Application of surface-modified XLPE nanocomposites for electrical insulation- partial discharge and morphological study

Abstract This paper investigates partial discharge (PD) characteristics of crosslinked polyethylene (XLPE) nanocomposites for unmodified, agglomerated, and Octylsilane-modified silica nanofillers (nano 1, 2, 3, 4, 5, 10 wt %) case. The surface modification of nanofiller helps to reduce the PD formation marginally. Octylsilane surface-modified XLPE/silica nano 3 wt % exhibits the lowest PD activity with highest discharge inception voltage and breakdown voltage. Also, the issue of change in the polymer structure due to the addition of nanofillers is reported here. The differential scanning calorimetry (DSC), thermo gravimetric analysis (TGA), fourier transform infrared (FTIR), and contact angle measurement study conducted suggests that the addition of nanosilica leads to the change in the melting point, thermal degradation temperature, heat of fusion, bonding structure and the contact angle of the polymer, respectively. These structural changes are explained with the supporting theory.


Introduction
Field of nanocomposite electrical insulations (nanofiller + electrical insulation) has gained a substantial attention of researchers in recent years. 1-3 The inclusion of nanofiller to insulation material leads to marginal improvement in its dielectric, electrical, thermal properties etc. [4][5][6][7][8] However, problem of agglomeration (or aggregation) of nanofillers has questioned its reliability and application. Agglomerated nanocomposites have reported the adverse performances such as decreased breakdown strength and other unfavorable electrical performance. This type of behavior biases the conclusion of different theories established to explain the role of nanofiller in the improvement of electrical, mechanical, thermal etc properties of nanocomposites. Recently, it has been found that surface modification (generally by Silane agents) of nanofillers acts as an adhesion between nanofillers and polymer matrix which improves the dispersion of nanofillers. This improves the dielectric properties such as AC and DC breakdown strength, electrical and water tree resistance while reducing the space charge, permittivity, dielectric loss etc. [9][10][11][12][13] This favors the application of surface-modified nanofillers with base materials such as epoxy resin, polyvinyl chloride, XLPE etc. for electrical insulation.
XLPE is considered as a suitable insulation material for the high voltage transmission of electrical power. [14][15][16] However, the increased voltage rating of XLPE based electrical cables has also increased the probability of formation of water tree, electrical tree, PD etc. inside insulation. Although the application of XLPE nanocomposite improves foresaid properties, very less research in the field of PD has been carried out. Authors have already reported the PD study of XLPE/ silica unmodified nanocomposites in 17 . In addition to this study, this paper presents the PD study of agglomerated and Octylsilane-modified XLPE/silica nanocomposites. It is expected that agglomerated nanocomposites may deliver almost same performance as that of unmodified or virgin samples.
Though the addition of nanofillers marginally improves the properties of base material, it also changes the morphology of polymer. [18][19][20][21][22] To observe these structural changes in the polymer matrix, characterization studies such as DSC (Maker-T. A Instruments, USA, SDT Q600 Conditions-performed from Room Temperature to 800 °C at 20 °C/min), TGA (Maker T. A Instruments, USA, SDT Q600, Conditions-performed from Room Temperature to 800 °C at 20 °C/min), FTIR (Maker-Shimadzu, Japan Conditions-measured between 4500 to 500 cm -1 ) and contact angle measurement (Instrumentcontact angle goniometer, Conditions-water drops are deposited on the surface of sample with the help of micro syringe attached to flat needle) are conducted which shows change in its melting point, degradation temperature, heat of fusion, bonding structure, and contact angle (in turn surface energy), respectively. Suitable theory has been proposed to describe the role of nanofillers in morphological changes of polymer.

Preparation and characterization
XLPE/silica nanocomposites (nanosilica dimensions -7 to 14 nm purchased from Reinste Nano Ventures Pvt. Ltd.) are prepared using twin screw extruder and injection molding (as shown below in the schematic) for following cases,

Case 1
PE granules with unmodified silica nanoparticles are mixed for 10 minutes and at 120 °C temperature. Dicumyl peroxide (DCP) is used as crosslinking agent. The nanocomposites prepared using this method are referred as unmodified nanocomposites.

Case 2
It should be noted that this case does not have a different preparation involved. Multiple number of specimens are prepared in case 1 for each wt %. It is evident from SEM that nanofillers cannot be dispersed in homogeneous manner. Some of specimens in each wt % show agglomeration as depicted in Figure 1

Case 3
Same method is followed as case 1 except Octylsilanemodified silica nanoparticles are used. Octylsilane is used due to its lesser size (7-14 nm), hydrophobicity and high specific surface area (150 m 2 /g) as compared to polydimethylsiloxane modified nanosilica (14 nm, 100 m 2 /g) provided by supplier which enables higher interactions at the interface of nanofiller and polymer matrix. Figure 1(c) and (d) shows that better dispersion of nanoparticles is achieved. In this case Octylsilane acts as a chemical adhesion between polymer matrix and nanosilica (attachment shown in Figure 1(e)). Generally, Silane coupling agent tries to separate nanoparticles from each other by acting as compatibilizer between nanoparticle surface and polymer matrix which eventually leads to homogeneous dispersion of nanofiller. 23,24 Sometimes, it also modifies the chemical bonds (hydrogen bonds) present at nanofiller surface. [25][26][27] Hereafter, nanocomposites prepared using this method are referred as Octylsilane-modified nanocomposites.

Experimental test
PD test is carried out using point-plane configuration with air clearance test setup as shown in Figure 2. PD performance is monitored using the measurement of discharge inception voltage (DIV) at which PD activity (or pulses) starts to appear, breakdown voltage (BDV) at which breakdown of specimen occurs and apparent charge (pC) Vs applied voltage (kV) characteristics following the same method reported in 17 . Ideal performance of insulation is that which has highest DIV and BDV with lesser PD activity. As discussed earlier, the surface treatment of nanofiller has considerable impact on its dielectric, electrical, thermal properties etc. With this context, present experimental study compares PD characteristics of unmodified and Octylsilane-modified XLPE/silica nanocomposites.

Results and discussion
Comparison of PD activity is made with virgin XLPE sample which is shown in Figure 3. PD characteristics and PD activity for nano 1, 2, 3, 4, 5, 10 wt % is illustrated in Figures 4-9, respectively. It can be seen that PD in all agglomerated nanocomposite samples remains almost same (nano 1, 2, 4, 10 wt %) or less (nano 3, 5 wt %) as compared to unmodified nanocomposite samples. This similarity is attributed to agglomerates present in the agglomerated nanocomposite samples. Here, agglomerates let the PD to reach toward the ground electrode. In case of XLPE/silica nano 3 and 5 wt % samples in Figures 6(a) and 8(a), the PD activity present in agglomerated samples is less than the unmodified sample. This is due to the point of the application of voltage i.e. the location of needle point. (explained later in the section Morphological changes due to the addition of nanofiller: theory of PD propagation).
The surface modification of nanofiller has marginal influence on the reduction of PD characteristics and activity. It can be seen from Figures 4-9(a) and (d) that in all wt % of Octylsilane surface modified nanocomposite samples, the PD remains well below the virgin and agglomerated nanocomposite samples. For unmodified case, XLPE/silica nano 5 wt % sample has the lowest PD activity with highest DIV and BDV. 17 This content is considered as the optimal content for PD characteristics. In case of Octylsilane surface modified XLPE/silica nanocomposites, the nano 3 wt % sample is having the lowest PD present as shown in Figure 6 with highest BDV and DIV as shown in Table 1. Clearly, this better performance is attributed to the dispersion of nanofiller due to surface modification. In a summary, the surface-modified nanocomposites have low optimal content of nanofiller toward PD characteristics as compared to the unmodified and agglomerated nanocomposites. Hence, the surface-modified XLPE/silica nanocomposites can also be considered as economically viable solution over the unmodified and agglomerated nanocomposites.

Theory of PD propagation
It is essential to understand the role of nanofillers in hindrance of PD. Generally in PD test, the sample to be tested is held between the high voltage and ground electrode as shown in Figure 2. It can be seen in Figure 10(a) that unmodified silica nanofillers are not dispersed in very good manner. Hence, frequency of encounter of PD with nanofiller (denoted by black color line) is very less and it is easily able to reach to ground

Morphological changes due to the addition of nanofiller
As the addition of nanofillers to base material enhances its properties (electrical, thermal, mechanical etc.), there is need to investigate its impact on the structure of base material (polymer in this case). Change in the morphology due to addition of nanofillers is explained hereafter with theory and supporting results.

Secondary crystallization and multiple melting points
Structural changes in XLPE due to addition of nanofillers mainly occur in amorphous phase. Nanofiller addition leads to decrease in surface energy which causes secondary or incomplete crystallization. 18,22,28 Incomplete or secondary crystallization can also be introduced due to degradation of polymer during the preparation and low molecular weight chain presence. 18,19 Peaks formed due to secondary or incomplete crystallization can be seen in Figure 11. To explain, virgin sample ( Figure 11(a)) is showing no phase transition from point 3 to 4 whereas nano 1 wt % unmodified (Figure 11(b)), agglomerated (Figure 11(c)) and Octylsilane-modified (Figure 11(d)) nanocomposite samples are showing multiple melting and/ or degradation peaks at 3′ and 3″ i.e. secondary or incomplete crystallization. In this study, melting peaks observed are due to annealing process followed in the preparation of XLPE nanocomposite samples. Process of annealing allows heated electrode. This may lead to the breakdown of sample which in turn reduces life of insulation. Agglomeration leads to concentration of nanofillers at several points. Hence, PD can reach to ground electrode easily as in case of unmodified samples as shown in Figure 10(b). To achieve a very good dispersion, the surface of the nanofillers is modified with Octyl-silane which acts as a coupling agent and improves the dispersion as well as the distribution of nanofillers marginally as shown in Figure 10(c). This helps to delay the propagation of PD and improves life of insulation. The surface modification of nanofillers enhances dispersion and avoids the agglomeration. But, it should be noted that ideal dispersion as shown in Figure 10(d) may not be achieved in case of surface-modified nanofillers. Hence, it is very much important to have standardized method of preparation to obtain ideal dispersion.

Figure 3 PD pulses appearing at 15.5 kV in virgin XLPE
melt compounding used in the preparation of nanocomposites. 33 It should be noted that increase in melting term is used with respect to addition of nanofiller. It should not be confused with the decrease with respect to Virgin XLPE sample.
Also, nanofiller addition to polymer accelerates the process of lamellar thickness inside polymer matrix which increases heat of fusion as shown in Figure 12(b). Whole process can be called as rearrangement of lamellae due to addition of nanofillers which is responsible for the multiple peaks in the DSC curves 34 shown in Figure 11. Theory of morphological changes in the structure of nanocomposites can be also supported by nucleation mechanism. Nanosilica acts as a nucleating agent which promotes the process of heterogeneous nucleation which reduces polymer chain mobility. Surface-modified nanocomposites are believed to increase the sites for nucleation and in turn crystallization. [35][36][37][38] Addition of nanosilica with polymer also changes the crystallinity of polymer as reported in previous research. 17 sample to cool down slowly and hence cannot be eliminated in nanocomposite preparation. In this process crystals which were not crystallized (incomplete crystallization) earlier during the heating at higher temperatures (during heating) gets crystallized (secondary crystallization) at lower temperatures (during cooling) due to nanofiller addition. Hence two melting peaks are observed due to the addition of nanofillers. 29 However, the melting points exhibit erratic trend in all types of nanocomposites here. In case of unmodified and agglomerated nanocomposites, melting point increases with the addition of nanofillers as shown in Figure 12(a). It may be due to increase in crystallization and crystal thickness. 30 However, in case of Octylsilane-modified nanocomposites no continuous trend of increase or decrease in melting point is observed. It may be due to change in molecular weight. This change causes crystals to melt at different temperatures. 31 These unpredictable changes can also be caused due to variable lamellar thickness due to re-crystallization 32 and of all nanocomposites. Contact angle measurement study shows that nature of polymer is changed from hydrophilic (Contact angle < 90°) to hydrophobic (contact angle > 90°). This change is due to replacement of Silanol group of unmodified nanosilica by octyl groups of Octylsilane-modified nanosilica. 22 Interestingly, contact angle of XLPE/silica nano 5 wt % unmodified samples shows considerable increase from 75.27° to 83.73° as shown in Figure 12(d). This change can be attributed to decrease in surface energy. Decrease in surface energy with the addition of nanofiller is reported in 27 . As it is already shown in our previous research reported in 17 , the optimal content of nanofiller is XLPE/silica nano 5 wt %. Also, for Octylsilane-modified optimal content of XLPE/silica nano 3 wt %, there is considerable change in contact angle from 94.59 o to 100.29 o . Also, nanofiller addition changes the polarity of the polymer surface from polar to non-polar. Hence, polymeric surface becomes more resistant (more hydrophobic) against the penetration of water. 27 However, relation between contact angle and surface energy with enhanced electrical properties can present significant results.

Thermal stability
Change in thermal degradation temperature of polymer with the addition of nanosilica can be clearly seen in Figure 12(c). It can be clearly seen that surface modification helps to achieve higher thermal stability than unmodified and agglomerated nanocomposites. The improved performance is ascribed to chemical changes of nanosilica due to its surface modification. Reactive groups (Si-O) present at the surface-modified silica interact with the polymer matrix which leads to enhancement in thermal stability. 39

Contact angle measurement
Contact angle is the angle between solid-liquid surface and is measured by using contact angle goniometer in this study. To measure contact angle, water drops are deposited on the surface of sample with the help of micro syringe attached to flat needle. Figure 12(d) depicts the contact angle of unmodified, agglomerated, and Octylsilane surface nanocomposites. Clearly, nanofiller addition increases the contact angle

FTIR analysis
Also, change in the bonding structure at the interface between nanoparticle and polymer matrix is shown in Figure 13. FTIR absorption spectra (Shimadzu, Japan) is measured between 4500 to 500 cm -1 . FTIR spectroscopy shows that surface modification introduces elongated peaks (except 1018 and 1091.57 cm −1 ) as compared to virgin and unmodified nanocomposites which may be due to extensive O-H bonding at nanofiller surface introduced due to Octylsilane. Peaks at 2914.44 and 2846.93 cm −1 represents C-H as well as C-O bonds of cumyl alcohol whereas 1463.97 cm −1 belongs to C=O of acetophenone. It should be noted that cumyl alcohol and acetophenone are the crosslinking by-products of XLPE. 40 Peak of 1018.41 cm −1 display Si-O bond. It proves that Silane structure is present inside Octylsilane-modified nanocomposites. 41 In a summary, the addition of nanofillers alter the kinetics of the crystallization and the morphology of the crystal growth.
Morphological changes are explained to show the role of different parameters on the performance of nanocomposites.

Conclusion
In case of PD study, Octylsilane-modified XLPE/silica nanocomposites have proved its efficacy over unmodified and agglomerated nanocomposites. From the results, PD activity can be summarized as, surface-modified case < agglomerated </≤ unmodified < Virgin XLPE. Octylsilane modified XLPE/silica nano 3 wt % is having better PD performance than the unmodified nano 5 wt %. Hence the Octylsilane-modified XLPE/silica nano 3 wt % is economically viable solution over the unmodified and agglomerated nanocomposites. Effect of nanofiller addition on the morphology of polymer matrix is also analyzed in this study which depends on its processing conditions, lamellar thickening process, and crystallinity. The improved performance of Octylsilane-modified nanocomposites have been explained with modified polymer chain alignment model which suggests that Octylsilane-modified nanofillers have better alignment with XLPE polymer chains which enhances the performance of Octylsilane-modified XLPE nanocomposites.
In this research, authors attempt to propose the modified polymer chain alignment model for XLPE nanocomposites. As it is already stated that DCP is used as a crosslinking agent, the chain alignment of the polyethylene will not begin unless DCP is added to polyethylene granules and mixed at standard conditions. After the addition of DCP, polymer chains will start aligning themselves with Octylsilane molecules. Assuming an ideal case, each Octylsilane molecule will get aligned with the one polymer chain as shown in Figure 14(a). As it can be seen, the addition of the nanosilica changes the morphology of the polymer. In case of unmodified nanosilica as shown in Figure 14(b), alignment of the polymer chains is very poor due to less interactions at interface of nanofiller. Also the overlapping of interfaces of neighboring nanofillers is the major reason behind these changes. This affects the chain mobility and chain length of polymer. In case of modified nanosilica, alignment of polymer chains will be improved as shown in Figure 14(c). This is attributed to the increased reactions at interface between nanofiller and polymer matrix. Same theory of unmodified nanocomposites applies to agglomerated. But, agglomeration of the nanofillers has very high probability of