Deep CO 2 fluids in nephelinite magmas

concentrations in Pliocene nephelinite compared to Miocene nephelinite can result also from very small carbonate-bearing xenoliths reported by Ibhi et al. (2002) but not observed in our thin sections.

Chapitre 3 – Les inclusions fluides et les minéraux riches en volatils comme traceurs des fluides

demonstrated that at high pressures and high temperatures, H2O is more soluble and tends to concentrate in the basaltic melt, whereas CO2 forms an immiscible fluid phase. The presence of phlogopite in Type 3 inclusions may however indicate that a small amount of water was present in the alkaline silicate melt. Saghro nephelinite magmas may have reached CO2-saturation at

Figure 4.14. P–T relations calculated for Type 1 (pure CO²) fluid inclusions. Miocene nephelinites: olivines containing (A) only Type 1A inclusions, (B) Type 1A and 2 inclusions, (C) Type 1A and 3 inclusions, (D) Type 1A, 2 and 3 of inclusions, and (F) Type 1B inclusions. Pliocene nephelinites: (E) olivines containing only Type 1A inclusions. Isochores have been calculated with equations from Span and Wagner (1996). The dotted line represents the crystallization temperature obtained for clinopyroxenes, equal to 1364 ºC for Miocene nephelinites and to 1320°C for Pliocene nephelinites. Dashed lines show the extreme minimum trapping pressures corresponding to crystallization temperatures calculated for clinopyroxenes.

depth but no saturation in other volatile species such as H2O, Cl, F, and S. The amount of components other than CO2 is then small enough to approximate P–T relations from pure CO2

systems.

The density of primary–pseudo-secondary fluid inclusions (Type 1B) of 0.81 ± 0.004 g/cm³ computed with microthermometric data is consistent with the 0.92 ± 0.02 g/cm³ density calculated based on the split of the Fermi diad measured with Raman microspectroscopy, using the method of Kawakami et al. (2003). Pressure of inclusion entrapment or re-equilibration corresponding to calculated densities can be derived from isochores using equations of state from Span and Wagner (1996) for pure CO2. This equation of state was chosen because the calibration ranges are 216–1100K and 0.5–800 MPa, which is suitable for our magmatic system. Assuming a model temperature of 1364 °C for Miocene nephelinites from clinopyroxene-melt thermometry, primary fluid inclusions (Type 1B) from Saghro Miocene nephelinite samples yield a minimum trapping pressure of 590 MPa (Fig. 4.14; Table 4.6). This pressure corresponds to a lithostatic depth of about 19 km.

Because it was not possible to carry out microthermometry on the fluid phases of Type 2 (fluid-solid) inclusions, we calculated the density of CO2 using the formula of Kawakami et al.

(2003) that returned densities of 0.45 ± 0.02 g/cm³. The lower density of the CO2 phase of Type 2 inclusions compared to that of Type 1B is probably due to the shrinkage caused by the crystallization of daughter phases in Type 2 inclusions, inducing expansion (hence lower density) of the CO2 phase. With respect to the density of coexisting fluid inclusions, a similar order of magnitude density drop was observed in fluid bubbles formed by in-situ immiscibility of CO2

volatile and silicate melt in melt inclusions from the subcontinental lithospheric mantle (Hidas et al., 2010). Nevertheless, in Saghro Miocene nephelinites, the minimum calculated pressures for Type 1B using Kawakami et al. (2003) is 775 ± 6 MPa, whereas it is irrelevant to estimate the trapping pressure of Type 2 inclusions because of heterogeneous entrapment of CO2 fluid and silicate melt and thus variable fluid/solid ratio (between Type 1B = pure fluid and Type 3 = pure melt) leading to a wide range of pressures. At the time of entrapment, the CO2-fluid in Type 2 inclusions had the same density as Type 1B inclusions. As the magma ascends and cools down, the silicate melt crystallized (i.e. shrinks) and the CO2 fluid expands in response to shrinkage, thus reducing its density.

It is worth noting that the large difference between the depths calculated from fluid inclusion densities and those from clinopyroxene-melt thermobarometry (ca. 2200 MPa for

Chapitre 3 – Les inclusions fluides et les minéraux riches en volatils comme traceurs des fluides

Miocene nephelinites) can be explained by the fact that fluid inclusions typically undergo post- entrapment modifications during magma ascent. These processes include leakage, decrepitation, stretching and necking-down, and occur in response to changing pressure and high temperature (e.g. Roedder, 1984; Hansteen et al., 1998; Andersen and Neumann, 2001). These processes can alter the internal pressure of the fluid inclusions towards the new ambient conditions, whether it is partial or total re-equilibration, faster than they alter mineral compositions. Indeed, both systems do not adapt at the same speed to changing pressures. Re-equilibration or formation of fluid inclusions can occur within hours to days (e.g. Wanamaker et al., 1990), whereas growth of a 5–10 µm wide clinopyroxene rim at equilibrium conditions or diffusive re-equilibration along 5–10 µm width needs months to years (Cashman, 1990; Putirka, 1997). Thus, clinopyroxene-melt equilibria indicate a major crystal fractionation event at depth, while fluid inclusions record short-term events at shallower levels during magma ascent. Moreover, since all post-entrapment processes lower the inclusion densities, we can assume that our calculated pressures represent minimum values.

Dixon (1997) used a closed-system degassing model to estimate the pressure at which a nephelinitic magma with initial H2O and CO2 contents of 2.0 and 1.0 wt.%, respectively, reached saturation. She constrained vapor saturation at ~400 MPa, which is relatively close to the trapping pressure of our primary Type 1B inclusions. Similarly, the VolatileCalc Excel^© spreadsheet from Newman and Lowenstern (2002) has been calibrated to estimate solubility curves for H2O–CO2 in basaltic systems. They used the model of Dixon (1997) for nephelinitic to tholeiitic magma compositions and thus it is suitable for our samples. Since the trapping pressure of our Type 1B inclusions is close to the upper limit of the model (~590 MPa), we used an extrapolation to estimate the solubility of a 40 wt.% SiO2 nephelinite magma at 590 MPa, considering a pure CO2 system, and we found a CO2 solubility of ~1.7 wt.%. It should be noted that a small amount of water in nephelinite melt (1 wt.%) at 590 MPa only slightly increases the CO2 solubility to 1.75 wt.% CO2 (1 wt.% in H2O-free system). The composition of the fluid inclusions remains globally unchanged since the exsolved vapor is composed of ~96 mol.% CO2

and ~4 mol.% H2O.

It is important to note that while the fluid phase in Type 2 is less dense than Type 1B inclusions (probably due to shrinkage), it is still denser than the fluid in Type 1A fluid inclusions, emphasizing the fact that the latter represents a different inclusion generation. Primary–pseudo- secondary CO2 fluid inclusions (Type 1B, 2 and 3) represent the greatest depth of trapping of our fluid inclusions dataset and Type 1B are present only in Miocene nephelinite. The occurrence of

these three types of inclusions together suggests heterogeneous entrapment at depth and that within a relatively short time span different fluid/melt ratios were entrapped in olivine. This event is present in Miocene nephelinite, whereas olivines in Pliocene nephelinite have only rare Type 2 and 3. The absence of Type 1B inclusions does not allow estimation of deep exsolving processes in Pliocene nephelinites.

Petrographic observations, microthermometric measurements and Raman analyses indicate that Type 1A fluid inclusions are secondary and the composition is CO2–dominated.

However, the low melting temperatures observed during microthermometric measurements of Type 1A inclusions suggest the presence of minor amounts of component(s) soluble in CO2, possibly CH4 or N2, but their concentration is below the detection limit for Raman spectroscopy (Fig. 4.10; Table 4.6). Assuming a model temperature of 1364 °C for Miocene nephelinites and 1320 °C for Pliocene nephelinites (clinopyroxene-melt thermometry), secondary (Type 1A) fluid inclusions yield minimum trapping pressures ranging between 10 and 90 MPa (i.e. 0.5–3 km), and between 55 and 150 MPa (i.e. 2–5 km), respectively (Fig. 4.14; Table 4.6). This indicates a shallow degassing event that affected both Miocene and Pliocene nephelinites.