Previous   |    Next

doi:10.2204/iodp.proc.314315316.201.2011

Introduction

Semiquantitative analysis of clay minerals is commonplace in such fields as sedimentology, paleoceanography, and paleoclimate (e.g., Petschick et al., 1996; Svensson et al., 2000; Fagel et al., 2003; Liu et al., 2003). Most techniques are based on X-ray diffraction (XRD), but accurate quantitative analysis of clays and clay minerals remains a formidable challenge (Brindley, 1980; Reynolds, 1989; Snyder and Bish, 1989; McManus, 1991; Moore and Reynolds, 1997). It is difficult to reproduce with precision the intensities generated by broad reflections of poorly crystalline clay minerals, so researchers typically use values of peak area in combination with sets of weighing factors (Biscaye, 1965; Cook et al., 1975; Heath and Pisias, 1979; Fagel et al., 2003; Underwood et al., 2003). Accuracy improves if the analytical methods include calibration with internal standards, use of single-line reference intensity ratios, and some fairly elaborate sample preparation steps to ensure uniformity of random particle orientations (Środoń et al., 2001; Omotoso et al., 2006). However, those kinds of approaches are too laborious to be practical for large suites of samples, such as those stemming from the Nankai Trough Seismogenic Zone Experiment (NanTroSEIZE).

In marine geology, the most widely used factors for calculating percentages of smectite, illite, and kaolinite + chlorite were established by Biscaye (1965). Unfortunately, the d-values of kaolinite (001) and chlorite (002) reflections are nearly identical at ~7 Å, as are the kaolinite (002) and chlorite (004) reflections at ~3.5 Å. This overlap makes it difficult to separate peak intensities or peak areas before calculating their relative abundances. Heating to 550°C or boiling with hydrochloric acid helps to confirm the presence/absence of the two minerals (Nelson and Roy, 1953; Brindley, 1961; Martin Vivaldi and Gallego, 1961), but those steps are time consuming, and comparisons of peaks by subtraction of intensity or area values before and after treatment are flawed because two separate specimens are being analyzed, thereby propagating the error. Consequently, if a specimen contains both kaolinite and chlorite, completing XRD scans before and after the treatment cannot resolve the individual contents of either mineral very accurately.

Biscaye (1964) showed that the ratio of heights of the overlapping 7.16–7.08 Å peaks are approximately the same as the ratio of heights of the double peaks with d-values of 3.58–3.54 Å. Most experts agree that the peaks at ~3.5 Å are better for semiquantitative analyses because of their slightly wider separation (e.g., Petschick et al., 1996; Fagel et al., 2003). The peak-intensity and fitted-peak-area ratios represent two ways to calculate the proportion of each mineral, although the fit between either one of the ratios and the true mineral abundance is nonlinear (Elverhøi and Rønningsland, 1978). Per unit weight, the kaolinite intensity is twice the chlorite intensity, so smaller contents of chlorite are detectable only as a shoulder on the double peak. To create better statistical fits and improve the XRD technique’s accuracy, we mixed and analyzed standards of chlorite and kaolinite in known proportions by weight, and we computed two regression curves for the relation between peak-area ratio and weight percent. This report documents those test results and provides a thorough analysis of the error. By reducing the error, we hope to improve the detail and reliability of XRD results during NanTroSEIZE.

Top of page   |    Previous   |    Next