The partitioning behaviour of trace metals between melts and H-O-C-Cl fluids: an experimental study

Models of volcanic processes and hydrothermal ore formation are increasingly reliant on the availability of data describing how trace elements partition between silicate melts and a volatile phase (including homogenous H-O-C fluids or coexisting vapours and dense saline brines). Since any partition coefficients will vary as a function of pressure (P), temperature (T), and the composition of both the melt and the volatile phase, then expanding the availability of such data should unlock a wealth of information for any volcanic or hydrothermal system for which one can acquire trace metal analyses. Here we present partitioning data acquired using a new experimental and analytical technique. The experimental setup is a variation on the fluid trap experiments of Stalder (1998) and Kessel (2004), in which a volume of vitreous carbon, corundum or quartz glass spheres form a pore space into which the fluids may migrate while maintaining constant communication with the melt. The pore space also performs the vital role of evenly distributing the precipitation of any solute load during quench. Fluids are introduced as free water, plus or minus NaCl, while CO2 is produced via the equilibrium reaction with the substrate in the vitreous carbon bearing experiments. Experiments are run at conditions of 100-300 MPa and 800°C using a rapid quench cold seal pressure vessel apparatus. Analysis is performed by Laser Ablation ICP-MS, using a thermo-electric cryo-stage to freeze the fluid sample so that it may be analysed as a solid. Access to the fluid is achieved by using the laser itself to cut through the capsule, thereby avoiding contamination or loss of the capsule’s contents during opening; the first communication with the outside world is to be directly ablated by the laser and passed into the mass spectrometer. A series of standards are prepared and analysed in the same manner as the experimental charge, and quantification is achieved using a five-point calibration curve. Results for Li show that partitioning is strongly dependent on salinity and pressure, with a sharp increase in partitioning into the fluid with increasing chloride concentration and decreasing pressure. CO2 content causes Li to partition more strongly into the melt than would be expected from previous experimental studies (e.g. Candela and Piccoli, 1995). We will also present new data on Cu, Ni and Zn partitioning using the same techniques. The application of simple models show how the data may be used to describe the behaviour of trace elements in natural volcanic systems, with a focus on the 1980-81 eruption of Mt Saint Helens.