Investigations of Carbon Nitride-Supported Mn₃O₄ Oxide Nanoparticles for ORR

Round 1

Reviewer 1 Report

The authors present a study of Mn oxide nanoparticles supported on carbon nitride using different immobilisation methods.  The authors tested their catalytic performance in the oxygen reduction reaction and carried out further characterisation using HAADF-STEM, EELS, XPS, PXRD and UV-Vis spectroscopy. The authors found that Mn oxide nanoparticles interacting with the N sites of the support increased the activity and selectivity towards the 4 e‐ reduction of O2 to H2O.

This is very interesting work and the manuscript is well written and presented. I suggest publication with minor corrections/suggestions as follows:

1. Page 4. When describing the four catalysts, the authors refer to the oleic acid route (OER). At other places in the manuscript it is referred to as the surfactant method preparation. It would help a non-specialist reader if the authors were consistent with this description.

 

2. Figure 1 (a). The main text and figure caption refer to an N2 saturated solution but the figure itself is labelled as Ar, which might confuse readers.

 

3. Figure 1 (b). There is a plot for GC but this isn’t mentioned in the main text.

 

4. Page 6, Figure 2. The HAADF-STEM images are referred to as Figure 2a and 3b in the main text, instead of Figure 2a and 2b. The authors say that Mn mapping shows clearly large particles with irregular morphologies for the impregnation route (figure 2(b)). Do the authors have any ideas about approximate sizes? The authors say that the surfactant method preparation (figure 2(a)) shows a homogenous distribution of Mn. The images in (a) and (b) have different length scales so it makes it quite difficult to judge the homogeneity of the Mn. Do the authors learn anything from the distribution of the N, C and O? If anything, the N K edge image seems to show a more irregular morphology in figure 2(a) (surfactant route) than figure 2(b) (impregnation method).

 

5. Page 7 and figure 3. In the main text, the authors refer to figure 3(b) and say the N2 peak at 399.9 eV is green, but in the figure it is blue. Similarly, the main text says that the N3 peak at 401 eV is blue, but in the figure it is green. The authors say that on a closer inspection of the N1s there is a small component at 398.9 eV in the impregnated samples. This is shown as an inset in figure 3(b) but it’s very confusing. There is no binding energy scale on the insets and the small component is shown in black in the inset but is red on the main plot (if I’m looking at it correctly). This needs to be made clearer for the reader.

 

6. Page 10. Should the XRD section have its own sub-heading? Currently the XRD data is under section 2.3 XPS characterisation.

 

Here’s a few typos I spotted:

Page 7, line 225. “XPs” should be XPS.

Figure 5. Axes label “wavelenght” should be wavelength.

Page 13, line 388. “phase phase” repeated word.

Author Response

Reviewer 1:

We very much thank this referee for their constructive and extremely helpful review. We have addressed fully the points 1-3 and corrections are highlighted on a yellow background.

Our comment on Point 4 is here below.

A: We welcome the suggestion of the referee very much and have changed the STEM-mapping (now Figure 3b) relative to the sample obtained via OAR such that both samples are now at the same magnification. We have also included additional TEM characterization, including HR images (Figure 2 and 4), in which the two methods show a completely different particle size distribution. Basically, in the OAR sample, the NPs are extremely small and below 2nm, the rather similar contrast in the TEM images makes it difficult to discriminate them from the support. In contrast on the impregnated samples the Mn oxide particles are visible in various size and shape. In this particular case, it is very difficult to provide a number because these samples are very difficult to visualize by TEM. As suggested by the referee we further analysed the elemental mapping and included the elemental composition expressed in At% in table 2. In relation to the point addressed on the inhomogeneity of the N distribution, we have no evidence that this is the case. The new mapping on this sample justifies our point. We found some interesting points on the Mn species. A discussion on this is also included now and highlight in the main text at pag. 6 lines 211-246.

point 5: Thank you very much for pointing this out. This is very very helpful.  This is now sorted and the image as been improved following the referee suggestion.

All the other points were addressed.

Reviewer 2 Report

The ms # catalysts-935499  ‘Investigations of carbon nitride-supported Mn oxide nanoparticles for ORR.’by Alexander Ian Large , Sebastian Wahl , Salvatore Abate , Ivan da Silva , Juan Jose Delgado Jaen , Nicola Pinna , Georg Held , Rosa Arrigo, addresses the oxygen reduction reaction over Mn oxide nanoparticles supported on carbon nitride. Two methods for the catalyst preparation were utilized and the properties of the resulting materials were compared.

The manuscript is well organized; however, there are a number of flaws that are described below. Please address the comments before preceding further publication.

 

1. ‘Earth‐abundant Mn oxide nanoparticles are supported on carbon nitride using two different immobilization methods (…)’ Please rewrite. Although manganese is the third most abundant transition metal in the Earth's crust, the Mn NPs are not.
2. What is the novelty of the work? A similar study has been presented using Mn porphiryn, where interaction/Mn stabilization of Mn NPs can be similar as in the described carbon nitride material. Journal of  the  Japan  Petroleum  Institute,  57,  (6),  237-250  (2014)
3. Line 35 ‘high abundance of N‐functional groups acting as anchoring sites for metals species”. The sentence is not entirely correct. Please address M. Antonietti, MCMillian works concerning carbon nitrides material, as well as recent work by Pieta et all (ACS Sustainable Chem. Eng. 2020, 8, 18, 7244–7255); where the material structure is addressed, and cavities sizes are fully calculated.
4. Why the catalyst based on porphyrins are not included for comparative study? There are many studies shoving excellent activity based on Co or Mn NPs supported on porphyrins. Instead, the Authors give less durable materials silica or other inorganic oxides based. This is very disappointing, as many relevant, very recent information is missed.
5. Based on XPS study, could you provide C:N ratio for obtained materials. Many publications have reported such materials as "carbon nitride" : however, the materials produced in many cases not been shown to be based on carbon and nitrogen in a 3:4 ratio.
6. Lines 315-323, please rewrite and give a proper explanation regarding the material used. Please comment on the nature of the carbon nitride support material. The Explanation given in many points is mixed and referred to different carbon nitride structures. Here in XRD, authors bases on features indicative for the stacking of -conjugated layers and inplane repeats of heptazine structural packing motifs with a spacing of ca 7 Å. Please revise the data analysis considering changes in long-range ordering for Mn modified and thermal treated material as well as interplanar pores. Surely the material crystallinity was affected upon heating.
7. Fig 5. Please provide scale appropriate for UV-Vis, at least until 1100-1500 nm. Any influence of the preparation route on influence on d-d transitions has been observed? Why the signals intensities are different? Please use the Absorbance scale instead of Reflectance, and provide Tauc plots, which will be more informative for potential readers.
8. Both the current divided by the electrode surfaces and by the mass of electroactive material deposited on the electrode surface should be shown in Figure 1.Why CV curves for T-MnOx_WI_CN shown in Fig. 1a (red curves) differ from those shown in Fig. 1b (blue curves)?
9. For a diffusion-limiting system, a well-defined plateau should be observed in the polarization current. Instead, a significant decrease in current is seen on the LV curves for potentials below ~0.5 V. This behavior is also clearly visible in the Tafel plot where three regions can be recognized, i.e., from 0.8 to 0.65, 0.65 to 0.55 and 0.55 to 0.2 V. Such behavior suggests another ORR mechanism than that proposed by the Authors. Moreover, CV curves show a peak current at ~0.5 V followed by a continuous decrease in currents. What is the origin of this behavior? CV for different potential scan rate should be performed and analyzed.

Author Response

We thank the referee for the work of reviewing our article and for the useful suggestions aimed to improve our contribution, which we welcome and address fully. Our point by point answer follows.

Point 1.

A: We agree with the referee that it is not the Mn oxide NPs that are earth abundant but rather Mn resources are widespread on the earth crust as opposed to Co, Li and so on. Following the referee´s suggestion we have reformulated the sentence as “Earth abundant Mn-based oxide NPs”. This would be a more precise way of describing our goal of using Mn as the redox element for ORR.

point 2.

A: We thank the referee for suggesting additional materials from the literature which we are including in the references to make our work more comprehensive. . This can be found as

48. Y. Yamanaka, Japan Petroleum  Institute, 2014, 57, 237-250

We are also grateful because his point has triggered further consideration and reasoning on the structure function correlation that we aimed to built in this contribution. Our work is substantially different from the case of the single site Mn porphyrin system in the paper mentioned above. Certainly, at the molecular level, the Mn-N interaction at the metal/support interface of the MnOx-CN electrocatalysts can be of similar nature to Mn-porphyrin systems but the implication of this on the reactivity of the oxide NPs is different as shown by the different reactivity of the two systems MnOx-WI_CN (large particles) and MnOx-OAR_CN (small particles and possibly Mn single or few atom clusters interacting with organic ligands) as well as the performance reported in the article suggested by the referee. In particular our results show that the large mixed valence Mn(II/III) oxide NPs favour the 4e- transfer reduction as opposed to the 2e- transfer reaction, which would occur on C materials and on our Mn-OAR system where single atoms could be inferred based on the additional TEM characterization which we have included in the new version of the manuscript in Figure 2 and 4.

Moreover, as we explained in the introduction, our target technology is reversible fuel cell technology and our goal is to generate a PSII biomimetic electrocatalysts by synthetizing mixed Mn oxide NPs encapsulated and protected by the CN matrix, which would simulate the role played by the pocket of proteins in the enzyme.  The synthesis of our material is also considerably cheaper than metal porphyrin, which is clearly another advantage.

point 3.

Whilst it is not clear to us why the referee considers the sentence incorrect; we are suggesting that the referee means probably that the N species are potentially acting as anchoring site but not necessarily are all involved in interactions with the metal species? We refer here to our previous extensive contributions on the nature of the interaction N-Metal including on the very same CN used in this work (ACS Catal. 2015, 5, 5, 2740–2753; ACS Catal. 2016, 6, 10, 6959–6966) performed using synchrotron based X-ray absorption spectroscopy which provide the highest resolution chemical state. In these articles we have proved the nature of the interactions. 

We are modifying the text to reflect this idea. Changes are now on a yellow background (page 1 line 38-39).

We do agree that seminal contributions from the group of McMillan and Antonietti are relevant in this field but we think that we did include already enough material from M. Antonietti ( see references (1) ACS Catal. 2016, 6, 10, 6959–6966; 2) Wang, X.; Blechert, S.; Antonietti, M. Polymeric Graphitic Carbon Nitride for Heterogeneous Photocatalysis. ACS Catal. 2012, 2, 1596–1606, doi:10.1021/cs300240x; 3) Su, F.; Mathew, S.C.; Möhlmann, L.; Antonietti, M.; Wang, X.; Blechert, S. Aerobic Oxidative Coupling of Amines by Carbon Nitride Photocatalysis with Visible Light. Angew. Chemie - Int. Ed. 2011, 123, 683–686, doi:10.1002/ange.201004365) and McMillan ( (4) Miller, T.S.; Belen Jorge, A.; Suter, T.M.; Sella, A.; Cora, F.; McMillan, P.F. Carbon nitrides : synthesis and characterization of a new class of functional materials. Phys. Chem. Chem. Phys. 2017, 19, 15613–15638, doi:10.1039/c7cp02711g., ).

We thank the referee for the suggestion of the work of Pieta, which we are now referring to as well. This can be found in reference 8b.

Point 4: 

This point is somehow very similar to point 2 and at the same time going against it. We would like to reiterate that our focus is on oxide NPs using very simple, cheap and scalable synthetic procedure. Our paper is more on understanding how we can improve/tailor the synthesis of carbon nitride-based materials to make them useful for electrocatalysis. Our results show a direction for improving the materials performances, but not stability. We are also targeting reversible fuel cell technology and not H₂O₂ direct electrosynthesis. We disagree with the referee on the fact that porphyrin systems must be included in this study simply because they do not enable us to achieve our goal, based on the performances reported in the article suggested by the referee. We also cannot help but notice that the referee at point 2 is questioning  the novelty of our work with respect to the porphyrin system, and whilst we have demonstrated that they are not the same, we also think that including porphyrin system (which is  anyway not our field of synthetic expertise) would significantly jeopardize the novelty of our contribution as suggested by the referee at point 2.

Point 5:

A: We thanks the referee for this valid point. Indeed, CN is a class of materials in which the building block units condense with each other to form structure with various degree of aromatization.

We welcome very much the suggestion of this referee and improved our contribution to make this point clearer as well as clarify the nature of the CN materials we are using in this work. (see for example line 27-30 in page 1).

In terms of the C:N ratio, we have now included the results of the elemental composition by bulk sensitive EELS characterization in table 2. We also describe this result therein indicating the atomic ration 1:1.

However, we do not think that the analysis of the C:N ratio either by XPS or EELS  is informative enough in this case for several reasons:

a)since the difference between the elusive C3N4 theoretical structure and our material is simply the increase of H terminated edge sites;

b) non-synchrotron based XPS measure the core level C1s and N1s at different excitation energy (being apart of circa 120 eV) and even if this difference is small, it still probe different depths, which together with a possible sample inhomogeneity at the surface as reported in our paper on the very same sample (ACS Catal.2016, 6, 10, 6959–6966) makes the quantification not useful.

c) XPS is a surface technique and we observed that the surface of this materials is different from the bulk ( due to presence of enriched C deposits thereby making this analysis incorrect.

The XPS analysis shows the N species expected for heptizine. This is now clarified and justified at 7 page lines 249-251 on a yellow background.

The XRD characterization reported in this work shows clearly high anisotropy in the in-plane and inter-planar stacking of the diffraction planes, which indicates that the structure of our material is “characterized by particles composed of many narrow ribbon-like platelet crystals stacked together”. This was already indicated in our first submission (see line 340-342 page 11).

Based on the XRD characterization included in this work and the NEXAFS characterization presented previously (Arrigo et al ACS Catalysis 2016) we conclude that our material is not the g-C3N4 materials which was identified theoretically as the most thermo-dinamically stable form of CN. This is in fact the reason why we have decided to name this material with the acronym CN standing for carbon nitride without any reference to its stoichiometry.

In our previous work, a detailed NEXAFS analysis has also indicated a deviation from the theoretical structure of C₃N₄, in which we have identified high abundance of C atoms, with limited aromatic π delocalization. Consistently terminal sp2 /sp3 −C=N−H ↔ =C−N−H enamine/imine N functionalities are highly abundant for this sample (ACS Catal. 2016, 6, 10, 6959–6966). We have included this information in the new version of the work at page 11 line 344-346.

Point 6

In terms of the XRD diffraction peak of the CN phase, we are referring to the comprehensive review by McMillan´ s group.

Therein the complexity of the analysis of the XRD data is addressed by referring to combined theoretical and experimental works from the literature. It was not our aim to analyse by XRD the change of the long range order upon Mn immobilization and we have observed that for the materials tested, the structure function correlation involves the nature of the Mn species rather than the long range order or the semiconducting properties of the resulting CN phase.

In this work, the changes of the core-level (local) electronic structure upon immobilization of Mn was analysed by XPS. The impact of Mn immobilization on the band structure and thus electronic properties of the catalysts was analysed by UV-Vis, which is also affected by a change of the long range order. The Mn phase was also chachertized by XPS and XRD.

However, we noted that the diffraction peaks related to the nitride phase are consistent with the poliheptazine-derived structure rather then the triazine one and therefore we included this in description of the data. This can be found at page 13 lines 356-362. Both the MnOx_WI_CN and MnOX_OAR_CN, which were investigated by XRD gave the same diffraction patterns in terms of peak position and broadening and therefore we believe we can safely say that the thermal annealing at 200C after the immobilization of Mn did not affect the long range order sensibly (note that between these two sample only the MnOx_WI_CN was thermally annealed after the immobilization), thus changes in the band gap can be attributed to the electronic  effect from Mn. 

Concerning the comment on the material crystallinity which is affected upon heating as seen in the in situ experiment,  We can confirm that the annealing at 500° would modify the long range order and as consequence the band gap compensating for the beneficial effect of Mn (Figure 7 sample MnOx_WI_CNa). We have added some comments on this (page 15 line 430-434). 

However we also believe that we did not make it clear the aim of the in situ XRD experiment. The main aim of the in situ XRD analysis is to unveil the structure of the Mn oxide NPs. These are very small and poorly crystalline, therefore giving a not very intense and broad diffraction patterns. By performing an in situ XRD annealing we were hoping to identify Mn oxide structural transformation that could help us to demonstates that the small peaks in the initial sample were relative to the Mn phase and upon annealing the expected phase changes were observed.

To make this point clear we have added an additional sentence in the main text. This is now more clearly explained at page 10 between line 325-330. 

Point 7

The figures has been modified following the suggestion of the referee. Regarding the d-d transitions, the F(R) don’t show clear peaks that may correspond to the d-d transitions. In fact, similar spectra are reported in the literature for CNx samples obtained at high pressures and temperatures. Therefore, we can´t conclude that there are modification on the d-d transitions. The experiments were carried out depositing the sample on a BaSO4 pellet to avoid any interference of the sample holder. Several experiments were carried for each sample increasing the amount of material to guarantee that we reach the highest absorption possible. Therefore, the intensity differences can be related to the intrinsic optics properties of the materials. However, small differences can also be related to some un-controllable experimental conditions such as the density of the pellet.

The details of the sample preparation are now explicitly mentioned in the experimental part.

Point 8

We are glad to the referee for noticing a change between the Fig 1a and b. Indeed there difference is that in  the first case the a mixture of CN and activated carbon was tested to compensate for the poor conductivity of the CN. However we found that this would lead to a electrocatalytic  behaviour dominated by activated carbon. Therefore to extract a more accurate structure function correlation we perform the tests on the pure CN samples. However we reckon that this was an oversight during the preparation of the manuscript and we have corrected the description of the figure accordingly (page 3 lines 138-142)  and included in the experimental part the preparation of the CN/AC mixture (lines 574-583). 

However, we are not convinced that both the current normalized by the electrode surface area and the mass of the electroactive material deposited should be reported. The quantities of the electroactive material are the same for every test so it would basically requires just dividing the data related to each sample by the same number, which is anyway indicated in the experimental part. We think that we would not gain any important insights but rather making the plots more crowded. If we are to chose the normalization parameter then we prefer to use the geometric surface because this is the only parameter that it is kept constant during the experiment. We are aware of the limitation of this testing approach as we have reported already in the original submission (lines 210-211 on the challenges of the thin film method).

Point 9:

We thank the referee for this comment and useful suggestion. We have reconsidered the discussion of the electrocatalytic performances and provide a more detailed explanation of the observed electrocatalytic behaviour. Changes are included at line 176-182. This is summarized here following:

In the diffusion limited region below 0.4 V a well-defined plateau would be expected. A monotone increase of the cathodic current is observed instead, commonly attributed to secondary reactions. This behavior is commonly observed when the the availability of sites that can break the O–O bond of O₂ is limited and H₂O₂ is the main reaction product [12].

Other changes are also included in the same section. Moreover a discussion based on the redox chemistry of Mn was included in the discussion section (line 506-515)

All the changes are highlighted on a yellow background. Finally we would like to point out that we did perform LV at different scan rate for all the samples and included in Figure 1d the one relative to the MnOx_WI_CN. We believe that we have provided enough information on the electrochemical performance of these systems and further analysis is not necessary for this contribution. We are currently attempting to improve the electrode preparation, which is a big challenge with the aim to explore furthemore the electrochemical performances of Mn modified CN systems in line with the referee suggestion. 

 

Reviewer 3 Report

The whole of this study is not of great interest to the community.
The low conductivity of these systems drastically limits their use in fuel cells

Author Response

We do agree with this referee that there is a need of finding a solution to improve the conductivity of this material, which has attracted the attention of the scientific community due to the high corrosion resistance, low cost and large availability, let alone the high versatility of this system to switching from photocatalytic to electrocatalytic technologies and viceversa. Our contribution fairly summarize the challenges we have encountered in working with these systems. We also identified directions and particularly the optimization of the Mn phase to trigger the 4e- transfer reaction, which is of great interest. Our conclusion is indeed that it is possible to realize it and we believe it is worth of disseminating this result. There are a plethora of examples showing that the cost of the overpotential could be reduced by other chemical approaches and we have several ideas on how to realize that based on the results presented here. We think that our honest and detailed contribution is beneficial for other researcher aiming to work with CN.

Overall this referee was positive in the judgement of our contribution except on the introduction that accordingly must be improved. We have attempted to improve it following other referee suggestions. All in all, we think that the feedback from this referee is positive, however, their only aim was to make the authors aware that at this current stage we are far from making this system of commercial interest for fuel cell application. This comment is fair  and welcome. 

 

Round 2

Reviewer 3 Report

This publication will serve as a working document for all groups studying on noble metal free catalysts.
The authors are aware of the limitations of their system but their arguments are convincing.
This publication will therefore help the scientific community to find the reliable environmental solution

Author Response

We thank the referee very much for their feedback.