JOURNAL OF PETROLOGY
VOLUME 39
NUMBER 11 & 12
PAGES 1931–1941
1998
Trace Element Composition of Mantlederived Carbonates and Coexisting Phases in Peridotite Xenoliths from Alkali Basalts DMITRI IONOV∗ GEMOC, SCHOOL OF EARTH SCIENCES, MACQUARIE UNIVERSITY, SYDNEY, 2109 N.S.W., AUSTRALIA
RECEIVED SEPTEMBER 30, 1997; REVISED TYPESCRIPT ACCEPTED MAY 21, 1998
The trace element compositions of carbonate, clinopyroxene, amphibole and silicate glass were determined in four mantle lherzolite xenoliths in alkali basalts from Spitsbergen and Mongolia by laser ablation ICP-MS. Carbonates in the xenoliths occur in fine-grained pockets that appear to have been produced by reaction of carbonaterich melts with the host peridotites. The carbonates are rich in Sr and Ba, but have low contents of rare earth elements. (Na,Al)-rich silicate glass commonly associated with the carbonates is a major host for many incompatible lithophile elements in the carbonatebearing pockets. The carbonates in the xenoliths do not appear to represent quenched carbonatite liquids but probably are crystal cumulates from carbonate-rich melts. Trace element patterns estimated for liquids that may have produced the carbonate-bearing pockets are consistent with general characteristics of carbonaterelated metasomatism (enrichments in light rare earth elements, Th, U, Ba and negative anomalies for high field strength elements). However, the absolute incompatible element abundances estimated for those liquids cannot provide the extremely strong enrichments invoked by some models of carbonate mantle metasomatism. Clinopyroxene and amphibole outside the carbonate-bearing pockets in the xenoliths from Spitsbergen have high contents of incompatible trace elements, indicating that the lherzolites also experienced metasomatic enrichment before the formation of the carbonates.
Carbonates may be an important accessory component in the mantle. They have been found in mantle peridotite,
pyroxenite and eclogite xenoliths as texturally equilibrated or interstitial grains (Wass, 1979), ‘globules’ in association with silicate glass (Amundsen, 1987; Ionov et al., 1993, 1996; Pyle & Haggerty, 1994; Kogarko et al., 1995; Norman, 1998) and components of fluid microinclusions (Frezzotti et al., 1994; Schiano & Clocchiatti, 1994). There is unambiguous experimental evidence that carbonates are stable in peridotitic assemblages at appropriate temperatures and pressures, and that smalldegree partial melting of carbonated peridotite at pressures [2 GPa produces carbonatitic liquids (Wyllie, 1987; Dalton & Wood, 1993a; Lee & Wyllie, 1998). Interaction of carbonate melts with the lithospheric mantle has been invoked to explain the formation and compositional characteristics of some metasomatized mantle peridotites (Wallace & Green, 1988; Yaxley et al., 1991; Dautria et al., 1992; Hauri et al., 1993; Ionov et al., 1993; Rudnick et al., 1993). Little is known about trace element composition of carbonates in mantle rocks and of primary carbonaterich liquids generated in the mantle. The models relating mantle metasomatism to carbonate-rich melts usually assume that these melts are similar in trace element composition to carbonatites exposed in the crust and, in particular, that such melts should be strongly enriched in incompatible trace elements. Unfortunately, crustal carbonatites show a wide range in chemical compositions and few workers believe that carbonatites reflect liquid compositions. Experimental data on element partitioning between immiscible silicate and carbonate liquids have also been used to assess the composition of mantle carbonates and primary carbonate-rich melts. It is not clear, however, whether such data are relevant in this
∗Telephone: (61-2) 98508378. Fax: (61-2) 98508428. e-mail:
[email protected]
Oxford University Press 1998
carbonate; clinopyroxene; LAM–ICP-MS; mantle metasomatism; (Na,Al)-rich silicate glass KEY WORDS:
INTRODUCTION
JOURNAL OF PETROLOGY
VOLUME 39
case because liquid immiscibility may not have played an important role in their formation (Lee & Wyllie, 1998). Direct data on trace element composition of mantle carbonates and related minerals in natural mantle rocks may provide important evidence for models of trace element enrichment by ‘carbonate’ metasomatism and generation of parental carbonate liquids. Ionov et al. (1993) provided analyses of whole-rock carbonate-bearing peridotite xenoliths from Spitsbergen and of acid leaches from these rocks done by solution inductively coupled plasma mass spectrometry (ICP-MS). They found enrichments in light rare earth elements (LREE) and Sr relative to heavy rare earth elements (HREE) and high field strength elements (HFSE) in plots normalized to their abundances in chondrites or primitive mantle (PM), and concluded that these trace element signatures are characteristic of the carbonate material in the xenoliths. Ionov et al. (1996) reported high contents of Sr and Ba in carbonates and Sr in clinopyroxene in these xenoliths determined in situ by proton microprobe analyses. The present work is based on laser ablation ICP-MS analyses for a large number of trace elements in carbonates and coexisting silicate minerals in mantle peridotite xenoliths in alkali basaltic rocks from Spitsbergen and Mongolia. This study examines the trace element compositions of mantle carbonates and an attempt is made to constrain the composition of mantlederived carbonate-rich liquids.
ANALYTICAL TECHNIQUES All samples were analysed by laser ablation microprobe (LAM)–ICP-MS ( Jackson et al., 1992) in 200 lm thick polished rock sections. The ICP-MS instrument is a Perkin–Elmer Sciex ELAN 5100 coupled with a UV (266 nm) laser. The laser was operated with 1 mJ/pulse energy and 4 Hz frequency for silicate minerals, and 2 Hz frequency with the laser beam focused above the sample surface for carbonates and silicate glass. Spot diameter for these analyses is 30–50 lm. NIST 610 glass was used as a calibration standard for all samples, with 44 Ca as an internal standard. Detection limits for most elements in clinopyroxene and amphibole were ~0·1 ppm. For carbonate and silicate glass, detection limits were higher when a low-energy, defocused beam was used. Analytical precision is Ζ5% at the ppm level. Details of ICP-MS and laser operating conditions have been published by Norman et al. (1996) and Norman (1998). Nonmatrix-matched calibration standards have been successively used for quantitative carbonate analysis by Feng (1994), who employed the NIST 612 silicate glass as a calibration standard to determine Sr, Y, Ba and REE in calcite and dolomite.
NUMBER 11 & 12
NOV & DEC 1998
In-run signal intensity for indicative major and minor elements was monitored during analysis of small mineral grains and glass pockets (Fig. 1a and b) to make sure that the laser beam stayed within the grain or phase selected. Ablation of carbonates was aborted when signal intensity for Al, Si, Cr or Ni increased well above background levels, and carbonate analyses with considerable levels of these elements were discarded. Rb and Ba were monitored for small clinopyroxene grains to check for contamination by silicate glass. Estimates of major element contents (Si, Al, Mg) obtained by LAM– ICP-MS agree well with electron microprobe (EMP) analyses of the same minerals. There is also a good agreement between LAM–ICP-MS data in this work and proton probe (PIXE) analyses for Sr, Ba, Y and Zr obtained earlier for Spitsbergen xenoliths 4-36-90 and 21-6 (Ionov et al., 1996). Average contents of Sr, Y and Zr determined by PIXE in primary and carbonate-related clinopyroxene in the same samples are largely within 15% of corresponding LAM–ICP-MS data. Some differences may be due to significant chemical zoning and grainto-grain variations in carbonate and carbonate-related clinopyroxene (Ionov et al., 1996).
SAMPLE DESCRIPTION Trace element data were obtained for three spinel lherzolite xenoliths from NW Spitsbergen and one spinel lherzolite xenolith from SE Mongolia (Dariganga). Table 1 gives a summary of petrography, mg-number [mgnumber = Mg/(Mg + Fe)at] of olivine, Cr2O3 content of spinel and estimates of equilibration temperatures for these samples. Detailed data on petrography and major element composition of the Spitsbergen xenoliths (including the samples in this study) have already been published (Ionov et al., 1996). The xenoliths are coarseto medium-grained rocks with microstructures ranging from coarse protogranular to mosaic equigranular. Typical grain sizes range from 2–6 mm for olivine and orthopyroxene to 1–2 mm for clinopyroxene and spinel. Xenolith 4-90-9 contains rare, small (0·5–1 mm) amphibole grains. Amphibole is more abundant and its grains are larger in xenolith 21-6, which also contains rare equant grains of apatite 0·1–0·5 mm in size. Carbonates analysed in this work occur in fine-grained pockets replacing minerals of host peridotite, most commonly spinel, orthopyroxene and amphibole (Ionov et al., 1996). These pockets also contain small (
