Breathable, transparent, waterproof, flexible and high-output triboelectric nanogenerators for sport monitoring and speech recognition

Abstract

Triboelectric nanogenerators (TENGs) with exceptional advantages are imperative for the development of next-generation wearable devices. In this study, we present a waterproof, flexible, transparent and breathable TENG fabricated by a cost-effective method. The innovative triboelectric layer can fold over 180°, retaining 83% transmittance compared to air, and achieve a water contact angle of 109.3°. Notably, we enhanced the nitrocellulose/silicon resin-based TENG's (NC/Si RS-TENG) output performance by 450%. Additionally, we explored the TENG performance with shape-adjustable electrodes, revealing a 5300% attenuation ratio between the electrode area and the electrical output performance and providing an approach to enhance the wearing comfort of the device. Furthermore, we introduce a compact triboelectric sensor (TES) for human–machine interface, sport monitoring and speech recognition. These findings broaden the prospects of shape-adjustable electrode TES in various domains, underscoring their significance in both fundamental research and practical applications.

---

1. Introduction

Recently, the number of sensors and smart wearable devices have shown dramatic growth due to their demand among humans in various fields,^1–8 such as sport monitoring,⁹ health care,¹⁰ human–machine interface^11–13 and auxiliary communication.¹⁴ Generally, the direct energy of those devices is electric energy stored in batteries, which may inevitably lead to environmental pollution due to the use of thousands of batteries.¹⁵ With the continuous efforts of researchers, several emerging energy harvesting strategies, such as piezoelectric nanogenerators,¹⁶ triboelectric nanogenerators¹⁷ and thermoelectric nanogenerators,¹⁸ have been innovated. Due to the diversity of materials, ease of fabrication and excellent output performance, TENGs have caught the eye and gained the interest of researchers.¹⁹ In addition, many researchers are focusing on improving the performance of TENGs. In terms of improving output performance, copper nano-wires,²⁰ silver nano-wires,⁶ MXene²¹ and copper powder¹⁰ are used to enhance the charge attraction ability. However, the breathability and transparency decrease as the particle content increases.²² Although the breathability and transparency can be optimized by combining the electrospinning and dielectric modulation methods,²³ some limitations, such as high energy consumption and low efficiency in fabrication still exist.

To address these challenges, we can explore two key approaches. First, from the perspective of the triboelectric layer, we favor the use of porous materials and appropriate modification techniques. Nitrocellulose (NC) films, for instance, represent a porous tribo-positive material,²⁴ which are both soluble in ester solvents and amenable to straightforward modification. Several studies have already been conducted to establish the feasibility of NC-based triboelectric nanogenerators (NC-TENGs).^25–27 However, there is a gap in the literature concerning a triboelectric layer that simultaneously achieves high transparency, breathability, and outstanding output performance. On the triboelectric device front, enhanced transparency and breathability can be achieved through the thoughtful design of suitable electrodes.²⁸ However, the use of customized electrodes in TENG wearable devices is expensive.²⁹ Besides, the conventional approach employed by most researchers involves using a fully solid electrode affixed to the back of the triboelectric layer, which is conducive to the maximum charge induction. A subset of researchers have explored alternative methods, such as finger-insertion electrodes or mesh electrodes in the construction of TENGs.^30,31 In addition, they have observed that the device performance is not reduced, which can be attributed to the fact that charge induction occurs even in areas of the triboelectric layer that are not directly covered by the electrode.³²

Nonetheless, to construct a high-output, transparent, and breathable TENG wearable device, several critical tasks remain. On one hand, the development of a cost-effective modification method for achieving high-performance triboelectric layers is of paramount importance. On the other hand, it is imperative to further elucidate the quantitative relationship between electrode properties and the performance of TENGs, seeking a balance between the electrical output performance and wearing comfort of devices.

Herein, a breathable and transparent TENG was fabricated by a low-cost method. The modified triboelectric layer possesses excellent transparency (83%, compared with air), resistance to moisture (water contact angle (WCA): 109.3°), breathability (water vapor transmittance rate: 650 g m⁻² d⁻¹) and flexibility (capable of folding over 180°). The NC/Si RS-TENG was constructed to investigate the output performance. The open-voltage and transferred charges are enhanced from 50 V, 21 nC to 225 V, 77 nC, and the mechanism of enhanced output performance is also demonstrated. Building upon the high-performance triboelectric layer, the quantitative relationship between the characteristics of electrodes and output performance is uncovered. Finally, the performance of the shape-adjustable electrode NC/Si RS-TENG in self-powered sensing, sport monitoring and speech recognition is demonstrated, the accuracy of speech recognition is 99.41% leveraging Deep Learning (DL) algorithms. These outcomes underscore the wide-ranging applicability of this technology across scientific disciplines, offering promising avenues for further research and practical implementations.

2. Results and discussion

2.1. Design and properties of modified NC (NC/Si RS)

The transparent, waterproof, and breathable triboelectric layer consists of modified NC, synthesized using a straightforward ‘one-pot’ method (Fig. 1a). In this process, a mixture was prepared by sequentially adding pure NC and various solvents into a beaker. The NC/Si RS was uniformly and firmly adhered to the bottom of Petri dish after evaporation of the solvent, forming a transparent and flexible triboelectric layer. The microscopic morphologies of the NC/Si RS film were shown in Fig. 1b. It is noteworthy that the introduction of Si RS into NC generates a host of micro/nano particles inside and outside of the NC (Fig. 1b) as the solvent evaporates. The surface roughness of NC/Si RS is improved by those micro/nano particles, which strengthens the contact of triboelectric layers and improved the triboelectric performance.³³ In addition, change in the roughness of NC/Si RS leads to a decrease in the surface energy, endowing a high hydrophobicity of NC/Si RS.³⁴ The fabricated triboelectric material combines good waterproofness, transparency, breathability, flexibility and triboelectric positivity, exhibiting excellent potential as a wearable device (Fig. 1c). As illustrated in Fig. 2a, the prepared NC/Si RS (2.5 cm × 2.5 cm) was placed on an A4 print paper, which is hardly noticed because of the excellent transparency. Benefiting from the excellent flexibility of NC/Si RS, it can be folded easily (Fig. 2b). Moreover, the thickness and weight of NC/Si RS were measured (Fig. S1†), and the light weight and thickness of NC/Si RS (0.0259 g, almost 60 μm) exhibit excellent wearability in wearable devices.

Fig. 1 Process of fabricating NC/Si RS and the diagram of NC-TES. (a) Diagram of the preparation procedures of NC/Si RS. (b) Schematic diagram of stratified NC/Si RS and the microstructure. (c) Potential application of NC-TES.

Fig. 2 Properties and performance of NC/Si RS. (a) NC/Si RS with a size of 2.5 cm × 2.5 cm. (b) Flexibility of NC/Si RS. (c) EDS of the Si element in NC/Si RS. (d) AFM images of the film. (e) XPS results of NC/Si RS. (f) Diagram of waterproof and breathable NC/Si RS. (g) Difference of waterproofness between NC films and NC/Si RS. (h) WVTR of NC/Si RS with Si RS and without Si RS. (i) Transmittance of NC/Si RS with different contents of Si RS. (j) Structure of the prepared NC/Si RS-TENG.

The micro/nano particles were found in the whole NC/Si RS (Fig. 1c), and those particles are adhered to inside and outside of NC/Si RS. Energy-dispersive spectroscopy (EDS) images validate that the silicon element exists in the NC/Si RS film (Fig. 2c). The surface roughness of NC/Si RS was assessed using an atomic force microscope (AFM), with results displayed in Fig. 2d. It is clear that the NC/Si RS shows enhanced roughness compared with the film without Si RS. To investigate the effect of Si RS in NC and analyze the chemical composition of NC/Si RS, the XPS analysis was performed. The full spectrum of NC/Si RS is shown in Fig. 2e(I) and S2.† Compared with the pure NC film, two new characteristic peaks of Si 2p and F 1s were generated, which is ascribed to the doped Si RS. High-resolution XPS spectrums of C 1s, N 1s, O 1s, F 1s and Si 2p on the surface of NC/Si RS (with Ri RS) were conducted (Fig. 2e(II–IV)). The spectrum of C 1s can be resolved into four characteristic peaks at 284.8 eV (C–C), 286.4 eV (C–O), 287.5 eV (C–O–C) and 288.9 (C–F), respectively.^35–37 The N 1s spectrum shows the peaks at 407.65 eV (–NO₃).³⁸ The O 1s spectrum was resolved into three peaks at 533.6 eV (O–NO₂), 532.5 eV (C–O–C) and 532.0 eV (SiO₂).^36,39 The spectrum of F 1s shows an obvious peak at 688.0 eV, which may relate to the C–F.⁴⁰ In addition, the spectrum of Si 2p exhibits the peak at 103.3 eV, indicating that the SiO₂ particles are distributed in NC/Si RS.³⁹

As the most fundamental requirements for wearable devices, waterproofness and breathability are significant factors.⁴¹ The mechanism of waterproof and breathability is illustrated in Fig. 2f, water-drops are blocked due to those particles adhered on the surface of NC/Si RS. On the contrary, the tiny water molecule passes through the film smoothly; therefore, the NC/Si RS film shows breathability. Fig. 2g and Video S1† display the difference of waterproofness. The water-drops dripped on the surface of the pure NC were absorbed, while the NC/Si RS film shows good waterproofness. To investigate the influence of Si RS on the waterproofness of the NC/Si RS film, the films doped with different contents of Si RS were evaluated. The NC (0.15 μm) film shows a WCA of 0°, while the WCA of NC/Si RS film could reach 109.3° (Fig. S3†). This is attributed to the increase in micro/nano particles in NC/Si RS, leading to a change in the surface roughness and surface energy,⁴² which is proved by scanning electron microscopic (SEM) images (Fig. S4†) and 3D laser confocal measuring microscopic images (Fig. S5†). Besides, the micro/nano silicon particles are intrinsic hydrophobic materials,⁴³ which increases the WCA of the NC/Si RS film further. To demonstrate the breathability of the NC/Si RS film, control experiments were conducted and the results are shown in Fig. S6.† Water vapor passed through the NC/Si RS film to a top beaker and adhered to the wall of the beaker after 1 minute, but the water vapor cannot pass the plastic wrap. To quantitatively study the breathability of the NC/Si RS film, the water vapor transmittance rate (WVTR) of the NC/Si RS film was measured based on the Fickian diffusion (the details of the calculation can be found in Note S1†). It is obvious that the WVTR of the NC/Si RS film shows a downward tendency as the environment temperature changes from 35 °C to 18 °C (Fig. 2h), which is due to the high temperature that boosts the molecular thermal motion, so that more water vapor passes through the film.⁴⁴ In addition, the film without doping Si RS possesses a higher WVTR and the gap of WVTR is more obvious at 35 °C, which is attributed to some microporous channels being hindered by Si RS and is proved by SEM images shown in Fig. S7.† Besides, the WVTR of single-electrode TENG was performed and the result is shown in Fig. S8.† It should be noted that the WVTR of NC/Si RS and the corresponding single-electrode TENG is 536 g m⁻² d⁻¹ and 521 g m⁻² d⁻¹ (35 °C), respectively, which is still higher than human skin (∼200 g m⁻² d⁻¹).⁴⁵ It is important to highlight that pure NC (0.15 μm) lacks transparency. The transmittance of NC/Si RS slightly decreases (from 83% to 72%) with the increase in Si RS content due to altered light diffusion by particles (Fig. 2i). This variation in transmittance is linked to changes in light refraction caused by differences in particle size and distribution within and outside the NC/Si RS film. Furthermore, the mechanical properties of NC and NC/Si RS were investigated; the tensile strength and elongation of NC/Si RS exhibits an increase of 333% and 540%, respectively (Fig. S9†). Based on the above discussion, the NC/Si RS shows good transmissivity (83% relative to air), stretchability of 45% (strain tolerance of human arm skin is about 27%⁶), waterproofness of 109.3° (human skin:⁴⁶ 83.61 ± 6.103°) and breathability of 536 g m⁻² d⁻¹ (human skin:⁴⁵ ∼200 g m⁻² d⁻¹), and these quantitative values indicate that our NC/Si RS are suitable for wearable devices and show good wearing comfortableness.

2.2. Optimization of triboelectric performance

The NC/Si RS film was employed for the preparation of positive triboelectric layer because of its positive triboelectric sequence.⁴⁷ The structure of the NC/Si RS-TENG is shown in Fig. 2j, a conductive fabric was used for the electrode of NC/Si RS due to the breathability, and the preparation details of NC/Si RS-TENGs are exhibited in Fig. S10.† The triboelectric performance of NC/Ri RS-TENGs was improved remarkably compared to the NC/Si RS-TENG fabricated by pure NC films. The NC/Si RS-TENG with Si RS of 0.2 g exhibits the best output performance, with open-circuit voltage (V_oc), short-circuit current (I_sc) and transformed charge (Q_sc) values of 225 V, 5.2 μA and 77 nC, respectively (Fig. 3a–c). In addition, the results indicate that the contents of Si RS are highly related to the output performance of NC/Si RS-TENGs, and increasing the contents of Si RS in the range of 0.00 g to 0.20 g enhances the output performance of NC/Si RS-TENGs obviously. The mechanism of the enhancement ascribes to the effect of changed surface roughness by micro/nano particles and induced F functional groups,⁴⁸ which can be validated by AFM, XPS and 19F NMR results (shown in Fig. 2d, e and S11†). However, increasing the contents of Si RS steadily causes a decrease in output performance, which is attributed to the reuniting of particles and too much F functional groups that lead to the reduction in tribo-positive sequences. It is proved by SEM figures, surface 3D scan images and XPS images of NC/Si RS shown in Fig. S2, S4 and S5.†

Fig. 3 Triboelectric performance, operation mechanism and working stability of the NC/Si RS-TENG. (a–c) V_oc, I_sc and Q_sc of the NC/Si RS-TENG with different amounts of Si RS. (d) Working mechanism of the NC/Si RS-TENG: (I) NC/Si RS-TENG is in electrostatic balance (II) the state of being pressed, (III) being released, (IV) being released and (V) being pressing again. (e) Finite simulation results of the NC/Si RS-TENG: (I) separated distance between positive and negative triboelectric layers −1 mm, (II) −11 mm, (III) −31 mm and (IV) −51 mm. (f) Impedance matching of the NC/Si RS-TENG. (g) Maximum power density of the NC/Si RS-TENG. (h) Using the NC/Si RS-TENG to charge different capacitances. (i) Working stability of the NC/Si RS-TENG in 10 000 times. (j) Stretched stability of the NC/Si RS-TENG.

To further investigate the performance of NC/Si RS-TENGs, the optimal content of 0.2 g Si RS was chosen. Based on the working principle of coupling effects of contact electrification and electrostatic induction, the mechanism of contact-separation NC/Si RS-TENG is illustrated in Fig. 3d. Initially, the system was in electrostatic equilibrium. Under the influence of external force, the top and bottom layers came into contact, leading to the generation of positive and negative charges due to the triboelectric sequence disparity between the NC/Si RS film and PDMS (Fig. 3d(I and II)). The two triboelectric materials started to separate gradually, while the external force disappeared; therefore, the electrostatic equilibrium was broken. Charges shifted from the top to the bottom layer via an external circuit, creating instantaneous current and potential difference. The potential difference peaked when the external force disappeared (Fig. 3d(III and IV)). When the force was reapplied to bring the layers closer, charges flowed in the opposite direction. Contact and separation cyclically generate periodical triboelectric signals. To better understand the process of charge transfer and confirm the working mechanism of NC/Si RS-TENGs, the simulation method was conducted using the COMSOL Multiphysics software; the results are shown in Fig. 3e.

For TENGs, triboelectric performance and work stability are significant factors. Fig. 3f shows the voltage and current of NC/Si RS-TENGs under the external load resistance that changed from 0.5 MΩ to 2 GΩ. The voltage was enhanced when the external load resistance was increased at the external circuit, while the current exhibited opposite tendency owing to the inhibitory effect caused by resistance. The instantaneous power density of NC/Si RS-TENGs is shown in Fig. 3g, which is 0.383 W m⁻² at an external load of 100 MΩ. The calculated details of the power density can be found in Note S2.† To assess practical electrical performance, capacitors of 1 μF, 4.7 μF, and 10 μF were used. The 1 μF capacitor was charged to 3 V in 7 seconds, while the 10 μF capacitor took 80 seconds (Fig. 3h). The NC/Si RS-TENG demonstrated working stability through 10 000 consecutive runs under constant conditions. As displayed in Fig. 3i, the V_oc is almost not changed violently, and the relative standard deviation (RSD) was calculated (RSD = 1.34%), which indicates that the NC/Si RS-TENG possesses good working stability and can operate for a long time. Additionally, the experiment of periodic fold was conducted, and the electrical output performance of NC/Si RS-TENGs is not changed obviously within 8000 times strain and stretch (Fig. 3j). Table S2† exhibits the comparison between this work and the other reports, indicating that the mechanical properties and output performance of NC/Si RS-TENGs are feasible to prepare self-powered sensing device with high performance.

2.3. Quantification of the relationship between electrode properties and triboelectric performance

The characteristics of the electrode are highly related to the performance of TENGs.⁴⁹ Different electrode materials were used to investigate the relation between the output performance of NC/Si RS-TENGs and electrode materials. The result indicates that the electrode material has little influence on the output performance of NC/Si RS-TENGs (shown in Fig. S12†). The area of the electrode is less than that of the triboelectric material, leading to a reduction in electrical output performance.^50,51 Seven square electrodes with different sizes (including 1 cm × 1 cm, 1.414 cm × 1.414 cm, 1.732 cm × 1.732 cm, 2 cm × 2 cm, 3 cm × 3 cm, 4 cm × 4 cm and 4.5 cm × 4.5 cm) were prepared to quantitatively investigate the relationship between electrode area and electrical output performance (Fig. 4a–d). The output performance of NC/Si RS-TENGs shows an upward trend with the increase in electrode area, but the tendency is blocked when the area of the electrode is larger than that of the triboelectric material since most charges are only induced at the point of contact. Fig. S13† exhibits the results of finite element simulation, showing the same trend as the experiment result.

Fig. 4 Quantification of the relationship of electrode properties and triboelectric performance. (a) Electrode with different areas. (b–d) Electrical output performance of the NC/Si RS-TENG equipped with different areas of the electrode. (e) Difference in output performance with the same area but unequal shape. (f) NC/Si RS-TENG with the circle electrode that the areas are π cm². (g) NC/Si RS-TENG with the circle electrode that the areas are π cm². (h) NC/Si RS-TENG with rectangle electrode (1 cm²) and the offset distances are changed from 0 to 20 mm.

In general, the properties of triboelectric devices such as transparency and breathability are indeed affected by the electrode characteristics, where a smaller electrode area tends to result in lower output performance. However, we observed that the TENG with a grid electrode exhibited nearly consistent output performance compared to a solid electrode (Fig. S14†), as charges can be induced due to the electrostatic induction even if a part of area of the triboelectric layer has no electrode.³² In addition, inspired by the different electrical output performance with diverse patterns of 3D printing,⁵² three electrodes with different shapes are designed (Fig. S15†). As shown in Fig. 4e, the NC/Si RS-TENG equipped with the “king” electrode displays a higher performance than that of the “cross” electrode and circle electrode, as the “king” electrode induces more charges.

From the aforementioned results, it can be inferred that the electrical output performance is not solely dependent on the electrode's surface area but is also influenced by its shape. To verify this guess, the solid and hollow electrodes with different shapes but same areas, including triangle, square and circular, were taken into consideration (Fig. S16†). It is interesting that the electrical output performance of the NC/Si RS-TENG equipped with diverse electrodes shows an obvious difference, whereas the area of the electrode is the same. The electrical output performance of hollow electrode TENGs is smaller than that of the solid electrode TENGs (Fig. 4f). In addition, the results indicated that the output performance is increased with the shape of electrode approaching the circle. Those results indicate the varied distribution of charges in the triboelectric layer and the circular electrode can induce more charges exactly than others shape. To further investigate the output performance of the circular electrode, four electrodes with the same area (π cm²) but different radii were fabricated (including one solid electrode and three hollow electrodes, shown in Fig. 4g). It should be noted that the electrical output performance shows a downward tendency with the increase in radius (Fig. 4g). Besides, as the control experiment, one group of square and hollow electrodes with the same area (π cm²) were prepared to further explore the phenomenon (Fig. S17†). With the increase in the diameter of hollow electrodes, the electrical output performance exhibits the same trend with the circular electrode NC/Si RS-TENG. Smaller electrodes (1 cm²) also show the same trend (Fig. S18†). In addition, those circular electrode TENGs possess a higher electrical output performance than that of square electrode TENGs (compared with the results of Fig. 4g and S17†). Taking those experimental results into consideration, there may exist a gradient that contributes to the difference in the electrode position and electrical output performance.

As a proof of concept, five electrodes (0.5 cm × 2 cm) with the same area were prepared, and each of the electrodes was pasted on the surface of the triboelectric material with the corresponding offset distance of 5 mm. Compared with the position of the electrode with zero offset distance, the lowest electrical output performance is observed at an offset distance of 20 mm (Fig. 4h). Increasing the offset distance leads to an obvious decrease in the output performance, which is regarded as a proof of the gradient. Benefiting from the tenacious electrical output performance even though the area of electrode has decreased markedly, it is going to be possible that it exhibits the triboelectric layer properties as far as possible at the cost of electrical output performance to some extent.

To demonstrate the performance of the shape-adjustable electrode NC/Si RS-TENG (2 cm × 0.5 cm), the output performance of the NC/Si RS-TENG working in diverse external forces was investigated. The output performance of the NC/Si RS-TENG displays an upward tendency with the increase in external forces (Fig. S19†), and the linear relationship that exists in external forces is at 20 N to 50 N. Fig. S20† displays the relationship between frequencies and output performance. V_oc remains stable, while I_sc notably increases when frequencies range from 0.25 to 2 Hz (the calculation can be found in Note S3†). In addition, the response properties of the NC/Si RS-TENG were also investigated (shown in Fig. S21†), and the device possesses excellent response performance (response time: 31 ms, recovery time: 10 ms). Combining the good performance in external force, frequencies and response properties of the TENG shows great potential in self-powered sensing.

2.4. Self-powered sensing and application

The NC/Si RS offers flexibility, waterproofing, breathability, and transparency, correspondingly, and the constructed NC/Si RS-TENG shows excellent performance. Based on those properties, a portable single-electrode nitrocellulose-based triboelectric sensor (SNC-TES) was designed to realize real-time sensing and human–machine interaction. The device was made up of NC/Si RS and conductive fabric, and a triboelectric signal was generated when finger was touched (Fig. 5a). The physical mechanism of SNC-TES is demonstrated in Note S4 and Fig. S22.† Due to the small area of the electrode, the NC/Si RS can be pasted on the surface of human skin without influencing the work of daily. To demonstrate the sensing capabilities of the SNC-TES, we designed a PCB integrated with LED lights. The SNC-TES can control these lights, and the PCB circuit is shown in Fig. S23,† while the interaction process is illustrated in Fig. 5b.

Fig. 5 Self-powered sensing performance of SNC-TES. (a) Structure of SNC-TES. (b) Process of triboelectric signal generation and processing. (c and d) Triboelectric signals are generated by using different external forces and different press times, respectively. (e) Triboelectric signals with both unique external forces and press time. (f) Working mechanism of SNC-TES in a gait monitor. (g and h) Dual-mode signals with different motion modes. (i) Signals of jump. (j) Signals of human jump with different heights.

It is critical for a single-electrode TENG sensing device to be able to respond the external stimulation quickly and possess high discrimination for external stimulation.⁵³ To study the sensing properties of the SNC-TES on human skin, various external forces and frequencies were applied to the SNC-TES. With the increase in touching pressure, the LED lights were lightened and the brightness was linked to the touching pressure (shown in Fig. 5c), which indicated that SNC-TES could clearly and quickly respond to the external pressure. Different touch frequencies caused the diverse hold time, which was distinguished easily (Fig. 5d). Additionally, the aptitudes and response frequencies of SNC-TES show good consistence with different pressures and touching frequencies (Fig. 5e and Video S2†).

Benefiting from the outstanding performance of SNC-TES, it was used as electric skin for sport monitoring. In this work, the SNC-TES was attached on the surface below the elbow (shown in Fig. S24†), and the polyurethane (PU) film was pasted on the SNC-TES to prevent the interference of clothes.⁵⁴ The dual-mode signal was generated by SNC-TES theoretically, and the operation mechanism of dual-mode SNC-TES in human sports monitoring was demonstrated as follows. Generally, the redundant charges flow to the ground, thus the whole body is in a static equilibrium state (Fig. 5f). Upon feet contact and separation with the ground, the static equilibrium state is broken and charges are transformed from body to ground to establish a new static equilibrium.⁵⁵ At the same time, the charges of SNC-TES are changed due to the electrostatic induction and the signal is generated. Besides, the arm contacts the SNC-TES when human is in motion and the external force is exerted on SNC-TES, which makes the dual-mode triboelectric signal generated. The electrical signals of walking or jogging process are shown in Fig. 5g–i and Video S3.† It should be noted that each signal (Fig. 5g, walk) possesses two peaks, including a large peak and a small peak since the signal between the foot and the ground is larger than the signal between arms and SNC-TES. The single triboelectric signal caused by the waving arm is displayed in Fig. S25.† Interestingly, the triboelectric signal generated by human motion is related to the contact intensity between human feet and ground. The different signals are generated when the experimenter's feet contact the ground unequal intensity (Fig. S26†). Fig. S27† clearly shows the falling signal after walking for 20 s, which reveals the potential applications of the portable SNC-TES in sport monitoring and healthcare.

To investigate the condition of feet off the ground, a jump test was performed by recording the triboelectric signal of portable SNC-TES (Fig. 5i). It is obvious that the signal of the jump is larger than walk or run, which is attributed to the larger contact area and flowed charges between human body and ground. Additionally, jump with different heights and frequencies was also studied (Fig. 5j and S28†). The generated signal almost shows a linear correlation with the jump height (Fig. 5j). In addition, the fast Fourier transform (FFT) of the gait waveforms was conducted to analyze the signals in the frequency domain. Compared to the walk, the frequency spectrum of run shows a higher peak frequency due to the faster-contracted speed between feet and ground (Fig. S29†). In addition, a wireless monitor device is fabricated to show the potential application of shape-adjustable electrode SNC-TES in sports monitoring (details can be found in Fig. S30 and Video S4†). According to the above-mentioned results, portable SNC-TES can be considered as a promising application for healthcare and sport monitoring.

To assess the impact of a portable SNC-TES on speech recognition, we affixed the SNC-TES onto the throat of a healthy individual to capture the vibration signals generated by the vocal cords. Triboelectric signals were generated immediately when humans pronounce some words, such as nano, energy, self-powered, wearable and device, those acquired signals possess different characteristics due to the unique vibration modes of vocal cords for each word. The generated process of signals is shown in Fig. 6a. The intensity of speech triboelectric signals is highly related to the stress of syllables. For instance, the word ‘energy: ['enədʒi]’ includes three obvious syllables of ‘['en]’, ‘[ə]’ and ‘[dʒi]’, where ‘[ə]’ is pronounced with light syllables, thus the peak of the signal shows an obvious reduction between the triboelectric signal of ‘['en]’ and ‘[dʒi]’ (presents in Fig. S31†). Those characteristics (such as intensity, frequency and amplitude) can be recorded accurately by SNC-TES (Fig. 6c–h). In recent years, ML algorithms have been developed rapidly as an emerging technology for classification and prediction. Particularly, one-dimensional convolutional neural network (1D CNN) is an effective method to extract subtle features of time-domain signals.⁵⁶ Based on the high performance of SNC-TES in speech triboelectric signal acquisition, a speech recognition model was established to distinguish the words.

Fig. 6 Potential application in speech recognition of SNC-TES. (a) Whole process of speech recognition. (b) Network structure of 1D CNN. (c–h) Signals acquired by SNC-TES (including nano, energy, nano-energy, device, self-powered and wearable). (i) Confusion matrix image and the accuracy of the classification.

The model's accuracy was assessed by testing six word categories: ‘nano,’ ‘energy,’ ‘nano-energy,’ ‘self-powered,’ ‘wearable,’ and ‘device.’ Each category comprised 120 training samples. Of these, 20% were allocated for testing, another 20% for validation, and the remaining 60% for training. Initially, a dataset was created by recording the triboelectric speech signals for each pronunciation of the words. After gaining a number of speech triboelectric signals, the 1D CNN model was trained. As shown in Fig. 6b, the model mainly consisted of an input layer, two convolutional layers, two max pooling layers, two full-connected layers and an output layer (the details of the structure could be found in Table. S1†). The triboelectric speech signals served as the input layer for the model, and prior to training, these signals underwent normalization to improve model stability and mitigate overfitting issues associated with limited sample sizes. Subtle features of signals are extracted after convolution, pooling and full-connected, the identified results of different words are shown in the output layer. Fig. S32† shows the accuracy of training and testing in the learning process. The accuracy of training and testing is improved markedly with the increase in epochs, the accuracy reaches the maximum (nearly 100%) and trend to be stable after training for 50 epochs, which indicates that the established model shows an excellent advantage. Fig. 6i shows the confusion matrix for different speech triboelectric signals and the accuracy of recognition is 99.41%, enabling us to identify different words. All the above-discussed results indicate that the SNC-TES possesses good performance in acquiring speech triboelectric signals generated by vocal cord vibration of humans, promising in portable wearable recognition devices.

3. Conclusion

In conclusion, we present a cost-effective method for producing a transparent, waterproof, flexible and high-performance triboelectric layer. The NC combined with Si RS exhibits impressive characteristics including a high water contact angle (109.3°), exceptional light transmittance (83% relative to air), and a remarkable water vapor transmittance rate (650 g m⁻² d⁻¹). The fabricated NC/Si RS-TENG demonstrates exceptional electrical output performance, including a V_oc of 225 V, I_sc of 5.2 μA, Q_sc of 77 nC, a maximum instantaneous power intensity of 0.383 W m⁻², and robust operational stability. We also conducted a comprehensive quantitative investigation into the impact of adjustable electrodes (such as area, shape, and position) on the NC/Si RS-TENG's electrical output performance. Our findings reveal that solid electrodes outperform hollow ones, with electrodes positioned centrally within the triboelectric layer exhibiting superior performance to those placed at the sides. Additionally, we observed a 30% reduction in output performance as the electrode area was reduced by a factor of 16. Leveraging the robust output performance of shape-adjustable NC/Si RS-TENGs, we explored their application in sacrificing output performance to enhance the wearing comfort. Furthermore, we prepared a SNC-TES for applications in self-powered sensing and sport monitoring. With the assistance of ML algorithms based on 1D CNN, the SNC-TES achieved an impressive 99.41% accuracy in speech recognition using triboelectric data. The work has a potential value, promising in energy harvesting, human–machine interfaces, sport monitoring, human healthcare and intelligence.

4. Experimental section

4.1. Fabrication of NC/Si RS as a positive friction layer

A certain quantity of NC films were added into a beaker containing ethyl acetate (EA, 2 g), isopropanol (IPA, 0.2 g), dioctyl phthalate (DOP, 0.04 g) and a Si RS (0–0.3 g) as the solvent. After a stirrer was put into the beaker and the beaker was covered with plastic wrap, the mixture was placed on an electromagnetic stirrer. The mixture was stirred vigorously until the NC was dissolved completely. After 1 hour, the mixture was poured onto a glass garden that has been pre-placed on a heating table stable at 50 °C. The film was peeled off from the glass garden after heating for 2 h to ensure complete solvent volatilization.

4.2. Fabrication of polydimethylsiloxane

The prepolymer and crosslinker of PDMS (Dow Corning) were mixed onto a beaker in a proportion of 10 : 1. Then, the mixture of elastomer was stirred for 0.5 h to ensure that the mixture was well mixed. The mixture was poured on the mold and the mold was placed in a vacuum drying oven at 80 °C for 3 h. Then, the PDMS was cut into different shapes.

4.3. Fabrication of NC/Si RS-TENGs

The NC/Si RS film was pasted on the surface of a piece of aluminum (Al) membrane, in which the NC/Si RS played the part of the positive triboelectric layer and the Al was the electrode for test. In addition, the PDMS was pasted on the surface of another piece of Al, where the PDMS was the negative triboelectric layer. The whole construction of the NC/Si RS-TENG can be found in Fig. S10.† In the practice experiment, two copper wires were connected to the electrodes of positive and negative triboelectric layers, respectively. For the SNC-TES, which worked in the single-electrode mode, the preparation of SNC-TES was demonstrated as follows. The conductive fabric (1 cm × 0.5 cm) was pasted on the surface of the NC/Si RS film, which was used as an electrode of SNC-TES to endow the good breathability of the whole device. The NC/Si RS film was pasted on the surface of the conductive fabric and a wire was connected to the electrode.

4.4. Fabrication of different shapes, areas and sizes of electrodes

All those different shapes, areas and sizes of conductive fabric electrode (is made up of nickel and copper) were designed using the industrial 3D design software (Solidworks, 2021) to ensure precision, and the models of electrode were processed using a laser cutting machine on the conductive fabric. The mesh electrodes are made up of red copper with different grid numbers (10, 20, 30, 60, 80, 120, 150, and 200). The size and area of all square mesh electrodes are 4 cm × 4 cm, and 16 cm², respectively.

4.5. Characterization and measurements

The microstructure, morphology and elemental component of NC in this paper were obtained using a scanning electron microscope (SEM, Nova 200 NanoLab) equipped with a EDX spectrometer. The surface chemical analysis of the NC/Si RS was conducted by X-ray photoelectron spectroscopy (XPS) utilizing a PHI 5000 VersaProbe III instrument. The nanocrystals' surface topography was meticulously probed using both an atomic force microscope (AFM), specifically the Dimension iCon system by Bruker, and a three-dimensional laser confocal measuring microscope (LEXT, OLS4100, Olympus). Additionally, the optical transparency of the NC/Si RS, in comparison to air, was quantified employing a Shimadzu UV2600 ultraviolet-visible spectrometer. Moreover, the water contact angles of a variety of NC films were meticulously ascertained using an optical water drop angle tester, specifically the LSA100, of Chinese provenance. All those electrical output performances (V_oc, I_sc, and Q_sc) of TENGs were measured using an electrometer (6514, Keithley) and recorded using a data acquisition card (NI, USB-6009).

4.6. Ethics declarations

The experimental participants are healthy men. All experiments were performed in compliance with the relevant guidelines (Declaration of Helsinki). The authors declare that human participants involved in the experiments (Tong Zheng and Linnan Zhang) are well informed and they have provided the informed consent.

Data availability

The data that support the findings of this paper are available from the corresponding author upon reasonable request.

Author contributions

Tong Zheng: conceptualization, writing – original draft preparation, methodology, investigation. Linnan Zhang and Yong Lei: construction of test platform, Guizhong Li: supervision, validation, writing – reviewing, revising and editing.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This research was supported by Wenzhou Association for Science and Technology (Grant No. jczc156); Graduate Scientific Research Foundation of Wenzhou University (XKC230429).

References

1. Y. X. Shi, F. Wang, J. W. Tian, S. Y. Li, E. G. Fu, J. H. Nie, R. Lei, Y. F. Ding, X. Y. Chen and Z. L. Wang, Sci. Adv., 2021, 7, eabe2943.
2. L. Manjakkal, L. Yin, A. Nathan, J. S. Wang and R. Dahiya, Adv. Mater., 2021, 33, 2100899.
3. W. A. D. M. Jayathilaka, K. Qi, Y. Qin, A. Chinnappan, W. Serrano-García, C. Baskar, H. Wang, J. He, S. Cui, S. W. Thomas and S. Ramakrishna, Adv. Mater., 2019, 31, 1805921.
4. Y. Song, J. H. Min, Y. Yu, H. B. Wang, Y. R. Yang, H. X. Zhang and W. Gao, Sci. Adv., 2020, 6, eaay9842.
5. K. Dong, X. Peng and Z. L. Wang, Adv. Mater., 2020, 32, 1902549.
6. X. Peng, K. Dong, C. Y. Ye, Y. Jiang, S. Y. Zhai, R. W. Cheng, D. Liu, X. P. Gao, J. Wang and Z. L. Wang, Sci. Adv., 2020, 6, eaba9624.
7. X. C. Qu, Z. Liu, P. C. Tan, C. Wang, Y. Liu, H. Q. Feng, D. Luo, Z. Li and Z. L. Wang, Sci. Adv., 2022, 8, eabq2521.
8. H. C. Ates, P. Q. Nguyen, L. Gonzalez-Macia, E. Morales-Narváez, F. Güder, J. J. Collins and C. Dincer, Nat. Rev. Mater., 2022, 7, 887–907.
9. Z. Wang, Z. R. Liu, G. R. Zhao, Z. C. Zhang, X. Y. Zhao, X. Y. Wan, Y. L. Zhang, Z. L. Wang and L. L. Li, ACS Nano, 2022, 16, 1661–1670.
10. X. Wang, Z. P. Feng, Y. S. Xia, G. Q. Zhang, L. A. Wang, L. Chen, Y. F. Wu, J. Yang and Z. L. Wang, Nano Energy, 2022, 102, 107710.
11. Z. D. Sun, M. L. Zhu, X. C. Shan and C. K. Lee, Nat. Commun., 2022, 13, 5224.
12. B. Zhou, J. Z. Liu, X. Huang, X. Y. Qin, H. Shao, C. Y. Tang and X. X. Zhang, Nanomicro. Lett., 2023, 15, 72.
13. X. Y. Qiu, J. Z. Liu, B. Zhou and X. X. Zhang, Adv. Funct. Mater., 2023, 33, 2300321.
14. F. Wen, Z. X. Zhang, T. Y. He and C. Lee, Nat. Commun., 2021, 12, 5378.
15. X. Pu, W. G. Hu and Z. L. Wang, Small, 2018, 14, 1702817.
16. H. Li, T. R. Chang, Y. S. Gai, K. Liang, Y. L. Jiao, D. F. Li, X. R. Jiang, Y. Wang, X. C. Huang, H. Wu, Y. M. Liu, J. Li, Y. M. Bai, K. Geng, N. R. Zhang, H. Meng, D. S. Huang, Z. Li, X. G. Yu and L. Q. Chang, Nano Energy, 2022, 92, 106786.
17. H. Shen, H. Lei, M. W. Gu, S. Miao, Z. Q. Gao, X. H. Sun, L. N. Sun, G. Y. Chen, H. B. Huang, L. G. Chen and Z. Wen, Adv. Funct. Mater., 2022, 32, 2204525.
18. B. Lee, H. Cho, K. T. Park, J. S. Kim, M. Park, H. Kim, Y. Hong and S. Chung, Nat. Commun., 2020, 11, 1289.
19. B. D. Chen and Z. L. Wang, Small, 2022, 18, 2107034.
20. G. Z. Li, Y. W. Cai, G. G. Wang, N. Sun, F. Li, H. L. Zhou, X. N. Zhang, H. X. Zhao, Y. H. Wang, J. C. Han and Y. Yang, Nano Energy, 2022, 99, 107423.
21. W. Y. Zhang, Y. Xi, E. G. Wang, X. C. Qu, Y. Yang, Y. B. Fan, B. J. Shi and Z. Li, ACS Appl. Mater. Interfaces, 2022, 14, 20122–20131.
22. R. Cao, X. J. Pu, X. Y. Du, W. Yang, J. N. Wang, H. Y. Guo, S. Y. Zhao, Z. Q. Yuan, C. Zhang, C. J. Li and Z. L. Wang, ACS Nano, 2018, 12, 5190–5196.
23. J. H. Zhou, Y. F. Zhang, J. W. Wang, D. S. Zhang, X. R. Zhou and J. Q. Xiong, ACS Nano, 2023, 17, 17920–17930.
24. R. H. Tang, M. Y. Xie, M. Li, L. Cao, S. S. Feng, Z. D. Li and F. Xu, Appl. Mater. Today, 2022, 26, 101305.
25. Y. Jie, H. R. Zhu, X. Cao, Y. Zhang, N. Wang, L. Q. Zhang and Z. L. Wang, ACS Nano, 2016, 10, 10366–10372.
26. S. Chen, J. X. Jiang, F. Xu and S. Q. Gong, Nano Energy, 2019, 61, 69–77.
27. T. Zheng, G. Z. Li, L. N. Zhang, W. S. Sun, X. M. Pan, T. H. Chen, Y. H. Wang, Y. Q. Zhou, J. L. Tian and Y. Yang, Nano Energy, 2023, 114, 108649.
28. Q. Wang, M. F. Chen, W. Li, Z. Li, Y. T. Chen and Y. M. Zhai, Nano Energy, 2017, 41, 128–138.
29. Y. M. Ye, H. C. Wang, Y. Q. Tian, K. P. Gao, M. H. Wang, X. Q. Wang, Z. K. Liang, X. L. You, S. Gao, D. Shao and B. W. Ji, Nanotechnol. Precis. Eng., 2023, 6, 045001.
30. C. Zhao, Z. Y. Wang, Y. W. Wang, M. Y. Xu and Y. C. Lai, Adv. Funct. Mater., 2023, 2306381.
31. C. Y. Li, Z. J. Xu, S. X. Xu, Z. L. Wang and W. Tang, Nano Res., 2023, 16, 11846–11854.
32. Q. F. Shi, Z. D. Sun, X. H. Le, J. Xie and C. K. Lee, ACS Nano, 2023, 17, 4985–4998.
33. X. Chen, Y. J. Zhao, F. Y. Wang, D. Q. Tong, L. X. Gao, D. X. Li, L. K. Wu, X. J. Mu and Y. Yang, Adv. Sci., 2022, 9, 2103957.
34. C. Y. Ye, D. Liu, X. Peng, Y. Jiang, R. W. Cheng, C. A. Ning, F. F. Sheng, Y. H. Zhang, K. Dong and Z. L. Wang, ACS Nano, 2021, 15, 18172–18181.
35. S. S. Kim, Y. S. Kang, H. D. Lee, J. K. Kim and S. C. Hong, J. Air Waste Manage. Assoc., 2012, 62, 235–241.
36. K. G. U. R. Kumarasinghe, W. C. H. Silva, M. D. A. Fernando, L. Palliyaguru, P. S. Jayawardena, M. Shimomura, S. S. N. Fernando, T. D. C. P. Gunasekara and P. M. Jayaweera, BioMed Res. Int., 2021, 2021, 6666642.
37. L. X. Cheng, S. Jandhyala, G. Mordi, A. T. Lucero, J. Huang, A. Azcatl, R. Addou, R. M. Wallace, L. Colombo and J. Kim, ACS Appl. Mater. Interfaces, 2016, 8, 5002–5008.
38. Y. Luo, T. Y. Li, H. Z. Zhang, W. Liu, X. B. Zhang, J. W. Yan, H. M. Zhang and X. F. Li, Angew. Chem., Int. Ed., 2021, 60, 11718–11724.
39. M. Garza, J. Liu, N. P. Magtoto and J. A. Kelber, Appl. Surf. Sci., 2004, 222, 253–262.
40. E. Yegen, U. Zimmermann, W. E. S. Unger and T. Braun, Plasma Processes Polym., 2009, 6, 11–16.
41. M. J. Zhou, F. Xu, L. Y. Ma, Q. L. Luo, W. W. Ma, R. W. Wang, C. T. Lan, X. Pu and X. H. Qin, Nano Energy, 2022, 104, 107885.
42. L. Y. Lan, J. Q. Xiong, D. C. Gao, Y. Li, J. Chen, J. Lv, J. F. Ping, Y. B. Ying and P. S. Lee, ACS Nano, 2021, 15, 5307–5315.
43. B. B. Fan, G. X. Liu, X. P. Fu, Z. Z. Wang, Z. Zhang and C. Zhang, Chem. Eng. J., 2022, 446, 137263.
44. Y. Li, S. Xiao, X. X. Zhang, P. Jia, S. S. Tian, C. Pan, F. P. Zeng, D. C. Chen, Y. Y. Chen, J. Tang and J. Xiong, Nano Energy, 2022, 98, 107347.
45. S. Chao, Y. M. Li, R. Zhao, L. Zhang, Y. Z. Li, C. Wang and X. Li, J. Drug Delivery Sci. Technol., 2018, 44, 440–447.
46. Y. Qiao, Q. Zhang, Y. Xiang, Z. Wang and X. Hu, J. Mater. Chem. A, 2023, 11, 20120–20128.
47. H. Y. Zou, Y. Zhang, L. T. Guo, P. H. Wang, X. He, G. Z. Dai, H. W. Zheng, C. Y. Chen, A. C. Wang, C. Xu and Z. L. Wang, Nat. Commun., 2019, 10, 1427.
48. Y. Y. Wang, X. Zhang, X. Y. Jin and W. J. Liu, Nano Energy, 2023, 105, 108021.
49. Y. L. Xu, Z. Q. Bai and G. B. Xu, Nano Energy, 2023, 108, 108224.
50. Y. H. Wu, Y. Luo, J. K. Qu, W. A. Daoud and T. Qi, Nano Energy, 2020, 75, 105027.
51. F. Yuan, S. Liu, J. Y. Zhou, S. Wang, Y. Wang, S. H. Xuan and X. L. Gong, Nano Energy, 2021, 86, 106071.
52. K. Ghosh, C. Iffelsberger, M. Konečný, J. Vyskočil, J. Michalička and M. Pumera, Adv. Energy Mater., 2023, 13, 2203476.
53. Y. Luo, Z. H. Wang, J. Y. Wang, X. Xiao, Q. Li, W. B. Ding and H. Y. Fu, Nano Energy, 2021, 89, 106330.
54. J. H. Park, C. Wu, S. Sung and T. W. Kim, Nano Energy, 2019, 57, 872–878.
55. C. L. Lee, M. Heo, H. Park, H. Joo, W. Seung and J. H. Lee, Nano Energy, 2022, 103, 107834.
56. Y. Xiong, L. Luo, J. H. Yang, J. Han, Y. Liu, H. S. Jiao, S. S. Wu, L. Q. Cheng, Z. Y. Feng, J. Sun, Z. L. Wang and Q. J. Sun, Nano Energy, 2023, 107, 108137.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ta05954e

---