Radiation-grafted anion-exchange membranes: the switch from low- to high-density polyethylene leads to remarkably enhanced fuel cell performance

Herein we detail the development of a new high-density polyethylene-(HDPE)-based radiation-grafted anion-exchange membrane (RG-AEM) that achieves a surprisingly high peak power density and a low in situ degradation rate (with configurations tailored to each). We also show that this new AEM can be successfully paired with an exemplar non-Pt-group cathode. A primary motivation for the development of anion-exchange membrane (AEM) fuel cells (AEMFCs) is the broader range of sustainable, non-precious-metal catalysts that are feasible; if costs are lowered enough, AEMFCs would be deployable in a range of stationary power sectors ( e.g. back-up and oﬀ-grid). However, as the performance of AEMFCs typically drop when Pt-based electrodes are replaced with non-Pt types, it is essential that the highest performing polyelectrolytes are developed, both membranes and ionomers (the latter incorporated to impart ionic conductivity in the electrodes). The findings with the high conductivity AEM reported herein will also be of interest to developers of AEMs for metal–air and redox-flow batteries, electrolysers (both H 2 O - H 2 and CO 2 - high-value chemicals and fuels), and salinity gradient power.

It is critical that anion exchange membranes (AEM), developed for use in anion exchange membrane fuel cells (AEMFCs), can support both high power outputs and in situ durability.The literature shows a dearth of options that have an acceptable combination of both, which is mandatory to push AEMFCs closer to widescale implementation and commercialisation.
In recent years, significant progress has been made in the development of AEMs generally, 1 and RG-AEMs specifically. 2G-AEMs produced via a high dose rate electron-beam modification of low density polyethylene (LDPE) films 2 have helped to advance the field, by both allowing for an enhanced understanding on how AEMs behave in AEMFCs, as well as their ability to support high H 2 fuel cell performances (41.0 W cm À2 at 460 1C with non-Pt cathodes). 3,4This performance was possible because such LDPE-based RG-AEMs have high conductivities, and kinetically fast water transport characteristics (that extends the region of mass transport power losses to higher current densities).However, a limiting aspect of thin LDPE-based RG-AEMs films are their modest mechanical properties under stress (Table 1), which needs to be improved for commercial applications.In this respect, the use of highdensity polyethylene (HDPE) was hypothesised to be a promising avenue of research, especially as prior reports suggest that HDPE can be used to make viable RG-AEMs 5 using low dose rate gamma ray methods (Sproll et al. 6a report low dose rates lead to a lower concentration of longer grafted chains, which leads to poorer performance characteristics).
We report for the first time a HDPE RG-AEM fabricated using a high dose rate electron-beaming method (amenable to large batch production), with comparable properties to a similarly fabricated LDPE RG-AEM, but with improved mechanical properties, fuel cell performance, and operational stability.
The RG-HDPE AEMs were fabricated and characterised using the methods detailed in our previous publications. 2,4,7n summary, HDPE films (10 mm thickness, ET321010, Goodfellow UK), and LDPE films (15 mm, ET311115, Goodfellow UK) were subjected to 100 kGy absorbed dose in air (peroxidation method) using a 4.5 MeV dynamic continuous electron-beam unit (STERIS Applied Sterilization Technologies, South Marston, UK, 10 kGy absorbed dose per rapid pass of the films under the e-beam).The e-beam irradiated films were then stored in a freezer at À40 1C until use.
Weighted e-beam irradiated films were grafted with vinylbenzyl chloride (VBC, mixture of 3-and 4-isomers, Sigma-Aldrich product 338729, no removal of any inhibitors) by immersion in a N 2 -purged aqueous solution of VBC (5 vol% with the further addition of 1 vol% octyl-2-pyrrolidone dispersant): 6 h grafting at 40 1C for LDPE and 4 h at 50 1C for HDPE.After thorough washing with toluene and drying at 16 h at room temperature, the grafted membranes were weighed before being submerged in aqueous trimethylamine (45 wt%) for 24 h.After subsequent multiple room temperature washings with ultrapure water (UPW, 18.2 MO cm), heating in UPW for 1 h at 60 1C, ionexchange by immersion in aqueous NaCl (1 mol dm À3 , immersion for 1 h in 3Â fresh solutions), and further washing in UPW (until no traces of free Cl À remained), the Cl À anion forms of the LDPE-and HDPE-based RG-AEMs were recovered (heron designated LDPE-AEM and HDPE-AEM, respectively).
For the grafting stage (pre-amination), we observed that the irradiated HDPE films could be grafted more repeatedly after storage at À40 1C for at least 6 months after beaming compared to the LDPE (Fig. 1): this is hypothesised to be due to differences in radical-based self-crosslinking rates during cold-storage.6b,c This more predictable ''shelf-life'' is clearly beneficial for repeated batch fabrication over 6 months (with predictable properties).
The key properties of the two RG-AEMs are summarised in Table 1.The HDPE RG-AEM exhibited very similar ex situ properties to the LDPE RG-AEM: ion-exchange capacity (IEC), hydrated thickness, water-uptake, and OH À conductivity.Both RG-AEMs (OH À form) also similarly retained their conductivities in a flowing humidified N 2 atmosphere at 80 1C (Fig. 2); the loss of conductivity was only (8 AE 1)% after 500 h for both.Comparing the Raman spectra before and after 500 h, Fig. 3 shows only minor RG-HDPE degradation, in good agreement with the reduction in conductivity during that time.
However, where the HDPE-AEM appears much more desirable compared to the LDPE-AEM was its tensile properties.The break stress of the HDPE-AEM was 52% greater than the LDPE AEM and it also showed an enhanced ability to stretch without failing.This was accomplished without having to increase the thickness of the final hydrated AEMs (achieved just by using a different substrate).The use of thin membranes is desirable as they enable fast water transport, meaning that a higher current  Fig. 3 The Raman spectra of the HDPE-AEM before and after the 500 h test presented in Fig. 2. A ThermoFisher DRX Raman microscope (532 nm laser) was used.Key diagnostic peaks 4 are labelled; * = HDPE-derived peaks.The spectra were normalised to the trimethylammonium peak at 753 cm À1 to aid visual comparison.a Ion-exchange capacity, IEC = mmol Cl À per g(dry RG-AEM, Cl À form).
f Gravimetric water uptake (Cl À form) at room temperature (= 100 Â (m hyd À m dehyd )/m dehyd , where m = mass/g).g The number of water molecules per Cl À anion in the fully hydrated AEM, calculated as: l H 2 O = WU/(100 Â 18.02 Â IEC).h The 4-probe (in-plane) OH À conductivity in a flowing 100% relative humidity N 2 atmosphere, 4 based on the method first reported by Ziv and Dekel. 8i Tensile properties of the Cl À form RG-AEMs (errors within 25%) in the ambient atmosphere.
][11] Fig. 4 shows the beginning-of-life AEMFC performances at 80 1C for cells operated with HDPE-and LDPE-AEMs using identical electrodes and benchmark fuel-cell-grade Pt-based catalysts (using the Surrey group's standard fuel cell testing protocols, which have been detailed extensively in recent publications 2,4 and allows us to maintain a local database of relative performances of different AEMs).Despite the LDPE-AEM and HDPE-AEM exhibiting similar ex situ thicknesses, water uptakes, and conductivities, the HDPE-AEM yielded a significantly higher H 2 /O 2 AEMFC performance: the HDPE-AEM exhibited a peak power density of 2.55 W cm À2 at 80 1C (cf.2.01 W cm À2 for the LDPE-AEM) with zero back-pressure gas feeds (with only ca.0.2 bar pressure drops).This is a dramatic demonstration of the findings by Sproll et al. related to the development of RG-proton-exchange membranes (RG-PEM), 12 who reported that conductivities, water uptakes, and resulting fuel cell (PEMFC) performances strongly depend on the micro-structure of the ETFE-base films used.The use of nominally identical base ETFE films (from two suppliers), differing only in microstructure, resulted in critical differences in the final membranes: larger crystalline sizes led to enhanced RG-PEM conductivities and durabilities of the resulting PEMFCs.The only significant variable in our comparison experiments (Fig. 4) was the starting base-material used.DSC experiments show that the supplied HDPE film had a crystallinity of 57% (cf.70-80% in supplier literature, HDPE is defined as having a low degree of branching), while the LDPE film had a lower crystallinity of 47% (ca.50% in the supplier literature, LDPE is defined as having a high degree of branching): these initial differences are clearly highly significant to the performances of the resulting AEMFCs.This also serves as a warning: a correctly selected combination of electron-beaming dose-rate and supplier (grade/additives) of base-material will be critical to any final, reproducible commercial production effort.
As there is no significant difference between the OH À conductivity or thicknesses of the HDPE-AEM and LDPE-AEM, we hypothesise that the improved performance is due to the enhanced water transport from the anode and cathode in the operating fuel cells [9][10][11] and this must be due to a change in the nanomorphology or microstructure between the LDPE-and HDPE-AEM.This hypothesis needs to be rigorously tested and so we are in the process of planning a series of quasi-elastic neutron scattering (QENS) and small-angle neutron scattering (SANS) experiments 13 to probe the ion-dynamics and nanomorphology in much more detail (comparing the new HDPE-based RG-AEM to both prior-art LDPE- 2,4 and ETFE-based 7 RG-AEMs).These results will be reported in a future, specific research paper.
These high performances were independently reproduced at the University of South Carolina with their own benchmarking parameters.The anode was PtRu(2 : 1)/C with a 0.7 mg cm À2 PtRu loading; the cathode was Pt/C with 0.6 mg cm À2 Pt loading; both catalyst layers contained 20 wt% radiationgrafted ETFE-based anion-exchange ionomer powder (IEC = 1.24AE 0.06 mmol g À1 ) and were sprayed onto Toray TGP-H-60 carbon-paper GDEs containing 5 wt% PTFE wet-proofing.The peak power density was 2.5 W cm À2 with a cell at 80 1C operating with H 2 /O 2 reacting gases.A 2.4 W cm À2 peak power density was also achieved in the same cell at an advantageously lower temperature of 70 1C (gases supplied at 1 dm 3 min À1 with an anode dew-point of 60 1C and cathode dew-point of 68 1C).The peak power density of this latter cell at 70 1C with CO 2 -free air at the cathode was 1.1 mW cm À2 .
Given the above data, a HDPE-AEM membrane-electrode assembly (MEA) was tested over 440 h at 600 mA cm À2 constant current discharge in a H 2 /air(CO 2 free) AEMFC at 70 1C (Fig. 5, caption details the test parameters used).‡ Both the cell voltage and area specific resistance (ASR) were recorded throughout the duration of the test.The ASR increased by 8.2 AE 0.2 mO cm 2 h À1 (95% confidence intervals, linear regression, R 2 = 0.55) during testing, indicating an excellent retention of the conductivities of the cell components.Secondly, over the 440 h of continuous operation, there was 7% voltage degradation when comparing the first and final data points (0.70 V -0.67 V), while a degradation rate of 68 AE 1 mV h À1 (95% confidence intervals, n = 15 386 data points, R 2 = 0.54) was estimated using a simple linear regression.At this stage we do not know the relative rates of degradation of each component (ionomer, catalyst, AEM) as this can't be elucidated using such a simple in situ durability test.This will be studied in much more detail in the future using a variety of more advanced techniques including operando tomography.
We also conducted an initial 100 h test with the LDPE-AEM for comparison (Fig. 5).This shows a much more rapid degradation rate of 790 AE 10 mV h À1 (95% confidence intervals, n = 3530 data points, R 2 = 087), which is why we terminated the test early.The failure mechanism with the LDPE-AEM clearly involved an increase in in situ ASR (38 -52 mO cm 2 ). ) and were sprayed directly onto Toray TGP-H-60 (10 wt% PTFE-treated) gas diffusion substrate; the H 2 /O 2 gases were supplied at 1 dm À3 min À1 at RH = 92% with no backpressure applied.This galvanostatic data was collected using a Scribner 850E fuel cell tester.

Fig. 1
Fig.1Degree of grafting (dog) of poly(VBC)-grafted membranes made from e-beamed LDPE and HDPE films that were stored (pre-grafted) at À40 1C for increasing periods of time.The dog (%) was calculated as: dog = 100 Â (m g À m i )/m i (where m i was the mass of the e-beamed pregrafted film and m g was the mass of the (pre-aminated) VBC-grafted membrane).7

Fig. 2
Fig. 2 Changes in OH À conductivity of the RG-AEMs when placed in a flowing relative humidity RH = 100% N 2 atmosphere at 80 1C for 500 h.Method discussed in detail in ref. 4 (where the LDPE-AEM data was first reported).

Fig. 4
Fig. 4 H 2 /O 2 AEMFC comparison tests (80 1C) for the RG-AEMs.Full details on the test conditions used are given in ref. 4 with the following summary data (the only variable in these tests was the RG-AEM used): PtRu/C anodes (50 wt% Pt and 25 wt% Ru) with a Pt-loading of 0.4 mg cm À2 ; Pt/C cathode (40 wt% Pt) with a Pt-loading of 0.4 mg cm À2 ; catalyst inks contained 20 wt% radiation-grafted ETFE-based anionexchange ionomer powder (IEC = 1.90 AE 0.06 mmol g À1) and were sprayed directly onto Toray TGP-H-60 (10 wt% PTFE-treated) gas diffusion substrate; the H 2 /O 2 gases were supplied at 1 dm À3 min À1 at RH = 92% with no backpressure applied.This galvanostatic data was collected using a Scribner 850E fuel cell tester.

Table 1
Key properties of the LDPE-and HDPE-AEMs compared.Error bars indicate sample standard deviations of measurements conducted on n = 3 different samples of each RG-AEM.Methods and parameters discussed in detail in ref. 4