Professional Documents
Culture Documents
1988,22,382-387
Discussion
The theoretical basis for a linear relationship between
log KO,and log a has been developed by Wasik et al. (30).
The high significance ( r = 0.951, Sy,x= 0.332,n = 30) of
the relationship between these parameters with our ex-
perimental results for selected diverse PCB congeners
suggests that log KO,values for other PCB’s could be es-
timated with reasonable precision from such a relationship.
It must be emphasized however that it would be expected
Table 11. log KO,Values for PCB Congeners
congener substitution TSA,b congener substitution TSA,b
no. patterna ~ 1 0 - 2 0m2 log KO, no. patterna X ~ O - ~ m2
O log KO,
0 biphenyl 184.43 4.09 75 2,4,4’,6- 241.85 6.05
1 2- 195.45 4.46 76 2,3’,4’,5’- 244.35 6.13
2 3- 201.95 4.69 77 3,3’,4,4’- 251.02 6.36
3 4- 202.12 4.69 78 3,3’,4,5- 250.85 6.35
4 2,2’- 200.80 4.65 79 3,3’,4,5’- 252.77 6.42
5 2,3- 210.34 4.97 80 3,3’,5,5’- 254.51 6.48
6 2,3’- 212.97 5.06 81 3,4,4’,5- 251.02 6.36
7 2,4- 213.14 5.07 82 2,2’,3,3’,4- 246.36 6.20
8 2,4‘- 213.14 5.07 83 2,2’,3,3‘,5- 248.10 6.26
9 2,5- 212.97 5.06 84 2,2’,3,3’,6- 241.60 6.04
10 2,6- 206.46 4.84 85 2,2’,3,4,4’- 249.16 6.30
11 3,3’- 219.47 5.28 86 2,2’,3,4,5- 247.07 6.23
12 3,4- 217.73 5.22 87 2,2’,3,4,5’- 248.99 6.29
13 3,4‘- 219.64 5.29 88 2,2’,3,4,6- 242.48 6.07
14 3,5- 219.47 5.28 89 2,2’,3,4,6’- 242.48 6.07
15 4,4‘- 219.81 5.30 90 2,2’,3,4’,5- 250.90 6.36
16 2,2’,3- 215.69 5.16 91 2,2‘,3,4‘,6- 244.40 6.13
17 2,2’,4- 218.50 5.25 92 2,2’,3,5,5’- 250.73 6.35
18 2,2’,5- 218.32 5.24 93 2,2’,3,5,6- 241.60 6.04
19 2,2’,6- 211.82 5.02 94 2,2’,3,5,6’- 244.23 6.13
20 2,3,3’- 227.86 5.57 95 2,2‘,3,5‘,6- 244.23 6.13
21 2,3,4- 226.11 5.51 96 2,2‘,3,6,6’- 232.06 5.71
22 2,3,4’- 228.03 5.58 97 2,2’,3,4’,5’- 248.99 6.29
23 2,3,5- 227.86 5.57 98 2,2’,3,4’,6’- 244.40 6.13
24 2,3,6- 221.35 5.35 99 2,2’,4,4’,5- 251.79 6.39
25 2,3‘,4- 230.66 5.67 100 2,2’,4,4’,6- 247.20 6.23
26 2,3’,5- 230.49 5.66 101 2,2’,4,5,5’- 251.62 6.38
27 2,3’,6- 223.99 5.44 102 2,2’,4,5,6’- 254.12 6.16
28 2,4,4’- 230.83 5.67 103 2,2’,4,5’,6- 247.03 6.22
29 2,4,5- 228.74 5.60 104 2,2’,4,6,6’- 234.87 5.81
30 2,4,6- 224.16 5.44 105 2,3,3’,4,4’- 259.41 6.65
31 2,4’,5- 230.66 5.67 106 2,3,3’,4,5- 259.24 6.64
32 2,4’,6- 224.16 5.44 107 2,3,3’,4’,5- 261.15 6.71
33 2,3’,4’- 228.75 5.60 108 2,3,3’,4,5’- 261.15 6.71
34 2,3‘,5’- 230.49 5.66 109 2,3,3’,4,6- 254.65 6.48
35 3,3‘,4- 235.25 5.82 110 2,3,3’,4’,6- 254.65 6.48
36 3,3’,5- 236.99 5.88 111 2,3,3’,5,5’- 262.90 6.76
37 3,4,4’- 235.42 5.83 112 2,3,3’,5,6- 253.76 6.45
38 3,4,5- 233.33 5.76 113 2,3,3’,5’,6- 256.39 6.54
39 3,4’,5- 237.16 5.89 114 2,3,4,4’,5- 259.41 6.65
40 2,2’,3,3’- 230.58 5.66 115 2,3,4,4’,6- 254.82 6.49
41 2,2’,3,4- 231.47 5.69 116 2,3,4,5,6- 250.10 6.33
42 2,2’,3,4’- 233.38 5.76 117 2,3,4‘,5,6- 253.93 6.46
43 2,2’,3,5- 233.21 5.75 118 2,3’,4,4’,5- 262.04 6.74
44 2,2‘,3,3‘- 233.21 5.75 119 2,3’,4,4’,6- 257.45 6.58
45 2,2‘,3,6- 226.71 5.53 120 2,3‘,4,5,5‘- 263.78 6.79
46 2,2’,3,6’- 223.71 5.53 121 2,3’,4,5’,6- 259.20 6.64
47 2,2‘,4,4’- 236.19 5.85 122 2,3,3’,4’,5’- 259.24 6.64
48 2,2’,4,5- 234.10 5.78 123 2,3,4,4’,5’- 262.04 6.74
49 2,2’,4,5’- 236.01 5.85 124 2,3’,4’,5,5’- 261.87 6.73
50 2,2’,4,6- 229.51 5.63 125 2,3’,4’,5’,6- 255.37 6.51
51 2,2’,4,6’- 229.51 5.63 126 3,3‘,4,4‘,5- 266.63 6.89
52 2,2’,5,5’- 235.84 5.84 127 3,3‘,4,5,5‘- 268.37 6.95
53 2,2’,5,6’- 229.34 5.62 128 2,2’,3,3’,4,4’- 262.13 6.74
54 2,2’,6,6’- 217.18 5.21 129 2,2’,3,3’,4,5- 261.96 6.73
55 2,3,3’,4- 243.63 6.11 130 2,2’,3,3’,4,5’- 263.88 6.80
56 2,3,3’,4’- 243.63 6.11 131 2,2’,3,3’,4,6- 257.37 6.58
57 2,3,3’,5- 245.38 6.17 132 2,2’,3,3’,4,6’- 257.37 6.58
58 2,3,3’,5’- 245.38 6.17 133 2,2’,3,3’,5,5’- 265.62 6.86
59 2,3,3’,6- 238.87 5.95 134 2,2’,3,3’,5,6- 256.49 6.55
60 2,3,4,4’- 243.80 6.11 135 2,2’,3,3’,5,6’- 259.12 6.64
61 2,3,4,5- 241.72 6.04 136 2,2’,3,3’,6,6’- 246.95 6.22
62 2,3,4,6- 237.13 5.89 137 2,2’,3,4,4’,5- 264.76 6.83
63 2,3,4’,5- 245.55 6.17 138 2,2’,3,4,4’,5’- 264.76 6.83
64 2,3,4’,6- 239.04 5.95 139 2,2’,3,4,4’,6- 260.18 6.67
65 2,3,5,6- 236.24 5.86 140 2,2’,3,4,4’,6’- 260.18 6.67
66 2,3’,4,4‘- 246.44 6.20 141 2,2’,3,4,5,5’- 264.59 6.82
67 2,3‘,4,5- 246.26 6.20 142 2,2’,3,4,5,6- 255.46 6.51
68 2,3‘,4,5‘- 248.18 6.26 143 2,2’,3,4,5,6’- 258.09 6.60
69 2,3’,4,6- 241.68 6.04 144 2,2’,3,4,5’,6- 260.00 6.67
70 2,3’,4’,5- 246.26 6.20 145 2,2’,3,4,6,6’- 247.84 6.25
71 2,3’,4’,6- 239.76 5.98 146 2,2’,3,4’,5,5’- 266.51 6.89
72 2,3’,5,5’- 248.01 6.26 147 2,2‘,3,4‘,5,6- 259.29 6.64
73 2,3’,5’,6- 241.50 6.04 148 2,2‘,3,4‘,5,6‘- 261.92 6.73
74 2,4,4’,5- 246.43 6.20 149 2,2’,3,4’,5’,6- 260.00 6.67
because it is nonselective and can be operated at high almost 1order of magnitude. To facilitate the comparison
temperatures (31). For small values of log a there is a close of TSA with log KO,and encompass such variation, it is
correspondencebetween experimental log a values on C-87 appropriate to accentuate the difference in TSA between
and those calculated for Apiezon L (which has McEEeynolds these isomers. This was achieved by consideration of the
constants close to those of C-87). But with larger log a surface areas of PCB molecules in a planar configuration,
values there is an increasing disparity (32). Retention with a Oo dihedral angle between the phenyl rings. In this
times of all PCB congeners on (2-87 are thus difficult to configuration, much of the exposed surface area of ortho
predict from results obtained with Apiezon L and more substituents, particularly chlorine, is reduced by mutual
difficult with more diverse stationary phases. Therefore, occlusion. It is emphasized that for most PCB congeners
the relationship expressed by eq 2 can only be effectively this planar conformation is a hypothetical one that cannot
used for estimation of unknown log KO,values, where the exist in reality because of excessive steric ortho-ortho
relative retention time is measured directly on (2-87 sta- interactions.
tionary phase. Surface areas of all PCB congeners have been calculated
Partition coefficients are related to the free-energy by Armstrong ( 2 4 2 9 )on the basis of a computer program
change on transfer between two phases. Also for groups using the van der Waals radii of component atoms, zero
of chemically similar compounds there is much empirical solvent radius, and appropriate bond distances and angles.
evidence to suggest that log KO,values are related to the The interplanar angle can be arbitrarily specified, and thus,
aqueous solvent cavity size or the size of the solute mol- an accurate computation of the TSA for all congeners in
ecule (23,33-35). For example, a linear relationship be- a planar configuration can be obtained (see Table 11). In
tween molar volume (cm3 mol-’) and log KO, has been addition, we have found using a program originally written
found for 17 ortho-substituted PCB congeners (27). by Hermann (QCPE No. 225) (35)that areas comparable
However, the molar volumes were calculated by the Le Bas to those listed in Table TI are obtained with similar input
method with a constant increment for chlorine regardless data.
of position. Such calculations cannot account for variations A plot of geherator column derived log KO, values
in substitution patterns between isomers that have been against TSA, in a planar configuration, is shown in Figure
observed to have marked effect on log KO,(see Table I). 2, yielding a significant linear relationship, expressed by
Pearlman (36)has indicated that there is in general, little eq 3 (r = 0.959, S,,% = 0.320, n = 46), for PCB’s ranging
difference between the use of surface area and the use of from the parent biphenyl to decachlorobiphenyl and con-
volume for these correlations. Surface area computations taining all levels of ortho substitution. The outlying points
with a range of compounds have proven reliable estimates in Figure 2 do not always represent those congeners for
of log KO,values and are also sensitive to structural var- which forced planarity is a poor model. This suggests that
iation (24, 26, 35, 37). this relationship could be used to provide reasonable es-
The experimental log KO,determinations for a series of timates of previously unknown log KO,values of PCB
tetrachlorobiphenyls (Table I) reveal that the values are congeners. The calculated log KO,values of all the PCB
smallest for the fully ortho substituted isomers and congeners are shown in Table 11. It is noteworthy that
greatest for the isomer with no o-chlorines and extend over these data include values for non-ortho-substituted PCB’s,
Ehviron. Sci. Technol., Vol. 22, No. 4, 1988 385
which are not present to any significant extent in com- For compounds that are liquid at the system temperature,
mercial products but have been shown to be the most toxic the melting point T M is set equal to T and hence the
(38). entropic term (ASf/2.303R)(l - TM/T) reduces to zero.
The inadequacy of the fragmental constant approach to Correlating these log S - (ASf/2.303R)(1- TM/T)data
prediction of log KO,values is revealed by consideration with TSA values from Table 11, an expression for solubility
of the TSA data. For example, TSA increases on sub- in terms of TSA results (r = 0.949, S,,x = 0.464, n = 17):
stitution in the 2-, 3-, and 4-positions of biphenyl are 11.0
X 17.5 X and 17.7 X m2,respectively, and log S = (-4.13 X 10-’)TSA +- (1 - $) + 3.48
with 2,2’-dichlorobiphenyl, TSA takes into account o- 2.303R
chlorine overlap and is not merely the sum of biphenyl and (6)
two ortho contributions. In addition, allowance is made
for overlap with adjacent chlorine substituents (e.g., 3,4- Assuming the entropy of fusion to be approximately 54.8
dichlorobiphenyl). However, there are few examples of J K-l mol-l (18),then the maximum aqueous solubility of
systematic increases in surface area with chlorine substi- a PCB congener can be estimated from its melting point
tution, one being biphenyl, 3-chlorobiphenyl, 3,3’-di- and TSA (Table 11) with eq 6.
chlorobiphenyl, 3,3’,5-trichlorobiphenyl,and 3,3’,5,5‘- Calculated solubilities compare favorability to published
tetrachlorobiphenyl. TSA calculations thus provide for data. The value for 4,4’-dichlorobiphenyl (1.95 X mol
a greater precision in recognizing subtle variations between L-l) is in good agreement with literature solubilities of 2.78
PCB congeners. X (IO),2.51 X (41),and 3.59 X mol L-l (42).
Of the experimentally determined log KO,values, only There is similar agreement between the calculated solu-
those for 2,3,4,5-tetra- (6.18) and 2,2’,3,3’,6,6’-hexachloro- bility of decachlorobiphenyl (1.62 X mol L-l) and
biphenyl (5.76) have been measured before, and the values published values of 1.49 X (18), 3.21 X 10-l’ (41),and
obtained in that study were 5.72 and 6.63, respectively (33). 8.02 X mol L-I (43).
There is little reliable partition coefficient data concerning
most PCB’s, and much of that we have used in deriving Conclusions
our correlation. In order to judge the validity and pre- These results indicate that there is a direct linear rela-
dictive utility of eq 3, however, the calculated log Kowvalue tionship between log KO,and log a,when a nonselective
of 4,4‘-dichlorobiphenyl (5.30) may be compared with gas chromatographic stationary phase is used, expressed
published values of 5.58 (IO),4.82 (12),4.92 (14),5.36 (In, by eq 2. It is suggested that this relationship could be used
and 5.33 (19) and that of decachlorobiphenyl (8.18) with to obtain the KO,values for all the PCB congeners if the
previous values of 8.26 (18) and 8.20 (19). Agreement is relative retention times are available.
quite reasonable considering that the partition coefficients Also, a relationship between log KO,and TSA has been
differ by 3 orders of magnitude. established for a diverse group of PCB congeners by cal-
The aqueous solubility of these compounds is also of culating TSA for molecules in which the two phenyl rings
interest. Knowledge or reliable estimates of log Kowpro- are coplanar, Le., eq 3. This equation has been used to
vide a method for determination of maximum aqueous calculate the log KO,values for all of the PCB congeners.
solubilities. Relationships between the logarithm of On the basis of this and other relationships, an expression
maximum aqueous solubility [log S (mol L-l)] and parti- for water solubility of the PCB’s was developed, Le., eq 6.
tion coefficient were originally theorized to be of an inverse In addition, it is suggested that this method could be
linear form (11,24,39). Including a term to account for used for prediction of log KO,values for other groups of
the crystal lattice interactions of solids, equations of the compounds such as polybrominated biphenyls and poly-
following general form were proposed: halodibenzodioxins and -furans. However, constants for
log KO, = the appropriate equation need to be established initially
-log ”-(
s + 2.303R 1- %) - log yo* - log vo* (4)
from experimental log KO,values for a small number of
compounds from each group.
Acknowledgments
where AS, is the entropy of fusion (J K-l mol-l), R is the
gas constant (8.314 J K-l mol-l), T M is the melting point We gratefully acknowledge the helpful assistance of
(K) of the solid solute, T i s the system temperature (K), David Armstrong, University of Wisconsin, for providing
yo* is the activity coefficient in water-saturated octanol, PCB surface area data.
and Vo*is the molar volume of water-saturated octanol Registry No. 1, 2051-60-7; 2, 2051-61-8; 3, 2051-62-9; 4,
(126.6 X L mol-’) (33). 13029-08-8; 5, 16605-91-7; 6, 25569-80-6; 7, 33284-50-3; 8,
These derivations, possessing a negative unitary slope, 34883-43-7; 9, 34883-39-1; 10, 33146-45-1; 11, 2050-67-1; 12,
assume a constant value of log yo*. Recently, it has been 2974-92-7; 13, 2974-90-5; 14, 34883-41-5; 15, 2050-68-2; 16,
found that for PCB’s and chlorobenzenes log yw (the 38444-78-9; 17, 37680-66-3; 18, 37680-65-2; 19, 38444-73-4; 20,
aqueous activity coefficient) and hence log S together with 38444-84-7; 21, 55702-46-0; 22, 38444-85-8; 23, 55720-44-0; 24,
55702-45-9; 25, 55712-37-3; 26, 38444-81-4; 27, 38444-76-7; 28,
log yo* are linearly related to solute molar volume due to 7012-37-5; 29, 15862-07-4; 30, 35693-92-6; 31, 15862-07-4; 32,
increasing size and nonideality (33). On this basis, it has 38444-77-8; 33, 38444-86-9; 34, 37680-68-5; 35, 37680-69-6; 36,
been proposed that log KO,to log S correlations should 38444-87-0; 37, 38444-90-5; 38, 53555-66-1; 39, 38444-88-1; 40,
have a slope slightly greater than -1, in agreement with 38444-93-8; 41, 52663-59-9; 42, 36559-22-5; 43, 70362-46-8; 45,
experimental correlations where the slope is commonly of 70362-45-7; 46, 41464-47-5; 47, 2437-79-8; 48, 70362-47-9; 49,
the order of -0.8 (11,24). 41464-40-8; 50, 62796-65-0; 51, 68194-04-7; 52, 35693-99-3; 55,
Using PCB solubility data from Miller (I@, log KO,is 74338-24-2; 56, 41464-43-1; 57, 70424-67-8; 58, 41464-49-7; 59,
related to log S for these compounds by 74472-33-6; 60, 33025-41-1; 62, 54230-22-7; 63, 74472-34-7; 64,
52663-58-8; 67, 73575-53-8; 68, 73575-52-7; 69, 60233-24-1; 70,
log KO, = -0.795
2.303R
(1 - %)]+ 0.960
32598-11-1; 71, 41464-46-4; 72, 41464-42-0; 73, 74338-23-1; 74,
32690-93-0; 75, 32598-12-2; 76, 70362-48-0; 78, 70362-49-1; 79,
41464-48-6; 80, 33284-52-5; 81, 70362-50-4; 82, 52663-62-4; 83,
(5) 60145-20-2; 84, 52663-60-2; 85, 65510-45-4; 86, 55312-69-1; 87,