Study on mechanical and electrical properties improvement of insulating paper modified by cellulose nanocrystals

The performance of oil–paper insulation is crucial for ensuring the safety and reliability of high-voltage power equipment. Therefore, enhancing its performance is of utmost importance to maintain the reliability of the power system.¹ Factors, such as mechanical and dielectric properties, are key factors that need to be improved to meet the demand for enhanced performance and ensure the long-term reliability of power equipment.

Recent studies have investigated the use of inorganic nanoparticles for enhancing the properties of insulating paper,² with researchers exploring homogeneous bulk doping methods to achieve these improvements.³ For instance, the addition of 1 weight percent (wt. %) of nano-SiO₂ increased the tensile strength and the breakdown field strength at power frequency,⁴ while nano-TiO₂ modified insulating paper showed reduced accumulation of space charge and damage caused by repeated impulses.^5,6 The incorporation of both nano-SiO₂ and ZnO into insulating paper led to the enhancement in dielectric properties, with the greatest observed with a 1 wt. % addition of ZnO.⁷ Additionally, molecular simulation and experimental analysis confirmed that nanoparticles, such as nano-Al₂O₃ and nano-SiO₂, can enhance the mechanical properties of cellulose insulating paper.^8,9 In summary, nanoparticle doping is an efficient approach for improving properties of insulating paper.

Although inorganic nanoparticles have been studied as common additives in insulating materials, there are also some drawbacks to using them. The doping process requires coupling agents for interaction with cellulose in insulating paper, which increases costs and environmental impact.¹⁰ Inorganic particle dispersion in cellulose paper is often unsatisfactory and may easily form aggregates under electric field forces, negatively impacting other properties of the material.

Nanocellulose, however, has arisen as a potential alternative due to its high compatibility with fibers in cellulose paper,¹⁰ environmentally friendly properties,¹¹ exceptional dispersibility and stability, and low concentration requirement or achieving desired outcomes. It can be effective in improving the performance of paper, either as a direct substitute or as an additive or a coating agent.^12,13 Nanocellulose also contributes to enhancing mechanical properties and chemical stability, making it an ideal additive for insulating paper. There has been some progress in the preparation of insulating paper using nanocellulose, and the pure nanocellulose insulating paper obtained from the preparation performs well in terms of mechanical properties,^14,15 as well as breakdown characteristics,¹⁶ but the cost is too high. Zhou et al. found that the tensile strength of cationic nanocellulose-modified insulating paper was improved by 24.77%, and the DC breakdown field strength was improved by 21%,¹⁷ but the doping of cationic nanocellulose increased the dielectric loss of the modified paper. Chen et al. used citric acid cross-linking and melamine amide-modified cellulose nanocrystals to prepare insulating paper, and the tensile strength was improved by 17.30%, and the AC breakdown field strength was enhanced by 40.28%.¹⁸ These works show the great potential of nanocellulose in the field of insulation paper modification, but there are still some problems. The mechanism of nanocellulose modification of insulating paper is still not clear enough, and issues, such as how to determine the optimal concentration of doping, still need to be studied in depth.

Cellulose nanocrystals (CNCs) are a type of nanocellulose possessing a rod-like shape with higher crystallinity and mechanical strength relative to other forms of nanocellulose, such as nanofibrillated cellulose (NFC) or bacterial cellulose. Thus, CNC-modified materials exhibit superior durability, deformation resistance, and stability compared to other nanocellulose-modified materials. CNC can self-assemble into a cholesteric liquid crystal structure in aqueous dispersion at a certain concentration, which is also retained after drying into a film. Although the application of CNC for insulation paper modification is not yet widespread, it still shows great potential.¹⁹ 

In this paper, a new doping modification method based on CNC for improving the performance of insulating paper is proposed. The study comprehensively investigates the microstructure, moisture absorption, mechanical properties, breakdown behavior, and dielectric response of the doped insulating paper, providing a theoretical explanation for the observed improvements. Overall, the findings of this study support the use of CNC as an effective additive for enhancing the performance of insulating paper, with the potential to improve the safety and reliability of power transformers.

The insulating paper pulp was obtained from the unbleached coniferous wood pulp board without any bleaching agent. CNC with a diameter of 20–30 nm and a length of 500–600 nm, obtained from ScienceK Co. Ltd., Huzhou, China, through weak acid hydrolysis, was used in all experiments. The concentration of the CNC solution was 5 wt. %. All experiments were conducted using de-ionized water and unpurified chemicals. The 25# transformer insulating oil was purchased from Liaoning Furun Lubricating Oil Co., Ltd. (China) for impregnation of insulating paper.

First, an unbleached coniferous pulp cardboard was weighed, soaked in de-ionized water for 15 min, and then torn into small pieces and put into the pulp tank of a Wally pulper for soaking. The beating degree of the pulp was measured regularly during pulping, and pulping was stopped when the beating degree reached between 40°SR (Schopper–Riegler number) and 60°SR. A certain amount of CNC was then weighed and ultrasonicated and poured into the dissociated pulp. Mechanical stirring was performed for 5 min, and the speed of the stirrer was set between 400 and 500 r/min. The treated pulp was placed in a sheet former, and the water was extracted after pneumatic stirring to obtain round wet paper sheets with a diameter of about 200 mm. Then, the wet paper sheet was subjected to two-stage pressing treatment to improve the density of the wet paper sheet. Finally, the sheets are vacuum dried, as shown in Fig. 1. The doping concentration of CNC is set to 0, 1, 3, 6, and 10 wt. %. First, the maximum concentration was set at 10 wt. %, which was determined after referring to other studies and pre-experiments. At concentrations higher than 10%, the agglomeration in the insulating paper is serious and the performance enhancement is not obvious with the increase of concentration. At low concentrations, the concentration gradient was set denser in order to better study the effect of interfacial interaction on performance enhancement.

The paper samples were first dried in a vacuum oven at 105 °C for 24 h, and the insulating oil was dried in a vacuum oven at 90 °C for 48 h. The paper samples were then impregnated with insulating oil for 48 h, stored in the insulating oil, and used for electrical performance testing.

The morphology of CNC is observed by a transmission electron microscope (TEM) using JEM-2100F. The micromorphology of the modified insulating paper was observed using a scanning electron microscope (EVO 10) after gold coating. The acceleration voltage used was 15 kV. The adsorption–desorption isotherms of the samples to water vapor were tested using a vacuum vapor/gas sorption analyzer (BSD-VVS) at 25 °C and 17 relative pressure points. Porosity was tested by a mercury intrusion method using MicroActive AutoPore V9600 equipment. The Fourier transform infrared (FTIR) spectra were recorded in a reflection mode by Thermo Scientific Nicolet IN10 and IZ10.

The tensile strength and Young's modulus of the samples were characterized through a mechanical strength tester (IMT-Tensile 01) and a microcomputer-controlled electronic universal testing machine (5 kN). The length of the sample used for the test is 100 mm and the width is 15 mm. The distance between fixtures is 50 mm, and the separation speed of the fixtures is 100 mm/min. Each sample was tested at least ten times.

The AC (alternating current) and DC (direct current) breakdown strength of the sample were measured according to IEC 60243 using 50 mm diameter brass circular plate electrodes and 25 mm diameter brass spherical electrodes, respectively. The diameter of samples was greater than 55 mm to avoid surface flashover. The frequency-domain dielectric spectrum of the sample was tested using a DIRANA dielectric response analyzer, and the wiring diagram was shown in Fig. 2. The test temperature was 40 °C, and the test frequency range was 1 mHz to 5 kHz. Measurements were made in 25# transformer oil, and the average results of two measurements were taken for each sample.

Figure 3 shows the transmission electron microscope (TEM) diagram of the CNC. As can be seen from the figure, the length of the CNC is in the range of 500–600 nm, the diameter is in the range of 20–30 nm, and the L/D ratio is around 20.

Scanning electron microscope (SEM) images of the hand sheet prepared in the laboratory and a commercial insulating paper made by Miri Tokushu Paper MFG. Co., Ltd. were obtained and compared, as shown in Fig. 4. The fiber widths and microstructures of the two samples were found to be similar, suggesting that the preparation method used in the laboratory can represent the properties of commercial insulating paper to some extent. This indicates that the results obtained from the laboratory-based doping modification method can provide useful insights for the development of commercial insulating paper.²⁰ 

The SEM images in Fig. 5 provide insights into the impact of CNC doping on the morphology of insulating paper. The reduction in the pore size observed upon doping could be attributed to the ability of CNC to fill the inter-fiber voids in the paper matrix.

At lower concentrations, the distribution of CNC was relatively uniform, indicating that the doping process was successful in dispersing the CNC evenly throughout the paper matrix. The uniform distribution of CNC at low concentrations may be related to their ability to disperse more readily in the aqueous suspension during the papermaking process. This can lead to a more homogeneous distribution of particles within the paper structure, which may be beneficial for achieving better performance properties. However, when the concentration of doped CNC reached 10 wt. %, some nanoparticle agglomeration was observed, which might negatively affect the performance of the paper. Hence, there is an optimal range of CNC doping concentration that enhances the performance of insulating paper.

Figure 6 shows the pore size distribution of unmodified insulating paper and 3 wt. % doped modified insulating paper obtained by measuring porosity by a mercury intrusion method. The 3 wt. % doped insulating paper with nanofibrillated cellulose shows a decrease in porosity due to the introduction of nanofibrillated cellulose, which fills the voids between the fibers and reduces the number of large pores while increasing the proportion of fine pores.

FTIR spectra of modified insulating paper were presented in Fig. 7. The same functional groups were observed for the unmodified paper and the CNC-modified paper, indicating that the doping of CNC had basically no effect on the chemical structure of the insulating paper. Compared with the unmodified paper, a slightly higher wide peak of peak absorption intensity was observed at 3330 cm⁻¹ for the modified paper, with the highest absorption intensity at 3330 cm⁻¹ for the 3% CNC-modified paper. 3 wt. % CNC-modified paper had a peak height of 0.087 at 3330 cm⁻¹ and unmodified paper had a peak height of 0.065 at 3330 cm⁻¹. This suggested that doping the right amount of CNC increased the hydrogen bond density of the insulating paper.

Moisture is a common defect in oil–paper insulation that can pose a threat to its reliability, caused by aging and external intrusion.²¹ Hence, stringent requirements are placed on the moisture content of oil–paper insulation, with a residual moisture content of 0.2%–0.5% being the accepted range when leaving the factory. For controlling the water content of oiled paper insulation, the water vapor adsorption and desorption properties of insulating paper play a notable role. The gas adsorption process is categorized into single-layer and multi-layer adsorption based on the number of gas molecular layers that the material surface covers.²² The IUPAC recommendation of 1985 classified physical adsorption isotherms into six types. These models are based on the specific characteristics of the material surface and can provide insights into the water content control of oiled paper insulation.²³ 

In this study, the adsorption isotherms of water vapor on insulating paper were measured with different doping concentrations of CNC, shown in Fig. 8. The results showed that after doping, the adsorption of water vapor on insulating paper decreased slightly, but the difference between different concentrations was insignificant.

Insulating paper is highly hygroscopic, and the free hydroxyl groups in the fibers make it highly polar adsorbent. The more hygroscopic the insulating paper is, the higher the water content of the oil–paper insulation system will be as the operating time of the equipment increases. The higher the water content of the insulating paper, the higher the content of polar molecules in the paper and the increase in dielectric loss.

Although CNC doping introduces a large number of hydroxyl groups, the experimental results of the adsorption reduction indicate that these groups mainly form hydrogen bonds with the cellulose in the insulating paper, occupying some previously exposed hydroxyl groups on the cellulose of the insulating paper, and increasing the spatial density of the hydrogen bonds within the insulating paper. As a result, the moisture adsorption capacity of the insulating paper was slightly reduced after doping.

The water vapor adsorption isotherm of the modified insulating paper showed no clear inflection point at low relative pressures. At high relative pressures (0.6–1), the slope of the adsorption isotherm increased with increasing relative pressure, indicating multilayer adsorption with large aperture and relatively strong bonding between the adsorbent (i.e., the CNC-doped insulating paper) and the adsorbate (i.e., water vapor). The lack of a closed loop between the adsorption and desorption curves indicated a strong adsorption capacity of insulating paper for water vapor, making it difficult to desorb completely.

To fit the adsorption isotherms of the modified insulating paper, five points were selected within the relative pressure range of 0.3–0.7, which is within the applicable range of the BET equation. The obtained R² value exceeded 0.99, indicating that the BET model can effectively describe the multilayer adsorption behavior of the modified insulating paper (Fig. 9). The fitting parameters for the four adsorption isotherms are presented in Table I.

The maximum adsorption capacity (q_m) for a monolayer in samples with varying mass fractions ranged between 37.15 and 36.49 mg/g. Comparatively, the unmodified paper exhibited a marginally higher q_m for a monolayer, as opposed to the modified insulating paper. The adsorption behavior of each sample was found to be physically similar, as indicated by the similarity in the number of adsorption layers, which was approximately five to six layers.

CNC doping has a limited effect in reducing water vapor adsorption on insulating paper, which implies that the mechanism of reducing water vapor adsorption is intricate and may involve other factors. The high R² value indicates that the BET model can effectively describe the adsorption behavior of the modified insulating paper, providing valuable insights into the fundamental mechanism of moisture absorption in insulating materials. Future studies should investigate the detailed mechanism of CNC doping on water vapor adsorption and its potential application in improving the moisture resistance of insulating paper.

Insulating paper is a pivotal component in an electric power equipment's oil–paper insulation system, which necessitates specific mechanical properties. The mechanical properties of CNC-doped insulating paper with different mass fractions, expressed in tensile strength and Young's modulus, are presented in Fig. 10.

The addition of CNC enhances the mechanical strength of the insulating paper. Tensile strength increases with the doping concentration of CNC up to 6 wt. %, beyond which the tensile strength decreases. The highest tensile strength of 10.30 N/mm, a 6.91% increase, is achieved at the 6 wt. % doping concentration. Furthermore, even at 10 wt. % doping concentration, the tensile strength of the modified paper is still greater than that of the unmodified paper. The doping of CNC also leads to an increase in Young's modulus of the insulating paper with a peak increase of 11.98% observed at 10 wt. % doping. This increase becomes flat above 3 wt. % doping.

The mechanical strength of insulating paper is affected by multiple factors, including the bonding strength between fiber bundles, the strength of the fibers themselves, and the fiber arrangement. The fiber bundles of insulating paper are bonded to each other mainly by a hydrogen bond.²⁴ From the results of vapor adsorption and spectroscopy, it can be seen that the doped CNC forms hydrogen bonds with the original bare hydroxyl groups in the insulating paper, increasing the number of hydrogen bonds in the insulating paper. The mechanical failure process of CNC-modified insulating paper involves fiber breakage and a series of hydrogen bonding ruptures and reformations. During the successive shear-slip process, alignment and misalignment between the interfacial hydroxyl functional groups occur alternately, and the potential energy of the whole system oscillates between the steady value and the transition state.²⁵ The periodic breaking and reorganization of hydrogen bonds dissipates a large amount of energy. The doping of CNC leads to an increase in the number of hydrogen bonds between fiber bundles, which enhances the bonding strength between the fibers of the insulating paper and improves the level of stress transfer in the insulating paper. On the other hand, CNC itself has very high mechanical strength. Moreover, the length of fibers in the pulp prepared by pulping is basically in the millimeter or micrometer scale, which is much larger than that of CNC.²⁶ Therefore, doping CNC enriches the composition of fibers of different lengths in insulating paper. Also, it can serve to fill the voids between fiber bundles in the insulating paper and improve the stress transfer level of the insulating paper.

The Page model is a semi-empirical and intuitive model that considers both individual fiber strength and bonding strength between fibers when predicting paper tensile strength. According to the Page model, the addition of CNC within an appropriate concentration range can increase the bond strength of insulating paper, improve the shear bond strength between fibers (τ), and increase the size of the fiber relative bond area, leading to an enhancement in tensile strength. However, the substitution of CNC for long fibers reduces the proportion of long fibers that can withstand higher tensile strength per unit amount of sample. According to the Page model, this results in a rapid decrease in the average fiber slenderness ratio (Pl/A), which exceeds the strengthening effect of increased bonding and has an adverse effect on the tensile strength of the modified paper. Furthermore, at high CNC concentrations, CNCs tend to agglomerate, leading to stress concentration at agglomerate sites, which can also affect the tensile strength of the doping system.

In general, Young's modulus of insulating paper ranges from 0.5 to 4.0 GPa, while that of nanocellulose is about 100–140 GPa. In particular, Young's modulus of CNC can reach 150 GPa. Based on the Halpin–Kardos model, CNC added to the insulating paper caused Young's modulus of CNC-modified insulating paper to increase with the doping concentration. However, at higher doping concentrations, the overall Young's modulus is negatively impacted by nanoparticle agglomeration, limiting its improvement. Due to the decreased inter-particle spacing within CNC agglomerates, Young's modulus of the CNC agglomerates and overall stiffness of the samples decrease. Simultaneously, the reduced contact area between CNC agglomerates and the insulating paper weakens the stress transfer effect. Considering the adverse impacts of CNC aggregation and the positive effects of CNC doping on enhancing mechanical properties, the resulting increase in Young's modulus becomes less pronounced at higher doping concentrations.

The outcomes shown in Fig. 11, pertaining to the AC and DC breakdown strength of the modified insulating paper, demonstrate a similar trend for both types of breakdown strength. The breakdown voltages show a rising trend initially as the doping concentration increases, followed by a declining trend, and peaking at a doping concentration of 3 wt. %. At this concentration, the maximum AC breakdown field strength for CNC-modified paper reached 74.3 kV/mm, which is 28.53% higher than the unmodified paper. Additionally, the DC breakdown strength increased by 8.72%. The dispersion of the breakdown strength first decreased, then increased, as the doping concentration increased. In conclusion, these results suggest that a doping concentration of 3 wt. % is optimal for achieving maximum breakdown strength in insulating paper.

P(E) is the cumulative failure probability as a function of the breakdown field strength (E), with the shape parameter (β) and the characteristic breakdown field strength (α) considered, where α corresponds to the field strength at which P(E) equals 63.28%. The two-parameter Weibull distribution of DC breakdown field strength of insulating paper is shown in Fig. 12, and the parameters of the Weibull distribution are shown in Table II.

Incorporating CNC into insulating paper can increase its breakdown field strength due to the nanoparticles' large specific surface area. Doping CNC introduces additional interface traps into the paper, which can impede charge flow and electric field establishment. Interface traps can effectively capture charges, reducing the electric field and suppressing charge flow, which can lead to a higher breakdown field strength.

When CNC is doped into insulating paper, it can fill the small pores between the micro-fibers, as demonstrated by SEM results. This leads to an improvement in the breakdown performance when the paper is immersed in insulating oil since the relative permittivity of cellulose is higher than that of insulating oil.

However, excessive CNC doping can cause nanoparticles to agglomerate in the matrix, leading to a reduction in breakdown strength, due to an enhanced interfacial polarization effect in the paper.

The breakdown strength of paper insulation is influenced by the presence of internal pores, which can cause some degree of dispersion in test results. It can have an impact on the accuracy and reliability of measurements. By filling the internal pores, CNC can help reduce this dispersion and improve the consistency of breakdown strength measurements. However, excessive doping can cause agglomeration of nanoparticles in the paper, leading to uneven dispersion and increased variability of breakdown strength measurements.

The dielectric properties of insulating paper modified with varying mass fractions of CNC were investigated, as shown in Figs. 13 and 14. The FDS curve of the modified insulating paper shows a discernible loss peak in the frequency range of 1–10 Hz. Within the measured frequency range, the loss tangent (tan δ) decreases with an increasing mass fraction of CNC, including the loss peak. This trend reaches its maximum reduction at 3 wt. %. For CNC concentrations exceeding 3 wt. %, the reduction in tan δ approaches saturation. As the doping concentration increases, a decreasing trend in the real part of the complex permittivity (ɛ_r′) of the insulating paper can be observed in the mid-to-low frequency range (1 mHz–10 Hz), with the decrease saturating when it exceeds 3 wt. %.

Dipoles in cellulosic materials are randomly oriented and freely arranged with no permanent dipole moment. When CNC is exposed to an applied electric field, the dipoles are oriented in the direction of the electric field. The hydroxyl group on the side chain has both advantages and disadvantages. It provides high polarization while, on the other hand, the hydrogen bonding network created by the hydroxyl groups limits the polarization.³⁰ This implies that the doping of nanocellulose should be in a suitable concentration range so that the hydroxyl groups introduced by nanocellulose can participate in the formation of a hydrogen bonding network as much as possible, and the number of exposed hydroxyl groups can be reduced.

On the other hand, for the oil–paper insulation system, CNC doping reduces the interfacial polarization between the insulating oil and the insulating paper, thus suppressing the dielectric loss in the insulating paper. Filling pores with CNC reduces the interfacial area between the insulating oil and paper, leading to a decrease in interfacial polarization loss and relative permittivity. Furthermore, CNC filling enhances the cross-linking between cellulose fibers through interactions such as hydrogen bonding, forming a denser and more interconnected spatial network structure, which reduces the relative permittivity. CNC doping reduces moisture content, thereby reducing dielectric loss due to low-frequency conductive loss at low frequencies and interfacial polarization loss at medium-to-high frequencies caused by moisture.³¹ However, at concentrations exceeding 3 wt. %, CNC aggregation reduces the interaction with cellulose fibers in insulating paper, causing the trend of decreasing dielectric loss and relative permittivity to reach saturation. These results show that CNC doping is an effective method to reduce the dielectric loss and the dielectric constant in insulating materials and enhance the reliability of oil–paper insulation systems.

With the addition of nanocellulose, there is a very significant decrease in the amplitude of both relaxation process A and relaxation process B, and the relaxation peaks are shifted toward the direction of a small frequency. There is also a decrease in the conductivity process. The amplitude of relaxation process C is slightly increased.

The low-frequency relaxation process is related to concentration polarization. Nanocellulose is able to combine with the free water in the insulating paper by hydrogen bonding, thus suppressing the interfacial polarization and conductivity loss in the low-frequency band. The mid-frequency band is mainly dominated by the interfacial polarization of the oil and paper. As the doping of nanocellulose fills the pores and reduces the interfacial area between the insulating oil and the insulating paper, the interfacial polarization between the insulating oil and the paper is suppressed, and the dielectric loss of the insulating paper is reduced.

This study investigates the micromorphology and moisture absorption of cellulose nanocrystal (CNC)-modified insulating paper and examines the impact and mechanism of CNC doping on the mechanical and dielectric properties of the paper. The results demonstrate that the introduction of CNC can effectively enhance the comprehensive performance of oil-immersed insulating paper. The effect of CNC doping on performance showed an initial increase followed by a decrease at higher doping concentrations, with the optimal concentration providing the maximum enhancement. Data variability increased and then decreased as the concentration increased. Based on the findings, the following conclusions can be drawn:

1. CNC doping in the fiber network of the insulating paper enhances mechanical properties through hydrogen bonding, characterized by tensile strength and Young's modulus. Tensile strength was increased by up to 6.91% and Young's modulus by up to 11.98%. Excessively high concentrations of CNC lead to a decrease in tensile strength and saturation of Young's modulus as a result of the reduced average fiber slenderness ratio and increased CNC aggregation.

2. CNC doping fills the small pores in insulating paper, improving the consistency of measurement data, while excessive doping of CNC can lead to an increase in data dispersion due to CNC aggregation. The hydrogen bonding between CNC and cellulose fibers reduces the ability of hydroxyl groups on both CNC and the insulating paper to bind with water, resulting in a slightly lower moisture absorption rate. The introduction of CNC reduces interfacial polarization in the insulating paper, owing to factors, such as filling of pores and reduction in water content.

3. CNC doping introduces more traps, which inhibit charge flow; thus, the AC and DC breakdown field strength was increased. The breakdown field strength of CNC-modified paper reached the highest at 3 wt. % concentration, with a 28.53% increase in the AC breakdown field strength and a 8.72% increase in the DC breakdown field strength. The aggregation of CNCs at excessive doping levels enhances the polarization effect, thereby reducing the breakdown strength.

Overall, the results of this study highlight the potential of CNC doping as a promising strategy for enhancing the comprehensive performance of transformer insulating paper, which is a critical factor in ensuring the safe and reliable operation of power transformers. Future research could focus on optimizing the doping concentration and evaluating the long-term stability and reliability of this innovative material.

This work was supported by the Key Laboratory of Engineering Dielectrics and Its Application (Harbin University of Science and Technology), Ministry of Education.

The authors have no conflicts to disclose.

Daning Zhang: Conceptualization (equal); Data curation (equal); Writing – review & editing (equal). Xinnan Zhai: Data curation (equal); Formal analysis (equal); Writing – original draft (equal). Siyu Wang: Writing – review & editing (equal). Xuan Li: Data curation (equal); Methodology (equal); Resources (equal). Pengjiang Xu: Funding acquisition (equal); Investigation (equal). Haoxiang Zhao: Conceptualization (equal); Data curation (equal); Writing – original draft (equal). Guan-jun Zhang: Conceptualization (equal); Methodology (equal).

The data that support the findings of this study are available within the article.