Effect of Intrinsic Properties of Anions on the Electrocatalytic Activity of NiCo2O4 and NiCo2O x S4-x Grown by Chemical Bath Deposition.

Electrochemical water (H2O) splitting is one of the most promising technologies for energy storage by hydrogen (H2) generation but suffers from the requirement of high overpotential in the anodic half-reaction (oxygen evolution), which is a four-electron process. Though transition-metal oxides and oxysulfides are increasingly researched and used as oxygen evolution electrocatalysts, the bases of their differential activities are not properly understood. In this article, we have synthesized NiCo2O4 and NiCo2O x S4-x by a chemical bath deposition technique, and the latter has shown better oxygen evolution performance, both in terms of stability and activity, under alkaline conditions. Comprehensive analysis through time-dependent cyclic voltammetry, microscopy, and elemental analysis reveal that the higher activity of NiCo2O x S4-x may be attributed to the lower metal-sulfur bond energy that facilitates the activation process to form the active metal hydroxide/oxyhydroxide species, higher electrochemically active surface area, higher pore diameter and rugged morphology that prevents corrosion. This work provides significant insights on the advantages of sulfur-containing materials as electrochemical precatalysts over their oxide counterparts for oxygen evolution reaction.

Introduction

With increase in global population, the demand for energy is sharply increasing. For energy generation, major economies are primarily dependent on the consumption of fossil fuels, which leads to the emission of toxic materials.[1] These materials are responsible for climate change that has destructive implications on the biodiversity and sustainability of life on earth. Creating a global scale sustainable energy generation and storage system is one of the most profound challenges humanity is facing today.[2−5]

Among clean fuels, hydrogen is considered as one of the best candidates because of its high mass-specific energy and zero carbon emission. Electrochemical water splitting has been considered as one of the most promising techniques for hydrogen generation. In this process, electrochemical energy splits water (H2O) into H2 and O2.[1,6] However, this process has a positive Gibb’s free energy (ΔG). Moreover, when an electrochemical cell is “driven” toward completion by applying reasonable potential, it becomes kinetically controlled because of activation energy, ion mobility (diffusion), surface hindrance, and entropy.[7,8] In practice, catalysts are generally employed to minimize the activation energy of the reaction.[9] While hydrogen evolution is a two-electron process, evolution of oxygen is a four-electron process, which makes the kinetics of overall water splitting unfavorable. This has encouraged scientists to develop efficient catalysts to lower the potential needed to perform oxygen evolution.[10] Among them, RuO2 and IrO2 have shown promising activities.[11,12] However, the scarcity and cost of these metals limit their use as catalysts in practical devices. The abundance and low-cost of transition-metal compounds make them a prudent choice to be used as electrocatalysts. In the last decade, enormous effort has been made in the synthesis, characterization, and electrocatalytic performance evaluation of transition-metal (mostly Ni, Co, and Fe) compounds.[7,13,14] Recently, both nickel hydroxide and cobalt hydroxide have shown a promising activity as bifunctional catalysts (efficient for both hydrogen and oxygen evolution).[14,15] For example, a facile synthesis of binary Ni–Fe oxide electrocatalyst supported on Ni foam for oxygen evolution has been reported by Liu et al.[16] NiCo2S4 nanowires synthesized on Ni foam have shown superior activity as a bifunctional catalyst.[17] Recently, Nandan et al. have developed Fe–NC/CN centers as potential candidates for both oxygen evolution reaction (OER) and oxygen reduction reaction and successfully employed them as air cathodes in metal–air batteries.[18,19] Although most articles solely attribute the improvement of electrocatalytic activity of a material to the decrease in charge-transfer resistance of a catalyst, we believe that the inherent properties (such as hard–soft nature, electronegativity, bond dissociation energy, and so forth) of different elements that are present in a catalyst can also be correlated with their respective electrocatalytic activity.[16,17] Moreover, recent reports indicate that the as-synthesized material actually acts as a precatalyst, which transforms during electrochemical measurements to produce the active catalyst layer. Therefore, it is necessary to relate the electrocatalytic activity of a material with the evolution of an active catalyst layer.[20,21] Another major challenge in water-splitting research is the preparation of the working electrode. For evaluating the electrocatalytic activity, catalysts are generally drop-casted on an expensive glassy carbon electrode in the presence of binders (such as Nafion).[14,22,23] However, in practical water-splitting devices, this may not be suitable as the materials often leach out over time. This demands for a thin, robust, and stable catalyst layer over a cheap conducting substrate. Techniques such as chemical vapor deposition or physical vapor deposition are generally employed to directly deposit catalyst layers onto conductive substrates during catalyst synthesis. However, these methods are quite tedious and difficult to control.[24,25] Alternatively, a thin layer of catalyst materials can also be directly grown on substrates using hydrothermal reactors by appropriate tuning of synthetic conditions.[14−17,26,27] However, it has been noticed that although the same reaction conditions are used, minor differences in hydrothermal setup across labs often lead to the formation of different materials. This motivated us to develop a more general method that will use relatively cheap and easily available equipment and is able to produce robust thin layers on substrates.

In this article, precursors of NiCo2O4 and NiCo2OS4– layers were synthesized on fluorine-doped tin oxide (FTO) glass substrates by a chemical bath deposition (CBD) technique that requires only a glass bottle (250 mL) and a heating mantle as the apparatus. This method produced a uniform and robust layer of precursors, which was then annealed under specific conditions to synthesize NiCo2O4 and NiCo2OS4–. This two-step method ensures that the density and quantity of metal centers present per unit area of the substrate remain the same for both catalysts, and thus, any difference in catalytic activities of the two materials cannot be attributed to the difference in loading of the catalyst material on the substrate. The catalysts were then evaluated for their activity toward OER. Apart from the traditional concept of considering the charge-transfer resistance of the materials as the sole descriptor for the difference in their catalytic activity, differences in their rate of activation, electrochemically active surface area (ECSA), metal–anion bond dissociation energy, porosity and resistivity toward material corrosion were probed, and the results indicate these parameters to be noteworthy for the proper understanding of the catalytic performance of a material. The obtained results clearly illustrate the advantages of using sulfur-containing electrocatalysts, both in terms of activity and stability, compared to their oxygen counterparts for OER under alkaline conditions.

As a further support, density functional theoretical (DFT) calculations were carried out and the computed results relate the increase in electrocatalytic activity with the increase in covalency between metallic and anionic orbitals, which in turn are dependent on the intrinsic properties of different constituent elements in a compound.

Results and Characterization

Phase and Composition Analysis

All synthesized samples were characterized by the powder X-ray diffraction (PXRD) technique to obtain information about their phase. The peaks (Figure b) coming from the sample prepared by annealing the precursor in atmospheric conditions matched with the standard pattern (JCPDS—20-0781) (Figure a) of NiCo2O4. However, in this spectrum, the peaks from the FTO substrate also appear at 26°, 34°, 52°, and 62°. Especially, the peak at 38° is enhanced in intensity because of the overlap between the peaks originating from both the sample and FTO. As shown in Figure d, most peaks in the PXRD pattern of the sample annealed under argon atmosphere in the presence of thiourea resemble with that of NiCo2S4 (JCPDS—43-1477) (Figure c). In this sample as well, the peaks at 26°, 38° and 52° originate from the overlap between sample peaks and FTO peaks. In addition, there are few peaks (41°, 62°) that do not match with that of NiCo2S4. These peaks have probably originated from a small amount of nickel oxide (NiO) that has formed as a side product during synthesis.[28]

Figure 1

PXRD analysis of as-synthesized samples coated of the FTO glass substrate: (a) standard pattern of NiCo2O4 (JCPDS—20-0781), (b) as-synthesized NiCo2O4, (c) standard pattern of NiCo2S4 (JCPDS—43-1477), and (d) as-synthesized NiCo2OS4–. (*) indicates peaks originating from the FTO glass substrate and (Δ) indicates peaks originating from the overlap between diffraction from both the sample and the FTO glass substrate.

To obtain information about the composition of different elements in each sample, energy-dispersive X-ray spectroscopy (EDAX) analysis was carried out. The samples were carefully scraped from the FTO substrate, and the collected powder was subjected to EDAX analysis. Each sample was synthesized twice, and EDAX was collected from five different regions of each sample. Upon averaging the obtained values, the ratios of different elements were found to be Ni/Co/O = 1:1.9:3.9 and Ni/Co/O/S = 1:2.1:0.6:3.29 for NiCo2O4 and NiCo2OS4–, respectively (as shown in Figure S1 and Tables S1 and S2). The presence of a small amount of O in the sample prepared under argon atmosphere in the presence of thiourea indicates that it is not possible to entirely eliminate oxygen by this method.

Morphology Analysis

To obtain information about the morphology of the synthesized materials, microscopic analysis of the products on FTO glasses was carried out. Scanning electron microscopy (SEM) images indicate the formation of very thin flakelike structures (about 2 μm long) with smooth surfaces for NiCo2O4 (Figure a,b). A closer inspection reveals the formation of relatively smaller (50–150 nm) flakes in between the larger flakes. As shown in Figure a, the CBD technique has yielded a uniform and dense layer of NiCo2O4 on the FTO glass substrate. SEM images of NiCo2OS4– show a similar flaky structure but with a rugged surface (Figure c,d). These flakes are however slightly thicker than those of NiCo2O4. Here again, the layer is dense and uniform in nature (Figure c).

Figure 2

SEM images of (a,b) NiCo2O4 and (c,d) NiCo2OS4– on the FTO glass substrate.

Inductively Coupled Plasma Mass Spectroscopy (ICP-MS) Analysis

During CBD, two FTO glass electrodes were always placed in the same bath simultaneously. Later, one of them was converted to NiCo2O4 and the other into NiCo2OS4–. This was done to ensure equal loading of the precursor on the electrodes. To further confirm this, the materials were characterized by the ICP-MS analysis, whose details (sample preparation, calculation, etc.) have been presented in the Supporting Information. The results show that the amount of catalyst loading per unit area of the electrode was 0.351 and 0.349 mg/cm2 for NiCo2O4 and NiCo2OS4– respectively. Moreover, the ratio of concentrations of metal ions (Co/Ni) were found to be 1.88:1 and 1.93:1 for NiCo2O4 and NiCo2OS4–, respectively. This indicates that any difference in electrocatalytic activity cannot be attributed to the difference in the number of metal ions.

Electrochemical Activity

To evaluate the oxygen evolution activity of both NiCo2O4 and NiCo2OS4– on FTO, they were characterized by electrochemical techniques as described in section . As shown in Figure , NiCo2OS4– attains the current density of 10 mA/cm2 at 1.60 V, whereas NiCo2O4 shows a similar activity at 1.66 V. Thus, to obtain the benchmark current density, NiCo2O4 requires about ∼60 mV more potential compared to NiCo2OS4– at pH 14. This clearly suggests that NiCo2OS4– is a better electrocatalyst compared to NiCo2O4 for OER under alkaline conditions. The OER overpotential value of NiCo2OS4– is either better or comparable to most other reports on NiCo-based electrocatalysts (as shown in Table S3).

Figure 3

CV curves (recorded at 10 mV s–1) of OER at pH 14 for (a) NiCo2O4 and (b) NiCo2OS4–. The inset shows the overpotential values for the electrocatalysts.

The surface area and porosity of a material are known to have significant effect on its catalytic activity toward any reaction. To garner information regarding these parameters, nitrogen adsorption–desorption measurements were conducted on catalyst nanopowders, which were prepared by appropriate treatment (details in Materials and Methods: Synthesis section) of the precipitate that had formed during CBD synthesis. As shown in Figure S2, the obtained nitrogen adsorption–desorption isotherms correspond to type IV and the H-3 hysteresis loop in the isotherms indicate mesoporosity with a slit-like pore geometry in both samples according to IUPAC classification.[29] The Brunauer–Emmett–Teller (BET) surface areas for NiCo2O4 and NiCo2OS4– were found to be 83 and 16 m2 g–1, respectively. However, the average pore diameters of NiCo2O4 and NiCo2OS4– were 6.5 and 14.2 nm, respectively. We believe that despite the smaller surface area, the larger pore diameter in NiCo2OS4– acts as the decisive factor behind its superior electrocatalytic activity toward OER as according to Knudsen’s diffusivity model, larger pores are known to cause less collisions between the fluid and pore walls, leading to higher mean free paths.[30] This assists in greater interpenetration of the electrolyte inside the catalyst material that in turn leads to a higher electrocatalytic activity.

The ECSA, which represents the extent of catalyst surface that is able to take part in electrochemical reactions, is another major parameter behind the electrocatalytic activity of a material. For noble metals, adsorption probes are generally employed to measure the ECSA.[31,32] However, such well-characterized adsorption probes are unavailable for most other materials. In such cases, double-layer capacitance (DLC) measurement, which reports the amount of accumulated charge on the electrode surface and is directly proportional to the ECSA, is a nondestructive alternative for estimating the ECSA.[32,33] To measure the ECSA of both electrocatalysts, cyclic voltammetries (CVs) at different scan rates were recorded in the non-faradic region in 1 M KOH. From the CV curves, the values of anodic (ia) and cathodic (iv) current densities at 0.85 V were measured and plotted with respect to the scan rate (as shown in Figure S3). From these linear plots, the capacitances of NiCo2O4 and NiCo2OS4– were found to be 10 and 1 F/cm2, respectively. Such a large difference in capacitance values despite the presence of the same amount of metal centers on electrodes indicates that the condition (1 M KOH) employed for the DLC measurement, though widely followed, has severe limitations. Similar concerns have been raised by Surendranath et al. in the recent past.[34] They have observed that in aqueous electrolytes, ion-transfer reactions at the electrode interface lead to additional current that convolutes the DLC measurements, resulting in a large variability in capacitance values, especially in the case of metal oxide and chalcogenide surfaces. Further, they suggested that the DLC data collected in polar aprotic electrolytes will be free from such convolution effects. In our case, upon using a polar aprotic solvent (acetonitrile containing 0.15 M KPF6) for DLC measurements, the specific capacitance values for NiCo2O4 and NiCo2OS4– were found to be 1.40 and 2.24 mF/cm2, respectively (as shown in Figures and S4). Thus, the higher ECSA can be considered as a primary factor behind the superior electrocatalytic activity of NiCo2OS4–.

Figure 4

DLC current vs scan rate measured in acetonitrile containing 0.15 M KPF6 for (a) NiCo2O4 and (b) NiCo2OS4–.

Recently, the major research focus on metal chalcogenide and metal phosphide electrocatalysts is toward identifying their actual catalyst centers. It has been already pointed out by researchers that during OER, the high oxidizing potential converts the surface of metal chalcogenides into amorphous metal hydroxide/oxyhydroxide, which in turn acts as the active catalyst.[20,21] Thus, probing the generation of the active catalyst on different materials may impart significant insights into their corresponding electrocatalytic activities. Moreover, recent theoretical calculations suggest that the metal center of an electrocatalyst constantly switches its oxidation state during the electrocatalytic process (as shown in Scheme ). In this regard, CV data can be particularly useful as they are able to provide detailed information related to metal redox processes, active catalyst generation, and their effects on the electrocatalytic performance.

Scheme 1

Schematic of the Change in the Oxidation State (n) of the Metal Center during OER

Thus, time-dependent CV, elemental analysis, and microscopy were carried out to probe the generation of the active catalyst layer and its influence on the OER activity of the two electrocatalysts. For NiCo2OS4–, though no significant redox activity during the anodic sweep was observed in the first cycle recorded during precondition, the cathodic sweep consists of a reduction peak at 1.25 V (as shown in Figure ). Interestingly, an oxidation peak appeared at ∼1.6 V second cycle onward, and its current density increased in the consecutive scans. The oxidation peak corresponds to the oxidation of Ni2+ to Ni3+ and Co3+ to Co4+, whereas the reduction peak can be attributed to their reduction. A large shift (∼110 mV) of the reduction peak was noticed during precondition (consisting of 10 cycles).

Figure 5

Precondition CV data (not iR-compensated) of NiCo2OS4– recorded at 50 mV/s.

For better understanding of this phenomenon, the NiCo2OS4– electrode was subjected to chronoamperometry (at 1.7 V) for 24 h and CV scans were recorded at regular intervals. As shown in Figure S5, the activity of the electrocatalyst increased ∼38 percent in the first 30 min, and CV scan recorded at this interval shows a huge increase for both oxidation and reduction peak current (Figure ). During this interval, while the reduction peak potential has shifted from 1.28 to 1.22 V, no significant shift of the oxidation peak potential was observed. Thereafter, the electrocatalytic performance of NiCo2OS4– remains stable and decreases only 4% in the next 24 h. No major change in the CV data was also noticed except a small shift of reduction peak from 1.22 to 1.19 V. The changes in CV response at 10 mA/cm2 exactly follow the change in chronoamperometry performance.

Figure 6

CV data (not iR-compensated; 10 mV/s) of NiCo2OS4– recorded at different intervals during chronoamperometry.

The gradual emergence of oxidation peak and the correlated changes of chronoamperometry and CV data can be directly related to the generation of the active catalyst (metal hydroxide/oxyhydroxide) layer. Such changes can be better explained by considering the nature of the anionic species in the electrocatalyst. Sulfur being a soft base favors metals in their lower oxidation states and thus in the initial scans, while the presence of a large amount of sulfur in the catalyst impedes metal ion oxidation (as is evident from the small oxidation peak), the reduction process starts from the very first cycle. The ability of sulfur to reduce metal centers is already widely known.[35,36] In consecutive scans, the generation of metal hydroxides makes the metal oxidation process relatively favorable by virtue of the hardness of oxygen, which is reflected in the increase in the anodic peak current. However, the presence of oxygen makes the metal center reduction process more difficult, which gets reflected in the shift of the reduction peak toward lower potential values. Because both electrocatalytic performance and current densities of redox peaks in CV reach their maximum after 30 min, we may conclude that the generation of the active catalyst layer completes in the first 30 min for NiCo2OS4–. No further activation results in the near-constant nature of electrocatalytic performance and CV data. This assumption is further supported by the elemental analysis and microscopic data. Elemental analysis shows that while the ratio of oxygen to sulfur has drastically changed from 8.54:47.11 to 51.65:5.99 after just 30 min of chronoamperometric treatment, data collected after 24 h show a slight increase in oxygen content along with a decrease in sulfur content (Tables S4 and S5). Transmission electron microscopy (TEM) images also reveal that while the fresh catalyst material was crystalline in nature, a marked decrease of crystallinity was observed on the catalyst surface after 30 min of chronoamperometry, indicating the formation of an amorphous metal hydroxide layer (Figure S6).

Similar time-dependent studies were performed with NiCo2O4 electrodes. In this case, the CV data show the appearance of oxidation and reduction peaks at 1.30 and 1.11 V, respectively, from the very beginning (as shown in Figure ).

Figure 7

Precondition CV data (not iR-compensated) of NiCo2O recorded at 50 mV/s.

The electrocatalytic activity of NiCo2O4 continuously increased for the first 5 h, followed by a steady decline (as shown in Figure S5). The CV data recorded at different intervals provide interesting insights (Figure ). While the reduction peak potential remains almost unperturbed with time, the oxidation peak continuously shifts to higher potential values. Moreover, the current density for both oxidation and reduction peaks increases up to 5 h, followed by a gradual decrease.

Figure 8

CV data (not iR-compensated; 10 mV/s) of NiCo2O4 recorded at different intervals during chronoamperometry.

Such changes in current density, once correlated with SEM images, may explain these observations. Once subjected to OER conditions, cracks start to appear on NiCo2O4 flakes, leading to the availability of more number of metal centers for electrochemical reactions (Figure S7a). However, after 5 h, the cracks become so wide that the electrocatalyst material starts to break out of the electrode, resulting in a gradual decrease in electrocatalytic activity (Figure S7b). Interestingly, we have also noticed the appearance of a small amount of black precipitate in the electrochemical cell after subjecting NiCo2O4 electrodes to prolonged OER conditions. It is worth mentioning at this point that the SEM images of NiCo2OS4– recorded after 24 h of usage do not show a similar damage of the electrode material (Figure S7c,d). Further, this explains the saturation of electrochemical performance after initial activation process in the case of NiCo2OS4–. While the number of active sites for NiCo2O4 increases for the first few hours because of cracking of flakes, it remains unaltered in the case of NiCo2OS4–.

These results indicate that the advantages of NiCo2OS4– over NiCo2O4 as the OER electrocatalyst arise from its faster activation as well as stable performance under continuous usage. The faster activation of NiCo2OS4– may be attributed to the difference in the bond energy of metal–sulfur and metal–oxygen bond.[37] As the bond dissociation energy of metal–sulfur is lower compared to that of metal–oxygen, the conversion of metal sulfide into metal hydroxide/oxyhydroxide, which is the active electrocatalyst, is relatively more facile. In addition, the rugged nature of the NiCo2OS4– flake (as is evident from SEM images) provides it the much-needed corrosion resistance (Figure S7c,d). As a result, they show near-constant performance for 24 h. On the contrary, the NiCo2O4 flakes are prone to cracks, which initially provide a larger number of catalytically active sites but ultimately lead to corrosion, resulting in a decrease in activity.

The better activity of NiCo2OS4– compared to that of NiCo2O4 is further supported by Tafel slope values, which are 78 and 43 mV/dec for NiCo2O4 and NiCo2OS4–, respectively (as shown in Figure S8).

To further understand the higher electrocatalytic activity of NiCo2OS4– over NiCo2O4, electrochemical impedance spectroscopy (EIS) of both the samples under OER condition was carried out and the obtained data were fitted in a modified Randle’s circuit (as shown in Figure ). While most parameters such as solution resistance (Rs), DLC (Cdl), and diffusion-related resistance (W) are similar, the charge-transfer resistance (Rct), that is, the resistance that electrons have to overcome for moving between the reactant and the catalyst, of the two materials is significantly different. The Rct for NiCo2OS4– was 23.07 Ω, whereas for NiCo2O4 it was 38.76 Ω. A relatively lower (16.69 Ω) Rct for NiCo2OS4– enabled faster charge transfer between the reactant/intermediates and the catalyst material, ultimately leading to a higher catalytic activity compared to that of NiCo2O4. However, the observed Rct values for both electrode materials were relatively higher than the reported values for similar materials in the literature. We believe that the flakelike [two-dimensional (2D) structure] morphology of both materials may be primarily responsible for such increased Rct because recent reports suggest that the charge-transfer phenomenon is more favorable in one-dimensional (1D) structures compared to their 2D counterparts.[13,38] For example, Qu et al. have suggested that 1D structures allow easier interpenetration of the electrolyte, leading to the lowering of Rct.[39] Further, the presence of fewer grain boundaries in 1D materials is able to provide adequate channels for faster charge transport compared to that in 2D materials.

Figure 9

Nyquist plots of (black) NiCo2OS4– and (red) NiCo2O4 electrodes. The table shows the contributions of different parameters in the EIS study for both samples. The potentials at which each sample shows 10 mA/cm2 activity were chosen for performing EIS.

Recent studies on the electrical conductivity of oxide and sulfide-based materials indicate that sulfide-based materials are generally more conductive compared to their oxide counterparts.[40] This may be attributed to the presence of relatively more labile d electrons in sulfur as compared to the p electrons of oxygen. We believe that the superior conductivity of sulfides may make more surface sites electrically accessible in NiCo2OS4–. This may be one of the reasons behind the better electrocatalytic activity of NiCo2OS4– compared to that of NiCo2O4.

Thereafter, theoretical calculations (DFT) were carried out in order to determine the electronic structure and covalency near the Fermi level of the materials. Recent reports suggest that covalency and the position of d-band center of the metals and p-band center of O/S relative to the Fermi level are the strongest descriptors for the oxygen evolution activity of a catalyst.[41] However, computation of the d-band center of transition metals by DFT often leads to a very high correlation error. Thus, we exploited the delocalized nature of the O/S p-band to obtain a more accurate picture of the electronic structural characteristics of catalyst materials while still reflecting the metal d-character through hybridized density of states. The relative position of these bands determines the hybridization and covalent mixing between the oxygen 2p and metal 3d orbitals. To perform theoretical calculations, some value of x for NiCo2OS4– is required. Because the composition of NiCo2OS4– was found to be Ni/Co/O/S = 1:2.1:0.6:3.29 from EDAX (as shown in Table S2), NiCo2OS3 was taken as the model for theoretical calculations. As shown in Figure , the p-band center of NiCo2OS3 is higher than that of NiCo2O4. Further, computation of d-band centers of the transition metal shows that the d-band center of NiCo2OS3 is higher than that of NiCo2O4. We believe that the uplift of the p-band center after sulfur incorporation enhances the M-O/S hybridization, which in turn increases the catalytic behavior. Because both covalency and position of band centers depend on the properties of constituent elements, the theoretical calculations also indicate that the presence of sulfur plays a crucial role behind the higher electrocatalytic activity of NiCo2OS4– compared to that of NiCo2O4.

Figure 10

Band centers of NiCo2OS3 and NiCo2O4.

Conclusions

In summary, this article compares NiCo-based oxide and oxysulfide materials for OER and details the advantages of using metal oxysulfides over their oxide counterparts. We have synthesized NiCo2O4 and NiCo2OS4– on the FTO substrate by CBD, which is a simple yet highly reproducible and scalable technique for the catalyst layer deposition on the substrate. This further helps in the comparative study as it ensures the presence of an equal number of metal centers on substrates. In comparison to NiCo2O4, the NiCo2OS4– sample shows a much better electrocatalytic activity for OER at pH 14. Moreover, NiCo2OS4– shows 30% increase in electrocatalytic performance within just 30 min of usage and is able to retain this activity for 24 h. In contrast, the electrocatalytic activity of NiCo2O4 decreases about 13% after 24 h usage. The reasons behind the better electrocatalytic performance of NiCo2OS4–, both in terms of activity and stability, has been probed through time-dependent CV, microscopy, and elemental analysis. The results indicate that the higher activity of NiCo2OS4– can be attributed to its higher pore diameter, higher ECSA, lower charge transfer resistance and lower metal–sulfur bond energy that helps in faster activation to form active metal hydroxide/oxyhydroxide species and rugged morphology that prevents corrosion. This work provides significant justification of using sulfur-containing materials as electrochemical precatalysts over their oxide counterparts for OER.

Materials and Methods

Materials

Nickel nitrate hexahydrate [Ni(NO3)3·6H2O], cobalt nitrate hexahydrate [Co(NO3)3·6H2O], urea [CO(NH2)2], thiourea [CS(NH2)2], potassium hydroxide [KOH], ammonium fluoride [NH4F], and acetonitrile were purchased from Merck. Potassium hexafluorophosphate [KPF6] was purchased from Aldrich chemicals. All chemicals were used without any further purification. FTO-coated glass was purchased from Sigma-Aldrich.

Synthesis

Precursor Synthesis by CBD

The precursor for both NiCo2O4 and NiCo2OS4– on FTO was synthesized by the CBD technique. Briefly, 0.5 mM Ni(NO3)3·6H2O (145.4 mg) and 1 mM Co(NO3)3·6H2O (291.04 mg) were dissolved in 25 mL water. Subsequently, a 25 mL aqueous solution containing 2.5 mM urea (150 mg) and 1 mM NH4F (37 mg) was added to it. The reaction mixture was stirred for 30 min and transferred to a reagent glass bottle, which was used as the reaction vessel during synthesis. FTO glasses were cut into 1 cm × 1 cm pieces and washed with soap solution, distilled water, and ethanol prior to usage. Two cleaned FTO glasses were then attached with a glass slide and kept at an angle of 45° with the conductive side facing downward in the same chemical bath. The bottle was tightly capped and kept undisturbed for 48 h at ∼90 °C (as shown in Scheme S1). The temperature of the chemical bath was controlled by an immersed temperature sensor. After completion, the reaction mixture was allowed to cool down, and the FTO glasses were carefully rinsed with copious amount of distilled water. On the conductive side of the FTO glass, formation of a grayish white layer was observed. This was used as the precursor for further synthesis of NiCo2O4 and NiCo2OS4–.

Synthesis of NiCo2O4

The FTO glasses with precursors formed on their conductive side were kept in a tube furnace and were subsequently annealed at 380 °C for 2 h under atmospheric condition. The ramping rate for attaining the desired temperature was set at 3 °C min–1. After the reaction, the grayish white layer turned black, which was later characterized to be NiCo2O4. These material-coated FTO glasses were used as electrodes during electrochemical measurements. Further, a similar treatment was performed on the precipitate that had formed during CBD, and the obtained nanopowder was subsequently used for the surface area analysis.

Synthesis of NiCo2OS4–

The FTO glasses with precursors formed on their conductive side were placed at the center of the annealing chamber of a tube furnace, and about 100 mg of thiourea was placed upstream in the same chamber. Argon was then passed for about 30 min to ensure an inert condition inside the tube. Subsequently, the furnace was heated to 380 °C at a ramping rate of 3 °C min–1. The reaction was carried out for 2 h under uninterrupted Ar flow (as shown in Scheme S2). During the reaction, thiourea decomposed and reacted with the precursor on the FTO glasses (placed downstream wrt thiourea) to form NiCo2OS4– (black in color). These material-coated FTO glasses were used as electrodes during electrochemical measurements. Further, a similar treatment was performed on the precipitate that had formed during CBD, and the obtained nanopowder was subsequently used for the surface area analysis.

Experimental Section

Electrochemical Measurements

All electrochemical measurements were carried out in a three-electrode cell comprising the materials on FTO glass substrates as the working electrode, a Pt wire as the counter electrode, and an Ag/AgCl (3.5 M KCl) as the reference electrode. A 1 (M) KOH solution was used as the electrolyte for OER measurements. The potential scale was calibrated to reversible hydrogen electrode and subsequently 85% iR-corrected (automatic mode) with respect to the Ohmic resistance of the solution, unless specified. Prior to characterization, the samples were preconditioned by CV (1–1.8 V; 10 cycles at 50 mV s–1). CV was then performed (1–1.8 V at 10 mV s–1) in order to evaluate the oxygen evolution activity of the samples. A current density of 10 mA/cm2 was used as the benchmark to compare the activities of different samples. EIS was carried out in the same setup within the frequency range 100 kHz to 100 Hz at voltages corresponding to the 10 mA/cm2 current density for each sample. Chronoamperometry was performed for 24 h in order to evaluate the stability of the catalyst layers, and CV was obtained at different intervals during the chronoamperometric experiment to extract information about the catalyst and its electrocatalytic activity. To measure the ECSA, an organic electrolyte (acetonitrile containing 0.15 M KPF6) solution was used as suggested by Surendranath et al.[34]

Computational Details

The theoretical analysis and predictions in this article were obtained with first principle calculations based on DFT with the ultrasoft method as implemented in the VASP package, which employs a plane-wave basis and pseudopotentials.[42−45] The exchange correlation energy was treated within the generalized gradient approximation, using the functional of Perdew, Burke, and Ernzerhof. Two structures, (a) NiCo2O4 and (b) NiCo2OS3, were taken under consideration to study the effect of sulfur on the electrocatalytic activity of the material. A 56 atom cubic supercell of eight units of inverse spinel (Fd3m) NiCo2O4 was considered, and 24 of the 32 atoms of O were randomly substituted by S to study NiCo2OS3. A cutoff energy of 50 Ry for truncating the kinetic energy and of 500 Ry for the representation of charges in the plane-wave basis set was used. This was adequate for both the structures. The structures were simulated using a (3 × 3 × 3) uniform grid of k point to sample the 2D Brillouin zone.

Characterization Techniques

The PXRD patterns were collected using a Rigaku-SmartLab diffractometer attached with a D/tex ultradetector and Cu Kα source operating at 35 mA and 70 kV. The scan range was set from 20 to 70° 2θ with a step size of 0.02° and a count time of 2 s. The samples coated on FTO substrates were placed on a quartz slide during measurements. Field emission SEM images and EDAX were acquired on a SUPRA 55-VP instrument with patented GEMINI column technology. Prior to loading the samples into the chamber, they were coated with a thin layer of gold–palladium in order to avoid charging effects. Nitrogen adsorption–desorption measurements were conducted at 77 K with a Micromeritics Gemini VII-2390t instrument. The powders were outgassed in vacuum at 150 °C for 2 h prior to measurements. All electrochemical characterizations were carried out in a CHI electrochemical workstation (CHI604D). TEM images were acquired on a JEM 2100F field emission transmission electron microscope operating at 200 kV. ICP-MS was performed in a Thermo Scientific XSERIES 2 ICP-MS instrument.