During Leg 209, we performed chemical analyses of samples selected by the shipboard scientific party using inductively coupled plasma–atomic emission spectrometry (ICP-AES) and gas chromatography. Various lithologies, including peridotite, serpentinite, and gabbro, were analyzed for major oxide and selected trace element compositions. Sampling and analytical procedures were adapted from those developed during Legs 147, 153, 176, 187, 197, and 203, and the overall strategy is described in ODP Technical Note 29 by Murray et al. (2000). The shipboard analytical facilities are described in the Leg 147 and 187 Initial Reports volumes (Shipboard Scientific Party, 1993 and 2001, respectively). The ICP-AES was first used during Leg 187, and additional details are given in the Leg 197 Initial Reports volume (Shipboard Scientific Party, 2002).
Selected representative samples were first cut with a diamond-impregnated saw blade and wet-ground on a diamond abrasive wheel to remove surface contamination. Samples were then washed in an ultrasonic bath containing methanol for ~10 min, followed by three consecutive ~10-min washes in an ultrasonic bath containing nanopure deionized water. They were then dried for ~12 hr in an oven at 110°C. The cleaned whole-rock samples (~10 cm3 in size) were reduced to fragments <1 cm in diameter by crushing them between two disks of Delrin plastic in a hydraulic press, followed by grinding for ~5 min in a Spex 8510 shatterbox with a tungsten carbide barrel. Approximately 1 g of sample powder was weighed on a Scientech balance and ignited at 1025°C for 4 hr to determine weight loss on ignition (LOI).
Aliquots of 100 ± 2 mg of the ignited whole-rock powders were mixed with 400 ± 0.4 mg of lithium metaborate (LiBO2) flux that had been preweighed on shore. All samples and standards were weighed on the Cahn Electrobalance under computer control. Weighing errors are conservatively estimated to be ±0.01 mg.
Mixtures of flux and rock powders were fused in Pt-Au crucibles at 1050°C for 10–12 min in a Bead Sampler NT-2100. Ten microliters of 0.172-mM aqueous lithium bromide (LiBr) solution was added to the mixture before fusion as an antiwetting agent to prevent the cooled bead from sticking to the crucible. Cooled beads were transferred to 125-mL polypropylene bottles and dissolved in 50 mL of 2.3-M nitric acid (HNO3) by shaking with a Burrell Wrist Action bottle shaker for 1 hr. After digestion of the glass bead, all of the solution was filtered to 0.45 µm into a clean 60-mL wide-mouth polypropylene bottle. Next, 2.5 mL of this solution was transferred to a plastic vial and diluted with 17.5 mL of 2.3-M HNO3 to bring the total volume to 20 mL. This solution-to-sample dilution, used for major and trace elements, is 4000. Dilutions were made using a Brinkman dispensette (0.25 mL).
Major (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, P, and K) and trace (Sr, V, Cr, Co, Ni, Cu, Sc, Y, Zr, and Ba) element concentrations of powder samples were determined with the JY2000 Ultrace ICP-AES. The JY2000 sequentially measures characteristic emission intensities (with wavelengths between ~100 and 800 nm). P was systematically below detection limits in peridotites and gabbroic rocks in Site 1268 samples and in ultramafic standards (JP-1, DTS-1, and ROA-3). Therefore, we did not analyze this element at the other sites. We routinely ran two element menus, major elements (Si, Fe, Mg, Mn, Al, Ca, and Na) and trace elements (Ti, K, and trace elements listed above). Standard rock powders, calibration and drift solutions, and chemical blanks were included with the unknowns for each sample run. Murray et al. (2000) developed protocols for dissolution and ICP-AES analysis of rock powders. The hard rock analytical procedure was refined during a previous expedition (Leg 197; Shipboard Scientific Party, 2002). The elements analyzed, emission lines used, limits of detection, and specific analytical conditions during Leg 209 are provided in Table T6.
The JY2000 plasma was ignited 30 min before each run to allow the instrument to warm up and stabilize. After the warm-up period, a zero-order search was performed to check the mechanical zero of the diffraction grating. After the zero-order search, the mechanical step positions of emission lines were tuned by automatically searching with a 0.002-nm window across each emission peak using BHVO-2 standard prepared in 2.3-M HNO3. During the initial setup, an emission profile was collected for each peak, using BHVO-2, to determine peak-to-background intensities and to set the locations of background points for each element. The JY2000 software uses these background locations to calculate the net intensity for each emission line. The photomultiplier voltage was optimized by automatically adjusting the gain for each element using the standard (either BHVO-2 or DTS-1) with the highest concentration for that element. Before each run, a profile of BHVO-2 was collected to assess the performance of the instrument from day to day. A typical analytical session for 12 samples lasted ~9 hr, depending on the number of replicate analyses.
All ICP-AES data presented in the site chapters were acquired using the Gaussian analytical mode (mode 2) of the Windows version 5.01 JY2000 software. This mode was used to fit a Gaussian curve to five points, each measured for 1 s, across a peak. The fit was then integrated to determine the area under the curve. The analysis time required in this mode is about three times that needed to simply measure a single peak intensity, but it leads to considerable improvement in analytical precision (Leg 187; Shipboard Scientific Party, 2001). Each unknown sample was run at least twice, nonsequentially, in all sample runs. We used the concentric nebulizer for the JY2000 because it delivers a finer aerosol to the plasma and results in a more stable signal (Leg 187; Shipboard Scientific Party, 2001). However, use of this nebulizer requires filtering the solutions and somewhat greater sample dilution factors to reduce clogging.
ICP-AES runs included the following:
A 2.3-M HNO3 wash solution was run for a minimum of 90 s between each of the samples and standards. The following is an example of a typical run menu for 12 samples, requiring ~9 hr of instrument time: drift, blank, DTS-1, BIR-1, drift, sample 1, sample 2, sample 3, ROA-3, drift, sample 4, sample 5, sample 6, BIR-1 (run as an unknown), drift, sample 7, sample 8, sample 9, ROA-3, drift, sample 10, sample 11, sample 12, blank, drift, DTS-1, BIR-1, drift. During the earlier parts of an analytical session the drift solution is run more frequently, after which it is run at regularly spaced intervals to simplify data reduction. Analytical sessions usually included two analyses of ROA-3, a pyroxenite from the Ronda ultramafic massif previously prepared and analyzed on shore. Use of this secondary standard and BIR-1, run as unknowns, provides a means to realistically assess both analytical accuracy and external precision.
Following each analytical session, the raw intensities were transferred to a data file and data reduction was completed using a spreadsheet to ensure proper control over standardization and drift correction. Once transferred, intensities for all samples were corrected for the procedural blank. A drift correction was then applied to each element by linear interpolation between the drift-monitoring solutions run before and after a particular batch of samples. The interpolation factor for each sample is based on the time of the analysis. Following blank subtraction and drift correction, element concentrations for each sample were calculated from the average intensity per unit concentration for the standards JP-1, BHVO-2, BIR-1 (two of three runs), and DTS-1. The blank was also included in the regression with both its intensity and concentration set to zero. The regression technique gave good correlation coefficients (>0.99 for all elements except Co [>0.96] and Ba [>0.97]) for most oxides and trace elements. It also revealed either important discrepancies and/or problems with sample preparation (e.g., in the case of Cu, which is not reported in the site chapters).
Estimates of accuracy and precision for major and trace element analyses are based on replicate analyses of ROA-3 and BIR-1, the results of which are presented in Table T7. In general, run-to-run relative precision by ICP-AES was better than 2% for the major elements. Run-to-run relative precision for minor and trace elements was generally <5% except when elements were near the detection limit of the instrument (see Table T6 for instrument detection limits).
To more fully investigate LOI for each sample, gas chromatographic separation of sample volatiles was carried out using a Carlo Erba NA 1500 CHS analyzer, in which the respective gaseous oxides of C, H, and S are quantitatively determined by a thermal conductivity detector. The sample introduction system has a vertically mounted quartz tube containing small pellets of reduced Cu, separated by a small amount of quartz wool from tungstic anhydride, which acts as a catalyst. Samples are dropped into the tube in tin boats and heated at 1010°C in the presence of oxygen for ~75 s. During this time, nitrogen, hydrogen, carbon, and sulfur released from the sample are oxidized and swept into the gas chromatograph (GC) using helium carrier gas. The sulfur from sulfide minerals is oxidized to SO2, and SO2 is released from any sulfate minerals present in the sample. Aliquots of sulfanilamide (C5H8N2O5S) weighing between ~1 and 3 mg were used for primary calibration of the instrument. A single sample analysis requires ~12 min. During this time, signal intensity at the detector was continuously recorded, and N, C, H, and S separated by the GC were measured sequentially at ~60, 120, 320, and 465 s. Following blank subtraction, concentrations for each sample were calculated from the integrated peak areas of the respective gases relative to those for the standard using a linear regression. The blank was included in the regression with both its intensity and concentration set to zero. The regression technique usually gave good correlation coefficients (>0.99).
Sample analyses were performed on rock powders dried at 110°C for 12 hr. For Sites 1268 through 1271, sample sizes were typically 30–40 mg. This relatively large amount of sample was initially chosen in an attempt to optimize the analysis of CO2 in serpentinized peridotite. However, 30 mg of serpentinized peridotite releases about an order of magnitude more H2O than the standard calibration using sulfanilamide. After Site 1271, it became very evident that analysis of the larger (30–40 mg) samples leads to a bias of 10%–20% in measured water content, so the sample size was reduced to 5–10 mg to optimize the analysis of H2O concentration. This bias estimate is based on comparison of measured and accepted values for three secondary standard rocks routinely analyzed as unknowns during each analytical session. The three secondary standards are a shipboard diabase standard BAS140 (Sparks and Zuleger, 1995; Bach et al., 1996), a peridotite standard JP-1, and an altered basalt from Leg 147 (Sample 147-895D-10W; Puchelt et al., 1996). Typical analytical sessions included multiple analyses of each of these standard rocks, to provide a test of the accuracy and reproducibility of the methods. Results of the GC analyses for BAS140, Sample 147-895D-10W, and JP-1 are presented in Table T8. Detection limits, calculated as three times the standard deviation of all the blank analyses, are CO2 = 0.04 wt%, H2O = 0.15 wt%, and S = 350 ppm. In comparing the GC volatile analyses to the LOI results, it is also important to bear in mind that conversion of Fe2+ to Fe3+ during ignition may produce a gain in weight that is 11.1% of the percentage of ferrous Fe contained in the sample. This can lead to LOI results that are less than the volatile concentrations determined by GC. Because of a technical problem, S was not analyzed in most of the Site 1270 rocks.
Complete sample analyses are available in data tables (see the "Supplementary Materials" contents list) and in the "Geochemistry" sections in each site chapter.