I was wondering if somebody could give me some insight into best ways to produce crystal structures, with a given force field, at 0 K. I am interested specifically in glycine, and have obtained the experimental structure from the CSD. I have seen in the literature people refer to annealing, where I equilibrate my crystal at a large temperature, and then anneal. However, I am concerned about the effects of such things as annealing rate here. I have also seen people mention performing energy minimization. However, my concern is typical energy minimization routines, at least to my knowledge, do not adjust the simulation volume according to a specified pressure. Thus, if my virial is positive for my experimental atomic coordinates using my force field, I don't think the minimizer would expand the volume to bring the pressure back to 1 atm. Could anyone give me some guidance. I am trying to prepare glycine crystal structures, at 0 K, using the GAFF force field with the CNDO charges.

  • 1
    $\begingroup$ I'd also have a look whether there are multiple crystal structures available - if so (one would believer there are), compare the different conformations manually, they might give you some insight into which conformations are allowed/preferred. $\endgroup$
    – Gerhard
    Oct 15, 2015 at 6:41
  • $\begingroup$ Glycine is indeed polymorphic, if thats what you were referring to. I don't think I am quite getting your point. Could you explain further? $\endgroup$ Oct 15, 2015 at 15:17
  • $\begingroup$ It is possible to perform a simulation with constant pressure or constant volume boundary conditions, so it is not a concern. However you should realize that prediction / simulation of accurate crystal structure is neither simple nor the purpose of most MM techniques. $\endgroup$
    – Greg
    Dec 3, 2015 at 0:47

1 Answer 1


I'm not entirely sure what you're asking. Are you attempting to match the geometry of a molecule from a crystal structure or the entire crystal packing?

Computational prediction of crystal structures is very much an unsolved problem. There have been multiple "blind tests" from the Cambridge Crystal Database, with varying levels of success. In general, the most accurate results come from dispersion-corrected DFT, not force field methods.

In some cases, to speed results and screening, teams will prepare "tailored force fields" by deriving optimized parameters for a particular compound on-the-fly from the DFT calculations. In no way are these intended for general purpose force field methods.

Also, most methods typically have to optimize both the unit cell parameters and the geometry of the compounds in them. Your note about expanding the volume is important - if you are working in this field, you may wish to use such heuristics to optimize the cell parameters, but typical minimizers will focus on either atomic coordinates or unit cells but not both.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.