30 μm Thin Anode Gas Diffusion Layers for Optimized PEM Fuel Cell Operation at 120 °C and Low Relative Humidity

Achieving higher temperature operation above 100 °C, while increasing the volumetric power density are key targets in current fuel cell development. In this work, a novel type of polyvinylidene fluoride and graphite based microporous film of only 30 μm thickness, acting as thin alternative to typical gas diffusion layers (GDLs) on the anode side of a hydrogen fuel cell is reported. Ex situ measurements (contact angle, electric conductivity, gas permeability measurements, and scanning electron microscopy (SEM)) reveal that the thin, porous layers achieve comparable chemical, electrical, and transport properties to commonly employed and significantly thicker gas diffusion layers. In‐situ fuel cell tests reveal considerable improvements, particularly in the target operation range beyond 100 °C: At 120 °C and 30% RH and 0.65 V cell voltage, the best thin anode layer enables a 51% higher current density (1342 vs 885 mA cm−2) and a 9% lower high frequency resistance (76 vs 84 mΩ cm−2) compared to a commercial state‐of‐the‐art GDL.


Introduction
Proton-exchange membrane (PEM) fuel cells are considered a key technology to tackle global warming on a short-term time scale due to their higher efficiency over combustion engines and zero CO 2 and NO x emissions. [1]In a fuel cell, the membraneelectrode-assembly (MEA) is a critical component, which is composed of anode and cathode gas diffusion layers (GDLs) and catalyst layers and a proton-conducting membrane separating both sides. [2]GDLs ensure gas transport and electron transfer from the flow fields/ bipolar plates towards the catalyst layers and water and heat transfer vice versa. [3]erformance of the PEM fuel cells is highly dependent on the operating conditions, such as humidity and temperature, mainly due to the proton conduction mechanism of the state-of-the-art perfluoro sulfonic acid ionomers in catalyst layers and the membrane. [4]Increasing the performance of the PEM fuel cells for hot and dry stack operating conditions is of great interest for the commercial use in heavy-duty applications because of the volumetric restrictions in a truck cabin for cooling systems urges for higher operating temperatures. [5,6]Based on these results, we focused in our work on this critical operation temperature range between 95 and 120 °C.
Commercial state-of-the-art GDLs typically consist of a carbon paper with randomly aligned 7-10 μm carbon fibers, usually referred to as the gas diffusion backing.This backing is typically coated with a thin microporous layer (MPL) based on carbon particles and polytetrafluoroethylene (PTFE) as polymer binder.The overall thickness of GDLs, including the MPL, typically ranges from 150 to 250 μm, which results in a major contribution to the cell pitch especially compared to the catalyst-coated membrane (typically around 30 μm total thickness for current fuel cell applications).
To reduce this considerable thickness, there have been several reports introducing thinner, self-standing diffusion layers/MPLs as an alternative to the established GDLs.It was reported, that self-standing MPLs, in addition to enable lower thickness, might have better gas transport properties due to shorter diffusion paths [7,8] and reduce mechanical membrane degradation [9] due to a lower surface roughness.Another key difference is the narrower pore size distribution of the self-standing MPLs compared to conventional non-woven papers, which was reported to enable better fuel cell performance than with MPLs with larger pore size distribution. [8,10]o yield suitable porosity in self-standing MPLs with polyvinylidene fluoride (PVDF) as polymer binder, a phase inversion approach can be chosen: non-solvent induced phase separation (NIPS) is a very widely used technique to produce porous Achieving higher temperature operation above 100 °C, while increasing the volumetric power density are key targets in current fuel cell development.In this work, a novel type of polyvinylidene fluoride and graphite based microporous film of only 30 μm thickness, acting as thin alternative to typical gas diffusion layers (GDLs) on the anode side of a hydrogen fuel cell is reported.Ex situ measurements (contact angle, electric conductivity, gas permeability measurements, and scanning electron microscopy (SEM)) reveal that the thin, porous layers achieve comparable chemical, electrical, and transport properties to commonly employed and significantly thicker gas diffusion layers.In-situ fuel cell tests reveal considerable improvements, particularly in the target operation range beyond 100 °C: At 120 °C and 30% RH and 0.65 V cell voltage, the best thin anode layer enables a 51% higher current density (1342 vs 885 mA cm À2 ) and a 9% lower high frequency resistance (76 vs 84 mΩ cm À2 ) compared to a commercial state-of-the-art GDL.polymeric materials, especially common in filtration membrane industry. [11]During NIPS, the dissolved polymer solution is introduced to a non-solvent of the polymer as a coagulation bath.The solvent of the polymer should be miscible in the used non-solvent.Solvents such as DMAc, DMSO, NMP, or DMF are commonly used to produce porous PVDF by immersing the PVDF solution in water as non-solvent.The pore morphologies are affected by solvent/non-solvent interactions as well as the solubility of the polymer.In addition, the concentration of the solution, solvent, non-solvent media, immersion temperature, immersion duration, period between casting, and immersion has a direct influence on the pore morphology. [12]VDF-bound porous MPLs and electrodes fabricated by NIPS for PEM fuel cells were first presented by Cabasso et al. in 1996. [13]Their patent covers a variety of fabrication conditions such as 50-300 μm layer as GDL or gas diffusion electrode, a variety of solvents in the casting ink, coagulation baths, and temperatures.Park et al. investigated MPLs fabricated with different PVDF/Vulcan XC72R ratios and compared the performance with PTFE-bound MPLs on carbon cloth as GDL substrate. [14]They found that the gas permeability reduced when increasing the PVDF to Vulcan XC7R concentration in the MPLs from 4:6 to 6:4.With their PVDF-based MPLs, they achieved a comparable cell performance to a conventional PTFE-bound MPLs.Ong et al. investigated the fabrication of MPLs with NIPS and analyzed the effect of different parameters as PVDF solvent, MPL ink concentration, MPL thickness, and conductive filler for PVDF-based MPLs on through-plane resistance and gas permeability. [15]ottino et al. investigated MPLs fabricated via NIPS with 40 wt% PVDF binder, 40 wt% graphite (TIMREX HSAG 300), and 20 wt% Vulcan XC72R.They compared DMSO and NMP as solvents for PVDF and varied the coagulation durations of the NIPS process and found highest permeability with the MPL with DMSO as solvent and 8 min immersion in the bath. [16]Both works, Ong et al. and Bottino et al. reported fuel cell data by combining the novel MPLs with a conventional fiber-based GDL backing.
The previous studies prove the versatility of NIPS to create porous films with tailored porosity and their practicability as MPL in fuel cells.However, the reported electrochemical characterization data for humidified H 2 /Air fuel cell operation (<0.5 A cm À2 @ 0.65 V) was not comparable to the technical state-of-the-art today (>1.5 A cm À2 @ 0.65 V).Further, the MPLs fabricated with NIPS were not self-standing but supported on a GDL substrate.In this work, we therefore build on the described state of the art and use NIPS to develop a novel, free-standing and 30 μm thin, anode MPL with a mixture of graphite powder and PVDF as polymer binder with particular focus on performance improvement at application-relevant intermediate operation temperatures >100 °C.

Experimental Section 2.1. Materials and Ink Preparation
PVDF was purchased from Sigma Aldrich as pellets with a molecular weight of 275 000 and NMP solvent was purchased from Carl Roth (99.8% for synthesis) and used without further purification.SFG6L is used as graphite powder, supplied from Imerys Carbon, and used without further purification.
For the preparation of the 10 wt% PVDF solution in NMP, it is mixed with a magnetic stirrer at 80 °C overnight.Imerys SFG6L graphite powder, 10 wt% PVDF solution in NMP and additional NMP are mixed in glass jars for 10 min with a Hielscher UP200Ht sonicator at 80 watts.All MPL inks were prepared with 25 wt% total solids content stock solutions regardless of their PVDF/graphite ratio.

Casting and Non-Solvent Induced Phase Separation
The MPL dispersions were casted on aluminum foil with a doctor blade as 50 μm wet film.After 20 s, casted foils were immersed in DI water at room temperature.After 2 min, partially delaminated MPL layers were transferred to PTFE sheets from aluminum foils, and dried overnight in a vacuum oven at 80 °C as shown in Figure 1.The dried MPL pieces were cut to 4 cm 2 and used as self-standing MPLs without backing layer.

CCM Fabrication
The CCMs were prepared inhouse by spraying catalyst inks onto Fumatech RS710 membranes (reinforced, 10 μm nominal thickness) with a Biofluidix Nebular ultrasonic spray coater.Both cathode and anode inks featured an ionomer to carbon ratio of 0.7 and a total ink solid content of 2 wt%.For both anode and cathode 1:4 IPA:DI Water was used as solvent and 3M 800EW as ionomer.For the cathode ink, Umicore Elyst 50 PtCo/C catalyst was used, and a total platinum loading of 0.4 mg cm À2 was sprayed.For the anode ink, Tanaka TEC10E50E Pt/C was used, and a total platinum loading of 0.1 mg cm À2 was applied.

Electrochemical Characterization
Scribner 850e test stations were used for electrochemical characterization.The experimental protocol shown in Table 1 was used to test all CCMs.Cell assembly structure is shown in Figure 2a.For the cathode side, a contemporary GDL-type with state-of-theart performance (Freudenberg H14Cx483) was used for all CCMs. [17]Our self-standing anode MPLs were compared with Freudenberg H14Cx483 GDLs at the anode side.Parallel flow fields were used for both electrodes.H14Cx483 were 30% compressed with 130 μm glass fiber reinforced PTFE gasket and SS-MPLs were 60% compressed with 12 μm PET gasket.

Permeability Measurements
The permeability of the diffusion layers was calculated by pressure drop measurements [15,18] and were conducted with nitrogen gas.A Scribner 850e fuel cell test bench station was used as mass flow controlling unit.In between the flow fields nothing but the diffusion layer were fixed in a set-up, similar to the procedure for performing PEM single cell testing suggested from the U.S. Department of Energy. [19]The tested gas diffusion layers were laminated with a 40 μm PAN foil leaving a window of 4 cm 2 to ensure gas separation of the anode and cathode compartments.The N 2 flow through the tested diffusion layer was varied from 0.25 to 1 L min À1 in steps of 0.05 L min À1 at room temperature.After 2 min of stabilization the pressure drop was measured with a Chauvin Arnoux CA852 Manometer with each flow as shown in Figure 2b.

Contact Angle Measurements
Static contact angles were measured with a Dataphysics OCA25 device.An average of 10 measurements with 13 μL DI water drops was reported.

Measurement of Electrical Resistivity
The electrical resistivity was measured by assembling the anode MPL with a reference GDL in a fuel cell fixture without catalyst coated membrane (CCM), while applying the same compression.The electrical resistance was determined with the Scribner 850e potentiostat by sweeping the current from 0 to 250 mA cm À2 and applying Ohms law.

Scanning Electron Microscopy
A Tescan MAIA3 XMH electron microscope was used with 5 kV acceleration voltage, approximately 0.25 nA beam current and 7 mm working distance with a secondary electron detector.

Porosity
Porosity was calculated by gravimetric measurements of multiple 4 cm 2 SS-MPLs with a Sartorious ME36S micro scale according to the equation below, in which M is the weight of SS-MPL portion, ρ is the density, and r is the volumetric ratio of substances in the dry SS-MPL formulas.Due to the difference of binder/particle compositions, the density of the SS-MPLs were also different and calculated according to the Equation ( 1) and (2).

Results & Discussion
As described in the methods section, a flexible self-standing MPL (SS-MPL) was fabricated via NIPS, which yields a PVDF foam holding graphite flakes after immersing the ink in the nonsolvent water.This PVDF foam with graphite flakes forms a flexible, SS-MPL due to the intrinsic flexibility of the PVDF binder and the porous structures created with NIPS process.Exemplary photographs, showing the appearance and high flexibility of the obtained material are displayed in Figure 3.

Micromorphology
In SS-MPL samples, 4-6 μm graphite particles are bound by microporous PVDF network.All SS-MPL inks comprised the same solid content and only the PVDF to graphite was varied.Formation of even larger pores observed due to impregnation of immersion solution to the polymer solution.20 μm view field images, left row of Figure 4, reveal that the PVDF network have 0.1-2 μm pores on the surface.With increasing PVDF concentration, reduction in the pore sizes of the skin layers were observed.Since PVDF becomes insoluble when the solvent (NMP) leaches through the immersion bath (water), PVDF amount per volume in these porous PVDF networks should increase with increasing PVDF concentration in the SS-MPL inks.In line with our hypothesis, 40% PVDF has the lowest pore size as shown in Figure 4e.PVDF is denser at the top and the bottom end of the SS-MPL, especially visible in 35% and 40% PVDF containing samples of the cross-section images of 50 μm view field (right row).Since the skin layer of the SS-MPLs are relatively thin (<1 μm) compared to the total thickness, a very distinct pore variation across the thickness were not observed.
With increasing PVDF concentration more macro-pores were observed on the top layer (Figure 4a-e).In Figure 4f shows the commercial fiber-based H14Cx483 GDL as comparison, which  exhibited a higher roughness and more inhomogeneous pore structure compared to the SS-MPLs.In a cross section of a SS-MPLs shown in Figure 5, randomly aligned graphite flakes can be observed, showing a more homogeneous pore distribution compared to the fiber-based structure of the reference GDL.Pores between randomly stacked graphite flakes were observed to be in the range of about 2-4 μm.
PVDF concentrations between 20% and 35% show an increasing presence of pores on the substrate surface.In line with this result, 4-8 μm pores are observed frequently in the cross-section images, both scaling with the amount of PVDF-binder.At 40% PVDF, homogeneously distributed larger voids appear on the surface.One reason might be that the PVDF to solvent ratio during the NIPS process has an impact on the structure formation.Another possibility might be the faster diffusion of NMP in the PVDF solution through the bath during immersion, due to less diffusion blocking graphite flakes supporting the PVDF matrix and shorter diffusion paths.However, in all SS-MPL samples, cross-section investigation reveals the higher void fraction compared to the skin layers as shown in Figure 4 right row.
Figure 5 further shows the significant thickness difference between the reference GDL (%160 μm) and an exemplary SS-MPL with 20% PVDF content (%30 μm).Both, the SS-MPLs and the reference GDL show a homogeneous thickness.A recent study investigating multiple commercial MPLs with X-ray tomography and, confirmed that the reference GDL type used in this study (H14Cx483) from Freudenberg is homogeneous in thickness, but inhomogeneous with regards to the pore size distribution confirming our results. [20]

Permeability and Porosity
All SS-MPLs feature a higher pressure drop and lower permeability compared to the reference GDL (black solid line) as shown in Figure 6a despite the lower thickness compared to the reference GDL.Chen reported that the porosity of the Freudenberg H14Cx483 is 68%. [20]According to gravimetric measurements, the porosities of SS-MPLs are around 15% higher than that of the H14Cx483 MPL.However, as shown in Figure 4f, H14Cx483 has larger pores, which majorly contribute to the high permeability despite lower overall porosity.
For the SS-MPLs the permeability was found to decrease with increasing PVDF concentration from 20% to 35% and increase again when for a concentration of 40% as shown in Figure 6b.This is most likely linked to the strongly increased amount of macropores in the 40% PVDF sample (Figure 4e).The calculated permeabilities range between 10 À4 and 10 À3 mol s À1 Pa À1 m À2 , which is in line with previously reported values for PVDF bound MPLs [15,16] with similar fabrication conditions (NMP as PVDF solvent and %10 wt% PVDF concentration in the ink).It was reported that in hot and dry operation of fuel cells the net water transport coefficient is positive, [21] i.e., the water transport is transported through the anode to the anode outlet.Thus, a less permeable anode MPL is believed to prevent drying out of the anode and resulting in better performance. [22]

Contact Angle Measurement Results
Most commercial state-of-the-art GDLs contain PTFE as a hydrophobic treatment, both in the MPL and the backing layer, which provide a strong non-wetting behavior as the contact angle of the pure PTFE films previously reported to be above 110°. [23,24][27][28] Compared to PTFE, PVDF features a significantly lower hydrophobicity and contact angle, most likely linked to the difference in a non-fully perfluorinated chemical structure.In line with the previously reported contact angle  data of pure binder films, our contact angle measurements of the SS-MPLs indicate a stronger wetting compared to the reference which includes PTFE as a binder.Increasing the PVDF content resulted in a reduction of the contact angle ranging from %90°f or 20% PVDF to below 70°for the 40% PVDF sample, as listed in Table 2.However, it was known that surface roughness and porosity also have an influence on the contact angle. [29]herefore, most likely, this trend is related to the different surface roughness with increasing PVDF content, as evidenced in the SEM images (Figure 4a-e).The considerable standard deviation for all samples might also be caused by porosity and local surface inhomogeneity.As increased performance of hydrophilic MPLs in dry conditions was reported previously, [30,31] these contact angles encourage the use of the SS-MPL for fuel cell operation at low relative humidity.

Electrochemical Performance Results
To probe the performance of the novel SS-MPLs in the fuel cell, they were tested at elevated temperatures and reduced humidification of 1) 95 °C and 80% RH; 2) 95 °C and 50% RH; and 3) 120 °C and 30% RH.At 95 °C and 80% RH, all SS-MPLs and the reference achieved similar performances of above 1.60A cm À2 at 0.65 V with a high frequency resistance in the range of 20-30 mΩ cm 2 (Figure 7a).Only the SS-MPL with 35% PVDF content showed a slightly reduced performance at higher current densities.Since the cathode GDLs were identical, a possible explanation could be that flooding occurred at the anode side due to water accumulation at the anode after back diffusion from the cathode side.
At 95 °C and 50% RH, the HFR of the H14Cx483 reference was increased to 40 mΩ cm 2 at 1.5 A cm À2 and the performance was decreased to below 1.6 A cm À2 at 0.65 V, while the SS-MPLs reached higher current densities of above 1.9A cm À2 at 0.65 V.At 120 °C and 30% RH, the SS-MPL with 40% PVDF achieved 1.34 A cm À2 , while the reference H14Cx483 showed only 0.88 A cm À2 at 0.65 V.The difference in performance can be mostly attributed to the lower HFR of the SS-MPL samples (65-76 vs 84 mΩ @ 1 A cm À2 ).Thus, it is hypothesized that the humidity retention in the MEA with SS-MPL is improved compared to the MEA with H14Cx483 GDL.The resulting reduced ionic membrane resistance of the SS-MPL MEAs could even compromise the worse electrical conductivity of the SS-MPL with 40% PVDF, indicating the high relevance of water management at temperatures beyond 100 °C towards cell operation.
In Table 3, a compact literature review of available fuel cell performance data at 120 °C and reduced relative humidity is provided.With 1342 mA cm À2 our SS-MPL exceeds the published state of the art for fuel cells operated at 120 °C and low relative humidity (factor of 2.5-10).It is important to emphasize that our work was based on an optimized CCM with a thin, reinforced perfluorosulfonic acid (PFSA) membrane (10 μm) and shortsidechain PFSA-based catalyst layers.This was not the case in the majority of the literature references, still relying on thicker membranes and long side chain ionomers like Nafion, which are less suitable for cell operation >100 °C.However, in direct comparison to our own reference cell with commercial GDL and identical CCM we see a 51% improvement at 0.65 V at

Conclusion
Gas diffusion layers are key components to tailor the water management in hydrogen fuel cells.In our work we introduced a novel self-supported anode MPL optimized for the fuel cell operation at temperatures >100 °C and low relative humidity.The thickness of the self-supported anode MPL was only 30 μm, which is around five times lower compared to current state-ofthe-art materials.The reduced thickness helps to increase volumetric power density of fuel cells, while maintaining a flexible structure.Further, the employed phase inversion approach enabled a homogeneous and well controlled tailoring of the MPL morphology and surface.The thin MPLs feature a comparable permeability and a more hydrophilic wetting behavior.In fuel cell tests, the novel anode MPLs enabled a more than 50% higher current density at 0.65 V and 120 °C and 30% RH compared to a cell with a conventional state-of-the-art GDL.The main reason for the improved performance is a lower high frequency resistance of 65-76 mΩ cm 2 compared to that of the commercial reference GDL (84 mΩ cm 2 ), that indicates the positive influence of the MPL on the CCM water management.The results highlight the potential of novel GDL compositions and asymmetric anode/cathode GDL-combinations to enable significant improvements of fuel cell operation at elevated temperatures > 100 °C.

Figure 1 .
Figure 1.Schematic illustration of the fabrication of self-standing micro porous layers via non-solvent induced phase separation (NIPS).

Figure 2 .
Figure 2. a) Cell assembly for electrochemical characterization, b) In-house pressure drop measurement setup.

Figure 5 .
Figure 5. Thickness comparison of the a) state-of-the-art reference and b) herein developed self-standing MPL with 20% volumetric PVDF.

Figure 6 .
Figure 6.a) Pressure drop measurements of the diffusion layers b) calculated (nitrogen) permeability values with respect to volumetric PVDF content.

Table 1 .
Fuel cell characterization protocol used in this work.

Table 2 .
Summary of ex situ characterization results of the different GDLs.À4 mol s À1 Pa À1 m À2 ]

Table 3 .
Overview of published fuel cell polarization data at 120 °C and low relative humidity.