You should describe a little more your system to get more help. For example, are you simulating this molecule in gas phase or solvent environment, are you using periodic boundary conditions? If you are using periodic boundary conditions it is recommended that the cutoff of the non-bounded interactions for the MM part to be smaller than half of the box size.

A good reference:

Frenkel and Smit, Understanding Molecular Simulations.

In addition, you should mention the software you are using.

I am trying to perform QM/MM simulation (Geometry optimization and then NVT simulation). I have a small organic molecule in a cubic box of water (box size 30*30*30). The organic molecule at the centre of the box is defined as the quantum region and the water molecules in MM criteria. I have performed a classical MD simulations, prior to the QM/MM one, following the standard MD simulation protocol.

The query is related to the setting up a QM/MM simulation by taking this final structure from the previous simulation. I came accross few literatures where some people have used periodicity in the QM part while others did not.

Choosing a proper box size for the quantum part (periodic or non-periodic) seems to be crucial for this type of calculations. Any suggestions in this regard would be highly appreciated.

Hi Tarak,

Yes, the boundary conditions will have non-negligible effects. In general, these kind of simulations will depend on the QM Hamiltonian, MM Hamiltonian, coupling between QM and MM regions, boundary conditions for MM region, and size of QM region.

Some codes treat QM-MM electrostatic interactions when they calculate electrostatics of MM region, therefore setup of proper cutoffs for QM/MM electrostatic interactions may be important if periodic boundary conditions are used for the MM region.

There is a paper that addresses some of your concerns:

Journal of Computational Chemistry, 33, pp 363, 2012.

Thank you Sir.

Based on your longest cut-off distance for non-bonded interactions you can judge the situation. Note that your organic molecule should not interact with its periodic image withing the cut-ff distance. Also consider that you cannot arbitrarily decrease your cut-off to meet this condition as you will get bunch of artifacts. If your computation power is fair you may also want no water molecule to be cut-off range of your organic material and one of its periodic image. But in the case of limited computational resources you can compromise. Also you can use dodecahedron box type to reduce number of water molecules required.

How do I choose the QM box from the entire simulation box.

I understand that the molecules within qunatum description should be included in the QM box. I want to have few water molecules in the qunatum region. Then how do I get the proper size of the QM box (containing a single organic molecule + x number of water molecules)?