As with any analytical protocol, a cavalcade of unanticipated difficulties associated with seagoing chemical analyses often arises to cause a change in the guidelines and data-reduction techniques of an analytical run. Shipboard analyses provide a different situation than shore-based analyses, where increased error can be introduced because of powder weighing, time constraints, increased variability in lithologies, and availability of materials.
Weighing of sample and lithium metaborate flux on board the JOIDES Resolution is relatively accurate when seas are calm. However, when seas are rough, it is difficult to accurately weigh without significant error. Flux, for the most part, was weighed previously on shore at Texas A&M University to save time; these values are relatively accurate. Flux-powder weight is not as critical as sample powder weight; therefore, the greatest error in weighing is introduced by inaccuracy in sample weight. Even though a weight correction is applied to the data, there is still an error that can not be accounted for. Although igneous geochemists tend to analyze relatively fewer samples (albeit, for a more comprehensive element menu), the analysis of hundreds of sediments on a given leg strictly requires that the flux aliquots be preweighed on shore.
ICP-AES analyses during transit and high seas cause problems that concern high drift, low precision and accuracy, and increased %RSD values. We observed an obvious contrast between instrumental operation during transit and instrumental operation while at site, with significantly poorer results occurring during the transit. We believe this poorer precision is due to the mechanism of nebulization. The nebulizer generates an aerosol mist and injects humidified Ar gas into the nebulizer along with the sample. The ICP-AES on board the JOIDES Resolution is bolted to a table to avoid movement of the instrument, and this is directly coupled to the ship's movement. Upon injection into the spray chamber, however, the aerosol is decoupled from the instrument, which moves around the aerosol (from the aerosol's reference frame). This creates a situation where the aerosol is inconsistently sampled by the plasma, resulting in high drift and low precision for a run. For example, data from Site 1222, which was analyzed during a transit, experienced drift between samples of 50% or more, and the standard deviation for each element concentration and sample was so large that no true pattern was apparent.
The precision of the analyses of P indicated low reproducibility between analytical runs (Table T1). Phosphorus requires the input of nitrogen gas to improve detection limits during sampling because of its low-emission wavelength, which is severely compromised by interference with air. A relatively new nitrogen system on board the JOIDES Resolution was employed for Sites 1215-1219, where nitrogen was supplied to the instrument directly from a nitrogen generator rather than from the storage cylinders. Initially, it was thought that the flow of nitrogen from the generator would be the best source because there would be no period of decreased nitrogen as there would be if a cylinder began to deplete. However, after various tests, we determined that nitrogen supplied via the generator caused precision to greatly decrease (~25%), and nitrogen supplied via the storage cylinders (Sites 1220-1221) caused precision to improve (~10%). Again, this nitrogen flush is only relevant for P because other elements are unaffected by the atmospheric interference.
During the leg, we encountered CaCO3-rich sediments (nannofossil ooze/chalk) that, when analyzed, indicated Ca values >40%, which would indicate >100% CaCO3. This is obviously impossible and is probably due to the fact that Ca concentrations in the SRMs used in the calibration did not span the concentration range of the samples measured. Our highest SRM for Ca values was NIST-1C with just over 35% Ca, which is significantly less than a pure CaCO3 end-member. Also, weighing errors may have also played a role. Regardless, for future legs targeting carbonate-rich sediments, we suggest the use of a pure calcium carbonate powder as a high-end calibration for Ca. Since this was seen in only a few samples (~5-10), it did not prove to be a significant thorn in the side of the shipboard program, and further shore-based analyses will help to constrain the problem.
Sample throughput was limited by the supply of Ar gas and the number of platinum crucibles. We recommend that legs considering a large ICP-AES analytical program should include sufficient Ar gas tanks as well as additional platinum crucibles. This will not only increase sample throughput but will also allow for increased analysis of standards, replicates for determination of precision, and other parameters that will improve the overall results.