Mineralogical Characterization Of Brazilian Kaolin Ore Using Diffuse Reflectance Spectroscopy

For most industrial minerals, the color is directly associated with economic value. In terms of physics, the principal property that defines the term ?color? is the intensity of light reflected (or transmitted) by an object over the visible band, comprehended between 400 and 700 nm. This reflectance spectrum serves as the starting point for a variety of color-describing factors, the most common in the mineral industry being the brightness (TAPPI, 1977 and 1986) and the L*a*b* system (HunterLab, 2008). Quality specifications for mineral products are usually based on these colorimetric parameters. This market is becoming increasingly more selective and demanding over the quality of raw materials and alongside the traditionally controlled characteristics, such as particle size distribution and chemical composition, a trend exists towards the establishment of increasingly more strict specifications for the optical properties of these materials. This is due to the fact that most plants consuming these industrial minerals incorporate them into products in which the visual appearance is controlled. For these plants, most of them being the manufacturers of plastics, paper, paints, and ceramics, variations in the color of mineral materials result in difficulties to achieve the right color in the final product. Therefore, these plants set strict specifications for colorimetric parameters such as whiteness and yellowness, as well as colorimetric coordinates like L*, a*, and b*. For the manufacturer of industrial minerals, this translates into difficulties simply because many of these parameters are usually alien to the tradition of the mining industry. The main quality parameter in the kaolin industry is brightness, which is a weighted average of reflectance values over the range of 400-510 nm (TAPPI, 1977). Due to the lack of a better property, producers tend to use brightness to also assess the quality of the in situ ore and along the processing operations. However, the unprocessed kaolin is often accompanied by ?contaminant? minerals. For Brazilian kaolins, which are the object of this study, the most important contaminants are hematite (Fe2O3), goethite (FeOOH), and anatase (TiO2). Each of these contaminant minerals has a distinct color and behaves differently during the processing of the kaolin ore. Hematite and goethite in particular are particularly strong pigmenting agents and at concentrations as low as 100 ppm they can make the difference between a high-quality white pigment product for paper coating and waste material. As each Fe-bearing contaminant has a different color and responds differently to processing operations (for example, magnetic separation, chemical bleaching, selective flocculation, and/or flotation in the case of kaolin), the in situ brightness tends to show a poor correlation with the brightness of the final product. Brightness measurements are sometimes complemented with X-Ray Fluorescence (XRF) analysis of Fe and Ti, but the total oxide content provided by XRF is also poorly correlated with product quality because Fe may be in the form of goethite, hematite, or inside the crystalline structure of kaolinite. Thus, a more robust technique capable of distinguishing each mineral phase is desirable. Diffuse reflectance spectroscopy has been used to analyze iron oxide/hydroxide content both qualitatively (Torrent and Barrón, 2003; Liu et al., 2011) and quantitatively (Jiet al., 2002 and 2006). These authors correlate slopes and inflection points at characteristic absorption bands with oxide/hydroxide content in natural soils. The approach presented here involves the extensive use of the Kubelka-Munk (K-M) theory, which has been used for color matching (Schabbachet al., 2009 and 2011) and particle size determination (Otsuka, 2004) in a simplified version. Unfortunately, the scattering constant S is very difficult to measure for dry powders, because the equations require a thin layer of material. Attempts have been made (Gonçalves and Petter, 2007, and Gonçalves, 2009) but the experimental errors tend to be high. Thus, this paper aims to develop a tool for the quantification of contaminant minerals in kaolin based on the K-M theory without resorting to measurements over thin layers. This tool is based on a new two-constant approach, which takes into account differences in scattering between materials over the whole measured spectrum. The experimental dataset consists of the reflectance spectra of mixtures between a bleached kaolin and synthetic Fe pigments.