WARNING: ENERGY DRIFT (INTERNAL CONSISTENCY ERROR IN THE CODE)

[joannewanghf]

Dear Prof. Dubbeldam and RASPA community,
I have used RASPA 2.0.37 for gas adsorption simulations in a Cu-MOF in last months.

I built myown forcefiled files based on Dubbeldam2012MIL-101, and got reasonable isotherm data for CH4 and CO2 adsorption. But encountered ENERGY DRIFT warning when I move to C2H2 (acetylene) adsorption, even at low pressure (e.g. 24Pa, 298K).

I found this warning would be eliminated when the charge of C_C2H2 and H_C2H2 equal to zero, but it leads to very low absolute adsorption (i.e., 0.02 cm^3/g. the experimental result is ~52 cm^3/g).  When I set the charge of C_C2H2 and H_C2H2 as 0.278, I got close adsorption as experiment but ENERGY DRIFT warning occurs.

I tried to change, respectively, radii of Cu, C_C2H2 and H_C2H2 in pseudoatom.def, L-J parameters of Cu, C_C2H2 and H_C2H2 in force_field_mixing_rules.def. All of attempts led to dramatically changes in amount of total ENERGY DRIFT and seems no rules. Besides, I performed two runs, respectively, with the same input files, it gives very different total energy drift.

So how can I fix this problem, and how to build a proper forcefield for specific systems, is there some guidelines or skills?

Some parameters used in one of my jobs with the smallest total energy drift (1.11544) is attached :
   

[joannewanghf]

I modified atom coordinates of C2H2 and problem solved. But I don't know why.

[David Dubbeldam]

It might be that if you choose the atomic positions wrong, that the molecules can overlap. This happened in particular with for example water where the larger oxygen shield the hydrogens. Often the hydrogens do not have a VDW parameter because the repulsion of the oxygen is sufficient. Choosing a wrong repulsion (too small sigma) leads to overlap with the hydrogens and hence numerical issues. So with classical force fields you should always be aware and avoid unphysical situations.