Deep learning-integrated SERS platform for accurate identification of diverse phthalate ester subtypes

Area of Science:

• Analytical Chemistry
• Materials Science
• Computational Science

Background:

• Phthalate esters (PAEs) are common plasticizers that can leach from polymers, posing environmental and health risks, including endocrine disruption.
• Current detection methods for PAEs can be time-consuming and lack the sensitivity required for regulatory compliance.
• There is a need for rapid, accurate, and field-deployable analytical tools for PAE identification and quantification.

Purpose of the Study:

• To develop a surface-enhanced Raman spectroscopy (SERS)-based analytical platform for the rapid and accurate identification and classification of seven representative PAEs.
• To integrate deep learning algorithms with SERS for enhanced detection capabilities.
• To validate the platform's performance in detecting PAEs in real-world consumer products.

Main Methods:

• Engineered plasmonic gold nanopillar (AuNP) substrates with nanogap structures to create intense electromagnetic hotspots for SERS signal amplification.
• Collected comprehensive SERS spectral datasets using a portable Raman spectrometer.
• Trained deep neural network (DNN) models on the spectral data for PAE classification and utilized SHapley Additive exPlanations (SHAP) for model interpretability.

Main Results:

• The SERS platform achieved a robust classification accuracy of 99.4% for all seven PAE species.
• SHAP analysis identified key Raman spectral features crucial for accurate PAE prediction.
• The platform successfully detected PAEs in commercial consumer products at concentrations near the 0.1% (w/w) regulatory threshold.

Conclusions:

• The integration of nanostructure-enhanced SERS and deep learning provides a powerful, high-throughput, and field-deployable method for reliable PAE detection.
• This approach offers significant potential for environmental monitoring, consumer product safety assessment, and regulatory compliance.
• The developed platform demonstrates a promising advancement in analytical techniques for hazardous chemical detection.