INVITED TALK: Making Failure Analysis FAIR: FA Ontology in Operation
INVITED TALK: Making Failure Analysis FAIR: FA Ontology in Operation
Tuesday, November 18, 2025: 10:30 AM
Ballroom A (Pasadena Convention Center)
Summary:
Failure Analysis (FA) is critical for fast time-to-market and semiconductor reliability, yet structuring FA knowledge to align with FAIR principles (Findable, Accessible, Interoperable, Reusable) remains a challenge. Inspired by medical ontologies, we propose a Failure Analysis Ontology (FAO) to represent FA knowledge in a machine- and human-readable format. This enables knowledge graphs that support AI-driven searches and analytics, delivering faster and more accurate insights across the FA process. The FAO spans the entire FA workflow, linking FA data with production information such as technology details and FA equipment. This unified framework ensures traceability, smarter analysis, and streamlined problem-solving throughout the semiconductor lifecycle. After building the ontology, we already have implemented a flow for content moderation as well as productive application to label FA analysis and subsequently extract statistical data based on the ontology structure. Moreover, as part of the EU-funded FA2IR project, we collaborate with European semiconductor manufacturers, FA service providers, and tool suppliers to create a modular, open-source FA ontology with the plan of creating an industry standard for FA process and result classification.
Failure Analysis (FA) is critical for fast time-to-market and semiconductor reliability, yet structuring FA knowledge to align with FAIR principles (Findable, Accessible, Interoperable, Reusable) remains a challenge. Inspired by medical ontologies, we propose a Failure Analysis Ontology (FAO) to represent FA knowledge in a machine- and human-readable format. This enables knowledge graphs that support AI-driven searches and analytics, delivering faster and more accurate insights across the FA process. The FAO spans the entire FA workflow, linking FA data with production information such as technology details and FA equipment. This unified framework ensures traceability, smarter analysis, and streamlined problem-solving throughout the semiconductor lifecycle. After building the ontology, we already have implemented a flow for content moderation as well as productive application to label FA analysis and subsequently extract statistical data based on the ontology structure. Moreover, as part of the EU-funded FA2IR project, we collaborate with European semiconductor manufacturers, FA service providers, and tool suppliers to create a modular, open-source FA ontology with the plan of creating an industry standard for FA process and result classification.