An exceptionally stable octacobalt-cluster-based metal–organic framework for enhanced water oxidation catalysis

The site, capping four coplanar cobalt ions, has a near-optimal OH– adsorption energy, which is beneficial to accelerate the reaction kinetics of water oxidation catalysis.


Introduction
The replacement of fossil fuels with hydrogen generated by water splitting is a very attractive solution to the present energy problem. In order to develop a practical technology based on these elements, durable and efficient water oxidation catalysts need to be developed. 1,2 However, the water oxidation process involves a four-electron process, leading to slow kinetics. 3,4 For this reason, improving the efficiency of water oxidation catalysts is still a challenging task, despite the considerable achievements that have been made in recent years. [5][6][7] The active oxygen species (namely, the Oc radical) has been accepted as the important intermediate for the formation of hydroperoxy (OOH) species and for the subsequent conversion to O 2 molecules. 8 Under alkaline conditions, the M-Oc (M ¼ metal) species is generated from the oxidation of the M-OH species, enhancing the adsorption energy of the reactant OH À , which might be benecial to the formation of *OH for the oxygen evolution reaction (OER) intermediates. However, if OH À binds too strongly, it will occupy available surface sites and poison the reaction. 9 Therefore, optimizing the OH À adsorption energy to a near-optimal value might be benecial to accelerate the reaction kinetics. 10 Obviously, regulating the coordination number of the oxygen atom of OH À could be the most effective strategy to optimize the adsorption energy (Scheme 1). For instance, Jin et al. incorporated gold clusters onto the CoSe 2 catalyst to increase the coordination number of the OH À from 1 to 2, resulting in the enhancement of activity. 11 Nevertheless, the binding affinity was still too weak, leading to a large amount of energy input required to produce *OH.
Metal-organic frameworks (MOFs), as crystalline porous materials with high surface areas and outstanding structural designability, can combine the advantages of both homogeneous and heterogeneous catalysts. Recently, MOFs have emerged as potential catalysts for water oxidation, 12,13 the hydrogen evolution reaction, 14-17 carbon dioxide reduction, [18][19][20][21][22] etc. 14,23-25 Nevertheless, as similar to traditional catalysts, the reported MOF catalysts also suffer from poor stability and low catalytic activity. 26,27 Among the various types of MOFs, metalazolate frameworks (MAFs) are famous for their extraordinary chemical stabilities. 28 In addition, the high connectivity of the metal cluster could enhance the stability of the framework. 29,30 Considering the relatively high activities and earth-abundance of cobalt ions, and that the multinuclear metal cluster might have a favourable OH À adsorption energy, a combination of the highly connected cobalt-hydroxide unit and the azolate bridging ligand is the best choice. Here, we report a highly stable, octacobalt cluster based MAF with both extraordinarily high activity and durability. We demonstrated that the metal site, capping four coplanar cobalt ions, indeed serves as a highly efficient active site for water oxidation.

Materials and methods
All reagents were commercially available and used without further purication. 1,4-Benzenedi(1H-1,2,3-triazole (H 2 bdt) was synthesized according to the method in the literature. Elemental analyses (EA) were conducted using an Elementar Vario EL analyzer. X-ray photoelectron spectroscopy (XPS) measurements were performed with a VG Scientic ESCALAB 250 instrument. Powder X-ray diffraction (PXRD) patterns were collected on a Bruker D8-Advance diffractometer with Cu Ka radiation and a LynxEye detector. Variable-temperature PXRD data were collected on a Rigaku SmartLab X-ray diffractometer (Cu-Ka, l ¼ 1.54056Å). Thermogravimetric (TG) analyses were performed on a TA Q50 thermogravimetric analyzer under nitrogen gas at a heating rate of 10 C min À1 . Scanning electron microscope (SEM) images were obtained from an ultra-highresolution electron microscope (FE-SEM, SU8010). Gas sorption isotherms were measured on a Micromeritics ASAP 2020M instrument. Before the sorption experiments, the assynthesized samples were rst solvent exchanged with MeOH, and then activated for 12 h at 150 C under vacuum. N 2 (99.999%) was used for all measurements. The temperature was controlled by a liquid nitrogen bath (77 K).

Crystal structure determination
Diffraction data of Co 4 -bdt were collected on a Rigaku XtaLAB P300DS-detector diffractometer (Cu Ka). All structures were solved by direct methods and rened with the full-matrix leastsquares technique on F 2 by the SHELXTL-2014 soware package. All non-hydrogen atoms were rened anisotropically. Hydrogen atoms were placed geometrically. The PLATON SQUEEZE treatment was applied, because all guest solvent molecules were extremely disordered and could not be modeled. Detailed structure determination parameters and crystallographic data are given in Table S1. †
The purity of Co 4 -bdt was preliminarily demonstrated by the scanning electron microscopy (SEM) images and transmission electron microscopy (TEM) images (Fig. S2 †). Thermogravimetric (TG) and powder X-ray diffraction (PXRD) measurements of methanol-exchanged Co 4 -bdt indicated that its guest molecules can be removed above 100 C, and the host framework can be stable up to 300 C ( Fig. S3 and S4 †). Moreover, Co 4 -bdt can remain intact in 6 M KOH (Fig. S5 †) for at least 3 days, representing the most alkali stable MOF. 32,33 A N 2 sorption isotherm was measured for Co 4 -bdt at 77 K ( Fig. S6 †), which shows typical type-I characteristics, with a saturation uptake of 594 cm 3 (STP) g À1 , corresponding to a pore volume of 0.928 cm 3 g À1 (crystallographic value of 0.924 cm 3 g À1 ). Furthermore, the Brunauer-Emmett-Teller (BET) and Langmuir surface areas of Co 4 -bdt were calculated to be 2266 and 2570 m 2 g À1 , respectively.
The photodriven water oxidation (PWO) experiment of Co 4bdt was performed under visible light in water with [Ru(bpy) 3 ] SO 4 (bpy ¼ 2,2 0 -bipyridine) as the photosensitizer and Na 2 S 2 O 8 as the sacricial electron acceptor, and these are the typical reaction conditions used in the literature. A Clark-type oxygen electrode was used to monitor in situ the amount of evolved O 2 dissolved in the solution (Fig. 2a and S7 †). O 2 rapidly formed at an initial turnover frequency (TOF) of 3.05 AE 0.03 s À1 . Due to the consumption of the sacricial electron acceptor, the production rates slowly decreased (Fig. S8 †). Except for PSII, the performance of Co 4 -bdt is higher than that for all other known heterogeneous catalysts, 26,27,[34][35][36][37][38][39] and is comparable to the best homogeneous catalysts under the same conditions [40][41][42][43] (Tables S2 and S3 †). Since the process that limits catalytic turnover is the oxidative quenching of the Ru excited state, 44-46 the photocatalytic water oxidation experiments using [Ru(bpy) 3 ] 3+ as the chemical oxidant without Na 2 S 2 O 8 were carried out. It can be seen that the TOF values increased and that the catalysts work under pseudo rst order conditions (Fig. S9 †). Importantly, the actual TOF value of Co 4 -bdt is as high as 15.7 s À1 (Fig. S10 †), which is higher than that of the best PWO catalyst (13 s À1 ). [47][48][49][50] Thanks to the excellent stability, the TOF of Co 4 -bdt remained 3.05 s À1 (Fig. 2c) aer 12 000 runs, indicating that the turnover number (TON) is larger than one million. Notably, the TON value of Co 4 -bdt is two orders higher than that of the best catalysts. 51 To directly determine the TON of the Co 4 -bdt, timedependent oxygen evolution experiments were carried out with chemical oxidant [Ru(bpy) 3 ] 3+ . As shown in Fig. S11, † Co 4 -bdt gave a TON of 1.2 Â 10 6 , which is consistent with that obtained from the recycling experiment (Fig. 2c). Aer the reactions, the Co 4 -bdt catalyst was recovered from the reaction mixture and was found to be almost unchanged according to the N 2 sorption isotherm measurements (Fig. S6 †), the PXRD patterns and the SEM images (Fig. S12 †). Furthermore, inductively coupled plasma-mass spectrometry showed that just 0.14% of the Co ions in Co 4 -bdt were leached into the reaction solution aer the reaction, and the ltrate showed negligible catalytic activity (Fig. S7 †), which conrmed the heterogeneous nature and the stability of the catalyst. The high stability of Co 4 -bdt might be ascribed to the high connectivity of the second building units (SBUs) 29 and the more stable cobalt-N (nitrogen atom) coordination bonds. 28,32 In order to study the mechanism, we selected ve other cobalt-based MOFs with variable metal-hydroxide units and an isostructural MOF, [Ni 8 (OH) 6 (bdt) 4 54 The square-pyramidal Co ion can coordinate to a terminal water molecule or to a hydroxyl anion to form a distorted octahedral mode; (v) [Co(mim) 2 ] (ZIF-67/Co-mim, Hmim ¼ 2-methylimidazole) is constructed of tetrahedral Co ions and mim À ligands. 55 In addition, the tetrahedral Co ion can coordinate to a terminal water molecule or to a hydroxyl anion to form a distorted trigonal-bipyramidal mode.
To demonstrate that the Co-manifold works with a four electron/four proton mechanism, the activity of Co 4 -bdt for water oxidation was studied by electrochemical characterization. Linear sweep voltammetry (LSV) was performed in water at pH ¼ 7 (Fig. 3a, S17 and S18 †). Assuming that the water oxidation reaction involves a four-electron process, the Faraday efficiency of Co 4 -bdt for water oxidation was measured to be virtually 100% (Fig. S19 and Table S6 †). Importantly, Co 4 -bdt afforded a current density of 2.0 mA cm À2 at an overpotential of 352 mV, which is much lower than that for all other reported catalysts (Table S7 †). The performance of Co 4 -bdt showed negligible changes aer OER tests at 10 mA cm À2 for 24 h (Fig. 3b). Furthermore, PXRD patterns (Fig. S12 †) and cyclic voltammetry curves (Fig. 3c) of Co 4 -bdt showed negligible changes aer the electrochemical OER tests for 24 h. The electrocatalytic activity follows the order: Co 4 -bdt (352 mV) > Co 4 -cpt (355 mV) > Co 3 -in (385 mV) > Co 2 -bbta (489 mV) > Co-dobdc (544 mV) > Co-mim (638 mV) ( Fig. S17a and Table S7 †) and this is consistent with results observed from the photodriven water oxidation experiment. This phenomenon demonstrates the high catalytic activity of Co 4 -bdt in the water oxidation reaction.
Isotope tracing experiments were carried out to investigate the role of m 4 -OH À during the water oxidation reaction. The extent of 18  indicates that the bridging OH À ligand does indeed participate in the reaction to offer an oxygen vacancy, which serves as the active site for water oxidation. This phenomenon was also observed for Co 3 -in and Co 2 -bbta (Fig. S21 †). In other words, the site, capping four coplanar cobalt ions, indeed serves as the Fig. 3 (a) LSV curves of Co 4 -bdt, Co 4 -cpt, Co 3 -in, Co 2 -bbta, Co-mim and Co-dobdc at pH ¼ 7. (b) The chronopotentiometry curves of Co 4 -bdt at an overpotential corresponding to the current density of 10 mA cm À2 at pH ¼ 7. (c) LSV curves of Co 4 -bdt before (black) and after (red) the electrochemical OER test at 10 mA cm À2 for 24 h at pH ¼ 7. highly efficient active site for water oxidation. Such an active site for water oxidation is the rst to be reported to date. It should be noted that not all oxygen atoms from the Co 4 cluster participate in the reaction at the same time, and the O coordinated on the Co 4 cluster will be replaced by the water molecule before the O 2 formation. Therefore, although the O from the Co 4 cluster indeed engages in O 2 formation, the structure of Co 4 -bdt remains during and aer the photodriven water oxidation reaction (Fig. 4a, S12 and S24 †).
To further understand the relationship between the coordination number of the oxygen atom in OH À and the properties, we analyzed the adsorption energy of OH À adsorbed on the active site by periodic density functional theory (PDFT). The adsorption energy (DE) of OH À follows the order, Co-mim (2.13 eV) > Co-dobdc (2.03 eV) > Co 2 -bbta (1.75 eV) > Co 3 -in (1.67 eV) > Co 4 -cpt (1.58 eV) z Co 4 -bdt (1.56 eV) > Ni 4 -bdt (1.19 eV) (Fig. 4b), and this implies that the OH À becomes more stable with an increase in the coordination number. The poor performance of Ni 4 -bdt is due to its too strongly OH À binding. By combining the PDFT simulation results and the isotope tracing experiments, it can be seen that the high catalytic performance of Co 4 -bdt might be ascribed to the fact that the OH À is appropriately stabilized by four coplanar cobalt ions.

Conclusions
In conclusion, through a combination of the highly connected cobalt-hydroxide unit and the azolate bridging ligand, we designed and synthesized cobalt azolate frameworks with high stability and excellent water oxidation performance. Since the oxygen atom is simultaneously coordinated by four coplanar cobalt ions, the reacting hydroxyl radical is appropriately stabilized during the water oxidation reaction, promoting the catalytic performance. These results should be insightful for understanding the structure-function relationship of water oxidation catalysts and for developing new MOF-based catalysts.

Conflicts of interest
There are no conicts to declare.