Data-Driven Approach to Modeling Microfabricated Chemical Sensor Manufacturing.

Area of Science:

• Materials Science
• Chemical Engineering
• Analytical Chemistry

Background:

• Micro pre-concentrator chips (μPC) are crucial for detecting volatile organic compounds (VOCs).
• Scaling up μPC chip production requires robust quality analysis (QA) and quality control (QC) methods.
• Existing QA/QC methods may not adequately address the complexities of microfabrication processes.

Purpose of the Study:

• To develop and validate a statistical model-based approach for QA/QC of scaled μPC chip manufacturing.
• To identify key manufacturing parameters influencing μPC chip performance.
• To enable real-time QA/QC during large-scale production.

Main Methods:

• A statistical model was developed for QA/QC of μPC chip performance.
• A production batch of 30 wafer chips underwent rigorous chemical performance testing.
• Principal Component Analysis (PCA) and multivariate regression models were employed for data analysis.

Main Results:

• The first two principal components in PCA explained 74.28% of chemical testing variance.
• 111 out of 118 viable chips fell within the 95% confidence interval.
• Sorbent mass, heater resistance, and RTD resistance were identified as critical manufacturing parameters affecting chemical performance.

Conclusions:

• The developed statistical model effectively scores individual chip performance and identifies influential manufacturing parameters.
• This data-driven, model-based approach represents a novel method for producing microfabricated chemical sensors.
• Future large-scale production can utilize statistical sampling for real-time QA/QC.