DOI: 10.1002/cctc.201901674 Full Papers 1 2 3 Visualizing Potential-Induced Pitting Corrosion of Ultrathin 4 Single-Crystalline IrO2(110) Films on RuO2(110)/Ru(0001)5 6 under Electrochemical Water Splitting Conditions 7 8 Tim Weber,[a, b] Till Ortmann,[a] Daniel Escalera-López,[c] Marcel J. S. Abb,[a, b] Boris Mogwitz,[a, b] 9 Serhiy Cherevko,[c] Marcus Rohnke,[a, b] and Herbert Over*[a, b] 10 11 12 Sophisticated IrO2(110)-RuO2(110)/Ru(0001) model electrodes that is formed by electrochemical oxidation of the metallic Ru13 are employed in the oxygen evolution reaction (OER) under (0001) substrate. The time evolution of the corrosion process at 14 acidic conditions. The potential-induced pitting corrosion of a fixed electrode potential (1.48 V vs. SHE) is followed via cyclic 15 such electrodes is confirmed by a variety of experimental voltammetry and SEM. The passivating IrO2(110) layer results in16 techniques, including scanning electron microscopy (SEM), an “induction period” for the pit growth that is followed by 17 time-of-flight secondary ion mass spectrometry (ToF-SIMS), and rapid corrosion of the RuO2(110)/Ru(0001) substrate. The18 operando scanning flow cell-inductively coupled plasma mass observed narrow and time-independent size distribution rela- 19 spectrometry (SFC-ICP-MS). The structure of the pits is reminis- tive to the mean size of the pits is attributed to a sluggish 20 cent of a cylinder (evidenced by focused ion beam scanning removal of the corrosion products by diffusion across the cracks 21 electron microscopy: FIB-SEM), where the inner surface of the of the pits covering IrO2 layer, leading to steady state corrosion22 pits is covered by hydrous RuO2 (cyclic voltammetry, ToF-SIMS) during a total polarization time of 20 to 60 minutes.23 24 25 1. Introduction materials. Currently, IrO2 and RuO2 are the state-of-the-art 26 catalysts for the OER in acidic environments. RuO2 is the most 27 In electrocatalysis missing or insufficient stability of the active electrocatalyst, albeit with insufficient long-term stability, 28 catalyst[1,2] is a major concern that occurs for instance in the while IrO2 is somewhat less active, [4] but substantially more 29 anodic half reaction of electrochemical water splitting,[3] where stable than RuO .[5–7]2 30 oxygen is evolved (OER: oxygen evolution reaction) as the Over the past decade, fundamental understanding in 31 counter half reaction to the desired hydrogen evolution electrocatalytic activity has been pushed forward by ab initio 32 reaction (HER) at the cathodic side. The OER is a coupled four- methods,[8–10] although the underlying approximations have 33 electron proton process with sluggish reaction kinetics that hardly been validated against benchmark experiments.[11] Ab 34 requires the use of efficient (active and stable) catalyst initio methods for stability issues in electrocatalytic reactions 35 are just emerging.[12–15] Therefore, to deepen our understanding 36 of the molecular processes in electrocatalysis (activity and 37 [a] T. Weber, T. Ortmann, M. J. S. Abb, Dr. B. Mogwitz, Dr. M. Rohnke, Prof. H. Over stability), kinetic and structural studies of well-defined model38 Institute of Physical Chemistry electrode materials with low structural complexity are required. 39 Justus Liebig University These experiments can ultimately serve as benchmarks for 40 Heinrich-Buff-Ring 17 Giessen 35392 (Germany) theoretical ab initio methods, thereby allowing to validate and41 E-mail: Herbert.Over@phys.chemie.uni-giessen.de to advance the theoretical methodology. 42 [b] T. Weber, M. J. S. Abb, Dr. B. Mogwitz, Dr. M. Rohnke, Prof. H. Over Recently, a dedicated model electrode consisting of a 43 Center for Materials Research Justus Liebig University single-crystalline IrO2(110) layer that fully covers a structure-44 [16] Heinrich-Buff-Ring 16 directing template RuO2(110)/Ru(0001) was developed to 45 Giessen 35392 (Germany) study the electrocorrosion under OER conditions.[17] The alter- 46 [c] Dr. D. Escalera-López, Dr. S. Cherevko Helmholtz Institute Erlangen-Nürnberg for Renewable Energy (IEK-11) ations of the crystalline structure and morphology of the47 Forschungszentrum Jülich GmbH IrO2(110) film were followed upon anodic polarization, employ- 48 Egerlandstr. 3 ing a combination of powerful in situ synchrotron radiation 49 Erlangen 91058 (Germany) based techniques including X-ray reflectivity (XRR) and surface 50 Supporting information for this article is available on the WWW under 51 https://doi.org/10.1002/cctc.201901674 X-ray diffraction (SXRD) and most notably ex situ techniques This publication is part of a Special Collection on “Advanced Microscopy and such as scanning electron microscopy (SEM) and X-ray photo-52 Spectroscopy for Catalysis”. Please check the ChemCatChem homepage for electron spectroscopy (XPS). In this study the potential-induced 53 more articles in the collection. pitting corrosion of the ultrathin capping layer IrO2(110) on 54 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. RuO (110)/Ru(0001) was observed when a pulse-rest potential 55 This is an open access article under the terms of the Creative Commons 2 56 Attribution License, which permits use, distribution and reproduction in any protocol was applied, where the electrode potential was medium, provided the original work is properly cited. stepwise increased from 1.30 to 1.94 V vs. the standard hydro- 57 ChemCatChem 2020, 12, 855–866 855 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 855/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers gen electrode (SHE) and held for 80 s each potential. In situ XRR The first sample, denoted here as IrO2(SFC), was studied with a 1 and SXRD experiments were performed in between two pulses scanning flow cell-inductively coupled plasma mass spectrometer 2 at a resting potential of 1.30 V vs. SHE. (SFC-ICP-MS) custom setup following the procedures reported 3 previously.[20] Briefly, a Perkin Elmer NexION 300x ICP-MS was An alternative approach to examine the anodic corrosion 4 calibrated daily with four standard solutions containing known process is to monitor the corrosion products in the electrolyte 5 elemental amounts of Ir and Ru (Merck Certipur) using 20 μg ·L 1 solution with operando scanning flow cell-inductively coupled 187Re and 103Rh as internal standards. After calibration, the ICP-MS 6 plasma mass spectrometry (SFC-ICP-MS), that has shown to be was connected by means of Tygon tubing (380 μm internal 7 extraordinarily sensitive.[7] For the layered model electrode diameter) to the LabVIEW-controlled SFC setup, comprising a 8 system IrO (110)-RuO (110)/Ru(0001), dissolved Ir and Ru spe- Gamry Reference 600 potentiostat (Gamry, USA), a double-junction2 2 9 Ag/AgCl reference electrode compartment (Metrohm, Switzerland; cies can be monitored and quantified separately. 10 outer compartment filled with 0.1 M HClO4, inner compartment Both experimental approaches are complementary to some with standard 3 M KCl electrolyte) and a graphite rod counter 11 extent, although the viewpoint on the corrosion process is electrode compartment (6 mm diameter, 99.995%, Sigma-Aldrich). 12 fundamentally different. While the operando SFC-ICP-MS meth- The V-shaped flow cell employed, CNC machined from a polycar- 13 od addresses the corrosion process from the product side, bonate block (CAM 4–02 Impression Gold, vhf camfacture AG, 14 in situ structural methods in combination with ex situ SEM Germany) presented an opening of 0.033 cm 2, effectively defining 15 the working electrode area. The freshly made 0.1 M HClO electro- follow the process from the educt side and how the structure 4 16 lyte (70%, Suprapur, Merck; pH=1) was pumped from a reservoir (thickness, crystallinity) and morphology of the active part of connected to the SFC setup downstream towards the ICP-MS pump 17 the electrode is varying in the course of corrosion. at a flow rate of 220 μL ·min 1. All electrolytes used in SFC-ICP-MS 18 From the previous corrosion study of the IrO (110)- measurements were prepared using ultrapure water (MilliQ IQ2 19 RuO2(110)/Ru(0001) model electrode [17] there have been impor- 7000, Merck). 20 tant scientific questions left that need to be settled: i) What are After the SFC-ICP-MS experiment the electrode surface was 21 the structure, morphology and chemical composition of the characterized by means of scanning electron microscopy (SEM: 22 pits? ii) How does the initial corrosion process on the micro- Zeiss Merlin apparatus). The SE micrographs were obtained with 23 scopic scale proceed? iii) How do the pits start to form and the secondary electron detectors (InLens or SE2), the accelerating 24 voltage was 2 kV while the probe current was 100 pA. After surface expand both laterally and vertically into the substrate with 25 characterization the sample was further treated galvanostatically at time? a fixed current density of 5 mA·cm 2 (cutoff value of the potential 26 In this study we present ex situ SEM results to follow and to scan during the SFC-ICP-MS experiment). For this purpose the 27 quantify the potential-induced corrosion progress with increas- sample was placed in an electrochemical (EC) glass cell utilizing a 28 ing corrosion time at a fixed electrode potential. This time series hanging-meniscus rotating disk electrode (RDE) setup so that only 29 the IrO2(110) surface was exposed to the electrolyte solution, acan be the starting point to validate a proposed reaction 30 0.5 M H2SO4 solution (pH=0.4) prepared from H2SO4 (Suprapur; mechanism for the corrosion process. The three-dimensional 31 Merck, Darmstadt, Germany) and ultrapure water (Milli-Q Direct 8, structure of the pits is reconstructed from focused ion beam Merck). An Ag/AgCl electrode (sat. KCl) was used as reference 32 scanning electron microscopy (FIB-SEM) and time-of-flight electrode while the counter electrode consisted of a glassy carbon 33 secondary ion mass spectrometry (ToF-SIMS) experiments. rod. The potential values are given with respect to the standard 34 Accompanying SFC-ICP-MS experiments allow us to disentangle hydrogen electrode (SHE) throughout the paper. Prior to the 35 electrochemical measurements the electrolyte solution was de- the contributions of Ir and Ru dissolution/removal in the 36 gassed by flushing with argon while during the measurements the corrosion process that can be correlated with the induced 37 atmosphere above the solution was kept in argon. As galvanostat morphology changes disclosed in the post SEM experiment. either a PGSTAT302 N (Autolab-Metrohm) or a SP-150 (BioLogic 38 Science Instruments) was employed. After galvanostatic treatment 39 the electrode surface was again characterized via SEM. This 40 Experimental Section sequence of galvanostatic treatment and surface characterization 41 was repeated several times. In addition, at certain points the 42 The IrO2(110)-RuO2(110)/Ru(0001) model electrodes were prepared electrode surface was studied via time-of-flight secondary ion mass under ultra-high vacuum conditions as described previously.[16] spectrometry (ToF-SIMS) and focused ion beam scanning electron43 Briefly summarized, the RuO2(110) layer serving as a template was microscopy (FIB-SEM).44 grown epitaxially on a single-crystalline hat-shaped Ru(0001) disk 45 [18,19] For ToF-SIMS analysis a ToF-SIMS 5–100 (IonTOF GmbH, Münster,(4.7 mm diameter, MaTecK, Jülich, Germany). Subsequently 46 Germany) was employed. The primary ion gun was operated iniridium was deposited on the RuO2(110)/Ru(0001) template by burst alignment mode with 25 keV Bi+ ions as analysis species (I= 47 physical vapor deposition (PVD) utilizing an electron beam 0.44 pA @ 55 μs cycle time). The 50×50 μm2 probing area was 48 evaporator (EMF 3, Omicron). Iridium was oxidized at 700 K during 7 scanned with 512×512 pixels. Two depth profiles were carried out 49 deposition (p(O2)=3×10 mbar) and additionally postoxidized in 5 with 500 eV Cs + ions (I=41.1 nA), one in interlaced mode with a 10 mbar of O2. The single-crystalline IrO2(110) films are fully 50 crater size of 170×170 μm2 and another one in non-interlacedcovering the model electrode surface, so that there is no Ru 3d 51 mode with a crater size of 150×150 μm 2. Data evaluation was signal visible in the XP spectra (cf. Figure S1 of the Supporting conducted with the Surface Lab 7.0 software (IonTOF Company). 52 Information). Since the preparation conditions of the samples were The obtained mass resolution is m/~m=180 @ m/z 117.9 (RuO ). 53 identical to those of a previous contribution we assume the [17] For distinct mass assignment exemplary measurements were 54 thickness of the IrO2(110) films to be ~10 nm. The model conducted in delayed extraction mode. Here the mass resolution is electrodes were studied in two different ways which are explained 55 m/~m=1393 @ m/z 117.9 (RuO ). FIB-SIMS cuts were prepared in the following. 56 with 30 keV Ga+ ions, I=5.6 nA. Core milling of the FIB craters was 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 856 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 856/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers carried out with a 300 μm aperture and dwell time of 50 ms with 1 512×512 pixels and 10 milling scans. For fine milling the 100 μm 2 aperture was used (dwell time 30 ms, 512×512 pixels, 3 milling 3 scans). 4 For FIB-SEM experiments a DualBeam XEIA3 Triglav microscope 5 (TESCAN) with a Xe-plasma FIB was employed. The milling and 6 polishing was done in one step with Xe+ ions at an energy of 7 30 keV (I=18 nA). SE micrographs of the FIB cut were subsequently 8 recorded within the Zeiss Merlin apparatus. 9 On the second sample, denoted here as IrO2(SEM), the progress of 10 the potential-induced pitting corrosion was systematically studied 11 by means of cyclic voltammetry and SEM. The experimental setup for electrochemistry was the same as described above for the 12 galvanostatic treatment. As potentiostat a PGSTAT302 N (Autolab- 13 Metrohm) was employed, equipped with modules enabling electro- 14 chemical impedance spectroscopy (EIS) and true analog voltage 15 sweeps. For a systematic investigation of the potential-induced 16 pitting corrosion the model electrode was polarized to 1.48 V vs. 17 SHE (the potential at which the pitting corrosion starts [17]) for a certain time. Subsequently, the sample was characterized via cyclic 18 voltammetry, removed from the EC cell and transferred to the SEM 19 apparatus (Zeiss Merlin) to obtain SE micrographs of the electrode 20 surface. The micrographs were obtained with the secondary 21 electron detectors (InLens or SE2), the acceleration voltage was 22 2 kV and the probe current was 100 pA. This protocol was repeated 23 several times up to a total polarization time of 82 min. Within the sequence of experiments it was realized that macroscopic oxygen 24 bubbles formed during the polarization. For this reason, during the 25 last four polarization steps the sample was rotated to remove 26 evolving, macroscopic oxygen bubbles. 27 28 29 2. Experimental Results 30 31 Recently, the IrO2(110)-RuO2(110)/Ru(0001) system was em- 32 ployed as a model electrode to gain insight into corrosion 33 Figure 1. SE micrograph (InLens) of the pitted IrO2(110)-RuO2(110)/Ru(0001) processes under OER conditions. The IrO2(110) film exhibits a model electrode surface (top) and corresponding schematic, three-dimen- 34 roughness on the mesoscale (yet the terraces are atomically sional representation (bottom). 35 flat) in a regular array reminiscent of “roofs”:[16,17] there is a 36 sequence of ascending terraces up to the top of the roof that is 37 followed by a symmetric descending series of terraces. For this 2.1. IrO2(SFC) 38 reason we have introduced the term “rooflike structure” in our 39 previous study.[17] Due to this rooflike structure of the IrO2(110) The SFC-ICP-MS setup allows to follow operando electrochem- 40 film the corrosion of the model electrode could be studied on ical dissolution processes. This technique was applied to the 41 different length scales ranging from atomic (in situ SXRD/XRR) IrO2(110)-RuO2(110)/Ru(0001) model electrode to gain insight 42 to mesoscale (ex situ SEM).[17] Based on these studies it was into the corrosion process from analysis of dissolved Ir and Ru 43 concluded that the IrO2(110) film is remarkably stable against species in the electrolyte solution. An overlay of the acquired 44 anodic corrosion in the OER potential region, while the data is provided in Figure 2a. The dissolution profiles at a 45 degradation of the model electrode proceeds via potential- magnified scale can be found in Figure S2 of the Supporting 46 induced pitting corrosion. Ex situ SEM experiments revealed Information (SI). 47 that the corrosion is initiated by Ir dissolution at so-called The top and middle panels of Figure 2a display the 48 “surface grain boundaries” where rotational domains of the electrode potential testing protocol selected for SFC-ICP-MS 49 IrO2(110) film meet. Once the underlying RuO2(110)/Ru(0001) is measurements and the corresponding recorded current den- 50 reached, pitting corrosion is accelerated due to the instability of sities normalized to the geometric surface as a function of time, 51 RuO2 and Ru under these anodic conditions. [17] A schematic respectively, while the bottom panel depicts the resulting 52 representation of resulting pits is shown in Figure 1. dissolution rates of Ir (green) and Ru (blue). First, the electrode 53 is held at open-circuit potential (OCP) for 5 min (cf. top of 54 Figure 2a) to approach and contact the SFC with the IrO2(110)- 55 RuO2(110)/Ru(0001) model electrode, as seen by the sudden 56 electrode potential drop from ca. 2 V to ca. 0.69 V vs. SHE after 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 857 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 857/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Figure 2. (a) Overlaid SFC-ICP-MS data of the IrO2(110)-RuO2(110)/Ru(0001) model electrode. [M] refers here to the monitored metal element dissolution, in 22 this work either Ir or Ru. (b) SE micrograph (InLens) of the freshly prepared electrode surface and (c) SE micrograph (InLens) of the electrode surface after the 23 SFC-ICP-MS experiment. 24 25 26 contact. Since no dissolution peak arising from SFC contact is comment on the calculation of the monolayer normalized 27 visible in the ICP-MS data (cf. bottom of Figure 2a), this dissolution can be found in the SI). This result is consistent with 28 indicates that the IrO2(110)-RuO2(110)/Ru(0001) electrode does the process of potential-induced pitting corrosion, where the 29 not present intrinsically unstable surface defects or native total area of cracks in the IrO2(110) film is small, while the 30 oxides prone to chemical dissolution once in contact with the under-corroded region in the Ru(0001) substrate is large and 31 acidic electrolyte employed.[21] Subsequently the electrode is set deep. This conclusion is corroborated by means of post SEM 32 to a potential of 1.141 V vs. SHE (1.2 V vs RHE, pre-OER analysis of the model electrode surface that reveals small pits 33 potential) for 3 min (cf. top of Figure 2a). Previous studies have (cf. Figure 2c and Figure S3). 34 shown that cathodic dissolution is negligible for thermally After the SCF-ICP-MS experiments the sample was further 35 prepared IrO ,[7]2 whereas for anodically formed IrO2 its exper- treated galvanostatically for 4.5 and 17.5 min in total at 36 imental onset potential is 1.041 V vs. SHE.[22] Induced by this 5 mA ·cm 2 and studied by means of ToF-SIMS and FIB-SEM to 37 potential variation there are dissolution peaks (cf. bottom of analyze the shape and the chemical composition of the pits. 38 Figure 2a) arising for both Ir (0.065�0.004 ng · cm 2) and Ru With a focused ion beam the electrode surface was cut 39 (0.4�0.2 ng · cm 2). These peaks decline with time to the perpendicular to the surface (either within the ToF-SIMS or the 40 baseline within the potentiostatic hold, indicating that steady- FIB-SEM apparatus) to be able to study the depth of the pit. SE 41 state dissolution is not present. The transient dissolution may micrographs were then acquired and are shown in Figure 3. 42 be due to restructuring of the oxides[23] and/or dissolution of Figure 3a shows a SE micrograph of the electrode surface 43 iridium exposed in surface grain boundaries.[17] Next, the prior to the ion beam cutting. Clearly visible is the rooflike 44 electrode potential is scanned positively at a 10 mV· s 1 scan structure of the IrO2(110) film both at the fully intact surface 45 rate until a cutoff current density value of 5 mA·cm 2 is reached and the locally undercut region. This is consistent with the 46 (cf. middle of Figure 2a) after which the potential is set back to previous finding that the rooflike structure on intact domains 47 OCP. During this potential scan dissolution is observed for both was not altered due to anodic polarization in the OER potential 48 Ir and Ru (cf. bottom of Figure 2a), with dissolution onset region.[17] Also clearly visible is a pit in form of the typical ring- 49 potentials estimated at 1.50�0.02 V and 1.47�0.1 V, respec- shaped contrast around the fractured IrO2(110) film which is 50 tively. It is noteworthy that the total dissolution for Ir (0.24� assumed to be due to erosion of the underlying Ru substrate. 51 0.04 ng ·cm 2) is almost two orders of magnitude lower than However, in the previous publication[17] this undercut region 52 that for Ru (20.4�0.8 ng · cm 2). was not explored in detail. With FIB cuts perpendicular to the 53 Given the well-defined, layered and highly-crystalline nature electrode surface it is possible to open the pits and image them 54 of the IrO2(110)-RuO2(110)/Ru(0001) electrode, we can also with SEM from the side (cf. Figure 3b, c). The dashed white lines 55 normalize the dissolution to the working electrode area, with indicate the edges of the FIB cuts, while the dotted red line 56 values of 0.075% and 12.1% for Ir and Ru, respectively (a illustrates the shape of the pit. The covering IrO2(110) film is 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 858 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 858/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers 1 2 3 4 5 6 7 8 9 10 Figure 3. (a) SE micrograph (InLens) of a pit at the electrode surface. (b) SE micrograph (SE2, stage tilted by 40°) of a pit truncated by a FIB cut (Ga+ ions) 11 done within the ToF-SIMS apparatus. The FIB cut was applied after 4.5 min at 5 mA ·cm 2. (c) SE micrograph (InLens, stage tilted by 30°) of a pit truncated by a FIB cut (Xe+ ions) done within the FIB-SEM apparatus. The FIB cut was applied after 17.5 min at 5 mA ·cm 2. The dashed white lines indicate the edges of the 12 FIB cuts while the dotted red line illustrates the shape of the pit. 13 14 15 removed by the focused ion beam, but the local erosion of the 16 Ru(0001) substrate becomes now clearly visible, whose three- 17 dimensional shape is reminiscent of a cylinder. From Figure 3b, 18 c it seems that the pit walls perpendicular to the electrode 19 surface are porous, while the base area is plain. The observed 20 porosity is reconciled with electrochemically formed hydrous 21 RuO2 at the pit walls. From the profile of the pit in the SEM 22 image (cf. Figure 3b) the radius fairly coincides with its depth. 23 This might indicate that the corrosion rates of the Ru(0001) 24 25 facet parallel and the Ru(10 1 0) facet normal to the IrO2(110) 26 film are at least comparable. 27 To gain further information on the chemical composition 28 within the pits, ToF-SIMS was utilized. In principle, we are able 29 to visualize the chemical composition of the layered structure 30 of the IrO2(110)-RuO2(110)/Ru(0001) model electrode (cf. Figure 31 S4). However, in the following we rather focus on the chemical 32 composition of the pits than on the layered structure of the 33 model electrode. 34 In Figure 4 the IrO 2 (green) and RuO (red) mass signals of 35 ToF-SIMS depth profiles are overlaid; additional mass signals are 36 provided in Figures S5 and S6. The profile in Figure 4a is 2.9 μm 37 deep, depicting the pits completely in terms of their depth. The 38 profile was recorded in interlaced mode, therefore the IrO2(110) 39 layer does not fully cover the surface anymore. In contrast, the 40 Figure 4. 3D maps recorded within two depth profiles utilizing ToF-SIMS of profile (1 μm deep) shown in Figure 4b was recorded in non- 41 the model electrode after galvanostatic treatment at 5 mA ·cm 2 for 17.5 min interlaced mode so that the IrO2(110) layer is visible within the in total: a) interlaced mode and b) non-interlaced mode. The maps show the 42 total scanned area. The difference between interlaced and non- overlaid mass signals of IrO 2 (green) and RuO (red). The signal intensity 43 increases from black (low intensity) to the respective color (high intensity). interlaced mode is the usage of the sputtergun. In interlaced 44 mode the sputtergun operates during the deadtime, that is 45 when the ions are passing the time-of-flight tube. After 46 scanning half of the sample area the topmost IrO2 layer is surface and the pits formed due to anodic corrosion. Even the 47 already removed by sputtering. In non-interlaced mode a C mass signal can be employed to image the pits, due likely to 48 complete analysis scan of one layer is carried out before the CO2 that is dissolved in the hydrous RuO layer [24,25] 2 at the inner 49 sputtergun erodes the sample area further. The IrO 2 mass surface of the pits. 50 signal with its high intensity at the top region clearly displays The pits are not visualized as structures with zero intensity 51 the IrO2(110) film at the electrode surface. In contrast, the RuO but “filled”. This can be explained by the primary ion beam 52 mass signal illustrates channel-like structures that are assigned impinging on the sample surface at an angle of 45°. For this 53 to be the pits. The detection of RuO ions indicates the reason in the pit region the beam ends up at the inner wall of a 54 presence of electrochemically formed hydrous RuO2 at the inner pit. Due to their trajectory the emitted secondary ions are 55 surface of the pits. Additionally, the O mass signal (cf. Figures detected in the center of the pit. A graphical representation of 56 S5 and S6) can be utilized to image both the oxide film at the this effect can be found in the SI (cf. Figure S7). 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 859 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 859/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers 2.2. IrO2(SEM) time of 82 min the pit’s diameter reaches about 3–4 μm. The 1 IrO2(110) film covering the pits exhibits cracks which might be 2 The IrO2(110)-RuO2(110)/Ru(0001) model electrode was polar- caused by mechanical stress due to oxygen bubbles forming in 3 ized at 1.48 V vs. SHE and a time series of SE micrographs of the the pits or by the relief of internal strain in the IrO2(110) layer. In 4 electrode surface was acquired. Already from the current addition, the number of pits is increasing with total polarization 5 increase in the single chronoamperometry experiments (cf. time as summarized in Figure 6. However, with increasing their 6 Figure S8) we can conclude that the electrode surface corrodes. diameter the pits are merging at a certain point (cf. Figure 6, 7 The increase in current density with polarization time is due after 72 min and 82 min). 8 partly to the high OER activity of hydrous RuO2 that is formed The SE micrographs are quantitatively evaluated utilizing 9 at the pit walls during corrosion. After each polarization step the software package ImageJ[26] (v. 1.52a). After each polar- 10 (for a specific time period) the electrode surface was charac- ization step ten SE micrographs were recorded at a magnifica- 11 terized by means of cyclic voltammetry and ex situ SEM. tion of 3,000× for evaluation. When the area of the pits had 12 Figures 5 and 6 compile SE micrographs for various total increased so that they became visible at lower magnifications, 13 polarization times. ten additional SE micrographs each were recorded at magnifi- 14 As visualized in Figure 5 the first pits appear after 630 s of cations of 1,000× and 2,000× additionally. The determination 15 polarization with a diameter of a few hundred nanometers, so of pit area and the number of pits is based on the intensity 16 that it seems that there is a kind of “induction period” of at contrast due to the pits and the resulting differences in their 17 least 310 s which is needed for the pits to form. Between 310 s grayscale. ImageJ allows to select a particular grayscale range 18 and 630 s no SE micrographs are available since at the of a grayscale picture and to convert it to a black-and-white 19 beginning of the time series the duration of the single picture. After that the pits are enumerated applying the 20 polarization steps was doubled each step (starting from 10 s) software’s feature “analyze particles” while the pixel size is 21 until the first pits appeared. From Figure 5 we recognize that calibrated to the scale bar of the micrographs. In order to avoid 22 the lateral shape of the pits is approximately a disk, its diameter erroneous counts, structures on the surface appearing black or 23 increases with total polarization time. After a total polarization in a similar grayscale range as the pits (but not being pits) are 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 Figure 5. SE micrographs (all InLens) of the model electrode surface in dependence of the total polarization time at 1.48 V vs. SHE. The first small pits appear 56 after 630 s. 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 860 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 860/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Figure 6. SE micrographs (all InLens) of the model electrode surface in dependence of the total polarization time at 1.48 V vs. SHE. 36 37 38 cut out prior to the evaluation and a threshold value for the number of pits rises abruptly. With total polarization time both 39 minimal pit area is set. In addition, fragmentarily depicted pits the number of pits and the mean pit area increase first (cf. 40 at the edges of the micrographs are excluded from analysis. Figure 7a, b and Figures 5, 6), while between 25 min and 41 Moreover, the voltammetric charge is derived from the 50 min (ii) the pit density remains roughly constant albeit the 42 cyclic voltammograms (CVs). Here CVs with a potential region mean pit area is still growing. Therefore, the existing pits are 43 from 0.5–1.3 V vs. SHE are used so that only the capacitive growing in diameter while only few pits are newly formed. With 44 current with no OER contribution is evaluated. Figure 7 provides ongoing polarization the pit density decreases (iii) which can be 45 a survey of the derived data, (a)–(c) are based on the evaluation explained by merging pits (cf. Figure 5). This causes ImageJ to 46 of the SE micrographs, whereas (d) gives the voltammetric count pits connected by at least one pixel as a single pit so that 47 charge derived from the CVs. the number of pits apparently declines. However, within the 48 The mean pit area and the total pit area relative to the time period (ii) (cf. Figure 7b) there is no significant merging of 49 whole electrode surface area (cf. Figure 7a, c) are increasing pits observed. 50 steadily with polarization time after a kind of “induction period”. With SEM we monitor the changes in surface morphology of 51 The pit density, that is the number of pits per surface area, the model electrode, while electrochemical alterations are 52 shows a different time evolution: after the “induction period” monitored with cyclic voltammetry. Figure 7d shows the 53 the pit density increases, then reaches a plateau and declines voltammetric charge as a function of the total polarization time. 54 afterwards (cf. Figure 7b). From these data one can derive a The series of cyclic voltammograms can be found in Figure S9. 55 “mechanism” for the potential-induced pitting corrosion: in the The mean pit area, the total pit area and the voltammetric 56 beginning (i) when the electrode surface starts to form pits the charge are steadily increasing with total polarization time. 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 861 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 861/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Figure 7. (a) Mean area of the pits, (b) number of pits per surface area, (c) relative pit area with respect to the total surface area and (d) voltammetric charge 29 as derived from the CVs as a function of the total polarization time at 1.48 V vs. SHE. The legend in (a)–(c) gives the different magnifications of the respective SE micrographs used for the evaluation. 30 31 32 33 When comparing the mean pit area and the voltammetric From visual inspection of Figure 6 the size distribution of 34 charge as a function of polarization time in Figure 7a, d one the pits is surprisingly narrow throughout the time series. This 35 can recognize a very similar behavior. This is reconciled with a impression can further be analyzed via a time series of 36 linear behavior of the voltammetric charge, derived from the histograms based on the SEM time series, similar to those 37 CVs, as a function of the mean pit area (cf. Figure 8). shown in Figure 6. Each histogram is given in relative units, i. e., 38 size of the pits relative to the averaged size, and a description 39 of how they are derived from the SE micrographs can be found 40 in the SI (cf. Figure S10). A total polarization time higher than 41 52 min was not subject to analysis since the pits start merging 42 (cf. Figure 7b (iii)). The histograms are fitted to normal 43 distributions (Gaussian), thereby providing the relative standard 44 deviation (cf. Figure 9) as a measure for the narrowness of the 45 pit size distribution. 46 These standard deviations of the pit size distributions as a 47 function of the total polarization time, depicted in Figure 9, are 48 roughly constant indicating that the pit size distribution does 49 not change very much with the total polarization time. 50 51 52 3. Discussion 53 54 55 Figure 8. Voltammetric charge as derived from the cyclic voltammograms as The present contribution provides an in-depth study of the a function of the mean pit area (as derived from the SEM series at a recently recognized potential-induced pitting corrosion 56 magnification of 3,000×). A linear correlation can be established. process[17] of ultrathin single-crystalline IrO2(110) films sup- 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 862 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 862/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers dissolution signal is visible with RuO2(110)/Ru(0001) exhibiting a 1 total dissolution around two orders of magnitude higher than 2 that of the IrO2(110) layer (20.4�0.8 ng · cm 2 vs. 0.24� 3 0.04 ng ·cm 2). This trend is in good agreement with analogous 4 experiments previously performed on thermally-oxidized sput- 5 tered Ir and Ru thin films, where RuO2 and IrO2 total dissolution 6 was 1.9�0.3 ng · cm 2 and 0.07�0.01 ng ·cm 2, respectively.[7] 7 The total dissolution values observed for the IrO2(110) and the 8 RuO2(110)/Ru(0001) are, however, one order of magnitude 9 higher than the aforementioned but still significantly lower 10 than those found at an identical voltage ramp protocol for 11 metallic Ru (253�11 ng · cm 2) and Ir (8.4�0.3 ng · cm 2),[7] as 12 well as electrochemically-grown hydrous Ir oxide films, which 13 range from 8.6 to 25.1 ng ·cm 2 depending on the number of 14 growth cycles applied to a pristine Ir electrode.[27] Therefore, we 15 can conclude that the discrepancies regarding total Ir and Ru 16 dissolution with respect to those found for rutile RuO2 and IrO2 17 Figure 9. Relative standard deviation of the normal distribution of the relative pit size as a function of the total polarization time. The dashed line surfaces arise from the potential-induced pitting corrosion18 indicates the mean value. mechanism. Besides Ir dissolution from IrO2(110) surface grain 19 boundaries mechanical instability of the IrO2(110) layer above 20 the pits may explain the higher total Ir dissolution for IrO2(110)- 21 RuO2(110)/Ru(0001), where the experimental Ir dissolution onset 22 ported on a RuO2(110)/Ru(0001) template, serving as model (1.50�0.02 V) is in good agreement with the potential under 23 electrode. In situ synchrotron radiation based techniques which pitting corrosion was observed (1.48 V vs. SHE).[17] The 24 (SXRD, XRR) revealed that the IrO2(110) film was not altered pitting corrosion process then propagates through the 25 upon anodic polarization up to 1.94 V vs. SHE in terms of RuO2(110) layer down to the underlying Ru(0001) substrate, 26 thickness, periodicity and domain size. The integral intensity yielding a total dissolution higher than that expected for rutile 27 derived from the in situ SXRD data decreased by around 85% RuO2 given the higher instability of the metallic Ru surface, 28 though, while ex situ XPS indicated that only around 50% of which is more prone to electrodissolution under OER 29 the initial Ir amount on the electrode surface was lost.[17] This potentials.[7] 30 apparent contradiction was unraveled utilizing ex situ SEM After the SFC-ICP-MS experiment the IrO2(SFC) sample was 31 which revealed potential-induced pitting corrosion, initiated at further treated galvanostatically at a fixed current density of 32 so-called surface grain boundaries, to be operative. Therefore, it 5 mA ·cm 2 for certain time steps and the pits were character- 33 was concluded that the IrO2(110) film was disordered rather ized ex situ via FIB-SEM and ToF-SIMS. With FIB-SEM we were 34 than dissolved during the in situ experiments.[17] However, the able to demonstrate that the Ru substrate is indeed eroded 35 process of potential-induced pitting corrosion of IrO2(110)- locally, virtually forming cylindrical pits with the depth of the 36 RuO2(110)/Ru(0001) itself has not been studied in detail yet. The pits roughly equal to its radius r. The inner surface of the pits is 37 present contribution aims at the elucidation of the corrosion covered by hydrous RuO2 as evidenced by 3D mapping via ToF- 38 process (cyclic voltammetry, SEM, SFC-ICP-MS) and of the SIMS (cf. Figure 4), the appearance of the pits in SEM (cf. 39 structure and chemical composition of the pits (FIB-SEM, ToF- Figure 3b), and by the increase in voltammetric charge with 40 SIMS). In order to gain insights, we first employ operando SCF- total polarization time (cf. Figure 7d). Also the ICP-MS data 41 ICP-MS measurements to quantify the amount of Ir and Ru indicate the presence of hydrous RuO2 since the total Ru 42 dissolved due to anodic polarization. Second, the three-dimen- dissolution (20.4�0.8 ng · cm 2) is significantly higher than what 43 sional morphology of the pits is revealed by FIB-SEM and ToF- one would expect for rutile RuO2 (1.9�0.3 ng ·cm 2). According 44 SIMS, with ToF-SIMS providing information on the chemical to the Pourbaix diagram of Ru[28] the metal undergoes electro- 45 composition within the pits. Third, the temporal evolution of chemical oxidation to hydrous RuO2 via a Ru(OH)3 species and 46 the forming pits is followed by means of cyclic voltammetry at even higher potentials dissolves via formation of soluble Ru 47 and SEM. (VI) and Ru(VIII) species like RuO 24 and RuO4, respectively. [28,29] 48 The SFC-ICP-MS experiment reveals two dissolution signals The thermodynamic stability is reconciled with the time 49 of Ir and Ru (cf. Figure 2a). The first signal, which declines to evolution of the pit morphology. As soon as the passivating 50 the baseline after a certain time, is due likely to transient IrO2(110) layer breaks down, the Ru metal substrate starts to be 51 dissolution[23] when the electrode potential is held at 1.141 V vs. electrochemically corroded forming first hydrous RuO2 that is 52 SHE. This transient dissolution might be correlated with surface further oxidized to form soluble Ru species in high oxidation 53 grain boundaries with highly under-coordinated Ir atoms that state. In this way the pits form. The lateral and vertical growth 54 are prone to dissolve. Here the first pits form first, so that a Ru of pits proceeds via corrosion of the inner surface of the pits, 55 dissolution signal emerges as well. When ramping the electrode forming a covering hydrous RuO2 layer from where soluble Ru 56 potential up to 1.508�0.005 V vs. SHE (at 5 mA ·cm 2) a second species are dissolved. 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 863 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 863/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers From the cylindrical form of the pits we can deduce that the For large polarization times the corrosion process should 1 corrosion process perpendicular to the surface normal in ( reach steady state conditions. The reasons behind this simple 2 model are (i) that the driving force according to Nernst 3 10 1 0) and symmetry-equivalent directions is almost independ- equation diminishes with increasing concentration of dissolved 4 ent of the lateral orientation. Would corrosion along the ( Ru species (as discussed above) and (ii) that the crack area 5 10 1 0) and symmetry-equivalent directions be much slower where the dissolved Ru species can escape from the pit is small 6 than other in-plane directions, a hexagonal instead of a disk compared to the wall area of the pit, where metallic Ru is 7 shape appearance of the pit base would be expected. Quite in electrochemically oxidized and dissolved and the processes to 8 contrast, the corrosion process along the surface normal in reach steady state are fast enough. This means that diffusion of 9 (0001) direction leads to quite flat base area, indicating that a the dissolved Ru species in the pits needs to be fast and the 10 partially corroded (0001) plane is unstable and rapidly transport outside the pits should be even faster to maintain a 11 corroded. constant concentration profile across the crack. 12 From Figure 8 it is clear that the voltammetric charge is A rough estimation based on a typical diffusion coefficient 13 proportional to the mean pit area and therefore proportional to in aqueous solutions of 10 5 cm2 · s 1 leads to a diffusion length 14 r2, with r being the radius of a pit. The increase in capacitive of some 100 μm in 100 s, the time resolution of our corrosion 15 current and hence the voltammetric charge is indicative of experiment. Since the pit dimensions of the undercutting 16 hydrous RuO2, a well-known supercapacitor material, [30] that is region is in the order of less than 10 μm even after a 17 electrochemically formed at the pit walls[17,24] as evidenced by polarization time of 1 h, steady state can quickly be accom- 18 the FIB-SEM data in Figure 3 and the 3D maps obtained from plished by diffusion. 19 ToF-SIMS in Figure 4. If the shape of a pit is assumed to be Assuming that the pits are cylindrical with radius r and 20 cylindrical with the height h of the cylinder being at least depth h the molar amount dn of dissolved Ru species per time 21 proportional to the radius r of the pits (FIB-SEM, cf. Figure 3b), dt is given by Equation (1) 22 then the inner surface area of the pits and hence the 23 voltammetric charge is proportional to r2 as observed in dn dr 24 Figure 8. dt ¼ c � 4 � p � rðtÞ � hðtÞ � dt¼ jdif � Acrack¼ const: (1) 25 As summarized in Figure 9, the standard deviation of the 26 relative size distribution of the pits does not vary with polar- with c being a constant, describing the molar concentration of 27 ization time. Since for a “simple” growth mechanism one would Ru in the sample. In steady state the produced amount needs 28 expect a broadening of the pit size distribution with growth to be identical to that which diffuses through the crack with the 29 time (= total polarization time), we infer that the growth area Acrack and diffusion flux jdif, given by Fick’s second law. This 30 “mechanism” of the pits is quite complex. This finding in corresponds to a mass balance. 31 conjunction with an apparent “induction period” is indicative of From experiments we know that the depth of the pits h is 32 at least three processes which take place on very different time proportional to the radius of the pits r. Therefore, we can easily 33 scales. During the “induction period” iridium from the IrO2(110) solve the differential Equation (1) and obtain Equation (2) 34 film might be dissolved at surface grain boundaries, therefore 35 locally thinning the “passivating” oxide layer. This process is rðtÞ3 rðt Þ3 ¼ c00 � ðt t0Þ (2) 36 presumably slow, since the corrosion rate derived from the SFC- 37 ICP-MS experiment is quite low for the IrO2(110) film (cf. with c’ being a constant. This relation can be reconciled with 38 Figures 2and S2). Once the significantly less stable RuO2(110)/ the experimental values: in the polarization time window 39 Ru(0001) substrate is exposed to the electrolyte solution, the between 25 min and 65 min the experimental data from 40 corrosion process is strongly accelerated: the dissolution of Ru Figure 7a can be linearly approximated by r(t) as a function of 41 proceeds about two orders of magnitude faster than that of Ir t1/3 (cf. Figure 10). 42 (cf. Figures 2and S2). However, to achieve the observed narrow The range where Equation (2) is applicable is roughly the 43 and constant relative size distribution of pits, the initially fast time range where the number of pits is constant. From 44 undercorrosion process must slow down for larger pits. The experiment we obtain r(t0)=0.25 μm and t0=25 min. 45 reason for this deceleration of the corrosion process of Ru may 46 be attributed to a sluggish removal of the corrosion products 47 through the narrow cracks in the IrO2 layer, namely soluble Ru 4. Conclusions 48 species with high oxidation states like RuO 2 and RuO .[28,29]4 4 As 49 a consequence, the concentration of Ru-based ions is high in Recently, the degradation of IrO2(110)-RuO2(110)/Ru(0001) mod- 50 the pits, shifting the reversible half-cell potential(s) Urev of the el electrodes under oxygen evolution conditions in an acidic 51 respective Ru redox transition(s) positively. For this reason, the environment was shown to proceed via potential-induced 52 overpotential(s) η, defined as η=U–Urev with U being the pitting corrosion. [17] The present contribution is dedicated to 53 applied electrode potential, decrease(s) and the resulting answer important remaining scientific questions regarding 54 corrosion current (density) decreases exponentially, according structure/morphology and chemical composition of the pits 55 to Butler-Volmer equation,[11,31] thus narrowing the size distribu- and the time evolution of the corrosion process on the 56 tion. microscopic scale. First, FIB-SEM and ToF-SIMS evidence the 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 864 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 864/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers In the steady state regime the corrosion rate is determined by 1 diffusion of dissolved Ru species through the cracks of the pits. 2 3 4 Supporting Information 5 6 – XP spectrum of the freshly prepared model electrode surface 7 (Ir 4d and Ru 3d binding energy region) 8 – zoomed-in view of the dissolution profile 9 – comparison of SE micrographs before and after the SFC-ICP- 10 MS experiment 11 – ToF-SIMS 3D map illustrating the layered structure of the 12 model electrode 13 – ToF-SIMS 3D maps for various anion mass signals 14 – current density as a function of the total polarization time at 15 1.48 V vs. SHE 16 Figure 10. The mean pit radius r as a function of t1/3, with t being the total – cyclic voltammograms recorded after polarization and used17 polarization time. for determining the voltammetric charge 18 – time series of histograms of the relative pit size distribution 19 20 21 existence of cylinder-shaped pits at the model electrode during Acknowledgements 22 the initial stage of potential-induced pitting corrosion. Electro- 23 corrosion of the unstable Ru(0001) substrate is in-plane nearly We acknowledge financial support by the BMBF (project: 24 independent of the direction. Operando analysis of the 05 K2016-HEXCHEM) and by DFG (SPP2080: Ov21-16 and CH1763/ 25 corrosion products by means of SFC-ICP-MS reveals dissolution 3-1). 26 of Ir and Ru, therefore the initially fully covering IrO2(110) film 27 has to be (locally) dissolved or mechanically disrupted so that 28 the electrolyte solution can reach the unstable RuO2(110)/Ru Conflict of Interest 29 (0001) substrate. Since post SEM analysis reveals small pits in 30 the model electrode surface we infer that the two orders of The authors declare no conflict of interest. 31 magnitude higher Ru than Ir dissolution signal supports pitting 32 corrosion to be operative from the “electrolyte-point of view”. 33 Keywords: anodic corrosion · electron microscopy · mass Utilizing ToF-SIMS the presence of hydrous RuO2 within the pits, 34 spectrometry · oxygen evolution reaction (OER) · single- as already suggested by XPS data,[17] is evidenced. The increase 35 crystalline electrodes in voltammetric charge as derived from the cyclic voltammo- 36 grams is attributed to the formation of hydrous RuO2, a well- 37 known supercapacitor material.[30] From the linear correlation of 38 [1] M. Carmo, D. L. Fritz, J. Mergel, D. Stolten, Int. J. Hydrogen Energy 2013, the voltammetric charge with the mean pit area, we conclude 38, 4901–4934. 39 that hydrous RuO2 grows only at the inner surface of the pits. [2] S. L. Scott, ACS Catal. 2018, 8, 8597–8599. 40 [3] T. Reier, H. N. Nong, D. Teschner, R. Schlögl, P. Strasser, Adv. Energy Changes in surface morphology and electrochemical behav- 41 Mater. 2017, 7, 1601275. ior at 1.48 V vs. SHE due to potential-induced pitting corrosion [4] K. A. Stoerzinger, L. Qiao, M. D. Biegalski, Y. Shao-Horn, J. Phys. Chem. 42 are studied systematically as a function of polarization time via Lett. 2014, 5, 1636–1641. 43 [5] R. Kötz, S. Stucki, Electrochim. Acta 1986, 31, 1311–1316. cyclic voltammetry and SEM. The pits are steadily increasing 44 [6] N. Danilovic, R. Subbaraman, K.-C. Chang, S. H. Chang, Y. J. Kang, J. with polarization time, as does the voltammetric charge. In Snyder, A. P. Paulikas, D. Strmcnik, Y.-T. Kim, D. Myers, V. R. Stamenkovic, 45 contrast, the pit density is increasing after an “induction N. M. Markovic, J. Phys. Chem. Lett. 2014, 5, 2474–2478. 46 [7] S. Cherevko, S. Geiger, O. Kasian, N. Kulyk, J.-P. Grote, A. Savan, B. R. period”, then reaching a plateau and decreasing afterwards. The 47 Shrestha, S. Merzlikin, B. Breitbach, A. Ludwig, K. J. J. Mayrhofer, Catal. decrease can be explained by merging pits, therefore reducing Today 2016, 262, 170–180. 48 the number of isolated pits. The lateral relative size distribution [8] J. Rossmeisl, A. Logadottir, J. K. Norskov, Chem. Phys. 2005, 319, 178– 49 184. of the pits is surprisingly narrow and practically constant for all 50 [9] A. B. Anderson, Electrocatalysis 2012, 3, 176–182. polarization times, indicating a complex growth behavior with [10] Z. W. Seh, J. Kibsgaard, C. F. Dickens, I. Chorkendorff, J. K. Norskov, T. F. 51 at least three processes on very different time scales involved: Jaramillo, Science 2017, 355, eaad4998. 52 [11] K. S. Exner, I. Sohrabnejad-Eskan, H. Over, ACS Catal. 2018, 8, 1864– After the slow induction period, where the passivating IrO2(110) 53 1879. layer is locally eroded, fast undercorrosion of the unstable [12] T. Binninger, R. Mohamed, K. Waltar, E. Fabbri, P. Levecque, R. Kötz, T. J. 54 RuO2(110)/Ru(0001) sets in. The dissolved Ru species in the pits Schmid, Sci. Rep. 2015, 5, 12167. 55 [13] J. A. Herron, A. Morikawa, M. Mavrikakis, PNAS 2016, 113, E4937-E4945. are trapped, thereby decelerating the further growth of the pits. 56 [14] C. Spöri, J. T. H. Kwan, A. Bonakdarpour, D. P. Wilkinson, P. Strasser, Angew. Chem. Int. Ed. 2017, 56, 5994–6021. 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 865 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 865/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Full Papers [15] Y.-F. Li, ChemSusChem 2019, 12, 1846–1857. [26] C. A. Schneider, W. S. Rasband, K. W. Eliceiri, Nat. Methods 2012, 9, 671– 1 [16] M. J. S. Abb, B. Herd, H. Over, J. Phys. Chem. C 2018, 122, 14725–14732. 675. 2 [17] T. Weber, J. Pfrommer, M. J. S. Abb, B. Herd, O. Khalid, M. Rohnke, P. H. [27] S. Cherevko, S. Geiger, O. Kasian, A. Mingers, K. J. J. Mayrhofer, J. 3 Lakner, J. Evertsson, S. Volkov, F. Bertram, R. Znaiguia, F. Carla, V. Vonk, Electroanal. Chem. 2016, 774, 102–110. E. Lundgren, A. Stierle, H. Over, ACS Catal. 2019, 9, 6530–6539. [28] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, 4 [18] B. Herd, M. Knapp, H. Over, J. Phys. Chem. C 2012, 116, 24649–24660. Pergamon Press, 1966. 5 [19] I. Sohrabnejad-Eskan, A. Goryachev, K. S. Exner, L. A. Kibler, E. J. M. [29] J. Juodkazytė, R. Vilkauskaitė, B. Šebeka, K. Juodkazis, Transactions of the Hensen, J. P. Hofmann, H. Over, ACS Catal. 2017, 7, 2403–2411. IMF 2007, 85, 194–201.6 [20] S. O. Klemm, A. A. Topalov, C. A. Laska, K. J. J. Mayrhofer, Electrochem. [30] D. R. Rolison, J. W. Long, J. C. Lytle, A. E. Fischer, C. P. Rhodes, T. M. 7 Commun. 2011, 13, 1533–1535. McEvoy, M. E. Bourg, A. M. Lubers, Chem. Soc. Rev. 2009, 38, 226–252. 8 [21] S. Cherevko, J. Electroanal. Chem. 2017, 787, 11–13. [31] J. O’M. Bockris, A. K. N. Reddy, Modern Electrochemistry, Vol. 2, Plenum 9 [22] S. Cherevko in Encyclopedia of Interfacial Chemistry, Vol. 5.1 (Ed.: K. Publishing Corporation, New York, 1973, pp. 991–1014. Wandelt), Elsevier, 2018, pp. 68–75. 10 [23] S. Cherevko, A. R. Zeradjanin, A. A. Topalov, N. Kulyk, I. Katsounaros, 11 K. J. J. Mayrhofer, ChemCatChem 2014, 6, 2219–2223. 12 [24] P. P. T. Krause, H. Camuka, T. Leichtweiss, H. Over, Nanoscale 2016, 8, Manuscript received: September 4, 2019 13944–13953. Revised manuscript received: October 15, 2019 13 [25] T. Weber, M. J. S. Abb, O. Khalid, J. Pfrommer, F. Carla, R. Znaiguia, V. Accepted manuscript online: October 16, 2019 14 Vonk, A. Stierle, H. Over, J. Phys. Chem. C 2019, 123, 3979–3987. Version of record online: December 5, 2019 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 ChemCatChem 2020, 12, 855–866 www.chemcatchem.org 866 © 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA Wiley VCH Freitag, 31.01.2020 2003 / 152686 [S. 866/866] 1 18673899, 2020, 3, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cctc.201901674 by Justus-Liebig-Universitat, Wiley Online Library on [01/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License