3.6 Adsorption energies with consistent references, coverage and dispersion
These are unexecuted teaching inputs and starting settings to test. Original figures are schematics, not computed results. Use licensed VASP and PAW data, replace every placeholder, record the executable version and validate convergence.
3.6.1 Model, units and provenance
Use eV for energy, Å for length and eV/Å for force; 1 kbar = 0.1 GPa. State normalization per atom, molecule, primitive cell or simulation cell. Record PAW identifiers, release, ZVAL, ENMAX and permitted hashes; never redistribute POTCAR. SCF convergence addresses the chosen electronic problem; convergence of the target property requires separate tests.
Original schematic. Curves explain concepts; blank data areas await verified learner results. No calculation is claimed.
3.6.2 Worked case: procedure, interpretation and checks
Intuition and prerequisites
Subtracting three large numbers can produce a chemically important small difference. The subtraction is meaningful only when its terms refer to compatible models and a clearly balanced process. A converged slab alone does not guarantee a converged adsorption energy; molecule reference, spin, surface relaxation, coverage and dispersion can all change the answer.
This case assumes that slab thickness, vacuum and a baseline lateral cell have already been studied. It extends that work to a matched energy cycle. Choose a neutral molecular adsorption example first. Charged adsorption, electrochemical potentials and solvent chemical potentials need additional formalism and are not silently covered by the neutral-vacuum formula.
For n intact adsorbates A on one model surface, define
\(E_{\mathrm{ads}/A}=\frac{E(\mathrm{slab}+nA)-E(\mathrm{clean\ slab})-nE(A_{\mathrm{ref}})}{n}\).
With this sign convention, a negative value means the final relaxed state is lower in electronic energy than the stated separated references. If adsorption is dissociative, rewrite the balanced reference reaction explicitly. Do not insert atomic energies into a molecular-reference expression and call the result the same adsorption energy. The official surface workshop provides adsorption as a practical teaching context. VASP surface workshop
Build the energy cycle
- State the reaction and coverage. Define θ using a documented site count or surface area. “One molecule” is not a coverage unless the cell is specified. For symmetric adsorption on both faces, count the total adsorbates and any factor of two explicitly.
- Create matched directories. Use
clean_slab,adsorbed_state, andmolecule_reference, plus a manifest of shared and intentionally different settings. Use the same lateral cell and slab constraint scheme for clean and covered slabs. - Relax candidate adsorption states. Explore chemically plausible sites and orientations; a single initial geometry finds at most a local basin. Keep the relaxation criteria comparable. For magnetic substrates, verify converged magnetic states and avoid comparing unrelated electronic metastable solutions accidentally.
- Validate molecular reference. Use a large enough isolated-molecule cell, suitable sampling, correct spin and geometry. Converge image interactions. Identical numerical k meshes are not mandatory across molecule and slab; each needs a justified converged integration strategy while the physical Hamiltonian and PAW reference remain consistent.
- Perform matched final static energies. Use the same energy convention and converged precision. Check smearing effects for the metallic slab and consistent extraction of the energy used in the subtraction. Never mix an F from one run with an E0 from another by accident.
- Evaluate dispersion consistently. If a selected correction is part of the method, include it in every relevant reference calculation and relaxation. A dispersion addition can change geometry and energy; comparing only a post hoc energy term does not fully compare optimized methods.
- Study coverage and lateral size. Repeat with larger lateral cells or explicitly different coverages. Periodic adsorbate interactions may be the physics of an ordered overlayer or an unwanted approximation to isolated adsorption; state which is intended.
- Propagate uncertainty honestly. Numerical convergence, competing minima and method dependence are different uncertainty sources. A tiny SCF tolerance does not erase functional uncertainty.
IVDW=12 selects the D3(BJ) option; it is not a universal endorsement for every metal surface or adsorbate. Screened dispersion, nonlocal functionals and other models have distinct assumptions. Check the functional and any required parameters rather than adding two dispersion treatments accidentally. IVDW
Dipole corrections require the correct surface-normal direction and sensible geometry. They do not substitute for vacuum convergence, and charged slabs need additional care. A molecule reference may require a different appropriate electrostatic treatment from an asymmetric slab. Electrostatic corrections
Separate interaction and deformation
For interpretation, evaluate fragments frozen in the adsorbed geometry as an optional additional energy cycle. Then decompose the adsorption change into interaction between deformed fragments and the cost of deforming clean fragments. Keep electronic states and conventions compatible. This can explain why stronger local attraction does not always yield more favourable total adsorption.
Do not call the resulting electronic adsorption energy a finite-temperature free energy. Zero-point energy, vibrational and configurational entropy, gas-phase chemical potentials, solvent and pressure effects require additional, stated approximations. Nor is adsorption energy a kinetic barrier: strong binding does not determine the pathway or rate.
Sanity checks: atom balance correct; total adsorbate count correct; slab area and coverage stated; same functional and potentials; same fixed-layer scheme; molecule spin checked; no accidental changes of magnetic state; converged final static energies; consistent dispersion; every reported value traceable to three or more source runs.
Traps: comparing the lowest adsorbed energy with a nonrelaxed clean slab; using different lateral cell strain in references; forgetting adsorbates on the second face; interpreting a coverage series as pure numerical convergence; confusing per molecule, per atom and per area; presenting a DFT electronic number as a desorption temperature.
Exercise. Prepare a blank adsorption-energy ledger with columns for run ID, atom count, area, coverage, method, spin, energy field, final energy, and source file. Derive the balanced formula before inserting your own calculated values. Then identify one additional calculation needed to separate lateral interactions from slab-size error.
3.6.3 Unexecuted inputs and analysis scaffolds
These are unexecuted teaching inputs and starting settings to test. Original figures are schematics, not computed results. Use licensed VASP and PAW data, replace every placeholder, record the executable version and validate convergence.
3.6.3.1 Input block 1
# UNEXECUTED final-static delta for a previously justified neutral slab model
IBRION = -1
NSW = 0
EDIFF = 1E-7
PREC = Accurate
# Example method choice ONLY if PBE+D3(BJ) is the intended validated model:
IVDW = 12
# Example asymmetric slab, normal to third lattice direction:
LDIPOL = .TRUE.
IDIPOL = 3
# Set DIPOL according to your actual slab geometry; do not copy a guessed centre.
# Retain validated ENCUT, k mesh, spin, smearing and PAW choices.
3.6.4 Related learning paths
- 1.1 Four input files, one physical question
- 1.2 A convergence laboratory with an error budget
- 8.9 Train, validate and deploy VASP-native machine-learned force fields
- 1.3 Reliable HPC execution, restart validation and reproducible automation
- 3.4 Slabs, adsorption, dipoles and work functions