IODP Proceedings    Volume contents     Search



Quantitative methods in powder X-ray diffraction analyses

Sediment mineralogy can be easily studied in the coarse fraction by petrographic analyses (Tucker, 1988). However, quantitative mineralogy of the bulk fraction (sand- to clay-size material) can be analyzed with a range of methods (Omotoso et al., 2006; Ortiz et al., 2009; Tucker, 1988), with powder qXRD analyses considered more robust (Moore and Reynolds, 1997; Tucker, 1988). Yet this method poses challenges primarily associated with producing a sample that has small and homogeneous grain size and is randomly orientated (Eberl, 2003, 2004; Kleeberg et al., 2008; Omotoso et al., 2006; Środoń et al., 2001).

Here we use a qXRD method (RockJock; Eberl, 2003) that uses an internal corundum standard with a sample preparation technique by Środoń et al. (2001) that homogenizes the sample and minimizes preferred orientation (Eberl, 2003). This sample preparation, analysis, and data reduction techniques have been used for several sediment provenance studies (Andrews and Eberl, 2007, 2011, 2012; Eberl, 2004; Ortiz et al., 2009). The procedures that we followed for this work incorporate several changes in sample preparation and data analysis (i.e., RockJock revision 11) that have been implemented by D. Eberl from earlier versions (Eberl, 2003), improving the random orientation of the sample grains as well as the analysis of the diffraction pattern.

The method was implemented and tested before the analysis of sediment samples from the expedition. We prepared several standards of known mineralogy (Table T1) that contained different ratios of the minerals expected to be most abundant in the sediment from Canterbury Basin. Because the presence of muscovite and chlorite in the sediment has shown to be key in identifying sediment sources within the basin (Adams and Kelley, 1998; Mortimer, 1993; Shapiro et al., 2007), sample standards were created containing these minerals. Average accuracy errors for individual minerals using this method are, after normalization, 3 wt% for quartz, 2 wt% for albite, 2 wt% for labradorite, 2 wt% for orthoclase, 5 wt% for biotite, 6 wt% for chlorite, 3 wt% for muscovite, and 2 wt% for calcite (Table T2). An overall average error from all standard analyses of ±3 wt% is the same as that observed by Eberl (2003) and is considered robust enough for comparison with other data sets (e.g., geochemistry, downhole petrophysics logs).

Precision of the method was tested by the analysis of three different splits of a particular sediment sample (Table T3). Each split was taken from an individual sample that went through all steps of sample preparation. Standard deviation between the three replicates’ mineralogy show that average errors are <1 wt% (after normalization), except in the case of illite, in which precision error is slightly higher (Table T3).

Sample preparation

Our sample preparation methods follow those of Eberl (2003; RockJock revision 11), and the reader is directed there for additional specifics and sources of standards/chemicals used. The method used 1–g of bulk freeze-dried sediment sample that is mixed with 0.25–g of a corundum standard (nominal grain size = 3.5 µm). The sample was mixed with 4 mL of ethanol and ground in a McCrone micronizing mill (ZrO cylinders) for 5 min. The sample was then oven dried for 48 h at 40°C, after which it was mixed with Vertrel (SPEX CertiPrep) and shaken in a plastic vial with three plastic balls in a vertical vortex for 10 min to homogenize the sample and minimize the potential for a preferred mineral orientation. The sample was sieved through a 250 µm mesh and side-loaded by tapping into a round aluminum sample holder mounted against 600 grit sandpaper. This mounting technique minimizes a preferred mineral orientation (Eberl, 2003; Środoń et al., 2001).

X-ray diffraction parameters and data analysis

Side-packed samples were loaded onto a sample-holder carousel and were analyzed in a Rigaku Ultima IV X-ray diffraction system operated at 45 kV, 35 mA, in which the incidence angle spanned from 5° to 85°2θ at 0.02°2θ step size with a scan speed of 0.5 s/step, resulting in 4000 data points (Fig. F2). Sample holders were rotated at 10 rpm during a scan. Each sample was analyzed three times to improve the signal-to-noise ratio, and the combined scans were then imported into the Microsoft Excel RockJock macro program (Eberl, 2003). This program uses stored XRD patterns of mineral standards to recreate the measured diffraction pattern. After selecting the minerals that may be present in the sample (Table T4), the diffraction pattern was analyzed using full-pattern fitting in the 19.0° to 64.5°2θ range to find integrated intensities for the minerals, which were determined from the proportion of each of the mineral standard pattern that results in the best fit. The process utilizes the Solver function in Excel to minimize the degree of fit parameter between the calculated and measured pattern, with values <0.1 considered optimal (Eberl, 2003). Our average degree of fit was 0.097, with a standard deviation of 0.011 (Table T4). The integrated intensities of the mineral standards were used as a reference to determine the weight percentages of the minerals in the sample. The output of this program included a list of minerals studied with their corresponding weight percent and degree of fit (Table T5).

In describing concentrations, mineral groups were presented in the report. Some minerals considered in the diffraction pattern analysis were included to account for variability in the shipboard observed mineralogy of Expedition 317 sediment. This was the case for biotite, which generally was not observed in pattern analysis (Table T5) but was observed visually in smear slides (see the “Expedition 317 summary” chapter [Expedition 317 Scientists, 2011a]). Consequently, the biotite content of the analyses samples may have been reflected in the combined concentrations of biotite, phlogopite, which has a similar diffraction pattern to biotite, and glauconite (only observed in samples from Hole U1351B), which may be a product of biotite alteration. Several polytypes of illite were included in the RockJock analysis, but to be conservative we combined them with muscovite because of the similarity in patterns (Moore and Reynolds, 1997), although it is reported as separate phases in Table T5. Plagioclase feldspars are reported as a total amount, and the most abundant phase observed was albite, with lesser amounts of oligoclase (~25% of total plagioclase feldspars), and trace amounts of labradorite. K-feldspars are reported as a total amount with roughly equal amounts of orthoclase and microcline, when present.