CS-1420 October, 1985 PARTITION COEFFICIENT (_n-OCTANOL/WATER) GENERATOR COLUMN METHOD OFFICE OF TOXIC SUBSTANCES OFFICE OF PESTICIDES AND TOXIC SUBSTANCES U.S. ENVIRONMENTAL PROTECTION AGENCY WASHINGTON, DC 20460 ------- CS-1420 (October, 1985) TABLE OF CONTENTS Page I. NEED FOR THE TEST...... 1 II. SCIENTIFIC ASPECTS 2 A. Rationale For the Test Method 2 B. Other Methods For Determining Kow 3 1. The Conventional Shake-Flask Method 3 2. Reverse-Phase High-Pressure Liquid Chromatography as a Method of Estimating Kow.... 4 3. Estimation from Water Solubility 6 4. Estimation Using the Fragment Constant Method... 8 5. Estimation Using Thin-Layer Chromatography 10 C. Rationale for the Test Conditions 10 1. Special Laboratory Equipment 10 2. Temperature Control 11 3. Purity of Octanol, Water and Other Solvents 11 D. Test Data and Reporting 12 1. Test Report 12 2. Analytical Procedures 13 III. REFERENCES 14 ------- CS-1420 (October, 1985) I. NEED FOR THE TEST The octanol/water partition coefficient (Kow) is one of the most frequently measured and most widely used parameters in assessing the environmental fate of organic chemicals. Its importance is due primarily to the fact that Kow is used to predict the general lipophilicity or hydrophobicity of a chemical. Of greater significance on the quantitative level, however, is the fact that KQW has been found to correlate with the water solubility, the soil/sediment sorption coefficient and the bioconcentration factor. Although all of the above mentioned properties are important in predicting the fate and residence time of a chemical in the environment, the bioconcentration factor (BCF) carries with it the greatest consequences. This is because the BCF is used to predict the potential for a chemical to bioconcentrate in living tissue. When applied to the aquatic environment, the BCF is indicative of the potential for a chemical to bioconcentrate in the tissue of fish and other lower aquatic organisms. In terms of the Kow, chemicals with a value less than 10 will not bioconcentrate while those with a value greater than 104 will. Thus, although a chemical may be present in a stream or lake at subtoxic concentrations, if its Kow is greater than 104 it could bioconcentrate and accumulate to levels that may be toxic to not only the organism itself, but also to the consumers of that organism, not the least important of which are Homo sapiens. 1 ------- CS-1420 (October, 1985) The direct determination of Kow, using the conventional shake-flask method, CS-1400 (EPA, 1983), is subject to numerous experimental difficulties. These include the formation of colloidal dispersions (emulsions) during the shaking step, incomplete separation of the octanol and water phases, difficulty in analyzing very hydrophobic chemicals in the water phase, the adsorption of the solute onto the surfaces of transfer vessels, and the loss of volatile solute into the atmosphere. In addition, the method is time consuming. II. SCIENTIFIC ASPECTS A. Rationale for the Test Method This test method uses a chromatographic technique for determining the octanol/water partition coefficient. The method, called the generator column method, was developed and validated by Wasik et al (1981) and Tewari et al (1982); using this method KQW is calculated Kow as the ratio of the molar concentration of a 1.0% (w/w) solution of the test substance in octanol to the molar concentration of the test substance in water, at equilibrium. The advantages of the generator column method include: (1) all the difficulties of the shake-flask method are avoided; (2) equilibrium is achieved rapidly; (3) easy to carry out and gives precise and accurate results (±3%), especially for very hydrophobic chemicals and (4) is applicable to a wide range 2 ------- CS-1420 (October, 1985) of chemicals with different functional groups and over a wide range of log Kow (1 to >6). Wasik et al (1981) and Tewari et al. (1982) have also provided a thermodynamic basis for their method which was validated on 62 organic test substances falling into seven general chemical classes. Their validation studies showed excellent agreement between the values of KQW obtained using the generator column method and those obtained using the standard shake flask method. Their results established that the data obtained using the generator column are precise and accurate to ±3% and that their method is the best currently available method for determining Kow. B. Other Methods for Determining 1. The Conventional Shake Flask Method The conventional method for determining the octanol/water coefficient involves shaking a solute with two immiscible solvents and then measuring the solute concentration in the two solvents after equilibration (Leo, et al. 1971). The distribution of the solute between the two solvents is a direct consequence of the thermodynamic requirements for equilibrium that apply only to dilute solutions. The resulting ratio of the two solute concentrations is the partition coefficient which is constant at a given temperature. Numerous researchers have used this method and published their results (Fujita et al. 1964; 3 ------- CS-1420 (October, 1985) Hansch and Anderson 1967; Leo et al. 1971; Chiou et al. 1977). Both EPA (1983) and OECD (1981) specify the use of this method in their protocols (see CG-1400). Although there is no ASTM Standard Test Method for Kow, the shake-flask method is the "standard method" used in industry when Kow is to be determined. As noted earlier, the shake flask method is not without problems, the most important of which are the formation of emulsions, surface adsorption of the solute, and the time consuming nature of the test procedure. However, with suitable care and patience these problems are minimized and basic experimental data are obtained that can be correlated with other important environmental fate parameters. 2. Reverse-Phase High-Pressure Liquid Chromatography as a Method of Estimating A rapid method based on reverse-phase high-pressure liquid chromatography has been developed by Veith (Veith et al. 1979) to estimate the octanol/water partition coefficient of organic chemicals. This method has been incorporated into test guideline CG-1410: Partition Coefficient (jv-Octanol/Water) Estimation by Liquid Chromatography (EPA, 1983). A summary of the method follows. Using the solvent mixture water/methanol (15/85 v/v) as the elutant, the log of the retention time [log (tR)] of organic chemicals on a permanently bonded (C-18) reverse-phase high- 4 ------- CS-1420 (October, 1985) pressure liquid chromatographic system has been found to be linearly related to log KQW. This relationship has been expressed by the equation log KQW = A log (tR) - B, (1) where A and B are constants determined from the experimental data for some organic chemicals. Using a mixture of the chemicals benzene, bromobenzene, biphenyl, -DDE [2,2-bis(p-chloro- phenyl) -1,1-dichloroethylene] and 2,4,5,2•,51-pentachloro- biphenyl, A and B were found to be 5.106 and 1.258, respectively, with a coefficient of determination of 0.975. It must be emphasized that this correlation is limited with respect to being representative of the organic chemicals encountered. This calibration mixture was selected largely on the basis of the log Kow values reported in the literature, and the correlation is linear over five orders of magnitude of Kow. To determine the accuracy of this method of estimating log Kow by comparision with data reported in the literature, Veith and coworkers measured the retention time of 18 chemicals, and the standards and log Kow values were calculated from the regression equation (1). More recently Garst and Wilson (1984) have used the HPLC method to determine the partition coefficient for 66 structurally diverse chemicals ranging in log Kow from 0 to near 8. Veith's results indicated that log Kow can be estimated to within (22.8 4^20.0) 5 ------- CS-1420 (October, 1985) percent when compared with the values reported in the literature from measurements using other methods. The percent error was calculated assuming the literature value is the correct log Kow; these researchers had some reservations about this assumption. It should be noted that some of the greatest relative errors were observed with polar chemicals that dissociate in water (e.g., nr-chlorobenzoic acid, 2, 4, 5-trichloro-phenol, and diphenylamine). This method has a definite advantage, since the estimation of Kow can be made rapidly and relatively easily in comparison to the determination of Kow by the conventional method. Furthermore, Kow can be estimated for individual chemicals in complex mixtures (e.g., solid wastes) without knowing the specific chemical structure of each chemical. Other researchers have developed high-pressure liquid chromatographic methods to determine Kow (Mirrless et al. 1976; Carlson et al. 1975; Hulshoff and Perrin 1976; McCall 1975). However, these methods are based on a very limited number of experiments and considerably more work is needed to develop them. 3. Estimation from Water Solubility By definition, the partition coefficient expresses the equilibrium ratio of an organic chemical partitioned between an organic liquid (i.e., j^-octanol) and water. This partitioning is, in essence, equivalent to partitioning of the organic 6 ------- CS-1420 (October, 1985) chemical between itself and water. Thus, one would expect that a correlation might exist between the partition coefficient (KQW) and water solubility(s). Indeed, as shown by Mackay (1977), both of these properties are a function of the aqueous phase activity coefficient of the compound, and the correlation between the KQW and S is based on the ratio of the activity coefficients in water and octanol. Chiou et al. (1977) studied the relationship between K , o w and the water solubility, S, and found that, for 34 organic chemicals, an excellent linear correlation was observed between log KQW and log S that extended to more than eight orders of magnitude in water solubility (10~^ to 10^ ppm), and six orders of magnitude in Kow (10 to 107). From their data the following regression equation was derived: log KQW = 5.00 - 0.670 log S, r2 = 0.970 where S is expressed in ymol/L, and r2 is the coefficient of determination. To date a total of 18 different regression equations have been derived that correlate water solubility with the octanol water partition coefficient. This large number of equations results from the fact that solubility varies with the functional group of the molecule. Lyman et al. ( 1982) have summarized these 7 ------- CS-1420 (October, 1985) equations along with the class and number of chemicals that apply to each equation, the respective correlation coefficient, the applicable range of S and Kow values and the temperature at which the solubility data were obtained. Recent developments in the correlation between S and Kow indicate that many of the above equations can be combined into a single general relationship (Lyman, 1984). This would greatly simplify the estimation procedure, but introduces greater error, particularly for certain classes of compounds. This method has a definite advantage in that KQW can be estimated from an experimentally determined parameter - water solubility. Although this approach also has a cost advantage in that two parameters are determined for the cost of one, it must be kept in mind that the Kow is an estimated value subject to the limitations usually associated with approximations. Thus, while the estimation of Kow from water solubility is definitely a valid method, the values obtained cannot be construed as being equivalent to those obtained from the conventional shake-flask method, the HPLC method, or the generator column method. 4. Estimation Using the Fragment Constant Method Hansch and Leo (1979) have developed a method to estimate Kow from empirically derived atomic or group fragment constants 8 ------- CS-1420 (October, 1985) (f) and structural factors (F). Using these, the log Kow is calculated using the following equation: log Kow = Sum of fragments (f) + structural factors (F) Of course, the critical piece of information required to apply this method is the structure of the chemical in question—which is not always known. However, if the structure is known then there are a total of over 200 fragments (f and F) available that take into account such structural factors as molecular flexibility, unsaturation, multiple halogenation, branching, and h-polar fragment interactions. These and the method of calculation, with examples, are reviewed extensively by Lyman et al. ( 1982). In addition, a large collection (about 15,000) of both measured and calculated Kow values has been compiled by Hansch and Leo (1979); the method is also available for use on computer (Chou and Jurs 1979). Estimation by the fragment constant is probably the best initial step in determining Kow since it can be done without experimentation and is quite reliable for a large number of common organic chemicals. However, for some functional groups and complex or highly substituted molecules the fragment constant method can give erroneous, misleading results. 9 ------- CS-1420 (October, 1985) 5. Estimation Using Thin-Layer Chromatography It has been reported that thin-layer chromatography can be used to estimate Kow (Mirrless et al. 1976; Hulshoff and Perrin 1976). However, the generator column method is far superior to thin-layer chromatography (TLC) because of its accuracy, i.e., definition of the peak, reproducibility; ease of detection in many cases; and above all the range of applicability: the generator column method is applicable over 5 orders of magnitude of Kow while TLC is only applicable over 1.5 orders of magnitude of Kow (Mirrless et al. 1976). C. Rationale for the Test Conditions 1. Special Laboratory Equipment For the determination of the molar concentration of the test substance in water, several pieces of special laboratory equipment are required. This equipment is the same as that used in the generator column method for water solubility, CG-1510 (EPA, 1983) and includes: 1) a specially designed generator column; 2) a constant temperature bath; 3) a high pressure liquid chroniatograph with detector and integrator; 4) a specially designed extractor column; 5) two high pressure rotary switching valves; 6) specially designed collection vessels; and 7) a gas chromatograph with detector. In addition, particular types or brand names of column packing are specified. All of the special 10 ------- CS-1420 (October, 1985) laboratory equipment is same as that used by Tewari et al (1982) who developed and validated the test method. Thus, in order to maintain the integrity of the test method the procedure should be conducted using the special equipment as described. Any changes in or modification to the equipment should be reported in detail. 2. Temperature Control From the theory of the distribution law as outlined in the test support document for determining Kow by the conventional shake flask method, CG-1400 (EPA, 1983), the distribution coefficient Kow is a function of the temperature, and is a constant at a fixed temperature. Hence, in performing the octanol water partition coefficient measurements using the generator column, the temperature should be controlled. Controlling the temperature to ±0.05 °C is easy to carry out and is inexpensive. 3. Purity of Octanol, Water and Other Solvents Dissolved salts and other impurities can affect the solubility of a compound in octanol, water or any solvent. Thus, the test guideline specifies the use of purified _rv-octanol and reagent grade water. Trace amounts of impurities present in jv-octanol tend to produce emulsions and must be removed. The purified _rv-octanol may either be prepared according to the procedure given in the guideline or purchased from Fisher 11 ------- CS-1420 (October, 1985) Scientific. The water is prepared according to ASTM 01193-77: Standard Specification for Water. All other solvents used in the test method should be reagent or HPLC grade and contain no impurities that could interfere with the determination of the test compound. D. Test Data and Reporting 1. Test Report As a matter of standard laboratory practice, three determinations of the KQW for each test substance are required along with the mean and standard deviation. In addition, the weights of test substance and octanol used in the preparation of the test solution are required, as is the molar concentration of the test solution. Other ancillary data required for a complete assessment of the results include: test temperature, and if the HPLC method is employed, the method used to determine the sample loop volume and its average and standard deviation from three runs, and the response factor and its mean and standard deviation. For the GC method, data on the calibration procedure, the resulting regression equation, and results from the replicate GC runs should be reported in order that the results can be vai idated. 12 ------- CS-1420 (October, 1985) 2. Analytical Procedures In those cases where a separate analytical procedure is used to determine the concentration of the test substance in n- octanol a description of the analytical method should be given in the test report. This will enable the accuracy and precision of the gravimetric method for preparing the test substance in n-octanol to be assessed. If any changes are made or problems encountered in the test method, they should be reported so that any necessary revisions can be incorporated in the test method. 13 ------- CS-1420 (October, 1985) III. REFERENCES ASTM. 1978. Annual Book of ASTM Standards, American Society for Testing and Materials, Philadelphia, PA., Part 31, Method D 1193. Carlson RM, Carlson RE, Kopperman HL. 1975. Determination of partition coefficients by liquid chromatography. J Chromatogr 107:219. Chou JT, Jurs PC. 1979. Computer assisted computation of partition coefficients from molecular structure using fragment constants. J Chem Inf Comput Sci 19:172. Chiou CT, Freed VH, Schmedding DW, coefficient and bioaccumulation of Environ Sci Technol 11:475. Kohnert RL. 1977. Partition selected organic chemicals. EPA. 1983. U.S. Environmental Protection Agency. Office of Toxic Substances. Chemical Fate Test Guidelines. PB 83-257717. Fujita T, Iwasa J, Hansch C. 1964. A new substituent constant, derived from partition coefficients. J An Chem Soc 86:5175. Garst JE, Wilson WC. 1985. An Accurate, Wide Range, Automated HPLC Method for Determination of Octanol:Water Partition Coefficients. To be published. Hansch C, Anderson SM. 1967. The effect of intramolecular hydrophobic bonding on partition coefficients. J Org Chem 23:2583. Hansch C, Leo A. 1979. Substituent constants for correlation analysis in chemistry and biology. New York: J. Wiley & Sons. Hulshoff A, Perrin JH. 1976. A comparison of the determination of partition coefficients of 1,4-benzodiazepines by high- performance liquid chromatography and thin-layer chromatography. J Chromatogr 129:263. Leo A, Hansch C, Elkins D. 1971. Partition coefficients and their uses. Chem Rev 71:525. Lyman WJ. 1984. Personal communication. Lyman WJ, Reehl WF, Rosenblatt DH. 1982. Handbook of Chemical Property Estimation Methods. Environmental Behavior of Organic Compounds. McGraw Hill Book Company. New York. 14 ------- CS-1420 (October, 1985) Mackay D. 1977. Environ Sci Technol 11:1219. McCall JM. 1975. Liquid-liquid partition coefficients by high- pressure liquid chromatography. J Med Chem 18:549. Mirrless MS, Moulton SJ, Murphy CT, Taylor PJ. 1976. Direct measurement of octanol-water partition coefficients by high- pressure liquid chromatography. J Med Chem 19:615. OECD. 1981. Organization for Economic Cooperation and Development (OECD). Guidelines for Testing Chemicals: No. 107- Partition Coefficient (jv-Octanol/Water). Director of Information, OECD; 2 Rue Andre-Pascal, 75775 PARIS CEDEX 16, France. Veith GD, Austin NM, Morris RT. (1979) A rapid method for estimating log P for organic chemicals. Water Res 13:43. Wasik SP, Tewari YB, Miller MM, Martire DE. 1981. Octanol/Water partition coefficient and aqueous solubilities of organic compounds. National Bureau of Standard (NBS), U.S. Department of Commerce, Washington, DC. NBS Report NBSIR 81-2406. Tewari YB, Miller MM, Wasik SP, Martire DE. 1982. Aqueous Solubility and Octanol/Water Partition Coefficient of Organic Compounds at 25.0° C. J Chem Eng Data 27:451-454. 15 ------- |