Analysis of sediment samples by X-ray diffraction (XRD) has been a routine part of shipboard and shore-based measurements by the Ocean Drilling Program (ODP) and the Deep Sea Drilling Project (DSDP). The presence of specific detrital and/or authigenic minerals can be detected easily through visual recognition of characteristic peak positions. It is more difficult, however, to estimate the relative abundance of a mineral with meaningful accuracy (Moore, 1968; Cook et al., 1975; Heath and Pisias, 1979; Johnson et al., 1985; Mascle et al., 1988). Fisher and Underwood (1995) developed a method during ODP Leg 156 to calculate abundances of common minerals in bulk powders using matrix singular value decomposition (SVD). They derived normalization factors based on the peak areas of diagnostic XRD reflections, as produced by standard mineral mixtures with known weight percentages of each component. In essence, this method accounts for changes in any given mineral's peak dimensions as a function of its own absolute abundance, as well as the abundance of every other mineral in the mixture.
When the method of Fisher and Underwood (1995) is applied to natural samples of marine sediment, the accuracy of absolute weight percent values deteriorates as additional minerals and amorphous solids (e.g., volcanic glass and biogenic silica) increase in number and amount. Consequently, the mineral percentages calculated by SVD are relative only with respect to the other minerals in the standard mixtures (e.g., weight percent quartz relative to total clay, plagioclase, and calcite). In the case of diatomaceous ooze or vitric mud, this limitation could lead to substantial errors in estimates of absolute mineral abundance relative to all solid phases. Another limitation of the SVD approach is the need to establish sets of normalization factors that match each indigenous mineral mixture within each study area. In other words, factors for a kaolinite-rich mineral suite from Barbados or Costa Rica will not work for an illite-chlorite assemblage in Nankai Trough or Cascadia. Thus, some advanced knowledge of the natural sediment's composition is a prerequisite to mixing appropriate mineral standards. A third significant limitation is imposed by design differences in X-ray diffractometers (e.g., a fixed-step vs. continuous-scan mode, or automatic vs. fixed slits). Separate normalization factors are needed for each type of instrument. This requirement is especially important if there is a desire to integrate shipboard and shore-based data sets generated by different instruments. A final consideration is the individual instrument performance (e.g., life span of X-ray tube and detector). If peak intensities change significantly with tube life, recalibration may be warranted.
Similar questions of accuracy arise during the semiquantitative calculations of clay mineral percentages. The most common approach in marine geology is to apply the Biscaye (1965) peak area weighting factors during calculations of the relative proportions of smectite, illite, and chlorite. The errors in such data can be substantial, however, and they change with the absolute abundance by weight of each mineral (Underwood et al., 1993). Results can also shift because of interlaboratory differences in sample disaggregation and chemical treatments, particle size separation, and the degree of preferred orientation of clay mounts (Moore and Reynolds, 1989; McManus, 1991). Even though the reproducibility of such data might be good, the typical estimates of accuracy are no better than ±10%.
The purpose of this data report is to document the acquisition of new normalization factors for analyses of bulk powder and clay-sized mixtures in sediments from the Nankai Trough and Shikoku Basin (Fig. F1). Recalibration of the methodology was required for several reasons: (1) to allow accurate comparisons among bulk powder data generated by shipboard (Philips) and shore-based (Scintag) XRD systems; (2) to improve mergers of data from DSDP Site 297 and ODP Leg 190; (3) to improve the accuracy of calculated relative mineral percentages within the clay-sized fraction (<2 µm); and (4) to improve the accuracy of calculated percentages of specific clay minerals (e.g., weight percent smectite) within the bulk sediment. We also include the results of bulk powder and clay-fraction analyses of samples from DSDP Site 297 to illustrate fully the utility of the method. The data from Site 297 are important for characterizing the composition of subduction inputs within the Ashizuri transect of Nankai Trough (Fig. F1), and they have been used to document how frictional properties change as a function of total clay content and clay mineralogy (Brown et al., in press).