Visible to Near‐Infrared Reflectance Spectroscopy of Asteroid (16) Psyche: Implications for the Psyche Mission's Science Investigations

Abstract

The NASA Psyche mission will explore the structure, composition, and other properties of asteroid (16) Psyche to test hypotheses about its formation. Variations in radar reflectivity, density, thermal inertia, and visible to near‐infrared (VNIR) reflectance spectra of Psyche suggest a highly metallic composition with mafic silicate minerals (e.g., pyroxene) heterogeneously distributed on the surface in low abundance (<10 vol.%). The Psyche spacecraft's Multispectral Imager is designed to map ≥80% of the surface at high spatial resolution (≤20 m/pixel) through a panchromatic filter and provide compositional information for about ≥80% of the surface using seven narrowband filters at VNIR wavelengths (∼400–1,100 nm) and at spatial scales of ≤500 m/pixel. We analyzed 359 reflectance spectra from samples consistent with current uncertainties in Psyche's composition and compared them to published reflectance spectra of the asteroid using a chi‐square test for goodness of fit. The best matches for Psyche include iron meteorite powder, powders from the sulfide minerals troilite and pentlandite, and powder from the CH/CBb chondrite Isheyevo. Comparison of absorption features support the interpretation that Psyche's surface is a metal‐silicate mixture, although the exact abundance and chemistry of the silicate component remains poorly constrained. We convolve our spectra to the Imager's spectral throughput to demonstrate preliminary strategies for mapping the surface composition of the asteroid using filter ratios and reconstructed band parameters. Our results provide predictions of the kinds of surface compositional information that the Psyche mission could reveal on the solar system's largest M‐type asteroid.

Key Points

• Visible to near‐infrared spectra of (16) Psyche are consistent with meteorites (irons and metal‐rich chondrites) and sulfides

• The Psyche mission's Multispectral Imager can identify and potentially discriminate such materials if present on the surface of Psyche

• Imager‐convolved data indicate that the instrument can accurately recover absorption band parameters in certain metal‐silicate mixtures

1. Introduction

Asteroid (16) Psyche is the largest of the M‐type asteroids and is the target of the NASA Psyche mission (Elkins‐Tanton et al., 2020). Within the last few decades, multiple studies have observed the surface of Psyche at ultraviolet to infrared wavelengths and have generally interpreted the surface to be dominated by Fe‐Ni metal with a minor (<10 vol.%) component of low‐Fe, low‐Ca pyroxene (e.g., DeMeo et al., 2009; Fornasier et al., 2010; Hardersen et al., 2005; Ockert‐Bell et al., 2010). Since NASA's selection of the Psyche mission in early 2017, there has been renewed interest in visible to near‐infrared (VNIR) spectroscopic study and interpretation of the asteroid's surface mineralogy (e.g., Becker et al., 2020; Cantillo et al., 2021; Dibb et al., 2022; Sanchez et al., 2017; Takir et al., 2017). In addition to spectroscopic observations, density estimates and measurements of both radar reflectivity and thermal inertia of Psyche are consistent with the presence of core‐like materials, such as iron meteorites and sulfide minerals like troilite (FeS) (Cambioni et al., 2022; de Kleer et al., 2021; Elkins‐Tanton et al., 2020; Landsman et al., 2018; Matter et al., 2013; Shepard et al., 2021; Siltala & Granvik, 2021).

A synthesis of observations of Psyche indicates that plausible bulk compositions could be far less metallic (∼30–55 vol.% metal) than previously thought, with significant pore space and a still unknown non‐metal component (Elkins‐Tanton et al., 2020, 2022). New studies since selection do not alter the mission's primary hypothesis, which is that the asteroid is a remnant core of a differentiated planetesimal that had its silicate mantle mostly stripped in a series of hit‐and‐run style collisions (e.g., Asphaug, 2010; Asphaug & Reufer, 2014; Sarid et al., 2015). Other hypotheses for the formation of Psyche include the reaccretion of mostly core materials from a catastrophically disrupted planetesimal, or preservation of a primitive, undifferentiated body with high metal content (Elkins‐Tanton et al., 2022). If Psyche indeed contains materials from a planetesimal core, it offers an unparalleled opportunity to explore the nature and evolution of planetary interiors. Data returned from the mission will offer strong constraints on models of solar system evolution and greatly inform our understanding of terrestrial planet formation.

To test these hypotheses, the Psyche spacecraft will be equipped with a set of science instruments that includes the Psyche Multispectral Imager system (henceforth called the “Imager”; Bell et al., 2016; Elkins‐Tanton et al., 2022). The Imager is equipped with one broadband and seven narrowband science filters at visible to near‐infrared wavelengths (∼400–1,100 nm) chosen to study the asteroid's surface geology, composition, and topography (Table 1; Bell et al., 2016; Dibb et al., 2018). The spacecraft's Gamma Ray and Neutron Spectrometer (GRNS) will measure the elemental composition of Psyche's surface at high energy resolution but relatively low spatial resolution, while the Imager will acquire high spatial resolution (∼20 m/pixel) information about the asteroid's surface mineralogy. Planned mapping orbits at 704, 293, 174, and 71 km average altitude correspond to estimated ideal pixel scales of approximately 35, 15, 9, and 4 m/px, respectively, for imaging with the broadband filter. Integration of data from the Imager and GRNS will be critical for understanding the surface composition of Psyche.

Table 1.

Psyche Imager Bandpasses

#	ECAS name a	Center wvl. (nm)	Effective center b (nm)	FWHM c (nm)	Justification
0	‐‐	‐‐	‐‐	‐‐	ND5 “Sun safe” blocking filter
1	CLR	540	516	280	Unfiltered CCD for OpNav, topography, and geology
2	b	437	442	50	Sulfide detection and blue component of true color
3	o	495	498	25	Sulfide detection
4	v	550	553	25	Sulfide detection and green component of true color
5	w	725	727	40	Typical peak reflectance continuum and red component of true color
6	x	850	850	50	Search for Fe‐bearing silicates and other 1,000 nm features
7	p	948	946	50	Search for Fe‐bearing silicates and other 1,000 nm features
8	z	1,041	1,015	90	Search for Fe‐bearing silicates and other 1,000 nm features

^a Filter names that are comparable to those from the Eight Color Asteroid Survey (Zellner et al., 1985).

^b Effective centers are computed after convolution of each bandpass with Imager quantum efficiency, solar emission, and representative spectra of Psyche.

^c Full width at half maximum.

2. Materials and Methods

2.1. Natural and Synthetic Laboratory Samples and Mixtures

To support future interpretation of Imager data, we compiled a library of 359 reflectance spectra from possible analog materials that fall within the range of expected Psyche surface minerals, given the hypotheses and uncertainties surrounding the asteroid's origin (Elkins‐Tanton et al., 2020, 2022). These materials include a variety of meteorites (irons, pallasites, mesosiderites, CB chondrites, ordinary and enstatite chondrites, aubrites, a diogenite and lodranite, and anomalous, ungrouped metal‐rich chondrites). We include spectra from sulfide minerals (troilite (FeS), pyrrhotite (Fe_1‐xS), pentlandite ((Fe,Ni)₉S₈), oldhamite (CaS), and a mixture of chalcopyrite (CuFeS₂) and pentlandite) and mixtures of iron meteorite metal and troilite (Dibb et al., 2021, 2022). We also included spectra of iron meteorites, enstatite chondrites, mixtures of silicate minerals, and mixtures of silicates and iron meteorite powder from the University of Winnipeg Planetary Spectroscopy Facility (Cloutis et al., 2009, 2010; Horgan et al., 2014; Izawa et al., 2010). We included reflectance spectra from the NASA RELAB Facility at Brown University database of synthetic olivine and pyroxene of a variety of compositions in anticipation of areas of differing silicate chemistry or abundance (Dyar et al., 2009; Klima et al., 2007). Finally, we included additional spectra of synthetic and natural sulfide minerals from the RELAB database to account for differences in spectra of natural and synthetic sulfides (e.g., Helbert et al., 2013; Varatharajan et al., 2019).

2.2. Sample Preparation and Measurements

The comparison of laboratory spectral data to telescopic spectral data is highly dependent on properties of both the measurement (primarily phase angle between the incident light source and detector but also ambient lab conditions like temperature and humidity) and the sample (composition and grain size or surface roughness). An informed comparison of laboratory and telescopic data requires documentation of these properties, especially for spectral data from public sources. The source, name, physical form and properties (e.g., hand sample, powder, slab), mineralogy, and measurement phase angle of the spectra in our library are shown in Table S1.

For spectral data collected at Arizona State University, we collected visible to near‐infrared (350–2,500 nm) reflectance spectra from samples loaned from the ASU Buseck Center for Meteorite Studies. Spectra were measured using an ASD FieldSpec 4 Standard‐Res spectroradiometer. The spectral sampling of this instrument is 1 nm, while the spectral resolution is 3 nm in the visible to near‐infrared (350–1,000 nm) and 10 nm in the shortwave infrared (1,000–2,500 nm). Spectra were measured relative to a Spectralon white reference target in air at a phase angle of 30°. We measured spectra of strongly specularly reflecting samples at an incidence angle of 38° and emission angle of 8° to avoid stray light artifacts from the order‐sorting filter in the FieldSpec 4 instrument. This has negligible impact on reflectance, spectral slope, or band parameters. For some samples, spectra were collected at additional phase angles of 15° and 45° to explore trends in spectral parameters with phase angles. For hand samples, slabs, and gravels, we created average spectra representative of the bulk sample by rastering across the available surface of each sample, given that the surface was visibly free of weathering, oxidation, or other contamination.

Several meteorites and minerals in the ASU collection were powdered as part of research to support spectral interpretations of metal‐rich asteroids (e.g., Dibb et al., 2021, 2022). These include sulfide minerals, iron meteorites, and a piece of the metal‐rich CH/CBb chondrite Isheyevo. Details for the production and verification of powder from these samples is given by Dibb et al. (2021, 2022). In summary, two iron meteorites (Gibeon and Sikhote‐Alin) were manually ground with diamond‐coated files, while the sulfides and the piece of Isheyevo were comminuted in a stainless steel impact mortar and pestle. The contents of the powders were examined for contamination both under a microscope and using powder X‐ray diffraction, which both indicated minimal contamination (Dibb et al., 2022). Spectra were collected from unsorted powders (i.e., all grains less than ∼1 cm) and from <75 μm to >75 μm sieved fractions. This grain size is consistent with hypothesized grain sizes of the surface of (16) Psyche (e.g., Landsman et al., 2018). Spectra of powder samples were measured with the powder in an aluminum sample cup. We rotated the sample cup at 90° increments, collecting spectra through a full rotation. We then emptied the powder onto clean weighing paper before repeating the process, then averaged all spectra to create a representative bulk spectrum.

2.3. Visible to Near‐Infrared Telescopic Spectral Data of (16) Psyche

Table 2 lists studies within the past several decades that have collected spectral data of Psyche at visible to near‐infrared wavelengths (e.g., DeMeo et al., 2009; Fornasier et al., 2010; Hardersen et al., 2005; Ockert‐Bell et al., 2008; Sanchez et al., 2017). These data show that the spectrum of Psyche is red‐sloped and relatively featureless, with a subtle absorption feature at ∼1,000 nm. This feature is interpreted as being caused by a relatively low abundance of low‐Fe, low‐Ca orthopyroxene (i.e., enstatite). Uncertainties on reported band centers allow possible band centers between ∼913 and ∼960 nm and band depths ranging from ∼0% to 2.9%.

Table 2.

Spectral Measurements and Properties of (16) Psyche

Study	Observation date (UTC time)	Solar phase angle (°)	Wvl. range (nm)	Visual albedo	Band 1 center (nm)	Band 1 depth (%)
SMASS/DeMeo et al. (2009)	2000 October 09	13.35 a	435–2,500 b	n.r.	n.r.	n.r.
Ockert‐Bell et al. (2010)	2005 November 15	9.9, 10.2	435–3,000 b	0.19	950 ± 10	0.015 ± 0.003
Fornasier et al. (2010)	2004 November 15 (19:31)	22.13 a	400–2,400	0.12	949 ± 8	2.9
Hardersen et al. (2005, 2011)	2002 March 22	7.12	750–2,500	0.1203	932 ± 8	∼1
Sanchez et al. (2017)	2015 December 08 (10:30)	1.8	690–2,500	n.r.	940 ± 10	1.3 ± 0.1
	2015 December 08 (11:43)	1.8	690–2,500	n.r.	940 ± 5	1.4 ± 0.1
	2015 December 08 (12:47)	1.8	690–2,500	n.r.	935 ± 10	1.0 ± 0.1
	2015 December 09 (14:01)	1.7	690–2,500	n.r.	940 ± 3	1.2 ± 0.1
	2016 February 10 (5:10)	19.5	690–2,500	n.r.	935 ± 4	1.3 ± 0.1
	2016 February 10 (5:30)	19.5	690–2,500	n.r.	919 ± 6	1.0 ± 0.1
	2016 February 10 (6:34)	19.5	690–2,500	n.r.	919 ± 3	1.3 ± 0.1
	2016 February 10 (8:05)	19.5	690–2,500	n.r.	928 ± 6	1.5 ± 0.1

^a Not reported (n.r.) but determined based on reported observation date using the JPL HORIZONS tool.

^b Used data from the SMASS II data set (Bus & Binzel, 2002).

Studies of the variation in pyroxene band center with changes in pyroxene chemistry suggest that the chemistry of the purported pyroxene on Psyche could vary significantly more than previously reported. Cloutis and Gaffey (1991) report that pyroxenes with Ca contents between 0 and 30 mol.% and Fe contents <65 mol.% have band centers between ∼910 and 950 nm. Similarly, spectra of synthetic pure Fe‐pyroxene from Klima et al. (2007) have band centers at wavelengths lower than 950 nm. To account for such uncertainty, we include pyroxenes in our spectral library samples with a wider range of chemistries than what has been reported for Psyche.

2.4. Analysis of Spectral Data

Once the library was assembled, we compared laboratory spectra with telescopic observations of Psyche using a chi‐square curve‐matching method and by comparing measured depths and centers of any absorption features. We then convolved the library spectra to Imager filter bandpasses and explored spectral characteristics that could support future Imager operations and mapping. All data processing was performed in Python 3.6.

Our first investigation was to compare the full wavelength range of each spectrum of Psyche to the laboratory spectra. We first normalized both asteroid and sample spectra to 1.0 at 750 nm. This wavelength was chosen because it exists in all the published spectra of Psyche, is typically a local maximum in reflectance spectra of Fe‐bearing silicates, and lies within a high signal‐to‐noise ratio region of the lab spectrophotometer used to collect data in this study. Next, we computed a chi‐square goodness‐of‐fit statistic using the following formula:

$${\chi }^2=\frac{1}{N_w}\sum _{i=1}^{N_w}\frac{{\left(R_i-w_i\right)}^2}{w_i}$$

where R _i and w _i are the reflectance at wavelength i of the meteorite and asteroid reflectance spectra, respectively, and N _w is the total number of wavelengths used. The use of similar chi‐square metrics to assess goodness‐of‐fit is well‐established in asteroid spectral studies (e.g., Fornasier et al., 2010; Nedelcu et al., 2007; Ockert‐Bell et al., 2010) and offers a simple way to filter a spectral library for best matches. In this context, the chi‐square metric is not used to evaluate a null hypothesis using a p‐value, but instead quantifies the difference between the spectra of Psyche and those in our library. It is essentially the same as the sum of squared residuals. Thus, a meteorite spectrum with the minimum chi‐square value for a certain spectrum of Psyche represents the best fit spectrum in our library. It is important to note that this metric weights every wavelength of the two spectra equally, and thus is not sensitive to specific absorption features. We propagate the measurement uncertainty for the asteroid and meteorite spectra to derive an uncertainty for the chi‐square metric. To match albedo values for Psyche, we filtered our library for spectra with >6% and <20% reflectance at 550 nm. The “reflectance‐filtered” library used in our chi‐square analysis contains 160 spectra. The chi‐square goodness‐of‐fit metric for all spectra in the full library is available in Table S2.

Further mineralogical interpretation of an asteroid's surface requires a comparison of the center and depth of any absorption features present in the spectrum to those in laboratory spectra (Gaffey et al., 2002). We report the depth and center of any ∼1,000 nm absorption features in the spectra in our library in Table S1. These parameters were calculated using the methods of Clark and Roush (1984) and Cloutis et al. (2015), which are depicted graphically in Figure 1. First, visually estimated shoulder wavelengths of any absorption feature were used to fit a straight line continuum across the absorption feature (dashed red lines in Figures 1a and 1b). The continuum was removed by dividing the sample's absolute reflectance by this continuum line (blue lines in Figures 1c and 1d). Similar to Cloutis et al. (2015), we fit a third order polynomial to the 20 wavelengths on either side of the minimum value of the continuum‐removed features (orange dashed lines in Figures 1c and 1d). The minimum value of the third order polynomial was taken to be the continuum‐removed band center. Band depth was determined using Equation 32 of Clark and Roush (1984):

$$D_b=\frac{\left(R_c-R_b\right)}{R_c}$$

where D _b is band depth, R _c is reflectance of the continuum at the band center, and R _b is the continuum‐removed reflectance at the band center. This process is repeated 50 times for each absorption feature in each spectrum, using a normally distributed set of shoulder wavelengths centered on the original input shoulder wavelengths with a variance of 10 nm. We report the mean and standard deviation of the 50 calculated values for each absorption feature in each spectrum (Table S1).

Figure 1.

Many of the spectra in our library contain absorption features at ∼1,000 nm that we compared to similar features in spectra of (16) Psyche. A spectrum of the diogenite Shalka (black line in a) exhibits strong absorption features, while spectra from the CH/CBb chondrite Isheyevo (black line in b) exhibit a weak feature comparable to Psyche. We divide the spectra by a linear continuum fit across the absorption feature (dotted red lines in panels [a and b]) and fit a third order polynomial to the minimum 40 values (dashed orange line in panels [c and d]) to determine band parameters (see text for details). For clarity, in panels (c and d) the third order polynomial is extended out to the minimum 100 values of the continuum‐removed absorption feature.

2.5. Diagnostic Spectral Parameters

To support Imager science goals, we developed a set of parameters that can be used to identify major compositions that could potentially be found on the surface of Psyche based on potential meteorite analogs and previous telescopic observations. First, we convolve the high‐resolution laboratory spectral data to the Imager's spectral throughput using the following equation:

$$I_{{\lambda }_1,{\lambda }_2}=\frac{\underset{{\lambda }_1}{\overset{{\lambda }_2}{\int }}R\left(\lambda \right)\ast QE\left(\lambda \right)\ast T\left(\lambda \right)\ast L_{sun}\left(\lambda \right)d\lambda }{\underset{{\lambda }_1}{\overset{{\lambda }_2}{\int }}QE\left(\lambda \right)\ast T\left(\lambda \right)\ast L_{sun}\left(\lambda \right)d\lambda }$$

where R(λ), QE(λ), T(λ), and L _sun (λ) are the sample reflectance, detector quantum efficiency, filter and telescope transmission, and solar radiance at wavelength λ, respectively. This process effectively downsamples the laboratory spectral data into an estimate of what the Imager would “see” if each library spectrum occurred on the surface of Psyche. To emulate probable measurement uncertainty of the Imager, we uniformly vary the reflectance at each wavelength of each input laboratory spectrum by 2%, perform the convolution 1,000 times, and then use the mean and standard deviation of the convolved values. An uncertainty of 2% is consistent with other flight instruments with similar designs and electronics to the Psyche Imager (e.g., Bell et al., 2021).

Because of variable band depths and centers of the spectra in our library, we developed three methods for measuring silicate absorption features in the Imager‐convolved spectra. These are graphically depicted in Figure 2. The first method (Figure 2a) is effective for spectra with strong absorption features centered near ∼948 nm (e.g., Fe‐rich pyroxene). In this case, we fit a cubic spline to the convolved data from the four longest wavelength filters (725, 850, 948, and 1,041 nm). We then generated a flat “continuum” connecting the convolved point at 725 nm to a point on the cubic spline longward of 1,041 nm that had the same reflectance as the 725 nm convolved point. Then, we divided the values of the cubic spline by the flat continuum. We used the wavelength of the minimum value of the quotient as the band center and computed the band depth as the difference between the minimum continuum‐removed reflectance value and 1.0.

Figure 2.

Three methods for measuring band depths and centers in spectra convolved to Imager bandpasses. High resolution laboratory spectra (gray lines) are downsampled to multispectral resolution data (blue diamonds) during the convolution process. See text for discussion.

The second method (Figure 2b) is similar to the first but is useful for spectra with strong near‐infrared slopes (e.g., mixtures dominated by metal). In this case, we fit a linear continuum between the convolved values for the 725 and 1,041 nm filters. We then divided a cubic spline fit to the four near‐infrared filters by this continuum. Similar to the first method, we assume the wavelength of the minimum value of the quotient as the band center and computed the band depth as the difference between the minimum continuum‐removed value and 1.0.

The third method (Figure 2c) is simpler and is useful for spectra with absorption features centered at wavelengths longer than ∼1,000 nm (e.g., Fe‐bearing olivine). First, the convolved data are normalized to unity at the 725 nm filter value. The band depth is then computed as the difference in the normalized convolved values for the 725 and 1,041 nm filters.

3. Comparison of Spectra of (16) Psyche to Laboratory Spectra

3.1. χ² Goodness‐Of‐Fit Comparison

The five laboratory spectra with the lowest chi‐square goodness‐of‐fit value for each spectrum of Psyche are listed in Table 3. This number of best‐fitting spectra was chosen to be shown in Table 3 to be consistent with past analyses using the chi‐square method (e.g., Nedelcu et al., 2007) and to show the spread of different compositions with spectra similar to Psyche (see Table S2 for all chi‐square values for each sample compared to each spectrum of Psyche). Brief descriptions of the samples, their physical form (grain size/surface roughness), and the phase angle of the measurement are also listed in Table 3. Consistent with Dibb et al. (2022), we find that spectral data of Psyche are quantitatively consistent with iron meteorite powders and slabs, sulfide powders and slabs, and powder from the CH/CBb chondrite Isheyevo (Table 3).

Table 3.

Best Chi‐Square Matches for (16) Psyche for Spectra in Our Library After Filtering for Reflectance >6% and <20% at 550 nm

(16) Psyche spectrum reference	ID/Spectrum name	Sample name (type)	Grain size (μm)/surface roughness	Phase angle (i/e)	χ2 (×106)
SMASS/DeMeo et al. (2009)	30_06	Isheyevo powder (CH/CBb)	>75	23/8	290 ± 2
	mar0207.060	MET01A (iron met.)	<45	60/−30	390 ± 29
	mar0207.036	MET01A (iron met.)	<45	60/60	400 ± 31
	30_02	Isheyevo powder (CH/CBb)	<1,000	38/8	420 ± 3
	MPMIX 1011	MET101A (iron met.)	<45	35/0	450 ± 31
Hardersen et al. (2005)	mar0707a.004	MET01A (iron met.)	<45	30/60	180 ± 23
	30_07	Isheyevo powder	<75	53/8	190 ± 7
	feb1607.006	MET02A (iron met.)	<45	30/0	390 ± 37
	feb1607.009	MET03A (iron met.)	<45	30/0	400 ± 34
	mar0107.004	MET01A (iron met.)	<45	0/30	400 ± 37
Fornasier et al. (2010)	25_07	Gibeon powder (iron met.)	>75	23/8	840 ± 23
	25_05	Gibeon powder (iron met.)	>75	38/8	1,250 ± 29
	39_05	Sikhote‐Alin powder (iron met.)	>75	23/8	1,400 ± 27
	39_07	Sikhote‐Alin powder (iron met.)	>75	53/8	1,430 ± 29
	feb1607.007	MET02B (iron met.)	45–90	30/0	1,790 ± 77
Sanchez et al. (2017)—2015 December 08 10:30	35_02	Auburn FeS powder (sulfide)	<1,000	38/8	140 ± 4
	35_03	Auburn FeS powder (sulfide)	<75	38/8	200 ± 6
	mar0207.012	MET01A (iron met.)	<45	30/30	310 ± 29
	41_02	MTMIX2 (metal‐sulfide mix)	<75	38/8	340 ± 6
	35_04	Auburn FeS powder (sulfide)	>75	38/8	340 ± 7
Sanchez et al. (2017)—2015 December 08 11:43	30_07	Isheyevo powder (CH/CBb)	<75	53/8	410 ± 33
	mar0107.004	MET01A (iron met.)	<45	0/30	410 ± 8
	mar0707a.004	MET01A (iron met.)	<45	30/60	430 ± 36
	35_02	Auburn FeS powder (sulfide)	<1,000	38/8	510 ± 10
	35_03	Auburn FeS powder (sulfide)	<75	38/8	600 ± 11
Sanchez et al. (2017)—2015 December 08 12:47	mar0707a.004	MET01A (iron met.)	<45	30/60	340 ± 30
	feb1607.009	MET03A (iron met.)	<45	30/0	390 ± 35
	feb1607.006	MET02A (iron met.)	<45	30/0	400 ± 35
	30_07	Isheyevo powder (CH/CBb)	<75	53/8	450 ± 6
	mar0107.004	MET01A (iron met.)	<45	0/30	650 ± 42
Sanchez et al. (2017)—2015 December 08 14:01	feb1607.010	MET03B (iron met.)	45–90	30/0	160 ± 21
	41_01	MTMIX1 (metal‐sulfide mix)	<75	38/8	180 ± 3
	feb1607.008	MET02C (iron met.)	90–250	30/0	240 ± 25
	feb1607.004	MET01B (iron met.)	45–90	30/0	240 ± 26
	25_08	Gibeon powder (iron met.)	<75	53/8	280 ± 3
Sanchez et al. (2017)—2016 February 10 05:10	36_01	Del Norte County FeS slab (sulfide)	1,200 grit	38/8	130 ± 1
	feb1607.010	MET03B (iron met.)	45–90	30/0	340 ± 30
	38_07	Minas Gerais Pyrrhotite (sulfide) powder	>75	53/8	340 ± 2
	41_01	MTMIX1 (metal‐sulfide mix)	<75	38/8	420 ± 2
	25_04	Gibeon powder (iron met.)	<75	38/8	430 ± 2
Sanchez et al. (2017)—2016 February 10 05:30	C1AG03	AG‐BXR‐003 Pentlandite powder (sulfide)	212–600	30/0	180 ± 72
	feb1607.005	MET01C (iron met.)	90–250	30/0	420 ± 33
	32_01	Nickel sulfide, hand sample (sulfide)	hand sample	30/0	530 ± 2
	feb1607.009	MET03A (iron met.)	<45	30/0	710 ± 42
	feb1607.006	MET02A (iron met.)	<45	30/0	730 ± 43
Sanchez et al. (2017)—2016 February 10 06:34	C1AG03	AG‐BXR‐003 Pentlandite powder (sulfide)	212–600	30/0	140 ± 55
	feb1607.005	MET01C (iron met.)	90–250	30/0	150 ± 21
	feb1607.009	MET03A (iron met.)	<45	30/0	270 ± 26
	feb1607.006	MET02A (iron met.)	<45	30/0	290 ± 27
	39_02	Sikhote‐Alin powder (iron met.)	<75	38/8	700 ± 2
Sanchez et al. (2017)—2016 February 10 08:05	32_01	Nickel sulfide, hand sample (sulfide)	hand sample	30/0	510 ± 2
	C1AG03	AG‐BXR‐003 Pentlandite powder (sulfide)	212–600	30/0	740 ± 129
	feb1607.005	MET01C (iron met.)	90–250	30/0	1,060 ± 55
	feb1607.009	MET03A (iron met.)	<45	30/0	1,060 ± 52
	feb1607.006	MET02A (iron met.)	<45	30/0	1,080 ± 53

The spectrum with the lowest chi‐square value (i.e., the “best” match) is plotted over the corresponding spectrum of Psyche in Figure 3. Similar to Psyche, these materials exhibit relatively featureless, red‐sloped spectra at visible to near‐infrared wavelengths (Figure 3). As mentioned previously, it is important to note that the chi‐square test used here weights the entire spectrum equally and is therefore not particularly sensitive to subtle absorption features. Therefore, the samples listed in Table 3 can only be interpreted as the most probable surface mineralogy for Psyche within the range of compositions sampled in our spectral library. The consistency with which certain materials have low chi‐square values when compared to multiple observations of Psyche provides increased support for a comparable mineralogy between that material and the surface of the asteroid.

Figure 3.

Reflectance spectra of (16) Psyche (black) plotted with their best fit spectrum in our library (red—CH/CBb chondrite, gray—iron meteorites, yellow—sulfide). “Best fit” was determined by which spectrum had the minimum chi‐square value. Spectral data of Psyche are from Fornasier et al. (2010) (solid black), Hardersen et al. (2005) (dotted black), the Small Main‐Belt Asteroid Spectral Survey (Bus & Binzel, 2002; DeMeo et al., 2009, dashed black), and Sanchez et al. (2017) (dash‐dot black). The multiple observations from Sanchez et al. (2017) are labeled by observation time (lower 8 spectra).

Consistent with previous studies, our results demonstrate similarity between reflectance spectra of Psyche and those of iron meteorite powder and/or low‐Fe silicates (e.g., enstatite). For example, Cloutis et al. (1990) demonstrated that the spectrum of Psyche was intermediate in spectral properties to powder from the Odessa IAB iron meteorite and powder from the acid‐insoluble fraction of the Happy Canyon enstatite chondrite. Happy Canyon is an impact melt breccia that consists of ∼85 vol.% enstatite (Fs_0.4), 5–10 vol.% An₂₆ plagioclase, 5 vol.% Fs_0.9 diopside, and minor Fe, Ni metal, and troilite (McCoy et al., 1995; Olsen et al., 1977).

We note, however, that enstatite chondrites are absent from the top five best spectral matches for all spectra of Psyche (Table 3). The enstatite chondrite with the lowest chi‐square value for a particular spectrum of Psyche is LEW 88180. The value for this sample ranks 27th out of the 160 spectra in our reflectance‐filtered library when compared to the SMASS/DeMeo et al. (2009) spectrum. The spectra of enstatite chondrites in this study have near‐infrared (∼1,000–2,500 nm) spectral slopes that are lower than spectra of Psyche, which causes a high chi‐square value relative to iron meteorites, sulfides, and the powder from Isheyevo. The spectral differences between enstatite chondrites and spectra of Psyche observed here are consistent with past comparisons of spectra of Psyche with meteorites (e.g., Cloutis et al., 1990; Fornasier et al., 2010; Hardersen et al., 2005). While they are a worse spectral match for Psyche than other materials in our study, the presence of an enstatite chondrite‐like mineralogy on the surface of the asteroid cannot be completely ruled out. Both radar and emissivity data of Psyche are consistent with an enstatite chondrite‐like surface mineralogy with areas of enhanced metal abundance (e.g., de Kleer et al., 2021; Landsman et al., 2018; Matter et al., 2013; Shepard et al., 2021). Additionally, enstatite chondrites are a likely analog composition for the only other M‐type asteroid visited by spacecraft, (21) Lutetia (Vernazza et al., 2011).

Fornasier et al. (2010) found the spectrum of Psyche to be best matched by iron meteorite powder and attributed a subtle absorption feature at ∼900 nm to low‐Fe, low‐Ca orthopyroxene. A connection between Psyche and the metal‐rich CB/CH chondrites (e.g., Isheyevo) was discouraged by Hardersen et al. (2011) because the long wavelength band center measured for Psyche (>940 nm) was thought to be inconsistent with the low‐Fe silicates seen in CB/CH chondrites. However, our results, in combination with studies showing the variation in band center with Fe in pyroxenes (e.g., Cloutis & Gaffey, 1991; Klima et al., 2007), suggests that a link could still exist. Furthermore, Shepard et al. (2021) and de Kleer et al. (2021) measured radar and millimeter emissivity and polarization that is consistent with a CB chondrite‐like composition with a near‐surface porosity of 75%. Isheyevo is made up of two distinct lithologies (metal‐rich/CBb‐like and metal‐poor/CH‐like) and exhibits laminar features interpreted as evidence of gentle reaccretion of material in the impact plume of a glancing collision between planetesimals (Garvie et al., 2017; Ivanova et al., 2008; Krot et al., 2007; Morris et al., 2015). Isheyevo is ∼60 vol.% metal and its ferromagnesian chondrules are typically Fe‐poor (Ivanova et al., 2008). If Psyche is indeed a mixture of core and mantle materials, identification of a metal‐silicate mixture like Isheyevo using both Imager and GRNS data could offer strong constraints on any impact(s) that led to Psyche's current state.

We note that it is unlikely that other types of carbonaceous chondritic material dominate the surface composition of Psyche. Generally, spectra from carbonaceous chondrites are too dark, too blue‐sloped (decreasing reflectance with increasing wavelength), and/or exhibit relatively strong absorption features not found in spectra of Psyche (Cloutis, Hiroi, et al., 2011; Cloutis, Hudon, et al., 2011; Cloutis et al., 2012a, 2012b, 2012c, 2012d, 2012e, 2012f). However, the variations in Psyche's albedo, spectral reflectance, radar reflectivity, and emissivity suggest that residual material from infalling carbonaceous chondritic impactors could be present on the surface. This would be consistent with observations of carbonaceous material on the surface of (4) Vesta (e.g., Reddy et al., 2012), dynamical studies of the possibility of infall of hydrated material (e.g., Avdellidou et al., 2018), and the results of Cantillo et al. (2021) demonstrating that a spectrum of Psyche is consistent with a mixture of metal, low‐Fe pyroxene, and powder from the Murchison CM2 chondrite.

Other meteorite types that have been associated with Psyche include mesosiderites (e.g., Sanchez et al., 2017; Viikinkoski et al., 2018) and ureilites (e.g., Goodrich et al., 2018; Hardersen et al., 2005). Mesosiderites are mixtures of metal, basaltic, gabbroic, and/or pyroxenitic silicates, and sulfides like troilite. These meteorites are typically >40 wt.% silicates, the majority of which is pyroxene with >20 mol.% Fe (Mittlefehldt et al., 1998). Spectra of mesosiderites exhibit 1,000 nm absorption features with band depths that exceed those measured in spectra of Psyche as well as ∼1,900 nm features absent in spectra of Psyche (this study and Burbine, Greenwood, et al., 2007). Thus, it is unlikely that Psyche's surface is dominated by a mesosiderite‐like lithology. Ureilites, which are carbon‐bearing ultramafic meteorites dominated by olivine and pyroxene, have been invoked as a possible precursor material for M‐type asteroids like Psyche that underwent a high‐temperature (>850°C) smelting reaction (Hardersen et al., 2005; Mittlefehldt et al., 1998). The olivine and pyroxene in these meteorites typically contain >10–15 mol.% Fe (Mittlefehldt et al., 1998). The smelting reaction would have reduced this Fe‐bearing olivine (fayalite) in the precursor material to Fe‐free pyroxene, metal, carbon dioxide, and silica, proceeding at a rate dependent on the limiting reactant (either carbon or fayalite; Hardersen et al., 2005). However, the presence of some Fe‐bearing pyroxene on Psyche (inferred from observed band centers) implies incomplete reduction of the fayalite component, which would theoretically remain on the surface. Spectra of ureilites and low‐Fe fayalite have band centers >1,000 nm, and thus its presence on the surface of Psyche is not supported by current observations of the asteroid. Thus, it is unlikely that ureilite‐like material currently exists on Psyche's surface (Hardersen et al., 2005).

The interpretation that the surface of Psyche could be enriched in (and possibly dominated by) sulfide minerals is relatively new. Fornasier et al. (2010) attributed some reflectance features at ∼490 nm in the spectrum of Psyche to sulfide minerals like troilite and/or oldhamite. The densities of troilite (∼4.8 g/cm³) and pentlandite (∼4.9 g/cm³) are within the range of values reported for Psyche (Elkins‐Tanton et al., 2020; Siltala & Granvik, 2021). Furthermore, de Kleer et al. (2021) measured thermal inertia and millimeter emissivity values of Psyche that are consistent with sulfides dominating the “metallic” portion of the asteroid, with a surface porosity of 40%. Sulfide minerals are common in both chondritic and achondritic meteorites (e.g., Mittlefehldt et al., 1998; Scott & Krot, 2003) and can form both directly from the solar nebula (Lauretta et al., 1997) and through both high and low‐temperature processes expected during planetary formation (Schrader et al., 2016 and references therein). While few meteorites with high abundances (tens of vol.%) of sulfide minerals exist in terrestrial collections, models of trace element behavior in iron meteorites suggest that sulfur may have been present in high enough abundance during early planetesimal core formation to form an immiscible sulfide liquid (Bercovici, 2020; Chabot, 2004; Goldstein et al., 2009; Ulff‐Møller, 1998). If Psyche is indeed composed of remnant core materials, a significant amount of sulfide material could remain detectable on the surface (e.g., Dibb et al., 2022). However, constraining the exact nature of any sulfide components on the surface of Psyche from reflectance spectroscopy remains challenging due to spectral differences between natural and synthetic sulfides and the limited number of spectra of natural oldhamite (e.g., Burbine et al., 2002; Helbert et al., 2013; Varatharajan et al., 2019). Further reanalysis of Psyche's properties in the context of sulfide minerals could help to constrain this hypothesis.

Finally, while Binzel et al. (1995) initially observed no rotational spectral variations for Psyche at visible wavelengths <900 nm, a more recent study in the near‐infrared by Sanchez et al. (2017) reported rotational variations that were detectable in the band center of the ∼1,000 nm feature. This variation in band center was interpreted by the authors as evidence for variation in the abundance and chemistry of a purported pyroxene component. We note that the two studies of rotational variation are not mutually exclusive: Binzel et al. (1995) observed Psyche at a nearly equatorial latitude while Sanchez et al. (2017) were observing at ∼−46° and ∼−52° latitude. Similar to Sanchez et al. (2017), our chi‐square results suggest that there could be significant compositional variation on the surface of the asteroid. This is further supported by variations in radar reflectivity and millimeter emissivity (e.g., Cambioni et al., 2022; de Kleer et al., 2021; Shepard et al., 2021) that are correlated with changes in optical albedo, consistent with potential variation in the metal‐silicate ratio across the surface of the asteroid.

3.2. Comparison of Band Parameters

Surface mineralogy can also be constrained by comparing the depths and centers of absorption features at ∼1,000 nm in the spectra in our library to telescopic spectra of Psyche (Figure 4). Similar to previous studies, our band parameter measurements suggest that the surface of Psyche is consistent with a metal‐rich composition mixed with a silicate component. An example of this is the 90/10 wt.% mixture of powder from the Odessa iron meteorite and Fs_45.6 orthopyroxene from Cloutis et al. (2009), labeled MPMIX1010 in Figure 4. We note that the pyroxene in this mixture is more Fe‐rich (En_50.8 Fs_45.6) than what is typically inferred from telescopic spectra of Psyche.

Figure 4.

The range of reported ∼1,000 nm band centers and depths of spectra of (16) Psyche (pink region) are consistent with band parameters from samples containing metal and/or low‐Fe, low‐Ca pyroxene (e.g., enstatite). Samples with the “30_…” prefix are samples of the CH/CBb chondrite Isheyevo. Sample 01_01 is a hand sample of the aubrite Mayo Belwa. MPMIX1010 is a powder mixture of 90 wt.% iron meteorite metal and 10 wt.% En_50.8 Fs_45.6 orthopyroxene (Cloutis et al., 2009). Samples C1DL01A and C1DL52A are synthetic pyroxenes with En₁₀₀ and En₉₀Wo₁₀ compositions, respectively (Klima et al., 2007). Sample 081222a.018 is powder from the EH4 chondrite MET 00783. The red line represents the theoretical minimum observable band depth measurable by the Imager.

Of the three types of materials with minimized chi‐square values (iron meteorites, sulfides, and powder from Isheyevo), only spectra from Isheyevo exhibit an absorption feature at ∼1,000 nm consistent with absorption features seen in spectra of Psyche. In Figure 4, points labeled with a “30_” prefix are samples from Isheyevo. The scatter in points from this sample is consistent with previous measurements of variation in band depth and center as grain size and phase angle vary. Samples 30_03 and 30_04 are from <75 micron and >75 micron fractions of powder from Isheyevo, respectively. Both spectra were measured at a phase angle of (i/e) 38°/8°. The finer fraction (sample 30_03) exhibits a shallower absorption feature (consistent with trends for finer grained pyroxene) and a longer wavelength band center (consistent with a shift caused by blue‐sloped spectra of finer grained metal; Cloutis et al., 1990; Cloutis et al., 2010). Additionally, it is possible that sample 30_03 has a higher proportion of iron‐bearing silicates (e.g., pyroxene and olivine) than the coarser fraction due to the ease with which silicates are comminuted compared to metal. The higher relative abundance of olivine compared to metal could shift the band center to longer wavelengths (e.g., Cloutis et al., 1990). Furthermore, spectra from the powders of Isheyevo at the same grain size but different phase angle (i.e., 30_04 at i/e 38°/8° and 30_06 at i/e 23°/8°) are consistent with published trends for metal‐rich compositions. For example, Cloutis et al. (2010) measured decreasing spectral slope with decreasing phase angle for iron meteorite powders. With a lower (or even negative) spectral slope, the band center of metal‐silicate mixtures could shift to longer wavelengths relative to a spectrum of the metal‐silicate mixture measured at a larger phase angle.

Absorption band parameters from a hand sample of the aubrite Mayo Belwa (sample 01_01 in Figure 4) are also consistent with those of Psyche. Like other aubrites, this meteorite is composed mostly of almost Fe‐free enstatite (97.5 vol.% En₉₉; Watters & Prinz, 1979). Another enstatite‐rich sample, the enstatite chondrite MET 00783 (sample 081222a.018), exhibits a band center slightly longward of the upper range of reported band parameters for Psyche. MET 00783 is mostly enstatite (67 wt.%) with troilite (12 wt.%), albite (6 wt.%), kamacite (6 wt.%), and gypsum (5 wt.%) (Izawa et al., 2010). The presence of albite and gypsum (terrestrial weathering products), which exhibit broad absorption features centered at ∼1,200 and 1,400 nm, respectively, may be the cause of the longer wavelength center of MET 00783. Finally, absorption features of En₁₀₀ and En₉₀Wo₁₀ synthetic pyroxenes from Klima et al. (2007) (samples C1DL01A and C1DL52A, respectively) are also consistent with reported band depths and band centers for Psyche.

Also shown in Figure 4 is a theoretical minimum observable band depth that the Imager can measure (∼0.3%). This limit is based on radiometric modeling of the Imager's performance during nominal Orbit B operations using equations from Tables 9–15 in Larson and Wertz (1999). We use this radiometric model to estimate the signal‐to‐noise ratio (SNR) for pixels in images taken through each filter. During Orbit B operations, we estimate the SNR in the two longest wavelength filters (centered at 948 and 1,041 nm) to be ∼118 and ∼98, respectively, assuming pixel summing (2 × 2 and 3 × 3, respectively) and the longest possible exposure times before the spacecraft's velocity causes smearing of a half‐pixel (167.4 and 251.1 ms, respectively). These SNR estimates translate to relative errors of 0.85% and 1.02%. Assuming Psyche's albedo at 1,000 nm is ∼0.14, these relative errors translate to absolute reflectance uncertainties of ∼0.12% and 0.14%. The sum of these two absolute reflectance uncertainties represents the theoretical minimum observable band depth (∼0.3%). This limit suggests that absorption features from very low‐Fe natural pyroxenes, like those in spectra of Mayo Belwa, could be undetectable by the Imager. However, very low‐Fe enstatite‐rich compositions (e.g., aubrites) are typically much brighter than the measured visible wavelength reflectance of Psyche, and thus current observations suggest that these compositions are not likely to occur on the asteroid's surface.

4. Preliminary Strategies for Using Imager Data to Assess Surface Composition

Three of the five major science objectives of the Psyche mission that require data from the Imager filters are: (a) to determine if Psyche is a remnant planetary core; (b) to determine if small bodies incorporate the same light elements as are expected in the Earth's core; and (c) to determine the redox conditions under which Psyche was formed (Elkins‐Tanton et al., 2016, 2020). To address these objectives, the Imager must assess the homogeneity of metal‐silicate mixing, identify any diagnostic sulfur‐bearing phases present, and discriminate between silicate or sulfide compositions that could help to constrain redox conditions (Bell et al., 2016; Elkins‐Tanton et al., 2020, 2022). Using the laboratory spectra in our library, we have developed a set of parameters using spectra convolved to the Imager's spectral throughput that could allow rapid processing of future Imager multispectral data to derive compositional information about the surface of Psyche at the resolution of single images, and thus to help address the mission's relevant science objectives.

While the generally featureless, red‐sloped spectra from most disk‐integrated telescopic measurements of the asteroid (Table 2) are typically interpreted as evidence of a surface dominated by metal but potentially with some low‐Ca, low‐Fe silicates, they do not rule out the possibility of small (unresolved) areas on the surface enriched in silicate minerals with higher Ca or Fe contents (e.g., MPMIX1010 in Figure 4). This possibility is consistent with the rotational spectral variation reported by Sanchez et al. (2017), who measured areas of differing silicate abundance and chemistry heterogeneously distributed on the surface of Psyche. Thus, it is prudent to explore how the Imager could discriminate various compositions with higher abundances of silicates and/or variable silicate chemistries.

For example, the nature of iron‐bearing silicates relevant to Psyche (e.g., pyroxene and olivine) can potentially be constrained via the ratio of the 948/1,041 nm filter reflectances plotted versus the 550 nm filter reflectance (Figure 5). Low‐Ca, Fe‐bearing pyroxenes (yellow squares in Figure 5) typically exhibit an absorption feature between ∼910 and ∼960 nm due to Fe²⁺ in the M2 crystallographic site (Adams, 1974; Cloutis & Gaffey, 1991; Klima et al., 2007). This is expressed in Imager data as a ratio in the convolved data from the 948/1,041 nm filters less than 1. Conversely, spectra of high‐Ca pyroxene (purple squares in Figure 5) and olivine (green circles in Figure 5) exhibit absorption features at wavelengths longer than ∼1,000 nm, and thus the 948/1,041 nm ratio is typically greater than 1. High‐Mg silicates (e.g., some of the olivines and pyroxenes shown in Figure 5 as well as the aubrites) have much shallower or no absorption features, and thus exhibit 948/1,041 nm filter ratios close to 1. Most pure silicates in our library exhibit higher reflectance than the range of values reported for Psyche, represented in Figure 5 by a gray box. The more metal‐rich compositions (e.g., iron meteorites, enstatite and CB chondrites) and sulfides also plot within the gray box in Figure 5. Finally, mixtures of iron meteorite metal and Fo_90.4 olivine (green diamonds in Figure 5) and iron meteorite metal and En_50.8 Fs_45.6 orthopyroxene (blue triangles in Figure 5) from Cloutis et al. (2009) are also shown in Figure 5. Both of these sets of mixtures exhibit lower 550 nm filter reflectance and become featureless (i.e., migrate toward a 948/1,041 nm filter ratio of 1) with increasing metal content.

Figure 5.

Low‐Ca pyroxene (Wo < 13, yellow squares) can be discriminated from other silicates that could be present on the surface of Psyche, like high‐Ca pyroxene (purple squares) or olivine (green circles) using the ratio between Imager Filters 7/8 (948/1,041 nm). Furthermore, these iron‐bearing silicates can be discriminated from lower reflectance, featureless materials using both the Filter 7/8 ratio and the Filter 4 (550 nm) reflectance. The range of values for Psyche, iron meteorites, sulfides, and metal‐rich chondrites are represented by a gray rectangle. Mixtures of pyroxene and metal (MET + OPX, dark blue triangle and solid line) and mixtures of olivine and metal (MET + OLV, dark green diamonds and solid line) at 10 wt.% intervals indicate that mixtures with increasing metal content exhibit lower 550 nm reflectance and become “featureless” (948/1,041 nm ratio of ∼1.0). Uncertainty is typically smaller than the size of the points.

Second, some relationships among absorption feature parameters that have been well‐established from previous high spectral resolution laboratory studies can also be recovered from spectra in our library convolved to the Imager's spectral throughput. For example, Figure 6 shows the band centers reconstructed from Imager‐convolved spectra of natural and synthetic pyroxenes plotted against the band center derived from the high‐resolution laboratory spectra. The pyroxenes in Figure 6 are those that have chemistries within the range measured by Sanchez et al. (2017). The residual mean squared error (RMSE) from this comparison is 6.6 nm. Using the equations developed by Burbine, Buchanan, and Binzel (2007) for relating the ∼1,000 nm band center in spectra of HED meteorites to their chemistry, our RMSE translates to an uncertainty in the inferred pyroxene chemistry of ±8.5 mol.% En, ±6.8 mol.% Fs, and ±2.7 mol.% Wo. Accurate measurement of absorption band centers may be used to constrain the Fe content in silicates as well as the pyroxene/olivine ratio on the surface, and thus provide insight into the redox state of the asteroid (e.g., Hardersen et al., 2005; McSween, 1992).

Figure 6.

Band centers from synthetic (circles) and natural pyroxenes (triangles) can be measured in spectra convolved to the Imager's bandpasses (RMS error = 6.6 nm). The samples shown here have compositions covering the range of inferred pyroxene chemistries on the surface of Psyche by Sanchez et al. (2017). This includes all pyroxenes in our library with Fs ≤ 51.5 mol.% and Wo ≤ 12.9 mol.%. Points are colored according to their chemistry using an R/G/B = Fe/Mg/Ca scheme. Because the pyroxene on Psyche reported in Sanchez et al. (2017) is mostly Mg‐rich, most points are green/brown. The blue line of slope equal to 1 indicates perfect recreation of the band center in the Imager‐convolved spectra.

Furthermore, the abundance of pyroxene in metal‐pyroxene mixtures can be accurately measured from Imager‐convolved spectra by comparing band depths of spectra convolved to the Imager's throughput to band depths from high‐resolution laboratory spectra (Figure 7a) (RMSE = 2.14%). Band depths of spectra from metal‐olivine mixtures (Figure 7b) can be measured from Imager‐convolved data using Method 3 in Figure 2. For mixtures with high abundance of metal, band depths become increasingly shallow and the overall spectra become more red‐sloped. Band depths from mixtures with >40 wt.% olivine measured from Imager‐convolved spectra can be accurately recreated, and a cubic polynomial can be used to infer band depth for mixtures with <40 wt.% olivine (Figure 7b). The RMS error for this cubic polynomial fit to the metal‐olivine band depths is 3.71%.

Figure 7.

Band depths from high‐resolution laboratory spectra and Imager‐convolved spectra of metal‐orthopyroxene mixtures (a) and metal‐olivine mixtures (b) from Cloutis et al. (2009). The ratio of the components (in wt.% intervals of metal/silicates) is shown next to each point.

In Figure 8 we offer a preliminary strategy for discriminating the low reflectance, “featureless” compositions that could be present on the surface of the asteroid as suggested by remote sensing and our chi‐square analysis. These compositions include iron meteorites, sulfides, and CB and enstatite chondrites. Figure 8 shows the Filter 4/8 (550/1,041 nm) ratio versus the Filter 3/4 (495/550 nm) ratio for spectra from these compositions in our library that had ∼1,000 nm band depths of <3.0% measured from Imager‐convolved spectra. This limit is approximately the highest reported band depth in spectra of Psyche (Table 2). The Filter 3/4 (495/550 nm) ratio effectively separates the iron meteorites and sulfides due to the typically higher visible wavelength slopes of sulfides. Britt and Pieters (1988) and Cloutis et al. (2010) demonstrated that spectra of iron meteorites at various grain sizes and compositions are relatively invariant at these wavelengths, and this is reflected by the clustering of iron meteorites in Figure 8.

Figure 8.

Compositions with “featureless” spectra that are consistent with telescopic spectra of (16) Psyche can be discriminated based on ratios between Filters 4/8 (550/1,041 nm) and Filters 3/4 (495/550). Spectra of powder from terrestrial sulfides from Del Norte County, California and an outlier enstatite chondrite, MAC 02747, are labeled. The Del Norte sulfide is dominated by troilite and pyrrhotite with minor cubanite (CuFe₂S₃). MAC 02747 has the lowest metal content of the enstatite chondrites in this study. See text for discussion.

Conversely, sulfides exhibit a larger range of Filter 4/8 (550/1,041 nm) and Filter 3/4 (495/550 nm) ratios (Figure 8), likely due to the textural and mineralogical variation in the sulfide samples in our library (Table S1). Unlike the other sulfides, a cluster of points from spectra of a sulfide sample from Del Norte County, California exhibits Filter 4/8 (550/1,041 nm) ratios >0.8. This sample is dominated by troilite and pyrrhotite, but also contains a significant amount (<∼20 vol.%) of cubanite (CuFe₂S₃). This mineral exhibits a shallower near‐infrared spectral slope consistent with other Cu‐bearing Fe‐sulfides, which is likely the cause of Filter 4/8 (550/1,041 nm) ratios closer to 1.0 (Pirard et al., 2008; Wood & Strens, 1979). While cubanite has been found in CI chondrites (Berger et al., 2011), material returned from comet 81P/Wild 2 (Brownlee, 2014), and in exogenous regolith grains in Itokawa (Burgess & Stroud, 2021), cubanite is only found in metal‐rich meteorites as a terrestrial alteration product (Rubin, 1997; Rubin & Ma, 2017). The presence of cubanite in extraterrestrial materials constrains the maximum temperature experienced to <∼250°C (Berger et al., 2015), and thus identification of cubanite on the surface of Psyche could significantly constrain the asteroid's formation history.

Finally, “featureless” spectra of the CB and enstatite chondrites in our library exhibit Filter 3/4 (495/550 nm) and Filter 4/8 (550/1,041 nm) ratios distinct from the iron meteorites. We include enstatite chondrites in this plot because their spectra at Imager‐relevant wavelengths (∼400–1,100 nm) are comparable to spectra of Psyche (despite their near‐infrared spectra deviating from those of Psyche) and because they have been previously invoked as possible Psyche analogs. Both the enstatite chondrites and CB chondrites are characterized by a high abundance of low‐Fe, low‐Ca pyroxene and/or olivine in chondrules (20–80 vol.%) mixed with metal, and sulfides. Some of the enstatite chondrites also contain several wt.% feldspar and terrestrial alteration products (e.g., gypsum). The enstatite chondrite MAC 02747 exhibits a Filter 3/4 (495/550 nm) ratio much lower than the other enstatite chondrites in our study. This sample has the lowest metal content of the enstatite chondrites in Figure 8 (Izawa et al., 2010), and thus may have an enhanced metal‐oxygen charge transfer absorption relative to the other enstatite chondrites. This enhancement may result in a stronger visible slope and thus lower Filter 3/4 (495/550 nm) ratio.

5. Conclusions

The Psyche mission will explore the solar system's largest M‐type asteroid, (16) Psyche. We have assembled a library of laboratory spectra that are consistent with current uncertainties in Psyche's surface composition based on spectral properties, density, radar reflectivity, and emissivity. We have performed a chi‐square analysis that has shown that available telescopic spectra of Psyche, while consistent with iron meteorites, are also consistent with sulfide minerals (e.g., troilite and pentlandite) and powder from the CH/CBb chondrite Isheyevo. Second, we have compared band centers of ∼1,000 nm absorption features found in spectra of Psyche to similar features in the spectra in our library. Results from this investigation are consistent with previous interpretations of the asteroid's surface being metal‐rich with a low‐Fe, low‐Ca pyroxene component, but are also consistent with a metal‐rich composition mixed with a small abundance of Fe‐rich pyroxene (e.g., 90 wt.% iron meteorite metal and 10 wt.% En_50.8 Fs_45.6 orthopyroxene). Our investigation provides new insight into the surface composition of the asteroid and is consistent with recent studies that suggest the surface composition of Psyche may not be as metallic as once thought.

To address the Psyche mission's science objectives and to resolve these various compositional possibilities, the Psyche spacecraft's Multispectral Imager will map the surface from orbit at visible to near‐infrared wavelengths and at high spatial resolution. We have demonstrated that spectra of silicates and metal‐silicate mixtures convolved to the Psyche Imager's filters can be discriminated based on their Filter 4 (550 nm) reflectance and Filter 7/8 (948/1,041 nm) ratio. Band centers derived from Imager‐convolved data are predicted to be accurate to within <10 nm, while band depths derived from spectra of metal‐silicate mixtures are predicted to be accurate to within about 5%. Finally, the low reflectance, “featureless” materials in our spectral library that are consistent with telescopic spectra of Psyche can be discriminated using the Imager Filter 3/4 (495/550 nm) and Filter 4/8 (550/1,041 nm) ratios. Multispectral images acquired by the Imager, in combination with data from the spacecraft's GRNS and magnetometer investigations, will allow hypotheses of the asteroid's formation history to be tested and help to constrain critical processes in terrestrial planet formation.

Supporting information

Table S1

Table S2

Dibb, S. D. , Bell, J. F., III , Elkins‐Tanton, L. T. , & Williams, D. A. (2023). Visible to near‐infrared reflectance spectroscopy of asteroid (16) Psyche: Implications for the Psyche mission's science investigations. Earth and Space Science, 10, e2022EA002694. 10.1029/2022EA002694

Data Availability Statement

All spectral data used in this study are available in the Open Science Foundation repository https://doi.org/10.17605/OSF.IO/Z6PHN. Spectral data from the NASA RELAB Facility at Brown University are available at https://sites.brown.edu/relab/relab-spectral-database/ and spectral data from the University of Winnipeg Planetary Spectrophotometer Facility are available at http://psf.uwinnipeg.ca/Sample_Database/. The reader is referred to those facilities for the most recent versions of the data and associated archiving information.