A double-crack structure for bionic wearable strain sensors with ultra-high sensitivity and a wide sensing range†

Di Zhu‡ ^a, Shengshun Duan‡ ^a, Jiachen Liu ^a, Shanyan Diao ^a, Jianlong Hong ^a, Shengxin Xiang ^a, Xiao Wei ^a, Peng Xiao ^b, Jun Xia ^a, Wei Lei ^a, Baoping Wang ^a, Qiongfeng Shi ^a and Jun Wu *^a
aJoint International Research Laboratory of Information Display and Visualization, School of Electronic Science and Engineering, Southeast University, 210096, China. E-mail: wujunseu@seu.edu.cn
^bState Grid Jiangsu Electric Power Co., Ltd, Research Institute, Nanjing, 211103, Jiangsu, P. R. China. E-mail: vodoco@foxmail.com

First published on 7th February 2024

---

---

1. Introduction

With the rapid development of electronic technologies, considerable attention has been paid to flexible sensors^1–3 because their emergence has significantly extended the forms of electronic devices.^4,5 Compared with traditional rigid sensors,^6,7 this kind of sensor shows excellent flexibility and ductility^8,9 so that they can be bent or folded freely. Furthermore, people can arbitrarily arrange them according to the requirements of measurement conditions due to the various structures of flexible sensors. At present, flexible strain sensors are extensively used in the development of electronic skin,^10–13 healthcare,^14,15 voice recognition,^16,17 sports equipment,^18,19 triboelectric nanogenerators,²⁰ textiles,^21,22 aerospace,²³ environmental monitoring^24,25 and other fields.^26,27 Generally, the sensing performance of flexible strain sensors can be evaluated using various performance parameters such as sensitivity [gauge factor (GF)], stretchability (sensing range), linearity, hysteresis, response time, and dynamic stability.^28–31 Among them, sensitivity and working range are the two most critical parameters for flexible strain sensors. To comprehensively detect human motion and physiological conditions,^32,33 strain sensors must have high sensitivity in a wide sensing range. Yet, it is often difficult to achieve both high sensitivity and a wide sensing range and it is also hard to maintain ultra-high sensitivity under small strain.

To improve the performance of flexible strain sensors, the existing methods can be divided into two categories. One is based on the material design or doping to improve the material performance, including carbon materials^34–37 (such as carbon nanotubes and graphene), metal materials,^38,39 and semiconductor oxides.⁴⁰ In addition to considering innovations in materials, people often make use of different device structures to obtain better performance, involving porous structures,⁴¹ hollow micro-structures,⁴² and fragmented micro-structures.⁴³ Structural engineering is an effective method for enhancing the performance of strain sensors. The performance of sensors varies with the structure of different geometric shapes because of the varying force deformation behaviors.

To balance sensitivity and the sensing range, researchers have conducted many studies. For example, Han et al.⁴⁴ designed a composite structure strain sensor based on nanomaterials and a conductive liquid that responds to multi-scale strains from 4% to over 400%, with high sensitivity and durability under small strain. Ji and co-workers⁴⁵ proposed a strategy to construct flexible strain sensors based on the V-groove/wrinkles hierarchical array. Compared with the sensors based on single-scale wrinkle structures, the sensors with the hierarchical array showed high sensitivity (GF ∼ 2557.71) and a wide sensing range (up to 45%). A cost-effective strain sensor with ultrahigh strain sensitivity under small strains (ε) was developed by Wu and his team,⁴⁶ which is based on the sub-micron to nano-scale voided clusters in Cu–Al alloy films. The sensor exhibits ultrahigh gauge factors as high as 584 (0% < ε < 0.5%), 10219 (0.5% < ε < 0.9%), and 43152 (0.9% < ε < 1.75%). Wang et al.⁴⁷ designed a breathable flexible strain sensor with a double-layered conductive network structure that exhibits a high sensitivity (up to 1477.7) and a very wide working range (up to 150%). Although previous research effectively enhanced the performance of strain sensors, it is obvious that sensors with ultra-high sensitivity have a narrow sensing range and sensors with a very wide sensing range have low sensitivity. Therefore, developing a versatile and straightforward preparation strategy for strain sensors with ultra-high sensitivity and a wide sensing range remains a challenge.

Bionics⁴⁸ means to mimic the functions and behaviours of biological systems and it often provides perfect ideas for people to solve technical problems. Arthropods such as spiders or scorpions in nature sensitively perceive small stimuli through the crack-sensing organs in the feet (lyriform organs). When cracks are created or healed, information is fed back to the brain through connected nerves, so the brain can respond rapidly to tiny changes. The operating principle of lyriform organs demonstrates that constructing surface cracks can improve the sensitivity of flexible sensors under a small strain.

Inspired by the structure of lyriform organs, we proposed a novel strategy to construct a flexible strain sensor with high sensitivity in a wide sensing range based on the Au–PDMS–MXene composites (APMCs) with a double-crack structure. The MXene and Au layers with cracks represent different resistances, forming a parallel circuit that can rapidly switch the resistance value. When a strain occurs, the rapid disconnection of the crack junctions on the Au layer causes a significant variation in the resistance value so that an obvious and rapid response can be achieved. The cracks improve the sensor's sensitivity, while the double-layer structure and the parallel circuit ensure its wide sensing range. Finally, a flexible strain sensor with an ultra-high sensitivity of 14373.6, a wide working range (up to 21%), a fast response time (183 ms), and excellent dynamical stability (almost no performance loss after 1000 stretching cycles and different frequency cycles) was obtained. In practical applications, the sensor was applied to different parts of the human body to sense the deformation of the skin, demonstrating great potential application value in human physiological detection, voice recognition, human–machine interactions, and sports equipment. This study could provide new ideas for preparing high-performance flexible strain sensors.

2. Experimental section

2.1. Materials

The PDMS prepolymer (Sylgard 184) and curing agent were bought from Dow Corning (Shanghai, China). MXene aqueous dispersion (5 mg mL⁻¹) was obtained from 11 Technology Co., Ltd. The Au target for sputtering was acquired from Shunkai Metal Materials Co., Ltd. Ethanol (C₂H₅OH, AR) and n-hexane (C₆H₁₄, AR) were supplied by Sinopharm Chemical Reagent Co., Ltd. Ultra-pure water (>18 MΩ cm⁻¹) was used for all experiments.

2.2. Fabrication of PDMS–MXene composites (PMCs)

49 mL ultra-pure water was added to 1 mL MXene aqueous dispersion (5 mg mL⁻¹) for dilution. The diluted MXene aqueous dispersion was added into a suction filter device for suction filtration. Therefore, the filter with MXene sheets attached was gained. Then, 3.3 g PDMS (prepolymer:curing agent = 10:1, w/w) mixed with 0.1 mL n-hexane was spin-coated on the filter-attached glass which was cleaned with ethanol at 200 rpm for 60 s, followed by heat curing at 85 °C for 2 h. Finally, PMCs could be obtained after removing the filter membrane.

2.3. Preparation of APMCs with a double-crack structure

Several micron-scale cracks (10–30 μm wide) were built on both sides of the PMCs through an engraving machine. A 200 nm-thick Au layer was formed on one side of the PMC which had no MXene sheets attached by sputtering the Au target. Therefore, there were hollow micron-scale cracks on the MXene layer and micron-scale cracks filled with conductive Au particles in the Au layer. Finally, APMCs with double-crack structures could be prepared.

2.4. Fabrication of the flexible strain sensor based on APMCs

In the process of fabrication of the flexible strain sensor based on APMCs with a double-crack structure, two pieces of copper foils were fixed on two sides of the APMCs by silver paste as electrodes, which connected the MXene layer and the Au layer. After leading the wires from the copper foil at both ends, the flexible strain sensor was fabricated.

2.5. Characterization

The morphologies and microstructures of the MXene sheets, contact surfaces, and cracks were characterized using a high-speed digital microscope (VW-9000, Keyence, Japan) and a scanning electron microscope (Quanta 200, FEI, China) equipped with an energy-dispersive X-ray spectroscopy (EDS) detector for mapping. The electrical characterization of each flexible strain sensor based on APMCs was conducted with a digital source meter (Keithley 2400, Tektronix, America). The digital tensile tester (STC-10kgAL, Pubtester, China) was used to provide controlled deformation. All the measurements were performed under working ambient conditions. To ensure the accuracy of the test data, each sample was measured three times, and the value of each data point represents the average value. The sensitivity of stretchable strain sensors was calculated as

	(1)

where R, R₀, and ε denote the real-time resistance, initial resistance, and applied strain, respectively.

3. Results and discussion

Fig. 1 shows the schematic diagrams of our proposed flexible strain sensor based on APMCs with a double-crack structure. The preparation of APMCs based on PMCs was mainly divided into two steps. One step was to build micron-scale cracks on the surface, and the other one was to form a metal film on the surface. And the carving of cracks in the Au layer needs to be completed before steaming gold, so a layer of Au will also be covered in micron-scale cracks (as shown in Fig. S1†). The gold in the cracks forms more conductive pathways, keeping the Au layer at a low resistance value in the untensioned state. The cracks were carved on both sides of the PMCs using an engraving machine. The knife was placed on the cantilever of the tensile tester, and the film was placed on the table of the tensile tester. Grid paper was placed on the workbench to locate the position of the PMCs, and the control of crack position was achieved. By precisely controlling the cantilever movement of the tensile tester, precise control of the crack depth can be achieved. The Au film was formed on the PDMS through the vacuum evaporation technique. Then the two wires were attached on the two sides of the sensors using copper foil and silver paste, where copper foil is used to conductively connect the upper Au film and the bottom MXene layer. The connection between Au and MXene forms a parallel circuit endows the sensor with switching characteristics, avoiding irreparable deformation of the sensor and an ineffective increase in resistance value. Fig. S2 in the ESI† shows the specific preparation process and Fig. S3† displays some physical photos of the preparation of PMCs.

Fig. 1a shows the whole structure of the strain sensor, consisting of Au–PDMS–MXene composites with a double-crack structure. The involved materials in this sensor included PDMS, Au, MXene, silver paste, and copper foil. The reason for using PDMS is that it has the characteristics of low cost, simple use, good adhesion, and good chemical inertness. MXene has high conductivity, high mechanical strength, good bonding stability, and high sensitivity to external pressure, and Au has excellent conductivity and good ductility. PDMS was chosen as the stretchable matrix to transfer strain. Au was located above PDMS as the strain-sensitive layer with rapidly changing resistance to strain. An MXene was used as a stable large resistance layer under the PDMS. Fig. 1b illustrates the photo of the strain sensor, which shows that the sensor was tightly attached to the arm. The cracks carved on different material surfaces are demonstrated in Fig. S4.† As seen, micron-scale cracks were respectively distributed on the PDMS film, MXene layer, and Au layer. When there was no deformation, the cracks were closed. But when tension was applied, the materials began to deform at the two ends, the corresponding cracks began to open, and the gap width gradually increased. This phenomenon is displayed in Fig. S5.†

The strain distribution of APMCs under different forces was simulated in COMSOL. Fig. 1c shows the simulation of strain distribution on the surface of the device under the action of 0.1 N force at both ends and the strain that occurred at this time was small (less than 2%). The strain was mainly focused on the bottom of the micron-scale cracks, resulting in a strong strain concentration. Therefore, the improvement of the sensitivity under small strain depends mainly on the opening and closing of the micron-scale cracks. Fig. 1d displays the simulation of strain distribution on the surface of the device under the action of 5 N force at both ends and the strain reached 8%. In this case, strain was also generated in the flat area of the device surface. Due to the mechanical modulus mismatch between the Au layer and the PDMS layer, nano-scale cracks caused by external forces appeared in the flat areas of the Au layer. Thus, when the strain is large, the increase in sensitivity is determined by both micron-scale cracks and nano-scale cracks.

Fig. 1e shows the variation of the resistance value with deformation for the sensor as a whole and different layers. The Au layer had a fast change in resistance to strain as well as the MXene layer with hollow cracks showed large resistance and the resistance changes slowly with strain. The upper and lower conductive functional layers were separated by PDMS film, but the two ends were connected by copper foil, thus forming a parallel circuit structure. Due to the parallel connection of these two resistors, the total resistance tends to be biased towards the resistance value of the MXene layer when the deformation is large, resulting in a gradual change. At the same time, the resistance of the Au layer changes rapidly. Fig. 1f illustrates an application demonstration of controlling a manipulator by sensors integrated into the hand for a simple human–machine interaction. In addition, the sensor was also applied to detect the physical movements of the throat, eyelids, ankles, elbows, wrists, and fingers, indicating that the sensor was a promising candidate in the field of full-range human motion detection.

3.1. Structure analysis

To further characterize the micro-structure of APMCs, a high-speed digital microscope, scanning electron microscope, and energy dispersive spectrometer (EDS) were used for observation and analysis. Fig. S6† shows the microstructure of the APMCs observed using a high-speed digital microscope. Fig. S6a† is an image obtained by observing the Au layer on the PDM film and Fig. S6b† displays the contact surface between PDMS and Au layer. Due to the ductility of the flexible matrix, the temperature change during the vacuum evaporation coating process causes the matrix to deform slightly. Therefore, the upper Au layer was not a flat and smooth material layer but was distributed with small nano-cracks like the land cracked by drought. As shown in Fig. S6c,† the micron-scale cracks on the Au layer coexisted with the nano-scale cracks naturally formed on the surface, which determined the sensing performance of the APMCs together. Fig. S6d–f† show the microstructure of the MXene layer at different magnifications. Due to the characteristics of the material itself, the attached MXene showed a stacked layered structure.

For APMCs, there are three kinds of cracks, which include artificial micron-scale cracks on the MXene layer, artificial micron-scale cracks, and natural nano-scale cracks on the Au layer. Fig. 2a and b show the structure of artificial micron-scale cracks on the MXene layer. It is seen that the width of the cracks was about 20 μm, and these cracks were hollow because they were carved after the formation of the MXene layer. Fig. 2c–f shows the structures of micron-nano cracks on the Au layer. The width of the micron-scale cracks was about 20 μm (Fig. 2d) and the width of the nano-scale cracks was about 500 nm (Fig. 2f). Unlike cracks on the MXene layer, the micron-scale cracks on the Au layer were filled with conductive Au because Au was plated after carving the cracks. In addition, the thickness of the Au layer was about 500 nm (Fig. S7a†), and the thickness of the MXene layer was about 2 μm (Fig. S7b†). Fig. S8† displays the SEM images of the contact surface between the Au layer and the PDMS layer, and the nano-scale cracks on the Au layer. Fig. 2g shows the EDS mapping images of a micron-scale crack on the MXene layer, which showed the distribution of Si, Ti, C, and O (Fig. S9†) elements. The titanium element was derived from the MXene sheets (Fig. S10a†) while the silicon element came from PDMS (Fig. S10b†), which proved the existence of the cracks on the surface.

3.2. Working mechanism

As shown in Fig. S11,† arthropods such as spiders or scorpions in nature sensitively perceive small stimuli through the crack-sensing organs in the feet (lyriform organs). When cracks are created or healed, information is fed back to the brain through connected nerves, so the brain can respond rapidly to tiny changes. The operating principle of the lyriform organs demonstrates that masterly constructing surface cracks can improve the sensitivity of flexible sensors under a small strain.

Inspired by lyriform organs, APMCs prepared by layered manufacturing methods have similar functional structures. Fig. 3a shows that multiple cracks are arranged in parallel on the Au layer with a fast change in resistance to strain. In addition, the MXene layer with hollow cracks shows large resistance and the resistance changes slowly with strain. The upper and lower conductive functional layers are separated by the PDMS film, but the two ends are connected by copper foil, thus forming a parallel circuit structure. Fig. 3b reflects the relationship between the resistance value of the branch and the total circuit in the parallel circuit (ideal state). The relationship of parallel circuits can be expressed as

	(2)

where R₁ is the resistance of the Au layer, which is the small resistance layer; R₂ is the resistance of the MXene layer, which is the large resistance layer; and R_Total is the total resistance of the APMCs. The model of the strain sensor with different stretches is proposed in Fig. 3c–e. When the sensor is in the initial phase, there are many conductive paths (Fig. 3c) because the upper micron-scale cracks are filled with Au and these cracks are in a closed state. Thus, R₁ is very small and this branch can be approximated as conducting. At the same time, most of the conductive paths are cut off after engraving cracks on the MXene layer, so R₂ is relatively large. When a small strain occurs, the cracks on the upper Au layer start to open and form tiny gaps, and the conductive paths begin to decrease (Fig. 3d), leading to an increase in resistance R₁. However, the lower MXene layer has a larger resistance value R₂, and fewer changes in the conductive paths, which is reflected in a slightly floating resistance value. At this stage, the dominant functional layer is the Au layer. R_Total and R₁ have almost the same changes, reflecting the rapid and obvious response to strain. When the stretching strain increases further, these micron-scale cracks and gaps gradually propagate, eventually resulting in the disruption of the conductive Au network (Fig. 3e), and the resistance of R₁ exceeds the resistance of R₂. At this stage, the conductive paths are now carried out by the MXene network and the MXene layer is the dominant functional layer. Ideally, the resistance of the strain sensor is reflected as a stable resistance. The resistance of parallel circuits tends to be smaller branch resistance values. When the sensor undergoes small deformation, it is equivalent to an Au layer resistance with a smaller resistance. When the deformation is larger, the Au layer resistance exceeds the MXene layer resistance, and the device is equivalent to the MXene layer resistance, showing a stable state of slow rise. Therefore, the resistance of the MXene layer acts as a clamping resistance, limiting the phenomenon of significant increase in resistance of the sensor under large deformation. The clamping resistor R₂ can be adjusted during the preparation process as needed. By changing the content of MXene or the number of cracks, the resistance of R₂ can be adjusted, realizing the setting of the clamping resistor. Within its working range, as the stretching strain is released, these opened cracks are recovered due to the excellent elasticity of the PDMS substrate (essentially reversible), so that the same conductive network is restored, with the resistance of the flexible strain sensor returning to its initial value.

Due to the mechanical modulus mismatch between the Au layer and the PDMS layer, some nano-scale cracks appear in the dense Au layer after pre-stretching. For the Au layer, the surface is divided into a normal distribution area, artificial micron-scale large cracks, and natural nano-scale small cracks. The width of micron-scale cracks is about 20 μm, and their resistance changes significantly, which is affected by a single micron-scale crack. The width of nano-scale cracks is approximately 500 nm, and significant changes in the resistance require changes in the conductive pathways of many nano-scale cracks. Since there will be no cracks in the normal area, the resistance value here will not change. The relationship between the various resistors is as follows:

R1 = RNormal + RMicron + RNano	(3)

RNormal = RNormal1 + RNormal2 + … + RNormalx	(4)

RMicron = RMicron1 + RMicron2 + … + RMicrony	(5)

RNano = RNano1 + RNano2 + … + RNanoz	(6)

where R_Normal, R_Micron, and R_Nano represent the resistance of the three parts. It is easy to observe the distribution and series relationship of the three types of resistance on the Au layer in Fig. S12.† For R_Normal, since the strain does not change the state of the smooth and flat area, it is a type of fixed-value resistor, which is approximately conductive for the Au layer. When no strain occurs, the nano-scale cracks on the Au layer are closed and the micron-scale cracks are filled with Au, so the corresponding R_Nano and R_Micron are very small. When a small strain begins to occur, micron-scale cracks will preferentially open and R_Micron will rise. Meanwhile, nano-scale cracks don't open so R_Nano remains unchanged. When the strain becomes larger, the nano-scale cracks gradually open and R_Nano begins to rise rapidly like R_Micron. The structure evolution process of the nano-scale cracks under different degrees of stretching was investigated by SEM (Fig. 3f–h). It is worth mentioning that, due to different cracks having different effects on the conductive paths, the resistance change caused by micron-scale cracks is greater than that by nano-scale cracks. Overall, during the straining process, the Au layer exhibits rapid resistance changes under the joint action of the two cracks.

3.3. Sensing performance characterization

To fully display the dynamic characteristics of the flexible strain sensor based on APMCs with a double-crack structure, the device was placed on a tensile machine (Fig. S13†), and strain of different magnitudes and frequencies was applied to test and characterize the device performance. Fig. S14† shows photos of the prepared flexible strain sensor under different strains, which reflect the recovery stability.

To evaluate the sensing performance of the sensor in terms of strain detection, the resistance changes (ΔR/R₀) under different strains were measured (Fig. 4a), where R and R₀ (about 100 Ω) are the resistance values of the sensor with and without strain, respectively. It is noticed that the sensing behavior could be divided into three regions: small strain region (0–3%), rapid change region (3–15%), and buffer (15–21%). The basis for classification is based on differences in sensing performance and application scenarios. In the small strain region, the change in the resistance of the sensor is mainly affected by the micron-scale cracks on the surface of the Au layer, resulting in a sensitive and obvious response. In the rapid change region, the change of sensor resistance is combined with micron-nano cracks. As shown in this region, the resistance value changes the fastest and the largest GF appears here, with a value of 14373.6. In the buffer, the resistance of the Au layer far exceeds the resistance of the MXene layer because the cracks are fully opened. At this time, the resistance of the sensor is mainly determined by the resistance of the MXene layer, so the overall resistance slowly rises and can reach a resistance change of about 2000 times. In general, the GF of the small strain region and buffer are both less than 3500, and the GF in the rapid change region is greater than 3500. In Fig. 4b, the maximum sensitivity (GF) and the sensing range of the flexible strain sensor based on APMCs with the double-crack structure are compared to the sensors in the reported literature studies.^{44–47,49–59} The existing strain sensors using a crack structure improve the sensitivity of the device but greatly reduce the operating range. Retaining a large operating range results in low sensitivity of the device. Due to micron-scale cracks, while retaining sensitivity, disperse the stress borne by nano-scale cracks and extend the working range of the device, the flexible strain sensor based on double-crack structure has both ultra-high sensitivity and a wide sensing range.

To verify the dynamic response of the strain sensor under different strains, periodic tensile/release tests of 0–16% strain were carried out, showing excellent stability and reliability, as shown in Fig. S15.† The reliability and durability of the sensor are essential factors affecting its application. Fig. 4c shows the resistance change of the sensor after 1000 cycles under 3% strain, and the resistance change curves of the early and late cycles are illustrated in the illustration. After many cycles, the shape of the response curve did not change significantly, proving that the strain sensor has good stability. The sensor has good dynamic response characteristics and is suitable for real-time monitoring. A tensile testing machine was used to quickly apply a tensile force on both ends of the strain sensor to make it achieve different deformations, and the resistance change was recorded during the whole process. Fig. 4d shows the response signal of the sensor when the applied strain is 9% and the response time here is 183 ms. As introduced in Fig. S16,† the strain of the device and its corresponding resistance change present a stepped distribution. The response time is stable at about 200 ms, which also proves that the strain sensor can adapt to the needs of applications about people. In addition, to characterize the cycling stability of the flexible strain sensor based on APMCs with a double-crack structure, the strain sensor was stretched back and forth to a 5% deformation at different frequencies. By applying a constant current (1 μA), the dynamic voltage curve can be obtained as shown in Fig. 4e. Output signals of the APMC strain sensor exhibit excellent stability at various frequencies in the range of 0.1–1 Hz.

Fig. S17† is a comparison diagram of the sensing performance of the sensor with and without the micron-scale cracks on the Au layer. The resistance of the sensor with cracks under the same deformation changes more, which is more conducive to detecting external deformation excitation. Affected by the micron-scale cracks, the sensor's resistance value exhibits larger changes under 0–3% deformation. Affected by micron-nano cracks, the sensor's resistance value changes faster under 3–15% deformation. Therefore, the crack structures could significantly decrease the detection limit and improve the sensitivity. Fig. 4f reflects the change in resistance under different deformations of sensors with different cracks. The resistance changes of sensors with cracks under slight deformation are about 5 times, 10 times, and 40 times, while sensors without cracks correspond to only about 1.02, 1.1, and 3 times. This further verified the important role of the crack structure in improving the sensitivity of the flexible sensor.

The key factor affecting sensor performance is the crack structure. We then test the influence of crack density on the sensor performance. Since the number of cracks was inversely proportional to the size of the crack interval, so crack density (d, the ratio of the whole length of the sensor to the distance between two cracks) was used as an independent variable. As shown in Fig. 4g, the resistance changes under 5% deformation are measured. It can be seen that as the crack density becomes smaller, the relative change in resistance of the sensor to a certain deformation variable is also decreasing. Therefore, by changing the number of cracks on the sensor, the sensing performance can be changed, which provides a new route for preparing flexible sensors with controllable performance.

3.4. Applications of strain sensors

The flexible strain sensor was then applied to monitor human motion, including large physical motion and weak physiological signals to prove its practicality. The detection of subtle motion signals usually requires sensors with high sensitivity, while monitoring large motion signals requires sensors with a wide working range. In addition, on-skin tests were carried out to prove the biocompatibility for long-term wearability (Fig. S18†). Three samples (Au layer, MXene layer and PDMS film) were attached to the forearms of one volunteer for one day of wearing. As shown in Fig. S18,† neither the Au layer nor the MXene layer caused any negative effect on the skin after one day of being attached. The schematic illustrates that the GMPC strain sensor was attached to various body parts to acquire posture information related to skin deformation, such as throat, eyelids, ankles, elbows, wrists, and fingers (Fig. 5a). Fig. 5b presents the resistance change caused by the vibration of the vocal cords transmitted to the sensor when a person makes three sounds of “S”, “E”, and “U”. The sensor detects the changes in the state of the eyelids during blinking as shown in Fig. 5c. Stability in frequency distinction and obvious resistance change under a large deformation of more than 3% are important components of the flexible strain sensor's performance. The flexible deformation sensor fixed at the ankle of the human body can sense the movement state of the human body by capturing the deformation of the skin at the ankle, as shown in Fig. 5(d). It can easily distinguish the state of walking and running. Fig. 5e–g show the rapid change in the resistance value of the deformation sensor when the elbow, wrist, and finger are bent. In addition, Fig. S19† shows that the flexible strain sensor can be used for the detection of pulse signals.

The flexible strain sensor can be employed for the accurate monitoring of human joints due to its high sensitivity, fast response, and flexibility. As proof-of-concept, we developed a human–machine interaction system of posture acquisition and interactive feedback. Fig. 6a presents the circuit scheme of the human–machine interaction system. Six strain sensors were fixed at the concave surface of the thumb joint, forefinger joint, middle finger joint, ring finger joint, little finger joint, and wrist joint and were serially connected with load resistors, respectively. Next, the microchip would acquire and process the voltages across the six strain sensors. Finally, the control commands in the manipulator communication protocol are processed and output by the single-chip microcomputer, to realize the interaction between different manipulators. Fig. 6b and c show the corresponding resistance changes of the strain sensors for different hand gestures and a schematic diagram of the normalized voltage signal for remote human–machine interaction combined with APMCs strain sensors. The picture of a sensing glove based on APMCs that could control the machine hand is shown in Fig. S20.† We displayed the use of fingers to control the gestures of “three”, “four”, “five” and “I love you” and the rotation of the manipulator with the wrist (‘+’ and ‘−’ indicate palm orientation). For example, for the palm opposite hand gesture “I love you”, the middle finger, the ring finger, and the wrist bend, cause the voltage across the strain sensors to increase, while for the remaining three fingers that remain straight, the voltage does not increase. Therefore, the voltage signal is encoded as “001101”. This system has a high accuracy rate for the recognition of different gestures and can adapt to better human–machine interaction application scenarios.

4. Conclusions

In the study, inspired by the structure of lyriform organs which can perceive tiny vibrations, we developed a flexible strain sensor with high sensitivity in a wide sensing range based on the double-crack structure. The MXene and Au layers with cracks represent different resistances, forming a parallel circuit that can rapidly switch the resistance value. When a strain occurs, the rapid disconnection of the crack junctions on the Au layer causes a significant variation in the resistance value so that an obvious and rapid response can be achieved. The cracks improve the sensor's sensitivity, while the double-layer structure and the parallel circuit ensure its wide sensing range. The related sensing performance test proves that it has excellent sensing performance and the effect of the crack structure on device performance is explored. Benefiting from the double-crack structure, the as-obtained flexible strain sensor showed an ultra-high sensitivity (maximum gauge factor of up to 14373.6) and a wide working range (up to 21%). In addition, the dynamic characteristics of the strain sensor were investigated in detail, indicating that the sensor had a fast response time (183 ms) and excellent dynamical stability (almost no performance loss after 1000 stretching cycles and different frequency cycles). Finally, the sensor was applied to detect the physical movements of the throat, eyelids, ankles, elbows, wrists, and fingers, indicating that the sensor was a promising candidate in the field of full-range human motion detection. Besides, a human–machine interaction system of posture acquisition and interactive feedback was built, which had a high accuracy rate for the recognition of different gestures and could adapt to better human–machine interaction application scenarios. This study could provide new ideas for preparing high-performance flexible strain sensors.

Conflicts of interest

The authors declare no competing interests.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (62075040), the Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX22_0230), the National Key R&D Program of China (2017YFB3600502, 2021YFF0701100), and the Start-up Research Fund of Southeast University (RF1028623164).

Notes and references

1. S.-T. Han, H. Peng, Q. Sun, S. Venkatesh, K.-S. Chung, S. C. Lau, Y. Zhou and V. A. L. Roy, Adv. Mater., 2017, 29, 1700375 CrossRef PubMed .
2. Q. Wang, J. Liu, G. Tian and D. Zhang, Nanoscale, 2021, 13, 14460–14468 RSC .
3. Q. Liu, Z. Liu, C. Li, K. Xie, P. Zhu, B. Shao, J. Zhang, J. Yang, J. Zhang, Q. Wang and C. F. Guo, Adv. Sci., 2020, 7, 2000348 CrossRef CAS PubMed .
4. S. Duan, Q. Shi, J. Hong, D. Zhu, Y. Lin, Y. Li, W. Lei, C. Lee and J. Wu, ACS Nano, 2023, 17, 1355–1371 CrossRef CAS PubMed .
5. S. Duan, Q. Shi and J. Wu, Adv. Intell. Syst., 2022, 4, 2200213 CrossRef .
6. K. Sengupta, T. Nagatsuma and D. M. Mittleman, Nat. Electron., 2018, 1, 622–635 CrossRef .
7. F. Ejeian, S. Azadi, A. Razmjou, Y. Orooji, A. Kottapalli, M. Ebrahimi Warkiani and M. Asadnia, Sens. Actuators, A, 2019, 295, 483–502 CrossRef CAS .
8. M. Segev-Bar and H. Haick, ACS Nano, 2013, 7, 8366–8378 CrossRef CAS PubMed .
9. P. Wang, M. Hu, H. Wang, Z. Chen, Y. Feng, J. Wang, W. Ling and Y. Huang, Adv. Sci., 2020, 7, 2001116 CrossRef CAS PubMed .
10. D. Zhao, B. Nie, G. Qi, S. Li, Q. Zhu, J. Qiu, Y. Hsu, Y. Zhang, W. Wang, Q. Zhang and Z. Wei, Nanoscale, 2022, 14, 12409–12417 RSC .
11. M. Zhu, J. Li, J. Yu, Z. Li and B. Ding, Angew. Chem., Int. Ed., 2022, 61, e202200226 CrossRef CAS PubMed .
12. J. Y. Oh and Z. Bao, Adv. Sci., 2019, 6, 1900186 CrossRef PubMed .
13. J. Chen, Y. Zhu, X. Chang, D. Pan, G. Song, Z. Guo and N. Naik, Adv. Funct. Mater., 2021, 31, 2104686 CrossRef CAS .
14. E. Aslanidis, E. Skotadis and D. Tsoukalas, Nanoscale, 2021, 13, 3263–3274 RSC .
15. T. Q. Trung and N.-E. Lee, Adv. Mater., 2016, 28, 4338–4372 CrossRef CAS PubMed .
16. B. Liang, Z. Zhang, W. Chen, D. Lu, L. Yang, R. Yang, H. Zhu, Z. Tang and X. Gui, Nano-Micro Lett., 2019, 11, 92 CrossRef CAS PubMed .
17. L. Xie, Z. Zhang, Q. Wu, Z. Gao, G. Mi, R. Wang, H.-b. Sun, Y. Zhao and Y. Du, Nanoscale, 2023, 15, 405–433 RSC .
18. J. Tang, Y. Wu, S. Ma, T. Yan and Z. Pan, Composites, Part B, 2022, 232, 109605 CrossRef CAS .
19. Q. Zheng, J.-h. Lee, X. Shen, X. Chen and J.-K. Kim, Mater. Today, 2020, 36, 158–179 CrossRef CAS .
20. Q. Shi, Z. Sun, Z. Zhang and C. Lee, Research, 2021, 2021, 6849171 CAS .
21. S. Seyedin, P. Zhang, M. Naebe, S. Qin, J. Chen, X. Wang and J. M. Razal, Mater. Horiz., 2019, 6, 219–249 RSC .
22. S. Duan, Y. Lin, C. Zhang, Y. Li, D. Zhu, J. Wu and W. Lei, Nano Energy, 2022, 91, 106650 CrossRef CAS .
23. S. Goossens, B. De Pauw, T. Geernaert, M. S. Salmanpour, Z. Sharif Khodaei, E. Karachalios, D. Saenz-Castillo, H. Thienpont and F. Berghmans, Smart Mater. Struct., 2019, 28, 065008 CrossRef CAS .
24. A. Maskay, D. M. Hummels and M. Pereira da Cunha, Measurement, 2018, 126, 318–321 CrossRef .
25. C.-C. Lu, W.-C. Gao, P. Li, W. Wu, R. K. Y. Li and H. Zhao, Chin. J. Polym. Sci., 2022, 41, 1037–1050 CrossRef .
26. M. Zhang, J. Li, L. Kang, N. Zhang, C. Huang, Y. He, M. Hu, X. Zhou and J. Zhang, Nanoscale, 2020, 12, 3988–3996 RSC .
27. Y. Yang and W. Gao, Chem. Soc. Rev., 2019, 48, 1465–1491 RSC .
28. J.-Y. Yu, S. E. Moon, J. H. Kim and S. M. Kang, Nano-Micro Lett., 2023, 15, 51 CrossRef CAS PubMed .
29. S. Duan, H. Yang, J. Hong, Y. Li, Y. Lin, D. Zhu, W. Lei and J. Wu, Nano Energy, 2022, 102, 107665 CrossRef CAS .
30. S. Pan, Z. Liu, M. Wang, Y. Jiang, Y. Luo, C. Wan, D. Qi, C. Wang, X. Ge and X. Chen, Adv. Mater., 2019, 31, 1903130 CrossRef PubMed .
31. X. Guo, W. Hong, T. Zhang, H. Li, T. Zhu, Q. Hong, D. Wang, L. Liu, Y. Zhao, D. Wang, Z. Mai, M. Wang, F. Yan, Y. Meng, Y. Xu and G. Xing, Adv. Mater. Technol., 2023, 8, 2201439 CrossRef CAS .
32. X. Wang, X. Liu and D. W. Schubert, Nano-Micro Lett., 2021, 13, 64 CrossRef PubMed .
33. Y. Hu, T. Zhao, P. Zhu, Y. Zhang, X. Liang, R. Sun and C.-P. Wong, Nano Res., 2018, 11, 1938–1955 CrossRef .
34. S. Duan, Y. Lin, Z. Wang, J. Tang, Y. Li, D. Zhu, J. Wu, L. Tao, C.-H. Choi, L. Sun, J. Xia, L. Wei and B. Wang, Research, 2021, 2021, 9861467 CAS .
35. J. Liu, P. Wang, G. Li, L. Yang, W. Yu, C. Meng and S. Guo, Nanoscale, 2022, 14, 12418–12430 RSC .
36. X. Li, M. Li, X. Li, X. Fan and C. Zhi, Research, 2022, 2022, 9892628 CAS .
37. S. Duan, X. Wei, F. Zhao, H. Yang, Y. Wang, P. Chen, J. Hong, S. Xiang, M. Luo, Q. Shi, G. Shen and J. Wu, Adv. Sci., 2023, 2304121 CrossRef PubMed .
38. J. Sun, Y. Yuan, G. Lu, T. Xue, J. Nie and Y. Lu, Adv. Mater. Interfaces, 2022, 9, 2102245 CrossRef CAS .
39. M. Xie, Y. Hu, J. Yin, Z. Zhao, J. Chen and J. Chao, Research, 2022, 2022, 9840131 CAS .
40. F. Li, T. Shen, C. Wang, Y. Zhang, J. Qi and H. Zhang, Nano-Micro Lett., 2020, 12, 106 CrossRef CAS PubMed .
41. J. Qiu, X. Guo, R. Chu, S. Wang, W. Zeng, L. Qu, Y. Zhao, F. Yan and G. Xing, ACS Appl. Mater. Interfaces, 2019, 11, 40716–40725 CrossRef CAS PubMed .
42. I. Plesco, M. Dragoman, J. Strobel, L. Ghimpu, F. Schütt, A. Dinescu, V. Ursaki, L. Kienle, R. Adelung and I. Tiginyanu, Superlattices Microstruct., 2018, 117, 418–422 CrossRef CAS .
43. H. Souri and D. Bhattacharyya, J. Mater. Chem. C, 2018, 6, 10524–10531 RSC .
44. S. Han, C. Liu, H. Xu, D. Yao, K. Yan, H. Zheng, H.-J. Chen, X. Gui, S. Chu and C. Liu, npj Flexible Electron., 2018, 2, 16 CrossRef .
45. J. Ji, C. Zhang, S. Yang, Y. Liu, J. Wang and Z. Shi, ACS Appl. Mater. Interfaces, 2022, 14, 24059–24066 CrossRef CAS PubMed .
46. X. Wu, H. Chen, X. Luo, D. Wang, P. Schaaf and G. Zhang, Adv. Mater. Technol., 2021, 6, 2100524 CrossRef CAS .
47. W. Wang, Y. Ma, T. Wang, K. Ding, W. Zhao, L. Jiao, D. Shu, C. Li, F. Hua, H. Jiang, S. Tong, S. Yang, Y. Ni and B. Cheng, ACS Appl. Mater. Interfaces, 2022, 14, 36611–36621 CrossRef CAS PubMed .
48. J. Wu, Y. Li, S. Duan, Z. Wang, X. Jing, Y. Lin, D. Zhu, W. Lei, Q. Shi and L. Tao, Research, 2023, 6, 0106 CrossRef CAS PubMed .
49. D. Kang, P. V. Pikhitsa, Y. W. Choi, C. Lee, S. S. Shin, L. Piao, B. Park, K.-Y. Suh, T.-i. Kim and M. Choi, Nature, 2014, 516, 222–226 CrossRef CAS PubMed .
50. Y. Wang, F. Wang, S. Yazigi, D. Zhang, X. Gui, Y. Qi, J. Zhong and L. Sun, Carbon, 2021, 173, 849–856 CrossRef CAS .
51. K.-H. Kim, T. M. Nguyen, S.-H. Ha, E. J. Choi, Y. Kim, W.-G. Kim, J.-W. Oh and J.-M. Kim, ACS Appl. Mater. Interfaces, 2020, 12, 45590–45601 CrossRef CAS PubMed .
52. P. Xue, C. Chen and D. Diao, Carbon, 2019, 147, 227–235 CrossRef CAS .
53. J. Lin, X. Cai, Z. Liu, N. Liu, M. Xie, B. Zhou, H. Wang and Z. Guo, Adv. Funct. Mater., 2020, 30, 2000398 CrossRef CAS .
54. S. Yang, C. Li, X. Chen, Y. Zhao, H. Zhang, N. Wen, Z. Fan and L. Pan, ACS Appl. Mater. Interfaces, 2020, 12, 19874–19881 CrossRef CAS PubMed .
55. B. Feng, X. Jiang, G. Zou, W. Wang, T. Sun, H. Yang, G. Zhao, M. Dong, Y. Xiao, H. Zhu and L. Liu, Adv. Funct. Mater., 2021, 31, 2102359 CrossRef CAS .
56. X. Zhao, H. Guo, P. Ding, W. Zhai, C. Liu, C. Shen and K. Dai, Nano Energy, 2023, 108, 108197 CrossRef CAS .
57. J. Li, R. Bao, J. Tao, M. Dong, Y. Zhang, S. Fu, D. Peng and C. Pan, Phys. Rev. Appl., 2020, 7, 011404 CrossRef CAS .
58. O. A. Araromi, M. A. Graule, K. L. Dorsey, S. Castellanos, J. R. Foster, W.-H. Hsu, A. E. Passy, J. J. Vlassak, J. C. Weaver, C. J. Walsh and R. J. Wood, Nature, 2020, 587, 219–224 CrossRef CAS PubMed .
59. L. Wang, X. Xu, J. Chen, W. Su, F. Zhang, A. Li, C. Li, C. Xu and Y. Sun, ACS Nano, 2022, 16, 12645–12655 CrossRef CAS PubMed .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3nr05476d

‡ Di Zhu and Shengshun Duan contributed equally to this work.

This journal is © The Royal Society of Chemistry 2024