The efficient separation between ilmenite and forsterite is critical for improving titanium resource utilization. However, the development of novel flotation collectors still heavily relies on inefficient trial-and-error experimentation. In this work, a high-throughput computational framework integrating Quantum Chemistry (QC), Machine Learning (ML), and Molecular Dynamics (MD) simulations was established for the design and screening of selective collectors. A dataset comprising 66 reported collectors was compiled, and 15 QC-derived descriptors were calculated at the PBE0/Def2-TZVP level. Among the evaluated ML algorithms, Gradient Boosting Regression achieved the optimal predictive performance, confirming a pronounced nonlinear structure–activity relationship between collector binding energies and their electronic descriptors. Feature importance analysis further identified the electron-donating ability of primary bonding atoms (ENDA1) and vertical ionization potential (VIP) as the dominant descriptors governing the interactions with Ti4+, Fe2+, and Mg2+ sites, revealing that electron-donating capacity and charge distribution play decisive roles in flotation selectivity. Accordingly, unsaturated hydroxamic acid collectors featuring C=C double bonds, including 2-OcHA (trans-2-Octenohydroxamic acid), 2-DeHA (trans-2-Decenohydroxamic acid), 2-DdHA (trans-2-Dodecenohydroxamic acid), 2-DyHA (trans-2-Dodecynohydroxamic acid), 9-HHA (trans-9-Hexadecenohydroxamic acid), and 9-OdHA (trans-9-Octadecenohydroxamic acid), were screened, exhibiting superior chelation selectivity toward Ti4+. MD simulations further elucidated the adsorption configurations, interfacial interaction energies, and dynamic adsorption behaviors that are inaccessible to conventional flotation experiments, corroborating the selective adsorption of the identified collectors onto ilmenite at the molecular level. This work not only provides an efficient computational methodology and model for flotation reagent screening but also elucidates the critical collector characteristics governing ilmenite flotation performance, offering a theoretical foundation for the molecular design of novel high-performance collectors.
