3.1. Synthesis and analysis
Nickel cobalt carbonate hydroxide (NiCoCH) and iron-containing nickel cobalt carbonate hydroxides (Fe
xNi
1–x)CoCH-(y) were synthesized from CoCl
2·6H
2O, NiCl
2·6H
2O, (NH
4)
2Fe(SO
4)
2·6H
2O and urea as precursors in hydrothermal reactions (Step 1 in
Scheme 1). Four different amounts of (NH
4)
2Fe(SO
4)
2·6H
2O were used to synthesize (Fe
xNi
1–x)CoCH-(y) while keeping the NiCl
2·6H
2O and CoCl
2·6H
2O amounts constant. The samples were named (Fe
xNi
1–x)CoCH-(0.025), (Fe
xNi
1–x)CoCH-(0.05), (Fe
xNi
1–x)CoCH-(0.075), and (Fe
xNi
1–x)CoCH-(0.1), representing the use of 0.025, 0.05, 0.075, and 1.0 mmol of the iron precursor. The sulfidation and selenization of the metal carbonate hydroxide precursors was achieved through a hydrothermal sulfidation process with Na
2S·9H
2O and a selenization process with SeO
2 (Step 2 in
Scheme 1). For the sulfoselenide sample, a mixture of Na
2S·9H
2O and SeO
2 was employed in the hydrothermal reaction.
Two different methods were used to determine the chemical formula of the as-prepared samples, a combination of AAS for the metal and CHNS analysis for the sulfur content (method 1) and SEM-EDX (method 2) (
Supplementary Materials, Table S2-S6). Method 1 provides more precise atomic ratios of metal and S content in the samples than EDX. In EDX the emitted X-rays give a 1-2 µm depth analysis but EDX as an X-ray spectroscopy experiences matrix effects and would need standards of similar composition as the sample for peak identification and accurate quantification. For the sulfoselenide and selenide sample, AAS for the metal content was combined with EDX for the Se content. Moreover, the chemical formulae obtained from AAS + CHNS + EDX were much closer to charge balance than the EDX-derived formulae. Based on the metal to sulfur ratios obtained from method 1 and the charge balance calculation of the samples, oxygen should also be incorporated into the structure of the samples. The presence of oxygen in the structure of samples was also proven from the EDX and XPS spectrum. Therefore, the chemical formulae are given here with their estimated oxygen content. The chemical formulas resulting from methods 1 and 2 are provided in
Table S6. The measured SEM-EDX data of the samples are provided in
Figures S1-S8.
The crystallinity of all sulfide samples was low, as evidenced by broad peaks of low intensity in the powder X-ray diffractograms, PXRDs (
Figure 1a). The crystalline phases in the iron-containing nickel cobalt samples were verified as spinels by matching to the known diffractograms of NiCo
2S
4 (ICDD no. 43-1477) and Co
3S
4 (ICDD no. 75-1561) (
Figure 1). The prominent diffraction peaks located at 26.8°, 31.5°, 38.1°, 50.4°, and 55.2° can be attributed to the (220), (311), (400), (511), and (440) planes of the spinel lattice [
17]. In addition, in the PXRD patterns of NiCo
2S
4 and iron-containing nickel cobalt samples, the diffraction peaks at 29.9° and 52.1° can be attributed to the (311) and (440) planes, respectively, of Co
9S
8 (ICDD no. 73-1442). It should be mentioned that the presence of an Co
9S
8 impurity in NiCo
2S
4 is reported in much previous literature, including Chen et al. work [
20], who first reported the formation of sea urchin-like NiCo
2S
4 using metal carbonate hydroxide as a precursor [
30,
37,
45,
46,
47,
48].
The content of the crystalline Co
9S
8 impurity decreases with increasing iron content and is only barely visible in Fe
0.3Ni
1.2Co
2.5(S
0.9O
0.1)
4 and no longer visible in Fe
0.6Ni
1.2Co
2.5(S
0.83O
0.17)
4. The corresponding selenides are of higher crystallinity (
Figure 1b). The prominent reflection peaks of NiCo
2Se
4 and Fe
0.4Ni
0.7Co
1.6(Se
0.81O
0.19)
4 match the simulation for NiCo
2Se
4 (ICDD no. 04-006-5241), where they correspond to the (002), (311), and (-313) crystal plane located at 33.3°, 44.9°, and 51.4°, respectively [
49]. By incorporating both sulfur and selenium in the structure, the crystallinity of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 from the PXRD pattern became too low and no clear crystalline phase analysis was possible anymore. Such low crystallinity was also seen in the literature for nickel sulfoselenide, oxygen-containing cobalt sulfide, and nickel sulfide nanoparticles, which were reported with excellent OER properties [
50,
51,
52].
The morphology of the as-prepared samples was studied using scanning electron microscopy, SEM. As shown in
Figure 2a, Ni
1.0Co
2.1(S
0.9O
0.1)
4 consists of needle-like structures combining parts with a sea urchin-like morphology. By increasing the iron content, the morphology became more sea urchin-like (
Figure 2b-f). In the selenides and the sulfoselenide Fe
0.6Ni
1.2Co
2.5(S
0.83O
0.17)
4 sample,
Figure 2 e-f, agglomerations of needle-like primary particles can be seen. The SEM-energy dispersive X-ray (EDX) mappings (
Supplementary Materials, Figure S9) support the AAS- and CHNS-based elemental analysis for the chemical formulae, and SEM-EDX was the analysis of choice to determine the selenium content.
To further investigate the sulfoselenide Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4, transition electron microscopy, TEM-EDX, was performed. The TEM images,
Figure 3, confirmed the needle-like microstructure. EDX-mapping also proved the uniform presence of Fe, Ni, Co, S, and Se in the sulfoselenide sample (see
Supplementary Materials, Table S1 for atom ratios).
Nitrogen adsorption-desorption isotherms, and specific surface area (BET) of the samples are given in
Figure S10 and Table S7.
X-ray photoelectron spectroscopy (XPS) was conducted to determine the valence state of the elements in Ni
1.0Co
2.1(S
0.9O
0.1)
4 and Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4. As shown in
Figure S11, X-ray survey spectra indicate the existence of Ni, Co, and S in Ni
1.0Co
2.1(S
0.9O
0.1)
4 and of Fe, Ni, Co, S, and Se in Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4. The high-resolution spectra of the metal atoms and of Se are shown in
Figure 4, those of S 2p and O1s of Ni
1.0Co
2.1(S
0.9O
0.1)
4 in
Figure S12a and Figure S13a, respectively. The positions of the XPS peaks are summarized in
Table S8 and S9. It should be noted that the analysis depth of XPS is only 0.7-11 nm as the detected photoelectrons can only escape from such a thin surface layer of the sample.
The metal ratios and the ratios between different oxidation states of nickel and cobalt (
Table 1) were calculated by integrating the fitted peak area for each metal valence state using the Ni 2p
3/2 and Co 2p
3/2 regions. In Ni
1.0Co
2.1(S
0.9O
0.1)
4 Ni
2+ and Co
3+ are the prevalent oxidation states.
Based on element ratios obtained from AAS and CHNS analysis and also the metal valence states for nickel and cobalt in the Ni1.0Co2.1(S0.9O0.1)4 sample, the chemical formula can be given as ((Ni2+)0.72(Ni3+)0.28)1.0((Co2+)0.27(Co3+)0.72)2.1(S0.9O0.1)4 which is anion-cation charge-balanced within rounding errors.
In contrast to the Ni
1.0Co
2.1(S
0.9O
0.1)
4 sample, Ni
3+ and Co
2+ are the dominant valence states in Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4, which might be one of the reasons for having an improved the OER performance of this catalyst. It is believed that Ni
3+ can improve electrophilicity and oxygen adsorption, which can increase the amount of NiOOH active sites during the OER reaction [
53].
The Fe 2p spectrum (
Figure 4c) for Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 represents two broad peaks at 712.5 eV and 725.0 eV, which can be assigned to Fe 2p
3/2 and Fe 2p
1/2, respectively [
54]. Furthermore, two satellite peaks were detected at 717.5 and 734.5 eV [
55,
56]. The 2p
3/2 spectrum range is 710 to 720 eV including the satellite peak, while the 2p
1/2 spectrum range is 720−735 eV with the satellite peak. For iron it should be noted that the Fe 2p spectral background has a contribution from a Co
LMM and Ni
LMM Auger peaks, making an unequivocal deconvolution and peak assignment difficult [
57,
58]. The Se 3d XPS spectra of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 (
Figure 4d) consist of two peaks centered at 55.0 and 57.0 eV, representing Se 3d
5/2, and Se 3d
3/2, respectively. The peak at 59.5 eV was attributed to SeO
X, forming on the surface due to exposure to air, and to the overlap with the Co 3p signal [
39]. The high-resolution spectrum of S 2p and O 1s of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 are shown in
Figure S12b and Figure 13b, respectively. Based on the elemental ratios obtained from AAS, CHNS and EDX(Se) analysis and also metal valence states for iron, nickel and cobalt in the Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 sample the chemical formula can be given as (Fe
2+)
0.5((Ni
2+)
0.1(Ni
3+)
0.9)
1.0((Co
2+)
0.87(Co
3+)
0.13)
2.0(S
0.57Se
0.25O
0.18)
4.
3.3. Oxygen evolution reaction performance
The OER activity of the mixed-metal sulfides, selenides, and the sulfoselenide was evaluated by a three-electrode setup in 1.0 mol L
–1 KOH solution. The working electrode was prepared according to a method used by Li et al.[
29], a slurry containing 8:1:1 mass portions of the active materials (mixed-metal sulfide, selenide, or sulfoselenide material), carbon black, Vulcan XC-72R, and polyvinylidene fluoride, PVDF, respectively, in
N-methyl-2-pyrrolidone, NMP (further details in experimental section). As is shown in
Figure 6a, polarization curves of the as-prepared electrodes revealed that the presence of iron could reduce the overpotential of the Ni
1.0Co
2.1(S
0.9O
0.1)
4 parent compound. In the polarization curves, the peaks around 1.3–1.4 eV are attributed to the oxidation of Ni
2+/Ni
3+ [
59].
The overpotential for Ni
1.0Co
2.1(S
0.9O
0.1)
4 of 346 mV at 50 mA cm
–2 was reduced with increasing iron content in Fe
0.1Ni
1.4Co
2.9(S
0.87O
0.13)
4, Fe
0.2Ni
1.5Co
2.8(S
0.9O
0.1)
4, Fe
0.3Ni
1.2Co
2.5(S
0.9O
0.1)
4, and Fe
0.6Ni
1.2Co
2.5(S
0.83O
0.17)
4 to 318, 310, 294, and 294 mV, respectively (
Figure 5a,b). The electronic interaction between Fe, Ni, and Co in the iron-containing samples alters the electronic structure, making Ni
2+ oxidation more difficult, resulting in a positive shift in the Ni
2+/Ni
3+ anodic peak at 1.3–1.4 eV [
30,
37]. The reduction of the OER overpotential in iron-containing samples can be attributed to reducing the charge transfer resistance through the synergistic electronic interaction between Fe, Co and Ni from a charge redistribution between active sites within the samples. Density functional theory (DFT) calculations in the literature traced the synergy to a decrease in the Gibbs free energy for the formation of a MOOH intermediate, which not only enhanced the intrinsic OER activity, but also significantly improved the intrinsic conductivity of iron-containing samples, greatly facilitating the charge transfer process [
41,
60].
In the next step, the effect of the coexistence of sulfur and selenium was investigated. The overpotential of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 (green line)of 277 mV at 50 mA cm
–2 presents a 17 mV and 22 mV reduction compared to Fe
0.6Ni
1.2Co
2.5(S
0.83O
0.17)
4 (294 mV) and RuO
2 (299 mV) (
Figure 5c,d), indicating that the insertion of selenium improves the performance of the transition metal sulfide. Moreover, the OER performance of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 at 100 mA cm
–2 only needed 299 mV while RuO
2 required 341 mV overpotential, which demonstrates the excellent electrocatalytic activity of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 even under a high current density. The boost of the OER activity of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 can be attributed to the effect of selenide incorporation that reduces the energy barrier of the OER reaction, optimizes the electronic structure of active sites via modulating of the d-band of the compound, and also accelerates the kinetics of the reaction [
61].
The Tafel slopes of the samples were obtained from LSV polarization curves at a scan rate of 5 mV s
–1 using the Tafel equation (1) [
62]:
η is the overpotential, b is the Tafel slope, j is the current density, and c is the intercept with the y-axis. The value of the Tafel slope is one of the most useful kinetic parameters and is inversely proportional to the kinetics of the OER reaction. Hence, as demonstrated in
Figure 6a,b, Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 with the lowest value of the Tafel slope presents the most favorable OER kinetics among the investigated samples. Besides, according to Krasil'shchikov's mechanistic paths (equation 2–5, M = active site, b = Tafel slope) [
62,
63] with their corresponding Tafel slope values, the rate-determining step of the OER reaction for Ni
1.0Co
2.1(S
0.9O
0.1)
4 (b = 132 mV dec
–1) is metal oxidation with hydroxide formation (reaction (2)). By increasing the iron content in the samples, the Tafel slope decreased and reached 87 mV dec
–1 for Fe
0.6Ni
1.2Co
2.5(S
0.83O
0.17)
4, suggesting that metal oxidation with hydroxide formation (2) and deprotonation of a metal hydroxide (3) both present rate-determining steps.
The lowest Tafel slope among the samples belongs to the sulfoselenide Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 with 75 mV dec
–1, which is again evidence for the role of selenium in enhancing the kinetics of the OER reaction. The Tafel value of 75 mV dec
–1 of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 is close to reaction (3) (b = 60 mV dec
–1) being then rate-determining in the overall OER process.
Electrochemical impedance spectroscopy (EIS) was conducted for the electron-transfer kinetics during the OER reaction and to justify the obtained overpotentials [
64].
Figure 6c shows Nyquist plots of the samples at the potential of 1.5 V vs. RHE. The semicircle diameter of Nyquist plots is inversely proportional to the charge transfer rate across the electrode and the electrolyte that accelerates reaction kinetics. Hence, a smaller semicircle diameter represents more favorable charge transfer kinetics [
24].
The smaller semicircle diameter of the Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 Nyquist plot indicates that the coexistence of iron and selenium in Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 can reduce the charge transfer resistance (Rct). Furthermore, a Voigt circuit model was applied to the Nyquist plots to evaluate the specific value for the charge resistance in the OER process. As shown in
Table 2, the value of charge resistance for Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 is the lowest (0.8 Ω) compared to the other investigated samples. The results from the Rct values are in line with the recorded OER overpotentials of the samples and the electrocatalytic performance.
To elucidate the importance of sulfur and selenium in the OER electrocatalysis performance, the (Fe
xNi
1–x)CoCH-(1.0) precursor for Fe
0.6Ni
1.2Co
2.5(S
0.83O
0.17)
4 and Fe
0.5Ni
1.0Co
2.0(S
0.57 Se
0.25O
0.18)
4 was investigated and found to have an OER LSV polarization curve with a higher overpotential of 330 mV vs. RHE to reach 50 mA cm
–2 and a Tafel slope of 98 mV dec
–1, higher than the iron-containing sulfide and sulfoselenide (
Table 2,
Figure S14). Furthermore, the larger Nyquist plot semicircle diameter and higher charge transfer resistance (R
ct = 2.5 Ω) of (Fe
xNi
1–x)CoCH-(1.0) imply the importance of S and Se in facilitating the charge transfer through the electrode-electrolyte interface in sulfide, selenides, and sulfoselenide samples (
Figure S15).
One of the critical parameters to evaluate the performance of electrocatalysts in practical applications is their long-term stability. Hence, a chronopotentiometry test at a current density of 50 mA cm
–2 for 20 h was conducted to evaluate the long-term stability performance of the Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 [
47].
Figure 6d illustrates that after 20 h of OER reaction, the overpotential for obtaining 50 mA cm
–2 increased only from 277 mV to 279 mV, which is essentially constant and supports the excellent electrocatalyst stability of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 for practical applications. The more so as for RuO
2 the overpotential increases from 300 mV to 375 mV at the same chronopotentiometry test for 20 h, at 50 mA cm
–2. A recent study by Jiang et al. [
65] proved that a SeO
x film on the surface could improve the catalyst stability in the OER reaction, which can be the reason behind the high stability of Fe
0.5Ni
1.0Co
2.0(S
0.57 Se
0.25O
0.18)
4 during the OER reaction (the overpotential increased from 277 to only 279 at 50 mA cm
–1 after 20 h OER reaction).
Figure 6.
(a) Tafel plots and (b) bar chart of Tafel slopes of the samples. (c) Nyquist plots of selected samples and Voigt circuit model. (d) Chronopotentiometry test of Fe0.5Ni1.0Co2.0(S0.57Se0.25O0.18)4 and RuO2.
Figure 6.
(a) Tafel plots and (b) bar chart of Tafel slopes of the samples. (c) Nyquist plots of selected samples and Voigt circuit model. (d) Chronopotentiometry test of Fe0.5Ni1.0Co2.0(S0.57Se0.25O0.18)4 and RuO2.
The superior OER activity and stability of Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 material can be attributed to several effects resulting from introducing Fe and Se
2– in the structure of the nickel cobalt sulfide base material: Introducing Fe sites can enhance OER performance a) by optimizing bond energies for OER intermediates adsorbing on the electrode surface, which facilitates the OER kinetics [
66], b) by overcoming the metal oxidation step and facilitating O
2 evolution [
67], c) by improving the conductivity of the electrode film [
68].
For comparison, the overpotential values of several high performance electrocatalysts at a current density of 50 mA cm
–2 using nickel foam as substrate are presented in
Table 3. Notably, the results demonstrate that the OER performance of the sulfoselenide Fe
0.5Ni
1.0Co
2.0(S
0.57Se
0.25O
0.18)
4 is on par with the best electrocatalysts reported in the literature, highlighting its comparable effectiveness.