The free-standing polyimide foil was cleaned by sonication in acetone and isopropanol for 5 min, respectively, and cured in a vacuum oven at 200°C overnight. 50-nm thick SiN _x was deposited through plasma-enhanced chemical vapor deposition (PECVD) on both substrate sides as a buffer layer to ensure good mechanical and chemical stability during the fabrication. Then, the process was started with a 35-nm Cr metal gate deposited using electron beam evaporation and patterned through standard photolithography. 25-nm thick Al₂O₃ gate dielectric was deposited using atomic layer deposition (ALD) and 15-nm thick a-IGZO was sputtered at room temperature to form the active layer. Wet etching was performed to structure the semiconductor island and the vias. 10-nm Ti adhesive layer and 60-nm Au were deposited through electron beam evaporation to form the Source/Drain contacts and patterned by lift-off. Finally, an additional 25-nm thick Al₂O₃ was deposited to provide the device passivation. The maximum temperature of the process was 150°C, reached during the ALD. More details on the fabrication process can be found in Munzenrieder et al. (2011). An optical picture of the finalized device is shown in Figure 1A.

The measurements were carried out within an Espec SH-262 climate chamber by setting temperature from −30.0°C to 50.0°C and RH from 0 to 95%. The experiment was conducted in dark to avoid light-induced instability on the IGZO active layer. The flexible TFT was fixed on a rigid glass substrate. The electronics characterization was performed through a Keithley 2614B source meter. The electrical interconnection between the device and the electronics setup was obtained by placing small pieces of Cu tapes at the edge of the polyimide stripe. Here, short Cu wires were glued through a silver conductive paste to connect the TFT pads and the tapes. Similarly, long Cu wires were glued and connected with the source meter. Data were recorded by means of a customized LabVIEW program.

The first measurement was performed at room T and RH (23.7°C, 55%) to evaluate the TFT performance. Figure 1B shows the TFT transfer characteristic and output. Note that the gate current (I _g ) (not reported here) was in the range of 1 μA and the drain current (I _d ) was 0.17 mA for V _GS = −6 V. These non-idealities were caused by the employed measurement setup (long Cu wires and source meter) since a I _g and an I _off in the order of pico-ampere (pA) were obtained by performing the electronics characterization in a probe station using a parameter analyzer (Agilent Technology B1500A).

The extracted parameters were obtained as follows: threshold voltage V _TH = −3.96 V, subthreshold swing SS = 314 mV/dec, on/off current ratio I _on /I _off = 1.39 × 10⁴, linear field-effect mobility μ _lin = 8.10 cm²/Vs. They were evaluated according to standard TFT model equations Sze and Ng (2006). The negative V _TH defined the TFT functionality in depletion mode.

The mechanical stability of the device was analyzed at room conditions Munzenrieder et al. (2011), Münzenrieder et al. (2013). The influence of both tensile and compressive strain on the electronic parameters was evaluated when the IGZO TFTs were bent to a bending radius of 4 mm (corresponding to a tensile strain of 0.63%) Münzenrieder et al. (2013) and 9 mm (corresponding to a compressive strain of 0.27%) Munzenrieder et al. (2011).

The TFT response to weather variations was evaluated by reproducing the most extreme climes on Earth planet, with their typical temperatures and relative humidity conditions as shown in Figure 2A. The transfer characteristics and the corresponding output were grouped for temperature below and above room T, in Figures 2B,C, respectively. For each weather condition, the transfer characteristic was acquired in the linear region (V _DS = 0.1 V) and saturation region (V _DS = 3 V). The gate current was not affected by climatic variations and remained constant in the microampere (μA) range. The electrical performance parameters are shown in Table 1 according to the order of the experiment. First, the TFT performances were evaluated from room T and RH (23.7°C, 55%) to negative temperatures. In particular, different climate conditions were simulated in the following order: Tundra (−2.0°C, 0%), Antarctic (−30.0°C, 0%), Taiga (−13.0°C, 0%), Greenlands (8.0°C, 60%), Rainforest (15.0°C, 80%). This protocol has allowed stable conditions (temperature and humidity) in the climate chamber during the TFT characterization. Within this region, a first positive shift of 0.22 V was observed from the initial condition to −30.0°C, 0% while the other parameters remained stable. Then, T was progressively increased up to 15.0°C as well as the RH from 0% to 80%. Here, an opposite shift of V _TH was observed to a more negative value. This trend of V _TH was in agreement with literature results showing how water vapour absorption induced charges accumulation on the active layer by depleting the channel Kim et al. (2017), Park et al. (2008). Coherently, the gradual reduction of V _TH was observed up to Jungle conditions (26.0°C, 90%). Finally, a major impact of RH was observed when the device was brought to the high temperature zones with a reduction of RH. Specifically, the following geographical locations were mimicked: Savannah (29.0°C, 60%), and Desert (50.0°C, 22%). Despite T increasing up to 50.0°C between these two locations, the overall effect on V _TH was a positive shift from −4.02 V to −3.86 V. An increase of the I _off was denoted at high variations of the RH (ΔRH>30%), in agreement with Park et al. (2008). Accordingly, the I _on /I _off ratio reduced of one order of magnitude down to 6.73 × 10³ and the SS deteriorated. Differently from the other parameters, the linear mobility was not influenced by RH since it increased with T, showing a maximum variation of 10% at the highest T (50.0°C, Desert).

The TFT suitability for wearable applications and epidermal electronics was evaluated by testing the device at three RH (low 35%, normal 60%, high 95%) and three temperatures to resemble the different body locations, as highlighted in Figure 3A. In particular, the transfer characteristics and the corresponding output are shown in Figure 3B for the forehead at 35.0°C, in Figure 3C for the arm at 32.0°C, and in Figure 3D for the foot at 29.0°C. The performance parameters are shown in Table 2. The V _TH dependence from both T and RH was coherent with expectations. The smallest value for V _TH (−4.12 V) was measured in the forehead at 35.0°C, 95% by proving the impact of increasing both T and RH on this parameter. On the other hand, the positive shift of the V _TH was observed according to the increase of RH, while keeping T constant Chowdhury et al. (2015). A clear influence on the TFT performance was observed at high humidity. Here, the I _off was dramatically increased, deteriorating the I _on /I _off ratio by two orders of magnitude. Furthermore, the SS increased to 1.5 V/dec at the hottest T and 95% meaning a less efficient TFT switching operation. Regarding the mobility variation, the results showed its dependence on both T and RH. A reduction of the linear mobility was observed when the RH changed from the normal condition (60%) to both high and low levels. This was coherent to previous experiments performed at constant T reporting the electronics characterization of two a-IGZO TFTs stored at ultra-dry condition (5%) and high humidity (95%) Lee and Jeong (2018).

After all the experiments, the TFT was again characterized at room T and RH (26.0°C, 55%). The results showed a variation of V _TH of 0.2 V (V _TH = −3.76 V), stable I _on /I _off ratio (I _on /I _off = 2.04 × 10⁴) and an increase of the μ _lin less than 5% (μ _lin = 8.37 cm²/Vs) compared to the starting values (V _TH = −3.96 V, I _on /I _off = 1.39 × 10⁴, μ _lin = 8.10 cm²/Vs). The stability of the IGZO TFT to T and RH conditions is linked with the presence of the Al₂O₃ passivation layer, preventing the device degradation due to moisture absorption Corsino et al. (2020), and oxygen vacancies formations Zhou et al. (2014).