Nanoscale kinetics of asymmetrical corrosion in core-shell nanoparticles

Designing new materials and structure to sustain the corrosion during operation requires better understanding on the corrosion dynamics. Observation on how the corrosion proceeds in atomic scale is thus critical. Here, using a liquid cell, we studied the real-time corrosion process of palladium@platinum (Pd@Pt) core-shell nanocubes via transmission electron microscopy (TEM). The results revealed that multiple etching pathways operatively contribute to the morphology evolution during corrosion, including galvanic etching on non-defected sites with slow kinetics and halogen-induced etching at defected sites at faster rates. Corners are the preferential corrosion sites; both etching pathways are mutually restricted during corrosion. Those insights on the interaction of nanostructures with reactive liquid environments can help better engineer the surface structure to improve the stability of electrocatalysts as well as design a new porous structure that may provide more active sites for catalysis.

The authors used an experimental and computational approach to study relationships between the coordination of surface atoms (shape of NPs) and dealloying-induced morphological changes of the Pt-Pd core shell structure. Although in situ TEM imaging is an important step in visualizing dealloyinginduced transformations, this paper will only be suitable for publication in Nature Communications after a major revision. The authors must address the shortcomings listed below: The abstract is misleading. This paper has nothing to do with understanding how to minimize the corrosion of ORR catalysts. It is, in fact, a study related to monitoring morphological changes during the dealloying of Pt/Pd systems; and this should clearly be stated in the paper. The conclusion that we have to learn how to protect unstable, low coordinated atoms is not new and has been comprehensively discussed in the literature that, unfortunately, the authors omitted to cite.
As mentioned above, it is important that it is now possible to use TEM for monitoring in situ morphological changes during dealloying of bimetallic systems. However, it is of paramount importance to use additional experimental probes that can provide atomic-scale information in order to be able to fully understand how the kinetics of dissolution depends on the density and nature of defects, the electrode potential, the electrochemical environment, and the history of the experiment. For example, in situ ICP-MS methods have been developed for studying single crystals, thin films and NPs, and in situ coherent X-ray diffraction has also been routinely used to study the role of defects in NPs. For details see: Angew. Chemie Int. Ed. 2012, 51 (50), 12613-12615;Electrochem. Commun. 48, 81-85 (2014); ACS Catalysis, 6 (2016) 2536-2544 and Science 356 (2017) This paper will be important to a very broad audience including fuel cell chemists, nanoparticles synthesis and TEM experts and is highly suitable for Nature Comm."

Response:
We would like to thank the reviewer for the appreciation of our effort on the study of etching of core-shell nanoparticles using liquid cell. The results contribute to our knowledge on: the formation of Pt nanoframe; the etching of the less-expensive metal; and the competition of different etching modes. To improve the presentation of our study, we did further analysis on our data based on the reviewer's suggestion.

Comments: "My main comment for improvement is
From the equation in the supporting information the rates are calculated from changes in area observed in the TEM. However the nanoparticles are 3 dimensional not 2 dimensional objects.

An effort to calculate rate changes in term of volumes and numbers of atoms etched would be more meaningful."
Response: We agree with the reviewer on this point. Because the TEM images are the projection of the 3D structure in 2D, it is challenging to estimate the etching volume based on the observed image, especially in such a highly dynamic nature, however, we can approach such estimation by assuming that the etched space is pseudo-spherical. This assumption is based on that: 1) the etched areas usually exhibit a rounding exterior contour; 2) we always observe that during the enlarging of the etched area, the contrast from that area becomes brighter, indicating that the etching is also proceeding in the depth direction along the electron beam. Therefore the etched volume is approximated as: The number of atoms being dissolved during the etching process is then: All the related calculation and analysis are now added in the Supplementary Information in the highlighted section of newly-added Supplementary Section 4 and Figure S2. Response: We thank the reviewer for acknowledging the importance of our work. The corrosion process studied here, we believe, is related to the potential failure of application of shape and composition controlled nanoparticles when being scaled up. However, this work is still useful and important and may be applied to practical catalysts at least to explain their behavior.
Response: Thanks for the referee's comment. We have read and cited Shinozaki's papers on the differences in RDE and MEAs, and corrected the description in the revised manuscript (see Line 49-51). We clarified that the poor performance of ORR catalyst in full fuel cell in comparison to ORR test is due to the use of different electrode, electrolyte, and the different evaluation protocols and operating conditions, specific to the following aspects: 1. The evaluation protocols are vastly different between RDE and MEA. Particularly, in RDE, the potential range from which kinetic information for the ORR can be obtained is narrow (>0.7-0.8 V vs. RHE).
2. The catalyst layer interface are different between two test conditions. In MEAs, the catalyst layer interface is porosity and ionomer coverage, while that in RDE is all acidic electrolyte.
3. Operating conditions, such as O 2 diffusion loss, temperature and relative humidity are all different in MEA and RDE.
For your convenience, we have highlighted the changes in the main text.

Reviewer #3 (Remarks to the Author):
Comments: "Report for MS by Shan et al.
The authors used an experimental and computational approach to study relationships between the coordination of surface atoms (shape of NPs) and dealloying-induced morphological changes of the Pt-Pd core shell structure. Although in situ TEM imaging is an important step in visualizing dealloying-induced transformations, this paper will only be suitable for publication in Nature Communications after a major revision. The authors must address the shortcomings listed below:" Response: We thank the reviewer for encouraging us that in situ TEM is critical due to its capability in visualizing the materials structure transformations. However, the reviewer also commented on the shortcomings of our manuscript. Here we provide a point-topoint reply and revision to these suggestions, we think our manuscript is much improved after the revision.
Comments: "The abstract is misleading. This paper has nothing to do with understanding how to minimize the corrosion of ORR catalysts. It is, in fact, a study related to monitoring morphological changes during the dealloying of Pt/Pd systems; and this should clearly be stated in the paper. The conclusion that we have to learn how to protect unstable, low coordinated atoms is not new and has been comprehensively discussed in the literature that, unfortunately, the authors omitted to cite".

Response:
We thank the reviewer for the suggestion. In our new introduction, we have cited and reviewed the works on protecting the under-coordinated atoms on the surfaces using depositing or alloying with Au, adsorption of Br -, annealing and engineering the mesoporous structures (Line 51-54). These methods also help to improve the ORR activity of the catalysts.
Our work on the in situ observation of Pd@Pt nanocube corrosion shows the entire process of how the Pd core is etched from surface defects and corners, the analysis reveals the mechanisms including the galvanic etching and halogen etching. These findings are important in understanding the catalyst evolution during operation, which are also helpful to provide insights on how to optimize the catalyst structure for better activity and stability. We have modified our abstract to better fit with this focus now. Response: We thank the reviewer for the suggestion. With extensive reading on the two in situ techniques using ICP-MS and grain Bragg coherent diffractive imaging (gBCDI), we now have better understanding on the strength and deficiency of different in situ approaches. In situ ICP-MS can measure the dissolution of metal electrodes in electrochemistry, while gBCDI can image the 3D strain and defect network in individual nanocrystals during different physical processes. However, in situ ICP-MS provides unique information on the average structure and composition, but the time resolution is only couple of seconds, which is beyond the time scale in the data from TEM with the time resolution of ~30 ms; gBCDI provide the morphology change of sub-200 nm individual grains, but local structure information including specific shape, localized etching, and the morphology and structure changes in the newly evolved areas can only be addressed by techniques with higher spatial resolution of ~1 nm in our case. Therefore we modified our introduction with more description on other important in situ techniques, and emphasized the advantage of in situ TEM study (Line 66-75). We also modified the conclusion (Line 289-297) to emphasize that the correlative study using various in situ approaches can offer more comprehensive understanding on the structure evolution and mechanisms. Thanks to the review's suggestion, in future we will definitely consider we will definitely consider combining these two techniques with in situ TEM when we study much slower materials evolution or chemical reactions at the scale of hundreds of nm.