SW potential-ATK and Lammps

[payam]

Dear all,
It seems that there is no one to one correspondence between SW parametrization in ATK and in Lammps. The number of cutoff distances in 3-body interaction terms of ATK is more than that of Lammps. The same thing is true for gammas.
This the list of parameters in Lammps:
Epsilon   Sigma           a           lambda     gamma           costheta0   A           B           p   q   tol

which is less than this list as an example:
potential = Stiwe3Potential(particleType1 = 'Se', particleType2 = 'Mo', particleType3 = 'Se', gamma0 = 0.425*Angstrom, gamma1 = 0.425*Angstrom, l = 8*eV, cosTheta0 = 0.2, type = 1, r_0 = 3.25*Angstrom, r_1 = 3.25*Angstrom, r_13 = 3.85*Angstrom)
Any clarification would be appreciated.

Best  regards,
Payam

[Anders Blom]

Actually there is a 1-to-1 correspondence, except that we have hardcoded q=0 as it seems to be the value chosen in the original SW article anyway.

The point in LAMMPS is that many of the parameters need to be specified not once, but for each atom/pair/triplet.
See http://lammps.sandia.gov/doc/pair_sw.html#stillinger

[payam]

Thank you Anders for your response.
BR,
Payam

[payam]

Dear Anders,
Still I am confusing.
The ATK considers only two types of three-body interactions in SW potential for MoS2 which are S-Mo-S and Mo-S-Mo. However, we know that we have two types of S atoms: S (up) and S(down). So, we need three types of three-body interactions such as:
1-S(u/d)-Mo-S(u/d),
2-Mo-S(u/d)-Mo
3-S(u/d)-Mo-S(d/u)
You notice that the cases 1 and 3 are not equal as the bond angles are different. So, my question is how do you handle it in ATK? Do you assume that they are equal?

Best regards,
Payam

[Anders Blom]

I'm not a deep expert in the internal mechanics of this potential, but as far as I can tell, no, they would not be treated equally, because there is an angle-dependent term in the potential itself (theta). Only theta0 is common, although I don't know what physical role it plays...
Of course, it may be possible find a better "monolayer MoS2" potential by explicitly considering the two S as inequivalent, and fitting to a lot of DFT calculations, but then you might as well do DFT for the problem at hand

[payam]

So, is there any possibility to add the ignored interaction to SW potential for MoS2 in ATK?

[Anders Blom]

There is no specific "ignored interaction" neither in the potential or ATK. What I was talking about was making a new type of potential, but that's not a simple task by any means (and not really possible either in ATK).

[Anders Blom]

I think if you evaluate the pairwise forces between the atoms using the potential you will see that it does indeed treat 1 and 3 differently, as you want.
It is however not possible to extract that information from ATK directly...

[payam]

Sounds good. Thank you Anders for immediate responses.

Best regards,
Payam