| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 3 | Pages 491-515 | 2003
© Oxford University Press 2003
Rates and Processes of Potassic Magma Evolution beneath Sangeang Api Volcano, East Sunda Arc, Indonesia
1 DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, WILLS MEMORIAL BUILDING, BRISTOL BS8 1RJ, UK
2 DEPARTMENT OF GEOLOGY AND GEOPHYSICS, THE UNIVERSITY OF ADELAIDE, ADELAIDE, SA 5005, AUSTRALIA
3 LABORATORIES OF THE GOVERNMENT GEOCHEMIST, QUEENS ROAD, TEDDINGTON PW11 0LY, UK
4 SCHOOL OF EARTH SCIENCES, UNIVERSITY OF TASMANIA, HOBART, TAS 7001, AUSTRALIA
5 DEPARTMENT OF EARTH SCIENCES, MEMORIAL UNIVERSITY OF NEWFOUNDLAND, ST. JOHNS, NEWFOUNDLAND A1B 3X5, CANADA
Telephone: (44) 0117 9545440. Fax: (44) 0117 9253385. E-mail: simon.turner{at}bristol.ac.uk
RECEIVED DECEMBER 7, 2001; ACCEPTED SEPTEMBER 17, 2002
 |
ABSTRACT |
|---|
U-series isotopes can provide unique insights into the physical processes of magma evolution by constraining the time scales over which they operate. This, however, requires rock suites that provide a clear and complete record of the liquid line of descent. Sangeang Api volcano, in the east Sunda arc, provides such an opportunity because it erupts potassic lavas (SiO2
47–55%), which contain a spectrum of xenoliths interpreted to represent the cumulates complementary to the lavas. Combined, the cumulates and lavas span a large compositional range, which major and trace element modelling suggests reflects 70% polybaric crystallization that began at sub-Moho depths and continued into the upper crust. The parental magmas can be successfully modelled by 3% dynamic partial melting, in the presence of 0–4% residual garnet, of a mid-ocean ridge basalt (MORB) source enriched by 3% subducted Sunda sediment in addition to a contribution of fluid-mobile elements from the subducting slab. The effects of fluid addition and partial melting on U–Th disequilibria appear to be competing processes. A moderate range in Sr, Nd and Pb isotopes is interpreted to reflect a combination of source heterogeneity and <15% assimilation of Indian MORB crust. Neither interaction with metasomatized arc lithosphere nor the presence of enriched, plume-type mantle in the mantle wedge is required by our data. The cumulates and lavas have largely indistinguishable (230Th/232Th) over a wide range of U/Th ratios and thus any age differences are minimal relative to the half-life of 230Th. All whole rocks and minerals are characterized by 226Ra excesses and modelling of the 226Ra–230Th–Ba data suggests that magmatic evolution beneath this arc volcano occurs on time scales of 2000 years. Combining this with simple numerical calculations we estimate that the Sangeang Api magma chamber is 6–10 km3 in size, cooling rates are 0·05°C/yr and minimum crystal growth rates are (2–7) x 10-13 cm/s. It is possible that a significant proportion of the crystal growth and differentiation occurred during isobaric decompression of magmas ascending through conduits. An implication is that the net magmatic flux across the Moho is of magmas that are already significantly evolved from primary magmas and this may be significant for why average continental crust has an andesitic bulk composition even though the flux out of the mantle wedge is basaltic. KEY WORDS: time scales; lava; xenolith; Sunda arc; potassic magma
|
INTRODUCTION |
|---|
Primary magmas are rarely erupted from island arc volcanoes and much effort has been expended in unravelling the mechanisms by which they evolve to produce the more differentiated magmas typically erupted. Of particular interest are the time scales of this magmatic differentiation and how those might vary with magma composition and differentiation path. Numerical models can be used to constrain both the rate of cooling and crystallization in crustal magma chambers (e.g. Marsh, 1998
). If magmatic differentiation occurs by crystal settling, the time scale will also be controlled by the size of the crystals, their density contrast with the host magma and the magma viscosity (Martin & Nokes, 1988). The behaviour of natural systems is clearly complex; phenocrysts may not necessarily settle in a tholeiitic magma because the liquid line of descent yields dense, iron-rich differentiates, whereas the higher silica content and thus greater viscosity of calc-alkaline magmas at a similar degree of differentiation may also inhibit phenocryst–liquid separation, albeit in a different way. Thus, there is a need for independent observational constraints on differentiation time scales.
In principle, time scale information can be obtained from studies of U-series disequilibria in magmatic rocks but their application to this problem remains largely in its infancy and most studies have concentrated on within-plate lavas [see Hawkesworth et al. (2000) for a recent review]. The available arc data indicate differentiation times of 102–103 years for basaltic to dacitic, tholeiitic and calc-alkaline lavas (Turner et al., 2000, 2001), although Heath et al. (1998) presented evidence for longer time scales of 60 kyr at Soufrière, St. Vincent, in the Lesser Antilles. In the case of potassic arc lavas no detailed studies have yet been undertaken. This is despite the fact that these are often the most explosive and hazardous volcanoes (e.g. Tambora, Self et al., 1984; Foden, 1986) for which time scale information might aid in hazard prediction. In within-plate settings, potassic lava differentiation time estimates range widely from 200 years for mugearite at Mt. Etna (Condomines et al., 1995) to 3000 years for phonolite at Mt. Erebus (Reagan et al., 1992) and 90–100 kyr for trachytes and phonolites from São Miguel and Laacher See, respectively (Widom et al., 1992; Bourdon et al., 1994). Finally, the origin of rear-arc potassic lavas themselves has been the source of considerable debate, with some studies concluding that an increase in the relative contribution from subducted sediment is sufficient to explain the observed data (e.g. Rogers et al., 1985) whereas other have argued that interaction with incompatible-element-enriched sub-arc lithospheric mantle saturated with potassic phases is required (e.g. Edwards et al., 1991).
Here we present the results of a detailed isotope study of Sangeang Api from the eastern Sunda arc in Indonesia. This volcano provides an opportunity to shed new light on the rates of processes of potassic magma evolution because its recent eruptive products include a series of potassic lavas and cumulate xenoliths whose mineral paragenesis is consistent with the progressive differentiation history shown by the lavas. U–Th–Ra isotope disequilibria are utilized to constrain the processes and time scale for this differentiation. Finally, neighbouring Tambora makes for an interesting comparison with Sangeang Api because its magmas appear to evolve along the same liquid line of descent.
|
SANGEANG API VOLCANO AND ITS ERUPTIVE PRODUCTS |
|---|
Sangeang Api volcano is an active alkaline volcano lying within an older caldera and has a known historical history of eruptions spanning 1512–1988. There have been at least 13 moderate-sized (105–107 m3), explosive eruptions in the last century and activity was almost continuous from 1953 to 1958, then was renewed by eruptions in 1964–1966 and most recently by more major (107–108 m3) eruptions in 1985–1998 (Simkin & Siebert, 1994
). The volcano is located at the rear of the east Sunda arc (Fig. 1a), 190 km above the Benioff zone and adjacent to the active volcano Tambora (Fig. 1b). The crust in this portion of the arc is composed of magmatically and tectonically thickened oceanic crust with the Moho lying at 14–17 km depth (Curray et al., 1977). The erupted products at Sangeang Api comprise potassic, volatile-enriched, silica-undersaturated lavas that exhume an abundant population of entrained clinopyroxene-rich, mafic and ultramafic xenoliths. The lavas range from potassic trachybasalt to trachyandesite in composition (47–55 wt % SiO2) with 5–15% normative-ne and are characterized by three distinct phenocryst assemblages: clinopyroxene–olivine; plagioclase–clinopyroxene–magnetite (Fig. 2a); and plagioclase–amphibole–clinopyroxene–magnetite. The latter two assemblages are typical of lavas with relatively evolved geochemical characteristics. Their compositions are similar to those of the potassic Pleistocene volcanoes Soromundi and Sangenges just to the west (Foden & Varne, 1980) and Batu Tara to the NE (Stolz et al., 1988; van Bergen et al., 1992), which also contain pyroxene-rich xenoliths (see Fig. 1 for volcano locations). The eruptive products of Sangeang Api contrast with those of the arc-front volcanoes, such as Rindjani (Fig. 1b), which are predominantly calc-alkaline in nature and generally lack xenoliths (e.g. Foden, 1983).
|
|
The xenoliths, which are interpreted as fragments of cumulate magmatic bodies within the arc crust, provide an insight into magmatic differentiation processes in the system feeding Sangeang Api. The cumulate xenoliths represent the products of progressive crystal–liquid differentiation of the parental magmas and can be divided into two main petrological groups. Group 1 are clinopyroxenites that consist of Ca-, and sometimes Cr-rich clinopyroxene with lesser olivine with variable magnetite contents (from <1% to
15%). They contain small amounts of pargasitic amphibole and/or phlogopite interstitial to the clinopyroxene. In some samples, amphibole occurs as a primary cumulate phase. Group 2 are gabbroic xenoliths with variable amphibole contents. These are plagioclase–magnetite–clinopyroxene rocks with amphibole contents that range from 4 to 50%. Included in this group are some amphibole-rich samples in which the amphibole has clearly undergone incongruent partial melting as a consequence of decompression (Foden & Green, 1992). This group includes the most recently exhumed xenoliths carried from the actively erupting cone by lahars in 1985–1988. These are distinctive from the other xenoliths in that they contain intergranular brown glass with hopper-quench-textured crystallites of plagioclase, clinopyroxene and amphibole (Fig. 2b). Thus, the cumulate xenoliths have broadly equivalent mineral assemblages to the phenocrysts of the lavas. A plagioclase crystal size distribution plot for one of the lavas is shown in Fig. 2c and is markedly concave upwards, indicating crystal accumulation (Marsh, 1998), consistent with some of the larger phenocrysts reflecting entrainment of crystals from disaggregated cumulate xenoliths.
Comparison between experimentally determined phase relations and those of the cumulate xenoliths suggests crystallization at temperatures around 1100–1000°C and at up to moderately elevated crustal pressures of 0·5–1 GPa from parental melts with >3% H2O (Foden & Green, 1992). Elevated H2O contents are also implied by the early precipitation of plagioclase-free assemblages, and somewhat elevated pressures are supported by the broad field of clinopyroxene crystallization in the xenoliths by comparison with the expanded field of olivine and plagioclase in the lavas.
|
ANALYTICAL TECHNIQUES |
|---|
The samples analysed here were selected from a larger sample suite of 33 lavas and 37 xenoliths ( J. Foden, unpublished data, 2002). They consist of eight lavas and seven undeformed cumulate xenoliths (one pyroxenite and six gabbros) all of which come from eruptions in 1953–1958 and 1985–1988 (the ages of the lavas are indicated in Table 1). For bulk-rock analyses, both lavas and cumulate xenoliths were crushed in a steel jaw-crusher and then milled in a tungsten-carbide mill at the University of Adelaide. Analysis of milled quartz indicates a Ta blank of
0·24 ppm for this mill and so the Ta data for the lavas can only be taken as maxima (italics in Table 1) and no data for the xenoliths are reported. The pyroxenite (SA24) and one of the gabbroic cumulate xenoliths (SA21) were separated into their constituent minerals. To obtain these mineral separates, the xenoliths were disaggregated by gently tapping with a hammer; between 200 and 2000 mg (or individual grains for 87Sr/86Sr analyses) of plagioclase, clinopyroxene, amphibole, magnetite and groundmass were then carefully hand-picked under a binocular microscope. Magnetite was separated and purified by repeatedly passing a small magnet over the crushed whole rock, followed by hand picking. The crystals in the cumulate xenoliths are remarkably free from inclusions (e.g. Fig. 2b) and the most likely form of impurity in the separates is small patches of adhering interstitial glass. The separates were visually estimated to be 99% pure and this was subsequently verified by comparison of in situ and bulk concentration data for Th and Ba (see Appendix). The mineral major element compositions were determined using a JEOL 733 electron microprobe at the University of Adelaide and ferric iron was calculated from stoichiometry for the clinopyroxenes and by using the program RECAMP for the amphiboles (Spear & Kimball, 1984).
|
Whole-rock major and selected trace elements (including Cl) were analysed by X-ray fluorescence with a Philips PW1480 100 kV spectrometer at the University of Adelaide. Reproducibility and accuracy are better than 10% for Ba, Ni, Cu and Cr, and better than 5% for all other elements. Cs, Zn, Co, Ta, Hf and rare earth element (REE) abundances were obtained either by inductively coupled plasma mass spectrometry (ICP-MS) or by instrumental neutron activation analysis (INAA). The lavas were analysed by ICP-MS at Boston or Cardiff universities following techniques similar to those described by Elliott et al. (1997)
and Turner et al. (1999). The xenoliths were analysed by INAA at the Open University following the techniques of Potts et al. (1985). Sr and Nd isotope ratios were measured dynamically on a single-collector Finnigan MAT 261 system at the University of Adelaide. Sr was fractionation corrected to 86Sr/88Sr = 0·1194 and Nd to 146Nd/144Nd = 0·7219. Measured values for multiple determinations of the NBS 987 and La Jolla standards over the period of the study were 0·710271 ± 20 and 0·511821 ± 14 (2
), respectively. Pb was analysed statically in temperature-controlled runs (1250°C) in multi-collector mode on a Finnigan MAT 262 system at the Open University and the ratios were corrected for 1
/a.m.u. mass-fractionation using the recommended values for NBS 981 (Todt et al., 1996). Blanks for Sr, Nd and Pb were typically <1 ng, 500 pg and 300 pg, respectively.
Th, U and 226Ra concentrations and 230Th/232Th isotope ratios were determined at the Open University by thermal ionization mass spectrometry on a high abundance sensitivity Finnigan MAT 262 system equipped with an RPQ-II energy filter. Samples were spiked with 229Th–236U and 228Ra tracers and dissolved using HF–HCl–HNO3 in Savillex beakers (the mineral separates were cleaned in an ultrasonic bath of purified water before dissolution). Treatment with HCl and H3BO3 was used to obtain completely clear solutions and to ensure sample-spike equilibration. Details of the U–Th separation and analysis techniques have been given by Turner et al. (1997) and procedures for 226Ra purification and analysis were identical to those described by Turner et al. (2000). U–Th mass spectrometric procedures were similar to those described by van Calsteren & Schwieters (1995). U, Th and 226Ra concentrations were determined to a precision of better than 0·5% (1) and the U and Th isotope ratios have an external reproducibility of 0·5% (1), which was monitored using the standard solution Th'U'std. Reproducibility of (226Ra/230Th) is estimated as 1·3% (Turner et al., 2000). Total procedural blanks for U and Th were typically 100 and 50 pg, respectively, and for 226Ra were <0·1 fg/g, and these are insignificant relative to the 30–300 ng of U and Th and 20–200 fg of Ra that was separated and loaded. Determinations of the AThO and TML Th standards yielded (230Th/232Th) = 1·017 ± 0·005 (n = 5) and (230Th/232Th) = 1·079 ± 0·005 (n = 2) during this period. One of the lavas and two of the xenoliths have 234U excesses that are outside of analytical error of unity by 1–2% and these may reflect minor contamination, although these samples do not have systematically higher 226Ra or 87Sr/86Sr. Repeat determinations of the Mt Lassen Ra standard yielded an average of 226Ra = 1065 ± 9 fg/g, which is within error of the values quoted by Volpe et al. (1991). Ba concentrations in the bulk mineral separates were determined by isotope dilution using a 135Ba tracer. Decay constants used in the calculation of activity ratios and age corrections were: 
238U = 1·551 x 10-10 a-1; 234U = 2·835 x 10-6 a-1; 232Th = 4·948 x 10-11 a-1; 230Th = 9·195 x 10-6 a-1; 226Ra = 4·332 x 10-4 a-1. All whole-rock (226Ra/230Th) ratios are age corrected to 1985, excepting sample 48067, which was corrected to 1953.
|
RESULTS |
|---|
Lavas
Major and trace element and isotopic compositions of the lavas are listed in Table 1. They range in composition from very primitive samples such as 48067 with 9·5% MgO and Mg-number = 0·70 [where Mg-number = Mg/(Mg + 0·85Fe2+tot)] to fairly evolved intermediate lavas with 2·6% MgO and 54% SiO2. All samples are much more potassic than the Sunda arc-front lavas, with between 1·8 and 3·9% K2O, and lie in the shoshonitic field in Fig. 3a. Olivine- and clinopyroxene-compatible trace element concentrations are low (Ni 49–5 ppm, Cr 232–5 ppm) and FeO*/MgO increases rapidly from one to two at low SiO2 contents and then increases more gradually as SiO2 begins to rise (Fig. 3b). Sc shows a continuous curvilinear decrease from 49 to 9 ppm with increasing SiO2 (Fig. 3c), whereas Sr levels remain almost constant, suggesting a bulk distribution coefficient around one (see inset to Fig. 4). Mantle-normalized incompatible trace element patterns show extreme enrichments of Ra, Ba, K and Sr, and negative Ta–Nb and Ti anomalies, all of which are characteristic of island arc lavas (Fig. 4). The concentrations of incompatible trace elements increase with SiO2 (see insets to Fig. 4) and range from high (e.g. Nb 5–7 ppm) to extremely high with increasing incompatibility (e.g. Ba 1057–1764 ppm). High halogen contents are indicated by chlorine contents ranging from 900 to 4850 ppm. 87Sr/86Sr and 143Nd/144Nd are negatively correlated and range from 0·7046 to 0·7050 and 0·51274 to 0·51261, respectively (Fig. 5a). 208Pb/204Pb, 207Pb/204Pb and 206Pb/204Pb are in the range of 38·87–39·17, 15·58–15·67 and 18·85–19·00, respectively, and 208Pb/204Pb and 206Pb/204Pb show a strong positive correlation at elevated 208Pb/204Pb relative to the Northern Hemisphere Reference Line (NHRL) but at lower 208Pb/204Pb and higher 206Pb/204Pb than the arc-front lavas (Fig. 5b).
|
|
|
|
|
Cumulate xenoliths
Major and trace element and isotopic compositions of the whole-rock cumulate xenoliths are given in Table 2. The major element composition of the xenoliths is largely dictated by their mineralogy and they invariably lie at the SiO2-poor end of the arrays defined by the lavas in Fig. 3. Most of the xenoliths have SiO2 <50% whereas MgO varies from 14·1 to 5·2% and K2O from 0·1 to 0·8% (Fig. 3). Four xenoliths from the larger database ( J. Foden, unpublished data, 2002) have K2O of 1·3–2·5%. Concentrations of the compatible trace elements Ni, Cr and Sc are remarkably low except in the pyroxenite SA24 (Ni 62 ppm, Cr 226 ppm), suggesting that most of them were precipitated from already differentiated magmas. Incompatible trace elements show a wide concentration range (e.g. Ba 208–830 ppm and Nb 0·8–8·7 ppm) but are generally lower than those in the host lavas (Fig. 4). The incompatible trace element patterns of the xenoliths are largely controlled by interstitial glass (see below) and so broadly mirror those observed in the host lavas (Fig. 4). An exception is for Sr in the gabbroic cumulate SA21, which is strongly enriched consistent with its high modal proportion of plagioclase. Isotopically, the xenoliths generally overlap with the lavas with 87Sr/86Sr, 143Nd/144Nd, 208Pb/204Pb, 207Pb/204Pb and 206Pb/204Pb in the range of 0·7047–0·7052, 0·51274–0·51262, 38·81–39·00, 15·57–15·59 and 18·81–18·93, respectively (Fig. 5).
Mineral analyses and apparent partition coefficients
Compositional and isotopic data for the mineral separates from cumulate xenoliths SA24 and SA21 are presented in Table 3 along with concentrations of Ba, Th, U and 226Ra, which were determined by isotope dilution. As noted above, the mineral separates will inevitably have contained some adhering glass that would have had an influence on the measured concentrations of highly incompatible elements. However, although these concentrations must therefore represent an upper limit for those in the pure minerals, calculations suggest that the amount of adhering glass was no more than 1% (see Appendix) and thus the bias caused by adhering glass is small. In contrast, comparison of the composition of the bulk cumulates and their separated mineral phases (Table 3) shows that the concentrations of Ba, Th, U and 226Ra in the bulk cumulates are clearly dictated by the interstitial glass component.
|
Sufficient glass and clinopyroxene were separated from the gabbroic cumulate xenolith SA21 to allow INAA measurements (in addition to the isotope dilution measurements of Ba, U, Th and 226Ra) and calculation of clinopyroxene–melt partition coefficients. The interstitial glass composition is more evolved, in terms of major elements, than that of any of the lavas, although its incompatible trace element concentrations are similar. This may reflect the presence of magmas evolved from a range of parental compositions (see below), but is consistent with the incompatible trace element concentrations in the cumulates being dictated by the interstitial glass. Apparent mineral–melt partition coefficient values calculated for the SA21 clinopyroxene (see Table 3) are rather high (e.g. 0·8 for Ti, 9·1 for Sc, 0·013 for Ba, 0·12 for La, 0·61 for Yb, 0·43 for Ta, 0·65 for Hf ) but are similar to those experimentally determined for aluminous clinopyroxenes from alkali basalts (Blundy et al., 1998
; Hill et al., 2000; Wood & Trigila, 2001). For cumulate xenolith SA21, four clinopyroxene, plagioclase and amphibole phenocrysts and four glass fragments were also analysed for 87Sr/86Sr composition to check for the presence of entrained, non-cognate, material and for isotopic equilibrium between the constituent phases. Table 3 shows that the results of the 16 measurements are essentially indistinguishable from each other and the whole-rock analysis within analytical error (±20).
|
MAGMATIC EVOLUTION BENEATH SANGEANG API |
|---|
Differentiation
The Sangeang Api lavas are volatile rich and incompatible element rich with very high Cl and Ba contents. In detail, the lavas show evidence for a segmented differentiation trend. As shown in Fig. 6a, the first stage of differentiation, involving lavas that contain olivine and clinopyroxene phenocrysts, produces rapid enrichment of Al2O3 with slight SiO2 enrichment whereas Sc (Figs 3c and 6b) and MgO (not shown) undergo continuous depletion. This trend then gives way, in the more plagioclase-phyric lavas, to a second stage in which Al2O3 remains constant with increasing SiO2. These two stages also correspond to an inflection in the FeO*/MgO vs SiO2 trend in Fig. 3b. Most importantly, the xenoliths show both geochemical and mineralogical affinities with the lavas, with low-Al2O3 olivine pyroxenites being linked to the first stage of differentiation and the Al2O3-rich gabbros to the second stage of differentiation. Thus, the liquid line of descent is one of protracted olivine and clinopyroxene fractionation through which low-alumina, undersaturated, potassic trachybasalt evolves to an iron-rich, high-alumina trachybasalt and subsequently to a trachyandesite once plagioclase comes onto the liquidus. The absence of plagioclase suggests that the first stage is likely to have occurred at pressures approaching 1 GPa, which would place it below the Moho in this part of the arc (Curray et al., 1977
), although elevated H2O contents may have helped to delay the crystallization of plagioclase. The assemblages in the gabbroic cumulate xenoliths indicate that the second stage is likely to have occurred at shallower depths and perhaps lower H2O contents (Foden & Green, 1992).
|
The inferred petrogenetic sequence was investigated more quantitatively via least-squares mixing calculations that used the mineral analyses from Table 3. The primitive sample 48067 was used as the ultimate parent and the first stage attempted to derive the composition of sample SA10 from this, after which the second stage attempted to derive the composition of the most evolved lava, SA5, from SA10. The best-fit proportions of phases are summarized in Table 4 and in Fig. 6a, and broadly demonstrate that
40% fractionation of olivine and clinopyroxene (equivalent to the pyroxenitic cumulate xenoliths) can generate the first stage lava trend, after which extraction of a gabbroic assemblage (equivalent to the gabbroic cumulate xenoliths) will yield the second stage of liquid evolution via a further 30% fractionation. The sums of residuals squared are <1 for both stages (Table 4). This can be significantly reduced to <0·1 for the second stage if amphibole is included in the calculation; however, these results are not presented because amphibole appears to occur as a primary phase only in some of the cumulate xenoliths and in the more differentiated lavas.
Trace element modelling was performed using the phase proportions determined from the least-squares modelling. This is inherently non-unique because of the range in published partition coefficients. However, models that do provide a reasonable approximation of the data highlight the need for elevated partition coefficients for Sc and Zr in clinopyroxene (Table 4, Fig. 6b), approaching the apparent partition coefficients that were calculated from the data in Table 3. The high compatibility of Sc and Zr reflects the Al-, Na- and Fe3+-rich nature of these clinopyroxenes (see Table 3) and recent experimental data show that this is a characteristic of clinopyroxenes crystallized from potassic magmas (Hill et al., 2000; Wood & Trigila, 2001). The major and trace element compositions of the bulk solids for stages 1 and 2 (calculated from Table 4) are plotted in Figs 3 and 6, where they can be compared with the compositional range of the cumulate xenoliths and the liquid line of descent implied by the array defined by the lavas.
It should be noted that the dataset plotted in Fig. 6 involves lavas and cumulate xenoliths of a range of ages (compared with the materials analysed for isotopes, which were restricted to those from the 1953 and 1985–1988 eruptions). Therefore, they cannot be comagmatic in the strictest sense, although the persistence of the trends does indicate that the magmas repeatedly evolved along similar liquid lines of descent. Additionally, incompatible trace element abundances vary somewhat irregularly and some of the most primitive samples, such as 48067, have higher abundances of incompatible trace elements than the main trend of the lavas (see Fig. 6b). This, and the isotopic variability discussed below, points to some geochemical diversity in the parental magmas and may indicate that there was that more than one magma batch present in the system at any given time. Finally, we note that many of the xenoliths contain intergranular glass and it seems likely that much of the scatter amongst the xenoliths relative to the calculated liquid and cumulate trends results from magma–cumulate mixing (see Fig. 6a).
Crustal interaction vs source heterogeneity
Although the major and trace element data can be explained by closed-system, polybaric differentiation, the lavas and cumulate xenoliths also span a range of radiogenic isotope compositions (Fig. 5). In principle, this could reflect either source-level heterogeneity or crustal-level assimilation. The inflections in the major and trace element differentiation trends rule out simple, two end-member mixing, and the variation in incompatible element abundances in the more mafic lavas and the presence of both 238U and 230Th excesses (see below) argue that at least two magma batches have been in the Sangeang Api plumbing system recently. Whether the full extent of the observed Sr, Nd and Pb isotope variation existed in the magma source region is harder to constrain, and the presence of correlations between radiogenic isotope ratios and indices of differentiation (Fig. 7a) has led us to also explore a model in which differentiation was accompanied by assimilation (e.g. DePaolo, 1981).
|
First, it is clear that the putative contaminant cannot have been radiogenic, silicic crust (e.g. Davidson, 1987
) because 87Sr/86Sr decreases with increasing differentiation (Fig. 7a). The low 87Sr/86Sr lavas are also those with the highest Rb/Sr (inset to Fig. 7a), suggesting that the most likely candidate for a contaminant is a partial melt of young and unradiogenic basalts that were formed in equilibrium with residual plagioclase at pressures <1 GPa within the crust. In principle, these could be old arc basalts similar to those currently erupting along the arc-front into which the low 87Sr/86Sr, high 143Nd/144Nd end of the Sangeang Api array projects in Fig. 5a. However, in Fig. 5b, the Sangeang Api lavas form a steep array that back-projects to a composition similar to many Indian mid-ocean ridge basalts (MORB) with 208Pb/204Pb 38·1 and 206Pb/204Pb 18·55 (Mahoney et al., 1992). Clearly, different contaminants could be invoked; however, because the intra-oceanic Sunda arc was built on Indian oceanic crust (Foden & Varne, 1980), these appear the most plausible candidates. An end-member model in which all of the variation was attributed to crustal-level assimilation of Indian MORB would, on the basis of Pb isotope mixing calculations (Fig. 5b), require 15% assimilation, which would equate to a rate of assimilation to fractionation (r) (DePaolo, 1981) of 0·5. We stress that this represents an upper limit and that we believe that source heterogeneity is responsible for much of the radiogenic isotope variation (see below).
Source composition and extent of partial melting
Recent studies have suggested that the Sunda arc-front lavas are derived from MORB-source mantle wedge peridotite that has been enriched through additions of both sediment and fluid components from the subducting plate (Hoogewerff et al., 1997; Turner & Foden, 2001). It has also been suggested that the rear-arc lavas show evidence for an increasing relative contribution from subducted sediment, a correspondingly weaker fluid signal and a decrease in the degree of partial melting. Here we explore to what extent the nature of the primitive Sangeang Api magmas can be modelled in an analogous manner. A plot of 143Nd/144Nd vs Th/Ce was used by Turner & Foden (2001) to investigate the sediment contribution to the source of the lavas because the rare earth elements (REE) and thorium are largely fluid-immobile. This diagram is reproduced in Fig. 7b and, as can be seen, none of the Sunda arc lavas can be explained by the bulk sediment–mantle mixing curve. On this basis, Turner & Foden (2001) suggested that the sediment component must have been added to the mantle wedge as a partial melt. In this case, the bulk sediment–mantle mixing curve provides an upper estimate of the amount of sediment contribution and the 143Nd/144Nd ratios of the more primitive Sangeang Api lavas (0·51265) indicate that this could reach 2·5%. More problematic is that the Sangeang Api array does not intersect either that defined by the arc-front lavas or the sediment–MORB-source mixing line in Fig. 5b. At present, the only potential explanation is that the sediment subducting beneath Sangeang Api is more radiogenic in Pb than estimated for the average Sunda sediment by Plank & Langmuir (1998).
The degree of partial melting and residual mineralogy are rarely well constrained in arc lavas, but using the most recent partition data for clinopyroxene on the mantle solidus (Blundy et al., 1998) we have used a plot of Tb/Yb vs La/Yb to estimate 2–4% partial melting assuming an I-MORB source containing 2·5% sediment as the source composition (Fig. 7c). Unlike La/Yb, which is sensitive to both the degree of partial melting and percentage of residual garnet, Tb/Yb is largely sensitive to only the percentage of residual garnet and so the higher Tb/Yb ratios in some of the Sangeang Api lavas imply the presence of a few percent residual garnet. A working model is presented in Fig. 8a, which compares the incompatible trace element composition of a model 3% batch melt of an I-MORB source containing 2·5% Sunda sediment and 3% residual garnet with the most primitive Sangeang Api lava. Although the patterns of the lava and the modelled composition are broadly similar in shape, there are obvious mismatches. However, the deficits in U and Sr in the modelled composition are inferred to reflect the need for additions of these elements in a fluid phase, whereas the excesses of Ta, Nb and Ti are taken as further evidence that the sediment component was added as a partial melt (assumed to be in equilibrium with residual rutile, which retained these high field strength elements).
|
Comparison with other rear-arc, potassic volcanoes
The Sunda arc is characterized by a number of rear-arc, potassic volcanoes, and early geochemical studies emphasized a positive correlation between K2O content and depth to Benioff zone (e.g. Hutchison, 1976
). However, the general applicability of this has been questioned (Foden & Varne, 1980) and here we review models for the origin of the potassium enrichment through a comparison of the Sangeang Api data with recent studies of Tambora (Foden, 1986), Batu Tara (van Bergen et al., 1992; Hoogewerff et al., 1997) and Muriah (Edwards et al., 1991). Tambora lies at a similar distance above the Benioff zone (180 km) to Sangeang Api (190 km) whereas Batu Tara and Muriah are located further back at 280 km and 370 km above the Benioff zone, respectively (Fig. 1a, Hutchison, 1982). Consistent with this, Tambora shows striking parallelism in its differentiation path to Sangeang Api in Figs 3 and 6 and has a near-identical incompatible trace element pattern in Fig. 4. In contrast, Fig. 3a shows that, with the exception of an older potassic series from Muriah, the Batu Tara and Muriah lavas are far more potassic than those from Sangeang Api or Tambora. Isotopically, the Muriah lavas have lower 143Nd/144Nd at a given 87Sr/86Sr and lower 206Pb/204Pb (Fig. 5) whereas the Batu Tara lavas are more radiogenic in Sr and Pb isotopes and have less radiogenic Nd (none of these lava suites show the negative correlation between 87Sr/86Sr and SiO2 observed for Sangeang Api in Fig. 7a). Preliminary radiogenic isotope data on a sample from the 1815 Tambora eruption (Turner & Foden, 2001) indicate much less radiogenic Sr and more radiogenic Nd compared with any of the other rear-arc potassic volcanoes (Fig. 5a), although this Tambora sample does lie at the extension of the Sangeang Api Pb isotope, 87Sr/86Sr–SiO2 and 87Sr/86Sr–Rb–Sr arrays in Figs 5b and 7a.
Potential role of enriched mantle and potassic phases
The Muriah and Batu Tara lava suites show 143Nd/144Nd–Th/Ce trends in Fig. 7b that are shallower than, and diverge from, the arc-front lava array. Mixing lines are straight in such diagrams and the implication is involvement of a component with low Th/Ce and 143Nd/144Nd 0·5126. The so-called EMI mantle component (Zindler & Hart, 1986) is inferred to have low Th/Ce and low 143Nd/144Nd, and van Bergen et al. (1992) invoked the presence of plume material in the mantle wedge to explain the enrichment of incompatible trace elements in lavas erupted at Batu Tara volcano to the NE of Sangeang Api. However, ocean island lavas erupted above mantle plumes are characterized by Nb/U ratios of 45–50 (Hofmann et al., 1986), whereas the Sangeang Api lavas have Nb/U ratios of 2–4. These low Nb/U ratios reflect the negative Nb anomalies in Fig. 4 and, although these are poorly modelled by bulk sediment addition in Fig. 8a, and may require addition of sediment melts (see above), this fit becomes far worse if plume-type mantle is used in the model. Thus, despite the overt simplicity of our modelling, we see no necessity to invoke the presence of plume material in the mantle wedge.
Edwards et al. (1991) suggested that the highly potassic Muriah lavas reflected interaction with phlogopite-bearing metasomatized, sub-arc, lithospheric mantle, and Foden & Green (1992) have argued for a role for the incongruent breakdown of amphibole in the genesis of some potassic arc magmas. In Fig. 3a, it is clear that K remains incompatible throughout the Sangeang Api and Tambora differentiation series rather than being buffered by a residual potassic phase. Furthermore, phlogopite has high Rb/Sr whereas amphibole is characterized by high Ba/Sr, and this allows a distinction to be made in diagrams such as Fig. 8b (Xu et al., 2001). The data from Muriah are too scattered for any discrimination to be made, whereas those from Batu Tara form steep trends in Fig. 8b, which may point to a role for phlogopite either as a residual source phase or else in the fractionating assemblage of these lavas. In contrast, the Sangeang Api and Tambora lava suites form inclined arrays in Fig. 8b, indicating that phlogopite and amphibole are unlikely to have been residual source phases or important fractionating phases. Of course, we cannot rule out the possibility that amphibole, present in the sub-arc, lithospheric mantle, melted out completely and thus contributed to the potassic nature of the Sangeang Api magmas. However, K is one of the better modelled elements in Fig. 8a and so, until compelling evidence to the contrary is presented, we conclude that a simple model involving small-degree partial melting of a sediment-enriched mantle source region is adequate to explain the observed data.
|
U–Th–Ra ISOTOPE DISEQUILIBRIA |
|---|
Given the well-constrained petrological and chemical relationships between the Sangeang Api lavas and cumulates, this suite provides an excellent opportunity for investigating the processes and time scales of magma generation and differentiation. To this end we selected lavas and cumulate xenoliths from the 1953 and 1985–1988 eruptions that span the compositional array in Fig. 6 for U–Th–Ra isotopic measurements.
Whole-rock U–Th–Ra results
The whole-rock results for the lavas and cumulate xenoliths are presented in Tables 1 and 2 and in Figs 9 and 10. The Th contents of the lavas range from 6·8 to 12·4 ppm and their (230Th/232Th) ratios exhibit a very restricted range (0·73–0·77) over a much greater (238U/232Th) range of 0·65–0·90. Thus, the lavas preserve both Th and U excesses, with (230Th/238U) ranging from 1·15 to 0·85 (Fig. 9a). The Tambora sample analysed by Turner & Foden (2001) has a significantly higher (230Th/232Th) ratio of 0·8 and lies closer to the U–Th equiline in Fig. 9a. 226Ra concentrations in the lavas range from 1687 to 2458 fg/g and (226Ra/230Th) varies from 1·64 to 3·06. These 226Ra excesses are unlikely to reflect contamination by seawater because the 87Sr/86Sr ratios of the lavas are too low (see Turner et al., 2000) and the largest 226Ra excesses occur in the lavas with the lowest Cl contents (Table 1). Two of the cumulate xenoliths (SA19 and SA20) have lower (238U/232Th) than any of the lavas (Fig. 9a) and their elevated P2O5 and LREE contents suggest that this reflects the presence of apatite in these two xenoliths. Together, the cumulate xenoliths span a wide range of Th contents and (238U/232Th) ratios (0·41–2·78 ppm and 0·40–0·86, respectively), although their (230Th/232Th) ratios are also relatively invariant (0·73–0·75) and indistinguishable from the lavas (see Table 3). The 226Ra concentrations of the cumulate xenoliths are lower than those observed in the lavas (60–492 fg/g) but (226Ra/230Th) ratios extend from 0·98 to 5·36.
|
|
Mineral U–Th–Ra results
The mineral separate data from the two cumulate xenoliths are presented in Table 3 and in Figs 9 and 10. Their (238U/232Th) ratios range from 0·60 to 0·92 whereas (230Th/232Th) remains the same as the whole rocks at 0·74–0·79, leading to a range in (230Th/238U) from 1·30 to 0·84 (Table 3). 226Ra–232Th disequilibria range from (226Ra/230Th) = 0·76 to 8·63. The Th, U and Ra contents of the constituent minerals are much lower than the whole-rock cumulate xenolith analyses. The implication is that the U–Th–Ra composition of the cumulate xenoliths is dictated by the interstitial glass (see Table 3) and this is consistent with the composition of the interstitial glass from cumulate xenolith SA21, which has 238U and 226Ra excesses similar to an evolved Sangeang Api lava. The corollary is that fractionation of the cumulate minerals would have little leverage on the melt composition and cannot, therefore, be responsible for the U–Th–Ra disequilibria observed in the lavas. In detail, the (230Th/232Th) ratios of the minerals from SA24 are indistinguishable from each other and the whole-rock analysis (Fig. 9b). The results from SA21 are more complex (Fig. 9c), with the glass and whole rock having slightly lower (230Th/232Th) than the minerals whereas the clinopyroxene appears to have slightly higher (230Th/232Th) than the other minerals. Nevertheless, the key point is that the minerals do not form a positively sloped array (see below).
Interpretation
Source processes
The U–Th isotope data are plotted on conventional (230Th/232Th) vs (238U/232Th) equiline diagrams in Fig. 9; the lavas straddle the equiline, exhibiting both 230Th excesses and 238U excesses (Fig. 9). Because these signals are likely to be developed only in the mantle (see below), the presence of both senses of U–Th disequilibria supports our preferred interpretation that much of the variation in Sr, Nd and Pb isotopes reflects source heterogeneity. Additionally, as U and Th are both highly incompatible in the minerals in the inferred fractionating assemblage, the fact that the lavas span a wide range in (238U/232Th) provides further evidence that they are not all derived from the same primary magma and therefore, even though different batches followed similar liquid lines of descent (see Fig. 6), more than one magma batch must have been present in the Sangeang Api system in the recent past.
238U excesses are characteristic of arc lavas and inferred to result from U addition by fluids from the subducting plate [see Hawkesworth et al. (1997) for a recent review]. Thus, the 238U excesses observed in five of eight of the analysed Sangeang Api lavas are interpreted to reflect fluid addition to their mantle source regions. Basalts erupted at mid-ocean ridges and ocean islands are characterized by 230Th excesses and these are usually interpreted to reflect a source where DU > DTh as a result of the effects of residual garnet (Beattie, 1993) or aluminous clinopyroxene (Wood et al., 1999). However, 230Th excesses are far less common in arc-front lavas and this has generally been interpreted to reflect an absence of residual garnet and the dominance of U addition by fluids (Hawkesworth et al., 1997). In the case of Sangeang Api, the depth to the Benioff zone is 190 km and there is evidence in elevated Tb/Yb ratios for the role of residual garnet (Fig. 7c), which could potentially overprint the effects of U addition by fluids in some magma batches. In Fig. 10a we plot Tb/Yb vs (238U/230Th), showing that the Sangeang Api lavas form a negative array and, by reference to Fig. 7c, it would appear that when the amount of residual garnet exceeds a few percent the effects of partial melting exceed those of fluid addition, resulting in magma batches with elevated Tb/Yb that have 230Th excesses. Thus, the effects of fluid addition and partial melting on U–Th disequilibria appear to be competing processes in the rear-arc environment. Importantly, the similarity in mantle bulk partition coefficients for U and Th (Wood et al., 1999) means that for melt fractions in excess of 0·5% 230Th excesses cannot be produced by simple batch or fractional melting but require in-growth in an upwelling matrix that is passing through the melting zone [see Elliott (1997) for a recent review]. Thus, given the 2–4% degrees of partial melting inferred from Fig. 7c, some form of dynamic partial melting and upwelling in the mantle beneath Sangeang Api seems to be required to explain the observed 230Th excesses.
A recent global survey has shown that 226Ra excesses in arc lavas show a positive correlation with fluid-sensitive indices such as Ba/Th (Turner et al., 2001) that are inferred to be added by fluids released during dehydration of the subducting plate. The inset to Fig. 10a shows that the Sangeang Api lavas are no exception to this and seawater contamination has already been ruled out as an origin for the 226Ra excesses. Furthermore, the largest 226Ra excesses are usually found in the more primitive lavas, implying a mantle derivation for this signal (Turner et al., 2001). We have considered the possibility that the 226Ra excesses derive from dissolution of potassic phases within the lithospheric mantle beneath the arc. However, as noted above, large U-series disequilibria in general require in-growth in an upwelling matrix (Elliott, 1997), which seems hard to envisage in a static, lithospheric mantle (Asmerom et al., 2000). In fact, because Ra (like Ba) will be partitioned into amphibole relative to Th (see Table 3), the excesses of 226Ra over 230Th in the Sangeang Api lavas provide additional evidence that amphibole was neither a residual source phase nor a major fractionating phase. In conclusion, we interpret the Sangeang Api 226Ra excesses to have resulted from fluid addition to their mantle source region, consistent with studies of 226Ra excesses in the Sunda lavas in general (Hoogewerff et al., 1997; Turner & Foden, 2001). Importantly, to preserve this signal to the surface requires that the combined time for melt extraction, transport through the mantle and time spent during differentiation within the lithosphere is significantly less than 8000 years (Turner et al., 2001).
Differentiation time scales
The U–Th equiline diagrams in Fig. 9 also emphasize that the lavas, cumulate xenoliths and their constituent minerals all lie on or close to a horizontal array and certainly there is no evidence of any positive correlation between (230Th/232Th) and (238U/232Th) that could be interpreted as an isochron. As we have shown, the crystallization of these minerals and the formation of the cumulates was the predominant driving mechanism for the evolution of the compositional array of the lavas and the simplest explanation for why all of the materials analysed lie on or close to a horizontal array in Fig. 9 is that this evolution occurred on a time scale that was very short relative to the half-life of 230Th (75 kyr). Such observations are not restricted to the rear-arc volcanoes. Rubin et al. (1989) presented U–Th isotope data from minerals from Batur volcano that formed an array whose slope was within error of the eruption age of the lava, implying similar short residence times for that magmatic system, and Turner & Foden (2001) have argued that recent differentiation beneath Galunggung volcano occurred in less than a few thousand years.
226Ra has a half-life of only 1600 years and can therefore be used to place tighter constraints on the time scale of differentiation. On a plot of (226Ra/230Th) vs Th (Fig. 10b), the lavas form a negative array. It should be noted that those lavas with 230Th excesses and 238U excesses have been identified with different symbols on this diagram as they are likely to have evolved from slightly different melt batches even though they demonstrably subsequently evolved along very similar differentiation trends (see Figs 3 and 6). Th is an incompatible element and so can be used as an index of differentiation but because Ra is even more incompatible (Tables 3 and 4), modelled Ra/Th ratios increase slightly during differentiation and cannot reproduce the negative array of the Sangeang Api lavas in Fig. 10b. An alternative explanation is that the negative array simply reflects the effects of the inferred assimilation of partial melts of Indian MORB crust that had very low or equilibrium (226Ra/230Th). However, the high concentrations of incompatible elements in such model melts result in mixing curves that are strongly convex-downwards and also do not match the data (Fig. 10b). In contrast to incompatible element concentrations, the Th isotope system is insensitive to assimilation because the 208Pb/206Pb isotope ratio inferred for the Indian MORB contaminant modelled in Fig. 5b suggests an equilibrium (230Th/232Th) ratio of 0·8, similar to the Sangeang Api lavas themselves. Even 20% assimilation will not shift the Th isotope ratio of the lavas by more than 2%. Thus, although it is likely that some assimilation occurred, it does not appear to have been the principal control on the U–Th–Ra isotope systematics.
Our preferred interpretation stems from the fact that the differentiation model plotted in Fig. 10b does not have any time dependence and therefore effectively simulates instantaneous differentiation. In a time-dependent model, the slope of the negative array reflects decay of 226Ra during differentiation and the time elapsed during differentiation can be constrained by the half-life of 226Ra to be of the order of 2000 years (see calibration in Fig. 10b). It should be noted that, to constrain the maximum likely amount of time elapsed during differentiation, our model assumes a similar magnitude of 226Ra–230Th disequilibria in all parental magma, which is similar to the largest observed in the Sunda lavas such as Iya (see Fig. 1a for location) using data from Turner & Foden (2001). If this value represents an overestimate, then the time scales inferred from Fig. 10b will also be overestimated. Similarly, any enhancing of the rate of decrease of (226Ra/230Th) through the effects of assimilation would mean that the inferred time scales represent maximum estimates.
In principle, the Ra–Th isotope data from the cumulate xenoliths can be used to verify the differentiation time scales inferred from Fig. 10b because they have been shown to be complementary to the liquid line of descent followed by the lavas (Fig. 6). The whole-rock cumulate xenolith data do not provide a particularly robust test because their Ra–Th disequilibria have been demonstrated to be largely dictated by the composition of the interstitial glass. However, using the method of Cooper et al. (2001) described in the Appendix, it is possible to calculate (226Ra)/Ba vs time evolution curves to constrain when the cumulate minerals crystallized. The intersection of the curves for melts in equilibrium with clinopyroxene and amphibole from the pyroxenitic cumulate xenolith SA24 suggests that crystallization occurred 1500–2300 years ago (Fig. 11a). The results from the gabbroic cumulate xenolith SA21 indicate crystallization between 1900 and 2000 years ago from the intersection of the curves for melts in equilibrium with clinopyroxene, plagioclase, magnetite and the glass (Fig. 11b). The overlap between the ages from the two cumulates suggests that the first stage of differentiation was rapid and arguably commenced no longer than 300 years before the second stage. It should be noted that the concordant intersection of the SA21 glass with the curves for melts in equilibrium with the clinopyroxene and plagioclase in this cumulate would not be expected if the glass represented an unrelated liquid that infiltrated the cumulate some time after its crystallization. This suggests that the interstitial glass is very close in 226Ra–230Th–Ba composition to the liquid from which the cumulate minerals crystallized. The SA21 amphibole is more complex to interpret because the calculated melt in equilibrium with the amphibole has (226Ra)/Ba ratios >1000. It intersects the curve for melt in equilibrium with the clinopyroxene over an age range of 100–400 years but does not intersect the curves for the other phases in Fig. 11b. There are three possible, and not mutually exclusive, explanations. First, the 226Ra–230Th–Ba systematics of the amphibole are likely to have been disturbed by the incongruent partial melting observed in thin section. Second, the amphibole might reflect reaction between the clinopyroxene and a hydrous fluid sometime between 100 and 400 years ago. Third, it is possible that the amphibole crystallized from a liquid with a different 226Ra–230Th–Ba composition to the rest of the minerals. At present we do not have enough information to choose between these alternatives and accordingly we caution against attributing significance to the age range implied by the intersection of the curves for melts in equilibrium with the amphibole and clinopyroxene.
|
Physical implications
The ages of crystals can often only constrain the maximum age of magma residence beneath volcanoes because of the potential for magmas to entrain old cumulate materials (Hawkesworth et al., 2000
). An important result from Sangeang Api is that, notwithstanding some of the complexities discussed above, the constraints from the cumulates appear to be in excellent agreement with the differentiation time scale inferred from the whole-rock lavas in Fig. 10b. This provides encouragement that the ages represent true magma residence times and that 70% differentiation, possibly coupled with some assimilation, occurred within a maximum of 2300 years beneath the volcano. It can be estimated that for a basaltic magma to undergo 70% crystallization requires a temperature drop (
T) of 100°C below the solidus (Grove & Bryan, 1983; Ghiorso & Carmichael, 1985; Grove et al., 1990). If this temperature drop was linear through time, then a cooling rate of 0·05°C/yr is implied by the Sangeang Api data, and this is similar to minimum recent estimates from a study of Kilauea (Cooper et al., 2001).
The volumes of magma chambers are difficult to constrain but a recent approach based on the thermal output at volcanoes (Hawkesworth et al., 2000) equates the volume of a magma chamber to
 |
cool is the time taken for cooling, 
is the density, c the specific heat capacity, L the latent heat of crystallization, T the temperature drop and 
the mass fraction of crystals grown. Assuming = 0·7 from the major element modelling, T = 100°C from above, cool = 2000 years from the Ra–Th data, and using values of 2600 kg/m3, 1500 J/kg per K and 4 x 105 J/kg for , c and L, respectively, that are appropriate for basalt magmas (Spera, 2000), we calculate a volume of 6 km3 for a power output of 100 MW. It is important to note that the actual power output from Sangeang Api volcano is currently unknown; however, we suggest that this calculation may represent a conservative estimate because 100 MW lies at the low end of the power output range estimated for volcanoes generally (100–1000 MW, Hochstein, 1995). Moreover, a similar volume is estimated assuming that the magma chamber is in steady state (e.g. Pyle, 1992; Condomines, 1994):
 |
0·005 km3 (Simkin & Siebert, 1994) and 3·5 x 10-4 for Qo and ft, respectively, yields a volume of 10 km3.
Given a magma chamber volume of 6 km3 and setting the chamber height to H = V1/3, the characteristic time scale for crystal settling in a convecting magma can be calculated using the model of Martin & Nokes (1988):
 |
the density contrast between the crystals and liquid, and r is the radius of the crystals. For µ = 104 Pa s, = 500 kg/m3 and r = 10-3 m, settle = 660 years, which is about a third of the total magma residence time and may imply some crystal retention, which would be consistent with the convex-upward nature of the crystal size distribution plot in Fig. 2c. The crystal growth rates can be calculated from the relationship (Marsh, 1998)
 |
is the average crystal residence time and m is the slope of the crystal size distribution plot. Because this crystal size distribution is curved, we have considered it in two halves, corresponding to fine and coarse crystal populations, which have slopes of -2·29 and -8·66 mm-1, respectively (see Fig. 2c). Using an average = 2000 years from the combined Ra–Th data leads to plagioclase growth rates of 7 x 10-13 and 2 x 10-13 cm/s for the coarse and fine crystal populations, which are likely to correspond to those crystallized at deeper and shallower crustal levels, respectively. It should be noted that this calculation assumes continuous growth over the full 2000 years. Moreover, we have presented evidence that the Sangeang Api magmas did not evolve isobarically in a single magma chamber but underwent polybaric crystallization. Isothermal decompression can lead to crystallization time scales that are much more rapid than those predicted by crystallization as a result of cooling alone and can approach those of eruptive periodicity (Blundy & Cashman, 2001). For these reasons, the calculated growth rates probably represent minimum estimates and we hypothesize that much of the differentiation of the Sangeang Api lavas may have occurred during ascent in conduits through the arc lithosphere. Crystallization would occur on the walls of the conduits and the depolymerizing effects of elevated halogen and alkali contents would enhance the segregation of the liquids from the phenocrysts. If magma moves in pulses, as suggested by seismic periodicity (e.g. Neuberg, 2000), pieces of this cumulate material lining the conduits would periodically become entrained in the magmas to be erupted as the cumulate xenoliths.
Magma residence times beneath Tambora volcano
The cataclysmic eruption of Tambora in April 1815 (e.g. Self et al., 1984) produced some 100 km3 of potassic magmas, which Foden (1986) suggested contained 3% H2O and underwent crystallization at 850–700°C at 0·5 GPa. As discussed above, Tambora shows a striking parallelism in its overall composition and liquid line of descent to Sangeang Api. Thus, the inferred degrees of partial melting, volatile contents and differentiation appear similar, whereas the eruptive temperatures and volumes for Tambora were lower and higher, respectively. However, the lower temperatures and higher volumes may be explained if the 1815 magma accumulated, resided and cooled for longer times in the crust. In fact, this is exactly what is suggested by the displacement to lower (226Ra/230Th) in Fig. 10b. Moreover, if the analogy with Sangeang Api holds, then the (226Ra/230Th) ratio of the 1815 Tambora lava implies a crustal residence time of 5000–6000 years, and we are intrigued that this is almost identical to the time elapsed since the preceding major eruption of Tambora 5000 years ago (Foden, 1986). A detailed U-series study of the Tambora lavas and cumulate xenoliths would be timely in testing this greater residence time hypothesis.
|
CONCLUDING REMARKS |
|---|
A working model for the evolution of magmas beneath Sangeang Api is summarized in Fig. 12. The composition of the most primitive lavas is consistent with 3% partial melting, in the presence of 1–4% residual garnet, of a MORB source enriched by 2·5% subducted Sunda sediment and a contribution of fluid-mobile elements from the subducting slab. Schmidt & Poli (1998) have shown that it is reasonable to expect some water to be subducted this deep and the increase in relative sediment contribution, compared with the arc-front lavas (Turner & Foden, 2001
), may reflect the increase in slab-surface temperatures at depth. The effects of fluid addition and partial melting on U–Th disequilibria appear to be competing processes in the rear-arc environment. Major and trace element data indicate that the lavas and cumulate xenoliths erupted from Sangeang Api volcano represent complementary parts of a polybaric liquid line of descent that involved some 70% equilibrium crystallization and potentially also some assimilation of Indian MORB arc crust. The time scales of this magmatic differentiation are inferred from U–Th–Ra isotope data to be of the order of 2000 years, similar to recent estimates based on along-arc data from the Tonga–Kermadec (Turner et al., 2000) and other oceanic arcs (Turner et al., 2001). The evidence that the first 30–40% of (pyroxenitic) differentiation appears to have occurred in less than 300 years at pressures >1 GPa may imply that this occurred during magma ascent in conduits. Moroever, our data suggest that the net magmatic flux across the Moho is of relatively evolved magmas that had left significant cumulate material behind in the upper parts of the lithospheric mantle (Fig. 12, Foden & Green, 1992). This is consistent with recent seismic studies that have identified mafic to ultramafic cumulates at the Moho in the Aleutian arc (Fliedner & Klemperer, 1999) and could help explain why average continental crust has an andesitic bulk composition even though the flux out of the asthenosphere is basaltic (Arculus, 1981; Ellam & Hawkesworth, 1988). Subsequent ponding and gabbroic fractionation within the crust occurred over a time scale of <2000 years, limiting the time available for crustal assimilation.
|
|
APPENDIX: (226Ra)/Ba VS TIME EVOLUTION DIAGRAMS |
|---|
 TOPABSTRACTINTRODUCTIONSANGEANG API VOLCANO AND...ANALYTICAL TECHNIQUESRESULTSMAGMATIC EVOLUTION BENEATH...U-Th-Ra ISOTOPE DISEQUILIBRIACONCLUDING REMARKS APPENDIX: (226Ra)/Ba VS TIME... REFERENCES |
|---|
Previously, ‘isochron’ ages have been determined from Ra–Th isotope data by normalizing to the chemically similar element Ba (e.g. Volpe & Hammond, 1991
), but it is now clear that the underlying assumption that DBa = DRa is erroneous (Wood et al., 1999). Following the elegant approach of Cooper et al. (2001), we determined the time-dependent (226Ra)/Ba composition of melts in equilibrium with the analysed mineral phases using the following equation:
 |
is the decay constant for 226Ra (4·332 x 10–4 a-1) and DRa/Ba values (Table 4) were calculated for a temperature of 1100°C using the method of Blundy & Wood (1994). Crystallization ages are then estimated from the point at which the (226Ra)/Ba ratios of the melts in equilibrium with the minerals are concordant with each other and/or the groundmass (Fig. 11).
An additional complexity is that application of these theoretical partition coefficients assumes that the bulk mineral separates are 100% pure, which is rarely the case because of small amounts of glass adhering to the mineral edges and/or the presence of mineral and/or melt inclusions. The former can be removed by treatment with HF, but this can also leach U-series nuclides and cannot be used to remove inclusions. The Sangeang Api cumulate crystals appear very homogeneous and free of inclusions (Fig. 2b), and we believe that adhering glass is the most likely possibility in this instance. Therefore, we adopted an approach whereby the concentrations of Ba and Th in the bulk separates and pure minerals were compared to identify the presence of adhering glass. The Ba and Th concentrations in the groundmass and bulk mineral separates were determined by isotope dilution, whereas the Ba and Th contents of the pure minerals were determined by ICP-MS and ion microprobe, respectively, taking care to avoid inclusions. The laser ICP-MS analyses were performed at the University of Bristol using methods similar to those of Norman et al. (1996) and using NIST SRM 612 as a standard. The ion microprobe data were obtained at the University of Edinburgh using 450 s count times and NIST SRM 610, 612, 614, 616 and Corning Glass W as standards. The results and the calculated percentage of adhering glass for each phase are listed in Table A1. The percentage of adhering glass was always close to 1%, consistent with optical estimates, and, where both measurements were made, the amounts calculated on the basis of Ba and Th contents are generally in good agreement with each other. The one exception is for the clinopyroxene from SA24, where the estimated percentage of adhering glass based on Th concentrations is 49% compared with 1% based on Ba (Table A1). The former estimate is untenable on the basis of optical observations and so the latter value was adopted in this case. These estimates were then used to back-extract the effects of glass and melt inclusions from the bulk mineral separate compositions by simple mass balance. This requires that the composition of the interstitial glass is known; for SA21 we used the composition of the analysed glass but this was not possible for SA24 because insufficient glass was present to facilitate analysis. Therefore, for SA24, we used the composition of the bulk xenolith as the only available estimate of the composition of the interstitial glass and because it appears likely that the composition of interstitial glass dominates the incompatible element concentrations of the xenoliths (Table 3). Although this partly compromises the results from SA24 the age constraints obtained were consistent with those from SA21, which also have a smaller estimated error (Fig. 11). No corrections were possible for SA24 amphibole or SA21 magnetite because concentration data for the pure minerals could not be obtained (Table A1).
|
|
ACKNOWLEDGEMENTS |
|---|
David Bruce, Bruce Paterson, Louise Thomas, Nick Rogers and Peter van Calsteren are all thanked for their assistance with the analytical work. Bernie Wood and Jon Blundy kindly calculated the Ra/Ba partition coefficient ratios, and we are grateful to Derek Vance, who performed the single-crystal Sr-isotope analyses for sample SA21 at ETH. Terry Plank and Julian Pearce kindly provided the ICP-MS trace element analyses, and Richard Hinton performed the ion microprobe analyses at the University of Edinburgh. The crystal size distribution was made by Neil Carpenter at Bristol. Careful reviews by Kari Cooper, Ian Parkinson, Mark Reagan and an anonymous reviewer helped to significantly improve the paper. We also thank Marjorie Wilson for her extensive editorial comments. This work was supported by NERC grant GR3/11701 to S.T., who is funded by a Royal Society University Research Fellowship.
|
FOOTNOTES |
|---|
Deceased. 
|
REFERENCES |
|---|
Arculus, R. J. (1981). Island arc magmatism in relation to the evolution of the crust and mantle. Tectonophysics 75, 113–133.[CrossRef][Web of Science]
Asmerom, Y., Cheng, H., Thomas, R., Hirschmann, M. & Edwards, R. L. (2000). Melting of the Earth's lithospheric mantle inferred from protactinium–thorium–uranium isotopic data. Nature 406, 293–296.[CrossRef][Medline]
Beattie, P. (1993). Uranium–thorium disequilibria and partitioning on melting of garnet peridotite. Nature 363, 63–65.[CrossRef]
Blundy, J. & Cashman, K. (2001). Ascent-driven crystallisation of dacite magmas at Mount St Helens. 1980–1986. Contributions to Mineralogy and Petrology 140, 631–650.[Web of Science]
Blundy, J. D. & Wood, B. J. (1994). Prediction of crystal–melt partition coefficients from elastic moduli. Nature 372, 452–454.[CrossRef]
Blundy, J. D., Robinson, J. A. C. & Wood, B. J. (1998). Heavy REE are compatible in clinopyroxene on the spinel lherzolite solidus. Earth and Planetary Science Letters 160, 493–504.[CrossRef][Web of Science]
Bourdon, B., Zindler, A. & Worner, G. (1994). Evolution of the Laacher See magma chamber: evidence from SIMS and TIMS measurements of U–Th disequilibria in minerals and glasses. Earth and Planetary Science Letters 126, 75–90.[CrossRef][Web of Science]
Condomines, M. (1994). Comment on: ‘The volume and residence time of magma beneath active volcanoes determined by decay-series disequilibria methods’. Earth and Planetary Science Letters 122, 251–255.[CrossRef][Web of Science]
Condomines, M., Tanguy, J.-C. & Michaud, V. (1995). Magma dynamics at Mt Etna: constraints from U–Th–Ra–Pb radioactive disequilibria and Sr isotopes in historical lavas. Earth and Planetary Science Letters 132, 25–41.[CrossRef][Web of Science]
Cooper, K. M., Reid, M. R., Murrell, M. T. & Clague, C. A. (2001). Crystal and magma residence at Kilauea volcano, Hawaii: 230Th–226Ra dating of the 1955 east rift eruption. Earth and Planetary Science Letters 184, 703–718.[CrossRef][Web of Science]
Curray, J. R., Shor, G. G., Raitt, R.W. & Henry, M. (1977). Seismic refraction and reflection studies of crustal structure of the Banda and eastern Sunda arc. Journal of Geophysical Research 82, 2479–2489.
Davidson, J. P. (1987). Crustal contamination versus subduction zone enrichment: examples from the Lesser Antilles and implications for mantle source compositions of island arc volcanic rocks. Geochimica et Cosmochimica Acta 51, 2185–2198.[CrossRef][Web of Science]
DePaolo, D. J. (1981). Trace element and isotopic effects of combined wallrock assimilation and fractional crystallisation. Earth and Planetary Science Letters 53, 189–202.[CrossRef][Web of Science]
Edwards, C., Menzies, M. & Thirlwall, M. (1991). Evidence from Muriah, Indonesia, for the interplay of supra-subduction zone and intraplate processes in the genesis of potassic alkaline magmas. Journal of Petrology 32, 555–592.
Ellam, R. M. & Hawkesworth, C.J. (1988). Is average continental crust generated at subduction zones? Geology 16, 314–317.
Elliott, T. (1997). Fractionation of U and Th during mantle melting: a reprise. Chemical Geology 139, 165–183.[CrossRef][Web of Science]
Elliott, T., Plank, T., Zindler, A., White, W. & Bourdon, B. (1997). Element transport from slab to volcanic front at the Mariana arc. Journal of Geophysical Research 102, 14991–15019.[CrossRef]
Fliedner, M. M. & Klemperer, S. L. (1999). Structure of an island-arc: wide-angle seismic studies in the eastern Aleutian islands, Alaska. Journal of Geophysical Research 104, 10667–10694.[CrossRef]
Foden, J. D. (1983). The petrology of the calc-alkaline lavas of Rindjani volcano, east Sunda arc: a model for island arc petrogenesis. Journal of Petrology 24, 98–130.[Web of Science]
Foden, J. D. (1986). The petrology of Tambora volcano, Indonesia: a model for the 1815 eruption. Journal of Volcanology and Geothermal Research 27, 1–41.[CrossRef][Web of Science]
Foden, J. D. & Green, D. H. (1992). Possible role of amphibole in the origin of andesite: some experimental and natural evidence. Contributions to Mineralogy and Petrology 109, 479–493.[CrossRef][Web of Science]
Foden, J. D. & Varne, R. (1980). The petrology and tectonic setting of Quaternary–Recent volcanic centres of Lombok and Sumbawa, Sunda arc. Chemical Geology 30, 201–226.[CrossRef][Web of Science]
Ghiorso, M. S. & Carmichael, I. S. E. (1985). Chemical mass transfer in magmatic processes II. Applications in equilibrium crystallization, fractionation and assimilation. Contributions to Mineralogy and Petrology 90, 121–141.[CrossRef][Web of Science]
Grove, T. L. & Bryan, W. B. (1983). Fractionation of pyroxene-phyric MORB at low pressure: an experimental study. Contributions to Mineralogy and Petrology 84, 293–309.[CrossRef][Web of Science]
Grove, T. L., Kinzler, R. J. & Bryan, W. B. (1990). Natural and experimental phase relations of lavas from Serocki volcano. In: Detrick, R., Honnorez, J., Bryan, W. B., et al. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 106/109. College Station, TX: Ocean Drilling Program, pp. 9–17.
Halliday, A. N., Lee, C.-C., Tommasini, S., Davies, G. R., Paslick, C. R., Fitton, J. F. & James, D. E. (1995). Incompatible trace elements in OIB and MORB and source enrichment in the sub-oceanic mantle. Earth and Planetary Science Letters 133, 379–395.[CrossRef][Web of Science]
Hart, S. R. (1984). A large-scale isotope anomaly in the Southern Hemisphere mantle. Nature 309, 753–757.[CrossRef]
Hawkesworth, C. J., Turner, S. P., McDermott, F., Peate, D. W. & van Calsteren, P. (1997). U–Th isotopes in arc magmas: implications for element transfer from the subducted crust. Science 276, 551–555.
Hawkesworth, C., Blake, S., Evans, P., Hughes, R., Macdonald, R., Thomas, L., Turner, S. & Zellmer, G. (2000). The time scales of crystal fractionation in magma chambers—integrating physical, isotopic and geochemical perspectives. Journal of Petrology 41, 991–1006.
Heath, E., Turner, S. P., Macdonald, R., Hawkesworth, C. J. & van Calsteren, P. (1998). Long magma residence times at an island arc volcano (Soufriere, St. Vincent) in the Lesser Antilles: evidence from 238U–230Th isochron dating. Earth and Planetary Science Letters 160, 49–63.[CrossRef][Web of Science]
Higgins, M. D. (2000). Measurement of crystal size distributions. American Mineralogist 85, 1105–1116.
Hill, E., Wood, B. J. & Blundy, J. D. (2000). The effect of Ca-Tschermaks component on trace element partitioning between clinopyroxene and silicate melt. Lithos 53, 203–215.[CrossRef][Web of Science]
Hochstein, M. (1995). Crustal heat transfer in the Taupo Volcanic Zone (New Zealand): comparison with other volcanic arcs and explanatory heat source models. Journal of Volcanology and Geothermal Research 76, 297–317.[CrossRef]
Hofmann, A. W., Jochum, K. P., Seufert, M. & White, W. M. (1986). Nb and Pb in oceanic basalts: new constraints on mantle evolution. Earth and Planetary Science Letters 79, 33–45.[CrossRef][Web of Science]
Hoogewerff, J. A., van Bergen, M. J., Vroon, P. Z., Hertogen, J., Wordel, R., Sneyers, A., Nasution, A., Varekamp, J. C., Moens, H. L. E. & Mouchel, D. (1997). U-series, Sr–Nd–Pb isotope and trace-element systematics across an active island arc–continent collision zone: implications for element transfer at the slab–wedge interface. Geochimica et Cosmochimica Acta 61, 1057–1072.[CrossRef][Web of Science]
Hutchison, C. S. (1976). Indonesian active volcanic arc: K, Sr and Rb variation with depth to the Benioff zone. Geology 4, 407–408.
Hutchison, C. S. (1982). Indonesia. In: Thorpe, R. S. (ed.) Andesites. New York: Wiley, pp. 207–224.
Mahoney, J., LeRoex, A. P., Peng, Z., Fisher, R. L. & Natland, J. H. (1992). Southwestern limits of Indian ocean ridge mantle and the origin of low 206Pb/204Pb mid-ocean ridge basalt: isotope systematics of the central southwest Indian ridge (17–50°E). Journal of Geophysical Research 97, 19771–19790.[CrossRef]
Marsh, B. D. (1999). Magma chambers. Annual Review of Earth and Planetary Sciences 17, 439–474.[CrossRef]
Marsh, B. D. (1998). On the interpretation of crystal size distributions in magmatic systems. Journal of Petrology 39, 553–599.[CrossRef][Web of Science]
Martin, D. & Nokes, R. (1988). Crystal settling in a vigorously convecting magma chamber. Nature 332, 534–536.[CrossRef]
Neuberg, J. (2000). Characteristics and causes of shallow seismicity in andesite volcanoes. Philosophical Transactions of the Royal Society of London 358, 1533–1546.
Norman, M. D., Pearson, N. J., Sharma, A. & Griffin, W. L. (1996). Quantitative analysis of trace elements in geological materials by laser ablation ICPMS: instrumental operating conditions and calibration values of NIST glasses. Geostandards Newsletter 20, 427–261.
Plank, T. & Langmuir, C. H. (1998). The chemical composition of subducting sediment and its consequences for the crust and mantle. Chemical Geology 145, 325–394.[CrossRef][Web of Science]
Potts, P. J., Thorpe, O. W., Isaacs, M. C. & Wright, D. W. (1985). High precision instrumental neutron-activation analysis of geological samples employing simultaneous counting with both planar and coaxial detectors. Chemical Geology 48, 145–155.[CrossRef][Web of Science]
Pyle, D. M. (1992). The volume and residence time of magma beneath active volcanoes determined by decay-series disequilibria methods. Earth and Planetary Science Letters 112, 61–73.[CrossRef][Web of Science]
Reagan, M. K., Volpe, A. M. & Cashman, K. V. (1992). 238U and 232Th-series chronology of phonolite fractionation at Mount Erebus, Antarctica. Geochimica et Cosmochimica Acta 56, 1401–1407.[CrossRef][Web of Science]
Rogers, N. W., Hawkesworth, C. J., Parker, R. J. & Marsh, J. S. (1985). The geochemistry of potassic lavas from Vulsini, Central Italy and implication for mantle enrichment processes beneath the Roman Region. Contributions to Mineralogy and Petrology 90, 328–353.
Rubin, K. H., Wheller, G. E., Tanzer, M. O., MacDougall, J. D., Varne, R. & Finkel, R. (1989). 238U decay series systematics of young lavas from Batur volcano, Sunda arc. Journal of Volcanology and Geothermal Research 38, 215–226.[CrossRef][Web of Science]
Schmidt, M. W. & Poli, S. (1988). Experimentally based water budgets for dehydrating slabs and consequences for arc magma generation. Earth and Planetary Science Letters 163, 361–379.
Self, S., Rampino, M. R., Newton, M. S. & Wolff, J. A. (1984). Volcanological study of the great Tambora eruption of 1815. Geology 12, 659–663.
Simkin, T. & Siebert, L. (1994). Volcanoes of the World. Washington, DC: Smithsonian Institution, 349 pp.
Spear, F. S. & Kimball, C. (1984). RECAMP—a Fortran IV program for estimating the Fe3+ contents in amphiboles. Computers and Geosciences 10, 317–325.
Spera, F. J. (2000). Physical properties of magma. In: Sigurdsson, H. et al. (eds) Encyclopedia of Volcanoes. New York: Academic Press, pp. 171–190.
Stolper, E. & Newman, S. (1994). The role of water in the petrogenesis of Mariana trough magmas. Earth and Planetary Science Letters 121, 293–325.[CrossRef][Web of Science]
Stolz, A. J., Varne, R., Davies, G. R., Wheller, G. E., Foden, J. D. & Abott, M. J. (1988). The geochemistry and petrogenesis of K-rich alkaline volcanics from the Batu Tara volcano, eastern Sunda arc. Contributions to Mineralogy and Petrology 98, 374–389.[CrossRef][Web of Science]
Todt, W., Cliff, R. A., Hanser, A. & Hofmann, A. W. (1996). Evaluation of a 202Pb–205Pb double spike for high-precision lead isotopic analyses. In: Basu, A. & Hart, S. (eds) Earth Processes, Reading the Isotopic Code. Geophysical Monograph, American Geophysical Union 95, 429–437.
Turner, S. & Foden, J. (2001). U–Th–Ra disequilibria, Sr–Nd–Pb isotope and trace element variations in Sunda arc lavas: predominance of a subducted sediment component. Contributions to Mineralogy and Petrology 142, 43–57.[Web of Science]
Turner, S., Hawkesworth, C., Rogers, N., Bartlett, J., Worthington, T., Hergt, J., Pearce, J. & Smith, I. (1997). 238U–230Th disequilibria, magma petrogenesis and flux rates beneath the depleted Tonga–Kermadec island arc. Geochimica et Cosmochimica Acta 61, 4855–4884.[CrossRef][Web of Science]
Turner, S. P., Platt, J. P., George, R. M. M., Kelley, S. P., Pearson, D. G. & Nowell, G. M. (1999). Magmatism associated with orogenic collapse of the Betic–Alboran Domain, SE Spain. Journal of Petrology 40, 1011–1036.[CrossRef][Web of Science]
Turner, S., Bourdon, B., Hawkesworth, C. & Evans, P. (2000). 226Ra–230Th evidence for multiple dehydration events, rapid melt ascent and the time scales of differentiation beneath the Tonga–Kermadec island arc. Earth and Planetary Science Letters 179, 581–593.[CrossRef][Web of Science]
Turner, S., Evans, P. & Hawkesworth, C. (2001). Ultra-fast source-to-surface movement of melt at island arcs from 226Ra–230Th systematics. Science 292, 1363–1366.
van Bergen, M. J., Vroon, P. Z., Varekamp, J. C. & Poorter, R. P. E. (1992). The origin of the potassic rock suite from Batu Tara volcano (East Sunda arc, Indonesia). Lithos 28, 261–282.[CrossRef][Web of Science]
van Calsteren, P. & Schwieters, J. B. (1995). Performance indications for TIMS with a deceleration lens system and post-deceleration detector selection. International Journal of Mass Spectrometry and Ionisation Processes 146/147, 119–129.[CrossRef]
Volpe, A. M. & Hammond, P. E. (1991). 238U–230Th–226Ra disequilibrium in young Mt. St. Helens rocks: time constraint for magma formation and crystallisation. Earth and Planetary Science Letters 107, 475–486.[CrossRef][Web of Science]
Volpe, A. M., Olivares, J. A. & Murrell, M. T. (1991). Determination of radium isotope ratios and abundances in geologic samples by thermal ionization mass spectrometry. Analytical Chemistry 63, 913–916.[Medline]
Widom, E., Schmincke, H.-U. & Gill, J. B. (1992). Processes and timescales in the evolution of a chemically zoned trachyte: Fogoa A, Sao Miguel, Azores. Contributions to Mineralogy and Petrology 111, 311–328.[CrossRef][Web of Science]
Wood, B. J. & Trigila, R. (2001). Experimental determination of aluminous clinopyroxene–melt partition coefficients for potassic liquids, with application to the evolution of the Roman province potassic magmas. Chemical Geology 172, 213–223.[CrossRef][Web of Science]
Wood, B. J., Blundy, J. D. & Robinson, J. A. C. (1999). The role of clinopyroxene in generating U-series disequilibrium during mantle melting. Geochimica et Cosmochimica Acta 63, 1613–1620.[CrossRef][Web of Science]
Xu, Y.-G., Menzies, M., Thirlwall, M. F. & Xie, G.-H. (2001). Exotic lithosphere mantle beneath the western Yangtze craton: petrogenetic links to Tibet using highly magnesian ultrapotassic rocks. Geology 29, 863–866.
Zindler, A. & Hart, S. R. (1986). Chemical geodynamics. Annual Review of Earth and Planetary Sciences 14, 493–571.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K. M. Cooper and M. R. Reid Uranium-series Crystal Ages Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 479 - 544. [Full Text] [PDF]
|
|
|
|
 |
|
 A. Dosseto, S. P. Turner, M. Sandiford, and J. Davidson Uranium-series isotope and thermal constraints on the rate and depth of silicic magma genesis Geological Society, London, Special Publications, January 1, 2008; 304(1): 169 - 181. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Kuritani, T. Yokoyama, and E. Nakamura Rates of Thermal and Chemical Evolution of Magmas in a Cooling Magma Chamber: a Chronological and Theoretical Study on Basaltic and Andesitic Lavas from Rishiri Volcano, Japan J. Petrology, July 1, 2007; 48(7): 1295 - 1319. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. K. Handley, C. G. Macpherson, J. P. Davidson, K. Berlo, and D. Lowry Constraining Fluid and Sediment Contributions to Subduction-Related Magmatism in Indonesia: Ijen Volcanic Complex J. Petrology, June 1, 2007; 48(6): 1155 - 1183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. MEDARD, M. W. SCHMIDT, P. SCHIANO, and L. OTTOLINI Melting of Amphibole-bearing Wehrlites: an Experimental Study on the Origin of Ultra-calcic Nepheline-normative Melts J. Petrology, March 1, 2006; 47(3): 481 - 504. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M.A. Elburg, M.J. van Bergen, and J.D. Foden Subducted upper and lower continental crust contributes to magmatism in the collision sector of the Sunda-Banda arc, Indonesia Geology, January 1, 2004; 32(1): 41 - 44. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. GEORGE, S. TURNER, C. HAWKESWORTH, C. R. BACON, C. NYE, P. STELLING, and S. DREHER Chemical versus Temporal Controls on the Evolution of Tholeiitic and Calc-alkaline Magmas at Two Volcanoes in the Alaska-Aleutian Arc J. Petrology, January 1, 2004; 45(1): 203 - 219. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||