Dear Lammps users,
I would like to reproduce results obtained using Rappe&Goddard QEq method (very simple potential for MoS2 with LJ + QEq, [Dallavalle et al., Langmuir 2012, 28, 7393-7400]). As someone suggested me in this mailing list, I tried to do this with gulp and everything has gone fine. Recently I noticed the new implementation of QEq methods in LAMMPS. Then I tried to use fix qeq/slater in order to reproduce results but I got charges way smaller than expected (+0.37 on Mo atoms vs +0.76 reported in the paper). I already triple-checked basic things (geometry, parameters & units). Anyway, minimalistic input files for reproducing results are enclosed.
So far, the only reason I could find is that in LAMMPS qeq/slater method is implemented following [A. Nakano, Comp. Phys. Comm., 104, 59-69 (1997)] that use 1s orbital (~exp(-zeta*r)) for charge distributions instead of more general ns Slater-type orbitals (~r^(n-1)*exp(-zeta'*r) ) as done in Rappe&Goddard's paper.
Am I right? In case, could it be possible/easy to implement Rappe&Goddard's QEq flavour in LAMMPS?
Thanks,
Paolo
param.qeq (89 Bytes)
system.data (1.07 KB)
system.in (929 Bytes)
Fix qeq/slater is from the Streitz-Mintmire potential, not Rappe&Goddard's
original formulation. They differ by the use of core charge, but they did
use the same Slater 1s orbital.
Try making the qcore sufficiently small (e.g. 0.001) and make sure you have
the chi and eta correct (eta multiplied by 2 or not).
Ray
Dear Lammps users,
I would like to reproduce results obtained using Rappe&Goddard QEq method
(very simple potential for MoS2 with LJ + QEq, [Dallavalle et al., Langmuir
2012, 28, 7393-7400]). As someone suggested me in
this mailing list, I tried to do this with gulp and everything has gone
Dear Ray,
thank you for your prompt answer. Rappe&Goddard does not consider core charges, thus I already set qcore to 0.0. As I can see in the R&G's paper, they do not use Slater 1s orbital (equation 15 of the paper).
Paolo
Their Slater 1s orbital is the same the one used in Streitz-Mintmire.
Ray
Totally agree (Slater 1s orbital is always in the form a*exp(-b*r) ). But, as far as I can understand, they use the 1s orbital only for hydrogen atoms. For all other elements, orbitals employed are different.
Paolo
Nevertheless, ReaxFF in all implementations uses the formulation provided in the Nakano paper.
Ray
OK, but the forcefield I want to implement does not belong to the ReaxFF family...
Back to my question: could you please give me some advice about how to modify fix_qeq_slater.cpp file in order to implement Rappe&Goddard's formulation of QEq? Is it possible/easy to do this in your opinion?
Thanks,
Paolo
You will have to change the 1s orbital to the ns orbital you are interested in. Formulating the product of Slater orbitals is the more difficult part as it involves complicated integrals. It is difficult, but definitely do-able.
Ray
Ok, thanks for the answers. I will give a try.
Paolo