Ni electrodes with 3D-ordered surface structures for boosting bubble releasing toward high current density alkaline water splitting

Highlights

• •Ni electrode with 3D-ordered surface structures is constructed by electroetching.
• •3D-ordered surface structures promote the release of bubbles from the electrode.
• •High-speed camera confirms that larger bubbles hinder the convection of electrolyte.
• •The durability of the micro-nano-rough electrode is verified by the accelerated test.

Abstract

The performance of alkaline water electrolysis (AWE) at high current densities is limited by gas bubble generation on the surface of electrodes, which covers active sites and blocks mass transfer, resulting in lower AWE efficiency. Here, we utilize electro-etching to construct Ni electrodes with hydrophilic and aerophobic surfaces to improve the efficiency of AWE. Ni atoms on the Ni surface can be exfoliated orderly along the crystal planes by electro-etching, forming micro-nano-scale rough surfaces with multiple crystal planes exposed. The 3D-ordered surface structures increase the exposure of active sites and promote the removal of bubbles on the surface of the electrode during the AWE process. In addition, experimental results from high-speed camera reveal that rapidly released bubbles can improve the local circulation of electrolyte. Lastly, the accelerated durability test based on practical working condition demonstrates that the 3D-ordered surface structures are robust and durable during the AWE process.

1. Introduction

Hydrogen produced from water electrolysis coupling with renewable energy sources will be one of the primary energy generation/storage options for a low-carbon society in the 21st century [1], [2], [3]. AWE, proton exchange membrane electrolysis cells (PEMECs), and solid oxide electrolysis cells (SOECs) are currently the most feasible technical routes for hydrogen production from water electrolysis [4], [5]. Compared with PEMECs and SOECs, AWE is characterized by its high technology maturity and low investment cost [6], [7]. According to the International Energy Agency (IEA), the global electrolysis production capacity in 2020 is ∼3 GW/year, of which alkaline electrolysis accounts for 85%. However, commercial AWE operates at relatively low current density (<400 mA cm⁻²) and low voltage efficiency (<60%), due to the high alkaline electrolyte resistance, high diaphragm resistance, and low catalyst activity [8], [9].

The hydrogen and oxygen bubbles are generated in the AWE process at the cathode and the anode, respectively. The stoichiometric volume of hydrogen at the cathode is twice that of oxygen at the anode according to the following reactions:

$$\text{Cathode}:4{\text{H}}_2\text{O}+4e^{-}\to 4{\text{OH}}^{-}+2H_2$$ (1)

$$\text{Anode}:4{\text{OH}}^{-}\to 2{\text{H}}_2\text{O}+{\text{O}}_2+4e^{-}$$ (2)

$$\text{Overall:}{\text{2H}}_{\text{2}}\text{O}\to {\text{2H}}_{\text{2}}{\text{+ O}}_{\text{2}}$$ (3)

Cathodic hydrogen evolution reaction (HER) is taken as an example to show the effect of bubbles on the efficiency of AWE. The total overpotential ${\eta }_{\text{total}}$ from anode to cathode consists of the activation overpotential for HER${\eta }_{\text{act}}$, cell ohmic overpotential ${\eta }_{\text{ohm, cell}}$, concentration overpotential ${\eta }_{\text{con}}$, and bubble overpotential ${\eta }_{\text{bub}}$:

$${\eta }_{\mathit{total}}={\eta }_{\mathit{act}}+{\eta }_{\mathit{ohm},cell}+{\eta }_{\mathit{con}}+{\eta }_{\mathit{bub}}$$ (4)

Complex bubble dynamics includes nucleation, growth, detachment, and micro-convection, which exerts influence on HER, such as 1) ${\eta }_{\text{act}}$ by the blockage of the electrochemical active area, 2) ${\eta }_{\text{ohm,cell}}$ by forming the bubble layer, and 3)${\eta }_{\text{con}}$ by blocking micro-convection and ion-conducting pathways [9], [10], [11]. The bubble-free design of the electrode has been demonstrated to avoid 70 mV of ${\eta }_{\text{bub}}$ at 1 A/cm², which contains 40 mV of elimination of gas bubbles and 30 mV of hydrophobic anode increases electrochemical active surface areas (ECSAs) and facilitates gas removal [12]. A novel ring microelectrode encircling a hydrophobic microcavity was effective in avoiding bubble coverage in AWE, and the cell voltage is lower than those associated with conventional microelectrodes [13]. In addition, the tuning of flow fields [14], magnetic fields [15], [16], acoustic fields [17], [18], and electrolyte formulation [19] can mitigate the impact of bubbles on the overpotential of the electrolysis system. Nonetheless, although the above methods can mitigate bubble-induced energy losses, they also add complexity due to the addition of auxiliary measures in the electrolysis systems [20].

The efficient removal of bubbles can also be achieved by designing the geometry of the electrode, such as a 3D printing electrode with periodic pore structure [21] or pores with adaptive pore-throat aspect ratios [8]. Furthermore, on a microscopic level, bubbles tend to form at heterogeneous interfaces (e.g., crevices, edges, and corners in the electrode surface), where the thermodynamic barrier for nucleation is lower for gas evolution reaction [22], [23]. This provides an opportunity for controlling bubble dynamics [24], [25]. Zhang et al. developed an integrated bundle electrode with wettability-gradient copper cones, which is endowed with the multifunction of continuous generation, direct transport, and efficient collection of hydrogen bubbles [22]. The microscale linear arrays with 10 μm wide ridges separated by a distance of 200 μm allowed higher mobility of both the bubbles and the electrolyte [26], and the separation size is consistent with the statistical results of bubble size in water electrolysis [27]. Sun et al constructed a “superaerophobic” rough surface that can effectively drive the gas produced from the catalyst surface at a small critical size (≈20 µm) [24]. Superhydrophilic array Ni electrode reduced the overpotential of about 250 mV in HER without changing the inherent activity of Ni [28]. In addition, the preparation of active catalyst on the substrate by chemical- or electro-deposition can improve both the activity and the roughness of the electrode [29], [30]. The roughened electrode can increase the hydrophilicity and gas repellency, which can reduce the bubble diameter, accelerate the detachment of the bubbles, and ultimately, improve the water electrolysis efficiency [24], [28], [31], [32]. For all that, designing an economical electrode structure is a major challenge for its application in AWE industry.

Due to the complex coupling relationship between electrochemical dynamics and bubble dynamics (nucleation, growth, and detachment), it is difficult to quantitatively distinguish the influence of bubbles on overpotentials in a real electrolysis cell [33]. The actual operating condition in AWE has a great impact on the durability of the electrode, such as bubbles generated by high current density scouring the electrode, frequent start-up and stop, and reverse current causing corrosion and falling off of the catalyst [5], [34], [35]. While the electrodes or catalysts are usually tested in a narrow range of current density (≤100 mA/cm²), which is difficult to reflect the practical application potential.

Here, we report a practical Ni electrode with 3D-ordered surface structures obtained by electro-etching polycrystalline Ni. Through anodic etching, Ni atoms can be peeled off orderly in the direction of the crystal plane, forming the ordered angular surface structures, and the roughness of 3D-ordered surface structures can be tuned by the electro-etching time. The 3D-ordered surface structures of the Ni electrode accelerate the detachment of bubbles and electrolyte mass transfer in the electrolysis process and reduces most of the polarization loss caused by bubbles. Compared with the smooth Ni mesh electrode, the Ni electrode with 3D-ordered surface structures shows lower HER overpotential in the electrochemical test. The 600 cycles of accelerated test at 400 mA/cm² on practical work condition proves enough structure stability in the long-term operation of the AWE system, indicating the durability and application prospects of our electro-etched Ni electrode.

2. Experiments and characterization

2.1. Ni electrode preparation

1 cm*1 cm of Ni meshes (99.9%, 60 mesh, Kangweisiwang, Anpingxian) were obtained by the wire cutting machine to ensure the same sample area. The Ni mesh was first ultrasonicated in pure water and ethanol (1:1) for 20 min to remove the oil and impurities, followed by washing sequentially with excessive pure water. Then, the Ni mesh was placed in 200 mL of 2 M H₂SO₄ (MODERN ORIENTAL FINE CHEMISTRY, Beijing), and a 50 mA/cm² anode constant current (powered by an electrochemical workstation) was applied to etch the Ni metal with a Pt foil used for the cathode electrode. The etching time was changed to obtain samples with different roughness. The etched Ni mesh was ultrasonicated in pure water to remove the residual electrolyte, followed by drying in the air for 4 h for future use.

2.2. Characterization

The morphologies of the Ni meshes were investigated by field emission scanning electron microscope (SEM). The distributions of elements of samples were detected by energy-dispersive X-ray spectroscopy (EDS). The crystalline structure was assessed using an X-ray diffractometer (XRD). A glass electrolytic cell with two windows in a line and a high-speed camera was used to observe bubbles on the electrode surface.

2.3. Electrochemical measurements

The kinetic activity of the Ni meshes was conducted on AUTOLAB PGSTAT302N in a three-electrode configuration in 1 M N₂-saturated KOH electrolyte at 25 °C, mercuric oxide electrode, and Pt mesh (1.5 cm*1.5 cm) were used as the reference electrode and auxiliary electrode, respectively. All potentials were calculated with respect to the reversible hydrogen electrode (RHE) scale according to the Nernst equation ($E_{\text{RHE}}=E_{\text{SCE}}+0.241+0.059\times \text{pH}$) at 25 °C. The linear sweep voltammetry (LSV) curves were carried out at a scan rate of 5 mV s-1 for HER and OER measurements, and the electrochemical data were recorded with iR compensation (90%). CV was performed in the non-Faradaic region which was evaluated around the open circuit potential (for the cathode) and 0.95–1.05 V (for the anode), and a scan rate of 25, 50, 100, 200, 300, 400, 500 mV/s was chosen, respectively. The electrochemical impedance spectroscopy (EIS) measurements were performed in a frequency range between 100 kHz and 0.1 Hz with an ac amplitude of 5 mV. The durability of Ni meshes was valued in 30 wt% KOH at 80 °C, with the consideration of the industrial conditions.

3. Results and discussions

3.1. Electrode morphology

The pristine Ni mesh is woven from 0.2 mm Ni wire with a smooth surface, as shown in scanning electron microscope (SEM) images (Fig. S1a and Fig. 1a, b). To increase the surface roughness, sandpaper is conventionally employed to polish the electrode, however, only the outer surface of the Ni mesh could be polished (Fig. S1b). In contrast, all exposed surfaces in the Ni mesh can be effectively roughened via electro-etching (Fig. 1c, d). Herein, Ni meshes at the anode are etched at 50 mA/cm² with 10, 20, 30, and 40 min, and denoted as Ni-E10, Ni-E20, Ni-E30, and Ni-E40, respectively. With the increase in etching time, the surface of Ni wire becomes more orderly (Fig. 1c, Fig. S2a-d). For example, the 3D-ordered angular structures are formed on the surface of Ni-E30 (Fig. 1d). The grain boundaries display ribs that are about 10 μm in length and 0.3 μm in width (the insert image in Fig. 1d). XRD results (Fig. S3) reveal the cubic crystal structure of both pristine and Ni-E30, including the (1 1 1) plane at 44.5°, (2 0 0) plane at 51.8°, and (2 2 0) plane at 76.4°. Combined with SEM images analysis (Fig. 1 and Fig. S2), the processes for the formation of the 3D-ordered surface structure are elucidated as follows: initially, the atoms on the outermost surface are oxidized and separated under the action of the anode current (pristine Ni to Ni-E10 or Fig. S2a to Fig. S2b). After the initial electro-etching, the different crystal phases and grain boundaries on the Ni surface are exposed. Then, the crystal plane energy of Ni increases with the increase of lattice index, and the ordered oxidation and detachment of Ni atoms occur according to the crystal plane energy (Ni-E10 to Ni-E20 or Fig. S2b to Fig. S2c) [36], [37]. Finally, the 3D-ordered surface structures are formed (Ni-E30 to Ni-E40 or Fig. 1d to Fig. S2d).

Fig. 1.

Morphology and structure characterization of the Ni electrode: a) pristine Ni, b) high magnification view of the surface of pristine Ni, c) Ni-E30, and d) high magnification view of the surface of Ni-E30.

The roughness factors of the samples can be indirectly derived by normalizing the ECSAs of the ideal metal plane by identifying the electrical double-layer capacitance of the sample [21], [38]. Here, the electrical double-layer capacitance of Ni mesh is obtained by cyclic voltammetry at different scanning speeds around open-circuit voltage, and the roughness factor of Ni mesh is obtained by normalizing the smooth pristine Ni with electrical double-layer capacitance (Fig. 2a). The roughness factors of all samples subjected to electro-etching are increasing and reaching the maximum at 20 min (Fig. 2b, Table S1). However, the roughness factors of the samples decrease with an electro-etching time of greater than 20 min due to the reduction of the overall surface area, as the Ni wires become thinner in diameter under long-time electro-etching.

Fig. 2.

The roughness factors of Ni elecrodes: (a) Capacitive current against the scan rate measured in an N₂-saturated 1 M KOH, with a scan rate of 25, 50, 100, 200, 300, 400, 500 mV/s around the open circuit potential, (b) Normalized roughness factor by pristine Ni.

3.2. Bubble dynamics

Metal Ni is an aerophobic material, and the aerophobicity of the Ni surface is related to its roughness factor (r) [39]. According to the Cassie–Baxter equation (5):

$$\cos {\alpha }^{\prime }=fs\left(\cos \alpha +1\right)-1$$ (5)

where α is the apparent bubble contact angle, ${\alpha }^{\prime }$ is the apparent bubble contact angle on the rough solid surface, and $\text{fs}$ is the solid fraction of the contact area. A higher roughness factor will result in a smaller contact area between gas and solid, leading to much smaller bubble adhesion force and smaller bubble detachment size (Fig. 3a and 3b) [39]. Thus, the aerophobicity of the material could be significantly enhanced due to the high roughness of the surface [40]. The bubble contact angle on the Ni electrode cannot be directly measured due to the curved structure of Ni mesh, however, according to the model of bubble detachment on the wire, the diameter ${\text{d}}_{\text{b}}$ can be represented as [41], [42]:

$$d_b=\sqrt{\frac{24\sigma }{g\left({\rho }_L-{\rho }_G\right)}\left(\frac{{\sin }^2\theta }{2+3\cos \theta -{\cos }^3\theta }\right)}$$ (6)

$$\theta =\frac{\pi -\alpha }{1+{\sin }^{-1}\left(d_b/D_e\right)}$$ (7)

where σ is the liquid surface tension, ${\text{D}}_{\text{e}}$ is the diameter of Ni wires, ${\rho }_L$ and ${\rho }_G$ are the liquid and gas density, respectively.

Fig. 3.

Illustration of the contact angle on smooth (a) and rough (b) surfaces, (c) minimum bubble detachment size based on the specific contact angle, (d) average bubble detachment size distribution histogram based on image recognition.

Although this transcendental equation has no analytical solution, the relationship between the bubble contact angle and the diameter of the detached bubble is obtained by assigning a specific value. As shown in Fig. 3c, the minimum bubble detachment size decreases with the increase of bubble contact angle, which is consistent with the abovementioned roughness analysis. To obtain the bubble dynamic characteristics of AWE, all the Ni meshes are tested at 400 mA/cm² in 30 wt% KOH. The images of bubbles are firstly preprocessed in MATLAB, including gray processing, filtering, enhancement, and morphological processing. Then the circular bubbles are detected through the grayscale difference with the background and the bubble diameters are measured and statistically analyzed (Figs. S4–S8). The average bubble detachment sizes of attached bubbles decrease in the order of pristine Ni > Ni-E10 > Ni-E20 > Ni-E30 ≈ Ni-E40 (Fig. 3d, Video S1a-e). The experimental data are consistent with the theoretical trend, demonstrating that the rough surface of Ni with a high roughness factor/contact angle can facilitate the releasing of bubbles in water electrolysis [43], [44].

Specifically, as shown in the images and video taken by high-speed camera (Fig. 4a and Video S2), the sizes of attached hydrogen bubbles (at −0.6 V) on the rough surface are smaller than that of on the smooth surface, demonstrating that the rough surface of Ni can facilitate the releasing of bubbles in water electrolysis. For the Ni meshes, a large number of bubbles are attached to the pristine Ni with an average bubble size of 25.51 ± 10.13 μm in diameter (Fig. 4b), and their sizes are consistent with that of other materials (Pt plate or stainless steel mesh) [45]. While the average size of the bubbles in Fig. 4c is 19.48 ± 3.11 μm, consisting with the critical size (≈20 µm) [24], and no large bubble (>30 μm) appears on the Ni-E30 surface. Therefore, the impact of bubbles on the electrochemical active area (${\eta }_{\mathit{act}}$) and the bubble layer thickness (${\eta }_{\mathit{ohm},cell}$) has been reduced by the small bubbles on the 3D-ordered surface structures of Ni-E30. In a microscopic view (Fig. 4d, Video S3a), a smooth surface with a small bubble contact angle results in larger contact areas (black dot circle) and higher bubble adhesion forces, leading to the attachment of large bubbles. Furthermore, small bubbles are blocked by large bubbles, which moved up slowly under buoyancy. For Ni-E30 (Fig. 4e, Video S3b), bubbles form a fast-flowing channel along the back of the Ni wire intersection, the rapid removal of small bubbles forces convective mass transfer and decreases the concentration gradients [33]. Thus, the behavior of small bubbles on microscale reduces the ${\eta }_{\mathit{con}}$ polarization during the electrochemical reaction. The accelerating bubble separation by the 3D ordered surface structures is also effective at 200 and 600 mA/cm² (Video S4a, b), indicating its practicability in a wide current range.

Fig. 4.

High-speed camera images of hydrogen bubbles on a) a Ni wire with both smooth and rough surface at −0.6 V, b) pristine Ni, c) Ni-E30, d) pristine Ni in high magnification view, e) Ni-E30 in high magnification view. Yellow circle: bubble size, black circle: contact area.

3.3. Water electrolysis performances

To study the relationship between electrolytic potential and bubble dynamics, the samples with different roughness factors were investigated through the three-electrode system. It should be noted that Ni-E20 shows the largest roughness factor in the as-prepared samples, indicating the largest electrochemical active surface areas of Ni-E20 among the samples, nonetheless, Ni-E30 exhibits the best HER performance (Fig. 5a, Table S1, −0.413 mV@400 mA/cm², 38 mV lower than that of pristine Ni). This phenomenon was also reported in other studies, in which the ECSAs depend on the content, species, and morphology of the catalyst [25], [38], [46]. Herein, morphology or surface structure of Ni-E20 show the rough surface without the 3D-ordered structure, however, Ni-E30 shows the best HER performance, not only because of its larger ECSAs but also due to the specific 3D-ordered surface structures that accelerate the bubble dynamics, which reduces the bubble overpotential. In addition, the Tafel slopes of the as-prepared samples after electro-etching all decreased (Fig. 5b), indicating facilitated HER kinetics of the 3D-ordered surface structure. The current density of samples beyond −0.4 V increases linearly, indicating that electrochemical performance at high current density is dominated by ohmic polarization and concentration polarization [11], [27]. The slopes of the linear sweep voltammetry (LSV) curves demonstrate that the rapid removal of bubbles reduces the solution resistance and facilitates the OH⁻ conduction. After the LSV test, EIS was performed subsequently. Some of hydrogen bubbles produced in LSV test were attached on the surface of Ni electrode, which resulted in the reduced ECSAs. The EIS performed around 0.01 V (Fig. 5c) showed Ni-E30 had the maximum capacitance and the EIS performed around −0.5 V (Fig. 5d) furtherly showed the smallest resistance of Ni-E30 (Rs = R_solution + R_bubbles = 0.86 Ω), indicting the bubbles could be removed rapidly by the rough surface. Thus, Ni-E30 performs better at high current density in HER. In addition, the active contents, such as NiO and/or NiOOH, are removed from the Ni mesh surface by electro-etching, the etched Ni samples show less OER activity (Figs. S9 and S10) [47], [48]. Nonetheless, the OER performance does not recover by the continuous anodic activation, indicating the excellent oxidation resistance of the 3D-ordered surface structures.

Fig. 5.

HER electrochemical performances of Ni electrodes: (a) LSV curves with a 90% IR-correction, (b) Tafel slopes. The Nyquist plots: c) at 0.01 V and d) at −0.5 V (the inset shows the equivalent circuit).

So far, we have obtained a Ni electrode with the optimized 3D-ordered surface structures for HER performance. The kinetic activity of the water electrolysis catalyst was tested in 1 M KOH electrolyte at room temperature. However, the actual working condition of the electrolytic cell is usually 20 ∼ 30 wt% KOH, 70 ∼ 90 °C. Moreover, practical working conditions include frequent starts and stops at high current density, which will cause the corrosion and shedding of active materials from the electrode surface [34], [49]. Thus, the durability issue is one of the most important challenges for the application of alkaline water electrolysis cells with micro-nano structured electrodes [50].

Start-stop accelerated durability tests, conducted at 400 mA/cm² for 60 s and turned off the current for 30 s, are performed 600-unit cycles in 30 wt% KOH at 80 °C to test the durability of the Ni-E30 electrode, as shown in Fig. 6a. The chronopotentiometry test proved that 60 s is sufficient for the electrode to reach a steady state in the polarization process (Fig. S11). Pristine Ni meshes were used for both cathode and anode in the reference group, while pristine Ni and Ni-E30 were used as anode and cathode for the accelerated durability testing, respectively. The results (Fig. 6b) show that the electrolytic voltage of the double pristine Ni electrodes increased by 26 mV after the accelerated durability test, while the electrolytic voltage of pristine Ni and Ni-E30 dropped by 6 mV. Herein, the final electrolytic voltage between Ni-E30 and pristine Ni was 19 mV lower than the double pristine Ni electrodes, due to the greatly reduced negative effect from bubbles. Accordingly, Ni-E30 can save 1.74 × 10⁵ kW·h of electricity per year (10 h per day) for a 5 MW AWE system.

Fig. 6.

Durability performances of Ni electrodes: (a) Schematic diagram of the unit cycles of the accelerated durability test, (b) The full-cell potential from the average of the last 10 points of each cycle under at a temperature of 80 °C in 30 wt% KOH solution. (c) SEM images of Ni-E30 after accelerated durability test, (d) SEM images of of Ni-E30 after 48-hour of standing in 30 wt% KOH.

The cathode Ni-E30 after the accelerated durability test exhibits metallic luster optically (Fig. S12a), on the contrary, the anode Ni mesh appears black (Fig. S12b). The cathode electrode was suffering corrosion by the reverse current in 30 wt% KOH during the shutdown, in which the metal Ni on the surface was oxidized and dissolved to 30 wt% KOH. However, the 3D-ordered surface structures are almost retained in microscopic view (Fig. 6c), illustrating its durability under working conditions. In addition, compared with pristine Ni (Figs. S13 and S14a-e), a large amount of Cu impurities (17.8at. %, Fig. S15a-e) have been deposited on Ni-E30 under the action of the cathode potential. Cu impurities source may come from in thermocouple of thermometer and/or impurities in KOH.

After 48-hour of immersion in 30 wt% KOH, the cathode Ni-E30 is partially oxidized, and the Cu impurities (Fig. S16a-e, 0.8at. %) are mostly dissolved and disappeared, nonetheless, the complete 3D-ordered surface structures are preserved (Fig. 6d). The rib position, where the Cu impurities are deposited, is the active position with the high local intensity of the electric field for HER [22], [23]. In the microscopic view, the crystal plane and grain boundary on the surface of the Ni substrate are exposed, better yet, the electrolysis reaction occurs more easily at bulges and cracks of the electrode surface. Those structures are also the locations for the electrodeposited metal catalysts and hydrogen production, according to the distribution of the electric field in the rough surface [33].

4. Conclusion

The key to improving the current density of hydrogen production from alkaline water is to optimize the performance of the electrode. Aiming at the bubble problem under high current density, Ni electrodes with 3D-ordered surface structures were successfully prepared by the electro-etching method. The 3D-ordered surface structures increase the exposure of active sites in the crystal and grain boundary and accelerate the growth and detachment of bubbles by reducing the contact area between bubble and electrode. Ni-E30 exhibits the best HER performance with a bubble-related overpotential of 38 mV via mitigating bubble agglomeration through optimizing 3D surface structures. After accelerated durability tests, the performance of Ni-E30 is further improved, with 19 mV overpotential dropping in the full cell test, and the 3D-ordered surface structures are mostly retained, indicating their high durability. This work provides a guidance in the design of practical and efficient electrodes with micro-nano-scaled surface structures in the AWE system.

5. Author statement

Fuyuan Yang and Minggao Ouyang supervised the research. Jugang Ma, Mingye Yang, Guanlei Zhao designed the experiment and wrote the paper. Jugang Ma and Mingye Yang synthesized the material and performed the catalyst characterizations and catalytic experiments. Yangyang Li, Guanlei Zhao, Biao Liu, Jian Dang, Junjie Gu, Song Hu contributed to scientific discussion of the article.

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.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.