The formation of (NiFe)S2 pyrite mesocrystals as efficient pre-catalysts for water oxidation

We have fabricated (NiFe)S2 pyrite mesocrystals in the form of nearly-single crystalline porous cubes that self-optimize for efficient catalysis of water oxidation under electrochemical conditions.


Synthesis of the pyrite porous cubes (PCs):
In a typical synthesis of the PC-1, 0.029 g of Ni(NO 3 ) 2 . 6H 2 O (0.1 mmol) was dissolved in 0.1 mL EtOH, then added into 3 mL of OAm and stirred for several minutes in an autoclave. At the meantime, 0.034 g of TAA (0.45 mmol) was dissolved in 1 mL of OAm, and added into the Ni(NO 3 ) 2 . 6H 2 O/OAm solution under vigorous stir. The stir was kept for another 5 min. Then the autoclave was sealed and heated at 180 o C for 4 hours, after which it was allowed to cool naturally to room temperature. The resulting products were separated by centrifugation and washed with cyclohexane and EtOH several times. The PC-2, PC-3 and PC-4 were prepared in a similar way, by adding 0.005 g (0.013mmol), 0.010 g (0.025mmol) and 0.020 g (0.

Synthesis of the (oxy)hydroxides:
In a typical synthesis of OH-1, 0.29 g of Ni(NO 3 ) 2 . 6H 2 O (1mmol) was dissolved in 15 mL of EtOH in an autoclave, then 3 mL of OAm were added into the solution, followed by adding 15 mL of cyclohexane. Then the autoclave was sealed and heated at 100 o C for 4 hours, after which it was allowed to cool naturally to room temperature. The resulting products were separated by centrifugation and washed with cyclohexane and EtOH several times. The OH-2 and OH-3 were prepared in a similar way, by adding 0.06 g (0.15mmol), 0.12 g (0.3 mmol) of Fe(NO 3 ) 3 . 6H 2 O together with 0.29 g of Ni(NO 3 ) 2 . 6H 2 O (1 mmol) at the beginning, respectively. The ratio of Ni to Fe determined by the inductively coupled plasma optical emission spectrometry (ICP-OES) of the OH-2 and OH-3 were 6.9:1 and 3.2:1, respectively.

Electrochemical tests:
The electrochemical tests were performed in 1M KOH solutions and conducted with a CHI650D electrochemical workstation using a three electrode configuration. Pt foil served as the counter electrode and Ag/AgCl electrode were used as the reference electrode. RHE potential was calculated as E(RHE) = E(vs Ag/AgCl) + 0.198 + 0.0592*pH. Before electrochemical test, the electrolyte solutions were purged with O 2 for at least 30mins. The working electrodes were prepared in this way: (1) materials (about 5 mg) were first loaded on Ketjin carbon (about 1mg) by dispersing them in cyclohexane, and sonicated for at least 30mins, followed by centrifugation and drying; (2) the carbon loaded materials were dispersed in water/ethanol/Nafion (5%) solutions (volume ratio:0.7:0.25:0.05) and sonicated for at least 30mins to prepare the ink; (3) then 14uL of the ink was dropped on the surface of a rotating disk electrode (0.196cm 2 ) and allowed to dry naturally. The loading amounts of metal species on the electrodes were determined by the ICP-OES (Results shown in Table S1). The current density provided in this paper was normalized by the geometry area of the electrode (0.196cm 2 ). The LSV plots were recorded by applying proper potential ranges at the scan speed of 10mV/s and the Tafel plots were recorded at the scan speed of 1mV/s at proper potential ranges. The rotating ring-disk electrode (RRDE) tests were conducted with a CHI 760D electrochemical workstation. Before electrochemical test, the electrolyte solutions (1M KOH) were purged with N 2 for at least 30mins. The collecting efficiency of the RRDE were measured in a solution of 0.1M KOH with 10 mM K 3 Fe(CN) 6 , and the results showed a collection efficiency of N=4.0.
The turnover frequency (TOF) were calculated based on all Ni and Fe amounts on the electrodes, by: 58.69 + ( , )

)
Characterizations: The X-ray diffraction (XRD) patterns were performed on a Bruker D8 ADVANCE X-ray powder diffractometer operated at 40 kV voltage and 40 mA current with Cu Kα radiation (λ = 1.5406 Å). The X-ray photoelectron spectroscopy (XPS) results were achieved by ESCALAB 250Xi (Thermo Fisher). The calibrations were done with respect to carbon C1s. The morphology and size of the products were determined by a Hitachi H-7700 transmission electron microscope (TEM) at 100 kV, a JEOL 2100F HRTEM equipped with energy dispersive X-ray spectroscopy (EDS) at 200 kV, was used to perform high-angle annular-dark-field (HAADF)-STEM and EDS mapping. Inductively coupled plasma optical emission spectrometry (ICP-OES) was used to determine the composition of the mass loading for electrochemical test. The soft X-ray adsorption spectrum (sXAS) measurements were carried out the at the beamline 4W9B in Beijing Synchrotron Radiation Facility (BSRF). The X-ray was monochromatized by a double-crystal Si (111) monochromator for BSRF. The energy was calibrated using a gold metal foil.   Although the carbon grid for TEM testing contains C and O, the shape of clusters embedded in gel can almost be identified from the signal of C and O in the mapping results. Moreover, the signal of N (results from OAm in the synthesis) can superimpose the gel. Together with the signals of S and Ni, the black precipitates when mixing all precursors can be inferred to be small nickel sulfide clusters of less than 2nm embedded in a gel.  The figure and table in the lower panel indicate the Randles model used in this report and the corresponding simulation results (the simulation was achieved using ZSimpWin software). The reason that imaginary impedance increases at high frequency is attributed to the contributions from cable and instrument capacitances and inductances. The R s is estimated by using the impedance at high frequency where the phase angle is closest to 0 degrees 1 . The upper row shows the CV scans of different materials with different scanning rate (10, 30, 50, 70, 90, 110 mV/s) at a potential range including their open circuit potentials. The ECSA were estimated from the electrochemical double layer capacitance 2 (C dl ). The 2C dl was estimated by plotting the ∆j (j anodic -j cathodic ) at a given potential against the scan rate (left graph in the lowest row).

Supporting
The middle row shows the CV scans of different materials with different scanning rate (10,30,50,70,90, 110 mV/s) at a potential range including their oxidation and reduction peaks. Generally, the oxidation peak current obeys the power law relation with the scan rate 3 (υ): . The value of b can be calculated by the slope of the logj verses logυ plot (right graph = in the lowest row, slopes are shown in the left). There are two well-defined conditions 4 : (1) If b is close to 0.5, the faradaic reactions happed in the bulk; (2) If b is close to 1.0, it indicates a surface-reaction limited response. Since the oxidation peak cannot be easily differentiated in the case of the PC-3 and PC-4, we used the reduction peak current here. From the results of the PC-1 and PC-2, the conclusion made based on oxidation peak current and reduction peak current was similar. All of them suggests a faradaic redox process in the bulk.

  
Where n app is the apparent number of electrons, and it was reported to be 2 at a rotating rate of 1600 rpm 2 ; i r and i d are the measured ring and disk currents, respectively, and N is the collection efficiency (4.0). The decrease of Faradaic efficiency with the increase of potential at the disk electrode was caused by the generation of oxygen bubbles 5 ; only dissolved oxygen can be reduced at the surface of the Pt ring electrode. When the current density (about 1 mA/cm 2 , 0.3mA in the graphs) at this potential is sufficiently large to generate fully dissolved oxygen without forming oxygen bubbles at the disk electrode. Thus, the calculated FE would decrease as the current increased. The calculated FE of 96% and 98% at 1.49V and 1.45V (of the PC-2 and PC-4, respectively) is representative of their OER FE.      The spectra of Ni 2p , Fe 2p , O 1s and S 2s scans are shown in the first, second, third and fourth row respectively. The analysis of Ni 2p and Fe 2p scans is showcased in Fig. S12. In the S 2s spectra, the peak centered at lower energy arises from S in pyrite 6 , the peak centered at higher energy (168-172 eV) arises from the surface absorption of oxygen with sulfur to form S-O bond 7 . The upper row showcases the analyzing of the Ni 2p 8 and Fe 2p 6, 9 data using the data of the PC-4 after CV scans. All the data were analyzed in the same way, and the results are shown in the lower row (the amount of O is considered in the left and omit in the right). The red color in the right panel indicate the amount of S, and other colors indicate metals, we can infer that the relative amount of remaining S (to total amount of metals) would be higher with higher content of Fe doping from this graph.      The values indicate the ratio of high valence to low valence, estimated by the deconvolution of the spectra (examples are shown below). It's worth to mention that this method can only qualitative tell the ratio of high valence, but cannot explicitly determine the valence of metals and their ratios. The qualitative results can suggest that: In the case of the PC-2 materials, the spectra would become more similar with that of the (oxy)hydroxide after CV scans, with similar ratio of high valence metals. However, the PC-4 materials showed different features, with lower ratio of high valence metals. But the PC-4 pre-catalysts indeed showed higher activity.