What is the meaning of Coulomb Subtraction?

Specifically, in the GULP 2003 paper, DOI: https://doi.org/10.1080/0892702031000104887, in the methods section on Two-body Short-range Interactions we have the following:

For covalently bonded atoms, it is often preferable to Coulomb subtract the interaction and to describe it with either a harmonic or Morse potential.

What exactly does that mean? Does that mean the Coulomb interaction has no effect at all when using a pair potential? That seems to be what it’s implying to me, but I can’t find an explanation of the terminology anywhere.

For 2 atoms that are bonded covalently with either a harmonic or morse potential then it’s better not to include the Coulomb interaction between these two atoms along the bond. The reason for this is that the potentials have an r0 value which is designed to be the equilibrium bond length, but if you were to leave the Coulomb term acting then this would no longer be the case since the potential would have to counteract the electrostatic part. The Coulomb terms will still act between atoms that aren’t covalently bonded to each other. Hope that clarifies things.

1 Like

Okay, so in a metal only system in which all the atoms interact through Morse potentials is there any Coulomb interaction at all? For example in systems I have set up I don’t have any charges on the atom species, does that mean I won’t have a coulomb interaction?

Does that mean that Couloumb Subtraction simply means “Don’t do the coulomb interaction”, or does it mean “Do the coulomb interaction, but then take it away again”?

For a metal you won’t have any Coulomb interaction and so the subtraction is irrelevant. In general, the way things work is that Coulomb subtraction means that the term will be calculated and then removed again. The reason for this is that if you are dealing with a 3D solid, for example, then you will evaluate the Ewald sum and then subtract a direct Coulomb interaction for that pair of atoms (i.e. the two terms are not the same when initially calculated since one has the complementary error function in real space and the other doesn’t).

1 Like