ALBERT

Description: Viscosities of liquid albite (NaAlSi3O8) and a Himalayan leucogranite were measured near the glass transition at a pressure of one atmosphere for water contents of 0, 2·8 and 3·4 wt.%. Measured viscosities range from 1013·8 Pa. s at 935 K to 109·0 Pa. s at 1119 K for anhydrous granite, and from 1010·2 Pa. s at 760 K to 1012·9 Pa. s at 658 K for granite containing 3·4 wt.% H2O. The leucogranite is the first naturally occurring liquid composition to be investigated over the wide range of T-X(H2O) conditions which may be encountered in both plutonic and volcanic settings. At typical magmatic temperatures of 750°C, the viscosity of the leucogranite is 1011·0 Pa. s for the anhydrous liquid, dropping to 106·5 Pa. s for a water content of 3 wt.% H2O. For the same temperature, the viscosity of liquid NaAlSi3O8 is reduced from 1012·2 to 106·3 Pa. s by the addition of 1·9 wt.% H2O. Combined with published high-temperature viscosity data, these results confirm that water reduces the viscosity of NaAlSi3O8 liquids to a much greater degree than that of natural leucogranitic liquids. Furthermore, the viscosity of NaAlSi3O8 liquid becomes substantially nonArrhenian at water contents as low as 1 wt.% H2O, while that of the leucogranite appears to remain close to Arrhenian to at least 3 wt.% H2O, and viscosity–temperature relationships for hydrous leucogranites must be nearly Arrhenian over a wide range of temperature and viscosity. Therefore, the viscosity of hydrous NaAlSi3O8 liquid does not provide a good model for natural granitic or rhyolitic liquids, especially at lower temperatures and water contents.Qualitatively, the differences can be explained in terms of configurational entropy theory because the addition of water should lead to higher entropies of mixing in simple model compositions than in complex natural compositions. This hypothesis also explains why the water reduces magma viscosity to a larger degree at low temperatures, and is consistent with published viscosity data for hydrous liquid compositions ranging from NaAlSi3O8 and synthetic haplogranites to natural samples. Therefore, predictive models of magma viscosity need to account for compositional variations in more detail than via simple approximations of the degree of polymerisation of the melt structure.

Description: Isobaric crystallisation paths obtained from phase equilibrium experiments show that, whereas in rhyolitic compositions melt fraction trends are distinctly eutectic, dacitic and more mafic compositions have their crystallinities linearly correlated with temperatures. As a consequence, the viscosities of the latter continuously increase on cooling, whereas for the former they remain constant or even decrease during 80% of the crystallisation interval, which opens new perspectives for the fluid dynamical modelling of felsic magma chambers. Given the typical dyke widths observed for basaltic magmas, results of analogue modelling predict that injection of mafic magmas into crystallising intermediate to silicic plutons under pre-eruption conditions cannot yield homogeneous composition. Homogenisation can occur, however, if injection takes place in the early stages of magmatic evolution (i.e. at near liquidus conditions) but only in magmas of dacitic or more mafic composition. More generally, the potential for efficient mixing between silicic and mafic magmas sharing large interfaces at upper crustal levels is greater for dry basalts than for wet ones. At the other extreme, small mafic enclaves found in many granitoids behave essentially as rigid objects during a substantial part of the crystallisation interval of the host magmas, which implies that finite strain analyses carried out on such markers can give only a minimum estimate of the total amount of strain experienced by the host pluton. Mafic enclaves carried by granitic magmas behave as passive markers only at near solidus conditions, typically when the host granitic magma shows near-solid behaviour. Thus they cannot be used as fossil indicators of direction of magmatic flow.

Description: Liquidus phase relationships at H2O-saturated and -undersaturated conditions and 2 kbar in the systems Qz-Or-Ab (SiO2-KAlSi3O8-NaAlSi3O8), Qz-Or-Ab-Al2O3, and subsystems are compared and discussed. In the peraluminous systems (i.e. when melts are saturated with respect to mullite) the liquidus temperatures are lowered by 40-55°C for compositions in the quartz primary field and by 15-25°C for cotectic compositions. The composition of the Qz-Ab eutectic and of the minimum are slightly shifted towards more Qz-rich compositions (minimum composition at P(H2O) = 2 kbar in the system Qz-Or-Ab-A12O3, saturated with respect to mullite: Qz40Or23Ab37). In melts saturated with sillimanite or mullite, the effect of high Al content may be lower for the Qz-Or than for the Qz-Ab eutectic.The depression of the liquidus temperatures may be partly related to the higher H2O solubility in melts saturated with respect to mullite. The solubility of H2O in a melt with a composition of Qz28Or34Ab38 at 2 kbar and 800°C is 5·77 ± 0·15 wt% H2O and 6·36 ± 0·30 wt% H2O in a melt with the same Qz/Or/Ab proportions but saturated with respect to mullite.The effect of high Al contents on the Mg and Fe contents of Ca-free granite melts was investigated at 775°C-3 kbar (melts coexisting with phlogopite), and at 820°C-2 kbar (melts coexisting with biotite and spinel), under NNO buffer conditions. Less than 0·15 wt% MgO is incorporated in subaluminous melts coexisting with phlogopite, whereas peraluminous melts (2·9 wt% normative corundum) contain 0·6–0·7 wt% MgO. A similar behaviour of the MgO content is observed for melts coexisting with biotite. In contrast, no significant effect of high Al contents on the FeO content of melt coexisting with biotite was observed. This suggests that the Fe/Mg ratio may be significantly lower in peraluminous than in subaluminous granitic melts.

Description: Melting experiments with plagioclases were performed in the systems Ab-An, Ab-An-H2O, Qz-Ab-An-H2O, and Qz-Or-Ab-An-H2O-CO2. The experimental products were analysed by electron microprobe, and the kinetics of the reactions were studied qualitatively.Melting of the plagioclase in the system Ab-An (P = 1 atm, T = 1420°C) is very fast in the first minutes but becomes slower with increasing run duration and is incomplete even after 1000 hours. The Ab/An fractionation between melt and residual plagioclase is similar to that described by Bowen (1913).Melting kinetics of plagioclase in the system Ab-An-H2O ( = 5 kbar, T = 1000°C) is controlled by the diffusion of water into the plagioclase structure. Melting is especially fast parallel to the a-axis. The experimental products show separation of melts and crystals.In the tonalite system Qz-Ab-An-H2O, equilibrium melting could be observed down to 830°C ( = 2 kbar) but not a lower temperatures. The kinetics of the reaction is enhanced by deficiency or excess of alumina in the aluminosilicate melt surrounding the plagioclase crystals. The fractionation of Ab and An between melt and plagioclase crystals is more pronounced in the presence of quartz than in the Ab-An-H2O system. The ratio An/An + Ab is approximately 0·35 in the melt and 0·85 in the coexisting plagioclase T = 880°C).In the haplogranodiorite system Qz-Or-Ab-An-H2O–CO2, melting reactions were performed at P = 0·5 kbar, T = 880°C, and of approximately 0·5. It is assumed that near equilibrium compositions of melt and coexisting residual plagioclase could be obtained in long duration runs (run time = 60 days). The distribution of Ab and An between melt and minerals is similar to that observed in the tonalite system. The partial melt coexisting with an An-rich plagioclase and Or-rich K-feldspar is relatively poor in An.