The distribution of lithium in peridotitic and pyroxenitic mantle lithologies — an indicator of magmatic and metasomatic processes

Abstract

Lithium concentrations in orthopyroxene, clinopyroxene, olivine, garnet or spinel from equilibrated spinel peridotite, garnet peridotite and garnet pyroxenite xenoliths and from metasomatised peridotites and pyroxenites were measured using the ion-microprobe (SIMS). With respect to changes in physical and chemical parameters, we present new data on lithium contents in mantle minerals and its partitioning behaviour in (1) equilibrated and (2) metasomatised samples. Lithium is preferentially incorporated into olivine, ranging between 1 and 2 ppm in equilibrated unmetasomatised peridotites. Pyroxenes from peridotitic xenoliths have Li concentrations on the order of several hundred ppb up to 1.3 ppm, while pyroxenes from pyroxenites have somewhat higher abundances (1–3 ppm with a maximum of 19 ppm). The following partitioning relationships have been established: ol>cpx≥opx≫grt or sp for garnet and spinel peridotites, respectively, and cpx≥opx>grt for garnet pyroxenites. The intercrystalline partitioning of Li is independent of T, P and bulk composition (for ultramafic to mafic compositions), making Li a suitable tracer element for chemical processes such as metasomatism. We estimate a bulk Li content of 1.0–1.5 ppm for fertile to moderately depleted lithospheric mantle. Low Li abundances in mantle peridotites and pyroxenites emphasise its incompatibility during partial melting and fractional crystallisation. However, elevated Li concentrations are present in some pyroxenites, presumably due to complete crystallisation of trapped partial melts. Metasomatised samples from peridotite massifs (Pyrenees and Ivrea Zone) and mantle nodules from two Victorian volcanic fields (Australia) clearly show enrichment of Li in both olivine and clinopyroxene, whereby the distribution of Li between olivine and clinopyroxene has generally not achieved equilibrium. Disequilibrium is manifested by preferential Li enrichment in either olivine or clinopyroxene depending on the type of metasomatic agent involved (carbonatite vs. mafic silicate melt). Differences in absolute Li abundances and in its partitioning behaviour allow the identification not only of metasomatic overprints, but also of magmatic processes such as partial melting, crystal fractionation and accumulation. The sensitivity of Li as such a chemical tracer gives an additional criterium for recognising cryptic metasomatism.

Introduction

Trace element data are an essential ingredient for the modelling of petrological processes such as partial melting, crystal fractionation, assimilation or mantle metasomatism. Differences in element partition behaviour permit the identification of various magmatic processes. A fundamental requirement is the knowledge of (1) element abundances in different mantle lithologies and (2) the way in which elements partition between coexisting minerals or between a crystalline and melt phase under given physical and chemical conditions.

Magmatic processes (e.g., partial melting, crystal fractionation and mantle metasomatism) will variably modify the inventory of lithium in the lithospheric mantle. During partial melting or fractionation processes Li will preferentially partition into the melt (e.g., Hart and Dunn, 1993; Blundy et al., 1998; Brenan et al., 1998; Taura et al., 1998). Consequently, Li is enriched in the differentiated crust relative to the primitive mantle. However, Li⁺ has a similar ionic radius to that of Mg²⁺ or Fe²⁺, permitting coupled substitution in olivine and pyroxene, potentially with trivalent cations, such as Al³⁺, Fe³⁺, Cr³⁺, Sc³⁺, V³⁺ and REE³⁺. Based on differences in the partition behaviour, also with respect to larger incompatible cations such as K and Rb, Li may offer a potential tool to identify and model partial melting, crystal fractionation and metasomatic processes.

The aim of this paper is to examine the behaviour of Li during such processes. Our approach was to subdivide our investigation into two steps. The first part involved the determination of Li abundances in mantle minerals from various equilibrated mantle lithologies to gain information about the overall mantle abundance of Li and the role that physical (P and T) and chemical parameters may potentially play in controlling the distribution of Li between olivine, orthopyroxene, clinopyroxene, garnet and spinel. For this purpose, a set of well characterised equilibrated mantle xenoliths (Seitz et al., 1999) was investigated with Secondary Ion Mass Spectrometry (SIMS). In a second step we used the information gained from these mantle xenoliths as a baseline with which to compare mantle samples that have been metasomatically overprinted.