The effect of Mg concentration in silicate glasses on CO2 solubility and solution mechanism: Implication for natural magmatic systems

Abstract

International audienceFollowing an experimental approach conducted between 0.5 and 1.5 GPa, we investigated the change in CO2 solubility as a function of the XMg (MgO/(MgO+CaO)) for a range of silicate glasses. The synthesised CO2-bearing glasses have XMg up to 0.72, stoichiometric NBO/T (degree of polymerization) up to 2.6 corresponding to highly depolymerized compositions analogues to kimberlites. Several samples were synthesised with 17O enrichment to investigate the CO2 dissolution mechanism via the change in O species environments by NMR spectroscopy.The experimental results show that CO2 solubility increases with NBO/T in agreement with previous works. In addition, increasing XMg strongly decreases CO2 solubility: from 18 to 7 wt.% CO2 as XMg ranges from 0 to 0.6 (1.5 GPa and NBO/T ~ 2).17O NMR results demonstrate that CO2 molecules dissolve as CO3^2- groups showing a signal at +146 ppm for which the intensity is linearly correlated to the wt.% CO2 determined by Raman. The analysis of the oxygen environments as a function of CO2 content for MgO^NBO (+62 ppm) and CaO^NBO (+103 ppm) show that CO2 dissolves preferentially in the vicinity of Ca2+ atoms. The difference in CO2 solubility is explained by the ability for Mg2+ cations to act as a weak network former and to be present in four-fold coordination or by the stronger affinity of CO2 molecules for Ca2+ rather than for Mg2+. We show that the CO2 solubility is negatively correlated to the melt ionic field strength which reflects the variation in the affinity of CO2 molecules for one cation or another.Strongly depolymerized mantle melts, such as kimberlites, melilitites, nephelinites and basanites will exhibit lower CO2 solubility than currently assumed due to their high MgO content which must imply degassing at greater depth, potentially in the sub-lithospheric mantle

17O NMR evidence of free ionic clusters Mn+CO32−in silicate glasses: Precursors for carbonate-silicate liquids immiscibility

“…The 17 O MAS and {X}-17 O J-HMQC NMR spectra are shown in Figure 1 for several samples: XE-2 ( Figure 1A), HK-1 ( Figure 1B), HK-M ( Figure 1C) and RB8E-12 ( Figure 1D). Several oxygen environments are witnessed in the 17 O MAS NMR spectra: ~+50 ppm assigned to 17 O nuclei in BO configuration associated with Si or Al tetrahedra; ~+100 ppm assigned to 17 O nuclei in NBO configuration (Stebbins et al 1999;Lee and Stebbins 2003;Kelsey et al 2008) and ~+160 ppm assigned to 17 O nuclei in CO 3 2environments (Morizet et al 2017). The peak at ~+100 ppm is less apparent in Mg-bearing compositions (HK and XE) whereas it is prominent in Mg-free Ca-rich RB8E composition consistent with the peak assignment to Ca 2+…”

“…In the FIC M n+ ..CO 3 2-,M n+ is a charge balancing cation which can be Mg 2+ , Ca 2+ and Na + to a lower extent (Na 2 O is ~5 wt.%) for the currently investigated compositions. Although,Morizet et al (2017) showed that CO 3 2groups dissolve preferentially in the vicinity of Ca 2+ cations, we do not exclude the possibility for Mg 2+ ..CO 3 2being formed considering that Mg 2+ could play a role comparable to Ca 2+(Cormier and Cuello 2013).At this point, we wish to raise the level of confidence in the above discussion. Indeed, one…”

Reply on the comment by X. Xue on «Towards the reconciliation of viscosity change and CO2-induced polymerization in silicate melt» by Yann Morizet, Michael Paris, David Sifré, Ida Di Carlo, Sandra Ory, and Fabrice Gaillard [chemical Geology 458, 38-47]

“…Finally, the comment points out the lack of agreement in between the XCO 3 2from 17 O NMR and XCO 2 from Raman spectroscopy as reported in Morizet et al (2017b). In the present reply, we propose a corrigendum to our work (Morizet et al, 2017c) and an addendum pointing towards the future scientific questions to be answered.…”

“…In our recent work (Morizet et al, 2017b), using 17 O NMR spectroscopy, we have shown that the peak intensity at ~+150 ppm was linearly correlated to the total CO 2 content determined by Raman spectroscopy (Morizet et al, 2013). The observed correlation helped us to assign the 17 O NMR peak located at ~+150 ppm to carbonate environments in an unambiguous manner.…”

“…In this correlation (based on 7 data points), we determined a slope of 0.8. As reported in Morizet et al (2017b) the XCO 3 2-17 O NMR cannot be straightforwardly related to XCO 2 Raman mainly because the quantitative calculations are dependent on too many parameters as pointed out by Xue (2019).…”