Accuracy of Automated SEM-EDS, FTIR, and XRD Mineral Identification and Quantification: A Discussion based on Measurements of Shale and Synthetic Mudrocks
The accuracy of rock mineral analysis is the degree of closeness between measured (calculated) and actual (true) mineral composition of the rock. Accuracy is dependent on multiple factors, including the analytical method, model assumptions, sample preparation and presentation, and the textural and compositional complexity of the sample. This makes accuracy assessment of modal mineralogy reports a difficult task, especially for shale rocks. Despite the importance in O&G applications, where numerical mineral data are required (e.g. downhole tool calibration), little has been published on this topic, and no comparison between different analytical approaches has been attempted. In this paper, we compare and discuss results from independent analyses using XRD, FTIR, QEMSCAN and a newly developed SEM-EDS based analytical technique performed on 20 compositionally diverse shale samples and 20 synthetic fine-grained ‘mudrocks’. The synthetic rocks were made from 25 minerals combined in strategic, chemically similar, but structurally different combinations. The mineral material was ground to >50 μm, weighted out, mixed into a homogenous paste with CH4O, and pressed into synthetic rocks. The results highlight that mineral misidentification and quantification is significant for all analytical techniques. This must be taken into account to define the most adequate level of reporting of minerals. Since all approaches are based on repositories of mineral reference standards, the accuracy of mineral identification is dependent on the presence or absence of relevant standard for minerals/phases present in the sample. Therefore, the existence of a procedure to report unmatched or poorly matched phases into an ‘unclassified’ category has significant potential for improving the overall accuracy of reported minerals. This concept is incorporated into our proposed technique, introducing threshold-based reporting of ‘unclassified’ phases, and interactive metadata maps to visualize the quality of the spectral matching across the rock surface. Accuracy evaluation of a range of analytical mineral identification and quantification approaches highlights the importance of understanding the specific limitations of numerical modal mineralogy reports. The complexity and fine-grained nature of mudrocks and shale will always result in less reliable assay data. New approaches and tools that make these limitations more transparent could prove to be the way forward.
AAPG Datapages/Search and Discovery Article #90189 © 2014 AAPG Annual Convention and Exhibition, Houston, Texas, USA, April 6–9, 2014