bader-charge-analysis

Bader Charge Analysis

Overview

Bader analysis partitions the continuous electron density from DFT into atomic basins defined by zero-flux surfaces of the density gradient. This gives physically meaningful atomic charges and charge transfer values.

What Bader Charges Tell You

• Charge transfer between adsorbate and surface
• Oxidation states of atoms in a material
• Electron donation/back-donation in catalytic bonds
• Ionic vs covalent character of bonds

VASP Settings for Bader Analysis

Bader analysis requires fine-grid charge density output:

LAECHG = .TRUE.    # Write core charge density (AECCAR0, AECCAR2)
LCHARG = .TRUE.    # Write valence charge density (CHGCAR)
NGXF, NGYF, NGZF   # Fine FFT grid (2x default, e.g., NGXF=2*NGX)

The all-electron charge density is: AECCAR0 + AECCAR2, which is summed with the Bader code to avoid errors from pseudopotential smoothing.

MCP Workflow

Step 1: Single-point with charge output

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "add_task", "workflow_id": "wf_bader",
  "task_type": "single_point",
  "params": {
    "software": "vasp",
    "ENCUT": 520,
    "LAECHG": true,
    "LCHARG": true,
    "PREC": "Accurate",
    "system_name": "charge density"
  }
}}

Step 2: Bader analysis post-processing

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "add_task", "workflow_id": "wf_bader",
  "task_type": "charge_analysis",
  "depends_on": "task_sp",
  "params": {"method": "bader", "system_name": "Bader charges"}
}}

Step 3: Get results

{"tool": "catgo_workflow_engine", "arguments": {
  "action": "get_result", "workflow_id": "wf_bader", "task_id": "task_bader"
}}

{"tool": "catgo_analyze", "arguments": {
  "action": "charges", "workflow_id": "wf_bader", "task_id": "task_bader"
}}

Python API

Basic Bader Analysis

from catgo.workflow import Workflow

wf = Workflow("Bader charge - CO on Pt(111)")

inp = wf.add_task("structure_input", structure=co_pt_json)

# Relax first
opt = wf.add_task("geo_opt",
    structure=inp.output.structure,
    software="vasp", ENCUT=520)

# Single-point with charge density output
sp = wf.add_task("single_point",
    structure=opt.output.structure,
    software="vasp", ENCUT=520,
    LAECHG=True, LCHARG=True, PREC="Accurate")

# Bader post-processing
bader = wf.add_task("charge_analysis",
    chgcar=sp.output.chgcar,
    aeccar0=sp.output.aeccar0,
    aeccar2=sp.output.aeccar2,
    method="bader")

wf.submit()

Charge Transfer Analysis

# Compare charges before and after adsorption
wf = Workflow("Charge transfer analysis")

# Clean slab Bader
slab_sp = wf.add_task("single_point",
    structure=slab_opt.output.structure,
    software="vasp", ENCUT=520, LAECHG=True, LCHARG=True)
slab_bader = wf.add_task("charge_analysis",
    chgcar=slab_sp.output.chgcar,
    aeccar0=slab_sp.output.aeccar0,
    aeccar2=slab_sp.output.aeccar2)

# Slab+adsorbate Bader
ads_sp = wf.add_task("single_point",
    structure=ads_opt.output.structure,
    software="vasp", ENCUT=520, LAECHG=True, LCHARG=True)
ads_bader = wf.add_task("charge_analysis",
    chgcar=ads_sp.output.chgcar,
    aeccar0=ads_sp.output.aeccar0,
    aeccar2=ads_sp.output.aeccar2)

wf.submit()

# After completion:
# dq = q_ads(atom) - q_clean(atom) for each surface atom
# Positive dq = atom lost electrons; Negative dq = atom gained electrons

DAG Structure

structure --> geo_opt --> single_point(LAECHG) --> charge_analysis

Output Format

Bader analysis returns per-atom data:

Interpreting Results

Common Reference Charges (VASP PAW, valence electrons)

Charge Transfer Upon Adsorption

dq_adsorbate = sum(net_charge of adsorbate atoms in slab+ads system)
             - sum(net_charge of same atoms in isolated adsorbate)

• dq < 0: adsorbate gains electrons (acceptor, e.g., CO on Pt)
• dq > 0: adsorbate loses electrons (donor, e.g., Na on surface)

Common Pitfalls

1. Always use LAECHG=.TRUE. to get all-electron charge density. Bader analysis on pseudocharge (CHGCAR alone) gives wrong atomic charges because core electrons are missing.
2. PREC=Accurate and a fine FFT grid improve Bader basin boundaries. Coarse grids can misassign charge near atomic boundaries.
3. Bader charges are NOT formal oxidation states. They are typically smaller in magnitude (e.g., Ti in TiO2 shows +2.3, not +4).
4. For charge transfer analysis, use the SAME computational settings for the reference and adsorbed systems.
5. The Bader program (Henkelman group) must be available on the HPC. CatGo calls it automatically during charge_analysis post-processing.
6. For spin-polarized systems, Bader can also partition spin density -- this gives magnetic moments per atom.