Dramatically enhanced electrical breakdown strength in cellulose nanopaper

Insulating paper made from natural wood fibers with diameter ranging from several micrometers to tens of micrometers has been widely used in power transformers, cables and capacitors.¹ Recently, with the development of ultra-high voltage power transmission system, insulation failures in ±800 kV convert transformers become a severe problem.² One of the major reasons is that the insulating properties of the paper insulation in use are not good enough and need to be improved. However, only a few studies have dealt with the reinforcements of breakdown behavior and tensile strength of insulating paper since the first use of paper as an electrically insulating material in power transformers.³ Among the available references, one remarkable investigation conducted by Ruiliao Jin et al⁴ reported the enhancement of AC breakdown strength of paper modified by nanoparticles. Unfortunately, mechanical performance of the nanomodified paper normally deceased due to the incompatible between the inorganic nanoparticles and organic wood fibers.³ In our recent study, we observed the improvement of both breakdown behavior and mechanical properties by introduction of nanofibrillated cellulose (NFC) to ordinary cellulose fibers.⁵ Although the enhancement is no more than 30%, this study inspires our interest in the electrically insulating performance of paper made from 100% nanofibers, namely nanopaper.

Because nanofiber has a width less than the wavelength of visible light, the pores inside nanopaper formed by nanofibers can be small enough to avoid light scattering. The major difference between nanopaper and regular paper is that the former is optical transparency. In addition to that, nanopaper also exhibits good mechanical and thermal properties.^6–8 Tensile strength of nanopaper can reach 200-400 MPa, which is obviously higher than that of regular paper.^7,9 Coefficient of thermal expansion (CTE) of nanopaper is lower than 8.5 ppm/K, whereas that of plastic is about 50 ppm/K.⁷ Even at 150°C, nanopaper can maintain its high optical transparency and smooth surface.⁶ And even at 260°C, nanopaper can exhibit advantageous mechanical properties.⁹ Due to its excellent properties, a lot of studies have discussed the potential applications of nanopaper, such as flexible displays, biosensors and energy-storage devices.^7,9,10 However, with respect to the possibility of being used as an electrically insulating material, one of the most important applications of regular paper, has been rarely reported.

For an electrically insulating material used in power equipment, electrical properties, mechanical strength and thermal durability are all of great importance. Based on the existing literature, we can infer that nanopaper has outstanding mechanical and thermal properties. R. Hollertz et al.³ compared the dielectric responses of nanopaper and kraft paper, she found that nanopaper had a larger relative permittivity and a higher dielectric loss and attributed these characteristics to the higher density of nanopaper. However, other electrical properties of nanopaper are still unknown and need further investigations. Among the electrical performances, breakdown strength is typically regarded as the decisively important property. Therefore, in order to determine the suitability of nanopaper for electrical insulation, evaluation on breakdown behavior of nanopaper is of great significance.

In this study, the suitability of using nanopaper as an electrically insulating material was discussed. Breakdown strength of nanopaper was evaluated and compared to that of conventional insulating paper. To explore the reason for the significantly enhanced breakdown strength of nanopaper, we conducted the SEM analysis and measured the pore size distribution inside the samples. Obviously decreased pore diameter was observed in nanopaper. In addition, theoretical breakdown strengths of nanopaper and regular insulting paper were calculated to evaluate the effect of pore size on breakdown behaviors. The good agreement between the theoretical values and experimental data indicates that the increased breakdown strength of nanopaper is probably attributed to the decreased pore size inside the fibrous network. Due to its outstanding breakdown strength, thermal and mechanical properties, nanopaper is considered to be a promising candidate for electrical insulation.

Nanofibrillated cellulose (NFC) was provided by Tianjin University of Science and Technology (Tianjin, China). It was prepared from enzyme-treated dissolving pulp by using a grinding and homogenization method. Average diameter of the obtained NFC was below 50 nm. For the preparation of nanopaper, 200 g NFC suspension with a concentration of 1.5 wt. % was diluted in pure water (electrical conductivity of $3μS/cm)$ to a final mass of 600 g (NFC concentration of 0.5 wt. %). Then the suspension was ultrasonically dispersed for 5 minutes by using the SCIENTZ JY 99-IIDN ultrasonic homogenizer with a power of 1260 W. After that, the above steps are repeated to obtain 1200 g suspension with a NFC concentration of 0.5 wt. %. Then the suspension was poured into a handsheet former with a stainless steel metal mesh (2000 mesh), filtered by vacuum to obtain the wet nanopaper sheet. Finally, the wet sheet was dried at room temperature without additional pressure. Prepared nanopaper has a thickness of about $70μm$ and a density of 0.62 g/cm³. The high optical transparency shown in Fig. 1(b) suggests the successful preparation of nanopaper.

DC breakdown tests were conducted according to IEC 60243. Equal diameter (25 mm) electrodes made of stainless steel were used. Size of the test samples was 50 mm × 50 mm. Spellman SL150 was used to supply the DC voltage. Measurements were performed at room temperature in the air. Prior to the tests, the samples were thermally treated at 105°C for 12 hours to remove the moisture. For both conventional paper and nanopaper, 15 samples were measured. Fig. 2 presents the breakdown strengths of conventional insulating paper and nanopaper. The conventional paper used in this study was commercial available. It had a thickness of about $70μm$ and a density of 0.85 g/cm³. Breakdown strength of nanopaper is 67.7 kV/mm, whereas that of conventional paper is only 20 kV/mm, indicating that nanopaper has a much better breakdown performance. Optical microscope images of the breakdown sites are shown in Figs. 3(a) and 3(b), respectively. Olympus BX 43 microscope at a magnification of 10x was used. For both conventional paper and nanopaper, only one breakdown position is observed. Diameters of the holes caused by electrical breakdown are typically less than $100μm$⁠. More detailed pictures obtained by scanning electron microscope (SEM) exhibit a difference in the edge of the holes, see Fig. 3(c) and Fig. 3(d). Quanta FEG 450 scanning electron microscope was used. All the samples were sputter-coated with gold before observations. It can be seen that in some nanopaper samples, the nanofibers in the breakdown holes are not completely melted and evaporated. A membrane can be observed.

Since paper can be regarded as a composite material composed of cellulose fibers and air voids. Properties of wood fibers and the porous structure probably have a great effect on the breakdown behaviors. Therefore, the following discussion will focus on these two aspects. Because nanofibers are prepared from regular wood fibers, chemical structures and properties of nanofibers and wood fibers are almost the same.⁷ Consequently, difference in fiber properties is less likely to be the reason for the increased breakdown strength. Porous structure is normally described by porosity and pore size distribution. Porosity of conventional paper or nanopaper can be calculated by¹¹ 

Where p is porosity; $ρsample$ and $ρcellulose$ represent the densities of test sample and cellulose fiber, respectively. Density of cellulose fiber is usually assumed to be 1.53 g/cm³.

According to Eq. (1), higher density means a lower porosity. Typically, breakdown strength decreases with the increase of porosity.^12–14 In this study, density of conventional paper is higher than that of nanopaper, which means that nanopaper has a higher porosity. Hence, porosity is not likely to be the reason. The dramatically enhanced breakdown strength of nanopaper is probably due to the variation of pore size distribution. As mentioned above, nanopaper is optical transparency, indicating that air voids in nanopaper should be much smaller than those in conventional paper. By comparing Fig. 3(e) and Fig. 3(f), we can see some obvious air voids in conventional paper although coating is used to fill the voids and make the surface smooth. In contrast, no big holes are observed in nanopaper. The rough surface of nanopaper is due to the absence of surface smoothing process.⁶ 

To determine the pore size distributions of conventional paper and nanopaper, mercury intrusion method was firstly used. An AutoPore IV 9500 (Micromeritics USA) was utilized. The advancing contact angle and the surface tension were defined as 130° and 0.485 N/m, respectively. The maximum pressure is 414 MPa, corresponding to pores with a diameter of 3 nm. The measurable pore size ranges from 3 nm to $380μm$⁠. Fig. 4 shows the pore size distribution and cumulative pore volume results of conventional paper and nanopaper. For conventional paper, pore size distribution exhibits two peaks. The first one is around $1.7μm$⁠, and the second one is between $10μm$ and $20μm$⁠. The second peak is due to the interspaces between the discrete pieces of test sample.¹⁵ The small variation of cumulative pore volume between $3μm$ and $10μm$ also indicates that pores larger than $10μm$ are attributed to the space between test pieces. Therefore, pores with a diameter high than $10μm$ should not be taken into account. Pore size of conventional insulating paper is around $1.7μm$⁠, which shows a good agreement with the available reference.¹⁶ With regard to nanopaper, two samples were measured, both of them show that no pore is observed in the range from 3 nm to $10μm$⁠. Because pores with a diameter of about $30μm$ are caused by the interspace between nanopaper pieces, we can infer that pores in nanopaper are smaller than 3 nm.

Gas adsorption technique was adopted to further determine the pore size distributions of nanopaper. An Autosorb IQ2 (Quantachrom USA) was used for the measurement. N₂ was utilized as the analysis gas at 77.35 K. The nanopaper sample was outgassed at 120°C for 11 hours. Fig. 5 shows the pore size distribution and cumulative pore volume of nanopaper in the range from 1 nm to 200 nm. It can be seen that most pores have a diameter of about 2.8 nm. On the basis of the above analyses, we conclude that pore size of conventional paper is about $1.7μm$⁠, whereas that of nanopaper is around 2.8 nm. As electrical breakdown field of porous material usually increases with the reduction of pore size.^13,17 It can be inferred that the increased breakdown strength of nanopaper is probably due to the decreased pore size.

In order to further verify the above speculation and evaluate the effect of pore size on breakdown behavior, the breakdown model proposed by Gerson and Marshall¹⁴ is adopted. In that model, the pores inside the specimen are assumed to be randomly distributed. Test samples is regarded as a huge cube composed of small cubes with an edge equals to the diameter of the air voids. For a given small cube, the probabilities of the cube being occupied by air voids or cellulose fiber are p (porosity) and 1-p, respectively. Fig. 6 gives a schematic representation for the electrical breakdown model by using 15×15×10 small cubes. The model has 15×15 columns, and each column has 10 cubes. Electrical breakdown will occur along the column containing the maximum number of air voids. The probability of finding a column of total n cubes that contains x cubes occupied by air voids is

Where p is porosity of the test sample and can be calculated according to Eq. (1). The maximum value of x, namely x_m, is determined by

Where N is the number of the columns. Once x_m is known, the breakdown field for the test sample is

Where E₀ is the breakdown strength of cellulose. For conventional paper and nanopaper, the value can be regarded as the same. Thickness of conventional paper and nanopaper is $70μm$⁠. Effective diameter of them is 25 mm. Porosity of conventional paper is 44%, and that of nanopaper is 59%. Pore sizes of conventional paper and nanopaper are $1.7μm$ and 2.8 nm, respectively. Therefore, the calculated value for conventional paper is

And that for nanopaper is

E_nanopaper/E_regular equals to 3.25. The ratio of the measured breakdown strength of nanopaper to that of conventional insulating paper is 3.39, indicating that the calculated values agree well with the experimental data. Therefore, it is concluded that the decreased pore size is the main reason for the enhanced breakdown strength of nanopaper.

In summary, breakdown strength of nanopaper was evaluated and compared to that of conventional insulating paper. Results show that breakdown strength of nanopaper is dramatically enhanced to 67.7 kV/mm, which is more than two times higher than that of conventional paper. Diameters of air voids in conventional paper and nanopaper are $1.7μm$ and 2.8 nm, respectively. The good agreement between the experimental data and the calculated values indicates that the increased breakdown strength of nanopaper is mainly due to the decreased pore size. Because of its outstanding breakdown strength, thermal and mechanical properties, nanopaper is considered to be a promising material for electrical insulation in ultra-high voltage electrical apparatus.

This work was supported by the Science and Technology Project of China Southern Power Grid (No. KY2014-2-0016).