Ultrathin, Graphene‐in‐Polyimide Strain Sensor via Laser‐Induced Interfacial Ablation of Polyimide

Laser‐induced graphene sensors have attracted considerable interest in various fields; however, the low sensitivity and conformability limit their further applications in measuring soft, large deformable structures. Here, an innovative method of interface ablation is presented to convert the interfacial polyimide into graphene by nanosecond ultraviolet laser (308 nm). Significantly different from the traditional laser surface ablation, interface ablation demonstrates its unique capacity to produce high‐quality graphene with limited ablation depth, which benefits from the combined effect of highly concentrated temperature distribution, the confinement of reaction product, and a unique ablation mode dominated by heat conduction. Using this method, an ultrathin (8 µm), graphene‐in‐polyimide (GiP) strain sensor is obtained, which is six times thinner than that prepared by the traditional surface ablation. The ultrathin GiP sensors exhibit excellent conformability (small bending radius of 400 µm), high strain sensitivity (24.8), and high force sensitivity (4.2 N−1). Demonstrations of this GiP strain sensor in the deformation measurement of the morphing aircraft (e.g., bending, twisting, and impact) illustrate its powerful abilities in the health monitoring of equipment, thus providing engineering opportunities for smart devices requiring accurate deformation measurement.


Introduction
In the era of intelligence, the ability to collect internal and external information is a fundamental requirement for advanced devices to enable themselves to interact with the outside world, [1] such as the electronic skin of intelligent robotics [2] and the smart skin of morphing aircraft. [3] Accurately and real-time monitoring of the deformation state of these intelligent devices often relies on strain sensors with ultrathin and high sensitivity. [4] Specifically, for thin objects (e.g., small soft robots [5] ), the thickness of the strain sensor has a significant effect on the deformation of the original structure. [6] As for large objects (e.g., morphing aircraft [7] ), the thickness of the sensor affects its conformability [8] and sensitivity as well as the measurement accuracy. [9] So far, although strain sensors based on metalsensitive layers and flexible substrate with a thickness of 10-100 µm (e.g., resin and polyimide [10] ) have been widely used in traditional consumer electronics (e.g., pulse wave monitor [11] ) and health monitoring of equipment (e.g., smart skin of the aircraft [12] ), the small fracture strain, low sensitivity, and large thickness limit their further applications. [13] Besides, these strain sensors are usually prepared by complicated lithography technology with a high cost. [10] In recent years, a new method to fabricate sensors by lasercarbonizing polymers into conductive graphene through surface ablation [14] has been widely used for human-computer interaction, [15] health monitoring, [16] energy storage, [17] and mechanical sensing [18] due to its facile preparation process [19] and large measuring range. [20] Using this method, various devices can be easily prepared, such as single-function or multifunctional sensors. [21] Many researchers have proved that a thick polyimide substrate (≈100 µm) is required for the fabrication of graphene strain sensors by traditional carbon dioxide laser. [22] To prepare a thinner strain sensor, one approach is using a laser with a shorter wavelength, such as 450, [23] 405, [24] and 355 nm. [25] For example, a 355-nm ultraviolet (UV) laser was used to prepare graphene strain sensors on a 50-µm polyimide substrate. However, the exfoliate phenomenon of as-formed graphene during surface irradiation results in the further ablation of the underlying polyimide, which limits the fabrication of a thinner graphene strain sensor. Another approach is to transfer the Laser-induced graphene sensors have attracted considerable interest in various fields; however, the low sensitivity and conformability limit their further applications in measuring soft, large deformable structures. Here, an innovative method of interface ablation is presented to convert the interfacial polyimide into graphene by nanosecond ultraviolet laser (308 nm). Significantly different from the traditional laser surface ablation, interface ablation demonstrates its unique capacity to produce high-quality graphene with limited ablation depth, which benefits from the combined effect of highly concentrated temperature distribution, the confinement of reaction product, and a unique ablation mode dominated by heat conduction. Using this method, an ultrathin (8 µm), graphene-in-polyimide (GiP) strain sensor is obtained, which is six times thinner than that prepared by the traditional surface ablation. The ultrathin GiP sensors exhibit excellent conformability (small bending radius of 400 µm), high strain sensitivity (24.8), and high force sensitivity (4.2 N −1 ). Demonstrations of this GiP strain sensor in the deformation measurement of the morphing aircraft (e.g., bending, twisting, and impact) illustrate its powerful abilities in the health monitoring of equipment, thus providing engineering opportunities for smart devices requiring accurate deformation measurement.
laser-induced graphene onto other ultrathin flexible sticky substrates, such as polydimethylsiloxane (PDMS). [26] Although this approach can improve the sensitivity to some extent, it loses the excellent performance of the polyimide substrate in chemical, thermal, and mechanical stability, which prevents it from being used in harsh environments. [8] Additionally, the transfer process increases the cost and time to prepare strain sensors. [27] Therefore, it is still necessary to develop a new method to fabricate graphene strain sensors with the features of ultrathin, high sensitivity, high stability, and broad practicability.
In this paper, we present a novel method of interface ablation to fabricate ultrathin (8 µm), graphene-in-polyimide (GiP) strain sensors, which benefit from the unique ablation reaction occurring at the interface between the transparent substrate and polyimide. Using this method, a sixfold decrease in the thickness of the polyimide substrate is achieved. Thanks to the unique advantages of interface ablation in terms of highly concentrated temperature distribution, the confinement of reaction product, and a unique ablation mode dominated by heat conduction, the conductive graphene can be produced with limited ablation depth (3.5 µm), and embedded in the polyimide substrate.
The prepared GiP strain sensor shows excellent conformability (small bending radius of 400 µm), high strain sensitivity (≈24.8), high force sensitivity (≈4.2 N −1 ), quick response times (50 ms for response and 60 ms for recovery), and high stability in the cycle tests over 1000 cycles (1% strain). These excellent performances have been well demonstrated in the deformation monitoring of large deformable structures, including morphing aircraft. Figure 1a shows the fabrication process of the ultrathin, GiP strain sensor. To enable interface ablation, a pulsed UV laser www.advelectronicmat.de with a wavelength of 308 nm [28] and a two-layer sample with laser-transparent quartz glass and light-absorbing polyimide ( Figure S1, Supporting Information) are selected. [29] First, a polyimide layer with a thickness of 8 µm is deposited on the surface of quartz glass. Then, a predesigned mask is pasted onto the surface of quartz glass to achieve patterned graphene by high laser flux. In this process, the polyimide at the interface absorbs sufficient energy to completely decompose into conductive graphene with an ablation depth of 3.5 µm (Figure 1b). Next, the sample is irradiated with a low laser flux to exfoliate the polyimide film from the quartz glass after removing the mask. During this process, the interfacial polyimide undergoes insufficient decomposition with an ablation depth of 0.25 µm (Figure 1c). Finally, the generated gas products promote the underlying unablated polyimide away from the quartz glass to achieve a freestanding ultrathin sensor array, as exhibited in Figure 1d.

Characterization of Carbonization Product
The laser flux absorbed by the interfacial polyimide is determined by laser energy density and accumulated pulse number (APN), as exhibited in Figure 2a (the shade of the color represents the amount of the absorbed laser energy). The detailed definition of APN [30] is illustrated in Figure 2b. Three different laser fluxes are used, i.e., the low laser flux (100 mJ cm −2 , APN = 10), medium laser flux (130 mJ cm −2 , APN = 100), and high laser flux (180 mJ cm −2 , APN = 1500). For low laser flux, polyimide could be peeled off from the quartz glass substrate, showing no visible color change ( Figure S2a, Supporting Information). The SEM-magnified image shows that there only exists a small number of protruding structures in the local area ( Figure 2c). As the laser flux increases, the visual color change from yellow to black, accompanied by the increased size and density of the protruding structures. Compared with surface ablation, the morphology of interface ablation is more uniform at high laser flux.
The ablation depths of the above three laser fluxes are also investigated. As the laser flux increases, the ablation depths increase for both two ablation modes, but the differences between them are more pronounced (Figure 2d). At low laser flux, the interface ablation depth is 0.25 µm, which is approximately half of the surface ablation depth of 0.51 µm. Under the condition of high laser flux, the surface ablation depth is 11.5 times larger than that of interface ablation. Although the ablation depth of interface ablation is limited, the transition from the nonconductive film into conductive electrodes can be also achieved by the graphitization of interfacial polyimide at medium laser flux (square resistance of 890 kΩ sq −1 ) and high laser flux (square resistance of 13 kΩ sq −1 ) ( Figure S2b, Supporting Information). This square resistance is larger than traditional surface ablation of carbon dioxide lasers, which is mainly caused by three reasons. First, the coupled photothermal and photochemical effects of the 308-nm laser lower the ablation temperature compared with the pure photothermal effect of the 10.6 µm laser. [31] Second, the smaller light penetration depth (≈100 nm for the 308 nm laser) allows very thin polyimide layers to be carbonized by the laser, while the carboniza-tion of the lower layer of polyimide depends on heat conduction. This feature results in a nonuniform quality of the carbonization product, with the upper layer having a higher quality than the lower layer. Finally, the low frequency (maximum frequency of 20 Hz) and narrow pulse width (20 ns) of the 308-nm laser results in heat preservation time (the heat generated by the previous pulse is largely dissipated before the next pulse) much less than that of kilohertz or megahertz lasers. [25b] Raman spectroscopy that can accurately detect the classes of carbon [32,33] is used to identify the specific species of the carbonization product at the three different laser fluxes mentioned above. As shown in Figure 2e, the polyimide after the irradiation with low laser flux shows no obvious peaks, which is consistent with the original polyimide. However, the polyimide is converted into the amorphous carbon with D-peak (≈1350 cm −1 ) and G-peak (≈1580 cm −1 ) after the irradiation with medium laser flux, and into the graphene with D-peak, G-peak, and 2D-peak (≈2700 cm −1 ) after the irradiation with high laser flux. Besides, the I D /I G (defect ratio) of the carbonization product at high laser flux is smaller than that at medium laser flux, indicating a higher quality of the carbonization product. At high laser flux, X-ray diffractometer (XRD) results show that the peak of (002) appears at 25.9°, indicating the more ordered carbonization product with the interplane lattice space of ≈3.4 Å (Figure 2f). Similar results can be obtained by X-ray photoelectron spectroscopy (XPS) analysis. The peak width of C (sp 2 ) and the impurity content of COC and COC are smaller at high laser flux, suggesting the carbonization product is purer with the increase of laser flux ( Figure S3, Supporting Information).
The process window for exfoliation, amorphous carbon and graphene has been obtained by batch experiments of the carbonization product analysis, as illustrated in Figure 2g. It can be found that clear exfoliation without carbonization can only be achieved under conditions of low laser flux. To generate graphene, suitable laser energy density is the prerequisite, and sufficient APN is also required. Besides, the laser energy density and APN affects the quality of graphene. The I D /I G (defect ratio) gradually decreases with increasing laser energy density at the same APN of 1500. At a given laser energy density of 180 mJ cm −2 , the I D /I G gradually decreases to a stable value with the increase of APN ( Figure S4, Supporting Information), suggesting that a certain amount of defect remains even at high APN. It is noteworthy that the I D /I G of 0.4 is comparable with the laser-induced graphene fabricated by the traditional method, e.g., surface ablation by a common carbon dioxide laser. [14]

Mechanism Analysis of Interface Ablation
According to the above results, the interfacial polyimide after laser irradiation exists mainly in three forms: no obvious visual variations, amorphous carbon, and graphene. Especially, conductive graphene can be also produced even at a limited ablation depth. To better understand these phenomena, a photothermal model is established to investigate the temperature distribution during laser irradiation (see the Experimental Section for details). The thinner ablation depth of interface ablation results from the following three key mechanisms. First, the heat conduction of the quarts glass substrate results in a more www.advelectronicmat.de uniform and localized temperature distribution for interface ablation. As seen in Figure 3a, the maximum temperature of interface ablation is not located at the interface between quartz glass and polyimide, but in the polyimide layer about 50 nm from the interface due to the excellent thermal conductivity of the upper quartz glass (Figure 3b), while the maximum temperature of surface ablation is right at the irradiation surface. The maximum temperature of surface ablation is 2.9 times that of interface ablation. Second, the smaller temperature stress and the confinement of the quartz glass avoid the exfoliation of the carbonization product. Figure 3c exhibits the temperature stress for interface ablation and surface ablation. As can be seen, the temperature stress between layers is lower and more uniform for interface ablation, which is beneficial to reduce the exfoliation of the as-formed carbonization product. Besides, due to the confinement of the upper quartz, the exfoliation of the carbonization product can be further avoided. Finally, for interface ablation, a large portion of the carbonization www.advelectronicmat.de product is generated by heat conduction, rather than direct laser irradiation, which leads to a smaller ablation depth. As illustrated in Figure 3d, compared with surface ablation, the interface ablation depth is much smaller, especially for higher APN. The ablation depth increases sharply at the early stage and then goes into a slow growth process. Unlike the surface ablation process where the carbonization product will exfoliate, the carbonization product of the previous interface ablation will www.advelectronicmat.de be collected to absorb the subsequent laser energy. Thus, the underlying polyimide could only be decomposed through heat transfer from the top carbonization product. This unique ablation mode reduces the ablation depth.
During laser irradiation, the maximum temperature of the polyimide varies with the laser energy density (Figure 3e). Compared with surface ablation, the temperature of interface ablation is much lower, e.g., a 2.9-fold reduction of the maximum temperature for interface ablation (3000 K) at 150 mJ cm −2 . Considering the lower temperature of interface ablation, the decomposition states of polyimide molecules are investigated using the reactive molecular dynamics method. The results show that the interfacial polyimide decomposes mainly into carbon clusters and some small molecules, such as water, CO, H 2 , and N 2 ( Figure S5, Supporting Information), and the specific composition varies with reaction temperature (Figure 3f). Specifically, the carbon clusters exist mainly in the planar graphene hexagons at 3000 K, indicating that the graphene can still be produced even if the interface ablation temperature is much lower than that of surface ablation. As the temperature decreases to 2500 K, most of the carbon atoms exist in the form of amorphous carbon chains. By further decreasing the temperature to 2000 K, only a small amount of carbon chains appear, while the benzene rings of the polyimide molecule still exist, suggesting that the exfoliation of polyimide with no obvious visual variations from the quartz glass substrate can be achieved by selecting the proper process parameters.
For interface ablation, the carbonization product will undergo repeated irradiation. [34] Four different reaction times (0.5, 1, 2, and 8 ns) are used to investigate this feature on the carbonization products, while the temperature is fixed at 3000 K ( Figure S6, Supporting Information). As illustrated in Figure 3g, for the reaction time of 0.5 ns, the carbon clusters exist mainly in the form of carbon chains. As the reaction time increases, these carbon chains change into planar graphene hexagons (8 ns), indicating that the interface ablation of repeated irradiation helps improve the quality of carbonized products.

Characterization of GiP Strain Sensor
Considering the graphene quality and processing efficiency, a laser energy density of 180 mJ cm −2 and an APN of 1500 are selected to fabricate GiP strain sensors with four different thicknesses (100, 20, 8, and 5 µm). Meanwhile, a commercial strain sensor with a thickness of 50 µm is used for comparison (see the Experimental Section for details). The results demonstrate that the strain sensitivities of the GiP strain sensors are similar, which are more than 30 times higher than that of the commercial strain sensor (Figure 4a). Table S2 (Supporting Information) exhibits the sensitivity and maximum strain of the strain sensors prepared by different methods. Although graphene (or CNT)/PDMS strain sensors show larger sensitivity and linear strain range, the procedure and fabrication cost increase. Especially, they are not suitable for harsh environments. Compared with the reported graphene/polyimide strain sensors with excellent stability, the significant reduction in substrate thickness does not affect the strain sensitivity. As for the commercial metal/polyimide strain sensor, the resistance-strain relationship is no longer linear once the strain is greater than 1.0% (Figure 4b). In contrast, the resistance variation rate of the GiP strain sensors (thicknesses of 100, 20, and 8 µm) increases linearly with increasing strain, even when the strain reaches 2% (Figure 4a). Once strain exceeds this value, this linear relationship becomes nonlinear because the deformation of polyimide enters into the plastic stage ( Figure S7c,d, Supporting Information). As the sensor thickness reduces from 100 to 8 µm, the fracture strain of the GiP sensor decreases from 15.4% to 3.2%. Particularly, the 5-µm thick sensor lost usability due to its very small fracture strain (no longer linear at 0.38% and breaks at 0.47%), which is due to the vulnerability of the underlying polyimide substrate.
When the thickness of the sensor is reduced from 100 to 8 µm, the cross-sectional area of the sensor reduces by 12.5 times with the fixed width of 30 mm, causing the force sensitivity to increase from 0.3 to 4.2 N −1 (Figure 4c). This feature enables the ultrathin GiP strain sensor to detect the tiny deformation caused by low force, such as tiny object impact and pulse waves that are difficult to measure with a commercial strain sensor or thick GiP strain sensor. [22a] In addition, the ultrathin strain sensor exhibits fast response times (50 ms for response and 60 ms for recovery in Figure 4d) and good cycling stability (≈1000 cycles at 1% strain in Figure 4e,f).
The rational surface wettability design of the skin-interfaced sensors contributes to the stability and integrity of the devices. Generally, hydrophobic properties are necessary for strain sensors that measure the deformation of the equipment. So far, several methods have been developed to adjust the hydrophilic/ hydrophobic properties of the devices, including surface modification and the creation of micro-/nanostructures to achieve wettability patterns. [35] For hydrophilic laser-induced graphene prepared in the atmosphere, oxygen-free processing environments (reduction of hydrophilic functional groups) [36] and surface modification (addition of hydrophobic functional groups) [21a] schemes have been widely adopted to tune it to hydrophobicity. The proposed interface ablation is a closed and oxygen-free processing environment, thus enabling the hydrophobic property of the GiP ( Figure S8a, Supporting Information). Thanks to this property, the resistance variation rate of the GiP sensor is negligible (<0.25%) when relative humidity changes from 25% to 90% ( Figure S8b, Supporting Information).
The design of rough surfaces has been wildly adopted in the fabrication of aircraft to reduce the icing effect. Generally, the roughness height ranges from 0.28 to 0.79 mm. [37] As a typical example, we select an object with a radius of 400 µm to show the excellent conformability of the ultrathin GiP sensor. To show this feature, polyimide films with three thicknesses of 100, 20, and 8 µm are attached to a curved surface with a radius of 400 µm, respectively. As seen in Figure 4g, for a 100-µm thick polyimide film, there is a large gap between the polyimide and the curved surface. As the thickness reduces to 20 µm, a polyimide film can bend to a certain extent, resulting in a smaller gap. Especially, for the 8-µm polyimide film, it can achieve conformal attachment. This favorable feature enables the sensor with good conformability to measurement areas with localized www.advelectronicmat.de large curvature, such as robot arm joints and human-wrinkled skin. Moreover, the thinner the strain sensor, the smaller the influence on the measured object. Particularly, paper-like electronics with a thickness of fewer than 300 µm (e.g., flexible displays) require an ultrathin strain sensor (<10 µm) to reduce the strain measurement error. For instance, for electronic devices with a length of 100 mm and a thickness of 200 µm, the strain measurement error decreases from 100% to 12% with the ratio of h object /h sensor increasing from 2 to 25 due to the less offset of the neutral layers (Figure 4h), which has been investigated in our previous work in detail. [10]

Demonstration of GiP Strain Sensor Array on a Morphing Aircraft
One specific case where high sensitivity, excellent working stability, and good conformability ( Figure S10b

www.advelectronicmat.de
According to the fluid analysis in Figure S9 (Supporting Information), a thinner sensor is necessary to reduce the influence of the attached sensor on the surrounding flow field. As proof of health monitoring of the large devices, a 3 × 3 ultrathin sensor array is fabricated and attached to the left half region of a morphing aircraft for monitoring the deformation during bending, torsion, and impact, as shown in Figure 5a,b. For simplicity, S ij (i,j denotes the row and column of the strain sensor, respectively) is hereafter used to represent the corresponding strain sensor.
For the bending of the wing (Figure 5c), the resistance variation rate of the strain sensor in the row direction increases linearly with the bending angle from 0° to 7.5° (Figure 5d). Especially, the resistance variation rate of S 31 near the wing root (1.1%) is greater than that of S 33 near the wing front (0.15%), mainly because the wing root is constrained while the wing front is free to be deformed. In addition, since the deformation in the column direction is the same during bending, the resistance variation rate of strain sensors in the same column is almost the same, such as for sensors in the first column (in Figure S10d, Supporting Information). Replacing S 33 with a commercial strain sensor (50 µm), the resistance variation rate of the GiP strain sensor is much larger (Figure 5e), which is beneficial to improve the measurement accuracy.
For torsion, the hydraulic module at the front of the wing drives this area to produce tensile deformation while the root of the wing remains stationary (Figure 5f). The variation rate of resistance increases linearly with the twisting angle (Figure 5g). Selecting three sensors in a row (S 13 , S 23, and S 33 ) at the front of the wing for comparison, it can be easily found that the resistance variation rate of S 33 (0.53%) is the largest, while the variation of S 31 is almost negligible (0.02% in Figure S10e, Supporting Information), indicating that the strain of the wing has symmetrical distribution along the middle line during torsion. Similarly, the resistance variation rate of the GiP strain sensor is much larger than that of the commercial strain sensor (50 µm) under the same condition (twisting angle of 5° at the attached position of S 33 , shown in Figure 5h).
During the flight of the morphing aircraft, detecting the object impact has important significance, which has been carried out by a block (5 g) falling test from a height of 0.1 m (Figure 5i). The results show that the GiP strain sensor near the impact position has a good response to the impact (Figure 5j), while under the same circumstances, the commercial strain sensor (50 µm) shows no variation (Figure 5k). This is due to the much higher force sensitivity of the ultrathin sensor.
The decoupling of multimodal sensors is generally challenging but important when two signals are input. At present, several strategies have been developed to decouple multiple inputs, including structure design, various sensing units, one sensing unit facilitated by the selection of materials with various principles, different response times, and combined materials and structures. [38] In this study, the mechanism of different response times can be used to decouple multiple inputs. The bending and impact can be decoupled when they are sequentially applied (e.g., the bending angle changes from 0° to 1.5° in Figure S10f, Supporting Information). Besides, when impacted during bending deformation, these two signals can be also decoupled due to their great difference in the time scale.
Specifically, the impact can be completed within ≈0.1 s, which is much faster than the normal wing deformation, i.e., it takes 0.38 s for 1.5° of deformation. The above studies demonstrate the promising applications of ultrathin strain sensors in the deformation monitoring of morphing aircraft.

Conclusions
In summary, an innovative method of interface ablation provides a scalable, flexible, and particularly efficient solution to prepare ultrathin (8 µm), GiP strain sensors through carbonization at high laser flux and exfoliation at low laser flux. Compared with traditional surface ablation, the thickness required for polyimide substrates can be reduced by six times. Mechanism analyses illustrate the unique advantages of interface ablation in terms of highly concentrated temperature distribution, collection of the reaction product, and thinner ablation depth by distinct ablation mode of heat conduction. Due to its smaller thickness, the prepared ultrathin strain sensor exhibits excellent conformability (small bending radius of 400 µm), high strain sensitivity (≈24.8), and high force sensitivity (4.2 N −1 ). Demonstration of these excellent performances is reflected in the monitoring of deformation (e.g., bending, twisting, and impact) of the morphing aircraft. In addition to strain sensors, the method of interface ablation can be also used to fabricate other GiP devices, including humidity sensors, temperature sensors, conductive polymer films, GiP circuits, etc.

Experimental Section
Fabrication of Two-Layer Sample: The polyimide solution, purchased from Beijing Bomi Technology Co., Ltd (China), was first spin-coated on the surface of the quartz glass with a speed of 2500 rpm min −1 and a time of 120 s. Then, the sample was placed on a hot plate at 130 °C and 90 s for precuring. Finally, it was placed in a furnace at 300 °C for imidization for 2 h to obtain the desired sample. Before imidization, the spin-coating and procuring could be repeated several times to obtain polyimide films with different thicknesses.
Fabrication of GiP Strain Sensor: To increase the stiffness of the ultrathin polyimide, a layer of commercial thermal release tapes (TRT) was first attached to the surface of the polyimide after the fabrication of the two-layer sample. Then, the GiP strain sensor was fabricated through high-laser flux carbonization and low-laser flux exfoliation. Finally, the peeled polyimide film was placed on a hot plate (120 °C) to remove the TRT tape. The commercial conductive silver paste (purchased from Shenzhen Zhenglan Industrial Co., Ltd., China) was used to connect the strain sensor with the conductive wire. The commercial strain sensor (BA 120-3AA, initial resistance of 120 Ω) was purchased from Shanghai Chengke Electronic Technology Co., Ltd. (China).
Sensitivity Test of GiP Strain Sensor and Deformation Monitoring of the Morphing Aircraft: The strain sensitivity and force sensitivity were tested by the tensile-testing machine (Instron 5944, USA). This equipment could obtain displacement and force at the same time. The morphing aircraft was designed and manufactured by Shenyang Hezhong Zhichuang Technology Co. Ltd. (China). The fabricated strain sensors were pasted on the surface of morphing aircraft with 3 m glue. The bending and twisting of the morphing aircraft were controlled by software.
Material Characterization and Testing: The absorbance of polyimide and quartz glass was measured by a UV spectrophotometer (Lambda 35, USA). A scanning electron microscope (SEM, SU8020, Hitachi, Japan) was used to study the morphologies of polyimide after laser www.advelectronicmat.de

www.advelectronicmat.de
irradiation with different laser fluxes and the contact degree of polyimide films with different thicknesses and curved surfaces. A high-resolution scanning electron microscope (SEM, FEI Helios G3, USA) was selected to investigate the ablation depth. Carbonization products of interface ablation at different laser fluxes were identified by using a Raman spectrometer with 532 nm (Renishaw inVia Reflex, UK). A multichannel digital external meter (Keysight 34460A, China) was used to measure the resistance variation rate under different strains and curvature radii. X-ray diffractometer (XRD-6100, Japan) was selected to test the crystal lattices of polyimide after interface ablation. By X-ray photoelectron spectrometer (AXIS-ULTRA DLD-600 W, Kratos, Japan), the elemental composition of polyimide after interface ablation at different laser fluxes could be obtained. The resistance variation rate of the GiP sensor in humid environments was tested by a constant temperature and humidity test chamber (Shenzhen Changxu Machinery Co., China). The contact angle of the GiP sensor was measured by the contact angle measurement instrument (JC2000D1, Shanghai Zhongchen Digital Technology Equipment Co., China). Height profile diagrams of the original surface and the surface with polyimide (attached to the wing surface with 3 m glue) were tested with a laser scanning confocal microscope (VK-X200K, Japan).
Multiphysics Temperature Simulation: The COMSOL software was used to obtain the temperature distribution of surface ablation and interface ablation. Two modules of radiation heat transfer and solid heat transfer were selected in this study. Specifically, the module of radiation heat transfer was used to analyze the interaction of laser and material, while the module of solid heat transfer was used for heat conduction. The material parameters of polyimide and quartz glass were obtained based on the previous literature, which was exhibited in detail in Table S1 (Supporting Information). A two-dimensional plane model with a size of 20 × 7 µm was selected with the thickness of quartz glass of 5 µm and polyimide of 2 µm. Previous studies had demonstrated that the polyimide could be carbonized once the temperature exceeds 873 K, [39] so this temperature was defined as the critical ablation condition in simulations. The polyimide above this temperature was considered to form carbon, so this region used the carbon material parameters in the analysis of the effect of APN on ablation depth. Considering that part of the carbon exfoliated during surface ablation, 20% of the carbon thickness was chosen to be retained in the simulation.
Reactive Molecular Dynamics Simulation: The LAMMPS software integrated with the ReaxFF potential that included four elements of C, H, O, and N [40] was used for molecular dynamics simulation. First, the molecule of polyimide (C 29 H 16 O 6 N 2 ) was established and geometrically optimized. Then, eight polyimide molecules were pressed into a cell with an initial density of 0.6 g cm −3 to avoid the overlap of the atom. Next, the above cell was compressed to 1.3 g cm −3 at 0.01 GPa using the NPT ensemble. Finally, it was decompressed at 0.1 MPa using the NPT ensemble to obtain the initial model for reactive molecular dynamics calculations. [41] In the calculation model, the periodic boundary conditions and NVT ensembled with a timestep of 0.25 femtosecond (fs) were selected, accompanied by the Nose-Hoover thermostat temperature control equation.

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