Influence of the acid–base stoichiometry and residual water on the transport mechanism in a highly-Brønsted-acidic proton-conducting ionic liquid

In this study, Brønsted-acidic proton conducting ionic liquids are considered as potential new electrolytes for polymer membrane fuel cells with operating temperatures above 100 °C. N-Methyltaurine and trifluoromethanesulfonic acid (TfOH) were mixed at various stoichiometric ratios in order to investigate the influence of an acid or base excess. The proton conductivity and self-diffusion of the “neat” and with 6 wt% water samples were investigated by following electrochemical and NMR methods. The composition change in the complete species and the relative proton transport mechanism based on the NMR results are discussed in detail. During fuel cell operation, the presence of significant amounts of residual water is unavoidable. In PEFC electrolytes, the predominating proton transfer process depends on the cooperative mechanism, when PILs are fixed on the polymer matrix within the membrane. Due to the comparable acidity of the cation [2-Sema]+ and the hydroxonium cation, with excess N-methyltaurine or H2O in the compositions, fast proton exchange reactions between the protonated [2-Sema]+ cation, N-methyltaurine and H2O can be envisaged. Thus, an increasing ratio of cooperative proton transport could be observed. Therefore, for polymer membrane fuel cells operating at elevated temperatures, the highly acidic PILs with excess bases are promising candidates for future use as electrolytes.


Introduction
Polymer electrolyte fuel cells (PEFCs) operating at elevated temperatures (>100 C) offer signicant improvements over lowtemperature PEFCs, such as no humidication of the feed gas, no water recirculation, a more efficient cooling of the cell and a higher tolerance against feed gas impurities. [1][2][3] The proton conductivity of NAFION®-based proton exchange membranes (PEMs), used in PEFCs for low operation temperatures, depends mainly on the polymer's water uptake. For operation at elevated temperatures (>100 C), the conductivity of a new membrane material should be maintained in anhydrous conditions. Currently, (high temperature-) HT-PEFCs are based on polybenzimidazole (PBI) membranes doped with phosphoric acid (H 3 PO 4 ). [4][5][6][7] However, the presence of H 3 PO 4 leads to slow cathodic oxygen reduction reaction (ORR) kinetics. 8 There is a specic adsorption of H 3 PO 4 species on active sites in the redox catalyst platinum, which causes an inhibition (poisoning) effect. 9,10 In addition, the insufficient solubility and diffusivity of oxygen is discussed. 11 Thus, there is a need for new nonaqueous proton-conducting electrolytes to be operational for temperatures of 100-120 C.
Proton-conducting ionic liquids (PILs) are promising candidates as non-aqueous electrolytes at operating temperatures >100 C. Ionic liquids (ILs) are ionic compounds with bulky cations and anions and thus a low lattice energy. 12 PILs have received much attention as a potential electrolyte in PEFCs due to their good conductivity, wide electrochemical windows and low ammability. [13][14][15][16][17][18] In a PIL, the cation or anion may act as a protonic charge carrier, and so either the cations or anions are Brønsted-acids. In the case of cations, i.e., a PIL of the type HB + A À , it consists of an (organic) base B, protonated by a very strong acid HA, respectively a super acid: The anions of super acids, such as triuoromethanesulfonic acid or bis-triuoromethylsulfonimid, have a less inhibiting effect on electrocatalytically-active electrode surfaces than H 3 PO 4 . The triimid (CF 3 SO 2 ) 2 N À and triate CF 3 SO 3 À anions interact only very weakly with metal atoms, resulting in weak adsorption on a Pt surface. 9,19,20 In a water-free PIL of the type HB + A À , protons can only move in an electric eld via the protonated cations HB + by means of a vehicle mechanism. A drawback of ILs or PILs is oen poor conductivity because of relatively high viscosity. A proton transfer back to the anion of the superacid, e.g., [TfO] À , has only a very small probability because the protolysis equilibrium in eqn (1) is on the far right side. Thus, a cooperative transport mechanism involving the anions is not possible. In the case of vehicular mechanism, conductivity and viscosity are coupled to each other according to the Stokes-Einstein relation. However, to avoid the leakage of the liquid electrolyte during operation, a PIL applied in a PEFC must be immobilized in a polymer matrix. A study of a [Dema][TfO]-doped PBI membrane by Liu et al. shows an activation energy of the conductivity in the range of the cooperative mechanism, 21 whereas the pure [Dema][TfO] is vehicular. 22 In this case, the vehicular transport of the cation HB + is constrained and a cooperative proton transport mechanism would be advantageous. This was shown in a study by Noda et al. that the excess in the base B of a PIL of the type HB + A À , which provides the cooperative proton transport and improves conductivity. 12 The excess base B acts as a proton acceptor that is protonated by the proton donor HB + . The cooperative transports through the excess base B only necessitate the reorientations of the involved particles B an HB + . This results in increased proton conductivity and reduced activation energy for the conduction process. 12,23,24 In particular, PILs contain strong Brønsted-acidic cations that are usually highly hygroscopic. The water absorption is difficult to prevent. Moreover, under fuel cell operation, water will be generated on the cathode side. The presence of residual water acts as a proton acceptor and gives rise to a protolysis equilibrium with the cation HB + : Its extent depends on the acidity of the cation. In a preceding NMR study, it was shown that in Brønsted-acidic PILs of the type HB + A À cooperative proton transfer will dominate, depending on the cation acidity and the residual water content. 25 There will also be fast exchange between HB + , B, H 2 O and H 3 O + . In general, as discussed above, the introduction of a proton acceptor would improve the cooperative transport. The coexistence of excess base B and residual water offer the possibility to improve the technically utilizable conductivity of PIL electrolytes.
In this experimental study, the effect of the PIL acid-base stoichiometry on the proton transport mechanism in a system with residual water is investigated. In general, cooperative proton transport in an IL system requires the presence of a proton acceptor and proton donor with comparable acidity. 26 A highly acidic PIL, 2-sulfoethylmethylammonum triate [TfO] (pK A1 z À1), 27 is used. The acidity of this is comparable to the hydroxonium cation (pK A ¼ 0). The 2-sulfoethylmethylammonium cation is prepared by means of the protonation of 2-methylaminoethanesulfonic acid (N-methyltaurine), which exists as a zwitterion due to tautomerism. Because of the presence of the sulfonic acid functionality, it is a very strong acid that can protonate the residual water at a signicant percentage. 8 Appropriate amounts of N-methyltaurine (MTau) and tri-uoromethanesulfonic acid (TfOH) are mixed at various molar ratios to vary the PILs compositions from the TfOH-excess to MTau-excess. The interactions between the cations, the excess base and H 2 O are determined by means of electrical conductivity measurements and 1 H NMR spectroscopy. Using a pulsed-eld gradient (PFG-) NMR technique, the self-diffusion coefficients of the individual protons in the PILs are obtained. The effect of stoichiometry and residual H 2 O on the prevailing proton transport mechanism 26 is discussed by comparing the measured macroscopic and microscopic properties.

Conductivity measurement
The AC conductivity measurements were performed in a fourprobe conductivity cell, using platinum electrodes. The cell constant as a function of the sample volume was determined by using a 0.1 M KCl solution for calibration. The intended water contents of the binary PIL + H 2 O mixtures were checked using Karl-Fischer titration at the beginning of each measurement. The total ohmic resistance s as a function of the temperature T of the neat PIL and of the PIL + H 2 O samples was determined by means of impedance spectroscopy. The temperature T was increased in increments of 10 C from 60 to 110 C and vice versa. The excitation amplitude was adjusted to 10 mV. The specic conductivity s was calculated by using the cell constant.

H NMR parameters
The acquisition of the NMR spectra was performed using a Bruker 600 MHz spectrometer, equipped with a 5 mm cryoprobe tuned to 1 H. A capillary lled with D 2 O was enclosed the sample tubes as a eld lock. The measurements were performed at 90 C, because at lower temperatures, the increasing viscosity leads to high relaxation times.

Measurement of the self-diffusion coefficients
The self-diffusion coefficients of the observable protons were measured using the diffusion-ordered spectroscopy (DOSY) technique at 90 C. The measurements were performed by applying 30 eld gradient increments with a gradient strength g from 1.3 to 32.5 G cm À1 . The values of the gradient pulse length d and the diffusion time intervals D were optimized to aim for least 85% signal attenuation at the strongest eld gradient. The value of the (self-) diffusion coefficient D i of a certain proton species i was obtained from the decay of its measured echo intensity vs. the gradient eld strength g.

Results and discussion
Total conductivity vs. stoichiometry and water content The measurements of the total conductivity s were performed in the temperature range between 60 and 110 C. The total conductivity includes cationic and anionic charge transport. In the x[MTau]$(1 À x)[TfOH] and the x[MTau]$+(1 À x)[TfOH] + 6 wt% H 2 O samples, the viscosity was strongly dependent on the composition. In general, the viscosity was rising with an increasing fraction x of the (at room temperature solid) base Nmethyltaurine, i.e., with increasing MTau-excess. In the case of an increasing TfOH-excess, the viscosity was generally decreasing, i.e., with decreasing x. A higher content of H 2 O also leads to a lower viscosity. The dependence of the total conductivity s, respectively of the specic total conductivity L on the viscosity h, can be explained according to the Stokes-Einstein and Nernst-Planck relations: Assuming an ionic compound A n 1 B n 2 C n 2 .X n i , n i denotes the stoichiometric factor, z i the charge number and r i the hydrodynamic radius of the ionic species i. In the case of a dissociation degree a Diss not being equal to unity, the concentration of the ionic species i is denoted with c i and the initial concentration of the ionic compound A n 1 B n 2 C n 2 .X n i with c 0 . Thus, a decrease in the viscosity will accelerate the vehicular proton transport by PIL cations and H 3 O + .
The dependency of the total conductivity s on the neat samples and samples with a water content of 6 wt% on the temperature T and the stoichiometry x is depicted in Fig. 2(a) and (b), respectively. The course of the conductivities corresponds to the change in the viscosity. In the case of the neat samples, depicted in Fig. 2(a), the sample with the smallest MTau molar fraction x ¼ 0.3 exhibits the highest total conductivity. Correspondingly, the total conductivities of the samples with a water content of 6 wt% are generally higher compared to the neat sample, as is depicted in Fig. 2 For all of the investigated samples, the total conductivity s increases as a function of the temperature T. However, the extent of the conductivity increase vs. temperature is different. The highest impact on the conductivity was found for the sample with the highest MTau molar fraction of x ¼ 0.7. For this MTau molar fraction, in the case of the neat samples, s increases by a factor of 8.4 when T rises from 60 C to 110 C. In the case of the sample with a molar fraction of x ¼ 0.3, the increase was only by a factor of 3. The samples with a water content of 6 wt% and a molar fraction of x ¼ 0.7 and x ¼ 0.3 show a similar behaviour, with the conductivity increasing by a factor of 8.52 and 2.75, respectively. The different factors may indicate that there is not only a change in viscosity but also of the proton transport mechanism responsible for the increase in the total conductivity when increasing the molar fraction of MTau. The total conductivity describes the bulk charge transport macroscopically and includes all mobile charge carriers ([2-Sema] + , [TfO] À and H 3 O + ). The underlying ionic charge (proton) transport mechanisms can only be discerned by techniques sensitive to the local environment of the mobile charge carriers, i.e., the NMR.
In the following, 1 H-NMR and 1 H-PFG-NMR are used to measure the local dynamics of the mobile protonic charge carriers, as well as their self-diffusion coefficients. As discussed above, in a PIL of the type HB + A À , an excess of the base B as well as a certain water content are able to provide cooperative transport and thus improve the proton conductivity. Therefore, depending on the excess of the base and water content, both transport mechanisms, vehicular and cooperative, may be present in a sample.
Due to the high viscosity of the samples at room temperature, especially in the case of MTau-excess, the NMR measurements were all performed at 90 C to avoid an FWHM of the signals too broad to evaluate.  Paper (d) and the CH 2 CH 2 (c, c 0 ) protons. In the case of samples with a base-excess, i.e., x > 0.5, the protons of the CH 2 CH 2 backbone (c, c 0 ) are difficult to distinguish due to a general increase of the FWHM in the spectra. The NMR chemical shi d of a proton depends on the local screening of the external magnetic eld by the local electron density. In the timescale of the NMR experiment, a chemical shi to a lower eld represents a generally lower local electron density or higher "delocalization" of the proton. The local electron density can be affected by intramolecular interactions with adjacent groups or intermolecular interactions by hydrogen bonds. A high acidic proton is accompanied by a low local electron density. In addition, the formation of a hydrogen bond leads to the deshielding of a proton. Thus, the (NMR-) active protons of a Brønsted-acidic PIL in particular are inuenced not only by the water content but also by the acid-base stoichiometry, as the active protons can easily form hydrogen bonds and are subject to intramolecular interactions. In the case of the water-free samples, the molar fractions of the species [2-Sema] + , MTau, TfOH and [TfO] À should vary as a function of the stoichiometry x, as depicted in Fig. 4. TfOH is a much stronger acid compared to the [2-Sema] + cation, and so we can safely assume a complete proton transfer from TfOH to MTau for the entire range of stoichiometry x. This should lead to a decreasing fraction of TfOH in the stoichiometric range from x ¼ 0 to 1/2 and an increasing fraction of MTau in the range from x ¼ 1/2 to 1.
As the acidity of the NH 2 + moiety in the [2-Sema] + cation and MTau is not sufficient to protonate SO 3 À moieties (intra/ intermolecular) to a noticeable extent, the residence time prior to re-transfer should be very short. Thus, protolysis equilibria should not contribute to the observed stoichiometry xdependent shi of these protons (pK A2 of NH 2 + in MTau, about 10.2). [27][28][29] As is shown in Fig. 4 In the stoichiometry range between x ¼ 0.5 to 0.7, the signal shis with a slope of 0.45 ppm/0.1 Dx; see Fig. 4.
In the case of TfOH-excess, i.e., for samples with a stoichiometry of x ¼ 0.3 to 0.5, the shi in the signal is comparably small and not monotonic. Due to the TfOH excess, the SO 3 H moiety of the MTau is fully protonated. There are only variable fractions of [2-Sema] + cations, TfOH and [TfO] À . Thus, there is primarily an intermolecular proton transfer between the anion [TfO] À and TfOH, which also results in a de-shielding of the proton: Changing the stoichiometry from x ¼ 0.5 to 0.4 leads to an increasing delocalization of the SO 3 H proton and to an initial downeld shi, as seen in the NMR chemical shi in Fig. 3(b). In the system TfOH/[TfO] À , the presence of [TfO] À should be equivalent to the excess base MTau in the system [2-Sema] + /MTau. Thus, a further increase in the molar fraction of TfOH may inhibit the delocalization again, resulting in the observed shi of the SO 3 H proton back towards the higher eld when changing the stoichiometry from x ¼ 0.4 to 0.3, due to the restricted mobility.
In the case of the neat, nearly water-free samples, a change in the stoichiometry will affect the probability and duration of forming hydrogen bonds to the NH 2 proton on cations in the timescale of the NMR measurement. In the case of the SO 3 H proton, the MTau-excess leads to a signicant delocalization between the SO 3 À /SO 3 H sites in the timescale of the NMR measurement and TfOH-excess results to a delocalization between the [TfO] À /TfOH. Both of these may explain the observed signal shi to high magnetic elds in the 1 H NMR spectrum. The acid-base stoichiometry also inuences the proton transport processes. The observation of a proton delocalisation on the NMR timescale indicates the possible presence of intermolecular (cooperative) proton transport. The inuence of the proton delocalization on the proton transport will be discussed in relation to the self-diffusion coefficient in the section " 1 H-PFG-NMR".

H-NMR of x[MTau]$(À x)[TfOH] samples with 6 wt% water
with a water content of 6 wt% at 90 C is depicted in Fig. 5(a) and the chemical shi d vs. stoichiometry x of the of SO 3 H and NH 2 + protons in Fig. 5 (6) and (7): TfOH is the much stronger acid compared to the [2-Sema] + cation and the acidity of the hydroxonium cation H 3 O + is on the same order as the acidity of the [2-Sema] + cation. † 30,31 A preceding work conrms the fast exchange of the H 2 O proton and SO 3 H proton, resulting in a single signal in the spectrum of SO 3 H/H 3 O + /H 2 O at an average NMR shi. 25 Moreover, the signal of the SO 3 H proton shows a corresponding increase in its integral area with increasing H 2 O content, which is not the case for the NH 2 + or alkyl protons.
Considering the protolysis reactions in eqn (6) and (7) can be estimated for a sample with a stoichiometric composition (x ¼ 0.5). 32 These assumptions are underpin the construction of the tentative plot in Fig. 6.
In  (6) and (7). Increasing the stoichiometry x from The proton transport mechanism is further discussed, together with the 1 H-PFG-NMR/DOSY measurements and the self-diffusion coefficient in the next section. The mobile protonic charge carriers in the PIL/H 2  If there is only vehicular transport, this should principally lead to a decrease in the (total) conductivity with increasing x. If there is also cooperative transport, the presence of H 2 O, acting as a proton acceptor, will also accelerate the intermolecular proton transfer between the MTau and the [2-Sema] + cation, leading to faster cooperative transport. For a stoichiometry of x > 0.3, the fraction of (free) H 2 O increases with increasing x. A maximum of an additional cooperative transport and thus of the (total) conductivity should be expected when the equimolar fractions of H 2 O and H 3 O + are present (the maximum probability for proton transfers between H 3 O + and H 2 O). This is approximately the case for a stoichiometry x z 0.4 and corresponds well to the measured values for the total conductivity which exhibits a maximum at this stoichiometry; see Fig. 2(b) and 6. However, the total conductivity is highly coupled with the viscosity. At a higher stoichiometric composition x > 0.4, the vehicular transport is again attenuated and thus the conductivity is decreased.

H-PFG-NMR/DOSY
The self-diffusion coefficient of the protons is measured by 1  As observed for the total conductivity s, the diffusion coef-cient D H + of the active proton and D cation of the [2-Sema] + cation are decreasing with increasing stoichiometry x (from 0.3 to 0.65, respectively, to 0.7). The dynamic viscosity h of the samples is directly coupled to the diffusion coefficient of the [2-Sema] + cation D cation due to the Stokes-Einstein relation. In the case of cooperative transport, there is a distinct decoupling from viscous processes. Thus, if vehicular and cooperative mechanisms are both present, an increase in viscosity will increase the impact of cooperative transport to the total proton transport. The share of cooperative processes in all of the samples is evaluated by calculating the ratio between D H + ,coop to D H + as follows: 25 The diffusion coefficient D H + ,vehicle for vehicular transport is identical to that of the [2-Sema] + cation. The ratio represents the share of cooperative transport in the total proton diffusion process. The share of cooperative transport is depicted in Fig. 7. With increasing stoichiometry x, i.e., with an increasing fraction of MTau, the share of cooperative transport is also increasing for the neat samples and for samples with a water content of 6 wt%. When comparing the course of both curves, as expected, the presence of the amphoter water generally increases the share of the cooperative mechanism. The ratio of D H + ,coop to D H + reaches a value of 82% in a sample with a stoichiometry of x ¼ 0.65 and 6 wt% water content. Corresponding to the chemical shi in the 1 H-NMR spectrum, the ratio relates to the ability of the protons to be delocalized on the NMR timescale and thus to participate in intermolecular transfer. A higher share of cooperative transport generally leads to a lower eld of the chemical shi, as discussed above. Conclusion N-Methyltaurine and TfOH were mixed at various molar ratios in order to obtain samples of the proton-conducting liquid [TfO] with various amounts of excess free acid TfOH or excess free base N-methyltaurine. The "nearly neat" samples and the samples with 6 wt% residual water were investigated regarding proton conductivity and self-diffusion using electrochemical and NMR methods. It could be observed that an excess of the free base MTau retards the vehicular proton transport due to an increase in the dynamic viscosity, which leads macroscopically to a lower (total) conductivity. In an PEFC electrolyte, based on a PIL immobilised in a polymer matrix, the proton conductivity depends on the presence and the share of cooperative transport as vehicular transport is signicantly hampered. The use of a base-excess high acidic PIL would allow a higher fraction of cooperative transport and thus a higher proton conductivity. For the future use as conductive electrolytes in PEFCs at elevated operation temperatures (100-120 C) and atmospheric (non-humidied) operation, a PIL with a high hygroscopicity to retain H 2 O formed during operation at the cathode and an excess of the base may be favorable. Enabling fast cooperative transport may help in reaching sufficient proton conductivities.

Conflicts of interest
There are no conicts to declare.