Dear Lammps Dev,
In order to use eff, we need to input the electron radius and positions into the data file. However, according to the manual, the electron radius of C-C bond is 1.258 bohr and C-H is 1.543 bohr. What is the exact meaning of electron radius? How can we calculate them? Is it the expected value of the electron distribution function? And the calculation of electron radius will be different from system to system e.g. B-N or Si-C system?
The first thing to note is that the electron radius in eff is an ‘electron’ property, not a bond property. In eff, electrons are represented by wave packets, which substituted into the time-dependent Schrodinger equation and assuming a locally harmonic potential produce the semiclassical equations of motion that describe the average position and SIZE of a wave packet (classical dynamics, Ehrenfest’s theorem).
The electron size is associated with the electronic (wavefunction) kinetic energy, which in turn is a manifestation of the Heisenberg uncertainty principle (inverse square of each electron’s size). This kinetic energy attempts to make the electron larger while any electrostatic interaction between the electron and protons, attempts to shrink the electron size. Thus the electron radius will depend on its environment (system), e.g. diffuse electrons will have a large radius, and localized ones a small one.
To get you going fast, go over some of the examples we have included for most elements currently supported by eFF (/examples/USER/eff). Keep in mind that the full electron representation is limited to low Z numbers, so if you’re going to do something like SiC, you should use the effective core pseudopotential (ECP) representation (all the info you need is under http://lammps.sandia.gov/doc/pair_eff.html). Furthermore, in contrast to the full electron case, which dynamically adjusts the electron radius (i.e. the force field will automatically adjust your electron radii for the particular system), we have fixed the ECP radius based on a QM training data set. You can see the suggested default ECP radii values in the settings section of the pair_eff_cut.cpp code.
of course, the ECP valence electrons are also dynamic variables