Lithospheric Mantle Source for Continental Arc Crust? Trace Element Compositions of Diopside in Spinel Lherzolites

Marc D. Norman

GEMOC, School of Earth Sciences, Macquarie University, North Ryde NSW 2109

ABSTRACT. Depletions of high-field strength elements (HFSE) such as Ti, Nb, and Ta relative to other trace elements with supposedly similar incompatibilities in basaltic systems (e.g., REE, Y, and alkalies) are well known but poorly understood features of arc volcanics. Slab-derived fluids are most likely responsible for enrichments of mobile elements such B, As, Sb, Rb, and Cs in arc volcanics (Morris et al., 1990; You et al., 1996), but the HFSE and REE are essentially immobile in slab-derived fluids (Keppler, 1996, You et al., 1996), so that the fractionations of these elements observed in arc volcanics apparently suggest either a unique mantle source compared to ridges or plumes, different conditions of melting beneath arcs, or addition of a silicate melt component from the slab to the mantle wedge.

Trace element compositions of diopside in spinel lherzolite xenoliths from SE Australia also show striking depletions of Nb and Ti relative to Th, REE, and Y (Fig. 1), raising the possibility that the lithospheric mantle in this region may have been created or modified in an arc environment. Alternatively, some of these xenoliths may represent appropriate mantle sources for arc magmas, or for the production of continental volcanic arc crust.

SAMPLING AND METHODS. A small suite of spinel lherzolites from Mt. Shadwell, western Victoria, has been studied using electron microprobe and laser ablation ICPMS to determine the major and trace element compositions of their constituent minerals. These data provide useful constraints on the processes affecting the lithospheric mantle in this region. For diopside, abundances of REE, Nb, Y, Zr, Hf, Th, U, Sr, Ga, Ti, Sc, V, Co, and Ni were determined by laser microprobe. Detection limits ranged from 10 ppb for U and Th, to 2 ppm for Ni. The NIST 610 glass was used for calibration of relative element sensitivities and each analysis was normalized using CaO values determined by electron microprobe as an internal standard. Calibration values for the NIST 610 glass, and a description of laser operating conditions and error analysis are given by Norman et al. (1996).

RESULTS. All of the xenoliths in this study are protogranular spinel lherzolites, derived from the lithosphere. Co-existing pyroxene and spinel compositions indicate equilibration temperatures of 850-950 oC, corresponding to relatively shallow depths within the lithosphere (25-35 km) by reference to the SE Australian geotherm (O'Reilly and Griffin, 1985). Two groups of lherzolites were found: one which is relatively fertile in bulk composition (diopside-rich) and has trace element patterns indicating these xenoliths have escaped significant metasomatism, and another group which is depleted in bulk composition (diopside-poor) but with trace element patterns indicating cryptic metasomatic enrichment.

The fertile lherzolites contain 12-13% modal diopside, corresponding to a bulk composition with 3% CaO. Olivines in these lherzolites are Fo89.3-90.6, and co-existing diopsides have Mg# of 91-92. Diopsides in these lherzolites have LREE-depleted patterns with striking depletions of Nb relative to La and Th, and less severe depletions of Ti and Zr relative to adjacent REE (Figs. 1, 2). These trace element patterns can be modelled by 2-5% batch melting of a primitive mantle source in the spinel facies (Fig. 2). The deep negative Nb anomalies observed in these diopsides appear to be a natural consequence of this melting, based on published distribution coefficients that indicate Nb is significantly more incompatible in diopside than either La or Th, and nearly as incompatible as Ba (Skulski et al., 1994; Hart and Dunn, 1993).

In contrast, the observed depletions of Ti relative to the HREE in these diopsides cannot be modelled using published distribution coefficients (Fig. 2). Ti abundances show well-defined melting trends (e.g., decreasing Ti in diopside with increasing Cr in spinel), and probably cannot be explained by subsolidus equilibration between high-Ca and low-Ca pyroxene.

The modest degree of depletion that has affected these fertile lherzolites is similar to that required to produce the Depleted Mantle source of N-MORB, which can be modelled as the residue after extraction of 1-2% continental crust from the primitive upper mantle (Hofmann, 1988). A link between the compositions of these fertile spinel lherzolite xenoliths and the processes that form continental crust is suggested by the trace element pattern of the melt that would be in equilibrium with these diopsides, calculated from the distribution coefficients. The pattern of this hypothetical melt bears a striking resemblance to that of bulk continental arc crust (Rudnick and Fountain, 1995), including enrichments in LREE/HREE, and a negative Nb anomaly (Fig. 3).

In contrast, 2-5% batch melts produced from a primitive mantle source by the melting model described above would have a positive Nb anomalies due to the extreme incompatibility of Nb relative to La and Th in mantle diopside, producing a pattern unlike any known magma type. However, prior extraction of a only a very small degree partial melt (e.g. 0.1-0.5%) from the primitive source would be sufficient to impose a negative Nb anomaly on any subsequent magmas. Other highly incompatible elements such as the alkalies would also have been extracted efficiently into this initial small degree melt.

This raises the question: how important are small degree partial melts in controlling mantle evolution and the formation of continental crust? Certainly, the negative Nb anomaly of the continental crust and the Nb/Th and Nb/LREE relations between depleted mantle and continental crust, would be difficult to understand if their compositions are controlled simply by percent-level degrees of partial melting, as suggested by a literal interpretation of trace element models for the continental crust and depleted mantle (Hofmann, 1988). If small degree melts are important, where are they in the geological record, and specifically, where is the material with positive Nb anomalies and accompanying enrichments of alkalies and other highly incompatible trace elements?

A related question concerns the mechanism of upper mantle depletion. Conventional wisdom holds that the Depleted Mantle is the complement to the continental crust. Generation of continental crust in volcanic arcs requires a multi-stage process involving extraction of a primitive basaltic crust from the mantle, segregation of the felsic continental component from this basaltic crust, and recycling and efficient mixing of the mafic residue back into the upper mantle.

Recycling of subducted oceanic crust into the lower mantle is thought to be an important process for producing the source regions of mantle plumes, but this produces extreme compositional heterogeneity in plume basalts, in contrast to the global homogeneity of the Depleted Mantle endmember as sampled by N-MORB. The scale of heterogeneities in mantle plumes appears to be on the order of 1-10's of km, based on the compositional variations observed in plume basalts such as those from Hawaii (Lassiter et al. 1996). In contrast, the fertile spinel lherzolites from Mt. Shadwell display compositional homogeneity among and within individual grains, apparently requiring the re-mixing of mafic residues after continental crust extraction back into the upper mantle on the scale of mm's to µm's, if this is the mechanism responsible for their depletion. While mixing is easier in the upper mantle compared to the lower mantle due to its lower viscosity, further consideration of the role of small degree partial melts in creating upper mantle depletions may be warranted (e.g., O'Nions and McKenzie, 1988; Galer and Goldstein, 1991).

The second group of mantle xenoliths in this study are more depleted in bulk composition, with only 2-3% modal diopside, corresponding to ² 1% CaO in the bulk rock. Trace element patterns of diopside in these rocks demonstrate a clear signature of cryptic metasomatic enrichment of LREE, Th, U, Sr, and Nb. Although the normalized trace element patterns of these diopsides retain a deep negative Nb anomaly, the absolute concentrations in these diopsides are 3-10x greater than that measured in diopside from the fertile lherzolites (Fig. 1).

Mass balance models show that the metasomatism produced an absolute enrichment of Nb, Th, and LREE in the rock compared to the fertile lherzolites, and that the high concentrations in the metasomatized diopsides are not simply a dilution effect caused by the presence of less total diopside in the rock. Although a subduction-related origin for the metasomatism has been suggested for some amphibole-apatite lherzolites of southeastern Australia (Griffin et al., 1988), Nb is virtually immobile in slab-derived fluids (Keppler, 1996, You et al., 1996) and is essentially unaffected by subduction-related fluxes into arc magmas (Pearce et al., 1995). In contrast, the addition of Nb into the metasomatized lherzolites studied here shows that Nb was mobile during this metasomatism.

Alternatively, a plume source for the cryptic metasomatism present in these Mt. Shadwell lherzolites may be possible. High Sm/Hf and Zr/Hf, and low Ti/Nb ratios in the metasomatized diopsides suggests that a carbonatitic fluid or melt may have been responsible for the trace element enrichment, which may also be consistent with a plume source for the metasomatism.

CONCLUSIONS. Trace element patterns of diopside in relatively fertile spinel lherzolite xenoliths from Mt. Shadwell, western Victoria, are LREE-depleted and have striking depletions of Nb and Ti relative to the REE, raising the possibility that they may have evolved in an arc environment, or represent a suitable source for arc basalts. The Nb depletions can be accounted for by extraction of a relatively small degree (2-5%) batch melt, but the Ti depletions are difficult to explain by melting using published distribution coefficients. A melt in equilibrium with these diopsides would have a trace element pattern similar to that of bulk continental arc crust, although a prior episode of melting appears necessary to avoid Nb enrichments relative to Th and LREE in the melt.

Cryptically metasomatized xenoliths from Mt. Shadwell are depleted in bulk compositions (i.e. diopside-poor). Diopsides from these xenoliths have trace element patterns indicating addition of Nb, LREE, Th, U, and Sr to these rocks. This may be difficult to account for by a subduction-related process because of the immobility of Nb in slab-derived fluids. Alternatively, a carbonatitic fluid or melt derived from a plume source may have been responsible for the cryptic metasomatism observed in this group of xenoliths.

REFERENCES

Galer S.J.G. and Goldstein S.L., 1991, Early mantle differentiation and its thermal consequences. Geochim. Cosmochim. Acta 55, 227-239

Griffin W.L., O'Reilly S.Y., and Stabel A., 1988, Mantle metasomatism beneath western Victoria: II. Isotopic geochemistry of Cr-diopside lherzolites and Al-augite pyroxenites. Geochim. Cosmochim. Acta. 52, 449-459

Hart S.R. and Dunn T., 1993, Experimental cpx/melt partitioning of 24 trace elements. Contrib. Mineral. Petrol. 113, 1-8

Hofmann A.W., 1988, Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust. Earth Planet Sci Lett 90, 297-314.

Keppler H., 1996, Constraints from partitioning experiments on the composition of subduction-zone fluids. Nature 380, 237-239.

Lassiter J.C., DePaolo D.J., and Tatsumoto M., 1996. Isotopic evolution of Mauna Kea volcano: results from the initial phase of the Hawaii Scientific Drilling Project. J. Geophys. Res. 101, 11769-11780

Norman M.D., Pearson N.J., Sharma A., and Griffin W.L., 1996, Quantitative analysis of trace elements in geological materials by laser ablation ICPMS: instrumental operating conditions and calibration values of NIST glasses. Geostandards Newsletter, in press.

O'Nions R.K. and McKenzie D.P., 1988, Melting and continent generation. Earth Planet Sci Lett 90, 449-456.

O'Reilly S.Y. and Griffin W.L., 1985, A xenolith-derived geotherm for southeastern Australia and its geophysical implications. Tectonophysics 111, 41-63

Pearce J.A., Baker P.E., Harvey P.K., and Luff I.W., 1995, Geochemical evidence for subduction fluxes, mantle melting and fractional crystallization beneath the South Sandwich island arc. J. Petrol. 36, 1073-1109.

Rudnick R.L. and Fountain D.M., 1995, Nature and composition of the continental crust: a lower crustal perspective. Reviews of Geophysics 33, 267-309

Skulski T., Minarik W., and Watson E.B., 1994, High-pressure experimental trace-element partitioning between clinopyroxene and basaltic melts. Chem. Geol. 117, 127-147

You C.-F., Castillo P.R., Gieskes J.M., Chan L.H., and Spivack A.J., 1996, Trace element behavior in hydrothermal experiments: implications for fluid processes at shallow depths in subduction zones. Earth Planet Sci Lett 140, 41-52.

Back to the GEMOC Abstract Titles Page