1. William P. Eckel, Ph.D. Office of Pesticide Programs (OPP) US EPA
QSAR and Calculators for Freundlich Adsorption Coefficient (Kf) Based on 18 Agricultural Soils American Chemical Society 254th National Meeting, August 24, 2017 Washington, DC
William P. Eckel, Ph.D.
Office of Pesticide Programs (OPP)
US EPA

2. Question: Is data in OPP’s Possession useful for new Quantitative Structure-Activity Relationships?
OPP has hundreds of guideline studies for mobility and metabolism of pesticide active ingredients
Could new QSARs complement methods already available to OPPT?
EPISuite methods based on logKow and Molecular Connectivity Index
Soil mobility (batch equilibrium guideline) as a pilot

3. Batch Equilibrium Study (835.1230 or 163-1)
Soil - water partitioning at several concentrations
Several agricultural soils required per chemical
FIFRA requirement for all conventional pesticides
Regression of results yields Freundlich coefficient (Kf) and exponent (1/n) for each soil

4. Question: Is Kf a Function of the Soil Used?
Same soils are used repeatedly in submitted studies
Is there a relationship between a property of the chemical, and its Kf on a particular soil?
Regressed Kf versus “sub-cooled liquid” solubility
Sub-cooled liquid solubility (S_scl) calculated by “fugacity ratio” method; requires melting point and solubility of solid

5. Fugacity Ratio for Subcooled Liquid solubility
ln(Fugacity ratio) = (ΔHfus/R)*((mp/298)– 1)
Melting point of solid in Kelvins
ΔHfus/R (or Sfus) estimated as J/mol-K
Subcooled liquid solubility (S-scl) = Solubility(solid)/fugacity ratio
Interpreted as Solubility if substance were a liquid at 25°C

6. Data Mining of Data Evaluation Records (DERs)
18 soils in recent OPP electronic files had at least 3 chemicals with Batch Equilibrium study DERs
41 chemicals total over the 18 soils (US and European)
Soils were identified by name (e.g., Don Uglem clay loam)
Soil identity verified by sand/silt/clay content and %organic carbon
Ranged from Sand (0.38 %OC) to Humic (19% OC)
Log-log plot (Kf vs. solubility) and power function regression

7. Summary of Soils and Chemical Data Used

8. Data for Speyer 2.2 Loamy Sand with Kf estimates vs. Measured Kf

9. Log-Log Plot of Kf vs. S_scl (r2 = 0.75)

10. Example of Subcooled Liquid Solubility Calculator: Atrazine. Avg
Example of Subcooled Liquid Solubility Calculator: Atrazine. Avg. Kfoc = 174; range =

11. Test of Subcooled Liquid Calculator
333 pesticide active ingredients of all classes from Footprint (EU) database
Each had reported Kfoc range/average, melting point and solubility
Calculated S_scl for all 333, then calculated average Kf and Kfoc
Compared Average Predicted to Average Reported
Prediction within 2x, 3x, 10x ?

12. Results of Footprint Database Analysis
Predicted/Reported Kfoc:
Ratio 0.1 – 10: (83%)
Ratio 0.33 – 3: (65%)
Ratio 0.5 – 2: (43%)

16. Effect of Ionized Chemicals (pKa reported)

17. Correlation of 1/n to mp (Laacherhof Axxa soil)

18. Initial Analysis of 1/n versus melting point

19. Prediction on Molar solubility basis

20. Individual classes: Carbamates

21. Individual classes: Conazoles

22. Individual Classes: Strobilurins

23. Individual Classes: Urea herbicides

24. Conclusions
Yes, OPP’s Batch Equilibrium data holdings are useful for constructing QSARs
Sub-cooled liquid Solubility is a useful independent variable to predict Kf for both neutral and ionized chemicals
Most predictions of Kfoc are within 10x; almost half within 2x
Class-specific regressions might improve predictions of Kf
May be possible to estimate exponent, 1/n