Optimized Ni, Co, Mn Oxides Anchored on Graphite Plates for Highly Efficient Overall Water Splitting

Abstract

Nickel, cobalt, and manganese oxides are easily obtainable non-noble metal catalysts for water splitting. However, the relationship between composition and catalysts’ performance still needs systematic studies. Herein, guided by theoretical calculations, a low overpotential, easily prepared Mn-doped Co₃O₄ was deposited on graphite plates for water splitting. The 30% Mn-doped Co₃O₄ (Co_2.1Mn_0.9O₄) required the lowest overpotential for oxygen evolution reaction (OER), in which the Co_2.1Mn_0.9O₄ reached 20, 30, and 50 mA cm⁻² in the overpotentials of 425, 451, and 487 mV, respectively, with 90% IR compensation. Under overall water-splitting conditions, the current density reached 30 mA cm⁻² at an overpotential of 0.78 V without IR compensation. Charge density difference analysis illustrates that doped Mn provides electrons for O atoms, and that Mn doping also promotes the electron fluctuation of Co atoms. XPS analysis reveals that Mn-doping increases the chemical valence of the Co atom, and that the doped Mn atom also exhibits higher chemical valence than the Mn of Mn₃O₄, which is advantageous to boost the form of based-OOH* radical, then decrease the overpotential. Considering the particular simplicity of growing the Co_2.1Mn_0.9O₄ on graphite plates, this work is expected to provide a feasible way to develop the high-performance Co-Mn bimetallic oxide for water splitting.

1. Introduction

In contemporary study giving equal importance to energy and the environment, electrolytic water has emerged as a critical avenue for the effective utilization of clean energy resources [1,2]. Noble metal catalysts containing Ru, Rh, Pd, Os, Ir, and Pt exhibit excellent catalyst performance. However, due to resource scarcity and high price, their large-scale applications are limited; therefore, it is of great importance to develop non-noble metal catalysts [3,4]. Numerous materials are nascent non-noble catalysts for water splitting, i.e., high-entropy materials, transition metal phosphides/oxides/sulfides/selenides/tellurides, perovskite composites such as LaSr_3−y, LaSr_2.7Co_1.5Fe_1.5O₁₀, and their composites [5,6,7,8,9,10,11]. Transition metal oxides have been extensively studied in recent years due to their easy preparation and low cost, especially Co, Ni, and Mn oxides [12,13,14,15,16]. For instance, Zhang et al. deposited Co/CoO on N-doped graphene, and as-prepared Co/CoO/N-G achieved 10 mA cm⁻² at an overpotential of 315 mV with 80% IR compensation [17]. Yang et al. deposited graphene sheets on carbon cloth and then deposited NiO on the graphene sheets to obtain the catalytic electrode. The prepared electrode achieved 10 mA cm⁻² at an overpotential of 425 mV (did not clarify whether IR compensation was utilized) [18]. Zhang et al. introduced oxygen vacancies into the surface of MnO₂ nanorods, and the prepared MnO₂ achieved a current density of 5 mA cm⁻² at an overpotential of 410 mV with 90% IR compensation [19]. However, monometallic oxides are still susceptible to corrosion by alkaline electrolytes and insufficient catalyst activity [20,21]. Doping by other elements is a commonly used method to enhance material stability and performance, which has been confirmed in many research fields. For example, in the field of energy storage, Poudel et al. reported that 3.0 at.% Mn-doped WS₂ delivered 107.79 mAh g⁻¹ for an asymmetric supercapacitor at the current density of 1.0 A g⁻¹, while the WS₂ only delivered 37.52 mAh g⁻¹ in the same condition [22]. Furthermore, the 2.0 at.% Gd-doped MnO₂ improved the capacity from 49.5 mAh g⁻¹ to 82.1 mAh g⁻¹, and the improved amplitude reached 65.8% [23]. It is believed that doping by other elements has introduced new active centers and regulated the electronic structure of the original catalysts, thus soundly improving the catalyst activity. However, for a surface-doped catalyst, the catalyst activity will decrease dramatically after the surface layer is corroded.

The bimetallic material is equivalent to a uniform solid solution, in which the activity of the catalyst does not decrease significantly after the surface layer is corroded. Recent studies have suggested that bimetallic compounds/composites with appropriate elemental ratios exhibit similar properties to doped materials. For example, Tang et al. discovered the OER performance of Co_δMn_2−δCH (CH: carbonate hydroxide) following the order of Co₁Mn₁CH > Co_1.5Mn_0.5CH > CoCH > Co_0.5Mn_1.5CH in 1 mol/L KOH electrolyte. This indicates that a small amount of Mn-doping CoCH can enhance the OER performance, but excessive doping will lead to inferior OER performance [24]. Similarly, Nai et al. investigated the OER performance of Ni-Co bimetallic hydroxides and found that the polarization voltage followed the order of Ni(OH)₂ > Ni₂Co(OH)_x > NiCo(OH)_x > Co(OH)₂ > NiCo_8.3(OH)_x > NiCo_5.7(OH)_x > NiCo_2.7(OH)_x. This suggests that a small amount of Ni substituting Co(OH)₂ can reduce the overpotential, and the optimal performance is achieved when approximately 1/4 Co atoms are substituted by Ni. However, substituting more than 50% of the Co atoms will increase the overpotential [25]. Based on the above studies, it is reasonable to assume that the utilization of Ni, Co, and Mn bimetallic oxides could be a promising approach for enhancing water electrolysis performance. However, to the best of our knowledge, a systematic, comprehensive, and in-depth investigation of the Ni, Co, and Mn bimetallic oxides for water splitting is still lacking.

In this work, theoretical calculations were conducted on Co₃O₄, Mn₃O₄, Co_xNi_1−xO, Ni_xMn_1−xO, and Co_1−xMn_xO (x = 0, 0.2, 0.4, 0.6, 0.8) to predict the catalytic activity variation as the elements proportion change. Calculations results predicted that the Ni-doping MnO, Co-doping MnO, Ni-doping Co₃O₄, and Mn-doping Co₃O₄ may exhibit lower overpotential. Usually, catalysts need to deposit on substrates for application. Recent reports indicate that graphite is a practical catalyst substrate due to its high electronic conductivity and low cost, and its low catalytic activity does not obscure the catalyst’s performance [26,27,28,29]. Herein, the graphite plates were applied as the support to investigate the catalytic activity of Co₃O₄, Mn₃O₄, Co_xNi_1−xO, Ni_xMn_1−xO, and Co_1−xMn_xO. The experiment results illustrated that Co_2.1Mn_0.9O₄ is most favorable for water splitting, which was consistent with the theoretical calculation predicted results.

2. Results and Discussion

2.1. DFT Calculation Results

Before conducting the experiments, band structure and density of states (DOS) of the Co₃O₄, Mn₃O₄, Co_xNi_1−xO, Ni_xMn_1−xO, and Co_1−xMn_xO were calculated (the results are shown in Figures S1 and S2, and Table S1). A smaller band gap value indicates better conductivity of the material and lower overpotential when the material is employed as a catalyst for water splitting [30,31,32]. It is worth noting that the band gap values obtained from theoretical calculations are significantly lower than those obtained from experiments, primarily due to the perfect crystal structures used for theoretical calculations, whereas materials tested in experiments contain crystallographic defects, such as impurities, atomic vacancies, and surface contamination. Despite the significant numerical discrepancy between the theoretical calculations and experiments, it does not impede the analysis of the variation of the materials’ properties. According to Figure 1a and Table S1, the band gap values of MnO, CoO, and NiO are 0, 0.157, and 0.236 eV, respectively. This suggests that the electronic conductivity of MnO is higher than that of CoO, and CoO also exhibits higher electronic conductivity than NiO. Furthermore, the band gap of Ni_0.2Mn_0.8O and Co_0.2Mn_0.8O is 0 eV, which proves that 20% Ni or Co substituted MnO maintains its original conductivity. Moreover, the electronic conductivity of Ni_0.8Co_0.2O is lower than that of NiO, which also illustrates that a small amount of Co doping NiO improved the electronic conductivity. However, if the doping amount exceeds a certain limit, the band gap will increase, i.e., Ni_0.4Mn_0.6O and Co_0.4Mn_0.6O show a higher band gap than MnO, and Ni_0.6Co_0.4O shows a higher band gap than NiO.

The d-band center energy calculated from DOS was listed in Table S2 and plotted in Figure 1b. According to the d-band center theory, the closer the d-band center value to the Fermi level, the stronger adsorption of adsorbate on the exposed surface of the adsorbent [33,34]. In the case of water electrolysis, the high adsorption energy resulted in higher overpotential, which originated from the challenge of electronically insulated O₂/H₂ desorption from the adsorbent. For monometallic oxides, the d-band center of MnO, Co₃O₄, NiO, Mn₃O₄, and CoO are 5.0143, 4.9905, 3.7307, 3.2971, 2.0131 eV, respectively, which implies that the catalysts’ activity of those oxides abides the rule of MnO ≈ Co₃O₄ > NiO > Mn₃O₄ > CoO. The d-band center of Mn-doped CoO moves away from the Fermi level with the Mn concentration increasing, suggesting a gradual decrease in the overpotential with the increase of Mn concentration. For Co-doped NiO, the d-band center moves closer to the Fermi level upon increasing Co concentration, indicating an increase in the overpotential with the increase of Co concentration. Similarly, for Ni-doped MnO, the d-band center moves closer to the Fermi level with increasing Ni concentration, indicating an increase in the overpotential with an increase in the Ni concentration. Notice that the d-band center value of MnO and Co₃O₄ is far away from the Fermi level and the difference is negligible. Combining the high electronic conductive property of MnO and Co₃O₄, we can conclude that MnO-based and Co₃O₄-based oxides will exhibit higher catalyst activity than other oxides; in other words, a small amount of Ni-doped MnO, Co-doped MnO, Ni-doped Co₃O₄, and Mn-doped Co₃O₄ will show lower overpotential for water splitting.

2.2. Electrode Preparation and Characterization

Guided by the theoretical calculations, graphite plates were wetted by the solutions containing Co²⁺/Ni²⁺, Co²⁺/Mn²⁺, or Ni²⁺/Mn²⁺, then dried and heated at 300 °C for 2 h to deposit the catalyst on the surface. Figure 2a schematizes the experimental process. To simplify the discussion, the samples were named in the form of M₁M₂-n, where M₁ and M₂ represent the metal cations used in the experiment, and n represents the proportion of M₂ in the M₁/M₂ solution. For example, CoNi-1 represents the electrode obtained using a Co²⁺/Ni²⁺ solution where Co accounts for 90% and Ni accounts for 10%. Similarly, CoNi-2 represents the electrode obtained using a Co²⁺/Ni²⁺ solution where Co accounts for 80% and Ni accounts for 20%, and so on. The solution was dried, then the obtained powder was heated at 300 °C for 2 h to test the XRD, considering that, if the electrodes were tested directly, the catalyst’s peak may be obscured by the diffraction peaks of graphite, resulting in difficult identification.

In the case of CoNi-n (Figure 2b), when the Ni content was between 10% and 40%, the diffraction peaks matched Co₃O₄ (PDF#74-1657), implying that the powder obtained was Ni-doped Co₃O₄. At a Ni content ranging from 50% to 60%, the XRD patterns matched NiO (PDF #78-0423) and Co₃O₄. The actual powder may be a mixture of Co-doped NiO and Ni-doped Co₃O₄. At a Ni content of 70% to 90%, the XRD peaks matched NiO and metallic Ni. The actual yield powder may be the mixture of Ni and Co-doped NiO. For CoMn-n (Figure 2c), when the Mn content was 90%, the XRD peaks matched Mn₃O₄ (PDF#80-0382), Co₃O₄, and MnO (PDF#75-0626); thus, the actual yield powder may be the mixture of Co-doped Mn₃O₄, Mn-doped Co₃O₄, and Co-doped MnO. When the Mn decreased to 50~80%, the XRD peaks matched Mn₃O₄ and Co₃O₄, which implies that the MnO or Co-doped MnO was erased. When the Mn content decreased to 10~40%, the XRD peaks matched Co₃O₄; thus, the yield powder was Mn-doped Co₃O₄. In the case of NiMn-n (Figure 2d), both NiO and metallic Ni were detected when the Ni content was between 70% and 90%; thus, the yield powder should be the mixture of Mn-doped NiO and Ni. When the Ni content decreased to 20~60%, the XRD peaks matched NiO and Mn₃O₄; therefore, the yield powder may be the mixture of Mn-doped NiO and Ni-doped Mn₃O₄. At a Ni content of 10%, the XRD peaks matched Mn₃O₄ and MnO; thus, the yield powder may be a mixture of Ni-doped Mn₃O₄ and Ni-doped MnO.

Notice that Ni/NiO, Co₃O₄, and Mn₃O₄ were obtained by heating Ni(CH₃COO)₂·4H₂O, Co(CH₃COO)₂·4H₂O, and Mn(CH₃COO)₂·4H₂O, respectively (Figure S3). In this case, the MnO cannot be synthesized by the experimental method of this work. According to Figure 2c,d, the single phase of Ni-doped MnO and Co-doped MnO failed to synthesize, while the Ni-doped Co₃O₄ and Mn-doped Co₃O₄ were successfully synthesized. According to Figure 2b,c, to successfully fabricate the Ni-doped Co₃O₄ and Mn-doped Co₃O₄, the concentration of Ni or Mn should be in the range of 0~40%.

Subsequently, electrolysis water tests were performed, and the linear sweep voltammetry (LSV) curves without IR compensation for CoNi-n, CoMn-n, and NiMn-n are shown in Figure 3a, Figure 3b and Figure 3c, respectively. In CoNi-n, CoNi-6 exhibited the lowest overpotential compared to others. In CoMn-n, CoMn-3 showed the lowest overpotential, and in NiMn-n, NiMn-3 showed the lowest overpotential. Due to the smaller difference in polarization voltage of those electrodes, more in-depth experimental research is still needed, especially for CoNi-n and CoMn-n. Based on the XRD analysis presented in Figure 2, CoNi-6 may be the mixture of Co-doped NiO and Ni-doped Co₃O₄, CoMn-3 was Co_2.1Mn_0.9O₄, and NiMn-3 may be the mixture of Mn-doped NiO and Ni-doped Mn₃O₄. To further investigate the catalytic activity of CoNi-6, CoMn-3, and NiMn-3, the LSV of OER with 90% IR compensation was tested and the results are shown in Figure 3d. We can see that CoMn-3 showed the lowest overpotential compared to CoNi-6 and NiMn-3. The LSV curve of CoMn-3 was also compared with Ni/NiO, Co₃O₄, and Mn₃O₄, and CoMn-3 still showed the lowest overpotential (Figure S4). It should be mentioned that the electronegativity of Co, Ni, and Mn are 1.91, 1.88, and 1.55, respectively. The higher the electronegativity, the stronger the ability to pull electrons. Recent literature proves that the HER is a two-step process, and OER is a four-step process, the overpotential of OER is higher than that of HER, and the forming of based-OOH* radical from based-O* radical is the rate-determining step (reaction equation: based-O* + OH⁻ → based-OOH*) [3,6,21]. The higher electronegativity of the reactive site, the easier it is to capture the electron from OH⁻, which makes easier the formation of the based-OOH* radical. It implies that Co₃O₄ will exhibit lower OER overpotential than NiO, and NiO will exhibit lower OER overpotential than Mn₃O₄. This conclusion has also been verified by the experiment (Figure S4). Herein, the Gibbs free energy of the four elementary reactions of the OER on Co₃O₄, Mn₃O₄, and CoMn-3 was calculated, and results are shown in Figures S5–S7. It can be seen that the calculated overpotential of OER on Co₃O₄, Mn₃O₄, Co-site of CoMn-3, and Mn-site of CoMn-3 are 0.753, 0.806, 0.594, and 0.619 V, respectively, which further proves that the OER on Co₃O₄ exhibits lower overpotential than on Mn₃O₄, and that the Co-site of CoMn-3 has more activity than the Mn-site. As for the Ni-doped Co₃O₄ or Mn-doped Co₃O₄, due to their similar molecular formula (Co_3−γM_γO₄, M = Ni, Mn, 0 ≤ γ ≤ 0.40), Co_3−γMn_γO₄ has more empty electron orbitals to accept external electrons; therefore, it is easier to transform based-O* to based-OOH*, and lower overpotential is required.

According to Figure 3d, the current densities of CoNi-6, CoMn-3, and NiMn-3 reached 10 mA cm⁻² at an overpotential of 378, 374, and 388 mV, respectively, and reached 100 mA cm⁻² at an overpotential of 477, 472, and 550 mV, respectively. The polarization voltage difference between CoMn-3 and CoNi-6 is within the measurement error range, so we are not entirely certain that CoMn-3 is better than CoNi-6. After this work, more research still needs to be done to distinguish the catalyst activity of those materials. In this work, we discuss and analyze the catalytic activity of those materials based on the measured data, as the CoMn-n and CoNi-n still need more in-depth experimental research. Figure 3e illustrates the turnover frequency (TOF) curves of CoNi-6, CoMn-3, and NiMn-3; the higher the TOF values, the higher the OER activity of the material. At an overpotential of 300, 350, 400, 450, and 500 mV, the TOF values of CoMn-3 were 0.09896 × 10⁻⁴, 0.20659 × 10⁻⁴, 0.6773 × 10⁻⁴, 2.25077 × 10⁻⁴, and 5.49996 × 10⁻⁴ s⁻¹, respectively, and the corresponding TOF values for CoNi-6 were 0.05932 × 10⁻⁴, 0.13946 × 10⁻⁴, 0.51648 × 10⁻⁴, 1.63669 × 10⁻⁴, and 4.19032 × 10⁻⁴ s⁻¹, respectively. Accordingly, CoMn-3 exhibited superior catalytic activity to CoNi-6, especially at high current densities. In contrast, NiMn-3 exhibited the lowest catalytic activity, demonstrating the inferior OER activity of NiMn-3. To assess the practical application property of the electrodes, current–voltage (i-t) tests without IR compensation were conducted at an initial current density of 10 mA cm⁻². Based on results shown in Figure 3f, CoMn-3 showed better stability than CoNi-6 and NiMn-3 according to the i-t curves. Furthermore, LSV curves without IR compensation were carried out on the electrodes that had undergone a 55 h i-t test and compared with the LSV curve before going through the i-t test. As shown in Figure 3g, after the i-t test, the LSV curves shifted to the higher voltage. The LSV curve of CoMn-3 only slightly shifted, NiMn-3 showed a relatively larger shift amplitude, and CoNi-6 showed the largest shift amplitude. These results demonstrate that CoMn-3 is more stable than CoNi-6 and NiMn-3 for OER.

To further confirm the difference between CoNi-6, CoMn-3, and NiMn-3, the HER performance was evaluated with 90% IR compensation, and the LSV curves are presented in Figure 3h. CoMn-3 manifested a significantly lower polarization than CoNi-6, while NiMn-3 displayed the highest voltage polarization. CoMn-3 achieved a current density of 100 mA cm⁻² with a 437 mV overpotential, whereas CoNi-6 required an overpotential of 471 mV, and NiMn-3 required an overpotential of 529 mV to reach the same current density. After the electrodes underwent a 55 h i-t test of OER, the LSV curve of HER without IR compensation was tested and is shown in Figure 3i. The LSV curve without going through the i-t test is also shown for comparison. It can be seen that CoMn-3 exhibited a high degree of consistency with the initial LSV curve, implying excellent stability of CoMn-3 in alkaline electrolytes, while CoNi-6 and NiMn-3 showed a remarkable voltage shift to higher overpotential.

To reveal the discrepancy between CoNi-n, CoMn-n, and NiMn-n, scanning electron microscopy (SEM) was adopted to reveal the morphology of those electrodes. As shown in Figure S8, CoNi-n were composed of flake-like sheets; the scale of the sheets was 0.1~2 μm. Figure S9 shows that CoMn-1, CoMn-2, CoMn-5, CoMn-6, CoMn-8, and CoMn-9 were composed of flake-like sheets, while CoMn-3 was composed of nanodots, CoMn-4 was composed of nanodots deposed on sheets, and CoMn-7 was composed of flocculent-like nanodots. In the condition of NiMn-n (Figure S10), except for CoMn-4, the others were composed of micrometer fiber. The higher-resolution images of CoNi-6, CoMn-3, and NiMn-3 are shown in Figure 4a–c. CoNi-6 exhibited a flake-like structure of less than 1 μm with a relatively uniform size. CoMn-3 exhibited spherical particles with a diameter of 100~200 nm, and NiMn-3 exhibited fibers with a diameter of 100~500 nm. Ni/NiO, Co₃O₄, and Mn₃O₄ deposed on graphite plates were also investigated the morphology (Figure S11), whereby the Ni/NiO was composed of sheets, Co₃O₄ was nanodots, and Mn₃O₄ was micrometer fiber. The CoMn-3 and Co₃O₄ exhibited similar structure and morphology, and the NiMn-3 and Mn₃O₄ exhibited similar structure and morphology, which proves that the small amount of Mn-doping Co₃O₄ or Ni-doping Mn₃O₄ would not devastate the structure of Co₃O₄ and Mn₃O₄.

The X-ray photoelectron spectroscopy (XPS) survey of CoNi-6, CoMn-3, and NiMn-3 is shown in Figure 4d. Apart from C and O elements, Ni and Co were detected in CoNi-6, Ni and Mn were detected in NiMn-3, and Co and Mn were detected in CoMn-3. The C 1s peak can be fitted to the C-C peak at 284.80 eV and the C-O/C=O peak at 285.1–285.3 eV (Figure 4e) [35]. Likewise, the O 1s peak can be deconvoluted to the O-Co/O-Ni, O-Co/O-Mn, O-Ni/O-Mn peaks at 529.3–529.8 eV and H-O-H bond for the residual water at 531.4–531.7 eV (Figure 4f) [12,36,37]. The Co 2p peak of CoNi-6 can be resolved into two peaks located at 779.85 eV and 795.02 eV for Co-Ni 2p_3/2 and Co-Ni 2p_1/2, respectively, and two other peaks located at 781.16 eV and 796.67 eV for Co²⁺ 2p_3/2 and Co²⁺ 2p_1/2, respectively. Three satellite peaks were observed at 785.40 eV, 789.46 eV, and 803.98 eV (Figure 4g) [38]. In addition, in 773.15 eV, a weak, blurry peak was also observed, which may have originated from the interaction of Co₃O₄ and NiO. Similarly, the Co 2p peak of CoMn-3 can be deconvoluted into two peaks located at 779.96 eV and 794.96 eV for Co-Mn 2p_3/2 and Co-Mn 2p_1/2, and two other peaks located at 781.07 eV and 796.16 eV for Co²⁺ 2p_3/2 and Co²⁺ 2p_1/2, respectively. Three satellite peaks were observed at 786.22 eV, 790.02 eV, and 803.12 eV for CoMn-3 (Figure 4g) [39]. The Ni 2p peak of CoNi-6 can be resolved into peaks of Ni-Co 2p_3/2 at 856.07 eV and Ni-Co 2p_1/2 at 873.26 eV, as well as Ni²⁺ 2p_3/2 at 861.67 eV and Ni²⁺ 2p_1/2 at 880.08 eV. Additionally, three peaks at 854.35 eV, 864.63 eV, and 871.47 eV are satellite peaks (Figure 4h) [38,40]. The Ni 2p peak of NiMn-3 can be deconvoluted into Ni-Mn 2p_3/2 at 855.76 eV and Ni-Mn 2p_1/2 at 872.61 eV, as well as Ni²⁺ 2p_3/2 at 861.34 eV and Ni²⁺ 2p_1/2 at 879.83 eV. The peaks located at 854.44 eV and 866.11 eV are two satellite peaks (Figure 4h) [41]. Finally, the Mn 2p peak of CoMn-3 was analyzed and fitted to the Mn-Co 2p_3/2 peak of Mn²⁺, Mn³⁺, and Mn⁴⁺ located at 641.35 eV, 642.73 eV, 645.26 eV, respectively, and Mn-Co bond of Mn 2p_1/2 peak of Mn²⁺, Mn³⁺, Mn⁴⁺ located at 652.92, 654.00, 655.81 eV, respectively [42]. Similarly, the Mn 2p peak of NiMn-3 was analyzed and fitted to the Mn-Ni bond of Mn 2p_3/2 peak of Mn²⁺, Mn³⁺, Mn⁴⁺ located at 641.24 eV, 642.59 eV, 644.71 eV, respectively, and the Mn-Co bond of Mn 2p_1/2 peak of Mn²⁺, Mn³⁺, Mn⁴⁺ located at 652.73 eV, 653.90 eV, 655.92 eV, respectively [43]. In addition, a weak peak located at 638.16 eV was also observed, which may have originated from the interaction of Mn₃O₄ and NiO. Comparing the Co 2p spectra of CoNi-6 and CoMn-3, it can be inferred that CoNi-6 has more components than CoMn-3. Similarly, NiMn-3 has more components than CoMn-3 by comparing the Mn 2p spectra. Combining previous XRD data, it can be confirmed that CoMn-3 is Co_2.1Mn_0.9O₄.

To further analyze the charge redistribution of the CoMn-3, the charge density difference of CoMn-3 was calculated based on DFT, and the results are shown in Figure S12. The result demonstrates that Mn atoms provided electrons to O atoms, and the Mn-doping promoted some Co atoms to gain electrons and some Co atoms to lose electrons. The chemical valence state of Mn and Co atoms was further investigated by XPS (Figure S13). Comparing the Co 2p spectra of Co₃O₄ and CoMn-3, it can be seen that the binding energy of Co 2p of CoMn-3 shifted to higher energy, proving that the Co site of CoMn-3 lost the electrons and increased the chemical valence, which is an advantage to capture the electron from OH⁻ to boost the based-O* transfer to based-OOH* radical. The Mn 2p spectra of CoMn-3 also showed higher binding energy than that of Mn₃O₄, which is also an advantage to boost the based-O* transfer to based-OOH* radical. Thus, the Mn-doping Co₃O₄ simultaneously improved the catalytic activity of Co and Mn sites compared to monometallic Co₃O₄ and Mn₃O₄.

Due to the best HER/OER performance, the overall water-splitting properties of CoMn-3 without IR compensation were tested in a two-electrode system to evaluate its potential practicality. Figure 5a shows the LSV curve. The current density reached 10 mA cm⁻² at 1.78 V and reached 30 mA cm⁻² at 2.01 V. At 1.80 V, the current density reached 11.19 mA cm⁻², and the graphite plate only contributed the current density of 0.86 mA cm⁻²; thus, the tested data reflect the catalytic activity of CoMn-3. At 2.0 V, the current density reached 29.30 mA cm⁻², and the current density contributed by the graphite plate is still less than 15% (Figure S14). The rate CV curves are shown in Figure S15, accordingly, and the double-layer capacitance was calculated to be 15.86 mF cm⁻² (Figure 5b). Considering that the theoretical value of double-layer capacitance for a perfect, defect-free planar electrode is 0.040 mF cm⁻², the active surface area of CoMn-3 was estimated to be 396 cm² (15.86/0.040) in a 1.0 cm² electrode, which indicates a remarkably high active surface area [44]. Additionally, Figure 5c presents the i-t curve obtained at 1.78 V, revealing that, even after 55 h of testing, CoMn-3 shows acceptable stability. Moreover, Figure 5d depicts the comparison of the LSV curve before and after the i-t test, with the curves almost overlapping, which suggests that CoMn-3 may be a potentially applicable water-splitting catalyst.

3. Materials and Methods

3.1. Theoretical Calculations

The crystal structures of NiO, CoO, MnO, Co₃O₄, and Mn₃O₄ were downloaded from the Materials Project website, and all downloaded crystal structures are the lowest energy structures. The oxides of Co_xNi_1−xO, Ni_xMn_1−xO, and Co_1−xMn_xO, (x = 0, 0.2, 0.4, 0.6, 0.8) were obtained by atomic substitution. By replacing atoms, Co_0.8Ni_0.2O, Co_0.6Ni_0.4O, Co_0.4Ni_0.6O, Co_0.2Ni_0.8O, Ni_0.8Mn_0.2O, Ni_0.6Mn_0.4O, Ni_0.4Mn_0.6O, Ni_0.2Mn_0.8O, Co_0.8Mn_0.2O, Co_0.6Mn_0.4O, Co_0.4Mn_0.6O, and Co_0.2Mn_0.8O were obtained. The academic version of CASTEP was used to calculate the band structure and density of states, and all crystal structures were read and visualized using the VESTA software. The structures were optimized before band structure and density-of-states calculations. The main parameters of theoretical calculations are as follows: GGA-PBE function was used, the energy convergence criterion was set to 1 × 10⁻⁵ eV/atom, the force convergence criterion was set to 0.03 eV/Å, the cutoff energy was set to 550 eV, and the band structure and DOS were calculated using the primitive cell. Spin polarization was considered in the calculation of band structure and density of states, and the LDA + U function was used to correct the strong correlation of Ni, Co, and Mn. The center values of the d-bands for both the up and down spins of each system were calculated using the following formula [45,46]:

$${\epsilon }_{\uparrow ,\downarrow }=\frac{{\displaystyle {\int }_{low}^{high}E\times PDOS(E)dE}}{{\displaystyle {\int }_{low}^{high}PDOS(E)dE}}$$

In the equation, E is the energy of electrons, and projected DOS (PDOS) is the electronic density of states. Considering the degeneracy of the up and down spin, the average of ε_↑ and ε_↓ is taken as the evaluation of the catalyst performance of the materials.

The Gibbs energy of the OER and charge density difference of CoMn-3 were calculated on the VASP code. The electron interaction function, K-point set, energy convergence, and force convergence criterion are the same as the band structure calculation. The oxygen evolution energy was calculated based on the following equation:

* + OH⁻ → OH* + e⁻, ΔG₁ = G(OH*) − G(*) − G(OH⁻)

OH* + OH⁻ → O* + H₂O + e⁻, ΔG₂ = G(O*) + G(H₂O) − G(OH*) − G(OH⁻)

O* + OH⁻ → OOH* + e⁻, ΔG₃ = ΔG(OOH*) − G(O*) − G(OH⁻)

OOH* + OH⁻ → O₂ + * + H₂O + e⁻, ΔG₄ = 4.92 − ΔG₁ − ΔG₂ − ΔG₃

where * represents the based, OH* represents the based OH free radical, and so on. The highest energy is the speed-determining step.

3.2. Preparation and Testing of Electrodes

Ni(CH₃COO)₂·4H₂O, Co(CH₃COO)₂·4H₂O, and Mn(CH₃COO)₂·4H₂O were purchased from Aladdin Chemistry Co., Ltd., Shanghai, China, and used directly without further treatment. A 2-millimeter-thick graphite plate was purchased from Dongguan Yongyao Graphite Material Co., Ltd., Dongguan, China. The experimental process was as follows: Prepare 1.0 mol/L aqueous solutions of Ni(CH₃COO)₂·4H₂O, Co(CH₃COO)₂·4H₂O, and Mn(CH₃COO)₂·4H₂O, then immerse the 1.0 × 2.0 cm² in size of graphite plates into the solution for 30 min. Remove the sucked graphite plates from the solution and dry, and then place them in a tubular furnace. Heat the furnace to 300 °C at a rate of 5 °C/min and hold at 300 °C for 2 h. After the heating is finished and cooled down to room temperature spontaneously, obtain the monometallic oxide growth graphite plates. Next, replace the solution with a mixed solution containing Co²⁺/Ni²⁺ to prepare CoNi-n grown graphite plates, the molar ratio of Co²⁺, Ni²⁺ changing from 0.8/0.2, 0.6/0.4, 0.4/0.6, and 0.2/0.8, while maintaining the total metal ion concentration at 1.0 mol/L. Prepare the CoMn-n grown graphite plates by using the Co²⁺/Mn²⁺ solution and repeat the same protocol, and prepare the NiMn-n grown graphite plates by using the Ni²⁺/Mn²⁺ solution by repeating the same protocol. Then use the oxides-grown graphite plates as the working electrode of the electrolyte water.

The Shimadzu-XRD-7000 XRD (Shimadzu, Japan) tester was used to determine the phase structure and composition of the metal oxides. The structure of the electrodes was observed by Nova 200 NanoSEM (FEI Company, Houston, TX, USA) scanning electron microscopes, and the atomic valence state of the metal oxide on the electrode’s surface was investigated using the Axis Ultra DLD X-ray photoelectron spectrometer (Kratos, Japan). The 1.0 mol/L NaOH aqueous solution was used as the electrolyte, and the EC-100B electrochemical workstation (Mesobiosystems Co., Ltd., Wuhan, China) was used to test the linear voltammetry curve (LSV) without IR compensation, current–time curve, and CV, while the CHI660E electrochemical workstation (Shanghai Chenhua Instrument Co., Ltd., Shanghai, China) was used to test the LSV with 90% IR compensation. For the OER and HER tests, the three-electrode system was adopted, wherein a Pt electrode was used as the auxiliary electrode, a saturated Ag/AgCl electrode was used as the reference electrode, and the scan rate for LSV was 10 mV/s. The working electrode voltage was normalized to the hydrogen standard electrode (RHE) voltage, and the normalization formula was [47]:

E(RHE) = E(Ag/AgCl) + 0.05916 × pH +0.1976 (V)

Herein, the pH value was 14; thus,

E(RHE) = E(Ag/AgCl) + 1.0236 (V)

The overpotential calculation formula was:

η = E(RHE) − 1.23 (V)

For the overall water splitting, only a two-electrode system was adopted, in which the as-prepared electrode was used as both the cathode and anode.

4. Conclusions

To summarize, this study employed theoretical calculations to predict the overpotential of Co_xNi_1−xO, Ni_xMn_1−xO, Co_1−xMn_xO, Co₃O₄, and Mn₃O₄ as catalysts for water splitting, indicating that doped MnO or Co₃O₄ would exhibit lower overpotential. Subsequently, a series of nickel, cobalt, and manganese oxides were deposited on graphite plates and the OER/HER performances were investigated, whereby Co_2.1Mn_0.9O₄ exhibited the highest catalytic activity. In overall water splitting, Co_2.1Mn_0.9O₄ reached 10 mA cm⁻² at 1.78 V and reached 30 mA cm⁻² at 2.01 V. XPS revealed the chemical valence of the Mn site, and the Co site was increased compared to Mn₃O₄ and Co₃O₄, which is an advantage to boost the form of based-OOH* radical. This work once again demonstrates the advantages of theoretical calculations in guiding experimental investigations.