Machine Learning-Enhanced SERS Sensor Using Microgroove Structures for Enriching and Confining Nanoplastics in...

Area of Science:

• Materials Science and Engineering
• Environmental Science
• Analytical Chemistry

Background:

• Trace detection of nanoplastics (NPs) is challenging due to random deposition and poor contact with detection hotspots.
• Precise localization of enriched analytes is a significant hurdle for on-site environmental monitoring.

Purpose of the Study:

• To develop a sensor for sensitive and precise detection and localization of nanoplastics.
• To overcome limitations in trace analyte deposition and hotspot contact for enhanced sensing capabilities.

Main Methods:

• Fabrication of a 3D Ag/In2O3/fluorine-doped tin oxide microcavity array with microgrooves.
• Utilizing microgroove pinning effect to control analyte deposition and direct enrichment.
• Employing surface-enhanced Raman spectroscopy (SERS) combined with machine learning (ML) for NP classification.

Main Results:

• Achieved a limit of detection (LOD) of 2.4 × 10^-13 M for 4-MBA with excellent spatial uniformity (RSD = 7.55%).
• Detected various nanoplastics (PS, PMMA, PET) with an LOD of 20 ng/mL for 100 nm PS.
• Demonstrated practical applicability with LODs of 360 ng/mL in river water and 2.94 μg/g in fish matrix for PS.
• Achieved high classification accuracy (100% in river water, 99% in fish matrix) using CNN with a small dataset.

Conclusions:

• The microgroove-confined enrichment and 3D hotspot design enable precise localization and trace detection of nanoplastics.
• Small-dataset machine learning significantly enhances the accuracy of nanoplastic classification in complex environmental matrices.
• This integrated approach offers a field-ready solution for detecting trace pollutants in real-world scenarios.