Defective graphene synergizes with NiO_x nanosheets to facilitate dual-defect-assisted two-electron oxygen reduction

Materials and synthesis methods

Every reagent was acquired from commercial sources and used as received without extra purification. Nafion (5%), Potassium nitrate (KNO₃, AR) and Melamine (C₃H₆N₆, 99%) were purchased from Aladdin. Potassium hydroxide (KOH, ≥ 85%) was purchased from Titan Technology. Nickel acetylacetone (C₁₀H₁₄NiO₄, 98%) was purchased from Merrier Biochemical Technology. Single-layer graphene powder (G, 0.8-1.2 nm) was purchased from Jiangsu Xianfeng Nano Material Technology. Anhydrous ethanol [anhydrous ethanol(C₂H₅OH)] was obtained from Sinopharm Chemical.

Materials characterizations

A Zeiss Geminisem 500 (Carl Zeiss AG, Germany) microscope operating at 5 kV was used to acquire scanning electron microscopy (SEM) pictures. Energy dispersive X-ray spectroscopy (EDX) mapping pictures, high-resolution transmission electron microscopy (HRTEM), high angle annular dark field scanning transmission electron microscopy (TEM), and TEM were acquired on the JEOL JEM-2100F (JEOL Ltd., Japan) 300 kV. The Rigaku Ultima IV (JEOL Ltd., Japan) was used to record X-ray diffraction (XRD) patterns. It operated at 40 kV with Cu Kα radiation (λ = 0.15406 nm), a scan rate of 10°/min, and a theta range of 5° to 90°. Raman spectra were obtained using a laser with a wavelength of 532 nm on a Raman spectrometer system (Raman spectrometer model: HORIBA HR Evolution, HORIBA, Ltd., France). The Brunauer-Emmett-Teller (BET) surface area of +the resultant materialthe resultant materialthe resultant material was calculated using a relative pressure of P/P₀ = 0.05-0.1. An Al-Ka radiation-powered Thermo Scientific K-Alpha device (Thermo Fisher Scientific, America) operating at 12 kV was used to perform X-ray photoelectron spectroscopy (XPS). The beamline 4B9A at the Beijing Synchrotron Radiation Facility (BSRF) in Beijing, China, has an optical emission end station where the Ni-K-edge X-ray absorption near-edge structure (XANES) spectra of the catalysts were examined.

Electrochemical measurements

Through the use of the CHI760E electrochemical workstation, a three-electrode system is used for all electrochemical testing. An Ag/AgCl electrode (saturated KCl solution), a carbon rod, and a rotating ring disc electrode (RRDE) (glassy carbon disc, 4 mm diameter, 5 mm inner diameter, 7 mm outer diameter, Pt ring) make up a three-electrode system. The platinum wire was used as the counter electrode and the Ag/AgCl electrode was used as the reference electrode. The reversible hydrogen electrode (RHE), E_RHE = E_Ag/AgCl + 0.197 + 0.0592^*PH^[32], was used to calibrate the reference electrode. Materials 5 mg of the sample, 10 uL of Nafion (5 weight percent), 700 uL of water, and 290 uL of isopropanol were sonicated for 1 h to create catalyst ink. The disc electrode was then covered with 5 uL of the dispersion, which was allowed to dry at room temperature. At the disc electrode, the catalyst’s mass loading was 0.20 mg·cm^-2. Before testing, O₂ was passed into 0.1 M KOH for 30 min to make sure O₂ saturation was achieved. Linear scanning voltammetry (LSV) tests were performed at 1,600 rpm in an O₂-saturated 0.1 M KOH electrolyte. Repeat at a scan rate of 10 mV·s^-1 over a potential range of 0.1 V vs. RHE to 1.0 V vs. RHE. The ring potential was also held at 1.2 V vs. RHE. Normalize all current densities to the area of the disc electrode.

The hydrogen peroxide yield [H₂O₂ (%)] and the number of electron transfers (n) are calculated based on^[33]

Where I_d represents the disk current, I_r denotes the ring current, and N = 0.42 signifies the current collection efficiency of the platinum ring.

Morphology and structure

NiO_x@DG was synthesized following the procedure depicted in Scheme 1, and detailed information can be found in the Supporting Information. In simple terms, NiO_x powder was obtained by heating nickel acetylacetone and KNO₃ in a muffle furnace at 350 °C for 90 min. In the XRD result [Supplementary Figure 1], it was concluded from the observation of characteristic peaks that the XRD of bulk NiO_x showed the characteristic peaks of NiO. This was because KNO₃ can act as an oxidizing base in this process, providing O^2- to form metal oxides^[34]. Furthermore, it was evident that the characteristic peaks of NiO_x bulk are broader and lower compared to those of NiO. This was attributed to the introduction of oxygen vacancies, which disrupt the order of the oxide’s lattice structure. In the SEM images [Supplementary Figure 2A], we can observe its rough surface, and the layered structure can be clearly seen (indicated by the white arrows in the image). However, no distinct particles were observed in the TEM images [Supplementary Figure 2B]. Nevertheless, HRTEM further describes the morphological structure of the NiO_x powder and the spatial distribution of the NiO particles within it. The presence of small black particles as NiO is indicated [enclosed by red dashed lines in Supplementary Figure 2C], and the lattice fringe spacing of 0.244 nm corresponds to the interplanar spacing of the (111) plane of NiO [Supplementary Figure 2D-F], a result that was consistent with the XRD analysis. The elemental mapping results [Supplementary Figure 2G] showed that Ni and O elements were uniformly distributed on the surface of the carbon carrier. Subsequently, the NiO_x powder was placed in ethanol and sonicated to exfoliate NiO_x NS, and the upper clear liquid was taken to remove larger NiO crystal particles and incompletely exfoliated bulk NiO_x. This was followed by hydrothermal treatment with DG (for detailed procedures, see Supporting Information) to obtain the final NiO_x@DG composite. In addition, control samples, including physical mixtures of NiO_x and DG (NiO_x-DG) and NiO_x@DG-9 composite with a changed mass ratio (NiO_x:DG = 9:1), were also prepared.

Scheme 1. The schematic procedure for synthesizing the NiO_x@DG; The gray balls represent C, the blue balls represent N, the red balls represent Ni, and the brown balls represent O. DG: Defective graphene; NG: nitrogen-doped graphene.

We characterized the composited sample to confirm the completion of exfoliation. From the XRD image in [Supplementary Figure 3], it can be observed that the characteristic peaks of the oxides disappear after the exfoliation and composite process. This indicated that centrifugation can separate the big crystalline particles and bulk NiO_x, leaving behind NiO_x NS in the upper clear liquid. In the SEM image [Figure 1A], NiO_x NS with a diameter of approximately 100 nm can be observed loaded on DG, and the presence of these nanosheets was also observed in the TEM image [Figure 1B]. The average thickness of NiO_x@DG was measured to be about 2 nm using atomic force microscopy (AFM) [Figure 1C and D]. HRTEM and corresponding inverse fast Fourier transform (IFFT) images were used to investigate the presence of crystalline structures of NiO [Figure 1E and F and Supplementary Figure 4]. In specific areas (enclosed by red boxes), irregular atomic arrangements were observed, indicating the presence of characteristic features of defective structures, which was caused by the overflow of lattice oxygen, consistent with the XRD results. Line scan analysis and elemental mapping spectra revealed the uniform distribution of C, O and Ni over the composite catalysts [Figure 1G and H and Supplementary Figure 5].

Figure 1. (A) SEM image, (B) TEM image, (C) AFM image of NiO_x@DG and (D), Height profile along the line of NiO_x@DG; (E) HRTEM image and (F) The enlarged HRTEM image within the red dashed frame marks the local area of disordered atoms; (G and H) SEM and line-scan element mapping for NiO_x@DG. SEM: Scanning electron microscopy; TEM: transmission electron microscopy; AFM: atomic force microscopy; DG: defective graphene; HRTEM: high-resolution transmission electron microscopy.

The powder XRD (PXRD) patterns [Figure 2A] verified that no clear typical diffraction peaks of metal oxides were present in NiO_x@DG and NiO_x@DG-9, with the exception of a broad peak at about 26° corresponding to the (002) plane of graphitic carbon. Subsequently, the chemical properties of the samples were thoroughly investigated using XPS. The high-resolution XPS spectra of Ni 2p for NiO_x@DG and NiO_x-DG are shown in Figure 2B and Supplementary Figure 6. These spectra can be well-deconvoluted into several sub-peaks that correspond to Ni2+ (nearly 855.7 eV) and two associated satellite peaks. Compared to NiO_x-DG, there is a noticeable negative shift of around 0.45 eV in the Ni2+ peak of NiO_x@DG, indicating that Ni acts as an electron acceptor to receive electrons transferred from DG. However, there is no significant shift in the Ni 2p XPS spectra between NiO_x NS-350 and NiO_x-DG [Supplementary Figure 7], implying that simple physical mixing cannot alter the electronic structure of the Ni centers. The Ni 2p XPS spectrum of NiO_x@DG-9 also shows two peaks [Supplementary Figure 8], the oxidation peak of Ni (2+) at a binding energy of ~ 855.7 eV, and the associated satellite peak, which is consistent with the results of Ni oxides observed in XRD and HRTEM. A larger negative shift in binding energy of 0.8 eV can be observed, suggesting that higher loading of NiO_x NS in NiO_x@DG-9 is beneficial for more oxygen vacancies to communicate with DG, thereby gaining more electrons. In the high-resolution XPS spectrum of O 1s [Figure 2C], three peaks can be observed, which were assigned to lattice oxygen at 529.93 eV, surface adsorbed oxygen at 532.63 eV, and lattice oxygen near oxygen vacancies at 531.4 eV. The O_d/O_t (O_d represents oxygen vacancy content and O_t indicates the total O content) ratio of NiO_x@DG was 20%, lower than that of NiO_x-DG (36.5%), which may be attributed to the increased surface adsorbed oxygen during the hydrothermal treatment to synthesize NiO_x@DG composite^[35]. Figure 2D showed the high-resolution XPS spectra of C 1s, which can be deconvoluted into three subpeaks, corresponding to C=C (sp² carbon, 284.6 eV), C-C (sp³ carbon, 285.3 eV), and C-O (287.3 eV). The electron paramagnetic resonance (EPR) results also confirmed the presence of oxygen vacancies, with a characteristic peak at g = 2.003, which was attributed to the formation of trapped electrons by oxygen vacancies [Figure 2E]^[36]. Moreover, it has been reported that oxygen vacancies play a role in enhancing the ORR activity^[37]. This further illustrates that KNO₃ was essential for the low-temperature construction of amorphous oxides [Supplementary Figure 9]. According to the reports, the electron transfer from the carrier to the active center also involves the synergistic participation of oxygen vacancies (The following text discusses in detail)^[38]. The I_D:I_G = 1.10 in the Raman spectrum further indicated that DG had a large number of defects [Figure 2F and Supplementary Figure 10].

Figure 2. (A) PXRD patterns; (B) Ni 2p XPS spectra; (C) N 1s XPS spectra and (D) C 1s XPS spectra of NiO_x@DG and NiO_x-DG; (E) EPR spectra of NiO_x@DG; (F) Raman spectroscopy of NiO_x@DG. PXRD: Powder X-ray diffraction; XPS: X-ray photoelectron spectroscopy; DG: defective graphene.

Extended X-ray absorption fine structure (EXAFS) and XANES studies were performed to further determine the local coordination environment of Ni in the catalyst. The Ni-K edge XANES spectra for the NiO_x@DG and reference materials are shown in Figure 3A. The absorption edges of these samples were located between the absorption edge of Ni foil and NiO. The Ni valence state was between 0 and +2, which was consistent with the XPS data. The Fourier-transformed EXAFS curves for NiO_x@DG, NiO_x@DG-9, and NiO_x NS-350 also showed the presence of Ni-O, with a peak in the R-space at approximately 1.68 Å, confirming that the Ni species primarily exist in the form of NiO clusters [Figure 3B]. The intensity of the peak is related to the number of atoms with the same average interatomic distance and the distribution of Ni centers^[39]. The introduction of oxygen vacancies led to the reduction of central Ni species, resulting in a decrease in coordination number and an increase in bond length compared to NiO, which was consistent with other literature reports^[40]. There is no Ni-Ni scattering channel at ~ 2.2 Å that corresponds to the crystal nickel metal structure. In the meantime, the Ni-O_xo backward scattering path of NiO is closely matched by another prominent peak at about 2.6 Å^[41]. The wavelet transform (WT)-EXAFS method can be used to distinguish backscattered atoms in R space and K space. Figure 3C and Supplementary Figure 11 demonstrated that the WT contour plots of the various reference samples show Ni-O at ~ 5.0 Å^-1; the intensity of Ni-O is strongest at ~ 7.3 Å^-1 in WT contour plots of different prepared samples, which corresponds to the Ni-Ni coordination structure of NiO and further indicates the presence of NiO_x. Finally, in order to accurately determine the local coordination environment of the Ni species, quantitative EXAFS fitting was carried out. The fitting results in Figure 3D and Supplementary Table 1 show that the Ni-O coordination number in NiO_x@DG is approximately 4.46, less than the Ni-O coordination number of bulk NiO (6.0). This result confirms the existence of oxygen vacancies.

Figure 3. (A) Normalized Ni K-edge XANES spectra of NiO_x@DG and reference materials; (B) FT of Ni K-edge EXAFS data of NiO_x@DG and reference materials. The boxed area is enlarged in Figure 3A’s inset; (C) Ni K-edge WT-EXAFS contour plots for NiO_x@DG and reference materials; (D) Ni K-edge EXAFS fitting curves for NiO_x@DG. XANES: X-ray absorption near-edge structure; DG: defective graphene; FT: fourier transform; EXAFS: extended X-ray absorption fine structure; WT: wavelet transform.

Performance testing

RRDE on a three-electrode system was used to assess the ORR performance of the catalysts in 0.1 M KOH solution. We first studied the effect of catalyst loading amount on ORR activity. As shown in Supplementary Figure 12, the optimal catalytic performance can be achieved when the loading amount is below 0.2 mg·cm^-2. The decrease in activity of high loading amounts may be attributed to a more severe catalyst stacking effect, which weakens the wettability and accessibility to the electrolyte. When the electrode loading is as low as 0.1 mg·cm^-2, although high 2e^- ORR acticity is preserved, systematic errors are inevitably introduced during the measurement process. Therefore, we choose 0.2 mg·cm^-2 as the testing loading amount for all the materials. The linear sweep voltammetry (LSV) curves in Figure 4A reveal that the disk-limited current density of NiO_x@DG exceeds all the other tested samples, with the value of about 2.4 mA·cm^-2 close to the theoretical mass transfer limit for the 2e^- ORR process (~ 3 mA·cm^-2)^[7,42]. For NiO_x@DG, NiO_x@DG-9, NiO_x NS-350, and NiO_x NS-550, the onset potential (defined as the potential at a current density of 0.1 mA·cm^-2) was all about 0.78 V, almost greater than the thermodynamic limit of 0.76 V (pH = 13 in 0.1 M KOH). Previous studies have shown that a potential related to the Nernst effect, exceeding the limit, can arise due to a combination of high 2e^- ORR catalytic activity and a low concentration of H₂O₂ generated in an alkaline environment^[10,43,44]. For NiO_x-DG, the higher onset potential (0.86 V), the higher disk current density (~ 2.1 mA·cm^-2), and the lower ring current density (~ 0.5 mA·cm^-2) indicated that they favor the kinetics of the 4e^- ORR pathway. According to Figure 4B, the 2e^- route consumes most of the electrons for NiO_x@DG throughout a broad potential range, offering a high H₂O₂ selectivity of almost 95% and an electron transfer number that is closest to 2 (2.1). This overwhelmed many reported catalysts for the electric production of H₂O₂ [Supplementary Table 2]. In contrast, NiO_x-DG obtained through physical mixing only has about 75% H₂O₂ selectivity, with an electron transfer number of n = 2.5. This was attributed to the difficulty in the physical mixing process for the electron cloud density of DG defects to be transported through oxygen vacancies as an electron bridge, resulting in a lower electron cloud density at the Ni sites. The high electron cloud density at the centers of activity sites was essential for the adsorption and desorption of intermediates^[45]. The rate-determining step (RDS) and reaction kinetics were analyzed using the Tafel slopes [Figure 4C]. The disk Tafel slopes for the various samples were 230.4 mV·dec^-1 (NiO_x-DG), 112.1 mV·dec^-1 (NiO_x@DG-9), 90.1 mV·dec^-1 (NiO_x NS-350), 82.6 mV·dec^-1 (NiO_x NS-550), and 76.4 mV·dec^-1 (NiO_x@DG), verifying that the RDS changed from the first ^*OOH production stage (O₂ → ^*OOH, ~ 120 mV·dec^-1) to the dissociation step (^*OOH → H₂O₂, ~ 70 mV·dec^-1)^[46,47]. Additionally, the NiO_x@DG sample exhibited the lowest Tafel slope for the ring current, indicating its capability for rapid H₂O₂ production [Figure 4D]. This was likely due to the oxygen vacancies acting as electron bridges, leading to a higher electron density at the Ni sites, which was relevant to the adsorption/desorption of intermediates^[48]. Accelerated deterioration tests (ADT) were subsequently used to assess the catalyst’s stability, as shown in Figure 4E and F. The outstanding endurance of NiO_x@DG for the electrosynthesis of H₂O₂ was confirmed by the LSV curves, which showed a negative shift after 5,000 ADT cycles but still maintained a high H₂O₂ selectivity of about 85%. For the chronoamperometry stability test at a constant potential of 0.2 V vs. RHE, the ring-disk current slowly decreases as ORR proceeds, but the selectivity is maintained at a relatively high level of 86%, which can be comparable with many reported excellent catalysts toward 2e^- ORR [Supplementary Figure 13 and Supplementary Table 3]. To investigate the reason for activity deterioration after cyclic voltammetry (CV) stability testing, we conducted series of post-mortem characterizations [Supplementary Figure 14]. According to the HRTEM images, distinct NiO lattice fringes can be observed, indicating most NiO crystal is preserved. However, extensive defect structures emerge in the NiO lattice. This confirms that lixiviation of Ni species occurs during the CV stability testing. The dissolution of partial Ni sites can also be demonstrated by the XPS spectra and inductively coupled plasma mass spectrometry (ICP-MS) data [Supplementary Table 4]. All these findings affirm that the activity decrease of NiO_x@DG during long-term working conditions comes from the slow Ni leaching into the electrolytes.

Figure 4. (A) Polarization curves of NiO_x@DG, NiO_x@DG-9, NiO_x-DG, NiO_x NS-350 and NiO_x NS-550 on RRDE at 1,600 rpm in O₂-saturated 0.1 M KOH; (B) The calculated H₂O₂ selectivity and Transfer electron n of catalysts; (C and D) Tafel slope of NiO_x@DG, NiO_x@DG-9, NiO_x-DG, NiO_x NS-350 and NiO_x NS-550; (E and F) Stability test of NiO_x@DG catalyst after 5,000 cycles. DG: Defective graphene; RRDE: rotating ring disc electrode; n: number; RHE: reversible hydrogen electrode.

To further confirm the synergistic effect of oxygen vacancies in NiO_x and topological defects in DG for the electrosynthesis of hydrogen peroxide, two single-defect samples (NiO_x@G and NiO@DG) were prepared for comparison. As illustrated in Supplementary Figures 15 and 16, both NiO_x@G and NiO@DG manifested significantly reduced H₂O₂ selectivity compared with NiO_x@DG composites, emphasizing the critical role of synergy between interfacial oxygen vacancies and carbon defects toward 2e^- ORR. Noteworthily, the ORR pathway on the NiO@DG sample is 4e^- process, while NiO_x@G still preserves certain electrosynthesis performance of H₂O₂. In addition, comparing the ORR performance between NiO@DG and NiO@G [Supplementary Figure 17] also confirms that only introducing intrinsic carbon defects in G cannot activate the Ni atoms in hybridized NiO materials. All the results demonstrate the dual defects promoted mechanism in NiO_x@DG can be depicted as follows: (1) Oxygen vacancies play a crucial role in activating the adjacent Ni atoms to serve as the main active centers, transferring 4e^- ORR route to approach 2e^- pathway; (2) The coupled topological defects in DG possibly optimize the electronic structures around active sites, finally attaining an optimal electrocatalytic performance toward 2e^- ORR.

Doping heteroatoms in carbon materials is also well-accepted to efficiently regulate the atomic and electronic configurations besides the intrinsic defect engineering. Hence, we further studied whether doping N₂ atoms in G can also boost the H₂O₂ electroproduction performance, similar to the effect of topological defects. As shown in Supplementary Figures 18 and 19, N dopants in G cannot lead to activity enhancement in 2e^- ORR as compared to NiO@G and NiO_x@G, which indicates its uselessness in such composite system toward ORR process. To directly uncover the influence of the G substrate itself on ORR processes, a series of G-based carbon materials were prepared with similar methods at different temperatures (800-1,150 °C, denoted as NG-800, NG-900, NG-1,000 and DG-1,150 °C) were prepared. The defect structure and elemental state were first characterized by Raman and XPS spectra, showing that N₂ content decreased while the defect density rose as the pyrolysis temperature increased, consistent with previous reports that removal of N₂ dopants in carbon lattice can induce the rearrangement of surrounding carbon atoms to form more topological defects [Supplementary Figure 20A and B]. The ORR activities of these carbon samples were then evaluated on RRDE and the results were presented in Supplementary Figure 20C and D. Apparently, the ORR processes on all the carbon materials without coupled NiO species are close to 4e^- transfer pathway, indicating that neither N dopants nor carbon defects can directly boost the desorption of generated H₂O₂ to prevent the cleavage of O-O bonds. This can further confirm our speculation that, in NiO_x@DG composites, Ni atoms beside oxygen vacancies act as the real catalytic centers for O₂ adsorption and activation, while the interfacial carbon defects contribute to optimizing the electronic structure of active Ni atom to achieve the final excellent 2e^- ORR activity.

Density functional theory (DFT) calculations were further conducted to gain a deep understanding on the activity mechanism. As pentagon defects in carbon lattice have been demonstrated to show excellent electrocatalytic performance in various reactions, here we choose G lattice possessing pentagon defects as the defective carbon model. Accordingly, three composite models were constructed: NiO with one oxygen vacancy coupled DG (NiO_x@DG) [Supplementary Figure 21A], NiO with one oxygen vacancy coupled perfect G (NiO_x@G) [Supplementary Figure 21B], and NiO lattice without oxygen vacancies coupled DG (NiO@DG) [Supplementary Figure 21C]. The models with adsorbed ^*OOH intermediates were also presented in Supplementary Figure 21. It can be seen that, only in NiO_x@DG model, the ^*OOH is laterally adsorbed on the NiO_x layer, which may facilitate the elongation of the O–O bond^[16], and lead to the consequent desorption process. The adsorption energy of the key intermediate ^*OOH is generally used as a key descriptor for predicting the 2e^- ORR.^[15] As shown in Supplementary Figure 21D, NiO_x@DG model has the lowest energy barrier (1.24 eV) under alkaline conditions as compared to the other two models, indicating its energetically favorable 2e^- ORR process due to the enhanced intermediate desorption capacity.