Experimental investigation of fluid transfer in
sub-arc mantle conditions
Cassian Pirard and Jörg Hermann
Research School of Earth Sciences, Australian
National University, Canberra, ACT 0200, Australia
Fluid transport from the subducted slab to the locus of partial melting
in the mantle wedge in volcanic arcs is a process which is still strongly
debated. Two end member mechanisms are considered: 1) porous fluid transfer
through the mantle or 2) focused fluid flow in dykes/channels. These
two processes are very different and the composition of reacted fluids
arriving at the locus of partial melting in the mantle wedge must have
different trace element signatures depending on which process is involved.
mixed experiment showing anhydrous phases (Olivine, Orthopyroxene)
and hydrous phases (Biotite, Amphibole) 1b. H2O-rich
layered experiment showing the contact zone between the olivine
and the glass, forming an orthopyroxene layer.
The main goal of this experimental study is to constrain the change
in composition of the fluid as a result of these two ways of fluid transport.
Experiments were performed on natural San Carlos olivine representing
a simplified mantle and various pre-synthesized, trace element doped,
hydrous felsic glass identified as slab-extracted melts (Fig.1)(Hermann & Spandler,
2008). Synthesis piston cylinder experiments were carried out in gold
capsules for a week in the range 700°-1100°
C and 35kbar which represent
average values for the extraction of slab-fluids into the mantle (Fig.2).
Porous fluid transport was simulated by mixing a 1 to 4 ratio of fine
grained hydrous felsic glass with fine grained olivine. One end of the
capsule was filled with carbon spheres in order to collect the reacted
quenched fluid at the end of the run. These mixed charges show an olivine-orthopyroxene-biotite±garnet±
amphibole assemblage in equilibrium with a fluid (Fig. 1a). Fluid traps
collected in K2O-rich experiments (amphibole barren) were analyzed with
laser ablation ICP-MS. Fluid composition was calculated using Ce as internal
standard and normalized on the initial felsic glass. It appears that
the crystallization of phlogopite has a strong impact on the composition
of the fluid. The K2O/H2O ratio is considerably diminished (Fig. 2) and
the LILE have a strong affinity to follow potassium in phlogopite whereas
LREE, MREE and HFSE tend to be enriched into the fluids. In the case
of the H2O-rich experiments, the presence of amphibole and biotite modify
the system. Fluids are less abundant and most of the initial starting
material is retained in a hydrous peridotitic mix.
Composition of the different K-bearing phases of these sets of mixed experiments. Full squares are starting compositions; crosses are phlogopite K2O/H2O ratios; circles are quenched fluids and triangles give the amphibole composition.
Focused fluid was simulated by a layered experiment of hydrous felsic
glass overlying coarse olivine grains. A carbon spheres fluid trap was
placed over the olivine layer. Significant differences are observed in
this type of experiment compared to the mixed experiments. A reaction
zone consisting of an orthopyroxenite layer ±garnet only occurs at the
interface between olivine and the felsic glass and neither phlogopite
nor amphibole has been observed (Fig.1b). In consequence, the glass composition
is very similar to the starting composition and the shielding provided
by the garnet-orthopyroxenite reduced strongly interactions with olivine,
keeping the K2O/H2O high. LILE remain high in the quenched glass and
REE and HFSE are less affected with respect to the initial starting glass.
These two types of experiments show that there are strong differences
in transport behaviour of LILE in the mantle wedge dependent on the fluid
flow mechanism. The high K2O/H2O and LILE contents observed in arc lavas
suggest that fluid transfer in sub-arc conditions can occurs by channelled
flow. In case of porous flow, fluids are strongly affected by the crystallization
of biotite and LILE are retained in the residue (Fig. 2). However, the
melting of such hydrous peridotite residues containing both micas and
amphiboles could potentially lead to the formation of arc lavas as well.
Hermann J, Spandler C (2008) Sediment melts at sub-arc depths: an experimental
study. Journal of Petrology 49:717-740