11:45 AM - 12:00 PM
[PPS03-10] Ryugu Sample Analysis by MicrOmega: Development and Application of the Fitting Method for Asymmetric Absorption Bands
Keywords:Hayabusa2, MicrOmega, spectrum analysis, infrared spectroscopy, hydroxy group, programming
Introduction: Hayabusa2 returned 5.4g of samples from the C-type asteroid Ryugu in 2020, and their initial description has been conducted at Extraterrestrial Sample Curation Center of JAXA [1][2]. As part of the initial description, the spectra measured by MicrOmega [3] have been analyzed, but the fitting method for asymmetric absorption bands has not been established. This study is to develop and apply the fitting method.
MicrOmega installed at the curation center for the initial description is the NIR hyperspectral microscope developed by IAS and almost identical to the model mounted on the small lander MASCOT to measure mineralogical and chemical composition of the surface of Ryugu. It takes microscopic images over an area of 5 mm square with 22.5 µm/pixel for the wavelength of 0.99-3.65µm [3]. A deep absorption band at 2.7µm has been shown as a common feature for Ryugu samples [2]. The 2.7µm asymmetric absorption band is thought to originate from the OH groups of hydrous minerals but is likely composed of multiple absorption bands.
Methods: In this study, a baseline estimation and the fitting with four Gaussian functions were applied to the analysis of the asymmetric absorption band at 2.7µm. Using multiple Gaussian functions was chosen since the sample is composed of multiple minerals and the major minerals have multiple energy levels.
Asymmetric least squares smoother was used for the baseline estimation and applied to 2.5-3.3µm range [4].
Then, four Gaussian functions were fitted to the spectra after baseline removal. In the fitting, initial values of the parameters (peak wavelength, depth, and FWHM of each Gaussian function) were set and the optimal parameters were obtained to minimize the sum of squared residuals with the measured data. In addition, the relationship was examined between the parameters obtained from the fitting analysis.
We used 139 spectral data (79 from Room A and 60 from Room C) archived in the DARTS Server.
Results: Fig.1 shows the Gaussian functions, composite waveforms, and residuals obtained from the fitting of the spectrum of Ryugu sample A0033. The Gaussian functions are designated as f1 to f4 from the shorter wavelength (the peak wavelengths for A0033 are f1:2.712µm, f2:2.752µm, f3:2.858µm, and f4:3.050µm).
Discussion: Even a single OH group is known to split into multiple absorption bands based on the relationship between the crystal axis and incident light [5]. The relationship between the peak wavelength f1 and the constant depth ratio of f1 to f2 indicates that f1 and f2 likely correspond to the absorption of the related OH groups but with different vibrational energies at different crystal axes of the same crystal. On the other hand, f3 and f4 do not show a similar trend and may be derived from different functional groups. In particular, f4 is not likely to originate from an OH group, since the absorption band around 3.1 µm is considered to originate from an NH group [2].
In the relationship between peak wavelength and depth of f1, depth tends to decrease as the peak wavelength shifts to longer wavelengths, confirming that the Room A sample is divided into two groups: a group with a longer peak wavelength and smaller depth and the other group with a shorter peak wavelength and larger depth. A similar relationship was observed for the 2.7 µm total absorption peak wavelength and depth [6].
Summary: The spectral analysis method used in this study enables more accurate fitting of asymmetric absorption bands and advances the physical and quantitative discussion, which should lead to a reconsideration of the nature of the future fitting analysis.
References: [1]Yada, T. et al. (2022) Nat. Astron. 6, 214, [2]Pilorget, C. et al. (2022) Nat. Astron. 6, 221, [3]Bibring, J.-P. et al. (2017) SSR 208, 401, [4]Eilers, P.H.C. & Boelens, H.F.M. (2005) Leiden Univ. Med. Cent. Rep., 1, 5, [5]Shoval, S. et al. (2001) Opt. Mat. 16, 301, [6]Le Pivert-Jolivet, T. et al. (2022), MetSoc2022, #6255.
MicrOmega installed at the curation center for the initial description is the NIR hyperspectral microscope developed by IAS and almost identical to the model mounted on the small lander MASCOT to measure mineralogical and chemical composition of the surface of Ryugu. It takes microscopic images over an area of 5 mm square with 22.5 µm/pixel for the wavelength of 0.99-3.65µm [3]. A deep absorption band at 2.7µm has been shown as a common feature for Ryugu samples [2]. The 2.7µm asymmetric absorption band is thought to originate from the OH groups of hydrous minerals but is likely composed of multiple absorption bands.
Methods: In this study, a baseline estimation and the fitting with four Gaussian functions were applied to the analysis of the asymmetric absorption band at 2.7µm. Using multiple Gaussian functions was chosen since the sample is composed of multiple minerals and the major minerals have multiple energy levels.
Asymmetric least squares smoother was used for the baseline estimation and applied to 2.5-3.3µm range [4].
Then, four Gaussian functions were fitted to the spectra after baseline removal. In the fitting, initial values of the parameters (peak wavelength, depth, and FWHM of each Gaussian function) were set and the optimal parameters were obtained to minimize the sum of squared residuals with the measured data. In addition, the relationship was examined between the parameters obtained from the fitting analysis.
We used 139 spectral data (79 from Room A and 60 from Room C) archived in the DARTS Server.
Results: Fig.1 shows the Gaussian functions, composite waveforms, and residuals obtained from the fitting of the spectrum of Ryugu sample A0033. The Gaussian functions are designated as f1 to f4 from the shorter wavelength (the peak wavelengths for A0033 are f1:2.712µm, f2:2.752µm, f3:2.858µm, and f4:3.050µm).
Discussion: Even a single OH group is known to split into multiple absorption bands based on the relationship between the crystal axis and incident light [5]. The relationship between the peak wavelength f1 and the constant depth ratio of f1 to f2 indicates that f1 and f2 likely correspond to the absorption of the related OH groups but with different vibrational energies at different crystal axes of the same crystal. On the other hand, f3 and f4 do not show a similar trend and may be derived from different functional groups. In particular, f4 is not likely to originate from an OH group, since the absorption band around 3.1 µm is considered to originate from an NH group [2].
In the relationship between peak wavelength and depth of f1, depth tends to decrease as the peak wavelength shifts to longer wavelengths, confirming that the Room A sample is divided into two groups: a group with a longer peak wavelength and smaller depth and the other group with a shorter peak wavelength and larger depth. A similar relationship was observed for the 2.7 µm total absorption peak wavelength and depth [6].
Summary: The spectral analysis method used in this study enables more accurate fitting of asymmetric absorption bands and advances the physical and quantitative discussion, which should lead to a reconsideration of the nature of the future fitting analysis.
References: [1]Yada, T. et al. (2022) Nat. Astron. 6, 214, [2]Pilorget, C. et al. (2022) Nat. Astron. 6, 221, [3]Bibring, J.-P. et al. (2017) SSR 208, 401, [4]Eilers, P.H.C. & Boelens, H.F.M. (2005) Leiden Univ. Med. Cent. Rep., 1, 5, [5]Shoval, S. et al. (2001) Opt. Mat. 16, 301, [6]Le Pivert-Jolivet, T. et al. (2022), MetSoc2022, #6255.