Calibrations

Before the cruise, a number of titrations were made on CRMs (#12) in the laboratory. The laboratory titration results for TALK, TCO₂, and pH are given in Table 2 along with the assigned TCO₂ and the pH measured relative to TRIS buffers (Dickson 1993) and spectrophotometrically (Byrne and Breland 1989; Robert-Baldo et al. 1985; Millero et al. 1993b; Clayton and Byrne 1993). It should be pointed out that the values of pH are on the seawater scale defined by Dickson (1984):

pH_sws = -log[H⁺]_sws = -log{[H⁺] + [HSO₄^-] + [HF]} (1)

The precision in the values of TALK (±2 µmol/kg), TCO₂ (±5 µmol/kg), and pH (±0.005) was quite good. The titration values of TCO₂ were 18 ± 4 µmol/kg higher than the values assigned and measured by coulometery. The titration values of pH (7.93 ± 0.01) were 0.01 lower than the values measured by spectrophotometric methods (7.940 ± 0.002) and using seawater buffers (Millero et al. 1993b). The differences in pH and TCO₂ are caused by the non-ideal behavior of the electrodes near a pH of 8 (Millero et al. 1993a). Calibration of the electrodes using TRIS seawater buffers yielded a pH of 7.942 ± 0.005 from the initial emf readings of the titration. These results are in excellent agreement with the spectrophotometrically determined pH and show a lower standard error than the values determined from the titrations.

The program used to calculate the TCO₂ parameters assumed that the electrodes would respond to a change in pH with an ideal slope of 59.2 mV at 25°C as determined from the Nernst equation. The slopes of the electrodes using buffers and titrating with HCl frequently gave non-ideal behavior. The parameters produced by varying this slope indicated (Millero et al. 1993a) that the deviations resulting from these changes were much greater for TCO₂ than for TALK. Errors of 1.0 mV in the slope yielded differences in TALK and TCO₂, respectively, of 2.1 and 22.8 µmol/kg when the pK1* was also allowed to vary (Millero et al. 1993a). The values of TALK were not strongly affected by the behavior of the electrodes. The values of TCO₂ and pH determined for the CRMs with the buffer-derived slope (58.4) were in good agreement with the correct values (pH = 7.935 and TCO₂ = 1984 µmol/kg). These calculations indicated that the deviations in the TCO₂ derived from titration were a result of errors in the slope of the electrode, and not a result of unknown protolytes (Bradshaw and Brewer 1988). If the slope determined from the buffers was used, the titrations yielded reliable values of pH, TALK and TCO₂. This fact was used to make sure that the field titration measurements yielded the most reliable values of pH and TCO₂.

During the cruise, the electrodes in each titration system were calibrated with TRIS seawater buffers (Millero 1986) of known pH_sws (8.057) determined with a H₂, Pt/AgCl, Ag electrode (Millero et al. 1993b). Titrations of CRMs (#12) were also made during the cruise. The results are given in Table 3. The average values, TALK = 2229 ± 7 µmol/kg, TCO₂ = 1984 ± 6 µmol/kg, and pH = 7.944 ± 0.01, are in good agreement with the laboratory results. The deviations are larger at sea than obtained in the laboratory (Table 2) but indicate that the titration systems performed well throughout the cruise. The large errors are related to problems in reproducing the volume in the glass cells. Presently, a plastic cell with a more reproducible volume is used, making it possible to reproduce the CRMs to ±2 µmol/kg for TALK, to ±3 µmol/kg for TCO₂, and to ±0.005 for pH (Millero et al. 1993a).

Table 3. Titrations of certified reference materials at sea (Batch #12)

^a Calculated from the initial emf using TRIS buffer calibration.
^b Certified value from weighted titrations.
^c Certified value from manometric extraction technique.
^d Results obtained in the laboratory using spectrophotometric methods.