Modelling multicomponent solute transport in structured soils

The mobility of contaminants in soil is an important factor in determining their ability to spread into the wider environment. For non-volatile substances, transport within the soil is generally dominated by transport of dissolved fractions in the soil water phase, via either diffusion or convection. During this process the mobility of reactive ions is strongly affected by adsorption. Adsorption processes regulate the distribution of ions over an immobile fraction that is attached to soil particles and a mobile fraction that is present in the soil water phase, and therefore determine which fraction of ions is available for transport via the water phase.In structured soil the conditions are more complex as water flow is not homogeneous. The flow of water is fast through cracks and macropores in between soil peds and aggregates, but at the same time a significant part of the soil water can be stagnant inside the smaller pores inside peds and aggregates. In order to reach adsorbing particles inside the peds and aggregates, ions have to diffuse through the smaller pores inside the aggregates or soil peds. Depending on the size of the aggregates, this can cause a significant delay in sorption and a marked different behaviour from the instantaneous equilibrium assumption that is usually adhered to for homogeneous systems. The effect this has on ion transport is referred to as physical nonequilibrium.Adsorption of ions onto charged surfaces in soil is a complex process. Soils are multicomponent systems in which the behaviour of contaminant ions is influenced by other ions, for example through competitive sorption. Recent developments in surface complexation models start to become increasingly successful in accounting for competitive effects during sorption.This thesis addresses the combination of multicomponent chemistry, diffusion-limited sorption and transport. Transport and sorption processes were studied in flow experiments with well-defined artificial model systems and by computer modelling. The modelling was stepwise validated against experimental results. An artificial flow system was developed in the first part of this thesis by gradually increasing the complexity of the system from non-reactive transport to multicomponent reactive transport. The model progressed accordingly by including the relevant sorption models into the transport model. The model simulations were then used to evaluate the effects of multicomponent chemistry on transport.The initial flow system was prepared from a column filled with spherical gel beads made from alginate gel. The gel acts as a stagnant water phase, in which ions can only move by diffusion. Transport through the column with gel beads was modelled with a two-region transport model. An immobile region represented the solution inside the beads and an immobile region represented the flowing solution in between the beads. Ion transport was modelled by convection in the mobile region and by diffusion in the immobile region. First, diffusion of nitrate and phosphate out of the gel beads was measured in a batch system. The model predicted the release of ions from the gel beads accurately using reported diffusion coefficients for diffusion of the ions in water. Leaching of nitrate through a bead filled column was measured at different flow rates. The concentration in the effluent was plotted against time to give breakthrough curves. The breakthrough curves predicted by the model matched the measurements accurately.The next experiments evaluated proton transport through a column with alginate gel beads. Proton sorption on alginate occurs by ion-exchange with the calcium ions that cross-link the alginate polymer structure. The sorption parameters were measured separately in acid-base titrations of the gel at different calcium levels. The ion-exchange model was incorporated in the two-region transport model. The combined model described the proton leaching curves at different calcium concentrations very well. The results showed the combined effects of nonlinear sorption and competition, which added to the effects of physical nonequilibrium.The third set of experiments considered sulphate and chromate sorption and transport in a column with goethite. Column experiments were performed with goethite embedded in polyacrylamide gel beads. The two-region transport model was combined with a surface complexation model to predict reactive transport in the goethite-gel bead system. Chromate and sulphate breakthrough curves were measured in a set of transport experiments, along with corresponding changes in the effluent pH. Model parameters for the surface complexation model were obtained from literature data and sorption measurements. The model predicted the breakthrough curves well for transport of chromate and sulphate in separate column experiments. However, the model overestimated the pH changes in the effluent, possibly because of proton buffering by the polyacrylamide gel. The effect of competitive sorption on transport was examined in experiments with both anions present. The model predicted the effect of competition very well in a system initially equilibrated with sulphate, followed by infiltration with chromate. However, when sulphate was infiltrated after equilibration with chromate, chromate desorption and sulphate adsorption were clearly underestimated by the transport model. The exchange between the more strongly bound chromate and the sulphate added subsequently may be too slow to cause a substantial chromate peak in the effluent.The second part of the thesis applied multicomponent chemistry and transport modelling to strontium sorption in microscopic aggregates of hydrous ferric oxide (HFO). Previous studies found that it takes days to months to reach equilibrium sorption probably due to slow diffusion into the porous particles. A novel mechanistic approach was suggested to explain slow diffusion into the small aggregates. A model was developed that relates diffusion to the electrostatic potential inside the aggregate pores. At pH values below 8, the surfaces are positively charged which causes an electrostatic potential that stretches across the whole pore diameter. Positively charged ions such as strontium are repulsed and remain at very low concentrations in the pore space. Here a Donnan electrostatic model was used to calculate the electrostatic potential inside the pores based on the assumption that the potential gradients in the small pores are overlapping. The electrostatic model is directly linked to the surface complexation model (CD-MUSIC) that describes the surface charge dependent on protonation of the surface groups and adsorption of counter ions and strontium. Strontium sorption is pH-dependent and in its turn influences the local pH inside the aggregates by causing desorption of protons from the oxide surface. Therefore the model accounts for sorption and diffusion of both ions. Diffusion is calculated from the local concentrations and electrostatic gradient in the pores using the Nernst-Planck diffusion equation. The diffusion flux of strontium is small at the low concentrations in the pores. The time it takes to reach equilibrium is strongly dependent on pH which determines the amount of strontium sorption, but above all, the mineral charge and the repulsion from the pores. The Donnan-diffusion model was compared with non-electrostatic pore diffusion, which does not take electrostatic interactions into account. The Donnan model predicts very low concentrations of strontium in the pores and diffusion rates that are up to 8000 times lower than predicted with a non-electrostatic model.The Donnan-diffusion mode] was validated against strontium transport in a column of hydrous ferric oxide. Microscopic aggregates of hydrous ferric oxide (230 nm) were prepared by freezing and thawing ferrihydrite. The resulting aggregates have pores with a number-average size of 2 nm and a size distribution ranging from 1 to 12 nm. The surface complexation parameters were assessed by acid-base titrations and strontium sorption experiments. Strontium transport in the columns was measured at pH 4 and 7. The Sr breakthrough curves showed that sorption was virtually instantaneous on part of the hydrous ferric oxide, bistantaneous sorption was explained by unrestrained diffusion in the largest pore fraction of the aggregate pores. The accessible pore fraction was determined by salt pulses and was found to be dependent on pH. Taking the fast accessible pore fraction into account in the transport simulations, the model matched the experimental breakthrough curves well at both pH values.This thesis demonstrates that the combination of multicomponent chemistry, diffusion-limited sorption and transport can be modelled by combining mechanistic process descriptions. There were often more influencing factors than initially anticipated, even in relatively simple systems. Nevertheless, mechanistic modelling provides valuable insights into the processes that are involved in multicomponent transport in soil systems.