Lunar and Planetary Science XLVIII (2017) 2127.pdf RADIATIVE TRANSFER MODELING OF LABORATORY DERIVED ILMENITE-SILICATE MIXTURES: INSIGHTS INTO THE CLASSIFICATION OF HIGH-TITANIUM LUNAR BASALTS. K.M. Robertson, S. Li., R.E. Milliken and C. M. Pieters. Dept. of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI, 02912. [email protected] Introduction: The measurement of Fe-Ti oxides in returned lunar samples has helped to refine our understanding of the processes responsible for the formation of mare basalts [1]. Limited sampling however, means that we must rely on orbital remote sensing techniques to understand the global distribution of high-Ti lunar basalts. To date, remote sensing analyses of Ti-bearing minerals have largely focused on an empirical correlation between TiO2 content and spectral slope [2,3] across the UV-VIS range that is not uniquely linked to material properties/mineral physics and can have relative uncertainties >50% [4]. Hapke’s radiative transfer model [5] is commonly used for un-mixing of spectra acquired for regolith on airless planetary surfaces, including the Moon [6]. Such models have the potential to provide quantitative information on Ti-bearing phases on the lunar surface, but important issues persist in using these models for mapping the distribution of Fe-Ti oxides. Ilmenite is the most common Fe-Ti oxide in lunar basalts, and has complex spectral properties that include λ dependent opacity, highly variable albedos across the λ range, and strong textural effects on the mixing systematics [7,8]. These effects must be better understood at the individual level in order to constrain Fe-Ti abundance from remotely-sensed data. Here, we present results of a systematic analysis of laboratory derived ilmenite-silicate mixtures to test the ability of RTM’s to quantify the ilmenite abundance and to evaluate textural effects on the mixing systematics. Fig. 1. Endmember VIS-NIR spectra for the sieved size ranges as measured on the bi-directional spectrometer at RELAB. A small (20-32um) and coarse (75-125 um) ilmenite was mixed with the 4 size fractions of the silicates to test the Hapke model’s ability to model ilmenite with varying particle size. Methods: Mineral endmembers included a synthetic ilmenite sample from Alfa Aesar, San Carlos olivine, Tanzanian enstatite and a gem quality labradorite (Fig. 1). Endmembers were ground and wet sieved to four particle sizes (20-38 μm, 38-45 μm, 63-75 μm, 75-125 μm) to be combined for spectral modeling. The coarse ilmenite (75-125 µm) and small ilmenite (20-38 µm) were mixed (5 wt% increments up to 25 wt%) with all 4 size fractions of each silicate to test the effect of ilmenite particle size on the modeling results. The ternary suite started with 60 wt% pyroxene and 40 wt% plagioclase with ilmenite being added in 5 wt% increments up to 25 wt%. Ternary mixtures used the same particle sizes (38-45 μm or 75-125 μm) for each endmember within a given sample. Samples were measured once on the Bi-directional spectrometer in RELAB at Brown University and 8 times on an ASD FieldSpec3 spectroradiometer using similar geometries (i=30º, e=0º, g=30º) relative to spectralon. Samples were also measured on a Bruker D2 Phaser XRD to validate abundance values. The overall parameterization of the Hapke model used in this study is the same as that of Li and Li (2011) [9] however; it is revised slightly to account for the complex nature of the ilmenite phase. Here we use single scattering albedo (SSA) of mineral endmembers as opposed to optical constants for the un-mixing. The model inputs are, the measured reflectance data, viewing geometry (i, e, g), end-member single scattering albedo and densities. Results: The reflectance spectra for the binary mixtures of pyroxene/olivine (63-75 µm) and ilmenite (20-32 µm) are presented in Fig. 2. There is a dramatic decrease in the band depth of diagnostic features as well as the overall albedo for all silicates with as little as 5% ilmenite added. While not unexpected, these results highlight the significant non-linear effects that opaque phases have on the mixing systematics. Fig. 2. ASD measurements of the pyroxene-ilmenite and olivineilmenite binary mixing suite. In the example given, a size fraction of 75-125 µm was used for pyroxene and olivine and 2032 µm for ilmenite Lunar and Planetary Science XLVIII (2017) 2127.pdf Fig. 5. A) Modeled abundance values for the 38-45μm ternary mixtures. Pyroxene is overestimated at the expense of plagioclase. Modeled spectra for the 5 and 20wt% ilmenite are shown below. Fig. 3. A) Averaged modeled ilmenite abundance values with standard deviation for the ilmenite (20-32 μm) - silicate (20-32 μm) binary mixtures. B) Averaged modeled ilmenite abundance values for ilmenite and pyroxene mixtures spanning the small (20-32 μm) and coarse (75-125 μm) size fractions. Fig. 4. Examples of the modeled (black) and the measured ASD spectra for the ilmenite and silicate mixtures (20-38 μm) from Fig. 3. All spectra are of the 20 wt% ilmenite mixtures. The Hapke un-mixing results of the BDR and ASD data for all size fractions are promising when using the SSA as opposed to optical constants. The values reported here are averages of 8 separate measurements on the ASD instrument with standard deviations. Modeled abundance values for the small ilmenite fraction (20-32 um) and silicates (20-32 um) are reported in Fig. 3a. The average ilmenite abundance across all 8 measurements are within 2% of the measured values while the standard deviations are all within ±5%.These results show little variation between measurements when similar particle sizes are used. The modeled spectra for the 20 wt% ilmenite mixtures (Fig. 4) all have RMS values ~ 10-5. Modeled ilmenite abundances for mixtures with variable particle sizes are shown in Fig. 3b. The average weight fraction estimates for both the small and coarse ilmenite suites are all within 5% of actual values. A larger error is associated with individual measurements however, when the mixtures involve large differences in particle size between endmembers. The variations are not systematic and likely a result of sample preparation effects. Images of the sample mounts suggest that settling of the finer grained particles between the larger grains is occurring and could be causing the observed variation. Particle size effects are a real issue that must be acknowledged in any results pertaining to evaluating opaque phases on the lunar surface. Ternary fits are shown in Fig. 5. These mixtures involved similar grain sizes between the ilmenite and silicates therefore there were no particle size effects to contend with from sample preparation. The abundance estimates for binary plagioclase-pyroxene mixture is accurate (Fig 5); however as ilmenite is added, the pyroxene is overestimated at the expense of plagioclase. These results do not appear to be unique to this work as they were also reported in [10]. Conclusions: The addition of small amounts of ilmenite causes suppression of absorption features, reduced albedo and a ‘red’ spectral slope a longer wavelengths [8]. The plagioclase-ilmenite mixtures of Fig 4 are particularly illuminating where ilmenite absorptions dominate. The RTM modeling of the ternary mixtures resulted in an offset of the plagioclase and pyroxene abundance. A similar result was observed with Apollo 17 basalt samples (70017 and 70035) that we modeled as well as observed in previous studies [10]. The revised RTM is promising for ilmenite-bearing binary mixtures but, results vary when the particle size differences increase. In this case, the variations can be attributed to sample preparation however; significant work is still needed to address the effect of very fine ilmenite phases in a basalt mixture. Specifically, how much of the red continuum is due to the absorption properties of fine-grained ilmenite vs scattering of generic fine-grained opaques (micron and nano phase) in a silicate matrix? Acknowledgments: The authors acknowledge the support of NASA SSWP grant NNS15AR68G. References: [1] Neal and Taylor, (1992) Geochim. Cosmochim. Acta. 79. [2] Gillis et al., (2003) JGR. 108b. [3] Lucey et al., (2000) JGR. 105. [4] Gillis-Davis et al., (2006) Geochim. Cosmochim. Acta.70. [5] Hapke, B., (2005) Cambridge university press. [6] Mustard, J.F. and C.M. Pieters, JGR 94. [7] Riner et al., (2009) Geophys. Res. Let.36. [8] Isaacson et al., (2011) Meteorit. Planet. Sci. 46. [9] Li and Li, (2011) JGR. E09001. [10] Hiroi et al., (2009) LPSC abstract # 1723.
© Copyright 2026 Paperzz