Development of quantitative structure activity relationships (QSARs) for predicting the aggregation of TiO 2 nanoparticles under favorable conditions.

This study developed multi-linear regression (MLR) quantitative structure-activity relationships (QSARs) to predict n -TiO2 aggregation in the presence of high concentrations of representative emerging organic contaminants (EOCs), which presented favorable conditions to interaction with n -TiO2 . The largest diameter change (Δ 517 nm at 0 h and Δ 1164 nm at 12 h) of n -TiO2 was observed by estrone, while the smallest diameter change (Δ -114 nm at 0 h and - 4 nm at 12 h) was observed by lincomycin during experimental periods. In addition, the zeta potential changes of n -TiO2 were observed that the biggest changes were observed by 17β-estradiol (-1.3 mV) and alachlor (-10.02 mV) at 0 h, while 17β-estradiol (-1.31 mV) and pendimethalin (-11.4 mV) showed the biggest changes at 12 h comparing to control. These changes of n -TiO2 diameter and zeta potential may implicate the effects of unique physico-chemical properties of each EOC on the surface modification of n -TiO2 . Based on the interaction results, this study investigated the QSARs between n -TiO2 aggregation and physico-chemical descriptors of EOCs with 7 representative descriptors (pKa, Cw, log Kow , M.W., P.S.A., M.V., # of HBD) for predicting n -TiO2 aggregation rate kinetics at 0 h and 12 h by applying MATLAB statistical methods (model 1 - fitlm and model 2 - stepwiselm). In a model 1, QSARs showed the good coefficients of determination (R2  = 0.92) at 0 h and (R2  = 0.87) at 12 h with 7 descriptors. In a model 2, QSARs showed the goodness of fit of a model (R2  = 0.9998) with 8 descriptors (pKa, Cw, log Kow , M.W., P.S.A., M.V., #HBD, pKa ⋅#H bond donors) at 0 h, while QSARs showed the coefficients of determination (R2  = 0.68) with 2 descriptors (pKa, M.V.) at 12 h. Particularly, we observed that some descriptors of EOCs such as pKa and # of HBD having polarity have more influenced on the n -TiO2 aggregation rate kinetics. Our developed QSARs demonstrated that the 7 descriptors of EOCs were significantly effective descriptors for predicting n -TiO2 aggregation rate kinetics in favorable conditions, which may implicate the complexity interactions between heterogeneous surfaces of n -TiO2 and physico-chemical properties of EOCs.