hello-qm/catgo-lrg/dos-analysis

DOS and Electronic Structure Analysis

Overview

Density of states (DOS) analysis extracts electronic structure information from completed DFT calculations. Key quantities:

• Total DOS: overall electronic structure, band gap identification
• PDOS: orbital-resolved contributions from specific atoms
• d-band center: catalytic activity descriptor (higher = stronger binding)
• Spin-resolved DOS: magnetic ordering, spin polarization

MCP Tool: catgo_analyze

Total DOS

{"tool": "catgo_analyze", "arguments": {
  "action": "dos",
  "workflow_id": "wf_abc",
  "task_id": "task_sp",
  "dos_type": "total"
}}

Projected DOS (PDOS)

{"tool": "catgo_analyze", "arguments": {
  "action": "dos",
  "workflow_id": "wf_abc",
  "task_id": "task_sp",
  "dos_type": "projected",
  "atom_indices": [0, 1, 2, 3],
  "orbitals": ["d"]
}}

d-Band Center

{"tool": "catgo_analyze", "arguments": {
  "action": "dos",
  "workflow_id": "wf_abc",
  "task_id": "task_sp",
  "dos_type": "dband",
  "atom_indices": [0, 1, 2, 3]
}}

Returns:

• d_band_center: energy relative to Fermi level (eV)
• d_band_width: standard deviation of d-band (eV)
• d_band_filling: fraction of d-band occupied (0-1)

Workflow for DOS Analysis

DOS requires a completed single_point or geo_opt calculation with appropriate VASP settings.

VASP Settings for DOS

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "add_task", "workflow_id": "wf_dos",
  "task_type": "single_point",
  "params": {
    "software": "vasp",
    "ENCUT": 520,
    "ISMEAR": -5,
    "NEDOS": 3001,
    "LORBIT": 11,
    "system_name": "DOS calculation"
  }
}}

Key VASP parameters:

• ISMEAR = -5: tetrahedron method with Blochl corrections (accurate DOS)
• NEDOS = 3001: number of DOS grid points (default 301 is too coarse)
• LORBIT = 11: write projected DOS (DOSCAR with atom/orbital decomposition)

Two-Step Pattern: Relax then DOS

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "add_task", "workflow_id": "wf_dos",
  "task_type": "geo_opt",
  "params": {"software": "vasp", "ENCUT": 520, "system_name": "relax"}
}}

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "add_task", "workflow_id": "wf_dos",
  "task_type": "single_point",
  "depends_on": "task_relax",
  "params": {
    "software": "vasp", "ENCUT": 520,
    "ISMEAR": -5, "NEDOS": 3001, "LORBIT": 11,
    "system_name": "DOS"
  }
}}

Python API

from catgo.workflow import Workflow
wf = Workflow("DOS analysis - Pt(111)")
inp = wf.add_task("structure_input", structure=pt_slab_json)
# Step 1: Geometry optimization
opt = wf.add_task("geo_opt",
    structure=inp.output.structure,
    software="vasp", ENCUT=520)
# Step 2: DOS single-point on relaxed structure
dos_sp = wf.add_task("single_point",
    structure=opt.output.structure,
    software="vasp", ENCUT=520,
    ISMEAR=-5, NEDOS=3001, LORBIT=11)
# Step 3: Post-process DOS
dos = wf.add_task("dos_analysis",
    doscar=dos_sp.output.doscar,
    atom_indices=[0, 1, 2, 3],
    orbitals=["d"],
    compute_dband=True)
wf.submit()

d-Band Center Theory

The d-band model (Hammer-Norskov) relates catalytic activity to the d-band center position relative to the Fermi level:

epsilon_d = integral(E * rho_d(E) dE) / integral(rho_d(E) dE)

Integrated over occupied states (up to Fermi level).

Surface vs Bulk d-Band

Surface atoms have narrower d-bands (fewer neighbors) and higher d-band centers than bulk atoms. Always select surface atom indices for catalysis analysis.

Spin-Resolved DOS

For magnetic systems (Fe, Co, Ni, oxides), enable spin polarization:

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "add_task", "workflow_id": "wf_dos",
  "task_type": "single_point",
  "params": {
    "software": "vasp", "ENCUT": 520,
    "ISPIN": 2, "ISMEAR": -5, "NEDOS": 3001, "LORBIT": 11,
    "system_name": "spin-DOS"
  }
}}

Spin-resolved DOS returns separate up/down channels. The magnetic moment per atom equals the integral of (rho_up - rho_down) up to E_F.

Orbital Channels

Available orbital projections for PDOS:

Specific sub-orbitals: "dz2", "dx2-y2", "dxy", "dxz", "dyz"

Common Pitfalls

1. Never use ISMEAR=1 (Methfessel-Paxton) for DOS -- it produces negative

DOS artifacts. Use ISMEAR=-5 (tetrahedron) for static DOS calculations.

1. NEDOS=301 (VASP default) gives very coarse DOS. Use at least 2001-3001.
2. LORBIT=11 is required for PDOS. Without it, only total DOS is available.
3. Always do DOS as a separate single_point after geo_opt. The DOS from

a relaxation run uses the smearing from NSW>0 and is unreliable.

1. For d-band center, select only surface layer atoms. Including bulk atoms

averages out the surface electronic signature.

1. Band gap from DOS can be noisy -- compare with the band structure if

precise gap values are needed.