Visible-light-photocatalytic reduction of CO₂ to valuable chemicals mimicking natural photosynthesis process is an effective way to mitigate the growing global energy crisis [1-6]. Metal-organic frameworks (MOFs), a type of porous crystalline material assembled by organic linkers and metal/metal cluster nodes through coordination bonds, have captured widespread interest in photocatalysis fields [7-9]. The metal nodes can act as catalytic active sites, and the porous structure shows high affinity toward the gaseous substrate. Therefore, MOFs present as ideal candidates to accumulate and then reduce CO₂ to CO and CH₄ by using the photogenerated electrons deriving from light excitation. However, the relatively low catalytic activity of MOFs has been a stumbling block for their potential applications. Therefore, improving the photocatalytic efficiency for CO₂ reduction is a long-lasting challenge for researchers in this field.

Post-modification syntheses have been proven as an effective strategy to tune MOF structures for improved catalytic performance [10-17]. Recently, a study reported that the divalent cobalt in an alkaline-stable MOF can be oxidized to trivalent cobalt through introducing OH⁻ into backbones via post-modification [18]. It has been established that the transition metal Co(Ⅲ) is very active for many catalytic reactions [19]. Additionally, the hydroxyl group has a high affinity toward CO₂ [20]. Inspired by these findings, we speculated that integrating OH⁻ and highly-active Co(Ⅲ) into a MOF structure may enhance the CO₂ photoreduction activity.

Herein, we report the successful post-modification of a cobalt MOF from Cl⁻ bridge to OH⁻ bridge by simple ion exchange. The influence of Cl⁻/OH⁻ on the CO₂ photocatalytic activity and selectivity was systemically investigated, and the underlying mechanism was revealed by X-ray photoelectron spectroscopy (XPS), cyclic voltammetry (CV), and other characterizations. These experimental results demonstrate that the introduction of hydroxyl groups not only greatly increases the CO₂ affinity but also alters the oxidation state of metal clusters, resulting in significantly optimized catalytic kinetics and improved photocatalytic activities. The efforts and discoveries described here are expected to provide some insight into the design of effective photocatalysts.The MOF [Co₂(μ-Cl)₂(bbta)] (H₂bbta = 1H, 5H-benzo-(1,2-d: 4,5-d′)bistriazole, denoted as Co-Cl) was obtained according to the previous report [18]. In brief, H₂bbta and CoCl₂·6H₂O were ultrasonically dissolved in a mixture of N, N-dimethylformamide (DMF) and methanol (MeOH), which was heated at 70 ℃ for 72 h to yield pink needle microcrystals of Co-Cl. And Co-Cl possesses a three-dimensional (3D) coordination framework with a honeycomb-like structure. Each Co ion is coordinated by three nitrogen atoms from three bbta²⁻ ligands, two μ-Cl anions, and each bbta²⁻ ligand coordinates to six Co ions. The Co ions are bridged by the μ₃-triazolate ring and μ-Cl to form a helical chain, which are connected by the phenyl backbone of bbta²⁻ to generate a pillared-rod structure with large 1D channels parallel to the helical chains. The replacement of bridging Cl⁻ with OH⁻ groups can be achieved by immersing Co-Cl in 1.0 mol/L KOH solution for 14 h at room temperature (denoted Co-OH). Powder X-ray diffraction (PXRD) pattern revealed that Co-OH maintained the pure and highly crystalline phase after the treatment (Fig. 1a). The stretching vibrations of OH⁻ at 3630 cm⁻¹ and Co-O at 561 cm⁻¹ were observed in infrared spectroscopy (IR) spectra (Fig. 1b), suggesting the successful introduction of hydroxo-bridge cobalt clusters in Co-OH. Moreover, X-ray photoelectron spectroscopy (XPS) analyses revealed that the Cl characteristic peak disappeared in the spectrum for Co-OH (Fig. S1 in Supporting information), indicating Cl⁻ was completely substituted by OH⁻. Meanwhile, the morphologies of the as-prepared Co-OH retained after soaking in KOH as revealed by scanning electron microscopy (SEM), indicating its desired stability for the subsequent catalytic reaction (Figs. S2a and b in Supporting information). Moreover, the corresponding elemental mappings of O and Cl illustrated the absence of Cl in Co-Cl, which matches well with the results obtained from XPS studies. The contact-angle tests showed that Co-OH exhibited a smaller contact angle (32.1°) than Co-Cl (36.3°) did (Figs. S2c and d in Supporting information), which indicates that the hydrophilicity of material does not greatly change with replacing Cl⁻ by OH⁻. The high hydrophilicity nature would be conducive to the water activation in the CO₂ reduction process [3]. The broad and strong light absorption at 400–700 nm for both Co-OH and Co-Cl ensure the full use of visible-light irradiation for photocatalysis (Fig. 2a). The bandgaps of Co-OH and Co-Cl were estimated to be 2.50 eV and 2.80 eV, respectively, from the Kubelka-Munk plots (Fig. 2b), indicating that OH⁻ modification barely impact on the energy gaps. Mott-Schottky measurements at 500, 1000, and 1500 Hz were used to investigate the semiconductor character and energy alignment of Co-OH and Co-Cl (Figs. 2c and d). The positive slope of the obtained C⁻² value indicates the n-type semiconductor feature of these materials.

The conduction band minimum (CBM) was evaluated to be −1.83 V for Co-Cl, −1.14 V for Co-OH vs. Ag/AgCl electrodes (−1.63 V and −0.94 V vs. NHE, pH 7), which is more negative than the redox potentials of CO₂/CO (−0.51 V vs. NHE, pH 7) and CO₂/CH₄ (−0.24 V vs. NHE, pH 7). Based on the bandgap information obtained from the UV–vis spectrum, the valence band maximum (VBM) was estimated to be around 1.56 V for Co-OH and 1.17 V for Co-Cl (vs. NHE, pH 7). The replacement of Cl⁻ with OH⁻ resulted in a considerable change on the CBM position of the MOF, which might be attributed to the change on the valence state of Co centers. Although Co-OH has low N₂ uptakes than Co-Cl (Fig. 2e), there exists a noticeable hysteresis loop between adsorption and desorption branches, indicating a stronger N₂ affinity of Co-OH frameworks. The stronger N₂ affinities of Co-OH may be ascribed to the defects, which resulted in a slightly larger pore size distribution for Co-OH than for Co-Cl (Fig. S4 in Supporting information). CO₂ adsorption uptake of Co-OH is unexpectedly higher than that of Co-Cl, in contrast to nitrogen adsorption. According to Fig. 2f, Co-OH has considerably improved CO₂ binding affinities, as evidenced by the steep slopes at low pressures, the higher uptakes throughout the entire pressure range, and the hysteresis loops in isotherm. We can expect that this high CO₂ affinity would considerably improve the CO₂ photoreduction efficiency.

All the photocatalytic reactions of CO₂ and H₂O were conducted under visible-light illumination (λ > 400 nm) without the presence of hole scavengers. As indicated in Figs. 3a and b, the CH₄ generation rate was 0.53 µmol g⁻¹ h⁻¹ and the CO generation rate was 2.0 µmol g⁻¹ h⁻¹ for Co-Cl. In contrast, Co-OH showed a much increased CH₄ evolution rate as high as 9.8 µmol g⁻¹ h⁻¹, although the CO generation rate is almost the same. To verify the origin of CH₄, N₂ instead of CO₂ was introduced into the reaction system as a reactant under similar conditions. In this case, only a trace amount (13 µmol/g) of CH₄ was obtained after 5 h. The maintained high crystallinity and the negligible decrease in the CH₄ and CO generation rate in four successive runs confirm the durability of Co-OH (Fig. 3d and Fig. S5 in Supporting information). H₂O₂ was confirmed as the oxidation product by the iodometric method [21-23]. The iodide ion (I⁻) reacts with H₂O₂ and form triiodide ion (I₃⁻) strongly absorbing at 352 nm (Fig. 3c) (more information can be seen in Supporting information). This unexpected increase in activity and selective illustrates the capability of this strategy for photocatalyst design, inspiring us to further look into the mechanism.

To better understand the charge transfer behavior of these samples, electrochemical impedance spectroscopy (EIS) studies were performed. Compared with Co-Cl, the semicircle radius of Co-OH curve dropped dramatically, showing a lower resistance (Fig. 4a). As a consequence, Co-OH has an improved conductivity and charge-carrier transfer efficiency, which is consistent with the photocatalytic activity trends. Meanwhile, Co-OH showed a lower fluorescence intensity than Co-Cl did, implying a greatly suppressed radiative electron-hole recombination in Co-OH (Fig. 4b). XPS spectra were acquired in order to comprehensively investigate the cause of the improved photocatalytic activity. The O 1s XPS spectra of Co-Cl shown in Fig. S6 (Supporting information) could be divided into two peaks centered at 531.7 and 533.3 eV, which may be attributed to adsorbed oxygen and surface hydroxyl groups, respectively [24, 25].

While the peaks of Co-OH at 531.0 and 533.1 eV can be attributed to the lattice oxygen (Co-O) and surface hydroxyl groups, respectively, confirming the successful substitution of Cl⁻ with OH⁻ [24]. On the other hand, the Co 2P peaks could be fitted into two peaks at 782.8 and 780.8 eV in Co-Cl (781.5 and 780.5 eV in Co-OH), which can be assigned to Co(Ⅲ) and Co(Ⅱ) (Fig. 4c). This is consistent with the prediction that the Co(Ⅱ) inside Co-OH bound to a terminal H₂O/OH⁻ ligand and was further oxidized to Co(Ⅲ)-OH species [18]. However, the Co(Ⅲ)/Co(Ⅱ) ratio in Co-OH is three times higher than that in Co-Cl, implying that the majority of Co(Ⅱ) of Co-OH was converted to Co(Ⅲ) (Table S1 in Supporting information). In addition, the Co 2p_1/2–2p_3/2 spin−orbit level energy spacing is 16 eV for Co-Cl and 15 eV for Co-OH, this is also an indicative of +3 oxidation state for the majority of the Co species in Co-OH [26, 27]. Other evidence came from the lower binding energy of Co(Ⅲ) (blue part) in Co-OH than that in Co-Cl, indicating that Co-OH has a higher electron cloud density. Based on the XPS data, we can conclude that there exists a large amount of Co(Ⅲ) spices in Co-OH after post-modification, which is assumed to an important reason for the increased activity [19]. Cyclic voltammetry (CV) can provide more information about the redox behavior of material and the possible catalytic process during reaction. As revealed in Fig. 4d, both Co-Cl and Co-OH show irreversible Co^Ⅱ/Co^I peaks at E = −0.51 V (vs. Ag/AgCl), while Co-OH shows an extra reversible reductive process peak at E_1/2 = 0.64 V which can be assigned to the redox of Co^Ⅱ/Co^Ⅲ [27-30]. Moreover, the peak current for Co^Ⅱ/Co^I peak in Co-OH is lower than that of Co-Cl, implying a decrease of Co(Ⅱ) component in Co-OH. The newly appeared Co^Ⅱ/Co^Ⅲ redox peak and the decreased Co(Ⅱ) content further confirm the transform from Co(Ⅱ) to Co(Ⅲ).

In this work, the bridging Cl⁻ ions in the cobalt clusters of a Co-MOF were successfully replaced by OH⁻ ions via post-modification, resulting in a highly efficient and reusable visible-light photocatalyst for CO₂ reduction. The mechanistic analyses demonstrated that, compared with Cl⁻, the introduction of OH⁻ causes a higher CO₂ affinity and partial oxidation of the cobalt nodes from Co(Ⅱ) to Co(Ⅲ), significantly boosting the with photocatalytic activity of materials. This work might shed light on the fabrication of highly active and stable MOF-based photocatalysts.

The authors declare no competing financial interest.

This work was financially supported by the National Key Research and Development Program of China (No. 2018YFA0208600), the National Natural Science Foundation of China (No. 22033008), Fujian Science & Technology Innovation Laboratory for Optoelectronic Information of China (No. 2021ZZ103).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.03.034.