Detection of Microplastics in Freshwater Sediments Based on Raman Spectroscopy and Convolutional Neural Networks

Area of Science:

• Environmental Science
• Analytical Chemistry
• Materials Science

Background:

• Microplastic (MP) pollution is a growing concern in aquatic environments.
• Microplastics accumulate in sediments, where their complex composition hinders analysis.
• Effective monitoring methods are crucial for MP pollution control.

Purpose of the Study:

• To develop a novel detection method for microplastics in freshwater sediments.
• To utilize Raman spectroscopy combined with convolutional neural networks (CNNs) for accurate MP identification.
• To assess the origin of microplastics in Changdang Lake sediments.

Main Methods:

• Density separation and vacuum-assisted filtration were used to isolate MPs from sediment samples.
• A CNN model was trained and tested using Raman spectra of MPs and sediment mixtures.
• The performance of the CNN model was compared against support vector machine (SVM) and random forest (RF) models.

Main Results:

• The developed CNN-based method achieved a high identification accuracy of 94.27% for MPs in freshwater sediments.
• The CNN model demonstrated superior performance compared to SVM and RF models.
• Application to Changdang Lake sediments indicated that local anthropogenic activities are the primary source of MPs.

Conclusions:

• Raman spectroscopy coupled with CNNs provides an accurate and efficient method for detecting microplastics in complex freshwater sediment matrices.
• The findings highlight the effectiveness of advanced analytical techniques in environmental monitoring.
• Understanding MP sources is vital for developing targeted mitigation strategies to reduce aquatic pollution.