The Nature of Proton Shuttling in Protic Ionic Liquid Fuel Cells

It has been proposed previously that protic ionic liquids (PILs) such as diethylmethylammonium triflate could be used as the electrolytes in nonhumidified, intermediate temperature H2 fuel cells, potentially offering the prospect of high conductivity and performance, even under anhydrous conditions. In this contribution, a combination of electroanalytical chemistry and fuel‐cell polarization analyses is used to demonstrate for the first time that the pure PILs cannot support proton shuttling between the electrodes of fuel cells. Only through the inclusion of dissolved acidic or basic proton shuttles can viable protic ionic fuel cells be fabricated, which has major consequences for the use of these neoteric electrolytes in fuel cells.

to Brønsted bases and exhibit very low vapor pressures and inherent conductivity. [6] The most promising PIL for use in fuel cells is diethylmethylammonium triflate, [dema][TfO] (Scheme 1), which is thermally stable to 360 °C ( Figure S1, Supporting Information), electrochemically stable over an almost 4 V wide potential window ( Figure S2, Supporting Information), and exhibits an anhydrous conductivity of 68 mS cm −1 at 150 °C. [5a] It has been proposed that PILs can shuttle H + ions between the electrodes of fuel cells via the cations of the PIL; protonation of the parent base diethylmethylamine (which is denoted dema) upon anodic oxidation of H 2 yields protonated [dema] + cations (Equation (1)), which are subsequently deprotonated at the cathode (Equation (2)) [7] H 2dema 2 dema 2e 2 [ ] In this communication, we demonstrate that this H + -ion shuttling mechanism cannot occur effectively in cells containing the pure PIL, and that fuel cells require the use of "nonstoichiometric" PILs containing added acids or bases. These observations have significant consequences for the development of protic ionic liquid-based fuel cells; addition of corrosive acids or volatile bases to PIL electrolytes could counter some of the reasons for using an ionic liquid-based electrolyte in the first place and we discuss these implications.
The dashed, red, and blue lines in Figure 1A show cyclic voltammograms (CVs) of [ [TfO] containing 0.1 mol dm −3 of the parent triflic acid (TfOH), respectively. The reference electrode in each case was a Pd/H reference electrode (see Experimental Section for details). [8] Comparison of the dashed and red lines shows that the oxygen reduction reaction (ORR) in [dema] [TfO] results in a large cathodic wave with an onset potential of about 0.4 V. The limiting current density, j L , is about 3.5 A cm −2 and corresponds to the formation of H 2 O (see ref. [9]). In the acidified PIL, the entire ORR wave appears at a more positive potential and has an onset potential of about 1.1 V. j L in the pure and acidified PILs is similar, indicating that the ORR in It has been proposed previously that protic ionic liquids (PILs) such as diethylmethylammonium triflate could be used as the electrolytes in nonhumidified, intermediate temperature H 2 fuel cells, potentially offering the prospect of high conductivity and performance, even under anhydrous conditions. In this contribution, a combination of electroanalytical chemistry and fuel-cell polarization analyses is used to demonstrate for the first time that the pure PILs cannot support proton shuttling between the electrodes of fuel cells. Only through the inclusion of dissolved acidic or basic proton shuttles can viable protic ionic fuel cells be fabricated, which has major consequences for the use of these neoteric electrolytes in fuel cells.

Fuel Cells
While low-temperature proton exchange membrane fuel cells are promising for vehicular transport, the cost, performance, and stability of the required Pt-based electrocatalysts are major commercial stumbling blocks. [1] Moreover, the low operating temperature of these devices makes heat and water management significant challenges. The development of "intermediate temperature" fuel cells that operate in the range 100-300 °C offers several potential advantages over low-temperature cells, including the use of nonprecious metal electrocatalysts, simplification of heat and water management, and improved tolerance to fuel contaminants such as CO and SO 2 . [2] The most common class of electrolyte membranes for fuel cells operating at >100 °C are H 3 PO 4 -doped polybenzimidazole (PBI) membranes, with proton conductivities of about 0.1 S cm −1 . [3] However, leaching of H 3 PO 4 from the membranes causes a drastic drop in membrane proton conductivity, and degradation of the polymer membranes can occur due to the attack by reactive oxygen species such as H 2 O 2 , •OH, and •OOH. [4] The limitations of electrolytes such as H 3 PO 4 -doped PBI have led to the use of protic ionic liquids (PILs) as electrolytes for intermediate temperature fuel cells, which was pioneered most notably by Watanabe and co-workers. [5] PILs can be formed by transferring protons from a range of Brønsted acids www.advenergymat.de www.advancedsciencenews.com the acidified PIL also yields H 2 O. The positive shift of the ORR potential upon acidification of the PIL can be explained by considering the proton donor taking part in the ORR in each case. In pure [dema][TfO], the only protons available to take part in the reaction are those on the PIL cations (pK a 10.6 [10] ), yielding dema (Equation (2)), whereas the ORR at positive potentials in the acidified PIL involves protons from the added TfOH (pK a −14 [10] ) To explore whether the ORR potential depends generally on the pK a of the added protic species, voltammograms of samples of O 2 -saturated [dema][TfO] containing various acids (including acetic acid, trifluoroacetic acid, and perchloric acid) were recorded ( Figure 1B). While a clear j L cannot be identified in all cases, j L generally reaches ≈3.5 mA cm −2 in the presence of each acid, indicating that the product is H 2 O in each case. The ORR half-wave potential, E 1/2 , shifts positive by about 50 mV per aqueous pK a unit in the range -3 < pK a < 10 ( Figure 1C), which is close to that expected for the ORR involving a generalized proton donor and to that observed previously using an aprotic ionic liquid. [10] The deviation from linearity at pK a < -3 can be attributed to solvent leveling, wherein the supporting PIL acts as a buffer. [10] To estimate the impact of protic additives on operation of  [8] H 2 TfO 2TfOH 2e The HOR proceeds according to Equation (5) in acidified and pure [dema] [TfO], as the only sites available for occupation by electrogenerated protons are on the [TfO]anions ( Figure S3A, Supporting Information). The ORR in the acidified PIL proceeds according to Equation (4). Therefore, the net reaction in a fuel cell containing the acidified PIL should be the reaction between H 2 and O 2 to yield H 2 O. In this case, H + ions should be shuttled between the anode and cathode of the fuel cell via the TfOH/[TfO]couple ( Figure 2B). As shown by Figure 2A, the difference between the onset potentials of the ORR and the HOR is close to 0.9 V, which should correspond to an open-circuit voltage of 0.9 V in a fuel cell (as observed for conventional [1b] and PIL-based [5e] H 2 fuel cells). Figure 2C shows  (1)) at a significantly lower onset potential (-0.5 V) than in the acidified PIL. The ORR proceeds according to Equation (2) Figure S3B, Supporting Information). Therefore, the net reaction in a fuel cell containing the PIL with added base should also be oxidation of H 2 to yield H 2 O, and the H + ions should be shuttled between the anode and cathode via the dema/[dema] + couple ( Figure 2D). As shown by the arrow in Figure 2C, the difference between the onset potentials of the HOR and ORR is also about 0.9 V.
Finally, we compare the HOR and ORR potentials in pure [dema][TfO] ( Figure 2E). The HOR-onset potential is about 0.05 V, and must involve concomitant protonation of the [TfO]ions (Equation (5) That is, the simultaneous HOR and ORR in the PIL could only involve concomitant formation of the parent acid and base from the PIL, a highly endergonic process that offsets the exergonic Adv. Energy Mater. 2019, 9,1900744  [TfO] containing (from left to right) no added acid, acetic acid, formic acid, trifluoroacetic acid, sulfuric acid, perchloric acid, and triflic acid. All voltammograms were recorded using a Pt electrode rotating at 1600 rpm and at a scan rate of 100 mV s −1 . The cell temperature was 22 °C. The additive concentration was 0.1 mol dm −3 in each case and all potential sweeps began in the negative direction. C) The ORR half-wave potential versus the aqueous pK a of (from right to left) [dema][TfO], acetic acid, formic acid, trifluoroacetic acid, sulfuric acid, perchloric acid, and triflic acid. reaction between H 2 and O 2 (in which ∆G ⦵ = −237 kJ mol− 1 ). [11] This loss of driving force is illustrated by the almostoverlapping HOR-and ORR-onset potentials in Figure 2E. This analysis was also carried out using the structurally similar PIL diethylmethylammonium bis[trifluoromethanesulfonyl]imide, and the same behavior was observed ( Figure S4, Supporting Information). Therefore, the PIL electrolyte of the fuel cell should contain either acidic or basic additives to shuttle protons between the electrodes and allow current flow between the anode and cathode. In the absence of a significant concentration of an added proton shuttle, sustainable current should not flow ( Figure 2F). The loss of driving force for current flow would prevent the build-up of a significant concentration of the required proton shuttle (as suggested by Equations (1) and (2)). To estimate the concentration of TfOH required to effect fast ORR electrochemistry, we changed the concentration of TfOH in [dema][TfO] and recorded a series of ORR voltammograms. The ORR reaches its mass-transport-controlled rate at a TfOH concentration of 88 mmol dm −3 ( Figure S5, Supporting Information), which is almost two orders of magnitude higher than the saturation concentration of O 2 in [dema][TfO]. [12] To demonstrate that the effects observed during voltammetry of the PILs translate to the operation of fuel cells, we constructed a series of Grove-type fuel cells containing Pt wires immersed in [dema][TfO] and over which H 2 and O 2 , respectively, were flowing. [5e] The red line in Figure 3A shows the polarization curve recorded using a cell containing acidified [dema] [TfO] and operating at room temperature. The general shape of the curve is similar to that recorded using conventional fuel cells, [1b]   approximately linearly as the current density increases due to Ohmic losses within the cell and, above 0.5 mA cm −2 , the voltage drops significantly due to mass-transport losses within the cell. [13] The blue line in Figure 3A shows a similar polarization curve recorded using a cell containing 0.1 mol dm −3 dema in [dema] [TfO]. The open-circuit voltage is similar to that observed using the acidified electrolyte and the general shape of the curve is similar to that recorded when using the acidified electrolyte, demonstrating that proton transport between the anode and cathode is possible. In contrast, when the electrolyte is pure [dema][TfO] (green line in Figure 3A), the cell voltage drops drastically as soon as current starts to flow and decreases to 0.0 V at about 0.1 mA cm −2 , demonstrating that the pure PIL could not shuttle protons between the electrodes of the cell effectively, in agreement with the behavior predicted using analytical voltammetry.
Finally, we consider why others have fabricated functioning cells that contained [dema][TfO] as electrolytes. [5h] We hypothesize that this may be due to the heating that PILs have undergone during synthesis, purification, or operation of PIL-based fuel cells. Heating PILs can increase the equilibrium concentration of parent acids and bases in PILs, and even loss of the most volatile component yielding nonstoichiometric PILs. [14] Such effects could potentially provide the shuttles required to transport protons between the anode and cathode of fuel cells. Indeed, a drastic increase in the ORR-onset potential in [dema][TfO] as the PIL temperature was raised from 25 to 160 °C was demonstrated previously, indicating that the ORR proceeds via different routes at high and low temperatures. [15] To test this hypothesis further, we heated a sample of [dema] [TfO] at 150 °C for 64 h and then allowed it to cool down to room temperature. Figure 3B,C shows photographs of the PIL before and after heating. The dark color of the sample that had been heated indicates that some decomposition of the liquid had occurred during heating, a phenomenon that we tentatively attribute to decomposition of dema formed at the elevated temperature. Electroanalysis of the sample that had been heated and then cooled reveals that the ORR potential shifted positive ( Figure S6, Supporting Information), indicating that a protic contaminant had formed irreversibly in the PIL, changing the mechanism by which O 2 is reduced and effecting the ORR at higher potentials. Therefore, it appears that the formation of nonstoichiometric PILs at elevated temperatures (potentially followed by decomposition and/or loss of the most volatile component) may be responsible for the positive results obtained previously using PIL-based fuel cells. This phenomenon may have been overlooked as the detection of neutral parent species in PILs (or determining the extent of "ionicity" of the liquids) is difficult and usually relies on conductimetric measurements, [16] due to the increase in conductivity caused by addition of neutral species to PILs. [17] As dissolved proton shuttles play such a key role in these devices, we believe they cannot be overlooked any more and that future research in this area should address the optimization of such systems. For example, a potential route to the formation of highly nonstoichiometric PILs that can support proton transport between anodes and cathodes in fuel cells is to use PILs in which the equilibrium composition favors a high concentration of the parent species. The equilibrium composition is related to the difference in the aqueous pK a of the acids and bases, and it is generally accepted that a "∆pK a " > 15 results in effectively complete proton transfer from the acids to bases. [18] It will be interesting to investigate whether the use of PILs in which ∆pK a is significantly less than 15 could provide the species required to transport protons in PIL-based fuel cells. Such cells could potentially even work efficiently at low temperatures. Of course, as ∆pK a and the ionicity of the PIL decreases, the liquid would become more acidic and its vapor pressure would increase, countering some of the reasons for using PIL electrolytes in the first place and raising concerns of cell corrosion and volatilization of essential electrolyte components, and potentially limiting cell lifetimes. [19] The deliberate addition of dissolved proton shuttles to highly-ionic PILs, while offering a route to effective proton shuttling within such cells, raises similar concerns. We believe that the solution lies in the identification of ionic liquid compositions that can support relatively high concentrations of dissolved proton shuttles (either added to the liquid or as part of the inherent equilibrium composition of the liquid) and high conductivity, but in which the liquids remain noncorrosive and nonvolatile.
In summary, we have described the impact of protic additives on the electrochemical behavior of PILs for use as electrolytes in intermediate temperature fuel cells. Electrochemically stable PILs must contain dissolved proton shuttles to effectively carry protons between the anode and cathode of PIL-based fuel cells. The use of such compositions raises new challenges that must be addressed, including potential loss of proton-conducting additives by evaporation from the electrolytes and cell corrosion. We believe that the challenge is to identify compositions that can retain the advantages offered by ionic liquid electrolytes (nonvolatility and noncorrosivity) while supporting facile proton transport between the anode and cathode of fuel cells.
[dema] [TfO] was prepared by the dropwise addition of 1 mol dm −3 aqueous triflic acid to 1 mol dm −3 aqueous N,N-diethylmethylamine with a 5% molar excess of the base. The reaction mixture was cooled in an ice bath. Excess water was removed by rotary evaporation. The product was then either held under vacuum (5 × 10 −2 mbar) at room temperature for 72 h and then at 70 °C for 24 h, or held under vacuum (5 × 10 −2 mbar) at 40 °C for 1 week, to remove residual water and base. The liquid was then transferred under vacuum to a glovebox containing a dry N 2 atmosphere (<0.1 ppm O 2 and <0.4 ppm H 2 O). Attenuated total reflection Fourier-transform infrared spectroscopy was carried out using a Bruker Alpha Platinum-ATR FTIR spectrometer at 30 °C ( Figure S7, Supporting Information). The liquid was characterized by 1 H, 13 C, and 19 F NMR spectroscopy (Bruker Ascend 400 MHz NMR Spectrometer) using a hexadeuterodimethyl sulfoxide solvent ( Figures S8-S10, Supporting Information). Thermogravimetric analysis was carried out using a TA Instruments Discovery instrument under a N 2 atmosphere, and residual water contents were determined using Karl-Fischer titrations (Mitsubishi CA-200 Moisturemeter). Residual water contents were < 100 ppm.
Electrochemical Measurements: The use of Pd/H reference electrodes for electrochemistry in PILs was introduced by Angell. [8] Pd/H electrodes provided stable potentials in PILs, which were defined by the hydrogen redox equilibrium at the interface between Pd and the PIL. Our reference electrode was fabricated by sealing a 0.5 mm diameter Pd wire (≈2.7 cm long) into borosilicate glass using a butane flame. Approximately 1.5 cm of the Pd wire was left protruding from the glass sheath to form a Pd rod electrode. A polished Ag contact wire was attached to the Pd wire inside the glass sheath using Ag epoxy (Chemtronics, Kennesaw, GA, USA). The Pd rod was then annealed in a butane flame and immersed in water before H 2 was bubbled over the Pd metal surface for 30 min. The rod was then rinsed with deionized water and either dried under a flow of N 2 or Ar or left covered in air until dry.
A three-electrode electrochemical cell containing a 4.6 mm diameter Pt disk working electrode, a Pt flag counter electrode (A ≈ 0.70 cm 2 ), and a Pd/H reference electrode was used for all electrochemical measurements, unless otherwise stated. Prior to use, the Pt electrode was polished using an aqueous suspension of 0.05 µm alumina (Buehler, Lake Bluff, Illinois) on felt polishing pads and rinsed thoroughly with deionized water. The Pt flag counter electrode was flame annealed in a butane flame before being rinsed with deionized water. All working and counter electrodes were dried in a glassware oven before use. The cell was charged with 5 mL of [dema][TfO] and bubbled with O 2 , H 2 , or Ar through a needle for a minimum of 40 min, while being stirred by a magnetic stirrer. During voltammetric measurements, a blanket of the appropriate gas was maintained above the PIL. All experiments were carried out at room temperature (22 °C). The effect of heating the PIL on the electrochemical behavior of the liquid was examined by heating a sample of [dema][TfO] to 150 °C in air for 64 hours using a heating block. After heating, the sample was left to cool to room temperature and was re-saturated with O 2 by bubbling with the gas for 40 min, while the liquids were stirred.
To mitigate against any changes in the potential of the reference electrode upon addition of acid, base, or gases to the PILs, potentials were adjusted such that the electrochemical window was the same during each analysis. Fuel-cell polarization curves were recorded using a H-cell from Pine Research Instrumentation, Inc (Durham, NC, USA), in which a glass frit separated the two chambers. Each chamber of the cell was filled with 8 mL of [dema] [TfO]. H 2 and O 2 were flowed over the negative and positive electrodes, respectively, and each side of the cell was stirred at 650 rpm during measurements. 0.5 mm diameter Pt wire electrodes were used as the anode and cathode. The current was increased from 0.0 A at a rate of 0.25 µA s −1 and cell voltage was recorded.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.