Skip Navigation | ANU Home | Search ANU | Directories
The Australian National University
Research School of Earth Sciences
Printer Friendly Version of this Document
Untitled Document

Accurate determination of the solidus of simplified spinel lherzolite in the system CaO-MgO-Al2O3-SiO2 (CMAS) at 11 kbar: forward and reversal experiments

X. Liu and H.StC. O’Neill

The simple system CaO-MgO-Al2O3-SiO2 (CMAS) is widely recognised as a good analogue for the Earth’s mantle, since, under the appropriate pressure-temperature conditions, the major mineral phases of the mantle all occur in this system. In the upper mantle, there are four such phases, namely, olivine, orthopyroxene, clinopyroxene, and an aluminous phase, plagioclase, spinel or garnet, depending on pressure. Accurate determination of melt compositions multiply saturated with four mineral phases at the solidus of the system CMAS are particularly desirable in order to understand how melt compositions change with pressure (i.e., the depth interval over which melting occurs), and to constrain the thermodynamic properties of silicate melts. However, the determination of such compositions is difficult experimentally, as the five phases (melt plus four crystalline phases) constitute an isobarically invariant assemblage in the component CMAS system, and therefore only coexist over an infinitely narrow temperature interval.

We have determined the solidus of simplified spinel lherzolite in the system CMAS at 11 kbar using three different experimental methods: 1. Normal forward partial melting experiments; 2. Sandwich forward partial melting experiments; 3. Reversal experiments.

Preliminary normal forward partial melting experiments were first undertaken using the typical salt-pyrex assembly. We bracketed the solidus between 1300°C (ol+opx+cpx+sp) and 1310°C (melt+ol+opx+sp). However, the solidus temperature in these preliminary experiments may be slightly low because of contamination by water. We were unable to obtain the full 5 phase assemblage due to the univariant nature of the melting reaction.

In the sandwich experimental technique, a layer of glass was put between two layers of the crystalline assemblage. The aim is to produce a relatively large volume of melt in equilibrium with crystals of the appropriate composition; this should eliminate problems caused by quench modification of the melt composition. The composition of the initial glass and the crystalline mixture containing Fo, Sp, Opx, Cpx came from the literature. In order to ensure near-anhydrous conditions, new salt-pyrex-Fe2O3 sleeve assembly was used.

As the solution to the problem of obtaining all five phases, we conceived the following strategy. Adding a highly incompatible component to the system (i.e., one that enters only the melt phase) turns the system into a divariant one, allowing the four crystalline phases plus melt to coexist over a finite temperature interval. The amount of the incompatible component and the temperature (the two are completely correlated) are varied, and the results so obtained are extrapolated to zero concentration of the incompatible component.

For the incompatible component we chose K2O, since this component can be important in other contexts in igneous petrogenesis; consequently, the information obtained on the effects of K2O on the melting process may prove useful, although it is but a by-product of our main purpose.

Two initial melt compositions were synthesised, with 1% or 3% K2O. Extrapolating the K2O content of the melt in those experiments displaying Fo+Sp+Opx+Cpx+Melt back to zero K2O gives a solidus temperature for isobarically invariant melting in the system CMAS at 11 kbar of 1320°C. The same principle allows a very accurate determination of the melt composition at the solidus.

The reversal experiments, using the new assembly arrangement, were performed with a glass of the composition derived from the sandwich forward partial melting experiments. The experiment at 1330°C displayed only glass, that at 1320°C and 1310°C displayed Melt+Opx+Cpx+Sp. Forsterite is in a reaction relationship with melt and is not expected to be observed in such reversal experiments. The compositions of the phases present in the experiment at 1320°C closely match those observed in the forward partial melting experiments.