ACCELERATED REAGENT DISCOVERY FOR FERTILIZER MINERAL PROCESSING *A. Eskanlou, J.K. Caers, D.Z. Yin Mineral-X, Departments of Earth & Planetary Sciences, Stanford University, USA, (*Presenting author: eskanlou@stanford.edu) Abstract Flotation performance in phosphate beneficiation is often limited by insufficient reagent selectivity between fluorapatite and carbonate gangue, and reagent selection is still largely trialand-error, which is costly and slow to improve as ores become more complex. This study presents a physics-informed screening workflow that links molecular electronic structure to reagent– mineral affinity using a low-data, computational dataset. We assembled 22 chemically diverse reagent molecules, computed quantum-mechanical descriptors (dipole moment, HOMO–LUMO gap, LUMO energy, polarizability), and calculated adsorption energies on fluorapatite (001) and dolomite (104) surfaces. To model the two correlated adsorption responses without a data-intensive multi-output Gaussian process, adsorption energies were transformed into principal-component space, modeled with Gaussian Process Regression (GPR) to provide interpolation and uncertainty estimates, and back-transformed to the original response space. The resulting surrogate model enables interpolation, uncertainty-guided prioritization of the next computation or laboratory tests, and identification of a Pareto-efficient frontier that supports down-selection of candidates that strengthen fluorapatite recovery while limiting dolomite. A representative match, 6methylheptanoic acid, shows close agreement with computed affinity for dolomite and comparable fluorapatite affinity. The workflow is modular and transferable to other mineral systems where selectivity bottlenecks constrain processing of lower-grade, more complex ores. Keywords Critical minerals processing; Froth flotation; Phosphate; Fertilizer; AI; Density Functional Theory (DFT). 1. Context and Problem Statement Phosphate is essential for fertilizers and increasingly important for lithium-iron-phosphate (LFP) batteries. Froth flotation is the most common processing technology to concentrate phosphate minerals from associated gangue minerals with a 60 % global share in marketable phosphate beneficiation (Zhang et al., 2001). This process uses surfactants (chemical reagents) to modify the surface wettability (hydrophobicity) of mineral particles, which facilitates the separation of the specific mineral of interest via air bubbles rising in an aqueous environment (Wills & Finch, 2016). Conventional reagent schemes (often fatty-acid based) lack sufficient selectivity, making direct flotation impractical and forcing reverse flotation that floats large volumes of gangue (silicates/carbonates) while depressing phosphate, increasing water and reagent
RkJQdWJsaXNoZXIy MTM0Mzk2