| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Phase Relations and Liquid Lines of Descent in Hydrous Ferrobasalt—Implications for the Skaergaard Intrusion and Columbia River Flood Basalts
Institut Für Mineralogie, Leibniz Universität Hannover, Callinstr. 3, D-30167, Hannover, Germany
RECEIVED JUNE 15, 2007; ACCEPTED AUGUST 19, 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Crystallization experiments using a hydrous ferrobasalt as starting material, conducted at 200 MPa, 940–1200°C, at a wide range of water activities (0·1–1) and redox conditions (QFM – 3 to QFM + 4, where QFM is the quartz–fayalite–magnetite oxygen buffer), show that H2O influences significantly the differentiation history of ferrobasaltic magmas. A combination of our data with published experiments on dry ferrobasalt at 1 atm provides an extensive experimental database for modeling and quantifying crystallization and differentiation processes within a typical Fe-rich tholeiitic system under both dry and hydrous conditions. The addition of H2O decreases liquidus temperatures and changes significantly the proportions, temperature range and sequence of crystallizing mineral phases. The dry liquidus is at about 1170°C whereas the liquidus for H2O-saturated melts is at
1060°C. The main phases crystallizing from H2O-bearing ferrobasalt at the investigated conditions are olivine (OL), clinopyroxene (CPX), plagioclase (PL), magnetite (MT), hematite (HM), ilmenite (ILM) and amphibole (AM). The phase assemblage is similar to that of the dry system except for the presence of HM at extremely oxidizing conditions and AM at low temperatures (< 950°C) and H2O-saturated conditions. The important observation made in this study is that the stability of Fe–Ti-oxides, and in particular MT, as well as the simultaneous coprecipitation of MT and ILM, are almost independent of the activity of H2O (aH2O) in the system, whereas the liquidus temperatures of the silicate minerals are dramatically depressed by increasing aH2O. The stabilities of oxides are controlled mainly by the redox conditions prevailing in the system. The most pronounced effect of aH2O on the liquidus temperatures of silicates is observed for PL, which shows a considerable delay in crystallization with progressive magma differentiation. Early crystallization of Fe–Ti-oxides in H2O-bearing ferrobasaltic compositions precludes any significant Fe enrichment during differentiation. As Fe enrichment is a characteristic feature of the Skaergaard intrusion, it implies that the Skaergaard parental magma did not contain considerable amounts of water. On the other hand, our experiments indicate that the differentiation of some ferrobasaltic series from the Columbia River flood basalt province might have occurred in magmatic systems containing significant amounts of volatiles (0·5–3 wt % H2O). KEY WORDS: ferrobasalt; Skaergaard; Columbia River flood basalts; experiment; differentiation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Basaltic magmas rich in Fe and Ti evolve by differentiation processes following the tholeiitic Fenner trend of differentiation (Fenner, 1929
). Early stages of magma evolution are characterized by an increase in FeO and TiO2 concentrations at almost constant SiO2 content of the melt, followed by enrichment in silica and depletion in iron at higher degrees of differentiation. The change in the differentiation path from Fe enrichment to Fe depletion occurs at the onset of Fe–Ti-oxide crystallization, which strongly depends on the conditions prevailing in the magmatic system, mainly the fO2.
Natural examples of magmas with FeO enrichment of up to 12–19 wt % can be found in volcanic tholeiitic suites (e.g. Carmichael, 1964), mid-ocean ridge basalts (MORB; e.g. Lehnert et al., 2000), late-stage ferrobasalts in gabbros (Natland & Dick, 2001), layered intrusions (e.g. Skaergaard, Wager, 1960; Kiglapait, Morse, 1981; Newark Island, Wiebe & Snyder, 1993; Fedorivka, Duchesne et al., 2006) and some flood basalts (e.g. Columbia River basalts, Hooper, 2000). However, most of the Fe-rich compositions are found in plutonic environments and only rare cases are known for erupted rocks, which is attributed to the high density of such magmas (e.g. Sparks et al., 1980; Stolper & Walker, 1980; Brooks et al., 1991; Stewart et al., 2003). The eruptive potential of voluminous flood ferrobasalts is considered to be a result of relatively high pre-eruptive concentrations of volatiles in the magmas (e.g. Columbia River basalts with [H2O + CO2] > 4 wt %, Lange, 2002). On the other hand, high concentrations of water in a magma can shift the tholeiitic differentiation to a calc-alkaline differentiation trend (e.g. Sisson & Grove, 1993; Grove et al., 2003; Berndt et al., 2005) as a result of the dramatic effect of volatiles on crystallization temperatures, phase relations and compositions of minerals. This implies that the accurate interpretation of natural data and modelling of differentiation processes in Fe-rich magmas (e.g. Ariskin, 2003) requires systematic investigation of natural systems and experimental quantification of the role of volatiles in crystallization processes. Although a number of experimental studies have focused on the influence of H2O on the crystallization of mafic compositions (e.g. Hamilton et al., 1964; Holloway & Burnham, 1972; Spulber & Rutherford, 1983; Sisson & Grove, 1993; Gaetani et al., 1994; Muntener et al., 2001; Pichavant et al., 2002b; Grove et al., 2003; Berndt et al., 2005; Di Carlo et al., 2006; Feig et al., 2006; Hamada & Fujii, 2008; Mercer & Johnston, 2008), little is known about the differentiation of typical ferrobasaltic magmas in the presence of volatiles.
The lack of such data has motivated our experimental investigations of the role of H2O–CO2-bearing fluids in the evolution of ferrobasalts, with particular attention to the differentiation paths and phase relations in the magmas. The starting material was chosen to be representative of the parental magma of the Skaergaard intrusion (Greenland), considering that the igneous complex of Skaergaard is one of the best investigated tholeiitic layered intrusions of ferrobasaltic composition in the world (resulting in over 500 publications as reported by Andersen & Brooks (2003); see also recent special issue of Lithos 2006; 92, 1–2) In particular, the magma differentiation processes that occurred at Skaergaard have been investigated experimentally for several decades. It is generally assumed that the Skaergaard magma was dry at the time of emplacement and, hence, the experimental approaches have focused on Skaergaard-like compositions at dry and low-pressure conditions (e.g. Snyder et al., 1993; Toplis & Carroll, 1995; Lattard & Partzsch, 2001; Partzsch et al., 2004; Thy et al., 2006). However, the role of H2O in the history of the Skaergaard could be underestimated, especially in the later stages of closed-system evolution of the Skaergaard magma. Recent studies have shown that late-stage granophyre magmas and pegmatites within the Skaergaard horizons were fluid-rich and produced typical H2O-bearing minerals such as amphibole and biotite (e.g. Larsen & Tegner, 2006). Primary fluid inclusions found in the granophyre minerals indicate the presence of fluids composed of aqueous saline solutions and CH4 (Larsen et al., 1992; Larsen & Tegner, 2006). Thus, additional experimental approaches are needed to evaluate the possible role of volatiles in the differentiation processes and evolution of Skaergaard ferrobasalts. Moreover, considering the large and detailed database obtained for dry ferrobasalts, an experimental approach simulating volatile-bearing conditions offers an opportunity for direct comparison of the data obtained in water-rich and dry ferrobasaltic systems, as well as an interpolation of experimental results for ferrobasalts with intermediate water contents. A combination of existing experimental data obtained under dry and water-bearing conditions is useful not only for the local interpretation of Skaergaard magma differentiation but also for general modeling and quantifying crystallization and differentiation processes within a typical Fe-rich tholeiitic system under dry to hydrous conditions. Here we apply our data to interpret the compositional evolution of some Fe-rich basaltic series from the Columbia River flood basalts.
|
EXPERIMENTAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Starting material
The starting material was a synthetic analogue of a ferrobasaltic melt (Table 1), which is assumed to be the composition of the parental magma of the Skaergaard intrusion (dike C, Brooks & Nielsen, 1978
). The same composition (hereafter SC1, following Toplis & Carroll, 1995) was used in experimental studies on phase relations and differentiation in a ferrobasaltic system at dry conditions (1 atm) in a system open and closed to oxygen (Toplis & Carroll, 1995, 1996; Lattard & Partzsch, 2001) and more recently in experiments investigating the oxidation state of Fe in dry (Partzsch et al., 2004) and in hydrous (Botcharnikov et al., 2005b) melts as well as the solubility of C-O-H fluids in ferrobasalts (Botcharnikov et al., 2005a).
|
The starting glass powder was prepared from a mixture of oxides (SiO2, TiO2, Al2O3, Fe2O3, MgO) and carbonates (CaCO3, Na2CO3, K2CO3) ground in a rotary mortar. After 2 h of melting in a Pt crucible at 1600°C, 1 atm, and at a log fO2 = –0.68 (air), the resulting melt was quenched and the obtained glass was ground in an agate mortar. The powdered glass was melted a second time for 0·5 h to obtain a homogeneous composition. The homogeneity of the silicate glass was verified by electron microprobe (see standard deviation of multiple analyses in Table 1).
For experiments under oxidizing conditions (with fO2 corresponding to 4 log-bar units higher than the quartz–fayalite–magnetite (QFM) oxygen buffer or log fO2 QFM + 4·2), the starting glass was used without any additional pre-experimental treatment. Because the preparation of the starting glass was performed at highly oxidizing conditions at 1600°C and 1 atm, it was necessary to pre-equilibrate the starting glass at the required fO2 for the experiments at reducing conditions. Such a treatment minimizes the production of excess H2O inside the capsule as a result of reduction of initially oxidized melt and decreases the duration of sample re-equilibration during the experiment. The glass powder was placed in a ceramic crucible and melted for 2 h at 1220°C in a 1 atm gas-mixing (Ar–H2–H2O) furnace at an fO2 corresponding to the desired fO2 of future high-pressure experiments (no significant contamination of melt with Al2O3 from the crucible was observed). The melted glass batch was quenched by dropping the crucible in cold water. The quenched glass was drilled out of the crucible, crushed, and powdered. Two fractions with grain sizes of <100 µm and 100–200 µm were mixed together in a ratio 1 : 1 to decrease the free volume between grains.
Experimental strategy
Our goal was to extend the database of Toplis & Carroll (1995), obtained for the ferrobasaltic composition SC1 at dry conditions, to fluid-bearing conditions at various water activities. To reproduce the possible wide range of storage conditions in natural magmatic systems (i.e. variations in water activity and redox state), we conducted phase equilibrium experiments in which a basaltic system was equilibrated with H2O–CO2-bearing fluids at three fixed hydrogen fugacities. This resulted in different redox conditions depending on the activity of H2O in the system.
The composition of the fluid phase in our experiments was varied by adding different proportions of H2O and CO2. Silver oxalate (Ag2C2O4) was used as a source of CO2. The initial mole fraction of H2O (XflH2Oini) in the fluid phase was varied from XflH2Oini = 0·0 (nominally dry, pure CO2 in the fluid) to 1· 0 (water-saturated), as listed in Table 2. For most of the samples, the initial amount of H2O, CO2 or H2O–CO2 mixture was kept constant on a molar basis and equivalent to 5 wt % of total pure water in the capsule. This amount of added volatiles was enough to produce a free fluid phase in most experimental runs at 200 MPa total pressure (except samples B1 and B177, which contained no initial fluid), which was verified using the weight-loss method after opening the capsules.
|
The temperature of the experimental runs was varied from 1175 to 940°C, in most cases in 25°C intervals, to cover a wide range of crystallization conditions in hydrous system. All experiments were conducted at 200 MPa, which is in agreement with recent estimates of the prevailing pressure during Skaergaard crystallization based on fluid inclusion data and amphibole–plagioclase geobarometry for primary granophyres of the Skaergaard intrusion (Larsen & Tegner, 2006
).
Pre-saturation of capsules with Fe
The experiments were performed under redox conditions varying from highly oxidizing (log fO2 QFM + 4) to reducing conditions (log fO2 QFM – 3). With decreasing fO2, the proportion of ferrous Fe in the melt and the solubility of Fe in the capsule material increase (especially in Pt capsules), and this may affect significantly the composition of silicate melt inside the capsule. To minimize the loss of Fe, we used capsules made of Au80Pd20 alloy and of Au at temperatures above and below 1000°C, respectively (Table 2). The Au80Pd20 capsules have been proved to be suitable containers for experiments with Fe- and H2O-bearing compositions at magmatic temperatures and pressures (e.g. Kawamoto & Hirose, 1994; Hall et al., 2004; Berndt et al., 2005; Feig et al., 2006). In addition, for experiments under reducing conditions (i.e. at log fO2 
QFM + 1) all Au80Pd20 capsules were presaturated with Fe following the procedure of Ford (1978). The capsule containers were placed in a ceramic crucible, filled and covered with a basaltic melt of similar composition to that of the starting glass. The crucible was held for 2 days at 1220°C in a 1 atm gas-mixing furnace with controlled oxygen fugacity at the desired fO2 of future experiments. After the pre-saturation, all glass remnants were mechanically removed from the capsules. The capsules were then cleaned in HF for 2 days. However, this procedure can only minimize the risk of iron loss and we were not able to completely avoid the problem in samples with low water activity. In some cases, when the redox conditions of Fe presaturation were more reduced than the actual redox conditions in the capsules, a slight increase in the Fe content of the melt was detected (up to 5 relative % as noted in Table 2). Thus, additional measures were taken to minimize the problem of iron loss or gain. In particular, run times were kept as short as possible to ensure a compromise between the attainment of local equilibrium and the migration of Fe. Additionally, the experimental products were mainly analyzed in the central part of the sample (far from the capsule wall). Although it has been suggested that Au capsules may absorb large amounts of Fe at low fO2 (Ratajeski et al.., 1999), no significant Fe loss to our Au capsules was detected (Table 2). Hence, no Fe presaturation of Au containers was carried out prior to the experiments.
Experimental technique
For each experiment, about 40–50 mg of dry glass powder was loaded in 15 mm long (inner diameter of 2·6 mm) Au80Pd20 or Au capsules. Water (0–2·5 µl) and a certain amount of silver oxalate (0–19 mg) were added to the glass powder to adjust the desired XH2Ofl in the capsule. The glasses for the experiments at nominally dry conditions (pure CO2 in the fluid) were put into the capsules and dried at 600°C for 1–2 h to minimize the amount of adsorbed water. Then, silver oxalate was added to the capsules and the capsules were welded shut with a graphite arc-welder.
The experiments were performed in internally heated pressure vessels (IHPV) oriented vertically [a detailed description of the apparatus has been given by Berndt et al. (2002)]. The total pressure was measured and recorded continuously with an uncertainty of about 1 MPa. The variations of pressure during the experiments were <5 MPa. Temperature was measured with four unsheathed S-type (Pt–Pt90Rh10) thermocouples to control the temperature gradient over a length of 30 mm inside the vessel. Temperature oscillations were below 3–5°C depending on the vessel and experimental run.
The IHPV used for oxidized experiments was pressurized with pure Ar gas. The experiments at reduced conditions were conducted in a second IHPV pressurized with a mixture of Ar and H2 gases. In the case of the pure Ar pressure medium, the intrinsic fO2 of the vessel was close to log fO2 QFM + 4·2 (Berndt et al., 2002). An Ar–H2 gas mixture was used as the pressure medium to adjust the required fH2 in the vessel and to perform experiments at the desired redox conditions [calculation of fO2 values is based on the equation of Schwab & Kuestner (1981)]. The f H2 prevailing in the IHPV at high P and T was controlled with a Shaw-membrane (e.g. Scaillet et al., 1992; Berndt et al., 2002). Different fixed hydrogen fugacities were applied in the experiments to maintain redox conditions corresponding to the nominal oxygen buffers at log fO2 = QFM + 1 and QFM in the systems with water activity (aH2O) equal to unity at a given T and P. Within the sample capsule, the hydrogen fugacity is fixed as a result of an inward diffusion of hydrogen. This, in turn, controls the fugacity of oxygen inside the capsule through the equilibrium reaction of water formation (H2 + 
O2 
H2O). Thus, in the capsules with aH2O <1, the redox conditions were more reduced than in the experiments with aH2O = 1 (e.g. Scaillet et al., 1992; Botcharnikov et al., 2005b).
The capsules were pressurized to 200 MPa and heated isobarically from room temperature to the temperature of the experiment at a rate of 30°C/min. Any overheating did not exceed 10°C and in most cases was less than 5°C. The run duration varied from 1 to 120 h, depending on the run temperature, expected fO2 and capsule material. The 1 h runs were conducted at high temperatures to minimize the Fe-loss problem, assuming that the kinetics of chemical transport and redox reactions in basaltic melts at high temperature is fast enough to attain equilibrium (e.g. Chekhmir et al., 1985; Berndt et al., 2002; Gaillard et al., 2002, 2003a, 2003b). The redox kinetics of Fe as a main heterovalent cation in our ferrobasalt is expected to be identical for hydrous and non-hydrous conditions and the rate of Fe oxidation or reduction should be not affected by the kinetics of H2 transfer through the capsules as reported by Gaillard et al. (2002). The samples were quenched using a sample holder equipped with a rapid quench facility (Berndt et al., 2002). The quench rate depends on the size of the sample and the T–P conditions inside the vessel but it was sufficiently fast to avoid quench effects (the quench rate is about 150 K/s).
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Determination of XH2Ofl and calculation of aH2O
A conventional weight-loss method was used to determine the mole fraction of water in the fluid phase (XH2Ofl) coexisting with melt and mineral phases at high pressure and temperature: (1) the capsule was weighed; (2) the fluid phase was frozen by placing the capsule in liquid nitrogen; (3) the capsule was pierced with a needle; (4) after warming to room temperature, the capsule was weighed to determine the mass of CO2 in the fluid; (5) the capsule was placed into a drying furnace at 110°C for 3–5 min and subsequently weighed to measure the mass of H2O, lost from the capsule. The amount of atmospheric nitrogen trapped in the experimental charge during preparation of the capsules was estimated to be low (see Tamic et al., 2001
) and not considered in the calculations of XH2Ofl.
It must be noted that H2O and CO2 are the dominant fluid components, and concentrations are at least one order of magnitude lower for CO and CH4 than for CO2 in a wide range of redox conditions at 200 MPa (e.g. Churakov & Gottschalk, 2003; Duan & Zhang, 2006). A significant effect of redox conditions on the speciation of carbon in the fluid phase and solubility of CO2 in the silicate melt is expected at log fO2 less than QFM – 1 only (e.g, Pawley et al., 1992; Holloway & Blank, 1994; Scaillet & Pichavant, 2004). Thus, the mole fraction of CO2 and H2O in the fluid phase in most of our experiments performed at log fO2 
QFM – 1 can be reliably calculated from the determined weight loss from the capsules. The calculated fO2 values for samples at more reduced conditions can be slightly overestimated as a result of the increasing proportion of CO (+ CH4) in the fluid phase.
The determined XH2Ofl provides estimates of the water activity prevailing in a capsule during the experiment. The relationship between XH2Ofl and aH2O can be derived from empirical (e.g. Aranovich & Newton, 1999) and thermodynamic (e.g. Churakov & Gottschalk, 2003; Duan & Zhang, 2006) models of the properties of C–O–H fluids. The models show that the deviation from an ideal behavior of C–O–H fluids is low at high temperatures and 200 MPa. The difference between determined XH2Ofl and calculated aH2O is typically comparable with the analytical uncertainty for XH2Ofl. Thus, for simplicity, we assumed that aH2O is approximately equal to XH2Ofl (Table 2).
The determination of XH2Ofl failed for several samples. In this case, water activity was calculated from the model of Burnham (1979) based on the estimated concentrations of dissolved H2O in a residual melt (see below). It should be noted that the model of Burnham (1979) slightly overestimates water activity in basaltic melts (e.g. Botcharnikov et al., 2005b).
Because the final XH2Ofl in the system depends on different parameters such as the initial mole fraction of water in the capsule, the degree of crystallinity, and the H2O solubility in the residual melt, which is a function of melt composition, the measured values in the experimental products vary over a wide range (Table 2). Hence, for simplicity, we used the initial mole fraction of H2O in the fluid to distinguish between the experimental series and to compare the results of runs.
Electron microprobe
Fragments of each sample were mounted in epoxy for electron microprobe analysis. The analyses of the experimental products were performed with a Cameca SX100 electron microprobe. Minerals were analyzed with focused beam at 15 kV, 15 nA beam current and counting times for major elements of 10 s. Sodium and potassium were analysed first with counting times of 5 s to minimize alkali loss. Glasses were measured with a defocused beam of 5–20 µm, 4 nA beam current and counting times of 4 s for Na and K, and 8 s for the other elements. In samples with low melt fraction, the microprobe beam was defocused as much as possible. No significant alkali loss (within the uncertainty) was detected using these analytical conditions. Multiple measurements were made for each phase within a sample to minimize possible analytical errors and check for homogeneity.
Mass balance calculations for phase proportions and Fe loss from the system
Mass balance was applied to calculate the proportions of coexisting phases using the determined phase compositions. Glass analyses were normalized to 100% to exclude the H2O dissolved in the melt from the mass balance. Water content of amphiboles was not considered in the calculations. Special attention was paid to the estimation of Fe loss or gain as a result of reaction with the capsule walls. In the case of superliquidus experiments, the calculation of Fe loss or gain was straightforward, whereas in subliquidus charges it was estimated by varying the initial Fe content of the charge in the calculations. The minimum value of the residual (R2) for the calculated mass proportions of coexisting phases was used as a criterion for the estimation of the bulk Fe loss or gain. Experimental products for which corrections related to loss of Fe have been taken into account are indicated in Table 2. The mass balance calculations show maximum Fe gain of about 5 relative % and Fe depletion of 18 relative %; the calculated proportions of phases with the minimum R2 are reported in Table 2.
Determination of water content in experimental glasses
Two methods were applied to determine H2O concentrations in the experimental quenched glasses. Selected samples, free of mineral phases or containing small amount of crystals, were analyzed using conventional near-infrared Fourier transform IR (FTIR) spectroscopy as described in detail by Botcharnikov et al. (2005b). The H2O content of the glasses in most samples was determined using the ‘by-difference’ method (e.g. Devine et al., 1995); that is, 100% minus the analytical total. To improve the quality of the ‘by-difference’ technique, two sets of standard hydrous MOR basaltic glasses and SC1 ferrobasaltic glasses with known water contents [from 0·89 to 6·27 wt % H2O after Berndt et al. (2002) and from 0·25 to 4·7 wt % after Botcharnikov et al. (2005b), respectively] were analysed during each electron microprobe session (to minimize the effect of possible drift in analytical conditions). The obtained calibration curve was used to calculate the actual H2O concentration of the sample. The uncertainty of the calibration includes mainly the counting statistics on the microprobe total. The typical error is usually 0·5 wt % H2O. The possible effect of the redox state of the glass on H2O determination was found to be within the error of the method. In addition, the comparison of H2O concentrations obtained by FTIR spectroscopy with the ‘by-difference’ data also showed an agreement within the uncertainty of the latter method. The highest errors are obtained for experimental products with low melt fractions, because glass analyses were possible only with focused or slightly defocused (2–5 µm) microprobe beam.
Calculations of fO2
At fixed hydrogen fugacity in IHPV, the fH2 inside the capsules is controlled by diffusion of H2 through the capsule wall and is identical to the fH2 in the vessel. Hence, the redox state of the system in each experiment depends on the external redox conditions in the IHPV and on the redox reactions in the capsule. Assuming a negligible effect of reactions between carbon-bearing species on redox conditions at the studied T and P, the dissociation of water is the main reaction controlling redox equilibria inside the capsules. Using the estimated aH2O values, the prevailing fO2 was calculated for each experiment as log fO2capsule = log fO2apparent + 2log (aH2O) (see also Botcharnikov et al., 2005b) where log fO2apparent is the oxygen fugacity that is expected in the system at aH2O = 1. The fO2 values were also used to calculate the ferric–ferrous ratio in the residual melts according to the model of Moretti (2005). The effect of aH2O in the system on ferric–ferrous ratio is found to be insignificant in ferrobasaltic melts and the Fe3+/Fe2+ variations are within the precision of the model predictions at 1200°C and 200 MPa (Botcharnikov et al., 2005b). The results of the fO2 and 
QFM calculations are presented in Table 2 (where QFM is the difference in log fO2 relative to the QFM buffer at a given T and P).
Attainment of equilibrium
Although we did not perform any reversal or time-dependent experiments (mainly because of the Fe-loss problem), to check the attainment of equilibrium in our systems we have several lines of evidence that phases in most experiments have been equilibrated during the run. The high volatile diffusivities and high crystallization kinetics in basaltic systems ensure fast equilibration between silicate melt, fluid and crystallizing mineral phases at high temperatures. At lower temperatures in systems close to the solidus, having a silica-rich residual melt composition, the equilibration kinetics slows down and requires an extremely long duration of the experiments. Hence, such systems may not reach equilibrium at the laboratory time-scale. With the exception of samples from the experiments at near-solidus conditions (noted in Table 2), most samples show evidence of attained equilibrium based on textural and compositional characteristics: (1) the distribution of mineral phases is homogeneous throughout the samples; (2) the morphology of the crystals does not indicate any quench crystallization (tails or skeletal forms); (3) the minerals and glasses have mostly homogeneous compositions (except Fe loss from some glasses close to the capsule wall); (4) the crystallization sequence, compositions and proportions of phases follow systematic trends; (5) the mineral–melt and mineral–mineral partition coefficients are in agreement with literature data (see details in following discussion).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Phase relations
The experimental products and phase proportions are listed in Table 2. The crystallizing phases are olivine (OL), clinopyroxene (CPX), plagioclase (PL), magnetite (MT), hematite (HM), ilmenite (ILM) and amphibole (AM). Commonly, the stabilities of different phases depend on pressure, temperature, water activity and oxygen fugacity in the system. At isobaric conditions and at a given external redox potential (fixed fH2), the redox state of the system is controlled by temperature and water activity in the capsule. To evaluate the effect of each of these three key parameters on phase relations, the determined stability fields of the minerals are plotted as a function of temperature, water activity (or water content of the melt, H2Om) and fO2 (
QFM) in Figs 1 and 2.
|
|
Figure 1 illustrates phase relations as a function of temperature (in the range from 900–1200°C) and water content of the melt (from
0·5 to 6 wt % H2Om) at three nominal redox conditions, corresponding to QFM + 4 (Fig. 1a), QFM + 1 (Fig. 1b) and QFM (Fig. 1c) at aH2O = 1. With a decrease in water content or activity, the redox conditions in the capsules become more reduced than the nominal conditions as shown by black arrows and vertical dotted lines in Fig. 1. These lines represent the log fO2 values relative to QFM buffer (shown as QFM) estimated from the determined aH2O (or H2Om). Redox conditions of the experiments vary in a wide range of about three logarithmic units of fO2 at a given nominal redox state of the system as shown in all three panels (a, b and c) of Fig. 1. The entire range of explored fO2 covers about seven logarithmic units relative to QFM. Figure 2 illustrates phase equilibria as a function of temperature and redox conditions at different nominal water activities in the system. Panels a, b, c and d show phase relations at nominal water activities, corresponding to the initial mole fractions of H2O in the fluid equal to 1, 0·6, 0·2 and 0, respectively. Figure 2d illustrates the stability fields of different phases at dry conditions after Toplis & Carroll (1995). As mentioned above, because of the dependence of phase stabilities on all three parameters, both types of diagrams (Figs 1 and 2) will be considered in the following description of results and the subsequent discussion.
Although experiments at very oxidizing conditions (>QFM + 2) shown in Figs 1a and 2a–d represent an extreme case, which can be rarely found in natural magmatic environments (in particular in basaltic systems), they allow extrapolation of the stability fields of different phases to more oxidizing conditions. At redox conditions of about >QFM + 2·5 and temperatures >1120°C, MT is the only liquidus phase, followed by HM at higher fO2 and lower T. The stability of MT is strongly controlled by the prevailing fO2 and change in redox state from QFM + 3 to QFM – 2 decreases the liquidus temperature of MT by 200°C (as observed in Fig. 2). It is noteworthy that water has almost no effect on the stability of MT. The apparent increase in the liquidus temperature as a function of H2Om observed in Fig. 1a is probably more related to an increase in fO2 than to the change in aH2O. A weak effect of aH2O is found at fO2 in the range <QFM + 1, where aH2O slightly decreases the temperature of the MT liquidus (compare slopes of MT liquidus in Fig. 1b and c as well as in Fig. 2a–e). Another notable feature of oxidizing conditions is a presence of HM as a member of ILM–HM solid solution at redox conditions >QFM + 3, followed by an ILM–HM compositional gap in the range fO2 QFM + 1 to QFM + 3 (see Figs 1a and 2a). ILM appears first at redox conditions <QFM + 1 and its stability field also shows a weak dependence on aH2O, although it is more pronounced than the dependence of the MT liquidus on water content. Remarkably, the redox conditions at which MT and ILM occur simultaneously on the liquidus vary in a narrow range from QFM – 0·5 to QFM (Fig. 2). In addition, their simultaneous crystallization in this range of redox conditions occurs independently of aH2O in the system. All these observations clearly indicate that the stability of Fe–Ti-oxides is predominantly controlled by fO2, whereas aH2O has only a minor influence on their crystallization temperatures.
In contrast to Fe–Ti-oxides, the stability fields of silicate minerals show that increasing aH2O dramatically depresses their liquidus temperatures (Figs 1 and 2). At redox conditions of <QFM + 1·5, the liquidus, defined by silicate minerals, in the dry system at 1 atm is at about 1170°C whereas the liquidus for H2O-saturated melts at 200 MPa is at 1060°C. In water-bearing systems, OL is typically the first liquidus phase, followed by CPX and PL. Amphibole is stable only at aH2O 1, T < 1000°C and it seems that this mineral crystallizes at higher T at oxidizing conditions when compared with reducing conditions (Figs 1a,c and 2a). Another remarkable difference between dry and water-bearing conditions is the OL-only stability field, which almost disappears in H2O-saturated magmas. Temperatures of the OL liquidus do not show any detectable dependence on fO2 at redox conditions <QFM + 1. The stability of CPX is also almost independent of fO2 in the investigated range, except that CPX saturation has a weak positive dependence on fO2 and T in dry and nominally dry magmas (Fig. 2d and e). The most significant influence of aH2O and fO2 on the crystallization temperatures of silicates is observed for the liquidus of PL. The largest effect of fO2 is found in H2O-saturated systems, where the PL liquidus temperature decreases from 1060°C to 950°C with an fO2 decrease from QFM + 4 to QFM – 2 (Fig. 2a). With decreasing water content, the influence of redox conditions decreases as well, diminishing almost completely in dry systems (see Fig. 2).
In general, the data for the dry SC1 system obtained by Toplis & Carroll (1995) show very good agreement with our data for the nominally dry system (compare Fig. 2a and e). The main differences are: (1) the temperature of the first crystallizing phases, which is slightly higher (20–30°C) in dry 1 atm runs compared with that in ‘nominally dry’ experiments at 200 MPa; (2) the absence of an OL–PL stability field in water-bearing experiments, which is visible in the dry system (Fig. 2e); (3) the occurrence of OL as a single phase in H2O-bearing magmas. All these observed differences can, presumably, be attributed to the presence of H2O and to the higher pressure in our experiments, both affecting the liquidus temperatures of silicate minerals and the stability of PL in particular.
Phase proportions
Calculated phase proportions are listed in Table 2 and shown in Fig 3. The fraction of residual melt decreases linearly with progressive crystallization down to about 10 wt % at 1050°C in the dry and nominally dry systems (Fig. 3a). The crystallization in H2O-saturated magmas starts at about 1060°C and from the obtained relationship between T and melt proportion we can estimate that the H2O-saturated magma should be completely crystallized at about 880–900°C. Moreover, the temperature range between the H2O-saturated liquidus and solidus is slightly larger when compared with the range in the dry system (compare slopes in Fig. 3a). The proportions of CPX and PL in general show parallel linear trends depending on XflH2Oini (Fig. 3b and c). No detectable effect of fO2 on the proportions of the melt, CPX and PL has been observed. On the other hand, the weight fractions of OL, MT and ILM–HM might have been affected by the redox conditions in the system and, hence, they show a scattering of the data plotted in terms of temperature and water activity in Fig. 3d–f. However, water has a much more significant influence on phase proportions and no pronounced systematics were observed as a function of fO2 for OL, MT and ILM–HM. An additional important factor influencing phase proportions is the resorption of OL as discussed by Toplis & Carroll (1995). However, this effect is not clearly visible from our data (Fig. 3d). The highest mineral proportions are reached for PL (up to 45 wt %), followed by CPX (<35 wt %), OL (<12 wt %) and Fe–Ti-oxides (<9 wt % for MT and <4 wt % for ILM). Proportions of AM were determined to be about 3 and 10 wt % in experiments B182 and B184, respectively (Table 2).
|
Phase chemistry
Here we present the compositions of experimental phases determined by electron microprobe (summarized in Tables 3–9
). The relationships between phase compositions and partitioning of elements between phases will be presented in the subsequent discussion.
|
|
|
|
|
|
|
Glass
The compositions of residual melts range from basalt to andesite as reported in Table 3 and shown in Figs 4 and 5, which illustrate variations in the concentrations of major elements with MgO content and temperature, respectively. It must be noted that both figures summarize all the obtained data regardless of the redox conditions of the experiment. However, changes in the redox conditions influence mainly the concentrations of FeO and SiO2 as well as, to a smaller extent, TiO2 and Na2O. At a given aH2O in the system and with increasing fO2, the variations in TiO2 and Na2O concentrations show a small decrease and increase, respectively. The decrease in TiO2 can be explained by the increasing stability of Fe–Ti-oxides. The small positive variation in Na2O is probably also related to the early crystallization of MT and delayed crystallization of PL at more reduced conditions, both leading to changes in bulk melt composition (see also Fig. 2). The relationship between the concentrations of FeO and SiO2 as a function of fO2 will be discussed below. Here we present the compositional variations of the residual melts caused by changes in water activity.
|
|
The relationship between the chemical composition of the melt and MgO content, which can be used as a magma differentiation index, is shown in Fig. 4. With decreasing MgO, the SiO2 concentrations show higher values compared with the data for the dry system, although no systematics with XflH2Oini are visible (Fig. 4a). However, as mentioned above and as be discussed below, oxidized melts have higher SiO2 contents compared with reduced ones at a given MgO. The evolution of TiO2 concentrations in Fig. 4b demonstrates the sharp peak at around 4 wt % MgO in the dry system and less pronounced TiO2 enrichment with further MgO depletion in hydrous systems. The maximum concentration of TiO2 is about 5·5 wt % and about 3 wt % in the dry and H2O-rich melts, respectively, and the concentration peak almost disappears in H2O-saturated melts. Such a behavior reflects the crystallization history of Fe–Ti-oxides, and in particular ILM. The ferrobasaltic system shows dramatic differences in the evolution paths of Al2O3 as a function of water activity (Fig. 4c). The Al2O3 concentration in the H2O-rich melts increases as a result of a delay in PL crystallization, in contrast to the rapid reduction of Al2O3 in the dry residual melts caused by the early crystallization of PL. The behavior of FeO is governed by the crystallization of minerals such as OL, CPX and Fe–Ti-oxides, and, hence, it strongly depends on aH2O and fO2. A significant increase of FeO in the residual melts is observed in the dry system whereas a wide scattering of the data is observed for the H2O-bearing melts (Fig. 4d). The evolution of CaO with decreasing MgO does not depend strongly on changes in XflH2Oini (Fig. 4e). Hydrous systems show a small enrichment in Na2O relative to dry ones (Fig. 4f). K2O content in the melts as a function of MgO is slightly lower in hydrous melts than in the dry ones, although no detectable systematic variations are observed (Fig. 4g). The lower CaO/Al2O3 ratios in the H2O-rich melts compared with those of the dry melts, as displayed in Fig. 4h, probably reflect the different pressures of the experiments (i.e. 200 and 0·1 MPa, respectively). No systematic variation in CaO/Al2O3 ratio as a function of aH2O can be resolved from our data.
The compositional trends as a function of temperature are shown in Fig. 5. As expected, the changes in concentrations of all elements are significantly delayed in H2O-saturated melts because of the depression of the liquidus temperatures of the main mineral phases with addition of H2O to the system (Fig. 5a–h). The most illustrative in this sense are the variations in concentrations of MgO, CaO and K2O (Fig. 5e, f and h, respectively), which show a systematic temperature-dependent compositional trends in systems with different aH2O. For H2O-poor conditions, the concentrations of both FeO and TiO2 show a significant increase with progressive crystallization and, after reaching a maximum, a rapid decrease with falling temperature in the dry melts (Fig. 5b andd). Such an increase in FeO and TiO2 is not observed in the H2O-saturated melts, regardless of fO2.
Olivine
The forsterite content of OL (Fo = Mg/[Mg + Fe], mol %) varies from Fo40 to Fo77 as a function of temperature, aH2O and fO2 (Table 4). The Fo content shows temperature-dependent linear trends that are almost parallel at high water activity (XflH2Oini > 0·6) (Fig. 6a). Although at lower aH2O the data are scattered, they also lie on a positive Fo vs temperature trends. The slope of the trend for the dry system is slightly steeper than the slope of the trend for the H2O-saturated system (Fig. 6a), reflecting the positive effect of aH2O on the temperature range of crystallization of OL. The Fo content of olivine at the liquidus (i.e. at T 1150°C) in the dry experiments of Toplis & Carroll (1995) agrees well with our data from the nominally dry runs. The data of Toplis & Carroll (1995) reach lower Fo values (down to Fo30) because of the higher degree of differentiation reached in their experiments (Fig. 3). However, the relationship between Fo and the magnesium number of the coexisting melt [Mg-number(Melt) = Mg/[Mg + Fe2+], mol %, which was determined accounting for the calculation of Fe2+/Fe3+ ratio in the melts using the model of Moretti (2005)] is identical for dry and hydrous systems (Fig. 6b). Two samples B111 and B112 (see Tables 3 and 4), which showed peculiar values, were excluded from this diagram. We suppose that the main reasons for this discrepancy are analytical problems with the analysis of the residual melt in highly crystallized samples and/or with the calculation of ferric–ferrous ratio of this melt (see also discussion on FeO and MgO partitioning between Melt and OL below). Figure 6c and 6d illustrates the effect of fO2 on the composition of OL. With increasing fO2 at a given temperature, the Fo content of OL increases significantly (Fig. 6c) as a result of decreasing activity of ferrous Fe in the coexisting melt (as evident from Fig. 6d).
|
Clinopyroxene
The compositions of pyroxenes (Table 5) are projected onto the triangular diagram wollastonite–(enstatite + forsterite)–tschermakite [Wo–(An + Fs)–CaTs, Fig. 7; Gaetani et al., 1993
], where the effects of pressure and aH2O on CPX chemistry can be separated (visualized). Similar to the observations of Gaetani et al. (1993), our experiments at 200 MPa demonstrate a small but systematic increase in CaTs content in CPX in comparison with the CPX compositions from the dry 1 atm experiments of Toplis & Carroll (1995). This shift is consistent with the general changes of CPX composition with increasing pressure (black arrows and outlined fields in Fig. 7; e.g. Gaetani et al., 1993). With increasing water activity, the Ca content of CPX increases and shifts the CPX compositions to the Wo corner of the triangular diagram (dashed arrow in Fig. 7).
|
The Mg-number(CPX) (= MgO/[MgO + FeO*], mol %, where FeO* is a total FeO content of CPX) in the dry system varies in a wide range from 44 to 72 with a temperature change of only 70°C (Fig. 8). With increasing aH2O, the Mg-number(CPX) becomes less sensitive to temperature variations and almost constant Mg-number(CPX) values (70–76) are observed within a T range of 110°C in the H2O-saturated system (Fig. 8a). However, the observed changes in Mg-number(CPX) are also related to the variation in fO2, as can be seen in Fig. 8b. With increasing fO2 at a given temperature, the Mg-number(CPX) values increase and the effect of redox conditions is very similar to the influence of water activity. Thus, the most oxidized and H2O-rich magmas will crystallize CPX with the highest magnesium numbers.
|
Plagioclase
The compositions of PL are reported in Table 6 and shown in Fig. 9 as a function of temperature and nominal aH2O in the system. As expected from experimental data for basalts (e.g. Sisson & Grove, 1993
; Panjasawatwong et al., 1995; Berndt et al., 2005; Takagi et al., 2005; Feig et al., 2006), the An values increase significantly with increasing aH2O (>30 mol %). The maximum An content of about 85 is determined in an experiment performed at T = 980°C and H2O-saturated conditions (run B191 in Table 6). No systematic effect of fO2 on the composition of PL was determined (Table 6).
|
The K2O contents of plagioclases vary from 0·03 to 0·33 wt %. At a given T, lower K2O concentrations are observed in PL from high-aH2O runs (Table 6). Concentrations of other minor elements such as FeO, TiO2 and MgO in PL vary over a wide range from
1 to 4 wt % for FeO and from 0·1 to 1·5 wt % for both TiO2 and MgO (Table 6). The maximum values are observed mainly in experiments with XflH2Oini = 0·6, although no systematic behavior is found as a function of temperature, fO2 or An content. The significant variations in the concentrations of FeO, TiO2 and MgO may be attributed to analytical problems related to the contamination of PL analyses by surrounding or underlying glass with high contents of FeO, TiO2 and MgO (e.g. Sugawara, 2000).
Fe–Ti-oxides
The compositions of Fe–Ti-oxides, recalculated following Stormer (1983), are listed in Tables 7 and 8 and shown in Fig. 10. The crystallization of minerals of the magnetite–ulvöspinel solid solution occurs in the entire range of investigated fO2 (i.e. from log fO2 = QFM –2 to log fO2 = QFM + 4) (Fig. 10a) and the magnetite mole fractions (XMT) cover a wide range from 0·1 to 1. Despite the scatter of the data, there is a general linear trend of XMT vs QFM that is consistent with the trend obtained for magnetite compositions from the dry experiments at 1 atm, implying that aH2O has a negligible effect on the composition of MT. The observed scattering at given QFM and aH2O (see, e.g. data at QFM + 4 and aH2O = 1) is attributed to differences in experimental temperatures, and our data indicate that XMT values increase with increasing temperature (see Table 7). Two distinct compositional ranges of oxides from the ilmenite–hematite solid solution are observed in Fig. 10b as a function of fO2. The ILM-rich compositions are restricted to the reduced conditions (log fO2 < QFM + 0·5) whereas HM-rich oxides are stable only in highly oxidized systems (log fO2 > QFM + 2·5). The ilmenite mole fraction (XILM) of ILM-rich compositions varies in a narrow range from 0·9 to unity with changes in QFM from c. +0·5 to –1·5, similar to the observations of Toplis & Carroll (1995). The effect of aH2O on XILM is not detected. HM-rich oxide compositions form a separate field at high fO2 with a wide compositional variation (XILM from 0·2 to 0·6) within a narrow log fO2 range (Fig. 10b).
|
The concentrations of MgO in MT and ILM–HM vary from
2 to >9 wt % and from 2·8 to 5·7 wt %, respectively (Tables 7 and 8). The MgO content of MT has a positive correlation with fO2 and a negative correlation with the Mg-number of the coexisting melt, whereas MgO of ILM–HM shows no systematics. The Al2O3 content varies from 3 to 6 wt % and from 0·3 to 1· 5 wt % in MTs and ILM–HM, respectively (Tables 7 and 8). In both solid solutions, the Al2O3 slightly decreases with decreasing fO2 but no dependence on T is visible.
Amphibole
Amphiboles are Mg-rich pargasites (Table 9) according to the classification of Leake et al. (2004). The ratio of tetrahedrally coordinated Al to Si in amphibole (Al/Si)AM correlates linearly with the Al/Si ratio in the coexisting melt (Al/Si)Melt, with the partition coefficient KdAl–SiAM–Melt (= [Al/Si]AM/[Al/Si]Melt; mol %) being very close to 0·94 as reported by Sisson & Grove (1993) for amphibole–melt equilibria in a wide range of magma compositions. Our small (four experiments only) dataset for AM prevents discussion of the effects of temperature and fO2 on the stability and compositional variations of AM. Mg-number(AM) (= Mg/[Mg + Fe2+]; mol %) crystallized at reducing conditions [Mg-number(AM) = 64] is slightly lower than that of AM [Mg-number(AM) = 72–74] formed at oxidizing conditions.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Partitioning of major elements between mineral and melt phases
Mg/Fe and Ca partitioning between olivine and melt
The relationships between the Fo content of OL and temperature and Fo vs Mg-number(Melt) were shown previously in Fig. 6. The apparent effect of increasing water activity and fO2 on the composition of OL at a given temperature is most probably related to differences in the degree of crystallization and changes in the ferric–ferrous ratio in the coexisting melt. The aH2O and fO2 are not expected to influence significantly the Mg/Fe partitioning between melt and OL. Similar observations were made in experimental studies on crystallization of MORB magmas (e.g. Berndt et al., 2005
; Almeev et al., 2007). Our OL–Melt pairs also follow the general Fo vs Mg-number(Melt) trend observed for OL–Melt equilibrium (Fig. 6b and d), independently of T, aH2O and fO2. The compositions of coexisting OL and melt, expressed as a partition coefficient KdOL–MeltFe–Mg (= [XOLFe/XOLMg]/[XMeltFe/XMeltMg]), are often used to test achievement of equilibrium in experimental products. In previous studies a constant KdOL–MeltFe–Mg value of about 0·3 (originally assumed to be independent of T and melt composition) was proposed as an indicator for equilibrium conditions (Roeder & Emslie, 1970). However, in a recent study, Toplis (2005) clearly illustrated a complex dependence of KdOL–MeltFe–Mg on different parameters, resulting in a significant deviation of equilibrium experimental pairs from the proposed value of 0·3 (i.e. from 0·15 to 0·45). The mole fraction of FeO in the melt in our experiments was calculated based on the determined fO2 and melt composition using the model of Moretti (2005). Most KdOL–MeltFe–Mg values for our OL–Melt pairs range between 0·29 and 0·39 [except for two samples with KdOL-MeltFe–Mg up to 44 (B193) and 57 (B112)]; see Table 4]. These high values most probably result from analytical problems with the determination of the composition of the coexisting melt and from the calculation of the ferric–ferrous ratio of the melt. The model of Moretti (2005) is calibrated mostly using the data for Fe2+/Fe3+ ratios obtained for superliquidus basaltic compositions. Although this model gave a prediction that was within the uncertainty of the experimental data for the redox state of Fe in ferrobasaltic melts (see Botcharnikov et al., 2005b), it may not always work correctly for the residual melts of highly crystallized magmas.
Calcium is a minor component of natural magmatic olivines with concentrations depending on the Fo content (e.g. Libourel, 1999), which provide constraints on the conditions of OL formation (e.g. Jurewicz & Watson, 1988; Libourel, 1999; Kamenetsky et al., 2006). The partition coefficient of CaO between OL and coexisting melt (DOL–MCaO = CaOOL/CaOmelt, wt %) increases with decreasing Fo content of OL, as illustrated in Fig. 11. Although our data are in general agreement with other experimental studies, the water-rich samples have significantly lower DOL–MCaO values compared with the model of Libourel (1999). In addition, the entire dataset has a steeper slope in Fig. 11, as recently noted by Berndt et al. (2005) and Feig et al. (2006), implying that the activity of CaO in the melt decreases with increasing activity of H2O (assuming that the equilibrium partitioning of Fe/Mg between OL and melt is not significantly affected by aH2O as discussed above). This can be presumably attributed to the preferred incorporation of Ca in PL, resulting in crystallization of An-rich PL in systems with high aH2O (see below).
|
Mg/Fe and Ca/Fe partitioning between clinopyroxene and melt
The data on Ca/Fe and Fe/Mg partitioning between CPX and coexisting melt are plotted in Fig. 12. The trend of XCaO/XFeO* ratios (where X is a mole fraction and FeO* is the total FeO content) in CPX–Melt pairs in the water-bearing samples deviates from the trend found for the dry samples of Toplis & Carroll (1995
) and, at a given XCaO/XFeO* in CPX, the XCaO/XFeOmelt ratio is higher in our study than that observed in the dry systems (Fig. 12a). The Ca/Fe partitioning is also controlled by fO2, as shown in Fig. 12b. Excluding a couple of outliers, which could be the result of analytical problems, the CPX minerals have generally lower values of XCaO/XFeO* at a given XCaO/XFeOmelt with increasing fO2. This is most probably related to the change in the activity of ferric and ferrous iron in the melt with change in redox conditions. The relationship between Mg-number of CPX and melt is shown in Fig. 12c and d as a function of aH2O and fO2, respectively. For these diagrams, we have calculated the Mg-number(CPX) taking into account only Fe2+ from the calculated structural formula of CPX. In general, the Mg-number(CPX) vs Mg-number(Melt) trend of H2O-bearing systems follows the trend of the dry system but it flattens with increasing Mg-number(Melt) (Fig. 12c). The highest Mg-number(CPX) values are also related to the most oxidized conditions in the system, as shown in Fig. 12d. In general, the relationship between Mg-number(CPX) and Mg-number(Melt) is very similar to that observed for the OL–Melt pairs plotted in Fig. 6b and d.
|
The variations of Ca/Al in CPX and melt in dry and H2O-bearing experiments are illustrated in Fig. 13. Both datasets have distinct range of values with significantly higher Ca/Al for CPX in the dry system. The wide scatter of data for the dry system (Toplis & Carroll, 1995
) is notable compared with the narrow Ca/Al range in the water-bearing system, which is almost independent of XflH2Oini. These results are in accord with the results of Feig et al. (2006) in a hydrous primitive tholeiitic system. These observations indicate that geothermometers based on calibration of CPX–Melt equilibria in dry systems (e.g. Putirka et al., 2003) need to be applied with caution for H2O-rich magmas (see also Feig et al., 2006, fig. 10).
|
Ca/Na partitioning between plagiolase and melt
The Ca/Na partitioning between coexisting PL and melt is illustrated in Fig. 14a. Continuous lines represent constant values of KdCa–NaPL–Melt (= [Ca/NaPL]/[Ca/NaMelt], mol %) and corresponding estimated H2O contents of the melt as proposed by Sisson & Grove (1993
). In this diagram the KdCa–NaPL–Melt = 3·4 line defines H2O-rich melts with an H2O content of about 4 wt %, whereas KdCa–NaPL–Melt = 1·7 and unity lines are drawn for a melt with 2 and 0 wt % H2O, respectively. With XflH2Oini increase from zero to unity, the KdCa–NaPL–Melt value increases from 0·5 to >3·5, as shown in Fig. 14b. Our data, together with those of the recent experimental studies of Berndt et al. (2005), Takagi et al. (2005) and Feig et al. (2006) for MOR and tholeiitic basalts, agree with the work of Sisson & Grove (1993) for calc-alkaline melt compositions. The good agreement between the datasets for different melt compositions indicates that the composition of plagioclase is mainly controlled by the Ca/Na ratio of the coexisting melt and the aH2O in the system (e.g. Panjasawatwong et al., 1995).
|
Liquid lines of descent and application to the Skaergaard intrusion
The experimentally determined relationships between the FeO* and SiO2 concentrations of the experimental hydrous and dry melts as a function of aH2O and fO2 are shown in Fig. 15a and b, respectively. At dry 1 atm conditions, the residual melts show an enrichment in FeO* from 13 to
19 wt % with progressive magma differentiation. The evolutionary trends of basaltic liquids from our hydrous experiments show the highest FeO* enrichment of the melt up to 16–17 wt % (Fig. 15; Table 2). It must be noted that some samples from our experiments have significant Fe loss compared with the starting SC1 composition (see also Tables 1 and 2). However, each sample must be considered independently because the loss of Fe from the system is different from sample to sample. In other words, the variations in the concentrations of FeO* and SiO2 should be interpreted taking into account not only the entire trend at a given aH2O or fO2 but also considering the Fe loss from each sample. In this sense, the samples with the highest measured FeO* concentrations (i.e. B62 and B63, Table 3) represent the actual FeO* enrichment of the melt, as the Fe loss from both samples is estimated to be about 4 relative % only. The FeO* enrichment in other samples from the experiments at reduced conditions may be underestimated, presumably by the relative amount of FeO* lost from the system (reported in Table 2). However, the exact variations in FeO* are not known, as they will depend on the entire crystallization history of the magma and it can be different in magmas with different bulk FeO* concentration. Despite the difference in the bulk FeO* content in the experimental ferrobasalts and the possible small underestimation of actual FeO* enrichment in the residual melts, the results clearly indicate that increasing aH2O and increasing fO2 considerably depress the Fe-enrichment trend (Fig. 15a and b). Although the H2O-poor and reduced samples follow, in general, the trend defined by the experimental data of Toplis & Carroll (1995), they do not reach such high values of FeO* concentrations. These significant differences in the compositional evolution of dry and hydrous ferrobasalts are attributed to the fact that the stability of MT is not affected by the presence of water in the system at a fixed fO2, whereas the stabilities of the silicate phases are strongly dependent on water activity (see also Figs 1 and 2). In other words, at given redox conditions and with increasing water activity, MT appears on the liquidus at a lower degree of differentiation, leading to significant scavenging of Fe from the melt and to the suppression of the Fenner differentiation trend. In water-rich systems, this effect becomes even more pronounced with increasing fO2, because MT appears as the first phase on the liquidus (see Fig. 2).
|
All the experimental data [including the recent data of Thy et al. (2006
)] are summarized in Fig. 15c and d, together with the predictions of the Skaergaard differentiation of Wager & Brown (1967) and Hunter & Sparks (1987) and with natural compositions. These compositions (shown as an outlined grey field) are those of dikes associated with the Skaergaard intrusion [see compilation of Thy et al. (2006)] and of flood basalts from Geikie Plateau Formation, East Greenland (Andreasen et al., 2004). Both rock types are plotted together as it is considered that the Skaergaard dikes and flood basalts are genetically related (Andreasen et al., 2004; Thy et al., 2006). As can be seen in Fig. 15c and d, the compositional evolution of the system is strongly defined by the choice of the starting composition [compare trends of Wager & Brown (1967), Hunter & Sparks (1987), Toplis & Carroll (1995) and Thy (2006)]. Nevertheless, the comparison of the proposed evolution of different Skaergaard magma compositions with the entire experimental dataset implies that the Fe enrichment observed in the Skaergaard ferrobasalts may occur only at very low aH2O and relatively reduced conditions.
This conclusion is also supported by the variation of FeO*/MgO ratios as a function of the SiO2 content of the melt as illustrated in Fig. 16. Both increasing aH2O (Fig. 16a) and increasing fO2 (Fig. 16b) move the differentiation trend of ferrobasaltic liquids away from the typical tholeiitic basaltic series to calc-alkaline compositions [according to the classification of Miyashiro (1974)]. It should be noted that the compositions of natural ferrobasalts from the Skaergaard and compositions of East Greenland flood basalts plot above the line discriminating tholeiitic and calc-alkaline series (i.e. in the tholeiitic field). Our experimental data are in agreement with the experimental results of Sisson & Grove (1993), who showed a change in the trend of liquid lines of descent from tholeiitic to calc-alkaline with increasing aH2O and fO2, recently confirmed by the experiments of Berndt et al. (2005) and the discussion of Koepke et al. (2007).
|
An important question, which is still open, is related to the initial water content of the Skaergaard parental magma. The absence of primary amphibole has commonly been used as evidence that magmatic differentiation occurred under ‘dry’ conditions. Our experiments clearly show that this observation is not the best argument for dry conditions, as amphibole is stable only below 950°C (under reducing conditions), a temperature at which most of the crystallization is already completed. However, our results indicate that the presence of significant amounts of water in the parental magma composition precludes any significant Fe enrichment during differentiation, which is characteristic for the Skaergaard intrusion (see Figs 4, 5, 15 and 16). Thus, the experimental data confirm that the Skaergaard parental magma did not contain significant amounts of water. However, the presence of fluid-saturated granophyre magmas and pegmatites, containing the H2O-bearing minerals amphibole and biotite, during the later stages of magma evolution within the Skaergaard intrusion (McBirney, 1989; Larsen et al., 1992
; Larsen & Tegner, 2006) raises a question about the role of H2O in the generation of such late-stage compositions. If we compare the compositions of granophyres with the compositional trends obtained in dry and hydrous experiments (Fig. 17), we can see that the granophyre magmas could simply have been produced by the differentiation of dry (or water-poor) ferrobasalts. In Fig. 17, the trends for K2O are not plotted because K is strongly incompatible, simply accumulating in the residual melts. The evolution of Na2O is also not very informative for the reconstruction of magma crystallization conditions (see Figs 4 and 5). It is noteworthy that for all major elements the compositional fields of granophyres lie close to the end of the compositional fields defining the differentiation trend of the Skaergaard magma obtained in the dry experiments. In particular, the large compositional gap in Al2O3 content (Fig. 17c) between the natural granophyres and the H2O-bearing experimental melts clearly indicates that such granophyres could not be generated from water-rich parental ferrobasaltic liquids. On the contrary, the small amounts of water that might have been present in the parental magmas could accumulate in the residual melts during magmatic differentiation and produce H2O-rich, late-stage granophyres. Thus, we can conclude that even the occurrence of granophyres is not an indication of a significant role for H2O in the evolution of the Skaergaard ferrobasaltic magmas.
|
Implications for the petrogenesis of the Columbia River flood basalts
Although it has been experimentally demonstrated that the parental Skaergaard magmas were almost dry during emplacement, our data for H2O-bearing ferrobasalts can be applied to understand the differentiation of other Fe-rich magma compositions; for example, some of the Columbia River flood basalts (CR). It is assumed that the CR basalts originated from the lower crust or mantle and estimations of the intensive parameters of crystallization give temperatures in the range from 1120 to 1222°C and pressures from 0 to 0·66 GPa (Caprarelli & Reidel, 2005
). Considering that Fe-rich magmas from the CR would have been too dense to erupt through the continental crust if they had been volatile-free, it has been suggested that the minimum total H2O + CO2 concentrations required for their ascent to the surface should be about 1–4 wt % based on density calculations (Lange, 2002). Another possible indication for the presence of relatively high amounts of H2O is the low proportion of phenocrysts in these magmas, although the bulk magma compositions are chemically evolved (Durand & Sen, 2004). Water, dissolved in the melt, might have kept the magmas close to their liquidus during ascent and eruption.
An extensive dataset of more than a thousand whole-rock analyses is available for the Columbia River flood basalts from the GEOROC database (GEOROC, 2008). Most of these samples have been systematically investigated in the study of Hooper (2000), who characterized the eruptive history and petrochemical trends of these basalts. The Grande Ronde (GR) basaltic series within the CR dataset is representative of 85 vol. % of the entire erupted basalt sequence. The eruption of these basalts was followed by Fe- and Ti-rich, but Si-poor, tholeiites of the Wanapum Formation (WF). The compositions of both GR and WF basalts are plotted in FeO* vs SiO2 diagrams in Fig. 15c and d. The comparison of natural GR and WF compositions with the experimental data and with the compositions of the Skaergaard intrusion indicates that the compositions of the CR basaltic magmas do not follow the dry tholeiitic trend. Although the least evolved composition of the WF ferrobasalts is close to the starting composition of the Skaergaard magma studied by Hunter & Sparks (1987), the evolutionary trend of the WF does not show any significant enrichment in FeO*. Figure 15c and d demonstrates that the compositional evolution of the CR basalts correlates best with the H2O-bearing liquid compositions obtained in the experiments. Similar observations can be made on the basis of Fig. 16, which illustrates the relationship between FeO*/MgO and SiO2 concentration. Again, the compositions of the CR basalts lie in the field of the H2O-bearing experimental compositions. It is notable that the most MgO-rich and some other compositions of the GR basalts lie close to or below the TH–CA discriminating line (Fig. 16), presumably implying that the CR magmas might have contained water and/or might have been relatively oxidized.
Figure 18 illustrates the compositions of the GR and WF basalts in comparison with the experimentally determined hydrous 200 MPa melts (closed circles for XH2Ofl = 1 and open circles for XH2Ofl <1,) and dry 1 atm basaltic compositions saturated with OL + PL, OL + CPX or OL + PL + CPX (for more details see Fig. 17). The Skaergaard evolutionary trend plots within the field with a dashed outline (for more details see Fig. 17). The least evolved composition of the WF basalts is very close to the SC1 composition used in our experiments as a starting material (Fig. 18a–f). The WF composition is slightly more evolved (i.e. has a lower CaO and MgO content; Fig. 18b) and, in addition, its Al2O3 concentration is lower by 2 wt % (Fig. 18c). In contrast, the GR basalts have very similar concentrations of CaO, MgO and Al2O3 but are significantly depleted in FeO* and TiO2 and enriched in SiO2. Hence, in the following discussion we mostly focus on the WF compositions and show the evolution of the GR series for comparison. The concentrations of CaO (Fig. 18b) and SiO2 (Fig. 18d) show a good correspondence between our data and natural WF compositions. However, these melt components do not provide clear constraints on the amounts of dissolved H2O. The important observation that can be made from Fig. 18 is that the WF basalts do not demonstrate any significant enrichment either in FeO* (Fig. 18a) or in TiO2 (Fig. 18e), unlike the evolutionary trend of the dry Skaergaard liquids or the GR series. Moreover, the WF compositional fields for FeO* and TiO2 overlap the compositional range of glasses obtained in H2O-bearing experiments (compare Figs 15 and 16). The concentration of Al2O3 is very sensitive to plagioclase crystallization, which in turn depends on the amount of dissolved water in the melt. In principle, the WF compositions do not show any significant enrichment in Al2O3, which might be predicted based on the H2O-rich melts from our experiments (Fig. 18c). On the other hand, the concentration of Al2O3 in the WF basalts does not decrease in the same way as observed for the dry basalts. Although the GR basalts show an enrichment in FeO* and TiO2, the concentrations of Al2O3 are similar to those in our water-bearing experiments, although they show a decrease with differentiation. It should be noted that the least evolved composition of the GR series is significantly different, especially in terms of FeO*, TiO2 and SiO2, from SC1 used in our experiments. Thus, a direct comparison of evolutionary trends on the Harker variation diagrams should be made with caution.
|
It must also be emphasized that the different trends defined by Al2O3 in the CR and Skaergaard basalts may be due to a difference in crystallization pressure. With increasing pressure, the stability of CPX increases, leading to an enrichment of the residual melt in Al2O3 and to a decrease in CaO/Al2O3. The calculated CaO/Al2O3 ratios (Fig. 18f) for the CR basalts agree with our data and they are lower than those of the Skaergaard magmas, indicating that both the WF and GR magmas evolved at relatively high pressures (200 MPa or more). This is in agreement with estimates based on CPX compositions (Caprarelli & Reidel, 2005
). Although other oxide concentrations and CaO/Al2O3 ratios in Fig. 18 do not confirm that the CR basalts differentiated at very high pressure under dry conditions, below we have compared the natural and experimental compositions in pseudo-ternary diagrams to better determine the effects of P and aH2O.
To distinguish between dry (at low to elevated pressure) and hydrous magmatic differentiation for the natural Columbia River basalt compositions we analyzed the liquid line of descent using simplified pseudo-ternary diagrams. Applying the recasting technique of Tormey et al. (1987) and Grove (1993), we recalculated our liquid compositions into mineral components and projected both the experimental data and the natural compositions onto CPX (diopside + hedenbergite)–PLAG (anorthite + albite)–QTZ (Fig. 19a) and OL (forsterite + fayalite)–CPX (diopside + hedenbergite)–PLAG (anorthite + albite) (Fig. 19b) pseudo-ternary diagrams through OL (+ Fe–Ti oxides) and QTZ (+ Fe–Ti oxides), respectively. These projections from the OL and QTZ apices of the basalt tetrahedron demonstrate the locations of the experimental basaltic liquids produced at dry (Toplis & Carroll, 1995, open triangles) and hydrous (this study, circles) conditions and saturated with OL + PL, OL + CPX, PL + CPX or OL + PL + CPX mineral assemblages. It should be noted that we were not aiming to constrain the positions of cotectics and the position of multiple-saturation boundaries at dry and hydrous conditions, as previously shown for high-alumina basaltic (HAB) composition (Sisson & Grove, 1993). Instead, we simply tried to demonstrate the general differences between dry and hydrous crystallization paths for the same parental basalt along OL + PL, OL + CPX, PL + CPX or OL + PL + CPX cotectics. Thus, the fields (and trends) in Fig. 19a outline melt compositions that are saturated with a different set of minerals, but have a similar range of aH2O, varying from dry to H2O-saturated conditions.
|
Under dry conditions basaltic melts experience extensive PL crystallization (PL prevails over OL and CPX in the solid phase) resulting in a shift of the residual melt composition from the PL apex towards CPX, and further towards silica-enriched compositions with progressive crystallization (Fig. 19a). In the OL + PL + CPX diagram the dry 1atm basaltic melts evolve from the PL corner in the direction of the CPX apex up to 60–70% crystallization; however, the most differentiated melts evolve in opposite direction, backwards to the PL apex. In contrast, and similar to observations made for the HAB system (Sisson & Grove, 1993
), hydrous ferrobasaltic melts are significantly displaced towards ‘PL-enriched’ compositions (Fig. 19a) and evolve toward the PL apex nearly parallel to the CPX–PL join in the projection from QTZ (Fig. 19b), as a result of the delayed crystallization of plagioclase. The natural GR (grey dashes) and WF (grey crosses) compositions do not follow the dry Skaergaard trend and overlap the melt compositions obtained in those hydrous experiments that have relatively low aH2O. Figure 19b also indicates that the compositional fields of the GR and WF basalts better correspond to H2O-bearing trends compared with the dry trend at 1 atm.
To check whether the experimental data obtained for ferrobasaltic systems are applicable to basaltic compositions and to prove that our conclusions for the CR basaltic series are valid, we performed additional comparisons using experimental data available in the literature. Using the INFOREX-3.0 database (Ariskin et al., 1996) we compiled experimental data for basaltic compositions similar to the GR and WF natural basalts (SiO2 < 55 wt %, Na2O + K2O <5 wt %). We selected only dry and hydrous basaltic liquids produced at pressures less than 600 MPa and saturated with OL + PL, OL + CPX, PL + CPX or OL + PL + CPX mineral assemblages. The recalculated compositions of melts from 77 hydrous and 303 dry experiments are plotted in Fig. 20 on similar pseudo-ternary diagrams to those in Fig. 19. The entire experimental database clearly illustrates that, despite the small overlap between hydrous and dry magmas, the position of the dry and hydrous cotectics is distinct. This confirms our conclusions made on the basis of experiments in ferrobasaltic systems.
|
To understand the possible effect of pressure on magma evolution trends, we used the recent experimental data obtained for the Snake River Plain (SRP) basalts from the study of Whitaker et al. (2007
). Their experiments were conducted over a wide range of pressures from 0·1 to 930 MPa and the experimental charges contained very low amounts of water (0·05 wt % H2O). It should be noted that the experiments with the SRP composition can be compared with the CR basalts because both basaltic provinces are genetically related to the same mantle plume (Yellowstone hotspot; Duncan, 1982; Draper, 1991). However, the starting experimental SRP composition (MgO 10·5 wt %) is significantly more primitive than the GR (MgO 6·6 wt %) and WF (MgO 5·8 wt %) parental magmas. Figure 21 summarizes the experimental data of Whitaker et al. (2007) and shows that increasing pressure influences significantly the differentiation pathway of the evolving basalts. Most of the trends defined for different pressures deviate from those obtained in our experiments or observed for the GR and WF basalts. Two compositions from the experiments at 0·1 MPa are very close to ones from natural rocks (Fig. 21a). The same compositions and one from experiments at 430 MPa plot in the field of the GR and WF series in Fig. 21b. However, it must be emphasized that these melt compositions correspond to experiments with 75–80 vol. % of crystals. Thus, these experimental data cannot be applied to the natural, mostly aphyric, basalts from the CR. In general, both diagrams in Fig. 21 demonstrate that the compositional trends of the CR basalts cannot be explained by the differentiation of dry magmas at high pressures.
|
In conclusion, examination of the possible influence of different factors such as aH2O, pressure and fO2 on the differentiation of basaltic magmas indicates that the magmas of the Grande Ronde and Wanapum Formation basaltic series of the Columbia River might have contained significant amounts of H2O (0·5–3 wt %), in agreement with previous suggestions (e.g. Lange, 2002
; Durand & Sen, 2004). |
ACKNOWLEDGEMENTS |
|---|
This work was funded by the DFG (projects Ko1723/3 and Ho1337/17). We acknowledge W. Hurkuck, B. Aichinger and O. Diedrich for technical assistance. We thank J. Berndt and M. Freise for the valuable help with experiments in the initial stage of the project, M.Portnyagin and I.Veksler for useful discussions of the experimental results, and G.Sen and S.Durand for the data on the compositions of Columbia River basalts. B. Scaillet, M. Toplis and E. Christiansen are gratefully acknowledged for their detailed, constructive and thoughtful comments, which significantly improved the scientific goals, interpretation of experimental results and implications for natural systems in this paper. The editorial work of M. Wilson is greatly appreciated.
*Corresponding author. Fax: + 49(0)511 762 3045. E-mail: R.Botcharnikov{at}mineralogie.uni-hannover.de
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Almeev RR, Koepke J, Holtz F, Parat F, Botcharnikov RE. The effect of H2O on olivine crystallization in MORB: Experimental calibration at 200 MPa. American Mineralogist (2007) 92:670–674.
Andersen JCØ, Brooks CK. The (virtual) Skaergaard intrusion. (2003) Available at www.skaergaard.org.
Andreasen R, Peate DW, Brooks CK. Magma plumbing systems in large igneous provinces: Inferences from cyclical variations in Palaeogene East Greenland basalts. Contributions to Mineralogy and Petrology (2004) 147:438–452.[CrossRef][Web of Science]
Aranovich LY, Newton RC. Experimental determination of CO2–H2O activity–composition relations at 600–1000°C and 6–14 kbar by reversed decarbonation and dehydration reactions. American Mineralogist (1999) 84:1319–1332.[Abstract]
Ariskin AA. The compositional evolution of differentiated liquids from the Skaergaard Layered Series as determined by geochemical thermometry. Russian Journal of Earth Sciences (2003) 5:1–29.[CrossRef]
Ariskin AA, Barmina GS, Meshalkin SS, Nikolaev GS, Almeev RR. INFOREX-3.0: A database on experimental studies of phase equilibria in igneous rocks and synthetic systems. 2. Data description and petrological applications. Computers and Geosciences (1996) 22:1073–1082.[CrossRef]
Bender JF, Hodges FN, Bence AE. Petrogenesis of basalts from the project Famous Area: experimental study from 0 to 15 kbars. Earth and Planetary Science Letters (1978) 41:277–302.[CrossRef][Web of Science]
Berndt J, Liebske C, Holtz F, Freise M, Nowak M, Ziegenbein D, Hurkuck W, Koepke J. A combined rapid-quench and H2-membrane setup for internally heated pressure vessels: Description and application for water solubility in basaltic melts. American Mineralogist (2002) 87:1717–1726.
Berndt J, Koepke J, Holtz F. An experimental investigation of the influence of water and oxygen fugacity on differentiation of MORB at 200 MPa. Journal of Petrology (2005) 46:135–167.
Botcharnikov R, Freise M, Holtz F, Behrens H. Solubility of C–O–H mixtures in natural melts: new experimental data and application range of recent models. Annals of Geophysics (2005a) 48:633–646.[Web of Science]
Botcharnikov RE, Koepke J, Holtz F, McCammon C, Wilke M. The effect of water activity on the oxidation and structural state of Fe in a ferro-basaltic melt. Geochimica et Cosmochimica Acta (2005b) 69:5071–5085.[CrossRef][Web of Science]
Brooks CK, Nielsen T. FD. Early stages in differentiation of Skaergaard magma as revealed by a closely related suite of dike rocks. Lithos (1978) 11:1–14.[CrossRef][Web of Science]
Brooks CK, Larsen LM, Nielsen T. FD. Importance of iron-rich tholeiitic magmas at divergent plate margins—a reappraisal. Geology (1991) 19:269–272.
Burnham CW. The importance of volatile constituents. In: The Evolution of the Igneous Rocks—Yoder HS, ed. (1979) Princeton, NJ: Princeton University Press. 1077–1084.
Caprarelli G, Reidel SP. A clinopyroxene–basalt geothermobarometry perspective of Columbia Plateau (NW USA) Miocene magmatism. Terra Nova (2005) 17:265–277.[CrossRef][Web of Science]
Carmichael I. SE. The petrology of Thingmuli, a Tertiary volcano in Eastern Iceland. Journal of Petrology (1964) 5:435–460.
Chekhmir AS, Persikov ES, Epelbaum MB, Bukhtiyarov PG. Experimental investigation of the hydrogen transport through the magmatic melt model. Geokhimiya (1985) 594–598.
Churakov SV, Gottschalk M. Perturbation theory based equation of state for polar molecular fluids: II. Fluid mixtures. Geochimica et Cosmochimica Acta (2003) 67:2415–2425.[CrossRef][Web of Science]
Devine JD, Gardner JE, Brack HP, Layne GD, Rutherford MJ. Comparison of microanalytical methods for estimating H2O contents of silicic volcanic glasses. American Mineralogist (1995) 80:319–328.[Abstract]
Di Carlo I, Pichavant M, Rotolo SG, Scaillet B. Experimental crystallization of a high-K arc basalt: the Golden Pumice, Stromboli volcano (Italy). Journal of Petrology (2006) 47:1317–1343.
Draper DS. Late Cenozoic bimodal magmatism in the Northern Basin and Range Province of Southeastern Oregon. Journal of Volcanology and Geothermal Research (1991) 47:299–328.[CrossRef][Web of Science]
Duan Z, Zhang Z. Equation of state of the H2O, CO2, and H2O–CO2 systems up to 10 GPa and 2573·15 K: Molecular dynamics simulations with ab initio potential surface. Geochimica et Cosmochimica Acta (2006) 70:2311–2324.[CrossRef][Web of Science]
Duchesne JC, Shumlyanskyy L, Charlier B. The Fedorivka layered intrusion (Korosten Pluton, Ukraine): An example of highly differentiated ferrobasaltic evolution. Lithos (2006) 89:353–376.[CrossRef][Web of Science]
Duncan RA. A captured island chain in the Coast Range of Oregon and Washington. Journal of Geophysical Research (1982) 87:10827–10837.[CrossRef]
Durand SR, Sen G. Preeruption history of the Grande Ronde Formation lavas, Columbia River Basalt Group, American Northwest: Evidence from phenocrysts. Geology (2004) 32:293–296.
Feig ST, Koepke J, Snow J. Effect of water on tholeiitic basalt phase equilibria—an experimental study under oxidizing conditions. Contributions to Mineralogy and Petrology (2006) 152:611–638.[CrossRef][Web of Science]
Fenner CN. The crystallization of basalt. American Journal of Science (1929) 18:225–253.[Web of Science]
Fisk MR, Bence AE. Experimental crystallization of chrome spinel in FAMOUS basalt 527-1-1. Earth and Planetary Science Letters (1980) 48:111–123.[CrossRef][Web of Science]
Fisk MR, Schilling JG, Sigurdsson H. An experimental investigation of Iceland and Reykjanes ridge tholeiites: I. Phase relations. Contributions to Mineralogy and Petrology (1980) 74:361–374.[Web of Science]
Ford CE. Platinum–iron alloy sample containers for melting experiments on iron-bearing rocks, minerals, and related systems. Mineralogical Magazine (1978) 42:271–275.[Web of Science]
Gaetani GA, Grove TL, Bryan WB. The influence of water on the petrogenesis of subduction-related igneous rocks. Nature (1993) 365:332–334.[CrossRef][Web of Science]
Gaetani GA, Grove TL, Bryan WB. Experimental phase relations of basaltic andesite from hole 839B under hydrous and anhydrous conditions. In: Proceedings of the Oceanic Drilling Program, Scientific Results, 135—Hawkins J, Paron L, Allan J, et al, eds. (1994) College Station: TX; Ocean Drilling Program. 557–563.
Gaillard F, Scaillet B, Pichavant M. Kinetics of iron oxidation–reduction in hydrous silicic melts. American Mineralogist (2002) 87:829–837.
Gaillard F, Pichavant M, Mackwell S, Champallier R, Scaillet B, McCammon C. Chemical transfer during redox exchanges between H2 and Fe-bearing silicate melts. American Mineralogist (2003a) 88:308–315.
Gaillard F, Schmidt B, Mackwell S, McCammon C. Rate of hydrogen–iron redox exchange in silicate melts and glasses. Geochimica et Cosmochimica Acta (2003b) 67:2427–2441.[CrossRef][Web of Science]
GEOROC. Geochemistry of rocks of the oceans and continents. (2008) Mainz: MPI für Chemie. Available at http://georoc.mpch-mainz.gwdg.de/georoc/.
Grove TL. Corrections to expressions for calculating mineral components in ‘Origin of calc-alkaline series lavas at Medicine Lake volcano by fractionation, assimilation and mixing’ and ‘Experimental petrology of normal MORB near the Kane Fracture Zone: 22°–25°N, Mid-Atlantic Ridge’. Contributions to Mineralogy and Petrology (1993) 114:422–424.[CrossRef]
Grove TL, Bryan WB. Fractionation of pyroxene-phyric MORB at low pressure: an experimental study. Contributions to Mineralogy and Petrology (1983) 84:293–309.[CrossRef][Web of Science]
Grove TL, Kinzler RJ, Bryan WB. Natural and experimental phase relations of lavas from Serocki Volcano. In:. In: Proc. ODP, Sci. Results, 106/109—Detrick R, Honnorez J, Bryan, et al, eds. (1990) College Station, TX: Ocean Drilling Program. 9–17.
Grove TL, Elkins-Tanton LT, Parman SW, Chatterjee N, Muntener O, Gaetani GA. Fractional crystallization and mantle-melting controls on calc-alkaline differentiation trends. Contributions to Mineralogy and Petrology (2003) 145:515–533.[CrossRef][Web of Science]
Hall LJ, Brodie J, Wood BJ, Carroll MR. Iron and water losses from hydrous basalts contained in Au80Pd20 capsules at high pressure and temperature. Mineralogical Magazine (2004) 68:75–81.
Hamada M, Fujii T. Experimental constraints on the effects of pressure and H2O on the fractional crystallization of high-Mg island arc basalt. Contributions to Mineralogy and Petrology (2008) 155:767–790.[CrossRef][Web of Science]
Hamilton DL, Burnham CW, Osborn EF. The solubility of water and effects of oxygen fugacity and water content on crystallization in mafic magmas. Journal of Petrology (1964) 5:21–39.
Holloway JR, Blank JG. Application of experimental results to C–O–H species in natural melts. In: Volatiles in Magmas. Mineralogical Society of America, Reviews in Mineralogy —Carroll MR, Holloway JR, eds. (1994) 30:187–230.
Holloway JR, Burnham CW. Melting relations of basalt with equilibrium water pressure less than total pressure. Journal of Petrology (1972) 13:1–29.
Hooper PR. Chemical discrimination of Columbia River basalt flows. Geochem. Geophys. Geosyst. (2000) 1:1024. doi:10.1029/2000GC000040.
Hoover JD. The chilled marginal gabbro and other contact rocks of the Skaergaard intrusion. Journal of Petrology (1989) 30:441–476.
Hunter RH, Sparks R. SJ. The differentiation of the Skaergaard intrusion. Contributions to Mineralogy and Petrology (1987) 95:451–461.[CrossRef][Web of Science]
Jurewicz A. JG, Watson EB. Cations in olivine. I: Calcium partitioning and calcium–magnesium distribution between olivines and coexisting melts, with petrologic applications. Contributions to Mineralogy and Petrology (1988) 99:176–185.[CrossRef][Web of Science]
Juster TC, Grove TL, Perfit MR. Experimental constraints on the generation of Fe–Ti basalts, andesites, and rhyodacites at the Galapagos spreading center, 85°W, and 95°W. Journal of Geophysical Research (1989) 94:9251–9274.
Kamenetsky VS, Elburg M, Arculus R, Thomas R. Magmatic origin of low-Ca olivine in subduction-related magmas: Coexistence of contrasting magmas. Chemical Geology (2006) 233:346–357.[CrossRef][Web of Science]
Kawamoto T, Hirose K. Au–Pd sample containers for melting experiments on iron and water-bearing systems. European Journal of Mineralogy (1994) 6:381–385.
Koepke J, Berndt J, Feig ST, Holtz F. The formation of SiO2-rich melts within the deep oceanic crust by hydrous partial melting of gabbros. Contributions to Mineralogy and Petrology (2007) 153:67–84.[CrossRef][Web of Science]
Kohut EJ, Nielsen RL. Low-pressure phase equilibria of anhydrous anorthite-bearing mafic magmas. Geochem. Geophys. Geosyst. (2003) 4:1057. doi:10.1029/2002GC000451.[CrossRef]
Lange RA. Constraints on the preeruptive volatile concentrations in the Columbia River flood basalts. Geology (2002) 30:179–182.
Larsen RB, Tegner C. Pressure conditions for the solidification of the Skaergaard intrusion: Eruption of East Greenland flood basalts in less than 300,000 years. Lithos (2006) 92:181–197.[CrossRef][Web of Science]
Larsen RB, Brooks CK, Bird DK. Methane-bearing, aqueous, saline solutions in the Skaergaard Intrusion, East Greenland. Contributions to Mineralogy and Petrology (1992) 112:428–437.[CrossRef][Web of Science]
Lattard D, Partzsch GM. Magmatic crystallization experiments at 1 bar in systems closed to oxygen: a new/old experimental approach. European Journal of Mineralogy (2001) 13:467–478.
Leake BE, Woolley AR, Birch WD, Burke E. AJ, Ferraris G, Grice GD, Hawthorne FC, Kisch HJ, Krivovichev VG, Schumacher JC, Stephenson N. CN, Whittaker E. JW. Nomenclature of amphiboles: additions and revisions to the International Mineralogical Association's 1997 recommendations. Canadian Mineralogist (2004) 41:1355–1362.[CrossRef][Web of Science]
Lehnert K, Su Y, Langmuir CH, Sarbas B, Nohl U. A global geochemical database structure for rocks. Geochemistry, Geophysics, Geosystems (2000) 1. 1999GC000026.
Libourel G. Systematics of calcium partitioning between olivine and silicate melt: implications for melt structure and calcium content of magmatic olivines. Contributions to Mineralogy and Petrology (1999) 136:63–80.[CrossRef][Web of Science]
McBirney AR. The Skaergaard Layered Series: I. Structure and Average Compositions. Journal of Petrology (1989) 30:363–397.
Mercer CN, Johnston AD. Experimental studies of the P–T–H2O near-liquidus phase relations of basaltic andesite from North Sister Volcano, High Oregon Cascades: constraints on lower-crustal mineral assemblages. Contributions to Mineralogy and Petrology (2008) 155:571–592.[CrossRef][Web of Science]
Miyashiro A. Volcanic rock series in island arcs and active continental margins. American Journal of Science (1974) 274:321–355.[Abstract]
Moretti R. Polymerisation, basicity, oxidation state and their role in ionic modelling of silicate melts. Annals of Geophysics (2005) 48:583–608.[Web of Science]
Morse SA. Kiglapait geochemistry: IV. The major elements. Geochimica et Cosmochimica Acta (1981) 45:461–479.[CrossRef][Web of Science]
Muntener O, Kelemen PB, Grove TL. The role of H2O during crystallization of primitive arc magmas under uppermost mantle conditions and genesis of igneous pyroxenites: an experimental study. Contributions to Mineralogy and Petrology (2001) 141:643–658.[Web of Science]
Natland JH, Dick H. JB. Formation of the lower ocean crust and the crystallization of gabbroic cumulates at a very slowly spreading ridge. Journal of Volcanology and Geothermal Research (2001) 110:191–233.[CrossRef][Web of Science]
Panjasawatwong Y, Danyushevsky LV, Crawford AJ, Harris KL. An experimental study of the effects of melt composition on plagioclase–melt equilibria at 5 kbar and 10 kbar—Implications for the origin of magmatic high-An plagioclase. Contributions to Mineralogy and Petrology (1995) 118:420–432.[CrossRef][Web of Science]
Partzsch GM, Lattard D, McCammon C. Mossbauer spectroscopic determination of Fe3+/Fe2+ in synthetic basaltic glass: a test of empirical fO2 equations under superliquidus and subliquidus conditions. Contributions to Mineralogy and Petrology (2004) 147:565–580.[Web of Science]
Pawley AR, Holloway JR, McMillan PF. The effect of oxygen fugacity on the solubility of carbon oxygen fluids in basaltic melt. Earth and Planetary Science Letters (1992) 110:213–225.[CrossRef][Web of Science]
Pichavant M, Martel C, Bourdier J.-L, Scaillet B. Physical conditions, structure, and dynamics of a zoned magma chamber: Mont Pelée (Martinique, Lesser Antilles Arc). Journal of Geophysical Research (2002a) 107:1–26.
Pichavant M, Mysen BO, Macdonald R. Source and H2O content of high-MgO magmas in island arc settings: An experimental study of a primitive calc-alkaline basalt from St. Vincent, Lesser Antilles arc. Geochimica et Cosmochimica Acta (2002b) 66:2193–2209.[CrossRef][Web of Science]
Putirka KD, Mikaelian H, Ryerson F, Shaw H. New clinopyroxene–liquid thermobarometers for maric, evolved, and volatile-bearing lava compositions, with applications to lavas from Tibet and the Snake River Plain, Idaho. American Mineralogist (2003) 88:1542–1554.
Ratajeski K, Sisson TW. Loss of iron to gold capsules in rock-melting experiments. American Mineralogist (1999) 84:1521–1527.[Abstract]
Roeder PL, Emslie RF. Olivine–liquid equilibrium. Contributions to Mineralogy and Petrology (1970) 29:275–289.[CrossRef][Web of Science]
Sano T, Fujii T, Deshmukh SS, Fukuoka T, Aramaki S. Differentiation processes of Deccan trap basalts: contribution from geochemistry and experimental petrology. Journal of Petrology (2001) 42:2175–2195.
Scaillet B, Pichavant M. Role of fO2 on fluid saturation in oceanic basalt. Nature (2004) 430. doi: 10.1038/nature02814.
Scaillet B, Pichavant M, Roux J, Humbert G, Lefevre A. Improvements of the Shaw membrane technique for measurement and control of fH2 at high temperatures and pressures. American Mineralogist (1992) 77:647–655.[Abstract]
Schwab RD, Kuestner D. The equilibrium fugacities of important oxygen buffers in technology and petrology. Neues Jahrbuch für Mineralogie (1981) 140:112–142.
Scoates JS, Cascio ML, Weis D, Lindsley DH. Experimental constraints on the origin and evolution of mildly alkalic basalts from the Kerguelen Archipelago, Southeast Indian Ocean. Contributions to Mineralogy and Petrology (2005) 151:582–599.[Web of Science]
Sisson TW, Grove TL. Experimental investigations of the size="95%"role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology (1993) 113:143–166.[CrossRef][Web of Science]
Snyder D, Carmichael I. SE, Wiebe RA. Experimental study of liquid evolution in an Fe-rich, layered mafic intrusion—Constraints of Fe–Ti oxide precipitation on the T–fO2 and T–
paths of tholeiitic magmas. Contributions to Mineralogy and Petrology (1993) 113:73–86.[CrossRef][Web of Science]
Sparks R. SJ, Meyer P, Sigurdsson H. Density variation amongst Mid-Ocean Ridge Basalts—Implications for magma mixing and the scarcity of primitive lavas. Earth and Planetary Science Letters (1980) 46:419–430.[CrossRef][Web of Science]
Spulber SD, Rutherford MJ. The origin of rhyolite and plagiogranite in oceanic crust: an experimental study. Journal of Petrology (1983) 24:1–25.
Stewart MA, Klein EM, Karson JA, Brophy JG. Geochemical relationships between dikes and lavas at the Hess Deep Rift: Implications for magma eruptibility. J. Geophys. Res. (2003) 108:2184. doi:10.1029/2001JB001622.[CrossRef]
Stolper E, Walker D. Melt density and the average composition of basalt. Contributions to Mineralogy and Petrology (1980) 74:7–12.[CrossRef][Web of Science]
Stormer J. CJ. The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multicomponent iron–titanium oxides. American Mineralogist (1983) 68:586–594.[Abstract]
Sugawara T. Thermodynamic analysis of Fe and Mg partitioning between plagioclase and silicate liquid. Contributions to Mineralogy and Petrology (2000) 138:101–113.[CrossRef][Web of Science]
Takagi D, Sato H, Nakagawa M. Experimental study of a low-alkali tholeiite at 1–5 kbar: optimal condition for the crystallization of high-An plagioclase in hydrous arc tholeiite. Contributions to Mineralogy and Petrology (2005) 149:527–540.[CrossRef][Web of Science]
Tamic N, Behrens H, Holtz F. The solubility of H2O and CO2 in rhyolitic melts in equilibrium with a mixed CO2–H2O fluid phase. Chemical Geology (2001) 174:333–347.[CrossRef][Web of Science]
Thy P, Lesher CE, Fram MS. Low-pressure experimental constraints on the evolution of basaltic lavas from site 917, southeast Greenland continental margin. In: Proceedings of the Ocean Drilling Program, Scientific Results, 152—Sunders AD, Larsen HC, Wise S. W. Jr., eds. (1998) College Station, TX: Ocean Drilling Program. 359–372.
Thy P, Lesher CE, Mayfield JD. Low-pressure melting studies of basalt and basaltic andesite from the southeast Greenland continental margin and the origin of dacites at site 917. In: Proceedings of the Ocean Drilling Program, Scientific Results, 163—Larsen HC, Duncan RA, Allan JF, et al, eds. (1999) College Station, TX: Ocean Drilling Program. 95–112.
Thy P, Lesher CE, Nielsen T. FD, Brooks CK. Experimental constraints on the Skaergaard liquid line of descent. Lithos (2006) 92:154–180.[CrossRef][Web of Science]
Toplis MJ. The thermodynamicsof iron and magnesium partitioning between olivine and liquid: criteria for assessing and predicting equilibrium in natural and experimental systems. Contributions to Mineralogy and Petrology (2005) 149:22–39.[CrossRef][Web of Science]
Toplis MJ, Carroll MR. An experimental study of the influence of oxygen fugacity on Fe–Ti oxide stability, phase relations, and mineral–melt equilibra in ferro-basaltic systems. Journal of Petrology (1995) 36:1137–1170.
Toplis MJ, Carroll MR. Differentiation of ferro-basaltic magmas under conditions open and closed to oxygen: Implications for the Skaergaard intrusion and other natural systems. Journal of Petrology (1996) 37:837–858.
Toplis MJ, Libourel G, Carroll MR. The role of phosphorus in crystallization processes of basalt: an experimental study. Geochimica et Cosmochimica Acta (1994) 58:797–810.[CrossRef][Web of Science]
Tormey DR, Grove TL, Bryan WB. Experimental petrology of normal MORB near the Kane Fracture-Zone—22–25°N, Mid-Atlantic Ridge. Contributions to Mineralogy and Petrology (1987) 96:121–139.[CrossRef][Web of Science]
Wager LR. The major element variation of the layered series of the Skaergaard intrusion and a re-estimation of the average composition of the hidden layered series and of the successive residual magmas. Journal of Petrology (1960) 1:364–398.
Wager LR, Brown GM. Layered Igneous Rocks (1967) San Francisco, CA: W.H. Freeman & Co. 588.
Walker D, Snibata T, DeLong SE. Abyssal tholeiites from the Oceanographer Fracture Zone II. Phase equilibria and mixing. Contributions to Mineralogy and Petrology (1979) 70:111–125.[CrossRef][Web of Science]
Whitaker ML, Nekvasil H, Lindsley DH, Difrancesco NJ. The role of pressure in producing compositional diversity in intraplate basaltic magmas. Journal of Petrology (2007) 48:365–393.
Wiebe RA, Snyder D. Slow, dense replenishments of a basic magma chamber—the layered series of the Newark Island layered intrusion, Nain, Labrador. Contributions to Mineralogy and Petrology (1993) 113:59–72.[CrossRef][Web of Science]
Yang HJ, Kinzler RJ, Grove TL. Experiments and models of anhydrous, basaltic olivine–plagioclase–augite saturated melts from 0·001 to 10 kbar. Contributions to Mineralogy and Petrology (1996) 124:1963–1973.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Thy, C. Tegner, and C. E. Lesher Liquidus temperatures of the Skaergaard magma American Mineralogist, October 1, 2009; 94(10): 1371 - 1376. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||