SUMMARY: Seawater is collected from known depths using CTD-rosette sampling protocols. Subsamples are drawn into precalibrated iodine flasks and dissolved oxygen is chemically bound by the formation of a manganese (III) hydroxide floc. The floc is subsequently dissolved under acidic conditions which stochiometrically converts the original dissolved oxygen (DO) oxidizing equivalents to triiodide. The latter is quantitatively titrated with sodium thiosulfate to a potentiometric end-point using a high-precision computer-controlled titration system.

1. Principle

The oxygen content of seawater is a fundamental measurement in oceanography providing information which can elucidate water mass movements, net primary productivity, atmosphere-ocean interactions and carbon remineralization processes. The oxygen content of a seawater sample is largely determined by a balance between: (a) the exchange of atmospheric oxygen with the upper mixed layer, (b) net increases due to photosynthetic processes and (c) net decreases due to respiratory demands and heterotrophic processes. From an oceanographic perspective, the measurement of dissolved oxygen is a parameter of fundamental importance.

In this procedure, a divalent manganese solution and a strong alkali are added to the water sample which results in the formation of a floc (1). In the presence of oxygen the manganese II is oxidized to manganese III (2). On subsequent acidification (1 < pH < 2.5), the manganese hydroxide floc dissolves, and I2 is produced in stoichiometric proportion to the original O2 concentration (3). The iodine, in the form of triiodide (4), is titrated with a standardized thiosulfate solution (5), and the oxygen content is calculated from the quantity of thiosulfate consumed.

(1) Mn2+ + 2OH ---------- Mn(OH)2
(2) 2Mn(OH)2 + ½O2 + H2O ---------- 2Mn(OH)3
(3) 2Mn(OH)3 + 2I- + 6H+ ---------- 2Mn2+ + I2 + 6H2O
(4) I2 + I- ---------- I3-
(5) I3- + 2S2O32- ---------- 3I- + S4O62-

2. Precautions

It is extremely important to prevent contamination of the sample with atmospheric oxygen during sampling, fixation and storage. Drawing tubes and sample flasks must be free of bubbles (3.1 and 3.2.1), as must the autopipette system which dispenses the reagents (3.2.2). If the water samples cool significantly or the seal dries during storage, air can infiltrate the flask. The rims of the flasks are filled with seawater and the samples are stored in a location where temperature fluctuations are minimized (3.4).

The solubility of oxygen in seawater is temperature-dependent and deep samples may warm up as they are brought to the surface (i.e. through the thermocline). It is therefore necessary to measure the temperature in the Niskin bottle at the time the sample is drawn in order to calculate the oxygen concentration in situ (3.3). Oxygen samples are always the first samples drawn from the Niskin bottles.

3. Field Sampling

3.1. Sample collection
3.1.1. Subsamples are drawn as soon as the rosette arrives on deck. DO samples are the first samples drawn from the water sampling bottle.
3.1.2. The drawing tube is flushed with sample and any bubbles are displaced. The tube is then inserted (with water flowing) almost to the bottom of the sample flask. Care is taken to avoid creating turbulence in the flask in order to minimize the intrusion of atmospheric oxygen.
3.1.3. The flask is overflowed with two to three volumes of sample. With the sample flowing, the tube is slowly withdrawn from the flask so that the bottle remains brimful when the tube is completely withdrawn.
3.2. Sample fixation
3.2.1. Before fixing the sample, the iodine flask is carefully examined for any bubbles that may have adhered to the walls. If any bubbles are present, the sample is discarded and drawn again.
3.2.2. Immediately after examining the flask, the two fixing reagents (manganous chloride and alkaline iodide) are added simultaneously using two 1 ml autopipettes mounted side- by-side and activated by a single injection lever arm. Care is taken so that bubbles are not added in this process. The pipette tips are positioned at least 3 cm below the surface of the sample, the lever arm is pressed slowly and gently and released while the pipette tips are still submerged, and the pipettes and tubing are checked frequently for bubbles.
3.2.3. The glass stopper is carefully inserted and the flask is shaken vigorously for at least 20 seconds with a rapid wrist action (it is extremely important that the floc be dispersed throughout the flask).
3.3. Sample temperature
3.3.1. The temperature of the water in the Niskin bottle is measured and recorded immediately following sample collection and fixation. This is accomplished by a second person to reduce the time interval between sampling and temperature determination.
3.3.2. An insulated holding container is rinsed thoroughly and filled with the sample. The temperature is then measured with a NBS traceable quick-response digital thermometer.
3.3.3. In situ temperature is measured at the time of bottle trip using a calibrated thermistor (see Chapter 4).
3.4. Storage
3.4.1. Approximately 20 minutes after the samples are collected, the sample bottles are shaken again to resuspend the floc. The flared mouth of the flask is then filled with seawater and bottles are stored in a cool location where temperature fluctuations are minimized.

4. Analysis

4.1. Computer controlled potentiometric titration

The computer program (DO05HOT.BAS) is written in GW Basic and located in the DO subdirectory on the AT 286 hard drive. This program interfaces and operates the Dosimat (model 665) and pH meter (Orion model 720A or EA940). The program parameters are chosen to optimize the equivalence point determination based on the computer controlled addition of µl quantities of titrant. The titration time is approximately 10 minutes per sample, which is a function of the response time of the electrode and the preset electrode stability criteria. The titration time is minimized by the manual addition of titrant by the operator until the color of the sample is a very light yellow (almost clear). At this point the computer-controlled functions are enabled. The computer terminates the titration when the end point criteria are met. The following steps outline the Incremental mv Titration program operation:

1. After the computer has booted and all peripheral instruments are connected to the computer and turned on, change directories to C:\DO. The titration program and GWBasic reside here. To initialize the program, type "GWBASIC DO05HOT" and return.
2. If the system is ready to go (i.e., all connections are "OK" and all peripherals are working) the program will list a description of the task keys available. When you are done with this screen, press return.
3. You will be presented with the option to load preset parameters into the program. For work at Sta. ALOHA you should answer "Y" or "y".
4. Next you will be asked to input the name of the file containing the parameters. The default answers are ALOHA05 and C:\DO. These default answers should be accepted for routine work at Sta. ALOHA.
5. The program will list the default operational parameters which have been found to optimize the sensitivity and precision of the equivalence point determination at Sta. ALOHA and will prompt you to either accept the default values or change them. In most instances the statement is self-explanatory.
Statement Comments
Sodium thiosulfate normality (N) = 0.05 This is the standardized normality of the Na2SO3 titrant. This together with the change in mv and volume added determines the slope.
Buret size (ml) = 5 Capacity of the Dosimat buret on the exchange unit.
Titration increment size (ml) = 0.02 This is the initial volume of titrant added before the first slope criteria is enabled.
Preadd volume (ml) = 0 This is set at 0 since we add the preadd manually.
Total vol. of titrant (ml) = 4 This is one of the three titration end point criteria. If 4 ml of titrant is reached, the titration is terminated.
Stop potential (mv) = 100 This is the second of the three end point criteria. If the potential reaches 100 mv, the titration is terminated.
Electrode drift time (sec/.2mv) = 5 Criteria for electrode stability.
First slope increment change (mv/meq) = This is a calculated value which controls the reduction in titrant increment size (300 mv/ titrant normality).
Second slope increment change (mv/meq) = calculated (750 mv/titrant normality)
Third slope increment change (mv/meq) = calculated (3000 mv/titrant normality)
First increment volume change (ml) = .01 Titrant volume increment when first slope criteria is enabled.
Second increment volume change (ml) = .002 Titrant volume increment when second slope criteria is enabled.
Third increment volume change (ml) = .0005 Titrant volume increment when third slope criteria is enabled.
6. If you wish to accept these values (which you should for routine work at Sta. ALOHA), then enter a "Y" or "y" and return. If you do not want to accept the default parameters enter a "N" or "n". You will have to input all values except those calculated.
7. Next you will be asked to input:
(a) the date of analysis
(b) the pH meter model used
(c) if you have a printer connected to the computer
(d) if you want a detailed or summarized printout of the titrations (use summarized for routine work at Sta. ALOHA).
(e) if you want an alarm to be activated at the end of each run
(f) the path of data storage (C:\DO\HOT##)
(g) the sample name. This is an 8 character max file name and for standardization purposes the following convention is used. For blanks and standards an alphanumeric designation is used which represents the date and sample type. For samples, an alphanumeric system is also used which represents the iodine flask number and the HOT cruise number (see examples below, assume the date of the analysis is 3/25/92 for these examples):
Reagent Blank 3252bl1 (this creates a prn file with the name 3252bl1.prn)
Primary KIO3 Standard 3252std1 (if you run more than 1 std then number sequentially)
CSK Certified KIO3 Standard 3252csk1
DO Sample 1H36 (iodine flask number followed by HOT cruise number).
DO Sample Backtitrated 1BH36
Different DO Sample analyzed in a previously used flask 1AH36
(h) your name or ID
8. At this point a statement will appear on the screen instructing you that the Dosimat is now in the manual titration mode and to titrate your sample to a pale yellow color. To reduce the titration time, manually titrate to a very pale yellow (almost clear), then press "x".
9. From this point on the computer controls the titration, adjusting the volume of titrant added in response to the slope criteria. The titration is terminated when one of three events occurs. Either the volume of titrant or mv criteria is reached or the calculated slope increases four consecutive times.
10. At this point, the results are printed out and you are prompted to make any comments which are specific to that sample. Record any problems, observations or concerns. This is a great help in interpreting the data. Entering return will close that file and you will be prompted if you would like to run another sample. By entering "Y" or "y" the process begins again. If you do not want to run any more samples press "N" or "n" and you will see "ok". Type "system" to return to the C: prompt. Type "run" to start the program again. Normally the program is not terminated until the end of the cruise. If all of the samples have been run but more are to be collected, do not terminate the program (i.e., do not enter "n"; leave it as is).

Other concerns:

(a) Backtitrating

The program adds incremental volumes of titrant to the sample based on selected slope criteria. If the operator gets too close to the equivalence point or passes it, during the manual titration step, then the program will not reach its end point criteria and will continue to run. To avoid this, check the mv reading. For our samples this reading should be approximately 320-360 mv at the end of the manual addition. If you find that you have titrated just past this point, go ahead and enable computer controlled titration and observe 3 or 4 cycles checking the slope. It should decrease to a minimum value (~-500,000 to -900,000), and then increase. If it initially increases and the mv readings are below 300 mv then you have gone too far. In this case, press F5 and add volumetrically a known amount of primary standard. Ideally, you want to add an amount which is equivalent to the normal manual addition end point. I have found this to usually be around 100- 200 µl, depending on how far you went over or how long the program ran. The program will have continued to add 20 µl if you passed the end point during the manual addition. If you walk away and come back 1/2 hr later, it will still be adding titrant at 20 µl every 2-3 minutes. In this case the 200-300 µl of titrant have been added. Therefore, you need to add enough primary standard to equal this amount plus an amount equal to the manual addition end point. Our primary standard is about 0.025 N and the titrant about 0.05 N so you would need to add 400-600 µl of primary standard to equal the added titrant, plus approximately 100 µl to reach the manual addition end point. From this point, the program should function properly.

(b) Interrupting an analysis in progress

If for some reason you need to interrupt an analysis, you can do so by pressing F1.

(c) Restarting an interrupted analysis

After the buret refills, the program will prompt you whether you want to run another sample. Continue in the normal manner.

(d) Volumetric delivery of standards

To obtain the maximum precision and accuracy it is critical to deliver the primary standard in a manner which is accurate and reproducible. At sea and in the laboratory we use class A volumetric pipets which have been gravimetrically calibrated. At the 5000 µl range, these glass pipets are accurate to within +4.0 µl (0.08%) with a precision of +2.5 µl (0.05%). This is equal to the precision of the Dosimat within the delivery range of 500-3000 µl. To obtain this precision it is necessary to be very reproducible in the delivery operation. A technique which produces good results is outlined below:

  1. Fill the pipet past the calibration mark and slowly deliver the contents into the container until the meniscus is level with the mark.

  2. Invert the pipet and wipe off any excess solution from the tip and barrel (the level of the solution within the pipet will be below the tip at this point).

  3. Invert again and touch the pipet tip to the sample vessel wall. Release finger pressure and allow pipet to drain. After the pipet has drained keep the tip against the vessel wall for an additional 15-20 seconds.

  4. Remove pipet and rinse any solution that may be on the vessel wall into the sample with DDW.

To maintain accuracy and precision it is necessary to have an absolutely clean pipet where none of the delivered solution clings to the pipet walls. It is also good practice to rinse the pipet with the filling solution prior to filling (unless this is the first use of the pipet).

4.2. Reagent blank determination

The reagent blank is determined by adding the fixing reagents in reverse order and titrating a known volume of primary standard three separate times in the same flask. The volume of the first titration includes the equivalents of primary standard added plus any oxidizing or reducing substances in the reagents (V1); the second titration should theoretically be equal to the equivalents of primary standard only (V2). However, it has been our experience that we rarely observe an obvious difference between V1 and V2 which could be attributed to the reagents and not a variable end point estimation or pipetting. To help quantify this relationship we have added a third aliquot titration (V3), which again should be equivalent to only the primary standard. A statistical comparison between the estimates of |V1-V2| and |V2-V3| is performed to determine whether a significant blank exists.

Fill an iodine flask with distilled water and add in the following order, mixing after each addition, 1 ml of sulfuric acid reagent (10 N), 1 ml alkaline iodide reagent (4 M sodium iodide in 8 N sodium hydroxide), and 1 ml manganese chloride reagent (3 M). [Care must be taken to dispense the reagents below the blank solution surface or in such a manner as to ensure that no precipitate forms in the neck of the flask. If this occurs, a high blank will be encountered if the precipitate subsequently dissolves in the acidified blank solution during the titration step. This is especially prevalent when using the redox electrode which displaces a larger volume of sample into the neck of the flask.] Add 1.0 ml of 0.01 N potassium iodate from a precalibrated glass class A volumetric pipette. Titrate manually, with adequate stirring, to a very pale yellow color, enable computer operation of the titration and run to completion. Record the volume at the equivalence point as V1. Add a second 1.0 ml aliquot of potassium iodate to the flask and titrate as above.

Record as V2 the sum of the volume at the equivalence point and the residual titrant from the first titration (this is equal to the total amount of titrant added minus the volume of titrant at the equivalence point). Add a third 1.0 ml aliquot of potassium iodate, titrate and record volume as V3 in the same manner as V2. Repeat this procedure a minimum of 3 times or until a clear pattern is established.

4.3. Standardization of the thiosulfate titrant

Prepare a blank solution as outlined section 4.1. Add a known volume (at least 5 ml) of potassium iodate primary standard (usually 0.02 N). It is necessary to know the volume added and the normality of the potassium iodate to the highest degree of accuracy that is reasonably possible. Therefore, class A volumetric pipets which have been gravimetrically calibrated should be used and the potassium iodate should be thoroughly dried, weighed in large aliquots relative to the sensitivity of the balance to reduce weighing errors, and the final primary standard volume determined gravimetrically. Titrate as outlined above. Since precision is usually increased with increasing number of titrations, it is our general practice to titrate 3-7 replicates at the beginning and end of each analytical run. In addition to these, we typically run 1-4 standards per day since our samples are analyzed at sea over a 3-4 day period.

4.4. Calculate the concentration of thiosulfate working solution using the following relationship:

VT * NT = VI * NI

where: VT= the blank corrected volume of thiosulfate used to reach the end-point
NT= the normality of thiosulfate solution
VI= the volume of potassium iodate added
NI= the normality of potassium iodate solution
4.5. Sample titration

Remove the ovelying seawater seal and remove the stopper. Add 1 ml of sulfuric acid reagent (10 N), and carefully slip a clean teflon-coated magnetic stir bar into the flask. Place the flask on a magnetic stirrer, insert the clean redox electrode and titrant delivery tube. Titrate as outlined above. Record the presence of bubbles or any unusual aspects of the titration or sample condition in the comment section. Rinse electrode and titrant delivery tube with DDW and repeat above procedure for the next sample.

5. Calculations

Calculate the DO concentration using the following formula:

                   (Vt-Vb) * Nt * E
      µmol O2 l-1 = ________________ - RDO
                       (Vf - Vr) 
where: Vt= volume of titrant (µl)
Vb= volume of blank (µl)
Nt= normality of titrant (µeq µl-1)
E= 0.2500 (µmol O2 µeq-1)
Vf= volume of flask (l)
RDO= dissolved oxygen content of the reagents = (0.804 * 0.14)/Vf
Vr= volume of fixing reagents (l)

This value is derived from Carpenter 1965 (i.e., DO content of fixing reagents when 1 ml of each reagent is used in 140 ml of sample equals 0.018 ml O2 l-1).

6. Precision and Accuracy

The method as outlined above is capable of a precision of 0.1% or less (as defined by the coefficient of variation for triplicate samples). The accuracy of the Winkler titration procedure, when the Carpenter modifications are employed, has been determined to be 0.1% (Carpenter, 1965a,b).

Currently, we are using a high quality commercially prepared potassium iodate certified reference standard to independently assess the accuracy of our potassium iodate primary standard normality. A batch of primary standard is prepared and the theoretical normality normalized to the certified primary standard value. Our primary standard is periodically checked against the certified standard until the batch or lot is exhausted.

7. Equipment

8. Reagents

Manganese chloride reagent (3 M)

Dissolve 600 g of manganese chloride tetrahydrate (MnCl2.4H2O) in 800 ml distilled water and make up to 1 liter in a volumetric flask.

Alkaline iodide reagent

Dissolve 320 g sodium hydroxide (NaOH) in 500 ml distilled water and, separately, dissolve 600 g sodium iodide in 500 ml distilled water. Mix the he two solutions 1:1, by volume.

Sulfuric acid reagent (10 N)

Mix 280 ml of concentrated sulfuric acid into distilled water using a 1 liter volumetric flask

Sodium thiosulfate reagent (0.05 N)

Dissolve 12.41 g of reagent grade sodium thiosulfate (Na2S2O3. 5H2O and make up to one liter with distilled water. Determine exact normality as described under 4.3 above.

Potassium iodate reagent (0.025 N)

Dissolve 0.8918 g of dry (100°C, 2 hours) KIO3 into 800 ml distilled water and bring up to 1 liter in a volumetric flask.

CSK 0.0100 N KIO3 1° standard

Wako Chemicals, Inc., 1600 Bellwood Rd., Richmond, VA 23237.

9. References