Previous   |    Next

doi:10.2204/iodp.proc.314315316.132.2009

Lithology

Visual core descriptions

We followed conventional Ocean Drilling Program (ODP) and IODP procedures for recording sedimentologic information on VCD forms on a section-by-section basis (Mazzullo and Graham, 1988). Core descriptions were transferred to section-scale templates using J-CORES software and then converted to core-scale depictions using Strater software. Texture (defined by the relative proportions of sand, silt, and clay) follows the classification of Shepard (1954). The classification scheme for siliciclastic lithologies follows Mazzullo et al. (1988).

The Graphic Lithology column on each VCD plots to scale all beds that are at least 2 cm thick. Combined graphic patterns show persistent interlayers ≤2 cm thick. It is difficult to discriminate between the dominant lithologies of silty clay and clayey silt without quantitative grain size analysis, so we grouped this entire range of textures into the category “silty clay” on all illustrations. We did not use separate patterns for more heavily indurated examples of the same lithologies (e.g., silty clay versus silty claystone) because the dividing line is arbitrary. Figure F7 displays graphic patterns for all lithologies encountered during Expedition 316. Also shown are symbols for internal sedimentary structures, soft-sediment deformation structures, and severity of core disturbance in both soft sediment and indurated sedimentary rock. Assessment of core disturbance by drilling and core processing is primarily based on the degree to which observed textures and fabrics depart from expected natural occurring textures and structures, and thus is a somewhat subjective determination, influenced by the experience of the observer. In tectonically deformed sediments, the distinction between natural and induced fabrics can be particularly ambiguous.

Smear slides

Smear slides are useful for identifying and reporting basic sediment attributes (texture and composition), but the results are not quantitative. We estimated the abundances of biogenic, volcaniclastic, and siliciclastic constituents with the help of a visual comparison chart (Rothwell 1989). Inherent errors are large, however, especially for fine silt and clay-size fractions, and reproducibility among different sedimentologists is poor. Smear slide analysis also tends to underestimate the amount of sand-size grains because they are difficult to incorporate evenly onto the slide. Thus, it would be misleading to report values as absolute percentages. Instead, our descriptive results are tabulated as visual percentage estimates with values grouped into the following range categories:

• D = dominant (>50%).
• A = abundant (>20%–50%).
• C = common (>5%–20%).
• P = present (>1%–5%).
• R = rare (0.1%–1%).
• T = trace (<0.1%).

The relative abundance of major components was also validated by X-ray diffraction (see “X-ray diffraction”), and the absolute proportion of carbonate was verified by coulometric analysis (see “Organic geochemistry”).

The sample location for each smear slide was entered into the J-CORES database with a sample code of SS using the Samples application. The position of each sample is shown on the VCD slide editor column of the VCD application. We tabulated data in an Excel spreadsheet because data entry into J-CORES is prohibitively time-consuming and the program will not accept ranges of values for individual compositional categories.

X-ray diffraction

We completed routine XRD analyses of bulk powders using a Philips PANalytical CubiX PRO (PW3800) diffractometer. The principal goal was to estimate relative weight percentages of total clay minerals, quartz, plagioclase, and calcite using peak areas. A secondary goal was to determine ratios of cristobalite to quartz using peak intensity. Most of the samples were selected from intervals adjacent to whole-round samples, and most are part of sampling clusters with associated physical properties and carbonate analysis. A few additional samples were collected periodically from such unusual lithologies as carbonate-cemented claystone and volcanic ash. Samples were freeze-dried, crushed with a ball mill, and mounted as random bulk powders. The instrument settings were as follows:

• Generator = 40 kV and 45 mA.
• Tube anode = Cu.
• Wavelength = 1.54184 Å (CuK_α).
• Step spacing = 0.01°2θ.
• Rate = 0.1 second/step.
• Slits = automatic.
• Measuring diameter = 10 mm.
• Scanning range = 2°2θ to 60°2θ.

In order for our results to match those of ODP Leg 190 as closely as possible, the choice was made to use MacDiff 4.2.5 software (www.ccp14.ac.uk/​ccp/​ccp14/​ftp-mirror/​krumm/​Software/​macintosh/​macdiff/​MacDiff.html) for data processing. Each peak’s upper and lower limits were adjusted following the guidelines shown in Table T2. Calculations of relative mineral abundance utilized a matrix of normalization factors derived from integrated peak areas and singular value decomposition (SVD) (Table T3). As described by Fisher and Underwood (1995), calibration of SVD factors depends on the analysis of known weight-percent mixtures of mineral standards that are appropriate matches for the natural sediments. Bulk powder mixtures for the Nankai Trough are the same as those reported by Underwood et al. (2003): quartz (Saint Peter sandstone), feldspar (Ca-rich albite), calcite (Cyprus chalk), smectite (Ca-montmorillonite), illite (Clay Mineral Society IMt-2, 2M1 polytype), and chlorite (Clay Mineral Society CCa-2). Examples of diffractograms for standard mixtures are shown in Figure F8.

Average errors (SVD-derived estimates versus true weight percent) are total clay minerals = 3.0%, quartz = 1.7%, plagioclase = 1.2%, and calcite = 1.8% (see CDEX XRD cookbook for thorough analysis of error). In spite of its precision with standard mixtures, the SVD method is only semiquantitative and results for natural specimens should be interpreted with some caution. One of the fundamental problems with any bulk powder XRD method is the difference in peak response between poorly crystalline minerals at low diffraction angles (e.g., clay minerals) and highly crystalline minerals at higher diffraction angles (e.g., quartz and plagioclase). Clay mineral content is best characterized by measuring the peak area, whereas peak intensity may be easier and more accurate to quantify quartz, feldspar, and calcite. Analyzing oriented aggregates enhances basal reflections of the clay minerals, but this step is time consuming and requires isolation of the clay-size fraction to be effective. Errors also propagate as more minerals and peaks are added to the procedure. For clay mineral assemblages, the two options are to individually measure one peak for each mineral and add the estimates together (thereby propagating the error) or to measure a single composite peak at 19.4° to 20.4°2θ. Another source of error is contamination of mineral standards by impurities such as quartz (e.g., the illite standard contains ~20% quartz). Peak interference is particularly problematic for estimates of cristobalite content because the (101) reflection at 22.003°2θ (d-value = 4.0397Å) is in close proximity to plagioclase peaks (e.g., albite [201] [4.032Å, 22.043°2θ] and anorthite [201] [4.042Å, 21.988°2θ]).

In the final assessment, calculated values of a mineral’s weight percent should only be regarded as relative percentages within a four-component system of clay minerals + quartz + plagioclase + calcite. How close those estimates are to their absolute percentages within the mass of total solids will depend on the abundance of amorphous solids (e.g., biogenic opal and volcanic glass), as well as the total of all other minerals that occur in minor or trace quantities. For most natural samples, the absolute errors are probably between 5% and 10%. Thus, the primary value of bulk powder XRD data should be to identify spatial and temporal trends in sediment composition and to assist with core-log integration.

X-ray fluorescence

XRF analysis of major and minor elemental composition was performed using the TATSCAN-F2 energy dispersive spectrometry (EDS)-based core scanner (Sakamoto et al., 2006). Analyses were obtained in two modes: scanning of the whole-core surface and analysis of whole-rock powders. Samples for XRF were chosen in a targeted manner to investigate specific questions concerning the spatial variation of elemental composition in particular core intervals.

The Rh X-ray source was operated at 30 kV accelerating voltage and a current of 0.170 mA. Data are reported as total counts on the peak and also as semiquantitative oxide weight percents. Semiquantitative analysis was performed using a 200 s accumulation. Oxide percentages were calculated from background-corrected integrated peak intensities using software provided by the vendor (JEOL) for the TATSCAN. Table T4 shows shipboard XRF results for the standard JSd-2 compared to the official values for this standard.

Scanning was performed at a spatial resolution of 1 cm and an accumulation live time of 80 s. For interval 316-C0004C-9H-5, 58–100 cm, the scan was repeated with a 5 mm offset (58.5–100.5 cm), yielding a sampling interval of 0.5 cm for the entire data set.

Four applications were undertaken during Expedition 316. In the first application, XRF scanning was performed across the unconformity observed at Section 316-C0004C-9H-5 (see “Lithology” in the “Expedition 316 Site C0004” chapter). Scanning was performed twice. The first scan was made at 1 cm intervals from 50 to 100 cm (Fig. F9). The surface of the core was too rough, leaving intervals of low overall X-ray intensity. To achieve better results, the archive half of the section was smoothed in small depth increments (1–8 cm) using a sharp tool. The core surface was carefully cleaned with a brush between each increment and the material saved for analysis (Table T3). The unconformity slopes steeply across the core, so samples both above and below the unconformity were obtained for three intervals. The scan was repeated on the smoothed surface in 0.5 cm increments from 50 to 100.5 cm (Fig. F10), with better results. Powders from the surface-smoothing process were used for both routine XRD and XRF (as described above) (Table T5; Fig. F11). More than 1 h of tedious effort was required to carefully smooth the archive half without damaging the unconformity (~1.5 h). Another hour of technician time was spent setting up the scanner and handling the data.

A second application of XRF scanning was performed to better understand the X-ray CT results. Examples of two common CT-defined lithologies were scanned at a spatial resolution of 1 cm (Figs. F12, F13). These lithologies differ in their CT number as well as in the textures visible on the CT images (homogeneity, burrowing, etc). “Type 1” lithology (Fig. F12) is characterized by a relatively high CT number, high uniformity, and a lack of microporosity; “Type 2” lithology (Fig. F13) has a relatively low CT number, inhomogenous density, and abundant small micropores. XRF data revealed that the X-ray CT images are relatively insensitive to changes in sediment composition. Proportional variations in common minerals (e.g., quartz, feldspar, and calcite) correspond to variations in bulk elemental composition, but they are not reflected in discernible differences in CT number. Highly localized concentrations of more dense minerals such as pyrite are visible, however. CT number tends to be dominated by porosity, which overwhelms variations in grain density (see “X-ray computed tomography”). XRF scans required ~1 h of technician time for setting up the scans and handling the data.

In the third application, whole-rock powders were analyzed to determine the bulk elemental composition of samples characterized by a relatively small range of XRD-determined compositions. Samples in the interval of interest contain no carbonate and have clay mineral content in a narrow range (48%–56%) (Table T6). XRF data reveal no trends for 10 of the 11 elements examined. Generating these data required more time than the other two applications: ~3 h for sample preparation, ~3 h for machine operation, and 1 h for data analysis.

In the fourth application, three fine-grained black siliceous clasts from the gravel in interval 316-C0007C-17H-CC, 0–5 cm, were analyzed to provide a comparison to the XRD analysis of the same grains (Table T7). Removal of the absolute amounts of Si, Al, Na, and Ca from the XRF data consistent with the XRD-determined relative amounts of quartz and feldspar (assuming all feldspar is plagioclase) leaves a residual that is in excess of the amount of clay determined from XRD (Table T7). This is not unexpected, as there may be components such as glass, mafic minerals, and oxides that are not accounted for in the XRD determination. Feldspar compositions for the individual clasts, calculated assuming that all the Na and Ca in the XRF analysis are affiliated with plagioclase, are Ab₉₂ (possible metasedimentary clasts with albitized grains?), Ab₇₆, and Ab₄₀. XRF indicates less total plagioclase than XRD.

Preliminary assessment of shipboard XRF

In the case of the unconformity, data of considerable utility were generated. The scans clearly show the decline in carbonate content above and the increase in Fe mineralization (pyrite) below the unconformity.

In the second application, XRF data revealed that the X-ray CT images are relatively insensitive to changes in lithology. Variations of lithology would be more readily detected by XRD, XRF, or even smear slides than by X-ray CT. The third application suggests that XRF may be viable as a “last resort” in cases where provenance changes are suspected but not apparent from XRD or smear slides in very fine grained materials. However, given the time required, XRF of this type may not be viable if other higher-priority tasks are pressing. Finally, in the case of the three gravel clasts, reconciling XRF with XRD would certainly benefit from petrographic information.

XRF scanning and XRF powder analysis are highly useful but nonroutine elements of shipboard analysis. Experiences with the applications described here suggest that the most productive use of these techniques may be highly targeted use in which questions about small-scale spatial variations in elemental composition are at issue. Unconformities, erosion surfaces, seafloor precipitation crusts, and alteration halos are examples of situations in which XRF scanning might reveal the nature of processes and products that are inaccessible to visual and simple petrographic observations and for which destructive analyses such as XRD may be undesirable.

Top of page   |    Previous   |    Next