Structural Reconstruction of a Cobalt- and Ferrocene-Based Metal–Organic Framework during the Electrochemical Oxygen Evolution Reaction

Metal–organic frameworks (MOFs) are increasingly being investigated as electrocatalysts for the oxygen evolution reaction (OER) due to their unique modular structures that present a hybrid between molecular and heterogeneous catalysts, featuring well-defined active sites. However, many fundamental questions remain open regarding the electrochemical stability of MOFs, structural reconstruction of coordination sites, and the role of in situ-formed species. Here, we report the structural transformation of a surface-grown MOF containing cobalt nodes and 1,1′-ferrocenedicarboxylic acid linkers (denoted as CoFc-MOF) during the OER in alkaline electrolyte. Ex situ and in situ investigations of CoFc-MOF film suggest that the MOF acts as a precatalyst and undergoes a two-step restructuring process under operating conditions to generate a metal oxyhydroxide phase. The MOF-derived metal oxyhydroxide catalyst, supported on nickel foam electrodes, displays high activity toward the OER with an overpotential of 190 mV at a current density of 10 mA cm–2. While this study demonstrates the necessity of investigating structural evolution of MOFs during electrocatalysis, it also shows the potential of using MOFs as precursors in catalyst design.


Characterisation Techniques
The morphology of the sample was characterised using scanning electron microscopy (SEM, Tescan Essence) with energy-dispersive X-ray spectroscopy (EDX).Raman measurements were performed using a Bruker Senterra benchtop Raman microscope using 785 nm laser excitation and a 20× objective lens.The spectrometer was calibrated using a polystyrene standard.Raman spectra were collected over the 50-2630 cm -1 spectral range at 1 mW power with 60 s exposure time and 5 accumulations.Fourier-transform infrared spectroscopy (FTIR) was performed on a Bruker Alpha II using ATR accessory.Powder X-ray diffraction data was collected on a Bruker D8 Discover using a Cu Kα radiation source (λ=1.54178Å).Thermal analysis was performed on Netzsch STA 449 F3 model attached with mass spectrometer QMS403.During the measurements, the purge gas flow (N2) and protective gas flow (N2) were 50 mL min -1 and 20 ml min -1 , respectively.The samples were heated from 30 °C to 600 °C at a rate of 10 °C min -1 with simultaneous mass spec analysis of the evolved gas.X-ray photoelectron spectroscopy (XPS) was performed on a Thermo Fisher Scientific K-alpha+ spectrometer.Samples were analysed using a micro-focused monochromatic Al x-ray source over an area of approximately 200 microns.Data were recorded at pass energies of 150 eV for survey scans and 40 eV for a high-resolution scan with 1 eV and 0.1 eV step sizes, respectively.Data analysis was performed in CasaXPS using a Shirley type background and Scofield cross sections, with an energy dependence of -0.6.ICP analysis was performed on a Thermo Scientific iCAP 7000 series ICP-OES Spectrometer, using the Qutegra software.
Atomic Force Microscopy (AFM) was performed on an AFM Workshop HR-AFM.The images were processed and analysed using Gwyddion.X Band (9.4GHz) continuous Wave Electron Paramagnetic Resonance (EPR) measurements were carried out using a Bruker EMXPlus Spectrometer equipped with a Bruker ER4122-SHQ resonator.Cryogenic temperatures were achieved using a ColdEdge Stinger closed-cycle cryocooler mated to an Oxford Instruments ESR900 cryostat.Temperature maintenance and control was handled via an Oxford Instruments MercuryITC.
Electrochemical measurements: Electrochemical tests were carried out in 1M KOH using a three-electrode system on a BioLogic VSP potentiostat.A Ag|AgCl|KCl (satd.)reference electrode (CHI111) was used for all experiments.Pt-mesh and NF were used as the counter electrode for testing with FTO and NF, respectively.The Ag|AgCl|KCl (satd.)reference electrode was regularly calibrated against a master reference electrode and any drift in the potential (E') of the Ag|AgCl reference was corrected using the equation: The reversible hydrogen electrode (RHE) potentials were obtained with the equation: E (V vs RHE) = 0.197 + Ecorrected (V vs Ag|AgCl) + 0.059 × pH Linear sweep voltammetry (LSV) measurements were performed at 10 mV s -1 using the reverse method, to avoid overlapping of oxidation peaks with the metal ion and water.Overpotentials (η) are calculated using the equation: Tafel slopes were calculated from an LSV performed at 10 mV s -1 utilising the Tafel equation: η = b log(j) + a; where η denotes overpotential, j denotes current density, a is a constant and b is the Tafel slope.Chronopotentiometry (CP) was performed at 10 mA cm -2 to evaluate the stability of the CoFc MOF.The variation in potential over the course of 24 h chronopotentiometry was determined to be ~1.3%.During chronopotentiometry, the potential stabilised at ~1.470 V (vs.RHE) after ~1.4 h, followed by a gradual decrease to ~1.455 V after 8h.Afterwards, the potential slowly increased to 1.475 V after 24 h electrolysis.The increase in potential was ~20 mV from 8h to 24 h corresponding to a ~1.3% increase.
The electrochemical active surface area (ECSA) was calculated by recording CVs at variable scan rates (5-100 mv s -1 ) in the non-faradaic region from 0.75 to 0.80 V (vs RHE).The doublelayer capacitance (Cdl) is obtained by plotting the capacitive currents at 0.775 V, with the Cdl equalling half the gradient.
XAS measurements: XAS data was collected at beamline B18 at the Diamond Light Source (beamtime number SP-34632-1).Pellets (13 mm) for pristine and KOH-soaked CoFc-MOF were prepared by homogeneously mixing MOF powders with an inert cellulose matrix and mounted on a sample holder using Kapton tape.For the post-catalysis sample, 0.25 mL catalyst ink containing CoFc-MOF (16 mg CoFc-MOF dispersed in 2 mL IPA containing 0.04 mL 5 wt% Nafion solution using 30 min sonication) was drop casted on 1×1 cm 2 carbon paper electrode (AvCarb GDS2120), which was subjected to 6 h controlled potential electrolysis at 1.5 V.The electrode was cut into four 0.25 cm 2 squares and stacked using Kapton tape for XAS measurements.Co(OH)2 (97%, Thermo Scientific Chemicals) and CoOOH (synthesised using a literature method) 1 were used as the cobalt standards and Ferrocene as the iron standard.The PXRD patterns of Co(OH)2 and CoOOH are shown in Figure S23.
Measurements of the pellets were performed in transmission mode (3 scans), while the postcatalysis electrode was measured in fluorescence mode (6 scans).XAS data processing and EXAFS analysis were performed using Athena and Artemis software, respectively.Samples were prepared for HPLC using the 50 μL aliquot of electrolyte and 950 μL of 1% acetic acid solution in 6:4 acetonitrile:water.Samples were analysed using a 6:4 acetonitrile:water solvent mix, with a flow of 0.2 mL s -1 .The run time was 15 minutes, with an oven temperature of 30°C and an analysis wavelength of 254 nm.

UV-vis
In situ ICP: In situ ICP-MS measurements were performed using an Agilent 7900 ICP-MS in combination with a custom electrochemical flow cell with an active surface area of 0.07069 cm 2 .Agilent's MassHunter 4.3 was used as the operating software.Samples were dropcast onto 5x5 mm carbon paper squares with a one-sided loading of 0.08 ± 0.05 mg.To apply the electric potentials, a BioLogic VSP-300 potentiostat with a Pt counter electrode and an Ag/AgCl reference electrode was used.A 10 mM KOH electrolyte solution was used (higher electrolyte concentrations would lead to clogging of the ICP-MS).The electrochemical stability tests were performed at constant potentials of 1.4 V and 1.5 V vs. RHE, with flow rates of 8.22 mg s -1 and 7.259 mg s -1 for total measurement times of 2 h.To quantify Fe and Co, calibration with Merck's ICP multi-element standard solution XVI was performed.

EPR and SQUID measurements: Typical measurement parameters used for EPR:
Microwave frequency 9.37GHz, microwave power 2 mW, modulation amplitude 5 G, sweep time 120 s, and sweep range 0-17,000 G.Additional high frequency EPR measurements were attempted using a Bruker ER5106QT (Q Band, 34GHz) and ER5106KT (K Band, 24 GHz) Flexline resonator inside an Oxford Instruments CF935 Cryostat cooled by the same ColdEdge Stinger.The bulk magnetic properties of the powders were studied on a Quantum Design MPMS3 SQUID magnetometer.Magnetic Susceptibility measurements in the range 1.8 K to 300 K along with isothermal magnetisation (at 300 K) between 0 and 7 T. Diamagnetic corrections were applied to account for the response of the gelatin capsule and sample holder.

Optimisation of CoFc-MOF structure:
The CoFc-MOF structure was optimised based on a ZnFc-MOF previously reported, 2 changing the Zn atoms with Co. Periodic Density Functional Theory (DFT) calculations were carried out using the CP2K software package. 36][7] A cut-off energy of 10.8 keV (800 Ry) and relative cut-off energy of 816 eV (60 Ry) to map the primary Gaussian basis set onto the auxiliary plane-wave basis.This converges unit cell energies to within 10meV.Calculations were performed using the Perdew-Burke-Ernzerhof (PBE) XC functional optimised for solid systems (PBEsol), 8,9 with Grimme's D3 dispersion correction. 10,11A 2×8×4 Monkhorst-Pack k-point mesh grid was used for all calculations.SCF convergence was set to the largest density matrix element changing by <1.0E-6 and fermi-dirac smearing was used with an effective electronic temperature of 500 K.
Geometry optimisations and cell optimisations were carried out unit maximum forces on atoms were <0.02 eVÅ -1 and pressure on the cell <10 MPa.S3.S4.

Supplementary Figures
Spectroelectrochemistry: In situ spectroelectrochemical analysis was performed in a glass cuvette (path length 1 mm) using a CP|CoFc-MOF working electrode (CP = carbon paper, AvCarb P50T), a Ag|AgCl|KCl (satd.)reference electrode, and a Pt wire counter electrode.An AvaLight-DHC (deuterium and halogen light source) was used as the light source, and the spectra were recorded using an AvaSpec-2048 spectrometer.The electrode was equilibrated in the electrolyte at the open-circuit potential for 1 hour before applying the oxidative potential.The UV-vis spectrum was recorded at regular intervals.Ex situ HPLC: HPLC was performed using a Shimadzu Prominence modular setup, comprising of a SIL-20A Auto Sampler, DGU-20A5 Degasser, LC-20AD Liquid Chromatograph, CTO-20A Column Oven and SPD-20A UV-Vis Detector connected with a CBM-20A Communication Bus.The LabSolutions software package was utilised for sample acquisition and post-processing.Chronoamperometry was performed at 10 mA cm -2 for 2 hours in 0.1M KOH, using a NF counter electrode and Ag|AgCl|KCl (satd.)reference electrode.50 μL aliquots of electrolyte were taken at minutes 2, 5, 10, 20, 30, 40, 50, 60 and 120.

Figure S3 .
Figure S3.SEM image of bulk CoFc-MOF.The smaller aggregates on the sheets indicate that the bulk material is a mixture of two types of particles with different morphologies.

Figure S4 .
Figure S4.(a and b) SEM image of NF|CoFc-MOF at different magnifications.

Figure S7 .
Figure S7.(a) AFM image of a CoFc-MOF flake, and (b) AFM height profile associated with black line.

Figure S8 .
Figure S8.Powder X-ray diffraction patterns of NF|CoFc-MOF, NF|NiFc-MOF, and NF|FeFc-MOF, demonstrating similar structure of the three surface-grown materials.Panel (a) shows the non-baseline corrected data and (b) shows the baseline corrected data in the range 2θ = 5-50°.The intense reflections at ~28°, ~44°, ~52°, and ~76° observed in all three electrodes originate from the sample holder and nickel foam support.

Figure S11 .
Figure S11.NMR spectrum of CoFc-MOF with terephthalic acid internal reference.100 µL D2SO4 and 500 µL DMSO-d6 were added to 2.5 mg CoFc-MOF.The resulting suspension was heated and sonicated until all particulate was dissolved.3 mg terephthalic acid (TPA) was added as an internal reference.

Figure S13 .
Figure S13.Co 2p XP spectra of as-synthesised, KOH-soaked, and post-catalysis CoFc-MOF.The solid black lines show the envelope of the fitted peaks.

Figure S14 .
Figure S14.Fe 2p XP spectra of as-synthesised, KOH-soaked, and post-catalysis CoFc-MOF.The solid black lines show the envelope of the fitted peaks.

Figure S15 .
Figure S15.(a) C 1s and (b) O 1s XPS spectra of as-synthesised, KOH-soaked, and postcatalysis CoFc-MOF.The two additional peaks at 292.9 eV and 295.7 eV in the C 1s XPS os KOH-soaked and post-catalysis samples (panel a) correspond to K 2p3/2 and 2p1/2 signals, respectively.(c) O 1s XPS peak fit for as-synthesised and post-catalysis samples with the solid black lines showing the envelope of the fitted peaks.

Figure
Figure S17.a) CV curves of NF|CoFc-MOF at different scan rates b) Plot of capacitive current of NF|CoFc at 0.7 V vs RHE.Cdl = 4.71 mF cm -2 c) CV curves of NF at different scan rates d) Plot of capacitive current of NF at 0.7 V vs RHE.Cdl = 1.15.

Figure
Figure S18.a) Continuous CV of NF|CoFc-MOF in 1M KOH in the non-Faradaic potential window.b) Selected voltammograms showing the change in shape of CV with increasing number of scans.The geometric surface area of the electrode is ~0.5 cm 2 .

Figure S19 .
Figure S19.Linear dependence of capacitive current on the scan rate for NF|CoFc-MOF after increasing number of continuous CV scans in the potential window 0.87-0.97V V (Figure a) and 0.87-1.37V V (Figure b).The capacitive current densities at varying scan rates were determined from the difference in cathodic and anodic current densities at 0.92 V.

Figure S20 .
Figure S20.Nyquist plots of NF|CoFc-MOF electrode after CV scans with different cycle numbers.The voltammetric scans were performed over the following potential windows: (a) 0.87-0.97V, (b) 0.87-1.37V, and (c) 0.87-1.57V.The EIS experiments were carried out with an applied potential of 0.87 V vs. RHE in 1.0 M KOH.

Figure S23 .
Figure S23.(a) PXRD pattern of KOH-soaked CoFc-MOF compared to that of commercial β-Co(OH)2 and CoOOH (synthesised according to literature protocol); the PXRD data for soaked CoFc-MOF was baseline corrected.(b) PXRD pattern of KOH-soaked MOF is compared against the patterns of β-Co(OH)2 (collection code 257275), α-Co(OH)2 (collection code 172037) and CoOOH (collection code 56288) obtained from the Inorganic Crystal Structure Database.(c) PXRD pattern of KOH-soaked CoFc-MOF (non-baseline corrected) showing the peaks in the smaller 2θ range.

Figure S24 .
Figure S24.Comparison of the Raman spectra of as-synthesised NF|CoFc-MOF (black trace) and post-catalysis NF|CoFc-MOF (red trace).For the post-catalysis sample, the electrode was used for chronopotentiometry at 10 mA cm-2 for 12 hours.The bottom panel shows 1600-2600 cm -1 region and the data were baseline-corrected using the in-built function in OPUS software.

Figure S25 . 1 .
Figure S25.(a) UV-vis spectra of 1,1'-ferrocene dicarboxylic acid (FcDA) in aqueous KOH (1 M) at different concentration.The inset shows the plot of absorbance at 445 nm versus concentration of FcDA which produced a molar extinction coefficient (ε445nm) of 314.7 M -1 cm - 1 .The 306 nm and 445 nm bands are attributed to charge-transfer and d-d transitions, respectively. 12(b) In situ UV-vis spectroelectrochemistry of the electrolyte during chronopotentiometry with CoFc-MOF electrode at different applied potential (vs.RHE).The catalyst loading on the carbon paper electrode was ~0.8 mg cm -2 .The electrode was held at each potential for 2 minutes with stirring before recording the spectrum.The black and red trace show the pre-electrolysis UV-vis spectra (at open circuit voltage, OCV) of the electrolyte after equilibriating the electrode for 1 min and 60 min, respectively.(c) The current vs. time trace for the spectroelectrochemistry experiment showing the anodic current at different applied potential.

Figure S26 .
Figure S26.Variable-field magnetization data for pristine and KOH-soaked CoFc-MOF collected at 300 K.The KOH-soaked sample was prepared by soaking the MOF in 1 M KOH for 2 h.

Figure S28 .
Figure S28.The Fourier transform of e k 3 -weighted EXAFS spectra of Co K-edge for (a) pristine CoFc-MOF, (b) KOH-treated CoFc-MOF, and (c) post-catalysis CoFc-MOF.The open circles show the experimental data and the solid lines present the simulations.The fit parameters for the simulations are given in TableS3.

Figure S29 .
Figure S29.The Fourier transform of e k 3 -weighted EXAFS spectra of Fe K-edge for (a) pristine CoFc-MOF, (b) KOH-treated CoFc-MOF, and (c) post-catalysis CoFc-MOF.The open circles show the experimental data and the solid lines present the simulations.The fit parameters for the simulations are given in TableS4.

Table S1 .
Experimental mass loss of CoFc-MOF observed in the TGA curve, and the calculated mass loss based on a molecular formula of [Co(OH)(FcDA)•DMF].

Table S2 .
Atomic and weight percentages from EDS for pre-and post-catalysis NF|CoFc

Table S4 .
EXAFS fitting parameters for k 3 -weighted Fe K-edge data.FEFF input was built using crystal structure data for ferrocene and Fe(III) layered double hydroxide (ICSD # 159700).