Author Topic: Numforce MP2 for high temperatures  (Read 4500 times)

Stegi

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Numforce MP2 for high temperatures
« on: April 22, 2013, 10:17:59 AM »
Hello everyone,

I am calculating reaction energies for various small molecules and thermodynamic corrections for high temperatures (650 - 950 K). On the MP2 level I am using numforce to calculate the vibr. force constants with the option -central. I used the default displacement of 0.02. All geometries have only real modes and the SCF and density convergence criteria are tight, respectively.

Here comes my problem: For 950 K the thermodynamic corrections (freeh) are pretty off. For some reactions, the Gibbs free energy rises at high temperature, as opposed to expectation (and GGA-level aoforce calculations). In most cases, however, the Gibbs free energies are way too low, so there is no similar behaviour or trend through all the reactions. This puzzles me. Some of the dG values are completely unrealistic (although corresponding MP2 energies and dG for lower temperatures from the same calculation are fine).

Is this odd behaviour of numforce thermodynamic properties at high temperatures known? Is there a solution or a way around? Or do I have to stick to lower temperatures when applying numforce?

Best regards and many thanks for suggestions!
Stegi

Stegi

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Re: Numforce MP2 for high temperatures
« Reply #1 on: May 16, 2013, 02:20:06 PM »
I narrowed down the misbehaviour to something that might be a bug in the freeh script:

Taking numerical MP2 force constants from numforce, there is a sudden drop of the thermodynamic corrections between 773 and 774 K (400 to 401 Celsius) leading to an interruption of the linear trends on each side of this temperature.
Does anyone have an explanation for this?

Stegi

Stegi

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Re: Numforce MP2 for high temperatures
« Reply #2 on: May 23, 2013, 05:03:23 PM »
Hello,

this issue is solved. It was a processing error on my side and nothing is wrong with numforce or freeh.

Sorry my bad, :-\
Stegi.