Skip to content

Protocols and specification

Jump to bottom
Jason.Yu edited this page Sep 27, 2021 · 2 revisions

The definition and specification of protocols

Common specifications: Structures

The input structure is not an optional parameter for all the verification workflows.

In Δ-factor calculation, most stable elemental system from Cottenier's work and rare-earth nitrides from Topsakal-Wentzkovitch work;

In convergence workflow Phonon, pressure, cohesive energy: Cottenier's structures (except SiF4 has been used for F because of convergence issues) and rare-earth nitrides; Use primitive cells.

  • Bands: Cottenier's structures reduced to primitive cells (except SiF4 has been used for F because of convergence issues) and rare-earth nitrides. PwbandWorkflow will make a primitive cell for band calculation (Remember to turn off the relax step).

The definition and usage of different protocols

Parameters of Δ calculations

  • wave function cutoffs: 200 Ry;
  • dual = 8 (PAW/US), 4 (NC); Mn/Fe/Co have larger duals tested as well; 12 and 16. We have gone in a mode where we do not use the dual, but we use ECUTRHO and ECUTWFC. However, dual is still used in simply setting the ecutwfc/ecutrho pairs.
  • k-points: 0.1A^-1;
  • smearing (degauss): Marzari-Vanderbilt, 0.01 Ry;
  • non spin-polarized calculations except Mn (antiferromagnetic), O and Cr (antiferromagnetic), Fe, Co, and Ni (ferromagnetic).

As for calculation of lanthenide, always increase nbnd to two times of the default number.

Parameters in phonon, pressure, cohesive energy calculations:

  • k-points: 0.15A^-1
  • smearing: Marzari-Vanderbilt, 0.01 Ry;
  • k-points for the isolated atoms: 1x1x1;
  • smearing for the isolated atoms: gaussian 0.01 Ry;
  • unit cell for the isolated atoms: 12x12x12 Å with atom sit in [6.0, 6.0, 6.0] the middle of the cell;
  • q-point: only calculate the phonon frequencies on Brillouin-Zone border q=(0.5, 0.5, 0.5).
  • all calculations non-spin-polarized.

In isolate atom energy calculation of cohesive energy evaluation. As for lanthenide, increase nbnd to three times of the default number. Moreover, use more RAM(by increase num_machine to 4).

NOTE: PWscf writes in the output something called total energy. This is NOT the total energy when you have smearing; it’s the total free energy E-TS. PWscf also writes -TS, so one can get back the total energy E. In general (for a metal) E-TS should be used. For an atom instead the total energy should be used, since the -TS term is not really physical (it comes from the entropy of fractional occupations on the atom). Check with Nicola if you have atoms where -TS is different from zero. (http://theossrv1.epfl.ch/Main/ElectronicTemperature)

Clone this wiki locally