How to calculate dielectric constant by MD and other ways?

LAMMPS LAMMPS General Discussion
, ,
1

Hi all:
I want to calculate the material dielectric constant. My way to solve this problem can be define following step:

  1. MD to get dipole, by code (compute dipole/chunk ), and fix ave/correlate to get time correlation function of dipole moment.
  2. and then I do not know how to calculate dielectric constant, maybe can use linear response theory to get it, but I have no ideas.
    Someone could tell me the correct way to deal with this problem. THANKS.
    BEST
    Wildon
2

You can use Maicos for dielectric profile calculation, or MDAnalysis for dielectric constant calculation if its a bulk system.

1 Like
3

@WildonCao the dielectric constant is computed from using a suitable fluctuation formula which are usually based on the Kirkwood G-factor G_k which is based on the fluctuations of the total dipole moment \left< \vec{\mathrm{M}}^2 \right>_t. What is suitable depends strongly on how you compute your electrostatic interactions. For ReaxFF (as you seem to be using from your tags to the post), this is Wolf summation. You have to search the published literature for a suitable fluctuation formula. I am not aware of one, when I studied computing electrostatic constants correctly and accurately during my time as a graduate student (many, many years ago) there was no ReaxFF and people only used coulomb cutoff of variants of the Ewald summation.

I have given some seminars about this a long time ago. For example see slide 18 to 21 in http://www.theochem.ruhr-uni-bochum.de/~legacy.akohlmey/files/talk-trieste2004-water.pdf (more details and references are in my PhD thesis, but that is written in German :frowning: so that is likely not a lot of help).

Please also note that properly converged data for computing dielectric constants requires very long simulations (like 10s of nanoseconds).

2 Likes
4

Thanks @akohlmey for your patient explanation it seems to be more difficult by REAXFF, there are many unknown problem.
I want to know what will happen and how to describe the change, when I fix efield in Polymer at z direction. In my opinion, there will happen two thing when fix efield, conduction and polarization，this point of view comes from Electrical Engineering.
conduction ,it seems impossible when I use REAXFF to do MD.
polarization,I just know how to calculate dipole’s change.
I have no other idea to analysis microscopic mechanism of electric field action. Could you give me some advice from a professional perspective. THANKS
Best,
Wildon

5

THANKS @simongravelle , I will try.

6

You need to find yourself a knowledgeable collaborator. Dielectric properties are a tricky business and it requires some time and a lot of care and patience to learn how to deal with them correctly. For some properties, very subtle changes can result in big differences. For some properties, system size is important, for others simulation time, and yet other require both.

The PDF I pointed out is the result of over one year of learning and two more years of patiently running very long simulations (at a time when computers were far less powerful than now).

1 Like
7

Well, then don’t use ReaxFF :slight_smile: after all, if partially-charged particles are constantly breaking free from their molecules then what you’re modelling is basically a conductive plasma, which is very different from a dielectric.

For a first study you can just throw an efield onto your system and measure the resulting polarisation; as long as the polarisation energy is much smaller than k_B T you should be safe. (Read this paper through and try to understand as much of it as you can: Modelling electrochemical systems with finite field molecular dynamics - IOPscience) But again, it’s not clear how well this would work for ReaxFF.

1 Like