Hosted Melt Inclusions Research Articles

Crystal–Melt Relationships and the Record of Deep Mixing and Crystallization in the ad 1783 Laki Eruption, Iceland

Concurrent mixing and crystallization of magmas has been identi- fied in the plumbing of Icelandic magmatic systems from investiga- tion of their erupted products. Thermobarometric calculations constrain the depth of these processes to the mid- to lower crust. It is currently unknown whether similar processes also occur in the magmas that feed the large fissure eruptions characteristic of the Eastern Volcanic Zone (EVZ). We therefore present the results of an investigation of crystal^melt relationships in the AD 1783 Laki (Skaftar Fires) eruption. The trace element composition of olivine- hosted melt inclusions indicates that concurrent mixing and crystal- lization of variable mantle melts has occurred in the deep parts of the Laki plumbing system. Magmatic evolution is also recorded in crystal zonation patterns. Zoned plagioclase macrocrysts are made up of three distinct domains: high-anorthite cores, oscillatory zoned mantles and low-anorthite rims. Published mineral^melt equilib- rium partition coefficients indicate that macrocryst rims are in equilibrium with the carrier liquid erupted at the surface. High- anorthite cores are more primitive than any other crystal or melt inclusion composition in the magma and are never in equilibrium with melts generated by fractional crystallization models that assume a single liquid line of descent related to the carrier liquid. High-anorthite crystals may have grown from depleted, high-Ca/Na melts of the shallow mantle, which may have been mixed into melts parental to the carrier liquid early in magmatic evolution.The crystal size distribution (CSD) of plagioclase macrocrysts suggests that plagioclase cores represent an assimilated or accumulated crystal population. Cores have acted as nuclei for the growth of oscillatory zoned plagioclase mantles. It is possible to relate the range of plagio- clase mantle compositions to the carrier liquid using fractional crystallization models involving eutectic co-crystallization with clinopyroxene and olivine macrocrysts. Negative correlation of titan- ium and anorthite content in plagioclase mantles confirms that oscil- latory zoning is driven by variations in melt composition, most probably the result of repeated recharge and fractionation events. Melt barometry indicates that the melt last equilibrated with olivine, plagioclase and clinopyroxene in the shallow crust at 1^2 kbar, whereas clinopyroxene^melt barometry suggests that the bulk of macrocryst growth occurred within the mid-crust at 2^5·4 kbar. Much of the macrocryst content of the Laki magma occurs as glomer- ocrysts, of which only rims in contact with the groundmass are in equilibrium with the carrier liquid.This indicates that the formation of glomerocrysts occurred before rim growth, during deposition of crystal mushes in the mid-crust, which subsequently disaggregated upon transport to the shallow crust.

Volatiles contents, degassing and crystallisation of intermediate magmas at Volcan de Colima, Mexico, inferred from melt inclusions

In volatile-saturated magmas, degassing and crystallisation are interrelated processes which influence the eruption style. Melt inclusions provide critical infor- mation on volatile and melt evolution, but this information can be compromised significantly by post-entrapment modification of the inclusions. We assess the reliability and significance of pyroxene-hosted melt inclusion analyses to document the volatile contents (particularly H2O) and evolution of intermediate arc magmas at Volcan de Coli- ma, Mexico. The melt inclusions have maximal H2O contents (B4 wt%) consistent with petrological estimates and the constraint that the magmas crystallised outside the amphibole stability field, demonstrating that pyroxene- hosted melt inclusions can preserve H2O contents close to their entrapment values even in effusive eruptions with low effusion rates (0.6 m 3 s -1 ). The absence of noticeable H2O loss in some of the inclusions requires post-entrapment diffusion coefficients (B1 9 10 -13 m 2 s -1 ) at least several order of magnitude smaller than experimentally determined H ? diffusion coefficient in pyroxenes. The H2O content distribution is, however, not uniform, and several peaks in the data, interpreted to result from diffusive H2O reequil- ibration, are observed around 1 and 0.2 wt%. H2O diffusive loss is also consistent with the manifest lack of correlations between H2O and CO2 or S contents. The absence of tex- tural evidence supporting post-entrapment H2O loss sug- gests that diffusion most likely occurred via melt channels prior to sealing of the inclusions, rather than through the host crystals. Good correlation between the melt inclusion sealing and volcano-tectonic seismic swarm depths further indicate that, taken as a whole, the melt inclusion popu- lation accurately records the pre-eruptive conditions of the magmatic system. Our data demonstrate that H2O diffusive loss is a second-order process and that pyroxene-hosted melt inclusions can effectively record the volatile contents and decompression-induced crystallisation paths of vapour- saturated magmas.

The effects of re-homogenization on plagioclase hosted melt inclusions

Melt inclusions trapped in phenocrysts provide a unique picture of magma systems prior to modification by crustal processes. However, post-entrapment crystallization complicates their interpretation. Re-heating the phenocryst to the temperature of entrapment is a commonly applied method to recover the original melt composition. To understand the effects of re-homogenization, we compared the composition of re-heated and naturally quenched melt inclusions and inclusion compositions that had been subjected to over-heating and under-heating to examine the degree to which anomalous compositions were produced. Our results on plagioclase hosted inclusions from ocean floor basalts indicate that the general patterns represented by naturally quenched inclusions are the same as observed for rehomogenized inclusions. Most important, the range of minor elements described for plagioclase hosted inclusions from basalts is found in naturally quenched inclusions, and is therefore not a consequence of the re-homogenization process.

Magmatic degassing of ore-metals at the Menez Gwen: Input from the Azores plume into an active Mid-Atlantic Ridge seafloor hydrothermal system

Menez Gwen is an active seafloor hydrothermal system located on the Mid-Atlantic Ridge at 37°50′N under the Azores plume influence and within enriched MORB type rocks. Two distinct highly plagioclase phyric (HPP) lavas occur at Menez Gwen, type I and type II. Type I HPP lavas are vesicular and highly enriched in incompatible trace elements; type II HPP lavas are less vesicular and less enriched. Melt inclusions in plagioclase and olivine phenocrysts in these HPP basalt lavas were studied by laser-ablation inductively-coupled plasma mass spectrometry (LA-ICP-MS). The LA-ICP-MS data support previous findings that mixing of melts occurred below Menez Gwen. Olivine crystals enclosing melt inclusions having compositions of both type I and type II magmas occur in type I HPP basalts. All plagioclase hosted melt inclusions have type II signatures regardless of their host rock character. The melt inclusions of the less incompatible element enriched type II lavas are more enriched in Cu than type I, an observation that is relevant to the origin of metals characteristic of these particular basalt-hosted seafloor hydrothermal systems. Furthermore, petrographic and analytical observations of melt inclusions indicate exsolution from the magma of an immiscible volatile phase carrying Cu and S. Subsequent injection into the magma chamber of the more enriched type I melt coupled with extensive pre-eruptive degassing of CO 2 and S likely triggered the mobilization of the exsolved ore-metals. High volatile contents and extensive magmatic degassing in a shallow magma chamber and mixing of enriched melts were contributors to the seafloor hydrothermal activity at Menez Gwen.

Mafic enclaves in the rhyolitic products of Lipari historical eruptions; relationships with the coeval Vulcano magmas (Aeolian Islands, Italy)

The recent finding of mafic enclaves in the Rocche Rosse (RR) lava flow, the last magmatic product on Lipari (Aeolian Islands, Italy) (AD 1230 ± 40), opens the possibility to investigate in detail the most recent magmatic system of the island, an important issue for the volcanic hazard assessment of the area. The RR lava flow is an aphyric rhyolitic coulee consisting of grey and black pumice and black and grey obsidian. Enclaves have ellipsoidal to spheroidal shape and vary from mm-sized in the central portion of the flow, to cm-sized, at the top and in the flow front, where they are also more abundant. Enclaves are shoshonitic-latitic (group A) and trachytic (group B) in composition. The mineralogy of group A consists of dominant clinopyroxene crystals with minor abundance of feldspar (plagioclase > K-feldspar), olivine and biotite, while group B is composed of feldspar (K-feldspar > plagioclase) with minor clinopyroxene, olivine and biotite. Geochemical modeling suggests that the host rhyolitic rocks could be the product of AFC (Assimilation plus Fractional Crystallization) of a magma compositionally similar to the associated shoshonitic-latitic enclaves, which, in turn, could be obtained, through an AFC process, from the primitive melts erupted as olivine hosted melt inclusions during the last 15 ka at Vulcano. The already-known last 42 ka relationship between Lipari and Vulcano Islands is here reinforced until historical time, especially for the last 1 ka. The geochemical and petrological overlap between Lipari and Vulcano is interpreted to reflect the existence of a similar magmatic system underneath the two islands. The nearly aphyric RR rhyolites are interpreted to be the products of a superheated (temperature far above the liquidus) and initially water-undersaturated magma that underwent degassing close to the surface inhibiting microlite crystallization.

Magma mixing in upper mantle: Evidence from high Mg# olivine hosted melt inclusions in MORBs near East Pacific Rise 13°N

Early formed high-Mg-# olivine phenocrysts during evolution of MORB magmas usually host melt inclusions, which record importantw information about the early-stage evolution of magma. Five MORB samples from near East Pacific Rise (EPR) 13 degrees N vary little in K/Ti (0.07-0.12), Tb/Lu (1.72-1.84) and Sm/Nd (0.310-0.332) and have similar REEs patterns, indicating that depleted upper mantle has similar mineral composition. Sixty-five initial melt inclusions derived by correcting olivine fractionation and "FeO-Loss" show averagely higher MgO contents than their host rocks. Melt inclusions have higher CaO/Al2O3 ratios than their host rocks, and these CaO/Al2O3 ratios are positively and negatively correlated with MgO and Na2O respectively, suggesting that these magmas have experienced high pressure crystallization of clinopyroxene. Average crystallization pressure, which is calculated based on the pressure dependence of clinopyroxene crystallization, is 0.83 +/- 0.25 GPa, and implys that these melt inclusions are averagely trapped in mantle depth of similar to 24 km. These melt inclusions show negative correlations of Ca-8/Al-8 and Na-8 with Fe-8, and wider ranges of Ca-8/Al-8, Na-8, Fe-8 and K/Ti than their host rocks, suggesting that these melt inclusions formed by mixing magmas of different melting degrees and depths. According to the average value and ranges of Ca-8/Al-8, Na-8, Fe-8 and K/Ti, these magmas would necessitate other mixing ends in shallow crust except in upper mantle. The compositional diversity of melt inclusions in MORBs phenocrysts cannot always be used to indicate magma mixing and crystallization in shallow crust, and melt inclusions in high Mg-# olivine formed under mantle pressure must be excluded in study of the magma process at crustal level. This study shows that, in EPR, MORBs have experienced mixing of magmas formed by different melting degrees and depths in the mantle.

The heterogeneous nature of the Southern Tyrrhenian mantle: Evidence from olivine-hosted melt inclusions from back-arc magmas of the Marsili seamount

We present major and trace element data of olivine (Fo 86–91) — hosted melt inclusions from five Island Arc Basalt (IAB) lavas from Marsili seamount, a large volcano in the active Marsili back-arc basin of the Southern Tyrrhenian. The geochemical composition of this suite of melt inclusions varies widely, extending the range of variation of the host lavas and providing further insights into the mantle components entering the melting region under the Marsili volcano. The trace element trends defined by melt inclusions with compositions similar to the host rock confirm the involvement of two different mantle domains, affected by different degree of subduction-related metasomatism. Melt inclusions generated from the ambient MORB mantle wedge are depleted in incompatible trace elements and show a pronounced subduction signature, resembling the Marsili IAB- group 2 lavas. Melt inclusions derived from heterogeneous mantle, where small blobs of an OIB enriched component are also involved in the melting, are less depleted in incompatible trace element, and are geochemically similar to the Marsili IAB- group 1 lavas. The olivine-hosted melt inclusions with a composition different to that of the host rock reveal two further distinct geochemical signatures. One signature resembles that of the OIB-like lavas from the nearby Ustica Island, and is also alike to that of the few OIB-like lavas so far recovered from the Marsili volcano. However, the Marsili OIB-like melt inclusions are richer in SiO 2, TiO 2, Na 2O, Zr, Y and HREE and depleted in CaO relative to these OIB-like lavas, suggesting an origin by melting of trachyandesitic bodies in the lower lithospheric mantle, which are themselves formed by high pressure fractional crystallization of parental OIB-like melts. The other detected signature is similar to that of the MORB lavas recording the early ocean-formation stage of the Southern Tyrrhenian back-arc basin. Nevertheless, the high CaO contents of these melts suggest that they probably resulted from melting of clinopyroxene-rich veins related to this early ocean-forming stage.

Lithogeochemistry of volcanic rocks and hosted melt inclusions in Wudalianchi, Heilongjiang, China

Wudalianchi (五大连池) volcanoes are famous Quaternary ones with fairly good volcanic landscape in Northeast China. The volcanic rocks are phonotephrite, tephri-phonolite, trachyandesite, and basaltic trachyandesite with SiO2 contents of 47.3 wt.%–54.2 wt.%. The characteristic of high K2O content in Wudalianchi volcanic rocks suggests that they probably share a common potassic magma source. The magma evolutions of Wudalianchi volcanic rocks are similar, while the new eruption products of Laoheishan (老黑山) and Huoshaoshan (火烧山) volcanoes are slightly more evolved than the old eruption materials, i.e., the magma of the new eruptions in 1719–1721 from the above two volcanoes is the evolved magma in underground chamber. The main phenocrysts of Wudalianchi volcanic rocks are olivines, clinopyroxenes, and some characteristic K-rich leucites. The various-shaped melt inclusions are found in olivine phenocrysts from new eruption products of Laoheishan and Huoshaoshan volcanoes. By electron microprobe (EMP) analyzing, the contents of SiO2 and total alkali show a wide range, suggesting that the pre-eruptive magma was probably more complicated, or the melt inclusions were rather unhomogeneous. On the basis of EMP results, it is believed that the sulfur degassing rate from this new eruption was much higher than that of the millennium eruption of Tianchi (天池) volcano, while the chlorine degassing rate was a magnitude lower than that of Tianchi volcano.

( 210Pb/ 226Ra) variations during the 1994–2001 intracaldera volcanism at Rabaul Caldera

Determining the timing and source of gas transfer during intermittent intracaldera volcanism can aid in our understanding of degassing in these large systems. Using ( 210Pb/ 226Ra) ratios, (parentheses denote activity ratios) as a time-sensitive tracer, injections of 222Rn and the subsequent time scales of gas accumulation and loss can be determined. Variations in ( 210Pb/ 226Ra) have been measured for 15 volcanic products erupted at Rabaul Caldera over the period 1994 to 2001. In addition, one basaltic enclave from the 1937 eruption was also analyzed. Water and carbon dioxide contents determined from olivine hosted melt inclusions erupted in 1997 are < 1% and suggest extensive shallow-level degassing. Both 210Pb excesses and deficits are found in andesites and dacites, whereas the basaltic enclave displays an ( 210Pb/ 226Ra) 0 ratio of 7. Between 1994 and 1997 three samples with ( 210Pb/ 226Ra) deficits were erupted which indicate open system gas loss since 1992 and 1994. No correlation exists between ( 210Pb/ 226Ra) and lava chemistry, eruptive style or date. 210Pb excesses are more common than deficits in Rabaul samples but cannot be explained by plagioclase feldspar accumulation, Pb sublimate accumulation or differentiation. Instead, a model of intra-magma 222Rn transfer can produce 210Pb excesses of the appropriate magnitude if gas transfer occurs over 1–5 years from an underlying body of magma that is 2–10 times larger than the volume of erupted material and that is consistent with geophysical estimates. Although intermittent gas transfer events can be inferred by the development of 210Pb excess, there is no evidence at Rabaul for a direct link between eruptive style, gas flux and ( 210Pb/ 226Ra).

Copper transport by high temperature, sulfur-rich magmatic vapor: Evidence from silicate melt and vapor inclusions in a basaltic andesite from the Villarrica volcano (Chile)

Abstract Silicate melt inclusions have been analyzed in co-crystallized plagioclase, olivine and clinopyroxene in a typical arc-related basaltic-andesite from the Villarrica volcano (Chile). The melt inclusions show identical compositions with respect to most major and trace elements in all host minerals. However plagioclase hosted melt inclusions are variably enriched in Cu (up to ~ 4800 ppm) and Ag (up to ~ 950 ppb). We interpret the elevated Cu and Ag concentrations to be the result of heterogeneous entrapment of silicate melt with a sulfur-rich, chlorine-poor high-temperature magmatic vapor phase in which Cu and Ag were strongly compatible. This assumption is supported by the presence of individual Cu-rich vapor inclusions in some plagioclase crystals. Such a magmatic vapor phase may play a significant role in the transport of S, Cu and Ag within or between magmatic reservoirs and, consequently, in the formation of porphyry-type ore deposits.

Determination of sulfur speciation and oxidation state of olivine hosted melt inclusions

In order to better understand what controls sulfur speciation in melt inclusions, and how that pertains to the original basalt composition, we have conducted a series of heating experiments on naturally quenched and crystalline olivine-hosted melt inclusions. Sulfur speciation was determined from S Kα peak shift measurements by electron microprobe on the experimentally heated inclusions as well as a series of naturally quenched inclusions, and matrix glasses. Naturally quenched olivine-hosted melt inclusions record a similar but more variable sulfur speciation relative to matrix glasses, (up to 45–50% variation in S 6+/S total). Much of this range can be attributed to the effect of degassing which may either increase or decrease the S 6+/S total. In addition, olivine melt re equilibration and H diffusion out of the inclusion both potentially result in the oxidation of melt inclusions. Heating of melt inclusions can have different effects on the sulfur speciation under different conditions. A slight decrease in S 6+/S total and oxygen fugacity (∼0.1 log units) can occur from overheating of inclusions (above the temperature of entrapment), resulting from excess ferrous iron in the melt. An increase in heating times should result in an oxidation of the inclusion generated by increased H diffusion out of the inclusion. However, results of heating experiments on melt inclusions from an Izu backarc basalt for less than 30 min do not show a significant increase in sulfur oxidation. In addition, experiments conducted at both IW and FMQ have measured sulfur speciation consistent with naturally quenched inclusions suggesting that at experimental temperatures near that of olivine crystallization the furnace atmosphere does not exert significant control on the melt fO 2. By taking these parameters into account, sulfur speciation and oxidation state of basaltic melt trapped within inclusions can be accurately determined from both naturally quenched and heated olivine hosted melt inclusions.

Plagioclase and pyroxene hosted melt inclusions in basaltic andesites of the current eruption of Arenal volcano, Costa Rica

In this study, we investigated melt inclusions hosted in pyroxene and plagioclase to find direct evidence for the composition of melt components in the pre-eruptive magmas of the current eruption of Arenal volcano and to further shed light on the petrogenetic history of remarkably uniform basaltic andesitic bulk compositions. Composition of melt inclusions ranges widely regardless of whether an inclusion is hosted by plagioclase or by pyroxene and whether crystals with inclusions came from tephra samples or from slower emplaced lava flows. However, inclusions from each type of host mineral have distinct evolutionary trends most consistent with being mainly introduced by post-emplacement crystallization of the enclosing host, which is also supported by the composition of groundmass glasses. At the least-modified compositions, plagioclase and pyroxene inclusion trends overlap allowing for identification of melt compositions entrapped by both phases and, thus, strongly suggesting that these melt compositions existed in the reservoir prior to entrapment. Most of these are “dacitic” (61 to 64 wt.% SiO 2) and strongly match phenocryst-poor dacitic magmas of earlier eruptive phases of Arenal [Borgia, A., Poore, C., Carr, M.J., Melson, W.G., Alvarado, G.E., 1988. Structural, stratigraphic, and petrologic aspects of the Arenal–Chato volcanic system, Costa Rica: evolution of a young stratovolcanic complex. Bull Volcanol, 50, 86–105], suggesting earlier dacitic magmas may have been generated by melt-extraction processes. Correction for host crystallization of some inclusions also suggests that melt components as mafic as ∼ 53 wt.% may have been entrapped. All melt components inferred to have existed in the magmatic reservoirs prior to entrapment have low Mg# (38–45) yielding evidence for liquid compositions required for crystallizing the bulk of observed pyroxene and likely also olivine. Water rich (> 5 wt.%, by difference) and probably undegassed (S = 400–1600 ppm, Cl = ∼ 2500 ppm) melt inclusions are found among inclusions from tephra. The close association of water-rich, dacitic melt inclusions with very anorthitic (85–91) plagioclase host compositions suggests that high An plagioclase may originate also from rather silicic liquids. Furthermore, our study shows that melt inclusions from Arenal, even if only those that come from rapidly chilled tephra samples are considered, are complicated. They suggest rather multi-stage and complicated assembling scenarios for erupting bulk compositions during the current activity and likely for prior activities as well.

Contamination and melt aggregation processes in continental flood basalts: constraints from melt inclusions in Oligocene basalts from Yemen

Melt inclusions from Oligocene continental flood basalts (CFB) erupted in Yemen provide unique insight into the timing and nature of the processes that lead to crustal contamination and melt aggregation in CFB magmas. Large variations in trace element indices that are sensitive to the degree and composition of assimilated crustal material (e.g. K2O=0.20–1.94 wt%, Ba=13–543 ppm, K/Nb=128–1603, Ba/Th=9–303) are evident in many inclusions, even where these derive from the same host lava, and reflect the complexity of the processes that lead to contamination within individual CFB melting and melt-transport systems. The compositions of melt inclusions relate to differences in the degree of contamination, but in addition require that there is substantial heterogeneity in the composition of the contaminant material itself. Many inclusions also appear to contain more primitive melts than typical Yemen CFB lava compositions, and as such would be highly sensitive to addition of crustal materials. Overall melt inclusions provide a markedly better record of the diversity of melt compositions present within given CFB magma systems than the bulk compositions of erupted lavas.Highly contaminated melts (with K/Nb and Ba/Nb up to 957 and 22, respectively) trapped within fosterite-rich olivines (Fo85–90) require very high rates of assimilation relative to crystal fractionation, with Ma/Mc values (the mass ratio of assimilated to crystallized material)≫1. Such rapid assimilation may reflect decoupling of heat and mass transfer at the margins of larger magma chambers, within feeder dyke complexes, or at other sites where primitive magma is juxtaposed against wall-rocks that are already heated to temperatures near, or above, their solidus. In addition, relatively little assimilation appears to have occurred after crystallization of the phases that host melt inclusions, consistent with a thermal link between assimilation and phenocryst formation. Melt inclusions also show trace element variations related to mantle source compositions and mantle melting processes. Two inclusions with unusually Sr-rich and rare earth element-poor compositions are similar to those recognized from Mauna Loa, Hawaii, and may be related to melting of recycled gabbroic material within the upwelling Afar plume. In addition, many melt inclusions with crustally contaminated compositions also show large variations in trace element ratios that are essentially insensitive to crustal contamination (e.g. Zr/Y=2–10) but are fractionated during progressive partial melting within an upwelling mantle column. The presence of both mantle and crustal-derived trace element signatures in the same inclusions demonstrates that melt transport systems in Yemen CFB were capable of transporting compositionally distinct melt batches, without complete mixing, through the asthenospheric and lithospheric mantle until final mixing and aggregation (and contamination) within crustal magma reservoirs. Thus, regardless of contamination, the ultimate compositions of many CFB lavas may be determined by magma mixing within the crust, rather than representing primitive compositions derived directly from the mantle.

Thermal and compositional evolution of the shallow magma chambers of Vesuvius: Evidence from pyroxene phenocrysts and melt inclusions

A large number of clinopyroxene (cpx) phenocrysts and hosted melt inclusions (MI) in pyroclasts from several Vesuvius eruptions were analyzed. Assuming that their temperature of homogenization (Thom) reflects that of crystallization of the host mineral and that after homogenization and quenching, composition of the MI represents the melt from which the host crystallized, each cpx‐MI pair records equilibrium conditions at Thom. These data were used to discuss the magma evolution within shallow magma chamber. They formed by means of the periodic arrival of mafic batches, recorded by the MI in diopside (Fs4–7) crystallized at 1150°–1200°C. Under open conduit conditions the chambers are small, and each magma pulse induces thermal and compositional variations recorded by oscillatory zoned pyroxenes (Fs4–15). Their MI and Thom summarize the evolution of the chamber resulting from cycles of magma injection, crystal settling, and magma extraction. The deposits of the explosive eruptions which reopen the conduit reflect variable mixing during magma withdrawal from zoned reservoirs, whose layering was deciphered through MI and Thom in salite (Fs15–30). We suggest that the Vesuvius magma chambers evolve from prolate toward subequant, changing their layering with increasing volume and age: (1) initial stage, high aspect ratio chamber, homogeneous mafic melt (T°C∼1100) crystal enriched downward; (2) young stage, medium aspect ratio, continuous gradation from mildly evolved (T°C∼1050) to felsic melt (T°C∼850–900); and (3) mature stage, low aspect ratio, twofold chamber with stepwise gradient separating lower, convective, mildly evolved portion (T°C∼1050) from upper, stratified, felsic portion (T°C∼800–950).