Fe–Mg diffusion in olivine II: point defect chemistry,change of diffusion mechanisms and a model for calculation of diffusion coefficients in natural olivine

Analysis of existing data and models on point defects in pure (Fe,Mg)-olivine (Phys Chem Miner 10:27–37,1983; Phys Chem Miner 29:680–694, 2002) shows that it is necessary to consider thermodynamic non-ideality of mixing to adequately describe the concentration of point defects over the range of measurement. In spite of different sources of uncertainties, the concentrations of vacancies in octahedral sites in (Fe,Mg)-olivine are on the order of 10⁻⁴ per atomic formula unit at 1,000–1,200 °C according to both the studies. We provide the first explicit plots of vacancy concentrations in olivine as a function of temperature and oxygen fugacity according to the two models. It is found that in contrast to absolute concentrations at ∼1,100 °C and dependence on fO₂, there is considerable uncertainty in our knowledge of temperature dependence of vacancy concentrations. This needs to be considered in discussing the transport properties such as diffusion coefficients. Moreover, these defect models in pure (Fe,Mg)-olivine need to be extended by considering aliovalent impurities such as Al, Cr to describe the behavior of natural olivine. We have developed such a formulation, and used it to analyze the considerable database of diffusion coefficients in olivine from Dohmen et al. (Phys Chem Miner this volume, 2007) (Part - I) and older data in the literature. The analysis documents unequivocally for the first time a change of diffusion mechanism in a silicate mineral—from the transition metal extrinsic (TaMED) to the purely extrinsic (PED) domain, at fO₂ below 10⁻¹⁰  Pa, and consequently, temperatures below 900 °C. The change of diffusion mechanism manifests itself in a change in fO₂ dependence of diffusivity and a slight change in activation energy of diffusion—the activation energy increases at lower temperatures. These are consistent with the predictions of Chakraborty (J Geophys Res 102(B6):12317–12331, 1997). Defect formation enthalpies in the TaMED regime (distinct from intrinsic defect formation) lie between −66 and + 15 kJ/mol and migration energies of octahedral cations in olivine are most likely ∼ 260 kJ/mol, consistent with previous inferences (Phys Chem 207:147–162, 1998). Plots are shown for diffusion at various constant fO₂ as well as along fO₂ buffers, to highlight the difference in behavior between the two. Considering all the diffusion data and constraints from the point defect models, (Fe–Mg) diffusion in olivine along [001] is best described by the Master equations: (1) At oxygen fugacities greater than 10⁻¹⁰ Pa:

Multicomponent diffusion in garnets I: general theoretical considerations and experimental data for Fe–Mg systems

We have carried out a combined theoretical and experimental study of multicomponent diffusion in garnets to address some unresolved issues and to better constrain the diffusion behavior of Fe and Mg in almandine–pyrope-rich garnets. We have (1) improved the convolution correction of concentration profiles measured using electron microprobes, (2) studied the effect of thermodynamic non-ideality on diffusion and (3) explored the use of a mathematical error minimization routine (the Nelder-Mead downhill simplex method) compared to the visual fitting of concentration profiles used in earlier studies. We conclude that incorporation of thermodynamic non-ideality alters the shapes of calculated profiles, resulting in better fits to measured shapes, but retrieved diffusion coefficients do not differ from those retrieved using ideal models by more than a factor of 1.2 for most natural garnet compositions. Diffusion coefficients retrieved using the two kinds of models differ only significantly for some unusual Mg–Mn–Ca-rich garnets. We found that when one of the diffusion coefficients becomes much faster or slower than the rest, or when the diffusion couple has a composition that is dominated by one component (>75 %), then profile shapes become insensitive to one or more tracer diffusion coefficients. Visual fitting and numerical fitting using the Nelder-Mead algorithm give identical results for idealized profile shapes, but for data with strong analytical noise or asymmetric profile shapes, visual fitting returns values closer to the known inputs. Finally, we have carried out four additional diffusion couple experiments (25–35 kbar, 1,260–1,400 °C) in a piston-cylinder apparatus using natural pyrope- and almandine-rich garnets. We have combined our results with a reanalysis of the profiles from Ganguly et al. (1998) using the tools developed in this work to obtain the following Arrhenius parameters in D = D ₀ exp{–[Q _1bar + (P–1)ΔV ⁺]/RT} for D _Mg^* and D _Fe^*: Mg: Q _1bar = 228.3 ± 20.3 kJ/mol, D ₀ = 2.72 (±4.52) × 10⁻¹⁰ m²/s, Fe: Q _1bar = 226.9 ± 18.6 kJ/mol, D ₀ = 1.64 (±2.54) × 10⁻¹⁰ m²/s. ΔV ⁺ values were assumed to be the same as those obtained by Chakraborty and Ganguly (1992).     

Peierls–Nabarro modelling of dislocations in diopside

The core structures of dislocations in diopside have been calculated within the Peierls model, which assumes a planar core. 1/2<110> dislocations can dissociate into two collinear partial dislocations. We show that [001] glide is very difficult in (010) and that a non-collinear dissociation of [001](100) (modelled within a Peierls–Nabarro–Galerkin approach) makes glide equally easy in (100) and {110}. A widely spread core structure corresponding to a low lattice friction has been found for [100](010) and [010](100) dislocations which is not supported by mechanical data and, together with TEM observations, suggests that another, probably non-planar core structure is possible for these dislocations.     

Fe–Mg interdiffusion rates in clinopyroxene: experimental data and implications for Fe–Mg exchange geothermometers

Chemical interdiffusion of Fe–Mg along the c-axis [001] in natural diopside crystals (X _Di = 0.93) was experimentally studied at ambient pressure, at temperatures ranging from 800 to 1,200 °C and oxygen fugacities from 10^?11 to 10^?17 bar. Diffusion couples were prepared by ablating an olivine (X _Fo = 0.3) target to deposit a thin film (20–100 nm) onto a polished surface of a natural, oriented diopside crystal using the pulsed laser deposition technique. After diffusion anneals, compositional depth profiles at the near surface region (~400 nm) were measured using Rutherford backscattering spectroscopy. In the experimental temperature and compositional range, no strong dependence of D ^Fe–Mg on composition of clinopyroxene (Fe/Mg ratio between Di₉₃–Di₆₅) or oxygen fugacity could be detected within the resolution of the study. The lack of fO₂-dependence may be related to the relatively high Al content of the crystals used in this study. Diffusion coefficients, D ^Fe–Mg, can be described by a single Arrhenius relation with $$D^{{{\text{Fe}} - {\text{Mg}}}} = 2. 7 7\pm 4. 2 7\times 10^{ - 7} {\text{exp(}}-3 20. 7\pm 1 6.0{\text{ kJ}}/{\text{mol}}/{\text{RT)m}}^{ 2} /{\text{s}}.$$ D ^Fe–Mg in clinopyroxene appears to be faster than diffusion involving Ca-species (e.g., D ^Ca–Mg) while it is slower than D ^Fe–Mg in other common mafic minerals (spinel, olivine, garnet, and orthopyroxene). As a consequence, diffusion in clinopyroxene may be the rate-limiting process for the freezing of many geothermometers, and compositional zoning in clinopyroxene may preserve records of a higher (compared to that preserved in other coexisting mafic minerals) temperature segment of the thermal history of a rock. In the absence of pervasive recrystallization, clinopyroxene grains will retain compositions from peak temperatures at their cores in most geological and planetary settings where peak temperatures did not exceed ~1,100 °C (e.g., resetting may be expected in slowly cooled mantle rocks, many plutonic mafic rocks, or ultra-high temperature metamorphic rocks).     

Response of the coccolithophores Emiliania huxleyi and Coccolithus braarudii to changing seawater Mg and Ca concentrations: Mg/Ca, Sr/Ca ratios and δCa, δMg of coccolith calcite

Calcium and magnesium concentrations in seawater have varied over geological time scales. On short time scales, variations in the major ion composition of seawater influences coccolithophorid physiology and the chemistry of biogenically produced coccoliths. Validation of those changes via controlled laboratory experiments is a crucial step in applying coccolithophorid based paleoproxies for the reconstruction of past environmental conditions. Therefore, we examined the response of two species of coccolithophores, Emiliania huxleyi and Coccolithus braarudii, to changes in the seawater Mg/Ca ratio (≈0.5 to 10 mol/mol) by either manipulating the magnesium or calcium concentration under controlled laboratory conditions. Concurrently, seawater Sr/Ca ratios were also modified (≈2 to 40 mmol/mol), while keeping salinity constant at 35. The physiological response was monitored by measurements of the cell growth rate as well as the production rates of particulate inorganic and organic carbon, and chlorophyll a. Additionally, coccolithophorid calcite was analyzed for its elemental composition (Sr/Ca and Mg/Ca) as well as isotope fractionation of calcium and magnesium (Δ^44/40Ca and Δ^26/24Mg). Our results reveal that physiological rates were substantially influenced by changes in seawater calcium rather than magnesium concentration within the range estimated to have occurred over the past 250 million years when coccolithophores appear in the fossil record. All physiological rates of E. huxleyi decreased at a calcium concentration above 25 mmol L⁻¹, whereas C. braarudii displayed a higher tolerance to increased seawater calcium concentrations. Partition coefficient of Sr was calculated as 0.36 ± 0.04 (±2σ) independent of species. Partition coefficient of Mg²⁺ increased with increasing seawater Ca²⁺ concentrations in both coccolithophore species. Calcium isotope fractionation was constant at 1.1 ± 0.1‰ (±2σ) and not altered by changes in seawater Mg/Ca ratio. There is a well-defined inverse linear relationship between calcium isotope fractionation and partition coefficient of Sr²⁺ in all experiments, suggesting similar controls on both proxies in the investigated species. Magnesium isotope ratios were relatively stable for seawater Mg/Ca ratios ranging from 1 to 5, with a higher degree of fractionation in Emiliania huxleyi (by ≈0.2‰ in Δ^26/24Mg). Although Mg/Ca ratios in the calcite of coccolithophores and foraminifera are similar, the former have considerably higher Δ^26/24Mg (by >+3‰), presumably due to differences in calcification mechanisms between the two taxa. These observations suggest, a physiological control over magnesium elemental and isotopic fractionation during the process of calcification in coccolithophores.     

Groundwater mixing and mineralization processes in a mountain–oasis–desert basin,northwest China: hydrogeochemistry and environmental tracer indicators

Hydrogeochemistry and environmental tracers (²H, ¹⁸O, ⁸⁷Sr/⁸⁶Sr) in precipitation, river and reservoir water, and groundwater have been used to determine groundwater recharge sources, and to identify mixing characteristics and mineralization processes in the Manas River Basin (MRB), which is a typical mountain–oasis–desert ecosystem in arid northwest China. The oasis component is artificial (irrigation). Groundwater with enriched stable isotope content originates from local precipitation and surface-water leakage in the piedmont alluvial–oasis plain. Groundwater with more depleted isotopes in the north oasis plain and desert is recharged by lateral flow from the adjacent mountains, for which recharge is associated with high altitude and/or paleo-water infiltrating during a period of much colder climate. Little evaporation and isotope exchange between groundwater and rock and soil minerals occurred in the mountain, piedmont and oasis plain. Groundwater δ²H and δ¹⁸O values show more homogeneous values along the groundwater flow direction and with well depths, indicating inter-aquifer mixing processes. A regional contrast of groundwater allows the ⁸⁷Sr/⁸⁶Sr ratios and δ¹⁸O values to be useful in a combination with Cl, Na, Mg, Ca and Sr concentrations to distinguish the groundwater mixing characteristics. Two main processes are identified: groundwater lateral-flow mixing and river leakage in the piedmont alluvial–oasis plain, and vertical mixing in the north oasis plain and the desert. The ⁸⁷Sr/⁸⁶Sr ratios and selected ion ratios reveal that carbonate dissolution and mixing with silicate from the southern mountain area are primarily controlling the strontium isotope hydrogeochemistry.     

Volume diffusion of Ytterbium in YAG: thin-film experiments and combined TEM–RBS analysis

In this study, we address volume diffusion of ytterbium in yttrium aluminum garnet (YAG) using thin-film single crystal diffusion couples. We employ analytical transmission electron microscopy (ATEM) as a tool for combined microstructural and microchemical analysis and compare the results to Rutherford backscattering (RBS) analysis. Given the high spatial resolution of the method, we focus on microstructural changes of the thin-film diffusant source during the diffusion anneal. We evaluate the potential influence of the associated changes in its transport properties on the evolution of concentration profiles in the single crystal substrate. This approach allows us to test the reliability of determination of volume diffusion coefficients from thin-film diffusion experiments. We found that for the chosen experimental setting, the influence of thin-film re-crystallization is small when compared with the experimental uncertainty and good estimates for the volume diffusion coefficients of Yb in YAG can be obtained using standard assumptions. Both Yb-concentration profiles analyzed with ATEM and with RBS give similar results. At 1,450°C and 1 bar, we infer log D _Yb (m²/s) values of −19.37 ± 0.07 (TEM) and −19.84 ± 0.02 (RBS). Although the change in thin-film transport properties associated with successive crystallization during the diffusion anneal does not play a major role for our experimental setup, this effect cannot generally be ignored.     

Messinian Ca–Cl Brines from Mediterranean Basins: Tracing Diagenetic Effects by Ca/Mg Versus Ca/Sr Diagram

In natural resource exploration, Ca–Cl basinal brines are important for understanding the origin and spatial and temporal distribution of hydrocarbons and sedimentary ore deposits. Little attention has been paid to the possible connection between fossil basinal brines and paleo-seawaters and to the implications for reconstructing paleo-seawater compositions. Secular variations of Ca/Mg and Ca/Sr ratios in seawater have been documented mainly using fluid inclusions in halite, calcareous fossils and mineral analyses. However, brines and other sedimentary records connected to paleo-seawater or its evaporated residues may be chemically affected by burial diagenesis or the effects of continental waters of meteoric origin, thus complicating interpretations of the analytical results. To investigate these effects on fluids and minerals related to the Messinian salinity crisis of the Mediterranean basin, we re-evaluate published data from: (1) brackish-to-brine waters from onshore (Northern Apennine foredeep; Levantine basin) and offshore (porewaters from the Deep Sea Drilling Project); (2) Messinian parental seawater deduced from calcareous fossils, fluid inclusions and sulfate minerals; (3) meteoric waters dissolving evaporites. The compositional trends related to seawater evaporation, diagenesis and mixing that affect the Ca/Mg and Ca/Sr molar ratios of the basinal brines are effectively discriminated on a binary plot depicting the proper fields for seawater and meteoric-derived fluids. Brines showing stronger dolomitization start from Ca/Mg and Ca/Sr molar ratios of Messinian seawater deduced from the published analysis of fluid inclusions and open ocean fossils, that are therefore here validated ex post.     

Phase relations in the carbon-saturated C–Mg–Fe–Si–O system and C and Si solubility in liquid Fe at high pressure and temperature: implications for planetary interiors

The phase and melting relations of the C-saturated C–Mg–Fe–Si–O system were investigated at high pressure and temperature to understand the role of carbon in the structure of the Earth, terrestrial planets, and carbon-enriched extraterrestrial planets. The phase relations were studied using two types of experiments at 4 GPa: analyses of recovered samples and in situ X-ray diffractions. Our experiments revealed that the composition of metallic iron melts changes from a C-rich composition with up to about 5 wt.% C under oxidizing conditions (ΔIW = ?1.7 to ?1.2, where ΔIW is the deviation of the oxygen fugacity (fO₂) from an iron-wüstite (IW) buffer) to a C-depleted composition with 21 wt.% Si under reducing conditions (ΔIW < ?3.3) at 4 GPa and 1,873 K. SiC grains also coexisted with the Fe–Si melt under the most reducing conditions. The solubility of C in liquid Fe increased with increasing fO₂, whereas the solubility of Si decreased with increasing fO₂. The carbon-bearing phases were graphite, Fe₃C, SiC, and Fe alloy melt (Fe–C or Fe–Si–C melts) under the redox conditions applied at 4 GPa, but carbonate was not observed under our experimental conditions. The phase relations observed in this study can be applicable to the Earth and other planets. In hypothetical reducing carbon planets (ΔIW < ?6.2), graphite/diamond and/or SiC exist in the mantle, whereas the core would be an Fe–Si alloy containing very small amount of C even in the carbon-enriched planets. The mutually exclusive nature of C and Si may be important also for considering the light elements of the Earth’s core.     

Fe–Mg partitioning between post-perovskite and ferropericlase in the lowermost mantle

Fe–Mg partitioning between post-perovskite and ferropericlase has been studied using a laser-heated diamond anvil cell at pressures up to 154 GPa and 2,010 K which corresponds to the conditions in the lowermost mantle. The composition of the phases in the recovered samples was determined using analytical transmission electron microscopy. Our results reveal that the Fe–Mg partition coefficient between post-perovskite and ferropericlase (K _D^PPv/Fp) increases with decreasing bulk iron content. The compositional dependence of K _D^PPv/Fp on the bulk iron content explains the inconsistency in previous studies, and the effect of the bulk iron content is the most dominant factor compared to other factors, such as temperature and aluminum content. Iron prefers ferropericlase compared to post-perovskite over a wide compositional range, whereas the iron content of post-perovskite (X _Fe^PPv, the mole fraction) does not exceed a value of 0.10. The iron-rich ferropericlase phase may have significant influence on the physical properties, such as the seismic velocity and electrical conductivity at the core–mantle boundary region.     

Partitioning of Mg, Ca, and Na between carbonatite melt and hydrous fluid at 0.1–0.2 GPa

Experimental studies of the element distribution between carbonatite melts and hydrous fluids are hampered by the fact that neither the fluid nor the melt can be isochemically quenched in conventional high-pressure vessels. In order to overcome this problem, we used a double-capsule technique to separate immiscible fluid and melt phases during and after the runs. The inner platinum capsules were charged with carbonate mixtures (CaCO₃, MgCO₃ and Na₂CO₃) and placed inside the outer capsules charged with distilled water and diamond powder. The latter was used as an inert trap for solids precipitating from the fluid on quenching. Carbonate melt and hydrous fluid equilibrated through a small hole left in the upper end of the inner capsule. The runs were performed in rapid-quench cold-seal pressure vessels at 0.1–0.2 GPa and 700–900 °C in the two-phase (fluid + melt) stability region. Both quenched melt and quenched fluid were dissolved in dilute HCl and analysed by inductively coupled plasma atomic emission spectroscopy. The results show that under all conditions investigated, fluid/melt partition coefficients for Ca and Mg are similar and several times smaller than those for Na. At 0.1 GPa and a water/carbonatite ratio of 1 (by weight), the partition coefficients are D_Na = 0.35 ± 0.02, D_Ca = 0.09 ± 0.02, and D_Mg = 0.13 ± 0.01. Between 700 and 900 °C, the effect of temperature on partitioning is negligible. However, D_Na increases significantly with decreasing water/carbonatite ratio in the system. Our data show that the release of a hydrous fluid enriched in sodium and simultaneous crystallisation of calcite can transform an alkaline, vapour-saturated carbonatite melt into a body of pure calcite surrounded by zones of sodium metasomatism. Thus, it is quite possible that carbonate magmas with substantial amounts of alkalies were common parental liquids of plutonic carbonatites. Received: 6 May 1999 / Accepted: 31 August 1999     

Fe–Mg Interdiffusion in (Mg,Fe)O

We have performed a series of interdiffusion experiments on magnesiowüstite samples at room pressure, temperatures from 1,320° to 1,400°C, and oxygen fugacities from 10^?1.0 Pa to 10^?4.3 Pa, using mixed CO/CO₂ or H₂/CO₂ gases. The interdiffusion couples were composed of a single-crystal of MgO lightly pressed against a single-crystal of (Mg_1-x Fe _x )_1-δO with 0.07<x<0.27. The interdiffusion coefficient was calculated using the Boltzmann–Matano analysis as a function of iron content, oxygen fugacity, temperature, and water fugacity. For the entire range of conditions tested and for compositions with 0.01<x<0.27, the interdiffusion coefficient varies as $$\tilde D\, =\,2.9\times10^{ - 6}\,f_{{\text{O}}_2 }^{0.19}\,x^{0.73}\,{\text{e}}^{ - (209,000\, -\,96,000\,x)/RT}\,\,{\text{m}}^{\text{2}} {\text{s}}^{ -1} $$ These dependencies on oxygen fugacity and composition are reasonably consistent with interdiffusion mediated by unassociated cation vacancies. For the limited range of water activity that could be investigated using mixed gases at room pressure, no effect of water on interdiffusion could be observed. The dependence of the interdiffusion coefficient on iron content decreased with increasing iron concentration at constant oxygen fugacity and temperature. There is a close agreement between our activation energy for interdiffusion extrapolated to zero iron content (x=0) and that of previous researchers who used electrical conductivity experiments to determine vacancy diffusivities in lightly doped MgO.     

Role of diffusion–precipitation reactions in authigenic pyritization

Models proposed for authigenic pyritization in the literature provide good indicators of the effect of very high concentrations of available iron on decaying organisms; however, the impact of lower concentrations of iron on an actively decaying system is not so well characterised. Gel-stabilised systems are used to model the effect of extremes of iron concentration on the precipitation of pyrite and the process of organic matter preservation. The experiments show the effect of sulphate reduction decay in an environment where iron is limited or dispersed, and in iron-rich environment where diffusion is limited. The formation of discrete sulphide bands in experiments where iron is limited indicates that negative feedback, or Liesegang, reactions play a role in the development of gaps between sites of organic matter preservation and pyrite precipitation, providing a mechanism for the formation of pyrite halos, concretion rims and overgrowths. In iron-rich environments, pyrite formation is confined to the decaying organism, and the Liesegang effect is limited due to the restricted diffusion of dissolved sulphide.     

Ternary Ca–Fe–Mg carbonates: subsolidus phase relations at 3.5 GPa and a thermodynamic solid solution model including order/disorder

Subduction carries atmospheric and crustal carbon hosted in the altered oceanic crystalline basement and in pelagic sediments back into the mantle. Reactions involving complex carbonate solid solutions(s) lead to the transfer of carbon into the mantle, where it may be stored as graphite/diamond, in fluids or melts, or in carbonates. To constrain the thermodynamics and thus reactions of the ternary Ca–Mg–Fe carbonate solid solution, piston cylinder experiments have been performed in the system CaCO₃–MgCO₃–FeCO₃ at a pressure of 3.5 GPa and temperatures of 900–1,100°C. At 900°C, the system has two miscibility gaps: the solvus dolomite–calcite, which closes at X _MgCO3 ~0.7, and the solvus dolomite–magnesite, which ranges from the Mg to the Fe side of the ternary. With increasing temperature, the two miscibility gaps become narrower until complete solid solutions between CaCO₃–Ca_0.5Mg_0.5CO₃ is reached at 1,100°C and between CaCO₃–FeCO₃ at 1,000°C. The solvi are characterized by strong compositional asymmetry and by an order–disorder mechanism. To deal with these features, a solid solution model based on the van Laar macroscopic formalism has been calculated for ternary carbonates. This thermodynamic solid solution model is able to reproduce the experimentally constrained phase relations in the system CaCO₃–MgCO₃–FeCO₃ in a broad P–T range. To test our model, calculated phase equilibria were compared with experiments performed in carbonated mafic protolithes, demonstrating the reliability of our solid solution model at pressures up to 6 GPa in complex systems.     

Generation of hydrogen ions and hydrogen gas in quartz–water crushing experiments: an example of chemical processes in active faults

To understand the fundamental chemical processes of fluid–rock interaction during the pulverization of quartz grains in fault zones, quartz grains were crushed within pure water. The crushing experiments were performed batch style using a shaking apparatus. The crushing process induced a decrease in pH and an increase in hydrogen gas with increased shaking duration. The amount of hydrogen ions generated was five times larger than that of the hydrogen gas, which was consistent with the amount of Si radicals estimated from electron spin resonance measurements by Hochstrasser and Antonini (1972). This indicates that hydrogen gas was generated by consuming most of the Si radicals. The generation of hydrogen ions was most likely related to the presence of silanols on the newly formed mineral surface, implying a change of proton activities in the fluid after pulverization of quartz.     

Experimental peridotite–melt reaction at one atmosphere: a textural and chemical study

Sieve-textured clinopyroxene and spinel are common in mantle xenoliths and have been interpreted to be the result of partial melting, mantle metasomatism and host magma–xenolith reaction during transport. In this paper, we test the latter hypothesis with a series of reduced and oxidized experiments at 1,200 and 1,156°C at one atmosphere using a synthetic leucitite melt and discs of natural peridotite. Our results show that sieve texture development on clinopyroxene and spinel in mantle xenoliths is the result of a multistage reaction process. In the first step, orthopyroxene undergoes incongruent dissolution to produce a silica and alkali-rich melt together with olivine. As this melt migrates along grain boundaries it causes incongruent dissolution of clinopyroxene and spinel. The incongruent dissolution mechanism involves complete dissolution of the clinopyroxene or spinel followed by nucleation and growth of a secondary clinopyroxene or spinel once the reacting melt is saturated. The reaction of orthopyroxene, clinopyroxene and spinel with infiltrated host magma results in a range of melt compositions that are very similar to those interpreted to be due to very small degrees of partial melting. Electronic supplementary material The online version of this article (doi:) contains supplementary material, which is available to authorized users.