P(VDF-TrFE) powder (PolyK Technologies, USA) with a molar ratio of 75/25 was mixed with N,N-dimethylformamide (DMF, Tianjin Fuyu Fine Chemicals Co., Ltd., China) solvent. A highly crystalline P(VDF-TrFE) film with a thickness of 0.048 mm was obtained via conventional solution casting at 200 °C followed by annealing at 140 °C. Subsequently, the P(VDF-TrFE) film underwent polarization treatment using a polarization device (RK2674A, made in China) and an oil bath piezoelectric polarization device (HYJH-L/20KV, made in China). All P(VDF-TrFE) piezoelectric films in this study were treated using the same polarization process to eliminate the effect of different polarization conditions on piezoelectric performance. Both sides of the film were metalized using silver nano-ink screen printing, and the final assembly formed a piezoelectric sensor, as shown in Figure 1A.

The new piezoelectric-electromagnetic hybrid energy harvester (PEEH) has overall dimensions of 60 × 90 × 40 mm³ and is encapsulated on both sides with transparent acrylic. It primarily consists of a dual piezoelectric cantilever (PEH), a dual electromagnetic module (EMHs), and a rotating assembly (Supplementary Figure 1 for assembly details). The PEH unit (29 × 20 × 0.1 mm³) consists of a P(VDF-TrFE) piezoelectric film bonded to a flexible substrate using epoxy resin (Purchased from Zhejiang Lixie Instrument & Equipment Co., Ltd., China), with a cylindrical magnet (Φ8 × 3 mm) attached to the top of its free end. The EMH module consists of a copper coil (inner and outer diameters of 8 and 36 mm, respectively, and a height of 7 mm) mounted on an acrylic plate. Additionally, the rotating assembly includes an external bell housing, a top sealing plate, and a suspension chassis embedded with eight cylindrical permanent magnets and connected via bearings.

1H Nuclear Magnetic Resonance spectroscopy (NMR) spectra were recorded on a Bruker Avance II 400 MHz NMR spectrometer (Bruker, Germany) using C₃D₆O (99.9%, Aladdin) as the solvent. Crystal structure characterization was performed on a Rigaku D/MAX-2400 X-ray diffractometer (XRD, Cu Kα, λ = 0.15406 nm, 40 kV) in Japan. Fourier transform infrared (FTIR) spectra were recorded on a Shimadzu 8400S FTIR spectrometer (400-2,000 cm^-1, 4 cm^-1 resolution). Thermal properties were tested on a Netzsch differential scanning calorimetry (DSC) 214 Polyma differential scanning calorimeter (N₂ atmosphere, 10 °C min^-1 heating/cooling rate). Surface morphology was observed and photographed using a Nikon LV100POL polarizing microscope (POM, crossed polarization mode, ×200 magnification). Polarization treatment was performed in a Chinese Huayuan HYJH-L/20 kV oil-bath high-voltage polarizer. Dielectric temperature and frequency spectra were measured using an American Agilent 4284A precision impedance analyzer (100 Hz-10 MHz, AC amplitude 100 mV) coupled with a Linkam THMS600 temperature controller. The ferroelectric loop recordings were obtained using the TF Analyzer 2000 ferroelectric testing system (frequency: 10 Hz) from the German company aixACCT. The piezoelectric constant d₃₃ was measured within 24 h after polarization using a ZJ-3 quasi-static d₃₃ tester (0.25 N, 110 Hz) from the Institute of Acoustics, Chinese Academy of Sciences.

In this study, ANSYS and Ansoft Maxwell software were used to perform simulation analyses on the piezoelectric cantilever beam and the electromagnetic module, respectively. A finite element simulation analysis of a piezoelectric cantilever beam was performed using ANSYS software. The geometric model was constructed using DesignModeler, a modeling tool within ANSYS. ANSYS software's Adaptive Mesh Refinement was used. During the calculation, the software automatically refines the mesh in key areas based on the magnetic field gradient and stops iterating once preset convergence criteria are met to ensure stable results. The fixed end was set as a fully constrained fixed support, while the forces between the magnets at the free end were applied through fluid-structure interaction or discrete dynamics. The fixed end was configured as a fully constrained fixed support, while the forces between the magnets at the free end were applied using fluid-structure interaction or discrete dynamics. The electromagnetic module was simulated using Ansoft Maxwell software, employing the 3D Transient Solver to accurately simulate the time-dependent changes in magnetic flux during rotor rotation. The boundary of the external air domain was set to “Radiation” to simulate an infinite space, while the coil was configured as “Winding excitation” with the actual number of turns and internal resistance defined. A rotating motion band was established around the rotor assembly, with the rotational speed strictly matched to the rotational speed under the wind speed conditions of the experimental test.

To characterize the output characteristics of the energy harvester, a dedicated vibration and wind tunnel testing platform was constructed [Supplementary Figure 2]. The vibration platform consists of a controller (Yiheng), a power amplifier (SA-PA080, Wuxi Shiao Technology Co., Ltd., China), a shaker, and an accelerometer (EV4200) to provide and monitor the reference vibration excitation; The wind tunnel platform combines a low-noise axial flow wind tunnel (for low wind speeds) with an axial flow fan (for high wind speeds), and is equipped with an anemometer (GM8910, Shenzhen BENETECH Biaozhi Instruments, China) and tachometer (Xima AR925 Contact-Type Tachometer, Xima Instruments SMART SENSOR, China) to monitor the flow field and prototype rotational speed in real time. Additionally, the electrical testing platform [Supplementary Figure 3] uses a digital oscilloscope (RIGOL DS1054Z, Shenzhen Huaqing Instrument Co., Ltd., China) and an electrometer (Keithley-6482, China) to measure and record the device’s output voltage and short-circuit current, respectively.

The chemical structure of the synthesized P(VDF-TrFE) copolymer was first corroborated using ¹H NMR spectroscopy [Figure 1B]. The multiplet observed at 2.2-3.3 ppm is assigned to the protons in the -CH₂-CF₂- groups of the VDF units, while the signal spanning 5.1-6.0 ppm corresponds to the -CHF-CF₂- groups of the TrFE units, accurately reflecting the expected monomeric configuration. Crystallographic analysis via XRD [Supplementary Figure 4] reveals a single, sharp (200/110) diffraction peak at 2θ = 20.9°. The complete absence of α-phase characteristic peaks indicates that the film crystallizes almost exclusively into the electroactive β phase, establishing a robust structural foundation for its high-voltage electrical response. This high β phase purity is further substantiated by FTIR spectroscopy [Figure 1C]. Characteristic vibrational bands of the β phase are prominent at 490, 840, 1,280, and 1,400 cm^-1, whereas the 880 cm^-1 band, which is indicative of the nonpolar α phase, is nearly absent^[35]. This result confirms the overwhelming dominance of the polar β conformation.

Thermal and morphological evaluations further validate the film's quality. DSC [Supplementary Figure 5] demonstrates a Curie temperature (T_c) of 98 °C and a melting point (T_m) of 156 °C. These values align with literature reports for highly crystalline P(VDF-TrFE), ensuring adequate thermal stability across the anticipated device operating temperature range. Macroscopically, this high crystalline quality is reflected in the polarized optical micrograph [Figure 1D and Supplementary Figure 6], which exhibits a typical lamellar grain morphology characterized by radially symmetric light-dark extinction bands, a hallmark of ordered lamellar crystal stacking.

The electrical properties, which are pivotal for electromechanical coupling, were systematically characterized. At room temperature and 100 Hz, the film exhibits a relative dielectric constant (ε_r) of approximately 14 and a low dielectric loss (tanδ < 0.04) [Figure 1E]. It is noteworthy that the pure P(VDF-TrFE) film exhibits a relatively high dielectric loss (~ 0.2) at 10⁷ Hz, which is inherently governed by the unavoidable intrinsic dipole relaxation of the polar polymer under ultra-high frequency electric fields, consistent with the standard physical characteristics of P(VDF-TrFE) films^[36,37]. As the frequency increases, ε_r monotonically decreases while tanδ slightly increases, a behavior entirely consistent with the classical dipole orientation polarization model. Moreover, the temperature-dependent dielectric spectrum [Figure 1F] displays a pronounced dielectric peak at T_c, capturing the classic ferroelectric-to-paraelectric phase transition. The intrinsic ferroelectricity of the film is explicitly verified by the polarization-electric field (P-E) hysteresis loop [Figure 1G]. Under an applied electric field of 180 MV/m, the remnant polarization (P_r) reaches 8.1 μC cm^-2 with a coercive field (E_c) of 67 MV/m, underscoring the film's robust dielectric strength and excellent polarization switching capability. Finally, the longitudinal piezoelectric coefficient (d₃₃) exhibits an initial decrease before stabilizing as a function of increasing film thickness [Figure 1H]. Notably, at an optimized thickness of 48 μm, the d₃₃ value peaks at -31 pC/N, representing the upper echelon of performance for flexible polymer piezoelectric. This superior electromechanical coupling directly empowers the high-sensitivity, self-powered sensing capabilities essential for subsequent microelectromechanical systems (MEMS) applications, particularly in wind speed monitoring and hybrid energy harvesting.

Building upon the aforementioned material advantages, this study proposes a novel PEEH specifically engineered for the concurrent extraction of ambient wind energy and the autonomous monitoring of wind speed. Figure 2A delineates the architectural flowchart of the PEEH-enabled wireless sensor node, which integrates four fundamental modules: the electromechanical structure, power management, signal processing, and wireless transmission. Operationally, ambient wind kinetic energy is simultaneously scavenged by the electromagnetic harvester (EMH) and the piezoelectric harvester (PEH). Crucially, the system intelligently bifurcates these harvested outputs based on their optimal functions. The EMH, comprising dual stationary coils, serves as the primary power generator. The induced alternating current (AC) is routed through a comprehensive power management circuit (PMC), which includes a rectifier, an undervoltage lockout (UVLO) mechanism, storage capacitors, and a voltage regulator to provide a stable, regulated direct current (DC) to the microcontroller unit (MCU). Conversely, the two well-designed P(VDF-TrFE) PEHs function as self-powered active sensors, directly correlating wind speed with their oscillatory output frequency. By integrating these power supply and sensing pathways, the MCU accurately processes the frequency signals and transmits the wind data via a radio frequency (RF) module. This highly integrated paradigm eliminates the reliance on wired power grids or frequent battery replacements, offering a robust, maintenance-free monitoring solution ideal for remote deployments, such as meteorological stations, bridges, and communication towers.

The structural configuration that actualizes these dual functions is detailed in Figure 2B. The EMH module features coils symmetrically anchored to a stationary base, concentrically aligned with the trajectory of the rotor magnets to maximize magnetic flux linkage. To enhance the system's responsiveness and reliability under low-wind conditions, the rotor disk incorporates an opposing magnetic suspension mechanism; these symmetrically positioned reverse magnets effectively minimize mechanical friction and starting torque. As the wind-driven rotor assembly spins, the cylindrical magnets attached to its periphery sweep across the stationary coils, inducing an electromotive force for efficient bulk energy harvesting. Concurrently, this rotational motion actuates the piezoelectric sensing module. As illustrated in Figure 2C, periodic alternating magnetic attraction and repulsion forces occur between the suspended magnets on the rotating disk and the permanent magnets affixed to the free ends of the PEH cantilevers. This non-contact magnetic plucking mechanism drives the PEH into sustained mechanical oscillation. Consequently, variations in ambient wind speed are precisely translated into proportional rotor speeds, which in turn dictate the excitation frequency and voltage response of the PEH. This elegantly coupled mechanical design ensures that dynamic wind variations are faithfully captured as electrical signals, thereby establishing a highly reliable foundation for self-powered sensing.

Finite element method (FEM) simulations were utilized to optimize the structural design, evaluate mechanical stability, and maximize the overall energy conversion efficiency of the PEEH. FEM simulations were mainly used to guide rotor structure design and to compare the relative differences in magnetic field distribution and magnetic excitation characteristics under different magnet configurations, based on which the optimal structure for subsequent experiments was selected. Initially, modal analysis was performed on the PEH cantilever. As illustrated in Supplementary Figure 7, the simulations revealed three distinct vibration modes: the fundamental bending mode, a symmetrically balanced mode, and a higher-order node-splitting mode. Notably, the fundamental (first-order) mode exhibits a relatively low resonant frequency of 56.75 Hz. This low-frequency characteristic ensures that the cantilever is highly responsive and can be readily excited by the low-frequency, small-amplitude mechanical stimuli generated by the wind-driven rotor, providing a robust mechanical foundation for subsequent energy harvesting.

Beyond structural dynamics, the configuration of the rotor significantly dictates the overall electromechanical performance of the system^[38]. Consequently, we systematically investigated the influence of the rotor's permanent magnet array-specifically the magnet count and spatial arrangement-on both electromagnetic induction and piezoelectric excitation. As depicted in Figure 3A, three configurations were modeled: six uniformly distributed magnets with identical polarities (Model A), six magnets with alternating polarities (Model B), and eight magnets with alternating polarities (Model C). The vector spatial distributions of the magnetic flux density for each configuration were examined [Figure 3B and Supplementary Figure 8]. FEM simulation results show that among the three studied rotor magnet configurations, Model C generates the highest magnetic flux density. This enhanced magnetic field helps improve electromagnetic induction, so Model C was chosen as the rotor structure for subsequent device fabrication and electrical performance testing.

To better understand how different rotor structures affect the piezoelectric excitation methods, we conducted a FEM analysis of the instantaneous magnetic forces on the free-end magnet of the PEH along the Y and Z axes at 350 rpm, to compare the magnetic excitation characteristics produced by different magnet configurations [Figure 3C]. Under Models A and B, the PEH tip experiences a continuously positive (unidirectional) magnetic force in the Z direction, imposing a static upward deflection on the cantilever that results in restricted oscillation and an irregular, asymmetric voltage output. In stark contrast, Model C subjects the PEH tip to strictly alternating positive and negative magnetic forces (a bidirectional “push-pull” effect). FEM shows that Model C generates periodic alternating magnetic excitation characteristics, which theoretically makes it better at driving the piezoelectric cantilever to form stable back-and-forth vibrations, providing a reasonable explanation for getting more stable piezoelectric output. Furthermore, piezoelectric potential simulations [Figure 3D] indicate that the generated voltage reaches its maximum near the fixed clamped end of the cantilever, which is entirely consistent with the classical mechanical stress distribution of a cantilever beam. Compared to the other configurations, Model C induces the largest deformation amplitude and, consequently, the highest voltage output. In summary, the sophisticated rotor architecture of Model C comprehensively optimizes energy harvesting efficiency while simultaneously enhancing signal fidelity. Therefore, this study chose Model C, which had the best performance in this work, for further device fabrication and electrical performance testing.

Guided by the preceding simulation results, the rotor configuration of Model C, which ensures optimal energy-harvesting efficiency and highly uniform piezoelectric excitation, was selected for fabrication and experimental validation. Initially, to assess the system's dynamic characteristics and its resilience to external environmental vibrations, frequency sweep experiments were conducted to determine the actual resonant frequency of the PEH cantilever. As illustrated in Supplementary Figure 9, frequency sweeps (45-65 Hz) under varying base accelerations experimentally confirm a resonant peak at ~56 Hz. This result aligns excellently with our theoretical modal analysis. To further isolate the vibration characteristics and evaluate the system's robustness against external mechanical crosstalk, the permanent magnet at the free end of the PEH was substituted with a non-magnetic copper disc of equivalent mass [Figure 4A]. A continuous frequency sweep (10-70 Hz at an acceleration of 9.8 m/s²) was applied to the base while maintaining a constant wind speed of 3.5 m/s. The results demonstrate that the overall voltage response is a direct superposition of the wind-induced excitation and the external base vibration. Crucially, once the external frequency sweep was terminated, the PEH immediately returned to its stable baseline output driven solely by the wind. These findings conclusively demonstrate that external environmental vibrations do not disrupt the fundamental electromechanical coupling of the PEEH, thereby guaranteeing robust operational stability in complex, real-world deployment scenarios.

Subsequently, the electromechanical performance of the fully assembled PEEH system was systematically evaluated across a broad spectrum of wind conditions [Supplementary Figure 10]. Figure 4B confirms its core sensing capability. The figure presents the frequency–wind speed linear response analysis of the PEH within the range of 2-13 m/s, demonstrating a strong linear correlation between the output oscillation frequency of the PEH and the ambient wind speed. This strict linearity confirms the PEH's exceptional sensitivity and reliability, establishing it as an ideal active wind speed sensor. Detailed analyses of the output voltage further elucidate the dynamic behaviors of both modules. As depicted in Figure 4C and Supplementary Figure 11, as the rotor-induced excitation frequency approaches the mechanical resonance frequency of the cantilever (~56 Hz), the PEH voltage initially increases with wind speed, reaching a total output of 112 V at 12 m/s. However, once the wind speed drives the excitation frequency beyond the resonance threshold, the output voltage generated by the cantilever’s vibration amplitude will naturally decay. In contrast, governed by Faraday’s law of electromagnetic induction, the output voltage of the EMH exhibits a continuous linear increase with wind speed [Supplementary Figure 12]. Short-circuit current measurements conducted at a reference wind speed of 11 m/s [Figure 4D and Supplementary Figure 13] confirmed that both modules produced stable and continuous current output. To maximize energy extraction, impedance matching tests were performed at a wind speed of 12 m/s [Figure 4E and Supplementary Figure 14], yielding optimal load values of 2 MΩ for the PEH and 200 Ω for the EMH, respectively. Notably, the effective operational range of the proposed hybrid energy harvester extends to 3-17 m/s, within which stable electrical outputs can be generated. When connected to a 200 Ω load, Figure 4F shows the EMH's best output power at different wind speeds in the 3-17 m/s range. At 17 m/s, the EMH's total output power reaches as high as 38.5 mW. Finally, the superior power generation capability of the proposed PEEH is further contextualized in Figure 4G, where it demonstrates a significant competitive advantage over other recently reported piezoelectric and hybrid energy harvesters^[30,39-47]. In summary, the synergistic integration of the PEH and EMH affords highly accurate wind sensing alongside continuous, high-efficiency energy harvesting, flawlessly fulfilling the power requirements of autonomous wireless sensor nodes.

To efficiently convert the harvested AC energy into a continuous and stable DC power supply for the wireless sensor node, a dedicated power management circuit (PMC) was engineered. Initial evaluations of the electromagnetic modules (EMH1 and EMH2) at a wind speed of 4.5 m/s [Figure 5A and B] reveal that connecting the EMH outputs in series prior to rectification yields the optimal electrical output. Taking into account the forward voltage drop of approximately 0.3 V across the rectifier diode, the 4.2 V input voltage yields a regulated DC output of approximately 3.9 V after rectification, corresponding to a rectification efficiency of approximately 92.86% (Figure 5C for the schematic diagram). At the heart of the PMC is the BQ25570 power management integrated circuit, which integrates an MPPT algorithm and a buck-boost energy management module to efficiently charge the supercapacitor. According to the BQ25570 data sheet, the converter achieves a conversion efficiency of approximately 80% under typical operating conditions. Therefore, the overall energy conversion efficiency of the PMC can be estimated based on the rectification efficiency and the BQ25570 conversion efficiency, which is approximately 74.3%. Then, various energy storage components were used to evaluate the charging dynamics under different wind speeds [Figure 5D]. At a wind speed of 4.5 m/s, capacitors with capacities of 1,000 and 2,200 μF both reached about 3 V within 100 s. And as the wind speed increased to 7 m/s, the charging voltage for both capacitors went up to around 6.0-6.5 V. Notably, at an elevated wind speed of 11 m/s, the system demonstrated exceptional charging capability, rapidly charging a substantial 0.1 F supercapacitor to 4 V in merely 275 s [Figure 5E]. The experimental test bench and the comprehensive PMC topology are provided in Figure 5F and Supplementary Figure 15, respectively.

To ensure the uninterrupted operation of the sensing node, the system employs strict hysteresis voltage threshold management utilizing the IC's UVLO feature. Regulated power delivery is triggered only when the supercapacitor voltage reaches an upper threshold of 4.0 V, and is suspended to initiate recharging when the voltage drops to a lower limit of 3.0 V. This autonomous management is experimentally validated by the continuous voltage waveform recorded at 11 m/s [Figure 5G]. Following an initial 275 s charging phase to 4.0 V, the PMC delivers a strictly regulated 3.6 V to the load. Once continuous discharge depletes the capacitor to 3.0 V, the UVLO mechanism seamlessly isolates the load, allowing the supercapacitor to recharge. A full charging and energy recovery cycle takes about 11.6 min, with the system providing continuous power for around 7.7 min. In conclusion, the intelligently managed PEEH system demonstrates exceptionally reliable operation across a broad spectrum of wind conditions. By autonomously managing the charging and discharging cycles, it provides a highly practical, maintenance-free power foundation tailored for self-powered wireless sensor nodes deployed in remote engineering infrastructures, such as bridges and communication towers.

To enable autonomous operation of wireless sensor nodes, we introduce a parallel resistor to adjust and stabilize the output voltage by connecting a 20 kΩ resistor [Figure 6A]. This configuration clamps the output voltage within the safe operating range of the MCU. The system integrates a CC2340 bluetooth low energy (BLE) module to minimize overall energy consumption. Initial power profiling revealed that, under a 3.6 V supply in a continuous active mode, the MCU and BLE module draw an average current of 8.14 mA, equating to a power consumption of 29.30 mW [Figure 6B]. Recognizing that this continuous draw exceeds the steady-state generating capacity of the PEEH, we implemented an event-driven duty-cycling power management strategy. Specifically, the sensor node operates on a 24 s periodic cycle, wherein the energy-intensive BLE module is briefly awakened solely during the active phase for data acquisition and wireless transmission [Figure 6C]. This optimized time-division approach drastically reduces the system's average current to 3.25 mA (lowering power consumption to 11.70 mW), successfully aligning the load demand with the PEEH's energy harvesting capacity and enabling sustained self-powered operation.

Having secured the system's power budget, the operational endurance and reliability of the complete PEEH were systematically evaluated. To simulate harsh continuous outdoor environments, the system was subjected to a high-speed (17 m/s) endurance test for 30 min [Figure 6D]. Throughout this rigorous monitoring period, the PEEH maintained a highly stable voltage output with no observable mechanical degradation or electrical performance fluctuations, substantiating its robust durability for long-term engineering deployments. Furthermore, the system's start capability under suboptimal conditions was verified. As depicted in Figure 6E, even under a gentle breeze of 4.5 m/s, the power management circuit successfully accumulated sufficient energy over 1,700 s to reach the 4.0 V threshold, thereby triggering the regulated VOUT to awaken the sensor node. This confirms the circuit's exceptional practicality and resilience in low-wind-speed scenarios.

Finally, the core function of PEEH, namely accurate wind speed monitoring, was evaluated. By extracting wind speed data from the vibration frequency of the P(VDF-TrFE) cantilever, the system exhibits inherent immunity to the voltage amplitude attenuation that plagues traditional amplitude-based measurement methods. Field validations [Figure 6F and Supplementary Figure 16] demonstrated that at a reference wind speed of 6.5 m/s, the oscilloscope-measured baseline vibration frequency was 33.22 Hz. The system’s wireless transmission and detected frequencies were 33.25, 33.57, and 33.37 Hz, respectively, with a maximum frequency error of only 1.05%. This error was primarily due to the uncertainty of the MCU clock reference and hardware noise in the signal conditioning circuit. Accordingly, the maximum relative error of the calculated wind speeds (6.47, 6.54, and 6.50 m/s) was extremely low, at only 0.6%. This is attributable to the rigorous calibration of high-precision commercial anemometers, the system algorithm’s effective filtering of random hardware noise through multi-point averaging, and the mathematical convergence of the calibration formula within the current wind speed range, thereby achieving wind speed measurement accuracy superior to that of frequency-based measurements. Subsequent synchronous benchmarking against a commercial standard anemometer (GM8910) confirmed that the device maintains an overall measurement error of ≤ 1.05% across a broad wind speed range of 3-17 m/s, fully satisfying the stringent accuracy requirements for engineering measurements. These compelling results conclusively demonstrate that the PEEH successfully unites stable, reliable energy harvesting with high-precision sensing, offering a highly viable hardware foundation for self-powered wireless IoT networks. This work verifies the device’s performance under laboratory conditions, and future work will further test it in outdoor environments and evaluate continuous operational stability to promote its practical engineering applications.