Hydroxylation of Platinum Surface Oxides Induced by Water Vapor

With its high stability and well-tuned binding strength for adsorbates, platinum is an excellent catalyst for a wide range of reactions. In applications like car exhaust purification, the oxidation of hydrocarbons, and fuel cells, platinum is exposed to highly oxidizing conditions, which often leads to the formation of surface oxides. To reveal the structure of these surface oxides, the oxidation of Pt in O2 has been widely studied. However, in most applications, H2O is also an important or even dominant part of the reaction mixture. Here, we investigate the interaction of H2O with Pt surface oxides using near-ambient-pressure X-ray photoelectron spectroscopy. We find that reversible hydroxylation readily occurs in H2O/O2 mixtures. Using time-resolved measurements, we show that O–OH exchange occurs on a time scale of seconds.


1.
Comparing the results obtained from a sputtered polycrystalline foil with those from a well-annealed single crystal to correlate activity with defect sites is not justified. In addition to defects, the sputtered foil will also contain non-111 surfaces with lower coordinated Pt atoms, which may explain the difference in reactivity between the foil and the single crystal.
The comparison that should be made is either between a sputtered and non-sputtered single crystal or between a sputtered and non-sputtered foil. Without these additional experiments, the increased reactivity cannot be unequivocally assigned to defects and the claim should be reduced to "lower coordinated Pt atoms are more reactive" (which is not very surprising).
Quantification of the increased density of defects induced by the sputtering using SPM would further strengthen this argument. Even more so, because the oxidation itself roughens the surface.

2.
The Pt4f fits are not very convincing. This is mostly because crucial information is missing a.
The fitting function should be given. The label "LF" is insufficient as the paper should be understandable without having to resort to the CasaXPS manual. Furthermore, the different fit parameters need to be clarified and for every fit it needs to be specified, which of them were free fitting parameters and which were constraint. Justification for constrains are also important. This is critical since the fitting function seems to contain 6 fitting parameters per peak. Finally, a motivation is required for the choice of this empirical function. b.
The presence of the clean surface Pt peak needs further supporting. Typically, the base pressure of NAP XPS systems is not that great and the presence of a fully empty surface may not be likely. At least, surface temperature and pressure need to be stated. Showing survey and C1s spectra would further enhance the reliability of this claim. At this given KE/IMFP is the surface-bulk ratio what you would expect? c.
The binding energy difference between bulk Pt and Pt-chem is only 0.09 eV, which is a small shift. In this light, I wonder how much the choice of the fitting function influences the outcome of this fitting model. This feeling is strengthened by the fact that the binding energy of the Pt-4O has a range of about 1 eV, which is ten times the shift of the Pt-chem peak with respect to the Pt bulk peak. All together, the results seem quite arbitrary. I realize given the fact that the Pt 4f spectra all look very similar, this is somewhat expected. Therefore, it would be good to link the intensity of the Pt-chem and Pt-4O peaks with their corresponding O1s counterparts to at least show that there is internal consistency in the fitting model. XAS spectra, important information is missing. How where these obtained (transmission, TEY, TFY)? Why are the spectral features of the gas phase not visible in these spectra? Similarly, why are the gas phase peaks not visible in the XPS spectra?
None of these points are trivial and deserve careful discussion and possibly some additional experiments.
Author's Response to Peer Review Comments: We are pleased to note that the reviewers are positive about the manuscript, noting that "it has the potential to become an influential paper" and that it "should be well suited for Journal of Physical Chemistry Letters after minor revisions". We thank the reviewers for their comments, which have helped to improve the manuscript. With the revisions outlined below, we believe that the manuscript is now suitable for publication in JPCL.
With kind regards,

Rik Mom
Reviewer: 1 Recommendation: This paper is publishable subject to minor revisions noted. Further review is not needed.
Comments: Report on "The hydroxylation of platinum surface oxides induced by water vapor" by Mom et al.
The paper is a well written article on the hydroxylation of PtOx, studied by synchrotron based near ambient pressure XPS. The paper should be well suited for Journal of Physical Chemistry Letters after minor revisions. Details are given below.
The discussion of the exchange reaction (hydroxylation / dehydroxylation) of the oxide is quite interesting. Could one also argue along the lines of the desorption temperature of either H2 or H2O?
Our initial interpretation went along the same line as the reviewer, that H2O adsorbs dissociatively on the oxide. However, as we note in the manuscript, the O coverage remains constant when water is introduced into the chamber. This does not fit with an adsorption reaction, where the coverage should increase. Rather, there is a replacement of adsorbates (Reaction 1 in the manuscript).
To describe dehydroxylation in terms of H2 desorption, one would assume the reaction: 2 OHads → 2 Oads + H2 Since there is essentially no H2 in the gas phase of the chamber, this would be an irreversible reaction. However, Figures 3 and S7 show that the reaction is reversible.
Hence, we conclude that discussing the hydroxylation/dehydroxylation behavior in terms of a desorption temperature would potentially set the reader on the wrong foot.
Please add a fit to the data in Fig. 3. This would allow for a better understanding of the changes during reaction.
Although we agree with the reviewer that a fit can be helpful for a more quantitative understanding of the changes of the surface state, we should note that the time-resolved measurements inherently had a lower data quality. Not only is the signal-to-noise ratio lower than for the steady-state spectra, we also had to employ a smaller binding energy range to decrease the acquisition time, making the baseline for the fit somewhat ambiguous. For these reasons, we feel that a fit of the data would not be sufficiently reliable and could potentially misguide the reader.
Please add the experimental resolution of the measurements to the SI We have indicated the experimental resolution in SI Section S2: The experimental resolution for the O 1s measurements was about 450 meV (based on Xe 5p3/2 measurements). For the O K-edge spectra, the resolution was about 220 meV. For the Pt 4f spectra, the resolution was about 200 meV.

Reviewer: 2
Recommendation: This paper may be publishable, but major revision is needed; I would like to be invited to review any future revision.
Comments: This manuscript is focussed on an interesting topic, the humid oxidation of Pt, which would appeal to both the heterogeneous catalysis and the electrochemistry communities. It has the potential to become an influential paper. However, I believe the following changes are required before the paper can be considered for publication in JPCL: 1. Comparing the results obtained from a sputtered polycrystalline foil with those from a well-annealed single crystal to correlate activity with defect sites is not justified. In addition to defects, the sputtered foil will also contain non-111 surfaces with lower coordinated Pt atoms, which may explain the difference in reactivity between the foil and the single crystal. The comparison that should be made is either between a sputtered and non-sputtered single crystal or between a sputtered and non-sputtered foil. Without these additional experiments, the increased reactivity cannot be unequivocally assigned to defects and the claim should be reduced to "lower coordinated Pt atoms are more reactive" (which is not very surprising).
Quantification of the increased density of defects induced by the sputtering using SPM would further strengthen this argument. Even more so, because the oxidation itself roughens the surface.
We thank the reviewer for this comment, which has indeed also been discussed internally. We considered the term "defects" as sufficiently general to highlight the structural differences between the sputtered foil and the single crystal. However, we agree that the term "lower-coordinated Pt atoms" is more suitable, as it more clearly includes other facets. We have replaced the term "defect sites" in the manuscript.
We should point out that the revised statement is not a trivial case of "lower coordinated Pt atoms are more reactive", since we are exchanging one adsorbate for another here. Hence, it is not about the reactivity of the lower coordinated sites, but rather the selectivity for the Oads or OHads adsorbate. The fact that lower coordinated sites are more selective towards OHads than (111) terraces does not seem trivial.
2. The Pt4f fits are not very convincing. This is mostly because crucial information is missing a. The fitting function should be given. The label "LF" is insufficient as the paper should be understandable without having to resort to the CasaXPS manual. Furthermore, the different fit parameters need to be clarified and for every fit it needs to be specified, which of them were free fitting parameters and which were constraint. Justification for constrains are also important. This is critical since the fitting function seems to contain 6 fitting parameters per peak. Finally, a motivation is required for the choice of this empirical function. b. The presence of the clean surface Pt peak needs further supporting. Typically, the base pressure of NAP XPS systems is not that great and the presence of a fully empty surface may not be likely. At least, surface temperature and pressure need to be stated. Showing survey and C1s spectra would further enhance the reliability of this claim. At this given KE/IMFP is the surface-bulk ratio what you would expect? c. The binding energy difference between bulk Pt and Pt-chem is only 0.09 eV, which is a small shift. In this light, I wonder how much the choice of the fitting function influences the outcome of this fitting model. This feeling is strengthened by the fact that the binding energy of the Pt-4O has a range of about 1 eV, which is ten times the shift of the Pt-chem peak with respect to the Pt bulk peak. All together, the results seem quite arbitrary. I realize given the fact that the Pt 4f spectra all look very similar, this is somewhat expected. Therefore, it would be good to link the intensity of the Pt-chem and Pt-4O peaks with their corresponding O1s counterparts to at least show that there is internal consistency in the fitting model.
The reviewer raises some just points about the fitting model. We agree that more details could be helpful for the reader. Therefore, we have added a detailed description of the fitting model and our choices in the SI (copied below for your convenience). A few points we should note: -The mathematical form of the LF line shape is rather complex. Therefore, we chose to explain the parameters rather than show the function and provided a reference to a more detailed treatment for the interested reader (the CasaXPS manual actually does not show the full function). -The Pt 4f fitting model was based on the model of Miller et al. and is merely used to check the consistency of our data with previously published results. The Pt 4f data in Figure 1a is not used to derive any new conclusions here. Nonetheless, we have justified the choices in our fit model.

New text in SI section S3:
The LF line shape chosen here is an asymmetric gaussian-lorentzian line shape.