Mark
Ingratta
a,
Matti
Elomaa
b and
Patric
Jannasch
*a
aDepartment of Chemistry, Polymer and Materials Chemistry, Lund University, P.O.B. 124, SE 221 00, Lund, Sweden. E-mail: patric.jannasch@polymat.lth.se
bLaboratory of Polymer Chemistry, University of Helsinki, P B 55, 00014, Finland
First published on 12th March 2010
Densely phosphonated electrolyte membranes were prepared from poly(2,6-dimethyl-1,4-phenylene oxide) (PPO) grafted with poly(vinylphosphonic acid) (PVPA) side chains. In the first step, PPO was lithiated in solution at room temperature by adding n-butyllithium to form an anionic macroinitiatior. Next, diethyl vinylphosphonate was anionically polymerized from the lithiated sites at −78 °C. This protocol gave good control over the density of the grafting sites and the copolymer composition. Films of copolymers containing between 35 and 74 wt% poly(diethyl vinylphosphonate) were first cast from solution and subsequently fully hydrolyzed to produce transparent flexible proton conducting membranes of PPO-graft-PVPA containing up to 6 mmol phosphonic acid groups per gram of dry copolymer. Thermogravimetric analysis showed anhydride formation at increasing temperatures above 100 °C with no copolymer degradation occurring until nearly 400 °C under air. Fully hydrated membranes reached proton conductivities above 1 mS cm−1 at −20 °C and 80 mS cm−1 at 120 °C.
The current state-of-the-art membrane Nafion® is a sulfonated fluoropolymer, which exhibits high stability and high proton conductivity within its optimal operating temperature window between 0 and 80 °C. While Nafion® has attracted strong interest as the standard for proton conductivity, it still has several drawbacks. For example, it has a relatively high price, and its conductivity drops drastically once outside its optimal temperature and humidity range. This drop is due to Nafion’s® dependence on the presence of water for proton transport within the hydrophilic channels in the membrane, and is caused by water freezing at low temperatures and evaporation at high temperatures.
One proposed strategy for conductive polymers with an increased operating temperature window at low relative humidity (RH) is to substitute the sulfonic acid groups for an intrinsically proton conducting functionality.10 In this way, water is not required for proton transport across the membrane. Polymers functionalized with phosphonic acid11–15 and various imidazoles16–21 have been investigated for this purpose, and the results have shown that the concentration and mode of immobilization of these groups are very important for the proton conductivity. We have previously studied poly(vinylphosphonic acid) (PVPA) in graft and block copolymers, in combination with poly(arylene ether sulfone) (PSU)22 and polystyrene (PS),23 respectively. Both of these copolymers showed relatively high proton conductivity in the range of 1–5 and 10–100 mS cm−1 in the nominally dry and hydrated state, respectively. Several recent investigations using 1H MAS (magic angle spinning) NMR and 31P NMR spectroscopy have been conducted to determine the effect of humidity and thermal history on the proton conducting properties of homopolymers of PVPA.24,25 It was shown that RHs above 40%, equivalent to 15 wt% water, or 0.8 water molecules per acid group, were required to minimize anhydride formation between phosphonic acid groups. Temperatures greater than 150 °C also favored anhydride formation, which decreased the proton conductivity since increasing cross-linking lowered the acid concentration and the chain mobility, and hence also the proton mobility. Several strategies have evolved in an attempt to reduce anhydride formation, including copolymers with both phosphonic acid and basic functions,26–28 blends29 containing PVPA and basic polymers, and the use of block copolymers30 containing poly(vinylbenzylphosphonic acid) and PS or polyetheretherketone (PEEK).
PPO is a versatile aromatic polymer, which can advantageously be used as a precursor in the preparation of graft and block copolymers.31 For example, chain end hydroxyl groups can be exploited for polycondensation reactions and have also been used to introduce initiator sites for the atom transfer radical polymerization (ATRP) of styrene to form PPO-block-PS.32 Alternatively, the aromatic ring and benzylic functions in the PPO main chain can be employed in several functionalization reactions, such as electrophilic substitution reactions, chlorination, and sulfonation, to name a few.31 PPO has also been used recently to synthesize PPO-graft-PS,33 PPO-graft-poly(methyl methacrylate),33 and PPO-graft-poly(styrene-sulfonic acid)34 by first brominating PPO and then using ATRP to grow the side chains.
In the current work, PPO was successfully grafted with diethyl vinylphosphonate (DEVP) in a single step via anionic polymerization to give PPO-graft-poly(diethyl vinylphosphonate) (PPOgPDEVP). n-Butyllithium was used to introduce anionic initiator sites along the PPO backbone,35,36 followed by the addition of DEVP. Films of PPOgPDEVP were cast and then hydrolyzed to form transparent, flexible PPOgPVPA membranes. The influence of the copolymer structure on the morphology and key membrane properties, such as thermal stability, water uptake and proton conductivity were studied.
Copolymer | Target DLa | Polymerization time/min | TMCS added | TMS in aryl position on PPOb | TMS on PDEVP chain endc | PDEVP contentd/wt% | Degree of polymerizatione |
---|---|---|---|---|---|---|---|
a DL = degree of lithiation, the average number of lithiated sites per PPO monomer unit. b Average number of TMS groups attached at the aryl position per PPO monomer unit (n.d. = not detected). c Average number of TMS groups attached at the PDEVP chain end per PPO monomer unit. d Determined by 1H NMR data. e Calculated using the PDEVP content and the number of TMS groups on the PDEVP chain ends as determined by 1H NMR data (n.c. = not calculated). | |||||||
PPOgPDEVP1 | 0.15 | 120 | y | 0.023 | 0.037 | 38 | 12 |
PPOgPDEVP2 | 0.15 | 120 | y | 0.024 | 0.040 | 44 | 15 |
PPOgPDEVP3 | 0.15 | 120 | n | — | — | 54 | n.c. |
PPOgPDEVP4 | 0.15 | 20 | y | 0.052 | 0.035 | 59 | 31 |
PPOgPDEVP5 | 0.15 | 120 | y | 0.028 | 0.047 | 60 | 23 |
PPOgPDEVP6 | 0.15 | 120 | n | — | — | 74 | n.c. |
PPOgPDEVP7 | 0.075 | 20 | y | 0.01 | 0.018 | 35 | 22 |
PPOgPDEVP8 | 0.075 | 20 | y | 0.01 | 0.018 | 44 | 31 |
PPOgPDEVP9 | 0.075 | 120 | y | n.d. | 0.016 | 73 | 124 |
1H NMR analysis was completed using a Bruker 400 MHz spectrometer. Spectra of the PPOgPDEVP samples were recorded at 400.13 MHz in deuterated chloroform. NMR analysis was not performed on the PPOgPVPA samples due to their insolubility. Representative 1H NMR spectra of PPO, PPO reacted with TMCS and PPOgPDEVP8 are shown in Fig. 1a–c, respectively. The weight percent of PDEVP was determined by integrating the aromatic protons of the PPO backbone at 6.47 ppm relative to those of the CH2 of the ester group of the PDEVP unit at 4.10 ppm. The peaks from the methyl protons of the trimethylsilyl (TMS) groups were used to quantify the number and distribution of active lithiated sites at the end of the grafting reaction, and appeared at 0.45, 0.17 and −0.069 ppm for Ar–Si(CH3)3, PDEVP–Si(CH3)3, and ArCH2–Si(CH3)3 respectively.
Fig. 1 1H NMR spectra of (a) neat PPO, (b) PPO substituted with TMS groups, and (c) PPOgPDEVP8. All of the spectra were collected using CDCl3 solutions. |
Hydrolysis of the membranes to obtain the PPOgPVPA membrane was achieved using concentrated aqueous HCl. The dry PPOgPDEVP membranes were placed in a round bottomed flask equipped with a reflux condenser and charged with 50 mL aqueous HCl. The solution was refluxed for 8 h, after which the acidic membranes were leached with Milli-Q water at least 5 times to remove traces of HCl. The membranes were finally dried in a vacuum oven for 24 h.
The water uptake (wwater) was determined according to eqn (1) for both PPOgPDEVP and PPOgPVPA membranes. Dry weights (wdry) were determined after drying the membranes in a vacuum oven at 50 °C for 48 h while wet weights (wwet) were determined by immersing the membrane in Milli-Q water for 24 h. The wet membranes were weighed after excess water was quickly removed using tissue paper.
wwater = [(wwet − wdry)/wdry] × 100% | (1) |
Differential scanning calorimetry (DSC) was used to determine the glass transition temperatures (Tgs) for all PPOgPDEVP and PPOgPVPA membranes and was performed using a TA Instruments Q1000 DSC. Each PPOgPVPA and PPOgPDEVP sample was examined using hermetically sealed pans with the following sequence of scans: 40 → 150 → 50 → 200 → 50 → 250 → 50 → 300 °C. The heating and cooling rates were 10 °C min−1 and Tgs were determined from the last heating scan.
Copolymer | Theor. IEC/mmol g−1 | PVPA content (wt%) | w water (wt%) | [H2O]/[–PO3H2] | T g/°C |
---|---|---|---|---|---|
PPOgPVPA1 | 2.5 | 28 | 16 | 3.5 | 231 |
PPOgPVPA2 | 3.2 | 34 | 23 | 4.1 | 234 |
PPOgPVPA3 | 4.0 | 43 | 32 | 4.5 | 255 |
PPOgPVPA4 | 4.5 | 49 | 43 | 5.3 | 249 |
PPOgPVPA5 | 4.6 | 49 | 44 | 5.4 | 251 |
PPOgPVPA6 | 6.0 | 65 | 111 | 11.1 | 253 |
PPOgPVPA7 | 2.4 | 26 | 19 | 4.3 | 245 |
PPOgPVPA8 | 3.1 | 34 | 24 | 4.2 | 224 |
PPOgPVPA9 | 6.0 | 65 | 152 | 14.3 | 257 |
An anionic polymerization technique was selected in order to develop a straightforward one pot synthesis where PPO can be lithiated to provide initiating sites for the subsequent DEVP polymerizations. This procedure may be compared to alternative polymerization methods involving techniques, such as ATRP, where a “macroinitiator” must first be synthesized and purified prior to the polymerization step. We have previously polymerized DEVP anionically from lithiated PSU,22 and from polystyryl anions in sequential polymerizations23 to form graft and block copolymers, respectively. One significant difference in the synthesis procedure used in the current work was that PPO was much less soluble in THF than PSU or PS, thus requiring a more carefully controlled lithiation procedure to avoid polymer precipitation. The mixture of PPO in THF was first warmed to 60 °C under stirring to fully dissolve the PPO. n-Butyllithium was then added at room temperature to lithiate the PPO before cooling to −78 °C. Non-lithiated PPO would precipitate out of solution upon reaching approx. −30 °C. However, lithiated PPO is much more soluble in THF up to a degree of lithiation (DL, the average number of lithiated sites per repeat unit of PPO) of approximately 0.5, at which point the polymer begins to gel out of solution. Also, in both previous cases where anionic polymerization of DEVP was performed, the polymerization was initiated by adding 1,1-diphenylethylene (DPE) to modify the reactivity of the anionic sites. In the present work, no additions of DPE were necessary, since the benzyllithium formed on PPO was found to be an efficient initiator site for DEVP at −78 °C (Scheme 1).
Scheme 1 Grafting PDEVP from PPO by anionic polymerization followed by acidic hydrolysis of the grafts to PVPA. |
In an initial study, 1H NMR analysis was used to estimate the number and positions of the titrated sites initiated by butyllithium, as done previously by Chalk et al.35,36 Two PPO samples were lithiated followed by the addition of TMCS. As shown in Fig. 1b, two new peaks were present for the Ar–H and Si–CH3 protons, respectively, at 5.98 and 0.45 ppm when TMS was attached at the aryl position, and at 6.22 and −0.069 ppm with TMS attached at the benzyl position. Two different levels of DL were targeted, 0.15 and 0.075. Using the peak intensities at 0.45 and −0.069 ppm in the NMR spectra, it was calculated that on average 0.12 and 0.05 lithiated sites per PPO repeat unit reacted with TMCS when targeting DLs of 0.15 and 0.075, respectively, in approximately a 65:35 ratio for the aryl vs. benzyl position. These values are in agreement with previous findings.35,36 In addition, a PDEVP homopolymer end-capped with TMS was synthesized using 1,1-diphenylhexyllithium.23 The chemical shift for TMS at the PDEVP chain end was confirmed at 0.17 ppm.
Most of the graft polymerizations were terminated by the addition of TMCS. This provided a way to measure the number of active PDEVP chain ends and any remaining lithiated sites along the PPO chain by NMR spectroscopy. Fig. 1c shows the 1H NMR spectrum of sample PPOgPDEVP8, which indicates the presence of TMS in aryl positions and at PDEVP chain ends. Table 1 contains the data on the average number of TMS groups found in these positions per PPO repeat unit in the respective graft copolymer. Importantly, no peak was found for TMS attached at the benzyl position of PPO in any of the samples, which strongly suggested that this was the initiation site for the graft polymerization and that it was thus completely consumed. The degree of polymerization of the PDEVP grafts was calculated by the use of the average number of TMS groups attached at the PDEVP chain ends. As seen in Table 1, this number was found to be between 20 and 30% of the DL. The number of active sites remaining on PPO main chain and the PDEVP side chain ends was found to be lower than when only TMCS was added to lithiated PPO, which indicated that a portion of the lithiated sites was deactivated by the addition of DEVP. Increasing the reaction time from 20 min to 120 min for DEVP had no effect on the monomer conversion or the number of active PDEVP chain ends, while a slight decrease in the number of lithiated aryl positions was noted (Table 1, samples PPOgPDEVP4 and -5). This indicated that the majority of the deactivation reactions occurred immediately upon the addition of DEVP and were likely due to traces of impurities in the feed monomer. The number of active chain ends for PDEVP was used to estimate the average graft chain length in combination with the PDEVP content determined from the integration of the characteristic –O–CH2–CH3 signals at 4.1 ppm in the 1H NMR spectra (Fig. 1c, Table 1).
Fig. 2 depicts partial FTIR spectra of pristine PPO and representative samples of the graft copolymers in the ester and acid form. As can be seen, the spectra of the copolymers in the ester form contained several new peaks, consistent with that expected after grafting with PDEVP. For example, there was a band for the (PO) stretch at 1230 cm−1, and bands clearly visible at 1052, 1023 and 776 cm−1 for absorptions from the (P)–O–C ester linkages. Also, there was the P–O–(C) vibration at 946 cm−1. After hydrolysis of the ester groups, these characteristic bands disappeared and new bands assigned to (P)–O–H were present at 990 and 2300 cm−1, respectively.
Fig. 2 FTIR spectra of (a) PPO, (b) PPOgPDEVP6 and (c) PPOgPVPA6. |
Fig. 3 The TGA traces of (a) PPOgPDEVP2 and PPOgPVPA2 recorded at 10 °C min−1 under nitrogen and (b) the TGA traces of several PPOgPVPA samples with PVPA contents ranging from 28 to 65 wt%, recorded at 1 °C min−1 under air. |
Fig. 3b shows the degradation of the PPOgPVPA membranes under air with PVPA contents ranging from 28 to 65 wt%. The slow weight loss beginning at ∼150 °C has already been discussed and was attributed to the reversible condensation of the phosphonic acid groups. Actually, a small weight increase was noted at approximately 250 °C, which may be due to oxygen uptake in connection with the formation of phosphates. The major degradation occurred above 350 °C for all membranes and was ascribed to the degradation of PVPA, which has been shown to occur at approximately 330 and 380 °C under atmospheric and inert conditions, respectively.25 Also, the magnitude of the weight loss above 400 °C decreases with increasing PVPA content. This is likely due to the flame-retardant mechanism of phosphoric acid.39
DSC was used to measure the Tgs of the copolymers. Graft and block copolymers with a distinct phase separation of the blocks typically show two distinct glass transitions.40 Block copolymers of PS and PVPA, as well as PSU grafted with PVPA, showed two distinct glass transitions, with the Tg from PVPA increasing with an increase in the annealing temperature.22,23 In the current work, both the PPOgPDEVP and PPOgPVPA membranes were studied. In the case of PPOgPDEVP, only a single Tg was detected at 166–186 °C, and only for the four samples with the lowest PDEVP content. This transition most probably originated from a PPO-rich phase, with the Tg significantly depressed in comparison with that of the unmodified PPO at 218 °C. This finding may indicate a partial miscibility between the PPO and the PDEVP.
The PPOgPVPA samples showed only a single glass transition with the Tg increasing with increasing temperatures during the thermal cycling in the DSC, as shown in Fig. 4 and summarised in Table 2. This indicated that the transition originated from the PVPA phase, where successively increasing temperatures during the cycling led to increasing formation and concentration of phosphonic acid anhydride bridges, which resulted in increased crosslinking and Tgs. Any glass transition from a PPO-rich phase was thus not detected for the PPOgPVPA copolymers. The use of dynamic mechanical analysis of the PPOgPDEVP and PPOgPVPA membranes would most probably increase the chance of observing two glass transitions.
Fig. 4 DSC heating traces of PPOgPVPA2 from 40 → 150 → 50 → 200 → 50 → 250 → 50 → 300 °C at 10 °C min−1. |
Fig. 5 AFM tapping mode topography (lower) and phase (upper) images of thin films of PPOgPDEVP8 (a) and -9 (b). |
Fig. 6 The water uptake of PPOgPDEVP and PPOgPVPA membranes as a function of the IEC, i.e., mmoles phosphonate per gram of dry polymer (X denotes –H and –CH2CH3 for PPOgPVPA and PPOgPDEVP, respectively; the lines are only shown to guide the eye). |
Also of note is the increasing water uptake with a decreasing DL for copolymers with similar, high IEC values. The water uptake of PPOgPVPA9 (DL = 0.075) was 50% higher than for the more densely grafted PPOgPVPA6 (DL = 0.15), although both copolymers had an IEC of 6.0 mmol g−1. At a constant IEC, a decrease in DL will increase the chain length and lead to an increased size of PVPA domains. The effect of the DL was, however, not observed for membranes with IEC values below 4.8 mmol g−1, where the water uptake was moderate with increasing IEC values.
The proton conductivity of the PPOgPVPA membranes fully immersed in water was measured from −20 up to 120 °C, and the results are plotted in Fig. 7. The corresponding data for Nafion® 117 was included for reference. As can be seen, the conductivity increased sharply between −20 and 0 °C because of the melting of the absorbed water and reached high values at higher temperatures. The effect of the melting water was most pronounced for the Nafion® 117 membrane. As expected, the PVPA content and the subsequent water uptake was the dominant factor in influencing the membrane conductivity, with a 40 fold increase from 2 to 80 mS cm−1 at 120 °C when increasing the IEC from 2.5 to 6.0 mol g−1. The effect of the PVPA content was most pronounced at low IEC values between 2.5 and 3.0 mmol g−1, increasing 10 fold, but there was still a strong increase from 3.0 to 6.0 mmol g−1, as shown in Fig. 8. On the other hand, the DL and PVPA chain length appeared to have had little influence on the conductivity, as all the membranes fell on the same trend line regardless of the DL. This is in contrast with the previously noted water uptake for the two copolymers with an IEC of 6.0 and is likely due to a dilution of the phosphonic acid moieties, which disfavors the membrane with the highest water uptake. The conductivity of 80 mS cm−1 for the highest PVPA contents at 120 °C is very close to the value of 93 mS cm−1 achieved for PSUgPVPA under the same conditions.22 Thus, the similar water uptake characteristics and conductivities of the two copolymers implied that their morphologies were equally efficient for proton transport.
Fig. 7 Conductivity data for the PPOgPVPA membranes measured by EIS under immersed conditions in a sealed cell from −20 to 120 °C. The corresponding data for Nafion® 117 has been added for reference. |
This journal is © The Royal Society of Chemistry 2010 |