Introduction

Carbonaceous asteroids are primitive Solar System small bodies that hold clues to understanding the origin and evolution of the Solar System. A combination of remote sensing of asteroid surfaces and analyses of returned samples from there by sample return missions provide much progress in evolution and metamorphic processes of materials in the early Solar System. Asteroids are classified by their spectral profiles, while their surface materials have been altered by space weathering since their formation1. Spectral changes due to space weathering on the Moon and S-type asteroids (i.e., Itokawa) have been reported, such as reddening in the visible to near-infrared (VNIR) wavelengths, with generation of iron nano (size) particles that were observed in returned samples2 and in the samples reproduced by laboratory experiments3 For C-type asteroids, spectral changes due to space weathering vary depending on the mineralogy of the samples and/or experimental conditions. For example, both bluing and reddening trends in the VNIR spectra were reported by pulsed laser ablation experiments on CM and CI chondrites simulating micro-meteoritic impacts4,5, as well as ion irradiation experiments (He + , Ar +) simulating solar winds6. In this study, we investigated the effects of space weathering remaining in the return samples of the C-type asteroid (162173) Ryugu.

Asteroid Ryugu explored by the JAXA Hayabusa2 mission is classified as a Cb-type asteroid with a flat and featureless spectral shape but with an intense 2.7 µm absorption band represented by OH groups, according to the results by the Telescopic Optical Navigation Camera (ONC-T) and the Near Infrared Spectrometer (NIRS3) onboard the spacecraft7,8,9. Hayabusa2 returned samples of Ryugu to Earth in 2020 after two touchdown (TD) sampling operations: TD1 and TD210,11. During the TD1, samples from the surface were collected at the equatorial ridge of Ryugu and stored into the chamber A of the sample container. During the TD2, the mixture of surface and subsurface samples were collected at the ejecta blanket where the excavated materials were sedimented after the artificial crater experiment using the Small Carry-on Impactor (SCI), and they were stored into the chamber C12. This different sampling location and methods imply that the different degrees of space weathering were expected for the samples in chambers A and C13.

Initial descriptions of these Ryugu samples have been performed under nitrogen-filled conditions at the JAXA’s Curation Center14. One of them is a near-infrared (NIR) hyperspectral microscopy using the MicrOmega, which acquires NIR spectra from 0.99 to 3.65 µm with a spatial resolution of 22.5 µm per pixel non-destructively, non-contact, and without contamination15. Typical spectral features observed by MicrOmega in most Ryugu individual particles of millimeter to centimeter size are a continuum at around 1.4–2.3 µm without significant absorption band, a relatively deep asymmetric absorption band at ~ 2.7 µm, and the other absorption bands at ~ 3.1 µm and ~ 3.4 µm15,16,17. Among these spectral features, the 2.7 µm band has been reported to show a decrease in the depth due to dehydration and a shift of its peak position towards longer wavelengths through space weathering by solar wind implantations and micrometeorite impacts13,18,19. In the previous study13, the peak position and depth of the 2.7 µm bands were investigated by fitting the short wavelength side with a Gaussian function. In this study, we adopted a method to use multiple Gaussian functions to fit the entire band feature more precisely and investigate it in more detail19,20. We investigated the difference between the chambers A and C particles of Ryugu in the trends of the peak position to the depth of the 2.7 µm band , which was not deeply discussed in the previous study, and propose a hypothesis for the various degrees of space weathering occurred on Ryugu. In particular, for the chamber C particles, we performed UV irradiation experiments on a carbonaceous chondrite meteorite and obtained the spectral data of the pre- and post-irradiated meteorite. We used this data to discuss the potential mechanism of the weathering effects on Ryugu by solar UV radiation for three months between the excavation by the SCI impact to the sampling during the TD2.

Results

Figure 1a shows the average NIR spectrum of the Ryugu particles used in this study measured by MicrOmega. The spectra of Ryugu particles commonly exhibit no prominent absorption bands over the 1.4–2.3 µm region (“2.0 µm continuum”, hereafter) and a 2.7 µm band corresponding mainly to OH stretching vibrations bonded to metal cations such as Fe and Mg in phyllosilicates8,9,13,16,21. The 2.7 µm band is asymmetric and is expected to consist of several bands due to H2O15,22,23 and NH groups15,16,24, and the other25, as well as OH groups. Therefore, we developed a fitting method for the entire band using multiple Gaussian functions to fit them precisely and make the detailed discussions possible. Fitting analysis with different numbers of Gaussian functions were performed for the average spectra of 207 Ryugu particles used in this study (119 and 88 particles from chambers A and C, respectively). Based on the derivation of the evaluation parameters: the Akaike Information Criterion (AIC)26 and the Bayesian Information Criterion (BIC)27, and the coefficient of determination R2 of each model, it was determined that the optimal number of Gaussians is 6, as shown in Fig. 1b (see Methods for details). In fact, all 207 particles are well fitted with 6 Gaussian components, which are named f1 to f6 from shorter wavelength. The peak position and the depth of each component and the composite waveform for the average spectra are presented in Table 1.

Fig. 1
figure 1

(a) Average spectra of 207 Ryugu particles used in this study. In common with all particles, there are a continuum around 2.0 (1.4–2.4) µm without prominent absorption bands and a strong absorption band at 2.7 µm originating from OH groups. (b) A result of multiple Gaussian fitting of the 2.7 µm band of an average spectrum (blue dots: measured data, black line: composite waveform, blue polyline (shifted to 2.5%): residual spectra). Individual component Gaussians in color shading: f1 (purple), f2 (blue), f3 (light blue), f4 (green), f5 (yellow), and f6 (orange).

Full size image
Table 1 Peak position and band depth of each Gaussian component and the composite waveform of the 2.7 µm band of the average spectra.
Full size table

Figure 2a shows the relationship between the peak position and the absorption band depth of the composite waveform of the 2.7 µm band for all 207 particles. This reveals that the 2.7 µm band tends to decrease in depth as the peak position shifts towards longer wavelengths, which is similarly found in the previous study using the single Gaussian fitting at the most prominent peak in the 2.7 µm band on Ryugu particles13. As shown in Fig. 2b, we divided this trend into two groups, one with short wavelengths—deep depths (group α) and the other with long wavelengths—shallow depths (group β), using the central black line as a threshold. The particles in each group are named Aα and Aβ, Cα and Cβ for the particles in chambers A and C, respectively. In Fig. 2b, each group is color-coded, and the respective averages are plotted. If the particles in chambers A and C are spectrally identical, the averages of the two groups (Aα and Cα, or Aβ and Cβ), should be almost the same. However, these groups are distributed differently and shift with a trend from deeper toward shallower in depth and from shorter toward longer in wavelength in the order of Aα, C (Cα to Cβ), and Aβ. The group Aα and Aβ are spatially separated in the plot, while group Cα and Cβ are rather continuous, which is also seen in the previous study13. Thus, we combined the group Cα and Cβ into group C, and focused on the difference between the group Aα, Aβ, and C. Here it would worth emphasizing the trend of the group C indicating the intermediate characteristics between the group Aα and Aβ because the group C was expected to consist of non or poorly space weathered particles as same as the group Aα13.

Fig. 2
figure 2

(a) Peak position vs. band depth of the composite waveform of the 2.7 µm band (red dots: chamber A particles, light blue dots: chamber C particles). Error bars show 2 sigma errors on the fitting analysis. The depth tends to be shallower at longer wavelengths. (b) For the trend observed in (a), the central straight black line passing through (2.71, 0) and (2.726, 30) is set as the threshold where the distributions of the particles are divided into two regions: short wavelength with deep depth, and long wavelength with shallow depth. Particles in each region are color-coded and named: long wavelength—deep depth region is named α (Aα: orange, Cα: blue), short wavelength—shallow depth region is named β (Aβ: pink, Cβ: dark blue). The mean value and the standard error of the peak positions for each group is Aα: 2.7143 ± 0.0001 µm, Aβ: 2.7193 ± 0.0003 µm, Cα: 2.7150 ± 0.0001 µm, and Cβ: 2.7186 ± 0.0004 µm. The depth is Aα: 18.0 ± 0.4%, Aβ: 10.9 ± 0.3%, Cα: 15.2 ± 0.3%, and Cβ: 12.1 ± 0.4%. There is a trend that the longer the peak position is, the shallower the depth is, in the groups from Aα to C (Cα to Cβ) to Aβ in the diagram, indicating that the chamber C particles mostly overlap with the Aα group, while partially extending toward the Aβ group.

Full size image

We also calculated the slopes of the 2.0 µm continuum (1.4–2.3 µm) derived by fitting with a linear function for the three groups of Aα, Aβ, and C, which are 5.82 ± 0.16, 5.57 ± 0.23, and 5.57 ± 0.18 in %/cm-1 × 10–5, respectively, where the mean values of the slope and the standard error are shown. Although they are overwrapped within the standard errors, it may indicate the tendency that the slopes of Aβ and C are gentler than that of Aα (Aα > C ≈ Aβ).

Figure 3 shows the relationship between the peak position of the composite waveform (mainly component waveform f1) and the depth ratio f1/f2 of the 2.7 µm band. Particles in groups Aα and C show a trend that f2 depth decreases as the peak position shifts longward. The group C tends to have a larger depth ratio of f1/f2 than the group Aα. On the other hand, particles in groups C and Aβ show a trend that f1 depth decreases as the peak position shifts longward. The group Aβ tends to have a smaller depth ratio of f1/f2 than the group C.

Fig. 3
figure 3

Peak positions of composite waveform vs. depth ratios of f1 and f2 (f1/f2) of the 2.7 µm bands (Orange dots: Aα in the chamber A particles, pink dots: Aβ in the chamber A particles, light blue dots: the chamber C particles in Fig. 2). The averaged mean values and the standard errors of the depth ratios (f1/f2) for each group are Aα: 1.43 ± 0.09, Aβ: 1.35 ± 0 .10, and C: 1.63 ± 0.09. These three averaged values suggest a preferential decreasing trend of f2 from Aα to C and f1 from C to Aβ, as indicated by the arrows.

Full size image

In terms of the spectral feature of group C indicating intermediate trend between group Aα and Aβ, we proposed a hypothesis that the particles of group C experienced the short-term (several month) space weathering because a large portion of the chamber C particles was expected to be exposed to solar UV radiation for three months from the excavation by the SCI impact to the sampling during the TD2. To examine whether the three months of solar UV radiation on Ryugu could have affected the NIR spectral properties, we conducted the UV irradiation experiment on a Ivuna CI chondrite under a nitrogen environment at ISAS, JAXA. The sample we selected was a chip of Ivuna CI chondrite meteorite which is most similar in composition to the Ryugu samples. The sample with the size of 1.5 mm and the mass of 2.5 mg was fixed on an aluminum stage with carbon tape and installed in the hermetic container with a UV transparent quartz window called FFTC (Facility to Facility Transfer Container)28. The sample was irradiated with UV light of 250–385 nm in wavelength inside a nitrogen-filled FFTC using a Xenon light source (MAX-303, Asahi Spectra), with a flux of 2.3 × 103 mW/cm2 for 1 h, which is equivalent to 258.4 h of solar UV exposure at 1 AU29. Pre- and post-irradiation spectra were measured using a microscopic FTIR spectrometer (FT/IR-6100, JASCO) with 500 µm × 500 µm aperture, 1.28–28.6 µm wavelength range, and 2 cm-1 sampling resolution in a nitrogen purged glove box. All procedures were conducted under a nitrogen condition to prevent atmospheric contamination (see the Method section for details).

Figure 4 and Table 2 show the comparison of NIR spectra of Ivuna before and after the UV irradiation. The broad absorption band centered at 2.8 µm in the spectra before the UV irradiation indicates the presence of adsorbed water originating from terrestrial contamination. The reduction of the 2.8 μm broad band was observed after the UV irradiation. The 2.7 µm sharp band (2.70–2.73 µm), typically observed in Ryugu particles, is expected to originate from the OH groups (~ 2.719–2.759 µm15,23,30) without the adsorbed water (2.8–2.9 µm, 3.1 µm15,22,23). Focusing on the 2.7 µm sharp band, a decrease in depth by ~ 3% and a longward shift by ~ 1 nm were observed after the UV irradiation (see Table 2, Supplementary Figure-S2, and Table-S1). The slope around the 2.7 µm band which is the slope between the reflectance at 2.50 µm and 4.00 µm also decreases by the UV irradiation, indicating a tendency toward bluing. These results suggest that the 2.7 µm sharp band could be changed by UV irradiation in a much shorter timescale (~ 10 days at 1 AU) than conventionally recognized timescales for space weathering such as ~ 104–106 years1,31 for solar winds and ~ 108–109 years3 for impacts of micrometeorites. Discussed below are the space weathering processes which occurred to the chamber C particles due to solar UV radiation.

Fig. 4
figure 4

(Top) NIR spectral comparison before and after 1 h UV irradiation on a chip of Ivuna meteorite (blue line: before irradiation, orange line: after 1 h UV irradiation). These reflectance spectra are normalized at 2.5 µm. The area at around the 2.7 µm absorption band is enlarged in the bottom right figure. (Bottom) The difference of the NIR spectra before and after 1 h UV irradiation on Ivuna (red line). For the peak positions and the band depths of the component Gaussians, see Supplementary Figure-S2. Adsorbed water due to the terrestrial environment was removed by the UV irradiation.

Full size image
Table 2 Peak position and band depth of the 2.7 µm sharp band (2.70–2.73 µm) and the spectral slope around the 2.7 µm band for before and after 1 h UV irradiation on Ivuna.
Full size table

Discussion

We discuss about the 2.7 µm absorption band of the average spectrum of all the 207 Ryugu particles used in this study, which consists of 6 components of Gaussians f1 to f6 shown in Fig. 1b and Table 1. The shorter components correspond to OH stretching vibrations bonded to metal cations such as Fe and Mg in phyllosilicates8,9,13,16,21. The peak position depends on the molar ratio of Mg/(Mg + Fe). Specifically, Mg3-OH is located at ~ 2.719 µm, Mg2Fe-OH at ~ 2.730 µm, MgFe2-OH at ~ 2.743 µm, and Fe3-OH at ~ 2.759 µm, which are located at longer wavelengths with higher Fe content15,23,30. Thus, f1 possibly corresponds to Mg3-OH (or mixed with Mg2Fe-OH), and f2 possibly corresponds to Mg2Fe-OH, MgFe2-OH, and Fe3-OH, and f1 is considered to have a larger Mg/(Fe + Mg) ratio than f2. There are several candidates of the origins for f3-f5. The first is the adsorbed water (H2O). Typical water absorption bands are the fundamental symmetric (ν1) and asymmetric (ν3) stretches at ~ 2.8–2.9 µm, and the first overtone (2ν2) at ~ 3.1 µm15,22,23. The next candidate is the carbonyl group (C = O), which is at ~ 2.80–3.0 µm15,32. The possibility of other organics also cannot be ruled out. For f6, there is an absorption band of NH4-bearing phyllosilicates at ~ 3.06 µm. F6 is considered to originate from the NH group, and possibly from H2O as well15,16,24. F6 may be further split into three or more absorption bands given that the shape of this band is asymmetrical. However, because the shape of the 3.1 µm band varies from particle to particle, the result of the fitting with a single Gaussian function was adopted in this study.

Similarly found in Ryugu returned samples in this study, the relatively deep asymmetric absorption band at 2.72 µm on the Ryugu is also observed by NIRS38,9. NIRS3 observed the inside, the surrounding area, and the far outside of the SCI artificial crater and found that the 2.7 μm band was deeper as more subsurface materials were thought to be present33. In addition, as a common spectral behavior for all crater floors, the 2.7 µm band depth is slightly deeper than that for the surface34. This suggests that, after the asteroid Ryugu was formed, space weathering and/or heating by solar flux could have caused the decrease of surface water content. It has been reported that solar flux heats and dehydrates 10–100 cm thickness of the Ryugu surface layer10. However, no evidence of intense heating or internal thermal metamorphism for returned samples has been found35,36,37. Therefore, space weathering is considered to have played a major role in the dehydration process on the Ryugu surface layer18. In fact, observations of the Ryugu returned samples have shown that space weathering caused dehydration and reduction of the 2.7 µm band depth18. In addition, it was suggested that the 2.7 µm band is shifted longward by space weathering. As reported in Hiroi et al. (2023), by comparing the 2.7 µm band of the chamber C particles in the laboratory and the Ryugu surface by NIRS3, the peak position on the Ryugu surface was shifted longward by 6 nm due to solar wind and/or micrometeorite effects, compared to inside the SCI artificial crater19. Le Pivert-Jolivet et al. (2023) also suggested a longward shift due to space weathering (solar wind and micrometeorites) by comparing the space-weathering-rich area and space-weathering-poor area in particle A0017 based on surface morphology13.

As in the previous studies, longward shift of peak position in the 2.7 µm band is most likely the effect of space weathering. On the other hand, there is a report that grain size affect to the absorption band depth. Cantillo et al. (2023) measured the UV to mid-infrared (0.2–14 µm) reflectance spectra of seven carbonaceous chondrites to investigate the effects of grain size on their spectral features. They used powdered samples with different grain sizes (45–1000 µm) and observed a trend of decreasing absorption band depth (~ 1 µm) with increasing grain size. The reduced band depth in larger grains was interpreted as less scattering of light caused by the decreased length of light path in the samples. As the samples used in this study are mm sized single particles, absorption band depth would be proportional to the path length of light passing through the particle, which would mainly be constrained by the thickness of the particles38. As shown in Supplementary Figure-S3, there is no significant relationship between the grain size and the absorption band depth, grain size would not affect to the spectral feature (band depth at 2.7 µm) in this study. Consequently, the trend of longward shift of peak position and reduction of the 2.7 µm absorption band depth in order of Aα, C, and Aβ is considered to reflect the degree of dehydration due to space weathering.

The spectral slope (reflectance ratio) may provide clues to the effects of space weathering. NIRS3 observation shows that the reflectance ratio at 2.6 µm to 2.2 µm (2.6 µm / 2.2 µm) is smaller for “Surface” than “Subsurface”39. “Surface” is thought to be more affected by space weathering than “Subsurface” except for the gardening effects. Majority of the Aα particles with a gentler slope around 2.0 µm may be more affected by space weathering than the C and Aβ particles. In addition to space weathering, there may also be effects of different porosity, grain size, and roughness on the spectral data. As for the nearly identical slope of C and Aβ, this suggests that the change of the spectral slope may be saturated at an early stage. However, the spectral slope can be affected by other factors such as grain size, porosity, and geometry of observation. Therefore, it is difficult to determine the degree of space weathering effects based on the spectral slope only.

It is known that the vibrational frequency correlates with the strength of the bond, as expressed in the following equation,

$$\nu = \frac{1}{2\pi c}\sqrt{\frac{k}{m}}$$

where ν is the vibrational frequency, k is the strength of the bond, m is the converted mass, and c is the velocity of the light. Figure 1b shows that the peak position of f2 is at longer wavelength than that of f1, so that the OH bond corresponding to f2 (f2 OH bond) is considered to be more easily broken than f1 OH bond by low energy irradiation. Both bands are considered to be broken for high energy processes involving morphological changes, such as melting by micrometeorite impacts and solar winds. Figure 3 shows that the f2 OH bond is preferentially broken (dehydroxylated) in the group C compared to the groups Aα and Aβ. The groups Aα and Aβ showed almost the same f1/f2 ratio. This might imply that short-term weak space weathering by UV radiation preferencially broke f2 OH bond in the group C particles, whereas long-term strong space weathering by micrometeorite impacts and solar winds broke both f1 and f2 OH bonds significantly.

As for the main types of space weathering effects by micrometeorite impacts, solar winds, and solar UV rays40, solar winds and micrometeorite impacts require ~ 104–106 years1,31 and ~ 108–109 years 3, respectively, which are much longer than solar UV rays of < 10 years41 (or might be much shorter as suggested here). Micrometeorite impacts and solar winds melt the uppermost surface of samples and distort the crystal structure39. In the experiment of micrometeorite impacts, a laser irradiation causes the molten surface layer “Frothy layer”4,5,42,43. The Frothy layers are also found in Ryugu samples18,44, which have a smaller Mg/(Fe + Mg) ratio than phyllosilicates inside the particle18. In the Ryugu sample A0017, the position of the 2.7 µm band in the “SpWe-rich” area is at longer wavelength, and the depth of the 2.74–2.8 µm region (considered to correspond to f2 in this study) is slightly deeper, compared to the “SpWe-poor” area13. Considering that the f1 and f2 OH bonds mainly corresponds to Mg-OH and Fe-OH, respectively, majority of the Aβ particles are considered to be the particles that contains components with less Mg-OH/(Fe-OH + Mg-OH) and its 2.7 µm band shift towards longer wavelengths due to micrometeorite impacts and solar winds irradiation compared to majority of Aα and C particles (Fig. 5).

Fig. 5
figure 5

A proposed weathering process on the chamber C particles (blue), explaining the potential cause of the majority of the chamber C particles having an intermediate characteristic between the Aα (orange) and Aβ (pink) particles. This hypothesis is that majority of the chamber C particles are Aα particles ejected by the SCI impact and have been slightly affected by space weathering such as solar UV rays until they were sedimented at the TD2 site and sampled from there during the TD212,18,41,45. Majority of Aβ particles are thought to have been present in the uppermost surface layer of Ryugu, which may have been affected by micrometeorite impacts and solar winds. Majority of Aα particles are thought to have been present in the subsurface layer or the inside of boulders crashed during the sampling, which may have been almost unaffected by micrometeorites and solar winds. This figure is a simplified layered model where the layers are represented by Aα and Aβ. The wavy line is used to represent the physical distance between TD1 and TD2, indicating that the details in between are omitted.

Full size image

Considering that the heterogeneity in spectral feature at 2.7 µm absorption band in Ryugu particles is derived from the degree of space weathering, the effect of space weathering is stronger in order of particles grouped in Aα, C, and Aβ. On the asteroid Ryugu, space weathering such as micrometeorite impacts and Solar winds affected only top surface of the asteroid and even on the snigle particles (micrometeorite impacts: ~ 103 nm, solar winds: < 102 nm)18. Thus, difference between the group Aα and Aβ can be interpleted by the presence of space weathering as they were collecleted from the surface of asteroid Ryugu, and this trend is consistent with the previous study13. Alternatively, even for a single particle, the effects of micrometeorite impacts and solar winds are limited to the surface of the particle only18. Therefore, the difference between group Aα and Aβ may result from the surface of the particle being stripped away due to crushing during sample collection and/or return to Earth13. On the other hand, part of the group C particles show intermediate characteristics between the group Aα and Aβ, and they seemed to be experienced weak space weathering. As the particles in goup C were collected at TD2 sampling site, ejecta blanket nearby the artificial crater, they are mixture of surface and subsurface (~ 1 m) materials. It is expected that most of the group C particles were not affected by space weathering like the group Aα and particle with space weathering, if present, show same characteristics with the group Aβ. Actually, the chamber C particles do not contain much solar wind noble gases37. Therefore, the presence of the group C with weak space weathering could not be explained by the space weathering originally occurred on the Ryugu asteroid, and we propoed another senario as follows.

After cratering by the SCI impact, the excavated Ryugu surface particles were slowly ejected and exposed to space for several hours to several days until they were sedimented to the nearby area of the crater due to a microgravity environment. These particles were exposed by irradiation of solar UV for that period and for a maximum of 3 months from the excavation by the SCI impact in April 2019 to the sampling during the TD2 in July 2019, which may cause the slight dehydration and peak shift of OH absorption band at 2.7 µm. The effect of UV irradiation on olivine and Murchison (CM2) meteorite was investigated in the previous studies29,41. They observed the change in visible spectra by UV irradiation equivalent to that for 2–3 years with the Sun at distance of 1 AU. Our preliminary experiment of UV irradiation corresponding to ~ 10 days of Solar UV irradiation at 1 AU (on the surface of the asteroid Ryugu (~ 1.3 AU), this timescale corresponding to ~ 18 days (436.7 h)) on Ivuna meteorite showed ~ 3% decrease of 2.7 µm absorption band depth and longward shift of its peak position by 1 nm. This trend seems to be consistent with the difference between the groups Aα and C in the Ryugu particles. On the other hand, the effect of temperature increase due to the UV irradiation in the experiment is estimated to be several tens of degrees Celsius, which should be considered in future experiments simulating solar UV irradiation. In particular, this temperature increase could lead to dehydration of the sample, meaning that the observed changes might not be solely attributed to UV-induced chemical alterations. Therefore, careful evaluation of the thermal effect is necessary in future studies. Weak space weathering may also include solar heating (possibly affecting the whole particle) in addition to solar UV rays. Additionally, since the Ivuna sample used in this study contained adsorbed water and suffered terrestrial alterations unlike the Ryugu samples, and the experiment was conducted under nitrogen atmosphere, although vacuum conditions would be ideal, further experiments will continue to be conducted to improve reproducibility and representativeness.

The Ryugu return samples contain particles exhibiting varying degrees of space weathering, as inferred from the relationship between the depth and peak position of the 2.7 µm absorption band. In particular, the particles from chamber A display a distinct bimodal distribution, with particles (Aβ) that appear to have undergone significant space weathering and particles (Aα) that show little to no evidence of such alteration. This bimodality may reflect a difference between particles that experienced prolonged space weathering on Ryugu’s surface and those that were fractured during sampling or transportation, exposing fresh surfaces. In contrast, most of the particles from chamber C exhibit spectral characteristics intermediate between Aα and Aβ. Notably, chamber C contains very few particles that show strong space weathering effects. This finding aligns with the fact that the SCI excavation efficiently sampled subsurface materials from Ryugu. However, compared to Aα, the chamber C samples exhibit slightly stronger dehydration, accompanied by a shift of the absorption band peak position towards longer wavelengths. One possible cause of this dehydration and peak shift is exposure to solar UV radiation after the SCI impact. The excavated surface particles of Ryugu were subjected to continuous UV exposure for up to three months until the TD2 operation. To test this hypothesis, a UV irradiation experiment was conducted using the Ivuna meteorite. The results revealed both a peak shift and a decrease in the depth of the 2.7 µm absorption band, with changes resembling those observed in the Ryugu particles. These findings suggest that approximately 18 days of solar UV exposure is sufficient to induce the spectral changes observed between Aα and the chamber C particles. However, these spectral changes cannot be attributed solely to UV irradiation, as thermal effects must also be considered. UV exposure may have increased the temperature of the particles, thereby promoting dehydration. Therefore, future studies should quantitatively assess the thermal effects associated with UV irradiation. Furthermore, improvements in experimental conditions are necessary. The Ivuna particle have undergone terrestrial weathering, and the experiment was conducted in nitrogen environment. To enhance the reproducibility of the results, future experiments should incorporate the following modifications: (1) conducting the experiment in vacuum environment and (2) using samples that have not been affected by terrestrial weathering.

Methods

The MicrOmega is a near-infrared hyperspectral microscope developed by the Institut d’Astrophysique Spatiale, Universite Paris-Saclay, Orsay, France (IAS)15. MicrOmega has a FOV of 5 × 5 mm2 (256 × 256 pix2) with a spatial resolution of 22.5 µm/pix and can acquire spectral data in the 0.99–3.65 µm range. The typical spectral sampling resolution over the spectral range is 20 cm-1 and is 10 cm-1 in the ~ 2.7 μm region. Features that can be identified in this NIR range are mainly hydrates, organics, hydroxylated material, carbonates, and mafic minerals. MicrOmega has been installed at the JAXA Curation Center for the initial description of Ryugu returned samples. MicrOmega measurements have been performed in an N2-purged acrylic chamber to prevent a contamination, i.e., under oxidation and moisture control. This acrylic chamber is also shielded with a black cover to prevent stray light during the measurement. A water-cooled copper plate is attached at the bottom of MicrOmega to maintain its temperature at 10 °C. The detector inside MicrOmega is cooled to 110 K by a cryogenic cooler. The sample chamber is also under the N2 conditions, and the sample is placed on an integrated gold mirror substrate. The sample was measured through a sapphire window inserted between the MicrOmega chamber and sample chamber and calibrated with an infragold and a spectralon46. In this study, a total of 207 spectral data (119 from chamber A particles and 88 from chamber C particles) from the MicrOmega-Curation-DARTS server47 were used (see the spreadsheet in the Data availability section).

For the slope analysis of a region about 1.4–2.3 µm without absorption bands (named a 2.0 µm continuum), at first, the spectral data were normalized at reflectance at 1.57 µm. Then, the slope was calculated by fitting a linear function to the 1.4—2.3 µm range. In the fitting, the parameters a and b were calculated to minimize the sum of the residual squares of the linear function (y = ax + b).

For the analysis of the 2.7 µm band, at first, the baseline estimation and the subtraction were performed, and then the entire absorption band was analyzed by fitting with multiple Gaussian functions. Since the Ryugu particle is composed of many minerals and these minerals have multiple vibrational energy levels (For example, the same functional group but with different bonding atoms), multiple Gaussian functions were used to analyze the entire absorption band. An asymmetric least squares smoother48 was used for the baseline estimation for the range of 2.55—3.25 µm. This method assumes that the measured spectral data (y) follows a smooth data sequence (z). In Eq. 1 with penalty term, the baseline (z) is estimated by applying to the least-squares problem, i.e., solving Eq. 248 The first term of the Eq. 1 is the goodness of fit of the baseline (z) to the measured data (y), and the second term is the roughness (smoothness) of the baseline (z) based on Eq. 3. Each of these terms contains a weight (w) and a penalty factor (λ), respectively. The weight (w) is an arbitrary weight to the relative position of the estimated value (z) relative to the measured data (y). The penalty coefficient (λ) is a parameter that sets the magnitude of the second term, the baseline smoothness contribution. Common values for these parameters were used for all particles (w = 0.02, λ = 105.5) in this study. Then, the processed reflection data were converted to intensity data using Eq. 4.

$$Q={\sum }_{i}{w}_{i}{\left({y}_{i}-{z}_{i}\right)}^{2} + \lambda {\sum }_{i}{\left({\Delta }^{2}{z}_{i}\right)}^{2},$$
(1)
$$\frac{\partial Q}{\partial z}=0,$$
(2)
$${\Delta }^{2}{z}_{i}=\left({z}_{i}-{z}_{i-1}\right)-\left({z}_{i-1}-{z}_{i-2}\right),$$
(3)
$$Intensity\left[\%\right]=\frac{Measured - Baseline}{Baseline}\times 100.$$
(4)

Next, the intensity data points were linearly interpolated by a factor of 6 to reduce the error bars of the acquisition parameters in the multiple Gaussian fitting. Then, fitting with multiple Gaussian functions was performed. In this analysis, initial values (peak wavelength, depth, and width of each Gaussian function) were set and the optimal parameters (peak wavelength, depth, and Full Width at Half Maximum: FWHM) of the Gaussian functions that minimize the sum of squared residuals with the measured data were calculated. To determine the optimal number of Gaussian functions, two statistical criteria were used: the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). AIC is a widely used indicator for model selection, balancing model fit and complexity. It is defined as Eq. 5, where \(k\) represents the number of parameters in the model, and \(L\) denotes the maximum likelihood of the model. A lower AIC value indicates a model that better fits the data with fewer parameters, thus minimizing the risk of overfitting26. BIC is another criterion for model selection, and it is defined as Eq. 6, where \(n\) is the number of data points. Like AIC, a lower BIC value suggests a more optimal model. BIC tends to favor simpler models compared to AIC due to its higher penalty on the number of parameters27. Models with 1 to 8 Gaussian functions were applied to the average spectra, and the best model was selected using the AIC and BIC, as well as the efficiency of determination: R2, as shown in Eq. 7. The results of applying each model and plotting the values of AIC, BIC, and the value of R2 for each model are shown in the Supplementary Figure-S1. The results show that the values of all three indicators (AIC, BIC, and R2) remained almost constant when using more than 6 Gaussian functions. In addition, some particles could not be fitted with more than 7 Gaussian functions. From the perspective of an integrated analysis of all particles, the results of the analysis with 6 Gaussian functions were used in this study.

$$AIC=2k-2\text{ln}(L)$$
(5)
$$BIC=k\text{ln}(n)-2\text{ln}(L)$$
(6)
$${R}^{2}=1-\frac{\sum {\left(y-{y}_{fit}\right)}^{2}}{\sum {\left(y-\overline{y }\right)}^{2}}$$
(7)

Error bars on the parameters of the composite waveform were calculated as follows. The Jacobian matrix is utilized to estimate the uncertainty (standard error) of fitting parameters in nonlinear least squares fitting. This method leverages the variability in the residuals to derive the uncertainty associated with each parameter. Initially, the Jacobian matrix \(J\) was calculated, comprising the partial derivatives of the residuals \(r\) with respect to each fitting parameter \(p\). Each element \({J}_{ij}\) of the Jacobian matrix represents the partial derivative of the \(i\)-th residual with respect to the \(j\)-th parameter in Eq. 8. Subsequently, the covariance matrix of the parameters \(Cov(p)\) was derived from the Jacobian matrix \(J\) and the variance of the residuals \({\sigma }^{2}\) in Eq. 9. Here, \({J}^{T}\) denotes the transpose of the Jacobian matrix, and \({({J}^{T}J)}^{-1}\) is its inverse. The variance of the residuals \({\sigma }^{2}\) was estimated by dividing the residual sum of squares (RSS) by the degrees of freedom, which is the number of data points \(N\) minus the number of parameters \(k\), shown in Eq. 10. Finally, the standard error (uncertainty) of each parameter was obtained by taking the square root of the diagonal elements of the covariance matrix as shown in Eq. 11.

$${J}_{ij}=\frac{\partial {r}_{i}}{\partial {p}_{j}}$$
(8)
$$Cov\left(p\right)={\sigma }^{2}{({J}^{T}J)}^{-1}$$
(9)
$${\sigma }^{2}=\frac{RSS}{N-k}$$
(10)
$$Standard Error\left({p}_{j}\right)=\sqrt{{Cov(p)}_{jj}}$$
(11)

The UV irradiation experiment was conducted at the Institute of Space and Astronautical Science (ISAS), Japan Aerospace Exploration Agency (JAXA). A chip of Ivuna carbonaceous chondrite meteorite (2.5 mg, and ~ 1.5 mm), was selected for the UV irradiation experiment in this study as Ivuna is known to have a similar composition to Ryugu. The Ivuna particle was used in its granular form and was fixed onto an aluminum stage using carbon tape. UV irradiation was performed in a nitrogen-filled facility-to-facility transfer container (FFTC), a sealed container designed to prevent atmospheric exposure28. The UV light was introduced through a synthetic fused silica glass window of the FFTC with a transmittance of 0.92. The UV irradiation system used in the experiment was an Asahi Spectra MAX-303, which emits light in the wavelength range of 250–385 nm29. To determine the correspondence between laboratory UV irradiation time and solar UV exposure at 1 AU, we considered the transmittance of the synthetic fused silica glass window, the spectral irradiance of the light source, and the degradation of the UV lamps over time. In this experiment, the Ivuna sample was irradiated at an intensity of 2.3 × 103 mW/cm2 for 1 h. The total energy delivered in this 1-h laboratory experiment is equivalent to 258.4 h of solar UV irradiation at 1 AU. Spectral data of the Ivuna, before and after irradiation experiment, were obtained using a Fourier-transform infrared (FTIR) spectrometer (JASCO FT/IR-6100) in a nitrogen-filled glove box at ISAS to avoid any terrestrial contamination (e.g., weathering). The measurement aperture was set to 500 µm × 500 µm, covering a wavelength range of 1.28–28.6 µm. The spectral resolution was 2 cm⁻1, and the cumulative count for the measurement was 600. All experimental procedures, including irradiation and spectral measurements, were performed under nitrogen conditions to avoid contamination from atmospheric components. In the analysis of the spectral data, we focused on the 2.7 µm sharp absorption band (2.70—2.74 µm) where the least terrestrial contamination is expected to examine the UV effects on the Ivuna. We analyzed the spectral data of non-irradiated and irradiated Ivuna using Multi-Gaussian Fitting applied to the Ryugu particles and calculated the parameters of the composite waveform for the Ivuna sample (Table 2, Supplementary Figure-S2, and Table-S1). The slope around 2.0 µm, which was investigated for the Ryugu particles, is too noisy in FTIR spectra. Instead, the slope between shorter wavelength side and longer wavelength side of the 2.7 µm band was calculated. First, the reflectance was normalized at 2.49–2.51 µm. Next, the average reflectance was calculated at the wavelength ranges of 2.49–2.51 µm and 3.99–4.01 µm, and the slope between the two points was calculated.