Material-level countermeasures for securing microfluidic biochips.

Area of Science:

• Biotechnology and Biomedical Engineering
• Materials Science
• Cybersecurity

Background:

• Flow-based microfluidic biochips (FMBs) are increasingly used in diagnostics and computing.
• Outsourced manufacturing exposes FMBs to stealthy material-level attacks, including PDMS degradation via solvents or altered curing ratios.
• These attacks threaten FMB functionality, enable counterfeiting, and facilitate intellectual property theft.

Purpose of the Study:

• To develop a dynamic material-level watermarking scheme for polydimethylsiloxane (PDMS)-based FMBs.
• To create a robust method for detecting material degradation and fabrication anomalies in FMBs.
• To safeguard FMBs against counterfeiting and intellectual property theft.

Main Methods:

• Implemented a perylene-labeled fluorescent dye in PDMS microvalves, creating a unique excimer intensity peak under 405 nm laser excitation.
• Quantified mechanical strain by measuring the downward shift in fluorescence intensity during pneumatic actuation.
• Utilized machine learning models trained on force-displacement data from mechanical punch tests to detect curing ratio anomalies.

Main Results:

• Demonstrated a high correlation (R² = 0.971) between normalized excimer intensity change and microvalve strain using fluorescence microscopy.
• Achieved over 99% accuracy in detecting curing ratio anomalies with machine learning models.
• Validated the effectiveness of the watermarking scheme and ML models in identifying material-level attacks.

Conclusions:

• The proposed dynamic watermarking scheme effectively detects material-level attacks on PDMS microfluidic biochips.
• Machine learning models provide highly accurate detection of fabrication anomalies, specifically curing ratio variations.
• These countermeasures offer proactive protection for FMBs against counterfeiting and ensure device integrity in critical applications like point-of-care diagnostics.