A Possible Dust Origin for an Unusual Feature in Io's Sodium Neutral Clouds

Io's Neutral Clouds are one of the brightest visible manifestations of the intense volcanic and magnetospheric activity in the Jovian system. All the material that fills the Io Plasma Torus and Io's Neutral Clouds ultimately comes from the intense volcanic activity of Io. Since the discovery of sodium at Io (Brown 1974), and its Neutral Cloud (Trafton et al. 1974), this minor but bright species has been used to monitor the response of Io's Neutral Cloud to the changes of its drivers (Io volcanic activity and Jupiter magnetosphere). The resonant scattering efficiency (the so-called g value) of sodium is orders of magnitude greater than the dominant species in Io's Neutral Clouds (S and O and their molecular compounds). Moreover, the wavelength of the sodium doublet (the D2 line at D1 line around 5900 Å) is such that it can be easily detected from the ground. Over the past decades, observations and models have identified structures in the Neutral Clouds, each related to a particular source process and interaction with the Jovian magnetosphere. The most prominent of these structures, and the first to be discovered, is the corona, sometimes referred to as the "banana cloud" (the "Region B" in Brown et al. 1975), composed of slowly escaping sodium atoms (a few km s⁻¹ relative to Io). Its peculiar morphology, with the leading part more extended (about 60° along the orbit) than the trailing part (30° along the orbit), is controlled by celestial mechanics and by ionization from the plasma torus, which is warmer outwards from Io's orbit compared to inwards (Nash et al. 1986). Another feature of the sodium Neutral Clouds is the "jet" (or directional feature; Goldberg et al. 1984), much more variable than the banana cloud, and composed of much more energetic atoms. The "jet" extends from Io in the anti-Jupiter direction, oriented approximately perpendicularly to the local Jovian magnetic field (Wilson & Schneider 1994). It is probably caused by prompt pickup ion neutralization very close to Io, its narrowness indicating an unperturbed Jovian magnetic field at Io (Wilson & Schneider 1999). The outer acceleration results from the corotational electric field applied to positively charged ions. Finally, a more diffuse feature is the "stream" (Schneider et al. 1991) of fast moving neutral sodium atoms resulting from dissociative recombination of fast molecular sodium-bearing ions (most likely NaCl⁺) corotating with Jupiter at ∼70 km s⁻¹. These atoms can be found all around Ios orbit. Over many orbits around Jupiter, this stream creates the vast, faint neutral sodium cloud detected hundreds of Jupiter radii from Jupiter (Mendillo et al. 1990). All these features contribute to the roughly 1 ton of plasma that is injected in the Io Plasma Torus every second (Schneider & Bagenal 2007). Their morphology depends on several factors, including Io's position (both orbital longitude and magnetic latitude) Jupiter's magnetic field, and Io's volcanic activity (see review by Bagenal & Dols 2020). For example, it has long been debated whether the sodium Neutral Clouds decrease in extent during eclipses behind Jupiter, due to condensation of Io's atmosphere and/or to a suppression of solar photons available for photodissociation of sodium-bearing molecular compounds. It was to test these possibilities that we performed ground-based observations of Io before and after eclipses, in 2007. We discovered that solar flux is an important factor in supplying the sodium atoms to the Neutral Clouds, and that suppression of photodissociation of sodium-bearing molecules (most likely NaCl) during eclipse drives a decrease in exospheric Na atoms soon after egress (Grava et al. 2014).

During those observing runs, we also discovered an additional spectroscopic feature: an emission of sodium (visible at both D-lines) blueshifted (i.e., directed toward the observer and thus toward Jupiter) by tens of km s⁻¹ relative to Io, hence moving much faster than the escape speed at Io (see Figure 1). We first reported this feature in Schneider et al. (2008) and performed additional observations at the same telescope in 2009. This paper describes the analysis of this feature using simulations of sodium atoms in Io's Neutral Clouds and suggests possible explanations.

Zoom In Zoom Out Reset image size

Spectra were collected by the now-decommissioned high-resolution échelle spectrograph SARG mounted at the 3.6 m Italian telescope Telescopio Nazionale Galileo (TNG) during several nights spanning 2 yr, and different geometric configurations of Io. Spectra taken in 2007 were analyzed by Grava et al. (2014) and included in this work for completeness, while spectra taken in 2009 are analyzed here for the first time. The spectrograph's slit was 26."7 long and 0."4 wide, had a spectral resolution of λ/Δλ ∼ 115,000, and was placed almost always parallel to Jupiter's rotational equator and centered on Io (or on other Galilean satellites, when used for calibration). The exposure time was almost always 600 s. The 2007 spectra were obtained with an interference filter which blocked all photons but those with wavelengths close to the sodium doublet. This procedure allowed us to use the full extent of the long slit without order overlapping. The 2009 spectra were obtained without such interference filter, to collect light from the entire bandpass of SARG (3700–10,000 Å) and to study, therefore, other species. As such, only the central ∼50 pixels were used, and the remaining ∼30 pixels at the edges were excluded due to order contamination. The effective length of the slit in this case was 17."7.

Data reduction of the spectra was described in depth in Grava et al. (2014) and is summarized here. Standard data reduction steps such as subtraction of bias, flatfielding, and wavelength calibration (using a Th–Ar lamp) were performed using IRAF routines. Specific data reduction steps pertinent to our data set were performed later and included: (1) removal of an interference pattern caused by the sodium filter (for 2007 observations only); (2) removal of a "pedestal" (residual background); (3) removal of a "ghost" of Jupiter's spectrum on top of Io spectra; (4) removal of telluric absorption, performed using rapidly rotating O- and B-type stars devoid of absorption lines observed at several airmasses that bracketed ours; (5) subtraction of the "continuum," i.e., the solar light reflected off Io's disk using spectra of other Galilean satellites (Europa, Ganymede, and, more rarely, Callisto) properly shifted to account for the different Doppler speeds; (6) conversion of sodium brightness from counts s⁻¹ to Rayleighs (1 R = 10⁶/4π photons cm⁻² s⁻¹ sr⁻¹; Hunten et al. 1956) using Jupiter spectra and the well-measured intrinsic brightness of Jupiter at the sodium wavelength (5.5 MR Å⁻¹; Brown & Schneider 1981). During most of the nights the seeing was good, never exceeding 16" as FWHM (calculated as a deconvolution of a Gaussian fit to Io's continuum and the theoretical diameter of Io's disk, 1."2). Details of observations are summarized in Table 1.

Figure 2 shows a typical spectrum of Io. On the left is the average of Io's brightness over detector columns 31–33, corresponding to a distance of 1.7–2.9 R_Io eastwards on the sky plane (toward Jupiter, in this case) of Io's center (1 R_Io ∼1820 km) and, in dashed line, the g values. The g value (or g factor) is the number of solar photons resonantly scattered each second by each sodium atom. In optically thin exospheres, like Io's neutral clouds, the brightness I (in Rayleighs) is directly related to the line-of-sight column density N (in atoms cm⁻²) by the g-value g (Brown & Yung 1976):

$$\begin{eqnarray}&&I=g\cdot N.\end{eqnarray} \tag{ 1 }$$

The g value (expressed in photons atoms⁻¹ s⁻¹) is related to the solar spectrum, and depends, among other things, on the heliocentric radial velocity Δv of sodium atoms, owing to the deep Fraunhofer absorption in the solar spectrum Hunten et al. (1988). We used g values for sodium from Killen et al. (2009). Δv is measured in our spectra by the Doppler shift Δλ of Io relative to the reference wavelengths λ of the sodium D-lines (5889.95 and 5895.92 Å for the D2 and D1 line, respectively):

$$\begin{eqnarray}&&{\rm{\Delta }}v=c\cdot \displaystyle \frac{{\rm{\Delta }}\lambda }{\lambda }\end{eqnarray} \tag{ 2 }$$

where c is the speed of light. The resulting Δv is referred to Io's reference frame, which is in itself moving relative to the Sun. An atom at rest at Io moving away from the Sun (like at western elongations) will "see" a redshifted solar spectrum. Therefore, to properly convert brightness into column densities it is necessary to blueshift the g values, i.e., shifting them toward negative velocities. This is shown in Figure 2, panel (a), which shows that the "dip" in the emission line (in Rayleighs) at −5 km s⁻¹ is caused by the "dip" of the Fraunhofer line of the g value at −2.5 km s⁻¹. The resulting division gives the column density without such "dip" (panel (b)), but with a "bump" on the blue side of the peak (negative velocities). This is the new emission feature. It is reminiscent of the "skirt" detected first by Trafton (1975) and then by Cremonese et al. (1992) and interpreted by Trafton & Macy (1977, 1978) to be due to sodium atoms streaming away from Io at moderate speeds (18 km s⁻¹ at most) in the leading direction. However, it is different from that because the "bump" lies on the opposite side. Our "bump" is more prominent on the short wavelength side when Io is west of Jupiter. Moreover, as we shall see briefly, the feature changes Doppler shift through the night. Note that in our observations the west direction corresponds to the orbital trailing side of Io. In terms of plasma flow, the trailing side corresponds to the upstream side, and East is the leading (or downstream) side.

Zoom In Zoom Out Reset image size

We performed the conversion from brightness to column density for all of our 2D spectra. An example of the resulting 2D spectra, together with the geometry of the observation, is shown in Figure 3 (the others can be found as a Figure set in the online Journal). In this figure, Io is about to enter Jupiter's shadow and the main emission feature (the "banana cloud"), being approximately at rest relative to Io (white horizontal dashed lines), dominates the spectra. The faint, straight emission line traversing the whole slit length is Earth's sodium layer at 90 km altitude resonantly scattering solar photons shortly before dawn. The streak of diffuse emission very close to Io on the Eastern side (toward Jupiter) extending to ∼75 km s⁻¹ is the stream of fast sodium atoms that feed the giant sodium nebula (e.g., Flynn et al. 1992). The feature modeled in this paper is the extended emission blueshifted by few of tens of km s⁻¹ with respect to Io, meaning these sodium atoms are moving toward the observer. The feature is seen here in the eastern half of the slit (left, in the figure), i.e., toward Jupiter.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (49 panels)

Tarball of figure set images

We applied our Monte Carlo model of sodium atom trajectories (Burger et al. 2014) to reproduce our observations as closely as possible, by including the location and direction of ejection and the velocity distribution of ejected atoms. These parameters are constrained by our observations. The goal is to reproduce the emission feature and its temporal variability.

Figure 4 illustrates the change in time of the new feature's morphology at different orbital (or subsolar) longitudes. This sequence shows that the blueshifted feature, more prominent toward the end of the observing run, early in the night is redshifted by a similarly high speed (>10 km s⁻¹). The transition from red to blueshifts suggests that Na is ejected in a direction that is roughly perpendicular to the Io-observer line when the subsolar longitude is ∼330° (between the 4th and the 5th spectrum in Figure 4). This red to blueshift can then be explained by the changing observing geometry as Io orbits Jupiter. This peculiar behavior inspired our modeling approach. We first modeled the trajectories of the neutral sodium atoms (Section 4.1) and then found a plausible physical mechanism that can bring these sodium atoms at the location and with the speed we observe them (Section 4.1).

Zoom In Zoom Out Reset image size

4.1. Neutral Sodium Atoms Trajectories

To simulate the observations we used the Monte Carlo model that is described in Burger et al. (2014). In this model, atoms are tracked under the forces of gravity and solar radiation pressure. The equations of motion are solved with a 5th order, adaptive step-size Runge–Kutta algorithm. Atoms are ejected at randomly selected times between the start and end of the model run so that the simulation contains a mixture of freshly released and older atoms. Loss processes include photoionization, electron-impact ionization, and charge exchange, though the atoms in the simulations shown below are mostly lost by leaving the small field of view. The observation geometry was recreated using the SPICE toolkit (Acton 1996).

We ran the model many times with different initial conditions in order to determine the direction and speed of Na that would reproduce the red to blueshift in Figure 4. Sodium atoms are ejected radially in intervals 1125 wide. The origin of the ejected atoms is labeled in Io's west longitude, where 0° corresponds to the sub-Jupiter meridian and 90° points along Ios orbital motion (the "leading" direction). Speeds values from 4 to 130 km s⁻¹ were used, each with a narrow distribution (each bin being 10 m s⁻¹ wide).

Each of the two free parameters we used to reproduce our observations affects only one aspect of the emission feature. The ejection speed distribution affects the Doppler shift (wavelength dimension), while the ejection location affects the morphology of the emission feature (spatial direction). This is illustrated in Figure 5, where we show the effects of changing these parameters, compared to the nominal, best model.

Zoom In Zoom Out Reset image size

We find that the best combination of parameters to reproduce our observations is a broad speed distribution, with ejection velocities relative to Io of 50–90 km s⁻¹, and a relatively narrow range of ejection location: 45°–68° of Io west longitude, meaning an ejection from the sub-Jovian/leading hemisphere. Models with the nominal direction but lower speed (40 km s⁻¹) or higher speed (130 km s⁻¹) fail to match the magnitude of the Doppler shift, though they maintain the red-to-blue transition at a subsolar point of ∼ 330° (left-right panels in Figure 5). The mismatches in direction include one angular bin toward the leading direction and one bin toward the sub-Jovian direction, and these both fail to reproduce the red-to-blue transition and the magnitude of the Doppler shift (top-down panels in Figure 5).

Figure 6 shows the best simulations compared to some of the observations. The rest of the data-model comparison can be found in the animation associated with Figure 6 in the online journal. The nominal model shown here (in both Figures 5 and 6) includes speeds of 57 km s⁻¹ and 90 km s⁻¹, though lower speeds could be included without significantly changing the result. Figure 6 illustrates how our model can reproduce the shift in velocity relative to Io (from red, or positive, to blue, or negative) within the same amount of time (about 4 hr).

Download figure:

Video Standard image High-resolution image

Figure 7 shows a top-down view of the nominal model of sodium trajectories near the beginning and end of the observing sequence. In this image, directions toward the observer (Earth, ⊕, and Jupiter, ♃) are indicated. The red-to-blue shift is seen here as a change in ejection direction as seen from Earth. Also, it can be noted that the ejection direction is roughly halfway, in terms of degrees, between the leading direction and Jupiter direction.

Zoom In Zoom Out Reset image size

4.2. Dust Grains Trajectories

Having found the characteristics of the sodium feature (speed distribution and direction), we need to identify a process that may bring sodium atoms at the location and speed observed. The direction of the blueshifted sodium feature (toward Jupiter) is consistent with negatively charged particles that move following Jupiter magnetosphere's corotational electric field (e.g., Horányi et al. 1997). There are two possible candidates: sodium-bearing negative ions that are promptly neutralized and negatively charged Na-rich dust grains from which sodium atoms are liberated. In this section we discuss the latter, as many of the grains' properties agree with our directional feature; at the end of Section 5.1 we briefly discuss the negative ions hypothesis.

Dust in the Jovian environment has been detected by multiple spacecraft: Ulysses (Grün et al. 1993), Galileo (Grün et al. 1996), and Cassini (Postberg et al. 2006). All these observations are consistent with electrically charged dust grain of radius ∼10 nm (Zook et al. 1996). The periodicity of the dust impact signal in Galileo data was one of the early indicators that Io, and not the gossamer rings, was the source of such dust grains (Graps et al. 2000). Subsequently, the Cosmic Dust Analyser (CDA) onboard Cassini was able to distinguish the composition of these grains. NaCl was the dominant species identified by CDA, followed by other components like Na₂SO₄ and K₂SO₄ (Postberg et al. 2006). This is in contrast with the main composition of Io's atmosphere and neutral clouds, where NaCl, Na⁺, and Cl⁺ are just trace species (NaCl fraction in composition is less than 1%, while the Na⁺ and Cl⁺ concentration in the torus is between 2% and 6%; Lellouch et al. 2003; Küppers & Schneider 2000) The inverse trend in composition for dust may be related to the process that ultimately ejects these particles in the first place, i.e., volcanic eruptions. Because of their high condensation temperature, NaCl and KCl are abundant condensates 20 minutes after outgassing from the vent; while sulfuric components, due to their lower condensation temperature, are still far from condensation (Zhang et al. 2004).

We performed simulations of the trajectory of negatively charged nanodust grains of various sizes. We assume that the dust particles 10 nm in radii start with a circular Keplerian motion at Io's orbit and an initial charge of 20 electric charge, corresponding to a surface potential of −3 V. The charging and dynamical evolution of nanometer-sized grains is modeled as described in Horányi et al. (1997). For individual dust grains, two equations are integrated simultaneously: the equation of motion and the grain charging equation. The equation of motion considers the gravity of Jupiter and the Lorentz forces acting on the grain from the corotating magnetosphere. The grain charging equation depends on plasma properties at the grain's location and determines the grain charge-to-mass ratio, essential to calculate the Lorentz forces. Grains can be charged negatively or positively, depending on the plasma environment, and the charging begins almost instantaneously.

A top-down view of the trajectory of a negatively charged dust grain of radius 10 nm, as well as a positively charged one (same size), is shown in Figure 8. Figure 9 shows the direction, the relative speed and distance with respect to Io of a negatively charged dust grain of radius 10 nm. The gray circle in that figure represents the parameter space (both velocity and distance to Io) consistent with our observations. The dynamical evolution of charged nanoparticles largely depends on the grain charge polarity: the positively charged particle is accelerated away from Jupiter in almost a straight trajectory, while the negatively charged one exhibits a cycloidal motion along Ios orbit similar to pickup ions. In general, negatively (positively) charged grains are accelerated toward (away from) Jupiter because of the outward-pointing corotating electric field (e.g., Horanyi et al. 1993). Within the first cycloid (within the first ∼15 minutes), when the grain is within ∼10 Io radii, both its direction and speed are comparable to the properties of the sodium jet. Grains of larger size (e.g., 100 nm), with a much smaller charge-to-mass ratio, are less accelerated by Lorentz forces compared to 10 nm radius dust grains. Therefore, their velocity relative to Io is much smaller than the observed one. Similarly, much smaller particles (size < 10 nm) have a much smaller gyroradius and tend to stay much closer to Io (a couple of Io radii at most) than our sodium feature. We shall return to this point in the next section.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Nanodust particles get charged at high altitude at Io by capture of ionospheric electrons or photoelectron production (Krüger et al. 2003b). Only the most energetic plumes are expected to bring grains to high altitudes so that they collect sufficient charge from ambient plasma to overcome Io's gravity (Johnson et al. 1980). Flandes (2004) showed that particles need to reach at least ∼400 km altitude to be able to escape, and the flight time from vent to this height is 15–20 minutes.

Once they get charged, these nanodust particles follow trajectories that are determined by the corotational electric field of Jupiter's magnetosphere. The dependence of dust trajectories on the highly variable magnetosphere implies a variability in the grains trajectories even if the dust ejection rate is constant. This variability, combined with the narrow field of view of our slit, might explain the variability seen in our data set within a couple of hours (Figure 4). Most previous Io nanodust detection were carried out at distances further away from Ios orbit. In this respect, our observations probe a much closer region to Io than all previous dust measurements.

The spatial information of the observed Na jets also provides constraints on the dynamics of its possible dusty source. The observed Na speeds correspond roughly to ∼50 km s⁻¹, but grains in the dust model (10 nm in radius with 20 e⁻ initial charge) need to be accelerated for some distance before reaching those speeds, at least 7 R_Io as seen from Earth.

A dust grain with lower initial charge gets charged later, and thus farther away from Io than our observations. Only grains with proper charge-to-mass ratio can be sufficiently accelerated by the Lorentz forces and maintain a direction comparable to the observed sodium jet. To first order, regardless of the grain size, grains in the same plasma environment will be charged to roughly the same electrostatic potential determined by the ambient plasma conditions. For a given electrostatic potential, the grain charge-to-mass ratio is inversely proportional to the square of the grain size. Negatively charged grains much larger than 10 nm have lower charge-to-mass ratio and would not be accelerated to the observed speed in the vicinity of Io. Negatively charged grains much smaller than 10 nm, on the other hand, have much smaller gyroradius and will produce both red and blue shifted components in the vicinity of Io, which is also not consistent with the observation. The grain dynamics thus suggests that the sodium jet can only be produced by charged grains with radius ∼10 nm. This is in good agreement with previous grain size estimate of the Jovian stream particles based on space dust instruments onboard Galileo and Cassini spacecraft (Postberg et al. 2006; Hsu et al. 2012).

It is possible to perform a rough estimate of the source rate from the spectra and compare it to the source rate from the neutral model. The sodium source rate r is the product of the peak column abundance, the vertical extent d of the emission, and the velocity v_⊥ perpendicular to the observer's line of sight (Schneider et al. 1991):

$$\begin{eqnarray}&&r=d\cdot {v}_{\perp }\cdot {\int }_{{v}_{1}}^{{v}_{2}}n(v){dv}.\end{eqnarray} \tag{ 3 }$$

In the above formula, we used Io's diameter as vertical extent d of our feature; n is the column density per unit velocity averaged between 3.0 and 7.0 R_Io; and the integral is performed between v₁ = 7 and v₂ = 22 km s⁻¹ relative to Io, where the blueshifted feature is more prominent. Finally, the velocity perpendicular to the observer's line of sight is defined as ${v}_{\perp }={v}_{\mathrm{los}}\cdot \tan \left(\theta \right)$, where v_los is the line-of-sight velocity (the one our spectroscopic observations are sensitive to) and θ is the angle between our line of sight and the jet (we assume our directional feature, or "jet," to be the axis of the cone of sodium atoms—see Figure 7). This angle is necessarily taken out of our modeling, since with spectroscopic observations we can only measure the radial (line-of-sight) velocity and have no information on the true orientation of the sodium feature. For each spectrum, we average the D2 and the D1 source rates and report these numbers in Table 2. The median of the resulting sodium source rates is 7.0 × 10²⁵ Na s⁻¹ from the data, corresponding to 2.7 kg s⁻¹. The source rate from the neutral model is somehow greater (2.6 × 10²⁶ Na s⁻¹, corresponding to 10.0 kg s⁻¹), but this was derived assuming a wide "jet" (30° in the vertical direction). A narrower feature would require proportionally less sodium.

5.1. Sputtering of Na from Nanodust Grains

For the dust hypothesis to work, we need a mechanism that releases sodium atoms from these nanodust grains. Possible processes include photon-stimulated desorption (PSD), electron-stimulated desorption, or ion sputtering. The first two have too low a source rate. From experiments of PSD of Na from surfaces that simulate the lunar silicates (Yakshinskiy & Madey 1999), scaling the solar flux to Jupiter's distance, and taking into account our derived distribution of dust grains, the source rate from PSD is 10⁹ s⁻¹. The same authors also showed that electron-stimulated desorption has an only slightly higher cross section compared to PSD. The other mechanism, ion sputtering, gives higher source rates, although there are several uncertainties that affect the calculation of the rate. First, an absolute value for sputtering yield of sodium atoms has only been measured on Na₂S: Chrisey et al. (1988) report a sputtering yield of 0.1 and 0.2 Na₂S ion⁻¹ from O⁺ and S⁺ ions, respectively. The molecules of interest in our case are sodium chloride (NaCl) and sodium sulfate (Na₂SO₄), given that these were the compositions of the Na-bearing dust grains detected by Cassini CDA (Postberg et al. 2006). Sodium sputtering yield for NaCl has been estimated by Johnson (2000) to be ∼1.0 Na ion⁻¹, assuming that NaCl has the same atomic ejection efficiency as Na₂S. Sodium sputtering yield for Na₂SO₄ has only been reported as a relative value: it is 100–1000 times greater than the sputtering yield for NaS (Wiens et al. 1997). Assuming the absolute sputtering yield of NaS to be the same as that of Na₂S, we can estimate the Na sputtering yield from Na₂SO₄ to be 100–1000 times greater than that of NaCl, or 100–1000 Na ion⁻¹. The flux Φ of precipitating ions responsible for the sputtering (mostly S⁺ and O⁺) can be calculated by multiplying the velocity of the ions (co-rotation, or 57 km s⁻¹) by the density of these heavy ions close to Io, or about 1500 cm⁻³ (Dougherty et al. 2017). We obtain Φ = 8.6 × 10⁹ Na cm⁻² s⁻¹.

The second uncertainty is the dust available for sputtering. Over the course of the 7 yr mission and from a distance between 13 and 400 R_J from Io, the Galileo Dust Detector System measured dust production rates at Io between 10⁻³ and 10² kg s⁻¹ (with an average value between 0.1 and 1.0 kg s⁻¹), possibly correlated to volcanic eruptions at Io (Figure 2 of Krüger et al. 2003a).

With these numbers, we can estimate the number of Na atoms sputtered by nanodust grains and compare it to our observed sodium production rate. We start by assuming a dust production rate of 1 kg s⁻¹, the upper value of the average range reported by Krüger et al. (2003a). Assuming a spherical shape and a density of 1.5 g cm⁻³ (Krüger et al. 2003a), this corresponds to 1.6 × 10²⁰ grains s⁻¹ of radius 10 nm. We now multiply this number by the time it takes for the dust grains to reach the point where we do see sodium emission, or 0.05 hr (from Figure 9), to get N_grains = 2.9 × 10²². The total area available for sputtering will be N_grains · A · c, where A is the area of a 10 nm radius spherical nanodust grain, and c the concentration of either NaCl (90%) or Na₂SO4₄ (10%), from Cassini CDA measurements (Postberg et al. 2006). These areas are 3.2 × 10¹¹ cm² and 3.6 × 10¹⁰ cm² for NaCl and Na₂SO₄, respectively. Multiplying them by the corresponding sputtering yields Y and the heavy ion flux, we get the sodium production rate for sputtering s:

$$\begin{eqnarray}&&s={N}_{\mathrm{grains}}\cdot A\cdot c\cdot {\rm{\Phi }}\cdot Y.\end{eqnarray} \tag{ 4 }$$

We obtain 2.8 × 10²¹ Na s⁻¹ from NaCl, and 3.1 × 10²² Na s⁻¹ from Na₂SO₄ from 1 kg s⁻¹ of dust material.

These numbers are between 3 and 4 orders of magnitude lower than our inferred rate of 7.0 × 10²⁵ s⁻¹ (from the observations). They can be reconciled with our values if we assume some combination of the highest estimate for dust production rate from Galileo (100 kg s⁻¹), the highest estimate for the sputtering yield of Na from Na₂SO₄ (1000 Na ion⁻¹), and a surface area for porous particles that is ∼10 times higher than the spherical area (this is the case for micron-sized lunar regolith particles; Taylor & Meek 2004). Additionally, the dust production rates from Galileo should be considered a lower limit, as dust is likely to undergo other destruction processes (not included in the calculations of Krüger et al. 2003a) between Io and the distance at which grains were detected by Galileo (between 13 and 400 R_J from Io). This is particularly true if some ejected dust has a more volatile composition than NaCl. Finally, it is not clear how the ion sputtering experiments, performed on flat surfaces, would change for dust composed of fluffy grains. To further investigate the discrepancy, a detailed calculation of the release of Na from dust particles and full treatment of dust destruction in Jupiter's highly complicated plasma environment would be required, and these are outside the scope of this paper.

We note that the new sodium feature was most prominent when Io was within 5° of the magnetic equator (Table 1). The bulk of the Io plasma torus is confined in the centrifugal equator, which is very close to the magnetic one. The plasma environment of that region (high plasma-electron density and low electron temperature; Meyer-Vernet et al. 1995) is consistent with the negative grain charge polarity inferred from the trajectory simulation. In addition, the sputtering erosion responsible to release sodium atoms from dust surfaces is also higher at low magnetic latitudes because of the higher ion density. Our results thus indicate an indirect pathway to deliver sodium atoms, and likely other species, from Io to the neutral clouds (and ultimately to the magnetosphere) by sputtering from dust surfaces.

We conclude this section by evaluating and discarding another possible candidate, mentioned in Section 4.2, that is a population of negative ions formed inside Io's plumes. In this case, the most likely process to create these ions is three-body electron attachment. The most likely species to undergo this process is NaSO₄, a radical species with a substantial electron affinity and the highest-rate coefficient among several Na-bearing species. The negative ions would then release the neutral sodium atoms we observe following electron detachment. The concentration of NaSO₄ inside Io's plumes is unknown, but to match our observed source rate per unit volume, a density similar to that of potassium would be required, or 9 orders of magnitude greater than the the expected concentration of Na₂SO₄ inside Io's plumes (Moses et al. 2002). For this reason, we favor the hypothesis of ion sputtering from negatively charged dust nanograins as the most plausible source of our new sodium emission feature.

We have performed a Monte Carlo simulation of sodium atoms under the influence of Io's gravity and solar radiation pressure to explain an unusual Jupiter-oriented feature (Schneider et al. 2008) we detected in our high-resolution spectra of Io's sodium Neutral Clouds from the TNG telescope. This feature is directed toward Jupiter and is rapidly variable in time. The best model is the one that has the sodium atoms ejected in the leading-sub/Jovian hemisphere of Io (45°–68° west longitude) with a broad velocity distribution (50–90 km s⁻¹). We propose that the mechanism most likely responsible for creating sodium atoms with that orientation and that speed is sputtering of Na from Na-bearing molecules (NaCl or Na₂SO₄, both detected near Jupiter by Cassini CDA) attached to negatively charged dust grains (10 nm in radius) that move under the influence of the corotational electric field of Jupiter's magnetosphere. This is consistent with modeling of trajectories of negatively charged nanodust grains, which present the right velocity and orientation consistent with our observations and our sodium model. The median sodium source rate inferred from our observations is 7.0 × 10²⁵ Na s⁻¹, to be compared with the theoretical estimate from ion sputtering of Na from NaCl or Na₂SO₄ between 5.5 × 10²¹ Na s⁻¹ and 6.2 × 10²⁶ Na s⁻¹, depending on the choice of several uncertain parameters, such as the sputtering yield, the dust production rate at Io, and surface area available to sputtering. We point out the need for detailed calculations of the release of Na from dust particles and full treatment of dust destruction in Jupiter's highly complicated plasma environment.

Our results uncover a new mechanism by which Io's sodium Neutral Clouds are replenished, and highlight the need for future observing campaigns to better constrain the escape rate of sodium atoms produced at Io by this mechanism; before the arrival of Europa Clipper spacecraft, whose dust counter will be able to study the dust population in the Jupiter environment at much closer range than any other spacecraft so far.

This work has been supported by NASA Solar System Workings grant 80NSSC18K0008, NSF's Planetary Astronomy Program, and the Astronomy Department and CISAS of University of Padua, through a contract by the Italian Space Agency (ASI). C.G. wishes to thank Rosemary M. Killen for providing the g-values used in this work, and Jane L. Fox and John M. C. Plane for insightful discussions on negative ions chemistry. Based on observations made with the Italian Telescopio Nazionale Galileo (TNG) operated by the Fundación Galileo Galilei of the Istituto Nazionale di Astrofisica at the Observatorio del Roque de los Muchachos (La Palma, Canary Islands, Spain). Files for observations in 2009 are available on the archive of the TNG telescope, at http://archives.ia2.inaf.it/tng. Files for observations in 2007 can be requested to the corresponding author.