Microlites and "nanolites" in rhyolitic glass: microstructural and chemical characterization

 We used transmission electron microscopy (TEM) and scanning electron microscopy (SEM) to study magmatic crystals in the Ben Lomond rhyolite lava dome, Taupo Volcanic Center, New Zealand. Using TEM and SEM to investigate the size distributions of these crystals, we identified three size populations: microphenocrysts (>1.2 μm wide), microlites (>0.6 μm wide), and smaller crystals (<0.6 μm wide) which we term "nanolites". The predominant mineral phases of the microlites and nanolites are augites, pigeonites, and hypersthenes. The compositions and microstructures within these pyroxenes indicate disequilibrium crystallization at approximately 850–900  °C and undercoolings as high as 300  °C from equilibrium crystallization temperatures. Complex microstructures resulting from subsolidus reactions in augite and pigeonite are consistent with moderate cooling rates within the upper obsidian layer of the Ben Lomond rhyolite dome. This study demonstrates the existence of sub-micron magmatic crystals in a rhyolite and illustrates the potential of TEM to provide unique information about the crystallization and cooling histories of glassy volcanic rocks. Received: May 8, 1995 / Accepted: November 27, 1995     

Compressibility of titanosilicate melts

The effect of composition on the relaxed adiabatic bulk modulus (K₀) of a range of alkali- and alkaline earth-titanosilicate [X ₂ ^n/n+ TiSiO₅ (X=Li, Na, K, Rb, Cs, Ca, Sr, Ba)] melts has been investigated. The relaxed bulk moduli of these melts have been measured using ultrasonic interferometric methods at frequencies of 3, 5 and 7 MHz in the temperature range of 950 to 1600°C (0.02 Pa s < _s < 5 Pa s). The bulk moduli of these melts decrease with increasing cation size from Li to Cs and Ca to Ba, and with increasing temperature. The bulk moduli of the Li-, Na-, Ca- and Ba-bearing metasilicate melts decrease with the addition of both TiO₂ and SiO₂ whereas those of the K-, Rb- and Cs-bearing melts increase. Linear fits to the bulk modulus versus volume fraction of TiO₂ do not converge to a common compressibility of the TiO₂ component, indicating that the structural role of TiO₂ in these melts is dependent on the identity of the cation. This proposition is supported by a number of other property data for these and related melt compositions including heat capacity and density, as well as structural inferences from X-ray absorption spectroscopy (XANES). The compositional dependence of the compressibility of the TiO₂ component in these melts explains the difficulty incurred in previous attempts to incorporate TiO₂ in calculation schemes for melt compressibility. The empirical relationship KV^-4/3 for isostructural materials has been used to evaluate the compressibility-related structural changes occurring in these melts. The alkali metasilicate and disilicate melts are isostructural, independent of the cation. The addition of Ti to the metasilicate composition (i.e. X₂TiSiO₅), however, results in a series of melts which are not isostructural. The alkaline-earth metasilicate and disilicate compositions are not isostructural, but the addition of Ti to the metasilicate compositions (i.e. XTiSiO₅) would appear, on the basis of modulus-volume systematics, to result in the melts becoming isostructural with respect to compressibility.     

The origin of reaction textures in mantle peridotite xenoliths from Sal Island, Cape Verde: the case for “metasomatism” by the host lava

Reaction zones around minerals in mantle xenoliths have been reported from many localities worldwide. Interpretations of the origins of these textures fall into two groups: mantle metasomatic reaction or reaction during transport of the xenoliths to the surface. A suite of harzburgitic mantle xenoliths from Sal, Cape Verde show clear evidence of reaction during transport. The reactions resulted in the formation of olivine–clinopyroxene and Si- and alkali-rich glass reaction zones around orthopyroxene and sieve-textured clinopyroxene and sieve textured spinel, both of which are associated with a Si- and alkali-rich glass similar to that in the orthopyroxene reaction zones. Reaction occurred at pressures less than the mantle equilibration pressure and at temperatures close to the liquidus temperature of the host magma. In addition, there is a clear spatial relation of reaction with the host lava: reaction is most intense near the lava/xenolith contact. The residence time of the xenoliths in the host magma, determined from Fe–Mg interdiffusion profiles in olivine, was approximately 4 years. Our results cannot be reconciled with a recent model for the evolution of the mantle below the Cape Verde Archipelago involving mantle metasomatism by kimberlitic melt. We contend that alkali-rich glasses in the Sal xenoliths are not remnants of a kimberlitic melt, but rather they are the result of reaction between the host lava or a similar magma and xenolith minerals, in particular orthopyroxene. The formation of a Si- and alkali-rich glass by host magma–orthopyroxene reaction appears to be a necessary precursor to formation of sieve textured spinel and clinopyroxene.     

Thermal expansivities of supercooled haplobasaltic liquids

The thermal expansion of supercooled liquids in the haplobasaltic (anorthite-diopside) system have been determined via methods of container-based dilatometry. The expansivity data obtained in this study agree well with estimates provided by previous dilatometric determinations in the system that have relied on alternative experimental strategies. The data have been combined with high-temperature, superliquidus determinations of melt density to obtain expressions for the volume-temperature (V-T) relationships of liquids in the anorthite-diopside system. The V-T data clearly indicate a nonlinear temperature dependence of volume for all melts investigated. The variation is most striking for diopside, where the coefficient of volume thermal expansion decreases ∼56% from temperatures near the glass transition to superliquidus temperatures. With increasing anorthite content, the degree of variation appears to decrease. An₄₂Di₅₈ exhibits a decrease of 39% of its coefficient of thermal expansion and An₉₈Di₀₂ of 33%, respectively. The expansivities obtained in this study cannot be reproduced by means of published models that are based on linear V-T relationships. They require instead a reanalysis of existing pressure-V-T equations of state models for silicate melts.     

Viscosity of hydrous Etna basalt: implications for Plinian-style basaltic eruptions

Water dissolved in a silicate melt can strongly influence its physical properties and thus magma behavior during crystallization, degassing, foaming and fragmentation. Etna is a basaltic volcano whose activity is dominated by effusive eruptions which have long represented a threat to the densely populated, surrounding area. Recently, recognition of the products of a Plinian eruption (122 B.C.) has raised further issues for hazard assessment at Etna and other basaltic volcanoes. Constraining the behavior of Etna magma under conditions relevant to both effusive and explosive hazards requires viscosity data under conditions near the glass transition. Here we have investigated the viscosity of hydrous Etna lava in order to better understand eruptive processes which characterize this volcano. The experimental methods which have been used include piston cylinder synthesis of the hydrated melts, micropenetration viscometry for low-temperature viscosity measurements, and near-infrared spectroscopy for the evaluation of sample homogeneity and measurements of water content. Additionally, scanning calorimetric determinations were performed to check whether incipient crystallization had occurred. Sample compositions were determined using electron microprobe analysis and ⁵⁷Fe Mössbauer spectroscopy. Results from this study are compared with previous reports of trachytic, phonolitic and model calc-alkaline rhyolite (HPG8) compositions. The viscosity of the basaltic melt (dry and wet) has been parameterized in terms of temperature and water content via the non-Arrhenian equation: log₁₀-=-4.643+(5,812.44-427.042H₂O)/(T(K)-499.31+28.742ln(H₂O)) where - is the viscosity in Pa s, H₂O is the water content in wt%, and T is the temperature in Kelvin. We observe that the viscosity of alkali basalt (at more than 0.5 wt% H₂O) is similar to that of an alkaline trachyte (Agnano-Monte Spina eruption, Phlegrean Fields) and much higher than that of a peralkaline phonolite (Teide, Tenerife) at similar silica contents and NBO/T. For water contents above 1.5 wt%, the viscosity of the basalt is similar to that of rhyolitic melts with similar water contents. At temperatures ranging from 1,050 to 1,150 °C and with water contents between 0.5 and 2.3 wt% (eruptive conditions), the viscosities calculated using the equation defined in this study are (1) in reasonable agreement with those calculated using Shaw's model, and (2) much lower than those experimentally determined in a previous study. However, outside these temperature and water content ranges, the agreement with Shaw's model (1972) breaks down.     

Effects of F,B2O3 and P2O5 on the solubility of water in haplogranite melts compared to natural silicate melts

The effects of F, B₂O₃ and P₂O₅ on the H₂O solubility in a haplogranite liquid (36 wt. % SiO₂, 39 wt. % NaAlSi₃O₈, 25 wt. % KAlSi₃O₈) have been determined at 0.5, 1, 2, and 3 kb and 800, 850, and 900°C. The H₂O solubility increases with increasing F and B content of the melt. The H₂O solubility increase in more important at high pressure (2 and 3 kb) than at low pressure (0.5 kb). At 2 kb and 800°C, the H₂O solubility increases from 5.94 to 8.22 wt. % H₂O with increasing F content in the melt from 0 to 4.55 wt. %, corresponding to a linear H₂O solubility increase of 0.53 mol H₂O/mol F. With addition of 4.35 wt. % B₂O₃, the H₂O solubility increases up to 6.86 wt. % H₂O at 2 kb and 800°C, corresponding to a linear increase of 1.05 mol H₂O/mol B₂O₃. The results allow to define the individual effects of fluorine and boron on H₂O solubility in haplogranitic melts with compositions close to that of H₂O-saturated thermal minima (at 0.5–3 kb). Although P has a dramatic effect on the phase relations in the haplogranite system, its effect on the H₂O solubility was found to be negligible in natural melt compositions. The concominant increase in H₂O solubility and F can not be interpreted on the basis of the available spectroscopic data (existence of hydrated aluminofluoride complexes or not). In contrast, hydrated borates or more probably boroxol complexes have been demonstrated in B-bearing hydrous melts.     

Rhyolite magma degassing: an experimental study of melt vesiculation

 The vesiculation of a peralkaline rhyolite melt (initially containing ∼0.14 wt.% H₂O) has been investigated at temperatures above the rheological glass transition (T _g≈530  °C) by (a) in situ optical observation of individual bubble growth or dissolution and (b) dilatometric measurements of the volume expansion due to vesiculation. The activation energy of the timescale for bubble growth equals the activation energy of viscous flow at relatively low temperatures (650–790  °C), but decreases and tends towards the value for water diffusion at high temperatures (790–925  °C). The time dependence of volume expansion follows the Avrami equation ΔV (t)∼{1–exp [–(t/τ_av) ⁿ ]} with the exponent n=2–2.5. The induction time of nucleation and the characteristic timescale (τ_av) in the Avrami equation have the same activation energy, again equal to the activation energy of viscous flow, which means that in viscous melts (Peclet number <1) the vesiculation (volume expansion), the bubble growth process, and, possibly, the nucleation of vesicles, are controlled by the relaxation of viscous stresses. One of the potential volcanological consequences of such behavior is the existence of a significant time lag between the attainment of a super-saturated state in volatile-bearing rhyolitic magmas and the onset of their expansion. Received: March 20, 1995 / Accepted: October 24, 1995     

Measurement and implication of "effective" viscosity for rhyolite flow emplacement

Hazardous explosive activity may sporadically accompany the extrusion of silicic lava domes. Modelling of the emplacement of silicic domes is therefore an important task for volcanic hazard assessment. Such modelling has been hampered by a lack of a sufficiently accurate rheological database for silicic lavas with crystals and vesicles. In the present study, the parallel-plate viscometry method was applied to determine the shear viscosity of five natural rhyolitic samples from a vertical section through the Ben Lomond lava dome, Taupo Volcanic Centre, New Zealand. Rheological measurements were performed at volcanologically relevant temperatures (780-950°C) and strain rates (10^-5-10^-7 s^-1). Although these samples are in the metastable state, viscosity determinations, melt composition, as well as water and crystal contents of samples were demonstrably stable during experiments. For samples containing up to 5 vol.% microlites, the composition of the melt, rather than the physical effect of suspended crystals, had greater influence on the effective viscosity of the silicic magma. Samples with 10 vol.% microlites and containing a flow banding defined by microlites show no significant orientational effects on apparent viscosity. The rheological measurements were used together with a simple cooling model to construct thermal and viscosity profiles revealing conditions during the emplacement of the Ben Lomond lava dome.     

The onset of non-Newtonian rheology of silicate melts

The viscoelastic behavior of silicate melts has been measured for a range of compositions (NaAlSi₃O₈, NaCaAlSi₂O₇, CaMgSi₂O₆, Li₂Si₄O₉, Na₂Si₄O₉, K₂Si₄O₉, Na₂Si₃O₇, K₂Si₃O₇ and Na₂Si₂O₅) using the fiber elongation method. A1l compositions exhibit Newtonian behavior at low strain-rates, but non-Newtonian behavior at higher strain-rates, with strain-rate increasing faster than the applied stress. The decrease in shear viscosity observed at the high strain-rates ranges from 0.3 to 1.6 log₁₀ units (Pa s). The relaxation strain-rates, _relax, of these melts have been estimated from the low strain-rate, Newtonian, shear viscosity, using the Maxwell relationship; _relax= ^–1=(_s/G)^–1. For all compositions investigated, the onset of non-Newtonian rheology is observed at strain-rates 2.5+0.5 orders of magnitude less than the calculated relaxation strain-rate. This difference between the non-Newtonian onset and the relaxation strain-rate is larger than that predicted by the single relaxation time Maxwell model. Normalization of the experimental strain-rates to the relaxation strain-rate predicted from the Maxwell relation, eliminates the composition. and temperature-dependence of the onset of non-Newtonian behavior. The distribution of relaxation in the viscoelastic region appears to be unrelated to melt chemistry. This conclusion is consistent with the torsional, frequency domain study of Mills (1974) which illustrated a composition-invariance of the distribution of the imaginary component of the shear modulus in melts on the Na₂O-SiO₂ join. The present, time domain study of viscoelasticity contrasts with frequency domain studies in terms of the absolute strains employed. The present study employs relatively large total strains (up to 2). This compares with typical strains of 10^–8 in ultrasonic (frequency domain) studies. The stresses used to achieve the strain-rates required to observe viscoelastic behavior in this study approach the tensile strength of the fibers with the result that some of our experiments were terminated by fiber breakage. Although the breakage is unrelated to the observation of non-Newtonian viscosity, their close proximity in this and earlier studies suggests that brittle failure of igneous melts, may, in general, be preceded by a period of non-Newtonian rheology.