News:

SMF - Just Installed!

Main Menu

Grid potentials for mobile cations in MC/GCMC simulations

Started by AHF, September 02, 2021, 09:47:10 PM

Previous topic - Next topic

AHF

Hello everyone,

I have two questions for the RASPA community regarding the use of grid potentials in MC/GCMC simulations:

(1) I am simulating adsorption of CH4 and N2 in NaY cation-exchanged zeolites which contain mobile Na+ cations. I was wondering if grid potentials in RASPA can be also used for mobile cations (such as Na+) in the same way that one would use them for neutral adsorbate molecules? In fact, I have created a grid potential for Na+ in my system, however when I run a GCMC simulation using the corresponding grid file, I get the following error from RASPA in the output file where it tests the accuracy of grids:

PseudoAtom 7 Framework-[Na]
=========================================================================================
        Boltzmann average energy VDW (table)                 :     0.000000000000
        Boltzmann average energy VDW (full)                  :               -nan
        Boltzmann relative error VDW                         :     0.000000000000
        Boltzmann average energy Coulomb (table)             : 3091865.680709035601
        Boltzmann average energy Coulomb (full)              :               -nan
        Boltzmann relative error Coulomb                     :               -nan
=========================================================================================
        Boltzmann average Force(x) VDW (table)               :     0.000000000000
        Boltzmann average Force(x) VDW (full)                :               -nan
        Boltzmann relative error VDW                         :     0.000000000000
        Boltzmann average Force(x) Coulomb (table)           : 90829.808190701209
        Boltzmann average Force(x) Coulomb (full)            :               -nan
        Boltzmann relative error Coulomb                     :               -nan
=========================================================================================
        Boltzmann average Force(y) VDW (table)               :     0.000000000000
        Boltzmann average Force(y) VDW (full)                :               -nan
        Boltzmann relative error VDW                         :     0.000000000000
        Boltzmann average Force(y) Coulomb (table)           : -410892.822263886570
        Boltzmann average Force(y) Coulomb (full)            :               -nan
        Boltzmann relative error Coulomb                     :               -nan
=========================================================================================
        Boltzmann average Force(z) VDW (table)               :     0.000000000000
        Boltzmann average Force(z) VDW (full)                :               -nan
        Boltzmann relative error VDW                         :     0.000000000000
        Boltzmann average Force(z) Coulomb (table)           : 25639.732840344324
        Boltzmann average Force(z) Coulomb (full)            :               -nan
        Boltzmann relative error Coulomb                     :               -nan

This is bizarre to me because in my system, Na+ interacts with the framework using both BUCKINGHAM2 and coulomb interactions.
So, why Boltzmann average energy for VDW is either zero or nan? Why Boltzmann average energy (full) for Coulomb is nan?

(2) I also wonder if one could model some of the interatomic interactions using grids, while the rest of them are modelled directly (without any grids) in the same simulation run?

Any assistance will be much appreciated.
Thank you.

David Dubbeldam

The grids are tested by random insertion. So there is no mechanism that avoids overlap of the ions.
You can get: large negative Coulombic energy + large repulsive energy, that nearly cancels out.
So, this is more a problem of the test then of the grid itself.
You can run a simulation with and without the grid and see if you get the same answers.

AHF