| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 45 | Number 2 | Pages 299-320 | 2004
© Oxford University Press 2004; all rights reserved
Silicate, Hydrous and Carbonate Metasomatism at Lherz, France: Contemporaneous Derivatives of Silicate Melt–Harzburgite Reaction
1 LABORATOIRE DE TECTONOPHYSIQUE, ISTEEM, CASE 49, CNRS ET UNIVERSITÉ DE MONTPELLIER 2, PLACE EUGÈNE BATAILLON, 34095 MONTPELLIER CEDEX 05, FRANCE
2 DEPARTMENT OF GEOLOGY, ROYAL HOLLOWAY, UNIVERSITY OF LONDON, EGHAM HILL, EGHAM TW20 0EX, UK
3 DEPARTMENT OF GEOLOGY AND GEOPHYSICS, WOODS HOLE OCEANOGRAPHIC INSTITUTION, WOODS HOLE, MA 02543, USA
4 DEPARTMENT OF EARTH, ATMOSPHERIC AND PLANETARY SCIENCES, MASSACHUSETTS INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MA 02139, USA
RECEIVED NOVEMBER 18, 2002; ACCEPTED SEPTEMBER 5, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
Complex multi-stage models involving silicate, hydrous and carbonate melts of distinct provenance have been invoked to explain the metasomatism observed in mantle rocks. In contrast, relatively simple models requiring polybaric crystallization of alkaline silicate melts have been proposed to explain the occurrence of veined mantle rocks. To address the spatial and temporal relationships between veins and wall-rocks, a sequence of drill cores was obtained from Lherz, France. In outcrop the vein (amphibole–garnet pyroxenite dyke) is spatially associated with hornblendite veinlets (lherzite), and proximal amphibole-bearing and distal apatite-bearing wall-rock peridotite. Considerable elemental and isotopic heterogeneity exists in these wall-rock peridotites, in many instances equivalent to, or greater than, that observed in mantle xenoliths from worldwide localities. A single stage of reactive porous flow best explains the elemental and isotopic heterogeneity in the wall-rock. In essence it is proposed that emplacement of the silicate melt (dyke) was inextricably linked to chromatographic fractionation/reaction of derivatives which led to the coexistence, in space and time, of silicate, hydrous and carbonate melts. This model elegantly and simply describes the formation of complex metasomatic aureoles around mantle veins and negates the need, in the case of basalt-hosted (and kimberlite-hosted) xenoliths, for complex multi-stage models involving several episodes of melt influx with each melt being of different provenance.
KEY WORDS: mantle metasomatism; trace-element enrichment; isotopic contamination; wall-rock peridotite; Lherz peridotite
|
INTRODUCTION AND AIMS |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
The Lherz massif, Etang de Lers in the French Pyrenees, is the type locality for lherzolite first described by Le Lievre in 1787, who noted the olivine-rich character of the massif and the similarity to peridotite xenoliths entrained in volcanic rocks in the Massif Central. De La Metherie coined the term ‘lherzolite’. Over 200 years later orogenic peridotites, and especially the Lherz massif, are pivotal to our understanding of the temporal and spatial aspects of melt transfer through mantle rocks. This massif has been the focus of many studies of melt transfer processes within mantle rocks and in recent years has been the source of debate concerning mantle processes (Polvé & Allègre, 1980
; Wilshire et al., 1980; Loubet & Allègre, 1982; Bodinier et al., 1987a, 1987b, 1988, 1990; Hamelin & Allègre, 1988; Vétil et al., 1988; Downes et al., 1991; Fabriès et al., 1991; Mukasa et al., 1991; Nielson & Wilshire, 1993; McPherson et al., 1996; Pattou et al., 1996; Woodland et al., 1996; Zanetti et al., 1996; Lorand et al., 1999).
In the Lherz orogenic peridotite, Bodinier et al. (1990) studied a 65 cm section of harzburgite wall-rock perpendicular to an amphibole–garnet pyroxenite dyke. Their results revealed significant mineralogical and chemical zonation of the wall-rock. At less than 
25 cm from the dyke–wall-rock boundary, hydrous amphibole-bearing peridotite displays patterns of decreasing concentrations of Fe, Ti, Mn, Al, Ca, Na and heavy rare earth elements (HREE), and increasing concentrations of Mg and Ni away from the dyke. In this zone, the chondrite-normalized REE patterns are convex-upward (0–15 cm) or U-shaped (15–25 cm) and relatively unfractionated (LaN/YbN 1). At >25 cm from the dyke, anhydrous harzburgite displays a selective light REE (LREE) enrichment and Woodland et al. (1996) noted the presence of apatite in this zone. The normalized REE patterns vary from U-shaped at 25–30 cm (LaN/YbN 1·7–2·6) to strongly LREE-enriched at 30–65 cm (LaN/YbN 3–16). This variation encompasses almost the whole range of LREE/HREE ratios in metasomatized mantle xenoliths from alkali basalts. Bodinier et al. (1990) explained wall-rock zonation as a single-stage process involving progressive chemical evolution of melts infiltrated from the dyke into the harzburgite wall-rock. In this model, the compatible and moderately incompatible elements are buffered by the wall-rock peridotite proximal to the dyke (<20 cm) whereas distal to the dyke the most incompatible elements remain controlled by the infiltrated melt. This mechanism results in a chromatographic fractionation of REE and accounts for the high LREE/HREE ratios observed distal to the dyke (see also Zanetti et al., 1996).
However, Nielson & Wilshire (1993) criticized this interpretation. In particular, they questioned the ability of mantle melts to percolate through lithospheric peridotites over distances greater than a few centimetres. This negated the very existence of chromatographic fronts in the upper mantle (Navon & Stolper, 1987). They suggested an interpretation for the Lherz wall-rock that involved multi-stage metasomatism with a requirement for vein conduits not seen in the peridotite block. At first sight, this interpretation appears to be supported by the isotopic data of Downes et al. (1991) in three samples from the Lherz wall-rock. These data indicate that the isotopic signature of the dyke does not extend over more than 20 cm into the wall-rock; yet the strong LREE enrichment typical of the host peridotite at >25 cm from the dyke has never been observed in peridotites unrelated to amphibole pyroxenite dykes, neither in Lherz nor in other Pyrenean peridotites (Bodinier et al., 1988; McPherson et al., 1996; Burnham et al., 1998; Fabriès et al., 1998). On the other hand, a similar composition has been found in harzburgites from the small Montau massif (western Pyrenees), which contain abundant and massive (metres thick) amphibole–feldspar segregates (J.-L. Bodinier, unpublished data, 2003).
To address some of these issues and the conflicting views, a new section was drilled in the Lherz wall-rock studied by Bodinier et al. (1990). Woodland et al. (1996) reported on the petrology and oxygen fugacity of these samples. Herein we report on a comprehensive geochemical study using data from several sources including X-ray fluorescence (XRF), inductively coupled plasma mass spectrometry (ICP-MS), secondary ion mass spectrometry (SIMS) and thermal ionization mass spectrometry (TIMS).
We are particularly interested in addressing several questions, as follows.
- What process best explains the mineralogical and chemical variability in the section from the dyke to proximal amphibole-bearing and distal anhydrous, apatite-bearing peridotites?
- Are we dealing with single- or multi-stage processes, and were the processes synchronous in time and space?
- If trace-element enrichment occurred during a single stage, can we reproduce with modelling the ultra-enrichment of LREE within the anhydrous peridotite (distal to the dyke) while keeping the isotopic contamination from the dyke restricted to the amphibole peridotite (proximal to the dyke)?
|
WALL-ROCK DESCRIPTION AND SAMPLING |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
The studied wall-rock is from a block of harzburgite that contains a dyke of amphibole–garnet pyroxenite, about 15 cm thick, and a complex swarm of bifurcating and anastomosing veins of ‘lherzite’ (= hornblendite). The lherzites are restricted to the wall-rock proximal to the main dyke (<25 cm). From this distance to the edge of the block (
80 cm from the dyke–wall-rock boundary), the harzburgite is devoid of dykes or veins (Woodland et al., 1996, fig. 1). The dyke and veins cross-cut the harzburgite at a high angle to the foliation. They belong to a suite of hydrous pyroxenites considered as high-pressure–high-temperature crystal segregates from alkaline basalts that migrated through the Pyrenean lithospheric mantle in the Cretaceous (Loubet & Allègre, 1982; Goldberg et al., 1986; Bodinier et al., 1987a; Vétil et al., 1988; Fabriès et al., 1991; Rossy et al., 1992). The conditions of emplacement of the dykes were estimated by Vétil et al. to be 1·0–1·5 GPa and 900–1000°C. Cross-cutting and overprinting relations indicate that the lherzite veins were emplaced at a slightly later stage than the pyroxenite dykes and were probably coeval with poikiloblastic amphibole developed at the expense of anhydrous primary minerals (pyroxene + garnet). The lherzites and the amphibole in the dykes are ascribed to the segregation of volatile-rich, evolved melts after crystallization of anhydrous garnet pyroxenites from the alkaline basalt (Wilshire et al., 1980; Bodinier et al., 1988a; Woodland et al., 1996).
Woodland et al. (1996) provided a detailed mineralogical description of the wall-rock samples collected for this study. The harzburgite wall-rock can be subdivided into two zones based upon textural, mineralogical and geochemical data: (1) an inner zone <25 cm from the main pyroxenite dyke and (2) an outer zone >25 cm from the main dyke (Bodinier et al., 1990). This boundary is basically coincident with the most distant lherzite vein from the main dyke. The harzburgite in the inner zone has a well-developed porphyroclastic texture and contains disseminated brown amphibole and rare phlogopite, testifying to its modally metasomatized character. The amphiboles, sometimes replacing primary orthopyroxene, have compositions similar to those found in modally metasomatized spinel peridotite xenoliths (e.g. Menzies et al., 1987). Several narrow (2–3 mm) irregularly bounded veinlets of amphibole and phlogopite occur sub-parallel to the lherzite veins in this zone. The outer zone, beyond the lherzite veins (>25 cm), has a granular to porphyroclastic texture and is essentially free of amphibole. The wall-rock in this zone appears to be texturally and mineralogically indistinguishable from ‘normal’ harzburgites from Lherz that are spatially unrelated to dykes. However, Woodland et al. (1996) identified rare grains of subhedral apatite in a number of samples from the outer zone, especially in the most distal part of the wall-rock (>50 cm from the dyke).
Thirty-one samples were drilled along a section perpendicular to the main dyke and veinlets, about 60 cm apart from the section studied by Bodinier et al. (1990). Among these samples, five were collected in the wall-rock of the main dyke opposite to that studied by Bodinier et al. (Woodland et al., 1996, fig. 1). This wall-rock contains a subordinate amphibole–garnet pyroxenite dykelet and lherzite veins. It is comparable with the amphibole harzburgite zone defined in the main wall-rock with regard to texture, mineralogy and geochemistry (McPherson, 1994; Woodland et al., 1996). These five samples will not be further considered in the present study. Among the other samples, four were drilled in the main dyke (samples EDL1, 2, 14 and 15), nine are from the amphibole harzburgite wall-rock (<25 cm from the dyke: samples EDL16, 17, 18, 19, 20, 22, 24, 25 and 27) among which sample EDL22 includes a lherzite veinlet, and 13 are from the distal harzburgite wall-rock (>25 cm: samples EDL28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39 and 40).
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
The samples were analysed as bulk rocks by XRF for major and several trace elements using the University of London XRF facility at Royal Holloway (RHUL). In addition, Rb, Sr, Y, Zr, Nb, Ba and Y were determined by quadrupole ICP-MS in selected samples at the NERC ICP-MS facility, RHUL. REE (La, Ce, Nd, Sm, Eu, Gd, Dy, Er, Yb and Lu) were determined by isotope dilution after separation, following the method of Thirlwall (1982)
modified for the VG354 multicollector mass spectrometer of the radiogenic isotope laboratory at RHUL. Sr and Nd isotopes were also run on bulk rocks as it was impossible to separate sufficient clinopyroxene from the harzburgite cores, because of the small sample size. To remove secondary grain boundary components that may affect Sr isotopes, rock powders were leached in screwtop capsules at 180°C for 2 h in 6 M HCl. Confirmation that the leaching technique was effectively removing secondary minerals was sought by analysing fine sieve fractions from four xenoliths previously analysed using leached clinopyroxene (Downes et al., 1992). It is also worth noting the consistency of our results with the Sr isotope data previously obtained by Downes et al. (1991) in three samples of leached clinopyroxene from the same wall-rock. These clinopyroxenes were separated from the whole-rock samples studied by Bodinier et al. (1990).
The isotope ratios were measured with multidynamic techniques (Thirlwall, 1991) on the VG354 at RHUL. Details of the analytical procedures have been given by McPherson et al. (1996), who also reported on the reproducibility and accuracy of trace-element and isotope analyses. Major-, trace-element and some Sr–Nd isotope data are given in Table 1 for 22 drilled cores.
|
In addition to the analyses performed in the new drilled section, a selection of 13 bulk-rock samples previously studied by Bodinier et al. (1990)
were re-analysed for trace elements (Rb, Sr, Y, Zr, Nb, Cs, Ba, REE, Hf, Ta, Pb, Th and U) by ICP-MS at ISTEEM, University of Montpellier. We followed the procedures described by Ionov et al. (1992) and Godard et al. (2000), who reported the accuracy and reproducibility of the method based on repeated analyses of international peridotite standards. The results are given in Table 2.
|
Finally, clinopyroxenes and a few amphiboles from the main dyke and from 12 wall-rock samples were analysed in thin section by ion microprobe for Ti, V, Sr, Cr, Zr, Y and REE (La, Ce, Nd, Sm, Eu, Dy, Er and Yb). The analyses were performed by SIMS on a Cameca IMS 3f system at the Woods Hole Oceanographic Institution, following the procedure described by Shimizu & Hart (1982)
. Previous studies have reported reproducibility of clinopyroxene analyses obtained with this procedure (e.g. Takazawa et al., 1996). SIMS data are given in Table 3.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
As illustrated in Fig. 1 for Ti and Ce, the wall-rock section sampled for this study compares favourably with that studied by Bodinier et al. (1990)
with regard to major-, minor- and trace-element variations in bulk rocks. This indicates that the metasomatic zones defined by Bodinier et al. show some continuity along the dyke, at least over the distance (60 cm) between the two sections. At less than 20–25 cm from the dyke–wall-rock boundary, the amphibole peridotite displays decreasing concentrations of Fe, Ti, Mn, Al, Ca, Na and HREE, and increasing concentrations of Mg and Ni away from the dyke (Table 1). Further in the wall-rock, these elements show nearly constant concentrations in the amphibole-free harzburgite. The Al, Ca, Na and HREE contents are low, consistent with the refractory character of the wall-rock (e.g. Al2O3 mostly <1% and HREE <0·1 x chondritic abundances). In contrast to the moderately incompatible elements, such as Ti and HREE, more incompatible elements, such as LREE and P, show increasing abundances from the dyke border throughout the wall-rock (0–70 cm, Fig. 1). The highest phosphorus content (>0·045% P2O5 at >50 cm from the dyke) is found in samples in which apatite was positively identified by Woodland et al. (1996).
|
, amphibole-free harzburgite. The samples in which apatite was positively identified by Woodland et al. (1996)The ICP-MS data for the section previously studied by Bodinier et al. (1990)
confirm the strong variability of trace elements that had been already demonstrated for REE (Table 2 and Fig. 2). The amphibole harzburgite shows flat, or slightly convex-upward, primitive mantle-normalized trace-element patterns. The trace-element patterns in the wall-rock mimic that of the dyke at lower concentration, except for higher Th and U relative to LREE and the lack of a positive Nb–Ta anomaly. The LaN/YbN ratio is in the range 0·8–1·3 for the amphibole harzburgite (<20 cm). Further in the wall-rock, the amphibole-free harzburgite is characterized by a strong enrichment in LREE relative to HREE (LaN/YbN = 7·1–16·6). Th and U are enriched in comparison with the amphibole harzburgite, although at a lesser degree than the LREE. Similarly, Sr tends to be less enriched than the LREE whereas Ba is strongly depleted in the amphibole-free harzburgite zone. The high field strength elements (HFSE: Nb, Ta, Zr and Hf in Fig. 2) show negative anomalies relative to REE and Th–U. LREE-enriched, HFSE-depleted trace-element patterns have been frequently reported from mantle xenoliths, where they are generally ascribed to carbonate metasomatism (Dautria et al., 1992; Hauri et al., 1993; Ionov et al., 1993; Rudnick et al., 1993). The resemblance to carbonate-melt metasomatism is enhanced by the presence of apatite (Woodland et al., 1996).
|
Two representative harzburgites from Lherz were also analysed by ICP-MS at Montpellier. As far as we can tell, these samples are spatially unrelated to dykes and veins. Like many mantle harzburgites worldwide (e.g. Bodinier & Godard, 2003
), they display U-shaped normalized trace-element patterns. In contrast to the wall-rock harzburgite, they are depleted in the LREE and lack negative HFSE anomalies (Fig. 2). In agreement with the bulk-rock trace-element data, the ion microprobe analyses (Table 3 and Fig. 3) show a progressive increase in the LREE/HREE ratio in cpx away from the dyke, with a maximum LaN/YbN ratio in the range 14–18 at 30–60 cm distance. This variation is merely a reflection of a significant increase in the concentration of the LREE in cpx. Chondrite-normalized La, Ce and Nd contents increase from 10–30 x chondrite proximal to the dyke (i.e. <20 cm), to 40–110 x chondrite distal to the dyke (i.e. >30 cm). Conversely, Zr and Ti in cpx (and to a lesser degree HREE) decrease markedly and reach a minimum ZrN = 3–5 distal to the dyke (30–60 cm) compared with 20–50 proximal to the dyke (<20 cm). These variations result in markedly different chondrite-normalized trace-element patterns for cpx. In the amphibole peridotite (<20 cm from the dyke) the cpx patterns are mildly fractionated, slightly convex and smooth in shape, except for a subtle positive anomaly of Zr (ZrN/SmN = 1·3–2·3). In contrast, cpx from the amphibole-free peridotite (>30 cm) has strongly fractionated (LREE-enriched) trace-element patterns with negative Zr and Ti anomalies. At 30–60 cm from the dyke, the ZrN/SmN and TiN/DyN ratios in cpx are systematically lower than 0·1. On the other hand, these anomalies at not present at >60 cm, where the HREE tend to increase. In the most distal samples (60–80 cm), the trace-element patterns of cpx show a steady increase from HREE to LREE without significant HFSE anomalies. The strong LREE enrichment of cpx in the anhydrous harzburgite rules out chemical equilibrium with the parental melt of the dyke. Coupled with the negative Zr and Ti anomalies, this enrichment suggests equilibrium with carbonate melt.
|
With the exception of Zr, the trace-element composition of the amphiboles (Table 3) mimics that of cpx from the same samples, but the concentrations are generally higher (Kamph/cpx = 0·8–1·5 for Y, V and Cr, 1·5–2·5 for HREE, 2–4 for LREE, and 3–4 for Sr and Ti). Zr is depleted in amphibole relative to cpx (Kamph/cpx = 0·5–0·6). These values are consistent with the amphibole/cpx relationships observed in other mantle rocks and experimental runs (e.g. Witt-Eickschen & Harte, 1994
; Ionov & Hofmann, 1995; Tiepolo et al., 2001), indicating that inter-mineral chemical equilibrium was achieved in the amphibole harzburgite wall-rock.
The contrasted geochemical signatures of the spatially juxtaposed amphibole and apatite peridotites are also emphasized by their Nd–Sr isotope compositions (Table 1 and Fig. 4a). Proximal to the dyke–wall-rock contact (<10 cm), the wall-rock has an isotopic signature (
Nd = 6–7; Sr = -20 to -25) comparable with that of the dyke itself and other amphibole pyroxenites in Lherz (Downes et al., 1991; Mukasa et al., 1991; McPherson et al., 1996). Distal to the dyke, the isotopic composition evolves towards more enriched values, reaching Nd = -5 and Sr = +10 at >20 cm. The low 143Nd/144Nd of the anhydrous harzburgite requires time-integration of low Sm/Nd ratios (i.e. LREE enrichment) over periods of 900–1100 Myr (DMM model age and dyke–wall-rock ‘isochron’ age, respectively) and, as such, cannot have formed in the time frame of dyke emplacement in the Cretaceous (100 Myr; Goldberg et al., 1986). Insufficient time has elapsed since the formation of the dyke to generate the isotopic variability by time-integration of the LREE enrichment. On the other hand, in Lherz and other orogenic lherzolites, the peridotites unrelated to dykes and pyroxenite layers show a negative correlation between Nd–Sr isotopic enrichment and ‘fertility’ (Fig. 4b; Downes et al., 1991; Reisberg & Lorand, 1995; Bodinier & Godard, 2003). The enriched isotopic signature of the distal anhydrous peridotite is consistent with the refractory, harzburgitic composition of the wall-rock and is therefore considered to represent the isotopic signature of the wall-rock protolith before dyke emplacement. The correlation of the 143Nd/144Nd ratio vs MgO suggests that the isotopic enrichment of the harzburgite protolith is somehow linked to the formation of the tabular harzburgites. Conversely, the proximal samples collected at <20 cm from the dyke plot away from the correlation and were clearly contaminated during dyke emplacement.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
Trace-element variations somewhat comparable with those observed in the Lherz wall-rock have been reported from several suites of mantle xenoliths, in several instances amphibole- and apatite-bearing peridotites (e.g. Hauri et al., 1993
; Bedini et al., 1997; Zangana et al., 1997; Baker et al., 1998; Xu et al., 1998; Ionov et al., 2002a). Bedini et al. (1997) observed similar variations in xenoliths from the East African Rift, which correlate with textural/mineralogical features and equilibrium temperatures. In mantle xenoliths, flat trace-element patterns (i.e. abundance normalized to primitive mantle) are generally observed in relatively high-T, apatite-free samples whereas the large ion lithophile element (LILE)-enriched, HFSE-depleted signature is typical of lower-T, apatite-bearing samples. Bedini et al. (1997) ascribed this variation to a kilometre-scale, single-stage process involving gradual solidification of melt infiltrated into the lithospheric mantle through successive melt–rock reactions down a conductive, thermal gradient. This process would be associated with strong chemical evolution of the melt, from an originally basaltic composition to progressively volatile-enriched compositions, culminating in low-T, small-volume carbonate melts. Harte et al. (1993) suggested that most metasomatic agents in mantle rocks may be formed by vein crystallization and ‘percolative fractional crystallization’ of ocean island basalt (OIB)-type primary melts. Similarly, Baker et al. (1998) attributed the amphibole–apatite-bearing peridotites at Ataq, Yemen, to interaction between lithosphere and small-volume melts derived from the Afar plume partial melts.
The reactive porous flow model is supported by trace-element modelling (Bedini et al., 1997; Vernières et al., 1997). However, the ultra-enriched LREE and Th–U compositions (e.g. 
100 x chondrite values for LREE in cpx) are well above the content expected for equilibrium with basaltic melt. This can be explained by the coupled effects of melt movement and reactions at decreasing melt mass, as also suggested by Harte et al. (1993). In contrast, HREE and compatible elements would be buffered by the peridotite minerals (olivine, opx, cpx), and the HFSE and Rb–Ba by very small amounts of Ti-oxides or amphibole, and phlogopite, respectively. Ionov et al. (2002a) used a similar approach to explain the coexistence of mildly LREE-enriched, amphibole-bearing peridotites and strongly LREE-enriched peridotites containing only sparse amphibole in a suite of mantle xenoliths from Spitsbergen. Such modelling can be applied to the Lherz wall-rock to account for the observed trace-element variation with a single-stage process. However, in Lherz the distance of percolation would be shorter than that envisioned in previous models, especially by Bedini et al. (1997). On the other hand, because the host peridotites were at a temperature of about 900–1000°C during dyke emplacement (Vétil et al., 1988; Fabriès et al., 1991), the wall-rock probably experienced a similar temperature gradient to that envisioned by Bedini et al. (i.e. a few hundred degrees).
The aim of modelling applied to the Lherz wall-rock is to reproduce the ultra-enrichment of LREE within the distal apatite-bearing peridotite while keeping the isotopic contamination from the dyke restricted to the proximal amphibole peridotite. The model assumes distinct origins and timing for chemical (LREE) and isotopic (Sr–Nd) enrichments in the anhydrous harzburgite. The LREE enrichment is considered to be related to dyke emplacement in the Cretaceous whereas the isotopic enrichment (low 143Nd/144Nd–high 87Sr/86Sr) is a signature of the wall-rock protolith. We shall thereafter discuss alternatives to this model, which consider that LREE enrichment occurred before dyke injection in the wall-rock protolith, being linked with isotopic enrichment.
|
THEORETICAL MODELLING |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
Previous studies used porous-flow models to explain trace-element variations in mantle xenoliths similar to those observed in the Lherz wall-rock (e.g. Bedini et al., 1997
; Ionov et al., 2002a). In the present study, we focused on reproducing bulk-rock and cpx Nd abundances and 143Nd/144Nd variations in the wall-rock. The following terms will be used to describe the various aspects of dyke–wall-rock reaction. Wall-rock protolith. Pre-existent harzburgite with a Nd content and 143Nd/144Nd ratio equivalent to the harzburgites spatially unrelated to dykes and veins in Lherz.
Dyke. Garnet–clinopyroxene assemblage formed by crystallization of a basaltic melt at high P and T.
Infiltrated zone. Wall-rock infiltrated by melt from the main dyke or volatile-rich derivatives where the protolith composition was modified by diffusional re-equilibration with infiltrated melt and/or percolation-reaction mechanisms.
Amphibole harzburgite reaction zone. Zone of modified wall-rock proximal (<20 cm) to the main dyke, where amphibole (±cpx) is precipitated in sheared peridotite. The Nd content and 143Nd/144Nd ratio are mainly controlled by variable degrees of re-equilibration with infiltrated melt.
Apatite harzburgite reaction zone. Zone of modified wall-rock distal (>20 cm) to the main dyke, where apatite is precipitated in less sheared peridotite. In this zone, Nd abundance and 143Nd/144Nd ratio are governed by the coupled effects of melt transport and amphibole precipitation. Consequently, the abundance of Nd is relatively high, but the 143Nd/144Nd ratio remains close to the protolith value.
Background and assumptions
The variation of 143Nd/144Nd in the Lherz wall-rock (Fig. 4a) can be simulated with a one-dimensional (1D) chromatographic model developed by Vasseur et al. (1991), which has been recently applied to isotope systems (Ionov et al., 2002b). The main assumptions in this model are explained elsewhere (Bodinier et al., 1990); some additional assumptions for the modelling of isotopes include the following.
Composition of the peridotite protolith. The Nd content of the pre-existent wall-rock protolith was fixed from the composition of harzburgites unrelated to dykes and veins in Lherz (Fig. 2). Based on the 143Nd/144Nd vs MgO correlation in Lherz peridotites (Fig. 4b), the 143Nd/144Nd ratio of the protolith is assumed to be that of the wall-rock at 30–50 cm from the dyke.
Composition of the infiltrated melt. The melt infiltrated in the wall-rock was probably evolved as a result of the precipitation of anhydrous minerals (pyroxene and garnet) in the dyke. This early derivative was very probably similar to the parental melt of the lherzites, suggesting that it was enriched in volatiles (water) and LILE compared with the parental melt of the dyke. However, the lherzites are virtually indistinguishable from the amphibole from the dykes in term of REE (Bodinier et al., 1987a). Moreover, REE partitioning between amphibole and cpx in the dyke is consistent with published values (Witt-Eickschen & Harte, 1994; Ionov & Hofmann, 1995; Tiepolo et al., 2001). This suggests that melt evolution as a result of mineral segregation in the dyke was associated with only minor LREE enrichment. Alternatively, the anhydrous minerals were re-equilibrated with the parental melt of the lherzites owing to pervasive percolation of the water-rich derivative in the dyke itself. This second alternative was suggested by Harte et al. (1993) for composite mantle xenoliths. It is supported by textural evidence for the replacive (metasomatic) character of the amphibole in the dyke. In both hypotheses, the composition of cpx in dyke can be used to establish the Nd content of the infiltrated melt. Its 143Nd/144Nd ratio is assumed to be that of the dyke.
Partition coefficients. The mineral/melt partition coefficients (Kd) of Nd are considered to be constant (although the ‘plate model’, for instance, allows variation of this parameter). The chemical evolution of melt in the wall-rock towards water-rich and/or carbonatitic compositions probably resulted in variations of the Kd values for trace elements. However, the variations are expected to be moderate for REE compared with other trace elements such as LILE and HFSE. Moreover, the exact composition of melt derivatives in the wall-rock is poorly constrained. Therefore, our approach was to try to reproduce the Nd variations with a simple model, assuming constant Kd values, and to modify the Kd values only if the simple approach proved to be unsuccessful.
Age corrections. Age corrections are moderate (
10%) because of the relatively young age of dyke injection and emplacement of the massif (100 Ma). Uncorrected 143Nd/144Nd ratios were used for modelling.
All simulations indicate the existence of a chromatographic Nd isotopic front at some distance behind a melt infiltration front (Fig. 5). Behind the Nd isotopic front, the isotopic composition of the harzburgite protolith has re-equilibrated with the infiltrated melt. In contrast, ahead of the front, the percolating melt has re-equilibrated with the harzburgite protolith. Nevertheless, it should be noted that the chromatographic front does not constitute a sharp boundary between these two domains. Rather it is a transition zone, 5–10 cm wide. The width of the chromatographic front is governed by element diffusivity in peridotitic minerals and interstitial melt. Wider fronts result either from slow diffusivity in minerals and/or efficient diffusivity in interstitial melt (Bodinier et al., 1990; Vasseur et al., 1991). Consistent with chromatographic theory (Navon & Stolper, 1987), our results indicate that the distance of the Nd chromatographic front behind the melt infiltration front depends on two main parameters: (1) bulk peridotite/melt Nd partition coefficient (D) and (2) melt fraction. Figure 5 illustrates the effect of these parameters. For a refractory mineralogy (i.e. 1% cpx, 0% amphibole; low D value) and a relatively high melt fraction (1%), the Nd front is close to the percolation front. To account for the existence of a chromatographic front at 20 cm from the dyke–wall-rock contact, the distance of melt percolation should not exceed >25 cm away from the contact. However, the presence of a few per cent amphibole, as is observed in the wall-rock, and/or melt fraction values significantly smaller than 1% result in a considerable change in the location of the Nd front relative to the melt front. A ‘best fit’ of the 143Nd/144Nd ratios with this model is shown in Fig. 6a. This was obtained with the following set of parameters: distance of the melt front from the melt source (dyke) 1 m, melt fraction 0·1% and amphibole fraction 4%.
|
|
(porosity, = melt fraction) = 0·1%, modal composition involving 4% amphibole. The other parameter values are given in Non-reactive models (i.e. no melt–rock reaction)
The application of the 1D chromatographic model to the 143Nd/144Nd isotopic variations in the Lherz section has several drawbacks. First of all, the modelled chromatographic front is steeper than the 143Nd/144Nd variation observed in the wall-rock (Fig. 6a). A ‘smoother’ pattern could be obtained by decreasing the Nd diffusion coefficient in the minerals, but this would require chemical disequilibrium between melt and minerals whereas SIMS data indicate near-equilibrium, as attested by almost identical Nd content in cpx in the dyke and in the amphibole harzburgite (Table 3 and Fig. 3). Alternatively, to explain smoother patterns one could invoke efficient ‘interstitial diffusive dispersion’ of the elements (Bodinier et al., 1990
, and references therein), but we lack the experimental data to constrain the actual value of this parameter. However, the main weakness in this application is apparent in Fig. 6b, where we compare the measured Nd abundance in cpx with the model predictions based on the best-fit parameters in Fig. 6a. It is clear from this diagram that the ‘best fit’ obtained for the isotopes does not fit with the Nd concentrations. The calculated Nd abundances in cpx are much lower than those measured by SIMS, which implies that the minerals are far from being in chemical equilibrium with the melt. This result is a reflection of the slow diffusivity of Nd in minerals and of the relatively low melt/rock ratios in the model (i.e. 1% at 10 cm). As illustrated in Fig. 6b (dashed lines), the Nd concentration in cpx can be modified as follows.
- Duration of the infiltration process. This was increased by a factor of ten (i.e. 1000 years), which requires a decrease in the interstitial melt velocity from 1 to 0·1 cm/year if the distance of percolation is kept constant at 1 m. However, in this case only the cpx adjacent to the dyke (<1 cm) is re-equilibrated (Fig. 6b). Further from the dyke, the Nd concentration in cpx decreases and reaches the lowest value of the ‘best-fit’ model at about 10 cm. Moreover, increasing the duration of the infiltration process results in a 143Nd/144Nd front that is too close to the dyke, when compared with the data for Lherz (Fig. 6a). A reduction in amphibole proportion is required to balance this effect (Fig. 5).
- Melt/rock ratio. Increasing the melt/rock ratio can be achieved by increasing the melt fraction to 1%, or the distance of percolation to 2 m. In both cases (Fig. 6b), the model accounts for the abundance of Nd in the amphibole peridotite (<25 cm from the dyke). However, the 143Nd/144Nd fronts are displaced to a greater distance from the dyke (Fig. 6a).
Most importantly, none of these models reproduces the strong enrichment in Nd observed in the distal amphibole-free peridotite at >30 cm. This result is not surprising, as the non-reactive model cannot account for trace-element abundances higher than those in equilibrium with the infiltrated melt (i.e. ‘ultra-enriched’ compositions).
Reactive models (with melt–rock reaction)
Strongly enriched trace-element abundances can be produced by melt infiltration mechanisms that involve melt–rock reactions at decreasing melt mass (Bedini et al., 1997; Ionov et al., 2002a). This results in gradual solidification of the melt in the infiltrated peridotite, a process also known as ‘percolative fractional crystallization’ (Harte et al., 1993). In the Lherz wall-rock, evidence for this process is provided by the presence of secondary amphibole and cpx in the amphibole harzburgite zone (<25 cm from the dyke; Bodinier et al., 1990; Woodland et al., 1996). Godard et al. (1995) predicted that the combination of melt transport and reaction would result in strong enrichments of incompatible elements that would be spatially disconnected from the reaction domain. For the most incompatible elements (i.e. for chromatographic fronts moving faster than the reaction front), the enrichment is expected to occur between the chromatographic front and the reaction front (Godard et al., 1995, fig. 1a and b). In this scenario, the enrichment in highly incompatible elements in the Lherz wall-rock would occur beyond the amphibole peridotite and may account for the enrichment in the amphibole-free peridotite.
To evaluate this hypothesis, we investigated the effect of amphibole precipitation on the variations in the abundance of Nd and 143Nd/144Nd isotope ratio in the wall-rock using the ‘plate model’ of Vernières et al. (1997). In this model, melt transport is not considered as a function of time and distance but simulated by propagation of melt batches through macrovolumes of mantle rock. The plate model cannot be used to provide space and/or time constraints for the considered process. On the other hand, this approach allows the simulation of complex melt–rock reaction processes (e.g. Bedini et al., 1997; Xu et al., 1998).
To illustrate the effect of reaction (i.e. amphibole precipitation), the model was initially run without reaction. This first step is useful because it helps establish the number of cells for the model. Because the model assumes homogeneous mineral and melt compositions within individual reaction cells after each increment of melt transport and reaction, the number of cells in a given percolation column is a reflection of the efficiency of melt homogenization by intergranular diffusion mechanisms. In other words, low cell numbers indicate efficient interstitial dispersion of the elements and isotopic signatures whereas high cell numbers indicate inefficient dispersion. This effect is shown in Fig. 7a, where the results of plate modelling applied to 143Nd/144Nd isotope variations in the wall-rock are shown for variable cell numbers. This indicates that high cell numbers (i.e. 100 and 1000) result in steeper chromatographic fronts than low cell numbers (i.e. 10 and 20). The data best fit a model involving 20 cells. If the distance of melt infiltration was around 1 m, the critical distance of interstitial melt homogenization by intergranular diffusion would be about 5 cm. This should be treated as a rough estimate because intergranular diffusion is likely to be more efficient close to the dyke (i.e. amphibole peridotite) than further in the host (i.e. anhydrous peridotite). The probable cause of such features is the existence of a thermal and porosity gradient. This was proposed by Bodinier et al. (1990) on the basis of major-element gradients (higher Al, Ca and Fe) in the amphibole peridotite.
|
Similar to the 1D approach, plate modelling without reaction does not account for the abundances of Nd observed in the wall-rock (Fig. 7b). Near the dyke (i.e. magma source), the calculated Nd contents in cpx are somewhat higher than obtained with the ‘best-fit’ 1D model. However, when no reaction is involved the plate model is no better than the 1D model in accounting for Nd variation in the wall-rock, especially in the distal anhydrous peridotite.
Logically the next step in modelling was to precipitate amphibole. The precipitation of cpx is ignored for simplicity, the effect of this mineral on Nd variations being basically the same as that of amphibole. In the model, amphibole precipitation was allowed to occur in the lower quarter of the percolation column, adjacent to the magma source (= dyke). This is consistent with the observation that amphibole and secondary cpx are restricted to the amphibole peridotite (<25 cm from the dyke). Based on textural observations indicating that amphibole occurs both in thin veinlets and as a replacive mineral after opx (Woodland et al., 1996), the amphibole is formed in the model at the expense of both interstitial melt and opx. Whereas subtle differences in the input parameters (e.g. the rate of amphibole precipitation, i.e. the amount of amphibole precipitated relative to the amount of melt infiltrated) account for the different results (Fig. 8), all the models result in amphibole fractions varying from 4–5% close to the melt source to zero at some distance from the dyke, within the wall-rock (Fig. 8a). The average proportion of amphibole in the lower quarter of the column (1–2%) is consistent with the values actually observed in the amphibole harzburgite (leaving out the main lherzite veins). Most importantly, the models account reasonably well for the abrupt variation in 143Nd/144Nd, the abundance of Nd, and the existence of two domains in the wall-rock, bounded by the reaction front of amphibole precipitation.
|
|
ALTERNATIVE HYPOTHESES |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
The model above assumes distinct origins for chemical (LREE) and isotopic (Sr–Nd) enrichment in the anhydrous harzburgite. The chemical enrichment is considered to be young, being related to dyke emplacement in the Cretaceous, whereas the isotopic enrichment would be much older, possibly linked to the harzburgite formation. An alternative to this model would be to ascribe the LREE enrichment to an early metasomatic stage. This scenario assumes a diffuse metasomatic event around 1000 Myr ago that would also account for modal (apatite), chemical and isotopic enrichment in the wall-rock protolith. Although rather unlikely—because strong LREE enrichment as observed in the studied wall-rock is not observed in harzburgites spatially unrelated to dykes—this model cannot be completely ruled out. Therefore, the hypothesis that the wall-rock was enriched in LREE before dyke injection must be examined. An important implication of this scenario is the requirement that LREE and other highly incompatible elements (P) were ‘drained’ from the wall-rock into the dyke to account for the decreasing abundance of these elements toward the dyke. This can be achieved by the following processes.
(1) Advective element transport into partial melt tapped from the wall-rock by the dyke
Heating by melt flowing in the dyke could have triggered incipient melting of the wall-rock and migration of small fractions of partial melt into the fracture–vein system. However, assuming a wall-rock protolith composition with 143Nd/144Nd = 0·5124, any melts generated within that protolith would have an identical ratio and migration of such melts toward the dyke would impart this isotopic composition on the wall-rock. This prediction is at variance with the systematic increase in 143Nd/144Nd observed from the enriched wall-rock toward the dyke. In addition, melting of the wall-rock could not explain the Fe-rich halo adjacent to the dyke, nor the overall fertile character of the amphibole harzburgite, compared with the anhydrous peridotite (i.e. the Al, Ca, Ti and HREE enrichment coupled with the presence of secondary cpx and amphibole).
(2) Diffusional transport via interstitial melt infiltrated from the dyke into the wall-rock
In this model, melt infiltration would explain Fe-enrichment and fertilization in the amphibole harzburgite, whereas interstitial diffusion in infiltrated melt would account for variable degrees of re-equilibration of the ultra-enriched wall-rock protolith to the less enriched compositions in equilibrium with melt in the dyke. Bodinier et al. (1990) suggested that intergranular diffusion via infiltrated melt might have been significant in the amphibole harzburgite (<25 cm), being enhanced by the thermal gradient and a high melt fraction. However, phosphorus shows a steadily increasing abundance much beyond the amphibole harzburgite zone, at a distance 75 cm from the dyke; this extended gradient of P is difficult to explain by diffusion (Fig. 1). Moreover, diffusion cannot explain the negative HFSE anomalies that occur in a transition zone (25–60 cm from the dyke) between amphibole harzburgite and the most distal apatite-bearing harzburgite, which both contain HFSE-undepleted cpx (Fig. 3).
(3) A combination of these two mechanisms: diffusion in infiltrated melt (amphibole harzburgite) and draining of incipient partial melt (apatite harzburgite)
A scenario involving diffusion in the infiltrated melt close to the dyke and draining of incipient melt further in the host might be envisioned to explain both the fertilization of the amphibole harzburgite, and the trace-element and isotopic variability in the wall-rock. However, even such a complex model would leave unexplained the negative HFSE anomalies in cpx at 25–60 cm from the dyke. On the other hand, these anomalies can be accounted for by a reactive porous flow model. They were interpreted by Bedini et al. (1997) and Ionov et al. (2002a) as ‘transient’ (chromatographic) signatures resulting from the precipitation of Nb-rich microphases (Ti-oxides or amphibole) along with melt percolation. Alternatively, these anomalies may be explained by the lower solubility of these elements in carbonate-rich derivatives of the infiltrated melt.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTION AND AIMSWALL-ROCK DESCRIPTION AND...ANALYTICAL METHODSRESULTSDISCUSSIONTHEORETICAL MODELLINGALTERNATIVE HYPOTHESESCONCLUSIONSREFERENCES |
|---|
Reactive chromatographic processes offer an explanation for the juxtaposition of the amphibole–garnet pyroxenite dyke, subordinate hornblendite veins, the amphibole harzburgite and the apatite-bearing harzburgite with a single-stage process. Both the amphibole and the apatite-forming metasomatic processes originated from the same dyke, which produced water-rich and carbonate-rich melts as derivatives from crystal segregation in a magma conduit (pyroxene–garnet in the dyke) and wall-rock reactions (amphibole ± cpx ± phlogopite in veins and wall-rock proximal to the dyke). In this particular example there is no necessity to invoke different and genetically distinct metasomatic melts or fluids (i.e. a hydrous silicate melt and a carbonate melt).
Modelling resolves the paradox of the spatial decoupling between isotopic contamination and LREE enrichment. Indeed, the simulation of isotopic variations during