The diamond trap method

The basic idea of the diamond trap method is to fill a portion of the capsule with a fine-grained diamond powder. Since diamond is mostly chemically inert and mechanically strong, this kind of trap provides some pores, preserved at high pressure, which are available for circulation of the fluid only. In this way, equilibration of the fluid with the mineral phases is facilitated as the system remains open for the entire duration of the experiment, unlike for synthetic fluid inclusions. At the same time, an efficient segregation of the fluid is attained since the solid starting material is unlikely to enter and contaminate the diamond trap if some precautions are taken. It is therefore reasonable to assume that in general, solid materials found inside the diamond trap represent fluid components precipitated upon quenching. This, however, may not be true in systems that are particularly sensitive to temperature. In these cases, the temperature gradients that typically develop in high-pressure experiments may lead to dissolution of minerals and their reprecipitation in the diamond trap. Moreover, this phenomenon is likely to be more severe when a metastable starting material is used, as a supersaturated solution that reaches equilibrium by precipitating crystals is expected to form during the early stages of the experiment. As these crystalline phases could also contaminate the diamond trap, they will be

22

erroneously considered to be fluid components, resulting in an overestimation of the solute concentrations in the fluid. This problem may be reduced, if the crystalline mineral assemblage stable at the experimental conditions is used as starting material. However, as discussed above, this would hinder the attainment of chemical equilibrium between fluid and minerals.

The diamond trap technique was first proposed by Ryabchikov et al. (1989). These authors introduced diamond powder, enclosed in a perforated inner capsule, inside an outer capsule also containing the solid starting material and distilled water. A later development of the method was proposed to study the composition of low fractions of partial melting (e.g. Johnson and Kushiro 1992, Kushiro and Hirose 1992, Hirose and Kushiro 1993, Baker and Stolper 1994, Baker et al 1995). In these experiments, to enhance the segregation of small amounts of melt, the diamond powder was inserted directly inside the outer capsule and different geometries were tested. The results from Kushiro and Hirose (1992) showed that one of the best designs is a thin layer of diamond sandwiched at the center of the capsule between the solid starting material. One important observation of this study is that the liquid must completely fill the pores between the diamonds in order to avoid the development of negative pressure gradients in the trap, which would prevent the attainment of equilibrium at the desired experimental conditions. Later, Stalder et al. (1997, 1998) used the diamond trap method combined with LA-ICP-MS for the determination of trace element partition coefficients between aqueous fluids and minerals. After high pressure experiments, the capsules were pierced and dried. Fluid compositions were therefore derived from analysis of the diamond trap portion containing the solid precipitates, assuming that none of the fluid components remained dissolved in the fluid that was lost from the capsule. However, this assumption is not always valid and may introduce large errors in the measured concentrations. This problem was largely solved by Kessel et al (2004), who introduced a new approach for the analysis of the diamond trap. These authors opened the capsules in frozen state and performed measurements of the diamond trap containing both the solid precipitates and the frozen fluid by LA-ICP-MS equipped with a freezing chamber. In this way, the bulk composition of the fluid, as it was at high pressure and temperature during the experiment, can be directly measured. This approach was tested through measurements of quartz solubilities in water by Aerts et al. (2010), who showed that for this system, the accuracy and precision of the method is similar to weight-loss experiments.

23

In the present study, the method developed by Kessel et al (2004) was used for the determination of fluid/eclogite partition coefficients. Moreover, in order to test the reliability of the method, the following three different analytical approaches were compared in simplified experiments conducted at 2 kbar and room temperature loaded with a fluid of known composition. Measurements of the diamond trap layer were performed in such experiments either: (a) after evaporation of the fluid at ambient condition, (b) after freeze-drying the sample, or (c) after opening the capsule in frozen state to directly analyze the fluid in solid state. In addition, mineral solubilities in simple and well-studied systems (quartz-water, forsterite-enstatite-water, albite-water, rutile-water and corundum-water) were determined using the diamond trap technique and compared to literature values in order to assess the reliability of the method. (see Chapter 6 for further details). Moreover, to confirm that equilibrium was attained in the fluid/eclogite partitioning experiments, several reversed experiments were performed. In such experiments, the trace elements were completely doped into the fluid phase instead of the solid starting material.

Figure 2.1. Picture of an experimental charge after a piston cylinder pressure and high-temperature diamond trap experiment. After LA-ICP-MS analysis of the fluid in frozen state, the sample was mounted in epoxy and polished for mineral analyses.

24