TY - JOUR
T1 - Carbon solubility in silicate melts in equilibrium with a CO-CO2 gas phase and graphite
AU - Yoshioka, Takahiro
AU - Nakashima, Daisuke
AU - Nakamura, Tomoki
AU - Shcheka, Svyatoslav
AU - Keppler, Hans
N1 - Funding Information:
The authors would like to thank Raphael Njul and Hubert Schulze for sample polishing and Detlef Krausse for help with the microprobe analyses. Masakuni Yamanobe and Takamichi Miyazaki assisted in SIMS analyses. Constructive comments by Mike Toplis and two anonymous referees helped to improve the manuscript. This work was supported by the DFG-JSPS International Research Training Group "Deep Earth Volatile Cycles" (DFG GRK 2156).
Publisher Copyright:
© 2019 Elsevier Ltd
PY - 2019/8/15
Y1 - 2019/8/15
N2 - The solubility of carbon in silicate melts (Fe-free MORB basalt, Fe-free andesite and Fe-free rhyolite) coexisting with graphite and a CO-CO2 fluid phase was measured to 3 GPa and 1500 °C. Experiments in the 0.2–0.5 GPa range were carried out in an internally heated pressure vessel equipped with a rapid-quench device. In these experiments, glass powder and pure CO gas were directly loaded into platinum capsules; during the runs, graphite precipitated, indicating that the equilibrium 2 CO = C + CO2 in the gas phase was reached. Experiments at higher pressures were carried out in a piston cylinder apparatus, starting with glass powder, graphite, and Ag2C2O4 as a source of CO2. Carbon contents in quenched glasses were measured by secondary ion mass spectrometry (SIMS). Bulk carbon solubility is directly proportional to pressure and follows Henry's law with good approximation. The effect of temperature on solubility is small. Henry coefficients of bulk carbon solubility obtained by fitting all data were 2.15 ppm C/MPa for MORB, 1.57 ppm C/MPa for andesite and 2.14 ppm C/MPa for rhyolite. In almost all samples, bulk carbon contents were higher than the content of oxidized carbon (CO2 and carbonate). The difference is interpreted to be due to dissolved CO, consistent with bands in the 2100–2200 cm−1 range of the infrared and Raman spectra. For MORB, the solubility cCO of CO (expressed as wt.% carbon) may be described by the relationship log cCO MORB = −5.20 + 0.80 log fCO (R2 = 0.83), where fCO is CO fugacity, while the rhyolite data are best described by log cCO Rhyolite = −4.08 + 0.52 log fCO (R2 = 0.72). Our data imply that the solubility of carbon in silicate melts depends very strongly on whether or not graphite saturation in the gas phase is reached, as the equilibrium with graphite affects the fugacity of CO and CO2. If the equilibrium 2 CO = C + CO2 is attained through graphite precipitation, bulk carbon solubility will be close to the solubility of CO2 over a wide pressure and temperature range, since the gas phase consists mostly of CO2. Only at low pressures of a few 100 MPa or below, the fraction of CO in the gas phase becomes so large that bulk carbon solubility is significantly reduced compared to CO2. Low carbon solubilities may also be observed in situations where graphite precipitation is kinetically inhibited. This could have been the case in the source region of lunar fire fountain eruptions, where dissolved CO may have caused vapor saturation up to 4 km below the lunar surface. We argue that in a Hadean magma ocean on Earth, extensive graphite precipitation and a low atmospheric pressure on the surface could have stabilized a CO-rich atmosphere that may have limited carbon sequestration in the core. We suggest that dissolved CO in glasses may be easily overlooked, because the infrared extinction coefficient for CO (966 ± 229 l mol−1 cm−2 in rhyolite glass) is more than one order of magnitude lower than that of CO2 or carbonate.
AB - The solubility of carbon in silicate melts (Fe-free MORB basalt, Fe-free andesite and Fe-free rhyolite) coexisting with graphite and a CO-CO2 fluid phase was measured to 3 GPa and 1500 °C. Experiments in the 0.2–0.5 GPa range were carried out in an internally heated pressure vessel equipped with a rapid-quench device. In these experiments, glass powder and pure CO gas were directly loaded into platinum capsules; during the runs, graphite precipitated, indicating that the equilibrium 2 CO = C + CO2 in the gas phase was reached. Experiments at higher pressures were carried out in a piston cylinder apparatus, starting with glass powder, graphite, and Ag2C2O4 as a source of CO2. Carbon contents in quenched glasses were measured by secondary ion mass spectrometry (SIMS). Bulk carbon solubility is directly proportional to pressure and follows Henry's law with good approximation. The effect of temperature on solubility is small. Henry coefficients of bulk carbon solubility obtained by fitting all data were 2.15 ppm C/MPa for MORB, 1.57 ppm C/MPa for andesite and 2.14 ppm C/MPa for rhyolite. In almost all samples, bulk carbon contents were higher than the content of oxidized carbon (CO2 and carbonate). The difference is interpreted to be due to dissolved CO, consistent with bands in the 2100–2200 cm−1 range of the infrared and Raman spectra. For MORB, the solubility cCO of CO (expressed as wt.% carbon) may be described by the relationship log cCO MORB = −5.20 + 0.80 log fCO (R2 = 0.83), where fCO is CO fugacity, while the rhyolite data are best described by log cCO Rhyolite = −4.08 + 0.52 log fCO (R2 = 0.72). Our data imply that the solubility of carbon in silicate melts depends very strongly on whether or not graphite saturation in the gas phase is reached, as the equilibrium with graphite affects the fugacity of CO and CO2. If the equilibrium 2 CO = C + CO2 is attained through graphite precipitation, bulk carbon solubility will be close to the solubility of CO2 over a wide pressure and temperature range, since the gas phase consists mostly of CO2. Only at low pressures of a few 100 MPa or below, the fraction of CO in the gas phase becomes so large that bulk carbon solubility is significantly reduced compared to CO2. Low carbon solubilities may also be observed in situations where graphite precipitation is kinetically inhibited. This could have been the case in the source region of lunar fire fountain eruptions, where dissolved CO may have caused vapor saturation up to 4 km below the lunar surface. We argue that in a Hadean magma ocean on Earth, extensive graphite precipitation and a low atmospheric pressure on the surface could have stabilized a CO-rich atmosphere that may have limited carbon sequestration in the core. We suggest that dissolved CO in glasses may be easily overlooked, because the infrared extinction coefficient for CO (966 ± 229 l mol−1 cm−2 in rhyolite glass) is more than one order of magnitude lower than that of CO2 or carbonate.
KW - Carbon
KW - CO
KW - CO2
KW - graphite
KW - lunar volcanism
KW - magma ocean
KW - primordial atmosphere
UR - http://www.scopus.com/inward/record.url?scp=85067601044&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85067601044&partnerID=8YFLogxK
U2 - 10.1016/j.gca.2019.06.007
DO - 10.1016/j.gca.2019.06.007
M3 - Article
AN - SCOPUS:85067601044
SN - 0016-7037
VL - 259
SP - 129
EP - 143
JO - Geochimica et Cosmochimica Acta
JF - Geochimica et Cosmochimica Acta
ER -