Toshitsugu Fujii

Melting phase relations of an anhydrous mid‐ocean ridge basalt from 3 to 20 GPa: Implications for the behavior of subducted oceanic crust in the mantle

High pressure melting experiments on an anhydrous abyssal tholeiite collected from the Mid-Atlantic Ridge have been conducted over the pressure interval 3 to 20 GPa to explore the fate of subducted oceanic crust in the mantle. The composition of the mid-ocean ridge basalt (MORB) is almost identical to the average basaltic layer of the oceanic lithosphere. The melting phase relations of the MORB are summarized as follows: (1) the liquidus temperature is about 1425°C at 3 GPa, and it rises almost linearly to above 1900°C at 10 GPa. The slope of the liquidus curve decreases slightly above 10 GPa. Nevertheless, it still has a positive slope with increasing pressure. At 20 GPa, the liquidus temperature is about 2200°C. (2) The liquidus phase changes from clinopyroxene to garnet above 3.5 GPa. (3) The solidus temperature rises almost linearly to 2100 °C at 20 GPa; consequently, the melting interval is slightly narrower at high pressures (<140°C at 5 GPa, <100°C at 20 GPa). (4) Silica minerals (coesite, stishovite) are stable near and below the solidus. (5) At shallow mantle conditions (2∼7 GPa), the liquidus temperature of the MORB is slightly lower than the solidus of the mantle material (peridotite, KLB-1); however, the solidus temperature of the MORB nearly equals or exceeds the liquidus temperature of the mantle material at about 16 GPa. The solidus temperature of the mantle material is expected to exceed the liquidus temperature of the MORB again at higher pressures where the liquidus phase in the peridotite system is majorite and/or modified spinel rather than olivine. In the ordinary thermal structure of the present mantle, subducted MORB is difficult to melt in the absence of volatiles because its dry solidus is much higher than the estimated mantle geotherm. However, in the ancient Earth, when the mantle was supposedly much hotter than today, complete or partial melting of subducting MORB was quite probable and this melting process could have played a significant role in controlling the movements of subducted slab and the evolution of the mantle and continental crust.

Melting relations of a magnesian abyssal tholeiite and the origin of MORBs

Melting relations of a glassy magnesian olivine tholeiite from the FAMOUS area have been studied within the pressure range 1 atm to 15 kbar. From 1 atm to 10 kbar, olivine is the liquidus phase, followed by plagioclase and Ca-rich clinopyroxene. Above 10 kbar, Ca-rich clinopyroxene appears on the liquidus, followed by orthopyroxene and spinel. Near 10 kbar, olivine, orthopyroxene, clinopyroxene, spinel and plagioclase crystallize within 10°C of the liquidus. This indicates that a liquid of this magnesian olivine tholeiite composition could coexist with mantle peridotite at about 10 kbar. This result is in agreement with the geochemistry of Ni; the Ni concentration of the studied sample corresponds to the theoretical concentration in a primary magma [14,15]. These data suggest that at least some magnesian mid-oceanic ridge basalts (MORBs) could be primary melts segregated from the mantle at depths near the transition zone between plagioclase lherzolite and spinel lherzolite (about 10 kbar). Based on this model, the residual mantle after extraction of MORBs should be lherzolite, not harzburgite. High-pressure (7–10 kbar) fractionation models involving olivine, plagioclase and clinopyroxene, which have been proposed by several workers (e.g. [36]) to explain the varieties of MORBs, were re-emphasized based on this melting study. The rare occurrence of clinopyroxene as a phenocryst phase in MORBs is explained by precipitation in a magma chamber at high pressure, or by dissolution of clinopyroxene formed earlier at high pressure.

Composition of aqueous fluid coexisting with mantle minerals at high pressure and its bearing on the differentiation of the Earth's mantle

In order to understand the role of aqueous fluid on the differentiation of the mantle, the compositions of aqueous fluids coexisting with mantle minerals were investigated in the system MgO-SiO2-H2O at pressures of 3 to 10 GPa and temperatures of 1000 to 1500°C with an MA8-type multianvil apparatus. Phase boundaries between the stability fields of forsterite + aqueous fluid, forsterite + enstatite + aqueous fluid, and enstatite + aqueous fluid were determined by varying the bulk composition at constant temperature and pressure. The composition of aqueous fluid coexisting with forsterite and enstatite can be defined by the intersection of these two phase boundaries. The solubility of silicate components in aqueous fluid coexisting with forsterite and enstatite increases with increasing pressure up to 8 GPa, from about 30 wt% at 3 GPa to about 70 wt% at 8 GPa. It becomes almost constant above 8 GPa. The Mg/Si weight ratio of these aqueous fluids is much higher than at low pressure (0.2 at 1.5 GPa) and almost constant (1.2) at pressures between 3 and 8 GPa. At 10 GPa, it becomes about 1.4. Aqueous fluid migrating upward through the mantle can therefore dissolve large amounts of silicates, leaving modified Mg/Si ratios of residual materials. It is suggested that the chemical stratification of Mg/Si in the Earth may have been formed as a result of aqueous fluid migration.

Preliminary report on the activity at Unzen Volcano (Japan), November 1990-November 1991: Dacite lava domes and pyroclastic flows

Abstract The eruption of Unzen Volcano commenced on 17 November 1990. Phreatic and phreatomagmatic eruptions occurred by early May 1991. No large-scale explosive eruptions preceded the extrusion of lava domes. Lava domes appeared in a summit crater on 20 May 1991, and they grew on the steep slope of Mt. Fugen at Unzen Volcano. Rockfalls from the margins of the domes frequently generated pyroclastic flows. Major pyroclastic flows occurred on 3 June, 8 June, and 15 September 1991. The 3 June pyroclastic flow killed forty-three persons. Many of the pyroclastic flows seem to have resulted from the simple rockfalls, except one flow on 8 June, which was accompanied by an explosion from the crater. Many of the rockfalls that generated pyroclastic flows were witnessed. As of November 1991. Unzen Volcano was still active with a nearly constant magma-supply rate of about 0.3 × 10 6 m 3 /d. The total magma output exceeded 45 × 10 6 m 3 by the beginning of November 1991. The volume of the lava domes is more than 23 × 10 6 m 3 .

The 15 September 1991 pyroclastic flows at Unzen Volcano (Japan) : a flow model for associated ash-cloud surges

Large-scale collapse of a dacite dome in the late afternoon of 15 September 1991 generated a series of pyroclastic-flow events at Unzen Volcano. Pyroclastic flows with a volume of 1×106 m3 (as DRE) descended the northeastern slope of the volcano, changing their courses to the southeast due to topographic control. After they exited a narrow gorge, an ash-cloud surge rushed straight ahead, detaching the main body of the flow that turned and followed the topographic lows to the east. The surge swept the Kita-Kamikoba area, which had been devastated by the previous pyroclastic-flow events, and transported a car as far as 120 m. Following detachment, the surge lost its force after it moved several hundred meters, but maintained a high temperature. The deposits consist of a bottom layer of better-sorted ash (unit 1), a thick layer of block and ash (unit 2), and a thin top layer of fall-out ash (unit 3). Unit 2 overlies unit 1 with an erosional contact. The upper part of unit 2 grades into better-sorted ash. At distal block-and-ash flow deposits, the bottom part of unit 2 also consists of better-sorted ash, and the contact with the unit 1 deposits becomes ambiguous. Video footage of cascading pyroclastic flows during the 1991–1995 eruption, traveling over surfaces without any topographic barriers, revealed that lobes of ash cloud protruded intermittently from the moving head and sides, and that these lobes surged ahead on the ground surface. This fact, together with the inspection by helicopter shortly after the events, suggests that the protruded lobes consisted of better-sorted ash, and resulted in the deposits of unit 1. The highest ash-cloud plume at the Oshigadani valley exit, and the thickest deposition of fall-out ash over Kita-Kamikoba and Ohnokoba, indicate that abundant ash was also produced when the flow passed through a narrow gorge. In the model presented here, the ash clouds from the pyroclastic flows were composed of a basal turbulent current of high concentration (main body), an overriding and intermediate fluidization zone, and an overlying dilute cloud. Release of pressurized gas in lava block pores, due to collisions among blocks and the resulting upward current, caused a zone of fluidization just above the main body. The mixture of gas and ash sorted in the fluidization zone moved ahead and to the side of the main body as a gravitational current, where the ash was deposited as surge deposits. The main body, which had high internal friction and shear near its base, then overran the surge deposits, partially eroding them. When the upward current of gas (fluidization) waned, better-sorted ash suspended in the fluidization zone was deposited on block-and-ash deposits. In the distal places of block-and-ash deposits, unit 2 probably was deposited in non-turbulent fashion without any erosion of the underlying layer (unit 1).

Groundmass pargasite in the 1991–1995 dacite of Unzen volcano: phase stability experiments and volcanological implications

Abstract Pargasite commonly occurs in the dacitic groundmass of the 1991–1995 eruption products of Unzen volcano. We described the occurrence and chemical compositions of amphibole in the dacite, and also carried out melting experiments to determine the low-pressure stability limit of amphibole in the dacite. The 1991–1995 ejecta of the Unzen volcano show petrographic evidence of magma mixing, such as reverse compositional zoning of plagioclase and amphibole phenocrysts, and we used a groundmass separate as a starting material for the experiments. Reversed experiments show that the maximum temperature for the crystallization of amphibole is 930°C at 196 MPa, 900°C at 98 MPa, and 820°C at 49 MPa. Compared with the experimental results on the Mount St. Helens dacite, present experiments on the Unzen dacitic groundmass show that amphibole is stable to pressures ca. 50 MPa lower at 850°C. Available Fe–Ti oxide thermometry indicates the crystallization temperature of the groundmass of the Unzen dacite to be 880±30°C, suggesting that the groundmass pargasite crystallized at >70 MPa, corresponding to a depth of more than 3 km in the conduit. The chlorine content of the groundmass pargasite is much lower than that of phenocrystic magnesiohornblende in the 1991–1995 dacite of Unzen volcano, indicating that vesiculation/degassing of magma took place before the crystallization of the groundmass pargasite. The present study shows that the magma was water oversaturated and that the degassing of magma along with magma mixing caused crystallization of the groundmass amphibole at depths of more than 3 km in the conduit.

Connectivity of aqueous fluid in the Earth's upper mantle

The geometrical distribution of the aqueous fluid in textural equilibrium with forsterite has been investigated by measurements of the dihedral angle, θ, at pressures of 3 and 5 GPa and at 1000°C. The measured θ at 3 and 5 GPa are 48° and 40° to 42°, respectively. Since a value of θ smaller than 60° indicates that the aqueous fluid can form an interconnected network along the three-grain and through the four-grain junctions, the existing aqueous fluid in the mantle can migrate by percolation even at the small volume fractions in the pressure range. Combining these results with previously published results at lower pressure, it is suggested that the interconnected network of aqueous fluid is stable only at higher pressure than 2 GPa for the commonly accepted water content of the upper mantle. The present results show that the large scale transport of aqueous fluid is possible above 2 GPa, and that the physical properties of the upper mantle may change drastically at that pressure.

The melting relation of the system, iron and carbon at high pressure and its bearing on the early stage of the Earth

The melting relation of iron-carbon system is studied to see how iron takes in or disgorges carbon at high pressure (up to 12GPa). The understanding on these points is closely related to the problem of the reservoir of carbon in the earths interior. The results are: (1) Iron-carbon system shows eutectic melting till at least 12GPa. (2) The eutectic composition is about 3 or 4 wt% of carbon, which does not vary very much with pressure. (3) The eutectic temperature slightly goes up as pressure increases at a rate of 7°C/GPa. This gradient is fairly lower than that of the melting temperature of silicates. Based on these results and other facts, the following scenario is inferred on the core formation during the early stage of the earth. Because carbon melts into iron forming eutectic system at low temperature, carbon within the accreted chondritic materials might be absorbed into iron melt near the surface of the magma ocean. The observation that the melting temperature of silicate goes up more rapidly as pressure increases than the eutectic temperature of iron-carbon system, indicates that the temperature within the magma ocean is maintained higher than the melting temperature of iron-carbon system. Therefore, the carbon-bearing iron melt may sink into the deep interior of the earth without solidifying and disgorging carbon. After all, it is strongly suggested that carbon may settle with liquid iron forming core.

The October 1983 eruption of Miyakejima volcano

Abstract Miyakejima volcano erupted at 1515 hours on October 3, 1985. During the first 2 hours, a 4.5-km-long flank fissure produced the curtain of fire at higher vents from 500 m to 100 m in altitude. Porous scoria drifted eastward to form a thin blanket across the island. Most of the erupted material from these vents formed aa lava flows which devastated the forest. One flow reached the sea shore and another engulfed the largest community of the island within several hours after the start of the eruption. About 430 houses were burned and buried but the 1500 residents escaped without casualties. An effort was made to control the advancing lava flow by spraying it with 4700 tons of sea water. However, by the time the watering operation started, the lava flow apparently had stopped moving so that the real effect of watering could not be confirmed. The fissure extended northward and southward at a rate of 40 m per minute during the first 30 minutes. The northward extension stopped at 505 m altitude but the southward extension continued at a rate of 26 m per minute to the 40 m deep sea floor off the southern coast of the island. Near the southern end of the fissure, violent phreatomagmatic explosions produced large craters and a pyroclastic ring on the coast. An 8000-m-high eruption column produced heavy scoria fall in the southeastern part of the island. The scoria was much denser than that erupted by the northern vents and caused great damage to farm lands and houses and broke car windows. The largest earthquake of M = 6.2 occurred at 2233 hours which shook off loose debris and gave rise to a revival of fountaining eruptions in the southernmost vents. The eruption ended before 0600 hours of the next day, October 4. The products are tholeiitic basalts (SiO 2 52–54%) containing a few percent of plagioclase phenocrysts. Much less abundant augite and titanomagnetite and very rare olivine and hypersthene phenocrysts are also found. These basalts fall on the general differentiation trend of Miyakejima volcano and are in a moderately advanced stage of differentiation. Highly vesiculated glassy xenoliths (SiO 2 56–68%) are found included in the juvenile ejecta. Chemistry of the xenoliths indicates that most of them are derived from mechanical mixtures (probably volcaniclastic sediments) of ancient ejecta of Miyakejima of various compositions.

Ascending subducted oceanic crust entrained within mantle plumes

Upward migration of subducted oceanic crust from deep in the upper mantle is discussed based on high-pressure experimental data. Numerical calculation reveals that a subducted oceanic crust, which is negatively buoyant in the present upper mantle, can ascend from the transition zone and be observed as a discrete magma source at the surface only when it has been broken into fragments comparable in dimension with its thickness and involved in a solid diapir of heated peridotite. For example, a solid diapir of heated peridotite with a temperature excess of 200°C can contain up to 15∼20 vol% of fragments of oceanic crust when it passes through the upper mantle, and it may cause extensive basaltic volcanism at the surface.