MINERAL COPRECIPITATION AND MAPPING FOR APPLICATIONS IN TOXIC ELEMENT TREATMENT AND CO2 SEQUESTRATION

Abstract

Carbon mineralization is one of the major approaches to safe CO2 removal and permanent sequestration. Mineralization reactions are currently engineered in alkaline waste materials or ultramafic rocks but improved understanding of carbonate precipitation in the presence of foreign elements will enable carbon mineralization in sulfidic and other heavy metal-contaminated waste materials. This dissertation advances understanding of the fundamental processes and mechanisms underlying incorporation of contaminants into calcium carbonate and their stability once coprecipitated. Whether carbonate coprecipitation reactions can simultaneously target hazardous elements from water and mitigate CO2 is the driving engineering motivation behind this dissertation. Experimental and synchrotron-based characterizations of coprecipitation and the reaction products revealed novel features of carbonate coprecipitates including heterogeneous internal distributions of metal contaminants (i.e., Cd, Zn, As) at the nano- and micrometer scales, morphology and polymorphism, and the mobility of the toxic elements upon dissolution of the hosting phase. Collectively, results from the experimental and multidimensional imaging studies suggest that carbonate precipitation is an effective process for water treatment via solid solution formation with calcium and can achieve near-complete mitigation of the contaminants from water. It was revealed that toxic elements remain sequestered during dissolution of the hosting carbonate phase, and this process is controlled by internal chemical gradients and the selective release of the more soluble calcium from the solid. Novel findings regarding the uptake of contaminants and selective leaching are extended to the immobilization of cadmium during a two-step carbonation process involving Cd-contaminated gypsum wastes. The dissertation also presents a new methodology for mineral mapping (SMART) and reactivity characterizations. This method uses machine learning to build a model that relates synchrotron micro x-ray fluorescence and diffraction data, which embed information about elemental abundances and mineralogical presence. The mineral classifier is used to spatially map and quantify minerals from an input of µXRF data and can reliably distinguish minerals of similar chemistry. Generalizability to new samples, extendibility to new elements and minerals, and reactivity characterizations defined by a combination of particle sizes, mineral accessibility, and trace element compositions, are highlighted via applications and quantitative interpretations of SMART-generated mineral maps.