Activation of subnanometric Pt on Cu-modified CeO2 via redox-coupled atomic layer deposition for CO oxidation

Improving the low-temperature activity (below 100 °C) and noble-metal efficiency of automotive exhaust catalysts has been a continuous effort to eliminate cold-start emissions, yet great challenges remain. Here we report a strategy to activate the low-temperature performance of Pt catalysts on Cu-modified CeO2 supports based on redox-coupled atomic layer deposition. The interfacial reducibility and structure of composite catalysts have been precisely tuned by oxide doping and accurate control of Pt size. Cu-modified CeO2-supported Pt sub-nanoclusters demonstrate a remarkable performance with an onset of CO oxidation reactivity below room temperature, which is one order of magnitude more active than atomically-dispersed Pt catalysts. The Cu-O-Ce site with activated lattice oxygen anchors deposited Pt sub-nanoclusters, leading to a moderate CO adsorption strength at the interface that facilitates the low-temperature CO oxidation performance.

1. Figure 2a shows the effect of Cu on the CO conversion of Pt1/Ce and Ptn/Ce catalysts. The strong positive effect of Cu on the CO conversion temperature is misleading as samples containing Cu also contain 2.5 higher concentration of Pt. Figure S8 illustrates that by diluting the catalysts the effect of Cu gets smaller but is still there. At the same time, it is not explained why the catalyst for these tests were diluted only by a factor of 1.6 and not by 2.5 based on Pt loading and similar Pt dispersion within the error bar. This observation questions whether the effect of Cu exists and whether it is significant considering the error bars on concentration of Pt and dispersion of Pt.
2. Kinetic characterization of the catalyst activity at low conversion regime requires additional details describing experiments and may be additional tests showing reproducibility. What kind of pre-treatment was used before measuring the activity of the catalysts to remove adsorbed species such as carbonates? Which material was used to dilute the catalyst during tests? Concerning Figure 2b, was the activity measured after initial deactivation tests of the catalysts? Were the activity data and the activation energies reproducible upon temperature cycling? The activation energy for Ptn/Ce seem to be very high, what can be the reason? For comparison, it would be useful to report also the activity of CuCe catalyst in the Figure 2b. 3. The formula in the line 363 contains parameter winterface corresponding to the fraction of Pt atoms at Pt-support interface, which are also called perimeter sites. This parameter is used for TOF calculation but there are no details how this parameter was estimated for each catalyst. Besides, Table S1 compares TOFs for different Pt containing catalysts measured in this work with those previously reported in the literature. Can the authors confirm that in all these works TOF parameter was estimated based on the number of above-mentioned perimeter sites and that this number was estimated in a similar way? Ptn/CuCe in comparison to Ptn/Ce catalyst. This effect is interesting and novel. It would make sense to confirm it by TPD experiments. Besides, measuring of the reaction orders for CO and oxygen would be useful, as weaker CO adsorption on Ptn/CuCe can affect the reaction mechanism in different ways. May be reaction on Ptn/CuCe takes place not only at the interface but also on Pt nanoparticles surface due to weaker CO adsorption? 5. The EXAFS analysis involving second coordination shell of Pt reported in Table S2 does not look very reliable: the fitted curves are not presented, the fitted deltaE of 21 -29 eV is too high, which suggests that fits might not be correct. Besides, basic description of XAS beamline is missing: type of monochromator, mirrors, detectors, flux, beam size… 6. Figure 3c compares the calculated Bader changes of interfacial Pt atoms to the structural parameters of Pt determined by XAS measured ex situ. What is the origin of these correlations from the theoretical point of view? Can the authors add the error bars to the plots to confirm that the changes are significant? How relevant are the reported correlations considering that in air most of Pt atoms in Ptn particles are oxidized by air, while during catalysis they should be partially reduced, thus large fraction of oxygen in their local coordination should be replaced by CO? 7. Concerning XPS measurements, were they performed in vacuum? Was the possibility of photo-reduction of Ce4+ during the measurements considered? 8. The term "modulation" used in the manuscript title seems to be not exact for the reported phenomena.
Reviewer #2: Remarks to the Author: The authors have studied CO oxidation on Pt deposited on Cu doped ceria nanorods. The results show high reactivity for Pt sub-nano clusters for CO oxidation. There is considerable interest in improving the reactivity of Pt catalysts for this reaction, so the work is potentially interesting. However, the authors have overlooked some recent literature, which causes me to question their interpretation. The approach they used to report reactivity, TOF based on interfacial sites, is not consistent with standard practice in this field. Finally, they state that Cu doped ceria is not active for CO oxidation at low temperatures, which is not true. For these reasons, I do not think the manuscript is not suitable for publication in its present form.
1) Cu doped ceria is known to be active for CO oxidation as seen in reference 1 and 2 below. Atomically dispersed Cu shows onset of CO oxidation reactivity starting at room temperature.

Author reply:
We thank the reviewer for taking his/her time evaluating our manuscript and agree with the reviewer that the catalytic mechanism studies of our prepared catalysts can be strengthened by adding certain in situ characterizations. Additional in situ DRIFTS experiment of CO oxidation of Pt n /CeCu catalyst has been performed at room temperature to investigate the catalytic mechanism. As shown in  The manuscript is of potential interest for the wide community of scientists in the field of catalysis and design of nanomaterials. At the same time, there are several critical points, which do not allow saying that all conclusions of this work are supported by the experimental or theoretical data. Therefore, major revision is recommended.
Addressing the following points can significantly improve understanding of the effects reported of this work and clarify their significance:

Author reply:
We appreciate the reviewer's recognition on the importance of this topic to the wide catalysis community. In the revised manuscript, we made our best efforts to improve the work by adding new experimental results, such as in situ characterizations, repeatability tests, reaction order tests, CO-TPD and H 2 -TPR. The detailed experimental data and discussion have been presented below in response to listed comments. We believe that we have addressed the reviewer's concerns and improved the quality of our work. We hope the revised manuscript will deem fit to be published in Nature Communications.

Author reply:
We thank the reviewer for carefully checking detailed experimental records. Indeed, we found that the weights of supported Pt catalysts and diluted Ce 0.99 Cu 0.01 O 2 supports have been labeled incorrectly in the caption of Supplementary Fig. 11. In order to eliminate the effects of Pt's mass loading, we diluted 20 mg Pt n /CeCu or Pt 1 /CeCu with 30 mg Ce 0.99 Cu 0.01 O 2 , which corresponded to the factor of 2.5 for the decrease of Pt's mass loading. We thank the reviewer for pointing out our typo and have revised the corresponding description of Supplementary Fig. 11. In terms of the concentration and dispersion of Pt, we agree with the reviewer that they are important for accurate activity evaluations of our prepared catalysts. To this end, the ICP-OES characterization has been performed to test Pt's mass loading variation of three samples of Pt n /CeCu that were prepared by the same process as described in our manuscript. The average mass loading of Pt for Pt n /CeCu is 1.77 ± 0.11 wt% based on the ICP-OES results (1.65 wt%, 1.79 wt% and 1.86 wt%) The consistency of Pt's mass loading can be attributed to the well-controlled process of atomic layer deposition method. The Pt size distribution analysis of AC-STEM images ( Supplementary Fig. 5) indicate that the sizes of Pt clusters in Pt n /CeCu and Pt n /Ce are similar. Therefore, we believe the slight variation on concentration and dispersion of Pt towards the activity evaluation of our prepared catalysts are negligible compared with the effect of Cu dopants.

Author reply:
We appreciate the reviewer's comments on the kinetic tests of supported Pt catalysts and have added relevant description of our experiments in details, as well as performed additional tests to address the reviewer's concern. In terms of the question on the pre-treatment of catalysts, the catalyst is tested as is after the ALD process, without deliberately performing additional pre-treatments. The catalyst is quite stable with respect to cycling tests with negligible change in light off temperature (shown in Fig. R2), implying the robustness of the fabrication process and reproducibility of the catalytic performance results. The negligible shifts of CO conversion curves also indicate the structural stability of our constructed Pt/oxide interface, which can be attributed to the gas-phase based atomic layer deposition method that has minimal detrimental effect on the surface structure of oxide supports as reported in our previous work (Chem. Sci. 2018,9,2469

Modification:
1. Added the result and discussion of the cycling tests of Pt n /CeCu. "The slightly shifts of CO conversion curves in cycling tests of Pt n /CeCu ( Supplementary Fig. 9) indicate the structural stability of the constructed Pt/oxide interface, which can be attributed to the gas-phase based ALD method that has minimal detrimental effect on the surface structure of oxide supports 35 ." (Second paragraph in page 7 in text and Supplementary Fig. 9 Fig. 15), close to that in previous studies 39,43,44 ." (First paragraph in page 8 in text and Supplementary Fig. 15  We thank the reviewer for his/her comment on our TOF calculation. We added detailed information and procedural description to clarify the calculation method of TOF in order to make a fair comparison in  Table R1, the TOF of Pt n /CeCu for CO oxidation at 80 ºC is 0.26 s -1 , which is one order of magnitude higher than that of atomically-dispersed Pt catalyst and rivals the best state-of-the-art Pt/CeO 2 catalyst. In the revised manuscript, we have recalculated the TOF values and revised the corresponding discussion.   In the revised manuscript, we have added the data and corresponding discussion to strengthen our catalytic mechanism study. Figure R6. The reaction orders of CO and O 2 over Pt n /CeCu.

Modification:
1. Added the result and discussion of CO-TPD. "The temperature-programmed desorption of CO has also been performed to investigate the desorption behavior of CO on Pt n /CeCu and Pt n /Ce ( Supplementary Fig. 21   We appreciate reviewer's sharp comment and agree with the reviewer that the fitted  Table R2. We thank the reviewer for pointing out that missing information of our used X-ray absorption spectroscopy (XAS) beamline. In the updated manuscript, we have added the corresponding information of XAS experiments in details. The incident photon beam was selected by a double-crystal Si (111) monochromator after a collimating mirror and focused by a toroidal mirror. All XAS measurements were conducted in transmission mode using a 19-element high-purity germanium solid-state detector. The X-ray beam size on our prepared catalysts was about 0.9 × 0.3 mm 2 at half-maximum (FWHM) with a photon flux of > 4 × 10 11 photons/s at 9 keV.        Author reply: We thank the reviewer for this suggestion. The term "modulation" emphasizes on the regularity of the results with changing parameters, which might not be suitable in this case. The key to activate the catalytic activity of supported Pt catalysts in our work is the precise control of interfacial reducibility and structure via oxide doping and redox-coupled atomic layer deposition process. To highlight the importance of this fabrication strategy and its beneficial effects, we have changed the title of our manuscript to "Activation of subnanometric Pt on Cu-modified CeO 2 via redox-coupled atomic layer deposition for low-temperature CO oxidation".

Modification:
1. The title of our manuscript has been revised to "Activation of subnanometric Pt on Cu-modified CeO 2 via redox-coupled atomic layer deposition for low-temperature CO oxidation".
(Title in Page 1)

Reviewer #2:
The authors have studied CO oxidation on Pt deposited on Cu doped ceria nanorods.

Author reply:
We appreciate the reviewer for his/her positive evaluation on our work and pointing us to some excellent works in this field. We have carefully read these references and improved the discussion of the catalytic mechanism part in our study. We have revised the corresponding discussion including the activity of Cu doped ceria and the catalytic mechanism, as well as the evaluation and comparison of TOF in the updated manuscript. We hope the updated manuscript will deem fit for the publication in  Supplementary Fig. 10 Fig. 19 2. Revised Fig. 2d and Supplementary Fig. 18 ( Fig. 2d in text and Supplementary Fig. 18 Table R3, the TOF of Pt n /CeCu for CO oxidation at 80 ºC is 0.26 s -1 , which is one order of magnitude higher than that of atomically-dispersed Pt catalyst. Note that the TOF values for reference 9-12 in Table S1 in supplementary information are calculated based on the fraction of interfacial or surface Pt atoms. Therefore, the catalytic activity of Pt n /CeCu still rivals state-of-the-art Pt/CeO 2 catalysts for low-temperature CO oxidation. In the updated manuscript, we have recalculated the TOF values and revised the corresponding discussion. Moreover, the TOF of our prepared Pt n /CeCu catalyst is higher than that reported in previous references (ACS Catalysis 2015, 5, 5164) and rivals the best state-of-the-art Pt/CeO 2 catalysts. We believe this method could be as well applied to other noble metal/oxide combinations and is of general interest to the broad catalysis community.
We have strengthened the discussion of precise interface control and structure-activity relationship in our revised manuscript and hope the revised manuscript is appropriate for publication in Nature Communications.
We agree with the reviewer that in-situ experiments such as the in-situ X-ray absorption spectroscopy (XAS) experiment as reported in the previous reference (ACS Catalysis 2015, 5, 5164) are of great significance to study the catalytic mechanism, which can monitor the change of chemical states and local structures of active sites.
Unfortunately, the in-situ XAS experiments could not be done due to the travel ban and the shut down/reduced operation of the beam center. To address the referee's concern, we conducted in-situ DRIFTS, H 2 -TPR and CO-TPD that can also shed light on the change of local chemical environments in the reaction. Fig. R13 Fig. 6)." (Second paragraph in page 6 in text and Supplementary Fig. 6 in SI)

2>> In the synthesis, can the Pt mass loading of Pt/CeCu be controlled?
Author reply: We appreciate the reviewer's question on the controllability of our preparation method for supported Pt catalysts. ALD is a self-limiting process that can precisely control the adsorption and nucleation of gas-phase precursors by tuning the ALD recipes,

Modification:
1. Revised the mass loading of Pt n /CeCu by the tested average value.
(Supplementary Table 1 in SI)

3>>
The reducibility of all catalysts should be characterized by temperature-programmed reduction.

Author reply:
We agree with the reviewer that H 2 or CO TPR can directly reflect the reducibility of catalysts. To this end, H 2 -TPR tests were performed for all catalysts using a chemisorption analyzer (AMI-300 series, Altamira Instrument). Specifically, 30 mg of the catalyst has been supported by quartz wool in a U-type quartz tube reactor, which is pretreated using 30 mL/min of Ar at 100 °C for 30 min. The feed is switched to 30 mL/min of 10% vol. H 2 balanced with Ar, when the catalyst is cooled down to room temperature. Then, the reactor is heated to 800 ºC with a ramp rate of 5 ºC/min and thermal conductivity detector is utilized to monitor the signal of H 2 consumption. As shown in Fig. R17, an obvious reduction peak below 200 ºC for Pt 1 /Ce and Pt 1 /CeCu can be attributed to the reduction of oxidized Pt single atoms.

Author reply:
We appreciate the reviewer for his/her constructive comment and we recheck on the point that the signals of adsorbed CO in the in-situ DRIFTS spectra decrease under He flow. The adsorption energy of CO molecule on the vertex site of Pt cluster is usually very strong as our previous study reported (Nanoscale 2019, 11, 8150).
However, the CO adsorption strength at the interfacial Pt site of Pt n /CeCu is weakened due to the electron transfer form Pt atoms to oxide support. Per the suggestion of reviewer, the temperature-programed desorption of CO (CO-TPD) has been performed to investigate by the desorption behavior of CO. As shown in Fig.   R18, the peaks at -75 and -67 ºC for Pt n /CeCu and Pt n /Ce can be assigned to CO desorption from Cu doped CeO 2 and CeO 2 nanorod supports, respectively, which are closely to that of Pt/CeO 2 in previous study (J. Phys. Chem. 1987, 91, 3310). The We have strengthened the corresponding discussion of CO-TPD and in situ DRIFTS result in our updated manuscript. Figure R18. TPD curves following a saturation adsorption of CO on (a) Pt n /CeCu and (b) Pt n /Ce catalysts at -100 ºC.

Modification:
1. Added the result and discussion of CO-TPD. "The temperature-programmed desorption of CO has also been performed to investigate the desorption behavior of CO on Pt n /CeCu and Pt n /Ce (Supplementary Fig. 21)

Author reply:
We appreciate the reviewer's constructive comments on our DFT calculation models.
As shown in Fig. R19   Author reply: We appreciate the reviewer's concern on valence states of Pt analyzed by the XPS results of our catalysts. As shown in Figure 3, the white line intensities of Pt 1 /CeCu Reviewer #2: Remarks to the Author: The authors have revised the manuscript in response to the reviewer comments and have addressed most of the concerns. The work shows that the presence of Cu dopants allows the Pt samples with sub-nanometer clusters to achieve a high reactivity. When the reactivity is properly normalized to total Pt atoms, the performance is comparable to the best Pt catalysts, and the work implies that oxygen activation of the support was achieved by the Cu dopants. The work is suitable for publication. I have only two revisions to suggest.
1) The authors need to check the indexing of ceria lattice planes in Figure 1. The (110) lattice plane is forbidden by symmetry rules, it is the (220) that is allowed. They need to check their calibrations and also the planes allowed by symmetry considerations. Furthermore, it has been shown in the literature that ceria rods expose (111) and (100) surfaces [1]. The authors do not show any evidence for (110) surface facets, so they should revise their description of the rods and correct this. Also, they circle a few of the bright dots in the image b and d, but too many of these dots have similar intensity. The numbers of bright dots should be consistent with the Pt loading, in terms of #of atoms/nm2. The contrast from the Ce columns makes it difficult to properly identify the Pt atoms, hence it is best to look at the thin regions of the sample where the contrast from the support is a minimum. Since the AC-STEM image is a slice through the sample, and the Pt atoms are all at the surface, the image can be used to quantify the Pt loading. See ref [2] for a description of how the numbers of Pt single atoms per unit area can be consistent with bulk loading and surface analysis.
2) The loss of the CO band after flowing helium (Figure 2 d) cannot be used to infer weaker bonding of CO to Pt. As shown in ref 46, the lattice oxygen can react with the CO causing formation of CO2 at ambient temperature. In that study, to check the strength of the CO binding with the Pt, the authors first depleted all interfacial oxygen by flowing CO during TPR. After that, the catalyst was cooled without exposing to any additional oxygen. It was found that the CO band could not be removed by flowing He. Only such a test could demonstrate the binding of CO to Pt, and whether that binding energy was altered by support doping. The experiment performed by the authors does not address this concern because the presence of active oxygen species at the Pt-ceria interface allows the strongly bound CO to be reacted easily at room temperature. Figure R1. X-ray diffraction pattern of CeO 2 nanorod support.

Author reply:
We appreciate the reviewer for taking his/her time to evaluate our revised manuscript.
We have seriously considered the reviewer's comments and made our best efforts to improve the work. We have emphasized the novelty of our reported strategy for the precise interface control, especially the role of Cu dopants on the enhanced catalytic activity. More realistic DFT calculations have been performed to study the CO oxidation at the interface of Pt 14 /Ce and Pt 14 /CeCu to strengthen the reliability of our proposed catalytic mechanism. We hope the revised manuscript will deem fit for the publication in Nature Communications. Below is our point by point response to reviewer's comments.
1>> Insights on catalytic mechanism are similar to previous studies, in which the sub-nanometer Pt clusters exhibit the best activity for CO oxidation.

Author reply:
We thank the reviewer's comment on the spatial distribution of Cu and Pt clusters.
Since Cu dopants are introduced with a low concentration during the hydrothermal synthesis procedure, Cu dopants mostly reside in the lattice of CeO 2 by replacing Ce atoms, which agrees well with the Cu K-edge XANES and H 2 -TPR results. Pt clusters are anchored to the surface of Cu doped CeO 2 supports as shown in Fig. 1.
Our in-situ DRFITS spectra, CO-TPD and H 2 -TPR results indicate that Cu dopants can not only activate the interfacial oxygen, but also weaken the CO bonding strength at interfacial Pt atoms. This is in line with the differential charge densities analysis of Pt 5 /Ce and Pt 5 /CeCu in Fig. R5. Since the low-valence Cu dopant shares less electron density with surrounding oxygen atoms than Ce atoms, the bonding between interfacial Pt and oxygen is strengthened with an enhanced polarization, which leads a weakened adsorption of CO molecule at interfacial Pt atoms. We have added the results and discussion of differential charge density in the revised manuscript. Figure R5. Differential charge density (Δρ) and the corresponding contours in the plane crossing the interfacial Pt and oxygen atoms of (a) Pt 5 /Ce and (b) Pt 5 /CeCu.

Modification:
1. Added the discussion of differential charge density. "The differential charge densities indicate that the bonding between interfacial Pt and oxygen is strengthened with an enhanced polarization due to the low-valence Cu dopant, which leads a weakened adsorption of CO molecule at interfacial Pt atoms (Supplementary Fig.   31)." (Second paragraph in page 13 in text) 2. Added the results of differential charge density in Supplementary Fig. 31.
( Supplementary Fig. 31) 4>> Contribution of Cu on the catalytic performance is still unclear. EDS cannot get the conclusion that Cu is enriched at Pt/CeO2 interfaces.

Author reply:
We thank the reviewer's insightful question on the role of Cu dopants on the catalytic performance and agree that the EDS elemental maps alone are not enough to support the conclusion of Cu enrichment at the Pt/CeO 2 interface. EDS elemental maps (