Fe-rich Zone Research Articles

Observations and simulations for phase separation process of immiscible Fe50Sn50 alloy droplets placed on a chilling surface

Liquid phase separation and microstructure evolution mechanisms of immiscible Fe50Sn50 alloy droplets placed on a chilling surface have been investigated by both arc melting experiments and lattice-Boltzmann simulations. The pattern selection of phase-separated morphologies strongly depended on the sample diameter. With the reduction in sample size, phase separation time was shortened, two-layer macrosegregation morphologies first transformed into intermediate phase-separated morphologies and finally changed into a homogeneously dispersed microstructure. The farther away from the sample bottom the longer the phase separation time and the larger the Sn-rich phase. A heat transfer equation was proposed to analyze the temperature field characteristics of Fe50Sn50 alloy droplets placed on the surface of a water-cooled copper mold. Theoretical analyses revealed that the sample gradually cooled from the bottom to the top of the sample. There existed a relatively high temperature gradient of several hundred Kelvins per millimeter between the top and the bottom of the arc melted sample, which induced the occurrence of an extremely intense Marangoni convection. The smaller the sample size the larger the temperature gradient and the greater deviation-degree of Fe-rich zone in two-layer macrosegregation structure from the bottom of the sample. The simulated two-layer phase-separated morphology agreed well with the experimental observations. Numerical simulations demonstrated that the effect of Marangoni convection on the phase separation was significantly stronger than the Stokes motion, and Fe-rich globules with a larger density tended to move upward to form a Fe-rich zone deviating from the bottom of the sample.

A comparative study of metastable phase separation for undercooled liquid Fe35Cu65 alloy under natural and forced cooling conditions

The influences of the cooling style on metastable phase separation (MPS) kinetics of undercooled liquid Fe35Cu65 alloy have been systematically investigated by glass fluxing technique and lattice-Boltzmann method. MPS was observed in the undercooling range of 13–285 K. With a rise in the undercooling, phase separation time increased linearly. Meanwhile, a microsegregation morphology, which appeared as Fe-rich globules near the sample surface and well-defined αFe dendrites far from the sample surface, transformed into the eccentric core-shell macrosegregation morphology composed of a floating Fe-rich core and a sinking Cu-rich shell. Under the forced cooling condition, the macrosegregation formation required a larger critical undercooling. In contrast with the naturally cooled sample, the phase separation time of forcedly cooled sample was much shorter, and both the eccentric level of the core-shell macrosegregation and the volume fraction of Fe-rich zone were smaller at the same undercooling. The MPS and microstructure evolution of liquid Fe35Cu65 alloy under the natural and forced cooling conditions were numerically simulated and experimentally confirmed. Theoretical analyses revealed that the fluid dynamics caused by the Stokes motion and Marangoni migration of Fe-rich globules were strongly dependent on the cooling style, the size and the position of Fe-rich globules during the MPS.

Metastable phase separation kinetics controlled by superheating and undercooling of liquid Fe-Cu peritectic alloys

The effects of superheating and undercooling on the metastable phase separation (MPS) kinetics for glass-fluxed Fe65Cu35 peritectic alloy were investigated by experiments and simulations. MPS occurred in the experimental undercooling range of 175–297 K, and induced the eccentric core-shell macrosegregation (ECSM) structure. It was first demonstrated that the relationship of the phase separation undercooling ΔTPS and the phase separation time tPS with the superheating ΔTH and the undercooling ΔT could be described by curved-surface functions. The increasing ΔTPS was observed under the smaller ΔTH or greater ΔT condition. Both the rising ΔTH and the ΔT contributed to the extension of the tPS, and then resulted in the expansion of Cu-rich zone as well as the microhardness improvement of Fe-rich zone, indicating the more thorough MPS. Numerical simulations showed that the surface segregation, Marangoni convection and Stokes sedimentation mainly pushed the evolution of the ECSM.

Distribution and Geochemistry of Arsenic in Sediments of the World’s Largest Choked Estuary: the Patos Lagoon, Brazil

The arsenic (As) contents of sediment cores from the Patos Lagoon in southern Brazil were measured to better understand the biogeochemical cycling of As in the estuary. Sediment cores (ca. 60 cm) were obtained from three locations within the estuary to capture possible changes in As content across the salinity gradient (i.e., where saline, brackish, and freshwater dominated). Two sediment cores were collected at each location, one beneath open water and the other from the fringing salt marsh. Along with As, we quantified the particle size; redox potential (Eh); manganese (Mn), iron (Fe), total organic carbon (TOC), and free (dissolved) sulfide concentrations; and acid volatile sulfide (AVS) and chromium reducible sulfide (CRS) contents. Bioturbation supports oxygen penetration to depths between 20 and 30 cm below the salt marsh surface, where an As- and Fe-rich zone was identified (3-fold higher As than mean As contents of the sediments). A similar subsurface peak of As, Fe, and Mn occurs in the open-water cores, albeit at greater depths between 40 and 50 cm below the surface. The subsurface peak has As concentrations that are 2-fold higher than the average for each core. Vertical profiles of Eh, free sulfides, and CRS for the shallow open-water cores showed similar distribution at depths of 50 cm, suggesting that pyrite formation is an important sink for As in the open-water cores. The data demonstrate clear differences in the geochemical conditions for salt marshes and shallow open waters that can have important implications for As distribution in estuarine sediments.

Chemical composition, genesis and exploration implication of garnet from the Hongshan Cu-Mo skarn deposit, SW China

In this study, we present new mineral chemical data on the garnet from the Hongshan Cu-Mo deposit, one of the largest skarn deposits in the Sanjiang Tethyan tectonic domain, in SW China. The aims are to unravel the formation mechanism of the oscillatory-zoned garnet at Hongshan, and to explore possible applications using garnet trace element to distinguish different types of skarn deposits.Two generations of garnets, i.e., poorly-zoned garnet I (Grt I) and the well-zoned garnet II (Grt II), are identified from the endoskarn and exoskarn, at Hongshan based on field and petrographic observations. EPMA and LA-ICP-MS analyses show that Grt I contains a narrow compositional range (Adr40.07-61.25 Grs36.65-57.08), and chondrite-normalized HREE-enriched and LREE-depleted patterns with minor positive Eu anomalies. In contrast, Grt II shows a wide compositional range (Adr27.78-67.53 Grs30.67-70.92 to almost pure andradite Adr94.41-99.99 Grs0.00-4.21) with interlayered Al-rich and Fe-rich zones. In Grt II, its Al-rich zone is chemically similar to Grt I, whereas its Fe-rich zone exhibits chondrite-normalized LREE-enriched and HREE-depleted patterns with distinct positive Eu anomalies. The positive REE3+ vs. Mg correlations, and the poor REE3+ vs. Y and Ca correlations, altogether indicate that the REE3+ incorporated into garnet was co-influenced by the menzerite-type substitution mechanism (Ca2+-1VIIIREE3++1VIIIAl3+-1VIMg2++1VI), fluid chemistry, and the physicochemical conditions of precipitation. The higher U contents and HREE-enriched patterns of Grt I suggest that it was formed under a relatively reducing and near-neutral pH condition. Varying U contents and REE patterns across different oscillatory zones indicate that Grt II was probably formed under alternating physicochemical fluctuations led by repeating crack-seal cycles in the hydrothermal system. Besides, chemical data compilation suggests that the garnet from different types of skarn deposits has different compositions, and can be used as mineral prospecting tools. We suggested that discrimination plots, such as δEu vs. V and δCe vs. U, can distinguish Cu skarn deposit from other skarn deposit types, whilst the U vs. W plot can discriminate between Cu, W, W-Sn and W-Mo skarn deposits.

Phase Separation of Fe-Cu-Pb Alloy and Recycling of Mixed Metals in Waste Printed Circuit Boards

The pyrolysis experiment was carried out on the waste printed circuit boards (WPCBs) of mobile phones to obtain the mixed metals which contains more than ten valuable metals. The main components in the mixed metals are elements Fe, Cu, Pb, Sn. The liquid-liquid phase separation behavior of (Fe0.4Cu0.6)100-xPbx ternary alloy has been studied. The introduction of Pb into the metastable immiscible Fe-Cu alloy can result in a stable liquid-liquid phase separation into L(Fe) and L(Cu,Pb) liquids. With the increasing of Pb content, the second phase separation in the residual L(Cu,Pb) liquid was detected, resulting in the formation of three-zone-separation structure. On this basis, a hierarchical separation system was designed to recycle mixed metals in super gravity field. The results show that the metals Cr, Co, Ni, Si are mainly enriched in the Fe-rich zone, the precious metals Au, Ag and a small amount of Zn are concentrated in the Cu-rich zone, while the low-melting-point metals Sn, Bi, Cd, and In are collected in the Pb-rich zone.

Fracture toughness of WC-Fe cermet in W-WC-Fe composite by nanoindentation

To enhance the strength of iron-based alloys, tungsten carbide (WC)-reinforced iron (Fe) matrix composite with a double-scale structure is proposed and produced through a general heat-treatment process at 1150 °C for 40 min. X-ray diffraction (XRD) and scanning electron microscopy (SEM) analyses reveal that the double-scale structure involves both reinforcement elements of WC particle and W-WC pillar, where the WC-Fe cermet enwraps the metal tungsten wire. The chemical composition of WC-Fe cermet is studied, and the contents of Fe and W elements present a gradient distribution. Moreover, the hardness, elastic modulus and fracture toughness of WC-Fe cermet are evaluated via nanoindentation. The evaluation of toughness is based on the indentation method and the appropriate selection of equation at the same load. The results obtained show that the fracture toughness value is within the range of 4.11 ± 0.91–7.89 ± 1.02 MPa m1/2, which is considerably dependent on the composition. The investigations of crack surface morphologies indicate that radial cracks at the corners of the indentation originate from the crossing of slip bands and that the toughening mechanism is mainly crack deflection, grain bridging and intergranular fracture in the Fe-rich zone.

Phase separation and subsequent solidification of peritectic Fe–Cu–Ge alloys subjected to substantial undercooling processing

Liquid ternary Fe50-x/2Cu50-x/2Gex (x = 5 and 15) alloys were highly undercooled by glass fluxing technique, and the achieved maximum undercoolings are 300 K (0.18TL) and 268 K (0.16TL), respectively. At small undercoolings, these two alloys solidify in a normal pathway of peritectic solidification. If the alloy undercooling continues to increase beyond a certain value, metastable liquid phase separation takes place. The critical undercooling for Fe47.5Cu47.5Ge5 alloy is 70 K and that for Fe42.5Cu42.5Ge15 alloy is 110 K. At larger undercoolings, the homogeneous liquid alloy separates into an Fe-rich zone and a Cu-rich zone. In such a case, the undercooled alloy displays a monotectic solidification structure. The rapid solidification of the Fe-rich zone is the keystone of alloy solidification, and the growth velocity of the primary phase rises with increasing alloy undercooling according to a power law relation. Within the experimental undercooling ranges, the primary γFe phase of Fe47.5Cu47.5Ge5 alloy attains a maximum growth velocity of 19 m/s, while the growth velocity of the primary Fe0.84Ge0.16 phase reaches 11 m/s in undercooled Fe42.5Cu42.5Ge15 alloy. Moreover, energy dispersive spectroscopy (EDS) analysis shows that the Fe-rich and Cu-rich liquid phases have a similar affinity with solute Ge.

Fluid convection and solidification mechanisms of liquid Fe50Cu50 alloy under electromagnetic levitation condition

In the electromagnetic levitation experiment, the liquid flow in the undercooled liquid alloy remarkably affects the relevant thermodynamic property measurement and solidification microstructure. Therefore, it is of great importance to understand the fluid convection inside the undercooled melt. Theoretical calculation and electromagnetic levitation experiment have been used to investigate the internal velocity distribution and rapid solidification mechanism of Fe50Cu50 alloy. Based on axisymmetric electromagnetic levitation model, the distribution patterns of magnetic flux density and inducted current for levitated Fe50Cu50 alloy are calculated together with the mean Lorenz force. The Navier-Stokes equations are further taken into account in order to clarify the internal fluid flow. The results of the theoretical calculation reveal that the fluid velocity within levitated melt is strongly dependent on three factors, i.e., current density, current frequency and melt undercooling. As one of these factors increases, the maximum fluid velocity decreases while the average fluid velocity increases. Meanwhile, the area with fluid velocity larger than 100 mm·-1 is significantly extended. Furthermore, the fluid flow within levitated melt displays an annular tubular distribution characteristic. The Fe50Cu50 alloy melt is undercooled and solidified under electromagnetic levitation condition. In this undercooling regime △ T50Cu50 alloy melt has suppressed phase separation substantially. Once the undercooling attains a value of 150 K, metastable phase separation leads to the formation of layered pattern structure consisting of floating Fe-rich zone and sinking Cu-rich zone. A core-shell macrosegregation morphology with the Cu-rich zone distributed in the center and outside of the sample and Fe-rich zone in the middle occurs if the undercooling increases to 204 K. With the enhancement of undercooling after phase separation, the grain size of α -Fe dendrites in Cu-rich zone presents a decreasing trend. In contrast to the phase separated morphology of Fe50Cu50 alloy under the glass fluxing condition, the phase separated morphologies show obviously different characteristics. In such a case, the forced convection induced by electromagnetic stirring results in the formation of wavy interface between Fe-rich and Cu-rich zones, the distorted morphology of the Cu-rich spheres distributed in the Fe-rich zone, and the increased appearance probabilities of Cu-rich spheres at the upper part of electromagnetically levitated sample. Experimental observations demonstrate that the distribution pattern of Cu-rich spheres in Fe-rich zone is influenced by the tubular fluid flow inside the melt.

Solidification pathways of ternary Cu62.5Fe27.5Sn10 alloy modulated through liquid undercooling and containerless processing

The active control of microstructure evolution is still a challenging factor for the development of advanced immiscible alloys. Here, we make an attempt to modulate the solidification pathways of undercooled Cu62.5Fe27.5Sn10 alloy by glass fluxing and drop tube techniques. Through regulating the liquid undercooling, three types of microstructures, dendrite, dispersive structure and macrosegregation pattern, were formed under the normal gravity condition. Below the first critical undercooling of 15 K, the alloy melt displayed the normal peritectic solidification. At moderate undercoolings above 15 K, the metastable liquid phase separation took place and the solidified microstructure appeared as homogeneously dispersed structure. If undercooling further overtook the second threshold of 107 K, macrosegregation occurred and the bulk alloy separated into an Fe-rich zone and a Cu-rich zone. Under the free fall condition, the alloy droplets with the droplet diameter beyond 805 μm showed the equilibrium peritectic solidification. If the droplet diameter decreased below 805 μm, the metastable liquid phase separation was induced and the microstructural morphology of Cu62.5Fe27.5Sn10 alloy droplet evolved from dendrite into dispersive structure. Furthermore, experimental and simulated results revealed that the temperature gradient had great influence on the size distribution of Fe-rich globules.

Liquid phase separation and subsequent dendritic solidification of ternary Fe35Cu35Si30 alloy

Liquid Fe35Cu35Si30 alloy has achieved the maximum undercooling of 328 K (0.24TL) with glass fluxing method, and it displayed triple solidification mechanisms. A critical undercooling of 24 K was determined for metastable liquid phase separation. At lower undercoolings, α-Fe phase was the primary phase and the solidification microstructure appeared as homogeneous well-defined dendrites. When the undercooling exceeded 24 K, the sample segregated into Fe-rich and Cu-rich zones. In the Fe-rich zone, FeSi intermetallic compound was the primary phase within the undercooling regime below 230 K, while Fe5Si3 intermetallic compound replaced FeSi phase as the primary phase at larger undercoolings. The growth velocity of FeSi phase increased whereas that of Fe5Si3 phase decreased with increasing undercooling. For the Cu-rich zone, FeSi intermetallic compound was always the primary phase. Energy-dispersive spectrometry analyses showed that the average compositions of separated zones have deviated substantially from the original alloy composition.

Dendritic Growth Characteristics of Cu-Rich Zone within Phase Separated Fe<sub>50</sub>Cu<sub>50</sub> Alloy

The undercooled Fe50Cu50 alloy experiences a metastable liquid phase separation and separates into a Fe-rich zone and a Cu-rich zone within the gravity field. The growth characteristics of the Cu-rich zone were investigated by the glass fluxing method, and the achieved undercooling range was 20−261 K. The volume fraction of the Cu-rich zone decreases with the enhancement of the bulk undercooling. The microstructural morphologies of the Cu-rich zone are similar at all the undercooling conditions, that is, αFe dendrites and particles are distributed inside (Cu) phase matrix. The secondary dendritic arm spacing of αFe dendrites decreases with the increase in bulk undercooling. The growth mechanism of αFe dendrites was analyzed by using the LKT/BCT dendritic growth theory. The dendritic growth in the Cu-rich zone is mainly controlled by solute diffusion so that the dendritic growth velocity is only several millimeters per second. Besides, the calculated results indicate that there is only inconspicuous solute trapping during the solidification of Cu-rich zone.

Solute redistribution during phase separation of ternary Fe–Cu–Si alloy

Ternary Fe48Cu48Si4 immiscible alloy was rapidly solidified under the containerless microgravity condition inside a drop tube. Liquid phase separation took place in the alloy melt and led to the formation of various segregated structures. The core–shell structure consisting of Fe-rich and Cu-rich zones and the homogenously dispersed structure were the major structural morphologies. Phase field simulation results revealed that the two-layer core–shell was the final structure of liquid phase separation. The solute redistribution of liquid Fe48Cu48Si4 alloy experienced the macroscopic solute distribution induced by liquid phase separation, the secondary phase separation within the separated liquid phases and the solute trapping during rapid solidification. Energy dispersive spectroscopy analysis showed that the solute Si was enriched in the Fe-rich zone whereas depleted in the Cu-rich zone. In addition, both αFe and (Cu) phases in the Fe-rich zone exhibited a conspicuous solute trapping effect. As compared with (Cu) phase, αFe phase had a stronger affinity with solute Si.

A comparative study of dendritic growth within undercooled liquid pure Fe and Fe50Cu50 alloy

The dendritic growth characteristics within undercooled liquid pure Fe and binary Fe50Cu50 alloy were investigated by glass fluxing and electromagnetic levitation methods. The growth velocity of pure Fe dendrite increases with undercooling, which reaches 69ms−1 at the maximum undercooling of 280K (0.15Tm). The microstructure appears as well-defined dendrites under slight undercooling, while it is characterized by a network of grain boundaries without discernible dendrites under substantial undercooling. A glass fluxing method was used to study the dendritic growth within the undercooled liquid Fe50Cu50 alloy. The maximum undercooling achieved was 261K (0.15TL), and the corresponding growth velocity was 15ms−1. A critical undercooling for metastable liquid phase separation of 7K was determined. At larger undercoolings, the sample separates into an Fe-rich zone and a Cu-rich zone. The microstructure of the Fe-rich zone displays well-branched dendrites in the condition of slight undercooling, whereas it transforms into monotectic solidification structures in the highly undercooled regime. Slender dendrites and spheres are characteristics of the structural morphology of the Cu-rich zone at all of the undercoolings. Energy-dispersive spectroscopy analysis shows that the average compositions of phase-separated zones have a large deviation from the original alloy composition, and conspicuous solute trapping has taken place.

Phase and microstructure of TbCu7-type SmFe melt-spun powders

The SmxZr0.3Fe9.1–xCo0.6 (x=0.8, 0.9, 1.0) powders were prepared by melt-spun method with different quenching velocities. The phase and microstructure were studied by X-ray diffraction (XRD), scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The Th2Zn17-type structure of the as-cast state changed to TbCu7-type after quenching to a rotating molybdenum roll under certain velocity, and the formation of TbCu7-type phase was strictly depending on the Sm content and roll speed. The SEM morphology showed that the Fe-rich zone was typically fish-bone structure and TEM diffraction pattern indicated the nano-scale crystal size with TbCu7-structure when x=0.9, and FCC type γ-Fe on the basis of α-Fe formed in the non-equilibrium solidification could be detected by selected area electron diffraction (SAED) indexing in the x=0.8 samples.

Metastable Phase Separation and Concomitant Solute Redistribution of Liquid Fe-Cu-Sn Ternary Alloy

Liquid Fe-Cu-Sn ternary alloys with lower Sn contents are usually assumed to display a peritectic-type solidification process under equilibrium condition. Here we show that liquid Fe47.5Cu47.5Sn5 ternary alloy exhibits a metastable immiscibility gap in the undercooling range of 51–329 K (0.19TL). Macroscopic phase separation occurs once undercooling exceeds 196 K and causes the formation of a floating Fe-rich zone and a descending Cu-rich zone. Solute redistribution induces the depletion of Sn concentration in the Fe-rich zone and its enrichment in the Cu-rich zone. The primary Fe phase grows dendritically and its growth velocity increases with undercooling until the appearance of notable macrosegregation, but will decrease if undercooling further increases beyond 236 K. The microsegregation degrees of both solutes in Fe and Cu phases vary only slightly with undercooling.

Macroscopic phase separation and primary FeSi compound growth within undercooled ternary Fe–Sn–Si monotectic alloy

Liquid ternary Fe37Sn32Si31 monotectic alloy in bulk form has been undercooled by 305 K (0.18T L) by the glass fluxing method. Macroscopic phase separation took place and led to the formation of a top Fe-rich zone and a bottom Sn-rich zone. The large degree of undercooling was found to facilitate the evolution of macrosegregation. Dendritic growth was the characteristic of the primary FeSi compound in an Fe-rich zone. The growth velocity was up to 0.42 m/s and the dendrite grains were efficiently refined by an increase in the level of undercooling. The solute trapping of Sn atoms could not occur in the FeSi compound within the undercooling range achieved here, resulting in the suppression of the dendritic growth velocity compared with that in binary Fe50Si50 alloy. The interdendritic microstructures mainly consisted of decomposed Fe5Si3 compound plus some α-Fe phase.

Spontaneous near-substrate composition modulation in electrodeposited Fe–Co–Ni alloys

Conventional and reverse depth profile analysis of electrodeposited Fe–Co–Ni alloys was performed by secondary neutral mass spectrometry (SNMS). It was found that the reverse sputtering method gave a much better depth resolution at the vicinity of the substrate. The reverse SNMS spectra showed that the deposition of Fe–Co–Ni alloys starts with the formation of an Fe-rich zone followed by an increase in Co concentration, then the nickel content increases and a steady-state alloy composition is achieved. At high current density, the initial depth pattern reproduces itself twice before the composition becomes stable. It was concluded that the varying depth profile is a consequence of the anomalous nature of the codeposition of the alloy components, the depletion of the electrolyte with respect to the metal salts, and the dependence of the intensity of the hydrogen evolution on the deposit surface composition.

Sulfur isotope characteristics of the Archean Cu-Zn Gossan Hill VHMS deposit, Western Australia

Gossan Hill is an Archean (∼3.0 Ga) Cu–Zn–magnetite-rich volcanic-hosted massive sulfide (VHMS) deposit in the Yilgarn Craton of Western Australia. Massive sulfide and magnetite occur within a layered succession of tuffaceous, felsic volcaniclastic rocks of the Golden Grove Formation. The Gossan Hill deposit consists of two stratigraphically separate ore zones that are stratabound and interconnected by sulfide veins. Thickly developed massive sulfide and stockwork zones in the north of the deposit are interpreted to represent a feeder zone. The deposit is broadly zoned from a Cu–Fe-rich lower ore zone, upwards through Cu–Zn to Zn–Ag–Au–Pb enrichment in the upper ore zone. New sulfur isotope studies at the Gossan Hill deposit indicate that the variation is wider than previously reported, with sulfide δ34S values varying between −1.6 and 7.8‰ with an average of 2.1 ± 1.4‰ (1σ error). Sulfur isotope values have a broad systematic stratigraphic increase of approximately 1.2‰ from the base to the top of the deposit. This variation in sulfur isotope values is significant in view of typical narrow ranges for Archean VHMS deposits. Copper-rich sulfides in the lower ore zone have a narrower range (δ34S values of −1.6 to 3.4‰, average ∼1.6 ± 0.9‰) than sulfides in the upper ore zone. The lower ore zone is interpreted to have formed from a relatively uniform reduced sulfur source dominated by leached igneous rock sulfur and minor magmatic sulfur. Towards the upper Zn-rich ore zone, an overall increase in δ34S values is accompanied by a wider range of δ34S values, with the greatest variation occurring in massive pyrite at the southern margin of the upper ore zone (−1.0 to 7.8‰). The higher average δ34S values (2.8 ± 2.1‰) and their wider range are explained by mixing of hydrothermal fluids containing leached igneous rock sulfur with Archean seawater (δ34S values of 2 to 3‰) near the paleoseafloor. The widest range of δ34S values at the southern margin of the deposit occurs away from the feeder zone and is attributed to greater seawater mixing away from the central upflow zone.

THE EFFECTS OF ECLOGITIC HETEROGENEITIES ON UPPER MANTLE MELTING

THE EFFECTS OF ECLOGITIC HETEROGENEITIES ON UPPER MANTLE MELTING