To enable the calculation of electric dipole moments, potfit has to be compiled with the dipole option. The dipole $\vec P_{i,n}$ of atom $i$ in iteration step $n$ is then calculated self-consistently by the use of the Tangney-Scandolo potential model¹⁾.

$$\vec P_{i,n} = \vec P_{i,\text{NF}} + \vec P_{i,\text{IND}}$$

The near field (NF) part,

$$\vec P_{i,\text{NF}} = \alpha \sum\limits_{j \neq i} \frac{q_j \vec r_{ij}}{r_{ij}^3} f_{ij}$$

is caused by the electric field of nearby charges. The induced (IND) part,

$$\vec P_{i,\text{IND}} = \alpha \vec E (\vec P_{j,n-1}),$$

is due to the electric field of the other dipole moments. $\alpha$ is the polarizability of the considered atom type and $f_{ij}$ is an ad hoc introduced function to account for multipole effects of nearest neighbors.

dipole implies option coulomb, because charges are needed to evaluate the dipole moments.

dipole can be used without specifying additional parameters in the parameter file, because everything works with default values. However, advanced users can specify two new parameters:

dp_tol - float - 1.e-7
   dipole iteration precision.

dp_mix - float - 0.2
   mixing parameter for dipole convergence during iteration.

• dipole has to be compiled with option apot.
• dipole implies coulomb.
• dipole can be used with stress, fweight, evo and can also be executed in parallel using option mpi.
• dipole can not be used together with other force-field approaches (pair, adp, eam, …).

When using dipole, the following parameters have to given in the potential file, straight after the charges:

• alpha polarisability for each atom type.
• b and c parameters of the short-range dipole-model, have to be given for each interaction.

Example for the diatomic oxide SiO₂ (contains coulomb-parameters):

#F 0 3
#C Si O
#I 0 0 0
#E
 
elstat
ratio       1 2
charge_Si   value min max 
kappa       value min max 
alpha_Si    value min max
alpha_O     value min max
b_SiSi      value min max
b_SiO       value min max
b_OO        value min max
c_SiSi      value min max
c_SiO       value min max
c_O         value min max

An entire potential file can be downloaded here: Examples

To describe a system with $N$ atom types you need $N(N+1)/2$ potentials.

The potential table is assumed to be symmetric, i.e. the potential for the atom types 1-0 is the same as the potential 0-1.

The order of the pair potentials in the potential file for $N$ atom types is:

$\Phi_{00}, \ldots, \Phi_{0N}, \Phi_{11}, \ldots, \Phi_{1N}, \ldots, \Phi_{NN}$