Mechanically Stable Poly(arylene ether) Anion Exchange Membranes Prepared from Commercially Available Polymers for Alkaline Electrochemical Devices

Our previous works details the preparation of chloromethyalted PSF (CMPSF) and brominated PPO (BrPPO) from commercially available Udel PSF and PPO polymers.^34,40 BrPPO was prepared via free radical bromination using azobisisobutyronitrile (AIBN) as the initiator and n-bromosuccinimide (NBS) as the brominating reagent. As shown in our previous studies and by Xu et al.,⁴¹ the reaction temperature (100 to 130°C) ultimately determined the relative selectivity of bromination to the benzyl and aryl position of PPO, while the amount of NBS added determined the amount of bromine functionalized to PPO.

Chloromethylation of PSF and PPO was performed using a Friedel-Crafts reaction developed by Avram et al.⁴² Yan and co-workers adopted this procedure for PPO and they observed that the chloromethylation kinetics in PPO were more facile when compared to PSF.⁴³ The Avram et al. procedure specifies stannic chloride as the Lewis acid catalyst and chlorotrimethylsilane and paraformaldehyde as the chloromethylating reagents. Unlike the free radical bromination reaction, the degree of functionalization (DF) to PSF during the reaction was determined by time and not the amount of chloromethylating reagents added. The electronic supplementary information section (ESI) provides a detailed account of the synthesis methods.

The AEMs were prepared by dissolving the halogenated poly(arylene ethers) in an aprotic solvent (either n-methyl-2-pyrrolidone (NMP) or N,N-dimethylformamide (DMF)) and the addition of a tertiary amine - N,N-dimethylhexylamine (DMH) or trimethylamine (TMA). The Menshutkin reaction proceeded for 12 hours at room temperature for TMA and 60°C for DMH. The AEMs were gathered by drop casting the AEM reaction solutions onto glass plates in an oven and evaporating the solvent for 12 hours at 70°C. The films were removed from the glass plate after immersion in deionized water using a knife. See Scheme 1 for the synthesis of the halogenated poly(arylene ether)s and the subsequent formation of AEMs.

Scheme 1. Preparation of chloromethylated PPO, brominated PPO, and chloromethylated PSF followed by reaction of the halogenated polymers with a tertiary amine to prepare AEMs.

Membranes to be characterized in the hydroxide (OH⁻) or chloride (Cl⁻) counter-ion form were ion-exchanged immediately prior to characterization. See our previous work for the procedure.³⁴ The prepared AEMs (before and after exposure to alkaline solutions) were characterized by one-dimensional and two-dimensional NMR methods and FTIR spectroscopy to confirm the structure. Integration of ¹H NMR spectra was used to estimate the theoretical ion-exchange capacity (IEC), while the Vohlard titration was used to measure the experimental IEC. Electrochemical impedance spectroscopy was used to determine the in-plane ion conductivity in deionized water (∼18 MΩ) under temperature control; a 4-point probe was employed for this measurement. Water uptake (WU) was measured gravimetrically for AEMs exposed to humidified water vapor. To evaluate mechanical properties, stress-strain curves were generated using a dynamic mechanical analyzer (DMA). See the ESI and our previous reports for full details of each experimental method and for a description of the alkaline stability tests in 1 M KOH at 60°C for 10 days.^{34,35,38,40,44}

BrPPO was synthesized by free radical bromination. Varying the reaction temperature yielded different ratios of bromine at the aryl and benzyl positions of the PPO backbone. In this work, two batches of BrPPO (BrPPO Batch #1 and #2) were obtained using different reaction temperatures: 115°C and 130°C, respectively. The amount of bromine at the aryl position in BrPPO Batch #1 was 15% (i.e., 15% of repeat units had bromine to the aryl position) and the amount of bromine at the aryl position in BrPPO Batch #2 was 4%. As noted by Xu et al.,⁴¹ high reaction temperatures dictated selectivity of bromine to the benzyl position over the aryl position. The authors observed that adding more bromine to the aryl position made the polymer more hydrophobic. This allowed the use of higher IECs to achieve higher ionic conductivity while retarding excess water uptake. Arges et al. established that AEMs prepared from BrPPO with bromine at the aryl position achieved hydroxide ion conductivities close to 40 to 50 mS cm⁻¹ over the temperature range of 30°C to 70°C, but that the AEMs became extremely brittle in 1 M and 6 M KOH solutions at 60°C.³⁴

Figure 1a reports the ¹H-¹³C HMQC NMR spectrum of BrPPO-DMH⁺ from BrPPO Batch #1. The ¹H-¹³C HMQC NMR spectrum of BrPPO-DMH⁺ from BrPPO Batch #2 resembled the spectrum given in Figure 1a and is presented in the ESI under Figure S1. The signal associated with the dimethyl groups adjacent to the quaternized nitrogen in DMH⁺ was observed at 3.0 ppm for ¹H and 50 ppm for ¹³C. This signal verified that DMH⁺ was successfully attached to brominated PPO. Integration of the ¹H NMR dimethyl signals in DMH⁺ for BrPPO-DMH⁺ Batch #1 demonstrated that 103 ± 1% of bromomethylated groups in BrPPO Batch #1 reacted with DMH. Integration of the ¹H NMR for BrPPO-DMH⁺ Batch #2 yielded 98 ± 2% conversion of the bromomethyl groups to DMH⁺ cations.

Zoom In Zoom Out Reset image size

Figure 1b presents the ¹H-¹³C HMQC of PSF-DMH⁺. Similar to Figure 1a, the dimethyl group adjacent to the quaternized nitrogen in DMH⁺ was observed at the ¹H-¹³C HMQC coupling of 3.0 ppm for ¹H and 50 ppm for ¹³C. Integrating the ¹H NMR for PSF-DMH⁺ groups revealed that 107 ± 2% of the chloromethylated groups reacted with DMH. The conversion apparently exceeded 100% because of the variability in the determined DF value in CMPSF and the error in the integration of the ¹H NMR spectra (which can be up to 10%).⁴⁵

Note: The assignments of the other signals associated with the PSF and PPO backbones seen in Figures 1a and 1b have been detailed in our previous work and are not repeated here.^34,38

Neither CMPPO-DMH⁺ nor CMPPO-TMA⁺ could be dissolved into aprotic solvents (DMF, dimethylsulfoxide, and N,N-dimethylacetamide) or chloroform. The lack of solubility prevented us from determining the conversion of chloromethyl groups to quaternary ammonium groups using NMR and from calculating the theoretical IEC. Therefore, FTIR spectroscopy was used to confirm successful preparation of these AEMs. Figures 2a and 2b present the FTIR spectra of pristine CMPPO-DMH⁺ and CMPPO-TMA⁺ AEMs. Both spectra show the C-N⁺ infrared vibration absorption at 1600 to 1610 cm⁻¹.^46,47 Additionally, both spectra showed broad absorption peaks in the range of 3300 to 3500 cm⁻¹ corresponding to the O-H vibrational stretching, arising from water solvating the quaternary ammonium cations and chloride anions in the AEM. The absorbance peaks observed in the range 2700 to 3000 cm⁻¹ for both spectra were attributed to the –CH₃ stretching from the PPO backbone and the TMA⁺ and DMH⁺ cations and the –CH₂– groups in DMH⁺ (in Figure 2b).⁴⁸ Additionally, the signal for C-Br stretching in bromobenzyl groups, expected at 740 to 750 cm⁻¹, was not observed in the FTIR spectra of the CMPPO-derived AEMs.^46,47 Considered together, these observations suggested that the CMPPO AEMs with TMA⁺ or DMH⁺ cations were indeed successfully prepared.

Zoom In Zoom Out Reset image size

Table I reports the DF values for the halogenated polymer precursors used for preparing the AEMs while Table II provides the theoretical and experimental IEC of the different AEMs prepared. The IEC values given for each AEM material were directly proportional to the DF values of their halogenated polymer precursors. For most samples where both experimental and theoretical IEC values were reported, the values were in agreement, suggesting that both the NMR integration and Vohlard titration methods were accurate and complimentary. However, there was a significant disagreement between the theoretical and experimental IECs of BrPPO-DMH⁺ AEM Batch #2 – 2.3 mmol g⁻¹ versus 1.5 mmol g⁻¹ respectively. A similar finding was reported by us previously on PSF AEMs with IEC values above 2 mmol g⁻¹,⁴⁰ where the PSF AEMs with large IECs (over 2 mmol g⁻¹) demonstrated larger theoretical IEC values versus the experimental IECs, while the experimental IEC and theoretical IEC of AEMs below 2 mmol g⁻¹ were in agreement. This trend suggests that the Vohlard titration method may have an upper limit with regard to its accuracy, which would not be unexpected because the accuracy of the method is predicated on complete ion-exchange between the chloride ions in the AEMs with the nitrate ions in the solution. AEMs with larger IEC values may suffer from incomplete ion-exchange. To verify this viewpoint, a two-stage Volhard titration was performed on the BrPPO-DMH⁺ AEM. The method and results (see Table S1) are given in the ESI section. However, the hypothesis of incomplete ion-exchange seems to be invalid based on the results (i.e., a single stage extraction was sufficient for a near complete ion-exchange). Future work will look to investigate the discrepancy in IEC values obtained by Vohlard titration technique and by integrating the ¹H NMR spectrum for AEMs with high IEC.

Table II also reports the vapor phase water uptake at 98% relative humidity at 25°C for each AEM. BrPPO AEMs Batches #1 and #2 had the lowest water uptake values. The water uptake for the PSF-DMH⁺ AEM was slightly higher than that of the BrPPO AEMs. CMPPO AEMs had the largest water vapor uptake (22% to 27%) despite having the lowest IEC. The water uptake of AEMs with large IECs can be suppressed through modification of the backbone by introduction of hydrophobic moieties. For instance, BrPPO-DMH⁺ Batch #1 AEMs had the highest IEC but the lowest water uptake. This was made possible by the incorporation of bromine to the aryl position of the polymer backbone. The vapor phase water uptake for BrPPO-DMH⁺ Batch #2 AEM at room temperature was 12%, while the vapor phase water uptake for CMPSF-DMH⁺ was 13%. Both of these AEM materials had similar IEC values (1.5 to 1.6 mmol g⁻¹) suggesting that PSF and brominated PPO Batch #2 were reasonably similar in their hydrophobic characteristics.

Figures 3a and 3b report the in-plane ionic conductivity of the PPO and PSF AEMs in the Cl⁻ form and OH⁻ form in DI water under temperature control. All AEMs, except PSF-DMH⁺_, had a modest increase in their ionic conductivity with increasing temperature. The PSF-DMH⁺ AEM, on the other hand, demonstrated a relatively larger increase in its ionic conductivity from 30 to 50°C but saw a rapid drop in ionic conductivity at 70°C. The decrease of the ionic conductivity at 70°C resulted from excess swelling in water at the higher temperature, which hindered ion mobility. We observed a similar effect (i.e., excess swelling compromising ionic conductivity) for PSF AEMs with high IECs and in the hydroxide form in our previous reports.⁴⁰ In future studies, we will seek to determine the appropriate IEC for PSF-DMH⁺ AEMs that maximize ionic conductivity while simultaneously minimizing water swelling. The hydroxide ion conductivity trend of the AEMs was similar to their chloride ion conductivity trend observed except that the ionic conductivity values were higher for the hydroxide form. The higher ionic conductivity stemmed from the 2.6× higher mobility of the hydroxide ion compared to the chloride ion.⁸ However, AEMs ion-exchanged to the hydroxide form mostly likely consisted of a mixture of hydroxide, carbonate, and bicarbonate anions despite efforts to purge the water of carbon dioxide by bubbling nitrogen gas through it. Figure S4 in the SI section reports the bicarbonate ion conductivity of one synthesized AEM variant – BrPPO-DMH⁺ (Batch #1). The bicarbonate ion conductivity was about the same as the hydroxide ion conductivity, thus confirming that the AEMs in the hydroxide form initially suffered from bicarbonate or bicarbonate formation. Hence, comparisons or trends between materials are best defined based on the chloride and bicarbonate ion conductivity.

Zoom In Zoom Out Reset image size

The PSF-DMH⁺ AEM demonstrated substantially higher ionic conductivity than the PPO AEMs (both BrPPO and CMPPO AEMs). In Li et al's work, the PPO AEMs with long alkyl tails (e.g., n = 16 carbons) gave rise to a micro-phase separation in the AEM structure – as revealed by small-angle X-ray scattering (SAXS). The authors also observed a broad peak with the smaller n = 6 alkyl tail in PPO-DMH⁺ AEMs, but did not observe a scattering pattern in the SAXS pattern of PPO-TMA⁺. Thus, the micro-phase separation of the ionic domains apart from the hydrophobic n-alky tail domains enabled better ionic conductivity. Because the PSF backbone is more flexible than the PPO backbone, the PSF AEM with DMH⁺ could display distinct phase separation behavior, which would explain the better ionic conductivity attained.

The mechanical properties of the AEMs were investigated using a dynamic mechanical analyzer that collected stress-strain curves. Table III reports the stress and strain at break for each pristine AEM sample in the chloride form. From Table III, it was apparent that the PPO AEMs prepared from chloromethylation rather than free radical bromination had remarkably better mechanical properties. Furthermore, the PSF AEMs had substantially better mechanical properties over the BrPPO AEMs. The poor mechanical properties of BrPPO AEMs were ascribed to the large, bulky bromine atom functionalized to the aryl unit in PPO. The significant difference in mechanical properties of CMPPO AEMs against BrPPO AEMs was attribute to residual, bulky Br present at the aryl position in BrPPO preventing efficient chain packing. Another possible explanation entails that the cation group orthogonal to the methyl group enables better chain packing over the cation attached directly to the methyl group. In this context, we note that the PSF AEMs had substantially better mechanical properties compare to the BrPPO AEMs. The strength of the PSF-DMH⁺ AEMs was good when compared with a commercially available Nafion 211 membrane (13.9 MPa). The mechanical properties of CMPPO AEMs were slightly better than the mechanical properties of CMPSF AEMs with the same cation. But, both the CMPSF AEMs and CMPPO AEMs demonstrated significantly better mechanical properties over BrPPO AEMs.

The alkaline stability of the AEMs was evaluated by monitoring the chloride ion conductivity, IEC, mechanical properties, and NMR spectra after exposure 1 M KOH at 60°C for 10 days. Using this combination of functional and spectroscopic tests allowed us to: i.) gauge which materials were resilient to alkaline solutions with a high level of confidence, ii.) determine the modes of hydroxide-ion induced degradation, and iii.) connect different degradation modes to the chemical structure of the AEM. Utilizing a multitude of test methodologies was the best strategy to understanding the diverse and complex degradation modes of AEMs.

The CMPPO-DMH⁺ and CMPPO-TMA⁺ AEMs were the only AEMs testable with respect to ionic conductivity after exposure to the 1 M KOH solution for 10 days at 60°C. Despite their mechanical resilience, these AEMs demonstrated significant losses in in-plane chloride ion conductivity. The chloride ionic conductivity of CMPPO-TMA⁺ dropped from 6.0 ± 0.3 mS cm⁻¹ to 2.1 ± 0.1 mS cm⁻¹ after immersion in 1 M KOH for 10 days at 60°C, while that of CMPPO-DMH⁺ dropped from 5.0 ± 0.6 mS cm⁻¹ to 1.2 ± 0.4 mS cm⁻¹ under the identical test condition. The BrPPO and PSF AEMs became too brittle to measure ionic conductivity even after 10 days of exposure to 1 M KOH at 60°C.

Ionic conductivity measurement methods cannot discriminate between anion and cation driven conductivity, stability data reported by using ion conductivity as the primary metric can sometimes be misleading.^44,49 For instance, if backbone degradation occurs yielding carboxylic acid functionalities but the fixed cation carriers are degraded, the carboxylic acid groups would contribute to conductivity of the sample even though the fixed cation carriers (which is what we are concerned with) were degraded. Hence, the experimental measurement of anion IEC by Volhard titration is one of the most reliable and simplest ways to confirm degradation of the fixed cation groups in an AEM. Table IV reports the IEC of the AEMs before and after exposure to 1M KOH at 60°C for 10 days. This table also reports the IEC data normalized to the AEMs' initial IEC values expressed as a percent. An unexplored consideration is the concentration of fixed ionic carriers (i.e., IEC) on the rate of AEM cation stability when exposed to alkaline solutions. Pan et al. showed that adding less ionic groups per repeat unit in Udel PSF backbone reduced the rate of cation triggered hydrolysis of the PSF backbone in alkaline media.⁵⁰ A limitation in the analysis of the alkaline stability data reported as the IEC value alone was that the initial IEC of each AEM material differed. Hence, the IEC data expressed as a relative percent provided a more useful comparison. The AEMs prepared from BrPPO-DMH⁺ Batch #1 retained 79% of their initial IEC after 10 days in 1M KOH at 60°C, while AEMs prepared from BrPPO Batch #2 only retained 13% of their initial IEC after being stored in 1 M KOH at 60°C for 10 days. These results suggested that adding more bromine to the aryl position stabilized the cation from hydroxide ion attack. However, we do recognize that a more systematic study exploring the amount of halogen to the PPO backbone and its influence on cation stability needs to be conducted to test this hypothesis. The PSF-DMH⁺ AEM retained only 27% of its initial IEC after 10 days after exposure in 1M KOH at 60°C and all the samples broke into pieces after 3 days. The cation groups of the CMPPO AEMs demonstrated better hydroxide ion attack resistance when compared to those of BrPPO AEM Batch #2 and PSF AEMs because their relative IEC drop after 10 days was smaller – the CMPPO AEMs retained 60% of their IEC.

Anion IEC and ionic conductivity experiments are practical metrics for assessing whether or not AEM materials are resilient in alkaline solutions. However, these experiments cannot conclusively confirm chemical degradation nor can they provide insight into the degradation products and their corresponding mechanisms. The latter is very important for understanding AEM degradation and for engaging in the rationale design of new AEM materials that are stable in alkaline media. Here, using 1D and 2D NMR spectroscopy, we have confirmed that the DMH⁺ cation degrades through dealkylation (a direct nucleophiic substitution method) in BrPPO and PSF AEMs and we have confirmed that this cation triggers ether-hydrolysis in PSF AEMs. Figure 4 is the ¹H-¹³C HMQC NMR spectrum of BrPPO-DMH⁺ Batch #1 after exposure to 1 M KOH at 60°C for 10 days. Figure S2 in the ESI is the ¹H-¹³C HMQC NMR spectrum of BrPPO-DMH⁺ Batch #2 after exposure to 1 M KOH at 60°C for 10 days. Figure S2 and Figure 4 were qualitatively identical. In both these spectra, a new peak appeared at the cross-section of 2.1 ppm for ¹H and 46 ppm for ¹³C (signal A) when compared to the pristine ¹H-¹³C HMQC NMR spectra of BrPPO-DMH⁺ Batches #1 and #2 (See Figures 1a and Figure S1 in the ESI). This peak was ascribed to the methyl group(s) in a benzyl tertiary amine moiety. Therefore, the DMH⁺ cation degraded via dealkylation of one of the methyl groups or the n-hexyl chain. Figure 5 illustrates the dealkylation pathway of degradation of DMH⁺. These AEMs, as previously stated, were observed to become brittle after exposure to alkali. Arges et al. linked the mechanical (in)stability of BrPPO AEMs to cation triggered chemical degradation of the backbone in alkaline media.³⁴ The peaks associated with the backbone degradation observed for the BrPPO AEMs in Arges et al.'s work could not be detected in Figures 4 and S2 in the ESI because the carbon atoms and protons in the DMH⁺ cation's n-hexyl alkyl chain had the same chemical shifts (1.3 ppm for ¹H and 30 ppm for ¹³C and 0.9 ppm for ¹H and 15 ppm for ¹³C) as the backbone degradation products observed in Arges et al.'s work. Li et al. demonstrated that this cation - DMH⁺ - and other variants with long alkyl chains and crosslinked alkyl chains, prevented chain scission in PPO AEMs, whereas PPO-TMA⁺ AEMs became brittle in alkaline solutions.^18,36,37 However, our findings challenge this result. Perhaps the difference lies in the fact that our BrPPO-DMH⁺ AEMs had residual bromine groups on the PPO backbone that spurred chain scission upon exposing the BrPPO-DMH⁺ AEMs to 1 M KOH solution at 60°C (Li et al.'s PPO AEMs did not have bromine attached to the PPO aryl position.)

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Similarly, the ¹H-¹³C HMQC NMR spectrum of PSF-DMH⁺ AEMs after exposure to 1 M KOH at 60°C for 10 days demonstrated the same tertiary amine signal (2.1 ppm for ¹H and 46 ppm for ¹³C - signal B) from the deakylation of DMH⁺ as seen in Figures 4 and S2 in the ESI for BrPPO-DMH⁺ AEMs. See Figure 6. Furthermore, phenyl alcohol was observed at 6.0 ppm for ¹H and 120 ppm for ¹³C (signal C) in the ¹H-¹³C HMQC NMR spectrum of PSF-DMH⁺ after exposure to alkaline solutions. The presence of signal C indicated the degradation of the PSF backbone in PSF-DMH⁺ AEMs via ether hydrolysis.³⁵ Hence, while DMH⁺ mitigated chain scission of the PPO backbone in PPO AEMs (provided there was no residual bromine in the backbone – as in Li et al.'s work), the same effect does not hold true when switching to an alternative poly(arylene ether) backbone like Udel PSF.

Zoom In Zoom Out Reset image size

CMPPO AEMs were examined by FTIR spectroscopy after exposing them to 1 M KOH for 10 days at 60°C. See Figures S3a and S3b in the ESI section. No significant changes were observed in the FTIR spectra. The experimental IEC measurements demonstrated that the cation groups remained in CMPPO AEMs after 10 days of exposure to 1 M KOH for 10 days at 60°C; the C-N⁺ stretching signal was still apparent at 1600 to 1610 cm⁻¹ in Figures S3a and S3b in the ESI. Cation degradation would result in benzyl alcohol groups or tertiary amines, but existing substituents in the PPO backbone and water would obscure those signals. For example, the 'C-N' stretch corresponding to tertiary amines would arise between 1050 cm⁻¹ to 1100 cm⁻¹. However, the 'C-O' stretching in aryl ethers also appear in this range. Likewise, the 'O-H' stretch in benzyl alcohol, resulting from direct nucleophilic substitution of the quaternary ammonium groups, would overlap with the 'O-H' stretching signal of water solvating remaining cation groups.

Table III also reports stress and strain results at the breaking point for BrPPO, CMPPO and CMPSF AEMs with DMH⁺ or TMA⁺ cations at 50°C and 50% RH as a function of exposure to alkaline solutions. After 10 days exposure to 1 M KOH at 60°C, PSF-DMH⁺ AEMs and BrPPO AEMs were too brittle to measure. Thus, neither of material was suitable for alkaline electrochemical applications at elevated temperatures. Conversely, CMPPO AEMs maintained adequate mechanical properties prior to and after exposure in alkaline media at 60°C for 10 days though their stress and strain values at break were slightly diminished after exposure to alkaline solutions. The mechanical stability of the CMPPO AEMs over the BrPPO AEMs and PSF AEMs stemmed from: a) the absence of electron withdrawing substituents in the CMPPO backbone (like bromine at the aryl position and sulfone), and b) the fact that each repeat unit containing a cation had two electron donating methyl groups rather than a single methyl group.