The soils investigated in this study were selected and sampled during the geological and pedological field survey of the “Saint-jean-Ligoure” diorite massif in Haute-Vienne (Limousin, France), 25 km in the South of Limoges (Figure 1). They are inceptisols where wet sewage sludges were deposited during the 90’s, each year for ten years, at an application rate of 2500 kg・ha⁻¹・year⁻¹ with mean Zn concentration of 498 mg・kg⁻¹. Two soils were sampled on the same dioritic parent-rock: the Amended Soil (AS) that received the sewage sludges for ten years and the Control Soil (CS) that is free of sludge spreading.

The two soils are moderately drained, deep with the weathered rock found at 100 - 150 cm depth and show the typical A, Bw and C horizons sequence of inceptisols (Figure 2(a), Figure 2(b)). The R horizon, i.e. the coherent unweathered parent rock (Figure 2(c)), is a diorite which appears to the naked eye as a “salt and pepper” coarse-grained rock made up of major amphiboles, plagioclases, and minor quartz. The two soils have

very similar physicochemical properties (Figure 2(d)) with slightly acidic to almost neutral pH in all horizons (6.1 - 6.8). The pH values increase with depth, suggesting some leaching of base cations or probably the effect of organic matter with 1.82% to 1.95 % organic carbon content in the A horizons, decreasing to 0.85% - 0.63 % in the Bw horizon and disappearing in the C horizon. The cation exchange capacities of the two soils are similar with medium values ranging between 26.8 and 31.9 cmol・kg⁻¹, whereas the clay contents, typical of the clay loam texture observed in the inceptisols, increase slightly in the Bw horizon (cambic horizon).

Each soil was carefully sampled using plastic cylindrical corers to prevent the soil metal contamination. The cylinder, 10 cm long with a diameter of 10 cm, had cutting edges. It was smoothly and continuously pushed into the soil to collect undisturbed blocks which retained the original soil structure. Each core was divided into two subsamples. The first was disaggregated by gentle shaking in water, sieved to 2 mm, air dried and reserved for the mineral grains separation, the bulk chemical and the X-ray diffraction (XRD) analyses.

The second subsample was preserved undisturbed for thin sections preparation, electron probe microanalyses (EPMA) and scanning electron microscope (SEM) study.

The organic matter in the first <2 mm subsample was removed using 33% hydrogen peroxide. 200 grams of this sample were shaken in water to destroy the aggregates, sieved into the sandy fractions 50 - 100 µm, 100 - 150 µm, 150 - 250 µm, 250 - 500 µm, and dried at 50˚C for 24 h. The observation of each sieved fraction under optical microscopy (OM) revealed that most euhedral monomineralic grains were found in the 250 - 500 µm fraction. The smaller sized fractions were made up of broken primary minerals. About 500 monomineralic grains, 250 to 500 µm in size, of amphiboles and plagioclases (the two most important minerals in the parent-rock and soils) were then needle-sorted under the stereomicroscope for the study of their specific weathering products. Separation efficiency was checked by XRD.

The bulk chemical analyses were obtained from the first subsample, freed of its organic matter. An aliquot quantity of matter (5 g) was collected by quartering with a riffle splitter and crushed down to less than 50 µm size into agate mortar. 300 mg were taken from the total powder, fused with LiBO₂ at 1050˚C and dissolved in 1N HNO₃ for the bulk chemical analyses. The major and trace elements were analysed using ICP-AES and ICP-MS at the Service d’Analyse des Roches et des Minéraux, (SARM), CRPG-CNRS (Vendoeuvre-lès-Nancy, France). The major and trace elements concentrations were expressed in wt.% and mg・kg⁻¹, respectively.

The bulk clay fractions, together with the clay minerals extracted from the weathered monomineralic grains by sonification, were analysed by XRD on Ca-saturated oriented preparations, air dried and glycolated. The diffractometer was a PHILIPS PW 1730 (40 kV, 40 mA) with a Fe-filtered CoKa radiation and a stepping motor driven with a DACO-MP recorder and the Diffrac-AT software (Socabim, Munich, Germany). The bulk clay fraction <2 µm was extracted from 10 g of the <2 mm first subsample. The 10 grams sampled were first dispersed in 350 ml deionized water, sonicated for 2 min at 300 W/20 kHz, then mechanically shaken for 3 h. The stable suspension was then centrifugated at 20˚C and 1000 rpm for 2 mn 30 s with a Jouan JR4.22 centrifuge in order to separate the <2 µm fraction. The effectiveness of this clay separation was then checked using a laser diffraction granulometer IP Malvern Mastersizer.

The second subsample was devoted to the EPMA and the SEM studies. Thin sections were first prepared according to the procedure of Camuti and McGuire [20] . Samples were hardened under vacuum (100 mbars) with an epoxy resin (ARALDITE 2020) mixed with 20% acetone thinner. After polymerization, samples were cut and thin sections obtained through polishing with silicon carbide (17 µm and 9 µm particle sizes) and diamond calibrated powders (6 µm, 3 µm, 1 µm, and 0.25 µm particle sizes).

Representative pedofeatures (clay coatings, clay cutans, etc.) and weathering microsites were first located on the thin sections, marked with black circle under an optical microscope and then “in situ” analysed for major and trace elements. EPMA were obtained using a CAMECA SX 50 electron microprobe (Service CAMPARIS, Université Paris VI) equipped with wavelength-dispersive spectrometers (WDS). The microprobe was calibrated using synthetic and natural oxides. Corrections were made with a ZAF program. A specific trace program with the electron microprobe was developed [21] to analyse heavy metals in small volumes with detection limits closed to 6 - 8 mg・kg⁻¹. Major elements analyses were performed at 15 kV and 4 nA prior to trace element analysis performed at higher voltage (30 kV) and higher beam current (100 nA and 500 nA for small and large minerals, respectively) in order to improve detection limits. The spot sizes varied from 1 to 5 - 10 µm to limit beam damage. The counting time was 10 s per major element and 1000s per trace element. Major and trace elements are expressed respectively in wt% and mg・kg⁻¹.

For each dominant primary rock forming mineral, i.e. amphibole and plagioclase, ten grains were selected from the 250 - 500 µm needle-sorted fraction and affixed to glass plates with double-sided adhesive tape. The two glass plates were then gold-coated for the micromorphological SEM study of the weathering products.

The geochemical mass balances were obtained from the calculation of enrichment factors (EF) using the method of Hernandez et al. [7] and Sterckeman et al. [4] with the equation:

where the content of element X is normalized to the content of the reference element R. A value of 1 for EF indicates no element enrichment or depletion. EF values >1 indicate element enrichment and when <1 indicate element depletion. The aluminium, with its narrow range of variation in the two soils (17.53% to 20.08% in Table 1) was chosen as the reference element. The CS8 sample with the lowest loss on ignition suffered very low weathering and was selected as the reference rock for major elements and Zn bulk analyses.

^aTotal iron expressed as Fe₂O₃. ^bLOI: Loss on ignition.

The two soils are derived from the same dioritic parent rock which outcrops at the bottom of the quarry exposing the control soil profile (R horizon in Figure 2(c)). The diorite appears to the naked eye as a “salt and pepper” coarse-grained rock made up of major dark-green amphiboles, white plagioclases and minor vitreous quartz. The modal rock composition was obtained from the sample CS8, i.e. the unweathered coherent rock with high apparent density (2.92 g・cm⁻³) and the lowest loss on ignition (3.77% in Table 1). The counting of 3000 points on thin section gives the mineralogical composition of 38% amphibole, 32% plagioclase, 12% orthoclase and minor quartz (8%), albite (7%), and titanomagnétite (3%). The EPMA of the major elements in the rock forming minerals show that the amphiboles are calcic hornblendes and plagioclases are andesine with 48% anorthite content. The EPMA of Zn (Table 2) reveal that the amphiboles are the major Zn-bearing minerals with high Zn concentrations of 180 mg・kg⁻¹ versus lower Zn contents in the plagioclases (17 mg・kg⁻¹), orthoclase (20 mg・kg⁻¹) and albite (14 mg・kg⁻¹).

The enrichment factors calculated using the Equation (1) and Table 1 are plotted versus depth in Figure 3. The chemical evolution trends from rock to saprolite and soil appear similar for the major elements in the control and amended soils. The first chemical change observed during weathering is the general increase in EF calculated for the loss on ignition (EF-L.O.I.) from bottom to top of the two soil profiles (Figure 3(a)). EF-L.O.I. increases

^aEach analysis is already the mean of ten measurements. ^bStandard error.

more than twice from the sample reference CS8 (EF-L.O.I. = 1) to the top of the soils (EF-L.O.I. reaches 2.41 in the sample AS2 at the top of the amended soil). This trend reflects the general increase in the bulk water content (loss on ignition at 1050˚C, i.e. the hydrate content in the absence of the CO₂ found in carbonates and after organic matter destruction), and is related to the increase in the hydrated clay minerals crystallizations as weathering proceeds. In the same way as aluminium that was chosen for reference element, the measured SiO₂ and Fe₂O₃ concentrations (Table 1) remain stable throughout the two soil profiles and give EF values close to 1 (0.91 - 1.12 for EF-SiO₂ and 0.93 - 1.08 for EF-Fe₂O₃). Conversely, the alkaline earth elements Mg and Ca decrease regularly from the bottom to the top of the two soils: The EF-MgO and EF-CaO reach respectively 0.69 and 0.34 in the sample AS2 at the top of the amended soil (Figure 3(b)). This regular and general decrease in the two soils is indicative of the weathering processes which affect the amphiboles (EF-MgO and EF-CaO) and the plagioclases (EF-CaO) in the saprolite and soil.

The chemical analyses of Zn in the bulk soil horizons (Table 1) reveal that, although the sewage sludges spread on the amended soil were heavily loaded with Zn (498 mg・kg⁻¹ as mean concentration), its concentration in the two soils never exceeds the threshold value for anthropogenic contamination (300 mg・kg⁻¹, [22] ). The Zn enrichment factors, when calculated from Table 1 and plotted versus depth in Figure 3(c), reveals two different trends depending on whether the control soil or the amended soil is considered: The distribution of EF-Zn in the control soil does not show any significant variation throughout the soil profile, whereas EFs in the amended soil indicate pronounced depletion of Zn in the surface horizons that may suggest migration of this metal to depth (down to 190 cm, sample AS7). This Zn migration to depth has been already observed by Scokart et al. [23] .

The clay minerals identified in the <2 µm granulometric fractions are similar in the control and amended soils. The XRD on Ca-oriented preparations reveals the typical 001 reflections of smectite, at 15.10 Å in air-dried state and shifting to 17.05 Å after ethylene-glycol solvation. The smectite is associated with a kaolinite which is typified, in air-dried and ethylene-glycol solvated states, by 001 and 002 reflections at 7.15 Å and 3.55 Å respectively. These reflections disappear after 450˚C heating. No reflection of iron oxide or hydroxide was detected on the XRD patterns of the random powders.

The Zn contents in the soil clay fractions were measured using the electron microprobe and are given with their mean standard errors in Table 3. The EPMA are plotted versus depth in Figure 4(a) to follow the Zn distribution in the two soil profiles. It appears in Table 3 that the Zn concentrations in the bulk clay fractions of the amended soil range from 144 to 203 mg・kg⁻¹ and are always higher than in the control soil where they range from 10 to 24 mg・kg⁻¹. The shape of the Zn distribution profile in the amended soil (Figure 4(a)) is typically that of a downward migrating element, with Zn concentrations about ten times that measured in the control soil and a regular decrease from the top to the bottom of the amended soil. The highest Zn concentrations are measured in the surface horizons, from 195 to 203 mg・kg⁻¹ at 20 cm (sample AS1) and 40 cm (sample AS2) depths, decreasing to 150 mg・kg⁻¹ at the bottom of the soil (190 cm depth, sample (AS7)). For comparison, the Zn distribution profile of the control soil does not show significant variations when standard errors are considered

^aEach analysis is already the mean of ten measurements. ^bStandard error.

(Figure 4(a)). The Zn concentrations are much lower than those measured in the amended soil, and range from 10 to 24 mg・kg⁻¹ at 20 cm (sample CS1) and 40 cm (sample CS2) depths, and from 14 to 15 mg・kg⁻¹ at 180 cm (sample CS6) and 485 cm (sample CS8) depths.

The first signs of amphiboles weathering are found in the C horizons of the two soils. The OM observation of the thin sections reveals that euhedral amphibole prismatic crystals begin to weather along the enlarged cleavage planes (Figure 5(a)) where SEM shows the crystallization of clay minerals with the typical honeycomb texture of the smectite (Figure 5(b)). These characteristic features are commonly observed in the weathered rocks and soils [24] [25] . The weathering intensity increases in the Bw horizons where amphibole crystals appears highly fragmented into 20 - 100 µm organized residues by the opening of the intramineral microcracks, up to 50 µm wide (Figure 5(c)). The SEM reveals that the clay minerals filling these microcracks exhibit the same honeycomb morphology of smectite clay minerals, already observed in the C horizons; they are covering the saw-tooth terminations of the dissolving amphibole crystals (Figure 5(d)). The most advanced weathering stages are found in the surface A horizons where the amphiboles crystals are almost entirely replaced by a clayey weathering plasma (Figure 5(e)) made up of smectites with the characteristic honeycomb texture (Figure 5(f)).

The clay minerals, extracted from the amphiboles grains by sonication, were identified by XRD. The XRD patterns are similar in the C, Bw, and A horizons with the same typical reflections of smectite. The Ca-oriented preparations reveals strong 001 reflections ranging from 14.50 Å to 15.49 Å in air-dried state, shifting to 16.87 - 17.14 Å with higher orders at 8.46 Å and 3.40 Å after ethylene-glycol solvation. This expansion to 17 Å with ethylene-glycol is typical of smectite layers.

The EPMA of the amphibole weathering products characterize two chemically distinct smectite populations. The first is detected in the C horizons of soils and is made up of magnesian smectites belonging to the trioctahedral group of saponites. The second population is found in more advanced weathering stages (Bw and A horizons) as aluminous smectites belonging to the dioctahedral smectite group of montmorillonites.

The Zn contents in the clay minerals produced by the weathering of amphiboles were measured using the electron microprobe and are given in Table 4 with their mean standard errors. The EPMA are plotted versus depth in Figure 4(b) to follow the behaviour of Zn in the two soil profiles. The early weathering of amphiboles into saponites (R horizon, samples CS8 and CS7) results into the partial release of Zn, from 180 mg・kg⁻¹ in the

fresh amphibole (Table 2) to 76 - 85 mg・kg⁻¹ in the saponites. Higher weathering levels in the C, Bw, and A horizons produce different Zn distributions depending on whether the control or the amended soil is considered.

When taking into account the mean standard errors, the Zn concentrations in saponites and montmorillonites appear almost constant throughout the control soil profile (Figure 4(b)). The behaviour of Zn in the amended soil appears much more complicated. Initially, the Zn concentrations measured in the saponites derived from amphiboles are very similar in the C horizons of the amended and control soil profiles: 72 to 82 mg・kg⁻¹ in AS7, AS6, AS5 versus 72 to 81 mg・kg⁻¹ in CS6 and CS5 (Table 4, Figure 4(b)). In a second stage, the Zn concentrations measured in the montmorillonites from the Bw and A horizons of the amended soil differ strongly from that in the control soil. The concentrations increase in the amended soil from 105 to 220 mg・kg⁻¹ in AS4 and AS1 versus 68 to 82 mg・kg⁻¹ in CS4 and CS1 (Table 4, Figure 4(b)).