Transparent and Stretchable Piezoresistive Strain Sensors with Buckled Indium Tin Oxide Film

Owing to excellent performance and a simple mechanism, stretchable piezoresistive strain sensors have been applied to human skin for monitoring physical activities, physiological activities, etc. However, it is still a challenge to simultaneously realize highly sensitive and stretchable piezoresistive strain sensors with high optical transparency. This study reports a transparent and stretchable piezoresistive strain sensor with 2D surface buckling by fabricating ultrathin indium tin oxide (ITO) film on the biaxially pre‐stretched polyacrylate (VHB) elastomer followed by pre‐stretch releasing. To the authors' knowledge, semiconductors are applied for a stretchable piezoresistive strain sensor for the first time. Furthermore, this strain sensor exhibits a high sensitivity of 569, a high transparency of 88% and a high biaxial stretchability of 110% at the same time. This device demonstrates the good long‐term stability over 500 stretching–relaxing cycles. The high sensitivity can be mainly attributed to the piezoresistive effect of the semiconductor where carrier mobility and the resulting resistivity can be significantly changed by the strain. The strain sensors attached to human skin are used to monitor many human motions such as chewing, swallowing, breathing, and walking. ITO‐based strain sensors pave the way toward the development of highly stretchable and sensitive wearable electronics.


Introduction
Stretchable piezoresistive strain sensors, which could seamlessly and conformally contact with skin surfaces and measure the DOI: 10.1002/aelm.202300197 physical and physiological signals of the human body in a simple work principle, [1] are crucial in wearable sensing applications such as human-motion detection, health monitoring, artificial prostheses, and soft robots. [2] Stretchable piezoresistive strain sensors have been intensively developed with two typical design strategies. The first strategy is the regulation of strainsensitive connection resistance between the discontinuous components of a conductive percolation network with a stretchable and insulative matrix or substrate versus the applied strain. [3][4][5][6][7][8] The second sensor design strategy is the modulation of intrinsic resistance of the material itself versus the applied strain by using the piezoresistive theory. [9] This strategy is mainly implemented with the method of directly measuring the resistance of intrinsically conductive and stretchable organic materials or by employing geometric engineering of brittle materials. [9][10][11] For geometric engineering, the conductive or semi-conductive brittle films are patterned or hollowed out to tolerate large-scale strain or deformation. The approach of geometric engineering can be categorized into structure hollowing (e.g., forming serpentine, [12] kirigami, [13] horseshoe shape, [14] etc.), discontinuous stiff components patterning (e.g., forming rigid islands), [15] and surface buckling. [9] Compared with other approaches, surface buckling is a simpler method, which could enable brittle thin film to adaptively form the stretchable crumpled geometric structure versus the applied strain, without any intentional patterning process. On the one hand, surface buckling approach could effectively improve the stretchability of piezoresistive strain sensors based on brittle materials. In this approach, the stiff thin film is attached to a pre-stretched stretchable elastomer substrate, forming a stiff film/compliant substrate bilayer mechanical system. When prestretching is released, the stiff film would bear the compression stress and flex or buckles into micro-folds on the soft substrate due to the highly nonlinear elasticity of the elastomeric substrate. [16] According to the beam and plate theory, when these buckled patterns form, the apparent Young's modulus of stiff thin film would be reduced by several orders of magnitude and its fracture strain would greatly increase due to the microscopic www.advancedsciencenews.com www.advelectronicmat.de bending behavior of stiff film, enhancing the stretchability of sensor. [17,18] Thus, the buckled stiff film can be stretched and unfolded to a relatively flat state within a "tension" strain range.
On the other hand, surface buckling could also effectively detect the strain in a simple working principle. According to the theory of piezoresistive effect, the sensitivity, namely gauge factor, of piezoresistive strain sensor is determined by the geometric and resistivity effects. [19][20][21] The device sensitivity is usually positively correlative to Poisson's ratio of the material and resistivity change.
In recent years, surface buckling has been explored to develop stretchable piezoresistive strain sensors with carbon or metal materials. [22][23][24][25][26][27][28] The sensitivity of these devices is mainly decided by the geometric effect since the resistivities of carbon or metal materials change little versus strain. Carbon-based strain sensors that are made of 2D materials exhibit the large stretchability due to the small mechanical strengths. S. Yan et al. developed a strain sensor with 1D-wrinkled reduced graphene oxide (rGO) film, exhibiting the large stretchability of 300%. However, this sensor shows a low sensitivity of about 0.03 at 300% strain. [22] The low sensitivity of this device can be attributed to the low Poisson's ratio of rGO between −0.567 and 0.27 according to the geometric effect theory mentioned above. [29] To solve the problem of low sensitivity, metals with the larger Poisson's ratios were applied for strain sensors. W. Zhang et al. developed a strain sensor with 1D-wrinkled silver (Ag) thin film, exhibiting a higher sensitivity (0.8) at 100% strain due to the higher Poisson's ratio (0.37). [25] M. Khine et al. further promoted the sensitivity of similar device up to 42 at 185% strain by using the 1D-wrinkled platinum (Pt) film with a higher Poisson's ratio (0.38). [23] However, this sensitivity of the sensor is not high enough for wide applications.
In a word, current piezoresistive strain sensors with metal or carbon-based active films exhibit the low sensitivity at the large strains. It remains a great challenge to simultaneously achieve the high sensitivity while keeping the high stretchability in a sensor. In addition, all the reported sensors were pre-stretched uniaxially so that they can only be stretchable in the pre-stretched direction rather than other directions. Nevertheless, actual human motions occur in the multi-direction rather than an ideal single direction. Therefore, the monitoring of multi-directional human motions is the other challenge to the current stretchable piezoresistive strain sensors.
To meet the challenge of low sensitivity, besides the geometric effect, the resistivity change of material versus strain could also be used to further improve device sensitivity according to piezoresistive effect theory. In fact, the resistivity effect of some active material (e.g., semiconductors) would greatly change device sensitivity by several orders of magnitude versus strain and the geometric effect can only linearly change the sensitivity versus strain according to the theory of piezoresistive effect. In fact, numerous rigid strain sensors based on semiconductors (e.g., silicon, indium tin oxide (ITO), etc.) exhibit the much higher sensitivity than metals and other conductive materials. [19,[30][31][32][33][34] However, to our knowledge, semiconductor has not yet been reported to act as the active material of any stretchable piezoresistive strain sensor.
Herein, we introduce ITO, which is the most widely used transparent and conductive semiconductor material, to develop a stretchable piezoresistive strain sensor with surface buckling.
As mentioned above, ITO has been widely used for conventional rigid strain sensors that exhibit the much higher sensitivity than those metal-based strain sensors. In addition, ITO also has the advantage of high optical transparency (>90% in the visible light range), facilitating to develop the transparent strain sensors. It would allow for direct observation of the skin or other components underneath the sensor and yield the integration of electric and optical devices.
In this work, we developed a new ITO-based stretchable piezoresistive strain sensor with surface buckling by depositing ultrathin ITO film on the biaxially pre-stretched polyacrylate (VHB) elastomer membrane. To our knowledge, it is the first time to apply semiconductor material and the underlying resistivity effect of piezoresistive effect theory to improve the sensitivity of a stretchable piezoresistive strain sensor. Our ITO-based strain sensor simultaneously exhibits the highest sensitivity of 569 and the highest transparency of 88% at the biaxial tension strain of 110%. We also apply for the first time the operation mode of biaxial stretching for piezoresistive strain sensors with surface buckling structures. Thus, the strain sensor exhibits excellent and robust performance for real-time monitoring the 2D human motion activities. These results pave the way to potentially apply semiconductor-based sensors for wearable and stretchable electronics. Figure 1a shows a schematic diagram of the whole fabrication process flow of strain sensors. The active material of the ultrathin ITO film of the sensor was deposited as a resistor on the pre-stretched VHB elastomer membrane. As the pre-stretching is partially released, ITO film is uniformly compressed so that surface buckling forms out of the plane at the device surface. When the ITO-based strain sensor is stretched again, the buckled ITO film would be unfolded and its resistance would also vary with the increasing strain. Figure 1b displays the surface microstructure of the device, in which pre-stretched is fully released, by using a scanning electron microscope (SEM). SEM image demonstrates that the buckling patterns are randomly and uniformly distributed on the device surface. The inset SEM image shows the patterns are suspected to be wrinkles, ridges, or folders. These patterns seem to be locally periodic but randomly orientated in a large scale. Figure 1c shows the optical image of the strain sensor (inset) and experimental setup for device characterization. The change of device resistance can be real-time monitored as the sensor is biaxially stretched. Figure 1d shows the relative change in device resistance, namely (R − R 0 )/R 0 versus biaxial strain. Here, R 0 and R stand for the original and the after deformation resistance of the device, respectively, while ΔR, namely R − R 0 , is the absolute change of device resistance. It is worth noting that the original state is defined as the one where the pre-stretching is fully released. As seen, ΔR/R 0 is small and grows slightly as strain increases in a small range, whereas ΔR/R 0 arises significantly with a further increase in strain. To estimate the sensitivity of the strain sensor, the gauge factor (GF) of the strain sensor is defined as where strain is defined as ɛ = (D − D 0 )/D 0 , D 0 and D are the original and after deformation diameter of ITO film, respectively. As seen, the ITO-based strain sensor exhibits a relatively low gauge factor below 10 in the small strain range while its GF can greatly grow up to several hundred at a high strain of over 100%, showing both high sensitivity and stretchability. The 3D surface morphology of the ITO-based strain sensor at different stretching strain was characterized by laser confocal microscopy (LCM). Figure 2a shows the surface morphology of the strain sensor at a pristine state in which the pre-stretch of the elastomer substrate is fully released. The sensor surface is highly crumpled and covered by high-density nonperiodic patterns that are oriented randomly. These patterns seem to be hy-brid and composed of one type of a higher buckling pattern in red and the other lower one in blue or green. As the sensor is biaxially stretched, both the density and height of buckling patterns decrease, as seen in Figure 2b,c. This variation trend indicates that the crumpled ITO film can be gradually unfolded again without cracks formation. The height profiles of surface patterns in Figure 2d confirm that the patterns at the crumpled state (e.g., 50% strain) are mainly composed of the periodic-like lower patterns, which are roughly interdigitated with the higher localized patterns. These two patterns are considered to be ridges and wrinkles, respectively, as seen in Figure 2e. The ridge patterns are localized and show a peak in the center and relatively flat sides. Oppositely, wrinkle patterns appear in a sinusoidal-like  shape and seem to be periodic in the local zones. It is found that only wrinkles appear at 100% strain while both patterns appear at the small stretching strain. The defined 100% strain and pristine states correspond to the stress-free and compressed states, respectively, of the ITO strain sensors. It means that surface patterns would convert from wrinkles to wrinkle-ridge double patterns with enhanced compressive strain. This phenomenon is attributed to the large Young's modulus ratio between ITO and VHB elastomer materials. The pattern would be a wrinkle at a small compression strain and it would be converted into a ridge at a high strain. [35] Figure 2a-d also demonstrates that the rough sensor surface at pristine would be flattened with the increase in stretching strain and pattern heights also reduce in the same process. Figure 2f shows that, as stretching strain increases, surface roughness gradually reduces from a high value of 8.32 μm at pristine state down to 0.179 μm at 110% stretching strain. These results confirm that the ITO film is compliant enough to be folded and flattened without destructive delamination or cracks formation. In fact, as the pre-stretch is released, the ITO film would endure the compressed stress inside the horizontal plane. The ITO film would break at this large compressive strain in conventional conditions. However, the buckling of ultrathin ITO film reduces its Young's modulus and makes it compliant on an elastomer membrane, forming stable 3D buckled patterns out of the horizontal plane. Thus, once the compressed state of the ITO film is defined as the pristine state for the strain sensor, the device could be stretched on a large scale until the ITO film is flattened. In this way, the ITO-based strain sensor exhibits a high stretchability over 100% strain due to the formation of stable 3D crumpled patterns out of the stress plane.

Results and Discussion
High stretchability is crucial but not easily realized for stretchable strain sensors because destructive deformation often occurs at high strain states. Figure 3a depicts the measured stress and ΔR/R 0 of the ITO-based strain sensor versus the applied biaxial stretching strain. As seen, the stress gradually increases with the increase in biaxial stretching strain until it collapses at 650%.This indicates that the sensor can endure large-scale biaxial mechanical stretching due to the elastomeric VHB substrate until the membrane is broken. However, the crumpled stiff ITO film would be flattened at 110% strain and further stretching would induce the destructive crack formation. As seen in Figure S1, Supporting Information, when the stretching strain is below the threshold strain of 110%, no cracks are observed in the 3D morphology image while the crossed cracks occur as strain exceeds the threshold. This transition is also reflected by the resistance response of the sensor. The ΔR/R 0 also drastically changes from several tens up to values beyond 1000 as strain is over 110% strain, as shown in Figure 3a. The formation of cracks would block or completely cut off the original transport path of electrons, dramatically increasing the resistance of the sensor. Therefore, we improve the stretchability of the ITO-based strain sensor by promoting the threshold strain of crack formation from the conventional less than 3% to over 100% employing surface buckling. In practice, the stretchable strain of 50% is sufficient for most human motions. This means that our strain sensors exhibit high enough stretchability for practical applications. Figure 3b shows the resistance responses (ΔR/R 0 ) of the strain sensor to strain in a quasi-steady state test mode. The strain sensor can quickly respond to the stretching strain and exhibit fabulous segment stability when loaded with the step-up strains from pristine to 87%. In addition, the strain sensor working in a dynamic mode (Figure 3c) also demonstrates the quick and reliable response to the applied stepwise/cyclic loadings at various strains. In addition, the strain sensors also respond to strain promptly and stably at different stretching/relaxing cyclic frequencies, indicating robust performance in response ( Figure S2, Supporting Information). Figure 3d shows the resistance responses (ΔR/R 0 ) of the strain sensors upon the applied biaxial stretching strain with different pre-stretches. The variation trends of ΔR/R 0 show the common feature versus strain and the process of resistance response can be divided into three stages (Figure 1d) within the whole strain range. In the first stage, ΔR/R 0 increases slightly with the increasing strain in the small strain range ( Figure S3, Supporting Information). ΔR/R 0 usually does not exceed 1 within the initial at least 20% strain sensing range. Thus the gauge factor of the sensor is below 3, suggesting low sensitivity at this stage. In the second stage, the resistance response of the sensor increases up to over ten versus the further increasing stain in a gradual way, exhibiting the medium sensitivity of several tens. In the final stage, as the stretching strain continues to grow up, the ΔR/R 0 of the sensor increases significantly within a narrow stain range, indicating a high sensitivity of several hundred (Figures 1d and 3d). This means that the strain sensor owns a threshold strain for medium and high sensitivity detection. In practice, we can shift the threshold to a small value by tuning the initial state of the sensor.
The pre-stretch of the elastomeric substrate also plays a significant role in the performance of the strain sensor. Figure 3d shows that the larger pre-stretch results in the increase in threshold strain of the second stage mentioned above. The promotion of pre-stretch leads to the increase in the sensing strain range from 65% at 200% pre-stretch up to 123% at 400% pre-stretch ( Figure S4, Supporting Information). Meanwhile, the gauge factor also significantly increases with pre-stretch from 204 at 200% pre-stretch to 569 at 300% pre-stretch and has a saturation trend with the further increase in pre-stretch. The strain sensor exhibits the highest sensitivity of 610 at 123% biaxial strain, indicating the excellent performance of the sensor with both high sensitivity and stretchability.
Besides, the long-term stability of strain sensors is also crucial for practical applications. Figure 3e demonstrates that the electrical responses of the strain sensor exhibit high stability during the long-term tests for about 500 stretching-relaxing cycles, indicating the fabulous electrical and mechanical reliability of the sensor. In Figure 3f, we compare the gauge factor and workable sensing strain range of our strain sensor with those of the previously reported piezoresistive strain sensors with the surface buckling method. [22][23][24][25][26][27][28] The work of these peers falls into three categories including metals, carbon materials, and semiconductor materials such as ITO in this work. It is found that the ITObased strain sensor in this work exhibits the highest sensitivity of 610 compared to the reported sensors, and promotes the recording sensitivity of the same type of sensors nearly by one order of magnitude. Meanwhile, the workable sensing range of our sensors is up to 123% strain, which is large enough for practical applications. In addition, it is worthy that all the reported strain sensors in Figure 3f own high stretchability in a single direction while our devices can be stretchable biaxially. This biaxial stretch is more suitable for monitoring complex 2D human motions.
Although ITO-based strain sensors with high sensitivity have been developed, the underlying physical mechanisms should be elucidated because it is of great both scientific and practical importance. According to the basic piezoresistive effect theory, the resistance response (ΔR/R 0 ) of the strain sensor is given by two terms representing the geometric and resistivity effects. [19][20][21] where is the Poisson's ratio of the material, 0 is the original resistivity, is the resistivity, and Δ is the change of resistivity ( − 0 ). For metals, their resistivity changes little versus applied strain so that the latter part of the formula is negligible. Thus, the geometric effect, especially the Poisson's ratio of the material, dominates the sensitivity of metal-based piezoresistive strain sensors. The Pt or Ag material with a large Poisson's ratio (0.38 or 0.37) exhibits high sensitivity. Here, Poisson's ratio of ITO is 0.35 which is slightly smaller than that of Pt or Ag. [36] Thus, ITO-based strain sensors should own a relatively lower sensitivity than Ag or Pt-based sensors. However, the sensitivity of ITO-based sensors is much high than metal sensors. Not like metals, ITO is one type of doped wide-bandgap semiconductor. In fact, ITO film is biaxially compressed at the defined pristine state in this work while the stretching is the relaxing of compression in the view of the mechanics of materials. As a compressive strain is applied to ITO film, the density of states and lattice symmetry of the crystals change versus strain, shifting energy bands of polycrystalline ITO material. [37] This result would induce bandgap shifts, band warping, and a changed intervalley scattering rate, leading to the promotion of carrier mobility and conductivity. [38] Thus, the resistivity of ITO film would reduce under compressive strain. Therefore, the second part of Formula (2), namely the resistivity effect, also significantly contributes to the overall piezoresistive effect so that the ITO-based strain sensor owns a much higher sensitivity than the reported metal sensors. The gauge factors of semiconductors are usually 1-2 orders of magnitude larger than those of metals in conventional rigid strain sensors. [19] This result also verifies that semiconductors (e.g., ITO) based strain sensors indeed have a higher sensitivity than metal-based sensors no matter if are rigid or stretchable devices. In a word, the high sensitivity of ITO-based strain sensors is attributed to both the geometric effect and resistivity factors of the piezoresistive effect, especially the resistivity factor, due to the semiconductor features of ITO material.
In addition, if the wearable device such as a strain sensor is transparent, we can directly observe the skin state or integrate the sensor with other optoelectronic devices. Therefore, high optical transparency is an additional advantage for strain sensors, yielding a visually perfect fit with the human body. [39] However, it is still a challenge to simultaneously realize the high transparency and high sensitivity of stretchable strain sensors. Figure 4a illustrates the optical direct transmittance spectra of ITO-based strain sensors between 350 and 800 nm. When biaxial strain is below 50%, the strain sensor exhibits a relatively low transmittance, which is attributed to the light scattering effect induced by the rough surface consisting of buckled patterns. As biaxial strain exceeds 50%, the transmittance of the sensor increases significantly because the light scattering is reduced due to the unfolding and flattening of the crumpled surface. Figure 4b shows that the transmittance of the sensor at a wavelength of 550 nm monotonically increases versus the biaxial stretching strain. The transmittance increases slightly from 0.1% at pristine state to 8% at 40% strain, and then significantly increases up to 45% at 43% strain. The inset shows that the image of the colorful picture behind the sensor looks fuzzy at 20% strain due to the low transmittance. The transmittance can continue to grow from 59% at 53% strain up to 88% at 100% strain. The insets in these states display that the colorful picture behind the sensor can be observed, suggesting the high transmittance of the sensor. It is worth noting that, as mentioned above, the strain sensor shows a high sensitivity of 569 at 100% strain at the same time. This means that the high transmittance of the strain sensor is indeed achieved simultaneously with high stretchability and high sensitivity.
Since ITO-based strain sensors show high sensitivity at the large biaxial stretching strain, they can be used as a stretchable and wearable device for motion monitoring. Here, we apply the strain sensor to obtain physical signals of human motions. This device can monitor the motions such as stretching, twisting, and bending at different parts of the human body. For example, chewing motions are detected by attaching an ITO-based strain sensor to the face (Figure 5a). Facial tissue involves numerous micromovements and deformations (≈20% strain) across the face during chewing. [40] The resistance response (ΔR/R 0 ) to strain shows a relatively stable signal as the mouth is closed. ΔR/R 0 significantly increases when the mouth is opened and it reduces as the mouth is closed. In fact, the sensor is stretched at the "mouth open" state and this stretching is relaxed at the "mouth close" state. Thus, the sensor can monitor the stretching-relaxing motions of facial muscles in the chewing process.
Similarly, if placed onto the throat, an ITO-based strain sensor can also be used to monitor the twisting motion of the neck when one turns the head around. The strain sensor would be twisted as the head is turned around (Figure 5b). The ITO-based strain sensor can not only detect the single-mode motions mentioned above but also monitor complicated human motions. For example, this device attached to the throat can monitor the whole process of swallowing by detecting throat deformations (Figure 5c).
The ITO-based strain sensor can not only detect the small strain range motions mentioned above but also monitor large strain of human motions. ITO-based strain sensors can detect the whole process of raising the arm (Figure 5d) and elbow bending (Figure 5e). Similarly, when the volunteer squats down, the strain sensor bends with the knee so that the crumpled patterns are unfolded while the sensor recovers to the buckled state when he stands up (Figure 5f). In a word, these results indicate that ITO-based strain sensors work well in detecting physical signals of human motions and have the potential for wearable electronics applications.

Conclusions
In summary, we developed a new stretchable piezoresistive strain sensor with 2D surface buckled patterns by coating ITO film onto the pre-stretched VHB elastomer substrate. It is verified that the crumpled surface is composed of localized ridges and wrinkles structures. To our knowledge, semiconductor such as ITO is applied for stretchable piezoresistive strain sensor with surface buckling for the first time. Furthermore, the ITO-based strain sensor exhibits the highest sensitivity of 610 compared with the previously reported sensors with similar structures. The high sensitivity of ITO-based strain sensors can be attributed to the piezoresistive effect of semiconductors where the carrier mobility and the resulting resistivity are significantly influenced by the strain. This device exhibits a sensitivity of 569 and a transparency of 88% at the biaxial stretching strain of 110%, simultaneously achieving high sensitivity, stretchability, and transparency. Besides, ITO-based sensors do not dramatically degrade in the 500 stretching-relaxing cycles, indicating long-term stability. In addition, the operation mode of biaxial stretching is applied for the first time to the piezoresistive strain sensor with surface buckling structures. Finally, ITO-based strain sensors are used to monitor human motions such as chewing, swallowing, breathing, and walking. These results show the great potential of ITO-based strain sensors for wearable and stretchable electronics.

Experimental Section
Preparation of Strain Sensor: First of all, the elastomer substrate of polyacrylate (3M VHB 4905) was biaxially pre-stretched at 300% strain. Second, a 15 nm thick ITO film was deposited onto the VHB substrate by magnetron sputtering at room temperature. Next, the ITO-buckled structure was formed when the biaxial pre-stretched substrate was released. Finally, the copper tapes were attached on the edge of the round-shaped strain sensor to act as the electrodes while liquid metal was painted to connect the ITO film and copper tapes.
Sample Characterization: The surface morphologies of the ITO-based strain sensors were characterized by using an SEM (OXFOORD, X-Max20). 3D images of surface morphologies were measured using LCM (KEYENCE, VK-X1000).
Evaluation of the Sensing Performance of the Strain Sensor: The sensor performance was evaluated in an ambient environment. The stretching and relaxing processes of strain sensors were performed by using a custom-made stretching machine. The electrical resistance of strain sensors was measured by Digit Multimeters (Agilent 34410A). The direct transmittance of strain sensors was tested by UV-vis spectrophotometer in the range between 300 and 900 nm.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.