The net chemical changes to the oceanic crust during plate creation and destruction are poorly understood because of our inability to constrain its bulk composition on length scales that are relevant to global geochemical budgets. Such large-scale chemical compositions can be relatively easily determined for thick homogeneous sedimentary strata or for unaltered lava flows that may represent total volumes of several cubic kilometers or more. However, rock units in the oceanic crust are heterogeneous either because of original conditions during their formation or as a result of secondary processes such as weathering or metamorphism. Special efforts have to be made to bridge the gap in scale from a geochemical sample size (centimeter scale) and a geologically mappable unit to allow extrapolation of geochemical data on a regional scale (hundreds to thousands of meters). This extrapolation requires a detailed analysis of the heterogeneity on the scale of an outcrop or drilled section. All representative units need to be sampled and integrated in their respective proportions to determine the bulk geochemical composition of a particular reservoir.
Integration of multiple samples for large-scale geochemical budgets can be done in two ways: (1) by physical integration (i.e., mixing) of samples in their respective proportions to form large composite samples for chemical analysis or (2) by geochemical analysis of each sample and numerical integration of the chemical data to determine large-scale chemical composition. The result of these two different procedures should yield an identical geochemical mean; nonetheless, the two approaches differ in terms of their practical and scientific benefits.
The physical composite approach has the advantage that it requires fewer chemical analyses. Small amounts of material from a large number of samples are combined in one or several large composite samples. This is particularly important in ODP sampling, where very little core often has to be shared by a large group of scientists. A small number of composites provides a focus of analytical effort that permits correlation of geochemical data that are usually performed by different laboratories on different samples.
However, composite samples also have potential problems. Once a composite is mixed, it is impossible to resolve whether a particular element is homogeneously distributed between contributing samples or whether it is dominated by a few samples with extreme composition. Thus, by analyzing each sample, bulk compositions can be constructed knowing the composition of each component and permitting cross-checks for samples of extreme composition. Analyzing each sample also has the advantage that the geochemical covariation between individual samples is known, which may help constrain the processes that control the bulk composition.
In our approach, we chose a path between these extremes: samples taken shipboard will be analyzed on shore for the less labor intensive analytical steps (such as major and trace elements), whereas composites will be analyzed for all chemical and isotopic parameters and curated for future investigators. Furthermore, we will prepare several composites that allow us to identify some of the heterogeneity, in particular between moderately altered lava flows and the extremely altered compositions in intercalated sediments, breccias, and hyaloclastites.
For the preparation of composites, we selected 118 individual samples from Hole 801C, ~25 samples per 100 m drilled, including 58 samples from cores drilled during Leg 129 (see the "Supplementary Materials" contents list). All samples were characterized and powdered on board; the mixing and homogenization of composite powders will be completed onshore. Individual core samples were chosen as representative samples of the different alteration domains. Typically, we selected one characteristic minimally altered sample per major igneous unit or at least every ~20 m drilled (total 28 samples). About 50 samples were taken to represent the average altered portion of the core in terms of its colors, halos, and type of veins. We also sampled characteristic, highly altered intraflow materials (IFMs) based on lithologic type (sedimentary or igneous), color, and texture. Most samples were taken as 4- to 6-cm-long, quarter-core sections and were subsequently split into three 4- to 6-mm-thick slabs for the composite powder, a thin section, and an archive sample. In some cases we used representative grains from the coarsely crushed samples for the archive sample.
We recorded the color, vein mineral content, and primary and secondary mineralogy of each sample included in composites. After describing each sample, we used an Al grit in a sandblaster to remove surface contamination, eroding 0.1-0.5 mm of the outermost surface of the sample. Subsequently, individual slabs were cleaned ultrasonically for ~10-20 min in DI and air dried. The dry slabs were sandwiched between two plastic discs, placed in a hydraulic press, and reduced to 2- to 4-mm-sized particles by applying 5-8 T of pressure. Coarsely crushed rocks were then freeze-dried for 3 hr and powdered for 15 min in an alumina-ball mill. Separately ground powder batches of individual samples were combined and mixed for an additional 5 min for homogenization in order to obtain ~10 g of powder for each composite sample.
Overall, we have chosen a procedure that allows us to compare composite data to previous estimates of this type at Sites 417 and 418. A total of 12 composites was prepared to define the bulk chemical composition on different scale lengths. The largest scale is represented in the top alkali basalt composite (Sequence I), the total tholeiite section composite (Sequences III-VIII), and the total drilled section at Site 801. In addition, we will prepare nine other composites, three from each of the three depth intervals (upper, intermediate, and lower tholeiite section). For each of these depth intervals, we will prepare a pillow/flow composite (FLO), an IFM composite, and a depth composite that is a mix of FLO and IFM materials in the proportions in which they occur in each interval. FLO composites are made of a representative mix of basalt samples and quenched margins (plus a minor amount of interpillow material), as is typical for most sequences of submarine pillows and flows. Here, flows may include some shallow intrusions that are hard to distinguish from flows where contacts are not completely recovered. Individual samples were chosen to represent the proportions of different alteration types and vein mineral abundances, as they were described for the whole core. IFMs include epiclastic, volcaniclastic, and hydrothermal units that are deposited between igneous cooling units. These IFMs are often poorly recovered, and their proportions are difficult to determine. Whenever possible, we used logging data to determine the volume of the different units. However, much of this analysis will be conducted onshore; thus, the final composition of the composites will be determined once the log analysis is completed.