Plasma and Thermal Processing Leading to Latitudinal and Temporal Variability of the Trapped O 2 at Europa and Ganymede

Abstract

We describe the physical processes that affect the formation, trapping, and outgassing of O₂ at Europa and Ganymede. Following Voyager measurements of their ambient magnetospheric plasmas, laboratory data indicated that the observed ions, which were mostly ejected from volcanic Io, would in turn impact and sputter their surfaces. This would decompose the ice and produce thin oxygen atmospheres. More than a decade later, Europa’s O₂ atmosphere was inferred from observations of the O aurora, and “condensed” O₂ bands were observed in Ganymede’s icy surface at 5773 and 6275 Å. More than another decade later, their atmospheres were shown to have a dusk/dawn enhancement, confirmed by recent Juno data. Although the incident plasma produces these observables, processes that occur within the topmost surface are still not well understood. Here, we note that the incident plasma particles produce a nonequilibrium defect density locally in the surface ice grains. Defect diffusion within these grains leads to the formation of voids and molecular products, some of which are volatile. Although some volatiles are released into the satellite atmospheres, others are trapped at defect sites or trapped in voids, which create gas bubbles whose lifetimes (in steady state) are limited by the plasma-induced destruction rate. Here, we discuss how trapping competes with the annealing of the radiation damage. We describe the differences observed at Europa and Ganymede and roughly determine the observed trend with latitude of O₂ bands observed on Ganymede’s trailing hemisphere. This understanding is used to discuss the relative importance of “condensed” O₂ and O₂ adsorbed on regolith grains as atmospheric sources, accounting for dusk/dawn enhancements and temporal variability reported in “condensed” O₂ band depths. Since plasma-induced damage and thermal annealing timescales drive oxidant variability on icy moons (likely also Callisto, Dione, and Rhea), they can help determine volatile downwelling, a potentially metabolic source for their oceans, and upwelling of other trapped oxidants (e.g., CO₂), suggestive of ongoing geologic activity.

Keywords

Europa—Ganymede—Ocean worlds—Oxygen

1. Introduction

Energetic ions and electrons in the Jovian magnetosphere (Lanzerotti et al. 1978; 1983) impact the icy surfaces on Europa and Ganymede (e.g., Cooper et al., 2001; Nordheim et al., 2018) and cause the sputter ejection of molecules from their surfaces (e.g., Davis et al., 2021; Johnson, 1990; Teolis et al., 2017). The concomitant radiolysis is manifested by the production of thin atmospheres that contain H₂ and O₂(Johnson et al. 1982), products of the decomposition of ice in their surface grains (e.g., Reimann et al., 1984) as inferred from spacecraft and remote sensing observations (e.g., de Kleer et al., 2023; Hall et al., 1995; Roth et al., 2016). Equally interesting are the absorption bands seen in reflectance, also indicative of radiolytic decomposition of ice: “condensed” O₂ (Calvin et al., 1996; Spencer et al., 1995; Trumbo et al., 2021) seen also at Callisto (Spencer and Calvin, 2002), H₂O₂ (e.g., Trumbo et al., 2019,2023; Wu et al., 2024), O₃ (Noll et al., 1996; Ramachandran et al., 2024), and oxidized contaminants (Carlson et al., 2017) all trapped in the surface ice grains. The “condensed” O₂ bands have been suggested to result from dimer absorption in solid O₂ (e.g., He et al., 2021; Michalsky et al., 1999), although some data suggest that the band shapes might be better fitted if the ice also contains CO₂ (Migliorini et al., 2022). These bands are also present in gas phase O₂, either by collision-induced absorption (CIA) by pairs of O₂ or from van der Waals bound (O₂)₂ dimers. CIA requires densities far higher than what exists in the Galilean satellite atmospheres, and bound dimers in the gas phase are challenging to isolate in laboratory experiments. Since the satellite atmospheres are tenuous and the surface temperatures are such that O₂ is volatile down to the depth of the thermal wave, the “condensed” O₂ bands were suggested to be due to O₂ trapped in radiation-produced voids (Calvin et al., 1996; Johnson and Jesser, 1997, hereafter JJ). Indeed, since O₃ is not readily produced during the irradiation of ice, its presence in Ganymede’s surface is consistent with its production by excitation of “condensed” O₂ trapped in the large, near-surface, ice grains (Johnson and Quickenden, 1997, hereafter JQ). These observed bands are also consistent with the preferential loss of H₂ from the irradiated icy surfaces as suggested by ambient plasma measurements (e.g., Cooper et al., 2001; Szalay et al., 2024) and modeling (e.g., Carberry Mogan et al., 2022). Below, we refer to energetic particles generically as a plasma; we acknowledge that the term generally refers to bulk electrons and ions in the satellite tori.

Since individual O₂ molecules trapped at defect sites or in vacancies and voids are difficult to detect, the “condensed” O₂ bands, which are associated with the excitation of pairs of O₂ molecules, have proven to be useful. However, these bands, associated with multiple O₂ inclusions and often referred to as dimer excitations, do not account for all the O₂. Therefore, in addition to O₂ being driven into the satellite atmospheres and O₂ trapped in ice grains, we suggest that O₂ is transiently adsorbed and desorbed on grain surfaces and exists as a gas that permeates their porous regolith (Johnson et al., 2003). That is, the spatial and temporal variations in the ambient atmospheres have been suggested to be due to thermal release of individual O₂ trapped at defects (e.g., Johnson et al., 2019; Teolis and Waite, 2016). Interestingly, the “condensed” O₂ band depths exhibit significant temporal variability with times short compared with the satellite orbital periods (e.g., Spencer et al., 2019; Trumbo et al., 2021). This aspect confirms that radiation processing is active; it produces both gas-phase and trapped species in competition with their destruction and thermal annealing of the radiation damage.

Motivated by such observations, and by the modeling of the dusk/dawn atmospheric enhancements (Oza et al., 2018,2019), confirmed by recent magnetic field observations by Juno (Addison et al., 2024), below we discuss the physics of O₂ trapping in and its ejection into the very thin atmospheres on Europa and Ganymede, and we address the observed temporal and spatial variations. We first review aspects of the now extensive laboratory data and then describe our understanding of “condensed” O₂ formation as well as its relevance to the satellite observations. Useful satellite data are given in Table 1, and the activation energies of the relevant physical processes are provided in Table 2.

Table 1.
Relevant Satellite Data for Europa and Ganymede’s Sunlit Trailing and Leading Hemispheres

Region T_eq^a (K) $\bar{T}$ ^b (K) Thermal annealing time/Orbital period^c Plasma flux, f_ion^d (ions/s/cm²) Plasma bubble impact time/Orbital period^e $\bar{A_{b}}$ Bond albedo^f O₂ band depths (%)^g

E (T) 135 126 0.06 3 × 10⁸ 0.02 0.45 ∼0.3

E (L) 125 116 1.4 10⁸ 0.06 0.65 ∼0.3

G (T) 147 137 1.2 × 10⁻³ 6 × 10⁴ 53 0.37 ∼2

G (L) 142 132 4.2 × 10⁻³ 10⁷ 0.3 0.43 ∼0.7

Region	T_eq^a (K)	$\bar{T}$ ^b (K)	Thermal annealing time/Orbital period^c	Plasma flux, f_ion^d (ions/s/cm²)	Plasma bubble impact time/Orbital period^e	$\bar{A_{b}}$ Bond albedo^f	O₂ band depths (%)^g
E (T)	135	126	0.06	3 × 10⁸	0.02	0.45	∼0.3
E (L)	125	116	1.4	10⁸	0.06	0.65	∼0.3
G (T)	147	137	1.2 × 10⁻³	6 × 10⁴	53	0.37	∼2
G (L)	142	132	4.2 × 10⁻³	10⁷	0.3	0.43	∼0.7

Spencer (1987) (equatorial values).

Average hemispheric temperatures: $\bar{T} = \int_{0}^{\frac{π}{2}} T (θ) c o s (θ) d θ .$

Cao (2021) (Schmitt et al., 1989, Fig. 12). Also, Figure 2 modeled as t_anneal ∼ 10¹³s e^{(−Eanneal/kT)}.

f_i_on: Plasma ion flux (Poppe et al., 2018).

$Plasma ion impact time for a 4 nm bubble [t_{b} \sim {(π r_{b}^{2} f_{ion})}^{- 1}] . Proportional to$ the average bubble destruction time (Appendix) within the average plasma penetration depth.

Spencer (1987).

Spencer et al. (1995); Calvin et al. (1996); Spencer and Calvin (2002); Trumbo et al. (2021) (rough averages).

Table 2.

Activation Energies Discussed: exp(−E_a/kT)

Process	E_a (eV)
E_b,O₂ adsorption on grain surfaces^a	∼0.11–0.14
E_vm, vacancy formation in ice^b	∼0.34 ± 0.07
E_plasma,Plasma particle ejection of O₂^c	∼0.03–0.06
E_anneal,Crystallization Rate: Lab samples^d	∼0.46
E_O₂, “condensed” O₂ Ganymede^e	∼0.3

Johnson et al. (2019) Europa Dusk/dawn enhancement; Teolis and Waite (2016) Dione, Rhea seasonal variability; Lab data: Laufer et al. (2017); He et al. (2016).

Experimental values in Eldrup (1976).

Reimann et al. (1984), Teolis et al. (2017): from fits of ice sputtering yield of O₂ versus temperature.

Schmitt et al. (1989); Cao (2021).

This article based on the Trumbo et al. (2021) equatorial data.

2. Radiolysis

Whereas crystalline ice has an equilibrium density of defects determined by the temperature (e.g., de Koning, 2020), incident energetic ions and electrons produce an over-density of displacements and interstitials that form a “track” of vacancies and radicals along their paths through ice. The region around the “track” is transiently heated by the excess energy deposited, which leads to diffusion of these defects and reactions. After that, transient heat dissipates, and defect diffusion continues, determined by the background temperature. As the solid anneals, the vacancies and interstitials can recombine (annihilate) or be incorporated into various sinks. The diffusing radicals that react with vacancies contribute to annealing (i.e., recrystallization), or they can react to form new chemical species. The diffusing vacancies can also accumulate and produce voids (e.g., Okada et al., 2020). Voids produced by electron irradiation are readily seen in electron microscope studies of ice up to the peak temperatures on Ganymede, with their production and growth affected by impurities (Heide and Zeitler, 1985). During annealing, small voids can be absorbed by larger voids, which reduces the net surface to volume ratio, a process often referred to as Oswald ripening. This process also produces segregated precipitates in solids (e.g., Gusak et al., 2006). Void disappearance with increasing temperature is seen in studies of small grain, microporous ice, presumably due to mobility in the presence of extensive grain surfaces. Diffusion and reactions are, of course, enhanced along the pathway formed by penetrating radiation (Bénit and Brown, 1990) as well as along surfaces: for example, the atmosphere interface, grain boundaries, and internal surfaces in voids. Such processes have been extensively discussed, particularly for room temperature solids exposed to energetic particle radiations (e.g., Jiang et al., 2022; Wiedersich, 1972), and the early studies of the radiolysis of ice are summarized in JQ.

The diffusion of H atoms, produced by proton implantation or by molecular dissociation, has been shown to form H₂ in several materials (e.g., Behrisch and Eckstein, 2007), including ice (Christianson and Garrod, 2021). Depending on the temperature, the resulting H₂ can also diffuse efficiently (e.g., Patterson and Saltzman, 2021), often along grain boundaries, and can desorb at an external surface. However, when H₂ is formed in or enters a void, it can become more stably trapped and form a gas bubble. This is well studied and leads to the swelling of highly irradiated, room temperature solids (e.g., Vook et al., 1975) as well as to the formation of H₂ bubbles in ultraviolet irradiated low temperature ice (Tachibana et al., 2017). Therefore, data on fast proton and alpha particle production of H₂ and He gas bubbles in reactor walls were used early on (e.g., JJ) to help understand the “condensed” O₂ observations at Ganymede. That is, like H, radiolytically produced O can diffuse along surfaces or defect pathways and react at defect sites to produce transiently trapped O₂. Or, if the reaction occurs on a surface in a void, it can lead to more stably trapped O₂, which we will refer to at times as a bubble. The presence of oxygen bubbles in Antarctic and Arctic ices, formed by co-deposition, is well established (e.g., Miller, 1969) and even confirmed in thin samples (e.g., Bahr et al., 2001). However, interior O₂ bubbles have not been directly identified in radiation-processed ice. Such formation and trapping of O₂ was suggested early on by its release on warming relatively thick irradiated samples (e.g., JQ). Also, a luminescence feature in photolyzed ice was used to suggest that O₂ can form at an interior surface from diffusing O atoms (Matich et al., 1993). But convincing laboratory studies are still needed. Although attempts have been made to explain the formation of O₂ in irradiated ice using gas-phase reaction rates (e.g., Li et al., 2022), the relevant processes in solids differ significantly. That is, even at the highest temperatures on Ganymede (∼160 K), O₂ is not soluble in crystalline ice, so that defect sites and surfaces play a critical role, for example, appendix of Johnson et al. (2003). Therefore, below we focus on processes that occur in solids.

Because of the relatively efficient diffusion of H in ice at satellite temperatures, its subsequent desorption from irradiated samples enhances the production of oxidants (Brown et al., 1982). This has been confirmed by many subsequent experiments, as summarized by Teolis et al. (2017), and rediscovered regularly (e.g., Abellan et al., 2023). Of course, in the presence of carbon, sulfur, and so on, other oxidized species are produced due to the preferential loss of H₂, as seen in reflectance spectra (e.g., Carlson et al., 2017). Laboratory studies indicate that the production and ejection of O₂ exhibits a dose dependence consistent with the dissociation, on average, of two water molecules per O₂ ejected (e.g., Reimann et al., 1984) or with the formation and the excitation of a precursor (Sieger et al. 1998; Orlando and Sieger, 2003). In addition, the O₂ yield is small and independent of T below ∼60–80 K (e.g., Davis et al., 2021; Teolis et al., 2017), roughly consistent with defect mobility becoming efficient above ∼0.3–0.5 times the melting temperature (e.g., JJ). With increasing temperature, the ejection of O₂ from irradiated laboratory ice samples increases, indicative of an activated process, but with an unexplained activation energy (e.g., Reimann et al., 1984). Here, we suggest that the extracted activation energy is determined by the surface tension of ice (see Appendix A1).

Primary products are formed in irradiated ice other than those discussed above. As extensively observed, the OH produced can react with impurities or react to produce H₂O₂ (e.g., Mifsud et al., 2022), a species seen in irradiated ice samples as well as in the reflectance spectra of Europa and Ganymede (e.g., Trumbo et al., 2021). Although not volatile, when formed in voids, H₂O₂ inclusions can also build up. Therefore, Cooper et al. (2003) suggested that H₂O₂ molecules trapped in voids might react with increasing temperature or be excited by radiation and contribute to the production of O₂ bubbles, a suggestion not yet confirmed. Although O₂ formation and ejection increase with increasing ice temperature (e.g., Reimann et al., 1984), above ∼150–160 K rapid recrystallization occurs, and the O₂ yield drops. In contrast, H₂O₂ appears to be a stable product at lower temperatures (e.g., Teolis et al., 2017). Indeed, at Ganymede, peroxide is preferentially concentrated in the cold polar regions (e.g., Bockelée-Morvan et al., 2024), while the “condensed” O₂ bands are seen to be dominant at the warmer lower latitudes. At Europa, however, peroxide is also seen in some low latitude regions, possibly collocated with other trace species such as CO₂ (Trumbo et al., 2019; Wu et al., 2024). Therefore, the solid-state chemistry that occurs in ice (Johnson et al., 2003) still requires considerable work. Ice irradiation experiments at ∼10 keV suggest peroxide-rich regions on Europa (e.g., Tara and Powys regions) are radiolytically enhanced by the presence of CO₂ (Mamo et al., 2025), also likely affected by the available oxidant trapping described here. After we describe below how irradiation-induced O₂ can be formed and trapped in voids, using data in Tables 1 and 2, we consider its contribution to the icy satellite atmospheres.

3. O₂ Bubble Formation: Satellite Versus Laboratory Surfaces

Although laboratory studies show that the radiation-induced decomposition of ice leads to the formation of O₂ and H₂, the absence of direct detection of the “condensed” O₂ bands in irradiated samples remains a problem. The samples typically studied differ significantly in porosity and grain size from those on satellite (e.g., Cassidy and Johnson, 2005) and comet surfaces (Oza and Johnson, 2024). The vapor-deposited laboratory samples tend to be small grain, microporous and thin, in order to reduce charging by the incident particles. On the contrary, the reflectance data for the icy Galilean satellites suggest their surfaces are composed of large, primarily crystalline, grains of ∼100 µm to 1 mm (e.g., King and Fletcher, 2022; Ligier et al., 2019; Stephan et al., 2020). These surfaces are presumed to be fairy castle structures with a very high pore space between such grains. Because of their high porosity, the satellite regoliths are permeated with atmospheric volatiles, and the area accessible to incident radiation is much larger than the satellites’ geometric surface.

Surface temperatures and “condensed” O₂ band depths are extracted from the reflectance at visible wavelengths. However, these spectra depend on grain size, porosity, contaminant concentration, and radiation damage; thus, the reflectance data, although critical, are not simple to interpret. For instance, since crystalline ice is transparent in the visible, surfaces observed in reflectance can appear dark or bright in the visible for a number of reasons: dark due to, for example, significant concentrations of absorbing contaminants or highly transparent with few absorption and scattering sites; bright due to, for example, small grains or large grains with a high density of defects that act as scattering sites. Surfaces that are bright in the visible can also be produced by radiation damage (e.g., Famá et al., 2010; Johnson, 1985) or by returning H₂O ejected by sublimation, sputtering, or venting (e.g., Li et al., 2020), producing a fine-grain frost (Trumbo et al., 2023; Villanueva et al., 2023). Although modeling indicates that O₂ can be formed and trapped, even at very low temperatures in small grains in irradiated ices in the interstellar medium (Jin and Garrod, 2020), the satellite surfaces differ considerably. Therefore, in the following, we assume the observed “condensed” O₂ is primarily formed and trapped in voids produced by irradiation of relatively large ice grains. We imagine a hierarchy of sizes: average grain size ≫ visible wavelength ≫ average bubble radius. We also presume that plasma particles that produce the “condensed” O₂ are primarily those with penetration depths comparable to the depths sampled by the reflected photons.

In addition to the physical differences discussed, radiation dose rates in laboratory studies are many orders of magnitude larger than those experienced by grains in the icy regoliths. Whereas thermally driven recrystallization rates in laboratory studies of irradiated low-temperature ice samples are typically not relevant, that is not the case for all regions of the icy Galilean satellite surfaces. Because the fate of radiation damage is determined by a variety of diffusion processes, here we use the thermal annealing rate as a crude proxy for the processes that occur following radiation damage of ice (Table 3). Although crystallization studies of amorphous or vapor-deposited ice exhibit significant scatter (e.g., Baragiola, 2003; Cao, 2021), we use the result in Schmitt et al. (1989) as a guide, which is roughly consistent with an activation energy E_anneal∼0.46 eV. These data suggest a large range of annealing rates across the satellite surfaces, varying from minutes at the peak temperature on Ganymede’s trailing hemisphere to on the order of the orbital period at Europa and at high latitudes on Ganymede, as indicated in Table 1. Therefore, the lack of direct laboratory detection of the “condensed” O₂ bands in irradiated ice samples we suggest is due to the disparity between annealing and radiation damage rates, as well as the grain size issue discussed above. Although critically important, the application of the available laboratory data to the icy satellite observations is not straightforward.

Early studies of the release of H₂ and O₂ on warming of thick samples exposed to penetrating radiation are suggestive (JQ). New studies, possibly using electron microscopy (e.g., Abellan et al., 2023), on relatively thick annealed samples are needed. This is critical as the absence of direct laboratory observations of radiolytically induced “condensed” O₂ bands has led to the suggestion that O₂ cold traps exist in the topmost ice layer of the icy Galilean satellites (e.g., Vidal et al., 1997; Baragiola and Bahr 1998). However, even for porous icy regoliths, the thermal skin depths are several centimeters, which implies that low temperatures sufficient to form O₂ are unlikely. This is especially the case at low latitudes on Ganymede’s sunlit trailing hemisphere where “condensed” O₂ band depths are strongest at the equator ∼147 K.

Since incident plasma particles transiently supersaturate the local defect density, the growth of voids that results in bubbles of trapped volatiles is governed by the flux of vacancies versus the recrystallization rate (e.g., JJ; Jiang et al., 2022) as discussed above. However, incident radiation that penetrates a gas-filled void (bubble) can also cause dissolution of the bubble and dispersion of the trapped species. That is, bubble damage, which is typically ignored, competes with aggregation and growth. Under steady irradiation, a quasi-equilibrium is established: higher dose rates produce a high density of smaller, on average, bubbles that have higher interior surface energies (e.g., Gittus, 1978; Tachibana et al., 2017). Therefore, in the presence of annealing, the average bubble dimension varies inversely, but slowly, with the radiation damage rate. Noting that the defect production rate is roughly proportional to the energy deposition rate, dE/dt, JJ referred to an estimate of the average radius, r_b, of gas “bubbles” produced under steady energetic particle implantation as

r_{b} α {[D (T) / (d E / d t)]}^{x}

(1)

Here, x is a small exponent, x ∼ 1/4 in Gittus (1978), and D(T) accounts for the relevant diffusion processes. In our discussions below, the temperature dependence of the recrystallization rate of ice is used as a proxy for these processes. In addition, a rough estimate for the average void radius under steady irradiation, accounting for growth by vacancy diffusion and void destruction, is derived in Appendix A2. It has a similar form to that in Eq. 1 but with x ∼ ½, verifying the inverse dependence on the irradiation rate. These expressions suggest that a low flux of highly penetrating plasma particles, with a lower energy deposition per unit path length, penetrating a higher temperature ice can lead, on average, to larger bubbles.

Whereas Europa’s surface experiences a more intense and diverse plasma flux, at low latitudes, Ganymede’s field tends to deflect the heavy thermal ions that cause surficial damage and brightening. In the presence of higher average surface temperatures and lower fluxes of primarily penetrating particles, larger band depths on Ganymede should not be surprising. These “low latitude” observations, discussed below, appear to include the open-closed field line boundary region (Fig. 1 in Trumbo et al., 2021) in which the Juno-JADE instrument detected a flux of 30–100 keV electrons (Ebert et al., 2022). From Galileo encounters in the Jovian plasma sheet, simulations also suggest Ganymede’s mini-magnetosphere shields electrons below 40 MeV at the equator (Liuzzo et al., 2020), which, based on Eq. 1, enhances bubble formation. In addition to producing the observed O aurora, these electrons, 4.5 keV to 100 MeV, primarily deposit their energy in the surface ice (Waite et al., 2024) to depths (∼10–100 µm) roughly consistent with the grains being modified by lightly ionizing radiation.

FIG. 1.

Jovian plasma sheet latitudes $θ_{Plasma - Torus}$ versus $ϕ_{III}$ at Ganymede (dashed) during four observing periods spanning 1999–2020, summarized in Trumbo et al. (2021), suggestive of differences in energetic particle fluxes. The date of each observation, probed Ganymede W longitudes $(W)$ , and approximate integrated O₂ band area $d A ($ centered at 5773 Å) are indicated in the color bar, maximum at Ganymede’s trailing (T) 1999 August 28 ( $\geq 4 Å)$ decreasing near Ganymede’s leading (L) 1–2 $Å$ (2020), 0.5 $Å$ (1999), and back to 2.5 $Å$ (1999 September 23). All 1999 observations are near Ganymede’s equator at latitudes $θ_{G} (1999) \sim {3.2}^{\circ}$ N and $θ_{G} (2020) \sim {- 1.4}^{\circ}$ S. The approximate torus angle is indicated locally at Ganymede for visualization.

Calvin et al. (1996) compared the visible bands observed on Ganymede’s trailing hemisphere to those of “condensed” O₂. Based on the band shape and the local temperature, they suggested the best fits were either solid-γ or liquid O₂, and they estimated absorbance maxima from the data in (Landau et al., 1962): α_D ∼1.15 or 4.8/cm, respectively). Estimating the photon path length from reflectance data, they obtained ∼0.1–1% concentration of O₂ to H₂O. Subsequently, Hand et al. (2006) used the depth, δI, of the “condensed” O₂ band and the intensity, I, of the visible reflectance to estimate the fraction of O₂ in Europa’s surface ice grains. Assuming wavelengths small compared with the grain radius, r_g, but large compared with internal scattering sites, they wrote

δ I / I \sim (α_{D} {f_{O2}}^{2}) [(4 r_{g} / 3) {(1 + r_{0})}^{2} / (1 - r_{0})]

(2)

The term in square brackets is an estimate of the average path length of the reflected visible photons determined by the average grain radius, r_g, and by internal scattering centers indicated by a “reflectance” parameter r₀. Although they referred to a clathrate, they used an estimate for α_D ∼ 1.6/cm based on solid-γ O₂ at 0.577 µm with r₀ ∼0.97, obtaining f_O2 ∼1.2–4.7% in Europa’s surface ice (Hand et al., 2006; Hesse et al., 2022). If correct, then “condensed” O₂ is clearly the dominant reservoir of O₂ in the near-surface region (e.g., Johnson et al., 2003). But (α_D f_O2²) in Eq. 2 suggests that individual O₂ molecules are randomly distributed within the grains, leading to an overestimate of condensed O₂ fraction. Therefore, f_O2² should be replaced by f_γO2, the fraction of the pathlength in solid-γ O₂ which would give f_γO2 < ∼0.2% at Europa. However, the applied parameters have significant uncertainties, and the brighter surface at Europa has a sampling depth in the visible that differs from that at low latitudes on Ganymede. Assuming similar grain sizes and absorption coefficients, the ratio of their observed average band depths (∼6.7) times the ratio of their albedos (0.67) suggests, very roughly, a “condensed” O₂ fraction on Ganymede’s trailing hemisphere in the region sampled by the reflected photons is ∼4.5 times higher than the fraction in the region sampled by the reflected photons at Europa.

Since the pathlength through the regolith grains in the visible is critical, more detailed modeling is needed using new absorption coefficients (Calvin, 2025). This is especially so since the related infrared bands (e.g., Bockelée-Morvan et al., 2024) have not been detected as discussed in Calvin et al. (1996). Recently, Tinner et al. (2024) used steady-state sputtering measurements to estimate a broad range of possible O₂ concentrations in irradiated samples for a range of grain sizes, but they did not obtain dimer concentrations.

4. Ganymede Low Latitude Observations

Focusing on the available observations, we note that voids are potential sinks for trapping diffusing radicals, which can lead to the formation of more stably trapped O₂, as discussed above. Void growth, controlled by vacancy migration, is an activated process depending on the background temperature (e.g., Eldrup, 1976). In their summary of “condensed” O₂ observations at Ganymede, Trumbo et al. (2021; Fig. 3A) noted a very rough correlation of a drop in the “condensed” O₂ band area (∼factor of 3) at low latitudes with an increase in the albedo (∼0.39→0.47). This trend is consistent with the darker trailing hemisphere having larger band areas on average than the brighter leading hemisphere (Calvin et al., 1996), as well as at the brighter high latitudes. These observations suggest that the regolith temperature plays a critical role in affecting “condensed” O₂ band formation as discussed above. That is, at low latitudes on Ganymede, where the annealing rate is rapid compared with its rotation rate (Table 1), the variation of the albedo is consistent with a small average temperature drop, ΔT(ϕ), over the observing region centered on longitude ϕ from that centered on regions near ϕ ∼ 270^°. If the local temperatures are roughly in radiative equilibrium, then at longitude ϕ, dT(ϕ)/T(270°) ∼ (1/4) dA/[1 − A(270°)] in which A is the visible albedo. Using the values above, the change in albedo from near 270^° to near 360^° west longitude is consistent with a fractional temperature change of ∼5 K, ΔT(360°)/T(270°) ∼ 0.037, close to the average temperature change suggested by observations (Table 1). Assuming the “condensed” O₂ bands are determined in part by thermal processing with an activation energy E_O₂, and initially ignoring differences in the plasma flux and photon sampling depth, we approximate an averaged e ratio of the observed band areas as

< R (ϕ, 270 > \sim e^{- (\frac{E_{O_{2}}}{k T (270 ° W)} \frac{Δ T (ϕ)}{T (270 ° W)})}

(3)

Using the temperature ratio above suggests an effective activation energy E_O₂ ∼ 0.3 eV. This is a value close to the experimental value for activation energy of vacancy migration (E_vm ∼ 0.34 ± 0.07 eV; Eldrup, 1976), which determines the growth rate of voids in which, we suggest, “condensed” O₂ is formed and trapped.

Of course, the particle flux, background temperature, and photo-sampling depth all factor into the “condensed” O₂ formation process. Therefore, it might not be surprising that in a different Ganymede observing period, a similar fractional drop in band area was observed in going from near 270^° to 180^° west longitude, but the band areas were somewhat smaller (Trumbo et al., 2021; Fig. 3B). Since the temperature excursion is close to that estimated above, we suggest that drop in band area is due to a difference in the plasma flux. That is, although the background temperature can affect the net “steady state” density of the gas-containing voids, the bubble formation and destruction rates depend on the radiation damage rate compared with the annealing rate (Appendix A2). Plasma variability at Ganymede is primarily due to the tilt of Jupiter’s dipole and Io’s activity. Ignoring the latter, Figure 1 shows the plasma torus latitudes at Ganymede during the observing periods summarized in Trumbo et al. (2021). Although a simple correlation with magnetic latitude was not observed for Europa’s brighter surface with weaker band depths (Spencer et al., 2019), Ganymede’s optical aurorae roughly follow the expected fall off in electron density with latitude during eclipse $ϕ_{obs} \sim 0$ ° (Milby et al., 2024), which implies a difference in the radiation rate of the surface ice grains (Waite et al., 2024). The optical aurorae show variability similar to in situ data of the Io plasma torus at $ϕ_{III} \sim 115$ ° and 295° (Bagenal et al., 2015) when the plasma sheet is ∼0.4–0.7 R_J ( $θ_{Plasma Torus} \sim$ 2–4°). Although this is suggestive of electron density and magnetospheric variations, a better correlation is needed, and a more detailed description is required. This includes accounting for the shielding by Ganymede’s magnetic field, as well as incorporating electron number density oscillations as measured by in situ spacecraft such as Voyager, Cassini, and Galileo (Bagenal et al., 2015). Ultimately, a better understanding at the molecular level of the radiation damage, void growth, and annealing rates within the photon sampling depth is also needed.

Finally, it has been shown that the gas pressure in a bubble, p_b, can affect its stability (e.g., JJ). Under steady irradiation, p_b can build in a bubble of radius r_b until it exceeds the interior surface tension, γ_s. For much higher pressures, the gas molecules would tend to become soluble, so that

p_{b} < \sim 2 γ_{s} / r_{b}

(4)

Using γ_s ∼80 mJ/m² and p_b ∼n_O2kT, with an average temperature on Ganymede’s trailing hemisphere (∼147 K), the maximum column of O₂ confined per bubble is ∼(r_b n_O2) < ∼10¹⁶ O₂/cm². If we assume that n_O2 approaches the liquid density, then r_b < ∼0.04 µm. Of course, as is well documented in Earth’s atmosphere (McKellar et al., 1972), the “dimer” absorption also occurs during low-temperature collisions of O₂ molecules. Such collisions would be frequent for any density of O₂ trapped in voids. Therefore, molecular-level modeling of the formation and trapping of O₂ is needed.

5. Gas-Phase O₂ Sources

Radiolytically produced O₂ molecules ejected from the surfaces of Europa and Ganymede primarily return (e.g., Oza et al., 2019) and rattle around in their highly porous regoliths as discussed earlier. At any time, some of these return into the atmosphere thermalized, while others are transiently trapped at defect sites on the grain surfaces (Johnson et al., 2019). Therefore, these irradiated regoliths are permeated with transiently adsorbed O₂, as well as other volatile products. This is in addition to the O₂ formed and trapped in voids described above. In this way, the sources of O₂ to the atmosphere consist of newly produced and ejected O₂, ambient O₂ in thermal equilibrium with the local regolith, and the plasma-induced release of O₂ from near-surface bubbles.

Globally, the radiolytic production of new O₂ from ice must be in equilibrium with its destruction in the atmosphere and regolith and, to a much smaller extent, to downward geologic transport (e.g., Chyba, 2000; Johnson et al., 2003). The net destruction must also be in rough equilibrium with the loss to space of its companion decomposition product, H₂ (e.g., Szalay et al., 2024). Due to the accumulation of O₂ in the regolith, thermal desorption dominates direct plasma-induced production of atmospheric O₂ (Johnson et al., 2019; Oza et al., 2018). As the satellite surface temperatures vary spatially and the plasma irradiation is nonuniform, the atmospheric column densities of thermalized O₂ are asymmetrical (Oza et al., 2019).

Thermal desorption of O₂ into these atmospheres is suggested by the modeling of observations. A binding energy to grain surfaces, E_b ∼ 0.11 eV, was estimated from the seasonal variability of O₂ detected by the Cassini Ion and Neutral Mass Spectrometer (INMS) instrument as emanating from Dione and Rhea (Teolis and Waite, 2016). A very similar value, E_b ∼ 0.14 eV, was estimated (Johnson et al., 2019) to account for the size of the dusk/dawn enhancement in the O₂ column density at Europa (Oza et al., 2018; Roth et al., 2016). These extracted values are remarkably consistent with the laboratory studies (He et al., 2016; Laufer et al., 2017) and represent adsorption of O₂ at surface defect sites, presumably dangling H bonds. The thermal desorption rate can be roughly approximated by using the following equation:

{t_{d}}^{- 1} \sim p_{d} ν_{b} exp (- E_{b} / kT)

(5)

where p_d is a desorption probability, and ν_b is the effective vibrational frequency (∼10¹²/s; He et al., 2016). Using E_b ∼ 0.14 eV, an average temperature on Europa of 100 K, then t_d ∼ (10⁻⁵ s) if p_d ∼ 1. This is an order of magnitude longer than the time spent between collisions with grains and far shorter than the ballistic hop time above the surface. It confirms the rapid recycling of atmospheric O₂. Therefore, the average atmospheric O₂ source rate ∼10⁸/cm²s estimated at Europa (Oza et al., 2018) suggests a rough transient adsorption coverage of O₂ on grain surfaces directly exposed to the vacuum of ∼10³ O₂/cm².

Ice that has a significant density of voids can become fluid like (Tachibana et al., 2017), so that, under stress or temperature gradients, O₂ bubbles can migrate to grain boundaries, as seen in Antarctic ice cores (Miller, 1969). Therefore, at an external surface, trapped gas can be released, aided by sputtering and thermal erosion (Dadic et al., 2010,2019). Since these processes are slow in large grains, it has been suggested that satellite surface ice with relatively stable inclusions of oxidized products could be delivered geologically to their underground oceans (e.g., Pappalardo et al., 1998), possibly supporting biology (Chyba, 2000; Hesse et al., 2022; Johnson et al., 2003).

Although bubble migration to an external surface can contribute to the local atmosphere, destruction by incident plasma primarily determines the steady-state density in ice grains (JJ). That destruction is occurring within the photon sampling depth is indicated by the temporal variability observed in the dimer band depths (Spencer et al., 2019; Trumbo et al., 2021). That is, a flux, Φ, of energetic plasma particles can penetrate bubbles of radius r_b, in a time, t_b ∼(π r_b Φ)⁻¹. This can destroy the trapped O₂, cause reactions, as suggested by the presence of O₃, and drive products into the ice matrix. Since Φ is roughly proportional to dE/dt in Eq. 1, the flux of plasma particles can destroy bubbles, as well as contribute to their formation (JJ; Appendix A2). Using r_b <∼4 nm and a flux of highly penetrating plasma onto Europa of ∼10⁸/(cm²s) (Cooper et al., 2001), the destruction time is short compared with its orbital period (t_b < ∼0.1 t_orbit). Since the plasma flux is variable (Bagenal et al., 2015), at a minimum due to the tilt of Jupiter’s dipole (e.g., Milby et al., 2024), and the lifetime of bubbles within the photon sampling depth is short compared with the orbital period, the apparent higher variability in the “condensed” O₂ absorption band depth at Europa as compared with that at Ganymede we suggest is due to the higher radiation rate and lower annealing rate. Indeed, the thermal annealing timescales in Ganymede’s equatorial region, following Eq. 5 for ν_b = 10¹³s, E_a = 0.46 eV (Schmitt et al., 2009; Cao et al., 2021) results in a bulk crystallization timescale of only ∼10–40 min or 1–4 × 10⁻³ of an orbital period (Table 2, Fig. 2), far shorter than Europa’s annealing timescales. While recent 3.1 µm JWST observations of Europa’s Tara Regio and Powys Regio (Cartwright et al., 2025) imply recent surface modification of <15 days, our estimates suggest Europa’s leading hemisphere equator crystallizes as fast as ∼5 days (1.4 europan orbits). Although far slower than Europa and Ganymede’s trailing hemispheres, the leading hemisphere equator is far faster than northern latitudes at ∼100 K approaching ∼1 Myr timescales, consistent with amorphous ice features in the infrared (Cartwright et al., 2025). The thermal region depicted in Figure 2 toward long, approximately decadal timescales is consistent with ∼5 K surface temperature changes due to Jupiter’s eccentric ∼11.86 year orbit driving seasonal annealing observed on Europa by JWST (Mergny et al., 2025) and Ganymede.

FIG. 2.

Thermal annealing timescale estimated in units of hours (left) and Europa orbital time (right: vertical axis) following t_anneal ∼ 10¹³s e⁽⁻ ^E ^/ ^kT ⁾ for bulk ice crystallization at E_a = 0.46 eV (Schmitt et al., 1989). The gray region bounds the approximate uncertainty in water ice annealing from experiments (Cao, 2021) roughly ±0.2 eV (dotted lines). Horizontal lines highlight approximate plasma destruction times for each moon. Europa and Ganymede’s equatorial sunlit leading and trailing hemisphere temperatures are indicated for each moon, corresponding to data in Table 2. Two approximate regimes are highlighted: (1) plasma regime (top left), similar to Europa’s northern latitudes where amorphous ice is observed, and (2) thermal regime (lower right), where the annealing time is >1000× orbital and plasma timescales similar to Ganymede’s trailing hemisphere. O₂ melting, boiling, and critical points are indicated.

FIG. 3.

Radiolytic processes in surface ice grains in time. Diurnal temperature variations on Europa (85–130 K) and Ganymede (70–150 K) and variable energetic particles (∼keV → MeV) lead to observed O₂ bubble formation (radiation damage vs. annealing) observed via the 5773 Å O₂ dimer absorption bands. Watercolor painting: Emily Costello (Hawaii Institute of Geophysics and Planetology). Note: bubbles painted in the figure are natural nitrogen/oxygen bubbles in watercolor paint.

6. Discussion

We briefly reviewed aspects of the physics associated with the radiolytically produced O₂ observed in the icy regoliths on Europa and Ganymede, and the sources of oxygen into their thin atmospheres. In this, we were guided by the modeling of radiation-produced voids and bubbles in refractory solids, as well as laboratory data on the radiolysis of ice. Although the latter has been critical to our understanding of spacecraft and telescopic observations, we point out that care must be taken in directly applying these data. That is, as discussed above, the satellite surface ice tends to be highly porous with relatively large grains that experience annealing and relatively low radiation rates, whereas the laboratory samples are typically thin, small-grain, microporous samples that experience relatively high irradiation rates in which annealing due to the background temperature is often irrelevant. Therefore, direct observation of the radiation-induced dimer bands in laboratory ice samples has been problematic, which has led to considerable speculation. This is unfortunate, as some estimates of the radiolytically produced O₂ reservoirs at the icy satellites suggest that the “condensed” O₂ is the dominant reservoir of O₂ (Johnson et al., 2003). This is the case although bands other than those in the visible have not been identified (Calvin, 2025; Calvin et al., 1996), which suggests more detailed modeling of the reflectance spectra is needed.

Observations suggest to us that O₂ is trapped in the satellite surfaces in two forms: O₂ bubbles (“condensed” O₂) and transiently adsorbed O₂ on grain surfaces, as indicated by modeling of the dusk/dawn asymmetry (Johnson et al., 2019). Here, we also suggest that the observed variability of the “condensed” O₂ bands is due to the destruction of bubbles within the photon sampling depth by the penetrating plasma particles. Therefore, the higher plasma fluxes and lower temperatures on Europa appear consistent with more rapid recycling of smaller bubbles (e.g., Appendix A2), which also contribute to light scattering in the visible. This is opposite to the case at low latitudes on Ganymede, which experiences a low flux of primarily penetrating ions and electrons, so that the energy deposition rate per unit path length in the ice is much lower than at Europa, while thermal annealing is more rapid. This is generalized and depicted in Figure 3, starting clockwise from a neutral ice grain (Costello and Oza, 2026).

Although bubble destruction and migration to grain surfaces can contribute to the atmospheric source rate, the timescales suggested by the variability in the dimer band depth (Spencer et al., 2019) are much longer than the times for the thermal desorption of the returning atmospheric O₂. That is, the modeling (Oza et al., 2019) of the observed dusk/dawn enhancement of the O₂-driven aurora at Europa (Roth et al., 2016), which is roughly consistent with plasma observations (Addison et al., 2022), indicates that the dominant supply of atmospheric O₂ is thermal desorption. These observations have led to the extraction of adsorption energies on ice grains in their regoliths of ∼0.11–0.14 eV (Johnson et al., 2019; Teolis and Waite, 2016), values that we pointed out are roughly consistent with laboratory data (He et al., 2016; Laufer et al., 2017). Finally, we suggested that the previously unexplained plasma-induced O₂ “activation energy”, seen extensively in laboratory studies (e.g., Orlando and Sieger, 2003; Reimann et al., 1984; Teolis et al., 2017) is determined by the surface tension of ice. This is consistent with trapping of O₂ seen at the surface in laboratory samples (Teolis et al., 2009) and the role of interior surfaces in the formation of trapped volatiles. Such aspects must be confirmed by detailed molecular-level modeling in preparation for the observations expected from the Europa Clipper and JUICE missions. Of particular interest is the Moon and Jupiter Imaging Spectrometer on JUICE, which may be able to probe oxidant features in shadow (Poulet et al., 2024). Future space observatories such as Habitable World Observatory might consider a spectrometer capable of studying the molecular oxygen feature described here with a large-aperture telescope. Thus, far space observatories such as James Webb Space Telescope (JWST) are limited to more complex oxidants on putative ocean worlds: Ariel at the Uranian system (Cartwright et al., 2024a), Europa (Trumbo et al., 2023; Villanueva et al., 2023), Ganymede (Bockelée-Morvan et al., 2024), and Callisto (Cartwright et al., 2024b) probed in situ by Galileo/Near Infrared Mapping Spectrometer (NIMS) at Europa (Hansen and McCord, 2008) and Cassini/Visible and Infrared Mapping Spectrometer (VIMS) at Enceladus (Brown et al., 2006; Combe et al., 2019), as well as SO₂ and SiO₂ at exoplanetary systems (Inglis et al., 2024; Oza et al., 2026). Understanding the processes discussed above can also help interpret JWST observations, suggestive of an active source at Europa (Kadoya et al., 2025) and beyond.

7. Summary

Following the early experiments of Brown’s and Lanzerotti’s group (e.g., Reimann et al., 1984), it became clear that the Jovian magnetospheric plasma ejected H₂ and O₂ from the surfaces of the icy Galilean satellites. However, the subsequent observations of the so-called “condensed” O₂ bands seen in reflectance spectra at Europa, Ganymede, and Callisto (e.g., Spencer and Calvin, 2002) remain an enigma. This is the case because the differences in depths and variability of the bands seen on Europa and Ganymede are not simply explained using the available laboratory data.

Expanding on earlier work (e.g., Johnson and Jesser, 1997), we have used our understanding of the radiation-induced processes that form, trap, and desorb O₂ at Europa and Ganymede. That is, incident plasma particles not only lead to the production of voids in which volatiles can be formed and trapped but can also lead to their destruction and to the production of scattering centers that affect the reflected photon sampling depths. Here, we describe how these processes are affected by thermal annealing rates in competition with the rate of plasma-induced damage production. Although we can understand the differences in the “condensed” O₂ observations on these neighboring icy bodies, a quantitative description of their formation in the ice grains is needed beyond the simple model in Appendix A2 and qualitative depiction in Figure 3 (Costello and Oza, 2026). This will require modeling of the molecular physics, as well as experiments involving spectroscopy on thicker ice samples exposed to low rates of penetrating electrons or light ions. A quantitative description is needed, as other volatile products are also observed to be trapped in near-surface ice grains. More importantly, such an understanding could help validate the suggestion that the multiple O₂ formed and trapped in the surficial ice might be a potential source of oxygen delivered geologically to their subsurface oceans, providing a metabolic source for potential biological activity underneath the ice as well as the suggestion of ongoing activity at Europa (Kadoya et al., 2025).

Footnotes

Acknowledgments

The authors thank Shane Carberry-Mogan, Sydney A. Willis, and the two referees for helpful comments. A.V.O. also thanks Rosaly M.C. Lopes for discussions on the importance of oxidants in habitable ocean worlds. A.V.O. expresses gratitude for the original artwork by artist and planetary scientist Emily Costello. This work was partly supported by the NASA Astrobiology Institute (NAI) project “Habitability of Hydrocarbon Worlds: Titan and Beyond” (principal investigator: Rosaly M.C. Lopes).

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Abbreviations Used

Appendix

In Appendix A1, we propose the process that determines the “unexplained” activation energy seen in studies of ejection of O₂ from irradiated ice, and in Appendix A2, we describe the dependence of the steady state void size on the rate of irradiation of ice grains.

Associate Editor: Sherry L. Cady

References

Abellan

, Gautron

, LaVerne

. Radiolysis of thin water ice in electron microscopy. J Phys Chem C 2023;127(31):15336–15345.

Addison

, Haynes

, Stahl AM,

et al. Magnetic signatures of the interaction between Europa and Jupiter's magnetosphere during the Juno flyby. Geophys Res Lett 2024;51(2):e2023GL106810

Addison

, Liuzzo

, Simon

. Effect of the magnetospheric plasma interaction and solar illumination on ion sputtering of Europa’s surface ice. JGR Space Physics 2022;127(2):e2021JA030136.

Bagenal

, Sidrow

, Wilson

, et al. Plasma conditions at Europa’s orbit. Icarus 2015;261:1–13.

Bahr

, Famá

, Vidal

, et al. Radiolysis of water ice in the outer solar system: Sputtering and trapping of radiation products. J Geophys Res 2001;106(E12):33285–33290.

Baragiola

, Bahr

. Laboratory studies of the optical properties and stability of oxygen on Ganymede. J Geophys Res 1998;103(E11):25865–25872.

Baragiola

. Water ice on outer solar system surfaces: Basic properties and radiation effects. Planet Space Sci 2003;51(14–15):953–961.

Behrisch

, Eckstein

. Sputtering by Particle Bombardment, Topics in Applied Physics. Springer: Berlin; 2007.

Bénit

, Brown

. Electronic sputtering of oxygen and water molecules prom thin films of water ice bombarded by MeV Ne+ ions. Nucl Instrum Methods Phys Res B 1990;46(1–4):448–451.

10.

Bockelée-Morvan

, Poch

, Leblanc

, et al. A patchy CO₂ exosphere on Ganymede revealed by the James Webb space telescope. A&A 2024;690:L11.

11.

Brown

, Clark

, Buratti

, et al. Composition and physical properties of Enceladus’ surface. Science 2006;311(5766):1425–1428.

12.

Brown

, Augustyniak

, Simmons

, et al. Erosion and molecule formation in condensed gas films by electronic energy loss of fast ions. Nucl Instrum Methods Phys Res 1982;198(1):1–8.

13.

Calvin

. Absorption coefficients of condensed O₂: Application to Ganymede. Planet Sci J 2025;6(9):216.

14.

Calvin

, Johnson

, Spencer

. O₂ on Ganymede: Spectral characteristics and plasma formation mechanisms. Geophys Res Lett 1996;23(6):673–676.

15.

Cao

. Formation and crystallization of low-density amorphous ice. J Phys D: Appl Phys 2021;54(20):203002.

16.

Carberry Mogan

, Tucker

, Johnson

, et al. Callisto’s atmosphere: First evidence for H₂ and constraints on H₂O. J Geophys Res Planets 2022;127.

17.

Carlson

, Calvin

, Dalton

, et al. Europa’s Surface Composition. University of Arizona Press: Tucson, AZ; 2017.

18.

Cartwright

, Holler

, Villanueva

, et al. JWST reveals CO ice, concentrated CO₂ deposits, and evidence for carbonates potentially sourced from Ariel’s interior. APJL 2024;970(2):L29.

19.

Cartwright

, Villanueva

, Holler

, et al. Revealing Callisto’s carbon-rich surface and CO₂ atmosphere with JWST. Planet Sci J 2024;5(3):60.

20.

Cartwright

, Hibbitts

, Holler

, et al. JWST reveals spectral tracers of recent surface modification on Europa. Planet Sci J 2025;6(5):125.

21.

Cassidy

, Johnson

. Monte Carlo model of sputtering and other ejection processes within a regolith. Icarus 2005;176(2):499–507.

22.

Christianson

, Garrod

. Chemical kinetics simulations of ice chemistry on porous versus non-porous dust grains. Front Astron Space Sci 2021;8; doi: 10.3389/fspas.2021.643297

23.

Chyba

. Energy for microbial life on Europa. Nature 2000;403(6768):381–382.

24.

Combe

J-P

, McCord

, Matson

, et al. Nature, distribution and origin of CO₂ on Enceladus. Icarus 2019;317:491–508.

25.

Cooper

, Johnson

, Mauk

, et al. Energetic ion and electron irradiation of the icy Galilean satellites. Icarus 2001;149(1):133–159.

26.

Cooper

, Johnson

, Quickenden

. Hydrogen peroxide dimers and the production of O₂ in icy satellite surfaces. Icarus 2003;166(2):444–446.

27.

Costello

, Oza

. Radiolytic processes in surface ice grains over time. Zenodo 2026; doi: 10.5281/zenodo.19878688

28.

Dadic

, Light

, Warren

. Migration of air bubbles in ice under a temperature gradient, with application to “snowball earth”. J Geophys Res 2010;115(D18); doi: 10.1029/2010JD014148

29.

Dadic

, Schneebeli

, Wiese

, et al. Temperature-driven bubble migration as proxy for internal bubble pressures and bubble trapping function in ice cores. JGR Atmospheres 2019;124(17–18):10264–10282.

30.

Davis

, Meier

, Cooper

, et al. The contribution of electrons to the sputter-produced O₂ exosphere on Europa. APJL 2021;908(2):L53.

31.

de Kleer

, Milby

, Schmidt

, et al. The optical aurorae of Europa, Ganymede, and Callisto. Planet Sci J 2023;4(2):37.

32.

de Koning

. Crystal imperfections in ice Ih. J Chem Phys 2020;153(11):110902.

33.

Ebert

, Fuselier

, Allegrini

, et al. Evidence for magnetic reconnection at Ganymede’s upstream magnetopause during the PJ34 Juno flyby. Geophys Res Lett 2022;49(23):e2022GL099775.

34.

Eldrup

. Vacancy migration and void formation in γ‐irradiated ice. J Chem Phys 1976;64(12):5283–5290.

35.

Famá

, Loeffler

, Raut

, et al. Radiation-induced amorphization of crystalline ice. Icarus 2010;207(1):314–319.

36.

Gittus

. Irradiation Effects in Crystalline Solids. Applied Science Publishers: London; 1978.

37.

Gusak

, Lutsenko

, Tu

. Ostwald ripening with non-equilibrium vacancies. Acta Mater 2006;54(3):785–791.

38.

Hall

, Strobel

, Feldman

, et al. Detection of an oxygen atmosphere on Jupiter moon Europa. Nature 1995;373(6516):677–681.

39.

Hand

, Chyba

, Carlson

, et al. Clathrate hydrates of oxidants in the ice shell of Europa. Astrobiology 2006;6(3):463–482.

40.

Hansen

, McCord

. Widespread CO₂ and other non-ice compounds on the anti-Jovian and trailing sides of Europa from Galileo/NIMS observations. Geophys Res Lett 2008;35(1) L01202.

41.

, Acharyya

, Vidali

. Sticking of molecules on nonporous amorphous water ice. APJ 2016;823(1):56.

42.

, Fang

, Shoshanim

, et al. Scattering and absorption cross sections of atmospheric gases in the ultraviolet–visible wavelength range (307–725 nm). Atmos Chem Phys 2021;21(19):14927–14940.

43.

Heide

, Zeitler

. The physical behavior of solid water at low temperatures and the embedding of electron microscopical specimens. Ultramicroscopy 1985;16(2):151–160.

44.

Hesse

, Jordan

, Vance

, et al. Downward oxidant transport through Europa’s ice shell by density-driven brine percolation. Geophys Res Lett 2022;49(5).

45.

Inglis

, Batalha

, Lewis

, et al. Quartz clouds in the dayside atmosphere of the quintessential hot Jupiter HD 189733 b. APJL 2024;973(2):L41.

46.

Jiang

, Xin

, Liu

, et al. Phase-field simulation of radiation-induced bubble evolution in recrystallized U–Mo alloy. Nucl Eng Technol 2022;54(1):226–233.

47.

Jin

, Garrod

. Formation of complex organic molecules in cold interstellar environments through nondiffusive grain-surface and ice-mantle chemistry. APJS 2020;249(2):26.

48.

Johnson

, Lanzerotti

, Brown

. Planetary applications of ion induced erosion of condensed-gas frosts. Nucl Instrum Methods Phys Res 1982;198(1):147–157.

49.

Johnson

, Oza

, Leblanc

, et al. The origin and fate of O₂ in Europa’s ice: An atmospheric perspective. Space Sci Rev 2019;215(1):20.

50.

Johnson

, Quickenden

. Photolysis and radiolysis of water ice on outer solar system bodies. J Geophys Res 1997;102(E5):10985–10996.

51.

Johnson

, Quickenden

, Cooper

, et al. The production of oxidants in Europa’s surface. Astrobiology 2003;3(4):823–850.

52.

Johnson

. Energetic Charged-Particle Interactions with Atmospheres and Surfaces. (Physics and Chemistry in Space). Springer: Berlin; 1990.

53.

Johnson

. Polar frost formation on Ganymede. Icarus 1985;62(2):344–347.

54.

Johnson

, Jesser

. O₂/O₃ Microatmospheres in the surface of Ganymede. Astrophys J 1997;480(1):L79–L82.

55.

Kadoya

, Sekine

, Kodama

. Equatorial CO₂ distribution suggests active CO₂ supply on Europa. Planet Sci J 2025;6:102.

56.

Ketcham

, Hobbs

. An experimental determination of the surface energies of ice. Philos Mag (Abingdon) 1969;19(162):1161–1173.

57.

King

, Fletcher

. Global modeling of Ganymede’s surface composition: Near-IR mapping from VLT/SPHERE. JGR Planets 2022;127(12):e2022JE007323.

58.

Landau

, Allin

, Welsh

The absorption spectrum of solid oxygen in the wavelength region from 12,000 Å to 3300 Å. Spectrochimica Acta 1962;18(3):1–19; doi: 10.1016/S0371-1951(62)80106-4

59.

Lanzerotti

, Brown

, Poate

, et al. On the contribution of water products from Galilean satellites to the Jovian magnetosphere. Geophys Res Lett 1978;5(2):155–158.

60.

Lanzerotti

, Maclennan

, Brown

, et al. Implications of voyager data for energetic ion erosion of the icy satellites of Saturn. J Geophys Res 1983;88(A11):8765–8770.

61.

Laufer

, Bar-Nun

, Greenberg

. Trapping mechanism of O₂ in water ice as first measured by Rosetta spacecraft. Mon Not R Astron Soc 2017;469(Suppl_2):S818–S823.

62.

, Gudipati

, Mishra

, et al. Oxidant generation in the ice under electron irradiation: Simulation and application to Europa. Icarus 2022;373:114760.

63.

, Gudipati

, Yung

. The influence of Europa’s plumes on its atmosphere and ionosphere. Icarus 2020;352:113999.

64.

Ligier

, Paranicas

, Carter

, et al. Surface composition and properties of Ganymede: Updates from ground-based observations with the near-infrared imaging spectrometer SINFONI/VLT/ESO. Icarus 2019;333:496–515.

65.

Liuzzo

, Poppe

, Paranicas

, et al. Variability in the energetic electron bombardment of Ganymede. JGR Space Phys 2020;125(9):e2020JA028347.

66.

Mamo

, Raut

, Teolis

, et al. Laboratory investigation of CO₂-driven enhancement of radiolytic H₂O₂ on Europa and other icy moons. Planet Sci J 2025;6(7):170.

67.

Matich

, Bakker

, Lennon

, et al. O₂ luminescence from UV-excited H₂O and D₂O ices. J Phys Chem 1993;97:10539–10553.

68.

McKellar

ARW

, Rich

, Welsh

. Collision-induced vibrational and electronic spectra of gaseous oxygen at low temperatures. Can J Phys 1972;50(1):1–9.

69.

Mergny

, Schmidt

, Keil

. The blinking crystallinity of Europa: A competition between irradiation and thermal alteration. Icarus 2025;441:116700.

70.

Michalsky

, Beauharnois

, Berndt

, et al. O₂-O_s absorption band identification based on optical depth spectra of the visible and near-infrared. Geophys Res Lett 1999;26(11):1581–1584.

71.

Mifsud

, Hailey

, Herczku

, et al. Laboratory experiments on the radiation astrochemistry of water ice phases. Eur Phys J D 2022;76(5):87.

72.

Migliorini

, Kanuchova

, Ioppolo

, et al. On the origin of molecular oxygen on the surface of Ganymede. Icarus 2022;383:115074.

73.

Milby

, Katherine de

, Schmidt

, et al. Short-timescale spatial variability of Ganymede’s optical aurora. Planet Sci J 2024;5(7):153.

74.

Miller

. Clathrate hydrates of air in Antarctic ice. Science 1969;165(3892):489–490.

75.

Noll

, Johnson

, Lane

, et al. Detection of ozone on Ganymede. Science 1996;273(5273):341–343.

76.

Nordheim

, Hand

, Paranicas

. Preservation of potential biosignatures in the shallow subsurface of Europa. Nat Astron 2018;2(8):673–679.

77.

Okada

, Ohkubo

, Maruyama

, et al. Characterization of irradiation-induced novel voids in α-quartz. AIP Advances 2020;10(12):125212.

78.

Orlando

, Sieger

. The role of electron-stimulated production of O₂ from water ice in the radiation processing of outer solar system surfaces. Surf Sci 2003;528(1–3):1–7.

79.

Oza

, Johnson

. Common origin of trapped volatiles in oxidized icy moons and comets. Icarus 2024;411:115944.

80.

Oza

, Johnson

, Leblanc

. Dusk/dawn atmospheric asymmetries on tidally locked satellites: O₂ at Europa. Icarus 2018;305:50–55.

81.

Oza

, Leblanc

, Johnson

, et al. Dusk over dawn O₂ asymmetry in Europa’s near-surface atmosphere. Planet Space Sci 2019;167:23–32.

82.

Oza

, Gebek

, Meyer zu Westram

, et al. Volcanic satellites tidally venting Na, K, and SO₂ in optical and infrared light. Mon Not R Astron Soc 2026;546(1) staf1526.

83.

Pappalardo

, Head

, Greeley

, et al. Geological evidence for solid-state convection in Europa’s ice shell. Nature 1998;391(6665):365–368.

84.

Patterson

, Saltzman

. Diffusivity and solubility of H₂ in ice I_h: Implications for the behavior of H₂ in polar ice. J Geophys Res 2021;126.

85.

Petrenko

, Whitworth

. Physics of Ice. Oxford University Press: Oxford, UK; 2010.

86.

Poppe

A R

, Fatemi

, Khurana

K K.

Thermal and Energetic Ion Dynamics in Ganymede's Magnetosphere. JGR Space Physics 2018;123(6):4614–4637; doi: 10.1029/2018JA025312

87.

Poulet

, Piccioni

, Langevin

, et al. Moons and Jupiter Imaging Spectrometer (MAJIS) on Jupiter Icy Moons Explorer (JUICE). Space Sci Rev 2024;220(3):27.

88.

Ramachandran

, Meka

, Rahul

, et al. Ultraviolet spectrum reveals the presence of ozone on Jupiter’s moon Callisto. Icarus 2024;410:115896.

89.

Reimann

, Boring

, Johnson

, et al. Ion-induced molecular ejection from D₂O ice. Surf Sci 1984;147(1):227–240.

90.

Roth

, Saur

, Retherford

, et al. Europa’s far ultraviolet oxygen aurora from a comprehensive set of HST observations. JGR Space Physics 2016;121(3):2143–2170.

91.

Schaible

, Johnson

, Zhigilei

, et al. High energy electron sintering of icy regoliths: Formation of the PacMan thermal anomalies on the icy Saturnian moons. Icarus 2017;285:211–223.

92.

Schmitt

, Espinasse

, Grim

RJA

, et al. Laboratory studies of cometary ice analogues. In Proceedings of an International Workshop on Physics and Mechanics of Cometary Materials. (ESA SP-302). ( Hunt

, Guyenne

, eds.) European Space Agency: Noordwijk; 1989; pp. 65–69.

93.

Sieger

, Simpson

, Orlando

. Production of O₂ on icy satellites by electronic excitation of low-temperature water ice. Nature 1998;394(6693):554–556.

94.

Spencer

. Icy Galilean satellite reflectance spectra: Less ice on Ganymede and Callisto? Icarus 1987;70(1):99–110.

95.

Spencer

, Calvin

. Condensed O₂ on Europa and Callisto. Astron J 2002;124(6):3400–3403.

96.

Spencer

, Calvin

, Person

. Charge-coupled-device spectra of the Galilean satellites—Molecular-oxygen on Ganymede. J Geophys Res Planets 1995;100:19049–19056.

97.

Spencer

, Grundy

, Schmidt

. Rapid temporal variability of condensed oxygen on Europa? In EPSC-DPS Joint Meeting 2019. EPSC-DPS: Geneva, Switzerland; 2019.

98.

Stephan

, Hibbitts

, Jaumann

. H₂O-ice particle size variations across Ganymede’s and Callisto’s surface. Icarus 2020;337:113440.

99.

Szalay

, Allegrini

, Ebert

, et al. Oxygen production from dissociation of Europa’s water-ice surface. Nat Astron 2024;8(5):567–576.

100.

Tachibana

, Kouchi

, Hama

, et al. Liquid-like behavior of UV-irradiated interstellar ice analog at low temperatures. Sci Adv 2017;3(9):eaao2538.

101.

Teolis

B D

, Loeffler

M J

, Raut U,

et al. Ozone Synthesis on the Icy Satellites. ApJ 2006;644(2):L141–L144; doi: 10.1086/505743

102.

Teolis

, Plainaki

, Cassidy

, et al. Water ice radiolytic O₂, H₂, and H₂O₂ yields for any projectile species, energy, or temperature: A model for icy astrophysical bodies. JGR Planets 2017;122(10):1996–2012.

103.

Teolis

, Shi

, Baragiola

. Formation, trapping, and ejection of radiolytic O₂ from ion-irradiated water ice studied by sputter depth profiling. J Chem Phys 2009;130(13):134704.

104.

Teolis

, Waite

. Dione and Rhea seasonal exospheres revealed by Cassini CAPS and INMS. Icarus 2016;272:277–289.

105.

Tinner

, Galli

, Bär

, et al. Electron-induced radiolysis of water ice and the buildup of oxygen. JGR Planets 2024;129(12):e2024JE008393.

106.

Trumbo

, Brown

, Bockelée-Morvan

, et al. Hydrogen peroxide at the poles of Ganymede. Sci Adv 2023;9(29):eadg3724.

107.

Trumbo

, Brown

, Hand

. H₂O₂ within chaos terrain on Europa’s leading hemisphere. AJ 2019;158(3):127.

108.

Trumbo

, Brown

, Adams

. The geographic distribution of dense-phase O₂ on Ganymede. Planet Sci J 2021;2(4):139.

109.

Vidal

, Bahr

, Baragiola

, et al. Oxygen on Ganymede: Laboratory studies. Science 1997;276(5320):1839–1842.

110.

Villanueva

, Hammel

, Milam

, et al. Endogenous CO₂ ice mixture on the surface of Europa and no detection of plume activity. Science 2023;381(6664):1305–1308.

111.

Vook

, Birnbaum

, Blewitt

, et al. Report to American-physical-society by study-group on physics problems relating to energy technologies—Radiation effects on materials. Rev Mod Phys 1975;47(S3):S1–S44.

112.

Waite

, Greathouse

, Mogan

, et al. Magnetospheric-ionospheric-atmospheric implications from the Juno flyby of Ganymede. JGR Planets 2024;129(4):e2023JE007859.

113.

Warren

. Optical properties of ice and snow. Philos Trans A Math Phys Eng Sci 2019;377(2146):20180161.

114.

Wiedersich

. On the theory of void formation during irradiation. Radiat Eff 1972;12(1–2):111–125.

115.

, Trumbo

, Brown

, et al. Europa’s H₂O₂: Temperature insensitivity and a correlation with CO₂. Planet Sci J 2024;5(10):220.

Plasma and Thermal Processing Leading to Latitudinal and Temporal Variability of the Trapped O 2 at Europa and Ganymede

Abstract

Keywords

1. Introduction

3. O2 Bubble Formation: Satellite Versus Laboratory Surfaces

7. Summary

Footnotes

Acknowledgments

Author Disclosure Statement

Funding Information

Abbreviations Used

Appendix

References

3. O₂ Bubble Formation: Satellite Versus Laboratory Surfaces