Discrete fCO₂ Measurements

A total of 1549 individual samples were collected in 60-mL serum bottles for discrete fCO₂ analysis. The analysis and calculations followed the method of Neill et al. (1997). The serum bottles were crimp-sealed, a small amount of water was withdrawn from the bottle, leaving a liquid phase of ~54 mL and a gas phase of -6 mL. The introduced gas phase was a mixture of CO₂ in N₂ drawn from a gas-sampling bag and was therefore at atmospheric pressure. Three different concentrations of CO₂ in N₂ were used in order that the partial pressure difference between the liquid and gas phases was kept to a minimum. Normally, near-surface samples (0-250 m) were exposed to a gas phase close to the atmospheric partial pressure of CO₂ (~380 ppm), whereas samples collected from deeper water (> 250 m) were exposed to a gas phase with fCO₂ of ~750 ppm. Toward the eastern end of the section (station numbers > 244), very high fCO₂ values were measured at intermediate depths and a third headspace gas with fCO₂ of ~1450 ppm was used for samples from between ~200 m and 1000 m.

Following headspace introduction, the sealed serum bottles were shaken (equilibrated) for ~3 hours at constant temperature [temperature was held constant to < 0.05°C and was measured with a thermistor equipped with a National Institute of Standards and Technology (NIST)-traceable calibration to an accuracy of 0.005°C]. Most of the samples were equilibrated at ~20°C, however near-surface temperatures along the section were as high as 29°C. In order to maintain a positive pressure in the headspace of the serum bottle during equilibration, samples collected from water depths with a potential temperature > 20°C were equilibrated at sample temperatures of 30 to 32°C. For samples that were equilibrated at the higher temperatures, there was a risk that water vapor would condense in the connecting tubing or sample loop, effectively decreasing the volume of gas injected into the gas chromatograph. This was controlled for and checked by periodically equilibrating duplicate samples at both temperatures and comparing the results following normalization to the same temperature (see discussion of correction in the following paragraphs).

After equilibration, the headspace pressure was measured with a quartz crystal pressure transducer (Paroscientific Inc.; model 216B; 0-45 PSIA). The barometer was connected to a fixed, low-dead-volume side-port needle that was pointed downward and inserted through the septum cap of the serum bottle. The dead volume of the transducer-needle assembly in use with the system was determined to be 290 µL (compared with a nominal headspace volume of ~6 mL), and all headspace pressure data were corrected accordingly.

Following this pressure measurement, the headspace was displaced and flushed through a 0.45-mL sample loop. The mole fraction of CO₂ in the headspace gas was measured by injecting the contents of this loop (at known temperature and pressure) into a gas chromatograph (GC) equipped with a flame ionization detector (FID) and using catalytic conversion of CO₂ to CH₄. The GC measurements were calibrated against a set of four separate CO₂ in air standards (CO₂ mixing ratios of 265, 352, 743, and 1536 pptv) that had previously been intercalibrated with standards maintained in the laboratory of Taro Takahashi and David Chipman at the Lamont-Doherty Earth Observatory (D. Chipman, personal communications, 2001). A calibration curve based on these four standards was run at the beginning and end of the analysis of the samples from a station. In addition, a check standard (743 pptv) was run after the analysis of every four water samples.

The process of equilibrating the water samples with an introduced headspace involves repartitioning of CO₂ between the liquid and gas phases. This in turn alters the TCO₂ of the water sample and its fCO₂. The fCO₂ measured after equilibration is therefore perturbed from the fCO₂ that the sample would have had if no headspace had been introduced. This effect was corrected for using a mass balance for inorganic carbon based on separate TCO₂ measurements on (not equilibrated) duplicate samples using the SOMMA system and knowledge of the introduced headspace gas content. The correction was made using the apparent dissociation constants of CO₂ in seawater (Roy et al. 1993) and the constraint that the total alkalinity of the water sample remained unchanged during the equilibration. The motivation for the use of variable headspace gases (see earlier discussion) was to minimize the magnitude of these corrections by closely matching the fCO₂ of the introduced headspace to that of the sample. The calculation procedure and associated errors are discussed in detail by Neill et al. (1997).

The mass-balance-corrected results are reported as fCO₂ at both the actual temperature of equilibrium and also, for convenience, at a constant temperature of 20°C. The actual temperature of equilibration is also reported. The correction to 20°C was made using the program CO₂SYS of Lewis and Wallace (1998) using the CO₂ solubility data of Weiss (1974) and the dissociation constants of Roy et al. (1993). Based on measurement of 93 duplicate samples, the precision of the fCO₂ analysis was ~1%.

The samples that were equilibrated in the warmer equilibration bath (30 to 32°C) consistently showed a small positive offset after normalization to a common equilibration temperature when compared with the replicates that were equilibrated in the cooler bath. It was hypothesized that the offset was the result of water condensing out of the headspace gas as it passed through the headspace sampling needle and transfer tubing, thereby increasing the mole fraction of CO₂ within the gas that filled the sample loop that was injected into the gas chromatograph. A theoretical calculation, which assumed that the headspace gas was 100% saturated with H₂O at the equilibration temperature as it entered the needle and was 100% saturated at room temperature as it entered the heated sample loop housing, was in good agreement with the observed offset. A correction for this effect has therefore been applied to all samples that were equilibrated in the warmer bath. Following this cruise, the system was modified to include heated transfer tubing to prevent such condensation.

The Partial pressure of CO₂ (pCO₂) data presented in this NDP was calculated by R. Key of Princeton University using the equations taken from Weiss 1974. Figure 5 summarizes the analytical results as a contour-section plot of the calculated pCO₂ data from the WOCE Section A8 in Atlantic Ocean along 11.3° S.