W H P Operation s an d M eth od s { Ju ly 1991

Dissolved Oxygen
C. H. Culberson
College of Marine Studies
University of Delaware
Newark, Delaware 19716 U.S.A.

1. Introduction
This chapter describes procedures to be used for the calibration of volumetric glass-
ware, for the calculation of oxygen concentrations, for the preparation of reagents, and for
water sampling in the Winkler method for dissolved oxygen. These procedures are based
on the Chesapeake Bay Institute technique for the Winkler dissolved oxygen method (Car-
penter, 1965). All participants in WOCE are expected to closely follow Carpenter's method
for the measurement of dissolved oxygen in discrete water samples.
Carpenter's (1965) modication of the Winkler technique is reproducible (2 standard
deviations) to 0.010 mL/L (0.4 mol/kg-sw: approximately 0.2% of air saturated O2
concentrations). Achieving this level of precision, however, requires scrupulous attention to
detail.
The elimination of iodine loss by volatilization is the essential element of Carpenter's
technique. This is accomplished by optimizing the concentrations of the pickling reagents
(MnC12 , NaOH/NaI, H2 SO4 ), and by adopting a whole bottle titration which eliminates
iodine loss during sample transfer.
Institutions which titrate aliquots, rather than the entire oxygen sample, must
demonstrate that their technique does not exhibit systematic errors, when compared to
whole bottle titrations. All institutions muse use the pickling reagent concentrations spec-
ied by Carpenter (1965); no exceptions are allowed.
This chapter is not meant to be a cookbook or recipe which lists each step necessary
to determine dissolved oxygen. Rather, its purpose if (1) to specify algorithms for calibration
of volumetric glassware, and (2) to discuss those aspects of Carpenter's method which are
ambiguous or which require modication.

2. Calibration of Volumetric Apparatus

For ease of use aboard ship, Carpenter's method uses volumetric techniques to dis-
pense solutions and to collect samples. To achieve the highest accuracy, all volumetric
glassware must be calibrated. For instance, it is not unusual for the volumes of Class A
analytical pipets to dier by 0.3% from the nominal volume.
2 Dissolved Oxygen { July 1991

Volumetric apparatus used in Carpenter's procedure, include the following:

a. oxygen asks (iodine asks, 125 cm3 nominal capacity)

b. 10 cm3 pipet used to dispense the iodate standard
c. 1 cm3 pipet used to dispense iodate for blank determinations
d. 1 cm3 volumetric ask used to prepare the iodate standard
e. piston buret

With the exception of item c, all volumetric apparatus must be calibrated to contain or
deliver. Since blanks are determined by dierence, the exact calibration of item c is not
essential, as long as the same volume is dispensed each time.
The volumetric apparatus should be calibrated by measuring the weight of pure
water contained or delivered at a reference temperature of 20.0 C. All weights must be
corrected for air buoyancy. The following equations should be used to calculate the true
volume from the measured weight of water in air.

2.1 Volume of Apparatus at Calibration Temperature

The volume of glassware at some calibration temperature, V(t ), is the quotient of
c

the mass of water contained or delivered to the density of pure water,

V(tc ) = Ww
fbuoy = w (1)
where Ww is the weight of water in air, fbuoy is the buoyancy correction, and w is the
density of pure water at the calibration temperature, tc .

2.1.1 Density of pure water

The density of pure water, w , is needed to transform the measured mass of water
into an equivalent volume of water. Over the temperature range, 0 - 40 C, the density of
pure water (units: g
cm,3) is (Bigg, 1967):

 w = 0:999842594 + 6:793952
10,5
t , 9:095290
10,6
t2 +

(2)
1:001685
10,7
t3 , 1:120083
10,9
t4 + 6:536332
10,12
t5
where t the water temperature in  C.
WHP Operations and Methods { July 1991 3

2.1.2 Buoyancy correction

Since the weight of objects in air is less than that in vacuum, a buoyancy correction
must be applied to the weight of pure water contained or delivered by volumetric glassware.
This correction is given by,

fbuoy = 1 ,1 ,air =weights

air=w (3)

where,
air = density of air at the temperature and pressure of the weighing oper-
ation ( 0.0012 g
cm,3 at 20 C),
 w = density of pure water ( 1.0 g
cm,3 for water), and
weights = density of balance calibration weights ( 8.0 g
cm,3 ) for stainless
steel,  8.4 g
cm,3 for brass).
For weighing temperature between 10 and 30 C, use of the approximate buoyancy
factor,
fbuoy = 1:00105 (4)
will result in an error of less than 0.01%.

2.2 Change of Glass Volume with Temperature

Since glass expends upon heating, the volume of glassware increases slightly with
increasing temperature. The volume at temperature t2  C, V(t2 ), is related to the volume
at temperature t1  C, V(t1 ), by the formula,

V(t2 ) = V(t1 )
[1 + v (t2 , t1 )] (5)

where v is the cubical coecient of thermal expansion for the particular glass. For borosil-
icate glass, v is approximately 1.0
10,5 ; for soft glass, v is approximately 2.5
10,5 .
The change in volume of glass apparatus with temperature is quite small. For borosilicate
glass, the volume changes only 0.01% for a 10 C temperature dierence.
Equation (5) is used to reduce calibrated glassware volumes to the standard tem-
perature, 20 C; and to reduce oxygen ask volumes from the standard temperature to the
sampling temperature.
4 Dissolved Oxygen { July 1991

2.2.1 Volume of apparatus at standard temperature

The volume of glassware, calibrated at temperature tc  C, at a reference temperature

of 20 C can be calculated from equation (5). Setting t1 equal to the calibration temperature,
tc  C; and t2 to the reference temperature, 20 C; yields,

V(20) = V(tc )
[1 + v (20 , tc )] (6)

where,

V(20) = volume of oxygen ask at reference temperature, 20 C,

V(tc ) = volume of oxygen ask at calibration temperature, tc  C, from
equation (1).

2.2.2 Volume of oxygen ask at sampling temperature

To calculate the concentration of oxygen in a seawater sample, the volume of seawater
contained in the oxygen ask when the sample was pickled must be known.
Due to the thermal expansion of glass, the volume of the oxygen ask will depend
slightly on the temperature of the seawater used to ll the ask (see equation (5) above),

Vbot = V(20)
[1 + v (tsw , 20)] (7)

where,
Vbot = volume of oxygen ask at temperature tsw  C,
V(20) = volume of oxygen ask at reference temperature, 20 C, from equation (6), and
tsw = seawater temperature at the time of pickling.

2.2.3 Sample calculations of glassware calibration

Tables 1 and 2 give actual calibration data and calculated volumes for an oxygen
ask and a 10 cm3 pipet. The density of pure water was calculated from equation (2), the
buoyancy factor from equation (3), the volume at the calibration temperature from equation
(1), and the volume at 20 C from equation (6).
WHP Operations and Methods { July 1991 5

Table 1
Sample calibration of an oxygen ask using WOCE equations. Calibration temperature,
21.7 C; buoyancy factor calculated using density of stainless steel weights, and the cubical
coecient of thermal expansion for borosilicate glass.

Weights in air Density Buoyancy Volume at

Empty Full Water water factor 21.7 C 20.0 C
84.51 230.58 146.07 0.997841 1.001054 146.540 146.538

Table 2
Sample calibration of a 10 mL pipet using WOCE equations. Calibration temperature,
24.0 C; buoyancy factor calculated using density of stainless steel weights, and the cubical
coecient of thermal expansion for borosilicate glass.

Weights in air Density Buoyancy Volume at

Empty Full Water water factor 24.0 C 20.0 C
17.2015 27.1273 9.9258 0.997299 1.001055 9.9632 9.9628
6 Dissolved Oxygen { July 1991

3. Calculation of Dissolved Oxygen Concentration

3.1 Titration of an Aliquot
Equations for calculating the concentration of dissolved oxygen from the results
of aliquot titrations of acidied oxygen samples are more complex than those for whole
bottle titrations (equation (8) below); the exact form of the equations will depend on the
titration and standardization procedures used by each institution. Consequently, no general
algorithm can be given for calculating the oxygen concentration from the results of aliquot
titrations. Institutions performing aliquot titrations must submit the equations that they
use to calculate dissolved oxygen concentration to the WOCE Hydrographic Programme
Oce for verication and approval.

3.2 Whole Bottle Titrations

The concentration of dissolved oxygen, determined from whole bottle titrations,
should be calculated form equation (8). Equation (8) diers from Carpenter's (1965) equa-
tion for whole bottle titrations in the correction for oxygen added with the reagents. In
equation (8), the correction for oxygen added with the reagents is a function of the ask
volume; whereas, in Carpenter's equation, a constant correction, 0.018 mL/L, is used.. The
use of a constant value for the concentration of oxygen added with the reagents is only
correct if all asks have identical volumes. The value of DOreg in (8) is a measured value
(Murray et al, 1968), rather than the estimated value used by Carpenter (1965).

,
(Vx Vblk;dw )
VIO3
NIO3
5598 , 1000
DO
,
O = (Vstd Vblk;dw ) reg
2
(Vbot , Vreg ) (8)

Terms in equation (8) have the following meaning. All volumes are expressed in
cubic centimeters.
WHP Operations and Methods { July 1991 7

Vx = thiosulfate titer of sample (cm3 )

Vblk dw = thiosulfate titer of pure water blank (cm3 )
;

Vstd = thiosulfate titer of standard (cm3 )

Vbot = volume of sample bottle (cm3 ) at the temper-
ature of sampling
Vreg = volume (2 cm3 ) of sample displaced by
reagents
VIO3 = volume of iodate standard (cm3 ) at tempera-
ture of standardization
NIO3 = normality of iodate standard (=6
molarity) at
temperature of standardization
DOreg = absolute amount of oxygen added with
reagents, 0.0017 mL (Murray et al, 1968)
O2 = oxygen concentration in sample (mL/L)

3.3 Conversion of Volumetric to Weight Concentrations

Oxygen concentrations must be reported in units of micromol per kilogram of sea-
water (mol/kg-sw). Oxygen concentrations in mL/L calculated from equation (8) can be
converted to units of mol/kg-sw by the following equation,

O2 (mol=kg , sw) = 44:660
O2 (mL=L)=sw (9)

where sw is the density of seawater at the temperature at which the oxygen sample was
pickled. If the seawater temperature at the time of pickling is not known, the density of
seawater in equation (9) should be calculated at the potential temperature of the water
sample. The value 44.660 equals (1000/molar volume of oxygen gas at STP) (Weiss, 1981).

3.3.1 Density of seawater

The density of seawater, sw , is needed to convert volumetric oxygen concentrations
(mL/L) to mass concentrations (mol/kg-sw). Between 0 and 40 C, and 0.5 to 43 salinity,
the density of seawater at 1 atmosphere (units: g
cm,3 ) is Millero and Poisson, 1981),

 = +A
S +B
S +C
S
sw w
1:5 2
(10)
8 Dissolved Oxygen { July 1991

where S is the salinity, t is the temperature, in  C, and

A = 8:24493
10, , 4:0899
10,
t + 7:6438
10,
t ,

4 6 8 2

8:2467
10,
t + 5:3875
10,
t

10 3 12 4
(11)

B = ,5:72466
10, + 1:0227
10,
t , 1:6546
10,
t

6 7 9 2
(12)

C = 4:8314
10, 7
(13)

3.3.2 Sample calculation of oxygen concentration

Table 3 provides actual titration data and calculated oxygen calculations for 2 whole
bottle titrations.

Table 3

Sample calculations of dissolved oxygen concentration.

Oxygen ask volume at 20 C: 146.538 mL
Parameters in equation (8): Vblk = -0.0004
VIO3 = 9.9784
NIO3 = 0.01014
Vstd = 0.7180
Vreg = 2
DOreg = 0.0017

Temperature Salinity Oxygen ask volume at Dissolved Oxygen

of sampling sampling temperature mL/L mol/kg-sw
10.78 C 35.398 146.524 mL 3.3738 147.06
2.19 C 34.915 146.512 mL 6.1839 269.64
WHP Operations and Methods { July 1991 9

3.4 Eect of Temperature on Mass of Thiosulfate and Iodate Dispensed

The thermal expansion of aqueous solutions is much greater than that of glass. Con-
sequently, the mass of solution dispensed by pipets and burets depends on the temperature
of the solution. In gure 1, the ratio of the mass of pure water dispensed at temperature
t  C to the mass dispensed at 20 C is plotted versus the solution temperature. The val-
ues plotted in the gure are corrected for the thermal expansions of both water and glass.
However, almost all of the variation is due to the thermal expansion of water.
The gure shows that, at 20 C, the mass of solution dispensed changes about 0.1%

for a 5 change in solution temperature. In equation (8), it is the masses of thiosulfate and
of the iodate standard which will be aected by thermal expansion of the solution.
The thiosulfate volumes (Vx , Vstd ) occur as a ratio in equation (8). Consequently,
mass changes due to thermal expansion of the thiosulfate titrant will cancel if temperatures
during standardization are close to those during sample analysis.
The eects of thermal expansion are more serious for the iodate standard. Room
temperature should be measured during each thiosulfate standardization so that the mass
of the iodate standard dispensed can be calculated. Assuming a reference temperature of
20  C, the volume of the iodate pipet at the standardization temperature, tstd  C;, is (see
equation (5)),

VIO3 = V(tstd = V(20)
[1 + v (tstd , 20)]: (14)

The iodate normality (= 6
molarity) at temperature, tstd  C, can be calculated as,

NIO3 (tstd = NIO3 (20)
[w (tstd )=w (20)] (15)
where,
NIO3 (20) = normality of iodate standard (= 6
molarity) at 20 C,
 (t )
w std = density of pure water at standardization temperature, and
 (20)
w = density of pure water at reference temperature, 20 C.
Substituting the pure water density for the density of dilute KIO3 in equation (15)
will not cause signicant error.
10 Dissolved Oxygen { July 1991

Figure 1: Eect of temperature on mass of water dispensed due to thermal expansion of

water and glass.
WHP Operations and Methods { July 1991 11

4. Preparation of Reagents
Reagents used in Carpenter's procedure include the following:

a. MnC12
4H2 O 600 g/liter, 3 molar

b. NaOH 320 g/liter, 8 molar
NaI 600 g/liter, 4 molar
c. H2 SO4 280 mL concentrated H2 SO4 /liter, 5 molar
d. Starch 1% solution
e. Na2 S2 O3
5H2 O 35 g/liter, 0.14 molar
f. KIO3 0.3567 g/liter, 0.001667 molar
With the possible exceptions of the starch, thiosulfate and iodate solutions, all in-
stitutions involved in WOCE oxygen measurements must use the reagent concentrations
specied by Carpenter (1965). Reagent solutions should be prepared from reagent grade
(or better) chemicals.

4.1 Manganous Chloride

The substitution of manganous sulfate for manganous chloride is allowed because
the solubility of oxygen in 3 molar MnSO4 does not appear to have been measured (Murray
et al, 1968). Consequently, the correction for oxygen added with the reagents, DOreg , in
equation (8) is uncertain when manganous sulfate is used in place of MnC12 .

4.2 Sodium Thiosulfate

Note that the chemical formula for sodium thiosulfate is misprinted in the table
labeled Reagent Solutions on page 141 of Carpenter's (1965) paper. The correct formula is
Na2 S2 O3
5H2 O.

4.3 Potassium Iodate Standard

The measured weight of KIO3 used to prepare the iodate standard must be corrected
for air buoyancy. The buoyancy factor for KIO3 calculated from equation (3), assuming the
density of KIO3 equals 3.89 g
cm,3 (Hodgman, 1958), is 1.000159. The molecular weight
of KIO3 equals 214.001 (Commission on atomic weights and isotopic abundances, 1988).
Commercially prepared KIO3 solutions are not allowed as standards. Experience has
shown that the concentration of commercial standards may be in error by 0.5%.
12 Dissolved Oxygen { July 1991

The normality (= 6
molarity) of the KIO3 standard at the temperature of preparation

should be calculated from the following equations,
NIO3 (tp ) = WKIO3
fbuoy KIO3
6
;
(16)
V(tp )
214:001

V(tp ) = V(20)
[1 + v (tp , 20)] (17)

where,

NIO3 (tp ) = normality of iodate standard (= 6
molarity) at tp  C,

WKIO3 = weight of KIO3 in air,
fbuoy KIO3
; = buoyancy correction for solid KIO3 (= 1.000159),
214.001 = 1987 molecular weight KIO3 ,
tp = preparation temperature of KIO3 solution,
V(tp ) = volume of volumetric ask at temperature, tp  C, and
V(20) = volume of volumetric ask at 20 C reference temperature.
The normality (= 6
molarity) of the KIO3 standard at the reference temperature
20 C, is calculated from the following equation,
NIO3 (20) = NIO3 (tp )
[w (20)=w (tp )]: (18)

Substituting the pure water density for the density of dilute KIO3 in equation (18)
will not cause signicant error.

5. Water Sampling and Oxygen Analysis

Except for the changes noted below, all participants in WOCE are expected to adhere
to the procedures given by Carpenter (1965) for the measurement of dissolved oxygen in
discrete water samples.

5.1 Procedure for Samples

A six-inch piece of tygon tubing slipped over the outlet valve of the water sampler
can be used as the drawing tube. Before lling each oxygen ask, run a small volume of
WHP Operations and Methods { July 1991 13

sample water through the drawing tube to remove air. Air bubbles frequently adhere to the
sides of the drawing tube. If this occurs, tap the drawing tube to remove them before lling
the oxygen ask. Bubbles are very likely to form on the sides of new tygon tubing. This
problem can be avoided by soaking the tubing in seawater for several days prior to use.
Rinse the oxygen ask several times with sample water ( 30 cm3 ) prior to lling;
then ll the ask; and, nally, over ow it with at least 3 bottle volumes of sample water
to minimize contamination with atmospheric oxygen. Over owing the ask with copious
amounts of sample water is the single most important factor in producing high-quality
oxygen results.
Prior to sampling, the MnC12 and NaOH/NaI reagents should be placed near where
the oxygen samples will be drawn. Add these two reagents immediately after each oxygen
sample is drawn. Do not stopper the oxygen asks before adding the MnC12 and NaOH/NaI
reagents. The tip of the reagent dispensers should extend below the neck of the oxygen
asks, so that precipitate does not form in the excess seawater above the neck of the asks.
The calculation of dissolved oxygen, equation (8), involves corrections for the vol-
umes of MnC12 and NaOH/NaI added, and for the oxygen that they contain. The reagent
dispensers must dispense exactly 1.00 cm3 . Their calibration should be checked, and ad-
justed if necessary, by dispensing ten 1 mL volumes into a 10 mL graduated cylinder. The
solubility of oxygen in the reagents is temperature dependent. Since the value used for
the concentration of oxygen in the reagents, DOreg , is the air solubility at 25.5 C, it is
important that the reagents be stored at a temperature as close to 25.5 C as possible. To
maintain their temperature as constant as possible, the MnC12 and NaOH/NaI reagents
should be stored in the laboratory, and only brought on deck when needed. These reagents
should never be stored in the cold, since this will signicantly increase their dissolved oxygen
content.

5.2 Standardization
Carpenter (1965) does not add 1 mL of MnC12 to standards, even though it is
added to blanks. Carpenter's standardization procedure should be modied to include the
addition of 1 mL of MnC12 after the addition of the NaOH/NaI reagent. Standards should
be determined in pure water, not in seawater.

5.3 Blank Determination

Blanks should be determined in pure water, not in seawater. The little information
available on the magnitude of seawater blanks suggests that they are not constant but vary
with depth and position. Rigorously then, seawater blanks should be determined at each
depth that oxygen samples are taken. This does not appear feasible. Ignoring the seawater
blank will introduce small, systematic errors into the oxygen calculations, but these errors
will be the same for all institutions and will not aect the comparison of results from
dierent institutions.
14 Dissolved Oxygen { July 1991

5.4 Replication and Precision

To provide a reliable estimate of analytical precision, duplicate oxygen samples
should be titrated for at least 10% of the total number of samples taken. Duplicates should
cover the entire range of observed oxygen concentrations, and should not be concentrated
at any one concentration. The reproducibility of oxygen measurements (2), calculated
from the above duplicates, must be 0.5% of the highest oxygen concentration encoun-
tered. Each set of oxygen data submitted to WOCE must state the standard deviation of
the data, and the number of replicates used to calculate the standard deviation.

5.5 Endpoint Detection

Photometric or amperometric endpoint detection may be substituted for the visual
starch endpoint specied by Carpenter (1965).

6. Acknowledgements
Must of the discussion in section 2, Calibration of Volumetric Apparatus, is taken
from the unpublished notes of Andrew Dickson (1990). Some of the comments in section 5,
Water Sampling and Oxygen Analysis, are based on the Marine Technicians Handbook from
the Scripps Institution of Oceanography (Anonymous, 1971).

7. References
Anonymous, 1971. Marine Technicians Handbook: Oxygen Analysis. Scripps Institution
of Oceanography Reference No. 71-8.
Bigg, P. H., 1967. Density of water in S. I. units over the range 0-40 C. British Journal
of Applied Physics,18, 521-537.
Carpenter, J. H., 1965. The Chesapeake Bay Institute technique for the Winkler dissolved
oxygen method. Limnol. Oceanogr. 10, 141-143.
Commission on Atomic Weights and Isotopic Abundances, 1988. Atomic weights of the
elements 1987. Pure & Applied Chemistry 60(6):841-854.
Dickson, A., 1990. Calibration of oxygen bottles from the weight of the contained water
when weighed in air. Unpublished notes. 4 pages.
Hodgman, C. D., editor, 1958. Handbook of Chemistry and Physics, 39th edition. Chemical
Rubber Publishing Company, Cleveland, Ohio.
Millero F. J. and A. Poisson, 1981. International one-atmosphere equation of state of
seawater. Deep Sea Res., 28, 625-629.
WHP Operations and Methods { July 1991 15

Murray, C. N., J. P. Riley, and T. R. S. Wilson, 1968. The solubility of oxygen in Winkler
reagents used for the determination of dissolved oxygen. Deep Sea Res., 15, 237-238.
Weiss, R. F., 1981. Oxygen solubility in seawater. Unesco Technical Papers in Marine
Science. No. 36:22