Improved EPMA of tephra glasses using TDI, combined EDS+WDS, MAN, a multi-standard blank correction, and a multi-standard normalization to facilitate smaller beams, more elements, greater precision, and better between-session reproducibility

Poster presented at the INTAV International Field Conference on Tephrochronology
at Moieciu de Sus, Romania in June 2018

Abstract:

Analysis of natural and synthetic glasses is a common application of electron probe microanalysis (EPMA). Since the late 1960s, EPMA has remained at the forefront of tephra glass analysis because of the technique’s high spatial resolution, analytical sensitivity, rapidity, cost-effectiveness, and potential for high precision. These attributes make it possible to chemically fingerprint and thereby identify and correlate the deposits of individual volcanic eruptions. This, in turn, enables the development of chronological frameworks using widely-dispersed tephra deposits. Additional applications include magmatic processes, eruption processes, volcanic gasses (e.g. via melt inclusions), atmospheric dispersal of tephra, and the environmental and human impacts of volcanic eruptions. In recent years, there have been an increasing number of tephra studies which target trace deposits of volcanic ash known as cryptotephra. This has greatly expanded the geographic range of tephra correlation to localities more than 5000 km from their volcanic sources. However, cryptotephra deposits often contain sparse, small grains. Additionally, even visible tephra beds may contain grains with very narrow walls of glass between bubbles or an abundance of microcrysts which also limit the size of target areas available for analysis. Thus, to maximize the amount of data that can be collected, it may be necessary to use smaller electron beam diameters. In addition, some volcanic centers tend to repeatedly erupt tephras with a very strong “family resemblance” in their geochemistry. Although such closely-clustered compositions may be helpful in attributing tephras to a particular source volcano, distinguishing between individual eruptions becomes much more difficult. Trace-element analysis, e.g. by LA-ICP-MS, is one approach for such situations, but increasing the precision and between-session reproducibility of EPMA analysis and adding more elements can also help. The more precise and additional data can also increase confidence in tephra correlations which rely upon such techniques. Utilization of smaller electron beams can, however, worsen the well-known problem of alkali element migration or “sodium loss” which degrades analytical accuracy. Increasing EPMA precision through greater analysis time and beam current poses a similar accuracy problem. To enable routine, precise, and accurate EPMA of silicate glasses at smaller beam diameters using a broad range of EPMA instrumentation, including older instruments, and to acquire data for additional elements, two analytical methods have been developed. Both utilize (1) a time-dependent-intensity (TDI) correction for time-varying X-ray intensities, (2) mean atomic number (MAN) modelled X-ray backgrounds, (3) a combination of several wavelength-dispersive (WDS) spectrometers with a single high count rate silicon drift detector (SDD) energy dispersive (EDS) spectrometer, (4) an offline multi-standard blank correction, and (5) an offline multi-standard whole-session normalization to better matche analyzed concentrations to reference concentrations. This has been implemented on a 6-WDS spectrometer ARL SEMQ electron microprobe - originally built in the 1980s - with a modern EDS and modern analytical and automation software. The TDI correction enables excellent accuracy for Na, and where needed for K. MAN backgrounds simultaneously reduce analytical time and improve analytical precision. The fast SDD EDS enables Si and Al to be analyzed to high precision in only eight seconds, before count rates for these elements change significantly. The blank correction improves trace-level accuracy, and the normalization improves session-to-session reproducibility. Both the blank and normalization use data from multiple reference materials in combination for reduced statistical variation and reduced dependence on the results from any single reference material. For routine work, a 3-minute analysis at 10 nA produces excellent results for 12 elements (SiO2, TiO2, Al2O3, FeO , MnO , MgO , CaO , Na2O, K2O, P2O5, Cl, and BaO) using beams as small as 6 microns on felsic glasses and 3 microns on mafic glasses. Where more precision and/or more elements are needed, a 5-minute analysis at 40 nA produces results for 20 elements (SiO2, TiO2, Al2O3, FeO , MnO , MgO , CaO , Na2O, K2O, P2O5, Cl, BaO, CoO, Cr2O3, NiO, Rb2O, SO3, SrO, VO2, and ZrO2), albeit at larger beam diameters.