PBI‐type Polymers and Acidic Proton Conducting Ionic Liquids – Conductivity and Molecular Interactions

Proton conducting ionic liquids (PILs) are discussed as new electrolytes for the use as non‐aqueous electrolytes at operation temperatures above 100 °C. During fuel cell operation the presence of significant amounts of residual water is unavoidable. The highly Brønsted‐acidic PIL 2‐Sulfoethylmethylammonum triflate [2‐Sema][TfO] is able to perform fast proton exchange processes with H2O, resulting from 1H‐NMR and pulsed field gradient (PFG)/diffusion ordered spectroscopy (DOSY) self‐diffusion measurements. Proton conduction takes place by a vehicle mechanism via PIL cations or H3O+, but also by a cooperative mechanism involving both species. Thus, highly Brønsted‐acidic PILs are promising candidates for the use as non‐aqueous electrolytes. To use [2‐Sema][TfO] as electrolyte in a proton electrolyte fuel cell (PEFC) it has to be immobilized in a host polymer. There is a (slow) uptake of the PIL by polybenzimidazole (PBI) up to a weight increase of ∼130%, due to a swelling process. A protonation of the basic imidazole moieties takes place. NMR analysis was applied to elucidate the molecular interactions between PBI, PIL, and residual water. Proton exchange, respectively an interaction between the polar groups and water can be observed in spectra, indicating a network of H‐bonds in doped PBI. Therefore, highly acidic PILs are promising candidates for the use as non‐aqueous electrolytes.


Introduction
Polymer electrolyte fuel cells (PEFCs), operational at an elevated temperature above 100°C, have attracted much attention recently, due to their superiorities compare to low temperature (LT)-PEFCs: (i) no feed gas humidification, (ii) a more efficient cooling system (easier water and heat management), (iii) the possibility of recovering high-grade waste heat, and (iv) a higher tolerance against feed gas impurities [1,2]. Currently, (high temperature) HT-PEFC, based on phosphoric acid doped polybenzimidazole (PBI) membranes, cannot compete with the performance characteristics of NAFION-based LT-PEFCs [3]. Despite the high operation temperature of 160°C, the presence of H 3 PO 4 causes a slow cathodic oxygen reduction reaction kinetics (ORR). This is primarily caused by an inhibition effect due to a poisoning by adsorption of H 3 PO 4 species onto active catalyst sites [4]. Also discussed is a low solubility of O 2 and a slow diffusion of O 2 through the H 3 PO 4 film which covers the platinum catalyst [5]. Thus, there is a necessity for new non-aqueous proton conducting electrolytes operational for the temperature range between 100-120°C.
PBI is widely used in HT-PEFCs as a proton-conducting electrolyte matrix, because of its high decomposition temperature and excellent thermal and chemical stability. The basic imidazole moieties of the PBI chains are protonated by H 3 PO 4 , the H 2 PO 4 anions interact with the polymer chains by strong coulomb forces and the additional H 3 PO 4 molecules by H-bonds [6]. Mobile charge carriers for protons are formed by a strong autoprotolysis, i.e., H 4 PO 4 + and H 2 PO 4 -, hence providing a high proton conductivity at low water concentrations.
Proton conducting ionic liquids (PILs) with acidic cations are promising candidates for the use as non-aqueous electrolytes at operation temperatures above 100°C and got much attention for applications in PEFCs [7]. PILs are ionic compounds with bulky cations and anions [8]. PILs with anions based on very strong acid respectively super acids as trifluoromethanesulfonic acid or bis-trifluoromethylsulfonimid have a less inhibiting effect than H 3 PO 4 because these anions, e.g., triflate CF 3 SO 3 or triflimid (CF 3 SO 2 ) 2 N -, are less strongly adsorbed on Pt than H 3 PO 4 or H 2 PO 4 -. In a neat (i.e., water-free) PIL, proton transport can inevitably only take place by a vehicle mechanism. However, PILs are normally hygroscopic. During fuel cell operation, also above 100°C, a water uptake will unavoidably take place. On the other hand, by utilising the hygroscopicity of the PIL and the water production at the cathode side of the fuel cell, a way to enhance the conductivity of the electrolyte may be provided. The amphoteric water will act as a proton acceptor and donor and participates in the proton transfer processes in the bulk of the PIL as a fast carrier.
In this contribution we present an experimental study on the proton transport mechanism in a high Brønsted-acidic PIL as a function of the amount of residual water. Moreover, the interaction of the high acidic PIL with a host polymer with basic moieties like m-PBI is investigated to ascertain its ability as a possible membrane material. A PIL with a high acidic cation, 2-Sulfoethylmethylammonum triflate [TfO] is prepared. The ability to protonate the residual H 2 O were investigated by mixing appropriate amounts of the PIL and H 2 O at various molar ratios to obtain compositions varying from neat PIL to H 2 O-excess conditions. The interactions between PIL cation and H 2 O were determined by measuring the macroscopic (total) conductivity and the self-diffusion coefficients by 1 H-NMR spectroscopy (PFG/DOSY). The neat PIL is introduced into PBI polymer membrane by a doping (swelling) process. The PIL uptake degree was monitored by the weight increase. The doping process was monitored by infrared (IR) spectroscopy, proving the protonation of base imidazole groups on PBI chains. Thermogravimetric alanysis (TGA) measurements were used to determine the thermal stability of the doped membranes.

Ionic Liquid Preparation
[2-Sema][TfO] was prepared by slowly adding trifluoromethanesulfonic acid (reagent grade, 98%, Sigma Aldrich) to 2-methylaminoethansulfonic acid (N-methyltaurine, 99%, Sigma Life Science). A more detailed description of the preparation process can be found elsewhere [9]. Coulometric Karl-Fischer titration (852 Titrando/Metrohm company) yielded a water concentration of 0.6 wt.%. By adding appropriate amounts of water, various compositions from neat to equimolar ratio between PIL and H 2 O molecules were achieved. In the case of [TfO], a water concentration of~6 wt.% corresponds to an equimolar ratio, i.e., 50 mol.% H 2 O.

1 H-NMR Parameters
The acquisition of the NMR spectra was performed by using a Bruker 600 MHz spectrometer, equipped with a 5 mm cryoprobe tuned to 1 H. The chemical shifts were determined by using an external deuterium reference (field lock), i.e., capillaries filled with D 2 O and enclosed together with the PIL in the sample tubes. The measurements were conducted at 90°C to achieve a lower viscosity compared to room temperature.
The NMR samples of the membranes were dissolved in DMSO-d6 and filled in 5 mm tube. The measurements were conducted at room temperature (25°C).

Measurement of the Diffusion Coefficients
The self-diffusion coefficients were measured by using the PFG/DOSY technique. The applied parameters were: 3.5 ms diffusion gradient length (d), 100 ms diffusion delay (Δ), 15 gradient increments with gradient strength (g) from 1.3 to 32.5 G cm -1 . The combination of d and Δ was optimized to give rise to at least 85% signal attenuation at strongest gradient field.

Ionic Liquid Doped Membrane Preparation
PBI membranes with a thickness of 60 mm (un-doped) were cut into samples of~5 cm 2 . The membrane samples were predried by a heat treatment at 150°C for 30 min. The membranes were immersed in the neat PILs for a doping (swelling) process at various temperatures to investigate the uptake kinetics. The membrane samples were weighed before and after immersion to obtain the mass difference. The uptake degree was calculated by the weight increase, as given in Eq. (1): where m a and m b are the mass of original dried PBI membrane and the after doping membrane, respectively.

Membrane Characterization
Attenuated total reflection (ATR) spectra of the undoped and doped membrane samples and of the neat PIL were measured in reflection mode in the range of 500-4,000 cm -1 (Monolithic diamond GladiATR, PIKE technologies). The experiments were carried out at room temperature. The thermal stability of the membranes and PIL were examined, by using TGA (Perkin Elmer STA 6000). The samples were heated from room temperature to 800°C with a heating rate of 5°C min -1 in air atmosphere. Weight loss was measured and reported as a function of temperature. The baseline correction was performed by using an empty crucible and an identical measurement program. The images of membranes were taken by a light microscope (ZEISS AXIO Imager, M1m).

(Total) Conductivity vs. Temperature and H 2 O Concentration
To investigate the influence of the water content on the proton transport in the PIL, 2-6 wt.% H 2 O was added. This corresponds to a molar ratio of 25-50 mol.% H 2 O. The conductivity data were reported in our previous work [10]. The AC conductivity measurements were performed in a four-probe conductivity cell, using platinum electrodes. The total ohmic resistance was determined by means of impedance spectroscopy. The dependency of the (total) conductivity s in the [TfO]/H 2 O system on the temperature T of nearly neat [TfO] up to a ratio of 50 mol.% H 2 O is shown in Figure 1. The conductivity depends on the H 2 O content w H 2 O and the temperature T. In the range of 60-100°C no distinct Vogler-Tammann-Fulcher (VTF) behavior is observed. The activation energy E a decreases considerably from 44.9 kJ mol -1 for the neat PIL to 37.6 kJ mol -1 for a molar ratio of 50 mol.% H 2 O.
Both, an increasing temperature T and an increasing H 2 O concentration lead to a decrease of the viscosity of a PIL. In the case of pure vehicular charge transport, the conductivity is coupled via the Stokes-Einstein relation to the (dynamic) viscosity h. However, due to the high acidity of the [2-Sema] + cation, a significant amount of the cations undergo an intermolecular proton transfer to the residual water molecules , as given in Eq. (2): Considering the pK A values of H 3 O + and of a SO 3 H moiety, the equilibrium concentration of H 3 O + is most likely in the same order of magnitude as the concentration of the remaining [2-Sema] + cations. Due to the protolysis equilibrium, the proton transport in the PIL/H 2 O systems can take place not only by the migration of the cation but also of the H 3 O + ions. It can assumed that the mobility of the smaller H 3 O + ions is higher than that of the cation. Thus, more mobile charge carriers are formed with increasing H 2 O concentration. Alternatively, an cooperative proton transport mechanism can take place, due to a fast excange of protons between H 3 O + , H 2 O, [2-Sema] + and N-Methyltaurine. The change of the activation energy vs. H 2 O concentration indicates either a change of the predominating charge carrier or of the mechanism. To decide this, 1 H-NMR measurements were performed.

1 H-NMR and Self-diffusion Coefficients
The prevailing mobile charge carrier, respectively the proton transport mechanism, can be evidenced by performing  Figure 7.
The two signals of (nearly) neat [TfO], appearing at high magnetic fields at a shift of 3.4 and 4.1 ppm, can be assigned to the CH 3 and CH 2 CH 2 protons and the signal at a medium field of 7.5 ppm is assigned to the NH 2 protons. The acidic proton of the SO 3 H group appears at a low magnetic field and a shift of 12.9 ppm. With increasing H 2 O content w H 2 O , up to an equimolar composition (~6 wt.%), it shifts about 1.9 ppm towards a higher field.
The latter observation can be caused by strong interactions between the high acidic protons of the [2-Sema] + cations and the H 2 O molecules. If the kinetic of the proton exchange between the SO 3 H group and the H 3 O + ions is very fast, both species cannot be separated by 1 H-NMR. This results in a single signal at an averaged position. Adding H 2 O will increase the H 3 O + concentration in the protolysis equilibrium because of the high acidity of the [2-Sema] + cations and will increase the total concentration of the active (mobile) protons.   As the protons of H 3 O + ions can be expected at a lower field than the protons of H 2 O molecules and at a higher field than the protons of SO 3 H groups [11], this may lead to the observed shift to a higher field and in an increase of the relative integral peak intensity (area). The proton self-diffusion coefficients D i of the different protonic species were measured by using the PFG/DOSY technique. In Figure 3 the evaluated self-diffusion coefficients for a temperature of 90°C are shown as a function of the H 2 O content. Because of the high viscosity of [TfO], the DOSY measurements were at this elevated temperature for a more precise measurement of the relaxation time.
The self-diffusion coefficient D SO 3 H=H 2 O the of active (acidic) proton differs from the self-diffusion coefficient D NH 2 þ , D CH 2 CH 2 , and D CH 3 of the other protons attached to the cation, see Figure 3. In the whole investigated H 2 O concentration range the active proton diffuses distinctly faster than other protons. With increasing H 2 O content, the self-diffusion coefficient of all protons are increasing, but in the case of the active proton (SO 3 H/H 2 O) the increase (slope) is much higher compared to the others (NH 2 + , CH 2 CH 2 and CH 3 ).
When extrapolating to ideally neat [TfO], i.e., w H 2 O ¼ 0, the difference between the self-diffusion coefficients vanishes. This confirms to the assumption made above on the possible transport mechanisms in a high acidic PIL. In the neat PIL the proton transport is restricted to the diffusional motion of the whole cation. A pure vehicle mechanism is present. A value of (7 + 2) Á 10 -8 cm 2 s -1 can be estimated for the [ With increasing H 2 O concentration and increasing concentration of H 3 O + due to protolysis, more mobile charge carriers are present. Moreover, due to a fast exchange of the proton between [2-Sema] + and H 3 O + , the predominating proton transport mechanism changes from vehicular to cooperative. The increase of the self-diffusion coefficients of the other protons may be only caused by the change of the viscosity, as already discussed above.

Uptake of PILs by m-PBI by a Swelling Process
There is a slow uptake of neat [TfO] by m-PBI, when performing a swelling experiment with a membrane sample. The doping process is strongly dependent on the doping temperature. The degree of uptake (increase in wt.%) at various doping temperatures T depends on the doping time t according to Fick's 2 nd law, as depicted in Figure 4. Thus, a higher temperature provides a faster doping process. However, the uptake reaches a certain limitation of about 125 to 135 wt.%. From the temperature dependence of the diffusion process, an activation energy E a of 122.7 kJ mol -1 can be estimated. Using a four-probe setup the obtained membranes exhibit a (total) conductivity of 2.68 mS cm -1 at 100°C and 30% rel. humidity (RH).
The diffusion coefficientsD [TfO] evaluated from the doping process are depicted in Table 1. The Fick's diffusion coefficientD [TfO] of [TfO] in m-PBI at a temperature of 100°C has a value of about 1.9 Á 10 -10 cm 2 s -1 . This is 3 orders of magnitude lower compared to the analogue doping process with H 3 PO 4 [12]. This may be caused by the apparent bigger molecular size of [TfO] and a spatial steric hindrance.
The same doping experiment of m-PBI is performed with the medium acidic PIL 1-Ethylimidazole triflate [TfO] (pK a = 7.26), and the low acidic PIL, Diethylmethylammonium triflate [Dema][TfO] (pK a = 10.55) [10]. No uptake of these PILs can be detected. The uptake of a PIL by a PBI-type polymer may be highly dependent on the cation acidity, and thus on the ability to protonate the basic imidazole moieties.

IR Spectra of [2-Sema][TfO] and [2-Sema][TfO] Doped PBI Membranes
The doping process of [TfO] into m-PBI was monitored by IR-ATR spectroscopy. The IR-ATR spectra of the PBI membrane, [TfO] and the [TfO] doped PBI membrane are depicted in Figure 5. In the case of the pure m-PBI sample, the band at 3,415 cm -1 is attributed to the isolated N-H stretching mode of the imidazole, i.e., nonhydrogen bonded ''free'' N-H groups, whereas the bands at 3,250-2,500 cm -1 are assigned to the stretching modes of selfassociated N-H bonds [13]. The N + -H vibrations will appear in this range, because of the protonation of the imine in the doped samples [14]. The C=N stretching mode in the imidazole ring appears in the region of 1,606 cm -1 . A band at 1,534 cm -1 is attributed to the in-plane deformation of the benzimidazole rings [15].
In the case of [TfO], there are no studies available concerning vibrational spectroscopy. The IR bands belonging to vibration modes of the triflate anion can be readily assigned by comparing, e.g., with literature data of aqueous sodium triflate or other triflate based ILs [16,17]. When considering the standard textbooks, the band at~3,200 cm -1 in neat [TfO] may be assigned to the O-H stretching mode of SO 3 H group in the cation and the band at~2,850 cm -1 to the N + -H or C-H stretching modes. The bands which show up at 1,280 and 1,020 cm -1 can be assigned to the asymmetric and symmetric stretching modes of the SO 3 moiety of the triflate anion. The asymmetric and symmetric stretching modes of the CF 3 moiety show up at 1,210 and 1,164 cm -1 . As well, the bands appearing at 632 and 512 cm -1 can be assigned to the anion, caused by the symmetric and asymmetric deformation vibration of the SO 3 moiety [18][19][20]. Thus, only the remaining bands at 573, 728, 910, 1,354, and 1,469 cm -1 are most probably caused by vibration modes of the cation.
The spectral region of the O-H and N-H stretching modes as shown in Figure 5, reveals the evolution of the protonation of the polymer by the PIL. In the doped samples, the broad band of the N + -H stretching mode at 3,250-2,500 cm -1 becomes stronger. Simultaneously, it shows a shift to lower wavenumbers, which may indicates an increasing fraction of H-bonded species [21]. The band which appears at 3,400 cm -1 can be probably attributed to the association of O-H from both cation and residual H 2 O. The C=N stretching mode at 1,606 cm -1 and the C-C stretching mode at 1,534 cm -1 of the imidazole ring are shifting towards higher wavenumbers to 1,632 cm -1 and 1,578 cm -1 , respectively. This may be caused by the protonation of imidazole ring. The increasing electron density on carbon of the heterocycle leads to a raising absorption frequencies of ring vibrations [22].

TGA Analysis of [2-Sema][TfO] and [2-Sema][TfO] Doped PBI Membranes
The thermal properties of neat m-PBI and [TfO], as well as of TfOH and [TfO] doped m-PBI membranes are investigated by TGA, as depicted in Figure 6. TfOH doped m-PBI membranes were also prepared. By comparing with the [TfO] doped m-PBI membranes this allows a limited insight into the absorption stages, respectively to the composition. For doping m-PBI with TfOH, aqueous solutions with a concentration of 20 wt.% were used to keep control to the process and avoid a complete dissolution of the polymer material. Neat m-PBI shows an excellent thermal stability, the polymer starts to decompose beginning from 570°C. The weight loss below 150°C might be attributed to residual water and solvent. In the case of neat [TfO] a weight loss of about 5 wt.% took place between 120°C and 197°C. This notable low decomposition temperature might be caused by the high acidity of the cation. As discussed, the residual water is subject to a protolysis equilibrium with the cation. The hydrate of triflic acid (i.e., hydroxonium triflate) is appreciable volatile (T b = 218-219°C) [21][22][23]. Alternatively, the anion may be re-protonated by the cation, leading to a direct loss of TfOH (T b = 162°C) [24].
In the case of TfOH doped m-PBI the polymer chains were fully protonated by TfOH, as known for H 3 PO 4 doped m-PBI, because of its high acidity [22]. Thus, a weight increase of about 98 wt.% can be predicted, assuming 2 TfOanions per m-PBI repeating unit. In the obtained ''polybenzimidazolium triflate'' m-PBI-H 2 + (TfO -) 2  The uptake of a PIL by m-PBI membrane in a swelling process is dependent on the acidity of cation. The protonation of m-PBI chain is obviously a prerequisite for the uptake of an electrolyte. In the case of the medium acidic PIL [TfO] and the very low acidic PIL [Dema][TfO] no interaction with m-PBI is observed, i.e., there is no uptake by/weight increase of the polymer (pK A cation < pK A BImH+ = 5.6).

NMR Analysis of [2-Sema][TfO] and [2-Sema][TfO] Doped PBI Membranes
The composition of the [TfO] doped m-PBI, indicated by the TGA measurements, can be proved by 1    According to the detected fractions of cation, anion and N-methyltaurine, it can be assumed that the free (conjugated) base N-methyltaurine diffuses out after protonation. There are two triflate anions per repeat unit of the PBI chains, which will provide~95 wt.% weight increase. Thus, a PBI Á triflate is formed. The remaining~30wt.% weight increase is due to the uptake of additional PIL, i.e., [TfO] and some N-methyltaurine. The additional species are absorbed by forming H-bonds in protonated PBI membrane. The low (effective) doping degree regarding the PIL may explained the poor (total) conductivity.

Analysis of [2-Sema][TfO] Doped PBI Membranes by Light Microscopy
The [TfO] doped PBI membranes were also investigated by light microscopy, a micrograph is depicted in Figure 8. Incremental density of mechanical defects can be observed with increasing doping level, which may cause a decay of the mechanical properties.

Conclusions
The high-Brønsted acidic proton conducting ionic liquid 2-Sulfoethylmethylammonium triflate [TfO] was investigated in this study. There is evidence that the proton transport mechanism in this PIL with a high acidic cation changes from a vehicular in the case of the neat PIL to a cooperative mechanism as a function of the H 2 O concentration.
There is an interaction of the high acidic PIL with m-PBI, a polymer with basic moieties, when performing an interdiffusion/swelling experiment. The uptake degree of [TfO] by a PBI membrane -expressed as a weight increase -reaches a value of about 130%. Using a membrane with 60 mm thickness the process takes several days due to a slow interdiffusion. A protonation of the base imidazole moieties takes place but there is an out-diffusion of the neutral base N-methyltaurine after the proton transfer from the cation to the polymer. Proton exchange, respectively an interaction between the polar groups and water can be observed in the NMR spectra, indicating a network of H-bonds in doped PBI. Thus, highly acidic PILs are promising candidates for the use as non-aqueous electrolytes. However, a simple swelling process to immobilize the electrolyte in a host polymer -as used successfully to prepare H 3 PO 4 /PBI membranes for HT-PEFCs -is no applicable. The resulting (total) conductivity is not sufficient. To reach a higher doping degree solution casting may be an alternative preparation method and should be investigated in a future work.