Perfluorooctanoic Acids (PFOA) removal using electrochemical oxidation: A machine learning approach.

Area of Science:

• Environmental Chemistry
• Electrochemistry
• Machine Learning

Background:

• Perfluorooctanoic Acid (PFOA) is a persistent pollutant requiring effective removal strategies.
• Electrooxidation (EO) is a promising technology for degrading environmental contaminants like PFOA.

Purpose of the Study:

• To evaluate and compare various machine learning (ML) models for predicting EO efficiency in PFOA removal.
• To identify key operational parameters influencing the PFOA degradation process using ML.

Main Methods:

• Evaluated K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), Gradient Boosted Decision Trees (GBDT), and Deep Learning (DL) models.
• Utilized 10-fold cross-validation to assess model performance.
• Analyzed feature importance using Gini impurity and Mean Decrease in Accuracy (MDA).

Main Results:

• The Random Forest (RF) model demonstrated superior performance with an RMSE of 7.7 and a correlation coefficient of 0.965.
• Electrolysis Time was identified as the most significant factor influencing PFOA removal efficiency.
• Current Density and Anode Material were also critical factors, with some variation in ranking between Gini and MDA analyses.

Conclusions:

• The RF model provides a robust and accurate tool for predicting EO efficiency in PFOA remediation.
• Understanding key operational parameters like Electrolysis Time is vital for optimizing electrochemical degradation processes.
• This study advances environmental remediation technologies by offering a reliable ML-based approach for pollutant degradation.