The biogenic silica extraction method we employ for both the Na2CO3 and the KOH treatments generally follows Mortlock and Froelich (1989) with some important differences. Before analysis, samples are freeze dried for 3-4 days under vacuum (~2 Torr), ground with an aluminum oxide mortar and pestle, stored in glass-shell vials with Tightseal caps (Kimble 60965-D), and placed in air-tight plastic containers lined with indicating Drierite to minimize rehydration. Approximately 20-50 mg of dried sample is weighed and placed in a 50-mL Nalgene (No. 3110-0500) centrifuge tube (with the exception of Site 1098 Consistency Standard: 10-15 mg was used because of high biogenic silica content). Samples are not "equilibrated" with ambient atmosphere before weighing, as recommended by Mortlock and Froelich (1989). Significant errors can result in the final weight percent calculation of components when samples are rehydrated with "ambient" moisture before weighing, especially for samples with significant clay contents. Seven to ten milliliters of a 10% H2O2 solution (reagent grade) is added to each tube (including standards and blanks), sonified, and allowed to effervesce for 24-48 hr. Two and a half milliliter of 10% HCl is added to each tube, sonified, and allowed to stand for several hours. Tubes are filled with distilled water to dilute the reagents, centrifuged at 4300 rpm for 20 min (IEC Centra GP8), and very carefully decanted with a 10-mL single-channel pipette down to the last 3-4 mL of solution. Sample tubes with remaining sediment are lightly covered with polyvinyl chloride film, placed in constant-temperature gravity oven at 60°C (Precision Model 25EG) until dry, then stored at room temperature until ready for analysis, which is typically a few days.
To each tube, 20 mL of a 2-M Na2CO3 or 2-M KOH (reagent grade) solution is added using a Repipet Jr. (Barnstead/Thermolyne No. 7010; 10 mL ± 0.1 mL) or a single-channel pipette (Fisher brand Finnpipette; 2-10 mL ± 0.10%), capped, sonified to disperse sediment, and weighed. Sample racks (Nalgene 5970) of 24 tubes are placed in a covered, shaking water bath (Precision Model 25) containing room-temperature distilled water that just covers the reagent in the sample tubes. A programmed timer is set to turn on 6-24 hr later to heat to a constant temperature of 85°C (±0.05°C) and constantly agitated at 70 rpm. After 8.5-10 hr at 85°C, sample tubes are removed from the bath and reweighed to monitor potential loss of reagent from evaporation or potential gain of bath water through the pinhole in the cap while in the shaking water bath. Samples are then centrifuged for 20 min at 4300 rpm, and before analysis, all samples will have cooled for ~1 hr. Sample cooling is preferable to taking aliquots while hot. We noticed significant pipetting errors when low-volume aliquots (50 µL) are taken from hot samples.
Dissolved silica is measured by the heteropoly blue method using a Hach Company DR/4000 spectrophotometer with a detection range from 0 to 1.600 mg/L SiO2 ±0.01 mg/L at a wavelength set to 815 nm. An aliquot of 50 µL of digested sample is pipetted (Fisher Finnpipette 40-200 µL ± 1.1%) using sterile tips into a precleaned polypropylene reaction vessel containing 9.95 mL deionized distilled water (ASTM Type II, VWR 3234-7, also used to prepare reagents). Additional reagents (molybdate 3, citric acid, and amino acid F) are added following a standard procedure for low-range silica ("Method 8186" [Hach, 1997]). Sample aliquots larger than 500 µL failed to develop a colored solution and results were erratic, suggesting that the alkaline Na2CO3 or KOH matrix interferes with the complexation process. Because silica reacts with molybdate under acidic conditions to form the unreduced (yellow) form of silicomolybdic acid, matrix interferences, which significantly raise the pH, can be avoided by drawing aliquots no larger than 100 µL.