| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 8 | Pages 1477-1502 | 2003
© Oxford University Press 2003
The Kinetics of Degassing-Induced Crystallization at Soufrière Hills Volcano, Montserrat
1 DEPARTMENT OF EARTH SCIENCES, BRISTOL UNIVERSITY, BRISTOL BS8 1RJ, UK
2 SCHOOL OF ENVIRONMENTAL SCIENCE, UNIVERSITY OF EAST ANGLIA, NORWICH NR4 7TJ, UK
3 DIPARTMENTO DE SCIENZE DELLA TERRA, UNIVERSITA DI CAMERINO, 62032 CAMERINO, ITALY
* Corresponding author. Present address: Department of Earth Sciences, Bristol University, Bristol BS8 1RJ, UK.E-mail: S.Couch{at}mail.com
RECEIVED MAY 31, 2002; ACCEPTED FEBRUARY 10, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
A series of decompression experiments at 875°C, with a 16 h anneal period at 160 MPa, and depressurizations to a final water pressure (Pf) of 125–30 MPa, were run for between 1 and 504 h. Using an experimentally derived plagioclase liquidus, the depressurizations are estimated to impose an undercooling (
T) of between 38 and 151°C. The experiments show that there is a delay of 
1–4 h in the nucleation of plagioclase crystals after decompression. Over the range of pressures considered, nucleation rates vary up to 32 sites/mm3 per s, and increase with undercooling. The growth rate of plagioclase increases with T, reaching a maximum of 1·7 x 10-6 mm/s at a Pf of 75 MPa. Comparison of the experimental results with natural samples of the Soufrière Hills andesite suggests that pumice samples did not experience any decompression-induced crystallization and therefore ascent from the magma chamber must have been completed within 4 h. Dome samples are variable in texture, but all contain numerous small microlite crystals with low plagioclase anorthite contents and larger microlites with albitic overgrowth rims. Experimental results show that these textural differences can be related to variations in ascent rate. KEY WORDS: crystallization kinetics; experiments; microlites; Montserrat; textural analysis
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Recent dome-forming eruptions, such as Mount St. Helens (USA), Mount Unzen (Japan) and Soufrière Hills (Montserrat), have provided opportunities for detailed study of many volcanic processes. Complex eruptive behaviour can be related to volatile loss during magma ascent (Eichelberger et al., 1986; Melnik & Sparks, 1999; Nakada & Motomura, 1999; Sparks et al., 2000; Blundy & Cashman, 2001). H2O is usually the predominant volatile, and water exsolution as a result of decompression raises liquidus temperatures. Thus magma that was at, or above, the liquidus at elevated pressures can become significantly undercooled during decompression, causing extensive groundmass crystallization. The exsolution of water from the melt may result in cooling (Sahagian & Proussevitch, 1996; Zhang, 1999a; Mastin & Ghiorso, 2001); however, this will be balanced by latent heat release during crystallization. For the Soufrière Hills andesite the latent heat release is estimated to be up to 40°C (Couch et al., 2003). At high undercoolings, crystallization is likely to be dominated by nucleation at new sites (Lofgren, 1980; Kirkpatrick, 1981), producing microlite crystals [arbitrarily defined as <100 µm in length by Murphy et al. (2000)]. These extra crystals can constitute a significant proportion of the crystal fraction; up to 22% in Pinatubo dacites (Hammer et al., 1999) and 18% in Mount St. Helens dacites (Geschwind & Rutherford, 1995). This crystallization, combined with the decreased water content of the melt phase, can cause the magma viscosity to increase by several orders of magnitude (Lejeune & Richet, 1995; Sparks et al., 2000). Magma can reach the surface as a hot, largely crystalline solid with little residual melt remaining (Voight et al., 1999; Watts et al., 2002).
Textural analysis has identified the significant role that late-stage microlite crystallization can play in shallow magmatic processes (Cashman, 1992; Hammer et al., 1999, 2000). Recently, decompression and its role in crystallization have been considered in the petrogenesis of andesites (Cashman & Blundy, 2000; Blundy & Cashman, 2001), utilizing work by Tuttle & Bowen (1958) and Holtz et al. (1992). Many previous experimental studies investigated crystallization induced by cooling (Lofgren, 1974; Fenn, 1977; Swanson, 1977; Muncill & Lasaga, 1987, 1988; Dunbar et al., 1995; McCoy & Lofgren, 1999). However degassing, not cooling, is the major cause of groundmass crystallization in many dome eruptions. Only a few experimental studies have investigated decompression-induced crystallization. Geschwind & Rutherford (1995) studied the crystallization of plagioclase microlites during the 1980–1986 eruption at Mount St. Helens, and Hammer & Rutherford (2002) carried out decompression experiments on dacite from Mount Pinatubo so as to understand crystallization kinetics. This paper describes experimental and petrological studies of degassing-induced crystallization using an analogue for the composition of the groundmass of the Soufrière Hills andesite. Experimental results are compared with natural samples from recently erupted Soufrière Hills lava. This paper compares the equilibrium experiments and petrological studies of Couch et al. (2003) with results from decompression experiments.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Experimental procedure
A starting groundmass composition (aMon6a) was estimated by averaging 200-rastered electron microprobe analyses of the dome rock sample MVO34 (Barclay et al., 1998) and was prepared by A.-M. Lejeune, as described by Couch et al. (2003). This composition was used because the groundmass is of particular interest in understanding decompression-induced crystallization. The temperature of all experiments was set at 875°C, a representative value of the pre-eruptive conditions based on previous estimates (Devine et al., 1998a; Murphy et al., 2000; Couch et al., 2001). The cold-seal experimental set-up has been described by Couch et al. (2003). All runs were annealed for 16 h at 875°C and a water pressure of 160 MPa. These conditions were slightly below the plagioclase liquidus of
910°C at 160 MPa. This procedure generated a melt composition thought to be representative of the melt phase in the magma chamber before ascent. It generated bubbles and also seed crystals to allow growth on crystals to occur. The annealing water pressure of 160 MPa is greater than the minimum estimated magma chamber water pressure (130 MPa, Barclay et al., 1998), but a lower anneal pressure would have resulted in a significant amount of crystallization, which would have made distinguishing anneal-related crystals from decompression crystals difficult.
After annealing, the experimental charge was decompressed. Depressurization was controlled by a manually operated piston-screw pump, which allowed near-instantaneous decompressions. Decompression induces an undercooling (T) based on the equilibrium plagioclase liquidus determined in equilibrium experiments (see fig. 4 of Couch et al., 2003). As the final pressure (Pf) decreases, T increases. However, plagioclase crystallization during the annealing stage changes the melt composition and shifts the temperature of the plagioclase liquidus of the residual melt. Contours of K2O content in the equilibrium glass compositions are approximately parallel to the plagioclase liquidus and can be used to estimate its position. The plagioclase liquidus of the annealed melt is shifted toward lower temperatures relative to the aMon6a plagioclase liquidus, indicating that the annealed melt represents a similar amount of plagioclase crystallization at different pressures and temperatures (Fig. 1). Experimental uncertainties of the plagioclase liquidus and undercoolings are estimated as ±10°C, based on the accuracy of the experimental set-up.
|
Two types of experiments were carried out; single decompression experiments (SDE) and multi-step decompression experiments (MDE). SDE experiments (Fig. 2a) involved depressurization from 160 to 125, 100, 75, 50 or 30 MPa, which imposes
T of 38, 66, 98, 132 or 151°C (±10°C), respectively. After decompression, the pressure line was closed and the experiment left to run for periods ranging from 1 h to 504 h (3 weeks) and then quenched. For the MDE experiments the total pressure change from the annealing pressure of 160 MPa to the final pressure was divided into eight equal decompression steps. The sample was left for a fixed period between each step (Fig. 2b). By varying the time between steps (1 h, 12 h, 24 h) a range of decompression rates can be approximated (13·75–0·18 MPa/h). These experiments investigated how crystallization kinetics are affected by different values of T, and also attempted to replicate eruptive processes. The SDE represents a sudden decompression, such as happens after a major dome collapse or an explosion, and the MDE simulates the type of pulsatory behaviour observed at Soufrière Hills Volcano (Voight et al., 1999; Wylie et al., 1999).
|
Analytical techniques
Polished and carbon-coated thin sections of experimental charges and natural samples were analysed on a Cameca Camebax Micro using SamX software with PAP correction procedures for plagioclase and on a JEOL JXA8600 with XAF correction procedures for glass at 15 kV accelerating voltage, as described by Couch et al. (2003). Analysis of core and rims of plagioclase microlites was carried out, and the length and width of the crystal measured. The number of analyses varied from eight to 15.
Image analysis
Experimental samples
To determine plagioclase proportions and crystal dimensions, five backscatter electron images at different scales were collected for each sample using a Cambridge Stereoscan MK250 scanning electron microscope. Plagioclase fraction (
) and number density (number/mm2, Na) were determined, using methods similar to those of Hammer et al. (1999). Typically about 50 crystals per field of view were measured, although the count was lower in some low-crystallinity samples. Crystals that were cut by the image edge were included in the count of nucleation sites. Crystals that were skeletal or spherulitic were considered to be a single nucleation site. Determination of nucleation sites in high-T experiments was difficult because of the dendritic morphology.
Characteristic crystal size, s (mm), and the volumetric number density, Nv (mm-3) were calculated from
 | (1) |
 | (2) |
Crystal sizes were adjusted for stereological effects, assuming an approximately spherical shape (Kellerhalls et al., 1975; Wiebel, 1980). The Nv calculation uses a standard method for correcting the area- to volumetric-density (Cheng & Lemlich, 1983). The incremental nucleation rate, Ii (mm-3 s-1), can be estimated from
 | (3) |
Here Nvb and Nva are the nucleation densities at times ta and tb for two experiments with the same T and different durations. As duration increases, the overall number of nucleation sites should remain constant, or should increase. As a result of experimental scatter, lines of best fit were plotted and used to calculate s and Nv as a function of time. The incremental nucleation rate was calculated only for SDE. For the MDE a time-averaged, and therefore minimum rate, Im, was calculated, where
 | (4) |
Incremental growth rate, Gi, is approximated as the average of the half-length (L) of the 10 longest plagioclase crystals, as employed by Fenn (1977) and Hammer & Rutherford (2002):
 | (5) |
 | (6) |
Hammer et al. (1999) compared these techniques with CSD analysis and found similar overall results. Furthermore, these methods allow analysis of more samples than by CSD analysis. Stereological corrections are needed for CSD analysis, which is difficult for hopper, swallowtail and skeletal morphologies (Hammer & Rutherford, 2002). Histograms of area data use bin sizes of area, starting at 20 µm2 and increasing at 20 µm intervals to a maximum of 1000 µm2. Areas greater than 1000 µm2 were classified as ‘more than 1000 µm2’. The crystal length to width (aspect ratio) was calculated using the dimensions of the 10 longest plagioclase crystals.
Natural samples
The textures of four representative samples from the recent eruption, ranging from vesicular pumice to dense dome lava, and two prehistoric Soufrière Hills andesites were analysed, using the same methods as for the experimental samples. However, the natural samples contained a wider range of plagioclase crystals, from several millimetres to <1 µm. For each sample, several (2–6) fields of view were selected and three backscattered electron images taken at different magnifications to measure the area fraction, nucleation density and dimensions of plagioclase crystals for the size ranges 50–100 µm, 10–50 µm and 0–10 µm. Proportions of crystals >100 µm, other mineral phases <100 µm (pyroxenes, amphiboles, oxides and quartz), vesicles and glass were also determined. Modal analysis of the thin sections determined the proportion of the various phenocryst and microphenocryst phases. The groundmass modes were converted from volume percent to weight percent on a vesicle-free and >100 µm crystal-free basis.
|
EXPERIMENTAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Time-scales of gas exsolution
Loss of water from the melt changes the liquidus temperature of plagioclase (and other phases), ultimately driving crystallization. If H2O exsolution is slow compared with the experimental duration, the actual undercooling is reduced, as a result of the melt remaining supersaturated. Consequently, nucleation of new crystals could be delayed and growth rates reduced because of the lower real undercooling. Thus time-scales for H2O exsolution need to be considered.
Experimental studies by Gardner et al. (1999) have found that decompressing rhyolitic melts results in equilibrium degassing for slow decompressions (<0·025 MPa/s), but supersaturation can occur at faster decompressions (>0·25 MPa/s). However, Mangan & Sisson (2000) carried out experiments decompressing silicic melts without nucleation sites, with homogeneous nucleation of gas bubbles. They found a delay in bubble nucleation until supersaturations (P) of 130–150 MPa, and consequently deduced that degassing was a disequilibrium process. Because bubbles are observed in all our experiments, these results suggest that the experimental decompressions involve near-equilibrium degassing.
A method for assessing time-scales of water exsolution is to consider the distance (
) between bubbles and to estimate the time for equilibrium by diffusion (Navon & Lyakhovsky, 1998):
 | (7) |
10-7 cm2/s at 6 wt % H2O, decreasing to 10-8 cm2/s at <1 wt % H2O. Water content is estimated to range from 6·4 to 3 wt % H2O in our experiments, using the by-difference method (Devine et al., 1995). Hence D is estimated to be 10-7 cm2/s. Scanning electron microscopy (SEM) images of experimental products give a maximum distance between bubbles of 70 µm. Hence a characteristic diffusion time of 2 min is estimated for typical experiments. As the melt was water saturated at the annealing conditions (160 MPa), pre-existing bubbles formed during the annealing period would have facilitated water exsolution. Water exsolution therefore typically occurs over short time-scales (minutes) compared with the total experimental duration (hours).
Stable phases and compositional trends
Table 1 summarizes the key experimental data on compositions and textures, distinguishing SDE, MDE and repeat experiments. Figures 3 and 4 display characteristic images of SDE and MDE at different T values. Two experiments (sc93 and sc101) resulted in greater degrees of crystallization than expected. It is thought that the position of the charge within the vessel changed relative to the other runs. These samples have been included, as there is generally only a slight impact on most results. These experimental results were, however, omitted in the estimation of nucleation rates. Clinopyroxene, oxides and plagioclase crystallized in all experiments. In some low-Pf experiments a silica phase crystallized.
|
|
|
Annealing conditions
The shortest duration (1 h) decompression experiments were used to estimate the starting melt composition after annealing at 875°C and 160 MPa, because no further crystallization occurred during decompression. Total crystallinity was estimated using both image analysis and mass balance using K2O contents in the glass. Crystallinities of these experiments vary from 3·1 to 11·9 wt %. Experiment sc212 (Pf 100 MPa, 1 h duration) is the most representative of the degree of crystallinity found in the equilibrium experiment run at the annealing conditions of Couch et al. (2003), and hence is deemed the starting composition for all the decompression experiments. The inferred starting melt composition is enriched in SiO2 and depleted in CaO, MgO and FeO compared with the aMon6a composition (Table 2), consistent with clinopyroxene, plagioclase and titanomagnetite crystallization during annealing. Crystallization during the annealing period yields a melt similar to the natural melts discussed by Couch et al. (2003).
|
) of analyses renormalized to 100% anhydrousCompositional variations in plagioclase
Plagioclase compositions in a crystal–melt assemblage can constrain crystallization conditions. The equilibrium experiments (Couch et al., 2003) established how plagioclase composition varies with pressure and temperature, and these results can be compared with the products of the decompression experiments. Crystals formed during the annealing period were identified by their cores of An
60; however, these were not identified in every experiment. In some experiments the An60 core was overgrown by a more albitic rim, caused by decompression-induced growth. In the MDE it was not always clear whether the core of a crystal was related to annealing or had grown during one of the first high-pressure decompression steps and so no distinction was made. Plagioclase composition in SDE at any given T is generally similar to equilibrium compositions (Fig. 5), regardless of duration. For the MDE the anorthite contents are significantly higher than those found in either the SDE or equilibrium experiments. At Pf 50 MPa the lowest anorthite content is observed for the shortest duration, fastest decompression rate, MDE experiment.
|
Compositional variations in glass
Variations in glass composition (Table 2) can be used with textural variations to understand how
T and time affect the degree of crystallization. For the SDE, K2O increases with T and time, particularly after silica saturation is reached (Fig. 6). The MDE glass compositions have a more limited range, as silica saturation is not reached. The trends of the glass composition are unaffected by the decompression path (i.e. a single step or multiple steps) and are controlled only by the total amount of crystallization. The absence of silica saturation for the MDE is due to the lower degree of crystallization. Melt proportions have been plotted using the projection scheme of Blundy & Cashman (2001). Both types of experiment follow a tightly constrained trend (Fig. 7), with a shift towards the orthoclase–quartz join for the SDE at the onset of saturation of crystalline silica.
|
|
Textural observations
Plagioclase morphology
Crystal morphology varies systematically in SDE with
T, from planar tabular, to hopper, swallowtail, skeletal, dendritic and finally chain morphology at the highest T (Table 3). A more limited variation in morphology is observed in the MDE (Figs 3 and 4). In the runs where no decompression-induced crystallization is observed, the anneal-period crystals are tabular, as expected with crystallization at small undercoolings.
|
Variation in plagioclase proportion
Figure 8 compares plagioclase proportions from the equilibrium results (Couch et al., 2003) with those induced by decompression as estimated by image analysis. The SDE are close to the equilibrium trend, showing a steady increase in plagioclase as Pf decreases. For MDE plagioclase proportion increases as Pf decreases, but proportions are consistently lower than for SDE by up to 8 vol. %. The proportion of plagioclase increases with duration for most series of experiments (Fig. 9). In the case of SDE with a Pf of 125–75 MPa (
T 38–98°C) there is little new plagioclase crystallization until an experiment has run for more than 4 h. Then the bulk of the crystallization takes place within 24 h, after which there is little further crystallization. For experiments decompressed to 50 MPa (T = 132°C), crystallization initiates after 1 h and stabilizes within 4–8 h. For the MDE the short duration (8 h) experiments all show similar amounts of crystallization (5–6 vol. %). For slower decompression rates over 96 and 192 h, two trends can be discerned, despite the scatter. For those experiments with large final undercoolings (T = 98 and 132°C), the proportion increases substantially. For the experiments with small final undercoolings (T = 38 and 66°C) the data indicate either slight or no increase in plagioclase proportion.
|
|
Aspect ratio of plagioclase crystals
There is an overall increase in aspect ratio as
T increases for both SDE and MDE (Fig. 10). For the MDE, as the duration of the experiment increases the aspect ratio decreases. Most SDE have an aspect ratio >10 and most MDE have an aspect ratio <10.
|
Nucleation density
Previous studies (Fenn, 1974; Swanson, 1977; Berkebile & Dowty, 1982) have noted crystal nucleation from the container walls. However, wall-based nucleation was not observed in our experiments, in agreement with the results of Hammer & Rutherford (2002). Heterogeneous nucleation on vesicles was identified (Fig. 3b and h), as also reported by Davis & Ihinger (1998) and Schmelzer et al. (1993). Nucleation density increases as
T increases (Table 1). For the SDE, all experiments that ran for 1 h lacked new crystals. For experiments with high T, nucleation of new crystals took place in <4 h, and for lower T, nucleation occurred within 8 h. Once nucleation began, the number of nucleation sites became approximately constant in <24 h. For MDE at high decompression rates (high Ttotal and a short experimental duration), nucleation density is large, whereas at low decompression rates (low Ttotal and long experimental duration) the nucleation density is small. The nucleation densities for the MDE are much lower than for similar experimental conditions (Pf and final T ) in SDE, and the difference is particularly pronounced for low decompression rates. Thus for MDE crystallization the role of crystal growth becomes increasingly important relative to new nucleation as decompression rate decreases.
Nucleation lag
The time delay between the decompression event and the onset of crystallization is known as a ‘nucleation incubation’ (Fenn, 1974). We infer a minimum nucleation lag for the SDE only. For 1 h experiments, no new nuclei have formed and the low nucleation density relates to crystals formed during annealing. For T of 38, 66 and 98°C nucleation densities are similar to the 1 h quench after 4 h, but much higher after 8 h, implying a nucleation lag of 4 h or more. For T of 66 and 98°C nucleation is observed in the 4 h experiments and for T 38°C nucleation is still absent. Experiments with T 132–151°C have high nucleation densities after 4 h, suggesting that the nucleation delay is at least 1 h. For the MDE no estimate of nucleation lag could be made because of the difficulty of distinguishing crystals formed during annealing from those formed by decompression.
Nucleation and growth rates
For the SDE, nucleation rate increases with T, possibly reaching a peak at T 150°C (Fig. 11a). Maximum nucleation rate occurs at around 4–8 h after decompression, and then rapidly decreases to low levels. For the MDE the nucleation rate appears more dependent on the duration of the experiment than final T (Fig. 11b) and the nucleation rates are considerably lower for the MDE than for the SDE.
|
For SDE and MDE, growth rates are mostly in the range 10-6–10-8 mm/s (Fig. 12). For SDE, the highest growth rates are observed in the short duration experiments, with growth rate decreasing with time for a given undercooling (Fig. 12a). The fastest growth rates in the short duration experiments do not vary greatly with
T. For longer times the growth rate decreases from a maximum at the smallest undercooling (T 38°C). For the MDE the growth rate is highest in the rapidly decompressed experiments and decreases for slower decompressions, which is partly an artefact of averaging growth rates over the duration of each experiment.
|
The nucleation and growth rates of the SDE can be considered in terms of the effective undercooling, which is an estimate of the likely undercooling still imposed after some crystallization has taken place. It is estimated from the bulk crystallinity of a given experiment, which in turn is determined by mass balance calculations based on the melt composition. With this, the position of the plagioclase liquidus after this crystallization can be approximated. Hence the effective undercooling is the difference between the ‘new’ plagioclase liquidus and the conditions of the experiment. The effective undercooling should trend towards zero as equilibrium is reached. Both nucleation and growth rates decrease as the effective nucleation rate decreases (Fig. 13), confirming that undercooling is the driving force of the crystallization process.
|
Crystal area distributions
For any
T the number of crystals increases relatively rapidly as the duration of the experiment increases and then stabilizes. As T increases the number density of crystals increases and for the SDE the area (mean size) of the crystals decreases. Crystals formed during the initial 160 MPa anneal period explain the tail of larger crystals observed in the SDE graphs (Fig. 14). In the MDE the number density of crystals is highest for the short duration experiments (1 h step) regardless of Pf. However, the number of crystals is considerably lower than for the SDE. The shape of these distributions is similar to those of some of the longer duration, small T, SDE, with a peak at low crystal area. For the experiments of longer duration (12 and 24 h steps) there is a relatively small, but even distribution of crystal areas, with a peak of very large crystals (>1000 µm2), which are different from the SDE.
|
Experimental precision
Two experimental conditions were repeated and found to be in good agreement for various kinetic indicators for the SDE experiments (Table 1), in particular the nucleation density, which is especially sensitive to slight changes in pressure and temperature. For MDE repeat experiments, the composition of new crystals is closely similar between the experiments, but the amount of plagioclase crystallization, nucleation density and the estimated growth rates are more disparate. The diminished reproducibility of the MDE may relate to the number of decompression events in a given experiment. By having eight decompression steps, errors relating to the exact amount and timing of depressurization are increased.
|
DISCUSSION OF EXPERIMENTAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
The role of
TDepressurization induces crystallization of magma to form plagioclase and other phases such as pyroxene and Fe–Ti oxides because the melt is moved below the liquidi for those phases as water exsolves during this decompression process. Evidence to support this includes the increase in plagioclase proportion as
T increases and also the changes in glass composition with T. The anorthite content decreases as T increases. Equilibrium experiments show that as water pressure decreases, the plagioclase becomes more sodic. Thus the decrease in anorthite content as experimental duration increases in the SDE can be interpreted as the system seeking equilibrium at the lower pressure. For the MDE experiments, the decrease in anorthite content with time is also related to the system attempting to re-equilibrate, but the anorthite content does not reach the values observed in the equilibrium experiments.
The lower crystal proportions of MDE compared with SDE, and the difference in anorthite content, can be explained in terms of the MDE experiments being closer to fractional crystallization and the SDE being closer to equilibrium crystallization. In SDE the total undercooling is instantly imposed and the plagioclase composition at that pressure and temperature forms. For MDE with identical final pressure and temperature conditions, crystallization occurs from the starting pressure through a range of T values. Crystals with high anorthite content crystallize and so melt evolution is different, resulting in less final crystallization and more strongly zoned crystals in the MDE than in SDE.
Figure 15 shows a classic kinetic relationship, providing a qualitative basis for the interpretation of experimental results. For SDE experiments, increasing T moves the system from growth confined to seeds (i.e. pre-existing anneal crystals), to growth-dominated crystallization, with low nucleation rates, to nucleation-dominated growth at large T. For the MDE, the experiments at higher decompression rates (short experimental duration) experience larger values of T than lower decompression rates (long experimental duration). The actual T values experienced decrease as decompression rate decreases, explaining why nucleation rate decreases as experimental duration increases. The T imposed at a given decompression interval influences crystal morphology. If T is high, long, thin crystals will grow, whereas at small T crystal growth occurs on existing sites, resulting in lower aspect ratios.
|
The importance of time
A nucleation lag of between 1 and 4 h has been identified. However, after this lag crystallization can be rapid and extensive and will re-establish equilibrium (Fig. 9). A rapid ascent rate (fast depressurization) will not permit significant crystallization, regardless of the value of
T, whereas, if the depressurization rate is sufficiently low, crystallization can be significant and maintain equilibrium. This capacity to switch from crystal-poor to crystal-rich groundmass with only slight changes in decompression rate is an important feature of the system.
Comparison with other experimental results
Textural trends are similar to those observed by Lofgren (1974) and Hammer & Rutherford (2002), who identified that, as T increases, feldspar morphology systematically varies from planar, to hopper, swallowtail and dendritic forms. Dynamic cooling experiments performed by Lofgren (1977) showed that the starting texture strongly controls the final texture. If there were numerous existing nucleation sites, crystallization was predominantly growth based. Where there were few nucleation sites at the start of cooling-induced crystallization, the texture was considerably more variable, with considerable nucleation of new crystal sites, which is in agreement with the results of Fokin et al. (1999).
The Montserrat experimental results agree with observations in other experimental studies (Fenn, 1977; Swanson, 1977; Muncill & Lasaga, 1987, 1988; Burkhard, 2002; Hammer & Rutherford, 2002) that both nucleation and growth rates increase with T, reaching a maximum before decreasing at very high T. The growth rate maximum is always observed at lower T than the nucleation maximum.
Hammer & Rutherford (2002) performed similar decompression experiments on Pinatubo dacite to study feldspar crystallization kinetics. The starting material was crushed pumice, containing phenocrysts and glass with 78 wt % SiO2 (anhydrous). Experiments were annealed at high pressure and then decompressed either in single- or multi-decompression steps. Results suggested that at Ttotal 34–93°C growth on existing crystals was dominant, and at Ttotal 125–241°C microlite nucleation was dominant. The peak in growth rate is at a similar T to that observed for the Montserrat composition. The peak nucleation rate for the Pinatubo dacite is estimated to be at T 210°C, whereas for Montserrat the peak is at T 150°C. It is possible that for Montserrat the nucleation rate is greater at higher T, but pertinent experiments were not carried out.
|
STUDIES OF NATURAL SAMPLES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Experimental products and natural samples were compared. Variations in plagioclase composition and glass composition have been presented by Couch et al. (2003). Textures of six samples were analysed (Fig. 16), and these results, together with important compositional values, are summarized in Table 4.
|
|
Textural results
Glass proportions were estimated using the K2O content of the glass (Couch et al., 2003). Comparisons with the image analysis estimates of glass content suggest that there is reasonable agreement between methods (Table 4), except for a pumice sample (MVO1109d). The image analysis method may underestimate vesicular samples. As plagioclase microlite content increases, glass content decreases (Fig. 17), whereas pyroxene and oxide proportions vary unsystematically. Crystalline silica is only found with high-plagioclase fractions. The increase in plagioclase content observed in some samples appears to be spread throughout the three size classifications. Nucleation densities vary considerably from
6000 to 33 000 sites/mm2. Lava dome samples have higher nucleation densities than pumice. Crystal-area distributions are similar (Fig. 18), except that dome samples have considerably higher numbers of sub-20 mm2 crystals. All samples have a similar distribution of larger crystals, regardless of sample type.
|
|
Discussion of natural samples
Textural differences can be related to ascent history. Samples that experienced relatively slow ascent and possible stalling in the conduit or dome have less glass, more plagioclase, the presence of crystalline silica and a high abundance of small crystals. In contrast, rapidly ascended pumice and glassy dome samples, inferred to be associated with fast dome growth, contain more glass, less plagioclase and few small crystals. The crystal-area distributions (Fig. 18) show that samples that experienced more extensive crystallization contain a predominance of sub-20 µm2 crystals. These trends must be related to the kinetics of the system.
The dominance of crystal areas <20 µm2 suggests that nucleation was the dominant crystallization process, so that crystals formed in the magma chamber experienced little growth during ascent. Had there been extensive growth rather than nucleation, the number of large microlites would be expected to increase in the dome samples. Overgrowth rims in these samples were identified but their thickness was not more than 5 µm. Thus the increase in plagioclase fraction from rim growth is not significant.
|
COMPARISON OF EXPERIMENTS AND NATURAL SAMPLES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Interpretations from compositional and textural information
Table 5 outlines inferences about ascent conditions made from compositional and textural data. The presence of crystalline silica indicates relatively slow ascent rates, with stalling at shallow levels (<50 MPa), whereas its absence means either that ascent to the surface was sufficiently rapid to avoid nucleation of silica crystals, or that stagnation at water pressures >50 MPa was followed by rapid eruption.
|
In all natural samples the microlites are predominantly tabular in shape, with some hopper and rare skeletal crystals. By comparison with morphologies in the experiments, the magma could not have experienced undercoolings >60°C. However, pumice sample MVO1109d must have experienced rapid ascent and high
T such that there was insufficient time for crystallization. Most natural samples crystallized considerably more plagioclase and display higher nucleation densities than in the experiments. The natural samples may contain considerably more magma chamber formed plagioclase microlites compared with the experiments. It is also probable that the magma crystallized at pressures <30 MPa, resulting in greater crystallization and higher nucleation densities. For the crystalline dome samples, the overall decompression rate must have been very low (<0·2 MPa/h). There are no discernible differences between recently erupted and prehistoric samples, implying similar eruptive processes for the past 3·6 ka.
An intriguing issue is that of the plagioclase compositions. The least calcic plagioclase observed in the natural samples is An40. If this represents the lowest pressure at which magma equilibrated (35 MPa), material must have stalled at a minimum depth of 1·4 km, assuming a magmastatic pressure gradient. Although the crystallization of plagioclase microlites can raise the bulk temperature by latent heat release, heating of up to only 30°C is expected from 15 wt % crystallization of plagioclase (Couch et al., 2003), which is insufficient to increase anorthite contents to those observed if the final pressure was very low. Furthermore, the latent heat will be balanced by the cooling effect of water exsolution during ascent. Alternatively, over-pressurization can reach 10 MPa at shallow depths in the conduit (Melnik & Sparks 1999; Voight et al., 1999). Hence a PH2O of 25 MPa with an overpressure of 10 MPa has a magmastatic overburden of 15 MPa and consequently an inferred depth of 600 m rather than 1000 m. Therefore, plagioclase can crystallize with higher An contents than the depth would initially imply. Alternatively, the system may not have reached equilibrium, and therefore the anorthite content does not reflect the true decompression path of the magma. As our experimental study does not extend to very low pressures, the evolution of the plagioclase composition is not yet known at these conditions.
Finally, it should be noted that the rhyolitic starting composition used was based on an estimate of the matrix composition (glass and crystals <100 µm). Therefore it is possible that disparities between the experiments and the natural samples may be associated with compositional differences.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Four ascent profiles are inferred by comparison of natural samples (Couch et al., 2003) with experimental results.
- Rapid ascent. Magma travels from the chamber at 130 MPa to the surface sufficiently rapidly that insignificant crystallization occurs. This style of ascent is recorded in pumice, with no overgrowth rims, an absence of crystalline silica and a low plagioclase microlite fraction. Experimental results suggest that ascent needs to occur in 4 h or less to prevent any decompression-induced crystallization.
- Moderate ascent—stagnation—rapid ascent. Magma rises at a moderate rate, stalls within the conduit for hours to days and then is rapidly erupted at the surface, vesiculating slightly. This style of ascent is inferred for MVO37, where the compositions of the plagioclase (An45) suggest some moderate-pressure crystallization (80 MPa). However, the low overall plagioclase fraction indicates limited crystallization time and the vesicular nature is consistent with rapid final ascent.
- Moderate ascent—stagnation—moderate ascent. Material rises at a moderate rate, stalls in the conduit, permitting some crystallization, before a steady ascent is resumed, permitting degassing with limited further crystallization. MVO47 follows this ascent profile, as indicated by the moderate-pressure plagioclase compositions (An51) observed. Its low vesicularity suggests that there was sufficient time for efficient degassing. However, the absence of low-pressure composition plagioclase indicates that only limited crystallization occurred during the final ascent. The trace quantities of crystalline silica support incipient sub-50 MPa crystallization.
- Slow steady ascent—stagnation at shallow levels. Material rises steadily (<0·2 MPa/h integrated decompression rate) to a shallow level (<50 MPa), where it stalls for a considerable time, resulting in extensive microlite crystallization, including crystalline silica. MVO25, -34 and -1217 are samples inferred to have experienced this kind of ascent history. The low anorthite contents of the plagioclase microlites (An41) and the presence of quartz both indicate extensive crystallization at low pressures.
The absence of experiments at low pressures (<30 MPa) does limit the strength of these interpretations, as the behaviour of the system in terms of phase compositions is unknown. It is possible that the anorthite compositions in the natural samples were not in equilibrium at their final pressure. Therefore some samples may have spent significant periods of time at lower pressures.
Implications for eruptive style
Variations in extrusive activity can be related to magma viscosity. Sparks et al. (2000) estimated that the viscosity of the andesite leaving the magma chamber (30–35 wt % melt) is 7 x 106 Pa s, and that highly crystalline samples (<10% melt) have a viscosity of 1013–1014 Pa s. Groundmass crystallization, together with gas exsolution, causes the dramatic increase in viscosity.
The extensive monitoring of the eruption of the Soufrière Hills since 1995 has resulted in considerable advances in the understanding of how lava domes grow. Sparks et al. (1998) reported a range of growth rates between 0·2 and 10 m3/s. Watts et al. (2002) identified a range of dome structures, which can be correlated with extrusion rate. At the slowest discharge rates (0·4 m3/s) near-vertical spines form, and as extrusion rate increases, structures become more massive as megaspines and the shear lobes form (-3 m3/s). At high discharge rates (8 m3/s) pancake lobes are observed; although these still have a blocky surface, lateral spreading at the summit has been identified, implying a more fluid-like emplacement. Explosive activity is observed above 9 m3/s (Druitt et al., 2002).
The observed dome growth patterns correlate with textural characteristics. Here we assume that the conduit has a radius of 15 m and cross-sectional area of 700 m2, based on a variety of observations and theoretical modelling [see Melnik & Sparks (2003) for details]. With this assumed cross-section, the magma discharge rates equivalent to different decompression rates can be estimated to range from 35 MPa/h (equivalent to an extrusion rate 285 m3/s) for pumices to <0·2 MPa/h (equivalent to 1·6 m3/s) for highly crystalline dome samples using experimental constraints. Thus the range of observed discharge rates (Sparks et al., 1998; Druitt et al., 2002) is consistent with the range inferred from textural analysis of samples and comparison with experimental results.
Estimates of maximum discharge rate for pumice samples, based on dome volumes (Druitt et al., 2002), are up to 10 m3/s, which implies that magma took 4 days to reach the surface. The absence of amphibole reaction rims in the pumice samples led Devine et al. (1998b) to suggest a maximum decompression period of 1–2 days. Druitt et al. (2002) have estimated that higher extrusion rates may occur after major explosive eruptions, up to 20 m3/s. Even at these rates magma would take 2 days to ascend from depth to the surface, and we suggest that to prevent degassing-induced crystallization, ascent should have taken place in <4 h. This upper limit of 4 h is based on the predicted nucleation delay from the experimental results. A further clarification of the 4 h limit is that this estimate is based on a single rapid decompression event, whereas the magma would have ascended and decompressed continuously. The MDE experimental results, which approximate continuous ascent, show that crystallization must have occurred within the 8 h regardless of ascent rate.
One way of reconciling the faster ascent rate deduced here with the lower discharge rates estimated by Druitt et al. (2002) is to postulate that the conduit is narrower at depth than at the surface. Constraints on the conduit cross-section arise from surface or near-surface observations (Robertson et al., 1998; Melnik & Sparks, 2002), and together with models constrain only the exit conditions during dome extrusion. They do not preclude a narrower conduit or smaller cross-section at depth. To conserve mass flux, a narrower conduit at depth requires an increase in ascent velocity for the same exit discharge and in this case the time to ascend to shallow depths would be much shorter. In the explosive eruptions the magma that explodes originates at typical depths of 0·4–1·4 km (Druitt et al., 2002) so that in periods of explosive eruptions the ascent time may be much shorter even though the discharge rate remains at similar values. The discrepancy can be resolved if the cross-section at depths of 1 km or more is about 3–5 times less than at the surface. The narrowing of the conduit at depth may encourage stagnation, as material ascending rapidly from depth will experience a sudden decrease in ascent rate as the conduit widens towards the surface. Thus this could result in a sudden burst of crystallization, increasing the viscosity and stalling the ascent of the magma.
An important issue when considering the experimental results is whether the temperature of 875°C used is appropriate. A lower temperature could help to reconcile the problem of the pumice samples, as at lower temperatures the system would be more sluggish, resulting in a longer nucleation lag. The experiments by Hammer & Rutherford (2002) using dacites from Pinatubo, were run at 780°C. They found that in experiments with durations <27 h, groundmass textures could not be resolved because of limited crystal sites. This can be used as an upper limit on the nucleation delay of the dacite system. Thus if the Montserrat groundmass was at a temperature closer to 840°C (the lower limit of the estimated magma chamber temperatures, Barclay et al., 1998; Murphy et al., 2000), the nucleation delay could be increased. However, even if the lag period is increased to 12 h, a magma discharge of 75 m3/s is needed to prevent decompression-induced crystallization, assuming a constant cross-section of 700 m2. The exsolution of water during decompression will result in cooling of the magma (Zhang, 1999a); however, this is balanced by latent heat release during crystallization (Couch et al., 2003), hence temperatures below 840°C are not anticipated. A final consideration for sample MVO1109d is that the plagioclase is An>60, suggesting that the sample had been considerably hotter than 860°C shortly before eruption (Couch et al., 2003). There is evidence to suggest that mixing occurred of the andesite with mafic magma (Rutherford & Devine, 2003), which may change the composition and temperature of the andesite melt and hence affect the crystallization kinetics. However, even allowing for possible increases in the nucleation delay, a significant reduction in conduit cross-section at depth is required to prevent crystallization during ascent.
A further complication in the interpretation of the results is that the annealing conditions for the decompression experiments were at higher pressures (160 MPa) than that estimated for the magma chamber (130 MPa). Thus the experiments experienced larger undercoolings at a given Pf than the natural samples, as a greater amount of decompression had to taken place. The higher-pressure anneal conditions may also help to explain the disparity between the nucleation densities of the experiments and the natural samples. Experiments by Sato (1995) have shown that slightly reducing the temperature of the annealing stage results in considerably higher nucleation densities. Therefore it is likely that the high-pressure anneal stage of our experiments in this study has a similar effect in reducing the nucleation density of the plagioclase crystals.
|
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
An experimental study of a representative groundmass composition of andesite of the Soufrière Hills Volcano, Montserrat, has been carried out to investigate decompression-induced crystallization. A sequence of experiments, run at 875°C, was held at PH2O of 160 MPa for 16 h and then decompressed to a final pressure varying between 125 and 30 MPa. According to equilibrium experiments these decompressions induce an undercooling (
T) of 40–150°C for plagioclase. The decompression event was either instantaneous or took place over eight equal instalments. The single decompression experiments were dominated by the nucleation and growth of new plagioclase crystals, whose morphology varied systematically with increasing T, from tabular to hopper, swallowtail, dendritic and chain forms. Crystallization of the multiple decompression experiments was dominated by growth on existing crystal sites. A nucleation lag was identified of between 1 and 4 h, depending on T. The groundmass textures that form between leaving the magma chamber and being erupted are controlled by ascent rate in an open-system degassing scenario, based on experimental results. The variations in microlite fraction, composition and texture are due to differences in the ascent history. Dome lavas ascended slowly, possibly stalling for brief periods (hours to days) before stagnating at shallow depths for several days, resulting in extensive crystallization. Pumice samples ascended rapidly over time-scales of <4 h, inhibiting decompression-related crystallization. Prehistoric samples show trends similar to those for recent eruptive products, confirming a consistency in ascent processes for several thousand years. |
ACKNOWLEDGEMENTS |
|---|
Thanks are due to members of the Montserrat Volcano Observatory for collecting the samples used, to Oleg Melnik, Jon Blundy and Jenni Barclay for helpful discussions, and to Malcolm Rutherford and Bruno Scaillet for reviews that helped to clarify the manuscript. S.C. acknowledges an NERC studentship, R.S.J.S. an NERC professorship and Royal Society Wolfson Award, MRC, support from Gruppo Nazionale de Vulcanologia.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSEXPERIMENTAL RESULTSDISCUSSION OF EXPERIMENTAL...STUDIES OF NATURAL SAMPLESCOMPARISON OF EXPERIMENTS AND...DISCUSSIONSUMMARYREFERENCES |
|---|
Barclay, J., Rutherford, M. J., Carroll, M. R., Murphy, M. D., Devine, J. D., Gardner, J. E. & Sparks, R. S. J. (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]
Berkebile, C. A. & Dowty, E. (1982). Nucleation in laboratory charges of basaltic composition. American Mineralogist 67, 886–899.[Abstract]
Blundy, J. & Cashman, K. V. (2001). Magma ascent and crystallisation at Mount St Helens, 1980–1986. Contributions to Mineralogy and Petrology 140, 631–650.[Web of Science]
Burkhard, D. J. M. (2002). Kinetics of crystallisation: example of micro-crystallisation in basalt lava. Contributions to Mineralogy and Petrology 142, 724–737.[Web of Science]
Cashman, K. V. (1992). Groundmass crystallisation of Mount St Helens dacite, 1980–1986: a tool for interpreting shallow magmatic processes. Contributions to Mineralogy and Petrology 109, 431–449.[CrossRef][Web of Science]
Cashman, K. V. & Blundy, J. (2000). Degassing and crystallisation of ascending andesite and dacite. Philosophical Transactions of the Royal Society of London, Series A 358, 1487–1513.[CrossRef]
Cheng, H. C. & Lemlich, R. (1983). Errors in the measurement of bubble size distribution in foam. Industrial Engineering Chemical Fundamentals 22, 105–109.[CrossRef]
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]
Couch, S., Harford, C. L., Sparks, R. S. J. & Carroll, M. R. (2003). Experimental constraints on the conditions of formation of highly calcic plagioclase microlites at the Soufrière Hills Volcano, Montserrat. Journal of Petrology 44, 1455–1475.
Davis, M. J. & Ihinger, P. D. (1998). Heterogeneous crystal nucleation on bubbles in silicate melt. American Mineralogist 83, 1008–1015.[Abstract]
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., Barclay, J., Sparks, R. S. J., Carroll, M. R., Young, S. R. & Gardner, J. E. (1998a). Petrologic evidence for pre-eruptive pressure–temperature conditions, and recent reheating, of andesitic magma erupting at the Soufrière Hills volcano, Montserrat, WI. 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.
Druitt, T. H., Young, S. R., Baptie, B., Bonadonna, C., Calder, E. S., Clarke, A. B., Cole, P., Harford, C., Herd, R. A., Luckett, R., Ryan, G. & Voight, B. (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 the Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 281–306.
Dunbar, N. W., Jacobs, G. K. & Naney, M. T. (1995). Crystallisation processes in an artificial magma: variations in crystal shape, growth rate and composition with melt cooling history. Contributions to Mineralogy and Petrology 120, 412–425.[Web of Science]
Eichelberger, J. C., Carrigan, C. R., Westrich, H. R. & Price, R. H. (1986). Non-explosive silicic volcanism. Nature 323, 598–602.[CrossRef]
Fenn, P. M. (1974). Nucleation and growth of alkali feldspars from a melt. In: Mackenzie, W. S. & Zussman, J. (eds) The Feldspars. Manchester: Manchester University Press, pp. 360–361.
Fenn, P. M. (1977). The nucleation and growth of alkali feldspars from hydrous melts. Journal of the Mineralogical Association of Canada 15, 135–161.
Fokin, V. M., Potapov, O. V., Chinaglia, C. R. & Zanotto, E. D. (1999). The effect of pre-existing crystals on the crystallisation kinetics of a soda–lime–silica glass. The courtyard phenomenon. Journal of Non-Crystalline Solids 258, 180–186.[CrossRef]
Gardner, J. E., Hilton, M. & Carroll, M. R. (1999). Experimental constraints on degassing of magma: isothermal bubble growth during continuous decompression from high pressure. Earth and Planetary Science Letters 168, 201–218.[CrossRef][Web of Science]
Geschwind, C. H. & Rutherford, M. J. (1995). Crystallisation of microlites during magma ascent: the fluid mechanics of 1980–1986 at Mount St Helens. Bulletin of Volcanology 57, 356–370.[Web of Science]
Hammer, J. E. & Rutherford, M. J. (2002). An experimental study of the kinetics of decompression-induced crystallisation in silicic melt. Journal of Geophysical Research 107, 148–227.
Hammer, J. E., Cashman, K. V., Hoblitt, R. P. & Newman, S. (1999). Degassing and microlite crystallisation during pre-climatic events of the 1991 eruption of Mt. Pinatubo, Philippines. Bulletin of Volcanology 60, 355–380.[CrossRef][Web of Science]
Hammer, J. E., Cashman, K. V. & Voight, B. (2000). Magmatic processes revealed by textural and compositional trends in Merapi dome lavas. Journal of Volcanology and Geothermal Research 100, 165–192.[CrossRef][Web of Science]
Holtz, F., Pichavant, M., Barbey, P. & Johannes, W. (1992). Effects of H2O on liquidus phase relations in the haplogranite system at 2 and 5 kbar. American Mineralogist 77, 1223–1241.[Abstract]
Kellerhals, R., Shaw, J. & Arora, V. K. (1975). On grain size from thin sections. Journal of Geology 83, 79–96.[Web of Science]
Kirkpatrick, R. J. (1981). Kinetics of crystallisation in igneous rocks. In: Lasaga, A. C. & Kirkpatrick, R. J. (eds) Kinetics of Geochemical Processes. Mineralogical Society of America, Reviews in Mineralogy 8, 321–397.
Lejeune, A.-M. & Richet, P. (1995). Rheology of crystal-bearing silicate melts: an experimental study at high viscosities. Journal of Geophysical Research 100, 4215–4229.[CrossRef]
Lofgren, G. E. (1974). An experimental study of plagioclase crystal morphology: isothermal crystallisation. American Journal of Science 274, 243–273.[Abstract]
Lofgren, G. E. (1977). Dynamic crystallisation experiments bearing on the origin of textures in impact-generated liquids. Proceedings of the Eighth Lunar Science Conference, Geochimica et Cosmochimica Acta Supplement, 2079–2095.
Lofgren, G. E. (1980). Experimental studies on the dynamics crystallisation of silicate melts. In: Hargraves, R. B. (ed.) Physics of Magmatic Processes. Princeton, NJ: Princeton University Press, pp. 487–551.
Mangan, M. T. & Sisson, T. (2000). Delayed, disequilibrium degassing in rhyolite magma: decompression experiments and implications for explosive volcanism. Earth and Planetary Science Letters 183, 441–455.[CrossRef][Web of Science]
Mastin, L. G. & Ghiorso, M. S. (2001). Adiabatic temperature changes of magma–gas mixtures during ascent and eruption. Contributions to Mineralogy and Petrology 141, 307–321.[Web of Science]
McCoy, T. J. & Lofgren, G. E. (1999). Crystallisation of the Zagami shergottite: an experimental study. Earth and Planetary Science Letters 173, 397–411.[CrossRef][Web of Science]
Melnik, O. & Sparks, R. S. J. (1999). Nonlinear dynamics of lava dome extrusion. Nature 402, 37–41.[CrossRef]
Melnik, O. & Sparks, R. S. J. (2002). Dynamics of magma ascent and lava extrusion at Soufrière Hills Volcano, Montserrat. In: Druitt, T. H. & Kokelaar, B. P. (eds) The Eruption of the Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 153–172.
Muncill, G. E. & Lasaga, A. C. (1987). Crystal-growth kinetics of plagioclase in igneous systems: one-atmosphere experiments and application of a simplified growth model. American Mineralogist 72, 299–311.[Abstract]
Muncill, G. E. & Lasaga, A. C. (1988). Crystal-growth kinetics of plagioclase in igneous systems: isothermal H2O-saturated experiments and extension of a growth model to complex silicate melts. American Mineralogist 73, 982–992.[Abstract]
Murphy, M. D., Sparks, R. S. J., Barclay, J., Carroll, M. R. & Brewer, T. S. (2000). Remobilisation of andesite magma by intrusion of mafic magma at the Soufrière Hills volcano, Montserrat, West Indies. Journal of Petrology 41, 21–42.
Nakada, S. & Motomura, Y. (1999). Petrology of the 1991–1995 eruption at Unzen: effusion pulsation and groundmass crystallisation. Journal of Volcanology and Geothermal Research 89, 173–196.[CrossRef][Web of Science]
Navon, O. & Lyakhovsky, V. (1998). Vesiculation processes in silicic magmas. In: Gilbert, J. S. & Sparks, R. S. J. (eds) The Physics of Explosive Volcanic Eruptions. Geological Society, London, Special Publications 145, 27–50.
Robertson, R. E., Cole, P., Sparks, R. S. J., Harford, C., Lejeune, A.-M., McGuire, W. J., Miller, A. D., Murphy, M. D., Norton, G. E., Stevens, N. S. & Young, S. R. (1998). The explosive eruption of Soufrière Hills volcano, Montserrat. Geophysical Research Letters 25, 3689–3692.
Rutherford, M. J. & Devine, J. D. (2003). Magmatic conditions and magma ascent as indicated by hornblende phase equilibria and reactions in the 1995–2002 Soufrière Hills magma. Journal of Petrology 44, 1433–1453.
Sahagian, D. L. & Proussevitch, A. A. (1996). Thermal effects of magma degassing. Journal of Volcanology and Geothermal Research 74, 19–38.[CrossRef][Web of Science]
Sato, H. (1995). Textural difference between pahoehoe and aa lavas of Izu–Oshima Volcano, Japan—an experimental study on population-density of plagioclase. Journal of Volcanology and Geothermal Research 66, 101–113.[CrossRef][Web of Science]
Schmelzer, J., Pascova, R. & Moller, J. (1993). Surface-induced devitrification of glasses—the influence of elastic strains. Journal of Non-Crystalline Solids 162, 26–39.[CrossRef]
Sparks, R. S. J., Young, S. R., Barclay, J., Calder, E. S., Cole, P., Darroux, B., 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]
Sparks, R. S. J., Murphy, M. D., Lejeune, A.-M., Watts, R., Barclay, J. & Young, S. R. (2000). Control on the emplacement of the andesite lava dome of the Soufrière Hills volcano by degassing-induced crystallisation. Terra Nova 12, 14–20.[CrossRef][Web of Science]
Swanson, S. E. (1977). Relation of nucleation and crystal-growth rate to the development of granitic textures. American Mineralogist 62, 966–978.[Abstract]
Tuttle, O. F. & Bowen, N. L. (1958). Origin of Granite in the Light of Experimental Studies in the System NaAlSi3O8–KAlSi3O8–SiO2–H2O. Geological Society of America Memoir 74.
Voight, B., Sparks, R. S. J., Miller, A. D., Stewart, R. C., Hoblitt, R. P., Clarke, A. B., et al. (1999). Magma flow instability and cyclic activity at Soufrière Hills Volcano, Montserrat, British West Indies. Science 283, 1138–1142.
Watts, R. B., Herd, R. A., Sparks, R. S. J. & Young, S. R. (2002). Growth patterns and emplacement of the andesitic lava dome at the Soufrière Hills volcano, Montserrat: July 1995 to March 1998. In: Druitt, T. H. & Kokelaar, B. P. (eds) The Eruption of the Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 115–152.
Wiebel, E. R. (1980). Stereological Methods, Volume 2. Theoretical Foundations. New York: Academic Press.
Wylie, J. J., Voight, B. & Whitehead, J. A. (1999). Instability of magma flow from volatile-dependent viscosity. Science 285, 1883–1885.
Zhang, Y. (1999a). Exsolution enthalpy of water from silicate liquids. Journal of Volcanology and Geothermal Research 88, 201–207.[CrossRef][Web of Science]
Zhang, Y. (1999b). H2O in rhyolitic glasses and melts: measurement, speciation, solubility and diffusion. Reviews of Geophysics 37, 493–516.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Ohba, K. Matsuoka, Y. Kimura, H. Ishikawa, and H. Fujimaki Deep Crystallization Differentiation of Arc Tholeiite Basalt Magmas from Northern Honshu Arc, Japan J. Petrology, June 1, 2009; 50(6): 1025 - 1046. [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]
|
|
|
|
 |
|
 J. E. Hammer Experimental Studies of the Kinetics and Energetics of Magma Crystallization Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 9 - 59. [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]
|
|
|
|
 |
|
 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]
|
|
|
|
|
|
M.-N. Guilbaud, S. Blake, T. Thordarson, and S. Self Role of Syn-eruptive Cooling and Degassing on Textures of Lavas from the AD 1783-1784 Laki Eruption, South Iceland J. Petrology, July 1, 2007; 48(7): 1265 - 1294. [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]
|
|
|
|
|
|
J. Blundy and K. Cashman Rapid decompression-driven crystallization recorded by melt inclusions from Mount St. Helens volcano Geology, October 1, 2005; 33(10): 793 - 796. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Couch and S. Couch Experimental investigation of crystallization kinetics in a haplogranite system American Mineralogist, October 1, 2003; 88(10): 1471 - 1485. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. COUCH, C. L. HARFORD, R. S. J. SPARKS, and M. R. CARROLL Experimental Constraints on the Conditions of Formation of Highly Calcic Plagioclase Microlites at the Soufrire Hills Volcano, Montserrat J. Petrology, August 1, 2003; 44(8): 1455 - 1475. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||