ABINIT Discussion Forums

Greetings,

I am currently working with ABINIT to calculate the cohesive energy of transition metals, specifically Palladium and Rhodium. The calculations are with LDA-TM and the corresponding pseudopotentials. When I finally determine the cohesive energies and compare them to experiment my results are off by 60% for Rhodium and 80% for Palladium. My questions are, is this a common result for transition metals? and is there a way to improve / handle a heavy transition metal calculation?

I will be happy to provide additional information if needed to help with this issue.

Thank you all.

Calculating cohesive energies is very ticky.
First, compare your results to the literature results obtained with the functional you use, not to experimental results.
In this way you validate your procedure (in order to avoid the situation when your incorrectly calculated numbers "agree" with "experiment").
Here is an example of a correct one: http://www.sciencedirect.com/science/ar ... 2805013154 (includes Pd).
The hard part will be to make sure your atom is in a right electronic state (non-spherical if needed) at zero temperature, ar as close as possible to zero temperature (tsmear and occopt options), in a large box.
Isolated transition metal atoms may show SCF convergence problems and trying to play with mixers/eigensolvers may result in getting convergence to spurius states.
Then, be aware that even if you calculate it using a correct procedure you may get a wrong answer due to the used pseudopotential.
See for example my post viewtopic.php?f=9&t=1365 where i'm trying (unsuccessfuly) to get PBE atomization energy of O2 right.