Effect of Dual Porous Layers with Patterned Wettability on Low-Temperature Start Performance of Polymer Electrolyte Membrane Fuel Cell

The low-temperature start problem of polymer electrolyte membrane fuel cells (PEMFCs) is a factor limiting their large-scale application. To improve the low-temperature start performance of a PEMFC, a novel microporous layer (MPL) and a gas diffusion layer (GDL) with planar wettability distribution, in which the hydrophilic and hydrophobic lines were arranged alternately in the in-plane direction, were investigated in this study. The influence of the dual planar-distributed wettability of the MPL and GDL on the normal temperature and low-temperature start performance of the PEMFC was investigated. Before performing the major experiment, the effect of the assembly pressure of the membrane electrode assembly (MEA), which has a significant effect on the PEMFC performance, was examined and determined to use in the experiment. The experimental results show that the dual hybrid MPL and GDL can further prolong the operation time of the PEMFC at different below-freezing temperatures owing to efficient water management and thus significantly improve the low-temperature start performance of the PEMFC.


Introduction
Fuel cell vehicles (FCVs) have attracted increasing attention in recent years owing to their high energy efficiency and environmental friendliness. However, some issues hinder their practical application, including the problem of low-temperature start. Because polymer electrolyte membrane fuel cells (PEMFCs) generate water inside the cell during FCV operation at below-freezing temperatures, the water generated is in a liquid supercooled state, which may freeze inside the cell [1]. Studies have shown that repeated freezing in the PEMFC causes irreversible damage to the cell. For example, cracks are formed in the proton exchange membrane after repeated freeze/thaw cycles [2], and platinum particles in the catalytic layer (CL) may also detach [3]. In addition, freezing causes gaps between the proton exchange membrane and the CL [4] and damage the fiber structure inside the gas diffusion layer (GDL) [5]. It is possible to improve the low-temperature start ability of PEMFCs using a microporous layer (MPL), because the MPL can contact the CL more closely and reduce the accumulation of water on the surface of the GDL [6]. However, when freezing occurs between the CL and the MPL, the generated ice blocks air transport to the CL, and the PEMFC fails to low-temperature start [7]. Therefore, to improve the adaptability of the PEMFC to below-freezing temperature environments, the management of water effectively inside the cell is important, particularly in the MPL and GDL. maintaining the performance in the normal temperature range. Furthermore, the characteristics of assembly pressure on the low-temperature start performance for PEMFC with both conventional and hybrid MPL were investigated and the adequate assembly pressure was decided for the experiment. The low-temperature start ability of a PEMFC with the dual hybrid structure was compared with those of PEMFCs with the single hybrid MPL structure and a conventional MPL and GDL structure under the optimum assembly pressure. Figure 1 shows the schematics of structure and functions of the dual H-MPL & GDL with planar wettability distribution applied to both the MPL and GDL. The hydrophilic and hydrophobic nano-porous and micro-porous media were striped and distributed in alternating parallel lines in the in-plane direction of the hybrid MPL and GDL, respectively. The position of the hydrophilic/hydrophobic MPL corresponded to that of the hydrophilic/hydrophobic GDL. For the hydrophobic MPL/GDL, the surface inside the pores was non-wettable, so the water was more easily separated from the surface. On the contrary, the surface inside the pores of hydrophilic MPL/GDL was wettable, so water was more likely to stay on the surface. It is precisely because of the difference in wetting characteristics between hydrophilic and hydrophobic pores that the capillary pressure difference is caused, as the liquid water can move from hydrophobic to hydrophilic regions. When the water generated by the CL stays in the narrow gap between CL and MPL, the liquid water moves from the hydrophobic area to the hydrophilic area in the MPL, owing to the capillary pressure difference based on the water repellent nature of the hydrophobic area and the absorbing nature of the hydrophilic area. Subsequently, the water mainly penetrates the hydrophilic areas from the MPL to the GDL. Furthermore, the hybrid GDL also contains hydrophobic and hydrophilic regions. When some amount of water penetrates the MPL and GDL of the hydrophobic areas, it migrates to the hydrophilic region from the hydrophobic region, owing to the capillary pressure difference. According to the previous research, the structure of planar wettability distribution can improve the oxygen diffusivity in GDL under different water saturation. Therefore, it is possible to keep channels as pathways for gas diffusion from the gas channel through the hydrophobic region to the CL. The channels enable the PEMFC to operate continuously, thus improving its low-temperature start performance.

Principle of Dual Hybrid MPL and Hybrid GDL
Energies 2020, 13,3529 3 of 16 conventional and hybrid MPL were investigated and the adequate assembly pressure was decided for the experiment. The low-temperature start ability of a PEMFC with the dual hybrid structure was compared with those of PEMFCs with the single hybrid MPL structure and a conventional MPL and GDL structure under the optimum assembly pressure. Figure 1 shows the schematics of structure and functions of the dual H-MPL & GDL with planar wettability distribution applied to both the MPL and GDL. The hydrophilic and hydrophobic nanoporous and micro-porous media were striped and distributed in alternating parallel lines in the inplane direction of the hybrid MPL and GDL, respectively. The position of the hydrophilic/hydrophobic MPL corresponded to that of the hydrophilic/hydrophobic GDL. For the hydrophobic MPL/GDL, the surface inside the pores was non-wettable, so the water was more easily separated from the surface. On the contrary, the surface inside the pores of hydrophilic MPL/GDL was wettable, so water was more likely to stay on the surface. It is precisely because of the difference in wetting characteristics between hydrophilic and hydrophobic pores that the capillary pressure difference is caused, as the liquid water can move from hydrophobic to hydrophilic regions. When the water generated by the CL stays in the narrow gap between CL and MPL, the liquid water moves from the hydrophobic area to the hydrophilic area in the MPL, owing to the capillary pressure difference based on the water repellent nature of the hydrophobic area and the absorbing nature of the hydrophilic area. Subsequently, the water mainly penetrates the hydrophilic areas from the MPL to the GDL. Furthermore, the hybrid GDL also contains hydrophobic and hydrophilic regions. When some amount of water penetrates the MPL and GDL of the hydrophobic areas, it migrates to the hydrophilic region from the hydrophobic region, owing to the capillary pressure difference. According to the previous research, the structure of planar wettability distribution can improve the oxygen diffusivity in GDL under different water saturation. Therefore, it is possible to keep channels as pathways for gas diffusion from the gas channel through the hydrophobic region to the CL. The channels enable the PEMFC to operate continuously, thus improving its low-temperature start performance. The main material of the H-MPL used in this research was carbon black. Adding polytetrafluoroethylene (PTFE) can make the MPL hydrophobic, whereas adding polyamide resin can make it hydrophilic. The detailed process of preparing the H-MPL with high reproducibility was described in a former study [25]. The contact angle of hydrophobic area in H-MPL was 135.0 ± 5.0°. The measurement of contact angle for hydrophilic MPL was difficult, because the water drop was absorbed immediately inside when touching the MPL surface. This shows that the hydrophilic MPL has a strong hydrophilicity. Toray 060 was used as the GDL, which could be treated with PTFE to make it hydrophobic. The detailed method of PTFE treatment for preparing hydrophobic stripes of the hybrid GDL (H-GDL) can be found in another study [21]. The main material of the H-MPL used in this research was carbon black.

Principle of Dual Hybrid MPL and Hybrid GDL
Adding polytetrafluoroethylene (PTFE) can make the MPL hydrophobic, whereas adding polyamide resin can make it hydrophilic. The detailed process of preparing the H-MPL with high reproducibility was described in a former study [25]. The contact angle of hydrophobic area in H-MPL was 135.0 ± 5.0 • . The measurement of contact angle for hydrophilic MPL was difficult, because the water drop was absorbed immediately inside when touching the MPL surface. This shows that the hydrophilic MPL Energies 2020, 13, 3529 4 of 16 has a strong hydrophilicity. Toray 060 was used as the GDL, which could be treated with PTFE to make it hydrophobic. The detailed method of PTFE treatment for preparing hydrophobic stripes of the hybrid GDL (H-GDL) can be found in another study [21].
The liquid water distribution on the surface of GDL side, after immerging the dual H-MPL & GDL in water for 30 s, is shown in Figure 2. The water distribution inside hybrid GDL was investigated by X-ray in previous research [21]. When the liquid water saturation in the hybrid GDL was not 100%, water would accumulate in the hydrophilic region, leaving channels for gas diffusion in the hydrophobic region. In addition, the distribution of liquid water on the hybrid MPL surface has been described in literature [25]. Water accumulated in the hydrophilic MPL while the hydrophobic MPL remained waterless. Such water management of dual H-MPL & GDL affects the performance of PEMFC. The liquid water distribution on the surface of GDL side, after immerging the dual H-MPL & GDL in water for 30 s, is shown in Figure 2. The water distribution inside hybrid GDL was investigated by X-ray in previous research [21]. When the liquid water saturation in the hybrid GDL was not 100%, water would accumulate in the hydrophilic region, leaving channels for gas diffusion in the hydrophobic region. In addition, the distribution of liquid water on the hybrid MPL surface has been described in literature [25]. Water accumulated in the hydrophilic MPL while the hydrophobic MPL remained waterless. Such water management of dual H-MPL & GDL affects the performance of PEMFC. The area of the MPL was 5.0 × 5.0 cm 2 , the thickness was 30 ± 5 μm, and the hydrophilic or hydrophobic stripe width of each H-MPL and H-GDL was 1.0 mm. Different stripe widths and angles can be studied in future research. The pore size of the conventional MPL was mainly around 77 nm, while that of the hybrid MPL was mainly around 96 nm. Table 1 lists the specification and experimental conditions for the low-temperature start of the PEMFC used in this experiment. A carbon bipolar plate with a parallel flow gas channel was used. The widths of the flow channel and rib ensured that the hydrophilic and hydrophobic lines of H-MPL coild be fixed parallel to the gas channel, so as to avoid the effect of location difference between the flow channel and the hydrophilic/hydrophobic region on the PEMFC performance. The boundaries between the hydrophilic and hydrophobic areas of the GDL were arranged such that they were located at the center of the rib and the channel. SGL 29BC with the MPL was used for the anode GDL, and the three types of structures (mentioned in Section 2.1) were used for the cathode GDL and MPL. In each set of tests, all other factors were the same except for the cathode GDL and/or MPL. It should be noted that the PEMFC operated at a low current density (0.04 A·cm −2 ) in order to avoid the rapid generation of supercooled water in CL and to prevent membrane dehydration, which are adverse to the low-temperature start performance of the cell [26]. Figure 3 shows the experimental system for the low-temperature start of the PEMFC. The PEMFC was positioned in the constant-low-temperature chamber. The temperatures of the fuel cell and reactant gas were controlled by exchanging heat with the air in the chamber. After the reactant gas was refrigerated in the cooling coil and reached the preset temperature, it entered the PEMFC. In addition, an electrochemical workstation was used to measure the variation in the internal ohmic resistance (highfrequency resistance) of the PEMFC under subfreezing conditions, and the frequency was 2500 Hz. The area of the MPL was 5.0 × 5.0 cm 2 , the thickness was 30 ± 5 µm, and the hydrophilic or hydrophobic stripe width of each H-MPL and H-GDL was 1.0 mm. Different stripe widths and angles can be studied in future research. The pore size of the conventional MPL was mainly around 77 nm, while that of the hybrid MPL was mainly around 96 nm. Table 1 lists the specification and experimental conditions for the low-temperature start of the PEMFC used in this experiment. A carbon bipolar plate with a parallel flow gas channel was used. The widths of the flow channel and rib ensured that the hydrophilic and hydrophobic lines of H-MPL coild be fixed parallel to the gas channel, so as to avoid the effect of location difference between the flow channel and the hydrophilic/hydrophobic region on the PEMFC performance. The boundaries between the hydrophilic and hydrophobic areas of the GDL were arranged such that they were located at the center of the rib and the channel. SGL 29BC with the MPL was used for the anode GDL, and the three types of structures (mentioned in Section 2.1) were used for the cathode GDL and MPL. In each set of tests, all other factors were the same except for the cathode GDL and/or MPL. It should be noted that the PEMFC operated at a low current density (0.04 A·cm −2 ) in order to avoid the rapid generation of supercooled water in CL and to prevent membrane dehydration, which are adverse to the low-temperature start performance of the cell [26].  Figure 3 shows the experimental system for the low-temperature start of the PEMFC. The PEMFC was positioned in the constant-low-temperature chamber. The temperatures of the fuel cell and reactant gas were controlled by exchanging heat with the air in the chamber. After the reactant gas was refrigerated in the cooling coil and reached the preset temperature, it entered the PEMFC. In addition, an electrochemical workstation was used to measure the variation in the internal ohmic resistance (high-frequency resistance) of the PEMFC under subfreezing conditions, and the frequency was 2500 Hz.  On the other hand, the performance of the PEMFC under normal temperature (70 °C) was tested using an experimental system ( Figure 3). The fuel cell temperature was controlled at 70 °C by pumping the circulating constant-temperature water into the end plate of the PEMFC. The stoichiometric ratio of hydrogen and air entering the PEMFC was 5.0, and the relative humidity of air and hydrogen was 70% and 100%, respectively. The low-temperature start experiment was conducted by first purging the cathode and anode of PEMFC with nitrogen (50 °C, 35% RH, 1000 mL/min) for 1.5 h to set the initial state of the moisture distribution inside the PEMFC. Second, the fuel cell was placed in the constant-temperature chamber to cool it down to the target temperature and was maintained there for 1 h. Third, hydrogen and air were pumped into the anode and cathode of the PEMFC, respectively, and the current density was set. When the PEMFC started to operate, relevant data were recorded. Finally, the recording was stopped after the PEMFC operated for 30 min or when it failed to start. The first step was repeated for the next experiment. On the other hand, the performance of the PEMFC under normal temperature (70 • C) was tested using an experimental system ( Figure 3). The fuel cell temperature was controlled at 70 • C by pumping the circulating constant-temperature water into the end plate of the PEMFC. The stoichiometric ratio of hydrogen and air entering the PEMFC was 5.0, and the relative humidity of air and hydrogen was 70% and 100%, respectively.

Experimental System for Low-Temperature Start of PEMFC
The low-temperature start experiment was conducted by first purging the cathode and anode of PEMFC with nitrogen (50 • C, 35% RH, 1000 mL/min) for 1.5 h to set the initial state of the moisture distribution inside the PEMFC. Second, the fuel cell was placed in the constant-temperature chamber to cool it down to the target temperature and was maintained there for 1 h. Third, hydrogen and air were pumped into the anode and cathode of the PEMFC, respectively, and the current density was set. When the PEMFC started to operate, relevant data were recorded. Finally, the recording was stopped after the PEMFC operated for 30 min or when it failed to start. The first step was repeated for the next experiment.

Characteristics of MEA Structural Change by Assembly Pressure
The assembly pressure on MEA is mainly controlled by bolt tightening torque on PEMFC assembly. In this research, pressure sensitive paper was used to test the assembly pressure distribution between CL and MPL under bolt torque of 1 N·m, 2 N·m, and 4 N·m, respectively. The range of assembly torque was set to prevent gas leakage from PEMFC and severe damage to GDL. As shown in Figure 4a, the assembly pressure acting on MEA increased with increasing bolt torque. When bolt torque was 1 N·m, 2 N·m, and 4 N·m, the corresponding average assembly pressure acting on CL and MPL were 0.30-0.75 MPa, 0.50-1.50 MPa, and 1.50-2.50 MPa, respectively. Particularly, the pressure in the region under the ridge was greater than in the region under the gas flow channel.

Characteristics of MEA Structural Change by Assembly Pressure
The assembly pressure on MEA is mainly controlled by bolt tightening torque on PEMFC assembly. In this research, pressure sensitive paper was used to test the assembly pressure distribution between CL and MPL under bolt torque of 1 Nm, 2 Nm, and 4 Nm, respectively. The range of assembly torque was set to prevent gas leakage from PEMFC and severe damage to GDL. As shown in Figure 4a, the assembly pressure acting on MEA increased with increasing bolt torque. When bolt torque was 1 Nm, 2 Nm ,and 4 Nm, the corresponding average assembly pressure acting on CL and MPL were 0.30-0.75 MPa, 0.50-1.50 MPa, and 1.50-2.50 MPa, respectively. Particularly, the pressure in the region under the ridge was greater than in the region under the gas flow channel. Figure 4b shows the different magnification photos of the GDL surface under different assembly pressure. Figure 4b shows that the GDL surface was flat when the assembly pressure was 1 Nm. When the assembly pressure increased, the surface of GDL with striped unevenness was seen. Particularly, under 4 Nm, the pattern was obvious. In addition, original thickness of GDL with MPL without the addition of the tightening force was 175 ± 5 μm. When the assembly pressure was 1 Nm, 2 Nm, and 4 Nm, the thickness was 150 μm, 130 μm, and 100 μm, respectively. Under the assembly pressure of 4 Nm, the thickness of GDL with MPL was 0.571-times thicker than original one.  The features of microstructure and pore in GDL at different assembly pressures are shown in Figure 5. The photos of GDL cross-section were taken by scanning electron microscopy (SEM), as shown in Figure 5a. It can be seen that when the assembly pressure was 1 Nm, there was no obvious fracture of the carbon fibers on the upper surface of GDL. There were slight destructions in the carbon fiber of   Figure 4b shows that the GDL surface was flat when the assembly pressure was 1 N·m. When the assembly pressure increased, the surface of GDL with striped unevenness was seen. Particularly, under 4 N·m, the pattern was obvious. In addition, original thickness of GDL with MPL without the addition of the tightening force was 175 ± 5 µm. When the assembly pressure was 1 N·m, 2 N·m, and 4 N·m, the thickness was 150 µm, 130 µm, and 100 µm, respectively. Under the assembly pressure of 4 N·m, the thickness of GDL with MPL was 0.571-times thicker than original one.
The features of microstructure and pore in GDL at different assembly pressures are shown in Figure 5. The photos of GDL cross-section were taken by scanning electron microscopy (SEM), as shown in Figure 5a. It can be seen that when the assembly pressure was 1 N·m, there was no obvious fracture of the carbon fibers on the upper surface of GDL. There were slight destructions in the carbon fiber of GDL under 2 N·m. However, when the assembly pressure was 4 N·m, the carbon fibers on the GDL surface were significantly broken. The GDL under the ridge was significantly thinner than that under the gas flow channel due to high compression. Figure 5b shows the porosity and pore size distribution of GDL and MPL under different assembly pressures. It should be noted that the pore size and porosity were measured using the mercury intrusion porosimetry (MIP) method. The two peaks in the curve for 1 N·m case indicate that the pore diameter of GDL and MPL were around 33.0 µm and 95.4 nm, respectively. However, when the assembly pressure was 2 N·m, in addition to a higher peak (indicating that the pore size of GDL was mainly 33.0 µm), there were two lower peaks in the curve. The peak at 77.1 nm represents the pore size of MPL, while the peak at 1.0 µm shows the pore size of GDL compressed under the ridge. Similarly, when the assembly pressure was 4 N·m, there were three peaks in the curve, indicating that the pore diameters of GDL and MPL were 24.2 µm (GDL under the gas flow channel), 0.8 µm (GDL under the ridge), and 62.5 nm, respectively. The comparison of the pore size distribution of GDL and MPL under different assembly pressures indicates that the larger the assembly pressure was, the smaller the pore size of MPL or GDL was. Moreover, under lager assembly pressure, the thicknesses of GDL under gas flow channel and under ridge were different, and the pore size distributions of GDL under the two regions were also different. In this study, since MPL was combined with GDL, the porosity of GDL or MPL could not be measured separately. The results of the test show that when the assembly pressure was 1 N·m, 2 N·m, and 4 N·m, the average porosity of GDL and MPL was 76.4%, 74.9%, and 64.0%, respectively. Thus, the increase in assembly pressure resulted in the decrease of porosity of GDL and MPL. When PEMFC was working, the changes in GDL directly affected the transfer of liquid water and the freezing of supercooled water inside.
The performance of PEMFC with two different combinations of MPL + GDL (C-MPL & GDL, H-MPL & C-GDL) was tested and compared under 1 N·m, 2 N·m, and 4 N·m to decide the optimum assembly pressure on low-temperature start ability. Then, the performance of PEMFC with different MPL and GDL structures at 70 • C and low-temperature start performance were investigated under the optimum assembly pressure.  Figure 6a,b PEMFC had a longer working time with hybrid MPL compared to conventional MPL for each supercooling degree and assembly pressure. In general, the cold startup process of PEMFC was affected by exothermic reaction and supercooling degree. When the degree of supercooling increased, the freezing probability of the liquid water inside the fuel cell increased and was one of the reasons why the PEMFC working time reduced.

The Effect of Assembly Pressure on Low-temperature Start Performance
Energies 2020, 13, 3529 8 of 16 size distributions of GDL under the two regions were also different. In this study, since MPL was combined with GDL, the porosity of GDL or MPL could not be measured separately. The results of the test show that when the assembly pressure was 1 Nm, 2 Nm, and 4 Nm, the average porosity of GDL and MPL was 76.4%, 74.9%, and 64.0%, respectively. Thus, the increase in assembly pressure resulted in the decrease of porosity of GDL and MPL. When PEMFC was working, the changes in GDL directly affected the transfer of liquid water and the freezing of supercooled water inside.   Figure 6a,b PEMFC had a longer working time with hybrid MPL compared to conventional MPL for each supercooling degree and assembly pressure. In general, the cold startup process of PEMFC was affected by exothermic reaction and supercooling degree. When the degree of supercooling increased, the freezing probability of the liquid water inside the fuel cell increased and was one of the reasons why the PEMFC working time reduced.
In addition, the influence of the tightening bolt torque can be seen from Figure 6a,b. With the increasing assembly pressure, the working time of PEMFC with both hybrid and conventional MPLs reduced at the same supercooling degree, because the capacity of the total water retention of MPL and GDL is considered a very important factor in cold start condition. Since the increase in assembly pressure leads to a decrease in the porosity of MPL and GDL due to thickness reduction by pressure and easier water saturation, it is easier to block its pores and prevent oxygen from reaching the cathode CL with the freezing of supercooled water in MEA and GDL. Excessive assembly pressure, however, damages the structure of carbon fiber in GDL. All of these changes decreased the working time of PEMFC. However, if the assembly pressure is too low, gas leaks from PEMFC. Therefore, 1 Nm was the most suitable assembly pressure in this experiment. When PEMFC works, liquid water first accumulates in CL and then passes through the cathode MPL and GDL to reach the gas flow channel. The accumulation or freezing of supercooled water in the cathode gas flow channel causes the pressure drop between the inlet and outlet to increase. Figure 7a Figure 7a,b the growth range of pressure drop increased with the increase of assembly pressure, indicating that liquid water can pass through MPL and GDL faster and then move into the gas flow channel. Although increasing the assembly pressure increased the transfer resistance of the liquid water, it also reduced the thickness of GDL and MPL. This resulted in a shorter moving distance for the liquid water, as the water or ice could gather in the gas flow channel faster. Comparing Figure 7a,b the ranges of the pressure drop increase of PEMFC with H-MPL & C-GDL were In addition, the influence of the tightening bolt torque can be seen from Figure 6a,b. With the increasing assembly pressure, the working time of PEMFC with both hybrid and conventional MPLs reduced at the same supercooling degree, because the capacity of the total water retention of MPL and GDL is considered a very important factor in cold start condition. Since the increase in assembly pressure leads to a decrease in the porosity of MPL and GDL due to thickness reduction by pressure and easier water saturation, it is easier to block its pores and prevent oxygen from reaching the cathode CL with the freezing of supercooled water in MEA and GDL. Excessive assembly pressure, however, damages the structure of carbon fiber in GDL. All of these changes decreased the working time of PEMFC. However, if the assembly pressure is too low, gas leaks from PEMFC. Therefore, 1 N·m was the most suitable assembly pressure in this experiment.
When PEMFC works, liquid water first accumulates in CL and then passes through the cathode MPL and GDL to reach the gas flow channel. The accumulation or freezing of supercooled water in the cathode gas flow channel causes the pressure drop between the inlet and outlet to increase. Figure 7a,b shows the variation of pressure drop in the cathode gas flow channel when PEMFC worked under different assembly pressures. From the comparison of −4.2 • C curves under different assembly pressures in Figure 7a,b the growth range of pressure drop increased with the increase of assembly pressure, indicating that liquid water can pass through MPL and GDL faster and then move into the gas flow channel. Although increasing the assembly pressure increased the transfer resistance of the liquid water, it also reduced the thickness of GDL and MPL. This resulted in a shorter moving distance for the liquid water, as the water or ice could gather in the gas flow channel faster. Comparing Figure 7a,b the ranges of the pressure drop increase of PEMFC with H-MPL & C-GDL were smaller than that of C-MPL & GDL. Since the hydrophilic area in hybrid MPL stored part of the generated water, the amount of liquid water entering the flow channel was smaller than that of C-MPL & GDL. The variations of ohmic resistance during PEMFC operation at different assembly pressures are shown in Figure 8. When PEMFC worked at low temperature, the ohmic resistance gradually decreased at the initial stage. When PEMFC failed to start at a low temperature, the voltage started to decrease and the ohmic resistance inside the fuel cell gradually increased. Since the supercooled water inside the PEMFC froze and an ice layer was formed in the MEA, the PEMFC resistance increased [6,26]. Comparison of Figure 8a,b shows that at freezing, the resistance of fuel cell with H-MPL & C-GDL increased gradually compared to C-MPL & GDL, since H-MPL & C-GDL contained water in hydrophilic MPL regions, and the water in the interface between MPL and CL was less than that of C-MPL & GDL. Figure 8 clearly shows that with increasing assembly pressure, the increase range of resistance decreased when the cold startup failed. A possible reason is that when assembly pressure increased, the porosity of MPL and GDL decreased, causing a decrease in the volume of liquid water drops inside them.

Performance of PEMFC with Different MPL/GDL Structures at Normal Temperature
As 1 Nm is the most suitable assembly pressure for PEMFC according to the research in Section 3.2, all of the following experiments were conducted under the assembly pressure of 1 Nm. Figure 9 shows the performance of PEMFC with different MPL or GDL structures at 70 °C in terms of the voltage V and power density P vs the current density I. Three measurements were performed under each condition to verify the repeatability. The error in each condition was within 2% at the maximum power density. Moreover, the trends in the curves were similar. The I-V and I-P curves show that there was no larger difference in the range of smaller I (−0.2 A/ cm 2 ) among the three types of MEAs. However, the differences in V and P can be seen in a wider range of I (0.2-1.9 A/ cm 2 ). In particular, The variations of ohmic resistance during PEMFC operation at different assembly pressures are shown in Figure 8. When PEMFC worked at low temperature, the ohmic resistance gradually decreased at the initial stage. When PEMFC failed to start at a low temperature, the voltage started to decrease and the ohmic resistance inside the fuel cell gradually increased. Since the supercooled water inside the PEMFC froze and an ice layer was formed in the MEA, the PEMFC resistance increased [6,26]. Comparison of Figure 8a,b shows that at freezing, the resistance of fuel cell with H-MPL & C-GDL increased gradually compared to C-MPL & GDL, since H-MPL & C-GDL contained water in hydrophilic MPL regions, and the water in the interface between MPL and CL was less than that of C-MPL & GDL. Figure 8 clearly shows that with increasing assembly pressure, the increase range of resistance decreased when the cold startup failed. A possible reason is that when assembly pressure increased, the porosity of MPL and GDL decreased, causing a decrease in the volume of liquid water drops inside them.

Performance of PEMFC with Different MPL/GDL Structures at Normal Temperature
As 1 N·m is the most suitable assembly pressure for PEMFC according to the research in Section 3.2, all of the following experiments were conducted under the assembly pressure of 1 N·m. Figure 9 shows the performance of PEMFC with different MPL or GDL structures at 70 • C in terms of the voltage V and power density P vs the current density I. Three measurements were performed under each condition to verify the repeatability. The error in each condition was within 2% at the maximum power density. Moreover, the trends in the curves were similar. The I-V and I-P curves show that there was no larger difference in the range of smaller I (−0.2 A/ cm 2 ) among the three types of MEAs. However, the differences in V and P can be seen in a wider range of I (0.2-1.9 A/ cm 2 ). In particular, the features of the curves were different when 1.3 < I < 1.9 A/cm 2 . The P values are the maximum for the dual hybrid MPL & GDL at a relatively low current density I (1.4 A/cm 2 ) compared with the other two cases. The PEMFCs with the H MPL & C-GDL and C-MPL & GDL exhibited the same maximum I of 1.9 A/cm 2 , which was greater than that in the dual H-MPL & GDL case.
The dual H-MPL & GDL could absorb water in the hydrophilic area and provide a straight path for oxygen diffusion in a relatively wide range of I (0.8-1.6 A/cm 2 ). On the other hand, the anti-flooding property was remarkable in the C-GDL (+ H-MPL or + C-MPL) with the hydrophobic property because of the formation of homogeneously distributed discrete droplets at I values ranging from 1.7 A/cm 2 to 1.9 A/cm 2 , though the diffusion path for oxygen was longer because of the formation of bent paths due to the discrete randomly distributed liquid. Moreover, as shown in Figures 9 and 10, although the generated water tended to stagnate in the gap between CL and MPL due to their hydrophobic properties for C-MPL, the water was quickly absorbed in the hydrophilic area in the MPL and only a little amount of water stagnated in the gap between the CL and MPL for the hybrid MPL. Therefore, the hybrid MPL gave higher V and P values because of the active electrochemical reaction taking place without any excess water on the CL in a wider range of I. In particular, while extension to largest I range (approximately 1.9 A/cm 2 ) with larger V and P was seen for the H MPL & C-GDL case, the dual H-MPL & GDL gave the highest V and P values at approximately I = 1.4 A/cm 2 as described previously. Thus, the PEMFC with the dual H-MPL & GDL structure showed the best performance under the normal temperature condition.  Energies 2020, 13, 3529 11 of 16 from 1.7 A/cm 2 to 1.9 A/cm 2 , though the diffusion path for oxygen was longer because of the formation of bent paths due to the discrete randomly distributed liquid. Moreover, as shown in Figure 9 and Figure 10, although the generated water tended to stagnate in the gap between CL and MPL due to their hydrophobic properties for C-MPL, the water was quickly absorbed in the hydrophilic area in the MPL and only a little amount of water stagnated in the gap between the CL and MPL for the hybrid MPL. Therefore, the hybrid MPL gave higher V and P values because of the active electrochemical reaction taking place without any excess water on the CL in a wider range of I. In particular, while extension to largest I range (approximately 1.9 A/cm 2 ) with larger V and P was seen for the H MPL & C-GDL case, the dual H-MPL & GDL gave the highest V and P values at approximately I = 1.4 A/cm 2 as described previously. Thus, the PEMFC with the dual H-MPL & GDL structure showed the best performance under the normal temperature condition.  Figure 11 shows the tested results of the low-temperature start performance of PEMFC with different MPL and GDL structures. The PEMFC with C-MPL & GDL operated for more than 30 min at −4.2 °C, and the operation time decreased with the decrease in the temperature. The operation time  Figure 11 shows the tested results of the low-temperature start performance of PEMFC with different MPL and GDL structures. The PEMFC with C-MPL & GDL operated for more than 30 min at −4.2 • C, and the operation time decreased with the decrease in the temperature. The operation time of the PEMFC with the dual H-MPL & GDL at below-freezing temperatures was even better than that of the PEMFC with H-MPL & C-GDL. This was because of the alternating structure of the hydrophilic or hydrophobic regions in the dual H-MPL & GDL, as the structure made the supercooled water at the interface of the CL/MPL quickly move from the hydrophobic to hydrophilic regions. Moreover, the water in the MPL and GDL could concentrate in the hydrophilic region, keeping channels for air diffusion to the cathode CL in the hydrophobic region. Therefore, the PEMFC can continue to work for a long time, even if the supercooled water freezes. The prolonged operation time of the PEMFC at below-freezing temperatures can improve the low-temperature start performance of FCVs under below-freezing conditions. Moreover, the performance of PEMFC at normal temperature conditions after 17-times of low-temperature start experiments was tested, and the data before and after low-temperature start were compared. As a result, the structure of the hybrid MPL or GDL can be maintained after operation at low temperatures, and there was no significant degradation in the performance of the PEMFC before and after several low-temperature start experiments.

Low-temperature Start Performance of PEMFC with Different Structures
Energies 2020, 13, 3529 12 of 16 below-freezing conditions. Moreover, the performance of PEMFC at normal temperature conditions after 17-times of low-temperature start experiments was tested, and the data before and after lowtemperature start were compared. As a result, the structure of the hybrid MPL or GDL can be maintained after operation at low temperatures, and there was no significant degradation in the performance of the PEMFC before and after several low-temperature start experiments.  Figure 12a shows the ohmic resistance changes of PEMFC with different MPL/GDL structures under different below-freezing temperatures. For the cell with C-MPL & GDL, it can be seen that at −4.2 °C, the ohmic resistance gradually decreased in the initial 200 s, which was due to the generated water which moistened the MEA, making the ohmic resistance of the cell decrease. As the PEMFC continued to work, the ohmic resistance tended to stabilize. However, at −10.0 °C to −5.5 °C, when the PEMFC stopped working, the ohmic resistance suddenly increased due to the internal freezing of MEA [26]. Moreover, at −5.5 °C and −6.5 °C, the ohmic resistance increased faster than that of −9.0 °C and −10.0 °C when the PEMFC stopped. This is because at higher temperatures (−5.5 °C and −6.5 °C), the water in the cell will remain super-cooled state and accumulate in the MEA. At the moment of stopping, the ohmic resistance will rise rapidly due to the sudden formation of the ice layer in the MEA. However, at lower temperatures (−9.0 °C and −10.0 °C), water will quickly freeze after generation and ice will gradually accumulate in MEA, so ohmic resistance will gradually increase until the PEMFC stops working.   Figure 12a shows the ohmic resistance changes of PEMFC with different MPL/GDL structures under different below-freezing temperatures. For the cell with C-MPL & GDL, it can be seen that at −4.2 • C, the ohmic resistance gradually decreased in the initial 200 s, which was due to the generated water which moistened the MEA, making the ohmic resistance of the cell decrease. As the PEMFC continued to work, the ohmic resistance tended to stabilize. However, at −10.0 • C to −5.5 • C, when the PEMFC stopped working, the ohmic resistance suddenly increased due to the internal freezing of MEA [26]. Moreover, at −5.5 • C and −6.5 • C, the ohmic resistance increased faster than that of −9.0 • C and −10.0 • C when the PEMFC stopped. This is because at higher temperatures (−5.5 • C and −6.5 • C), the water in the cell will remain super-cooled state and accumulate in the MEA. At the moment of stopping, the ohmic resistance will rise rapidly due to the sudden formation of the ice layer in the MEA. However, at lower temperatures (−9.0 • C and −10.0 • C), water will quickly freeze after generation and ice will gradually accumulate in MEA, so ohmic resistance will gradually increase until the PEMFC stops working.    Figure 12b shows the change in the pressure drop in the PEMFC cathode flow channel when using different MPL/GDL structures. The pressure drop of the cathode channel did not increase significantly in the first 800 s (this is common to all the three PEMFCs), and after approximately 800 s, the pressure drop gradually increased. As the temperature was −5.5 • C or lower for the PEMFC with C-MPL & GDL, the pressure drop did not augment significantly during the first 1000 s of cell operation, indicating that water mainly existed and froze in the MEA area during the operation of the PEMFC. The PEMFCs with the dual H-MPL & GDL and H MPL & C-GDL operated for more than 800 s at −5.5 • C and −6.5 • C. Water reached the cathode flow channel and accumulate therein because of the increase in the pressure drop. The increasing rate of the pressure drop was reduced, because the water easily froze in the MEA with decreasing temperature, i.e., the amount of water moving to the flow channel reduced [11]. This trend also indicates that the supercooling degree affected the distance travelled by the supercooled water in the MEA. As the distance travelled by the supercooled water was shorter with the decrease in the temperature, the freezing mainly occurred in the area near the CL. Figure 13a,b summarizes the effect of different structures on the PEMFC operation time at different below-freezing temperatures. The operation time was defined as the duration from the start of operation until the voltage becomes zero. As shown in Figure 13a, when only applying the hybrid MPL structure at −8.0 • C, the operation time of the PEMFC was 950 s, which is 2.14-times greater than that of the PEMFC with C-MPL & GDL (444 s). Furthermore, in the case of the dual H-MPL & GDL structure, the operation time was 1533 s, which is 3.45-times greater than that of the PEMFC with the C-MPL & GDL. Similarly, when the low-temperature start was at −9.0 • C, the operation times were 615 s and 1118 s for the PEMFCs with the H-MPL & C-GDL and dual H-MPL & GDL, respectively. As shown in Figure 13b, in which the enhancement ratio of operational time denotes the magnifications for C-MPL & GDL, the enhancement ratios of operation time were improved by 1.90-and 3.45-times. As described previously, the improvements can be attributed to the wettable area of the H-GDL, which mainly absorbed and kept the supercooled water inside the hydrophilic region. This also helped maintain an air diffusion channel in the hydrophobic area of the GDL, thus further prolonging the operation time at below-freezing temperatures. However, at −10.0 • C, the enhancement ratios of operation time of the PEMFCs with the H MPL & C-GDL and dual H-MPL & GDL were 347 s and 411 s, respectively, which are not significantly better (1.59-and 1.89-times) than that of the PEMFC with the C-MPL & GDL (218 s). The enhancement ratio was reduced with lowering temperature, because the water more likely froze with the increased supercooling of the water at such low temperatures. The amount of water reaching the GDL at −10.0 • C was less than that at −8.0 • C and −9.0 • C. Therefore, the influence of the hybrid MPL became more important, and the effectiveness of the H-GDL in improving the low-temperature start performance of the PEMFC decreased with the decrease in the temperature. When the temperature is below −10 • C, water may freeze in CL [6]. Therefore, H-MPL and H-GDL can improve the PEMFC's low-temperature start performance above −10 • C under the condition of this study, in which the effectiveness of hybrid MPL/GDL is merely shown. However, a relatively thick bipolar plate was used in this study and the thermal insulation was not enough. Therefore, with thin bipolar plate and strict insulation, such as the central cell in the stack, the rate of temperature rise due to power generation becomes larger, and it is likely to operate in the further lower temperature range. temperature, because the water more likely froze with the increased supercooling of the water at such low temperatures. The amount of water reaching the GDL at −10.0 °C was less than that at −8.0 °C and −9.0 °C. Therefore, the influence of the hybrid MPL became more important, and the effectiveness of the H-GDL in improving the low-temperature start performance of the PEMFC decreased with the decrease in the temperature. When the temperature is below −10 °C, water may freeze in CL [6]. Therefore, H-MPL and H-GDL can improve the PEMFC's low-temperature start performance above −10 °C under the condition of this study, in which the effectiveness of hybrid MPL/GDL is merely shown. However, a relatively thick bipolar plate was used in this study and the thermal insulation was not enough. Therefore, with thin bipolar plate and strict insulation, such as the central cell in the stack, the rate of temperature rise due to power generation becomes larger, and it is likely to operate in the further lower temperature range.

Conclusions
The performance of a PEMFC with a novel H-MPL and H-GDL structure exhibiting planardistributed wettability (dual H-GDL & MPL), in which hydrophilic and hydrophobic lines were alternately arranged in the in-plane direction, was investigated. The experiment was performed under the adequate assembly pressure of the MEA as an affecting factor on the PEMFC performance. The experimental results of the study are as follows.
(1) The effect of the assembly pressure of MEA was first determined to use in the experiment, since increasing the assembly pressure between MEA reduces the thickness of GDL and MPL and causes a decrease in porosity. As a result, the working time of PEMFC at subfreezing temperature was longer at the lowest assembly pressure in this experiment.
(2) The PEMFC with the dual H-GDL & MPL exhibited a better low-temperature start performance than those with the H-MPL & C-GDL and C-MPL & GDL. Therefore, it was shown

Conclusions
The performance of a PEMFC with a novel H-MPL and H-GDL structure exhibiting planar-distributed wettability (dual H-GDL & MPL), in which hydrophilic and hydrophobic lines were alternately arranged in the in-plane direction, was investigated. The experiment was performed under the adequate assembly pressure of the MEA as an affecting factor on the PEMFC performance. The experimental results of the study are as follows.
(1) The effect of the assembly pressure of MEA was first determined to use in the experiment, since increasing the assembly pressure between MEA reduces the thickness of GDL and MPL and causes a decrease in porosity. As a result, the working time of PEMFC at subfreezing temperature was longer at the lowest assembly pressure in this experiment.