Halogen Volatility from Silicate Melt

The Earth, the Moon and other bodies in our solar system (e.g., Vesta) are depleted in volatile elements [e.g., 1]. The causes of these depletions, however, are uncertain: did the Earth form as a volatile-free body only to have a volatile-rich component added towards the end of accretion (a kind of late-volatile veneer)[2], or was the process of partial melting and vaporization on proto-planets responsible [3]?

To better understand the process of volatile loss from planetary materials, we measured the volatile behavior of the halogens (F, Cl, and Br) from a silicate melt at fixed temperature and oxygen fugacity. To do this, we performed experiments in a furnace where a crucible of melt was constantly stirred, thereby minimizing the role of diffusion. Calculations suggest the gas species of metallic volatile elements above a silicate melt are generally atomic gases (i.e., Na⁰)[ 4]. Thus, one would expect, and indeed it is observed [3] that the volatilities of these metallic components increase with decreasing oxygen fugacity. For example at 1300°C a one-hour experiment run half a log unit below the NNO buffer, retained ~ 100% and 22% of its original Zn and Cd, respectively; whereas an experiment run at the same temperature and duration, but run at an oxygen fugacity 5.5 log units below the NNO buffer retained only 13% and 2% of its Zn and Cd [3]. We find that non-metallic halogens, on the other hand, are not very volatile and show little dependence on oxygen fugacity. All other conditions being equal, at half a log unit below the NNO buffer, ~ 74% of the Br and 89% of the F were retained; and at an oxygen fugacity 5.5 log units below the NNO buffer ~77% of the Br and 96% of the F remain. This volatile trend is consistent with the inference that the halogens exist as anions in the silicate melt and volatilize to monatomic or dimeric gases, F, Cl₂, etc.

[1]Dreibus et al. (1977) Proc of Lunar Sci Conf 8th:211-227

[2] Schönbächler et al. (2010) Science 328, 884-887.

[3] Norris and Wood (2017) Nature 549, 507-510.

[4] Miguel et al. (2011) Astrophys J Lett 742(2):L19

[5] Clay et al. (2017) Nature 551, 614-618