I am trying to simulate diffusion of light gas molecules in a solid porous silicalite structure. The structure is rigid and not supposed to move.
I ran MD for a couple of million steps in order to minimize potential energy of the system using NPT. However, I noticed that potential energy of the system does not decrease that much and becomes equilibrated at a very high positive energy level which is clearly not correct. In fact in this type of systems, at a reasonably equilibrated system, the potential energy should be negative, considering I have only employed simple LJ potential to model interatomic interactions and coloumbic potential for atoms ( oxygen and Si charge are -1 and +2 respectively ). Moreover, initial configuration of system are neural and also periodic.
i also put down my commands :
I am trying to simulate diffusion of light gas molecules in a solid porous silicalite structure. The structure is rigid and not supposed to move.
I ran MD for a couple of million steps in order to minimize potential energy of the system using NPT. However, I noticed that potential energy of the system does not decrease that much and becomes equilibrated at a very high positive energy level which is clearly not correct. In fact in this type of systems, at a reasonably equilibrated system, the potential energy should be negative, considering I have only employed simple LJ potential to model interatomic interactions and coloumbic potential for atoms ( oxygen and Si charge are -1 and +2 respectively ). Moreover, initial configuration of system are neural and also periodic.
i also put down my commands :
On the science side the first and more obvious Q that comes to mind is:
Why in the world are you trying to use the NPT ensemble if your zeolite framework is rigid? Being this the case note that with a rigid framework your pressure definition (whatever you think that means) could be meaningless. As an extra bonus you may want to know that employing bare bones MD to equilibrate this type of systems is, in general, a bad idea. Especially when the molecules diffusing have more than a few degrees of freedom. Gran Canonical Monte Carlo techniques are commonly used to equilibrate systems of this type. Once equilibrated, then you can run MD to compute the diffusion coefficient (if that’s what you are after).
Carlos