Structure-activity relationship study on the phytotoxicity of polyether lubricants: Experimental investigation and theoretical prediction based on machine learning models

Due to the increasing awareness of environmental protection among consumers and the tightening of environmental regulations, more and more attention has been paid to environment-friendly lubricants. As a consequence, evaluating the toxicity of lubricants is essential as a way to gauge their performance. In this study, the structure–activity relationship between the structure and properties of polyether lubricants was systematically explored by studying their phytotoxicity (terrestrial/aquatic plants), physicochemical properties and lubrication performances. The results indicated that the toxicity of polyether lubricants on Pak choi and both algae was confirmed in the following order according to the EC₅₀ values: PEG 200 < PEG 400 < PEG 600, PEG 400 > PPG 400, and PEGME 200 < PEGDME 250. Nevertheless, for wheat, PEG 400 < PPG 400. These consequences may be related to molecular polarity, cell structure or both. Considering the high cost and heavy workload of toxicity testing, three machine learning models were also established to compare their toxicity prediction ability on limited data sets under different plant tests. It is found that the ET model produced the best prediction performance in most cases. Overall, the toxicity of polyether lubricants greatly influenced by their structures, and it behaves very differently on different organisms. It is necessary to establish a comprehensive evaluation system that can objectively demonstrate the toxicity of lubricants in the environment. Among the testing oils, the viscosity-temperature, friction-reducing and antiwear performances of PEG 400, PEG 600 and PEGDME 250 are relatively satisfying, but these three oils usually have higher negative environmental impacts. So, there is a contradiction between the performances and environmental friendliness. To resolve this contradiction, there is still a lot of research work to be done.