Dear Amir,

Increasing your simulation box(while keeping the density) will increase the number of particles in the system And therefore the interactions to be computed will increase. In classical MD, you need to compute N2 interactions. So, increasing too much your box will slow down your calculations.

Depending on what you want to study, the choice of the system size must be made as a compromise between efficiency and accuracy. In MD you will have finite size effects, you can study this for instance analyzing properties for different sizes, let’s say increasing the Volume to be V, 2V, 4V... and trying to study what the properties for an infinite size(like in a real system) would be. So you can safely chose the larger, but at the same time minimum doable by computer time.

other than that, you can increase the size as long as you can afford the computation time.

Thanks Henao

But I deal with a system of given number of particles, such as attached

I'd like to study the interfacial behavior of water and oil while the distance between the top of water layer and oil bottom being larger than cut-off distance. I know this.

But, what if taking box z-dimension larger than enough? 

Does it effects simulation results or computation speed? How?

Thanks in advance. 

Well, in the case of interfaces like this you have two choices that depend on the real system you are studying.

1. In a real system of an oil (Organic for instance) droplet in water, you have only one interface water/oil. If you introduce a vacuu, it means you are adding two interfaces water/air and oil/air. In that case it could be a good idea to increase enough the length of both phases in the z direction without vacuu. It would mean that you will have only wate:oil interfaces repeating periodically and therefore representing better your system. In that case the length should be enough to avoid artifacts.

2. As you do it in the picture, the length of z should guarantee that bottom waters don’t “see” top oil. In this kind of systems it is possible that you will have some molecules evaporating, so you should also take that into account. Making larger the vacuum part could maybe affect the part of the long range electrostatics. But I don’t know how much it would affect the efficiency.

Hello,

You can check my thesis (i worked on interfaces using MD simulation) :

Thesis Numerical study of the physical factors responsible for the ...

OR this article

Article A Predictive Calculation of the Crystallization Tendency of ...

Best regards,

Joseph