2021 Volume 62 Issue 6 Pages 797-806
A time-resolved measurement system, which combined a commercially available solution-flow cell and an inductively coupled plasma mass spectrometer, was established for determining the dissolution rate of platinum (Pt) and palladium (Pd) in this study. The detection limit of the system was successfully improved by thinning the cell channel, and the limits for Pt and Pd were 0.13 pg cm−2 s−1 and 0.39 pg cm−2 s−1, respectively. When the system was applied to Pt and Pd under potential cycling that mimicked the start/stop conditions of a polymer electrolyte fuel cells (PEFCs), it was revealed that the dissolution of Pt and Pd started from a more negative potential than that reported in previous studies. In addition, we succeeded in obtaining time-resolved measurements of the Pt and Pd dissolution rates below the open circuit potential of the PEFCs (approximately 1.0 V) and clarified that the dissolution mechanisms of Pt and Pd were different.
This Paper was Originally Published in Zairyo-to-Kankyo 69 (2020) 221–230. Abstract and captions of figures, and Fig. 6 are slightly modified.
Platinum (Pt),1) palladium (Pd),2) and their alloys (Pt–Pd alloys)3–7) exhibit excellent catalytic activity for the oxygen reduction reaction (ORR). Thus, they are considered candidates for use as cathode catalysts in polymer electrolyte fuel cells (PEFCs) to realize the downsizing and low-temperature operation of PEFCs. PEFCs have attracted attention as a clean energy conversion device because they do not emit toxic gases, such as carbon dioxide, and are expected to be widely prevalent in the future. However, the scarcity and high cost of Pt hinder its widespread use in PEFCs; thus, Pt alloy catalysts should be developed to decrease the use of Pt in cathode catalysts. Pt–Pd alloys along with Pt alloys containing transition metal elements M (iron (Fe), cobalt (Co), nickel (Ni), and copper (Cu)) show higher ORR activity than Pt catalysts.8–13) Therefore, alloy catalysts play a crucial role in decreasing the cost of a PEFCs. On the other hand, compared with a Pt catalyst, a Pt–M alloy catalyst is assumed to quickly deteriorate due to the dissolution of M because M is usually less noble than Pt. In other words, achieving a high durability of alloy catalysts under PEFCs operating conditions is necessary before their practical application in a PEFCs. Hence, investigating the dissolution mechanism of a Pt–M alloy is an urgent issue.
One of the problems in investigating the dissolution mechanism of Pt and Pt–M alloys is that the amount of dissolved Pt is extremely trivial. Moreover, in the case of Pt–M alloys, a Pt-enriched layer forms on the alloy surface because of the selective dissolution of the less noble metal in the alloy.14) Therefore, the further dissolution of M is suppressed by the Pt-enriched layer, resulting in a decrease in the amount of dissolved M, which also becomes a problem. Furthermore, the formation of a Pt oxide on the surface of the Pt and Pt–M alloys and the reduction current of this Pt oxide mainly appear in a cyclic voltammogram (CV) under unsteady-state conditions, which are similar to PEFCs operating conditions (potential cycling). Therefore, it is difficult to estimate the dissolution rate of Pt and M using a current measured in CV because the small current due to the dissolution of Pt and M is superimposed and hidden by the formation/reduction current of the Pt oxide.
Our group introduced inductively coupled plasma mass spectrometry (ICP-MS) to quantify the trivial amounts of dissolved Pt and M.15–19) Pt–M alloys were subjected to 100 cycles of potential cycling in a stationary solution, and the amount of dissolved Pt and M was directly quantified by analyzing the test solution using ICP-MS. Using this method, we clarified the effects of M species,15) range of potential cycling,16) Co composition on the dissolution mechanism of Pt–M alloys,17,18) and the correlation between the dissolution of M and the morphological change of the surface.19) However, the amount of dissolved Pt and M obtained by ICP-MS was an integrated value in the 100 cycles of potential cycling. Thus, the potential-resolved data of the amount of dissolved Pt and M could not be obtained. Therefore, Wang and Ooi et al.20–23) focused on a channel flow multi-electrode (CFME), which is a type of solution-flow electrochemical cell. In the CFME, mass transfer was controlled by flowing a laminar-flow solution into a channel. The dissolved Pt and Fe ions from the upstream Pt or Pt–Fe alloy electrode in the channel could be detected on the downstream detection electrode by electrochemical reactions. Since the gap between the two electrodes was 100 µm, the amounts of dissolved Pt and Fe were quantitatively detected in situ during potential cycling. The detection response of the CFME at a fast potential sweep rate was better than that of the online ICP-MS analysis (as described later). Since carbon (C)-supported Pt (Pt/C) catalysts are used as PEFCs catalysts, catalyst degradation caused by the detachment of Pt from the support due to carbon corrosion under PEFCs operating conditions is also a concern.24–26) A CFME can distinguish between Pt dissolution (electrochemical reaction) and Pt detachment from the electrode (nonelectrochemical reaction). Moreover, the dissolved valence of Pt (Pt2+ and Pt4+) is also distinguishable using a CFME.20)
At approximately the same time as our group started to analyze the dissolution mechanism of Pt–M alloys using ICP-MS, Mayrhofer et al.27) reported an online ICP-MS analysis of Cu dissolution by a solution-flow cell directly combined with ICP-MS. Furthermore, they developed a unique cell, which had a V-shaped flow circuit (test area: 1.5 mm2) called a scanning flow cell (SFC) and could measure an arbitrary position of a sample with an X-Y-Z stage.28) By using a combination of SFC and ICP-MS (SFC-ICP-MS), the dissolution mechanisms of precious metals, such as Pt,29–31) Pd,32) gold,33) and rhodium,34) and alloys, such as Pt–Cu35) and Pt–Ni,36) were systematically investigated. The detectability of SFC-ICP-MS depended on the detection limit of ICP-MS and was approximately 3 pg cm−2 s−1 for Pt.29) This value was higher than the detectability of the CFME (approximately 11 pg cm−2 s−1); thus, a time (potential)-resolved measurement of the extremely slow dissolution rate of precious metals could be possible. With the construction of SFC-ICP-MS as a basis, studies toward improving online ICP-MS analysis drastically increased. Instead of a SFC, a commercially available solution-flow cell37,38) and a stationary probe (SP) with a rotating ring electrode (RDE)39) can be successfully combined with ICP-MS, and SPRDE-ICP-MS has a detectability that is one order of magnitude lower than SFC-ICP-MS (approximately 0.4 pg cm−2 s−1).
Due to detection limits, most of the analyses of the dissolution mechanism of Pt and Pt–M alloys using CFME and SFC-ICP-MS focus on Pt dissolution at a noble potential (>1.0 V) above the open circuit potential (OCP) of the PEFCs because the dissolution of Pt will be more significant. In contrast, it has been reported that a small amount of Pt dissolution occurs around the OCP, which is the load cycle of a PEFCs.40) The required lifetime for fuel cell vehicles is considered to be 5,000 hours or 10 years of operation,41) and thus, even a slight dissolution of Pt may have a significant effect on the performance loss. Cherevko et al.31) proposed a stagnant method using SFC-ICP-MS to detect a trivial amount of dissolved Pt around the OCP. Pt was dissolved in an environment where the solution flow was stopped, and Pt ions were concentrated (stagnated) above the detection limit in a small area inside the cell. Finally, the preconcentrated Pt ions were detected by ICP-MS after resuming the solution flow. However, this stagnant method resulted in the lack of time-resolved information about Pt dissolution, which is the greatest advantage of SFC-ICP-MS. Therefore, if the lower detection limit of SFC-ICP-MS is improved, the time-resolved measurement of trivial Pt dissolution will be possible. Moreover, it will also be possible to detect the onset of Pt dissolution at a less noble potential than the onset potential previously reported.
In this study, a commercially available solution-flow cell was coupled with ICP-MS to establish an online ICP-MS analysis system with high detectability. This system was applied to Pt and Pd, and the detectability of the system and dissolution behavior of Pt and Pd were investigated.
Figure 1(a) shows the overall image of the measurement system that combines a solution-flow cell and ICP-MS for online analysis of a trivial amount of dissolved metal. In reference to the previous report by Gaberšček et al.,37,38) the measurement system consisted of a solution tank, a commercially available solution-flow cell (Cross flow cell, BAS Inc.), and an ICP-MS instrument. After deaerating the test solution with argon (Ar) gas, the valve of the tank was released, and then the peristaltic pump rotated. The solution finally flowed into a channel inside the cross-flow cell. The cross-flow cell consisted of a Teflon sheet (t = 100 µm) sandwiched between an electrode block and a flow-path block, and the solution flowed through the channel between the two blocks (Fig. 1(b)). Pt (99.95%) or Pd (99.95%), with a diameter of 3 mm as a working electrode (WE), and glassy carbon, with a diameter of 3 mm as a counter electrode (CE), were parallelly embedded in the electrode block with a gap of 1 mm. Paths for the solution inlet and outlet were open in the flow-path block, and a reference electrode (RE) was screwed into the outlet side of the solution. In this study, a screw-type silver/silver chloride (Ag/AgCl) electrode (RE-3VT, BAS Inc.) was used for the RE. The internal solution of the RE was 3 M NaCl, and thus, the potential difference between RE-3VT and the standard hydrogen electrode (SHE) was approximately +195 mV. Figure 1(c) schematically illustrates the ellipsoidal channel (V = 7.7 mm3) in the cell. The flow direction of the solution was perpendicular to the WE–CE. This design was to minimize the effects of (i) the pH change at the CE and (ii) the redeposition of dissolved ions from the WE on the CE during electrochemical measurements.
Schematic drawing showing the experimental setup for the online detection of metal dissolution using a cross-flow cell and ICP-MS.
The electrode block was mirror-polished with 0.25 µm diamond paste and then ultrasonically cleaned in Milli-Q (18 MΩ cm) water before experiments. Next, 0.5 M H2SO4 solutions (pH = approximately 0.5) were prepared by diluting Merck Suprapur 96% H2SO4 in Milli-Q water. All the electrochemical measurements were conducted in Ar-purged 0.5 M H2SO4 at 298 K by using a potentiostat (SP-300, Bio-logic Science Instruments). In this paper, the measured potential was mostly converted to the SHE; however, the measured potential was also partially converted to the reversible hydrogen electrode (RHE) to compare to the potential values measured in previous studies. In that case, the potential converted to the RHE was described as VRHE so that the SHE and RHE could be distinguished.
Before a test, 50 cycles of potential cycling were performed between 0.05 and 1.4 V at 100 mV s−1 to electrochemically clean the electrode surface, and then the WE underwent potential cycling over the same range at 10 or 1 mV s−1 or were potentiostatically polarized at 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0 V for 300 s.
2.3 Online ICP-MS measurementsOnline measurements of electrochemically dissolved Pt and Pd from the WE were conducted by using the time-resolved measurement mode in ICP-MS (7700x, Agilent Technology). When the time-resolved measurements of Pt (Pd) dissolution were performed, thallium (Tl) (indium (In)) was used as the internal solution, and the ICP-MS signals of Pt195 (Pd105) and Tl205 (In115) were alternately recorded at an interval of 0.03 s. The flow rate (vFlow) was determined by the ratio of the test solution weight, which was sampled for 10 min from the 3-way joint (Fig. 1(a), * part), to the solution density. Almost no deviation of vFlow appeared in the experiments, and the average value of vFlow was approximately 110 µL min−1 (1.8 µL s−1).
Standard curves for Pt and Pd were prepared to convert the ICP-MS signals to concentration (cM) [ppb] (= [µg L−1]) or dissolution rate (v) [pg cm−2 s−1] values. Standard solutions (concentrations of 0.1, 0.3, 0.5, and 1.0 ppb) of Pt and Pd were prepared with a commercial Pt standard solution (10 ppm, Agilent Technology) and a commercial Pd standard solution (1000 ppm, Agilent Technology) in 0.5 M H2SO4. The cM and v have the following correlation:
| \begin{equation} v = \frac{c_{\text{M}} \times v_{\text{Flow}}}{S_{\text{Geo}}}, \end{equation} | (1) |
Fifty cycles of electrochemical cleaning (0.05–1.4 V at 100 mV s−1) were conducted for Pt and Pd in Ar-purged 0.5 M H2SO4. Cyclic voltammograms (CVs) for Pt and Pd are shown in Figs. 2(a) and 2(b), respectively. As shown in Fig. 2(a), the hydrogen adsorption/desorption peaks, double layer region, and oxide formation/reduction peaks typically appeared in the initial CV of Pt. The shape of these peaks gradually changed with increasing cycle number before finally remaining almost constant, which indicated that a clean and stable surface was obtained by electrochemical cleaning. Similar to the CV of Pt, Pd oxide formation/reduction peaks typically appeared in the initial CV of Pd (Fig. 2(b)). In contrast, the hydrogen absorption peak overlapped with the hydrogen adsorption peak in the CV of Pd below 0.3 V in the cathodic scan. Therefore, the reduction current of Pd was much higher than that of Pt. Absorbed hydrogen inside the Pd bulk in the preceding cathodic scan desorbed in the anodic scan, which resulted in a large oxidation current. The shape of these peaks exhibited almost no change at the 50th cycle, and thus, a clean and stable surface of Pd was obtained.
CV profiles of (a) Pt and (b) Pd in Ar-purged 0.5 M H2SO4 at 100 mV s−1 between 0.05 and 1.4 V. CV profiles of (c) Pt and (d) Pd in Ar-purged 0.5 M H2SO4 at 100 mV s−1 between 0.05 and 1.4 V with different gasket thicknesses.
In a sufficiently deaerated solution, the double layer region that appeared in the CVs of Pt and Pd should ideally be almost symmetrical to the x-axis (i = 0). However, the double layer region in the CVs in this work shifted in a negative direction. This result was considered to be the effect of thinning a cell channel (100 µm) to detect a trivial amount of dissolved Pt and Pd. Levich42) reported a value of diffusion limiting current under laminar-flow conditions, which was analyzed by hydrodynamics, and the value increased with as the channel became thinner. The ellipsoidal flow circuit shown in Fig. 1(c) does not strictly satisfy a laminar-flow condition; however, the diffusion limiting current of residual oxygen in the test solution might cause a negative shift in the double layer region. Furthermore, this shift decreased as the channel thickened; therefore, the channel thickness was closely related to the detection limit of the established system. Thus, we determined that 100 µm was a suitable channel thickness for the system. When the channel thickness was increased from 100 µm to 1 mm, almost no negative shift was observed for the double layer region in the CVs of Pt and Pd (Figs. 2(c) and 2(d), respectively), and the onset potential for Pt (Pd) oxidation and the peak potential for Pt (Pd) oxide reduction could be estimated.
Figure 3 represents the online dissolution profiles of Pt and Pd during electrochemical cleaning. Prior to cleaning, the WE was polarized (i) at 1.4 V for 300 s, and then (ii) at 0.45 V for 300 s (Fig. 3(a)). This polarization was conducted to obtain Pt or Pd dissolution signals in ICP-MS. Pt and Pd dissolution occurred when they were polarized to noble potentials such as 1.4 V, and the Pt and Pd oxides that formed during polarization at 1.4 V were reduced to bare metals at 0.45 V.22,30,32) As shown in Fig. 1(a), there was some distance between the cross-flow cell and ICP-MS; thus, a time delay for the metal ions dissolved in the cell to reach ICP-MS should be considered to obtain accurate time-resolved data. Since Pt and Pd dissolved immediately after polarization at (i) 1.4 V and (ii) 0.45 V, if the ICP-MS measurement was started before polarization, the time for the Pt or Pd signal to increase from the baseline in the ICP-MS profile could be judged as the start time of the electrochemical measurement. The dissolution rates of Pt and Pd (Figs. 3(b) and 3(c), respectively) were plotted after correcting the delay between the ICP-MS and electrochemical measurements using the above calibration method.
(a) Potential sequence applied to the Pt and Pd electrodes. The measured (b) Pt and (c) Pd dissolution rates during the potential sequence.
Figure 3(b) confirmed that Pt dissolution accelerated immediately after polarization at 1.4 and 0.45 V. The Pt dissolution signal drastically decreased over time and finally reached its baseline within 300 s regardless of the polarized potential. Then, when Pt was subjected to potential cycling after polarization at 0.45 V, Pt dissolution increased again during potential cycling, and many peaks appeared in the dissolution profile. Since the number of peaks matched the number of cycles, it was suggested that the change in Pt dissolution rate with the potential change during the cycle could be monitored. However, previous research using a channel flow double electrode (CFDE)21) reported that two peaks appeared in the Pt dissolution profile within 1 cycle. Therefore, these two peaks might overlap in the online ICP-MS profile during potential cycling at a scan rate of 100 mV s−1. This phenomenon was also reported in the SFC-ICP-MS analyses using a V-shaped flow circuit cell,30) and these two peaks could not be separated unless the scan rate was decreased.
The dissolution rate of Pd is illustrated in Fig. 3(c); note that the scale for Pd dissolution was 10 times different than that for Pt dissolution. Comparing the dissolution rates of Pt and Pd, a similar tendency was observed in the profile; however, the dissolution behavior during potentiostatic polarization was quite different. In the case of Pt, the dissolution rate polarized at 0.45 V was higher than that at 1.4 V. In other words, cathodic dissolution during Pt oxide reduction was a dominant process. In contrast, anodic dissolution polarized at 1.4 V was dominant for Pd. Moreover, the Pd dissolution signal did not return to baseline for at least 300 s, indicating that unlike Pt, Pd continuously dissolved during polarization at 1.4 V. Hence, it was clear that the dissolution behavior of Pt and Pd differed significantly in the potential range up to 1.4 V, which corresponded to the PEFCs start-up/slow-down conditions. Furthermore, the amounts of dissolved Pt and Pd during potentiostatic polarization and electrochemical cleaning were 0.39 and 7.2 µg cm−2, respectively. The amount of dissolved Pd was approximately 20 times larger than that of Pt. However, the weights of Pt and Pd that existed in 1 monolayer (ML) corresponded to 0.40 and 0.22 µg cm−2, respectively; thus, the amounts of dissolved Pt and Pd were 1 and 32 MLs, respectively. Therefore, only the outermost layer of Pt and 6 nm-thick Pd dissolved during electrochemical cleaning.
3.2 Effect of scan rate on the dissolution rate of Pt and PdFigure 4(a) exhibits the online dissolution profile of Pt during potential cycling between 0.05 and 1.4 V at a scan rate of 10 mV s−1. Potentiostatic polarization was performed before potential cycling to calibrate the delay time between ICP-MS and the electrochemical measurement as described in Section 3.1, and the profile during potentiostatic polarization was omitted in the graph. Moreover, even though a total of 10 cycles of potential cycling was conducted, the dissolution profile of Pt was stable through 10 cycles after electrochemical cleaning (Section 3.1); thus, the result of the 3rd cycle was typically represented. Unlike the result of Fig. 3(b), two peaks were observed in the dissolution rate of Pt through 1 cycle. These were classified as PA (starting from approximately 1.00 V (0.97 VRHE) in the anodic scan) (see Fig. 4(b)) and as PC (approximately 0.62 V in the cathodic scan). Mayrhofer et al.30) reported that the onset potential of PA (i.e., the onset of Pt dissolution) was approximately 1.1 VRHE during potential cycling, which was almost the same value found in this work. Thus, Pt dissolution was successfully detected at a less noble potential compared to the previously reported value using the established experimental setup in this study. The amount of dissolved Pt during potential cycling was calculated to be approximately 12.1 ng cm−2 cycle−1. The amounts corresponding to PA and PC were 0.14 and 11.9 ng cm−2 cycle−1, respectively, and thus, remarkably, Pt dissolution occurred in the cathodic scan. This tendency was in good agreement with the Pt dissolution during potentiostatic polarization for the calibration of the delay time (Fig. 3(b)).
The measured (a) Pt and (c) Pd dissolution rates during the 3rd potential cycle at 10 mV s−1 and between 0.05 and 1.4 V. (b) and (d) are magnified views of (a) and (c) from approximately 40–70 s and 10–40 s, respectively.
Figure 4(c) represents the online dissolution profile of Pd during potential cycling between 0.05 and 1.4 V at a scan rate of 10 mV s−1. Two peaks, which could not be distinguished at a scan rate of 100 mV s−1, also appeared in the dissolution profile of Pd, similar to that of Pt. These were classified into PA (starting from approximately 0.73 V (0.70 VRHE) in the anodic scan) (see Fig. 4(d)) and PC (approximately 0.57 V in the cathodic scan). Mayrhofer et al.32) reported that the onset potential for Pd dissolution was 0.80 VRHE during potential cycling at a scan rate of 2 mV s−1 in 0.1 M H2SO4. Thus, Pd dissolution was also successfully detected at a less noble potential than the previously reported value in this study. There is no distinguishable border between PA and PC in the profile, and thus, Pd continuously dissolved from the anodic to the cathodic scan. The amount of dissolved Pd during potential cycling was calculated to be approximately 231 ng cm−2 cycle−1. Similar to the electrochemical cleaning results, the amount of dissolved Pd was approximately 20 times larger than that of Pt.
To clarify the correlation between the surface changes on Pt (Pd) and the dissolution rate of Pt (Pd), the detection profile in Fig. 4, which is against the potential, is replotted in Fig. 5 together with their CVs. As mentioned in Section 3.1, the current shift in the negative direction below 0.9 V was due to the residual oxygen that appeared in the CVs (Figs. 5(a) and 5(b)). Thus, it was difficult to estimate the precise onset potential for Pt and Pd oxidation. Moreover, the reduction peak of Pt oxide completely vanished in the CV. Although the Pd oxide reduction peak was observed in the CV, an accurate peak potential for Pd oxide reduction was also difficult due to the current shift. In general, the scan rate difference slightly affected oxide formation and reduction potential; however, these potentials could be approximately estimated from the CVs at a scan rate of 100 mV s−1 (Figs. 2(c) and 2(d)). The oxide formation potential of Pt and Pd was not clear; however, it was determined to be approximately 0.80 and 0.70 V, which was in good agreement with the previous results.32,43) On the other hand, the peak potential for Pt and Pd oxide was found to be approximately 0.73 and 0.64 V, respectively. These potentials were very close to the potentials for PA and PC in the dissolution profile of Pt and Pd, which suggested that Pt and Pd dissolution during potential cycling was strongly correlated with the surface oxidation and reduction of Pt and Pd.
CV profiles of (a) Pt and (b) Pd in Ar-purged 0.5 M H2SO4 at 10 mV s−1 between 0.05 and 1.4 V. The corresponding mass cyclic voltammograms of (c) Pt and (d) Pd. The solid and broken lines show the anodic and cathodic scans, respectively.
Even if the scan rate decreased, the dissolution behavior of Pt and Pd was almost unchanged, and PA and PC were observed in their dissolution profiles during potential cycling (Fig. 6). In the case of the dissolution profile of Pt, the onset potential for PA in the anodic scan was approximately 0.94 V, and the peak potential for PC in the cathodic scan was approximately 0.73 V. The onset potential for PA showed almost no dependence on the scan rate; however, the peak potential for PC shifted by more than 100 mV in the positive direction when the scan rate decreased. These trends were also recognized in the dissolution profile of Pd. The amounts of dissolved Pt and Pd during potential cycling are summarized in Table 1. For Pt dissolution, the ratio of PA to PC did not depend on the scan rate. The dissolution in the cathodic scan was a dominant process. In contrast, Pd dissolution in the anodic scan was a dominant process, although PA and PC were not clearly distinguished in the profiles (Figs. 4 and 6). Moreover, the amount of dissolved Pt and Pd per 1 cycle increased with decreasing scan rate. Note that the amount of dissolved Pt and Pd was extremely small, and the amount of dissolved Pt corresponded to 0.03–0.04 ML cycle−1. In other words, only 3–4% of Pt dissolved from the outermost surface during potential cycling. In addition, the amount of dissolved Pd was 1.0–1.7 ML cycle−1, and thus, only the Pd atoms near the outermost surface dissolved.
The measured (a) Pt and (c) Pd dissolution rates during potential cycling at 1 mV s−1 between 0.45 and 1.4 V. (b) and (d) are magnified views of (a) and (c) from approximately 400–600 s and 200–400 s, respectively.
Since the baseline fluctuation appeared in the ICP-MS profiles (e.g., Fig. 4(b)), it was difficult to strictly determine the initiation time of dissolution. In the case of a fast scan rate, this error significantly influenced the determination of the dissolution onset potential. Thus, potentiostatic polarization was preferable to obtain an accurate onset potential of dissolution.
Figure 7 shows the dissolution profiles of Pt and Pd during potentiostatic polarization. Every time the upper potential limit (EU) was changed, 300 s of potentiostatic polarization was performed at 0.45 V to remove the Pt or Pd oxide that formed in the preceding polarization. As evidenced in Fig. 7(b), Pt dissolution was not detected during potentiostatic polarization up to 0.7 V. The dissolution rate of Pt suddenly increased immediately after polarization at 0.8 V; however, this rate was significantly higher than the subsequent dissolution rate at 0.9 and 1.0 V. Moreover, the dissolution rate of Pt at 0.8 V was also higher than that of Pd, which contradicted the results shown in the previous sections. Since Mayrhofer et al.31) reported that Pt dissolution slightly occurred at 0.79 V, Pt might dissolve by potentiostatic polarization at 0.8 V; however, its rate should be investigated in the future. As shown in the inset of Fig. 7(b), the Pt dissolution rate reached a maximum of 3 pg cm−2 s−1 immediately after polarization at 0.9 V and then gradually decreased to the baseline. Subsequently, the Pt oxide formed at 0.9 V was reduced by potentiostatic polarization at 0.45 V. The trace of the Pt dissolution profile at 0.45 V was similar to that at 0.9 V. The dissolution behavior of Pt polarized at 1.0 V was the same as that at 0.9 V; however, the dissolution rate of Pt at 0.45 V after oxide formation at 1.0 V was higher than that after oxide formation at 0.9 V. Namely, the cathodic dissolution of Pt was the dominant process, which corresponded well to the dissolution behavior of Pt under potential cycling, as discussed in Section 3.2. As mentioned in the ‘Introduction’, Pt dissolution that occurred below 1.0 V was only detected by the stagnant method using SFC-ICP-MS; however, we succeeded in monitoring the changes in the Pt dissolution rate below 1.0 V in this study.
(a) Potential sequence applied to the Pt and Pd electrodes. The measured (b) Pt and (c) Pd dissolution rates during the potential sequence. The insets are magnified views of the highlighted areas in each Pt and Pd dissolution profile.
Under potentiostatic polarization conditions, Pd dissolution was first detected at 0.7 V (inset of Fig. 7(c)). Pd dissolution started immediately after polarization at 0.7 V and continued at a rate of approximately 0.5 pg cm−2 s−1 for 300 s. Then, the Pd dissolution rate increased with increasing EU. Additionally, the Pd dissolution rate suddenly increased, similar to the sudden increase in the Pt dissolution rate, at the initial stage of polarization. However, the dissolution rate of Pd remained over the detection limit for at least 300 s, although the rate decreased over time. The Pd dissolution rate seemed to remain at a steady state depending on the polarization potential. Polarization tests for longer periods of time should be conducted to determine whether Pd dissolution was finally suppressed below the detection limit. Interestingly, the dissolution of Pd was not accelerated even if the polarization potential was changed from a given potential (>0.7 V) to 0.45 V. This phenomenon was quite different from the dissolution behavior of Pt, thereby supporting that the anodic dissolution of Pd was the dominant process.
3.4 Dissolution mechanism of Pt and PdIn this study, we developed a measurement system with a higher detection sensitivity than a conventional online ICP-MS system and confirmed that Pt and Pd started to dissolve at a more negative potential than that previously reported under potential cycling.29,30) Moreover, we successfully measured the time-resolved dissolution rate of Pt and Pd under potentiostatic polarization below 1.0 V and revealed that the dissolution behavior of Pt and Pd was different under potentiostatic polarization. Based on the above results, the dissolution mechanism of Pt and Pd is discussed.
The onset of Pt dissolution under potential cycling occurred at ≈0.94–1.0 V in the anodic scan. In the past, it has been reported that Pt dissolution in the anodic scan is strongly related to the formation of surface oxides.30,31) As evidenced in the CV of Pt (Fig. 2(c)), the onset of Pt oxide formation was approximately 0.8 V in the anodic scan. The surface coverage of Pt oxide (PtO) became 1 near 1.1 V,44,45) and then repulsive interactions among the adsorbed oxygen atoms induced place-exchange phenomena beyond 1.1 V.46) Adsorbed oxygen atoms snuck into the lattice by the place-exchange reaction between Pt and O atoms, and a Pt oxide with a higher oxidation state (PtO2) formed on the surface:
| \begin{equation} \text{Pt} + \text{H$_{2}$O} \rightarrow \text{PtO} + \text{2H$^{+}$} + \text{2e$^{-}$}, \end{equation} | (2) |
| \begin{equation} \text{PtO} + \text{H$_{2}$O} \rightarrow \text{PtO$_{2}$} + \text{2H$^{+}$} + \text{2e$^{-}$}. \end{equation} | (3) |
| \begin{equation} \text{PtO} + \text{2H$^{+}$} \rightarrow \text{Pt$^{4+}$} + \text{H$_{2}$O} + \text{2e$^{-}$}. \end{equation} | (4) |
In the case of Pt dissolution below 1.1 V, the direct electrochemical dissolution without Pt oxide formation seemed to be the most rational,48) according to the following equation:
| \begin{align} & \text{Pt} \rightarrow \text{Pt$^{2+}$} + \text{2e$^{-}$},\\ & E^{0}\ (\text{Pt/Pt$^{2+}$}) = 1.188 + 0.0295 \log[\text{Pt$^{2+}$}]. \end{align} | (5) |
In addition, a massive dissolution of Pt was observed in the Pt oxide reduction region in the cathodic scan. This potential range was less noble than the estimated equilibrium potential for direct electrochemical dissolution: thus, the dissolution mechanism was quite different between the anodic and cathodic scans. In the cathodic scan, the PtO2 that was formed in the preceding anodic scan was reduced due to the reverse reaction of eqs. (2) and (3); therefore, Pt was considered to partially dissolve by the following equation:20,22,30)
| \begin{align} & \text{PtO$_{2}$} + \text{4H$^{+}$} + \text{2e$^{-}$} \rightarrow \text{Pt$^{2+}$} + \text{2H$_{2}$O},\\ & E^{0}\ (\text{PtO$_{2}$/Pt$^{2+}$}) = 0.837 - \text{0.1182 pH} \\ &\qquad\qquad\qquad\qquad - 0.0295 \log[\text{Pt$^{2+}$}]. \end{align} | (6) |
It has been reported that Pd dissolution is also strongly related to surface oxide formation/reduction.32) The onset of Pd dissolution under potential cycling occurred at ≈0.70–0.73 V in the anodic scan. The onset of Pd oxide formation was approximately 0.7 V in the anodic scan. However, the surface coverage of Pd oxide (PdO) became 1 at 1.4–1.5 V,49) which was different from PtO. Hence, almost no Pd oxide with a high oxidation state (PdO2) formed on the surface in the potential range employed in this study; therefore, Pd directly dissolved from the surface without passing through the oxide by the following equation:48)
| \begin{equation} \text{Pd} + \text{H$_{2}$O} \rightarrow \text{PdO} + \text{2H$^{+}$} + \text{2e$^{-}$}, \end{equation} | (7) |
| \begin{align} & \text{Pd} \rightarrow \text{Pd$^{2+}$} + \text{2e$^{-}$},\\ & E^{0}\ (\text{Pd/Pd$^{2+}$}) = 0.987 + 0.0295 \log[\text{Pd$^{2+}$}]. \end{align} | (8) |
Unlike Pt dissolution under potential cycling, Pd continuously dissolved from the preceding anodic scan to the cathodic scan (Figs. 4 and 6). Nevertheless, it was confirmed that Pd dissolution accelerated in the Pd oxide reduction region. Additionally, almost no PdO2 formed on the surface under potential cycling up to 1.4 V; thus, Pd might not dissolve due to PdO2 reduction shown below:
| \begin{equation} \text{PdO$_{2}$} + \text{4H$^{+}$} + \text{2e$^{-}$} \rightarrow \text{Pd$^{2+}$} + \text{2H$_{2}$O}, \end{equation} | (9) |
| \begin{equation} \text{PdO} + \text{2H$^{+}$} \rightarrow \text{Pd$^{2+}$} + \text{H$_{2}$O}. \end{equation} | (10) |
The dissolution rate of Pd was more than 20 times larger than that of Pt, in good agreement with previous results.32,50,51) This remarkable behavior was explained by the ionic radii difference between the respective cations.52) Pt ions were more easily hydrated than Pd ions because the ionic radii of Pt are smaller than that of Pd, resulting in the increase in Pd dissolution.49) A more detailed mechanism is discussed in Ref. 49).
The dissolution behaviors of Pt and Pd were investigated under potentiostatic polarization and potential cycling, and the following conclusions were drawn.
We acknowledge Y. Ohtsuka (Tokyo Institute of Technology, Open Facility Center, Materials Analysis Division) for assistance with the online ICP-MS analysis.
