Se contents and S/Se ratios of spinel peridotite xenoliths from the Massif Central (France)

Jean-Pierre Lorand, Ambre Luguet: Laboratoire de Minéralogie, Paris, France; Reid R. Keays : Laurentian University, Ramsey Lake Road, Sudburry, Canada, Now at Monash University, Australia; Olivier Alard : GEMOC, Macquarie

Introduction

Selenium and sulfur are volatile chalcophile elements in meteorites. The S/Se of chondrites is well constrained at 2560±200 (Dreibus et al, 1995). In the terrestrial mantle the abundances of S and Se, often referred as chalcogenides (Morgan, 1986), are relevant to core-mantle segregation models and the late accretionary history of the Earth (e.g. O'Neill, 1991). Upper mantle peridotites show these two elements concentrated into low-melting discrete sulfide microphases belonging to the Fe-Ni-S system. Nevertheless, the behaviour of S and Se is far from being fully understood. Interpretations of S concentration data are strongly debated because of the poor stability of host sulfides with respect to supergene weathering and late-stage oxidations that preferentially affect mantle xenoliths uplifted by within-plate alkali lavas (Lorand, 1990; Ionov et al, 1992). Reliable Se data are almost lacking probably because this element is three order of magnitude less concentrated than S. Here we present Se analyses for 43 spinel lherzolite xenoliths from the Massif Central Cenozoïc volcanism with in view a better understanding of Se and in corrolary S abundances in the sub-continental lithospheric mantle-SCLM. Selenium is considerably more stable than S during low temperature alterations. It thus provides a mean to overcome late-stage perturbations of chalcogenide abundances.
Sampling
The Massif Central and neighbouring areas of recent volcanic activity (Bas-Languedoc-Causses) offers a complete section across a thinned, mainly Middle to Upper Proterozoïc SCLM reactivated by the mantle plume that triggered major Plioquaternary alkali volcanism. The samples analysed for Se come from 19 volcanic vents either older (e.g. Languedoc-Causses-Livradois; 9-25 Ma) or contemporaneous to the impigement of the mantle plume beneath the MCF (e.g. Sioule, Devès, Velay; Vivarais; 5-0.1 Ma). Modal compositions range from fertile lherzolites (3.75 wt.% CaO) to harzburgites (0.3 wt.% CaO; Fig. 1). The sample studied can be subdivided into three groups according to textures, equilibration temperatures and incompatible trace element patterns. Group I samples are low-T (<1000°C) protogranular to porphyroclastic peridotites from the Causse-Languedoc area. Group II peridotites come from volcanic vents less than 5 Ma old; these are medium T (1000-1100°C) protogranular to granular/granuloblastic peridotites. Group III peridotites, all from the Quaternary volcanism, display high-temperature (1200°C) poikiloblastic textures. Incompatible trace element patterns of group I and group II peridotites range from DMM-like to enriched in the most incompatible elements. Some samples display complex ITE patterns (strong LILE enrichments and HFSE negative anomalies) usually ascribed to percolation of carbonated or volatile-enriched small volume melts. U-shaped REE patterns ascribed to chromatographic fractionation of the LILE are not uncommon in group II peridotites. Textures and trace element patterns of group III peridotites (regular enrichment from the less incompatible to the most incompatible trace element) are considered to be due to extensive re-equilibration with large volumes of magmas at the base of a thinned (<60 km) SCLM above the MCF mantle plume (Alard et al, 1996).

Results

Selenium was analysed at the Laurentian University by Hydride Generation and Flameless AAS from 10 g whole-rock splits, which allowed 0.2 ppb detection limits and precision of about 5%. S data come from Lorand (1990) or were generated by iodometric method for the purpose of this study. The Se concentration range (0.2 to 67 ppb) conform with the few results previously reported for similar mantle xenoliths (Jagoutz et al, 1979; Morgan, 1986). However, distinctions can be established between the three groups (Fig. 1)

Group I lherzolites have broadly similar Se contents (17-67 ppb) as the Southwestern USA highly fertile mantle xenoliths analysed by Morgan (1986). In the Montferrier neck (Languedoc) fertile lherzolites displaying ITE patterns of the "carbonated" metasomatism are slightly higher in Se than the other samples showing LILE-depleted ITE patterns (40-67 pb vs. 17-48 ppb). Group 2 protogranular and granular/granuloblastic xenoliths are on average Se-poorer than Group I samples, although some overlaps are observed in Fig. 1. Most group II Se contents span the range 5-22 ppb irrespective of ITE patterns ; fertile lherzolites with up to 5 wt.% Ti pargasite and high Ba, Rb and Sr contents and lherzolites with DMM-like REE patterns may show similar Se contents. Group III poikiloblastic xenoliths are markedly depleted in Se (<5 ppb) compared to the other two groups, with Se contents as low as 0.2 ppb. This sharp Se decrease may be observed in xenoliths from a single volcanic vent (e.g. Montbriançon, Devès).



 

Although usually believed to be moderately incompatible, Se does not correlate with lithophile elements of similar incompatibility (e.g. Al2O3; CaO, used as Fertility Indexes). Lherzolites display the largest scatter. However, poikiloblastic Group III peridotites preserve almost constant Se abundances in spite of the largest variation of CaO (0.3-3 wt.% ). In contrast, the Se contents are proportional to the number of sulfide grains per polished thin sections, as recomputed from the abundance of iron hydroxide pseudomorphs replacing ancient magmatic sulfides (r= 0.75). A broadly positive correlation is also observed between S and Se (r=0.75). Extremely little xenoliths have preserved chondritic S/Se ratios. These are either small chips encasted in massive basaltic lava flows or group III peridotite xenoliths which contain only fully enclosed sulfides in olivine and pyroxenes. There, sulfides display no petrographic evidence of weathering. Some samples that have been metasomatized by volatile-rich small volume melts (''carbonated metasomatism''; e.g. Montferrier; Montboissier) and amphibole-rich lherzolite display higher than chondritic S/Se (up to 10 4 in Montferrier sample PG6; Fig. 2) typical of hydrothermal sulfide ores. Half of the samples analysed display lower than chondritic S/Se. These are xenoliths from scoria, cinder cones, tuffs and oxidized magmatic breccias where petrographic evidence of wheathering of host sulfides are widespread (Lorand, 1990)

Discussion

Fig. 2 provides indisputable evidence of modification of S-Se systematics relative to chondrites. Volatile-rich ("carbonated") metasomatic melts produced S gains relative to Se whereas S-losses resulted from supergene weathering. Assuming chondritic initial S/Se, the amount of S lost, probably as sulfates, through weathering of sulfides can reach 100%. According to Fig. 2 about half of the sulfur data measured in MCF xenoliths are not reliable and care must also be exercised in recomputing pre-entrapment S contents assuming chondritic S/Se ratios because some samples clearly had higher than chondritic S/Se ratios prior to supergene weathering. Nevertheless, Se data seriously question the conclusion that supergene weathering was responsible for the low and uniform S contents of continental mantle xenoliths (c.f. Lorand, 1990). The relationships between Se abundances and petrographic characteristics of the investigated xenoliths argue for large scale variations of chalcogenide abundances in the SCLM beneath the Massif Central, unrelated to post-entrapment alteration processes.

As it was only marginally affected by the MCF plume, group I peridotites could have preserved the Se contents of the SCLM before reactivation by the plume. Assuming chondritic S/Se ratios, the Se data (20-50 ppb) translate into primary S of about 50-130 ppm. These latter are at the lower bound of the MORB-source Mantle sampled by orogenic lherzolites, although significantly depleted (see also Alard et al, this volume). The uniformly low Se (and S) contents of poikiloblastic peridotites can be explained by an almost complete removal of molten sulfides by the melt infiltrated from the MCF plume. The fact that both harzburgites and lherzolites are similarly Se-depleted suggests that such high-T percolation processes involving large melt fractions could generate sulfide-depleted haloes at the base of thinned SCLM. The significance of Group II Se data is more controversial. The lack of positive correlation between Se and FI indicates that Se has been mobile at some stages postdating the ancient melting event, perhaps in cunjunction with highly evolved melts infiltrated from the MCF plume. However, the Group II S contents (30-60 ppm) that can be recomputed from Se contents fit well the range reported for xenolith suites from worldwide provenances (c.f. Lorand, 1990; Ionov et al, 1992). In short, the S/Se systematics of MCF peridotite xenoliths highlight the difficulty of ascribing a unique S and Se concentration range for the SCLM.

References

Alard, O., Dautria, J.M. & Bodinier, J.L., (1996) C.R.Acad. Sci. Paris, 323, 763-770.

Ionov, D.A., Hoef, J., Wedepohl K.Hans and Wiechert, U. (1992), Earth Planet. Sci. Lett. 111 : 269-286.

Lorand, J.P. (1990), Geochim. Cosmochim. Acta, 54 : 1487-1492.

Morgan, J.W., (1986), J. Geophys. Res., 91 : 12375-12387.

O'Neill, H.St. C., (1991), Geochim. Cosmochim. Acta, 55 : 1159-1172.