Ferric iron partitioning between plagioclase and silicate liquid: thermodynamics and petrological applications

A series of Fe and Mg partition experiments between plagioclase and silicate liquid were performed in the system SiO₂-Al₂O₃-Fe₂O₃-FeO-MgO-CaO-Na₂O under oxygen fugacities from below the IW buffer up to that of air. A thermodynamic model of plagioclase solid solution for the (CaAl,NaSi,KSi)(Fe³⁺,Al³⁺)Si₂O₈-Ca(Fe²⁺,Mg)Si₃O₈ system is proposed and is calibrated by regression analysis based on new and previously reported experimental data of Fe and Mg partitioning between plagioclase and silicate liquid, and reported thermodynamic properties of end members, ternary feldspar and silicate liquid. Using the derived thermodynamic model, FeO^t, MgO content and Mg/(Fe^t+Mg) in plagioclase can be predicted from liquid composition with standard deviations of ǂ.34 wt% (relative error =9%) and ǂ.08 wt% (14%) and ǂ.7 (8%) respectively. Calculated Fe³⁺-Al exchange chemical potentials of plagioclase, m_{Fe^{3 +} ( Al )- 1} ^Pl{\rm \mu }_{{\rm Fe}^{{\rm 3 + }} \left( {{\rm Al}} \right)_{{\rm - 1}} }^{{\rm Pl}} agree with those calculated using reported thermodynamic models for multicomponent spinel, m_{Fe^{3 +} ( Al )- 1} ^Sp{\rm \mu }_{{\rm Fe}^{{\rm 3 + }} \left( {{\rm Al}} \right)_{{\rm - 1}} }^{{\rm Sp}} and clinopyroxene, m_{Fe^{3 +} ( Al )- 1} ^Cpx{\rm \mu }_{{\rm Fe}^{{\rm 3 + }} \left( {{\rm Al}} \right)_{{\rm - 1}} }^{{\rm Cpx}} . The FeO^t content of plagioclase coexisting with spinel or clinopyroxene is affected by Fe³⁺/(Fe³⁺+Al) and Mg/(Fe+Mg) of spinel or clinopyroxene and temperature, while it is independent of the anorthite content of plagioclase. Three oxygen barometers based on the proposed model are investigated. Although the oxygen fugacities predicted by the plagioclase-liquid oxygen barometer are scattered, this study found that plagioclase-spinel-clinopyroxene-oxygen and plagioclase-olivine-oxygen equilibria can be used as practical oxygen barometers. As a petrological application, prediction of plagioclase composition and f_O2 are carried out for the Upper Zone of the Skaergaard intrusion. The estimated oxygen fugacities are well below QFM buffer and consistent with the estimation of oxidization states in previous studies.     

The effect of liquid composition on the partitioning of Ni between olivine and silicate melt

We report the results of experiments designed to separate the effects of temperature and pressure from liquid composition on the partitioning of Ni between olivine and liquid, \(D_{\text{Ni}}^{\text{ol/liq}}\). Experiments were performed from 1300 to 1600 °C and 1 atm to 3.0 GPa, using mid-ocean ridge basalt (MORB) glass surrounded by powdered olivine in graphite–Pt double capsules at high pressure and powdered MORB in crucibles fabricated from single crystals of San Carlos olivine at one atmosphere. In these experiments, pressure and temperature were varied in such a way that we produced a series of liquids, each with an approximately constant composition (~12, ~15, and ~21 wt% MgO). Previously, we used a similar approach to show that \(D_{\text{Ni}}^{\text{ol/liq}}\) for a liquid with ~18 wt% MgO is a strong function of temperature. Combining the new data presented here with our previous results allows us to separate the effects of temperature from composition. We fit our data based on a Ni–Mg exchange reaction, which yields \(\ln \left( {D_{\text{Ni}}^{\text{molar}} } \right) = \frac{{ -\Delta _{r(1)} H_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } }}{RT} + \frac{{\Delta _{r(1)} S_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } }}{R} - \ln \left( {\frac{{X_{\text{MgO}}^{\text{liq}} }}{{X_{{{\text{MgSi}}_{ 0. 5} {\text{O}}_{ 2} }}^{\text{ol}} }}} \right).\) Each subset of constant composition experiments displays roughly the same temperature dependence of \(D_{\text{Ni}}^{\text{ol/liq}}\) (i.e.,\(-\Delta _{r(1)} H_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } /R\)) as previously reported for liquids with ~18 wt% MgO. Fitting new data presented here (15 experiments) in conjunction with our 13 previously published experiments (those with ~18 wt% MgO in the silicate liquid) to the above expression gives \(-\Delta _{r(1)} H_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } /R\) = 3641 ± 396 (K) and \(\Delta _{r(1)} S_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } /R\) = ? 1.597 ± 0.229. Adding data from the literature yields \(-\Delta _{r(1)} H_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } /R\) = 4505 ± 196 (K) and \(\Delta _{r(1)} S_{{T_{\text{ref}} ,P_{\text{ref}} }}^{ \circ } /R\) = ? 2.075 ± 0.120, a set of coefficients that leads to a predictive equation for \(D_{\text{Ni}}^{\text{ol/liq}}\) applicable to a wide range of melt compositions. We use the results of our work to model the melting of peridotite beneath lithosphere of varying thickness and show that: (1) a positive correlation between NiO in magnesian olivine phenocrysts and lithospheric thickness is expected given a temperature-dependent \(D_{\text{Ni}}^{\text{ol/liq}} ,\) and (2) the magnitude of the slope for natural samples is consistent with our experimentally determined temperature dependence. Alternative processes to generate the positive correlation between NiO in magnesian olivines and lithospheric thickness, such as the melting of olivine-free pyroxenite, are possible, but they are not required to explain the observed correlation of NiO concentration in initially crystallizing olivine with lithospheric thickness.     

Redox equilibria of iron and silicate melt structure: Implications for olivine/melt element partitioning

Olivine/melt partitioning of ΣFe, Fe²⁺, Mg²⁺, Ca²⁺, Mn²⁺, Co²⁺, and Ni²⁺ has been determined in the systems CaO-MgO-FeO-Fe₂O₃-SiO₂ (FD) and CaO-MgO-FeO-Fe₂O₃-Al₂O₃-SiO₂ (FDA3) as a function of oxygen fugacity (f_O2) at 0.1 MPa pressure. Total iron oxide content of the starting materials was ∼20 wt%. The f_O2 was to used to control the Fe³⁺/ΣFe (ΣFe: total iron) of the melts. The Fe³⁺/ΣFe and structural roles of Fe²⁺ and Fe³⁺ were determined with ⁵⁷Fe resonant absorption Mössbauer spectroscopy. Changes in melt polymerization, NBO/T, as a function of f_O2 was estimated from the Mössbauer data and existing melt structure information. It varies by ∼100% in melts coexisting with olivine in the FDA3 system and by about 300% in the FD system in the Fe³⁺/ΣFe range of the experiments (0.805-0.092). The partition coefficients ( in olivine/wt% in melt) are systematic functions of f_O2 and, therefore, NBO/T of the melt. There is a -minimum in the FDA3 system at NBO/T-values corresponding to intermediate Fe³⁺/ΣFe (0.34-0.44). In the Al-free system, FD, where the NBO/T values of melts range between ∼1 and ∼2.9, the partition coefficients are positively correlated with NBO/T (decreasing Fe³⁺/ΣFe). These relationships are explained by consideration of solution behavior in the melts governed by Qⁿ-unit distribution and structural changes of the divalent cations in the melts (coordination number, complexing with Fe³⁺, and distortion of the polyhedra).     

An experimental study of Fe-Mg partitioning between garnet and olivine and its calibration as a geothermometer

The partitioning of Fe and Mg between coexisting garnet and olivine has been studied at 30 kb pressure and temperatures of 900 ° to 1,400 °C. The results of both synthesis and reversal experiments demonstrate that K _D (= (Fe/Mg)_gt/(Fe/Mg)_OI) is strongly dependent on Fe/Mg ratio and on the calcium content of the garnet. For example, at 1,000 °C/30 kb, K _D varies from about 1.2 in very iron-rich compositions to 1.9 at the magnesium end of the series. Increasing the mole fraction of calcium in the garnet from 0 to 0.3 at 1,000 ° C increases K _D in magnesian compositions from 1.9 to about 2.5.The observed temperature and composition dependence of K _D has been formulated into an equation suitable for geothermometry by considering the solid solution properties of the olivine and garnet phases. It was found that, within experimental error, the simplest kind of nonideal solution model (Regular Solution) fits the experimental data adequately. The use of more complex models did not markedly improve the fit to the data, so the model with the least number of variables was adopted.Multiple linear regression of the experimental data (72 points) yielded, for the exchange reaction: 3Fe₂SiO₄+2Mg₃Al₂Si₃O₁₂ olivine garnet 2Fe₂Al₂Si₃O₁₂+3Mg₂SiO₄ garnet olivine H ° (30kb) of –10,750 cal and S ° of –4.26 cal deg^–1 mol^–1. Absolute magnitudes of interaction parameters (W _ij ) derived from the regression are subject to considerable uncertainty. The partition coefficient is, however, strongly dependent on the following differences between solution parameters and these differences are fairly well constrained: W _FeMg ^ol -W _FeMg ^gt 800 cal W _CaMg ^gt -W _CaFe ^gt 2,670 cal.The geothermometer is most sensitive in the temperature and composition regions where K _D is substantially greater than 1. Thus, for example, peridotitic compositions at temperatures less than about 1,300 ° C should yield calculated temperatures within 60 °C of the true value. Iron rich compositions (at any temperature) and magnesian compositions at temperatures well above 1,300 °C could not be expected to yield accurate calculated temperatures.For a fixed K _D the influence of pressure is to raise the calculated temperature by between 3 and 6 °C per kbar.     

Experimental partitioning of uranium between liquid iron sulfide and liquid silicate: Implications for radioactivity in the Earth’s core

Measurable uranium (U) is found in metal sulfide liquids in equilibrium with molten silicate at conditions appropriate for a planetary magma ocean: 1-10 GPa, 1750-2100 °C, 0-28 wt% S, and fO₂ 2 log units below IW. However, the transfer of U from metal sulfide to silicate under our experimental conditions is so complete that insufficient U would remain so as to be of any importance to the core’s heat budget. U content in the sulfide phase increases strongly with S content but shows no significant variability with either pressure or temperature. Maximum is 0.001 while most values are considerably lower.     

An experimental study of Fe-Mg partitioning between olivine and orthopyroxene at 1173, 1273 and 1423 K and 1.6 GPa

The partitioning of Mg and Fe²⁺ between coexisting olivines and orthopyroxenes in the system MgO-FeO-SiO₂ has been investigated experimentally at 1173, 1273, 1423 K and 1.6 GPa over the whole range of Mg/Fe ratios. The use of barium borosilicate as a flux to promote grain growth, and the identification by back-scattered electron imaging of resulting growth rims suitable for analysis by electron microprobe, results in coexisting olivine and orthopyroxenene compositions determined to a precision of±0.003 to 0.004 in molar Fe/(Mg+Fe). Quasi-reversal experiments were performed starting with Mg-rich olivine and Fe-rich orthopyroxene (low K_D) and vice versa (high K_D), which produced indistinguishable results. The distribution coefficient, K_D, depends on composition and on temperature, but near Fe/(Mg+Fe)=0.1 (i.e. mantle compositions) these effects cancel out, and K_D is insensitive to temperature. The results agree well with previous experimental investigations, and constrain the thermodynamic mixing properties of Mg-Fe olivine solid solutions to show small near-symmetric deviations from ideality, with between 2000 and 8000 J/mol. Multiple non-linear least squares regression of all data gave a best fit with (implying 5450 J/mol at 1 bar) and , but the two W _G parameters are so highly correlated with each other that our data are almost equally well fit with , as obtained by Wiser and Wood. This value implies , apparently independent of temperature. Our experimental results are not compatible with the assessment of olivine-orthopyroxene equilibria of Sack and Ghiorso.     

The partitioning of Fe and Mg between olivine and carbonate and the stability of carbonate under mantle conditions

We have investigated the effect of Fe on the stabilities of carbonate (carb) in lherzolite assemblages by determining the partitioning of Fe and Mg between silicate (olivine; ol) and carbonates (magnesite, dolomite, magnesian calcite) at high pressures and temperatures. Fe enters olivine preferentially relative to magnesite and ordered dolomite, but Fe and Mg partition almost equally between disordered calcic carbonate and olivine. Measurement of K _d (X _Fe ^carb X _Mg ^ol /X _Fe ^ol X _Mg ^carb ) as a function of Fe/ Mg ratio indicates that Fe–Mg carbonates deviate only slightly from ideality. Using the regular solution parameter for olivine W _FeMg ^ol of 3.7±0.8 kJ/mol (Wiser and Wood 1991) we obtain for (FeMg)CO₃ a W _FeMg ^carb of 3.05±1.50 kJ/mol. The effect of Ca–Mg–Fe disordering is to raise K _d substantially enabling us to calculate W _CaMg ^carb -W _CaFe ^carb of 5.3±2.2 kJ/mol. The activity-composition relationships and partitioning data have been used to calculate the effect of Fe/Mg ratio on mantle decarbonation and exchange reactions. We find that carbonate (dolomite and magnesian calcite) is stable to slightly lower pressures (by 1 kbar) in mantle lherzolitic assemblages than in the CaO–MgO–SiO₂(CMS)–CO₂ system. The high pressure breakdown of dolomite + orthopyroxene to magnesite + clinopyroxene is displaced to higher pressures (by 2 kbar) in natural compositions relative to CMS. CO₂. We also find a stability field of magnesian calcite in lherzolite at 15–25 kbar and 750–1000°C.     

Systematics of calcium partitioning between olivine and silicate melt: implications for melt structure and calcium content of magmatic olivines

A systematic characterization of the chemical factors that control calcium partitioning between olivine and melt in a magmatic environment was undertaken using experiments performed on compositionally simple systems (CaO-MgO-SiO₂, CaO-MgO-Al₂O₃-SiO₂, CaO-MgO-Al₂O₃-SiO₂-Cr₂O₃, CaO-MgO-Al₂O₃-SiO₂-TiO₂, CaO-MgO-Al₂O₃-SiO₂-Na₂O, CaO-MgO-Al₂O₃-SiO₂-FeO, CaO-MgO-Al₂O₃-SiO₂-FeO-Na₂O) over a wide range of temperature (1050–1530 °C) at one bar pressure. The calcium concentration of olivines is shown to be dependent not only on the forsterite content of the olivine but to a large extent on melt composition. For a fixed CaO content of the melt, these results show that the CaO concentration of olivine is strongly sensitive to the amount of alumina, alkali and ferrous iron present in the coexisting melt. Oxygen fugacity and temperature are not found directly to affect Ca partitioning. It is thus proposed that the systematic variations of the calcium content of olivine may be used as an “in-situ chemical potentiometer” of the lime activity of the melt. Based upon these data in synthetic systems, an empirical model describing Ca partitioning between olivine and melt is developed. When applied to natural olivines this model reproduces their Ca content, where melt composition is known, to within ±10% relative. The model may therefore be used to predict changes in melt composition during olivine crystallization and/or to assess whether an olivine is in equilibrium with its host magma. Finally, the wide range of Ca partitioning observed at fixed crystal composition confirms that minor element partitioning between crystal and melt cannot be predicted from the physical characteristics of the crystal alone, and that the non-ideality of the melt has to be taken into account. Received: 12 June 1998 / Accepted: 1 February 1999     

Copper partitioning between olivine and melt inclusions and its content in primitive island-arc magmas of Kamchatka

Melt inclusions and hosting them highly magnesian olivine from rocks of Kamchatka and the Western Aleutian island arc were analyzed for copper content by LA-ICP-MS to determine the copper partition coefficient in primitive island-arc magmas. Based on measurements of 45 olivine–melt pairs, this coefficient was determined to be 0.028 ± 0.009 (2σ), which is the lowest value among previously published data. Mass-balance calculations of copper in a typical mantle peridotite using obtained partition coefficient indicate that its content in peridotite and primary mantle magmas is mainly determined by mantle sulfide. The Cu partition coefficient was also used to calculate the copper content in parental magmas of volcanoes of the Central Kamchatka Depression. Estimates obtained using copper content in phenocrysts of primitive olivine (Fo > 88 mol %) from these rocks are, on average, 139 ± 58 ppm (2σ), which exceed copper contents in primitive basalts (MgO > 8.5 wt %) of mid-ocean ridges (MORB 93 ± 31 ppm). This suggests the primary enrichment of Central Kamchatka magmas in copper and correlates with their more oxidizing conditions of formation as compared to MORB.     

The effect of sulfur on the partitioning of Ni and other first-row transition elements between olivine and silicate melt

The effect of sulfur dissolved as sulfide (S²⁻) in silicate melts on the activity coefficients of NiO and some other oxides of divalent cations (Ca, Cr, Mn, Fe and Co) has been determined from olivine/melt partitioning experiments at 1400 °C in six melt compositions in the system CaO-MgO-Al₂O₃-SiO₂ (CMAS), and in derivatives of these compositions at 1370 °C, obtained from the six CMAS compositions by substituting Fe for Mg (FeCMAS). Amounts of S²⁻ were varied from zero to sulfide saturation, reaching 4100 μg g⁻¹ S in the most sulfur-rich silicate melt. The sulfide solubilities compare reasonably well with those predicted from the parameterization of the sulfide capacity of silicate melts at 1400 °C of O’Neill and Mavrogenes (2002), although in detail systematic deviations indicate that a more sophisticated model may improve the prediction of sulfide capacities.The results show a barely discernible effect of S²⁻ in the silicate melt on Fe, Co and Ni partition coefficients, and also surprisingly, a tiny but resolvable effect on Ca partitioning, but no detectable effect on Cr, Mn or some other lithophile incompatible elements (Sc, Ti, V, Y, Zr and Hf). Decreasing Mg# of olivine (reflecting increasing FeO in the system) has a significant influence on the partitioning of several of the divalent cations, particularly Ca and Ni. We find a remarkably systematic correlation between and the ionic radius of M²⁺, where M = Ca, Cr, Mn, Fe, Co or Ni, which is attributable to a simple relationship between size mismatch and excess free energies of mixing in Mg-rich olivine solid solutions.Neither the effect of S²⁻ nor of Mg#^ol is large enough by an order of magnitude to account for the reported variations of obtained from electron microprobe analyses of olivine/glass pairs from mid-ocean ridge basalts (MORBs). Comparing these MORB glass analyses with the Ni-MgO systematics of MORB from other studies in the literature, which were obtained using a variety of analytical techniques, shows that these electron microprobe analyses are anomalous. We suggest that the reported variation of with S content in MORB is an analytical artifact.Mass balance of melt and olivine compositions with the starting compositions shows that dissolved S²⁻ depresses the olivine liquidus of haplobasaltic silicate melts by 5.8 × 10⁻³ (±1.3 × 10⁻³) K per μg g⁻¹ of S²⁻, which is negligible in most contexts. We also present data for the partitioning of some incompatible trace elements (Sc, Ti, Y, Zr and Hf) between olivine and melt. The data for Sc and Y confirm previous results showing that and decrease with increasing SiO₂ content of the melt. Values of average 0.01 with most falling in the range 0.005-0.015. Zr and Hf are considerably more incompatible than Ti in olivine, with and about 10⁻³. The ratio / is well constrained at 0.611 ± 0.016.     

Extension of lattice strain theory to mineral/mineral rare-earth element partitioning: An approach for assessing disequilibrium and developing internally consistent partition coefficients between olivine, orthopyroxene, clinopyroxene and basaltic melt

Olivine/melt and orthopyroxene/melt rare-earth element (REE) partition coefficients consistent with clinopyroxene/melt partition coefficients were determined indirectly from subsolidus partitioning between olivine, orthopyroxene, and clinopyroxene after suitable correction for temperature. Heavy- and middle-REE ratios for olivine/clinopyroxene and orthopyroxene/clinopyroxene pairs correlate negatively with effective cationic radius, whereas those for the light REEs correlate positively with cationic radius, generating a U-shaped pattern in apparent mineral/clinopyroxene partition coefficients versus cationic radius. Lattice strain models of partitioning modified for subsolidus conditions yield negative correlations of olivine/clinopyroxene and orthopyroxene/clinopyroxene with respect to cationic radii, predicting well the measured partitioning behaviors of the heavy and middle REEs but not that of the light REEs. The light-REE systematics cannot be explained with lattice strain theory and, instead, can be explained by disequilibrium enrichment of the light REEs in melt inclusions or on the rims of olivine and orthopyroxene. Realistic light-REE partition coefficients were thus extrapolated from the measured heavy- and middle-REE partition coefficients using the lattice strain model. Light REE olivine/melt and orthopyroxene/melt partition coefficients calculated in this manner are lower than most published values, but agree reasonably well with partitioning experiments using the most recent in situ analytical techniques (secondary-ionization mass spectrometry and laser ablation inductively coupled plasma mass spectrometry). These new olivine/melt and orthopyroxene/melt partition coefficients are useful for accurate modeling of the REE contents of clinopyroxene-poor to -free lithologies, such as harzburgitic residues of melting. Finally, the application of the lattice strain theory to subsolidus conditions represents a framework for assessing the degree of REE disequilibrium in a rock.     

Ferric-ferrous ratio in liquid iron oxides: Analysis and applications to natural basaltic melts

Experimental data on the proportions of ferrous and ferric iron in pure liquid oxides (Darken and Gurry, 1946) were used to test different redox models. The obtained inferences were used to evaluate possible problems in describing the dependence of Fe³⁺/Fe²⁺ on oxygen fugacity in natural basaltic melts.     

The partitioning of Pt-Re-Os between solid and liquid metal in the Fe-Ni-S system at high pressure: Implications for inner core fractionation

Coupled ¹⁸⁶Os/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os enrichments of plume-derived lavas have been suggested to reflect contributions of materials from the outer core (Brandon et al., 1998). This hypothesis is based on the assumption that the Earth’s liquid outer core has high Pt/Os and slightly high Re/Os ratios as a result of the crystallization of the solid inner core, and shows coupled enrichments in the ¹⁸⁶Os/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios, reflecting the decay of ¹⁹⁰Pt and ¹⁸⁷Re to ¹⁸⁶Os and ¹⁸⁷Os, respectively. Partitioning experiments of Pt-Re-Os between solid and liquid metal were performed at 5-20 GPa and 1250-1400 °C, to examine the effects of pressure in the Fe-Ni-S system. The ratios (D_Os/D_Pt, D_Os/D_Re) of measured partition coefficients of Pt, Re and Os are almost constant with increasing pressure. D_Os/D_Pt increases significantly, whereas D_Os/D_Re decreases, with increasing sulphur content in the liquid metal. On the basis of the present experimental results, it is unlikely that the required Pt-Re-Os fractionation is generated during inner core crystallization, assuming that the light element in the Earth’s core is sulphur.     

An experimental investigation of the partitioning of REE between garnet and liquid with reference to the role of defect equilibria

The partitioning of samarium and thulium between garnets and melts in the systems Mg₃Al₂-Si₃O₁₂-H₂O and Ca₃Al₂Si₃O₁₂-H₂O has been studied as a function of REE concentration in the garnets at 30 kbar pressure. Synthesis experiments of variable time under constant P, T conditions indicate that garnet initially crystallizes rapidly to produce apparent values of D _Sm (D _Sm=concentration of Sm in garnet/concentration of Sm in liquid) which are too large in the case of pyrope and too small in the case of grossular. As the experiment proceeds, Sm diffuses out of or into the garnet and the equilibrium value of D _Sm is approached. Approximate values of diffusion coefficients for Sm in pyrope garnet obtained by this method are 6 × 10^–13 cm² s^–1 at 1,300 ° C and 2 × 10^–12 cm² s^–1 at 1,500 ° C, and for grossular, 8.3 × 10^–12 cm² s^–1 at 1,200 ° C and 4.6 × 10^–11 cm² s^–1 at 1,300 ° C. The equilibrium values of D _Sm have been reversed by experiments with Sm-free pyrope and Sm-bearing glass, and with Sm-bearing grossular and Sm-free glass.Between 12 ppm and 1,000 ppm Sm in pyrope at 1,300 ° C and between 80 ppm and >2 wt.% Tm in pyrope at 1,500 ° C, partition coefficients are constant and independent of REE concentration. Above 100 ppm of Sm in garnet at 1,500 ° C, partition coefficients are independent of Sm concentration. At lower concentrations, however, D _Sm is dependent upon the Sm content of the garnet. The two regions may be interpreted in terms of charge-balanced substitution of Sm₃Al₅O₁₂ in the garnet at high Sm concentrations and defect equilibria involving cation vacancies at low concentrations. At very low REE concentrations (< 1 ppm Tm in grossular at 1,300 ° C) D_REE garnet/liquid again becomes constant with an apparent Henry's Law value greater than that at high concentrations. This may be interpreted in terms of a large abundance of cation vacancies relative to the number of REE ions.The importance of defects in the low concentration region has been confirmed by adding other REE (at 80 ppm level) to the system Mg₃Al₂Si₃O₁₂-H₂O at low Sm concentrations. These change D _Sm in the defect region, demonstrating their role in the production of vacancies.Experiments on a natural pyropic garnet indicate that defect equilibria are of importance to REE partitioning within the concentration ranges found in nature.