AKARI/IRC Near-Infrared Asteroid Spectroscopic Survey: AcuA-spec

Knowledge of water in the solar system is important for understanding of a wide range of evolutionary processes and the thermal history of the solar system. To explore the existence of water in the solar system, it is indispensable to investigate hydrated minerals and/or water ice on asteroids. These water-related materials show absorption features in the 3-$\micron$ band (wavelengths from 2.7 to 3.1 $\micron$). We conducted a spectroscopic survey of asteroids in the 3-$\micron$ band using the Infrared Camera (IRC) on board the Japanese infrared satellite AKARI. In the warm mission period of AKARI, 147 pointed observations were performed for 66 asteroids in the grism mode for wavelengths from 2.5 to 5 $\micron$. According to these observations, most C-complex asteroids have clear absorption features ($>10\%$ with respect to the continuum) related to hydrated minerals at a peak wavelength of approximately 2.75 $\micron$, while S-complex asteroids have no significant feature in this wavelength range. The present data are released to the public as the Asteroid Catalog using AKARI Spectroscopic Observations (AcuA-spec).


Introduction
Water is found in various forms in our solar system and is one of the most important ingredients in the origin of life. It also has vital implications on the exploration of extrasolar planets and provides evidence for the evolution of the solar system, especially its thermal history. Silicate minerals account for a large fraction of solid materials in the solar system and water exists as ice in small solar system bodies beyond Jupiter. Hydrated minerals (any mineral that contains H2O or OH) are formed in environments where anhydrous rock and liquid water exist together with a cer-tain pressure and temperature, resulting from aqueous alteration (e.g., Brearley 2006 and references cited therein). Because hydrated minerals are stable even above the sublimation temperature of water ice, they become an important tracer of water present in the history of the solar system unless they were reset by a temperature change after formation. The study of hydrated minerals is therefore important for understanding of the origin of Earth's water and unravelling of the earliest thermal processes in the solar system. Most asteroids have not experienced sufficient thermal evolution to differentiate into layered structures like terrestrial planets since their formation; thus, aster-oids are considered to record the initial conditions of our solar nebula of 4.6 Ga ago. To explore the existence of water in the present solar system, it is necessary to investigate the presence of hydrated minerals and water ice on various types of asteroids.
Meteorites collected on the Earth have typically fallen from asteroids (e.g., Morbidelli & Gladman 1998) and bring useful information for asteroid research. However, it is difficult to correctly measure the water content because meteorites have been contaminated by atmospheric water (e.g., Beck et al. 2010;Mogi et al. 2017). Observations with astronomical telescopes are needed to investigate asteroids without terrestrial alteration. Hydrated minerals and water ice exhibit diagnostic absorption features in the so-called 3-µm band (approximately 2.5-3.5 µm wavelength range, Rivkin et al. 2002). Features at ∼ 2.7 µm are attributed to hydrated minerals and those at ∼ 3.05 µm to water ice. Other materials also display spectral features at these wavelengths, such as ammonium-bearing phyllosilicate (King et al. 1992), mineral brucite (magnesium hydroxide; Beck et al. 2011), and absorbed water molecules in regolith particles (e.g., in the lunar rocks or soils; Clark 2009). Many spectroscopic surveys have been conducted in the 3-µm band using ground-based observatories. Takir & Emery (2012) observed 28 outer main-belt asteroids with the NASA Infrared Telescope Facility (IRTF) on the summit of Mauna Kea, Hawaii, and classified them into four spectral groups based on the absorption shapes: sharp, rounded, Ceres-like, and Europa-like, and discussed the distribution and abundance of hydrated minerals in the outer main-belt region. However, space-borne telescopes can perform more accurate observations for the identification of mineral species, because the spectrum observed with a ground-based telescope (2.5-2.85 µm) is severely affected by telluric absorption (e.g., Rivkin et al. 2002).
AKARI (Murakami et al. 2007), launched in 2006, is a Japanese satellite mission fully dedicated to a wide range of infrared astronomy, including galaxy evolution, stellar formation and evolution, interstellar media, and solar system objects. As part of our research, we conducted two types of asteroid survey with AKARI. One is a mid-infrared asteroid survey. Using the mid-infrared part of all-sky survey data obtained with the Infrared Camera (IRC; Onaka et al. 2007;Ishihara et al. 2010) on board AKARI, we constructed a size and albedo catalog of 5120 asteroids (Usui et al. 2011), which is summarized in the Asteroid Catalog using AKARI (AcuA) 1 . This is an unbiased asteroid catalog 1 This catalog, as well as the infrared flux data of individual asteroids (Alí-Lagoa et al. 2018), is open to the public at the following URLs: http://darts.isas.jaxa.jp/astro/akari/catalogue/AcuA.html and at two mid-infrared bands (9 µm and 18 µm), which fully covers objects with a diameter of > 40 km in the main-belt region (Usui et al. 2014). It was conducted using observations made during the cryogenic phase (Phase 1 and 2). The other survey is a near-infrared spectroscopic survey, which is described in this paper. Low-resolution spectroscopic observations were performed using the near-infrared channel (2.5-5 µm) of the IRC, which provide valuable data thanks to its high sensitivity and unique wavelength coverage (Ohyama et al. 2007). Note that the Infrared Space Observatory (ISO; Kessler et al. 1996) has a spectroscopic sensitivity to detect only the largest objects in the main-belt region (Dotto et al. 1999;Rivkin 1997) and that the Infrared Spectrograph (IRS) of the Spitzer space telescope (Werner et al. 2004) only covers wavelengths longer than 5 µm.
In this study, we report a spectroscopic survey of asteroids using the IRC on board AKARI. During the warm mission period of AKARI (Phase 3; Onaka et al. 2012), 147 pointed observations were performed for 66 asteroids in the grism mode at wavelengths from 2.5 to 5 µm. The observed objects comprise C-complex (×23), S-complex (×17), Xcomplex (×22), D-complex (×3), and V-type (×1) asteroids, all of which are in the main-belt region and have diameters of 40 km or larger. This paper is organized as follows: section 2 describes the observations of asteroids made with AKARI and the data reduction process used to derive reflectance spectra; section 3 presents the characteristics of the obtained spectra; and the results are discussed in section 4.

Observations and Data Reduction
2.1 NIR grism spectroscopic observations with the AKARI/IRC AKARI was launched on 2006 February 21 UT, and its liquid helium cryogen boiled off on 2007 August 26 UT, 550 days after the launch (Murakami et al. 2007). This cryogenic phase is referred to as Phase 1 and 2. In the posthelium phase (Phase 3), the telescope and its scientific instruments were maintained at approximately 40 K by the mechanical cooler and only near-infrared observations were carried out until 2010 February. The main purpose of AKARI was to perform all-sky surveys during Phase 1 and 2 by continuous scanning of the sky at a constant scan speed; i.e., in the survey observation mode. AKARI also has a capability of a pointed observation, which is occasionally inserted into the continuous survey observation (the observational strategy of AKARI is discussed in http://www.ir.isas.jaxa.jp/AKARI/Archive/Catalogues/Asteroid Flux V1/ Matsuhara et al. 2005). In Phase 3, only pointed observations with IRC were available, in which the same attitude control mode was employed as that during Phase 1 and 2 operations. This pointed observation mode is employed for deep imaging or spectroscopy. Our targets were 66 objects selected from main-belt asteroids, the Cybeles, the Hildas, and one near-Earth object, 4015 Wilson-Harringtion, according to the visibility from AKARI. Most of these observations were performed as part of the AKARI Mission Program entitled "Origin and Evolution of Solar System Objects" (SOSOS), while four observations were executed within the framework of the AKARI director's time program (DT): two observations were made for 6 Hebe and the other two for 128 Nemesis. The physical properties and taxonomic classification of our targets are summarized in tables 1 and 2. All objects have entries of the size and albedo (Usui et al. 2011;Mainzer et al. 2011), rotational period (Warner et al. 2009), and taxonomic information from visible to nearinfrared wavelengths (from 0.4-2.5 µm; hereafter vis-NIR) compiled by Hasegawa et al. (2017). The distribution of the orbital elements and the size and albedo diagrams of our targets are shown in figures 1 and 2, respectively.
Observations were conducted 147 times from 2008 May to 2010 February. Observation logs are given in table 3. Typically, two or three pointed observations were performed for each target, whereas three targets (8 Flora, 69 Hesperia, and 4015 Wilson-Harrington) were observed only once. Most objects were observed within ∼ 100 min intervals, which were the orbital period of the satellite. Exceptions are 216 Kleopatra and 704 Interamnia, both of which were observed at intervals of more than 14 months because of the scheduling constraint, but not for any scientific reason. The direction of AKARI observations was severely limited to within 90 • ± 0.7 • of the solar elongation angle due to the design of the orbit and attitude control system to avoid radiation from the Earth and the Sun. Within this geometric constraint, our targets in the mainbelt region were located between phase angles of 15 • and 32 • . We made observations with the AKARI Astronomical Observation Template (AOT) of spectroscopy dedicated to Phase 3 (IRCZ4; see IRC Data User Manual 2 ), which is equivalent to IRC04 in Phase 1 and 2 (Onaka et al. 2007;Ohyama et al. 2007). The grism mode covered a wavelength range of 2.5-5 µm with a spectral resolution of R ∼ 100. Within a single AOT operation lasting approximately 10 min, 8 or 9 spectroscopic images were taken with the grism, as well as a photometric image (called a reference image) through a broadband filter in the wavelength range of 2.7-3.8 µm centered at 3.2 µm (the N3 2 http://www.ir.isas.jaxa.jp/AKARI/Observation/support/IRC/ band). The effective exposure time for each frame was 44.41 s (see IRC Data User Manual). Total exposure time of one pointed observation is 310-400 sec depending on the attitude stability of the satellite; thus our observations provided a flux-limited sample. The targets were put on the 1 ′ × 1 ′ aperture mask (the "Np" window; Onaka et al. 2007, also see figure 3) to minimize contamination from nearby objects (e.g., Ootsubo et al. 2012). AKARI was also equipped with narrow slits ("Nh" of width of 3 ′′ and "Ns" of width of 5 ′′ ); however, these modes were designed for diffuse sources. The absolute accuracy of the telescope pointing of the satellite is a few arcseconds; thus, we used the "Np" window instead of "Nh" or "Ns" slits for point sources. For the "Np" observations, it was not necessary to consider slit loss because of the large width of this mask. Note that the full width at the half maximum of the point spread function (PSF) in the N3 band is 3.2 pixels or 4. ′′ 67 (Onaka et al. 2008(Onaka et al. , 2010; also see IRC Data User Manual). Figure 3 shows an example of data reduction for C-type asteroid 511 Davida (ObsID:1520065-001).

Extraction of one-dimensional spectra
The raw data were reduced using the IDL-based software package in the IRC Spectroscopy Toolkit for Phase 3 data, version 20170225RC (Ohyama et al. 2007;Usui et al. 2018; also see IRC Data User Manual). Standard array image processing, such as dark subtraction, cosmic-ray removal, linearity correction, and various image anomaly corrections (bad-pixel masking, correction of tilt of the spectrum, and the first-order sky subtraction), were first performed for each exposure frame with the toolkit. Then, multipleexposure frames were added, taking into account the relative image shift due to the satellite attitude drift among the frames. Because AKARI does not have a tracking mode for moving objects, the resulting two-dimensional spectra with the standard toolkit was blurred due to motion of the object. Note that the velocities of apparent motion of the targets were typically less than 7. ′′ 5 / 10 min (or ∼ 5 pixel / 10 min); thus, the targets were located within the field of view of the "Np" window during the observations (with the exception of 4015 Wilson-Harrington, which was 30 ′′ / 10 min). The movements of the target asteroid during the observation of each frame were calculated to add to the relative image shift, and multiple-exposure frames were shift-and-added to extract two-dimensional spectral image of the objects. This function was implemented in the toolkit for the data of solar system objects (details are given in Ootsubo et al. 2012). Using the combination of shift-and-add with the movement of the asteroid and the 3-σ clipping methods, the effects of hot pixels were reduced in the stacked images, although the number of hot pixels in Phase 3 increased compared with Phase 1 and 2. The fluctuation of the background signals causes a random noise of the flux uncertainties of the target spectra. With the standard toolkit, the sky background was estimated from the adjacent region of the target in the "Np" window and subtracted from the stacked images. In some cases, however, the background signals are not able to be estimated properly from the adjacent region because of contamination of neighboring stars in the "Np" window. In this study, wavelength-dependent sky background was estimated from the diffuse sky spectrum of the "Ns" slit for the same pointed observation which was checked as contamination-free "blank" sky by visual inspection.
The absolute flux calibration (spectral response calibration) of the AKARI/IRC spectroscopy is based on the observation of active galactic nuclei (AGNs) and spectroscopic standard stars obtained with the same instrumental setup. The AGNs without significant emission or absorption features are used as calibrators of redder objects (contributed in longer wavelengths), and the standard stars whose spectra are already templated are used as bluer objects (contributed in shorter wavelengths). 13 featureless AGNs and six K-and A-type standard stars were selected as flux templates. The observations of these objects were performed within the framework of the DT program, separately from individual observations. Comparing the modeled templates of the objects with the observed flux count in analog-to-digital units (ADU), the spectral response was obtained as a function of wavelength. This spectral response was implemented in the toolkit. Details are given in Baba et al. (2016) and Baba et al., submitted. In general, the spectroscopic flat-fields are made by gathering a large number of blank sky spectroscopy images that are combined and normalized so that any faint object spectra are removed by clipping averaging techniques. For the AKARI observations, flat data were not taken in individual pointed observations because the observational time was severely limited due to the avoidance constraints of the satellite. As the natural background is faint in the near-infrared for small-aperture spectroscopy of the IRC (Np, Ns, and Nh), the quality of the flat data cannot be improved substantially even after stacking of multi-pointed observational data. Thus, the quality of the processed spectra is, unfortunately, limited by the quality of the flat data, not by the dark current nor photon noise. If the spectra have low S/N ratios, then applying a flat could degrade the data. Therefore, we skipped performing flat fielding in the data reduction process (with an option called /no slit flat prepared to disable the flat correction in the toolkit: see IRC Data User Manual).
It is reported that the sensitivity decreased by a maximum of approximately 10% during Phase 3 with the increase of the IRC detector temperature (Onaka et al. 2010). This can be approximated by a linear function of the detector temperature T (Baba et al., submitted) as: where f (T ) is the correction factor, a = −0.0290 K −1 , and T0 = 43.55 K. For ObsID:1521116-001 observed on 2009 November 18, the recorded detector temperature was the highest (T = 45.13 K) among all asteroid observations during Phase 3, and the correction factor was given as f (T ) = 0.954. Then the flux density is scaled by 1/f (T ) = 1.048 to obtain the corrected value. This correction factor is assumed to not have wavelength-dependency. In the spectroscopic analysis, the flux and the wavelength accuracy strongly depend on the accuracy of the wavelength zero point, which was estimated on the reference image of the N3 band. It was estimated to be, at worst, 1 pixel in our analysis. Thus, the flux and the wavelength uncertainties were estimated by calculating how much the spectrum changed when the wavelength zero point was shifted by ±1 pixel (Shimonishi et al. 2013). Finally, one-dimensional spectra were extracted from the two-dimensional images by summing signals over 7 pixels (approximately 10. ′′ 5) in the spatial direction to reduce the effect of hot pixels and/or cosmic rays that hit the detector. Then, one-dimensional spectra were extracted. The obtained one-dimensional spectra were smoothed along their wavelength with 5 pixels (i.e., 1.5× PSF) for further analyses. In this study, we focus on broader features (bandwidth of > 0.1 µm) appeared at around 2.7 µm and 3.1 µm in the spectra (c.f., table 1 in . There are wavy patterns in other wavelengths remaining in some extracted spectra, which have not seen in meteorite spectra. It is likely that most of them are spurious due to the contamination of neighboring stars or insufficient background subtraction, although we cannot completely rule out that they are real features. In the following, we do not discuss other features. The obtained spectra of each pointed observation are summarized in supplementary data. After visual inspection, it was found that some observational data were contaminated by ghost patterns (Egusa et al. 2016) or by nearby stars that happened to be in the same field of the "Np" window. The data of the following observations were therefore removed from further analyses (see table 3); ObsID:1521233-001 for 9 Metis, ObsID:1520189-001 for 44 Nysa, ObsID:1521191-001 for 49 Pales, ObsID:1521212-001 for 145 Adeona, and ObsID:1520161-001 for 250 Bettina.

Instrument linearity and saturation
Saturation of the observed signal makes severe problems for larger asteroids, which need to be handled with care. The saturation level of the detector during Phase 3 observations was reported as 2000 ADU per pixel (Onaka et al. 2008; see also IRC Data User Manual), which roughly corresponds to ∼ 1 Jy for the grism mode. A brighter observed flux than this value cannot be assured in the linearity correction of the detector. This was the case for the largest objects: 1 Ceres, 2 Pallas, and 4 Vesta. In this work, the following empirical method was used to determine the available wavelength range and extract the spectra.
The minimum observation unit of the IRC is called the exposure cycle, which consists of one short exposure (∼ 4.6 sec) and one long exposure (∼ 44.4 sec) for the near-infrared channel. One pointed observation of spectroscopy contains 8-9 exposure cycles, as described in section 2.1. Short and long exposure data are reduced separately by the toolkit with each set of the calibration parameters. Unusual behaviors due to non-linearity or saturation of the detector can be found by comparing short-and long-exposure spectra. Figure 4 shows the short-and long-exposure spectra of 1 Ceres, 2 Pallas, and 4 Vesta. Comparison of the short-and long-exposure spectra suggests that the available wavelength range in which both spectra match to an accuracy of 10% of each flux is λ < 3.7 µm for 1 Ceres and λ < 3.75 µm for 2 Pallas. The spectra within these wavelengths can be used for further analyses. However, for of 4 Vesta, the short-and longexposure spectra agree only within 3.2 < λ < 3.5 µm, which does not provide necessary information for the present study. Thus, the 4 Vesta spectra cannot be used for further analyses.

Thermal component subtraction
The spectrum obtained in the previous section essentially consisted of two components in the wavelength range of 2.5-5 µm: thermal emission of the asteroid itself (Fe) and reflected sunlight (Fr). To obtain the reflectance spectrum, it was necessary to remove the thermal component from the spectrum, which depends on the distribution of surface temperature at the epoch of the observation. In this work, the Near-Earth Asteroid Thermal Model (NEATM; Harris 1998) was used to calculate the thermal flux at longer wavelengths by fitting the spectrum. The NEATM is a refinement of the standard thermal model (STM; Lebofsky et al. 1986). The NEATM solves simultaneously for the diameter (D) and the geometric albedo (pv) as well as the beaming parameter (η) to fit the observed (infrared) flux. η was originally introduced in the STM to allow the model temperature distribution to fit the observed enhancement of thermal emission at small solar phase angles (e.g., Lebofsky et al. 1986). In practice, η can be treated as a model parameter that allows a first-order correction for any effect that influences the observed surface temperature distribution. The geometric information; i.e., the heliocentric distance (r h ), the geocentric distance (∆), and the phase angle (α), were obtained from JPL Horizons (table 3). The absolute magnitude (H) and the slope parameter (G) were employed as the visible flux (Bowell et al. 1989). The emissivity was assumed to be constant at ǫ = 0.9 throughout the wavelengths considered in this study (c.f., Lebofsky et al. 1986). D and η were parameterized to fit the thermal flux of the asteroid. After subtracting the thermal component fitted by NEATM, the spectrum was divided by the solar spectrum based on the corrected Kurucz model (Berk et al. 1999) to obtain the reflectance spectrum of the asteroid. The obtained reflectance spectra of each pointed observation are summarized in supplementary data.
2.2.4 Criterion to reject spectra severely contaminated by thermal emission In this research, we aimed to detect small absorption features (∼ 10%) on the spectrum of the reflected sunlight component. The thermal component of the spectra was subtracted as described in the previous section. However, it was still necessary to carefully handle the "contamination" of thermal emission from the asteroid itself. It is hard to extract reflectance spectra with a sufficient S/N ratio from spectra with large thermal emission contribution.
The grism spectroscopy of the AKARI/IRC covers a wavelength range of 2.5-5 µm. At these wavelengths, thermal emission has an equivalent contribution to the total spectrum of a main-belt asteroid to the reflected component. As mentioned above, thermal emission was estimated by NEATM; that is, thermal emission was assumed as a gray body with a fixed emissivity of ǫ = 0.9. Realistically, the emissivity is not necessarily constant. This ambiguity of emissivity could affect the detailed features of the reflectance spectrum. We examined the contribution of the thermal component in the total flux density as follows.
Let us consider the relationship between the bihemispherical reflectivity (p h ) and the hemispherical emissivity (ε h ) at a certain wavelength.
As stated by Kirchhoff's law of thermal radiation, p h and ε h are complementary as (Hapke 1993) : Note that both p h and ε h have wavelength dependency; p h = p h (λ), and ε h = ε h (λ). We assume that the spectrum has an absorption feature with a band depth of s. Here, the band depth is defined as a fraction of the depression from the continuum. The bi-directional reflectivity to be observed with the absorption feature (p d ) is given as: where p d0 is the continuum of the bi-directional reflectivity. In the same way, we assume that the spectrum has an emission feature with an emission strength of e. The directional emissivity to be observed with the emission feature (ε d ) is given as: where ε d0 is the continuum of the directional emissivity. It is assumed that an absorption feature appears as downward from the continuum, while an emission feature appears as upward. Therefore, an absorption is denoted negative (−s) in equation (3) and an emission is positive (+e) in equation (4). The bi-hemispherical reflectivity (p h ) and the bi-directional reflectivity (p d ) are assumed to have the same relation as the Bond albedo to the geometric albedo in the IAU H-G function model (Bowell et al. 1989) as: where q is the phase integral empirically given by the slope parameter (G) as: It should be noted that the applicability of the H-G function in the infrared wavelengths is not well studied. Lederer et al. (2008) reported the photometric properties of S-type asteroid 25143 Itokawa based on ground-based UBVRIJHK broadband observational data (0.36-2.15 µm); the difference of the phase integral by wavelength is small (q = 0.11 ± 0.01 in V-band and 0.13 ± 0.01 in K-band).
There are no other reports to date of the phase integral measured in the near-infrared wavelengths. Here the wavelength difference of the phase integral of asteroids observed in this study is assumed to be small. Thus equations (5)-(6) is applied to all asteroids in our analysis. In addition, the angular variation of the directional emissivity (ε d ) is assumed to be neglected (e.g., Sobrino & Cuenca 1999;García-Santos et al. 2012), thus to be identical to the hemispherical emissivity (ε h ) as: and where ε h0 is the continuum of the hemispherical emissivity which is given by Kirchhoff's law as where p h0 is the continuum of the bi-hemispherical reflectivity and again From these equations, Thus, we have An asteroid spectrum (Ftot) comprises two components: reflected sunlight (Fr) and thermal emission (Fe) as Flux densities of the absorption feature (∆Fr) and emission feature (∆Fe) are written as |∆Fr| = sFr , |∆Fe| = eFe .
To detect the absorption feature (i.e., measuring s) with more than x × 100 % accuracy, the contribution of the emission feature of the thermal component should be suppressed as Here, we consider that x is the same value as the typical uncertainty of the flux density of AKARI spectroscopy. The geometric infrared albedo (pIR) is defined as the ratio of the brightness at zero phase angle to the brightness of a perfect Lambert disk of the equivalent radius at a given infrared wavelength. We assume that the bi-directional reflectivity of the continuum (p d0 ) is equal to the infrared albedo (pIR) at 2.45 µm which is given as: where pv, R2.45, and R0.55 are the geometric visible albedo, the relative reflectance at 2.45 µm, and the relative reflectance at 0.55 µm, respectively. Note that the second equal sign of equation (13) comes from the assumption that the ratio of the geometric infrared albedo to the geometric visible albedo is as the same as the ratio of the relative reflectance at 2.45 µm and that at 0.55 µm. The relative reflectance values at 0.55 µm and 2.45 µm are obtained from the vis-NIR spectra (spectral data are compiled by Hasegawa et al. 2017). Fe Ftot , which can be calculated by the thermal model described in the previous subsection, is a function of wavelength. Thus, equation (12) with x = 0.04 (i.e., 4% uncertainty; see section 3.1) is treated as a condition of the available wavelength range (upper limit of wavelength, λtrunc) to extract the reflectance spectrum with sufficient accuracy. λtrunc values for each observed target are listed in table 4. It is natural that 4015 Wilson-Harringtion, which is the near-Earth object, has a high surface temperature; thus, the thermal component fully occupies the wavelength coverage of the spectroscopy (λtrunc = 3.0 µm). We cannot retrieve any valid reflectance spectra from this object; instead, the spectral data of this object were utilized for studying the thermal properties of the asteroid (e.g., Bach, Ishiguro, & Usui 2017).

3-µm-band depth measurements
The measurement method for band depth described in Takir & Emery (2012) cannot be applied to our data set because our spectra only cover wavelengths longer than 2.5 µm, which is the starting wavelength for absorption features associated with hydrated minerals. Instead, the band depth (D) was measured as: 1. before measuring the band depth, two or three reflectance spectra of each asteroid were averaged, 2. the continuum (Rc) was defined as the connection of two local peaks of the reflectance spectrum across the peak wavelength of the absorption feature, 3. the band depth (D) was defined as (see figure 3) where R λ is the reflectance spectrum (c.f., Clark & Roush 1984).
The peak wavelength was defined as the wavelength with the maximum value of the band depth. To increase the S/N, two or three spectra obtained were averaged in step 1. Thus, the spectral variability due to heterogeneities of the surface material or any other reason was smoothed out. After a visual inspection of our results, obvious features appeared at around 2.7 µm and 3.1 µm in the spectra. Thus, two band depths, D2.7 and D3.1, are treated as absorption features in this study and small features in other wavelengths are not considered.

General trend of spectra
The obtained reflectance spectra are summarized in figures 5-8. Figure 9 shows the distribution of uncertainties of all the data points in the reflectance spectra of 64 asteroids.
Most of the data points in the reflectance spectra have uncertainties smaller than 10% (the median value is 3.8%).
The uncertainties in the reflectance spectra are caused by the uncertainty in the absolute flux calibration (∼5%), the ambiguity of the wavelength zero point on the reference image (∼8%), and the fluctuation of the sky background (∼3%), which are described in section 2.2.1. Uncertainty in the thermal model calculation described in section 2.2.3 is not taken into account in the reflectance spectra; this uncertainty contributes to the spectra at λtrunc or longer wavelengths.
As described above, the observed spectrum of 4 Vesta (V-type) is saturated in the 3-µm band (section 2.2.2), and the thermal contamination of 4015 Wilson-Harringtion (Btype) is not fully corrected (section 2.2.4). Therefore, the spectra of these two objects cannot be used for further analysis. In total, the reflectance spectra of 64 objects are available in this study.
Diagnostic spectral features appear at wavelengths from 2.6 µm to 2.9 µm or longer with a band center of ∼ 2.75 µm (hereafter the 2.7-µm band), and from 2.8-2.9 µm to 3.2 µm or longer with a band center of ∼ 3.05 µm (hereafter the 3.1-µm band). Our results confirm that no significant spectral feature with broad peak centered at 2.95 µm is found in the asteroid spectra, while such a feature sometimes appears in the spectra of meteorites, which is attributed to adsorbed water (Beck et al. 2010;. There are other absorption features in the 3-µm band; for example, those associated with organic materials at 3.4-3.6 µm (e.g., De Sanctis et al. 2017), which is another aspect of interest to this study.
Most C-complex asteroids have clear absorption features in the 2.7-µm band, as well as some in the 3.1-µm band. On the contrary, most S-complex asteroids show no significant features at these wavelengths. Some X-complex asteroids have absorption features in the 2.7-µm band, but among D-complex asteroids, only one asteroid in our data shows an absorption feature in the 2.7-µm band.
The band depths and the peak wavelengths of spectra in the 2.7-and 3.1-µm bands for each asteroid are summarized in table 5. The 3-µm band shape in this table is determined by a combination of the spectral features in 2.7 µm (D2.7) and 3.1 µm (D3.1) bands as: • sharp spectral features with only D2.7, • w-shape spectral features 3 with both D2.7 and D3.1, • 3-µm dent features with only D3.1, • unclassified.
Here, detection of the band depth is identified by an S/N signal > 2. The number of classified objects is summarized in figure 10; sharp (×23), w-shape (×4), and 3-µm dent (×8) out of 64 asteroids. The relationships between band depth and other physical parameters are shown in figures 11-19. Figure 11 shows the relationship between the band depth in the 2.7-µm band and the geometric albedo. From this figure, a general trend of absorption in the 2.7-µm band is observed, whereby most C-complex asteroids (17 out of 22), three X-complex asteroids, and one D-complex asteroid have significant absorption features, which is typically associated with hydrated minerals. All these objects have low albedo (pv < 0.1), except for 2 Pallas (pv = 0.15; B-type). However, removing the top three absorption objects (51 Nemausa: 47%, 13 Egeria: 41%, and 106 Dione: 36%), there is no clear correlation found. Figure 12 reveals the relationship in the 3.1-µm band, in which no clear correlation is observed. Figures 13 and 14 show the distribution of peak absorption wavelengths for the 2.7-and 3.1-µm bands, respectively. The mean value of peak wavelength in the 2.7-µm band of 27 objects with S/N > 2 is 2.75 ± 0.03 µm, and that in the 3.1-µm band of 12 objects is 3.08 ± 0.02 µm. The uncertainty of the peak wavelength is estimated by a 1-pixel resolution of approximately 0.01 µm.  Takir & Emery (2012) reported that 31 Euphrosyne is classified as Europa-like with a 13.63% band depth in the 3.00-µm band. These largest objects, with a diameter > 250 km, have a band depth in the 2.7µm band of moderate to weak (30 > D2.7 > 10%), while some 150-250 km objects have greater band depths (D2.7 > 30%). On the other hand, no clear trend is observed in the distribution of the 3.1-µm band in figure 16.
Figures 17 and 18 show the relationship with the semimajor axis of objects. There are no clear trends in this distribution because the scatter is too large and the data suffered from significant observational bias. Figure 19 displays the spectral slope measured in the 3µm band. In this study, the spectral slope (S) is defined as the slope from 2.6 µm to λtrunc described in section 2.2.4. The mean value of the slope of the asteroids which have an absorption feature in the 2.7-µm band (D2.7 > 10%) is S = 0.004 ± 0.033, which is almost flat. In contrast, that of the asteroids which do not have a feature (D2.7 < 10%) is a positive slope (S = 0.058 ± 0.044), which is likely to be connected with the slope of vis-NIR spectra.

Individual objects
Brief descriptions of 64 individual asteroids are given in this section.

C-complex asteroids
Most C-complex asteroids (17 out of 22), especially all Ch-, Cgh-, B-, and Cb-type asteroids, have obvious absorption features (> 18%) at around 2.75 µm. Among Ccomplex asteroids, four types of spectral feature are observed: sharp, w-shape, 3-µm dent, and no clear significant feature. Moreover, in the sharp group, two subgroups exist: spectra with relatively large band depth ( > ∼ 24%), whose features extend toward 3 µm or longer wavelengths, and spectra with moderate band depth ( < ∼ 24%), whose features end at around ∼ 2.9 µm. The former group either does not exhibit the 3.1-µm band feature or it is obscured by the strong 2.7-µm band feature. The latter may have other features longer than 3 µm (weak 3-µm dent). The former comprises Ch-and Cgh-types, while the latter belongs to other C-complex asteroids.
• C-type asteroids 1 Ceres Ceres, classified as a dwarf planet, is the largest object in the main-belt region. This object is classified as a C-type asteroid and was the first object observed in the 3-µm band; i.e., Lebofsky (1978) clearly detected the 3-µm depth on its spectrum, King et al. (1992) discussed the existence of ammoniated phyllosilicate on Ceres, and  discussed phase angle-induced spectral effects on the 3-µm absorption band. Recently, Ceres has been comprehensively studied using in-situ observations of the spacecraft Dawn (Russell et al. 2004;Russell & Raymond 2011); an absorption band centered near 3.1 µm has been attributed to ammoniated phyllosilicates widespread across its surface (de Sanctis et al. 2015). The characteristics of the spectrum taken with AKARI are essentially consistent with those of previous researches: the band centers located at 2.73 µm (2.72 µm, de Sanctis et al. 2015) and 3.08 µm (3.06-3.07 µm, . In this study, this spectral feature is categorized as the w-shape. There is a bumpy structure at around 3.2 µm of uncertain origin, which is the similar feature found in  observed at the same phase angle of α ∼ 22 • . 10 Hygiea Hygiea is a C-type asteroid, the fourth largest object (diameter of 430 km) in the main-belt region and is the largest member of its own family (Carruba 2013 (2012) classified the 3-µm band spectral feature of this object into the Ceres-like group. Our AKARI results also show that this object has two significant features at 2.72 and 3.08 µm; thus, this spectral feature is also categorized as the w-shape.

Themis
Themis is a C-type asteroid. It is the largest member of its own family (e.g., D. Nesvorny 2015 4 ). Fornasier et al. (1999) reported that this object has a signature of aqueous alteration based on ground-based observations at visible wavelengths. Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the rounded group. The rounded shape has been previously identified by Campins et al. (2010). Rivkin & Emery (2010) found that the spectrum of Themis is matched by a spectral model of a mixture of ice-coated pyroxene grains and amorphous carbon, suggesting that the surfaces of these asteroids with rounded features include very fine water frost in the form of grain coatings. Our AKARI results show that its spectrum has two absorption features: at around 2.76 µm, which is associated with hydrated minerals, and at around 3.07 µm, a wide feature related to water ice; thus, this spectral feature is categorized as the w-shape. The 3.1-µm feature of this object, which is thought to be associated with water ice, is broader than that of 1 Ceres or 10 Hygiea.

Europa
Europa is one of the largest C-type asteroids (diameter of 350 km). This object has a detailed 3-D shape  (2012) classified the 3-µm band spectral feature of this object into the Europa-like group using its own name. Our results show a small 3.1-µm feature as well as a weaker feature at 2.7 µm, which is below the detection limit (S/N< 2). 81 Terpsichore Terpsichore is a C-type asteroid and the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ). The albedo is 0.048 according to AKARI (Usui et al. 2011) or 0.034 according to WISE (Masiero et al. 2012), which is relatively dark as C-complex asteroids. There are no reports to date of observations in the 3-µm band. Our AKARI results show the 2.7-µm feature, which is classified into the sharp group. 94 Aurora Follow-up observations at 1-2.5 µm were performed by Hasegawa et al. (2017), and this object was classified as C-type according to the Bus-DeMeo taxonomy.
No features have been observed yet in the 3-µm band; our results also show no significant feature in the 3-µm band.

Nemesis
Nemesis is a C-type asteroid and the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ). Nemesis is a slow rotator, taking 77.81 hr for one revolution (Warner et al. 2009). Heterogeneous surface properties were reported for this object by Scaltriti et al. (1979). Our AKARI results show a significant feature in 2.74 µm, which is classified into the sharp group.

Eunike
Eunike is a C-type asteroid and a slow rotator, taking 21.812 hr for one revolution (Warner et al. 2009). Albedo variegation of this object was reported by Pilcher et al. (2014). Fornasier et al. (1999) reported that this object does not exhibit an aqueous alteration signature according to ground-based observations at visible wavelengths. Rivkin et al. (2003) reported an absorption feature of 5.3% towards 2.7 µm, but our results show no feature in the 3-µm band. Note that the signal of this object observed with AKARI was faint (∼ 9 mJy in 3 µm, or APmag = 13.35) and no significant spectral feature was observed above the detection limit. 419 Aurelia Aurelia is a C-type asteroid (or F-type in the Tholen taxonomy). The signal observed by AKARI was faint (∼ 6 mJy in 3 µm, or APmag = 13.73). It has an absorption feature at around 2.8 µm, but it was too noisy and below the detection limit. 423 Diotima Follow-up observations in 1-2.5 µm were performed by Hasegawa et al. (2017), and this object was classified as C-type according to the Bus-DeMeo taxonomy. This object is a member of the Eos collisional family (D. Nesvorny 2012 5 , not in D. Nesvorny 2015 4 ), but is likely to be an interloper based on a spin state analysis (Hanuš et al. 2018). Jones et al. (1990) conducted 3-µm band observations with IRTF and detected no signal associated with hydrated minerals and/or water ice. Our AKARI results show an absorption feature at around 2.79 µm, but its spectral shape is shallower and wider than that of typical hydrated minerals found in other asteroids in this study.

Patientia
Patientia is a C-type asteroid, or Cb-type in the Bus taxonomy (Lazzaro et al. 2004), it has a very flat light curve, indicating a spherical body (Micha lowski et al. (2012) classified the 3-µm band spectral feature of this object into the Europa-like group. Our AKARI results show absorption features at around 3.06 µm, which are classified as the 3-µm dent. A spectral feature appears at around 2.78 µm, but is below the detection limit.

Davida
This object is a C-type asteroid and a member of the Meliboea family (D. Nesvorny 2015 4 ). A shape model of this object was given by Conrad et al. (2007) based on Keck AO observations, and its volume and bulk density are discussed by Viikinkoski et al. (2017). Its bulk density (2.1 ± 0.4 g cm −3 ) suggests some degree of differentiation within the interior of the object. Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the sharp group. Our AKARI results show absorption features at around 2.73 µm and, to a lesser degree, the 3.1-µm feature, which is below the detection limit. Its spectral shape is classified into the sharp group.
• B-and Cb-type asteroids 2 Pallas Pallas is a B-type asteroid and the third largest asteroid in the main-belt region; it is associated with a collisional family (Lemaitre & Morbidelli 1994) that resulted from a cratering event. It is reported that  (2012) discussed that water may play an important role in the thermal-physical evolution of this object. Surface heterogeneity of this object was detected with ground-based telescopes and Hubble observations (Schmidt et al. 2009;Carry et al. 2010).
Our AKARI results show a significant feature found at around 2.74 µm, which appears similar to the sharp group in Takir & Emery (2012).  discussed that the spectral shapes of Ch-type asteroids have the same spectral shape as that of 2 Pallas; i.e., "Pallas-type" spectral group, which is consistent with the presence of phyllosilicates. In the AKARI observations, it is sufficiently bright and (partly) saturates the observed signal on the detector as described in section 2.2.2, while no unusual behavior is found in the spectra compared with the literature.

Interamnia
Interamnia is a Cb-type asteroid with a diameter of 320 km. It is one of the largest bodies in the mainbelt, but is not associated with a family (Rivkin, Asphaug, & Bottke 2014). In the Tholen taxonomy, it is classified as F-type. Polarimetric observations suggest that this object has peculiar surface properties (Belskaya et al. 2005). Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the sharp group. Our AKARI results also show absorption features at around 2.74 µm, which is classified as the sharp group.
• Cgh-and Ch-type asteroids 51 Nemausa Nemausa is a Cgh-type asteroid (or Ch-type in the Bus taxonomy). Fornasier et al. (1999) reported that this object displays an aqueous alteration signature according to ground-based observations at visible wavelengths.  reported a strong absorption feature for this object in the 3-µm band. Our AKARI results also show a significant absorption feature at around 2.77 µm, which belongs to the sharp group. The band depth at 2.7 µm for this object is the strongest among all targets in this study (∼ 47%).
106 Dione Dione is a Cgh-type asteroid, for which Rivkin et al. (2003) reported evidence of hydration in the 3-µm band with a ground-based telescope. Our AKARI results show a significant absorption feature at around 2.76 µm, which is classified as the sharp group.
Egeria is a Ch-type asteroid. Burbine (1998) discussed that the band shape at around 0.7 µm is associated with CM chondrites. Rivkin et al. (2003) reported evidence of hydration in the 3-µm band with a ground-based telescope.  classified the 3-µm band spectral feature of this object into the sharp group, which is confirmed by our AKARI results to be at around 2.76 µm.

Pales
Pales is a Ch-type asteroid and a slow rotator, taking 20.70 hr for one revolution (Warner et al. 2009). No studies to date report observations in the 3-µm band; however, our AKARI results show an absorption feature at around 2.75 µm, which is classified into the sharp group. 50 Virginia Virginia is a Ch-type asteroid or X-type in the Tholen taxonomy, Virginia has no observations in the 3-µm band so far. Conversely, our AKARI results show an absorption feature at around 2.74 µm, which is classified into the sharp group; however, the signal observed with AKARI is faint (∼ 9 mJy in 3 µm, or APmag = 13.21).

Hermione
Hermione is a Ch-type asteroid in the Cybele group, Hermione is a binary system with a moon S/2002 (121) 1 with a diameter of ∼ 12 km (Marchis et al. 2006), which converts to a diameter ratio for this system of 0.06. Takir & Emery (2012) classified the 3µm band spectral feature of this object into the sharp group. Hargrove et al. (2012) found a deep 3-µm absorption feature with IRTF observations and no 10µm emission feature with Spitzer/IRS. Our AKARI results also show an absorption feature at around 2.78 µm, which is classified into the sharp group. Two observations were performed with AKARI at 9.9-h intervals; no significant difference is observed. 127 Johanna Hasegawa et al. (2017) reported that this object is classified as Ch-type in the Bus-DeMeo taxonomy and  reported an absorption feature of this object in the 3-µm band. The signal for this object observed with AKARI is faint (∼ 5 mJy in 3 µm, or APmag = 13.91). One feature appears at around 2.81 µm, which is classified into the sharp group, but it is almost indiscernible from the noise.

Adeona
Adeona is a Ch-type asteroid and the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ). Fornasier et al. (1999) reported an aqueous alteration signature according to ground-based obser-vations at visible wavelengths. The signal according to AKARI is faint (∼ 9 mJy in 3 µm, or APmag = 13.03). One feature appears at around 2.76 µm, which is classified into the sharp group.
• Low-albedo X-complex asteroids 46 Hestia Hestia is an Xc-type asteroid (or P-type in the Tholen taxonomy), Hestia is a slow rotator, with one revolution lasting 21.04 hr (Warner et al. 2009). There are no reports to date of features in the 3-µm band. Fieber-Beyer & Gaffey (2015) discussed the relationship between these objects and CR chondrites based on the vis-NIR observations. Our AKARI results show absorption features at around 2.74 µm, which is classified as the sharp group, and some features at around 3.1 µm, but the spectral shape appears bumpy. The observed signal is faint (∼ 12 mJy in 3 µm, or APmag = 13.42) and its spectrum suffered from noise. Two observations were performed with AKARI at 100-min intervals with no significant difference found between them.

Melete
Melete is an Xk-type asteroid (or P-type in the Tholen taxonomy). No features have yet been observed in the 3-µm band. Our AKARI results show absorption features at around 2.73 µm (sharp). There is also a small feature at around 3.1 µm, which is below the detection limit.

Cybele
Cybele, an Xk-type asteroid (or P-type in the Tholen taxonomy) with a diameter of 300 km, belongs to the Cybele group bearing its own name, which is located in the outer main-belt region. Licandro et al. (2011) made spectroscopic observations with IRTF and found an absorption band centered at 3.10 µm, which is associated with water ice as frost, but no feature associated with hydrated silicates in the 2-4 µm band. The former feature is similar to that of C-type asteroid 24 Themis. Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the rounded group. Our AKARI results also show an absorption feature at around 3.1 µm, but its spectral shape is sharper than previous works. 87 Sylvia Sylvia is an X-type asteroid (or P-type in the Tholen taxonomy), Sylvia is a triple system with two moons, Remus and Romulus, with respective diameters of ∼ 7 km and ∼ 18 km (Marchis et al. 2005). The respective diameter ratios of this system are 0.03 and 0.07. There are no previous reports of spectroscopic observations in the 3-µm band. The signal of this object observed with AKARI is faint (∼ 13 mJy in 3 µm, or APmag = 13.30); it shows some features in the 3-µm band, but they are below the detection limit. 140 Siwa Siwa is an Xc-type asteroid (or P-type in the Tholen taxonomy) and a slow rotator, taking 34.45 hr for one revolution (Warner et al. 2009). No absorption features were found in previous studies (Takir & Emery 2012); i.e., it is a featureless object. Our AKARI results also show no significant features in the 3-µm band.

Hilda
Hilda is an X-type asteroid (or P-type in the Tholen taxonomy) and the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ). Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the rounded group. The signal observed by AKARI is faint (∼ 7 mJy in 3 µm, or APmag = 14.01); thus, we cannot identify any significant features with a sufficient S/N ratio.

Ino
Ino is an Xk-type asteroid and the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ). Jones et al. (1990) reported some hydration of this object. Our AKARI results show an absorption feature at around 2.75 µm, which is classified into the sharp group. A small dent shape of uncertain origin appears at around 2.97 µm.

Lacadiera
Lacadiera is an Xk-type asteroid (or D-type in the Tholen taxonomy). There are no previous reports of hydration detected in the 3-µm band (e.g., Cruikshank et al. 2002). The signal observed by AKARI is too faint (∼ 4 mJy in 3 µm, or APmag = 14.24); thus, no significant features were observed with a sufficient S/N ratio.

Hedwig
Hedwig is an Xk-type asteroid (or P-type in the Tholen taxonomy), Hedwig is a slow rotator, with one revolution lasting 27.33 hr (Warner et al. 2009). No spectroscopic observations have been made in the 3µm band. The AKARI signal is faint (∼ 11 mJy in 3 µm, or APmag = 13.42); however, this object has a sharp spectral feature at around 2.75 µm, which is typically found in C-complex asteroids in AKARI data.
There is no dynamical family associated with this object (Davis et al. 1999). Takir et al. (2017) detected a 3-µm band feature at approximately a ∼ 3% level, possibly associated with water or hydroxyl, which is proposed to have exogenic origins (e.g., Landsman et al. 2018;Avdellidou et al. 2018). In contrast, our AKARI data indicate no significant absorption feature in the 3-µm band. Note that a ∼ 4% band depth at around 3.1 µm is found in our data, but it is below the detection limit. Two observations were performed with AKARI at 100-min intervals with no difference found between them.

Lutetia
Lutetia is an Xc-type asteroid (or M-type in the Tholen taxonomy), its bulk density was determined from the spacecraft Rosetta flyby as 3.4 ± 0.3 g cm −3 (Pätzold et al. 2011). Rosetta also performed spectroscopy of Lutetia (Coradini et al. 2011) and reported no absorption features in the spectral range from 0.4 to 3.5 µm. IRTF observations (Vernazza et al. 2011) also revealed absence of 3-µm features. On the other hand, Rivkin et al. (2011) reported the detection of a 3-µm band feature at 3-5% level in the southern hemisphere, the other side of the asteroid which is not visible to Rosetta Barucci & Fulchignoni 2017). Our AKARI results show no absorption features in the 3-µm band. A small dent shape with a ∼ 2% band depth appears at around 2.7 µm, which might come from contamination due to background stars.

Kalliope
Kalliope is an X-type asteroid (or M-type in the Tholen taxonomy) with an albedo of 0.24 (Usui et al. 2011) and bulk density of 3.35 g cm −3 (Descamps et al. 2008), Kalliope is a binary system with a moon, Linus (22 Kalliope I), with a diameter of 28 km and an orbital period of 3.5954 days (Descamps et al. 2008).
The diameter ratio of this system is 0.2. Rivkin et al. (2000) showed evidence for hydrated minerals on Kalliope with a ∼ 10% band depth in the 3-µm band based on IRTF observations. Our AKARI results show a small feature with a ∼ 4% band depth at around 2.78 µm. Two observations were performed with AKARI at 3.3-hr intervals with no significant difference.

Pandora
Pandora is an Xk-type asteroid (or M-type in the Tholen taxonomy) with a relatively high albedo: 0.34 observed by AKARI (Usui et al. 2011) or 0.20 according to WISE (Masiero et al. 2012). Jones et al. (1990) reported an absorption feature in the 3-µm band, and Rivkin et al. (2000) also showed a hydration feature with a band depth of ∼ 9% and temporal variation in this spectral feature. Our AKARI results show a feature at around 2.8 µm, which is below the detection limit. Three observations were performed with AKARI within 6.6 hr but all of them suffered from contamination due to background stars.

Hesperia
Hesperia is an Xk-type asteroid (or M-type in the Tholen taxonomy). Landsman et al. (2015) reported a 3-µm absorption feature with a band depth of 6.2%. Our AKARI results show a feature at around 2.8 µm that is below the detection limit. Note that only one pointed observation was conducted for this object with AKARI.

Undina
Undina is an Xk-type asteroid. Rivkin et al. (2000) reported a 3-µm absorption feature with a band depth of 9.3%. Our AKARI results show a small feature with a ∼ 6% band depth at around 2.76 µm.

Antigone
Antigone is an Xk-type asteroid (or M-type in the Tholen taxonomy). Rivkin et al. (2000) reported a 3-µm absorption feature with a band depth of 14.4%. Our AKARI results show a peculiar wavy structure related to contamination by background stars (Kmag=13.5 in the data of ObsID:1520147-001, and Kmag=15.7 in the data of ObsID:1520148-001).
135 Hertha Hertha is an Xk-type asteroid (or M-type in the Tholen taxonomy), Hertha is one of the largest members in the Nysa-Polana complex (e.g., Dykhuis & Greenberg 2015). Rivkin et al. (2000) reported a 3µm absorption feature with a band depth of 10.2% and Rivkin et al. (2002) discussed the 0.7 µm ab-sorption of this object. Our AKARI results show a peculiar wavy structure of uncertain origin, which is below the detection limit.

Athor
Athor is an Xc-type asteroid (or M-type in the Tholen taxonomy). Rivkin et al. (2000) reported absence of 3-µm features. The signal observed with AKARI is too faint (∼ 12 mJy in 3 µm, or APmag = 13.31) to identify any significant feature with a sufficient S/N ratio.

Kleopatra
Kleopatra is an Xe-type asteroid (or M-type in the Tholen taxonomy), which is famous for its peculiar bilobate, or "dog-bone" shape (Ostro et al. 2000). This is a triple system with two moons, Cleoselene and Alexhelios, with diameters of ∼ 7 km and ∼ 9 km, respectively (Descamps et al. 2011). Diameter ratios of this system are 0.06 and 0.07. Landsman et al. (2015) showed evidence for hydrated minerals in Kleopatra using IRTF observations; approximately 5% on average, as well as rotational variability in the depth of its 3-µm feature. Our AKARI results show a small feature with a ∼ 3% band depth at around 2.78 µm.
Three observations were performed with AKARI; the first two were separated by 5-hr intervals and the last was conducted 1.46 years after the first two because of observational scheduling. Among these three, no significant spectral differences are observed above the noise level.

Bettina
Bettina is an Xk-type asteroid (or M-type in the Tholen taxonomy). Vernazza et al. (2009) discussed a similarity of spectral shape between this object and mesosiderite, a stony-iron meteorite, at 0.4-2.5 µm.
No observations exist in the 3-µm band. The signal observed by AKARI is faint (∼ 13 mJy in 3 µm, or APmag = 13.31) and no significant absorption feature is found in the 3-µm band. One data (ObsID:1520161-001) is removed by contamination so only one observational data is used for this object.
• High-albedo X-complex asteroids 44 Nysa Nysa is an Xn-type asteroid (or E-type in the Tholen taxonomy) with a high albedo of 0.48 (Usui et al. 2011;Masiero et al. 2012), Owing to its high albedo, the contribution of thermal emission is small at 2.5-5 µm. It is the largest member of the Nysa-Polana family (D. Nesvorny 2015 4 ). Rivkin et al. (1995) detected the 3-µm band depth on Nysa with 14%, which is attributed to hydrated minerals. By removing one data (ObsID:1520189-001) due to contamination as described above, only one observational data from AKARI is used for this object. It shows a peculiar wavy structure of uncertain origin. An unidentified feature is also observed with a ∼ 14% band depth at around 3.08 µm.

Angelina
Angelina is an Xe-type asteroid (or E-type in the Tholen taxonomy) with a high albedo; 0.52 according to AKARI (Usui et al. 2011) or 0.48 for WISE (Masiero et al. 2014). Owing to its high albedo, the thermal emission contribution is small at 2.5-5 µm. There are no reports of observations in the 3-µm band. The signal observed by AKARI is faint (∼ 12-13 mJy in 3 µm, or APmag = 12.83) and its spectrum is flat with no features in the 3-µm band.

D-complex asteroids
D-complex asteroids are darker and located further from the observer; thus, these asteroids are fainter (APmag > 13). One has an absorption feature in the 2.7-µm band, like C-complex asteroids, and the other two do not.

Irmintraud
Irmintraud is a T-type asteroid, and there are a couple of reports suggesting the existence of water on this object (e.g., Lebofsky et al. 1990, Merényi et al. 1997). Kanno et al. (2003) performed near-infrared photometric and spectroscopic observations of this object and concluded that the Tagish Lake meteorite is related to D-type asteroids. The signal observed by AKARI is faint (∼ 5 mJy in 3 µm, or APmag = 14.36) and the data quality is not sufficient to interpret any pattern in the spectral shape at 2.5-3.5 µm, which is almost hidden by noise; thus, no significant features are identified with a sufficient S/N ratio.

Polyxo
Polyxo is a T-type asteroid. Rivkin et al. (2002) reported that most T-types have a 3-µm band feature. Hiroi & Hasegawa (2003) made spectral fitting between Polyxo with ground-based observations and the Tagish Lake meteorite. Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the sharp group. The AKARI signal for this object is faint (∼ 13 mJy in 3 µm, or APmag = 13.34); however, it has a sharp spectral feature with a ∼ 15% band depth at around 2.76 µm, which is typically found in C-complex asteroids in AKARI data.

Bononia
Bononia is a D-type asteroid and belongs to the Hilda group. Takir & Emery (2012) classified the 3-µm band spectral feature of this object into the rounded group. The AKARI signal of this object is faint (∼ 6 mJy in 3 µm, or APmag = 14.26). A pattern is observed in the spectral shape at 2.5-3.5 µm, which is almost indistinguishable from noise; thus, no significant features have a sufficient S/N ratio.

S-complex asteroids
Our AKARI results show that only a few S-complex asteroids have an absorption feature with a few percent band depth in the 3-µm band.

Astraea
Astraea is an S-type asteroid. Jones et al. (1990) observed Astraea with IRTF and reported no absorption feature in the 3-µm band. Our AKARI results also show no significant feature in the spectrum of this object.

Hebe
Hebe is an S-type asteroid. Rivkin et al. (2001) detected a 3-5% band depth in the 3-µm band of this object with UKIRT. This level of feature is below the detection limit of AKARI observations, and our results indicate no significant feature in this object.

Iris
Iris is an S-type asteroid. Rivkin (1997) did not detect any hydrated minerals on this object with IRTF. Our AKARI results also show no significant feature in the spectrum of this object in the 3-µm band.

Polyhymnia
Polyhymnia is an S-type asteroid. No previous observations have been reported for the 3-µm band.
Our AKARI results show a small feature with a ∼ 3% band depth at around 2.68 µm.

Harmonia
Harmonia is an S-type asteroid, and our AKARI results show no significant feature in the spectrum of this object in the 3-µm band.

Eurynome
Eurynome is an S-type asteroid. Our AKARI results show a slightly wavy structure with noise, but no significant feature in the 3-µm band.

Gallia
Gallia is an S-type asteroid and the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ). Gallia is a slow rotator, taking 20.66 hr for one revolution (Warner et al. 2009). Jones et al. (1990) observed Gallia with IRTF and reported no absorption feature in the 3-µm band. Our AKARI results show a wavy structure with noise, but no significant feature in the 3-µm band.

Bohemia
Bohemia is an S-type asteroid. The signal observed by AKARI is faint (∼ 6 mJy in 3 µm, or APmag = 13.89) and the data quality is insufficient to indicate a pattern in the spectral shape at 2.5-3.5 µm, which is almost hidden by noise; thus, no significant features are identified with a sufficient S/N ratio.

Herculina
Herculina is an S-type asteroid. Jones et al. (1990) observed Herculina with IRTF and reported no absorption feature in the 3-µm band. Our AKARI results also show no significant feature in the spectrum of this object in the 3-µm band.
8 Flora Flora is an Sw-type asteroid and has a high spectral slope at 0.45-2.45 µm. Eaton et al. (1983) observed Flora with UKIRT and found a steep slope of the spectrum at 3-4 µm. This spectral shape was considered a continuation of that in shorter wavelengths. Our AKARI data also has a positive spectral slope of S = 0.102 µm −1 , which is higher than the mean value of S-complex asteroids. A small dent shape of uncertain origin with a ∼ 6% band depth appears at around 3.09 µm. Note that only one observation was conducted for this object with AKARI.
89 Julia 89 Julia is an Sw-type asteroid, the largest member of the asteroid family bearing its name (D. Nesvorny 2015 4 ), and considered a parent body of the quasicrystal-bearing CV meteorite Khatyrka (Meier et al. 2018). Our AKARI results show a slightly wavy structure, but no significant feature in the 3µm band. It has a large positive spectral slope from 2.5 µm to 3.5 µm (S = 0.112 µm −1 ).

Asporina
Asporina is an A-type asteroid, this object is considered a member of the rare olivine-dominated asteroids (e.g., Sanchez et al. 2014). In our AKARI data, a small dent shape with a ∼ 9% band depth appears at around 3.08 µm; however, the AKARI signal is faint (∼ 11 mJy in 3 µm, or APmag = 13.82) and the data quality is not sufficient for further analysis.

Eleonora
Eleonora is an A-type asteroid. There is a possibility that the surface materials of this object contain a pallasite assemblage . In our AKARI data, a small dent shape of uncertain origin with a ∼ 4% band depth appears at around 3.06 µm.

Metis
Metis is an L-type asteroid (or S-type in the Tholen taxonomy). Our AKARI results show a bumpy structure at around 2.9 µm, which might be attributable to be contamination by background stars. These are below the detection limit; thus, no obvious features are found in our AKARI data.

Aquitania
Aquitania is an L-type asteroid (or S-type in the Tholen taxonomy), Aquitania is a slow rotator, taking 24.14 hr for one revolution (Warner et al. 2009). From polarimetric observations (Masiero & Cellino 2009), it is considered a member of the so-called Barbarians (Cellino et al. 2006), which represent some of the oldest surfaces in the solar system. Our AKARI results show a bumpy structure at around 3.1 µm, which might come from contamination due to background stars. These are below the detection limit, and thus no obvious features are found in our AKARI data. 42 Isis Isis is a K-type asteroid (or S-type in the Tholen taxonomy). An unidentified feature with a ∼ 8% band depth is observed at around 3.09 µm in AKARI data.

Advantages and limitations of the AKARI spectroscopic observations
The greatest advantage of the AKARI observations is free from disturbance by telluric absorption and thus able to obtain spectra continuously from 2.5 µm to 5 µm, which can fully cover the peak wavelength of the 2.7-µm band.
On the other hand, the effective aperture size of the AKARI telescope is 68.5 cm and the exposure time of each pointed observation is only 10 min in total, which is limited by severe constraints of the attitude control. Thus the present survey provides a flux-limited sample. The detec-tion sensitivity is ∼ 1.3 mJy at 3 µm with 10σ (see the IRC Data User Manual). Note that the sensitivity during Phase 3, the warm mission phase, is likely worse than this because of the temperature change due to degradation of the cryocooler (Onaka et al. 2010). Specifically, the number of hot pixels on the detector increases in Phase 3. To constitute data redundancy, data from two or three pointed observations are averaged to generate reflectance spectra in this study. Spectroscopic observations in the "Np" window are a type of slitless spectroscopy. The distribution of the target positions in the "Np" window in the reference image is shown in figure 20. This "Np" window is effective for observations of point sources, but can be vulnerable to contamination by neighboring stars and ghost images due to nearby bright sources. Our observations were scheduled to avoid the region at around the galactic plane, i.e., our targets were located at galactic latitudes of |b| > 16 • . Insufficient flat-fielding is also an issue for the present dataset (see section 2.2).

Identification of the 3-µm band shape
In this study, the 3-µm band shape is classified into sharp, w-shape, and 3-µm dent (table 5). This is determined by a combination of D2.7 and D3.0 described in section 3.1. This criterion assumes that asteroid spectra contain no narrow or steeply peaking features like atomic or molecular line spectra, but have broad features of typically ∼ 0.1 µm or wider in the 2.7-or 3.1-µm band.
Classification of the 3-µm band shape is not yet fully established compared to the taxonomies based on the vis-NIR spectra (e.g., Tholen 1984;Bus & Binzel 2002;Lazzaro et al. 2004;DeMeo et al. 2009). Takir & Emery (2012) and  classified 3-µm band shapes into sharp, rounded, Europa-like, and Ceres-like (hereafter Takir class), while Rivkin et al. (2012) classified them into Pallas-like, Ceres-like, Themis-like, and Lutetia-like (hereafter Rivkin class). These classifications are summarized in table 6. Our classification is generally consistent with Takir class and Rivkin class. Observations with the space telescope in this study detected the 2.7-µm band feature without atmospheric disturbance, which is a big help to classify asteroids into groups of the 3-µm band spectral shape.
There are five asteroids among our targets that are classified into the sharp group of the Takir class: 13 Egeria (Ch), 121 Hermione (Ch), 308 Polyxo (T), 511 Davida (C), and 704 Interamnia (Cb). Takir & Emery (2012) reported that this spectral shape is most consistent with the presence of hydrated silicates. Certainly, our AKARI results for these objects show significant absorption features at 2.7 µm. It should be noted that small features appear at around 3.1 µm, which might be obscured by the strong features at 2.7 µm. There are also four asteroids among our targets that are classified into the roundedshape of the Takir class: 24 Themis (C), 65 Cybele (Xk), 153 Hilda (X), and 361 Bononia (D). Based on AKARI data, 24 Themis has a wider absorption feature at both 2.7 µm and 3.1 µm and is thus classified into the w-shape in this study. 153 Hilda and 361 Bononia were not observed with a sufficient S/N with AKARI. Ceres-like objects include 1 Ceres (C) itself and 10 Hygiea (C). These objects display two clear features at 2.7 µm and 3.1 µm with comparable band depths, which belong to the wshape. Europa-like objects include 52 Europa (C) itself and 451 Patientia (C). Europa seems to have two features at 2.7 µm and 3.1 µm, but the former is too weak to be classified as the w-shape. Patientia also has a low S/N spectrum in our data.
In general, it is difficult to accurately classify an object into the featureless category. To make reliable classification, it is necessary to ensure that no features exist within a certain level of accuracy, which requires high S/N data throughout the wavelength of interest. The reflectance spectra obtained in this study include remnant wavy structures on a level of a few percent; thus, it is difficult to distinguish real features from these structures. There are three candidates for featureless spectra in the AKARI data, which have a band depth of less than 2% at 3-µm: 6 Hebe (S-type) 140 Siwa (Xc-type), and 532 Herculina (S-type). Note that 140 Siwa is classified as featureless according to the Takir class (Takir & Emery 2012), which does not show any feature above their noise level in the 3-µm band. 21 Lutetia is reported to have a broad absorption feature in the 3-µm band in the Rivkin class unlike Pallas, Ceres, or Themis . Our AKARI data, on the other hand, is truncated at λtrunc = 3.57 µm, and thus the broad spectral shape cannot be fully covered. Featureless spectra can constrain the abundance of surface materials, especially once they are combined with other spectral data. For example, Emery & Brown (2003) reported featureless spectra of 20 Trojan asteroids in the 3-µm region observed with IRTF and these featureless are attributed to the presence of anhydrous silicates (Emery & Brown 2004).

Peak wavelength in the 2.7-µm band feature
As seen in figure 13, the peak wavelength of the 2.7-µm band feature is concentrated at around 2.75 µm. In particular, C-complex asteroids have a trend between the peak wavelength and the band depth with S/N > 2. Figure 21 shows this trend for 17 C-complex asteroids. There are four outliers with longer peak wavelengths: 24 Themis at 2.76 µm, 121 Hermione at 2.81 µm, 127 Johanna at 2.85 µm, and 423 Diotima at 2.79 µm. Except for these four, there is a correlation between the peak wavelength and the band depth among 13 C-complex asteroids: where D2.7 is in units of percent (%) and λ peak is the peak wavelength in units of µm. The correlation coefficient is 0.88.
The correlation should be used with caution because only 13 asteroids comprise this trend; thus, it may be affected by observational bias. Nevertheless, this correlation may be important for considering asteroid spectra in meteorite research. Based on heating experiments of meteorites in the laboratory, Yamashita et al. (in prep) reported a peak-wavelength shift in the 2.7-µm band of hydrated minerals because of the dehydration process. The band depth of the 2.7-µm band indicates an abundance of phyllosilicate, and the peak wavelength of the 2.7-µm band indicates the Mg/Fe ratio in phyllosilicate. During the dehydration process by heating, phyllosilicate decreases and the Mg/Fe ratio simultaneously increase because of the progressive replacement of the phyllosilicates interlayer cations Fe 2+ by Mg 2+ (e.g., Rubin et al. 2007;Beck et al. 2010;Nakamura et al. 2017). This leads to a decrease of the band depth along with a peak wavelength shift toward shorter wavelengths. Therefore, equation (15) can be interpreted in terms of the meteorite dehydration history.
Some reports detected hydrated minerals on the lunar surface using in-situ observations from spacecrafts: Cassini (Clark 2009), Deep Impact (Sunshine et al. 2009), and Chandrayaan-1 (Pieters et al. 2009). These observations found peak wavelengths near 2.8-3.0 µm. Similar absorption was also detected in the lunar soil of Apollo 16 and 17 (Ichimura et al. 2012). It is considered that the mechanisms forming these lunar hydrated minerals are different from those on asteroids: they may be associated with interaction of the lunar regolith with solar wind proton implantation. These minerals appear to have absorption features near 2.8-3.0 µm. On the other hand, phyllosilicate absorption caused by aqueous alteration in meteorites appears in the vicinity of 2.7 µm (e.g., . Features found in most asteroids observed with AKARI are likely associated with this aqueous alteration.

Spectral slope in the 3-µm band
The spectral slope measured in the available wavelength range is shown in figure 19. The mean value of the spec-tral slope of total 64 objects is 0.047 ± 0.047 µm −1 , which is almost flat at these wavelengths, indicating that the thermal component subtraction described in section 2.2.3 works well; if the subtraction was insufficient, it may have caused a significant slope toward longer wavelengths. The mean value of the slope of each complex is: S(C) = 0.000 ± 0.031 µm −1 , S(X) = 0.063 ± 0.041 µm −1 , and S(S) = 0.059 ± 0.045 µm −1 . Spectral slope makes a major contribution to derive a taxonomic classification of asteroids base on the vis-NIR spectral data (e.g., DeMeo et al. 2009). Our results in this study do not conflict with the trend at shorter wavelengths, that is, C-complex asteroids have a general trend of the flat slope, and X-, S-complex asteroids have a trend of the medium to steep slope. For S-type asteroids, it is known that the space weathering causes a significant change in vis-NIR spectra, in which the albedo decreases (the spectrum becomes darker) and the spectral slope increases (redder) (Yamada et al. 1999). For the space weathering of C-complex asteroids, many laboratory experiments have been conducted (for recent studies, e.g., Matsuoka et al 2017;Lantz et al. 2017). However, it is still in debate whether the space weathering makes the spectral slope of asteroids other than S-type bluer or redder, and darker or brighter.

From dry to wet: characteristics of hydrated asteroids
In the field of meteorite study, it is widely agreed that CM, CI, and CR carbonaceous chondrites, which originate in Ccomplex asteroids, experienced aqueous alteration in their parent bodies (Brearley 2006 and references cited therein). This aqueous alteration occurred from the reaction of anhydrous rock and liquid water with the isotope decay heat (McAdam et al. 2015 and references cited therein). Correspondingly, from the standpoint of astronomical observations, hydration in asteroids has been discussed with a 0.7-µm absorption feature; C-complex asteroids with an absorption feature in the broad 0.7-µm band are categorized as Ch-or Cgh-type (Bus & Binzel 2002), where suffix "h" represents a hydrated subclass (e.g., Burbine & Bell 1993). This 0.7-µm feature is indicative of an Fe 2+ → Fe 3+ charge transfer transition in oxidized iron in phyllosilicates formed through aqueous alteration processes (e.g., Vilas & Gaffey 1989). On the other hand, absorption in the 2.7-µm band is attributed to OH-stretch in hydrated minerals and is a more direct indicator of the presence of water. Figure 22 shows the number of C-complex asteroids observed in the 0.7-and 2.7-µm band in this study (c.f., Rivkin et al. 2002. Figure 23 shows the distribution of band depths in the 0.7-and 2.7-µm bands. The 0.7-µm band depth is measured from the visible spectra compiled by Hasegawa et al. (2017).  discussed the correlation presented in the distribution of the 0.7-and 2.7-µm band depths based on the combined set of asteroid and meteorite measurements. No clear trends were found in figure 23, partly due to a lack of adequate samples for the statistical study. Fornasier et al. (2014) pointed out the relationship between presence/absence of the 0.7-and 2.7-µm band features as; if the 0.7-µm band is present on the spectrum, it is always accompanied by the 2.7-µm feature. On the other hand, even if the 0.7-µm band is not found, the 2.7-µm band may still be present on the spectrum. In our measurements, the 0.7-µm band depth is generally weaker by a few percent than that of the 2.7-µm band. Our results confirm those of Fornasier et al. (2014); in addition, not a small number of objects are found without both the 0.7-and 2.7-µm band features. Rivkin (2012) reported that 60-70% of C-complex asteroids are expected to have hydrated minerals. Our results also indicate that 77% of the observed C-complex asteroids have the 2.7-µm feature, although it may suffer from observational bias.
To discuss observational signatures with the aqueous alteration, we classify asteroids into three stages in this study as: • "dry" objects have no feature both in the 0.7-and 2.7µm bands, • "moist" objects have feature only in the 2.7-µm band, • "wet" objects have features both in the 0.7-and 2.7-µm bands.
It should be stressed that the terms "moist" and "wet" are figurative and not literal meaning; it does not mean the presence of water in the form of vapor or liquid, but describe the relative amount of H2O or OH in minerals. Note that water vapor was detected on 1 Ceres (classified into "moist") with the Herschel Space Observatory (Küppers et al. 2014). Figure 22 can be interpreted in terms of the sequence of aqueous alteration and dehydration process on asteroids with the three stages described above as: 1. An asteroid is "dry" if the signature of aqueous alteration does not appear on its surface -it may be formed in an environment without the presence of water, i.e., dry condition. 2. As aqueous alteration progresses to a certain degree, it appears as the 2.7-µm band feature. It becomes a "moist" (or moderately wet) object. 3. When aqueous alteration progresses sufficiently, the spectrum shows the 0.7-µm feature and the object becomes "wet". The progress of aqueous alteration does not necessarily progress completely to this stage, and the reaction may stop halfway without the appearance of the 0.7-µm feature. 4. Dehydration occurs by solar-radiation heat or impactinduced heat. As dehydration progresses, the 0.7-µm feature disappears and the asteroid changes from the "wet" to the "moist" stage. 5. As dehydration progresses further, the 2.7-µm feature also disappears and the object returns to the "dry" stage.
Let it be added, in this context, that a certain amount of evidence from meteorite study suggest the aqueous alteration of carbonaceous chondrites (e.g., Brearley 2006). These carbonaceous chondrites are considered to originate from C-complex asteroids, or more specifically, formed inside C-complex asteroids (e.g., McSween 1999). In other words, hydrated minerals in C-complex asteroids, which show the 2.7-µm band feature, are considered to be of an internal (endogenic) origin. This suggests that C-complex asteroids were formed in the environment where anhydrous rock and water ice existed together at low temperature in the protoplanetary disk at the time of planetesimal formation. On the other hand, the degree of aqueous alteration in ordinary chondrites is found to be relatively small (e.g., Brearley 2006). It is consistent with that only a few Scomplex asteroids show signatures of aqueous alteration in this study.
The situation of X-and D-complex asteroids is more complicated. The major reason why interpretation of these asteroids is difficult is that there are few spectral counterparts of meteorites found to X-or D-complex asteroids (c.f., Kanno et al. 2003;Hiroi & Hasegawa 2003). Nevertheless, it can be considered that low-albedo Xcomplex (i.e., P-type) and D-complex asteroids may possess properties similar to C-complex in the sense of primitive objects (e.g., Vernazza et al. 2015). Thus it is natural that some of these asteroids have signatures of hydration as seen in some stages of C-complex asteroids as the 2.7-µm band feature in this study (e.g., 56 Melete, 476 Hedwig, and 308 Polyxo).
There is an exceptional case found from our results that 349 Dembowska (R-type; olivine-rich, basaltic asteroid, Leith et al. 2017) has a 2.7-µm band feature with 4%. It is considered to originate from external (exogenic) materials, which was brought to its surface by hydrated impactors or created by solar wind interactions with silicates. Recent studies also reported that exogenic materials are detected in such traditionally dry asteroids, for example, 4 Vesta (V-type; McCord et al. 2012;De Sanctis et al. 2012), 16 Psyche (M-type: Takir et al. 2017;Avdellidou et al. 2018), and 433 Eros and 1036 Ganymed (S-type; Rivkin et al. 2018). This "contamination" of exogenic materials may be not rare in the main-belt region.

Summary
We conducted a near-infrared spectroscopic survey with the AKARI satellite to obtain reflectance spectra of 66 asteroids in the 2.5-5 µm range. These observations successfully fill the gap in the 2.5-2.85 µm data, which cannot be observed with ground-based telescopes. Based on the spectra obtained with AKARI, we found that most C-complex asteroids have clear absorption features related to hydrated minerals at a peak wavelength of approximately 2.75 µm, while no S-complex asteroids have clear absorption in this wavelength.
This data set, comprising direct observations of absorption features in the 2.7-µm band, is quite unique. This study will provide important information on whether asteroid features determined by spacecraft exploration are universal or exceptional. Combining this data set with spectra in shorter wavelengths (< 2.5 µm), and comparing it to meteorite spectra measured in the laboratory are both interesting research subjects. These will be discussed at length in future papers. Our spectral data are summarized in the Asteroid Catalog using AKARI Spectroscopic Observations (AcuA-spec) and open to the public on the JAXA archive 6 . band, which is used to derive the wavelength reference position in the spectral images. (b) Spectroscopic image with the grism (NG). One pointed observation consists of four spectroscopic frames, one reference frame, and four or five spectroscopic frames (total number of frames depends on the attitude stability of the satellite). The target is placed on the 1 ′ × 1 ′ "Np" window to avoid contamination from background stars. (c) Spectrum extracted from the target using the toolkit. The red, blue, and green curves denote the thermal component of the spectrum calculated by the NEATM that is removed to derive the reflectance spectrum, the spectrum of the reflected sunlight component, and the total modeled spectrum, respectively. (d) Reflectance spectrum normalized at 2.6 µm. Spectrum beyond the truncated wavelength (λtrunc ) cannot be used due to uncertainty of the thermal model (see section 2.2.4). The green dotted line denotes the continuum and the arrow shows the point where the band depth is measured. ObsID:1520188-001 1000 500 5000 5000 5000 Fig. 4. AKARI IRC 2.5-5.0 µm spectra of 1 Ceres, 2 Pallas, and 4 Vesta of each pointed observation. The red and blue dots denote the long exposure and short exposure data, respectively. The gray region denotes the unreliable wavelength range determined by inconsistency between the long and short exposure data.