8.8 Finite-temperature surfaces and liquids: sampling before storytelling

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.

8.8.1 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.

Finite-temperature surfaces and liquids: sampling before storytelling — original conceptual schematic

Original schematic. Curves explain concepts; blank data areas await verified learner results. No calculation is claimed.

8.8.2 Worked case: procedure, interpretation and checks

Intuition

At finite temperature, the system visits a distribution of adsorption geometries, orientations, coordination environments and solvent arrangements. A dramatic event in one short movie is a hypothesis generator. It is not automatically a rate, an equilibrium population or a free-energy barrier. A useful question is narrow: “How does the distribution of adsorbate height change between two conditions?” is more directly testable than “What is the entire reaction mechanism?”

Prerequisites. Use a previously converged slab or liquid model. This lesson does not rebuild the basic slab. Recheck coverage, lateral cell size, vacuum, dipole treatment, fixed layers and electronic settings at the intended thermal state. A solvent interface also needs a physically justified amount of solvent, accessible volume, composition and boundary model. Report whether the model represents a closed fixed-composition cell, a vacuum interface, or a more elaborate reservoir approximation.

A practical finite-temperature surface study

Choose two or more physically distinct initial adsorbate/solvent arrangements. Random velocity seeds alone may not cross a slow configurational barrier.
Keep the slab lattice fixed at a justified value for the substrate and intended temperature; explain whether it is the 0 K relaxed lattice, experimental lattice or a thermally expanded estimate.
Decide which substrate layers are mobile. Frozen layers alter heat capacity, phonons and energy dissipation; they are an approximation, not inert bookkeeping. Test their influence on the observable.
Equilibrate each arrangement under the same model. Apply conservative timestep and SCF checks, especially when protons, strong bonds or close solvent contacts are present.
Record observables that connect directly to the question: adsorbate height relative to a moving substrate reference plane, bond lengths, orientation angle, coordination, solvent density along z, and local residence indicators.
Define analysis before inspecting the most dramatic frames. Set state boundaries with hysteresis or persistence criteria when needed, and test sensitivity to those definitions.
Compare distributions and time blocks across independent starts. If replicas remain in distinct long-lived states, report metastability and incomplete mixing; do not simply pool them with equal weights and call the result equilibrium.
Extract selected snapshots for high-accuracy static checks. Ensure their electronic model is consistent or explicitly label a higher-level correction. A mean of electronic energies alone is not an adsorption free energy.

This is a fixed-cell illustration, not a recommendation to thermostat all species identically in every physical surface-dynamics study. If investigating dissipation or collision dynamics, the thermostat placement itself becomes part of the physical model and must be validated. The official transition-state tutorial demonstrates that dynamic approaches and reaction-coordinate monitoring are distinct workflows; rare-event free energies need dedicated methodology beyond an unbiased short trajectory. VASP dynamic approaches tutorial

Liquids: density, mixing and rare events

For a liquid, a relaxed starting box can retain the memory of packing. Compare multiple initial packings, not just velocities. A structural average can converge before a slow conformational or compositional rearrangement. If aiming at a bulk density, use a justified pressure treatment; if using fixed density, report its origin. Classical nuclei, finite cell size and exchange-correlation approximation may each bias water or hydrogen-bond networks; more trajectory time cannot eliminate a systematic Hamiltonian error.

A reaction observed once gives neither a stable event rate nor a guaranteed dominant pathway. Conversely, no reaction during an AIMD run can simply mean the barrier is too high for the accessible timescale. If enhanced sampling is needed, define a collective variable and validate reweighting or free-energy integration separately; never present a biased-time trajectory as an unbiased kinetic measurement.

Sanity checks: no artificial vacuum collapse; no unintended drift of the substrate reference plane; state populations stable by block if equilibrium is claimed; comparable results from different arrangements; constrained degrees of freedom documented; close contacts checked; temperature means and fluctuations sensible for active degrees of freedom.

Traps: measuring adsorbate height against the fixed origin while the slab drifts; claiming desorption energy from one departure; interpreting a constrained lower slab as a fully flexible substrate; averaging incompatible metastable replicas; treating hot-trajectory discovery as room-temperature kinetics.

Exercise. Design a study of a surface-bound molecule’s orientation distribution. Specify angle definition, substrate reference, starting orientations, independent replicas, burn-in and a criterion for incomplete mixing. Explain what additional work would be needed to turn a population ratio into a free-energy difference.

8.8.3 Unexecuted inputs and analysis scaffolds