Space Weathering Effects in Troilite by Simulated Solar-Wind Hydrogen and Helium Ion Irradiation.

Space weathering is a key process in the interpretation of airless planetary surfaces. As we engage new missions to planetary objects with potentially novel surfaces such as 16 Psyche, there is renewed interest in expanding our knowledge of space weathering effects to a wider variety of analog materials, including the physical/chemical effects of solar-wind ions on planetary regoliths. We have experimentally simulated the effects of solar ions on two polished thick sections of meteoritic troilite (FeS) via irradiation with 1 keV hydrogen (H+) and 4 keV helium (He+), to investigate effects resulting from different ion species. We detected depletion of sulfur over the course of each irradiation using in situ X-ray photoelectron spectroscopy. Sulfur depletion rates were surprisingly similar for H+ and He+, interpreted as a function of subsurface ion-activated diffusion. By comparing XPS-derived elemental abundances with SDTrimSP computer simulations, we further quantified sulfur diffusion, sputtering yield, and altered-layer composition with respect to incident-ion fluence, and accounted for the influence of surface oxidation due to atmospheric sample storage. Using scanning electron microscopy, we detected an increase in nanoscale surface roughness resulting from the irradiation, which we quantified using atomic force microscopy. Based on these results, we estimate that an exposure time of order 103 Earth-years is required for troilite on Psyche to reach equilibrium sulfur depletion within the first atomic layer.

Introduction

Airless planetary surfaces are constantly bombarded by energetic particles, photons, and meteorites from the surrounding space environment, altering their optical, physical, or chemical properties; these interactions are known as “space weathering.” Space weathering by solar‐wind ions—the continuous stream of hydrogen, helium, and trace amounts of heavier elements radiating out from the sun with a peak energy of ∼1 keV/amu—was identified as a mechanism to explain spectral variation of Lunar crater rays with age and distance, and optical differences between returned Apollo soils and crushed regolith rocks (e.g., Farrell et al., 2015; Gold, 1955; Hapke, 2001; Pieters & Noble, 2016). More recently, the spectral differences between asteroids and meteorite samples (Chapman, 1996; Dukes et al., 1999) were also attributed to solar‐wind irradiation. Radiation‐exposed surfaces of mafic minerals decrease in albedo and become redder in overall reflectance over time, with commensurate attenuation of absorption features (Hapke, 2001; Loeffler et al., 2009). On lunar grains and in Fe‐bearing minerals such as olivine, these spectral effects have been associated with the formation of nanophase iron (npFe) particles within the amorphous rims of weathered surfaces (Gaffey et al., 1993; Keller & McKay, 1997; Matsumoto et al., 2021; Pieters et al., 2000). Although a number of additional mechanisms have been proposed to play a role in formation of space weathering textures, including micrometeoroid impacts (Loeffler et al., 2016; Sasaki et al., 2001) and other sources of heating (Thompson et al., 2017), irradiation of surfaces by solar‐wind ions is considered the principal weathering mechanism on asteroids (Schläppi et al., 2008), directly linked to the formation of npFe and iron‐cladding on exposed surfaces (Matsumoto et al., 2020; Noguchi et al., 2014). Moreover, research on metals and oxides has identified other damage modes associated with light‐ion irradiation at energies similar to solar‐wind ions, such as phase‐change, pitting, and blistering (Auciello, 1984; Kajita et al., 2016; Meyer, 2018; Tanyeli et al., 2015).

Much space weathering work historically focused on silicate minerals, the major mineral components of ordinary chondrites—the most abundant meteoritic type—which are linked to S‐type asteroids (e.g., Clark et al., 1992; Gaffey et al., 1993; Nakamura et al., 2011; Pieters et al., 2000). Other minerals have received less attention. Only a small number of studies have investigated space‐weathering effects on iron meteorites or their major mineral components (Bezaeva et al., 2015; Harries & Langenhorst, 2014; Loeffler et al., 2008; Matsumoto et al., 2020, 2021; Prince et al., 2020; Wu et al., 2017). While there have been extensive engineering studies on the effect of the space environment on structural metals for spaceflight applications, these synthetic materials are poor analogs for iron meteorites due to their orders‐of‐magnitude smaller crystal sizes and differing microstructures, both resulting from the inability to synthetically crystallize materials over geologic timescales (Petrovic, 2011). Given that we have not yet visited a metallic asteroid with spacecraft (Elkins‐Tanton et al., 2020) and the difficulties in identifying asteroids that have metallic surfaces (Elkins‐Tanton et al., 2020; Landsman et al., 2015; Rivkin et al., 2000; Sanchez et al., 2017; Takir et al., 2017), any understanding of space weathering effects on iron meteorites as an analog for naturally occurring metallic planetary surfaces must begin from laboratory experiments.

Troilite is the predominant sulfide mineral present in chondritic meteorites, one of the most common inclusions in iron meteorites (Grady et al., 2014), and possibly present on the Psyche surface or bulk (Elkins‐Tanton et al., 2020). Previous studies invoked space weathering by solar wind to explain sulfur depletion observed at the near‐Earth asteroid 433 Eros (Killen, 2003; Kracher & Sears, 2005; Nittler et al., 2001). Experimental work to test this hypothesis has used helium and/or argon ion irradiation to simulate solar wind and laser irradiation to simulate meteoritic impact space weathering. These experiments identified a two‐step mechanism—sputtering initiated by incident solar‐wind protons, and subsequent partial melting and vaporization by micrometeoroid impacts—depleting sulfur by a factor of 1.5–2.5 over a period of 104–105 years (Loeffler et al., 2008; Thompson et al., 2019). Recent laboratory studies point to the evolution of a surficial layer of metallic iron and subsurface iron‐rich sulfide with ion irradiation (Keller et al., 2013). Additional studies have characterized space weathering effects on composition and microstructure of natural troilite grains returned from 25143 Itokawa (Harries & Langenhorst, 2014; Matsumoto et al., 2020) and Apollo lunar samples (Matsumoto et al., 2021), identifying micron‐ and sub‐micron‐scale iron whiskers forming on the porous, amorphized, and sulfur‐depleted surfaces of troilite grains.

We report the results of laboratory irradiation experiments using troilite thick‐section samples from iron‐meteorite inclusions, using protons and helium ions to better simulate the solar wind and identify species‐specific differences. Using X‐ray photoelectron spectroscopy (XPS) measurements and SDTrimSP computer simulations, we identify and characterize changes to surface composition and chemistry as a function of ion fluence and compare differences in sputtering by 1 keV/amu H+ and He+. We also perform scanning electron microscopy (SEM) and atomic force microscopy (AFM) measurements to determine roughness originating during irradiation, which may play a role in altering the albedo and reflectance spectra at short wavelengths for surfaces exposed to space, beyond the formation of npFe and loss of sulfur. Results from these experiments will be useful for the interpretation of observations from the NASA Psyche mission (Elkins‐Tanton et al., 2020).

Methods

Samples

The Center for Meteorite Studies (CMS) at Arizona State University provided two polished thick sections of troilite‐nodule inclusions taken from the Canyon Diablo and Toluca iron meteorites. We used polished thick sections instead of pressed powder slabs to optimize measurement of potential irradiation‐induced surface roughness. While loose or pressed powder is generally considered an analog for fine‐grained regoliths on airless planetary objects (e.g., Loeffler et al., 2008), Psyche may present us with a potentially novel type of planetary surface (Elkins‐Tanton et al., 2020) for which a powder sample may not be appropriate. Both Canyon Diablo and Toluca sections have been characterized with wavelength‐dispersive X‐ray spectroscopy calibrated against standardized pyrite and olivine to confirm their bulk composition is 50 at% S, 50 at% Fe, and <1% for others.

Irradiation and In Situ Characterization

Irradiation and in situ characterization of the samples were performed at the University of Virginia. Samples were ultrasonically cleaned in an isopropanol bath before irradiation to remove any organic contaminants from sample handling. We irradiated the Canyon Diablo sample with only He+ using an electron‐bombardment‐type ion gun on a PHI Versaprobe III XPS. We then irradiated the Toluca sample first with 4 keV helium (He+) ions, followed by 1 keV hydrogen (H+) ions using a similar ion gun on a highly customized PHI‐560 XPS system. The use of an initial He+ irradiation on both samples allowed the removal of surficial oxide and ensured that we could compare the results of H+ and He+ irradiation on similarly prepared sulfide surfaces. Details of the experimental setup can be found in Text S1 of Supporting Information S1 and have been described in earlier studies (e.g., Dukes et al., 1999; Laczniak et al., 2021).

Both irradiations were designed to reach a comparable target fluence (∼3 × 1018 ions/cm2), enabling direct comparison between proton and helium irradiation. The target fluence falls in the saturation range of S‐depletion observed previously for 4 keV He+ irradiation of FeS pressed pellets (Loeffler et al., 2008). The average solar‐wind flux at 1AU from the Sun is 2 × 108 ions/cm2/s, consisting of ∼96% H+, ∼4% He+, and heavier highly charged ions less than 0.1% (Johnson & Baragiola, 1991). Since solar‐wind flux decreases proportionally to the square of the distance from the Sun, the target fluence of 3 × 1018 ions/cm2 simulates exposure times of ∼100,000 years for He+ and ∼4000 years for H+ at 16 Psyche (semimajor axis ∼2.9 AU). To reach the target fluence in a practical laboratory time frame, we used an average flux of ∼1 × 1013 ions cm−2 s−1. In both systems, low energy electrons (<5 eV) from an electron flood gun prevent surface charging of the sample during irradiations and XPS analysis. The samples were at room temperature (300 K) during irradiation and characterization.

X‐Ray Photoelectron Spectroscopy

XPS is a quantitative surface analytical technique, providing concentration and chemical state information for elements within the top monolayers of a material (less than 10 nm). On both PHI560 and Versaprobe III, Al‐K X‐rays (E X‐ray = 1486.6 eV) interact with the samples, producing photoelectrons ejected into the vacuum with a characteristic kinetic energy (E k) given by E  = EX‐ray – E B – Φ, where Φ is the measured work function of the spectrometer, and E B is the binding energy of the core‐level (or valence) electron with respect to the Fermi level. Photoelectrons from deeper in the material lose energy as they travel through the solid and contribute to the spectral background. The measured energy distribution of the ejected electrons is displayed as counts/s versus E B. The area of an individual photoelectron peak after background subtraction, together with the corresponding normalization (“sensitivity”) factor, is used to obtain elemental abundance, while the peak position (E B) and shape give information on the chemical state of the element.

In this work, the S‐2p, O‐1s, C‐1s, and Fe‐3p photoelectron peaks were selected for troilite compositional determination; spectral background due to inelastically scattered electrons was subtracted using the iterative‐Shirley method (Shirley, 1972). We chose the Fe‐3p feature with its low background for use in quantification, as the Fe‐3p and S‐2p features provide information from a similar sampling depth—critical for direct comparison of S:Fe abundance. However, the Fe‐2p region, which is complicated by dual spin‐orbital features, multiplet splitting and energy‐loss satellites, exhibits greater sensitivity to chemical change and was monitored in the high‐resolution XPS data sets. For more details, see Figures S1 and S2 in Supporting Information S1, which show survey and high‐resolution XPS spectra, respectively, including Fe‐3p and Fe‐2p photoelectron peaks.

We confirmed our derived elemental concentrations utilizing manufacturer‐provided instrument‐specific sensitivity factors (Multipak v9.8, Wagner et al., 1981), by comparison with results utilizing the Fe‐2p3/2 and full Fe‐2p feature with both Wagner et al. (1981) and Loeffler et al. (2008) sensitivity factors; compositions were in agreement within 26% for each element/method. Versaprobe instrument sensitivity factors were confirmed by measurement of freshly cleaved pyrite, analyzed in the same manner (Fe‐3p and S‐2p) immediately upon fracture and found to be appropriately stoichiometric (S:Fe = 2.05 ± 0.05); Loeffler et al. (2008) did not measure a sensitivity factor for the Fe‐3p.

Sample Imaging and Surface Roughness Characterization

We imaged the polished samples using the FEI XL30 Field‐Emission Environmental SEM at ASU's Eyring Materials Center. We obtained images at >20,000x magnification of the sample surfaces before irradiation as well as images at the center of the irradiation spot and outside the irradiation spot after irradiation to identify any radiation‐induced changes to the surface roughness, topography, or texture. Because troilite is weakly conductive, we were able to obtain high‐resolution images operating the XL30 at low electron‐beam voltage (5 kV) without carbon‐coating the sample.

After confirming that the irradiated surface had roughened in the SEM measurements, we quantified the change in surface roughness using the Bruker Dimension 3000 AFM at ASU's Eyring Materials Center with a NCHV‐A probe in tapping mode. AFM measures the height (Z) of successive points across the sample surface by measuring the slight deflection of the nanoscale probe as it interacts with the surface. The probe rasters across a square subdivided into 512 × 512 probe measurement points. By decreasing the size of the square, the distance between probe measurement points decreases, enabling roughness characterization across a range of length scales. We selected squares with sizes 2 × 2, 5 × 5, and 20 × 20 μm, at locations within the irradiated area and without any visible large scratches or other features not obviously caused by irradiation. Once the raster image is obtained, MM‐8 software included in the Dimension 3000 calculates the Z range, average, and standard deviation, as well as the projected surface area of the rastered region. To minimize the statistical influence of pre‐irradiation roughness, we then selected smaller subregions within the 512 × 512 point raster images, which displayed minimal height variation from sample polishing scratches and recalculated the same statistics.

SDTrimSP Modeling

SDTrimSP 6.0 is a 1‐D Monte Carlo simulation that uses the binary‐collision approximation to model atomic‐collision processes for projectiles incident on amorphous targets, both gaseous and condensed (Mutzke et al., 2019). This method improves upon some of the known limitations with the SRIM simulations, which do not model accurately the effects of non‐normal projectile incidence, sputtering yield angular distributions, multicomponent sputtering yields, and interpolated stopping powers; comparisons of SDTrimSP output to experimental data are detailed in Hofsäss et al. (2014) and others (e.g., Arredondo et al., 2019; Wittmaack & Mutzke, 2017) along with a discussion of limitations and error. The program can output ion ranges, sputtering yields, reflection coefficients, damage cascades, target compositions as functions of depth or fluence, and specific target atom/molecule information such as trajectories, energies, exit angles, vacancies, and displacements for comparison with measured parameters (e.g., Stadlmayr et al., 2018; Szabo, Biber, Jäggi, Brenner, et al., 2020; Wittmaack & Mutzke, 2017). Unlike its predecessor, SRIM (Ziegler et al., 1985), SDTrimSP provides (a) the option for a dynamic mode, which periodically updates the stoichiometry of the target throughout the simulation, (b) incorporates outgassing and diffusion of both projectiles and target atoms, (c) includes chemical sputtering for carbon, (d) allows additional changes in atomic interaction potential, (e) facilitates multiple “concurrent” projectile types and energy/velocity distributions, and (f) provides a new inelastic energy model for projectiles exceeding 25 keV. The user can modify many parameters including the method to solve the scattering angle and time integrals, the surface binding‐energy model, and the atomic/molecular interaction potential model. Our simulations typically used the recommended Gauβ‐Legendre integration method; atom‐specific surface binding‐energy “model two;” and the KrC interaction potential. The resulting target‐compositional output of the SDTrimSP simulations of H+ and He+ irradiation on troilite can be directly compared to experimental data—such as XPS spectra—and used to predict the effects of solar‐wind ion impact on airless planetary surfaces, such as Psyche.

We ran SDTrimSP with variations of 4 keV He+ and 1 keV H+ normally incident on a planar target, using the laboratory flux of ∼1 × 1013 ions/cm2/s to a fluence of ∼3 × 1018 ions/cm2 in order to identify optimal parameters for surface binding energy and diffusion. The upper‐surface composition of the Canyon Diablo and Toluca samples were modeled with 150 and 46 Å oxide layers, respectively (see Section 3.4). Both oxides are atop a target composed of FeS. The outermost surfaces of the modeled targets were ∼3 Å of adventitious carbon, typical of most atmosphere‐exposed materials (Barr & Seal, 1995; Sinha & Mukherjee, 2018). The sample temperature was defined to be 300 K, consistent with laboratory measurements. Projectile ions and sulfur target atoms were allowed to diffuse via radiation‐enhanced diffusion and outgassing, where levels were controlled via their respective diffusion coefficient inputs. Undefined inputs, such as diffusion coefficients, surface binding energies, and sample oxide characteristics, were manipulated in these simulations to produce data that most closely resembles that from our experiments.

Conversion of SDTrimSP Output to XPS Information Depth

SDTrimSP provides compositional atomic fractions on a monolayer scale, while XPS provides elemental atomic fractions of the integrated upper surface (∼50 Å) as a function of fluence. Thus, to directly and quantitatively compare the simulation results to those from the lab, SDTrimSP outputs require conversion to match the XPS information depth.

Photoelectrons from deeper in the target are less likely to be detected in XPS, due to their increased interactions within the solid. Thus, the likelihood of detecting a photoelectron (and thus, its parent atom) is dependent on depth and the electron inelastic mean free path (IMFP) within the target (e.g., Jablonski & Powell, 2020). The IMFP was calculated for each (element) photoelectron using the NIST SRD71 TPP‐2M model (Powell & Jablonski, 2010) with a density of 4.74 g/cm3 for FeS (Skala et al., 2006). The signal intensity (I ) from each atomic layer can be described as where I is the photoelectron intensity from the surface layer, d is the depth, θ is the photoelectron detection angle (45°) with respect to the surface normal, and λ is the IMFP. To convert modeled data to XPS composition, an individual SDTrimSP simulation must be run where the final fluence is the respective XPS fluence. For each of those simulations, the final atomic fraction at each depth was weighted per Equation 1 for the individual element through the first 7.5 nm, then summed and normalized. Because XPS cannot detect helium or protons, the atomic fractions of the implanted helium or protons were not included in this normalization.

Results

Ion‐Induced Surface Roughness

SEM images of both Canyon Diablo and Toluca sample postirradiation clearly show the development of nanoscale ion‐induced surface texturing. Figure 1 shows high‐magnification images (20,000x) comparing outside and inside the irradiated region on both Canyon Diablo and Toluca. For both samples, the surface outside the irradiated area is smooth apart from striated scratches made during sample polishing. These polishing scratches are observed also within the irradiated area, overlain by additional pitting and other roughness features at a length scale smaller than 10 nm, a result of ion bombardment.

Figure 1

High‐magnification (20,000x) scanning electron microscopy images of the Toluca and Canyon Diablo polished surfaces outside the irradiated area and within the irradiated surfaces showing nanoscale texturing by both 1 keV H and 4 keV He.

The AFM surface roughness measurements were similar across both samples (Figure 2). On the unirradiated surfaces, the average of the Z standard deviations across eight AFM measurement sites was 2.36 ± 0.25 nm for Canyon Diablo and 2.26 ± 0.25 nm for Toluca. On the irradiated surfaces, the average of the Z standard deviations was 5.94 ± 0.36 nm for Canyon Diablo and 5.49 nm ± 0.41 for Toluca.

Figure 2

Atomic force microscopy images showing the unirradiated (left) and irradiated (right) surfaces of Canyon Diablo; both images are 2 micron squares.

Surface Compositional Change

Figure 3 shows the elemental abundances of Fe, S, O, and C with fluence as measured by XPS on the Canyon Diablo and Toluca samples. At low fluences on both samples, we observed a S:Fe ratio lower than the expected 1:1 stoichiometric ratio for troilite due to the presence of a surface oxide. Oxide formation occurs rapidly on iron sulfide exposed to O2 in the atmosphere as well as with the polishing required for SEM and AFM measurements. As this oxide layer was removed by sputtering during the early stage of the irradiation, the S:Fe ratio reached a relative maximum, after which it decreased as sulfur was preferentially removed.

Figure 3

Atomic abundances of the major elements on the (a) Canyon Diablo and (b) Toluca surface at varying fluences of H+ and He+ ions. The gray bar indicates the fluence where the interface between the surface oxide coating and bulk troilite is reached.

On the Canyon Diablo sample, after the removal of the surface oxide layer (described below), we measured a maximum S:Fe ratio of 0.60 at a fluence of 4.8 × 1017 He+/cm2, after which the S:Fe ratio decreased with continued irradiation (Figure 3a) to a final S:Fe ratio of 0.36 at the end of irradiation. The oxide‐troilite interface is identified as the region where the surface oxide is reduced to 50%, after irradiation by 3.2 × 1017 He+/cm2.

For the Toluca sample, irradiated with primarily protons, the S:Fe ratio is largest at 0.51 with 2.6 × 1017 ions/cm2 (Figure 3b). Again, after removal of the surface oxide (described below), we observed sulfur loss for fluences above 2.6 × 1017 ions/cm2. The final S:Fe ratio for the Toluca sample after a fluence of 6 × 1016 He+/cm2 and then at 2.8 × 1017 H+/cm2 was 0.22. The oxide‐troilite interface at 50% of the maximum oxygen concentration is found at a total fluence of 2.4 × 1017 ions/cm2. Carbon appears within and above the oxide layer for Toluca, and with proton irradiation both carbon and oxygen appear well beyond the fluence identified as the interface, suggesting a broader interface region potentially due to a reduced sputtering yield by protons in the oxide layer. Indeed, we found that the O was retained in the Toluca FeS at higher concentrations after the interface than for Canyon Diablo, even after ∼3 × 1018 ions/cm2. This is consistent with sputtering yield measurements and models for metal oxides, which find that protons sputter 10−1 less efficiently compared with helium (Betz & Wehner, 1983; Schaible et al., 2017).

Fe and S Surface Chemistry

Photoelectron binding energies are sensitive to the difference between the surface‐atom final‐state and ground‐state configurations and reflect the local chemistry. We examined the chemical state of the Canyon Diablo and Toluca surface layers, using high‐resolution (low pass‐energy) XPS spectra. Figure 4 shows the Fe2p and S2p regions measured at 1.5 × 1017 He+/cm2, just before the oxide‐surface interface for Canyon Diablo, and again at the end of the 4 keV He irradiation (3.6 × 1018 He+/cm2). Spectra for the O1s and C1s peaks were also collected (see Figure S5 in Supporting Information S1). For Toluca, the surface chemistry was analyzed after 6.6 × 1016 He+/cm2 and again after additional irradiation with 2.8 × 1018 H+/cm2. No difference in the chemical state was noted between proton and helium irradiations to imply the formation of new hydrogenated species, although the extent of chemical reduction was greater for 4 keV He+. Results are described below in detail for the Canyon Diablo sample; the Toluca spectra are broadly similar.

Figure 4

High‐resolution X‐ray photoelectron spectroscopy spectra for S (a) and Fe (b) observed on Canyon Diablo describe changes in chemistry with fluence, where the differently shaded photoelectron peaks indicate distinct chemical states. (a) The S‐2p3/2 and 2p1/2 photoelectron peaks show that the surface sulfur remains primarily troilite 𝑆2‐, with a minor increase in concentrations of 𝑆2 2− and 𝑆𝑛 2− ion‐induced defects. (b) At low fluence significant Fe(II)‐O (cerulean) and Fe(III)‐O (light olive) features are apparent from atmospheric exposure along with FeS, while at higher fluence the oxide and troilite (blue) are chemically reduced to Fe0 (dark olive). Energy loss features (lightest green) also appear when oxide is present. Although the magnitude of both S‐2p and Fe‐2p features was observed to increase with removal of surface oxide and adventitious carbon, the ratio of S:Fe is reduced (e.g., Figure 3).

As the precise surface potential for troilite is unknown during XPS acquisition, the binding‐energy scale must be referenced for each spectral‐acquisition data set. Adventitious carbon (C‐1s = 284.8 eV) was used to affix the absolute binding‐energy scale, with subsequent spectra adjusted to the dominant sulfur (S‐2p) feature, which showed no significant chemical change with irradiation (Figure 4a). The S‐2p3/2 and S‐2p1/2 photoelectron peaks were fit with an area ratio of 2:1 to three doublets corresponding to, S  (troilite), and (surface defects) with a binding‐energy separation constrained by troilite measurements of Terranova et al. (2018). With irradiation, the fractional abundance of the sulfur initially increases as the sulfur‐poor surface oxide is removed and the underlying troilite is sampled by XPS.

Once the material sampled is exclusively troilite, the S:Fe ratio is observed to decrease under both He+ and H+ impact. The primary (∼70%) S‐2p3/2 peak at 161.5 eV is indicative of native Fe‐S bonds as expected for troilite (Figure 4a) (Loeffler et al., 2008; Schaible et al., 2019). Features due to defects as well as bond‐breaking and reformation with irradiation, for example, , increased ∼3.7% with 4 keV He+ (Figure 4a) and 1 keV H+. We observed no spectral shifts in the S‐2p feature that would indicate the formation of new species such as SOH (∼163 eV) or adsorbed/trapped H2S (∼162.8 eV) with proton impact. No indication of S‐O bonding was apparent at ∼168–169 eV that would indicate S‐O bonds (Loeffler et al., 2008; Terranova et al., 2018; Thomas et al., 2003); thus, any surface sulfates have been removed by a fluence of 1.5 × 1017 He+/cm2.

The Fe‐2p photoelectron peaks, a well‐separated spin couplet, within Canyon Diablo and Toluca show a convolution of features from the iron‐oxide surface, the bulk troilite, and sulfur‐depleted iron after irradiation. For the Canyon Diablo sample just prior to the oxide‐troilite interface at ∼1.5 × 1017 He+ cm−2, Fe‐S bonds (Fe2p3/2: 707.25 eV) make up only ∼10.5% of the observed Fe; Fe(II)‐O bonds (∼17.5% Fe2p3/2: 708.3 eV with ∼19% Fe2p3/2 shakeup: 713.6 eV) and Fe(III)‐O (∼53% Fe2p3/2: 710.7 eV) in the surface oxide comprise the remainder (Bagus et al., 2021; Biesinger et al., 2011; Brion, 1980; Descostes et al., 2000).

The prevalence of Fe(III)‐O is consistent with a measured O:Fe ratio of ∼1.5 in the survey spectra (Figure 3a), indicating a native Fe2O3‐type surface oxide. With irradiation, Fe‐molecular bonds are broken by the kinetic energy transferred to the solid, preferentially sputter removing oxygen (Figure 4b) (Dukes et al., 1999; Loeffler et al., 2008). After 3.6 × 1018 He+/cm2, ∼15% of the Fe remains as intact Fe‐S, while the primary (65% of the signal) Fe2p3/2 photoelectron peak appears at 706.4 eV, indicative of Fe‐Fe bonds (Fe0) based on electron binding energy (Dukes et al., 1999; Mcintyre & Zetaruk, 1977; Skinner et al., 2004). This interpretation is also consistent with Fe‐Fe at ∼706 eV for a subsequent measurement of Fe0 in 99.95% Fe metal slug (Alfa Aesar) as well as with published values (e.g., Biesinger et al., 2011; Idczak et al., 2016). The remaining Fe2p3/2 photoelectron feature(s) (∼20%) indicates a small amount of Fe‐oxide remains within the near surface, likely due to ion‐induced shadowing (roughness) during sputtering and sputter‐induced layer mixing. This interpretation is consistent with information from the O1s photoelectron feature, which decreases in intensity by >90% (see Figure S5 in Supporting Information S1). A small shift to lower electron binding energy (283.8 eV) is observed with irradiation for recoil‐implanted C‐1s; a change in surface potential was considered, but contraindicated by resulting inconsistent O, S, and Fe chemistry. Instead, the C‐1s shift suggests the formation of a minor concentration (<1 at%) of iron carbide (Goretzki et al., 1989) or a transition in the carbon sp‐hybridization. While no differences in chemical outcome were observed between H+ and He+ irradiations, the extent of change over the course of the irradiation was slightly lower for protons (22% Fe‐Fe; 32% Fe‐S; ∼45% Fe oxides). The reduced rate of change for protons in comparison to helium is consistent with previous work on olivine (Dukes et al., 1999).

Oxide Layer Composition and Thickness Estimate

To estimate the thickness of the oxide layers on our samples, we used the XPS‐measured surface composition. This is 45.4 at% Fe, 44.9 at% O, 7.0 at% S, 1.3 at% N, 1.0 at% F, and 0.3 at% Ca for Canyon Diablo, as detected before any ion irradiation—without the contribution of adventitious carbon, which resides atop the oxide layer. After defining the oxide composition, we varied the thickness of the oxide layer in SDTrimSP to best fit the ion‐induced depletion rate of oxygen to that of the XPS data, while also matching the measured Fe and S data. Based on the minimized chi2 for all data points, oxide layers with thickness ∼15 and ∼4.6 nm, respectively, atop the Canyon Diablo and Toluca samples most accurately matched the SDTrimSP simulation to the XPS measurements (Figure 5). XPS showed some remaining oxygen (2.5% and 7.5%) in the samples after irradiation to a fluence of ∼3 × 1018 He+/cm2, while the SDTrimSP simulations indicate <0.1% remaining—likely resulting from the 1‐D simulation not accounting for surface roughness (Szabo et al., 2018).

Figure 5

SDTrimSP modeling was used to estimate the surface‐oxide depth for each meteoritic troilite section by modifying the oxide‐layer thickness in the simulation, then comparing computational results to measured composition. XPS data (black squares) are compared to oxide layers of varied thickness (dashed lines) as a function of ion fluence. Based on a chi2 analysis, oxide layers of ∼15 and ∼4.6 nm in the SDTrimSP simulation most accurately match XPS‐measured O‐concentration with fluence on the (a) Canyon Diablo and (b) Toluca samples. As in Figure 3, the gray bar indicates the fluence where the interface between the surface oxide coating and bulk troilite is reached in the experimental results.

Discussion

Ion‐Induced Surface Roughness and Microstructure

In SEM images, the polished meteorite surfaces showed nanoscale surface roughening, which evolved with irradiation (Figure 1). Although these nanofeatures were difficult to resolve individually even at high magnification, they appear to be uniformly distributed across the irradiated area with minimal variation in scale. AFM analysis indicated that this postirradiation nanotopography was approximately 5 nm in height and confirmed minimal variation in feature size across the irradiated area. We did not observe a significant difference in feature size or distribution between the Canyon Diablo sample (irradiated only with He+) and the Toluca sample (irradiated with both He+ and H+). With microscopy alone we could not conclude that these features differed in composition from the bulk troilite. However, as noted in the discussion of the Fe and S chemistry derived from XPS, we do observe the formation of surficial Fe0, and nanofeatures are known to result from subsurface defects and deformation that occur with the implantation of H/He (Auciello, 1984; Scherzer, 1983). We do not directly observe any physical structures larger than nanoscale via SEM/AFM, and the roughness we do observe is orders of magnitude smaller than previously observed iron whiskers and nanoparticles in naturally space‐weathered samples (e.g., Matsumoto et al., 2020, 2021). This is not too surprising, given the importance of diffusion and surface conditions in controlling the mechanism of sulfur depletion as noted by Matsumoto et al. (2021) and as we further discuss below.

XPS Sulfur Depletion

Physical sputtering is largely driven by elastic and inelastic collisions that exchange momentum between the projectile ions and atoms/electrons in the material, and is strongly affected by the elemental composition of the surface material, the existence of adsorbed layers, the chemical state (e.g., oxidation state of Fe), and the presence of implanted ions/atoms. Atoms within the solid are sputtered from a target when light keV ions are backscattered from heavier target atoms within the solid, transferring momentum toward the surface in subsequent interactions (Baragiola, 2004; Betz & Wehner, 1983). The proportion of momentum transferred is a function of mass, energy, and scattering angle. With the onset of irradiation, the near‐surface region can become transiently enriched in the lower‐mass component of a binary target by momentum‐exchange with backscattered light (H, He) ions. With continued irradiation, a steady‐state low‐mass atom depletion at higher fluence (∼1017 ions/cm2) typically results (Betz & Wehner, 1983; Dukes et al., 2011). In dual‐component materials such as troilite, the non‐stoichiometric (preferential) removal of one species over another is anticipated, until a surface compositional equilibrium is reached. Note that this ion‐induced equilibrium is not necessarily equivalent to the material bulk composition (e.g., Taglauer & Heiland, 1976).

We observed Fe enrichment within the surface (<10 nm) of irradiated meteoritic troilite (Figure 3), with an average concentration of ∼70 at% Fe after ∼3 × 1018 ions/cm2, resulting in an S:Fe ratio of ∼0.36 and 0.22, for He+ and H+, respectively, with normal ion incidence on a polished target. This is roughly twice the S‐depletion measured by Loeffler et al. (2008) on 4 keV He+ irradiated‐troilite pressed pellets (S:Fe ∼0.60). Increased S‐loss from our polished sample was expected, as surface roughness reduces sputtering yields due to topographic shadowing and enhanced sputter redeposition. Models estimate a 25%–90% reduction in sputtering yield (Carey & McDonnell, 1976; Cassidy & Johnson, 2005; Hapke & Cassidy, 1978), while experimental tests using granular surfaces suggest a 15%–67% decrease (Boydens et al., 2013; Loeffler et al., 2009). Strong surface enhancement of Fe in troilite and in pyrite after 1.5 keV Ar+ and O+ irradiation has been previously observed by Tsang et al. (1979). Energy‐Dispersive X‐ray spectroscopy (EDS) analyses have also shown increasing S‐loss in synthetic FeS and MgS, inversely proportional to depth after irradiation by 4 keV He+ and 5 keV Ga+, suggesting universal S‐depletion in metal sulfides with ions (Keller & Rahman, 2014; Keller et al., 2013).

Comparing Proton and Helium‐Induced Sulfur Depletion

To identify resultant compositional variation between proton and helium irradiation, the contribution of the surface oxide must be removed. Data normalization is done by shifting the total fluence of sample with the thicker oxide (Canyon Diablo) such that the oxide‐substrate interface occurs at the same position; this position becomes the nominal null fluence. We used the oxygen composition of ∼50%, to mark this location, which appears in the Canyon Diablo sample at 3.2 × 1017 He+/cm2 and at 2.4 × 1017 (He+, H+)/cm2 for Toluca; this is well beyond the initial irradiation by He+ (6 × 1016 He+/cm2) in the He+/H+ irradiated sample, allowing a direct comparison of proton to helium irradiation effects. Error bars in Figure 6 describe the maximum uncertainty in S/Fe atomic concentration, inclusive of photoelectron transition and sensitivity factor selection. However, the measured relative error for S and Fe within each individual data set is less than 13.5% (as detailed in Text S3 of Supporting Information S1).

Figure 6

Comparison of the troilite surface composition as measured by X‐ray photoelectron spectroscopy after removal of the oxide layer, shows similar S and Fe removal rates for 1 keV protons (open points) and 4 keV helium ions (closed points), as described by the change in (a) concentration with fluence and (b) the S:Fe ratio. Data are shifted so that fluence at the oxide‐troilite interface (50 at% O) is located at the origin. Interpolated curves between points are meant to guide the eye. Error bars represent the maximum compositional variation due to selection of Fe photoelectron feature (Fe‐2p vs. Fe‐3p vs. Fe‐2p3/2) and experimental error, derived from multiple data acquisitions at adjacent locations. Relative error within each data set is less than <13.5% for Fe and S, confirming near‐surface sulfur depletion with respect to iron for both H+ and He+.

We observe that the Fe and S curves trend similarly with fluence (1015−1018 ions/cm2) for both He+/H+ irradiations (Figure 6). From the XPS‐derived elemental abundance data (Figure 6), comparing the three data points for H+ to the overall trend for He+ suggesting rates of sulfur depletion for each ion species may not be as dissimilar as their measured sputtering yields would suggest (Roth et al., 1979). For a multicomponent material such as troilite, the equilibrium or steady‐state sputtering yield generally follows the monoatomic species component with the lowest yield; this is most often the constituent with the greatest mass, as lighter elements are typically preferentially removed (Betz & Wehner, 1983; Roth, 1983). Therefore for FeS, one would expect Fe enrichment with extended irradiation, and the sputtering yield would follow the trend of monoatomic Fe, measured as 10−1 atoms/ion for He+ and 10−2 atoms/ion for H+ (Bohdansky et al., 1977; Yamamura & Tawara, 1996). SDTrimSP calculations for 1 keV/amu He+ and H+ on stoichiometric troilite using standard surface binding energies identify similar order‐of‐magnitude differences, calculating YHe+/YH+ ratios of ∼8 (0.08/0.01) and ∼11 (0.045/0.004) for sulfur and iron, respectively, at low fluence (Figure S7 in Supporting Information S1).

These values imply that the Fe‐enrichment should occur roughly a decade faster for He+ irradiated troilite. However, this is not the case, as the final S:Fe ratio is 0.32 for the He+ irradiation (Canyon Diablo) and 0.22 for the H+ irradiation (Toluca) after ∼3 × 1018 ions cm−2. While only a minimal number of measurements with protons were acquired (due to the oxide overlayer), the similarity between He+/H+ curves over a wide range of fluences is intriguing and merits further investigation.

We found no evidence of new chemical species formation in the high‐resolution XPS data (e.g., Figure 4) to a level ∼100 ppm, suggesting that chemical sputtering enhancement of yield is unlikely, nor indication of beam contamination with heavier ions. Identification of a thermal dependence could provide more definitive information, as reactivity is a strong function of temperature (Roth, 1983). Irradiation by diatomic molecular ions such as H2 + have been shown to increase the sputtering from surfaces at high energies (>10 keV) due to overlapping of collision cascades; similarly, at energies <1 keV the sputtering yield deviates from twice the monatomic yield due to variation in the molecule effective mass (Dobes et al., 2011; Yao et al., 1998). However, low‐energy ion scattering measurements for 2 keV H2 + and microbalance analysis for 2, 4 keV O2 + have shown that for singly ionized diatomic molecules, the kinetic energy is divided quite equally between atoms (Dukes et al., 1999; Szabo et al., 2020). Since irradiations in the current project fall in the low keV range, between 0.5 and 10 keV, we do not expect molecular ions to enhance the sputtering rate. Lastly, radiation‐enhanced diffusion, the increase in target‐atom self‐diffusion precipitated by the growth of irradiation‐induced defects in the solid, can increase the mobility of S‐atoms to the surface to be sputter removed. This mechanism will be investigated for H+ and He+ irradiation in detail in Section 4.4.2.

Investigating Sulfur Depletion With SDTrimSP

The preferential removal of S over Fe with irradiation results from either (a) surface composition changes due to differences in constituent masses or surface binding energy (Us), (b) compositional changes in the altered layer through the ion penetration depth, or (c) both. Elemental concentrations through the ion range derive from thermal diffusion, radiation‐enhanced diffusion, recoil implantation and mixing, species surface segregation and radiation‐enhanced segregation (Betz & Wehner, 1983). Diffusion mechanisms assist in depletion by transporting the more volatile species (e.g., S) in vacancies and interstitials through a concentration gradient toward the surface, where it is subsequently sputter removed. Segregation traps the S at the solid‐vacuum interface where free energy is minimized. Both diffusion and segregation mechanisms are a function of temperature, and thus may perform differently on the surface of Psyche and other objects within the Asteroid Belt, where temperatures can reach from ∼240 K (e.g., Tosi et al., 2018) to significantly below. To identify the most‐influential parameters driving ion‐induced sulfur depletion in troilite, we simulated the process using SDTrimSP.

Effect of Surface Binding Energy

We evaluated the effect of surface binding energy (Us) on the FeS composition with ion irradiation by modifying the atom‐specific parameters in SDTrimSP. For the knock‐on collision regime of keV protons and helium interactions, the sputtering yields for troilite are highly dependent on internal kinetics governed by the target mass distribution, and to a lesser extent, by the atom surface binding energy. We confirmed this by independently varying Us for sulfur, while monitoring the effect on the target composition with irradiation and holding all other variables constant. We found that even large changes in Us (0 vs. 10 eV) for sulfur relative to iron did not affect the depletion profile significantly using either independent values for Us (model 1) or composition‐weighted values (model 2) (see Figure S6 in Supporting Information S1). Thus, we conclude that Us had a minor effect on the preferential sputtering of S, insufficient to explain the magnitude of the observed S:Fe depletion. Therefore, sulfur segregation or diffusion effects within the ion‐altered region must account for discrepancies between the simulation and XPS data. In subsequent SDTrimSP simulations, we retained Us default values, equivalent to the heat of sublimation.

Effect of Ion‐Induced Diffusion of Sulfur

SDTrimSP has recently been updated to include diffusion within the target, including (a) solid‐state thermal diffusion, where given sufficient thermal energy, target atoms move through the solid‐state lattice filling vacancies or occupying interstitial regions; and (b) radiation‐enhanced diffusion, in which the mobility of target atoms is enhanced by the greater ion‐induced defect density. In addition, simulations may also include (c) diffusion of the implanted projectile via pressure‐driven transport and by (d) outgassing of the projectile due to sputter‐induced compositional gradients. At 300 K target atoms are mobilized exclusively by concentration gradients within the damage region, diffusing toward the surface at rates dependent on concentration and on the temperature‐dependent diffusion coefficient with a probability dependent on location within the ion range (nearby atomic displacement). SDTrimSP employs variations of Fick's laws to describe the diffusion of target atoms toward the surface, as detailed in Supplementary Data section (see Text S5 in Supporting Information S1) and in von Toussaint et al. (2016).

We evaluated the effect of S‐diffusion by comparing the simulated target composition with fluence to XPS laboratory data, enabling and disabling diffusion in SDTrimSP. The compositional fit to data was vastly improved when damage‐driven diffusion was included in the model (Figure 7), suggesting that radiation‐induced diffusion plays a critical role in establishing S:Fe equilibrium values. We were able to estimate the sulfur diffusion coefficient ηT by running multiple SDTrimSP simulations with different settings and obtaining a best fit against our experimental data. SDTrimSP does not provide default diffusion coefficients except for H, Ar, and Xe. Since the ratio of diffusivity between helium and hydrogen was found to be 0.897 through gypsum (Yang et al., 2013) and 1.012 through air (Cussler, 1985), both close to unity, we equated the He and H diffusion coefficients. For target atoms, no published values appear in the literature for sulfur diffusion through troilite; thus, we extracted these values from SDTrimSP‐XPS data best fits. Radioactive tracer measurements of diffusion through metals show that the diffusivity for S can be orders of magnitude larger than that of Fe (Haynes et al., 2016); thus Fe remained static in our models.

Figure 7

Inclusion of damage‐driven diffusion for sulfur in the SDTrimSP simulation has a strong effect on modeled surface compositional changes in meteoritic troilite with ion irradiation. With no sulfur diffusion (η S = 0) in the model (a and b: dotted curves), the simulations do not replicate the measured surface composition (squares) for fluences beyond 5 × 10−17 ions cm−2. (a) Adjustment of the sulfur diffusion coefficient (dashed and solid curves) improves the model, where a chi2 best fit to S, Fe, O, and C compositions was identified for η S = 1 × 103 Å4 per He+(solid curve). (b) Similarly, inclusion of damage‐driven diffusion and variation of the coefficient for sulfur (dashed and solid curves) in the Toulca troilite identified a best fit with ηS = 5 × 102 Å4 per H+ (dash‐dot curve). Implanted H and He were also allowed to diffuse in the simulations. As in Figures 3 and 5, the gray bar indicates the fluence where the interface between the surface oxide coating and bulk troilite is reached in the experimental results.

Using the order of magnitude estimations, the values for the damage‐driven sulfur diffusion coefficients that have the lowest chi2 when compared to the Canyon Diablo and Toluca XPS data were, respectively, ∼1 × 103 and ∼5 × 102 Å4 per atom, roughly 10−3 of the helium/hydrogen coefficients; thus, we determined η (S) = 7.5  ×  102 Å4 per atom (Figure 7).

To quantify the sulfur diffusion in standard units (cm2/s), we compared the concentration of sulfur at a depth of 150 Å in the Canyon Diablo target after two different He+ fluences, (1.2 × 1018 He+/cm2 and 3.6 × 1018 He+/cm2), resulting in an estimated sulfur flux through the sample of 7.6 × 107 atoms/cm2/s (see Text S5 in Supporting Information S1). Thus, we find that the sulfur‐diffusion coefficient D (S) through the damage region of irradiated amorphous FeS is approximately 5.6 × 10–22 cm2/s at 300 K. This is significantly larger than diffusion measurements for isotopically labeled 34S through natural crystalline pyrite as measured by Rutherford backscattering for temperatures 500–725°C; extrapolating the coefficient to 300 K yields DS = 1.75 × 10–37 cm2/s (Watson et al., 2009), although such an extrapolation below the explicitly defined temperature range introduces substantial uncertainty. There is not a large body of literature comparing diffusion rates for amorphous versus crystal structures (Lazarus, 1985). Coffa et al. (1992) found that the diffusivity for transition‐metal species through silicon was larger for amorphous phases and suggested that transitory trapping at defects may inhibit interstitial diffusion in amorphous structures. While troilite demonstrates continued crystalline phases with irradiation (Christoffersen & Keller, 2011), defects and damage may similarly inhibit diffusion, resulting in a larger sulfur‐diffusion coefficient.

Effect of Elemental Composition Within the Altered Layer

With the inclusion of diffusion processes for sulfur and helium (or hydrogen), the altered layer directly below the surface through the ion range was inspected in the SDTrimSP simulations (Figure 8). The simulation results were compared with our XPS measurements, as well as Keller et al. (2013) interpretations for 4 keV He+ (45° incidence; 2.3 × 1018 He+/cm2) irradiated FeS and Matsumoto et al. (2020) observations of naturally space‐weathered Itokawa FeS grains. In transmission electron microscopy cross sections, Keller et al. (2013) identified an Fe‐enriched surface region of 2–3 nm with S‐depletion throughout the outermost ∼10 nm of the irradiated troilite, which was interpreted as the formation of a metallic skin atop a strongly S‐depleted region roughly equivalent to the ion range. Similar results were found in natural Itokawa FeS grains exposed to the solar wind, along with vesicles throughout the damaged region to a depth of 40–50 nm (Chaves et al., 2021; Matsumoto et al., 2020). The vesicles, which in some cases are open to the surface, are expected to result from the aggregation of implanted solar gases, since they appear through the ion penetration region. Partially amorphized sulfur‐depleted rims were also found on lunar FeS grains (Matsumoto et al., 2021), although these were significantly thicker (∼80–100 nm), suggesting a contribution from other space weathering mechanisms such as meteorite‐impact vapor deposition or more energetic ions.

Figure 8

SDTrimSP model for the final composition of troilite after 3.6 × 1018 (H, He)/cm2 as a function of depth, including implanted projectile species, with flux of 1 × 1013 (H, He)/s/cm2. The ratio of S:Fe illustrates the significant S‐depletion in the near‐surface with helium irradiation (∼5 nm; dashed blue) and with proton irradiation (∼20 nm), with minor depletion (5%–10%) extending even beyond the projectile penetration range due to diffusion resulting from the compositional gradient within the solid. The projectile implantation depth extends to ∼60 nm for 1 keV H+ (solid green) and ∼130 nm for 4 keV He+ (dashed green).

Both Matsumoto et al. (2020) and Keller et al. (2013) results are consistent with our XPS measurements of S:Fe concentration, where 95% of the S‐2p and Fe‐3p peak intensity derived from the surface‐most, ∼8 nm, based on the calculated IMFP (Powell & Jablonski, 2010). These results are also consistent with SDTrimSP results for 4 keV He+ irradiation to the fluence used in our experiments on meteoritic troilite (3.6 × 1018 He/cm2), where a strong Fe‐enrichment (S:Fe ∼0.1) is seen at the surface, decreasing swiftly with depth such that S:Fe ∼0.7 at 5 nm and 95% bulk stoichiometry is reached at ∼10 nm (Figure 8). The implanted He appears to be mainly distributed through depths of 10–110 nm—directly below the significant S‐depletion region and diffusing well beyond the classical penetration depth of 48 nm (range + straggle) calculated via TRIM (Ziegler et al., 1985). The ability for projectile ions to appear well beyond the ion range has been noted by many researchers and has been ascribed to ion‐enhanced diffusion, lattice swelling, discrete breathers, and other mechanisms (de Broglie et al., 2015; von Toussaint et al., 2016; Zhang et al., 2010). The helium concentration decrease with proximity to the surface is due to the low Us, which results in preferential sputtering at the outermost monolayer; this creates a concentration gradient for helium—from bulk to the surface through the damage region—driving diffusion outward. Pressure‐driven outgassing of the unbound projectiles additionally increases their transport from the implantation position.

For 1 keV protons the damage layer thickness differs due to the decreased penetration depth, although the absolute near‐surface Fe‐enrichment is enhanced, similar to our experimental data. We suggest that, for identical fluence, the decreased S:Fe for protons is due to an increase in the density of recoil displacements localized in the near‐surface region (<10 nm) facilitating S‐diffusion with subsequent sputter removal.

Similar to helium, SDTrimSP describes the presence of hydrogen, enhanced in the 10–50 nm range after irradiation by 3.6 × 1018 H+/cm2. This correlates with the embedded deep‐vesicle region (40–50 nm) identified in the Itokawa troilite particles (Matsumoto et al., 2020). This trapped hydrogen is then available for the formation of new molecular species, e.g. H2O and H2S, in the irradiated grain rim region (Schaible & Baragiola, 2014; Starukhina, 2001; Zhu et al., 2019).

Solar‐Ion Weathering of 16 Psyche

Having benchmarked our model parameters with XPS laboratory data, we are able to predict the effect of solar‐wind irradiation on pure troilite at Psyche. A simultaneous mixture of 4 keV He+ (4%) and 1 keV H+ (96%) at a total ion flux (2.9 A.U.) of 2.4 × 107 ions cm−2 s−1 were used in the simulation. Since S‐diffusion plays the major role in the depletion process, the contribution of heavy ions (Ar+; O+, etc.) was not included in this simulation. However, appropriate flux is critical, as concentration gradients are driven by the counterpoint between the sputtering rate and subsurface radiation‐induced diffusion (von Toussaint et al., 2016), changing significantly over the course of the irradiation as a function of surface atomic concentration (Figure S7 in Supporting Information S1). Initially, S sputters more rapidly (0.08 S/He; 0.01 S/H) from troilite than Fe (0.05 Fe/He; 0.005 Fe/H), preferentially removing sulfur from the near surface, such that by the time steady‐state conditions—beyond 3 × 1018 ions cm2—are reached, SDTrimSP sputtering yields are dominated by Fe (protons: S: 0.004; Fe: 0.009; H: 0.004 atoms/H+ and helium: S: 0.03; Fe: 0.08; He: 0.02 atoms/He+) at Psyche solar‐wind fluxes (Figure S8 in Supporting Information S1).

Our results suggest a rapid depletion of surficial sulfur on Psyche's surface over the first 100 years (Figure 9), consistent with sulfur depletion observed on other airless asteroid surfaces, such as Eros (Nittler et al., 2001) and Itokawa (Arai et al., 2008), where S‐depletion appears to correlate with space‐weathered regions. From the simulation, we expect that the S:Fe ratio in surficial fresh‐regolith troilite on Psyche should decay from unity to a level of ∼7% as a function of time. The depletion curve is best fit with a logistic model: y = A 2 + [(A 1 − A2)/(1 + (t/t 0) )], where A 1 is the initial S:Fe ratio (1), A2 is the equilibrium ratio after extended irradiation (0.067), t 0 is the midpoint of change (1.21 years), and ρ is the fitting power (0.70). This type of fit is frequently used as an empirical model for sputter depth profiling to describe concentration changes at the interface between two dissimilar layers, such as FeS and the resultant S‐depleted Fe (e.g., Kirchhoff, 2012).

Figure 9

Sulfur depletion was quantified via SDTrimSP in the outermost monolayer (1 ML) for exposed troilite deposits within Psyche's surface regolith over time (Earth‐years), given a solar‐wind flux of 2.4 × 107 ions cm−2 s−1 for a combination of 1 keV/amu protons (96%) plus helium (4%) is calculated via SDTrimSP. The S:Fe ratio decays at a geophysically rapid rate from unity to a level of ∼7% within 10 kyr, fit with a logistic curve. The ability to observe this depletion using the Psyche Mission's GRNS Spectrometer will depend on multiple parameters, including the asteroid surface composition, troilite abundance, average grain size, and the regolith gardening rate.

Psyche's uppermost monolayer is expected to record the greatest extent of sulfur‐depletion from solar‐wind ion irradiation—of high‐significance for exospheric production; thus, we focus our SDTrimSP studies on depletion in this region. Secondary‐ion mass spectrometers directed toward the surface of an airless body are able to identify species sputtered from the surface monolayer of an airless body such as the Moon (e.g., Elphic et al., 1991; Johnson, 1989; Yokota et al., 2009) or Mercury (e.g., Raines et al., 2013; Zurbuchen et al., 2008). However, while Schläppi et al. (2008) identify solar‐wind ion sputtering as the dominant mechanism for exospheric production at asteroids 2867 Steins and 21 Lutetia and transient OH and H2O has been noted at Ceres; no exosphere has been definitively detected yet around an asteroid (Grava et al., 2021). Thus, the potential to identify a thin sulfur‐exosphere sputtered from the surface of Psyche is intriguing.

In addition, while ion irradiation directly depletes surficial (10 nm) sulfur in laboratory samples, regolith gardening processes at Psyche should cycle irradiate surface particles, thus propagating the sulfur‐depleted region to greater depth. Although gardening rates and depths have yet to be quantified on a metallic surface, studies of S‐complex asteroid surfaces suggest time scales ranging from 104 years (Loeffler et al., 2008) to 105 years (Willman et al., 2010) and depths from 1 cm (Loeffler et al., 2008) to 1 km (Housen et al., 1979).

Two instruments aboard the Psyche spacecraft will measure surface properties potentially relevant to space weathering. The Gamma Ray and Neutron Spectrometer (GRNS) will measure bulk elemental abundances in the upper 10–100 cm across large‐footprint (∼tens of km diameter) regions of Psyche's surface beginning in Orbit D, with the ability to detect sulfur with 15% measurement precision if it is present at >3 wt% abundance (Peplowski et al., 2018). The Multispectral Imager will infer surface mineralogy using eight filter bands in the visible to near‐infrared (∼400 to ∼1,100 nm) built into twin CCD cameras (Dibb & Bell, 2018). Changes in surface micro‐roughness and chemistry may both play a role in altering the optical reflectance, slope, and albedo of troilite and other minerals present on the surface of Psyche (Yon & Pieters, 1987). The size and relative abundance of troilite grains in the regolith as well as gardening rates and other concurrent sulfur‐depleting processes (e.g., Prince et al., 2020), will influence the detectability of weathering‐induced sulfur depletion by both instruments, while not being directly measured.

Once the GRNS measures the sulfur content of a region and mineralogy is derived from multispectral images, a suite of compositions with different mixtures of rock, metal, and sulfide will nonuniquely match these measurements. In a compositional mixing model, the troilite fraction may be assumed to have reached full depletion of its sulfur content, or the degree of weathering may be included as a free parameter. Compositional models must also account for sputtered sulfur, which if it does not escape into space (Schläppi et al., 2008), redeposits across a wide area of the asteroid surface. Finally, the presence, strength, and structure of a magnetic field at Psyche would affect the flux of solar wind at the asteroid surface (Elkins‐Tanton et al., 2020; Oran et al., 2018).

Conclusion

We experimentally simulated space weathering by solar‐wind ions on meteoritic troilite samples, irradiating with both 1 keV H+ and 4 keV He+. Irradiation resulted in an ion‐induced increase in sample surface roughness to ∼5 nm uniformly distributed across the irradiated area, with H+ and He+ ions appearing equally effective at roughening polished surfaces. We observed loss of sulfur due to ion‐induced sputtering, with H+ irradiation surprisingly similarly effective at removing sulfur relative to iron as He+. Sputtering yields for troilite are fluence‐dependent, with near‐surface sulfur preferentially sputter removed at low fluences leaving an Fe‐enriched surface.

Minimal chemical variation is observed in S, while Fe is reduced from air‐formed oxide and native troilite at high fluences to Fe0. We modeled these irradiations with SDTrimSP, concluding that surface binding energy has only a minor effect on the distribution of S within the solid; rather, sulfur diffusion effects in the ion‐altered region account for the magnitude of the observed S:Fe depletion. Without invoking gardening or other processes, detecting S‐depleted troilite with the Psyche spacecraft's instruments may be challenging. These findings should be generally applicable to any troilite‐containing airless planetary body.

Supporting Information S1

Click here for additional data file.