# Justification of Molecular Dynamics (MD)!

1. ### Jatsui

3
Guys,

The question is as follows:

Q) I currently simulate organic systems using the Molecular Dynamics (MD) program. Previously, I have used MD to simulate primarily Silicon and Boron interaction at relatively high energies which is fine with the classical approach used with MD. I was wondering what justification can be provided for modelling organic systems, with MD, where energies go down to fractions of eVs (~0.1 eV), for which quantum mechanical effects dominate? In short, MD is good for classical mechanics but how can it be justified for modelling certain systems that sometimes follow quantum mechanics OR is it ok to do this?

Thanks, appreciate it!

2. ### bcrowell

5,805
Staff Emeritus
This should probably be posted in the "Atomic, Solid State, Comp. Physics" forum. If you wait a little bit, I would guess that an administrator will move it there.

3. ### alxm

1,864
Well, from the QM point-of-view, the fundamental approximation that underlies MD is the Born-Oppenheimer approximation; This tells us that the energy of the system/molecule can be parametrized in terms of the nuclear coordinates and more precisely, that this approximation is valid through the second-order in terms of energy. (IOW, treating a chemical bond like a harmonic oscillator is a valid second-order expansion around its equillibrium) Non-BO corrections (i.e. vibronic coupling) come in first at higher orders, and higher energies. So unless the system is in a highly excited vibronic state, it's not a problem. As for the electronic state, chemical systems are almost entirely in an electronic ground-state at room temperature, so that's not a major issue either.

So we're justified in treating the system's energy as a potential-energy surface as long as the system is not highly excited, or put another way, the coordinates aren't far from their equilibrium positions (unstretched chemical bonds). If that holds and the parametrization is good, it's a decent approximation. But if you stretch the bonds too far, the electronic state and the PES changes of course, so it all falls apart. (and describing how the PES and electronic state changes with a change of nuclear coordinates requires nothing less than an explicit QM treatment)

So in short: MD models are fine as long as you're not breaking or forming any chemical bonds, and your system is not in an excited electronic state or a highly level of vibronic excitation.

4. ### Jatsui

3
bcrowell - i think the shift has been made, cheers.

alxm - thanks for the advise!
Thing is that we are damaging the organic systems and are breaking bonds. The potential used is the Brenner REBO potential....

5. ### Useful nucleus

278
Some force fields used in MD such as REBO, REAXFF are designed to be reactive and hence they can account for some chemistry. The challenge is to know the chemistry of your material very well so that you can parametrize a force field to describe this.
The computational cost of reactive for fields are still much less than the quantum mechanical appracohes such as Car-Parinello MD or Born-Oppenheimer MD.

6. ### Jatsui

3
thanks useful nucleus!
useful comment indeed ;)