ANDYPENG09/catalyst-design-skill

Catalyst Design

When to use

• Users asking how to design a catalyst, which composition to pick, how to optimize performance, synthesis-route advice, catalyst formulation, electrocatalyst design, or how to choose synthesis conditions.
• Typical input: design goal (reaction type + material system / performance need). Optional: catalyst-search literature results.
• Covers HER, OER, ORR, PEMWE, overall water splitting, photocatalysis, and plastic upcycling, among other systems.

When NOT to use

• Pure literature search — use catalyst-search; this skill does not re-search.
• Non-catalysis material design (e.g., battery cathodes, semiconductor devices) — not applicable.
• Looking up a specific DOI or full text — use WebSearch / WebFetch directly.

Prerequisites

• Ships with its own knowledge base (references/) and templates (templates/); runs with no external dependencies.
• Optional: when WebSearch / WebFetch are available, the skill may trace a specific rule's origin (not required).

Input interface

• Required: design goal — reaction type plus material system or performance need.
• Optional: the structured literature output of catalyst-search (the literature_matrix.md table, supporting conclusions, and validation suggestions).
  • With literature — give targeted advice using its specific systems and metrics, and cite the support.
  • Without literature — give general advice from the built-in methodology (see references/design_methodology.md).

Methodology system (layered, multi-source, traceable, evolvable)

The methodology continuously fuses multi-source evidence, organized in four layers, and evolves dynamically with new literature.

• Full layered knowledge base: references/design_methodology.md (each entry carries a source tag, confidence, and update date).
• Dynamic update mechanism: references/methodology_update_protocol.md.
• Traceability registry: references/methodology_registry.md (trace by ID to source / DOI).

Four-layer framework

• L1 — Theory: structure–property relations, d-band center, Sabatier principle, mechanism typing (AEM / LOM; Volmer–Heyrovsky), activity–stability trade-off, defects as activity, bifunctional synergy.
• L2 — Methods and tools: synthesis library (SEA, low-melting doping, ligand anchoring, single-source pyrolysis, Joule heating, templating, grain-boundary engineering), characterization and computation (XRD, STEM, XPS, EXAFS, DFT, ML), structural design (size, ordering, core–shell, support, single- and dual-atom sites).
• L3 — Parameters: annealing temperature (450–1150 °C), cooling rate (slow cooling ≤ 2 °C/min), atmosphere, particle size, doping level, composition ratio, post-treatment.
• L4 — Validation: RDE → MEA tiered evaluation, applicable standards (TCASMES 400-2024, T/CRES 0030-2025), per-system benchmarks, ordering / stability / scale-up validation.

Source tags (traceability)

[CS] catalyst-search retrieval (with DOI) · [AI] user-acquired via AI / skill · [EXT] external channel (conference, patent, standard, internal data) · [EXP] empirical guess (to verify). Confidence: ★–★★★.

When generating advice, surface the source tag of each cited entry and distinguish "literature-confirmed" from "empirical guess".

Workflow

1. Parse input. Extract the design goal and collect multi-source evidence — catalyst-search matrix [CS], user-provided [AI], external [EXT]. If the input is incomplete (missing reaction type or material system), ask the user to clarify first; do not guess.
2. Match rules by layer. Map the goal to L1 theory (mechanism direction) → L2 methods (synthesis / characterization) → L3 parameters (windows) → L4 validation (evaluation path).
3. Generate advice. Output composition, structure, synthesis, conditions, and validation. Each citation must note its source tag (and DOI) and confidence, and distinguish confirmed from guessed.
4. Dynamic backfill (evolve, opt-in). Only when the user explicitly asks to update the knowledge base (e.g., "update your methodology" / "remember this"): validate new rules per methodology_update_protocol.md and backfill the skill's own references/design_methodology.md and references/methodology_registry.md (source, DOI, confidence, status). Otherwise keep the built-in methodology read-only and do not modify any skill files. For under-evidenced (only [EXP]) directions, proactively suggest a catalyst-search re-check to upgrade to [CS].
5. Output. Write the design proposal per templates/design_proposal.md and cite in GB/T 7714 (see catalyst-search's templates/citation_gb7714.md or this skill's own copy).

Output interface

• Design proposal covering composition, structure, synthesis, conditions, and validation; format per templates/design_proposal.md.
• When citing literature, give a GB/T 7714 citation and label each item "literature-supported" or "empirical guess".

Division of labor

• Literature search is catalyst-search's job; this skill does not re-search, only consumes that output or uses its built-in knowledge.
• To trace a specific rule's origin, WebSearch / WebFetch may be used (optional).

Notes

• Advice must be grounded in the built-in methodology or supplied literature; no ungrounded claims.
• Clearly distinguish "literature-confirmed" from "empirical guess".
• Advice must include an actionable validation path (characterization, testing) for user verification.
• Safety and permissions: by default this skill only reads its own references/ and templates/; it runs no system commands and makes no network requests beyond optional WebSearch / WebFetch; it collects or exfiltrates no user data. Optional self-update: step 4 ("Dynamic backfill") can write to the skill's own references/design_methodology.md and references/methodology_registry.md — but only when you explicitly enable it (e.g., "update your methodology"). It never writes to your own files or project directories.

Example

• Input: design an acidic OER catalyst with η₁₀ < 200 mV and stability > 500 h.
• Output:
  • Composition: Ru@IrOₓ core–shell (high-activity Ru core, dissolution-resistant IrOₓ shell). [CS] ★★★
  • Structure: core–shell interfacial charge redistribution, optimizing oxygen-intermediate adsorption on Ru. [CS] ★★★
  • Synthesis: Adams fusion for the IrOₓ shell, followed by reductive Ru deposition. [CS] ★★
  • Conditions: anneal at 400–500 °C (too high drives Ru into the shell, too low gives insufficient ordering). [EXP] ★
  • Validation: RDE for η₁₀ and Tafel slope; ICP-MS for Ru / Ir dissolution monitoring; 30 000-cycle ADT. [CS] ★★★

Project home

• GitHub: https://github.com/ANDYPENG09/catalyst-design-skill — source, updates, and issue tracker.
• Companion skill: https://github.com/ANDYPENG09/catalyst-search-skill — catalysis literature search.