| IODP Proceedings Volume contents Search | |||||||
|
|||||||
| Expedition reports Research results Supplementary material Drilling maps Expedition bibliography | |||||||
|
doi:10.2204/iodp.proc.308.214.2009 MethodsA total of 276 samples were taken onboard the R/V JOIDES Resolution during Expedition 308 (77 samples from Hole U1320A and 199 samples from Hole U1324B). Sample volume was 10 cm3, and the sample frequency was 1 sample per 3 m (one every two core sections, typically taken in odd-numbered sections). XRD analyses were carried out at the Geological Institute of the University of Neuchâtel, Switzerland. The samples were prepared following procedures described in Kübler (1987) and Adatte et al. (1996). Random powder of the bulk sample was used for characterization of the whole-rock mineralogy. Nearly 20 g of each sample was ground with a “jaw” crusher to obtain small chips (1–5 mm) of rock. Approximately 5 g was reserved for bulk sediment analyses and dried at 60°C and then ground again to a homogeneous powder with particle sizes <40 µm. An aliquot of 800 mg of this powder was pressed (20 bar) in a powder holder covered with a blotting paper and analyzed by XRD. Whole-rock composition was determined by XRD (SCINTAG XRD 2000 diffractometer). The scans were done using the following conditions:
The files generated with SCINTAG are raw data (.RD) that are transformed in a routine by the software (DMS program, v. 2.63, graphic-normal display) into calculated (.NI) files. The calculations are fast fourier noise filter, background subtraction, and Kα2 stripping. The measurements are made on calculated files. An internal standard (Kimmeridgian micritic limestone from the Ain region [France] composed of 99% calcite) was run at the beginning of each day to track the aging of the X-ray source tube by looking at the intensity of the peak at 29.43°2θ and the data were corrected accordingly. The resulting diffractograms were analyzed using MacDiff 4.2.5, and each mineral in the assemblage was identified by the hkl reflection of characteristic peaks (Table T1). The intensities of these characteristic peaks were collected for the semiquantitative analysis (Table T2). We chose peaks with no interference as much as possible, but this was not possible for plagioclase and K-feldspar, and we had to discriminate between their characteristic peaks by means of a profile-fitting function using MacDiff 4.2.5 and the Pearson VII method provided in this software. Semiquantitative analysis was based on the following equation (Ferrero, 1965, 1966; Klug and Alexander, 1974; Kübler, 1983):
where
The peak intensity (I) of a mineral “m” in a given assemblage is a function of the abundance of “m” (Cm), the peak intensity (in counts per second) of an external standard containing 100% “m” (Io), and the coefficient of mass absorption (µm) of “m” as well as of the whole assemblage (µe). This relationship is nonlinear, except when µm = µe. Determining µe (or the coefficient of the so called “matrix”) is the principal difficulty in using this technique, as it depends on the proportion of each mineral present in the matrix, which is unknown, as it is the quantity that we want to measure. Here, we follow the method of Ferrero (1966). We assume that the matrix is composed exclusively of clay minerals (micas, chlorites, kaolinites, and smectites), and we use a µe = 47.0, which was determined to be a good average value for a matrix dominated by illite in sedimentary rocks. We can then rewrite Equation 1 to calculate the weight percent of each mineral in the assemblage using
A list of the 100% intensity (Io) and coefficient of mass absorption (µm) of characteristic peaks for the external standards used is given in Table T1. This semiquantitative method leaves us with an “unquantified” fraction of the rock matrix equals to 100 – ΣCm. This fraction averages <5% but could be as high as 15% in some samples (Table T3). Two approaches with respect to the unquantified matrix can be considered. We can either assume that a portion of the rock matrix is either uncrystallized, poorly crystallized, or composed of phases present below detection limit and was thus not taken into account during the quantification process (this fraction could comprise organic matter or poorly crystallized minerals such as opal, phosphates, etc.) or we can assume that we underestimated the amount of phyllosilicates. Phyllosilicates represent a family of very different clay minerals (e.g., kaolinite, smectite, chlorite, etc.). We used the broad peak at 19.90°2θ as the characteristic peak for phyllosilicates, and we quantified them using a “standard” µm of 47 (Table T1). But the specific mixing of these minerals present in our sediment is likely to deviate from the “average” composition estimated for µm, potentially explaining the amount of unquantified minerals. Both options are equally valid, and the correct quantification values are probably affected by a combination of both effects. Here, we decided to give the percent phyllosilicates values both with and without the added unquantified fraction (Figs. F1A, F2A; red curve represents weight percent phyllosilicates alone and the blue curve represents weight percent phyllosilicates with the unquantified fraction added). This gives a bracket for the “true” phyllosilicate content. In general, the difference between the two curves is small, giving us good confidence in our quantification results. Instrumental error, caused by the fluctuation of the X-ray source, detector, and amplification of the signal, has been estimated by measuring the same sample (100% quartz) six times under the same conditions. The mean peak surface error is <3% and the mean intensity error is <1%. Errors associated with the quantification step are more difficult to accurately predict as they depend on the actual matrix effect of each sample, which is unknown. Experience with this method suggests that a conservative error estimate for the quantified values would be 5%–10% for phyllosilicates and 5% for grain minerals. |
||||||
