Energetic proton losses reveal Io’s extended and longitudinally asymmetrical atmosphere

The SO₂ frost sublimation (e.g. \citeAIngersoll1985,Moreno1991,Doute2001), volcanic plumes (e.g. \citeAMorabito1979,Spencer2000,Walker2010), atmospheric sputtering (e.g. \citeAMcGrath1987MagnetosphericAtmosphere,Pospieszalska1996) and physical chemistry [Dols \BOthers. (\APACyear2008), Dols \BOthers. (\APACyear2012)] contribute to the formation of an extended atmosphere of SO₂, S and O around Io. Minor species are also present such as SO [Lellouch \BOthers. (\APACyear1996)] and NaCl [Lellouch \BOthers. (\APACyear2003)]. The atmosphere extends into Hill’s sphere ($\sim$6 Io radii R_Io, 1821 km) as a corona, which is mainly composed of S and O [Wolven \BOthers. (\APACyear2001)] and is thought to be produced by atmospheric sputtering [McGrath \BBA Johnson (\APACyear1987)] and physical chemistry processes [Dols \BOthers. (\APACyear2008)].

The atmosphere of Io is thought to be asymmetrical along different axes (e.g. \citeABlocker2018MHDAtmosphere). Being sublimation dominated, the dayside is more dense than the nightside (e.g. \citeATsang2016TheEclipse). It is thought that the atmosphere could partially collapse on the nightside due to the condensation of atmospheric SO₂ in the cold night temperature. The atmospheric density at the poles is thought to be only a few percent of the atmosphere at the equator, due to the equatorial distribution of frost, the preferential heating of the sub solar point and the location of some main volcanoes closer to the equator (e.g. \citeAStrobel2001TheHydrogen,Spencer2005Mid-infraredAtmosphere). The anti‐Jovian atmosphere is denser and latitudinally more extended than the sub‐Jovian side (e.g. \citeAFeaga2009IosAtmosphere,Spencer2005Mid-infraredAtmosphere,Moullet2010SimultaneousArray). Lastly the atmospheric density varies longitudinally with a denser upstream than downstream atmosphere (e.g. \citeASaur2002InterpretationPasses). Major uncertainties remain in our understanding of the radial profile, longitudinal asymmetries, minor species, and how much the atmosphere collapses at night or in eclipse.

Some of the neutrals escape Io’s gravity field to form giant neutral clouds of S and O (several Jupiter radii in size, 71492 km) that extend along Io’s orbit. The neutral clouds at Io’s orbital velocity are continuously passed by the plasma flow of the Io torus at a relative velocity $\sim$57km/s [Kivelson \BOthers. (\APACyear2004)]. They experience charge exchange and electron-impact ionization, which supply the torus with fresh plasma. The flux of plasma directly from Io is about $\sim 300$ kg/s (e.g. \citeASaur2003TheIo,Dols2008ATorus). The plasma torus has an average electron density $\sim$2000 cm^-3 and an average thermal temperature $\sim$60-100 eV [Kivelson \BOthers. (\APACyear2004), Bagenal \BBA Dols (\APACyear2020)]. Centrifugal forces lead the plasma to flow radially outwards along Jupiter’s magnetic equator thereby forming a plasma sheet [Bagenal (\APACyear1989)].

While circulating through the magnetosphere, charged particles are accelerated to suprathermal energies [Paranicas \BOthers. (\APACyear2000)] and are able to diffuse back radially inwards [Khurana \BOthers. (\APACyear2004)] towards Jupiter’s moons. These particles are referred to as energetic particles in the rest of the text. In summary, Io is exposed to a flow of corotating low energy plasma ($\sim$100 eV) and to a population of energetic charged particles, including the energetic protons ($>100$ keV) on which this study focuses.

The Energetic Particle Detector (EPD) onboard Galileo [Williams \BOthers. (\APACyear1992)] detected dropouts of energetic ions during close flybys of Io. In this work we investigate specifically the cause of energetic proton losses. Several causes of energetic ion depletion have previously been proposed for various moons: impact of the energetic protons on the moon’s surface, deflection of the protons by inhomogeneous electromagnetic fields, charge exchange of the protons with atmospheric neutrals and wave particle interactions.

Energetic ions that impact the surface are absorbed and therefore result in a depletion, such losses have been widely reported at Io [Selesnick \BBA Cohen (\APACyear2009)] and various moons of Jupiter and Saturn e.g. \citeAParanicas2000EnergeticEuropa,Kivelson2004MagnetosphericSatellites,Roussos2022ABelts,Kotova2015ModelingDione,Krupp2020Magnetospheric20052015. Furthermore, a magnetic field gradient can restrict access of energetic ions in certain regions and cause apparent losses. Such features have been reported at Io [Selesnick \BBA Cohen (\APACyear2009)] and other moons of Jupiter and Saturn e.g. \citeAParanicas2000EnergeticEuropa,Roussos2012EnergeticInteraction,Huybrighs2023. We will refer to losses due to perturbed fields as “forbidden regions”.

When energetic protons charge exchange with atmospheric neutrals they turn into energetic neutral atoms (ENA) which will result in a depletion of the energetic protons. Atmospheric charge exchange has been investigated as a loss process of energetic ions at moons of Jupiter and Saturn e.g. \citeAMitchell2005EnergeticMagnetosphere,Wulms2010EnergeticFields,Kotova2015ModelingDione,Addison2021InfluenceWeathering. Specifically at Europa \citeAHuybrighs2020AnDepletions demonstrated, using a Monte Carlo particle tracing code, that proton depletions (115-244 keV) measured during the E26 Galileo flyby were affected, in addition to absorption by the surface, by atmospheric charge exchange and inhomogeneous fields. This study builds on the modelling efforts of \citeAHuybrighs2020AnDepletions.

Specifically at Io, ion dropouts during three close Galileo flybys (I27, I31 and I32) have been analysed by \citeASelesnick2009ChargeIo. Using ion back-tracing simulations they demonstrated that losses of energetic oxygen and sulfur ions ($>$5 MeV/nucleon) measured by Galileo’s Heavy Ion Counter (HIC) could be attributed to absorption by the surface and the effect Io’s Alfvén wings. \citeASelesnick2009ChargeIo accounted for the Alfvén wings by using an idealized analytical description, but did not account for other perturbations such as magnetic field pile-up. The addition of the Alfvén wings provides a better reproduction of the measured ion dropouts during polar flybys I31 and I32. At certain segments of the flyby, the Alfvén wings deflect ions that would otherwise collide with Io’s surface, while enhancing losses in other segments. For flyby I27 the effect of the Alfvén wing appeared to be small.

Ion dropouts have also been reported away from Io, but near the Io orbit, using Voyager 1, Voyager 2 and Galileo data [Armstrong \BOthers. (\APACyear1981), Cheng \BOthers. (\APACyear1983\APACexlab\BCnt2), Williams \BOthers. (\APACyear1996), Mauk \BOthers. (\APACyear1998), Lagg \BOthers. (\APACyear1998)]. Specifically, \citeACheng1983EnergeticOrbit and \citeALagg1998DeterminationMeasurements proposed charge exchange with neutrals as a loss process for the energetic ions, More recently, \citeAMauk2022 argued, based on [Smith \BOthers. (\APACyear2019)], that near Io’s orbit charge exchange with low energy ions would dominate over charge exchange with neutrals. \citeAMauk2022 also argues that near Io’s orbit charge exchange losses dominate for protons (roughly below 200 keV) over losses from scattering. Furthermore, they suggest that at energies above 300 keV the scattering dominates over charge exchange.

Lastly, ion dropouts have also been measured along field lines connected to Io, but not in a close flyby of Io. The JEDI instrument on Juno measured losses of energetic protons with energies in the 10-800 keV range over a region larger than Io along field lines connected with Io [Paranicas \BOthers. (\APACyear2019)]. They attribute these losses to wave-particle interaction, building on \citeANenon2018AOrbit which proposes that near Io ion cyclotron waves could resonate with energetic protons and cause proton losses. \citeANenon2018AOrbit suggest that protons at energies between about 0.3 and 4 MeV are affected by electromagnetic ion cyclotron waves near Io’s L shells. However, \citeAParanicas2019IosData also suggests that below 100 keV, proton losses to charge exchange may become significant due to the large charge exchange cross section.

In this study we investigate the cause of energetic proton losses measured close to Io ($<2R_{Io}$) by the Energetic Particle Detector (EPD) on the Galileo spacecraft. While the effect of charge exchange has been considered as a loss process for Io’s orbital environment, previous studies investigating particle losses near Io [Selesnick \BBA Cohen (\APACyear2009)] have not yet considered the effect of charge exchange with the atmosphere, and they employed a highly simplified model of the perturbed fields. Therefore, we opt to include charge exchange with Io’s asymmetrical atmosphere as a loss process as well as a detailed MHD model of the perturbed fields. To investigate the cause of the losses we employ a particle back-tracing model which we use to reproduce the flux of energetic protons as measured by Galileo during three Io flybys (I24, I27 and I31).

We consider this study in the first place as a feasibility study to demonstrate the sensitivity of the proton dropouts to the atmospheric structure and magnetic field configuration. Using previously validated models of fields and atmosphere we will show that the models are not able to fully reproduce the energetic proton dropouts and that the proton data can thus provide new observational constraints to modelers. Our aim is to provide hypotheses and directions to encourage further modelling studies that exploit the additional constraint offered by the energetic ions to the fullest.

In this work, we will focus on two models of the SO₂ distribution around Io: one radially symmetrical model (referred to as the ’symmetrical atmosphere’ in the rest of the text) and one asymmetrical model that takes into account longitudinal and latitudinal asymmetries (referred to as ’asymmetrical atmosphere’ in the rest of the text), leaving to future work a more comprehensive study of atmospheric and field perturbation models. A comparison between the two models will reveal the sensitivity of the proton losses to different atmospheric configurations. Profiles of the two atmospheric models for the different flybys are plotted in Figure 2. The global distribution of the asymmetrical atmosphere is visualised for the three flybys in Figure 3.

In our simulations we will consider both homogeneous and inhomogeneous electromagnetic fields. In the homogeneous electromagnetic fields, no perturbations due to the interaction of Io and the plasma that is corotating with Jupiter will occur. In the inhomogeneous fields the perturbations are simulated using an MHD approach. By simulating the particle flux for both cases we can determine the contribution of the perturbed fields to the measured flux, e.g. by determining if forbidden fields cause any of the measured proton losses. We will demonstrate in the next section that the inhomogeneous fields (as opposed to the homogeneous fields) can have a strong influence on the energetic charged particle depletion.

The homogeneous magnetic field B is obtained from the Galileo MAG data from Io and does not take into account the tilt of the background field. The electric field E is calculated using the relation $\textbf{E}=-\textbf{v}\times\textbf{B}$, in which v is the velocity of the corotational plasma. The values for B and v in the homogeneous case are the same as the initial values used for the inhomogeneous case (see Table 3).

The inhomogeneous magnetic field is obtained using the MHD simulations of \citeABlocker2018MHDAtmosphere, consistent with the asymmetrical atmospheric description proposed in Section 3.1. The unique feature of the \citeABlocker2018MHDAtmosphere MHD simulations is the absence of a prescribed magnetic field induced in Io’s putative magma ocean, as initially proposed by \citeAKhurana2011EvidenceInterior. \citeABlocker2018MHDAtmosphere propose a convincing match to the Galileo Magnetometer data with the simple assumption of an asymmetrical atmosphere and no induced dipole. Quoting their paper, they claim that “we find that the measured perturbations can be primarily caused by the plasma interaction with the longitudinally asymmetric atmosphere. This implies that a significant magnetic induction signal from a partially molten magma ocean is not necessarily required to explain the Galileo magnetometer data.” These simulated magnetic perturbations, consistent with the Galileo observations, are used in our study of the proton losses. For further details on the MHD simulation we refer the reader to the paper. We will demonstrate in the next section that the inhomogeneous fields (as opposed to the homogeneous fields) can have a strong influence on the energetic charged particle depletion.

The MHD simulations are used to self-consistently model the plasma interaction of Io’s atmosphere with the corotating plasma from Jupiter’s magnetosphere. The applied MHD model has already been used to study Io’s plasma interaction by \citeABlocker2018MHDAtmosphere. The authors adjusted the model to Io’s plasma interaction from the model which was developed for Ganymede’s plasma interaction by \citeADuling2014. A detailed description of the model is given in \citeABlocker2018MHDAtmosphere. The model solves a set of three-dimensional equations consisting of the continuity equation, the momentum equation, the induction equation, and an equation for the internal energy on a spherical grid. It takes into account collisions between ions and neutrals, plasma production and loss due to electron impact ionization and dissociative recombination, and the ionospheric Hall effect. Io’s atmosphere is implemented as a globally asymmetric static SO₂ neutral atmosphere (see Equation 3) which is described in detail in Section 3.1. The atmosphere description is introduced in the MHD model in the terms of ion-neutral collisions, electron impact ionization, and the ionospheric Hall effect. An overview of the parameter settings for the initial and boundary conditions in this model is given in Table 3.
Figure 4 shows the magnetic perturbations around Io for each Galileo flyby addressed here. In the three panels we see a strong increase of the magnetic field magnitude upstream of the moon due to the pile-up of the magnetic field lines. The formation of the Alfvén wings is clearly visible in the reduced plasma flow velocity (arrows in third plot) north and south of the moon. They are bent back with respect to the unperturbed background magnetic field. The plots in the equatorial plane (I24 and I27) show that the plasma flow is diverted around the moon and accelerated at the flanks. Most of the plasma flow into the ionosphere is reduced and is swept around the moon and the wings. Upstream of the moon the plasma flow velocity is decreased and downstream of the moon it increases again to the background value due to the Lorentz force. The asymmetry of the plasma flow and the magnetic field between the Jupiter-facing side and anti-Jupiter facing side shown in the first two plots is due to the plasma interaction with an asymmetric atmosphere. The comparison between the plasma flow between the I24 and I27 case shows that the plasma flow upstream of the moon is slowed down more effectively for the I27 case than for the I24 case. This is an effect of the asymmetric atmosphere (see Figure 3). The Galileo flyby I31 provided a close pass of the north polar region as presented in the trajectory in the third panel of Figure 4. The spacecraft crossed the northern Alfvén wing. Due to its close flyby to Io the electromagnetic perturbations are not only affected by the Alfvénic interaction but also by the ionospheric current system.

In \citeABlocker2018MHDAtmosphere a comparison between the MHD model and the measured components of the magnetic field are shown in Figure 7 for I24 (atm. Model 2) and Figure 10 for I27. Atm. model 2 is not shown in the plots for I27, but the results are similar as to atm. Model 3. For those flybys a comparison is also made to electron density derived from Galileo PWS and plasma velocity components from Galileo PLS (PWS data is shown in Figures 9 and 12). In D we provide the MHD simulations for I31, and a comparison with the magnetic field data which were not provided in \citeABlocker2018MHDAtmosphere.

We simulate the measured flux of energetic protons using a Monte Carlo particle back-tracing code, developed from \citeAHuybrighs2017OnMission,Huybrighs2020AnDepletions,Huybrighs2023. Particle tracing has been employed previously at various Galilean moons to investigate the ion dynamics as affected by field perturbations (e.g. \citeASelesnick2009ChargeIo,Poppe2018ThermalMagnetosphere,Liuzzo2019EnergeticCallistob,Breer2019EnergeticEuropa,Carnielli2019FirstIonosphere,Plainaki2020,Addison2022EffectIce,Nordheim2022MagnetosphericSurface,Huybrighs2023) and atmospheric charge exchange (e.g. \citeAHuybrighs2020AnDepletions,Addison2021InfluenceWeathering).

Because the energetic protons have no influence on the electromagnetic fields they can be treated as test particles. In the simulation protons are traced back in time from the EPD detector. For each particle it is determined if it impacts and/or charge exchanges. If, at any point along the particle trajectory, a surface impact ($r\leq R_{\text{Io}}$) or charge exchange with the atmosphere occurs, the particle is considered lost. This analysis is performed for many particles to determine the depletion as a fraction of the (normalized) undisturbed flux. All the steps of this process, from initiating the particle to the possible loss, are described in the following paragraphs. The relevant input parameters are summarized in Table LABEL:tab_particle.

We take into account the energy distribution of the proton flux in the simulation. Particles are initiated at a random energy within the energy range of the TP1 or TP3 channel, using a Monte Carlo approach. Once the particles are initiated they are binned according to their initial energy (10 bins) and pitch angle (64 bins). The energetic proton flux decreases with increasing proton energy, though less than an order of magnitude over both the TP1 or TP3 channels [Paranicas \BOthers. (\APACyear2003)]. We account for this decrease in flux with increasing energy by weighting the particles according to the energy distribution of the flux. Since we found in \citeAHuybrighs2020AnDepletions that the simulation is not very sensitive to the specific drop off profile in one energy channel, we have used the same relative change in flux ($\sim$40 percent decrease in flux from the beginning to the end of the energy channel) as used in \citeAHuybrighs2020AnDepletions,Huybrighs2021ReplyDepletions,Huybrighs2023 for Europa in this study.

We use a leap frog numerical integration scheme of the Lorentz force to trace the protons back in time [Huybrighs \BOthers. (\APACyear2017), Huybrighs \BOthers. (\APACyear2020), Huybrighs \BOthers. (\APACyear2021), Huybrighs \BOthers. (\APACyear2023)]. We simulate the trajectories of energetic protons under a homogeneous and an inhomogeneous electromagnetic field (the latter is discussed in the next section). In the homogeneous case we neglect any perturbations in the electromagnetic field resulting from the interaction of the corotational plasma with Io. Despite that the situation with homogeneous fields is not expected to occur at Io in reality, we consider including this case relevant so that we can determine relative the contribution of the inhomogeneous fields to the proton losses.

We neglect the pitch angle distribution of the protons, thus our pitch angle distribution is uniform. The consequence of this simplification is that our simulations will not be able to reproduce losses that are the result from variations in the background pitch angle distribution. However, we consider this approach suitable because our simulations will show that the deviations from the background distribution are much more pronounced features than the measured variations in the data upstream of Io.

In the simulation back-traced particles are assumed to contribute to a flux decrease, if they impact on Io’s surface or charge exchange with neutrals from its atmosphere. This is because when a back-traced particle that impacts Io or charge exchanges, is traced forward it would have to originate from the atmosphere or the surface, and neither of those are sources of energetic protons. It is then assumed that the corresponding particle does not exist and should thus count towards a loss of flux. Perturbed fields can form ’forbidden regions’ which will lead simulated particles to impact on Io’s surface, resulting in an apparent loss as measured by the instrument. In reality the particle has effectively been deflected away from the ’forbidden region’. In the simulation a particle is considered to have impacted Io’s surface when its trajectory intersects the moon’s surface. If the energetic proton charge exchanges, it is assumed to turn into an energetic neutral atom (ENA). ENAs are not bound by the electromagnetic fields, and will escape from the system at a high velocity, thereby contributing to the proton loss. The probability of charge exchange with the atmospheric neutrals is determined using Equation (7) from \citeABirdsall1991:

We neglect the following loss processes: charge exchange with the low energy ions, energy loss due to Coulomb collisions with ions and electrons, and wave particle interaction. \citeAMauk2022 suggest that charge exchange with the low energy ions could dominate over charge exchange with neutrals near Io’s orbit. However, in this study we are dealing with the environment very close to Io ($\sim 2$R_Io) where the neutral density of the atmosphere is orders of magnitude larger than the ion density ($\sim$2000 cm^-3; \citeABagenal2020TheEuropa). Furthermore, the charge exchange cross section of the energetic protons on the low energy ions is two orders of magnitude smaller than the cross section of energetic protons and the neutrals (see Table 9 in [Mauk \BOthers. (\APACyear2022)]). We therefore expect charge exchange with the low energy plasma to be only a minor effect compared to charge exchange with the neutral environment close to Io.

We also assume that the effect of Coulomb collisions as a loss process is negligible. Coulomb collisions could result in energy losses of energetic protons, which would result in apparent losses of the protons. However, \citeANenon2018AOrbit has shown that for energies of 0.1 MeV the loss rates due to charge exchange with neutrals of the Io and Europa torus are orders of magnitude larger than the losses due to Coulomb collisions. They specifically modelled Coulomb collisions with bound electrons of the neutral particles of the Jovian hydrogen corona and Io and Europa gas torus, with bound electrons of the cold plasma ions and with free electrons of the cold plasma of the inner magnetosphere of Jupiter. We also neglect Coulomb collisions as a loss process for the TP3 channel, if Coulomb losses with the particle environment (which at the distances of this work is dominated by the atmospheric density) were a major loss process in this channel for the data segments we considered, then we would expect to see a correlation between the density of the atmosphere and the dropout, which is not what we observe for TP3.

We also neglect wave particle interaction as a loss process that plays a major role in the proton loss features we observe within $\sim 2$R_Io. We build on \citeAMauk2022 which argues that near Io’s orbit charge exchange losses dominate for protons (roughly below 200 keV). As we will demonstrate later, the depletion features at 540-1250 keV can be explained by the field configuration, indicating that also at these higher energies wave particle interactions do not play a major role in proton depletion within $\sim 2$R_Io.

We ignore the possibility of protons re-entering the simulation box after exiting, because the bounce periods are larger than the time it takes the corotation to cover the spatial scales of the relevant proton loss features. Therefore, we consider it unlikely that losses occurring closer to upstream would create the measured losses further downstream by modifying the proton distributions. See C for a more detailed discussion.

Considering the complexity of the processes leading to proton depletion in Io’s atmosphere (electric and magnetic field perturbation, neutral density etc.) the atmospheric density cannot be directly inferred from the measurements. Thus, we analyze the data with a forward modeling approach, prescribing an atmospheric distribution and electric and magnetic field perturbations around Io to compute the proton depletion and compare it to the Galileo data.

In Figure 5 we show the measured energetic proton flux (solid orange line) for the three flybys in the TP1 and the TP3 channel. The vertical red dashed line indicates the point of closest approach for each flyby and the grey dashed line indicates the radial distance to Io. The flux in these plots has been normalized to the mean of two cycles of the instrument from the start of the flyby segment. This allows for comparison with the particle tracing code which expresses depletion numbers in percentages. The periodic variation of the flux measurements is due to the rotation and scanning motion of the spacecraft (as mentioned above, this takes about 140 s).

We observe that the region over which depletions occur along the spacecraft trajectory is larger in the TP1 channel than the TP3 channel for I27 and I31. Furthermore, for flyby I24 no depletion is observed in the TP3 channel (b) at all. This decrease in size or disappearance of the depletion region in TP3 is suggestive of charge exchange playing a major role in the depletion because the charge exchange cross section decreases significantly from the TP1 to the TP3 channel (see Figure 12). For I24 in TP1 the depletion region is symmetrical with the distance to Io, while in I27 and I31 the symmetry breaks down. During both I27 and I31 (TP1) the depletion region is more extended on the day/downstream side (right side in panels a c and e). For TP3 I24 does not show a trend of depletion. The extent of the depletion region in I27 and I31 (TP3) is roughly symmetrical with respect to the distance to Io. However, during I31 there is a complicated structure, with a near complete depletion occurring after the closest approach from 05:02 to 05:04 that does not occur post closest approach.

Note that we do not attribute the measured depletions to the gyroradius effect. That is, the effect where particles impact on Io’s surface because they encounter it during their gyromotion along the field line. Figure 13 shows that the gyroradius of the respective protons is smaller than the closest approach altitude, even during the closer flyby I27. Furthermore, the gyroradii shown in Figure 13 are for the upstream conditions, and represent a lower estimate. Near Io the magnitude of the field will increase due to magnetic pile up, therefore the gyroradius will be even smaller near Io.

An artifact that introduces nonphysical dropouts can occur in the EPD data, when measurements are taken on motor position 7 [Jia \BOthers. (\APACyear2021), Huybrighs \BOthers. (\APACyear2021)]. For flybys I24 and I27 this artifact does not occur because the EPD was kept in a fixed motor position, which is different from motor 7. For the data segment of I24 and I27, considered in this study, the motor position is fixed to position number 4. During I31 motor position 7 is one of the motor positions that is scanned, however we find that the EPD data gathered, when the motor is in position 7, does not appear to fall outside of the general trend. We conclude that the data artifact will not affect the conclusions of this study, because we focus on the general trend and structure of the depletion rather than depletion features measured at single motor positions.

In Figure 6 the results are shown for the back-tracing simulations for the TP1 energy channel (155-224 keV). In the plots on the left of the figure (a, c, and e) the results are shown for the homogeneous fields, while the right side (b, d, and f) shows the results for the inhomogeneous fields. The first row (a and b) shows the results for I24, the second row (c and d) for I27, and the third row (e and f) for I31. In each plot, the EPD measurement is shown in orange. The depletion predicted while assuming no atmospheric interactions is shown in black, and the simulations considering charge exchange with the atmosphere are shown in dark blue for the symmetrical atmosphere and light blue for the asymmetrical atmosphere. In Figure 7 simulation results for the TP3 energy channel (540-1250 keV) are shown. The same colors and labels are used as in Figure 6.

We will argue that at 155-224 keV charge exchange is the dominant loss process (I24) or a main loss process combined with surface absorption (I27 and I31). The measured losses are also sensitive to the description of the electromagnetic field (I27 and I31). Due to the collision cross section decreasing exponentially with energy, the likelihood of charge exchange occurring with high energy particles is smaller than for lower energy particles. Therefore the 540-1250 keV (TP3) simulation results do not show any significant difference when atmospheric charge exchange in included, all the solutions are overlapping each other (black, dark blue en light blue line). We will argue that 540-1250 keV protons during I24 are not affected by Io and that inhomogeneous electromagnetic fields are responsible for the proton depletion observed during the two closer flybys I27 and I31. Due to I31’s polar trajectory surface absorption is also an important loss process during that flyby.

As described in the previous sections a periodic discrepancy occurs between the simulated losses and the measured loss at 155-224 keV (TP1). While the simulated flux periodically returns to the unperturbed level, the data does not. We attribute this periodicity to the approximately 20 second spin of the Galileo spacecraft, which makes EPD scan different directions. Meaning that in certain directions the losses are predicted better than in others. We further interpret this periodic discrepancy as an indication that our atmospheric model can be improved in the directions where we underpredict the loss, in particular by increasing the scale height of the atmosphere.

Firstly, we point out that the periodic matching and mismatching of the data by the model does not occur at 540-1250 keV (TP3) and is therefore an energy dependent effect. In fact, the overall agreement between the simulation with perturbed fields and the data is notably better at 540-1250 keV (where charge exchange is negligible) than for the simulations at lower energies (155-224 keV) without charge exchange. Considering that atmospheric charge exchange is more likely to occur in the lower energy range than the higher energy range. The fact that the model more accurately reproduces the data in the higher energy regime compared to the lower energy regime suggests that the model is not sufficiently capturing the entirety of the charge exchange occurring at Io. This implies that if the atmospheric model is improved, charge exchange at lower energies would be enhanced and the depletion features at 155-224keV will be better reproduced.

Next we will discuss the discrepancy for flyby I24 as an example. Because charge exchange is the dominant loss process during this flyby, the interpretation of the discrepancy is simpler. During the entire duration of I24 a periodic matching and mismatching of the depth of the depletion associated with the asymmetrical model occurs. In Figure 9 we show particle trajectories for four points along the trajectory where the depletion reaches approximately the depth of the data (pink trajectory), and where it does not (green trajectory). The trajectories corresponding to points where the depth of depletion is underestimated have a smaller gyroradius, or equivalently, higher speed parallel to the magnetic field. The trajectories corresponding to those points where the depth of the depletion is approximately attained, have larger gyroradius, or equivalently lower speed parallel to the magnetic field.

The periodicity in the simulation (light blue line, Figure 6, panel a) has three dips within 20 seconds (the spin of the spacecraft). This is because during a full rotation of 20 seconds the detector viewing direction changes from perpendicular to the field line (larger gyroradius, large dip #1), to parallel (smaller gyroradius, no or small dip), perpendicular again but in opposite direction (larger gyroradius, large dip #2), parallel again but in opposite direction (smaller gyroradius, no or small dip) and back to the initial viewing angle (larger gyroradius, equivalent to large dip #1).

Next we discuss flyby I27, during which a different pattern of agreement between the data and the model than for I24 occurs. At the extremities of the I27 the depletion pattern is similar to I24: i.e. three approximately equal dips are visible every 20 seconds, with an underprediction of the measured depletion in between the dips (e.g. near the pink and green line on the right side of panel c in Figure 6). However, near the closest approach the pattern of depletion becomes different. Within the 20 second spin we see an underprediction of the depletion (green line on the left of the closest approach), a deep depletion that matches the data (purple line) and an undeprediction again. Thus, compared to I24, or the extremities of I27, we see only one deep dip per spin. In fact, what is different is that in the closer part of I27 within each 20 second cycle there is one of the two more field aligned directions (e.g. purple line left of the closest approach in in Figure 6, panel c) along which the depletion is now also approximately reproduced. Combined with the directions perpendicular to the field that were already depleted (previously referred to as deep dips #1 and #2) this depleted and field aligned direction forms the single deep dip we observe during one spin. The other field aligned direction (e.g. green line left of the closest approach, Figure 6 left column, panel c) still underpredicts the loss. In I24, or at the extremities of I27 (e.g. purple and green line right of the closest approach in Figure 6, panel c), both of the field aligned directions produce the losses about equally poorly. Corresponding particle trajectories are shown in Figure 10. The field aligned direction which is depleted in the closer part of I27 corresponds to the direction southward of the trajectory. Considering that the I27 flyby is located north of Io’s equator, particles going in these directions appear to have a higher chance of charge exchanging, even when they are more field aligned and have smaller gyroradii. We argue that this is because they traverse a larger section of the low altitude and therefore denser atmosphere, which results in a higher chance of charge exchange.

The underprediction in the northward field aligned directions appears to hold even in the case with the inhomogeneous fields (Figure 6, panel d). In the northward directions (e.g. closest to the green line on the left of panel c) the effect of the field configuration seems negligible. However, the southward field aligned directions (e.g. closest to the pink line on the left of panel c) is fully depleted in panel d. This means that we likely cannot derive any direct information about the atmospheric losses in those directions, since they are already fully saturated even without atmospheric charge exchange.

Lastly, we discuss the periodic discrepancy for flyby I31. In Figure 11 we show particle trajectories for four points along the I31 trajectory where the depletion reaches approximately the depth of the data (pink trajectory), and one where it does not (green trajectory). First considering the two trajectories in the left column of Figure 11. The direction in which the simulation matches the depth of depletion (pink) corresponds to a direction that intersects with the surface. As discussed in the previous section such a particle is considered lost. The green trajectory is essentially moving in the opposite direction and does not intersect the surface, implying that the loss process associated with this direction has to be different from surface impact. In the right panel the pink trajectory (depth of depletion reproduced) does not intersect the surface, rather the cause of the depletion is charge exchange. As for I24 and I27 it appears that both atmospheric models underestimate the losses in the more field aligned directions that do not encounter the denser part of the atmosphere. The conclusions for the ’green’ particles hold even when including perturbed fields, at those points the depletion seems unaffected by the perturbed fields (black line).

However, we also observe that during the two minutes around the closest approach the cyclical mismatch disappears in the case with the inhomogeneous fields. This suggest that at least part of the mismatch could be attributed to a mismatch between the simulated field and the real configuration. Nonetheless, the simulations suggest that the role of the electromagnetic field in producing the losses only matters in a segment along the trajectory that is more confined than where atmospheric charge exchange plays a role. Suggesting that both charge exchange and field perturbations are required to reproduce the observed loss.

In conclusion, we find that the periodic discrepancy can be attributed to the properties of the atmospheric model. Directions in which the depletion is underpredicted correspond to ions that have small gyroradii and are travelling away from Io or the denser part of its atmosphere. Losses along these directions thus contain more information about possible charge exchange at larger distances from Io. We will discuss the implications of this finding further in Section 5.4.

In this section we argue that atmospheric charge exchange is a necessary factor in order to explain the energetic proton depletion seen in the TP1 data during the Galileo flybys of Io. In fact, we suggest that near Io atmospheric charge exchange is the dominant loss process for flybys of 600 km altitude, while in closer flybys of 200km the electromagnetic field configuration and surface absorption also play a role in the measured proton depletion. In this sense the environment closest to Io is different than the environment along Io’s orbit, but not close to Io, wherein charge exchange with the low energy plasma is thought to be the dominant loss process [Mauk \BOthers. (\APACyear2022)]. Table LABEL:tab_summary summarises the effects contributing to the proton loss and the derived atmospheric properties.

As discussed in Section 4.1, at higher energies (540-1250 keV, TP3) the depletion region decreases in size along the flyby (I27 and I31) or disappears altogether (I24). This energy dependent size of the depletion region itself is already a strong indication for a role of charge exchange with the atmosphere as a loss process. Furthermore, the simulation results discussed in the previous section indicate that the inclusion of atmospheric charge exchange leads to deeper energetic proton depletion for all flybys in all cases, including those with homogeneous electromagnetic fields, with perturbed fields and for both atmospheric models, in the energy range of the TP1 channel of the EPD (155-224 keV), thereby further supporting our hypothesis regarding the role of charge exchange.

The effect of charge exchange is most apparent for flyby I24, for which we argue that charge exchange is the dominant loss process of energetic protons. Our simulations can reproduce the extent of the depletion region, as well as attain the depth of observed depletion, while neither perturbed fields or impact of the surface play a significant role during this flyby. Furthermore, the lack of any depletion in the TP3 channel is fully consistent with the energy dependent charge exchange cross section.

For flyby I27 the picture is more complex, as both electromagnetic fields which guide particles to the surface and charge exchange play a role in the formation of the depletion features, according to the simulations. The role of electromagnetic fields is confined to the region closest to Io, while the effect of charge exchange extends further. Furthermore, the depletion region in TP1 extends well beyond the depletion region in TP3, which is fully consistent with the energy dependence of charge exchange. We conclude that during I27 charge exchange is the major loss process during most of the flyby segment.

For flyby I31 the picture is different due to the polar geometry of the flyby. Here surface impact, perturbed fields and charge exchange play a role in the formation of the depletion features. However, also here the depletion region in TP1 extends well beyond the depletion region in TP3, which is fully consistent with the energy dependence of charge exchange. We conclude that during I31 charge exchange is a major loss process, becoming the dominant process on the outbound part of the trajectory which is on Io’s dayside/downstream side.

Nearest Io during I27 and I31 we observe that situations occur where half of the spin cycle or the entire cycle is fully depleted due to perturbed fields. The corresponding directions are difficult to interpret in terms of charge exchange, considering that any atmospheric model would fit them. Furthermore, the directions along the trajectory nearest Io where fields have no effect are aimed away from Io, thus this limits the sensitivity of the analysis in interpreting the density of Io’s atmosphere, nearest to Io.

Previously \citeAHuybrighs2020AnDepletions,Huybrighs2021ReplyDepletions have established that field perturbations and atmospheric charge exchange are required to explain particle dropouts measured during the E26 flyby of Europa (closest approach 350 km), in addition to losses by particles impacting on Europa’s surface. Whereas the same processes take place at Io, we point out that the relative importance of atmospheric charge exchange is much larger at Io, due to its substantially denser neutral environment. This is true for all flybys considered in this study, from the lower altitude flybys I27 (closest approach at 198 km) and I31 (closest approach at 193 km) as well as for the higher altitude flyby I24 (closest approach at 611 km).

Here we argue that an appropriate choice of electromagnetic field is also necessary to explain the proton depletion seen at 155-224 keV (TP1), together with charge exchange and surface impact, and at 540-1250 keV (TP3) together with surface impact, in flybys closer to the surface such as I27 (198 km) and I31 (193 km). At higher altitudes, such as flyby I24 (611 km), the choice of field configuration does not appear to cause losses. Losses due to the field configuration are caused when the field guides protons to the surface and makes the impact. Perturbations in the electromagnetic field can also form ’forbidden regions’ to which the access of the energetic proton is restricted (I31).

During flyby I27 (closest approach at 198 km), the field configuration causes losses, up to 100%, near the point of closest approach of Io in both TP1 and TP3 (Figures 6 and 4 in panel d, black curve). This is in contrast to flyby I24 (closest approach 611 km), which has a qualitatively similar geometry with respect to the upstream flow and Io as I27 (see Figure 1), however the choice of fields has no major effect on this flyby. Thus, due to the higher altitude the electromagnetic fields are unable to make the back-traced particles intersect with Io’s surface.

It should be noted that for flyby I31, the perturbed fields also reduce the range along the trajectory within which the depletion occurs. This is due to the fact that the fields bend the trajectories of the charged particles around Io, allowing a greater number of particles to pass from the south to north side of the moon than if homogeneous fields are assumed.

Even though the fields on their own do not cause proton losses during flyby I24, they do have a (minor) effect on the proton losses due to charge exchange. The simulated depletion pattern for I24 with atmospheric charge exchange and homogeneous fields (Figure 6, panel a) differs from the depletion pattern calculated with atmospheric charge exchange and perturbed fields (panel b in the same figure) in both shape and depth. With the inclusion of the perturbed fields points of maximum depletion are deeper by about 10%. This implies that even if the field configuration does not lead to charged particle depletion directly, it does affect the depletion caused by the atmospheric charge exchange. Hence, we argue that perturbed fields are also an influential, though minor, factor in the charged particle depletion seen in TP1 during flyby I24.

Here we argue for the existence of an extended downstream/dayside atmosphere during flybys I27 and I31, as well as the lack of evidence for complete atmospheric collapse on the upstream/nightside, based on the energetic proton measurements.

Figure 5 reveals some evidence for an asymmetry during flyby I27. By visual inspection it can be determined that the trend of the loss in TP1 (155-224 keV) strongly correlates with the radial distance on the outbound (right side) part of the trajectory, while the trend of the loss on the inbound (left side) of the figure is lagging the trend of the radial distance. Considering the evidence for atmospheric charge exchange is the main loss process, during most of the segment, this suggests that the atmosphere is less expanded on the nightside than on the dayside.

Flyby I31 crosses from the night side into the dayside over Io’s pole (see Figure 1), as well as from the upstream to the downstream plasma environment. Figure 6 shows that the measured depletion region during I31 extends further on the dayside/downstream region than on the nightside region/upstream. Thus, the energetic protons measurements are indicative of a dayside/upstream asymmetry in the proton losses. The simulation with the asymmetrical atmosphere (see Section 3.1), which incorporates a day-night asymmetry as well as an upstream-downstream asymmetry, demonstrates that the extended depletion from approximately 05:04:30, corresponding to the dayside, is caused by atmospheric charge exchange alone. Whereas the radially symmetrical atmosphere is unable to reproduce the extended loss on the dayside. Therefore we interpret the extended loss region on the dayside as an indication of an extended atmosphere on the downstream/dayside.

The atmosphere is expected to be denser on the dayside due to enhanced sublimation (e.g. \citeATsang2016TheEclipse). It is also thought to have a higher scale height on the downstream side (e.g. \citeASaur2002InterpretationPasses). Because the downstream and dayside coincide during I24 and I27 we cannot distinguish between these two possible physical explanations for the extension on the downstream/dayside.

For flyby I24 the trend of the loss follows the radial distance closely, we conclude that there is no clear sign of an atmospheric asymmetry in this flyby. The lack of a strong asymmetry does not necessarily imply that no asymmetry exists. We propose that it is the specific geometry of the I31 flyby, polar from the upstream/nightside to the downstream/dayside, that is most favourable to reveal the presence of asymmetries between the upstream/nightside and the downstream/dayside as compared to the other two. However, since I24 and I27 have similar geometries with the downstream/dayside, but different altitudes, it appears that the visibility of the asymmetry between the upstream/nightside and the downstream/dayside could be limited to lower altitudes.

We note that even though the atmosphere on the upstream/nightside extends less, there is still some loss on the upstream/nightside. From the proton measurements there is thus no evidence that the atmosphere is fully collapsed on the upstream/nightside.

Additional measurements, in particular those on the dayside when the dayside and the upstream side are not coinciding, could provide clarity on the role of enhanced sublimation in the formation of Io’s extended atmosphere on the dayside, additional flybys on the Jovian side could reveal the Jovian/anti-Jovian asymmetry.

In Section 4.3 we argued that for each of the three flybys there are viewing directions for which we underpredict the proton losses, that is where the modelled loss returns to unperturbed level whereas the data does not. We observe for the three flybys that the directions in which we underpredict the depth of the depletion corresponds to more field aligned particles that have a higher parallel velocity and thus smaller gyroradius. (An exception occurs when those particles pass through the denser part of the atmosphere closer to Io or the extended atmosphere on the upstream/dayside where they are depleted.) In contrast, the directions where the depth of depletion matches the data approximately correspond to particles that are less field aligned and have larger gyroradii. We suggest that a more extended atmosphere could compensate for the lack of depletion in the field aligned directions corresponding to smaller gyroradii, while retaining the relative agreement in other directions.

The protons with the larger gyroradius spend more time in the high density part of the atmosphere (due to a lower velocity parallel to the magnetic field) and thus have a higher chance of charge-exchanging in this region than a particle with the same total energy that has a smaller gyroradius and higher parallel velocity. If an additional extended atmospheric component were to be included, the larger gyroradius particles would still charge exchange at a similar position along their trajectory in back-tracing sense (since they are most sensitive to the denser part). The smaller gyroradius particles, which would have had a low chance of charge exchange in the dense part of the atmosphere, would get a higher chance of charge exchanging if there were an extended atmosphere that is of a larger physical scale. In this sense an extended atmosphere could increase the depletion for smaller gyroradius particles while having less effect on the larger gyroradius particles. Indeed, during I27 we observe that field aligned directions could be depleted when the densities are increased. At the pink marker during I27 (Figure 6, left side, panel c) the direction corresponds to particles that are closer to field aligned (see Figure 10), indeed the depletion at this point is reasonably well reproduced. On the other hand, at the green marker (left side) in the same figure, the field depletion is underpredicted. The difference between the two is that the pink marker travels through the densest part of the atmosphere, that is located at lower altitudes.

Additional simulations for a wide range of atmospheric properties are needed to constrain the properties of the extended atmosphere, we leave such a comprehensive enterprise as a subject for a future study. However, we do consider it likely that the missing atmospheric component extends beyond 2 Io radii, based on the fact that we observe the periodic discrepancy between the data and simulations throughout all flyby segments we considered.

The presence of such an extended atmosphere is supported by remote and in-situ observations of Io. Using the Hubble Space Telescope imaging spectrograph, [Wolven \BOthers. (\APACyear2001)] detected Far Ultraviolet (FUV) emissions of O and S neutral atoms extending to at least 10 R_Io around Io. Based on the direction of the instrument aperture, it can be inferred that these emissions are probably spherically distributed. Assuming nominal torus values for the electron density and temperature, Wolven et al. infer an Oxygen density $\sim$ 1-2 $10^{5}$ cm^-3 at 2 R_Io. The charge exchange cross sections of protons on O follow the same trend as H on O₂, but have somewhat lower values [Basu \BOthers. (\APACyear1987)]. We therefore consider charge exchange with the S and O corona as a possible explanation. Follow-up studies should consider a multi-species atmospheric model that includes the O and S corona and appropriate cross-sections.

Along most flybys of Io, Galileo observed Electro-Magnetic Ion Cyclotron (EMIC) waves at frequencies close to the SO${}_{2}^{+}$ and SO⁺ gyrofrequencies [Huddleston \BOthers. (\APACyear1998), Russell \BBA Kivelson (\APACyear2000), Russell \BOthers. (\APACyear2001), Blanco-Cano \BOthers. (\APACyear2001)]. Such EMIC emissions are produced when neutrals are freshly ionized, forming a typical PickUp ring-distribution in phase space. These EMIC emissions were observed mainly downstream of Io and sometimes far from the moon: along the J0 flyby in Io’s wake, EMIC waves were observed extending to $\sim$ 6-20 R_Io in the inbound-outbound directions of the Galileo trajectory. These EMIC detections suggest the presence of a tenuous corona of SO₂ and SO extending far from Io, mainly in the downstream direction. Under simplifying assumptions, \citeAHuddleston1997 infer a lower limit of the SO₂ density $\sim$ 10 cm^-3 at 4 R_Io but this estimate is very model-dependent. Although EMIC waves were also observed, at the S⁺ gyrofrequencies, such waves are usually damped by the torus thermal species and more difficult to observe. Based on Galileo PLS measurements of the plasma density and temperature along the I32 flyby, \citeADols2012AsymmetryFlybys propose the presence of SO₂ and SO coronae extending far downstream of Io and that such downstream coronae could be produced by SO₂ resonant charge exchange in Io’s atmosphere. On the contrary, our proton absorption data indicates an extended atmosphere both on the inbound (upstream) and outbound (downstream) parts of the spacecraft trajectories. In summary, there is observational evidence that Io’s atmosphere is very extended, which might explain some of the discrepancies between the observations and the simulations discussed above.

Above we argue that the more field aligned particles are more sensitive to an extended atmospheric component, unless they pass through a denser part of the lower altitude atmosphere and are already depleted there. However, above a few 1000 km altitude the density of the atmosphere no longer exceeds that of the plasma torus ($\sim 2000$ cm^-3) and charge exchange losses with the torus become more important. We don’t consider charge exchange with the low energy torus ions as a likely cause of the depletion features that we have attributed to a possible extended atmospheric component. The features that we focus on in this work only occur within 2 R_Io and are therefore likely an aspect of the interaction with Io’s local environment.

Our results show that the apparent proton losses during the flybys of Io are sensitive to both the electromagnetic field configuration and the atmospheric properties, because of neutral-ion charge exchange. These data thus provide an additional unexploited avenue to constrain the electromagnetic field and atmosphere structure around Io. A comprehensive parameter study to investigate the effect of the atmospheric and field properties on the proton flux forms the next logical step. Such studies should eventually take into account constraints provided by all Galileo datasets, including magnetic field (MAG), low energy plasma (PLS), plasma waves (PWS) and energetic ions (EPD). A multi-species atmospheric model or induced dipole, such as in \citeADols2012AsymmetryFlybys, should also be considered. However such a comprehensive endeavour is computationally intensive and we therefore consider it outside of the scope of this study. Instead this study aims to provide hypotheses and directions to guide such future modelling efforts.

Thanks to the fact that the TP3 channel of the EPD is sensitive to the field configuration but not sensitive to depletion through atmospheric charge exchange, data from this channel may be used as an initial constraint on the perturbed field model around Io. Once constraints are placed on what type of electromagnetic fields are present around Io, these may then be used in combination with atmospheric models to place constraints on the atmospheric structure of Io by looking at the depletion in the TP1 channel of the EPD.

When comparing atmospheric models, it is of great use to have different flight geometries. Flybys that have a point of closest approach similar to that of I24 (611 km) allow us to fit atmospheric models with less influence on the depletion by the electromagnetic fields. Furthermore, when combining these observations with lower flybys at the same latitude (such as I27), information can be gained on the vertical structure of the atmosphere allowing us to fit surface density and scale height accordingly. Polar flybys, such as I31, could be used for the testing of different longitudinally dependent models.

The Galileo data is of high value in this context, and will retain this value in the absence of a dedicated Io mission. The next missions destined for the Jupiter system, JUICE and Europa Clipper, are not scheduled to encounter Io from close by. However, the energetic particle data set has been expanded by two flybys by Juno with their closest approach at approximately 1500 km. Juno caries the energetic particle detector JEDI [Mauk \BOthers. (\APACyear2017)] which would be able to measure the proton flux.

Interpretation of the ion measurements is hampered by a lack of appropriate charge exchange cross sections. Here we used the cross section of H⁺ on O₂ as a proxy for SO₂. This assumption leads to an underestimation of the charge exchange cross section, which makes it more difficult to accurately estimate the absolute values of atmospheric densities. However, our conclusions on the atmospheric configuration depend on relative differences between inbound and outbound segments of the trajectory, as well as relative differences between viewing directions that already represent the depletion well and those that don’t, we therefore consider that these conclusions will hold, even for somewhat different charge exchange cross sections. Further laboratory studies could confirm our assumptions about the cross sections at higher energies as well as provide species specific cross sections, in particular cross sections for S and O ions are lacking.

Nenon2018AOrbit has modelled Jovian trapped protons with kinetic energies higher than 1 MeV. In the \citeANenon2018AOrbit model Io is considered as an inert absorber of protons. This work shows that the electromagnetic field configuration is important in determining the exact size of Io as a sink. For example, we have shown that from approximately equatorial flybys of 200 km (e.g. I27) protons can be guided to the surface and impact. On the other hand, the particle tracing simulations presented in this work show that near Io the energetic protons of energies above $\sim 100$ keV can also be partly deflected by the perturbed fields (e.g. flyby I31). Thus when treating Io as an inert body the losses at these energies could also be overestimated. Furthermore, \citeANenon2018AOrbit does not take into account charge exchange with Io’s atmosphere. Indeed, our simulations show that at energies of 1 MeV the effect of charge exchange will be minor. However, at lower energies where charge exchange is more efficient (e.g. below 224 keV) charge exchange will increase the size of Io as a proton sink in Jupiter’s magnetosphere, causing charge exchange losses up to at least 1 Io radius away.

The extended loss region around Io could be measured along the field lines connected to the loss region. Indeed losses of energetic protons with energies in the 10-800 keV range over a region larger than Io have been measured by Juno along field lines connected with Io [Paranicas \BOthers. (\APACyear2019)]. Whereas \citeAParanicas2019IosData focuses on wave particle interactions as a source of loss, this work illustrates that charge exchange (at energies below 224 keV) and the electromagnetic field configuration near Io will also play a role in the proton depletion. In fact, we suggest that the field configuration is a more important effect than losses due to wave particle interaction because our simulations at 540-1250 keV (TP3), have a good qualitative agreement with the measured loss regions, without the need of invoking wave-particle interactions.

In conclusion, we propose that the losses due to charge exchange near Io and apparent losses due to perturbed fields will also help explain losses observed along field lines connected to Io and its immediate environment.

Previously \citeAFutaana2015Low-energyTori have demonstrated that low-energy S and O energetic neutral atoms (referred to as LENA by the authors), of 460eV and 920eV respectively, resulting from the charge exchange of $<$1 keV plasma from the plasma torus with Io’s neutral particle environment are detectable by the Jovian Neutrals Analyzer (JNA) instrument on the JUpiter ICy moon Explorer (JUICE).

High energy ($>$10 keV) ENA emissions were first reported by \citeAKirsch1981 based on Voyager 1 LECP measurements. However, the mass or charge state could not be determined. Later, \citeAKrimigis2002 reported ENA emissions, possibly 50-80 keV hydrogen atoms, from Jupiter, based on Cassini INCA measurements. Subsequently, \citeAMauk2003 attributed the origin of the ENA emissions to Jupiter’s atmosphere and a possible neutral cloud slightly outside of the orbit of Europa. The emissions were not attributed to Io’s torus or Io itself.

This study suggests that energetic protons with energies exceeding 155 keV will charge exchange with Io’s atmosphere, and produce ENAs of $\sim 150$ keV energies. When the energetic protons become Energetic Neutral Atoms (ENA) they are no longer bound by the electromagnetic field and will escape at high energies that are similar to their energies before neutralisation [<]e.g.¿[]Rees1989PhysicsAtmosphere. While the proton flux at $\sim$ 100 keV is about two orders of magnitude less at Io than at Europa, the total mass of Io’s atmosphere is 100 times that of Europa [Bagenal \BBA Dols (\APACyear2020)]. Therefore, we recommend that future studies on the interpretation of ENA imaging of Jupiter’s magnetosphere also consider Io as a possible source of $\sim$100 keV ENAs.

These $\sim$100 keV ENAs could be detected by the Jovian Energetic Neutral and Ions (JENI) detector part of the Particle Environment Package (PEP) on the JUICE mission [Barabash \BOthers. (\APACyear2013)]. Variations of keV ENA flux at Io could provide information about variations in the energetic proton flux and variations in the neutral environment of Io. While the LENAs will reach JUICE within a few hours [Futaana \BOthers. (\APACyear2015)], the ENAs could reach JUICE in a Ganymede orbit within just a few minutes, thereby providing a more instantaneous assessment of the neutral-ion interaction at Io.

Proposals are put forward for new dedicated missions to Io [<]e.g.¿[]McEwen2021FutureIo. This study demonstrates the added value of including an energetic ion detector to investigate the electromagnetic field configuration and atmospheric structure around Io. While a magnetometer or neutral mass spectrometer only measures the field and atmospheric density along the flyby, the energetic ions are sensitive to the wider environment of Io. Their flux dropouts will thus provide information about the field configuration outside of the flyby trajectory. Thus, we recommend to include an energetic particle detector on future missions to Io.

A combination of lower and higher altitude flybys would be beneficial for isolating the relative importance of electromagnetic field perturbations and atmospheric charge exchange on the resulting proton loss. Flybys at different longitudes, latitudes and local times could help disentangle the atmospheric asymmetries.