Integrating cyclodextrin fluorescence sensing array and machine learning for rapid screening and detection of...

Area of Science:

• Analytical Chemistry
• Environmental Science
• Materials Science

Background:

• Per- and polyfluoroalkyl substances (PFASs) are persistent environmental toxins found in drinking water.
• Current detection methods like LC-MS are costly, complex, and not portable for rapid screening.
• Existing fluorescence sensors lack multiplexing capabilities or require complex fabrication.

Purpose of the Study:

• To develop a portable, cost-effective sensor for rapid, multiplexed detection of PFASs in water.
• To overcome the limitations of current single-analyte sensors and lab-bound analytical instruments.
• To integrate host-guest chemistry with machine learning for enhanced PFAS analysis.

Main Methods:

• Constructed a 4x6 sensor array using β-cyclodextrin polymer (β-CDP) complexes with four dyes.
• Utilized differential competitive binding of six target PFASs to generate unique fluorescence patterns.
• Employed Linear Discriminant Analysis (LDA) and a deep learning platform for discrimination and quantification.

Main Results:

• Simultaneously detected and quantified six PFASs (PFOA, PFOS, PFDA, PFNA, PFHpA, PFHxA) within 10 minutes.
• Achieved low detection limits (38 ng/L for PFOA, 31 ng/L for PFOS).
• Demonstrated high accuracy in real water samples (0.66% relative error vs. LC-MS) and classified PFAS mixtures.

Conclusions:

• Established a host-guest fluorescence array coupled with machine learning for multiplex PFAS detection.
• Provided a portable, cost-effective platform for on-site screening with LC-MS-level accuracy.
• Highlighted the potential for field-deployable devices for rapid water quality assessment and regulatory compliance.