MCP HubMCP Hub
스킬 목록으로 돌아가기

interpret-ir-spectrum

pjt222
업데이트됨 Yesterday
1 조회
17
2
17
GitHub에서 보기
기타general

정보

이 스킬은 진단 영역과 지문 영역을 분석하여 기능성 작용기를 식별하는 방식으로 IR 스펙트럼을 체계적으로 해석합니다. 수소 결합 효과를 감지하고 신뢰도가 평가된 기능성 작용기 목록을 생성합니다. 초기 화합물 스크리닝, 반응 모니터링, 또는 기능성 작용기의 존재 여부 확인에 사용하세요.

빠른 설치

Claude Code

추천
기본
npx skills add pjt222/agent-almanac -a claude-code
플러그인 명령대체
/plugin add https://github.com/pjt222/agent-almanac
Git 클론대체
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/interpret-ir-spectrum

Claude Code에서 이 명령을 복사하여 붙여넣어 스킬을 설치하세요

문서

Interpret IR Spectrum

Analyze IR absorption → id functional groups, assess H-bonding, inventory structural features.

Use When

  • ID functional groups in unknown (first screen)
  • Confirm presence/absence (e.g., rxn converted OH → ketone?)
  • Monitor rxn progress → appear/disappear of absorptions
  • Distinguish similar compounds by functional group
  • Complement NMR + MS w/ vibrational info

In

  • Req: IR data (abs freq cm-1 + intensities, %T or Abs)
  • Req: Prep method (KBr, ATR, Nujol, thin film, soln cell)
  • Opt: Mol formula / expected class
  • Opt: Known frags from other spectra
  • Opt: Instrument params (res, range, detector)

Do

Step 1: Spectrum Quality + Format

Verify suitability before peak analysis:

  1. y-axis format: %T (peaks down) / Abs (peaks up). Keep consistent.
  2. Wavenumber range: ≥ 4000-400 cm-1 for mid-IR. Note truncation.
  3. Baseline: Flat + near 100%T (or 0 Abs) in no-abs regions. Slopes/noise → reduce reliability.
  4. Resolution: Adjacent peaks < instrumental res → can't distinguish. Typical FTIR: 4 cm-1.
  5. Prep artifacts: KBr → broad OH ~3400 cm-1 (moisture). Nujol obscures CH stretch. ATR distorts low wavenumbers. Note.

→ Spectrum suitable; format, range, artifacts documented.

If err: Severe baseline probs, saturation (flat-bottom peaks → too-conc sample), prep artifacts obscuring critical regions → note limitation + flag regions unreliable.

Step 2: Diagnostic Region (4000-1500 cm-1)

High-freq region → most functional groups:

  1. O-H (3200-3600 cm-1): Broad abs. Sharp ~3600 → free OH; broad 3200-3400 → H-bonded OH (alcohols, acids, water).
  2. N-H (3300-3500 cm-1): Primary amines → 2 peaks (sym+asym); secondary → 1. Sharper than OH.
  3. C-H (2800-3300 cm-1):
Frequency (cm-1)Assignment
3300sp C-H (alkyne, sharp)
3000--3100sp2 C-H (aromatic, vinyl)
2850--3000sp3 C-H (alkyl, multiple peaks)
2700--2850Aldehyde C-H (two peaks from Fermi resonance)
  1. Triple-bond (2000-2300 cm-1):
Frequency (cm-1)AssignmentNotes
2100--2260C triple-bond CWeak or absent if symmetric
2200--2260C triple-bond NMedium to strong
~2350CO2Atmospheric artifact, disregard
  1. Carbonyl (1650-1800 cm-1) — most diagnostic single region:
Frequency (cm-1)Assignment
1800--1830, 1740--1770Acid anhydride (two C=O stretches)
1770--1780Acid chloride
1735--1750Ester
1700--1725Carboxylic acid
1705--1720Aldehyde
1705--1720Ketone
1680--1700Conjugated ketone / alpha-beta unsaturated
1630--1690Amide (amide I band)
  1. C=C + C=N (1600-1680 cm-1): Alkene C=C → 1620-1680 (weak-med). Aromatic C=C → multiple 1450-1600. C=N (imine) → 1620-1660.

→ All abs in diagnostic ID'd w/ func group + confidence (strong/tentative/absent).

If err: Carbonyl obscured (water in KBr, atm CO2) → note gap. Expected group absent → confirm w/ 2nd prep before concluding absent.

Step 3: Fingerprint (1500-400 cm-1)

Low-freq region → confirmation + structural detail:

  1. C-O (1000-1300 cm-1): Ethers, esters, alcohols, acids → strong C-O. Esters → characteristic strong band 1000-1100 + carbonyl.
  2. C-N (1000-1250 cm-1): Amines + amides; overlap C-O → tentative w/o other evidence.
  3. C-F, C-Cl, C-Br:
Frequency (cm-1)Assignment
1000--1400C-F (strong)
600--800C-Cl
500--680C-Br
  1. Aromatic subst pattern (700-900 cm-1): OOP C-H bending → substitution:
Frequency (cm-1)Pattern
730--770Mono-substituted (+ 690--710)
735--770Ortho-disubstituted
750--810, 860--900Meta-disubstituted
790--840Para-disubstituted
  1. Fingerprint comparison: Region unique per compound. Ref spectrum avail → overlay + compare → identity confirm.

→ Confirmatory assignments for Step 2 groups + structural detail (subst patterns, C-O/C-N).

If err: Fingerprint inherently complex + overlapping. Ambiguous → flag tentative + rely on diagnostic + other spectra.

Step 4: H-bonding + Intermolecular Effects

Evaluate sample state + interactions:

  1. H-bonding broadening: Compare width+pos of OH, NH. Free OH sharp ~3600; H-bonded broad + shifted 3200-3400. Acid dimers → very broad OH 2500-3300.
  2. Conc + state effects: Soln spectra at diff conc → distinguish intramolecular (conc-indep) from intermolecular (conc-dep) H-bonds.
  3. Fermi resonance: 2 overlapping bands → doublet. Classic: aldehyde C-H pair ~2720 + 2820. Recognize → avoid mis-assign as separate groups.
  4. Solid-state effects: KBr + Nujol → solid packing → broadens bands + shifts 10-20 cm-1 vs soln. ATR closest to neat liquid.

→ H-bonding characterized, prep artifacts accounted, anomalous band shapes explained.

If err: H-bonding unresolved (overlap OH + NH) → note ambiguity. D2O exchange / var-temp → helps, requires add'l data.

Step 5: Compile Func Group Inventory

Assemble findings → structured report:

  1. Confirmed groups: Strong unambiguous abs in diagnostic (e.g., sharp C=O at 1715 = ketone/aldehyde).
  2. Tentative: Weaker evidence / overlap → >1 possible group.
  3. Absent: Characteristic strong abs clearly missing (no broad OH → no free alcohol/acid).
  4. Discrepancies: Abs not fitting proposed groups, or expected abs missing.
  5. Cross-ref: Compare IR inventory vs NMR, MS, UV-Vis if avail.

→ Complete inventory by confidence, specific freqs + intensities cited as evidence.

If err: Inventory incomplete/contradictory → ID which add'l exps (ATR vs KBr, var conc, D2O exchange) resolve ambiguity.

Check

  • Quality assessed (baseline, res, artifacts, y-axis)
  • Solvent, prep, atm artifacts ID'd + excluded
  • All abs in diagnostic (4000-1500) assigned / flagged
  • Carbonyl region → sub-type assignment where possible
  • Fingerprint examined for confirmation
  • H-bonding evaluated + peak shape/pos impact documented
  • Inventory compiled w/ confidence
  • Absent groups explicit (neg evidence informative)
  • Cross-ref vs other spectra

Traps

  • Ignore prep artifacts: KBr moisture (broad 3400), Nujol C-H (2850-2950), ATR distortion at low wavenumbers → mimic/obscure real. Always consider prep.
  • Over-interpret fingerprint: Region < 1500 complex+overlapping. Use for confirm not primary ID. Don't assign every peak.
  • Confuse atm CO2 w/ sample: Sharp doublet ~2350 = atm CO2 usually, not sample. BG subtraction removes, verify.
  • Neglect intensity+width: Strong broad ≠ weak sharp at same freq. Report intensity (str/med/weak) + shape (sharp/broad) + freq.
  • Single-peak assignment: Never ID func group from single abs. Carbonyls → supported by additional bands (C-O for esters, N-H for amides, C-H for aldehydes).
  • Absence from weak abs: Some groups → inherently weak IR (sym C=C, triple bonds sym alkynes). Absence ≠ always absence of group.

  • interpret-nmr-spectrum — detailed connectivity + H environments
  • interpret-mass-spectrum — mol formula + fragmentation
  • interpret-uv-vis-spectrum — chromophores complementing IR
  • interpret-raman-spectrum — complementary vibrational → IR-inactive modes
  • plan-spectroscopic-analysis — select + sequence techniques pre-acquisition

GitHub 저장소

pjt222/agent-almanac
경로: i18n/caveman-ultra/skills/interpret-ir-spectrum
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

연관 스킬

llamaguard

기타

LlamaGuard는 폭력 및 혐오 발언 등 6가지 안전 범주에서 LLM 입력과 출력을 조정하기 위한 Meta의 70-80억 파라미터 모델입니다. 94-95% 정확도를 제공하며 vLLM, Hugging Face 또는 Amazon SageMaker를 사용해 배포할 수 있습니다. 이 기술을 사용하여 AI 애플리케이션에 콘텐츠 필터링 및 안전 가드레일을 손쉽게 통합하세요.

스킬 보기

cost-optimization

기타

이 Claude Skill은 리소스 적정화, 태깅 전략, 지출 분석을 통해 개발자들이 클라우드 비용을 최적화할 수 있도록 지원합니다. AWS, Azure, GCP에서 클라우드 비용을 절감하고 비용 거버넌스를 구현하기 위한 프레임워크를 제공합니다. 인프라 비용을 분석하거나, 리소스를 적정화하거나, 예산 제약을 충족해야 할 때 사용하세요.

스킬 보기

quantizing-models-bitsandbytes

기타

이 스킬은 bitsandbytes를 사용하여 LLM을 8비트 또는 4비트 정밀도로 양자화하며, 최소한의 정확도 손실로 50-75%의 메모리 감소를 달성합니다. 제한된 GPU 메모리에서 더 큰 모델을 실행하거나 추론을 가속화하는 데 이상적이며, INT8, NF4, FP4와 같은 형식을 지원합니다. 이 스킬은 HuggingFace Transformers와 통합되어 QLoRA 학습 및 8비트 옵티마이저를 가능하게 합니다.

스킬 보기

dispatching-parallel-agents

기타

이 Claude Skill은 3개 이상의 독립적인 문제를 동시에 조사하고 해결하기 위해 다중 에이전트를 배치합니다. 공유 상태나 의존성 없이 해결 가능한 무관련 장애 시나리오에 맞게 설계되었습니다. 핵심 기능은 병렬 문제 해결로, 각 독립 문제 영역마다 하나의 에이전트를 할당하여 효율성을 극대화합니다.

스킬 보기