A Freundlich-based model for prediction of pH-dependent sulfate adsorption in forest soil.

Sammanfattning: The period of industrialization after the second World War in Europe released SO2 and NOx by combustion of fossil fuels and contributed the formation of S and N compounds in the forest ecosystem. The Swedish forest soil systems were influenced by emissions of SO2 followed by H2SO4 deposition, consequently the pool of SO42- had increased in the forest ecosystem. This thesis studied SO42- adsorption in a podzolic Bs horizon soils taken from a Swedish forest soil system. The soil samples from five different sampling sites were collected and the results revealed different amounts of adsorbed SO42- in response to changes in equilibrium concentration and pH. This study found that the amount of adsorbed SO42- (mmol/kg) increased with an added equilibrium concentration of SO42- (mmol/l) and with a decreasing pH. This was determined by equilibration experiments. Based on the results a Freundlich-based model was developed to predict the pool of adsorbed SO42- in the soil samples. The model predicted the pool of adsorbed SO42- (mmol/kg) as a function of pH and the equilibrium concentration of SO42- (mmol/l) in the soil solution system. The extended Freundlich model was optimized in three different ways: by use of unconstrained, constrained and simplified two-point calibration. The results showed that the adsorption of sulfate in the Kloten Bs1 and Risbergshöjden B soils was higher as compared to the Tärnsjo B, Österström B, and Risfallet B soils. The coefficient of determination (R2) determined from an unconstrained fit of the extended Freundlich model (with three adjustable parameters) for Risbergshöjden B and Kloten Bs1 were R2 =0.998 and R2=0.993. Nearly as good fits were found in a constrained fit with two adjustable parameters when it was assumed that nearly 2 protons (2 H+) are co-adsorbed with one SO42- ion (Risbergshöjden B; R2=0.997 and Kloten Bs; R2=0.992). The simplified two-point calibration with two adjustable parameters showed similar parameter values for all most soils and was considered the best optimization method of extended Freundlich model, especially as it requires only limited input data.