Specific guidelines for measuring dissolved oxygen concentrations in seawater were given by WOCE (Culberson 1991) to maximize the quality of the data collected. Oxygen concentrations were to be determined titrametrically using the Carpenter modification of the Winkler (1888) method. The techniques developed by Carpenter (1965) maximized overall precision and accuracy and were reproducible (two standard deviations) to within +/- 0.4 µmol/kg (Culberson 1991), which translates to a precision of approximately 0.2 % of air saturated dissolved oxygen concentrations.

     To obtain an accurate estimate of analytical precision for dissolved oxygen samples collected during hydrographic cruises, WOCE protocols state that at least 10% of the total number of samples should be duplicates and that the oxygen concentrations of the duplicates should encompass the entire range of observed values (Culberson 1991). The precision of the duplicates (two standard deviations) must be < 0.5% of the largest dissolved oxygen concentration found, and oxygen data submitted to WOCE must include the standard deviation and the number of replicates with which it was calculated (Culberson 1991).

The basic methodology for measuring bottle dissolved oxygen in HOT samples has remained the same and relies on the Winkler titration (Winkler 1888). In essence, the dissolved oxygen present within a seawater sample is coerced under alkaline conditions to quantitatively oxidize divalent manganese to a trivalent state. The solution is then acidified, which converts iodide ion to iodine in an amount stoichiometrically proportional to the amount of dissolved oxygen contained within the original sample. The amount of iodine is then determined by titration with a thiosulfate solution of known concentration. The end result of the chemical reactions involved is that one mole of dissolved oxygen in the seawater sample will manifest as four moles of thiosulfate, and the original oxygen concentration of the seawater sample can be calculated.

     The Carpenter method of dissolved oxygen concentration determination (Carpenter, 1965) minimized the amount of error that was introduced into the titration process from the volatization of iodine and the difference between the titration end point and the equivalence point. Primarily through the optimization of reagent concentrations and the use of whole bottle titrations, Carpenter was able to achieve accuracies better than 0.1%.

     Accuracy of the dissolved oxygen titrations is difficult to estimate as there are currently no reference standards available for dissolved gases in seawater (NRCNACROS, 2002) due to the difficulty of preparing a stable solution. Carpenter (1965) created his own reference solutions to measure accuracy by dissolving a known quantity of pure oxygen into a volume of deoxygenated water. The process required very controlled conditions as well as generous quantities of mercury making this process impractical for routine use. For practical purposes, a process of standardization is followed instead.

     The Winkler titration method is an indirect measurement for dissolved oxygen determination. Due to the lack of a stable dissolved gas reference standard (Sect. 2.1.3), standardization of the process is based upon a potassium iodate (KIO3) solution prepared to a known normality (usually near 0.02 N for measurement of samples collected on HOT cruises [Karl et al. 1990]). Solutions of the KIO3 standard are then used to determine the exact normality of the sodium thiosulfate (Na2S203.5H2O) titrant (~0.1 N). According to the protocols of the HOT-BEACH group, a commercially available reference KIO3 standard (CSK Standard) is titrated to verify the results obtained with the first standard (Karl et al. 1990).

     There have been two significant changes in the methods employed to determine dissolved oxygen concentrations in HOT seawater samples. The first change was the adoption of the Carpenter modification beginning with samples from the eleventh cruise (HOT-11). The second change began with HOT-31 when the way in which the end of the titration was determined was switched from a visual to a potentiometric method.

2.2.1 Switch from Strickland and Parsons to Carpenter

     Up through HOT-10, dissolved oxygen concentrations were determined using the Winkler methods given by Strickland and Parsons (1972). With this method, 50 mL aliquots of a sample collected in a 300 mL BOD bottle were titrated, and the effects of sample temperature were not considered.

     Beginning with HOT-11, the Carpenter method was employed. This method primarily differs from the Strickland and Parsons method in two ways: the amount of sample collected and titrated, and the consideration of the sample temperature. The Carpenter method calls for the collection of a sample in a 125 mL BOD bottle, then the titration of the entire bottle. The method also requires that the temperature of the sample be measured at the time it is collected so that density effects related to temperature change could be accounted for.

2.2.2 End Point Determination and Automatic Titration

     The second major change in the HOT methods of dissolved oxygen measurement was the switch from a manual visual endpoint to an automated potentiometric method. Titrations of samples analyzed before HOT-31 were considered complete based upon color changes of the sample. As the sample neared a neutral pH, the yellow color of the tri-iodide faded to clear. The visual cue was enhanced with the addition of a starch indicator solution that turned the sample a dark blue which faded to clear at the end of the titration.

     Beginning with samples collected during HOT-31, the endpoint of the titration was determined potentiometrically with an electric probe. The titration was considered complete when the change in the redox potential of the sample reached a minimum.

     It was at this same time (HOT-31) that the addition of titrant became controlled by computer software. The probe measurements are continuously monitored by a program that varies the amount of titrant added to the sample based upon the rate of change of the redox potential (mVolts) of the sample. Smaller amounts of titrant are added from a Dosimat as the titration nears completion. The software alerts the technician performing the measurements when the process is finished, and the software calculates the corresponding dissolved oxygen concentration of the bottle sample.

Back to Table of Contents

Previous Next