Robust and stretchable indium gallium zinc oxide-based electronic textiles formed by cilia-assisted transfer printing

Electronic textile (e-textile) allows for high-end wearable electronic devices that provide easy access for carrying, handling and using. However, the related technology does not seem to be mature because the woven fabric hampers not only the device fabrication process directly on the complex surface but also the transfer printing of ultrathin planar electronic devices. Here we report an indirect method that enables conformal wrapping of surface with arbitrary yet complex shapes. Artificial cilia are introduced in the periphery of electronic devices as adhesive elements. The cilia also play an important role in confining a small amount of glue and damping mechanical stress to maintain robust electronic performance under mechanical deformation. The example of electronic applications depicts the feasibility of cilia for ‘stick-&-play' systems, which provide electronic functions by transfer printing on unconventional complex surfaces.

is torn more easily during the peel test than the thicker cilia (thickness = 1.4 or 3.5 µm). As a result of the mixed effect of the cilia thickness on the flexibility and mechanical strength, we believe that the optimal thickness for maximum overall adhesion is approximately 1.4 µm.

Supplementary Note 3: SEM images after detachment test
Supplementary Figure 9a shows photographs of the substrates after the detachment test using an air gun. The samples with cilia were detached at a higher air pressure. In the case of only cilia with no glue, most of the cilia were unwrapped from the threads upon intensive airblowing over the critical pressure (Supplementary Figure 9b). On the other hand, when using both cilia and glue prepared with a PDMS precursor solutions of not less than 3.3 wt.%, the cilia were torn under intensive air-blowing (Supplementary Figure 9c). These results suggest that the realistic adhesion between the cilia and textile is much higher than the measured critical pressure when using glue.

Supplementary Note 4: The origins of adhesion between textile and cilia
The origin of the increase in Pc should be considered in terms of physical, chemical, geometrical and mechanical aspects. The first concerns the gravimetric loading effect of the ultrathin PI substrate. For all the specimens used in Fig. 2a  surface, which can be determined by the thickness, surface roughness, length and density of the cilia. Assuming that the ideal plane surface of a cilium wraps a cylinder with radius (R), the critical adhesion energy per unit area (γc), which is the minimum energy per unit area required for conformal wrapping, can be calculated ( Supplementary Fig. S12 and S13) 4 . The γc becomes 0.28, 60, 280 and 950 mJ/m 2 for 300 nm, 1.8 µm, 3 µm and 4.5 µm thick cilia, respectively, which shows that thinner cilia result in more efficient wrapping behaviour and thereby larger contact area. When the surface of PI cilia is treated with O2 plasma, it becomes rougher, which generally induces a smaller contact area 2 ; the roughness factor changes from 1 nm to 67 nm after treatment for 7 min (Supplementary Figures 14 and 15). With the same thickness and surface roughness of cilia, which can be provided under the same RIE conditions for the surface treatment, longer and denser cilia induce more conformal contact area near the threads, as shown by the SEM images in Fig. 2a and 2b. The fourth relates to the residual stress of the samples generated during the fabrication process. When PI was treated with RIE and the supportive layer washed away, the free-standing film becomes concave in shape (Supplementary Figure S16), which is consistent with results previously reported in the literature using a PI cantilever tip 5 , suggesting that the bending occurs by the applied compressive stress for the bottom surface and tensile stress for the top surface of the PI film in the fabrication process. This phenomenon lowers the value of γc. Friction during the detaching process also contributes to the increase in Pc because the detaching directions are not always normal to the cilia surfaces due to the turbulence of air-blowing and complex geometry of the cilia and the textile 6 . In the case of using PDMS glue, we believe the additional elastic layer near the cilia not only enhances the interfacial adhesion between the cilia and the textile but also protects the cilia mechanically, thereby enhancing the value of Pc; the larger amount of residual glue reinforces the cilia.

Supplementary Note 5: Surface analysis of PI film with and without O 2 plasma treatment
We investigated the surface of the PI film before and after O2 plasma treatment by X-ray photoelectron spectroscopy (XPS). Fig. S11a shows the XPS survey scan spectra of the PI film with and without O2 plasma treatment. The O2 plasma increased the O1s/C1s peak ratio from 0.5 to 1.7, suggesting increased oxygen components. In the C1s XPS spectra, the peak at 286.3 eV attributed to C-O-C or C-OH increased sharply after O2 plasma treatment ( Supplementary Figures 11b and 11c). In particular, the peaks were much broader after O2 plasma treatment, indicating increased full width at half maximum (FWHM) values. These results should be induced by the binding energy shift due to the effect of the adjacent structures and were also observed in other literature 7 .

Supplementary Note 6: Calculated critical adhesion energy per unit area for conformal wrapping of cilia
We assume a thin film on a cylinder with a radius of R, as shown in Supplementary Figure   12a. For a thin film, the bending stiffness of homogeneous thin film was represented as where γc is the critical adhesion energy per unit area, which is the minimum energy required for the conformal wrapping of a thin film. At a fixed cilia width = 10 µm, γc was calculated for cases with different thicknesses of cilia and cylinder radii, and the values were compared to the experimental results ( Supplementary Figures 5 and 12c). We estimated the γ required We also assume a thin film on two overlapping cylinders with a radius of R to simplify the complex surface, as shown in Supplementary Figure 13a  When we transfer a thin film onto a complex substrate, the required adhesion energy is much higher than in the case shown in Supplementary Figure 12, as the rough morphology limits the conformal contact, requiring an extremely small bending radius of the thin film.
Supplementary Figures 13b and 13c show two cases for cylinders with R = 150 or 10 µm.
The results show that a rough surface (with smaller R) requires much higher adhesion energy for the conformal wrapping of thin film. Therefore, we introduced glue for conformal contact with the complex surfaces to enhance the adhesion energy, as shown in Figure 3.

Supplementary Note 8: Numerical mechanical modelling
To investigate the damping behavior of the peripheral cilia on the structure of serpentine electrode with textile under mechanical deformation, FEM analysis was performed using a commercial ANSYS program. FEM modeling was constructed on the base of the SEM images, which is illustrated in Supplementary Figures 23 and 24. The textile has plain weave shape and is modeled as the mesoscopic scale type. The approximate shape, size and spacing of the thread were obtained by analyzing the SEM images. Based on the tensile direction of textile, the specific area of the electrodes, which has horizontal and vertical orientation, was modeled. Anisotropic properties of thread was derived from an ideal thread model of continuous fibers and calculated from the thread constitutive equation 8 . Material properties data used in the study are described in Supplementary Table 2. All materials except thread were regarded as elastic and isotropic materials. A frictionless contact was defined for threadthread, thread-electrode and thread-cilia, and the attached area of cilia on thread was assumed for perfect bonding state. The left edge of the textile was constrained for the x, y, and z axis, and the right edge of the textile was constrained for the y and z axis respectively. The load displacement of the x-axis direction was applied on the right edge of the textile with a 9% tensile strain. The structure was modeled with the SOLID185 element which has 8 nodes, and it was assumed that there is no deformation by the residual stress.

Supplementary Note 9: Heat dissipation test
Supplementary Figures 28a and 28b show the temperature distribution of the electrode at DC 15 V with no strain and with tensile strain ε = 15.3%. The maximum temperature for the stretched electrode (~50 °C) was higher than for the non-stretched electrode (~45 °C). We believe that the increased resistance caused by tensile strain, as observed in the results shown in Fig. 5, induces the higher maximum temperature.
Heat dissipation is an issue in flexible electronics 9 ; it is desirable to use a minimal amount of polymer with comparably low thermal conductivity (κ) and high specific heat capacity (c) to allow efficient air-cooling. We investigated the cooling behaviour of the encapsulated electrode on a textile with PDMS glue (PDMSg) and PDMS film (PDMSf) with the application of air blowing by fans at different speeds. The maximum temperature distribution was observed upon applying DC 15V, followed by removing the voltage source, as shown in Supplementary Figures 28c and 28d. From the cooling curves, the time constant values, τ1 and τ2, and thermal decay time, t, required to reach 80% at maximum temperature (48-51 °C) for the samples were extracted as shown in Supplementary Table 4. In a lumped heat capacity system, assuming that the temperature of the whole body is constant at any point and changes where is the density, cp is the specific heat capacity, V is the volume of the metal body, A is the surface area of the body, and h is the heat transfer coefficient between the body and medium. According to Supplementary equation 7, as τ is proportional to the mass (m = ρ⋅V) and c p , using a small amount of PDMS with low thermal conductivity (κ) and high c p should induce faster thermal decay (shorter τ). The κ and cp values for the materials are shown in