Manipulating the oxygen reduction reaction pathway on Pt-coordinated motifs

Electrochemical oxygen reduction could proceed via either 4e−-pathway toward maximum chemical-to-electric energy conversion or 2e−-pathway toward onsite H2O2 production. Bulk Pt catalysts are known as the best monometallic materials catalyzing O2-to-H2O conversion, however, controversies on the reduction product selectivity are noted for atomic dispersed Pt catalysts. Here, we prepare a series of carbon supported Pt single atom catalyst with varied neighboring dopants and Pt site densities to investigate the local coordination environment effect on branching oxygen reduction pathway. Manipulation of 2e− or 4e− reduction pathways is demonstrated through modification of the Pt coordination environment from Pt-C to Pt-N-C and Pt-S-C, giving rise to a controlled H2O2 selectivity from 23.3% to 81.4% and a turnover frequency ratio of H2O2/H2O from 0.30 to 2.67 at 0.4 V versus reversible hydrogen electrode. Energetic analysis suggests both 2e− and 4e− pathways share a common intermediate of *OOH, Pt-C motif favors its dissociative reduction while Pt-S and Pt-N motifs prefer its direct protonation into H2O2. By taking the Pt-N-C catalyst as a stereotype, we further demonstrate that the maximum H2O2 selectivity can be manipulated from 70 to 20% with increasing Pt site density, providing hints for regulating the stepwise oxygen reduction in different application scenarios.


<b>REVIEWER COMMENTS</B>
Reviewer #1 (Remarks to the Author): In this work, the authors prepared a series of Pt-X-C single atom catalysts and investigated the influence of local coordination environment on acidic oxygen reduction reaction pathway tuning. Isolated Pt-S motif is identified as the most selective one toward 2e pathway while 4e pathway is predominant on Pt-C motif. The experimental RRDE and bulk electrolysis researches combined with energetic analysis provide precise answer as to how the neighboring dopants influencing the reaction pathway, which is largely controversial in recent literatures. This present methodology and electrocatalytic results could make a great significance to the active motifs engineering in single atom catalysts as well as to shed light on the structure-activity relationship. I would suggest its publication at Nat. Commun. after addressing the comments below: 1. For XPS spectra fitting, detailed parameters like the peak position, FWMH should be tabulated. 2. In Supplementary Figure 12, the authors considered the kinetic barrier of Pt-S-C motif for either 4e or 2e pathway, for which the latter is significantly lower than the former, suggestive both thermodynamically and kinetically favorable O2-to-H2O2 conversion. The authors are encouraged to supplement a similar kinetic analysis on Pt-C and Pt-N-C coordination structure. 3. Regarding the H2O2 quantification from bulk electrolysis, what's the detection limit for this KMnO4 titration method, i.e., 1 ppm, 10 ppm? 4. Open question, since O-doped carbon is widely reported to effectively convert O2 into H2O2 in electrochemistry, how to differentiate the contribution of Pt active motif from O-doped CNT substrate?
Reviewer #2 (Remarks to the Author): This manuscript describes the synthesis of Pt-(C, S, N)-CNT catalysts and their catalytic performance for 4e-pathway or 2e-pathway ORR. In terms of catalytic performance for ORR, the reported results are inferior to many recently published studies. The highest selectivity towards H2O2 is below 80%, much lower than many recent studies on Pt catalysts, oxidized carbon catalysts, and transition metal singleatom catalysts. Thus, the potential novelty of the current work is on the atomic structural tunning of single-atom sites and the understanding of their catalytic behaviors. Although different structures of M-N-C sites have been studied in several previous studies, the specific Pt-(C, S, N)-CNT comparison has not been reported before.
The authors used high-temperature thermal treatment methods to prepare Pt-(C, S, N)-CNT catalysts. They also provided XAS, XPS, and TEM images to prove that only single atom sites were obtained. XAS provides the average coordination environment of bulk samples. The shifts observed in the R space may originate from the mixture of small Pt nanoparticles and single Pt atoms in different coordination environments. TEM images only provide tiny areas on a sample. I am not convinced that only single atom sites exist in these catalysts, considering the high temperatures (900 and 800 oC) used in the synthesis. How could the nucleation of Pt particles be prevented entirely when the samples were annealed at 900 oC for 60 min? Further, the authors compared catalysts with different mass Pt loadings. Why can no Pt nanoparticles be formed in the 2.8 wt% sample while the 7 wt% sample has nanoparticles? Suppose you roughly estimate the specific surface area of carbon nanotubes. What would be the average distance between two Pt atoms if 2.8 wt% of Pt are all atomically dispersed on the carbon surface? I would suggest that the better way to quantify the Pt-based active sites is to use CO chemisorption or electrochemical stripping (Current Opinion in Electrochemistry 2018, 9:198-206). Then, the authors could compare TOF of different active sites and draw conclusions related to their intrinsic activities.
Overall, I think the work may potentially provide useful information in the field. But the current experiential data and discussion are insufficient to support the claimed conclusions.
Reviewer #3 (Remarks to the Author): The authors report here the oxygen reduction reaction pathway tuning over CNT supported Pt single atom catalysts coordinated with different metalloid dopant, and their active sites density effect on products selectivity. The materials are well characterized by HAADF-STEM, EXAFS and XPS, which confirm the atomically dispersed feature of Pt central atoms and the different electronic structure as arisen from different coordination motifs. I love the logic and the comprehensive consideration of multiple perspectives, for both experimental setup and theoretical simulations of thermodynamics and kinetics investigation. Overall, the manuscript is well organized and the conclusion is well supported by the present results. It can be accepted at Nature Communications with a minor revision. 1. For Pt-N-C coordination, the annealing temperature effect on the ORR onset potential has been screened. How about Pt-S-C? Is there any room to further improve the H2O2 generation performance? 2. What's the sulfur and oxygen content within bare S-doped CNT? In Fig. S3, O signal in the Pt-S-CNT is significantly higher that other 2 samples, while earlier studies suggest that O-doped carbon can intrinsically convert O2 into H2O2 at high activity, see Nat. Catal. 1, 156-162 (2018) andNat. Catal. 1, 282-290 (2018) for reference. Relevant active motif assignment for either Pt central sites or O dopants should be properly addressed. 3. In Figure 4 of long-term electrolysis test, the leaching ratio of Pt should be quantitatively discussed. 4. In Experimental section, details regarding the sample preparation and testing condition for MicroCT should be supplemented. 5. XPS experimental data and the fitting results are not well consistent. 6. The authors used monolayer graphite model to describe the surface of CNT, ignoring the effect of curvature, which may affect the final conclusion.

Response to reviewers' comments:
We thank the reviewers and editor for constructive comments which have helped us to greatly improve our research and the quality of our manuscript. Below, we carefully address the points raised by reviewers one by one, with additional experimental inputs on statistic Pt sites distribution by ATR-IR and high-resolution XRD, site density and TOF analysis via electrochemical CO stripping, and quantitative Pt leaching, as well as theoretical simulations on the kinetic barrier comparison of 2e -/4e -ORR and potential substrate curvature effect.

Reviewer 1
"In this work, the authors prepared a series of Pt-X-C single atom catalysts and investigated the influence of local coordination environment on acidic oxygen reduction reaction pathway tuning.
Isolated Pt-S motif is identified as the most selective one toward 2e pathway while 4e pathway is predominant on Pt-C motif. The experimental RRDE and bulk electrolysis researches combined with energetic analysis provide precise answer as to how the neighboring dopants influencing the reaction pathway, which is largely controversial in recent literatures. This present methodology and electrocatalytic results could make a great significance to the active motifs engineering in single atom catalysts as well as to shed light on the structure-activity relationship. I would suggest its publication at Nat. Commun. after addressing the comments below:" ---We much appreciate the reviewer's overall positive evaluation to our work. Please see below our point-by-point response to each of the raised suggestion.  For the kinetic analysis, we noted that Pt-S4 exhibits the lowest kinetic energy barrier for 2epathway (0.868 eV), followed by Pt-N4 of 1.006 eV and Pt-C4 of 2.024 eV.
For the thermodynamic analysis, we also plotted the Gibbs free energy change of OOH* to  times, and the amount of H2O2 solution consumed was 505 μL, 510 μL and 507 μL, respectively, which was close to the theoretical value 510.5 μL that calculated from the reaction equation: 2MnO4 -+ 5H2O2 +6H + → 6Mn2 + + 5O2 + 8H2O. The solution in the test tube of a control group (rightmost one) was still purple with 510 μL of water added, indicating that dilution does not make KMnO4 colorless. To titrate the unknown H2O2 content in the bulk electrolysis solution, 1 mL of electrolyte was collected instead of the above standard 16.6-ppm H2O2 solution, therefore, the detection limit of this KMnO4 titration can be calculated as 16.6×510.5/1000=8.5 ppm, which is lower than the first data point we measured in Figure 4e (ca. 71.7 ± 2.5 ppm from two independent measurements).
This result has been added as Supplementary Figure 24 on Page 25 of revised SI, and copied below for the reviewer's reference:    (2021)) for example.
We agree with the reviewer that STEM images provide the dispersion information on localized area, while EXAFS spectra and fittings demonstrate the average coordination information of bulk sample. Though there're some uncertainties from experimental viewpoint, these two tools are the most widely recognized methodology for probing the metal distribution in single atom catalysts.
To further probe if any Pt clusters and/or nanoparticles present in Pt-X-CNT, we have carried out additional high-resolution XRD (HR-XRD) and in situ electrochemical attenuated total reflection  reduction toward the 4epathway into H2O product. Therefore, this simulation result in turn reinforces our assignment of highly dispersed Pt-X coordination motifs as active center for selective O2-to-H2O2 conversion.  Fig. 5c to avoid any misunderstandings as suggested.    At last but not least, we calculated the TOFs under different Pt site densities using the same way.

Supplementary Figure 18. (a) Illustration of Pt1-S4-C and
For 200-Pt-N-CNT, the TOFH2O2 decreases with increasing overall catalyst loading from 0.1 mg cm -2 to 0.4 mg cm -2 (Fig. R3), suggestive the negative effect of a longer diffusion path for H2O2 generation. Similarly, at a fixed catalyst loading of 0.1 mg cm -2 , the determined TOFH2O2 decreases with increasing electrochemical active Pt site density from 0.29 wt.% to 0.51 wt.% as derived from CO stripping measurements, suggestive that the densely distributed Pt sites are beneficial for the total reduction of O2 into H2O rather than H2O2 generation.
These intrinsic TOFs analysis on the diffusion path effect and the Pt site density effect are actually in good agreement with our main conclusion as derived from electrochemical selectivity analysis.