An ion microprobe study of the partitioning of trace elements between clinopyroxene and liquid in the system diopside-albite-anorthite

The partitioning of trace elements (Sc, Ti, Sr and Sm) between diopsidic clinopyroxene and liquid was studied experimentally in the system diopside-albite-anorthite at 1250°C, 1300°C and 1345°C at 1 atm. Twelve different bulk compositions were selected to study the effects of temperature and chemical composition. A Cameca ion microprobe was used to determine trace element concentrations in both clinopyroxene and liquid. Experiments of different run duration $\text{1}\text{4}\text{8}\text{?}$ days) showed that equilibrium was approached in less than 4 days at 1275°C. Equilibrium was also evaluated by a reversal run. A series of runs of constant bulk composition but with variable trace element contents showed that Henry's Law was obeyed over concentration ranges of the trace elements similar to those encountered in natural systems. The partition coefficients show significant ranges: Sc, 0.345~2.61; Ti, 0.084~0.214; Sr, 0.075~0.136; Sm, 0.054~0.328; the values are comparable with those obtained experimentally by other investigators. The partition coefficients vary as a function of both temperature and chemical composition. The experimental results are discussed in terms of exchange equilibria using the Bottinga-Weill silicate melt model. It is demonstrated that analytical uncertainties of both major and trace elements play an important role in understanding trace element exchange equilibria; propagation of analytical errors in the thermodynamic treatment is equally important.     

Experimental determination of rare earth element partitioning between hydrous silicate melt,amphibole and garnet peridotite minerals at upper mantle pressures and temperatures

Partition coefficients of Ce, Sm and Tm involving garnet peridotite minerals, amphibole and hydrous silicate melt have been determined experimentally in the temperature and pressure ranges 950–1075°C and 10–25 kbar.Only several parts per million to several tens of parts per million of rare earth element (REE) can dissolve in the minerals before the crystal-liquid partition coefficients begin to vary as a function of REE content. The concentration ranges of constant partition coefficient increase with increasing temperature and are also positively correlated with the magnitude of the crystal-liquid partition coefficients. The upper concentration limits of constant partition coefficient and the value of the crystal-liquid partition coefficient for REE decrease in the order garnet > clinopyroxene > amphibole > orthopyroxene > olivine.Partition coefficients may vary by at least an order of magnitude as a function of bulk composition of the liquid phase (e.g. changing from basaltic to andesitic). The approximate ranges of the values of the partition coefficients as a function of bulk liquid composition are as follows:

$\text{Ce}\text{Sm}\text{Tm}\text{K}{}^{\mathrm{<mtext>ga-liq</mtext>}}\text{0.01–0.1}\text{0.3–3.4}\text{1–10}\text{K}{}^{\mathrm{<mtext>cpx-liq</mtext>}}\text{0.05–0.4}\text{0.09–0.7}\text{0.04–0.4}\text{K}{}^{\mathrm{<mtext>amph-liq</mtext>}}\text{0.04–0.4}\text{0.08–0.8}\text{0.07–0.7}\text{K}{}^{\mathrm{<mtext>opx-liq</mtext>}}\text{0.04–0.1}\text{0.05–0.1}\text{0.08–0.1}\text{K}{}^{\mathrm{<mtext>ol-liq</mtext>}}\text{0.01–0.02}\text{0.01–0.02}\text{0.01–0.02}$

where the values increase with increasing acidity of the melt.     

Rare earth diffusion kinetics in garnet: Experimental studies and applications

We determined the diffusion coefficient of Sm in almandine garnet as function of temperature at 1 bar and fO₂ corresponding to that of wüstite-iron buffer, and to a limited extent, that of a few other selected rare earth elements in almandine and pyrope garnets. Both garnets were demonstrated to have metastably survived the diffusion annealing at conditions beyond their stability fields. The experimental diffusion profiles were analyzed by secondary ion mass spectrometry, and in addition, by Rutherford back scattering spectroscopy for two samples. Transmission electron microscopic study of an almandine crystal that was diffusion-annealed did not reveal any near-surface fast diffusion path. Using reasonable approximations and theoretical analysis of vacancy diffusion, the experimental data were used to develop an expression of rare earth element (REE) diffusion coefficient in garnet as a function of temperature, pressure, fO₂, ionic radius, and matrix composition. Calculation of the closure temperature for the Sm-Nd decay system in almandine garnet in a metamorphic terrain shows very good agreement with that constrained independently. Modeling of the REE evolution in melt and residual garnet suggests that for dry melting condition, the REE pattern in the melt should commonly conform closely to that expected for equilibrium melting. However, for much lower solidus temperatures that would prevail in the presence of a H₂O-CO₂ fluid, the concentration of light REE in the melt could be significantly lower than that under equilibrium melting condition. A reported core and rim differences in the REE content of a garnet crystal in a mantle xenolith in kimberlite have been reproduced by assuming that the REE zoning was a consequence of entrapment in a magma derived from an external source for ∼32,000 yr before the eruption.     

CaAl ratio and composition of the Earth's upper mantle

Undifferentiated meteorites (chondrites) have the same relative abundances of refractory lithophile elements (Ca, Al, Ti, Sc, REE, etc.), despite variable absolute concentrations. The reasonable assumption of chondritic ratios among refractory elements in the bulk Earth is used to constrain the chemical composition of the upper mantle in the following way: Correlations of the compatible refractory elements Ca, Al, Ti, Sc and Yb with MgO are worldwide very similar in suites of spinel-lherzolite xenoliths from basaltic rocks. Such suites represent upper mantle material depleted to differing degrees by extraction of partial melts. From these refractory elements vs. MgO correlations, ratios of pairs of refractory elements were calculated at various MgO contents. Chondritic $\text{Al}\text{Ti}$ and $\text{Sc}\text{Ti}$ ratios were only obtained for MgO contents below 36%. A chrondritic $\text{Sc}\text{Yb}$ ratio requires an MgO content above 35%. We therefore accept 35.5% as the most reasonable MgO content of undepleted upper mantle. This MgO content is slightly below the spinel-lherzolite with the lowest measured MgO content (36.22%). The corresponding Al₂O₃ content of 4.75% is higher than in previous estimates of upper mantle composition. The concentrations of other elements were obtained from similar correlations at a MgO content of 35.5%. The resulting present upper mantle composition is enriched in refractory elements by a factor of 1.49 relative to Si and Cl and by a factor of 1.12 for Mg relative to Si and Cl. These enrichments are in the same range as those for the Vigarano type carbonaceous chondrites. The Mg/Mg + Fe ratio of 89 is slightly lower than previous estimates.The $\text{Ca}\text{Al}$ ratio in spinel lherzolite suites is, however, uniformly higher worldwide than the chondritic ratio by about 15%. Orogenic peridotites as well as komatiites appear to have similar non-chondritic $\text{Ca}\text{Al}$ ratios. It is therefore suggested that this non-chondritic $\text{Ca}\text{Al}$ ratio is a characteristic of the upper mantle, possibly since the Archean. A minor fractionation of about 4% of garnet in an early, global melting event (deep magma ocean?) is presented as the most likely cause for the high $\text{Ca}\text{Al}\text{-}\text{ratio}$. In this case the addition of 4% of such a garnet component to the undepleted present upper mantle would be required to obtain the composition of the primordial upper mantle. The $\text{Ca}\text{Al}\text{-}\text{ratio}$ of this primordial mantle would be 15% higher than that of the undepleted present upper mantle, resulting in an enrichment of refractory elements of 1.70 ($\text{Al}\text{Si}$ relative to Cl) for the primordial upper mantle.     

Chemical equilibration of the Earth's core and upper mantle

The oxygen fugacity (fO₂) of the Earth's upper mantle appears to lie somewhat above that of the iron-wüstite buffer, its fO₂ is assumed to have been similar to the present value at the time of core formation. In the upper mantle, the Fe-rich liquid protocore that would form under such conditions of fO₂ at elevated temperatures would lie predominantly in the system Fe-S-O. Distribution coefficients for Co, Cu, Ni, Ir, Au, Ir, W, Re, Mo, Ag and Ga between such liquids and basalt are known and minimum values are known for Ge. From these coefficients, upper mantle abundances for the above elements can be calculated by assuming cosmic abundances for the whole Earth and equilibrium between the Fe-S-O protocore and upper mantle. These calculated abundances are surprisingly close to presently known upper mantle abundances; agreements are within a factor of 5, except for Cu, W, and Mo. Therefore, siderophile element abundances in the upper mantle based on known distribution coefficients do not demand a late-stage meteoritic bombardment, and a protocore formed from the upper mantle containing S and O seems likely.As upper mantle abundances fit a local equilibrium model, then either the upper mantle has not been mixed with the rest of the mantle since core formation, or else partition coefficients between protocore and mantle were similar for the whole mantle regardless of $\text{P}$, $\text{T}$, and fO₂. The latter possibility seems unlikely over such a $\text{P-T}$ range.     

Experimental investigation of the partitioning of phosphorus between metal and silicate phases: implications for the Earth,Moon and Eucrite Parent Body

The solid metal/silicate melt partition coefficient for P, D(P), has been determined experimentally at 1190°C and 1300°C. The dependence of the partition coefficient on oxygen fugacity has been investigated, and is consistent with a valence state of 5 for P in the silicate melt. The experimental partition coefficients are given by: $\text{log D(P) = ?1.21 log ?O}{}_2\text{? 15.95 at 1190°C}$$\text{log D(P) = ?1.53 log ?O}{}_2\text{? 17.73 at 1300°C}$The experimentally determined partition coefficients may be used to interpret the low $\text{P}\text{La}$ ratios of the Earth, Moon and eucrites relative to C1 chondrites. The low $\text{P}\text{La}$ ratios in the eucrites may be explained by partitioning of P into 5% to 25% of a sulfur-bearing metallic liquid assuming equilibration and separation of the liquid metal from the silicates at low degrees of partial melting of the silicates. The low $\text{W}\text{La}$ ratios in the eucrites compared to C1 chondrites require the separation of an additional 2% to 10% solid metal.The lowering of both $\text{P}\text{La}$ and $\text{W}\text{La}$ ratios in the Moon may be explained by partitioning of P and W into metal during formation of a small core by separation of liquid metal from silicate at low degrees of partial melting of the silicates. The $\text{W}\text{La}$ ratios in the Earth and Moon are virtually indistinguishable, while $\text{P}\text{La}$ ratios differ by a factor of two. The concentrations of FeO also appear to be different. These observations are difficult to reconcile with the hypothesis of a terrestrial origin of the Moon following formation of the Earth's core, but are consistent with independent formation of the Earth and Moon.     

Garnet + liquid equilibrium

New experiments were performed to determine saturation conditions for garnet and silicate liquid. Starting compositions were natural basalt powders ranging from komatiite to nephelinite, which were partially melted at pressures between 25 and 100 kbar. Rounded grains of natural pyrope or grossular were added to some experiments to induce garnet saturation, and to aid the segregation of liquid pools for microprobe analysis. Simple expressions describing K _eq as a function of P, T and liquid composition were calibrated by linear least squares analysis of the data from this, and other, studies. Since garnets do not often occur as phenocrysts, equations were designed to predict garnet compositions when P, T and a silicate liquid composition are given. The regression data have a pressure range of 20–270 kbar, and compositions as diverse as nephelinite and komatiite. These models should thus apply to a broad range of geological problems. The majorite component in garnet was found to increase with increasing P, but compositional effects are also important. A garnet saturation surface applied to liquids with chondritic compositions shows that such liquids crystallize garnet with Mj contents of 0.27–0.42 at 200 kbar. Models of Earth differentiation thus need to account not only for fractionation of majorite, but also for Fe-, Ca-, Na- and Ti-bearing garnet components, which occur in non-trivial quantities at high pressure. Since many models of igneous petrogenesis rely on mineral-melt partition coefficients for the minor elements Na, Ti, and Cr, partition coefficients for these elements were also examined. The K _d ^gar/liq for Na was found to be P-sensitive; Na contents of basalts may thus potentially yield information regarding depths of partial melting. Received : 28 May 1997 / 25 November 1997     

SIMS analyses on trace and rare earth elements in coexisting clinopyroxene and mica from minette mafic enclaves in potassic syenites crystallized under high pressures

Trace and rare earth element contents were determined by SIMS technique in clinopyroxene and mica crystals from minette lamprophyric enclaves in a potassic syenite host. This co-mingled system was crystallized at high pressures, which varied about 3–5 GPa, as indicated by the presence of K-clinopyroxenes and pyrope-rich garnet with measurable amounts of K₂O and Na₂O, among the near-liquidus phases. Major and trace element composition of these lamprophyric enclaves is quite similar to those observed in silica-rich lamproites, suggesting that similar sources were involved in their origin. In a general view, the concentrations of most trace and rare earth elements in clinopyroxene of the studied enclaves are higher than those referred to by other authors. Clinopyroxene/melt partition coefficient for most trace elements are close to determinations in alkali-basalts and lamproites from Leucite Hills, with considerable differences relative to Gaussberg lamproites. Furthermore, these partition data are completely different from those determined for potassic lavas crystallized under crustal pressures. Spidergrams for clinopyroxenes exhibit negative-Sr anomalies relative to LREE, which have been associated by most authors to crystallization under low-pressures, out of garnet stability field. The presence of pyrope together with K-clinopyroxene excludes such hypotheses for the studied enclaves. Y and HREE are concentrated in clinopyroxene, whilst the other trace elements have Kd<1. LIL elements, except Rb, have incompatible (Kd<1) behavior in phlogopite. The high partition coefficient for Nb (Kd>3) determined in the studied phlogopite is unusual in lamproites, lamprophyres, and basalts, but frequently observed in phlogopite from metasomatic mantle samples, as well as in acid magmas. This partition value may indicate the lack of other mineral phase with high partition for this element during crystallization, and may be enhanced by the liquid composition progressively closer to alkali feldspar, an unsuitable structure for six-coordinated cations. Ce/Yb, Rb/Sr, and Zr/Hf ratios in clinopyroxene and mica suggest that the minettic magma could produce the host Piquiri potassic syenite by fractional crystallization. This hypothesis is not consistent with Ba concentrations in clinopyroxene and mica, which suggest that a Ba-bearing phase (e.g. alkali feldspar) should be among the fractionated phases in order to produce the potassic syenites.     

The genetic significance of almandine-pyrope phenocrysts in the calc-alkaline Borrowdale Volcanic Group,Northern England

A suite of garnet-bearing andesites and dacites from the Ordovician of N. W. England is described and major- and trace-element analyses of the garnet phenocrysts are presented. The garnets are of almandine-pyrope composition, with minor amounts of spessartine and grossular, and often show marked reversed zoning; the crystal becoming progressively enriched in pyrope towards the margin. Garnets from the dacites are consistently richer in almandine and spessartine than are those from the andesites.From a consideration of the chemistry of the garnet phenocrysts and host rocks, especially La and Y abundances, it is shown that garnet could not have been removed from the magma in quantities sufficient to affect the liquid composition. Consequently the magma must have evolved by some process other than crystal fractionation. It is proposed that the magma was generated by the partial melting of oceanic crust along an ancient Benioff zone, stored at depth (possibly at the crust/mantle interface) long enough for garnet to nucleate, and then transferred rapidly to the surface. Isobaric crystallisation of the garnet phenocrysts at depth could explain the reversed zoning observed.     

Matrix corrections in trace-element analysis by X-ray fluorescence: An extension of the Compton scattering technique to long wavelengths

Mass absorption coefficients (A₂) for a series of standard rocks, have been calculated in the wavelength region 0.492–3.03 $\text{a}\text{?}$. Plots of these data against the intensity of the Compton scattered peak [(I) Compton] give an excellent correlation for the wavelengths 0.429 $\text{a}\text{?}$ to the Fe-absorption edge (1.74 $\text{a}\text{?}$); the data confirm the observations of Reynolds. Hence, routine measurement of one peak will give the mass absorption coefficient of a sample in an analytically important region (Sn/1bK_α to Ni/1bK_α). A₂ has also been directly measured on three of the samples and systematic differences between calculated and measured are attributed to the measuring technique. At wavelengths longer than the Fe-absorption edge, (up to 3.03 $\text{a}\text{?}$) A₂ can be estimated using a combination of (I) Compton and Fe/1bK_α c.p.s. This technique enables meaningful matrix corrections to be carried out on the elements Co, Mn, Cr, V, Ti, Sc (K spectra) and Ba (L spectra). Cr and Ba results are presented for some standard rocks.     

Trace element abundances and mineral/melt distribution coefficients in phonolites from the Laacher See volcano (Germany)

Twenty six whole rocks, seven matrix and fifty three mineral separates from the compositionally zoned late Quaternary Laacher See tephra sequence (East Eifel, W Germany) were analyzed by instrumental neutron activation. These data document the chemical variation within the Laacher See magma chamber prior to eruption with a highly fractionated phonolite at the top and a more mafic phonolite at its base as derived from other data. Incompatible elements such as Zn, Zr, Nb, Hf, U, light and heavy rare earths are extremely enriched towards the top whereas compatible elements (e.g. Sr, Sc, Co, Eu) are strongly depleted. Semicompatible elements (Ta and some middle REE) are depleted at intermediate levels. This chemical variation is shown by whole rock and matrix data indicating the phonolite liquid was compositionally zoned regardless of phenocryst content. Hybrid rocks (phonolite-basanite) show the largest concentrations for compatible elements. All elements (except Rb) display continuous compositional variations with regard to the stratigraphic position of pumice samples. From these data we are able to distinguish three main units: An early erupted highly fractionated magma, the main volume of evolved phonolite and a mafic phonolite as the final products. The extreme variation of trace element distribution coef ficients (K) for 9 mineral phases with respect to stratigraphic position (resp. matrix composition) cannot be explained by conventional mechanisms. We postulate a significant modification of the trace element content of the phonolite melt by liquid-liquid controlled differentiation processes subsequent to and/or contemporaneous with (fractional) crystallization which caused disequilibrium between phenocrysts and host matrix. Therefore, our “distribution coefficients” deviate from equilibrium partition coefficients equivalent to the amount of this post crystallization modification of the matrix composition. The relationship between varying K and matrix composition is demonstrated by a ΔK-ΔM-diagram (variation of K versus variation of matrix, M). Different parts of this diagram relate to different parameters (T, P, polymerization, complex-building, equilibrium crystallization in a zoned magma column and post crystallization disequilibrium effects) which are responsible for the variation of distribution coefficients in general. The ΔK-ΔM-diagram may allow to distinguish between different processes affecting the distribution coefficients measured in natural volcanic rocks from a differentiating magma system.     

Henry's law behavior in simple systems and in magmas: Criteria for discerning concentration-dependent partition coefficients in nature

Although numerous questions still surround the topic of Henry's law (HL) as it applies to trace element partitioning, there now exist sufficient experimental data to make some generalizations regarding HL behavior in minerals. The most important of these is that the commonly-observed failure of HL at low concentration occurs at distinctly different levels even for chemically-similar elements in a single mineral. This observation in turn provides a basis for discerning effects of HL failure in natural systems: through examination of element ratios in minerals and rocks, it is possible, in principle, to distinguish HL effects from changes in partition coefficients due to variations in other magmatic parameters such as temperature and the compositions of phases. Initial applications of this approach to plagioclase/ liquid partitioning of REE and to the general behavior of $\text{Zr}\text{Hf}$ and $\text{Ba}\text{Rb}$ during basalt production suggest that HL usually does hold in nature.     

Simulation of thermodynamic mixing properties of garnet solid solutions at high temperatures and pressures

We present results of high temperature, high pressure atomistic simulations aimed at determining the thermodynamic mixing properties of key binary garnet solid solutions. Computations cover the pressure range 0–15 GPa and the temperature range 0–2000 K. Through a combination of Monte-Carlo and lattice-dynamics calculations, we derive thermodynamic mixing properties for garnets with compositions along the pyrope–almandine and pyrope–grossular joins, and compare these with existing experimental data. Across the pressure–temperature range considered, simulations show virtually ideal mixing behaviour in garnet on the pyrope–almandine join, while large excess volumes and enthalpies of mixing are predicted for garnet along the pyrope–grossular join. Excess heat capacities and entropies are also examined. These simulations shed additional light on the link between the behaviour at the atomic level and macroscopic thermodynamic properties: we illustrate the importance of certain atomistic Ca–Mg contacts in the pyrope–grossular solid solutions. For simulation techniques of this type to become sufficiently accurate for direct use in geological applications such as geothermobarometry, there is an urgent need for improved experimental determinations of several key quantities, such as the enthalpies of mixing along both joins.     

Trace element partitioning and melt structure: An experimental study at 1 atm pressure

Diopside-melt and forsterite-melt rare earth (REE) and Ni partition coefficients have been determined as a function of bulk compositions of the melt. Available Raman spectroscopic data have been used to determine the structures of the melts coexisting with diopside and forsterite. The compositional dependence of the partition coefficients is then related to the structural changes of the melt.The melts in all experiments have a ratio of nonbridging oxygens to tetrahedral cations ($\text{NBO}\text{T}$) between 1 and 0. The quenched melts consist of structural units that have, on the average, 2 (chain), 1 (sheet) and 0 (three-dimensional network) nonbridging oxygens per tetrahedral cation. The proportions of these structural units in the melts, as well as the overall $\text{NBO}\text{T}$, change as a function of the bulk composition of the melt.It has been found that Ce, Sm, Tm and Ni crystal-liquid partition coefficients ($\text{K}{}^{\mathrm{crystal?liq}}{}_{\mathrm{i}}\text{=}\text{C}{}^{\mathrm{crystal}}{}_{\mathrm{i}}\text{C}{}^{\mathrm{liq}}{}_{\mathrm{i}}$) decrease linearly with increasing $\text{NBO}\text{T}$. The values of the individual REE crystal-liquid trace element partition coefficients have different functional relations to $\text{NBO}\text{T}$, so that the degree of light REE enrichment of the melts would depend on their $\text{NBO}\text{T}$.The solution mechanisms of minor oxides such as CO₂, H₂O, TiO₂, P₂O₅ and Fe₂O₃ in silicate melts are known. These data have been recast as changes of $\text{NBO}\text{T}$ of the melts with regard to the type of oxide and its concentration in the melt. From such data the dependence of crystal-liquid partition coefficients on concentration and type of minor oxide in melt solution has been calculated.     

Trace and major elements in the sea-surface microlayer

Sea-surface microlayer samples were obtained in the North sea, by collecting droplets ejected by bubbles bursting at the sea-surface. The samples were analysed for some trace and major elements, mainly by neutron activation and atomic absorption spectrophotometry, and the results were compared with those from samples of bulk seawater taken at the same time. For any element X, the results may be expressed as a concentration factor related to Na, thus: $\text{CF =}\text{ratio of X and Na concentrations in microlayer sample}\text{similar ratio in corresponding bulk seawater sample}$The CF values for trace elements showed wide fluctuations from sample to sample. The only two elements for which relatively unambiguous CF values were obtained were Sc (from 2.5 to 100) and Pb (from 140 to 410). Other CF values were obtained for Co (up to 76). Zn (<50), La (up to 3000) and Ce (up to 500). The major ions Mg, Cl, K, Ca, Br, gave CF values between 0.54 and 2.2 in all cases measured. It is concluded that large enrichments of some trace elements can occur in the surface microlayer, but enrichment of major ions has not been observed.Concentrations of about 30 trace elements in particulate form in bulk seawater were measured in the course of the study.     

Silicate solid solutions and geothermometry

Principal component analysis, using eigenvalues and eigenvectors, of the encountered variation in the chemistry of 153 garnets indicates the following: (1) In low grade metamorphic rocks, spessertite and almandine are distinctly isomorphous. The variability of pyrope content does not influence the binary Fe-Mn relationship. However there may be some influence of the grossularite content. (2) In high grade metamorphic rocks, pyrope and almandine are distinctly isomorphous. Variability of grossularite content does not affect the binary Mg-Fe relationship. However, variability in spessertite content may influence the linearity particularly when there is little pyrope.Similar statistical analysis of the chemical data on 119 samples of clinopyroxenes indicates that only significant changes in the concentrations of Al in octahedral and/or of Al in tetrahedral sites and Ti could cause some change in the FeMg ratio in the mineral.Principal component analysis of the data on coexisting garnet and clinopyroxene could be used to classify rocks into their petrogenetic types.Distribution coefficient calculated by assuming ideal binary solution of Mg and Fe members in pyroxene and garnet is useful to indicate the P-T of the formation of the rocks.     

Garnet-cordierite-biotite equilibria in the Steinach aureole,Bavaria

Three garnet-biotite pairs and eleven garnet-cordierite-biotite triplets from the Steinach aureole (Oberpfalz, North-East Bavaria) were analyzed using an electron probe microanalyzer.The regional metamorphic muscovite-biotite schists contain garnets strongly zoned with Mn-Ca-rich centers and Fe-rich edges, the average composition being almandine 67 — spessartine 4 — pyrope 4 — grossular (+andradite) 25.The first contact garnet that is formed in mica schists of the outermost part of the aureole is small, virtually unzoned, and has an average composition of almandine 52 — spessartine 37 — pyrope 8 — grossular (+andradite) 3. With increasing metamorphic grade, there is a consistent trend to form garnets richer in Fe ending up with a composition almandine 84.5 — spessartine 5.5 — pyrope 7.5 — grossular (+andradite) 2.5. This trend is accompanied by a general increase in grain size and modal amount of garnet. Associated cordierites and biotites also become richer in Fe with increasing grade. While the garnets in the highest grade sillimanite hornfelses are poorly zoned, the transitional andalusite-sillimanite hornfelses contain garnets with distinct but variable zonation profiles.These facts can possibly be explained by the time-temperature relationships in the thermal aureole. In a phase diagram such as the Al-Fe-Mg-Mn tetrahedron, the limiting mineral compositions of a four-phase volume or a three-phase triangle are fixed by T and P (the latter remaining effectively constant within a thermal aureole). Thus, in garnet-cordierite-biotite assemblages, garnet zonation should be controlled by temperature variation rather than by a non-equilibrium depletion process. Taking into account the experimental data of Dahl (1968), a zoned garnet from a transitional andalusite-sillimanite hornfels would reflect a temperature increase of about 40° C during its growth. A hypothetical P-X diagram is proposed to show semi-quantitatively the compositional variation of garnet-cordierite pairs with varying pressures (T constant).     

Experimental determination of partition coefficients for rare earth and high-field-strength elements between clinopyroxene,garnet, and basaltic melt at high pressures

Clinopyroxene/melt and garnet/melt partition coefficients have been determined for Ti, Sr, Y, Zr, Nb, Hf, and rare earth elements from 19 doped experiments on 1921 Kilauea basalt. The experiments were carried out from 2.0 to 3.0 GPa and 1310° to 1470 °C. The purpose was to derive a set of partition coefficients for high-field-strength elements (HFSE) and rare earth elements (REE) in a systematic, linked set of experiments at P and T conditions relevant to basalt petrogenesis. These data are used in melting models to understand the development of negative HFSE anomalies observed in many abyssal peridotite clinopyroxenes. It is shown that melting can account for the observed trace element patterns in some residual peridotites, but that other processes may also be needed to account for most residual mantle compositions in mid-ocean ridge systems. It is also shown that REE are more strongly fractionated by garnet at these P-T conditions than previously thought. Received: 1 July 1997 / Accepted: 11 May 1998