Experiments on boron-bearing systems are made difficult by the reactivity and diffusivity of boron. Boron is lost through or to most capsule materials that are conventionally used in experimental petrologic studies (e.g. noble metals and, particularly, graphite). The loss of boron is so acute in many of our experiments conducted in noble metal capsules that equilibrium was almost certainly not maintained. Boron apparently does not react with tin and nor does it diffuse through a Sn capsule within a reasonable time-frame. A customised Sn crucible is constructed and loaded with boron-bearing hydrous assemblage. A cap is placed on the crucible and the capsule, enclosed in NaCl, is inserted into a piston-cylinder assembly. The apparatus is pressurised without any heating; this serves to cold-weld the cap to the crucible so the capsule does not need to be welded before applying pressure. Although the capsules deform at run pressures and temperatures, they appear to remain sealed throughout the run. Furthermore, the capsules are strong enough that they may be opened and run products removed quantitatively.
The chief problem with Sn as a capsule material is its low thermal stability. At 1 atm, Sn melts at 232°C. We are restricted to using Sn capsules for synthesising boron-bearing starting materials, determining solubility of boron in phases, etc. We have also experimented with use of tin oxide as a barrier to migration of boron. Tin oxide, while not a suitable capsule material itself can be applied as a coating to the inside of a capsule. It has much grater thermal stability than Sn metal (melting point
= 1630°C). it is, however, difficult to produce a continuous coating of tin oxide inside a sealed capsule and there is some loss of boron from the experimental charge. Experiments have been performed in two steps. Initial synthesis of an equilibrium boron-bearing assemblage is attempted by conducting an experiment for several to many days in a Sn capsule. Excess water is added to these starting materials. The solid assemblage is removed after the experiment is completed and reloaded into a capsule lined with tin oxide for a shorter-duration, higher-temperature experiment. The two-step process can be avoided if a continuous layer of tin oxide can be reliably created. We are currently experimenting on the best ways to produce an appropriate capsule.