interpret-raman-spectrum
について
このClaudeスキルは、ラマンスペクトルを分析し、分極率に基づく選択則を適用して分子振動を特定し、分子対称性を帰属します。補完的な赤外線データを統合して包括的な振動解析を行い、蛍光干渉などの実用的な問題にも対応します。開発者はこれを使用して、ラマン散乱データを体系的に解釈し、参照スペクトルとの照合を行うことができます。
クイックインストール
Claude Code
推奨npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/interpret-raman-spectrumこのコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします
ドキュメント
Interpret Raman Spectrum
Read Raman scattering spectra. Identify molecular vibrations. Apply selection rules complementary to IR absorption. Integrate Raman data with IR for comprehensive vibrational analysis.
When Use
- Analyze samples difficult for IR (aqueous solutions, sealed containers, remote sensing)
- Identify symmetric vibrations weak or inactive in IR
- Complement IR data using mutual exclusion principle for centrosymmetric molecules
- Characterize carbon materials (graphene, carbon nanotubes, diamond) via characteristic Raman bands
- Analyze inorganic compounds, minerals, crystalline phases where Raman often more informative than IR
- Non-destructive, in situ analysis (no sample prep required for many Raman measurements)
Inputs
- Required: Raman spectrum data (Raman shift in cm-1 vs. intensity)
- Required: Excitation laser wavelength (e.g., 532 nm, 633 nm, 785 nm, 1064 nm)
- Optional: IR spectrum of same sample for complementary analysis
- Optional: Polarization data (parallel and perpendicular spectra for depolarization ratios)
- Optional: Known molecular formula or compound class
- Optional: Sample physical state (solid, liquid, solution, gas, thin film)
Steps
Step 1: Assess Spectrum Quality and Identify Artifacts
Evaluate Raman spectrum for reliability before analyzing peaks:
- Laser wavelength and fluorescence: Fluorescence = most common interference in Raman. Makes broad intense background that obscures Raman peaks. Shorter-wavelength lasers (532 nm) excite more fluorescence. Longer-wavelength lasers (785 nm, 1064 nm) reduce it at cost of weaker Raman signal (intensity scales as lambda^-4)
- Signal-to-noise ratio: Raman peaks clearly distinguishable from noise? Weak Raman scatterers may need longer acquisition or higher laser power
- Cosmic ray spikes: Sharp, narrow spikes at random positions = cosmic ray artifacts, not Raman peaks. Appear in only one spectrum of time-averaged set. Remove by spike filters
- Baseline correction: Sloping or curved baseline (from fluorescence or thermal emission) should be subtracted before measuring peak positions and intensities
- Photodegradation: High laser power can damage or transform sample. Check for spectral changes between successive acquisitions at same spot. Reduce power if degradation observed
- Spectral range: Standard Raman spectra cover 100-4000 cm-1 Raman shift. Low-frequency cutoff depends on edge or notch filter used to block Rayleigh line. Note if any region truncated
Got: Spectrum quality assessed. Fluorescence level documented. Artifacts (cosmic rays, baseline drift) identified or corrected. Usable spectral range confirmed.
If fail: Fluorescence dominates spectrum (broad background >> Raman peaks)? Recommend re-measurement with longer-wavelength laser (785 or 1064 nm) or surface-enhanced Raman spectroscopy (SERS). Sample degrades? Reduce laser power or use rotating sample stage.
Step 2: Identify Raman-Active Modes and Apply Selection Rules
Determine which vibrations are Raman-active and how they complement IR data:
- Raman selection rule: Vibration Raman-active if it involves change in polarizability of molecule. Symmetric stretches (often change molecular volume) typically strong in Raman
- IR selection rule (for comparison): Vibration IR-active if involves change in dipole moment. Asymmetric stretches typically strong in IR
- Mutual exclusion principle: For molecules with center of inversion (centrosymmetric), no vibration can be both Raman-active and IR-active. Band appears in both spectra? Molecule lacks center of symmetry
- General complementarity: Even for non-centrosymmetric molecules, vibrations strong in Raman tend to be weak in IR, and vice versa. Complementarity makes combined Raman + IR dataset more informative than either alone
- Identify Raman-favored modes: Symmetric stretches (C-C, C=C, S-S, N=N), breathing modes of rings, stretches of homonuclear bonds (no dipole change, IR-inactive) typically strong in Raman
Got: Selection rules applied. Raman-active vs. IR-active modes distinguished. Mutual exclusion tested if molecule centrosymmetric.
If fail: Molecular symmetry unknown? Use combined Raman and IR data to infer it. Band appears in both spectra with comparable intensity? Molecule not centrosymmetric.
Step 3: Analyze Raman Shift Positions
Assign observed Raman bands to specific vibrational modes using characteristic frequencies:
- C-H stretching region (2800-3100 cm-1): Similar to IR, but Raman intensities differ. Aromatic and olefinic C-H (3000-3100 cm-1) often stronger in Raman than aliphatic C-H
- Triple bonds (2100-2260 cm-1): C triple-bond C symmetric stretch strong in Raman, often weak or absent in IR. C triple-bond N active in both
- Double bond stretches:
| Shift (cm-1) | Assignment | Raman Intensity |
|---|---|---|
| 1600--1680 | C=C stretch | Strong |
| 1650--1800 | C=O stretch | Medium (weaker than IR) |
| 1500--1600 | Aromatic C=C | Medium to strong |
- Aromatic ring modes:
| Shift (cm-1) | Assignment | Notes |
|---|---|---|
| 990--1010 | Ring breathing (monosubstituted) | Very strong, diagnostic |
| 1000 | Ring breathing (sym. trisubstituted) | Strong |
| 1580--1600 | Ring stretch | Medium |
| 3050--3070 | Aromatic C-H stretch | Medium |
- Other characteristic Raman bands:
| Shift (cm-1) | Assignment |
|---|---|
| 430--550 | S-S stretch (disulfide) |
| 570--705 | C-S stretch |
| 800--1100 | C-C skeletal stretch |
| 630--770 | C-Cl stretch |
| 500--680 | C-Br stretch |
| 200--400 | Metal-ligand stretch |
- Carbon materials: G band (~1580 cm-1, graphitic sp2) and D band (~1350 cm-1, defect/disorder) diagnostic for carbon allotropes. 2D band (~2700 cm-1) characterizes graphene layer count. Diamond shows sharp peak at 1332 cm-1
Got: All significant Raman bands assigned to vibrational modes with reference to characteristic frequency ranges.
If fail: Band cannot be assigned from tables above? Consult spectral databases (RRUFF for minerals, SDBS for organics). Unassigned bands may belong to combination modes, overtones, lattice vibrations in crystalline samples.
Step 4: Compare Raman with IR Data
Integrate two complementary vibrational techniques:
- Tabulate corresponding bands: Create comparison table listing each vibrational mode with Raman shift, IR frequency, relative intensity in each technique
- Identify modes observed in only one technique: Modes present in Raman but absent in IR (or vice versa) give symmetry info. Symmetric stretches of non-polar bonds (S-S, C=C in symmetric environments) appear only in Raman
- Resolve ambiguities: IR assignments tentative (e.g., overlapping C-O and C-N stretches in fingerprint region)? Check whether Raman gives clearer picture due to different relative intensities
- Functional group confirmation: Confirm IR-identified functional groups via Raman counterparts. Ester should show C=O in IR (~1735 cm-1) and C-O-C in Raman. Carboxylic acid should show broad O-H in IR and C=O in both techniques
- Assess overall consistency: Raman and IR data should be mutually consistent. Any contradictions (e.g., band assigned as symmetric stretch appearing strong in both spectra for allegedly centrosymmetric molecule) = error in assignment or symmetry assumption
Got: Unified vibrational analysis table combining Raman and IR data. Functional group assignments confirmed or refined by complementary info.
If fail: IR data unavailable? Raman spectrum alone still gives useful info, with reduced certainty. Note which assignments would benefit from IR confirmation.
Step 5: Evaluate Polarization Data and Document Results
Use depolarization ratios for symmetry assignment. Compile final analysis:
- Depolarization ratio (rho): rho = I_perpendicular / I_parallel, measured from polarized Raman experiments
- rho = 0 to 0.75: Polarized band (rho < 0.75). Totally symmetric vibrations (A-type) polarized
- rho = 0.75: Depolarized band. Non-totally-symmetric vibrations give rho = 0.75
- Symmetry assignment: Polarized bands must belong to totally symmetric irreducible representation of molecular point group. Helps distinguish between modes of different symmetry at similar frequencies
- Compile results: Assemble complete table of all observed Raman bands with:
- Raman shift (cm-1)
- Relative intensity (strong/medium/weak)
- Depolarization ratio (if measured)
- Assignment (vibrational mode)
- Corresponding IR band (if observed)
- Compare with reference spectra: Compound known? Compare observed Raman spectrum with published reference spectra (databases: RRUFF, SDBS, NIST). Agreement in peak positions within +/- 3 cm-1 and matching relative intensities confirms identity
- Report uncertainties: Flag any assignments tentative. Note which additional experiments (temperature-dependent Raman, resonance Raman, SERS) could resolve ambiguities
Got: Complete Raman analysis with all bands assigned. Polarization data interpreted for symmetry. Results integrated with IR and other spectroscopic data.
If fail: Polarization data unavailable? Symmetry assignment relies on frequency and intensity patterns alone. Note limitation. Recommend polarized measurements if symmetry info critical.
Checks
- Spectrum quality assessed (fluorescence, cosmic rays, baseline, photodegradation)
- Raman selection rules applied. Raman-active modes identified
- Mutual exclusion principle tested if molecule centrosymmetric
- All significant Raman bands assigned to vibrational modes
- Raman data compared and integrated with IR data where available
- Depolarization ratios interpreted for symmetry assignment (if polarization data available)
- Assignments consistent with known molecular structure or proposed structure from other techniques
- Results compared with reference spectra where possible
Pitfalls
- Fluorescence overwhelming Raman signal: Single most common problem. Switch to longer-wavelength laser or use time-gated detection. Do not try to interpret broad fluorescent humps as Raman bands.
- Confuse cosmic ray spikes with real peaks: Cosmic rays make sharp, intense spikes at random positions. Present in single acquisitions but disappear in averaged spectra. Always check for reproducibility.
- Neglect polarizability selection rule: Modes strong in IR (asymmetric stretches of polar bonds) may be weak or absent in Raman, and vice versa. Do not expect same intensity pattern as IR.
- Ignore sample degradation: High laser power can char, polymerize, phase-transform sample. Spectrum changes between successive measurements at same spot = degradation.
- Assume all Raman bands are fundamentals: Overtones (2x fundamental frequency) and combination bands can appear in Raman spectra. Typically weaker than fundamentals but can cause confusion if not considered.
- Overlook low-frequency modes: Lattice vibrations, torsional modes, metal-ligand stretches appear below 400 cm-1. Many conventional Raman setups do not access this region. Verify instrument's notch/edge filter allows measurement in low-frequency range if these modes relevant.
See Also
interpret-ir-spectrum— complementary vibrational technique for dipole-active modesinterpret-nmr-spectrum— determine molecular connectivity for complete structure assignmentinterpret-mass-spectrum— establish molecular formula and fragmentationinterpret-uv-vis-spectrum— characterize electronic transitions and chromophoresplan-spectroscopic-analysis— select and sequence analytical techniques before data acquisition
GitHub リポジトリ
関連スキル
llamaguard
その他LlamaGuardは、暴力やヘイトスピーチなど6つの安全性カテゴリーにおいて、LLMの入力と出力をモデレートするMetaの70-80億パラメータモデルです。94〜95%の精度を提供し、vLLM、Hugging Face、Amazon SageMakerを使用してデプロイ可能です。このスキルを使用して、AIアプリケーションにコンテンツフィルタリングと安全策を簡単に統合できます。
cost-optimization
その他このClaudeスキルは、リソースの適正サイジング、タグ付け戦略、支出分析を通じて、開発者がクラウドコストを最適化することを支援します。AWS、Azure、GCPにわたるクラウド支出の削減とコストガバナンスの実施のためのフレームワークを提供します。インフラコストの分析、リソースの適正サイジング、または予算制約への対応が必要な際にご利用ください。
quantizing-models-bitsandbytes
その他このスキルは、bitsandbytesを使用してLLMを8ビットまたは4ビット精度に量子化し、精度の低下を最小限に抑えつつ50〜75%のメモリ削減を実現します。限られたGPUメモリでより大規模なモデルを実行したり、推論を高速化するのに理想的で、INT8、NF4、FP4などのフォーマットをサポートしています。HuggingFace Transformersと統合され、QLoRAトレーニングや8ビットオプティマイザーを可能にします。
dispatching-parallel-agents
その他このClaudeスキルは、複数のエージェントを配備し、3つ以上の独立した問題を並行して調査・修正します。共有状態や依存関係がなく解決可能な、無関係な障害が発生するシナリオ向けに設計されています。中核となる機能は並列問題解決であり、効率を最大化するために独立した問題領域ごとに1つのエージェントを割り当てます。
