Radiation-induced grafting of a butyl-spacer styrenic monomer onto ETFE : synthesis of the most alkali stable radiation-grafted anion-exchange membrane to date

For the preparation of all the radiation-grafted anion-exchange membranes (AEM) in this study, poly(ethyleneco-tetrafluoroethylene) (ETFE) films of 25 μm (supplied by Nowofol Kunststoffprodukte GmbH, Germany) were first irradiated in air (containing O2) using a 4.5 MeV Dynamatron Continuous Electron Beam Unit (STERIS Synergy Health, South Marston, UK). The absorbed were doses were controlled by the number passes (10 kGy per pass). The films employed in this study were exposed to a total of 40 kGy absorbed dose. After irradiation and transport back to Surrey in dry-ice, films were stored in a freezer at -40 °C until the monomer grafting step was performed.

The use of aromatic polymer components in anion exchange membranes (AEMs) is widespread due to the availability of facile polymerisation and functionalisation procedures.However, the cationic head-groups that are covalently bound to benzylic positions are sensitive to substitution by hydroxide (leading to alkali unstable AEMs). 1 The major degradation products from alkali degradation of benzyltrimethylammonium cations are benzyl alcohol and trimethylamine (from direct S N 2 Bn mechanisms and via ylide formation). 2,3A strategy to reduce degradation via these pathways is to introduce an aliphatic (spacer) chain between the aryl and the cationic (ionexchanging) groups.As reviewed recently, 4 this strategy has proven successful for a wide variety of polymers containing aromatic group, [5][6][7][8][9][10][11] including poly(phenylene ethers), poly (phenylene) and polysulfones.In many cases, an improved ionic conductivity has also been observed on addition of spacergroups, which was ascribed to the increased mobility of the resulting quaternary ammonium-containing side-chains. 6,12,13igh performance AEMs containing a methylene-(C1)-group between the benzene ring and the cationic head-group have already been prepared by radiation-induced graing of vinylbenzyl chloride (VBC) monomer onto commodity ETFE polymer lms. 14The treatment of such lms in air with an e À -beam source leads to the functionalisation of the lms with peroxide groups (Scheme 1). 15Heating these radiation-activated polymer lms in the presence of VBC (the graing step) produces gra copolymers, ‡ which can be subsequently functionalised with amines to yield pendant quaternary ammonium groups.These AEMs give high performances in H 2 /O 2 anion-exchange membrane fuel cells (AEMFCs) but they also degrade during ex situ treatment in aqueous alkali solutions at >70 C. [16][17][18] We recently found that the efficiency of the graing step could be increased by the use of water in the grating mixture. 17However, it still remained impossible to radiation-gra linear aliphatic monomers, such as 6-chloro-1-hexene, onto ETFE, which would have eliminated the presence of base-labile benzylic groups in the resultant AEMs: radiation-graing is a radical polymerisation process, which enables polymerisation of styrenic monomers but not linear aliphatic vinyl monomers.As a compromise solution, a styrenic monomer was selected for Scheme 1 An outline of the synthesis of the pyrrolidinium-grafted-ETFE-based C4-AEM developed in this study.‡ The C1-AEM benchmark contains a single -CH 2group between the benzene ring and the cationic pyrrolidinium N-atom.
use that contains an alkyl-spacer-chain located between the vinylbenzene group and the cationic quaternary ammonium group: this eliminates the presence of (alkali-susceptible) benzylic quaternary ammonium groups while retaining the availability of readily-polymerizable vinylbenzene groups.
In this study, we report the synthesis of a radiation-graed AEM containing a (C4) butyl-spacer between the aromatic and quaternary ammonium groups (C4-AEM), which exhibits a reduced level of alkali degradation at 80 C. Theoretical and experimental studies on small cationic molecules have found that trimethylalkylammoniums bearing alkyl chains containing 4-6 carbon atoms in length show higher alkaline resistance than those with shorter or longer alkyl chains. 19,20We opted for a monomer, 1-(4-chlorobutyl)-4-vinylbenzene, containing a butyl-spacer located at the para-aromatic position (C4-monomer, Scheme 1), as larger monomers (e.g. with a C6-spacer) risk impeded diffusion into the ETFE lms in the graing process.The bromide form of this monomer has been previously prepared by Tomoi et al. and converted into cross-linked quaternary ammonium anion-exchange resins. 21The hydroxide form of these resins retained their ion exchange capacities at 100 C more than the (C1-linked) benzyltrimethylammonium analogues.Despite the presence of C-H bonds at the b-positions to the quaternary ammonium groups, these resins did not undergo high levels of the anticipated degradation via Hoffman elimination.
Tomoi et al.'s monomer synthesis involved Cu-mediated cross-coupling of vinylbenzylmagnesium chloride with 1,3dibromopropane. 21We applied the same method using 1bromo-3-chloropropane to give 4-(1 0 -chlorobutyl)styrene in 58% yield aer distillation (details in the ESI †).A divinyl by-product (Fig. S1 in the ESI †), resulting from homocoupling of the vinylbenzylmagnesium chloride, was also present in 23% molar ratio in the crude reaction mixture: this could be reduced to 2% on distillation (calculated from the 1 H NMR spectra, Fig. S2 in the ESI †).This by-product can act as a crosslinker during the graing step.Finally, 50 ppm of 4-tert-butylcatechol (radical polymerisation inhibitor) was added to the monomer (+2% crosslinker) mixture before use in the graing step: this matched the inhibitor concentration in the commercially available VBC monomer (supplied as a mixture of paraand meta-isomers) used to produce the C1-AEM benchmark.
The graing conditions employed were comparable to those described previously 17 but with the additional use of an ultrasound bath during the graing step.Nasef et al. recently reported that the yield and kinetics of radiation-graing could be enhanced by the use of sonication and that this additionally reduces the formation of undesired homopolymer by-products. 22With our synthetic protocol, a comparison between ultrasound-aided and ultrasound-free radiation-graing of VBC indicated that the use of ultrasound did not improve graing yield but did allow a reduction in graing time (6 h with sonication vs. 16 h without) with the elimination of poly(VBC) homopolymer formation (previously observed as a white solid on the walls of the graing vessel).The C4-monomer also underwent ultrasound-assisted radiation-graing but resulted in lower graing yields compared to the VBC, due to its higher molecular weight.Non-ultrasound graing of C4-monomer was even less successful and led to high levels of undesired homopolymer (conrmed by Raman spectroscopy, see Fig. S3 in the ESI †).
The optimum conditions for graing C4-monomer onto ETFE lms (yielding ETFE-polyC4) involved immersion of preirradiated ETFE (40 kGy in air) in N 2 -purged monomer-water mixtures containing the wetting agent 1-octyl-2-pyrrolidone (1% vol) and heating for 6 h at 70 C with sonication (35 kHz, 225 W).To account for the 28% higher molecular weight of C4-monomer, it was used at a concentration of 5% vol, while 4% vol VBC was used to synthesise the ETFE-polyVBC membrane that was then aminated to form the C1-AEM benchmark (all other graing conditions were identical).Graing was conrmed using Raman spectroscopy where signicant differences were observed between the two monomers (Fig. 1).The benzylic CH 2 Cl deformation (1270 cm À1 ) was not visible in the spectrum of ETFE-polyC4.The bands characteristic of meta-disubstituted benzene rings (1000 cm À1 and 700 cm À1 ) were also absent, 16,23 since the C4-monomer was synthesised from a paradisubstituted-only starting material.Homogeneous graing, throughout the core of both pre-aminated membranes, was conrmed by cross-sectional Raman maps of the ratio between the areas of the 1615 cm À1 /830 cm À1 bands, since these represent the aromatic ring/ETFE components, respectively (see Fig. S4 in the ESI †). 16,24or the amination of ETFE-polyVBC and ETFE-polyC4, to form C1-AEM and C4-AEM respectively, we selected N-methylpyrrolidine (MPY): this yields AEM cationic head-groups with superior alkaline stabilities and enhanced fuel cell performances compared to the traditional use of trimethylamine (TMA). 16We suspect that the reason behind the enhanced ex situ stability of MPY membranes is related to their higher l water (average number of H 2 O molecules per anion), being around 1.5 times those of TMA-based AEMs. 18As recently published, the number of water molecules solvating the OH À anions has a substantial effect on AEM chemical stability. 25,26Furthermore, attempts to aminate ETFE-polyC4 at room temperature with TMA proved unsuccessful due to the higher activation energies required for S N 2 reactions on non-benzylic primary alkyl chlorides (the high volatility of aqueous TMA solutions impedes the use of elevated temperatures on safety grounds).A prior optimisation study for the MPY amination of ETFE-polyVBC intermediates concluded that using 15% vol MPY in water for 6 hours at 60 C was the mildest and cheapest effective option (details in the ESI †).However, this did not translate to ETFE-polyC4 (containing lower reactivity, non-benzyl head-groups): 70 C and 24 h were required with the use of aqueous MPY (15% vol).
C4-AEM had 20% lower ion-exchange capacity (IEC) than the C1-AEM benchmark due to its lower Degree of Graing (DoG) (Table 1).However, the Water Uptakes (WU) and l water values were similar for both AEMs.Likewise, no major differences in the swelling degrees and the mechanical properties were seen between the two AEMs (Tables 1, S2, Fig. S6 and S7 in the ESI †).The Cl À conductivity of C4-AEM was lower than that of C1-AEM (Table 1 and Fig. S5 in the ESI †) but, as expected for AEMs containing similar pyrrolidinium chemistries and l water values, the IEC-normalised conductivities 27 were similar (Fig. 2).
The H 2 /O 2 AEMFC performances at 60 C of membrane electrode assemblies (MEA) comprising both AEMs were evaluated using gas diffusion electrodes (GDEs) that contained 20% wt ETFE-benzyltrimethylammonium-based anionexchange ionomer powder 28 and 80% wt electrocatalysts: the Pt/C cathodes and PtRu/C anodes (all 0.40 AE 0.01 mg Pt cm À2 ) were prepared as previously reported (details in the ESI †). 18The low current performances of both AEMs were identical, con-rming similar electrocatalytic activities (Fig. 3).However, the performance curves deviate at higher currents, where internal ohmic resistances and mass transport losses control performances.C1-AEM achieved a peak power density of 1.22 W cm À2 , whereas C4-AEM obtained 1.02 W cm À2 (under identical test conditions): this 17% lower power density is still good considering C4-AEM was 26% lower in conductivity at 60 C than C1-AEM.The peak power increased to 1.12 W cm À2 at 70 C with C4-AEM but no further improvement was observed when raising the temperature to 80 C (Fig. S8 in the ESI †).
The ex situ chemical stabilities were tested by ageing individual samples of each AEM in aqueous KOH (1 mol dm À3 ) at 80 C for 28 d.Post-aged samples were converted back into the Cl À -form and the IECs were re-measured (Table 2).The loss of IEC with C4-AEM was half that of the C1-AEM benchmark.For more insights into the degradation mechanisms, CHN + Cl elemental analyses were also performed (Fig. 4).The % Cl loss correlates with the % IEC loss as all degradation mechanisms lead to loss of Cl À anions.However, the differences between the % Cl and % N losses indicate multiple degradation mechanisms are operating (discussed previously: 16 see Fig. S9 in the ESI †).
Around a third of the degradation of C1-AEM involves loss of the N atoms due to OH À derived displacement of complete MPY head-groups (S N 2 Bn ).Degradation by this pathway is essentially eliminated with C4-AEM, as the partial negative charges on the a-C atom (attached to the quaternary ammonium N) are not stabilized by resonance with a phenyl ring (conrms the stabilising effect of the spacer-group). 4For C4-AEM, degradation via Hoffman elimination at the spacer-chain is also possible, which would lead to loss of N content: however, the near zero % N loss indicates that this, along with nucleophilic loss of MPY, does not occur to a signicant extent (in accordance with Tomoi et al.). 21Adding a spacer-chain appears to have "switched   off" the degradation mechanisms involving loss of N atoms (see the degradation mechanisms in Fig. S9 in the ESI †).Nevertheless, 13% degradation still occurs with C4-AEM, which is predominantly due to loss of positive charges (loss of Cl À anions) but with retention of N atoms on the polymer gras.This can proceed via nucleophilic substitution on the N-methyl group (S N 2 NMe ) or ring-opening of the 5-membered heterocyclic ring. 29A previous study of ETFE-based VBC-graed AEMs, with different head-groups, 16 observed 12-14% degradation that did not involve loss of N content (approximately constant for TMA-, MPY-and N-methylpiperidinium-based head-groups).To obtain further conrmation of the degradation mechanisms, solid-state 13 C and 14 N NMR spectra were recorded on the preand post-alkali-aged samples, but no discernible changes were observed on ageing (see Fig. S10 and S11 in the ESI †).

Conclusions and future research directions
The introduction of a (C4) butyl-spacer between the quaternary ammonium and the benzene rings of a ETFE-based radiation-graed anion-exchange membrane (AEM) led to a signicant enhancement in ex situ alkali stability (compared to the nonspacer methylene-(C1)-linked benchmark).A more realistic evaluation of chemical stability would involve alkali treatment of AEM samples at low relative humidity since, as recently reported, dehydration of the cathode during fuel cell operation accelerates degradation of the polymer electrolytes. 25,26,30These experiments will be considered for future, more extensive studies on these AEMs.The additional use of ultrasound during the graing step was essential for successful synthesis of the new C4-AEM.A lower degree of graing (cf. the C1-benchmark) led to a decrease in conductivity and fuel cell performance (peak power densities still exceeded 1 W cm À2 ).Next steps involve increasing the ion-exchange capacity (degree of graing) of the C4-spacer AEMs and development of a C4-type radiation-graed anion-exchange ionomer powder to enable in situ fuel cell durability studies with spacer-only anion-exchange polymer electrolytes in the membrane electrode assemblies (the lack of spacer-type ionomer powders currently precludes in situ durability studies as the current non-spacer ionomer powder degrades faster).

Instrumentation
High-resolution NMR spectra were obtained on a Bruker 500 MHz spectrometer. 1 H NMR spectra were referenced to TMS = 0 ppm.All NMR data is presented as follows: chemical shift (δ) in ppm, multiplicity, coupling constants (J / Hz), and integration.The butyl-spacer monomer (C4-monomer) was prepared in one step reaction between 4-vinylbenzyl chloride and 1-bromo-3-chloropropane according to the method described by Seki and co-workers. 1 4-Vinylbenzyl chloride was added dropwise to a mixture of Mg turnings (2.76 g, 115 mmol) in dry diethyl ether (52 mL) at 0 °C.Once the addition was finished, the mixture was stirred at room temperature for 1 h.LiCl (0.084 g, 2.0 mmol), anhydrous CuCl 2 (0.133 g, 0.10 mmol) and 1-bromo-3-chloropropane (50 mL, 500 mmol) were dissolved in THF (90 mL) in a separate flask, degassed and added to the Grignard reagent solution via syringe.The reaction mixture was stirred under N 2 at room temperature overnight.After quenching with 10 mL methanol, the resulting suspension was filtered and concentrated under reduced pressure.The resulting oil was dissolved in diethyl ether and washed with water.The organic phase was dried over anhydrous MgSO 4 and concentrated to yield a yellow oil that was purified by distillation at reduced pressure (0.09 mbar, 125 °C) to yield 4CVBC as a mixture containing 2% crosslinker (11.98 g, 60 % yield). 1

Optimised grafting procedure
For the grafting step with VBC monomer, a mixture containing 71.25 mL of deionized water, 0.75 mL of 1-octyl-2-pyrrolidone, and 3 mL of VBC (4 %vol of grafting mixture) was purged with nitrogen for 30 min at room temperature in a Schlenk tube.Following this, one pre-irradiated ETFE film sample (size ca. 13 cm × 13 cm) was taken from the freezer, weighed (m o ) and placed in the mixture in the Schlenk tube.The mixture was then purged for another 2 h and the tube was connected to the Schlenk line before immersing it in an ultrasound bath preheated to 70 °C.After 6 h of sonication at 70 °C (35 kHz and 30% of the instruments 750 W maximum power), the grafting mixture was discarded and the membrane was washed with toluene and sonicated at 70 °C for 5 min.The ETFE-polyVBC grafted film was finally dried in a vacuum oven at 70 °C for 4 h and weighed (m graft ) in order to determine the Degree of Grafting (DoG).
(%) = − × 100 For the C4-monomer grafted films (ETFE-polyC4), the exact same procedure was applied but with a 5 %vol dispersion of monomer in the grafting mixture.

Optimised amination procedure (yielding the anion-exchange membranes)
The grafted intermediates were rolled and placed into Schlenk tubes.A mixture of MPY (15 %vol) in UPW was added and the tube was heated in a water bath with stirring: 60 °C for 6 h for the ETFE-polyVBC and 70 °C for 24 h for the ETFE-polyC4.The resulting AEMs were thoroughly washed with UPW and heated in water at 60 °C for 1 h.The as-synthesised AEMs were then treated with aqueous NaCl (1 mol dm -3 ) for 2 h (changing the ionexchange solution 2 times during this period) and then washed thoroughly with UPW.The final Cl --forms of C1-AEM and C4-AEM, were stored in UPW until required for further analysis or testing.For the experiments on the determination of the optimal amination conditions, the same procedure was applied but with changes in amine concentration and reaction times.

Raman spectroscopy and microscopy
A DXR Raman microscope (Thermo Scientific) was used in this study, with a λ = 532 nm (10 mW) excitation laser focused through a confocal microscope.As well as the recording of Raman spectra of various films and membranes on the surface, the microscope mode was used to map the different components of (through thickness) cross-sectional samples of the (desiccator dried) AEMs.A 50× objective was used yielding a laser spot diameter of a theoretical minimum of 1 μm.Spectra were collected using the Thermo Scientific OMNIC TM Software with the use of the Array Automation function.The cross-sectional area maps of the AEMs were recorded with samplestage step sizes of 1 μm in the x and y directions (with the y direction being the cross-sectional thickness direction in this study); the vertical z displacement was fixed.A single spectrum at each sampling point (each pixel in the Raman maps) was recorded with a spectral range of 3350 -350 cm -1 with averaging of 4 acquisitions per spectrum (10 s per acquisition).

Water uptake (WU) and Swelling degree (SD)
AEM(Cl -) samples were removed from the storage UPW and excess surface water was removed by dabbing the surfaces of the samples with filter paper.The hydrated masses (m hyd ), area (A hyd ), and thicknesses (T hyd , measured using a digital micrometer) were speedily recorded (to avoid dehydration on prolonged exposure to the atmosphere that typically has a relative humidity RH < 100 % in the laboratories).The AEM samples were subsequently dried in a vacuum oven at 50 °C for 15 h, after which the dry masses (m dry ), area (A dry ), and thicknesses (T dry ) were recorded.All measurements were repeated on n = 4 samples of each AEM(Cl -).The WU and SD values were then calculated using the following equations: Fig. S2 1 H NMR spectra of C4-monomer: crude (top) and purified by distillation (bottom).Table S2 Tensile mechanical properties of both AEMs (extracted from the data in Figure S7 above).

Fig. 1
Fig.1Raman spectra of the pre-aminated radiation-grafted ETFEfilms synthesised using the C4-monomer and commercially available VBC (meta-and para-isomer mixture).All spectra have been normalized to the 830 cm À1 band (related to the ETFE backbone) to aid visual comparison.Laser l ¼ 532 nm.

Fig. 4 %
Fig. 4 % loss of IEC and % losses of Cl/C and N/C molar ratios, the latter extracted from elemental analyses, of the AEM samples before and after ex situ alkali treatment in aqueous KOH (1 M) at 80 C for 28 days.

Fig. S4
Fig.S42D cross-sectional Raman maps of both (pre-aminated) grafted membranes.Laser λ = 532 nm.The numbers on the scale-bar represent the band area ratio of the 1615 cm -1 / 830 cm -1 (aromatic ring / ETFE moieties): note the intrinsic intensity of the 1615 cm -1 band is different for the different monomers so this Raman map data does not indicate higher levels of grafting with the C4-monomer (this data is just an indication of grafting homogeneity through the membrane cores).

Fig. S5
Fig. S5 Cl --anion conductivities (4-probe, in-plane, measured in water) for both AEMs.Errors bars are from repeat measurement on n = 3 samples of each AEM.

Fig. S6
Fig. S6 Graphical comparision of the key properties of both AEMs.

Fig. S7
Fig. S7 Tensile stress-strain curves obtained at room temperature with n = 3 samples of each AEMs.

Fig. S9
Fig. S9The alkali degradation mechanisms for C1-AEM and C4-AEM that were discussed in the main paper.

Figure S10 13 C
Figure S1013 C SS-NMR spectra of the AEMs before (black) and after the alkali treatment (red) in aqueous KOH (1 mol dm -3 ) at 80 °C for 28 days.The spectrum of ETFE-polyC4 is plotted in grey (bottom panel).All spectra have been normalized to the intensity of the δ = 22 band (corresponding to the methylene CH 2 groups in the ETFE backbone) for visual comparison.

Figure S11 15 N
Figure S11 15 N SS-NMR spectra of the before and after alkali treatment.

Table 1
Key properties of both radiation-grafted AEMs studied (errors are from repeat measurement on n ¼ 3 or 4 samples of each AEM: n indicated in brackets) a Average number of H 2 O molecules per Cl À anion calculated as: l water ¼ WU (%)/(100 Â 18.02 Â IEC), where IEC is in mol g À1 .b Swelling degree in the x-y (in-plane) direction.c Swelling degree in the z (through-plane, thickness) direction.d The Cl À anion conductivities at 60 C of the fully hydrated AEMs (4-probe, in-plane measurements with the RG-AEM submerged in water).

Table 2
The IEC values of AEM samples (n ¼ 4) recorded before and after ex situ alkali ageing in aqueous KOH (1 mol dm À3 ) at 80 C for 28 d

Table S1
Data from the optimisation study for the amination of ETFE-polyVBC using N-methylpyrrolidine (MPY).All AEMs were prepared from the same batch of ETFE-polyVBC intermediate (83% degree of grafting).