Natural Rubber-TiO2 Nanocomposite Film for Triboelectric Nanogenerator Application

In this research, natural rubber (NR)-TiO2 nanocomposites were developed for triboelectric nanogenerator (TENG) application to harvest mechanical energy into electrical energy. Rutile TiO2 nanoparticles were used as fillers in NR material to improve dielectric properties so as to enhance the energy conversion performance of the NR composite TENG. The effect of filler concentration on TENG performance of the NR-TiO2 composites was investigated. In addition, ball-milling method was employed to reduce the agglomeration of TiO2 nanoparticles in order to improve their dispersion in the NR film. It was found that the TENG performance was significantly enhanced due to the increased dielectric constant of the NR-TiO2 composite films fabricated from the ball-milled TiO2. The TENG, fabricated from the NR-TiO2 composite using 24 h ball-milled TiO2 at 0.5%wt, delivered the highest power density of 237 mW/m2, which was almost four times higher than that of pristine NR TENG. Furthermore, the applications of the fabricated NR-TiO2 TENG as a power source to operate portable electronics devices were also demonstrated.


Introduction
Energy harvesting technologies have attracted great attention because of the significance in producing sustainable energy sources to overcome energy crisis and climate change. In addition, the rapidly increasing number of personal electronic devices and other components for the Internet of Things (IoT) platform leads to the increasing demand for energy. Triboelectric nanogenerator (TENG) is a mechanical energy harvesting device based on the combination of contact electrification and electrostatic induction effects [1]. TENG has gained much interest due to its high energy conversion efficiency with high power output, straightforward fabrication process, and low cost [2]. Apart from energy harvesting applications, TENGs also have the potential to be used for many self-powered sensor applications, including physical, chemical, gas, and liquid sensors [3][4][5][6].
A wide range of materials can be used to fabricate TENG; most of them are polymeric materials [7]. The common known materials are polytetrafluoroethylene (PTFE) [8,9], polydimethylsiloxane (PDMS) [10,11], polyvinylidenefluoride (PVDF) [12,13], and polymethyl methacrylate (PMMA) [14,15]. Natural rubber (NR) or polyisoprene is one of the natural polymers with good flexibility and strength employed in a wide range of applications [16]. Most of NR products, such as car tires, gloves, shoe insoles, and mattresses, involve the applications in direct contact with mechanical energy sources. NR is one of the triboelectric materials located in the triboelectric series possessing slightly negative polarity [7]. In this

Preparation of NR-TiO 2 Composite Films
The commercial NR latex (purchased from the Thai Rubber Latex Group Public Co., Ltd., Samut Prakan, Thailand) with a dry rubber content of 61% and rutile TiO 2 nanoparticles (Briture Co., Ltd., Hefei, China) were used in this work. NR latex and the as-received TiO 2 nanoparticles at 0.1, 0.2, 0.3, 0.4, and 0.5%wt were mixed by magnetic stirring for 5 min to ensure a homogeneous mixing. Then, 2 mL of the mixture was cast on an FTO substrate (Bangkok Solar Power Co., Ltd., Chachoengsao, Thailand) with an area of 4 cm × 4 cm so as to control the film thicknesses of approximately 0.5 mm. Three samples were prepared for each of the experimental conditions. The cast samples were then left to dry at room temperature for 4 days and cured at 80 • C for 2 h. In the present work, low curing temperature with long curing durations were employed in order to control the uniformity of the film top surface. Then, the samples were tested for the TENG performance, as described in Section 2.3.
In addition, the TiO 2 nanoparticles were dry ball-milled prior to mixing with NR latex. Yttria-stabilized zirconia balls and TiO 2 nanoparticles were put in a polyethylene (PE) plastic vial at the ball to a powder weight ratio (BPR) of 4:1. The ball-milling process was performed at a milling speed of 250 rpm for 6, 12, and 24 h. The ball-milled TiO 2 nanoparticles were then incorporated to NR latex following the same procedure, as described above. The composite films with ball-milled TiO 2 for 6, 12, and 24 h were labeled as "NR-TiO 2 -B6h, NR-TiO 2 -B12h, and NR-TiO 2 -B24h", respectively.

Material Characterizations
The morphologies and crystal structure of the composite films were investigated using a SEM (FEI, Helios Nanolab, Waltham, MA, USA) and an X-ray diffraction technique (XRD) Polymers 2021, 13, 2213 3 of 12 (PANalytical EMPYREAN, Malvern, UK), respectively. Dielectric constants were measured using an impedance analyzer (Keysight, E4990A, Colorado Springs, CO, USA) at room temperature. Chemical functional group analysis was performed using a fourier transform infrared spectroscopy (FTIR) (TENSOR27).

TENG Output Measurement
The output performances of the NR-based TENGs were tested by measuring electrical output voltage and current using a vertical contact-separation mode with a single electrode configuration. A PTFE sheet was used as a contact triboelectric material. The voltage and current output signals were acquired under the mechanical impact force of 10 N with impact frequency of 5 Hz using an oscilloscope (Tektronix DPO2002B, Tektronix China Ltd, Shang Hai, China) and a digital ammeter (Kiethley DMM6500, Tektronix China Ltd, Shang Hai, China), respectively.

Results
The electrical output of the NR-TiO 2 at 0.1-0.5%wt were measured under a vertical contact-separation mode with a single electrode configuration, as presented in Figure 1. PTFE was used as a contact material with negative triboelectric polarity. The electrical voltage and current were generated by the physical contact-separation of the NR-TiO 2 film and PTFE surfaces. When the surfaces are in contact, the electrification effect causes electrons to be transferred between the two materials, resulting in the formation of positive and negative charges on surfaces of NR-TiO 2 film and PTFE, respectively. When the two surfaces were separated, electrostatic induction of triboelectric charges allowed free electrons in the electrical contact to flow, neutralizing triboelectric charges on the surface. Under the repeated contact-separation, the alternative current was generated.

TENG Output Measurement
The output performances of the NR-based TENGs were tested by measuring ele cal output voltage and current using a vertical contact-separation mode with a single trode configuration. A PTFE sheet was used as a contact triboelectric material. The vol and current output signals were acquired under the mechanical impact force of 10 N impact frequency of 5 Hz using an oscilloscope (Tektronix DPO2002B, Tektronix C Ltd, Shang Hai, China) and a digital ammeter (Kiethley DMM6500, Tektronix China Shang Hai, China), respectively.

Results
The electrical output of the NR-TiO2 at 0.1-0.5%wt were measured under a ver contact-separation mode with a single electrode configuration, as presented in Figu PTFE was used as a contact material with negative triboelectric polarity. The elect voltage and current were generated by the physical contact-separation of the NR-TiO2 and PTFE surfaces. When the surfaces are in contact, the electrification effect causes trons to be transferred between the two materials, resulting in the formation of pos and negative charges on surfaces of NR-TiO2 film and PTFE, respectively. When the surfaces were separated, electrostatic induction of triboelectric charges allowed free trons in the electrical contact to flow, neutralizing triboelectric charges on the surface. der the repeated contact-separation, the alternative current was generated. The generated voltage and current of the fabricated NR-TiO2 TENGs using a ceived TiO2 powders are presented in Figure 2a,b, respectively. The electrical outpu the NR-TiO2 TENGs with as-received TiO2 increased with increasing TiO2 content were at the highest in the NR-TiO2 0.5%wt TENG, which were 78.4 V and 7.0 µA, res tively. However, the improvement of electrical output was not significant. It was pected that the as-received TiO2 nanoparticles were agglomerated, giving rise to the p dispersion in the NR matrix. Figure 1. Schematic diagram of the device configuration for measuring energy conversion performance with working mechanism of the fabricated TENG under a vertical-contact separation mode with single electrode configuration.

PTFE
The generated voltage and current of the fabricated NR-TiO 2 TENGs using as-received TiO 2 powders are presented in Figure 2a,b, respectively. The electrical outputs of the NR-TiO 2 TENGs with as-received TiO 2 increased with increasing TiO 2 content and were at the highest in the NR-TiO 2 0.5%wt TENG, which were 78.4 V and 7.0 µA, respectively. However, the improvement of electrical output was not significant. It was suspected that the as-received TiO 2 nanoparticles were agglomerated, giving rise to the poor dispersion in the NR matrix. In order to improve the dispersion in the NR matrix, TiO2 nanoparticles were milled for 6, 12, and 24 h periods, prior to mixing with the NR latex to form comp materials. The ball-milled TiO2 at 0.1-0.5%wt (same as above experiment) were add NR latex. Electrical output voltage and current of all the ball-milled NR-TiO2 TEN displayed in Figure 3 and are summarized in Figure 4. It was found that ball-milled helped to improve the electrical outputs of NR-TiO2 TENG, which increased with milling time. The dependence of electrical output on TiO2 concentration of the ball-m TiO2 TENG exhibited the same trend, as electrical output increased with increasing concentration. The addition of the 24-h-ball-milled TiO2 nanoparticles into NR si cantly improved TENG performance, and the highest output voltage of 113 V and cu of 9.8 µA was achieved from the NR-TiO2-B24h-0.5%wt TENG. The enhanceme TENG performance was attributed to the disintegration of TiO2 nanoparticles at long milling times, producing the well-dispersion in the NR polymer matrix. The role of nanoparticles on TENG performance will be further discussed in the dielectric prop in the following section. In order to improve the dispersion in the NR matrix, TiO 2 nanoparticles were ballmilled for 6, 12, and 24 h periods, prior to mixing with the NR latex to form composite materials. The ball-milled TiO 2 at 0.1-0.5%wt (same as above experiment) were added to NR latex. Electrical output voltage and current of all the ball-milled NR-TiO 2 TENG are displayed in Figure 3 and are summarized in Figure 4. It was found that ball-milled TiO 2 helped to improve the electrical outputs of NR-TiO 2 TENG, which increased with ball-milling time. The dependence of electrical output on TiO 2 concentration of the ballmilled TiO 2 TENG exhibited the same trend, as electrical output increased with increasing TiO 2 concentration. The addition of the 24-h-ball-milled TiO 2 nanoparticles into NR significantly improved TENG performance, and the highest output voltage of 113 V and current of 9.8 µA was achieved from the NR-TiO 2 -B24h-0.5%wt TENG. The enhancement of TENG performance was attributed to the disintegration of TiO 2 nanoparticles at long ball-milling times, producing the well-dispersion in the NR polymer matrix. The role of TiO 2 nanoparticles on TENG performance will be further discussed in the dielectric properties in the following section. In order to improve the dispersion in the NR matrix, TiO2 nanoparticles w milled for 6, 12, and 24 h periods, prior to mixing with the NR latex to form co materials. The ball-milled TiO2 at 0.1-0.5%wt (same as above experiment) were a NR latex. Electrical output voltage and current of all the ball-milled NR-TiO2 TE displayed in Figure 3 and are summarized in Figure 4. It was found that ball-mil helped to improve the electrical outputs of NR-TiO2 TENG, which increased w milling time. The dependence of electrical output on TiO2 concentration of the bal TiO2 TENG exhibited the same trend, as electrical output increased with increasi concentration. The addition of the 24-h-ball-milled TiO2 nanoparticles into NR cantly improved TENG performance, and the highest output voltage of 113 V and of 9.8 µA was achieved from the NR-TiO2-B24h-0.5%wt TENG. The enhance TENG performance was attributed to the disintegration of TiO2 nanoparticles at lo milling times, producing the well-dispersion in the NR polymer matrix. The role nanoparticles on TENG performance will be further discussed in the dielectric pr in the following section.    The SEM images of the plain NR film, NR-TiO2, NR-TiO2-B6h, NR-TiO2-B NR-TiO2-B24h composite films at TiO2 0.5%wt are displayed with the insets of th nanoparticle fillers in Figure 5. Clearly, the dispersion of TiO2 without the bal treatment was poor, as evidenced by the large agglomeration size of particles obs SEM images of TiO2 powders and NR composite film. The agglomeration of TiO particles was less observed in the ball-milled TiO2 powders, which was reduced creasing ball-milling times, contributing to the better dispersion in the NR compos   The SEM images of the plain NR film, NR-TiO2, NR-TiO2-B6h, NR-TiO2-B12h, and NR-TiO2-B24h composite films at TiO2 0.5%wt are displayed with the insets of their TiO2 nanoparticle fillers in Figure 5. Clearly, the dispersion of TiO2 without the ball-milling treatment was poor, as evidenced by the large agglomeration size of particles observed in SEM images of TiO2 powders and NR composite film. The agglomeration of TiO2 nanoparticles was less observed in the ball-milled TiO2 powders, which was reduced with increasing ball-milling times, contributing to the better dispersion in the NR composite films nanoparticles was less observed in the ball-milled TiO 2 powders, which was reduced with increasing ball-milling times, contributing to the better dispersion in the NR composite films accordingly. The physical appearances of the NR and NR-TiO 2 -B24h 0.1-0.5%wt composite films are presented in Figure 6. The transparency of the pure NR film decreased as the TiO 2 content increased. 2021, 13, x FOR PEER REVIEW 6 of 12 accordingly. The physical appearances of the NR and NR-TiO2-B24h 0.1-0.5%wt composite films are presented in Figure 6. The transparency of the pure NR film decreased as the TiO2 content increased.  The rutile phase of as-received and ball-milled TiO2 at 6, 12, and 24 h samples were confirmed by the XRD patterns as shown in Figure 7a (JCPDS No. 21-1276). This suggested that the ball-milling process did not change the crystal structure of the TiO2 nanoparticles. In this study, ball-milling was employed to break up the agglomerated particles and rutile phase is the most stable structure of TiO2; therefore, it should not cause the microstructural change of the particles. FTIR analysis of the NR and NR-TiO2-B24h 0.5%wt was performed and presented in Figure 7b. FTIR spectra of the NR and NR-TiO2-B24h 0.5%wt film are relatively similar, consisting of C-H stretching at 2850-2960 cm −1 and 1300-1400 cm −1 and C= C stretching at 839 cm −1 of polyisoprene molecules [30], and some C-O hydroxyl groups from non-rubber components in latex such as inorganic substances, proteins, phospholipids, carbohydrates, and fatty acids [16,31]. This suggested that no accordingly. The physical appearances of the NR and NR-TiO2-B24h 0.1-0.5%wt site films are presented in Figure 6. The transparency of the pure NR film decrease TiO2 content increased.  The rutile phase of as-received and ball-milled TiO2 at 6, 12, and 24 h sampl confirmed by the XRD patterns as shown in Figure 7a (JCPDS No. 21-1276). Th gested that the ball-milling process did not change the crystal structure of the TiO particles. In this study, ball-milling was employed to break up the agglomerated p and rutile phase is the most stable structure of TiO2; therefore, it should not ca microstructural change of the particles. FTIR analysis of the NR and NR-TiO 0.5%wt was performed and presented in Figure 7b. FTIR spectra of the NR and N B24h 0.5%wt film are relatively similar, consisting of C-H stretching at 2850-2960 c 1300-1400 cm −1 and C= C stretching at 839 cm −1 of polyisoprene molecules [30], an C-O hydroxyl groups from non-rubber components in latex such as inorganic sub proteins, phospholipids, carbohydrates, and fatty acids [16,31]. This suggested chemical bond was formed between TiO2 and NR polymer. The rutile phase of as-received and ball-milled TiO 2 at 6, 12, and 24 h samples were confirmed by the XRD patterns as shown in Figure 7a (JCPDS No. . This suggested that the ball-milling process did not change the crystal structure of the TiO 2 nanoparticles. In this study, ball-milling was employed to break up the agglomerated particles and rutile phase is the most stable structure of TiO 2 ; therefore, it should not cause the microstructural change of the particles. FTIR analysis of the NR and NR-TiO 2 -B24h 0.5%wt was performed and presented in Figure 7b. FTIR spectra of the NR and NR-TiO 2 -B24h 0.5%wt film are relatively similar, consisting of C-H stretching at 2850-2960 cm −1 and 1300-1400 cm −1 and C=C stretching at 839 cm −1 of polyisoprene molecules [30], and some C-O hydroxyl groups from non-rubber components in latex such as inorganic substances, proteins, phospholipids, carbohydrates, and fatty acids [16,31]. This suggested that no chemical bond was formed between TiO 2 and NR polymer. where ɛ0, S, d0, x(t), and v(t) are electrical permittivity of free space, contact area fective thickness constant, separation distance, and contact electrode velocity, tively.
Triboelectric charge density depends on the material contact couple, contact well as the charge storing ability of the surface. In the latter case, it refers to the d constant of the material. For a contact-separation mode TENG which can be con by a capacitive model, triboelectric charge is proportional to the capacitance of the which is given by [18] = where εr is dielectric constant and d is thickness of triboelectric material.
Dielectric constants of the NR-TiO2-B24h 0.1-0.5%wt films measured at the f cies ranging from 10 2 -10 8 Hz is presented in Figure 8. The dielectric constant at 1 the NR-TiO2-B24h was found to increase with TiO2 concentration. The improve dielectric constant in the NR-TiO2-B24h films with increasing TiO2 concentration cribed to the fact that TiO2 has a greater dielectric constant than NR. The additio creasing TiO2 filler concentration to NR polymer matrix gave rise to the increasing tric constant of the composites. The dielectric constant contributed to the charge tance at the surfaces of triboelectric materials, which intensified triboelectric char and short circuit current (I sc ) is given by where ε 0 , S, d 0 , x(t), and v(t) are electrical permittivity of free space, contact area size, effective thickness constant, separation distance, and contact electrode velocity, respectively. Triboelectric charge density depends on the material contact couple, contact area, as well as the charge storing ability of the surface. In the latter case, it refers to the dielectric constant of the material. For a contact-separation mode TENG which can be considered by a capacitive model, triboelectric charge is proportional to the capacitance of the device, which is given by [18] where ε r is dielectric constant and d is thickness of triboelectric material. Dielectric constants of the NR-TiO 2 -B24h 0.1-0.5%wt films measured at the frequencies ranging from 10 2 -10 8 Hz is presented in Figure 8. The dielectric constant at 1 kHz of the NR-TiO 2 -B24h was found to increase with TiO 2 concentration. The improvement of dielectric constant in the NR-TiO 2 -B24h films with increasing TiO 2 concentration was ascribed to the fact that TiO 2 has a greater dielectric constant than NR. The addition of increasing TiO 2 filler concentration to NR polymer matrix gave rise to the increasing dielectric constant of the composites. The dielectric constant contributed to the charge capacitance at the The dependence of the output performance on the contact-separation frequency were also studied. The voltage and current outputs of the NR-TiO2-B24h 0.5%wt TENG were measured at operation frequencies ranging from 2-10 Hz, as presented in Figure 9a,b respectively. It was found that electrical outputs depended on working frequency, and that the highest peak-to-peak voltage and current were 204 V and 13 µA, respectively, a a working frequency of 10 Hz. The increased electrical output was caused by charge re tention on the surface due to a short contact-separation cycle at high frequencies.
The delivered power density of the NR-TiO2 TENG was also studied by measuring voltage and current at different load resistances ranging from 1-100 MΩ. The plot of volt age and current versus load resistances is shown in Figure 10a. The working power den sity of 200-237 mW/m 2 was achieved at load resistances ranging from 3-20 MΩ and the maximum power density of 237 mW/m 2 was achieved at a matched load resistance of 7 MΩ (Figure 10b), which was 3.6 times larger than that of pristine NR TENG (66 mW/m 2 ) This electrical output was enough to charge up a 10, 22, and 47 µF capacitors, as presented in a voltage profile in Figure 10c, and was able to charge a 99 µF to operate a portable calculator and light up 60 green LEDs, as demonstrated in Figure 10d and Video S1 in the Supplementary Materials. In addition, a TENG device was fabricated which was able to light up 21 green LEDs by hand pressing, as demonstrated with the inset showing the schematic diagram of device components in Figure 10e. The dependence of the output performance on the contact-separation frequency were also studied. The voltage and current outputs of the NR-TiO 2 -B24h 0.5%wt TENG were measured at operation frequencies ranging from 2-10 Hz, as presented in Figure 9a,b, respectively. It was found that electrical outputs depended on working frequency, and that the highest peak-to-peak voltage and current were 204 V and 13 µA, respectively, at a working frequency of 10 Hz. The increased electrical output was caused by charge retention on the surface due to a short contact-separation cycle at high frequencies. The dependence of the output performance on the contact-separation frequency also studied. The voltage and current outputs of the NR-TiO2-B24h 0.5%wt TENG measured at operation frequencies ranging from 2-10 Hz, as presented in Figure  respectively. It was found that electrical outputs depended on working frequency that the highest peak-to-peak voltage and current were 204 V and 13 µA, respective a working frequency of 10 Hz. The increased electrical output was caused by char tention on the surface due to a short contact-separation cycle at high frequencies.
The delivered power density of the NR-TiO2 TENG was also studied by meas voltage and current at different load resistances ranging from 1-100 MΩ. The plot of age and current versus load resistances is shown in Figure 10a. The working power sity of 200-237 mW/m 2 was achieved at load resistances ranging from 3-20 MΩ an maximum power density of 237 mW/m 2 was achieved at a matched load resistanc MΩ (Figure 10b), which was 3.6 times larger than that of pristine NR TENG (66 mW This electrical output was enough to charge up a 10, 22, and 47 µF capacitors, as pres in a voltage profile in Figure 10c, and was able to charge a 99 µF to operate a po calculator and light up 60 green LEDs, as demonstrated in Figure 10d and Video S1 Supplementary Materials. In addition, a TENG device was fabricated which was a light up 21 green LEDs by hand pressing, as demonstrated with the inset showin schematic diagram of device components in Figure 10e. The delivered power density of the NR-TiO 2 TENG was also studied by measuring voltage and current at different load resistances ranging from 1-100 MΩ. The plot of voltage and current versus load resistances is shown in Figure 10a. The working power density of 200-237 mW/m 2 was achieved at load resistances ranging from 3-20 MΩ and the maximum power density of 237 mW/m 2 was achieved at a matched load resistance of 7 MΩ (Figure 10b), which was 3.6 times larger than that of pristine NR TENG (66 mW/m 2 ). This electrical output was enough to charge up a 10, 22, and 47 µF capacitors, as presented in a voltage profile in Figure 10c, and was able to charge a 99 µF to operate a portable calculator and light up 60 green LEDs, as demonstrated in Figure 10d and Video S1 in the Supplementary Materials. In addition, a TENG device was fabricated which was able to

Discussion
In the present work, the development of NR TENG with enhanced performance was demonstrated by the incorporation of TiO 2 nanoparticles. The improved TENG performance was attributed to the enhanced triboelectric charge density by enhancing the dielectric constant of materials, as discussed in the previous section. TiO 2 nanoparticles were employed as an effective filler for improving dielectric constant of NR composite film due to the high dielectric constant of TiO 2 . However, the agglomeration of nanoparticles suppressed the dispersion of nanoparticles in the NR matrix leading to an insignificant improvement of TENG performance, as presented in Figure 2. In this work, the simple and efficient approach to reduce the agglomeration of TiO 2 nanoparticles using ball-milling was proposed. TiO 2 nanoparticles were ball-milled prior to mixing with NR latex, which was found to effectively reduce the agglomeration of nanoparticles, as evidenced by SEM images (Figure 5), which then consequently produced the well-dispersion of TiO 2 in NR-TiO 2 composite films. In this work, the milling time of 24 h was found to efficiently reduce the agglomeration and produce the uniformly dispersed TiO 2 in the NR films. The power output enhancement of the NR-TiO 2 -B24h was attributed to the improved dielectric constant due to the good dispersion of TiO 2 nanoparticles. This suggested that ball-milling was an effective treatment to alleviate the agglomeration of TiO 2 nanoparticles, which magnified the TENG electrical output to about 1.5 times higher than the untreated TiO 2 composite TENG.
Comparing to other previous reports, the fabricated TENG showed a superior performance than the PDMS-Kapton-implanted TENG with a power density of 8.44 mW/m 2 [33], the 2D woven wearable TENG fabricated from nylon and polyester threads with a power density of 2.33 mW/m 2 [34], and approaching a propeller TENG made of PTFE and Al triboelectric materials with a power density of 283.95 mW/m 2 [35]. In addition, comparing to the NR-based TENG, the NR-TiO 2 TENG exhibited the comparable output power to the NR-Ag TENG in our previous report which was 262.4 mW/m 2 [36]. The slightly lower TENG electrical output of the NR-TiO 2 composite than that of the NR-Ag one was attributed to the lower dielectric constant of the NR-TiO 2 . The conductive Ag filler produced stronger interfacial polarization than the TiO 2 semiconductor filler in the NR insulating matrix [37]. Therefore, the main contribution for the improved dielectric constant of NR-TiO 2 was from the intrinsic dielectric property of TiO 2 , as described earlier.
One of the most attractive aspects for employing NR as triboelectric material is the ability to scale up the production for large-area energy harvesting, owing to its low fabrication cost and feasibility to form composite with other materials. Comparing to other triboelectric polymers mentioned above, the costs of NR and TiO 2 are much lower. In addition, the fabrication process of NR-TiO 2 composite in the present work is straightforward, low cost and effective, which is promising for the development of large-scale energy harvesting device.

Conclusions
The NR-TiO 2 TENG for harvesting mechanical energy into electricity was successfully fabricated. The addition of rutile TiO 2 nanoparticles at 0.5%wt of NR latex to form NR-TiO 2 composite was found to enhance energy conversion efficiency of the TENG. The modification of TiO 2 by the ball-milling technique for 24 h prior to mix with NR materials was found to effectively disintegrate TiO 2 nanoparticles which consequently helped the dispersion of the nanoparticle fillers in the polymer matrix. Owing to the high dielectric constant of TiO 2 fillers, the dielectric constant of the NR-TiO 2 -B24h film increased with increasing TiO 2 concentration. The NR-TiO 2 -B24h film with improved dielectric constant attributed to the enhancement of TENG electrical output with the highest power density of 237 mW/m 2 . This work showed the potential applications of NR-TiO 2 TENG as an environmentally friendly power source for portable electronic devices.