Total Alkalinity Measurements
Dr. Andrew Dickson's group (SIO) was responsible for the TALK measurements during section A22_2003. Samples for TALK were collected in glass bottles made from Schott Duran® glass. They were preserved by the addition of 0.02% by volume of a saturated mercury (II) chloride solution (DOE 1994 - SOP 01), and analyzed-typically within 24 h-on board ship.
TALK measurements were made using an open-cell, two-stage, potentiometric titration procedure similar to that used to certify reference materials for TALK (see Dickson et al. 2003), except that samples were not weighed into the titration vessel but instead were dispensed using a 120-mL glass syringe. A metal frame attached to the syringe barrel and plunger controlled the maximum extent the plunger could be withdrawn in the barrel. This ensured that a reproducible amount of seawater was dispensed.
The analytical procedure was as follows:
- An aliquot of seawater was dispensed into the titration vessel (a jacketed glass beaker with its temperature controlled to ±0.02 °C at about 20.0 °C), a stirrer bar added, and the temperature probe and burette tip inserted in the solution
- The solution was then acidified to a pH of about 3.6 with a single aliquot of the titration acid and stirred vigorously while CO2-free air was bubbled through for about 6 min to remove CO2
- The main titration was then started and the solution was titrated using 0.05-mL increments to a pH of about 3.0. Data from the pH range 3.5-3.0 were used in a non-linear least squares process that corrects for the reactions with sulfate and fluoride ions to estimate the TALK of the sample-see Dickson et al. (2003) for more details
The equipment used for this is listed in Table 3.
The hydrochloric acid used for the titration was made up in bulk and then stored in 1-L Pyrex bottles with greased ground-glass stoppers. The acid strength was approximately 0.100 mol/kg. The acid was made up in a 0.6 mol/kg sodium chloride background so as to approximate the ionic strength of seawater. Selected bottles of the acid were then analyzed coulometrically (Dickson et al. 2003) to assign a concentration to the batch.
Quality Control and Calibration of Reported Results
The at-sea repeatability of the method was estimated by analyzing duplicate samples, collected on each cast. These results were used to estimate a standard deviation using the standard expression (DOE 1994 - SOP 23). The repeatability was 1.06 µmol/kg based on 89 pairs of analyses.
In addition, analyses were made of the alkalinity of CO2 reference material. These analyses were carried out regularly throughout the cruise, typically a pair of analyses every 12 h. The results are shown in Fig. 2.
An examination of An examination of Fig. 2 suggests that there was a change in the system calibration at around the 100th reference material measurement (October 7, 2003). It seems from a review of contemporaneous notes that the syringe that was being used to dispense the seawater samples for analysis was changed on that date. It also appears that there may be an error in calibrating the volume dispensed from the syringe (or perhaps in the acid concentration value).
It thus seemed appropriate to treat the data in Fig. 2 as comprising two groups: stations 1-37 and 38-88 (i.e., before and after the syringe change):
Δ1-37 = 3.92 ± 1.23 µmol/kg (86); Δ38-88 = 1.44 ± 0.74 µmol/kg (66).
A decision was thus made to also treat the cruise data as being in the same two groups, and to calibrate the reported data by adjusting the measured results so as to correct these CRM results to a Δ of zero. The adjustment chosen was multiplicative (as would be expected if the deviation was indeed due to a poorly calibrated dispensing system). The reported data for stations 1-37 have thus been multiplied by a calibration factor of 0.99822, and those for stations 38-88 by a factor of 0.99935.
Finally, the adjusted alkalinity data results were multiplied by a factor of 1.0002 to correct for the dilution inherent in adding mercuric chloride to the sample to preserve it for analysis.
Once the at-sea alkalinity measurements had been adjusted in this fashion, they were salinity normalized to a salinity of 35 and the resulting values plotted in Ocean Data View to help identify any questionable data. As a result of this analysis, 16 points were identified as either questionable or bad, and flagged accordingly.
Dr. Frank Millero's group of Rosenstiel School of Marine and Atmospheric Science (RSMAS), University of Miami, was responsible for the TALK measurements during section A22_2003. The titration systems (Fig. 3) used to determine TALK, TCO2, and pH consisted of a Metrohm 665 Dosimat titrator and an Orion 720A pH meter that is controlled by a personal computer (Millero et al. 1993b). Both the acid titrant in a water jacketed burette and the seawater sample in a water jacketed cell were controlled to a constant temperature of 25 ± 0.1°C with a Neslab constant temperature bath. The Plexiglas water jacketed cell used is shown in (Fig. 3). These cells had fill-and-drain valves that increased the reproducibility of the cell volume.
The TALK system consisted in a manifold which allows the automated measurement of eight samples. A set of pumps, valves and relays are used to rinse, fill and drain the TALK cell ((Fig. 3)). The titration is controlled programmatically using National Instrument's Labwindows/CVI environment. The titration is made by adding HCl to seawater past the carbonic acid end point. A typical titration records the electro-magnetic fields (emf) reading after the readings become stable (± 0.05 mV) and adds enough acid to change the voltage to a pre-assigned increment (10 mV). In contrast to the delivery of a fixed volume increment of acid, this method gives more data points in the range of rapid increase in the emf near the endpoint. A full titration (25 points) takes about 20 min. Using two automated systems a 32-bottle station cast can be completed in 8 h.
The electrodes used to measure the emf of the sample during a titration consisted of a ROSS 8101 glass pH electrode and an Orion 90-02 double junction Ag/AgCl reference electrode.
The HCl used throughout the cruise were made, standardized, and stored in 500 cm3 glass bottles in the laboratory for use at sea. The 0.23202 M HCl solutions were made from 1 M Mallinckrodt standard solutions in 0.45 M NaCl to yield an ionic strength equivalent to that of average seawater (~0.7 M). The acid was standardized using a coulometric technique by our group and Dickson (Taylor and Smith, 1959; Marinenko and Taylor, 1968). Both results agree to ± 0.0001 M.
The volumes of the cells used at sea were determined in the laboratory by assuming a volume of 200cm3, then running many measurements of seawater with a known TALK. Once the TALK values agree to ± 1 µmol/kg, the known TALK of the sample is used to back-calculate the volume of the cell. The volume is reproducible to ± 0.01 cm3. Measurements of the TALK of CRM throughout the cruise are used to confirm the volume on each cell.
The volume of HCl delivered to the cell is traditionally assumed to have small uncertainties (Dickson 1981) and equated to the digital output of the titrator. Calibration of the burette of the Dosimat with Milli-Q water at 25°C indicate that the system delivers 3.000 cm3 (the value for a titration of seawater) to a precision of ± 0.0004 cm3. This uncertainty results in an error of ± 0.4 µmol/kg in TALK and TCO2. The accuracy of the volume of acid delivered by the Dosimat, however, is ten times bigger than the precision. Since the titration systems are calibrated using standard solutions, this error in the accuracy of volume delivery will be partially canceled and included in the value of cell volumes assigned.
The TALK of seawater was evaluated from the proton balance at the alkalinity equivalence point, pHequiv = 4.5, according to the exact definition of TALK (Dickson 1981)
TALK = [HCO3-] + 2[CO32-] + [B(OH)4-] + [OH-] + [HPO42-] + 2[PO43-] + [SiO(OH)3-] - [H+] - [HSO4-] - [HF] - [H3PO4] (1)
At any point of the titration, the TALK of seawater can be calculated from the equation
(V0 TA - VN)/(V0 + V) = [HCO3-] + 2[CO32-] + [B(OH)4-] + [OH-] + [HPO42-] + 2[PO43-] + [SiO(OH)3-] - [H+] - [HSO4-] - [HF] - [H3PO4] (2)
where V0 is the volume of the cell, N is the normality of the acid titrant, and V is the volume of acid added. In the calculation all the volumes are converted to mass using the known densities of the solutions.
A computer program has been developed in Labwindows/CVI to calculate the carbonate parameters (pHsw, E*, TALK, TCO2, and pK1) in seawater solutions. The program is patterned after those developed by Dickson (1981), Johansson and Wedborg (1982) and DOE 1994. The fitting is performed using the STEPIT routine (J.P. Chandler, Oklahoma State University, Stillwater, OK 74074). The STEPIT software package minimizes the sum of squares of residuals by adjusting the parameters E*, TALK, TCO2 and pK1. The computer program is based on equation (2) and assumes that nutrients such as phosphate, silicate and ammonia are negligible. This assumption is valid only for surface waters. Neglecting the concentration of nutrients in the seawater sample does not affect the accuracy of TALK, but does affect the carbonate alkalinity.
The pH and pK of the acids used in the program are on the seawater scale, [H+]sw = [H+] + [HSO4-] + [HF] (Dickson, 1984). The dissociation constants used in the program were taken from Dickson and Millero (1987) for carbonic acid, from Dickson (1990a) for boric acid, from Dickson and Riley (1979) for HF, from Dickson (1990b) for HSO4- and from Millero (1995) for water. The program requires as input the concentration of acid, volume of the cell, salinity, temperature, measured emf (E) and volume of HCl (VHCl). To obtain a reliable TALK from a full titration at least 25 data points should be collected (9 data points between pH 3.0 to 4.5). The precision of the fit is better than 0.4 µmol/kg when pK1 is allowed to vary and 1.5 µmol/kg when pK1 is fixed. The titration program has been compared to the titration programs used by others (Johansson and Wedborg 1982, Bradshaw and Brewer 1988) and the values of TALK agree to within ± 1 µmol/kg.
The spectroscopic pH and potentiometric TALK of CRM used during the cruise have been measured in the laboratory before the cruise to characterize the pH of the standard and ensure the titration systems were performing to the desired precision. During the cruise, titrations on CRM were made to ensure that the two titration systems were giving consistent values. The values of pH, TCO2, and TALK for CRM No. 61 are summarized in Table 4. The precision of the measurements was ± 3.6 µmol/kg for TALK, ± 3.4 µmol/kg for TCO2, and ± 0.009 for pH. The average values agreed to the certified values to ± 1.8 µmol/kg in TALK, ± 4.7 µmol/kg in TCO2, and ± 0.009 for pH. The deviations in TALK, TCO2 and pH for all the CRMs are shown in Fig. 5, Fig. 6, and Fig. 7. The deviations are within 2σ for most of the measurements. Since the average offset between the TALK measurements of CRM agreed within experimental error, no corrections were made in our field measurements of TALK. A small correction factor was made for TCO2 (a factor of 0.9951 was multiplied to System 2) and pH (a factor of 0.004 was added to System 1 and 0.015 to System 2) to the values for each titration system. The TCO2 measurements made on the titrations system have been compared to the values measured with the SOMMA system on the same samples. These results are shown in Fig. 8. The average differences of the adjusted values of the titration TCO2 on all the measurements made on the cruise agree with the SOMMA values to ± 3.0 µmol/kg, which is within the precision of the measurements. These comparisons indicate that the titration values of TCO2 from the alkalinity systems can yield reasonable values if the system is calibrated with CRM in agreement with earlier studies (Millero et al. 1993b).
Note: the TCO2 and pH values that have been measured on the alkalinity system are not present in the dataset for section A22_2003.
The precision of the instruments was also tested by making duplicate measurements of samples throughout the cruise. These duplicates were taken from the same niskin bottle, equilibrated for an equal amount of time, and then measured on the same instrument. A total of 117 duplicate samples was made, and the results showed that the average delta was 0.2 ± 1.8 µmol/kg for TALK, 0.2 ± 1.3 µmol/kg for TCO2, and 0.000 ± 0.003 for pH. Table 5 and Fig. 9, Fig. 10, and Fig. 11 summarize these results.
The carbonate system is characterized by four parameters: TALK, TCO2, partial pressure of carbon dioxide (pCO2), and pH. Knowing two of these parameters, one can calculate the other two. If more than two parameters are known, a comparison of calculated and measured values will tell if the measured value is internally consistent with the two used in the calculation. We have examined the internal consistency of pH and TALK measurements and the SOMMA values of TCO2. The "CO2sys.bas" basic program used to make these calculations was written by Lewis and Wallace (1998) and modified by Denis Pierrot to run in Excel. We used the carbonic acid constants of Mehrbach (1973) and refit by Dickson and Millero (1987) for all calculations, as well as the constant of Dickson (1990b) for bisulfate all on the seawater pH scale. We examined an input of pH and TALK to calculate TCO2, pH and TCO2 to calculate TALK, and TALK and TCO2 to calculate pH. The results of these calculations are summarized in Table 6 and the deviations are shown in Fig. 12, Fig. 13, and Fig. 14.
Once the data have been proven accurate and precise, as well as internally consistent, a comparison of the 1997 and 2003 cruises were made.
In looking at the changes in the carbonate system of the ocean, the surface measurements are of the greatest concern. These values change much more than do the deep water values. The following figures are a comparison between the 1997 and 2003 cruises along section A22 showing a variety of measurements versus latitude. Fig. 15 details the salinity measurements that were obtained from the ship's CTD measurements. The overall trend of salinity seems to remain fairly consistent. Fig. 16 shows the values for temperature, which were also obtained from the ship's CTD measurements. The overall trend for the temperature is slightly askew because the 1997 cruise took place in late summer and the 2003 cruise took place in the fall. Fig. 17 compares the normalized TALK of the two cruises (normalized TALK meaning corrected to a salinity of 35, or NTA = TA × 35/S, see Millero 1996). The overall trend of NTA remains fairly consistent between the two cruises. Fig. 18 shows the normalized TCO2 (normalized meaning corrected to a salinity of 35, or NTCO2 = TCO2 × 35/S, see Millero 1996). The trend has remained consistent, but the concentration of TCO2 has increased due to the uptake of anthropogenic carbon dioxide from industrialized countries. This effect is more predominant in colder or northern waters as these waters are able to hold more dissolved gases. Fig. 19 and Fig. 20 show the in situ pH and in situ pCO2 of the two cruises, respectively. These values were calculated from the temperatures and pressures at which the samples were taken, with pCO2 being calculated from the TALK and TCO2. The figures are basically inverses of one another, because carbon dioxide acts as a weak acid; therefore, as the concentration of pCO2 increases, the pH decreases and vice versa.
After reviewing the accuracy, precision and internal consistency of the data, we feel very confident in the level of quality of these data. In comparison with the 1997 cruise, it is very clear that the anthropogenic output of CO2 has been increasing in the atmosphere, and in turn, has increased the concentration of TCO2 and pCO2 in the surface of the ocean. This effect is more pronounced in northern latitudes where there is colder water, but the increase is also noticeable in the equatorial regions. The increase in CO2 has caused a slight decrease in the pH, because CO2 acts as a weak acid, effectively lowering the pH. However, there were no significant findings in increase or decrease of the TALK. Such changes in the oceans need to be constantly monitored to better understand and predict what may happen due to human activity on earth.