Undoped Polybenzimidazole Membranes Composited with CeP₅O₁₄ for Use in Hydrogen Fuel Cells at 200 °C

Abstract

Intermediate-temperature (IT) proton-exchange membranes (PEMs) play vital roles in hydrogen and direct liquid fuel cells, electrolyzers, and other electrochemical membrane reactors at elevated temperatures of higher than 150 °C. This article reports the fabrication and performance assessment of a type of new IT polymer–inorganic composite (PIC) PEMs that were made of cerium ultraphosphate (CeP₅O₁₄-CUP) as the durable solid-state proton conductor and undoped polybenzimidazole (PBI) as the high-temperature (HT) polymeric binder. The proton conductivity and electrochemical performance of the PIC PEMs were characterized at 200 °C with varying membrane thickness, processing parameters, and operating conditions using a single-stack hydrogen fuel cell connected to a fuel cell test station. Experimental results show that the PIC membranes (with CUP of 75 wt.%) carried high mechanical flexibility and strength as well as noticeably reduced water uptake of 4.4 wt.% compared to pristine PBI membranes of 14.0 wt.%. Single-stack hydrogen fuel cell tests at 200 °C in a humidified hydrogen and air environment showed that the proton conductivity of the PIC PEMs was measured up to 0.105 S/cm, and the electrochemical performance exhibited its dependence upon the membrane thickness with the power density of up to 191.7 mW/cm². Discussions are made to explore performance dependence and improvement strategies. The present study expects the promising future of the IT-PIC-PEMs for broad applications in high-efficiency electrochemical energy conversion and value-added chemical production at elevated temperatures of 200 °C or higher.

1. Introduction

Proton-exchange fuel cells (PEMFCs) have emerged as a type of leading electrochemical energy conversion device in clean energy technology, owing to their high efficiency, low emissions, and sound scalability [1,2,3]. Yet, conventional PEMFCs are operated at relatively low temperatures (LT) (typically below 100 °C), which limits their performance in terms of efficiency, fuel flexibility, and long-term durability. To overcome these limitations, growing research interests have been evidenced in developing intermediate-temperature (IT) PEMFCs that are designed to operate in the temperature range from 120 to 400 °C [4,5,6,7,8]. Fuel cells operated in the IT range exhibit several technical advantages including enhanced electrochemical kinetics, high tolerance to contaminating CO, H₂S, and other fuel impurities, lowered electrocatalyst loading, simplified heat and water management, as well as the potential for favorable waste heat recovery and use of low-cost non-platinum-group (NPG) catalysts and liquid fuels (e.g., ethanol) [9,10,11,12,13,14]. Yet, to develop and deploy high-performance PEMFCs to be operated at elevated temperatures, a number of technical barriers still need to be resolved in order to achieve high-performance IT proton-exchange membranes (PEMs) that carry sufficient mechanical and thermochemical durability, high electrochemical stability and proton conductivity (>0.02 S/cm), and low gas crossover [15,16,17,18,19,20,21,22].

As the core component of IT-PEMFCs, PEM directly influences the proton conductivity and the overall electrochemical performance of the resulting membrane electrode assemblies (MEAs) of the PEMFCs. To date, polybenzimidazole (PBI) as one of the common high-temperature (HT) polymers has been extensively investigated for developing acid-doped IT PEMs for IT-PEMFCs due to its exceptional thermal stability (with T_g of 425–436 °C), high mechanical strength, and ability to retain the high proton conductivity at elevated temperatures [10,11,12,13]. PBI, doped with phosphoric acid (H₃PO₄-PA), has demonstrated reasonably high proton conductivity at temperatures of above 150 °C. Nevertheless, the proton-conducting performance of PA-doped PBI membranes is often hindered due to unavoidable water loss at elevated temperatures, leading to decreasing proton conductivity and degrading fuel cell efficiency [11,13,23]. To date, several technical approaches have been devoted to synthesizing novel polymers, e.g., poly(pyridobisimidazole) (PPI), which exhibited an excellent acid uptake of over 550%, resulting in exceptional proton conductivity of 0.23 S/cm at 180 °C [24]. PEMs made of these HT polymers offer advantages in terms of mechanical flexibility, proton conductivity, and HT structural integrity and thermochemical stability. However, their scalability and stability in long-term PEMFC operation, especially in the presence of fuel impurities and extended cycling, are still an outstanding issue under intensive investigation [10,11,12,13,14,15,16]. A brief comparison of work temperatures, electrolytes, technical features, and outstanding issues among LT, IT, and HT fuel cells is summarized in

Table 1. Comparison of LT, IT, and HT fuel cells [3,5,6,7,10,11,12,13,14,15,16,17,18,19,20,21,22].

	LT PEMFCs	IT Fuel Cells	HT Fuel Cells
Work Temperature	50–100 °C	150–700 °C (e.g., IT-SOFCs, PAFCs †)	600–1000 °C (e.g., SOFCs, MCFCs ‡)
Electrolytes	● Polymer electrolyte membranes (e.g., Nafion®). ● Aqueous alkaline solutions (e.g., KOH), etc.	● H3PO4 for PAFCs. ● Metal phosphates, e.g., CsH2PO4, Zr(HPO4)2.nH2O, metal pyrophosphates (CeP2O7, Ti P2O7, Sn2O7); CsHSO4, etc. ● Heteropolyacids, e.g., H3PW12O40, H4SiW12O40, H4GeW12O40, H6P2W18O62, H5BW12O40, H6CoW12O40, H3PMo12O40, etc. ● Ceramic materials, e.g., doped ceria (CeO2) or stabilized zirconia (ZrO2), etc.	● Ceramic materials, e.g., yttria (Y2O3)-stabilized zirconia (YSZ). ● Molten carbonate salts, e.g., Li2CO3/K2CO3 mixture.
Advantages	● Fast start-up time and rapid dynamic response, suitable for mobile and portable applications, e.g., vehicles, personal electronics, etc. ● High power density, compact size. ● Zero-emissions.	● Higher tolerance to fuel impurities, e.g., CO, H2S, etc., compared to LT PEMFCs. ● High energy efficiency via recovering waste heat for combined heat and power (CHP) systems. ● High energy conversion rate due to faster reaction kinetics than LT PEMFCs. ● Simpler water and thermal management than LT PEMFCs. ● Lower degradation rates compared to SOFCs. ● Use of low-loading Pt or non-Pt-group catalysts. ● Suitable for hydrogen and direct liquid fuel cells, e.g., methanol, ethanol, etc.	● Fuel flexibility, suitable to use a wide variety of fuels, e.g., natural gas, biogas, liquid fuels, etc., through internal reforming. ● Noble metal catalysts are not required, leading to lower material costs. ● Not vulnerable to CO poisoning. ● High electrical efficiency up to 60–65%. ● High-quality waste heat for CHP.
Disadvantages	● Complex and costly water/thermal management. ● High costs due to use of expensive noble metal catalysts, e.g., Pt, and high-purity H2. ● High sensitivity to CO poisoning.	● Slower start-up than LT PEMFCs. ● Materials degradation at higher temperatures, existing a trade-off between efficiency and component degradation.	● Very slow start-up. ● Request expensive and heat-resistant materials. ● Significant challenges with thermal expansion and system sealing.
Challenges	● High cost of catalysts. ● Long-term durability. ● Building a costly hydrogen infrastructure of production, delivery, and storage.	● Developing new IT cathode materials with faster kinetics. ● Degradation mitigation. ● Improvement of long-term stability.	● Material degradation at HT. ● Thermal stress management. ● Complex system design.

Note: PAFCs ^†: Phosphoric acid fuel cells; MCFCs ^‡: Molten carbonate fuel cells.

[3,5,6,7,10,11,12,13,14,15,16,17,18,19,20,21,22], which shows the unique technical features and promising future of IT-PEMFCs for low-cost, high-efficiency and durable electrochemical energy conversion.

On the other hand, inorganic solid-state proton conductors have attracted particular attention in recent years due to their potential applications in electrochemical energy conversion and storage devices, including fuel cells and electrolyzers operated at elevated temperatures [7,9,12,13,14,15,16,17,18,19,20,21,22]. Among broad materials under exploration, cerium-based compounds demonstrate their unique electrochemical properties beneficial to enhance the proton conductivity of the resulting PEMs. Cerium (Ce), a rare-earth metal, is the most abundant element in the lanthanide series and carries fascinating redox behavior due to its unique atom electronic structure with four outer valence electrons and similar 4f, 5d, and 6s electron energy levels, which plays a crucial role in facilitating proton and oxide ion transport [4,7,9,15,16,17,18,19,25,26,27,28,29,30,31]. For instance, by means of so-gel method, Bhanu et al. [30] synthesized Nd³⁺ and Dy³⁺ co-doped cerium oxide (CeO₂) systems of Ce_0.8Nd_yO_{2-(x−y/2+y)} (x = 0.2; y = 0.04, 0.08, 0.1) as IT solid-state electrolytes for IT solid-oxide fuels cells (SOFCs) with the superior oxide ion conductivity of 2.2 × 10⁻² S/cm and activation energy of 0.83 eV at 600 °C. The performance is comparable to those of SOFCs based on doped metal transition perovskites and working at lowered temperatures (e.g., 700 °C) [32,33,34,35,36,37,38].

CeO₂ is often coined as ceria and has been widely studied for its oxygen storage capacity and ability to undergo reversible transformations between Ce⁴⁺ and Ce³⁺ oxidation states. This redox flexibility not only contributes to its versatile catalytic properties but also influences the proton-conducting mechanisms [39,40,41,42,43]. Integration of protons into the CeO₂ structure, particularly under hydrated or acidic conditions, enables it to act as a proton-conducting medium [15,16,17,18,27]. Yet, inorganic proton conductors are typically brittle and fragile, and it is necessary to composite their finely ground particles with HT polymers (e.g., PBI, polytetrafluoroethylene—PTFE, etc.) as polymeric binders to form polymer–inorganic composite (PIC) membranes to improve the membrane processibility, structural integrity, and mechanical flexibility and strength [15,17,18,44]. Fabrication of the PIC membranes utilizes the advantages of both inorganic solid-state proton conductors (high proton conductivity) and HT polymers (high mechanical performance). In addition, more recent investigations have been concentrated on design and fabrication of various composite membranes made of PBI composited with a variety of inorganic materials, e.g., metal phosphates and pyrophosphates, silica, and cerium-based compounds (CeO₂, Ce(SO₄)₂, Ce₂(SO₄)₃, Ce(HPO₄)₂, CePO₄, CeP₂O₇, etc.) [15,16,17,18,27,45,46]. These PIC membranes have demonstrated improved structural integrity, electrochemical stability, proton conductivity, and better chemical tolerance to fuel impurities (e.g., CO, SO₂, CH₄, etc.) [47,48].

Furthermore, recent studies also indicated that PIC proton-conducting membranes, e.g., those made of inorganic solid-state proton conductors and PBI, demonstrated noticeably improved membrane proton conductivity, structural durability, and PEMFC performance. For instance, PIC membranes incorporating silica or phosphate nanoparticles (NPs) have exhibited enhanced proton conductivity, increased mechanical strength, and better resistance to fuel impurities [25,27,48]. In addition, PIC membranes have also demonstrated the potential for improved long-term performance durability. However, challenges still remain in optimization of the composite formulation to balance the proton conductivity, durability, and manufacturing costs. Other inorganic proton conductors, e.g., doped CeO₂ and pyrophosphates, have also been under exploration for fabrication of IT-PIC-PEMs for use in PEMFCs. For instance, PBI membrane incorporated with Sn_0.95Al_0.05P₂O₇ (SAPO) fine particles was able to achieve the proton conductivity of 0.1 S/cm at 240 °C, with a peak power density of 840 mW/cm² in fuel cell tests [49]. While these PIC membranes have exhibited high cell performance and good stability in the IT range, outstanding technical challenges remain in enhancing their mechanical properties and reducing the manufacturing costs [15,27]. On the other hand, acid-doped PBI membranes demonstrated high proton conductivity and overall electrochemical performance. Yet, the level of acid doping in PBI is typically limited by various material and process factors, e.g., the acid retainability of PBI under fuel cell operating conditions, particularly at elevated temperatures. Acid loss due to evaporation or leaching with time can result in decreasing proton conductivity, thereby compromising the long-term membrane stability and the fuel cell performance [4,11,13,15,50]. Additionally, high acid content that is required to achieve the optimal proton conductivity can accelerate the mechanical degradation of the PEMs and therefore shorten their lifespans [14,24].

Moreover, the chemical compatibility of acid-doped PBI with other fuel cell components, e.g., the anode and cathode electrodes, is still an outstanding technical issue. The acidic environment can accelerate the degradation of the catalyst layers and other cell components and further negatively influences the overall durability and performance of the fuel cells [51,52]. Beyond these technical challenges, the cost and complexity of the acid-doping process itself also pose technical barriers to large-scale production in industry. Synthesis of high-quality highly acid-doped PBI membranes requires expensive chemicals and precise control of the reaction conditions, which makes the process both costly and time-consuming.

The present experimental study was the extension of our recent research of developing and characterizing acid-free (e.g., phosphoric acid) IT-PIC-PEMs for use in PEMFCs [45]. The present PIC membranes were made of CUP particles as the solid-state proton conductor and undoped PBI as the HT polymeric binder. Specifically, the experimental study was conducted to investigate the effects of membrane thickness and processing conditions as well as the fuel cell operating conditions on the proton conductivity and electrochemical performance of the PIC membranes at 200 °C. The research was aimed at providing valuable insights into the potential of IT-PIC-PEMs for use in next-generation IT, high-efficiency, durable electrochemical energy conversion devices.

2. Experimental

2.1. Materials

PBI S26 solution (with 26.2 wt.% PBI of Mw= ~35,000, dissolved in 72.3 wt.% dimethylacetamide (DMAc) in the presence of 1.5 wt.% lithium chloride (LiCl) stabilizer) was purchased from PBI Performance Products Inc. (Charlotte, NC, USA). CeO₂ nanoparticle (NP) powder, H₃PO₄ (86 wt.%), and DMAc anhydrous (99.8 wt.%) were purchased from MilliporeSigma (St. Louis, MO, USA). Platinum (Pt) catalyst NPs (60 wt.% Pt on Vulcan XC 72) and gas diffusion layers (GDLs) (CT carbon cloths with microporous layers) were purchased from the Fuel Cell Store (College Station, TX, USA).

2.2. Synthesis of CUP

Cerium ultraphosphate (CUP-CeP₅O₁₄) as the solid-state proton conductor was synthesized according to our recent study [45]. In brief, CeO₂ and H₃PO₄ (with the molar ratio of Ce/P = 1:12) were mixed and heated at 100 °C for 4 h in a tubular furnace under an argon atmosphere. Then, the mixture was heated to 800 °C in a Pt crucible and was further maintained at this temperature for 24 h in the argon atmosphere. Finally, the product was cooled down to 500 °C at a ramping temperature rate of 2 °C/h and air-quenched to ambient temperature. The obtained crystalline CUP powder had been further purified by rinsing in boiling deionized (DI) water three times, followed by drying in an air-circulated oven at 130 °C. Finally, the raw CUP powder was stored in a sealed glass vial before use.

2.3. Solution-Casting of PIC Membranes

The as-synthesized raw CUP powder was made up of irregular CUP particles within a wide range of particle sizes from a few dozen micrometers down to tens of nanometers. In order to achieve finer and more uniform CUP particles for preparing quality CUP-PBI PIC membranes in this study, the raw CUP powder was finely ground using a mortar and pestle for two hours. After grinding, a field-emission scanning electron microscope (FE-SEM) (JEOL JSM-7600F, JEOL Ltd., Tokyo, Japan) was employed to measure the particle size of the sampled CUP powder. For the convenience and accuracy of the particle size measurement by SEM, the sampled CUP powder was separated into three groups according to their relative CUP particle sizes, designated as “Large,” “Medium,” and “Small,” respectively, which was realized by using a lab-made multi-grade sedimentation setup, as illustrated in Figure 1. Therefore, for each designated group of the CUP particles, its mass was measured by a high-resolution digital balance, and its mean particle size and size deviation were determined via statistical analysis of the particle size data obtained from computerized image analysis of the CUP particle SEM images during SEM characterization of an ultrathin layer of CUP powder with the particle size of either “Large,” “Medium,” or “Small.”

The targeted CUP-PBI PIC membranes with the CUP/PBI mass ratio of 3:1 (75 wt.%) were fabricated using a solution-casting method, as illustrated in Figure 2A [45]. During the process, the finely ground CUP powder was dispersed in a DMAc solvent in a glass vial to form a CUP-DMAc colloid after sonication for 15 min, followed by magnetic stirring for 2 h. Then, a proper amount of PBI-DMAc solution was added to the CUP-DMAc colloid under vigorous stirring, followed by magnetic stirring for 24 h at room temperature. The formed slurry of the trinary CUP-PBI-DMAc mixture was cast onto a flat steel panel (Q-panel, the Q-Lab Co., Ltd., Westlak, OH, USA) with a controlled wet layer thickness using a drawdown bar with specific specifications (the Paul N. Gardner Co., Ltd., Pompano Beach, FL, USA) [53]. The as-cast PIC membranes attached to the steel panels had been placed in an air-circulating oven at 125 °C for a few minutes to evaporate the DMAc solvent and form a dry PIC membrane. Then, the steel panel covered with the dried PIC membrane was submerged in DI water and the PIC membrane was peeled off easily from the steel panel. Consequently, the newborn free-standing PIC membranes were further annealed at 125 °C in the same air-circulating oven for a few minutes to yield the undoped CUP-PBI PIC membranes ready for tests and characterization. Figure 2B–D shows the typical PIC membrane, gas diffusion electrode (GDE), and IT-MEA prepared in this study. For the purpose of comparison, undoped pure PBI membranes were also prepared using the same aforementioned procedure, which were used as the control membrane samples.

2.4. Water Uptake Testing

Samples of the PIC membrane (designated as CUP_75%) and pure PBI membrane prepared above were soaked in a DI water bath for specified time intervals from 2 to 24 h. After soaking, the membrane samples were taken out of the water bath, dried using lab tissue paper, and weighed to record their weights (m_wet). Weights of the dehydrated membrane samples (m_dry) were measured after desiccating the membrane samples in an air-circulating oven at 110 °C until no change in the weights was detected. The water uptake at 0 h represents the water content in the samples at the ambient temperature and humidity condition. The water uptake percentage of the membrane sample is calculated as

$$\mathrm{W}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r} \mathrm{u}\mathrm{p}\mathrm{t}\mathrm{a}\mathrm{k}\mathrm{e}\left(\mathrm{\%}\right)={\displaystyle \frac{{\mathrm{m}}_{\mathrm{w}\mathrm{e}\mathrm{t}}-{\mathrm{m}}_{\mathrm{d}\mathrm{r}\mathrm{y}}}{{\mathrm{m}}_{\mathrm{d}\mathrm{r}\mathrm{y}}}}\times 100\mathrm{\%},$$

(1)

which is used to indicate the water uptake property of the PIC membranes.

2.5. Preparation of GDEs and IT-MEAs

The GDEs used in the present fuel cell tests were prepared via air-brushing the Pt/C catalyst ink onto the procured gas diffusion layers (GDLs). In a typical process of MEA fabrication, the catalyst ink was lab-prepared by mixing 30 mg Pt/C catalyst (the Pt/C mass ratio = 60:100), 2.0 mg of the CUP-PBI suspension (with the CUP/PBI mass ratio = 98:2), and 0.8 g of DMAc as the carrier. The formed ink in a glass vial had been sonicated for 30 min for uniform mixing and then air-sprayed onto the surface of a square-shaped GDL with the areal area of 25 cm² that was placed onto a heated vacuum table at 120 °C. As a result, the as-fabricated GDEs carried the Pt-catalyst load of 1.0 mg/cm².

The MEA used in this study was assembled following a common hot-pressing procedure. During a typical process, an MEA laminate of five layers was assembled using a hand lay-up. The resulting MEA laminate with the areal dimensions of about 4.0 cm × 4.0 cm was placed in between the two GDEs prepared above (i.e., the anode and cathode, respectively), each of which carried an active area of 5 cm², and then hot-pressed at 150 °C with a pressure of 2.5 MPa for 10 min. After hot-pressing, the thickness of each GDL was reduced to ~83% of its original thickness. Consequently, the resulting MEA was installed into a single-stack fuel cell fixture (SAI Cell-525, the Scribner Associates Inc., Southern Pines, NC, USA) for the present hydrogen fuel cell testing at 200 °C. Figure 2B–D show the optical images of a typical CUP-PBI PIC membrane sample, a GDE covered with Pt/C catalyst, and an as-prepared IT-MEA in this study.

2.6. Structural Characterization of GDEs

Surface and cross-sectional morphologies of the GDEs were characterized by using a field-emission scanning electron microscope (FE-SEM) (JEOL JSM-7600F, JEOL Ltd., Tokyo, Japan). Prior to SEM characterization, small patches of GDE samples were scissored from the prepared GDE and sputter-coated with an ultrathin layer of carbon to avoid possible charge accumulation onto the samples during the SEM characterization. The elemental composition of the GDE samples was analyzed by using energy-dispersive X-ray spectroscopy (EDS) after the SEM micrographs were taken.

2.7. Single-Stack Hydrogen Fuel Cell Testing

A Scribner^® 850e fuel cell test station (the Scribner Associates Inc., Southern Pines, NC, USA) was utilized for characterization of the electrochemical performance of the IT-MEA and the proton conductivity of the IT-PIC-PEM prepared in this study. Figure 3 shows the fuel cell testing setup, in which a lab-made thermal chamber installed with a proportional–integral–derivative (PID) temperature control unit was used to maintain the fuel cell temperature at 200 °C. A single-stack hydrogen fuel cell (the Scribner^® SAI Cell-525) installed with the IT-MEA was operated under the testing condition: Inlet reactant gas temperatures of T_H2 = T_Air = 80 °C (humidified with relative humidity (RH) of ~47%), fuel cell temperature of T_cell = 200 °C, backpressure = 1.0 bar (for either H₂ or air), a minimum mass flow rate of 0.08 L/min and the stoichiometry of 1.2 applied for H₂ as well as a minimum mass flow rate of 0.15 L/min and the stoichiometry of 2.5 applied for air. Prior to data acquisition of the single-stack fuel cell performance, the MEA was first activated via a vendor-recommended break-in procedure at the cell temperature of 200 °C, such that the cell had been held via applying a pseudo-cyclic voltage with the alternative amplitudes of 0.6 V, 0.3 V, and then an open circuit condition, respectively, for a constant duration of 60 s for 20 cycles. Application of the break-in procedure in a fuel cell test was to eliminate the possible tiny amount of DMAc residues (with the boiling point of 165 °C) that might remain in the PIC membranes after fabrication and also to stabilize the MEA before a fuel cell test. The proton conductivity of the PIC membrane was extracted from the electrochemical impedance spectroscopy (EIS) measurement of the single-stack hydrogen fuel cell, which was operated at the constant direct current (DC) density of 0.1 A/cm², with 10 steps/decade in the frequency range from 0.1 to 10 kHz, and an alternative current (AC) with the amplitude of 5% that of the DC superimposed onto the DC. The Z-plot of the EIS measurement was recorded for spectra analysis, and the value of proton conductivity of the PIC membrane is calculated as

$$\sigma (\mathrm{S}/\mathrm{c}\mathrm{m})={\displaystyle \frac{L\left(\mathrm{c}\mathrm{m}\right)}{R\left(\mathrm{O}\mathrm{h}\mathrm{m}\right)\times A\left({\mathrm{c}\mathrm{m}}^2\right)}},$$

(2)

where σ (S/cm) stands for the proton conductivity of the PIC membrane, L (cm) is the membrane thickness, R (ohm) is the membrane ionic resistance that was extracted from the Nyquist plot, and A (cm²) is the active areal area of the membrane (5 cm²). During an EIS measurement, the single-stack fuel cell was maintained at an open circuit condition between two consecutive measurements.

3. Results and Discussion

The present IT-PIC-PEMs were fabricated by means of a low-cost solution-casting technique via casting the slurry of the CUP particles uniformly dispersed in a PBI/DMAc solution as aforementioned. To reduce the size of the particles, the CUP powder was carefully ground using mortar and pestle before preparing the slurry for casting. The present PEM fabrication technique can be conveniently employed for processing thin CUP-PBI PIC membranes with the CUP mass concentration as high as 85 wt.% [45].

3.1. CUP Particle Size Characterization

Figure 4 shows the typical SEM micrographs of as-synthesized CUP powder that was made up of irregular CUP particles with the particle sizes in a wide range. It can be observed that some CUP particles even carry the particle length of over 100 μm. Figure 5 shows the typical SEM images of three groups of a batch of finely ground CUP powder after being ground for 2 h, which are designated as “Large,” “Medium,” and “Small,” respectively. The statistical results of the CUP particle size of the three CUP particle groups are shown in

Table 2. The statistical results of the CUP particle size after grinding for hours.

CUP Powder Samples	Fractions	Average Size (µm)
As-synthesized raw CUP powder	Initial	16.18 ± 5.54 2.4 ± 0.9 † 0.3 ± 0.1 ‡
After 2 h grinding using mortar and pestle	Large	19.3 ± 5.7 2.1 ± 0.9
	Medium	7.5 ± 2.6 0.3 ± 0.1
	Small	2.0 ± 0.6 0.3 ± 0.1

Note: ^†: Relatively smaller-sized CUP particles picked by mouse during SEM imaging; ^‡: Much smaller-sized CUP particles picked by mouse during SEM imaging.

, and the mass distribution of the three CUP particle groups is shown in

Table 3. Mass distribution of three CUP particle groups (Large-, Medium-, and Small-sized) of finely ground CUP powder sample after sedimentation-based separation.

Particle Fractions	Large	Medium	Small	Total
Mass (g)	3.1379	0.7609	0.7996	4.6984
Mass Percent (%)	66.8	16.2	17.0	100

. It can be found that after having ground the raw CUP particles using mortar and pestle for two hours, the average particle size was obviously reduced. In addition, the lab-made three-grade sedimentation setup can effectively separate the CUP particles into three groups of CUP particles according to their particle sizes, which provides a simple, effective way to further grind the CUP particles in the larger particle groups into smaller ones.

3.2. GDE Characterization

To characterize the electrochemical performance of the IT-PIC-PEMs, each PIC membrane sample was assembled with two as-fabricated Pt-catalyst-covered GDEs to form an IT-MEA, which was further assembled into a single-stack hydrogen fuel cell that was connected to a fuel cell test station (Figure 3). Prior to assembling the IT-MEA, pieces of the typical as-prepared GDEs were scissored and sampled for characterization of the GDEs. SEM was employed to characterize the cross-sectional microstructures and surface morphology of the Pt-catalyst layer of the GDEs. EDX analysis was conducted to estimate the element distribution on the GDE surface. The typical SEM micrographs and EDX color mapping of the GDEs are shown in Figure 6. Surface morphology of the GDE indicates the porous nanostructure of the surface layer with visible Pt NPs. It is evidenced that the Pt-catalyst NPs were evenly distributed at the GDE surface with no visible agglomeration. The backscattered SEM micrograph of the cross-sectional GDE layer shows a visible boundary between the Pt-catalyst layer and the carbon substrate. The thickness of the Pt-catalyst layer (with the load of 1.0 mg/cm²) is estimated ~12–14 µm.

EDX color mapping of the element distribution on a sampled catalyst layer was made in terms of general elements and specific elements of Pt, N, O, and Ce on the surface, respectively, which shows the existence of the representing elements Pt, Ce, and N and their even distribution. In addition, elemental composition of the Pt-NP-covered GDE sample was further analyzed by means of energy-dispersive X-ray spectroscopy (EDS) after taking the SEM micrographs (Figure 7 and

Table 4. Elemental composition at the GDE surface.

Element	C	N	O	P	Ce	Pt
Wt. %	11.18 ± 0.82	0.62 ± 0.14	3.75 ± 0.77	1.23 ± 1.20	0.79 ± 0.89	81.40 ± 1.99

). The gained EDS results confirm the even distribution of the Pt-catalyst NPs as well as other elements at the surface of the catalyst layer, corresponding to the theoretical predictions.

3.3. Proton Conductivity Measurements of IT-MEAs Made of CUP-PBI PIC PEMs

The values of proton conductivity of the lab-made IT-PIC-PEMs were extracted from the EIS measurements. During a single-stack fuel cell test, the measured value of the proton conductivity of an IT PIC membrane typically increased with time until reaching a stable value at a certain point. Figure 7B shows the variation in the proton conductivity with increasing time of a typical IT-PIC-PEM, which was measured in a hydrogen fuel cell test at 200 °C using a Scribner 850e fuel cell test station. After having been tested for ~20 h, the measured proton conductivity reached 0.105 S/cm, which is at the higher side of the range of proton conductivity of Nafion^® membranes used in low-temperature PEMFCs [29].

Table 5. Proton conductivities of the IT CUP-PBI PIC PEMs at 200 °C.

Sample Specification	Membrane Thickness (μm)	Proton Conductivity (S/cm)
CUP_75%_82 µm †	82	0.048
CUP_75%_134 µm	134	0.063
CUP_75%_sprayed_166 µm ‡	166	0.046
CUP_75%_sprayed_105 µm	105	0.105
CUP_75%_spr_2 h_94 µm *	94	0.051
CUP_75%_spr_2 h_95 µm (0.5 mg/cm2 Pt/C)	95	0.066

Note: ^†: 75 wt.% CUP, thickness: 82 μm, time for grading CUP prior to casting: 15 min. ^‡: 75 wt.% CUP, thickness: 166 μm, time for grading CUP prior to casting: 15 min, two sides sprayed with CUP particles. *: 75 wt.% CUP, thickness: 94 μm, time for grading CUP prior to casting: 2 h, two sides sprayed with CUP particles.

summarizes the values of proton conductivity of several IT-PIC-PEM samples prepared in this study with different membrane thicknesses and processing parameters. These PIC membranes exhibited consistently high proton conductivity at 200 °C and 47% RH, which endorses the reliable processing technique of the IT-PIC-PEMs in this study.

Table 6. Comparison of proton conductivity of various PEMs reported in the literature and this work.

Proton Conductors	Proton Conductivity (S/cm)	Measurement Temperature (°C)	Refs.
Nafion® 112	0.092 0.06 ~0.15–0.16 0.0175	20 (in liquid water) 30 (100% RH) 80–100 (90–100% RH) 120 (50% RH)	[16] [16] [54] [54]
H3PO4-doped PBI	10−9–10−5 0.04–0.07	160 (1.9 H3PO4 per unit) 200 (4–6 H3PO4 per unit)	[55] [56]
CsH2PO4	10−7 10−2	Room temperature 232	[57,58]
CaHPO4.2H2O	10−3	327	[59]
In0.1Sn0.9P2O7 (pellet)	0.01–0.195	750–300	[60,61]
CeP2O7 (pellet)	0.001–0.018 0.002–0.035	50–200 (dry) 60–200 (PH2O = 0.15 atm)	[62] [27]
CeP5O14 (pellet)	0.03	250 (PH2O = 0.15 atm)	[26]
CUP-PBI membrane (CeP5O14, 75 wt.%)	0.046–0.105	200	this study

compares the values of proton conductivity of the present IT-PIC-PEMs at 200 °C with several common LT and IT PEMs reported in the literature. Compared to other LT and IT PEMs in Table 6, the present CUP-PBI PIC membranes exhibited excellent proton conductivity at 200 °C. Moreover, the present measurement of proton conductivity of the PIC membranes was based on in situ single-stack hydrogen PEMFC testing, and the PIC membranes were prepared using the low-cost solution-casting technique, which can be conveniently scaled up for low-cost, massive production of large-sized IT PEMs for industry.

Figure 8 shows the orthorhombic crystal structure of CUP (CeP₅O₁₄) that was determined and analyzed by X-ray diffraction (XRD) and related software package in our recent study [45]. Due to its condensation state of the phosphate state, the ultraphosphate crystal structure is more complex than common orthophosphates that are made up of isolated PO₄ tetrahedra sharing corners with metal oxide polyhedra. The overall structure of the (P₅O₁₄)³⁻ anion contains eight-membered conjugate rings, while the coordination state of all rare-earth cations in ultraphosphates is usually eight [26,63], in which five tetrahedral PO₄ units are linked by sharing oxygen atoms and the pendant oxygen anion in each PO₄ unit can easily form the hydrogen bond with one proton. Thus, protons bonded with the pendant oxygen anions can easily transport through the ultraphosphate crystal via Grotthuss mechanism, i.e., proton hopping, due to the short distance between two neighboring pendant oxygen anions in metal ultraphosphates, e.g., CUP. In addition, the main advantage of CUP is its ability to conduct protons at IT range of 100–400 °C with relatively low humidity, superior to the common LT PEMs like Nafion^®, which require a high level of hydration and are operated effectively only below 100 °C. Unlike many other proton conductors where water molecules act as the primary proton carriers, i.e., the vehicle mechanism, CUP’s proton conductivity is not solely dependent on a high concentration of mobile water molecules. Instead, the proton transport is an intrinsic property of the solid-state material itself, involving the movement of protons within the phosphate framework [26]. This allows the material to be operated at temperatures where liquid water would evaporate, thus avoiding those water management problems commonly confronted in traditional low-temperature fuel cells. In the present PIC membrane, the CUP particles provide the primary proton conduction pathway, while the polymer provides mechanical stability and flexibility. In this context, even in a slightly humidified environment, the proton transport is still dominated by the solid-state Grotthuss mechanism in the CUP-PBI membrane, since virtually PBI is not proton conductive. Nevertheless, additional efforts are still desired to explore more in-depth understanding of the proton-transporting mechanisms in the present IT-PIC-PEMs.

Furthermore, in the present fuel cell test, the measured proton conductivity of the IT-PIC-PEMs grew with increasing time until it reached the plateau with the value as high as >0.105 S/cm. This observation shows that the fuel cell needed an interval of time to hydrate the PIC membrane by moisture in the system. As reported in the literature [53], the proton conductivity of PIC membranes is closely correlated to the water content in the membranes. To fully understand the fuel cell behavior, the impedance response was measured at different current densities and RH levels (controlled by backpressure), as shown in Figure 7C. The impedance behavior of the CUP_75% membrane samples varies with the current density. As expected, the change in the high-frequency real-axis intercepts corresponding to the membrane ohmic resistance is negligible. However, the diameters of the loops vary depending on the testing conditions. The activation loop decreases with increasing current density. At lower current densities, the activation losses are dominant compared to the ohmic and mass transport losses. At higher current densities, the activation kinetics improve, resulting in a decrease in the loop diameter. The loop diameter also changes with the change in the backpressure, which corresponds to the RH level of the system. Increasing the RH level results in the improved activation kinetics of the system, as evidenced by the decrease in the loop diameter. However, no significant changes were observed in the high-frequency real-axis intercepts.

3.4. Water Absorption of the CUP-PBI PIC PEMs

The water uptake levels of a typical PIC membrane (with CUP 75 wt.%) and pure PBI membrane were measured, as shown in Figure 9. As expected, the water uptake of the pure PBI membrane is higher than that of the PIC membrane. After 12 h of soaking the membrane samples in water, the water uptake of the pure PBI membrane was about 14 wt.%, while it was 4.4 wt.% for the PIC membrane. Such low water uptake of the PIC membrane can be responsible for the negligible change in the proton conductivity with varying RH level. However, increasing the water content in the system can improve the activation kinetics and the overall performance of the fuel cell.

3.5. Fuel Cell Performance Characterization

To evaluate the electrochemical performance of the IT-MEA developed in this study, the polarization curves were measured at varying membrane parameters and fuel cell operating conditions.

Table 7. The fuel cell performance data of typical lab-made IT-PIC-PEMs at 200 °C.

PIC Membranes	Flow Rates (H2 and Air) (L/min)	OCV (V)	Max. Current Density (A/cm2)	Max. Power Density (mW/cm2)
CUP_75%_82 µm (1.0 mg/cm2 Pt/C)	0.08, 0.2	0.31	439.6	46.6
	0.25, 0.5	0.33	543.6	58.3
CUP_75%_134 µm (1.0 mg/cm2 Pt/C)	0.08, 0.2	0.58	572.9	114.3
CUP_75%_sprayed_166 µm (1.0 mg/cm2 Pt/C)	0.08, 0.2	0.95	439.3	104.2
	0.15, 1.0	0.96	410.5	87.1
CUP_75%_sprayed_105 µm (1.0 mg/cm2 Pt/C)	0.08, 0.2	0.31	693.8	70.5
	0.15, 0.3	0.32	763.3	87.1
	0.15, 1.0	0.35	1010.1	129.1
CUP_75%_spr_2 h_94 µm (1.0 mg/cm2 Pt/C)	0.08, 0.2	0.58	877.9	186.4
	0.15, 0.3	0.59	902.1	176.3
	0.15, 1.0	0.59	915.9	171.1
CUP_75%_spr_2 h_95 µm (0.5 mg/cm2 Pt/C)	0.08, 0.2	0.51	640.4	155.3
	0.15, 0.3	0.52	759.0	164.5
	0.15, 1.0	0.51	981.3	191.7

summarizes the typical polarization V-I curves of several IT-MEAs that were tested on the single-stack hydrogen fuel cell at 200 °C with humidified H₂ and air. Below we examine the effects of the PIC membrane parameters and operating conditions on the fuel cell performance.

3.5.1. Membrane Thickness Effect

Recent studies demonstrated that PEM thickness can noticeably impact the fuel cell performance [64,65]. The thickness can affect the membrane polymer hydration and structure, thermal and mechanical stability, gas impermeability (gas crossover), and ion transport, among others. A thick membrane exhibits higher ionic (proton) resistance and, therefore, a lower value of proton conductivity. However, decreasing membrane thickness may negatively increase the hydrogen crossover rate and thus reduces the cell performance, e.g., the open circuit voltage (OCV).

Figure 10A shows the fuel cell polarization (V-I) curve of a typical IT-MEA made of the PIC membrane with thickness of ~125 μm. Its maximum current density, peak power density, and OCV were measured as 566.8 mA/cm², 126.9 mW/cm², and 0.75 V, respectively. For the purpose of comparison, a thin PIC membrane with a thickness of 82 μm was also fabricated and assembled into an IT-MEA for electrochemical characterization on the single-stack fuel cell. The comparative results show that decreasing the membrane thickness down to 82 μm resulted in reduced ionic resistance. As observed in Table 2, the values of proton conductivity of the thinner PIC membranes are slightly lower than those of the thicker ones (with thickness of ~125–132 μm). However, such an effect is not significant. In addition, the electrochemical characterization results, as shown in Figure 10B, indicate a drastic decrease in OCV (0.33 V) and a decrease in power density (58.3 mW/cm²) due to the high hydrogen crossover rate and other possible side reactions in this thin PIC membrane (82 μm). In addition, reduced electrochemical performance and low OCV in the thin PIC membranes can also be induced by the large CUP particle size. Such observations were also confirmed by other experimental studies reported in the literature [66,67], where pure polymer membranes exhibited performance limitations due to thickness reduction. In addition, ultrathin membranes commonly display a severe performance degradation that can be overcome by properly increasing membrane thickness.

3.5.2. Effect of Spraying CUP Particles onto Membrane

To improve the hydrogen impermeability and surface proton conductivity of a PIC membrane, surface reinforcement was applied via spraying a thin layer of CUP-rich PBI/DMAc solution at its two sides. The thickness of the sprayed CUP-PBI surface layer after drying was ~20 µm. Figure 11A shows the V-I diagram of the IT-MEA installed with a CUP_75%_sprayed PIC membrane and Pt-NP-coated GDEs with the thickness of 166 µm. The maximum current density and peak power density at the gas flow rates of 0.08 L/min for H₂ and 0.2 L/min for air were 439.3 mA/cm² and 104.2 mW/cm², respectively. The corresponding overall OCV in this case was 0.96 V, close to the theoretical estimate of ~1.13 V, which demonstrates the low hydrogen crossover rate of the membrane with a large thickness. In addition, the IT-MEA performance abnormally decreased after increasing the reactant flow rates to 0.15 L/min for H₂ and 1.0 L/min for air.

Furthermore, in the case of the reactant flow rates of 0.15 L/min for H₂ and 1.0 L/min for air, an abnormal drop in the cell power density (P-I) diagram is observed at the current density between 300 and 400 mA/cm², which could be caused by the data acquisition system affected by electronic noise relevant fluctuations, which are also found in other diagrams. In this case, the system voltage efficiency at 200 °C is

$${\epsilon }_V={\displaystyle \frac{E_{cell}}{E_{theor}}}={\displaystyle \frac{0.96}{1.13}}=0.85.$$

(3)

The polarization diagrams also indicate that a thicker membrane is correlated to higher ohmic losses, which negatively influence the overall electrochemical performance of the fuel cell and are in agreement with the proton conductivity results as shown in Table 5.

Figure 11B shows the polarization (V-I) and power density (P-I) diagrams of an IT-MEA installed with a thin CUP_75%_sprayed_105 µm membrane that was sprayed with CUP-PBI particles at two sides and carried the membrane thickness of ~105 μm. The measurements show that this PIC membrane exhibited higher current densities. The maximum current density and peak power density at the reactant flow rates of 0.15 L/min for H₂ and 1.0 L/min for air were 1010.1 mA/cm² and 129.1 mW/cm², respectively. Yet, the overall OCV of this IT-MEA was measured as only 0.35 V due to the high hydrogen crossover in this thinner PIC membrane integrated with 75 wt.% CUP particles. The present experimental observations indicate that additional research is desired to enhance the gas impermeability of the PIC membranes at high CUP mass fraction (75 wt.%) and low membrane thickness (<50 μm or less), which may lead to a type of commercially viable IT-PIC-PEM with high proton conductivity and gas impermeability, sufficient mechanical strength and integrity, and excellent thermal and oxidative stability.

In the above figure, the fuel cell performance results show that high hydrogen crossover can be exerted due to the low membrane thickness and the large size of the CUP particles as the solid-state proton conductor. Thus, additional efforts can be made to further reduce the CUP particle size, which can improve the membrane microstructure, gas impermeability, and proton conductivity. One handy approach employed in this study was to finely grind the as-synthesized CUP powder for 2 h using mortar and pestle prior to the PIC membrane processing. This method can be further combined with the multi-grade sedimentation method to separate the ground CUP powder into large- and small-sized particle groups for the next step of finer grinding. In addition, the as-fabricated PIC membrane was sprayed at two sides with a thin layer of the finely ground CUP particles mixed with a small quantity of PBI/DMAc solution. The finely ground CUP particles were further utilized in GDE fabrication to form the effective three-phase catalyst layers. Figure 12A shows the polarization (V-I) and power density (P-I) diagrams of an IT-MEA installed with such a refined IT-PIC-PEM and tested at 200 °C.

Similarly, the polarization and power density diagrams of an IT-MEA installed with CUP_75%_spr_2h membrane and Pt-NP-loaded GDEs were obtained at three reactant flow rates: (a) 0.08 L/min for H₂ and 0.2 L/min for air, (b) 0.15 L/min for H₂ and 0.3 L/min for air, and (c) 0.15 L/min for H₂ and 1.0 L/min for air, respectively. The overall thickness of the PIC membrane after spraying with CUP-rich/PBI particles at two sides is ~ 94 μm. The maximum current density, peak power density, and overall OCV of the IT-MEA in case (a) were 877.9 mA/cm², 186.4 mW/cm², and 0.58 V, respectively. After increasing the reactant flow rates as shown in cases (b) and (c), the maximum current density and OCV of the IT-MEA were slightly improved, while the peak power density was slightly decreased. In case (b), these performance values were 902.1 mA/cm², 176.3 mW/cm², and 0.59 V, and in case (c) were 915.9 mA/cm², 171.1 mW/cm², and 0.59 V. The decrease in the peak power can be related to the larger gas crossover with increasing reactant flow rates.

In the above, the fuel cell performance results were obtained using the IT-MEA installed with the PIC membrane and GDEs that were both fabricated with the finely ground CUP particles (grinding for 2 h). It was shown that size reduction in the CUP particles did obviously improve the electrochemical performance of the IT-MEA, e.g., the current and power densities and OCV compared to those of the IT-MEA made of a thin PIC membrane [Figure 7B]. In addition, CUP particles with reduced size can even benefit the electrochemical performance of an IT-MEA installed with a thin PIC membrane (~ 94 μm), compared to the one installed with the PIC membrane of relatively large CUP particles, as shown in Figure 6B. In principle, small-sized CUP particles can enhance the specific CUP surface area (at a fixed CUP wt.%) and thus improve the interaction between the proton conductor and the catalyst layers in GDEs, which benefits the efficient proton transport across the GDEs and ensures better reaction kinetics at the catalyst sites [34]. In addition, small-sized CUP particles can lead to better integration with the catalyst layer, improving the interaction between the catalyst particles and the proton-conducting particles, and thus enhancing the catalyst utilization.

Moreover, reduction in the catalyst load in fuel cells can favorably reduce the unit costs per fuel cell stack. In this study, effect of the GDE catalyst load, i.e., 1.0 mg/cm² (See above) and 0.5 mg/cm², was examined on the electrochemical performance of the present IT-MEAs. Figure 12B shows the fuel cell performance of the IT-MEA installed with a CUP_75%_2h_spr PIC membrane (thickness: ~95 μm, catalyst load: 0.5 mg/cm² (Pt/C)). It is shown that the IT-MEA performance increased with increasing H₂ and air flow rates to 0.15 and 1.0 L/min, respectively. The maximum current density, peak power density, and OCV were measured as 981.3 mA/cm², 191.7 mW/cm², and 0.51 V, respectively. The fuel cell performance results show that halving the Pt-catalyst load in the IT-MEAs did not noticeably influence the cell performance, which offers the technical opportunity to explore the effect and related mechanisms of catalyst load reduction in the fuel cell performance, a promising strategy to reduce the costs of IT fuel cell stacks.

4. Conclusions

A comprehensive experimental study was conducted to characterize the proton conductivity and electrochemical performance of novel IT-MEAs installed with acid-free PIC proton-conducting membranes developed in this study. These PIC membranes were made of CUP particles as the solid-state proton conductor and undoped HT PBI as the polymeric binder, which fundamentally overcome the technical disadvantages of acid-doped PBI membranes. The test results indicate that the mechanically flexible CUP-PBI PIC membranes (loaded with CUP of 75 wt.%) carried a reduced water uptake of 4.4 wt.% compared to pristine PBI membranes (14.0 wt.%). The reduced water uptake can benefit less membrane swelling during the fuel cell cycling. In addition, the PIC membranes carried high proton conductivity of up to 0.105 S/cm, which was measured at RH 47% and 200 °C. In addition, the PIC membrane performance exhibited dependence upon the membrane thickness, process parameters, fuel cell operating conditions, etc.

The present experimental results also indicated that increasing reactant flow rates benefit the IT-MEA performance, while reducing membrane thickness drastically decreased the overall OCV of the IT-MEA due to the high hydrogen crossover and other side reactions. As a handy technical approach, spraying a thin layer of the mixture of the fine CUP particles and a small amount of PBI at both sides of the PIC membranes can noticeably enhance the overall cell OCV of up to 0.95V and resulted in the voltage efficiency of 0.85. Furthermore, reduction in the CUP particle size in the PIC membranes can improve the cell performance with the maximum power density of 191.7 mW/cm², which prospects a great opportunity to adopt nanostructured CUP particles to further enhance the IT-MEA performance while suppressing the hydrogen crossover rate of the PIC membranes. The present single-stack hydrogen fuel cell tests also indicated that reduction in the Pt-catalyst load from 1.0 to 0.5 mg/cm² did not noticeably influence the IT-MEA performance at 200 °C. Therefore, the present experimental study predicts the promising future of IT-PIC-PEMs for broad applications in IT electrochemical energy conversion and chemical production, e.g., IT hydrogen and direct liquid PEMFCs, steam electrolyzers, electrolytic CO₂ capture, low-pressure electrolytic ammonia production, etc.