The Anisotropic Chemical Reaction Mechanism of 1,3,3-trinitroazetidine (TNAZ) under Different Shock Wave Directions by ReaxFF Reactive Molecular Dynamics Simulations
Round 1
Reviewer 1 Report
The authors conduct MSST simulations of TNAZ along different orientations to assess chemical reactivity differences, showing mechanical/crystallographic influence on kinetics that also directly alters reaction pathways. While this work in performed properly and the conclusions drawn from the data are valid. Significant work and a major revision of the manuscript is needed. Additional analysis of the data and potentially additional MD runs would considerably enhance this article.
A few minor comments:
- Is the pressure in Figure 3 the hydrostatic pressure (the trace of the tensor) or the pressure tensor component in the shock direction. Please make this more clear.
- Table 1, while useful information, could probalby be placed in supplemental information
-Table 2 provides not additional information beyond what is already shown in Figure 3 and should either be deleted or placed in supplemental.
Major issues:
-Even with the 0.05 fs timestep used here, ReaxFF suffers from horrible total energy conservation at high temperatures and pressures. The so defined "slow reaction" zone is quite likely heating the system due to energy drift, not exothermic reaction. Please provide plots of system total energy in the supplemental (or even main manuscript) to confirm or deny that the total energy is not increasing in time.
- Why is the 8 km/s run stopped at 150 ps? This seems extraordinarily arbitrary, as they appear to be entering the fast reaction stage from Figure 4, and should be fully reacted in under 50ps of additional run. It seems an odd decision to not extend these runs briefly and be able to add additional data points to the following sections of the paper
- These system cells, especially in the [001] is extremely small. The authors cite a mechanical influence on the molecular packing as the primary reason for anisotropic reactivity. However, [100] case is only a few planes of molecules wide, and the shock direction in [001] is only a few planes long. This calls into question what system size effects can be controlling this anisotropy at this level. These cells are too small for the nucleation of dislocations or shear bands to resolve stresses that deform the molecule and the small number of molecules provides very little statistical variance in squeezing of the Y shape. Using larger system sizes or running ensembles of the same shock with different starting configurations would go a long way to confirming the chemical influence of this molecular strain mechanism.
- The definition of clusters in this work is somewhat weak (anything bigger than the original molecule). Additionally, the maximum cluster size in these runs are roughly 80% of all atoms in the cell. Typically, clustering in shocks and detonation soots have a minimum size on the order of 10s-100s of nanometers, well outside the range of the cells used here. Additionally, I wonder how sensitive these 'clusters' are to the reaxff bond order parameters. At extremely high pressures, 60 GPa in the 9 km/s case, how much of these clustering reactions are thermodynamically favorable, and how much of it is just due to compressing the atoms to extreme densities. (In TATB detonation, a lot of the clustering reactions occur well after the sonic point).
- My main issue with this work is the defined reactions in Tables 6-8. These seems to be mostly nonsense. Firstly, i'm assuming the 'reaction times' listed here is the span of times in which these reactions occur, not the time it takes to occur or lifetimes. In most of these cases, a reaction and its reverse reaction happen almost always the same amount of times. What is the criteria for a reaction to occur? Even with ReaxFF bond orders instead of distance cutoffs, there are a large number of bonds that reax will consider 'broken' only to have it reform a few steps later. Considering that the reaction times for the forward and reverse reactions are very similar, I would hazard a guess that most of these are just reaxff saying the bond is gone, and then the same bond reforms. This is not a reaction. A would suggest re-performing the reaction analysis with a significant lifetime criteria. The bond breaks and it must stay broken for XX ps. A sensitivity test of this lifetime will probably be needed.
This will greatly cut down on the number of 'reactions' that occur, which will also help with my other issue of tables 6-9: they are massive and most of the data is not used in forming conclusions. Only showing reactions with significant difference from orientation to orientation will help make the manuscript more readable, and the rest of the reactions that still occur after using a lifetime criteria can go in supplemental. How these reactions are found and counted should also be explained in detail somewhere in the manuscript.
Author Response
Dear reviewer：
Thanks for your comments and advice！We have responded to your questions and suggestions and revised the manuscript.Please see the attachment.
Sincerely，
Wu
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This work is well organized and well-written, providing a solid investigation of the anisotropic reaction mechanism of TNAZ. Here are some comments for the authors to address:
1. From Fig. 3, Fig. 4, and Table 2, it seems that TNAZ only has an anisotropic reaction at 9 and 10 km/s, but it is not obviously anisotropic at 8 and 11 km/s. Can the authors provide additional discussion on these differences at different velocities?
2. Line 212, "there are 5, 2, 1 and ...". What does "5, 2, 1" mean?
3. In section 2 Methods, can the authors explain how they selected the shock conditions?
4. After investigating the anisotropic reaction mechanism of TNAZ, can the authors make comments on how to apply this anisotropic mechanism in the applications of TNAZ?
5. The authors should check the references. There are some format mistakes in the reference list.
Author Response
Dear reviewer：
Thanks for your comments and advice！We have responded to your questions and suggestions and revised the manuscript.Please see the attachment.
Sincerely
Wu
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have mainly fixed issues from my previous concerns in this manuscript, but I am more confused than before now seeing the total energy curves from the MSST runs. MSST thermodynamic output of total energy sometimes includes additional MSST terms. LAMMPs used to have a fix modify to exclude this, it now handles this via 'ecouple' and 'econserve' sections of the total energy, which I have not had the time to assess the differences.
However, independent of this, at late times during the 'slow reaction' regime, assuming that the density is not changing much, these systems should be approximately adiabatic. (MSST does not employ a thermostat so where does the energy go?) So these large changes in total energy are either due to a large volume shift or a total energy drift from ReaxFF, but continued reaction should not cause such as massive total energy change.
This is mostly a minor detail, as the results are what they are and correct for the methods chosen, I am just somewhat puzzled by these total energy plots and the overall thermodynamics at late times.
Author Response
Dear reviewer：
Thanks for your comments and advice！We have responded to your questions.Please see the attachment.
Sincerely，
Wu
Author Response File: Author Response.pdf
Reviewer 2 Report
It looks good to me.
Author Response
Thank you sincerely.
