A Deep Learning Approach to Organic Pollutants Classification Using Voltammetry.

Area of Science:

• Electrochemistry
• Nanomaterial Science
• Machine Learning

Background:

• Accurate detection of organic water pollutants is crucial for environmental monitoring.
• Traditional methods face challenges with similar electroactive compounds.
• Low-cost disposable electrodes offer potential but require enhancement.

Purpose of the Study:

• To develop a deep learning technique for accurate detection and classification of organic water pollutants.
• To enhance the performance of low-cost screen-printed electrodes using nanomaterials.
• To address the challenge of classifying pollutants with similar electroactivity.

Main Methods:

• Cyclic voltammetry characterizations using modified screen-printed electrodes.
• Modification of electrodes with nanomaterials (e.g., carbon nanotubes) to improve sensitivity.
• Deep learning approach using convolutional neural networks (CNNs).
• Transformation of voltammetry data into Gramian angular field (GAF) images for CNN input.

Main Results:

• Nanomaterial modification (carbon nanotubes) improved detection sensitivity by approximately 25 times.
• The CNN model, utilizing GAF transformations, achieved 100% classification accuracy for hydroquinone and benzoquinone.
• The method effectively distinguished between pollutants with overlapping voltammetry peaks.

Conclusions:

• Deep learning combined with nanomaterial-enhanced electrochemistry provides a powerful tool for water pollutant analysis.
• Gramian angular field transformation enables effective classification of complex electrochemical data.
• This approach offers a sensitive, accurate, and reliable method for environmental monitoring of organic pollutants.