Articles
Rutile/melt partition coefficients for trace elements and an assessment of the influence of rutile on the trace element characteristics of subduction zone magmas

https://doi.org/10.1016/S0016-7037(99)00355-5Get rights and content

Abstract

Fractionation of some or all of the high field strength elements (HFSE) Nb, Ta, Zr, Hf, and Ti relative to other trace elements occurs in igneous rocks from convergent margins and in the average continental crust, and is generally attributed to a process occurring during subduction. The experimental partitioning of an extensive array of trace elements between rutile/melt pairs is presented which enables the effect of rutile during melting in subduction zones to be directly assessed. DNb and DTa are in the range 100–500, DZr and DHf are about 5, whereas all other trace elements analyzed have Drutile/melt less than 0.1. Published D patterns for Nb and Ta between rutile and water-rich fluids are similar to those for melt, whereas the values for Zr and Hf are significantly higher. DNb and DTa values for clinopyroxene and garnet are much lower than for rutile, and cannot cause the fractionation of HFSE from other elements seen in island arcs. The presence of rutile in the subducted slab residue during dehydration may be essential in the production of the geochemical signatures of arc magmas, whereas that of the continental crust, including higher Zr/Sm, may be produced by melting of eclogite.

Introduction

The trace element signatures of island arc magmas and the continental crust deviate from those of oceanic basalts, showing that one or more processes in addition to melting of peridotite must contribute to their generation. Island arc magmatic rocks exhibit characteristically low abundances for the high field strength elements Nb, Ta, Zr, Hf, and Ti relative to those of other incompatible elements, resulting in characteristic negative anomalies in normalized incompatible element variation patterns, in particular for Nb and Ta. The mechanism for this is debated, although it almost certainly takes place in or above the subducted slab. Many studies of the origin of the continental crust conclude that it also formed in association with subduction zones because of the overall similarity in composition between andesites and the continental crust (Taylor and McLennan, 1985). The continental crust also shows negative anomalies for Nb, Ta, and Ti, but it differs from many island arc lavas in having no relative depletion of Zr and Hf (Rudnick and Fountain, 1995), and may for upper continental crust have positive anomalies relative to Sm. These differences must be accounted for in the genetic mechanisms proposed. Although the growth curves of the continental crust are debated, much of the continental crust appears to have formed in Archean and early Proterozoic times McLennan and Taylor 1982, Nelson and DePaolo 1985 when the processes involved may have been different due to higher geothermal gradients at that time. A better understanding of the underlying mechanisms for the fractionation of HFSE in subduction zones is also important for the assessment of the role of recycled subducted material in oceanic magmatic rocks.

Most authors favor one of two hypotheses for the origin of the HFSE anomalies. In the first, partial melting of either the subducted slab or of the mantle wedge overlying it can cause the anomalies due to the stability of rutile or another titanate mineral which retains the HFSE. It is now established on the grounds of experimental solubility measurements of rutile Green and Pearson 1986, Ryerson and Watson 1987 and also from geochemical studies of island arc volcanics Woodhead et al 1993, Thirlwall et al 1994, that rutile cannot coexist with basaltic melts arising from partial melting of peridotite in the mantle wedge. It may, however, be present during melting of the subducting oceanic crust (Ringwood, 1990) or of non-peridotitic parts of the mantle wedge (Foley and Wheller, 1990). The second hypothesis appeals to fluid loss by dehydration reactions in the subducting plate: this is consistent with thermal models for subduction zones which suggest that the melting points of subducted materials are not usually attained at relevant depths Peacock 1996, Poli and Schmidt 1995. In this scenario, the HFSE anomalies in the volcanic melt compositions are obtained from melting of a mantle wedge which itself is depleted in the HFSE due to preferential enrichment of all other incompatible elements by the slab-derived fluid. Recent experimental work contains evidence both for (Keppler, 1996) and against Brenan et al 1995, Stalder et al 1998 the existence of HFSE anomalies in fluids without the need for residual titanate minerals. There has been a tendency to associate the necessity of residual titanate minerals with melting processes, although it applies equally to dehydration reactions.

The full understanding and quantification of the development of HFSE anomalies in subduction zone magmatism has been hampered by the lack of partitioning data for large trace element sets between rutile and melt, for which we here present new data. The application of laser ablation microprobe–ICP–MS facilitated the collection of partitioning data for a more comprehensive palette of trace elements than was previously available. These data complement recently acquired data for rutile/fluid pairs Brenan and Watson 1991, Brenan et al 1994, Ayers et al 1997, Adam et al 1997, Stalder et al 1998, Ayers 1998, allowing assessment of the relative importance of melting and dehydration reactions.

Section snippets

Experimental and analytical techniques

To produce minerals and melts representative of subduction zone processes, we have conducted melting experiments in the pressure range 1.8–2.5 GPa on the Lalkaldarno tonalite from Victoria, Australia, previously used in experiments by Jenner et al. (1993). Experiments were conducted in a single-stage piston-cylinder apparatus at the University of Göttingen using Pt capsules with graphite inner linings, and CaF2 as the pressure medium. Small amounts of water were added in liquid form. Two

Rutile/melt partitioning

Partition coefficients for trace elements between rutile and melt from the tonalite experiments are given in Table 1. The overall pattern of partitioning is depicted in Figure 1, in which the element order follows that for upper mantle melting (Sun and McDonough, 1989). Downward-pointing open triangles indicate maximum values, meaning that the elements involved were analyzed in the glass but below the limit of detection (used in calculation of maximum DRu/Lq) in rutile. Values for Ta are

Acknowledgements

Funding by Deutsche Forschungsgemeinschaft grant Fo 181/2-3 to S.F.F. is gratefully acknowledged. We thank S. M. Eggins, A. J. Crawford, P. Robinson, and B. Spettel for providing materials (geological or electronic) or analyses, and I. Horn for assistance during acquisition of LAM–ICP–MS data. Comments on this paper from J. Ayers and S. R. van der Laan, three anonymous reviewers, and the Harvard geochemistry group are much appreciated.

References (43)

  • T Plank et al.

    The chemical composition of subducting sediment and its consequences for the crust and mantle

    Chem. Geol.

    (1998)
  • R.P Rapp et al.

    Partial melting of amphibolite/eclogite and the origin of Archean trondhjemites and tonalites

    Precamb. Res.

    (1991)
  • A.E Ringwood

    Slab-mantle interactions. 3. Petrogenesis of intraplate magmas and structure of the upper mantle

    Chem. Geol.

    (1990)
  • F.J Ryerson et al.

    Rutile saturation in magmasImplications for Ti–Nb–Ta depletion in island-arc basalts

    Earth Planet. Sci. Lett.

    (1987)
  • R Stalder et al.

    Mineral-aqueous fluid partitioning of trace elements at 900–1200°C and 3.0–5.7 GPaNew experimental data for garnet, clinopyroxene, and rutile, and implications for mantle metasomatism

    Geochim. Cosmochim. Acta

    (1998)
  • J Woodhead et al.

    High field strength and transition element systematics in island arc and back-arc basin basaltsEvidence for multi-phase extraction and a depleted mantle wedge

    Earth Planet. Sci. Lett.

    (1993)
  • J Adam et al.

    Trace element partitioning between aqueous fluids, silicate melts and minerals

    Eur. J. Mineral.

    (1997)
  • J Arth et al.

    Quartz diorites derived by partial melting of eclogite or amphibolite at mantle depths

    Contrib. Mineral. Petrol.

    (1972)
  • J.C Ayers

    Trace element modeling of aqueous fluid–peridotite interaction in the mantle wedge of subduction zones

    Contrib. Mineral. Petrol.

    (1998)
  • Ayers J. C. and Watson E. B. (1993) Rutile solubility and mobility in supercritical aqueous fluids. Contrib. Mineral....
  • Ayers J. C., Brenan J. M., Watson E. B., Wark D. A., and Minarik W. G. (1992) A new capsule technique for hydrothermal...
  • Cited by (552)

    View all citing articles on Scopus
    View full text