MR31B-0362
Synthesis and Crystal Structure of Ferric-Rich MgSiO3-Perovskite
Previous Mössbauer studies have shown that up to 50% of Fe in Mg silicate perovskite can be Fe3+ (McCammon 1997, Jackson et al. 2005). In order to measure the solubility of Fe3+ and the effect of Fe3+ on the crystal structure of Mg silicate in the lower mantle, we conducted X-ray diffraction and Mössbauer spectroscopy measurements in the laser-heated diamond-anvil cell using an Ar medium at sectors GSECARS and 3 at APS, and at beamline 12.2.2 at ALS. We observed the formation of orthorhombic (Pbnm) perovskite from MgSiO3 enstatite and Fe2O3 hematite (25 and 50%) crystalline mixtures at 45-50 GPa after laser heating to 2000 K for 30 min. We also confirmed the synthesis of orthorhombic perovskite from glass starting materials with compositions of 90% MgSiO3 + 10% Fe2O3 and 97.5% MgSiO3 + 2.5% Fe2O3 at 50 GPa. This indicates that Mg silicate perovskite may have a large storage capacity for Fe3+ in the lower mantle. Our X-ray diffraction measurements show that 10% Fe2O3 expands the unit-cell volume of perovskite by as much as 2% at low pressure compared to pure Mg-endmember, but is more compressible than Mg-endmember perovskite. We also found that 10% Fe2O3 expands the b-axis by 0.5-0.7%, whereas the other axes are in agreement with those of Mg-endmember within experimental uncertainties to at least 65 GPa, indicating that Fe3+ expands the structure anisotropically.
MR31B-0363
High-temperature compression of ferropericlase and effect of temperature on iron spin transition
High-temperature compression experiments of ferropericlase (Fp) with a composition of (Mg0.81,Fe0.19)O were operated in an externally-heated diamond anvil cell from 19 to 85 GPa at a constant temperature of 859-869 K at the beamline BL10XU, SPring-8. Room-temperature experiments with laser-annealing technique were also carried out on the same material. Anomalous volume reductions which cannot be explained by the normal compression behavior were observed at 60-82 GPa and 58-64 GPa at high temperature and room temperature, respectively. These volume reductions are likely to be related to the spin transition of ferrous iron in Fp. The spin transition pressure interval expands with increasing temperature. At the base of the mantle, in subducted cold slabs, Fp may have full low-spin iron while in hot upwelling plumes, Fp has relatively high-spin rich iron. This difference in the spin state will result in the density contrast at the core-mantle boundary. We will discuss the temperature dependence of the nature of the spin transition.
MR31B-0364
Spin transition in ferrous iron in MgSiO3 perovskite under pressure
We present a density functional study of the pressure-induced spin transition in ferrous iron in MgSiO3 perovskite. We address the influence of iron concentration and configuration (structural and magnetic), as well as technical issues such as the nature of the exchange correlation (XC) functional (CA-LDA versus PBE-GGA) on the spin transition pressure. Supercells containing up to 160 atoms were adopted to tackle these issues. We show that there are preferred configurations for high-spin and low-spin iron and that the spin transition pressure depends strongly on iron concentration and XC functionals. We also address changes of atomic structure around Fe atoms and electronic structure including the blue shift accompanying the spin transition. Research supported by NSF/EAR 013533, 0230319, and NSF/ITR 0428774 (VLab). Computations were performed at the Minnesota Supercomputing Institute and Indiana Universityfs BigRed system.
MR31B-0365
Fe-Mg partitioning between perovskite, post-perovskite, and ferropericlase at the lowermost mantle
The lower mantle consists mainly of iron-bearing magnesium silicate perovskite and ferropericlase. The Fe-Mg partition coefficient between these minerals is important to understand the chemical and physical properties of the mantle. The lowermost 200 km of the mantle is called D" layer and it is considered to be a thermal and chemical boundary layer between the silicate mantle and outer core. Recent high pressure studies [e.g., Murakami et al., 2004] revealed that (Mg, Fe)SiO3 perovskite transformed to post-perovskite phase at the pressure and temperature conditions of D" layer. In this study, high pressure and high temperature partitioning experiments were performed up to 140 GPa at 2000 K using a laser heated diamond anvil cell (LHDAC). Powdered or single crystalline Al-free San Carlos olivine (Mg0.88, Fe0.12)2SiO4 was used as a starting material in order to avoid complicated compositional effects. The starting material was embedded in sodium chloride which is pressure medium. Pressures were determined by both the ruby fluorescence method [Mao et al., 1978] and the Raman shift of the first-order Raman spectra of diamond anvil [Akahama and Kawamura, 2004]. Temperatures were measured by spectroradiometric method. The recovered samples were analyzed using the technique of combination of FIB and ATEM (JEOL JEM-3000F (FEG TEM-STEM)). The result shows that post- perovskite phase exhibits very small iron content, Fe# = 0.01 at 140 GPa and 2000 K. Therefore, the partition coefficient was K = 0.03, which indicates that iron prefers ferropericlase strongly rather than post-perovskite phase, which is consistent with the prediction of the ab initio calculation [Iitaka et al., 2004], and the high-spin/low- spin transition arguments of ferropericlase [Badro et al., 2003]. Kobayashi et al. (2005) showed higher partitioning coefficient of K= 0.30 in spite of the lower temperature condition of 1600 K compared to the present experiment. They reported that significant iron depletion occurred in the sample heated at the high pressure condition, i.e., bulk Fe# (FeO/(MgO+FeO) in molar ratio) reduced to 0.09 from 0.12. On the other hand, the bulk iron content in the present experiment conducted at 140 GPa and 2000 K does not show a significant iron depletion. Therefore, the significant difference in the partition coefficients between the present result and that by Kobayashi et al may be caused by the compositional dependency of the partitioning behavior between post- perovskite and ferropericlase. Further study is necessary to clarify the compositional effect together with the effect of the spin transition on the partitioning behavior.
MR31B-0366
Magnetic and electronic thermodynamics of fayalite
Unique among the major elements, iron produces magnetic and electronic contributions to the thermodynamic properties or minerals, which influence phase stability and physical properties, including elasticity. To better understand these contributions, we have focused on a material that is widely studied, that may exhibit many features in common with high pressure iron-bearing phases, and which illustrates the richness and geophysical significance of magnetic and electronic excitations. We have used inelastic neutron-scattering measurements to study the magnetic excitations in the antiferromagnetic and paramagnetic phases of polycrystalline fayalite, Fe2SiO4. Sharp, nondispersing excitations are found in the ordered state, at 3.3, 5.4, 5.9, and 11.4 meV, and are interpreted as arising from the spin-orbit manifold of the high-spin Fe2+ ions. These excitations are increasingly damped with increasing temperature, merging into a quasielastic continuum near the 65 K Neel temperature, although their energy does not vary with temperature. We have calculated the contribution of the heat capacity arising from these magnetic excitations and found that it compares favorably with the magnetic heat capacity deduced experimentally. Our analysis indicates that the M1 and M2 sites behave distinctly. The M1 site behaves quasi-locally and appears in the heat capacity as a Schottky anomaly that explains the shoulder in the heat capacity curve near 20 K, while the M2 site contributes predominantly to the critical lambda anomaly. The behavior of fayalite illuminates the nature of magnetic states in several related minerals, including others that also show shoulders and lambda anomalies in the heat capacity (tephroite), those that show only lambda anomalies (cobalt olivine and liebenbergite), and those that show only non-lambda anomalies (bronzite, anthophyllite, and almandine). We find no evidence to support the recent claim that some transition metal silicate and germanate olivines exhibit strong geometric frustration.
MR31B-0367
Pressure-Induced Phase Transitions In Gadolinium Iron Borate
An understanding of spin crossover (SC) dynamics is relevant to understanding of a role or participation of SC in natural systems including lower Mantle minerals, heme proteins as well as from fundamental science of view. For example, pressure-induced electronic spin transitions of Fe2+ and Fe3+ iron occur in magnesiowustite, silicate perovskite and post-perovskite which are abundant minerals in the Earth's lower mantle [1-3]. Such a SC phenomenon has recently been observed in a number of magnetic minerals FeBO3 [4, 5], BiFeO3 [6], Fe2O3 [7], and Y3Fe5O12 [8], (La, Pr)FeO3 [9, 10]. In those cases, iron ions are in the trivalent state Fe3+ and the high-spin-low-spin (HS–LS) crossover is manifested as the collapse of the local magnetic moment and as the transition of the antiferromagnet to a paramagnetic state. For example, in FeBO3 at low temperatures a spin-crossover and some magnetic transitions with two triple points were found [4, 5]. Gadolinium iron borate, GdFe3(BO3)4 is also a system with SEC and recently, we have reported on phase transitions induced by high pressures in this material [11, 12]. We studied the structural and magnetic behavior of GdFe573(BO3)4 at high pressures and temperatures using a diamond anvil cell and a Synchrotron Mossbauer Spectroscopy technique. The hyperfine parameters and results obtained from the experiments are discussed. Based on our experimental data and theoretical calculation a tentative magnetic P–T phase diagram and an equation of states of GdFe573(BO3)4 are proposed. Important features of the phase diagram are a spin crossover, insulator-semiconductor transition and possible presence of two triple points where magnetic and paramagnetic phases of the high-spin and low-spin states coexist. 1. J. Badro, J.-P. Rueff, G. Vankó, et al., Science 305, 383 (2004). 2. J. M. Jackson, W. Sturhahn, G. Shen, et al., American Mineralogist 90, 199 (2005). 3. J.Li, V.V. Struzhkin, H.-K. Mao, et al., PNAS 101, 14027 (2004). 4. I.A. Troyan, A. G. Gavrilyuk, et al., JETP Lett. 74, 24 (2001). 5. A.G. Gavriliuk, I.A. Trojan. et al., JETP 100, 688 (2005). 6. A.G. Gavriliuk, V.V. Struzhkin, et al., JETP Lett. 82, 224 (2005). 7. M.P. Pasternak, G.Kh. Rozenberg, et al., Phys. Rev. Lett. 82, 4663 (1999). 8. I.S. Lyubutin, A.G. Gavrilyuk, I. A. Troyan, et al., JETP Lett. 82, 702 (2005). 9. G.R. Hearne, M.P. Pasternak, et al., Phys. Rev. B 51, 11 495 (1995). 10. W.M. Xu, O. Naaman, G.Kh. Rozenberg, et al., Phys. Rev. B 64, 094411 (2001). 11. A.G. Gavriliuk, SA. Kharlamova, et al., JETP Lett. 80, 426 (2004). 12. A.G. Gavriliuk, S.A. Kharlamova, et al. J. Phys.: Condens. Matt. 17. 1-6 (2005)
MR31B-0368
Effects of Si on the crystal structure and elastic property of Fe at Earth's inner core pressures
Earth's inner core is generally thought to consist of Fe-Ni alloy. Some amount of light elements are also expected to be dissolved. Recent high-pressure experiments and theoretical simulations have suggested Si to be a highly possible candidate of impurity elements in the inner core. In this study, we investigated the effects of Si on the crystal structure and elastic property of iron at the Earth's inner core pressures by first-principle calculations. Calculations show that Si changes the elasticity of the hcp and fcc phase of iron, in particular shear components, while it marginally affects bcc phase. As a result, VS of hcp iron unexpectedly decreases even by incorporating light Si. Velocity-density relations of Fe-rich alloys are found very different from those anticipated from B2-type FeSi. Research supported by the Ehime Univ Project Fund.
MR31B-0369
Melting curve of iron-silicon alloy to the core-mantle boundary pressure and the thermal structure of the Earthfs core
Earth's core consists of iron-Ni alloys and 10 wt.% of light elements, such as Si, S, O, C, H. It is composed of the solid inner core and the liquid outer core. Melting relation of the iron alloy is essential to estimate the thermal structure and temperature of the core, since the temperature of the inner core-outer core boundary (ICB) corresponds to the melting temperature of iron alloy at ICB pressure. Furthermore, the temperature of the outer core-mantle boundary (CMB) has to be higher than the melting temperature of iron alloy at CMB pressure. Therefore, the melting temperature of iron alloy at high pressures is significant for estimating the thermal structure of the Earth. In this study, we determined the melting temperature of Fe-17 wt.% Si alloy up to 119 GPa based on change of laser heating efficiency and the texture of the recovered samples using the sided laser heated diamond anvil cell. The measured melting curve of Fe-17 wt.% Si is slightly lower than that of pure Fe and the melting temperature was 3300 K at 120 GPa (CMB pressure). The melting temperature of the alloy is expressed as (P- P0)/a=(Tm/T0)c-1 by Simonfs equation , i.e., P0= 0 GPa, Tm= 1500 K, a= 3.54+1.1, and c= 4.53+0.4. In order to clarify the solid phase of Fe-17 wt.% Si, in-situ X-ray diffraction study was also carried out at the BL13A beamline in PF of KEK. The high-pressure, and high-temperature behavior of Fe-17 wt.% Si alloy was investigated up to 96 GPa and 2000 K. Fe-17 wt.% Si has a bcc structure and there was no phase transition in the range of the present experimental pressures and temperatures. By extrapolation of the melting temperature using Simonfs equation, the melting temperature of Fe-17 wt.% Si alloy is estimated to be 3300 K at the core-mantle boundary, 135 GPa, and 4000 K at the inner-core boundary around 330 GPa, which may provide the lower bound of the temperatures at CMB and ICB if the light element in the core is silicon.
MR31B-0370
Effect of Hydrogen and Carbon on the Melting Temperature of the Core
The temperature of the Earth's outer core has been discussed based on the melting temperature of Fe- O-S alloys (e.g., Boehler, 1996). Although hydrogen and carbon are the possible candidates of the core component, their effects on the melting temperature of iron at high-pressures are unclear. Using a Kawai-type multi-anvil apparatus at SPring-8 synchrotron, we carried out a series of melting experiments on FeH and Fe3C up to 20 and 28 GPa, respectively. In the experiments on FeH, Fe sponge mixed with MgO was packed into a NaCl container with a hydrogen source, LiAlH4 (e.g., Fukai et al., 1989). During heating under high-pressures, hydrogenation of iron was observed by volume change. The phase boundary between ε'-phase (low-temperature phase) and γ-phase (high-temperature phase) of iron-hydride was determined using both cooling and heating experiments. Hydrogen concentrations in the γ-FeHx and ε'-FeHx were calculated based on the excess volume data from that of pure iron. It is found that γ-FeHx and ε'-FeHx synthesized in our experiments at pressures between 10 and 20 GPa are nearly stoichiometric FeH. Melting temperature of the γ-FeH was determined by the abrupt change in the X-ray diffraction patterns (crystalline to amorphous). The melting temperatures were determined to be 1473, 1473, 1493, 1573 and 1593 K at 10, 11.5, 15, 18 and 20 GPa, respectively. In the experiments using Fe3C, the synthesized Fe3C powder was encapsulated in a MgO container. In the diffraction sequences during heating, the peaks of Fe3C disappeared, and the new peaks identified as those of Fe7C3 were observed with halo caused by liquid. Finally, the Fe7C3 peaks disappeared, and only the halo pattern was observed. Based on these observations, the incongruent melting of Fe3C to Fe7C3 and liquid is estimated to occur at 1823 and 1923 K at 19.7 and 27.0 GPa, respectively. The liquidus temperatures of the Fe3C composition are found to be at 2098 and 2198 K at 19.5 and 26.8 GPa, respectively. The melting temperatures of Fe3C determined by our experiments are >700 K lower than that of the previous estimation based on thermodynamic calculation (Wood, 1993). Our experimental results show a possibility that the hydrogen and carbon lower the melting temperature of iron (outer core) dramatically. The melting temperatures of γ-FeH and Fe3C at 20 GPa are already 500 K lower than that of pure iron estimated by Anderson and Isaak (2000). Extrapolating our experimental melting curves for FeH and Fe3C to core pressures using Lindemann's melting law, we obtained the melting temperatures to be ~2600 and ~2900 K at the core-mantle boundary (CMB), respectively. In the presence of both hydrogen and carbon, melting temperature of the Earth's outer core could be >1500 K lower than that of the previous estimates, implying that the temperature gap at CMB could be much smaller than the current estimates.
MR31B-0371
Partitioning of Potassium Between Fe-S and K-rich Silicate at High Temperature
Partitioning of potassium between liquid iron and silicate melt at high pressure and temperature is important to estimate the amount of potassium in the Earth's core. In the previous works, the partitioning experiments were performed below 26 GPa using a large volume press (e.g. Ito et al., 1993; Gessmann et al., 2002; Murthy et al., 2003), except for the work by Hirao et al. (2006) using a laser heated diamond anvil cell (LHDAC). The aim of this work is to clarify the effects of pressure, temperature and sulfur content on the partitioning coefficients of potassium (DK) in a wide pressure and temperature range using LHDAC. We performed partitioning experiments at 50 GPa and the temperature range of 2500-3000 K using a LHDAC. The starting material was foil of Fe79S21, which was sandwiched by a natural Aduralia (KAlSi3O8). Powder X-ray diffraction experiments were carried out at the ambient pressure to identify the reaction phases at BL13A and BL18C Photon Factory, KEK. Chemical analysis was performed using EPMA and ATEM (Institute for Materials Research, Tohoku Univ.). DK increases from 0.011 to 0.136 with increasing temperature (2500-3000 K) at 50 GPa. This trend is consistent with previous studies (e.g. Gessmann and Wood, 2002; Murthy et al., 2003). Our result indicates that DK decreases with increasing pressure. DK at 50 GPa and 3000 K in the sulfur- bearing system is larger than that reported by Ito et al.(1993) at 26 GPa and 2900 K in the sulfur-free system. This results show that sulfur may increase the solubility of potassium into metallic iron melt. DK of 0.005-0.12 at 135 GPa and 3000-4000 K is estimated by extrapolation of the DK to the condition of the core-mantle boundary (CMB) using the dependence of temperature and pressure observed in this study. This value suggests that the total potassium content in the core is about 1.2-29 ppm assuming that a primitive mantle concentration of potassium is 240 ppm (McDounough and Sun, 1995). It can produce only the radiogenic heat of 0.005-0.12 TW in the core, which is 1.2-6.0 % of the total current heat flux from the core.
MR31B-0372
Magnetic transition and sound velocities of Fe3C at high pressure
Carbon is a candidate light element in the Earth's core. Fe3C (cementite) has the lowest carbon content among all known Fe-C compounds. Under ambient conditions, Fe3C is ferromagnetic. A pressure-induced magnetic transition from a magnetic phase to non-magnetic phase has been found in previous studies; however, there is a controversy concerning the transition pressure. In this study, we carried out synchrotron Mossbauer spectroscopy (SMS) and nuclear resonant inelastic x-ray scattering (NRIXS) studies on Fe3C up to 52 GPa at room temperature at Sector 3 of the Advanced Photon Source (APS), Argonne National Laboratory. The starting material was synthesized from 57Fe-enriched iron powder and graphite powder. X-ray diffraction measurements revealed that the run product contains a non-negligible amount of iron. Our 1 bar SMS spectrum is well fitted by assuming one iron site with a magnetic hyperfine field of 20 T, consistent with that of Fe3C. Above 9.3 GPa our Mossbauer spectra revealed the disappearance of the 20 T site, indicating that Fe3C has lost its magnetism around 9.3 GPa or below. This is consistent with the magnetic transition at pressure of around 10 GPa reported in a Fe K-edge x- ray circular dichroism study (Duman et al., 2005), and is consistent with x-ray emission spectroscopy data (Lin et al., 2004), showing a significantly reduced magnetic moment at 12 GPa. A magnetic to paramagnetic phase transition is also observed at 483 K under room pressure (Wood et al., 2004). The observed magnetic transition under high pressure and high temperature indicates that the ferromagnetic phase is not stable under the Earth's core condition, and that the non-magnetic phase of Fe3C is more applicable to the Earth's core. We have also derived the Debye velocity of the sample from parabolic fitting to the low-energy range of the nuclear resonance inelastic x-ray scattering spectra at ambient condition. Combined with equation of state of Fe3C (Scott et al., 2001, Li et al., 2002), the compressional and shear wave velocities of Fe3C at ambient condition have also been derived.
MR31B-0373
High-Pressure Stability and Equations of State in the Fe-P System: Implications for Iron Meteorites and Planetary Cores
The density, elasticity and structural stability of iron and iron alloys are areas of long-standing interest in planetary science due to the known ubiquity of iron-rich cores in planetary bodies, and the observation that Earth's core is ~10% less dense than pure iron. Furthermore, iron-nickel meteorites often contain inclusions of iron carbide, sulfide and phosphide minerals, which has led to considerable interest in C, S and P (as well as many others) as potential "light" constituents of planetary cores. As part of our ongoing investigation of Fe3P, known as schreibersite when found in iron meteorites, we have measured the room-temperature bulk modulus (K0T) of Fe2P to 8 GPa using synchrotron X-ray diffraction coupled with Diamond Anvil Cells. Our motivation is to investigate systematics in the Fe-P system due to our previous observation that Fe3P is not stable above ~15 GPa (Scott et al., 2007). We used the same mixed-phase starting material (both Fe3P and Fe2P) as in our previous Fe3P Equation of State (EoS) study, but examined the bulk sample rather than a selection of pure Fe3P grains. Accordingly, our data contain information from both phases; we used diffraction peaks from Fe3P and its EoS to calibrate measurements on Fe2P relative to Fe3P. We imposed a constant \frac{c}{a} ratio on this hexagonal structure based on ambient-pressure measurements to reduce fitting parameters, which is consistent with observations of tetragonal Fe3P over this pressure range. A second order fit (i.e., dK/dP fixed at 4) to the Birch Murnaghan EoS produces a K0T of 140 ± 4 GPa. This value is 12% lower than that of Fe3P, and will be used to assess elasticity systematics in iron phosphide phases and potentially other AxY systems as well.
MR31B-0374
Sound velocities of iron sulfide at high pressure and high temperature
Iron sulfide, FeS, is widely believed to be a possible constituent of the cores of terrestrial planets. Stoichiometric FeS has been investigated extensively from the viewpoints of crystal structures, phase diagram, equation of state, and electronic properties at high pressure and/or high temperature using diamond anvil cell and large volume press apparatus. Though static compression studies have explored high pressure behavior, they provided only indirect determination of the bulk modulus K0 and its pressure derivatives dK/dP, and no information on the shear properties. In this study the elastic compressional (P) and shear (S) wave velocities of FeS have been measured, for the first time, across several successive phase transitions (FeS I, FeS II, FeS III, and FeS IV) along various P-T paths up to 11 GPa and 723 K, using the combined state of the art ultrasonic interferometry, X-ray diffraction and X-ray radiography method. Previous studies indicate that the Fe in FeS III is in a low-spin state, and structural phase transition of FeS II to FeS III involves the electronic transition of Fe from a high-spin to a low-spin state. Our measurements allow us to investigate the possible velocity changes associated with such an electronic transition. The elastic bulk and shear moduli for other polymorphs of FeS at high pressure and high temperature will also be presented.
MR31B-0375
Percolation experiments of FeS melts in partially molten peridotites
The series of percolation experiments have been done on partially molten fertile garnet peridotite xenolith by using the centrifuging piston-cylinder press. Powders with 100-200 μm and 20-30 μm grain size were mixed with 5-30 vol% Fe-FeS eutectic composition. The deformed high-T garnet peridotite with Mg# ~ 0.90 is composed from 60 vol% Ol, 15 vol% Opx, 6 vol% Cpx and 19 vol% Gar. The centrifuge experiments revealed a negligible percolation of Fe70S30 melts through the partially molten peridotite matrix. At 1150° C and 4.5 vol% of FeS the separation is 0.7 vol%/mm. Only at 1260° C and starting with 5 vol% of Fe70S30 the vertical gradient achieved 1-2 vol%/mm, and in samples wiith initial 15 vol% FeS the vertical separation achieved 2-2.5 vol%/mm after 10 h of centrifuging at 500 g. At 1280° C and 25 vol% of melting the initial 30 vol% of FeS have been removed during 0.5 h at 600 g. The partial melting of peridotite contributes in the increase of Fe70S30 droplet size, in agreement with Yoshino and Watson (2005) and in the increase of the effective velocity of FeS percolation. The permeabilities of partially molten peridotite with FeS using the viscosity of melt 0.1 Pa s are ~ 10-16 \div 10-17 m2, which is one order of magnitude lower than estimations (Roberts et al., 2007) from static experiments. The slow percolation of FeS in the absence of high degree of partial melting of silicates ( <15 vol%) preclude the scenario of the metallic core formation before the partial melting of silicate mantle in planetary bodies is started. The segregation velocities, scaled to 1 g are 10-5 mm/h. At higher degree of partial melting (>15 vol%) the interconnected FeS phase coagulates in large spherical clusters. At further partial melting ( ~30 vol%) the large clusters of FeS disintegrated into smaller droplets and moved in silicate melt channels downwards whereas the silicate melt migrates upward. The rate of FeS percolation according to this mechanism is 5× 10-3 mm/h. After flow of a large amount of FeS melt the system of silicate melt channels remains aligned vertically and the permeability of partially molten matrix is anisotropic. To remove the rest of FeS (the last 3-5 vol%) from the silicate matrix is not possible even applying the large centrifugal accelerations. The grain size reduction due to the phase transition Gar into Sp does not block the segregation of FeS, the smaller crystals are easier to shift and to rotate under the force of the sinking FeS droplets.