Major Element Partitioning Research Articles

Effects of fractional crystallization and cumulus processes on mineral composition trends of some lunar and terrestrial rock series

A plot of Mg of mafic minerals versus An of plagioclase in cumulate rocks from various lunar and terrestrial rock series shows each series to have a distinct curvilinear trend. The slopes of these trends vary from nearly vertical in the case of lunar anorthosites and Mg‐norites to nearly horizontal in the case of gabbros from the mid‐Atlantic ridge. Calculations based upon known major element partitioning between mafic minerals, plagioclase and subalkaline basaltic liquids indicate that fractional crystallization coupled with cotectic accumulation of mafic minerals and plagioclase will produce mineral composition trends on the Mg versus An diagram with slopes >1 for cases where An ≳ Mg. Furthermore, fractional crystallization of basaltic magmas with alkali concentrations approaching zero will produce near vertical Mg versus An trends. Therefore, the steep slopes of the lunar rock series are consistent with relatively simple fractionation processes. The relatively flat slope of mineral compositions from gabbros collected from the mid‐Atlantic ridge at 26°N is inconsistent with simple fractionation processes, and calculations show that periodic refilling of a fractionating magma chamber with picritic magma cannot simply explain this flat slope either.

Zonation des éléments en traces au cours de la croissance des cristaux dans les bains silicatés: l'exemple de Rb, Cs, Sr et Ba dans le système Qz-Ab-Or-H 2O

An indirect method was used to study Na, K, Rb, Cs, Sr and Ba partition coefficients between crystals and silicate melt. Equilibria between a hydrothermal solution and the melt at 800°C and 2 kb and between a hydrothermal solution and crystals at 750°C and 2 kb were separately achieved. For major element partitioning (Na and K), the results obtained here are in good agreement with those of Tuttle and Bowen (1958) which allow us to follow crystal evolution during a fractional crystallization process where the growth of zoned crystals takes place. For minor elements Rb, Cs, Sr, Ba, melt/aqueous solution partition coefficients depend on Na/K as well as the silica content of the melt. These effects are rather small for Rb and Cs, but are much more important for the alkaline earths. The feldspar/aqueous solution partition coefficients also depend on Na/K. The variations of the partition coefficients feldspar/melt are complex in the part of the Qz-Ab-Or diagram located below the cotectic line. During fractional crystallization following the Rayleigh law (assuming that there are no kinetic phenomena) Sr ( D > 10) is almost totally removed from the melt in the early stages whereas Cs ( D < 0.1) remains in the melt during the whole process. Rb and Ba have partition coefficients closer to unity. The variation of these coefficients, due to changes in bulk composition of liquid and crystals during fractional crystallization, can lead to complex zoning with possible concentration maxima at some stages. Similar phenomena can be expected in non-ideal natural solid solutions, even if no discontinuities can be detected in the physicochemical evolution of the parent magma.

First‐order major element variation in basalt glasses from the Mid‐Atlantic Ridge: 29°N to 73°N

Basaltic glasses from 29°N to 73°N on the Mid‐Atlantic Ridge define two geographic and compositional groups that are characterized by different major element concentrations, phase assemblages and partition coefficients. Group A glasses occur on the ridge segment north of Gibbs Fracture Zone, from 54°N, along the Reykjanes Ridge, the western volcanic zone of Iceland and Kolbeinsey Ridge, up to 70°N. Similar glasses occur also on the ridge from 29°N to 34°N. Group A is characterized by pronounced alumina depletion and iron‐enrichment trends, lower silica and alkalis and higher iron and high CaO/Al2O3. Clinopyroxene is only present in the more evolved group A glasses (Mg‐value &lt; 58). Olivine‐glass partition coefficients for Mg and Fe are systematically lower in group A than group B (Schilling and Sigurdsson, 1979), reflecting effects of melt composition on structure of the liquids. Group B glasses define the ridge segment north of 35°N over the Azores platform and up to Gibbs F. Z., at 53°N. They are also present on the Mohns Ridge north of the Jan Mayen F. Z. at 71°N and at least as far north as 73°. Group B glasses are characterized by higher silica and alkalis, low iron and lack of alumina depletion and iron‐enrichment trends. They contain calcic clinopyroxene (Cpx) throughout the compositional range, whereas the sub‐calcic augite is absent. The major‐element variation within each group can be modeled quantitatively by fractional crystallization involving the phenocryst phases. In group A, the model involves only olivine (Ol) and plagioclase (Pl) extraction in the early stages, but a total of 60% crystallization of the most primitive magma is required, yielding a final solid extract of Ol‐Pl‐Cpx in the proportions 1:3.3:1.5. Group B fractionation models require clinopyroxene extraction throughout the compositional range, and 45% crystallization of the primitive magma is required to account for the most evolved glasses giving a total solid extract of Ol‐Pl‐Cpx in proportions 1:5:3.9. The range of glass compositions in the two groups forms two distinct but parallel cotectic trends within the basalt tetrahedron, which are believed to define the quaternary univariant line Ol‐Pl‐Cpx‐Liq along which the magmas evolved during fractionation at low to intermediate pressure. It is proposed that the shift in phase boundaries and univariant lines reflects differences in melt and source compositions of the two groups. The distribution of the two groups is thus taken to indicate two major petrographic provinces of the Mid‐Atlantic Ridge.

Exsolved garnet-bearing pyroxene megacrysts from some South African kimberlites

Clinopyroxene and orthopyroxene megacrysts containing garnet lamellae up to 1.2 mm thick as an exsolved phase are found rarely in kimberlites from Frank Smith and Bellsbank. Chemically the clinopyroxenes are characteristically subcalcic, being within the range of 100 Ca/Ca + Mg + Fe = 27 to 36, and the orthopyroxenes are characterized by high Al 2O 3 and Cr 2O 3. Immediately after crystallization during very slow cooling, clinopyroxene and orthopyroxene exsolve wide-spaced orthopyroxene and clinopyroxene phases parallel to (100) of the host phases, respectively, then both host and exsolved phases exsolve garnet lamellae. Topotactic relations between pyroxenes and garnet are determined by X-ray for the first time. Partitioning of major and minor elements among the coexisting clinopyroxene, orthopyroxene and garnet in pyroxene megacrysts is the same as that of the granular-type garnet peridotite xenoliths in Lesotho and South African kimberlies. Mineralogy and chemistry indicate that subcalcic clinopyroxene and orthopyroxene megacrysts contain respectively about 10 and 3 mole % of the garnet molecule in solid solution.

Crystal/liquid partitioning in augite: effects of cooling rate

The partitioning of major and minor elements between augite and melt was determined as a function of cooling rate for two high-titanium basalt compositions. The results of this study of lunar rock systems 10017 and 75055 were compared with the results of other kinetic studies of augite-liquid partitioning in other rock systems. It was found that the partitioning of major elements (i.e., Ca, Fe, Mg) is essentially rate independent and is insensitive to bulk rock composition and to the nature and order of appearance of coexisting phases for cooling rates of less than 100°C/hr. The partitioning behavior of minor elements (i.e., Al, Cr, Ti) for the same range of cooling rates is complex, being dependent on cooling rate and bulk rock composition. Consideration of these factors is important when augite chemistry and/or partitioning behavior are used in modeling certain magmatic processes or in estimating the thermal history of basaltic rocks.

Petrogenesis of some Ligurian peridotites—I. Mineral and bulk-rock chemistry

Bulk-rock and mineral compositions of eleven lherzolitic specimens have been determined. Eight samples are from the Erro-Tobbio thrust sheet of the Gruppo di Voltri, western Liguria; three samples are from allochthonous eastern Ligurian masses within the External Ligurides. All specimens consist of spinel (±plagioclase) in addition to ol + opx + cpx, and have been derived from mantle portions of the South Alpine-Apennine lithospheric plate. Electron microprobe analyses of 11 olivines, 17 orthopyroxenes, 13 clinopyroxenes, eight spinels and one plagioclase have been performed. Except for Ca-rich and Ca-poor lamellae in some pyroxenes, and iron-enriched rims in a few spinels, the analyzed phases are remarkably homogeneous. Mineral compositions vary insignificantly from rock to rock. Consideration of major element partitioning in coexisting opx + cpx pairs, combined with the petrogenetic grid for aluminous lherzolite bulk compositions, yield the following provisional P- Tequilibration conditions: Erro-Tobbio complex, 1150 ± 50° C, 16 ± 6 kbar; External Ligurides, 1000 ± 50° C, 14 ± 6 kbar. Evidently the western Ligurian peridotites equilibrated at higher temperatures and possibly at slightly greater depths with the upper mantle than those from eastern Liguria. The lherzolite samples have nearly identical bulk-rock major element compositions. Comparisons with estimated compositions for the upper mantle, and for model pyrolite indicate that these Ligurian peridotites are primitive or only very slightly depleted in basaltic constituents. Thus, based on major element chemistry, the Ligurian peridotites cannot be regarded as residua left after extensive partial fusion, hence evidently cannot have produced large proportions of oceanic tholeiite liquid; on the other hand, they seem to represent a fertile mantle source for such melts. Relics of garnet have not been found, and apparent temperatures for both Erro-Tobbio and eastern Ligurian ultramafic suites exceed the classical intraplate oceanic geotherm; accordingly, these bodies may represent rising mantle material that recrystallized during ascent in the general vicinity of a spreading center, without having undergone extensive partial melting.

Clinopyroxene-orthopyroxene major and rare earth elements partitioning in spinel peridotite xenoliths from Assab (Ethiopia)

Major element and rare earth element (REE) partitioning among coexisting clinopyroxene-orthopyroxene pairs from mantle xenoliths of the Assab Range (Ethiopia) are discussed in terms of crystal-chemistry. Major element partitioning indicates relatively uniform conditions of subsolidus equilibration over a narrow range of temperatures (mean value about 1100 C) in the spinel peridotite stability field. Major element distributions and correlations, moreover, seem to indicate that the mantle material studied underwent slightly different depletions prior to the metamorphic equilibration. In spite of the rather homogeneous major element compositions for both cpx and opx, clinopyroxenes show chondrite-normalized REE patterns which are widely variable both in shape and absolute values, whereas orthopyroxenes exhibit more restricted ranges and concordant profiles. REE activity ratios have been investigated by applying Iiyama's ( Bull. Soc. fr. Minéral. Christallogr. 97, 143–151) thermodynamic model: the estimated activity patterns exhibit a good coherence for the different pyroxene pairs, in spite of the contrasting features of their REE concentration ratios. The wide ranges in the measured partition values for the same rare earth element in different pyroxene pairs have been related to coupled substitutions involving A1 in the Z site and REE in the M 2 site of clinopyroxene.

Irreversible Reactions in High-Grade Metapelitic Rocks

At least two discontinuous reactions and additional divariant (continuous) compositional shifts were overstepped in pelitic rocks of the inner aureole around the Ronda peridotite massif, southern Spain, probably as a result of upward drag and rapid decrease of Pt during high-grade metamorphism. The resulting reactions were garnet or biotite+A1-silicate↑cor-dierite+hercynite±excess component phases present throughout reaction (quartz, K-feld-spar, ilmenite, plagioclase). Both reactant and product phases are preserved. The extent of reaction, estimated from the amount of cordierite and hercynite produced by reaction, can vary markedly within a single thin section. Coexisting phases were analysed in twelve domains representing varying degrees of reaction in one thin section. A domain, the unit of comparison, was taken as the smallest volume that contained the major reacting phases. Compositions of products in a thin section vary little from domain to domain and probably can be considered a function only of the intensive variables (P, T, μH2o); product phases appear to approach equilibrium as determined by partitioning of major elements. However, reactant compositions are a function of extent of reaction and are not generally in equilibrium with the products. Thus, the reactions are irreversible. The original assemblage before reaction (quartz-K-feldspar-sillimanite and kyanite-garnet-biotite) is preserved within unzoned cores of garnet grains. As the extent of reaction increases, Mg/Fe of biotite in the matrix migrates from the initial value toward a more Fe-rich composition. The analytical data are used to reject the local equilibrium or domain equilibrium model of reaction on the scale of a domain. Product compositions do not vary as expected in this model as a function of extent of reaction, variable reactant composition, or distance from non-equilibrium reactants. The reactant grains must be chemically isolated from the equilibrium matrix products and fluid in partial equilibrium; this model is distinguished as the 'dissolution model'. Diffusion gradients still can be present in this model and are responsible for the distribution of product phases; however, the reaction rate is not diffusion-controlled. Reaction rate can be limited by interface reactions or by diffusion within crystals. Two models of reaction are examined assuming that reaction rate is controlled by reactant dissolution. The first model of irreversible reaction examined, assuming the reacting composition to be that for which the decrease of free energy of reaction is greatest, does not satisfactorily explain the data. A second, descriptive model is proposed in which the overall irreversible reaction is divided into component reactions and the rate of these reactions is assumed to be a monotonic function of free energy change or growth rate of product phases. The model can be shown qualitatively to explain observed garnet zoning and changing biotite composition during reaction. Because reactant phases and compositions of the primary assemblage are preserved along with the equilibrium product phases in these samples, more information is available from these ‘disequilibrium’ rocks than from samples that have completely equilibrated. The variability of extent of reaction in these samples emphasizes that contact of phases in a sample is not evidence of mutual compatibility.