The results of XRD analyses of the bulk powder standards are shown in Table T3. Diagnostic peak areas increase in a fairly consistent manner as a function of that mineral's absolute abundance by weight (Fig. F4). To reduce the effects of heterogeneity among the three batches of standards, we used the average values of peak area to calculate the normalization factors shown in Table T4. Table T3 also lists the calculated values of relative weight percent for each component in each mineral mixture, as well as the error of each value relative to the measured weight. The errors for total clay are the largest (average = 3%), but with one exception are 5% or less. Errors for plagioclase are the smallest (average = 1%). The errors associated with the shore-based standards are smaller, on average, than those produced during shipboard measurements (Shipboard Scientific Party, 2001), and they are more consistent in size. The shipboard errors average 4% for clay, 3% for quartz, and 2% for plagioclase and calcite. Contamination of the smectite standard and peak interference between chlorite and kaolinite contributed to those larger and more erratic errors.
Because of our desire to compare shore-based and shipboard data with confidence, we selected a suite of specimens to reanalyze at the University of Missouri. Table T5 shows the samples selected, the new XRD peak areas, and the calculated mineral abundances. Roughly one-half of the reruns were completed on exactly the same bulk powder that was analyzed shipboard, but the others utilized adjacent intervals within sampling "clusters" that may have extended 5 to 10 cm up and down a core. In addition, during the shipboard calculations of total clay, we added together the individual responses of clay peaks generated by four clay minerals, whereas shore-based calculations relied on a single composite peak. Neither approach is without flaw because the intensity of the composite peak depends on both total clay and which specific clay mineral is most abundant. The differences in methodology, as expected, led to systematic shifts in calculated mineral abundances for the natural sediments (Fig. F5). Results from the JOIDES Resolution (Moore, Taira, Klaus, et al., 2001) shift total clay and plagioclase lower (by an average of 6 and 4 wt%, respectively) as compared to shore-based replicates. Replicates shift values of calcite and quartz higher for shipboard data by an average of 2 and 8 wt%, respectively.
The results of XRD analyses of the clay-sized standards are shown in Table T6. As with the bulk powders, linear regression shows that the diagnostic peak areas increase in a fairly consistent manner as a function of each mineral's absolute abundance (Fig. F4). Because of heterogeneity among the three batches of oriented clay aggregates, we used the average values of peak area to calculate the normalization factors shown in Table T4. Table T6 also lists the calculated values of relative weight percent for each component of each mineral mixture, as well as the error for each value relative to the true measured weight. The errors for smectite are the largest (maximum = 5%; average = 3%). Errors for illite are 2% or less, and those for chlorite are 3% or less.
Table T6 also lists normalized percentages for clay minerals only (i.e., %smectite + %illite + %chlorite = 100%). These data permit direct comparisons with values calculated using the Biscaye (1965) peak area weighting factors: 1x for smectite (001) peak area, 4x for illite (001), and 2x for chlorite (002). The errors using SVD are no greater than 5% and are typically less than 4%, with no systematic shifts. The Biscaye peak area method, conversely, consistently underestimates by 7% to 17% the amount of smectite by weight. Overestimates of illite are as high as 16 wt% using Biscaye (1965) for the standards, and the calculated values for chlorite are typically 5 to 8 wt% higher than their true weight percentages.
The new SVD normalization factors have been applied to analyses of both bulk powders and oriented clay-sized aggregates using samples from DSDP Site 297. We characterized their composition as part of a pilot study to show how the coefficient of internal friction and shear strength change within the Shikoku Basin facies (Brown et al., in press). Our results for bulk powders are listed in Table T7. The data for the clay-sized fraction are listed in Table T8.
Figure F6 shows how relative abundances of total clay minerals, quartz, and plagioclase change as a function of stratigraphic position at Site 297. Contents of calcite are trivial, and smear slides show very little biogenic silica. The assemblage of clay minerals (smectite + illite + chlorite) increases in relative abundance toward the bottom of the Lower Shikoku Basin turbidite facies and throughout the volcaniclastic-rich facies. Analyses of the <2-µm size fraction demonstrate that this enrichment of total clay is caused by an increase in smectite. Percentages of smectite by weight within the clay sized fraction are as high as 50%-99%.
The absolute values displayed in Figure F7 are, of course, method-dependent. Figure F8 compares calculations of relative weight percent for each clay mineral using the SVD normalization factors vs. percent by weighted peak area using the Biscaye (1965) weighting factors. Percentages of smectite increase systematically (by as much as 15 to 20 wt%) using the SVD factors, whereas percentages of illite decrease systematically. Values of chlorite change modestly as a function of method. Analysis of error for the standard mineral mixtures (Table T6) indicates that the SVD-based data are more accurate indicators of mineral abundance by weight or volume.
One of the goals of shore-based research associated with ODP Leg 190 is to determine how sediment frictional properties change as a function of mineralogy, especially the abundance of smectite. Another goal is to determine how smectite dehydration affects fluid pressure and fluid flow within and beneath the accretionary prism. The reliability of empirical studies and numerical simulations will improve as our estimates of "absolute" clay-mineral abundance become more accurate. This objective is difficult to achieve using XRD, however, because of the inability to measure abundances of amorphous solids. In addition, the relation between clay in the mineral assemblage and clay in the grain size distribution is complicated. Sand and silt fractions within the trench-wedge and Shikoku Basin facies contain substantial amounts of altered volcanic rock fragments, mudstone-shale fragments, detrital mica and chlorite, and low-grade meta-sedimentary fragments (e.g., Fergusson, this volume). Thus, the difference between %clay in the bulk powder and %clay in the clay-sized fraction is impossible to pinpoint. With this caveat in mind, we estimated the amount of each clay mineral in the bulk powder by multiplying the weight percent of total clay by the relative percentage of each clay mineral in the <2-µm size fraction (Fig. F9) (where %smectite + %illite + %chlorite = 100%). Values of weighted peak area based on Biscaye (1965) weighting factors are 5 to 10 wt% lower than those based on SVD, but both sets of data indicate that smectite abundance within the lower Shikoku Basin is typically greater than 30 wt% of the bulk sediment.