Previous   |    Next

doi:10.2204/iodp.proc.313.203.2019

Methods and materials

XRF measurements on split-core surfaces (referred to hereafter as “XRF-core”) and on individual core samples (referred to hereafter as “XRF-sample”) were acquired for all three Expedition 313 holes (Table T1; Fig. F1B), focusing on the interval between seismic reflectors m4.5 and m4.1. In Hole M0027A, both XRF-core and XRF-sample measurements were acquired for all of lithostratigraphic Unit II and extend from seismic reflector m5.2 at the base uphole to reflector m4.1 (Table T1). In Hole M0028A, the upper part of Unit II was measured, with XRF-core measurements acquired in the clays between seismic reflectors m4.5 and m4.1 supported by interspersed XRF-sample measurements to correlate key trends across sites (Table T1). In Hole M0029A, in the thicker sequences of Unit II a program of targeted sampling was followed for XRF-sample measurements between seismic reflectors m4.5 and m4.1, with a small interval of XRF-core measurements in the uppermost clay near seismic reflector m4.1 (Table T1).

Core depth shifts

All core and section depths were obtained from the Expedition 313 legacy information found in the site chapters of this volume (see the “Site M0027,” “Site M0028,” and “Site M0029” chapters [Expedition 313 Scientists, 2010c, 2010d, 2010e]). At each site, notably in the clay sequences, core recovery was often recorded as >100% (Inwood, 2018). In this report, the method of choice to avoid confusion and loss of resolution in overlapping intervals was to use the depth map provided in Tables AT1, AT2, and AT3 in “Appendix A” to scale core from its original depth (core depth below seafloor, Method A [CSF-A]) to core depth below seafloor, Method B (CSF-B) (see IODP Depth Scales Terminology at http://www.iodp.org/top-resources/program-documents/policies-and-guidelines). Overlap affects 10 of 26 cores recovered from Hole M0027A, 9 of 11 cores from Hole M0028A, and 18 of 53 cores from Hole M0029A (maximum extra recovery of 27%, 14%, and 15%, respectively), and without scaling, considerable interpretation error could arise. Scaled depths (CSF-B) are used for all measurements taken on the core unless otherwise stated. Downhole logging depths are left unchanged from those reported (wireline depth below seafloor (WSF)/wireline matched depth below seafloor (WMSF) during Expedition 313 (see the Methods chapter [Expedition 313 Scientists, 2010b]).

Note that in Hole M0027A, Core 59H was left unscaled because it is in an interval of very low core recovery, with core positioning correspondingly uncertain and no overlap with adjacent cores. Overlaps with the underlying Cores 16R in Hole M0028A (following a reaming operation) and 104R in Hole M0029A (described as a slipped core) are considered to have resulted from drilling disturbance, and both cores were thus moved upward by the amount of overlap.

Geochemical measurements on split core surfaces (XRF-core measurements)

XRF core scanning provides a nondestructive analysis system for relatively fast, high-resolution analysis of major and minor elements (from Mg through U) by scanning the surface of split sediment cores. The core surface was observed carefully to avoid collecting data from around cracks. XRF data were collected directly from the split core surface of the archive half every 1 cm downcore over a 1.2 cm² area with an XRF Core Scanner II (AVAATECH; serial Number 2) at MARUM–University of Bremen using the following settings:

• Slit size = 10 mm,
• Generator
• = 50 kV at 1 mA current (only for Hole M0027A)
• = 30 kV at 1 mA current
• = 10 kV at 0.25 mA current
• Sampling time = 20 s

The split-core surface was covered with a 4 µm thin SPEXCerti Prep Ultralene1 foil to avoid contamination of the XRF measurement unit and desiccation of the sediment. The data were acquired by a Canberra X-PIPS silicon drift detector (SDD; Model SXD 15C-150-500) with 150 eV X-ray resolution, a Canberra digital spectrum analyzer (DAS 1000), and an Oxford Instruments 50W XTF5011 X-ray tube with rhodium (Rh) target material. Raw data were processed by analysis of X-ray spectra with iterative least-squares software (WIN AXIL) from Canberra Eurisys.

XRF count rates for the continuous core surfaces are influenced by factors that include sample matrix, sediment thickness, porosity (water content), and surface characteristics (e.g., Tjallingi et al., 2007). Direct count rates of individual elements must therefore be interpreted with caution, especially when comparing light and heavy elements or where measurements are closer to the noise level.

In “Appendix B,” geochemical XRF core measurements are presented as downhole plots for Holes M0027A (Fig. BF1), M0028A (Fig. BF2), and M0029A (Fig. BF3) and as raw and normalized values (counts per second) (Table BT1). Elements that were measured but are unreliable or below the detection threshold were excluded.

Geochemical measurements on individual samples (XRF-sample measurements)

Conventional XRF measurements were obtained from individual samples from Holes M0027A (118 samples), M0028A (21 samples), and M0029A (60 samples). All whole rock samples were prepared using standard methods (Pickering et al., 1993; Tarney and Marsh, 1991). Major and trace element data were obtained from fusion beads and pressed powder pellets, respectively, by XRF analysis using a PANalytical Axios Advanced X-Ray fluorescence spectrometer at the University of Leicester, United Kingdom. Total loss on ignition (LOI) was measured on predried powders. The PANalytical Axios runs a 4 kW Rh anode end-window supersharp ceramic technology X-ray tube. Samples are loaded from a 96-position sample changer when configured for 32 mm diameter fusion beads or pellets. The control and processing software was the PANalytical SuperQ system with IQ+, WROXI, and ProTrace extensions. Calibrations were set using international rock reference materials under the same conditions and regressing the measured count ratios against recommended concentrations (after Govindaraju et al., 1994; Imai et al., 1995; see also http://georem.mpch-mainz.gwdg.de).

Geochemical XRF-sample measurements are presented as downhole plots for Holes M0027A, M0028A, and M0029A (Figs. CF1, CF2, CF3 in “Appendix C”) and as major element oxides in weight percent and trace elements in parts per million in Tables CT1, CT2, and CT3 in “Appendix C.” Where values approach the noise level or were inappropriate for analysis for other reasons, Br, Cl, Cs, Sb, Se, Sn, W, and SO₃ were excluded from further analysis; however, the values were retained in the tables for completeness and to show small intervals where measurements were above the noise level.

Petrophysical, lithologic, mineralogic, and total organic carbon data

In this report, petrophysical logs are plotted and described with the geochemical data where they are relevant, namely magnetic susceptibility, both measured downhole and on the recovered core, and downhole spectral gamma radiation. The acquisition method for these data and for the lithologic description, digital linescan images, color reflectance, and mineralogic and total organic carbon (TOC) measurements are described in the Expedition 313 “Methods” chapter (Expedition 313 Scientists, 2010b).

Statistical analyses

For this study, 33 elements were analyzed for XRF-sample measurements and 14 elements were analyzed for XRF-core measurements. Statistical analyses are extremely beneficial for identifying or corroborating trends in geochemical data, especially for larger geochemical data sets. However, geochemical data represent a compositional data set (e.g., the major oxides are summed to 100%) and therefore suffers from the closure effect, so caution needs to be applied in the selection of appropriate statistical techniques (Aitchison, 1986; Pearson 1897; Chayes, 1960). Classical correlation plots for such constrained data can result in spurious artificial correlations, and a more applicable technique is to present correlations as a log-ratio variation matrix, a log-ratio covariance matrix, or a centered log-ratio covariance matrix (Rollinson, 1992). In this report, correlation matrixes for the quantitative XRF-sample data have been constructed based on centered log-ratios for Holes M0027A, M0028A, and M0029A (left panels in Fig. F2). The diagonal of the correlation plot shows each major element oxide and trace element against itself, with the darker the shading indicating higher the correlation (using absolute coefficient values); the nature of the correlations can be seen clearly irrespective of the reader’s familiarity with the precise statistical technique.

The statistical evaluation of the data was enhanced by running a principal component analysis (PCA) for major element oxides and trace elements, again using centered log-ratios. This analysis transforms the data into orthogonal components and represents a way of reducing dimensionality. Plots of the first two principal components (PC1 and PC2) are shown for Holes M0027A, M0028A, and M0029A (right panels in Fig. F2). Elements that are grouped together on these plots can be inferred to have a higher probability of being located within the same mineral or to be affected by similar processes.

The statistical results shown in Figure F2 are included at a larger scale in Figures DF1, DF2, and DF3 in “Appendix D,” with tables of the numerical values for the first 11 principal component values in each hole in Figures DF4, DF5, and DF6.

Each site was treated as a distinct and complete data set; therefore, the interpretation needs to take into consideration the different sampling intervals and intensities in each hole, as well as the range of lithologies encountered. As such, the results should be viewed as an initial pass that is beneficial for identifying key trends rather than a rigorous analysis for a more sophisticated interpretation. More advanced analyses could divide the data by lithology and focus on selected intervals. For example, PCA run on a smaller zone would be anticipated to show more distinct groupings.

XRF-core measurements are semiquantitative. To partially account for the various factors that can affect elemental counts, raw values are routinely divided by the total counts at each measurement point and therefore also represent compositional data. Statistical analyses have not been performed on this data set because preference was given to the quantitative XRF-sample data, but elements were analyzed for correlation with the XRF-sample measurements for Ba, Rb, Sr, Zr, Pb, Al, Si, P, S, K, Ca, Ti, Mn, and Fe, indicating a good correlation throughout and enabling future calibration of some of this data set for use as quantitative data.

Top of page   |    Previous   |    Next