In uence of the Semiconductive Composites Doped with Li4Ti5O12 on Space Charge Implantation in LDPE

Hongxia Yin (  13455018936@163.com ) Institute of Advanced Electrical Materials, Qingdao University of Science and Technology https://orcid.org/0000-0001-9892-8066 Shiyi Zhao Institute of Advanced Electrical Materials, Qingdao University of Science and Technology Chuncheng Hao Institute of Advanced Electrical Materials, Qingdao University of Science and Technology Qingquan Lei Institute of Acvanced Electrical Materials, Qingdao University of Science and Technology


Introduction
High-voltage direct current (HVDC) transmission systems have many practical applications owing to their advantages, including low line costs, large long-distance transmission capabilities, and the ability to transmit high power underground or underwater [1][2][3][4]. However, operation safety is compromised by two issues that must be urgently resolved. The rst issue is the space charge accumulation in the insulating material that leads to numerous problems such as degradation and aging, partial discharge, and premature insulating failure [5][6][7][8]. The second issue is a positive temperature coe cient (PTC) effect that is always observed for semiconductive nanocomposites. This effect increases the resistivity of a semiconductive compound with an increase in temperature. When the temperature reaches a critical value, it increases drastically, leading to a local overheating of the semiconductive screen, which deteriorates its ability to suppress charge implantation [9,10]. Hence, the development of safe HVDC cables requires mitigating the PTC effect and inhibiting space charge implantation to the insulating layer [11][12][13].
The idea of minimizing space charges in the insulation layer has attracted considerable attention in the past decades. For its realization, researches focused on the mechanisms of the charge injection and charge trapping processes; as a result, several methods for suppressing space charge accumulation, including the addition of certain types of nanoparticles to the insulation material, were developed [14][15][16][17]. However, few studies have investigated semiconductive screen layers, although they play a critical role in homogenizing the high electric eld and inhibiting charge implantation to the insulating layer.
Usually, a semiconductive screen consists of two layers in a single core HVDC cables: an inner layer sandwiched between the copper conductor and the insulating layer and an outer layer sandwiched between the insulating layer and the wire metallic shield. Furthermore, doping a screen material with ionic conductors such as La 0.8 Sr 0.2 MnO 3 , La 0.6 Sr 0.4 CoO 3 , or SrFe 12 O 19 magnetic particles represents an effective way of enhancing the conductive network and inhibiting space charge implantation [18][19][20]. A typical semiconductive screen is fabricated from a composite material consisting of a polymeric matrix, low-density polyethylene (LDPE), ethylene-vinyl acetate copolymer (EVA), and carbon black (CB) particles [21][22][23][24], which are dispersed across the matrix to form a conductive network.
The ionic conductor Li 4 Ti 5 O 12 (LTO) with a spinel structure has been examined by various researchers as a promising electrode material [25,26]. It possesses characteristic properties, including small structural changes during charge-discharge cycles. In particular, the insertion of Li + ions into the Li 4 Ti 5 O 12 lattice or their extraction from this lattice causes minor changes in the lattice dimensions; therefore, Li 4 Ti 5 O 12 is called "a zero-strain insertion compound." Note that Li 4 Ti 5 O 12 has a high Li + diffusion coe cient [27,28].
The objective of this study was to investigate the inhibition of the PTC effect and carrier transport to the insulating layer by a Li 4

Synthesis of Li 4 Ti 5 O 12 nanopowder
To prepare an ionic conductor with uniform morphology, narrow particle size distribution, and high homogeneity, a sol-gel technique was used to synthesize a Li 4 Ti 5 O 12 precursor, which was subsequently sintered to obtain Li 4 Ti 5 O 12 compound with a spinel structure. An appropriate amount of anhydrous lithium acetate and TBT were weighed at a Li + : Ti 4+ molar ratio of 0.82:1. Anhydrous lithium acetate was dissolved in a mixture composed of anhydrous ethanol, acetic acid, and deionized water (V: V: V = 120: 10: 5) at a molar ratio of 1:50 (Liquid A). TBT was dissolved in anhydrous ethanol at a molar ratio of 1:60 (Liquid B). Liquid B was placed into an oil bath at 55°C and subjected to vigorous stirring. Liquid A was gradually dropped into liquid B to produce a colorless and transparent solution. To obtain a stable colloidal solution, the reaction solution was aged for 4 h at 55°C under vigorous stirring. The resulting gel was heated to 80°C to extract excess anhydrous ethanol, water, and an organic precursor. Next, the organic precursor was dried at 150°C, transferred to a mu e furnace, and carbonized for 4 h at 500°C.
After that, the furnace temperature was increased to 800°C, and the sample was sintered for 6 h. As a result, lithium titanate with a spinel structure was obtained [25,26,29].
The synthesized Li 4 Ti 5 O 12 was ground in a planetary ball mill (MiQi YXQM-4L) for a certain period (10 hours, 20 hours and 30 hours). According to the ball milling procedure described elsewhere [30], the sample was divided into two equal parts and placed into a pair of nylon cans, followed by the addition of zirconia balls with different sizes at a weight ratio of 1:20 and anhydrous ethanol in the amount corresponding to 75% of the reactor capacity. The rotating speed was set to 300 rpm, and the ball milling time was increased at 10-h intervals until Li 4 Ti 5 O 12 was milled into nanopowder with uniform particle sizes. Samples were taken out every 10 h for structural and morphological analysis by X-ray diffraction (XRD, D-MAX 2500/PC) and eld emission scanning electron microscopy (FE-SEM, FEI/Nova Nano SEM 450), respectively, to determine the optimal ball milling time. Li 4 Ti 5 O 12 semiconductive composites were manufactured in two steps by a melt blending method.

Preparation and characterization of Li 4 Ti 5 O 12 semiconductive composites
Initially, 45wt.% LDPE, 30wt.% EVA, and 25wt.% CB particles were fused together at 150°C with a two-roll miller (HaPro). Next, 1 wt.%, 2 wt.%, 3 wt.%, 4 wt.%, and 5 wt.% of the Li 4 Ti 5 O 12 nanopowder and 2wt.% of the crosslinking agent were blended with the pre-mixed materials for 12 min at 120°C using a mixed torque rheometer (RM-200C, HaPro) to obtain LTO semiconductive materials. Simultaneously, the premixed product was blended without Li 4 Ti 5 O 12 during the rst step under the same conditions. The produced nanocomposites with different Li 4 Ti 5 O 12 contents were shaped into sheets with various sizes for different electric performance analyses using a hot press apparatus (SKZ401) as follows. First, a nanocomposite was placed in a at vulcanizer at a temperature of 130°C no pressure for 4 min and then under a pressure of 10 MPa for 6 min. In the next step, a cross-linking reaction was performed at a pressure of 10 MPa and temperature of 180°C for 15 min. Finally, the specimen was cooled for 10 min under a pressure of 10 MPa.
To obtain the distribution of Li 4 Ti 5 O 12 nanoparticles in the composite matrix, the microstructures of the LTO semiconductive composites with Li 4 Ti 5 O 12 contents of 1 wt.%, 4 wt.%, and 5 wt.% were observed by FE-SEM and HRTEM (FEI·Tecnai·G2F30) after being sliced in liquid nitrogen, and their elemental energy dispersive spectroscopy (EDS) analyses were performed.

Resistivity measurements
Because the thermal expansion coe cient of the polymer is three times larger than that of CB particles, the polymer matrix expands with increasing temperature. As a result, the CB particles become diluted, inhibiting their charge transfer and leading to the blockage of conductive channels, which increases the resistivity of the nano-compound [31,32]. Hence, resistivity represents an important characteristic of the conductive channels inside semiconductive composites.
The specimen for resistivity testing was a sheet with a length of 110 mm, width of 50 mm, and thickness of 0.5 mm. Its resistivity was measured at temperatures of 30, 40, 50, 60, 70, 80, 90, 100, and 110°C using a semiconductive rubber and plastic material resistivity tester (DB-4). The speci ed temperature range corresponds to the operational conditions of the HVDC cable.

TSDC testing
TSDC can accurately re ect the space charge amount in an insulating material; thus, TSDC measurements represent a widely used technique for studying the parameters of space charges in the insulation layer [33]. To examine the in uence of Li 4 Ti 5 O 12 concentration on the suppression of carrier transport under the action of a powerful electric eld, the TSDC generated in the insulation was determined. For this purpose, the insulating layer was polarized by high voltage under the shield of a LTO semiconductive nanocomposites. The utilized polarization device included a high negative voltage DC source, a couple of copper cylindrical electrodes with diameters of 25 mm. The LTO semiconductive layer was placed between the cathode and the insulating sheet used as the shelter, and a plastic shielding box used for screening (Fig. 1). The thickness of the LTO semiconductive layer was 0.5 mm. The insulating layer is an LDPE lm with a thickness of 0.3 mm and uniform structure. The surfaces of the insulating and semiconductive layers were cleaned with anhydrous ethanol and dried for 4 h at 60°C before testing. The polarizing electric eld strength was varied between 10, 20, and 30 kV/mm, and the polarization time was 30 min.
After the polarization process was complete, the LDPE depolarization current was measured using a Novocontrol Concept-80 wide band dielectric spectrometer. The starting temperature was 20°C, the heating rate was 3°C/min, and the nal temperature was 90°C.

Space charge distribution measurements
The space charges in the insulation layer can be classi ed into two types: homo-charges originated from the electrode (metal core) emission, and hetero-charges caused by the material impurities ionized under high stress [34]. When the electric eld strength is su ciently high (for example, 10 kV/mm or more), the implanted charges from the core become predominant, which is the major reason for the space charge accumulation in the insulation layer.
To investigate the inhibitory effect of the produced LTO semiconductive composites on the space charge implantation to the insulating material, PEA measurements were performed. During the test, an LTO semiconductive sheet was placed between the cathode and the insulating layer not only for acoustic impedance matching but also for suppressing the simultaneous charge injection into the insulation layer [35]. In this process, a high voltage was applied through the upper electrode to simulate charge injection, and a high-voltage electric pulse was applied at the same time to cause a small displacement of the charge in the insulating sheet. This slight movement was transferred to the piezoelectric sensor as a sound wave that was subsequently detected and transmitted to a computer for analysis. Thus, the distribution of space charges in the insulating layer was represented by the obtained charge density. A schematic of the utilized PEA measurement setup is shown in Fig. 2.
In this test, the in uence of Li 4 Ti 5 O 12 content in the semiconductive composites on the space charge distribution was explored, and an LDPE layer with a thickness of 0.3 mm was used as the insulation medium. The test electric eld strength was varied between 10, 20, and 30 kV/mm, and the applied voltage was maintained for 30 min.  large grain aggregates are present in the sample that was ball-milled for 10 h (Fig. 4b). After ball milling for 20 h, the particle size becomes more uniform, ranging between 300 and 500 nm, and all particle aggregates disappear (Fig. 4c). A comparison of the images presented in Fig. 4c and d Figure 5d displays the EDS spectrum of the specimen with a Li 4 Ti 5 O 12 content of 4 wt.%. It shows that the carbon amount is the highest, followed by oxygen and titanium. Hence, the white particles in Fig. 5a-c correspond to Li 4 Ti 5 O 12 particles.   As shown in Fig. 7a, the resistivity of the LTO semiconductive sheet increases more slowly than that of the specimen without Li 4 Ti 5 O 12 , and the corresponding curve exhibits the slowest ascending trend at a According to the results of previous studies [10,31], crystalline and amorphous phases coexist in polymers, and the majority of ionic conductors can merely disperse at their interface at room temperature. Owing to the large molecular weight difference between inorganic compounds and polymers, the molarity of inorganic particles is much higher than that of polymers. Therefore, if the concentration of ionic grains is high enough and these inorganic grains can be scattered across the matrix, they may collide with each other through thermal motion and thus exchange and transfer charges to form conductive channels that lower the composite resistivity. When the temperature is below the melting point of the polymer, its macromolecular chains cannot move freely, and the three composite phases remain relatively stable; as a result, the composite resistivity increases. When the composite is heated to the polymer melting point, the polymeric phase begins to melt while stretching macromolecular chains, which expands the polymer matrix and dilutes the conductive particles. If the conductive ller contains only CB particles, its conductance would decrease with increasing temperature, accompanied by the partial rupture of the conductive channel caused by CB dilution, which leads to a resistivity increase [31, 32,36].

Resistivity of LTO semiconductive composites
Representative diagrams of the charge delivery networks of the prepared semiconductive composites are presented in Fig. 8. Figure 8a depicts the charge transport network formed only by the CB particles in the polymer matrix at room temperature. As the temperature increases, the polymers start to in ate, and the distance between CB particles gradually increases, leading to a decrease in the number of conductive channels. As the temperature rises continuously, an increasingly large number of CB particles become isolated and much more conductive network fracture. In this case, charge delivery becomes more di cult, causing the PTC effect (Fig. 8b). According to the results of previous studies, not only the migration of Li + ions, but also the empty d-orbits of Ti 4+ ions contribute to the electron transfer process [27,28]. This facilitates the creation of a stronger conductive network inside the nanocomposites structure after the addition and scattering of Li 4 Ti 5 O 12 particles. It is well known that ionic conductivity increases with an increase in temperature, which compensates for the decreased electronic conductivity of CB particles.
According to the obtained results, the specimen containing 4 wt.% Li 4 Ti 5 O 12 has the lowest peak resistivity near the melting point, which indicates that the ionic conductivity of Li 4 Ti 5 O 12 and the electronic conductivity of CB complement each other and that an optimal conductive network is constructed, as shown in Fig. 8c. However, the PTC effect becomes strong again when the Li 4 Ti 5 O 12 concentration exceeds 4 wt.%. As reported in previous studies, the majority of inorganic compounds can exist only at the interface between the crystalline phase and the amorphous phase of polymers [9,36], which limits the space for conductive particles. As a result, excessive Li 4 Ti 5 O 12 nanoparticles begin to agglomerate and form oversized grains, as shown in Figs. 5c and 6c. This destroys the conductive network and enhances the PTC effect (Fig. 8d). A is obtained at an electric eld strength of 10 kV/mm. Its value is reduced to 2.9×10 − 12 A at 20 kV/mm and to 2.60×10 − 12 A at 30 kV/mm, corresponding to relative decreases of 44%, 37%, and 42%, respectively. Figure 9d shows the relationship between the charge quantity in the insulation material and the Li 4 Ti 5 O 12 concentration in the semiconductive compound (charge quantities in this work were calculated by integrating the depolarization current curves depicted in Fig. 9a-c). Its value rst decreases with an increase in Li 4 Ti 5 O 12 concentration and then increases as the latter exceeds 4 wt.%. The minimum charge quantity at electric eld strength of 10 kV/mm equals 5.33×10 − 9 C; its magnitude is further decreased to 3.11×10 − 9 C at 20 kV/mm and to 3.04×10 − 9 C at 30 kV/mm, which correspond to relative decreases of 49.7%, 58%, and 46.9%, respectively. Thus, the produced semiconductive composites inhibit space charge implantation most effectively at a Li 4 Ti 5 O 12 content of 4 wt.%. Figure 9 also shows that the depolarization current decreases when the polarizing stress increases from 10 kV/mm to 20 or 30 kV/mm, which is accompanied by the reduction in the charge quantity. These results suggest that the screening effect of the LTO semiconductive composites is enhanced under higher stress because the stress promotes the polarization of Li 4 Ti 5 O 12 molecules. Correspondingly, the electronic cloud of Li 4 Ti 5 O 12 becomes more distorted to generate an electric eld that partially offsets the applied electric eld. Hence, the movement of space charges is inhibited by the induced electric eld generated by the polarized Li 4 Ti 5 O 12 particles as they pass through the semiconductive layer. Moreover, a fraction of these charges cannot pass through the interfacial barrier between the semiconductive layer and the insulating layer due to energy losses. As a result, the space charge amount in the insulating layer is decreased more e ciently at an electric eld strength of 20 or 30 kV/mm, indicating that the prepared LTO semiconductive composites can be potentially used in HVDC cables operated at 20-30 kV/mm.

TSDC testing
As the Li 4 Ti 5 O 12 content exceeds 4 wt.%, the charge quantity in the insulation layer increases again, which implies that the screening ability of the LTO semiconductive compound is weakened due to the agglomeration of Li 4 Ti 5 O 12 particles, which is con rmed by the corresponding SEM and HRTEM images.
Hence, it can be concluded that the optimal Li 4 Ti 5 O 12 concentration is equal to 4 wt.%. Li 4 Ti 5 O 12 nanoparticles are polarized by E ex , the electronic cloud becomes deformed and generates an induced electric eld E in , the direction of which is opposite to that of E ex . As a result, F 1 becomes weakened. According to the rules of electron arrangements outside the nuclei, Ti 4+ ions have the tendency to attract free electrons to ll their empty orbitals, which implies that Li 4 Ti 5 O 12 particles can attract the moving charges and in uence the charge movement via the Coulombic force F 2 . Meanwhile, the moving charges are affected not only by F 1 but also by F 2 when they enter the LTO semiconductive composites. Figure 11b presents the results of force analysis of the moving charges in the LTO semiconductive composites, indicating that charge implantation is inhibited by F 2 .

Space charge distribution
It is now established that Li 4 Ti 5 O 12 particles agglomerate at Li 4 Ti 5 O 12 contents above 4 wt.%, which reduces the effective concentration of Li 4 Ti 5 O 12 and weakens the E in eld. Furthermore, the oversized grains increase the interface roughness, which will distort the interfacial electric eld. This changes the force on the moving charges, increases F 1 , and decreases F 2 , as shown in Fig. 11c.
In summary, the force on the moving charges in the semiconductive screen is strongly related to the

Conclusions
The results of this study suggest that the addition of a certain amount of ionic conductor Li 4 Ti 5 O 12 can enhance the composites conductive network, mitigate the electric eld distortion, and effectively weaken the space charge injection into the insulating layer. Its main conclusions are summarized as follows: When the Li 4 Ti 5 O 12 content is 4 wt.%, both the PTC effect and charge implantation process are signi cantly inhibited. The peak resistivity of the obtained composite was 48.4% lower than that of the sample without Li 4 Ti 5 O 12 . Simultaneously, the minimum charge amount in the insulation layer was reduced by 49.7% at an electric eld strength of 10 kV, by 58.9% at 20 kV/mm, and by 46.9% at 30 kV/mm. Moreover, the minimum peaks of the depolarization currents obtained at 10, 20, and 30 kV/mm were reduced by 44%, 37%, and 42%, respectively. Because the space charge transport is strongly in uenced by the external electric eld, the induced electric eld and Coulombic force on the polarized Li 4 Ti 5 O 12 particles can counteract a fraction of the external electric eld. The horizontal component of the Coulombic force may also prevent charge migration to the insulating layer, and the vertical component of the Coulomb force de ects the charge migration direction in the LTO semiconductive composites. Thus, a considerable number of charges become blocked at the interface between the semiconductive and insulating layers.
When the Li 4 Ti 5 O 12 concentration reaches 5 wt.%, the PTC effect becomes more pronounced (as compared with that of the sample containing 4 wt.% Li 4 Ti 5 O 12 ), and the charge accumulation process is facilitated by the agglomeration of Li 4 Ti 5 O 12 particles. This decreases the effective concentration of Li 4 Ti 5 O 12 particles and increases the interfacial roughness of the nanocomposites, which in turn promotes the electric eld distortion and charge implantation processes. Therefore, the ndings of this work may help develop novel and e cient semiconductive screening materials for HVDC cables.