Fe atom ejected when press slab

Hello,everyone, this is Rufer. I am modelling lubrication of hexane and Stearic acid confined in iron oxide slab as article( Nonequilibrium Molecular Dynamics Simulations of Organic Friction Modifiers Adsorbed on Iron Oxide Surfaces) mentioned. I use Berro’s model to simulate iron oxide: the Fe and O atoms are attached together by harmonic spring as mentioned below:
The hematite slab atoms were restrained in the corundum crystal by harmonic bonds between atoms within 3 Å. The force constant of these bonds was chosen to be 130 kcal mol–1 Å–2, which has been shown previously to keep the surface structure suitably rigid but not to adversely affect the thermostatting
In my model,I repeat this setting,but when I give a pressure on the fixed layer of upper slab,the Fe atoms of fixed layer will be ejected.they are not move simultaneously as O atoms.here are vibrations of system snapshot when applying pressure on the upper slab.

that’s my file below,any advice will be appreciated.
best wishes.
my file

I don’t think there is going to be much help that you can expect with this.
There is no indication that this is happening due to an error in LAMMPS and there is no simple telling about issues with your setup, since your model is far too complex to be able to tell from just looking at it.

So you will have to go about debugging it yourself since you have the most interest in solving this.
Obviously, it is not a setup that is easy to debug as is. Rather you will need to validate each part individually to behave as desired when modeling a small subset of it. After resolving all potential issues with that, you could go about and gradually assemble the more complex system from the pieces and thus make certain that the interactions between the subsets work as expected. This is all common sense and should be standard simulation procedure for modeling complex systems.

I will try to validate subset of my system as you said,thanks for your advice.