A JWST Study of CO₂ on the Satellites of Saturn

Solid state CO₂ is seen throughout the outer solar system, even at temperatures where crystalline CO₂ should not be stable over the age of the solar system (C. Ahrens et al. 2022). It is present on at least one Jupiter Trojan asteroid (I. Wong et al. 2024) and on all three icy Galilean satellites (T. B. McCord et al. 1998; S. K. Trumbo & M. E. Brown 2023; G. L. Villanueva et al. 2023; D. Bockelee-Morvan et al. 2024; R. J. Cartwright et al. 2024) and has been detected on most of the larger icy Saturnian satellites (B. J. Buratti et al. 2005; R. N. Clark et al. 2005, 2008; R. H. Brown et al. 2006; D. P. Cruikshank et al. 2010). CO₂ is also seen on Centaurs well inside the frost stability line of approximately 15 au (L. A. Lebofsky 1975). The presence of CO₂ above its stability temperature implies that CO₂ molecules are trapped within some other material. Suggestions have included trapping in crystalline or amorphous water ice (D. P. Cruikshank et al. 2010; D. Bockelee-Morvan et al. 2024; L. Schiltz et al. 2024), in organics (G. L. Villanueva et al. 2023; R. J. Cartwright et al. 2024), in mineral structures (C. A. Hibbitts et al. 2012), and in other materials (C. Ahrens et al. 2022). The source of the CO₂ is likewise unclear. Possibilities suggested include that the CO₂ is native to the body (C. A. Hibbitts et al. 2003), that it is created from irradiated organics or carbonates (D. Bockelee-Morvan et al. 2024; R. J. Cartwright et al. 2024), or that it might be a product of exogenous infall from comets or chondritic material (R. W. Carlson et al. 2009). Neither the source of the CO₂ nor its trapping mechanism has been firmly identified for any of these bodies.

One powerful method for understanding the trapping mechanism (which might be distinct from the source mechanism) of CO₂ is the precise position of the ν₃ asymmetric stretch band and of the ν₁ + ν₃ combination band (C–O stretch + asymmetric stretch). In crystalline CO₂ at 80 K, these bands appear at 4.268 μm and 2.698 μm, respectively (A. Oancea et al. 2012). When a CO₂ molecule is trapped within a foreign material, the molecular shape and bonds are distorted slightly, yielding wavelength shifts of several nanometers or more. Each trapping mechanism yields a characteristic shift, though in some cases different mechanisms can yield similar shifts, rendering identification ambiguous. Shifts in the 4.268 and 2.698 μm bands are independent, and when they are combined, we can better distinguish different trapping mechanisms.

While CO₂ was originally detected from the Galileo spacecraft (for the Galilean satellites) and the Cassini spacecraft (for the Saturnian satellites), recent observations of the icy Galilean satellites from JWST have shown that at higher sensitivity and higher spectral resolution, CO₂ on the surfaces of these bodies is considerably more complex than previously understood. Europa has CO₂ most strongly concentrated in Tara Regio—a region of extensive chaos—with a ν₃ doublet at 4.249 and 4.268 μm (S. K. Trumbo & M. E. Brown 2023; G. L. Villanueva et al. 2023). The longer wavelength is consistent with crystalline CO₂, which should be extremely unstable at equatorial regions like Tara Regio. Interestingly, a 2.697 μm singlet also appears, again at a similar wavelength to crystalline CO₂. G. L. Villanueva et al. (2023) model the CO₂ doublet as a combination of trapping in methanol ice and the presence of crystalline CO₂, while S. K. Trumbo & M. E. Brown (2023) point to lack of a detection of any organic materials and the instability of crystalline CO₂ and suggest trapping of oceanic CO₂ gases or the irradiation of carbonates as possible sources and trapping mechanisms.

Ganymede shows a broad, variably asymmetric ν₃ band with central wavelengths between 4.257 and 4.260 μm at low latitudes, where the wavelength is correlated with the optical albedo and thus possibly related to the presence of water ice. At high latitudes on the leading hemisphere, the line shifts to 4.2695 μm, close to the expected wavelength of crystalline CO₂ (D. Bockelee-Morvan et al. 2024). While the high-latitude CO₂ is plausibly trapped in amorphous water ice, the source and trapping mechanism of the low-latitude CO₂ remains unknown. Callisto has the strongest CO₂ absorption of the Galilean satellites and shows a 4.250 μm band on the trailing hemisphere that shifts as far redward as 4.258 μm in the regions near the Asgard basin (R. J. Cartwright et al. 2024). No satisfactory explanation of the source or trapping of the CO₂ has been made.

For the Galilean satellites, precise identification of sources and trapping mechanisms is stymied by the range in temperatures, compositions, irradiation environment, and subsurface structure of each of the satellites. The icy Saturnian satellites, in contrast, present a more uniform case study in trapped CO₂. The inner satellites (interior to Titan) have surfaces that are nearly pure water ice, with only trace amounts of localized dark material, while the outer satellites (external to Titan) include more abundant dark material, presumably sourced from Phoebe and its associated ring (A. R. Hendrix et al. 2018). The similarities and differences in CO₂ abundances and wavelength shifts on these bodies can allow us to disentangle the effects of ice versus non-ice material, the effects of organics, and the effects of the sources and trapping mechanisms throughout the Saturnian system. While observations from Cassini/VIMS were capable of detecting the CO₂ on these objects (D. P. Cruikshank et al. 2010), observations with JWST allow both significantly higher sensitivity and higher spectral resolution. Here, we take advantage of these characteristics to study trapped CO₂ across the Saturn system.

We obtained JWST NIRSpec spectra from 1.8 to 5.2 μm of Mimas, Tethys, Dione, and Rhea—mid-sized icy satellites interior to Titan, and of Hyperion, Iapetus, and Phoebe—mid-sized satellites exterior to Titan with at least some dark material. In all cases except Hyperion and Phoebe, which orbit nonsynchronously, we separately obtained spectra of the leading and of the trailing hemispheres. Observational details are given in Table 1. The observations were obtained using the G235H grating at the shortest wavelengths to prevent saturation, and the G395M grism at the longer wavelengths to increase the sensitivity to CO₂. In addition, we analyze data from Program 1250 of G235H and G395H spectra of the leading hemisphere (only) of Enceladus. In our observations, all targets were observed with four dithers in the NIRSpec integral field unit (the Enceladus observations obtained only two dithers). The target was visually confirmed in the data, then the data were processed through the empirical point-spread-function-fitting pipeline described in M. E. Brown et al. (2025) using JWST CRDS context file jwst_1252.pmap and pipeline version 1.14.0 (H. Bushouse et al. 2022). Observations of the solar analog P330E (L. Colina & R. Bohlin 1997), from Program 1538, were reduced identically. The stellar spectrum was resampled at the wavelengths corresponding to the combined Doppler shift of the star and the satellite geocentric and heliocentric velocity. The satellite spectrum was then divided by the stellar spectrum to yield a reflectance spectrum with the solar lines removed, as shown in Figure 1. While the JWST pipeline returns uncertainties, we find that empirically derived uncertainties determined by the scatter of the data about a polynomial fit (for regions with no absorption features) provide a more robust result, and such uncertainties are used throughout.

Table 1. JWST Observational Parameters

Satellite	Date and Time	Exposure	r	$\dot{r}$	Δ	$\dot{{\rm{\Delta }}}$	Longitude
		(s)	(au)	(km s−1)	(au)	(km s−1)	(deg)
Mimas (leading)	2023 Oct 31 05:47	291	9.75	−12.3	9.32	14.4	52
Mimas (trailing)	2024 Jul 24 11:03	291	9.67	13.9	8.95	−6.5	262
Tethys (leading)	2023 Oct 30 18:39	116	9.75	−11.2	9.31	15.7	78
Tethys (trailing)	2024 Jul 20 08:03	116	9.67	10.7	9.00	−11.1	271
Dione (leading)	2023 Nov 9 01:38	116	9.75	−10.4	9.46	18.3	98
Dione (trailing)	2024 Jul 22 01:36	116	9.67	9.37	8.98	−11.9	274
Rhea (leading)	2023 Nov 27 05:27	116	9.74	−8.30	9.76	20.7	76
Rhea (trailing)	2024 Jul 12 13:24	116	9.68	7.88	9.10	−16.3	261
Hyperion	2023 Oct 29 05:57	350	9.75	−4.9	9.29	21.7	⋯
Iapetus (leading)	2023 Oct 16 17:28	116	9.76	−3.6	9.11	19.8	80
Iapetus (trailing)	2024 Jul 25 15:56	116	9.67	2.6	8.93	−17.5	267
Phoebe	2023 Nov 20 22:50	1050	9.77	−2.4	9.69	26.9	65

Note. Exposure times are identical for the G235H and G395M settings. r and Δ refer to the heliocentric and JWST-centric distances, respectively, with $\dot{r}$ and $\dot{{\rm{\Delta }}}$ the heliocentric and JWST-centric velocities. Longitude is the subobserver longitude at the time of observation. Hyperion has no defined longitude system.

Download table as:  ASCIITypeset image

Zoom In Zoom Out Reset image size

At the broad scale discernible in Figure 1, the main features visible are the distinct absorptions due to water ice, with bands at 2, 3 and 4.5 μm. For the high-albedo nearly pure water ice surfaces, the 3 μm absorption feature goes to essentially zero reflectivity at 2.87 μm. Phoebe and the leading hemisphere of Iapetus have the least coverage of water ice, as can be seen in their elevated 3 μm reflectivities, and Hyperion and the trailing hemispheres of Dione and Rhea also have slightly nonzero 3 μm reflectivities. Also visible is the 3.1 μm Fresnel reflection peak. The wavelength of this peak shows the presence of ice in crystalline form, though the 3.2 μm secondary peak expected for pure crystalline ice is subdued (except on the trailing hemisphere of Iapetus), allowing the possibility of at least some amorphous ice in the top layers (R. M. Mastrapa et al. 2009). Finally, slight shifts in the ∼3.6 μm peak between the 3 and 4 μm bands show temperature differences, with the trailing hemispheres of Dione and Rhea shifted to slightly longer wavelengths, indicative of higher temperatures, as would be expected for these slightly darker surfaces.

An absorption band near the 4.27 μm ν₃ CO₂ region is detected on every hemisphere of every satellite observed, even where the band was not detectable in Cassini data. To more closely examine this spectral region, we divide each spectrum by an estimated continuum, obtained by performing a second-order polynomial fit to the region between 4.15 and 4.40 μm, excluding the region between 4.20 and 4.30 μm where CO₂ absorption might be present. We perform the same fit on data from the leading hemisphere of Enceladus, taken at higher spectral resolution in Program 1250. For the Enceladus spectrum, we rebin by a factor of 3 to approximate the same spectral resolution as the other satellites. The results are shown in Figure 2. All observations show absorptions near 4.25, 4.27 μm, or both. The absorptions on the outer satellites (defined by those beyond the orbit of Titan) are significantly stronger than those on the inner satellites and at slightly different wavelengths. We perform a two-Gaussian fit to each of the continuum-divided spectra in the 4.27 μm region to obtain best-fit models and uncertainties (where, again, the uncertainties in the data are estimated as the scatter in the region outside of the absorption features). The results of the fits are given in Table 2.

Zoom In Zoom Out Reset image size

In addition to the ∼4.27 μm ν₃ absorption, Hyperion and the trailing hemisphere of Iapetus have narrow absorptions near the location of the 2.7 μm ν₁ + ν₃ combination band. Many of the other satellites appear to have a broader feature at slightly longer wavelength. To examine these features more closely, we perform a similar analysis as above to the data, now performing a third-order polynomial fit to the region between 2.660 and 2.760 μm, excluding the regions between 2.690 and 2.703 and between 2.705 and 2.717 μm. Results are shown in Figure 3.

Zoom In Zoom Out Reset image size

From the continuum-divided spectra, narrow absorptions on Hyperion and the trailing hemisphere of Iapetus are evident as well as a broad absorption centered around 2.71 μm on many of the satellites. Uniquely, Phoebe and the dark leading hemisphere of Iapetus show no clear signs of absorption near 2.7 μm, while only Hyperion and the bright trailing hemisphere of Iapetus show the narrow band near 2.70 μm. We determine precise wavelengths, strengths, and widths of these absorption features by fitting the continuum-removed data to a single Gaussian. The results are given in Table 2.

The combination of the spectral characteristics near 4.27 μm and near 2.7 μm makes a crucial distinction clear: while the inner and outer satellites share similar sets of absorptions in the 4.27 μm region, the differences in the 2.7 μm region—including the narrow 2.70 μm absorptions on Hyperion and the trailing hemisphere of Iapetus and the lack of absorption in the 2.7 μm on Iapetus and Phoebe—require that the trapped state of the CO₂ on the inner and outer satellites is different, regardless of the similarity of the wavelengths of the 4.27 μm absorptions. We thus consider the inner and outer satellites separately below.

3.1. The Inner Icy Satellites

The CO₂ signatures of the mid-sized icy satellites interior to Titan have strong similarities. In all cases, an absorption near 4.272 μm is present, while on the trailing hemispheres of Dione and Rhea, a feature near 4.250 μm also appears, and the feature near 4.272 μm appears diminished. Likewise, a broad feature near 2.708 μm is present on the inner satellites, with the weakest such feature seen on Dione’s trailing hemisphere.

Based on the spectral similarities of the satellites, we assume that they each have the same sets of spectral features. We can thus obtain a higher signal-to-noise spectrum—allowing a more detailed view of the absorptions—by taking spectral averages. We first average both hemispheres of Mimas and Tethys and the leading hemispheres of Dione and Rhea to represent the spectra of surfaces dominated by the feature near 4.272 μm. Second, we average the trailing hemispheres of Dione and Rhea to represent the surfaces more dominated by the feature near 4.250 μm. These average spectra are shown in Figure 4. Performing Gaussian fits identical to those done for the individual spectra, we find that in both averaged spectra, the short wavelength bands are centered at 2.708 ± 0.002 μm, while the strongest longer-wavelength band is found at 4.2724 ± 0.0002 μm for the first average and 4.249 ± 0.001 μm for the second set of averaged spectra.

Zoom In Zoom Out Reset image size

Based both on optical imaging (P. Schenk et al. 2011) and the depth of the 3 μm absorption, the trailing hemispheres of Dione and Rhea have the strongest concentrations of dark nonwater ice material. The presence of the shorter-wavelength 4.249 μm absorption on the trailing sides of these two satellites leads to the hypothesis that this absorption is associated with the trapping of CO₂ in this dark material and that the weakness of the 4.272 μm absorption on these trailing hemispheres is due to this extra covering of dark material. While this shorter-wavelength feature is strongest on these two satellites, it appears to be present on all of the inner satellite spectra, where the asymmetric nature of the 4.273 μm band, with a broad wing to shorter wavelengths, is plausibly caused by the blending of a small amount of the shorter-wavelength absorption. Indeed, optical imaging shows that the dark material is seen throughout the inner satellites, even if dominant on the trailing hemispheres of Dione and Rhea (P. Schenk et al. 2011).

The composition of the dark material on the inner satellites remains unidentified, but the shift of the ν₃ line to shorter wavelength can be caused by the trapping of the CO₂ molecule into a tightly confined space (S. A. Sandford & L. J. Allamandola 1990). For example, cryogenic experiments with trapping in zeolites and clays showed absorption features ranging from 4.219 to 4.264 μm (C. Hibbitts & J. Szanyi 2007). R. N. Clark et al. (2005, 2008) have argued that the dark material throughout the Saturnian system is a fine-grained material, possibly a nanophase hematite or iron. While no relevant spectroscopic studies of CO₂ trapped by these types of materials or studies at these temperatures have been performed, such small iron oxides could be efficient at CO₂ capture (J. Baltrusaitis et al. 2011; A. Hakim et al. 2016).

The association of the shorter-wavelength 4.249 μm absorption with the dark material suggests that the longer-wavelength feature and the 2.708 μm absorption could be due to CO₂ trapped in the water ice, which dominates the remainder of these surfaces. Indeed, the measured wavelength for both spectral lines (2.708 ± 0.002 and 4.2724 ± 0.0002 μm) is nearly coincident with the values measured (2.705 and 4.2737 μm) for CO₂ absorption in a 100 K codeposited 50:1 mixture of amorphous water ice and CO₂ (S. A. Sandford & L. J. Allamandola 1990).

The inner satellites are all embedded within Saturn’s E ring, which is sourced from the geysers on the south pole of Enceladus (F. Spahn et al. 2006). These particles themselves are composed of water ice with possible inclusions of silicates, organics, and CO₂ (J. K. Hillier et al. 2007; F. Postberg et al. 2008). While observations of satellite albedo, color, infrared spectra, and particulate grain sizes have led to varying conclusions about the interaction between the surfaces of the satellites and the E-ring particles (i.e., C. J. A. Howett et al. 2018), Cassini radar observations suggest that these inner satellites are coated with decimeters of fresh clean water ice from the E ring (A. Le Gall et al. 2019). If correct, it is likely that the observed CO₂ is trapped in amorphous water carried in from the E-ring particles. The actual source of the CO₂ is less certain. While it could plausibly be carried in with the E-ring particles or formed from radiolysis of embedded organic materials, it is also possible that the CO₂ is native to each individual satellite and is released from the surface or subsurface only to be trapped by the ice.

3.2. The Outer Satellites

The three observed satellites that are exterior to Titan have stronger CO₂ absorptions and appear to have different forms of CO₂ than those interior to Titan. Figure 5 shows the continuum-removed spectra from the 2.7 μm ν₁ + ν₃ CO₂ absorption region and the 4.27 μm ν₃ absorption region for these satellites. As with the inner satellites, there appear to be two separate trapped states of CO₂, reflected in a shorter- and longer-wavelength component of the ν₃ absorption. The two lines both contribute strongly and overlap across the band, so precise estimation of band centers is less certain, but the shorter-wavelength feature averages 4.2487 ± 0.0004 μm on the two spectra where the lines are strongest. Likewise, the longer-wavelength feature averages 4.2730 ± 0.0003 μm on the two spectra where this component is strongest.

Zoom In Zoom Out Reset image size

On Phoebe, spatially resolved observations from Cassini found that the CO₂ absorption is most strongly associated with the dark material (R. N. Clark et al. 2005). The leading hemisphere of Iapetus is known to be coated with this dark material, which spirals inward through the large Phoebe ring (A. J. Verbiscer et al. 2009). The location of the dominant CO₂ absorption on the leading hemisphere of Iapetus matches that of Phoebe to the 2σ level. The 4.2730 ± 0.0003 average position of this line is close to, but not coincident with, the 4.2724 ± 0.0002 line of the inner satellites, though the strong blending of the lines for the outer satellites makes the reliability of the differences between these values less certain.

We thus associate this 4.2730 μm absorption with the dark material on Phoebe and Iapetus. The brighter trailing hemisphere of Iapetus and the interiors of the craters on Hyperion contain minor amounts of the dark material, and indeed, a long-wavelength absorption appears weakly in the spectra of both the leading hemisphere of Iapetus and of Hyperion, where they are blended with the stronger, shorter-wavelength absorption at 4.2487 μm. While the lower signal-to-noise Phoebe data show some variability around 2.7 μm, the higher signal-to-noise data on the dark hemisphere of Iapetus make it clear that this dark material shows no clear indications of any absorption due to the 2.7 μm ν₁ + ν₃ band.

A plausible source for CO₂ in the dark material of Phoebe is the irradiation of organic material in close proximity to water ice. Laboratory experiments have found that irradiation of ice-covered organics at temperatures similar to that in the Saturnian system produce a ν₃ CO₂ band at 4.271  ±  0.002 μm, consistent with the wavelength measured here (O. Gomis & G. Strazzulla 2005). Unfortunately, these experiments do not examine the 2.7 μm region, so it is unclear if the lack of ν₁ + ν₃ absorption seen on Phoebe and the dark hemisphere of Iapetus is also reproduced. Interestingly, the C–H stretch absorption features in the 3.4 μm region suggested to be present by D. P. Cruikshank et al. (2014) do not appear in these JWST data, though smaller irregular satellites do appear to have features in this region (M. Belyakov & M. E. Brown 2025). More analysis is required to understand the presence and state of organic material on these satellites, but the existence of at least some organic materials on a dark outer solar system body such as Phoebe seems likely.

The association of the longer-wavelength 4.2730 μm absorption with the dark material on the outer satellites suggests that either the Phoebe-sourced dark material and its associated CO₂ are transported to these other satellites or that the dark material is transported without the CO₂ (or loses its CO₂ upon impact) and the CO₂ is then regenerated on the satellites themselves through subsequent irradiation. Some evidence for the latter hypothesis is that spatially resolved Cassini spectra of these satellites suggest that the CO₂ absorption is strongest in the transition regions between the bright and dark materials, where the dark material would have the highest access to water that can aid the radiolytic production of CO₂ (E. E. Palmer & R. H. Brown 2011).

If CO₂ is being continuously created on these satellites from the irradiation of freshly delivered organic material, the strong CO₂ signature on the icy parts of these bodies are then plausibly produced from the same mechanism but then separately trapped in that icy material. It is surprising that CO₂ trapped in ice has a different wavelength in the outer satellite system than in the inner, but the presence of the 2.6975 μm absorption exclusively on the brighter, icier trailing hemisphere of Iapetus and on water-ice-rich Hyperion suggests that the CO₂ is associated with ice on these satellites but in a physically different way than on the inner satellites.

While some experiments on trapping of CO₂ in water ice have shown small secondary absorption features near the 4.248 μm feature seen here (M. S. Gudipati et al. 2023; L. Schiltz et al. 2024), no trapping mechanism in either crystalline or amorphous water ice has been measured to have a prominent peak at this position. Additionally, no trapping mechanism has ever been found to make a narrow 2.6975 μm feature like those seen here (though this weaker combination band is often not studied in laboratory experiments). These observations point to the need for substantially more laboratory work on the trapping of CO₂ and point to plausible materials on which to focus future work.

3.3. ¹³CO₂

With the detections of the strong CO₂ bands on the outer satellites, we consider the possibility of the detection of the 4.384 μm absorption due to the isotopologue ¹³CO₂. Figure 6 shows the continuum-divided region near the predicted ¹³CO₂ line and Gaussian fits to the Hyperion, Iapetus (leading and trailing), and Phoebe spectra. No absorptions are seen in most of the spectra, but the trailing hemisphere of Iapetus shows a 4.3% ± 0.5% absorption centered at 4.383 ± 0.001 μm, consistent with the expected ¹³CO/¹²CO₂ band. While ratios of band strengths between the ¹³CO₂ and ¹²CO₂ bands cannot be used to directly infer the ¹³C/¹²C ratio, as the trapping mechanisms can affect these molecules differently, the values are informative for comparison between the bodies. We find 1σ upper limits of 0.020, 0.008, and 0.012 for Hyperion, the leading hemisphere of Iapetus, and Phoebe, respectively, and we measure a ratio of 0.019  ±  0.02 for the trailing hemisphere of Iapetus. R. N. Clark et al. (2019) report values measured from VIMS data on Iapetus (dark material) and Phoebe (full disk) of 0.012 ± 0.001 and 0.053 ± 0.006, respectively. While the VIMS Iapetus value is a plausible combination of some fraction of the leading and trailing hemisphere values measured here, the high abundance of ¹³CO₂ reported for Phoebe would require a 4.9% absorption feature, which can be ruled out here at the 5σ level.

Zoom In Zoom Out Reset image size

The ¹³CO₂/¹²CO₂ ratio measured for the trailing hemisphere of Iapetus is higher than the upper limits measured for both Phoebe and for the leading hemisphere of Iapetus, reinforcing the conclusion above that the Iapetus leading hemisphere CO₂ differs in source, trapping mechanism, or both from that of the trailing hemisphere and of Phoebe. The less stringent Hyperion upper limit remains consistent with the measured value for the leading hemisphere of Iapetus, in agreement with our suggestion that the CO₂ at these locations are related. While we cannot directly infer the ¹³C/¹²C ratio from these results, we note that the ratios are broadly consistent with the measured interstellar isotopic ratio of 0.014 ± 0.003 (A. C. A. Boogert et al. 2000) and with the terrestrial value of 0.011.

The satellites of Saturn have at least two separate sources of CO₂ and at least four separate trapping mechanisms. The inner satellites have a CO₂ component with a ν₃ absorption at 4.2724 ± 0.0002 μm and a broad ν₁ + ν₃ combination band at 2.708 ± 0.002 μm. This CO₂ appears to be associated with the ice component of these satellites, with the likely source either the E-ring particles or CO₂ native to each satellite. A second component on the inner satellites has a ν₃ absorption at 4.249 ± 0.001 μm and no corresponding ν₁ + ν₃ absorption. This CO₂ appears to be trapped in the dark materials most strongly seen on the trailing hemispheres of Dione and Rhea. The source could again be native or could be the dark material itself. The outer satellites have a CO₂ component with a ν₃ absorption at 4.2730 ± 0.0003 μm that has no measurable ν₁ + ν₃ component. This CO₂ appears to be associated with the abundant dark material present on Phoebe, the leading hemisphere of Iapetus, and parts of Hyperion. The CO₂ is plausibly created from irradiation of organics associated with this dark material. A second CO₂ component on the outer satellites has a ν₃ absorption at 4.2487 ± 0.0004 μm and a narrow ν₁ + ν₃ band at 2.6975 ± 0.0001 μm as well as the same broad 2.708 ± 0.002 μm band seen in the inner satellites. This component appears associated with the icy regions of the outer satellites, and the source could either be native or CO₂ released from the darker regions.

Zoom In Zoom Out Reset image size

These observations have interesting implications for the icy Galilean satellites and the state of their CO₂ as well (Figure 7). Interpretations for the CO₂ detected on the Galilean satellites are sometimes similar to the interpretations we have made here for the Saturnian satellites, though in some cases the similarity of the interpretation is in spite of large spectral differences. The CO₂ signature on Europa appears intriguingly similar to that on Hyperion and on the trailing hemisphere of Iapetus. All three bodies exhibit the ν₃ band as a doublet, with wavelengths of approximately 4.25 and 4.27 μm, and a narrow ν₁ + ν₃ feature near 2.7 μm. On the Saturnian satellites, the differences between Phoebe, Iapetus, and Hyperion demonstrate that the longer-wavelength ν₃ feature is associated with the dark, plausibly organic material, while the shorter ν₃ peak and the narrow ν₁ + ν₃ band are associated with the water ice (or at least the icy regions). On Europa, however, the 2.7 μm combination band appears more spatially correlated with the longer-wavelength ν₃ peak (G. L. Villanueva et al. 2023), suggesting that a different interpretation is necessary for these very similar signatures. Indeed, Europa shows no spectral signatures of organic material or evidence of Phoebe-like dark material (S. K. Trumbo & M. E. Brown 2023). No satisfactory explanation for the spectral signatures of CO₂ on Europa has yet to be presented, and the resemblance to Iapetus and Hyperion remains a mystery, though upcoming additional spatial coverage of Europa by JWST may help to resolve some of these ambiguities. On Ganymede, the CO₂ at the poles has been interpreted to be trapped in amorphous water ice, and the 4.27 μm band appears similar to that of the inner Saturnian satellites (D. Bockelee-Morvan et al. 2024). Unfortunately, no shorter-wavelength (G235H) JWST NIRSpec data of Ganymede exist, so no comparison to the 2.7 μm region can be made. The equatorial CO₂ on Ganymede resembles none of the features on Saturn nor on the other Galilean satellites and remains a mystery. On Callisto, the ν₃ CO₂ band is similar both to that associated with the dark material on the trailing hemispheres of Dione and Rhea (which has no associated 2.7 μm absorption) and to that associated with the water ice on the outer Saturnian satellites (which has a narrow 2.7 μm absorption). R. J. Cartwright et al. (2024) suggest that, at least on the trailing hemisphere, the CO₂ forms radiolytically in and is trapped by carbonaceous material, though laboratory studies of the irradiation of organic materials show longer-wavelength ν₃ bands (i.e., O. Gomis & G. Strazzulla 2005). The similarity of the ν₃ absorptions on the dark terrains of Callisto to that of the CO₂ associated with the dark contaminants of the inner Saturnian satellites is intriguing, and, again, a comparison of the 2.7 μm regions would be instructive, though, again, no 2.7 μm data for Callisto have yet been obtained. On the leading hemisphere of Callisto, the ν₃ band center shifts in a way consistent with the inclusion of a CO₂ component mixed in water ice, reminiscent of the longer-wavelength ν₃ on the inner satellites of Saturn where a broad 2.71 μm absorption is also present.

Observations of the Saturnian system, and a comparison across satellites and across wavelengths, have finally allowed a consistent picture of the sources of CO₂ and the mechanisms of its trapping to begin to be assembled. Significant work remains, however, to fully understand CO₂ here and elsewhere in the solar system, including additional laboratory work exploring trapping materials and mechanisms proposed here, a more complete set of observations of the Galilean satellites, and a more in-depth examination of the dark material on the outer satellites of the Saturnian system.

We thank two anonymous referees for helping us to strengthen the arguments put in this paper. This work is based on observations made with the NASA/ESA/CSA James Webb Space Telescope. The data were obtained from the Mikulski Archive for Space Telescopes (MAST) at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-03127 for JWST. These observations are associated with program #3716. Support for program #3716 was provided by NASA through a grant from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-03127. The specific observations analyzed can be accessed via doi:10.17909/30px-vq56.

Facility: JWST (NIRSpec) - .