I do calculations for Al2O3 with the QEq method in GULP. I want to use the QEq method to determine the charge for Al and O. In the QEq method, the Coulomb interaction is estimated by the Coulomb overlap integral. I hope such Coulomb overlap integral is only used during the determination of the charges. I want to use usual point charge and Wolf sum for the Coulomb interaction during the calculation of the mechanical properties. However, it seems that the qeq keyword results in the application of the Coulomb overlap integral to both the determination of charges and the calculations of mechanical properties. Are there any options to restrict the Coulomb overlap integral in QEq to only the process of determining charges?
Hi JWJ,
If you specify QEq then GULP use this method to compute everything self-consistently. If you just want the charges from QEq but then want to do a standard calculation, all you have to do is run a single point calc with QEq, and then copy and paste the charges (or use the restart file) into an input that no longer has the QEq keyword.
Regards,
Julian
Thanks for your suggestion. That works for a single static calculation. However, I want to optimize the structure, in which the QEq method should be used at each optimizing process. So, I need to use the QEq for the whole calculation.
If I use the QEq in GULP, then all Coulomb interaction is calculated by the Coulomb overlap integral. The Coulomb force from the Coulomb overlap integral seems to be much smaller than the Coulomb force calculated by the Wolf sum of point charges.
Initially, I have a Born-Mayer model and the Coulomb interaction for point charge summed by the Wolf. This initial model gives reasonable mechanical properties. If I use QEq, then the Coulomb interaction is calculated by the Coulomb overlap integral, and I find that I need to make a large modification for the Born-Mayer parameters. I though the difference between the Coulomb overlap integral and the Wolf sum of point charge should be small, considering that the Al-O bonds are not so small and there is no H element here. Look like I am wrong. The difference is not small.
Hi JWJ,
There is a fundamental problem with what you are looking to do, unfortunately. If the charges are being determined at every geometry via QEq then this gives rise to a force between the atoms that arises from the charges being a function of the geometry. If you then throw away the QEq method for interactions then you are also throwing away the forces that result from this & so your energy and forces will no longer be consistent & your calculation will go horribly wrong. So I’m afraid fixing the charges is the only way (or switch to a different method such as EEM that uses the full Coulomb potential).
Regards,
Julian