Fabrication of honeycomb-structured composite material of Pr₂O₃, Co₃O₄, and graphene on nickel foam for high-stability supercapacitors

Abstract

Hydrothermal synthesis and annealing process were used to create a honeycomb-structured composite electrode material on nickel foam substrate that contained praseodymium oxide (Pr₂O₃), cobalt oxide (Co₃O₄), and reduced graphene oxide (rGO) (named as Pr₂O₃/Co₃O₄/rGO/NF). The Pr₂O₃/Co₃O₄/rGO/NF (PCGN) composite electrode material was created without the need for a binder and can be employed right away to create a supercapacitor. This honeycomb composite is further grown by the mesh structure of Pr₂O₃ and Co₃O₄, which has good electrical conductivity. This nanomorphology not only improves the specific surface area of the material but also facilitates the improvement of electron transfer efficiency. The synthesized electrode material has excellent electrochemical properties due to its special morphology and excellent conductivity, and its performance is better than that of praseodymium oxide and cobalt oxide alone, with a specific capacitance of 3316 F g⁻¹ in the three-electrode system at a current density of 1 A g⁻¹. After forming a symmetrical supercapacitor device, the energy density is 74.8 W h kg⁻¹ and the power density is 300 W kg⁻¹ at 0.5 A g⁻¹. The capacitance retention is 82% even after 35 000 cycles at 2 A g⁻¹.

---

1. Introduction

Energy is an important component that drives the operation and development of society. Clean and sustainable energy sources should be utilized immediately to take the place of fossil fuels due to the current energy crisis and environmental damage brought on by the overuse of these fuels. And the development of new and efficient energy storage devices has become the key to the problem.^1–3 Due to their high power density, high energy density, and extended cycle life, supercapacitors are becoming a hot topic in academia. Supercapacitors (SCs) are used in a variety of industries, like aerospace, wearable technology, electric automobiles, and other areas. Supercapacitors are primarily split into three groups based on how energy is stored: double-layer supercapacitors (EDLC), pseudocapacitors, and hybrid supercapacitors.^4–6 Double-layer supercapacitors are mainly used for energy storage by using the high specific surface area of carbon-based materials.⁷ In contrast to double-layer supercapacitors, the principal method of energy storage in pseudocapacitors is a quick and reversible Faraday reaction on the surface of electrode materials.^8,9 As a result, it is more capacitive and has a higher energy density than a double-layer supercapacitor.

Metal oxides, particularly transition metal oxides such as MnO₂,^10,11 RuO₂,¹² Co₃O₄,^13–16 and others, are good electrode materials for pseudocapacitors due to their high theoretical specific capacitance and abundance. As a transition metal oxide Co₃O₄, has a very optimal theoretical specific capacitance as well as the benefits of environmental protection and low cost, making it an excellent material for supercapacitors.^14,16–18 Li et al. claim that they created Co₃O₄/MnO₂/Co(OH)₂ composite electrodes using a straightforward hydrothermal process along with the proper heat treatment and electrodeposition and that these electrodes had a high specific capacitance of 3022.2 F g⁻¹ at a current density of 10 A g⁻¹.¹⁸ Mn@Co₃O₄-MSs modified electrodes were created by Chen et al. using solvothermal reaction and annealing treatment, and they displayed a high specific capacitance of 773 F g⁻¹ at 1 A g⁻¹.¹⁹ Co₃O₄/NiO/Mn₂O₃ has a significant surface area (123.9 m² g⁻¹) and high specific capacitance of 3652 mF cm² at 1 mA cm² current density.¹¹ With its simple aggregation, poor conductivity, and significant volume change during charging and discharging, Co₃O₄ has a limited range of applications. Currently, adding other compounds can help to solve the issues of low multiplicity and poor conductivity, and materials based on carbon are a great choice. Graphene conductor is one of the carbon-based materials that are particularly well suited for combining with the Co₃O₄ phase because it has excellent electrical performance, a large specific surface area, and good flexibility.^20,21 For instance, the RGO/Co₃O₄ composite demonstrated excellent electrochemical cycling stability with capacitance retention of more than 122% after 5000 continuous charge and discharge cycles and displayed a high specific capacitance of 331 F g⁻¹ at a current density of 5 A g⁻¹.²² By using a straightforward hydrothermal process, Liu et al. created GO(rGO)/Co₃O₄ composites.²³ The addition of rGO increased the electrode material's specific surface area and conductivity, and the composite electrode demonstrated a capacitance of 263.0 F g⁻¹ in a two-electrode supercapacitor at a current density of 0.2 A g⁻¹. A multi-composite material also performs better than a single metal material when compared to it because of the synergistic impact between the many materials, which not only helps to compensate for the shortcomings of a single material but also promises to enhance performance even further.^11,13,15,18 The specific capacitance values of Co₃O₄–Ni₃S₄–rGO electrodes were 899 F g⁻¹ at a current density of 0.5 A g⁻¹;13 rGO/Co₃O₄@Fe₂O₃ hybrid network as electrodes revealed a high capacitance of 784 F g⁻¹ at 1 A g⁻¹;15 NiO/Co₃O₄/NCDs nanosheets as electrodes exhibited a high specific capacity of 976.3 C g⁻¹ (1775 F g⁻¹) at 1 A g⁻¹.17 Overall, creating composite nanomaterials by mixing graphene with the oxides of other metals is a good solution to address current issues of Co₃O₄.

Rare earth elements are abundant in the crust of the earth and have a distinct 4f electronic structure, which allows them to exhibit outstanding electrochemical capabilities through a multi-energy leap.^24–26 In addition, the use of rare earth aids in modifying and enhancing the morphological structure and improving electrode performance.^27–30 They receive a lot of attention in the field of energy storage because of all their distinctive qualities. La–Co LDHs as the working electrode had a high specific capacitance of 2162 C g⁻¹ at a high current density of 10 A g⁻¹;³¹ MnO₂/La₂O₃ composites synthesized by the hydrothermal method had a specific capacitance of 245.4 F g⁻¹ at 0.3 A g⁻¹,²⁸ which was 1.5 times higher than MnO₂; MnO₂/CeO₂ has a specific capacitance of 274.3 F g⁻¹ at a current density of 0.5 A g⁻¹ which is much higher than MnO₂ (190.2 F g⁻¹) and CeO₂ (66.2 F g⁻¹).²⁹ Due to their incomplete 4f shells, rare earths operate as transient electron carriers and can speed up redox reactions when combined with cobalt trioxide. The structural flaws that result from compounding together can be seen by scanning electron microscopy (SEM).²⁷ Therefore, composing with rare earth elements effectively increases specific capacitance and cycle life and optimizing electrodes by producing oxygen vacancies, reducing charge transfer resistance, facilitating easier electrolyte ion diffusion, accelerating oxygen ion migration, controlling morphology and grain size.^28,30 Praseodymium, which is a rare earth with a high content and also has a distinctive 4f electron energy level structure, it has been widely employed in many industries and is a suitable material for supercapacitors,³² certain studies like Pr₆O₁₁@NiCo₂O₄ have a capacitance of 1635 F g⁻¹ at a current density of 0.5 mA cm⁻²;³³ NiFeCoPrO–Au/NF electrode exhibits a capacitance of 1792 F g⁻¹ at a current density of 10 A g⁻¹;34 PANI–Pr₂O₃–NiO–Co₃O₄ has a capacitance of 905 F g⁻¹ at current density 1 A g⁻¹ all confirm that.^35,36

In this study, a nickel foam substrate was used to create a thin-layered PCGN electrode material with good multiplicative performance through a straightforward hydrothermal reaction with the coexistence of nanosheet layers and nanowires. This experiment's synthesis method is very simple, and the composites are grown directly on NF without the use of conductive agents, surfactants, or adhesives to connect, which reduces damage to the three-dimensional skeleton of nickel foam,³⁵ avoids clogging and damage to the active material, ensures higher conductivity, and reduces contact resistance between the electroactive material and NF. As a result of the aqueous symmetric supercapacitor (SSC) device made of two PCGN electrodes, which has 300 W kg⁻¹ power density and 74.8 W h kg⁻¹ excellent energy density as well as remarkable stability even after 35 000 cycles. It brings hope for the preparation of environmentally friendly and inexpensive storage devices.

2. Experiment

2.1. Materials and reagents

Shanghai Aladdin Biochemical Technology Co. Ltd supplied urea (CH₄N₂O) and cobalt acetate tetrahydrate (C₄H₆CoO₄·4H₂O). Saen Chemical Technology Shanghai Co., Ltd supplied praseodymium acetate pentahydrate (CH₃CO₂)₃Pr·4H₂O). Pristine graphite powder, hydrogen peroxide (H₂O₂, 30 wt%), potassium permanganate (KMnO₄), hydrogen chloride (HCl, 36–38 wt%), sulfuric acid (H₂SO₄, 95–98 wt%), potassium hydroxide (KOH), sodium nitrate (NaNO₃), acetone (CH₃COCH₃, 99.5 wt%) and ethanol (C₂H₅OH, 99.5 wt%) was provided by The Beijing Fine Chemical Co. Ltd. Purchased the nickel foam (NF) from Kunshan Toll Hui Electronics Technology Co. Ltd. All of the compounds were analytical grade, and they were used just as they were received with no additional purification.

2.2. Material preparation and processing

2.2.1. Pre-experimental pretreatment. The large pieces of nickel foam were cut into small sheets (20 mm × 10 mm × 1 mm) and sonication washed with 3 mmol hydrochloric acid, acetone, ethanol, and deionized water for 30 min, respectively, and dried before the reaction. A modified Hummers’ method was used to make GO, and before the reaction, a graphene oxide solution with a concentration of 2 mg mL⁻¹ was sonicated for 1 h.

2.2.2. Synthesis of Pr₂O₃/Co₃O₄/rGO/NF. The electrode material was synthesized by the hydrothermal method. Weigh 0.85 mmol of cobalt acetate, praseodymium acetate, and 1 mmol of urea, add 7.5 mL of water, stir magnetically for 40 min and then add the same volume of graphene oxide solution with a concentration of 2 mg mL⁻¹ and continue to stir for 40 min. Place the mixture, along with the washed and dried nickel foam, in a 25 mL autoclave lined with polytetrafluoroethylene (PTFE), and reacted at 180 °C for 12 h. Following the completion of the reaction, the reaction kettle was cooled to room temperature, and the electrode is washed with deionized water and ethanol before being dried at 90 °C for 4 h (Fig. 1).

Fig. 1 The preparation process of Pr₂O₃/Co₃O₄/rGO/NF composite electrode.

Finally, the Pr₂O₃/Co₃O₄/rGO/NF precursor was placed in a crucible and annealed at 300 °C for 2 h. On nickel foam, the average amount of Pr₂O₃/Co₃O₄/rGO growth was 1.5 ± 0.07 mg cm⁻². The preparation process of Pr₂O₃/Co₃O₄/NF, Pr₂O₃/rGO/NF, Co₃O₄/rGO/NF and rGO/NF is shown in Fig. S1–S4 (ESI†).

2.3. Characterization techniques

X-ray electron spectrometer (XPS, model PHI 5300, Physical Electronics, USA) and Bruker D8 Advance X-ray powder diffractometer equipped with Cu Kα radiation (λ = 0.15418 nm) in the range of 2θ = 10°–80° (XRD), to analyze the composition and crystal structure of the material. A Leica DMLM microscope, a Renishaw InVia confocal Raman spectrometer, and an argon ion laser (514.5 nm, model Stellar-REN, Modu-Laser) as the excitation source was used to collect Raman spectra. The surface morphology and structure of the sample electrodes were examined using scanning electron microscopy (SEM, Quanta 600, FEI) with an energy dispersive spectrometer (EDS, Oxford, Gemini 300), transmission electron microscopy (TEM, JEM-2100, JEOL), and high-resolution transmission electron microscopy (HRTEM) to analyze. The surface area was determined by N₂ adsorption and desorption tests using Brunauer–Emmett–Teller (BET) apparatus, and the pore size distribution was computed by the Barret–Joyner–Halenda (BJH) method.

2.4. Electrochemical testing

A three-electrode system in an electrochemical workstation (CHI 760E, CH Instruments) was used to measure the electrochemical properties of the electrode materials. The saturated calomel electrode served as the reference electrode, the platinum electrode served as the counter electrode, and the prepared electrode served as the working electrode. All electrochemical tests were performed in an aqueous electrolyte of 6 mol L⁻¹ KOH.

2.4.1. Cyclic voltammetry (CV). CV tests were used on the electrode materials to compare the electrochemical performance changes of each electrode. The scan rate was 5–100 mV s⁻¹ and the potential window was −0.2 to 0.5 V.

2.4.2. Galvanostatic charge/discharge (GCD). A three-electrode system was used for GCD testing to compare the specific capacitance of the electrode materials. For GCD testing, a voltage window of −0.1 V to 0.4 V was used with current densities ranging from 1 A g⁻¹ to 10 A g⁻¹.

Eqn (1) gives the mass-specific capacitance measured in the three-electrode system.

Characterization of symmetric supercapacitors using a two-electrode system. Where eqn (2) is used to calculate the energy density (E, W h kg⁻¹) and eqn (3) is used to calculate the power density (P, W kg⁻¹).where m is the mass of the active material on the electrode, C is the specific capacitance of the electrode, I represents the constant discharge current, Δt is the discharge time, and ΔV indicates the total potential window.

2.4.3. Electrochemical impedance spectroscopy (EIS). EIS was employed to assess the mass transfer resistance and electrode interfacial resistance, as well as to investigate the causes of the enhancement of electrochemical characteristics. EIS was measured at open circuit voltages from 100 kHz to 0.01 Hz with an amplitude of 5 mV s⁻¹.

2.4.4. Cycling stability test. The electrochemical stability of the Pr₂O₃/Co₃O₄/rGO/NF composite electrode was explored using a constant current charge and discharge cycle test, and 35 000 stability tests showed that the composite electrode has high electrochemical performance and cycling stability.

3. Results and discussion

3.1. Composition and structure

Fig. 2a displays the results of an X-ray diffraction (XRD) analysis used to confirm the crystal structure of Pr₂O₃/Co₃O₄/rGO/NF. According to the PDF standard card (JCPDS No. 22-0880), the diffraction peaks of 2θ at 33.7° and 39.1° correspond to the crystallographic planes of Pr₂O₃ (3 1 1), (−1 1 3) and belongs to the monoclinic phase.³⁷ Also, a comparison of the standard card (JCPDS No. 42-1467) demonstrates that the Co₃O₄ (1 1 1) and (5 1 1) crystal faces are represented by the diffraction peaks of 2θ at 19° and 59.4°, and belongs to the cubic phase.³⁸ According to the PDF standard card (JCPDS No. 04-0850), the broad diffraction peak of 2θ at 20° corresponds to the (0 0 2) crystal face of rGO.³⁹

Fig. 2 (a) XRD pattern of the Pr₂O₃/Co₃O₄/rGO/NF, (b) Raman spectra of Pr₂O₃/Co₃O₄/rGO/NF, Pr₂O₃/Co₃O₄/NF and GO.

The diffraction peaks of 2θ at 44.5°, 51.8°, and 76.4° correspond to the (1 1 1), (2 0 0), and (2 2 0) crystallographic planes of singlet Ni. Since no additional impurity peaks were discovered in the samples, it is presumed that Pr₂O₃/Co₃O₄/rGO/NF was successfully prepared by the experimental scheme.

Further study of whether graphene has been successfully loaded on the electrode material has been carried out by using Raman spectroscopy for Pr₂O₃/Co₃O₄/rGO/NF, Pr₂O₃/Co₃O₄/NF, and rGO. All three samples were tested. It can be seen from Fig. 2b that Pr₂O₃/Co₃O₄/rGO/NF and rGO both exhibit more pronounced D-band and G-band characteristics of graphene, which are located at 1330 cm⁻¹ and 1586 cm⁻¹. The D-band is generated by the stretching vibration of graphene sp³ hybridization and the G-band is caused by the stretching vibration of sp² hybridization.³⁹ The regularity of the graphene structure can be examined by calculating the intensity ratio of the D-band and G-band. Calculations indicated that rGO has I_D/I_G ratio of 0.97, whereas Pr₂O₃/Co₃O₄/rGO/NF has its I_D/I_G ratio of 1.38. According to earlier research, the rise in the I_D/I_G value is a clear indication that the sp² hybridization structure has been disturbed. Therefore, it can be concluded that a portion of the sp² heterostructure in rGO is disrupted as a result of the heightened electronic interaction between Pr₂O₃, Co₃O₄, and rGO. The characteristic peaks associated with rGO are not present in the sample Pr₂O₃/Co₃O₄. Meanwhile, the characteristic peaks at 534 cm⁻¹ and 662 cm⁻¹ in Pr₂O₃/Co₃O₄ can be attributed to F_2g and A_1g of Co₃O₄,^40,41 and the characteristic peak at 385 cm⁻¹ is mainly owing to E_g of Pr₂O₃.^37,42 The study can demonstrate both the successful synthesis of ternary composite samples and the presence of rGO in Pr₂O₃/Co₃O₄/rGO/NF samples.

In addition, XPS analysis of Pr₂O₃/Co₃O₄/rGO/NF was also performed (Fig. 3a–e). The existence of the four elements Pr, Co, C, and O in the manufactured material is indicated by the scan region in Fig. 3a from 0 to 1100 V. The high-resolution spectra of C 1s are shown in Fig. 3b, with the peaks at 282.6 eV, 283.9 eV, and 286.7 eV, which correspond to C=C, C–C, and O–C=C. Three peaks in the O 1s spectra, with energies of 529.5 eV, 531.3 eV, and 532.9 eV, seen in Fig. 3c. Where 529.5 eV corresponds to the bonds in metal oxides, 531.3 eV corresponds to the hydroxide oxygen-containing bonds, and 532.9 eV corresponds to physisorbed oxygen.⁴³ It can be seen that some oxygen elements in PCGN exist as metal oxides. In Fig. 3d, the results of Co 2P tests show that Co has two valence states. The peak of Co 2p_3/2 spin orbitals is 778.1 eV and the peak of Co 2p_1/2 spin orbitals is 794.1 eV, and there are two weak satellite peaks located at 784.1 eV and 800.7 eV near both of them, the difference in binding energy between Co 2p_1/2 and Co 2p_3/2 is about 16 eV, indicating the existence of the interconversion of Co²⁺ and Co³⁺ during the redox reaction.⁴⁴

Fig. 3 (a) XPS spectra of the Pr₂O₃/Co₃O₄/rGO composite electrode material; high-resolution XPS spectra of (b) C 1s; (c) O 1s; (d) Co 2p; and (e) Pr 3d of the Pr₂O₃/Co₃O₄/rGO composite electrode material.

According to previous studies, when rGO is present a weakly oscillating satellite peak of about 784.1 eV is caused by a strong interaction between Co₃O₄ and rGO via Co–O.⁴⁵ Further analysis using the Gaussian fitting approach reveals that the characteristic peaks of Co³⁺ contain 777.9 eV and 793.1 eV, while the characteristic peaks of Co²⁺ contain 779.6 eV and 794.9 eV, confirming the presence of Co(II)/Co(III) in the sample.⁴⁶ Finally, it can be seen from Fig. 3e Pr 3d_3/2 binding energies (BEs) of Pr₂O₃ range from 929.5 to 952.6 eV.⁴⁷ The outcomes are in line with that the Pr 3d_5/2 and the claimed binding energy, proving that Pr₂O₃ was loaded onto nickel foam satisfactorily.

They were put through energy dispersive spectrometry (EDS) examinations in Fig. S5 and S6 (ESI†). As seen in the picture, Co, C, O, Pr, and Ni are evenly distributed throughout the Pr₂O₃/Co₃O₄/rGO/NF electrode sheet. This further verifies the existence of Pr elements on the Pr₂O₃/Co₃O₄/rGO/NF nanoflake.

Then, SEM and TEM tests were carried out to further determine the morphology and dimensions of the electrode material.

A scanning electron microscope (SEM) are displayed in Fig. 4. NF, rGO/NF, Pr₂O₃/NF, Pr₂O₃/rGO/NF, Co₃O₄/NF, Co₃O₄/rGO/NF, Pr₂O₃/Co₃O₄/NF, and Pr₂O₃/Co₃O₄/rGO/NF, are all depicted in Fig. 4a–i, with Fig. 4i displaying a detail magnification of Fig. 4h. It can be seen that the treated NF shown in 4a and b has a three-dimensional skeleton structure with a smooth surface, while the rGO grown on the NF shows a muslin shape with a folded surface. The nanosheet structure of Pr₂O₃ is tightly stacked; the nanowire structure of Co₃O₄ is disordered and the specific surface area is limited; both structures are not conducive to the entry and action of electrolyte, which limits the material's electrochemical performance.⁴⁸Fig. 4d–f show that the dispersion of Pr₂O₃ nanosheets was helped by the addition of rGO. On the other hand, the nanowire morphology formed by Co₃O₄ on NF after the addition of rGO was more dispersed compared. This suggests that rGO can assist the electrode material in dispersing and changing shape, which helps in increasing the specific surface area. In Fig. 4g, SEM images of Pr₂O₃/Co₃O₄/NF are a honeycomb morphology generated by the co-growing of Pr₂O₃ and Co₃O₄. On nickel foam, the honeycomb shape exhibits a high growth density and a large specific surface area.⁴⁹Fig. 4h and i show a honeycomb architecture in the Pr₂O₃/Co₃O₄/rGO on NF. The co-growing of Pr₂O₃ and Co₃O₄ after the addition of graphene leads to a mesh-like structure that is different from the nanosheets alone, as illustrated when the morphology is magnified.^49,50 The overall specific surface area of the honeycomb shape is further increased due to the appearance of the mesh-like structure. This intertwining of mesh-like structures into a honeycomb-like morphology not only helps to increase the active specific surface area and ion transport channels of the material and improve the ion transport and electrolyte transport efficiency, but it also allows the electrode material to have sufficient expansion/contraction area during charging and discharging.⁵¹ This enhances the electrochemical performance of the active electrode material. However, rGO cannot be seen clearly in the materials, which may be because it combines with Pr₂O₃ and Co₃O₄ to form a mesh structure that makes it difficult to distinguish between them.

Fig. 4 SEM of (a) bare NF, (b) rGO/NF, (c) Pr₂O₃/NF, (d) Pr₂O₃/rGO/NF, (e) Co₃O₄/NF, (f) Co₃O₄/rGO/NF, (g) Pr₂O₃/Co₃O₄/NF and (h and i) Pr₂O₃/Co₃O₄/rGO/NF the (i) is the plot of Pr₂O₃/Co₃O₄/rGO/NF magnification.

Pr₂O₃/Co₃O₄/rGO TEM images are displayed in Fig. 5a. The thin sheet shape of PCGN can be seen in the figures, which is in line with the SEM images that were captured. This morphology aids in increasing the electrode's conductivity and contact area with the electrolyte, and it is assumed that a comparable morphology aids in the electrode's energy storage and conversion rate. HRTEM imaging was used to further evaluate the composition of the PCGN, and Fig. 5c and d make it abundantly evident that the material has good textural qualities. Among these, the lattice spacings of 0.2303 nm and 0.2739 nm, which correspond to the (−1 1 3) and (3 1 1) crystal planes of Pr₂O₃, are depicted in Fig. 5c and d. The lattice spacing of 0.1556 nm in Fig. 5c corresponds to the (5 1 0) crystal planes of Co₃O₄. The results of TEM are in agreement with those of XRD, Raman spectroscopy, and XPS, all of which demonstrate that the electrode material is composed of Pr₂O₃, Co₃O₄, and rGO.

Fig. 5 (a) TEM, (b) SEAD pattern of Pr₂O₃/Co₃O₄/rGO/NF, (c and d) HRTEM images, the inset of (a) is the magnified view of nanostructures.

The nitrogen adsorption–desorption isotherms can be used to calculate the precise surface area and pore size of electrode material. It can be seen from the trends of the adsorption curves and the wedge-shaped openings created by the adsorption–desorption curves in Fig. 6a and b that the electrode materials exhibit a strong H3-type hysteresis line at high relative pressures, indicating that the isotherms have a typical type IV behavior.⁵²Fig. 6a shows the adsorption–desorption curve of Pr₂O₃/Co₃O₄/rGO. The sample Pr₂O₃/Co₃O₄/rGO has adsorption pores that are primarily concentrated between 6 nm and 16 nm, as can be observed by combining the pore size distribution map with the data report. The BET model estimates that the electrode material has a specific surface area of 83.85 m² g⁻¹, while the specific surface area of Pr₂O₃/Co₃O₄ is about 69.77 m² g⁻¹. Combined with SEM and TEM images, it can be seen that the specific surface area of the electrode material itself is larger, and the addition of GO makes Pr₂O₃ and Co₃O₄ more uniformly dispersed and changes their morphological structure, which not only provides a large number of channels for a charge but also helps the rapid transfer of electrons, so that the electrode can be more fully in contact with the electrolyte, which in turn makes the electrode showed better electrochemical performance.

Fig. 6 Nitrogen adsorption and desorption isotherm of (a) Pr₂O₃/Co₃O₄/rGO and (b) Pr₂O₃/Co₃O₄ with the pore size distribution in the insets respectively.

3.2. Electrochemical testing and analysis

To investigate the electrode performance of the developed electrode materials, cyclic voltammetry (CV), galvanostatic charging and discharging (GCD), and electrochemical impedance spectroscopy (EIS) tests were performed using 6 mol L⁻¹ KOH as the electrolyte in a three-electrode system. Results as shown in Fig. 7a–g.

Fig. 7 (a and b) CV curves of Pr₂O₃/Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF electrodes under different scan rates; (c) comparison CV curves of Pr₂O₃/Co₃O₄/rGO/NF, Pr₂O₃/Co₃O₄/NF, Pr₂O₃/rGO/NF, Co₃O₄/rGO/NF, rGO/NF and bare NF electrodes at 5 mV s⁻¹; (d and e) GCD curves of Pr₂O₃/Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF electrodes at different current densities; (f) comparison GCD curves of Pr₂O₃/Co₃O₄/rGO/NF, Pr₂O₃/rGO/NF, Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF electrodes at 1 A g⁻¹; (h) specific capacitances of Pr₂O₃/Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF electrodes at different current densities; (i) the coulombic efficiencies of Pr₂O₃/Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF at different current densities; (g) the Nyquist plots of Pr₂O₃/Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF electrodes, the circuit inset is the high-frequency region.

All sample's CV curves had the potential windows between −0.2 and 0.5 V; the scan rate was 5–100 mV s⁻¹. Fig. 7a depicts the CV curves of the target electrode, and the overall shape of all curves showed evident redox peaks, showing that the specific capacitance of the electrode was primarily caused by pseudocapacitance behavior. The oxidation peak potential moves to a higher level as the scan rate rises, while the reduction peak potential moves to a lower level.¹⁹ All of these changes are accompanied by an increase in the peak current of the CV curves. This may be due to the fact that an increase in scan rate causes a rise in electrode surface current, which causes the active material's reaction to be incomplete and the reaction's reversibility to diminish.¹⁸ The fact that the shape of the CV curve is mostly unaffected by changes in scan rates. This shows the Pr₂O₃/Co₃O₄/rGO/NF electrode has good reversible and stable.

The possible charge storage mechanisms of Pr₂O₃ and Co₃O₄ in alkaline electrolytes are as follows.^31,33

Co₃O₄ + OH⁻ + H₂O ⇌ 3CoOOH + e⁻

CoOOH + OH⁻ ⇌ CoO₂ + H₂O + e⁻

Pr₂O₃ + H₂O ⇌ 2PrOOH

PrOOH + 2OH⁻ ⇌ PrO(OH)₃⁺ + 3e⁻

Therefore, the redox peaks in the CV curves can be attributed to the conversion of cobalt and praseodymium in different oxidation states. Fig. 7c depicts a comparison plot of CV curves for several electrodes of Pr₂O₃/rGO/NF, Co₃O₄/rGO/NF, Pr₂O₃/Co₃O₄/rGO/NF, rGO/NF, NF, Pr₂O₃/Co₃O₄/NF, where the CV curves of rGO/NF and NF have relatively small areas and contribute less to capacitance. Comparing the electrochemical properties before and after addition, the CV curves of Pr₂O₃/Co₃O₄/rGO/NF electrodes show lower redox peak currents and relatively larger enclosed areas, demonstrating that nanocomposite samples have better specific capacitance than single samples. This honeycomb structure created by the composite electrode is more conducive for the ions in the electrolyte to enter the electrode than the sheet or wire structure alone. The electrochemical performance before and after the addition of graphene also demonstrates that rGO increases the number of active sites in the electrode material and enhances the electrode's electrochemical performance.

The GCD curves for the electrode materials are shown in Fig. 7d, a voltage window of −0.1 V to 0.4 V was used with current densities ranging from 1 A g⁻¹ to 10 A g⁻¹. The Pr₂O₃/Co₃O₄/rGO/NF electrodes can be seen to have unique plateau areas during charging and discharging at various current densities, which is a feature of pseudocapacitor supercapacitors and is consistent with the CV curves. The duration of the plateau grows with decreasing current density, and its presence shows the occurrence of redox processes during charging and discharging. The charging and discharging time are the longest at the current density of 1 A g⁻¹ in the diagram, and according to the formula for calculating specific capacitance, that specific capacitance is 3316 F g⁻¹. For the current densities of 2, 4, 6, 8, and 10 A g⁻¹, the corresponding specific capacitances are 3197 F g⁻¹, 2981 F g⁻¹, 2798 F g⁻¹, 2614 F g⁻¹, and 2498 F g⁻¹. It has been discovered that as the current density increases, the specific capacitance of the Pr₂O₃/Co₃O₄/rGO/NF electrode dramatically decreases. The primary reason is that when the current density is low, the active material can be fully redoxed, resulting in a higher capacitance; as the current density increases, however, the rate of redox reaction increases, while the movement of ions is constrained by the diffusion rate; as a result, the redox reaction on the electrode can only be carried out on the electrode surface, and the specific capacitance decreases.

Fig. 7f shows the CGD curves of several electrodes at 1 A g⁻¹, and it is clear that the Pr₂O₃/Co₃O₄/rGO/NF electrode has a considerably higher specific capacitance than the other sand is 1.5 times higher than that of the Pr₂O₃/Co₃O₄/NF electrode (2120 F g⁻¹), indicating that the addition of graphene and other metal oxides facilitates the improvement of the electrode performance.

The coulombic efficiency (η) of the electrode can be calculated by the following equation.

where the discharge and charging times, respectively, are denoted by t_d (s) and t_c (s). The electrode's coulombic efficiency is displayed in Fig. 7i, and it is demonstrated that the pseudocapacitance reaction is very well reversible, demonstrating the good electrochemical properties of the electrode material.

Electrochemical impedance spectroscopy (EIS) studies were carried out to further understand the performance of the Pr₂O₃/Co₃O₄/rGO/NF electrode. In the EIS test, the stabilization potentials of Pr₂O₃/Co₃O₄/NF and Pr₂O₃/Co₃O₄/rGO/NF were 0.007 V and −0.038 V, respectively. When comparing the Pr₂O₃/Co₃O₄/rGO/NF and Pr₂O₃/Co₃O₄/NF Nyquist plots, it can be seen that both images can be split into two categories: the high-frequency region and the low-frequency region. The initial impedance (R_s) is where the curve in the high-frequency region meets the horizontal axis; the charge transfer resistance (R_ct) is where the arc diameter is in the high-frequency region, and the slope of the line in the low-frequency region is a crucial parameter for calculating the Warburg impedance resulting from proton diffusion.¹⁸

In Fig. 7g, it can be shown that the Pr₂O₃/Co₃O₄/NF electrode's semicircular arc direction in the high-frequency area is significantly bigger than that of the Pr₂O₃/Co₃O₄/rGO/NF electrode, indicating that the resistance of Pr₂O₃/Co₃O₄/rGO/NF is lower than that of Pr₂O₃/Co₃O₄/NF. Pr₂O₃/Co₃O₄/rGO/NF has a slope that is steeper than Pr₂O₃/Co₃O₄/NF in the low-frequency region, which suggests that the it has a lower Warburg impedance resistance and has a faster ion diffusion rate as well as lower electrolyte diffusion resistance. This may be related to the honeycomb structure of the Pr₂O₃/Co₃O₄/rGO/NF electrode The combination of rGO can significantly increase electrical conductivity, according to all the studies above.

Two identical electrodes were employed as the cathode and anode, respectively, and a sheet of filter paper was used as the septum to create a supercapacitor by utilizing the high capacitance and strong multiplication capability of Pr₂O₃/Co₃O₄/rGO/NF electrodes. Electrochemical studies were also carried out using 6 mol L⁻¹ KOH as the electrolyte to gauge the performance of the symmetric supercapacitor.

According to Fig. 8a–g, the supercapacitor (SC) device's test parameters were that scan rate of 5–100 mV s⁻¹, and a voltage window between −0.6 V and 0.6 V for the CV curve; a voltage window of −0.6 V to 0.6 V, and a current density of 0.5–20 A g⁻¹ for the GCD curve; and parameters for the EIS that ranged from 100 kHz to 0.01 Hz. The performance of the composed SC is depicted in Fig. 8a–g. In Fig. 8a the CV plot of assembled SC device is nearly rectangular and devoid of redox peaks, showing good multiplicity performance and optimum capacitance performance.²⁸ The triangular shape of the GCD curve in Fig. 8b, which is symmetrical, shows that the device has good capacitive performance. From Fig. 8d, it was determined that the specific capacitance values of SC devices were 373.9, 355, 323.3, 290.3, 266, 260.7, 252.5, and 196.7 F g⁻¹ for 0.5–20 A g⁻¹.

Fig. 8 (a) CV curves of the symmetric supercapacitor under different scan rates; (b) GCD curves of the supercapacitor at different current densities; (c) the Nyquist plot of the symmetric supercapacitor before and after 35 000 cycles, the inset is the plot in the high frequency region; (d) specific capacitances of the supercapacitor at different current densities; (e) the Coulombic efficiencies of the supercapacitor at different current densities; (f) Ragone plot related to power and energy densities of the supercapacitor; (g) the capacitance retention at a current density of 2 A g⁻¹ for 35 000 cycles; (h and i) the GCD curves for the first 10 cycles and the last 10 cycles out of 35 000 cycles.

It is evident that as the scan rate is increased, the specific capacitance of the SC likewise drops. Only the outer surface of the electrode can be employed for charge storage since the diffusion rate limits the passage of ions at higher scan rates.

In addition, it is necessary to verify the supercapacitor's cycle stability since the amount of time it is used is a crucial factor in determining how well it performs. The initial 10 cycles and the last 10 cycles’ photos do not differ significantly, as seen in Fig. 8h and i. The capacitance retention of the device diminishes as the number of charge/discharge cycles rises. The supercapacitor's specific capacitance retention rate is still above 82% after 35 000 cycles of charging and discharging at a current density of 2 A g⁻¹, demonstrating the constructed supercapacitor's strong cycling stability. The material is not completely activated at the start of the cycle, but after repeated charge and discharge cycles, a slight rise in specific capacitance occurs, which may be caused by the electrodes becoming activated throughout the constant charging and discharging process, exposing additional active sites.^53,54 But the active electrode material undergoes volume expansion and is partially detached during prolonged cyclic charging and discharging; this causes an increase in resistance and a slight decrease in capacitor retention.⁵⁵ This is why the curve falls, then rises, then falls again.

In Fig. 8c, the Nyquist plot for the SC device during the loop test is displayed. R_s does not significantly alter as the number of cycles rises. R_ct rises after 35 000 cycles, which may be a result of longer charging and discharging times that cause some active electrode materials to shed, while electrode material volume expansion impedes electrolyte diffusion and raises resistance.¹⁷ The resistance change is comparatively minor after the number of cycles, which shows that the SC device has strong stability.

Meanwhile, it is calculated that when the power density of SC is, its energy density is, and the energy density and power density values obtained from this experiment are compared with the recently reported related studies, and the energy density and power density of the symmetric supercapacitor are found to be excellent.

Using the equation, it is determined that SC has an energy density of 74.8 W h kg⁻¹ at a power density of 300 W kg⁻¹. Even at a power density of 12 000 W kg⁻¹, SC also has an energy density of 39.3 W h kg⁻¹. When comparing the results of our experiment's energy and power density measurements to those of recently published research in the same field, it is found that the symmetric super symmetrical supercapacitor has good stability (Table 1).

Table 1 Comparison of electrochemical performances of Pr₂O₃/Co₃O₄/rGO/NF composite electrode in our work with previous reports

Electrode materials	Specific capacitance	Cycling performance	Energy density at power density	Ref.
Co3O4 nanoball/carbon aerogel	350 F g^−1 at 1 A g^−1	210% (6000 cycles)	23.82 W h kg^−1 at 95.96 W kg^−1	56
Co3O4/rGO	865 F g^−1 at 1 A g^−1	93.2% (5000 cycles)	No data	16
CuO/Co3O4	806.25 F g^−1 at 2 A g^−1	99.75% (2000 cycles)	71.66 W h kg^−1 at 800 W kg^−1	57
Co3O4/CeO2	1037.5 F g^−1 at 2.08 A g^−1	94.4% (5000 cycles)	No data	58
Co3O4/MnO2/Co(OH)2	3022.2 F g^−1 at 10 A g^−1	87.9% (5000 cycles)	11.4 W h kg^−1 at 666.7 W kg^−1	18
Pr-doped Co3O4 composites	1625 F g^−1 at 2 A g^−1	98% (5000 cycles)	42.3 W h kg^−1 at 240 W kg^−1	32
NiFeCoPrO–Au/NF	1792 F g^−1 at 10 A g^−1	99.5% (30 000 cycles)	20.94 W h kg^−1 at 589 W kg^−1	34
Pr6O11@NiCo2O4	1635 F g^−1 at 0.5 mA cm^−2	82% (1000 cycles)	6.6 W h kg^−1 at 150 W kg^−1	33
PANI–Pr2O3–NiO–Co3O4	905 F g^−1 at 1 A g^−1	No data	87.99 W h kg^−1 at 2.6 kW kg^−1	35
Pr2O3/Co3O4/rGO/NF	3197 F g^−1 at 1 A g^−1	91.8% (5000 cycles)	74.8 W h kg^−1 at 300 W kg^−1	This work
		82% (35 000 cycles)

---

4. Conclusions

In conclusion, using a straightforward hydrothermal method to load Pr₂O₃, Co₃O₄, and graphene onto nickel foam and annealing the materials, we successfully created honeycomb-structured Pr₂O₃/Co₃O₄/rGO/NF electrode materials. According to an XRD test, Pr₂O₃ is in the monoclinic phase while Co₃O₄ is in the cubic phase. According to the study, the composite of Pr₂O₃, Co₃O₄ and rGO has superior electrochemical characteristics, and the inclusion of graphene can significantly reduce the build-up issue and increase electrical conductivity. Ion diffusion is enhanced by the material's honeycomb structure, which also offers enough room for expansion and contraction. The developed composite electrode material exhibits an exceptional specific capacitance of 3316 F g⁻¹ at a current density of 1 A g⁻¹. Furthermore, a symmetrical supercapacitor made of Pr₂O₃/Co₃O₄/rGO/NF performs admirably, with a specific capacitance of 373.9 F g⁻¹ at 0.5 A g⁻¹ and an energy density of 74.8 W h kg⁻¹ at a power density of 300 W kg⁻¹. In terms of stability, the current density of 2 A g⁻¹ retains 82% of its charge after 35 000 cycles. This research sheds new light on the selection of supercapacitor materials.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We gratefully acknowledge the National Natural Science Foundation of China (No. 21271027) for their support to this work.

References

1. J. R. Miller and P. Simon, Science, 2008, 321, 651–652.
2. M. Ren, J. Di and W. Chen, Batteries Supercaps, 2021, 4, 1279–1290.
3. Y. Zhou, P. Jin, Y. Zhou and Y. Zhu, Sci. Rep., 2018, 8, 9005.
4. V. Augustyn, P. Simon and B. Dunn, Energy Environ. Sci., 2014, 7, 1597–1614.
5. Y. Jiang and J. Liu, Energy Environ. Mater., 2019, 2, 30–37.
6. Y. Wang, Y. Song and Y. Xia, Chem. Soc. Rev., 2016, 45, 5925–5950.
7. N.-S. Choi, Z. Chen, S. A. Freunberger, X. Ji, Y.-K. Sun, K. Amine, G. Yushin, L. F. Nazar, J. Cho and P. G. Bruce, Angew. Chem., Int. Ed., 2012, 51, 9994–10024.
8. R. Kötz and M. Carlen, Electrochim. Acta, 2000, 45, 2483–2498.
9. P. Simon and Y. Gogotsi, Nat. Mater., 2008, 7, 845–854.
10. B. Liu, Y. Liu, H. Chen, M. Yang and H. Li, ACS Sustainable Chem. Eng., 2019, 7, 3101–3110.
11. S. Li, Y. Duan, Y. Teng, N. Fan and Y. Huo, Appl. Surf. Sci., 2019, 478, 247–254.
12. Y. Wang, D. Gu, J. Guo, M. Xu, H. Sun, J. Li, L. Wang and L. Shen, ChemElectroChem, 2020, 7, 928–936.
13. M. B. Askari, S. M. Rozati, P. Salarizadeh and S. Azizi, Ceram. Int., 2022, 48, 16123–16130.
14. R. Zou, B. Wang, L. Zhu, L. Yan, F. Shi, Y. Sun, B. Shao, S. Zhang and W. Sun, Diamond Relat. Mater., 2022, 126, 109060.
15. Y. Wang, J. Zhou, Z. Zhou, H. Lv, B. Gu, K. Wang, Z. Chen, X. Yan, J. Zhang, W.-W. Liu and Y.-L. Chueh, Nanoscale, 2021, 13, 15431–15444.
16. M. Aadil, M. F. Warsi, P. O. Agboola, M. F. Aly Aboud and I. Shakir, Ceram. Int., 2021, 47, 9008–9016.
17. Z. Ji, K. Liu, N. Li, H. Zhang, W. Dai, X. Shen, G. Zhu, L. Kong and A. Yuan, J. Colloid Interface Sci., 2020, 579, 282–289.
18. X. Li, R. Miao, B. Tao, F. Miao, Y. Zang and P. K. Chu, Solid State Sci., 2019, 95, 105941.
19. H. Chen, J. Wang, F. Liao, X. Han, Y. Zhang, C. Xu and L. Gao, Ceram. Int., 2019, 45, 11876–11882.
20. W. Gu and G. Yushin, Wiley Interdiscip. Rev.: Energy Environ., 2014, 3, 424–473.
21. G. Xiong, C. Meng, R. G. Reifenberger, P. P. Irazoqui and T. S. Fisher, Electroanalysis, 2014, 26, 30–51.
22. Q. Li, X. Hu, Q. Yang, Z. Yan, L. Kang, Z. Lei, Z. Yang and Z. Liu, Electrochim. Acta, 2014, 119, 184–191.
23. G.-J. Liu, L.-Q. Fan, F.-D. Yu, J.-H. Wu, L. Liu, Z.-Y. Qiu and Q. Liu, J. Mater. Sci., 2013, 48, 8463–8470.
24. S. Arunachalam, B. Kirubasankar, D. Pan, H. Liu, C. Yan, Z. Guo and S. Angaiah, Green Energy Environ., 2020, 5, 259–273.
25. H. Zhao, J. Xia, D. Yin, M. Luo, C. Yan and Y. Du, Coord. Chem. Rev., 2019, 390, 32–49.
26. H. Huang and J.-J. Zhu, Analyst, 2019, 144, 6789–6811.
27. J. Theerthagiri, G. Durai, T. Tatarchuk, M. Sumathi, P. Kuppusami, J. Qin and M. Y. Choi, Ionics, 2020, 26, 2051–2061.
28. Y. Li, B. Guan, A. Maclennan, Y. Hu, D. Li, J. Zhao, Y. Wang and H. Zhang, Electrochim. Acta, 2017, 241, 395–405.
29. H. Zhang, J. Gu, J. Tong, Y. Hu, B. Guan, B. Hu, J. Zhao and C. Wang, Chem. Eng. J., 2016, 286, 139–149.
30. M. Majumder, R. B. Choudhary, A. K. Thakur and I. Karbhal, RSC Adv., 2017, 7, 20037–20048.
31. D. B. Bailmare, P. Tripathi, A. D. Deshmukh and B. K. Gupta, Sci. Rep., 2022, 12, 3084.
32. D. Yang, M. Xu, X. Liang, J. Wang, W. Fang, C. Zhu and F. Wang, Electrochim. Acta, 2022, 406, 139815.
33. N. Chen, A. Younis, S. Huang, D. Chu and S. Li, J. Alloys Compd., 2019, 783, 772–778.
34. D. Zhang, L. Peng, Z. Yang, Y. Yang and H. Li, Inorg. Chem., 2019, 58, 15841–15852.
35. M. Naveed Ur Rehman, T. Munawar, M. S. Nadeem, F. Mukhtar, A. Maqbool, M. Riaz, S. Manzoor, M. N. Ashiq and F. Iqbal, Ceram. Int., 2021, 47, 18497–18509.
36. K. T. Kubra, A. Javaid, B. Patil, R. Sharif, A. Salman, S. Shahzadi, S. Siddique and S. Ghani, Ceram. Int., 2019, 45, 6819–6827.
37. T. Munawar, S. Yasmeen, M. Hasan, K. Mahmood, A. Hussain, A. Ali, M. I. Arshad and F. Iqbal, Ceram. Int., 2020, 46, 11101–11114.
38. Y. Li, W. Qiu, F. Qin, H. Fang, V. G. Hadjiev, D. Litvinov and J. Bao, J. Phys. Chem. C, 2016, 120, 4511–4516.
39. D. Mandal, P. Routh and A. K. Nandi, ChemistrySelect, 2017, 2, 3163–3171.
40. B. Xiao, K. Zhao, L. Zhang, T. Cai, X. Zhang, Z. Wang, J. Yuan, L. Yang, P. Gao and D. He, Catal. Commun., 2018, 116, 1–4.
41. J. Li, G. Lu, G. Wu, D. Mao, Y. Guo, Y. Wang and Y. Guo, Catal. Sci. Technol., 2014, 4, 1268–1275.
42. J. E. Slimak, A. Sethi, T. Kolodiazhnyi and S. L. Cooper, Phys. Rev. Res., 2020, 2, 043169.
43. H. Al Kutubi, L. Rassaei, W. Olthuis, G. W. Nelson, J. S. Foord, P. Holdway, M. Carta, R. Malpass-Evans, N. B. McKeown, S. C. Tsang, R. Castaing, T. R. Forder, M. D. Jones, D. He and F. Marken, RSC Adv., 2015, 5, 73323–73326.
44. S. Xiong, C. Yuan, X. Zhang, B. Xi and Y. Qian, Chemistry, 2009, 15, 5320–5326.
45. X. Ren, H. Fan, J. Ma, C. Wang, M. Zhang and N. Zhao, Appl. Surf. Sci., 2018, 441, 194–203.
46. S. Zhou, Z. Ye, S. Hu, C. Hao, X. Wang, C. Huang and F. Wu, Nanoscale, 2018, 10, 15771–15781.
47. H. Ogasawara, A. Kotani, R. Potze, G. A. Sawatzky and B. T. Thole, Phys. Rev. B: Condens. Matter Mater. Phys., 1991, 44, 5465–5469.
48. H. Zhang, Z. Mo, R. Guo, N. Liu, M. Yan, R. Wang, H. Feng and X. Wei, J. Mater. Res., 2019, 34, 1200–1209.
49. W. Jia, J. Li, Z. Lu, Y. Juan and Y. Jiang, Materials, 2018, 11, 1560–1580.
50. J. Mei, W. Fu, Z. Zhang, X. Jiang, H. Bu, C. Jiang, E. Xie and W. Han, Energy, 2017, 139, 1153–1158.
51. O. Moradlou, H. Ansarinejad, M. Hosseinzadeh and H. Kazemi, J. Alloys Compd., 2018, 755, 231–241.
52. C.-B. Wang, C.-W. Tang, S.-J. Gau and S.-H. Chien, Catal. Lett., 2005, 101, 59–63.
53. L. Sun, Z. Xie, A. Wu, C. Tian, D. Wang, Y. Gu, Y. Gao and H. Fu, J. Mater. Chem. A, 2021, 9, 26226–26235.
54. Q. Cheng, J. Tang, J. Ma, H. Zhang, N. Shinya and L.-C. Qin, Phys. Chem. Chem. Phys., 2011, 13, 17615–17624.
55. L. Zhang, K. N. Hui, K. San Hui and H. Lee, J. Power Sources, 2016, 318, 76–85.
56. M. Zomorodian Esfahani, A. Aghaei, M. Khosravi, N. Bagheri, Z. Khakpour and M. Javaheri, New J. Chem., 2017, 41, 11731–11741.
57. H.-J. Kim, S. Y. Kim, L. J. Lim, A. E. Reddy and C. V. V. M. Gopi, New J. Chem., 2017, 41, 5493–5497.
58. J. Cui, X. Zhang, L. Tong, J. Luo, Y. Wang, Y. Zhang, K. Xie and Y. Wu, J. Mater. Chem. A, 2015, 3, 10425–10431.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nj05192c

---