Multiscale Comparison Study of Void Closure Law and Mechanism in the Bimetal Roll-Bonding Process
Round 1
Reviewer 1 Report
The manuscript analyzes the void closure mechanism during roll-bonding processes.
I enjoyed reading this manuscript and the results are very interesting. However, I have a couple of recommendations for the authors.
First of all, the English needs a thorough revision.
Secondly, too many things are taken for granted and some more explanations or at least a few more words would definitely help the reader understand the content of the manuscript. As it is, it is only written for specialists.
Examples:
At some point, model A, B and C are referred to in the text without having been clearly defined previously.
Dislocation lines are clearly indicated in Fig.15, while it would make much more sense to do it for the first figure in which they appear (Fig.13).
Section 3.2 is not very clear: where is carbon in the FCC-Fe used to simulate Q235 carbon steel?
Page 11: point A, B, C, D are discussed. It would make no harm to specify that the authors are referring to the points in Fig. 14.
Author Response
Response to Reviewer 1 Comments
The authors wish to thank the reviewer for your helpful comments and suggestions. The paper has been carefully revised, and the main changes are highlighted in bule color. Below is an itemized response to review.
Point 1: First of all, the English needs a thorough revision.
Response 1: The grammar and vocabulary in the manuscript are carefully reconsidered by us, and a professional English editing service was used to check the manuscript.
Point 2: Secondly, too many things are taken for granted and some more explanations or at least a few more words would definitely help the reader understand the content of the manuscript. As it is, it is only written for specialists.
Response: We re-examined the manuscript and added necessary explanations. All changes are marked in blue font.
Examples:
At some point, model A, B and C are referred to in the text without having been clearly defined previously.
Response: Model B and C refer to model 2 and 3. The manuscript has been written incorrectly and has been unified.
Dislocation lines are clearly indicated in Fig.15, while it would make much more sense to do it for the first figure in which they appear (Fig.13).
Response: The meaning of dislocation line in Fig.13 is explained. (Page 10, Lines 278-279)
Section 3.2 is not very clear: where is carbon in the FCC-Fe used to simulate Q235 carbon steel?
Response: As the content of carbon in Q235 carbon steel is very small (mass fraction is 0.14% - 0.20%), carbon is neglected in the process of modelling and FCC-Fe was used to simulate Q235 carbon steel. We have added statement to explain it. (Page 5, Lines 158)
Page 11: point A, B, C, D are discussed. It would make no harm to specify that the authors are referring to the points in Fig. 14.
Response: Thank you for your suggestion and we have added and revised the sentence. (Page 10, Lines 301, Page 11, Lines 305, Page 12, Lines 334)
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors study the pore closure during roll bonding of two different steel sheets. To accomplish this, they use laboratory experiments as well as finite element (FE) and molecular dynamics (MD) simulations. However, they do not attempt to use the MD results as input in the FE model, nor are the experiments analyzed on the microscopic scale, such that terming the approach “multiscale” is somewhat daring.
The paper is well-organized, but the language is partly difficult to understand, and a proper proof reading is required. For example, it remains unclear whether the term “rough peak” is used in fact for the asperities on the rough surfaces. Temperatures are partly given in units of °C or K. It is suggested to conform to one unit system.
More importantly, the load case of the experiments and simulations is not sufficiently clear: While the authors speak of roll bonding, they seem to apply a static pressure, by what they call “indentors”. The boundary conditions of both, experiment and models, must be made clear and reproducible.
The authors fail to properly introduce the main properties they use. The term “void closure rate” sometimes seems to indicate the total fraction of closed voids rather than the according rate (Figures 9 and 11). However, in the text, the term is sometimes being described as a rate, which obviously goes to zero as all pores are being closed, whereas the total pore closure goes monotonously to 1.
In Figure 12, the authors refer to stresses and strains, which are not trivially defined in the given complex loading situation. A clear mathematical definition of those terms is required.
Concerning the numerical methods: Do the FE simulations use ideal plasticity? No hardening terms are described. Why is an image analysis software applied to assess the pore closure instead of using the contact information of the FE model?
Are the volume-energy curves shown in the MD part obtained for the high temperature of 1000K? In any case, such curves are only meaningful, if several crystal structures are compared, in this case at least a comparison of fcc and bcc Fe would be necessary. The atomistic model is very small (no quantitative dimensions are given however). It is known that in such small atomistic ensembles the interpretation of plastic processes is extremely difficult. Also in these simulations, the stacking faults are spanning the entire simulation box. Is this realistic? Concerning the results, it is unclear why pure Fe reveals a higher yield strength than FeCrNi. Why is there no solid solution strengthening? Is this an artefact of the used potential?
The conclusion that the pore closure rate on the micro-scale is small during the elastic deformation is extremely misleading. This would of course also be true for the macroscopic FE simulations. However, at the applied temperature, the strength of the material is so small that there is no significant elastic regime. In the MD simulations the strength is on the level of GPa at a temperature of 1000K, which is of course an artefact of the high loading rates and the small sample size. This should not be subject of any conclusion about the physics of the underlying problem.
Some smaller issues:
Figure and table captions are completely inappropriate. More information needs to be given here.
Some unclear statements or terms:
Lines 34, 39, 41: Passages are unclear
Line 133: “Triangular” should probably be “sinusoidal”?
Line 165: refer to Fig. 6 instead of Fig. 1
Figure 1: Which material is shown and which method has been used to characterize the surface topology? This is also not mentioned in the text.
Figure 8: no units in colorbar
Figure 9: no units
Author Response
Response to Reviewer 2 Comments
The authors wish to thank the reviewer for your helpful comments and suggestions. The paper has been carefully revised, and the main changes are highlighted in bule color. Below is an itemized response to review.
Point 1: The authors study the pore closure during roll bonding of two different steel sheets. To accomplish this, they use laboratory experiments as well as finite element (FE) and molecular dynamics (MD) simulations. However, they do not attempt to use the MD results as input in the FE model, nor are the experiments analyzed on the microscopic scale, such that terming the approach “multiscale” is somewhat daring.
Response 1: In this manuscript, we observed the closure process of real composite interface voids through experiments and studied the void closure behaviour at different scales by the finite element method and molecular dynamics method. In order to summarize the research topics of this paper as far as possible. We changed the name of the article from “multiscale study” to “multiscale comparative study”. (Page 1, Lines 2)
Point 2: The paper is well-organized, but the language is partly difficult to understand, and a proper proof reading is required. For example, it remains unclear whether the term “rough peak” is used in fact for the asperities on the rough surfaces. Temperatures are partly given in units of °C or K. It is suggested to conform to one unit system.
Response: We have carefully revised the expression of the article, proofread and corrected some vocabulary, and marked the revision with blue font.
“Rough peak” is the asperities on the rough surfaces and relevant sentences have been revised for a better understanding. (Page 1, Lines 36-38)
We unified the units in the manuscript, and replaced K with ℃.
Point 3: More importantly, the load case of the experiments and simulations is not sufficiently clear: While the authors speak of roll bonding, they seem to apply a static pressure, by what they call “indentors”. The boundary conditions of both, experiment and models, must be made clear and reproducible.
Response 3: As it is very difficult to carry out rolling experiments under laboratory conditions, we use cylindrical indenter to simulate rolls and carry out static pressure composite experiments. To keep the unit rolling force in the experimental process consistent with that in the actual production, the diameter of the indenter is determined to 70mm. We have supplemented the relevant explanations. (Page 1, Lines 83-84)
Point 4: The authors fail to properly introduce the main properties they use. The term “void closure rate” sometimes seems to indicate the total fraction of closed voids rather than the according rate (Figures 9 and 11). However, in the text, the term is sometimes being described as a rate, which obviously goes to zero as all pores are being closed, whereas the total pore closure goes monotonously to 1.
Response 4: we have add the definition of void closure rate in the manuscript. The void closure rate is defined as the ratio of void area to the total area of the interface, in which the slope of the curve represents the speed of void closure. We have supplemented the relevant explanations. (Page 7, Lines 220,Page 8, Lines 253)
Point 5: In Figure 12, the authors refer to stresses and strains, which are not trivially defined in the given complex loading situation. A clear mathematical definition of those terms is required.
Response 5: The stress is the sum of the stresses of all atoms in Z direction and the strain is the change rate of the size in Z direction. The accordingly sentence was modified. (Page 9, Lines 265-266)
Point 6: Concerning the numerical methods: Do the FE simulations use ideal plasticity? No hardening terms are described. Why is an image analysis software applied to assess the pore closure instead of using the contact information of the FE model?
Response 6: Yes, The specimens are assumed ideal elasticity-plasticity, which has been explained in this paper. (Page 4, Lines 133-134)
The contact information of finite element method can be used to calculate the length of contact line, but the Image analysis method more directly reflects the change of voids in the roll-bonding process.
Point 7: Are the volume-energy curves shown in the MD part obtained for the high temperature of 1000K? In any case, such curves are only meaningful, if several crystal structures are compared, in this case at least a comparison of fcc and bcc Fe would be necessary. The atomistic model is very small (no quantitative dimensions are given however). It is known that in such small atomistic ensembles the interpretation of plastic processes is extremely difficult. Also in these simulations, the stacking faults are spanning the entire simulation box. Is this realistic? Concerning the results, it is unclear why pure Fe reveals a higher yield strength than FeCrNi. Why is there no solid solution strengthening? Is this an artefact of the used potential?
Response 7: In order to verify the validity of the potential function, we calculated the energy of the system under different lattice constants at room temperature, and compared the lattice constants at the lowest energy with the experimental results in the literature to verify the consistency. We can calculate the volume energy curve at 1000K, but the lattice constant at 1000K is not clear, so we calculate it at room temperature.
Since the potential function [1] used in this manuscript is only applicable to FCC system, we only simulate the bonding process of Fcc-Fe and Fcc-FeCrNi.
The size of three models is 9.26 nm×5.35 nm×9.26 nm and has been added to the manuscript. (Page 6, Lines 182)
The essence of metal plastic deformation is the movement of dislocations. Dislocations slip along the glide plane during plastic deformation, leaving stacking faults behind. Therefore, with the movement of dislocations, it is possible for stacking faults to fill the whole simulation system. Similar results have been found in the relevant literature [2-4].
At the microscale, yield point means the first dislocation appear. Compared with pure Fe, the presence of alloying elements in FeCrNi makes the structure more inhomogeneous, and it is easier to produce regions with larger local stress to induce dislocation. Similar results have been obtained in some related studies. For example, literature [5] analyses the effect of C on the tensile properties of Fe-C alloys. It is found that dislocations first occur in the region where C atoms are located, which induces the decrease of system strength. Fe-C alloys have lower yield strength than Fe alloys.
The phenomenon of solid solution strengthening is that the existence of solid solution atoms pins dislocations and hinders the movement of dislocations, which makes the plastic deformation of metals difficult. Compared with yield strength, the phenomenon of solid solution strengthening has more influence on the yield strength ratio of materials.
[1] Bonny G.; Castin N.; Terentyev D. Interatomic potential for studying ageing under irradiation in stainless steels: the FeNiCr model alloy. Model. Simul. Mater. Sci. Eng. 2013, 21, 5897
[2] Weng S Y, Ning H M, Hu N, et al. Strengthening effects of twin interface in Cu/Ni multilayer thin films – A molecular dynamics study[J]. Mater. Des., 2016, 111: 1
[3] Shao S, Medyanik S N. Interaction of dislocations with incoherent interfaces in nanoscale FCC-BCC metallic bi-layers[J]. Model. & Simul. in Mater. Sci. Eng. 2010, 18: 055
[4] Wang J, Misra A. An overview of interface-dominated deformation mechanisms in metallic multilayers[J]. Curr. Opin. Solid State Mater. Sci. 2011, 15: 20
[5] Peng W, Gang T. Molecular dynamics simulation on the uniaxial tension property of metal Fe interacting with C [J]. Chinese Journal of Applied Mechanics,2015,32:915-921+1097.
Point 8: The conclusion that the pore closure rate on the micro-scale is small during the elastic deformation is extremely misleading. This would of course also be true for the macroscopic FE simulations. However, at the applied temperature, the strength of the material is so small that there is no significant elastic regime. In the MD simulations the strength is on the level of GPa at a temperature of 1000K, which is of course an artefact of the high loading rates and the small sample size. This should not be subject of any conclusion about the physics of the underlying problem
Response 8: Following your comments, relevant statements were deleted in the conclusion, and the first point in the conclusion was completely revised as:
(1) At the macroscale, the closure rate of voids decreases with the increase of the real contact area between interfaces. The shape of voids changes from rectangular to circular or elliptical, and finally disappears completely. At the microscale, the arrangement of atoms near the interface become disorder under certain pressure and move towards to the voids, finally become ordered again with the voids completely healed. (Page 13, Lines 362-366)
Point 9: Some smaller issues:
Figure and table captions are completely inappropriate. More information needs to be given here.
Some unclear statements or terms:
Lines 34, 39, 41: Passages are unclear
Line 133: “Triangular” should probably be “sinusoidal”?
Line 165: refer to Fig. 6 instead of Fig. 1
Figure 1: Which material is shown and which method has been used to characterize the surface topology? This is also not mentioned in the text.
Figure 8: no units in colorbar
Figure 9: no units
Response 9: Thank you for your helpful suggestion and we have revised these sentence.
Lines 34: The accordingly sentence“These primitive rough peaks form the voids between the interfaces and shrink continuously until they completely closed during the roll-bonding process.” was modified as“When the two materials stacked, only the asperities on the surface contact each other, thus forming a row of irregular voids, which gradually shrink until completely close during roll-bonding process.” (Page 1, Lines 36-38)
Lines 39-41: The accordingly sentence “Void closure occurs simultaneously at various scales in the roll-bonding process. At the macro scale, microscale rough peaks deform and close contact is produced in most areas of the interface. At the microscale, the nanoscale voids gradually shrink to achieve interfacial heteroatomic bonding.” was modified as “The whole roll-bonding process can be regarded as the closure process of interface voids at various scales.” (Page 1, Lines 42-43)
Line 133: “Triangular” was modified as “sinusoidal”. (Page 4, Lines 136)
Line 165: “Fig. 1” was modified as “Fig. 6”. (Page 6, Lines 169)
Figure 1: “A 3D figure of specimens’ surfaces is shown in Figure 1” was modified as “The 3D surface morphology of stainless steel specimen scanned by white light interferometer microscope is shown in Figure 1.” (Page 2, Lines 82)
Figure 8、9: Units are added to the figures.
Author Response File: Author Response.pdf
Reviewer 3 Report
Do all changes required in attached file.
Comments for author File: Comments.pdf
Author Response
Response to Reviewer 3 Comments
The authors wish to thank the reviewer for your helpful comments and suggestions. The paper has been carefully revised, and the main changes are highlighted in bule color. Below is an itemized response to review.
Point 1: Title is not clear…please consider to re-write it in a different way. Interfacial Void Closure Behavior…the meaning is not clear.
Response 1: In order to summarize the research topics of this paper as far as possible. We changed the title of the article from “Interfacial Void Closure Behavior” to “Void Closure law and mechanism”. (Page 1, Lines 2)
Point 2: State of the art: people making burnishing are very closed to the approach, and there is not any hint of it. Methods, results, figures are quite the same. Materials and Manufacturing Processes 26 (8), 997-1003 for instance showed some good ideas. Please introduce it.
Response 2: Thank you for the relevant literature you provided, which makes up for the deficiency of the article. We carefully read the relevant literature and quoted them in this article. (Page 2, Lines 81-82)
Point 3: Very nice molecular dynamics (MD) simulations, however more discussion must perform about the modeling of friction.
Response 3: The discussion of dislocation evolution during deformation is supplemented. (Page 10, Lines 308-309; Page 11, Lines 318-320; Page 12, Lines 341-344)
Point 4: In all works fixtures and machine can deform, under the high process forces. Please consider it The International Journal of Advanced Manufacturing Technology 37 (7-8), 649-656 introduced a good line of models.
Response 4: Following your comments, we have supplemented the description of the process of indenter processing and added the relevant literature. (Page 2, Lines 89-92)
Point 5: Ref 29: the titles in some references are missed.
Response 5: Thank you for your suggestion and we have added and revised the sentence. (Page 14, Lines 456)
Round 2
Reviewer 2 Report
The authors addressed many points raised by the reviewer, in particular, the language is now at least understandable. However, there are still some open points concerning the atomistic simulations:
The atomic stress and strain is still not properly introduced in the text. In lines 265, the global stress is defined as the sum of the atomic stresses, but the latter is not defined. Why is the strain the "change rate"? This would yield a strain rate rather than a strain.
Concerning the occurrence of stacking faults, the authors compare their work to work, where explicitly nano-structured material has been investigated (refs [2-4] in their answer). In this case, in fact, many authors described that nano-structured material deforms by motion of partial dislocations, leaving behind stacking faults. However, the authors claim that their results are valid for bulk material, in which full dislocations will dominate the plastic deformation. This needs to be discussed in much more depths.
Also the explanations concerning the yield point are not yet satisfactory. Again, the authors need to explain in much more detail, why they consider the stress at which dislocations are nucleated to be the yield stress of the material. This definition is commonly used only for nano-structured materials, in which no free dislocations are present and need to be formed first. In such cases increasing the disorder of a system will support the dislocation nucleation and make the nano-structured material softer. However, in bulk metals, there are always sufficiently many free dislocations such that the yield strength is the stress at which these dislocations start to move. This bulk yield strength should always increase with the disorder of the system.
Points 2 and 3 clearly indicate that the atomic samples used by the authors are too small and that the findings cannot be transferred simply to cases where bulk deformation mechanisms are relevant. This is a serious flaw in their work and the authors should carefully address these concerns.
Author Response
Response to Reviewer 2 Comments
The authors wish to thank the reviewer for your helpful comments and suggestions. The paper has been carefully revised, and the main changes are highlighted in bule color. Below is an itemized response to review.
The authors addressed many points raised by the reviewer, in particular, the language is now at least understandable. However, there are still some open points concerning the atomistic simulations:
Point 1: The atomic stress and strain is still not properly introduced in the text. In lines 265, the global stress is defined as the sum of the atomic stresses, but the latter is not defined. Why is the strain the "change rate"? This would yield a strain rate rather than a strain.
Response 1: Thank you for your reminder and our previous statement was incorrect. The strain is the ratio of total deformation to the initial dimension of the material in Z direction. We have revised the sentence in manuscript. (Page 9, Lines 271-273)
Point 2: Concerning the occurrence of stacking faults, the authors compare their work to work, where explicitly nano-structured material has been investigated (refs [2-4] in their answer). In this case, in fact, many authors described that nano-structured material deforms by motion of partial dislocations, leaving behind stacking faults. However, the authors claim that their results are valid for bulk material, in which full dislocations will dominate the plastic deformation. This needs to be discussed in much more depths.
Response 2: Thank you for your patient comments. In the "Introduction" section of the manuscript, we introduce the background of molecular dynamics simulation.
“Modern roughness measurements show that the surface profile of many engineering surfaces has a multiscale fractal nature; that is, when the surface is properly amplified, the enlarged morphology is similar to the original morphology. Therefore, there are nano-scale rough peaks exist on the micron-scale rough peaks based on this assumption.” (Page 1, Lines 40-43)
Based on the fractal theory, we assume that there are countless rough peaks with different scales on macroscale rough peaks. Not only the rough peaks of the slab are nano-scale, but also the slab of Fe or FeCrNi in molecular dynamics simulation are nano-scale. We have revised the relevant statements and vocabulary of the manuscript. (Page 1, Lines 12-14; Page 6, Lines 182-184)
Point 3: Also the explanations concerning the yield point are not yet satisfactory. Again, the authors need to explain in much more detail, why they consider the stress at which dislocations are nucleated to be the yield stress of the material. This definition is commonly used only for nano-structured materials, in which no free dislocations are present and need to be formed first. In such cases increasing the disorder of a system will support the dislocation nucleation and make the nano-structured material softer. However, in bulk metals, there are always sufficiently many free dislocations such that the yield strength is the stress at which these dislocations start to move. This bulk yield strength should always increase with the disorder of the system.
Points 2 and 3 clearly indicate that the atomic samples used by the authors are too small and that the findings cannot be transferred simply to cases where bulk deformation mechanisms are relevant. This is a serious flaw in their work and the authors should carefully address these concerns.
Response 3: We deeply approve of you opinion. The unique properties of nano-materials are different from those of macro-materials in the process of deformation. Such as the partial dislocations are more likely to occur in nanostructured metals during deformation.
As mentioned in the previous response, "bluk" in the manuscript is also nanoscale. Of course, we neglected many factors in the simulation process, such as dislocations exist in bulk metals, which will certainly affect the deformation of rough peaks on the micro-scale. In order to make it clearer, we supplement the shortcomings of the research in this manuscript.
“However, there are still some shortcomings in the MD model, such as ignoring the initial dislocations that may exist in the micro-rough peaks, and the influence of matrix deformation on the deformation behavior of micro-rough peaks is not considered. Further research should be done in the future to solve these problem.” (Page 13, Lines 384-387)
In addition, we tried to modify the non-standard language of the article as much as possible. The discussion of dislocation evolution during deformation is supplemented. (Page 10, Lines 308-309; Page 11, Lines 318-320; Page 12, Lines 341-344)
Thank you again for your patient comments and suggestion on this manuscript.
Author Response File: Author Response.pdf
Reviewer 3 Report
Paper is OK
Author Response
The authors wish to thank the reviewer for your helpful comments and suggestions.
Round 3
Reviewer 2 Report
The authors addressed all concerns satisfactorily. The only open point is the definition of the atomic stresses, which was already requested in the previous review: "In line 265, the global stress is defined as the sum of the atomic stresses, but the latter is not defined." (now line 270).
Author Response
Response to Reviewer 2 Comments
The authors wish to thank the reviewer for your helpful comments and suggestions. The paper has been carefully revised, and the main changes are highlighted in bule color. Below is an itemized response to review.
Point 1: The authors addressed all concerns satisfactorily. The only open point is the definition of the atomic stresses, which was already requested in the previous review: "In line 265, the global stress is defined as the sum of the atomic stresses, but the latter is not defined." (now line 270).
Response 1: Thank you for your patient comments. The stress is the sum of the all atoms stresses in Z direction, in which atomic stress equals the potential density in the resultant force field generated by the neighboring atoms. (Page 9, Lines 270)
