Photodegradation behavior, toxicity assessment, and machine learning model of liquid crystal monomers on soil minerals and soils.

Liquid crystal monomers (LCMs) are proposed as persistent, bioaccumulative, and toxic emerging contaminants of global concern; however, their transformation behavior and associated risk in soil have rarely been reported. This study investigates the photodegradation kinetics, mechanisms, and toxicity evolution of a representative terphenyl LCM (4''-ethyl-2',3,4,5-tetrafluoro-1,1':4',1''-terphenyl, 2teFT) on soil minerals (SiO2 and goethite), and 8 biphenyl/triphenyl LCMs on soil. The photolysis rate constants (kobs) of 2teFT in various systems followed the order kobs,SiO2 > kobs,goethite > kobs,soils, and triphenyl LCMs possessed larger kobs than biphenyl LCMs. The reduced photo-degradability in soil mainly stems from lower light transmittance and organic matter's light-shielding effects. A machine learning model integrating the soil composition and molecular parameters was further developed to predict kobs of pollutants including LCMs on soil. The model exhibits satisfactory goodness-of-fit (R2tr = 0.945, RMSEtr = 0.277, MAEtr = 0.206), predictive performance (R2test = 0.786, RMSEtest = 0.442, MAEtest = 0.353), robustness and generalization ability (Q2cv = 0.701). Dipole moment, carbon atomic charge, and light sources are the important factors on kobs of pollutant on soil. Mechanistic insights demonstrated the reactive oxygen species (∙OH, ∙O2- and 1O2) were responsible for the degradation of diphenyl/terphenyl LCMs, with the main photodegradation pathways of 2teFT being defluorination, oxidation, and ring cleavage. Critically, some degradation products retaining diphenyl/terphenyl structures exhibit toxicity comparable to parent compound 2teFT, highlighting their ongoing environmental risks. These results are helpful for understanding the fate and risk of LCMs in soil.