Marine Science Faculty Publications

Water and Carbon Dioxide in Basaltic Magmas

Document Type


Publication Date



igneous petrology, experimental petrology, volatiles, basalt, degassing, trace element geochemistry

Digital Object Identifier (DOI)


Experiments were conducted in which basaltic melts were equilibrated with a vapor phase consisting of pure water, pure carbon dioxide, and water-carbon dioxide mixtures at 1200°C and 200 to 980 bars in order to develop a basis for interpreting the behavior of these volatiles during the evolution and degassing of submarine magmas. Molar absorptivities for the 4500 cm^(-1) band for hydroxyl groups and for the 5230 and 1630 cm^(-1) bands of molecular water were calibrated to be 0.67 ± 0.04, 0.62 ± 0.08, and 25 ± 3 l/mole-cm, respectively. The solubility of water in MORE liquid was determined from the experiments in which MORB melt was equilibrated with pure H_20 vapor. Results are in agreement with the higher-pressure results of Hamilton et al. (1964) on Columbia River basalt. Trends observed in the concentrations of molecular water and hydroxyl groups with respect to total water concentration in the quenched, experimental, basaltic glasses are similar to those observed in albitic glasses (Silver and Stolper, 1989). Moreover, the concentration of molecular water measured in the quenched basaltic glasses is approximately proportional to water fugacity in all samples regardless of the composition of the vapor (X_CO_2), demonstrating that molecular water solubility in basaltic melts is closely approximated by Henry's law at pressures less than 1 kbar. Total water concentrations and the speciation of water in vapor-saturated basaltic melt are fit by a regular ternary solution model with the coefficients for albitic glasses (Silver and Stolper, 1989), where the activity of water in the melt is given by Henry's law for molecular water. At pressures higher than about 1 kbar, the effect of the molar volume of water in the melt (V^(0,m)_(H_2O)) on the activity of water in vapor-saturated melts is no longer negligible; a (V^(0,m)_(H_2O)) ~12 cm^3/mole fits the data of Hamilton et al. (1964).

Concentrations of CO_2 dissolved as carbonate in the experimental glasses range from 63 to 315 ppm CO_2. Carbonate was the only species of dissolved carbon observed. The mole fraction of CO_2 in the vapor varied from 0.39 to 0.93. Concentrations of CO_2 dissolved as carbonate in the melt for all the experiments are proportional to fCO_2. The data for pure CO_2-saturated and mixed H_2O-CO_2-saturated experiments are fit with a straight line through the origin with a slope of 40 ppm/100 bar fCO_2 (equivalent to 47 ppm/km water depth). These results suggest Henrian behavior for CO_2; that is, the solubility of CO_2 in the basaltic melt is essentially proportional to the fugacity of CO_2 with the same constant of proportionality whether the vapor contains pure CO_2 or H_2O + CO_2. These results do not support the widely held view that water enhances the solubility of carbon dioxide in basaltic melts.

Results of degassing calculations show that the vapor phase in equilibrium with MORB magmas at typical midoceanic eruption depths is CO_2 -rich and that the dissolved CO_2 contents should vary linearly with depth of eruption. Basaltic magmas containing < 1.0 wt. % H_2O will not degas significant quantities of water until pressures < 100 bars are reached. As water contents increase either through fractional crystallization or variations in the initial water contents. an inverse correlation is predicted between dissolved CO_2 and H_20 contents in melts saturated with a mixed H_2O-CO_2 vapor phase. These predictions were tested by examining the water and carbon dioxide concentrations in suites of basaltic glasses from the Juan de Fuca Ridge and Hawaii.

Concentrations of dissolved H_2O and CO_2 were measured in a suite of basaltic glasses from the Juan de Fuca Ridge. CO_2 contents dissolved as carbonate range from about 45 to 360 ppm by weight. In contrast to the predictions based on vapor-saturated degassing samples empted at a given depth exhibit a large range in dissolved CO_2 contents that we interpret to be the result of variable amounts of degassing. The lowest CO_2 contents at each depth are in reasonable agreement with the experimentally determined CO_2 solubility curve for basalt at low pressures. All glasses with CO_2 values higher than the experimentally determined solubility at the emption depth are oversaturated because of incomplete degassing. The highest CO_2 contents are spatially associated with the local topographic highs for each ridge segment. Lavas from relatively deep areas may have had greater opportunity to degas duIing ascent from a relatively deeper magma chamber or during late ral flow in dikes or seatloor lava flows. The highest observed CO_2 concentrations are from the axial seamount and lead to an estimate of a minimum depth to the magma chamber of 2.7 kilometers beneath the ridge axis. Water contents were not modified during degassing and were found to behave incompatibly duIing partial melting and crystal fractionation. Variations in ratios of water to other incompatible elements suggest that water has a bulk partition coefficient similar to La duIing partial melting (~D0.010).

Major, minor, and dissolved volatile element concentrations were measured in tholeiitic glasses from the submarine portion (puna Ridge) of the east lift zone of Kilauea Volcano, Hawaii. Dissolved H_2O and S concentrations display a wide range relative to nonvolatile incompatible elements at all depths. This range cannot be readily explained by fractional crystallization, degassing of H_2O and S during eruption on the seafloor, or source region heterogeneities. Dissolved CO_2 concentrations, in contrast, show a positive correlation with eruption depth and typically agree within error with the solubility at that depth. Magmas along the Puna Ridge can be modelled as resulting from (1) mixing of a relatively volatile-rich, undegassed component with magmas that experienced low pressure (perhaps subaerial) degassing during which substantial H_2O, S, and CO_2 were lost, followed by (2) fractional crystallization of olivine, clinopyroxene, and plagioclase from this mixture to generate a residual liquid; and (3) further degassing, principally of CO_2 for samples erupted deeper than 1000 m, during eruption on the seafloor. The degassed end member may form at upper levels of the summit magma chamber (assuming less than lithostatic pressure gradients), during residence at shallow levels in the crust, or during sustained summit eruptions. The final phase of degassing during eruption on the seafloor occurs slowly enough to achieve melt/vapor equilibrium during exsolution of the typically CO_2-rich vapor phase. According to the model, an average Kilauean primary magma with 16.0 % MgO should contain ~0.47 wt. % H_20 and ~900 ppm S. The model predicts that submarine lavas from wholly submarine volcanoes (i.e., Loihi), for which there is no opportunity to generate the degassed end member by low pressure degassing, will be enriched in volatiles relative to those from volcanoes whose summits have breached the sea surface (i.e., Kilauea and Mauna Loa).

Was this content written or created while at USF?


Citation / Publisher Attribution

Water and Carbon Dioxide in Basaltic Magmas, California Institute of Technolog, 275 p.