Based on the mean contents of elements and their standard deviations estimated in the first part of this contribution (Naumov et al., 2023), we compared in detail the geodynamic settings distinguished earlier. In order to compare the compositions of mafic melts, a correction was introduced for changes related to the fractionation of the main minerals. Using numerical criteria, the elements were arranged in a sequence of the increasing degree of compatibility during melting and crystallization, and regular variations in element contents normalized to the mean composition of oceanic-island melts were distinguished. The melts of mid-oceanic ridges show a monotonous increase in normalized contents from the most incompatible (Cs, Ba, U, La, etc.) to compatible elements (Sc, Ni, and Cr). The settings of convergent plate boundaries show relative enrichment in the most incompatible elements and significant negative Ta−Nb anomalies relative to neighboring elements. The magmas of continental rifts show the most significant enrichment in the most incompatible elements, as well as Pb, Li, and some other elements. Indicator element ratios showing significant variations between the settings were distinguished for mafic melts. Some element ratios are almost identical (within observed variations) in mafic melts from all the settings. The mean element ratios in mafic, intermediate, and silicic magmas show three types of behavior. Some ratios (including the canonic ratios Nb/Ta, Zr/Hf, etc.) in intermediate and silicic magmas are inherited from the composition of mafic melts. Some ratios show irregular changes from mafic to silicic melts (Sr/Cr, F/Th, etc.). Some other ratios change monotonously and significantly in the sequence from mafic to silicic melts (Ni/Yb, Lu/P, etc.). The variations in element ratios are related to the crystallization differentiation of melts and contributions of geochemically contrasting reservoirs.

The partition of trace elements between minerals (olivine, orthopyroxene, clinopyroxene, and feldspars) and silicate melts is analyzed based on experimental data within broad P–T ranges (from 1 atm to 10 GPa and ∼1000–2000°C) and the compositions of melts (from ultramafic to ultrasilicic) and minerals. The dependences of the logarithmic partition coefficients (lnDi) on P–T parameters and compositions are approximated by linear functions of 1/T, P/T (where P is pressure and T is temperature in K) and compositional parameters of the minerals and melts. The Di/Dj ratios of a large number of pairs of elements are found out to be independent of experimental parameters and vary within narrow ranges. The parameters of the dependences of Di on P–T and compositions are estimated by minimizing the squared deviations of model Di and Di/Dj values from experimental ones. The dependences thus derived make it possible to calculate Di for numerous elements accurate to a factor of 1.2–2.0. As an illustrative example, a model is discussed for the derivation of mafic basaltic melts in mid-oceanic ridges at the melting of a peridotite source and crystallization of primary magmas under crustal parameters.

We analyzed published datasets relating to the composition of glasses from melt inclusions in minerals of volcanics from the Eastern Volcanic Belt and Sredinny Range of Kamchatka. A significant difference was found between the distribution of silica concentration in the rocks and melts: intermediate and basic compositions are most common among the rocks, whereas the glasses in melt inclusions are predominantly acidic. The distribution of major and trace elements was analyzed. It was shown that the contents of some elements are environment-specific (e.g., Nb and light REEs). We identified trace element ratios in melts that most strongly correlate with the geodynamic setting.

— Knowledge of mineral–melt partition coefficients ( D ) is necessary for geochemical modeling of magma formation and evolution. The main source of these parameters is experiments on equilibrium between minerals and silicate melt. The database on mineral–melt equilibrium experiments has grown continuously, which allows one to refine partition coefficients and reveal the most important factors affecting them. This paper reports the analysis of available experimental data on trace element partitioning between olivine and melt (7000 experiments from 587 publications were used). Based on the statistical processing of the data array, the dependence of D on melt and olivine composition and P–T conditions was evaluated. It was found that most of incompatible elements are either insensitive or moderately sensitive to these parameters. Among the compositional parameters, CaO content in melt is most significant. It was shown that D estimates can be significantly improved by using ratios of D values for different elements. Such ratios are often independent of experimental parameters and much less variable than D values for particular elements. The obtained D estimates for basaltic compositions vary within six orders of magnitude, from <10 –5 (U, Th, and La) to ~5–10 (Co and Ni). The low D values for most elements (<0.1) indicate that many trace element ratios in melts do not change significantly even at high degrees of olivine crystallization. The obtained estimates were compared with data on element partitioning between high-pressure (Mg,Fe) 2 SiO 4 phases (wadsleyite and ringwoodite) and silicate melt and between olivine and carbonate–silicate melts. In both cases, there is a close similarity to element partitioning between olivine and silicate melt. One exception is REE. D REE values for the wadsleyite/ringwoodite–melt and olivine–carbonate melt systems are approximately one order of magnitude lower than those for olivine–silicate melt partitioning.

Data from our original database, which includes more than 2 600 000 analyses for 75 elements of mineral-hosted melt inclusions and quench glasses in volcanic rocks, are generalized to calculate the mean concentrations of major, volatile, ore, and trace elements in magmatic melts from the following dominant geodynamic environments: (I) spreading zones of oceanic plates (mid-oceanic ridges), (II) environments affected by mantle plumes in oceanic plates (oceanic islands and lava plateaus), (III, IV) environments related to subduction processes (III is zones of arc magmatism on the oceanic crust, and IV is zones of magmatism in active continental margins in which magma-generating processes involve the continental crust), (V) environments of continental rifts and areas with continental hotspots, and (VI) environments of backarc spreading. A histogram of SiO2 distribution in natural magmatic melts shows a bimodal distribution: one of the maxima falls onto SiO2 concentrations of 50–52 wt

Data from our original database, which includes more than 2 600 000 analyses for 75 elements of mineral-hosted melt inclusions and quench glasses in volcanic rocks, are generalized to calculate the mean concentrations of major, volatile, ore, and trace elements in magmatic melts from the following dominant geodynamic environments: (I) spreading zones of oceanic plates (mid-oceanic ridges), (II) environments affected by mantle plumes in oceanic plates (oceanic islands and lava plateaus), (III, IV) environments related to subduction processes (III is zones of arc magmatism on the oceanic crust, and IV is zones of magmatism in active continental margins in which magma-generating processes involve the continental crust), (V) environments of continental rifts and areas with continental hotspots, and (VI) environments of backarc spreading. A histogram of SiO2 distribution in natural magmatic melts shows a bimodal distribution: one of the maxima falls onto SiO2 concentrations of 50–52 wt % and the other onto 72–76 wt %. The most widely spread melts contain 62–66 wt % SiO2. Mean temperatures and pressures are calculated for each of the environments. The normalized multielemental patterns presented for environments I through VI show the ratios of the mean concentrations of elements in magmatic melts of mafic, intermediate, and felsic composition to the concentrations in the primitive mantle. Mean ratios of incompatible, trace, and volatile components (H2O/Ce, K2O/Cl, Nb/U, Ba/Rb, Ce/Pb, etc.) are evaluated for the melts of each of the environments. The variations in these ratios are calculated, and it is demonstrated that the ratios of incompatible elements are mostly statistically significantly different in the different environments. The differences are particularly significant between the ratios of the most differently incompatible elements (e.g., Nb/Yb) and some ratios involving volatile components (e.g., K2O/H2O).

The hypothesis of redox freezing is based on the assumption that a Fe–Ni metal phase becomes stable in the peridotite mantle at increasing pressure and can serve as a reducer for carbonate–silicate melts. $${\text{CO}}_{3}^{{2 - }}$$ reduction and formation of elementary C (graphite or diamond) result in an increase in the solidus temperature and melt freezing. Thermodynamic calculations show that equilibrium oxygen fugacity in peridotite with carbon and magnesite is significantly lower than the values buffered by the mineral assemblages of metasediments (garnet–kyanite–SiO2–aragonite–elemental carbon) or eclogites (pyroxene–garnet–magnesite–elemental carbon). Hence, redox interaction between carbon-bearing peridotites and metasediments or eclogites may occur in the absence of metal and even in a Fe-free system. To explore this suggestion, we conducted experiments on interaction between forsterite (as a peridotite proxy) with synthetic mixtures simulating carbonatized metasediment (SiO2 + CaCO3 + Al2O3) and carbonatized eclogite (SiO2 + MgCO3 ± Al2O3 ± CaO) at 10 GPa and 1200–1500°C. To reduce the transport of major components, the mixtures were separated by a graphite disc, which also served as a source of C. The interaction resulted in the decarbonation of the carbonate-bearing metasediment or eclogite with diamond formation on the surface of the graphite disc. The graphite disc was dissolved at contact with peridotite, and metasomatic zoning developed. Pyroxene and magnesite with low Ca contents appeared in the distal metasomatic zone. The contents of Ca in the newly formed pyroxene and carbonate increases toward the graphite disc, and high-Ca pyroxene and garnet were observed in the proximal metasomatic zone. The obtained results indicate that coupled redox reactions occur in peridotite and metasediment (or eclogite): Mg2SiO4 + C + O2 = MgSiO3 +MgCO3 and CaCO3 + 1/3Al2S-iO5 + 2/3SiO2 = 1/3Ca3Al2Si3O12 + C + O2, respectively. The reactions occur owing to oxygen diffusion along intergranular melt channels. The interaction also involves the transfer of major cations, which resulted in the formation of carbonatized lherzolite and a diamond-bearing eclogite assemblage. Such a process is possible in nature at a contact of carbonated metasediment or eclogite with peridotite. The obtained results indicate that the presence of Fe–Ni metal is not necessary for redox freezing. The processes modeled in the experiments provide a possible explanation for the existence of diamond-rich eclogites and the scarcity of diamond in peridotite xenoliths.

In the second part of this study, we analyze how crystallization differentiation can affect concentrations of elements and their ratios in melt inclusions and glasses of rocks in major geodynamic environments. The paper presents analysis of experimental data on the partition coefficient of elements between minerals (olivine, pyroxenes, garnet, amphibole, biotite, sulfide, apatite, spinel, ilmenite, rutile, and zircon) and silicate melts, which were discussed in the first part of this study. It is demonstrated that the crystallization of major minerals only insignificantly affects the ratios of incompatible elements. The partition coefficients of some elements between accessory minerals and melts can be very high, but the effects of crystallization differentiation cannot be significant because of the small amounts of the crystallizing phases. These effects are the most significant for chalcophile elements (Cu, Ni, and others), at the separation of sulfides, and for Nb and Ta, at the crystallization of rutile. Differences in concentrations of various elements and their ratios to Cs concentrations are discussed with reference to various geodynamic environments. The maximum values of the ratios of practically all elements to Cs were found in melts in mid-oceanic ridges. The melts of oceanic islands and backarc basins are characterized by relatively low ratios, without any significant anomalies. The lowest ratios of elements to Cs were found in melts of continental and marginal environments. These melts are also characterized by clearly seen geochemical anomalies that are typical of rocks of the corresponding environments (negative Ta–Nb, positive Pb, and other anomalies).

Published data on the composition of mineral-hosted inclusions and quenched glasses of rocks were used to estimate the mean concentrations of 45 volatile, trace, and ore elements in silicate igneous melts from the main geodynamic settings of the Earth and in natural fluids. The following geodynamic settings were distinguished according to the conditions of formation and evolution of the igneous melts: (I) oceanic spreading zones (mid-oceanic ridges), (II) oceanic mantle-plume zones (oceanic islands and lava plateaus); (III and IV) subduction-related settings (III is island-arc zones, and IV is active continental margins); (V) continental rifts and hotspot zones; and (VI) backarc spreading basins related to subduction. The contents of the elements in basic and felsic melts were compared in settings III, IV and V. It was shown that differences in the enrichment factors of ore elements between the geodynamic settings could be caused by variations in the contribution of fluids to element transport and accumulation. Ratios of element contents in each of the geodynamic environments to the global mean values were calculated.

The mean concentrations of volatiles, major, and trace elements are estimated in the magmatic melts of Kamchatka based on an our database that includes analyses of melt inclusions and quenched glasses of rocks for 75 elements (the database comprised 1 900 000 analyses as of late 2018). The determined concentrations are compared with analogous characteristics of melts from island arcs and active continental margins. The distribution of SiO2 concentrations (more than 105 000 analyses) in natural magmatic melts from all geodynamic environments is obviously bimodal, with maxima at SiO2 = 50–52 and 72–76 wt %. The paper presents binary diagrams that show the concentrations of major, volatile, trace elements, and REE and diagrams of the normalized patterns average concentrations of elements in the magmatic melts. The diagrams show distinguishing features of the melts of Kamchatka, for example, the elevated Ba/Nb ratios of Kamchatka mafic melts compared to those of melts in other zones, which may reflect a high content of subduction-related material during the derivation of the Kamchatka melts, with their Th deficit controlled by specifics of the fluid regime.

Inclusions of breyite (previously known as walstromite-structured CaSiO3) in diamond are usually interpreted as retrogressed CaSiO3 perovskite trapped in the transition zone or the lower mantle. However, the thermodynamic stability field of breyite does not preclude its crystallization together with diamond under upper-mantle conditions (6–10 GPa). The possibility of breyite forming in subducted sedimentary material through the reaction CaCO3 + SiO2 = CaSiO3 + C + O2 was experimentally evaluated in the CaO–SiO2–C–O2 ± H2O system at 6–10 GPa, 900–1500 ∘C and oxygen fugacity 0.5–1.0 log units below the Fe–FeO (IW) buffer. One experimental series was conducted in the anhydrous subsystem and aimed at determining the melting temperature of the aragonite–coesite (or stishovite) assemblage. It was found that melting occurs at a lower temperature (∼1500 ∘C) than the decarbonation reaction, which indicates that breyite cannot be formed from aragonite and silica under anhydrous conditions and an oxygen fugacity above IW – 1. In the second experimental series, we investigated partial melting of an aragonite–coesite mixture under hydrous conditions at the same pressures and redox conditions. The melting temperature in the presence of water decreased strongly (to 900–1200 ∘C), and the melt had a hydrous silicate composition. The reduction of melt resulted in graphite crystallization in equilibrium with titanite-structured CaSi2O5 and breyite at ∼1000 ∘C. The maximum pressure of possible breyite formation is limited by the reaction CaSiO3 + SiO2 = CaSi2O5 at ∼8 GPa. Based on the experimental results, it is concluded that breyite inclusions found in natural diamond may be formed from an aragonite–coesite assemblage or carbonate melt at 6–8 GPa via reduction at high water activity.

Ferropericlase is a common lower mantle phase that can occur as inclusions in “superdeep” diamonds. However, it has been proposed that some of these inclusions could be formed in the upper mantle (Brey et al. 2004). To explore this hypothesis, two types of experiments on ferropericlase crystallization were conducted at 5–12 GPa and 1500–1700 °C: (1) equilibrium experiments with Ca–Mg–Fe carbonate + MgO + olivine mixtures and (2) “sandwich” experiments containing a layer of metallic iron overlain by olivine, with a carbonate layer on top. The equilibrium experiments produced a series of carbonate–silicate melts saturated in olivine and ferropericlase. These melts contain 2–12 wt% SiO2, which correlates negatively with CaO and CO2 contents and positively with temperature. A comparison of the composition of the carbonate–silicate melts with literature data for melts saturated in olivine and low-Ca pyroxene indicate that melts saturated in ferropericlase cannot be produced by crystallization of melts derived by melting carbonated harzburgite or lherzolite at upper mantle conditions. On the other hand, the sandwich experiments reveal that ferropericlase and diamond (or metastable graphite) can crystallize simultaneously during the reduction of carbonate–silicate melt. The resulting ferropericlase crystallizing in equilibrium with olivine will be richer in Fe compared with lower mantle ferropericlase in equilibrium with bridgmanite. Thus, the considerable variation in Mg# values observed for some suites of ferropericlase inclusions in diamond could in part be attributable to ferropericlase formation in the upper mantle.

As a continuation of our previous study, we estimated the mean contents of volatile, major, and trace components in silicic (>66 wt % SiO2) magmatic melts from main terrestrial geodynamic settings on the basis of our database, which includes (as of middle 2017) more than 1 500 000 determination of 75 elements in melt inclusions and quench glasses from rocks. Among the geodynamic settings are those related to subduction processes (III, island-arc zones originated on oceanic crust and IV, magmatic zones of active continental margins, where continental crust is involved in magma formation) and intracontinent rift and continental hot-spot regions (V). For each geodynamic setting, we calculated the mean contents of elements with confidence limits separately for melt inclusions and groundmass glasses and for the entire data set. Systematic differences were found between the mean compositions of melt inclusions and groundmass glasses from these geodynamic settings. Primitive mantle normalized spider diagrams were constructed for all geodynamic settings. Some ratios of elements and volatile components (H2O/Ce, K2O/Cl, La/Yb, Nb/U, Ba/Rb, Ce/ Pb, etc.) in silicic and mafic melts were compared. Variations in the ratios of various elements to Th, which is one of the most incompatible elements in silicic and mafic melts, were discussed.

Minor amounts of C may have a profound effect on the melting of mantle peridotites.Previous experimental studies in simple and complex systems showed that the presence of carbonate phases strongly depresses the solidus of mantle peridotite at pressures >3 GPa (e.g.Wyllie & Huang, 1976;Dasgupta & Hirschmann, 2006).The composition of low-degree melts changes from dolomitic to kimberlite-like with increasing temperature (e.g.Wallace & Green, 1988;Dalton & Presnall, 1998).Therefore, experimental studies of CO 2 -bearing mantle systems may have intriguing applications for mantle metasomatism and origin of primary carbonatite and kimberlite melts.In addition to CO 2 , the low-degree melts in the upper mantle probably contain considerable amounts of other volatile and incompatible elements.This study focuses on the influence of water and fluorine on the melting of magnesite-bearing mantle peridotite and is an extension of our previous experiments in the peridotite-CO 2 system (Brey et al., 2008, hereafter B08).

Sulfur solubility in Ca-Mg-Fe silicate-carbonate melt was experimentally determined at 5-10.5 GPa and 1400-1600 degrees C. Melt was produced by reaction between natural carbonates (calcite and magnesite) and San Carlos olivine (X-Fo similar to 0.9), which was used as a container, and equilibrated with either Fe-Ni sulfide liquid under reducing conditions (in the presence of graphite) or CaSO4 liquid under oxidizing conditions (Re-ReO2 buffer). To overcome the problem of the presence of sulfide droplets within the silicate-carbonate melt and to reverse the equilibrium, two sample configurations were employed: (1) a carbonate-sulfide mixture placed in the center of the olivine container and (2) sulfide placed in the olivine container and carbonate initially positioned outside the container near the wall of the external Pt capsule. In the former case, silicate-carbonate melt with suspended sulfide melt droplets infiltrated through the olivine capsule, and the two melts were separated owing to their different physical properties. In the latter case, silicate-carbonate melt migrated through thin fractures in the olivine container to come into contact with sulfide. The results of the two experimental series constrained the solubility of sulfide S in silicate-carbonate melt at 0.02-0.10 wt%. The S-solubility increases with increasing FeO content in the melt and is less sensitive to pressure, temperature and other compositional parameters. The solubility of S in the presence of the Re-ReO2 buffer is much higher (up to 2-3 wt%) and dominated by S6+. The obtained results indicate that S is relatively inert during carbonate mantle metasomatism under reduced conditions, but can be transported under oxidizing conditions.

Diamond nucleation and growth were investigated experimentally at 7.5 and 10.5 GPa and temperatures up to 1500 degrees C. Samples consisted of two layers: i) H2O- and CO2-bearing model sediment and ii) graphite-bearing garnet harzburgite comprising natural minerals. Two experimental series were conducted, one under a controlled temperature gradient with the sedimentary layer usually in the cold zone and the other under isothermal conditions. In the latter case, diamond seeds were added to the sedimentary mixture. During the experiments, the sedimentary layer partially or completely melted, with the melt percolating and interacting with the adjacent harzburgite. The graphite-to-diamond transition in the peridotite was observed above 1300 degrees C at 7.5 GPa and 1200 degrees C at 10.5 GPa in the temperature-gradient experiments, and at temperatures similar to 100 degrees C lower in the isothermal experiments with diamond seeds. Newly formed diamond occurs mostly as individual grains up to 10 pm in size and is separate from graphite aggregates. In some cases, an association of diamond with magnesite was observed. Diamond nucleation occurs in hydrous and CO2-bearing silicate melt following graphite dissolution and reaystallization. In the case of the diamond-magnesite association, diamond was probably formed through carbonate reduction coupled with graphite oxidation. The composition of the melts ranged from "carbonatitic" with similar to 10 wt% SiO2 and >50 wt% volatiles to hydrous silicate with similar to 40 wt% SiO2 and <10 wt% volatiles. This variation has no strong effect on diamond nucleation or growth. (C) 2018 Elsevier B.V. All rights reserved.

To better understand processes above subducted oceanic slabs, we have undertaken experiments with juxtaposed sediment and peridotite layers at pressures of 7.5 and 10.5 GPa at a controlled temperature gradient from similar to 100 to similar to 500 degrees C per a sample length of similar to 3 mm. The sediment starting material contains H2O (6.9 wt%) and CO2 (5.9 wt%) and has a major-element composition similar to GLOSS (Plank and Langmuir, 1998) doped with trace elements at 10-100 ppm levels. Several experiments were conducted with similar to 0.5 wt% Cl or F. The peridotite layer is composed of natural olivine (66 wt%), orthopyroxene (27 wt%) and garnet (7 wt%) mixed with similar to 15 wt% graphite. Several experimental configurations were investigated, but the "basic" setup has the sediment layer at the bottom in the cold zone (400-1200 degrees C) overlain by peridotite at 900-1500 degrees C. The temperature distribution was determined by two thermocouples and orthopyroxene-garnet thermometry. Features common to many experiments are (1) the development of multiple layers of various lithologies and a pool of hydrous silicate or carbonate-silicate melt in the hottest part of the capsule; (2) replacement of olivine by orthopyroxene in the metaperidotite; (3) preservation and growth of garnet and local development of magnesite in the metaperidotite layer; (4) enrichment in garnet within the metasediment layer at the contact with the metaperidotite; (5) formation of a clinopyroxene-garnet assemblage at the bottom (the coldest part); (6) presence of K-bearing phases (phlogopite or phengite) and carbonates in the metasediment layer only at temperatures <700 degrees C; and (7) occurrence of accessory zircon, rutile and phosphates in the coldest regions. In terms of element redistribution, the peridotite becomes strongly enriched in SiO2 compared to the starting composition, and the sediment gains MgO, FeO and Cr2O3. Potassium is fully extracted into the melt, while Na and Ca are largely retained in the coldest part of the metasediment layer in clinopyroxene, Ca-rich garnet and aragonite. The melt is a product of interaction between partial melt or fluid from the sediment and peridotite. It has a silico-carbonatite composition with variable SiO2, MgO, FeO and CaO contents and low Al2O3. The addition of Cl has almost no effect on element distribution, whereas the addition of F results in the appearance of humite-group minerals containing significant amounts of Ti. Trace-element distribution is controlled by pressure, temperature and mineral assemblages. At low temperatures in the sediment layer (<700 degrees C) Ba, Rb, Sr and Li are much more mobile than REE and HFSE, which results in high Ba/La, Ba/Nb, Sr/Nb etc. (fluid metasomatism). At higher temperatures in the sediment layer, the melt ismarkedly enriched in Ba, Rb, Sr, LREE and U relative to Ti, MREE and HREE. Negative Nb-Ta and Zr-Hf anomalies in melts are caused by the retention of rutile, zircon and humite-group minerals in the solid residue. Thermodiffusion may affect the ratios of some highly incompatible elements (e.g., Ta/La). Possible applications of the results to natural deep subduction are discussed in view of variations in mineral assemblages and trace element ratios. (C) 2017 Elsevier Ltd. All rights reserved.