Mercury Again, This Time the Crustal Composition

From MESSENGER's third flyby. NASA 

First, I’ll now use “Mercurian” rather than “Mercutian”, though I find both in the literature. The papers I’ve preferred use the former. Second, in addition to the special 2008 Science issue, there is an excellent special issue of Earth and Planetary Science Letters. Both contain papers based on the first MESSENGER flyby and both links are abstract-only. I’ll continue to link to limited content like that because it still offers an overview and a place to start if the reader wishes to follow up on a topic.

I want to cover one of the results from the third flyby I referenced in yesterday’s post, and that is the high abundance of iron and titanium found on the Mercurian surface by MESSENGER’s neutron spectrometer. This was noticed in the first flyby in 2008, but now it’s supported by new data. This is significant because:

(1) On the whole, Mercury is Fe-rich – it has the highest density of any planet. The Fe-rich core was once thought to be ~80% of Mercury’s mass but the discovery that it has a liquid outer core has led to a reworking of the models. Regardless, it’s still damn big and holds a lot of iron, BUT

(2) Spectral reflectance data of the Mercurian crust indicate low-Fe silicate minerals predominate. So where are the Fe and Ti?

For Mercury, Mariner images and ground-based reflectance spectroscopy suggested that the surface was composed of Fe-poor silicate minerals. MESSENGER data, as discussed by McClintock et al. (2008), Robinson et al. (2008), and Blewett et al. (2009), improved on this, showing that the bulk of Mercurian crust is likely composed of basaltic/gabbroic material rich in plagioclase feldspar (which never contains much iron), Mg-rich pyroxenes with less than 2-3 weight% FeO, and a variable amount of Fe- and Ti-bearing “opaque” minerals (oxides, sulfides, and native metals). Pyroxenes (and other mafic silicates like olivine) can have variable Mg/Fe and the spectra indicate that Mercurian pyroxenes have very low Fe. There are complications to modeling the abundances of opaques from reflectance spectra, and there are space-weathering effects (from inundation by micrometeorites and the solar wind) which cause secondary formation of microscopic or even nano-phase metallic Fe, but by observing fresh and old surfaces you can partly account for this. In short: almost all of the Fe (and Ti) must be in opaque phases and not the silicates; though different surface units could be distinguished based on relative amounts of opaques, the absolute abundances were unclear. Based on reflectance, iron was thought to be, on average, quite low (see McClintock et al., 2008).

But then come the data from MESSENGER’s neutron spectrometer, which provides a measurement of elemental Fe and Ti rather than detecting minerals like the reflectance spectra, and tells the mission team that the surface of Mercury is quite rich in Fe and Ti. The content from the press briefing was presented by David Lawrence and can be found as Presentation #3 here. The Mercury neutron signatures are best fit by a model composition similar to the Luna 16 sample (FeO = 16.3 wt.%, TiO₂ = 3.4 wt.%) or the Apollo 11 high-Ti regolith (FeO = 15.8 wt.%, TiO₂ = 7.5 wt.%) from the Moon. For comparison, terrestrial mid-ocean ridge basalts have ~15% FeO and <1% TiO₂. (Lunar data are compiled in New Views of the Moon.)

Data from MESSENGER's neutron spectrometer plotted against models for varying amounts of Fe and Ti. From the 1st and 3rd flyby. The high or highest Fe and Ti models best fit the data. A more complete explanation can be found in David Lawrence's presentation material, here.

The Apollo 11 regolith contains 14% ilmenite (FeTiO₃) by volume but its silicates contain much more Fe than the Mercurian silicates seem to, so even if the chemistry is superficially similar, the Mercurian crust is a different beast. Assuming (a) these data are correct, which they seem to be, and (b) the mineralogy determined by reflectance is broadly correct, then this calls for some remodeling of Mercury’s internal processes. It’s possible to get that much Ti into a magma (Apollo 11 high-Ti basalt is thought to form when a mantle melt incorporates an ilmenite cumulate source in the lower Lunar crust or upper mantle), but you need to have some significant control from oxygen fugacity and maybe some weird starting compositions to get that much Fe-Ti oxide crystallizing with such Fe-poor silicates. You can concentrate opaque minerals by crystal settling/density stratification in cumulates, but then you need a mechanism to mix them with low-Fe silicates. I’m only now wading into the literature on Mercury’s crustal formation and I found a recent paper by Brown and Elkins-Tanton that has some good ideas and provides a nice review of the models, but I’m not sure any of them can satisfactorily reconcile these data. It's exciting.