Platinum Dissolution and Redeposition from Pt/C Fuel Cell Electrocatalyst at Potential Cycling

Extremely sensitive on-line detection of metal ions concentration was used to investigate potential-resolved platinum dissolution from commercial fuel cell electrocatalyst. The experiments were carried out in an electrochemical ﬂow cell connected to inductively coupledplasmamassspectrometer.Avarietyofelectrochemicaltreatmentsusingdifferentvoltagescanratesconﬁrmedthepreviouslyobservedprimaryplatinumdegradationmechanism-theso-called“transientdissolution”.Importantly,theredepositionofdissolved platinumisnowshowntoplayanimportantroleintheoveralleffectivePtdissolution.Ptredepositiontrendsexhibitasigniﬁcantdependenceonvoltagescanrate.

Platinum dissolution has been extensively studied before-either in the form of a polycrystalline disk [1][2][3] or as a nanoparticulate catalyst. [4][5][6][7][8][9][10][11] Those studies have revealed that Pt electrochemical dissolution is predominantly a transient phenomenon occurring due to an interplay of Pt oxidation and reduction. The amount of dissolved platinum can be manipulated via different electrochemical treatments (scan rate, width of anodic and cathodic potential window), 1 gas atmosphere, 12 electrolyte, presence of organic molecules, 6,13 alloying metals, 8,14,15 thickness of the catalyst layer, 10,16 type of support material 16 and, last but not least, by the Pt particle size. 4 Much is owed to the development of advanced on-line analytical tools, especially electrochemical scanning flow cell (SFC) in combination with inductively coupled plasma mass spectrometer (ICP-MS). This particular methodology has importantly contributed to significant progress in the fundamental understanding of Pt dissolution during the past years. 1,2 Different versions of this setup have been developed in the meantime. 4,17,18 All of studies on this subject show consistently that the transient electrochemical Pt dissolution consists of an anodic and a cathodic counterpart. A comprehensive model has been reported describing the anodic dissolution as a consequence of a direct electrooxidation of the initially low coordinated Pt sites and more coordinated ones at higher potentials. 1 On the other hand, the cathodic dissolution, which is more intensive than the anodic counterpart, occurs due to reduction of sub-surface platinum oxide that weakens the Pt-Pt interaction by roughening and restructuring the surface. Ultimately Pt forms defect (low coordinated) sites that are prone to dissolution. 1 In the current study, we examine the effect of voltage scan rate on Pt dissolution from a benchmark Pt/C fuel cell electrocatalysts and report on several intriguing insights.

Experimental
A BASi electrochemical flow cell (Cross-Flow Cell Kit MW-5052) was coupled with an Agilent 7500ce ICP-MS instrument (Agilent Technologies, Palo Alto, USA), equipped with a MicroMist glass concentric nebulizer and a Peltier cooled Scott type double-pass quartz spray chamber. A 1500 W forward RF power was used, with the following argon gas flows: carrier-0.85 Lmin −1 , makeup 0.28 Lmin −1 ; = These authors contributed equally to this work. z E-mail: nejc.hodnik@ki.si plasma 1 Lmin −1 and cooling-15 Lmin −1 . 0.1 molL −1 HClO 4 (Aldrich 70%, 99.999% trace metals basis) carrier solution for the electrochemical experiments was pumped at 263 μLmin −1 by a peristaltic pump through the BASi EC flow cell into the ICP-MS nebulizer. The volume of the flow cell was defined by selecting the thickness of teflon gasket (in our case 0.14 mm).
Catalyst thin films were prepared by drop casting the suspension of well-dispersed catalyst in Milli-Q water (1 mg/mL) on glassy carbon electrodes. After the solvent had evaporated, a freshly prepared Nafion in 2-propanol (5 μL, 0.1%) was coated over the catalyst layer. The final total mass of catalyst was 10 μg (5 μg per electrode). The industrial benchmark consisted of 3 nm Pt particles (mean size) dispersed on Vulcan XC72 with a metal loading of 28.6 wt% and was the same as used in References 4,6,19.

Results and Discussion
Pt dissolution in general.- Fig. 1 shows the results of typical cyclic voltammetry experiments on a standard Pt-3 nm electrocatalyst. Simultaneously, the concentration of dissolved Pt from this sample as detected using online ICP-MS is displayed. The potential is swept from 0.05 V vs. RHE till increasing upper potential limits (UPL), specifically from 0.9 till 1.6 V vs. RHE, with the sweeping rate held constant at 5 mV s −1 . We note that the rather high UPL value of 1.6 V could induce carbon corrosion which would (indirectly) lead to Pt detachment. However, the rate of electrochemical oxidation of carbon support has been shown to be insignificant at room temperature, unless fast transitions (0.5 Vs −1 ) from low to high potential regimes (0.6 to 1.5 V vs RHE) are employed, as demonstrated by Pizzutilo et al. 20 It is seen from Fig. 1 that with increasing UPL, the platinum dissolution signal also increases. The first dissolved amounts of Pt are detected already at the initial UPL (0.9 V) which is in line with the work of Cherevko et al. 5 The main Pt dissolution features are in line with those found for polycrystalline Pt 1,2 and Pt nanoparticles. 4,6,8,14 The anodic Pt dissolution is presumably governed by the inversion of the Pt-O dipole in the Pt surface via the place-exchange mechanism, which leads to roughening and restructuring of Pt atoms at the surface. 1,[21][22][23][24][25][26] Once the UPL of 1.2 V has been reached, two peaks can be distinguished. The smaller one is evolving in the anodic and the bigger one in the cathodic scan (Fig. 1a). It has been reported that the position of the cathodic dissolution shifts toward lower potentials when the upper potential is increased. 4 This is in line with the irreversible nature of PtO and PtO 2 formation/reduction that is shifting the Pt-oxides reduction peak potential to lower potentials. A more detailed inspection reveals that the anodic Pt dissolution consists of two maxima labeled here as peak A1 and A2 (Fig. 1b). The former is ascribed to Pt dissolution due to Pt oxides formation, which eventually passivate Pt surface and inhibit its dissolution. A1 is the only anodic dissolution peak at UPL values ≤ 1.4 V vs. RHE. 2,3 At higher potentials, i.e., in the oxygen evolution region (OER) peak A2 starts to appear, which has been reported to be related to OER dynamics. 3,27,28 During the second cycle till UPL 1.6 vs. RHE the dissolution profile is notably lower. We presume that this is a measurement artifact, possibly due to formation of a small oxygen bubble blocking the surface. Other profiles and their areas are within the measurement error, namely the relative standard deviation (RSD) of 13%.
A discussion on dynamics of Pt dissolution: the scan rate experiment.-In order to get a more comprehensive insight into Pt dissolution under potentiodynamic conditions, the dissolution of Pt-3 nm was monitored under different voltage scan rates, varying from 2 to 500 mVs −1 . The upper potential limit was always 1.6 V vs. RHE. This (rather high potential limit) was chosen in order to capture as wide as possible potential range. The following separate experiments were performed: Anodic and cathodic scan rates were fixed within the same cycle and gradually decreased (from 500 to 2 mV s −1 ) in subsequent cycles.-These results confirm the already reported trend of Pt dissolution, which linearly scales with the amount of formed and subsequently reduced platinum oxide. 1 The total platinum dissolution amounts are shown to be inversely proportional to the voltage scan rate (Fig. 2 inset). 1 Separately determined, the anodic Pt dissolution rates show a different trend (Fig. 2b). This indicates that by increasing the voltage scan rate the Pt dissolution rate increases. Since less Pt oxide is formed under high scan rates, 29,30 Pt is less passivated with oxide, hence a direct dissolution of Pt (Pt → Pt 2+ + 2e − ) is promoted. This observation is not completely in line with the results on polycrystalline Pt reported by Topalov et al. 1 nor with the recent study by Lopes et al. where dissolution of Pt (111) surface was investigated. 31 Both studies showed a very small effect on the scan rate, or the amount of oxide formed, on the anodic Pt dissolution. However, we note that -at least as far as the process of Pt dissolution is concerned -single crystalline and polycrystalline surfaces differ importantly from the nanoparticles film in four aspects: i) orientation and distribution of crystal surface planes, ii) oxophilicity, iii) surface energy of Pt atoms -according to the Gibbs-Thomson effect 32 and iv) much lower Pt surface area. All four characteristics support the higher measured dissolution amounts from nanoparticles in any potentiodynamic treatment in comparison to the single crystalline or polycrystalline surfaces. In the case of the latter, it was shown that only Pt surface atoms with the lowest coordination, already present on the surface at the initial state, dissolve in the anodic scan. Hence this process is not significantly dependent on the scan rate, but only on the amount of oxide formed. 1,31 In the cathodic branch (not shown), the trend as regards the totally dissolved amount of Pt (note that this is different from the dissolution rates) is the same as in the anodic counterpart. This is in line with the amount of oxide formed, which is increasing with decreasing the scan rate. 1 As more time is given for this relatively slow process, more of Pt oxide gets formed. 29,30 Formation of the oxide should also be considered in the cathodic scan. Namely, the oxide continues to grow even after switching the potential vertex to cathodic direction. 21 Hence, to a certain extent, the increasing Pt dissolution amounts with decreasing scan rate could be ascribed to the oxide induced dissolution in the cathodic scan, at least in the high potential region of the scan. In the low potential region, the reduction of oxide is governing Pt dissolution.
Apart from the Pt oxide reduction dynamics, one has to additionally consider the event of redeposition of dissolved Pt ions, which can suppress the effective Pt cathodic dissolution. Namely, by increasing the scan rate at least three factors favor the redeposition: (i) more favorable potentials for redeposition are reached faster thus the rate of redeposition is faster, (ii) diffusion has less time to transfer Pt ions away from Pt surface and (iii) higher Pt ion concentrations occur near and even in the Pt catalyst thin film that favor redeposition. 1,31 The interplay of the effects of oxide reduction and redeposition is demonstrated by the trend of concentration maximums in the cathodic scan.
This trend reveals that the maximum, which is the highest concentration of Pt in the downstream electrolyte, is obtained at 25 mVs −1 (Fig. 2a). This is perfectly in line with the fact that at slow scan rates the oxide reduction proceeds sufficiently slow to prevent any relevant increase of the concentration of Pt in the electrolyte. With increasing scan rate the dissolution maximum steadily grows. However, at fast enough scan rates the oxide formation also becomes too slow for relevant dissolution to occur. Due to an interplay of these effects, which are presently only recognized qualitatively, it is not possible to predict the effect of short exposure to oxide formation/reduction and more effective redeposition in a reliable way.
Given the results discussed above, we tried to distinguish the effect of the timescale of experiment from the effect of the amount of formed oxide on the amount of dissolved Pt. Thus the following advanced experimental protocol was applied: The anodic scan rate was fixed at 5 mVs −1 followed by variable cathodic scan rates (from 500 to 2 mV s −1 ).-The aim of this experiment was to ensure that the amount and the nature of formed oxide were the same for all experiments. Therefore the "the anodic effect of oxide" on Pt dissolution was constant, and the only variable was the voltage scan rate in cathodic direction. Here we note again that the amount of oxide formed in cathodic direction ("cathodically formed oxide") is still a changing parameter as it is dependent on the scan rate. A reference cycle (20 mVs −1 ) was recorded after each sequence in order to verify if any irreversible changes, for example, the presence of unreduced oxide etc., occurred on Pt nanoparticles (see orange peaks in Fig. 3a). The results of Fig. 3a indicate that such changes, if any, were insignificant, as the dissolution profiles areas are within the measurement error (RSD 13%). The dissolution data reveal that by increasing the cathodic scan rate, the total dissolution amount decreases (Fig. 3a inset). We note that this trend is not uniformly expressed throughout all scan rates, which should be ascribed to the measurement error (RSD 13%). Nevertheless, the general dissolution trend is in line with a) a more effective redeposition of Pt ions, b) a lesser amount ("cathodically formed oxide" or c) both i) and ii). The dissolution maximums (Fig. 3a), however, show a different trend in comparison to Fig. 2; here the maximum concentration is consistently decreasing with decreasing the voltage scan rate whereas in Fig. 2 a maximum was observed at 25 mV s −1 . This eliminates the possibility that "cathodically formed oxide" could have a significant effect on cathodic dissolution and is in line with all of the three aspects that were assumed above to favor redeposition: (i) faster kinetics, (ii) less time for Pt ions to diffuse away and (iii) higher Pt ion concentrations near and in the catalyst thin film. In addition, a higher Pt concentration also causes a lower Pt dissolution due to the Nernstian behavior of this reaction. 16,33 In order to get more insight into these mechanisms, another set of experiments (see section Linear sweep anodic scan rate was fixed at 5 mVs −1 followed by potential holds at different lower potential limits (LPL)) was performed.
Linear sweep anodic scan rate was fixed at 5 mVs −1 followed by potential holds at different lower potential limits (LPL).-As in the previous experiments, the anodic scan rate and LPL were fixed to assure that the amount of oxide formed was always the same. After reaching the UPL of 1.6 V, the potential was lowered to different LPLs and held there for 300 s (Fig. 3c). Different LPLs were chosen in order to observe the effect of the kinetics of Pt oxide reduction; the duration of 300 s was chosen in order to give enough time for reduction to reach or at least to come close to the steady state through the entire oxide layer. Dissolution maximums are again increasing with decreasing the LPL value, which is in line with the more complete oxide reduction as a result of the sluggish kinetics of Pt oxide reduction. 31 On the other hand, the total amounts of Pt dissolved found in the downstream electrolyte (Fig. 3d) are decreasing with decreasing LPL indicating that redeposition is more effective for the same three reasons as mentioned previously. Note that also in this experiment the "cathodic oxide growth" is not possible as potentiostatic conditions are used. Furthermore, the potentials are sufficiently low, hence no oxide can be formed under such conditions.

Conclusions
Employing EFC-ICP-MS, new insights into the dissolution of Pt/C fuel cell catalyst were obtained. Specifically, under carefully selected electrochemical conditions, the anodic Pt dissolution from nanoparticles can be described as a scan rate-dependent process. Furthermore, in the case of the more aggressive cathodic dissolution it was found that redeposition is competing with Pt dissolution and greatly affects the overall effective amount of dissolved Pt. This points toward reconsideration of PEM fuel cell accelerated degradation tests.