Turning a native or corroded Mg alloy surface into an anti-corrosion coating in excited CO2

Despite their energy-efficient merits as promising light-weight structural materials, magnesium (Mg) based alloys suffer from inadequate corrosion resistance. One primary reason is that the native surface film on Mg formed in air mainly consists of Mg(OH)2 and MgO, which is porous and unprotective, especially in humid environments. Here, we demonstrate an environmentally benign method to grow a protective film on the surface of Mg/Mg alloy samples at room temperature, via a direct reaction of already-existing surface film with excited CO2. Moreover, for samples that have been corroded obviously on surface, the corrosion products can be converted directly to create a new protective surface. Mechanical tests show that compared with untreated samples, the protective layer can elevate the yield stress, suppress plastic instability and prolong compressive strains without peeling off from the metal surface. This environmentally friendly surface treatment method is promising to protect Mg alloys, including those already-corroded on the surface.

Mg is not MgO.... its nominally Mg(OH)2, often with some traces of MgO. It is a hydarated oxide that forms rapidly in (moist) air and water.

Reply:
We thank the reviewer for pointing out this inadequate statement.
The second sentence in abstract is "One primary reason is that the native surface oxide on Mg formed upon exposure to air consists of mainly MgO, which is porous and unprotective, especially in the humid environment." Initially, we intended to emphasize the first product upon exposure to air is mainly MgO. According to the XPS and TEM results of Nordlien et al. (Nordlien, J. H., et al. "A TEM investigation of naturally formed oxide films on pure magnesium." Corrosion science ,1997), the initial film formed immediately after exposing fresh surface by scratching in air contains 50%~60% wt% (40%~50% at%) magnesium hydroxide, and the other composition is MgO. Certainly, our claim is not entirely right in context. So, we revised this statement as "One primary reason is that the native surface film on Mg formed in air mainly consists of Mg(OH) 2 and MgO, which is porous and unprotective, especially in humid environment." Elsewhere in the manuscript, we also replaced MgO with oxide/hydroxide or native layer formed in air.
In addition, it should be noted that Nordlien et al. also reported the water evaporation from the hydrated zones of films formed on Mg and subsequent crystallization of the dehydrated zones into MgO occurring in TEM analysis due to high energy electron irradiation. (Nordlien, Jan Halvor, et al. "Morphology and structure of oxide films formed on MgAl alloys by exposure to air and water." Journal of the Electrochemical Society ,1996. Nordlien, J. H., et al. "A TEM investigation of naturally formed oxide films on pure magnesium." Corrosion science ,1997). So, the carbonation of the air-formed film on our pillars' surface we observed in TEM should be dominated by the reaction MgO+CO 2 =MgCO 3.

Comment 4:
A timeline of Mg (and corrosion), which incidentally corrects such overstatements, and incidentally shows that CO2 is not novel, is given by Esmaily et al (Progress in Materials Science).
The reduction of Mg corrosion rate by CO2 (to form carbonates) has been well and truly published many times before. Many industrial processes even use CO2 as a corrosion inhibitor......

Reply:
The reviewer thought that our work is not novel because the inhibitive effect of CO 2 on Mg corrosion has been reported. In the following, we will summarize the relevant researches progress and demonstrate their huge differences with our work.
It was reported that CO 2 inhibits Mg corrosion by 1) lowering pH (when dissolved in water, CO 2 forms carbonic acid), impeding the development of macroscopic corrosion cells and resulting in inhibition of pitting corrosion; 2) forming a magnesium hydroxy carbonate or carbonate-containing surface film, which can lower the conductivity of the electrolyte, block the anodic and cathodic reactions.
(Martell A E, Smith R M. Critical stability constants [M]. New York: Plenum Press, 1974. R. Lindstr€om, J.-E. Svensson, L.-G. Johansson, J. Electrochem. Soc. 149, 2002. Lindström, Rakel, et al. "Corrosion of magnesium in humid air." Corrosion Science,2004  Esmaily, M., et al. "Fundamentals and Advances in Magnesium Alloy Corrosion." Progress in Materials Science, 2017.) Based on these, we see that formation of Mg carbonate-containing film is only a partial reason for the inhibitive effect of CO 2 gas on the Mg corrosion. Moreover, the following two points should be noted: 1) The mechanism of CO 2 as corrosion inhibitor has not been fully understood. A study even shows that WE43 Mg alloy dissolved ambient levels of CO 2 slightly accelerated the corrosion rate and increased general surface corrosion compared to a CO 2 -free environment: the average corrosion rate in the ambient CO 2 was 0.16 mg/cm 2 ·day compared to 0.09mg/cm 2 ·day in the absence of CO 2 (Kaminski, Daniel Thomas.

Comment 6:
The claim that "This environment-friendly surface treatment method is expected to be universal to protect Mg-based materials, even those already-corroded Mg alloy workpieces." is entirely overstated. Cladding with aluminium can also do the same, so such a statement is not adding new science, or even impact.

Reply:
We disagree the statement "This environment-friendly surface treatment method is expected to be universal to protect Mg-based materials, even those already-corroded Mg alloy workpieces" is entirely overstated. It is a reasonable conclusion and expectation according to our experimental results. The reviewer's comment that "Cladding with aluminum can also do the same, so such a statement is not adding new science, or even impact" is too harsh. We can add some references from Kish et al., if the reviewer still feels they are necessary, after seeing our concerns above.

Comment 9:
All that Figure 3b shows is a galvanic couple, not protection at all. One side is the cathode, one the anode. This is how corrosion works..... All it takes is a small potential difference and this occurs. This is not so whopping.

Reply:
We disagree with the reviewer that Figure 3b is only a galvanic couple instead of the protection effect from MgCO 3 film. Figure 3a has obviously shown that the Mg pillar treated in excited CO 2 is well protected compared with its counterpart without any treatment. And in Figure 3a case, there exists no "galvanic couples" because the entire Mg pillar was coated with MgCO 3 . Figure 3b aims to further prove the protection of MgCO 3 film using the same Mg alloy pillar to exclude any possible discrepancies between different samples.
Increased corrosion potential of the treated part unambiguously shows the protective effect of MgCO 3 coating, and this increase is significant, as shown in our supplementary electrochemical tests on bulk Mg samples (see Figure S12 and Table   S1 in Supporting Information); meanwhile the corrosion current density also decreases by two orders of magnitude. So, the MgCO 3 film formed in excited CO 2 can greatly decelerate the corrosion rate and obviously improve the corrosion resistance of Mg alloys.

Comment 10:
The claim in Figure 4 that "Turning corroded Mg alloy surface into protective coating." is an overstatement. There is quite a lot of metal loss.
Reply: Figure 4 compares the corrosion resistance of the native oxide/hydroxide layer and MgCO 3 film on the same Mg alloy pillar. We deliberately pre-corroded an Mg alloy pillar and meanwhile made sure that one part of the pillar was uncorroded by tuning the immersion time. And then the corroded part was exposed to e-beam illumination and the flurry corrosion product reacted with excited CO 2 to produce MgCO 3 film wrapping this part (see supplementary movie 3). At that time, the uncorroded top part was bare. After the immersion in water and subsequent exposure in humid air, the originally corroded lower part kept intact with the protection of MgCO 3 , while the unprotected top part was seriously corroded, and that caused the so-called "metal loss".
Therefore，the summary of Figure 4 "Turning corroded Mg alloy surface into protective coating" is drawn from the real and solid experimental evidence.

Comment 11:
Irrespective of much of the above, the authors describe a coating... in essence. It is not a protective coating of value unless it is protective in a defect, or self forming (like a passive film). The Mg-Li film previously demonstrated on Mg-Li alloys is self healing (by self reforming), so it protects defect. Coating Mg with paint of eplating with Cu or Ni is common. None of the latter protect defects, and neither does the coating decribed here. This means it is not actually interesting, or useful.

Reply:
We agree that a self-reforming or self-healing protective film like the passive oxide on the surface of aluminum or titanium would be great. But there are applications where other surface treatments would be desirable, as long as the corrosion rate can be judiciously controlled. And every method has its own advantages and shortcomings. For example, the common chemical conversion coatings with self-healing ability face the toxicity, environmental contamination and poor mechanical properties, etc. Even the "stainless" Mg-Li alloy reported by Xu et al. is not always perfect because it is not suitable as the bio-degradable material. The surface treatment method we reported here is not perfect either, but it is easy to operate, environmentally-benign and especially suitable for pretreating small sized devices and bio-medical implants.
Nowadays, most of Mg alloys are still facing serious poor corrosion resistance problem. Therefore, to develop new corrosion prevention methods is not only necessary but also in a press need. Each Mg alloy used in different applications should has its own specific and optimized corrosion prevention method.

Response to the reviewer #2:
Comment 1: This is an interesting paper about the use of room temperature pretreatments to improve the protective oxide of Mg. These authors demonstrate via in situ TEM (environmental and mechanical) that the native oxide on Mg can be dramatically improved by exposure to ionizing CO2 gas (using electron beam) to change the native oxide to a protective MgCO3 layer. The authors claim that this oxide is more resistant to corrosion and more adherent to the underlying Mg. The work is of high quality, especially the in situ microscopy. Sufficient details in the paper and the supplement are given for others to reproduce the data.

Reply:
We thank the reviewer for pointing out the significance of our work! And we are glad to hear that the reviewer considers our work to be of high quality. We also highly appreciate the valuable suggestions given by the reviewer. In the following, we respond point-to-point to each comment.

Comment 2:
However, the corrosion experiments were done by only immersion in deionized water.
John T. Yates, Jr. paper in Langmuir 2001, 17, 2146-2152 also shows that using different oxidation conditions can improve corrosion behavior as well, where they used an electrochemical cell to verify improved corrosion behavior. This paper could be used as an example of how to further characterize the MgO versus the MgCO3.
The authors could also demonstrate improved oxidation behavior as well, as adhesion of the oxide is important for high temperature corrosion field, where cyclic oxidation using TGA is a standard method to demonstrate improved oxidation behavior.

Reply:
We thank the reviewer for the valuable suggestions. Our original goal of the present work was to explore an environment friendly and efficient method to improve the corrosion resistance of Mg alloys especially at small scale. So, we focused on the method development and proof processes with the convincing in-situ experimental evidences. But it is necessary to further verify the improved corrosion resistance using standard electrochemical corrosion tests. The anti-corrosion properties in 3.5% NaCl solution of the MgCO 3 protective films on the bulk Mg samples were evaluated by using potentiodynamic polarization measurements.
Considering that the native film on Mg surface is only tens of nanometers, the formed These bulk Mg samples with pre-corrosion surface film were treated in CO 2 plasma for 1 hour or 2 hours. The corroded surface with plate-like morphology were carbonated resulting in a relatively flat and dense MgCO 3 surface (see below Figure   S12). And then the anti-corrosion properties of MgCO 3 films were evaluated in 3.5 wt.% NaCl solution by using potentiodynamic polarization measurements. Compared with their counterparts, the corrosion resistance of bulk Mg samples with MgCO 3 protective film has been enhanced dramatically: the corrosion potential (E corr vs.SCE ) increases, and the corrosion current density (i corr ) can decrease by two orders of magnitude (see below Table S1).
After corrosion tests, large corrosion craters can be clearly observed even by naked eyes on pristine Mg surface. But the surfaces of the samples treated in CO 2 for 2 hours are still intact and there are no visible cracks ( Figure S12). The anti-corrosion effects of MgCO 3 protective film can be better than the anodized coating, and even can be comparable with the composite coatings obtained by micro-arc oxidations, see The optimal parameters, such as the immersion time in water and CO 2 plasma treatment time should be further explored so as to get optimized anti-corrosion properties.
The newly formed MgCO 3 protective film also displays significant oxidation resistances. See our reply to the comment 4 for experimental details and results.
Since the present work mainly focuses on the mechanism and microscopic characterization, we have supplemented in-situ high-temperature oxidation experiments on the Mg pillar with MgCO 3 protective film.
As for the further characterizations of MgO/Mg(OH) 2 vs. MgCO 3 , such as thermogravimetric analysis (TGA) to compare their thermal stability, nano-indentation or nano-scratching to show the mechanical properties of the film on Mg and so forth, we are going to report them in another work systematically and comprehensively.  Another point is that more specifics of the SAED analysis could be given to be more convincing that the oxide formed is MgCO3 (or mixed with MgO).

Reply:
We thank the reviewer for this good advice. We made the pure nanoscale MgO crystal react with excited CO 2 ，and by indexing the diffraction pattern of the product we verified the product was MgCO 3 . Accordingly, it is reasonable to conclude that the oxide layer on Mg pillar surface reacting with excited CO 2 should be MgCO 3 . We also tried to use the EELS analysis (carbon element signal) and SAED to further confirm that the product is MgCO 3 (see Figure S4 in supporting information). As for more specifics of the SAED, it is impossible with our TEM to select the MgCO 3 film only, which is several nanometers thick using the selected area aperture. Therefore, we took the high resolution TEM images of the surface areas of the Mg pillar treated in excited CO 2 , and then we could get the diffraction information of very small

Action taken:
We have included this part in supporting information as Figure S5. And the figure indices in SI have been updated accordingly.

Comment 4:
As the authors have access to an ETEM, could the authors expose the samples in situ to O2 or H2O vapor to determine if the MgCO3 in contrast to the initial MgO does slow down or change oxidation behavior at higher temperatures?

Reply:
We appreciate this good suggestion and carried out the in situ heating tests in gas environment. Because water vapor is not one of the standard gases recommended by the Hitachi environmental TEM, we exposed the samples in O 2 at high temperature to observe the different responses of the Mg pillars with and without the MgCO 3 coating.
The two figures below were extracted from the in situ videos recorded during the heating processes (from 20 to 200 ) of the Mg pillars with native oxide layer and MgCO 3 layer, respectively. Clearly, we can see that the native oxide layer couldn't protect the substrate metal from the oxygen attack at high temperature and the Mg pillar was seriously oxidized. In contrast, the morphology of Mg pillar with MgCO 3 protective film almost kept unchanged and the film was also intact to some extent at 200 , which proves the protective effect of the contact MgCO 3 film once again unambiguously. Presumably, this is because the compact and stable MgCO 3 film can effectively isolate the oxygen gas from Mg metal for the tested temperature range.
It should be noted that reaching 200 is a quite harsh environment for the nanoscale pillars. Action taken: We have included this part in supporting information as Figure S8.

Comment 5:
In the conclusion section, it would also be beneficial to the general audience to reiterate the broad application of Mg in technology and what environments that Mg would be exposed to in these technological applications. It would also be helpful to again provide some thermodynamics or kinetics of this reaction (MgO+ CO2 -> MgCO3) in the conclusion section in order to give a broader implication of how to select the pre-treatments to create better protective oxides on other metallurgical systems.

Reply:
Thanks a lot for the reviewer's advices. In the conclusion section, we have reiterated the broad applications and exposure environment of Mg alloys in service, and also the thermodynamics of the carbonation reaction.

Action taken:
In the conclusion section, we have modified the part before "The entire surface treatment process has been recorded in real time……" as "In summary, Mg alloys have broad uses in 3C products, automotive, aerospace and biomedical industries.
However, all of these applications face corrosion resistance problem in service, especially in the humid or aqueous environments. We developed an easy, environment-benign and effective anti-corrosion method: carbonation of the air-formed oxide/hydroxide film or hydrate corrosion products on Mg alloys' surface into a smooth, compact MgCO 3 protective layer by ionizing CO 2 gas using either high energy electron beam or plasma. The excited CO 2 accomplishes at room temperature the reaction MgO+ CO 2 →MgCO 3 , which usually occurs above 400 at atmospheric pressure. No extra heating or pretreatments are needed, rendering this method especially suitable for protecting small-sized or complex-shaped Mg alloy workpieces and for replenishing those already-corroded. Rather than having to mechanically clean away prior corrosion damage, one can consume it directly to create new protective surface."

Comment 6:
Some minor points: SM video 2: the beginning should be edited out.
Subtitles should be removed (e.g., Verification of Idea).

Reply:
We have made corrections accordingly in the manuscript and supplementary materials.

Action taken:
The beginning part with e-beam blocked off in movie 2 has been edited out.
Subtitles in manuscript have been removed. Figure 2b has been rewritten as "……MgO flakes did not change……"

Comment 7:
To summarize, this is an interesting paper that should influence the field of corrosion, oxidation, and nanoscience and technology, as the authors clearly demonstrate that different pretreatments especially reactive gas species can improve properties significantly. Although the TEM is excellent and convincing, the claim that the corrosion properties are improved through exposure to deionized water could be better. It would benefit the paper to carry out the more standard electrochemical corrosion or higher temperature cyclic oxidation tests to demonstrate that this pretreatment improves Mg resistance to corrosion or oxidation properties.

Reply:
We thank the reviewer for this concise summary of our work. We also really appreciate these valuable suggestions.
We have modified our conclusion as "the corrosion resistance of Mg alloys in deionized water can be improved". And as shown above, we have supplemented the standard electrochemical corrosion tests and higher temperature oxidation experiments to prove the superior anti-corrosion properties and stability of the MgCO 3 protective film.

Response to the reviewer #3:
Comment 1: This paper demonstrated that magnesium oxide could be converted into magnesium carbonate by excited CO2 in an environmental TEM, and the formed magnesium carbonate film could enhance the mechanical properties and inhibit the corrosion of magnesium in water. These are innovative findings. The most impressive experiment in this study was the in-situ observation of the formation process of magnesium carbonate. Therefore, the reviewer would like to recommend publication of this paper.

Reply:
Thank the reviewer for the accurate and comprehensive summary of our work. We are glad to hear that the reviewer considers our work to be scientifically sound and to be of general interests to the readers of Nature Communications.

Comment 2:
However, this paper can be further improved if the following issue could be properly addressed.
Corrosion resistance is critical for a surface film on magnesium. Since the formed magnesium carbonate film in this paper was claimed to have superior corrosion resistance in water, it would be worthwhile to compare this film with a simple coating, e.g. an anodized coating, on the magnesium under the same condition. The reviewer believes that such a comparison will interest many more scientists and engineers. If These bulk Mg samples with pre-corrosion surface film were treated in CO 2 plasma for 1 hour or 2 hours. The corroded surface with plate-like morphology were carbonated resulting in a relatively flat and dense MgCO 3 surface (see below Figure   S12). And then the anti-corrosion properties of MgCO 3 films were evaluated in 3.5 wt.% NaCl solution by using potentiodynamic polarization measurements. Compared with their counterparts, the corrosion resistance of bulk Mg samples with MgCO 3 protective film have been enhanced dramatically: the corrosion potential (E corr vs.SCE ) increases, and the corrosion current density (i corr ) can decrease by two orders of magnitude (see below Table S1). After corrosion tests, large corrosion craters can be clearly observed even by naked eyes on pristine Mg surface, but the surfaces of the samples treated in CO 2 for 2 hours are still intact and there are no visible cracks ( Figure S12).
Note that the anti-corrosion effects of MgCO 3 protective film are much better than the anodized coating, and even can be comparable with the composite coatings obtained by micro-arc oxidations, see Table S1.   The revision is worse than the original, as the authors are oblivious to the shortcomings. Right from the title, the paper is focused on corrosion resistant coating. As such, it can only be viewed in that context (and not viewed in the context of what the paper isn't). In terms of corrosion resistance, the Mg-carbonate is not self-healing, it is likely to be soluble when wet, and it is not anywhere near as novel as the authors claim. All that figure 3B is showing is a galvanic couple, not corrosion protection. This really is a very poor paper. Once again, the authors have avoided the might of the literature and opted to cite irrelevant textbooks in oxidation, etc. This paper would not survive pre-screening at a corrosion journal, and I cannot in good faith say it is anywhere near suited to Nature Comms.
My recommendation is a firm reject.

Reviewer #3 (Remarks to the Author):
This reviewer is happy with the revision.
Reviewer #4 (Remarks to the Author): The submitted manuscript introduces a detailed experimental study showing that the native surface oxide on Mg can be converted MgCO3 via e-beam irradiation (or plasma) in a CO2 atmosphere. With a set of corrosion tests, the authors further show that the resulting MgCO3 surface layer has improved corrosion resistance compared to the native surface oxide. These results are interesting and the experimental evidence is convincing. However, the following issues including the fundamental mechanism regarding the e-beam assisted MgCO3 formation require careful clarification: 1) The e-beam assisted MgCO3 formation is attributed to the excitation of CO2 gas molecules. This point can be incorrect and is against the experimental results. As shown in Figs. 3 and 4, the MgCO3 formation happens only in the local area under the e-beam irradiation, suggesting clearly that only adsorbed CO2 species are activated by the e-beam. In other words, the e-beam ionizes the CO2 molecules that have already adsorbed on the surface instead of CO2 molecules in the gas phase. Otherwise, MgCO3 formation would not be limited only to the e-beam irradiation area due to the fast and random motion of gas molecules. This effect can be similar to the electronbombardment effect on promoting the surface oxidation of Al(111), as shown by John Yates (PRL 89, 276101 (2002). In this sense, the authors should consider carefully the fundamental difference between e-beam and plasma in their experiments, because the latter may excite CO2 molecules in the gas phase, provided that the MgCO3 formation also happens in hidden areas (not directly exposed to the plasma).
2) Fig. 5 shows that the CO2 treated Mg pillar has improved mechanical properties compared to the untreated one. This is reasonable because the MgCO3 layer examined in their in-situ TEM experiment is very thin. The reviewer agrees with the authors' statement "there should be a critical thickness above which the MgCO3 scale would behave similarly to its brittle bulk counterpart". This also points to a question about the necessities of controlling the plasma-assisted carburization of the corroded product to avoid mechanical failure for thick MgCO3. The authors may add some comments.
3) There are some minor points that need to fix such as: i) the definition of the PB ratio is incorrect. It should be the ratio of the molar volume of the oxide with the molar volume of the metal.
ii) the statement "the waterproof carbonate can interfere with both anodic and cathodic reaction …" is vague and confusing. Does the surface layer chemically react or non-react with water? Improved clarity is needed.
4) there are some English errors that require careful proof read, such as the missing of "the" at many places.
5) this reviewer also went through the review comments and the authors' response from the first round of review. Most of the comments from reviewer 1 appear focusing on the practical aspect of the work reported in this manuscript, which I feel are minor issues in view of the fundamental implication of this study.

Response to the reviewer #1:
Comment 1: The revision is worse than the original, as the authors are oblivious to the shortcomings. Right from the title, the paper is focused on corrosion resistant coating.
As such, it can only be viewed in that context (and not viewed in the context of what the paper isn't).
This really is a very poor paper…… This paper would not survive pre-screening at a corrosion journal, and I cannot in good faith say it is anywhere near suited to Nature Comms. My recommendation is a firm reject.

Reply:
We feel sorry that we did not convince the reviewer in our previous reply. However, the comments from other reviewers clearly contradict the judgement from referee 1.
For example, they remarked that "The work is of high quality, especially the in situ microscopy. Sufficient details in the paper and the supplement are given for others to reproduce the data.", "this is an interesting paper that should influence the field of corrosion, oxidation, and nanoscience and technology, as the authors clearly demonstrate that different pretreatments especially reactive gas species can improve properties significantly", "These are innovative findings. The most impressive experiment in this study was the in-situ observation of the formation process of magnesium carbonate. Therefore, the reviewer would like to recommend publication of this paper", "These results are interesting, and the experimental evidence is convincing.", and "Most of the comments from reviewer 1 appear focusing on the practical aspect of the work reported in this manuscript, which I feel are minor issues in view of the fundamental implication of this study." In the following, we address the concerns from reviewer 1 in detail.

Comment 2:
Once again, the authors have avoided the might of the literature and opted to cite irrelevant textbooks in oxidation, etc.

Reply:
Based on the reviewer's comments from last round, we guess "the might of the In terms of corrosion resistance, the Mg-carbonate is not self-healing, it is likely to be soluble when wet, and it is not anywhere near as novel as the authors claim.

Reply:
We agree that Mg-carbonate is not self-healing. However, this does not preclude it as a useful surface protection. The MgCO 3 solubility in water at room temperature is as low as 0.02%. This explains why MgCO 3 can exist in nature for years. In addition, one proposed application of our finding is to manufacture bio-degradable Mg alloys with controllable corrosion rate. This means that even though the materials are not self-healing, they can be still very useful in certain circumstances if they are designed and fabricated in an appropriate manner.
Our work has significant novelty, which has been resonated by many experts in the field, including other reviewers. Firstly, our work is based on an in-depth understanding as to how MgO reacts with activated CO 2 . The applications and effects we showed met our design expectations. Even though previous work indicated that magnesium hydroxy carbonate or carbonate produced in humid or aqueous environment have certain anti-corrosion effect, the effects are usually weak, and the working principles are buried under complex conditions. Secondly, almost all Mg alloys in the atmospheric environment containing CO 2 are still facing serious corrosion problem, that's why so many researchers keep looking for different methods to improve their corrosion resistance. Our novelty mainly lies in understanding the nature of the reaction between the activated CO 2 and Mg oxide or hydroxide at room temperature and then apply it purposely. The environmental TEM technique, which is rarely used in the corrosion community, played a key role in inspiring our findings.

Comment 4:
All that figure 3B is showing is a galvanic couple, not corrosion protection.

Reply:
We cannot agree with the reviewer that Figure 3b is only a galvanic couple instead of the protection effect from MgCO 3 film. Firstly, the corrosion potential of the protected top part is indeed higher than the unprotected bottom part, which can be concluded from our potentiodynamic polarization tests results. So, these two parts with different corrosion potentials form a non-typical galvanic couple, which is usually developed when two different metals are separated by electrolytes. The increased corrosion potential of the top part with MgCO 3 film unambiguously vindicates our point that the produced MgCO 3 film has obvious protective effect.

Response to the reviewer #4:
Comment 1: The submitted manuscript introduces a detailed experimental study showing that the native surface oxide on Mg can be converted MgCO 3 via e-beam irradiation (or plasma) in a CO 2 atmosphere. With a set of corrosion tests, the authors further show that the resulting MgCO 3 surface layer has improved corrosion resistance compared to the native surface oxide. These results are interesting and the experimental evidence is convincing.

Reply:
We thank the reviewer for pointing out the significance of our work! And we are glad to hear that the reviewer considers our work to be convincing. We also highly appreciate the valuable suggestions given by the reviewer. In the following, we respond point-to-point to each comment.

Comment 2:
However, the following issues including the fundamental mechanism regarding the e-beam assisted MgCO 3 formation require careful clarification: The e-beam assisted MgCO 3 formation is attributed to the excitation of CO 2 gas molecules. This point can be incorrect and is against the experimental results. As shown in Figs. 3 and 4, the MgCO 3 formation happens only in the local area under the e-beam irradiation, suggesting clearly that only adsorbed CO 2 species are activated by the e-beam. In other words, the e-beam ionizes the CO 2 molecules that have already adsorbed on the surface instead of CO 2 molecules in the gas phase. Otherwise, MgCO 3 formation would not be limited only to the e-beam irradiation area due to the fast and random motion of gas molecules. This effect can be similar to the electron-bombardment effect on promoting the surface oxidation of Al (111), as shown by John Yates (PRL 89, 276101 (2002). In this sense, the authors should consider carefully the fundamental difference between e-beam and plasma in their experiments because the latter may excite CO 2 molecules in the gas phase, provided that the MgCO 3 formation also happens in hidden areas (not directly exposed to the plasma).

Reply:
We really appreciate the constructive comments and excellent advices from the reviewer. We fully agree that it is the adsorbed CO 2 species ionized by the electron beam that play the key role in forming the MgCO 3 layer. The Yates' paper (PRL 89, 276101, 2002) mentioned by the reviewer demonstrated that electron bombardments can enhance the oxidation rate of Al by inducing the negative electrostatic potential stored on the outer oxide film. In addition, the electrostatic field results in "memory effect" for prior electron irradiation. In other words, the enhanced oxidation continues even after e-beam bombardment is stopped. In our case, the carbonation discontinued as soon as the e-beam was blocked off, and it restarted immediately after the e-beam was turned on again (see below Figure S3), which suggests that the reaction of MgO with excited CO 2 species has no obvious memory for prior electron irradiation at all.
Therefore, the reaction of MgO with CO 2 at room temperature should be attributed mainly to the electron beam-stimulated excitation of adsorbed CO 2 species on the oxide surface. At the same, e-beam irradiation-induced defect sites or the electrostatic field effect within the oxide film mentioned in Yates's work, even if existed, should not play any significant role in our case.
Based on aforementioned consideration, we added the following discussions to compare the different effects of electron beam and plasma on the formation mechanism of MgCO 3 film: "It should be noted that the reaction mechanism of MgO with e-beam excited CO 2 is fundamentally different with CO 2 plasma. The glow discharge activated CO 2 molecules in the gas phase directly form the CO 2 plasma with ionic fragments and radicals of CO 2 * . Due to the fast motion of gas species, the formation of MgCO 3 occurs in all the areas that can be reached by the CO 2 plasma.
However, the situation is different for high energy e-beam irradiation. Compared to the CO 2 species adsorbed on the native oxide surface, the ionization rate of those free moving CO 2 gas molecules is much lower 23 . Therefore, MgCO 3 forms only in the local areas with absorbed CO 2 species and being exposed to the e-beam irradiation." Actions taken: 1> We have added a new figure in supporting information as Figure S3 to show that the reaction of MgO with excited CO 2 species has no obvious memory effect for the prior e-beam irradiation. with the e-beam off for 6 minutes, indicating that prior e-beam irradiation effect disappears as soon as e-beam is discontinued. The reaction restarted again as soon as the e-beam was turned on. (d) The morphology of products after exposure in 2 Pa CO 2 with the e-beam irradiation for additional 6 minutes.
2> We modified the right-side schematic in Figure 1 to demonstrate that the excited CO 2 * species mainly distribute on the oxide surface rather than suspend in the gas environment. Figure 1 in the revised manuscript 3> All statements related to the formation mechanism of the MgCO 3 film have been modified in the manuscript. Those revised parts are highlighted in blue on page 4, 5, 6, and 8, respectively. We also added discussions to compare the different effects of electron beam and plasma on the formation mechanism of MgCO 3 film (see page 10).
Comment 3: Fig. 5 shows that the CO 2 treated Mg pillar has improved mechanical properties compared to the untreated one. This is reasonable because the MgCO 3 layer examined in their in-situ TEM experiment is very thin. The reviewer agrees with the authors' statement "there should be a critical thickness above which the MgCO 3 scale would behave similarly to its brittle bulk counterpart". This also points to a question about the necessities of controlling the plasma-assisted carburization of the corroded product to avoid mechanical failure for thick MgCO 3 . The authors may add some comments. Reply: