Difficulties converging Acetonitrile MD run

[Michela Benazzi]

Hello everyone,

I am running NPT MD on a system of 25 acetonitrile molecules at 310 K and cannot get it to converge.

The purpose: to have an equilibrate system of acetonitrile that I can later add to a previously equilibrated slab of liquid Ga + vacuum. The acetonitrile would replace the vacuum space (10.8x10.8x21 A^3), and I would run MD again on the new integrated unit cell.

Some methods:

1. Unit cell: used the density of acetonitrile at 298.15 K. Boiling point is 355.1 K, so I assumed the density difference would be negligible.
2. Periodicity: XY to mirror the slab I will be using later (metal layer + vacuum on top).
   > Perhaps I don't need to use the same periodicity?
3. Used PBE, but did not account for Grimme D3 dispersion.
   > Because it is a polar molecule, I probably need to add for DFTD3. This paper mentions Coulomb F + Lennard Jones F having an effect on the potential.

I am rerunning it right now after adding DFTD3 and XYZ periodicity, but would still appreciate any advice. Sometimes I feel like I can "get away" with simplifying my system, and it only comes back to bite me later.

Thank you kindly for your help,

Michela

[Marcella Iannuzzi]

Hi Michela

Are you sure that the initial coordinates in MeCN.xyz are OK? They look weird. How did you get them. 

To equilibrate the density of the liquid one should consider a bulk system.

The initial pressure is very suspicious, indeed the temperature increases enormously. 

My first guess is that the initial coordinates are problematic.

The DFT settings are also not particularly good, but the main problem is the structure.

I would advise to first get a reasonable model of the bulk liquid and then extract from it the slab model you need. 

Probably it would help to first equilibrate with force fields a large acetonitrile box and only later run AIMD starting from the pre-equilibrated structure.

Regards

Marcella