A conductive catecholate-based framework coordinated with unsaturated bismuth boosts CO2 electroreduction to formate

Bismuth-based metal–organic frameworks (Bi-MOFs) have received attention in electrochemical CO2-to-formate conversion. However, the low conductivity and saturated coordination of Bi-MOFs usually lead to poor performance, which severely limits their widespread application. Herein, a conductive catecholate-based framework with Bi-enriched sites (HHTP, 2,3,6,7,10,11-hexahydroxytriphenylene) is constructed and the zigzagging corrugated topology of Bi–HHTP is first unraveled via single-crystal X-ray diffraction. Bi–HHTP possesses excellent electrical conductivity (1.65 S m−1) and unsaturated coordination Bi sites are confirmed by electron paramagnetic resonance spectroscopy. Bi–HHTP exhibited an outstanding performance for selective formate production of 95% with a maximum turnover frequency of 576 h−1 in a flow cell, which surpassed most of the previously reported Bi-MOFs. Significantly, the structure of Bi–HHTP could be well maintained after catalysis. In situ attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) confirms that the key intermediate is *COOH species. Density functional theory (DFT) calculations reveal that the rate-determining step is *COOH species generation, which is consistent with the in situ ATR-FTIR results. DFT calculations confirmed that the unsaturated coordination Bi sites acted as active sites for electrochemical CO2-to-formate conversion. This work provides new insights into the rational design of conductive, stable, and active Bi-MOFs to improve their performance towards electrochemical CO2 reduction.

then capped and sonicated for 30 minutes until the solid was dissolved. The reaction mixture was heated in an oven at 120°C for 12 h to produce small dark crystals. The autoclave was allowed to cool naturally, and the obtained crystals were washed with deionized water for five times.

Synthesis of In-HHTP.
A solid mixture of HHTP (7 mg) and In(NO 3 ) 3 ·5H 2 O (10 mg) was added to deionized water (4 mL) in a glass vial. The reaction mixture was then heated in an oven at 85°C for 12 h to produce small dark crystals. The vial was allowed to cool naturally and the obtained crystals were washed with deionized water for five times.
Big crystals of Bi-HHTP. The big crystals could be synthesized as a phase mixture under hydrothermal conditions by adding Bi(NO 3 ) 3 ·5H 2 O (18 mg) and HHTP (7 mg) to a Teflon-lined autoclave containing 9 mL deionized water. The reactor was then heated to 120°C for 100 h. The products were then filtered off and washed with deionized water for three times and then dried at 60°C overnight.
Working electrode preparation for electrochemical CO 2 reduction. Catalyst (10 mg) was uniformly ground to powder and dispersed into a mixed solution of 100 µL of Nafion solution (5 wt.%) and 850 µL of isopropanol solution, then sonicated for 60 min to form a homogeneous ink. 100 µL of the well-dispersed ink was loaded onto a CP with an area of 1 cm 2 (size of CP: 1.5 × 1.5 cm 2 ) and dried under ambient conditions. Synthesis of Ni-HHTP-NF. The self-supported Ni-HHTP-NF was fabricated by a facile bath oiling approach. Typically, NF with the size of 1.5 × 1.5 cm 2 was firstly treated with 2.0 M HCl solution for 1 h and then washed with deionized water for 3 times and ethanol for 3 times. After drying at room temperature, the pre-treated NF was immersed into an autoclave containing HHTP (7 mmol), Ni(CH 3 COO) 2 ·4H 2 O (10 mg) in deionized water (4 mL). Then the autoclave was sealed and heated at 85°C for 12 h.
The resulting Ni foam was washed with the mixture of water and ethanol (v/v= 1:1) for three times and finally dried in an ambient environment. Ni-HHTP was synthesized by the same procedure without NF.
Synthesis of Ni-BDC-NF. The self-supported Ni-BDC-NF was fabricated by the previous work, where NF was used as both the metal source and substrate. 1 Typically, NF was firstly treated with 2.0 M HCl solution, and then washed with deionized water for 3 times and ethanol for 3 times. Then, the pre-treated NF was immersed into an autoclave containing 1.20 mmol H 2 BDC, 25 mL DMA and diluted HCl solution (5.0 mL, 0.05 M). Then the autoclave was sealed and heated at 120°C for 20 h. After cooling, the resulting Ni foam was taken out, and then ultrasonically treated in deionized water for 1 min, washed with the mixture of water and ethanol solution (v/v= 1:1) for three times, and finally dried in the ambient environment. Thermogravimetric analysis (TGA). Thermogravimetric analysis data were gathered on a sample of Bi-HHTP using a TA Instruments Discovery TGA. The sample was put into a platinum crucible and heated in the air from 28°C to 800°C with a heating rate of 10°C min -1 .

Electrochemical CO 2 Reduction Measurements. Electrochemical measurements
were performed with a CHI 760E electrochemical workstation (Shanghai, Chenhua, China) using a three-channel flow cell comprising Bi-HHTP/CP as the gas diffusion layer (GDL), NF as a counter electrode, and Ag/AgCl (saturated KCl electrolyte) as a reference electrode. For electrochemical CO 2 reduction experiments, the anion exchange membrane separated the cathode and anode chamber (Fumasep, FAB-PK-130). And the electrolyte was 1 M KOH without special treatment. The electrolyte in the cathode and anode was circulated by a peristaltic pump (Kamoer, F01A-STP), with a flow rate of 10 mL min -1 . The high-purity CO 2 (99.9995%) gas was continuously passing through the flow chamber with the flow rate of 20 mL min -1 via a mass flow controller (Sevenstar, D07-7B). All potentials were measured versus an Ag/AgCl reference electrode without iR compensation. The potentials were converted to the RHE scale with the following equation: E (versus RHE) = E (versus Ag/AgCl) + 0.197 + 0.059 × pH, and the presented current density was normalized to the geometric surface area (1 × 1 cm 2 ). All experiments were carried out at room temperature.
Product Quantification. The liquid products were quantified by 1 H NMR spectroscopy (Bruker-DRX 400 MHz) using dimethyl sulfoxide (DMSO) as an internal standard. The pre-saturation method was used to suppress the water peak. The FE was calculated using the following equation: Where 96485 is the Faraday constant (C mol -1 ), n is the number of products per milliliter (mol mL -1 ), N is the electron transfer number (formate, CO, and H 2 ), 20 is the flow rate of CO 2 (mL min -1 ), 60 is reaction time (min) and Q is the total charge obtained from chronoamperometry. was estimated by plotting the Δj (j a -j c ) at -0.7 V vs. RHE against the scan rates, in which j a and j c are the anodic and cathodic current densities, respectively. The linear slope was equivalent to the C dl .   The thermal stability of Bi-HHTP was measured by TGA in Fig. S1c, and Bi-HHTP exhibited a high temperature (325°C). Subsequently, the specific surface area of Bi-HHTP was then structurally characterized using Brauner-Emmet-Teller (BET) analysis utilizing N 2 adsorption and desorption isotherms obtained on a Micromeritics 3FLEX instrument, and the BET of Bi-HHTP is 49.40 m 2 g -1 (Fig. S1d). In our work, besides Bi-HHTP, Ni-HHTP, and In-HHTP were also synthesized to investigate the relationship between metal elements and performance. TEM image indicates that Ni-HHTP shows a nanorod-like morphology (Fig. S2a), and the uniform pillar structure and lattice spacing of 0.89 nm shown in the HRTEM reveal the prominent crystal characteristics of Ni-HHTP (inset of Fig. S2a). The XRD result further certifies the crystal characteristics of Ni-HHTP (Fig. S2b). The peaks at ∼380 nm and ∼700 nm in UV-vis spectrum of Ni-HHTP respond to π-π* and LMCT, respectively (Fig. S2c). The stability of Ni-HHTP was also evaluated by subjecting the material to the KOH aqueous solution (Fig. S2d), and the result demonstrated that it remained intact even in the strong base. Similarly, TEM image shows that In-HHTP displays a nanorod-like morphology (Fig. S3a). Moreover, the structure of In-HHTP was similar to that of Ni-HHTP (Fig. S3b). The EDS result in Fig. S3c shows that In ions account for 25.56% of the total structure. Meanwhile, the uniform distribution of the In, C, and O elements throughout the whole structure was shown in the corresponding element mapping (Fig. S3d).    As illustrated in Fig. S5 and S6, the electrocatalytic performances of In-HHTP and Ni-HHTP towards CO 2 RR were also tested. Firstly, In-HHTP exhibited a more positive onset potential and dramatically larger current densities than those in the N 2 -purged electrolytes, signifying that the In-HHTP could also lower the reaction barriers for electrocatalytic CO 2 RR (Fig. S5a). The current density of In-HHTP was around 40 mA cm -2 at -1.1 V vs. RHE (Fig. S5b) and the best FE formate was 80%, indicating its worse performance compared to Bi-HHTP. The TOF of Bi-HHTP was approximately three times that of In-HHTP (Fig. S5c). The production rate of formate for Bi-HHTP was five times greater than that for In-HHTP at -1.1 V vs. RHE. However, Ni-HHTP showed poor performance towards electrocatalytic CO 2 RR (Fig. S6), in terms of current density and selectivity (the main product was H 2 ).  Given the unique nanostructure, Bi-based MOFs were usually employed as excellent pro-electrocatalysts, generating the metallic Bi for CO 2 RR via the cathodic in situ reductions. 16 In response to the question of whether reduction and reconstruction occurred in Bi-HHTP during CO 2 RR, a series of structural characterizations on catalysts were conducted. Firstly, as shown in Fig. S8a and S8b, the morphology of Bi-HHTP did not undergo significant change, compared with that before the catalysis.
XRD results showed that no diffraction character of metal Bi substance (No.  existed after excluding the signals of CP (Fig. S8c), indicating that Bi-HHTP was indeed stable in the process of CO 2 RR. The FTIR spectra of Bi-HHTP further demonstrated that the characteristic peak of Bi-HHTP did not disappear and that the Bi-O vibration existed after electrolysis, which also certified the maintaining topology of Bi-HHTP (Fig. S8d). Furthermore, XPS was further performed to shed light on the surface electronic states of the elements before and after electrolysis, where the binding energies at 159.2 and 164.5 eV could be assigned to Bi 4f 7/2 and Bi 4f 5/2 of Bi 3+ , respectively. 17 The Bi peak of reacted Bi-HHTP did not shift, compared with the pristine counterpart (Fig. S8e). Moreover, the binding energies of O 1s at 530.1 and 531.8 eV in Fig. S8f were respectively associated with the lattice oxygen (Bi-O) and chemically adsorbed oxygen (Bi-OH), where the chemically adsorbed oxygen dominated in Bi-HHTP. 18 The obvious peak that occurred at 535.4 eV after the reaction may be attributed to oxygen vacancy, implying that a large number of defects were further produced during the reconstruction of catalysts. Note that the coordinatively unsaturated metal sites could effectively promote ion transport, 19 which is beneficial to CO 2 RR. The conventional electrochemical CO 2 conversion system includes the oxygen evolution reaction (OER) and CO 2 RR. However, the OER with high thermodynamic potentials and slow kinetics could consume about 90% of the power input, which reduced the economic efficiency of this system. Furthermore, compared to OER, the thermodynamically favorable methanol oxidation reaction (MOR) can significantly diminish the power input with fast kinetics and positive potentials to reduce the energy consumption in the overall system. The Ni-based metal-organic frameworks (MOFs) with permanent porosity and the exposed metal sites showed a prominent performance for MOR. 1   As shown in Fig. S9, the OM and SEM images show that the surface of the pretreated NF was smooth ( Fig. S9a and S9c). Subsequently, an obvious dendrite on the NF was observed after the growth of Ni-HHTP ( Fig. S9b and S9d). The morphology of Ni-HHTP on NF is nanorod-like with a size of ~200 nm (Fig. S10a and S10b). The EDS data in Fig. S10c displayed that the Ni element only accounted for 6.96 wt% on the surface of materials, and all elements (C, O, Ni) are evenly distributed on the surface of Ni-HHTP-NF (Fig. S10d).  The XRD pattern of Ni-HHTP-NF is consistent with the simulated one and the asprepared MOF powder, confirming its high purity (Fig. S11a). 38  As a comparison, we also synthesized Ni-BDC-NF via the reported method, 44 and its characterization was displayed in Fig. S15. SEM images in Fig. S12a and S12b show that the Ni-BDC displays a nanosheet-like morphology, where Ni-BDC is uniformly distributed on the surface of NF by layer-by-layer stacked arrays. XRD pattern of Ni- Fig. S12c is well consistent with that of the reported, further indicating that Ni-BDC is successfully grown on NF. 42, 43 A strong absorption peak occurred at around 200 to 400 nm in the UV-vis spectrum (Fig. S12d), attributed to the transfer of the charge from the oxygen center of the organic bridge to the metal center. 44 It was reported that Ni 2+ species can be transformed into Ni 3+ species (NiOOH) during electrooxidation reactions, which were considered as the real active species. 45, on the electrocatalyst are considered to be the catalytically active species. 47 That means more stable electrocatalysts can be utilized in this field and nickel non-hydroxides could exhibit better performance. In situ Raman spectroscopy was used to investigate the changes in the catalyst during the OER process. The in situ Raman spectrum of Ni-HHTP-ET for OER under different potentials (1 M KOH electrolyte) was shown in Fig.   S13. In the OER process, the broad peak observed in the 400 -600 cm -1 wavenumber region was attributed to the formation of NiOOat high potentials (~1.97 V). 48 S14c). Interestingly, the real part at low frequency was contracted at ~1.5 V (Fig. S14d), attributed to the adsorption of electroactive species. activity and kinetics. 45 The ECSA was then estimated by the C dl based on CVs recorded at different scan rates in the non-faradaic region from 0.5 -1 V vs. RHE (Fig. S16). As shown in Fig. S16a and S16b, the current densities increased with increasing scan rates over Ni-HHTP-ET and Ni-BDC-ET. The higher C dl of Ni-HHTP-ET (13.84 mF cm -2 ) illustrated that the Ni-HHTP-ET could provide more effective active sites for MOR, compared to Ni-BDC-ET (5.94 mF cm -2 ). The higher C dl may also attribute to the characteristic organic ligand (HHTP) which could form a large π-π conjugated system and enhance ion transport or promote the penetration of electrolytes and strong adsorption of ions.  In situ FTIR spectroscopy as a powerful tool was also employed to explore the mechanistic studies of MOR in Ni-HHTP-ET electrodes by the detection of surface radicals and reaction intermediates using a commercial in situ attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) device (Fig. S17) and

BDC-NF in
providing information about characteristic vibrations of molecules on surfaces during the reaction. As shown in Fig. S18a, from the onset of the reaction at 1.35 V vs. RHE, some bands appeared in the spectra. The band at 1382 cm -1 was assigned to δ(CH) or ρr(COO) of HCOO -/HCOOCH 3 -, and the band at 1254 cm -1 was assigned to v(C-O) of formate. 49,50 The band at 1640 cm -1 can be assigned to antisymmetric stretching vibration bands of OCO. 51 Furthermore, no evidence was found of the presence of either adsorbed CO species (bridge-bonded CO or linearly-bonded CO, 1700-2000 cm -1 ) in the in situ ATR-FTIR (Fig. S18b). 50,52 Besides, the band of CO 3 2at around 1400 cm -1 can be found, indicating that further oxidation after the formation of formate may also occur in this reaction. 47 To elucidate the underlying reason for the activity of Ni-HHTP-ET for MOR, density functional theory (DFT) calculations were performed. The Gibbs free energy profiles for the MOR process on Ni-HHTP are illustrated in Fig. S18c. The whole process follows (I-VII step): CH 3 OH adsorption (II), CH 3 O * , CH 2 O * , CHO * , * HCOOH, COOH * adsorption and HCOOH generation. It can be seen that the potential-determining step for MOR is all contributed by the process of * CH 3 OH → * CH 3 O+H + + e − . VII is a competition reaction, with HCOOH working as the main production, which is consistent with the result of the band at 1382 cm -1 assigned to δ(CH) or ρr(COO) of HCOO -/HCOOCH 3 -, the band at 1254 cm -1 assigned to v(C-O) of formate, and the band at 1640 cm -1 assigned to antisymmetric stretching vibration bands of OCO, as confirmed by in situ ATR-FTIR.