An advanced chemometric modeling framework was developed by integrating an ensemble-based random forest regression (RFR) model. This framework was used to predict the desalination performance of ultrathin titania nanosheet (TNS) and Fe-tannic acid functionalized TNS (Fe-TNS) interlayered thin-film nanocomposite (iTNS and iFe-TFNS) poly(piperazine-amide) membranes. TNS exhibited a wrinkled morphology with a large surface area. Morphological analyses showed that TNS/Fe-TNS membranes formed rougher active layers via modulated monomer transport. A membrane containing 0.05 wt % Fe-TNS achieved a permeability of 6.96 ± 0.17 L m−2 h−1 bar−1, 95.29 ± 0.11% rejection for Na2SO4, and hydrophilic tendency (contact angle 33.9 ± 0.31°). The superior performance arises from the synergistic effect of TNS hydroxyl groups and Fe3+–phenolic coordination of tannic acid. It enhanced hydrophilicity, surface charge, and nanochannel formation, facilitating improved water permeability, salt rejection, superior chlorine resistance, and robustness. The chemometric RFR modeling approach showed strong correlation for smaller data sets (R2 > 0.95) in flux/rejection prediction and temporal fouling dynamics based on membrane characteristics, electrolyte properties, and process conditions. The integrated RFR framework achieved even higher accuracy (R2 > 0.98) for the prediction of poly(piperazine-amide) spectra. This study highlights that integrating a chemometric RFR model is an effective tool for predicting desalination performance.
Yogarathinam et al. (2026) studied this question.