[QE-users] Underestimated Adsorption Energy with optB88-vdW

Tamas Karpati tkarpati at gmail.com
Wed Oct 7 12:57:30 CEST 2020


Dear Gabriel,

1/2, are your sims. reproducing *all* of the ref. params? (eg. are
cell parameters the same?)
      any missing params. would contribute by adding an *unknown*
magnitude of error...
2/2, my guess would be that your evaluation of dE_a is wrong due to
*different cell sizes*
      (basis set superposition error?); this would explain the
systematic character of the deviation
      and suggest (pain indeed) to increase cell sizes up to the same
(ie. the largest) of
      E_slab+ads, E_slab and E_adsorbant.
+1, your eq. for the heat is recursive (kidding): LHS is to be renamed!

HTH,
  t

On Wed, Oct 7, 2020 at 12:19 PM Gabriel Bramley <G.A.Bramley at soton.ac.uk> wrote:
>
> Dear Users,
>
> Problem:
>
> I am performing calculations with the optB88-vdW functional to obtain adsorption energies of small aromatic molecules on metallic surface facets (Pt(111) and phenol for the cases discussed here). Unfortunately, the adsorption energies obtained from these calculations are systematically 50 kJ mol^-1 below comparable studies performed with VASP. I have replicated their parameters as closely as possible - as such I am also using PBE PAW pseudopotentials from the psl pseudopotential library.
>
> The above underestimation occurs for a fully periodic calculation. Using the dipole corrections (2D and ESM (bc1)) appears to exacerbate these issues, leading to a 70 kJ mol^-1 underestimation of the adsorption energy. (The system is centred in the centre of the cell with 8 Ang of vacuum either side of the system for 2D, and the system is centred around z=0 for ESM).
>
> The calculation for adsorption energy is performed as:
> E_ads = E_slab+ads - (E_slab + E_ads)
> Where the adsorbate is calculated in a 30x30x30 Ang cell with the Markov-Payne dipole correction (Though this give negligable difference compared to full PBC).
>
> Question:
>
> The PBE+rvv10 functional performs significantly better with PBC and 2D Coulomb Cutoff, giving sensible and expected values for the adsorption energy. However, I am cautious in trusting these results given the large errors for the optB88-vdW functional. Is there something obvious I am missing in these calculations or is this a known issue?
>
> A previous email to this list has pointed out other issues with this functional (https://www.mail-archive.com/users@lists.quantum-espresso.org/msg35965.html). This might help, but could someone please clarify this as well?
>
> Additional information:
>
> I am using QE v.6.5 compiled with Intel-2018 with the ELPA 2017 library. Enivron Module installed but not used in these calculations. All QC tests pass without error.
>
> Please find below the input file:
>
> ___________________________________
>
> &CONTROL
>    calculation      = 'relax'
>    restart_mode     = 'from_scratch'
>    prefix           = 'Pt'
>    disk_io          = 'low'
>    pseudo_dir       = '/home/gab1u17/pslib-kj-paw/'
>    nstep            = 300
> /
> &SYSTEM
>    ibrav            = 0
>    ecutwfc          = 45
>    ecutrho          = 360
>    occupations      = 'smearing'
>    degauss          = 0.05
>    smearing         = 'gaussian'
>    input_dft        = 'vdw-df-obk8'
>    assume_isolated  = '2D'
>    ntyp             = 4
>    nat              = 48
> /
> &ELECTRONS
>    electron_maxstep = 150
>    conv_thr         = 8e-08
>    mixing_mode      = 'local-TF'
>    mixing_beta      = 0.2
> /
> &IONS
>    ion_dynamics     = 'bfgs'
> /
> &CELL
> /
>
> ATOMIC_SPECIES
> Pt 195.084 Pt.pbe-n-kjpaw_psl.1.0.0.UPF
> C 12.011 C.pbe-n-kjpaw_psl.1.0.0.UPF
> O 15.999 O.pbe-n-kjpaw_psl.1.0.0.UPF
> H 1.008 H.pbe-n-kjpaw_psl.1.0.0.UPF
>
> K_POINTS automatic
> 6 6 1  0 0 0
>
> CELL_PARAMETERS angstrom
> 7.42718991274627 0.00000000000000 0.00000000000000
> 3.71359495637314 6.43213514316980 0.00000000000000
> 0.00000000000000 0.00000000000000 29.000000000003126
>
> ATOMIC_POSITIONS angstrom
> Pt 2.5641489000 0.4594382200 8.0000000000  0 0 0
> Pt 5.2167167200 1.3783146700 8.0000000000  0 0 0
> ...
> Pt 4.1556895900 0.1531460700 17.1683222700
> Pt 6.8082574200 1.0720225200 17.1683222700
> Pt 2.0336353300 1.9908989700 17.1683222700
> ...
> H 0.3060672802 1.6388023741 19.1376040000
> H 2.7167357186 1.0284278006 19.1376040000
> H 4.4562891110 2.8078620154 19.1376040000
>
> ___________________________________
>
> Thank you in advance for your assistance.
>
> Kind regards,
> Gabriel Bramley
> PhD. Candidate
> University of Southampton
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users


More information about the users mailing list