Quantitative evaluation of mineral grains using automated SEM-EDS analysis and its application potential in optically stimulated luminescence dating
Optically stimulated luminescence (OSL) dating allows constraining the depositional age of sediments with good accuracy and precision. A fundamental requirement in OSL dating is to use purified sub-samples (i.e. mono-mineralogic aliquots composed of e.g. quartz or potassium feldspar only), because of the different OSL properties and dosimetry characteristics of each mineral phase. Where multiple mineral phases are present on an aliquot, a mixed OSL signal might be obtained, with potentially adverse effects on the robustness of the resulting optical ages. Detailed evaluation of the mineralogical composition of the hundreds or even thousands of individual mineral particles that constitute an aliquot in OSL dating has – until recently – not been reasonably feasible with current analytical techniques. Here we report on the use of an automated mineralogy system that combines scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) and facilitates ultra-fast analysis of particulate mineral phases with a spatial resolution on the micron scale. The method is applied to mono-mineralogic coarse-grained (100–250 μm) and poly-mineral fine-grained (4–11 μm) OSL samples, respectively and cross-checked with electron probe micro analysis (EPMA). It is shown that (i) some coarse-grained mineral extracts that underwent standard physico-chemical preparation to isolate quartz for OSL dating, still suffer from mineralogical contamination, mainly in the form of feldspar and mica inclusions and, that (ii) polymineral fine-grained samples reveal a complex mineralogical composition with a significant percentage of mica (mainly muscovite). Implications of these quantitative mineralogical observations for OSL dating are discussed. QEMSCAN is further used to examine the efficiency of different physico-chemical preparation strategies to isolate a restricted range of mineralogies and to optimize single preparation steps. We conclude that the clear advantage of automated SEM–EDS systems lies in the rapidity with which accurate high-resolution maps of hundreds or even thousands of mineral particles can be generated, i.e. at a level statistically representative of the bulk OSL sample. Automated SEM–EDS techniques might thus be helpful in OSL dating for quality assurance and investigation of problematic OSL samples.