Development of a qsar model to predict soil organic carbon-water partition coefficients of polycyclic aromatic hydrocarbons / Renu Jailuk = การพัฒนาคิวซาร์โมเดลเพื่อทำนายค่าสัมประสิทธิ์การดูดซับ สารโพลีไซคลิกอะโรมาติกไฮโดรคาร์บอนของอินทรีย์คาร์บอนในดิน / เรณู ใจหลัก
The study focussed on the sorption of 9 PAH (NAP, ACE, FLU, ANT, PHE, FLA, PYR, BaA, and BaP) which was used as the representative for the PAH with log K[subscript ow] ranged from 3.4 to 6.5. the sorption of PAH was undertaken on four soil organic carbon contents over four concentrations of the compounds. The objectives were to evaluate the partitioning behavior of PAH between the soil and the water, to determine the relationships between log K[subscript oc] and log K[subscript ow], MW, log S, log VP, and log H, and to validate the developed models. The soil organic carbon-water partition coefficients (log K[subscript oc]) values of PAH obtained agree well with other studies. The values increased with increasing molecular weight and log K[subscript ow] of the compounds. Deviation of log K[subscript oc] values could be the influence of experimental condition, temperature, soil particle size, suspended or colloidal organic matter, biodegradation, and photodegradation. Good relationships of log K[subscript oc] which those of log K[subscript ow], MW, and log H were obtained which indicated these properties as good indicators for predicting the log K[subscript oc] values. This was evident by the values of log K[subscript oc] of other PAH with log K[subscript ow] ranged from 3.4 to 6.5 derived from the models were close to those in the literatures, especially, the log K[subscript oc] values derived from log K[subscript ow]. The application of the models with similar physicochemical property, i.e., PCDD, PCB is not good as for PAH compounds except the model derived from log K[subscript ow]. Therefore, the application of the models to these groups may be limited. Other models developed from the relationships of log K[subscript oc] with S, VP may not be appropriate since their correlation coefficients were not justified. This was probably due to variation in the S and VP values.