Upper soil horizons control the rare earth element patterns in shallow groundwater

Geoderma 239–240 (2015) 84–96

Contents lists available at ScienceDirect

Geoderma journal homepage: www.elsevier.com/locate/geoderma

Upper soil horizons control the rare earth element patterns in shallow groundwater Mathieu Pédrot ⁎, Aline Dia, Mélanie Davranche, Gérard Gruau CNRS-UMR 6118, Géosciences Rennes, University of Rennes 1, Avenue du Général Leclerc, 35042 Rennes Cedex, France

a r t i c l e

i n f o

Article history: Received 9 April 2014 Received in revised form 2 September 2014 Accepted 27 September 2014 Available online xxxx Keywords: Soil chemistry Groundwater Rare earth elements Ce anomaly Fe oxyhydroxides Organic carbon

a b s t r a c t Several studies dedicated to the aquatic geochemistry of rare earth elements (REEs) have displayed a wide topography-related spatial variability in the REE signatures of shallow groundwater. The aim of this study was to understand the processes leading to this specific REE signature, notably with regard to the size of the Ce anomaly. Soils were sampled in order to encompass the expected topographic variability in the organic carbon (OC)/ Fe(Mn) ratio. Leaching experiments were performed with the uppermost horizons of the soil. The REE patterns that developed in the soil leaching solution were similar to the REE patterns for the shallow groundwater collected in this catchment. The negative Ce anomaly evolves in a similar manner according to the topography. This spatial variation is strongly correlated with the soil OC/Fe ratio. For a low OC/Fe ratio, the negative Ce anomaly amplitude in the soil solution is large, whereas a high OC/Fe ratio generates a small or insignificant Ce anomaly. Reductive dissolution experiments using soil with low OC/Fe ratios demonstrated that the REE pattern for soil Fe oxyhydroxides exhibited a positive Ce anomaly and HREE enrichment, indicating a preferential association of these elements with Fe-oxyhydroxides. The rare earth element signature observed in the shallow groundwater is affected by Fe oxyhydroxides present in the upper soil horizons. In contrast, in soil with a high OC/Fe ratio, the REE pattern obtained under reducing conditions did not exhibit any Ce anomaly, suggesting that in the bottomland, the REE signature is affected by the OC content in the uppermost soil. This study highlights the impact of organic matter on the Fe pedofeatures, which control the development of a negative Ce anomaly in shallow groundwater. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Numerous studies have been dedicated to understanding the geochemistry of rare earth elements (REEs) in groundwater over the last three decades (Smedley, 1991; Viers et al., 1997; Johannesson et al., 1997, 1999; Johannesson and Hendry, 2000; Braun et al., 1998; Dia et al., 2000; Duncan and Shaw, 2003; Tang and Johannesson, 2005; Pourret et al., 2010). Rare earth elements have coherent physical and chemical properties which make them sensitive tracers of water–rock interactions and groundwater mixing. Nowadays, several important aspects of the REE geochemistry in groundwater are fairly well understood, including the role of pH with regard to the REE concentrations, the role of organic ligands in REE speciation and REE transfer, and the processes involved in the development of negative Ce anomalies (Smedley, 1991; Braun et al., 1998; Dia et al., 2000; Pourret et al., 2010). Many groundwater areas exhibit a negative Ce anomaly which has been demonstrated to be a consequence of Ce removal in response to the oxidation of Ce(III) into Ce(IV). One of the processes by which Ce(III) is oxidized into Ce(IV) and removed from the groundwater is the so-called “oxidative scavenging” by Fe and Mn oxides (e.g. ⁎ Corresponding author. E-mail address: [email protected] (M. Pédrot).

http://dx.doi.org/10.1016/j.geoderma.2014.09.023 0016-7061/© 2014 Elsevier B.V. All rights reserved.

Koeppenkastrop and De Carlo, 1992; De Carlo et al., 1998; Bau, 1999; Ohta and Kawabe, 2001; Davranche et al., 2005). The oxidative scavenging mechanism may be regarded as a three-stage process consisting of (i) an initial sorption of REE, including Ce(III), onto Mn and Fe oxyhydroxides, (ii) an oxidation of part of the adsorbed Ce(III) into Ce(IV), and finally (iii) a preferential desorption of the remaining Ce(III) and all other REE(III) as compared to Ce(IV) (Bau, 1999). Cerium oxidation can occur abiotically through the oxidative scavenging of dissolved Ce(III), as well as biotically, where the bacteria directly oxidize Ce(III) into Ce(IV), or catalyze the oxidation of Mn(II) into Mn(IV), which itself abiotically oxidizes Ce(III) (Moffett, 1990; Ohnuki et al., 2008; Tanaka et al., 2010). However, one essential feature of the REE geochemistry in groundwater that remains partly unexplained is the origin and significance of the systematic, topography-related variation in the Ce anomaly that occurs worldwide in many shallow groundwater areas (e.g. Braun et al., 1998; Gruau et al., 2004; Pourret et al., 2010). In these shallow groundwater areas, the same progressive disappearance of the negative Ce anomaly is observed from the top to the bottom of the catchment. The lack of a Ce anomaly in bottomland groundwater is likely to be an effect of the input of soil-derived groundwater in which REE are bound to organic colloids that do not display a Ce anomaly (Viers et al., 1997; Dia et al., 2000; Pourret et al., 2007). The amount of organic colloids in

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

85

Kerroland transect Gunolay transect

Fig. 1. Sketch showing the topography, channel network geometry, location of the Kervidy/Coët-Dan catchment.

shallow groundwater increases downhill due to the decrease in the distance between the water table and the uppermost organic-rich soil horizons (Pourret et al., 2010). The binding of REE by the organic colloids in the soil solution suppresses the oxidative-scavenging of Ce by Fe and Mn oxyhydroxides (Davranche et al., 2004, 2005, 2008). The distance between the soil organic-rich horizons and the water table downhill and the increased flux of organic colloids entering into the groundwater are therefore the major controls of the insignificant amount or the lack of a negative Ce anomaly (Pourret et al., 2010). A question then arises: do all of the topographic gradients observed in the Ce anomaly development have a shallow origin? More specifically, is the origin of the negative Ce anomaly found in uphill groundwater coming from the upper soil horizons? These questions are worth examining to the extent that the composition of the soils evolves with the topography. Soils in bottomland domains are expected to be Fe- and Mndepleted but enriched in organic matter (OM) as compared to their uphill equivalents (Walter and Curmi, 1998). This results from differences in moisture and the redox conditions in bottomland soils, as compared to the well-drained and permanently oxidized uphill soils (Riou, 1999). Therefore, the organic carbon (OC)/Fe(Mn) ratio in soils decreases from uphill to the bottomland. Thus, the major source of dissolved REE in the uphill soil solution is probably mineral surfaces, leading to the possible development of Ce oxidative scavenging. Should this be the case, soil water in uphill domains may exhibit a negative Ce anomaly, the amplitude of which may be closely linked to the soil OC/Fe(Mn) ratio. If so, the topography-related variation of Ce anomaly observed in shallow groundwater could have a surficial origin. In this study, we explored the possibility of a Ce anomaly variability with a fully surficial origin by conducting percolation experiments and reductive dissolution experiments on soils relative to a toposequence (Kervidy/Coët-Dan catchment, France). This catchment provides a typical example of shallow groundwater that exhibits the characteristic Ce anomaly topographic variations. Soils were sampled in order to encompass the expected topographic variability in the OC/Fe(Mn) ratio. The aims of this study were to (i) experimentally investigate whether or not the variations in the upper soil horizon composition along the toposequence, and notably in the OC/Fe(Mn) ratio, are likely to generate

variations in the Ce anomaly development in the groundwater recharge waters, (ii) discern the key factors controlling the development of a Ce anomaly in soil solutions, and (iii) identify the source of the REE signature variability in shallow groundwater.

2. Materials and methods 2.1. Site description Soil samples were collected in the Kervidy/Coët-Dan catchment (north-western France) from the Kerroland and Gunolay transects in January 2008 (Fig. 1). The region is marked by a humid temperate climate with a mean annual rainfall of 909 mm. The Kervidy/Coët-Dan is a part of the regional observatory (ORE) AgrHyS dedicated to the study of the effects of intensive agriculture on water quality (Molénat et al., 2008 and references therein). The bedrock is constituted of fissured and fractured upper Proterozoic schists (Dabard et al., 1996). Detailed information about soils and prevailing hydrogeochemistry can be found elsewhere (Mérot et al., 1995; Durand and Juan Torres, 1996; Curmi et al., 1998; Dia et al., 2000; Molénat et al., 2002; Pourret et al., 2010).

2.1.1. Soil samples The soils are developed from loamy material and eolian Quaternary deposits. They are locally dominated by silt, clay or sandstone materials (Pellerin and Van Vliet-Lanoë, 1998). A large number of secondary mineral phases including illite, smectite, kaolinite, various Fe oxides and Fe oxyhydroxides and Mn oxides can be observed in the soil horizons (Pauwels et al., 1998). With regard to the topography, soils are welldrained Dystric Cambisols and Luvisols (Dystrochrepts and Alfisols respectively, USDA Soil Taxonomy) from the upland to the bottomland areas. Epistagnic Luvisols and Epistagnic Albeluvisols (Aqualfs, USDA Soil Taxonomy) are also developed in the poorly-drained domains, in which Mn and Fe-oxyhydroxides are depleted due to seasonal waterlogging by the rising groundwater.

86

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

Table 1 Composition and location of the soil samples along the Kerroland and Gunolay transects. LREE/HREE ratio is ΣLREEUCC/ΣHREEUCC.

Sample 1 (S1) Sample 2 (S2) Sample 3 (S3) Sample 4 (S4) Sample 5 (S5) Sample 6 (S6)

OC %

N%

Fe203%

MnO %

Σ REE (ppm)

LREE/HREE ratio (UCC)

Ce anomaly

Sampling depth

Distance to the stream

0.47 1.40 2.97 4.51 1.98 5.88

0.14 0.18 0.23 0.38 0.18 0.28

4.20 4.09 3.86 3.08 4.00 3.33

0.09 0.11 0.03 0.03 0.08 0.05

168.39 161.43 137.50 133.61 147.70 133.62

0.67 0.64 0.59 0.61 0.63 0.62

1.02 0.96 0.96 0.95 0.98 0.97

30–50 0–30 0–30 0–30 0–30 0–30

350 350 200 25 175 300

2.1.2. Hydrochemical background The Kervidy/Coët-Dan groundwater can be divided into two spatially distributed hydrogeological and hydrochemical domains (Dia et al., 2000; Durand and Juan Torres, 1996; Molénat et al., 2002, 2008). The first domain includes bottomland or wetland domains. Here the water table usually reaches the organic-rich upper soil horizons during the wet season (winter and spring) leading to the development of temporary reducing conditions and colored, dissolved organic carbon (DOC)-rich (5 b DOC b 40 mg L−1) groundwater. By contrast deeper in the water table (i.e., N 1 m deep), the groundwater is colorless and DOC-poor (i.e. b5 mg L−1). The rare earth element concentrations are up to 15 μg L−1 in the shallow organic-rich groundwater, and then decrease with depth (i.e., b 4 m) to 0.15 μg L−1. The second domain includes the hillslope domains, where the water table is always a few meters below the soil surface (Fig. 1). Groundwater is oxidizing, colorless and DOC-poor (DOC b 5 mg L−1). The REE concentration vary from 30 μg L−1 to concentration below 0.1 μg L−1 at depth b15 m. The cerium (Ce) anomaly is characterized by the (Ce/Ce*) value, CeN ffi with CeN, LaN and PrN normalized where (Ce/Ce*) is defined as pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi LaN þ Pr N

to upper continental crust (UCC) (Taylor and McLennan, 1985; Akagi and Masuda, 1998; Riou, 1999). No Ce anomaly occurs in the schist aquifer (Ce/Ce* ≈ 1.0; Dia et al., 2000). However, previous investigations of the REE geochemistry of the groundwater (Dia et al., 2000; Pourret et al., 2007, 2010) showed a topography-related variation in the REE signatures in the upper part (i.e., at a depth ranging between ca. 3 and 15 m) of the water table; in particular, an upslope development of a large, negative Ce anomaly.

2.2. Soil sample characteristics Five samples weighting approximately 10 kg (Samples 2 to 6, called S2 to S6) from the organo-mineral horizon (0–30 cm; A-horizon) and one sample (Sample 1: S1) from the mineral horizon were collected along the Kerroland transect (Sample 1 to S4) and Gunolay transect (Samples 5 and 6: S5 and S6) (Fig. 1 and Table 1), thereby representing the upper soil horizons of the catchment. The land use analysis shows that 80% of the area is covered by cultivated land (S1, S2 and S3 and S5; covered by cereal), 10% by wetland (S4; covered by meadows) and 5% by forest land (S6; covered by fir). All of the experiments described below were conducted on the b2 mm soil fraction (NF-ISO11464, 2006). The major (including Fe and Mn) and trace element compositions of the soil samples were determined at the SARM laboratory in Nancy, France, by inductively-coupled plasma optical emission spectrometry (ICP-OES, Thermo Elemental IRIS radial) and inductivelycoupled plasma mass spectrometer (ICP-MS, Thermo Elemental X7), respectively, using a combination of fusion techniques with LiBO2 and acid digestion with HNO3 (Carignan et al., 2001). The organic carbon (OC) and N contents were determined at the Ecobio laboratory in Rennes (France), using an oxygen combustion method with a CS Analyzer (LECO SC 144DRPC). Both sets of analyses were combined to calculate the OC/Fe and OC/Mn ratios (mass ratio).

cm cm cm cm cm cm

m m m m m m

2.3. Experimental setup The experimental procedure was based on a combination of two series of experiments: (i) leaching experiments, carried out to determine the distribution pattern and fractionation of the water-extractable REE, and (ii) reductive dissolution experiments, performed to discern the nature of the mineral phases hosting REE. Speciation determinations were also carried out using ultrafiltration techniques with regard to the size distribution through decreasing cut-off sizes. 2.3.1. Leaching experiments Leaching experiments were conducted with dynamic column systems on the six soil samples using the experimental set-up described in detail in Pédrot et al. (2008). The column systems consisted of three separated Plexiglas columns, each with an inner diameter of 21 mm and a length of 370 mm. These columns were placed parallel to each other and were connected to each other by flexible tygon hoses that had an internal diameter of 0.2 mm. The size of the columns was selected to compensate the spatial variability of the hydraulic conductivity, especially concerning the edge effect of the columns. During each experiment, approximately 50 g of dry soil (prepared according to standard NF-ISO-11464, 2006) was placed in each column, so that the total amount of leached soil corresponded to 150 ± 0.1 g. Each Plexiglas column was equipped with an inlet and an outlet to allow the solution to circulate. The synthetic percolating solution used in the experiments was placed in a solution reservoir consisting of a 1500 mL polypropylene bottle equipped with an outlet so that the solution could circulate and an air vent to compensate for pressure variations. A horizontal perforated Teflon disk (pore size b 100 μm) was inserted at the bottom of each column to prevent any soil particles from entering the reservoirs. The solutions percolating at the outlets of the columns were mixed and collected in a single reservoir called the “sample reservoir”, which was used to collect the sample solution for chemical analysis. The sample reservoir consisted of a 500 mL polypropylene bottle equipped with a syringe to sample the percolating solution. The samples were filtered at b0.2 μm to determine the composition of the aqueous phase. The total porosity in the soil columns (i.e. pore water volume) and the effective porosity were estimated in order to efficiently compare these different soil samples. The total porosity was determined by measuring the maximum quantity of water that the soil can contain and the effective porosity corresponds to the gravimetric water content (Musy and Soutter, 1991). A peristaltic pump (Ismatec Ecoline) was used to continuously drive the solution from the solution reservoir to the columns. The solution gently percolated through the soil sample and was then driven to the sample reservoir by means of a second peristaltic pump. Under equilibrium conditions, the percolation rate was set to approximately 1.2 mL/min in order to optimize the soil–solution interface reactions. Air vents on top of the solution reservoir and on top of the soil reservoir allowed to preserve the aerobic conditions during the experiments, and also prevented pressure variations from occurring within the system. The synthetic solution consists of a NaCl solution (from 1 to 2.10− 3 mol L− 1), the concentration of which corresponds to the ionic strength of the soil water within the investigated soil horizons

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

(Dia et al., 2000). At the end of the leaching experiment, 1250 mL of the synthetic solution has percolated through the soil sample. This allowed the collection of between 11 and 12 soil solution samples during one elapsed time leaching experiment. 2.3.2. Reductive dissolution experiments Batch reducing experiments were carried out using a chemical reducing agent: L-ascorbic acid (Sigma Aldrich; N99%). Previous batch studies have shown that ascorbic acid facilitates the reductive dissolution of Fe(III)-bearing minerals, including ferrihydrite, hematite, goethite, and lepidocrocite (Larsen and Postma, 2001; Larsen et al., 2006). Ascorbic acid has typically been used as a proxy for the various reductants in the environment that can result in the dissolution of ferricbearing iron oxides (Debnath et al., 2010). Only three of the six studied samples were used in this experimental set-up. The selection was made on the basis of the OC/Fe ratio. The following samples were included: S1 with a very low OC/Fe ratio (0.16), S2 with a low OC/Fe ratio (0.49), and S6 with a high OC/Fe ratio (2.53). All of the soil samples are located in the hillslope area, and none of them exhibit any redox oscillations, in contrast to what can be observed in the wetland area (Curmi et al., 1998; Walter and Curmi, 1998). The experimental set-up consisted of a 250 mL polypropylene bottle enclosed by aluminium foil to protect the soil suspensions from UV irradiation. The soil suspensions were prepared with 1) a dry soil/solution ratio of 1:100 (wt/wt) (a soil concentration of 10 g L−1), and 2) a dry soil/solution ratio of 1:20 (wt/wt) (a soil concentration of 50 g L−1) for experiments 1 and 2, respectively. Three soils were tested in experiment 1, whereas only soil sample S1 was tested in experiment 2 in order to obtain longer dissolution kinetics. Reducing experiments were conducted using 9.5 × 10−2 mol L−1 L-ascorbic acid (Postma, 1993). In all of the cases, the experiments without acid ascorbic were carried out using the same batch method and soil/solution ratio but with only 2 × 10−2 mol L−1 NaCl. The soil suspensions were stirred during 26 days and 37 days for experiments 1 and 2, respectively. The pH and Eh were recorded throughout the experiments. Soil suspension samples were collected four times throughout the experiments. The samples were filtered at b 0.2 μm to determine the composition of the aqueous phase. All of the experiments were made in duplicate. 2.4. Chemical analyses The pH was measured with a combined Mettler InLab® electrode after a calibration performed with WTW standard solutions (pH = 4.01 and 7.00 at 25 °C). The accuracy of the pH measurement was ±0.05 pH units. The Eh was measured using a combined Pt electrode (Fisher scientific Bioblock). The Eh values are presented in millivolts (mV) relative to the standard hydrogen electrode. The dissolved organic carbon (DOC) concentration was analyzed on a Total Organic Carbon Analyzer (Shimadzu TOC-5050A). The accuracy of the DOC measurement was estimated at ±3% (by using a standard potassium hydrogen phthalate solution). Specific Ultra Violet Absorbance (SUVA) (SUVA = UV Abs254 nm (cm−1) ∗ 100 / DOC (mg L−1)) — strongly correlated with the degree of aromaticity of the OM (Weishaar et al., 2003) — was used as an indicator of the chemical composition and reacand tivity of the dissolved organic carbon. The major anion (Cl−, SO2− 4 NO− 3 ) concentrations were measured by ion chromatography (Dionex DX-120) with an uncertainty below 4%. Major- and trace-element concentrations were determined by ICP-MS (Agilent 4500), using indium as an internal standard. The international geostandard SLRS-4 was used to check the validity and reproducibility of the results. Organic-rich samples were digested with sub-boiled nitric acid (HNO3 14.6 mol L−1) at 85 °C, and then re-solubilized in HNO3 0.37 mol L−1 after complete evaporation before any measurement of the major and trace element concentrations was taken in order to avoid any interference with the DOC during the mass analysis. All of the measurements were made in triplicate. Typical uncertainties, including all sources of error, were

87

below ±5% for all of the trace elements, whereas for the major cations, the uncertainty was between 2% and 5%, depending on the measured concentrations (Yéghicheyan et al., 2001; Davranche et al., 2004). In addition to the analysis of the b0.22 μm sample solutions, ultrafiltration was performed on the samples in order to separate the colloidal fraction from the true dissolved phase following the decreasing pore size cut-off. Fifteen mL centrifugal tubes (Vivaspin) equipped with permeable membranes of decreasing pore sizes (30 kDa, 5 kDa and 2 kDa) were used. Each centrifugal filter device was washed and rinsed with HCl 0.1 N and ultra-pure water three times before use. Centrifugations were performed using a Jouan G4.12 centrifuge with a swinging bucket at about 3000 g for 20 min for the 30 kDa devices, and 3500 g and 3750 g for 30 min for the 5 kDa and 2 kDa devices, respectively (Pédrot et al., 2009, 2010). Therefore, for the leaching experiment, at least three soil solution samples collected during the leaching of each soil were ultrafiltered. All of the procedures (sampling, filtration and analysis) were carried out in order to minimize contamination. Blank tests were carried out during this study to determine possible contamination due to the filtration and analysis and were always negligible. 3. Results 3.1. Soil chemical composition The soil chemical composition data are presented in Table 1. The organic carbon (OC) concentrations varied widely from 0.47% to 5.88%. The OC concentrations were maximum in the forest and wetland soil samples (5.88 and 4.51%, respectively) and minimum in the deep, mineral soil sample (0.47%). The iron concentrations, expressed in weight % Fe2O3, were less variable and ranged from 3.08% to 4.2%. The iron concentrations were inversely correlated to the OC concentrations, and were maximum in the deep mineral sample and in the sample taken in the cultivated areas and minimum in the forest and wetland soil samples. The manganese concentrations were low and weakly variable, ranging from 0.03% to 0.11%. The rare earth element patterns in the bulk soil samples are presented in Fig. 2a. All of the REE patterns were similar, exhibiting a nearly flat distribution of REE, with either no Ce anomaly or an insignificant Ce anomaly (Ce/Ce* ranging from 0.95 to 1.02). The Light REE (LREE; i.e., from La to Eu) over the Heavy REE (HREE; i.e., from Gd to Lu) ratios (Long et al., 2010) appeared to be similar between these six soils, varying from 0.59 to 0.67. These patterns were similar to the patterns previously reported by Dia et al. (2000) for deeper soil horizons in the Kervidy/Coët-Dan catchment or for the deep fresh schist from which these soils arose. The elemental composition of the soils shows that the REE concentration decreases with the increase in the OC/Fe ratio. 3.2. Leaching experiments pH and Eh values of the soil solutions at the end of the leaching experiment and for the two hydrochemical domains of the Kervidy/CoëtDan groundwater are displayed in Table 2. pH values of soil solutions were similar and close to pH 6, except for S6 soil solution sample. The strong concentration of organic carbon (OC = 5.88%, Table 1) associated with vegetation acidophilus (fir) can explain an acidic pH (pH = 4.5) for this station. The DOC, Fe and REE concentrations through time versus the size fraction for the S2, S3 and S4 samples are displayed in Fig. 3. There was a larger amount of DOC in the upper soil horizon with a high OC/Fe ratio. The Fe concentrations in the dissolved phase were low (b300 μg L− 1). Moreover, the ultrafiltration showed that the Fe concentrations in the colloidal fraction (i.e. N5 kDa fraction) are weak (b100 μg L−1) except for the organic-rich samples S4 and S6 ([Fe] = 400 μg L−1) (Fig. 3). This specific Fe distribution in the N 5 kDa fractions suggest that Fe is mainly present in the colloidal fraction N5 kDa as bound ion or ferric phases stabilized by organic molecules, as previously observed by Pédrot et al. (2008, 2009). The amount of leached REE was

88

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

REE/UCC*10 6

10000

Sample 1 Sample 2 Sample 3

1000

Sample 4 Sample 5 Sample 6

a)

100 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

1000

REE/UCC*10 6

Sample 1

100

Sample 2 Sample 3 Sample 4

10

Sample 5 Sample 6

b)

1 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Fig. 2. Upper continental crust (UCC) normalized (Taylor and McLennan, 1985) REE patterns a) displayed by the bulk soil samples and b) recovered in experimental percolation solutions (b0.22 μm).

larger in the soil sample with a high OC/Fe ratio, and evolved with regard to the leached DOC. A positive correlation can be observed between the concentrations of the organic molecules in the N 5 kDa fraction and the REE concentrations (Fig. 4) as soon as the calculated SUVA is high (SUVA N 3.5 for organic molecules). The b2 kDa and b5 kDa fractions were REE-depleted, whereas the b30 kDa and b0.2 μm fractions were REE-enriched (Fig. 3). Thus, the dissolved REEs were mostly associated with organic colloids N5 kDa. The normalized REE patterns for the percolating waters (b 0.2 μm) at the end of the leaching experiments are presented in Fig. 2b. All of the patterns were quite similar, exhibiting a middle rare earth element (MREE) downward concavity, but they differ from each other with regard to the Ce anomaly development, displaying Ce anomalies of variable amplitude (Ce/Ce* ranging from 0.42 to 1.03). These Ce anomalies stabilized through time irrespective of the soil sample location. As soon as the beginning of the leaching experiments, the Ce anomaly amplitude differs in the various soil solutions with regard to the soil sampling location. In addition, as shown for sample S4 in Fig. 5, the REE patterns for the soil solution were identical in the various

Table 2 pH and Eh of soil solutions of the soil samples along the Kerroland and Gunolay transects and of groundwater along the Kerroland transect. Values of groundwater are obtained from Riou (1999) and Dia et al. (2000).

Sample 1 (S1) Sample 2 (S2) Sample 3 (S3) Sample 4 (S4) Sample 5 (S5) Sample 6 (S6) Groundwater in the fresh schist Groundwater uphill wells Groundwater closer to the stream

pH

Eh

6.5 6 5.8 5.9 6.8 4.5 5.3 to 5.6 5.2 to 5.4 6 to 6.4

410 415 391 400 397 371 230 to 360 430 286 to 410

size fractions. Furthermore, the REE data showed that the REE pattern remained constant through time, regardless of the ultrafiltration cut-off.

3.3. Reductive dissolution experiment In the experiments without any addition of a reducing agent, the ΣREE concentration in the dissolved phase ranged from 0.46 to 2.65 μg L− 1, with the highest concentration in sample S6: ΣREE = 2.65 μg L−1 versus 0.51 and 0.46 μg L−1 for samples S2 and S1, respectively. The concentration of the dissolved Fe was always low, indicating no dissolution of the Fe oxyhydroxides (the Fe concentration ranged from 30 to 310 μg L− 1). When the reducing agent was added, the Fe concentration increased strongly to reach a reduction of 58% to 85% in the total Fe soil content (the Fe concentration ranged from 171 to 209 mg L−1) (Table 3). This increase in Fe was clearly favored by the dissolution of the Fe oxyhydroxides produced by the reducing agent and was combined with a strong increase in the REE concentration in all of the soil samples. This rise was 13 to 350 times larger with ascorbic acid than without it (ΣREE = 49.7 μg L−1, 166.6 μg L−1, 221.6 μg L−1 for samples S6, S2 and S1, respectively) (Table 3). The rare earth element patterns in the experiments without ascorbic acid were identical to the REE pattern developed in the solutions used in the leaching experiments (Figs. 2 and 6). In particular, the LREE/HREE ratios and the Ce anomaly amplitudes were similar (with the following LREE/HREE ratios for the leaching and reductive experiment: 1.66/1.63 for sample S1; 0.79/0.85 for sample S2; 0.87/0.93 for sample S6, respectively). In the reductive experiments with ascorbic acid, the REE pattern and REE concentrations evolved through time, especially for mineral soil sample S1. Whereas the LREE/HREE ratio for soil samples S2 and S6 remained basically unchanged during the reduction, the LREE/HREE ratio for the mineral soil sample S1 decreases to 1.14 at t = 26 days (Table 3). The results showed a contrasted evolution of the Ce anomaly for the three soil samples with and without ascorbic acid (Table 3). For example, sample S6 yielded solutions showing a lack of a Ce anomaly, even after 85% of the total Fe was reduced. In contrast, a slight negative

70

400

5

50

4

<2 kDa

30

<30 kDa

<5 kDa

<2kDa <5kDa

200

<30kDa

<0.2 µm

<0.2 µm

20

REE (ppb)

300

40

Fe (ppb)

DOC (mg/L)

60

100

0 0

a)

10

20

30

40

<5kDa <30kDa

2

<0.2 µm

1

10 0

<2kDa

3

0

0

50

10

20

30

40

50

0

5

10

15

20

Pore volumes

Pore volumes

25

30

35

40

45

50

Pore volumes

400

50

5 4

<5 kDa <30 kDa

30

<0.2 µm

<2kDa <5kDa

200

<30kDa <0.2 µm

20

REE (ppb)

300 <2 kDa

40

Fe (ppb)

DOC (mg/L)

60

100

0

b)

0

10

20

30

40

<30kDa <0.2 µm

0

0

50

<5kDa

2 1

10 0

<2kDa

3

10

20

30

40

50

0

10

20

Pore volumes

Pore volumes

30

40

50

Pore volumes

350

<2kDa

1200

<2kDa

<30 kDa <0.2 µm

800

100

400

50

200

0

0 0

5

10

15

Pore volumes

20

25

30

<0.2 µm

600

REE (ppb)

<5 kDa

150

<5kDa

4

<0.2 µm

<30kDa

<30kDa

<2 kDa

200

5

<5kDa

1000

250

Fe (ppb)

DOC (mg/L)

300

c)

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

70

3 2 1

0

5

10

15

20

25

30

0 0

5

Pore volumes

10

15

20

25

30

Pore volumes

Fig. 3. Release of the DOC, Fe and REE concentrations throughout the leaching experiment for samples a) S2, b) S3 and c) S4.

89

90

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

50 R² = 0.84

DOC (mg L-1)

40

30

20

10

0

0

1

2

3

4

5

REE (µg L-1) Fig. 4. Diagram illustrating the positive correlation (Pearson's coefficient) between the DOC and REE concentrations in N5 kDa fractions in the upper horizons of the soil along the Kerroland transect.

Ce anomaly is observed in sample S2 (Ce/Ce* = 0.81) without ascorbic acid, which disappeared in the presence of the reducing agent after 76% of the total Fe was reduced. Most importantly, the marked negative Ce anomaly (Ce/Ce* = 0.58) developed in the sample S1 solution in the experiment without ascorbic acid completely disappeared in the presence of the reducing agent and was replaced by a strong positive Ce anomaly (Ce/Ce* = 1.82) at t = 26 days after 58% of the total Fe was reduced. The samples collected in experiments 1 and 2 were used to monitor the kinetics of the Fe and REE solubilization from soil sample S1. The Fe solubilization was concomitant to that of REE, and especially to that of Ce, as shown by the positive linear relationship between REE and Fe and between Ce and Fe in Fig. 7. The Mn concentrations increased at the same time, especially during the beginning of the experiment with ascorbic acid (Table 3). After 7 days, the Mn concentrations increased very slightly, which differs strongly from the REE and Fe behavior. Ultimately, the Mn concentrations increased in a weak proportion probably because the soil samples are depleted in Mn (Table 1). Moreover, based on the sequential extraction procedure applied to soil sample S4, a previous study has shown that a significant fraction of Mn occurred in the exchangeable and residual fractions thereby decreasing the reducible fraction (Pédrot et al., 2009). 4. Discussion 4.1. Rare earth element signatures in soil solutions The soil chemical composition analysis shows a relationship between the soil OC and Fe concentrations and the land use. Land cover

soils located in uncultivated areas, under forests or in wetlands, show a high OC/Fe ratio (OC/Fe ratio ≥ 2). In contrast, soils from cultivated areas exhibit a low OC/Fe ratio (OC/Fe ≤ 1). This correlation probably arises from a combination of factors and processes including differences in the (i) vegetation-sourced OC input to the soil, (ii) mineralization rates, (iii) soil moisture, and (iv) lack of soil mixing caused by tillage in uncultivated soils. The results from the soil leaching experiment showed a MREE downward concavity in the REE patterns and a variation in the amplitude of the Ce anomaly (Fig. 2). Moreover, the amount of leached REE was larger in the soil sample with a high OC/Fe ratio. Several studies dedicated to the distribution of REE in natural organic-rich groundwater have demonstrated that variations in the aquifer-rock composition or any anthropogenic input play a minor role in the recorded groundwater REE signature (Viers et al., 1997; Braun et al., 1998; Olivié-Lauquet et al., 2001; Gruau et al., 2004). The dominant factors appear to be the OM content and the redox status of the waters, notably the establishment of reducing conditions, which can lead to the reductive dissolution of soil Fe and Mn oxyhydroxides and to the subsequent release of the REE adsorbed or/and complexed to these mineral phases. Both Fe and Mn oxyhydroxides are able to strongly bind REE (Bau, 1999; Bau and Koschinsky, 2009; Ohta and Kawabe, 2001; Quinn et al., 2006; Schijf and Marshall, 2011). The REE fractionation that occurred with the decreasing pore-size cut-offs indicates that the dissolved REEs were mostly solubilized with the medium molecular weight colloids and, more precisely, by the molecular weight assemblages between 5 kDa to 30 kDa (Fig. 3). This fraction is DOC-enriched, and the amount of DOC in the N 5 kDa fraction strongly affected the amount of leached REE (Fig. 3). A positive relationship between REE and DOC in the N5 kDa fraction has been observed (Fig. 4). Leached REEs were therefore significantly bound to organic colloids, as confirmed by the MREE downward concavity observed in the REE patterns in all of the size fractions (Figs. 2 and 5), which remained unchanged across all of the experiments. This specific feature reflects a predominance of organic colloidal REE complexes, as previously observed in DOC-rich waters (Viers et al., 1997; Dia et al., 2000; Gruau et al., 2004; Johannesson et al., 2004; Pédrot et al., 2008; Pourret et al., 2010; Davranche et al., 2011). Therefore, the results of the leaching experiment confirmed the key role played by OM — through the colloidal pool as a vector — in the transfer of REE from the upper soil to the surrounding shallow groundwater. The amount of DOC is the major controlling factor of the dissolved REE concentration in the soil solution. The amplitude of the Ce anomaly decreases in the soil solutions between the upland and the bottomland along the Kerroland transect, suggesting a topographic control (Fig. 2). The same upslope development of a large negative Ce anomaly and a topography-related variation in the REE signatures for the surrounding shallow groundwater were previously observed (Dia et al., 2000; Pourret et al., 2010). Several

< 2 kDa

100

< 5 kDa < 30 kDa

REE/UCC*10 6

< 0.2 µm

10

1 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Fig. 5. Upper continental crust (UCC) normalized (Taylor and McLennan, 1985) REE patterns after the successive ultrafiltration of experimental percolation solutions interacting with sample S4.

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

91

Table 3 Chemical composition of the soil solutions during the reductive experiments 1 and 2. LREE/HREE ratio is ΣLREEUCC/ ΣHREEUCC.

Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 2 Experiment 2 Experiment 2 Experiment 2 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1

Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 2 Experiment 2 Experiment 2 Experiment 2 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1 Experiment 1

Sample location

Sample time

Fe (ppm)

Mn (ppm)

Ce (ppb)

REE (ppb)

%Ce on REE

Ce anomaly

LREE/HREE ratio (UCC)

S1 S1 S1 S1 S1 S1 S1 S1 S2 S2 S2 S2 S6 S6 S6 S6

Reference t=1h t = 4 days t = 26 days t = 6 days t = 12 days t = 23 days t = 35 days Reference t=1h t = 4 days t = 26 days Reference t=1h t = 4 days t = 26 days

0.03 0.07 38.79 171.17 305.4 363.36 634.88 718.39 0.31 0.51 83.80 209.48 0.27 0.33 78.51 171.80

0.01 0.02 4.45 5.49 33.92 31.64 39.84 34.97 0.01 0.01 6.17 7.68 0.36 0.66 1.98 2.88

0.14 0.19 50.52 125.79 224.899 377.64 573.78 688.73 0.19 0.29 22.00 72.04 1.15 1.63 12.15 21.93

0.46 0.60 100.53 221.56 516.20 777.28 1038.23 1216.34 0.51 0.79 51.57 166.62 2.65 3.81 27.38 49.70

29.7 32.1 50.3 56.8 43.6 48.6 55.3 56.6 37.1 36.2 42.6 43.2 43.2 42.8 44.4 44.1

0.58 0.63 1.35 1.82 1.12 1.38 1.75 1.81 0.86 0.77 1.00 1.01 1.06 1.04 1.08 1.10

1.63 1.63 1.30 1.14 1.05 1.11 1.18 1.16 0.82 0.93 0.85 0.92 0.93 0.96 1.00 0.79

Sample location

Sample time

REE (ppb) La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

S1 S1 S1 S1 S1 S1 S1 S1 S2 S2 S2 S2 S6 S6 S6 S6

Reference t=1h t = 4 days t = 26 days t = 6 days t = 12 days t = 23 days t = 35 days Reference t=1h t = 4 days t = 26 days Reference t=1h t = 4 days t = 26 days

0.068 0.116 14.055 25.942 71.101 94.890 119.382 144.447 0.083 0.163 10.297 32.928 0.445 0.666 5.014 9.319

0.136 0.193 50.524 125.790 224.899 377.637 573.779 688.734 0.189 0.288 21.995 72.045 1.147 1.632 12.148 21.927

0.036 0.040 4.959 9.202 27.591 38.400 44.356 50.236 0.029 0.044 2.461 8.047 0.134 0.192 1.303 2.236

0.160 0.178 19.908 36.663 117.224 162.288 183.671 202.592 0.123 0.177 9.688 31.093 0.554 0.792 5.250 8.787

0.024 0.032 3.824 7.495 24.010 33.652 37.891 41.302 0.024 0.035 1.862 6.156 0.107 0.155 1.079 1.836

0.004 0.005 0.714 1.518 4.808 6.740 7.588 8.304 0.005 0.007 0.385 1.304 0.022 0.032 0.225 0.416

0.014 0.016 2.422 5.089 16.911 23.474 25.534 28.117 0.019 0.027 1.516 4.898 0.084 0.118 0.825 1.587

0.001 0.002 0.330 0.762 2.391 3.305 3.790 4.163 0.003 0.004 0.232 0.747 0.012 0.016 0.118 0.255

0.007 0.008 1.754 4.186 12.878 17.579 20.134 22.321 0.017 0.021 1.347 4.236 0.064 0.088 0.642 1.501

0.001 0.002 0.300 0.718 2.187 2.988 3.390 3.863 0.003 0.004 0.258 0.775 0.012 0.017 0.118 0.278

0.0031 0.0046 0.8323 1.9701 5.8273 7.9872 9.0224 10.4317 0.0078 0.0102 0.7287 2.0905 0.0331 0.0472 0.3145 0.7601

0.0004 0.0007 0.1129 0.2723 0.7790 1.0550 1.2066 1.4390 0.0011 0.0014 0.1000 0.2850 0.0046 0.0071 0.0431 0.1041

0.0040 0.0041 0.6961 1.7078 4.9252 6.4263 7.4607 9.0916 0.0066 0.0108 0.6128 1.7555 0.0297 0.0453 0.2651 0.6154

0.0004 0.0008 0.0989 0.2433 0.6684 0.8580 1.0263 1.2962 0.0014 0.0013 0.0903 0.2611 0.0048 0.0064 0.0384 0.0843

previous studies evidenced that pH variations can promote fractionation of dissolved REE (Koeppenkastrop and De Carlo, 1992, 1993; Ohta and Kawabe, 2001). Thus, REE pattern shapes in groundwaters are influenced — to a large extent — by pH-driven adsorption/desorption reactions between solution and mineral surfaces (Verplanck et al., 2004; Gammons et al., 2005; Olias et al., 2005; Welch et al., 2009). Nevertheless, Pourret et al. (2010) showed that the slight pH variation between the top and bottomland shallow groundwater cannot explain the observed downwards decrease in the Ce anomaly amplitude in groundwaters. Since another positive relationship was observed between pH of soil solution and Ce anomaly present in bulk soil or in dissolved phase of leachates during the leaching experiment of soil samples (Tables 1, 2 and Fig. 2b), we suggest that in the Kervidy/CoëtDan catchment, the Ce anomaly amplitude is not controlled by preferential adsorption onto mineral surfaces which was pH dependent. This downslope decrease in the negative Ce anomaly is correlated with a DOC-enrichment of the groundwater, as well as with a change in the REE speciation. Rare earth elements in groundwater from the bottomland are dominantly bound by organic colloids, whereas in the upland groundwater, they also occur as inorganic species (Pourret et al., 2010). Furthermore, the DOC concentration in the bottomland groundwater increases when the water table rises in the soil organic horizons after a rainfall event, suggesting an input of organic colloids into the groundwater. These findings — combined with the presence of organic REE-enriched colloids that exhibit either no Ce anomaly or an insignificant negative Ce anomaly (Dia et al., 2000; Pourret et al., 2007; Pédrot et al., 2008) — are the basis of the hypothesis developed by Pourret et al. (2010): the lack of a negative Ce anomaly in the bottomland groundwater results from the input of soil-derived REE-bearing

organic-rich colloids. Therefore, through its ability to control the DOC groundwater content at the catchment scale, it has been proposed that topography is the ultimate key parameter to explain the REE patterns (Pourret et al., 2010). The results in the present study showed that the Ce anomaly amplitudes varied in the dissolved fractions (i.e. fractions b 0.2 μm) between 0.42 and 1.03 for the six investigated soil samples (Fig. 2). While the negative Ce anomaly amplitude varied along the Kerroland transect according to the topography, with the highest negative Ce anomalies in the upland, the Ce anomaly corresponding to soil sample S6 differs widely. Thus, although sample S6 is also located upland, no Ce anomaly developed in the leachate solution which can be explained by the forest location of sample S6 and its high OC content (Table 1). Therefore, the largest negative Ce anomalies are produced in soil with a low OC content (i.e. samples S1, S2 and S5). This observation is confirmed by the mineral soil sample S1 which shows the largest anomaly. The amplitude of the Cerium anomaly is plotted against the OC/Fe ratio in Fig. 8. Quite clearly, a positive linear relationship (R2 = 0.89) is obtained between the Ce anomaly amplitude and the soil OC/Fe ratio. The negative Ce anomaly regularly increased with the decreasing OC/Fe ratio, which itself depends on the topography. Therefore, the development of a negative Ce anomaly in soil leachates appears here to be controlled by the soil OC/Fe ratio, and this anomaly develop in soils with a low OC/Fe ratio (i.e. Fe-rich and OC-poor soils) or OC-depleted soils. High OC concentrations decrease the amplitude of the negative Ce anomaly, whereas high Fe concentrations generate a negative Ce anomaly. Thus the organic carbon and Fe oxyhydroxide contents are determinant factors with regard to the development of a Ce anomaly both in soil solutions and shallow groundwater. Since the OC/Fe ratio increases

92

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

a) No reductant

REE/UCC*10^6

100

10

1 Sample 1 Sample 2 Sample 6

0 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

b) Reductant -Time: 4 days

10000

REE /UCC*10^6

Sample 1 Sample 2 Sample 6

1000

100 La

Ce

Pr

Nd

Sm

Eu

Gd

c) Reductant

10000

Tb

Dy

Ho

Er

Tm

Yb

Lu

-Time: 26 days

REE/UCC*10^6

Sample 1 Sample 2 Sample 6

1000

100 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Fig. 6. Diagrams comparing the shapes of the REE pattern obtained in the dissolution experiments either without (a) or with the addition (b and c) of a chemical reducing agent.

1400

Ce

1200

REE

1000

1000

800

800

600

600

400

R² = 0.98

200

200 0

400

Released REE (µg L -1)

Released Ce (µg L -1)

1200

1400

R² = 0.98

0

100

200

300

400

500

600

700

0 800

Released Fe (mg L-1) Fig. 7. Diagram illustrating the relationship (Pearson's coefficient) between the released Ce and Fe and between the released REE and Fe from the S1 soil sample. Error bars correspond to triplicates.

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

1.2 1.0 R² = 0.89

Ce/Ce*

0.8 0.6 0.4 0.2 0.0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

OC/Fe Fig. 8. Diagram illustrating the relationship (Pearson's coefficient) between the development of the Ce anomaly and the OC/Fe ratios in the studied soil samples. Error bars correspond to triplicates.

along the Kerroland transect with the topography, this latter could be a major parameter controlling the spatial variability of the Ce anomaly. However, this study demonstrates that the soil geochemical composition, influenced by the topography and land use, remains the ultimate controlling factor of the REE signature. A question then arises: why is the amplitude of the Ce anomaly linked to the soil OC/Fe ratio?

4.2. Mechanisms of REE fractionation In the experiment without ascorbic acid as a reducing agent, the dissolved REE concentration varied according to the soil samples considered. As assessed by the relationship between the dissolved REE and DOC in the leaching experiment, the OC content in the soil controls the dissolved REE concentration. The dissolved REE concentration is six times higher in the OC-rich soil sample S6 than in the OC-depleted soil sample S1 despite a higher REE concentration (+ 25% in S1) (Table 3). The dissolved REE concentrations, in the experiment without ascorbic acid, are much lower than in the reducing experiment (19 to 480 times higher in samples S6, S2 and S1 with ascorbic acid) (Table 3). This increase in the REE concentration is explained by the reductive dissolution of Fe oxyhydroxides allowing the dissolution of the Fe soil total concentration by 58% to 85% (Tables 1 and 3). A linear relationship was observed between Fe and REE, and notably, between Fe and Ce (Fig. 7). It is commonly assumed that Fe and Mn oxyhydroxides are the major scavengers of trace elements in the natural environment (Cornell and Schwertmann, 2003). Iron and Mn oxyhydroxides display the largest sorption capacity among environmental materials (Dzombak and Morel, 1990) and, as evidenced by several studies (De Carlo et al., 1998; Bau, 1999; Ohta and Kawabe, 2001; Bau and Koschinsky, 2009), sorption is a major mechanism of REE scavenging by Fe and Mn oxyhydroxides; higher proportions of REE are sorbed by the latter than by Fe oxyhydroxides. Therefore they have significant control over REE mobility and, consequently, they can sometimes be the main contributor to their release (Johannesson and Zhou, 1999; Yan et al., 1999; Cao et al., 2001; Zhang and Shan, 2001; Feng, 2010). Sequential soil extraction experiments demonstrated that the REEs were partially extracted in the reducible fraction, e.g. bound with Fe and Mn oxyhydroxides (Leleyter et al., 1999; Pédrot et al., 2009; Rao et al., 2010). However, these methods cannot be used to individually discriminate the respective contribution of the Fe and Mn oxyhydroxides in the REE solubilization. In soil pH conditions, Mn-oxides are more likely to have a significant negative residual charge (the point of zero charge for birnessite or vernadite is around 2, whereas the point of zero charge for goethite, lepidocrocite, maghemite or ferrihydrite ranges from 5.3 to 8) (Parks, 1965; Schwertmann and Fechter, 1982; Appel et al., 2003;

93

Cornell and Schwertmann, 2003). Manganese oxides thus show a higher capacity (Piper, 1974; Elderfield and Greaves, 1981; Walter, 1991) and faster REE sorption kinetics (Koeppenkastrop and De Carlo, 1992) than Fe-oxides. However, all of the soils samples have low Mn concentrations and the released Mn concentration under reductive conditions in the present study is low and not correlated with REE and Ce. Therefore, the reduction of Mn oxyhydroxides might not account for the release of REE and Ce. With regard to the Fe oxyhydroxides, Koeppenkastrop and De Carlo (1993) and Bau (1999) used experiments to show that Fe oxyhydroxides have a strong ability to trap REE. Several natural field observations also confirmed the potential of Fe oxyhydroxide in terms of REE scavenging in soils or rivers (Ingri et al., 2000; Compton et al., 2003; Steinmann and Stille, 2008; Sultan and Shazili, 2009; Stolpe et al., 2013). In the experiment without ascorbic acid, the REE patterns exhibited a MREE downward concavity as for the leaching experiments, with the same Ce anomaly scheme between the soil samples. The hillslope soil samples, which did not exhibit redox oscillations, showed a wide range in the Ce anomaly amplitude according to the OC/Fe ratio, as for the leaching experiment. In the experiment with ascorbic acid for the S1 and S2 soil samples, the Ce anomaly increased positively with the reductive dissolution of the soil Fe oxyhydroxides, providing evidence that the Fe oxyhydroxides had a positive Ce anomaly (Fig. 6). In contrast, in the soil with a high OC/Fe ratio (N 2), no change in the Ce anomaly was observed. These Ce anomaly evolutions highlight a preferential scavenging of Ce by Fe oxyhydroxides, as compared to the other REE in the S1 and S2 samples to a lesser extent. However, the whole soils exhibit no or very small negative Ce anomalies (Fig. 2a) and no Ce anomalies were observed in the fresh schist (Dia et al., 2000). Therefore, a REE fractionation occurs between the minerals and organic soil components during the weathering/alteration processes. The weathering of the primary minerals releases Fe2 +, which partially precipitates as Fe oxyhydroxides (Sposito, 1989) that bind REE, especially in the S1 sample. However, Fe oxyhydroxides are not well-known to allow the development of a positive Ce anomaly. It is often assumed that Mn oxyhydroxides develop higher positive Ce anomalies than Feoxyhydroxides since Ce oxidation typically occurs by the surface catalysis of Mn oxyhydroxides (Rankin and Childs, 1976; Steinmann and Stille, 1997; Ohta and Kawabe, 2001; Palumbo et al., 2001). Nevertheless, Bau (1999) experimentally demonstrated that Ce oxidation can occur with Fe oxyhydroxides. The sorption–oxidation of Ce by Fe oxyhydroxides is followed by the preferential loss of Ce(III) and all sorbed REE(III) in response to the aging of the Fe oxyhydroxides. The desorption increased the Ce(IV)/Ce(III) ratio and favored the development of a positive Ce anomaly. The Ce-enrichment that appeared in the present study, when the Fe oxyhydroxides were reductively dissolved, confirms this process. At the beginning of the reductive experiment, Ce represented 30% of the ΣREE, while at the end of the experiment, Ce represented 57% of the ΣREE (Table 3). The regular release of these REE and Ce during the Fe oxyhydroxide dissolution suggested that these preferentially sequestered elements are preferentially associated (i.e., adsorbed and/or incorporated) with Fe oxyhydroxides (Fig. 7). The differences in the effective ionic radius between REE (average ionic radius of REE(III) = 108.3 pm) and Fe (average ionic radius of Fe(III) = 59.7 pm) (Shannon, 1976) imply that REE can hardly substitute Fe in the Fe-oxide lattice. As explained by Takahashi et al. (2007), it is likely that the oxidative sorption of dissolved Ce(III) is the main pathway of Ce incorporation in ferromanganese oxides. The sorbate remains at the Fe–Mn oxide surface after oxidation took place, gradually resulting in Ce enrichment. Based on the low amount of Ce leading to an unsaturation of the reactive surface sites by Ce(IV), the exposed surface is gradually buried during the growth of the ferromanganese oxide. Ce(IV) is too large to replace Fe within the Fe oxyhydroxide structure, but may be incorporated into defects or nanopores within the Fe oxyhydroxides during Fe oxide

94

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

crystallization. This process has been demonstrated with lead, which provides a higher ionic radius than Ce (Vu et al., 2013). Bau (1999) showed that the oxidation rates for Ce(III) decreased significantly during the first minutes after the Fe oxyhydroxide precipitation, indicating that the oxidation capacity of Ce(III) is significantly higher for fresh Fe oxyhydroxides than for pre-formed Fe oxyhydroxides. The same conclusion was made by Steinmann and Stille (2008) but in a stream context for a mixed basaltic–granitic catchment basin. This type of specific REE fractionation occurred when the dissolved REE precipitated with Fe oxyhydroxides, favoring the development of a positive Ce anomaly and, in turn, a Ce depletion in the dissolved b0.45 μm fraction. These observations suggest that the oxidation catalysis of Ce only takes place during the growth of the Fe oxyhydroxides. However, in natural systems, the reactive sites of the Fe oxides would be unavailable or less easily accessible to dissolved Ce3+ or Fe2+ due to the adsorption of major elements or OM to the mineral surface, thereby reducing the adsorption of Fe(II) directly onto the Fe(III) mineral surface (Jones et al., 2009). The adsorption of major elements or OM to the Fe(III) minerals inhibits the direct adsorption of Fe(II) and, therefore, reduces the extent of electron exchange between Fe(II) and Fe(III) required for the Fe(II)-catalyzed transformation. Wilson et al. (2013) showed that the degree of OM impregnation of the Fe pedofeatures increases with the increasing OM concentrations. The maximum sorptive capacity of Fe oxyhydroxides was calculated as below 0.22 wt./wt. (g kg− 1 OC/g kg− 1 Fe) (Wagai and Mayer, 2007). In the S1 mineral soil sample, the OC/Fe ratio is below this ratio (0.16), suggesting that OC does not cover the entire surface of the Fe oxyhydroxides. Thus, Fe oxyhydroxide sites are not totally unavailable for binding. Therefore, the low OC/Fe ratio provides a larger number of Ce(III) oxidizable sites and Fe oxyhydroxides are able to scavenge and oxidize Ce(III) and subsequently exhibit a positive Ce anomaly. Davranche et al. (2008) used OM/Mn oxyhydroxide competitive experiments involving REE binding to demonstrate that REE are firstly bound to OM and that, in a second stage, REE are slowly desorbed from OM to be re-adsorbed by the competitive Mn oxyhydroxides. This competition results in the development of a small positive Ce anomaly onto the Mn oxyhydroxides. In their experiments, the OC/Mn ratio varied between 0.02 and 0.32, which was in the same ratio range than the OC/Fe ratio for the S1 and S2 soil samples (OC/Fe ratio: 0.16 and 0.49 for S1 and S2, respectively). Therefore, in samples S1 and S2, the low OC/Fe ratio could allow the development of a positive Ce anomaly by the soil Fe oxyhydroxides despite an initial organic REE speciation. In contrast, in soil with a high OC/Fe ratio, namely the OC-enriched soil, several authors have demonstrated that Fe is preferentially present as an ion or nanoparticle bound to colloidal OM (Gaffney et al., 2008; Pédrot et al., 2011). Upon the dissolution of the Fe(III) mineral structure following an electron exchange with aqueous Fe(II), OM prevents the recrystallization into a more stable Fe(III) mineral, due to their ability to inhibit the polymerization of thermodynamically stable Fe(III) minerals (Jones et al., 2009). Hiemstra et al. (2010) showed in several agricultural top soils that Fe oxyhydroxide particles could be embedded in a soil OC matrix forming an OM–Fe nanoparticle association with an average OM volume fraction of 80%. Moreover, Davranche et al. (2011) demonstrated in regularly flooded organic-rich soil that Fe is preferentially bound to OM as ionic species rather than as particles. These authors showed that in this type of situation, REEs are preferentially bound to the OM part of the Fe–OM associations. Therefore, in soil samples with a high OC/Fe ratio (OC/Fe ratio N 2), the oxidative potential of Ce by the Fe oxyhydroxides is low due to an OM coating and the preferential binding of REE to OM (Pédrot et al., 2009). Both processes prevent the development of a Ce anomaly. The LREE/HREE ratio in the S1 soil solution sample decreased from 1.63 to 1.14 between the beginning and end of the reduction experiment (Table 3). This decrease highlighted the preferential scavenging of HREE as compared to LREE. Weathering favors the precipitation of Fe oxyhydroxides which have more affinity for HREE than LREE, notably

at an acidic pH, resulting in a soil solution that is depleted in HREE (Öhlander et al., 1996; Bau, 1999; Verplanck et al., 2004; Gammons et al., 2005; Quinn et al., 2006). The groundwater in the fresh schist, at the second alteration front, is characterized by a more acidic pH (pH = 5.5) than in the soil solution for the upper soils (pH = 6) (Riou, 1999). These acidic pH conditions favor a preferential sorption of HREE, in particular by Fe oxyhydroxides thereby resulting in a significant REE fractionation, as already observed in the acidic continental environments (Bau, 1999; Ohta and Kawabe, 2001; Quinn et al., 2006). 4.3. Consequence with regard to the REE topography-related fractionation The present experimental dataset shows that the upper soil horizons are an important control parameter for the REE signatures of the shallow groundwater and in particular, the amplitude of the Ce anomaly. The present results demonstrated that the OC/Fe ratio for the upper soil horizon related to the soil location in the topo-sequence is able to control the REE fractionation. Not only does this control occur in the soil solution, but also it occurs in the surrounding shallow groundwater. In soils exhibiting a low OC/Fe ratio — namely OC-depleted soils — the REE fractionation is mainly controlled by Fe oxyhydroxides. The REE patterns for these Fe oxyhydroxides are enriched in HREE and exhibit a Ce positive anomaly. The reductive dissolution of these Fe oxyhydroxides produces a continuous solubilization of the whole REE, particularly Ce and HREE, indicating a preferential association of these elements. In turn, the soil solution and the surrounding shallow groundwater displayed a REE pattern that was depleted in HREE and exhibited a negative Ce anomaly. In the upland area of the catchment, the REE signature observed in the shallow groundwater is therefore sourced in the Fe oxyhydroxides occurring within the upper soil horizons. For a high OC/Fe ratio — namely in OC enriched soil — the REE pattern displayed a decreasing or even a lack of a Ce anomaly. Previous studies have demonstrated that the inhibition of the Ce anomaly is related to the preferential binding of REE by the particulate or dissolved colloidal OM, which prevents Ce(III) oxidative scavenging by Fe or Mn oxyhydroxides (Davranche et al., 2004, 2005, 2008). The lack of any negative Ce anomaly in the solution can also be explained by the low concentration of particulate Fe oxyhydroxides or oxidative sites, induced by either an OM coating on the Fe oxyhydroxides or the inhibition of their growth in the presence of OM (Gaffney et al., 2008; Pédrot et al., 2011). Therefore, in the organic-rich upper layers of the soils, negative Ce anomalies cannot occur with the same amplitude as in the mineral OC-depleted soils. In this case, the source of the REE signature is in the soil OC content and its solubilization as colloidal OC in the soil solution. These processes strongly constrain the REE pattern for the surrounding shallow groundwater as shown in the synthetic scheme displayed in Fig. 9. 5. Conclusions Leaching experiments were carried out on soil samples to understand the origin of the spatial variability displayed by the REE signatures in shallow groundwater. A special effort was made to understand the observed topography-linked variations in the Ce anomaly previously observed in natural water samples taken in the investigated transects. The results showed that the upper part of the soil represents the major source of REE in shallow groundwater, as fingerprinted by the REE patterns in the groundwater. The upper part of the soil provides the organic colloidal phase, which is the major source and transfer vector of REE. Moreover, the results for the soil sample leachates combined with the related soil chemical compositions showed that the soil OC/Fe ratio controlled the development of the negative Ce anomaly in the soil solutions. The reductive experiments showed that Fe oxyhydroxides exhibited a REE fractionation underlined by HREE enrichment and a Ce positive anomaly in the soils characterized by a low OC/Fe ratio. The regular release of these REE during the dissolution of Fe oxyhydroxide suggested

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

95

Elevation from the stream (m)

20

15

Soil OC/Fe = 0.2

10

Soil OC/Fe = 1

Controlling factor: Soil Fe-oxides

5

Soil OC/Fe = 2

Soil

Controlling factors: Soil OM and Fe-oxides competition

Controlling factor: Organic colloids

Middleland water

Bottomland water

Upland water

0 -5

DOC ≈ 1 mg L-1 pH ≈ 5.2

Weathered schist

DOC ≈ 3 mg L-1 pH ≈ 5.5

DOC ≈ 5 mg L-1 pH ≈ 6

Fresh, fractured schist

-10 250

200

150 100 Distance to the stream (m)

50

0

Fig. 9. Sketch summarizing the processes responsible for the topography-related REE pattern in a theoretical catchment. REE patterns in bold-faced correspond to REE patterns of the reducible fraction of soil. In the top of the catchment, a negative Ce anomaly in the shallow groundwater is sourced in the presence of Fe-oxide in the uppermost soil.

that these elements are preferentially associated with Fe oxyhydroxides. However, these preferential trapping of Ce, and HREE to a lesser extent, as compared to other REE, varied across the soil samples with regard to the OC/Fe ratio. This type of process occurs due to both (i) the impact of OM on the Fe pedofeatures, notably by the inhibition of their growth and by the increase in their degree of organic impregnation and (ii) the complexation of REE by organic molecules. Acknowledgments The authors would like to acknowledge Mrs. Martine Bouknik-Le Coz for the analytical assistance and Mr. Patrice Petit-Jean for the technical assistance during the laboratory work. Thorough comments of reviewers improved significantly the first version of this contribution. This research was funded by the French ANR, through the ‘Programme Jeunes Chercheuses-Jeunes Chercheurs’/‘SURFREE: Rare earth elements partitioning at solid-water interface: Impact on REE geochemical behaviour and tracing properties’. References Akagi, T., Masuda, A., 1998. A simple thermodynamic interpretation of Ce anomaly. Geochem. J. 32, 301–314. Appel, C., Ma, L.Q., Rhue, R.D., Kennelley, E., 2003. Point of zero charge determination in soils and minerals via traditional methods and detection of electroacoustic mobility. Geoderma 113, 77–93. Bau, M., 1999. Scavenging of dissolved yttrium and rare earths by precipitating iron oxyhydroxide: Experimental evidence for Ce oxidation, Y-Ho fractionation, and lanthanide tetrad effect. Geochim. Cosmochim. Acta 63, 67–77. Bau, M., Koschinsky, A., 2009. Oxidative scavenging of cerium on hydrous Fe oxide: Evidence from the distribution of rare earth elements and yttrium between Fe oxides and Mn oxides in hydrogenetic ferromanganese crusts. Geochem. J. 43, 37–47. Braun, J.J., Viers, J., Dupré, B., Polve, M., Ndam, J., Muller, J.-P., 1998. Solid/liquid REE fractionation in the lateritic system of Goyoum, East Cameroon: The implication for the present dynamics of the soil covers of the humid tropical regions. Geochim. Cosmochim. Acta 62, 273–299. Cao, X.D., Chen, Y., Wang, X.R., Deng, X.H., 2001. Effects of redox potential and pH value on the release of rare earth elements from soil. Chemosphere 44, 655–661. Carignan, J., Hild, P., Mevelle, G., Morel, J., Yéghicheyan, D., 2001. Routine analyses of trace elements in geological samples using flow injection and low pressure on-line liquid chromatography coupled to ICP-MS: A study of geochemical reference materials BR, DR-N, UB-N, AN-G and GH. Geostand. Newslett. 25, 187–198. Compton, J.S., White, R.A., Smith, M., 2003. Rare earth element behavior in soils and salt pan sediments of a semi-arid granitic terrain in the Western Cape, South Africa. Chem. Geol. 201, 239–255.

Cornell, R.M., Schwertmann, U., 2003. The iron oxides: Structures, properties, reactions, occurrence and uses. Wiley-VCH, Weiheim (664 pp.). Curmi, P., Durand, P., Gascuel-Odoux, C., Mérot, P., Walter, C., Taha, A., 1998. Hydromorphic soils, hydrology and water quality: Spatial distribution and functional modelling at different scales. Nutr. Cycl. Agroecosyst. 50, 127–142. Dabard, M.P., Loi, A., Peucat, J.J., 1996. Zircon typology combined with Sm-Nd whole-rock isotope analysis to study Brioverian sediments from the Armorican Massif. Sediment. Geol. 101, 243–260. Davranche, M., Pourret, O., Gruau, G., Dia, A., 2004. Impact of humate complexation on the adsorption of REE onto Fe oxyhydroxide. J. Colloid Interface Sci. 277, 271–279. Davranche, M., Pourret, O., Gruau, G., Dia, A., Le Coz-Bouhnik, M., 2005. Adsorption of REE(III)-humate complexes onto MnO 2 : Experimental evidence for cerium anomaly and lanthanide tetrad effect suppression. Geochim. Cosmochim. Acta 69, 4825–4835. Davranche, M., Pourret, O., Gruau, G., Dia, A., Jin, D., Gaertner, D., 2008. Competitive binding of REE to humic acid and manganese oxide: Impact of reaction kinetics on development of cerium anomaly and REE adsorption. Chem. Geol. 247, 154–170. Davranche, M., Grybos, M., Gruau, G., Pédrot, M., Dia, A., Marsac, R., 2011. Rare earth element patterns: A tool for identifying trace metal sources during wetland soil reduction. Chem. Geol. 284, 127–137. De Carlo, E., Wen, X.Y., Irving, M., 1998. The influence of redox reactions on the uptake of dissolved Ce by suspended Fe and Mn oxide particles. Aquat. Geochem. 3, 357–389. Debnath, S., Hausner, D.B., Strongin, D.R., Kubicki, J., 2010. Reductive dissolution of ferrihydrite by ascorbic acid and the inhibiting effect of phospholipid. J. Colloid Interface Sci. 341, 215–223. Dia, A., Gruau, G., Olivié-Lauquet, G., Riou, C., Molénat, J., Curmi, P., 2000. The distribution of rare earth elements in groundwaters: Assessing the role of source-rock composition, redox changes and colloidal particles. Geochim. Cosmochim. Acta 64, 4131–4151. Duncan, T., Shaw, T., 2003. The mobility of rare earth elements and redox sensitive elements in the groundwater/seawater mixing zone of a shallow coastal aquifer. Aquat. Geochem. 9, 233–255. Durand, P., Juan Torres, J.L., 1996. Solute transfer in agricultural catchments: The interest and limits of mixing models. J. Hydrol. 181, 1–22. Dzombak, D.A., Morel, F.M.M., 1990. Surface complexation modeling. Hydrous ferric oxide. J. Wiley, New York (393 pp.). Elderfield, H., Greaves, M.J., 1981. Negative cerium anomalies in the rare earth element patterns of oceanic ferromanganese nodules. Earth Planet. Sci. Lett. 55, 163–170. Feng, J.L., 2010. Behaviour of rare earth elements and yttrium in ferromanganese concretions, gibbsite spots, and the surrounding terra rossa over dolomite during chemical weathering. Chem. Geol. 271, 112–132. Gaffney, J.W., White, K.N., Boult, S., 2008. Oxidation state and size of Fe controlled by organic matter in natural waters. Environ. Sci. Technol. 42, 3575–3581. Gammons, C.H., Wood, S.A., Nimick, D.A., 2005. Diel behavior of rare earth elements in a mountain stream with acidic to neutral pH. Geochim. Cosmochim. Acta 69, 3747–3758. Gruau, G., Dia, A., Olivié-Lauquet, G., Davranche, M., Pinay, G., 2004. Controls on the distribution of rare earth elements in shallow groundwaters. Water Res. 38, 3576–3586. Hiemstra, T., Antelo, J., Rahnemaie, R., Riemsdijk, W.H.v., 2010. Nanoparticles in natural systems I: The effective reactive surface area of the natural oxide fraction in field samples. Geochim. Cosmochim. Acta 74, 41–58. Ingri, J., Widerlund, A., Land, M., Gustafsson, Ö., Andersson, P., Öhlander, B., 2000. Temporal variations in the fractionation of the rare earth elements in a boreal river; the role of colloidal particles. Chem. Geol. 166, 23–45.

96

M. Pédrot et al. / Geoderma 239–240 (2015) 84–96

Johannesson, K.H., Hendry, M.J., 2000. Rare earth element geochemistry of groundwaters from a thick till and clay-rich aquitard sequence, Saskatchewan, Canada. Geochim. Cosmochim. Acta 64, 1493–1509. Johannesson, K.H., Zhou, X.P., 1999. Origin of middle rare earth element enrichments in acid waters of a Canadian High Arctic lake. Geochim. Cosmochim. Acta 63, 153–165. Johannesson, K.H., Stetzenbach, K.J., Hodge, V.F., 1997. Rare earth elements as geochemical tracers of regional groundwater mixing. Geochim. Cosmochim. Acta 61, 3605–3618. Johannesson, K.H., Farnham, I.M., Guo, C., Stetzenbach, K.J., 1999. Rare earth element fractionation and concentration variations along a groundwater flow path within a shallow, basin-fill aquifer, southern Nevada, USA. Geochim. Cosmochim. Acta 63, 2697–2708. Johannesson, K.H., Tang, J., Daniels, J.M., Bounds, W.J., Burdige, D.J., 2004. Rare earth element concentrations and speciation in organic-rich blackwaters of the Great Dismal Swamp, Virginia, USA. Chem. Geol. 209, 271–294. Jones, A.M., Collins, R.N., Rose, J., Waite, T.D., 2009. The effect of silica and natural organic matter on the Fe(II)-catalysed transformation and reactivity of Fe(III) minerals. Geochim. Cosmochim. Acta 73, 4409–4422. Koeppenkastrop, D., De Carlo, E.H., 1992. Sorption of rare-earth elements from seawater onto synthetic mineral particles: An experimental approach. Chem. Geol. 95, 251–263. Koeppenkastrop, D., De Carlo, E.H., 1993. Uptake of rare earth elements from solution by metal oxides. Environ. Sci. Technol. 27, 1796–1802. Larsen, O., Postma, D., 2001. Kinetics of reductive bulk dissolution of lepidocrocite, ferrihydrite, and goethite. Geochim. Cosmochim. Acta 65, 1367–1379. Larsen, O., Postma, D., Jakobsen, R., 2006. The reactivity of iron oxides towards reductive dissolution with ascorbic acid in a shallow sandy aquifer (RØmØ, Denmark). Geochim. Cosmochim. Acta 70, 4827–4835. Leleyter, L., Probst, J.-L., Depetris, P., Haida, S., Mortatti, J., Rouault, R., Samuel, J., 1999. REE distribution pattern in river sediments: Partitioning into residual and labile fractions. C. R. Acad. Sci. Ser. IIA Earth Planet. Sci. 329, 45–52. Long, K.R., Van Gosen, B.S., Foley, N.K., Cordier, D., 2010. The principal rare earth elements deposits of the United States: A summary of domestic deposits and a global perspective. U.S. Geological Survey Scientific Investigations Report 2010–5220, (96 pp.). Mérot, P., Durand, P., Morisson, C., 1995. Four-component hydrograph separation using isotopic and chemical determinations in an agricultural catchment in western France. Phys. Chem. Earth 20, 415–425. Moffett, J.W., 1990. Microbially mediated cerium oxidation in sea-water. Nature 345, 421–423. Molénat, J., Durand, P., Gascuel-Odoux, C., Davy, P., Gruau, G., 2002. Mechanisms of nitrate transfer from soil to stream in an agricultural watershed of french Brittany. Water, air, & soil pollution. Springer, Netherlands, pp. 161–183. Molénat, J., Gascuel-Odoux, C., Ruiz, L., Gruau, G., 2008. Role of water table dynamics on stream nitrate export and concentration in agricultural headwater catchment (France). J. Hydrol. 348, 363–378. Musy, A., Soutter, M., 1991. Physique du sol. Presses Polytechniques et Universitaires Romandes, Lausanne (335 pp.). NF-ISO-11464, 2006. Qualité du sol — Prétraitement des échantillons pour analyses physico-chimiques. Öhlander, B., Land, M., Ingri, J., Widerlund, A., 1996. Mobility of rare earth elements during weathering of till in northern Sweden. Appl. Geochem. 11, 93–99. Ohnuki, T., Ozaki, T., Kozai, N., Nankawa, T., Sakamoto, F., Sakai, T., Suzuki, Y., Francis, A.J., 2008. Concurrent transformation of Ce(III) and formation of biogenic manganese oxides. Chem. Geol. 253, 23–29. Ohta, A., Kawabe, I., 2001. REE(III) adsorption onto Mn dioxide (δ-MnO2) and Fe oxyhydroxide: Ce(III) oxidation by δ-MnO2. Geochim. Cosmochim. Acta 65, 695–703. Olias, M., Cerόn, J.C., Fernández, I., De la Rosa, J., 2005. Distribution of rare earth elements in an alluvial aquifer affected by acid mine drainage: the Guadiamar aquifer (SW Spain). Environ. Pollut. 135, 53–64. Olivié-Lauquet, G., Gruau, G., Dia, A., Riou, C., Jaffrezic, A., Hénin, O., 2001. Release of trace elements in wetlands: Role of seasonal variability. Water Res. 35, 943–952. Palumbo, B., Bellanca, A., Neri, R., Roe, M.J., 2001. Trace metal partitioning in Fe–Mn nodules from Sicilian soils, Italy. Chem. Geol. 173, 257–269. Parks, G.A., 1965. The isoelectric points of solid oxides, solid hydroxides and aqueous hydroxo complex systems. Chem. Rev. 39, 177–198. Pauwels, H., Kloppmann, W., Foucher, J.-C., Martelat, A., Fritsche, V., 1998. Field tracer test for denitrification in a pyrite-bearing schist aquifer. Appl. Geochem. 13, 767–778. Pédrot, M., Dia, A., Davranche, M., Bouhnik-Le Coz, M., Hénin, O., Gruau, G., 2008. Insights into colloid-mediated trace element release at the soil/water interface. J. Colloid Interface Sci. 325, 187–197. Pédrot, M., Dia, A., Davranche, M., 2009. Double pH control on humic substance-borne trace elements distribution in soil waters as inferred from ultrafiltration. J. Colloid Interface Sci. 339, 390–403. Pédrot, M., Dia, A., Davranche, M., 2010. Dynamic structure of humic substances: Rare earth elements as a fingerprint. J. Colloid Interface Sci. 345, 206–213. Pédrot, M., Le Boudec, A., Davranche, M., Dia, A., Henin, O., 2011. How does organic matter constrain the nature, size and availability of Fe nanoparticles for biological reduction? J. Colloid Interface Sci. 359, 75–85. Pellerin, J., Van Vliet-Lanoë, B., 1998. Le bassin du coët-Dan au cœur du Massif armoricain. 2. Analyse cartographique de la région de Naizin. In: Cheverry, C. (Ed.), Agriculture intensive et qualité des eaux. INRA, Paris, pp. 17–24 (in French). Piper, D.Z., 1974. Rare earth elements in ferromanganese nodules and other marine phases. Geochim. Cosmochim. Acta 38, 1007–1022. Postma, D., 1993. The reactivity of iron oxides in sediments: A kinetic approach. Geochim. Cosmochim. Acta 57, 5027–5034.

Pourret, O., Davranche, M., Gruau, G., Dia, A., 2007. Organic complexation of rare earth elements in natural waters: Evaluating model calculations from ultrafiltration data. Geochim. Cosmochim. Acta 71, 2718–2735. Pourret, O., Gruau, G., Dia, A., Davranche, M., Molénat, J., 2010. Colloidal control on the distribution of rare earth elements in shallow groundwaters. Aquat. Geochem. 16, 31–59. Quinn, K.A., Byrne, R.H., Schijf, J., 2006. Sorption of yttrium and rare earth elements by amorphous ferric hydroxide: Influence of pH and ionic strength. Mar. Chem. 99, 128–150. Rankin, P.C., Childs, C.W., 1976. Rare-earth elements in iron–manganese concretions from some New Zealand soils. Chem. Geol. 18, 55–64. Rao, C.R.M., Sahuquillo, A., Lopez-Sanchez, J.F., 2010. Comparison of single and sequential extraction procedures for the study of rare earth elements remobilisation in different types of soils. Anal. Chim. Acta. 662, 128–136. Riou, C., 1999. Géochimie des Terres Rares et des éléments traces associés dans les nappes et l'eau des sols hydromorphes, (Thesis of the University of Rennes 1, Rennes, 293 pp.). Schijf, J., Marshall, K.S., 2011. YREE sorption on hydrous ferric oxide in 0.5 M NaCl solutions: A model extension. Mar. Chem. 123, 32–43. Schwertmann, U., Fechter, H., 1982. The point of zero charge of natural and synthetic ferrihydrites and its relation to adsorbed silicate. Clay Miner. 17, 471–476. Shannon, R.D., 1976. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr. Sect. A 32, 751–767. Smedley, P.L., 1991. The geochemistry of rare earth elements in groundwater from the Carnmenellis area, southwest England. Geochim. Cosmochim. Acta 55, 2767–2779. Sposito, G., 1989. The Chemistry of soils. Oxford University Press, England (304 pp.). Steinmann, M., Stille, P., 1997. Rare earth element behavior and Pb, Sr, Nd isotope systematics in a heavy metal contaminated soil. Appl. Geochem. 12, 607–623. Steinmann, M., Stille, P., 2008. Controls on transport and fractionation of the rare earth elements in stream water of a mixed basaltic–granitic catchment basin (Massif Central, France). Chem. Geol. 254, 1–18. Stolpe, B.r., Guo, L., Shiller, A.M., 2013. Binding and transport of rare earth elements by organic and iron-rich nanocolloids in Alaskan rivers, as revealed by field-flow fractionation and ICP-MS. Geochim. Cosmochim. Acta 106, 446–462. Sultan, K., Shazili, N.A., 2009. Rare earth elements in tropical surface water, soil and sediments of the Terengganu River Basin, Malaysia. J. Rare Earths 27, 1072–1078. Takahashi, Y., Manceau, A., Geoffroy, N., Marcus, M.A., Usui, A., 2007. Chemical and structural control of the partitioning of Co, Ce, and Pb in marine ferromanganese oxides. Geochim. Cosmochim. Acta 71, 984–1008. Tanaka, K., Tani, Y., Takahashi, Y., Tanimizu, M., Suzuki, Y., Kozai, N., Ohnuki, T., 2010. A specific Ce oxidation process during sorption of rare earth elements on biogenic Mn oxide produced by Acremonium sp. strain KR21-2. Geochim. Cosmochim. Acta 74, 5463–5477. Tang, J., Johannesson, K.H., 2005. Adsorption of rare earth elements onto Carrizo sand: Experimental investigations and modeling with surface complexation. Geochim. Cosmochim. Acta 69, 5247–5261. Taylor, S.R., McLennan, S.M., 1985. The continental crust: Its composition and evolution. Blackwell Scientific, Oxford (312 pp.). Verplanck, P.L., Nordstrom, D.K., Taylor, H.E., Kimball, B.A., 2004. Rare earth element partitioning between hydrous ferric oxides and acid mine water during iron oxidation. Appl. Geochem. 19, 1339–1354. Viers, J., Dupré, B., Polvé, M., Schott, J., Dandurand, J.L., Braun, J.J., 1997. Chemical weathering in the drainage basin of a tropical watershed (Nsimi-Zoetele site, Cameroon): Comparison between organic-poor and organic-rich waters. Chem. Geol. 140, 181–206. Vu, H.P., Shaw, S., Brinza, L., Benning, L.G., 2013. Partitioning of Pb(II) during goethite and hematite crystallization: Implications for Pb transport in natural systems. Appl. Geochem. 39, 119–128. Wagai, R., Mayer, L.M., 2007. Sorptive stabilization of organic matter in soils by hydrous iron oxides. Geochim. Cosmochim. Acta 71, 25–35. Walter, A.-V., 1991. Caractérisation géochimique et minéralogique de l'altération de la carbonatite du complexe alcalin de Juquia (Brésil) — Comportement des terres rares dans les minéraux phosphates, (Thesis of the University of Aix-Marseille, AixMarseille, 247 pp.). Walter, C., Curmi, P., 1998. Les sols du bassin versant du Coêt-Dan. In: Cheverry, C. (Ed.), Agriculture intensive et qualité des eaux. INRA, Paris, pp. 85–105 (in French). Weishaar, J.L., Aiken, G.R., Bergamaschi, B.A., Fram, M.S., Fujii, R., Mopper, K., 2003. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environ. Sci. Technol. 37, 4702–4708. Welch, S.A., Christy, A.G., Isaacson, L., Kirste, D., 2009. Mineralogical control of rare earth elements in acid sulfate soils. Geochim. Cosmochim. Acta 73, 44–64. Wilson, C.A., Cloy, J.M., Graham, M.C., Hamlet, L.E., 2013. A microanalytical study of iron, aluminium and organic matter relationships in soils with contrasting hydrological regimes. Geoderma 202–203, 71–81. Yan, X.P., Kerrich, R., Hendry, M.J., 1999. Sequential leachates of multiple grain size fractions from a clay-rich till, Saskatchewan, Canada: Implications for controls on the rare earth element geochemistry of porewaters in an aquitard. Chem. Geol. 158, 53–79. Yéghicheyan, D., Carignan, J., Valladon, M., Le Coz, M.B., Cornec, F.L., Castrec-Rouelle, M., Robert, M., Aquilina, L., Aubry, E., Churlaud, C., Dia, A., Deberdt, S., Dupré, B., Freydier, R., Gruau, G., Hénin, O., de Kersabiec, A.-M., Macé, J., Marin, L., Morin, N., Petitjean, P., Serrat, E., 2001. A compilation of silicon and thirty one trace elements measured in the natural river water reference material SLRS-4 (NRC-CNRC). Geostand. Newslett. 25, 465–474. Zhang, S.Z., Shan, X.Q., 2001. Speciation of rare earth elements in soil and accumulation by wheat with rare earth fertilizer application. Environ. Pollut. 112, 395–405.