Disposable and flexible smart electronic tapes for long-term biopotential monitoring

Introduction

Physiological electricity is of great importance to a large amount of physiological activities for human body. Monitoring the electrical signal of physiological activities through a wearable system has various applications in disease prevention, health monitoring, and human–computer interaction^1,2,3,4,5. As an essential component of the wearable system, electrodes with high stability and durability are critical for a better user experience. At present, electrodes for bioelectric detection can be divided into four categories, i.e., wet, semi-dry, dry, and flexible electrodes. As a representative wet electrode, silver-silver chloride (Ag/AgCl) electrodes have been widely utilized in clinical medicine, which typically needs conductive gel and adhesive. Because of the possible irritation of skin as well as the detachment, the wet electrode is rarely employed for long-term monitoring of bioelectrical signals^6,7. On the contrary, the traditional dry electrodes do not require conductive liquid or adhesive, which is relatively convenient to wear. The major issue is that noise generated from the movement artifacts of such electrodes is relatively large^8,9. In addition, microneedles or finger structures are typically required to reduce the contact impedance via increasing the contact area between electrodes and skin. Moreover, long-term wearing is also unsuitable due to infections and other skin symptoms. Semi-dry electrode possesses the advantages of both wet electrodes and dry electrode. However, a conductive liquid still needed to be coated between the electrodes and skin^10,11.

In recent years, flexible electrodes for detecting physiological electricity have been developed due to their good biocompatibility, high comfortability, and long-term stability^{12,13,14,15,16,17,18,19}. Compared with the aforementioned electrodes, flexible electrodes cause fewer skin irritations and have a more stable bonding interface with the skin, resulting in a more convenient and comfortable wearing^{1,20,21,22,23}. Ultra-thin-film electrodes are favored by researchers for better contact with skin to enhance signal quality^{24,25,26,27,28}. Several examples include ultra-thin nano-network structured electrodes with excellent air permeability^{29,30,31,32,33,34}, serpentine wire structure with good stretchability^35,36, and customizable patterned graphene electrodes^37,38. These devices are thin and conformable with good contact with skin. However, the fabrication processes are typically complicated and high cost. In addition, viscous polymers are often used to glue the films with skin, which is unable to provide comfortability for continuous collection of signals^13,39,40,41. In addition to ultra-thin films, liquid metals have also been widely used as physiological electrical electrodes due to their excellent plasticity as well as the seamless contact with skin^42,43,44,45. The main problem for the liquid metal-based electrode is its complex operation. In addition, its biocompatibility is uncertain, making it unsuitable for long-term wear. Thus, it is still challenging to develop self-adhesive, ultra-stable and durable, and low-cost electrodes with the capability of seamless contact with the skin for long-term monitoring of physiological electricity.

Herein, we report a series of smart electronic tapes (SET) comprising carbon nanotubes (CNTs) electrode, which take advantages of an optimal second-order serpentine design on biocompatible and chemically inert substrate. Superior stretchability, high stability and durability, and excellent breathability are achieved simultaneously. In addition, cytotoxicity test has been conducted to evaluate the safety, demonstrating a harmless behavior to organ cells. Moreover, SETs do not require any extra gels, adhesives, and polymers, which enabled firm adhesion to skin without any delamination or detachment after wearing for a long time (7 days). Compared with the traditional Ag/AgCl gel electrode and other electrodes, SET has realized long-term and effective monitoring of physiological signals under various human conditions with a low noise level. Further, we design a wearable virtual reality (VR) gesture interaction system utilizing the SETs to collect arm’s muscle electrical signals to control the speed and direction of a virtual vehicle in real-time.

Results

Fabrication of smart electronic tapes

As shown in Fig. 1a, the shape of smart electronic tape looks like Scotch tape, which could be rolled into a roll. Fig. 1b presents the SETs-based wearable and flexible physiological electrical monitoring system, which has been used to record electrocardiogram (ECG) signal for all-day health monitoring, and collect electromyography (EMG) signal for real-time VR interaction. The schematic of the fabrication process of SET is displayed in Supplementary Fig. 1. Firstly, laser-based direct-writing technique was used to etch an aluminum plate with the optimal pattern, followed by a thorough cleaning with deionized water and ethanol. Next, the prepared mold is filled by a certain amount of a mixture of CNT and polydimethylsiloxane (PDMS) to form conductive patterns after curing process (see Methods). Then, a solution of target flexible substrate material is spin-coated on the cured PDMS/CNT network. Finally, the PDMS/CNT network is transferred to target substrate after peeling off the electrode. Utilizing this process, we can transfer the electrodes to various flexible substrates for different potential applications (Fig. 1c and Supplementary Fig. 2). For example, the electrodes possess high-temperature resistance after transferring to polyimide (PI) tape, while degradable electrodes are realized on paper tape. When the electrode is transferred to the ultra-thin PU substrate, it can easily fit perfectly with the fingerprint (Fig. 1d). Fig. 1e illustrates the working process of each module of the monitoring system. The flexible monitoring system has three modules, including a sensing module, a signal acquisition and processing module, and an application module. The workflow of the flexible health monitoring system is as follows: Firstly, the smart electronic tape is applied to the corresponding part of the human skin. Secondly, the flexible signal acquisition module is used to collect the physiological electrical signals by the intelligent electronic tape, and are processed in combination with the artificial intelligence algorithm. Finally, it is used for human health monitoring, virtual reality interaction, sleep state monitoring and motion state monitoring. The soft, flexible tape allows them to be placed on a variety of target locations on the body (Fig. 1f). SETs with different substrates expand their application scenarios, and are expected to be useful for people with rough skin, skin folds, damaged skin, and dry skin.

Disposable and flexible smart electronic tapes for long-term biopotential monitoring — **Fig. 1: Image and schematic diagram of smart electronic tapes.**

We also evaluate the adhesion force of the prepared electrodes on different substrates as the adhesion is a key to reducing noise signals in motion artifacts. As displayed in Fig. 2a, the electrodes prepared by our method all showed strong adhesion, among which, the electrodes on polyurethane (PU) tape, polypropylene (PP) tape, and PI tape showed stronger adhesion compared with the other four tapes. We tested the water permeability of some representative tapes because breathability is essential to improving the comfort performance of wearable devices (Fig. 2b). Our aim is to provide a universal method for electrode preparation on different substrates to meet various usage scenarios according to the properties of the substrate materials. Therefore, we did not conduct performance characterizations on all types of tapes. Instead, we selected a relatively typical tape for detailed study. The PU tape was selected as the substrate for following experiments due to the large adhesive force, high breathability (159 g m⁻² day⁻¹) and stretchability. The adhesive force of the PU tape on the skin has also been investigated (Fig. 2c), exhibiting a value of 0.55 ± 0.019 N/cm. The strong stickiness results in a firm contact on skin, which is beneficial for long-term wearing. To test the adhesion of PU-SET on different types of skin, we respectively attached PU-SET to neutral skin, oily skin, and dry skin. As shown in Supplementary Fig. 3, the adhesion of PU-SET on neutral skin is the strongest (0.55 ± 0.019 N/cm), followed by that on dry skin (0.45 ± 0.095 N/cm), and the weakest on oily skin (0.31 ± 0.037 N/cm). Due to the fact that the stratum corneum on dry skin is prone to peeling off, the test results exhibit relatively high fluctuations. Overall, there are certain differences in the adhesion of PU-SET on different types of skin, but it can all meet the requirements of daily use. In addition, our SETs can be made with multiple substrates. Therefore, when dealing with different types of skin, we can select a more suitable substrate to maximize the adhesion of SETs.

**Fig. 2: Performance characterization of smart electronic tapes.**

Compared with traditional metals or other rigid conductive materials, PDMS/CNT mixtures have better stretchability and chemical inertness. However, the electrical resistance of the PDMS/CNT mixture is still susceptible to strain. In addition, the poor air and sweat permeability of PDMS/CNT film makes it difficult for long-term and comfortable wearing on skin. Therefore, we have designed various PDMS/CNT patterns to reduce the strain-induced disturbance and improve the permeability of the electrode. Fig. 2d displays four different electrode patterns including second-order serpentine line, linear line, low-density serpentine line, and high-density serpentine line. Finite element method (FEM) has been used to simulate and analyze the strain distribution of the designed patterns on PU substrate under tensile strains of 33%, 66%, and 100% (Supplementary Fig. 4). It can be seen from the simulation results that under the same stretching conditions, compared with the linear pattern and the serpentine pattern, the second-order serpentine pattern has a multi-order curve structure, making it difficult for the applied stress to concentrate. This also means that the second-order serpentine pattern exhibits more stable performance under the same strain, which can greatly reduce the low-frequency baseline noise when the device is deformed. To verify the simulation results, we have prepared the corresponding electrode patterns (right panel in Fig. 2d) and investigated their resistance changes under 100% tension. As displayed in Fig. 2e, the linear electrode has the worst tensile performance and the largest resistance change rate about 160% under 100% tension, followed by the serpentine line types. In contrast, the resistance of the second-order serpentine pattern under strain is more stable, exhibiting a change in resistance of less than 40% under 100% tension. These results are consistent with the FEM simulation results, which verify the stability of the second-order serpentine pattern (Fig. 2e). It should be noted that the mechanical stability and the electrical conductivity of the electrodes can be further improved by increasing the CNT content in the PDMS/CNT composite. Specifically, the relative resistance change of the electrodes with 13% CNT is only 40% compared with that of 7% CNT under 100% strain (Fig. 2f). The second-order serpentine pattern in our design can be stretched to more than 300% (Supplementary Fig. 5), and human skin is soft and stretchable with a maximum strain of 180%^46,47, which meet most requirements of daily activities. In addition to the highly stable behavior, the prepared electrode also demonstrates excellent long-term durability. The resistance change rate is less than 15% after 45,000 stretching/releasing cycles under a strain of 50%. (Fig. 2g and Supplementary Fig. 6). We further prepare a CNT-based circuit with a commercial light-emitting display (LED) patch mounted on the pattern. As shown in Fig. 2h, the LED can maintain its high illumination even after being stretched to 100% (Supplementary Movie 1). In daily life, human skin is in a complex and changeable environment. In order to verify the performance of the second-order serpentine wire electrode, we tested the electrode-skin contact impedance under different skin conditions. The electrode was applied to the volunteer’s skin, and the electrode-skin contact impedance was tested using an impedance analyzer under normal, squeezed and stretched states, respectively. Fig. 2i shows that the resistance-skin contact impedance of second-order serpentine wire electrode were identical the same under different skin conditions, which further verifies the electrode can be used for signal monitoring in complex skin environments.

Temperature is also one of the important factors that affect the normal use of the equipment. As shown in Supplementary Fig. 7, considering the actual usage scenarios of SETs, we monitored the electrical properties of SETs within the range of 20–50 °C. The experimental results show that when the temperature rises from 20 °C to 50 °C, the resistance change rate of SET is lower than 3%, demonstrating relatively stable performance. In addition, within the common temperature range of the human body (35–40 °C), the resistance change rate of SETs is ~0.33%, having almost no impact on the performance of SETs.

Biopotential detection using SETs

The excellent stability and durability make the SETs to be promising for various applications. For example, we attached the electrode on skin to record ECG signal for 5 days. This is of great significance because the long-term monitoring ECG signals is favorable for early detection of heart diseases (i.e., ischemic disease, stroke, cardiac arrest, and abnormal heart rhythm), which will help reduce the mortality of heart disease patients. As displayed in Fig. 3a, a smart conductive tape on PU base (PU-SET) with a side length of about 2 cm is attached to arm and chest. A commercial Ag/AgCl gel electrode has also been adhered for comparison. It is obvious that PU-SETs needs a smaller area for the attachment to skin, which has less impact on the skin, easier to wear and more comfortable. In addition, PU-SETs are in lighter weight and higher comfort compared with commercial gel electrodes (Fig. 3b). The accumulation of sweat after long-term wearing will affect comfort as well as signal quality. The breathability experiment indicates that the PU-SET has a water permeability of more than 150 g m⁻² day⁻¹ (Fig. 2b). The electrodes have reasonable contact with the skin, which ensures the collection of high-quality ECG signals. The ECG signals collected by the PU-SET leads on the chest and arms can clearly identify the characteristic waveforms of electrocardiogram (i.e., P, Q, R, S, and T), which is better than that of commercial gel electrodes (Fig. 3c). We further calculate the signal-to-noise ratio (SNR) of the ECG signals for both electrodes. In the static state, the SNR of PU-SET (23.5 ± 0.45 dB) on the arm lead is slightly higher than that of the standard Ag/AgCl gel electrodes (22.65 ± 0.5 dB), which indicates that the performance of PU-SET is at least as good as the commercial gel electrodes (Fig. 3d). The skin with gel electrodes showed severe redness and swelling after ten minutes of wearing. In contrast, PU-SETs have negligible influence on skin even after long-term wearing for seven days (Fig. 3e). This is ascribed to the high biocompability of both CNT and PDMS. To verify the safety of PU-SET on skin, L929 cells have been used for cytotoxicity experiments. As displayed in Fig. 3f, a regular cell morphology has been observed from fluorescent live staining images with PU-SET and the control. Further, the absorption at 450 nm in the CCK8 assay from day 1 to day 3 has increased in a similar trend for PU-SET and the control (Fig. 3g). Since the absorption is proportional to the cell numbers, we can conclude that there has no obvious toxicity of the PU-SET, which ensures a safe wearing for a long time.

**Fig. 3: ECG signal detection and cytotoxicity verification of carbon-based tape.**

In addition to the biocompability, we also investigate the reliability of the PU-SETs for long-term monitoring of the ECG signals. It is observed from Fig. 3h that the PU-SETs can still collect ECG signals with high quality even after regular wearing for five consecutive days. The signal-to-noise ratios of PU-SET after being worn for one (22.2 ± 0.42 dB), three (21.5 ± 0.35 dB), and five (19.8 ± 0.48 dB) days are shown in Fig. 3i. After five days of normal wearing, the performance of the PU-SET decreased by approximately 10%, and it was still able to collect ECG signals with high quality. The experimental results have verified that PU-SETs are highly stable and reliable for long-term wearing.

During daily activities, human skin is prone to different degrees of deformations, such as bending, stretching, and compressing. To verify the performance of the PU-SET during human activities, we investigate the contact impedance between electrode and skin, and ECG signals when the skin is compressed and stretched (Supplementary Movie 2). The experimental results indicate that the impedance is retained without obvious change under deformations (Fig. 2i). The ECG waveforms have also been clearly identified during stretching and compressing conditions, which was further verified by PU-SET that can achieve ECG signal monitoring under complex skin conditions (Fig. 3j).

As mentioned, PU-SET exhibits excellent accuracy, stability, durability, and biocompatibility, which is favorable for real-time and long-time monitoring ECG signals. We evaluate the ability of all-day biopotential monitoring by recording the ECG signals of volunteers in various daily activities, such as exercise, relax, and sleep. High-quality ECG signals have been obtained in different states, along with well-distinguished characteristic waveforms. Even in a dynamic state, the signal-to-noise ratio of PU-SET can reach 17.1 ± 1.43 dB, and it is still able to achieve the monitoring of electrocardiogram signals (Fig. 3k). In addition, the heart rates of volunteers can also be calculated by counting the number of R waves in ECG signals. The initial heart rate in relaxing state is about 88 min⁻¹, while the values are changed to 65 and 131 during sleeping and after exercise, respectively (Fig. 3k, l), which is consistent with the actual values. PU-SETs can also be used in other human state monitoring. Snoring is quite common in humans, which significantly affects the quality of sleep and physical health. We have conducted ECG monitoring on the volunteers during normal sleep and snoring sleep. It is found that the ECG signal of snoring sleep has irregular fluctuations compared with normal sleep (Supplementary Fig. 8). All the results have demonstrated that PU-SET is capable of monitoring all-day activities.

In addition to ECG monitoring, PU-SET can also record other biopotentials. EMG is an electrophysiological phenomenon generated by muscle activity. By analyzing EMG signals, we can evaluate the muscle state (stretching/contraction) and related diseases (i.e., muscle atrophy, Parkinson’s, muscle rigidity)⁴⁸. Recently, EMG signals have also been employed in the fields of controlling prosthetic movements, rehabilitation training, and human-machine interaction^36,48,49. We have attached PU-SET to the forearm and monitored EMG signals through a flexible wearable monitoring system (Fig. 4a, b). Wherein EMG response is generated via the movement of four kinds of muscles, including flexor carpi ulnaris, superficial finger flexor, deep finger flexor, and flexor pollicis longus (Fig. 4c). As expected, EMG signals can be detected when fingers are under bending conditions. More importantly, different postures can also be accurately recognized by monitoring the change in EMG potential. For example, the bending of ring and little finger results in a much larger EMG potential than that of middle finger when the tape is placed on flexor carpi ulnaris (Fig. 4d, e). This is because that ulnar nerve is mainly distributed in the ring finger and little finger, which controls the muscle of flexor carpi ulnaris. Therefore, no obvious EMG signal is observed in the flexor carpi ulnaris when middle finger is bent. In contrast, there are strong EMG signals in channel one (the muscle of flexor pollicis longus) when bending the thumb and middle fingers (Fig. 4d, e). This dual-lead method can accurately recognize the bending of different fingers, while the traditional single-lead can only detect the movement of fingers. However, the EMG signals generated during finger movements are relatively weak. Therefore, we can only detect whether each finger is bent or not, and have not yet identified the bending angles of each finger. Compared with the commercial gel electrodes, PU-SET exhibits lower noise to detect the grip strength (Supplementary Fig. 9). Moreover, different grip strength from 15 to 25 kg have also been accurately identified by monitoring the EMG potential from the tape (Supplementary Fig. 9). In addition to the monitoring of finger bending, we have also placed the PU-SET on the front arm to conduct grip strength test (Supplementary Fig. 10).

**Fig. 4: Virtual reality interactive system based on EMG signals.**

We have designed an EMG virtual reality interactive system (Fig. 4f), which contains various units, including data acquisition, data processing, and wireless data transmission. The working process can be divided into three steps: 1) the originally acquired signal is filtered and amplified by an AD8236 chip, followed by 2) the conversion into digital domain by the built-in Analog Digital Converter (ADC), and finally 3) wirelessly transferred to a smartphone or personal computer (PC) through the built-in WIFI module for further analysis and processing (Supplementary Fig. 11). The dataset includes five gesture categories: right-hand swing right, right-hand swing left, right-hand clenched fist, right five fingers open, and natural resting state (Fig. 4g). For each gesture, 130 pieces of EMG data are collected, while 100 pieces are randomly selected as the training set and the remaining 30 pieces are the test set. Therefore, the dataset contains 500 pieces of training data and 150 pieces of testing data. We have designed an artificial intelligence algorithm for gesture recognition and VR interaction, which includes five convolutional layers and two fully connected layers (Fig. 4f and Supplementary Fig. 12). The size of the input EMG picture is 2 × 1000. After each convolution layer, the feature map corresponding to the number of convolution kernels will be outputted. Then, the distributed feature learned by the network will be mapped to the sample label space through the fully connected layer. Finally, the gesture category will be predicted. To train the network for faster convergence, we have used stochastic gradient descent with momentum. The training parameters are set as follows: Nesterov momentum of 0.9, weight decay of 0.0001, the initial learning rate of 0.001, and the learning rate of 0.1 times of the original at the 15th, 35th, 50th, and 65th epochs. Given the optimal parameters, the average accuracy of the five gestures is about 96.7%, wherein thumb, ring, and little finger have a recognition rate of 100% (Fig. 4h). As another example of potential application of the disposable tape, we develop a VR system with gesture interaction. The Unity engine has been used to build a 3D dynamic scene, in which the movement of a car is controlled by the defined gesture commands (Fig. 4i). For example, the car will turn left and right under the gestures of swinging the right hand to left and right, respectively. Similarly, a fist of right hand can accelerate the speed car and the car will be braked if the five fingers are open (Fig. 4j and Supplementary Movie 3–6). In the actual measurement, the average delay time from signal acquisition to real-time transmission to PC via wireless network is less than10 ms, while, the time is only 6.6 ms to provide gesture commands through the neural network (Supplementary Fig. 13). The negligible delay time (16.6 ms) makes the VR system interaction practically real-time.

Discussion

Here, we develop a SET to acquire various biopotentials. An ultra-stretchable CNT-based electrode with optimal second-order serpentine pattern has been easily manufactured and transferred to various kinds of tapes. The SETs demonstrate superior stretchability (more than 300%), stability (less than 40% change in resistance under 100% strain), durability (over 45,000 cycles of stretching/releasing), water permeability (150 g m⁻² day⁻¹), and biocompatibility. Despite the long-term and comfortable wearing, both ECG and EMG signals have been successfully acquired by placing the SETs on human arm with comparable accuracy and efficiency to commercial gel electrodes. In addition, highly accurate and real-time recognition of finger bending and controlling of a vehicle in VR system have also been realized through detecting the EMG signals with negligible delay time. More importantly, the SETs are completely disposable, opening up a novel avenue for real-time, long-term, and low-cost biopotential acquisition.

We have compared various properties of different types of electrodes, and summarized the values in Supplementary Table 1. As shown in Supplementary Table 1, the SETs exhibit lower contact impedance, higher signal-to-noise ratio and higher peak-to-peak value, which enables our devices to have the ability to collect high-quality bioelectric signals. In addition, the SETs also show higher adhesion and stable electrical properties, which further ensures the applicability of our devices in complex environments. However, the patterning of the conductive material further reduces its effective contact area with the skin. In addition, the SET still has difficulty working effectively in skin areas with more hair.

In the future, we could explore and develop flexible substrates that are temperature-insensitive and have excellent directional sweating capabilities to address the negative effects caused by sweat and extreme temperatures. Additionally, we could investigate the integrated design and combination of the wireless signal transmission module and the long-distance wireless energy transfer unit with the smart electronic tape, which would further improve the portability and continuity of wearing the smart electronic tape. Regarding environmental protection, we can either choose or develop new degradable base materials and optimize product design to allow the reuse of key components, extending the service life and reducing waste. At the same time, we need to establish and perfect a recycling mechanism to alleviate the environmental impact of the disposable design.

Methods

Mold preparation

The patterns are designed using AutoCAD software. A laser direct writing equipment has been employed to etch the aluminum plate using the designed pattern. A smooth aluminum plate is chosen as the substrate of the mold, and various designed patterns are etched on the aluminum substrate by the laser etching equipment to form second-order serpentine lines with a pattern width of 200 ± 20 μm and a depth of about 30 μm. The laser-etched template was then rinsed with deionized water and dried at 80 ^oC for 2 h. Finally, to ensure that the release agent can completely cover the surface of the mold, continuous spraying of the fluorine release agent (DR-518) is carried out for 5 to 6 s at a distance of 20–30 cm from the surface of the mold. Then, it is heated in an oven at 100 °C for 2 h, and the above operation is repeated three times to complete the preparation of the mold.

Dry electrode printing

PDMS (purchased from Aladdin) solution was mixed with CNTs (13 wt%, XFNANO, INC) and stirred for 30 min at room temperature to form a well-dispersed PDMS/CNT mixture. Subsequently, the mixture was filled into the mold using the squeegee printing method. Then, various pastes were evenly printed on the mixture and dried at 80 °C for 2 h. Finally, the resultant tape was peeled off after cooling down. Conductive carbon oil is used to connect the wires and smart electronic tape, and the interface is encapsulated with ultra-thin PU tape.

Biocompatibility test

L-929 cells, purchased from Shanghai cell bank of Chinese Academy of Sciences, were used to examine the in vitro cytotoxicity of the tape and control group according to the ISO 10993-5: 2009 (E) standard. In brief, 8 × 10⁴ L-929 cells were seeded in a 24-well plate with 1 mm medium, which contains High Glucose Dulbecco’s Modified Eagle’s Medium (DMEM, Hyclone, SH30243.01) supplemented by 10% fetal bovine serum (FBS, Gibco, 10099141) and 1% penicillin/streptomycin (PS, Hyclone, SV30010). 0.2 cm² sterile PU-SET were put in the plate indicated as the experiment groups and the group without any material served as control. All the groups were cultured at 37 °C in an incubator with 5% CO₂. The cell confluency and morphology were investigated with an inverted microscope, and the medium was changed 24 h after the cell-seeded. After 24 h of incaution, CCK-8 and live/dead assays were arranged to quantify the cell proliferation. The CCK8 (Dojindo, PL701) assay was diluted by 10 times the volume of the medium, and 500 µl diluted CCK8 was added to each well substituting medium. After 2 h incubation, 200 µl supernatants were collected into the 96-well plate. The absorption at 450 nm was detected using an absorption spectroscopy. The live staining kit (Beyotime, C2013M) was used to observe the live cell rates of different groups according to the manufacturer’s protocol. Briefly, 250 µl diluted calcein-AM was added to each group. After 30 min incubation and rinsingd with PBS, the samples were observed and recorded using a fluorescent microscope. The image information was processed by ImageJ software.

Device test

The tensile and mechanical tests of the device are completed by the test system we built. The system is composed of a source meter (Keithley 2400, Tektronix) for real-timely recording the change of electrical properties, and a high-precision stepper motor (MTS316, BOCIC) with clamp for stretch. The force gauge (HANDPI HP-5) was used to detect the real-time force change that was applied to materials and devices. The accuracy of the moving displacement could reach 2.5 µm. When conducting the adhesion test, loading-displacement curves were conducted with a tensile mold at a peeling test speed of 2.5 mm s⁻¹. When testing the electrical properties of the SET under different strain and 45,000 cyclic tensile experiments, the moving speed of the stepping motor was 25 mm s⁻¹. The physiological electrical signal test (ECG and EMG) is completed by our own flexible intelligent monitoring system and source meter, the sampling frequency is 2000 Hz.

Permeability test

First, cut different types of tapes of the same size. Then, select transparent glass bottles of the same size and add an equal amount of deionized water into each bottle. Next, seal the mouths of the glass bottles with the tapes and fix the surroundings with sealing films to prevent air leakage. After that, mark each glass bottle clearly and weigh it with a high-precision balance to obtain the initial weight. Finally, weigh the samples at regular intervals and calculate the corresponding water permeability by using the formula 1.

$$q=Delta m/left({A}_{0}tright)$$

(1)

Among them, ({Delta{m}}) represents the weight of water permeated through the device, ({A}_{0}) represents the area of the water-permeable region of the device, and (t) represents the time of water permeation of the device.