Information-Theoretic Causal Modelling of Semiconductor Process Dynamics
A novel framework has been established for deducing causal links within semiconductor manufacturing processes, utilizing information theory to scrutinize raw time-series data from equipment log files. The research, available on arXiv, conceptualizes tool dynamics as a stochastic dynamical system comprising both deterministic and stochastic elements, and calculates entropy transfer rates among variables using the Liang-Kleeman and Pires methodology. Initial findings indicated that 7.5% of the inferred dependencies were already known, 36.0% appeared plausible, 17.5% indicated previously uncharacterized relationships, and 39.0% contradicted established process knowledge. These results highlight the framework's potential to reveal new causal insights and emphasize the necessity for enhanced process control as the semiconductor sector evolves towards more complex computing devices and stricter process tolerances.
Key facts
- The framework models semiconductor tool dynamics as a stochastic dynamical system.
- It uses the Liang-Kleeman and Pires formalism to estimate entropy transfer rates.
- Preliminary results: 7.5% known dependencies, 36.0% plausible, 17.5% uncharacterised, 39.0% inconsistent.
- The study is available on arXiv under the category Electrical Engineering and Systems Science > Signal Processing.
- The work aims to improve advanced process control in semiconductor manufacturing.
- The framework infers causal relationships directly from raw equipment log-file time-series data.
- The paper is titled 'Information-Theoretic Causal Modelling of Semiconductor Process Dynamics'.
- The source URL is https://arxiv.org/abs/2608.14678.
Entities
Institutions
- arXiv