Skip to content

3.2 Aluminum: why metals need smearing care

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.

3.2.1 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.

 Aluminum: why metals need smearing care — original conceptual schematic

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

3.2.2 Worked case: procedure, interpretation and checks

Goal and intuition

Silicon has a gap; a metal has states crossing the Fermi level. Small changes in sampling can change which states are occupied. Smearing smooths the numerical integration, but a convenient smooth calculation is not necessarily the zero-temperature answer you want. Aluminum is a relatively simple first metallic case because its fcc structure and nonmagnetic starting model keep the focus on occupations and k sampling.

The target is a two-dimensional convergence study in k-grid density and smearing width. Prerequisites: 3.1 Silicon: from a finite object to a crystal and an understanding of energy normalization. Treat σ as a numerical width unless you deliberately choose and justify a physical electronic-temperature model; Methfessel–Paxton width should not be casually relabeled sample temperature.

Starter model and workflow

4.05 Å is a trial conventional lattice parameter, not an output.

  1. Hold geometry and cutoff fixed. Run Γ-centered meshes such as 8³, 12³, 16³ and 20³, extending as necessary.

What the outputs teach

The entropy term is a useful warning signal for a smearing approximation. Small entropy alone does not prove the k mesh is adequate; a mesh can undersample a Fermi surface even when a chosen width yields a small correction. Watch the joint dependence of energy, stress and iteration behavior. For a metal relaxation, the forces are tied to the variational smeared functional; do not assume they are derivatives of whichever extrapolated energy field you happen to plot.

Energy convergence may oscillate instead of approaching monotonically. That is not automatically a failed program. Changes in the relation between mesh points and the Fermi surface can produce nonmonotonic integration errors. Keep mesh centering consistent, record the full mesh and the irreducible count, and compare multiple adjacent refinements.

Pitfalls, limits and exercise

Avoid applying ISMEAR>0 indiscriminately to molecular and insulating cases. Avoid “σ→0 at any cost” on a coarse mesh: reducing width without adding k points can make integration and SCF behavior worse. If comparing surface and bulk energies later, test errors in the difference and use compatible numerical accuracy for both. Exercise: submit a k-grid-by-σ heat map of an explicitly labeled energy difference relative to your finest validated point, with no interpolation across unconverged runs. Explain which region is stable and why one fast run is insufficient evidence.

3.2.3 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.

3.2.3.1 Input block 1

Al fcc primitive trial cell
4.05
0.0 0.5 0.5
0.5 0.0 0.5
0.5 0.5 0.0
Al
1
Direct
0.0 0.0 0.0

3.2.3.2 Input block 2

SYSTEM = Al metal sampling study
ENCUT = <ECUT_EV>
PREC = Accurate
EDIFF = 1E-7
LREAL = .FALSE.
ISPIN = 1
ISMEAR = 1
SIGMA = 0.15
IBRION = -1
NSW = 0

3.2.5 Technical sources