Patterned catalyst layer boosts the performance of proton exchange membrane fuel cells by optimizing water management

Yingjie Zhou,Wenhui Zhang,Shengwei Yu,Haibo Jiang,Chunzhong Li

Key Laboratory for Ultrafine Materials of Ministry of Education,Shanghai Engineering Research Center of Hierarchical Nanomaterials,School of Materials Science and Engineering,East China University of Science &Technology,Shanghai 200237,China

Keywords:Water management Mass transfer Patterned catalyst layer Proton exchange membrane fuel cells Finite element analysis

ABSTRACT Mass transport is crucial to the performance of proton exchange membrane fuel cells,especially at high current densities.Generally,the oxygen and the generated water share same transmission medium but move towards opposite direction,which leads to serious mass transfer problems.Herein,a series of patterned catalyst layer were prepared with a simple one-step impressing method using nylon sieves as templates.With grooves 100 μm in width and 8 μm in depth on the surface of cathode catalyst layer,the maximum power density of fuel cell increases by 10% without any additional durability loss while maintaining a similar electrochemical surface area.The concentration contours calculated by finite element analysis reveal that the grooves built on the surface of catalyst layer serve to accumulate the water nearby while oxygen tends to transfer through relatively convex region,which results from capillary pressure difference caused by the pore structure difference between the two regions.The separation of oxidant gas and generated water avoids mass confliction thus boosts mass transport efficiency.

1.Introduction

With the advantages of higher energy conversion efficiency than combustion engines,higher energy density than batteries and zero pollutant emission,fuel cells have been regarded as one of the most promising power generation systems of the 21st century [1].Among them,the proton exchange membrane fuel cell(PEMFC) shows a good start-stop capability and operates under a relatively low temperature,which make it suitable to be used in vehicles and portable devices [2,3].Despite so many advantages,some problems remain to be solved for the further commercialization of PEMFCs,such as high catalyst cost[4],unsatisfactory durability [5–7] and poor water management [8].Many efforts have been done to synthesize catalysts with high activities and low Pt loadings,for instance,alloying platinum with other transition metals,constructing core–shell structures or fabricating single-atom catalysts[9,10].Novel catalyst layer(CL)structures have also been built to enlarge the ‘‘triple-phase boundaries (TPB)”[11,12].Besides a higher catalytic activity,an optimized mass transport is also necessary for an ideal fuel cell performance.A certain humidity is needed to facilitate the proton transfer across the polymer electrolyte membrane (also called proton exchange membrane,PEM) [13].Meanwhile,excessive water generated in the cathode catalyst layer should be efficiently drained so as not to hinder oxygen transport onto the surface of catalyst particles.Unfortunately,the oxygen and the generated water share the same transmission medium but move towards opposite directions,which both impedes the feed in of oxygen and drainage of water [14].

Forner-Cuencaet al.demonstrated that a gas diffusion layer(GDL)with intermittent hydrophilic regions has the ability to provide a separate pathway for product water to leave the cell [15].Similarly,Wanget al.perforated GDL with laser to build ordered holes with appropriate diameter,which served as ways to collect the water nearby thus enhancing water removal [16].Mark?tteret al.found that carbon fibers in the GDL act as guiding rails for water droplets,which provides the idea of building oriented electrodes to improve mass transport[17].In addition to the modification of GDL,efforts have also been made on the structure of catalyst layer to optimize mass transport.By depositing Pt nanoparticles on vertically aligned carbon nanotubes,Tianet al.prepared an oriented catalyst layer,which shows a better performance under high current densities [18].Leeet al.designed a catalyst layer with in-plane flow channel perpendicular to the flow field direction,which demonstrates significant enhancement of mass transport property because of facilitated molecular and Knudsen diffusion [19].Modifying the pore structure of catalyst layer has been proven to be another practical way to optimize gas diffusion [20,21].Meanwhile,numerical simulation and machine learning are increasingly used to provide validation and guidance for experiments [22,23].Bazylaket al.illustrated the dynamic eruption mechanism of water droplets in GDL with the help of computational fluid dynamics [24].Jiaoet al.developed a highaccuracy data-driven surrogate model which realizes the efficient multi-physics-resolved digital of PEMFCs [25].

In this work,a simple one-step impressing process was introduced to prepare patterned cathode catalyst layer.Nylon sieves of different sizes were used as transfer printing templates.Several impressing pressures were exerted to construct grooves of different depths.The single cell performances of the patterned and untreated catalyst coated membranes (CCMs) were tested and compared.With the help of finite element analysis (FEA) simulation,the reason for better mass transport management was explained.

2.Experimental

2.1.Preparation of surface-patterned cathode catalyst layer

The patterned catalyst layer was prepared through a simple impressing process as shown in Fig.1.Gore PRIMEA CCM,with a platinum loading of 0.15 mg?cm-2on the anode side and 0.40 mg?cm-2on the cathode side,was chosen for the impressing process.The CCM was placed between two thin polytetrafluoroethylene (PTFE) films to avoid physical damage while pressed and to make sure it can be easily peeled off after the procedure.A nylon sieve was then placed on the cathode side of the PTFE film coated CCM and these layers together were pressed for 10 s under certain pressure (0–8 MPa) at room temperature.After stripping off nylon sieve and PTFE layers,a CCM with patterned cathode catalyst layer was got.

2.2.Characterization of the patterned catalyst layer

The surface of the CCMs was photographed by a digital camera.The cross-section morphology of the catalyst layer was characterized by scanning electron microscope (SEM,Hitachi S-4800 N,Japan).The depth of the patterning structure was detected using a laser scanning confocal microscope (Keyence VK-X100K,Japan).The hydrophobicity of the gas diffusion layer and the microporous layer (MPL) was measured with a contact angle meter (Harke SPCA-X3series,China).The pore distribution within the CCM was measured by mercury intrusion porosimetry (Micromeritics AutoPore IV 9500,USA).

Fig.1.Schematic diagram of(a)impressing process,(b,c)side view of the plain and patterned CCM,(d) perspective view of the patterned CCM.

2.3.Single cell test

The CCM was placed between two gas diffusion layers with hydrophobic microporous layers from Shanghai Hesen Electric Co.LTD to form a membrane electrode assembly (MEA) with an active area of 4 cm2.A graphite bipolar plate with serpentine flow channel was chosen for fuel and oxidant gas feeding (Digital photographs of the single cell test fixture are shown in Fig.S1).The cell was tested at 333 K under a back pressure of 50 kPa[26].The flow rates of hydrogen and air were 110 and 275 ml?min-1,respectively.Both fuel and oxidant were humidified to 100%RH before entering the cell.Before testing,the cell was activated under a series of current densities until its performance no longer obviously improved over time.Polarization curve of the cell was measured using Chroma 63123A electronic load.The polarization curves were continuously tested three times and averaged.Then the gas for cathode was changed from air into nitrogen.The flow rates for both hydrogen and nitrogen were altered to 200 ml?min-1.The back pressure was altered to 100 kPa [27].Cyclic voltammetry (CV)and linear sweep voltammetry (LSV) were measured by Gamry interface 5000E electrochemical workstation.

3.Results and Discussion

3.1.Patterned catalyst layer with different groove spacing and depth

Nylon sieves with different hole sizes(50,80,100,300 and 800 mesh)were used to construct catalyst layers with different groove spacing.The digital photographs of sieves and patterned catalyst layers are shown in Figs.2(a),(b)&S2.The shape of the nylon template was well transferred onto the catalyst layer after the impressing procedure while grooves built on the surface of the catalyst layer are slightly wider than the diameter of the nylon wire.The width of nylon wires used in 50 mesh sieve is wider than others,which resulted in an excessively compressed CL (Fig.S3(a)) and mechanically damaged PEM.The hollow out ratio in 800 mesh sieve is extremely low,which impeded the formation of the convex regions (Fig.S3(b)).When using 100 mesh sieve as template,both the groove area proportion and the groove depth are moderate,which optimized the mass transport in the cell (Fig.S4).In that case,100 mesh nylon sieve was chosen as the template for the following experiments.

The cross-sectional views of the untreated CCM and patterned CCM are shown in Fig.S5.The untreated CCM has a relatively plain surface while an undulating structure can be seen on the surface of the patterned CCM.Photographs taken by laser scanning confocal microscope demonstrates the concavoconvex structure built on the surface of the cathode catalyst layer more clearly.The depth of grooves generated with different impressing pressure (2,4,6 and 8 MPa) is approximately 4–6,8–10,10–12 and 12–14 μm,respectively (Figs.2(c),(d) &S3(c)– (f)).

Fig.2.Digital photograph of (a) 100 mesh sieve,(b) patterned CCM.Laser scanning confocal microscope photograph of (c) plain CCM,(d) patterned CCM.

Fig.3.Single cell performance of the CCMs with different impressing pressure:(a) polarization curves,(b) comparison of rated and maximum power density,(c) cyclic voltammetry curves from which ECSA can be calculated,(d) linear sweep voltammetry curves which show the hydrogen crossover current density.

Polarization curves of MEA with cathode catalyst layer treated with different pressure are shown in Fig.3(a).Cells with patterned CCMs show better performances compared with the untreated plain one,especially in the concentration polarization region.While the groove depth of the catalyst layer reaches 8 μm (imprinted with 4 MPa pressure),the cell shows the best performance—a maximum power density of 705 mW?cm-2,which is 10% higher than the plain one’s.Further increase in impressing pressure may lead to a decline in performance (Fig.3(b)).Fig.3(c)shows cyclic voltammetry curves (scanning potential range from 0.1 to 1.2 V at a scan rate of 100 mV?s-1) of the cell with plain and patterned catalyst layers.The peak positions in the CV curves are almost the same,and electrochemical surface area (ECSA) also shows no significant difference(approximately 40.0 m2?g-1).Since the pressing procedure was done under macro scale and hasn’t changed the micro structure of the catalyst,no additional active site is formed,which is consistent with the similar performance shown in the low current density region in polarization curves.To characterize the integrity of the PEM,hydrogen crossover was measured (linear sweep voltammetry from 0.1 to 0.8 V at a scan rate of 10 mV?s-1).Treated with appropriate pressure (less than 6 MPa),the CCM shows similar or even lower hydrogen crossover current densities (Fig.3(d)),indicating that the PEM hasn’t been physically damaged during the treatment [28].However,when the pressure increased to 8 MPa,the crossover current density increases sharply,which may be the reason for its unsatisfactory single cell performance.

Mercury intrusion porosimetry results demonstrate that the pore size distribution of plain and patterned CCMs are similar,with a peak at 50 nm and another peak at 10 μm.The main difference lies in the porosity.The volume of 20–100 nm-diameter pores of the patterned CCM is lower than that of plain CCM(Fig.S6),which can be attributed to the condensation of catalyst regions impressed by nylon wires(regions under the grooves).The groove on the surface of the catalyst layer has a width of 65–70 μm,which cannot be detected by the mercury intrusion porosimetry.In that case,the patterned catalyst layer can be divided into three parts:the compressed part (which has a small pore size and low porosity),the convex part(which has a medium pore size and medium porosity)and the groove part(which can be considered to have a large pore size and extremely high porosity).The relatively porous structure near the GDL facilitates the mass transport while the relatively compact structure near the PEM meets the demand of more active sites,which has been proved beneficial for PEMFCs [29].On the other hand,due to lower capillary pressure of the larger pores,water tends to accumulate towards the groove part,thus the smaller pores in the condensed and convex part of the catalyst layer are left free for the transport of oxygen [30],which ensures enough oxygen to be transferred to the surface of catalyst particles for electrochemical reaction.

To evaluate the durability of the surface-patterned CCM,an accelerated durability test (ADT) was carried out (IEC TS 62282-7-1:2017).The cell ran alternately under two loads,20% rated power density and 100% rated power density,each lasting 30 min,and the resulting polarization curves of the plain and patterned CCM after certain ADT time are shown in Fig.4(a),(b).For the plain CCM,the single cell performance slightly declined in the first 80 h.When the ADT time had reached 130 h,the performance degraded instantly,losing approximately 20%of the original power density,indicating the end of the cell life.The polarization curves of cell with patterned CCM (8 μm groove depth) shows a similar trend,stable during the first 100 h and reaching an end of life after 140 h (losing more than 10% original power density).Both plain and patterned CCMs show obvious ECSA decrease(30% for the plain one and 10% for the patterned one) after the ADT(Figs.4(c) &S7(a),(b)),which may be a result of catalyst particle agglomeration [31].Compared to a stable loading,catalysts deteriorate faster under fluctuating loads.The performance of patterned CCM is more stable under high current density due to a better mass transport efficiency,which means the cell endures less drastic load fluctuation during the ADT (Fig.S8),so the decay rate of catalysts in the patterned CCM is rather slower compared with the plain CCM.On the other hand,the hydrogen crossover current of both remain approximately 3–4 mA?cm-2and haven’t shown remarkable uptrend(Figs.4(d)&S7(c),(d)),indicating the mechanical integrity of CCM throughout the ADT.In summary,a patterned CCM gains a higher power density without causing additional durability loss compared with plain one.

Fig.4.Single cell performances during the accelerated durability test:(a)polarization curves of the plain CCM,(b)polarization curves of the patterned CCM,(c)ECSA of the plain and patterned CCM,(d) hydrogen crossover current density of the plain and patterned CCM.

3.2.FEA simulation

To further explain the reasons for the improvement of single cell performance,Ansys Fluent was used to simulate charge transport and mass transport in the PEMFC.A three-dimensional single cell model with plain CCM was built and divided into hexahedral grids (Fig.5(a)).The size of the single cell model and the physicochemical parameters used in the simulation are listed in Table 1.

Table 1 Cell dimensions,operating conditions and material properties.

First,the grid independence was tested with four different mesh systems (84,586,676,544,1,937,380 and 4,489,790 elements).When the mesh elements have reached 2 million,further doubling the element number may only result in a relative error less than 1% (Fig.5(b)),indicating this kind of mesh density is already enough for the simulation.Then,a comparison between the experimentally measured and numerically predicted single cell polarization curves for the PEMFC with plain cathode catalyst layers was made(Fig.5(c)).The simulation result shows a good agreement with the experimental one,confirming the validation of the model and calculating parameters.

To investigate the groove built on the cathode catalyst layer surface’s influence on the mass transport in detail,a model with a patterned catalyst layer should be built.As the scale of the concavoconvex structure is relatively smaller than the plain layer,the mesh density was further increased(four times the density used in the above-mentioned model) to ensure the grid independence.To save the computing resource,the 2 cm×2 cm model with serpentine flow channel was simplified to a 9.76 mm × 2.56 mm model with one straight flow channel(Fig.5(d)–(f)).A model of same size with plain CCM was also built to be compared with the patterned one.

Simulated polarization curves of plain and patterned model are shown in Fig.S9.The cell with a patterned catalyst layer shows a20% improvement in maximum power density compared with the plain one.Although the absolute value of the power density isn’t exactly same as the experimentally measured one due to model simplification,the performance improvement after the impressing treatment (especially in the mass transport region)agreed with our experiment.

The simulation results are shown in Fig.6(a)displays the O2distribution within the cathode catalyst layer,microporous layer and gas diffusion layer of a cell with plain CCM.The O2molar concentration under the flow channel is obviously higher than that under the rib.It can also be seen that the O2concentration decreases progressively from gas diffusion layer to catalyst layer,which is consistent with the fact that O2was reacted in the catalyst layer.Fig.6(b) demonstrates the H2O distribution within the same place as Fig.6(a).Unlike O2,the concentration of H2O within catalyst layer is higher than that in gas diffusion layer as a result of water generation in the catalyst layer.

Fig.6(c) shows O2distribution within a cell with patterned CCM.The molar concentration of O2in the grooves is much lower than that in the convex parts,indicating oxygen pretend to transfer in relatively convex area.In contrast,molar concentration of H2O in the concave grooves is much higher (Fig.6(d)),indicating water pretend to accumulate in groove regions of the patterned catalyst layer.This kind of mass distribution can be extended to the microporous layer and the gas diffusion layer.Based on these results,we can conclude that the impressing procedure has built separated transport highways for oxygen and water to go through,thus avoiding the confliction between oxygen flow and water flow.With such ordered mass transport,a higher power density is finally generated compared to the plain one.

4.Conclusions

A simple method to prepared CCM with patterned structure is presented.With a nylon sieve template of proper size and imprinted under proper pressure,ordered grooves can be built on the surface of the catalyst layer.Single cell test results show that a catalyst layer with grooves 100–120 μm in width,8 μm in depth and spacing distance of 200–220 μm generates a superior performance of 705 mW?cm-2,which is 10% higher than that of the plain one’s.Cyclic voltammetry test results prove that ECSA remains almost the same after patterning.The grooves serve to accumulate the water generated nearby thus small pores in the catalyst layer are left free for oxygen to transfer through,which optimizes the mass transfer in CL.Accelerated durability tests prove that the patterned CCM with a better mass transport efficiency has a longer life.FEA simulation further verifies that water tends to transfer through groove area while oxygen tends to transfer through relatively convex region.The separation of oxidant gas and generated water in MEA avoids mass flow confliction thus boosts mass transport efficiency.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by the National Natural Science Foundation of China(21838003,91834301),the Shanghai Scientific and Technological Innovation Project(18JC1410600,19JC1410400),the Social Development Program of Shanghai (17DZ1200900),the Innovation Program of Shanghai Municipal Education Commission,and the Fundamental Research Funds for the Central Universities(222201718002).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2021.06.001.