Enhanced Power Generation by Piezoelectric P(VDF-TrFE)/rGO Nanocomposite Thin Film

In this study we fabricated a piezoelectric nanogenerator (PENG) of nanocomposite thin film comprising a conductive nanofiller of reduced graphene oxide (rGO) dispersed in a poly(vinylidene fluoride-co-trifluoroethylene) (P(VDF-TrFE)) matrix that was anticipated to show enhanced energy harvest performance. For the film preparation we employed the Langmuir-Schaefer (LS) technique to provide direct nucleation of the polar β-phase without any traditional polling or annealing process. We prepared five PENGs consisting of the nanocomposite LS films with different rGO contents in the P(VDF-TrFE) matrix and optimized their energy harvest performance. We found that the rGO-0.002 wt% film yielded the highest peak-peak open-circuit voltage (VOC) of 88 V upon bending and releasing at 2.5 Hz frequency, which was more than two times higher than the pristine P(VDF-TrFE) film. This optimized performance was explained by increased β-phase content, crystallinity, and piezoelectric modulus, and improved dielectric properties, based on scanning electron microscopy (SEM), Fourier transform infrared (FT-IR), x-ray diffraction (XRD), piezoelectric modulus, and dielectric property measurement results. This PENG with enhanced energy harvest performance has great potential in practical applications for low energy power supply in microelectronics such as wearable devices.


Introduction
Poly(vinylidene fluoride) (PVDF) and its copolymers such as poly(vinylidene fluorideco-trifluoroethylene) (P(VDF-TrFE)), have potential for self-powered flexible electronic devices due to their properties such as thermal stability, mechanical strength, resistance to acids and halogens, light weight, and flexibility, as well as their ferroelectric, dielectric, and piezoelectric properties. Mechanical energy from human body motion can be harvested using piezoelectric, pyroelectric, or triboelectric materials through piezoelectric nanogenerators (PENGs), pyroelectric nanogenerators (PyENGs), and triboelectric nanogenerators (TENGs), respectively [1][2][3]. Among them, the PENGs, which transform mechanical energy into electricity, are advantageous because they are known to have superior energy performance due to high effectiveness and efficiency in harvesting energy from environmental situations such as human motion and structural vibration [4], and thus they can be integrated with device components such as capacitors and rechargeable batteries to store energy with high efficiency.

Materials
Poly(vinylidenefluoride-co-trifluoroethylene) (P(VDF:TrFE)= 70:30 mol%) and reduced graphene oxide (rGO, research grade powder, surface area of 103 m 2 /g) were used as the polymer matrix and the nanofiller, respectively. An indium tin oxide-coated polyethylene terephthalate (ITO-PET) sheet (8 cm × 2.5 cm) was used as a solid substrate for film deposition by the LS technique. N,N-dimethylformamide (DMF, 99.8%) and acetone (99.7%) were used as solvents. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) and used as received without further purification.

Preparation of P(VDF-TrFE)/rGO Solution and LS Thin Film
A nanocomposite solution of 50 g was prepared by dissolving 25 mg of P(VDF-TrFE) and dispersing a specific amount of rGO in a mixed solvent of DMF/acetone (v/v 40/60) 59 mL. Four solutions were prepared by dispersing different amount of rGO in the P(VDF-TrFE) solution, which were P(VDF-TrFE) solutions containing 0.002, 0.004, 0.02, and 0.04 wt% rGO, and then they were sonicated for 15 min. A clean Langmuir trough of Teflon was filled with deionized water (18 MΩ•cm), and the P(VDF-TrFE)/rGO solution was spread on the air/water interface at room temperature (23 °C) using a microsyringe of 500 μL. A monolayer of P(VDF-TrFE)/rGO was compressed to a target surface pressure by moving two barriers inward at 10 mm/min, and was horizontally transferred to the ITO/PET surface. A thin film of multilayers was prepared by repeating the monolayer deposition procedure five times. The thin film was then dried in the air at room temperature and stored in a desiccator for further analysis and/or experiments. This procedure of thin film preparation is illustrated in Figure 1a.

Fabrication of PENG Device
A PENG device was fabricated from the LS film. The film deposition area (about 6 cm × 2.5 cm) of two ITO-PET sheets were cut and placed together touching each other at the coating area. To make a sandwich structure, two new ITO-PETs were placed on the top of the coating area at each side. The ITO coatings of the two ITO-PETs (about 2 cm × 2.5 cm) were positioned to protrude to the opposite side and were used as electrodes connected with alligator clips. Figure 1b shows a picture of the PENG device and an illustration of the sandwich structure. Nonconductive adhesive tape was wrapped tightly around the device to prevent any breakage during bending and releasing.

Fabrication of PENG Device
A PENG device was fabricated from the LS film. The film deposition area (about 6 cm × 2.5 cm) of two ITO-PET sheets were cut and placed together touching each other at the coating area. To make a sandwich structure, two new ITO-PETs were placed on the top of the coating area at each side. The ITO coatings of the two ITO-PETs (about 2 cm × 2.5 cm) were positioned to protrude to the opposite side and were used as electrodes connected with alligator clips. Figure 1b shows a picture of the PENG device and an illustration of the sandwich structure. Nonconductive adhesive tape was wrapped tightly around the device to prevent any breakage during bending and releasing.

Characterization
Surface morphology and microstructure were observed using a scanning electron microscopy (SEM, JSM-6700F, JEOL Ltd., Tokyo, Japan). To examine the crystalline structure of the thin film, an X-ray diffractometer (XRD, D/Max2500, Rigaku, Tokyo, Japan) and FT-IR spectroscopy (Nicolet iS50, Thermo Fisher Scientific, Waltham, MA, USA) were used. The XRD was operated with Cu target radiations in the 2θ range of 10 to 30 • with a step width of 0.02 • and a scan speed of 1.0 • /min. A quasi-static d 33 m (Han Tech, PM 3500, Seoul, South Korea) was used to measure the average d 33 constant after taking five measurements for each sample. A LCR meter (HIOKI 3532 LCR HITESTER, Plano, TX, USA) was used to calculate dielectric properties using a two-point probe system. The output voltage generated by the PENG was measured using a functional digital storage oscilloscope (EDUX1002G) at 1 and 2.5 Hz bending and releasing frequencies using a motorized bending machine Exi-Servo (EzS-NDR-42L-A-D).

Characteristics of Nanocomposite Monolayer
As the monolayer is compressed, the area (A) of spread monolayer decreases with almost same number of slowly soluble P(VDF-TrFE) molecules and rGO particles on the surface. Thus, the distance among the molecules and the particles decreases and the surface pressure (π) increases due to a higher repulsive intermolecular force among monolayer molecules and/or particles. This relationship between π and A is usually measured at a constant temperature during the compression, which is called the surface pressure-area (π-A) isotherm. Essential information concerning the nanocomposite monolayer can be acquired from the surface pressure-area (π-A) isotherm, and morphological properties of the film are usually affected by the deposition conditions [33]. Figure 2 shows the π-A isotherms of spread monolayers of five samples. As shown in Table 1, as the rGO content increased from 0 to 0.02 wt%, the surface area increased from 79.6 to 106.9 cm 2 (34.3%) at π = 5 mN/m and from 69.7 to 102.4 cm 2 (46.9%) at π = 15 mN/m, respectively. This suggests that the more rGO particles disperse in the unit area of monolayer, the higher the repulsive intermolecular force exerted among P(VDF-TrFE) molecules and rGO particles. More surface area was taken by a higher content of rGO because of the higher repulsive force among molecules and particles. It should be noted that a spread monolayer of P-rGO-0.001 was also examined and yielded a π-A isotherm undistinguishable from that of pristine P(VDF-TrFE), within error. We presumed that a smaller rGO content than 0.002 wt% did not affect monolayer properties significantly, and thus other film properties, so we did not investigate other characteristics and piezoelectric performance of this thin film further.

Characterization
Surface morphology and microstructure were observed using a scanning electron microscopy (SEM, JSM-6700F, JEOL Ltd., Tokyo, Japan). To examine the crystalline structure of the thin film, an X-ray diffractometer (XRD, D/Max2500, Rigaku, Tokyo, Japan) and FT-IR spectroscopy (Nicolet iS50, Thermo Fisher Scientific, Waltham, MA, USA) were used. The XRD was operated with Cu target radiations in the 2θ range of 10 to 30° with a step width of 0.02° and a scan speed of 1.0°/min. A quasi-static d33 m (Han Tech, PM 3500, Seoul, South Korea) was used to measure the average d33 constant after taking five measurements for each sample. A LCR meter (HIOKI 3532 LCR HITESTER, Plano, TX, USA) was used to calculate dielectric properties using a two-point probe system. The output voltage generated by the PENG was measured using a functional digital storage oscilloscope (EDUX1002G) at 1 and 2.5 Hz bending and releasing frequencies using a motorized bending machine Exi-Servo (EzS-NDR-42L-A-D).

Characteristics of Nanocomposite Monolayer
As the monolayer is compressed, the area (A) of spread monolayer decreases with almost same number of slowly soluble P(VDF-TrFE) molecules and rGO particles on the surface. Thus, the distance among the molecules and the particles decreases and the surface pressure (π) increases due to a higher repulsive intermolecular force among monolayer molecules and/or particles. This relationship between π and A is usually measured at a constant temperature during the compression, which is called the surface pressurearea (π-A) isotherm. Essential information concerning the nanocomposite monolayer can be acquired from the surface pressure-area (π-A) isotherm, and morphological properties of the film are usually affected by the deposition conditions [33]. Figure 2 shows the π-A isotherms of spread monolayers of five samples. As shown in Table 1, as the rGO content increased from 0 to 0.02 wt%, the surface area increased from 79.6 to 106.9 cm 2 (34.3%) at π = 5 mN/m and from 69.7 to 102.4 cm 2 (46.9%) at π = 15 mN/m, respectively. This suggests that the more rGO particles disperse in the unit area of monolayer, the higher the repulsive intermolecular force exerted among P(VDF-TrFE) molecules and rGO particles. More surface area was taken by a higher content of rGO because of the higher repulsive force among molecules and particles. It should be noted that a spread monolayer of P-rGO-0.001 was also examined and yielded a π-A isotherm undistinguishable from that of pristine P(VDF-TrFE), within error. We presumed that a smaller rGO content than 0.002 wt% did not affect monolayer properties significantly, and thus other film properties, so we did not investigate other characteristics and piezoelectric performance of this thin film further.

Morphology Analysis by SEM
Morphology and crystalline structure were observed with a SEM. Figure 3 shows SEM images for the thin-film samples of P-rGO-0, P-rGO-0.002, P-rGO-0.004, P-rGO-0.02, and P-rGO-0.04. As seen in Figure 3, P-rGO-0.002, P-rGO-0.004, and P-rGO-0.02 showed some aggregations of less than one micron, but P-rGO-0.04 showed much larger particles with the size of a few microns (see the circles in the figure), which were presumably aggregations of rGO particles, whereas P-rGO-0 formed a film of better quality with high uniformity and homogeneity. This trend was supported by the π-A isotherm results. The π-A isotherms in the Figure 2 can be classified into three groups according to their relative position. The first group (P-rGO-0) was positioned in the left, the second one (P-rGO-0.04) was positioned in the right, and the third group (P-rGO-0.002, P-rGO-0.004, and P-rGO-0.02) was positioned in the middle. The SEM morphology results suggest that the thin film start to form large aggregates of micron dimensions at a rGO content of 0.04 wt%, presumably due to agglomeration of rGO particles. In this study, the P-rGO-0.04 sample was expected to provide lower piezoelectric performance because nonuniformity and heterogeneity in the film have been reported to reduce piezoelectric efficiency in the literature [34].

Morphology Analysis by SEM
Morphology and crystalline structure were observed with a SEM. Figure 3 shows SEM images for the thin-film samples of P-rGO-0, P-rGO-0.002, P-rGO-0.004, P-rGO-0.02, and P-rGO-0.04. As seen in Figure 3, P-rGO-0.002, P-rGO-0.004, and P-rGO-0.02 showed some aggregations of less than one micron, but P-rGO-0.04 showed much larger particles with the size of a few microns (see the circles in the figure), which were presumably aggregations of rGO particles, whereas P-rGO-0 formed a film of better quality with high uniformity and homogeneity. This trend was supported by the -A isotherm results. The -A isotherms in the Figure 2 can be classified into three groups according to their relative position. The first group (P-rGO-0) was positioned in the left, the second one (P-rGO-0.04) was positioned in the right, and the third group (P-rGO-0.002, P-rGO-0.004, and P-rGO-0.02) was positioned in the middle. The SEM morphology results suggest that the thin film start to form large aggregates of micron dimensions at a rGO content of 0.04 wt%, presumably due to agglomeration of rGO particles. In this study, the P-rGO-0.04 sample was expected to provide lower piezoelectric performance because nonuniformity and heterogeneity in the film have been reported to reduce piezoelectric efficiency in the literature [34].

FT-IR Analysis
FT-IR was used to analyze the crystalline phase of the P(VDF-TrFE) matrix and the interaction between rGO particles and P(VDF-TrFE) molecules. Figure 4 shows the FT-IR spectra of five thin films with detailed peaks in the range of 750-1500 cm −1 , and the analysis of the β phase content (%).

FT-IR Analysis
FT-IR was used to analyze the crystalline phase of the P(VDF-TrFE) matrix and the interaction between rGO particles and P(VDF-TrFE) molecules. Figure 4 shows the FT-IR spectra of five thin films with detailed peaks in the range of 750-1500 cm −1 , and the analysis of the β phase content (%).
(c) As seen in the Figure 4a, there were several distinct peaks in the FT-IR spectra. There were stretching vibration peaks of -CH2-around 2900 cm −1 , and peaks for P(VDF-TrFE) crystalline phases in the range of 750 to 1500 cm −1 . Table 2 shows characteristics of the latter peaks. Among those peaks, the peak by symmetric stretching of CF2 and CC was reported to have the β-phase, whereas the peak by asymmetric stretching of CC, and wagging of CH2 and CF2, was the α-phase [35][36][37][38]. These characteristics can be used to calculate the percent content of the β-phase. The percent β-phase content, F(β) can be calculated by the following equation [39]: As seen in the Figure 4a, there were several distinct peaks in the FT-IR spectra. There were stretching vibration peaks of -CH 2 -around 2900 cm −1 , and peaks for P(VDF-TrFE) crystalline phases in the range of 750 to 1500 cm −1 . Table 2 shows characteristics of the latter peaks. Among those peaks, the peak by symmetric stretching of CF 2 and CC was reported to have the β-phase, whereas the peak by asymmetric stretching of CC, and wagging of CH 2 and CF 2 , was the α-phase [35][36][37][38]. These characteristics can be used to calculate the percent content of the β-phase. The percent β-phase content, F(β) can be calculated by the following equation [39]: where A β and A α are the absorption intensities at 846 and 761 cm −1 , respectively, and K β and K α are the absorption coefficients of 7.7 × 10 4 and 6.1 × 10 4 cm 2 mol −1 , for the β-phase and the α-phase, respectively. The calculation results are shown in the Figure 4b (see the insert).  [36,37] (ν a ) asymmetric stretching, (ν σ ) symmetric stretching, (b) bending, ( w ) wagging, (r) rocking. The sign "−" and "+" indicate the phase relation between the symmetry coordinates.
With a 0.002 wt% rGO concentration, the β-phase percent content showed a maximum of 98%, and then decreased to 87% as the rGO concentration increased up to 0.04 wt%. This implies that at the concentration of rGO 0.002 wt% the P(VDF-TrFE) polymer chains are mostly oriented parallel to the substrate [40]. A tiny absorption peak around 3400 cm −1 suggests intermolecular hydrogen bonding between the P(VDF-TrFE) chain, -CH 2 -dipoles, and oxygen-containing functional groups of rGO. The stretching vibration peaks of -CH 2around 2900 cm −1 shifted towards lower frequency region, which indicated stronger dipolar interactions of -CH 2 -CF 2 -in the polymer matrix. These two attractive forces of hydrogen bonding and dipolar interaction helped dipole orientation in P(VDF-TrFE) chains in the nanocomposite film of P(VDF-TrFE)/rGO [41,42]. This result of optimum rGO content can be explained in terms of interaction between surface charge due to oxygencontaining functional groups such as C-O, C=O, COO-, OH and delocalized π electrons of rGO and -CH 2 -CF 2 dipoles in polymer chains [15,43]. This interaction accommodates the P(VDF-TrFE)/rGO nanocomposite film to form the β-phase easily. With a specific content of rGO, the interaction between surface charge of rGO and dipoles in polymer chains maximizes all-trans (TTTT) molecular conformation (i.e., the β-phase in the polymer matrix), but with a higher content of rGO it results in rGO particles oppositely aligned to the chain alignment, thus reducing β-phase development [24,44]. The β-phase of alltrans molecular conformation in the polymer matrix might be further enhanced by the preparation method in this study. As an explanation, the HO-H groups in the water subphase help build hydrogen bonds with C-F groups of the P(VDF-TrFE) chain in the spread monolayer, and thus the P(VDF-TrFE) molecules should be oriented with C-F groups inward to the water subphase [26], which is similar to the all-trans molecular conformation in the β-phase. This explanation is illustrated in Figure 4c.

XRD Analysis
An XRD was used to analyze the crystal structure of the nanocomposite thin films. Figure 5a shows XRD patterns for five nanocomposite thin films. All the patterns showed distinct peaks at about 2θ = 19.7, corresponding to the (110/220) planes, recognized as polar β-phases [45]. It is noted that the peak with the rGO concentration of 0.002 wt% was distinctly larger than those with other concentrations. This trend can be explained in terms of degree of crystallization due the interaction between rGO and polymer chains. At the rGO concentration of 0.002 wt%, there seemed to be a proper degree of interaction between rGO and the polymer chains, leading to optimum β-phase with the highest order and crystallinity in the polymer matrix. However, the presence of rGO more than the optimum amount triggered aggregation by creating more nucleation seeds, which reduced the formation of the β-phase crystal structure. Importantly, the increase in the βphase diffraction peak intensity was significantly influenced by uniform and homogeneous distribution of rGO in the polymer matrix.
of interaction between rGO and the polymer chains, leading to optimum β-phase with the highest order and crystallinity in the polymer matrix. However, the presence of rGO more than the optimum amount triggered aggregation by creating more nucleation seeds, which reduced the formation of the β-phase crystal structure. Importantly, the increase in the β-phase diffraction peak intensity was significantly influenced by uniform and homogeneous distribution of rGO in the polymer matrix. The slight peak shift that took place at 0.002 wt% was ascribed to the change in lattice parameter of P(VDF-TrFE). The crystallinity of thin film appeared to be increased compared to the pristine P(VDF-TrFE) thin film. The percentage crystallinity (XC) was calculated using the following equation [35]: where ∑ACrys and ∑AAmor are the total integral area under the crystalline and amorphous zones, respectively. As expected, the P-rG-0.002 showed a maximum degree of crystallinity of 56%, and P-rGO-0.04 a minimum of 25% (see Table 3). Similar trends of XRD pattern and crystallinity have been reported with a spin-coated P(VDF/TrFE)/rGO film in the literature [44]. Table 3. Degree of crystallinity (%) calculated from XRD patterns of five thin films.

Sample
Percentage Crystallinity (%) P-rGO-0 35 P-rGO-0.002 56 P-rGO-0.004 36 P-rGO-0.02 29 P-rGO-0.04 25 The slight peak shift that took place at 0.002 wt% was ascribed to the change in lattice parameter of P(VDF-TrFE). The crystallinity of thin film appeared to be increased compared to the pristine P(VDF-TrFE) thin film. The percentage crystallinity (X C ) was calculated using the following equation [35]: where ∑A Crys and ∑A Amor are the total integral area under the crystalline and amorphous zones, respectively. As expected, the P-rG-0.002 showed a maximum degree of crystallinity of 56%, and P-rGO-0.04 a minimum of 25% (see Table 3). Similar trends of XRD pattern and crystallinity have been reported with a spin-coated P(VDF/TrFE)/rGO film in the literature [44]. Table 3. Degree of crystallinity (%) calculated from XRD patterns of five thin films.

Piezoelectric Modulus and Dielectric Properties
The piezoelectric modulus (d 33 ) for five thin films of ten nanocomposite monolayers was measured by employing a quasi-static d 33 m at a constant force of 250 × 10 −3 N. The results are shown in Figure 5b. Among the five films, P-rGO-0.002 yielded the highest d 33 value of 98 (±2.2) pC/N, which is consistent with previous results such as β-phase content and crystallinity. According to the principle of the piezoelectric effect, an applied external force produces an electric potential field in the longitudinal direction, which is nonuniformly distributed in the nanocomposite thin film due to different dielectric properties of rGO and P(VDF-TrFE) matrix (see Figure S2). These results indicate that P-rGO-0.002 had the highest degree of additional potential field in the film because it is known that additional lateral potential field increases the piezoelectric modulus (d 33 ) [46].
The dielectric constant (ε) and dielectric loss (tan δ) were measured for five thin films at room temperature using a two point-probe LCR meter in the AC signals frequency range of 10 4 -10 6 Hz. The ε and tan δ values were calculated from the measured capacitance (C) using the following equation [47]: where ε o , d, and A denote dielectric constant of a vacuum (8.85 × 10 −12 Fm −1 ), the distance (i.e., film thickness), and the surface area of the electrode, respectively [47]. At 10 4 Hz frequency, the dielectric constant value was 17 for the pristine P(VDF-TrFE) film and reached a maximum value of 21 at a rGO concentration of 0.002 wt%. As shown in Figure S1a, this trend can be explained by a model of micro-capacitor formation [43]. In the model, the nanocomposite thin film with the optimum amount (0.002 wt%) of rGO shows higher dielectric constant than that for any other films because rGO particles form proper number of micro-capacitors. In other words, two neighboring rGO platelets act as electrodes, and the polymer P(VDF-TrFE) as a dielectric medium in the thin films. At the optimum concentration of rGO, a proper number of micro-capacitors are formed to increase the film capacitance with an enhanced dielectric constant by aligning the dipoles. However, above the optimum amount of the rGO concentration, the rGO particles start to aggregate, forming an electrically conductive percolated system, thus reducing dielectric properties. Moreover, a further increase in filler content creates loss of insulating property and the charge carriers are accumulated at the interfaces between filler and polymer [48,49]. According to studies in the literature, the higher ε value may be due to dipoles having enough time to align at lower frequencies (10 4 ). At higher frequencies (10 6 Hz), the dipoles do not have enough time for alignment with the applied AC field direction. Thus, the dielectric constant becomes reduced. The dielectric loss (tan δ) followed the same trend as dielectric constant (ε), as shown in Figure S1b. The highest dielectric loss at the rGO concentration of 0.002 wt% was ascribed to formation of more conductive pathways. The low value of tan δ might be due to polarization loss, conduction loss, and produced heat after colloidal formation of dipoles at high AC field frequencies [46]. Table 4 summarizes all the characteristics of the five nanocomposite thin films.

Energy Harvesting
Five different PENG devices with a sandwich structure and thin films of ten nanocomposite multilayers of different rGO content were subjected to the open-circuit voltage (V OC ) and short-circuit current (I SC ) measurements. An oscilloscope was used for the measurements when the devices were tested with different bending and releasing frequencies of 1 Hz and 2.5 Hz, and a bending angle of 50 • . As a result of bending and releasing, a piezoelectric potential is generated at the bottom and top electrodes of the PENG devices. Compressive deformation vanishes upon releasing the bending force; hence, the generated electrons flow back to regain the potential, which causes the voltage to change in the opposite direction. Therefore, positive and negative output voltages are generated upon bending and releasing a PENGs, and the current has the same increasing and decreasing trend as the output voltage. Figure 6 shows V OC signals and average peak-peak V OC of five PENGs at π = 5 and 15 mN/m. As seen in Figure 6a,b, the PENG with ten layers of P-rGO-0.002 showed the highest peak-peak V OC of 88 V at 2.5 Hz and π = 5 mN/m, whereas that with P-rGO-0 was about half the V OC at 42 V under the same conditions. The PENGs with same number of layers of P-rGO-0.004, P-rGO-0.02, and P-rGO-0.04 showed peak-peak V OC values of 74, 36 and 34 V, respectively. The decrease in the peak-peak V OC value with increased rGO concentration for P-rGO-0.004, P-rGO-0.02 and P-rGO-0.04 PENGs can be ascribed mainly to the aggregation of rGO. More aggregation of nanofiller particles has been known to decrease the flow rate of electrons by reducing the effective electrostatic fields in between nano-flakes and polymer molecules [45]. In addition, the large surface area of rGO may accelerate the generation of CO dipoles, which increases micro-capacitor formation to hamper piezoelectric performance, as discussed previously.
To check the contribution of ITO-PET substrate to voltage creation, V OC for the ITO-PET was measured under the same conditions, but no significant signals were detected.
electrons flow back to regain the potential, which causes the voltage to change in the opposite direction. Therefore, positive and negative output voltages are generated upon bending and releasing a PENGs, and the current has the same increasing and decreasing trend as the output voltage. Figure 6 shows VOC signals and average peak-peak VOC of five PENGs at π = 5 and 15 mN/m. As seen in Figure 6a,b, the PENG with ten layers of P-rGO-0.002 showed the highest peak-peak VOC of 88 V at 2.5 Hz and π = 5 mN/m, whereas that with P-rGO-0 was about half the VOC at 42 V under the same conditions. The PENGs with same number of layers of P-rGO-0.004, P-rGO-0.02, and P-rGO-0.04 showed peak-peak VOC values of 74, 36 and 34 V, respectively. The decrease in the peak-peak VOC value with increased rGO concentration for P-rGO-0.004, P-rGO-0.02 and P-rGO-0.04 PENGs can be ascribed mainly to the aggregation of rGO. More aggregation of nanofiller particles has been known to decrease the flow rate of electrons by reducing the effective electrostatic fields in between nano-flakes and polymer molecules [45]. In addition, the large surface area of rGO may accelerate the generation of CO dipoles, which increases micro-capacitor formation to hamper piezoelectric performance, as discussed previously. To check the contribution of ITO-PET substrate to voltage creation, VOC for the ITO-PET was measured under the same conditions, but no significant signals were detected. Figure 6. (a) VOC signals generated by five PENGs at π = 5 mN/m; (b) average peak-peak VOC of five PENGs at π = 5 mN/m; (c) VOC signals generated by five PENGs at π = 15 mN/m; (d) average peak-peak VOC of five PENGs at π = 15 mN/m. Figure 6. (a) V OC signals generated by five PENGs at π = 5 mN/m; (b) average peak-peak V OC of five PENGs at π = 5 mN/m; (c) V OC signals generated by five PENGs at π = 15 mN/m; (d) average peak-peak V OC of five PENGs at π = 15 mN/m. When the bending and releasing frequency was reduced to 1 Hz, the peak-peak V OC values decreased to lower values of less than half. When the surface pressure increased to π = 15 mN/m, the peak-peak V OC values for P-rGO-0.004, P-rGO-0.02, and P-rGO-0.04 were not significantly changed within error (less than 10%), whereas that for P-rGO-0 significantly increased (more than 50%), and that for P-rGO-0.002 decreased (about 20%). This indicates that the surface pressure was not a critical factor influencing piezoelectric performance. Integration of all these piezoelectric performance test results indicates that the thin film of ten nanocomposite monolayers with 0.002 wt% rGO deposited π = 5 mN/m yielded the best piezoelectric performance, and thus this film is called 'optimum' or 'optimized' thin film hereafter.
As shown in Figure 7a, the short-circuit I SC values showed similar trends to V OC . In addition, Figures S2 and S3 show similar trends of peak-peak V OC values for other film thicknesses with different number of layers. As shown in Figure 7b, the thicker film (as much as 40 nm (i.e., thickness of ten layers)) [50] yielded the larger V OC values. This result suggests that a more conductive network is formed with a thicker film, and the aggregates in the network presumably have a less negative effect on the piezoelectric network in the thicker film.
As shown in Figure 7a, the short-circuit ISC values showed similar trends to VOC. In addition, Figures S2 and S3 show similar trends of peak-peak VOC values for other film thicknesses with different number of layers. As shown in Figure 7b, the thicker film (as much as 40 nm (i.e., thickness of ten layers)) [50] yielded the larger VOC values. This result suggests that a more conductive network is formed with a thicker film, and the aggregates in the network presumably have a less negative effect on the piezoelectric network in the thicker film. For a practical application, resistance dependencies of voltage and current were investigated by measuring voltage (VL) as a function of resistance load and by calculating current (IL) in a short-circuit. Figure 7c shows the VL and IL results for the optimum thin film PENG. With increasing resistance, the VL steadily increased, and the IL decreased, and the rate of increase and decrease became smaller around the inflection point of RL = For a practical application, resistance dependencies of voltage and current were investigated by measuring voltage (V L ) as a function of resistance load and by calculating current (I L ) in a short-circuit. Figure 7c shows the V L and I L results for the optimum thin film PENG. With increasing resistance, the V L steadily increased, and the I L decreased, and the rate of increase and decrease became smaller around the inflection point of R L = 20 MΩ. Figure 7d shows the power density profile as a function of resistance with a maximum of 16.5 µW/cm 2 at R L = 20 MΩ, which was calculated using the following equation: The optimum PENG with high power density (P) can store the electrical potential in conjunction with different capacitors for practical use. To demonstrate feasibility of practical use of the optimum PENG, a simple circuit was designed to comprise several light emitting diodes (LEDs), a capacitor, and four rectifiers, as shown in the Figure 8a. Using the circuits with different capacitors (0.1, 0.22, 1, and 10 µF), the electrical energy could be successfully stored, and the LEDs could be lit as shown in the Figure S4. As shown in the potential-time and energy storage-time profiles (Figure 8b,c), the maximum voltage of 65 V (DC) could be stored using a 0.1 µF capacitor, and with a 1 µF capacitor the maximum energy storage was calculated to be 999 µJ by the formula E C = 1 2 CV 2 where V and C denote the voltage and the capacitance, respectively [51]. The energy generated from the optimized PENG can be used to drive a low power-consuming electronics such as wearable device. As such, compared to conventional PENGs reported in the literature, this optimum PENG with enhanced performance showed several distinct advantages, such as using a minimal amount of rGO nanofiller, the capability of coating a large area (LS film), a facile and cheap fabrication process, and superior energy harvesting performance. Table 5 compares characteristics and performance of the optimum PENG with those of PVDF-based PENGs reported in the literature. These advantages of the optimum PENG make it a promising candidate for practical applications of low energy power supply in microelectronics, such as wearable devices.
of 65 V (DC) could be stored using a 0.1 μF capacitor, and with a 1 μF capacitor the maximum energy storage was calculated to be 999 μJ by the formula EC = CV 2 where V and C denote the voltage and the capacitance, respectively [51]. The energy generated from the optimized PENG can be used to drive a low power-consuming electronics such as wearable device. As such, compared to conventional PENGs reported in the literature, this optimum PENG with enhanced performance showed several distinct advantages, such as using a minimal amount of rGO nanofiller, the capability of coating a large area (LS film), a facile and cheap fabrication process, and superior energy harvesting performance. Table 5 compares characteristics and performance of the optimum PENG with those of PVDF-based PENGs reported in the literature. These advantages of the optimum PENG make it a promising candidate for practical applications of low energy power supply in microelectronics, such as wearable devices.

Conclusions
A piezoelectric nanogenerator (PENG) of nanocomposite thin film comprising rGO particles dispersed in a P(VDF-TrFE)) matrix was fabricated by arranging the film coated ITO-PET substrates in a sandwich structure. The nanocomposite thin film was prepared by the Langmuir-Schaefer (LS) technique to maximize the content of the polar β-phase without any traditional polling or annealing process. Several PENGs with different rGO contents, surface pressures, and different film thickness were subject to SEM observation, FT-IR spectroscopy, XRD, piezoelectric modulus, dielectric properties, and piezoelectric properties, such as open-circuit output voltage (V OC ) measurements, to optimize film composition, density, and thickness. A PENG with a film of ten layers of nanocomposite monolayer containing 0.002 wt% rGO at π = 5 mN/m was determined to have optimal performance of peak-peak V OC of 88 V, energy storage of 999 µJ with 1 µF capacitor. These results are discussed in terms of the β-phase, crystallinity, piezoelectric modulus d 33 , and dielectric properties. The energy generated from the optimum PENG can be used to directly drive low-power-consuming electronics such as wearable devices. Importantly, the novel optimum PENG fabricated in our study revealed several advantages, including superior energy harvesting performance with a minimal amount of rGO nanofiller, the capability of coating a large area, and a facile and cheap fabrication process. Therefore, these advantages of the optimum PENG make it a promising candidate for practical applications of low energy power supply in microelectronics.

Data Availability Statement:
The data in this study are available upon request from the corresponding author.