Reaxff

Under the reaction force field, why do I use different units of real and metal to produce different phenomena? The speed scaling method controlled the Pt circular nanoclusters at 1000K. Under real, the Pt nanoclusters did not maintain their structure (FIG. 1) and seemed to melt, while the metal could maintain its normal configuration (FIG. 2). The results are as follows:
图1
图2

Yes, why?

Without sharing more details about your simulation script is hard to tell, but if you change the units command and not the force field parameters (not to mention the timestep, temperature, etc), then you are basically running two simulations with the same starting point, but with very different physics.

1 Like

Dear teacher, thank you for your reply. According to your requirements, I have attached my in file, but I still haven’t found any mistakes. I hope you can give me some suggestions.
in1.file (1.9 KB)
in2.file (1.9 KB)

There are two problems with what you are doing.

It is an error to use “metal” units for ReaxFF force fields, since all known parameter files are set up for “real” units.

It is very problematic to run charge equilibration the way you do across all atoms and thus all atom types while reading the parameters for the ReaxFF pair style, since you have no parameters for atom type 1 in your ReaxFF parameter file (and no mapping in the pair style). So you may have corrupted memory or may be accessing unitialized memory and thus your charge equilibration will be wrong.

In general, to use ReaxFF successfully, you will need to find a parameterization for all elements and be specific for the system and conditions that you are modeling. ReaxFF parameters are not very transferable. Using ReaxFF with a hybrid potential is asking for trouble and has to be done with very great care. e.g. you need to enter the charge equilibration parameters manually.