| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 45 | Number 3 | Pages 531-554 | 2004
Journal of Petrology 45(3) © Oxford University Press 2004; all rights reserved.
Ultra-calcic Magmas Generated from Ca-depleted Mantle: an Experimental Study on the Origin of Ankaramites
1 INSTITUTE OF MINERALOGY AND PETROLOGY, ETH ZURICH, 8092 ZURICH, SWITZERLAND
2 RESEARCH SCHOOL OF EARTH SCIENCES, ANU, CANBERRA, A.C.T. 0200, AUSTRALIA
RECEIVED JANUARY 1, 2003; ACCEPTED JULY 31, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Ultra-calcic ankaramitic magmas or melt inclusions are ubiquitous in arc, ocean-island and mid-ocean ridge settings. They are primitive in character (XMg > 0·65) and have high CaO contents (>14 wt %) and CaO/Al2O3 (>1·1). Experiments on an ankaramite from Epi, Vanuatu arc, demonstrate that its liquidus surface has only clinopyroxene at pressures of 15 and 20 kbar, with XCO2 in the volatile component from 0 to 0·86. The parental Epi ankaramite is thus not an unfractionated magma. However, forcing the ankaramite experimentally into saturation with olivine, orthopyroxene and spinel results in more magnesian, ultra-calcic melts with CaO/Al2O3 of 1·21–1·58. The experimental melts are not extremely Ca-rich but high in CaO/Al2O3 and in MgO (up to 18.5 wt %), and would evolve to high-CaO melts through olivine fractionation. Fractionation models show that the Epi parent magma can be derived from such ultra-calcic experimental melts through mainly olivine fractionation. We show that the experimental ultra-calcic melts could form through low-degree melting of somewhat refractory mantle. The latter would have been depleted by previous melt extraction, which increases the CaO/Al2O3 in the residue as long as some clinopyroxene remains residual. This finding corrects the common assumption that ultra-calcic magmas must come from a Ca-rich pyroxenite-type source. The temperatures necessary for the generation of ultra-calcic magmas are
1330°C, and their presence would suggest melting regimes that are at the upper temperature end of previous interpretations made on the basis of picritic magmas. KEY WORDS: ankaramite; refractory lherzolite; ultra-calcic magma
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Ultra-calcic magmas or ankaramites (CaO/Al2O3 weight ratio
1·1) occur as lavas in the Vanuatu arc (Barsdell, 1988
; Barsdell & Berry, 1990); similar compositions occur as melt inclusions in rocks of several other arcs (Schiano et al., 2000). Olivine phenocrysts with XMg >90, and coexisting chrome-rich or aluminous spinel, identify the Vanuatu ankaramites as primitive magmas. There are similarities in composition between island-arc ankaramite lavas and ultra-calcic melt inclusions in magnesian olivines from arcs, ocean-island and mid-ocean ridges (Fig. 1). Ultra-calcic melts have been divided into two groups (Schiano et al., 2000; Kogiso & Hirschmann, 2001): a relatively alkali-poor, silica-rich hypersthene-normative suite (Na2O + K2O generally 
3 wt %, SiO2 >48 wt %) and an alkali-rich, silica-poor nepheline-normative suite (Na2O + K2O = 2·5–6 wt %; SiO2 as low as 45 wt %). The latter is found uniquely in the arc environment, whereas the hypersthene-normative type occurs in arcs, ocean islands and mid-ocean ridge settings (Schiano et al., 2000).
|
, nepheline-normative alkaline-rich ultra-calcic melts [compiled by Schiano et al. (2000)
, Mid-Atlantic Ridge at 43°N (Kamenetsky et al., 1998The Vanuatu ankaramite lavas range from slightly nepheline-normative (island of Gaua) to hypersthene + olivine normative, and for this study we have chosen the hypersthene-normative Epi suite for investigation (Barsdell & Berry, 1990
). The purpose of this study is to test whether these ultra-calcic magmas may be derived from lherzolitic or harzburgitic mantle sources. We emphasize that, on the island of Epi, ankaramite magma volumes are at a 10–100 km3 scale and that melt inclusions that occur at grain-size scales (and may or may not reflect large-scale processes) may result from very different processes (Kamenetsky et al., 1998). The bulk-rock compositions of island-arc ankaramites may be influenced by accumulation and/or fractionation of olivine, clinopyroxene and chrome spinel phenocrysts. However, previous workers have used petrographic criteria and known crystal partitioning relationships to infer the liquid compositions required to precipitate the observed olivine (Mg94), diopside (Mg94) and chrome spinel phenocrysts (Barsdell, 1988; Barsdell & Berry, 1990; Eggins, 1993). These liquid compositions are our starting point for the experimental study and we use an olivine-rich ankaramite from Western Epi (sample 71046, Barsdell & Berry, 1990) that is close to the estimated parent magmas for Western Epi and Meralava (Barsdell & Berry, 1990) as a primitive, hypersthene-normative ultra-calcic magma.
The melting of lherzolitic mantle is experimentally well studied (Figs 1 and 2). Anhydrous batch melting of fertile or slightly refractory lherzolite cannot yield magmas with CaO/Al2O3 significantly higher than unity. In all of the lherzolite melting studies, liquid compositions increase in CaO and CaO/Al2O3 up to the point of clinopyroxene disappearance (Fig. 2) leaving orthopyroxene and olivine in the residue. With further melting, they then decrease in CaO more strongly than in Al2O3 through melting of relatively aluminous orthopyroxene, so that the CaO/Al2O3 decreases slightly. The maximum CaO/Al2O3 of 1·0 at the clinopyroxene-out stage remains far below the values achieved in ultra-calcic magmas (maximum 1·5–1·7). It is thus difficult to obtain ultra-calcic magmas from mantle compositions with excess orthopyroxene (relative to clinopyroxene). Alternatively, melting of Ca-rich bulk compositions such as olivine pyroxenites and wehrlites has been suggested as a mechanism for generating ultra-calcic magmas. Kogiso & Hirschmann (2001) studied the dry melting of various wehrlitic compositions and obtained ultra-calcic melts. However, these melts were nepheline-normative, were obtained only at melt fractions above 30% (see Discussion) and were unsuitable as parents to the ultra-calcic melt inclusions, leading Kogiso & Hirschmann to discuss the possibility of obtaining ultra-calcic melts from depleted peridotite.
|
Dissolved CO2 could shift the olivine + orthopyroxene + clinopyroxene + spinel saturation surface to higher normative diopside and CaO/Al2O3 than for H2O-bearing or dry melting processes. In the study of Brey & Green (1977)
, dissolved (CO3)2- and (OH)- caused melilitites to saturate in orthopyroxene in addition to clinopyroxene + olivine, allowing olivine melilitite to form from residual garnet lherzolite at pressures of 25–35 kbar. Our experimental approach was to determine the influence of variable CO2 and H2O on temperature and phase relationships at the liquidus of a parental island-arc ankaramite with the scope to produce melts with high normative diopside and CaO/Al2O3. In a second step, we designed our methods to force the parental Epi melt into saturation with olivine, orthopyroxene and spinel with varying CO2:H2O in the dissolved volatile component. The objective was to find a melt composition for each CO2:H2O just at the elimination of clinopyroxene from the residue, i.e. at the temperature at which the highest CaO contents and CaO/Al2O3 necessarily occur. In the following we use ‘ultra-calcic’ for melts with a weight ratio CaO/Al2O3 > 1·1 irrespective of the CaO content (we will show that high CaO/Al2O3 is the essential characteristic of parental ultra-calcic magmas; high CaO can be achieved in a later step).
|
EXPERIMENTAL TECHNIQUES AND STRATEGY |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Liquidus experiments
As a first step, the liquidus surface of the parental Epi magma (Barsdell & Berry, 1990
) was determined as a function of volatile composition. Two H2O–CO2 sections were investigated at 15 and 20 kbar (Table 2). This pressure range is based on the absence of garnet in the source of the Epi ankaramites (Barsdell & Berry, 1990), which results in an upper pressure limit of c. 25 kbar. The thickness of the intra-oceanic arc crust beneath the Vanuatu arc is not well known, but in the very similar Izu–Bonin–Mariana arc system, it is >20 km (Suyehiro et al., 1996). This sets a lower pressure limit for mantle melting of roughly 10 kbar, assuming that the Epi magmas are mantle-derived.
A glass with a composition equivalent to the Epi parental magma (Table 1) but deficient in MgO was prepared at controlled oxygen fugacity to obtain an Fe3+/Fetot of 0·2. The deficiency in MgO was then compensated for by adding previously dried Mg(OH)2 and/or MgCO3. This procedure results in a constant molar volatile/non-volatile oxide ratio and in a varying XCO2 in the volatile component. Addition of Mg(OH)2 only resulted in 4·1 wt % H2O and XCO2 = 0·0 in the starting material. Addition of reagent grade MgCO3 always results in addition of some H2O as well, as synthetic MgCO3 is difficult to dry without causing decarbonization. We determined the H2O content of MgCO3 to be 3·1 wt % after drying at 110°C. This results in 9·3 wt % CO2 + 0·57 wt % H2O and XCO2 = 0·87 for addition of ‘MgCO3 only’ to the MgO-deficient glass. Batches with the original MgO content of the Epi parental magma were prepared with CO2/(H2O + CO2) fractions of 0·0, 0·24, 0·46, 0·66, 0·78 and 0·87. Primitive arc magmas generally do not contain more than 2–3 wt % H2O (Sobolev & Chaussidon, 1996) and our high volatile contents were chosen to maximize the effect of volatiles on the liquidus within reasonable bounds. Experiments were performed in an end-loaded 
-inch piston cylinder apparatus using a standard salt–Pyrex–graphite–crushable alumina assemblage. The inner furnace diameter was 3·2 mm and the capsules were separated from the furnace by an alumina sleeve. A PtRh type-B thermocouple was used for temperature measurement. Most experiments were conducted in welded graphite–Pt double capsules (1·0 mm i.d. graphite, 1·8 mm i.d. Pt, 2·3 mm o.d. Pt). Run times were 30 or 60 min. The high temperatures of this experimental study resulted in reducing conditions within graphite capsules, and a few experiments were repeated in welded 2·0 mm o.d. Au80Pd20 capsules. Fe loss to the Au80Pd20 capsule was insignificant (see below) and melt compositions in super-liquidus experiments were identical to initial bulk compositions within analytical uncertainty.
|
|
Saturation experiments
A second set of experiments was designed to force the parental Epi melt into saturation with olivine, orthopyroxene, clinopyroxene and spinel under a range of XCO2 values of the dissolved volatile component (Table 3). The objective of this series was to determine melt compositions at the temperature at which clinopyroxene disappears from the solid residue for each XCO2.
|
A layer of Epi melt starting material was placed below a peridotite layer within the capsule(s), maintaining a roughly constant peridotite:ankaramite proportion of 4:6 (Table 3). The peridotite was synthesized from an oxide + iron mix at 10 kbar, 1200°C. The resulting fine-grained (<10 µm) mix of olivine, enstatite, diopside and spinel (for actual mineral compositions and proportions, see Table 1) was finely ground before further use as a peridotite layer. The bulk composition of the refractory peridotite layer was based on mineral compositions from the phenocryst phases of a high Ca-boninite (Tongan Arc, Sobolev & Danyushevsky, 1994
, sample 26-2), which are thought to be appropriate for refractory melt-depleted mantle. The synthetic peridotite had low clinopyroxene and high orthopyroxene contents because the ankaramite melt-layer was already clinopyroxene saturated, but considerable dissolution of orthopyroxene in the melt was expected. The first two experiments revealed dissolution fronts of olivine within the peridotite layer so that the melt layer was not in contact with olivine. To ensure unzoned capsules and that all melts were in contact (and thus saturated) with olivine, all subsequent experiments had an additional 5 wt % San Carlos olivine (XMg = 0·90) mixed into the melt layer. These experiments (Table 3) did not show any zonation in the peridotite layer.
With this experimental setup we were able to obtain melt and mineral compositions at 15 kbar for XCO2 values of 0·0, 0·24, 0·46, 0·66 and 0·78 for melts saturated in ol–opx–cpx–sp and, within a temperature interval of 20°C, melts saturated in ol–opx–sp. A series of experiments was also performed with a dry glass, i.e. a volatile-absent bulk composition, that was directly fused from oxides. Experiments carried out with welded graphite–Pt double capsules yielded low oxygen fugacities (from fO2 QFM - 2·7 = IW + 0·9 at 1290°C to XCO2 QFM - 2·9 = IW + 0·6 at 1415°C, where QFM is quartz–fayalite–magnetite and IW is iron–wüstite). Because of the presence of graphite, redox exchange between C–H–O species dissolved in the melt and the capsule occurs. We have not as yet been able to obtain direct 
and OH- contents of the glass pools (by IR spectroscopy) and in the following, XCO2 and XH2O values refer to the initial mixes. To avoid the graphite capsule and reducing conditions, a few runs were repeated employing welded Au80Pd20 single capsules, with the intention that during relatively short run times the initial Fe3+/Fetot should more or less be maintained. After the experiment, Fe was measured in the AuPd capsules. A zone of 20–40 µm width with Fe above detection limit was identified; the maximum measured Fe content in the metal was 0·4 wt %, resulting in insignificant Fe loss to the sample.
Recently, it was suggested (Liu & O'Neill, 2002) that Cr/Al of the source may significantly influence the Ca/Al of melt compositions without significant increase of the Cr2O3 concentration in the melt. To evaluate this parameter, two additional peridotite mixes with (molar) XCr = Cr2O3/(Cr2O3 + Al2O3) = 0·466 and 0·294 were prepared, i.e. with Cr2O3 contents lower than our initial mix (with XCr = 0·748). The experimental ‘clinopyroxene-out bracket’ at XCO2 = 0·78 and a bulk XCr of approximately 0·187 (exact bulk compositions depend on the exact peridotite: ankaramite layer thicknesses) was repeated at identical conditions but with bulk XCr of 0·066 and 0·033 (Tables 1 and 3).
|
EXPERIMENTAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The liquidus of the Epi ankaramite ‘parental’ magma
Clinopyroxene is the single liquidus phase for all investigated XCO2 at 15 and 20 kbar (Fig. 3). Liquidus temperatures rise from 1300 to 1330°C with XCO2 increasing from 0·0 to 0·87 at 15 kbar and from 1360 to 1410°C with XCO2 increasing from 0·0 to 0·87 at 20 kbar. Characteristically, the first experiment below the liquidus temperature has 50–70% clinopyroxene present, although this observation may be moderated by some quench rim growth on primary clinopyroxene. Dendritic quench clinopyroxene is always abundant in the interstitial melt, and, in pools of melt, 50–200 µm wide zones of quench clinopyroxene are common. The most extreme examples of high degrees of crystallization over a small temperature interval occur in experiments (e.g. D-33, 1320°C, 15 kbar, Fig. 4) in which it appeared that one part of the capsule was at super-liquidus conditions (the measured melt composition in this part being that of the bulk composition) and the remainder of the capsule contained >50% clinopyroxene crystals. The crystallization of clinopyroxene in only one part of the capsule is ascribed to a temperature gradient over the capsule length (typically 1·2–1·5 mm inner length) and cannot be attributed to chemical differentiation within the capsule. The observed extensive crystallization over a narrow temperature interval is similar to the diopside–orthopyroxene–olivine ternary in the CaO–MgO–SiO2 system in which the interval between super-liquidus and sub-solidus does not exceed 20°C where clinopyroxene is in excess of orthopyroxene and olivine (Kushiro, 1969
). The reason for crystallization behaviour similar to that in the simple synthetic system is probably a reflection of the low concentrations of TiO2 and Na2O, and, compared with most other mafic magmas, relatively low Al2O3 contents of the bulk composition and clinopyroxenes.
|
|
The composition of clinopyroxenes at the liquidus surface is in the range 0·62–0·78 Ca p.f.u., 0·10–0·32 Al p.f.u. and XMg = 0·80–0·91 (Table 4). Strong zonation is observed in clinopyroxene, with the highest XMg and Ca contents and lowest Al contents in the cores. This zonation is a consequence of the large amount of crystallization occurring within an experiment. The first clinopyroxene nucleates in equilibrium with a melt close to the bulk composition. However, as clinopyroxene grows, the residual melt evolves and becomes enriched in Fe, Al and Na. Thus, with increasing degree of crystallization, clinopyroxene crystallizes in equilibrium with an increasingly Fe-, Na- and Al-enriched and Ca- and Mg-depleted melt without diffusional equilibration (Cherniak, 2001
) of previously formed clinopyroxene cores. Ca contents of clinopyroxene on the liquidus decrease with increasing XCO2 from approximately 0·74 to 0·64 Ca p.f.u. Nevertheless, clinopyroxene cores of this series of experiments are at least 0·1–0·2 Ca p.f.u. higher than a clinopyroxene that would lie on the orthopyroxene–clinopyroxene solvus. This might be taken as an indication that even at the highest XCO2 contents of the experiments, we are still ‘far’ from orthopyroxene saturation.
|
The conclusion of this series of experiments is that the estimated parental liquid of the Epi ankaramites is not in equilibrium with any possible orthopyroxene-bearing mantle residue over the investigated pressure range. A similar conclusion was reached by Della Pasqua & Varne (1997)
who investigated ankaramitic melt inclusions from Rindjani volcano, Indonesia, which have a range of CO2 and H2O dissolved in the melts.
Ultra-calcic melts in equilibrium with a mantle restite
To understand the chemical differences between a melt derived from a lherzolite source and the Epi parental magma, a layer of the latter melt was forced into saturation with olivine, orthopyroxene and spinel. At each XCO2 and with rising temperature, this second series of experiments resulted in a characteristic succession of three run products (Fig. 5):
- at temperatures more than 15°C below the clinopyroxene-out reaction, a large amount of clinopyroxene (30–60%) crystallized in the ankaramite layer, leaving a small melt pool around the rim of the capsule (Fig. 6a and b; D-76, D-77, D-79 in Table 3). In these experiments melt reacted with the peridotite layer to increase modal pyroxene and infiltrated only into the lower portion of the peridotite layer. We cannot be certain that the analysed melts in the melt pool have fully equilibrated with the peridotite layer. The extensive crystallization of clinopyroxene is analogous to that observed in the liquidus experiments and melt compositions from these experiments have relatively low CaO/Al2O3 (0·93–1·23).
- At temperatures less than 15°C below the clinopyroxene-out reaction, the ankaramite layer was fully molten, interstitial melt was present in the entire peridotite layer and modal amounts of clinopyroxene in the bulk decreased (Fig. 6c and d). In some experiments a layer of clinopyroxene ± spinel was present at the bottom of the capsule.
- At temperatures above the clinopyroxene-out reaction, residual orthopyroxene and olivine often grew to grain sizes of 100 µm (Fig. 6e–g). The former peridotite layer was typically reduced to half of its initial thickness and in most experiments a few 100–200 µm crystals of orthopyroxene or olivine were observed. These appeared to have sunk through the melt layer towards the bottom of the capsule (Fig. 6f). At the bottom of the capsules, an orthopyroxene–olivine–spinel layer is present. This layer could result from crystal settling, but could also have formed through a dissolution–reprecipitation process.
|
|
Melt compositions
Melt compositions at XCO2 values of 0·00, 0·24, 0·46, 0·66 and 0·78 were measured in experiments in which melts were saturated with olivine–orthopyroxene–clinopyroxene–spinel and with olivine–orthopyroxene–spinel. These are melts that are just below and above the clinopyroxene-out reaction, respectively (Table 5). For each XCO2, the highest CaO/Al2O3 is obtained at the clinopyroxene-out reaction, and melt compositions vary systematically as a function of XCO2 and capsule material. Similar to the liquidus experiments, clinopyroxene in the peridotite layer melts over a narrow (<20°C) temperature interval, causing significant variations in melt compositions just below the clinopyroxene-out reaction as a result of different amounts of coexisting clinopyroxene. In the following, only melt compositions just above the clinopyroxene-out reaction are discussed. These melts all have high CaO/Al2O3 ranging from 1·31 at XCO2 = 0·0 to a maximum value of 1·57 at XCO2 = 0·78. Although CaO/Al2O3 is as high as in most natural ultra-calcic melts, the experimental melts do not have extreme CaO contents. CaO contents are in the range 14·1–12·3 wt % and decrease with increasing XCO2. Al2O3 in the melt also decreases with increasing XCO2 (from 10·3 to 7·9 wt %). SiO2 contents are fairly constant at 50·0 ± 0·8 wt % and melts are hypersthene-normative. The experimental melts have XMg values ranging from 0·77 to 0·80 and high MgO contents of 16·3–18·6 wt %, the latter increasing with XCO2. Variation of the peridotite layer XCr from 0·75 to 0·47 resulted in only minor changes of phase and melt compositions, but with an XCr value of 0·29 in the peridotite layer spinel was eliminated from the residue and melts remained at high CaO/Al2O3 (Table 5).
|
At dry conditions, the volume mixing ratio for peridotite:ankaramite of 4:6 resulted in complete melting of orthopyroxene. Thus, the ankaramite melt was forced into saturation with opx + ol + sp in additional experiments, in which the mixing ratio was increased to 7:3. Even at 1400°C, we were not successful in saturating the resulting liquid in opx + ol + sp only, as clinopyroxene was still present. Nevertheless, the resulting melt (Table 5) in equilibrium with opx + cpx + ol + sp has a CaO/Al2O3 of 1·48 and a relatively low CaO content of 13·3 wt %.
In the normative olivine–diopside–quartz–CaTs + jadeite basalt tetrahedron (Yoder & Tilley, 1962; Green & Falloon, 1998) the experimental melts plot to normative diopside contents about twice those of previous determinations of melts from lherzolites or harzburgites (projected from olivine, Fig. 7). In projection from olivine, the experimental melts cover the field of parental melts from the Vanuatu arc. However, the latter have lower normative olivine contents, pointing towards significant olivine fractionation prior to eruption of the ‘parental’ melt compositions. In projection from diopside, our melts have higher normative olivine than experimental melts at comparable pressures in CO2-free systems, reflecting a shift of the multiple saturation surface. A similar increase in olivine solubility at high CO2 contents was previously observed for the olivine + orthopyroxene cotectic in the olivine–nepheline–quartz system (Taylor & Green, 1987) and in the olivine–kalsilite–quartz system (Edgar & Vukadinovic, 1992). This effect occurs also in complex natural basalts, in which the liquidus appearances of orthopyroxene, olivine, garnet and clinopyroxene shift as functions of CO2 and H2O contents (Brey & Green, 1977).
|
Residual phase compositions
Experimental phase compositions in equilibrium with ankaramitic melt are presented in Tables 5–7. Near the clinopyroxene-out reaction, all phases have high XMg values, olivines range from 0·90 to 0·94, orthopyroxenes from 0·91 to 0·95, and clinopyroxenes from 0·89 to 0·93. Ortho- and clinopyroxenes have low Al2O3 contents of 0·4–2·0 wt % and 0·9–2·9 wt %, respectively (all compositional values discussed in the text are for bulk compositions with XCr = 0·75). Cr2O3 contents range from 0·8 to 1·7 wt % in orthopyroxenes and from 0·7 to 2·0 wt % in clinopyroxenes. Orthopyroxenes have Cr2O3 contents that are similar to or higher than Al2O3 contents (in wt %). Spinels are characterized by high chromium contents with an XCr value of 0·70–0·79 and have a small magnetite component (0·02–0·13 Fe3+ p.f.u.; see below).
|
|
The variation of phase compositions over the investigated temperature and XCO2 range is small except for clinopyroxenes. Clinopyroxenes in equilibrium with orthopyroxene vary from 0·62 Ca p.f.u. at 1300°C to 0·36 Ca p.f.u. at 1340°C and to 0·21 Ca p.f.u. at 1380°C. In three experiments at temperatures from 1300 to 1330°C, three pyroxenes coexist: an orthopyroxene with
0·12 Ca p.f.u., a ‘pigeonite’ with 0·40–0·46 Ca p.f.u. and a ‘high-Ca’ clinopyroxene with 0·57 Ca p.f.u. The existence of two solvi at temperatures close to the stabilization of pigeonitic clinopyroxene is consistent with models derived from experiments on simple systems (Nickel & Brey, 1984). In our experiments, the Ca-rich solvus (i.e. the pigeonite–high-Ca clinopyroxene solvus) closes at 1340°C and at higher temperatures orthopyroxenes with roughly constant Ca of 0·10–0·11 a.p.f.u. coexist with ‘pigeonites’ that evolve from 0·36 Ca p.f.u. at 1340°C to 0·21 Ca p.f.u. at 1400°C.
Influence of capsule material and oxygen fugacity
Experiments with Au80Pd20 capsules yielded significantly higher Fe3+ contents in spinels (0·08–0·13 Fe3+ p.f.u.) than experiments with graphite–Pt double capsules (0·02–0·06, Fig. 8). KD values for the Fetot–Mg distribution between melt and olivine are systematically lower than the equilibrium value of 0·337 at 15 kbar (Ulmer, 1989), and calculated Fe3+/Fetot (Table 5, Fig. 8) in melts from graphite–Pt capsules are 0–4%, in contrast to 6–15% in melts from AuPd capsules (typical errors are ±2%). XMg values of minerals were 0·01–0·02 units higher in experiments with AuPd capsules compared with graphite capsules. Cr contents in melts from graphite capsules are almost twice those in melts from AuPd capsules, indicating a significant amount of Cr2+ in the former (Berry & O'Neill, 2000, and personal communication, 2001). Finally, the clinopyroxene-out reaction in the saturation experiments and the liquidus of the tested parental ankaramite are about 20°C lower in AuPd capsules compared with graphite capsules. These observations suggest that in graphite capsules, Fe was almost completely reduced to Fe2+ and Cr3+ was partly reduced to Cr2+. In AuPd capsules, the initial oxidation state was approximately conserved (even without redox reactions, the initial Fe3+ fraction of 0·2 in the ankaramite layer is lowered by melting of solid phases that contain almost only ferrous iron) and the higher XMg values in the minerals are a consequence of higher Fe3+/Fetot in the melt. The relative lowering of the phase boundaries could be explained by (1) more hydrogen in an oxidized compound (i.e. H2O) in the Au80Pd20 capsules as compared with an increase of reduced compounds (i.e. CH4) in the graphite capsules and (2) less hydrogen loss (and thus more H2O dissolved in the melt) in the AuPd capsules as compared with C–Pt capsules. Finally, the observed differences between experiments in graphite–Pt double capsules and in Au80Pd20 single capsules is partly also due to the shorter run-times in the latter (2–4 h compared with 20·5–24 h in C–Pt), which imply less diffusive hydrogen loss through capsule walls.
|
and Fe3+ fractions in melt than experiments in Pt–C capsules.CaO/Al2O3 values in the experimental melts differ for AuPd and C–Pt capsules in that the former have ratios that are c. 0·1 units lower. MgO contents are also slightly lower in experiments from AuPd capsules. Whether the reason for this difference is in the difference of oxidation state, the probably higher H2O contents in the melts from AuPd capsules, and/or in a different speciation of volatile components will be the subjects of further investigation. Whereas the oxidation state in the C–Pt capsules is probably appropriate for average mantle, the somewhat oxidized mantle wedge above subduction zones might be better represented by the experimental configuration with Au80Pd20 capsules.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Genesis of the Epi ankaramite suite
The most significant difference between the experimental ultra-calcic melts and the Epi parental melt is the significantly higher MgO contents in the experimental melts. Fractionation of olivine would conserve the high CaO/Al2O3 values, increase CaO (and Al2O3) contents and decrease MgO contents and XMg values of the liquid. Fractionation models were tested for olivine, clinopyroxene and Cr-spinel/magnetite starting from various experimental melt compositions. Experimental melts with residual clinopyroxene yielded fits with an R2 of 0·2–1·2, but those without clinopyroxene in the residue did not reproduce the Epi ankaramite parental magma (R2 > 2·5). The fit results (two examples in Table 8) show the following:
- starting from the relatively reduced experimental melts from Pt–C capsules, and from melt compositions formed at low to moderate XCO2, fractionation of 4–6 wt % olivine and about 1 wt % of magnetite produces melts of the Epi parent composition. If clinopyroxene fractionation or accumulation is allowed for, fit residuals do not significantly improve, resulting in 1–2 wt % clinopyroxene accumulation.
- Starting from the less reduced melts from AuPd capsules, fractionation of 4–6 wt % olivine and accumulation of 3–4 wt % magnetite result in melts close to the Epi parent composition. When allowing for clinopyroxene variation, accumulation of 3–5 wt % does improve the fit quality slightly.
- Some Cr-spinel fractionation would be necessary to account for the difference in Cr2O3 concentration between experimental melts and the parental Epi magma. However, the small excess of Cr in the experimental melts rather reflects fairly reducing conditions in the experiments than natural conditions and is thus regarded as an ‘experimental artefact’.
|
The fractionation of olivine during magma ascent is expected for melts with dissolved CO2 and H2O, high MgO and high normative olivine contents. When the magma ascends from 15 or 20 kbar, CO2 solubility in the magma decreases markedly around 10 kbar. As a consequence, C–H–O fluid dominated by CO2 is degassed. The olivine saturation surface therefore moves towards lower MgO and normative olivine contents, and olivine is precipitated at the fluid-saturated liquidus (e.g. Taylor & Green, 1987
; Edgar & Vukadinovic, 1992).
The mass balance calculations, which relate our experimental melts to the Epi parent, yield good fits within the uncertainty of choices of both the parental Epi magma and the refractory peridotite residue. We consider that our experimental melts at low to moderate XCO2 fractions are suitable parents for the ankaramite suites in the Vanuatu arc. The major implication is that primitive parents to the Vanuatu arc ankaramite suites were not ultra-calcic melts in terms of CaO contents, but melts high in MgO and CaO/Al2O3. Subsequent fractionation of olivine, as a result of pressure decrease and degassing of CO2, has increased CaO concentrations to values at which clinopyroxene precipitation reverses this trend (Barsdell, 1988; Barsdell & Berry, 1990; Eggins, 1993). The CaO/Al2O3 values observed in the ankaramites of the Vanuatu arc are not extreme in terms of our experimental range and lie at the lower end of the experimentally obtained melts.
A model source (and residue) for hypersthene-normative ankaramitic magmas or tuning the CaO/Al2O3 ratio of the source
The compositions of minerals equilibrated with the experimental ultra-calcic melts are characteristic for refractory harzburgitic mantle. Low Al2O3 in ortho- and clinopyroxene and low Na2O contents in clinopyroxene (if residual) are consistent with a mantle that underwent prior melt extraction; high XCr in spinels is also typical of refractory mantle. Our experiments do not constrain the modal composition of the source and thus we calculate a suitable source.
The most favourable way to generate high CaO/Al2O3 residues from first-stage melting of fertile lherzolite is to extract a moderate amount (6–18%) of melt with low CaO/Al2O3. Melts of this character are typical of low pressures and are derived from plagioclase lherzolite. A likely source (and residue) for ultra-calcic melts is exemplified in Table 9. We calculated possible residues after a first-stage melt extraction by subtracting published experimental melt compositions for well-studied mantle compositions (Hawaiian Pyrolite, HPy, Jaques & Green, 1980; Falloon et al., 1988; MM3, Baker & Stolper, 1994; Hirschmann et al., 1998; KLB-1, Hirose & Kushiro, 1993; garnet lherzolite PHN1611, Kushiro, 1996; Tinaquillo Lherzolite, Tql, Jaques & Green, 1980; Falloon et al., 1988, 1999) employing reported melt fractions that leave at least a few percent clinopyroxene in this first-stage residue (Fig. 9). Although we recognize that there are some problems in some experimental datasets of liquid composition and reacting bulk composition [as illustrated in Fig. 2 and discussed by Hirschmann et al. (1998) and Falloon et al. (1999)], a clear picture emerges from Fig. 9. Subtracting relatively low-pressure melts from fertile lherzolites leads to a strong increase of CaO/Al2O3 in the residue. The CaO/Al2O3 values of the resulting, somewhat refractory mantle would range from 1·2 to 1·8 for the respective compositions MM3, Tql, HPy and KLB-1, and to even higher values for PHN1611. These ratios are achieved by batch melting; fractional melting would increase the resulting CaO/Al2O3 as, at the relevant conditions, Al2O3 partitions more strongly into the melt than CaO (Hirose & Kushiro, 1993). Phase assemblages (ol + opx + cpx + sp + melt or ol + opx + sp + melt) from our experiments were then fitted to the first-stage residues. The best fit of our experimental phase compositions was obtained for the residue RS, which has a CaO/Al2O3 of 1·52, and results from Hawaiian Pyrolite with a 5 kbar, 1200°C melt (F = 17·5%, Jaques & Green, 1980; Falloon et al., 1988) extracted. Second-stage melting of this residue yields melt fractions of 4–5% for melts with CaO/Al2O3 near 1·3 and an olivine–orthopyroxene– clinopyroxene–spinel residue. Complete melting of clinopyroxene in the refractory source RS occurs at F = 10–12% for melts with CaO/Al2O3 near 1·5 (Table 9).
|
|
The evolution of melt compositions in experimental studies on fertile mantle is illustrated in Figs 1 and 2. In each study, maximum CaO/Al2O3 in melts is reached at the elimination of residual clinopyroxene and melts then evolve with near-constant or decreasing CaO/Al2O3 (Fig. 2) along the olivine + orthopyroxene + spinel cotectic. In projection from olivine, this cotectic represents an almost straight line into the orthopyroxene compositional field (Figs 7 and 9). Source compositions and their derivative melts in equilibrium with harzburgite residue lie along this line passing through the bulk composition and the composition of residual orthopyroxene (Figs 1 and 7). This line defines the maximum CaO/Al2O3 for partial melts from a particular composition; liquids with lherzolite residue lie below the prolongation of this line (i.e. away from the Di apex). Thus, in projection from olivine, a mantle source composition for ultra-calcic melts must lie on or at higher normative diopside contents than a line from orthopyroxene to the experimental melt compositions. The XCr of the source composition must be greater than those of experimental fertile compositions (e.g. HPy, MPy, MM3, KLB-1, PHN1611) so that residual spinel at clinopyroxene elimination has XCr > 0·7.
If only melt extraction is considered as a means of ‘tuning’ the source composition, then modal clinopyroxene must be low in suitable sources, i.e. typically between 2 and 6 wt %, and complete melting of clinopyroxene requires melt fractions only slightly larger. When melting occurs up to the clinopyroxene-out curve, melt compositions necessarily inherit a CaO/Al2O3 close to the bulk CaO/Al2O3 and melts acquire the complete inventory of the more incompatible elements previously concentrated in clinopyroxene. Residual spinel will increase CaO/Al2O3 in the melt slightly and the most extreme ratios might be obtained at small melt fractions and high Cr2O3 content in the residue (and thus high amounts of Cr-rich spinels). In the model under discussion, the highest possible CaO/Al2O3 in a melt depends on the amount and mode of previous melt extraction from the mantle source.
In Fig. 9, we calculate a range of such melt extraction models, seeking to generate residues that lie on or above the line joining residual orthopyroxene and the ultra-calcic melts (e.g. the Epi parent magma). The figure demonstrates that appropriate residues are achieved if the initial fertile lherzolite already has relatively high normative diopside and CaO/Al2O3. Appropriate residues with sufficiently high CaO/Al2O3 and XCr from mantle with subchondritic or chondritic CaO/Al2O3 values are difficult to reach.
Previous experiments on the genesis of melts with high CaO/Al2O3
Our experimental methods have established that island-arc ankaramites characterized by CaO/Al2O3 up to 
1·5 do not require an orthopyroxene-absent wehrlitic or pyroxenitic source but can be derived from a lherzolitic source, leaving residual refractory clinopyroxene-bearing or clinopyroxene-free harzburgite. Conditions near 15 kbar and temperatures of 1300–1360°C are appropriate in the presence of some dissolved CO2 and H2O, which lower the liquidus temperatures. In dry systems, temperatures of 1400°C are necessary to produce ultra-calcic melts with a CaO/Al2O3 of 1·5 from refractory lherzolite.
As an alternative to a refractory mantle source, ultra-calcic melts were also produced by melting of wehrlitic sources at high melt fractions (Pickering-Witter & Johnston, 2000; Kogiso & Hirschmann, 2001; Schwab & Johnston, 2001; see Fig. 10). In fact, any wehrlitic source must produce melts that will, with high degrees of melting, move towards the diopside apex in Fig. 10 and have a maximum CaO/Al2O3 close to the clinopyroxene composition in the starting material. The composition of the first melt to form from a given wehrlite depends on the exact composition of the clinopyroxene in a two-phase starting material or on the exact composition of clinopyroxene + minor phases (spinel, small amounts of orthopyroxene) in a multi-phase starting material. In the study of Kogiso & Hirschmann (2001), the absence of orthopyroxene in combination with moderate amounts of Na2O and Al2O3 in the clinopyroxene (e.g. 0·8 and 5·6 wt %, respectively) resulted in nepheline-normative melts (Fig. 10). Melts with CaO/Al2O3 > 1·1 are reached near 1350°C at melt fractions around 30%. Pickering-Witter & Johnston (2000) and Schwab & Johnston (2001) both studied the melting systematics of modally variable peridotites. In both studies, one wehrlitic starting material contained 7 wt % orthopyroxene, which immediately melted out (and is partly accommodated as clinoenstatite component in clinopyroxene). Nevertheless, the presence of orthopyroxene in the starting material shifted initial melt compositions towards the Opx/Qz apex in Fig. 10. The 7 wt % orthopyroxene combined with a clinopyroxene containing 1·5 and 6·4 wt % Na2O and Al2O, respectively, results in a melting trend towards the diopside apex, which is slightly nepheline-normative (FER-B, Pickering-Witter & Johnston, 2000), and ultra-calcic melts occur near 1350°C at melt fractions around 30%. Combination of 7 wt % orthopyroxene with a refractory clinopyroxene having 0·5 and 3·1 wt % Na2O and Al2O (INT-B, Schwab & Johnston, 2001) results in hypersthene-normative melts, which become ultra-calcic near 1350°C at melt fractions of 10–15%, melts with CaO/Al2O3 near 1·5 are reached at 1380°C and melt fractions of 20%. Finally, it should be pointed out that this starting material (INT-B, ol:opx:cpx:sp = 50:7:40:3) is made of refractory mantle phases. As a result of the high normative orthopyroxene content in the first melt of this material, its melting trajectory, projected from olivine, passes through the projection points representing our experimental melt compositions.
|
, OLCPX1; 
, OLCPX2; 
, Pyrox2B). x, FER-B, Pickering-Witter & Johnston (2000)All of the experimental studies on wehrlite sources produced ultra-calcic melts. This is a mere consequence of the starting compositions, which have little or no orthopyroxene and CaO/Al2O3 ranging from 1·7 to 3·4. Melt compositions evolve along the olivine + clinopyroxene cotectic (Fig. 10) towards the bulk CaO/Al2O3 at high melt fractions. However, only the melt compositions of Schwab & Johnston (2001)
at melt fractions near 20% are suitable parents for hypersthene-normative melts. Their starting material is wehrlitic, but contains refractory orthopyroxene and clinopyroxene and has a CaO/Al2O3 of 3·4.
Nepheline-normative ultra-calcic melts
As pointed out above, ultra-calcic melts can be subdivided into a hypersthene- and a nepheline-normative group (Schiano et al., 2000; Kogiso & Hirschmann, 2001), the former including the Epi suite. The nepheline-normative suite, which occurs only in the arc setting, has not only low SiO2 and high total alkalis, but also lower MgO contents, higher CaO and Al2O3, and lower CaO/Al2O3 than the hypersthene-normative group. XMg values of nepheline-normative ultra-calcic magmas are generally less than 0·72 whereas hypersthene-normative ultra-calcic magmas range to 0·79.
Generally, the alkali-rich character of the nepheline-normative suite is incompatible with the refractory nature of our proposed source. The low SiO2 contents of the nepheline-normative melts are also incompatible with saturation in orthopyroxene. In addition, the XMg values of the nepheline-normative suite preclude equilibration with refractory mantle minerals. The nepheline-normative suite is limited to the arc environment and it is possible that slab-derived fluids add alkalis to the source region. However, such fluids are not only Na2O- and K2O-rich but also SiO2-rich and thus are unsuitable for generating low-SiO2 magmas. We thus suggest that the group of nepheline-normative ultra-calcic melts is generated through a process different from second-stage melting of refractory mantle. A suitable source could involve significant addition of a mobile component (fluid or carbonatite melt) and metasomatic phases (e.g. amphibole or/and phlogopite), or a possibly wehrlitic cumulate (with significantly lower XMg than refractory mantle) different from clinopyroxene-bearing refractory mantle, or a combination of both (Medard et al., 2002).
A model for hypersthene-normative ultra-calcic melts
In the model discussed, hypersthene-normative ultra-calcic melts are produced from refractory lherzolite mantle at low melt fractions. In any major magmatic environment, i.e. in mid-ocean ridge, arc and ocean-island settings, refractory mantle is readily available. The highest bulk CaO/Al2O3, and thus the most suitable preconditioning of the mantle with respect to generation of ultra-calcic magmas, would occur in a refractory mantle containing small amounts of clinopyroxene. The most extreme ultra-calcic melts would represent melt fractions of less than a few percent from this latter source. In the volatile-bearing experiments we have arbitrarily used a bulk composition with CaO/Al2O3 = 1·35–1·39, yielding melt compositions with CaO/Al2O3 up to 1·57. In our dry experiments, the bulk CaO/Al2O3 was varied from 1·35 to 1·80, yielding melts with a maximum CaO/Al2O3 of 1·48. The latter melts still coexist with cpx + opx + ol + sp. This study does not cover the entire range of CaO/Al2O3 values observed, i.e. it does not include the melt inclusions from the Lau back-arc basin (Kamenetsky et al. 1997). However, extreme CaO/Al2O3 values could be reached by fractional melting: extraction of 10% first-stage melt from fertile mantle (CaO/Al2O3 0·9, 10 wt % clinopyroxene) leaves a mantle residue with CaO/Al2O3 = 1·2–1·5 and 5 wt % clinopyroxene. Further melting to exhaustion of clinopyroxene yields melts with CaO/Al2O3 up to 1·6 at melt fractions of 7–10%. Second-stage melting with extraction of 3–6% melt leaves some clinopyroxene (2 wt %) in the residue, which would reach a bulk CaO/Al2O3 of 2·0–3·5 (Table 9). Melting out the rest of clinopyroxene then would result in melts with CaO/Al2O3 above those observed in nature.
The natural limit to such a fractional melting process is the temperature necessary to produce melts with extreme CaO/Al2O3. Regardless of their source, the production of melts with CaO/Al2O3 values characteristic of ultra-calcic melts requires temperatures around 1350°C for CaO/Al2O3 near 1·3 and almost 1400°C for a CaO/Al2O3 of 1·5–1·6 [i.e. refractory peridotite in this study, wehrlite in the studies of Pickering-Witter & Johnston (2000), Kogiso & Hirschmann (2001) and Schwab & Johnston (2001)]. Addition of realistic amounts of volatiles diminishes these temperatures by only 20–30°C. The very existence of alkali-poor, hypersthene-normative ultra-calcic magmas and melt inclusions demonstrates that such temperatures are occasionally reached in all major mantle melting environments.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This study demonstrates that ‘preconditioning’ of a mantle source by melt extraction allows generation of magmas with high CaO/Al2O3 in equilibrium with olivine + orthopyroxene + spinel. Whatever the source CaO/Al2O3 in a refractory lherzolite, it will be inherited by a melt that is generated at temperatures at or above the clinopyroxene-out boundary. Residual spinel and orthopyroxene produce only a slight change to CaO/Al2O3 in the melt. We propose that the primitive parents of hypersthene-normative ultra-calcic magmas do not have high CaO contents, but have high CaO/Al2O3 and high MgO contents. Once such a parental magma is generated, olivine fractionation will increase CaO contents. SiO2 contents of the parental melts are already high and the hypersthene-normative character will be increased by further olivine fractionation.
Most of the experiments were conducted with dissolved CO2 + H2O. If a fluid was added from a subducting slab, it would also add some minor incompatible elements to the source and could account, for example, for slightly higher K2O and Na2O contents than could be expected to derive from a depleted mantle source. Alternatively, CO2 and H2O might be added to the source through a dolomitic carbonatite melt formed in the wedge environment, which also leads to a significant increase of CaO/Al2O3 in the source (Green et al., 2004). Nevertheless, dry olivine + orthopyroxene-saturated melts are also suitable parents for ultra-calcic melts from mid-ocean ridges and ocean islands, as our dry experiments resulted in ultra-calcic melts with a CaO/Al2O3 to 1·48 at 1400°C. Higher temperatures and exhaustion of clinopyroxene would have caused an even higher CaO/Al2O3 for these melts; on the other hand, small amounts of CO2 and H2O as present in intraplate environments would slightly lower the temperatures required.
Our results do not constrain whether the first-stage melt extraction(s) happened immediately before the event producing the ultra-calcic melts or billions of years before. The first-stage melt extraction could occur in the same melting cycle and the ankaramites could simply be the final product of a fractional melting process. Alternatively, the first melting event could be completely unrelated to generation of the ankaramites and the diversity of magmas in a given setting would then argue for diversity of peridotitic source compositions in the mantle.
|
ACKNOWLEDGEMENTS |
|---|
The early stage of this work greatly profited from animated discussions and e-mail exchanges with Pierre Schiano, Hugh O'Neill and Marc Hirschmann. We would also like to thank I. A. Nicholls and R. C. Price for their constructive reviews, and M. Pertermann for comments. M.S. acknowledges RSES, ANU, for a fellowship supporting his sabbatical leave at ANU during which the experiments of this study were performed. Thanks also go to all members of the experimental group at RSES for support and discussions.
|
FOOTNOTES |
|---|
* Corresponding author. Telephone: xx41-1-632-7988. Fax: xx41-1-632-1088. E-mail: max.schmidt{at}erdw.ethz.ch
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL TECHNIQUES AND...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Baker, M. B. & Stolper, E. M. (1994). Determining the composition of high-pressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta 58, 2811–2827.[CrossRef][Web of Science]
Barsdell, M. (1988). Petrology and petrogenesis of clinopyroxene-rich tholeiitic lavas, Merelava volcano, New Hebrides. Journal of Petrology 29, 927–964.
Barsdell, M. & Berry, R. F. (1990). Origin and evolution of primitive island arc ankaramites from Western Epi, New Hebrides. Journal of Petrology 31, 747–777.
Berry, A. J. & O'Neill, H. St. C. (2000). The oxidation state of chromium in silicate glasses as a function of oxygen fugacity, composition, temperature, and pressure. Goldschmidt 2000. Journal of Conference Abstracts 5(2), 214.
Brey, G. & Green, D. H. (1977). Systematic study of liquidus phase relations in olivine melilitite + H2O + CO2 at high pressures and petrogenesis of an olivine melilitite magma. Contributions to Mineralogy and Petrology 61, 141–162.[CrossRef][Web of Science]
Cherniak, D. J. (2001). Pb diffusion in Cr diopside, augite, and enstatite, and consideration of the dependence of cation diffusion in pyroxene on oxygen fugacity. Chemical Geology 177, 381–397.[CrossRef][Web of Science]
Della Pasqua, F. N. & Varne, R. (1997). Primitive ankaramitic magmas in volcanic arcs: a melt-inclusion approach. Canadian Mineralogist 35, 291–312.
Edgar, A. D. & Vukadinovic, D. (1992). Implications of experimental petrology to the evolution of ultrapotassic rocks. Lithos 28, 205–220.[CrossRef][Web of Science]
Eggins, S. M. (1993). Origin and differentiation of picritic are magmas, Ambae (Aoba), New Hebrides. Contributions to Mineralogy and Petrology 114, 79–100.[CrossRef][Web of Science]
Falloon, T. J., Green, D. H., Hatton, C. J. & Harris, K. L. (1988). Anhydrous partial melting of a fertile and depleted peridotite from 2 to 30 kb and application to basalt petrogenesis. Journal of Petrology 29, 1257–1282.
Falloon, T. J., Green, D. H., Danyushevsky, L. V. & Faul, U. H. (1999). Peridotite melting at 1·0 and 1·5 GPa: an experimental evaluation of techniques using diamond aggregates and mineral mixes for determination of near-solidus melts. Journal of Petrology 40, 1343–1375.[CrossRef][Web of Science]
Green, D. H. & Falloon, T. J. (1998). Pyrolite: a Ringwood concept and its current expression. In: Jackson, I. (ed.) The Earth's Mantle. Cambridge: Cambridge University Press, pp. 311–378.
Green, D. H., Schmidt, M. W. & Hibberson, W. O. (2004). Island arc ankaramites: primitive melts from fluxed refractory lherzolite mantle. Journal of Petrology (in press).
Hirose, K. & Kawamoto, T. (1995). Hydrous partial melting of lherzolite at 1 GPa: the effect of H2O on the genesis of basaltic magmas. Earth and Planetary Science Letters 133, 463–473.[CrossRef][Web of Science]
Hirose, K. & Kushiro, I. (1993). Partial melting of dry peridotites at high pressures: determination of compositions of melts segregated from peridotite using aggregates of diamond. Earth and Planetary Science Letters 114, 477–489.[CrossRef][Web of Science]
Hirose, K. & Kushiro, I. (1998). The effect of melt segregation on polybaric mantle melting: estimation from the incremental melting experiments. Physics of the Earth and Planetary Interiors 107, 111–118.[CrossRef][Web of Science]
Hirschmann, M. M., Baker, M. B. & Stolper, E. M. (1998). The effect of alkalis on the silica content of mantle-derived melts. Geochimica et Cosmochimica Acta 62, 883–902.[CrossRef][Web of Science]
Jaques, A. L. & Green, D. H. (1980). Anhydrous melting of peridotite at 0–15 kb pressure and the genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology 73, 287–310.[CrossRef][Web of Science]
Kamenetsky, V. S., Crawford, A. J., Eggins, St. M. & Mühe, R. (1997). Phenocryst and melt inclusion chemistry of near-axis seamounts, Valu Fa Ridge, Lau Basin: insight into mantle wedge melting and the addition of subduction components. Earth and Planetary Science Letters 151, 205–223.[CrossRef][Web of Science]
Kamenetsky, V. S., Eggins, St. M., Crawford, A. J., Green, D. H., Gasparon, M. & Falloon, T. J. (1998). Calcic melt inclusions in primitive olivine at 43°N MAR: evidence for melt–rock reaction/melting involving clinopyroxene-rich lithologies during MORB generation. Earth and Planetary Science Letters 160, 115–132.[CrossRef][Web of Science]
Kogiso, T. & Hirschmann, M. M. (2001). Experimental study of clinopyroxenite partial melting and the origin of ultra-calcic melt inclusions. Contributions to Mineralogy and Petrology 142, 347–360.[Web of Science]
Kushiro, I. (1969). The system Forsterite–Diopside–Silica with and without water at high pressures. American Journal of Science 267, 269–294.
Kushiro, I. (1996). Partial melting of a fertile mantle peridotite at high pressures: an experimental study using aggregates of diamond. In: Basu, A. & Hart, S. (eds) Earth Processes: Reading the Isotopic Code. Geophysical Monograph, American Geophysical Union 95, 109–122.
Liu, X. & O'Neill, H. St. C. (2002). Accurate determination of the solidus of simplified spinel lherzolite in the system CaO–MgO– Al2O3–SiO2 (CMAS) at 11 kbar: traditional and new experimental techniques. EMPG IX. Journal of Conference Abstracts 7(1), 67.
Medard, E., Schmidt, M. W. & Schiano, P. (2002). High-Ca primitive liquids: partial melts of clinopyroxene-rich cumulates? Journal of Conference Abstracts 7, 70.
Nickel, K. G. & Brey, G. P. (1984). Subsolidus orthopyroxene– clinopyroxene systematics in the system CaO–MgO–SiO2 to 60 kbar—a re-evaluation of the regular solution model. Contributions to Mineralogy and Petrology 87, 35–42.[CrossRef][Web of Science]
Pickering-Witter, J. & Johnston, A. D. (2000). The effects of variable bulk composition on the melting systematics of fertile peridotitic assemblages. Contributions to Mineralogy and Petrology 140, 190–211.[CrossRef][Web of Science]
Schiano, P., Eiler, J. M., Hutcheon, I. D. & Stolper, E. M. (2000). Primitive CaO-rich, silica-undersaturated melts in island arcs: evidence for the involvement of clinopyroxene-rich lithologies in the petrogenesis of arc magmas. Geochemistry, Geophysics, Geosystems 1, Paper number 1999G000032.
Schwab, B. E. & Johnston, A. D. (2001). Melting systematics of modally variable, compositionally intermediate peridotites and the effects of mineral fertility. Journal of Petrology 42, 1789–1811.
Sigurdsson, I. A., Steinthorsson, S. & Grönvold, K. (2000). Calcium-rich melt inclusions in Cr-spinels from Borgarhraun, northern Iceland. Earth and Planetary Science Letters 183, 15–26.[CrossRef][Web of Science]
Slater, L., McKenzie, D., Grönvold, K. & Shimizu, N. (2001). Melt generation and movement beneath Theistareykir, NE Iceland. Journal of Petrology 42, 321–354.
Sobolev, A. V. & Chaussidon, M. (1996). H2O concentrations in primary melts from supra-subduction zones and mid-ocean ridges: implications for H2O storage and recycling in the mantle. Earth and Planetary Science Letters 137, 45–55.[CrossRef][Web of Science]
Sobolev, A. V. & Danyushevsky, L. V. (1994). Petrology and geochemistry of boninites from the north termination of the Tonga Trench: constraints on the generation conditions of primary high-Ca boninite magmas. Journal of Petrology 35, 1183–1211.
Suyehiro, K., Takahashi, N., Ariie, Y., Yokoi, Y., Hino, R., Shinohara, M., Kanazawa, T., Hirata, N., Tokuyama, H. & Taira, A. (1996). Continental crust, crustal underplating, and low-Q upper mantle beneath an oceanic island arc. Science 272, 390–392.[Abstract]
Takahashi, E., Shimazaki, Y., Tsuzaki, Y. & Yoshida, H. (1993). Melting study of peridotite KLB-1 to 6·5 GPa and the origin of basaltic magmas. Philosophical Transactions of the Royal Society of London, Series A 342, 105–120.
Taylor, W. R. & Green, D. H. (1987). The petrogenetic role of methane: effect on liquidus phase relations and the solubility mechanism of reduced C–H volatiles. In: Mysen, B. O. (ed.) Magmatic Processes and Physiochemical Principles. Geochemical Society Special Publication 1, 121–138.
Ulmer, P. (1989). The dependence of the Fe2+–Mg cation-partitioning between olivine and basaltic liquid on pressure, temperature and composition. An experimental study to 30 kbar. Contributions to Mineralogy and Petrology 101, 261–273.[CrossRef][Web of Science]
Yoder, H. S. & Tilley, C. F. (1962). Origin of basaltic magmas: an experimental study of natural and synthetic rock systems. Journal of Petrology 3, 342–532.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Georgiev, P. Marchev, C. A. Heinrich, A. Von Quadt, I. Peytcheva, and P. Manetti Origin of Nepheline-normative High-K Ankaramites and the Evolution of Eastern Srednogorie Arc in SE Europe J. Petrology, October 1, 2009; 50(10): 1899 - 1933. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. D. Putirka Thermometers and Barometers for Volcanic Systems Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 61 - 120. [Full Text] [PDF]
|
|
|
|
|
|
E. MEDARD, M. W. SCHMIDT, P. SCHIANO, and L. OTTOLINI Melting of Amphibole-bearing Wehrlites: an Experimental Study on the Origin of Ultra-calcic Nepheline-normative Melts J. Petrology, March 1, 2006; 47(3): 481 - 504. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||