スキル一覧に戻る

develop-gc-method

pjt222
更新日 Yesterday
1 閲覧
17
2
17
GitHubで表示
開発general

について

このClaudeスキルは、カラム選択、温度プログラム設定、検出器構成をガイドすることで、開発者がガスクロマトグラフィー法を一から作成するのを支援します。新規GC分析を開始する場合や、既存の方法を異なる装置に適応させる場合に設計されています。揮発性および半揮発性化合物を扱え、初期性能検証も含まれています。

クイックインストール

Claude Code

推奨
メイン
npx skills add pjt222/agent-almanac -a claude-code
プラグインコマンド代替
/plugin add https://github.com/pjt222/agent-almanac
Git クローン代替
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/develop-gc-method

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

Develop a GC Method

Build gas chromatography method step-by-step. Column choice, temp program, carrier gas + detector, initial perf check. Volatile + semi-volatile analytes.

When Use

  • Start new GC analysis for volatile or semi-volatile compounds
  • Adapt published method to different instrument or matrix
  • Replace existing method that no longer meets perf needs
  • Develop method for compounds with known boiling points + polarities
  • Move from packed-column to capillary method

Inputs

Required

  • Target analytes: Compound list with CAS numbers, molecular weights, boiling points
  • Sample matrix: Sample type (air, water extract, solvent solution, biological fluid)
  • Detection limits: Required LOD/LOQ per analyte

Optional

  • Reference method: Published method (EPA, ASTM, pharmacopeial) as starting point
  • Available columns: Column inventory on hand
  • Instrument config: GC model, available detectors, autosampler type
  • Throughput needs: Max run time per sample
  • Regulatory framework: GLP, GMP, EPA, or other compliance context

Steps

Step 1: Define Analytical Objectives

  1. List all target analytes + physical properties (boiling point, polarity, molecular weight).
  2. Identify sample matrix + expected interferents or co-extractives.
  3. Set required detection limits, quantitation range, acceptable resolution between critical pairs.
  4. Decide if method must meet regulatory standard (EPA 8260, USP, etc.).
  5. Document throughput needs: max run time, injection volume, sample prep constraints.

Got: Written spec lists analytes, matrix, detection limits, resolution needs, regulatory/throughput constraints.

If fail: Analyte volatility data unavailable? Estimate boiling points from structural analogs or do scouting run on mid-polarity column to establish elution order.

Step 2: Pick Column

Pick column dimensions + stationary phase by analyte polarity + separation difficulty.

Column TypeStationary PhasePolarityTypical Use Cases
DB-1 / HP-1100% dimethylpolysiloxaneNon-polarHydrocarbons, solvents, general screening
DB-5 / HP-55% phenyl-methylpolysiloxaneLow polaritySemi-volatiles, EPA 8270, drugs of abuse
DB-170114% cyanopropylphenylMid polarityPesticides, herbicides
DB-WAX / HP-INNOWaxPolyethylene glycolPolarAlcohols, fatty acids, flavors, essential oils
DB-6246% cyanopropylphenylMid polarityVolatile organics, EPA 624/8260
DB-FFAPModified PEG (nitroterephthalic acid)Highly polarOrganic acids, free fatty acids
DB-3535% phenyl-methylpolysiloxaneMid-low polarityPolychlorinated biphenyls, confirmatory column
  1. Match analyte polarity to stationary phase: like dissolves like.
  2. Pick column length (15-60 m): longer = more plates, longer run.
  3. Pick inner diameter (0.25-0.53 mm): narrower = better efficiency, wider = more capacity.
  4. Pick film thickness (0.25-5.0 um): thicker films retain volatile analytes longer.
  5. Complex matrices? Consider guard column or retention gap.

Got: Column spec (phase, length, ID, film thickness) justified by analyte properties + separation needs.

If fail: No single column resolves all critical pairs? Plan confirmation column with orthogonal selectivity (e.g., DB-1 primary, DB-WAX confirmatory).

Step 3: Optimize Temperature Program

  1. Set initial oven temp at or below boiling point of most volatile analyte (hold 1-2 min for solvent focusing).
  2. Apply linear ramp. Starting points:
    • Simple mixtures: 10-20 C/min
    • Complex mixtures: 3-8 C/min for better resolution
    • Ultra-fast screening: 25-40 C/min on short thin-film columns
  3. Set final temp 10-20 C above boiling point of least volatile analyte.
  4. Add final hold (2-5 min) for complete elution + column bake-out.
  5. Critical pairs co-elute? Insert isothermal hold just before elution, or reduce ramp rate in that region.
  6. Verify total run time meets throughput needs.

Got: Temp program (initial temp, hold, ramp rate(s), final temp, final hold) separates all target analytes within acceptable run time.

If fail: Critical pairs still not resolved after ramp opt? Revisit column selection (Step 2) or try multi-ramp program with slower rates in problem region.

Step 4: Pick Carrier Gas

PropertyHelium (He)Hydrogen (H2)Nitrogen (N2)
Optimal linear velocity20-40 cm/s30-60 cm/s10-20 cm/s
Efficiency at high flowGoodBest (flat van Deemter)Poor
Speed advantageBaseline1.5-2x faster than HeSlowest
SafetyInertFlammable (needs leak detection)Inert
Cost / availabilityExpensive, supply concernsLow cost, generator optionVery low cost
Detector compatibilityAll detectorsNot with ECD; caution with some MSAll detectors
  1. Default to helium for general work + regulatory methods specifying He.
  2. Consider hydrogen for faster analysis or when helium constrained. Install hydrogen-specific leak detection + safety interlocks.
  3. Use nitrogen only for simple separations or cost-driven work.
  4. Set carrier gas flow to optimal linear velocity for gas + column ID.
  5. Measure actual linear velocity with unretained compound (e.g., methane on FID).

Got: Carrier gas picked, flow at optimal linear velocity, verified via unretained peak measurement.

If fail: Efficiency lower than expected at set flow? Generate van Deemter curve (plate height vs linear velocity) using 5-7 flow rates to find true optimum.

Step 5: Pick Detector

DetectorSelectivitySensitivity (approx.)Linear RangeBest For
FIDC-H bonds (universal organic)Low pg C/s10^7Hydrocarbons, general organics, quantitation
TCDUniversal (all compounds)Low ng10^5Permanent gases, bulk analysis
ECDElectronegative groups (halogens, nitro)Low fg (Cl compounds)10^4Pesticides, PCBs, halogenated solvents
NPD/FPDN, P (NPD); S, P (FPD)Low pg10^4-10^5Organophosphorus pesticides, sulfur compounds
MS (EI)Structural identificationLow pg (scan), fg (SIM)10^5-10^6Unknowns, confirmation, trace analysis
MS/MSHighest selectivityfg range10^5Complex matrices, ultra-trace, forensic
  1. Match detector to analyte chemistry + required sensitivity.
  2. Quantitative work, simple matrices → FID default (robust, linear, low maintenance).
  3. Trace analysis, complex matrices → MS in SIM mode or MS/MS in MRM mode.
  4. Halogenated compounds at trace → ECD gives best sensitivity.
  5. Set detector temp 20-50 C above max oven temp to stop condensation.
  6. Optimize detector gas flows per manufacturer.

Got: Detector picked + configured. Right temps + gas flows for target analytes.

If fail: Detector sensitivity insufficient at required detection limits? Concentrate sample (bigger injection, solvent evaporation) or switch to more sensitive/selective detector.

Step 6: Validate Initial Performance

  1. Prep system suitability standard with all target analytes at mid-range conc.
  2. Inject standard 6x consecutive.
  3. Evaluate:
    • Retention time RSD: < 1.0%
    • Peak area RSD: < 2.0% (< 5.0% for trace-level)
    • Resolution between critical pairs: Rs >= 1.5 (baseline) or >= 2.0 for regulated
    • Peak tailing factor: 0.8-1.5 (USP criteria T <= 2.0)
    • Theoretical plates (N): verify vs column manufacturer spec
  4. Inject blank to confirm no carryover or ghost peaks.
  5. Inject matrix blank to find potential interferents at target retention times.
  6. Document all params in method summary sheet.

Got: System suitability criteria met for all analytes across replicate injections. No carryover or matrix interferences at target retention windows.

If fail: Tailing? Check active sites (re-condition column, trim 0.5 m from inlet end, replace liner). RSD over limits? Investigate autosampler precision + injection technique. Resolution insufficient? Return to Step 3 to refine temp program.

Checks

  • All target analytes separated with Rs >= 1.5 for critical pairs
  • Retention time RSD < 1.0% over 6 replicate injections
  • Peak area RSD < 2.0% over 6 replicate injections
  • Peak tailing factors within 0.8-1.5 for all analytes
  • Blank shows no carryover > 0.1% of working conc
  • Matrix blank shows no interferents at target retention windows
  • Total run time meets throughput needs
  • Method params fully documented (column, temps, flows, detector settings)

Pitfalls

  • Ignoring column bleed temp limits: Above max isothermal temp of stationary phase → elevated baseline, ghost peaks, accelerated column degradation. Always check column spec sheet.
  • Oversized injection volumes: Too much solvent → fronting peaks, poor resolution for early eluters. Match injection volume to column capacity (usually 0.5-2 uL for 0.25 mm ID in split mode).
  • Wrong liner for injection mode: Splitless = single-taper or double-taper deactivated liner. Split = liner with glass wool. Mismatched liners → poor reproducibility.
  • Neglecting septum + liner maintenance: Septum coring + liner contamination = most common sources of ghost peaks + tailing. Replace septa every 50-100 injections, liners on documented schedule.
  • Skipping van Deemter optimization: Running at manufacturer default flow instead of measured optimum wastes efficiency, especially when switching carrier gases.
  • Insufficient column conditioning: New columns must be conditioned (ramp to max temp under carrier gas flow, no detector) to remove manufacturing residues before use.

See Also

  • develop-hplc-method -- liquid chromatography for non-volatile or thermally labile analytes
  • interpret-chromatogram -- reading + interpreting GC + HPLC chromatograms
  • troubleshoot-separation -- diagnose + fix peak shape, retention, resolution problems
  • validate-analytical-method -- formal ICH Q2 validation of developed GC method

GitHub リポジトリ

pjt222/agent-almanac
パス: i18n/caveman/skills/develop-gc-method
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

関連スキル

qmd

開発

qmdは、BM25、ベクトル埋め込み、およびリランキングを組み合わせたハイブリッド検索を用いて、ローカルファイルのインデックス作成と検索を可能にするローカル検索・インデックス作成CLIツールです。コマンドラインでの使用と、Claudeとの統合のためのMCP(Model Context Protocol)モードの両方をサポートしています。このツールは埋め込みにOllamaを使用し、インデックスをローカルに保存するため、ターミナルから直接ドキュメントやコードベースを検索するのに最適です。

スキルを見る

subagent-driven-development

開発

このスキルは、各独立したタスクに対して新規のサブエージェントを起動し、タスク間でコードレビューを実施しながら実装計画を実行します。レビュープロセスを通じて品質基準を維持しつつ、迅速な反復を可能にします。同一セッション内で主に独立したタスクに取り組む際に本スキルをご利用いただくことで、組み込まれた品質チェックを伴う継続的な進捗を確保できます。

スキルを見る

mcporter

開発

mcporterスキルは、開発者がClaudeから直接Model Context Protocol(MCP)サーバーを管理および呼び出せるようにします。このスキルは、利用可能なサーバーの一覧表示、引数を指定したツールの呼び出し、認証およびデーモンのライフサイクル管理を行うコマンドを提供します。開発ワークフローにおいてMCPサーバーの機能を統合およびテストする際に、このスキルをご利用ください。

スキルを見る

adk-deployment-specialist

開発

このスキルは、A2Aプロトコルを使用してVertex AI ADKエージェントをデプロイおよびオーケストレーションし、AgentCardの発見、タスク送信、およびコード実行サンドボックスやメモリバンクなどのサポートツールを管理します。Python、Java、またはGoで、順次、並列、またはループのオーケストレーションパターンを用いたマルチエージェントシステムの構築を可能にします。Google Cloud上でADKエージェントのデプロイやエージェントワークフローのオーケストレーションを求められた際にご利用ください。

スキルを見る