Design of saccharide based organic binder for low-grade iron ore pelletization using atomistic simulations and machine learning methods

Inorganic binders like bentonite, used for pelletization of low-grade iron ore, generate iron ore slimes with comparatively high silica and alumina content necessitating extra steps for their removal during iron making process. This demands the usage of organic binders as full or partial replacement of bentonite for iron ore pelletization. In this work, adsorption of organic binders with saccharides skeleton and -H, -OH, -CH₂OH and -CH₂CH₂OH as polar substituents, on goethite surface was studied using density functional theory, molecular dynamics and machine learning. It was observed that adsorption energy of binders on goethite surface had weak dependence on number of hydrogen bonds between them. With this favorable interaction in mind, a library containing 64 organic binders was constructed and adsorption energy of 30 of these binders was computed using molecular dynamics, followed by training of a linear regression model, which was then used to predict the adsorption energy of rest of the binders in the library. It was found that the introduction of -CH₂CH₂OH at R2 position resulted in statistically significant higher adsorption energy. Binder34 and Binder44 were identified as viable candidates for both goethite and hematite ore pelletization and adsorption of their n-mers on goethite and hematite surfaces was also quantified.