Integration of complex electronics into our everyday outfit is a key challenge for the tomorrow-wearable technologies. When performing data processing and communications, these circuits are an essential part of sensing and actuating smart systems. The choice of an appropriate electronic substrate is essential for robust and reliable, yet imperceptible systems. In this paper, we study and mechanically improve the conventional substrates of the flexible circuit board (FCB) industry for highly flexible and washable textile-integrated circuits. First, we study the mechanical limits of polymeric substrates to better apprehend how permanent damages occur during washing cycle. Measurements of plasticization threshold of our films under bending showed that bending radii below 1 mm occur during washing, and that 13 μm-thick polyethylene terephthalate (PET) film can be elastically deformed down to 0.4 mm of bending radius. Second, we focus on the electrical and mechanical behavior of a stack composed of a PET substrate, aluminum conductive tracks, and polydimethylsiloxane encapsulation. An analytical model of stacks under bending is correlated with experimental results to demonstrate the impact of the thicknesses of the different layers and to improve the reliability under washing. Using the same model, we investigate different materials for encapsulation to further optimize the mechanical reliability of FCB while decreasing its overall thickness.
Seamless integration of complex electronics into textile is a keystone for wearable technologies to become integral part of our lives. As for any other processing systems, complex circuitry with high computational performances is required to achieve multiple tasks, such as signal acquirement, processing, and transmission. This study particularly is part of the development of smart clothes for presymptomatic detection of cardiovascular diseases. Such wearable systems should fulfill two main characteristics. On the one hand, electronics should be imperceptible for the user, and therefore, they should not alter the textile’s natural wearability and comfort. This implies that the module shows compliancy, with a limited overall thickness. On the other hand, circuits must be robust enough to endure the stress our everyday outfits are subjected to, such as crumpling, especially during domestic washing cycle. The challenge consists in the concurrency between circuit’s complexity, imperceptibility, and mechanical robustness.
Researchers have explored different approaches to build textile electronics. Functionalized fibers1–8 meet the most advanced degree of integration, with transistors merged directly inside the core of the fibers. Textile mechanical properties are greatly preserved, but at the cost of computational capabilities, which limits current fiber-based electronics to basic logic circuits. In order to increase computational performances on textile substrates, rigid silicon-based integrated circuits are directly bonded to the surface of fabrics.9–12 However, due to the natural loose properties of textile substrates, misalignment and spatial resolution issues occur when integrating microelectronic components. On the contrary, flexible circuit boards (FCB) have the advantage to rely on the conventional industrial process of microelectronics and can be implemented in textiles.13–19 The resulting flexible devices achieve spatial resolution, component density, and computational performances similar to those of conventional printed circuit boards.
FCB are composed of a flexible substrate (most often a polymer film) and a functional layer containing tracks design (metallic layer). Rigid, conventional silicon-based integrated circuits (microcontrollers, Bluetooth communicating devices, etc.) are integrated on the stack to create a flexible electronic device. In this work, we focus on the flexible stack itself, without any integrated component.
Laminated films of aluminum on polyethylene-terephthalate (PET), and copper on polyimide (PI), are widely available and standard substrates for FCB. However, their mechanical properties are hardly sufficient for textile-integrated electronics. Indeed, applications in FCB industry so far (bus ticket, antitheft stickers, ribbon cable, etc.) imply limited stress and/or short lifetime. Textile electronics on the contrary are subjected to high and dynamic mechanical stress during daily domestic washing. Folding or creasing of polymer substrates is reported after conventional cleaning process.20 Cracks and delamination of the metallic tracks appear after washing and can lead to failure of the device.
Encapsulation of flexible electronics integrated into clothes is mandatory for two reasons. First, encapsulation protects electronics from moisture, chemicals, and water, to allow washing of the device. Second, it can improve the mechanical reliability of the device, for example, by preventing high deformations. Several encapsulation methods were explored to enhance mechanical performances of FCB under washing. Stiffening structures such as transfer molding or glob top encapsulation using epoxy resin9,11,12,14,21 offer good reliability through 20 washing cycles at 40 °C.22 Unfortunately, the resulting devices are no longer flexible. Soft silicone molding encapsulations, such as polydimethylsiloxane (PDMS), were investigated to improve mechanical behavior under washing, with a final thickness of a few millimeters.13,23 Whereas such encapsulations offer good chemical and mechanical protection, they diminish the compliancy of the module and can hinder the natural feeling of the garment.24 A tradeoff must be made between stack stiffness preventing high strain, and compliance required for user’s comfort.
In this work, we study how the conventional substrates of the FCB industry can be adapted for robust washable electronics, while preserving compliancy and thinness of the final substrate. In a first part, we focus on the polymeric substrates. Despite their high flexibility, one washing cycle is enough to induce permanent damages in the films. We study the films’ elastic limitations under bending with a customized setup allowing measurements of radius down to 100 μm.
In the second part, we focus on the electrical behavior of a stack composed of PET substrate, aluminum conductive tracks, and PDMS encapsulation. We demonstrate the impact of the thickness of the different layers on the mechanical reliability of the stack under washing. We show that our experimental results are consistent with an analytical model of stacks under bending.
We consider two kinds of deformations: non-permanent (elastic) deformations and permanent (plastic) deformations. It is important for flexible electronics to undergo only elastic deformations under use, as plastic deformations would shorten lifetime of the device and discomfort for the user. Table I shows different substrates with and without metallization, before and after washing cycle in domestic machine. For easier reading, films and stacks are identified by the material name followed by the thickness in microns: a stack composed of 18 μm-thick copper film on 50 μm-thick PI substrate is referred to as PI50/Cu18. Several small creases in PET50 and PI50 films, and coiling of PI50/Cu18 sample, can be visually observed after wash and show that plasticization occurred during washing. Whereas those materials show good flexibility and can be fully bent without fracture, one washing cycle is enough to induce permanent deformations in films.
Standard substrates of FCB industry. Samples dimensions are 20 mm × 50 mm. Damages induced by washing are visually noticeable: PET and PI films show multiple small creases, and metallized sheets have larger permanent folds.
Substrate’s thickness | PET 50 µm | PI 50 µm | PET 50 µm | PI 50 µm |
Metal’s thickness | … | … | Al 9 µm | Cu 18 µm |
Sample | PET50 | PI50 | PET50/Al9 | PI50/Cu18 |
Before wash | ![]() | ![]() | ![]() | ![]() |
After wash | ![]() | ![]() | ![]() | ![]() |
Substrate’s thickness | PET 50 µm | PI 50 µm | PET 50 µm | PI 50 µm |
Metal’s thickness | … | … | Al 9 µm | Cu 18 µm |
Sample | PET50 | PI50 | PET50/Al9 | PI50/Cu18 |
Before wash | ![]() | ![]() | ![]() | ![]() |
After wash | ![]() | ![]() | ![]() | ![]() |
To our knowledge, no standard test or modeling of strain implied in domestic cleaning machine is currently available. To evaluate and compare materials’ mechanical performances, we assume that most part of damages induced in flexible films is due to clothes’ crumpling during cleaning process, which translates to the films as flexion movements. Therefore, the elastic limits of materials under bending condition determine their mechanical reliability through washing cycles. The high flexibility of our materials presages radius thresholds in the millimeter range. In order to address this measurement, we built a setup using a uniaxial moving stage with side-view imaging of the sample under bending, as shown in Fig. 1(a). This setup allows minimum measurements of 100 µm, whereas more conventional rolling procedure is limited to few millimeters in radius.25–27 A first LabVIEW program synchronizes the moving stage with the camera pictures. A second LabVIEW program allows measurements of the bending radius from the pictures. Both programs are shown in the supplementary material in Fig. S1.
(a) Setup for the measurement of bending radius. The camera monitors the side view of the sample. The arrows on each sides of the sample indicate the bending movement. (b) Method to determine plasticization threshold of substrates under bending. Graph on the left depicts the radii of curvature through entire bending cycles of two samples. Green plot shows typical elastic deformation of the film (rfinal = rinitial) when bending down to rmin = 1.15 mm. Red plot translates plasticization of the substrate (rfinal < rinitial) after bending down to rmin = 0.48 mm. The inset of second sample shows permanent crease in the film. (c) Measurement of plasticization threshold of PI-based and (d) PET-based substrates under bending. Please note that scales are different between PI and PET samples, as plastic deformations occur faster in PI than in PET. (e) Theoretical plot of maximum strain in films under bending shows that thinner layer endures lower strain at same bending radius.
(a) Setup for the measurement of bending radius. The camera monitors the side view of the sample. The arrows on each sides of the sample indicate the bending movement. (b) Method to determine plasticization threshold of substrates under bending. Graph on the left depicts the radii of curvature through entire bending cycles of two samples. Green plot shows typical elastic deformation of the film (rfinal = rinitial) when bending down to rmin = 1.15 mm. Red plot translates plasticization of the substrate (rfinal < rinitial) after bending down to rmin = 0.48 mm. The inset of second sample shows permanent crease in the film. (c) Measurement of plasticization threshold of PI-based and (d) PET-based substrates under bending. Please note that scales are different between PI and PET samples, as plastic deformations occur faster in PI than in PET. (e) Theoretical plot of maximum strain in films under bending shows that thinner layer endures lower strain at same bending radius.
We measured the bending radius threshold for plasticization of the substrates as explained in Fig. 1(b). We define the plasticization ratio α as α = rfinal/rinitial, with rinitial and rfinal the natural, or stress-free, radii of curvature of the film, respectively, before and after bending. Samples are bent down to different radii of curvature, and the ratio α decreases when permanent deformations occur in the film. Figures 1(c) and 1(d) show α ratio for PET and PI films under bending. Each plot has two distinct regions. A first horizontal line translates typical elastic behavior where material returns to its initial state (rfinal = rinitial) after bending. The second region conveys plastic behavior (rfinal < rinitial) with decreasing α while increasing strain (decreasing bending radius). Transition between the two regions indicates the plasticization threshold of the material.
For PI50 and PET50 film, these thresholds are around 2.3 mm and 1.1 mm, respectively. Therefore, plastic deformations of PET50 foil in Table I suggest that radii of 1.1 mm or less apply to the flexible module under washing. Thinner layer of PET13 shows the best performance as it can be bent down to 0.4 mm without plastic deformations. Laminated polymer/metal stacks were also tested after the metal was completely removed by laser-engraving, to investigate the impact of the laser process on the polymer material. These films showed similar results than polymer substrates, which suggests that laser ablation does not alter the polymer’s mechanical properties upon folding. In comparison, mechanical characterization of plain polymer/metal laminated films revealed very low elasticity, with plasticization over 10 mm of curvature, as shown in Fig. S2 in supplementary material. To understand the differences in elastic domains, we used a model of strain implied in films under bending depicted in Fig. 1(e). For a layer bending downward, top surface is in tension whereas bottom surface is in compression. The neutral plane hb is the surface inside the film where the strain is null. Maximum tensile strain arises at the top surface of the film and depends on bending radius R and thickness of the film,
with R the radius of curvature of the neutral plane. In a uniform monolayer, the neutral axis lies in the middle of the film, and therefore,
Equation (2) implies that εtop at constant radius increases with the thickness of the film. This result is in correlation with the plots in Fig. 1(d), where PET13 can elastically withstand smaller radius than PET50. On the other hand, PI50 and PET50 are subjected to the same surface strain. The difference between their plasticization thresholds is consistent with that in their mechanical properties, as elastic limit is around 4% of strain for PET28 and 0.8% for PI.29 Those thresholds are reported in Fig. 1(e).
Because of its higher elastic limit, PET film can be bent to smaller radius while remaining in its elastic domain. Analytical model and experimental results show that reducing the substrate’s thickness widens the elastic domain under bending. Moreover, decreasing the thickness lowers the bending stiffness,30 which allows more conformable electronics. This conclusion is consistent with our ambition of thin and imperceptible electronics. In the following study, we focus on PET-based substrates because of their lower plasticization threshold.
Mechanical stresses during washing induce cracks and opening in the conductive tracks, causing failure of the device. In this part, we study the impact of the stack’s composition on electrical reliability of the functional layer under washing. Tested circuits were matrix composed of 800 μm-wide lines, wide enough for future integration of surface mounted components such as light-emitting diodes, and to apply probes for electrical measurements. The lines have 0°, 45°, and 90° orientations, which are standard geometrical features in microelectronics. Using a four-point probing, we measured the change in electrical resistance of the tracks as an indicator of the interconnections’ degradation. Our samples were integrated into shirts in two different ways for washing tests, to approach the behavior of wearable devices embedded in textile. First, they were put free-standing inside a closed pocket, which allow circuit moving independently from the surface of fabric. In a second approach, samples were stitched to the fabrics on both shorter sides, which allows partial transmission of stretching movement from the fabric to the stack. In both cases, the films are not attached to the fabric on their whole surfaces. Moreover, the impact of PDMS encapsulation on the samples’ mechanical behavior was tested, as integrated electronic circuits will further be encapsulated for chemical resistance.
Figures 2(a) and 2(b) reports the relative electrical resistance of aluminum lines after washing for PET50/Al9 and PET12/Al9 samples. For both stacks, pocket configuration samples show slightly better electrical results than stitched samples. This could be explained by the fact that, in stitched-configuration, bending and stretching movements of the fabric are transferred to the stack during washing, whereas the samples embedded in pocket only suffer from bending movements. This highlights that further integration of flexible electronics in textile should minimize physical ties with the fabric, and preferably group interconnections on the same side of the device.
For PET50/Al9 stack, pocket and stitched-samples, respectively, show an increase in electrical resistance of 5% and 7% after washing without encapsulation. 800 μm-thick PDMS encapsulation clearly improves the stack’s mechanical behavior, since maximum increase in resistance is about 2% after washing. Regarding global deformations, PET50/Al9 samples are slightly less plastically deformed when they are encapsulated. If we focus on the thinnest stack PET12/Al9, a PDMS encapsulation drastically improves the visual aspect of the samples after washing, however with almost no effect on the electrical resistance change. Non-encapsulated PET12/Al9 samples, despite their completely shriveled aspect (see inset pictures in Fig. 2), show similar electrical results to the encapsulated PET12/Al9 samples.
Electrical resistance of aluminum tracks after one washing cycle, normalized to the electrical resistance before washing, for (a) PET50/Al9 and (b) PET12/Al9 samples. Absolute electrical measurements can be found in Fig. S3 in the supplementary material. A washed sample is depicted in inset for each diagram bar. Regarding electrical measurements, PDMS encapsulation has a positive impact on PET50/Al9 samples, whereas no change is noticeable in case of PET12/Al9. Nevertheless, global deformation of PET12/Al9 samples is greatly reduced with PDMS encapsulation.
Electrical resistance of aluminum tracks after one washing cycle, normalized to the electrical resistance before washing, for (a) PET50/Al9 and (b) PET12/Al9 samples. Absolute electrical measurements can be found in Fig. S3 in the supplementary material. A washed sample is depicted in inset for each diagram bar. Regarding electrical measurements, PDMS encapsulation has a positive impact on PET50/Al9 samples, whereas no change is noticeable in case of PET12/Al9. Nevertheless, global deformation of PET12/Al9 samples is greatly reduced with PDMS encapsulation.
We investigate the evolution the neutral plane position using a model of the stack under bending. As a matter of fact, in case of multilayer structures, neutral plane hb shifts through the stack depending on its composition. da Silva proposes an analytical modeling of the elastic deformation of multilayer structures due to applied external uniform bending.31 If we consider a uniaxial geometry (film bending in only one direction), the neutral plane position hb through the stack is given by,
with Yi and ti, respectively, Young modulus and thickness of layer i, and hmi, the coordinate of the middle of layer i. Maximum tensile strain is then calculated using Eq. (1) as a function of radius of curvature.Based on this analytical model, we plot on Fig. 3(a) the neutral plane position in the aluminum layer as a function of PDMS thickness, for PET/Al/PDMS stacks. We do not include the textile’s layer in the model, as the stack is not monolithically integrated onto the fabric. It is important to lower as much as possible the strains inside the metal, as it is the functional layer of the stack. Here, we aim at placing the neutral plane in the middle of the aluminum layer, to reduce compressive and tensile strains inside the metal upon folding. In PET50/Al9, the neutral plane lies in the PET substrate, approximately 5.3 µm away from the middle of the aluminum. An 800 μm-thick layer of PDMS improves the electrical robustness, as it brings the neutral plane closer to the aluminum layer, at PET-Al interface. According to the model, 2 mm of PDMS coating is needed to ideally reach the middle of the functional layer. This thickness counters to our goal to provide imperceptible flexible electronics integrated into clothes. On the contrary, non-encapsulated PET12/Al9 stack naturally possesses a neutral plane close to the middle of the metallic layer, approximately 0.5 µm below it. This explains why PET12/Al9 samples show good electrical measurements despite the completely shriveled aspect of the samples, indicator of high deformations. An 800 μm-thick PDMS encapsulation shifts the neutral plane to the same distance but above the middle of the aluminum. We can assume that strain distribution in the metallic layer is not highly impacted by the displacement of the neutral plane, which explains the low change in electrical behavior in Fig. 2(b). A 600 μm-thick PDMS coating is needed to reach optimum position of the neutral plane, for an overall thickness below 1 mm. This results in thin, compliant, and electrically robust stack, suitable for integration in washable textiles.
(a) Analytical model applied to PET50/Al9 and PET12/Al9 stacks. PET12/Al9 stack naturally possesses a neutral plane close to the middle of aluminum layer, whereas it requires around 2 mm of PDMS encapsulation for PET50/Al9 films. (b) Analytical model applied to stacks of PET12/Al9 and PET50/Al9 with different encapsulating materials. The softer (low Young modulus) the coating, the thicker it needs to be to shift the neutral plane.
(a) Analytical model applied to PET50/Al9 and PET12/Al9 stacks. PET12/Al9 stack naturally possesses a neutral plane close to the middle of aluminum layer, whereas it requires around 2 mm of PDMS encapsulation for PET50/Al9 films. (b) Analytical model applied to stacks of PET12/Al9 and PET50/Al9 with different encapsulating materials. The softer (low Young modulus) the coating, the thicker it needs to be to shift the neutral plane.
Based on this model, one may determine the optimal encapsulation depending on stack’s composition and desired final thickness. Figure 3(b) compares the different thicknesses required to bring the neutral plane of PET12/Al9 stack in the middle of aluminum layer, considering the encapsulation material. Softer materials will have lower mechanical impact on stack and need thicker layer to shift the neutral plane. Whereas materials such as parylene C (Young modulus = 2.8 GPa) and polyvinylidene fluoride (PVDF) (Young modulus = 2.7 GPa) already impact the mechanics of the stack with thickness below 20 µm. These two materials should be more suitable for thicker layer of PET substrate, as they can mechanically balance the stack while keeping a reduced overall thickness.
In this work, we have demonstrated that commercialized flexible substrates could be mechanically improved to bring their properties closer to the requirements of washable electronics. We approached the issue of plastic deformations in polymer substrates under washing, by correlating it to the elastic limit of materials under bending. Due to the high flexibility of our films, we used a customized setup on a moving stage with camera monitoring allowing measurements of bending radius down to 0.1 mm. Comparison of curvature before and after bending reveals the plasticization threshold, which decreases with Young Modulus and thickness of the film. Therefore, 13 μm-thick PET film elastically bends down to 0.4 mm, whereas 50 μm-thick PI film plasticizes at 2.3 mm. Then, we measured the change in electrical resistance of aluminum tracks on polymer substrate, to monitor the damages induced in the metallic layer under washing. For PET50/Al9 samples, an 800 μm-thick PDMS encapsulation reduces the change in electrical resistance, from 5% to 2%, after one washing cycle. An analytical model of stacks under bending shows how the encapsulation layer brings the neutral plane closer to the aluminum layer, lowering the strains applied upon folding. For thinner and more compliant PET12/Al9 stack, the neutral plane is naturally close to the middle of the aluminum layer and the stack does not require a PDMS encapsulation to show good electrical results after washing. Nonetheless, a thin layer helps counterbalance the high deformability of the thin film and prevent from high deformations. As a result, the mechanical behavior of conventional films of FCB industry can be improved by controlling of the stack and with final overall thickness below 1 mm. Further materials with balanced thicknesses and Young moduli in regard to the analytical model could be investigated for even thinner devices.
Patterning of electrical tracks: Tested substrates were commercialized laminated films of aluminum on PET from ADDEV Materials. To create the aluminum tracks, unwanted metallic parts were removed by laser-hatching using the laser equipment Protolaser S from LPKF, with a 1064 nm wavelength beam, powered at 9 W with a frequency of 75 kHz. The diameter of the laser beam is 25 µm.
Washing test and electrical measurements: Washing cycles were performed in a domestic Samsung frontload washing machine WD0804W8E. To mimic real-life conditions, “daily” washing program was selected with a washing temperature of 30 °C and 800 rpm tumble dry step. A whole cycle was 1 h and 4 min long. No detergent was added to limit the washing effects to mechanical strain. For each cycle, the machine was filled in with same textile loading (2 kg). After wash, the samples were manually removed from textile and air-dried at room temperature before electrical testing.
Electrical tests were performed using a four-probe setup, with a National Instruments PXIe-4141 4-channel precision source measure unit and a PXI-4071 digital multimeter. Each washing configurations was tested through 5 samples. After washing, 6 electrical measurements were performed on each sample. Average values and error bars were calculated from those 6 measurements.
Mechanical characterization under bending: Bending tests were performed on a custom-designed XY motion table. The samples were placed in customized clamps featuring guiding curved ramps, in order to promote sharpest bending in the middle of the sample. A Foculus IEEE 1394 digital camera allowed high-resolution pictures of the samples during bending tests. Bending of the sample and image capture were synchronized and performed with a LabVIEW program. Bending radii were extracted from pictures using LabVIEW Vision module.
The supplementary material shows some additional data on the bending test (LabVIEW programs for monitoring of the X-Y table and bending tests on plain metal-polymer films) and additional data on the washing tests (raw electrical measurements after washing).