Enhancing Pt/C Catalysts for the Oxygen Reduction Reaction with Protic Ionic Liquids: The Effect of Anion Structure

Protic ionic liquids (ILs) have been recently studied as a potential approach to enhance oxygen reduction reaction (ORR) activity of carbon supported platinum catalysts (Pt/C) for application in polymer electrolyte membrane fuel cells. The high oxygen solubility in the ILs was suggested as one of the main reasons for the accelerated reaction rates. Because the nature of the anion of the IL has been associated with increased oxygen solubility, in this work we survey a number of ionic liquids with various anions to study this effect. While the search for direct correlation between the ORR activities and the oxygen solubilities does not produce any conclusive results, by contrast, the specific activity showed dependence on the availability of oxygenated species free Pt sites. This finding indicates that the inhibition of Pt oxidation and less adsorption of non-reactive species may also play an important role in the enhanced ORR activity. Moreover, the degree of IL coverage on the Pt surface was estimated using (bi)sulfate ions as an indicator. The surface coverage not only affected the ORR activity, but also the Pt dissolution process. This suggests that an optimal balance between activity and stability can be achieved on a partially covered Pt surface. © The Author(s) 2017. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.1071713jes] All rights reserved.

The oxygen reduction reaction (ORR) that occurs on the cathode poses a major hurdle for the efficient utilization of polymer electrolyte fuel cells (PEMFC). Extensive studies have focused on developing novel catalysts to improve the efficiency of this reaction. 1,2 The ORR involves multiple steps, among which O 2 + H + + e − ↔ OOH ads and OH ads + H + + e − ↔ H 2 O act as the potential rate determining steps. Simultaneously optimizing the Gibbs free energies for both steps to completely eliminate the overpotential is very challenging, and a ∼350 mV overpotenial is commonly observed, which is independent of the identity of the catalyst. 3,4 The large overpotential has been generally attributed to the sluggish kinetics of the ORR and adsorption of oxygenated species (e.g. OH ad ) or other anions. 5 The coverage of the adsorbed oxygenated species (θ OH ad ) is unfavorable for the reaction, and the availability of the free metal surface (as expressed by the (1-θ OH ad ) term) is one of the governing factors for the ORR activity. 6,7 As a result, much effort has been dedicated to weakening the bonding of OH to the catalyst surface either by shifting the d-band center of Pt 8,9 or by lateral repulsion from the supports (e.g. metal oxides). 10,11 Anion adsorption on Pt also affects the ORR activity, and it is agreed that the occupation of active sites deactivates the sites and reduces the activity. 12 In real-world membrane-electrode assembly (MEA), the Nafion membrane and ionomer, constituted by a Teflon-like backbone and an anionic cluster of sulfonic groups, 13 are widely employed as indispensable components. The interface between the Nafion and the metal has received great attention due to the strong irreversible sulfonate anions adsorption on the Pt. 12,[14][15][16] The degree of sensitivity to the anions highly depends on the Pt facets. 12 For example, the facet Pt(111) has been reported to be most sensitive to the anion adsorption. 17,18 To accommodate operation of PEMFC under low humidity conditions, high ionomer to carbon ratios have been adopted to ensure sufficient proton transport, 19 but the excessive sulfonic groups might reduce the catalyst performance.
In addition to the adsorption of non-reactive species, the substantial O 2 diffusion resistance through the interface of ionomer and Pt was recently recognized and extensively investigated. [20][21][22] Greszler et al. 23 found that the O 2 transport resistance in the thin ionomer film was much higher than in its bulk form. A DFT simulation also revealed a quick drop of O 2 concentration when close to the Nafion/Pt interface. 24 Although still a subject of active investigation, recent studies have pointed out that the reduced degree of freedom due to the adsorption of sulfonic acid and the low O 2 solubility are the main contributors to the diffusion resistance. 22 Ionic liquids (ILs) have been studied as a non-aqueous reaction media for electrochemical applications due to their low volatility, wide electrochemical windows, good ionic conductivity, and good chemical stability. [25][26][27][28][29] Quite recently, high O 2 solubility in some ILs has been utilized to address the issues of insufficient O 2 concentration near the Pt surface. Snyder et al. pioneered the studies of using ILs as an ORR promoter. 30,31 A protic IL, [MTBD][beti], was encapsulated in porous PtNi nanoparticles, which showed improvements in both mass activity (mA ug Pt −1 ) and specific activity (mA cm Pt −2 ) by ∼50% in the kinetically controlled regime. The higher O 2 solubility in the IL than in aqueous electrolyte was suggested as one of the major causes for the performance enhancement. Zhang et al. also reported a simple IL impregnation method to modify a carbon supported Pt electrocatalyst. 32 It was found that by filling catalyst pores with IL, the specific activity could be boosted. Higher O 2 solubility was also considered as a contributor.
In one of the previous publications, the anion structure of the IL was reported to be associated with the O 2 solubility. 33 Therefore, in this work a number of ILs with various anion structures and O 2 solubility were synthesized and studied in order to test the hypothesis that O 2 solubility, as controlled by the choice of anion, determines the ORR activity. The chemical structures of ILs are displayed in Fig. 1. In order to understand how much the O 2 solubility affects the ORR activity, a series of IL impregnated Pt/C catalysts were constructed and tested. In addition, adsorbed oxygenated species (θ OH ad ), IL coverage on the Pt/C, and catalyst durability were investigated.
Similar synthesis routes and purification processes were used to prepare the rest of the ILs: 1-methyl-2, 3,4,6,7,8- Impregnation of IL into Pt/C catalysts.-A modified impregnation method was adopted. 32 Typically, IL stock solutions with a concentration of 5 mg IL in 1 mL of isopropanol were prepared. 9.5 mg Pt/C catalyst was placed in a vial and first wet by 0.5 ml DI water, followed by adding the desired amount of IL stock solution to produce IL to carbon weight ratios (IL/C) of 0, 1.28, 2.56, and 3.84. The mixture was ultrasonicated for 20 min, and the solvent was slowly evaporated at 45 • C into the ambient atmosphere. Finally, the obtained powder was further dried under high vacuum (−1 bar, room temperature) overnight. O 2 solubility and diffusion coefficient measurements.-A gravimetric method was used to determine O 2 solubility in the ILs. The sample weight change during O 2 ab/desorption was monitored at a fixed temperature and pressure in a magnetic suspension balance (MSB) (Rubotherm GmbH). Initially, about 1 g of IL sample was added to the sample bucket and degassed to c.a. 10 −5 bar. After evaporating the water and other volatile impurities, the chamber was pressurized with O 2 to a certain pressure. The vapor-liquid equilibrium between O 2 and the IL sample was reached and confirmed by constantly weighing for at least 20 mins, and an absorption isotherm was recorded. The pressure in the sample chamber was then decreased in a stepwise manner to obtain a desorption isotherm. The absorption and desorption isotherms constituted a full isotherm. The values were corrected for buoyancy. The O 2 diffusion coefficient was obtained from a time-dependent absorption profile, and the O 2 absorption was mathematically modeled by a simplified mass diffusion process. 34,35 Electrochemical measurements.-All electrochemical measurements were performed on an EC-LAB SP-300 (BioLogic) and the electrolytes were prepared with Millipore (Milli-Q Synthesis) water with a resistance greater than 18.2 M · cm. A Pt wire (Pine) and a hydrogen reference electrode (Hydroflex,eDAQ) were used as counter electrodes and reference electrode, respectively. All potentials in this work are reported against a reversible hydrogen electrode (RHE) unless specified otherwise. 8 mL DI water, 4.5 mL Isopropanol and 50 ul 5% Nafion dispersion (D520, Ion Power) were added into the vial containing the prepared Pt/C-IL. The catalyst ink was subjected to 15 min ultrasonication in an ice bath. A 10 ul aliquot of this ink was pipetted onto the glassy carbon (GC) disk (5 mm diameter, Pine) and rotationally dried in air to form a uniform catalyst layer. 36 For the control sample, a Pt/C catalyst layer were prepared in exactly the same manner as the Pt/C-IL catalyst, except without any IL. The calculated final Pt loading on the disk is 18 ug/cm 2 disk . The catalyst coated GC was pre-conditioned in N 2 -saturated 0.1 M HClO 4 . Typically, the working disk was scanned from 0.05 V to 1.2 V with a scan rate of 100 mV s −1 until the cyclic voltammetry (CV) curve didn't change. After that, a stable CV was recorded from 0.05 to 1.2 V at 50 mV s −1 . To assure the ionic liquid in Pt/C-IL is equilibrated with electrolyte and to avoid any unwetted portion, CV before ORR and after ORR was compared. It is found that the change is insignificant and pre-condition is sufficient to recover the electrochemical surface area (ECA) (Fig. S1). ECA was obtained by integrating the charge under the hydrogen adsorption region (H upd ), assuming 210 uC cm Pt −2 . Electrochemical impedance spectroscopy (EIS) was performed at 0.45 V in N 2 -saturted HClO 4 to measure the ohmic resistance and protonic resistance. 36 The coverage of OH ad on Pt surface (θ OH ad ) at 0.9 V was calculated using an equation suggested by Zhang et al. 37,38 where Q OH ad is the oxidative charge builds from at 0.9 V and Q H upd is the charge under hydrogen adsorption region. The ORR measurements were carried out in O 2 -saturated 0.1 M HClO 4 . Multiple rotation rates were chosen and the Koutecky-Levich equation was used to determine the kinetic current (i k ). All the calculations were IR compensated and N 2 -background corrected, to fully reveal the i k . 36 At least 4 independent data sets were collected to ensure the reproducibility. Finally, the Pt/C-ILs were cycled for 5000 times in N 2 -saturated 0.1 M HClO 4 in a potential range of 0.6 V to 1.0 V at a scan rate of 50 mV s −1 , a protocol suggested by U.S. Department of Energy (DOE) to assess the Pt dissolution performance. Any changes in the CV and ORR from the initial samples to after cycling were evaluated by repeating the procedures described above.
The Pt/C-ILs were also studied in 0.5 M H 2 SO 4 to investigate the IL coverage on the Pt surface. The characteristic hydrogen adsorption peak on Pt(110) and Pt (100) in the presence of (bi)sulfate ions was used as a direct indictor to qualitatively estimate the IL coverage. To study the anion adsorption effect of [MTBD][C 4 F 9 SO 3 ], KC 4 F 9 SO 3 salt was used to simulate the free anion [C 4 F 9 SO 3 ] − . 0 mM, 0.001 mM, 0.01 mM, 0.1 mM and 1 mM KC 4 F 9 SO 3 in 0.1 M HClO 4 and 0.5 M H 2 SO 4 were prepared, respectively. CV and ORR were examined at each KC 4 F 9 SO 3 concentration.
CO displacement was conducted in 0.1 M HClO 4 with potential holding at 0.4 V, followed by CO stripping. The anion adsorption and CO stripping process were investigated. [39][40][41]

Results and Discussion
First we present the solubility and diffusivity of O 2 in the ILs. This is followed by electrochemical characterization of the Pt/C-ILs for all five ILs, as well as more detailed experiments with the most promising of the ILs, The NMR spectra shown in Fig. 2  Since the solubility of O 2 increases linearly with pressure in the ILs, Henry's Law was used to analyze the dissolution behavior of O 2 and the Henry's Law Constant (H) was used to describe the gas solubility in the ILs. O 2 solubility was measured at room temperature and pressures up to 45 bar. Fig. 3a gives an example of the O 2 solubility measurements in [MTBD][TFSI]. As mentioned above, the relationship between pressure and mole fraction is very linear, so the O 2 mole fraction at 1 bar can be accurately extrapolated from these data.
A one-dimensional mass diffusion model was employed to calculate the diffusion coefficient (D), 34,35 which was mathematically expressed as, Initial condition: C = C 0 , when t = 0, and 0 < z < L Boundary condition: C = C s , when t > 0, and z = 0, and ∂C ∂z = o where C is the O 2 concentration in IL; C 0 is the initial homogeneous O 2 concentration; C s is the O 2 saturation concentration; L is the depth of ionic liquid in the sample container; and D is the diffusion coefficient, which is assumed to be a constant. The solution for the space averaged O 2 concentration ( C ) is shown in Equation 2, and this equation was used for fitting the experimental data. The experimental and fitting results for [MTBD][TFSI] are shown in Fig. 3b. Clearly, the data is fit very well by the model.
where λ n = (n+ 1 2 )π L . The O 2 diffusion coefficients reported here are at 30 bar rather than 1 bar because significantly smaller uncertainty can be achieved at high pressure where the oxygen solubility is higher. This is a reasonable approach since pressure has little effect on the O 2 diffusion coefficient when O 2 is sparingly soluble, as is the case with ionic liquids. This is because there are no concerns that the viscosity of the O 2 -IL mixture would decrease dramatically with increasing O 2 partial pressure. 42,43 Because  Table I.
Where X O 2 is O 2 solubility in the IL mixture, X 1 and X 2 are mole fraction of individual ILs in the mixture, and X O 2 ,1 and X O 2 ,1 are the ) unless CC License in place (see abstract   Fig. 4a contains the cyclic voltammograms. It was found that the Pt/C-ILs suffered ECA loss compared to pristine Pt/C. The ECA loss for most ILs was moderate (10-13%), except for [MTBD][C 4 F 9 SO 3 ], which showed a value of 28% (Fig. 4c). The differences in the ECA behavior might be due to the competition adsorption of ions in IL with hydrogen. Meanwhile, the onset potentials of Pt-oxides formation of Pt/C-ILs were positively shifted, in agreement with observation of suppression of Pt oxidation. 31 Where n is number of electrons transferred, F is Faraday's constant, K is a chemical rate constant, (1-θ) is the available surface, β is a symmetry factor, E is the applied potential, and ω is the energy parameter for the Temkin isotherm. Equation 4 illustrated that the current density should be proportional to the oxygen concentration ([O 2 ]), and a higher oxygen solubility in the ILs is expected to yield a proportional increase in the current density. The solubility study showed that the average oxygen concentration in ILs was improved by a factor of ∼5-6 over aqueous 0.1 M HClO 4 solution. Unfortunately, neither the mass activity nor the specific activity show a six-fold improvement (Fig. 4d). For example, the highest mass activity of Pt/C-IL at 0.9 V is ∼250 mA/mg Pt , an increase of just ∼27% compared to the Pt/C baseline. The rate of the heterogeneous reaction occurring at the Pt-electrolyte interface can be expressed as is the O 2 mole flux and j is the current density. 48   [beti] did not increase by a factor of 5. Thus, it appears that slow O 2 diffusion through the ILs and long mean free path within the ILs counteract the high O 2 concentration benefit at the electrolyte/IL interface and limit the O 2 concentration at the Pt interface. Unlike the nanoporous PtNi nanoparticle that can encapsulate the IL inside the particles, the selected Pt/C platform in this work has no inner space to accommodate IL. As a consequence, the IL would only form a thin layer covering the catalyst surface. (Fig. 5) The difference of collision frequency of oxygen with inner wall of nanoporous Pt and with flat Pt might be one of the main reasons that we did not observe the proportional increase of current density. Even though the effect of high O 2 solubility in the ILs on the ORR activity was not significant in the  selected Pt/C platform, its potential benefit should not be neglected in the specific catalyst structure. 50 Besides, ILs with wider oxygen solubility window may be more helpful to better assess the effect of O 2 solubility.
Interestingly, the specific activity (SA, mA cm Pt −2 ) correlated to the activity per available Pt site. Motivated by the fact that comparable mass activities were achieved with less accessible Pt surface and less Pt-oxides formation. The specific activity was plotted against the fraction of oxides free Pt sites (1-θ OH ), which is shown in Fig. 4e. The specific activity clearly depends on (1-θ OH ) and a linear relationship can be seen. The θ OH of most Pt/C-ILs are between 0.4∼0.5, except for Pt/C-[MTBD][C 4 F 9 SO 3 ], which has significantly lower coverage by oxides. As mentioned above, previous studies mainly attributed the promotion of ORR activity to the higher O 2 concentration in ILs  The red dashed line drawn on the image is the limit of the carbon boundary. and the increased attempt frequencies, 30-32 but Zhang et al. recently suggested a different mechanism. They integrated Pt/C with two ILs with different O 2 solubility and found that the one with higher O 2 solubility actually did not outperform the other one, and the performance differences were very small. 38 Their study suggested that the ORR is not sensitive to the O 2 solubility, but that suppression of nonreactive species (e.g. Pt-oxides) is what contributes to the activity enhancement in the presence of ILs. We hypothesize that the IL providing a lateral repulsion of oxygenated species and weakening of the -OH adsorption is the mechanism for the IL coating improving specific activity. CO displacement and CO stripping results also shows a reduced anion adsorption on Pt and a delay of Pt-OH formation (Fig. S2).
Since the Pt/C modified with [MTBD][C 4 F 9 SO 3 ] showed the highest specific activity, further investigations were conducted with this IL. The morphology of [MTBD][C 4 F 9 SO 3 ] on the Pt/C was studied by TEM and the image is shown in Fig. 5. A thin layer can be detected on the Pt/C. Owing to its amorphous contrast, this is evidence of the presence of IL. The IL formed a continuous layer (1∼2 nm) covering the catalyst surface. Fig. 6 displays the electrochemical characteri-zation of Pt/C-[MTBD][C 4 F 9 SO 3 ] with various IL/C ratios. One can easily see that the ECA decreases quickly with increasing IL/C ratio and a 40% loss is observed at IL/C = 3.86. Meanwhile, Fig. 6b shows that the onset potential of Pt-oxides formation was positively shifted and the resulting amount of oxides (Q OH ) under potential range (0.6 V -1.2 V) was suppressed accordingly. The loss of ECA and reduction of Pt-oxides formation will be comprehensively discussed below.
Subbaraman et al. coated planar Pt with a thin Nafion film and observed the formation of Pt oxides was hindered compared to the Nafion free Pt. 12 It was claimed that the adsorption of sulfonate anions was competing with adsorption of oxygen species on the Pt surface, resulting in a positive shift of the onset potential of Pt oxidation. Since the [MTBD][C 4 F 9 SO 3 ] contains a similar SO 3 − group, it's speculated that the anion of the IL might impose a similar effect. To elucidate this possibility, the anion precursor, K[C 4 F 9 SO 3 ], was considered to simulate the [C 4 F 9 SO 3 ] − adsorption on Pt/C. The [C 4 F 9 SO 3 ] − concentrations varied from 0 mM to 1 mM in 0.1 M HClO 4 . One has to mention that the total mole content of available [C 4 F 9 SO 3 ] − in the aqueous electrolyte solution is far higher than that in the catalyst layer when [MTBD][C 4 F 9 SO 3 ] is used (e.g. the mole content of [C 4 F 9 SO 3 ] − in a 1 mM aqueous solution is ∼3000 times higher than in the catalyst layer at IL/C = 3.86 when using [MTBD][C 4 F 9 SO 3 ]), ensuring an adequate anion supply for the adsorption process. Fig. 7 presents the voltammograms and polarization curves of Pt/C in the presence of various [C 4 F 9 SO 3 ] − concentrations. It was found that the changes of CV between 0 M and 1 mM [C 4 F 9 SO 3 ] − were insignificant, especially in the hydrogen under potential deposition (H upd ) and Pt redox regions. A slight negative shift of linear sweep voltammetry (LSV) was observed at 1 mM compared to 0 M. K[C 4 F 9 SO 3 ] generated a lot of foam in the electrolyte due to its the hydrophobic fluoroalkyl chain, and purging with N 2 or O 2 accelerated the bubble formation and bursting. The situation became even worse in rotational operation. The disruption of local gas transport near the rotating disk might be the cause of the small change of ECA and of LSV. Unlike polymers, it is anticipated that the anion and cation of the IL would populate the surface but be mobile to a certain degree. It is also possible that the ion orientations are changed by applying an electrical potential. 51 Therefore, during an anodic potential scan, the negatively charged anion might head toward the Pt and occupy the sites. However, the small changes of CV and ORR imply that the binding strength of [C 4 F 9 SO 3 ] − on the Pt surface is weak and would not affect the Pt electrochemical behavior very much. In summary, the anion adsorption study suggests that the loss of ECA and decrease of Ptoxides formation found in Pt/C-[MTBD][C 4 F 9 SO 3 ] is not due to anion occupation.  Electrochemical Impedance Spectroscopy (EIS) is commonly adopted to measure the effective proton conductivity within the catalyst layer in MEA. 52 While in the rotating disk electrode (RDE) setup, due to the thinness of the catalyst layer and the abundance of electrolyte, the associated protonic resistance has been largely neglected. Nonetheless, it is essential to study the proton diffusion in Pt/C-IL. Three electrodes with different catalyst layers on GC were fabricated according to the RDE preparation described in the Experimental section. The nafion and IL amounts are exactly the same as detailed in the Experimental section. One has to mention that the experiment was not designed to measure the proton conductivity of either nafion ionomer or individual ionic liquid, but the overall proton diffusion resistance within the catalyst layer. The three electrodes are as follows: (1) Nafion only; the catalyst layer consisted of Pt/C and Nafion only (Pt/C-Nafion); (2) IL only; the catalyst layer consisted of Pt/C and IL ([MTBD][C 4 F 9 SO 3 ]) only; and (3) Nafion-IL; the catalyst layer consisted of Pt/C, IL and Nafion (see Experimental section on Pt/C-IL ink preparation for details). Fig. 8a presents the Nyquist plots, and the intercept of the linear portion on the x-axis was extracted to calculate the effective catalyst layer protonic resistance, R H+ (the inset shows an example how to obtain R H+ /3; details can be found in Ref. 36). All the protonic resistance was normalized to the area of catalyst layer. The values extracted from the data for Pt/C-Nafion is 1.65 · cm 2 , which agrees well with the reported value. 36 Unlike the Nafion only catalyst layer, the EIS response of the IL only catalyst layer shows a sluggish increase at low frequency, implying an increase in the proton diffusion resistance. The resulting R H+ was 5.14 · cm 2 , which is much higher than the Pt/C-Nafion value. Interestingly, the Nafion-IL prepared catalyst layer showed a much smaller resistance c.a. 0.17 · cm 2 . In the IL modified membrane system, it has been reported that the proton generated from the polymer -SO 3 H group associates with the IL, and hopping and diffusion mechanisms carry the proton to the SO 3 − site of the IL. [53][54][55] In this study, the proton may follow a similar transport mechanism. While the exact mechanism of reduced R H+ is still under investigation, we hypothesize that a more uniform catalyst layer in the co-presence of IL and Nafion may contribute to a decrease in proton diffusion resistance. EIS was also performed at various IL/C ratios and the results are summarized in Fig. 8b. It was found that in the IL/C range studied, the additional IL does not restrict sufficient proton supply within the catalyst layer.
In real MEA, the Nafion ionomer distribution and its coverage on the Pt surface largely determines the fuel cell performance, because the transport of O 2 and protons is controlled by the thickness of the ionomer. 19,56 Also, the carbon structures greatly affect the ionomer dispersion on the catalyst surface. 57,58 In an attempt to estimate IL coverage on the catalyst surface, we note that hydrogen adsorption behavior on low index Pt(hlk) facets differs dramatically in HClO 4 and H 2 SO 4 solutions, such as the Pt(100) and Pt(110). 6,17,59,60 With the aid of sensitivity of hydrogen adsorption region, we designed a method to study the IL coverage on the Pt. Fig. 9a showed the CV behaviors of Pt/C in 0.1 M HClO 4 and 0.5 M H 2 SO 4 aqueous solutions, respectively. Pt/C in H 2 SO 4 showed two sharper peaks in the H upd region, which we assign to the signature hydrogen adsorption on Pt(100) and Pt(110), respectively. In contrast, the Pt(110) adsorption peak was less sharp and Pt(100) became more broader in HClO 4 . The different hydrogen adsorption on low index Pt facets behavior was mainly due to the attraction of (bi)sulfate anion into the diffusion layer below 0.4 V and its coupling effect with hydrogen for adsorption. 59 Meanwhile, the onset potential of Pt oxidation shifted positively in H 2 SO 4 relative to in HClO 4 (Figs. 9a and 9b), indicating a competition mechanism between the adsorption of oxygenated species and (bi)sulfate ions. 60,61 For the Pt/C-[MTBD][C 4 F 9 SO 3 ] (IL/C = 2.56) catalyst (Fig. 9b), the H upd regions were virtually identical in the two acids. The sharp adsorption peak on Pt(110) was absent, while peak for adsorption on Pt (100) remained broad. The peak for Pt(110) was shifted slightly to a higher potential, which is due to a pH effect. 62 This finding implies that the IL might be preventing Pt from having direct contact with the (bi)sulfate ions. Taking advantage of the sensitivity of  To explore the additional loss of ECA of Pt/C-[MTBD][C 4 F 9 SO 3 ] in H 2 SO 4 , same anion adsorption studies using K[C 4 F 9 SO 3 ] were conducted and these are shown in Fig. 9e. Throughout the K[C 4 F 9 SO 3 ] concentration range studied, the changes in the CV curves in 0.5 M H 2 SO 4 were clearly different from those observed in 0.1 M HClO 4 . The ECA loss was much larger (21% in H 2 SO 4 vs. 8% in HClO 4 at 1 mM), and the decrease of Pt oxidation/Pt-oxides reduction region was more noticeable. All these findings provide evidence that the co-presence of [C 4 F 9 SO 3 ] − and (bi)sulfate affect the processes of hydrogen adsorption on Pt and Pt-oxides formation. This suggests that between IL/C = 1.28 and 2.56, Pt was partially covered by IL but it was still accessible to the (bi)sulfate ions. Increasing the IL/C ratio to 3.84, the ECAs and Pt-oxide formation were identical in both electrolytes, implying that contact between the Pt and the (bi)sulfate ions was limited. In summary, with the aid of the different CV behaviors in HClO 4 and H 2 SO 4 we are able to roughly estimate the IL coverage on Pt. At low IL/C ratios, Pt is partially covered by IL. Further increasing the IL/C ratio to 3.84 or beyond yields full coverage on the Pt surface.
We have shown that the IL loading effect on CV (Fig. 6) is closely associated with surface coverage. Therefore, the influence of IL loading on the ORR activity was further studied. The polarization curves are displayed in Fig. 10a. It was found that the half-wave potential of Pt/C-[MTBD][C 4 F 9 SO 3 ] shifted positively by ∼11 mV at IL/C = 1.28 and ∼20 mV at IL/C = 2.56 compared to the pristine Pt/C, indicating an enhanced ORR performance. However, excessive IL content caused a negative shift of half-wave potentials. Correspondingly, a volcano dependence of ORR activity on IL/C was revealed (Fig. 10c), that is, MA and SA initially increased with increasing IL content followed by a decrease. As discussed in Fig. 4, the (1-θ) heavily influenced the specific activity; therefore, the SA was plotted against the (1-θ) and this is illustrated in Fig. 10d. A similar volcano-dependence of SA on (1-θ) was observed. Between IL/C = 0 and IL/C = 2.56, SA increased with a decrease in θ OH ad . The IL coverage study revealed that the Pt is partially covered for those IL/C ratios, and a triple solid/liquid/gas (Pt/electrolyte/O 2 ) reaction phase was maintained. Even though the Pt/C-[MTBD][C 4 F 9 SO 3 ] suffered a loss of active surface area (Fig.  9d), the protection of the surface from nonreactive species by the IL leads to an increase in SA. The mass activity is a product of ECA and SA, and the beneficial increase in SA is offset somewhat by the decrease in ECA. Consequently, the gain of mass activity was limited to ∼50 mA mg −1 Pt . Tafel plots (Fig. 10b) also indicate that in the kinetic region of 0.9 V or above, the differences in intrinsic activity (j k ) between IL/C up to 2.56 were small. When the Pt surface was fully coated by IL (IL/C = 3.56 or above), the slow O 2 diffusion more than offsets the suppression of nonreactive species and this becomes the dominant factor. As detailed above, the lower O 2 diffusion coefficient in the IL and a longer diffusion pathway appears to restrict the efficient O 2 flux, leading to a dramatic decrease in the ORR activity.
Finally, the durability of Pt/C-[MTBD][C 4 F 9 SO 3 ] was examined using a protocol suggested by US DOE. Fig. 11 shows the polarization curves and cyclic voltammetry before and after 5000 cycles in the applied potential. The ECA loss and ORR activity loss at 0.9 V are summarized in Table II. Pronounced CV and ORR behavior changes were found for Pt/C, companied by 17.4% ECA loss, 30.3% specific Figure 11. Durability measurements of Pt/C-IL systems. ORR polarization curves before and after 5000 potential cycles: (a) Pristine Pt/C, and (b-c) Pt/C-IL with various IL/C ratios. The insets show the corresponding CV curves before and after the cycles. The cycling was carried out in N 2 -saturated 0.1 M HClO 4 over a potential range from 0.6 V to 1.0 V at a scan rate of 50 mV s −1 . activity loss, and 42.4% mass activity loss. In contrast, the losses were alleviated and less noticeable degradations were found at high IL content, demonstrating the protective effect of the IL. We adopted X-ray photoelectron spectroscopy analysis with N1s as an indicator to study the IL leaching in each step of electrochemical measurement. The results showed that the majority IL leaching occurred at the initial pre-conditioning, but was drastically reduced afterwards and the loss became less significant (not shown here). One can also observe that the change of CV of Pt/C-IL before and after cycling was small (inset in Figs. 11b-11d), suggesting the remaining IL in the catalyst layer helped to mitigate the Pt dissolution. Extensive research efforts have been undertaken to probe the Pt dissolution mechanism, and it is generally accepted that Pt dissolution occurs during anodic scan and cathodic scan, respectively. 63,64 Basically, Pt-oxides are formed during the anodic scan through Pt + H 2 O → PtO + 2H + + 2e − and Pt + 2H 2 O → PtO 2 + 4H + + 4e − , where an anodic dissolution takes place. The reductive dissolution follows a reaction pathway PtO 2 + 4H + + 2e − → Pt 2+ + 2H 2 O. One can easily find the presence of water is one of the key factors to generate the Pt-oxides and initiate the subsequent dissolution. Unlike Nafion, which has a hydrophilic domain, the [MTBD][C 4 F 9 SO 3 ] IL is relatively hydrophobic, with low water solubility that can provide a lateral repulsion force to reject water. Thus, reduced adsorption of water molecules on Pt is expected when the IL is present. The IL coverage can also affect the Pt dissolution process. In a partially covered Pt surface, a three-phase system remains and water can still attach to the Pt. Under full IL coverage, the water in contact with Pt will be reduced, except for the small amount of water dissolved in the IL. Overall, the alleviation of Pt dissolution is mainly attributed to protection by the IL. One has to mention, the smallest performance loss at an IL/C of 3.84 was achieved at a sacrifice of ORR activity' thus, a compromize of performance and stability should be taken into consideration.

Conclusions
We have examined a series of [MTBD]-based ILs with various anion structures to study the effect of oxygen solubility on ORR activity. ILs with higher oxygen solubility do not improve the ORR activity in the kinetic controlled region (either mass activity or specific activity). On the other hand, the specific activity correlates well with the availability of Pt sites (1-θ). A study of surface coverage revealed that ILs actually form a protective layer that reduces the adsorption of non-reactive species, which boosted the specific activity in return. Investigation of the influence of IL loading suggests that excess IL should be avoided in order to maintain three-phase contact between the solid catalyst, liquid and gaseous reagent, and a reasonable IL coverage on Pt should be chosen to balance the ORR activity and stability.