Total CO2 Measurements
During the TUNES-3 Expedition, 652 samples were analyzed for TCO2 concentrations in seawater. The sampling frequency for measurements of the carbonate parameters was reduced to a complete depth profile (36 samples) ~ every fourth hydrographic station (Fig. 2). This reduction was implemented not according to any prearranged geophysical criterion but to accommodate the time constraints for two analysts on board to perform CO2 sampling and measurements. In other words, the adopted CO2 sampling strategy was to measure as many samples as was technically and humanly possible.
For TCO2 measurements, the seawater samples were drawn into 500-mL borosilicate glass bottles equipped with Rodaviss joint closure systems, poisoned with 100 µL of a saturated solution of mercuric chloride (HgCl2), and analyzed on board generally within 18 h. TCO2 concentration was measured by semiautomated coulometry using an improved version of the instrument earlier described by Johnson et al. (1985, 1987) and calibrated using the procedure described in Goyet and Hacker (1992). This early "SOMMA-type" system did not have gas loops for calibration. Consequently, plans were to calibrate the system with standard solutions as described in Goyet and Hacker (1992); however, uncontrollable events (i.e., a hurricane occurred in Woods Hole a few days before the cruise) destroyed standard solutions that were prepared for the cruise. As a result, the certified reference materials (CRMs) were used as standards to calibrate the TCO2 extraction/coulometer system. Precision of the measurements was estimated to be better than 0.01%; the desired accuracy was better than 4 µmol/kg (Goyet et al. 1995). The automated coulometric system forced the sample into the pipet using a pressurized headspace gas. Pure nitrogen (N2) headspace gas was used for standard solution measurements, and CO2 headspace gas (290 ppm in air) was used for seawater sample measurements. The volume of the pipet was calibrated with distilled water and seawater (volume was ~30 mL, depending on the individual pipet used), and there was no significant difference in the delivery volume as a result of possible differences in surface tension at different salinities. The sample was drained from the pipet into a stripper containing 1.5 mL of 8.5% phosphoric acid.
This chamber and the added acid were purged of any CO2 with pure N2 carrier gas before the sample was added. In the stripper, the CO2 gas was extracted from the acidified sample by a continuous flow of pure N2 gas through a frit at the bottom of the stripper. The gas (mainly CO2, N2 and water vapor) was passed through a condenser thermostated with 4°C water and magnesium perchlorate [Mg(ClO4)] to remove water vapor. It was then passed through silica gel to remove residual aerosols and traces of hydrogen sulfide (H2S) and phosphoric acid (H3PO4) before being bubbled into a commercially available coulometric solution containing ethanolamine [NH2(CH2)2OH], dimethyl sulfoxide [(CH3)2SO], and thymolphthalein dye (made by UIC, Inc., Joliet, IL, USA). A coil made from glass tubing with thermostated water flowing through it was placed in the cell to maintain the solution at 24°C. The pH of the solution was monitored on an UIC, Inc., total CO2 coulometer by monitoring the thymolphthalein-absorbance indicator at ~610 nm. Hydroxide (-OH) ions were generated by the coulometer circuitry to maintain absorbance of the solution at a constant value. The analytical procedure was controlled by a microcomputer that also recorded the coulometric titration and computed the total CO2 extracted from the sample based on the amount of -OH generated to reach the end point.
Figure 3 summarizes the analytical results as a contour section plot of the TCO2 data from the WOCE Section P16C along ~150° W.