Blow-up during NPT Equilibration

GROMACS version: 2022.2
GROMACS modification: No

I simulate molecular dynamics of 1024 water. Water molecule is made by quantum calculation using Orca with B3LYP functional and def2-TZVP basis-set. I set GAFF by AmberTools. The simulation consists of minimization; NVE, NVT (310 K), and NPT (1 bar) equilibration; and NVE production.

The results seems good for minimization, NVE and NVT equilibration. Potential energy is converged and temperature is steady. But, during NPT equilibration, the system blow-up from 31.3 nm^3 to 38.5 nm^3. It affects the density of the system. The density should be 0.99 g/cm^3, but the result is 0.80 g/cm^3.

I’ve tried using different barostat: Berendsen (tau-p = 0.1) and C-rescale (tau-p = 2 and 4). The volume and density doesn’t give any difference, but the pressure is. Pressure for Berendsen is 1.44 bar (close to 1) and for C-rescale is 6.53 fot tau-p = 2 and 10.30 for tau-p = 4.

Can anyone help me how to overcome this problem?

The force field parameters calculated through quantum chemistry may not be suitable for direct use in molecular dynamics. Some existing water models, such as SPC/E or OPC, have optimized parameters, making them better suited for simulating pure water systems.