Application of Mineral/Fluid/Melt Trace Element Partitioning Data to Models of Arc Magma Genesis

Trevor H. Green, GEMOC, School of Earth Sciences, Macquarie University, NSW, 2109

It is widely accepted that most island arc basalts (IAB) come from a peridotitic source in the mantle wedge overlying the subduction zone (SZ). Geochemists commonly compare IAB with mid-ocean ridge basalts (MORB) because both are argued to represent relatively large degrees of melting of a mantle source, and both provide significant contributions to the earthís crust from the mantle. Striking chemical contrasts between IAB and MORB include an overall enrichment in IAB of SiO2 and large ion lithophile elements (LILE), typified by Ba, Rb, Sr, U, Th and Pb, and depletion of high-field-strength elements (HFSE), typified by Nb, Ta and to a lesser extend Zr and Hf. Models attempting to explain these chemical differences have usually proposed a critical contribution of LILE (but not HFSE) to the mantle wedge via a fluid phase from the subducted slab. This H2O-rich fluid also has the effect of depressing the mantle solidus and enhancing the field of crystallization of olivine, yielding relatively SiO2-rich basaltic melts. Alternatively, a contribution from a high-SiO2 hydrous melt from the subducted oceanic crust, or from subducted pelagic sediments, interacts with and modifies the mantle wedge to give it the distinctive source characteristics needed for IAB. Another suggestion has been the chemical modification of the wedge by an upwelling carbonatitic melt.

High pressure experiments have provided phase and major element constraints on these models, and importantly have outlined the restricted conditions where key accessory minerals (e.g. rutile) may have a critical role in controlling trace element (especially HFSE) distribution in derivative melts. High solubility of rutile in basaltic magmas at high pressure (P) is well established, so that rutile is not a residual phase to these magmas in their source regions. However, the marked decrease in solubility of rutile with decreasing temperature (T) and increasing Si02 dictates that rutile will be residual to silicic magma derivation from melting of subducted crust, and so may control the trace element content of the silicic melts that subsequently modify the composition of the peridotitic mantle wedge. Similarly, low rutile solubility in aqueous fluids at appropriate slab P and T suggests that rutile could be important in controlling the HFSE content of fluids entering the mantle wedge.

New experimentally-obtained trace element partition coefficients between minerals and melt or fluid allow further constraints on arc petrogenetic models, and provide the possibility of distinguishing between a dominant aqueous fluid, silicate melt or carbonatitic melt role in causing the distinctive trace element characteristics of the IAB source region. In particular, careful determination of Nb/Ta and Zr /Hf ratios in island arc volcanics may provide pointers to the relative importance of these different trace element enrichment agents. Recent high precision results for arc magmas indicate variation of Nb/Ta from 11 (in the most Nb-depleted IAB) to 20 (in less Nb-depleted IAB) (Eggins et al, 1996) to 33 in high-K IA volcanics (Stolz et al, 1996) (compared with mantle Nb/Ta = 17), whereas Zr /Hf varies from 30 to 48 (compared with mantle Zr /Hf = 36).

GEOCHEMICAL EVIDENCE FROM TRACE ELEMENT CHARACTERISTICS OF FLUIDS IN SUBDUCTION ZONES

Using different approaches, several published estimates of fluid trace element content are presented in Fig. 1, normalized to N-MORB and to Sr = 1 to allow clearer relative comparison. There is remarkable consistency in the patterns, pointing to enrichment of the fluid in LILE (Cs to U and Pb). La, Ce and Sr are generally suggested to be slightly enriched, but the HFSE, Y and REE (Nd to Lu) are relatively depleted in the fluid. Unfortunately the data do not allow evaluation of Nb/Ta or Zr /Hf ratios.

FLUID/SILICATE MELT PARTITIONING DATA

Determinations of trace element partitioning between aqueous fluid and silicate melts (see Table) ranging from basaltic to silicic are plotted in Fig. 2. All pairs show relative enrichment of Rb in the fluid, and relatively flat patterns for most other elements except for depletion of Th relative to U, Pb is enriched in fluid relative to andesitic melt (Keppler, 1996). Overall, elements partition much more strongly into fluids coexisting with silicic melts (where in fact the fluids approach the melt in composition) than into fluids coexisting with basaltic melts. Apart from the preceding points concerning Rb and Pb, fluids do not generally favour trace elements relative to melt, but Fig. 2 suggests that fluids may cause changes in element ratios, such as an increase in Rb/La, U/Th and possibly Pb/Sr.

TABLE: Major element contents of silicate melt starting compositions used in fluid/melt partitioning experiments.

T.E. denotes sum of trace elements added

1. Ayers & Eggler, 1995 (NaCl - H2O fluids)

2. Keppler, 1996 (H2O or (Na, K)Cl - H2O fluids)

3. Adam et al, 1996 (trondhjemite) (H2O or H2O-F, H2O-Cl fluids)

4. Adam et al, 1996 (basanite) (H2O fluid)

MINERAL/FLUID OR MELT PARTITIONING DATA

Partition coefficient (D) data for clinopyroxene, amphibole, garnet and rutile/fluid or melt pairs are given in Figs. 3-6, in order to evaluate any behaviour contrasts between minerals and variously fluid, silicate melt or carbonate melt. In general, mineral/fluid Ds are higher than mineral/melt values. However there are some significant exceptions and points of different behaviour, detailed as follows.

For clinopyroxene (cpx) (Fig. 3) mineral/fluid Ds for Pb and Ba are similar to mineral/melt values. Cpx/fluid fractionates U/Th more strongly than cpx/melt. Cpx/silicate and cpx/carbonate melt Ds appear similar, except for Zr and Hf, which are fractionated in opposite direction. For amphibole (amph) (Fig. 4) Rb and Pb mineral/fluid and mineral/melt Ds are close in value, but Rb/Ba behaviour is distinctly different for amph/fluid (<1) and each melt (>1 for carbonate melt, ~1 for silicate melt). Also Nb/Ta is fractionated more by amph/fluid and amph/carbonate melt than by amph/silicate melt, and HFSE/REE is higher for amph/carbonate pairs than for either amph/silicate or amph/fluid pairs. For garnet (gt) (Fig. 5) mineral/fluid Ds for Ba and Sr are lower than mineral/melt values, and U/Nb and Pb/Sr will decrease in fluids but will increase in melts through gt fractionation. Gt/silicate or carbonate melt Ds show very similar behaviour. Although only a small number of rutile/fluid and melt D values is available, the very high D values are striking, so that a relatively small volume of rutile may have a significant effect on trace element behaviour. Rutile/fluid fractionates Nb/Ta in the opposite direction to rutile/melt and to cpx, amph or gt/melt. Thus rutile/fluid fractionation will show a decrease in Nb/Ta compared with an increase in Nb/Ta for all the mineral/melt fractionating cases. Also rutile/fluid DU >> DTh, that is opposite to cpx/fluid, but similar to (though much higher than) gt/fluid.

CONSTRAINTS ON FLUID VS MELT ROLE IN SUBDUCTION ZONE PROCESSES

The cpx, amph, gt/fluid or melt D data indicate that for potential fluids or melts that could affect the peridotitic mantle wedge source region for SZ volcanics, relatively lower Rb/Ba and HFSE/REE in the SZ volcanics point to a carbonatitic melt modifying role, whereas higher U/Th and Rb/Ba and lower U/Nb suggests a fluid role. The similarity of trace element behaviour in fluid and trondhjemite indicates that there will be relatively little difference discernible between the role of high-pressure aqueous fluid and a low-degree trondhjemitic high-pressure melt, in terms of modifying the trace element composition of the IAB source region. This generalization does not hold if trace element-enriched accessory minerals (e.g. rutile) remain residual during the derivation of the fluid or melt. The relatively lower T of sub-solidus fluid-related processes, compared with melt-related processes, enhances the likely role of accessory minerals in the source regions for the fluids. This contrasts with the higher T melting situation where the greater solubility of the accessory minerals may strictly limit their potential for affecting the trace element contents of the derived melts.

If rutile/fluid partitioning behaviour exerts an important control on the geochemistry of the IAB or SZ volcanics source region, then derived magmas may have Nb/Ta < model mantle, whereas if rutile/melt control (together with cpx, amph or gt) is more significant than Nb/Ta will be > model mantle. Thus the recently obtained Nb/Ta data for IAB of 11 to 33 may reflect this contrasting rutile/fluid or melt Nb and Ta partitioning behaviour. An important corollary is that a model continental crustal value of Nb/Ta ( 11 would suggest that any major contribution to the growth of continental crust from SZ volcanism should come from magmas derived from a rutile/fluid affected source region. Additional evidence for a fluid rather than a silicate melt role may come from careful assessment of Rb/La, U / Th and Pb/Sr relative to Nb/Ta. The fluid/melt partitioning data summarized here suggest that a negative correlation of these ratios would confirm that fluid-linked trace element behaviour was the controlling factor.

REFERENCES

Adam, J., Green , T.H., Sie, S.H., & Ryan, C.G., 1996. Trace element partitioning between aqueous fluids, silicate melts and minerals. Eur.J.Mineral. (submitted).

Ayers, J.C., & Eggler, D.H., 1995. Partitioning of elements between silicate melt and H20-NaCl fluids at 1.5 and 2.0 GPa pressure: Implications for mantle metasomatism Geochim.Cosmochim. Acta, 59, 4237-4246.

Brenan, J.M., Shaw, H.F., Phinney, D.L., & Ryerson, F.J., 1994. Rutile-aqueous fluid partitioning of Nb, Ta, Hf, Zr, U and Th: implications for high field strength element depletions in island-arc basalts. Earth Planet. Sci. Lett. 128, 327-339.

Brenan, J.M., Shaw, H.F. Ryerson, F.J., & Phinney, D.L., 1995. Mineral-aqueous fluid partitioning of trace elements at 900(C and 2.0GPa: Constraints on the trace element chemistry of mantle and deep crustal fluids. Geochim.Cosmochim. Acta 59, 3331-3350.

Eggins, S.M., Woodhead, J.D., Kinsley, L., Sylvester, P., McCulloch, M.T., Hergt, J., & Handler, M., 1996. A simple method for the precise determination of 40 or more trace elements in geological samples by ICP-MS using enriched isotope internal standardisation. Chem.Geol. (in press).

Green, T.H., 1994. Experimental studies of trace-element partitioning applicable to igneous petrogenesis ó Sedona 16 years later. Chem.Geol. 117, 1-36.

Hawkesworth, C.J., Hergt, J.M., Ellam, R.M., & McDermott, F., 1991. Element fluxes associated with subduction related magmatism. Phil.Trans.R.Soc.Lond. A. 335, 393-405.

Jenner, G.A., Foley, S.F., Jackson, S.E., Green, T.H., Fryer, B.J., & Longerich, H.P., 1993. Determination of partition coefficients for trace elements in high pressure-temperature experimental run products by LAM-ICP-MS. Geochim.Cosmochim. Acta 57, 5099-5103.

Keppler, H., 1996. Constraints from partitioning experiments on the composition of subduction-zone fluids. Nature 380, 237-240.

McCulloch, M.T., & Gamble, J.A., 1991. Geochemical and geodynamical constraints on subduction zone magmatism. Earth Planet.Sci.Lett. 102, 358-374.

Maury, R.C., Defant, M.J., & Joron. J-L, 1992. Metasomatism of the sub-arc mantle inferred from trace elements in Philippine xenoliths. Nature 360, 661-663.

Pearce, J.A., Baker, P.E., Harvey, P.K., & 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.

Stolper, E., & Newman, S., 1994. The role of water in the petrogenesis of Mariana trough magmas. Earth Planet.Sci.Lett. 121, 293-325.

Stolz, A.J., Jochum, K.P., Spettel, B., & Hofmann, A.W., 1996. Fluid-and melt-related enrichment in the subarc mantle: evidence from Nb/Ta variation in island-arc basalts. Geology,24, 587-590.

Acknowledgments: The high-pressure experimental research involving determination of partition coefficients between minerals and melts or fluids has been supported by research grants from the Australian Research Council and Macquarie University. All of the data from Macquarie University used in this review has been obtained in collaboration with Drs. J. Adam, A. Chekhmir, A. Fujinawa, G. Nichols, N. Pearson, C. Ryan, S. Sie and E. Vicenzi and their contribution and interest is gratefully acknowledged.

Table and figures available from Trevor Green.

Back to the GEMOC Abstract Titles Page