Machine Learning Prediction of Adsorption Behavior of Xenobiotics on Microplastics under Different Environmental Conditions
Michael Bryant, Xingmao Ma
Abstract
There have been mounting concerns over microplastics as a vector of environmental xenobiotics recently. Adsorption plays a pivotal role in this process, which varies with the properties of xenobiotics, the characteristics of microplastics, and environmental conditions. The vast number of xenobiotics and the diversity of microplastics, as well as the continuous weathering of microplastics in the environment, make it unrealistic to measure the adsorption capacity and affinity of each combination of xenobiotics, microplastics, and environmental conditions in laboratory studies. Random Forest (RF) and Artificial Neural Network (ANN) algorithms were used to predict the adsorption affinity of xenobiotics on microplastics and elucidate the impact of environmental parameters. pH is responsible for a large variation in the results through its effect on the dissociation of ionizable xenobiotics and the surface charge of microplastics. The aging status of microplastics had a smaller but still significant impact on adsorption affinity, with pristine particles generally having a higher affinity. The results shed light on the potential alteration of the fate and impact of xenobiotics by microplastics. As more data become available in the future, the precision of machine learning (ML) models can be further improved. Overall, our study demonstrated the potential of ML in predicting the adsorption of a wide range of xenobiotics on microplastics.