Whole Fabric‐Assisted Thermoelectric Devices for Wearable Electronics

Abstract Flexible thermoelectric generators (f‐TEGs) have demonstrated great potential in wearable self‐powered health monitoring devices. However, the existing wearable f‐TEGs are neither flexible enough to bend and stretch while maintaining the device's integrity with a good TE performance nor directly compatible with clothes materials. Here, ultraflexible fabric‐based thermoelectric generators (uf‐TEGs) are proposed with conductive cloth electrodes and elastic fabric substrate. The patterned elastic fabric substrate fits the rigid cuboids well, together with serpentine structured cloth electrodes, rendering uf‐TEG with excellent integrity and flexibility, thereby achieving a highly functional TE performance when strain reaches 30% or on arbitrarily shaped heat sources. The uf‐TEGs show a large peak power of 64.10 μW for a temperature difference of 33.24 K with a high voltage output of 111.49 mV, which is superior compared to previously reported fabric‐based TEG devices, and it is still functional after the water immersion test. Besides the energy harvesting function, with both the temperature sensing ability and the touch perception, this uf‐TEG is demonstrated as the electrical skin when mounted on a robot. Moreover, due to the wind‐sensitive performance and self‐power ability, the uf‐TEGs are assembled on cloth as wearable health and motion monitoring devices.

loosely to the cloth electrode surface, with some regions still connected due to the sticky gel of the cloth tape material after heating. Then, it could be peeled off using a tweezer under the hot gun of setting a temperature of 120 °C. As demonstrated in the red circle of Figure S1e, two mutual pastes were connected to two sides of uf-TEG. In this way, the uf-TEG can be tightly worn on arms or legs and put on cylinder-shaped heat sources like cups with good contact. Figure S2. Testing set-up for measuring the output performance of uf-TEGs.

Testing set-up for measuring the output performance of uf-TEGs
To test the output performance of the uf-TEG, the uf-TEG was put on top of the hot plate, the surface temperature of the hot plate was measured by a thermocouple (k type) that connected to the data acquisition card (NI). The positive and negative anodes were also connected to the data acquisition card to record the voltage output data logs. ∆T was the temperature difference of the top and bottom cuboid temperature.The temperatures on the top and the bottom of the cuboid were measured from the infrared camera. Figure S3. Temperature distribution on different regions of the human body. The infrared images and the voltage output pictures captured when the uf-TEG was worn on a) arm, b) leg, and c) hand.  Figure S4. a) Test set-up for the wind speed affection. b) The voltage output of the uf-TEG under a wind speed of 0.5, 1.5, and 2.5 m/s. c) Nose respiration rate testing.

Temperature distribution on different test regions
The effect of the wind speed on the output performance of uf-TEGs was also conducted through the experiment set-up in Figure S4a. The electric fan was put next to the hot plate at a distance of 5 cm and the voltage source controlled wind speed. The bottom side of the uf-TEG was set at the 35 °C (±0.5°C) with ΔT of 10 °C. The output voltage of the uf-TEG under a wind speed of 0.5, 1.5, and 2.5 m/s was recorded in Figure S4b, and the output voltage increased suddenly after opening the fan. The wind took away the heat from the upper side and increased the temperature gradient between the upper and inner sides for all TE cuboids, thereby increasing the output voltage. The nose respiration rate was measured in Figure S3c through uf-TEGs with 8 TE pairs based on the same working principle. Figure S5. Temperature sensing glove testing. a) Temperature sensing glove with a beaker in hand (200 ml water). b) Temperature sensing with uf-TEG based glove (48 pairs) on hand with water temperature variated from 0°to 60°. c) water temperature in the beaker increased from 0 to 60 °C.

Temperature sensing glove
The temperature sensing glove was shown in Figure S5a. The water temperature was set from 0 to 60 °C with an interval of 5 °C and the temperature was cooled or heated by the refrigerator or hot plate. The voltage output under different water temperatures is shown in Figure S5b. The temperature was measured from the thermometer (Fisher brand) ( Figure S5c).  To experimentally verify the electrode's flexibility and stretchability, a single electrode has been transferred on a cloth substrate to test its resistance variation under stretching, bending and twisting situations ( Figure. S8a). As we stretched the electrode to 37.5% and 50.0 %, the corresponding resistance variations were 5.83% and 12.56%. When the electrode was bent on acrylic molds with different bending radii from 1 cm to 3 cm, with 0.5 cm intervals in between, the resistance variation was relatively small as it increased by 2.34% at the largest bending degree (bending radius of 1.0 cm). In addition, the electrode was clamped on both side and twisted from 0° to 360°, the resistance dropped about 2.05% and 6.81% for 270° and 360°, respectively. The small resistance change during the stretching, bending, and twisting for every single electrode pave the way for the stable TE performance after assembling with TE elements on cloth substrate. FEA simulation was used to analyze the stress distribution for a single electrode with a bending radius of 1 cm, 2 cm, and 3 cm, respectively. As shown in Figure S8b, the largest stress within the connected region is located on the electrode's inner parts (pointed out in red box) with relatively small stress with three orders of magnitudes of Pascal (Pa). FEA simulation was used to analyze the stress distribution for a single electrode under the strain of 10%, 20%, and 30%, respectively. As shown in Figure S8c, the largest stress is located on the electrode's inner parts (pointed out in red rectangles), with a value of 0.53 MPa under the strain of 30%. And the stress distribution also accounted for the small folds on the top and bottom sides of the electrodes during a stretching process (Figure 4b).

Human motion detection
6. Temperature sensing for electrical-skin application Figure S9. a) Schematic diagram of the working principle for this electrical skin of robot hand. b) Fitting result for the voltage output with the increasing water temperature.
When earing the uf-TEG as the e-skin on the robot hand, the inner temperature ℎ was close to the ambient temperature. And the electrical skin was functional as the Temperature difference ℎ existed when the robot hand grabbed a hot or cool bottle ( Figure S9a). As shown in Figure S9b, the fitting curve (red line) demonstrates good linearity for the relation between the voltage output and the input temperature difference. We have conducted experiments to study the impact of the fill factor on the uf-TEG. As shown in Figure R10a, three uf-TEGs were fabricated with different fill factors. Together with the one with 48 pairs device in the manuscript, these four uf-TEGs occupied the same area but with different TE pairs. To connect them, three types of electrodes with different lengths were designed and the resistance for different types of electrodes and the whole devices were summarized in Figure S10b and S10c. These two figures illustrate that the longer the serpentine electrode, the larger the electrode resistance. In the manuscript, the resistance of a single electrode designed for the uf-TEGs of 48 pairs is ~0.5 Ω, and the whole device resistance is ~54.5 Ω. As shown in Figure S10d, the output voltage of the 48 pairs is 3.77 times as that of the one with 12 TE pairs, but the inner resistance is only 1.48 times, according to the equation of the maximum power output ( = 2 4 ), the P max of uf-TEGs of 48 pairs is 9.60 times that of the 12 pairs, which further demonstrates that the larger the fill factor, the higher the device output power will be. Therefore, with the premise of designing a stretchable TEG, the 48 pairs device (fill factor = 0.16) with only one serpentine cycle could achieve the best device performance. The electrode resistance after different stretching cycles (from 50 to 2000 cycles) is shown in Figure S11a. After 2000 stretching cycles, the electrode resistance shows a slightly upward trend but with only 0.04Ω increase (increased by 7.1%). The SEM images were captured to show the surface morphology of the conductive fiber surface ( Figure S11b and S11d).

Stability
Comparing two electrodes with and without 2000 stretching cycles ( Figure S11c and S11e), especially when we focus on the largest deformation regions, we did not find any fracture on the electroplated fibers. Therefore, we speculate the woven electroplated fibers could share the localized stress, which had kept the surface metal layer from broken.