Water–granite interaction: Clues from strontium, neodymium and rare earth elements in soil and waters

Water–granite interaction: Clues from strontium, neodymium and rare earth elements in soil and waters

Applied Geochemistry Applied Geochemistry 21 (2006) 1432–1454 www.elsevier.com/locate/apgeochem Water–granite interaction: Clues from strontium, neod...

431KB Sizes 0 Downloads 21 Views

Applied Geochemistry Applied Geochemistry 21 (2006) 1432–1454 www.elsevier.com/locate/apgeochem

Water–granite interaction: Clues from strontium, neodymium and rare earth elements in soil and waters Philippe Ne´grel BRGM, Avenue C. Guillemin, BP 6009, 45060 Orle´ans Cedex 02, France Received 14 November 2005; accepted 25 April 2006 Editorial handling by W.M. Edmunds Available online 17 July 2006

Abstract Strontium-, Nd-, and rare-earth-element-isotope data are presented from rock, weathered rock (arene) and saprolite, sediment and soil, shallow and deep groundwater (e.g. mineral-water springs), and surface waters in the Margeride massif, located in the French Massif Central. Granitoid rock and gneiss are the main lithologies encountered in the Margeride, which corresponds to a large and 5-km-deep laccolith. Compared to bedrock, the Sr isotopes in arene, regolith, sediment and soil strongly diverge with a linear increase in the 87Sr/86Sr and Rb/Sr ratios. Neodymium isotopes fluctuate least between bedrock and the weathering products. In order to characterise the theoretical Sr isotopic signature IRf(Sr) of water interacting with granite, a dissolution model was applied, based on the hypothesis that most of the Sr comes from the dissolution of plagioclase, K-feldspar and biotite. Similar to the Sr model, an approach was developed for modelling the theoretical Nd isotopic signature IRf(Nd) of water interacting with a granite, assuming that most Nd originates from dissolution of the same minerals as those that yield Sr, plus apatite. The IRf(Sr) ratio of water after equilibration with the Sr derived from minerals was calculated for the Margeride granite and compared to values measured in surface- and groundwaters. Comparison of the results shows agreement between the calculated IRf(Sr) and the observed 87Sr/86Sr ratios. When calculating the IRf(Nd) ratio of water after equilibration with the Nd derived from minerals of the Margeride granite, the results indicated good agreement with surface-water values, whereas mineralised waters analysed within the Margeride hydrosystem could not be directly linked to weathering of the granite alone. Because the recharge area of deep groundwater is located on the Margeride massif, very deep circulation involving interaction with other rocks (e.g. shales) at depths of >5 km must be considered.  2006 Elsevier Ltd. All rights reserved.

1. Introduction Weathering is the breakdown and alteration of rocks and minerals at or near the Earth’s surface into products that tend towards equilibrium with the conditions found in this environment. WeatherE-mail address: [email protected]

ing reactions supply solutes to surface- and groundwater, and promote or inhibit development of pathways for groundwater flow. As such, water resources in hard-rocks (e.g. granite or gneiss) commonly involve different hydrogeological compartments, such as overlying sediments, weathered rock, the weathered-fissured zone, and fractured bedrock.

0883-2927/$ - see front matter  2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.apgeochem.2006.04.007

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

The weathering process causes rocks with different chemical characteristics and ages to release Sr into water. Faure (1986) has shown that 87Sr/86Sr ratios vary according to Rb/Sr ratios and the age of the weathered material. Because Sr isotopes are not fractionated by any natural process, the measured differences in the 87Sr/86Sr ratios are due to the mixing of Sr derived from various isotopically different sources. The 87Sr/86Sr ratio variations within a hydrosystem can provide information about the sources of Sr and the different mixing processes involved (Ne´grel, 1999; Tricca et al., 1999; Probst et al., 2000; Petelet-Giraud et al., 2003). If, in a given hydrosystem, Sr comes from two components with distinct isotopic compositions, the proportion of the Sr derived from each component can be determined (Ne´grel et al., 2000; Probst et al., 2000). The Sm–Nd isotopic systematic is also related to radioactive decay, where the 143Nd/144Nd ratio varies according to the Sm/Nd ratio and the age of the weathered material. The Sr and Nd isotopic ratios are, in any case, not altered by terrestrial processes and thus provide a clear indication of the ¨ hlander et al., 2000). source material (O In contrast to Sr isotopes, Nd isotopes have not been extensively employed in hydrogeological studies. Some data are available on the compositions of river water (Goldstein and Jacobsen, 1987; Martin and McCulloch, 1999; Tricca et al., 1999; Andersson et al., 2001) and hydrothermal fluids (Michard et al., 1987), but very little information is available for spring waters (Tricca et al., 1999; Ne´grel et al., 2000, 2001a; Leybourne and Cousens, 2005). Because various minerals in silicate rocks weather differently, it is possible that Nd isotopic compositions may not reflect the bulk parent rock. In their study of Nd isotopes in major rivers around the world, Goldstein and Jacobsen (1987) found small differences between eNd(0) of both the dissolved and suspended phases in the same rivers. In rivers draining igneous and metamorphic terranes, preferential dissolution of silicate minerals (plagioclase, pyroxene, amphibole and garnet) may be more important and may induce a weak shift in the Nd isotopic composition. However, the Nd isotopic composition mainly appears to be a good indicator of the weathered parent rock (Martin and ¨ hlander et al., 2000). The physMcCulloch, 1999; O icochemical processes (i.e. weathering conditions and stability of primary REE-carrying minerals) leading to the release and mobilisation of rare earth elements (REE) during weathering are well

1433

documented (Nesbitt, 1979; Braun et al., 1993; Condie et al., 1995; Tricca et al., 1999; Aubert et al., 2001). The present study deals with chemical weathering in the Margeride granite (Massif Central, France); its main focus is on water chemistry and isotope tracing, granitic parent rock composition, sediment and soil mineralogy, and REE geochemistry. The aim of this study was to determine the isotopic signature in deep groundwater from a granite matrix in terms of water–rock interaction, and to incorporate isotopic- and chemical-tracing data and constraints into methods for evaluating groundwater circulation. 2. Site description, hydrogeological context The granite-gneiss Margeride massif is located in the SE of the French Massif Central (Fig. 1). The Desges, a tributary of the Allier River, flows over approximately 40 km between the upper part of the Margeride massif and its confluence with the Allier (Ne´grel, 1999; Fig. 1b). The Desges watershed covers 89 km2 and ranges in altitude from 1250 m (headwaters) to 650 m (outlet into the Allier). The Desges river-course is long and tortuous with a dense N–S-oriented drainage pattern mainly on the Margeride plateau. Average annual precipitation (P) rate for the Margeride massif is 1000– 1200 mm/a with less than 20% falling as snow. The Margeride massif consists of light and dark granite as a result of the fractional crystallisation of a crustal magma in a subhorizontal laccolith, aged 323 ± 12 Ma (Couturie´ et al., 1979) with leucogranites dated at 298 ± 2 Ma intruding this granite. The average mineral composition is 37% quartz, 30% oligoclase, 23% K-feldspar and 10% biotite (light facies; de Goe¨r de Herve et al., 1991) and 31% quartz, 30% andesine, 20% K-feldspar and 19% biotite (dark facies). Whole-rock 87Sr/86Sr ratios range from 0.7211 to 0.7704 (Couturie´ et al., 1979; initial 87Sr/86Sr ratio close to 0.7136). Complementary studies (Couturie´ and Vachette-Caen, 1980; Downes and Duthou, 1988) on leucogranite intruding the Margeride granite gave a whole-rock 87Sr/86Sr ratio of 0.722–1.981 (initial 87Sr/86Sr ratio of 0.7130). The Upper Continental Crust normalised REE patterns of Margeride granites are flat to HREE-enriched with a (La/Yb)N ratio ranging from 0.64 to 1.06; the patterns have a Eu anomaly ranging from 0.52 to 2.83, attributed to the decreasing amount of K-feldspar (Williamson

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1434

a

b Fig. 1. Location map of the Massif Central (a) and detailed map of the Desges watershed (b). Stars represent the different sampling points. The bedrock of Desges watershed is granite (c), gneiss (f) and micaschist and rocks of the leptynite–amphibolite group (n); they are separated on the figure by the dashed grey lines. Mineral springs are Mazel and Fontaine Basses (M.F) and Ranc (R).

et al., 1996). The eNd(0) are 4.8 for the leucogranite (Williamson et al., 1996) and 9.5 for the granite (Pin and Duthou, 1990).

The gneiss is composed of quartz, oligoclase, muscovite and biotite in proportions similar to the granite, and minor sillimanite (de Goe¨r de Herve

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

et al., 1991). For gneiss (orthogneiss) collected elsewhere in the Massif Central (Downes and Duthou, 1988), the 87Sr/86Sr ratio ranges from 0.73255 to 0.8689 and the eNd(0) ranges from 6.3 to 9. The soils were described by Bourrie (1978) as podzolic soils and peat, varying in thickness from a few centimetres at the top of the massif to a few metres at lower altitudes. The solid weathering products that form in this area are mostly kaolinite, Fe hydroxides and vermiculite (Bourrie, 1978). Sarazin (1979) identified two main levels in a 1-m-thick soil profile on the granite: in the lower part (45–100 cm depth) the texture of the parent rock is preserved. Here, the main transformations are chloritisation of biotite and kaolinisation of plagioclase, and Fe-oxyhydroxides are present. In the upper part the coherence of the parent rock is destroyed, K-feldspar and biotite are isolated from the matrix, and organic and amorphous phases are common. The soil profile is underlain by about 2 m of granite sand where kaolinisation of K-feldspar is strong and biotite is largely oxidised. Weathering of the parent rock is isovolumetric. Concerning water resources, the French Massif Central region is known as an area containing numerous mineral-water springs with salinities up to 6 g L1 (Fouillac, 1983), many of which are used for health cures or for the bottling of mineral water. Regionally, mineralised waters are characterised by neutral pH (6–7) and high CO2 concentrations corresponding to 0.1 < PCO2 < 1.0 bar. Most of the sampled waters represent natural springs (Ne´grel et al., 2000). In the Margeride area (Fig. 1), Na– HCO3 springs emerge with low flow rates near the western boundary fault of the massif. Three springs are located in the vicinity of Le Mazel (Mazel, FB 1 and FB 2) and two near Le Ranc (Ranc 1 and 2) as summarised by Ne´grel et al. (2000). 3. Water, bedrock, sediment and soil sampling, and analytical methods Water samples for REE, Nd and Sr isotopic determinations were collected in acid-washed polyethylene bottles, after filtering through pre-washed 0.2 lm Sartorius cellulose filters in cleaned Teflon apparatus under N2 (Ne´grel et al., 2000). Samples were acidified with ultra pure HNO3 to pH <2. Water samples were collected at two different locations (D19 and D8); sampling point D8 was chosen as more representative than the earlier sampling

1435

point D17, where sediment and soil samples were collected as well (Ne´grel, 1999). Stream-bed sediments were collected at the same water sampling points with plastic spatulas and stored in polypropylene boxes following the procedure described in Ne´grel (1999). Residual soil samples (top 25 cm) were collected in the surrounding fields and stored in plastic bags. Parent-rock and weathered-rock samples were collected near Mazel spring (Fig. 1). Bedrock samples were powdered prior to analysis; the fresh bedrock minerals were separated by using magnetic separation, heavy liquids, and hand-picking techniques under an optical binocular microscope. Weathered rock (Mazel) and surface saprolites (A0 and B0 samples) were powdered and the <63 lm fraction was examined by X-ray diffraction (XRD). The other soil and sediment samples were oven-dried at 70 C and powdered prior to analysis. In order to obtain representative aliquots for the analyses, samples were homogenised, quartered and dry-sieved through a 165 lm nylon mesh and examined by XRD. X-ray diffraction is a qualitative to semi-quantitative method that defines the abundance classes for each mineral. Estimation of the mineral quantities in samples was obtained by the relative intensity of each mineral reflection in the X-ray diffractograms. The results were arranged into four main classes as follows: trace (5%); low (20%); present (50%) and abundant (>65%). The analytical procedures and methods are described in detail by Ne´grel and Deschamps (1996) and Ne´grel (1999). All sediment and soil samples were analysed for major and trace elements by X-ray fluorescence energy-dispersive spectrometry (XRF). Two certified reference materials (GBW 07311; GBW 07306) were run to verify the calibration. Analytical accuracy was monitored by the repeated analysis of an in-house standard of similar mass and physical conditions as the experimental samples: precision was between ±2% (Zr) and ±5% for K2O and CaO. After sample digestion in closed Teflon bombs containing a hot tri-acid mixture (HNO3, HF and HClO4), the chemical compositions of the granite, separate minerals, weathered rock and soil were determined by using ICP-AES for major elements and ICP-MS for the trace elements and rare earth elements (REE) using a VG Plasma Quad 2 Plus spectrometer (see method described in Ne´grel et al., 2000), with a precision close to 5%.

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1436

For water samples, standard cation-exchange chemistry was used for chemical separation, and mass spectrometry was used for Sr and for the Nd isotopic compositions. Both were determined following the method described in Ne´grel et al. (2000). Ten milligrams of ultrapure Fe was added to each litre of acidified sample. The pH was then brought to 7.5 with concentrated ultrapure NH4OH. After vigorous shaking for one day, flocculation occurred over the course of one week. More than 95% of the Nd was recovered in the precipitate, whereas major cations remained in the supernatant. After extraction of the supernate followed by several centrifugation steps, the precipitate was dissolved with 2.5 N HCl and evaporated to dryness. Neodymium was then separated using standard cation exchange followed by HDEHP reverse chromatography (White and Patchett, 1984). The total procedural blank was <100 pg. The same method was also used for Nd purification from sediment and soil samples before analysis. The 87Sr/86Sr and 143Nd/144Nd ratios were measured on the sediment and soil samples following standard acid-dissolution procedures, and on filtered and acidified waters. Standard cation-exchange chemistry was adopted for chemical separation and mass spectrometry for Sr (Ne´grel and Deschamps, 1996). Mass-spectrometry analyses of Sr were made on a Finnigan MAT 262 multiple-collector mass spectrometer using single W filaments with a Ta activator. The total blank for Sr was <1 ng for the entire chemical procedure. The measured 87Sr/86Sr ratios were normalised to a 88Sr/86Sr ratio of 0.1194 and then adjusted to the NBS 987 standard value of 0.710240. The reproducibility of the 87Sr/86Sr ratio measurements was tested by duplicate analysis of the NBS 987 standard, with a mean value close to 0.710227 ± 0.000017 (2r; n = 70). Neodymium isotopic ratios were measured using a Finnigan MAT

262 multiple-collector mass spectrometer with double Re filaments. The 143Nd/144Nd ratios were normalised to a 146Nd/144Nd ratio of 0.7219 and then adjusted to the La Jolla international standard value of 0.511860. Repeated measurements of the La Jolla standard during the analyses yielded a mean 143 Nd/144Nd of 0.511826 ± 0.000011 (2r, n = 33). The 143Nd/144Nd ratios are expressed as eNd(0), which represents the deviation in parts per 104 (e units) from 143Nd/144Nd in a chondritic reservoir with a present-day CHUR value of 0.512638 (de Paolo and Wasserburg, 1976). 4. Results 4.1. Mineralogical compositions Within a region of temperate climate, weathering of granitic rock usually results in arenisation (Sequeira Braga et al., 2002). Mineralogical compositions are summarised in Table 1. The Mazel weathered-rock (arene) sample was taken from the top of the granite outcrop and samples A0 and B0 were collected from the surface saprolite. The Mazel arene sample retains the original granitic structure and skeletal fabric, and is mainly composed of quartz (20%), K-feldspar (20%), plagioclase (20%) and chlorite (5%). For this sample, the classification of Begonha and Sequeira Braga (2002) and Sequeira Braga et al. (2002) was adopted and thus it was considered as ‘‘weathered rock’’. Samples A0 and B0 are not a direct skeletal-fabric material of the granite, and sand and silt are the main grainsize fractions. These saprolite samples are mainly composed of quartz, K-feldspar, plagioclase and illite-micas, with strongly fluctuating relative abundances. Quartz is present at around 35% in sample A0 and increases to 65% in sample B0, K-feldspar and illite-smectite display a similar abundance

Table 1 Mineralogical analysis for the Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, and soil samples A2 and B2 Sample

Ref.

Quartz

K-feldspar

Plagioclase

ALT1

Mazel arene

25

20

20

5

5

D19-3/A0 D19-3/A1 D19-3/A2 D17-3/B0 D17-3/B1 D17-3/B2

A0 A1 A2 B0 B1 B2

35 50 20 65 25 15

20 5 20 20 20 20

20 5 20 5 20 20

20 20 20 20 20 20

5 20 5 5 5 5

Results are expressed in %.

Illite/micas

Chlorite

Amorphous

15 5

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

(20%), but plagioclase fluctuates between 20% in sample A0 and 5% in sample B0. Samples A1 and B1 correspond to sediments collected on river banks and can be classified as sandy-silt. The analysed fraction is different with more quartz in sample A1 (50%) than in B1 (25%). Contrary to that, K-feldspar and plagioclase are more abundant in B1 (20%) than in A1 (5%). Illite-mica is similar in the two samples (20%). Sample A1 is richer in chlorite (20%), whereas only 5% is found in sample B1. An amorphous phase, around 15%, occurs in sample B1. Samples A2 and B2 were collected in fields bordering the streams and can be classified as silty-

1437

representative granites from the Margeride plateau (the average value is shown in Table 2a). Of the major elements, K2O, SiO2 and TiO2 fluctuate the least (0.9%, 2.2% and 2.5%, respectively) and CaO (17%) the most. The trace elements Ba and Zr exhibit a standard deviation of >10%. Comparison between these data and those for the same rock type show good agreement for major and trace elements (de Goe¨r de Herve et al., 1991). In order to evaluate the extent of chemical mobility during weathering of the granite, the percentage change relative to Ti following Eq. (1) (Nesbitt, 1979) was applied to various elements:

%change ¼ 100  ð½ðelement=TiÞsample =ðelement=TiÞgranite   1Þ

sand. They contain the lowest amount of quartz (15–20%); K-feldspar, illite-micas and plagioclase are estimated to be around 20%. However, the variations in quartz, K-feldspar and plagioclase cannot be directly used as parameters for defining the weathering stage, as Begonha and Sequeira Braga (2002) did not define distinct ‘‘more or less’’ trends in different weathering profiles developed on granite. 4.2. Major-element and REE compositions in whole rock, mineral separates, sediment and soil The chemical compositions and rare earth element (REE) concentrations of ‘‘arene’’, saprolite, soil and bedrock are given in Tables 2a and 2b. In the initial study on the Margeride massif, Ne´grel (1999) used average chemical compositions of two

ð1Þ

These elements were classified as K–Rb–Ca–Sr in Fig. 2a, according to the affinity between K and Rb, and Ca and Sr in silicate rocks, and as SiO2– Fe–Ba in Fig. 2b. Potassium, Rb, Ca and Sr are largely depleted in all samples. Depletion expressed as percentage changes reaches 60 to 90% for Ca, Sr and K, except in sediment sample B0, where the percentage change is around +15% for K and around +67% for Rb. The distribution of percentage changes for Rb closely follows those of K as demonstrated by Middelburg et al. (1988), but the depletion is less marked in the percentage changes, which agrees with their work (Middelburg et al., 1988) that argued that Rb is expected to be less depleted than K. The percentage change for the second group shows similar evolution for SiO2 and Ba as illustrated in Fig. 2b. In addition, they show variations that mimic those of K and Rb, but with

Table 2a Chemical composition of average Margeride granite (referred to as whole rock), Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, and soil samples A2 and B2 Sample

Ref.

SiO2 (%)

K2O (%)

CaO (%)

TiO2 (%)

Fe2O3 (%)

Zr (ppm)

Ba (ppm)

M1/POU ALT1

Whole rock Mazel arene

68.50 48.70

5.73 2.83

1.26 0.74

0.57 0.78

3.88 5.24

175 365

1057 333

D19-3/A0 D19-3/A1 D19-3/A2 D17-3/B0 D17-3/B1 D17-3/B2

A0 A1 A2 B0 B1 B2

59.81 66.49 57.06 64.57 59.02 57.11

4.08 3.36 3.33 4.55 3.51 2.76

0.47 0.38 0.31 0.30 0.33 0.55

0.65 0.68 0.82 0.35 0.62 0.53

3.66 2.49 4.46 1.31 2.43 2.78

341 441 301 129 398 206

631 595 417 454 555 292

Results are from Ne´grel (1999) and are expressed in wt% oxides for major elements and in ppm for trace elements.

0.09 0.26 0.3 0.07 0.16 0.25 0.55 1.72 1.97 0.49 1.09 1.67 0.09 0.25 0.3 0.08 0.16 0.26 0.55 1.61 1.81 0.45 1.02 1.59 0.19 0.57 0.62 0.15 0.35 0.56 1.03 3.14 3.21 0.78 1.85 2.93 0.2 0.6 0.6 0.15 0.35 0.6 1.5 5.1 4.5 1.1 2.6 4.12 0.46 0.8 0.67 0.2 0.36 0.61 1.96 6.59 5.49 1.18 3.08 4.23 10.7 36.1 28 6.23 15.9 20.2 2.65 9.06 7.5 1.5 3.9 4.6 24 84.6 72 14 37.3 43.1 A0 A1 A2 B0 B1 B2 D19-3/A0 D19-3/A1 D19-3/A2 D17-3/B0 D17-3/B1 D17-3/B2

13.2 42.5 35.4 8 20.2 22.9

4835 117 341 12 26.7 0.1 0.22 0.05 206 0.77 1.58 0.09 35.7 0.12 0.26 0.05 247 0.88 1.78 0.1 89.7 0.33 0.73 0.05 459 1.83 4.25 0.2 75.7 0.43 1.05 0.05 398 3.69 9.5 0.25 5.53 2.17 0.25 0.08 365 3.55 11.1 0.36 964 19.7 63.7 1.91 190 5.18 16.5 0.5 1297 51.3 156 5.35 Apatite Plagioclase Biotite K-feldspar M1/Ap M1/Pl M1/Bi M1/KF

476 27.4 74.3 3.3

195 217 0.28 0.26 1.85 1.86 0.3 0.29 1.92 1.92 0.75 0.7 4.03 3.86 0.8 0.8 6.1 6.5 1.24 0.98 7.11 7.11 36.2 37.7 9.34 9.93 84.2 97.5 Whole rock Mazel arene M1/POU ALT1

41.3 47.6

P REE (ppm) Lu (ppm) Yb (ppm) Tm (ppm) Er (ppm) Ho (ppm) Dy (ppm) Tb (ppm) Gd (ppm) Eu (ppm) Sm (ppm) Nd (ppm) Pr (ppm) Ce (ppm) La (ppm) Ref. Sample

Table 2b Rare earth element concentrations for parent rock, arene, separate minerals, sediment and soil, expressed in ppm

57 193 162 34 88 108

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1438

different degrees of depletion. ‘‘Arene’’ and soil sample A2 show the highest degree of depletion for SiO2 and Ba, both around 80%. Sediment sample B0 shows the lowest degree of depletion around 30% for SiO2 and Ba. Iron is depleted in all samples with a roughly similar degree of depletion (70% to 80%). Albare`de and Semhi (1995) demonstrated that minerals like sphene and rutile stay in the sediment while Fe (as Fe-oxyhydroxides) moves out. The Ba distribution is comparable to that of K, reflecting the distribution of Ba in K-feldspar as demonstrated by Middelburg et al. (1988) and Albare`de and Semhi (1995). P The REE content ( REE) of the whole rock is 195 ppm and only a part of the REE content in granite is present in the major mineral phases, the accessory minerals monazite, sphene, zircon and apatite accounting for the remaining REE content (e.g. Gromet and Silver, 1983). Apatite typically contains the highest REE content (4835 ppm) in minerals (Aubert et al., 2001), biotite is P the second REE carrier in the minerals with a REE of around 340 ppm, and plagioclase and K-feldspar have lower REE contents (117 and 12 ppm, respectively). The arene is slightly enriched in REE (217 ppm) compared to the whole rock; the samples of sediment A1 and soil A2 are richer in REE (193 and 162 ppm, respectively) compared to sediment B1 and soil B2 (88 and 108 ppm), whereas sediment samples A0 and B0 have the lowest REE contents (57 and 34 ppm). The REE contents in soil and sediment are in agreement with those given by Albare`de and Semhi (1995) for the Meurthe River. Reconstructing the whole-rock composition using the abundance of the analysed minerals plus quartz, yields 95.5% of the analysed whole rock. Therefore, P when calculating the REE budget with mineral abundances, the trace minerals (sphene, rutile, epidote, zircon, etc.) should be considered. Calculating the amount of REE to balance between the whole rock and the separated minerals gives 1326 ppm of REE for the trace minerals that represent 4.5% of the rock. The percentage changes P calculated according to Eq. (1) for the group REE–La–Sm–Nd is very similar for the whole REE, as well as for individual REEs as illustrated in Fig. 3. Sediment samples A0 and B0 have the highest degree of depletion (70% to 80%); arene and sample A1 have the lowest degree of depletion (around 20%). The REE depletion in arene, sediment and soil can be related to the breakdown of major and/or acces-

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1439

80 K Rb Ca Sr

60 40

% change

20 0 -20 -40 -60 -80 -100 ne

a el az

are

A0

A1

A2

B0

B1

B2

A1

A2

B0

B1

B2

M

-20 Ba Fe SiO2

% change

-40

-60

-80

-100

b

ne

l ze Ma

are

A0

Fig. 2. Percentage changes relative to Ti for K, Rb, Ca and Sr (a) and Ba, Fe and Si (b) to evaluate the extent of chemical mobility for the Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, and soil samples A2 and B2.

sory mineral phases present in the granite during weathering (Nesbitt, 1979; Braun et al., 1993; Condie et al., 1995; Aubert et al., 2001). The large changes in REE content should be important with regard to mass balance as they reflect large variations in the weathered minerals. This is especially

seen in samples A0 and B0 compared to the other samples.P The REE content and individual REE in sediments usually indicate dilution by quartz and K-feldspar through the La vs. Si and/or Ti relationship (Albare`de and Semhi, 1995). Changes in trace

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1440 0

REE La Sm Nd

-20

% change

-40

-60

-80

-100

z Ma

el

ar e

ne

A0

A1

Fig. 3. Percentage changes relative to Ti for the

minerals such as sphene and rutile would induce a shift in La content. Likewise, changes in zircon, apatite and monazite, generally largely fractionated in LREE relative to HREE, would induce a shift in the La/Sm ratio. However, as shown in Fig. 4, there is only a decrease in the La and Sm contents from whole rock and arene to soil samples A0 and B0, without a change in the La/Sm ratio of 6.3. A roughly similar relationship was shown by Albare`de and Semhi (1995), but their La/Sm ratio was 5 in the Meurthe sediments. The constant La/Sm ratio implies no addition or subtraction of trace minerals (apatite, monazite, sphene or rutile). Similarly, there is no evidence of fractionation between Sm and Nd (the linear relationship gives a coefficient of determination R2 of 0.99). 4.3. REE patterns in whole rock, separate minerals, sediment and soil The granite-normalised REE patterns of the mineral separates are presented in Fig. 5. Apatite shows an REE pattern with LREE depletion as the (La/ Yb)N and (Pr/Yb)N ratios demonstrate (0.1 and 0.2, respectively). The (La/Yb)N and (Pr/Yb)N ratios are defined by the ratio Lasample/Ybsample to the ratio Lagranite/Ybgranite, and Prsample/Ybsample to the ratio Prgranite/Ybgranite. Biotite has REE dis-

A2

P

B0

B1

B2

REE. Samples are the same as in Fig. 2.

tribution patterns that are similar to that of plagioclase with comparable (La/Yb)N and (Pr/Yb)N (Fig. 5) with exception of Eu anomaly. The (La/ Yb)N ratio is close to 1.2 in biotite and to 1.6 in plagioclase, the (Pr/Yb)N is close to 2.1 in biotite and to 1.3 in plagioclase; the difference reflects higher HREE depletion in plagioclase than in biotite. Kfeldspar shows a roughly flat REE pattern with (La/Yb)N and (Pr/Yb)N ratios close to 1.6 and 1.1, respectively. The Eu anomaly of apatite is negative and similar to that of biotite (0.1). The Eu anomaly, after de Baar et al. (1985) is defined as Eu/Eu* = 2EuN/ [SmN + GdN]. EuN, SmN and GdN correspond to the normalised granite concentrations. The Eu anomaly of plagioclase is positive but small (1.4). Therefore, K-feldspar is the only mineral in this granite carrying a strong positive Eu/Eu* of 3.2. The Ce anomaly after de Baar et al. (1985) is defined as Ce/Ce* = 2CeN/[LaN + PrN], where CeN, LaN and PrN, correspond to the normalised granite concentrations. The Ce anomaly is consistently close to unity {Ce/Ce* = 0.9–1}. The REE patterns in sediment and soil clearly differ as shown in Fig. 6a (REE patterns from site D19 plus arene) and in Fig. 6b (REE patterns from site D17 plus the arene). The arene, A1 (sediment collected on river bank) and A2 (soil sample) pat-

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1441

60

Mazel arene

D19-3/A1

Whole rock

40

La (ppm)

D19-3/A2

/ La

Sm

=

6.

3

D17-3/B2 D17-3/B1

20

D19-3/A0 D17-3/B0

0 0

2

4 Sm (ppm)

6

8

Fig. 4. Plot of the La vs. Sm content in Margeride granite, mineral separates and Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, and soil samples A2 and B2.

terns are comparable without any large REE fractionation in the arene and A1 samples whereas soil sample A2 shows a larger HREE fractionation and sediment sample A0 from the surface saprolite has a different REE pattern. Compared to site D19, the sediments show different REE patterns, being more HREE-enriched. Sediment sample B1 collected on the river bank has (La/ Yb)N and (Pr/Yb)N ratios close to 0.8, whereas these ratios are close to 0.6 in soil sample B2. Sediment sample B0 from the surface saprolite again has a different pattern, similar to that of sediment sample A0, with the exception of the Eu anomaly. The pattern for B0 is flat, whereas that for A0 mimics the plagioclase pattern. This agrees with the mineralogical composition of 20% plagioclase in sample A0. The soil REE patterns for samples A2 and B2 differ from those given by Aubert et al. (2001) in the system soil–soil solution–stream water in a small catchment of the Vosges Mountains. They demonstrated that REE patterns normalised to granite in a soil profile had decreasing HREE depletion deeper

in the profile, but the most closely related pattern was found at 180 cm depth whereas the soils sampled for the present study are from the uppermost 20 cm. However, as the mineral separates in the Vosges granite display different REE patterns than those in the Margeride, different patterns can be created by the weathering process (Aubert et al., 2001). Moreover, because none of the sediment and soil samples have a positive Ce anomaly, no redoxcontrolled processes leading to the formation of CeO2 (cerianite) can be invoked, similar to the process described by Aubert et al. (2001). 4.4. Sr and Nd isotopes in separate minerals, sediment and soil The Rb–Sr and Sm–Nd data of separate minerals from the Margeride granite are given in Table 3 with the associated abundance of each mineral in the rock matrix. Biotite is the poorest Sr-bearing mineral with 10 ppm Sr and conversely the richer Rb-bearing mineral with a concentration of

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1442 1000

Apatite Plagioclase Biotite K-Feldspar

CSample /CGranite

100

10

1

0.1

0.01 La

Ce

Pr

Nd Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm Yb

Lu

Fig. 5. REE patterns of the mineral separates normalised to the Margeride granite.

816 ppm. Apatite is more concentrated in Sr with 110 ppm but poor in Rb (9.5 ppm). Potassic feldspar has Sr concentrations of 338 ppm and a similar Rb concentration of 433 ppm, whereas plagioclase is richer in Sr with a concentration of 546 ppm and 203 ppm Rb. Apatite and plagioclase have the lowest, but clearly distinct, 87Sr/86Sr ratios (0.71250 and 0.71594, respectively). Biotite has high 87 Sr/86Sr ratios, reflecting its high Rb/Sr ratios (1.769). Potassic feldspar has an intermediate 87 Sr/86Sr ratio (0.72575), higher than the 0.718 value for feldspar in Margeride granite given by Beaucaire and Michard (1982). The whole-rock value is 0.72269, agreeing with the lowest part of the range given by Couturie´ et al. (1979); 0.7211–0.7704, but higher than the one value (0.714) given by Beaucaire and Michard (1982) for the Margeride granite. Apatite is the richer Nd-bearing mineral with a concentration of 964 ppm, associated with a Sm concentration of 365 ppm, both reflecting the affinity of phosphatic minerals for REE. Conversely, K-feldspar shows the lowest Nd (1.91 ppm) and Sm (0.36 ppm) concentrations. Neodymium concentrations increase in plagioclase (19.7 ppm) and biotite (63.7 ppm), as do Sm concentrations (3.55 and

11.1 ppm, respectively). The ratios are expressed as 143 Nd/144Nd and eNd(0) in Table 3. The lowest eNd(0) is observed for biotite (-9.9), low and similar values are observed for plagioclase and K-feldspar (-9.4) and the highest value is observed for apatite (-4.9). The whole-rock value is eNd(0) = 9.1, comparable with the value given by Pin and Duthou (1990) of 9.5Pfor the same granite. As for the REE, the Rb–Sr budget calculated with the mineral abundances has to take into consideration the input from trace minerals (sphene, rutile, epidote, zircon, etc.) to reach 100% (e.g., Oliva et al., 2003). The calculation yields an amount for these trace minerals of 960 ppm Rb and 851 ppm Sr, a Rb/Sr ratio of 1.128 and a reconstructed 87Sr/86Sr of 0.7198. The budget for the Sm–Nd systematics was calculated in the same manner as for Rb–Sr. The amount of Sm in trace minerals is 67 ppm and that of Nd is 415 ppm with a Sm/ Nd ratio of 0.1622. The reconstructed epsilon Nd value ratio of trace minerals is 9.8. The Sr isotopic composition of arene, sediment and soil presented vs. the Rb/Sr ratios in Fig. 7 plot along a line, with the exception of mineralised-water samples Ranc 1 and 2 that differ by their Rb/Sr

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454 1.2

1.2

1

1

0.8

CSample /CGranite

CSample /CGranite

0.8

0.6

0.4

0.6

0.4

0.2

0.2

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

a

1443

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

b

D19-3/A0 D19-3/A1 D19-3/A2 Arene Mazel

D17-3/B0 D17-3/B1 D17-3/B2 Arene Mazel

Fig. 6. REE patterns normalised to Margeride granite of the Mazel arene sample (a,b), sediment samples A0 (a) and B0 (b) from surface saprolite, sediment samples A1 (a) and B1 (b) collected on river bank, and soil samples A2 (a) and B2 (b).

Table 3 Rb, Sr concentrations, 87Sr/86Sr ratios, Sm and Nd concentrations, minerals, sediment and soil Sample

Ref.

Sr (ppm)

Rb (ppm)

M1/POU ALT1

Whole rock Mazel arene

278 123

281 246

M1/Ap M1/Pl M1/Bi M1/KF

Apatite Plagioclase Biotite K-feldspar

110 546 10 338

9.54 203 816 433

D19-3/A0 D19-3/A1 D19-3/A2 D17-3/B0 D17-3/B1 D17-3/B2

A0 A1 A2 B0 B1 B2

79 92 77 45 76 57

275 268 331 289 365 285

Rb/Sr

143

87

Sr/86Sr

Nd/144Nd ratios and eNd(0) for parent rock, arene, separate Sm (ppm)

Sm/Nd

143

36.2 37.7

0.196 0.189

0.51217 0.512168

9.1 9.1

Nd (ppm)

Nd/144Nd

eNd(0)

1.01 2.00

0.72269 0.731338

7.11 7.11

0.09 0.37 78.46 1.28

0.712504 0.715943 1.768812 0.725749

365 3.55 11.1 0.36

964 19.7 63.7 1.91

0.379 0.180 0.174 0.188

0.512387 0.512154 0.51213 0.512155

4.9 9.4 9.9 9.4

3.50 2.90 4.29 6.49 4.79 4.97

0.750848 0.744083 0.752901 0.785077 0.766495 0.762957

1.96 6.59 5.49 1.18 3.08 4.23

10.7 36.1 28 6.23 15.9 20.2

0.183 0.183 0.196 0.189 0.194 0.209

0.512131 0.512127 0.512119 0.512176 0.512171 0.512197

9.9 9.9 10.1 9.0 9.1 8.6

All concentrations in ppm. See text for the eNd(0) calculation.

ratio. The 87Sr/86Sr and Rb/Sr ratios both increase between the whole rock (0.72269 and 1.01, respectively) and the Mazel arene (0.73134 and 2.0, respectively), reflecting the weathering of the low 87 Sr/86Sr- and Rb/Sr-bearing minerals such as plagioclase and K-feldspar, as suggested by the mineralogy of the arene (20% of each in this sample). Sediment A1 collected on the river bank at site D19 shows a 87Sr/86Sr ratio of 0.74408 and a Rb/ Sr ratio of 2.90, higher than that of the arene. An increase is noted in the sediment from surface sapro-

lite A0 (87Sr/86Sr = 0.75085, Rb/Sr = 3.50) to soil A2 (87Sr/86Sr = 0.75290, Rb/Sr = 4.29). At site D17, soil B2 displays the lowest 87Sr/86Sr ratio (0.76296) associated with the lowest Rb/Sr ratio (4.96). As for site D19, there is an increase in the 87 Sr/86Sr and Rb/Sr ratios from sediment B1 collected on the river bank (87Sr/86Sr = 0.76649, Rb/ Sr = 4.79) up to the surface saprolite sample B0 (87Sr/86Sr = 0.78508, Rb/Sr = 6.49). The Nd isotopic compositions of arene, sediment and soil are presented in Fig. 8 vs. the Sm/Nd ratios

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1444

87

0.8

Biotite Sr/86Sr = 1.7688 Rb/Sr = 78.46

D17-3/B0

0.78

D17-3/B1 D17-3/B2

Sr/ Sr

0.76

6 09 01 0 . .9 9 = 0 Y ²= R

*X

+

71 0.

11

9

86

D19-3/A2 D19-3/A0

87

D19-3/A1

0.74 Mazel arene

K-Feldspar Whole rock

Whole rock-arene Separate minerals Sed/soils Surface waters Mineralized waters

0.72 Plagioclase Apatite

0.724 Whole rock

trace minerals

0.7 0

2

4

6

8

Rb/Sr

86

Sr/ Sr

0.72

87

Ranc2

Plagioclase

0.716 Ranc1

Mazel FB1 FB2

Apatite

0.712 0

0.4

Rb/Sr

0.8

1.2

Fig. 7. Relationship between the 87Sr/86Sr ratios and Rb/Sr ratios for the Margeride granite, separate minerals and Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, soil samples A2 and B2, and surface and mineralised-spring waters. Note the extended view (Rb/Sr ratios <1.2).

(Faure, 1986). The eNd(0) values for the soil B2, sediment B1 collected on the river bank and the surface saprolite B0 from site D17 are within bedrock and the Mazel arene. For site D19, all samples show low eNd(0), around 10. Sediment A1 collected on the river bank and the surface saprolite A0 are similar, whereas soil sample A2 shows a clear increase in the Sm/Nd ratio. Fig. 9 shows the relationship between the 87 Sr/86Sr ratios and eNd(0). The divergence in saprolite, sediment and soil compared to bedrock is more marked for the Sr isotopes than for Nd isotopes. All samples from site D19 have a slightly divergent eNd(0) compared to the bedrock and separate minerals. This can be explained by the influence of residual trace minerals (sphene, rutile, zircon, etc.) as indicated by their Sr- and Nd-isotope compositions,

plus a more than likely input from weathering of the gneiss unit (Fig. 1) because the river drains this unit. 4.5. REE, Nd and Sr isotopes in ground and river waters Five mineralised waters were analysed P in the Margeride area (Ne´grel et al., 2000), whose REE ranges from 255 ng/L (SDF, Source Fe´e) to 1065 ng/L (Mazel). These REE contents agree with previous investigations carried out in the Massif Central by Michard et al. (1987) and Sanjuan et al. (1988). Dissolved REE concentrations are presented as granite-normalised patterns in Fig. 10. In the Margeride area, the dissolved REE patterns are HREE-enriched {(La/Yb)N = 0.03–0.12, (Pr/Yb)N = 0.03–0.16}. Europium shows a slight positive

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1445

-4 Whole rock-arene Separate minerals Sed/soils Surface waters Mineralized waters

Apatite

εNd(0)

-6

Mazel

-8

FB2

-10

-8 FB1 Ranc1

D17-3/B2

-8.5

-12 0

εNd(0)

D17-3/B1 Mazel arene

0.8

1.2

1.6

Sm/Nd

D17-3/B0

-9

0.4

Whole rock

K-Feldspar

-9.5

Plagioclase Biotite D19-3/A0

-10

D19-3/A1

D19-3/A2

-10.5 0.16

0.18

0.2

0.22

Sm/Nd

Fig. 8. Relationship between the eNd(0) and Sm/Nd ratios for Margeride granite, separate minerals and Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, soil samples A2 and B2, and surface and mineralised-spring waters. Note the extended view (Sm/Nd ratios <0.22).

anomaly with Eu/Eu* ranging from 1 to 2.8, whereas Ce shows a negative anomaly with Ce/Ce* ranging from 0.19 to 0.83. The conclusion of the descriptive REE study (Ne´grel et al., 2000) of the mineralised-spring waters in the Massif Central is that most are characterised by HREE enrichment {(La/Yb)N < 0.4, (Pr/Yb)N < 0.6} and large positive (Er/Nd)N values (Tables 1 and 3). Similar REE patterns occur in the Vichy basin located farther north and at Vals-les-Bains located to the east (Michard et al., 1987; Sanjuan et al., 1988). For surface P waters the data are summarised in Table 4. The REE ranges from 431 (D19) to 400 ng/L (D8), clearly comparable to that of mineralised-spring waters. The granite-normalised REE patterns (Fig. 10) are HREE-enriched {(La/ Yb)N = 0.29–0.18, (Pr/Yb)N = 0.23–0.20}. Europium shows a slight positive anomaly with Eu/Eu*

close to 1.91–1.70, whereas Ce shows a slight negative one with Ce/Ce* close to 0.76–0.91. Comparable REE contents as well as REE patterns for the stream waters have been found by Tricca et al. (1999) and Aubert et al. (2001). The Sr isotopic data from mineral waters in the Margeride have been presented elsewhere (Ne´grel et al., 2000). Such mineral waters appear to fall into two groups: three springs (Mazel, FB1 and FB2) have the lowest Sr isotopic signature around 0.71295–0.71317, whereas mineral waters from the Ranc have the highest 87Sr/86Sr (0.71624–0.71636). The isotopic composition of Nd in these waters ranges from eNd(0) = 11.4 to 7.6 (Ne´grel et al., 2000). The most strongly negative values are for the Ranc and FB1 waters, whereas the least negative value is for Mazel. Comparing the solid samples and the water (either surface water or mineralised-spring

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1446

0.79 D17-3/B0

0.785

Whole rock-alterite minerals Soils

0.78 0.775

trace minerals

0.77 0.765

D17-3/B1 D17-3/B2

87

Sr/86Sr

0.76 0.755 0.75

D19-3/A2 D19-3/A0

0.745

D19-3/A1

0.74 0.735

Mazel arene

0.73

K-Feldspar

0.725 Whole rock

0.72 Plagioclase

0.715

Apatite

0.71 -10.5 -10

-9.5

-9

-8.5

-8

-7.5

εNd(0)

-7

-6.5

-6

-5.5

-5

-4.5

Fig. 9. Relationship between the eNd(0) and 87Sr/86Sr for Margeride granite, separate minerals and Mazel arene sample, sediment samples A0 and B0 from surface saprolite, sediment samples A1 and B1 collected on river bank, soil samples A2 and B2, and surface and mineralised-spring waters. The eNd(0) and 87Sr/86Sr of the remaining trace minerals (sphene, rutile, epidote, zircon) are indicated in the diagram (see text).

water) shows: (i) low eNd(0) for the two mineralised waters Ranc 1 and FB 1), with a eNd(0) lower than 11, and (ii) all waters have significantly higher Sm/Nd ratios. It is obvious that some springs from the Margeride area have a eNd(0) lower than the measured values of the parent rock.

5. Discussion

Part of the Sr mass balance in a crystalline environment should be controlled by Sr supplied by chemical weathering of Sr-bearing phases from the host rock, either primary phases or their alteration products, e.g. epidote and clay minerals (Franklyn et al., 1991; Blum et al., 1994; Brantley et al., 1998). Among the minerals typically found in granite, those that most commonly influence the Sr isotopic budget in waters are apatite, plagioclase, K-feldspar, and biotite and muscovite.

5.1. Strontium sources in groundwater 5.2. The weathering model for Sr and Nd Strontium-isotope ratios vary in nature because one of the Sr isotopes (87Sr) is formed by the radioactive decay of naturally occurring Rb (87Rb). The 87Sr/86Sr ratios can be used as tracers of water–rock interaction (Blum et al., 1994; Ne´grel et al., 2000; Oliva et al., 2003). The primary sources of Sr in groundwater are atmospheric input, dissolution of Sr-bearing minerals, and anthropogenic input.

In water–rock interaction studies, mineral-dissolution rates have been described by kinetic studies (Berner, 1978; Lasaga, 1984) that provide insight into the mechanisms of mineral transformations (Murphy and Helgeson, 1987). However, a thermodynamic approach like that of Hegelson (1968), allows prediction of the chemical composition of the fluid when the system involves low fluid–rock

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454 100 D19-water D8-water

CSample /CGranite

10

1

0.1 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er TmYb Lu 100

CSample /CGranite

10

1

FB1-ETM Ranc-ETM Mazel-ETM FB2-ETM

0.1 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er TmYb Lu

Fig. 10. Dissolved REE patterns of the surface waters and mineralised springs in the Margeride normalised to Margeride granite (data from Ne´grel et al., 2000).

ratios and assumes thermodynamic equilibrium between fluids and neogenic phases. In order to characterise the theoretical Sr isotopic signature of water interacting with granite, the dissolution model given by Ne´grel et al. (2001b) was applied, based on the hypothesis that most of the Sr comes from the dissolution of plagioclase, Kfeldspar and biotite (Zuddas et al., 1995; Bullen et al., 1997; Probst et al., 2000). Moreover, for newly formed phases that are in equilibrium with their parent solution, only the Sr concentration of the fluid is modified, whereas the 87Sr/86Sr ratio remains constant. Such a dissolution model requires the following parameters: (1) the Sr content of the three mineral phases (Srmx); (2) the Sr isotopic com-

1447

position of the three mineral phases (iSrmx); (3) the proportion of each mineral in the rock (%mx); and (4) the weatherability of each mineral phase (Wmx), where ‘mx’ corresponds equally to plagioclase, biotite and K-feldspar. The relative weatherability of the minerals, taking that of plagioclase as being 1, is 0.25 for biotite and 0.1 for K-feldspar as summarised by Ne´grel et al. (2001b). A sensitivity analysis of this model was realised by testing various mineral-weatherability ratios (Blum et al., 1994; Zuddas et al., 1995); the difference in the computed 87 Sr/86Sr ratios is lower than 7 · 104 (Petelet-Giraud et al., 2003). The theoretical Sr isotopic composition (iSrth) of water in equilibrium with granite is thus defined by: P ðiSrmx  Srmx  %mx  W mx Þ P iSrth ¼ ð2Þ ðSrmx  %mx  W mx Þ The model can be verified by comparing the iSrth calculation for given conditions of Sr abundance and isotopic characteristics of separate minerals, with the 87Sr/86Sr ratio of water that has interacted with the rock. Such a validation method was applied to seven granitoids and two basalts (Ne´grel et al., 2001b). This model was recently computed for the Chirac granite (Petelet-Giraud et al., 2003), which for water draining this granite gave a theoretical 87 Sr/86Sr signature intermediate between the plagioclase and whole-rock signatures, reflecting the contribution of all minerals with a predominance of plagioclase. Waters draining the Chirac granite are in good agreement with the calculated ratio, in that they plot within, or only slightly below, the error bar of the calculated value. The model was computed for the Margeride granite using data from Table 3: the theoretical signature for water draining the Margeride granite is close to 87Sr/86Sr = 0.7177 ± 7 · 104 according to the sensitivity analysis. In a similar method to that applied for Sr, an approach has been developed for determining the theoretical Nd isotopic signature of water interacting with granite, based on the hypothesis that most Nd derives from the dissolution of apatite, plagioclase, K-feldspar and biotite. The theoretical Nd isotopic composition (iNdth) of water in equilibrium with granite is defined by the following equation : P ðiNdmx  Ndmx  %mx  W mx Þ P iNdth ¼ ð3Þ ðNdmx  %mx  W mx Þ where ‘mx’ corresponds to apatite, plagioclase, biotite and K-feldspar, Ndmx is the Nd content of each

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1448

Table 4 pH, Sr, Rb, REE concentrations, Ce and Eu anomalies and La/ YbN and Er/NdN ratios, 87Sr/86Sr and 143Nd/144Nd ratios for surface waters collected on the Margeride Sample

D19

D8

pH Sr (lg/L) Rb (lg/L) La (ng/L) Ce (ng/L) Pr (ng/L) Nd (ng/L) Sm (ng/L) Eu (ng/L) Gd (ng/L) Tb (ng/L) Dy (ng/L) Ho (ng/L) Er (ng/L) Tm (ng/L) Yb (ng/L) Lu (ng/L) P REE (ng/L) Ce/Ce* Eu/Eu* La/YbN Pr/YbN 87 Sr/86Sr 143 Nd/144Nd

7.14 19 1.6 90 130 16 65 25 9 25 5 25 5 16 3 14 3 431.0 0.74 1.69 0.47 0.35 0.71486 0.512232

6.89 8 0.8 65 124 16 72 24 7 20 4 25 5 16 3 16 3 400.0 0.88 1.50 0.30 0.31 0.71425 0.512232

model, at least for the Margeride granite. Therefore, as stated by Aubert et al. (2001), it is assumed that the Nd isotopic composition of the granite–water interaction can be estimated through the weathering of apatite, plagioclase, biotite and K-feldspar. For apatite dissolution, a microbial role cannot be discarded as evidenced by Taunton et al. (2000). As very few studies deal with the application of Nd and its isotopes to weathering processes, there is no possibility, as was done for Sr, to evaluate the suitability of the model for reconstructing the Nd isotopic composition of the water after interacting with granite. Aubert et al. (2001) suggested on the basis of Sr and Nd isotope ratios that a major part of the Sr and Nd in waters interacting with granite originates from leached or dissolved phosphate minerals, like apatite. They stated that both biotite and K-feldspar play a minor role in controlling the isotopic composition of water. The model was computed for the Margeride granite using data from Table 3: the theoretical signature for water draining the Margeride granite is close to eNd(0) = 7.7 ± 0.2 according to the sensitivity analysis.

5.3. The atmospheric-input correction mineral, iNdmx is the isotopic composition of each mineral, %mx is the proportion of each mineral in the rock, and Wmx is the weatherability of each mineral phase. The last two were defined for the Sr model and are also used in the Nd model. Harlavan and Erel (2002) established the fact that apatite dissolves faster than sphene, based on P and Ti concentrations, apatite being at least 2–3 times less resistant to weathering than sphene. Biotite and hornblende are important when studying the release of REE during granitoid weathering, biotite because of its accessory-phase inclusions and hornblende because it is the most dominant primary mineral that includes REE. Harlavan and Erel (2002) also established that hornblende is twice as resistant to weathering as biotite and that allanite (REE-rich) is even less resistant than apatite. As Harlavan and Erel (2002) demonstrated, allanite should play an important role in controlling the REE in waters and therefore the Nd isotopic composition of the water. However, allanite is not present in the Margeride granite (Couturie´ et al., 1979; Couturie´ and Vachette-Caen, 1980) which precludes the consideration of allanite in the weathering

The Sr input in surface waters is from dissolved material in rainwater, dissolved material contributed by the weathering of different Sr-bearing mineral phases, and anthropogenic disturbance. In order to identify the water signature related only to water–rock interaction with the Margeride granites, it is first necessary to subtract that part of the mineralisation carried by rainwater, as demonstrated by Ne´grel and Deschamps (1996) and Petelet-Giraud et al. (2003). The atmospheric-input correction consists of quantifying and subtracting the portion of elements carried by rainwater into runoff. The chemical composition of rainwater is thus needed for calculating a geochemical mass-balance budget. According to work done on atmospheric corrections (Ne´grel and Deschamps, 1996; Ne´grel, 1999; Oliva et al., 2003; Petelet-Giraud et al., 2003), the atmospheric contribution to stream water for any element ‘Z’ is estimated by reference to the Cl concentration in the stream, multiplied by the Z/Cl ratios in rainwater. In the absence of evaporites in the Desges catchment, Cl ions in the atmosphere originate from wind-blown sea salt and human activity. In this

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

case, the mean chemical composition of rainfall in the Massif Central over 1.5 a (Ne´grel and Roy, 1998) was used, weighted by the quantity of each rain event. The rain collector used was located about 30 km SW from the study site (Ne´grel and Roy, 1998), roughly at the same distance from the ocean and under the same climatic conditions. For mass-balance equations, the highest concentrations of Cl ions from rainwater must be determined. The method of Ne´grel and Deschamps (1996) was used in which the mean weighted Cl content of rainwater is multiplied by an evapotranspiration concentration factor F, defined by annual precipitation rate P/(annual precipitation rate P  annual evaporation rate E), here close to 1.6. This gave a maximum Cl concentration of 32 lmole/L from rainwater input. If the Cl content in the stream water is greater than this value, the difference is attributed to human activity. For surface waters, three sampling points (D8, D17 and D19) were investigated by Ne´grel (1999) and two of them were re-sampled once for this study (D8 and D19). Indices 1–3 in Table 5 refer to the different sampling periods for the same sampling point. The samples were collected during short periods in May 1994 (index 1), September 1994 (index 2) and May 1995 (index 3). Applying the atmospheric correction to the surface-water samples taken at D8, D17 and D19, showed that four samples derived their total Cl content from atmospheric input, and thus reflect no anthropogenic influence. They are listed in Table 5 and correspond to D8 samples collected in May 1994 and May 1995, D17.3 collected in May 1995, and the D8 sample collected during this study. The Sr content derived from atmospheric input varies from 23% (D19, this study) to 57% (D8.3, May

1449

1995). Correction on the 87Sr/86Sr ratio is always significant, and varies from 3.7 to 6.1 · 104. Some samples have a residual Cl content that is clearly related to anthropogenic input, mainly from agricultural activity. Petelet-Giraud et al. (2003) defined a way to correct such input and the anthropogenic correction for this study was done according to their protocol and in the same way as the atmospheric correction. Considering that all residual Cl after atmospheric correction derives from anthropogenic input, the ultimate correction leads to all samples having zero Cl content. The Sr derived from agricultural input varies between 6 (D8.2, September 1994) and 13% for D19 samples (this study and D19.3, May 1995). The influence of anthropogenic correction on the 87Sr/86Sr ratio is particularly sensitive at 8 · 104 (D8.2, September 1994) to 3 · 103 (D19.3, May 1995). Once corrected for atmospheric and anthropogenic inputs, the 87Sr/86Sr ratio characterising water–granite interaction ranged from 0.7169 (D8.2, September 1994) to 0.72185 (D19.3, May 1995). When coupled with Sr concentrations, Sr isotope systematics can be used to investigate the mixing of different waters. By plotting the 87 Sr/86Sr ratio vs. the reciprocal Sr concentration in two component mixtures, straight lines are produced (Faure, 1986). The 87Sr/86Sr ratio in water thus reflects the different sources of Sr originating from rock weathering, and certain constraints on the mixing from these sources are provided by Sr isotope systematics. The 87Sr/86Sr ratio is plotted vs. 1/Sr in Fig. 11 for surface waters, after atmospheric and anthropogenic correction and the result of the Sr-weathering model. Fluctuations between the different sampling points in the watershed are noticeable. Compared to the results

Table 5 pH, 87Sr/86Sr, Cl and Sr concentrations (in lmol/L) at three sampling points (D8, D17 and D19) investigated for this study and by Ne´grel (1999) Samples

pH

87

Sr/86Sr

Cl (lmol/L)

Sr (lmol/L)

D8 D19

This study This study

6.89 7.14

0.714252 0.714862

28.2 62.0

0.09 0.22

D8 D8.2 D8.3

May 1994 September 1994 May 1995

7.98 6.86 6.94

0.714879 0.714126 0.714906

28.2 42.3 30.7

0.07 0.16 0.09

D17.3 D19.3

May 1995 May 1995

7.08 7.04

0.714891 0.715858

27.3 54.9

0.10 0.15

Indices 1–3 in this table refer to different sampling periods for the same sampling point. The samples were collected during short periods in May 1994 (index 1), September 1994 (index 2) and May 1995 (index 3).

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1450

of the weathering model, some points plot within the field of uncertainties associated to the model calculation (D19 and D8 from this study, D10.3, D17.3 and D8.2 from Ne´grel, 1999). Four points plot outside the weathering-model result (D8, D8.3, D10.1 and D19.3 from Ne´grel, 1999), and the influence of a stronger weathering of K-feldspar, biotite and/or accessory phases (sphene, rutile, epidote, zircon) as suggested by Oliva et al. (2003) should be pointed out. For Nd, few studies on rainwater have been done. Aubert et al. (2002) measured 0.8 ng/L Nd in rainwater collected in the Vosges area (eastern France), a granite massif fully comparable with the Margeride massif as exemplified by the roughly similar Cl content in rainwater (average 15 lmol/L; Probst et al., 2000). The atmospheric correction for Nd can therefore be estimated by multiplying the river’s Cl concentration by the Nd/Cl ratio of rainwater, this ratio being estimated by using the data of Aubert et al. (2002) and Probst et al. (2000) for the

Vosges catchment. For samples D19 and D8 this led to less than 3% Nd originating from rainwater input. As no characterisation of the Nd isotopic composition of rainwater is available, it is not possible to correct the 143Nd/144Nd ratio in the river water.

5.4. Implications for water–rock interactions and water circulation In order to avoid any influence of the Sr and Nd concentrations, both Sr (corrected for atmospheric input) and Nd isotopic compositions can be used as water–rock interaction tracers. Fig. 12 is a plot of the Sr and Nd isotopes in sediment, soil, surface water (corrected and uncorrected for atmospheric input) and groundwater, from separate minerals and incorporating the result of the Sr and Nd weathering models. Several features can be stated from this graph:

D19.3 0.7215 D10.1 D8 0.72075

D8.3

0.71925

87

86

Sr/ Sr

0.72

D10.3 D17.3 0.7185 D8

D19

Model

0.71775

0.717 D8.2

0

10

20

30

40

1/Sr Unpolluted samples (Négrel, 1999) Polluted samples (Négrel, 1999) This study

Fig. 11. Discrimination diagram, in terms of 87Sr/86Sr vs. 1/Sr, for surface waters from Margeride (data from Ne´grel, 1999, and this study). Sr concentrations and 87Sr/86Sr ratios are corrected for atmospheric and anthropogenic inputs (see text).

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

1451

0.732 Mazel arene

0.731

Whole rock-arene separate minerals Surface waters Mineralized waters

0.73 0.729 0.728

trace minerals

0.727 K-Feldspar

0.726 0.725

87

Sr/86Sr

0.724 Whole rock

0.723 0.722 0.721 0.72

Weathering Model

0.719 0.718

ATM*

0.717 Ranc2

0.716 0.715

Plagioclase

Ranc1

0.714 Apatite

0.713 0.712

FB1

Mazel

FB2

-11.5 -11 -10.5 -10 -9.5

-9 -8.5

-8 -7.5 -7 -6.5 -6

εNd(0)

-5.5 -5

-4.5

Fig. 12. Relationship between the eNd(0) and 87Sr/86Sr for Margeride granite, separate minerals and Mazel arene sample, surface waters and mineralised springs in the Margeride. ATM* indicates the position with regard to the 87Sr/86Sr ratios of surface waters corrected for atmospheric and anthropogenic inputs. The eNd(0) and 87Sr/86Sr of the remaining trace minerals (sphene, rutile, epidote, zircon) are indicated in the diagram (see text). The result of the weathering model is indicated by the grey circle (see text).

– For surface waters, the eNd(0) and 87Sr/86Sr agree with the weathering model, even if Nd is not corrected for atmospheric input. – For mineralised-spring waters, there is no agreement with the weathering model. The Mazel spring has the same eNd(0), but lower 87Sr/86Sr, as the weathering model. FB1, FB2, Ranc 1 and 2 springs have lower eNd(0) and 87Sr/86Sr than the weathering model. – A larger influence corresponding to an increase in weathering of one or more minerals cannot solely explain the values observed for eNd(0) in the mineralised FB1, Ranc 1 and 2 waters. Indeed, no mineral is known from the Margeride granite with such a low eNd(0). This is particularly true for the residual trace minerals that cannot control the eNd(0) of the water. – For mineralised-spring waters, both their eNd(0) and 87Sr/86Sr suggest in favour of other component(s) that could lead to the observed signatures. Deep groundwater circulation can be explained through connected fracture networks, which allow

water to interact at depth. One common method to estimate the depth of water circulation is to use geothermometers on mineralised waters. Dissolved chemical-species concentrations in mineralised fluids are a function of water temperature in the aquifer and of the weathered mineralogical assemblage (Fouillac, 1983), and species whose concentrations are controlled by temperature-dependent reactions can theoretically be used as geothermometers. Such geothermometers, applied to mineralised-spring waters from the Margeride, predict reservoir temperatures of around 200 C (Fouillac, 1983). The Massif Central in general and specifically the Margeride system, are characterised at 4–5 km depth by an extrapolated temperature ranging from 200 to 250 C, suggesting a hot geothermal anomaly zone. Geothermometer temperatures obtained on Margeride mineralised-spring waters agree with water circulation reaching depths of 4–5 km (Fouillac, 1983). Thus, the low eNd(0) values observed for some of the mineralised waters, which cannot be explained by the water–granite interaction, may result from interaction with the metamorphic gneiss and shale basement below the granite (de Goe¨r de

1452

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

Herve et al., 1991). However, as there is no direct measurement of Nd isotopes in such basement rocks near the Margeride massif, the eNd(0) can solely be approached through the work of Simien et al. (1999) on the Montagne Noire, west of the Massif Central, which forms the extension of the rocks capped by the Margeride granite. They gave whole-rock eNd(0) values lower than 12, which would be consistent with the values observed in the mineralised waters. Thus, deep water circulates through the Margeride granite and underlying metamorphic rocks, suggesting a deep-seated root of the hydrothermal system whose influence should be tested at a larger regional scale, including the spectacular hydrothermal Chaudes Aigues springs that emerge at more than 80 C.

These results highlight the complexity of deep groundwater circulation in the French Massif Central, known for its numerous mineral-water springs, many of which are used for health cures or for bottling mineral water. A good knowledge of water circulation and interaction with the surrounding rocks is important for ensuring an economic use of deep groundwater in terms of drinkable mineral waters, as well as in terms of future geothermal-energy development. Moreover, Sr corrections for atmospheric and anthropogenic input are important to understand which proportions in the dissolved and particulate load stem from natural background. Such work can be of fundamental interest in the EU integrated project AquaTerra where fluxes of anthropogenic pollutants are considered in more detail.

6. Conclusions Acknowledgements Chemical weathering was studied using geochemistry and Sr and Nd isotopes, together with the mineralogy and chemistry of rocks, sediments and soils on the Margeride granite (Massif Central, France). Compared to bedrock, the Sr isotopes in weathered rock (arene), regolith, sediment and soil strongly diverge with a linear increase in the 87Sr/86Sr and Rb/Sr ratios. Neosymium isotopes fluctuate least between bedrock and the weathering products. The weathering model applied to Sr and Nd isotopes, allows characterisation of the theoretical isotopic signature of water interacting with granite. This model, developed by Ne´grel et al. (2001b), is based on the hypothesis that most of the Sr comes from the dissolution of plagioclase, K-feldspar and biotite. It was adapted for Nd in this study, assuming that Nd originates from the same minerals as Sr, plus apatite. The Sr and Nd isotopic compositions of surfaceand (mineralised) groundwater were compared to those resulting from the weathering model. Once corrected for atmospheric and anthropogenic inputs, the Sr isotopic composition of surface water fully agrees with that of the weathering model. Neodymium can be considered as non-sensitive to atmospheric input and the Nd isotopic composition in surface water is comparable to that of weathered rock. However, there is no agreement between the Sr and Nd isotopic compositions of the weathering model and those of the mineralised waters. Thus, very deep circulation, involving interaction with another deep-seated rock below the granite must be considered.

This work was funded by the BRGM Research Division and by the European Union (integrated project AquaTerra, Project 505428). Although this work did not directly investigate any of the research areas in AquaTerra, it reveals important information about natural inputs to ground and surface water via rock–water interactions. This work therefore contributes in a larger context to the EU AquaTerra under the thematic priority ’’Sustainable Development, Global Change and Ecosystems (FP6) Water Cycle and Soil Related Aspects. Mineralogical, chemical and isotopic analyses were performed in the BRGM Laboratories, France. The author gratefully acknowledge Dr. F. Darbyshire, Dr. K.H. Johannesson and Professor W.M. Edmunds and an anonymous reviewer for significant improvements of the manuscript; thanks are due to the BRGM Translation Service for improving the English. References Albare`de, F., Semhi, K., 1995. Patterns of elemental transport in the bedload of the Meurthe River (NE France). Chem. Geol. 122, 129–145. ¨ ., 2001. Andersson, P.S., Dahlqvist, R., Ingri, J., Gustafsson, O The isotopic composition of Nd in a boreal river: a reflection of selective weathering and colloidal transport. Geochim. Cosmochim. Acta 65, 521–527. Aubert, D., Stille, P., Probst, A., 2001. REE fractionation during granite weathering and removal by waters and suspended loads: Sr and Nd isotopic evidence. Geochim. Cosmochim. Acta 65, 387–406. Aubert, D., Stille, P., Probst, A., Gauthier-Lafaye, F., Pourcelot, L., Del Nero, M., 2002. Characterisation and migration of

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454 atmospheric REE in soils and surface waters. Geochim. Cosmochim. Acta 66, 3339–3350. Beaucaire, C., Michard, G., 1982. Origin of dissolved minor elements (Li, Rb, Sr, Ba) in superficial waters in a granitic area. Geochem. J. 16, 247–258. Begonha, A., Sequeira Braga, M.A., 2002. Weathering of the Oporto granite: geotechnical and physical properties. Catena 49, 57–76. Berner, R.A., 1978. Rate control of mineral dissolution under earth surface conditions. Am. J. Sci. 278, 1235–1252. Blum, J.D., Erel, Y., Brown, K., 1994. 87Sr/86Sr ratios of Sierra Nevada stream waters: implications for relative mineral weathering rates. Geochim. Cosmochim. Acta 58, 5019–5025. Bourrie, G., 1978. Acquisition de la composition chimique des eaux en climat tempe´re´. Application aux granites des Vosges et de la Margeride. Sci. Ge´ol. 52. Brantley, S.L., Chesley, J.T., Stillings, L.L., 1998. Isotopic ratios and release rates of strontium measured from weathering feldspars. Geochim. Cosmochim. Acta 62, 1493–1500. Braun, J.J., Pagel, M., Herbillon, A., Rosin, C., 1993. Mobilization and redistribution of REEs and thorium in a syenitic lateritic profile: a mass balance study. Geochim. Cosmochim. Acta 57, 4419–4434. Bullen, T., White, A., Blum, A., Harden, J., Schulz, M., 1997. Chemical weathering of a soil chronosequence on granitoid alluvium: II. Mineralogic and isotopic constraints on the behaviour of strontium. Geochim. Cosmochim. Acta 61, 291– 306. Condie, K.C., Dengate, J., Cullers, R.L., 1995. Behavior of rare earth elements in a paleoweathering profile on granodiorite in the Front Range, Colorado, USA. Geochim. Cosmochim. Acta 59, 279–294. Couturie´, J.P., Vachette-Caen, M., 1980. Age westphalien des leucogranites recoupant le granite de Margeride (Massif Central franc¸ais). C.R. Acad. Sci. 291, 43–45. Couturie´, J.P., Vachette-Caen, M., Vialette, Y., 1979. Age namurien d’un laccolite granitique diffe´rencie´ par gravite´: le granite de Margeride (Massif Central franc¸ais). C.R. Acad. Sci. 289, 449–452. de Baar, H.J.W., Bacon, M.P., Brewer, P.G., Bruland, K.W., 1985. Rare earth elements in the Atlantic and Pacific Oceans. Geochim. Cosmochim. Acta 49, 1943–1959. de Goe¨r de Herve, A., Burg, J.P., Couturie´, J.P., Delpuech, A., Duthou, J.L., Etienne, R., Mercier-Batard, F., Pe´richaud, J.J., Pin, C., Tort, M., Turland, M., 1991. Carte ge´ologique de la France au 1/50.000, Ham quadrangle and Explanatory Notes. BRGM Editions. de Paolo, D.J., Wasserburg, G.J., 1976. Interferences about magma sources and mantle structure from variations of 143 Nd/144Nd. Geophys. Res. Lett. 3, 743–746. Downes, H., Duthou, J.L., 1988. Isotopic and trace element arguments for the lower-crustal origin of Hercynian granitoids and pre-Hercynian orthogneisses, Massif Central (France). Chem. Geol. 68, 291–308. Faure, G., 1986. Principles of Isotope Geology. J. Wiley and Sons. Fouillac, C., 1983. Chemical geothermometry in CO2-rich thermal waters. Example of the French Massif Central. Geothermics 12, 149–160. Franklyn, M.T., McNutt, R.H., Kamineni, D.C., Gascoyne, M., Frape, S.K., 1991. Groundwater 87Sr/86Sr values in the EyeDashwa Lakes pluton, Canada: Evidence for plagioclase-

1453

water reaction. Chem. Geol. 86, 111–122 (Isotope Geoscience Section). Goldstein, S.J., Jacobsen, S.B., 1987. The Nd and Sr Isotopic systematics of river-water dissolved material: implications for the sources of Nd and Sr in seawater. Chem. Geol. 66, 245– 272. Gromet, L.P., Silver, L.T., 1983. Rare earth element distributions among minerals in a granodiorite and their petrogenetic implications. Geochim. Cosmochim. Acta 47, 925–939. Harlavan, Y., Erel, Y., 2002. The release of Pb and REE from granitoids by the dissolution of accessory phases. Geochim. Cosmochim. Acta 66, 837–848. Hegelson, H.C., 1968. Evaluation of irreversible reactions in geochemical processes involving minerals and aqueous solutions. I. Thermodynamic relations. Geochim. Cosmochim. Acta 32, 853–877. Lasaga, A.C., 1984. Chemical kinetics of water–rock interaction. J. Geophys. Res. 89, 4009–4025. Leybourne, M.I., Cousens, B.L., 2005. Rare earth elements (REE) and Nd and Sr isotopes in groundwater and suspended sediments from the Bathhurst Mining Camp, New Brunswick: water–rock reactions and elemental fractionation. In: Johannesson, K.H. (Ed.), Rare Earth Elements in Groundwater Flow Systems. Springer, Dordrecht, pp. 253–293. Martin, C.E., McCulloch, M.T., 1999. Nd–Sr isotopic and trace element geochemistry of river sediments and soils in a fertilised catchment, New South Wales, Australia. Geochim. Cosmochim. Acta 63, 287–305. Michard, A., Beaucaire, C., Michard, G., 1987. Uranium and REE in CO2 rich-waters from Vals les Bains (France). Geochim. Cosmochim. Acta 51, 901–909. Middelburg, J.J., van der Weijden, C.H., Woittiez, J.R.W., 1988. Chemical processes affecting the mobility of major, minor, and trace elements during weathering of granitic rocks. Chem. Geol. 68, 253–273. Murphy, W.M., Helgeson, H.C., 1987. Thermodynamic and kinetics constraints on reaction rates among minerals and aqueous solutions. III. Activated complexes and the pHdependence of the rates of feldspar, pyroxene, wollastonite and olivine hydrolysis. Geochim. Cosmochim. Acta 51, 3137– 3153. Ne´grel, Ph., 1999. Geochemical study of a granitic area – the Margeride Mountains, France: chemical element behavior and 87Sr/86Sr constraints. Aquat. Geochem. 5, 125–165. Ne´grel, Ph., Deschamps, P., 1996. Natural and anthropogenic budgets of a small watershed in the Massif Central (France): chemical and strontium isotopic characterization in water and sediments. Aquat. Geochem. 2, 1–27. Ne´grel, Ph., Roy, S., 1998. Rain chemistry in the Massif Central (France). A strontium isotopic and major elements study. Appl. Geochem. 13, 941–952. Ne´grel, Ph., Guerrot, C., Cocherie, A., Azaroual, M., Brach, M., Fouillac, C., 2000. Rare earth elements, neodymium and strontium isotopic systematics in mineral waters: evidence from the Massif Central, France. Appl. Geochem. 15, 1345– 1367. Ne´grel, Ph., Casanova, J., Blomqvist, R., 2001a. Nd isotope variation in groundwater and mixing phenomena from Palmottu (Finland). Water Res. 35, 1617–1623. Ne´grel, Ph., Casanova, J., Aranyossy, J.F., 2001b. Strontium isotope systematics used to decipher the origin of groundwaters

1454

P. Ne´grel / Applied Geochemistry 21 (2006) 1432–1454

sampled from granitoids: the Vienne case (France). Chem. Geol. 177, 287–308. Nesbitt, H.W., 1979. Mobility and fractionation of rare earth elements during weathering of a granodiorite. Nature 279, 206–210. ¨ hlander, B., Ingri, J., Land, M., Scho¨berg, H., 2000. Change of O Sm–Nd isotope composition during weathering of till. Geochim. Cosmochim. Acta 64, 813–820. Oliva, P., Viers, J., Dupre´, B., 2003. Chemical weathering in granitic environments. Chem. Geol. 202, 225–256. Petelet-Giraud, E., Ne´grel, Ph., Casanova, J., 2003. Variability of 87 Sr/86Sr in water draining granite revealed after a double correction for atmospheric and anthropogenic inputs. Hydrol. Sci. J. 48, 729–742. Pin, C., Duthou, J.L., 1990. Sources of Hercynian granitoids from the French Massif Central: inferences from Nd isotopes and consequences for crustal evolution. Chem. Geol. 83, 281– 296. Probst, A., El Gh’mari, A., Aubert, D., Fritz, B., McNutt, R., 2000. Strontium as a tracer of weathering processes in a silicate catchment polluted by acid atmospheric inputs, Strengbach, France. Chem. Geol. 170, 203–219. Sanjuan, B., Michard, A., Michard, G., 1988. Influence of the temperature of CO2-rich springs on their aluminium and REE contents. Chem. Geol. 68, 57–67. Sarazin, G., 1979. Ge´ochimie de l’aluminium au cours de l’alte´ration des granites et basaltes sous climat tempe´re´. Doctoral thesis, Univ. Paris VI.

Sequeira Braga, M.A., Paquet, H., Begonha, A., 2002. Weathering of granites in a temperate climate (NW Portugal): granitic saprolites and arenization. Catena 49, 41–56. Simien, F., Mattauer, M., Alle`gre, C.J., 1999. Nd isotopes in the stratigraphic record of the Montagne Noire (French Massif Central): no significant Paleozoic juvenile inputs and preHercynian paleogeography. J. Geol. 107, 87–97. Taunton, A.E., Welch, S.A., Banfield, J.F., 2000. Microbial controls on phosphate and lanthanide distributions during granite weathering and soil formation. Chem. Geol. 169, 371– 382. Tricca, A., Stille, P., Steinman, M., Kiefel, B., Samuel, J., Eikenberg, J., 1999. Rare earth elements and Sr and Nd isotopic compositions of dissolved and suspended loads from small river systems in the Vosges mountains (France), the river Rhine and groundwater. Chem. Geol. 160, 139–158. White, W.M., Patchett, J., 1984. Hf–Nd–Sr isotopes and incompatible element abundances in Island Arcs: implications for magma origin and crust mantle evolution. Earth Planet. Sci. Lett. 67, 167–185. Williamson, B.J., Shaw, A., Downes, H., Thirlwall, M.F., 1996. Geochemical constraints on the genesis of Hercynian twomicas leucogranites from the Massif Central, France. Chem. Geol. 127, 25–42. Zuddas, P., Seimbille, F., Michard, G., 1995. Granite–fluid interaction at near equilibrium conditions: experimental and theoretical constraints from Sr contents and isotopic ratios. Chem. Geol. 121, 145–154.