ABINIT Discussion Forums

Hi...
Im doing LDA & GGA calculations using troullier martin pseudopotential for some of II-VI semiconductors (ZnS,ZnSe,...) but my results are not accurate comparing with other LDA & GGA results,, and when i use HGH psp my results are good for the same other input variables ,,, is troullier martin pseudopotential are always not accurate or there is something wrong with my input file?

Nawzad A.

Depends on the pseudopotential. In general various types of NC potentials can give comparable accuracy. PAW will be more precise.

Many of the Troullier-Martin pseudopotentials have core radii whose sum is greater than the typical bond length. If the cores overlap, the results are generally bad. GTH pseudopotentials will avoid this
problem, but the cutoff energy is much higher. PAW pseudopotentials offer accuracy and low cutoff, but often boxcut must be must greater than 2.

Mike

tetermp wrote:Many of the Troullier-Martin pseudopotentials have core radii whose sum is greater than the typical bond length. If the cores overlap, the results are generally bad. GTH pseudopotentials will avoid this
problem, but the cutoff energy is much higher. PAW pseudopotentials offer accuracy and low cutoff, but often boxcut must be must greater than 2.

Mike

I have GTH pseudopotentials but for LDA there are no GTH pseudopotentials for all elements in abinit website and for GGA the pspcode integer equal to 10 but in abinit should be between 1 and 8
could I change pspcode??

Nawzad A.