Journal of Petrology

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (47) Request Permissions Google Scholar Articles by RUTHERFORD, M. J. Articles by DEVINE, J. D. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Journal of Petrology | Volume 44 | Number 8 | Pages 1433-1453 | 2003
© Oxford University Press 2003

Magmatic Conditions and Magma Ascent as Indicated by Hornblende Phase Equilibria and Reactions in the 1995–2002 Soufrière Hills Magma

MALCOLM J. RUTHERFORD^* and JOSEPH D. DEVINE

DEPARTMENT OF GEOLOGICAL SCIENCES, BROWN UNIVERSITY, PROVIDENCE, RI 02912, USA

^* Corresponding author. E-mail: Malcolm_Rutherford{at}brown.edu

RECEIVED JUNE 15, 2002; ACCEPTED FEBRUARY 20, 2003

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
A study of amphiboles and associated minerals in samples of Soufrière Hills andesite erupted from 1995 to 2002 shows significant compositional variations within hornblende phenocrysts, a separate set of small pargasitic crystals in the groundmass, and two types of reaction rims on the phenocrysts. The composition of the amphiboles and coexisting phases defines the thermal history of the erupting magma. As many as seven zones (<200 µm wide) in the hornblende phenocrysts begin with a sharp increase in Mg and Si, and then change gradually to a more Fe- and Al-rich hornblende, a transition that is consistent with a temperature rise. Analyses of the hornblende phenocrysts and associated Fe–Ti oxides verify previous conclusions that the pre-eruption magma was at 130 MPa and 830 ± 10°C, but was variably heated before eruption. The heating occurred within 30 days of eruption for all magmas erupted, based on the width of Ti-rich rims on titanomagnetite phenocrysts. Experimental phase equilibria for the andesite confirm that the natural hornblende phenocrysts would be stable between 825 and 855°C at a P_H2O of 130 MPa, and would be even more Al rich if crystallized at higher pressure. Pargasite is not stable in the andesite, and its presence, along with high-An plagioclase microphenocrysts, requires mafic magma mingling and hybridization with pre-existing andesite. Experimental melts of the andesite at 130 MPa and 830 and 860°C compare well with melt inclusions in quartz and plagioclase, respectively. Reaction rims on a few hornblende crystals in each andesite sample are rich in high-Ca pyroxene and are produced experimentally by heating the andesite above the stability limit for hornblende. Decompression-induced breakdown rims occur in some samples, and the rate of this reaction has been experimentally calibrated for isothermal andesite magma ascent at 830–860°C. The average ascent rate of magma during much of the 1995–2002 eruption has been >0·02 m/s, the rate that allows hornblende to erupt free of decompression-induced reaction rims.

KEY WORDS: hornblende; magma ascent; magma mingling; pargasite; magmatic conditions

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
The current volcanic eruption at the Soufrière Hills Volcano in Montserrat began with extrusive dome-forming activity in autumn 1995, and the eruption continues today at the end of 2002. The andesitic magma eruption has produced dome lavas, numerous pyroclastic flows, primarily from dome collapse, and occasionally there have been ash column-producing explosions within the dome and the upper conduit that have resulted in pumice and ash deposits (Druitt et al., 2002). Samples collected regularly by the Montserrat Volcano Observatory (MVO) staff from dome-collapse flows and from ash produced by vulcanian explosions indicate that the erupting magma is coarsely porphyritic andesite with 57–61 wt % SiO₂ (Devine et al., 1998a; Murphy et al., 2000). The andesite contains phenocrysts (<1·5 cm) of plagioclase, hornblende, low-Ca pyroxene, Ti-magnetite, ilmenite, and quartz in order of decreasing abundance. The groundmass is variably vesiculated, yielding samples that range from dense dome rock to less common pumice. The phases present in the phenocryst assemblage are also present in the groundmass except that hornblende is replaced by high-Ca pyroxene, and there is always some glass remaining in the groundmass. The amount of groundmass crystallization depends on the rate of magma ascent from the pre-eruption magma storage zone to the surface (Devine et al., 1998b; Rutherford et al., 1998). Inclusions of mafic magma constitute up to 1 vol. % of the material being erupted. Many of the mafic inclusions were angular and nonvesicular during the early phases of the eruption. As the eruption progressed, however, the relative abundance of inclusions that are identifiable as blobs of an injected mafic magma, based on their shape and texture, increased at the expense of angular inclusions.

The erupted Soufrière Hills andesite contains sparse macroscopic evidence of magma mixing beyond the glass-bearing inclusions, but substantial microscopic evidence of a heating event is evident in the thin sections (Devine et al., 1998a, 2003; Murphy et al., 2000). Quartz phenocrysts, sparsely present in all thin sections, are highly embayed and commonly surrounded by a Ca-rich pyroxene reaction zone. Some low-Ca pyroxene also has an overgrowth of small, high-Ca pyroxene crystals, and complex resorption and overgrowth histories are recorded in plagioclase phenocrysts. However, the main evidence of heating is the fact that magnetite phenocrysts have a uniform core composition that typically increases to higher TiO₂ in the outer 20–30 µm. This zoning indicates a higher temperature of equilibration of the rim with surrounding melt and ilmenite, and possibly a small accompanying decrease in _O2 (Devine et al., 2003). Recently, Couch et al. (2001) proposed a model involving basalt emplacement beneath the pre-existing andesite, and a heating-induced convection within the andesite to explain the evidence for heating and what they concluded was a lack of evidence for magma hybridization. Although this model does appear to explain many of the mineralogical and textural observations, the amount of andesite heating allowable appears very restricted by the textural and phase equilibrium data for hornblende in this magma composition (Barclay et al., 1998) compared with that required in the model to develop convection. Specifically, the presence of hornblende shows that the heating did not drive the andesite magma temperature beyond the limit of hornblende stability, which appears to be 855°C at the 130 MPa estimated for the storage zone depth.

In this paper we present analyses of recently erupted samples of andesite from the Soufrière Hills eruption, and describe a series of experiments on the andesite that were performed to better determine hornblende–plagioclase–melt equilibria over a range of possible pre-eruption conditions. The natural andesite contains abundant evidence of disequilibria in the form of strongly zoned plagioclase and hornblende phenocrysts (Devine et al., 1998a), and also evidence for several injections of a more mafic magma over a period of time. On the basis of these observations, the andesite magma erupted never existed as a totally molten magma, and large portions of the phenocryst interiors are not in equilibrium with the interstitial pre-eruption melt. For this reason, experiments were not carried out on a totally molten (glassy) andesite, a common approach in some experimental studies (Martel et al., 1999; Scaillet & Evans, 1999). Instead, melting and crystallization experiments were designed to investigate crystal–melt equilibria by looking at the phenocryst compositions in equilibrium with a given melt following a small change in the andesite crystallinity. We also performed additional experiments designed to measure the extent of reaction between phenocrysts and melt as a function of time after a given change in pressure or temperature was imposed. Because the compositions of coexisting hornblende and plagioclase are demonstrably dependent on both P and T (Holland & Blundy, 1994; McCanta et al., 2001), hornblende composition and stability are likely to be important in resolving questions concerning the timing of pressure and temperature variations in the Soufrière Hills andesite. This is especially true given that individual hornblende phenocrysts in the erupted andesite are compositionally zoned, and the amphibole microphenocrysts in the magma have compositions that differ from the large hornblende phenocrysts. Reaction rims on hornblende crystals can also yield information about the ascent history of erupted magma (Rutherford & Hill, 1993; Rutherford et al., 1998), but to relate measurements of the observed reaction rims to magma ascent rates, experimental data for rim development are required for this specific magma. Decompression experiments have been carried out on the Soufrière Hills andesite as part of this project.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
Experimental
Phase equilibrium experiments were carried out using two compositionally identical, but texturally different, samples of the Soufrière Hills andesite as starting material. Early experiments used a sample of dome lava (MONT138, erupted in early 1996) with fairly extensive groundmass crystallization and thick reaction rims on the hornblende phenocrysts. Most experiments were performed using sample MVO573 erupted in September–October 1997. The MVO573 sample used is a large, light gray, single pumiceous block that contains unrimmed hornblendes, and abundant groundmass glass as a result of the rapid ascent of this magma from the 130 MPa storage region (Rutherford et al., 1998). Both samples were powdered, but experiments on the MVO573 starting material were run using a lightly crushed powder in which the phenocrysts remained largely intact. Melting experiments were carried out with the crushed pumice, and crystallization experiments with a starting material consisting of products from a previous experiment that had been at higher P and/or T, such that the approach to the final P–T conditions involved crystal growth. The techniques for the hydrothermal experiments have been previously described (Rutherford et al., 1985; Rutherford & Hill, 1993). High-temperature experiments were performed using AgPd tubes in TZM pressure vessels with an Ar and CH₄ pressure medium; lower-temperature experiments were performed using Ag tubes and Waspaloy (TM) pressure vessels with water as the pressure medium. Oxygen fugacities were controlled at NNO + 1 ± 0·5 log unit (where NNO is nickel–nickel oxide), the _O2 recorded by magnetite and ilmenite phenocryst cores in the Soufrière Hills andesite (Devine et al., 1998a, 2003). In the TZM pressure vessels, a mixture of CH₄ and Ar, calibrated to maintain this _O2, was used as the pressurizing medium. The _O2 in the Waspaloy vessels is maintained at the NNO + 1 log unit level by using a Ni filler rod (e.g. Gardner et al., 1995). Vessels were quenched in water at the end of experiments, a technique that produces a glass free of quench crystals when using this andesite starting material. The rate-controlled decompression experiments were carried out using the methods described by Rutherford & Hill (1993).

Analytical
Polished thin sections were prepared from Soufrière Hills samples collected by MVO as often as freshly erupted material became available from an explosion or dome collapse. Most of the samples were gray, relatively low-vesicularity andesite from the collapsed dome, but several samples erupted in the August–September 1997 period (including MVO573) were more pumiceous. The polished thin sections were used for petrographic and textural study and for electron microprobe (EMP) analyses of glasses and minerals using previously described methods (Rutherford & Devine, 1988; Devine et al., 1995). A fraction of the same sample was ground to a powder, fused to a glass using a molybdenum foil capsule in a sealed SiO₂ glass tube, and then analyzed by EMP to obtain bulk magma compositions.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
Andesite petrology
The phenocryst phases in the Soufrière Hills andesite have been studied and analyzed in samples erupted over the period from November 1995 to September 2002. As described previously (Rutherford et al., 1998; Murphy et al., 2000), the main phenocrysts present in the 1996–1997 eruptions are plagioclase, hornblende, low-Ca pyroxene, titanomagnetite, ilmenite, quartz, and apatite in order of decreasing abundance (Fig. 1). The total phenocryst abundance is 49 vol. % (vesicle-free) based on our thin section point counts (10 thin sections of three samples), assuming the microphenocrysts (>100 and <300 µm) were present in the magma when the magma ascent began. This assumption about microphenocrysts is supported by the observation that the microphenocrysts are present in MVO573, but this rapidly ascended sample is essentially free of microlites (<100 µm). The above volume estimate of crystals present in the pre-eruption magma is consistent with estimates from mass balance calculations using bulk-rock, matrix glass, and phenocryst compositions (e.g. Murphy et al., 2000). Variations in the bulk composition of the erupted andesite are small (SiO₂ ranges between 57·50 and 60·82 wt %) in samples collected over the entire eruption period, and no systematic variation is seen with time (Devine et al., 1998a; unpublished data, 2001). The groundmass in the erupted andesite has been variably crystallized, particularly in magma that ascended more slowly to the surface as indicated by the thicker reaction rims developed on hornblende, the reduced glass abundance, and decreased sample vesicularity (Devine et al., 1998b). However, some explosively erupted samples consist of magma that ascended rapidly and was quenched on eruption; glassy melt inclusions and, where possible, matrix glass have been analyzed in some of these samples.

View larger version (153K): [in this window] [in a new window]   Fig. 1. Photomicrographs of amphiboles in the Soufrière Hills andesite. Field width is 5 mm in all except (d), (e), and (k) where it is 0·7 mm, and (m) where it is 0·3 mm. (a) A large hornblende phenocryst with a decompression-induced rim 100 µm thick. (b) Section showing adjacent rounded and euhedral hornblende phenocrysts and small pargasite crystals in the groundmass. Rounded and embayed quartz grain has the remnants of a large melt inclusion. (c) Hornblende phenocryst in MVO573 showing cyclic growth zones from slightly higher Si and Mg (light) to higher Fe and Al (dark). (d) Expanded view of the upper end of the zoned phenocryst in (c) showing growth zones and some evidence of resorption following crystallization of some dark layers. Field width in this photograph is 0·7 mm. (e) Sample of MVO1216 (February 2001) showing phenocrysts of plagioclase, low-Ca pyroxene, and both large and small amphiboles. (f) Expanded view of (e) showing small hornblende phenocryst and laths of pargasite (p) in the groundmass. (g) Contact between the andesite on the left, and an enclave of vesicular pargasite-rich basaltic andesite. A hornblende from the andesite is partially enclosed in the enclave. (h) Large hornblende phenocryst with a Ca-pyroxene-rich rim of the type produced by hornblende decomposition produced by rising temperature. (i) A perfectly euhedral hornblende phenocryst adjacent to a hornblende with a thick reaction rim, which is interpreted to have formed during a prolonged stay in the conduit. (j) Reflected light view of hornblende phenocryst containing a significant amount of opacite development along internal fractures and cleavages. (k) Plagioclase phenocryst with two blade-shaped pargasite microphenocrysts trapped in a growth period of An56–60 plagioclase during a prior heating–hybridization event. Pargasite crystals are 200 µm long. (m) Photomicrograph of crystallization experiment M-56 at 850°C and 130 MPa showing a large low-Ca pyroxene just below a pool of glass containing plagioclase with An56 rims, small titanomagnetite, and hornblende crystals, some of which occur as overgrowths on Ca-rich pyroxene. Field is 300 µm wide.

 
Hornblende
Hornblende in the Soufrière Hills andesite shows considerable textural and compositional variability. The main population of crystals is very large (0·5–1·5 cm across), green to brown, and euhedral (Fig. 1a–c). Compositional traverses from rim to rim across some phenocrysts show alternations between zones of higher Mg and Si grading gradually back to darker-colored zones relatively depleted in these elements and enriched in Fe and Al (Table 1). As many as seven of these alternations occur in large phenocrysts (Fig. 2). Although this compositional zonation is generally not optically discernible in hornblende phenocrysts, it is visible in many sections cut parallel to the crystallographic c-axis, and the zonation is clearly parallel to crystal growth faces (Fig. 1c and d). When optically visible, the beginning of the light-colored layer, which is richer in Mg and Si, appears to occasionally follow a period of minor crystal resorption. The new growth zone is thick (<200 µm) on the (001) crystal faces and much thinner on faces parallel to the c-axis. This internal compositional zoning in large hornblende phenocrysts occurs in samples erupted throughout the 1996–2002 period. It should be noted that although we refer to zones as Mg or Fe rich, the total compositional variation in these compositional bands is small, with a maximum of 2 wt % change each in SiO₂, MgO, FeO and Al₂O₃ (Fig. 2).

View this table: [in this window] [in a new window]   Table 1: Representative natural and experimental amphibole compositions

 

View larger version (40K): [in this window] [in a new window]   Fig. 2. Core-to-rim compositional profile along the length of the zoned hornblende phenocryst pictured in Fig. 1c and d based on electron microprobe analyses (Table 1). The compositionally defined zones are present in similar sections of hornblende phenocrysts in most samples of the andesite, but are visible optically only in sections cut parallel to the crystallographic c-axis.

 
Reaction rims developed at the expense of large hornblende phenocrysts where they were in contact with melt (Fig. 1a) in magmas erupted during early 1996. This type of reaction rim is thin (2–20 µm) on crystals erupted during spring 1996 relative to those in the initial eruptions, and it is totally absent (Fig. 1c) on crystals erupted during July–October 1996 (Devine et al., 1998b), and in many more recent magmas. The origin of these rims is attributed to hornblende breakdown during slow magma ascent from the storage zone at depth (Rutherford et al., 1998), an interpretation supported by the association of high magma vesicularity and low groundmass crystallinity in samples containing unrimmed hornblende. Almost all of the magma erupted during and after summer 1996 contains some hornblende phenocrysts with reaction rims along with the main population of rim-free crystals (Fig. 1b, h). The rimmed crystals are interpreted as the result of magma mixing in the conduit involving new magma, material previously plated on the conduit walls by rising magma, and possibly some assimilation of the lower part of the lava dome (Rutherford et al., 1998). Although most unrimmed hornblende phenocrysts are bounded by planar crystal faces, a few crystals in some samples can be seen to be noticeably rounded (Fig. 1b), possibly indicating an episode of resorption or dissolution that accompanied the variable heating of the magma.

In samples of the andesite erupted during summer 1997 and in 2000 and 2001, after the 1998–1999 hiatus in dome growth, small (<0·4 mm), euhedral to subhedral crystals of pargasitic hornblende (12–16 wt % Al₂O₃; Table 1) occur together with the low-Al (6·3–8·3 wt % Al₂O₃) hornblende phenocrysts (Fig. 1e and f). Some of these crystals appear to be embayed and/or have reaction rims. Further study of the andesite samples has shown that corroded remnants of small pargasite crystals are also present in the andesite erupted during autumn 1996 (e.g. MONT153). The presence of these pargasitic crystals is correlated with the appearance of small (<50 cm), rounded, vesicular enclaves of basalt to basaltic andesite composition in the andesite (Fig. 1g). The composition of the pargasitic amphibole in the andesite is similar to the composition of crystals found in the basaltic andesite enclaves (Table 1). Blade-shaped crystals of compositionally similar amphibole are also observed as inclusions in the outer growth zone of plagioclase phenocrysts in the andesite (Fig. 1k). The Al-rich, pargasite crystals in the andesite may originate from the break-up and mixing-in of mafic magma injections, but we must also consider the possibility that they could represent growth of new crystals in the andesite as a result of changing P–T conditions.

One or two crystals of hornblende in each sample are surrounded by a thick (200–500 µm) rim composed primarily of high-Ca pyroxene (Fig. 1h) that is distinctly different texturally and compositionally from rims produced during decompression (Fig. 1a). The high-Ca pyroxene crystals are large (200–500 µm) compared with those in decompression rims (1–20 µm), are commonly elongated parallel to the hornblende c-axis, and often appear in near optical continuity with the hornblende. Experimental evidence is presented below to suggest that this reaction is the product of a thermal decomposition of hornblende, in contrast to reaction produced by decompression or oxidation.

The hornblendes in andesite lava dome samples have commonly experienced slight to extensive oxidation effects that are here referred to as opacitization (Garcia & Jacobson, 1979). The opacite is characterized by the appearance of increasing amounts of fine-grained magnetite, and occurs along crystal margins, as well as along fractures and cleavage planes throughout the hornblende (Fig. 1j), making crystals totally opaque in extreme cases. In contrast to the decompression-induced rims, the opacite reaction is not restricted to places where the hornblende is in contact with interstitial melt.

Plagioclase
The plagioclase phenocrysts in the Soufrière Hills andesite (Fig. 1) are complexly zoned (Murphy et al., 2000). Cores of phenocrysts are both sodic (An_50–60) and calcic (An_70–85), but the rims are generally in the range of An_53–58. Profiles across two crystals in andesite samples MVO1217 and 1234C (Fig. 3) are representative of phenocrysts in samples throughout the eruptive series. The rim compositions of some phenocrysts show a compositional spike to An₈₀, but on many the outermost layer is An_53–58. Crystal aggregates are common in this magma, and it is possible that the high-An rim of some plagioclase phenocrysts now exposed to the magma matrix (Fig. 3a) were attached to other crystals before the eruption, and the crystals were separated during magma ascent. As is clear from the compositional profiles in Fig. 3, there are commonly several spikes to more calcic compositions in the interior of the plagioclase phenocrysts. The outer An-rich rim on the plagioclase phenocrysts may be associated with the last heating and magma hybridization event to affect the andesite, the event that produced compositional zoning in the outer margins of Fe–Ti oxides and amphiboles in the same magma. The earlier spikes to high-An plagioclase composition may be the remnants of similar events earlier in the history of this magma system (Devine et al., 1998a). This theory needs to be tested by additional experiments on the relative rates of plagioclase, titanomagnetite, and hornblende growth in such magmas.

View larger version (39K): [in this window] [in a new window]   Fig. 3. Rim-to-rim traverses showing compositions across two typical plagioclase phenocrysts in the Soufrière Hills andesite. Data are from microprobe analyses of (a) MVO1217 and (b) MVO1234 erupted in February 2001 and September 2002, respectively.

 
Plagioclase in the Montserrat andesite also occurs as microphenocrysts (100–300 µm) that are similar to the phenocrysts in composition, and as microlites (<100 µm) in most samples. The microlites clearly grew during magma ascent because they are sparse to absent in the more pumiceous samples erupted explosively during mid-1996 and late 1997 (e.g. samples MONT153 and MVO573). Microlite compositions range widely from An₄₈ to An₇₅ (Murphy et al., 2000). Microphenocryst compositions are in the range An_65–85 and An_48–58 (Murphy et al., 2000). Small phenocrysts of plagioclase enclosed within the hornblende phenocrysts have rim compositions that range up to An₇₅.

Fe–Ti oxides
Titanomagnetite is the main Fe–Ti oxide among the phenocrysts in the Soufrière Hills andesite, but ilmenite grains are also present. As is typical of intermediate composition volcanic rocks, the Fe–Ti oxide phenocrysts (<1·6 mm diameter) are significantly smaller than the hornblende, pyroxene and plagioclase. The cores of the magnetite phenocrysts are compositionally homogeneous, and similar in samples erupted throughout the 1995–2002 period. However, the outer 0–50 µm margins of these phenocrysts commonly have a steadily increasing TiO₂ content compared with the average low-Ti core composition (7·78 ± 0·28 wt % TiO₂; Devine et al., 2003). Typical profiles across magnetite phenocrysts from two samples are plotted in Fig. 4. When in contact with ilmenite, the TiO₂ content of the magnetite rim is commonly somewhat higher than in magnetite in contact with groundmass. Devine et al. (2003) have suggested that the magnetite adjacent to ilmenite is actually a mixture of magnetite and ilmenite, and that all ilmenite phenocrysts are essentially unzoned. Ilmenite phenocrysts are similar in composition from sample to sample, and homogeneous except for zoning in grains adjacent to magnetite (Devine et al., 2003). The increased TiO₂ in the outer rim of most magnetite phenocrysts is a real diffusion gradient, however, and the width of the zone tends to be similar in grains in the same sample. The increase in TiO₂ toward the crystal rim appears to be the product of titanium exchange for iron produced by a heating of the magma (Devine et al., 2003). Generally there is no Ti-rich rim on the titanomagnetite phenocrysts in pumiceous samples such as MVO573 (Fig. 4), the sample we have chosen for most of our experiments. However, heating samples of this pumiceous andesite at 860°C for 22 days produces zoning from 7·8 wt % TiO₂, the core composition, to 10·5 wt % in the outer margin of magnetite phenocrysts. High-Ti rims are produced on MVO573 magnetite in as little as 2 days at 870°C (Devine et al., 2003). Although the Ti-rich rims of the natural titanomagnetite phenocrysts are interpreted as indicating a temperature rise, a small relative decrease in the _O2 would tend to produce a similar change in magnetite (Frost & Lindsley, 1991). A combined temperature rise with a small relative decrease in _O2 would explain both Ti-rich rims on magnetite and the observed lack of zoning in ilmenite, and this is our preferred interpretation. Interestingly, the widths of the Ti-rich rims on titanomagnetite grains have not changed in samples erupted during the course of the 1995–2002 eruption, and core TiO₂ contents have remained essentially constant at 7·78 ± 0·28 wt % TiO₂ (Devine et al., 2003).

View larger version (25K): [in this window] [in a new window]   Fig. 4. Plot of TiO2 compositions in traverses across three phenocrysts of Ti-magnetite: naturally occurring crystals from samples erupted in 1997 and 2001, and an experimentally annealed sample of a previously unzoned magnetite. Temperatures are calculated using Andersen & Lindsley (1988) and Andersen et al. (1993). The temperature is calculated using the composition of ilmenite phenocrysts in the sample, but it is for reference only and is only accurate for the uniform phenocryst core compositions. The zoned titanomagnetite rim is not generally in equilibrium with ilmenite and so the temperature is ‘apparent’ (Devine et al., 2003).

 
Pyroxenes
Low-Ca pyroxene is the third most abundant phenocryst phase in the Soufrière Hills andesite, after plagioclase and hornblende. The phenocrysts are generally euhedral and range up to 5 mm across (Fig. 1e). The phenocrysts are also generally homogeneous (En₆₀Fs₃₈Wo₂; Mg-number 58–62), but some have outer zones up to several tens of micrometres thick of more magnesian pyroxene (En₆₈Fs₃₁Wo₃; Mg-number 68). Other low-Ca phenocrysts have an outer rim of 100–200 µm Ca-rich pyroxene crystals. All of these types occur in the same sample (Murphy et al., 2000). High-Ca pyroxene does not occur as a phenocryst in the Montserrat andesite, but it does occur as microphenocrysts in rims around quartz as well as a few hornblende and low-Ca pyroxene phenocrysts. It also occurs a microphenocrysts in the groundmass. The average composition of the high-Ca pyroxene is En₄₆Fs₂₀Wo₃₄, but the En and Wo vary by as much as ±5 mol %.

Glasses
Glasses occur in the groundmass of the Soufrière Hills andesite, and as glassy remnants of melt inclusions trapped in phenocrysts. The groundmass of most samples is filled with plagioclase, pyroxene, and Fe–Ti oxide microlites, making the glass very difficult to analyze, and also somewhat variable compositionally depending on the amount of microlite crystallization that occurred during magma ascent. However, a matrix glass composition from rapidly quenched sample MVO243 is identical within analytical error to the anhydrous composition of melt trapped in quartz phenocrysts (Devine & Rutherford, in prep.). The averages of 30 plagioclase- and six quartz-hosted melt inclusion compositions analyzed in rapidly quenched samples MONT153 and MVO573 are given in Table 2. Melt inclusions trapped in plagioclase phenocrysts contain an average of 75·23 ± 0·80 wt % SiO₂ (anhydrous), suggesting that the main growth of plagioclase in the andesite occurred when the melt was less evolved than it was at quartz saturation (average SiO₂ 78·18 ± 0·40 wt %, anhydrous). Previous analyses of melt inclusions by Fourier transform infrared (FTIR) spectroscopy and by the difference method indicated that there was 4·3 ± 0·5 and 4·7 ± 0·5 wt % volatiles dissolved in the melt trapped during growth by quartz and plagioclase, respectively (Barclay et al., 1998). The FTIR data also indicated that the CO₂ abundance was below detection (50 ppm) in these melt inclusions. Analyses of the additional plagioclase melt inclusions (Table 2) support the earlier results, indicating a total volatile content by the difference method of 4·9 ± 0·5 wt %.

View this table: [in this window] [in a new window]   Table 2: Starting material, melt inclusion, and representative experimental glass compositions

 

	EXPERIMENTAL RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
Experiments have been carried out to determine the phase equilibria of the Soufrière Hills andesite composition. An earlier study by Barclay et al. (1998) was based on a limited number of experiments of relatively short duration. Additional long duration experiments were carried out to better estimate equilibrium compositions of the phenocrysts and coexisting melt (glass) as a function of crystallization conditions. The experiments are described in the Appendix. Because of the complexity of zoning in the plagioclase and hornblende phenocrysts, these experiments were primarily carried out by characterizing the changes in the natural sample assemblage produced by small, well-defined changes in the P–T conditions. Previously obtained data suggested that the total pressure in the magma was 130 MPa (Devine et al., 1998a) and that the dissolved water content of the melt trapped by quartz and plagioclase phenocrysts required a P_H2O of 130 MPa (Barclay et al., 1998). This evidence suggests that the magma was saturated with an H₂O-rich fluid before the eruption, and hence all of the phase equilibrium experiments have been water saturated at the NNO + 1 oxygen fugacity determined for the Soufrière Hills andesite (Devine et al., 1998a).

The phase boundaries in the new P_H2O–T phase diagram (Fig. 5) have not changed significantly from previous determinations (Barclay et al., 1998). However, the longer duration experiments have more closely defined the upper stability of hornblende and the temperatures at which quartz and biotite become stable with cooling in the 130–200 MPa pressure range. These are critical phase boundaries in assessing the pre-eruption magma conditions, and the history of conditions in the Soufrière Hills andesite. For example, hornblende appears to be unstable at 860°C, but growth does occur at 850°C and 130 MPa (Fig. 1m). New growth hornblende is also present in both melting and crystallization experiments between 810 and 870°C at 200 MPa. Ca-rich pyroxene does not occur in melting experiments except those above the upper stability limit of hornblende. In crystallization experiments that contain high-Ca pyroxene in the starting material, some of this phase is commonly still present as grains surrounded by new-growth hornblende at the end of experiment. These two sets of experiments indicate that the reaction of high-Ca pyroxene + melt to form hornblende occurs over a very small temperature interval near the temperature marking the upper stability limit of hornblende. The presence of biotite in low-temperature experiments places a lower temperature limit (>780°C) on conditions in the Soufrière Hills magma system over time, as no remnants of biotite have been seen in any of the samples studied. The other benefit of the new experiments is a much better estimate of the compositions of the stable hornblende and plagioclase at a given temperature and P_H2O. Plagioclase composition contours are drawn on the phase diagram based on analyses of new crystal growth in the experiments (Appendix).

View larger version (28K): [in this window] [in a new window]   Fig. 5. Pressure–temperature phase diagram for the Soufrière Hills andesite for PH2O = Ptotal conditions. The fugacity of oxygen is fixed at the NNO + 1 log unit condition indicated by core of Fe–Ti oxides in the natural samples. Triangles indicate the final P and T of experiments; they point in the direction of approach to final condition. Hornblende is stable in the experiments marked by filled triangles. The filled bar at 130 MPa represents the temperature, Ptotal, and PH2O based on analyses of the Fe–Ti oxides and melt inclusions (see text). Plagioclase composition contours are a summary of rim compositions obtained from the experiments (Appendix).

 

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
Conditions in the pre-eruption andesite
We previously concluded that the pre-eruption andesitic magma ascended from a storage zone that had been at 830 ± 10°C until just before eruption, and that the total pressure as well as the P_H2O in this storage zone was 130 ± 25 MPa (Barclay et al., 1998; Devine et al., 1998a). The higher TiO₂ contents toward the rims of magnetite phenocrysts, and the rimming of quartz with high-Ca pyroxene, was cited as evidence that the magma had experienced a heating event before eruption. Now that additional samples from the continuing eruption have been studied, and longer duration experiments have been completed to determine crystal–melt equilibria, these conclusions can be re-examined.

The estimate of 130 MPa total pressure on the pre-eruption Soufrière Hills magma is based on Al-in-hornblende geobarometry using the 6·4 wt % Al₂O₃ that is so common in the large hornblende phenocrysts in this magma. The hornblende rim compositions tend to be low Al₂O₃; however, as illustrated in the compositional profile across a hornblende phenocryst (Fig. 2), there are episodic swings in the Al₂O₃ content of these grains up to as high as 8·3 wt %, measurably higher than the average. We use the average composition of low-Al zones and the Johnson & Rutherford (1989) geobarometer to obtain the 130 MPa pressure, recognizing that the high-Al zones must be explained, and that not all requirements for the geobarometer are met by the Montserrat phenocryst–melt assemblage. One of these requirements is that the hornblende used in the geobarometry must be in equilibrium with quartz. If the melt in equilibrium with hornblende is not quartz saturated, the Al₂O₃ content of the coexisting hornblende is invariably higher than it would be otherwise (Johnson & Rutherford, 1989). Although there are several quartz crystals in every thin section of the Soufrière Hills andesite, these crystals became rimmed with high-Ca pyroxene at some time before the eruption, indicating that the magma was probably not continuously quartz saturated. Of the hornblende compositions present in the zoned phenocrysts, the low-Al hornblende is the most likely to be crystallized from an SiO₂-saturated melt. The geobarometer also requires the coexisting melt to be in equilibrium with K-rich alkali feldspar, and this is not the case in the Montserrat andesite. However, the fact that the Al-in-hornblende estimate of total pressure is equivalent to the P_H2O required to produce the H₂O observed in melt inclusions (Barclay et al., 1998) suggests that the total pressure estimate is accurate, but it requires experimental corroboration.

Additional evidence for the above pressure and temperature estimate for the pre-eruption Soufrière Hills magma comes from a comparison of hornblende and natural glass compositions with the experimental data for the same phases. Figure 6 shows the composition of experimental glass compositions (Table 2) for the Soufrière Hills andesite bulk composition equilibrated at a series of temperatures and a P_H2O of 130 MPa. Also shown are experimental glass compositions for two temperatures (810 and 840°C) at 200 MPa, and the compositions of glassy melt inclusions trapped in plagioclase and quartz phenocrysts in more pumiceous samples of the natural andesite. All melt (glass) compositions given in Table 2 were normalized to an anhydrous basis (i.e. 100 wt % totals) for plotting in Fig. 6. Melt–phenocryst rim equilibrium appears to be essentially achieved in the experiments judging from the close approach of glass compositions in the melting and crystallization experiments at a given pressure and temperature. Quartz saturation at 130 MPa occurs at 825°C, as indicated by the presence of small SiO₂ crystals in the crystallization experiments at <830°C, and by the fact that the melt SiO₂ does not increase in experiments at temperatures lower than 825°C. At 200 MPa, quartz saturation occurs at 790 ± 10°C. The melt compositions produced at 130 MPa in experiments at 830 and 800°C are very similar to the average melt inclusion in quartz grains in the andesite (Fig. 6; Table 2), but are somewhat lower in SiO₂. Their very high SiO₂ suggests that the quartz-hosted melt inclusions in this magma have probably experienced some dissolution of the host-crystal walls as a result of the recent heating experienced by the magma. The 790°C required for quartz saturation at 200 MPa is significantly lower than the 830 ± 10°C temperature obtained from Fe–Ti oxide geothemometry for the pre-eruption andesite, evidence against pressures >130 MPa for the pre-eruption magma storage zone.

View larger version (23K): [in this window] [in a new window]   Fig. 6. Plot of glass compositions produced in 130 MPa experiments on the Soufrière Hills andesite. Squares represent reversal experiments at 200 MPa for 810 and 840°C. Also plotted are melt inclusions in plagioclase separates from two samples, and five melt inclusions in quartz found in one of these samples. Analyses are given in Table 2.

 
The compositions of plagioclase melt inclusions (Fig. 6 and Table 2) in two rapidly quenched andesite samples (MONT153 and MVO243) are equivalent to the experimentally produced interstitial melt in the andesite at 860–830°C and 130 MPa. This observation suggests that most of the plagioclase growth took place at this temperature, 5–30°C above where quartz would be growing in this magma. This observation is not particularly surprising, as the interstitial melt present in the andesite at 860°C and 130 MPa is already rhyolitic in composition. A mass balance calculation shows that the melt trapped by quartz can be created by crystallizing 10 wt % plagioclase from the 860°C melt; no additional hornblende crystallization is required. The additional plagioclase crystallization that occurs as an andesite composition such as this cools from 860 to 830°C could be accommodated as a very thin layer on existing crystals without trapping melt inclusions of any measurable size.

Hornblende growth was observed in both melting and crystallization experiments of long duration on the Soufrière Hills andesite, although re-equilibration of pre-existing hornblende in a melting experiment was never complete, nor was growth of hornblende in crystallization experiments (Fig. 1m). The compositions of the new growth hornblende in the experiments (Table 3) show the effect of imposed changes in both temperature and pressure when plotted as a function of Al^IV, Al^VI, and Mg-number (Fig. 7). The data at 130 MPa clearly overlap the range of compositions obtained from profiles across large phenocrysts in the andesite. The composition of hornblende in these experiments illustrates that the low-Al₂O₃ (6·4 wt %) compositions that occur throughout the core of natural phenocrysts are not stable in the andesite magma until the temperature falls to 825°C at 130 MPa. At 850°C and 130 MPa, the equilibrium hornblende in this andesite composition magma is more Al rich, approaching the 8·2 ± 0·3 wt % Al₂O₃ found in the Al-rich zones throughout the phenocryst cores (Fig. 3; Table 1). The compositions of hornblendes produced in the 200 MPa experiments fall on the trend defined by the natural phenocrysts and the 130 MPa experimental hornblende crystals (Fig. 7a), a trend controlled in part by temperature. However, many of the 200 MPa compositions are clearly higher in Al^VI than any of the natural phenocrysts or the crystals produced experimentally at 130 MPa. The observed high-Al^VI compositions in amphiboles produced in 200 MPa experiments compared with natural amphibole compositions lend support to the conclusion that the total pressure in the andesitic magma was near 130 MPa, and not as high as 200 MPa.

View larger version (28K): [in this window] [in a new window]   Fig. 7. (a) Plot of Sum A site occupancy vs AlIV for the hornblendes produced in 130, 200, and 250 MPa experiments on the andesite composition (Appendix; Table 2), and the hornblende and pargasite present in typical samples of the natural andesite. (b) Plot of AlVI vs AlIV for the same sets of amphibole composition data as plotted in (a). (c) Plot of Mg# vs tetrahedral Al in the experimental and natural amphiboles.

 
The Holland & Blundy (1994) hornblende–plagioclase geothermometer can be used to calculate the temperature of the Montserrat andesite assuming a pressure of 130 MPa and H₂O saturation. Using the low-Al hornblende (6·4 wt % Al₂O₃) in the andesite and An₅₀ plagioclase (Fig. 5), this model yields a temperature of 810°C at pressures of 100–200 MPa. The high-Al hornblende composition in the andesite and the coexisting An₅₃ plagioclase yield a temperature of 845 ± 10°C (130 MPa) when run with either the quartz-saturated or the quartz-absent models of the Holland & Blundy (1994) geothermometer.

The Holland & Blundy (1994) solutions are thus consistent with the idea that the high-Al hornblende in Montserrat andesite equilibrated with An₅₃ plagioclase and melt at 845°C, about 30°C above the temperature at which the low-Al hornblende would crystallize in the same magma. The temperature estimates derived from the Holland & Blundy model and the experiments in this project are equivalent considering the errors in the two determinations.

Pre-eruption heating of the andesite: nature, origin, and timing
The evidence of a heating event that affected the 1995–2002 Soufrière Hills magma late in its pre-eruption history appears clear. Devine et al. (2003) have shown that the diffusion-like profile of increased Ti in the outer margin of magnetite phenocrysts (Fig. 4) is very common in samples erupted from June 1996 until the present. Phenocrysts in different samples show some variations in the Ti profile, however, and some samples (MVO573, and others erupted in early autumn 1997) contain compositionally unzoned titanomagnetite grains. These differences are interpreted as representing variations in the extent and timing of the heating of different batches of the erupted magma. The Ca-rich pyroxene rims on quartz grains and on some low-Ca pyroxene crystals can also be best explained by a rise in magma temperature to just above 855°C at 130 MPa, where Ca-rich pyroxene is stable rather than hornblende. The zoning to higher TiO₂ observed in the outer edge of magnetite phenocrysts can be experimentally produced in as little as 2 days at 870°C, or 27 days at 860°C (Fig. 4; Devine et al., 2003). However, many hornblende phenocrysts in the erupted andesite have not experienced any breakdown reaction, and only a few (1–5% of hornblende crystals in any sample) show evidence of the kind of breakdown that occurs with rising temperature. Hornblende phenocrysts that have experienced partial breakdown as the result of a temperature rise are the rare crystals that are rimmed with coarse Ca-rich pyroxene (Figs 1h and i, and 8b). A similar reaction zone (Fig. 8a) is produced on hornblende in experiments of 12–36 h duration at 870–925°C at 130 MPa, and the reaction zone is thicker at the higher temperatures for an experiment of a given duration. The high-Ca pyroxene reaction rim produced at 900°C and 200 MPa, 15°C above the hornblende breakdown temperature, is thicker (150 µm) and coarser grained than the rim produced at 860°C (10–20°C above the hornblende reaction curve) and 130 MPa. This is interpreted to be a result of the higher melt viscosity and lower reaction temperature in the latter experiment.

View larger version (186K): [in this window] [in a new window]   Fig. 8. Photomicrographs of rims on hornblende. (a) Ca-pyroxene-rich rim 70 µm thick produced experimentally on rim-free phenocryst fragment at 875°C (Run 42; Appendix). (b) Ca-pyroxene-rich rim of 300 µm on a natural andesite phenocryst. (c) Hornblende with a rim of 28 µm thickness produced by a 10 day decompression from 130 to 4 MPa at 860°C (Run 29). (d) Rims of 13 µm thickness on small hornblende fragments produced in an experiment (Run 50) similar to (c), but of 6 days duration.

 
We conclude that relatively small amounts of the Soufrière Hills andesite were raised to temperatures outside the hornblende stability field, but for no more than a few hours. However, this magma is now well mixed on the scale of a thin section with similar magma that was not raised above the hornblende stability field for sufficient time to produce hornblende breakdown. The heating of the pre-existing andesite was apparently produced by an intrusion of basaltic to basaltic andesite composition into the storage zone of the 830 ± 10°C andesite. This mafic magma appears in small amounts (1 vol. %; Murphy et al., 2000) as vesicular enclaves of 2–50 cm diameter in many samples of the erupted andesite. Heating could have taken place by heat transfer across the boundary between the two magmas (Couch et al., 2001; Devine et al., 2003) as the mafic intrusion pooled beneath the pre-existing andesite. The temperature of the erupting magma would also have been raised during andesite hybridization by the mafic magma, a process suggested by the presence of small pargasite laths and calcic-plagioclase microphenocrysts in the andesite. The experimental phase equilibria indicate that plagioclase more calcic than An₇₀ will not crystallize in the andesite at 130 MPa or the lower pressures (Fig. 5) achieved during eruption. Even if the ascending magma were to experience heating because of decompression-induced crystallization and the release of latent heat, the maximum heating would be of the order of 40°C (Devine et al., 2003), insufficient to cause crystallization of An_80–85 plagioclase. The presence of such high-An microphenocrysts in the mafic enclaves (Murphy et al., 2000) suggests that they were probably added to the andesite during the magma mingling process.

Origin of Soufrière Hills hornblende phenocrysts
The different compositions of the hornblende phenocrysts and the associated pargasitic hornblende in the Soufrière Hills andesite clearly contain valuable information concerning the magmatic history of the andesite. The potential importance of amphibole compositions was demonstrated by Helz (1979), Spear (1981), Hammerstrom & Zen (1986) and recently by Holland & Blundy (1994). As described above, the pargasitic hornblende in the Soufrière Hills magma occurs as small (<200 µm) blade-shaped crystals in mafic enclaves, in the andesite groundmass (also see Fig. 1e and f), and trapped in plagioclase phenocrysts in some erupted magma samples (Fig. 1k). The possibility that all pargasitic crystals came from the mafic magma is suggested by the fact that pargasite crystals in the two magmas are similar in size, shape, and composition (Fig. 7; Table 1). Additionally, experiments on the andesite bulk composition do not produce amphibole with Al₂O₃ contents as high as the 12–16 wt % observed in Montserrat pargasite anywhere within the hornblende stability field up to 250 MPa P_H2O (Table 1; Fig. 5). Mixing–hybridization events appear to be required to explain the presence of the blade-shaped pargasite in the andesite as well as the high-An plagioclase microphenocrysts. The small but distinct compositional variability observed in the bulk composition of erupted magma (57·5–60·8 wt % SiO₂) can be explained by variable amounts of mafic magma hybridization of the pre-existing andesite. Assuming the pre-existing andesite is close in composition to the most SiO₂-rich magma erupted, and assuming the mafic magma contains 52 wt % SiO₂, the amount of mafic magma hybridization would be in the 20–30 wt % range.

The alternation between Mg- and Si-rich zones and Al- and Fe-rich zones parallel to the growth surface in the large hornblende phenocrysts (Fig. 1b and c) can be explained as a record of temperature alternations in the magma similar to the one documented as occurring during the present eruption. There is approximately a one-for-one substitution of Al^IV for Si in each of these growth bands (Fig. 2), and the corresponding compositional exchange vector Si^IV + ( )^A = Al^IV + (Na, K)^A is considered to be strongly controlled by temperature (Helz, 1979; Blundy & Holland; 1990; Holland & Blundy, 1994). The increase in A site occupancy (Na + K) does not completely charge balance the tetrahedral Al-for-Si exchange that occurs (Fig. 7a), but the covariance of Ti and Fe³⁺ with increasing Al (Table 1) suggests that those cations may play such a role. The calculated structural formula of the end-member hornblende compositions in the phenocrysts indicates that the octahedral Al abundance is small and essentially equivalent in each end-member (Fig. 7b). Thus, the Tschermak substitution, Mg^VI + Si^IV = Al^IV + Al^VI, that is demonstrably indicative of pressure change at constant temperature (Johnson & Rutherford, 1989) explains little or none of the compositional variation within the hornblende phenocrysts. In contrast, hornblende crystallized in the andesite at 200 and 250 MPa contains higher Al^VI than the natural phenocrysts. The one aspect of the hornblende growth-zone chemistry that is somewhat surprising is the fact that the Mg-number of the low-temperature hornblende is higher than that of the high-temperature material (Fig. 7). However, the same two compositions are reproduced in the low- vs high-temperature experiments, verifying the initial conclusion. The trend in Mg-number with temperature may be the result of the complex reaction occurring to produce new-growth hornblende. Below we argue that the reaction involves the breakdown of pargasite, crystallization of melt that came from the mafic magma with the pargasite, and the dissolution of Fe–Ti oxides that were stable in the andesite at the lower temperature.

Although the experiments indicate that the Al-rich hornblende is stabilized by raising the temperature of andesitic magma equilibrated at 825°C and 130 MPa (water-saturated conditions) to 855°C (Fig. 7), very little new growth would occur in a closed-system magma without an equivalent amount of hornblende dissolution. The maximum temperature the andesite could have reached during heating is 860°C. The melt in the andesite at 860°C (Fig. 6) is rhyolitic (MgO 0·38 wt %) with essentially no capacity to crystallize hornblende. Although rounded outlines of a few phenocrysts (e.g. Fig. 1b) suggest there may have been minor hornblende dissolution during the present eruption, internal zone boundaries are remarkably planar and parallel to growth faces (Fig. 1d), suggesting that dissolution was not extensive in previous heating and cooling cycles. At the same time, some zones involve as much as 200 µm of newly added hornblende on crystals of 1 cm length (Fig. 1d). Each of the zones observed in the phenocrysts thus requires the addition of a significant hornblende component to the andesite. The pargasite-bearing mafic inclusions and the pargasite crystals in the groundmass described above appear to represent what remains after such a mafic magma addition. Pargasitic hornblende from the mafic magma will break down over time in the andesite because it is not stable in this magma, and the surface energy of the small grains is high. Pargasite breakdown, together with the crystallization of any melt from the mingled mafic magma, must be the source of the new growth, Al-rich hornblende on pre-existing crystals, as we observe only rare hornblende microphenocrysts in the andesite, and small hornblende crystals trapped in plagioclase or pyroxene phenocrysts are also rare. However, it is possible that the addition of a new growth layer on large hornblende phenocrysts also involves the growth of small hornblende crystals as a result of the hybridization process, and these crystals subsequently break down in an Ostwald-ripening process. On the basis of a crystal size distribution analysis, Higgins & Roberge (2003) reached a somewhat similar conclusion regarding the growth of the large hornblende phenocrysts in the Soufrière Hills magma, although they are not aware of the groundmass pargasite.

The time involved in the formation of one of the hornblende phenocryst growth zones is as yet undetermined. One possibility is that the outer growth zones represent magma influxes that have been identified on the basis of three seismic crises identified at Soufrière Hills in the past century (Shepherd et al., 1971). Presumably there were other magma injections between the Castle Peak eruption that occurred 400 years ago (Young et al., 1996) and the first of these historic seismic crises in 1896. Analytical (EMP and secondary ionization mass spectrometry) traverses across the zoned hornblende for elements with very different diffusivities may hold clues to the timing, and will be reported in a later paper. Another task that has yet to be completed is a correlation of these data for hornblende zoning with the zoning observed in plagioclase phenocrysts (Fig. 3) in these rocks. Some plagioclase phenocrysts clearly contain a series of high-An zones that might correlate with the heating and hybridization events recorded in the hornblende phenocrysts. Stewart & Fowler (2001) concentrated on the outermost An-rich zone in plagioclase phenocrysts, and suggested an origin involving mingling and non-equilibrium crystallization during magma ascent. This suggestion needs to be modified to accommodate the multiple growth zones observed in both plagioclase and hornblende. Ascent, at least from the storage zone at 5 km depth to the surface, does not appear to be required to produce the zoning. Although the zoning in titanomagnetite phenocrysts indicates a heating–hybridization just before the eruption of each batch of 1995–2002 magma (Devine et al., 2003), the presence of the cyclically zoned hornblende phenocrysts demonstrates that hybridization of unerupted magma must also occur in the Montserrat magma storage zone. It is interesting to note that similarly zoned hornblende phenocrysts have been observed in the 5000 km³ Fish Canyon tuff magma (Bachmann & Dungan, 2002) and in the 1980 Mount St. Helens dacite (Rutherford, 2002). This suggests that the hybridization-induced phenocryst growth process observed in the Soufrière Hills magma is common and occurs on large as well as small scales (Clynne, 1999; Venezky & Rutherford, 1999).

Petrologic indicators of Soufrière Hills magma ascent rates
Rutherford et al. (1998) noted the growth of reaction rims on hornblende in the Soufrière Hills andesite where the crystals were in contact with the groundmass, and interpreted them to have formed during magma ascent from the storage zone at 130 MPa (5–6 km depth). The reaction rims develop because of the decrease in water in the melt surrounding the phenocrysts with decreasing pressure, and the rim thickness is a function of the time the magma spends outside the hornblende stability field during magma ascent. Changes in the thickness of these reaction rims were measured for different batches of erupted magma, particularly those erupted from January–July 1996 (Devine et al., 1998b) when the rim thickness changed from 100 µm to only a few microns. We have continued to monitor and record the thickness of rims on hornblende in samples of magma obtained as the eruption progressed into mid-2002.

The original interpretation of the reaction rim thickness data in terms of magma ascent rate (Rutherford et al., 1998) was based on experiments that were performed at 900°C to investigate the Mount St. Helens dacite (Rutherford & Hill, 1993). The results of new decompression experiments carried out using the Soufrière Hills andesite as a starting material (Appendix) at 830, 845, and 860°C are plotted in Fig. 9, along with 900°C data for the Mount St. Helens dacite. Also shown are curves that illustrate the thickness of a reaction rim that would develop on hornblende in the Soufrière Hills andesite if some of the magma were held for various times at either 830 or 860°C and a pressure of 50 MPa. These data suggest that it is very unlikely that any of the erupted magma was held for any extended length of time at a pressure less than the 130 MPa pressure of phenocryst–melt equilibration (5–6 km), otherwise extensive hornblende reaction rims would exist on every grain in that sample. The rate of development of hornblende reaction rims in the constant rate decompression experiments from 130 to 4 MPa at 860°C in the Soufrière Hills andesite is very similar to that in a 160 to 2 MPa decompression of a Mount St. Helens dacite at 900°C (Fig. 9). In contrast, the rate of rim growth is measurably slower in a similar decompression of the andesite at 830°C. These observations are consistent with a temperature and melt viscosity control proposed for this reaction (Rutherford & Hill, 1993), although it is a little surprising that there is no measurable decrease in the growth rate in going from the 900°C dacite decompression to the 860°C curve for andesite. We attribute this observation to the difference in the position of the reaction that marks the upper stability of hornblende in the two magmas, the curve for andesite at any temperature (Fig. 5) lying at somewhat lower pressure.

View larger version (27K): [in this window] [in a new window]   Fig. 9. Plot of the rim thickness vs experiment duration for experiments on hornblende-bearing magma. The line for a 900°C decompression of Mount St. Helens dacite is from Rutherford & Hill (1993). The three bold lines show the rim thickness for decompression experiments (Appendix) of Montserrat andesite at different temperatures. Curves are also plotted for constant P and T annealing experiments on the andesite for 830 and 860°C temperatures.

 
If the temperature of the Soufrière Hills andesite magma had generally been raised to an average of 845°C before eruption, then the rate of magma ascent is given by the curve in Fig. 10 based on the thickness of reaction rims observed in the erupting andesite over time. If any of the erupting magma had a temperature as low as 830°C, the average ascent rate calculated for that magma would be 10% lower than is indicated in Fig. 10. Several features in this figure are noteworthy. First, the 0·019 m/s ascent rates shown for the periods just before and shortly after the 1998–1999 eruption hiatus are minimum rates because there were no rims on the hornblende crystals in these magmas. The relatively rapid rate of magma ascent during these periods is consistent with the fact that titanomagnetite phenocrysts present in the magma erupted during these periods show little or none of the zoning caused by late-stage heating. In addition, the occurrence of many vulcanian explosions at Soufrière Hills in the 1996–1997 rapid magma ascent period is what would be expected for relatively high magma ascent rates; equilibrium magma degassing with relatively low rates of gas loss. Clark et al. (2002) explained the variability and modulation of these explosions by calling for variable near-surface loss of the gas generated in the ascending magma in this period. An interesting question is why there have been so few explosions associated with the rapid ascent of magma in 2000 and 2001 compared with 1996–1997. Perhaps the average rate of ascent was actually significantly higher than the minimum of 0·019 m/s indicated by hornblende rims in the latter period. Another possibility is that the recently erupted magma is more successful at losing the gases that were exsolved during decompression because the lava dome and the upper conduit are now much more segmented and fractured by the continuing process of dome build-up and collapse.

View larger version (24K): [in this window] [in a new window]   Fig. 10. Plot of erupted volume and magma ascent rate for the Soufrière Hills andesite eruption from its initiation to September 2002. The ascent rates are calculated from measurements of hornblende breakdown rim thickness in thin sections of andesite samples and the 845°C curve in Fig. 9. In many samples there is a significant fraction of the hornblendes that have no rims, and the ascent rate of 0·019 m/s for these samples is a minimum estimate. The volume data are from Sparks et al. (1998) and the MVO staff (personal communication, 2002).

 
The dome volume curve (Fig. 10) is an extension of the data from Sparks et al. (1998) based on measurements by that group (R. S. J. Sparks, personal communication, 1998) and various MVO staff. Several points are interesting in a comparison of the ascent rate and dome volume curves. There was a continuing increase in the volume extrusion rate coincident with the increase in magma ascent rate throughout much of 1996 and 1997, indicating that the volume increase was actually the result of an ascent rate increase and not just an enlarging of the conduit. The sharp drop in magma ascent rate in autumn 1996 came after a major explosion and collapse of the existing dome. Shortly after the dome collapse there was a brief drop in the dome volume growth curve as well, but this was followed by almost a year of increasing dome volume growth rate during which the magma ascent rate was very high. This behavior appears consistent with the petrological model proposed above in which the magma storage zone at 5 km depth contains a limited amount of eruptable magma at any time, but is continuously being intruded by mafic magma that produces a hybridized andesite capable of erupting. A near-surface or conduit disturbance such as occurred in autumn 1996 may have briefly depleted the storage region of eruptable magma. A major crater-wall and lava dome collapse occurred in late December 1997, and approximately 3 months later surface eruption at Soufrière Hills ceased for a period extending from March 1998 to November 1999. Again, the petrology of the erupting magma suggests that a good explanation for this break in the eruption would be a break in the flow of mafic magma into the andesite storage zone. Less than 3 months after this break in the eruption, magma was again ascending from depth at a relatively very high rate (Fig. 10). One of the most interesting questions still to be answered is how the Soufrière Hills magma system was able to return to a rapid extrusion rate so quickly after a 20 month period of little or no flow through the conduit.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
In summary, a number of petrogenetically significant compositional and textural characteristics of amphibole have been identified in the Soufrière Hills andesite. Most of the hornblende in the andesite is present as large phenocrysts that contain several to as many as seven internal zoning cycles, each beginning with a sharp increase in MgO and SiO₂, and then becoming enriched in Al₂O₃ and FeO. Small lath-shaped pargasitic hornblende crystals are also present in the groundmass of magma erupted throughout the 1995–2002 period, as are microphenocrysts of An_80–90 plagioclase.

Phase equilibrium experiments on the andesite show that the phenocryst assemblage is stable in the range between 825 and 855°C at a pressure of 130 MPa, the pressure indicated by Al-in-hornblende geobarometry for the quartz-saturated magma. The 130 MPa pressure is consistent with quartz-hosted melt inclusion data for dissolved H₂O (Barclay et al., 1998) that indicate the quartz crystallized from a water-saturated interstitial melt at 130 MPa. An experimentally produced liquid line of descent for the andesite at 130 MPa reaches a matrix glass at 825°C that is compositionally equivalent to the average quartz-hosted melt inclusion. Compositions of hornblende produced in the experiments indicate that the Al₂O₃-rich zones in the hornblende phenocrysts formed at higher temperature (855°C) relative to the MgO- and SiO₂-rich zones. Pargasite crystals in the andesite are equivalent in composition and shape to crystals present in vesicular enclaves of mafic magma in the andesite, and those in the andesite are interpreted to represent mingling of the two magmas, and hybridization of the pre-existing andesite. Pargasitic amphibole is not stable in the andesite, and with time the small crystals present will break down and their components will be added to hornblende phenocrysts as new growth layers. The zoning in the interiors of hornblende phenocrysts in the andesite is interpreted to indicate that the hybridization process has occurred at least seven times during the history of this magma storage system.

On the basis of the time required to experimentally produce the high-Ti rims observed on titanomagnetite phenocrysts in the andesite, each magma sample erupted in the 1995–2002 period experienced a similar heating–hybridization event <30 days before eruption. These observations require a model of two-magma interaction similar to that proposed by Snyder (2000) and place severe time constraints on the timing of the interaction relative to magma eruption (Devine et al., 2003).

The reaction rims observed on the hornblende phenocrysts in the Soufrière Hills andesite are of two types. The rims consisting primarily of Ca-rich pyroxene were reproduced experimentally in andesite held only 2 days at 10–15°C above the hornblende stability field at 130–200 MPa. The presence of a few crystals with similar rims in many andesite samples is interpreted to mean that a small fraction of most erupted andesite was raised briefly (a few hours) to temperatures above the hornblende stability field. Decompression-induced reaction rims are present on many hornblende crystals in some samples of the andesite. A new calibration of the rim growth rate in a constant-rate, 845°C decompression of the andesite is considered appropriate for the average magma erupted at Soufrière Hills. Using this calibration, the hornblende breakdown rims in the andesite indicate that the magma ascent rate increased slowly from 0·001–0·002 m/s in early 1996 to >0·02 m/s in late 1996 and stayed high through 1997 and part of 1998. The rate dropped rapidly to zero at the 1998–1999 hiatus of the eruption, and then rapidly increased when the eruption began in November 1999, reaching the maximum detectable ascent rate again early in 2000. Because essentially all of the erupting magma is hybridized by injection of a mafic magma into the pre-existing andesite shortly before eruption, this drop in the magma ascent rate can best be explained by a break in the flow of mafic magma into the andesite storage zone.

	APPENDIX: HYDROTHERMAL EXPERIMENTS ON SOUFRIÈRE HILLS ANDESITE

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 

View this table: [in this window] [in a new window]

 

	ACKNOWLEDGEMENTS

 
This paper has been significantly improved by reviews by Dawnika Blatter, John Blundy, and Bruno Scaillet, although the interpretations of the data are those of the authors and are not necessarily endorsed by the reviewers. The authors are indebted to the staff of the Montserrat Volcano Observatory, who collected samples from the eruption whenever possible and made samples available for study.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS EXPERIMENTAL RESULTS DISCUSSION CONCLUSIONS APPENDIX: HYDROTHERMAL... REFERENCES

 
Andersen, D. J. & Lindsley, D. H. (1988). Internally consistent solution models for Fe–Mg–Mn–Ti oxides. American Mineralogist 73, 714–726.[Abstract]

Andersen, D. J., Lindsley, D. H. & Davidson, P. M. (1993). QUILF: a Pascal program to assess equilibria among Fe–Mg–Ti oxides, pyroxenes, olivine, and quartz. Computers and Geosciences 19, 1333–1350.

Bachmann, O. & Dungan, M. A. (2002). Temperature-induced aluminum zoning in hornblendes of the Fish Canyon magma, CO. American Mineralogist 87, 1062–1076.[Abstract/Free Full Text]

Barclay, J., Rutherford, M. J. & Carroll, M. R. (1998). Experimental phase equilibria constraints on pre-eruptive storage conditions of the Soufrière Hills magma. Geophysical Research Letters 25, 3437–3440.[CrossRef][Web of Science]

Blundy, J. D. & Holland, T. J. (1990). Calcic amphibole equilibria and a new amphibole–plagioclase geothermometer. Contributions to Mineralogy and Petrology 104, 208–224.[CrossRef][Web of Science]

Clark, A. B., Voight, B., Neri, A. & Macedonio, G. (2002). Transient dynamics of vulcanian explosions and column collapse. Nature 415, 897–401.[CrossRef][Medline]

Clynne, M. A. (1999). A complex magma mixing origin for rocks erupted in 1915, Lassen Peak, California. Journal of Petrology 40, 105–132.[CrossRef][Web of Science]

Couch, S., Sparks, R. S. J. & Carroll, M. R. (2001). Mineral disequilibrium in lavas explained by convective self-mixing in open magma chambers. Nature 411, 1037–1039.[CrossRef][Medline]

Devine, J. D., Gardner, J. E., Brack, H. P., Layne, G. D. & Rutherford, M. J. (1995). Comparison of microanalytical methods for estimation of H₂O contents of silicic volcanic glasses. American Mineralogist 80, 319–328.[Abstract]

Devine, J. D., Murphy, M. D., Rutherford, M. J., et al. (1998a). Petrologic evidence for pre-eruptive pressure–temperature conditions, and recent reheating, of andesitic magma erupting at the Soufrière Hills Volcano, Montserrat, W.I. Geophysical Research Letters 25, 3669–3672.[CrossRef][Web of Science]

Devine, J. D., Rutherford, M. J. & Gardner, J. E. (1998b). Petrologic determination of ascent rates for the 1995–1997 Soufrière Hills Volcano andesitic magma. Geophysical Research Letters 25, 3673–3676.

Devine, J. D., Rutherford, M. J., Norton, G. E. & Young, S. R. (2003). Magma storage region processes inferred from geochemistry of Fe–Ti oxides in andesitic magma, Soufrière Hills Volcano, Montserrat, W.I. Journal of Petrology 44, 1375–1400.[Abstract/Free Full Text]

Druitt, T. H., Young, S. R., Baptie, B., et al. (2002). Episodes of repetitive vulcanian explosions and fountain collapse at Soufrière Hills Volcano, Montserrat. In: Druitt, T. H. & Kokelaar, B. P. (eds) The Eruption of Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoir 21, 281–306.

Frost, B. R. & Lindsley, D. H. (1991). Occurrence of iron–titanium oxides in igneous rocks. In: Lindsley, D. H. (ed.) Oxide Minerals: Petrologic and Magnetic Significance. Mineralogical Society of America, Reviews in Mineralogy 25, 433–468.

Garcia, M. O. & Jacobson, S. S. (1979). Crystal clots, amphibole fraction, and the evolution of calc-alkaline magmas. Contributions to Mineralogy and Petrology 69, 319–327.[CrossRef][Web of Science]

Gardner, J. E., Carey, S., Rutherford, M. J. & Sigurdsson, H. (1995). Petrologic diversity in Mount St. Helens dacites during the last 4,000 years: implications for magma mixing. Contributions to Mineralogy and Petrology 119, 224–238.[Web of Science]

Hammerstrom, J. M. & Zen, E-A. (1986). Aluminum in hornblende: an empirical igneous geobarometer. American Mineralogist 71, 1297–1313.[Abstract]

Helz, R. T. (1979). Alkali exchange between hornblende and melt: a temperature-sensitive reaction. American Mineralogist 64, 953–965.[Abstract]

Higgins, M. D. & Roberge, J. (2003). Crystal size distribution of plagioclase and amphibole from Soufrière Hills volcano, Montserrat: evidence for dynamic crystallization–textural coarsening cycles. Journal of Petrology 44, 1401–1411.[Abstract/Free Full Text]

Holland, T. J. & Blundy, J. D. (1994). Non-ideal interactions in calcic amphiboles and their bearing on amphibole–plagioclase thermometry. Contributions to Mineralogy and Petrology 116, 431–447.

Johnson, M. C. & Rutherford, M. J. (1989). Experimental calibration of the aluminum-in-hornblende geobarometer with application to Long Valley caldera (California) volcanic rocks. Geology 17, 837–841.[Abstract/Free Full Text]

Martel, C., Pichavant, M., Holtz, F., Scaillet, B., Bourdier, J.-L. & Traineau, H. (1999). Effects of _O2 and H₂O on andesite phase relations between 2 and 4 kbar. Journal of Geophysical Research 104, 29453–29470.[CrossRef]

McCanta, M., Rutherford, M. J. & Hammer, J. (2001). Black Butte dacite dome, CA: a study of magma ascent rates using amphibole and plagioclase reactions and crystal size distributions. EOS Transactions, American Geophysical Union 82, 47.

Murphy, M. D., Sparks, R. S. J., Barclay, J., Carroll, M. R. & Brewer, T. S. (2000). Remobilization of andesite magma by intrusion of mafic magma at the Soufrière Hills Volcano, Montserrat, West Indies. Journal of Petrology 41, 21–42.[Abstract/Free Full Text]

Robinson, P., Spear, F. S., Schumacher, J. C., et al. (1982). Phase relations of metamorphic amphiboles: natural occurrence and theory. In Veblen, D.R., & Ribbe, P.H. (eds) Amphiboles: Petrology and Experimental Phase Relations. Reviews in Mineralogy 9B, 3–32.

Rutherford, M. J. (2002). Nature and timing of magma interaction processes in arc volcanic systems: data from rocks and from phase equilibria experiments. EOS Transactions, American Geophysical Union 83, 47.

Rutherford, M. J. & Devine, J. D. (1988). The May 18, 1980, eruption of Mount St. Helens, III. Stability and chemistry of amphibole in the magma chamber. Journal of Geophysical Research 93, 11949–11959.[CrossRef]

Rutherford, M. J. & Hill, P. M. (1993). Magma ascent rates from amphibole breakdown: an experimental study applied to the 1980–1986 Mount St. Helens eruptions. Journal of Geophysical Research 98, 19667–19685.[CrossRef]

Rutherford, M. J., Sigurdsson, H. & Carey, S. (1985). The May 18, 1980 eruption of Mount St. Helens, 1. Melt compositions and experimental phase equilibria. Journal of Geophysical Research 90, 2929–2947.

Rutherford, M. J., Devine, J. D. & Barclay, J. (1998). Changing magma conditions and ascent rates during the Soufrière Hills eruption on Montserrat. GSA Today 8, 1–7.

Scaillet, B. & Evans, B. W. (1999). The 15th of June eruption of Mount Pinatubo, l, Phase equilibria and pre-eruption P–T–_O2–P_H2O conditions of the dacite magma. Journal of Petrology 40, 381–411.[CrossRef][Web of Science]

Shepherd, J. B., Tomblin, J. F. & Woo, D. (1971). Volcano-seismic crisis in Montserrat, West Indies. Bulletin Volcanologique 35, 143–163.

Snyder, D. (2000). Thermal effects of the intrusion of basaltic magma into a more silicic magma chamber and implications for eruption triggering. Earth and Planetary Science Letters 175, 257–273.[CrossRef][Web of Science]

Sparks, R. S. J., Young, S. R., Barclay, J., et al. (1998). Magma production and growth of the lava dome of the Soufrière Hills Volcano, Montserrat, West Indies: November 1995 to December 1997. Geophysical Research Letters 25, 3421–3424.[CrossRef][Web of Science]

Spear, F. S. (1981). An experimental study of hornblende stability and compositional variability in amphibolite. American Journal of Science 281, 697–734.[Abstract/Free Full Text]

Stewart, M. L. & Fowler, A. D. (2001). The nature and occurrence of discrete zoning in plagioclase from recently erupted andesitic volcanic rocks, Montserrat. Journal of Volcanology and Geothermal Research 106, 243–253.[CrossRef][Web of Science]

Venezky, D. Y. & Rutherford, M. J. (1999). Petrology and Fe–Ti oxide re-equilibration of the 1991 Mount Unzen mixed magma. Journal of Volcanology and Geothermal Research 89, 213–230.[CrossRef][Web of Science]

Young, S. R., Hoblitt, R. P., Smith, A. L., Devine, J. D., Wadge, G. & Shepherd, J. B. (1996). Dating of explosive volcanic eruptions associated with dome growth at the Soufrière Hills volcano, Montserrat, West Indies. In: Second Caribbean Conference on Natural Hazards and Hazard Management. MVO Open File Report 96/22, Kingston, Jamaica, 1996.

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				C. D. Deering Cannibalization of an amphibole-rich andesitic progenitor induced by caldera-collapse during the Matahina eruption: Evidence from amphibole compositions American Mineralogist, August 1, 2009; 94(8-9): 1162 - 1174. [Abstract] [Full Text] [PDF]

				K. Genareau, A. B. Clarke, and R. L. Hervig New insight into explosive volcanic eruptions: Connecting crystal-scale chemical changes with conduit-scale dynamics Geology, April 1, 2009; 37(4): 367 - 370. [Abstract] [Full Text] [PDF]

				O. Reubi and J. Blundy Assimilation of Plutonic Roots, Formation of High-K 'Exotic' Melt Inclusions and Genesis of Andesitic Magmas at Volcan De Colima, Mexico J. Petrology, December 24, 2008; (2008) egn066v1. [Abstract] [Full Text] [PDF]

				F. Aydin Contrasting complexities in the evolution of calc-alkaline and alkaline melts of the Nigde volcanic rocks, Turkey: textural, mineral chemical and geochemical evidence European Journal of Mineralogy, February 1, 2008; 20(1): 101 - 118. [Abstract] [Full Text] [PDF]

				J. Blundy and K. Cashman Petrologic Reconstruction of Magmatic System Variables and Processes Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 179 - 239. [Full Text] [PDF]

				M. J. Rutherford Magma Ascent Rates Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 241 - 271. [Full Text] [PDF]

				M. J. Streck Mineral Textures and Zoning as Evidence for Open System Processes Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 595 - 622. [Full Text] [PDF]

				M. B. Turner, S. J. Cronin, R. B. Stewart, M. Bebbington, and I. E.M. Smith Using titanomagnetite textures to elucidate volcanic eruption histories Geology, January 1, 2008; 36(1): 31 - 34. [Abstract] [Full Text] [PDF]

				N. G. Lensky, R. S. J. Sparks, O. Navon, and V. Lyakhovsky Cyclic activity at Soufriere Hills Volcano, Montserrat: degassing-induced pressurization and stick-slip extrusion Geological Society, London, Special Publications, January 1, 2008; 307(1): 169 - 188. [Abstract] [Full Text] [PDF]

				K. Genareau, R. Hervig, and A. Clarke Geochemical variations in late-stage growth of volcanic phenocrysts revealed by SIMS depth-profiling American Mineralogist, August 1, 2007; 92(8-9): 1374 - 1382. [Abstract] [Full Text] [PDF]

				Y. Suzuki and S. Nakada Remobilization of Highly Crystalline Felsic Magma by Injection of Mafic Magma: Constraints from the Middle Sixth Century Eruption at Haruna Volcano, Honshu, Japan J. Petrology, August 1, 2007; 48(8): 1543 - 1567. [Abstract] [Full Text] [PDF]

				M. Viccaro, C. Ferlito, and R. Cristofolini Amphibole crystallization in the Etnean feeding system: mineral chemistry and trace element partitioning between Mg-hastingsite and alkali basaltic melt European Journal of Mineralogy, July 1, 2007; 19(4): 499 - 511. [Abstract] [Full Text] [PDF]

				M. C. S. HUMPHREYS, J. D. BLUNDY, and R. S. J. SPARKS Magma Evolution and Open-System Processes at Shiveluch Volcano: Insights from Phenocryst Zoning J. Petrology, December 1, 2006; 47(12): 2303 - 2334. [Abstract] [Full Text] [PDF]

				C. Martel, A. Radadi Ali, S. Poussineau, A. Gourgaud, and M. Pichavant Basalt-inherited microlites in silicic magmas: Evidence from Mount Pelee (Martinique, French West Indies) Geology, November 1, 2006; 34(11): 905 - 908. [Abstract] [Full Text] [PDF]

				C. BEIER, K. M. HAASE, and T. H. HANSTEEN Magma Evolution of the Sete Cidades Volcano, Sao Miguel, Azores J. Petrology, July 1, 2006; 47(7): 1375 - 1411. [Abstract] [Full Text] [PDF]

				C. ANNEN, J. D. BLUNDY, and R. S. J. SPARKS The Genesis of Intermediate and Silicic Magmas in Deep Crustal Hot Zones J. Petrology, March 1, 2006; 47(3): 505 - 539. [Abstract] [Full Text] [PDF]

				J. D. DEVINE, M. J. RUTHERFORD, G. E. NORTON, and S. R. YOUNG Magma Storage Region Processes Inferred from Geochemistry of Fe-Ti Oxides in Andesitic Magma, Soufriere Hills Volcano, Montserrat, W.I. J. Petrology, August 1, 2003; 44(8): 1375 - 1400. [Abstract] [Full Text] [PDF]

				S. COUCH, R. S. J. SPARKS, and M. R. CARROLL The Kinetics of Degassing-Induced Crystallization at Soufriere Hills Volcano, Montserrat J. Petrology, August 1, 2003; 44(8): 1477 - 1502. [Abstract] [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (47) Request Permissions Google Scholar Articles by RUTHERFORD, M. J. Articles by DEVINE, J. D. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Online ISSN 1460-2415 - Print ISSN 0022-3530

Copyright © 2009 Oxford University Press

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics