Supersonically Sprayed Flexible ZnO/PVDF Composite Films with Enhanced Piezoelectricity for Energy Harvesting and Storage

A ﬂ exible piezoelectric nanogenerator (PENG) was developed to convert ambient mechanical energy into electrical energy. Highly crystalline template-free zinc oxide (ZnO) microrods were prepared using a wet chemical method at a low temperature. As-synthesized ZnO microrods were blended with polyvinylidene ﬂ uoride (PVDF) to prepare a piezoelectric composite ﬁ lm using the supersonic spraying method. A PENG electrode comprising ZnO microrods (0.5g) produced 15.2V under a tapping force of 20 N at 5Hz. In addition, an output current of 32 μ A was produced with a maximum power density of 12.5 μ Wcm – 2 . The output voltage considering various body movements was investigated to demonstrate the ﬂ exibility and durability of the proposed PENG. Furthermore, the supercapacitive properties of ZnO/reduced graphene oxide were studied.


Introduction
The development of energy-harvesting technologies to meet the increasing energy demand has garnered considerable attention in the research community [1][2][3]. The harvesting of ambient energy, e.g., vibrational, wind, thermal, and solar energies, is being actively explored for advanced energy-harvesting applications based on the pyroelectric, piezoelectric, and triboelectric principles [4,5]. In particular, mechanical energy harvesting using piezoelectric nanogenerators (PENGs) is a promising alternative source of clean energy owing to their stability in moist and dusty environments [6]. Piezoelectricity is defined as the potential difference that develops across the two surfaces of a piezoelectric film owing to the deformation of crystal struc-tures under a mechanical force that causes electrical charge to flow through an electrical circuit.
Wang and Song [20] used piezoelectric ZnO nanowire (NW) arrays to convert mechanical energy into electrical energy. In their experiment, a strain field was generated owing to the bending of ZnO NWs, leading to the distribution of charge across the NWs. Based on the aforementioned study, multifunctional PENGs have been developed for intelligent wearable devices and sensors owing to their low weight, scalability, high flexibility, and simple fabrication process [21][22][23][24]. However, despite the high piezoelectric coefficient (d 33 ) of inorganic ceramic materials, their rigidity and fragility pose a challenge to wearable PENG devices.
To overcome the brittleness of piezoceramics, the addition of piezopolymers to piezoceramics has been investigated; this addition enhances the overall piezoelectric performance regarding durability and output voltage [25]. Semicrystalline PVDF is an important piezoelectric polymer used in flexible PENGs owing to its reasonable piezoelectric coefficient, excellent thermal and chemical resistance, biocompatibility, and high flexibility [25,26]. Currently, five crystalline polymorphs of PVDF (α, β, γ, δ, and ε phases) have been identified [27]. Among them, the nonpolar α phase polymorph contributes to the mechanical strength of a PENG, whereas the β phase contributes to piezoelectricity. Consequently, the β phase amount must be increased to maximize power output. Several techniques, e.g., stretching, poling under a high electric field, electrospinning, and impregnation with inorganic and organic materials, have been employed to increase the polar and piezoactive β phase content in PVDF [25,26]. The incorporation of inorganic materials with different morphologies into PVDF enhances its piezoelectric behavior owing to an increase in the β phase content and subsequent interfacial polarization [28][29][30][31]. Similarly, adding PVDF to rod-like ZnO results in a composite with enhanced piezoelectric behavior, and wearable electronic devices based on the resultant ZnO/PVDF (ZP) composite exhibit enhanced flexibility and longevity. Moreover, the piezoelectric properties of ZP composites depend on the ZnO content, preparation, and interactions between the two materials [32,33].
In this study, ZnO microrods with PVDF were supersonically sprayed to fabricate piezocomposite films. Supersonic spray deposition is scalable, nonvacuum, and fast and, thus, considered a commercially viable method for industrial production [34][35][36][37][38]. In addition, the α phase of PVDF can be transformed into the polar β phase based on the shear stress inflicted by supersonic impact between ZnO and PVDF, thereby increasing the piezoelectricity and output power of the films [39]. This study determines the optimal combination of ZnO and PVDF that produces excellent piezoelectric properties. In addition, the excellent mechanical stability and durability of PENGs are demonstrated by introducing tapping, bending, and other body mechanical loads. Finally, the turbid white solution was stirred in a closed container at 80°C for 5 h (Figure 1(a)). The resulting white precipitate was washed several times with a 1 : 1 water-ethanol mixture and dried overnight in an oven at 80°C.  Table 1. Each suspension was vigorously stirred at 70°C for 10 h to homogenously distribute the ZnO microrods in the PVDF/DMF mixture and prevent agglomeration. The ZP composite films are hereafter denoted as ZP4, ZP5, and ZP6 based on the masses of the ZnO microrods (0.4, 0.5, and 0.6 g, respectively (Table 1)).

Experimental Methods
The obtained solutions were used to fabricate ZP composite films via supersonic spraying [40,41]. A typical supersonic spraying setup comprises an air compressor, air heater, and convergent-divergent supersonic nozzle (de Laval nozzle). An atomizer connected to a syringe pump was used to atomize the prepared solutions. In other words, the atomized solution of the composite material was accelerated by a supersonic air stream and deposited on a copper foil. The preheating temperature and pressure of the carrier gas (i.e., air in this study) during supersonic spraying were 250°C and 0.4 MPa, respectively. The supersonically blown composite material was deposited on a Cu foil attached to a X-Y moving stage (Figure 1(b)). The supersonic spraying technique has been explained in other studies [42,43].
2.4. Fabrication of ZnO/PVDF-Based PENGs. The PENGs developed herein were fabricated using a 3 cm × 3 cm ZP composite film, 5 cm × 5 cm Cu foil, and 5 cm × 5 cm bare Cu tape. Two 0:5 cm × 5 cm Cu foil stripes acted as the positive and negative terminals (Figure 1(c)). Finally, the Cu foil | ZP composite film | Cu tape was encapsulated and 2 International Journal of Energy Research insulated using a nonconductive polyimide tape (Figure 1(c)). A tapping instrument (Piezoelectric Generator Measurement System, SnM, Republic of Korea), bending machine (Ocean Science, Coad. 722, Republic of Korea), and oscilloscope (DS1052E, RIGOL, China) were used to explore the kinetic energy-harvesting performance of the ZP-based PENGs. To measure the current and power, different load resistances (0.01-10 MΩ) were connected in series to the PENG. As shown in Figure 1(   3 International Journal of Energy Research annealed in an Ar environment at 500°C for 1 h at a heating rate of 3°C/min. The annealed specimens were punched into circles with a diameter of 14 mm and used as electrodes in a CR2032 symmetric coin cell. A 6 M KOH was used as the electrolyte, and both electrodes were electrically separated using a polymer membrane (Celgard, 3501, Chungbuk) [44]. Electrochemical measurements were conducted using cyclic voltammetry (CV) and galvanostatic chargedischarge (GCD) measurements. A VersaSTAT-3 potentiostat (Princeton Applied Research, USA) with a twoelectrode symmetric cell system was used for the CV measurements. GCD measurements were conducted using a battery cycler (WBCS3000, WonATech Inc.).
2.6. Characterizations. The crystallinity of the ZnO microrods and the resulting ZP composite films were investigated using X-ray diffraction (XRD, SmartLab, Rigaku, Japan) considering Cu-Kα radiation (λ = 1:5406 Å) at a rate of 2°/ min. Fourier-transform infrared spectroscopy (FTIR; Spectrum GX, PerkinElmer Inc., USA) was used to analyze the interaction between ZnO and PVDF in the ZP-based composite films. The vibrational mode study confirms the structural transformation of PVDF from the α phase to the β phase [45]. Raman spectra were obtained using a Raman spectrometer and 532 nm laser (LabRam ARAMIS IR2, HORIBA Jobin Yvon, Japan) to observe the influence of ZnO microrods on the phases of PVDF. X-ray photoelectron spectroscopy (XPS; Thermo Fisher Scientific Co., USA) was performed with monochromatic Al K α as an X-ray source to analyze the chemical state of the specimens. The morphology of the ZnO was examined using field-emission scanning electron microscopy (FE-SEM; S-5000, Hitachi, Ltd., Japan) and transmission electron microscopy (TEM; JEM 2100F, JEOL, Inc., Japan). The specimens were prepared for TEM analysis using the focused ion beam technique (FEI Helios G4 UX, Thermo Fisher Scientific, USA). The elemental mapping of the specimens was performed using an energydispersive spectrometer (EDS) installed in the transmission electron microscope. High-resolution TEM (HRTEM) and selective area electron diffraction (SAED) were performed to confirm the crystal structures of the specimens. Atomic force microscopy (AFM; XE-100, Park System, South Korea) was performed in the piezoelectric force microscopy (PFM) mode to analyze the surface roughness and effective piezoelectric coefficient.  [46]. Furthermore, the XRD patterns of the ZP-based composite films revealed typical and distinct ZnO diffraction peaks of PVDF at 18.4°, 20.1°, and 20.9°, corresponding to the semicrys-talline nature of PVDF. The most prominent peak, observed at 20.1°, represents the (110) plane of PVDF; particularly, this peak corresponds to the β phase of PVDF and is associated with the piezoelectric characteristics of PVDF [47].

Results and Discussion
The FTIR spectra of the composite films were characterized to compare the interactions of the ZnO microrods and chemical environment of the PVDF phases. The FTIR spectra of pristine PVDF, ZP4, ZP5, and ZP6 are presented in Figure 2(b). The peaks at 613, 761, 796, 975, and 1381 cm -1 are assigned to the nonpolar α phase of PVDF, whereas the peaks observed at 839, 1275, and 1457 cm -1 are assigned to the β phase [48,49]. The characteristic vibrational peak of the β phase at 839 cm -1 corresponds to CH 2 rocking, CF 2 stretching, and skeletal C-C stretching, whereas that at 1275 cm -1 corresponds to CF out-of-plane deformation [49]. The incorporation of ZnO microrods into the PVDF matrix led to the sharpening of peaks related to the β phase of PVDF. Specifically, with an increase in the number of ZnO microrods, the peaks at 839 and 1275 cm -1 became more prominent, whereas the peak at 761 cm -1 decreased in the cases of the ZP4-, ZP5-, and ZP6-based composite films (Figure 2(b)). The β phase fraction of the composite films was calculated based on the characteristic absorbance peaks of the α and β phases using Eq. (1) [49] where FðβÞ denotes the percentage composition of the electroactive β phase, A α and A β denote the area of absorbance peaks at 761 and 839 cm -1 , respectively, and 1.26 is the ratio of the absorbance coefficient of the β phase (K β ) at 839 cm -1 (7:7 × 10 4 cm 2 mol -1 ) to that of the α phase (K α ) at 761 cm -1 (6:1 × 10 4 cm 2 mol -1 ). The β phase amounts of the pristine PVDF-, ZP4-, ZP5-, and ZP6-based composite films were 37.1%, 42.2%, 47.0%, and 45.5%, respectively. This result indicates that the β phase concentration of the PVDF polymer increased owing to the effective interaction between the ZnO microrods and PVDF molecules. Various fabrication processes, e.g., electrospinning, solvent casting, and spin coating, contribute to the phase transformation of PVDF polymers owing to polymer chain stretching or elongation in the presence of fillers [50]. Similarly, β phase enhancement is possible owing to the stretching of PVDF in the presence of a filler (ZnO) during supersonic spraying. In contrast to ZP4 and ZP5, ZP6 composite films exhibit a slight reduction in the β phase, which could be attributed to the reduced stretching of PVDF in a dense ZP precursor solution [51].
In addition to the XRD and FTIR analyses, the ZP-based composite films were studied using Raman spectroscopy. The Raman spectra of the ZP4-, ZP5-, and ZP6-based composite films were obtained in the range of 600-1500 cm -1 to analyze the influence of the ZnO microrods on the PVDF matrix (Figure 2(c)). With an increase in the number of ZnO microrods, the Raman bands related to the α phase at 796.8 and 876.3 cm -1 were diminished, whereas the Raman band at 840 cm -1 corresponding to the β phase was enhanced. This observation shown in the inset of 4 International Journal of Energy Research Figure 2(c) illustrates the enlarged spectra in the range of 770-900 cm -1 . The peak at 1430 cm -1 corresponds to the three phases of the PVDF polymer. As the α-peak intensities diminished with an increase in the ZnO content and the β phase intensities increased, the all-transplanar zigzag (TTTT) conformation increased [30,52]. According to the FTIR results, ZP5 with 0.5 g of ZnO microrods exhibited the highest percentage of β phase PVDF, and hence, its elemental oxidation states were analyzed using XPS. The surface chemical composition and interfacial interaction between ZnO and PVDF were investigated in the binding energy range of 0-1200 eV. The XPS survey spectra (Figure 2(d)) of the synthesized particles and ZP5 confirm the presence of Zn and O. Moreover, the XPS survey spectrum of ZP5 indicates the dominant presence of C and F originating from PVDF. XPSPEAK41 was used to analyze the core XPS spectra of various elements by subtracting the Shirley background. The Zn 2p core spectrum of the ZnO powder exhibits spin-orbital splitting in Zn 2p 3/2 and Zn 2p 1/2 , with peaks at 1021.4 and 1044.5 eV, respectively (Figure 2(e)) [53,54]. Furthermore, the binding energy difference between the two orbital levels is 23.1 eV, confirming the existence of the Zn 2+ oxidation state of ZnO [54]. The core spectrum of the ZP5 film for Zn exhibits a slight shift in the binding energies of Zn 2p 3/2 (1021.2 eV) and Zn 2p 1/2 (1044.2 eV). This indicates that the Zn 2+ atoms originating from ZnO are inclined toward PVDF owing to their affinity for CF 2 . The C 1s core spectrum of ZP5 exhibits two dominant peaks at 290 and 285.5 eV, corresponding to -CF 2and -CH 2 - (Figure 2(f)). The deconvoluted peaks at 283.9, 285.4, and 286.7 eV correspond to C-C, CH 2 , and C-O, respectively [55]. Figure 2(g) shows two peaks in the deconvoluted O 1s core level spectrum of the ZP5-based composite film. The peak at 531.4 eV corresponded to covalently bonded Zn-O-Zn, indicating the presence of ZnO microrods in the PVDF matrix [56]. Another peak at 532.2 eV was attributed to the externally adsorbed H-O-H [3,57]. The XPS spectra of F 1s ( Figure S1) for the ZP5- 10  Intensity (a.u.) 5 International Journal of Energy Research based composite exhibit a peak corresponding to -CF 2 -, located at 687.5 eV, and possessed a lower binding energy than that of the pure PVDF polymer. The shift indicates that the β phase orientation of the PVDF polymer improves owing to ZnO-PVDF interactions [58].
The SEM images of the ZnO microrods and ZP5-based composite films are shown in Figure 3(a). The length and width of the ZnO microrods synthesized using the wet chemical method at 80°C were 2 μm and approximately 500 nm, respectively. Figure 3(b) shows the TEM images of the highly crystalline ZnO microrods embedded in the PVDF matrix. The high-angle annular dark-field imaging output used for elemental analysis is shown in Figure 3   International Journal of Energy Research thereby confirming the presence of the ZnO microrods and PVDF in the ZP5-based composite film. In addition, the HRTEM image and SAED patterns were investigated to analyze the crystal phase and orientation of the ZnO microrods. Diffraction patterns with well-defined lattice fringes were observed based on the HRTEM image (Figure 3(h)). The distances between two lattice fringes, known as interplanar d-spacings, were 0.25 and 0.28 nm, as marked in the enlarged HRTEM images of Figure 3(h); these values correspond to the (101) and (100) lattice fringes of hexagonal wurtzite ZnO, respectively [59]. The interplanar d-spacing was calculated as the reciprocal of the distance between the central bright spot and the other diffraction spots in the SAED pattern (Figure 3(i)). Thus, the calculated d-spacing values of 0.25 and 0.28 nm were assigned to the (101) and (100) lattice planes, respectively [59]. These results agree well with the XRD patterns shown in Figure 2(a). The piezoelectric properties of ZP5, which possessed the highest β phase content, and a PVDF film were determined using PFM to evaluate the improvement in the piezoelectric characteristics of the composite film owing to the presence of ZnO. PFM can effectively probe local polarization and determine piezoelectric coefficients [60]. The PFM measurements of the PVDF-and ZP5-based composite films were conducted using a Pt-Cr-coated cantilever (Multi75E_G) as an AFM tip. A change in the electric potential between the sample and AFM tip occurred during the electromechanical response owing to the converse piezoelectric effect (Figure 4). A 10 μm × 10 μm area of the ZP5-based composite film was scanned at an external bias voltage of 10 V and 1 Hz. Figure 4 shows the surface topography and PFM out-of-plane amplitude for the pristine PVDF-and ZP5-based composite films. The pristine PVDF film (Figure 4(a)) exhibits a smooth surface with a roughness value (R a ) of 59 nm. The topographical image of ZP5 (Figure 4(b)) exhibits a considerable increase in roughness (261 nm) owing to the incorporation of ZnO microrods into the PVDF matrix. The amplitude image (Figure 4(c)) of the PVDF sample exhibits nearly uniform bright regions, considering an average amplitude of 0.872 mV. The mean amplitude of the ZP5 film (Figure 4(d)) is 1.356 mV, which is higher than that of the pristine PVDF film. This indicates that the piezoelectric surface potential increased for the ZP5 composite film with ZnO microrods. The piezoelectric coefficient (d 33 ) values of the bare PVDF and ZP5-based composite films can be measured based on the strain produced at an applied voltage. The piezoelectric coefficient (d 33 ) (pm V -1 ) was obtained based on the PFM analysis using Eq. (2) [60,61]  where D denotes the deflection of the cantilever tip (in meters) between points A and B; D is calculated using the PFM amplitude value. V AC denotes the AC tip bias voltage at 1 V. The calculation of the deflection D is expressed in Eq. S1 of Section S1. The d 33 values of the bare PVDF and ZP5-based composite films calculated based on the forward-amplitude images were 23.3 and 36.3 pm V -1 , respectively. The higher piezoelectric coefficient of the composite indicated higher polarization and enhanced piezoelectric potential [62].

Piezoelectric Performance of ZnO/PVDF-Based
Composite Materials. Figure 5(a) shows the open-circuit potential (V oc ) values of the ZP4-, ZP5-, and ZP6-based PENGs under a tapping force of 20 N at 5 Hz. The V oc values of the ZP4-, ZP5-, and ZP6-based PENGs were 11.2, 15.2, and 10.8 V, respectively. The higher V oc value of ZP5 indi-cates that the 0.5 g of ZnO included in the precursor was optimal. The force distribution impact varied with the content of ZnO microrods and electroactive β phase in PVDF, which is confirmed by the increase in the piezopotential of ZP5. In the case of ZP6-based PENG, the piezopotential reduced under the same force (20 N) owing to the decrease in the β phase in the PVDF matrix, as confirmed by the FTIR analysis. Consequently, the piezoelectric performance of ZP5 was investigated in detail. The V oc values of the ZP5-based PENG at a tapping force frequency of 5 Hz under varying force magnitudes of 5, 10, 15, and 20 N were 6.1, 11.0, 13.1, and 15.2 V, respectively ( Figure 5(b)). This observation indicates that as the mechanical force increases, the piezoelectric potential increases. The increase in V oc was closely related to the increase in the number of dipoles caused by the higher mechanical strain imparted on ZnO and PVDF in the compressed state. A higher compression  International Journal of Energy Research leads to a higher amount of polarized charge at the interfacial sites of ZnO and PVDF [63].
Similarly, the tapping frequency-dependent piezoelectric potential was evaluated under a constant tapping force of 20 N (Figure 5(c)). At 1, 3, 5, and 7 Hz, the output voltage increased owing to the rapid tapping of the ZP5-based PENG. The V oc value at 1 Hz was 3.2 V and gradually increased to 15.2 V at 5 Hz, subsequently decreasing to 13.8 V at 7 Hz. The maximum V oc values were obtained at 20 N and 5 Hz. This observation can be attributed to fast polarization and the remnant electrons in the composite film. However, rapid tapping may result in insufficient time for restoring the polarized dipoles to their original forms, thereby decreasing the voltage at 7 Hz [64,65]. Accordingly, a bare PVDF filmbased PENG was tested to confirm the enhanced piezoelectric behavior originating from the ZnO microrods. The V oc value of the bare PVDF film-based PENG was 4.3 V, which is approximately one-third of that of ZP5-based PENG ( Figure 5(d)). The V oc value of ZP5-based PENG measured over 900 tapping cycles ( Figure 5(e)) exhibited a stable, consistent value of~15 V, indicating that the device possesses good piezoelectric stability. The inset of Figure 5(e) shows the tapping machine and oscilloscope. Table 2 lists the comparative results of previous PENGs and the current supersonically sprayed PENG, thereby demonstrating the superior performance of the current PENG.
The piezoelectric behavior was evaluated based on bending tests at different bending radii. Figure 5(f) shows the bending test performed for the ZP5-based PENG with bending radii (R b ) of 4, 5, 6, and 7 mm at 1 Hz; the output

10
International Journal of Energy Research voltages were 320, 219, 128, and 90 mV, respectively. As the slider in the bending machine moved inward, the bending radius of the flexible ZP5-based PENG decreased, thereby generating piezoelectric potential owing to the relative displacement of atoms within the composite film; this led to a higher voltage because of the internal strain of the composite film. In the flat position, no piezoelectric voltage was generated. The piezoelectric voltage produced during bending was caused by elastic deformation due to the internal strain of the piezoelectric ZnO microrods [69]. The internal strain during bending was not comparable to the strain produced during tapping. This resulted in a lower V oc [70].
To explore the energy-harvesting and voltage generation performance of the ZP5-based PENG during kinetic motions, we considered mechanical motions and body vibrations for potential motion-powered energy-harvesting applications (Figures 6(a)-6(f)). The maximum V oc values recorded for finger tapping (Figure 6(a)), palm tapping ( Figure 6(b)), and foot tapping (Figure 6(c)) were 2.20, 4.40, and 11.0 V, respectively. These values agree with the previous tapping results, indicating that V oc is directly proportional to the applied mechanical force [71]. In addition, body stretching tests, e.g., finger, elbow, and knee bending, were performed at various bending angles. Specifically, the bending tests were performed at 0°, 45°, and 90°( Figures 6(d)-6(f)). At 0°, no V oc was observed, whereas continuous periodic finger bending at 45°and 90°resulted in maximum V oc values of 165 and 500 mV, respectively. Similarly, elbow bending at 45°and 90°produced 420 and 580 mV, respectively, whereas for knee bending at the same These results indicate that ZP5-based PENG can harvest small-scale mechanical energy from human body movements, thus demonstrating its potential for practical applications.
The output current (I) was calculated based on the measured maximum piezopotential across the load resistance for the ZP5-based PENG (Figure 6(g)). The maximum output voltage (V max ) was recorded by varying the external load resistance R in the electrical circuit ( Figure 6(h)). A periodic tapping test was performed under a tapping force of 20 N at 5 Hz to measure the electric signals. The graph in Figure 6(h) illustrates the V max and I max values obtained with external resistances in the range of 0.01-100 MΩ. Figure 6(i) shows that a maximum power of 112.5 μW was obtained at 0.08 MΩ, and the power density of the ZP5-based PENG with an area of 3 cm × 3 cm was 12.5 μW cm -2 [72]. Moreover, the performance of the ZP5-based PENG was further investigated by lighting commercial LEDs based on the continuous tapping of the device at 20 N and 5 Hz (Figure 1(c)). These findings reveal the possibility of using the developed ZP-based PENG as a touch sensor and in LED lighting. The long-term cycling test under tapping and bending suggests the excellent durability of the PENG. The energy generated by the physical body activities was harnessed by attaching the PENG to the finger, forearm, and knee, confirming the flexibility of the ZP film, which yielded sufficient output voltage to power microelectronic devices.
The electrochemical properties of ZnO/rGO (Section 2.5) were explored by CV and GCD using a two-electrode symmetric coin cell. The CV measurements were performed at scan rates of 10, 20, 50, and 100 mV s -1 with a potential window of 0-1.0 V for the symmetric supercapacitor. Figure 7(a) displays the current as a function of the applied voltage at different scan rates, where nearrectangular CV curves are observed. These quasirectangular curves indicated the pseudocapacitive characteristics of the ZnO microrods coupled with rGO sheets. Figure 7(b) shows the anodic and cathodic curves as square root functions of the scan rate. The positive and negative curves denote anodic and cathodic peak currents, respectively. The linearity of the peak current in the graph indicated the occurrence of reversible faradaic reactions [16]. The GCD technique was used to calculate specific capacitance associated with ZnO covered with rGO. The potential response was recorded at a constant current density (Figure 7(c)). The applied constant current density ranged from 0.5 to 5 mA cm -2 . The GCD results revealed a discharge time of 16.4 s at 0.5 mAcm -2 . The plots of the charge and discharge cycles with variations in current density exhibited a nearly triangular shape, indicating that the electrode possesses good capacitive properties. Figure 7(d) shows the capacitance, energy density, and power density with respect to the specific current obtained from the GCD measurements. The maximum areal-specific capacitance was 33 mF cm -2 . Furthermore, the corresponding maximum energy and power densities were 5.38 μWh cm -2 and 1.16 mW cm -2 , respectively. The calculations of capacitance, energy density, and power density are expressed in Eqs. (S2), (S3), and (S4), respectively (Section S2). The relatively poor electrochemical performance of ZnO/rGO implies that further optimization of the fabricated electrode is necessary.

Conclusion
In this study, a flexible, durable, self-powered PENG electrode was fabricated using a supersonic spraying deposition technique. Inorganic piezoelectric ZnO microrods and PVDF polymer were combined to enhance their piezoelectricity and power density. The concentrations of the ZnO microrods and PVDF polymer were optimized to obtain the ZP composite, thereby exhibiting the maximum open circuit potential of 15.2 V and power density of 12.5 μW cm -2 . The flexibility and mechanical durability of the PENG were demonstrated via bending and tapping tests.

Data Availability
The numerical data used to support the findings of this study are available from the corresponding authors on reasonable request.