Atomic Layer Deposition and Etching Process Schemas Extracted with Schema-Miner

This data repository contains the extracted JSON schemas for Atomic Layer Deposition (ALD) and Atomic Layer Etching (ALE) processes, extracted using the Schema-Miner tool. The schemas are categorized under two distinct use cases: experimental and simulation, to reflect the differing perspectives in ALD/E process modeling.

Each schema captures essential process properties, along with their respective constraints, data types, and other structural details necessary for standardized representation and interoperability. The schema extraction was conducted from a curated set of scientific publications related to ALD/E, leveraging large language models (LLMs) in combination with domain-expert insights to ensure both accuracy and relevance.

The complete methodology is described in our paper “LLMs4SchemaDiscovery: A Human-in-the-Loop Workflow for Scientific Schema Mining with Large Language Models,” presented at ESWC Conference 2025. Read the paper here.

The Schema-Miner tool is publicly available on GitHub: Access the repository.

The extracted schemas have also been uploaded as templates to the Open Research Knowledge Graph (ORKG), and can be accessed via the following links:

Atomic Layer Deposition: 

• Experimental Schema: https://orkg.org/templates/R1434174
• Simulation Schema: https://orkg.org/templates/R1368294

Atomic Layer Etching

• Experimental Schema: https://orkg.org/templates/R1379646
• Simulation Schema: https://orkg.org/templates/R1434008

What’s new in Version 4?

Version 4 revises all four ALD/ALE JSON Schemas: ALD experimental, ALD simulation, ALE experimental, and ALE simulation. The release includes correctness fixes, semantic refinements, improved QUDT grounding, stronger validation constraints, structural revisions, and migration to a newer JSON Schema specification. A complete per-schema record of the changes is provided in the accompanying changelog.md file.

The main changes are:

1. JSON Schema specification upgrade: All four schemas were migrated from Draft-07 to Draft 2020-12. Each schema now declares the Draft 2020-12 meta-schema and has been checked for conformance to the updated specification.
2. Correctness fixes: Invalid required constraints, inconsistent property names, incorrect property types, and mismatches between inner and outer quantity-kind declarations were corrected.
3. QUDT grounding corrections and harmonization: All QUDT IRIs were normalized to the canonical, http://qudt.org/vocab/quantitykind/ and http://qudt.org/vocab/unit/ namespaces. Several quantity kinds and units were corrected for physical accuracy and harmonized so that equivalent quantities are represented consistently across the four schemas.
4. Semantic and description improvements: Property descriptions were clarified or corrected, and every declared property now includes a description. Fixed quantity-kind values now use the JSON Schema const keyword instead of single-value enumerations.
5. Structural and validation improvements: QUDT quantity objects now require quantityValue and quantityKind, while their nested value and unit objects require the corresponding internal fields. Non-empty object groups and arrays now use constraints such as minProperties, minItems, and minLength where appropriate. A consistent additionalProperties policy distinguishes closed fixed-structure objects from extensible domain-level groups, and physically motivated numerical bounds were added to quantitative properties.
6. Modeling improvements: Selected properties were restructured to represent their semantics more accurately. For example, elemental concentrations are now represented per element, while previously unstructured quantities such as synergy and ion or oxidant doses are represented using the full QUDT-grounded quantity structure.
7. Metadata: Each schema now includes a stable $id, a top-level description, and a consistent, schema-specific title.

Compatibility
Version 4 is not fully backward-compatible with Version 3. Because of the migration from Draft-07 to Draft 2020-12, corrected property names and constraints, structural changes, and revised QUDT groundings, instances created for Version 3 may require migration before they validate against Version 4. Users with existing Version 3 data should review changelog.md before upgrading.

Full details of every change, organized by schema and category, are provided in changelog.md.