1 Introduction
Fuel cells, as a new type of energy power system solution [
1], have a superior performance to pure electric ones in terms of driving range, energy density, and convenience of replenishing fuel [
2]. However, its commercialization is limited by the high cost and durability of fuel cell stacks [
3]. Consequently, materials, components, and systems which could prepare highly durable fuel cells were investigated to accelerate the commercialization of fuel cell vehicles. However, these studies focused on the active area of membrane electrode assembly (MEA), and ignored the frame sealing structure at the edge of MEA. Some recent reports have shown that the frame sealing structure can influence the durability of the membrane electrode, and early fuel cell failure usually occurs at the edge of the MEA [
4–
6]. Therefore, the frame seal structure and attenuation behavior under actual working conditions should be modified to improve the MEA durability. A reasonable frame seal structure design will improve the durability of fuel cell, and is a prerequisite to achieving a 5000-h durability goal [
7].
There are four types of sealing structures based on the positional relationship between the frame sealing structure and the active area component [
8], including the membrane direct sealing structure, the MEA-wrapped frame sealing structure, the proton exchange membrane (PEM)-wrapped frame sealing structure, and the rigid protective frame sealing structure. The rigid protective frame structure often uses adhesives to connect a plastic frame to the outer periphery of the proton exchange membrane. A schematic representation of a typical hard protective frame structure is shown in Fig.1. The main roles of the frame are facilitating sealing, insulation, protecting the proton exchange membrane, and preventing excessive compression of MEA [
9]. Currently, the structure of the rigid protection frame is widely used due to maturity and low-cost advantages associated with the structure. However, the joint and active areas of the frame structure are weak. In addition, there are gaps between the frame and the active area components of MEA, which are caused by assembling and manufacturing errors. Consequently, the PEM in the joint-area may be damaged, thereby causing early failure. Qiu et al. [
10] used the finite element analysis method to study stress evolution in PEM during assembly, operation, and charging of the fuel cell. They concluded that the main cause of failure to the fuel cell was the damage to the membrane at the junction of the frame and the active area at different temperature, humidity, and air pressure filling conditions. Therefore, mechanical damage is considered the main reason for early failure of the PEM. When the fuel cell is working, the membrane in the gap is directly exposed to the hygrothermal environment, leading to expansion and contraction due to suction/dehydration and temperature changes [
11,
12]. The resulting hygrothermal stress reduces the fatigue life of the membrane [
13,
14]. In addition, expansion and/or contraction of the component generates shear stress on the adhesive at the junction of the frame and the membrane electrode, which may cause the frame to peel off from the membrane electrode, resulting in sealing failure in severe cases [
15].
A good frame structure design and materials increase the frame sealing structure [
16]. Ye et al. [
9] adopted a two-dimensional finite element model to study the effect of frame material and structure on membrane stress and found that a material with a large elastic modulus has a better performance compared to the membrane. Elsewhere, Shizuku et al. [
17] showed that a frame material for MEA should have a good dimensional stability. Strategies such as addition of a gasket layer [
18–
21] and application of fillers [
22–
25] at the gap between the frame and the active area have been used to limit membrane deformation and prevent its exposure to the service environment. Overall, improving the continuity of the transition from the frame to the active area can effectively increase the reliability of the frame.
The durability of the frame is generally tested by high-low temperature change or humidity cycle test, and the degree of failure of the frame is evaluated by the crossover flow rate. Kawasumi & Ikeda [
26] conducted a tensile shear bond strength test to quantify the bond strength between the electrode and the frame of the membrane. Elsewhere, Mitsuda and others [
15,
22] designed the peel strength test of the resin frame, in which low- and high-temperature nitrogen gas was alternately passed through a single cell to test the bonding strength between the membrane electrode and the frame. In addition, the US Department of Energy (DOE) proposed a humidity cycle test for evaluating the ability of membrane electrodes to resist mechanical damage [
27]. Other methods for testing the durability of MEA have also been described, in which both mechanical and chemical degradation techniques are applied as a reference for
in situ aging test of the frame [
28–
32].
To date, research on the frame has mainly been performed using the finite element simulation methods, and found no direct evidence of attenuation at the frame. Therefore, in the present paper, the mechanisms underlying mechanical attenuation at the frame were experimentally explored and the effectiveness of the improved method was evaluated. Specifically, a thermal shock rig was developed to perform the MEA in-place accelerated aging experiment to determine the effect of frequent temperature changes on the reliability of the frame structure. In addition to the basic hard protection frame structure, an improved double-layer frame structure was established. It was found that the improvement measures were effective for the two structures, and the mechanism underlying frame attenuation was revealed.
2 Experiment
2.1 Thermal shock acceleration test method
Numerous studies have reported the effect of temperature and humidity on durability of fuel cells. For instance, high and low temperature nitrogen cycle [
15,
22] and humidity cycle [
27] tests were conducted to evaluate the ability of the membrane electrode to resist changes caused by different temperatures and humidity levels. However, both methods have been associated with challenges such as the requirement of complex equipment and long test cycles. In the present paper, a thermal shock test bench was developed, which could rapidly and efficiently perform high and low temperature shock durability tests on fuel cell membrane electrodes and other components. Instead of passing high temperature/low temperature gas into the stack, the thermal shock bench alternately passes constant temperature high temperature/low temperature liquid working fluid through the cooling water channel of the original stack. Thus, the same temperature liquid working fluid can either increase or decrease the stack temperature more effectively, compared to the gas, indicating that it has the potential to achieve a shorter high-temperature/low-temperature switching cycle.
A schematic representation of the working principle of the test device is shown in Fig.2. Briefly, the test device is composed of the cold/heat source, water pump, solenoid valve and control device, tested single cell and clamping device, and a temperature data acquisition equipment (Fig.2(a)). The working fluid of the test uses ethylene glycol antifreeze for vehicles, while the cold and heat sources are used to refrigerate and heat the working fluid to a specified temperature, respectively. On the other hand, the solenoid valve is used to control switching of the low/high temperature working fluid into the single cell (Fig.2(b)).
The low- or high-temperature working fluid enters the single cell, from the original cooling flow field plate, gets divided into two parts (on both sides of the cathode and anode), then flows through both sides of the membrane electrode. This results in either a decrease or increase in cell temperature. The temperatures of the low and high temperature working fluid are set at −20 and 80 °C, respectively. In addition, the on and off valve on the solenoid is used to control temperature-switching cycle, thereby allowing access to low or high temperature working fluid. The test ends after reaching a set limit of 10000 cycles. A schematic representation of the tested single cell is shown in Fig.3(a). In summary, a single cell has an effective area of 25 cm2 and includes the PEM, gas diffusion electrodes (GDE), the anode and cathode bipolar plates (BPPs), and a water field plate outside the bipolar plates. Both plates (bipolar and water field) are made of graphite, while there are epoxy resin end plates on both sides of the single cell, which reduces the heat capacity of the device. In addition, there are rubber sealing strips between the bipolar plate and the MEA, the bipolar and the water field plates, as well as the water field and the end plates, which serve as sealing materials. The test cell is compressed with an air pump, at a pressure of 5 bar, with air used as the gas.
Several temperature measurement points are arranged in the entire test device. When the liquid enters the fuel cell, the temperature of the working fluid flowing into it is detected. Notably, the refrigerant and heating machine detect temperatures at both cold and heat sources. The temperature measurement point of the thermocouple, in a single cell, is adhered to the inlet of the flow field plate using an adhesive, while that at the cold/heat source is placed in the water tank. In the single-cell thermal shock experiment, temperatures for the cold and heat sources are set at −30 and 80 °C, respectively, which correspond to the low-/cold-start temperatures of the fuel cell of the vehicle and the internal temperature of the fuel cell during operation. The temperature at the inlet is either slightly increased or decreased, due to the energy loss of pipeline transmission. The actual impact temperature is about −20 to 80 °C. The temperatures recorded at the inlet, cold and heat sources are summarized in Fig.3(b). Both high and low temperature working fluids are passed in one shock cycle for 20 s, while the whole shock cycle lasts 40 s. Methods for testing thermal shock are shown in Tab.1.
2.2 Test specimens and methods
The general form of a single-layer frame hard protection frame structure and a cross-sectional schematic representation of the positional relationship with the bipolar plate are shown in Fig.4(a), while the double-layer frame structure with sub-frames added to the cushion layer is shown in Fig.4(b). Notably, the overall thickness of the two frame structures is suppressed, with the latter exhibiting a two-layer structure, and a thinner sub-frame lined under the gas diffusion layer (GDL). The polar plate used in the test is made of graphite, and has a rubber sealing strip in the sealing groove that presses the frame to form a seal. Both frame substrates are typical polyethylene naphthalate (PEN), using the same type of cold-pressed adhesive. The PEM material is enhanced with a commercial material, but is not coated with a catalyst as it lacks electrochemical reaction. GDL uses a commercial carbon paper, and is coated with a microporous layer. In general, the MEA components of across both frame structures are made of similar materials, with the only difference being that the double-layer frame structure GDL covers 2 mm above the sub-frame, implying that it has a larger size.
3 Results and discussion
3.1 Influence of different frame structures
The results of the tensile test, which used the INSTRON universal testing machine, revealed mechanical attenuation of the membrane at the frame after MEA temperature shock aging. The test sample was taken from the MEA sample, after accelerated aging by thermal shock. Its size was about 10 mm × 30 mm, which reflected the size of the cut membrane. Generally, the test method pulls the test specimen at a speed of 5 mm/min until it breaks. One end of the test sample is connected to the frame, while the other is only a membrane. To avoid damage to the sample and related stress concentration caused by the fixture, both ends are padded with frame substrate PEN sheets. When testing, it is necessary to ensure that the sample is aligned up and down without distortion. Upon completion of the test, the maximum load is recorded prior to breaking. The two samples are then taken from each aging sample for tensile testing, due to the limited size of the active area of the sample. The maximum load during tensile testing is shown in Fig.5.
In general, the new sample has a maximum load of about 7 N during the breaking process. Notably, there are no significant differences between single and double layers, and no evidence of the effect on the frame structure. However, thermal shock-based accelerated aging caused a decrease in the breaking load of all specimens, albeit at varying degrees. There was a slight decrease in the double-layered frame, with a breaking load of about 6 N after 10000 thermal shocks. On the other hand, the single-layer frame exhibited a marked drop, as evidenced by a load of about 4.9 N. This may be attributed to the fact that the single-layer frame is weaker and easier to break at the gap, relative to its double-layer counterpart. After aging, the single-layer frame is usually near the frame, while the double-layer one appears near the middle of the sample.
3.2 Mechanism underlying aging at the frame
To evaluate the effect of thermal shock on MEA, the sample was rinsed with water and dried. After the test, the sample was cut into small pieces, sprayed with gold onto its surface, whose structure was then observed under a scanning electron microscope (SEM). Thereafter, surface and cross-sectional methods were employed to analyze the effects of thermal shock. The former approach allows for the analysis of the thickness direction of the MEA component, whereas the latter entails surface observation. The line between the border and active areas was mainly focused on. A summary of cross-sectional and top view results for both new and aged samples for the single-layer frame made in the same batch is presented in Fig.6. Fig.6(a) and Fig.6(c) show the top views for the border between the frame and active area. Fig.6(b) and Fig.6(d) depict cross-sectional views of the frame. Briefly, there exists a gap between GDL and the frame, with no alignment between GDL of the cathode and anode (Fig.6(a)). Moreover, some stray fibers can also be seen protruding from the end of the GDL, which may be in contact with the PEM and damage the PEM when the fuel cell is working. The frame has an uneven end, while its thin structure makes the manufacturing process difficult. The proton exchange membrane in the gap produces surface cracks, after thermal shock, as evidenced by presence of cracks of different sizes and shapes (Fig.6(c)). Generally, the crack is parallel to the gap between the frame and the active area, mainly because temperature changes cause the membrane in the gap to repeatedly fold and deform at an expansion and contraction direction that is parallel to the gap. Here, the structure of the frame cross-section frame-adhesive-PEM-adhesive-frame is evident, and the PEM has a uniform thickness. Moreover, the interfaces have an excellent adhesion between them. The results of the cross section of the frame, near the gap, after thermal shock aging is shown in Fig.6(d). Notably, the frame and adhesive are well bonded, while the bonding interface between them has been destroyed.
The profiles of cross-sectional and top views of the unaged double-layer frame structure are presented in Fig.7. In summary, Fig.7(a) and Fig.7(c) demonstrate the cross-sectional view of the frame, with the left and right sides of the figure denoting the frame and active area, respectively. From the results, it is evidently observed that the GDL covers the sub-frame, the GDL on the cathode and anode are not aligned, while the MEA is slightly bent at the end of the upper frame (also the end of the GDL). The porous layer of the GDL wraps the sub-frame, which represents the desirable effect for limiting PEM deformation. Notably, both sides of the frame exhibit an excellent adherence. The analysis of the top views, after GDL removal, reveals that the microporous layer carbon powder of the GDL covers the end of the sub-frame, while carbon powder particles adhere to the PEM (Fig.7(b) and Fig.7(d)).
Furthermore, the components of the frame are in good condition after thermal shock aging. Notably, both the adhesive and membrane at the border appear wrinkled, changing from the original straight to a wavy state (Fig.7(c)). This is possibly due to the fact that the large deformed sealant strip squeezes the adhesive to both sides, following a rise in temperature. Moreover, wrinkle marks are evident on the membrane (Fig.7(d)), and these are attributed to the bending of the end position of the sub-frame following clamping of the bipolar plate. In general, the double-layered frame structure exhibits an excellent proton exchange membrane protection effect. Notably, addition of a sub-frame limits the deformation of the membrane at the frame, which subsequently prevents mechanical damage.
Differences in shape and appearance of cracks are attributed to variations in MEA positions, from which samples were taken during analysis of cracks between the border area and between the frame and the active zone. The results of characterization at each position reveal three types of cracks and damage patterns (Fig.8). Fig.8(a) is a whisker-shaped crack, Fig.8(b) is a three-pronged crack, whereas Fig.8(c) is a pinhole-shaped damage. Whisker cracks were the most predominant among them. Gradual expansion and contraction of the junction, due to asynchronous deformation across different components, caused the frame and the active area to pull each other and produce a whisker-shaped crack that was approximately parallel to the frame (GDL). The pinhole-shaped damage may have been caused by either the own damage of the PEM or the GDL fiber stabbing the PEM. Notably, there was continuous development of early cracks or damage, which caused serious damage, and compromised the material properties of the membrane. This may even cause perforation, leading to leakage.
A cross-sectional view revealed distortion of the PEM in both frame structures. Fig.8(d) and Fig.8(e) were taken from near the sealing strip of the single-layer frame structure and the edge of the double-frame GDL, respectively. Notably, the double-layered frame structure also appears in the vicinity of the sealing strip, in a similar fashion to the single-layered one. This may be attributed to the fact that a rise in temperature induces significant changes in the size of the sealing strip, which is made of silicone rubber, and increases in clamping force on the frame. In addition, there was no synchronization of deformation of the components at the frame, while the softened adhesive and the membrane were simultaneously squeezed outward. Interestingly, although lowering the temperature could not restore the wrinkles of the membrane to their original shape, it caused it to form a wavy shape in the cross section. At the junction between the frame and the active area, the membrane in the gap is subjected to the deformation of the components at the frame. Moreover, frequent wrinkles cause gradual development of cracks on the membrane. In the double-layered frame structure, the gap between the frame and the active area is eliminated by the sub-frame lined under the GDL, while GDL compresses the sub-frame, thereby limiting the wrinkles on the membrane, to effectively prevent fatigue-induced damage.
4 Conclusions
In this paper, a built thermal shock test bench was used to test two types of structures, i.e., single- and double-layered frame MEA for accelerated aging. A combination of mechanical performance tests of the PEM underlying mechanism of attenuation at the frame. Based on these results, the following conclusions can be drawn:
1) A single-layered frame structure MEA exhibits significantly lower membrane properties than its double-layered counterpart, after thermal shock aging. Notably, single-layered frame structure can easily break at the gap between the frame and the active area.
2) The membrane at the gap of single-layered structure exhibited cracks, after thermal shock. However, this was not the case in the double-layered frame structure. The shape of the crack can be attributed to a constrained relationship between frame structure and the membrane.
3) Attenuation at the frame, caused by damp heat, is mainly caused by expansion and contraction of the components, due to changes in temperature and humidity during the operation of the fuel cell. This phenomenon ultimately causes fatigue-induced damage to the material. Gradual accumulation of mechanical damage compromises reliability of MEA materials, and subsequently causes problems, such as adhesive debonding, development of membrane cracks and pinholes. These ultimately cause leakage of cathode and anode.
4) Improving the continuity between the frame and the active area, coupled with limiting membrane deformation, is key to enhancing frame durability. Notably, the double-layered frame structure not only increases the GDL area but also limits the deformation of the membrane through addition of sub-frames. Consequently, its appearance remains intact with excellent effect even after exposure to thermal shock.
5) In future research, the failure mechanism under the combined effect of mechanical stress and chemical corrosion should be focused on, and the sealing structure be further optimized based on the optimization of the rim substrate and adhesive material.
PDF
(3492KB)
