Introduction

With the rapid development of Internet of Things (IoTs), the vast of wireless sensor network nodes present great challenges in distributing, scheduling, and managing power sources1,2,3,4. Traditional wiring and battery methods are inadequate for high-density network nodes, leading to increased maintenance costs and environmental threats5,6,7,8. Triboelectric nanogenerators (TENGs) offer a promising solution, by efficiently converting the surrounding mechanical energy into electricity9,10,11. Their versatility, compactness, and affordability make them increasingly significant in power and sensor systems12,13,14. In principle, enhancing the overall energy output and energy utilization efficiency of TENGs are two crucial aspects in practical applications15,16,17. To date, researchers have developed various effective strategies, including charge excitation18,19,20, local discharge21,22,23, space charge accumulation24,25,26, charge migration15,27, etc., to boost the output performance of TENGs. The average power density of TENGs has been raised to 87.26 W m−2 Hz−1, which is sufficient to meet the power requirements of most electronic devices in IoTs28. However, the high intrinsic impedance of TENGs (from megaohms to gigohms) greatly mismatches with that of the back-end nodes (typically in the kiloohm range), resulting in extremely low energy utilization efficiency29,30,31. For addressing this issue, researchers have employed various energy management (EM) strategies to modulate the power form of TENGs, including transformers29,32, switch capacitors33,34, Buck converters35,36,37, and other energy extraction mechanisms38,39,40,41, achieving a significant improvement in energy utilization efficiency. Nevertheless, in the current EM systems for TENGs, there are still two critical issues that have not been adequately addressed.

First, most existing researches employ contact separation mode TENGs (basic unit in TENGs) to assess the conversion efficiency of proposed EM strategies, claiming efficiencies exceeding 80%, or even 90%16,17. However, when utilizing these EM circuits directly for sliding (rotating) mode TENGs, the empirically proven conversion efficiency is measured <10% (Considering the significant disparities in the existing methods for calculating energy conversion efficiency, we adopt the final practical energy utilization rate to evaluate this parameter here, aiming for better comparability, as detailed in Supplementary Note 1 and Supplementary Table 1). The substantial difference observed in the same EM circuit applied to different TENG modes underscore the need for a more in-depth exploration of the intricate energy conversion mechanisms between TENGs and EM circuits. Second, the prevalent EM strategies mainly focuses on addressing the challenge of high output impedance in TENGs, subsequently providing real-time energy to electronic devices. However, the real-time nature of this power supply form renders it impractical for TENGs reliant on harvesting irregular mechanical energy from the environment to stably power electronic devices, which presents a significant impediment to the broader practical application of TENGs. Therefore, there exists an urgent demand to introduce systematic output regulation measures to engineer a TENG-based power supply equipped with energy storage and output regulation functionalities.

Herein, on one hand, a detailed energy transfer mechanism and maximum energy extraction strategy are proposed. We systematically introduce the two pivotal factors, namely voltage loss and charge loss, induced by switches (crucial components in EM circuits) during the energy transfer process. Subsequently, by synergistically optimizing the TENG and switch configurations, the maximum energy extraction effect is attained. As a result, energy conversion efficiency of 42.5% is observed in the rotating TENG, which is 8.5 times higher than that of directly using the current EM circuit. On the other hand, by rationally combining EM circuit with a commercial energy processing chip, a universal power supply strategy with energy storage and output regulation functionalities for all modes of TENGs is formed. As the demonstrations, a rotating TENG that only works for 21 s can make a hygrothermograph work stably for 417 s. Even under irregular hand driving, various types of TENGs can consistently provide stable power to electronic devices such as calculators and mini-game consoles. This work offers a comprehensive investigation of the energy transfer and conversion mechanism between TENGs and EM circuits, and presents a straightforward and effective energy storage and output regulation strategy for all-mode TENGs, which may provide valuable guidance and technical reference for engineering TENG based self-powered system.

Results

Key factors affecting energy conversion efficiency

To directly illustrate the optimization impact of the EM strategy proposed in this work on the energy utilization efficiency of TENGs, we select a conventional rotating free-standing mode TENG with 6 pairs of electrodes (RF-TENG-6) as the energy source, as depicted in Fig. 1a, and its intrinsic outputs and physical images are shown in Supplementary Fig. 1. The detailed output mechanism of RF-TENG is presented in Supplementary Fig. 2 and Supplementary Note 2. Figure 1b and Supplementary Fig. 3 illustrate the matching impedance characteristics of RF-TENG-6, with its peak root mean square (RMS) output power of 14.84 mW under the load resistance of 500 MΩ. Considering the equivalent impedance of common electronic devices is around ~KΩ, substantial energy wastage produces when directly powering electronic devices.

Fig. 1: Key factors affecting energy conversion efficiency.
figure 1

a 3D structure of RF-TENG-6. b RMS current, voltage, and power under different resistances. c Comparison of charging effects. Insets (i) and (ii) depict the circuit diagram and voltage curve of RF-TENG-6 directly charging a capacitor. Insets (iii) and (iv) depict the circuit diagram and voltage curve of RF-TENG-6 charging a capacitor through an EM circuit. Inset (v) depicts the calculation of energy conversion efficiency values for Insets (ii) and (iv). EM stands for Energy Management. d Potential simulation, (e) schematic and (f) output voltage of a sliding mode TENG unit. \({{{{\rm{\varphi }}}}}_{1}\) and \({{{{\rm{\varphi }}}}}_{2}\) represent the potentials of the two output electrodes. g Circuit diagram for measuring discharge current. Discharge current at gap distances of (h) 0 mm, (i) 0.1 mm, and (j) 1 mm respectively. k The trade-off of maximizing energy transfer. Source data are provided as a Source data file.

For demonstration, a 100 μF capacitor is chosen as the load. Figure 1c (i) and (ii) showcase the direct charging performance by RF-TENG-6. The voltage across the capacitor reaches 15 V after 212.5 s, indicating an average charging power of 52.9 μW, as calculated using the formula:

$${P}_{{out}}=\frac{{E}_{C}}{t}=\frac{\frac{1}{2}C{V}^{2}}{t}$$
(1)

where \({E}_{C}\), \(C\), \(V\), and \(t\) represent the stored energy, capacitance value, charging voltage, and charging time, respectively. Meanwhile, Fig. 1c(iii) and (iv) display the charging performance when employing the widely used EM circuit, with its detailed workflow presented in Supplementary Fig. 4 and Supplementary Note 3. Supplementary Movie 1 further demonstrates the discharge status of the switch, and Supplementary Fig. S5 and Supplementary Note 4 provide a specific description of the switch selection here. The results reveal that the average output power has increased by 14 times, reaching 745.03 μW. However, in comparison to the maximum RMS power in Fig. 1b, its energy utilization efficiency is only 5.02% (Fig. 1c(v)), approximating the actual conversion efficiency documented in previous works (Supplementary Table 1). Practically, the switch, a crucial component of EM, plays a decisive role during the energy conversion process. Therefore, a more detailed investigation into the energy conversion mechanism between the switch and RF-TENG-6 is conducted.

Figure 1d–f presents the numerical calculation results of the output voltage of the basic unit of RF-TENG, simulated using COMSOL, with detailed parameters outlined in Supplementary Fig. 6. The outcomes reveal that the voltage between two electrodes increases with the slider moving rightward. Consequently, discharge occurs when the output voltage reaches the breakdown threshold of the switch. Worth noting that, the threshold voltage can be regulated by adjusting the gap distance between two needle tips, as illustrated in Supplementary Fig. 7. In this case, the discharge behavior of the switch during a single motion cycle can be adjusted accordingly. Figure 1g–j depicts the discharge current at different switch gap distances, and the corresponding voltage curves is illustrated in Supplementary Fig. 8. As the results, the discharge intensity is higher while frequency is lower at 1 mm gap distance than these at 0.1 mm gap distance, respectively, which reflects a trade-off on energy conversion efficiency (Fig. 1k). Therefore, the switch state significantly influences the energy transmission effect, and its configuration optimization is pivotal for attaining high energy conversion efficiency.

Achieving maximum transferred energy through reducing voltage and charge loss

According to the theoretical energy conversion process16,42,43, when the output voltage of RF-TENG reaches its peak value (as the slider completes moving over a basic unit), the switch should synchronously be closed, as illustrated in Supplementary Fig. 9a. This condition is essential for achieving maximum energy release from the RF-TENG. In theory, the maximum output energy can be expressed as (Supplementary Fig. 9b):

$${E}_{max}=\frac{1}{2}{V}_{{OC},\, {max}} \cdot {Q}_{{SC},\, {max}}$$
(2)

Where \({V}_{{OC},\, {max} }\) is the maximum open-circuit voltage of RF-TENG, and \({Q}_{{SC},\, {max} }\) is the maximum transferred charge under short-circuit condition. To align with this energy maximization scheme, we can adjust the needle-gap distance so that its breakdown voltage matches \({V}_{{OC},\, {max} }\). In this case, the RF-TENG discharges once per motion cycle, maximizing the output energy (Supplementary Fig. 9c). If the gap is reduced to lower the breakdown voltage to one-third of \({V}_{{OC},\, {max} }\), the RF-TENG would discharge three times per motion cycle (Supplementary Fig. 9d). In this scenario, the total output energy of these three discharges is only one-third of the maximum output energy. This significant energy loss results from voltage-related losses (Fig. 2a).

Fig. 2: Achieving maximum transferred energy through reducing voltage and charge loss.
figure 2

a Schematic and theoretical V-Q curves of the basic RF-TENG with discharge switch during one sliding motion. b Transferred charge at different gap distances. c Charging curves under different inductances (gap distance: 0.9 mm). d Schematic and theoretical V-Q curve of RF-TENG after subdividing electrodes. e Actual V-Q curves under different gap distances (RF-TENG-12). f Matching inductance under 0.6 mm gap distance (Summary of charging time in Supplementary Fig. S18). g Reducing charge loss by optimizing discharge atmosphere. h Transferred charge under different air pressure (RF-TENG-12). i Charging time under different inductors. j Synergic optimization strategy for maximizing the energy conversion efficiency of RF-TENG. Inset (i) The RMS power as the resistance changes (RF-TENG-18). Inset (ii) The charging curve at optimized configurations. Source data are provided as a Source data file.

In practice, although the discharge voltage of the needle tips can be controlled to \({V}_{{OC},\, {max} }\), the actual transferred charge \({Q}_{1}\) should be significantly less than \({Q}_{{SC},\, {max} }\), as illustrated in Supplementary Fig. 9e. This type of energy loss can be defined as charge-related loss. As shown in Fig. 2b and Supplementary Fig. 10, when the gap distance increases from 0 to 1.2 mm, the time required for transferring 20 μC of charge extends from 2.2 s to 7.4 s. This indicates that a larger gap distance leads to significant charge loss, primarily due to extra discharges between the internal electrodes of the RF-TENG and collisions with air molecules. We hypothesize again by adjusting the gap to lower the discharge voltage to one-third of \({V}_{{OC},\, {max} }\), as shown in Supplementary Fig. 9f, the total transferred charge \({Q}_{2}\) would be greater than \({Q}_{1}\) in Supplementary Fig. 9e, because a lower voltage corresponds to lower charge loss. Thus, a larger gap can minimize voltage loss but results in significant charge loss, while a smaller gap minimizes charge loss but leads to substantial voltage loss. Therefore, an optimal gap distance exists for maximizing energy transfer.

In the experiment, the V-Q measurement at the gap distance of 0.3 mm is outlined in Supplementary Fig. 11, and the results at other gap distances are summarized in Supplementary Fig. 12. It can be seen that the maximum energy transfer occurs at the gap distance of 0.9 mm. Worth noting that, due to the limited accuracy in collecting ultra-high frequency discharge signals and potential crosstalk issues of the instrument, the measured V-Q curve is employed for qualitative analysis rather than quantitative calculations of the energy conversion rate. Consequently, the inductance matching experiments are further employed to determine the actual maximum output power at the optimal gap distance of 0.9 mm, as illustrated in Fig. 2c and Supplementary Fig. 13. In comparison to Fig. 1c (iv), the energy conversion efficiency is enhanced by 1.63 times in this case.

Based on the understanding above, we propose two distinct effective strategies for addressing the voltage loss and charge loss. First, building on the preceding analysis, it is evident that, although more transferred charges can be retained at a small gap distance, significant voltage loss occurs. In principle, reducing the intrinsic output voltage of RF-TENG is anticipated to ameliorate this situation. Consequently, the strategy of electrode segmentation is proposed, as depicted in Fig. 2d. In theory, dividing an electrode unit into three segments allows RF-TENG to transfer an equal charge within one motion cycle (Supplementary Fig. 14), while the maximum output voltage is reduced to one-third of the original (Supplementary Fig. 15 and Fig. 1f). It may seem that the RF-TENG with subdivided electrodes has the 1/3 energy of the original in one motion cycle, it is essential to note that the sliding distance is also 1/3 to the original. Therefore, the electrode segmentation strategy can effectively reduce the intrinsic output voltage of RF-TENG without altering the total output energy per unit time. Supplementary Fig. 16 demonstrates the effectiveness of the electrode subdivision strategy using RF-TENG with 12 pairs of electrodes (RF-TENG-12). Figure 2e and Supplementary Fig. 17 display the V-Q curves of RF-TENG-12 at different gap distances, indicating the optimal gap distance of 0.6 mm. Subsequently, the inductance matching experiments are shown in Fig. 2f and Supplementary Fig. 18. As the result, the optimal output power of 1.89 mW is achieved under the inductance value of 470 μH, which is 155% higher to the optimal output power of RF-TENG-6 (Fig. 2c).

Second, as described above, the collision of electrons and air molecules is another significant threat to energy loss (Fig. 2g). Therefore, improving the atmosphere in the needle tip gap would contribute to a higher energy conversion efficiency. As shown in Fig. 2h and Supplementary Fig. 19, with a fixed gap distance of 0.3 mm, as the air pressure decreases from 97 kPa to 1 kPa, the time required for charge transfer (20 μC) diminishes, which indicates that less air molecules is benefit to charge transfer. Moreover, even ensure the same discharge voltage, low air pressure reduces the charge loss, as two specific conditions (Gap: 0.9 mm, air pressure: 97 kPa and Gap: 1.5 mm, air pressure: 45 kPa) demonstrated in Supplementary Fig. 20. The matrix measurements of the transferred energy on gap distance and air pressure are further carried out (Supplementary Fig. 21). As the results, the maximum transferred energy occurs at 35 kPa pressure and 1.2 mm gap distance. Additionally, as the air pressure continues to decrease, due to the lacking of air molecules for building discharge tunnel, the transferred energy changes to decrease. Subsequently, a maximum output power of 3.17 mW (an increase of 168% compared to Fig. 2f) is achieved under the optimized discharge and inductance conditions (Fig. 2i and Supplementary Fig. 22).

Based on the above strategy, a RF-TENG-18 (with 18 pairs of electrodes) is fabricated for evaluating the actual energy conversion efficiency (Fig. 2j and Supplementary Fig. 23). Figure 2j(i) illustrates the matching impedance characteristics of the device. The transferred energy of RF-TENG-18 at different gap distances under various air pressures is depicted in Supplementary Fig. 24. Figure 2j(ii) showcases the optimal charging effect after inductance matching, resulting in the charging power of 7.35 mW. Comparatively, this achievement signifies a significant improvement in actual energy conversion efficiency compared to previous work (Supplementary Fig. 25 and Supplementary Table 1), once again affirming the efficacy of the proposed strategy. Indeed, if further steps are taken following the aforementioned logic, such as subdividing more units, reducing the size of TENG devices (resulting in lower intrinsic voltage), selecting a more suitable discharge atmosphere type, etc., the energy utilization efficiency can be further enhanced.

System circuit design of the TENG based power supply

As shown in Fig. 3a, b, to ensure the normal operation of electronic devices, the Zener diode is always used to stabilize the voltage of the power supply. However, it results in huge energy loss when the output of TENG exceeds a certain value and inapplicability for irregular energy situation. In this case, developing the TENG based power supply with energy storage and output regulation functionality holds great necessity for practical applications (Fig. 3c, d). The power management chip LTC-3588-1 (Hereinafter referred to as 3588) as the promising solution has been attempted in various types of TENGs as a voltage regulator module44,45,46. However, the direct integration of TENGs with 3588 has shown a remarkably low energy utilization rate. In this work, as dipected in Fig. 3e, through rational combination with our proposed EM circuit, the surprising features of the chip are explored. Supplementary Fig. 26a illustrates the circuit diagram of TENGs directly connected to 3588, with fixed values for C1, C2, C3, and L provided in Table S2. Cin is selected as the input capacitor with a value of 220 µF. By changing the connection nodes of D1 and D0, as depicted in Supplementary Fig. 26b, c, the output voltage can be set to 1.8 V, 2.5 V, 3.3 V, and 3.6 V. Subsequently, Supplementary Fig. 27 demonstrates that the output voltage reaches the set value when the voltage of Cin exceeds a certain threshold. To further analyze the energy conversion mechanism of TENGs combined with 3588, an output capacitor Cout with a capacitance of 3.3 mF is connected in parallel to the output port (Supplementary Fig. 28a). The dynamic voltage variation curves of Cin and Cout in Fig. 3f displays that, once the voltage of Cin drops, the voltage of Cout climbs, providing insights into the details of energy transfer from Cin to Cout. Therefore, the energy transfer efficiency can be expressed as:

$$\eta=\frac{{E}_{O}}{{E}_{I}}=\frac{\frac{1}{2}{C}_{{out}}({{V}_{O2}}^{2}-{{V}_{O1}}^{2})}{\frac{1}{2}{C}_{{in}}({{V}_{I1}}^{2}-{{V}_{I2}}^{2})}$$
(3)

Where \({E}_{O}\) and \({E}_{I}\) are the energy variation of Cout and Cin, respectively. The energy transfer efficiency during 11 cycles gradually increases with each cycle and reaches the maximum of 93.6% (Supplementary Fig. 28b), which fully proves the low power consumption of 3588. The output situation under different loads after removing Cout is further investigated. Although 3588 can output a constant voltage (Supplementary Fig. 26), it needs to meet certain requirments. The power consumption at the load end can be expressed as:

$${P}_{{Load}}=\frac{{U}^{2}}{R}$$
(4)

Where \(U\) and \(R\) denote the output voltage and load resistance values, respectively. When the power consumption exceeds the input power of 3588 (load resistance is below a certain threshold), the voltage stabilization function of 3588 cannot be maintained, as depicted in Fig. 3g. Supplementary Fig. 29 illustrates the output voltage curves as the load resistance increases from 1 kΩ to 200 kΩ. As the resistance value rises, the continuity of the output from 3588 improves, and when the resistance value exceeds 60 kΩ, the output voltage maintains a constant form. Supplementary Figs. 30 and 31 outline the calculation method for RMS power at 20 kΩ and the RMS power under different loads, respectively. The output power at 60 kΩ reaches 56.8 µW. In comparison to the maximum output power (17.31 mW) of RF-TENG-18 in Fig. 2j(i), the overall energy conversion efficiency of 3588 is merely 0.3%.

Fig. 3: System circuit design of the TENG based power supply.
figure 3

a A condition that the output power of TENGs is greater than the required power. b A stabilizing voltage scheme. c An overall plan for stable energy supply. d A metaphorical technical roadmap. e Circuit diagram for achieving energy storage and stable energy supply. f Dynamic voltage variation curves of \({{{{\rm{C}}}}}_{{{{\rm{in}}}}}\) and \({{{{\rm{C}}}}}_{{{{\rm{out}}}}}\). g Output voltage under different resistances. h Dynamic voltage variation curves of \({{{{\rm{C}}}}}_{{{{\rm{in}}}}}\) and \({{{{\rm{C}}}}}_{{{{\rm{out}}}}}\) using the optimized 3588. i Energy transfer efficiency of these 10 cycles in Fig. 3h. j RMS output current, voltage, and power under different loads. Source data are provided as a Source data file.

It seems that the low power consumption and low energy conversion efficiency of 3588 are contradictory. Actually, the low power consumption of 3588 refers to the low energy loss from Cin to the output end. Hence, the ultra-low conversion efficiency is attributed to the inefficient energy transfer from RF-TENG to Cin. This perspective is validated by the internal circuit diagram of 3588, as depicted in Supplementary Fig. 32. Directly rectifying the output of TENGs through the pins of PZ1 and PZ2 to charge Cin inevitably results in an extremely low energy extraction rate. Therefore, to enhance the charging efficiency for Cin and improve the overall energy conversion efficiency of 3588, the efficient EM strategy implemented above is applied as the front-end circuit (Fig. 3e). As shown in Supplementary Fig. 33 and Fig. 3h, i, the energy transfer process from Cin to Cout is measured while employing EM circuit. Although the voltage curves are similar to Fig. 3f and Supplementary Fig. 28b, the time needed for energy conversion is largely reduced. Subsequently, Supplementary Fig. 34 illustrates the output voltage under different loads, indicating that the optimized 3588 can achieve voltage stabilization when the load exceeds 500 Ω. Figure 3j summarizes the RMS output current, voltage, and power under different loads. The output power of RMS reached 6.9 mW at 500 Ω, performing 121.5 times improvement in the overall energy conversion efficiency.

Application of the EM strategy for various TENGs

As analyzed above, when the power consumption is lower than the input power of the energy module, the output terminal can uphold a constant voltage to ensure stable load operation. Meanwhile, the excess energy would be stored in the Cin for later use. As attested in Fig. 4a, the voltage of Cin keeps growing when operating RF-TENG-18, while the output remains stable at the load even RF-TENG-18 stops. The power supply should shut down only when the voltage of Cin drops to 2.9 V. The experimental results underscore that the EM strategy proposed here accomplishes the function of energy storage and output regulation, presenting significant practical value for self-powered system based on harvesting irregular mechanical energies. Figure 4b, c and Supplementary Movie 2 demonstrates the practical application of RF-TENG-18 in delivering stable energy to a hygrothermograph under discontinuous working conditions. It showcases that, the operation of RF-TENG-18 for 21 s ensures the stable operation of the hygrothermograph for impressive 417 s. In fact, this energy management solution that effectively combines TENGs with existing integrated circuit technologies has broad applicability, as stated in Supplementary Fig. 35 and Supplementary Note 5.

Fig. 4: Application of the EM strategy for various TENGs.
figure 4

a Input voltage \({{{{\rm{V}}}}}_{{{{\rm{in}}}}}\) and output voltage \({{{{\rm{V}}}}}_{{{{\rm{out}}}}}\) curves when the load resistance is 1 kΩ. b Specific \({{{{\rm{V}}}}}_{{{{\rm{in}}}}}\) and \({{{{\rm{V}}}}}_{{{{\rm{out}}}}}\) curves when powering a hygrothermograph. c Optical photograph of the application. d 3D structure of a CS-TENG. e Working mechanism of CS-TENG. f Powering a calculator by hand-driven CS-TENG. g 3D structure of DC-TENG. h Working mechanism of DC-TENG. i Powering a hygrothermograph by hand-driven DC-TENG. j Dynamic voltage curves and (k) actual photo of power supply to a game console (by hand-driven DC-TENG). l Widespread adoption of the TENG based power system in practical applications. Source data are provided as a Source data file.

To demonstrate the universality of the proposed EM strategy for various TENGs, two typical modes are fabricated. For each TENG device, the capacitance value of Cin needs to be thoughtfully selected to align with the output characteristic (Supplementary Table 3 provides reference values for selection). In addition, Supplementary Fig. 36 and Supplementary Note 6 summarize the entire energy management setups for all types of TENGs. Figure 4d, e, Supplementary Fig. S37, and Supplementary Note 7 illustrate the 3D structure and working principle of a contact-separated TENG (CS-TENG), while the circuit connection with 3588 is depicted in Supplementary Fig. 38. Demonstration of manually driving CS-TENG and powering a calculator stably are presented in Fig. 4f and Supplementary Movie 3. Subsequently, a direct current TENG (DC-TENG) is further produced, as shown in Fig. 4g, h, Supplementary Fig. 39, and Supplementary Note 8, along with the circuit connection to 3588 in Supplementary Fig. 40. As can be found in Fig. 4i and Supplementary Movie 4, a DC-TENG with intermittent operation can also provide continuous power to a hygrothermograph. Besides, Supplementary Movie 5 and Fig. 4j, k demonstrate an interesting application wherein the energy produced by human exercise can be harnessed to power entertainment systems during periods of rest. Therefore, the broad applicability of this EM strategy is fully demonstrated, contributing significantly to the widespread adoption of TENGs in practical applications, such as self-powered sensing systems, intelligent Internet of Things, and leisure entertainment (Fig. 4l). Finally, based on the foregoing description, the capacitor Cin emerges as a pivotal electronic component for energy storage. The average self-loss power of Cin, with different capacitance types and values, is measured, as depicted in Supplementary Figs. 4143, revealing a loss power ranging from several to tens of microwatts. Thereby, the potential for enhancing the energy storage capacity of Cin in the future is anticipated.

Discussion

In summary, we have tackled two practical challenges in the EM system of TENGs. Firstly, we identify that switches in the EM circuit induce significant losses in voltage and charge during energy transfer process, leading to ultra-low energy conversion efficiency. By synergistically optimizing the TENG and switch configurations, the maximum energy extraction effect is attained. Ultimately, we successfully enhance the actual energy conversion efficiency of the rotating TENG to 42.5%, which is 8.5 times higher than that of directly using the current EM circuit. Secondly, we propose an efficient energy storage strategy applicable to multi-mode TENGs by integrating a commercial energy processing chip, which enabled stable power supply for electronic devices under irregular mechanical stimuli. Above all, this work not only provides an in-depth energy transfer mechanism between TENGs and energy management circuits but also establishes a TENG-based constant voltage power supply system with energy storage capabilities. This holds significant guiding implications for the subsequent development of TENG energy management. Moreover, the successful integration of TENGs with existing circuit technologies paves the way for enhancing the performance and application flexibility of TENGs.

Methods

Fabrication of the RF-TENG

The RF-TENG consists of a stator (lower part) and a rotor (upper part). To begin, two circular acrylic substrates with a diameter of 200 mm and a thickness of 4 mm were prepared using a laser cutting machine, each featuring an 8 mm diameter hole in the center for bearing installation. For the lower part, two sets of output electrodes, each containing 6 units, were affixed in a complementary radial arrangement onto one of the aforementioned acrylic substrates. Each output electrode has a center angle of 26°, with inner and outer radii of 30 mm and 184 mm, respectively. Subsequently, a 1 mm thick PU foam film was applied to cover all electrodes. Moving on to the upper part, a 2 mm thick foam cushion layer, mirroring the shape of a set of output electrodes in the lower part, was adhered to the acrylic substrate. Next, a 30 µm thick PTFE film was affixed to the foam. This completes the RF-TENG with 6 pairs of electrodes. The production process of RF-TENGs with different electrode pairs was similar to the above.

Fabrication of CS-TENG

The CS-TENG comprises two parts: upper and lower. For the upper part, begin by affixing an aluminum foil with an area of 6 cm by 6 cm onto an acrylic board measuring 3 mm by 7 cm by 7 cm. Next, apply a layer of PTFE film with a thickness of 30 µm onto the aluminum foil. Turning to the lower part, start by pasting foam with dimensions of 3 mm by 7 cm by 7 cm onto an acrylic board with dimensions of 3 mm by 7 cm by 7 cm. Subsequently, attach a copper electrode with an area of 6 cm by 6 cm onto the foam. This completes the production of the CS-TENG.

Fabrication of DC-TENG

The DC-TENG consists of a stator (lower part) and a slider (upper part). To create the lower part, affix a foam layer measuring 1 mm by 25 cm by 10 cm onto an acrylic board with dimensions of 3 mm by 25 cm by 10 cm. For the upper part, begin by attaching a 100 µm thick PTFE film onto an acrylic board with dimensions of 3 mm by 6 cm by 4 cm. Subsequently, adhere copper electrodes to the two sides of the acrylic. This completes the production of the DC-TENG.

Construction of management circuit

The needle tip discharge switch is crafted from the needle tube of two syringes. The diode in the rectifier bridge is of model 2CL77A, and the flyback diode is of model MUR 460. H-shaped inductors and electrolytic capacitors are chosen for the LC in the circuit.

Measurement and characterization

The rotation process of the RF-TENG is achieved using a stepper motor (86BYG250D). The sliding process of the DC-TENG and the contact separation process of the CS-TENG are both controlled by a linear motor (LinMot PS01-37-120-C). Current and charge measurements are performed using an electrometer (Keithley 6514). All discharge related voltages and TENGs output voltages were measured by a high-speed electrostatic voltmeter (Trek 370) or a voltage probe (Tektronix P6015A) and an oscilloscope (Keysight DSOX4024A), with the exception of voltages measured by Keithley 6514. The regulation of atmospheric pressure is achieved in a vacuum box equipped with a vacuum pump.