Geoscience Reference
In-Depth Information
of metallic iron in the lower mantle (Frost
et al
.,
2004) implies that carbon will be dissolved in
solid solution in the metal phase or perhaps as
iron carbide (Fe, Ni)
7
C
3
, if the ratio of carbon
over metallic iron is high enough (Dasgupta &
Hirschmann, 2010).
Morizet
et al
. (2001) and Nowak
et al
. (2003)
showed that very likely, the equilibrium between
carbonate and molecular CO
2
shifts towards the
CO
2
at high temperature, so that even in basaltic
melts at high temperature, carbon dioxide is
present partially as molecular CO
2
. This would be
consistent with the observation that the diffusion
coefficient of CO
2
in silicate melts at 1350
◦
Cis
virtually the same for rhyolitic and for basaltic
compositions. In-situ infrared spectroscopy (Kon-
schak & Keppler, 2009) has entirely confirmed
these predictions. The enthalpy of the reaction
1.4.3 Carbon solubility in silicate melts
The solubility of CO
2
in silicate melts very
strongly depends on melt composition, as has
already been noted in early studies (e.g. Wyllie
& Tuttle, 1959). While CO
2
is poorly soluble
in silica-rich and fully polymerized melts, it is
highly soluble in basic and ultrabasic, depolymer-
ized melts (Fogel & Rutherford, 1990; Blank &
Brooker, 1994; Brooker
et al
., 2001; Shishkina
et al
., 2010). This is very different from wa-
ter, where the solubility depends only slightly
on melt composition. As a result of this large
compositional effect on solubility, the effect of
CO
2
on solidus temperatures is also very differ-
ent for felsic and for basic or ultrabasic systems.
In granitic compositions in the crust, CO
2
has lit-
tle effect on the solidus and in presence of a mixed
H
2
O
O
2
−
melt
CO
2
melt
CO
2
−
melt
3
+
=
(1.7)
(where O
2
−
is some nonbridging oxygen atom
in the melt) increases with the depolymerization
of the melt. At near solidus temperatures, there
are major differences in the CO
2
/
carbonate ratio
for silicic and basic melts; however, these differ-
ences nearly disappear at 1600
◦
C, where mostly
molecular CO
2
is present.
1.4.4 C-H-O fluids
The first samples of carbon-bearing mantle fluids
were probably described by Brewster (1823), who
found ''a mysterious fluid with remarkable phys-
ical properties'' in fluid inclusions in minerals
from a peridotite xenolith. Most mantle fluids are
probably some mixture of carbon species and wa-
ter, with variable amounts of dissolved salts and
silicates. Under oxidizing conditions, CO
2
-H
2
O
mixtures can only exist in the uppermost man-
tle, since in the 2.5-3.5 GPa range, a series of
carbonation reactions consumes the CO
2
(Wyl-
lie & Huang, 1976). At greater depth, oxidizing
fluids may only be water-rich, but since the man-
tle is generally undersaturated with water, they
can exist only under unusual circumstances in a
very water-rich environment, such as in the man-
tle wedge above a subducting plate. However,
in the deeper mantle, reducing conditions pre-
vail which may stabilize methane CH
4
instead
of CO
2
. Figure 1.11 shows a calculated speci-
ation model for a C-H-O fluid along an upper
mantle adiabat, assuming that oxygen fugacity
CO
2
fluid phase, the solidus rises with the
molar fraction of CO
2
, in other words, the ef-
fect of CO
2
is essentially to reduce water activity
(Keppler, 1989). In peridotitic systems, however,
CO
2
causes a drastic depression of melting tem-
peratures (e.g. Wyllie & Huang, 1976; Dalton &
Presnall, 1998). In contrast to CO
2
, methane CH
4
is probably poorly soluble in silicate melts (Taylor
& Green, 1988; Morizet
et al
., 2010).
Wyllie and Tuttle (1959) had already noted that
the strong effect of melt composition on CO
2
solubility must be related to the stabilization
of carbonate in depolymerized melts. This was
later fully confirmed by infrared spectroscopy of
quenched silicate glasses. In polymerized, silicic
glasses, carbon dioxide is dissolved as molec-
ular CO
2
only, while basic and depolymerized
melts contain only dissolved carbonate (Fine &
Stolper, 1985; Blank & Brooker, 1994). Speciation
in glasses, however, records only the equilibria at
the glass transition temperature, not in the melts
above the liquidus. By annealing experiments,
−
