Jonathan D. Blundy

Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase feldspar, silicate melts and hydrothermal solutions

The isothermal (750°C) experiments of Lagache and Dujon (1987) reveal that the partitioning of Sr between plagioclase feldspar and hydrothermal solutions is a funtion of the anorthite (An) content of the plagioclase, indicating that crystal chemistry may exert a powerful influence on trace element partitioning. In order to compare these results with those on trace element partitioning between plagioclase and silicate melts we have compiled from the literature a large dataset of experimental and volcanic distribution coefficients (Ds) for Sr (and Ba). These data, which span a compositional range from lunar basalt to high silica rhyolite and a temperature range of over 650°C, show a relationship between DSr (and DBa) and mole fraction An (XAn) which is similar to that exhibited by the hydrothermal results obtained at constant temperature. Plots of In DSr and In DBa versus XAn are linear with negative slope, indicating that both elements are more compatible in albite than anorthite. In terms of molar distribution coefficients (DSr∗) the hydrothermal and silicate melt data display an identical linear relationship between RT In DSr∗ (where T is the absolute temperature in K and R is the gas constant, 8.314 JK−1 mol−1) and XAn. We conclude therefore that crystal chemistry provides the dominant control on partitioning of Sr and Ba into plagioclase and that the effects of temperature, pressure, and fluid composition are minor. Apparent relationships between DSr (and DBa) and the reciprocal temperature (1/T) are artefacts of the linear relationships between XAn and 1/T in the experimental studies. By defining a Henrys law standard state for the silicate melts and hydrothermal solutions, and considering plagioclases to be ternary regular solutions, we are able to relate the observed relationships between RT In Di∗ (where i is Ba or Sr) and XAn to the excess free energies of the trace element partitioning reactions between plagioclase and melt or hydrothermal solution. The interaction parameters are consistent with simple models in which the larger Ba or Sr cations are accommodated by lattice strain in the host plagioclase lattice, which is assumed to be perfectly elastic and isotropic. Thus Di∗ is a function of the Youngs modulus of the host crystal and the size mismatch between trace and host cations. The greater elasticity of albite relative to anorthite accounts for the observed preference of Sr and Ba for sodic plagioclases over calcic plagioclases. For geochemical purposes the weight fraction partition coefficient Di is of more value than its molar counterpart. Regression of the Di data versus XAn yields the semi-empirical relationships RTIn DSr = 26,800 − 26,700 · XAnRTIn DBa = 10,200 – 38,200 · XAn. Thus measurement of the An and trace element (Ba, Sr) contents of a magmatic plagioclase enables calculation of the Ba and Sr contents of the coexisting liquid, which can be extremely important in the deciphering of igneous processes. By reference to plagioclase fractionation in the simple An-Ab binary we show that failure to take into account the compositional dependence of DSr can result in erroneous interpretations of geochemical trends. We also consider applications to three natural igneous suites: the Aden Volcanics; the layered Kiglapait Intrusion, Labrador; and the southern Actamello Massif, Italy.

Experimental constraints on major and trace element partitioning during partial melting of eclogite

Isobaric partial melting experiments were performed on an Fe-free synthetic composition to simulate partial melting of subducted oceanic crust. Nominally anhydrous experiments at 3.0 GPa yielded melts in equilibrium with garnet (13 to 16 mol.% grossular) and aluminous clinopyroxene (14 to 16 wt.% Al2O3). Melt compositions show decreasing Si and alkalis and increasing Ca, Mg, and Ti contents with increasing temperatures. Experiments at 1200 and 1300°C were rutile saturated, whereas experiments at 1400°C contained no residual rutile. We argue that during the initial stages of subduction, accessory rutile is likely to be stable in subsolidus eclogites of average midocean ridge basalt composition and that only large degrees of partial melting will eradicate rutile from an eclogitic source. At 3 GPa, any eclogites with a bulk TiO2 content of 1.5 wt.% rutile will produce rutile-saturated partial melts, except at very high degrees of melting. At higher pressures, all bulk Ti may dissolve in clinopyroxene and garnet, leaving no accessory rutile. Trace element partition coefficients for 24 trace elements between clinopyroxene, garnet, and melt were determined by secondary-ion mass spectrometry analysis of experimental run products at 1400°C and 3 GPa. Partition coefficients for the rare earth elements agree well with previous studies and have been evaluated using the lattice strain model. Partitioning data for high-field strength elements indicate complementary DZr/DHf for clinopyroxene and garnet. Partial melting of an eclogitic component of different modal compo- sitions may therefore explain both subchondritic and superchondritic Zr/Hf ratios. Superchondritic Zr/Hf has recently been observed in some ocean island basalts (OIB), and this may be taken as further evidence for components of recycled oceanic crust in OIB. The data also indicate slight Nb/Ta fractionation during partial melting of bimineralic eclogite, which is not, however, sufficient to explain some recently observed Nb/Ta fractionation in island arc rocks. Accessory rutile, however, can explain such fractionation. Copyright

Mineral-Melt Partitioning of Uranium, Thorium and Their Daughters

The uranium and thorium decay series (hereafter “U-series”) include the nuclides of ten elements, all of which can be found at trace levels in rocks and minerals. The relatively short half-lives of the U-series nuclides give them considerable potential to decipher a wide variety of natural processes. The common observation of secular radioactive disequilibrium between parent and daughter nuclides provides a time dimension that is not possible with the more commonly used trace elements. However, just like conventional trace elements, the behavior of U-series elements depends on their partitioning between coexisting phases, such as minerals and melts. Interpreting radioactive disequilibrium behavior of the U-series critically requires an understanding of how parent and daughter nuclides of these elements are fractionated one from another under the conditions of interest. Without appropriate partition coefficients ( D ) it is difficult to separate that part of any disequilibrium signal that is due to process and that part which is due to time. This problem is minimized, but by no means eliminated, by the use of activity ratios rather than concentration ratios, as conventionally used for trace elements. But still, there is very little point in determining isotopic concentrations at the sub-femtogram level, if the data themselves cannot be interpreted or modeled with comparable precision and accuracy. Unfortunately, with the partial exception of U, Th and Pb, our knowledge of partitioning of the U-series elements lags well behind our ability to measure them. The problem is exacerbated by the fact that, because nearly all of the elements of interest are highly incompatible ( D << 1) in all common silicate and oxide minerals, and many lack stable or long half-life isotopes, there are serious technical difficulties associated with determining their partition coefficients experimentally The purpose of this chapter is to first establish a case for the importance of …

The effect of Ca-Tschermaks component on trace element partitioning between clinopyroxene and silicate melt

Abstract We have studied the influence of Ca-Tschermaks (Calcium Tschermaks or CaTs) content of clinopyroxene on the partitioning of trace elements between this phase and silicate melt at fixed temperature and pressure. Ion probe analyses of experiments carried out in the system Na2O–CaO–MgO–Al2O3–SiO2, at 0.1 MPa and 1218°C, produced crystal-melt partition coefficients (D) of 36 trace elements (Li, Cl, Sc, Ti, V, Cr, Fe, Co, Ge, Sr, Y, Zr, Nb, Mo, Ru, Rh, In, Sn, Sb, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, Ta and W), for clinopyroxene compositions between 10 and 32 mol% CaTs. Partition coefficients for 2+ to 5+ cations show, for each charge, a near parabolic dependence of log D on ionic radius of the substituting cation, for partitioning into both the M1 and M2 sites of clinopyroxene. Fitting the results to the elastic strain model of Blundy and Wood [Blundy, J.D., Wood, B.J., 1994. Prediction of crystal-melt partition coefficients from elastic moduli. Nature 372, 452–454] we obtain results for the strain-free partition coefficients of theoretical cations (D0), with site radius r0, and for the sites Youngs Modulus (E). In agreement with earlier data our results show that increasing ivAl concentration in cpx is matched by increasing D, EM1, EM2 and D0 for tri-, tetra- and pentavalent cations. The degree of fractionation between chemically similar elements (i.e. Ta/Nb, Zr/Hf) also increases. In contrast, D values for mono-, di- and hexavalent cations decrease with increasing ivAl in the cpx. The large suite of trace elements used has allowed us to study the effects of cation charge on D0, r0 and E. We have found that D0 and r0 decrease with increasing cation charge, e.g. r0=0.66 A for 4+ cations and 0.59 A for 5+ cations substituting into M1. Values of EM1 and EM2 increase with cation charge as well as with increasing ivAl content. The increase in EM2 is linear and close to the trend set by Hazen and Finger [Hazen, R.M., Finger, L.W., 1979. Bulk modulus-volume relationship for cation–anion polyhedra. J. Geophys. Res. 84 (10) 6723–6728] for oxides. EM1 values are much higher and do not fit the trend predicted by the Hazen and Finger relationship.

THE ROLE OF CLINOPYROXENE IN GENERATING U-SERIES DISEQUILIBRIUM DURING MANTLE MELTING

Abstract We develop recent models of crystal-liquid partitioning Blundy and Wood 1994 , Wood and Blundy 1997 to predict the effects of crystal chemistry on the partitioning of U-series elements between clinopyroxene and coexisting silicate melt. With increasing pressure along the mantle solidus, clinopyroxene becomes richer in Na2O and Al2O3 and these components are predicted to reduce the size of the large M2 site. This means that, because U4+ is smaller than Th4+, clinopyroxene should progressively discriminate in favour of U4+ with increasing pressure. We find that the effect is sufficiently great that DU/DTh should change from 1.0 above 1.5 GPa. Thus, melting in the stability field of clinopyroxene should produce excess 238U relative to 230Th at low pressure and excess 230Th above 1.5 GPa. Observed excesses of 230Th in MORB may not, therefore, require the presence of garnet in the source region. We performed experiments at 1.5–1.9 GPa on compositions doped with U and Th to test the calculations. Crystals and quenched melts were analysed by ion microprobe. On the mantle solidus at 1.5 GPa, DU/DTh for clinopyroxene was found to be 1.07 ± 0.08, while a value of 1.19 ± 0.04 was obtained at 1.9 GPa. These results confirm the prediction that mantle clinopyroxenes have DU/DTh > 1.0 at high pressure. Absolute values of DU for clinopyroxene are 0.029 at 1.5 GPa and 0.023 at 1.9 GPa in good agreement with earlier measurements on aluminous clinopyroxenes. Corresponding values for orthopyroxene at 1.5 GPa are 1.7 × 10−3 for DU and 8 × 10−4 for DTh while olivine gave DU of 1.8 × 10−5 and DTh of 1.2 × 10−5 under similar conditions. When the new data are used in dynamic melting calculations we find that 230Th/238U activity ratios of up to about 1.23 in the liquid are consistent with melting in the spinel lherzolite field provided porosity is approximately 10−3 and melting rate ≤ 3 × 10−5 kg. m−3 yr−1. This confirms our model prediction that residual garnet is not required to generate excess 230Th in mantle melts. Solid-liquid partition coefficients for other members of the U decay series, Ra and Ac may be calculated from the model. For clinopyroxene we obtain DRa of about 10−6 and DAc of 5 × 10−4. The value for Ac is large enough that 227Ac activities may provide information on melt extraction processes. The M2 site of orthopyroxene is found to be elastically similar to the corresponding clinopyroxene site, which enables us to calculate for this phase, DRa of 4 × 10−10 and DAc of 2 × 10−6.

The effect of cation charge on crystal–melt partitioning of trace elements

In this paper we develop a simple theoretical model to explain two of the most important observations about trace element partitioning. The first of these is the apparent dependence of crystal–melt partition coefficients on charge such that when a cation such as Ca2+ is replaced by ions of approximately the same radius but different charge, partitioning follows the approximate relationship: D2+Ca>Di3+≈Dj1+>Dk4+≥Dn0+ where n refers to a noble gas atom. The second observation is the strong correlation between partition coefficients for highly charged ions and the bulk composition of the crystal, e.g. the correlation between DTh and Al content of clinopyroxene. Starting from the substitution of an ion of charge Zc into a site which normally accommodates a major ion of different charge Zo, we find, using a simplified continuum model [Born, Z. Phys. 1 (1920) 45–48] that the electrostatic work of substitution should depend on (Zc−Zo)2. Observed partitioning between melt and the clinopyroxene M2 site for ions of similar radius to Ca2+ follows an approximately parabolic dependence on cation charge, implying that electrostatic work is a major influence on partitioning behaviour. From the form of the parabola we derive an electrostatic energy of substitution ΔGelec in clinopyroxene M2 of approximately 28 kJ mol−1. One implication is that, if noble gases enter lattice sites in minerals, they should have broadly similar crystal–melt partition coefficients to U and Th. In any given natural pyroxene, with random mixing of major ions on each sublattice, the ‘best-fit’ charge at the M2 site depends on the composition of the local environment. From the bulk crystal composition we can compute the proportions of M2 sites which, for local electroneutrality, should contain 1+, 2+, 3+, 4+ or 0+ ions. By summing the proportions of the different configurations and weighting them according to their relative electrostatic energies, the effects of crystal composition on trace element partitioning can be predicted. The model predicts that the clinopyroxene–melt partition coefficients of rare earth elements (REE3+) should increase by a factor of 2.3 as the Al content of tetrahedral sites (Aliv) increases from 0.001 to 0.3. For more highly charged cations the effect is more dramatic, the partition coefficient for Th4+ being predicted to increase by a factor of 300 over the same composition range. In both cases the predictions agree well with experimental observations. When the effects of pressure are combined with those of composition we find that noble gas partition coefficients should increase with pressure relative to those of U and Th and plausibly become larger than the latter. We conclude that the electrostatic work of substitution exerts a large and predictable influence on the partitioning behaviour of trace cations. Together with lattice strain energy, it should be explicitly accounted for in the development of thermodynamic descriptions of partitioning behaviour.

Trace element evidence for plagioclase recycling in calc-alkaline magmas

Abstract Trace element zoning in plutonic plagioclase feldspars from the Val Fredda Complex, southern Adamello Massif, Italy has been measured using secondary-ion mass spectrometry (SIMS). The studied samples range from hornblende-gabbro through diorite to granitoid with an overall range in plagioclase composition from An96 to An15. Individual plagioclases display extensive zonation in anorthite through a considerable variety of zoning textures. There is equally extensive zoning in trace elements, notably Sr and Ba. Broadly similar trace element zoning patterns and concentrations are displayed by all plagioclases irrespective of host rock composition. Using a relationship betweenXAn and the distribution coefficientsDSr andDBa trace element concentrations in plagioclase are converted to those of the melt which precipitated each crystal zone. The calculated variations in Ba and Sr are greater than can be attributed to changes in pressure or temperature alone, indicating that compositional variation in the coexisting melt was the principal cause of zoning. The Ba Sr melt trend for each crystal describes a two-stage crystallisation history. The calcic cores of all crystals show melt trends of increasing Sr and Ba, which are attributed to mixing and fractionation in parental hydrous picrobasalt magmas. In contrast the melt trends recorded by the rims vary from sample to sample, but essentially show decreasing Sr and Ba consistent with fractionation of sodic plagioclase, alkali feldspar and/or biotite. The plagioclase cores are thought to have crystallised in a common magma chamber at 6–10 kbar pressure, while the rims crystallised at the emplacement level (1.5 kbar) from interstitial melt whose composition is determined by that of the host magma. The data suggest that early-formed calcic plagioclase was “recycled” by the magma during fractionation. Vigorous convection is invoked to both retain relatively dense calcic plagioclase within the evolving magma and to periodically re-entrain cumulus crystals into more evolved magmas. The occurrence of calcic plagioclase cores in felsic derivative magmas indicates that fractionation from picrobasalt to tonalite occurred on a relatively short time scale.

Asymmetric mantle dynamics in the MELT region of the East Pacific Rise

Abstract The mantle electromagnetic and tomography (MELT) experiment found a surprising degree of asymmetry in the mantle beneath the fast-spreading, southern East Pacific Rise (MELT Seismic Team, Science 280 (1998) 1215–1218; Forsyth et al., Science 280 (1998) 1235–1238; Toomey et al., Science 280 (1998) 1224–1227; Wolfe and Solomon, Science 280 (1998) 1230–1232; Scheirer et al., Science 280 (1998) 1221–1224; Evans et al., Science 286 (1999) 752–756). Pressure-release melting of the upwelling mantle produces magma that migrates to the surface to form a layer of new crust at the spreading center about 6 km thick (Canales et al., Science 280 (1998) 1218–1221). Seismic and electromagnetic measurements demonstrated that the distribution of this melt in the mantle is asymmetric (Forsyth et al., Science 280 (1998) 1235–1238; Toomey et al., Science 280 (1998) 1224–1227; Evans et al., Science 286 (1999) 752–756) at depths of several tens of kilometers, melt is more abundant beneath the Pacific plate to the west of the axis than beneath the Nazca plate to the east. MELT investigators attributed the asymmetry in melt and geophysical properties to several possible factors: asymmetric flow passively driven by coupling to the faster moving Pacific plate; interactions between the spreading center and hotspots of the south Pacific; an off-axis center of dynamic upwelling; and/or anomalous melting of an embedded compositional heterogeneity (MELT Seismic Team, Science 280 (1998) 1215–1218; Forsyth et al., Science 280 (1998) 1235–1238; Toomey et al., Science 280 (1998) 1224–1227; Wolfe and Solomon, Science 280 (1998) 1230–1232; Evans et al., Science 286 (1999) 752–756). Here we demonstrate that passive flow driven by asymmetric plate motion alone is not a sufficient explanation of the anomalies. Asthenospheric flow from hotspots in the Pacific superswell region back to the migrating ridge axis in conjunction with the asymmetric plate motion can create many of the observed anomalies.

Effect of Fe2+ on garnet–melt trace element partitioning: experiments in FCMAS and quantification of crystal-chemical controls in natural systems

Garnet–melt trace element partitioning experiments were performed in the system FeO–CaO–MgO–Al2O3–SiO2 (FCMAS) at 3 GPa and 1540°C, aimed specifically at studying the effect of garnet Fe2+ content on partition coefficients (DGrt/Melt). DGrt/Melt, measured by SIMS, for trivalent elements entering the garnet X-site show a small but significant dependence on garnet almandine content. This dependence is rationalised using the lattice strain model of Blundy and Wood [Blundy, J.D., Wood, B.J., 1994. Prediction of crystal–melt partition coefficients from elastic moduli. Nature 372, 452–454], which describes partitioning of an element i with radius ri and valency Z in terms of three parameters: the effective radius of the site r0(Z), the strain-free partition coefficient D0(Z) for a cation with radius r0(Z), and the apparent compressibility of the garnet X-site given by its Youngs modulus EX(Z). Combination of these results with data in Fe-free systems [Van Westrenen, W., Blundy, J.D., Wood, B.J., 1999. Crystal-chemical controls on trace element partitioning between garnet and anhydrous silicate melt. Am. Mineral. 84, 838–847] and crystal structure data for spessartine, andradite, and uvarovite, leads to the following equations for r0(3+) and EX(3+) as a function of garnet composition (X) and pressure (P): Accuracy of these equations is shown by application to the existing garnet–melt partitioning database, covering a wide range of P and T conditions (1.8 GPa<P<5.0 GPa; 975°C<T<1640°C). DGrt/Melt for all 3+ elements entering the X-site (REE, Sc and Y) are predicted to within 10–40% at given P, T, and X, when DGrt/Melt for just one of these elements is known. In the absence of such knowledge, relative element fractionation (e.g. DSmGrt/Melt/DNdGrt/Melt) can be predicted. As an example, we predict that during partial melting of garnet peridotite, group A eclogite, and garnet pyroxenite, r0(3+) for garnets ranges from 0.939±0.005 to 0.953±0.009 A. These values are consistently smaller than the ionic radius of the heaviest REE, Lu. The above equations quantify the crystal-chemical controls on garnet–melt partitioning for the REE, Y and Sc. As such, they represent a major advance en route to predicting DGrt/Melt for these elements as a function of P, T and X.

The effect of H2O on crystal-melt partitioning of trace elements

Abstract Many experimental data demonstrate distinct differences between crystal-liquid partition coefficients Di measured under high temperature, anhydrous conditions and those determined at lower temperatures in the presence of H2O. In this study we develop a thermodynamic method of separating the effects of H2O from those of temperature. We then apply the method to predict partitioning of REE between clinopyroxene, garnet and silicate melt over wide ranges of temperature, pressure and H2O content. Our initial inputs are the model of Wood and Blundy (1997) for REE partitioning between clinopyroxene and anhydrous melt and the melting temperatures of diopside on the join CaMgSi2O6-H2O (Eggler and Rosenhauer, 1978) . We then make the hypothesis that the effect of H2O on the activity of REE clinopyroxene component (REEMgAlSiO6) in the melt is the same as the measured effect on CaMgSi2O6 component. This leads to predictions of REE partition coefficients for clinopyroxene coexisting with hydrous melt at any P,T and H2O content up to 45 weight %. The results agree with observed REE partition coefficients with a standard deviation which is the same as that for the anhydrous data. We conclude, therefore, that the ‘H2O-effect’ may, in this case, be accurately predicted. We extend the approach to garnet by using the join Mg3Al2Si3O12-H2O to estimate the effects of H2O on all ‘garnet-like’ components in the melt. This enables calculation of garnet-melt REE partition coefficients for melts containing up to 25% H2O. The observation that H2O influences major and trace component activities in a similar manner enables us to make some generalisations about the combined effects on partitioning of decreasing temperature and increasing water content of the melt. The relative enthalpies of fusion ΔHf of major and trace components dictate whether trace element partition coefficients increase or decrease with H2O addition: ΔH f trace > ΔH f major gives increasing D trace with addition of H 2 O and ΔH f trace f major leads to decreasing D trace with addition of H 2 O. Note that Dtrace is strictly the ratio of mole fractions of charge-balanced components such as REEMgAlSiO6 in solid to melt phases. In the cases considered here, however, Dtrace closely approximates D = [trace] solid [trace] melt where [trace] refers to weight concentration. For clinopyroxene ΔHfREE