Question on Morse and three-body potential

Hi everyone,
I am trying to fit potential parameters for Na2SO4. As depicted in the previous essay (doi: 10.1016/j.jcrysgro.2006.05.032 & 10.1039%2FFD9939500273), the SO4 ions could be described by a Morse potential and S-O-S three-body term without Coulombic interactions. I am wondering if I give the species and Morse potential, will GULP perform coulombic interaction automatically? Or should I write some command to stop the coulombic computation?
Besides, I also don’t know how to determine the rmin and rmax parameters for Morse and three-body potential. As I could see from the examples listed below, these parameters of Morse and three-body potential are often shorter than those for Buckingham potential. Is that always true?

Example6 for O-O Buckingham potential:

O shel O shel 1280.0 0.3 27.88 0.0 10.0

Example10 for C-O Morse potential:

morse intra
C core O core 6.288000 2.060000 1.23000 0.0000 1.600

Example10 for N-C-O three-body potential:

C    core N    core O    core   5.897000 121.000000  1.600  1.600  3.000

My input file:

fit opti prop conp
5.385 5.385 7.26 90 90 120
Na 0 0 0 
Na 0 0 0.5 
Na 2/3 1/3 7/8 
S  2/3 1/3 0.27 
O  2/3 1/3 0.46 
O  0.2 -0.2 0.2 
Na  0.6
S   3.6
O   -1.2
Na O    4383.7555 0.243838 30.70 0.0 10.0 0 0 0
O  O    1844.7458 0.343645 192.58 0.0 10.0 0 0 0
morse intra
S O 5.0 1.2 1.505 0 1.6 1 1 1
S O O 5.5 109.5 1.600  1.600  3.000 1 0

Thanks for any possible guidance in advance.
Ji Yao

Hi Ji Yao,
I’d strongly recommend reading the literature and using one of the existing models for Na2SO4 rather than fitting before being familiar with the details of the potential model. Since you’re trying to set up a model that is essentially the same as existing ones, why not start from these and only change if there is a problem?
To answer a few points:

  1. GULP automatically computes the Coulomb interaction if there are charges. Where charge interactions aren’t appropriate (covalent bonds for example) then you can use the “molecule” keyword or similar (or the Coulomb subtraction options) to remove this contribution.
  2. There is no S-O-S threebody term in sulphate since there is only 1 S atom
  3. The cutoffs for things like morse and three-body terms will always be less than for Buckingham - just think about what they physically represent. The former are bonded interactions, while Buckingham includes van der Waals.
    My advice would be to study the literature models first & only having critiqued their strengths and weaknesses in detail consider changing the model.
    Regards, Julian

Hi Julian,
Thanks so much for your reply. The partial charge of S and O in previous literature is not consistent with what I need in a multi-component glasses and that’s why I am trying to fit some parameters by myself on the basis of these studies. Thanks for your advice and I will read more relevant literatures and enrich my theoretical knowledge.
Ji Yao

Hi Ji Yao,
As long as the total charge on sulphate is the same (-2) then the precise charges on S vs O has a much weaker effect on the interactions & so you might be able to use the other models still.

Hi Julian,
Thanks for your suggestion and I will try to use the other models to see the MD results.
Ji Yao