The ABINIT Users Mailing List ( CLOSED )

[paul.boulanger@umontreal.ca]

Hi,
   you are correct: to calculate the work function of a surface in Abinit, 
you must take the difference between the maximum of the total potential in 
the vacuum and the HOMO (or Fermi) energy. This is because in periodic 
systems the potential is not completely determined, one can always add a 
constant shift. Hence, the eigenvalues calculated can arbitrarily shift in 
energy. Taking the difference of these eigenvalues with the maximum of the 
potential permits to fixe the origin and obtain an absolute value. While if 
one makes a calculation without Born-von Karman periodic boundary conditions, 
i.e. like Gaussian, this is not necessary.

   Within Abinit, you can set the variable prtpot to 1, which will output the 
total potential in a binary file _POT (see input variables on Abinit web 
site). You can then use CUT3D to project this potential on a line, taken in 
the direction  perpendicular to the slab. Using XMGRACE you can then 
visualize the potential versus direction from the slab. One must make sure 
that the distance between slabs is converged, by looking a the smoothness and 
the flatness of the potential near the maximum, in the vacuum.

    I must warn you that in general, the work function is generally 
underestimated within LDA and GGAs. This is because, in these functionals, 
the potential falls off faster then the real one. But, the values are 
self-consistent, so you can compare different LDA work functions.

                          cheers,
                              Paul