A polymer electrolyte membrane fuel cell (PEMFC) has been developed as a clean and efficient energy conversion system for decarbonized mobility and distributed power generation because of its high efficiency and power density, low operating temperature, and fast start-up capacity relative to conventional energy conversion devices [1–4]. For realizing the widespread commercialization of PEMFCs, there are still technological hurdles including high cost, performance loss, and inferior durability with long-term cycle [5]. Among the key components of the membrane-electrode-assembly (MEA) in PEMFCs, a gas diffusion layer (GDL) plays a crucial role in simultaneously providing mechanical support, electrical conduction, and gas/water transport pathways between the catalyst layer (CL) and the flow channels of bipolar plate [6]. As GDLs govern the distribution of reactant gases and water within the electrode, the change in their physical properties and microstructures or structural states under compression has significant effects on the electrochemical reaction and overall cell performance [6–8]. Thus, understanding the compression–transport coupling behavior of GDLs is essential for their design to optimize the performance and long-term durability of MEAs in PEMFC applications.

Thin GDLs have attracted great interests in advanced PEMFC systems to achieve high volumetric power density and enhanced water and heat management. Thin GDLs facilitated water removal from the cathode, enabling rapid recovery from flooding during dynamic operation [7,8]. In addition, reducing the GDL thickness decreased the overall stack weight/volume and the electronic conduction path, and improved through-plane (TP) electrical conductivity [9,10]. However, thin GDLs have several critical drawbacks. With decreasing the GDL thickness, the relative fraction of the MPL-penetrated region increased and the pore connectivity within the carbon fiber substrate decreased [9–15]. Thin GDLs exhibit higher compression sensitivity: their pore structures collapse more easily under high compression and this lead to decrease gas permeability and mass-transport capability, while insufficient compression results in poor interfacial contact and elevated through-plane electrical resistance [12–15]. It has been widely recognized reported that the electrochemical performance of thin GDLs critically depends on the compression-induced microstructural deformation, including thickness reduction, pore-volume loss, and MPL intrusion, which influences in-plane (IP) and TP transport phenomena [11–13]. Furthermore, because thin GDLs are characterized by a shallower porous network, they are more susceptible to both flooding and dehydration, which narrows their operational window and degrades long-term durability [16–18]. In this regard, the balance between microstructures and gas diffusion under compression is also considered in thin GDL design for PEMFC applications.

Early studies have reported the fundamental trade-off between compression and performance. Wang and Nguyen [7] demonstrated that higher clamping pressures improved interfacial contact but mass-transport limitations were occurred under over-compression, and Nitta et al. [12] quantified this trade-off relation by means of compression–thickness–resistance curves. Kaviany et al. [11] investigated the modeling on effective diffusivity and water saturation to compression-induced morphological changes. Zenyuk et al. [15] observed the non-uniform fiber deformation under mechanical load via X-ray tomography, and identified abrupt pore-volume loss beyond critical stress levels during severe densification. On the multiphase perspective, it has been reported that the increased compression deteriorated water accumulation and diffusion resistance under highly humidified conditions [16–19]. Using high-energy operando X-ray Compton scattering imaging, Miyazawa et al. [18] have directly visualized heterogeneous liquid-water distribution inside operating PEMFCs, revealing localized water accumulation within GDLs that leads to transport blockage and oxygen starvation. For thin GDL design, Yildirim et al. [10] reported that the self-standing ultra-thin (~30 μm) carbon papers delivered high-current performance at elevated temperatures and low humidity when pore-size distribution and mechanical stiffness were properly optimized. Numerical and theoretical studies reflected the non-linear relationship between compression, contact resistance, and transport behavior. Eller et al. [19] reported that the controlled pore connectivity and MPL integrity were crucial for sustaining oxygen diffusion in thin GDLs by experimental verification. Wen et al. [20] employed a multi-physics model incorporating non-linear stress-strain behavior to identify the optimal compression ranges to minimize the contact resistance without sacrificing the mass transport. Collectively, the existing literatures have highlighted the consistent findings, including (a) the trade-off between contact resistance and gas diffusion under compression, (b) the influence of MPL penetration and pore connectivity, and (c) the enhanced compression sensitivity in thinner GDLs. Nevertheless, the quantitative interrelation between IP and TP transport, the impact of MPL-penetration ratio on diffusion and conduction, and the mechanical design guidelines for simultaneous optimization of thickness, compression pressure, and pore connectivity remain insufficiently resolved. More recent studies on thin GDLs for PEMFCs have focused on compression effects and gas transport characteristics, and employed advanced imaging, finite element method (FEM), computational fluid dynamics (CFD) simulations, and lattice boltzmann method (LBM) modeling to quantify structural deformation and its impact on mass transport and performance [21–25]. Chen et al. [22] demonstrated that compression ratio yields best trade-off between resistance and performance, and optimal performance was obtained at compression ratio of 37.5%. Advanced modeling and experimental validation provide clear guidelines for industrial application of thin GDLs [23].

This study aims to address the compression and transport coupling behavior in thin GDLs. We investigated the effect of microstructures and gas permeability of GDLs on the compression-related transport behavior by using different types of GDLs with identical MPLs but different backing-layer (BL) thicknesses. The structural characteristics of GDLs were analyzed using scanning electron microscopy (SEM) and mercury intrusion porosimetry (MIP), and the compression-dependent variations in thickness, electrical resistance, and IP/TP gas permeability were measured. Furthermore, I–V polarization and electrochemical impedance spectroscopy (EIS) analyses were performed to clarify the cell resistance components within the context of compression–transport behavior. In this study, we suggest the design guide of thin GDLs, emphasizing the need for sufficient mechanical strength, effective pore connectivity, and optimal compression to prevent excessive densification.

To clarify the influence of GDL characteristics on the electrochemical performance of the MEA, different types of GDLs employing the same MPL design were compared. The characteristics of GDLs used in this study were summarized in Table 1. As expected, total thickness and basis weight decreased proportionally with decreasing the BL thickness. Fig. 1 showed cross-sectional SEM images of different types of GDLs, and overall thickness of GDLs was consistent with the manufacturer’s specifications. However, it was difficult to identify the distinct boundary between the BL and the MPL. It can be seen that MPL penetrated approximately 50~100 μm into the carbon-fiber substrate in GDLs. Assuming that the penetration depth is similar across the GDL, it can be deduced that the relative fraction of the penetrated MPL increased with decreasing the total thickness of GDLs. In the case of GDL#C, MPL seems to be extended deeply into the carbon fiber-based substrates, likely reducing the effective pore volume available for gas diffusion.

Fig. 2(a) and (b). showed the pore volume and pore-size distribution of different types of the GDL obtained by the MIP. In this study, the pore size ranges were classified into three regions of a) ≤ 0.5 μm, b) 0.5~5 μm, and c) 5~50 μm, and the volumetric ratio of each region was then evaluated. The pore region of 5~50 μm mainly corresponds to macropores or cracks originating from the BL. Among the GDL samples, GDL#B exhibited a slightly higher pore volume in the 5~50 μm region and high total pore volume, which can be inferred to the different degree of MPL penetration into BL. For GDL#/C, a larger MPL penetration and the reduced pore connectivity were prone to constrain the effective TP pathways, whereas GDL#A retained more continuous networks of large pores beyond the MPL penetrated region, compensating the TP diffusion.

As shown in Fig. 2(c) and (d), IP permeability of GDLs decreased with increasing compression pressure. In the low-pressure region (≤ 0.5 MPa), GDL#B exhibited the highest IP permeability and TP electrical resistance, implying that open fiber networks promoted lateral diffusion but resulted in incomplete electrical contact. Conversely, GDL#C with thinner BL and deeper MPL penetration showed denser internal structures and lower TP resistance at the same compression. With further compression (> 1 MPa), there was a pronounced decrease in TP resistance of GDLs, accompanied by the reduction in IP permeability. Beyond a threshold pressure, carbon-fiber networks compact and enhance the interfacial contact between GDL and bipolar plate, but narrowing the gas-diffusion pathways. The differences in gas permeability and electrical resistance among the GDL samples diminished, indicating convergence to mechanically stabilized and electrically uniform interfaces. This revealed the mechanical–electrical relation within GDLs: more open pore architectures enhanced the mass transport containing IP diffusion, but compromised TP electrical conductivity at low level of compression. The TP gas permeability of GDLs decreased with decreasing the GDL thickness, which is consistent with the results for their microstructure analysis.

Fig. 3(a) shows the I–V polarization curves of the MEA with different types of GDLs under atmospheric conditions at 65°C and 100%RH. The MEA with GDL#B showed the best performance in the overall current density region. The voltage difference between MEA samples was observed at around 0.7 A∙cm^-2, and was more pronounced at > 1.5 A∙cm^-2. It implicated that larger voltage drops for GDL#A and GDL#C were caused by water vapor accumulation and gas diffusion limitations occurred in the MEA. Appropriate pore continuity and MPL penetration of GDL#B led to uniform oxygen transport with suppressing water vapor accumulation, resulting in the highest MEA performance even at > 1.5 A∙cm^-2. We deduced that there was a complex interplay of insufficient gas distribution capability within the electrode, the increased water vapor accumulation, and drainage inefficiency by the weakened capillary gradient. The MEA with GDL#C exhibited similar diffusion limitation, which was attributed to the reduction in effective pore size by large MPL penetration and the localized barrier formation by water vapor. The effect of GDL thickness on the MEA performance resulted from the combined effect of compression sensitivity, gas diffusion, and water management, not merely electrical resistance and geometric thinning. The GDL#B that balanced these factors, ensured uniform gas transport and effective drainage stability under actual assembling pressure condition (≈ 1 MPa), and thus maintained higher cell performance at > 1.5 A∙cm^-2 relative to GDL#A and GDL#C.

Fig. 3(b) shows the Nyquist plot of the MEA with different GDLs, recorded at 1.0 A∙cm^-2 under 100%RH. All the measurements were performed at different current density (Supporting Information, Fig. S1), and the measured impedance diagram was separated into the resistance components such as R_m, R_ct, and R_tr using an equivalent circuit model. The MEA samples exhibited similar resistances in the high-frequency range, while different impedance responses were observed in the mid- and low-frequency ranges. There are more notable changes in the resistance at the mid- and low-frequency region compared to the high-frequency region. The MEA with GDL#B showed the lowest overall resistance and significant decrease of semicircular radius in the mid-frequency region and low-frequency region. The kinetic resistance observed at high-frequency region for the MEA with different types of GDLs showed a similar behavior, while the mass transport resistance observed at low frequency region was significantly increased. This result indicated that the decrease in the MEA performance mainly resulted from the increase in the mass transport resistance. The MEA with GDL#A and GDL#C showed higher mass transport resistance compared to the MEA with GDL#B. In general, R_ct is directly related to the charge transfer rate and the effectiveness of the active area in the CL [26]. The MEA with GDL#B showed the lowest R_ct among MEA samples, resulting from maintaining the oxygen transport to CL because of proper compression characteristics and uniform gas distribution. However, GDL#A had relatively low IP gas permeability and led to the non-uniform oxygen supply across the electrode, resulting in low oxygen transport and high R_ct. The MEA with GDL#C showed a relatively high R_ct and R_tr, which was consistent with IP gas permeability and compression sensitivity. The GDL#A containing a large pore volume exhibited a relatively weak capillary pressure gradient, leading to the retardation of water drainage. In addition, a large MPL penetration of GDL#C limited the actual gas passage, and increased the gas diffusion resistance. To analyze the influence of different types of GDL on the MEA performance, we evaluated activation, ohmic and concentration overpotentials and compared them. As shown in Fig. 3(c) and (d). the MEA with GDL#B showed lower overpotential value than those of GDL#A and GDL#C. The MEA with different types of GDLs showed almost similar activation and ohmic overpotentials values (Supporting Information, Fig. S2). In high current density regions, MEA performance is mainly governed by the concentration overpotential. As the MEA with GDL#B exhibited the lowest total overpotential among MEA samples, it presented the highest MEA performance at > 1.5 A∙cm^-2.

The variation of the MEA performance with different types of GDLs mostly stemmed from the interplay between mechanical compression and gas transport in GDLs (as illustrated in Fig. 4). The transport and accumulation of water vapor/liquid water generated in the cell during operation has a significant effect on the MEA performance, largely governing by GDL characteristics. During cell operation, water generated in the CL migrated through hydrophobic MPLs toward the BL of the GDL and eventually to the flow channels of bipolar plates. Most of the accumulated waters tend to condensate under the rib region facing the BL of the GDL, whereas the condensation rarely occurs inside the flow channels. Such water redistribution and local saturation affect the MEA performance at high current density region, where the extent of accumulation is determined primarily by the microstructures and hydrophobicity of the GDL. Water condensation limitation diminished with decreasing BL thickness and pore size, and led to the decrease in MEA performance under high RH and high current density region. In this regard, it should be required for the balance between the controlled hydrophobicity for facilitating water diffusion from CL to BL and the optimized GDL microstructures to alleviate water condensation on the ribs as well as to compensate mass transfer near the flow channel. In the case of GDL#B, the balance between MPL penetration and pore continuity enabled efficient oxygen transport and water drainage under high RH conditions, leading to the lowest R_ct and R_tr values. Therefore, this study demonstrated that the MEA performance were optimized by balancing the gas diffusion with mechanical compression of the GDL, providing design guide of thin GDL with great potentials for PEMFC applications.

We investigated the effect of the compression behavior, microstructure, and electrical properties of different types of GDL with identical MPL on the electrochemical performance of the MEA, and clarified the influencing factors for adapting thin GDL in the PEMFC applications. As the GDL was thinner, the compression sensitivity increased and structural stability reduced. The intermediate GDL (GDL#B ≈ 215 μm) tended to maintain an intimate contact with the balance of compressibility and mechanical flexibility. However, the thin GDL (GDL#C ≈ 185 μm) showed the reduced effective porosity due to the increased MPL penetration and structural collapse. The TP gas permeability of GDLs was reduced with decreasing the GDL thickness, and the GDL#B showed the highest IP gas permeability. The electrochemical analysis revealed that the higher MEA performance of GDL#B was attributable to the lowest charge R_ct and R_tr. The lower MEA performance of GDL#C resulted from the oxygen diffusion limitation and water vapor accumulation caused by MPL penetration and pore collapse. the precise control of MPL penetration and distribution for efficient pore continuity, the mechanical reinforcement for compression stability, and the improvement in TP and IP gas permeability should be required for achieving the commercial-level MEA performance by applying thin GDL. This study suggested that MEA performance was optimized by balancing the gas diffusion continuity with mechanical compression of GDLs, providing design guide for engineering thin GDLs for high performance MEAs in PEMFC applications.