Mat Raman Spectra

Raman Spectra Calculation

<!-- mcp-tools-note -->

[!NOTE] Steps written server.tool are MCP tool calls: mace.load_model is the load_model tool of the mace server (mcp__mace__load_model, or mcp__plugin_atomistic-skills_mace__load_model when installed as a plugin). Without a connected server, run the same tools from the shell. Tools named in one command share a process, so a model loaded by load_model stays loaded:

${CLAUDE_SKILL_DIR}/../../venv/run mlip python -m src.mcp_server.cli mace load_model key=value relax_structure key=value
${CLAUDE_SKILL_DIR}/../../venv/run cpu python -m src.mcp_server.cli atomate2 run_atomate2_vasp_calculation key=value get_atomate2_job_status key=value

Goal

To calculate the Raman spectrum of a crystalline material by:

1. Identifying Raman-active phonon modes via group-theory symmetry analysis of Γ-point phonons (MLIP-only path)
2. Optionally computing full Raman scattering intensities using DFT-level Born effective charges and dielectric tensors (DFT path)

Physical background: Raman scattering intensity of mode $\nu$ is proportional to $|(\hat{e}i \cdot \alpha\nu \cdot \hat{e}s)|^2$, where $\alpha\nu = d\boldsymbol{\alpha}/dQ_\nu$ is the Raman tensor (derivative of polarizability with respect to mode coordinate $Q_\nu$). Computing $\alpha_\nu$ requires DFT-level dielectric calculations; phonon frequencies alone can be obtained with MLIPs.

[!IMPORTANT] Two-tier approach:

• MLIP tier (Step 1–3): Predicts Raman peak positions (frequencies). Intensities are set equal as a placeholder. Sufficient for peak assignment and comparison with experiment when known structure is available.
• DFT tier (Step 4): Computes actual Raman intensities via Born charges and dielectric tensor from VASP DFPT. Required for quantitative intensity matching.

Prerequisites

• Pymatgen, phonopy installed in cpu (for symmetry analysis and plotting)
• MLIP environment (mlip for MACE or MatGL, or fairchem) for phonon calculation

Instructions

1. Relax Structure

Before computing phonons, ensure the structure is fully relaxed. Use the MCP tool for your chosen MLIP:

mace.load_model(model_name="MACE-MH-1")
mace.relax_structure(
    structure_data="input_structure.cif",
    relax_cell=True,
    fmax=0.001,       # tight convergence for phonons
    output_dir="relaxation/"
)

[!IMPORTANT] Use a tight force convergence (fmax ≤ 0.001 eV/Å) for phonon calculations. Poorly relaxed structures produce imaginary frequencies that indicate a false structural instability.

2. Calculate Phonons (MLIP)

Use the mat-phonon skill to compute Γ-point phonons. The output phonon.yaml is the required input for this skill.

${CLAUDE_SKILL_DIR}/../../venv/run mlip python ${CLAUDE_SKILL_DIR}/../mat-phonon/scripts/calculate_phonon.py \
    --structure relaxation/relaxed_structure.cif \
    --model_type mace \
    --model_name MACE-MH-1 \
    --supercell_matrix '[[2,0,0],[0,2,0],[0,0,2]]' \
    --output_dir phonon_results/

Verify the output: check phonon_results/phonon.yaml exists and there are no large imaginary modes at Γ.

3. Analyse Raman-Active Modes and Simulate Spectrum (MLIP Tier)

${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/analyze_raman_modes.py \
    --phonon-yaml phonon_results/phonon.yaml \
    --structure relaxation/relaxed_structure.cif \
    --output-dir raman_results/ \
    --broadening 5.0 \
    --freq-max 1000

Parameters:

Output files:

• raman_modes.json — table of all Γ-point modes with symmetry label, frequency, and Raman-activity classification
• raman_spectrum.png/svg — simulated spectrum (MLIP tier uses equal intensities; DFT tier uses computed intensities)
• raman_modes_table.csv — CSV table for further analysis

[!NOTE] On Raman intensity accuracy: The MLIP tier assigns equal intensity to all Raman-active modes. Peak positions are reliable; relative intensities are not. For quantitative comparison with experiment, proceed to Step 4.