The transit of chemical elements within the different parts of Orleans valley’s aquifer is studied by two complementary methods. Those methods rely on the fractionation of lanthanides (Ln) during their migration in natural waters. The first method consists in studying natural lanthanides patterns within the watershed, at its entries and exits. The second one lies on multi-tracer experiments with Ln-EDTA complexes. This work is completed through an observation network consisting of 52 piezometers set on a sand and gravel quarry, and the natural entries and exits of the aquifer.
Orleans valley’s aquifer, which is made of an alluvial watershed lying on a karstic aquifer, is mainly fed by the Loire river via a large karstic network. At the entries of the aquifer (Loire river at Jargeau), the Ln concentrations in the dissolved fraction (< 0,22 μm) vary with the flow of the river. During floods, Loire river waters display bulk continental crust-like Ln compositions with a slight enrichment in heavy Ln from Dy to Lu. When the Loire river flow becomes low level, the crust-normalised Ln patterns show a depletion in light Ln whereas Lu concentrations remain identical. The same evolution spatially occurs between the entries and exits of the karstic network. Spring waters are depleted in light Ln relative to the Loire river whereas heavy Ln (Yb, Lu) remain constant during transit. Furthermore, the depletion in light Ln increases with the distance between entries and exits.
Tracer experiments using EDTA-complexed Ln within and between the alluvial and calcareous parts of the watershed have shown that complexed Ln are fractionated across all these geological strata. The recoveries of tracers always follow the order light Ln < heavy Ln. Moreover, both sediments analyses and filtering experiments at a porosity of 0,02 μm show that, in the presence of EDTA, Ln adsorb onto sediments and colloids in the order light Ln > heavy Ln. On the other hand, the filtration of alluvial groundwater with high colloids content induces no significant Ln fractionation when the solution contains no strong chelating agent.
Hence, the transit of natural and artificial Ln in Orleans valley aquifer can be explained by two complementary processes. (1) Decanting/filtering or, on the opposite, stirring of colloids. Those processes induce no important Ln fractionation. (2) Exchanges of Ln between solute complexes, colloids and sediments due to the presence of strong chelating agents. Those exchanges fractionate the Ln in the order of their stability constants. Considering the natural Ln fractionation that occurs in the Loire river and in the studied aquifer, the carbonates, the stability constants of which follow the order light Ln < heavy Ln, are the best candidates as natural strong chelating agents.
From the hydrodynamic point of view, both tracer experiments and natural Ln concentrations show that the transfer of elements within the alluvial watershed is pulsed by the Loire river movements. During an ascent phase, the elements migrate away from and perpendicularly to the karstic channels direction. During the river descent, horizontal flows are quasi absent and migrations are mainly vertical from the alluvia down to the calcareous part of the aquifer. Due to those hydrodynamic characteristics, alluvia and non fissured limestone have a high dynamic confining capacity. Elements with high affinity for solid or colloidal phases (e.g. light Ln) have an increased confining capacity in the whole aquifer, by sorption and colloid filtration within the alluvia and at the alluvial-calcareous interface, and by colloid decanting within the karstic channels. Overall, this model combines two components. The first one, hydrodynamical, results from the repartition of the loads pulsed by river Loire through the karst. The second one physico-chemical, results from the element distribution mainly controlled by colloïde/solute complexes exchange coefficients.