- Email: [email protected]
A practical method for measuring the ion exchange capacity decrease of hydroxide exchange membranes during intrinsic degradation
Journal of Power Sources xxx (2017) 1e6
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
A practical method for measuring the ion exchange capacity decrease of hydroxide exchange membranes during intrinsic degradation Klaus-Dieter Kreuer a, *, Patric Jannasch b a b
Max-Planck-Institute for Solid State Research, Heisenbergstrasse 1, D-70569 Stuttgart, Germany Department of Chemistry, Polymer & Materials Chemistry, Lund University, SE-22 100 Lund, Sweden
h i g h l i g h t s A new method for quantifying IEC decrease of hydroxide exchange membranes. Degradation rate of HEMs depends on temperature, hydration number and total water content. Method could help to resolve debates about relative durability of various HEMs.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 June 2017 Received in revised form 25 July 2017 Accepted 26 July 2017 Available online xxx
In this work we present a practical thermogravimetric method for quantifying the IEC (ion exchange capacity) decrease of hydroxide exchange membranes (HEMs) during intrinsic degradation mainly occurring through nucleophilic attack of the anion exchanging group by hydroxide ions. The method involves measuring weight changes under controlled temperature and relative humidity. These conditions are close to these in a fuel cell, i.e. the measured degradation rate includes all effects originating from the polymeric structure, the consumption of hydroxide ions and the release of water. In particular, this approach involves no added solvents or base, thereby avoiding inaccuracies that may arise in other methods due to the presence of solvents (other than water) or co-ions (such as Naþ or Kþ). We demonstrate the method by characterizing the decomposition of membranes consisting of poly(2,6dimethyl-1,4-phenylene oxide) functionalized with trimethyl-pentyl-ammonium side chains. The decomposition rate is found to depend on temperature, relative humidity RH (controlling the hydration number l) and the total water content (controlled by the actual IEC and RH). © 2017 Elsevier B.V. All rights reserved.
Keywords: Hydroxide exchange membrane AEM Degradation TGA Fuel cell
1. Introduction The potential application of anion exchange membranes (AEMs) in their hydroxide (OH) form (sometimes denoted by HEM) as separators in low temperature fuel cells is a matter of ongoing research. In contrast to acidic proton exchange membranes (PEM) that require the use of noble metal electro-catalysts to form a fuel cell electrode, the basic conditions under which HEMs operate allow less costly non-noble metal catalysts to be used [1]. OH conductivity values approaching the proton conductivity of PEMs (such as Nafion®) are common [2] for high levels of hydration, and bicarbonate (HCO 3 ) formation from the reaction of OH with CO2 present in air (used as oxidant in fuel cells) may be reduced by
* Corresponding author. E-mail address: [email protected] (K.-D. Kreuer).
using CO2 absorbers and adequate electrochemical management (self-purging [3]). The major problem stems from the mandatory presence of highly nucleophilic hydroxide ions as conducting ion. Quaternized ammonium (QA) groups are commonly used as positive countercharge within the polymeric structure; however, as typical leaving groups, QAs are well-known to react with OH through different energetically similar pathways [4] including nucleophilic substitution, b-elimination, and rearrangement reactions such as Stevens rearrangement in the absence of b-protons. As a consequence, HEMs inherently decompose, thereby losing ion exchange capacity (IEC) and ionic conductivity [5e10]. The degradation rate depends not only on the kind of QA but also on its environment within the polymeric structure. To deal with this complexity, we studied the degradation rates of a series of QA-salts in concentrated aqueous solutions of NaOH to identify
http://dx.doi.org/10.1016/j.jpowsour.2017.07.106 0378-7753/© 2017 Elsevier B.V. All rights reserved.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
2
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
suitable candidates for ionic groups in HEMs [11]. In this small molecule study, we also investigated the effects of different types of attachment, e.g. benzylic attachment is found to be extremely critical with respect to stability which is consistent with the results of another small-molecule study showing higher stability for alkylsubstituted compared to benzylic-substituted cations [12]. In the end, however, it is critical to test the durability of membranes under conditions similar to those in an operating fuel cell. These conditions differ from the conditions provided by aqueous solutions of NaOH (KOH) in various ways: i) within a HEM, OH counter ions are consumed in degradation reactions while the concentration (activity) of OH in excess NaOH solution is virtually unaffected by membrane degradation. ii) For high molarity, significant co-ion uptake (which corresponds to an uptake of excess NaOH) may affect the degradation rate through the presence of Naþ in the membrane. iii) The molar ratio [H2O]/[OH] in aqueous solutions of NaOH may be higher than for the low hydration conditions which may occur in running fuel cells. Especially the cathode side is expected to dry out as a result of electroosmotic water drag [13,14] from the cathode to the anode side at high ionic (OH) current density. Ion solvation (type of solvent and degree of solvation) actually affects degradation rates, and especially heavy hydration is known to strongly reduce the reactivity of the highly nucleophilic OH ion [11]. In a recent small-molecule NMR study, Dekel et al. [15] have quantified this effect in ex situ tests for hydration levels l ¼ 0 (dry) e 4 [H2O]/[OH] by adding defined amounts of water to dry solutions of QA and KOH in DMSO and following the degradation by 1H NMR spectroscopy. Here we present a gravimetric method which allows one to follow the IEC decay of HEMs at controlled temperature and hydration levels. The approach captures all effects that originate from the polymeric structure of the membrane, while avoiding possible inaccuracies arising from the unrealistic presence of solvents (other than water), co-ions (such as Kþ), or an excess of counter ions (OH). Moreover, the method is straight-forward and requires only a thermogravimetric analyzer [16]. 2. Basic idea As generally observed for PEMs, also the hydration number (l ¼ [H2O]/[ionic group]) for HEMs is approximately proportional to the IEC for a given temperature T and relative humidity RH (RH < 65% where hydration is exothermic). This approximately holds for different types of QA [data not yet published] and is especially true when a given type of ionomer with different concentration of a given type of QA is concerned. Therefore, also the difference of hydration numbers for two different values of RH (Dl) is a good measure for the actual IEC. In the proposed method, degradation occurs under controlled T and RH conditions while continuously recording the sample weight. This decreases with time for two reasons:
at that time. In this way, the evolution of IEC during degradation is recorded as a function of time. From this, information about reaction order and reaction rate may directly be deduced. Conceptually, the method works provided hydroxide ions merely react with QA functional groups (which is mostly the case). If hydroxide ions are also consumed in reactions with the polymer backbone, e.g. through nucleophilic attack of aryl-sulfones [17] or aryl-ethers [18], a decrease of Dl is expected as well because the reaction products (e.g. phenylates) are less hygroscopic than hydroxide ions. In order to exclude such reaction pathways, the IEC should be double checked at the end of the experiment by an independent method (see below). An IEC value higher than the one obtained from TGA would then be indicative for backbone degradation.
3. Experimental All experiments were carried out with a HEM consisting of a poly(2,6-dimethyl-1,4-phenylene oxide) (PPO) backbone functionalized with trimethyl-pentyl-ammonium side chains (Scheme 1). Since there is no benchmark membrane for AEMs (like in the case of PEMs for which Nafion is generally used for benchmarking) we have selected this homemade membrane because its chemical structure is related to the chemical structures of several AEMs for which reasonable stability in aqueous NaOH solutions have been reported [19e22]. The ionomer has been prepared as described earlier [23] and membranes were formed from NMP solution at T ¼ 80 C. The IEC of aseprepared membranes was determined in the Br form by displacing Br with NO 3 (~100 mg membrane in 40 ml 0.2 M NaNO3 for three days) and titrating the released Br with AgNO3 using a silver-electrode. Membranes in their Br form were converted into the OH form by ion exchange in 1 M NaOH (200 mg membrane in 40 ml solution) where the solution was replaced five times every 24 h under CO2-free conditions (in a homemade glove box [24]). This turned out to be necessary in order to reduce the Br content to less than 3% of the IEC. The residual Br content of OH exchanged membranes was actually determined by displacing Br with NO 3 after each exchanging step and subsequent Br titration (see above) using small pieces of the membrane. For the ion exchange procedure, one should keep in mind that ionic association in the NO 3 , Br and especially the I forms of AEMs leads to a severe stabilization compared the OH form [2]. The hydroxide exchanged membranes were rinsed with deionized water until neutrality and then kept in pure water not longer than a few days before carrying out the stability experiments. For this, a thermogravimetric balance with magnetic coupling is used (for details see ref. 13) membrane samples (100e200 mg) were placed in a quartz crucible within the glove box and then
i) volatile reaction products may be set free, ii) hydration water is released because of decreasing IEC (degradation of ionic QA groups). The first contribution depends on the kind of reaction products (alcohols, amines) and the polymeric host, degree of hydration and temperature determining the rate at which the degradation products are released. The second contribution, however, is mainly controlled by the decreasing IEC. Both contributions can easily be separated by intermittently measuring the hydration number for a higher RH. The increase of l (Dl) then is a measure of the actual IEC
Scheme 1. PPO grafted with trimethyl-pentyl-ammonium cations.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
3
transferred to the balance using a gas tight glass container. Nitrogen gas was passed through a CO2 absorber (sodalime with indicator, Merck No. 1.06733.0501) before entering the humidification system of the balance. In the present study, we have first recorded a hydration isotherm at T ¼ 40 C by equilibrating the membrane at RH ¼ 80, 65, 50, 30 and 10% for 10 h at each humidification level. Since some decomposition is observed for RH < 10% even at this low temperature, the weight of the dry sample could only be determined in the F form in a separate experiment (on the same ion exchanged membrane) by drying the membrane in a stream of dry nitrogen at T ¼ 100 C in the balance. The dependence of the hydration number in the OH form was measured for two RH values (50 and 65%) in the interval T ¼ 40e100 C (above this temperature, there was no reversibility as a consequence of decomposition). Stability tests were carried out at two temperatures (T ¼ 60 and 100 C) and two relative humidity values (RH ¼ 10 and 50%) in order to separate the dependence of the reaction rate on both parameters. In all experiments, the decomposition intervals were 10 20 h interrupted by 5 h intervals with higher RH (50 or 65%) for obtaining the transient IEC. Only for the most aggressive conditions (T ¼ 100 C, RH ¼ 10%) shorter decomposition intervals were chosen in order to increase the time resolution. After long term TGA (approx. 250 hrs.), the membranes were exchanged into the Cl form (in an excess of 1 M NaCl for five days at T ¼ 40 C) before determining the residual IEC in the same way as described before. This was done in order to cross-check the IEC decrease as determined by TGA. In order to compare the proposed methodology with the commonly used stability test in caustic aqueous solutions we have also immersed membrane pieces in 1 and 10 M aqueous KOH at T ¼ 100 C (approx. 200 mg membrane in 50 ml KOH solution in a Teflon crucible placed in a closed steel autoclave). Since we are interested in the evolution of the degradation with time, the autoclave was cooled and open after regular time intervals for taking out pieces of about 50 mg. These were rinsed with deionized water and then transferred into a 1 M NaCl solution for ion exchange followed by a determination of the residual IEC (see above). 4. Results and discussion Since the suggested method relies on defined relative weight changes as a consequence of degradation, the hydration isotherm for conditions under which HEMs are virtually stable (T ¼ 40 C, RH > 10%) is discussed first. In an earlier study we had already found that anion exchange membranes (AEM) may significantly swell in water for anions with high hydration enthalpy. This is the case for anions like OH and F in contrast to more hydrophobic anions such as Cl, Br and I, which tend to form ionic aggregates with the immobile quaternary ammonium cations [2]. For the present AEM with an IEC of close to 1.6 meq./g, even at room temperature, water uptake in pure water is very high (l ¼ 54 corresponding to a weight gain of 165% with respect to the dry membrane). Then, drying the membrane at a defined RH does not lead to a stable hydration level even after 10 h (Fig. 1a). However, subsequent rehydration leads to stable hydration numbers l for not too high (RH < 65%) and not too low RH > 10% (Fig. 1b). For the membrane used in this study, even for a temperature as low as T ¼ 40 C, some degradation is indicated by a slow weight loss under very dry conditions (RH ¼ 10%, Fig. 1a). As in the case of acidic membranes, l depends on temperature T. Since the ion exchanging group (tetra-methyl-ammonium hydroxide) is only a moderately strong base, its affinity toward water decreases with T. This is clearly visible for l (T, RH) which decreases from l ¼ 3.80 at T ¼ 40 C to l ¼ 2.92 at T ¼ 100 C for RH ¼ 50% (Fig. 2).
Fig. 1. Membrane hydration at T ¼ 40 C: a) TGA trace for step wise drying (RH ¼ 65, 50, 30, 10%) and rehydration (RH ¼ 30, 50, 65%). Note, that the weight loss at RH ¼ 10% indicate the onset of decomposition. b) Hydration isotherm compared to the hydration isotherm of Nafion [13].
For higher RH, l is getting increasingly dependent on the membrane's pre-treatment, for higher T there is significant decomposition even at high RH. Prior to the actual stability test, it is important to ensure a stable hydration number at the respective temperature at an RH where decomposition is slow. This hydration number is then considered to be the hydration number l of the membrane with its initial IEC . Depending on T and RH under which decomposition is recorded, we have chosen RH ¼ 50 or 65% for probing the transient IEC. Fig. 3 shows the TGA traces for T ¼ 60 and 100 C. While the weight was virtually stable during the probing regimes (RH ¼ 50, 65%), decomposition is indicated by both the decreasing difference in l for the two RH levels and the weight decrease during the low RH regimes (decomposition regime). Only for the most aggressive conditions (T ¼ 100 C, RH ¼ 10%), very fast initial decomposition is observed before the nominal decomposition RH (10%) is attained. The transient IEC(t) as obtained from: IEC(t) ¼ (Dl(t)/Dl ) IEC
(1)
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
4
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
is shown in Fig. 4. The fact that the IEC values obtained by titration after the degradation tests are virtually identical to the last data point (at about T ¼ 200 h, Fig. 4) gives confidence that the proposed procedure provides reliable data for IEC(t). In particular, the IEC obtained by titration is never higher than this deduced from the TGA data suggesting relative backbone durability under the given conditions. Only for the most severe degradation conditions (T ¼ 100 C/RH ¼ 10%, see above), membranes did not show any ion exchange activity after the test, although thermogravimetric analysis indicates some residual hygroscopicity (Fig. 3). It is not surprising that the observed degradation rates do not follow first order reaction kinetics: IEC(t) ¼ IEC exp (-k t)
Fig. 2. Membrane hydration as a function of temperature for RH ¼ 50% and 65%. The corresponding data for Nafion [13] are given for comparison.
(2)
as observed in small molecule degradation studies in aqueous solutions with a large excess of NaOH [10]. In the present case, both QA cations and OH counter ions are consumed in the degradation reaction which may suggest second order reaction kinetics: 1/IEC(t) ¼ 1/IEC þ k t
(3)
But this is not confirmed for any degradation conditions chosen in this study. For these, the degradation rate seems to decrease with increasing degree of degradation, i.e. with decreasing water content. In fact, there is indication that the decreasing amount of water decreases the global dynamics of the system and as a consequence also the rate of degradation (see below). Another part of the complexity of the reaction kinetics is probably the fact that degradation products are more or less released depending on T and RH. If all reaction products were retained in the membrane, the weight loss at a given RH would simply correspond to the reduced water uptake as a consequence of IEC decrease. Then, IEC(t) is: IEC(t) ¼ (l(t)/l ) IEC
(4)
In fact, this is approximately the case for low T and high RH. Except for T ¼ 60 C and RH ¼ 50%, however, the total weight loss is about 2e3 times higher than expected for this case (Fig. 3). In other words, about 36e54 g reaction products per mole ionic groups are released. This is also reflected in the negative values for l in Fig. 3 (note that for this representation, l is calculated assuming a
Fig. 3. TGA traces for T ¼ 60 C and 100 C. Relative humidity RH values are given for the degradation regimes and the probing regimes (in brackets). The weight changes are expressed in terms of hydration number l assuming a constant IEC ¼ IEC (see text).
Fig. 4. IEC decay for different degradation conditions as obtained from TGA traces (Fig. 3). IEC data obtained by titration after the experiments are also included.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
constant ion exchange capacity IEC ¼ IEC ). This finding underlines the need to periodically apply higher RH for probing the actual IEC(t) from Dl for two RH values (1) and suggests that sluggish release of reaction products may contribute to the complexity of the degradation kinetics. Other reasons may be competing reaction pathways or more than one OH may be involved in the degradation of one ionic group. The latter may explain the observation that even under the most severe degradation conditions (T ¼ 100 C, RH ¼ 10%), there is still some IEC left after more than 100 h (Fig. 4). In any case, degradation cannot be described by a single degradation rate k (any order), and this is what is also expected for membrane degradation in a running fuel cell. In all cases the degradation rate seems to decrease with proceeding degradation. It is not the aim of the present work to identify details of the reaction scheme on a molecular level. This is expected to be different for different types of membrane. But the degradation of all HEMs goes along with a decrease in IEC, and therefore, IEC(t) for defined T/RH conditions is a meaningful parameter (function) for comparing the durability of different types of membrane.
5. Comparison to other durability tests and some practical considerations We assume that the present methodology provides the most realistic and precise way to obtain rates for the intrinsic degradation reactions in HEMs. The T/RH conditions can be chosen close to those occurring in a running fuel cell, and the obtained ion exchange capacity transients (IEC(t)) appear to well reproduce data obtained by titration. It goes without saying that other reactions, e.g. with peroxyl and hydroxyl radicals, which may also take place in a fuel cell, are not captured by this method. Compared to tests in aqueous solutions of KOH (NaOH), no additional cations, anions, and water are present in the membrane. Especially at high molarity, Donnan exclusion is less efficient, i.e. significant amounts of KOH may enter the membranes from solution, the water uptake being mainly determined by the membrane's viscoelastic properties [16]. Naively thinking, one might choose a 10 M KOH for a stability test because for this the ratio [H2O]/[KOH] is about 4.7 which corresponds to a similar value for l at RH ¼ 60%. But the comparison of experimental results for a test in 10 M KOH and at RH ¼ 50% (l ~ 3.7) both at T ¼ 100 C (Fig. 5) show that
5
degradation in solution is faster. Part of this difference is surely the large excess of hydroxide ions which may compensate for OH consumed in the reaction, but the fact that the initial IEC decay is significantly stronger in solution also suggests a difference in the rates of the underlying elementary reactions. In solution, the total uptake of water and ions is higher although the water to ion ratio (l) is similar, and this may increase the dynamics in general and therefore also the rate of degradation. For degradation tests in solution of low KOH molarity, KOH uptake is surely lower, but such tests do not provide a defined degree of hydration l. E.g. for a 1 M KOH, the molar water to base ratio is about 54 which is usually much higher than the hydration number l attained by the membrane immersed in this solution (again dependent on the membrane's viscoelastic properties). But since the degradation rate strongly depends on water content (hydration number), KOH molarity alone does not guarantee defined degradation conditions. In the present case, degradation in 1 M KOH is again surprisingly fast (Fig. 5) considering the high molar ratio of water to ion in solution. Obviously the hydration number is lower in the membrane, and the degradation rate is expected to be even faster for membranes with the same ion exchanging group but a tougher backbone more effectively limiting swelling (hydration number l). Since fixing T/RH (RH < 65%) allows setting l in a very narrow range, the method presented here is expected to produce more defined results. It should be noted that the activity (chemical potential) of the mobile species are also defined for membranes in contact with solutions, but not the number of water molecules per ion. This is because of the membrane's internal pressure is another degree of freedom entering into the species chemical potential [16]. The present results actually suggest that the rate of degradation is more dependent on the hydration number l than on the chemical potential of water. The key finding of the degradation test of trimethyl-benzylammonium salts recently published by Dekel et al. [15] is the extent to which the degradation rate is affected by lowering hydration at room temperature. Comparing their data for l ¼ 1 and 4 (Fig. 3 in Ref. [15]) with the data we have obtained at T ¼ 60 C with RH ¼ 10% and 50% (corresponding to l ¼ 1.15 and 3.75), confirms this effect in an almost quantitative manner. However, the methodology presented here allows recording the degradation rate of membranes (as opposed to small molecules/ions) and easily reveals its temperature dependence. For the type of membrane considered here, increasing T from 60 to 100 C increases the degradation at RH ¼ 50% more than decreasing RH from 50% to 10% does (Fig. 4). It seems that both hydration level (RH, l) and T are the major parameters controlling inherent membrane degradation. Ideally, the test suggested here may be made by a TGA system which allows to control T and RH in the range T ¼ 40e100 C and RH ¼ 10e65%. The conditions for which data are reported here (Fig. 3) may be used as guide line for choosing appropriate values for T and RH. But even TGA systems which allow to set low RH at relatively low T only may provide useful results since for many membranes significant degradation rates may occur e.g. at T ¼ 60 C and RH ¼ 10% (Fig. 4). These conditions may be accessible by virtually any TGA device. Acknowledgement
Fig. 5. Membrane degradation (IEC decay) as determined by TGA and by treatments in caustic aqueous solutions. Note that the ratio [H2O]/[OH] is about 54 for 1 M NaOH and 4.7 for 10 M KOH.
The authors would like to thank A. Fuchs and U. Klock (MaxPlanck-Institute for Solid State Research) and H.-S. Dang (Department of Chemistry, Lund University) for technical assistance. We are also very grateful to R. Usiskin and T. Saatkamp (Max-PlanckInstitute for Solid State Research) for reading the proofs. We also would like to thank Alexey Serov (University of New Mexico) for giving us the chance to publish in this Special Issue.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
6
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
References [1] J.R. Varcoe, R.C.T. Slade, Fuel Cells 5 (2005) 187e200. [2] M.G. Marino, J.P. Melchior, A. Wohlfarth, K.D. Kreuer, J. Membr. Sci. 464 (2014) 61e71. [3] M. Inaba, Y. Matsui, M. Saito, A. Tasaka, K. Fukuta, S. Watanabe, H. Yanagi, Electrochemistry 79 (2011) 322e325. [4] S. Chempath, J.M. Boncella, L.R. Pratt, N. Henson, B.S. Pivovar, J. Phys. Chem. C 114 (2010) 11977e11983. [5] Y. Ye, Y. A. Elabd, Chemical Stability of Anion Exchange Membranes for Fuel Cells, Chap. 15, pp. 233 e251. URL http://pubs.acs.org/doi/abs/10.1021/bk2012-1096.ch014. [6] T. Sata, M. Tsujimoto, T. Yamaguchi, K. Matsusaki, J. Membr. Sci. 112 (1996) 161e170. [7] G. Couture, A. Alaaeddine, F. Boschet, B. Ameduri, Prog. Polym. Sci. 36 (2011) 1521e1557. [8] G. Merle, M. Wessling, K. Nijmeijer, J. Membr. Sci. 377 (2011) 1e35. [9] C.G. Arges, J. Parrondo, G. Johnson, A. Nadhan, V. Ramani, J. Mater.Chem. 22 (2012) 3733e3744. [10] D. Chen, M.A. Hickner, ACS Appl. Mater. Interfaces 4 (2012) 5775e5781.
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
M.G. Marino, K.D. Kreuer, ChemSusChem 8 (3) (2015) 513e523. A.D. Mohanty, C. Bae, J. Mater. Chem. A 2 (2014) 17314e17320. Y.S. Li, T.S. Zhao, W.W. Yang, Int. J. Hydrogen Energy 35 (2010) 5656e5665. X. Wang, J.P. McClure, P.S. Fedkiw, Electrochim. Acta 79 (2012) 126e132. D.R. Dekel, M. Amar, S. Willdorf, M. Kosa, S. Dhara, C.E. Diesendruck, Chem. Mater. 29 (10) (2017) 4425e4431. K.D. Kreuer, Solid State Ionics 252 (2013) 93e101. E.A. Weiber, P. Jannasch, J. Membr. Sci. 481 (2015) 164e171. C. Fujimoto, D.-S. Kim, M. Hibbs, D. Wrobleski, Y.S. Kim, J. Membr. Sci. 423e424 (2012) 438e449. S.A. Nunez, C. Capparelli, M.A. Hickner, Chem. Mater. 28 (8) (2016) 2589e2598. W.H. Lee, A.D. Mohanty, C. Bae, ACS Macro Lett. 4 (4) (2015) 453e457. Z.J. Yang, J.H. Zhou, S.W. Wang, J.Q. Hou, L. Wu, T.W. Xu, J. Mater. Chem. A 3 (29) (2015) 15015e15019. H.S. Dang, E.A. Weiber, P. Jannasch, J. Mater. Chem. A 3 (10) (2015) 5280e5284. H.S. Dang, P. Jannasch, Macromolecules 48 (16) (2015) 5742e5751. M.G. Marino, PhD thesis (2015), https://elib.uni-stuttgart.de/handle/11682/ 6896.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
A practical method for measuring the ion exchange capacity decrease of hydroxide exchange membranes during intrinsic degradation Klaus-Dieter Kreuer a, *, Patric Jannasch b a b
Max-Planck-Institute for Solid State Research, Heisenbergstrasse 1, D-70569 Stuttgart, Germany Department of Chemistry, Polymer & Materials Chemistry, Lund University, SE-22 100 Lund, Sweden
h i g h l i g h t s A new method for quantifying IEC decrease of hydroxide exchange membranes. Degradation rate of HEMs depends on temperature, hydration number and total water content. Method could help to resolve debates about relative durability of various HEMs.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 June 2017 Received in revised form 25 July 2017 Accepted 26 July 2017 Available online xxx
In this work we present a practical thermogravimetric method for quantifying the IEC (ion exchange capacity) decrease of hydroxide exchange membranes (HEMs) during intrinsic degradation mainly occurring through nucleophilic attack of the anion exchanging group by hydroxide ions. The method involves measuring weight changes under controlled temperature and relative humidity. These conditions are close to these in a fuel cell, i.e. the measured degradation rate includes all effects originating from the polymeric structure, the consumption of hydroxide ions and the release of water. In particular, this approach involves no added solvents or base, thereby avoiding inaccuracies that may arise in other methods due to the presence of solvents (other than water) or co-ions (such as Naþ or Kþ). We demonstrate the method by characterizing the decomposition of membranes consisting of poly(2,6dimethyl-1,4-phenylene oxide) functionalized with trimethyl-pentyl-ammonium side chains. The decomposition rate is found to depend on temperature, relative humidity RH (controlling the hydration number l) and the total water content (controlled by the actual IEC and RH). © 2017 Elsevier B.V. All rights reserved.
Keywords: Hydroxide exchange membrane AEM Degradation TGA Fuel cell
1. Introduction The potential application of anion exchange membranes (AEMs) in their hydroxide (OH) form (sometimes denoted by HEM) as separators in low temperature fuel cells is a matter of ongoing research. In contrast to acidic proton exchange membranes (PEM) that require the use of noble metal electro-catalysts to form a fuel cell electrode, the basic conditions under which HEMs operate allow less costly non-noble metal catalysts to be used [1]. OH conductivity values approaching the proton conductivity of PEMs (such as Nafion®) are common [2] for high levels of hydration, and bicarbonate (HCO 3 ) formation from the reaction of OH with CO2 present in air (used as oxidant in fuel cells) may be reduced by
* Corresponding author. E-mail address: [email protected] (K.-D. Kreuer).
using CO2 absorbers and adequate electrochemical management (self-purging [3]). The major problem stems from the mandatory presence of highly nucleophilic hydroxide ions as conducting ion. Quaternized ammonium (QA) groups are commonly used as positive countercharge within the polymeric structure; however, as typical leaving groups, QAs are well-known to react with OH through different energetically similar pathways [4] including nucleophilic substitution, b-elimination, and rearrangement reactions such as Stevens rearrangement in the absence of b-protons. As a consequence, HEMs inherently decompose, thereby losing ion exchange capacity (IEC) and ionic conductivity [5e10]. The degradation rate depends not only on the kind of QA but also on its environment within the polymeric structure. To deal with this complexity, we studied the degradation rates of a series of QA-salts in concentrated aqueous solutions of NaOH to identify
http://dx.doi.org/10.1016/j.jpowsour.2017.07.106 0378-7753/© 2017 Elsevier B.V. All rights reserved.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
2
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
suitable candidates for ionic groups in HEMs [11]. In this small molecule study, we also investigated the effects of different types of attachment, e.g. benzylic attachment is found to be extremely critical with respect to stability which is consistent with the results of another small-molecule study showing higher stability for alkylsubstituted compared to benzylic-substituted cations [12]. In the end, however, it is critical to test the durability of membranes under conditions similar to those in an operating fuel cell. These conditions differ from the conditions provided by aqueous solutions of NaOH (KOH) in various ways: i) within a HEM, OH counter ions are consumed in degradation reactions while the concentration (activity) of OH in excess NaOH solution is virtually unaffected by membrane degradation. ii) For high molarity, significant co-ion uptake (which corresponds to an uptake of excess NaOH) may affect the degradation rate through the presence of Naþ in the membrane. iii) The molar ratio [H2O]/[OH] in aqueous solutions of NaOH may be higher than for the low hydration conditions which may occur in running fuel cells. Especially the cathode side is expected to dry out as a result of electroosmotic water drag [13,14] from the cathode to the anode side at high ionic (OH) current density. Ion solvation (type of solvent and degree of solvation) actually affects degradation rates, and especially heavy hydration is known to strongly reduce the reactivity of the highly nucleophilic OH ion [11]. In a recent small-molecule NMR study, Dekel et al. [15] have quantified this effect in ex situ tests for hydration levels l ¼ 0 (dry) e 4 [H2O]/[OH] by adding defined amounts of water to dry solutions of QA and KOH in DMSO and following the degradation by 1H NMR spectroscopy. Here we present a gravimetric method which allows one to follow the IEC decay of HEMs at controlled temperature and hydration levels. The approach captures all effects that originate from the polymeric structure of the membrane, while avoiding possible inaccuracies arising from the unrealistic presence of solvents (other than water), co-ions (such as Kþ), or an excess of counter ions (OH). Moreover, the method is straight-forward and requires only a thermogravimetric analyzer [16]. 2. Basic idea As generally observed for PEMs, also the hydration number (l ¼ [H2O]/[ionic group]) for HEMs is approximately proportional to the IEC for a given temperature T and relative humidity RH (RH < 65% where hydration is exothermic). This approximately holds for different types of QA [data not yet published] and is especially true when a given type of ionomer with different concentration of a given type of QA is concerned. Therefore, also the difference of hydration numbers for two different values of RH (Dl) is a good measure for the actual IEC. In the proposed method, degradation occurs under controlled T and RH conditions while continuously recording the sample weight. This decreases with time for two reasons:
at that time. In this way, the evolution of IEC during degradation is recorded as a function of time. From this, information about reaction order and reaction rate may directly be deduced. Conceptually, the method works provided hydroxide ions merely react with QA functional groups (which is mostly the case). If hydroxide ions are also consumed in reactions with the polymer backbone, e.g. through nucleophilic attack of aryl-sulfones [17] or aryl-ethers [18], a decrease of Dl is expected as well because the reaction products (e.g. phenylates) are less hygroscopic than hydroxide ions. In order to exclude such reaction pathways, the IEC should be double checked at the end of the experiment by an independent method (see below). An IEC value higher than the one obtained from TGA would then be indicative for backbone degradation.
3. Experimental All experiments were carried out with a HEM consisting of a poly(2,6-dimethyl-1,4-phenylene oxide) (PPO) backbone functionalized with trimethyl-pentyl-ammonium side chains (Scheme 1). Since there is no benchmark membrane for AEMs (like in the case of PEMs for which Nafion is generally used for benchmarking) we have selected this homemade membrane because its chemical structure is related to the chemical structures of several AEMs for which reasonable stability in aqueous NaOH solutions have been reported [19e22]. The ionomer has been prepared as described earlier [23] and membranes were formed from NMP solution at T ¼ 80 C. The IEC of aseprepared membranes was determined in the Br form by displacing Br with NO 3 (~100 mg membrane in 40 ml 0.2 M NaNO3 for three days) and titrating the released Br with AgNO3 using a silver-electrode. Membranes in their Br form were converted into the OH form by ion exchange in 1 M NaOH (200 mg membrane in 40 ml solution) where the solution was replaced five times every 24 h under CO2-free conditions (in a homemade glove box [24]). This turned out to be necessary in order to reduce the Br content to less than 3% of the IEC. The residual Br content of OH exchanged membranes was actually determined by displacing Br with NO 3 after each exchanging step and subsequent Br titration (see above) using small pieces of the membrane. For the ion exchange procedure, one should keep in mind that ionic association in the NO 3 , Br and especially the I forms of AEMs leads to a severe stabilization compared the OH form [2]. The hydroxide exchanged membranes were rinsed with deionized water until neutrality and then kept in pure water not longer than a few days before carrying out the stability experiments. For this, a thermogravimetric balance with magnetic coupling is used (for details see ref. 13) membrane samples (100e200 mg) were placed in a quartz crucible within the glove box and then
i) volatile reaction products may be set free, ii) hydration water is released because of decreasing IEC (degradation of ionic QA groups). The first contribution depends on the kind of reaction products (alcohols, amines) and the polymeric host, degree of hydration and temperature determining the rate at which the degradation products are released. The second contribution, however, is mainly controlled by the decreasing IEC. Both contributions can easily be separated by intermittently measuring the hydration number for a higher RH. The increase of l (Dl) then is a measure of the actual IEC
Scheme 1. PPO grafted with trimethyl-pentyl-ammonium cations.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
3
transferred to the balance using a gas tight glass container. Nitrogen gas was passed through a CO2 absorber (sodalime with indicator, Merck No. 1.06733.0501) before entering the humidification system of the balance. In the present study, we have first recorded a hydration isotherm at T ¼ 40 C by equilibrating the membrane at RH ¼ 80, 65, 50, 30 and 10% for 10 h at each humidification level. Since some decomposition is observed for RH < 10% even at this low temperature, the weight of the dry sample could only be determined in the F form in a separate experiment (on the same ion exchanged membrane) by drying the membrane in a stream of dry nitrogen at T ¼ 100 C in the balance. The dependence of the hydration number in the OH form was measured for two RH values (50 and 65%) in the interval T ¼ 40e100 C (above this temperature, there was no reversibility as a consequence of decomposition). Stability tests were carried out at two temperatures (T ¼ 60 and 100 C) and two relative humidity values (RH ¼ 10 and 50%) in order to separate the dependence of the reaction rate on both parameters. In all experiments, the decomposition intervals were 10 20 h interrupted by 5 h intervals with higher RH (50 or 65%) for obtaining the transient IEC. Only for the most aggressive conditions (T ¼ 100 C, RH ¼ 10%) shorter decomposition intervals were chosen in order to increase the time resolution. After long term TGA (approx. 250 hrs.), the membranes were exchanged into the Cl form (in an excess of 1 M NaCl for five days at T ¼ 40 C) before determining the residual IEC in the same way as described before. This was done in order to cross-check the IEC decrease as determined by TGA. In order to compare the proposed methodology with the commonly used stability test in caustic aqueous solutions we have also immersed membrane pieces in 1 and 10 M aqueous KOH at T ¼ 100 C (approx. 200 mg membrane in 50 ml KOH solution in a Teflon crucible placed in a closed steel autoclave). Since we are interested in the evolution of the degradation with time, the autoclave was cooled and open after regular time intervals for taking out pieces of about 50 mg. These were rinsed with deionized water and then transferred into a 1 M NaCl solution for ion exchange followed by a determination of the residual IEC (see above). 4. Results and discussion Since the suggested method relies on defined relative weight changes as a consequence of degradation, the hydration isotherm for conditions under which HEMs are virtually stable (T ¼ 40 C, RH > 10%) is discussed first. In an earlier study we had already found that anion exchange membranes (AEM) may significantly swell in water for anions with high hydration enthalpy. This is the case for anions like OH and F in contrast to more hydrophobic anions such as Cl, Br and I, which tend to form ionic aggregates with the immobile quaternary ammonium cations [2]. For the present AEM with an IEC of close to 1.6 meq./g, even at room temperature, water uptake in pure water is very high (l ¼ 54 corresponding to a weight gain of 165% with respect to the dry membrane). Then, drying the membrane at a defined RH does not lead to a stable hydration level even after 10 h (Fig. 1a). However, subsequent rehydration leads to stable hydration numbers l for not too high (RH < 65%) and not too low RH > 10% (Fig. 1b). For the membrane used in this study, even for a temperature as low as T ¼ 40 C, some degradation is indicated by a slow weight loss under very dry conditions (RH ¼ 10%, Fig. 1a). As in the case of acidic membranes, l depends on temperature T. Since the ion exchanging group (tetra-methyl-ammonium hydroxide) is only a moderately strong base, its affinity toward water decreases with T. This is clearly visible for l (T, RH) which decreases from l ¼ 3.80 at T ¼ 40 C to l ¼ 2.92 at T ¼ 100 C for RH ¼ 50% (Fig. 2).
Fig. 1. Membrane hydration at T ¼ 40 C: a) TGA trace for step wise drying (RH ¼ 65, 50, 30, 10%) and rehydration (RH ¼ 30, 50, 65%). Note, that the weight loss at RH ¼ 10% indicate the onset of decomposition. b) Hydration isotherm compared to the hydration isotherm of Nafion [13].
For higher RH, l is getting increasingly dependent on the membrane's pre-treatment, for higher T there is significant decomposition even at high RH. Prior to the actual stability test, it is important to ensure a stable hydration number at the respective temperature at an RH where decomposition is slow. This hydration number is then considered to be the hydration number l of the membrane with its initial IEC . Depending on T and RH under which decomposition is recorded, we have chosen RH ¼ 50 or 65% for probing the transient IEC. Fig. 3 shows the TGA traces for T ¼ 60 and 100 C. While the weight was virtually stable during the probing regimes (RH ¼ 50, 65%), decomposition is indicated by both the decreasing difference in l for the two RH levels and the weight decrease during the low RH regimes (decomposition regime). Only for the most aggressive conditions (T ¼ 100 C, RH ¼ 10%), very fast initial decomposition is observed before the nominal decomposition RH (10%) is attained. The transient IEC(t) as obtained from: IEC(t) ¼ (Dl(t)/Dl ) IEC
(1)
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
4
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
is shown in Fig. 4. The fact that the IEC values obtained by titration after the degradation tests are virtually identical to the last data point (at about T ¼ 200 h, Fig. 4) gives confidence that the proposed procedure provides reliable data for IEC(t). In particular, the IEC obtained by titration is never higher than this deduced from the TGA data suggesting relative backbone durability under the given conditions. Only for the most severe degradation conditions (T ¼ 100 C/RH ¼ 10%, see above), membranes did not show any ion exchange activity after the test, although thermogravimetric analysis indicates some residual hygroscopicity (Fig. 3). It is not surprising that the observed degradation rates do not follow first order reaction kinetics: IEC(t) ¼ IEC exp (-k t)
Fig. 2. Membrane hydration as a function of temperature for RH ¼ 50% and 65%. The corresponding data for Nafion [13] are given for comparison.
(2)
as observed in small molecule degradation studies in aqueous solutions with a large excess of NaOH [10]. In the present case, both QA cations and OH counter ions are consumed in the degradation reaction which may suggest second order reaction kinetics: 1/IEC(t) ¼ 1/IEC þ k t
(3)
But this is not confirmed for any degradation conditions chosen in this study. For these, the degradation rate seems to decrease with increasing degree of degradation, i.e. with decreasing water content. In fact, there is indication that the decreasing amount of water decreases the global dynamics of the system and as a consequence also the rate of degradation (see below). Another part of the complexity of the reaction kinetics is probably the fact that degradation products are more or less released depending on T and RH. If all reaction products were retained in the membrane, the weight loss at a given RH would simply correspond to the reduced water uptake as a consequence of IEC decrease. Then, IEC(t) is: IEC(t) ¼ (l(t)/l ) IEC
(4)
In fact, this is approximately the case for low T and high RH. Except for T ¼ 60 C and RH ¼ 50%, however, the total weight loss is about 2e3 times higher than expected for this case (Fig. 3). In other words, about 36e54 g reaction products per mole ionic groups are released. This is also reflected in the negative values for l in Fig. 3 (note that for this representation, l is calculated assuming a
Fig. 3. TGA traces for T ¼ 60 C and 100 C. Relative humidity RH values are given for the degradation regimes and the probing regimes (in brackets). The weight changes are expressed in terms of hydration number l assuming a constant IEC ¼ IEC (see text).
Fig. 4. IEC decay for different degradation conditions as obtained from TGA traces (Fig. 3). IEC data obtained by titration after the experiments are also included.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
constant ion exchange capacity IEC ¼ IEC ). This finding underlines the need to periodically apply higher RH for probing the actual IEC(t) from Dl for two RH values (1) and suggests that sluggish release of reaction products may contribute to the complexity of the degradation kinetics. Other reasons may be competing reaction pathways or more than one OH may be involved in the degradation of one ionic group. The latter may explain the observation that even under the most severe degradation conditions (T ¼ 100 C, RH ¼ 10%), there is still some IEC left after more than 100 h (Fig. 4). In any case, degradation cannot be described by a single degradation rate k (any order), and this is what is also expected for membrane degradation in a running fuel cell. In all cases the degradation rate seems to decrease with proceeding degradation. It is not the aim of the present work to identify details of the reaction scheme on a molecular level. This is expected to be different for different types of membrane. But the degradation of all HEMs goes along with a decrease in IEC, and therefore, IEC(t) for defined T/RH conditions is a meaningful parameter (function) for comparing the durability of different types of membrane.
5. Comparison to other durability tests and some practical considerations We assume that the present methodology provides the most realistic and precise way to obtain rates for the intrinsic degradation reactions in HEMs. The T/RH conditions can be chosen close to those occurring in a running fuel cell, and the obtained ion exchange capacity transients (IEC(t)) appear to well reproduce data obtained by titration. It goes without saying that other reactions, e.g. with peroxyl and hydroxyl radicals, which may also take place in a fuel cell, are not captured by this method. Compared to tests in aqueous solutions of KOH (NaOH), no additional cations, anions, and water are present in the membrane. Especially at high molarity, Donnan exclusion is less efficient, i.e. significant amounts of KOH may enter the membranes from solution, the water uptake being mainly determined by the membrane's viscoelastic properties [16]. Naively thinking, one might choose a 10 M KOH for a stability test because for this the ratio [H2O]/[KOH] is about 4.7 which corresponds to a similar value for l at RH ¼ 60%. But the comparison of experimental results for a test in 10 M KOH and at RH ¼ 50% (l ~ 3.7) both at T ¼ 100 C (Fig. 5) show that
5
degradation in solution is faster. Part of this difference is surely the large excess of hydroxide ions which may compensate for OH consumed in the reaction, but the fact that the initial IEC decay is significantly stronger in solution also suggests a difference in the rates of the underlying elementary reactions. In solution, the total uptake of water and ions is higher although the water to ion ratio (l) is similar, and this may increase the dynamics in general and therefore also the rate of degradation. For degradation tests in solution of low KOH molarity, KOH uptake is surely lower, but such tests do not provide a defined degree of hydration l. E.g. for a 1 M KOH, the molar water to base ratio is about 54 which is usually much higher than the hydration number l attained by the membrane immersed in this solution (again dependent on the membrane's viscoelastic properties). But since the degradation rate strongly depends on water content (hydration number), KOH molarity alone does not guarantee defined degradation conditions. In the present case, degradation in 1 M KOH is again surprisingly fast (Fig. 5) considering the high molar ratio of water to ion in solution. Obviously the hydration number is lower in the membrane, and the degradation rate is expected to be even faster for membranes with the same ion exchanging group but a tougher backbone more effectively limiting swelling (hydration number l). Since fixing T/RH (RH < 65%) allows setting l in a very narrow range, the method presented here is expected to produce more defined results. It should be noted that the activity (chemical potential) of the mobile species are also defined for membranes in contact with solutions, but not the number of water molecules per ion. This is because of the membrane's internal pressure is another degree of freedom entering into the species chemical potential [16]. The present results actually suggest that the rate of degradation is more dependent on the hydration number l than on the chemical potential of water. The key finding of the degradation test of trimethyl-benzylammonium salts recently published by Dekel et al. [15] is the extent to which the degradation rate is affected by lowering hydration at room temperature. Comparing their data for l ¼ 1 and 4 (Fig. 3 in Ref. [15]) with the data we have obtained at T ¼ 60 C with RH ¼ 10% and 50% (corresponding to l ¼ 1.15 and 3.75), confirms this effect in an almost quantitative manner. However, the methodology presented here allows recording the degradation rate of membranes (as opposed to small molecules/ions) and easily reveals its temperature dependence. For the type of membrane considered here, increasing T from 60 to 100 C increases the degradation at RH ¼ 50% more than decreasing RH from 50% to 10% does (Fig. 4). It seems that both hydration level (RH, l) and T are the major parameters controlling inherent membrane degradation. Ideally, the test suggested here may be made by a TGA system which allows to control T and RH in the range T ¼ 40e100 C and RH ¼ 10e65%. The conditions for which data are reported here (Fig. 3) may be used as guide line for choosing appropriate values for T and RH. But even TGA systems which allow to set low RH at relatively low T only may provide useful results since for many membranes significant degradation rates may occur e.g. at T ¼ 60 C and RH ¼ 10% (Fig. 4). These conditions may be accessible by virtually any TGA device. Acknowledgement
Fig. 5. Membrane degradation (IEC decay) as determined by TGA and by treatments in caustic aqueous solutions. Note that the ratio [H2O]/[OH] is about 54 for 1 M NaOH and 4.7 for 10 M KOH.
The authors would like to thank A. Fuchs and U. Klock (MaxPlanck-Institute for Solid State Research) and H.-S. Dang (Department of Chemistry, Lund University) for technical assistance. We are also very grateful to R. Usiskin and T. Saatkamp (Max-PlanckInstitute for Solid State Research) for reading the proofs. We also would like to thank Alexey Serov (University of New Mexico) for giving us the chance to publish in this Special Issue.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106
6
K.-D. Kreuer, P. Jannasch / Journal of Power Sources xxx (2017) 1e6
References [1] J.R. Varcoe, R.C.T. Slade, Fuel Cells 5 (2005) 187e200. [2] M.G. Marino, J.P. Melchior, A. Wohlfarth, K.D. Kreuer, J. Membr. Sci. 464 (2014) 61e71. [3] M. Inaba, Y. Matsui, M. Saito, A. Tasaka, K. Fukuta, S. Watanabe, H. Yanagi, Electrochemistry 79 (2011) 322e325. [4] S. Chempath, J.M. Boncella, L.R. Pratt, N. Henson, B.S. Pivovar, J. Phys. Chem. C 114 (2010) 11977e11983. [5] Y. Ye, Y. A. Elabd, Chemical Stability of Anion Exchange Membranes for Fuel Cells, Chap. 15, pp. 233 e251. URL http://pubs.acs.org/doi/abs/10.1021/bk2012-1096.ch014. [6] T. Sata, M. Tsujimoto, T. Yamaguchi, K. Matsusaki, J. Membr. Sci. 112 (1996) 161e170. [7] G. Couture, A. Alaaeddine, F. Boschet, B. Ameduri, Prog. Polym. Sci. 36 (2011) 1521e1557. [8] G. Merle, M. Wessling, K. Nijmeijer, J. Membr. Sci. 377 (2011) 1e35. [9] C.G. Arges, J. Parrondo, G. Johnson, A. Nadhan, V. Ramani, J. Mater.Chem. 22 (2012) 3733e3744. [10] D. Chen, M.A. Hickner, ACS Appl. Mater. Interfaces 4 (2012) 5775e5781.
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
M.G. Marino, K.D. Kreuer, ChemSusChem 8 (3) (2015) 513e523. A.D. Mohanty, C. Bae, J. Mater. Chem. A 2 (2014) 17314e17320. Y.S. Li, T.S. Zhao, W.W. Yang, Int. J. Hydrogen Energy 35 (2010) 5656e5665. X. Wang, J.P. McClure, P.S. Fedkiw, Electrochim. Acta 79 (2012) 126e132. D.R. Dekel, M. Amar, S. Willdorf, M. Kosa, S. Dhara, C.E. Diesendruck, Chem. Mater. 29 (10) (2017) 4425e4431. K.D. Kreuer, Solid State Ionics 252 (2013) 93e101. E.A. Weiber, P. Jannasch, J. Membr. Sci. 481 (2015) 164e171. C. Fujimoto, D.-S. Kim, M. Hibbs, D. Wrobleski, Y.S. Kim, J. Membr. Sci. 423e424 (2012) 438e449. S.A. Nunez, C. Capparelli, M.A. Hickner, Chem. Mater. 28 (8) (2016) 2589e2598. W.H. Lee, A.D. Mohanty, C. Bae, ACS Macro Lett. 4 (4) (2015) 453e457. Z.J. Yang, J.H. Zhou, S.W. Wang, J.Q. Hou, L. Wu, T.W. Xu, J. Mater. Chem. A 3 (29) (2015) 15015e15019. H.S. Dang, E.A. Weiber, P. Jannasch, J. Mater. Chem. A 3 (10) (2015) 5280e5284. H.S. Dang, P. Jannasch, Macromolecules 48 (16) (2015) 5742e5751. M.G. Marino, PhD thesis (2015), https://elib.uni-stuttgart.de/handle/11682/ 6896.
Please cite this article in press as: K.-D. Kreuer, P. Jannasch, Journal of Power Sources (2017), http://dx.doi.org/10.1016/j.jpowsour.2017.07.106