The Composition of Near-solidus Partial Melts of Fertile Peridotite at 1 and 1·5 GPa: Implications for the Petrogenesis of MORB
1ARC Centre of Excellence in Ore Deposits and School of Earth Sciences, University of Tasmania, Private Bag 79, Hobart, Tasmania 7001, Australia
2Research School of Earth Sciences, The Australian National University, Canberra, Act 0200, Australia
RECEIVED MAY 14, 2007; ACCEPTED FEBRUARY 12, 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
We have determined the near-solidus melt compositions for peridotite MM-3, a suitable composition for the production of mid-ocean ridge basalt (MORB) by decompression partial melting, at 1 and 1·5 GPa. At 1 GPa the MM-3 composition has a subsolidus plagioclase-bearing spinel lherzolite assemblage, and a solidus at
1270°C. At only 5°C above the solidus, 4% melt is present as a result of almost complete melting of plagioclase. This melting behaviour in plagioclase lherzolite is predicted from simple systems and previous experimental work. The persistence of plagioclase to > 0·8 GPa is strongly dependent on bulk-rock CaO/Na2O and normative plagioclase content in the peridotite. At 1·5 GPa the MM-3 composition has a subsolidus spinel lherzolite assemblage, and a solidus at 1350°C. We have determined a near-solidus melt composition at 2% melting within 10°C of the solidus. Near-solidus melts at both 1 and 1·5 GPa are nepheline normative, and have low normative diopside contents; also they have the highest TiO2, Al2O3 and Na2O, and the lowest FeO and Cr2O3 contents compared with higher degree partial melts. Comparison of these near-solidus melts with primitive MORB glasses, which lie in the olivine-only field of crystallization at low pressure, indicate that petrogenetic models involving aggregation of near-fractional melts formed during melting at pressures of 1·5 GPa or less are unlikely to be correct. In this study we use an experimental approach that utilizes sintered oxide mix starting materials and peridotite reaction experiments. We also examine some recent studies using an alternative approach of melt migration into, and entrapment within ‘melt traps’ (olivine, diamond, vitreous carbon) and discuss optimal procedures for this method. KEY WORDS: experimental petrology; mantle melting; near-solidus; fertile peridotite; MORB
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The determination of peridotite partial melt compositions places an important constraint on models of magma genesis and geodynamics (Green & Ringwood, 1967a
; Kushiro et al., 1968; Jaques & Green, 1979, 1980; Stolper, 1980; Takahashi & Kushiro, 1983; Baker & Stolper, 1994; Green & Falloon, 1998). In recent years one important focus of experimental studies has been the determination of near-solidus partial melts for a range of peridotite compositions to help calibrate mantle melting models (e.g. pMELTS, Ghiorso et al., 2002). Although conceptually simple, the determination of melt composition in peridotite at low to moderate degrees of melting is experimentally difficult and different approaches have resulted in some controversy, which is difficult for the non-specialist to evaluate. From the earliest studies quoted above, two methods have developed, one using synthesized sintered oxide mixes and glasses of selected composition, and the other using crushed natural rocks or mixtures of separated, crushed and remixed minerals from lherzolites. A third approach, that of increasing chemical complexity of chemical systems from MgO + SiO2, through CaO + MgO + Al2O3 + SiO2 (CMAS) to addition of Na2O, FeO, or Cr2O3 to CMAS, is similar to the ‘sintered oxide’ approach above and focuses on varying bulk compositions around multi-phase cotectics, determining both mineral phases and liquid composition (O’Hara & Yoder, 1967; Walter & Presnall, 1994).
A major experimental difficulty was recognized (Jaques & Green, 1979, 1980) in the modification of melt compositions during quenching by outgrowth of residual phases, ensuring that pools of glass analysed by electron microprobe gave quench-modified disequilibrium compositions and not the composition of the equilibrium melt at the experimental P, T. A first step to solve this problem was to increase the proportion of melt in the composition by preparing olivine-depleted compositions (i.e. pyrolite–40% olivine). However, quench modification of glass was still apparent (Jaques & Green, 1979). The quench modification of glass in hydrous peridotite melting studies had also been demonstrated by preparing compositions of the analysed glass and showing that these compositions had liquidus temperatures and liquidus phases incompatible with those of the peridotite melting experiment (Nicholls, 1974; Green, 1976). These studies demonstrated both the need for ‘forward’ (peridotite melting) and ‘reversal’ (‘basalt liquidus’) experiments (i.e. an iterative approach). Stolper (1980) used an innovative method of enclosing a natural basaltic glass fragment within peridotite, thus creating a large pool of melt surrounded by peridotite. Later work built on this using an iteration of forward melting experiments, estimation of melt composition and preparation of this composition from oxide mixes. Falloon et al. (1988, 2001) used a ‘sandwich’ or layered experimental charge, in which the composition of the best estimates of the liquid composition from the forward experiment [in particular, the results of Jaques & Green (1980)] were prepared as sintered oxides or glasses and run as layers against or between the original peridotite composition. This approach was used systematically to obtain melt compositions from ‘enriched’ (Hawaiian Pyrolite), ‘fertile’ MOR (mid-ocean ridge) Pyrolite, and depleted lherzolite (Tinaquillo Lherzolite) using sintered oxide mixes as starting materials and a two-step approach of initial estimation of melt composition in the direct peridotite melting experiment, followed by a ‘sandwich’ of estimated melt + peridotite (Falloon et al., 1988, 2001). This method yields large melt pools thus avoiding quench modification of glass, and the use of sintered oxide or glass starting materials avoids persistence of relict minerals or disequilibrium melting of natural minerals. Recent papers by Dasgupta & Hirschmann (2007) and Hirschmann & Dasgupta (2007) have provided both theoretical and empirical arguments that (1) estimates of melt compositions, (2) determination of residual phase compositions, and (3) iterative use of the ‘sandwich’ approach are necessary to reliably determine the composition of near-solidus melts of peridotite. In this paper we provide further argument supporting this convergence in methods.
In attempting to test the validity of alternative approaches, we have studied the composition MM-3 (Baker & Stolper, 1994; Baker et al., 1995) a composition closely matching estimates of fertile asthenospheric mantle [i.e. source for modern mid-ocean ridge basalt (MORB)] such as the MOR Pyrolite (Green & Falloon, 1998) or Primitive Modern Mantle (Hart & Zindler, 1986). This composition (MM-3) has been extensively studied using a crushed mineral mix and ‘diamond melt trap’ to determine compositions of near-solidus melts (Baker & Stolper, 1994; Baker et al., 1995). Because the initial data on near-solidus melts from these studies differed so strongly from results on similar compositions (MOR Pyrolite, Tinaquillo Lherzolite) using sintered oxide mixes and iterative ‘sandwich’ approach (Falloon et al., 1988), we conducted an additional investigation of the melting behaviour of this composition using our preferred sintered oxide and iterative ‘sandwich’ approach (Falloon et al., 1997, 1999a). Correction of the composition of the melt initially obtained by Baker et al. (1995) at 1 GPa, 1250–1260°C, by Hirschmann et al. (1998), established agreement that the near-solidus melt at 1 GPa was a nepheline-normative Na-rich composition of alkali olivine basalt affinity (hawaiite–mugearite) but differences between the natural minerals–diamond trap and sintered oxide–iterative ‘sandwich’ approaches remained in higher temperature experiments (Falloon et al., 1999a).
In this paper, we extend the work on MM-3 by confirming the near-solidus melt compositions at 1 GPa and at 1·5 GPa. At 1 GPa, 1250°C, MM-3 is subsolidus lherzolite with both plagioclase and aluminous spinel present but at 1·5 GPa, 1300°C the subsolidus lherzolite assemblage contains only spinel. Using the results of peridotite reaction experiments (layered charges), Falloon et al. (1999a) calculated near-solidus melts at 1 GPa, 1260°C, and 1·5 GPa, 1325°C, in equilibrium with plagioclase lherzolite and spinel lherzolite, respectively. In this paper we test these calculated compositions by a further iteration of the ‘sandwich’ technique.
The recommended approach from Falloon et al. (2001) is to perform direct melting experiments on a fine-grained, sintered oxide mix of the composition of interest, establishing the composition of residual phases, composition of interstitial glass (quench-modified melt) and an estimate (using mass balance) of the true liquid composition and phase proportions. Surprisingly, Dasgupta & Hirschmann (2007) and Hirschmann & Dasgupta (2007) did not begin with the ‘forward melting’ experiment on the target composition [the peridotite or ‘bread’ of Hirschmann & Dasgupta (2007)] to constrain the composition of the initial melt layer [the basalt or ‘meat’ of Hirschmann & Dasgupta (2007)]. The forward experiment, particularly using olivine-depleted compositions (i.e. MOR Pyrolite–40% olivine, etc.; Jaques & Green, 1979, 1980; Green & Falloon 1998), uses modal estimates, element partitioning, including Fe/Mg for olivine/liquid, and techniques of focused and scanning beam microanalysis to estimate the melt composition. This step considerably shortens the number of iterations required to converge on the near-solidus melt, as opposed to starting with liquids far removed from equilibration with the residual mineralogy as illustrated by Hirschmann & Dasgupta (2007). Dasgupta & Hirschmann (2007), in their demonstration example of the iterative sandwich experiment, used a mix of carbonate and oxides for their ‘meat’ layer, which shows a large shift in composition at the first interaction stage. An additional difference, the use of crushed minerals rather than sintered oxide as the ‘bread’ layer, is briefly addressed below.
As the experimental pressures used in this study (1–1·5 GPa) encompass the plagioclase to spinel lherzolite transition for fertile mantle peridotite, the results have important implications for some models of MORB petrogenesis based on simplified model systems (Presnall et al., 1979, 2002). It is important to test whether the results in more complex systems are consistent with the data and models of MORB petrogenesis based on simple system data.
|
EXPERIMENTAL AND ANALYTICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Experimental techniques
Starting compositions
All starting compositions used in the experimental study are presented in Table 1. The starting compositions were prepared from a mixture of analytical grade oxides and carbonates (Ca, Na, K), mixed under acetone in an agate mortar. This mixture was pelletized and sintered overnight (
16–20 h) at 950°C. The required proportion of FeO in the target composition was added as synthetic fayalite, previously prepared from sintered oxides under reducing conditions. The mixture was ground, before storage in glass vials in a 110°C oven.
|
A scanning electron micrograph of the sintered oxide powder is given as Fig. S1 in the Supplementary Data (available for downloading at http://www.petrology.oxfordjournals.org/), and shows single mineral crystals of <1 µm. X-ray diffraction shows the presence of olivine, pyroxenes, plagioclase and spinel mineral species, and very minor amounts of unreacted oxides (periclase, corundum, rutile). Sintering at a temperature of 950°C, we do not expect an equilibrated mineral assemblage but argue that the fine grain size and intimate mixing of phases produces a highly reactive starting material with very low probability of survival of initial phase compositions in the final products, which have grain sizes from 5 to 200 µm (Supplementary Data, Figs S2 and S3).
We have not attempted to verify the composition of our peridotite starting material by direct analysis, as it is very difficult to experimentally melt the peridotite bulk composition without significant loss of oxides and contamination from containers. However, our previous experimental work using the MM-3 starting material (Falloon et al., 1999a) demonstrated very good mass balance for high-degree melting experiments (e.g. run T-4264, which consists of 40% melt plus olivine; Falloon et al., 1999a) and very close matching to the high-temperature, high-degree melting experiments of Baker & Stolper (1994), which would not have been possible if our nominal composition had been significantly incorrect.
Other laboratories have conducted peridotite melting studies using crushed natural peridotite or compositions prepared by first separating minerals from peridotite, then crushing them to very fine grain size and recombining the crushed minerals to give a desired peridotite composition (Baker & Stolper, 1994; Schwab & Johnston 2001; Wasylenki et al., 2003; Dasgupta & Hirschmann 2007). In the early 1960s at the beginning of a period of intensive experimental study of natural rock compositions, including basalts, eclogites and peridotites (Yoder & Tilley, 1962; Green & Ringwood, 1967a, 1967b, 1973), thorough investigation was made of the use of crushed natural rocks, and of mineral separation and fine crushing (e.g. Raheim & Green, 1974). The experience was summarized by Green & Falloon (1998, pp. 321–324) and followed earlier published reasons [by, for example, Green (1976)] for the selection of sintered oxide mixes and glasses as starting minerals.
In the present study we directly compare results from sintered oxide starting materials with those from nominally the same composition (MM-3) run as a crushed mineral mix. We have a number of concerns about the mineral mix approach, as follows.
- Fine crushing to <10 µm, including elutriation to attempt to homogenize grain size, introduces unavoidable contamination from the crushing medium; that is, SiO2 enrichment from agate, Al2O3 enrichment from corundum, etc. The contamination is extremely difficult (Fe) or impossible (SiO2, Al2O3) to remove. Bulk compositions prepared from finely crushed minerals or rocks may differ significantly from the ‘theoretical’ composition (Raheim & Green, 1974).
- Relict cores of starting minerals persist so that analyses of new equilibrium phases must select rims and avoid core overlap. Mass balance and modal proportion calculations or mapping are subject to uncertainty. The presence of mineral zoning and relict cores is usually identified and its effects are minimized by long run times, care in analyses, etc. However, our point is that these problems can and should be avoided. The effect of relict cores is clearly evident in a number of studies utilizing natural mineral mixes [e.g. mineral mixes PHN1611 and HK-66; see discussion by Falloon et al. (1999a)].
- Disequilibrium melting occurs in natural minerals (particularly clinopyroxene) that initially crystallized at high P, low T and are then placed at lower or similar P, but higher T (Doukhan et al., 1993). Once present, particularly if melt traps of diamond, olivine or other aggregate are being used, such disequilibrium melts may persist in the experiment.
Sintered oxide mixes are not difficult to prepare and minor elements (K, Cr, P, Mn, Ni, Ti) can be added with accuracy and precision to 1–2 g mixes at the 0·01% level (i.e. at 0·1 mg in a 1 g mix). Thorough mixing minimizes the ‘nugget’ effect for such minor components and effective doping of mixes with trace elements is achieved using the tetra-ethyl silicate gel approach as a precursor to sintering. The preparation of silicate standards for analytical purposes is a standard technique and with the ready availability of high purity reagents, the experimental petrologist has no good reason to choose the additional problems inherent in the use of crushed mineral mixes. Reluctance to prepare oxide mixes or glasses of chosen composition may also be the reason for rarity of experiments to test and confirm the composition and phase relations of melts obtained in a ‘forward’ melting experiment by preparing a bulk composition of the melt and placing the melt charge (or melt + source composition layer) at the P, T of the forward experiment. If the liquidus temperatures and liquidus phases (including possible reaction relationships at the liquidus) do not match (within ±20°C) those of the forward experiment then there is a problem of quench modification of melt, or disequilibrium melting or charge inhomogeneity in the ‘forward’ experiment. Further discussion of the use of natural mineral mixes in experimental studies is presented in the Supple-mentary Data.
Run assemblies and temperature control
All experiments were performed using standard piston-cylinder techniques in the High Pressure Laboratory, School of Earth Sciences, University of Tasmania (UTas). All experiments used NaCl–Pyrex assemblies with graphite heaters and a W97Re3–W75Re25 (W–Re, Type D) thermocouple (calibrated against the melting point of Au and the e.m.f. of a Type S, Pt–Pt90Rh10 (Pt) thermocouple, at atmospheric pressure in an Ar atmosphere). Temperatures were controlled to within ±1°C of the set point using a Eurotherm type 818 controller. No pressure correction was applied to the thermocouple calibration. Calculated experimental temperatures using the anhydrous olivine liquidus geothermometer of Herzberg & O’Hara (2002) show a very close matching to our nominal temperatures (Table 2). All experiments employed graphite capsules with fired pyrophyllite and alumina spacers, and mullite and alumina surrounds. The thermocouple enters the assembly through a composite two- and four-bore alumina sheath and is shielded from the graphite capsule by a 1 mm alumina disc. The thermocouple junction is formed by crossing the thermocouple wires utilizing the four-bore alumina sheath, which forms the top 5 mm of the alumina thermocouple sheath. All experimental components and starting materials were stored in an oven at 110°C. Experiments were performed using the hot piston-out technique (Johannes et al., 1971). For experiments at 1 GPa a nitrogen gas flow over the thermocouple exit from the end plate of the piston-cylinder apparatus prevented thermocouple oxidation (Walter & Presnall, 1994; Falloon et al., 2001).
|
Analytical techniques
Electron microprobe microanalysis
At the end of each experiment, the entire experimental charge was mounted and sectioned longitudinally before polishing. Experimental phase compositions presented in Table 3 were analysed by wavelength-dispersive microanalysis using a Cameca SX-50 microprobe housed in the Central Science Laboratory, UTas (operating conditions 15 keV, 20–25
A). Counting times for all elements were 20 s for the peak and 10 s for the background on both sides of the peak. We routinely used a beam spot size of 1 µm for analysis. All glass analyses have been normalized to the composition of international glass standard VG-2 (Jarosewich et al., 1980), which was analysed together with the glasses under the same analytical conditions. An average of analyses of the VG-2 glass standard is presented at the end of Table 3. As can be seen from the back-scattered electron (BSE) images presented in the Supplementary Data our experimental techniques and run times have produced well-equilibrated (we use the term ‘well-equilibrated’ where there is melt continuum between crystals, phases are well crystallized and we do not observe compositional zoning in 5–30 µm crystals) unzoned run products. However, because of the small size of many phases (especially spinel), the tabular lath-like morphology of pyroxenes, quench crystallization on primary phases, and the parallel growth of clinopyroxene relative to orthopyroxene can all potentially result in significant chemical variation in the primary phases caused by overlap during electron microprobe analysis. This is a possible alternative explanation for the low-Ca compositions interpreted as metastable unreacted cores in clinopyroxene by Robinson et al. (1998) in experiments that also used fine-grained sintered oxide starting materials. If, because of the nature of sectioning of the experimental charge, a thin tabular otherwise unzoned clinopyroxene is attached to orthopyroxene crystals beneath the analytical surface, then overlap will occur, resulting in low-Ca compositions. We have presented in Table 3 average analyses, selected analyses and in some cases calculated analyses of all primary run products that we believe are free from overlap with other phases.
|
Calculation of melt fraction (F) in peridotite melting and reaction experiments
We employed the method of Falloon et al. (2001
) to calculate the melt fraction (F) for peridotite MM-3 and these values are listed in Table 3. For peridotite reaction experiments we also present the calculated melt fraction (F) using the technique of Robinson et al. (1998) in Table 3. |
EXPERIMENTAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
We undertook several experiments designed to test whether the near-solidus calculated equilibrium melts at 1 GPa and 1·5 GPa of Falloon et al. (1999a
) met the test of liquidus temperature and phase saturation at the designated P, T. The results of our experiments are summarized in Table 2 and phase compositions and mass balance calculations are presented in Table 3.
Near-solidus melt at 1 GPa
In run T-4353 the calculated equilibrium 1 GPa near-solidus melt composition of Falloon et al. (1999a), composition IE-1·0 GPa (Table 1), was reacted with MM-3 at the designated target temperature of 1260°C. Run T-4353 resulted in a well-equilibrated run assemblage of oliv + opx + cpx + plag + liq (Fig. S2m–o). The plagioclase has a composition of An52·3, slightly more sodic than value of An56 for the subsolidus plagioclase (at 1250°C) in the forward melting experiment (run T-4193, Falloon et al., 1999a; see Fig. S2r). The olivine has a composition of Mg-number 89·8, significantly less than the determined subsolidus olivine composition of Mg-number 90·6 (run T-4193, Falloon et al., 1999a). This result indicates that the mix IE-1·0 GPa is not an appropriate composition for the near-solidus melt composition of peridotite MM-3. Although all target phases + liquid are present, the lower anorthite content of plagioclase and lower forsterite content of olivine show that the added liquid (IE-1·0 GPa) is not an exact match for the near-solidus melt of MM-3 at 1260°C, 1 GPa. Using the analysed phase compositions of run T-4353 an excellent mass balance is obtained for the bulk composition of the mixture (40% MM3 + 60% IE-1·0 GPa). However, if we use the liquid composition of the sandwich experiment and its residual phases in a mass balance against the MM-3 composition itself, we obtain a negative value for the liquid proportion, confirming that the composition (IE-1· 0 GPa) is not that of a near-solidus melt of MM-3. The experiment T-4353 shows that the mass of the melt component decreased during run T-4353 (MLi/MLf value of 1·22, Table 2) indicating that the resulting phase equilibria are dominated by the added melt component not by the peridotite MM-3 composition. This result is consistent with the crystallization experiments performed on the IE-1·0 GPa composition (Table 2), which demonstrate that it becomes entirely liquid only between 1270 and 1280°C. At the target temperature for the calculated near-solidus melt (1260°C) the IE-1·0 GPa composition is partially crystallized (Table 3). Based on the variation of crystallization with temperature, the liquidus (i.e. zero crystallization) is estimated to lie at 1271°C at 1 GPa for this composition. The crystallization sequence for the IE-1·0 GPa melt composition at 1 GPa is Plag (An52·2) + Cpx (Mg-number 89·8) at 1270°C, followed by Plag (An52) + Cpx (Mg-number 88·2) + Oliv (Mg-number 88·1) at 1260°C.
The results of T-4353 suggests that the calculated IE-1·0 GPa composition is too enriched in plagioclase components to be an equilibrium near-solidus melt of peridotite MM-3 at 1270°C. To experimentally determine the near-solidus melt composition, we performed a layered reaction experiment, with 13% added melt component (IE-1·0 GPa), resulting in a well-equilibrated run assemblage consisting of oliv + opx + cpx + sp + plag + liq (run T-4377, Table 2, Fig. S2j–l). The plagioclase in run T-4377 is present only in trace amounts and was very difficult to analyse with the electron microprobe because of overlap with other phases. Our best estimate of its composition is An60·6; that is, slightly more calcic than the subsolidus plagioclase of T-4193 (Falloon et al., 1999a). The olivine has a composition of Mg-number 90·8, slightly higher than the determined subsolidus olivine composition of Mg-number 90·6 (run T-4193, Falloon et al., 1999a). Mass-balance calculations (Table 3) on both the bulk composition of the charge and the MM-3 composition confirm trace amounts of plagioclase, 18 ± 1% melting of the charge and 5 ± 2% melting of MM-3 [4 ± 1% using the Robinson et al. (1998) calculation], which indicates virtually total melting of all subsolidus plagioclase (4 ± 1%, Falloon et al., 1999a). Mass-balance calculations indicate that the melt component increased during run T-4377, indicating that melting of the MM-3 composition occurred.
Both runs T-4353 and T-4377 have essentially identical calculated temperatures using the Herzberg & O’Hara (2002) olivine geothermometer (calculations were performed using the software PETROLOG, Danushevsky, 2001) of 1271 and 1272°C. As the phase assemblage for run T-4353 is too fertile for MM-3 at 1271°C and mass balance indicates no involvement of MM-3, whereas T-4377 at 1272°C indicates melting of MM-3, we consider the melt composition in run T-4377 as the closest experimentally determined approximation to a near-solidus melt of peridotite MM-3 (within 5°C of the solidus of MM-3). The true melt composition at the solidus of MM-3 must therefore lie between the composition of the glass in run T-4353 and T-4377. An average of the glass compositions in runs T-4353 and T-4377 (Table 4) gives an excellent mass balance using the subsolidus phase compositions from run T-4193 (Falloon et al., 1999a) with 2% melting (Table 3), and we therefore consider that this composition is the best estimate of the near-solidus melt for peridotite MM-3 at 1 GPa in equilibrium with a plagioclase–spinel lherzolite residue.
|
Our reaction experiments at 1 GPa confirm previous work (Presnall et al., 1979
; Jaques & Green, 1980) indicating that plagioclase will enter the melt within a few degrees above the solidus. In the case of the MM-3 composition, which contains only a minor amount of subsolidus plagioclase, plagioclase melts completely within 5°C above the solidus (solidus 1270°C) forming 4–5% melt. Layered reaction experiments, runs T-4375 and T-4380 (Table 2), were performed at higher nominal and calculated temperatures (1270 and 1295°C, and 1279 and 1297°C, respectively) to determine melt compositions in equilibrium with a spinel lherzolite assemblage at 1 GPa for the MM-3 composition. Both runs resulted in well-equilibrated run assemblages (Tables 2 and 3, Fig. S2g–i, p, q) and represent between 6 and 12% melting of MM-3.
Near-solidus melt at 1·5 GPa
The solidus of MM-3 at 1·5 GPa was bracketed by runs at 1300°C (ol + opx + cpx + sp) and 1350°C (ol + opx + cpx + sp + liq) but the composition of liquid (IE-1·5 GPa, Falloon et al., 1999a) was calculated from a reaction experiment [T-4293; second iteration in terms of Hirschmann & Dasgupta (2007)] at 1325°C, 1·5 GPa, which yielded (ol + opx + cpx + liq). The composition IE-1·5 GPa (Table 1) is a calculated near-solidus melt of MM-3 at 1325°C, 1·5 GPa (Falloon et al., 1999a). To test this composition at 1·5 GPa we performed a mixed reaction experiment using 60 wt % of the IE-1·5 GPa mix at 1325°C, 1·5 GPa (run T-4343, Tables 2 and 3). Run T-4343 resulted in a well-equilibrated run assemblage consisting of oliv + opx + cpx + sp + liq. However, mass-balance calculations (Tables 2 and 3) indicate that the melt phase decreased during reaction (MLi/MLf = 1·62) and thus the run assemblage is a result of crystallization and reaction below the solidus of the MM-3 composition. This is supported by the olivine composition of Mg-number 88·0, significantly more Fe-rich than the determined subsolidus olivine composition at 1·5 GPa (Mg-number 90·6, run T-4260, Falloon et al., 1999a). The composition of the glass in run T-4343 is too fertile to be a near-solidus melt of MM-3. This is confirmed by mass-balance calculations with the MM-3 composition, which failed to achieve positive phase proportions. The results of run T-4343 indicate that the near-solidus melt for MM-3 lies at a temperature >1325°C. Indeed, crystallization experiments on the IE-1·5 GPa composition (Tables 2 and 3) showed that it is a liquid between 1360 and 1370°C. Clinopyroxene was the only liquidus phase for all of the crystallization experiments performed on the IE-1·5 GPa composition.
To experimentally determine the near-solidus melt at 1·5 GPa we performed a layered reaction experiment with 24 wt % of the IE-1·5 GPa composition at a nominal temperature of 1350°C (run T-4352, Tables 2 and 3). Run T-4352 resulted in a well-equilibrated run assemblage consisting of oliv + opx + cpx + sp + liq (Fig. S3i–l). The olivine composition has a Mg-number of 90·7 slightly higher than the subsolidus olivine composition (Mg-number 90·6) and mass-balance calculations indicate that the melt phase increased slightly during the course of reaction (MLi/MLf = 0·94, Table 2). Mass-balance calculations using the MM-3 composition demonstrate positive phase proportions with 2% melting (1·7 ± 0·7%, Table 3). We therefore consider that the glass composition in run T-4352 is a near-solidus melt of peridotite MM-3 composition at 2% melting at a calculated temperature of 1360°C, 10°C above the solidus at 1·5 GPa (1350°C). Layered reaction experiments, runs T-4376 and T-4381, were performed at higher nominal temperatures (1370 and 1385°C, respectively) to determine melt compositions in equilibrium with a spinel lherzolite assemblage at 1·5 GPa for the MM-3 composition. Both runs resulted in well-equilibrated run assemblages (Tables 2 and 3, Fig. S3c–f) and represent between 7 and 14% melting of MM-3 (Table 3).
It is helpful to relate the studies seeking to define the near-solidus melts of MM-3 composition to the modified iterative sandwich approach advocated by Hirschmann & Dasgupta (2007). Hirschmann & Dasgupta (2007) demonstrated using theoretical calculations that there is a rapid convergence towards the true peridotite melt composition utilizing peridotite reaction experiments if the reactant melt composition is a calculated composition based on the results from an initial reaction experiment. In the recommended technique of Falloon et al. (2001) the initial reactant melt is a calculated composition based on the results of direct melting experiments on the peridotite composition of interest. For example, in the study of Falloon et al. (1988) the reactant melt compositions were all calculated melt compositions based on the direct melting study of Jaques & Green (1980). In the present series of experiments on peridotite MM-3 (Falloon et al., 1997, 1999a, this study) the initial reactant melt was based on the direct melting diamond trap experiments of Baker & Stolper (1994) and Baker et al. (1995). Thus we would expect a rapid approach to the true equilibrium melt of MM-3 composition based on the theoretical analysis of Hirschmann & Daguspta (2007). In Fig. 1 we demonstrate how our experiments have indeed converged towards what we believe is the composition of the true near-solidus melt compositions of MM-3 at 1 and 1·5 GPa. In Fig. 1 we have arranged the experimental results from UTas with those from CalTech into a chronological sequence that shows the results of direct melting experiments (Baker et al., 1995; Hirschmann et al., 1998), reaction experiments and calculations (Falloon et al., 1997, 1999a, this study) and how the compositions have changed towards the ‘true’ near-solidus melt composition. The sequence number in Fig. 1 we believe is equivalent to the sequential iterative steps in the theoretical approach of Hirschmann & Dasgupta (2007). Figure 1 demonstrates that the compositions of near-solidus melts at both 1 and 1·5 GPa have converged to an approximately constant composition by sequence number 3 (see caption to Fig. 1 for detailed explanation of compositions plotted and data sources used). Figure 1 also demonstrates systematic changes in melt compositions with sequence number. Initial melt compositions at both 1 and 1·5 GPa have relatively higher SiO2 and Al2O3, and lower TiO2, FeO, Cr2O3 and MgO compared with the true near-solidus melt compositions (black arrows, Fig. 1). In the case of Na2O the trends are less systematic at 1 GPa compared with 1·5 GPa. At 1·5 GPa Na2O remains relatively constant whereas at 1 GPa Na2O is highly variable, similar in behaviour to that predicted by Hirschmann & Gupta (2007). In the case of CaO, there is a decrease in CaO at 1·5 GPa and an increase at 1 GPa. In summary, Fig. 1 demonstrates that it is unlikely that further reaction experiments would produce compositions significantly different from the ‘true’ near-solidus melt compositions presented in Table 4, and shown as black arrows in Fig. 1 and that our experiment techniques have, in accordance with the theoretical analysis of Hirschmann & Dasgputa (2007), successfully determined the compositions of near-solidus melts for a fertile peridotite (MM-3) similar to MOR Pyrolite at 1 and 1·5 GPa.
|
, 1 GPa data; •, 1·5 GPa data. Arrows indicate the near-solidus melt compositions from |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Melting of peridotite MM-3 at 1 and 1·5 GPa
Utilizing the results of this study and those of Falloon et al. (1999a
) we present in Figs 2–4 the compositions of partial melts of peridotite MM-3 from near-solidus to melts in equilibrium with a dunite residue. In Fig. 2 we project the compositions of partial melts into the normative tetrahedron using projections from Olivine (Fig. 2a) and Diopside (Fig. 2b) and use the melt compositions to define melting cotectics appropriate for MM-3 at 1 and 1·5 GPa. We also plot the field for our analyses of the VG-2 glass standard (average presented in Table 3) as a means of demonstrating the effect of analytical uncertainty on the plotted positions of glass compositions within the tetrahedron. As can be seen from Fig. 2, analytical uncertainty does not have a significant effect on the plotted positions of glasses analysed by electron microprobe. In general, the size of the symbols used in Fig. 2 encompasses the complete range of individual analyses that were used to calculate the average glass compositions (Table 3).
|
|
|
, 1 GPa data; Figure 2 shows that the glass compositions in both the projections define smooth melting trends starting from the nepheline normative near-solidus melts determined in this study. In the projection from Olivine (Fig. 2a), at both 1 and 1·5 GPa, the near-solidus melt compositions lie at relatively low normative diopside contents, and as melting progresses melt compositions increase their normative diopside contents, reaching a maximum at the point of clinopyroxene elimination from the residue. After the elimination of clinopyroxene, normative diopside contents of the melts decrease along an oliv + opx + liq cotectic and at the point of orthopyroxene elimination the melt compositions overlie the MM-3 bulk composition in this projection. The maximum normative diopside content in the melting trends moves to higher diopside contents and closer to the Di–Plag–Oliv plane of critical undersaturation with increasing pressure. In the projection from Diopside (Fig. 2b), melts in equilibrium with lherzolite and harzburgite residues at both 1 and 1·5 GPa define continuous smooth trends that terminate at the point of orthopyroxene elimination, when, with further melting, melt compositions lie along an olivine control line through the peridotite MM-3 composition.
In Fig. 3 we compare the melting trends for MM-3 with MOR Pyrolite (a very similar composition to MM-3), a more refractory composition (Tinaquillo Lherzolite) and an enriched composition (Hawaiian Pyrolite). Figure 3 shows that the melting trends from MM-3 are very similar to MOR Pyrolite except that melt compositions extend to higher normative Di in the projection from Olivine (Fig. 3a) because of the higher CaO/Al2O3 value of MM-3. We emphasize that the subsolidus plagioclase compositions at 1 GPa mark the departure points of the melting trends from the plagioclase lherzolite cotectic. The melting trend for Hawaiian Pyrolite crosses the plagioclase lherzolite cotectic defined by K2O-free data towards relatively more Qz-rich compositions as a result of the effect of K2O (2 wt % K2O in near-solidus melts; T. J. Falloon, unpublished data), consistent with the experimental work in simple systems (Conceição & Green, 2000). For K2O-free systems, small differences in bulk Cr-number [100Cr/(Cr + Al)] have no detectable effect on the position of the plagioclase lherzolite cotectic at 1 GPa (Cr-number varies from 6·5 in MOR Pyrolite to 10·3 in MM-3 lherzolite).
In Fig. 4 we plot the major element compositions and calculated melt fraction of the 1 and 1·5 GPa partial melts against calculated temperature. The data in Fig. 3 define smooth trends that demonstrate some important features of isobaric melting at 1 and 1·5 GPa. As is observed by many other experimental studies in both simple and complex systems, near-solidus melts relative to higher temperature melts have the highest TiO2, Al2O3 and Na2O (Fig. 4b, c and g) and the lowest FeO and Cr2O3 contents (Fig. 4d and h). The behaviour of SiO2 and CaO, however, is more complex. Near-solidus melts have relatively high SiO2 (Fig. 4a) contents compared with higher temperature melts in equilibrium with clinopyroxene (although it should be emphasized that the high SiO2 contents are accompanied by high Na2O contents so that the liquids are silica-undersaturated and nepheline-normative). As melting progresses SiO2 contents decrease, reaching a local minimum at the point of clinopyroxene exhaustion. SiO2 contents then increase with continued melting in equilibrium with a harzburgite residue, reaching a maximum at the point of orthopyroxene exhaustion. SiO2 contents then decrease with further melting in equilibrium with a dunite residue, decreasing to the point of 100% melting, where the SiO2 content of the bulk peridotite is obtained. In contrast to SiO2, near-solidus melts have relatively low CaO (Fig. 4f) contents compared with higher temperature melts in equilibrium with clinopyroxene. As melting progresses CaO contents increase, reaching a maximum at the point of clinopyroxene exhaustion. CaO contents then decrease continuously with further melting, with no inflexion point at the point of orthopyroxene exhaustion.
In Fig. 4e we plot the calculated melt fraction of MM-3 against calculated temperature. As can be seen from Fig. 4e the melt fraction at both 1 and 1·5 GPa, for spinel lherzolite to dunite residues, shows smooth trends with calculated temperature, and these trends can be fitted with the following polynomials:
|
| (1) |
|
| (2) |
Average isobaric melt productivity (
F/T)P at both 1 and 1·5 GPa were calculated, in a similar manner to Wasylenki et al. (2003), by linear fits to T(°C)calc vs F data both below and above cpx-out. At 1 GPa the average melt productivity is 0·23/°C between 1270 and 1360°C for lherzolite melting and 0·14/°C between 1360 and 1470°C for harzburgite melting.
Calculated melt productivity for plagioclase lherzolite melting is 0·42/°C (1260–1272°C). Thus melting in the plagioclase stability field is about two times more productive than melting in the spinel lherzolite stability field. Calculated melt productivities at 1·5 GPa are higher than those at 1 GPa, with 0·34/°C for spinel lherzolite melting (temperature range 1360–1420°C) and 0·20/°C for harzburgite melting (temperature range 1420–1500°C).
In Fig. 5 we present the results of our mass-balance calculations for MM-3 peridotite (Table 3) to obtain melting reactions at 1 and 1·5 GPa. To calculate the melting reactions we have used the methods of Baker & Stolper (1994). The melting reactions listed below were determined from the slopes of linear fits for the solid phase trends in Fig. 5. The calculated melting reactions are as follows:
|
Plagioclase lherzolite
|
|
Spinel lherzolite
|
|
|
|
|
|
|
|
Harzburgite
|
|
|
|
|
|
|
|
), and ** indicates melt reaction calculated from Baker & Stolper (1994) data, this study.
The melting reaction at 1 GPa for spinel lherzolite differs significantly from that calculated using the data of Baker & Stolper (1994). This difference is most readily apparent in the significantly higher Cpx/Opx ratio of 10 vs 3 (this study). This result supports our repeated assertions (Falloon et al., 1996, 1997, 1999a) that melt compositions in diamond aggregate trapping experiments utilizing natural mineral mixes are dominated by a disequilibrium melt component derived from clinopyroxene [see Supplementary Data for further discussion and comparisons with the data of Baker & Stolper (1994)]. Baker & Stolper (1994), however, published a very different spinel lherzolite melting reaction (see reaction labelled * above) to that calculated using their data (this study, see reaction labelled ** above). This difference is because we used all of their data and did not force the linear fits to go through the starting modes of the MM-3 mineral mix as was the case for Baker & Stolper (1994). We believe that it is inappropriate to use the starting mineral mix mode as a constraint on the slopes of the linear fits, as the starting mineral compositions and modal proportions are not an equilibrium assemblage at 1 GPa. Because of the very slow reaction rates in natural mineral mixes they cannot be used to determine an equilibrium assemblage at subsolidus conditions.
However, the melting reaction at 1 GPa for harzburgite melting is similar to that calculated with the Baker & Stolper (1994) data. This result also confirms the conclusions of Falloon et al. (1999a) that only the higher temperature experiments on the MM-3 mineral mix utilizing a diamond aggregate trap are equilibrium melts of MM-3 composition
Petrogenesis of MORB; are MORB generated at the plagioclase–spinel lherzolite transition between 1 and 1·5 GPa?
Based on the experimental results of studies on model peridotite compositions in the chemical systems CMAS (Presnall et al., 1979), CMASN (Walter & Presnall, 1994) and CMASF (Gudfinnsson & Presnall, 2000), Presnall et al. (2002) have proposed a model for MORB petrogenesis that involves melting of mantle peridotite at relatively low pressures (0·9–1·5 GPa) and temperatures (1260°C). This model is a modification of the model originally proposed by Presnall et al. (1979), wherein MORB is produced by melting at a ‘cusp’ on the solidus of fertile mantle peridotite where spinel lherzolite changes to a plagioclase lherzolite assemblage, at 0·9 GPa. An earlier model advocating extraction of high-Al MORB at 1 GPa was proposed by Green & Ringwood (1967a) but retracted by Green et al. (1979) based on a detailed experimental investigation of the primitive MORB glass DSDP-3-18-7-1 (Green et al., 1979).
In the model of Presnall et al. (2002) the major element compositions of MORB are controlled by polybaric fractional melting over the pressure interval encompassing the spinel–plagioclase lherzolite phase transition. Melts generated in this pressure interval are compositionally distinct from MORB but reach the compositional field of MORB via cotectic olivine–plagioclase crystallization. Presnall et al. (2002) used the MORB glass data of Melson (1992) and viewed these data within the CIPW normative tetrahedron using projections from Di and Plag. When the MORB glasses are viewed from the projection from Plag onto the face Ol–Di–Q (see Presnall et al., 2002, fig. 8) they lie at higher values of Di than the divariant surface in the CMASNF system for liquids in equilibrium with ol, opx, cpx, pl and sp. Presnall et al. (2002), on the basis of this observation, concluded that all MORB glasses must have undergone cotectic fractionation crystallization of olivine and plagioclase if their model of magma genesis is correct. An important argument of Presnall et al. (2002) was that there is no evidence for olivine fractionation in MORB glasses when projected into the normative tetrahedron. In the Supplementary Data we provide detailed arguments to show that some MORB glasses show clear evidence for initial olivine-only fractionation and that some glasses are formed while the erupting melt is in the olivine-only field.
In Fig. 6a we project the compositions of MORB glasses from Table 5, for which we have mineralogical and experimental evidence (see Supplementary Data) for olivine-only crystallization before an olivine–plagioclase cotectic, into the normative projection from the olivine apex onto the face JCL–Di–Qz. We compare their plotted positions with our experimental results and other relevant experimental data. The projection from olivine onto the face JCL–Di–Qz (Fig. 6a) is particularly useful in discussions of MORB petrogenesis (Green & Falloon, 2005), as projected compositions will not change position with olivine addition or subtraction. Thus the position of glasses in the olivine-only field is fixed in this projection relative to their parental mantle-derived compositions.
|
|
As can be seen from Fig. 6a, the MORB glasses have significantly higher normative Di in the olivine projection compared with near-solidus melts at 1–1·5 GPa from fertile MORB mantle represented by peridotite MM-3. They also plot at higher normative diopside contents than melt compositions in equilibrium with plagioclase lherzolite residues at 1 GPa determined in complex systems (Falloon et al., 1997
; P. Nimis et al., unpublished data). It should be noted that the plagioclase lherzolite cotectic extends towards the plotted position of Na2O-free melt in equilibrium with plagioclase lherzolite in the simple system CMASF (at An100, Fig. 6). It is important to note that crystal fractionation of near-solidus melts from MM-3 at both high and low pressure will not move derivative liquids towards the composition of the MORB glasses (Fig. 7). In particular, the near-solidus melts from MM-3 at 1 and 1·5 GPa have low CaO (Fig. 7a) and FeO (Fig. 7c) and significantly higher Na2O (Fig. 7b) contents compared with the majority of MORB glasses. The low FeO contents of experimentally determined melts of MM-3 at 1 and 1·5 GPa result in significantly higher Mg-number compared with the most magnesian of MORB glasses (Mg-number >73 vs Mg-number <70; see Tables 4 and 5). It should be noted that melting of more depleted peridotite at low pressure will produce melts significantly lower in FeO and higher in MgO (and hence with higher Mg-number, >78; see Table 5) than melts from relatively fertile peridotite, as can be seen from the plotted position of the 1 GPa near-solidus melt for plagioclase lherzolite in the simple system CMASF (Fig. 7c, Table 5). This observation of high Mg-number in 1 GPa near-solidus peridotite melts compared with MORB glasses was used as an argument by Baker & Stolper (1994) for the non-primary nature of MORB glasses and the significant role of olivine fractionation from higher pressure primary melts (see Baker & Stolper, 1994, fig. 13). Similar arguments have also been made by Green et al. (1979).
|
It is also impossible for either oliv or oliv + plag fractionation to move the most refractory near-solidus melt of plagioclase lherzolite at 1 GPa (CMASF, Gudfinsson & Presnall, 2002) towards the plotted position of MORB glasses in Fig. 6. For the Presnall et al. (2002
) model to be consistent in Fig. 6, it requires that the MORB glasses in Fig. 6 obtained their plotted positions via oliv + plag crystallization from a very restricted range of melts in equilibrium with plagioclase lherzolite with An60–65 in the residue (approximately at the position of the oliv + plag vector in Fig. 6a). However, MORB glasses of these localities crystallize significantly more calcic plagioclase (An>80) than expected for melts derived from this restricted range of melts from the plagioclase lherzolite cotectic at 1 GPa. The calcic plagioclase crystallizing from these MORB glass suites, however, is consistent with a prior history of olivine crystallization from parental melts that have equilibrated at moderate to high melt fractions with lherzolite to harzburgite residues [see discussion by Green & Falloon (2005)]. As discussed above, the parental or primary melts for the MORB glasses from Table 5 will plot in Fig. 6a in exactly the same position as the glasses within the olivine-only field. Assuming that MM-3 is a suitable source for the parental melts of the plotted glasses, they separated from the mantle at pressures of 1·2–1·5 GPa leaving a lherzolite residue. However, on the projection from diopside (Fig. 6b), the position of glasses is affected by olivine fractionation, which moves them away from the olivine apex (Fig. 6b). As can be seen in Fig. 6b, all MORB glasses under consideration plot at a lower pressure than in Fig. 6a and require olivine addition to bring them into coincidence. This observation, together with the Mg-number differences noted above, is another confirmation of the early olivine-only crystallization from the parental MORB melts.
An important conclusion from the above discussion is that it is impossible for the selected MORB glasses (Table 5) to be the result of near-fractional melting of fertile to depleted peridotite in the pressure range of the spinel–plagioclase lherzolite transition (1–1·5 GPa) as suggested by the Presnall et al. (2002) model. The high normative diopside of these MORB glasses can be obtained only by a melting or equilibration process, essentially identical to isobaric batch melting, along an oliv + cpx + opx + liq cotectic towards the point of cpx exhaustion. In summary, our experimental results on near-solidus melt compositions from fertile peridotite MM-3 and the compositions of MORB glasses that lie in the olivine-only field of crystallization at low-pressure demonstrate that the model of Presnall et al. (2002) cannot explain the observed MORB compositional range. A detailed examination of a range of MORB glass suites has been presented by Green & Falloon (2005). They concluded that MORB glasses with >9·5 wt % MgO require primary melt compositions controlled by equilibrium with spinel lherzolite to harzburgite residues mostly at pressures of between 1·5 and 2 GPa. Green & Falloon (2005) found no evidence for primary MORB melts produced by low-pressure, low-degree melting of plagioclase lherzolite–spinel lherzolite compositions. This result is consistent with the conclusions of Niu (2004) that the plagioclase peridotite facies is within the cold thermal boundary layer beneath ocean ridges where it is too cold to melt and that MORB melts are produced in the higher pressure spinel peridotite facies. However, Green & Falloon (2005) did conclude that some MORB glasses require equilibrium with refractory harzburgite at pressures approaching 1 GPa.
The evidence for an isobaric batch-melting like process controlling the major element compositions of primary MORB melts is consistent with melting models that invoke the addition of variable proportions of small-degree volatile-rich (C–H–O) melts produced in the ‘incipient’ melting regime (Green & Falloon, 1998), which may be aggregated from a larger volume of the asthenosphere. This fugitive component is added to an upwelling lherzolite entering the ‘major melting regime’ (Eggins, 1992; Niu, 1997, 1999; Green & Falloon, 1998; Falloon et al., 1999b; Faul, 2001; Presnall et al., 2002; Presnall & Gudfinnsson, 2005). In such models the major element chemistry of parental MORB melts reflects a relatively large melt fraction or aggregation (F = 0·15–0·20) and equilibration with lherzolite to harzburgite residues, whereas the trace element and radiogenic geochemistry, especially U–Th series disequilibria (Rubin & Macdougall, 1988; Williams & Gill, 1989; Rubin et al., 2005) reflects the history of addition or extraction (McKenzie, 1985; Iwamori, 1994; Zou & Zindler, 2000; Speigelman, 2000) of small-degree melts from the ‘incipient melting regime’ and shallow wall-rock reaction or magma chamber processes.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
- At 1 GPa the rapid melting of subsolidus plagioclase controls the composition of near-solidus melts. For the MM-3 composition 4% melt is produced within 5°C of the solidus, which lies at 1270°C. The near-solidus melt has a Na2O content of 4·35 wt %, TiO2 content of 1·45 wt % and a low FeO content of 5·42 wt %. The 1 GPa near-solidus melt is nepheline normative and has a very low normative diopside content.
- At 1·5 GPa, the solidus of MM-3 lies at 1350°C, and 2% melting is produced within 10°C of the solidus. The near-solidus melt has a Na2O content of 3·56 wt %, TiO2 content of 1·22 wt % and low FeO of 6·91 wt %. The 1·5 GPa near-solidus melt is also nepheline normative and has low normative diopside content.
- Detailed comparisons (see Supplementary Data) of data obtained on MM-3 and other peridotite compositions using mineral mixes rather than sintered oxides as starting materials confirm the difficulties in obtaining equilibrated phase compositions and reliable melt compositions, whether or not melt traps (diamond, carbon spheres, olivine) are used. We confirm the need for both forward (peridotite melting) and reversal (liquidus studies of inferred melts), and these compositions are easily prepared from high purity reagents (and trace element components can be included). This iterative approach, also advocated by Dasgupta & Hirschmann (2007) and Hirschmann & Dasgupta (2007), is shortened if the first forward experiment can be used effectively to estimate the melt composition, and melt traps can be very helpful for this, together with phase composition, modal estimation and mass-balance calculations. At higher pressures the use of diamond or carbon spheres adds C as a potential component in the melting processes. The most effective technique for demonstrating near-solidus melt compositions (i.e. melts in equilibrium with four or more residual phases in peridotite) is to follow the forward melting experiment with one or two iterations of a layered or sandwich approach with peridotite layer and estimated melt, the melt thus forming a major part of the experimental charge.
- Comparing the near-solidus melts, produced from fertile to refractory lherzolite source compositions, at 1 and 1·5 GPa, with primitive MORB demonstrates that such MORB cannot be produced by aggregation of near-fractional melts from a progressively melted fertile lherzolite source to refractory residue. Such aggregated melts will have significantly lower Ca/Al values than those observed in primitive melts. The high Ca/Al values of the primitive MORB are consistent with their equilibrium with harzburgite to lherzolite residues at moderate (10–15%) melt fractions and pressures between 1 and 2 GPa. Our experimental results do not support the MORB petrogenetic model of Presnall et al. (2002).
- Estimates of mantle potential temperature of 1280°C or less are inconsistent with the generation of the most magnesian MOR glasses or with inferred parental MORB by upwelling of fertile asthenosphere of composition of MM-3 or MORB pyrolite. With this low potential temperature, upwelling mantle intersects the major melting regime of peridotite + (C, H, O) or the volatile-free lherzolite solidus at pressures close to 1 GPa and melt compositions are nepheline normative and with low CaO/Al2O3, low MgO and low FeO contents compared with the observed primitive glasses.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
We acknowledge the technical assistance of Keith Harris, Wieslav Jablonski, David Steele and Karsten Goemann. We acknowledge support from the Australian Research Council. We acknowledge support of the Museum of Natural History, Washington, DC, which provided electron microprobe standards. We thank Gudmundur Gudfinnsson, Yaoling Niu, Dana Johnston and Mike Baker for their constructive and critical reviews.
*Corresponding author. Telephone: +61-3-62262270. Fax: +61-3-62232547. E-mail: trevor.falloon{at}utas.edu.au
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...EXPERIMENTAL RESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Baker MB, Stolper EM. Determining the composition of high-pressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta (1994) 58:2811–2827.[CrossRef][Web of Science]
Baker MB, Hirschmann MM, Ghiorso MS, Stolper EM. Compositions of near-solidus peridotite melts from experiments and thermodynamic calculations. Nature (1995) 375:308–311.[CrossRef]
Bryan WB, Finger LW, Chayes F. Estimating proportions in petrographic mixing equations by least squares approximation. Science (1969) 163:926–927.
Conceição RV, Green DH. Behavior of the cotectic curve En–Ol in the system leucite–olivine–quartz under dry conditions to 2·0 GPa. Geochemistry, Geophysics, Geosystems (2000) 1(9). paper number 2000GC000071.
Danyushevsky LV. The effect of small amounts of H2O on crystallization of mid-ocean ridge and backarc basin magmas. Journal of Volcanology and Geothermal Research (2001) 110:265–280.[CrossRef][Web of Science]
Danyushevsky LV, Perfit MR, Eggins SM, Falloon TJ. Crustal origin for coupled ultra-depleted and plagioclase signatures in MORB olivine-hosted melt inclusions: evidence from the Siqueiros Transform Fault, East Pacific Rise. Contributions to Mineralogy and Petrology (2003) 144:619–637.[Web of Science]
Dasgupta R, Hirschmann MM. A modified iterative sandwich method for determination of near-solidus partial melt compositions. II. Application to determination of near-solidus melt compositions of carbonated peridotite. Contributions to Mineralogy and Petrology (2007) 154:647–661.[CrossRef][Web of Science]
Doukhan N, Doukhan J.-C, Ingrin J, Jaoul O, Raterron P. Early partial melting in pyroxenes. American Mineralogist (1993) 78:1246–1256.[Abstract]
Eggins SM. Petrogenesis of Hawaiian tholeiites: 2. Aspects of dynamic melt segregation. Contributions to Mineralogy and Petrology (1992) 110:398–410.[CrossRef][Web of Science]
Falloon TJ, Green DH, Hatton CJ, Harris KL. Anhydrous partial melting of a fertile and depleted peridotite from 2 to 30 kbars and application to basalt petrogenesis. Journal of Petrology (1988) 29:1257–1282.
Falloon TJ, Green DH, O’Neill H. StC, Ballhaus CG. Quest for low-degree mantle melts. Nature (1996) 381:285.[CrossRef]
Falloon TJ, Green DH, O’Neill H. SC, Hibberson WO. Experimental tests of low degree peridotite partial melt compositions, implications for the nature of anhydrous near-solidus peridotite melts at 1 GPa. Earth and Planetary Science Letters (1997) 152:149–162.[CrossRef][Web of Science]
Falloon TJ, Green DH, Danyushevsky LV, Faul UH. 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 (1999a) 40:1343–1375.[CrossRef][Web of Science]
Falloon TJ, Green DH, Jaques AL, Hawkins JW. Refractory magmas in back-arc basin settings—experimental constraints on the petrogenesis of a Lau Basin example. Journal of Petrology (1999b) 40:255–277.[CrossRef][Web of Science]
Falloon TJ, Green DH, Danyushevsky LV. Peridotite melting at 1 GPa; reversal experiments on partial melt compositions produced by peridotite–basalt sandwich experiments. Journal of Petrology (2001) 42:2363–2390.
Faul UH. Melt retention and segregation beneath mid-ocean ridges. Nature (2001) 410:920–923.[CrossRef]
Ghiorso MS, Hirschmann MM, Reiner PW, Kress VC. The pMELTS: a revision of MELTS for improved calculation of phase relations and major element partitioning related to partial melting of the mantle to 3 GPa. Geochemistry, Geophysics, Geosystems (2002) 3. paper number 2001GC000217.
Green DH. Experimental testing of ‘equilibrium’ partial melting of peridotite under water-saturated, high pressure conditions. Canadian Mineralogist (1976) 14:255–268.
Green DH, Falloon TJ. Pyrolite: a Ringwood concept and its current expression. In: The Earth's Mantle: Composition, Structure, and Evolution—Jackson I, ed. (1998) Cambridge: Cambridge University Press. 311–378.
Green DH, Falloon TJ. Primary magmas at mid-ocean ridges,‘hotspots’, and other intraplate settings: Constraints on mantle potential temperature. In. Plates, Plumes, and Paradigms. Geological Society of America, Special Papers—Foulger GR, Natland JH, Presnall DC, Anderson DL, eds. (2005) 338:217–247. doi:10.1130/2005.2388(14).
Green DH, Ringwood AE. The genesis of basaltic magmas. Contributions to Mineralogy and Petrology (1967a) 15:103–190.[CrossRef]
Green DH, Ringwood AE. The stability fields of aluminous pyroxene peridotite and garnet peridotite. Earth and Planetary Science Letters (1967b) 3:151–160.[CrossRef][Web of Science]
Green DH, Ringwood AE. Significance of a primitive lunar basaltic composition present in Apollo 15 soils and breccias. Earth and Planetary Science Letters (1973) 19:1–8.[CrossRef][Web of Science]
Green DH, Hibberson WO, Jaques AL. Petrogenesis of mid-ocean ridge basalts. In: The Earth: Its Origin, Structure and Evolution—McElhinny MW, ed. (1979) London: Academic Press. 265–290.
Gudfinnsson GH, Presnall DC. Melting behavior of model lherzolite in the system CaO–MgO–Al2O3–SiO2–FeO at 0·7–2·8 GPa. Journal of Petrology (2000) 41:1241–1269.
Hart SR, Zindler A. In search of a bulk-earth composition. Chemical Geology (1986) 57:247–267.[CrossRef][Web of Science]
Herzberg C, O’Hara MJ. Plume-associated ultramafic magmas of Phanerozoic age. Journal of Petrology (2002) 43:1857–1883.
Hirschmann MM, Baker MB, Stolper EM. The effect of alkalis on the silica content of mantle-derived melts. Geochimica et Cosmochimica Acta (1998) 62:883–902.[CrossRef][Web of Science]
Hirschmann MM, Dasgupta R. A modified iterative sandwich method for determination of near-solidus partial melt compositions. I. Theoretical considerations. Contributions to Mineralogy and Petrology (2007) 154:635–645.[CrossRef][Web of Science]
Iwamori H. 238U–230Th–226Ra and 235U–231Pa disequilibria produced by mantle melting with porous and channel flows. Earth and Planetary Science Letters (1994) 125:1–16.[CrossRef][Web of Science]
Jaques AL, Green DH. Determination of liquid compositions in high-pressure melting of peridotite. American Mineralogist (1979) 64:1312–1321.[Abstract]
Jaques AL, Green DH. Anhydrous melting of peridotite at 0–15 kb pressure and the genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology (1980) 73:287–310.[CrossRef][Web of Science]
Jarosewich EJ, Nelen JA, Norberg JA. Reference samples for electron microprobe analyses. Geostandards Newsletter (1980) 4:257–258.[CrossRef]
Johannes W, Bell PM, Boettcher AL, Chipman DW, Hays JF, Mao HK, Newton RC, Seifert F. An interlaboratory comparison of piston-cylinder pressure calibration using the albite-breakdown reaction. Contributions to Mineralogy and Petrology (1971) 32:24–38.[CrossRef][Web of Science]
Kushiro I, Syono Y, Akimoto S. Melting of a peridotite at high pressures and high water pressures. Journal of Geophysical Research (1968) 73:6023–6029.
McKenzie DP. 230Th–238U disequilibrium and the melting processes beneath ridge axes. Earth and Planetary Science Letters (1985) 72:149–157.[CrossRef][Web of Science]
McNeill AW. The crystallization history of normal mid-ocean ridge basalts from the eastern Pacific Ocean and implications for the composition of primary mid-ocean ridge magmas: evidence from mineralogy, pillow-rim glasses and melt inclusion studies. In: Ph.D. thesis (1997) Hobart: University of Tasmania. 258.
McNeill AW, Danyushevsky LV. Composition and crystallization temperatures of primary melts from Hole 896A basalts: evidence from melt inclusion studies. In. In: Proceedings of the Ocean Drilling Program, Scientific Results, 148—Alt JC, Kinshita H, Stokking LB, Michael PJ, eds. (1996) College Station, TX: Ocean Drilling Program. 21–35.
Melson WG. VGGP Database of Basalt Glass Analyses (1992) Washington, DC: Smithsonian Institution.
Nicholls IA. Liquids in equilibrium with peridotitic mineral assemblages at high water pressures. Contributions to Mineralogy and Petrology (1974) 45:289–316.[CrossRef][Web of Science]
Niu Y. Mantle melting and melt extraction processes beneath ocean ridges: Evidence from abyssal peridotites. Journal of Petrology (1997) 38:1047–1074.[CrossRef][Web of Science]
Niu Y. Comments on some misconceptions in igneous/experimental petrology and methodology: A reply. Journal of Petrology (1999) 40:1195–1203.
Niu Y. Bulk-rock major and trace element compositions of abyssal peridotites: Surprises and implications for mantle melting, melt extraction and post-melting processes beneath ocean ridges. Journal of Petrology (2004) 45:2458.
Niu Y. Generation and evolution of basaltic magmas: some basic concepts and a new view on the origin of Mesozoic–Cenozoic basaltic volcanism in eastern China. Geological Journal of China Universities (2005) 11:9–46.
O’Hara MJ, Yoder HS. Formation and fractionation of basic magmas at high pressures. Scottish Journal of Geology (1967) 3:67–117.
Presnall DC, Gudfinnsson GH. Carbonate-rich melts in the oceanic low-velocity zone and deep mantle. In. Foulger GR, Natland JH, Presnall DC, Anderson DL, eds. (2005) 338:207–216. Plates, Plumes, and Paradigms. Geological Society of America, Special Papers, doi:10.1130/2005.2388(13).
Presnall DC, Dixon JR, O’Donnell TH, Dixon SA. Generation of mid-ocean ridge tholeiites. Journal of Petrology (1979) 20:3–35.
Presnall DC, Gudfinnsson GH, Walter MJ. Generation of mid-ocean ridge basalts at pressures from 1 to 7 GPa. Geochimica et Cosmochimica Acta (2002) 66:2073–2090.[CrossRef][Web of Science]
Raheim A, Green DH. Experimental determination of the temperature and pressure dependence of the Fe–Mg partition coefficient for coexisting garnet and clinopyroxene. Contributions to Mineralogy and Petrology (1974) 48:179–203.[CrossRef][Web of Science]
Robinson J. AC, Wood BJ, Blundy JD. The beginning of melting of fertile and depleted peridotite at 1·5 GPa. Earth and Planetary Science Letters (1998) 155:97–111.[CrossRef][Web of Science]
Rubin KH, Macdougall JD. 226Ra excesses in mid-ocean-ridge basalts and mantle melting. Nature (1988) 335:158–161.[CrossRef]
Rubin KH, van der Zander I, Smith MC, Bergmanis EC. Minimum speed limit for ocean ridge magmatism from 210Pb–226Ra–230Th disequilibria. Nature (2005) 437:534–538. doi:10/1038/nature03993.[CrossRef][Medline]
Schwab BE, Johnston AD. Melting systematics of modally variable, compositionally intermediate peridotites and the effect of mineral fertility. Journal of Petrology (2001) 42:1789–1811.
Spiegelman M. UserCalc: a web-based Uranium series calculator for magma-migration problems. Geochemistry, Geophysics, Geosystems (2000) 1(8). doi:10.1029/1999GC000030.
Stolper E. A phase diagram for mid-ocean ridge basalts: preliminary results and implications for petrogenesis. Contributions to Mineralogy and Petrology (1980) 74:13–27.[CrossRef][Web of Science]
Takahashi E, Kushiro I. Melting of a dry peridotite at high pressures and basalt magma genesis. American Mineralogist (1983) 68:859–879.[Abstract]
Walter MJ, Presnall DC. Melting behavior of simplified lherzolite in the system CaO–MgO–Al2O3–SiO2–Na2O from 7 to 35 kbar. Journal of Petrology (1994) 35:329–359.
Wasylenki LE, Baker MB, Kent A. JR, Stolper EM. Near-solidus melting of the shallow upper mantle: partial melting experiments on depleted peridotite. Journal of Petrology (2003) 44:1163–1191.
Williams RW, Gill JB. Effects of partial melting on the uranium decay series. Geochimica et Cosmochimica Acta (1989) 53:1607–1619.[CrossRef][Web of Science]
Yoder HS, Tilley CE. Origin of basalt magmas: an experimental study of natural and synthetic rock systems. Journal of Petrology (1962) 3:342–529.
Zou Z, Zindler A. Theoretical studies of 238U–230Th–226Ra and 235U–231Pa disequilibria in young lavas produced by mantle melting. Geochimica et Cosmochimica Acta (2000) 64:1809–1817.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||