Development of a Process to Recycle NdFeB Permanent Magnets Based on the CaO-Al₂O₃-Nd₂O₃ Slag System

Round 1

Reviewer 1 Report

This study identifies the pyro- and hydrometallurgy process of NdFeB permanent magnet recycling and is worth publishing. However, there are some deficiencies in the considerations listed below, which need to be corrected and confirmed. Major revisions are suggested.

1) [Line 102, 2. Materials and Methods] In 1. Introduction section, the authors described that the production of REE-oxides from magnets using SiO2-based flux. However, the CaO-Al2O3-REE2O3 system was chosen as a collector phase. What is the reason that Al2O3-based flux?  

2) [Line 143, 2. Materials and Methods] Why the authors use the corundum crucible in the CaO-Al2O3-REE2O3? In high temperature, the corundum crucible dissolves into CaO-based slags. The final slag composition is changed by the solubility of Al2O3 in corundum crucible. Please check the following reference. 

(Huo, Y. Z., Gu, H. Z., Huang, A., Ma, B. Y., Chen, L. G., Li, G. Q., & Li, Y. W. (2022). Characterization and mechanism of dissolution behavior of Al2O3/MgO oxides in molten slags. Journal of Iron and Steel Research International, 29(11), 1711-1722. )

(Liu, J., Liu, Z., Feng, J., Li, B., Chen, J., Ren, B., ... & Yin, S. (2022). Reaction Mechanism of CA6, Al2O3 and CA6-Al2O3 Refractories with Refining Slag. Materials, 15(19), 6779)

3) [Line 149, 2. Materials and Methods] Why the authors add the P2O5 in the CaO-Al2O3-REE2O3? It is better to show the phase diagram of CaO-Al2O3-P2O5 for readers. 

4) [Figure.1, 2. Materials and Methods] The authors represented the flowchart of the proposed two-step process (pyrometallurgy using CaO-Al2O3 flux and hydrometallurgy using H2SO4 or MSA. In oxidative smelting and reductive smelting processes, it is better to show the partial pressure of oxygen for readers. What is critical pO2 for reduction of Nd2O3? In reductive smelting process, did the Nd concentrate to metal phase? (Nd2O3 + 3C -> 2Nd + 3CO) 

5) [Table 2. 2. Materials and Methods] The type of reactions are carbothermic reactions. Did the recycling process of NdFeB permanent magnets consider the CO2 emssions?

6) [Figure 3. 3.1 Modelling results] The authors represented the CaO-Al2O3-REE2O3 phase diagram. Isn’t necessary CaO-Al2O3-FeOx-REE2O3 phase diagram for oxidative smelting process? 

7) [Line 328. 3.3 Hydrometallurgical process] The authors used the HCl, HNO3, MSA for precipitation of Ca in hydrometallurgical process. What is full name of MSA? Please describe the MSA for readers.

8) [Figure 11 and 12. 3.3 Hydrometallurgical process] It is better to use higher resolution figure for readers. 

 

Author Response

Dear Reviewer,

Thank you for your review and the comments. We changed the manuscript according to your suggestions. Below you can find an explanation and change overview for each comment. Especially the suggested additional phase diagrams would of course be very nice to see, but they have to be the result of following research, further citations where given where applicable.

1) [Line 102, 2. Materials and Methods] In 1. Introduction section, the authors described that the production of REE-oxides from magnets using SiO2-based flux. However, the CaO-Al2O3-REE2O3 system was chosen as a collector phase. What is the reason that Al2O3-based flux?  

Most authors using SiO2 focus on only the pyrometallurgical step. As stated now without brackets SiO2 is detrimental due to the formation of silica gel in the chosen hydrometallurgical step, which increases process complexity and could lead to REE losses. Of course adding SiO2 to the list of potential fluxes can be subject to further research in a quaternary system.

-> Change: Clarified why SiO2 is not used and that we purposefully diverge from what other authors did previously.

2) [Line 143, 2. Materials and Methods] Why the authors use the corundum crucible in the CaO-Al2O3-REE2O3? In high temperature, the corundum crucible dissolves into CaO-based slags. The final slag composition is changed by the solubility of Al2O3 in corundum crucible. Please check the following reference. 

A slight dilution of Al2O3 from the crucible is happening. As visible from the results in table 3 Al2O3 is slightly higher for Setup 1 and 3, while nearly the same amount of Al2O3 and CaO powder where supplied initially. This is not the case for the Setup 2 slag. I would argue that the solution of Al2O3 is so slow, that in the oxidative first process step this effect is below our measurement accuracy. For the second reductive step graphite crucibles are chosen.

Change: We mentioned that corundum crucibles are chosen over graphite or SiC crucibles, because we work in an oxidative state. Other ceramics would introduce new elements (e.g. MgO) that would further increase the complexity of the system. Al2O3 solution is an effect to be aware of.

(Huo, Y. Z., Gu, H. Z., Huang, A., Ma, B. Y., Chen, L. G., Li, G. Q., & Li, Y. W. (2022). Characterization and mechanism of dissolution behavior of Al2O3/MgO oxides in molten slags. Journal of Iron and Steel Research International, 29(11), 1711-1722. )

(Liu, J., Liu, Z., Feng, J., Li, B., Chen, J., Ren, B., ... & Yin, S. (2022). Reaction Mechanism of CA6, Al2O3 and CA6-Al2O3 Refractories with Refining Slag. Materials, 15(19), 6779)

3) [Line 149, 2. Materials and Methods] Why the authors add the P2O5 in the CaO-Al2O3-REE2O3? It is better to show the phase diagram of CaO-Al2O3-P2O5 for readers. 

As stated, the idea behind this step was to possibly create a monazite like REE rich mineral, (REE phosphate) since this did not happen in our system, this slag was not further processed.

The phase diagram of the CaO-Al2O3-P2O5 does not present interest for the present work, because it does not show which phases can concentrate Nd2O3. Besides this, P2O5 was used only in one mixture of oxidative smelting. Thermodynamic description for the system CaO-Al2O3-P2O5 is not available and experimental data are scarce.

However, from a scientific perspective the creation of such a phase diagram would be interesting for future work.

4) [Figure.1, 2. Materials and Methods] The authors represented the flowchart of the proposed two-step process (pyrometallurgy using CaO-Al2O3 flux and hydrometallurgy using H2SO4 or MSA. In oxidative smelting and reductive smelting processes, it is better to show the partial pressure of oxygen for readers. What is critical pO2 for reduction of Nd2O3? In reductive smelting process, did the Nd concentrate to metal phase? (Nd2O3 + 3C -> 2Nd + 3CO) 

Changes: Partial pressures for O2 are defined in the graphs (for the oxidative step we blew in pressured air, p(O2) is around 0.21 atm; for the reductive step we started carbothermically reducing Al2O3 meaning, that at 1700 °C we are below 10^-12 atm p(O2), this is included in the flowchart now.

The critical p(O2) for Nd2O3 reduction is far lower, but since our goal is to concentrate Nd as Nd2O3 as intermediate for molten salt electrolysis, this is not relevant for the featured process. The Nd left the metal phase (REPM alloy) towards the oxidic slag to up to 99.9 % (in the best case, as stated in results, discussion and conclusion) and is never reduced carbothermically. Even if the reaction Nd2O3 + 3C -> 2Nd + 3CO would have taken place, the resulting Nd would have been oxidized by Al2O3 and metal Al would have been formed.

5) [Table 2. 2. Materials and Methods] The type of reactions are carbothermic reactions. Did the recycling process of NdFeB permanent magnets consider the CO2 emssions?

CO2 emissions where considered, but not directly measured. At those high temperatures most C, that was lost (mass weight balance) will have been lost as CO. We believe that the carbothermic reduction is just the first step in understanding the system and proposing one process.

An industrial process would have many benefits if done metallothermically. The possibility of metallothermic reduction is proven in initial scouting experiments. This is stated in methods briefly. Since this process is available and featured in the paper, we do not dive deeper in CO/CO2 emission numbers.

Changes: We clarify in the discussion and conclusion part, that we see the future in a green metallothermic process and not in the rather old fashioned carbothermic process since no CO2 is emitted here.

6) [Figure 3. 3.1 Modelling results] The authors represented the CaO-Al2O3-REE2O3 phase diagram. Isn’t necessary CaO-Al2O3-FeOx-REE2O3 phase diagram for oxidative smelting process? 

The presenting phase diagrams for the CaO-FeO-Al2O3-Nd2O3 system seems to be more reasonable, but only very limited data on phase relations are available for this system at 1300°C [1979Jan]. Therefore, experimental investigations of constituent sub-systems and quaternary system as well as optimization of thermodynamic parameters are necessary. Such kind of investigation can be the subject of several future papers.

[1979Jan] C.M. Jantzen, F.P. Glasser Solid-state reactions in the system CaO-Nd2O3-Fe2O3-Al2O3, Mat Res. Bul. 14 (1979) 1601-1607.

The modelling results are viable for systems where the Fe-Nd separation happened nearly perfectly. Since in our best case setups, only very low amounts of Fe (0,4 %) are within the slag, the CaO-Al2O3-Nd2O3 phase diagram accurately represents our system.

Changes: Cited Source for a CaO-Nd2O3-Fe2O3-Al2O3 phase system and mentioned, that FeO is not relevant for a process where Fe and Nd2O3 are separated in two different phases in the methods chapter.

7) [Line 328. 3.3 Hydrometallurgical process] The authors used the HCl, HNO3, MSA for precipitation of Ca in hydrometallurgical process. What is full name of MSA? Please describe the MSA for readers.

MSA is methane sulfonic acid (introduced in the methods chapter)

Changes: Abbreviation is introduced in the abstract already

8) [Figure 11 and 12. 3.3 Hydrometallurgical process] It is better to use higher resolution figure for readers. 

Figures are exchanged with higher resolution pictures; they will be supplied as .png/.tiff as well.

We hope the changes we applied further increase the quality of the manuscript. If there are further questions/suggestion please feel free to contact us:
[email protected]

Kind regards

Ludwig Blenau

Reviewer 2 Report

General comments: The manuscript was well-written and very interesting.

 

Specific comments

 

Line 40: As the usual practice, it is important to write PRC and EU in full first before using these acronyms.

 

Line 64 to 82: The story flow in this paragraph is incoherent. There should be proper link from one study by the particular researcher to another.

 

Line 95: I guess the authors meant THAN as opposed to THEN.

 

Line 106: The authors have tried to justify the use of CaO and Al2O3 to avoid the formation of silica gel without mentioning that CaO when leached in acid will lead to high consumption of acid so it is Al2O3 in basic solution.

 

 All figures in this manuscript are of very low resolution. I think the author relied on screenshotting. I suggest they put high resolution figures.

 

The captions for Figure 1, 3, and 6 are too long. Some information should be taken to the main body. 

 

 

Author Response

Dear Reviewer,

Thank you for your review and the comments. We changed the manuscript according to your suggestions. Below you can find an explanation and change overview for each comment.

 

Line 40: As the usual practice, it is important to write PRC and EU in full first before using these acronyms.

Changes: Changed according to the suggestion

 

Line 64 to 82: The story flow in this paragraph is incoherent. There should be proper link from one study by the particular researcher to another.

This paragraph is meant to give a very brief overview of different pyrometallurgical REPM recycling approaches. Since they did not necessarily build up on each other, it there is no factual link between them. We sorted them by year of publication.

Change: Clarified how we sorted

 

Line 95: I guess the authors meant THAN as opposed to THEN.

We did, and modified accordingly

 

Line 106: The authors have tried to justify the use of CaO and Al2O3 to avoid the formation of silica gel without mentioning that CaO when leached in acid will lead to high consumption of acid so it is Al2O3 in basic solution.

 We introduced a new sentence at line 110 clarifying the problem. There is no link between preventing SiO2 silica formation and consumption of acid by CaO or Al2O3. SiO2 and any other flux would also consume acid. That’s why we need high RE₂O₃ slag ratio. This is discussed in the paper.

 

 All figures in this manuscript are of very low resolution. I think the author relied on screenshotting. I suggest they put high resolution figures.

 Figures 7, 9, 10, 11 and 12 are low resolution, they are now redone. All figures are increased in size.

Figure 7 is reworked, Figure 8 is only unsharp for the scaling below the SEM-images due to an older SEM-software and con not be increased in resolution to much. The other pictures resolution is according to standards, if they still appear low res, due to the MS word compression, for the final publication we will include a folder with all figures in .png/.tiff format.

 

The captions for Figure 1, 3, and 6 are too long. Some information should be taken to the main body. 

This is true, the text also reads more like main body, we modified accordingly

 

We hope the changes we applied further increase the quality of the manuscript. If there are further questions/suggestion please feel free to contact us:
[email protected]

Kind regards

Ludwig Blenau

Reviewer 3 Report

This is good topic to recycle NdFeB permanent magnets scrap materials by this proposed new process, i.e. production of pure REE-Oxides from the REPMs, which in turn can be used to produce new REPMs. The authors developed the two-step pyrometallurgical process separating Fe and Nd from REPM and leading to a homogeneously oxidized slag. After second step carbothermic reduction an iron lean slag concentrating about 10% RE oxides was successfully generated. After the steps, leaching by NaOH or NH4OH and by H2SO4 were all experimentally studied.

The topic is important to metallurgists and recycle researchers. The quality of the conducted study is very high. The research background is solid. A literary review has been made quite fully. The results of this manuscript are well presented and organized. The presented results of the experiments have scientific meaning.

Overall, the authors have done a very high level research paper with all aspects perfect.

I recommend to accept this manuscript after the minor revisions.

1.       The CaO-Al2O3-RE2O3 system was chosen to avoid SiO2 which is detrimental due to the formation of silica gel in the following hydrometallurgical step. This is good choice. In oxidative smelting, P2O5 was used, will it affect the phase diagram and oxidation ability of slags? This may be illustrated.

2.       Were the slags in leaching experiment come from the pyrometallurgical process? Or as I learned they are made by pure oxides. The metal may enriched in the slag in the process. That’s may be different from pure oxide mixtures. This may be addressed.

3.       The leaching of CaO-Al2O3-RE2O3 slag system by NaOH or NH4OH and by H2SO4 were all experimentally studied. However, the leaching results are not deep. The process steps are put forward but may be concepts. It requires future work as the authors stated. This is not mandatory for revisions.

4.       In table 3, the compositions of setup 1 to 3 could be slag compositions? It’s better to use mass percentage of oxides CaO-Al2O3-RE2O3-P2O5 etc. That’s easy to understand for readers. This also applies to table 4.

5.       The descriptions in the figure tittles for example figure 2 to 6 and figure 8 should be moved to the manuscript. Leave figure tittle or name along like figure 9 and 10 did.

6.       Besides, figure 7, 9, 10, 11 are in a low quality of resolution. Please revise.

7.       Show the full name of abbreviations when it appears first time, many abbreviations were not given although it was common sense. MSA in abstract, CALPHAD

8.       Typing error, no.3 in conclusions, ammounts.

9.       Some references of RE oxides may be referred for example a systematic review of properties and application of RE oxides https://doi.org/10.3390/cryst13020202

Author Response

Dear Reviewer,

Thank you for your review and the comments. We changed the manuscript according to your suggestions. Below you can find an explanation and change overview for each comment.

1. The CaO-Al2O3-RE2O3 system was chosen to avoid SiO2 which is detrimental due to the formation of silica gel in the following hydrometallurgical step. This is good choice. In oxidative smelting, P2O5 was used, will it affect the phase diagram and oxidation ability of slags? This may be illustrated.

As stated, the idea behind this step was to possibly create a monazite like REE rich mineral, (REE phosphate) since this did not happen in our system, this slag was not further processed.

The phase diagram of the CaO-Al2O3-P2O5 does not present interest for the present work, because it does not show which phases can concentrate Nd2O3. Besides this, P2O5 was used only in one mixture of oxidative smelting. Thermodynamic description for the system CaO-Al2O3-P2O5 is not available and experimental data are scarce.

However, from a scientific perspective the creation of such a phase diagram would be interesting for future work.

P2O5 does not hinder the oxidation of the slag.

2. Were the slags in leaching experiment come from the pyrometallurgical process? Or as I learned they are made by pure oxides. The metal may enriched in the slag in the process. That’s may be different from pure oxide mixtures. This may be addressed.

Change: line 514: “Real slags may accumulate impurities such as dissolved metal phase or ferrous oxide, which were not investigated in this research.”

3. The leaching of CaO-Al2O3-RE2O3 slag system by NaOH or NH4OH and by H2SO4 were all experimentally studied. However, the leaching results are not deep. The process steps are put forward but may be concepts. It requires future work as the authors stated. This is not mandatory for revisions.

We agree. More research is needed. It was just a demonstration, but this is already mentioned in the text.

4. In table 3, the compositions of setup 1 to 3 could be slag compositions? It’s better to use mass percentage of oxides CaO-Al2O3-RE2O3-P2O5 etc. That’s easy to understand for readers. This also applies to table 4.

We agree, that giving the slag components as oxides would perhaps be easier to understand, we chose not to do so because i) REPM (which is of course metallic and not oxidic) would need another table and could not be compared directly and ii) the ICP does not measure oxygen, mathematically adding oxygen would be an interpretation already not a result, this is especially true for Fe where we assume Fe2O3 after the oxidative step, but have no clear measurement pointing at the Fe2O3 to FeO ratio.

Changes: Commented on the metallic/oxidic state of the materials features in table 3 and 4.

5. The descriptions in the figure tittles for example figure 2 to 6 and figure 8 should be moved to the manuscript. Leave figure tittle or name along like figure 9 and 10 did.

Is changed for 3, 6 and 8; 2,4 and 5 are in our understanding brief and relevant enough for the figure to be kept in place.

6. Besides, figure 7, 9, 10, 11 are in a low quality of resolution. Please revise.

Changes: Improved 7, 9, 10 and 11, we are also going to supply higher resolution .png/.tiff with the final version of this manuscript.

7. Show the full name of abbreviations when it appears first time, many abbreviations were not given although it was common sense. MSA in abstract, CALPHAD

Is added for both cases according to suggestion

8. Typing error, no.3 in conclusions, amounts.

Is changed according to suggestion.

9. Some references of RE oxides may be referred for example a systematic review of properties and application of RE oxides https://doi.org/10.3390/cryst13020202

The given doi is for a paper on removing Ce2O3 in slags for improved steelmaking. While this is very interesting, Ce plays no role in REPM recycling. We added a reference for Nd2O3-CaO-Al2O3-FeO systems, which we believe might be more relevant.

We hope the changes we applied further increase the quality of the manuscript. If there are further questions/suggestion please feel free to contact us:
[email protected]

Kind regards

Ludwig Blenau