Michael R. Carroll

Mineral disequilibrium in lavas explained by convective self-mixing in open magma chambers

Characteristic features of many porphyritic andesite and dacite lavas are that they are rich in crystals and display a range of disequilibrium features, including reversely zoned crystals, resorption surfaces, wide ranges of mineral compositions and minerals which are not in equilibrium with the surrounding rock matrix. These features are often interpreted as evidence of the mixing of magmas of contrasting composition, temperature and origin. Here, however, we propose that such features can also be caused by convection within a magma body with a single composition, that is heated from below and cooled from above. We describe petrological observations of andesite lava erupted at the Soufrière Hills volcano, Montserrat, which indicate a heating event and the intermingling of crystals that have very different thermal histories. We present experimental data on a representative groundmass composition of this lava, which indicate that it is difficult to explain the calcic compositions of plagioclase overgrowth rims and microphenocrysts unless parts of the magma were at temperatures much higher than the inferred average temperature. The concept of convective self-mixing allows us to explain the occurrence of compositions of minerals that apparently cannot coexist under equilibrium conditions.

Experimental phase equilibria constraints on pre-eruptive storage conditions of the Soufriere Hills magma

New experimental results are used to constrain the P. T, X(H 2 O) conditions of the Soufriere Hills magma prior to ascent and eruption. The experiments were performed on a powdered andesite erupted in January, 1996, at an fO 2 corresponding to ∼NNO+1 with P H2 O and temperatures in the range 50 to 200 MPa and 800 to 940°C. Amphibole is stable at P H2 O >115 MPa and temperatures 72 wt% SiO 2 in residual melt) at P H2 O >115 MPa. Analyses of rhyolitic glass inclusions in quartz and plagioclase from recently erupted samples indicate melt water contents of 4.27±0.54 wt% H 2 O and CO 2 contents <60 ppm. The evolved Soufriere Hills magma would therefore be H 2 O-saturated at pressures <130 MPa. These results suggest that the Soufriere Hills magma containing the stable assemblage amphibole, quartz, plagioclase, orthopyroxene, magnetite and ilmenite was stored at P H2 O of 115-130 MPa, equivalent to a minimum depth for a water-saturated magma chamber of 5-6 km depth. Magma temperatures were initially low (820-840°C). Quartz is believed to have been destabilised by a heating event involving injection of new basaltic magma. The stability field of hornblende provides a useful upper limit (∼880°C) for the extent of this reheating.

Experimental constraints on degassing of magma: isothermal bubble growth during continuous decompression from high pressure

Numerical models predict that rapid ascent of hydrous magma can lead to supersaturation of dissolved volatile constituents, possibly leading to explosive eruption. We have performed controlled decompression experiments to investigate the ascent rates required to maintain bubble–melt equilibrium. High-silica rhyolitic melts were saturated with water at 200 MPa and 825°C, decompressed to lower pressures at constant rates of 0.025, 0.25, 0.5, and 1.0 MPa s−1, and then rapidly quenched isobarically. Other samples were saturated with water over the pressure range investigated to determine equilibrium water solubility in order to quantify degassing efficiency during decompression. At a decompression rate of 0.025 MPa s−1, melt–vapor equilibrium was maintained over the entire pressure range examined: 200 to 17.5 MPa. A single bubble nucleation event occurred in response to decompression, and quenched bubble sizes can be modeled by a equilibrium bubble growth model that takes into account the number density of bubbles. At decompression rates of 0.25, 0.5, and 1.0 MPa s−1, rhyolitic melts could not degas in equilibrium when pressure decreased from 200 MPa to 140 MPa, and water supersaturation (ΔP) in the melt reached up to 60 MPa, with higher values at faster decompression rates. Further pressure release resulted in near equilibrium degassing and ΔP dropped significantly. In each case, ΔP decreased when bubbles exceeded 10 vol.%. A single, heterogeneous bubble nucleation event occurred in each experiment when ΔP<20 MPa; no other bubbles nucleated despite ΔP reaching 60 MPa, which is probably too low to trigger homogeneous nucleation. Compared to estimates for magma decompression rates during lava dome eruptions, our results indicate that magmas can degas efficiently throughout their ascent to the surface. In explosive eruptions, decompression rates may exceed those of this study and hence melts may become supersaturated with water. Such fast decompressions are expected, however, only when magma is highly vesicular, which would aid approach to equilibrium degassing.

The role of magma mixing in triggering the current eruption at the Soufriere Hills Volcano, Montserrat, West Indies

The andesite lava currently erupting at the Soufriere Hills volcano, Montserrat, contains ubiquitous mafic inclusions which show evidence of having been molten when incorporated into the andesite. The andesite phenocrysts have a range of textures and zonation patterns which suggest that non-uniform reheating of the magma occurred shortly before the current eruption. Reheating resulted in remobilisation of the resident magma and may have induced eruption.

Solubilities of carbon dioxide and water in rhyolitic melt at 850°C and 750 bars

Concentrations of carbon dioxide and water dissolved in glasses quenched from rhyolitic melts equilibrated with H_2O-CO_2 fluids at 850°C and 750 bar were measured using infrared spectroscopy; concentrations of H_2O and CO_2 in the quenched fluids were measured manometrically. The mole fraction of CO_2 in the quenched fluid ranged from 0.06 to 0.91. Concentrations of CO_2 in the coexisting rhyolitic melt increased from 23(±6) ppm for the sample equilibrated with the most CO_2-poor fluid to 515(±16) ppm for that equilibrated with the most CO_2-rich fluid. The water content of the melt varied from 0.51(±0.06) to 3.34(±0.08) wt%. Our results show that concentrations of molecular CO_2 and H_2O in the glasses obey Henrys Law; i.e., the mole fractions of molecular CO_2 and molecular H_2O in the quenched melts are proportional to their fugacities in the coexisting vapor. CO_2 contents of vapor-saturated melts are not enhanced by addition of water to CO_2-rich vapor, contrary to previous reports for silicate melts at higher pressures. The Henrian behavior of CO_2 and H_2O at low pressure considerably simplifies modeling of the degassing of silicic magmas.

Is the Iceland hot spot also wet? Evidence from the water contents of undegassed submarine and subglacial pillow basalts

Abstract Water contents have been measured in basaltic glasses from submarine and subglacial eruption sites along the Reykjanes Ridge and Iceland, respectively, in order to evaluate the hypothesis of Schilling et al. [Phil. Trans. R. Soc. London A 56 (1980) 147–178] that hot spots are also wet spots. Having erupted under pressure the water contents measured in these samples are potentially unaffected by degassing. After correcting these water contents for the effects of crystallisation (to give H2O(8) values) they indicate that the concentration of water in the source regions increases from 165 ppm at the southern end of the Reykjanes Ridge to between 620 and 920 ppm beneath Iceland. This suggests that Iceland is a wet spot and the H2O(8) values indicate that its influence on basalt compositions increases northwards along the Reykjanes Ridge from ∼61°N (650 km from the plume centre) towards Iceland. The existence of wetter Icelandic source regions have important implications for mantle melting, as enrichments of this magnitude depress the mantle solidus, increasing the degree of melting at a given temperature. Therefore the enhanced rates of volcanism on Iceland may be a result of wetter sources in addition to a thermal anomaly beneath Iceland.

Noble gas solubilities in silicate melts and glasses: New experimental results for argon and the relationship between solubility and ionic porosity

New measurements of the solubility of Ar in basaltic, rhyolitic, orthoclasic, and albitic melts and glasses ar Ar pressures of 250–10,000 bar and temperatures of 400–1300°C are presented and combined with other solubility measurements for a wider range of melt compositions to parameterize the effects of pressure, temperature, and melt composition on Ar solubility. Argon solubility in melts and glasses is roughly linear with Ar pressure under these conditions. At near-liquidus temperatures, solubility in melts is approximately independent (within ∼ 10%) of temperature, while some results below 600–700°C show an increase in solubility with decreasing temperature, perhaps reflecting differences in the nature of Ar solubility in glasses and melts. There is also a positive, linear correlation between the “ionic porosity” of melt and the logarithm of Ar solubility. This correlation is better than previously noted correlations between inert gas solubility and melt density and volume, and provides a useful means of predicting how Ar solubility varies with melt composition. The solubilities of He, Ne, Kr, and Xe are also positively correlated with ionic porosity, but are increasingly sensitive to ionic porosity as the size of the gas atom increases, suggesting that with more efficient packing of the melt structure the availability of sites that can incorporate inert gas atoms decreases more rapidly for larger atoms than for smaller atoms. Comparison of inert gas solubilities with those of molecular CO_2 and molecular H_2O in rhyolitic melts shows that solubilities decrease in the order H_2O_(mol) ⪢ He > Ne > Ar > CO_(2,mol) ≈ Kr > Xe. The much higher solubility of molecular H_2O compared to the other neutral gas species (and molecular CO_2) suggests that it is not merely passively occupying interstitial “holes” in the melt structure as is thought to be the case for the rare gases (and likely for molecular CO_2), but rather it is stabilized in the melt structure by chemical bonds (e.g., by hydrating cations or through hydrogen bonds).

Petrologic evidence for pre‐eruptive pressure‐temperature conditions, and recent reheating, of andesitic magma erupting at the Soufriere Hills Volcano, Montserrat, W.I.

The recent eruption of the Soufriere Hills Volcano in Montserrat (July, 1995, to present; September, 1997) has produced an andesitic dome (SiO2 ∼ 59–61 wt.%). The eruption has been caused by invasion of mafic magma into a preexisting andesitic magma storage region (P ∼ 130 MPa; ≥5 km depth). The composition of the andesite has remained essentially constant throughout the eruption, but heating by the mafic magma increased the andesite temperature from ≤830°C to ≤880°C. Prior to being heated, the stable mineral assemblage in the andesite was plagioclase + amphibole + orthopyroxene + titanomagnetite + ilmenite + quartz. The rise in temperature from ≤830°C to ≤880°C (fO2 ∼ 1 log unit above NNO) has caused quartz to become unstable, and has also caused changes in silicate and Fe-Ti oxide mineral compositions. The andesitic magma is likely saturated with an H2O-rich vapor phase in the upper part of the magma storage region. Melt H2O content is ∼4.7 wt.%.

The solubility of H2O in phonolitic melts

Abstract We have calibrated the IR spectroscopic technique for measurement of H2O dissolved in phonolitic glasses as hydroxyl and ELO molecules using manometric and weight-loss methods. The resulting molar absorptivity coefficients are 1.25+0.33-0.22 (for absorbance due to OH- at 4500 cm-1) and 1.10+0.12-0.10 (for absorbance due to molecular H2O at 5200 cm-1). These values are similar to those previously determined for hydrous jadeitic glasses. We have applied our calibration to a new set of solubility experiments in which H2O and a natural phonolitic glass were equilibrated at near-liquidus temperatures (85-973 °C) and pressures of 191-1500 bars for periods of 38-272 h. We used a regular solution model to develop an equation of state for the solubility of H2O in phonolitic melts. Our experimental results demonstrate that H2O solubility is appreciably higher in phonolitic melts compared with basaltic and rhyolitic melts at the same pressures and near-liquidus temperatures; e.g.. the solubility of H2O at 1000 bars is 4.9 wt% in phonolitic melt (850 °C). 4 wt% in rhyolitic (850 °C). and 3.2 wt% in basaltic (1200 °C) melts. The calculated partial molar volume of dissolved H2O in phonolitic melt (8.5 ± 2.5 cm3/mol) falls between that determined by similar methods for rhyolitic and basaltic melts, but we note that the significance of this number is unknown because speciation changes during quenching are not sufficiently well characterized.

The OH content of pyrope at high pressure

The OH contents of pyrope garnets synthesised at 1000°C and pressures between 2 and 13 GPa have been measured by infrared spectroscopy. We find that under the same conditions of pressure, temperature, aH2O and aSiO2 the OH content of pyrope is similar to that of grossular. This shows that observed differences in nature between water contents of pyrope and grossular are related to paragenesis rather than reflecting higher intrinsic solubility in grossular. In the presence of excess SiO2 the H2O content increases with pressure to 5 GPa (1000 ppm) then decreases to below the detection limit at pressures above 7 GPa. Thus garnet becomes more hydrous with pressure to some critical value, beyond which dehydration occurs even under H2O-saturated conditions. This observation is consistent with the measured partial molar volume of water in hydrogarnet which becomes greater than that in fluid water within the pressure range of this study. When corrected to upper mantle conditions we find that pyrope should dehydrate with increasing depth below about 250 km. Pyrope is unlikely therefore to be a major site of water storage in the transition zone and upper mantle.