Surface-Enhanced Raman Spectroscopy-Machine Learning for Multiplex Naphthenic Acid Profiling in Water.

Area of Science:

• Environmental Chemistry
• Analytical Chemistry
• Spectroscopy

Background:

• Naphthenic acids (NAs) are toxic pollutants in industrial effluents, posing risks to aquatic ecosystems.
• Accurate detection and quantification of NAs are crucial for environmental monitoring and risk assessment.

Purpose of the Study:

• To develop a sensitive, data-driven method for detecting and quantifying diverse naphthenic acids in water.
• To leverage surface-enhanced Raman spectroscopy (SERS) and machine learning (ML) for simultaneous NA identification and quantification.

Main Methods:

• Utilized highly uniform silver (Ag) nanoparticles with a cationic surfactant to enhance SERS signals for naphthenic acids.
• Employed machine learning models, including random forest (RF), ridge regression, and Siamese convolutional neural networks (SNN), for spectral data analysis.
• Applied spectral transformations like fast Walsh-Hadamard transform (FWHT), scaling, and principal component analysis (PCA) for model training.

Main Results:

• Achieved detection limits as low as 10^-4 to 10^-5 M for eight distinct NA types.
• RF model showed 86.3% accuracy for single acid identification; ridge regression models achieved average R² of 99.5% for concentration prediction.
• SNN model demonstrated 95% accuracy for identifying multiple NAs in complex mixtures, with averaged F1 scores around 95%.

Conclusions:

• SERS coupled with ML provides a sensitive and effective method for simultaneous identification and quantification of multiple naphthenic acids.
• The developed approach eliminates the need for sample extraction or separation, offering a practical solution for environmental monitoring.
• This work serves as a proof of concept for the rapid and sensitive detection of naphthenic acids in complex environmental matrices.