[CP2K-user] [CP2K:22373] Question about charge settings and binding energy calculations for oleate/carboxylate adsorption on CsPbBr3 slabs in CP2K

[Thomas Kühne]

Dear Haotian,

the key point is that "remove H but keep CHARGE 0" does not represent RCOO-.

Removing a neutral H atom while keeping the remaining system neutral produces
a neutral carboxyl radical or a different redox state. If the resulting
electron count is odd, it also requires UKS and an explicit multiplicity. Its
strong distortion is therefore not surprising.

Setting the total CHARGE to -1 gives the intended electron count for an
isolated slab+RCOO- system, but it also creates a net-charged periodically
repeated slab. Its electrostatic energy is not a vacuum-independent adsorption
energy unless a physical compensating charge, counterion or electrode model is
specified. The excess electron may also occupy a CsPbBr3 conduction state
instead of remaining localized on the ligand.

Changing SCF mixing settings cannot remove this physical ambiguity.

For the chemistry described in the question, I would first use a neutral,
atom- and charge-balanced dissociative adsorption reaction:

  slab + RCOOH -> slab-RCOO + H_surface

and calculate

  E_ads^diss =
      E(slab-RCOO + H_surface)
    - E(slab)
    - E(RCOOH)

with CHARGE 0 for all three calculations.

Place the hydrogen on the chemically plausible surface site or sites and
compare the alternatives. If the experimental process is ligand exchange,
construct the corresponding balanced exchange reaction instead. If a
counterion, defect or electrolyte provides charge compensation experimentally,
it should be included explicitly.

For a genuinely two-dimensionally periodic neutral slab, CELL and POISSON
periodicity must agree:

&DFT
  CHARGE 0
  &POISSON
    PERIODIC XY
    POISSON_SOLVER ANALYTIC
  &END POISSON
END DFT

&CELL
  ...
  PERIODIC XY
&END CELL

Alternatively, use three-dimensional periodicity with sufficient vacuum and a
dipole correction for a neutral asymmetric slab:

&DFT
  CHARGE 0
  SURFACE_DIPOLE_CORRECTION TRUE
  SURF_DIP_DIR Z
  &POISSON
    PERIODIC XYZ
    POISSON_SOLVER PERIODIC
  &END POISSON
&END DFT

&CELL
  ...
  PERIODIC XYZ
&END CELL

The dipole correction does not make the energy of a net-charged slab
well-defined.

An expression such as

  E(slab+RCOO-) - E(slab) - E(RCOO-)

is only meaningful after choosing and consistently applying the same physical
compensation model, cell, boundary conditions and finite-size treatment.

Newer CP2K versions provide PLANAR_COUNTER_CHARGE for explicit
counterelectrode models in three-dimensionally periodic charged-slab or GCE
calculations. This is a physical countercharge model, not a universal
numerical correction for arbitrary charged adsorption calculations.

Before comparing ligands, converge the lateral cell and coverage, slab
thickness, vacuum, cutoff and REL_CUTOFF, k-points and dispersion treatment.
Inspect the spin density, charge density and frontier orbitals to determine
where the extra electron or proton is located.

For adsorption from solution, solvent, counterions, proton chemical potential
or pH, and entropic contributions are additional parts of the adsorption free
energy.

I also checked the electrostatics numerically with CP2K. A neutral 2D test is
unchanged when the vacuum is increased, whereas the energy of an otherwise
equivalent charged 2D test changes by about 2.65 eV when the vacuum is
increased from 15 to 30 angstrom. This is why the very large charged-slab
binding energy is not reliable as currently defined.

Best regards,
Thomas

> Am 07.07.2026 um 10:20 schrieb '舒夜' via cp2k <cp2k at googlegroups.com>:
> 
> Dear all,
> 
> I am a beginner with CP2K. Recently, I have been trying to calculate the adsorption/binding energy of carboxylate ligands on a CsPbBr3 perovskite slab surface. Specifically, I would like to model the interaction between deprotonated carboxylate ligands, such as oleate (OA−) or benzoate (BA−), and surface Pb sites, and compare the relative binding strengths of different ligands.
> 
> The main issue I am facing is how to properly treat the charge of the deprotonated ligand in CP2K, and how to define a reliable binding energy.
> 
> So far, I have tested several models:
> 
> I removed the H atom from the carboxyl group of the acid molecule and placed the deprotonated ligand near a surface Pb site of the CsPbBr3 slab, while setting the total charge of the whole system to 0.
> The energy can converge in this case, but the ligand structure becomes significantly distorted after geometry optimization. Therefore, I am not sure whether this “H-removed but neutral total charge” model is physically meaningful.
> 
> I treated the deprotonated carboxylate ligand as an anion and set the total charge of the whole system to -1.
> In this case, the SCF/energy convergence shows strong oscillations, and the calculated binding energy becomes abnormally large. The ligand may also become distorted during optimization.
> 
> I also tried setting the slab periodicity to XY and modifying the Poisson/electrostatic boundary condition settings, but I am still not sure what the correct setup should be for a slab + charged adsorbate system.
> 
> I would like to ask the following questions:
> 
> What is the recommended way to model the adsorption of carboxylate ligands such as OA−/BA− on a CsPbBr3 slab surface in CP2K?
> 
> For a deprotonated carboxylate ligand, should I directly calculate a charged slab + ligand anion system? Or is it more appropriate to construct an overall neutral dissociative adsorption model, for example:
> 
> neutral slab + neutral acid → slab–carboxylate + H_surf
> 
> where the H atom from the carboxylic acid is transferred to a surface Br site, so that the whole system remains at CHARGE 0?
> 
> If a charged system such as slab + OA− is used, how should CHARGE, PERIODIC, POISSON, and possible dipole/electrostatic corrections be set in CP2K? Would the adsorption energy obtained from such a charged slab system be reliable?
> 
> For the binding energy, which reference state is more appropriate?
> 
> E_ads = E(slab + OA−) - E(slab) - E(OA−)
> 
> or
> 
> E_ads^diss = E(slab–OA− + H_surf) - E(slab) - E(neutral OA)
> 
> I would greatly appreciate any suggestions on the proper modeling strategy and binding energy definition.
> 
> Best regards,
> Haotian Chen
> 
> 
> -- 
> You received this message because you are subscribed to the Google Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com <mailto:cp2k+unsubscribe at googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/cp2k/964c77e8-3b8c-4748-a9b0-407e0e9a6accn%40googlegroups.com <https://groups.google.com/d/msgid/cp2k/964c77e8-3b8c-4748-a9b0-407e0e9a6accn%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/748B81EC-CCF9-455B-AEA9-DD50FA932EE5%40gmail.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260806/e953a986/attachment.htm>