Elsevier

Geochimica et Cosmochimica Acta

Volume 68, Issue 22, 15 November 2004, Pages 4687-4703
Geochimica et Cosmochimica Acta

CO2 solubility in dacitic melts equilibrated with H2O-CO2 fluids: Implications for modeling the solubility of CO2 in silicic melts

https://doi.org/10.1016/j.gca.2004.04.019Get rights and content

Abstract

The solubility of CO2 in dacitic melts equilibrated with H2O-CO2 fluids was experimentally investigated at 1250°C and 100 to 500 MPa. CO2 is dissolved in dacitic glasses as molecular CO2 and carbonate. The quantification of total CO2 in the glasses by mid-infrared (MIR) spectroscopy is difficult because the weak carbonate bands at 1430 and 1530 cm−1 can not be reliably separated from background features in the spectra. Furthermore, the ratio of CO2,mol/carbonate in the quenched glasses strongly decreases with increasing water content. Due to the difficulties in quantifying CO2 species concentrations from the MIR spectra we have measured total CO2 contents of dacitic glasses by secondary ion mass spectrometry (SIMS).

At all pressures, the dependence of CO2 solubility in dacitic melts on xfluidCO2,total shows a strong positive deviation from linearity with almost constant CO2 solubility at xCO2fluid > 0.8 (maximum CO2 solubility of 795 ± 41, 1376 ± 73 and 2949 ± 166 ppm at 100, 200 and 500 MPa, respectively), indicating that dissolved water strongly enhances the solubility of CO2. A similar nonlinear variation of CO2 solubility with xCO2fluid has been observed for rhyolitic melts in which carbon dioxide is incorporated exclusively as molecular CO2 (Tamic et al., 2001). We infer that water species in the melt do not only stabilize carbonate groups as has been suggested earlier but also CO2 molecules.

A thermodynamic model describing the dependence of the CO2 solubility in hydrous rhyolitic and dacitic melts on T, P, fCO2 and the mol fraction of water in the melt (xwater) has been developed. An exponential variation of the equilibrium constant K1 with xwater is proposed to account for the nonlinear dependence of xCO2,totalmelt on xCO2fluid. The model reproduces the CO2 solubility data for dacitic melts within ±14% relative and the data for rhyolitic melts within 10% relative in the pressure range 100–500 MPa (except for six outliers at low xCO2fluid). Data obtained for rhyolitic melts at 75 MPa and 850°C show a stronger deviation from the model, suggesting a change in the solubility behavior of CO2 at low pressures (a Henrian behavior of the CO2 solubility is observed at low pressure and low H2O concentrations in the melt). We recommend to use our model only in the pressure range 100–500 MPa and in the xCO2fluid range 0.1–0.95. The thermodynamic modeling indicates that the partial molar volume of total CO2 is much lower in rhyolitic melts (31.7 cm3/mol) than in dacitic melts (46.6 cm3/mol). The dissolution enthalpy for CO2 in hydrous rhyolitic melts was found to be negligible. This result suggests that temperature is of minor importance for CO2 solubility in silicic melts.

Introduction

Volatiles dissolved in silicate melts influence dramatically the chemical and physical properties of magmas. Because of its high solubility in silicate melts, water is of particular interest for understanding properties of magmas. CO2 is the second most abundant volatile in natural magmas. Although CO2 is usually subordinate in concentration in silicic magmas, it is often the first component to reach saturation due to its low solubility in the melt. Even small amounts of dissolved CO2 in water bearing melts shift the fluid saturation limit to higher pressures and thus greater depth (Holloway, 1976). Because vesiculation, initiated by exsolution of CO2-H2O-bearing fluids from the melt, is the driving force in many eruptive situations, it is important to investigate the saturation limit of fluids in the system C-H-O.

Similar to H2O, CO2 dissolves in silicate glasses and melts in form of at least two different species, as an unreacted molecular species (CO2 molecules, hereafter referred to as CO2,mol) and as species formed by reaction between molecular species and the silicate framework (carbonate groups). Whereas water speciation in glasses depends mainly on the total water content but only weakly on anhydrous composition (e.g., Silver et al., 1990; Behrens et al., 1996), the speciation of CO2 is found to be independent on total CO2 concentration but varies strongly with anhydrous composition of the glass (Fine and Stolper, 1985; Blank and Brooker, 1994; Brooker et al., 2001). In highly polymerized rhyolitic glasses CO2 is incorporated exclusively as molecular CO2 (e.g., Blank et al., 1993; Tamic et al., 2001), whereas only carbonate is present in more depolymerized glasses (Blank and Brooker, 1994). Glasses of intermediate compositions such as dacite, andesite and phonolite contain both types of species (Blank and Brooker, 1994; King and Holloway, 2002; Morizet et al., 2002).

Although CO2 solubility data (if not specified otherwise, the term CO2 refers to the total carbon dioxide) are available for a variety of natural and synthetic melt compositions (Mysen et al., 1975, 1976; Blank and Brooker, 1994; Dixon et al., 1995; Jakobsson, 1997; Brooker et al., 1999; Tamic et al., 2001; King and Holloway, 2002; King et al., 2002; Morizet et al., 2002), general models to predict the solubility of CO2 in silicate melts of mafic to silicic compositions (Spera and Bergman, 1980; Papale, 1999; Brooker et al., 2001) have large errors. This is due mainly to the uncertainty of the input data for the thermodynamic models. Especially, the early studies on CO2 solubility in silicate melts suffer from imprecise analytical techniques.

CO2 solubility in melts equilibrated with mixed H2O-CO2 fluids was investigated only in a few studies. In the pioneering works of Mysen et al. (1975, 1976), the highest CO2 solubilities in andesitic, basaltic and albitic compositions were not found in melts in equilibrium with pure CO2 fluids but for fluids with mole fractions of CO2 (xCO2fluid) of 0.6–0.7. Mysen et al. (1975, 1976) suggested that the formation of OH groups depolymerizes the melt structure, resulting in an increase of carbonate groups in the melt. However, as discussed by Tingle (1987), there may be substantial errors with the 14C β-track autoradiography which was used by Mysen et al. (1975, 1976) to determine carbon quantitatively. In more recent studies, IR and NMR spectroscopy was used to measure CO2 speciation and concentration in glasses. Rhyolitic compositions (containing only molecular CO2) were examined by Blank et al. (1993) at 75 MPa and 850°C, and by Tamic et al. (2001) at 200 and 500 MPa and 800 and 1100°C. Basaltic compositions (containing only carbonate) were studied by Dixon et al. (1995) at 1200°C and pressures up to 98 MPa. To our knowledge, the effect of water on CO2 speciation in glasses containing both CO2,mol and carbonate groups has been systematically investigated only in the studies of Kohn and Brooker (1994) and King and Holloway (2002) using samples synthesized at higher pressures in piston cylinder apparatus. Using 13C MAS NMR spectroscopy, Kohn and Brooker (1994) found a maximum in the ratio of CO2,mol/carbonate at a water content of 1–2 wt% in albitic and jadeitic glasses. On the other hand, IR spectroscopy on andesitic glasses showed a continuous decrease of CO2,mol/carbonate with increasing water content of the melt from 0 to 3.4 wt% (King and Holloway, 2002).

In this paper, we report new experimental results on the solubility of CO2 in dacitic melts in equilibrium with mixed H2O-CO2 fluids at 100, 200 and 500 MPa and 1250°C. To understand the solubility behavior of CO2 in the melts, we have also examined the effect of water on the speciation of CO2 in the quenched dacitic glasses using IR micro-spectroscopy. However, because the quantification of carbonate concentration in the glass by IR spectroscopy was found to be too imprecise, we applied secondary ion mass spectrometry (SIMS) to measure quantitatively the total CO2 concentration in the samples. On the basis of the new solubility and speciation data for dacitic composition and data for rhyolitic composition from previous studies, a thermodynamic model is proposed to describe the dependence of the CO2 solubility in silicic melts (rhyolite to dacite) on temperature, pressure, fugacity of CO2 (fCO2) and the water content in the melt.

Section snippets

Starting material

The starting composition was a synthetic dry glass with a composition close to the bulk composition of the dacite of the Unzen Volcano, Japan (Chen et al., 1993). The homogenous dry glass was synthesized by melting oxides and carbonates at 1600°C for more than 4 h, grinding and remelting for additional 4 h at the same temperature. The composition of the glass, determined by electron microprobe, is given in Table 1 (average analysis using a Cameca microprobe, with 15 kV accelerating voltage, 5

Karl-Fischer titration (KFT)

The water content of glasses (Cwater) from the solubility experiments were measured after thermal extraction using Karl-Fischer titration (Behrens, 1995). Single wafers of 10–20 mg were wrapped tightly in a Pt foil to avoid explosion during heating. To account for unextracted water in the samples after analysis, 0.13 ± 0.07 wt% H2O were added to the quantity measured by KFT. This value was found to be typical for polymerized rhyolitic glasses (Leschik et al., 2004) as well as for depolymerized

MIR spectroscopic investigation of CO2

Figure 1a shows IR spectra of selected CO2 bearing dacitic glasses in comparison to the starting glass synthesized in air. The sharp absorption band at 2348 cm−1 is attributed to the ν3 asymmetric stretching vibration of CO2,mol dissolved in the glass (Fine and Stolper, 1985). This peak is distinct from that for free gaseous molecular CO2, which would rise to a doublet of unresolved rotational structure centered at ∼2350 cm−1 (Brooker et al., 1999). The band system at ∼1530 cm−1 and ∼1430 cm−1

Speciation of CO2

As shown in Figure 6a, the ratio of CO2,mol (measured by IR spectroscopy) to CO2,total (measured by SIMS) in dacitic glasses strongly decreases with increasing water content. Up to 30% of carbon dioxide is dissolved in molecular form in water-poor glasses but less than 3% in glasses containing more than 3 wt% water (Table 4). This trend is consistent with data of King and Holloway (2002) for andesitic glasses. On the other hand, synthesis pressure and total CO2 content appears to have a minor

Conclusions

New CO2 solubility data are presented for dacitic melts equilibrated with H2O-CO2 fluids at 1250°C and pressures of 100, 200 and 500 MPa. SIMS was calibrated to quantify the total carbon content of the quenched glasses using self-made glass standards. The relative sensitivity factor for 12C/28Si was found to decrease systematically with water content of the glasses. Due to variations in SIMS measurement conditions calibration lines have to be determined for each analytical session.

A strong

Acknowledgments

This study was supported by the Deutsche Forschungsgemeinschaft (DFG grant Ho 1337/3+7). Fruitful comments of Hans Keppler, Yang Liu and two anonymous reviewers helped to improve the manuscript.

References (63)

  • J.D. Kubicki et al.

    Structural role of CO2 and [CO3]2− in fully polymerized, sodium-aluminosilicate melts and glasses

    Geochim. Cosmochim. Acta

    (1995)
  • D.P. Mattey

    Carbon dioxide solubility and carbon isotope fractionation in basaltic melt

    Geochim. Cosmochim. Acta

    (1991)
  • K.D. McKeegan et al.

    Ion microprobe isotopic measurements of individual interplanetary dust particles

    Geochim. Cosmochim. Acta

    (1985)
  • Y. Morizet et al.

    CO2 in haplophonolitic meltSolubility, speciation and carbonate complexation

    Geochim. Cosmochim. Acta

    (2002)
  • M. Nowak et al.

    Carbon dioxide speciation in silicate meltsA restart

    Earth Planet Sci. Lett.

    (2003)
  • S. Ohlhorst et al.

    Compositional dependence of molar absorptivities of near-infrared OH- and H2O bands in rhyolitic to basaltic glasses

    Chem. Geol.

    (2001)
  • V. Pan et al.

    The pressure and temperature dependence of carbon dioxide solubility in tholeiitic basalt melts

    Geochim. Cosmochim. Acta

    (1991)
  • P. Richet et al.

    Water and the viscosity of andesitic melts

    Chem. Geol.

    (1996)
  • N. Tamic et al.

    The solubility of H2O and CO2 in rhyolitic melts in equilibrium with a mixed CO2-H2O fluid phase

    Chem. Geol.

    (2001)
  • T.N. Tingle

    An evaluation of the carbon-14 beta track techniqueImplications for solubilities and partition coefficients determined by beta track mapping

    Geochim. Cosmochim. Acta

    (1987)
  • Y. Zhang et al.

    H2O diffusion in rhyolitic melts and glasses

    Chem. Geol.

    (2000)
  • E. Zinner et al.

    Interstellar SiC in the Murchison and Murray meteoritesIsotopic compostion of Ne, Xe, Si, C and N

    Geochim. Cosmochim. Acta

    (1989)
  • L.Y. Aranovich et al.

    Experimental determination of CO2-H2O activity-composition relations at 600–1000°C and 6–14 kbar by reversed decarbonation and dehydration reactions

    Am. Mineral.

    (1999)
  • H. Behrens

    Determination of water solubilities in high-viscosity meltsAn experimental study on NaAlSi3O8 and KAlSi3O8 melts

    Eur. J. Mineral.

    (1995)
  • H. Behrens et al.

    Quantification of water speciation in silicate glasses and melts by IR spectroscopy—In situ vs. quench technique

    Phase Transitions.

    (2003)
  • H. Behrens et al.

    Quantification of H2O contents in silicate glasses using IR spectroscopy—A calibration based on hydrous glasses analyzed by Karl-Fischer titration

    Glass Sci. Techn.

    (2003)
  • H. Behrens et al.

    Determination of molar absorption coefficient for the IR absorption band of CO2 in rhyolitic glasses

    Am. Mineral.

    (2004)
  • J. Berndt et al.

    A combined rapid-quench and H2-membrane setup for internally heated pressure vesselsDescription and application for water solubility in basaltic melts

    Am. Mineral.

    (2002)
  • J.G. Blank

    An experimental investigation of the behaviour of carbon dioxide in rhyolitic melt

    (1993)
  • J.G. Blank et al.

    Experimental studies of carbon dioxide in silicate meltsSolubility, speciation and stable isotope behavior

    Rev. Mineral.

    (1994)
  • C.-H. Chen et al.

    Relationship between eruption volume and neodymium isotopic composition at Unzen volcano

    Nature

    (1993)
  • Cited by (78)

    • Raman spectroscopy to determine CO<inf>2</inf> solubility in mafic silicate melts at high pressure: Haplobasaltic, haploandesitic and approach of basaltic compositions

      2021, Chemical Geology
      Citation Excerpt :

      More recently, higher pressures were reached but, for mafic compositions, few studies achieve pressures higher than 1 GPa, and very few higher than 3 GPa (review in Iacono-Marziano et al., 2012). On the theoretical side, often to interpret experimental data, a first approach developed classical models based on macroscopic thermodynamics yielding a very abundant literature (e.g., Khitarov and Kadik, 1973; Dixon et al., 1995; Papale, 1999; Behrens et al., 2004; Iacono-Marziano et al., 2012). These models are essentially empirical and include parameters that are adjusted to reproduce experimental data.

    View all citing articles on Scopus

    Associate editor: C. Romano

    View full text