Band And Dos Analysis

Allowed tools

• generate_kpath
• vasp_prepare
• vasp_band_prepare
• vasp_execute_batch
• analyze_vasp_results
• render_structure_views

Workflow

1. Accept the relaxed bulk first

• Do not start DOS or band work from a bulk structure that has not passed the base convergence screen.
• Do not treat the DOS written during an ionic relaxation as the final DOS artifact; VASP Wiki recommends a separate static/DOS stage for the converged structure.

2. Build DOS as a dedicated post-relax stage

• DOS should be prepared as a separate static-like branch after the structure is finalized.
• For more accurate DOS, increase the k-point density relative to the relaxation/static baseline when that is the stated goal.
• If you want a non-self-consistent DOS (ICHARG=11), the preceding self-consistent stage must have produced a reusable CHGCAR.
• State whether the DOS branch is self-consistent or fixed-charge-density.
• Band-path generation is a separate primitive: generate_kpath proposes the line-mode KPOINTS, and vasp_band_prepare owns the dedicated band input deck.

3. Choose the DOS occupation method explicitly

• For bulk DOS and very accurate static energies, VASP Wiki recommends tetrahedron integration with Blöchl corrections (ISMEAR=-5) on a Gamma-centered mesh that can actually form tetrahedra.
• Do not blindly keep the tetrahedron starter for slab, gas, line-mode, or otherwise low-dimensional / too-sparse meshes. In those cases use user_incar_patch to switch to Gaussian smearing (ISMEAR=0) with small SIGMA, typically 0.03-0.1.
• When the system class or gap status is uncertain, prefer the Gaussian fallback over Methfessel-Paxton. For clearly gapped or unknown-gap cases, EFERMI=MIDGAP is a reasonable explicit stabilizer to surface.

4. Keep projection and fixed-density caveats visible

• LORBIT controls PDOS output, but fixed-density reuse also needs the charge-density provenance to be scientifically traceable.
• For ICHARG=11/12, VASP Wiki recommends LMAXMIX matched to the highest angular momentum in the PAW data (2/4/6 for s,p / d / f dominated cases). The tool now auto-fills a baseline when absent; surface any deliberate override.

5. Keep provenance visible

• Report the exact KPOINTS artifact used for the band job.
• Report the DOS smearing choice, whether the run is self-consistent or ICHARG=11, and the source of any reused CHGCAR.
• If the upstream static stage was created specifically to feed DOS/bands, say so and keep the CHGCAR-writing choice explicit.
• Use render_structure_views when the bulk cell standardization or orientation is visually ambiguous.

Method-critical defaults

• Do not compare DOS or bandgaps across jobs that changed spin, DFT+U, smearing, or k-point density silently.
• For DOS, VASP Wiki recommends a dedicated post-relax static branch; do not interpret relax-stage DOSCAR as the final electronic-structure result.
• Keep the tetrahedron default for the cases where it is justified; do not quietly apply it to slab/gas or tetrahedron-ineligible meshes.
• When the electronic character is uncertain, prefer ISMEAR=0 with small SIGMA over ISMEAR>0.
• Surface any user_incar_patch changes that alter DOS/band interpretation.
• For DOS reuse with CHGCAR, report the charge-density source explicitly and make sure the upstream stage actually wrote that CHGCAR.
• For line-mode band runs, keep the generated KPOINTS artifact and any reused CHGCAR source explicit.

Output Contract

Return:

• DOS or band job root
• generated KPOINTS path when relevant
• DOS mode and smearing choice when DOS is in scope
• VASP analysis summary path
• any parsed bandgap value or convergence caveat

References

• Use bulk-relax-and-reference to establish the accepted bulk baseline before this skill.