1. Introduction
Electrochemical water splitting is a promising method for clean and sustainable hydrogen production [
1,
2]. However, a bottleneck in water splitting applications is its high energy consumption, including the energy loss required to drive the sluggish oxygen evolution reaction (OER) process [
3,
4]. Efficient catalysts, such as Ru [
5,
6], Fe [
7], and Ni-based [
8,
9] materials, have been exploited to alter reaction pathways and consequently, lower energy barriers [[
10], [
11], [
12], [
13], [
14]].
Metallic Ni is a promising OER catalyst with good catalytic activity, electrical conductivity, abundance, low cost, and remarkable stability. Nevertheless, metallic Ni is a precatalyst that converts to more active oxides/hydroxides during the OER [
15,
16]. Therefore, considerable attention has focused on direct Ni oxidation. For example, pristine Ni foam (NF) is soaked in HCl solution to expose surface Ni atoms and oxidize them into NiO in air [
17]. Oxidants, such as S
2O
82− [
18], oxygen radicals [
19], and Fe
3+ [
20], have also been considered to activate the NF surface, where active Ni oxides, hydroxides, and oxyhydroxides sediments are formed as sheet-, wire-, or dendrite-like films.
Regardless of the oxidation treatment, oxidized Ni species undergo surface reconstruction by charging/discharging, as illustrated by the Bode scheme [
21] (e.g., α-Ni(OH)
2, β-Ni(OH)
2, γ-NiOOH, and β-NiOOH), depending on the applied potential and electrolyte pH. Oxidation-activated NiOOH, commonly presented as γ-NiOOH and β-NiOOH with different OER performances, is considered the true active species for OER [
22]. γ-NiOOH contains high-valence Ni
4+ and offers higher intrinsic OER activity than β-NiOOH with only Ni
3+ [
23]. The synthesis approach determines the reconstructed NiOOH structure. Zhong et al. [
24] evaluated the conversion of NiOOH from three Ni-based precatalysts using chronopotentiometry. Despite their chemical similarity, various extents of distortion are observed in the NiO
6 octahedra of these NiOOH samples owing to the strains, resulting in distinct OER performance. Therefore, modulating oxidized Ni species that can be further reconstructed into OER-favorable NiOOH species remains difficult.
In this study, electrochemically exfoliated graphene (EG) was used to oxidize and modulate the surface composition of NF via an interfacial redox strategy. The EG-activated NF surface exhibited desirable Ni2+ species that can be converted to the ideal γ-NiOOH species, instead of β-NiOOH species. Moreover, single Ni atoms and clusters were simultaneously anchored to EG after the redox process, acting as active sites for enhanced OER performance and inhibiting the continuous oxidation of the underlying metallic Ni. Therefore, the resultant NF can promote the OER performance, requiring an overpotential of 290 mV to reach 10 mA⋅cm−2. This interfacial redox modulation strategy is versatile for oxidation-degree-controllable EG- and Ni-based metals, leaving sufficient space for an increase in the OER performance. As a proof-of-concept, the optimal NiFe system decreased the overpotential by 10 mA⋅cm−2 at 243 mV and exhibited remarkable stability at 500 mA⋅cm−2 for 100 h.
2. Results and discussion
The synthesis procedure of the EG-modulated NF is illustrated in
Fig. 1(a). Graphene was exfoliated from a graphite foil precursor in (NH
4)
2SO
4 solution via a facile electrochemical method. The interfacial redox reaction started when the EG suspension was drop-cast onto the NF. As NF has a lower reduction potential, electrons transferred from interfacial Ni to EG [
25]. Ionized Ni (Ni
2+) and negatively charged EG approached each other owing to the electrostatic attraction. Thus, the EG layers were tightly attached to the NF surface and simultaneously reduced by the electrons released from the NF. After 24 h of drying, the color of the EG-oxidized NF (denoted as EG-NF) darkened (Fig. S3 in Appendix A). Subsequently, it was collected and used as the working electrode for the OER catalysis.
The EG layers of EG-NF were observed via scanning electron microscopy (SEM). The NF surface is initially smooth (
Fig. 1(b)). The decoration of the surface with curved and interconnected EG layers increased its roughness (
Figs. 1(c) and
(d)). The X-ray diffraction (XRD) patterns of EG-NF exhibit clear peaks ascribed to NF (
Fig. 1(e)). Meanwhile, in the XRD pattern of EG, a broad peak is observed at 26°, which differs from the known characteristic peak (about 10°) of graphene. This is ascribed to the partial restacking of the EG layers into thin films [
26] or removal of the intercalated oxygen-containing groups on the EG [
27]. In a typical Raman spectrum of a carbon material, D and G bands arise from the breathing mode of the sp
2 atoms in the rings and bond stretching of the graphite lattice, respectively [
28]. A high intensity ratio of the D and G bands (
ID/
IG) implies the existence of disordered structures and defects on graphene [
29]. Based on the Raman spectrum shown in
Fig. 1(f), the
ID/
IG ratio of EG is 1.05, denoting the enriched structural disorders and defects in EG owing to the small lateral size and heterodoping (e.g., O, S, and N, as shown in Fig. S4(a) in Appendix A) on the basal and edge planes [
30,
31]. After reduction by NF, the
ID/
IG of EG-NF increased to 1.09 because of the formation of isolated sp
2 domains and topological defects [
32,
33].
X-ray photoelectron spectroscopy (XPS) was used to analyze the presence of Ni
2+ species. In the Ni 2p
3/2 spectra (
Fig. 1(g)), the intensity of the metallic Ni peak in EG-NF sharply decreased from 29.05% to 3.92%. The predominant peak at 855.7 eV is attributed to the Ni
2+ species, such as Ni(OH)
2. In contrast, in the Ni 2p
3/2 spectrum of NF, a strong feature peak of metallic Ni and two weaker peaks at 853.7 and 855.8 eV, corresponding to Ni
2+, such as NiO and Ni(OH)
2 [
34], are noted after deconvolution. In addition to the changes in the Ni 2p
3/2 spectrum, the intensity of the O peak in the XPS survey spectrum of EG-NF (Fig. S4(a)) is higher than that of EG. The increased amount of O is mainly attributed to the generation of oxidized Ni layers owing to the EG. This contribution was quantified using the lattice O peak shown in Fig. S4(b) in Appendix A. In the O 1s XPS spectrum of EG-NF, a new peak at 530.3 eV is noted after deconvolution owing to lattice O, which mostly originates from the oxidized Ni
2+ species, and the O content is approximately 20.9%. In the C 1s spectrum of EG-NF, the peak intensities of the oxygen-containing functional groups are lower than those of EG (Fig. S4(c) in Appendix A). These observations conform with the prediction of the interfacial redox interaction between NF and EG, where the activation of Ni
2+ species on NF and EG reduction simultaneously occurred to form stacked layers on the NF.
Ni
2+ species undergo oxidation before the OER, and the reconstructed Ni species (i.e., Ni
3+ and Ni
4+) are active sites for OER catalysis [[
35], [
36], [
37]]. Differential pulse voltammetry (DPV) was performed to detect the reconstructed Ni species in EG-NF and NF during the OER [
38]. The DPV curves of both EG-NF and NF (
Fig. 2(a)) show a reduction peak ascribed to the Ni
3+/2+ redox pair at 1.31 V. However, the DPV curve of EG-NF has an additional Ni
4+/3+ reduction peak at approximately 1.38 V, suggesting that the actual OER-active species in EG-NF also contained Ni
4+. The distinctive Ni
3+ and Ni
4+ species resulted in various OER performances.
In the
iR-corrected linear scanning voltammetry (LSV) curves (where
i stands for current flow and
R stands for resistance), EG-NF required lower potentials to reach the same current densities as NF (
Fig. 2(b)). The overpotential to reach 10 mA⋅cm
−2 is 290 mV, which is lower than that of NF (370 mV). The improved OER performance of EG-NF contributes to the oxidation ability of EG, thereby inducing the formation of more active Ni
4+ species during OER [
39] and increasing the amount of oxidized Ni
2+ species. The latter is confirmed by the comparison of the electrochemical surface area (ECSA) in terms of the double-layer capacitance (Fig. S5(a) in Appendix A). As shown in Fig. S5(b) in Appendix A, the line slope is proportional to the ECSA. The ECSA of EG-NF is 10 times higher than that of NF.
To eliminate the current contribution from the number of active sites and further evaluate the intrinsic activity, the LSV curves were normalized to the electrochemically accessible Ni atoms derived from the oxidation peak in the cyclic voltammetry (CV) curves (details are given in Section S1.5 in Appendix A). A similar tendency is observed for the OER performance. The superb OER performance of EG-NF is also illustrated by the turnover frequency (TOF) at 400 mV. EG-NF presents a TOF of 0.86 s
−1 which is 6.6 times higher than that of NF. Moreover, EG-NF has a lower Tafel slope of 54 mV per decade (
Fig. 2(c)), indicating the enhanced OER kinetics of NF after EG oxidation. This finding is also related to the accelerated charge-transfer ability of the NF enabled by EG, as highlighted in the Nyquist plots (Fig. S6 in Appendix A).
In situ electrochemical impedance spectroscopy (EIS) and Raman spectroscopy were used to elucidate the decisive factors of the improved OER performance of NF after EG oxidation. As shown in the Bode phase plots of EG-NF (
Fig. 2(d)), the response in the middle-frequency region (10
2-10
3 Hz) is related to the surface double-layer capacitance, whereas the response in the low-frequency region (below 10 Hz) is associated with the nonhomogeneous charge distribution caused by the surface-reconstructed Ni species (e.g., Ni
3+ and Ni
4+). With increasing potential, the phase angle at 0.1 Hz sharply decreased at 1.40 V and subsequently increased because of the NiOOH formation. However, the potential for this phase angle change in NF is lower (Fig. S7 in Appendix A), indicating the difficulty of EG-NF to generate NiOOH.
In situ Raman spectroscopy was used to reveal the structural changes in NF and EG-NF with the potential (
Figs. 2(e) and
(f)). As the oxidation potential of the Ni
2+ species is reached, two bands, related to the bending δ(Ni
III-O) modes and stretching ν(Ni
III-O) vibrations in NiOOH, are noted. The intensity ratio of the δ(Ni
III-O) band to ν(Ni
III-O) band (
Fig. 2(f)) is higher, which is typical for γ-NiOOH. In contrast, β-NiOOH is obtained by NF and is maintained at even higher potentials. The statistic Ni valence in γ-NiOOH is +3.6. Moreover, highly oxidized Ni
4+ species are noted in γ-NiOOH, whereas Ni
3+ dominated β-NiOOH. This explains the higher potential required for the EG-NF to form NiOOH in the Bode plots and agrees with the DPV result, in which a Ni
4+/3+ reduction peak is observed. Along with the Bode scheme (i.e., α-Ni(OH)
2 is transferred to γ-NiOOH by charge), we infer that the oxidized Ni
2+ species in EG-NF contain α-Ni(OH)
2 after the interfacial redox interaction between EG and NF.
The difference in the OER activity between EG-NF and NF is indicated by the altered reconstructed NiOOH species. Theoretical calculations were performed to understand the activity distinction between β-NiOOH and γ-NiOOH, which are true active phases in the OER catalysis. The density-of-states calculation results (Fig. S8(a) in Appendix A) denote the metallic electronic structure of γ-NiOOH, which ensures its rapid charge-transfer ability during OER catalysis. A conventional OER mechanism of γ-NiOOH is presented in
Fig. 2(g), where three intermediate adsorbates (*OH, *O, and *OOH) are involved (the asterisk represents the active site). The specific reaction energies and overpotentials calculated by density functional theory were then used to estimate the catalytic activity. As shown in the reaction step diagram (
Fig. 2(h)), the γ-NiOOH(001) surface exhibits lower elementary reaction energies than those of the β-phase. The rate-determining steps of both NiOOH species occurred during the formation of the *OOH intermediate. The reaction energies of β-NiOOH and γ-NiOOH are 1.75 and 1.07 eV, respectively (0 V). Therefore, the overpotential (0.4 V) of γ-NiOOH (0.66 eV) is lower than that of β-NiOOH (1.35 eV). The charge density differences (Fig. S8(b) in Appendix A) of the OER intermediates (*OH, *O, and *OOH) with γ-NiOOH further confirmed the good adsorption properties of the γ-NiOOH surfaces, suggesting that γ-NiOOH is more favorable for OER catalysis under alkaline conditions.
EG reduction by NF contributed to the enhanced OER performance of the active Ni species, except for the altered NiOOH. During the redox interaction between EG and NF, Ni species were deposited on the graphene layers. Ni species were noted on EG, as confirmed by high-angle annular dark-field scanning transmission electron microscopy (STEM) [
40,
41] and the results are shown in
Fig. 3(a). This suggests that Ni atoms are partially trapped by the defects [
6] on EG (e.g., oxygen defects) during the redox process at the interface between NF and EG. Further energy dispersive spectroscopy (EDS) mapping (
Fig. 3(b)) indicates the presence of Ni on the peeled graphene layers after oxidation, denoting the wide distribution of Ni species throughout the surface of the peeled graphene layers. In addition, the graphene layers peeled from EG-NF by sonication featured a pair of redox peaks of Ni and OER activity in the CV curve (
Fig. 3(c)). The characteristic peaks of Ni are also observed in the Ni 2p spectrum (Fig. S9 in Appendix A). Both results confirm the existence of Ni species on the EG after oxidizing the NF. These Ni single atoms and clusters on the graphene layers are also active sites for OER catalysis, as supported by the decreased OER performance of EG-NF after sonication (Fig. S10 in Appendix A).
In the long-term stability test of EG-NF at 10 mA⋅cm
−2, the potential slightly increased in the initial period (Fig. S11 in Appendix A) because of the detachment of EG, which was peeled off by the oxygen bubbles at the interfaces. This results in the loss of active sites from the detached graphene layers, thereby degrading the OER activity of EG-NF after the stability test (
Fig. 3(d)). However, the potential of NF continuously increased and fluctuated during the stability test owing to the gradual oxidation of the internal metallic Ni. Consequently, this thickened the surface oxidized layer of NF. Therefore, the oxygen content of NF is increased (Fig. S12 in Appendix A), and the atomic ratio of Ni to O decreased from 0.37 to 0.12.
The Ni 2p spectrum of NF after the stability test (
Fig. 3(e)) clearly shows a diminished metallic Ni peak and a dominant peak of Ni
2+ species. However, in the XRD pattern of NF (Fig. S13 in Appendix A), no new characteristic peaks are observed, which can be ascribed to the low crystallinity or amorphous features of these Ni
2+ species [
42]. In the Ni 2p spectrum, EG-NF exhibits an inverse pattern, in which the metallic peak is well preserved after the stability test (
Fig. 3(e)). The inhibited surface oxidation can be related to the single Ni atoms and clusters on EG, where the adsorption of OH
− is kinetically preferential than that of γ-NiOOH, as shown in
Fig. 3(f). Consequently, the oxidation of the underlying Ni was significantly reduced. Moreover, EG layers were still closely attached to the NF after the long-term stability test (Fig. S14 in Appendix A), suggesting the strong interaction between EG and NF.
The interfacial redox process of NF with the oxidant EG stimulated the active Ni species on the NF and EG surfaces, thereby accelerating the OER kinetics and stabilizing metallic Ni during the long-term operation of NF. The versatility of this interfacial redox modulation strategy was explored by initially replacing EG with EG(Na+), which is another type of graphene prepared in a Na2SO4 electrolyte using a similar procedure. In the XRD patterns (Fig. S15(a) in Appendix A), EG(Na+)-NF and EG(Na+) exhibit the same features as EG-NF and EG, respectively. After the interfacial redox interaction with NF, EG(Na+) covered the NF, as shown in the SEM images (Fig. S15(b) in Appendix A). However, based on the XPS survey spectrum (Fig. S15(c) in Appendix A) and Raman spectrum (Fig. S15(d) in Appendix A), EG(Na+) has inferior oxidation ability compared to EG. In particular, its C-to-O ratio (10.07) is higher, whereas its ID/IG ratio (0.85) is lower owing to fewer oxygen defects. Therefore, the potential difference between EG(Na+) and NF was lower than that between EG and NF (Table S2 in Appendix A), resulting in a less intensive interfacial redox interaction between NF and EG(Na+). Consequently, the OER performance of EG(Na+)-NF is inferior to that of EG-NF (Fig. S15(e) in Appendix A).
In addition to the controllable modulation of graphene, other Ni-based metal foams were used to demonstrate the potential of this interfacial redox modulation strategy for enhancing the OER performance. The morphology of the NiFe foams after the interfacial redox interaction with EG (Fig. S16 in Appendix A) is similar to that of EG-NF (
Fig. 1(c)). Moreover, the NiFe foams exhibit a significantly enhanced OER performance compared to the other samples (
Fig. 4(a)). The optimal EG-NiFe(7:3) requires a potential of 243 mV to reach 10 mA⋅cm
−2 and its Tafel slope is as low as 47 mV⋅dec
-1, as shown in
Fig. 4(b). Table S3 in Appendix A presents a comparison of the OER performance of Ni-based catalysts, suggesting the attractive catalytic activity of the metals after oxidation by EG. In particular, EG-NiFe(7:3) demonstrated impressive stability even at 500 mA⋅cm
−2 for 100 h, regardless of the fluctuating potential due to the bubble accumulation or electrolyte consumption and refilling (
Fig. 4(c)). This corroborates the practical application of EG-NiFe(7:3) in industrial alkaline water splitting. EG(Na
+) and NiFe foams were used to illustrate the versatility of the interfacial redox strategy. We believe that the OER activity could be further enhanced by fine-tuning the Ni-to-Fe ratio of the NiFe foam or using more complex metal compositions.
3. Conclusions
The NF surface was oxidized by EG through interfacial redox modulation. Electrochemical voltammetry, in situ EIS, Raman spectroscopy, STEM, and theoretical calculations were employed in the analysis, revealing the generation of active Ni species. The Ni species includes Ni2+ species on NF for the reconstruction of more OER-favored γ-NiOOH and Ni single atoms and clusters on EG with superb OH− adsorption abilities. Therefore, the EG-modulated NF promoted the OER performance, delivering an overpotential of 290 mV at 10 mA⋅cm−2 and TOF of 0.86 s−1 at 400 mV. This interfacial redox modulation strategy was tunable in terms of both the oxidation degree of the EG layers and metallic Ni-based substrates. To verify the versatility of this strategy, an optimized NiFe system was demonstrated. An overpotential of 243 mV was needed to achieve 10 mA⋅cm−2, and the stability at 500 mA⋅cm−2 for 100 h was demonstrated. We believe that this strategy can be extended to other catalytic systems, providing an appealing route for the efficient design of electrode materials in energy-related devices.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was partly supported by the National Natural Science Foundation of China (22208296, 22278364, 22425805, U22A20432, 22238008, 22211530045, 22578393, and 22178308), the Research Funds of Institute of Zhejiang University-Quzhou (IZQ2021RCZX020, IZQ2021RCZX022, IZQ2021KJ2016, IZQ2021KJ2017, and IZQ2021RCZX040), the Development Project of Zhejiang Province’s “Jianbing” and “Lingyan” (2023C01226), the National Key Research and Development Program of China (2022YFB4002100), the financial support from the Fundamental Research Funds for the Central Universities (226-2024-00060), the Key Technology Breakthrough Program of Ningbo “Science and Innovation Yongjiang 2035” (2024H024), the Science Foundation of Donghai Laboratory (DH-2022ZY0009), the State Key Laboratory of Fine Chemicals, the Dalian University of Technology (KF 2113), and the Guangdong Basic and Applied Basic Research Foundation (2020A1515110338).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.04.028.