It is well known that aquifer structural properties and the resulting heterogeneous distribution of hydraulic conductivity and porosity significantly control groundwater flow and spreading of solutes. In addition to this, physico-chemical aquifer heterogeneity, i.e. different intra-particle sorption and diffusion properties for different source rocks of the aquifer material (lithological components) grouped in different grain size fractions, influence the interaction of reactive solutes with the aquifer material. To be able to consider both types of heterogeneity, a new 3D finite-difference reactive solute transport modeling approach was developed, being an essential component of a methodology allowing for the upscaling of small-scale laboratory measurements and for the assessment of parameter uncertainty. Sorption and desorption are introduced at grain scale through the simulation of a retarded intra-particle diffusion process in the heterogeneous aquifer material for each lithological component and each grain size fraction in every model cell. For a practical application of the code the data needed may be introduced into each model cell following a facies-based geostatistical approach. First modeling results emphasize the strong impact of the lithological aquifer material composition and confirm the need for a geostatistical process-based reactive transport modeling approach with spatially variable hydraulic and hydrogeochemical aquifer parameters.