d.o. in hyper saline water
TRANSCRIPT
-
7/24/2019 D.O. in Hyper Saline Water
1/16
Limnoi. Oceanogr., 36(2),
1991,235-250
0 1991, by the American
Society of Limnology and Oceanography,
Inc.
Dissolved oxygen concentrations in hypersaline waters
J. E. Sherwood,F. Stagnitti, and M. J. Kokkinn
Faculty of Applied Science and Technology, Warrnambool Institute of Advanced Education,
Warrnambool, Victoria 3280, Australia
W. D. Williams
Department of Zoology, University of Adelaide, G.P.O. Box 498, Adelaide, South Australia 5001
Abstract
Henrys law constants (kO) and equilibrium concentrations (CO*) of dissolved oxygen (DO) at 1
atm were measured in NaCl solutions of concentration (S) up to -260%~ and at temperatures (7)
between 273 and 308K. An equation of the form
In Co* = a, +
$ + a,ln T + a,T + a,T2 + S(a, + a,T + a7P) + asS2
was found to predict DO values to within the experimental uncertainty. An equation of the same
form satisfactorily described the variation of In kO over the same temperature and concentration
ranges. In order to develop these equations it was also necessary to develop ones to describe the
variation of density and vapor pressure of NaCl solutions with T and S. The equations can be
used to generate ables of oxygen solubility values that can be used for hypersaline waters dominated
by NaCl. Theoretically, DO values based on NaCl can be corrected for the presence of other ionic
salts in natural waters. At present this correction is limited by the availability of DO data for these
subdominant electrolytes.
It is a century since L. W. Winkler pub-
lished his pioneering studies on the solu-
bility of oxygen in water. Since then, equi-
librium concentrations of dissolved oxygen
(DO) in marine and estuarine waters have
been measured over a range of salinity and
temperature with his technique or minor
modifications of it. In these saline waters,
as well as in freshwaters, his basic technique
has become the standard to determine the
concentration of DO, and tables are avail-
able that document the relationship be-
tween salinity, temperature, and DO at
equilibrium in such waters (Weiss 1970;
Benson and Krause 1980; Mortimer 198 1).
It is only relatively recently that other meth-
ods of measuring DO have been developed;
they involve diffusion to a reducing elec-
trode. They have not supplanted Winklers
method where simplicity and accuracy are
important.
Acknowledgments
R. Evans prepared the figures. Fred Post provided
literature otherwise inaccessible to us. D. J. J. Kinsman
donated a copy of a paper we found difficult to obtain.
W.D.W. thanks the Australian Water Research Ad-
visory Council for support (Scientific Merit Grant).
Remarks by anonymous referees are also ackuowl-
edged.
Concentrations of DO in hypersaline wa-
ters, that is waters with salt concentrations
>40%, have been much less studied, al-
though it has long been known that DO con-
centrations decline markedly with increas-
ing salinity. Perhaps, at least in part, the
difficulties of measuring DO by the Winkler
method in hypersaline waters, alluded to by
Walker et al. (1970) and Hammer (1986),
explain why so few studies have taken place.
In any event, for limnologists studying sa-
line lakes there are no tab les available of the
sort that freshwater limnologists or marine
and estuarine scientists have. The few stud-
ies that have been made of DO in hyper-
saline waters genera lly involve experimen-
tal techniques that can be criticized on
methodological grounds, measurements at
only one or two temperatures, and (or)
par-
ticular
bodies of salt water (Laguna Ta-
maulipas, marine brines, Great Salt Lake)
(e.g. Macarthur 1916; Copeland 1967; Loe-
flich 1972; Kinsman et al. 1974; Jones et
al. 1976). For the most part, studies of DO
in hypersaline waters have been published
in journals lacking wide circulation among
limnologists.
There is a need to remedy this situation
given the wide distribution and abundance
235
-
7/24/2019 D.O. in Hyper Saline Water
2/16
236
Sherwood et al.
of inland saline waters (Hammer 1986; Wil-
liams 1986a) and the biological significance
Notation
__-
of DO to thkm. This papei attempts to do
so. We report a systematic-study of DO con-
centrations in hypersaline waters between
temperatures ,3f 0 and 35C. NaCl solutions
ranging up to near saturation (- 26OY&$ ere
selected as model solutions because most
natural saline waters are dominated by this
salt. The effect of other salts on DO levels
is also examined.
Theory
Gas solubi/ity and Henrys Iaw-Appli-
cation of Hellrys law to the solubility of
atmospheric gases n natural waters has been
thoroughly outlined in papers by Benson
and Krause (1980), Weiss (1974), and Weiss
and Price (19 80). According to Henrys law
the solubility of a gas, , in a solvent o f fixed
composition .atconstant temperature (T), is
given by
A= kix,
(1)
whereA is the fugacity of the gas n the vapor
phase, Xi is its mole fraction in the liquid
phase, and k, is the Henrys law coefficient
(units given in list of notation).
Following the treatment of Benson and
Krause for o:uygen,
where Znj is lbe sum of the moles of oxygen
(n,J, water, and all other dissolved sub-
stances in some volume, V,, of solution at
the equilibrillm temperature. But
i.e. the solution mass (m3 divided by its
density at the equilibrium temperature (p,).
The concentration of dissolved oxygen
(C,) is commonly measured in molar units,
i.e.
c-&y.
s
Thus from Eq. 2 and 4
Co( mol liter- ) =
fOG%)
k, v, ,
(5)
and by Eq. 3
(6)
B
B
co
CO
G
c
MX
Bunsen coefficient for a dissolved gas
Second virial coefficient for oxygen gas,
atm-
Dissolved oxygen concentration in equilibri-
um with an oxygen-containing atmosphere
of total pressure, Pa,, mg liter-l
Dissolved oxygen concentration in equilibri-
um with an atmosphere of standard com-
position and saturated with water vapor at
a total pressure of 1 atm, mg liter-l
Co* in distilled water, mg liter-
CO* in a solution of the electrolyte MX, mg
liter-l
Co* in a salt mixture, mg liter
Fugacity of the real gas i, atm
Setschenow constant for an ion of fype i, li-
ters mol-*
Henrys law coefficient for disgolved gas i,
atm
Ion-specific Setschenow constant, kg mol-
Temperature-dependent Setschenow constant
for an electrolyte, liters mol-
A modified Henrys law constant defined by
the equation: CO = K&, mol liter- aim-
Number of moles of component i, mol
Gas pressure, atm
Equilibrium vapor pressure of water, atm
Solution density at a known temperature, g
liter-l
Coefficient of determination adjusted for de
it is an approximate unbiased estimate of
the population rz
rms error of a regression analysis
Salinity, measured as grams of salt per kilo-
gram of solution, %
Celsius temperature, C
Absolute or Kelvin temperature, K
Solution volume, liters
Mole fraction of component i in a solution
Valency of an ion of type i
Application to NaCl solutioris-In the salt
solutions used in our experiments the fol-
lowing approximations can be made:
Znj N n, + nsalt
(7)
ms = mw + %an
(8)
where the subscripts w and salt refer to wa-
ter and NaCl.
Neglecting dissolved gases (chiefly oxy-
gen and nitrogen) introduces an error of
~0.01% in Eq. 7 and 8. Thus,
znj_ (n, + n,.d
-
7/24/2019 D.O. in Hyper Saline Water
3/16
Oxygen in hypersaline waters
For NaCl solutions whose concentrations
are measured as salinity (S in ?&)), he fol-
lowing substitutions are possible for Eq. 9:
m, = l,OOOg
m,
= (1,000 - 5) g
m
salt
=sg
M, = 18.015 g mol-l for Hz0
M
N&l
= 58.443 g mol-l for NaCl,
i.e.
$ = (5.5509 x 1O-2 - 3.8399
s
x 10-5s).
(10)
Substituting Eq. 10 into 6 gives
Co(mo1 liter-)
=fp (5.5509 x 1O-2 - 3.8399
0
x 10-5s).
(11)
Fugacity of oxygen-similar expressions
for the fugacity of a real gas have been de-
rived by both Benson and Krause (1980)
and Weiss (1974):
fo = Poexp(PB) (12)
where PO is the partial pressure of oxygen
in the gas mixture (atm), P is the total pres-
sure of the gas mixture (atm), and B is the
second virial coefficient for oxygen at tem-
perature T(atm-I).
For oxygen, Benson and Krause derived
the following expression for B (although they
represented the virial coefficient by the sym-
bol -0) for 0 3, the
SE is given.
Results (Table 1) indicate that iodine loss
was not significant during the pipetting step.
An uncertainty of f 0. 1C in water bath
temperature introduces a maximal uncer-
tainty of 0.04 mg liter- in Co at 0C. This
uncertainty decreases as temperatures and
salinity increase.
With the exception of the value for YC,
all measurements of Co* for distilled water
lie within 0.04 mg liter- of accepted values
(Table 2). The reason for the discrepancy at
5C is not known. Values for other temper-
atures indicate that systematic errors are not
significant when compared to the random
errors of the experiment.
DO determinations reported here are thus
considered reliable to at least +0.05 mg li-
ter-. The experimental technique does ap-
pear to lead to a true solubility equilibrium,
despite reports of a tendency for this meth-
od of aeration to lead to supersaturation
(Carpenter 1966). A similar conclusion was
reached by Khomutov and Konnik (1974).
An equation of state or k0 and Co* values
in NaCl solutions-The data set for Henrys
law constants (ko) and DO concentrations
(Co*), consisting of 102 paired values (Ta-
ble 3), was the basis of our mathematical
models. Multiple regression was used to de-
velop a model of best fit for predicting In
k0 and In Co* given concentration S(?&)and
temperature T(K). Initial attempts to model
our data set were to fit it to an equation
derived for dissolved gases in seawater
(Weiss 1970). This equation has the form:
lnx=aO+a,l + aJn T + a,T
+ a,F T S(a, + a,T + a,F) (24)
where x = k0 or Co*.
These equations proved to be statistically
ineffective and of insufficient accuracy. Their
major problems were an overall lack of fit,
lack of statistical significance of the partial
regression coefficients ai, and a strong cur-
vilinear response in salinity terms. After
considerable analysis and testing of regres-
sion equations consistent with the Weiss
formulation, it became apparent that the fit
could not be improved unless higher order
salinity terms were included. Consistently
better results were obtained by introducing
an P term. In this paper, we shall refer to
equations with an additional s2 term as the
modified Weiss equations. They are of the
form
In x = a, + a, f + aJn T + a,T
+ a,T2 % S(a, + a,T + a7F) +
asp
(25)
where x = k, or Co*.
Both the Weiss and modified Weiss equa-
tions have a considerable degree of multi-
collinearity (Table 4), as is to be expected
with fitting polynomials of this nature.
Therefore, the values of the partial regres-
sion coefficients are highly sensitive to the
inherent accuracy of the experimental data.
In other words, changing the data set may
produce a different set of values. The sta-
bility of the coefficients was checked by
splitting the data set into two (by random
draw) and then running the regressions
through each of the subsets. The changes n
the coefficients were generally ~2% and
comparable accuracy to the original data set
was maintained. These results indicated that
in spite of the multicollinearity, the coeffi-
cients were reasonably stable and reliable.
Another consequenceof multicollinearity
is that extrapolation outside the range of
values for S and T presented here is unre-
liable and should be avoided. Our equations
cannot be expected to perform satisfactorily
for values oft outside the range O-35C and
S > 2607~. Bearing in mind the conditions
stated above, we find the modified Weiss
equations will produce accurate predictions
of our measured dissolved oxygen concen-
trations and are significantly better than the
Weiss equation. Calculation of the absolute
-
7/24/2019 D.O. in Hyper Saline Water
8/16
242
Sherwood et al.
Table 3. Expknental determinations of k. and
Co*. These data kere used in deriving the regression
equations.
Temp.
(C)
COltCll 10 k,
co*
w (atm)
(mg liter-)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
5
5
5
5
5
5.4
5.4
5.4
5.4
5.4
5.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
19.4
19.4
19.4
19.4
19.4
19.4
19.4
19.4
19.4
0
0
0
0
0
0
0
50.37
50.37
57.4
.57.4
f00.25
100.25
141.9
141.9
143.52
$43.52
152.6
152.6
200.43
1 00.43
253.15
:;53.15
154.63
X54.63
0
1Oi.61
102.61
i158.58
2158.58
55.55
55.55
160.92
160.92
: 03.28
: 03.28
0
0
63.05
63.05
106.46
: 06.46
:i47.9
il47.9
:195.19
:195.19
250.79
250.79
0
0
0
0
54.13
54.13
98.44
98.44
103.13
2.508
14.73
2.537
14.56
2.500
14.77
2.516
14.68
2.509 14.72
2.525
14.63
2.545
14.51
3.568 10.38
3.569
10.38
3.78
9.80
3.77
9.83
5.09
7.28
5.09
7.28
7.07 5.23
7.11
5.20
7.12
5.20
7.08 5.23
7.54
4.91
7.53
4.9 1
11.1
3.34
10.9
3.39
17.2
2.13
17.0
2.16
17.5
2.10
17.2
2.13
2.913
12.65
2.920 12.62
5.82 6.35
5.80 6.37
19.1
1.91
18.5 1.97
4.17
8.84
4.20
8.79
8.80
4.18
8.83 4.17
12.1
3.03
12.2
3.01
3.30 11.13
3.283 11.18
4.84
7.60
4.80
7.66
6.51 5.64
6.49
5.66
8.56
4.28
8.52 4.30
11.9 3.06
12.2
3.01
18.7
1.95
18.7 1.94
3.95 9.19
3.92 9.26
3.95 9.18
3.93 9.25
5.38 6.75
5.36
6.78
7.16
5.07
7.06 5.14
7.28
4.98
Table 3. Continued.
Temp.
(0
19.4
103.13
7.24 5.01
19.4 147.96 9.75 3.72
19.4 147.96
9.78 3.70
19.4 195.69
13.5
2.67
19.4
195.69 13.4 2.70
19.4 247.43
19.5
1.84
19.4 247.43
19.3
1.87
19.4
250.59 20.2
1.78
19.4 250.59
19.6
1.83
27.25
0 4.49 7.96
27.25
0 4.51
7.93
27.25
53.76 6.12
5.84
27.25
53.76
6.12 5.84
27.25
104.31 8.10
4.41
27.25
104.31 8.09
4.42
27.25 160.39 11.6 3.08
27.25
160.39 11.6 3.08
27.25
215.14 16.6
2.14
27.25 215.14
16.6
2.14
27.25
258.49 22.2 1.59
27.25
258.49 22.2
1.59
34.1
0 4.98
7.03
34.1
0 4.96 7.06
34.1
158.39 11.8
2.96
34.1 158.39
11.8
2.97
35.2 0 5.06
6.90
35.4 0 5.09 6.85
35.4 0 5.00
6.98
35.4
0
5.07
6.88
35.4 0 5.05
6.9L
35.4 52.78 6.64
5.26
35.4
52.78 6.59 5.29
35.4 102.6
8.74 3.99
35.4 102.6 8.80
3.96
35.4 114.89 9.31
3.75
35.4
114.89
9.31 :
3.75
35.4
159.56 12.2 2.86
35.4 159.56 12.2
2.85
35.4
199.48 15.6
2.23
35.4
199.48 15.3
2.28
35.4
251.73 21.2
1.64
35.4 251.73 21.5
1.62
35.4 256.54
22.4 1.55
35.4
256.54 22.4
1.55
differences between experimental and pre-
dicted values of C,* showed ,that for the
Weiss equation 22 of the 102: data points
differed by ~0.05 mg liter-, 48 by ~0.10
mg liter-, and 87 by 10.20 mg lil.er-l. By
contrast, for the modified Weiss equations,
85 differences were 10.05 mg liter-, 96
were 10.10 mg liter-, and all were within
0.20 mg liter-l.
Modeling errors for the modified Weiss
equations thus compare very favorably to
-
7/24/2019 D.O. in Hyper Saline Water
9/16
Oxygen in hypersaline waters
243
Salinity (g kg-)
Fig. 1. Variation of Co* with salinity at, in de-
scending order, O, 5, lo, 20, 30, and 40C. Curves
are drawn using Eq. 25; 95% of the experimental data
points lie within 0.10 mg liter-l of the predicted curves.
the experimental precision of the data.
Equation 25 was thus used to construct the
curves of Fig. 1. Note that the relationship
between DO and salinity approaches lin-
earity at higher temperatures. In all cases
DO concentration decreasesas salinity in-
creases, although the effect is reduced at
higher temperature.
Violation of the Setschenow relation-
ship-Equation 25 developed as the best-fit
model for our data violates the Setschenow
equation. This empirical relationship is that
at constant temperature
= K[MX]
CW
where C, = Co*
in distilled water, C,, =
Co* in an electrolyte solution of concentra-
tion [MX], and K = temperature-dependent
Setschenow constant (varies with the nature
of the electrolyte, MX).
Equation 26 indicates that the logarithm
of DO should be a
linear
function of salin-
ity, not a
quadratic
as we have found. The
validity of Eq. 26 has been demonstrated
for a wide range of solutes in various elec-
trolyte solutions (Long and McDevit 1952;
Battino and Clever 1966; Weiss 1970).
We examined the effect of the Sz carefully
using data at each of our constant temper-
atures (see Table 3). For Co* at lower
temperatures (OV, 5C) the rms error of re-
gression (.s) or a quadratic in NaCl concen-
tration was two-thirds that for a linear re-
gression in concentration. At higher
temperatures (10.4-35.2C) s for the qua-
dratic regression was similar in magnitude
-mo
OclO
00 mb-m-
m OOCIN~
m ~000
6 id66
* 4 + I(
* * * *
***xc*+*+
*wP-CINbl-w
000000000
PWb-C-lhlbr-
00000000
-
7/24/2019 D.O. in Hyper Saline Water
10/16
244
Sherwood et al.
Table 5.
Setschenowonstants
K) calculated from published data on the solubility of oxygen in NaCl solutions
and from this inve$tigation.
Temp.
(c)
0
10
25
40
MA paa]
(m(bl liter)
5.37
5.11
2.49
5.35
5.10
5.31
2.0
5.06
1.2
4.0
1.92
,1.53
.5.28
5.03
(W2LP)
0.1600f0.0008
0.174~0.011
0.1699t0.0007
0.1486+0.0006
0.154f0.010
0.1336a0.0003
0.125+0.010
0.136+0.007
0.136
0.1374+-0.0004
0.1416+0.0008
0.145LO.002
0.1210~0.001
0.125+0.004
n
11
3
4
11
4
11
4
4
8
7
3
11
4
--
-
Reference ;
-
This study
Cramer 1980
Eucken and Hertzber@,1950
This study
Cramer 1980
This study
Finn 1967
Cramer 1980
Khomutov and Konnik 1974t
Macarthur 1916 :
Geffcken 1904
Eucken and Hertzberg 1950
This study
Cramer 1980
to, but general\y smaller than, s for the lin-
ear regression*.:This s to be expected since
our analysis qldicates that the Sz term is
always statisti:cally significant. What our
analysis also sljows, however, is that at each
temperature our data fit a linear regression
very well (P = 1 OO,s values 0.003-0.006,
12 I n 5 25).,
The questionsmay be asked whether the
S term resu1t.s rom curvature in the ex-
perimental dai.a set that is caused by some
nonideal behayrior at very high salinities. To
check we performed regressions or both the
Weiss and modified Weiss equations to a
number of da;ta subsets containing values
for salinities i:n the ranges O-100$& inclu-
sive, O-l 507~ nclusive, and 0-2007~ inclu-
sive. In all capes, he term s2 was statisti-
cally significant and the modified Weiss
equations pedormed better than the Weiss
model, althou:+ its influence on the regres-
sion diminishlzd in the data sets containing
lower salinity values, e.g. 0-100~~. One
would expect that the S term would have
little influence on the outcome of the re-
gression in the salinity ranges (< 4Oo/oo)on-
sidered by Weiss (1970).
It is possiblie that the significance of the
P term has been generally overlooked by
previous workers because inear regression
fitted their data sets to within experimental
uncertainty. Also, at lower salinities and
higher tempeTatures the influence of the s2
term on the r,egressiondiminishes.
In order to facilitate comparison of our
data with the available literature we have
calculated Setschenow constants from it.
They are consistent with findings by other
workers (Table 5) and have standard devi-
ations of 0.5% or less.
Comparison with other stud&s-The ef-
fect of electrolyte concentrations on the sol-
ubility of gases n aqueous solutions is an
area of cons iderable theoretical and app lied
interest. Two reviews have summarized the
literature up to 1966 (Long and McDevit
1952; Battino and Clever 1966). This lit-
erature includes several studies on the sol-
ubility of oxygen in solutions of NaCl and
other salts of limnological significance. A
number of single-temperature studies (25C)
have been published (Macarthur 1916; Finn
1967; Kinsman et al. 1974; Khomutov and
Konnik 1974). Studies of NaCl solutions at
several temperatures include those of
Geffcken (1904: 15C, 25(Z), Eucken and
Hertzberg (1950: OC, 15C, 20C 25Q
and Cramer (1980: 0-300C). Only Cra-
mers study, involving distilled water and
three NaCl concentrations (-50, 150, and
2507) covers the entire concentration and
temperature ranges of our work. IJnfortu-
nately, the solubility data of ihese earlier
studies are presented with a number of dif-
ferent concentration units for both DO and
electrolyte.
Comparison of our results with olthersne-
cessitated conversion of all measurements
-
7/24/2019 D.O. in Hyper Saline Water
11/16
Oxygen in hypersaline waters
245
to a common set of units. The empirical
Setschenow Eq. 26 was selected for this
comparison, Molarity was used rather than
molality because t typically gives linear Set-
schenow plots to higher concentrations
(Long and McDevit 1952). Several inves-
tigators have reported Setschenow con-
stants in these units.
Equation 25 was used to generate C,*
values at 10 salt concentrations (O-250?&)
and four temperatures. Linear regression of
h3ul co*
against NaCl molarity was then
used to determine the Setschenow constants
for our data and those of other workers, after
unit conversion if necessary (Table 5 ). As
previously discussed, small standard errors
(< lo/o) associated with constants found by
us and others indicate a good fit of the data
to Eq. 26 over the entire concentration range
to saturation. A lack of overlap of constants
measured at the same temperature, how-
ever, indicates significant systematic errors
among experiments.
Cramers Setschenow constants a re with-
in 2-3% of ours (except at OC), but have a
larger SE. His measured DO values for dis-
tilled water differ by up to 0.9 mg liter-
from accepted values and are consistently
lower than ours at other salt concentrations
(max difference, 0.9 mg liter-). In contrast,
Macarthurs value for Cc,* in distilled water
is within 0.2% of the currently accepted val-
ue, and at higher salt concentrations his val-
ues and those of Kinsman et al. are in ex-
cellent agreement. There is also good
agreement with our values; over the con-
centration range of 0-2OOY&NaCl, the max-
imal difference between their DO values and
values predicted w ith Eq. 25 is 0.08 mg li-
ter-. Agreement between all three investi-
gations is thus very good. Values of K found
by Geffcken (1904), Eucken and Hertzberg
(1950), and Finn (1967) are up to 8% dif-
ferent from ours.
NaCl solutions as models or natural wa-
ters-In most saline lakes worldwide, so-
dium and chloride ions predominate (see
appendix C, Hammer 1986). In Australia
especially, there are very few exceptions to
this pattern of ionic dominance (Williams
1967, 1986b; Hart and McKelvie 1986). It
was for these reasons that NaCl was chosen
as the principal experimental solution. Nev-
ertheless, numerous exceptions to the gen-
eral pattern of ionic dominance occur, and
it is of interest therefore to consider the na-
ture of differences in DO solubility in salt
solutions other than pure NaCl.
Studies of oxygen solubility in salt
mixtures are rare. A comparison was made
between the data of Weiss (1970) for sea-
water and the predictions of Eq. 25, using
the coefficients of Table 4. Differences be-
tween the two C,* values were calculated
at temperatures of O, 1O,20, 30, and 40C
and salinities of 0, 10,20, 30, and 40%~ In
all 25 cases, he differences were I 0.14 mg
liter-i, with the values of Weiss being less
than ours in 2 1 cases.As later results show,
these lower values are qualitatively consis-
tent w ith the presence of ions such as Mg2+
and SOA2- n seawater solutions. In 9 of the
25 cases he difference was ~0.02 mg liter-,
in 15 it was 10.06 mg liter-, and in only
4 caseswas the difference >O. 10 mg liter-.
All the latter were at 40C (1O-40?&). Equa-
tion 25 appears able to predict C,* in sea-
water or diluted seawater o within 0.15 mg
liter- over the temperature range 0-4OC.
Extrapolation of the seawater equation
(Weiss 1970) above 4Oo/oos of doubtful va-
lidity. At high salinities (-200?7m)solubili-
ties predicted by the equation were lo-20%
higher than those measured for NaCl so-
lutions. Similarly high predictions have been
found for He and Ar in Dead Sea brines
(Weiss and Price 1989).
DO concentrations in MgSO, (Fig. 2A)
and other electrolyte solutions (Fig. 2B) de-
crease less rapidly with increasing concen-
trations of these salts than is the case with
NaCl. Because he concentrations of DO in
pure NaCl and MgSO, solutions differ by
< 1 mg liter- at all but the lowest temper-
atures and highest concentrations (Fig. 2A),
it would seem hat DO solubility in mixtures
of these salts should not differ significantly
from DO solubility in pure solutions of
NaCl.
To test this, we studied mixtures of MgS04
and NaCl, as well as solutions made from
coarse sea salt. The differences of measured
DO concentrations from those predicted
with Eq. 25 for pure NaCl solutions are ~0.3
mg liter- (Table 6). As expected from Fig.
2, the predicted DO value is less than that
-
7/24/2019 D.O. in Hyper Saline Water
12/16
246
Sherwood et al.
I
0
,- I
I
50 : 100 150
200 250 300
Conceidration of Solution (g kg-l)
Fig. 2. Effect pf the nature of electrolytes on the
variation of DO wjth salt concentration. A. This study.
B. Macarthur (1916) at 25C.
measured for e~ach f the salt mixtures. Most
of the differen&esare within 0.12 mg liter-.
It should be iioted that the experimental
uncertainty of each of the single determi-
nations of DC) in Table 6 is -0.05 mg li-
ter-l. Maxim@ differences (0.2-0.3 mg li-
ter-) were foynd for the higher proportions
of MgSO, (7 : 3). It seems, hen, that values
predicted from pure NaCl solutions are like-
ly to provide near but low estimates (within
-0.1 mg liter- in many cases)of DO con-
centrations in Australian salt lakes.
Copeland (1967) carried out a limited
study of DO levels in a Mexican hypersali ne
lagoon, Laguna Tamaulipas. This coastal la-
goon is probably dominated by NaCl, al-
though details of ionic composition are not
given. For salinities < 150?&, Copelands
measured values are within 0.2 mg liter-l
of the predicted values using Eq. 2.5, with
three of the four values being within 0.1 mg
liter- (Table 7). Agreement at 220% is poor
and may be due to peculiarities in the ionic
composition of the lagoon.
;
Predictions of DO for salt mixtures-In
electrolyte solutions there is evidence that
to a good approximation each ion can be
considered to contribute independently to
the salting out effect on DO. As a con-
sequence, tables of ion-specific Setschenow
constants have been produced by Schumpe
et al. (1978) and Pawlikowski and Praus-
nitz (1983, 1984). These tables open the
possibility of predicting DO values in com-
plex salt mixtures in which ion,concentra-
tions are known. In practice, the approach
is presently limited by insufficient data.
Setschenow constants at 25C are available
for most dominant ions in natural waters
(Na+, K+, Mg2+, Ca2+, Cl-, SOd2-, C032-).
Only a few ions have had constants deter-
mined at other temperatures (Na+, K+, Cl-).
Interstudy differences in the magnitude of
Setschenow constants for a particular salt
are up to 200/6 e.g. Table 5; see also Kho-
mutov and Konnik 1974). This disparity
limits the accuracy of ion-specific con-
stants derived from these data. Despite
these restrictions, it has been possible, in
a few cases, to test the potential of this
technique for limnologists.
i
For the artificial salt mixtures given in
Table 6, the assumption of independent
ionic contributions to salting out gives
= KN,,,[NaC1l+ ~M~~,[M~SQI
= oglo + oglo(+--)
07)
-
7/24/2019 D.O. in Hyper Saline Water
13/16
Oxygen in hypersaline waters
247
Table 6. DO concentrations in artificial lake mixtures compared to those in pure NaCl solutions of the same
concentration.
DO Ima liter-9
Temp.
0
0
23.3
35.2
Concn
6s kg- )
53.4
103.0
140.4
192.6
100.0
250.0
51.8
101.3
143.1
199.6
253.2
Predicted
Fraction NaCl* MWtSUId
Eq. 25
At
Eq. 28
At
0.723 10.30 10.13 0.17 10.42 -0.12
0.717 7.36 7.08 0.28 7.49 -0.13
0.910 5.41 5.34 0.07 5.48 -0.07
0.907 3.64 3.55 0.09 3.68 -0.04
4.81 4.77 0.04
1.86 1.75 0.11
0.699 5.38 5.30 0.08 5.41 -0.03
0.707 4.15 4.05 0.10 4.24 -0.09
0.902 3.21 3.18 0.03 3.27 -0.06
0.905 2.37 2.25 0.12 2.35 +0.02
0.712 1.87 1.58 0.29 1.86 +0.01
* Fraction by mass of NaCl in salt mixtures.
t Difference between measured and predicted values.
I) Solutions prepared from coarse sea salt.
Thus,
crnix =
c
NaCl cMgSOzt
G
(28)
where Cmix s the DO concentration in the
salt mixture, C,,,, that in a pure NaCl so-
lution of concentration [NaCl], and C,,,,
that in a pure MgS04 solution of concen-
tration [MgSO,].
From the data of Fig. 2A, Setschenow
constants were determined for pure MgSO,
at 35.2Y (0.228, s = 0.004 kg mol-I,
n =
6) and 5.2% (0.279, s = 0.008 kg mol-l, n
= 6). They were used to estimate a value
for MgSO, at 0C (0.288 kg mol-I), assum-
ing K is a linear function of temperature.
We have found this assumption to be true
for NaCl at these temperatures
(see Table
5) as have other workers (Pawlikowski and
Prausnitz 1983, 1984). Values of the Set-
schenow constants at 0C and 35.2C were
used to predict CMgsoq or each of the
mixtures in Table 6. Values for C,,, were
determined with Eq. 25. Substituting them
into Eq. 28 allowed prediction of the CmiX
values given in Table 6. The differences be-
tween measured and predicted values are
generally smaller than those found with pre-
dicted DO values for NaCl solutions of the
same total salinity (i.e. based on Eq. 25).
This improvement is particularly noticeable
at higher salinities.
Schumpe et al. (1978) have determined
Setschenow constants at 25C for individual
ions. Their equation for predicting DO in
salt mixtures is
where Hi is the Setschenow constant for an
ion of type i, [Ml, the concentration (mol
liter-) of ions of type
i,
and zi the valency
of ions of type i. Since they list values of Hi
for all major ions in seawater, t was possible
to predict DO for seawater and Laguna Ta-
maulipas at 25C (Table 7). For the latter-
a coastal lagoon- we assumed a relative
ionic composition the same as seawater and,
since I& values are given in liters mol-l, a
density for lagoon water equal to a NaCl
solution of the same salinity. The salting out
effects of C0,2- and HCO,- were also ne-
glected. Although agreement between mea-
sured and predicted DO values is better than
that given by Eq. 25 for seawater and 957~
lagoon water, it is -0.15 mg liter- worse
at 150 and 22OY& Given the assumption
outlined above, the agreement should be
considered satisfactory but not superior to
results obtained from DO values in pure
NaCl solutions.
Pawlikowski and Prausnitz (1983, 1984)
have also given ion-specific Setschenow
constants at 25C although electrolyte con-
centrations are expressed as molality (mol
kg-). Unfortunately for limnologists, no
data are given for Mg2+ or C032-, greatly
-
7/24/2019 D.O. in Hyper Saline Water
14/16
248
Sherwood et al.
Table 7. Comparison of equilibrium DO concentrations measured in seawater and Laguna Tamaulipas,
Mexico (Copeland 1967) and predicted with Eq. 25 and 29.
--
----
--
W (mg iter-)
--
Temp. COOCll
(cl
(g Is-l)
25 ,zst
s5*
1508
220*
37 105*
l iO$
2> 0$
MWlSURd
Eq. 25 A*
6.60 6.79 -0.19
4.62 4.79 -0.17
3.50 3.40 +0.10
2.81 2.13 +0.68
3.85 3.90 -0.05
3.12 3.01 f0.11
2.50 2.96 +0.54
Predicted
----
Eq. 29
CL*
--
6.75 -0.1s
4.69 -0.01
3.27 40.23
1.98 to.83
* Difference between m c;lsured and predicted DO values.
f Seawater.
# Laguna Tamaulipas, h&&o.
limiting application of these results to nat-
ural
waters. They have, however, produced
equations to predict the variation of Set-
schenow constants with temperature over
the range 0-6OC. These equations are based
partly on theoretical considerations and have
the form
k = aIs + a2J + (hs + ~2S~Wk)gas
(30)
where
k,
is the ion-specific Setschenow con-
stant (kg mol-*),
T
is Kelvin, (e/k)gas s the
Lennard-Jones parameter for O2 (118 K),
and a s, a2s,&, b,, are ion-specific coeffi-
cients.
DO values were predicted for a 2007~ so-
lution at 0C (3.02 mg liter-), 25C (2.46
mg liter-), and 40C (2.19 mg liter-) using
Eq. 30 and th-e coefficients for NaCl. They
differ by -0.32, 0.01 and 0.01 mg liter-
respectively worn values predicted by Eq.
.25. Although coefficients for Eq. 30 are giv-
en for Na+, K+, and Cl-, they are insufficient
to allow its application to most natural wa-
ters. We are .currently working with lim-
nologically important salts to overcome this
deficiency.
One final Iboint should be made about
predicting DQ concentrations. Because of a
lack of consensus among researchers, values
for DO, salt concentrations, and Setsche-
now constants are reported in both volume-
based (e.g. ml; liter-) and mass-based con-
centrations units (e.g. mol kg-). Although
it may be de&able for all concentrations to
be given in temperature-independent units
based on mass (e.g. mg kg- or mol kg-),
the convenience of volumetric glassware in
laboratory procedures means that this goal
commonly is not achieved. Accordingly, to
get most access to published data, limnol-
ogists must have accurate density tables for
a particular water body. Ideally, equations
of state that give the variation of density
with salinity and temperature should be cal-
culated. Such equations are known, for ex-
ample, for seawater (Miller0 and Poisson
198 1) and Dead Sea brines (Krumgalz and
Miller0 1982). Changes in relative ionic
composition over time may complicate de-
termination of the equation of state for a
body of water.
Conclusion
The constancy of relative ionic compo-
sition of estuarine and marine waters has
made possible the preparation of precise
(a0.02 mg liter-) solubility tables for ox-
ygen (e.g. Weiss 1970). A general table of
comparable precision and simplicity for hy-
persaline lakes cannot be achieved at pres-
ent because of the great variability in their
ionic composition. This variability affects
the equilibrium concentration of oxygen.
Nevertheless, since most saline lakes are
dominated by NaCl, a table based on DO
solubility in pure NaCl solutions has some
utility. A table of this sort can be generated
with Eq. 25. The data given shpuld be gen-
erally reliable to within 0.2 mg liter- DO
for lakes dominated by NaCl (L70% by
mass).
It is possible, in theory, to predict DO
concentrations for salt mixtures such as hy-
-
7/24/2019 D.O. in Hyper Saline Water
15/16
Oxjrgen in hype73aline waters
persaline natural waters. In practice, re-
stricted data mean that this approach pres-
ently offers no advantages over the use of
DO concentrations estimated from model
NaCl solutions. Research to determine ion-
specific Setschenow constants over a range
of temperatures for ions of limnological sig-
nificance is needed to change this situation.
References
AMERICAN PUBLIC HEALTH ASSOCIATION.1985. Stan-
dard methods for the examination of water and
wastewater, 16th ed.
BAITINO, R., AND H. L. CLEVER. 1966. The solubility
of gases n liquids. Chem. Rev. 66: 395-463.
BENSON, B. B., AND
D. KRAUSE,JR. 1980. The con-
centration and isotopic fractionation of gasesdis-
solved in fresh water in equilibrium with the at-
mosphere: 1. Oxygen. Limnol. Oceanogr. 25: 662-
671.
CARPENTER, . H. 1966. New measurements of ox-
ygen solubility in pure and natural water. Limnol.
Oceanogr. 11: 264-277.
CARRIIT, D. E., AND J. H. CARPENTER. 1966. Com-
parison and evaluation of currently employed
modifications of the Winkler method for deter-
mining dissolved oxygen in sea water; a NASCO
report. J. Mar. Res. 24: 286-318.
COPELAND, . J. 1967. Environmental characteristics
of hypersaline lagoons. Publ. Univ. Texas Int. Mar.
Sci. 12: 207-218.
CRAMER, S. D. 1980. The solubility of oxygen in
brines from 0 to 300C. Ind. Eng. Chem. Process
Design Dev. 19: 300-305.
EUCKEN,A., AND G. HERT~BERG.1980. Aussalzeffekt
und Ionenhydration. Z. Phys. Chem. (Leipzig) 195:
l-23.
FINN, R. K. 1967. Agitation and aeration, p. 69-99.
In N. B lakebrough [ed.], Biochemical and biolog-
ical engineering science. V. 1. Academic.
GEFFCKEN,G. 1904. Beitrage zur Kenntnis der Los-
lichkeitsbeeinflussung. Z. Phys. Chem. (Leipzig)
49: 257-302.
GREEN,E. S., AND D. E. CARRITT.
1966. An improved
iodine determination flask for whole-bottle titra-
tions. Analyst
91: 207-208.
HAMMER, U. T. 1986. Saline lake ecosystems of the
world. Junk.
HANDBCIOKOF Pnvs~cs AND CHEMISTRY. 1980. 6 1st
ed. CRC.
HART, B. T., AND I. D. MCI(ELVIE. 1986. Chemical
limnology in Australia, p. 3-3 1. In P. De Deckker
and W. D. Williams [eds.], Limnology in Austra-
lia. CSIRO and Junk.
JONES,C. T., C. G. CLYDE, W. J. GRENNEY,AND J. P.
RILEY. 1976. Development of a water quality
simulation model applicable to Great Salt Lake.
Utah. Utah State Univ. Water Res. Lab. Rep:
PRJEWOlB-1. 129 D.
KINSMAN, D. J. J., M. BOARDMAN,AND M. BORCSIK.
1974. An experimental determination of the sol-
ubility of oxygen in marine brines, p. 325-327. In
249
Proc. 4th Int. Symp. Salt. V. 1. Northern Ohio
Geol. Sot.
&IOMUTOV, N. E., AND E. I. KONNIK. 1974. Solu-
bility of oxygen in aqueous electrolyte solutions.
Russ. J. Phys. Chem. 48: 359-362.
KRUMC+ALZ, . S., AND
F.
J. MILLERO. 1982. Physico-
chemical study of Dead Sea waters. 2. Density
measurements and equation of state of Dead Sea
waters at 1 atm. Mar. Chem. 11: 477-492.
L~EFLICH, L. A. 1972. Studies on the brine flagellate
Dunufielfa salina. Ph.D. thesis, Univ. California,
San Diego.
LONG. F. A.. AND W. F. MCDEVIT. 1952. Activity
coefficients of nonelectrolyte solutes in aqueous
salt solutions. Chem. Rev. 51: 119-169.
MACARTHUR,C. G. 19 16. Solubility of oxygen in salt
solutions and the hydrates of these salts. J. Phys.
Chem. 20: 495-502.
MAJOR, G. A., G. DALPON~,J. KYLE, AND B. NEWELL.
1972. Laboratory techniques in marine chemis-
try, p. 6-9. CSIRO D iv. Fish. Oceanogr. Rep. 51.
MILLERO,F. J. 1986. Solubility of oxygen in seawater.
Zn UNESCO Tech. Pap. Mar. Sci. 50, Appendix 1.
-,
AND
A. POISSON. 198 1. International one-
atmosphere equation of state of seawater. Deep-
Sea Res. 28: 625-629.
MORTIMER, C. H. 198 1. The oxygen content of air-
saturated fresh waters over ranges of temperature
and atmospheric pressure of limnological interest.
Mitt. Int. Ver. Theor. Angew. Limnol. 22, p. l-
23.
PAWLIKOWSKI,E. M., AND J. M. PRAUSNITZ. 1983,
1984. Estimation of Setschenow constants for
gases n common salts at moderate temperatures.
Ind. Eng. Chem. Fund. 22: 86-90. Correction. 23:
210.
PERRY,R. H., AND D. W. GREEN.
1984. Perrys chem-
ical engineers handbook, 6th ed. McGraw-Hill.
SCHUMPE, ., I. ADLER, AND W. D. DECKWER.
1978.
Solubility of oxygen in electrolyte solutions. Bio-
technol. Bioeng. 20: 145-150.
STECHER, . G.,
M.
J. FINKEL, AND 0.
H.
SEIGMUND
[eds.]. 1960. The Merck index of chemicals and
drugs, 7th ed.
VOGEL, A. I. 1961. A textbook of auantitative in-
organic analysis, 3rd ed. Longmans.
WALKER,K. F., W. D. WILLIAMS. AND U. T. HAMMER.
1970. The Miller methodfor oxygen determi-
nation applied to saline lakes. Limnol. Oceanogr.
15: 814-815.
WASHBURN,E. W. [ed.] 1928. International critical
tables. V. 3. McGraw-Hill.
WIESS,R. F. 1970. The solubility of nitrogen, oxygen
and argon in water and seawater. Deep-Sea Res.
17: 721-735,
-. 1974. Carbon dioxide in water and seawater:
The solubility of a non-ideal gas. Mar. Chem. 2:
203-215.
-
AND
B. A. PRICE. 1980. Nitrous oxide solu-
biiity in water and seawater. Mar. Chem. 8: 347-
359.
-,
AND
-.
1989. Dead Sea gas solubilities.
Earth Planet. Sci. Lett. 92: 7-10.
WILLIAMS, W. D. 1967. The chemical characteristics
-
7/24/2019 D.O. in Hyper Saline Water
16/16
250
Sherwood et al.
and their fauni: l&en studies. Aust. Natl. Univ.
-. 1986a. Clmnology, the study of inland wa-
ters: A comment on perceptions of studies of salt
lakes, past andpresent, p. 47 l-484. In P. De Deck-
ker and W. D. ,Williams [eds.], Limnology in Aus-
tralia. CSIRO and Junk.
of lentic surIa