The identification of minerals is key for adequately characterizing geological materials at different scales, from the study of outcrop surfaces to extraplanetary exploration. One of the most widely used identification methodologies in recent years is the spectroscopy in the visible, near infrared and short-wave infrared wavelength ranges (VNIR-SWIR spectroscopy). In polymineralic samples the position, geometry and intensity of the absorption features are parameters that can depart from the single-mineral samples as interference phenomena occur between the spectra of each mineral present. In this work it was studied and compared the reflectance spectra of binary mixtures to determine the detection limits of each mineral. For that, binary mixtures of well crystallized kaolinite and poorly crystallized kaolinite mixed with calcite, dolomite, gypsum, quartz and feldspar were prepared. The selected minerals are common in the Earth's crust in different geological contexts, mainly in sedimentary environments, in as alterations associated with ore deposits, and they have also been described on the surface of several extraterrestrial terrains. Binary samples spectra, their continuum removed-spectra and their second derivatives were compared, and identification limits, expressed as a percentage of each mineral, were obtained. These identification limits vary depending on the minerals mixed and the normalization method applied. The results after continuum removal showed that the identification of both kaolinites, in mixtures with calcite and dolomite, is possible if its content is ≥5%; and ≥15%, in mixtures with gypsum. The content to identify calcite must be ≥75%; whereas for dolomite is ≥60%; and ≥20% for gypsum. The implementation of the second derivative entailed some variations in the previous identification limits. Thus, kaolinite is identifiable if its content is ≥5% in all the mixtures; whereas carbonates have identification limits of 90%; and 5% in the case of gypsum. The identification of kaolinite in the presence of both tectosilicates is relatively easy (only 5% content is necessary regardless of the normalization method used). These limits were obtained from a constrained strategy based only in the study of binary mixtures, and easily obtained determination parameters. In that sense, they will probably diverge from the limits obtained in real cases, as other factors such as the number of constituent minerals, grain size, homogenization, and/or crystallinity, among others, may influence the identification. However, the results offer a diagnosis, based on a systematic study, of the arising issues to identify different minerals in polymineralic samples in laboratory or by remote sensing approaches. It is particularly notable the high detection limits of both carbonates in the presence of kaolinite. Therefore, it highlights the ease with which a misidentification of important constituents can occur in SWIR spectroscopy.

The authors would like to express their gratitude to the reviewers and the editor for their constructive comments and suggestions. This study was supported by Junta de Castilla y León, Spain, and Fondo Europeo de Desarrollo Regional (FEDER) (grant number SA0107P20). We acknowledge the technical support provided by the “Servicio de preparación de rocas” of Universidad de Salamanca. E. Manchado is especially recognized for helping with the spectroradiometer data acquisition.

Peer reviewed