For this study, the hydrophilicity of the anode-side’s catalyst layer was reinforced to ensure that the anode was able to absorb more water than the cathode, which improved the back-diffusion effect of water produced by the cell cathode. To produce the anode (hydrogen electrode), the following procedure was performed. First, the Pt/C catalyst, SiO₂, Nafion solution, deionized water, and isopropyl alcohol were mixed and prepared in the appro- priate proportions to produce the catalyst ink (Pt/C catalyst and Nafion solution weight ratio, 3:1). The catalyst ink was then stirred (120 rpm) at 80˚C until a certain degree of viscosity was achieved. The transfer paper was coated with the catalyst ink (0.3 mg/cm² Pt content) to form the hydrogen electrode. To produce the cathode (oxygen electrode), similar steps were performed, except no SiO₂ was added to the catalyst ink, which was coated directly onto the GDL.

To produce the MEA using the transfer method, a coating fixture with a size identical to that of the electrode was first produced. Next, the coating fixture was placed onto the transfer paper, which was coated with the catalyst ink. Subsequently, the transfer paper was placed into a circulation oven and dried for 1 h at 80˚C. After drying, the paper was measured to determine whether the required loading level had been achieved. The entire manufacturing process followed the traditional method combined with the hot-pressing technique of the transfer method, where the anode was constructed using transfer paper that was coated with various mixture ratios of the Pt/C- SiO₂ ink, and the cathode was constructed using a gas diffusion electrode that was coated with the Pt/C catalyst ink. A Nafion 112 membrane was inserted between the transfer paper and the gas diffusion electrode. Next, the upper and lower presses were preheated for 60 s at 140˚C, and the pressure was increased to 40 kgf/cm² for 90 s before removal. Finally, self-humidifying MEA was produced.

For the electric current and voltage (I-V) performance curve test in this experiment, hydrogen and air were used as the reaction gases, and humidifying bottles were used to control the humidity of the reaction gases. Prior to the test, the MEA was left to activate for 10 h. To reduce the influence of various operation variables, a fixed H2 flow rate of 200 sccm, an air flow rate of 500 sccm, and a zero gas back-pressure were applied. The I-V test was repeated 15 times, and the data obtained during the 15th iteration was selected to plot the I-V curve to compare how adding the SiO₂ affected the fuel cell performance under the various fuel cell temperatures and humidifying conditions.

After coating the SiO₂-loaded catalyst ink directly onto the GDL, EDS mapping was performed to analyze the composition and observe the distribution or dispersion of the SiO₂. Figure 1 shows the composition analysis results from the EDS mapping of the SiO₂-coated GDL. Figure 1(a) shows the emergence of several particles on the specimen surface. Combining Figures 1(a)-(c) indicates an increasing presence of Si and O elements proximal to these particles. Therefore, we concluded that these were SiO₂ particles. Because the SiO₂ employed in this study was nanopowder, we assumed that these particles were also created by the aggregation of the SiO₂ nanopowder. The EDS mapping analysis showed a slightly worse SiO₂ dispersion and an uneven distribution, which could cause uneven water absorption, thereby influencing the self-humidification effect. Thus, the ink technique for the dispersion of hydrophilic nanopowders requires further improvement.

The water contact angle test is used to measure the hydrophilicity of a surface; the lower the angle, the greater the hydrophilicity. During the experiment, the water contact angle was calculated by dripping water onto the GDL and calculating the average value after 3 measurements. The measurement results in Figure 2 show that the water contact angle value of the non-coated GDL was 118˚, which decreased to 45˚ after SiO₂ coating. This

indicates that adding the SiO₂ nanopowders enhanced the hydrophilicity. In other words, after adding the SiO₂, the specimen was better able to absorb water, which supports our initial assumption.

For the electric current and voltage (I-V) performance curve test in this experiment, hydrogen and air were used as the reaction gases, and common humidifying bottles were employed as the humidifier. We investigated how adding various amounts of SiO₂ to the hydrogen electrode’s catalyst layer affects the cell performance under the following 3 independent conditions: 1) the temperature of the humidified bottles was 70˚C; 2) the temperature of the humidified bottles was 50˚C; and 3) the reaction gas was not humidified.

Figure 3 shows the I-V performance test results for adding varying amounts of SiO₂ when the gas was humidified at 70˚C and the fuel cell temperature was maintained at 70˚C (Figure 3(a)) and 50˚C (Figure 3(b)). A fuel cell temperature of 70˚C is common for the majority of PEMFC operations. Based on the properties of the fuel cell test station used in this experiment, the 70˚C gas humidification temperature delivered the appropriate amount of water vapor into the fuel cell. Consequently, the tests conducted under these conditions facilitate the assessment of the general performance of PEMFCs. When the fuel cell temperature was maintained at 70˚C (Figure 3(a)), the test results showed that the SiO₂-loaded catalyst layer outperformed the non-loaded layer; however, the difference was minor. When operated at 0.5 V, the measured current density was approximately 535 mA/cm² for both fuel cells (i.e., with and without the SiO₂).

Figure 3(b) shows that adding SiO₂ to the catalyst layer had minimal influence on the fuel cell performance. However, the fuel cell with the SiO₂-loaded catalyst layer outperformed the fuel cell with the non-loaded layer. When operated at 0.5 V, the non-loaded fuel cell produced a current density of 494.1 mA/cm². The optimum performance was achieved after adding 0.06 mg/cm² SiO₂, which produced a current density of 527.1 mA/cm². Because the gas humidifying temperature was 20˚C higher than the fuel cell temperature under these test conditions, supersaturated water vapor conditions were observed and the water vapor in the fuel cell was likely to undergo condensation. The slightly superior performance of the fuel cell with the SiO₂-loaded catalyst layer indicates that despite the excessive gas humidity, the addition of hydrophilic SiO₂ did not cause any observable flooding problems and negative effects on the fuel cell performance. Therefore, we hypothesize that when employing the transfer method to construct the anode catalyst layer, it is difficult to disrupt the hydrophobicity of the original GDL. This may be considered an advantage for constructing self-humidifying PEMFC using the transfer method.

Figure 4 shows the I-V performance tests for adding various amounts of SiO₂ when the gas was humidified at 50˚C, and the fuel cell temperature was maintained at 70˚C (Figure 4(a)) and 50˚C (Figure 4(b)). The results in Figure 4(a) show that the gas humidifying temperature was 20˚C lower than that of the fuel cell. Consequently, this difference in temperature caused a shortage of water vapor when the gas entered the fuel cell from the humidifying bottles; accordingly, this was considered a low-humidification test condition. The test results showed that the fuel cells with the SiO₂-loaded catalyst layer outperformed the fuel cells with the non-loaded layer;

however, the practical value of the performance was minor and the difference was minimal. When operated at 0.5 V, the fuel cells with 0.06 mg/cm² SiO₂ attained the optimum performance, and the current density was measured at 85.3 mA/cm².

Figure 4(b) shows that identical gas humidifying temperature and fuel cell temperatures should produce sufficient water vapor to hydrate the fuel cell. Furthermore, the fuel cells with the SiO₂-loaded catalyst layer outperformed the fuel cells with the non-loaded catalyst layer under these testing conditions. When operated at 0.5 V, the current density was measured at 453.4 mA/cm² for the non-layered fuel cells. The optimum performance was attained by adding 0.12 mg/cm² SiO₂, the current density of which was measured at 550.1 mA/cm². The fuel cells with a cell temperature set at 50˚C performed significantly better those with a cell temperature set at 70˚C. This result explains the difference in performance under high and low humidification.

The I-V performance tests when the gas was not humidified could directly verify whether the fuel cells were capable of self-humidification. Figure 5 shows the I-V performance test results when the gas was not humidified and the fuel cell temperature was maintained at 70˚C (Figure 5(a)) and 50˚C (Figure 5(b)).

Figure 5(a) shows that the performance of the fuel cell was extremely poor; thus, few data were recorded for the tests without humidification. Figure 5(b) shows the results for the I-V performance test where the fuel cell temperature was maintained at 50˚C. The rationale for conducting the 50˚C fuel cell test was that we believed the loss of the water produced by the fuel cells was caused by high temperatures within the cells. This also indicates that satisfying the PEMFC self-humidifying requirement would be difficult, regardless of whether a self-humidifying functional design was added to the fuel cell. Therefore, an ideal temperature for self- humidifying PEMFC operations must be established and identified. The results in Figure 5(b) clearly show that the SiO₂-loaded catalyst layer outperformed the non-loaded layer. When operated at 0.5 V, the current density for the non-loaded layer was only 2.43 mA/cm², which offers little to no electricity generation value. After 0.06 mg/cm² SiO₂ was added, the current density increased to 9.36 mA/cm²; however, this current density also has no practical value. After adding 0.12 mg/cm² of SiO₂, the current density increased to 93.9 mA/cm², which approaches actual practical values. Compared to the fuel cells with the non-loaded catalyst layer, the current density of this fuel cell is higher and shows a clear improvement.

At a fuel cell temperature of 50˚C, the addition of a specific amount of SiO₂ could enhance the performance of fuel cells. However, for fuel cells at 70˚C, the effect was non-existent. We contend that this phenomenon was caused by the increased fuel cell temperature, which contributed to evaporation rather than retention of the water produced within the fuel cell. The addition of SiO₂ water adsorption effects offers an insufficient explanation for the water loss caused at high temperatures. This also shows the limitations that self-humidifying fuel cells have regarding fuel cell temperatures. During the electrochemical reaction, the PEMFC also released heat; therefore, standard fuel cell stacks are typically equipped with air- or water-cooling mechanisms to ensure the fuel cell temperature is maintained within a specified range. Thus, understanding the appropriate reaction temperature for PEMFCs is also extremely crucial. Although fuel cells are only operable at 50˚C, if a sufficient and consistent performance can be verified, feasible applications for self-humidifying PEMFCs can be identified.