Challenges in Identifying Artificial Objects in the Near-Earth Object Population: Spectral Characterization of 2020 SO

Since the dawn of the Space Age, hundreds of payloads have been launched into heliocentric space. As near-Earth object (NEO) surveys search deeper for small asteroids, more artificial objects in heliocentric orbits are being discovered. We now face a challenge to identify the true nature of these objects and avoid contaminating the NEO catalog. Here, we present the methods used to characterize one such object. 2020 SO was discovered by the Pan-STARRS1 survey on 2020 September 17. Originally classified as a NEO, the object’s artificial nature became evident due to its low velocity relative to Earth and solar radiation pressure affecting its orbit about the Sun. Based on a backward propagation of its orbit, 2020 SO is thought to be a Centaur rocket body (R/B) from the launch of the Surveyor 2 mission to the Moon. We characterized 2020 SO using a range of ground-based optical and near-infrared telescopes to constrain its true nature. We find that its reflectance spectrum is consistent with that of other Centaur R/B launched during a similar time frame, and we identify 1.4, 1.7, and 2.3 μm absorption bands consistent with polyvinyl fluoride used on the aft bulkhead radiation shield exterior of Centaur-D R/B at the time.


Introduction
Since the Luna 1 spacecraft was sent toward the Moon in 1959, several rocket upper stages have been disposed into heliocentric orbits with the rationale that these objects would not return to Earth's vicinity anytime soon.Sixty years later, we are now rediscovering these long-lost objects as our efforts to discover smaller near-Earth objects (NEOs) continue to ramp up.Planetary defense surveys routinely discover objects whose true nature (artificial versus natural) is unknown.These objects are primarily in cislunar space, which includes the volume of space between Earth and the Moon, orbits around the Moon, and lunar-transfer orbits that may extend far beyond the Moon's orbit.It is worth noting that this region of space includes distances over 10 times farther than geostationary Earth orbits (GEOs) and therefore a volume 1000 times larger than the previous scope of space situational awareness (Holzinger et al. 2021).
When the Vera Rubin Observatory comes online in 2025, it will discover dozens of NEOs each night, on average, and will likely increase the chances of these misidentifications in cislunar space (Jones et al. 2015).The problem is exacerbated by upwards of a hundred lunar missions currently planned to launch in the next decade. 13,142020 SO is a recent example of one such object that was misidentified as an NEO, with other examples including XL8D89E,15 2018 AV2, 16 2023 NM,17 and J002E3.Object J002E3 was the first heliocentric-orbiting, artificial object to be characterized with rotational periods and visible spectra through an extensive observational campaign in 2002 (Lambert et al. 2004).
2020 SO was discovered on 2020 September 17 by the Pan-STARRS1 planetary defense survey.Its motion was initially regarded as unremarkable and 2020 SO was reported to the Minor Planet Center (MPC) as a NEO candidate. 182020 SO was removed from the MPC listings as the object's status as an artificial body became evident due to its low velocity relative to Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Earth and changes to its orbit around the Sun due to solar radiation pressure. 19The backward propagation of 2020 SO's orbit puts its closest approach to Earth in 1966 late-September, with a predicted distance close enough that the object may have originated from the Earth at this time.This time frame for the closest approach to Earth correlates strongly with the launch of NASA's Surveyor 2 mission to the Moon, which is now considered to be the origin of 2020 SO. 20 The Surveyor 2 mission was intended to land and characterize the lunar surface ahead of the Apollo missions, but a thruster failure caused the payload to crash into the Moon and sent the upper stage of the rocket into heliocentric space. 20Although the dynamics of 2020 SO are strongly suggestive of its artificial nature, characterization information in the form of lightcurves and spectroscopy would be the definitive proof necessary to identify the true nature of 2020 SO.
2020 SO made two passes near the Earth during its 4 monthlong stay in cislunar space.During this time, 2020 SO became bright enough for characterization with multiple techniques and instruments to study the object.The first manuscript to characterize the lightcurve of 2020 SO was published by Campbell et al. (2022) and determined the object's spin state and used photometric data to find a rotation rate, estimated to be 9.328 ± 0.275 s at a 2σ confidence level.Bondarenko & Marshalov (2022) measured a radar albedo for 2020 SO of 0.95, a rotation rate of ∼9.5 s, estimated the object's size as approximately 10 m long and 3 m wide, and estimated a circular polarization ratio of ∼0.31, consistent with surface irregularities at centimeter sizes.The results from these studies support the classification of the object as an artificial body.In this study, we present spectral characterization of 2020 SO as diagnostic evidence of its artificial nature and identify key absorption bands that may assist future identifications of artificial materials without detailed modeling efforts.The data discussed here shows that near-infrared (NIR) spectroscopy (0.7-2.5 μm) is a powerful tool for differentiating between natural and artificial objects as unique absorption bands are present that best align with artificial materials and differ from absorption bands seen on asteroids.

Observation Campaign
The close encounters with 2020 SO placed the object in near-Earth space for 4 months, allowing progressively better data to be collected over time.Several telescopes were utilized to collect data before it returned to heliocentric orbit in 2021 March.During the observation window, 2020 SO's brightness varied from approximately visual magnitude 21.5 to 14, with brightness peaks occurring in 2020 November and 2021 January.Different sensor systems were utilized to gather as much information within the constraints of the object's brightness at that time.Figure 1 shows how 2020 SO's brightness and phase angle varied over time, with the dates and instruments of observations for various stages of this project demarcated with vertical lines.
Due to the faintness of the object, initial characterization was performed with photometric observations taken by the Large Binocular Telescope (LBT) located at Mount Graham outside Safford, AZ (see Table 1).This system consists of twin 8.4 m mirrors and was used to collect 60 s Sloan Digital Sky Survey (SDSS) g, r, i, and z filter images with the Large Binocular Camera (LBC; Giallongo et al. 2008;Speziali et al. 2008;Rothberg et al. 2016).Images were obtained in a g − r − g − r filter sequence in the LBC Blue system and i − z − z − i filter sequence in the LBC Red system to minimize potential lightcurve effects on color estimates.
Typical CCD reduction processes of dark and bias subtraction and flat-field division were used for the images from LBC (e.g., Howell 2006).Each image was plate solved with astrometry.net and a 2D background map was constructed via Photutils's median-background estimator (Lang et al. 2010;Bradley et al. 2019).The summed flux was collected for the target aperture in the background-subtracted frame using a 3σ aperture radius to increase the signal-to-noise ratio of the data (Howell 1989).Instrumental magnitudes were calculated using the standard magnitude equation and then calibrated by calculating the zero-point magnitude (the offset from instrumental to catalog magnitude) from at least 10 (typically >20) field stars in each frame from the SDSS Data Release 12 (DR12) catalog (Alam et al. 2015).Additional code packages for source detection, flux estimation, and FITS file manipulation were used in Python including astroquery, CCDProc, and astropy (Robitaille et al. 2013;Craig et al. 2017;Price-Whelan et al. 2018;Ginsburg et al. 2019).The weighted mean magnitude was calculated for each filter and the uncertainty was estimated as the square root of the inverse of the sum of the weights (e.g., Selvin 2015).For these calculations, the weights were taken as the uncertainty in the zero-point magnitude for each frame.These data were used to estimate the visual spectral slope (R = 4) of the object while it was too faint to be observed by smaller systems or with higher-resolution spectroscopy (e.g., R ∼ 100).
To make comparisons between the 2020 SO spectral slope observed by the LBT and potentially similar objects (derelict Centaur stages in Earth orbit), visible spectra of comparison objects were collected with the 0.6 m Robotic Automated Pointing Telescope for Optical Reflectance Spectroscopy (RAPTORS) I system (see Table 1) on the University of Arizona campus in Tucson, AZ (Battle et al. 2022b).The system is equipped with a 30 line mm −1 diffraction grating and a Finger Lakes Instrumentation Proline 4710 deep depletion CCD.This combination allows RAPTORS I to take lowresolution (R ∼ 30) visible spectra (0.438-0.950 μm) of targets to approximately visual magnitude 15.Solar-analog stars are observed each night to produce reflectance measurements.The system and its data-processing methodology are described in Battle et al. (2022b).
As 2020 SO became brighter, it was possible to obtain low-resolution (R ∼ 100) NIR (0.69-2.54 μm) spectra with the SpeX instrument on NASA's 3.2 m Infrared Telescope Facility (IRTF) located on Maunakea, Hawai'i (see Table 1; Rayner et al. 2003).Observations were performed with a 0 8 slit oriented along the parallactic angle to minimize differential atmospheric refraction.Spectra were obtained in an ABBA nodding pattern to allow sky background subtraction between frames.A local, G-type standard star was observed before and after each target and roughly every hour during observations of a specific target to correct for telluric absorption bands.To obtain reflectance measurements and correct the spectral slope of the standard stars, a solar analog was observed each night.These spectra were processed using SpexTool, and solar-analog stars were observed each night to obtain reflectance spectra of 2020 SO (Rayner et al. 2003;Vacca et al. 2003;Cushing et al. 2004).The IRTF was also used to collect spectra of a known Centaur-D rocket body (R/B) for comparison to 2020 SO as supporting evidence of its artificial nature.
Other sensor systems were utilized for a photometric study to estimate 2020 SO's rotation rate, spin state, and radar surface properties.In this study, we present only the data related to spectroscopy and spectrophotometry.The systems and their observational circumstances for the data used in this study are presented in Table 1.Known artificial objects are listed by their unique North American Aerospace Defense (NORAD) ID number.

Visible-wavelength Results
Visible-wavelength spectrophotometry was acquired as an initial characterization while 2020 SO was still too faint for higher-resolution spectroscopy.This allowed a spectral slope to be estimated from calibrated measurements of the target's brightness in each of the SDSS g, r, i, and z filters.Astronomical color indices are then calculated by subtracting the brightness of the object in two filters and are typically indicated with the bluer wavelength first (e.g., the g − r color index).The color indices used for the Sun21 were estimated using listings from SDSS DR12 (Alam et al. 2015), which The brightness axis is inverted so that brighter magnitudes (smaller numbers) appear higher.Data were retrieved from JPL Horizons prior to the object's removal as an NEO from their database.(2006).Normalizing to the object's g-filter magnitude, m g , the reflectance at the central wavelength, R(λ), of any other observed filter, F, can be calculated using Equation (1), where m F,e indicates the Sun's magnitude in the indicated filter: Equation (1) can be derived from the magnitude equation, and variations of it are seen in the literature such as in DeMeo et al. (2009b).Performing this calculation for each observed filter produces four reflectance values that serve as a low-resolution color "spectrum."For all visible spectra in this study, we normalize the reflectance to 0.7 μm since this wavelength provides overlap with the NIR measurements and is a central wavelength to the visible-wavelength systems.Because none of the SDSS filters has a central wavelength of 0.7 μm, a line was fit across the reflectance values of the four filters in order to estimate R(0.7 μm) for normalization.
Figure 2 shows the result of the reflectance computed using the measured spectrophotometry from LBT along with the average asteroid reflectance values for the Bus-DeMeo S-and D-type asteroids (DeMeo et al. 2009a;Slivan 2013).S-type asteroids were chosen as the most common spectral type among near-Earth asteroids and D-types were chosen as the near-Earth asteroid spectral type with the highest spectral slope in an attempt to match the steep spectral slope observed on 2020 SO (DeMeo et al. 2009a;Binzel et al. 2019).Vertical bars represent the 1σ uncertainty for the measured magnitudes of 2020 SO and the 1σ variation among each asteroid taxonomy in the Bus-DeMeo system (DeMeo et al. 2009a;Slivan 2013).
Although no features are discernible from this color "spectrum" of 2020 SO, it lacks any flattening in the reflectance between the i-and z-band filters that might imply an absorption band near 0.9 μm associated with S-type asteroids.The spectral slope is steeper than the average for the D-type asteroids, indicating that 2020 SO has a redder spectral slope than most planetary materials.Although centaur planetary bodies and very rare main-belt asteroids may have comparable or redder spectral slopes than D-types, we would not expect to find these in near-Earth space and do not make comparisons with these to 2020 SO (Hasegawa et al. 2021).These measurements were the first spectral evidence that 2020 SO might not be an asteroid.
To verify the proposed dynamical and photometric link between 2020 SO and the Surveyor 2 Centaur-D R/B, three other Centaur-D R/Bs in Earth orbit were observed.These objects were chosen to be the same model of Centaur upper stage and were launched during a similar time frame as the proposed Surveyor 2 origin of 2020 SO.Each R/B observed is identified with a unique NORAD ID, with object 3598 being launched in 1968 December, 4882 launched in 1971 January, and 6155 launched in 1972 August. 22igure 3 shows the visible spectra of these comparison objects measured with RAPTORS I. Variations at the shortest and longest wavelengths are due to phase angle effects.The Centaur R/B slopes follow the expected trend with the highestphase object having the steepest slope (object 3598), consistent with phase reddening effects known to occur in asteroid spectra (Sanchez et al. 2012).The comparison Centaur R/Bs are observed at relatively high phase angles (∼50°and higher) due to their low-Earth orbits, while 2020 SO is observed at both moderate phase angles (∼20°-25°) and near opposition (∼5°).The overall slope of the Centaur-D R/B spectra is a closer match to 2020 SO than that of the D-type asteroids, further supporting the hypothesis of the link between 2020 SO and the Surveyor 2 mission.The fact that 2020 SO's spectral slope remains steeper than the observations of the high-phase comparison R/Bs implies another factor, such as space aging (similar to space weathering on asteroids), may be the dominating contributor to 2020 SO's high spectral slope.

Near-infrared Results
Although the spectrophotometry and comparison observations of Centaur-D R/B slopes is useful, spectral slopes are not diagnostic of surface-material composition and can be affected by many alteration effects such as phase angle and space aging (Sanchez et al. 2012;Pieters & Noble 2016;Pearce et al. 2020;Battle et al. 2022a).Both planetary and artificial materials produce absorption bands at NIR wavelengths (0.69-2.54 μm) that can help identify the origin and composition of the  observed object.By quantifying the spectral characteristics of 2020 SO, we find key features to identify future objects with unknown origins.

Spectral Comparisons
Three common visible and NIR absorption bands are particularly useful for discriminating different planetary materials: a 0.7 μm phyllosilicate hydration band, a ∼0.9-1.0 μm absorption due to Fe 2+ in olivine and Ca 2+ in pyroxene, and a ∼2.0 μm absorption band due to Fe 2+ in pyroxene (Adams 1974;Cloutis et al. 1986;Vilas & Gaffey 1989;Gaffey et al. 1993;Ostrowski et al. 2010).Ultimately, many of the materials found on planetary surfaces are comprised of silicates and have similar spectra to one another with well-characterized variances.Conversely, artificial materials make up a much wider range of material types including paints, metals, photovoltaic cells, multilayer insulation, and more.Not many studies have been dedicated to the foundation of a comprehensive database of spectra of these artificial materials, but some relevant studies have focused on polyethylene terephthalate (or PET) materials, solar cell material, solar cell cover glass, aluminum, copper, and thermal control paints measured with various degrees of space-aging effects in a simulated GEO environment (Plis et al. 2020(Plis et al. , 2021;;Reyes et al. 2021;Shah et al. 2023).The lack of a unified spectral database along with the increased number of potential material combinations and appropriately estimated alteration effects to consider makes comparisons of artificial materials challenging (Bengtson et al. 2022).This study expands upon the number of telescopic visible and NIR spectra which may better inform the needs of future laboratory studies.
The first night of NIR data obtained on 2020 SO is shown in Figure 4 along with the visible spectrophotometry from LBT, visible spectra of the three Centaur-D R/Bs, and the average visible-NIR spectrum of D-type asteroids from Bus-DeMeo taxonomy (DeMeo et al. 2009a;Slivan 2013).The spectrum of 2020 SO has a steep spectral slope from 0.69 to ∼1.5 μm, which is consistent with the LBT measurements.At wavelengths longer than ∼1.5 μm, there is a sharp change in the reflectance to a less steep spectral slope.No obvious absorption bands are discernible, which further supports the claims that this is not a natural, planetary body since we might expect to see a ∼1.0 or ∼2.0 μm absorption band for many asteroid types.The lack of absorption bands does not exclude asteroid types with featureless spectra, but the earlier spectral slope comparison makes this less probable for all asteroid types.Additionally, most asteroid visible and NIR spectra do not have a sharp change in the spectral slope.A D-type asteroid in the Bus-DeMeo taxonomy can have a "gentle kink" or curvature in spectral slope near 1.5 μm, and is the best match for this spectrum when analyzed with the SMASS online tool (DeMeo et al. 2009a;Slivan 2013). 23The spectral slope of 2020 SO is still steeper than the average D-type asteroid, however, as shown in Figure 2, and the change in spectral slope is much sharper than the average D-type asteroid, as shown in Figure 4.
The spectrum shown in Figure 4 was taken while 2020 SO was near the limits of the IRTF's capabilities (visual magnitude ∼19.3), resulting in a noisier spectrum.As 2020 SO came closer to the Earth, it became much brighter, including observations at magnitude 14.9 on 2020 November 30 at a distance of ∼153,000 km.The increase in target brightness allowed for a higher signal-to-noise ratio spectrum than previously observed.Figure 5 shows all three sets of NIR spectra observed with the IRTF and their associated phase angles during observations.The major absorption bands seen in the spectra and their approximate wavelengths are also indicated with vertical arrows.
The higher signal spectra shown in Figure 5 show a rise in reflectance in the range of ∼0.7-1.6 μm with a neutral slope at longer wavelengths.The visible colors obtained by the LBT are consistent with the NIR data from the IRTF.In the November 29 and November 30 spectra, absorption bands are visible near approximately 1.4, 1.7, and 2.3 μm.To further characterize these features, the band centers and band depths were measured following the procedures of Sanchez et al. (2012Sanchez et al. ( , 2020)), and the band parameters are shown in Table 2.

Material Identification
NASA historical documents on the Centaur R/B systems identify the primary materials used on the exterior of the Centaur-D R/Bs used in the launch of the Surveyor 2 mission (Convair Aerospace Division of General Dynamics & Martin Marietta Aerospace 1973).These materials include a stainlesssteel fuselage plus a rigid radiation shield assembly of laminated nylon fabric insulated with aluminized mylar on the interior and white polyvinyl fluoride (PVF) covering the exterior.Figure 6 shows the stainless-steel fuselage with the white PVF covering the aft bulkhead of a Centaur-D R/B being prepared for launch. 24 sample of PVF was acquired in order to make direct comparisons between 2020 SO and the materials known to be on the Centaur-D R/Bs.Spectra were obtained with an ASD LabSpec 4 Hi-Res spectrometer, which has 3 nm resolution at 0.70 μm and 6 nm resolution at 1.40 and 2.10 μm.The sample was illuminated at a 0°incidence angle by a 120 W quartztungsten bulb, while the reflected signal was measured at a reflected angle of 30°.Spectra were measured relative to a baseline disk and corrected for dark current.The semitransparent PVF sample was placed over a Spectralon substrate only the PVF absorption bands would be measured and contributions from the substrate would be calibrated out.Data were processed with a Python script that removed a known infrared feature of Spectralon and corrected for detector sensitivity offsets at 1.0 and 1.8 μm (Kokaly et al. 2017).The spectrum of stainless steel from Jorgensen (2000) and Jorgensen et al. (2001) was acquired with a similar method and instrument to what we have described here.A single, stray data point at 1 μm has been removed when compared to the literature spectrum.The stainless-steel spectrum has some ringing near 1.4 μm likely from specular reflections exceeding the detector's dynamic range.The visible-NIR spectra of these two end-members are shown in Figure 7.
The red spectral slope observed on 2020 SO is broadly consistent with the red slope of stainless steel, although 2020 SO's spectrum has a steeper slope than pure stainless steel.This is likely due to a combination of phase angle and spaceaging effects that are difficult to discern.The PVF lab spectrum shows a sharp rise in reflectance over wavelengths up to 0.5 μm followed by a negative spectral slope at longer wavelengths.Four absorption bands are present in the laboratory PVF spectrum, and the position of the three most prominent (deepest) absorption bands are shown in Table 2 along with the corresponding band parameter measurements for 2020 SO.Band parameters were measured using a custom Python code described in Sanchez et al. (2020) that follows the procedures outlined in Cloutis et al. (1986).Band centers are calculated by dividing out a linear fit across the peaks on either side of the band and then fitting a 3rd-order polynomial to the bottom third of the band.Band depth is calculated as the percent drop between this linear-fit continuum and the reflectance of the band center, as described in Clark & Roush (1984).Band parameter uncertainties are estimated by measuring the band parameter 10 times and calculating the standard deviation.Because 2020 SO is rotating much faster (∼9.5 s) than the integration time for IRTF spectra (200 s), the observed spectrum is expected to be an average of the properties of the materials on the object's surface.
In order to make a better comparison to a known object, the IRTF was used to observe a Centaur-D R/B (NORAD ID 4882) in NIR wavelengths to verify the PVF absorption bands detected on 2020 SO. Figure 8 shows the NIR spectrum of 2020 SO compared with Centaur-D R/B 4882.As noted before, differences between the spectral slopes can come from a combination of phase angle and space-aging effects.While the two objects were launched at similar times, 2020 SO spent  Object 4882 lacks the absorption bands at shorter wavelengths, however the 2.29 μm band is nearly identical to 2020 SO's absorption band in both band center (2.30 ± 0.01 μm for object 4882, 2.290 ± 0.010 μm for 2020 SO) and band depth (6.0% ± 0.1% for object 4882, 6.3% ± 0.1% for 2020 SO).

Discussion
Our analysis of 2020 SO has shown that spectroscopy can be a powerful tool for differentiating between artificial and natural objects and confirms that 2020 SO is a Centaur-D R/B instead of a natural body such as an asteroid.This hypothesis is also supported by other studies that find the object to be a fast rotator with a cylindrical shape, both of which are consistent with a R/B (Campbell et al. 2022).Area-to-mass ratio estimates from astrometric measurements also point to 2020 SO being an artificial object due to the increased effects of solar radiation pressure on the object's orbit. 25,26Radar measurements show the object has a very high radar albedo, implying a metallic surface consistent with a R/B fuselage, and its size approximately matches that of a Centaur R/B (Bondarenko & Marshalov 2022).
Visible spectra and spectrophotometry can be used to identify spectral slopes that are too steep to likely be from natural objects.NIR spectra show the absence of the prominent 1 and 2 μm absorption bands due to olivine and pyroxene and identify absorption bands for artificial materials on the object.If objects are bright enough for spectrometry, this method can provide faster differentiation between natural and artificial objects than the long observational arcs required for area-tomass ratio measurements.Specifically, the ∼2.3 μm absorption band is indicative of artificial materials, and we believe it can be used to conclusively differentiate natural from artificial objects in space.This method also provides more information about what materials might be on the surface than radar, which will only indicate a likely metallic surface.
Absorption bands consistent with PVF were found in the NIR spectrum of 2020 SO and a nearly identical absorption band was seen in the NIR spectrum of Centaur-D R/B 4882, helping solidify the conclusion that 2020 SO is a Centaur-D R/ B from the Surveyor 2 mission.More NIR spectra of known spacecraft on orbit should be made to understand if these features are common among spacecraft and to better differentiate between natural and artificial satellites.
Although we were able to identify primary materials on 2020 SO from historic documentation, this will not be the case for all  objects.One major limitation is the availability of spacecraft material spectra that include multiple phase angles and spaceaging states to better recreate observed telescopic spectra.Future work could include spectral-mixing modeling to determine the ratios of each material present in the spectra of these objects, though not unless the laboratory data are available to support that work.

Summary
We present results from our detailed characterization observational campaign for 2020 SO using the LBT for spectrophotometry, RAPTORS I for visible spectroscopy, and the NASA IRTF for NIR spectroscopy.Here, we summarize our findings and present evidence supporting that 2020 SO is a Centaur-D R/B from the Surveyor 2 mission.
1. LBT spectrophotometric observations show a spectral slope at visible wavelengths that is redder than the average for any near-Earth asteroid taxonomic type.2. Visible spectroscopy of Centaur-D R/Bs still in Earth orbit show they have similar spectral slopes as 2020 SO. 3. The increased spectral slope of 2020 SO compared to the Centaur-D R/Bs launched at a similar time potentially highlights the increased space-aging effects of heliocentric space.4. NIR spectra taken with the NASA IRTF show that 2020 SO lacks the 1 and 2 μm absorption bands seen in the spectra of many silicate-dominated asteroids.5. Laboratory spectra of PVF were found to have absorption bands that are consistent with those in 2020 SO's spectrum.PVF was a spectrally dominant component on Centaur R/Bs and was used for thermal insulation.6.The NIR spectrum of an Earth-orbiting Centaur-D R/B, the same model as 2020 SO, was measured and found to have a similar spectral shape and the same 2.3 μm PVF absorption band as 2020 SO. 7. We successfully use visible and NIR spectra to differentiate between a natural and artificial object.8.The 2.3 μm absorption band may be a useful diagnostic absorption band critical for differentiating between artificial and natural bodies and may still be detectable at moderate signal-to-noise ratios.9. Future work will benefit from a larger spectral database of known artificial objects as well as laboratory studies of spacecraft materials at different phase angles and spaceaging stages.

Figure 1 .
Figure 1.Phase angle (top) and brightness (bottom) of 2020 SO over time with UTC dates of campaign observations marked by vertical lines that span both subplots.The brightness axis is inverted so that brighter magnitudes (smaller numbers) appear higher.Data were retrieved from JPL Horizons prior to the object's removal as an NEO from their database.

Figure 2 .
Figure 2. Photometric color spectrum of 2020 SO taken with the LBT compared to the average visible spectrum for S-and D-type asteroids in the Bus-DeMeo taxonomy.Error bars represent 1σ uncertainties and all spectra are normalized at 0.7 μm.

Figure 3 .
Figure 3. Visible spectra of 2020 SO compared to several Centaur-D R/Bs, normalized at 0.7 μm.Variations in the slope of the Centaur spectra are likely due to phase angle variations between observations.

Figure 4 .
Figure 4. Visible spectra of 2020 SO and several Centaur-D R/Bs overlaid on the NIR spectrum of 2020 SO, showing overall agreement in the visible wavelengths.The average Bus-DeMeo D-type asteroid visible-NIR spectrum from DeMeo et al. (2009a) is also shown for comparison.All of the spectra are normalized at 0.7 μm.

Figure 5 .
Figure 5. Near-infrared spectra of 2020 SO taken during its approach and normalized at 1.5 μm.The phase angle (PA) during observations is indicated in the legend.Vertical arrows point to the approximate location of observed absorption bands.

Figure 6 .
Figure 6.Photograph of a Centaur upper stage being installed and readied at the General Dynamics factory in 1962.The photo shows the stainless-steel body and engines with the white polyvinyl fluoride on the aft bulkhead radiation shield.Image credit: NASA GRC.

Figure 7 .
Figure 7. End-member spectra for two materials found on Centaur-D R/Bs.Stainless steel (left) fromJorgensen (2000) andJorgensen et al. (2001) shows a red spectral slope with a broad, subtle 1 μm feature.The double peak near 1.3-1.4μm is due to detector dynamic range limitations and is not a real feature of stainless steel.PVF (right) shows a blue spectral slope after 0.5 μm with strong absorption bands near 1.7 and 2.3 μm plus fainter bands near 1.2 and 1.4 μm.Both spectra are normalized at 1.5 μm.Band parameters for the three deeper PVF bands are given in Table2.

Figure 8 .
Figure 8. Near-infrared spectra of 2020 SO and a Centaur-D R/B showing similar features.Both spectra are normalized at 1.5 μm.Two absorption bands from PVF consistent with those seen on 2020 SO are indicated with vertical arrows.

Table 1
Observational Circumstances for Characterizing 2020 SO and Associated ObjectsNotes.Phase angle is given in degrees.Range is given in kilometers and lunar distances (LD).The average of the east-west component of the phase angle is calculated for objects with an * due to their large movement across the sky during observations.2020 SO's rotation rate is ∼9.3-9.5 s; the rotation rates for Centaur R/B in low-Earth orbit are unknown.The solar analog SAO 93936 was used for all IRTF data, and standard stars are listed.
a Data collected with RAPTORS I use only one pointing of the solar-analog star, which is listed in place of the standard star for that night.utilizes transformations from Bilir et al. (2005) and Rodgers et al.

Table 2
Absorption Band Centers Measured for the Three Deepest Bands in PVF and 2020 SO's NIR Spectrum of its time in heliocentric space and may have experienced a harsher radiation environment, leading to more of a spaceaged surface (e.g., Reitz 2008; Restier-Verlet et al. 2021). most