19]. This is particularly advantageous for constructing single‐atom‐site catalysts (SASCs), as the well‐defined pores and coordination environments of COFs effectively prevent metal agglomeration, ensuring high dispersion and stability [20, 21]. The resulting isolated metal centers, characterized by uniform electronic structures and maximized atom utilization, offer precisely controlled active sites and often exhibit superior catalytic selectivity [22, 23]. Anchoring transition metal single atoms such as cobalt within the COF matrix, where they coordinate with framework heteroatoms such as nitrogen or oxygen, can alter the adsorption and conversion behavior of key CO₂ reduction intermediates, including *COOH and *CO, thereby influencing the reaction pathway and product selectivity toward CO formation.

Within SASCs, the catalytic performance is strongly influenced by the coordination microenvironment of the metal center, although the density of accessible metal sites can also contribute to the overall activity. As shown in Scheme 1, variations in coordination atom types (e.g., N [24, 25, 26, 27], O [28, 29], and S [30]), coordination modes (e.g., bidentate [31, 32], tridentate [33, 34], and tetradentate [23, 35, 36]), and framework substituents collectively regulate the electronic structure and local chemical environment of the metal center, thereby influencing CO₂ activation and product distribution. Cobalt‐based SASCs, with their suitable d‐orbital configuration and intrinsic activity for CO₂ adsorption, are excellent candidates for this reaction. For instance, a tridentate Co−N₁O₂ coordination structure has been shown to offer superior metal stabilization compared to bidentate motifs, while providing more accessible active sites than tetradentate configurations. This specific environment not only maintains atomic dispersion but also enhances the formation kinetics of the *COOH intermediate, lowering the energy barrier of the rate‐determining step and promoting highly selective CO generation [37, 38, 39]. Notably, the cobalt required for such catalysts could be sustainably sourced from the leachates of spent lithium‐ion batteries. This approach would mitigate the environmental hazards associated with battery waste, create a synergistic link between energy storage and conversion cycles, and enhance the overall economic viability. Therefore, the development of novel COF‐based SASCs with rationally designed coordination structures, potentially incorporating recycled metals, represents a forward‐looking direction for advancing efficient, selective, and sustainable photocatalytic CO₂ reduction [40].

SCHEME 1.

Schematic illustration of the previously reported COF structures, the synthesis of the COFs in this work, and the coordination‐based capture/incorporation of cobalt species from spent batteries into the COF frameworks.

Herein, a series of COFs featuring N₁O₂‐tridentate coordination sites were synthesized by condensing 2,4,6‐trihydroxybenzene‐1,3,5‐tricarbaldehyde with hydrazide‐ or amine‐containing building blocks, including the hydrazone‐linked AH‐COF and AH‐Ph‐COF as well as the imine‐derived IM‐BQ‐COF. Combined experimental and theoretical studies reveal that the N₁O₂‐tridentate coordination structure constructed via acylhydrazone linkages can more effectively adsorb and stabilize cobalt ions, while modulating the electronic structure of cobalt to lower the reaction energy barrier in the photocatalytic CO₂ reduction. Consequently, the Co‐loaded catalyst Co‐AH‐COF exhibits the highest photocatalytic performance for CO₂RR, with a CO production rate of up to 34.2 mmol g⁻¹ h⁻¹ surpassing that of Co‐AH‐Ph‐COF and far exceeding that of Co‐IM‐BQ‐COF. Furthermore, AH‐COF was employed to recover cobalt species from spent lithium‐ion battery leachate at room temperature. The resulting cobalt‐loaded material, Co_R‐AH‐COF, demonstrates catalytic activity comparable to that of the pristine Co‐AH‐COF prepared from conventional cobalt sources, achieving an excellent conversion rate of 33.6 mmol g⁻¹ h⁻¹.

2. Results and Discussion

2.1. Structural Characterizations

Scheme 1 and Figure S1 present the synthetic pathways for the target COFs, AH‐COF, AH‐Ph‐COF, and IM‐BQ‐COF, prepared by solvothermal Schiff base condensation between 2,4,6‐trihydroxybenzene‐1,3,5‐tricarbaldehyde (TP) and the respective linkers ethanedihydrazide, benzene‐1,4‐dicarbohydrazide, and 2,5‐diaminohydroquinone hydrochloride. Subsequently, Co species were introduced into the three frameworks at room temperature through a simple coordination‐based Co loading process, affording Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF. Inductively coupled plasma mass spectrometry (ICP‐MS) analysis determined their cobalt contents to be 4.70%, 3.86%, and 0.55%, corresponding to theoretical loadings of 16.6%, 17.4%, and 2.2%, respectively, demonstrating that the hydrazone‐linked COFs (AH‐COF and AH‐Ph‐COF) exhibit a much stronger cobalt‐coordination capability than the imine‐linked IM‐BQ‐COF.

The structures of the synthesized materials were first characterized by Fourier‐transform infrared (FT‐IR) spectroscopy. As shown in Figure S2, the AH‐COF spectrum exhibits the disappearance of the aldehyde (−CHO) characteristic peaks at 2894 and 1645 cm⁻¹ from TP [41], along with the emergence of characteristic signals for the β‐ketoenamine linkage, including the C═O bond at 1650 cm⁻¹, the C─N stretching vibration at 1208 cm⁻¹, and the C═C bond at 1630 cm⁻¹, confirming the occurrence of Schiff base condensation and enol‐to‐keto tautomerization [42, 43, 44]. After introducing Co species into AH‐COF, the intensities of the characteristic absorption bands assigned to the C═O and N─H groups markedly decreased, while the stretching vibration of the C─N bond shifts from 1208 to 1217 cm⁻¹, indicating successful incorporation of Co species into the AH‐COF framework via Co─O/Co─N coordination [45, 46, 47, 48]. Similar results are observed for AH‐Ph‐COF, showing nearly identical spectral behavior. In the case of IM‐BQ‐COF, the disappearance of the aldehyde characteristic peaks from TP and the emergence of β‐ketoenamine linkage signals are similarly observed. In contrast, after the adsorption of Co species, the C─O stretching band at 1262 cm⁻¹, assigned to the phenolic hydroxyl group, completely disappeared, while a new C═O stretching peak attributed to the β‐diketone structure emerged at 1640 cm⁻¹ [49, 50, 51, 52]. These spectral changes suggest that, during Co incorporation, the hydroquinone‐derived unit in IM‐BQ‐COF is likely converted into a quinone‐like structure, thereby generating an O/N‐containing tridentate coordination environment similar to those in AH‐COF and AH‐Ph‐COF. As shown in Figure S3, the results were further confirmed by ¹³C cross‐polarization magic angle spinning nuclear magnetic resonance (¹³C CP/MAS NMR) spectroscopy, with characteristic β‐ketone carbon signals observed at 195.2 ppm for AH‐COF, 192.6 ppm for AH‐Ph‐COF, and 189.3 ppm for IM‐BQ‐COF [53, 54, 55].

The ordered structures of these COFs were investigated by powder X‐ray diffraction (PXRD). Plausible crystal models, constructed using the Materials Studio software package, confirmed that all three COFs adopt an AA stacking mode. Pawley refinement of the experimental PXRD pattern for AH‐COF yielded unit cell parameters of a = b = 21.55 Å, c = 3.25 Å, α = β = 90°, γ = 120°, with agreement factors of R _p = 2.70% and R _wp = 3.56%. Accordingly, the experimental peaks observed at 4.7 and 27.5° for AH‐COF were assigned to the (100) and (001) facets, respectively, Figure 1b. Similarly, for AH‐Ph‐COF, Figure S4a, the peaks at 3.6, 6.2, 7.3, and 25.0° were indexed to the (100), (110), (200), and (001) facets, with refined unit cell parameters of a = b = 31.69 Å, c = 3.6 Å, α = β = 90°, γ = 120°, R _p = 4.41%, and R _wp = 3.00%. For IM‐BQ‐COF, Figure S4b, the peaks at 4.5, 7.1, 9.0, and 25.0° were assigned to the (100), (021), (022), and (001) facets, with unit cell parameters of a = b = 22.68 Å, c = 3.57 Å, α = β = 90°, γ = 120°, R _p = 3.96%, and R _wp = 3.03%. All three COFs are insoluble in common solvents, Figure S4d, including acetic acid, water, acetonitrile, DMF, and THF, and maintain their crystallinity after being soaked in these solvents for seven days, demonstrating their robust chemical stability. Moreover, after cobalt ion incorporation, the PXRD patterns of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF largely retain the diffraction features of the pristine frameworks, respectively, confirming that cobalt incorporation did not compromise the structural integrity of the frameworks.

FIGURE 1.

(a) Topology representation of the sql net of AH‐COF. (b) Experimental and refined PXRD patterns of AH‐COF. (c) Co 2p XPS spectrum of Co‐AH‐COF. (d) Co K‐edge XANES spectra and (e) FT EXAFS spectra of Co‐AH‐COF, Co foil, CoPc, and CoO. (f) WT EXAFS spectra of Co‐AH‐COF, Co‐AH‐Ph‐COF, Co‐IM‐BQ‐COF, and CoO. (g) TEM image and (h) EDS mapping of Co‐AH‐COF. (i) CO₂ sorption isotherms of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF measured at 298 K.

X‐ray photoelectron spectroscopy (XPS) was employed to investigate the chemical structures and elemental oxidation states of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF. As shown in Figure S5, the survey XPS spectra confirm the presence of C, N, O, and Co elements in all three COFs, indicating the successful incorporation of cobalt. In the high‐resolution Co 2p spectrum of Co‐AH‐COF, Figure 1c, the Co 2p^3/2 and Co 2p^1/2 binding energies observed at 781.1 and 796.5 eV, respectively, are assigned to Co²⁺, with no characteristic peaks of other cobalt oxidation states detected, confirming the divalent state of cobalt within the COF framework. Furthermore, compared to the metal‐free AH‐COF, Figure S6, slight shifts are observed in the high‐resolution N 1s and O 1s spectra, suggesting the formation of Co─N and Co─O coordination bonds. Similar spectral features are obtained for Co‐AH‐Ph‐COF and Co‐IM‐BQ‐COF, Figures S7 and S8. To further elucidate the local coordination environment of cobalt in the Co‐based COFs, X‐ray absorption spectroscopy (XAS) was performed on Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF. As shown in Figure 1d, the Co K‐edge X‐ray absorption near‐edge structure (XANES) spectra of these three COFs exhibit similar near‐edge shoulder positions, all closely aligned with that of CoO, indicating a Co valence state near +2, which is consistent with XPS results. The Fourier‐transformed (FT) extended X‐ray absorption fine structure (EXAFS) spectra of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF exhibit dominant peaks at 1.51, 1.45, and 1.47 Å, respectively, Figure 1e, which fall between the characteristic lengths of Co─N (1.45 Å) and Co─O (1.74 Å) bonds, and show no significant Co−Co scattering contributions in the 2.10∼2.62 Å range, thereby confirming predominant Co─O/N coordination environments, with cobalt stabilized as atomically dispersed isolated sites throughout the frameworks. Wavelet transform (WT) EXAFS analysis further confirms the absence of Co─Co bonds and the dominant presence of atomically dispersed cobalt in Co‐AH‐COF, Figure 1f, as evidenced by a distinct maximum at 1.41 Å in the R‐space contour plot that is characteristic of Co─O/N coordination, together with the lack of WT signatures corresponding to the Co─Co scattering observed at approximately 2.14 Å in Co foil and 2.58 Å in CoO. Similar Co─O and Co─N coordination features are observed in the WT‐EXAFS analyses of Co‐AH‐Ph‐COF and Co‐IM‐BQ‐COF, Figure S10. Quantitative EXAFS fitting was conducted to determine the precise coordination configuration of the Co atoms, as shown in Figures S11 and S12, revealing a Co–N₁O₂ coordination structure for Co(II) in Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF, which aligns well with previously reported single‐crystal structures of analogous molecular complexes [56, 57].

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) characterization results indicate that AH‐COF exhibits a loose porous network structure, Figures S13a and S14a, while AH‐Ph‐COF and IM‐BQ‐COF display typical rod‐like morphologies, Figure 1g and Figure S14b, with distinct lattice fringes clearly identifiable in their TEM images. After adsorbing Co²⁺, the morphological features of all three COF materials remain consistent with the original COFs, showing no significant structural distortion. Energy‐dispersive X‐ray (EDX) elemental mapping results confirm that C, O, N, and Co elements are uniformly distributed in the Co‐AH‐COF composite material, Figure 1h, further verifying the successful loading and homogeneous distribution of Co ions without noticeable aggregation. Thermogravimetric analysis (TGA) results show that the onset decomposition temperature is 270°C for Co‐AH‐COF and Co‐AH‐Ph‐COF, and 250°C for Co‐IM‐BQ‐COF, Figure S15, indicating good thermal stability for all three materials. N₂ adsorption‐desorption isotherms were measured at 77 K to evaluate the permanent porosity of the as‐prepared COFs. AH‐COF/Co‐AH‐COF and IM‐BQ‐COF/Co‐IM‐BQ‐COF show Type II(H3) isotherms, and AH‐Ph‐COF/Co‐AH‐Ph‐COF show Type IV(H1) isotherms. The adsorption isotherms were fitted based on the Brunauer–Emmett–Teller (BET) model, which revealed that after introducing Co(II), the specific surface area of the material decreased while the pore size remained essentially unchanged. Taking AH‐COF as an example, its specific surface area is 265.23 m² g⁻¹ while that of Co‐AH‐COF is 227.57 m² g⁻¹, with both materials showing a pore size of approximately 2.00 nm. The same trend can be observed for the AH‐Ph‐COF and IM‐BQ‐COF systems: the specific surface areas of AH‐Ph‐COF and Co‐AH‐Ph‐COF are 534.0 and 488.0 m² g⁻¹, respectively, while those of IM‐BQ‐COF and Co‐IM‐BQ‐COF are 191.74 and 112.90 m² g⁻¹, respectively, Figure S16. Furthermore, as shown in Figure 1i, the CO₂ sorption performance of the samples was measured at 1 bar to characterize their affinity toward CO₂. At 273 K and 298 K, the CO₂ uptake capacities of Co‐AH‐COF reaches 57.7 and 38.4 cm³ g⁻¹, respectively, markedly higher than the 30.5 and 25.1 cm³ g⁻¹ of Co‐AH‐Ph‐COF and the 30.1 and 24.3 cm³ g⁻¹ of Co‐IM‐BQ‐COF under the same conditions. These results indicate that the pore size and structural features of Co‐AH‐COF are more conducive to enhancing CO₂ adsorption and mass transfer, thereby offering potential advantages for subsequent CO₂ conversion.

2.2. Photoelectrochemical Behavior

UV–Vis diffuse reflectance spectroscopy (UV–Vis DRS) reveals that all three COFs exhibit promising light‐harvesting characteristics in the visible region, Figure 2a, indicating their potential for natural light‐driven catalytic reactions. Based on Tauc plot analysis, the band gaps of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF are estimated to be 2.04, 2.05, and 1.85 eV, respectively, Figure S18. The slightly narrower band gap of Co‐IM‐BQ‐COF highlights the significant influence of ligand structure on the optical properties of the materials. Mott–Schottky (M‐S) measurements show positive slopes for these COFs, identifying them as n‐type‐like semiconductors, Figure 2b. The flat‐band potentials were used to estimate the conduction band (CB) positions at −1.50, −1.62, and −0.90 V versus normal hydrogen electrode (NHE), and the valence band (VB) positions at +0.54, +0.43, and +0.95 V vs. NHE for Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF, respectively. The resulting band structures are depicted in Figure 2c. Notably, the CB edges of all COFs are positioned below the CO₂/CO redox potential of −0.53 V versus NHE, indicating that photocatalytic CO₂ reduction is thermodynamically feasible on these materials.

FIGURE 2.

(a) UV–Vis DRS of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF. (b) M‐S plot of Co‐AH‐COF. (c) Schematic illustration of the CB and VB positions for photocatalytic CO₂RR. (d) PL spectra of [Ru(bpy)₃]Cl₂ and [Ru(bpy)₃]Cl₂ in the presence of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF. (e) Transient photocurrent responses and (f) EIS plots of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF

The photoinduced charge carrier transfer behavior was investigated by conducting steady‐state photoluminescence (PL) and time‐resolved photoluminescence (TRPL) spectroscopy on the photosensitizer tris(2,2'‐bipyridine)ruthenium(II) chloride ([Ru(bpy)₃]Cl₂) before and after the introduction of the three cobalt‐loaded COFs. As shown in Figure 2d, the quenching of [Ru(bpy)₃]Cl₂ fluorescence in the presence of cobalt‐loaded COFs indicates efficient electron transfer from the photosensitizer. Notably, the quenching efficiency is markedly greater for Co‐AH‐COF and Co‐AH‐Ph‐COF compared to Co‐IM‐BQ‐COF. Furthermore, TRPL measurements show that the fluorescence lifetime of [Ru(bpy)₃]Cl₂ decreases from 212 to 183, 187, and 194 ns after the addition of Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF, respectively, Figure S20. These results indicate that Co‐AH‐COF is the most effective among the three samples in extracting photoinduced electrons from the photosensitizer. Subsequently, transient photocurrent responses under xenon lamp irradiation were measured to probe the photogenerated charge carrier transfer efficiency (Figure 2e). All three COFs exhibit prompt and stable photocurrent responses, with Co‐AH‐COF displaying the highest photocurrent density, indicative of the most efficient charge separation. The EIS data further show that both Co‐AH‐COF and Co‐AH‐Ph‐COF exhibit lower interfacial charge‐transfer resistance than Co‐IM‐BQ‐COF, indicating more favorable interfacial electron‐transport kinetics in the two hydrazone‐linked systems. It is worth noting that the marginally higher interfacial charge transfer resistance of Co‐AH‐COF compared to Co‐AH‐Ph‐COF arises from its poorer crystallinity and lack of the z‐axis π‐electron transfer pathway present in Co‐AH‐Ph‐COF via its benzene ring moiety.

The photocatalytic CO₂RR was performed under visible‐light irradiation at wavelengths of 420 nm or longer in a CO₂‐saturated MeCN/H₂O mixed solvent, using [Ru(bpy)₃]Cl₂ as the photosensitizer and triisopropanolamine (TIPA) as the sacrificial agent. Figure 3a shows that no CO was produced in control experiments lacking any essential component (light, catalyst, photosensitizer, or sacrificial agent), confirming that CO generation requires the full photocatalytic system. Systematic optimization of the reaction conditions revealed that, as shown in Figure 3b and Figure S21, the hourly CO production rates were essentially identical when using 1.0 and 2.0 mg of the Co‐AH‐COF catalyst under otherwise identical conditions. This suggests that a catalyst loading of 1.0 mg is sufficient under the present reaction conditions, and that further increasing the catalyst amount does not improve the overall reaction efficiency. Although the mass activity for CO₂RR reaches 90.2 mmol g⁻¹ h⁻¹ at a catalyst loading of 0.1 mg, potential systematic errors associated with such low quantities were considered. Therefore, a consistent catalyst loading of 1.0 mg was adopted for all subsequent experiments to ensure reliable comparison. As shown in Figure 3c and Figure S21, key reaction conditions, including the solvent ratio and sacrificial agent dosage, were systematically optimized. Under the optimized conditions in Figure 3d, Co‐AH‐COF demonstrates the highest activity, yielding 34.2 mmol g⁻¹ h⁻¹ of CO and 8.2 mmol g⁻¹ h⁻¹ of H₂. It substantially outperforms Co‐AH‐Ph‐COF with 25.4 and 11.4 mmol g⁻¹ h⁻¹ as well as Co‐IM‐BQ‐COF with 12.8 and 2.6 mmol g⁻¹ h⁻¹ for CO and H₂ evolution, respectively. Notably, although Co‐AH‐Ph‐COF has a Co loading comparable to that of Co‐AH‐COF, its lower CO₂RR activity indicates that the catalytic differences are not governed solely by the amount of Co species, but are also closely related to the intrinsic properties of the Co sites. These results indicate that the differences in CO₂RR activity and product distribution cannot be explained by Co loading alone, but are instead related to the combined effects of Co loading and the framework‐dependent local environment of the Co sites.

FIGURE 3.

(a) Control experiments for photocatalytic CO₂RR under different catalytic conditions. (b) Effect of catalyst dosage on CO₂RR performance of Co‐AH‐COF. (c) Effect of the solvent ratio on the CO₂RR performance of Co‐AH‐COF. (d) Time‐dependent CO evolution for COFs‐catalyzed CO₂RR under visible‐light irradiation. (e) Stability test of Co‐AH‐COF. (f) GC‐MS spectra for ¹³C isotope tracer tests. (g) Battery recycling flow chart. (h) Co K‐edge XANES spectra of Co‐AH‐COF and Co_R‐AH‐COF. (i) Time‐dependent CO evolution during Co_R‐AH‐COF‐catalyzed CO₂ reduction under visible‐light irradiation.

To further confirm the carbon source of CO, isotope labeling experiments were performed under a ¹³CO₂ atmosphere, and the gas‐phase products were analyzed by gas chromatography‐mass spectrometry (GC‐MS). Figure 3f shows a characteristic signal at m/z = 29 for ¹³CO, while no corresponding peak at m/z = 28 for ¹²CO is observed, confirming that the generated CO originates from the reduction of ¹³CO₂ rather than from other potential carbon sources in the system. Meanwhile, no liquid‐phase products were detected by ¹H NMR analysis (Figure S23a), further confirming that CO and H₂ constituted the only detectable products in this photocatalytic system. Furthermore, the apparent quantum yield (AQY) of Co‐AH‐COF was evaluated at different incident wavelengths, reaching 1.87% at 420 nm, Figure S23c. The catalyst stability was examined through four consecutive recycling experiments (Figure 3e), which showed that the CO production rate of Co‐AH‐COF remained essentially constant over multiple cycles, demonstrating good operational stability. Comparative analyses of the samples before and after the reaction by XPS, XRD, FT‐IR, and ICP‐MS revealed no significant changes in the chemical valence states, crystal structure, or cobalt content (Figures S4c, S24, and S25 and Table S1). This further confirms that Co‐AH‐COF possesses good structural integrity and elemental retention under photocatalytic CO₂ reduction conditions, highlighting its potential for recycling and practical application.

To explore cobalt resource recycling and the green preparation of COF‐based catalysts, AH‐COF was used to capture cobalt species derived from spent‐battery leachate, affording Co_R‐AH‐COF. ICP‐MS analysis confirmed a Co content of 5.2% in Co_R‐AH‐COF. The Co K‐edge XANES spectrum of Co_R‐AH‐COF closely resembles that of Co‐AH‐COF, indicating a similar local coordination environment around the Co centers. Under identical reaction conditions, Co_R‐AH‐COF exhibits photocatalytic CO₂ reduction performance essentially comparable to that of Co‐AH‐COF, delivering a CO generation rate of 33.6 mmol g⁻¹ h⁻¹ and an H₂ evolution rate of 9.6 mmol g⁻¹ h⁻¹. These results demonstrate the feasibility of reusing spent‐battery‐leachate‐derived cobalt species as a viable precursor for the construction of functional photocatalytic materials. It should be noted, however, that the present study does not quantify the concentrations of coexisting metal ions before and after treatment, nor does it systematically establish the selectivity of AH‐COF toward Co²⁺ in a multicomponent leachate. The detailed composition of residual impurity ions and their possible influence on catalytic behavior therefore remain to be clarified in future work.

2.3. Photocatalytic Reaction Mechanisms

To elucidate the reaction mechanism of Co‐AH‐COF during photocatalytic CO₂ reduction, in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was employed to dynamically monitor the evolution of reaction intermediates. As shown in Figure 4a, prior to illumination (0 min), no characteristic absorption signals associated with CO₂ reduction intermediates were detected. Under light irradiation, several distinct peaks emerged in the 1200–2200 cm⁻¹ region: peaks at 1414 and 1666 cm⁻¹ were assigned to symmetric and asymmetric stretching vibrations of bicarbonate (HCO₃ ⁻), while other features were attributed to dissolved CO₂, bidentate carbonate (b‐CO₃ ²⁻ at 1254 and 1584 cm⁻¹), and monodentate carbonate (m‐CO₃ ²⁻ at 1374 and 1506 cm⁻¹). Notably, the intensities of the peaks at 1288, 1337, and 1476 cm⁻¹ increased progressively with irradiation time, indicating the formation of the *COOH intermediate [58]. Concurrently, the absorption peak at 2122 cm⁻¹ [59], assigned to the *CO intermediate, also intensified with prolonged illumination [38, 60]. These spectroscopic observations collectively support the conclusion that the CO₂ photoreduction to CO catalyzed by Co‐AH‐COF follows the reaction pathway:

$${\mathrm{CO}}_2\to *{\mathrm{CO}}_2\to *\mathrm{COOH}\to *{\mathrm{COOH}}_2\to *\mathrm{CO}\to \mathrm{CO}.$$

FIGURE 4.

(a) In situ DRIFTS spectra of Co‐AH‐COF collected at different illumination times. (b) In situ XANES spectra at the Co K‐edge and (c) corresponding first‐derivative spectra of Co‐AH‐COF during photocatalytic CO₂ reduction under light and dark conditions. (d) Gibbs free energy changes and corresponding coordination structures for AH‐COF, AH‐Ph‐COF, and IM‐BQ‐COF before and after Co²⁺ adsorption. (e) Gibbs free energy changes for the photocatalytic CO₂RR on COFs, in which * means the COF skeleton. (f) Schematic illustration of the proposed reaction mechanism for photocatalytic CO₂ reduction to CO over Co‐AH‐COF.

In situ XAFS measurements were further performed to dynamically monitor the oxidation state of Co sites in Co‐AH‐COF and to gain insight into their role during the CO₂RR process. The test was conducted with the sample immersed in a CO₂‐saturated acetonitrile/aqueous solution containing a photosensitizer [Ru(bpy)₃]Cl₂ and a sacrificial agent TIPA. Upon illumination, Figure 4b, the XANES spectra exhibit a gradual shift of the Co absorption edge toward higher energies with increasing irradiation time. This spectral evolution indicates a light‐induced change in the electronic state of the Co centers during photocatalysis, which is fully consistent with the transfer of photogenerated electrons to the Co active sites in the photocatalytic system. The observed energy shift of the absorption edge can be unambiguously attributed to the effective capture of photogenerated electrons by the active Co centers: as Co accepts photogenerated electrons, the electron density around the Co species increases, thereby raising the energy required for X‐ray absorption. This process reflects the strong electron‐trapping capability of the Co sites, providing key spectroscopic evidence for their role as active centers in the photocatalytic reduction reaction. To more precisely capture the shift of the absorption edge, Figure 4c, the near‐edge absorption profiles were converted into first‐derivative spectra. The derivative peak exhibits a distinct negative shift upon illumination, and reverts to its initial position when the system is returned to darkness. This reversible behavior confirms that the light‐induced reduction of Co is a dynamically driven process, rather than an irreversible chemical transformation of the Co species, thereby strongly supporting the cyclic “photo‐reduction/dark‐recovery” characteristic of the Co active centers.

Subsequently, density functional theory (DFT) calculations were performed to elucidate the differences in photocatalytic CO₂RR activity among Co‐AH‐COF, Co‐AH‐Ph‐COF, and Co‐IM‐BQ‐COF, and to uncover their underlying origins at the atomic level. Since CO₂ adsorption and activation predominantly occur near the metal coordination centers, a representative cluster model containing a single Co‐N₁O₂ coordination motif was employed to enhance computational efficiency and highlight local reaction characteristics. To evaluate the thermodynamic feasibility of anchoring Co precursors onto the COF frameworks, a ligand‐exchange model in an aqueous environment was constructed. Using [Co(OAc)(H₂O)₄]⁺ as the solvated Co(II) precursor, its interaction with the COF coordination site was simulated to form the anchored product COF−Co(OAc). To account for the stabilization effect of site deprotonation in water, [(H₂O)₃·H₃O]⁺ was adopted as a representative hydrated proton species. The corresponding reaction is simplified as: COF−H + [Co(OAc)(H₂O)₄]⁺ → COF−Co(OAc) + [(H₂O)₃·H₃O]⁺. As shown in Figure 4d, the calculated Gibbs free energy change, obtained at the same level of theory and with consistent solvation models, confirms that this anchoring process is thermodynamically favorable, supporting the spontaneous immobilization of Co species at the N₁O₂‐tridentate sites of the COFs under aqueous conditions. Further comparison shows the Gibbs free energy change for Co‐AH‐COF is more negative (ΔG = −0.04505 Hartree), comparable to that of Co‐AH‐Ph‑COF (ΔG = −0.04580 Hartree) and significantly lower than that of Co‐IM‐BQ‐COF (ΔG = −0.03815 Hartree). This trend aligns with the experimentally measured Co loading amounts.

Based on the stable Co coordination center configurations, the free energy changes along the reaction pathway for CO₂ reduction to CO were further calculated in Figure 4e. The reaction initiates with CO₂ adsorption at the Co active site to form the *CO₂ adsorbed species, followed by two consecutive PCET steps that sequentially generate the *COOH intermediate and subsequently convert it to *CO, accompanied by H₂O formation. Finally, CO desorbs into the gas phase. The computational results indicate that the free energy barrier for COOH formation (ΔG_{* COOH}) on Co‐AH‐COF is lower than those on Co‐AH‐Ph‑COF and Co‐IM‐BQ‐COF. This suggests a greater thermodynamic advantage for Co‐AH‐COF in the *COOH generation step, thereby more effectively promoting CO₂RR toward CO production. The trend is consistent with the regulatory effect of the Co−N₁O₂ local coordination environment on the metal center's electronic structure, indicating that stronger local coordination and charge redistribution help stabilize the *COOH adsorption state and reduce the thermodynamic barrier of the rate‐determining step.

Based on the combined experimental characterization and DFT results, a plausible reaction mechanism for photocatalytic CO₂ reduction to CO over Co‐AH‐COF is proposed, as shown in Figure 4f. Under visible light irradiation, the photosensitizer [Ru(bpy)₃]²⁺ (PS) is excited to PS*, followed by interfacial electron transfer from PS* to the Co–COFs. This injects and accumulates electrons at the Co−N₁O₂ coordination center, rendering the Co site electron‐enriched and more effective for CO₂ activation. Meanwhile, the oxidized photosensitizer PS⁺, generated after electron transfer, is rapidly reduced back to PS by the sacrificial electron donor TIPA, thereby suppressing charge recombination and sustaining continuous electron supply. CO₂ adsorbed near the Co site accepts an electron to form the *CO₂ (or CO₂ ^•−) adsorbed state. Subsequently, with protons supplied by the aqueous medium (H₂O), it undergoes successive PCET steps to form the key *COOH intermediate, which is further reduced to *CO and H₂O. Finally, CO desorbs from the catalytic site, completing the catalytic cycle for CO₂‐to‐CO conversion.

3. Conclusion

In summary, a series of N₁O₂‐tridentate COFs were constructed to host Co(II) single‐atom sites for photocatalytic CO₂ reduction. Experimental and DFT studies show that the framework linkage strongly influences Co‐anchoring, the local/electronic environment of the Co sites, and the resulting catalytic behavior. Among the three catalysts, the hydrazone‐linked Co‐AH‐COF exhibits the most favorable overall combination of Co‐site density, local coordination environment, charge‐transfer behavior, and CO₂ affinity, leading to the best photocatalytic CO₂RR performance. In addition, AH‐COF can capture cobalt species from spent‐battery leachate to produce Co_R‐AH‐COF, which shows CO₂RR performance comparable to that of Co‐AH‐COF under identical conditions. These results demonstrate the feasibility of employing waste‐battery‐derived cobalt species as a reusable precursor for COF‐based catalyst preparation and provide a sustainable route for transforming waste metal resources into functional catalytic materials.

Funding

This work was supported by the National Natural Science Foundation of China (Nos. 22501020 and 22235001), Fundamental Research Funds for the Central Universities (No. FRF‐TP‐24‐014A), Guizhou Provincial Key Laboratory Platform Project (ZSYS [2025] 008), and Talent Program of Guizhou University (No. [2024] 11).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73910‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.