Spectral Evidence for a Brucite-carbonate Alteration Assemblage on Ceres

Introduction: The dwarf planet Ceres is the largest object in the asteroid belt and is a primary target of NASA’s Dawn mission [1]. Recent modeling has shown that Ceres is likely differentiated and consists of a thin silicate+ice crust underlain by a thicker water ice mantle [2,3]. The presence of hydrated phases on Ceres, as inferred from the 3 μm absorption feature present in surface reflectance spectra, was discovered ~30 years ago. However, the specific hydrated phases, their origin, and links between Ceres and carbonaceous chondrites are still of considerable debate. Specifically, the presence of a narrow absorption feature at 3.06 μm has been attributed to water ice, clays, ammoniated clays, and cronstedtite, to name a few [4-7]. In this work we show that this feature is best matched by brucite, Mg(OH)2, and that the surface of Ceres is best explained by a brucite-Mg carbonate alteration assemblage. Methods: Laboratory spectra of carbonate (e.g., magnesite, dolomite), cronstedtite, brucite, and saponite powders were measured using an FTIR spectrometer (Fig. 1). The brucite spectrum is of a reagent grade powder, the ammoniated saponite is from the USGS database, the cronstedtite is from the Caltech mineral collection, and the carbonates are from the CRISM online spectral library. Brucite reacts readily with CO2 to produce magnesite, thus it is nearly impossible to obtain a pure powdered brucite sample under normal atmospheric conditions [8]. Indeed, the spectrum of our brucite sample exhibits trace amounts of magnesite as visible by the weak carbonate bands at 3.4 and 3.9 μm (Fig. 1). Variations in particle size between samples can strongly affect absorption strength and albedo at nearinfrared (NIR) wavelengths, so the lab spectra were scaled to 1 prior to use in our NIR model. Differences in particle size have a weaker effect at longer wavelengths, thus we did not scale the lab spectra when modeling the Ceres data past 5 μm. We performed a simple linear mixing model using the reflectance spectra of these particulate samples as inputs, also including positive and negatively sloped lines to account for continuum slopes and to adjust band strengths when modeling the NIR data. The Ceres spectrum was acquired with the SpeX instrument at NASA’s Infrared Telescope Facility on Mauna Kea and is discussed in greater detail by [7]. Emissivity data from 5-13 μm were acquired by the Kuiper Airborne Observatory [9] and were converted to reflectance using Kirchoff’s Law prior to use in the model. Results: Although linear mixing models based on reflectance spectra do not account for complex multiple scattering effects in intimate mixtures, our results show that this simple method produces excellent fits to the Ceres data from 2.85 – 4.1 μm (Fig. 2). All major spectral features can be modeled by a mixture of brucite and carbonate. Although the absorptions at ~3.4 and ~3.9-4 μm have previously been modeled with carbonates [7], to our knowledge this is the first time the feature at 3.06 μm has been modeled with brucite. The carbonate features are fit best by the Mg-bearing varieties (as opposed to Fe-bearing siderite), consistent with the types of carbonates found in some carbonaceous chondrites. The presence of ammoniated saponite [6] is not required to fit the 3.06 μm feature. Although the 3.06 μm feature can be modeled by NH4 saponite when isolated [6], this phase does not produce a reasonable fit when brucite is excluded and the full wavelength range is modeled (Fig. 3). Therefore, all features in the Ceres specrum can be explained by the presence of OHbearing materials; there is no clear evidence for H2O-bearing phases. However, we note that including phyllosilicates such as cronstedtite or smectites in the model improves the fit at wavelengths shorter than 3.0 μm (Fig. 2) due to their increased reflectance in this region. Our model results for the 5-13 μm region are significant improvements over previous attempts [7,9], possibly because the particulate samples used in this study better approximate the particle size distribution on the surface of Ceres. Almost all major absorptions can be explained by the presence of carbonate and, to a lesser extent, cronstedtite. Pure brucite lacks strong diagnostic features at these wavelengths, but the minor amounts of carbonate in our sample produce a slight improvement in the fit when included in the model.