Pwyll and Manannán Craters as a Laboratory for Constraining Irradiation Timescales on Europa

Bland et al. (2021) recently updated the available Galileo/SSI data set of Europa to account for uncertainties in the spacecraft's position and pointing which resulted in poor alignment of overlapping images, with some images displaced by more than 100 km from their correct location. We used the individual processed and projected "level 2" Galileo/SSI images with updated spatial information from Bland et al. (2021), which we acquired from the USGS-hosted SpatioTemporal Asset Catalog (STAC). ¹ We selected three images for our analysis. The first is a cutout from a larger image taken on 1996 December 19. We selected a region near Pwyll crater which covers an area ∼700 km wide with a spatial resolution ∼1.3 km pixel⁻¹ and shows the low-albedo crater and proximal ejecta surrounded by an extensive blanket of bright ray ejecta. The second image, taken on 1997 December 16, shows a close-up view of Pwyll, including the dark crater and proximal ejecta at a spatial resolution of ∼150 m pixel⁻¹. The last image, taken on 1998 March 29, shows Manannán crater, Belus linea, and the surrounding region at a spatial resolution of ∼1.65 km pixel⁻¹. All of the selected SSI images were taken with the clear filter at a relatively high solar illumination angle, which allowed for the best visible contrast between the dark material in the craters and proximal ejecta and the higher-albedo material of the bright ray systems. These SSI images can be seen in Figures 6(a), (d), and (g), respectively.

We used both of the existing high-spatial-resolution Galileo/NIMS observation of Pwyll crater and its ejecta. The first is NIMS cube 12ENCPWYLL01A, taken on 1997 December 16, which covers the crater, the dark proximal ejecta, and only a very small amount of the surrounding bright ray ejecta at a spatial resolution of ∼1.5 km pixel⁻¹. We also used NIMS cube E6ENSUCOMP01A, taken on 1997 February 20, which covers a larger area to the west of Pwyll crater and includes the bright ray ejecta and some older background terrain at a spatial resolution of ∼4.5 km pixel⁻¹. Finally, we used NIMS cube C3ENLINEA01B taken on 1996 November 6, which covers Manannán crater, its ejecta, and much of the surrounding terrain, including Belus linea, at a spatial resolution of ∼11 km pixel⁻¹. The NIMS cube 12ENCPWYLL01A has been previously analyzed by Fanale et al. (2000), while the other two NIMS observations have remained unpublished. The data were downloaded from the NASA Planetary Data Archive (Carlson 1998) as calibrated I/F mosaics. From the Galileo/NIMS spectral image cubes PDS data set (Carlson 1998), we used the individual PDS products E6ENSUCOMP01A-MSY02.IOF, 12ENCPWYLL01A-MSY02.IOF, and C3ENLINEA-01B-MSY04.IOF.

The included spatial information was calculated based on projection onto Europa according to spacecraft position, target position, and scan platform orientation, which are known to have relatively large uncertainties. We found that there were small offsets between the spatial information in the NIMS cubes and the recently updated SSI images from Bland et al. (2021). We therefore applied small shifts in latitude and longitude to each of the NIMS data cubes to more closely match the SSI images. We empirically determined the optimal offsets using the results of our linear spectral modeling efforts (see Section 2.3 below). In particular, we mapped the total water-ice abundance of each pixel, where the higher water-ice abundances should correspond to the higher-albedo regions such as the bright ray ejecta in the SSI images (e.g., Fanale et al. 2000) and the lower water-ice abundances correspond with the darker, less icy regions such as the crater interior, proximal ejecta, and older background terrains. The shapes of the bright ray ejecta are taken from Leonard et al. (2024). We applied an offset of −0.25° in latitude and +0.75° in longitude for cube 12ENCPWYLL01A, +0.2° in latitude and −0.25° in longitude for cube E6ENSUCOMP01A, and −0.75° in latitude and −0.75° in longitude for cube C3ENLINEA01B. All of these applied offsets are small when compared with the field of view for their respective observations.

We performed a simple linear spectral modeling analysis to explore the relative contributions of water ice, hydrated sulfuric acid, and salts to the spectra in and around Pwyll and Manannán craters. We based our spectral library on that of King et al. (2022), who modeled much of Europa's surface from a combination of global Galileo/NIMS and Very Large Telescope/SPHERE observations. This spectral library includes crystalline and amorphous water ice, several hydration states of sulfuric acid, and a variety of cryogenic salts. The specific laboratory spectra included in the library and their relevant properties are listed in Table 1. Water-ice spectra, for both crystalline and amorphous water ice, were calculated using laboratory optical constants (Mastrapa et al. 2008) and the Shkuratov model (Shkuratov et al. 1999) for grain sizes of 5 μみゅーm, 50 μみゅーm, 200 μみゅーm, and 1 mm, and a volume filling fraction of 0.7.

Each NIMS spectrum was analyzed by fitting a linear combination of laboratory spectra from the reference library, where the model spectrum, M_λらむだ , is described by

$$\begin{eqnarray}&&{M}_{\lambda }=C\displaystyle \sum _{i}{w}_{i}\cdot {S}_{i,\lambda }\ \ \ \mathrm{where}\ \displaystyle \sum _{i}{w}_{i}=1,\end{eqnarray} \tag{ 1 }$$

where w_i are the weights associated with each laboratory end-member spectrum, S_i . The additional scalar scaling factor, C, accounts for small variations in the reflectance between the calibrated NIMS observations and the measured laboratory spectra. These differences may be due to photometric effects caused by differing illumination of the laboratory samples and Europa's surface, as well as errors associated with the use of a linear mixing model. Typical values of C range from ∼0.4 to 0.7 for NIMS cube 12ENCPWYLL01A, ∼0.3–0.5 for NIMS cube E6ENSUCOMP01A, and ∼0.8–1.2 for NIMS cube C3ENLINEA01B. We used the nonnegative least-squares algorithm implemented in SciPy v1.10.1 to solve for the best-fit spectral model at each NIMS pixel.

We map the relative contribution of sulfuric acid and salts to the best-fit spectral models for each NIMS cube. We calculate the abundance of sulfuric acid in each pixel by summing the individual coefficients of all three laboratory sulfuric acid spectra, which have different grain sizes and hydration states. Similarly, we report the relative abundance of the "salty" material by summing the coefficients of all of the salts included in our spectral library. While our library includes a variety of hydrated salts which are potentially present on Europa's surface, the specific composition of Europa's dark endogenic material is not well known. The near-infrared spectra of Europa's dark hydrate material, as well as many of the candidate salt species, are mainly characterized by asymmetric distortions to the water-ice bands and lack specific characteristic absorptions. This lack of distinguishing spectral features, particularly at the low spectral resolution of NIMS, results in degeneracies between various salt endmembers, which make positive identification of specific species on Europa's surface difficult (e.g., King et al. 2022). We therefore choose not to report the specific coefficients for each individual salt, but rather only consider the total contribution of the non-water-ice and nonsulfuric-acid "salty" hydrated material to the spectral model, hereafter referred to as the salt abundance.

The measurement uncertainty in the NIMS observations is not well known, and the influence of nonquantifiable radiation noise means that a simple error estimate based on the inherent scatter of each spectrum cannot robustly predict the true error. We therefore adopt a Monte Carlo bootstrap method to estimate the error in our model abundances, where we use the idea of a constant chi-squared boundary to predict the 1σしぐま confidence limit for each of our model weights, as described in Press et al. (1992, Section 15.6). We first use a bootstrap sampling method to generate 1000 synthetic spectra for each NIMS spectra, with the same number of data points as the original spectrum, by drawing N (wavelength, I/F) pairs with replacement from the original spectrum. We then use the same spectral library and chi-squared minimization algorithm as before to estimate a new set of model abundances for each synthetic spectrum and calculate a set of corresponding chi-squared values, with respect to the original NIMS spectra for which we want to find error bars. Each of these new chi-squared values, calculated for a set of model coefficients which have been perturbed away from the best-fit abundances, are necessarily larger than the minimum chi-squared for the original NIMS spectrum. As described in Press et al. (1992), there is some region within which the χかい² value increases by no more than a set Δでるたχかい² which defines an M-dimensional confidence region. We find the approximate location of the constant Δでるたχかい² boundary that corresponds to the 1σしぐま confidence region by considering the range of model abundances which have a corresponding chi-squared value in the lowest 68% of the set of chi-squared values. We then use the range of model abundances present in the set to estimate 1σしぐま error bars for each of the M model abundances corresponding to the specific library spectra, as well as the error bars for our combined compositional categories of water ice, sulfuric acid, and salts. These estimated 1σしぐま error bars are reported throughout the paper along with the best-fit abundance values.

At both Pwyll and Manannán, the dark material in the crater floor and proximal ejecta and the bright ray ejecta blankets are depleted in sulfuric acid relative to nearby older terrain. Pwyll, the younger of the two craters, appears to be more significantly depleted in sulfuric acid. Assuming the background terrain concentrations represent the equilibrium concentration of sulfuric acid for the irradiation environment near each crater, this suggests that the radiolytic sulfur cycle has not yet had enough time to build up an equilibrium concentration of H₂SO₄ at either Pwyll or Manannán. This is consistent with the results of Hendrix et al. (2011), who found that a 280 nm absorption feature attributed to SO₂, a possible byproduct of the radiolytic sulfur cycle, had a lower band strength near Pwyll with increasing band depth farther from the crater. Together, these results place a strong lower limit of the age of Pwyll and Manannán craters on the equilibrium timescale of the sulfuric acid cycle operating on Europa's trailing hemisphere.

While the exact ages of Pwyll and Manannán craters remain unknown, we can use their estimated ages and make some simplifying assumptions to calculate a rough, order-of-magnitude estimate for the equilibrium timescale of Europa's radiolytic sulfur cycle. Bierhaus et al. (2001) estimate Pwyll is ∼3–18 Myr old based on solar system cratering rates. However, estimates of ion sputtering erosion from Paranicas et al. (2001) indicate that the top 1 cm of ice should be redistributed in approximately 0.6 Myr, which we would expect to degrade Pwyll's hemisphere-crossing bright rays and suggests that Pwyll may be even younger, perhaps less than 1 Myr (Bierhaus et al. 2009). This young age is consistent, within a factor of 2, with the Zahnle et al. (2003) estimate that a 20 km crater should form once every ∼2 Myr and that Pwyll appears to be the freshest and youngest large crater on Europa. While a concrete age estimate for Manannán crater does not exist in the literature, Manannán is expected to be relatively young, postdates all nearby terrains including Belus linea, and likely formed under similar ice shell properties as Pwyll (Schenk & Turtle 2009), suggesting it may be the second youngest crater >20 km on Europa. We therefore use the cratering rates of Zahnle et al. (2003) to roughly estimate its age to be ∼2 Myr older than Pwyll. This is generally consistent with its observed limited rays, which may be the remnants of a once larger ray system which have subsequently been degraded by sputtering erosion, but has not yet fully degraded.

Considering the low end of the age estimates for Pwyll and Manannán, we assume that Pwyll is ∼1 Myr old and Manannán is ∼3 Myr old. We now carry out a very simple calculation to estimate a lower bound on the equilibrium timescale of the sulfuric acid cycle on Europa. As seen in Figure 6, Pwyll appears to have reached a sulfuric acid concentration which has a model abundance of ∼50% the equilibrium abundance of the background terrain. If we assume that a 50% deficit in the best-fit spectral model corresponds to a 50% deficit in the physical concentration of sulfuric acid, then, assuming a linear growth in concentration, we would expect it to take at least 2 Myr to reach the equilibrium concentration at Pwyll. Similarly, the Manannán crater and ejecta has a sulfuric acid abundance of ∼70% of its equilibrium abundance. If it has taken ∼3 Myr to produce this amount of sulfuric acid, then we might expect to reach an equilibrium concentration at Manannán after ∼4.3 Myr. Taking these together, we suggest that the equilibrium timescale for the radiolytic sulfur cycle on Europa's trailing hemisphere is likely to be at least a few million years, and perhaps even longer depending on the true ages of Pwyll and Manannán craters.

It is important to note that the equilibrium timescale for the radiolytic sulfur cycle may not be consistent across Europa's trailing hemisphere. Differences in the flux of bombarding particles and changes in the surface composition may lead to differences in the equilibrium concentration of hydrated sulfuric acid and the timescale for reaching that concentration. Pwyll and Manannán are located at similar angular distances from the trailing hemisphere apex (25^◦ and 30^◦, respectively). We therefore expect both craters to receive a similar flux of cold sulfur ions, based on the irradiation models of Cassidy et al. (2013), who suggest that the flux of cold sulfur plasma to Europa's trailing hemisphere drops off by a $\cos \theta$ relationship, where θしーた is the angular distance from trailing apex. However, in our simple spectral models the background terrain near Manannán has a higher equilibrium concentration of hydrated sulfuric acid (∼50%–65%) than the terrain near Pwyll (∼40%–55%), consistent with the overall distribution of sulfuric acid hydrate found by Ligier et al. (2016). This suggests that the equilibrium concentration of sulfuric acid, and likely the timescale to reach that concentration, may depend on more than the sulfur ion flux. For example, the flux of 0.1–25 MeV electrons, as estimated by Paranicas et al. (2009), is expected to be higher at Manannán than Pwyll and may help explain the higher equilibrium concentration of sulfuric acid near Manannán. This is consistent with the hypothesis proposed by Dalton et al. (2013) that the production and steady-state concentration of hydrated sulfuric acid may depend on both the concentration of Iogenic sulfur ions and the flux of energetic electrons.

Laboratory estimates for the sulfur cycle equilibrium timescale on Europa's trailing hemisphere are typically ∼10⁴ yr (e.g., Carlson et al. 2002; Strazzulla et al. 2007; Ding et al. 2013). However, our results suggest that it may take several orders of magnitude longer to reach an equilibrium concentration of H₂SO₄ on Europa. It is very unlikely that both Pwyll and Manannán craters are young enough to be consistent with a sulfuric acid equilibrium timescale of a few thousand years, so we must conclude that the equilibrium timescale for the radiolytic sulfur cycle is significantly underestimated by these laboratory estimates. Strazzulla et al. (2007) and Ding et al. (2013) calculate these timescales using their experiment-derived reaction rates, estimates for the column density of hydrated H₂SO₄ molecules in the observable surface layers on Europa from Carlson et al. (1999), and estimates for the KeV–MeV sulfur ion flux from Cooper et al. (2001) and Dalton et al. (2013), respectively. These calculations would therefore underestimate the equilibrium timescale if the actual sulfuric acid column density at Europa was higher than that estimated by Carlson et al. (1999), or if the real sulfur ion flux is lower than that estimated by Cooper et al. (2001) and Dalton et al. (2013). However, it is difficult to achieve a 2 order of magnitude increase in the equilibrium timescale by simply adjusting the sulfuric acid column density and sulfur ion flux, which suggests that understanding the discrepancy between our observational constraints and the equilibrium timescales derived from laboratory experiments may require more careful analysis.

Laboratory experiments must necessarily replicate Europa's irradiation environment over much shorter timescales, perhaps reaching an equivalent irradiation dose of a few thousand to a million years on Europa in a matter of hours. Additionally, most laboratory experiments are performed for a single bombarding particle and energy, while various species and energies are acting simultaneously at Europa. Indeed, laboratory experiments have demonstrated that the specific intermediary products and steady-state concentration of H₂SO₄ may depend strongly on the temperature, energies, and projectiles involved (e.g., Moore et al. 2007; Strazzulla et al. 2007; Loeffler et al. 2011). Thus, using a laboratory-derived reaction rate for a single bombarding particle and energy may not provide a robust estimate for the actual equilibrium timescale at Europa, which may help explain the discrepancy between our observational constraints and the estimates of Carlson et al. (2002), Strazzulla et al. (2007), and Ding et al. (2013). Additionally, while radiolysis is responsible for the formation of H₂SO₄ on Europa, it can also destroy these molecules once formed. The balance of the production and destruction of hydrated H₂SO₄ on Europa is not well known, but is likely to be heavily particle and energy dependent. If H₂SO₄ is more readily destroyed within the complex irradiation environment on Europa's trailing hemisphere than in laboratory experiments with single-particle bombardment, then these destructive processes may act to prolong the timescale over which sulfuric acid reaches a steady-state concentration on Europa. Additionally, it is possible that thermal processes which are unimportant at laboratory timescales may become important on Europa, and could serve to slow down the equilibrium timescale of the radiolytic sulfur cycle. Indeed, Tribbett & Loeffler (2022) demonstrated that H₂O + H₂S + O₃ ice mixtures at Europa temperatures can undergo thermal oxidation reactions, even on laboratory timescales, which suggests that thermal reactions may affect both the steady-state composition and intermediary products of the radiolytic sulfur cycle.

We find that the dark, red material in the crater and proximal ejectas of both Pwyll and Manannán contain higher salt abundances than the surrounding bright ray ejecta. This dark material was likely excavated from Europa's subsurface during impact (Moore et al. 1998; Schenk & Turtle 2009), and is therefore expected to contain endogenic material which may have been further concentrated via preferential vaporization of ice during the impact process (Tomlinson & Hayne 2022). Indeed, Fanale et al. (2000) analyzed the NIMS observations shown in Figures 6(f), (g), and (h), and found that the visually dark and red material in the crater floor and proximal ejecta have a spectral signature dominated by asymmetric absorption bands, which they broadly attributed to non-ice hydrated material. In the framework of our spectral modeling analysis, these asymmetric bands can be explained by a variety of hydrated salts but are not well reproduced by a simple combination of water ice and hydrated sulfuric acid. Endogenic, presumably salty material within Europa's other geologically young features like the chaos terrains, bands, and ridges become progressively darker and redder toward the trailing hemisphere apex, potentially due to irradiation-induced reddening of salts. We expect that the endogenic salts within the Pwyll and Manannán craters and proximal ejecta have also been darkened and reddened via irradiation, based on their low-albedo and visually red coloration in the Galileo/SSI false-color images. If this is the case, the timescale for irradiation-induced reddening of salts on Europa's trailing hemisphere is apparently shorter than the ages of the Pwyll and Manannán craters. Taking the lower end of the age estimates for Pwyll, as in Section 4.1, we can place an approximate upper limit of ∼1 Myr on the irradiation-induced reddening timescale of salts on Europa's trailing hemisphere.