Technical report 123

APPENDIX B: MEASUREMENT OF PARTITIONING (KOW)

OECD 107 – Partition coefficient (n-octanol/water): Shake flask

This method is appropriate for the determination of the partition coefficient of substances with log KOW values in the range of -2 to 4 (OECD, 1995a). Measurements of the partition coefficient should be made on ionisable substances only in their non-ionised (neutral) state and this can be achieved by the use of an appropriate buffer with a pH of at least two units below (free acid) or above (free base) the pKa.

In order to determine a partition coefficient duplicate samples of water, n-octanol and test compound are equilibrated with each other via mechanical shaking. The two solvent phases are then separated by centrifugation and the concentrations of the test compound in the two phases, aqueous and octanol, are determined using an appropriate analytical technique. In total there should be three duplicate runs with different solvent ratios, and the six KOW values should be within a factor of two of each other.

Briefly, this method was developed for measuring KOW of neutral compounds and is of limited value when dealing with ionisable or multi-protic compounds.

OECD 107 – Partition coefficient (n-octanol/water): Shake flask at pH 5, 7 and 9

The shake flask method conducted @ pH 5, 7 and 9 is a variation of the OECD 107 method and takes account of the partitioning behaviour of ionisable compounds over the environmentally relevant pH range (pH 5-9). Typically, three pH values (pH 5, 7 and 9) are employed. The principle of the test is the same except that three flasks, each buffered at a specific pH, are employed rather than a single flask buffered to ensure that only the neutral form of the molecule is present. Depending on the pKa of the compound, changes in pH will alter the proportion of non-ionised and ionised species distributed between the two different solvent phases. The reported value from this method is DOW, and should always be associated with the relevant pH value at which it was measured. Unfortunately, much of the scientific literature has conflated the terms KOW and DOW and it is not unusual for the distribution coefficient to be expressed as KOW at pH 7 rather than log DOW at pH 7. This can lead to potential confusion, since KOW should strictly refer to the neutral species, which may or may not exist at pH 7. Importantly, DOW varies with pH in accordance with the relative extent of speciation whereas KOW is a constant value based upon the neutral undissociated form of the compound. The value for the dissociated molecule determined around a pH of 7 is considered more realistic for PBT and chemical safety assessment (ECHA, 2008b).

OECD 117 – Partition coefficient (n-octanol/water): High performance liquid chromatography (HPLC)

This is a relatively quick way of estimating KOW. It is not measured directly, but from a correlation between log k (capacity factor) and a series of reference chemicals with known log KOW values (OECD, 2004). It therefore depends on the quality of the log KOW measurement of the reference chemicals (often measured by the shake flask method). A series of reference compounds with similar chemical functionality to the test material should be used to generate the correlation between log k and log KOW. The test material is injected onto a reverse phase column and partitions between the mobile solvent phase and the hydrocarbon stationary phase as it is transported along the column by the mobile phase. The chemicals are retained in proportion to their hydrocarbon-water partition coefficient, with hydrophilic chemicals eluted first and hydrophobic chemicals last. The retention time is used to derive log K, which is then used to calculate log KOW of the test substance, based on linear regression analysis of log K with the log KOW values of the reference substances.

In general, the HPLC method is less sensitive to impurities than the shake flask method. The HPLC method is also very suitable for measuring the KOW of mixtures of chemical homologues. However, charged molecules have more complex retention behaviour than the neutral species, and therefore this method is not recommended as being suitable for determining the KOW of ionisable compounds.

OECD 122 – Partition coefficient (n-octanol/water): pH-Metric for ionisable substances

This is the method of choice for determining the partitioning behaviour of multiprotic or ionisable substances with log KOW values between -2 to 7 (OECD, 2000a). A derived estimation of log KOW is based on the apparent shift in pKa when n-octanol is added to a solution of the test substance in water. The pH-metric technique consists of two linked potentiometric titrations. The pKa of the test substance is determined by acid-base titration of an aqueous solution of the substance using potentiometric measurement of pH. To measure the log KOW, one or more additional titrations are done. In these additional titrations, n-octanol is added to a solution of the substance in water and the apparent pKa of the substance in the two-phase system is measured. This apparent pKa value is denoted by the term poKa. The pKa values derived from the aqueous and the two-phase titration curves are different and this difference is related to the value of log KOW. Using established equations, an estimation of the partition coefficient is obtained. This method derives values for pKa, log KOW and log DOW across the ionisable pH range and therefore represents the most comprehensive data outputs of all methods reviewed here and has been recommended by the EMA ERA guideline Q+A technical review (EMA, 2010). For ionisable solutes, a major advantage of the method over the shake-flask procedure is that it affords a distribution profile (i.e., a pH-log DOW curve) rather than single points. However, the method requires sophisticated analytical technology and there appears to be limited availability at Contract Research Organisations. The method also struggles with insoluble compounds and self-evidently is not suitable for neutral compounds.

OECD 123 – Partition coefficient (n-octanol/water): Slow-stirring

The ‘shake-flask’ method (OECD 107) is prone to artefacts due to transfer of octanol micro droplets into the aqueous phase. With increasing values of KOW the presence of these droplets in the aqueous phase leads to an increasing overestimation of the concentration of the test substance in the water. Therefore, the use of the shake flask method is limited to substances with log KOW < 4. The experimental difficulties associated with the formation of micro droplets can be overcome in the slow-stirring method (OECD, 2006). In this method, water, octanol and the test compound are equilibrated in a stirred reactor. Exchange between the phases is accelerated by stirring. In that manner turbulence is introduced and the exchange between octanol and water is enhanced without micro droplets being formed. The time to equilibrium, however, will vary depending on the hydrophobicity of the test substances. For very hydrophobic substance the time to equilibrium may take several days. Once equilibrium is reached the KOW is determined directly as in the shake flask method. In comparison with the shake flask method, this method may also be adapted to investigate the DOW of ionisable substances at pH 5, 7 and 9.

OECD 107/EU A.8 – Estimation of log KOW from saturated solutions

Due to the low solubility and the multi-protic nature of some substances, conventional methods for determining KOW, such as the shake flask procedure or the HPLC screen, are not technically feasible or considered appropriate. For substances with KOW values outside the range of applicability, test methods that utilise the saturation solubility values in octanol and water, which are determined by visual inspection and which may be supported by turbidity measurements, may be appropriate. In these instances, KOW can be estimated based on the ratio of the solubility of the chemical in octanol and water at saturation.

This estimation method does have a serious drawback in that KOW is not based on an interaction between water and solvent (octanol) and as a consequence the correlation between octanol solubility and KOW may be perceived as insufficient. In the event of insurmountable technical challenges this is an acceptable alternative to the shake flask method, but it is recommended that data obtained using the solubility approach be treated with caution.