TCO2 was determined using two automated dynamic headspace sample processors (SOMMAs) with coulometric detection of the CO2 extracted from acidified samples. A description of the SOMMA Coulometry System and its calibration can be found in Johnson et al. (1987); Johnson and Wallace (1992); and Johnson et al. (1993). A schematic diagram of the SOMMA analytical system and sequence may be found in earlier publications (Johnson et al. 1993), and further details concerning the coulometric titration can be found in Huffman (1977) and Johnson et al. (1985). Samples were collected in 300-mL precombusted (450±C for 24 h) glass standard Biological Oxygen Demand (BOD) bottles, poisoned with 200-ul of a 50% saturated solution of HgCl2, and analyzed for TCO2 within 24 hours of collection (DOE Hand Book of Methods, 1994). Before analysis, they were stored in a refrigerator in darkness at ~15±C until analyzed. Analyses of duplicate samples separated in time by up to 8 hours showed no evidence of any significant biological consumption or production of CO2 during storage under the above conditions. CRMS were routinely analyzed according to DOE prescribed methods (1994). The CRMs were supplied by Dr. Andrew Dickson of the SIO, and during Section P6 batches 10 and 11 were used. The certified values for batch 10 were S =34.5722 and Certified TCO2=1960.67 ± 0. 39 µmol/kg (n=5). The corresponding numbers for batch 11 were S=38.5 and TCO2 =2188.77 µ 0.56 µmol/kg (n=5). The CRM TCO2 concentration was determined by Vacuum-Extraction/Manometry in the laboratory of C.D. Keeling at SIO.
Seawater introduced from an automated "to-deliver" pipette into a stripping chamber was acidified, and the resultant CO2 from continuous gas extraction was dried and coulometrically titrated on a model 5011 UIC Coulometer with a maximum titration current of 50 mA in the counts mode [the number of pulses or counts generated by the Coulometer's voltage to frequency converter (VFC) during the titration was displayed]. In the coulometer cell, the acid (hydroxyethylcarbamic acid) formed from the reaction of CO2 and ethanolamine was titrated coulometrically (electrolytic generation of OH-) with photometric endpoint detection. The product of the time and the current passed through the cell during the titration (charge in Coulombs) is related by Faraday's constant to the number of moles of OH- generated and thus to the moles of CO2 that reacted with ethanolamine to form the acid. The age of each titration cell was logged from its birth (time that electrical current is applied to the cell) until its death (time when the current is turned off). The age was measured in minutes from birth age was measured in minutes from birth (chronological age) and in mgC titrated since birth (carbon age).
Each system was controlled with an IBM-compatible personal computer equipped with two RS232 serial ports (coulometer and barometer), a 24-line digital Input/Output (I/O) card (solid state relays and valves), and an Analog to Digital (A/D) card (temperature, conductivity, and pressure sensors). The A/D cards were manufactured by Real Time Devices (State College, Pa.). The temperature sensors (model LM34CH, National Semiconductor, Santa Clara, Calif.), with a voltage output of 10 mV°F built into the SOMMA were calibrated against thermistors certified to 0.01°C (PN CSP60BT103M, Thermometrics, Edison, N.J.) using a certified mercury thermometer as a secondary standard. These sensors monitored the temperature of SOMMA components including the pipette, the gas sample loops, and the coulometer cell. The SOMMA software was written in GWBASIC Version 3.20 (Microsoft Corp., Redmond, Wash.), and the instruments were driven from an options menu appearing on the PC monitor. Since the coulometers operated in the counts mode, conversions and calculations were made using the SOMMA software rather than the programs and the constants hardwired into the coulometer circuitry.
The "to-deliver" volumes (Vcal) of the sample pipettes were determined (calibrated) gravimetrically prior to the cruise and were checked periodically during the cruise by collecting aliquots of deionized water dispensed from the pipette into preweighed serum bottles. The serum bottles were crimp sealed and weighed immediately during the on-shore laboratory calibrations, or were returned to shore and reweighed on a model R300S (Sartorius, Gottingen, Germany) balance as soon as possible. The apparent weight (g) of water collected (Wair) was corrected to the mass in vacuo (Mvac) from Mvac = Wair + Wair (0.0012 / d - 0.0012 / 8.0) , where 0.0012 is the sea level density of air at 1 atm, d is the density of the calibration fluid at the pipette temperature and sample salinity, and 8.0 is the density of the stainless steel weights. The "to-deliver" volume was Vcal = Mvac / d . The calibration volumes (Vcal) at the calibration temperature (tcal) of the SOMMA System pipettes for the three P6 Sections are given in Table 4.
The sample volume (Vt) at the pipette temperature was calculated from the expression Vt = Vcal [1 + av (t - tcal)] , where av is the coefficient of volumetric expansion for Pyrex-type glass (1 X 10 -5/°C) and t is the temperature of the pipette at the time of a measurement. Table 4 shows a small decrease in Vcal for system 006 over time. This is consistent with other systems used daily for periods exceeding 30 days (Johnson et al. 1998a). 1998a). The mean pipette temperature (t) during the P6 cruise was 15.12 ± 0.41°C (n=3580).
The Coulometers used to detect CO2 were periodically electronically calibrated as described in Johnson et al, (1993, 1996) and DOE (1994) throughout the P6 Section. For the calibration, at least two levels of current (usually 50 and 2 mA) were passed through an independent and very precisely known resistance (R) for a fixed time. The voltage (V) across the resistance was continuously measured, and the instantaneous current (I) across the resistance was calculated from Ohm's law and integrated over the calibration time. Then the number of pulses (counts) accumulated by the VFC during this time was compared with the theoretical number computed from the factory calibration of the VFC [frequency = 105 pulses (counts) generated per second at 200 mA] and the measured current. If the VFC was perfectly calibrated at the factory, the electronic calibration procedure would yield a straight line passing through the origin with an intercept (Intec) of 0 and a slope (Slopeec) of 1. The factory-calibration of the VFC and the value of the Faraday (96489 Coulomb/mol) yields a scaling factor of 4.82445 X 103 counts/µmol, and the theoretical number of micromoles of carbon titrated (M) after extraction from water samples or the gas loops was M = [Counts / 4824.45 - (Blank X Tt) - (Intec X Ti)] / Slopeec, where Tt was the length of the titration in minutes, Blank is the system blank in µmol/min, Intec the intercept in µmol/min, and Ti the time in minutes during the titration where current flow was continuous. Note that the slope obtained from the electronic calibration procedure applied for the entire length of the titration but the intercept correction applied only for the period of continuous current flow (usually 3 to 4 min) because the electronic calibration can only be carried out for periods of continuous current flow. The results of the electronic calibrations are given in Table 5.
The SOMMA-Coulometry systems were also calibrated with pure CO2 (calibration gas) using hardware consisting of an 8-port gas sampling valve (GSV) with two sample loops of known volume (determined gravimetrically by the method of Wilke et al. 1993) connected to the calibration gas through an isolation valve with the vent side of the GSV plumbed to a barometer. When a gas loop was filled with CO2, the mass (moles) of CO2 contained therein was calculated by dividing the loop volume (V) by the molar volume of CO 2 at the ambient temperature (T) and pressure (P). The molar volume of CO2 [V(CO2)] was calculated iteratively from T, P, and the first viral coefficient B(T) for pure CO2: V(CO2) = RT / P X 1 + B(T) / V(CO2). The gas calibration factor (CALFAC) the ratio of the calculated mass to that determined coulometrically was used to correct the subsequent titrations for small departures from 100% recoveries (DOE 1994). Pressure was measured with a barometer, model 216B-101 Digiquartz Transducer (Paroscientific, Inc., Redmond, Wash.) that is factory-calibrated for pressures between 11.5 and 16.0 psia. The standard operating procedure was to make gas calibrations daily for each newly prepared titration cell [normally, one cell per day and three sequential calibrations per cell at a carbon age of 3 to 6 mgC with the result of the third calibration taken as the CALFAC if consistent with the second (i.e., agreement to ±0.1% or better)]. The CALFAC data for the P6 section are summarized in Table 6.
For water samples, the TCO2 concentration in \265mol/kg was calculated from TCO2 = M X CALFAC X (1 / (Vt X p )) X dhg, where p is the density of seawater in g/mL at the measurement temperature and sample salinity calculated from the equation of state given by Millero and Poisson (1981) and dHg is the correction for sample dilution with bichloride solution (for P6 dHg = 1.00066).
As noted above, the daily CALFAC determined for System 004 on the P6E Leg was too high (indicating lower recovery of CO2), and when substituted into the last equation it led to over-estimates of the CRM TCO2 concentration by 3 to 4 µ mol/kg (> or = 0.1%). Unfortunately, the cause of this problem was not discovered until a later cruise aboard the R/V Meteor (WOCE Section A10) in early 1993 when a leaky plumbing fitting was found and replaced on System 004 as described by Johnson et al. (1998b). For convenience, the line carrying the calibration gas (CO2) to the GSV had been plumbed (prior to the P6 Section) with a tee connection on the upstream side of the GSV with one branch connected to the GSV through an isolation valve (IV) and the other branch to a Quick-Connect Fitting (Swagelok, Crawford Fitting). This plumbing configuration facilitated the rapid connection of the calibration gas to an external flow-meter and flow rate adjustments, however, the quick-connect fitting apparently allowed a very small amount of air to infiltrate into the calibration gas line slightly diluting the calibration gas. The Quick-Connect and tee fittings were replaced early on the A10 Section and System 004 was successfully gas-calibrated thereafter.
For the P6E and P6C Legs, a "CRM-based calibration factor" was also calculated for System 004 by taking the resulting counts for the first CRM analyzed on each cell and substituting it along with the certified TCO2 into the last equation and solving it for CALFAC. At the end of each leg, a leg-specific mean CRM-based CALFAC was calculated for System 004, and these data along with the mean CALFAC determined for System 006 are also given in Table 6. The TCO2 measurements from sample analyses made on SOMMA 004 were calculated using the mean CRM-based CALFAC shown in Table 7, while TCO2 on SOMMA 006 was calculated using the CALFAC determined daily using pure-CO2 gas for each new cell born according to DOE (1994).
Taken together, Tables 4 ("to-deliver" pipette volume), 5 (electronic calibration), and 6 (gas or CRM calibration) show that the response of Systems 004 and 006 remained constant throughout the three legs of the P6 Section. In addition, water samples were collected periodically in duplicate and one of the duplicates was analyzed on each system (see Table 4).
The SOMMA 006 was equipped with a conductance cell (Model SBE-4, Sea-Bird Electronics, Inc., Bellevue, Wash.) for the determination of a salinity measurement as described by Johnson et al. (1993). Whenever possible SOMMA and CTD salinity were compared to ensure that the salinity of the analyzed samples matched the CTD salinity. The agreement between CTD and SOMMA salinity was 0.04 or better, and several mistrips of the Niskin bottles were quickly identified using this comparison.
Quality Control-Quality Assurance (QC-QA) was assessed from the results of 293 CRM analyses made on Systems 004 and 006. The mean differences between the measured and certified TCO2 (measured amount minus certified amount) were calculated for each leg, and the mean differences are summarized in Table 7. The individual differences for the Section P6 are plotted in Fig. 5.
The accuracy of the CRM analyses was much better than 1 µmol/kg on both systems throughout the P6 Section, and the overall mean difference is -0.16 µmol/kg (n = 293). The precision of the CRM determination on the P6 Section is the standard deviation of the difference between determined and certified TCO2 (±1.49 µmol/kg). The imprecision of the CRM analyses was slightly higher on the P6C Section for both systems, whereas the best precision was obtained on the final Leg P6W on System 006. There were two CRM analyses during Section P6 that were considered to be outliers, and these results have not been included in Table 7 or Fig. 5. An outlier was defined as an analytical difference between the measured and Certified TCO2 which exceeds ±5.0 µmol/kg. The two outliers were measured on System 006: CRM No. 171 on June 6 at a carbon age of 38.8 mgC; and CRM No. 170 on June 21 at an age of 8.6 mgC. Additional CRMs were analyzed on both cells to within 1 µmol /kg of the certified TCO2 even at very advanced carbon ages (55.5 and 46.1 mgC, respectively). Hence, the cause of the outlier values was likely not the behavior of the coulometric titration or cell, but may have been related to a temporary problem with the sample delivery system (pipette). Alternatively, the CRM may have been compromised during preparation or storage. Sample duplication with these cells was observed to be excellent, and accordingly the data obtained with them has been included in the data set described herein.
The second phase of the QC-QA procedure was the assessment of sample precision on each system (single-system precision) and the assignment of an overall precision to the P6 TCO2 samples. This was the second cruise where two independent SOMMA systems were deployed side-by-side, and the conventions employed for the estimation of precision given for WOCE Sections A1E and A10 data (Johnson et al. 1996; 1998b) have been retained in Table 8.
The single-system precision was determined from samples with duplicates analyzed on the same system (either 004 or 006). The sample precision was calculated using duplicates that were analyzed on both systems (004 and 006).
Single-system and sample precision have been separately assessed in Table 8 as:
Single-system precision provides a measure of drift in system response during a sequence of sample analyses. This is because the time elapsed between duplicate analyses on the same system using the same coulometer cell was deliberately kept at between 3 and 12 hours. Any temporal drift in system response would therefore be reflected in the single-system precision by decreased precision of the duplicate analyses. Sample precision, on the other hand, provides an estimate of overall sample precision for the section(s) independent of which analytical system was used. It was estimated because TCO2 data were measured using two separate systems during the cruise. Sample precision is the most conservative estimate of precision, incorporating several sources of random or systematic (bias) error including errors associated with the inability to gas calibrate System 004.
It should be noted that the sample precision calculation includes the results for some samples (K = 13) that were analyzed in triplicate (i.e., two replicates analyzed on one system and the third on the second system). For these cases, the mean of the two replicate analyses was used for calculating Sp2. Averaging the replicate results reduced the degrees of freedom term by making n = 1 for each system no matter how many replicates were actually run on each system. Without averaging (d.f. = 50), Sp2 was ±1.50 µmol/kg, whereas with averaging (Table 8, d.f. = 37) Sp2 increased to ±1.65 µmol/kg. For the P6 Section, the more conservative estimate of ±1.65 µmol/kg was chosen as the precision of the TCO2 determination. This was in excellent agreement with the precision of the CRM determination (±1.49 µmol/kg).
As with other sections completed by the BNL measurement group using SOMMA-Coulometer systems run in parallel, the sample precision was slightly less precise than the single- system precision (1.65 µmol/kg vs < 1.0, respectively). This indicates that any error introduced from changes in system response (drift) during the coulometer cell lifetime were within the overall precision of the method. The excellent agreement between "between-sample" and "between-Niskin" precision suggests that there were no significant analytical effects caused by the gas exchange with the overlying headspace of the Niskin bottles during the on-deck sampling. This is consistent with the data collected during other cruises (Johnson et al. 1996; 1998b). The P6 sample precision (±1.65 µmol/kg) was also in good agreement with the sample precision for the BNL WOCE Sections A1E (±1.65 µmol/kg), A10 (±1.92 µmol/kg), A8 (±1.17 µmol/kg), and the North Atlantic sections (±1.59 µmol/kg).
The difference between sample and single-system precision may be the result, in part, of the inability to perform daily gas-calibrations on System 004 during P6. Use of a single average CALFAC for an entire leg may, for example, have masked real cell-to-cell variations in CALFAC. This would have the effect of increasing imprecision but not necessarily altering overall accuracy (the positive and negative differences would cancel) as is suggested by Table 9. Table 9 provides additional proof of the overall accuracy and the absence of a bias between the two systems. Each system yields, within statistical precision, the same result for the same samples even though one system (006) was gas-calibrated daily and the other (004) was not (see also Tables 6, 7, and 8). Table 6 shows that the two calibration procedures gave stable and nearly identical results during the entire P6 Section. However, analyzing independent water sample duplicates on each system is the definitive test for accuracy, precision, and the calibration procedures used. The accuracy of System 004 was validated in this way. This is similar to the situation previously described for the A10 Section (Johnson et al. 1998b) where only one of the two systems used possessed a gas-calibration unit and the gas-calibrated system served as the reference system.
The final step in the QC-QA procedure was the ship-to-shore comparison. Here sample duplicates (commonly called the "Keeling Samples") were analyzed "in real time" at sea by continuous gas extraction/coulometry and later, after shipment and storage, on-shore by vacuum extraction/manometry at the laboratory of C. D. Keeling at SIO (Guenther et al. 1994). The "Keeling Samples" were collected in specially provided threaded 500-mL glass bottles with 4 mL of headspace volume, poisoned with 100 µL of a saturated HgCl2 solution, and then sealed air-tight with a greased ground glass stopper that was secured to the bottle with a threaded plastic screw cap. The latter was bored out to fit over the top of the stopper and mated to the bottle threads so that an air-tight seal was made by gently tightening the cap until a secure seal between the stopper and bottle was achieved. This procedure was carried out with 21 samples collected at 15 stations during P6. The results of the comparison are given in Table 10.
The mean ship-to-shore analytical difference was -2.64 µmol/kg (n = 21). The lower ship- based results for P6 are consistent with the ship-to-shore comparisons from Sections A9, A1E, and A10 previously reported (Johnson et al. 1995; 1996; 1998b). This negative bias for water samples was greater than the sample precision and the analytical difference observed for the CRM analyses (Wallace 2001). The reason for the tendency of the ship-based results to be lower than the shore-based results is not known at this time.
In aggregate, Tables 7-10 show an internally consistent data set with excellent accuracy, high single-system precision (<±1.0 µmol/kg), and a slightly higher imprecision for the sample precision (±1.65 µmol/kg). Based on Tables 7-10 and following the precedent of previous data submissions no correction for instrumental bias or CRM analytical differences has been applied to the sample data. Fig. 6 summarizes the analytical results as a contour section plot of the TCO2 data from the WOCE Section P6 along 32.5°S.
tbeaty 4/19/2002