Research Article

Dust grains fall from Saturn’s D-ring into its equatorial upper atmosphere

See allHide authors and affiliations

Science  05 Oct 2018:
Vol. 362, Issue 6410, eaat2236
DOI: 10.1126/science.aat2236

Cassini's final phase of exploration

The Cassini spacecraft spent 13 years orbiting Saturn; as it ran low on fuel, the trajectory was changed to sample regions it had not yet visited. A series of orbits close to the rings was followed by a Grand Finale orbit, which took the spacecraft through the gap between Saturn and its rings before the spacecraft was destroyed when it entered the planet's upper atmosphere. Six papers in this issue report results from these final phases of the Cassini mission. Dougherty et al. measured the magnetic field close to Saturn, which implies a complex multilayer dynamo process inside the planet. Roussos et al. detected an additional radiation belt trapped within the rings, sustained by the radioactive decay of free neutrons. Lamy et al. present plasma measurements taken as Cassini flew through regions emitting kilometric radiation, connected to the planet's aurorae. Hsu et al. determined the composition of large, solid dust particles falling from the rings into the planet, whereas Mitchell et al. investigated the smaller dust nanograins and show how they interact with the planet's upper atmosphere. Finally, Waite et al. identified molecules in the infalling material and directly measured the composition of Saturn's atmosphere.

Science, this issue p. eaat5434, p. eaat1962, p. eaat2027, p. eaat3185, p. eaat2236, p. eaat2382

Structured Abstract


Ring material has long been thought to enter Saturn’s atmosphere, modifying its atmospheric and ionospheric chemistry. This phenomenon, dubbed “ring rain,” involves the transport of charged dust particles from the main rings along the planetary magnetic field.


At the end of the Cassini mission, measurements by onboard instruments tested this hypothesis as well as whether ring material falls directly into the equatorial atmosphere. The final 22 orbits of the Cassini mission sent the spacecraft through the gap between the atmosphere and the innermost of the broad ring system, the D-ring.


The Magnetospheric Imaging Instrument—designed to measure energetic neutral atoms, ions, and electrons—recorded very small dust grains [8000 to 40,000 unified atomic mass units (u), or roughly 1- to 3-nm radius] in two sensors. At 3000-km altitude, a peak rate of ~300,000 counts s–1 was detected by one sensor as Cassini crossed the equatorial plane. At lower altitude (1700 to 2000 km), a second sensor recorded positively charged dust in the upper atmosphere and ionosphere over a size range of ~8000 to 40,000 u (~1 to 2 nm, assuming the density of ice).

Consistent with this observation, larger dust in the 0.1- to 1-µm range was detected by the Cassini Dust Analyzer and the Radio and Plasma Wave Science instrument.


We modeled the interaction of dust with the H and H2 exospheric populations known to populate the gap. Collisions between small dust grains and H atoms provide sufficient drag to de-orbit the dust, causing it to plunge into the atmosphere over ~4 hours. The analysis indicates that at least ~5 kg s−1 of dust is continuously precipitating into the atmosphere. At 3000-km altitude, the dust is distributed symmetrically about the equator, mostly between ±2° latitude with a peak density of ~0.1 cm−3. On the wings of the distribution, consistent with ring rain transport along the magnetic field, almost all of the dust was observed to be charged. At the 2000 to 1700 km altitude, the dust has reached a diffusive terminal velocity and, although showing some bias toward the equator, is ordered mostly by a scale height of ~180 km in altitude. The most probable source for this dust population is the innermost bright ringlet of the D-ring, known as the D68 ringlet.

We predict that this kinetic process generates a highly anisotropic neutral hydrogen population, concentrated near the equatorial plane with periapses between ~4000 and 7000 km, and apoapses ranging to as high as 10 Saturn radii, with a small fraction on escape trajectories.

Ring dust.

(Top left) Data and model fits for the equatorial dust population near 3000-km altitude for three orbits through the D-ring gap. HV, plate detector high voltage. Red line uses left scale (percent). (Bottom left) Dust counts (blue) from ~2000 to 1700 km (modulated by sensor energy/charge steps, green), ordered by altitude and latitude, consistent with diffusive transport. (Right) Model trajectory of a dust particle in three frames of reference, as collisions with exospheric hydrogen degrade its velocity. Saturn has been shrunk to expand the gap for clarity.


The sizes of Saturn’s ring particles range from meters (boulders) to nanometers (dust). Determination of the rings’ ages depends on loss processes, including the transport of dust into Saturn’s atmosphere. During the Grand Finale orbits of the Cassini spacecraft, its instruments measured tiny dust grains that compose the innermost D-ring of Saturn. The nanometer-sized dust experiences collisions with exospheric (upper atmosphere) hydrogen and molecular hydrogen, which forces it to fall from the ring into the ionosphere and lower atmosphere. We used the Magnetospheric Imaging Instrument to detect and characterize this dust transport and also found that diffusion dominates above and near the altitude of peak ionospheric density. This mechanism results in a mass deposition into the equatorial atmosphere of ~5 kilograms per second, constraining the age of the D-ring.

Particle influx from Saturn’s rings to its upper atmosphere, termed “ring rain,” was invoked to account for ionospheric electron densities measured by the Voyager spacecraft, which were lower than expected from solar extreme ultraviolet ionization (1, 2). Ring rain is expected to occur at altitudes below ~0.5 RS [where 1 RS (Saturn radius) = 60,268 km]. In this region, magnetic drag on charged grains can slow them below local Keplerian orbital velocity, and the gravitational force dominates over the electromagnetic and centrifugal forces that otherwise keep such small (~100-nm- to 1-μm-sized) charged dust grains in orbit. The charged grains are then pulled along magnetic field lines into the planet’s atmosphere (36). The water-based ice grains in the atmosphere absorb electrons, thereby reducing the ionospheric electron density, primarily at latitudes that lie at the atmospheric feet of magnetic field lines connected to the rings (2). This reduction has been remotely observed from ionospheric occultation measurements (7, 8) and from ground-based telescope observations (6). Simulations of ring rain suggest that charged dust precipitation should be greatest at the magnetic foot-points of the main rings (>15° from the equator), whereas the ionospheric electron densities are observed to be lowest near the equator (810). The ring rain mechanism does not predict this, because the equator is both close to the region of highest ionizing solar flux and inaccessible to the charged dust that would follow the magnetic field.

Dust particles have previously been detected with Cassini instruments—including the Low Energy Magnetospheric Measurements System (LEMMS) instrument, the Cassini Plasma Spectrometer (CAPS) electron spectrometer (ELS) and ion mass spectrometer (IMS), and the Radio and Plasma Waves Spectrometer (RPWS)—which all detected dust in the plumes emanating from the moon Enceladus (1113). The ELS also detected large negative molecules in the ionosphere of large moon Titan (14).

We report in situ measurements of ring-material precipitation measured during the final, proximal orbits of the Cassini spacecraft. Referred to by rev (orbit) number, these orbits took the spacecraft very close to the planet within the D-ring. Of the 293 revs during the Cassini mission, the final 23 were the proximal orbits. Our measurements were made during revs 273, 284, 286, and 288 through 293 with two subsystems of the Magnetospheric Imaging Instrument (MIMI) (15), namely, the ion and neutral camera (INCA) and the charge energy mass spectrometer (CHEMS).

Dust measurements in the gap between Saturn and its D-ring

Only three of the proximal orbit equator crossings were useful for INCA measurements, because the dust ram velocity vector (the vector sum of prograde circular Keplerian orbital motion and the spacecraft velocity) must be within INCA’s 90°-by-120° field of view (FOV). The INCA observations during these three orbits—at 2850-, 3000-, and 3100-km altitude—showed distinct signals, which we attribute to dust (Fig. 1). INCA records particles both when they enter the detector at an initial microchannel plate (MCP, a start count) and when they strike a final MCP (a stop count) (16). The dust signals are inconsistent with known backgrounds such as penetrating electrons or sunlight, both of which produce approximately proportional responses on the INCA start and stop MCP detectors, whereas the dust signal is exclusive to the INCA start MCP. All three of these equator crossings also had similar (though not identical) features on the outbound orbital egress. For revs 273 and 284, the count rates dropped much more quickly on the egress than they rose on the ingress, owing to a rotation of Cassini about the spacecraft’s high-gain antenna (z) axis, which is a planned maneuver used to facilitate calibration of the magnetometer for other studies (17). Just after the peak counting rates were reached (at the equator), the spacecraft rotation took the dust ram vector out of INCA’s FOV, so the rate dropped quickly as the dust stream no longer entered the instrument. On rev 286, the response was much more symmetric because Cassini maintained a nearly fixed attitude relative to the dust ram direction, but there is a jump in the rate curve on the egress, where the start rate abruptly nearly doubled before continuing its decline. This jump was produced by a sudden drop in the INCA charged-particle deflection plates voltage from 2 to 0 kV, which was caused by the immersion of the spacecraft in ionospheric density high enough to short-circuit the deflection plates as Cassini approached its minimum altitude during this rev. To exclude ions when INCA is in its energetic neutral atom imaging mode, these deflection plates were biased at 2 kV for rev 286 but were held at 0 V (ion mode) during revs 273 and 284 (16).

Fig. 1 INCA detector counting rates.

(A to C) Raw INCA start and stop counting rates for Cassini equator crossings on revs 273 (A), 284 (B), and 286 (C). As Cassini traverses through the equatorial region during the proximal orbits, the INCA start rates climb, owing to increased dust near the ring plane. Counting rates on the start and stop plates for these three proximal-orbit periapses that had the ram vector in the INCA FOV show evidence of dust, evident as enhancements in start rates (red traces) without changes in the stop rates (blue traces) as Cassini crossed the equator. In each instance, the start rate reaches its maximum at the equator crossing time.

In this near-equatorial region containing Saturn's inner radiation belt, the counting rates on INCA’s stop MCP are driven primarily by penetrating particles. The inner radiation belt (18) is composed primarily of very-high-energy protons (≥300 MeV) that easily penetrate the walls of the INCA sensor and produce counts in the detector proportional to both the start and stop areas. The correspondence between the stop rates and the fraction of the start rates that are driven by penetrators can be used to predict and then subtract the contribution from penetrators to the start rates, producing a background-corrected version of the start rates that we attribute to dust impacts (16).

Figure 2 shows a plot of dust count rate as a function of spacecraft planetocentric latitude for the three different equator crossings. The rates have been background corrected (16) but otherwise are the raw rates. The count rate curves for revs 273 and 284 are almost identical, despite being taken more than 2 months apart. From this, we infer that the dust population is constant over this time and altitude scale. For revs 273 and 284, no voltage was applied across the INCA deflection plates, so the dust is counted independent of its charge state. We performed a trial-and-error fit to the derive a count rate function (CRF) for these dust count profiles, using the sum of two Gaussian distributions: CRFTotalDust = A exp{½[(λ – 0.17°)/σA]2} + B exp{½[(λ – 0.17°)/σB]2}(1)We found that the constant that scales the peak amplitude of the equatorial counting rate (A) = 316,000, the constant that scales the amplitude of the higher latitude wings of the distribution (B) = 6000, the Gaussian width of the equatorial peak (σA) = 1.4°, and the Gaussian width of the higher latitude component (σB) = 5°; λ is the planetographic latitude. The small 0.17° offset may derive from the known northward offset of the magnetic equator from the spin equator of Saturn, although it is formally only a parameter needed for a good fit. Uncertainties in A and B are ~5 and ~10%, respectively; uncertainties on σA and σB are ~5%, and the uncertainty on λ is ~±0.2°.

Fig. 2 Dust counting rates and analysis.

Dust counting rates (left scale) for revs 273 (light cyan), 284 (green), and 286 (blue) after background subtraction, plotted against Saturn latitude. The spacecraft approached the equator from the north, so time runs from right to left in this representation. The rates for revs 273 and 284 have been truncated at the point where the dust ram vector begins to leave the INCA FOV. Also shown are model fits to the rev 273 and 284 data (dashed line, total dust fit); the fitted ingress and peak of the rev 286 data (dashed-dotted line, neutral dust fit); and curves derived from those models. The charged dust counts (purple) curve represents the difference between the total dust fit and the rev 286 measured dust flux, which we take to be due to dust that is both charged and in the E/q range of less than 180 to 200 keV/e. The red curve at the bottom represents the percent (also read on the left scale) of the total dust contributed by the charged dust counts. All of the curves (with the exception of the red one) can also be read from the right-hand scale, which converts them to the measured dust density (see text). The label “HV drops” indicates that the deflection plate high voltage (HV) dropped from 2 to 0 kV at this point; “HV returns” indicates that over this latitude location, the deflection plate HV recovered from 0 to 2 kV. cts, counts.

For rev 286, the voltage across INCA deflection plates was 2 kV. This setting removed all charged dust up to about 80 or 90 keV per elementary charge (keV/e) and most of the charged dust up to roughly 200 keV/e. Given the combined velocity of the spacecraft and the dust, the impacts on the INCA start foil are at about 31 km s−1. By using this velocity, 200 keV/e is equivalent to a particle mass of just under 40,000 unified atomic mass units (u). Because the particle velocity is the same for all particles dominated by the spacecraft velocity, particle mass scales linearly with energy. The ingress for the rev 286 distribution was also fitted by using two Gaussian components (Fig. 2), as below:CRFNeutralDust = C exp{½[(λ – 0.17°)/σC]2} + D exp{½[(λ – 0.17°)/σD]2}(2)where C = 250,000, D = 1550, σC = 1.22°, and σD = 2.15°. The definition and uncertainties for these parameters are similar to those for Eq. 1.

During the egress on rev 286 toward south latitudes, Cassini was moving to lower altitudes (the minimum altitude was reached at about −5° latitude). When the ionospheric density short-circuited INCA’s deflection plates, the voltage dropped suddenly from 2 to 0 kV and resulted in the entry of the charged dust that had been excluded by the deflection system. This accounts for the abrupt increase in the dust counting rates by about a factor of two, as seen in Fig. 2. The resulting count rate jumped up to the total dust fit curve, confirming the constancy of the dust population for all three revs and giving us additional confidence in our approach to estimating the fraction of the population that is both charged and below the INCA deflection system maximum sensor energy/charge (E/q) limit. The INCA deflection system removed charged dust independent of its sign—both negative and positive dust are removed with equal effectiveness. Therefore, we cannot determine from these data what proportion of the charged dust is positive or negative, only the fraction that is charged and has E/q ≤ ~200 keV/e, or, for this velocity, ~40,000 u/e. Previous work has shown that, in this size range, grains will acquire only three likely charge states, −1 e, 0, or +1 e, with the most probable charge state at higher altitudes being neutral [e.g., (13, 1921)], and the charging times are hours to days (19, 21, 22).

Near the equatorial peak, only 20% of the dust measured is charged, but that fraction rises to 100% by ±6° latitude. This can be attributed to different dynamics for the neutral and charged components. Charged dust in this size range is constrained by the magnetic field, so it cannot fall radially downward but instead moves to lower altitude along the magnetic field (15). This process shifts an equatorial dust source to off-equatorial latitudes, which we invoke to explain the 100% charged-dust population detected in the higher-latitude wings of the total dust distribution. The absence of neutral dust in this higher-latitude population reflects both that this charged population is all less than 40,000 u (otherwise the INCA deflection system would not have removed it in rev 286) and that at this altitude, the drag-related transport time for a grain from the D-ring to the 3000-km altitude is much shorter than the time for a grain to change charge state. Because this charged-dust transport process has been studied previously (3, 4) and this higher-latitude component forms ~1% of the equatorial peak, we hereafter focus on the neutral component, which feels no electromagnetic forces, and thus its transport is dominated by gravity and collisions with the atmosphere.

Figure 2 includes a conversion of the count rate into density. The primary assumption in calculating the density is that the efficiency for dust detection by INCA is 100% above some minimum energy that can penetrate the INCA start foil. The foil has an areal mass density of 10 μg cm−2, so we estimate this energy to be about 50 to 60 keV. At the dust ram velocity, this is probably a valid assumption, because the penetration of the foil and the generation of a plasma by the impact of the dust with the foil is almost unavoidable. Under this assumption, the density (N) is simplyNDust = CRDust/(VDustRamAFoil)(3)where CRDust is the INCA start counting rate after background subtraction, given by Eqs. 1 or 2; VDustRam is the dust ram velocity (~31.5 km s−1); and AFoil is the start foil area, corrected for the angle between the ram velocity and the normal to the foil and corrected for the transmission of the foil support grid. The foil area is 2 cm2, the angle from the foil normal ranges from about 50° to 55° for the attitude configurations near the equator crossings, and the grid transmission is 82%. The corrected foil area then ranges between 1.05 and 0.94 cm2, so we adopt 1.0 cm2 for this calculation. The peak density at the equator is then NDust = 0.1 cm−3 for all three revs, despite the fact that Cassini’s equatorial altitude varied between 2850 and 3100 km. As we show below from dust flux–continuity considerations, this is the altitude range where the transport transitions from nearly free falling to collision dominated—where the vertical velocity becomes constant and density reaches a local minimum—so density remains approximately constant at these altitudes.

Dust-grain trajectories

In Fig. 3, A and B, we show the angular distribution of the dust from the two equator crossings during which Cassini was rolling about the spacecraft z axis. The phase of the roll angle was identical for both revs 273 and 284 to optimize ionosphere and neutral atmosphere observations by the Ion Neutral Mass Spectrometer (INMS) (23) at the lowest altitude in the orbit, which was reached at about −5° latitude on each rev. Thus, the position of the INCA FOV varied with latitude in the same way over both revs. A comparison of the measured counting rates with the extrapolated fit to the ingress and peak reveals a close match for both revs until about a minute past the peak, when the signal begins to deviate from the fit. As the angular distribution of the dust that is ramming into the instrument (from the combination of spacecraft and dust orbital motion) begins to cross outside the edge of the INCA FOV, the counting rates drop quickly. The ratio between the data and the fit (F10, where F10 = 10 × CR/CRF) drops from ~1 to ~0 as the dust leaves the INCA FOV (Fig. 3, A and B). A similar signature appears at the onset of the signal, about a minute after the start of each plot, but because the dust density is much lower, the statistics on the rise are less well characterized and more subject to background subtraction uncertainties. The angle αKeplerian that the edge of the INCA FOV makes with the calculated ram vector corresponding to the local circular Keplerian motion is also plotted in Fig. 3, A and B. If the dust velocity was centered on the value corresponding to circular Keplerian motion, then the point at which the ratio of the measured dust signal dropped to 0.5 relative to the fit would be centered at αKeplerian = 0. Instead, the ratio drops to ~0 at αKeplerian = 0. To determine the angular distribution of the dust ram directions, we differentiate the curve describing the cutoff with rotation, F10. To do this, we plot F10 against αKeplerian and fit a polynomial to the resulting points (Fig. 3C). We then differentiate the polynomial and plot that function (normalized) against αKeplerian (Fig. 3D). The resulting ram angle distribution has a peak at about 5°, dropping to near zero at 0° and ~14°. The (azimuthal) velocities that would result in this distribution range from 100% Keplerian at 0° to 85% Keplerian at 5° to 72% Keplerian at 14° (the distribution is not symmetric about its peak).

Fig. 3

Determination of dust angular distribution. (A and B) For rev 273 (A) and rev 284 (B), the red curve is a double (dbl)-Gaussian function fit to the dust count rate data (left scale); the blue curve is the dust count rate measured by INCA (left scale); the black curve is the ratio of the INCA dust count rates to the fit, multiplied by 10 (F10 = 10 × CR/CRF, right scale); and the cyan curve is the angle between the sharp edge of the INCA FOV and the calculated ram vector, assuming circular Keplerian dust (αKeplerian), as the spacecraft rotates about its z axis. (C) The blue points show F10 (for rev 284; rev 273 is essentially identical) plotted against αKeplerian; the red curve is a polynomial fit to F10 versus αKeplerian. (D) The (normalized) differentiation of the polynomial fit function is shown, where rev 273 is in red and rev 284 is in blue. This represents the angular distribution of the dust ram response relative to αKeplerian, showing that rather than being centered on αKeplerian = 0, as would be expected for dust in circular Keplerian motion, the dust ram is centered 5° away from zero.

At this altitude, any neutral dust particle with velocity less than about 98.5% local circular Keplerian (virtually all of the neutral particles measured by INCA) has an orbital periapsis below the top of Saturn’s atmosphere (defined as the height at which the atmospheric pressure is 1 bar). Regardless of its subsequent collision history, any such dust particle is destined to precipitate into the atmosphere. Any additional collisions experienced on the way down simply determine where, not whether, it will precipitate. We consider the effect of collisions below.

Entry of dust into Saturn’s atmosphere

We have modeled the trajectories of (uncharged) Keplerian dust grains starting just above the altitude of the D68 ringlet in Saturn’s D-ring, as they collide with atoms (primarily H at the highest altitude, with H2 and He becoming important below 4000 to 5000 km). We track their changes in velocity due to the collisions and the effects of gravity. Figure 4 shows the results of this calculation for a specific dust grain size, 19,300 u, which corresponds to ~100 keV at the spacecraft ram velocity. The collisions with H, H2, and He are shown, based on an atmospheric model (24), with H2 densities adjusted to conform with in situ measurements by INMS (25). We sum the collisions with H, H2, and He, weighted by their atomic masses, by modeling collisions with a species of mass 2 u for H2 and scaling the number of collisions with H and He by one-half and two, respectively. The horizontal component of the velocity is referenced to the atmospheric corotation velocity—that is, if the velocity drops to zero, the grain is no longer moving with respect to the rest frame of the atmosphere. The vertical velocity peaks at about 2700-km altitude, below which collisions modify its velocity profile until about 1700 km (depending on particle radius) where the particle has reached its terminal velocity, and its transport becomes more fully diffusive.

Fig. 4

Collisions of dust grains with atmospheric atoms and resulting velocities. Cumulative number of collisions (left scale, solid curves) between a dust grain (19,300 u and 1.8-nm radius, in this example) and Saturn’s atmosphere, as a function of altitude above Saturn’s 1-bar level. At higher altitudes, collisions with H (thin, black line) dominate the horizontal collisions. Below ~3300 km, H2 collisions (blue line) dominate. The two dotted curves show the velocity components (horizontal and vertical, right axis scale) of the grain as it is slowed by collisions with atmospheric atoms. Vertical collisions (red line) become important below ~2800 km, where the vertical velocity profile begins to plateau.

From this calculation, we also determine the vertical velocity for the dust as a function of altitude. The vertical velocity peaks just below 3000-km altitude at about 2.5 km s−1. By using this velocity, a dust density of 0.1 cm−3, assuming an average grain size of 15,000 u (~2.5 × 10−23 kg) for the INCA measurements and integrating over an area between ±1° latitude and over the circumference of Saturn at 3000-km altitude, we find (2.5 × 10−23 kg) × (0.1 cm−3) × 2π × (63,280 × 105 cm) × 2° × (π/180°) × (63,280 × 105 cm) × (2.5 × 105 cm s−1) = ~5.5 kg s−1 mass flux into the atmosphere. This is the flux for the particles that INCA can measure, that is, those larger than approximately 10,000 u. As described above, the upper limit is poorly constrained but is likely less than 40,000 u. Both the assumption of the average grain size and the ~10% uncertainty in average vertical velocity contribute to uncertainties as high as 50% in the mass flux estimate. Given the likelihood that the size distribution of the dust is probably a power law with a slope of at least 3 and possibly as steep as 6 (26), the total mass flux of all dust particles is probably much higher.

This mass flux (~1.7 × 10−13 g m−2 s−1) is about an order of magnitude higher than that estimated for 0.5- to 100-μm interplanetary dust into Saturn’s atmosphere (27). However, there are some important differences. First, this is D-ring dust, and this rate of influx may be a relatively recent development, beginning in about 2014 with a disruption event in the D68 ringlet (25, 26). Second, this D-ring dust comprises much smaller grains than those previously modeled (27). Third, the dust velocity relative to the atmosphere at Saturn is much lower than for interplanetary dust, either for Saturn or for Earth, because it is falling from a nearby Keplerian orbit rather than from infinity. The average neutral-atmosphere atomic mass is much smaller at Saturn than at Earth. This, together with the lower relative velocity, means that these particles do not ablate as interplanetary dust does in Earth’s upper atmosphere and so do not break up or directly affect the chemistry of the atmosphere in the same way as interplanetary dust does in Earth’s mesosphere. Finally, this influx is limited to the equatorial region [although charged dust flux from the main rings also likely exceeds the interplanetary dust flux (28)].

The distribution in latitude measured by INCA arises directly from the angular scattering caused by the collisions that degrade the grain velocities. The grains—which we argue below begin their descent from the inner edge of the D-ring, most likely from the D68 ringlet [see, for example, (29)]—are initially confined very close to the equator, likely within a few kilometers or less. The angular scattering due to H atoms during their deceleration will result in a random walk relative to the equatorial source, gradually spreading the distribution as it descends.

Figure 5 shows the model dust density profile (for the size range to which INCA is sensitive) as a function of altitude, normalized to the INCA measurement of 0.1 cm−3 at 3000 km. The continuity calculation does not include latitudinal diffusion, so the predicted densities below the altitude of the INCA measurements are, in that sense, upper bounds. Within the grain size range measured by INCA, the predicted density dependence can be modeled using a scale-height formalism at the lower altitudes (see below), with a smaller grain size distribution resulting in higher densities at the lower altitudes. Because the dust grain size is only loosely constrained by the INCA measurements, the scaling by the observed number density of 0.1 cm−3 at 3000-km altitude may represent a size range that is larger than that appropriate to the MIMI CHEMS observations (below) and much larger than the molecular grains (25). Because we have no direct measurements of neutral H, the model values for H are not well constrained. In Figs. 4 and 5, we used the higher-density model for H (24), which has a considerable effect on the collision history. Because by the 3000-km altitude the collision dynamics are dominated by H2, reducing the H to much lower (0.1%) than the predicted values (24) renders the impact of collisions with H negligible, yet only alters these velocity profiles at and below 3000 km slightly, with the horizontal velocity at 3000 km changing from 9.3 km s−1 in the atmosphere frame to 10.2 km s−1, a ~10% difference. Thus, the observations do not strongly constrain the H density. Such a reduction in the H density would have an effect on the higher-altitude dust density profile predicted by the model (Fig. 5), leading to a stronger upward gradient in dust density and higher dust densities above 3000 km than the model currently indicates. Because we have no direct measurements of the equatorial dust density above 3000 km, we cannot presently constrain the H density with much precision.

Fig. 5 Predicted altitude profile of dust density based on dust flux continuity.

The blue line uses the continuity of the dust grain flux determined from the calculated vertical velocities combined with the dust density (in the size range measured by INCA) at ~3000 km (0.1 cm−3) to predict the densities above and below the altitude of the INCA measurements. The blue diamond shows the altitude of the D68 ringlet, which we suggest is the primary source for the dust observed by Cassini. The upturn at low altitude depends on the dust radius and material density. Atmospheric mass density is plotted in red on the right scale. The decreasing dust density from high to low altitude reflects the increasing vertical velocity as the particles deorbit, until, at ~2800 km, the atmospheric mass density is sufficient to begin decelerating the falling particles, transitioning the transport to diffusive. For the 19,300-u particle modeled, the vertical transport is dominated by collisions below ~2200 km, leading to the density being well characterized by a scale height at those altitudes, as was observed (close fit to the dotted green curve, based on a 180-km scale height).

Influence on Saturn’s atmosphere

This influx of dust grains and their collisions with the very high atmosphere has implications for the atmospheric structure both below and above the altitudes sampled by Cassini. The effects on the ionosphere are substantial (25), and there are also effects on the high-altitude exosphere. In slowing from Keplerian speeds to corotational speeds, each dust grain undergoes thousands of kinetic collisions with H atoms and H2 molecules at altitudes where these constituents’ mean free paths are very long. Each such (assumed elastic) collision, if effective in slowing the grain, is also very effective in speeding up the colliding atom or molecule. For example, a head-on collision between a Keplerian (velocity ~24 km s−1) grain of 10,000 u and a H atom moving at the 10–km s−1 corotational speed in the same direction (plus some small thermal velocity) leaves the grain moving at 23.997 km s−1 and the atom moving at 38 km s−1. The latter is above the escape velocity for Saturn. This implies that this process will produce perhaps thousands (at least hundreds) of high-velocity H atoms for every dust grain that precipitates into the atmosphere. However, these atoms do not have an isotropic distribution. Because the dust is confined to the equator, the collisions which impart the greatest velocity are (nearly) head-on, and the dust motion is mostly horizontal, the atoms that receive high speed boosts will be moving nearly tangentially with the top of the atmosphere and so are likely to continue to very high altitudes without additional collisions. Their trajectories are expected to be roughly parallel to the equator, so they should constitute a quasi-equatorial population of neutral hydrogen with periapsis at the top of the atmosphere and apoapsis at several RS. Some H2 may also achieve similar trajectories, although the probabilities are lower because of the lower H2 densities at the highest altitudes where escape without collision is most likely.

This energetic, anisotropic H and H2 population is very unlikely to be lost to collision with the rings, which constitute a very thin target for atoms or molecules launched from the equator. We suggest that this is the mechanism behind the equatorial hydrogen plume detected by the Cassini Ultraviolet Imaging Spectrograph (UVIS) instrument (21).

Dust at lower altitudes

Whereas the ram vector entered INCA’s aperture on the revs discussed above, for several later revs at lower near-equatorial altitudes, the dust ram vector entered the CHEMS sensor aperture. The response of CHEMS to dust grains is similar to that of INCA (16) after the grains’ transit of CHEMS’ electrostatic analyzer (ESA). Dust particles with the requisite E/q to transit CHEMS’ ESA reach the start foil, vaporize, and, if they are of sufficient energy, produce a plasma cloud at the foil exit plane that triggers the start MCP only. Electrons from this plasma trigger the start MCP but not the stop MCP. Measurements during the proximal orbits show that only starts, but no discernable stops, are generated as the dust particle is destroyed by the impact with the start foil.

During the close approaches when dust was detected (revs 288 through 292, as well as rev 293, the final plunge by Cassini into Saturn’s atmosphere), CHEMS telescope 2 (T2) was pointed in the corotation ram direction, defined as the spacecraft velocity added to the rotation velocity of the planetary atmosphere. CHEMS telescopes 1 (T1) and 3 (T3) were centered at ~60° above and below the ram direction. This spacecraft attitude was designed to provide optimum pointing for the INMS instrument (25) but also proved beneficial for CHEMS because its T2 aperture was co-aligned with the INMS FOV. Although the CHEMS FOV faced the ram vector during the rolls in revs 273 and 284, that geometry was only reached well past the equator where dust densities were lower and the radiation backgrounds higher at the higher altitude of those periapses.

The start rates for the ram-facing T2 telescope produced a well-defined signal (Fig. 6). The weak signal seen in T1 is not well understood (16). There was no discernable signal in any of the CHEMS stop rates or solid-state detector rates, other than the backgrounds from penetrating particles. This region is devoid of any trace of energetic charged particles in the design range for CHEMS or INCA (18), and at this altitude, the intensity of the energetic radiation belt particles is also very low, resulting in background low enough to measure this relatively weak signal. We have inferred that the response in T2 was produced by positively charged grains with E/q > 40 keV/e, up to the maximum measurable value of 220 keV/e. Similar responses were recorded by CHEMS during transits of the icy dust plumes emanating from the south polar region of Enceladus and in measurements by the Cassini CAPS instrument (13). See (16) for additional discussion of the CHEMS response to dust.

Fig. 6 CHEMS dust counting rates for Rev 288.

(A) Raw start counts per second from the CHEMS T1, T2, and T3 telescopes as a function of latitude for the periapsis pass on rev 288 (time runs from right to left). The strongest signal is in T2, which was oriented to detect positively charged dust grains in the combined spacecraft and corotation vector ram flow. The square wave pattern is created as the CHEMS ESA steps through its E/q selection voltages—grains selected in the lower end of the CHEMS E/q range do not have sufficient energy to penetrate the start foil and generate start electrons. On this orbit, the inner radiation belt (18) creates background counts from high-energy penetrating charged particles, especially at latitudes beyond about 10° from either side of the equator. (B) Using the T3 start counts to represent background (bkgd) from penetrating particles and scaling the T3 rates to the same level as the observed backgrounds in T2, we subtract this background from the T2 rates, leaving the dust counts as signal. Also plotted are the spacecraft altitude (black, right axis) and the corresponding ESA HV step for each measurement (green, left axis), with the highest-energy step number, 31, corresponding to 220 keV/e.

This energy-per-charge threshold, combined with the assumption that the grains are singly charged and that their energy is dominated by the ram velocity, indicates that the grains have masses of 8000 to 40,000 u. Assuming the density of water ice (this does not assume the grains are ice, as many organics have similar densities) implies radii of ~1.2 to 2.3 nm.

Figure 7 shows the CHEMS data for the six revs for which the pointing allowed CHEMS to face the dust ram vector in T2 and at low enough altitude so that the dust signal was above the radiation backgrounds in the CHEMS start rates. As before, we fitted the data with an exponential function of altitude, which results in a scale height of approximately 180±30 km. For revs 288, 289, and 290, an additional Gaussian in latitude is applied as a multiplying function to achieve an adequate description of the CHEMS dust profiles, F(z, λ) = F0 exp[−(zz0)/HD] exp{½[(λ − λ0)/σλ]2}(4)where z is altitude above the 1-bar region, HD is a scale height, and λ and σλ have the same meaning as in Eq. 1. The scale height HD is held constant at 180 km, whereas the σλ parameter and amplitude F0 were allowed to vary in this latitude-dependence fitting process. The z0 was fixed at 1700 km, and λ0 was set to 0.17° on the basis of the INCA measurements but is consistent with zero. Over several orbits, the σλ parameter changed from ~5° to 20° (Fig. 7). The profiles evolved from requiring an equatorial bias to being independent of latitude. The peak amplitudes of the profiles varied by up to a factor of two from one rev to the next, a much larger variability than that observed near 3000 km by INCA. The uncertainties in these parameters are similar to those indicated for Eqs. 1 and 2.

Fig. 7 CHEMS dust counting rates for six revs.

Charged dust counting rates for all six revs for which CHEMS measured the charged dust signal. These include rev 293, the final plunge into Saturn’s atmosphere, which reached higher rates at the lowest altitude and stopped when communication with the spacecraft was lost. Only the data at and above ESA step 19 (~40 keV/e, or about 8000 u/e) are shown, to reduce confusion introduced by noise or low-efficiency measurements in the lower ESA steps. Two model fits to the data are shown for each rev except rev 293. In green are exponentials in altitude with a common scale height of 180 km and varying amplitudes to match the measured rates. In black are the same exponentials but multiplied by a Gaussian in latitude centered at the equator with width σLat as indicated (with amplitude again adjusted to fit the data). Revs 291 and 292 are consistent with, but do not strongly constrain, the equator-centered Gaussian. Rev 293, with the steep entry into the atmosphere, did not constrain equator weighting. The first three revs favor the Gaussian fit. A small constant level was added to each fit to accommodate the wings (ingress and egress).

CHEMS measured positive dust particles near 1700-km altitude, which we compared to the dust measured by INCA near 3000 km. INCA made measurements for three revs, and these measurements were highly repeatable, with variations in density <5% and identical spatial and velocity distributions. Although sub-Keplerian, the velocities measured by INCA were well above the atmospheric (corotation) velocity and confined to a region close to the equator. CHEMS has a smaller 4° acceptance angle in the direction discriminating between ram velocities consistent with corotation or Keplerian and so would have detected nothing if the dust velocity differed from the corotation speed (for which the spacecraft attitude was tuned) by more than about 10%. This, and the observation that altitude dependence is well characterized by an exponential with a scale height (~180 km) similar to that of the atmosphere itself, leads to the conclusion that at this altitude, the transport is diffusive in nature, at least for positive dust grains.

We infer that the dust seen by INCA at high altitude is the source for the dust seen by CHEMS at lower altitude (with the original source being the D-ring, and likely the D68 ringlet). Why does the distribution vary in apparent density at the lower altitude? Continuity of mass flux suggests that it should be fairly constant, at least when integrated over all latitudes. Clearly the transport changes appreciably between 3000 and 1700 km, and asymmetries in the neutral atmosphere density relative to the equator may be involved. With more solar input north of the equator during this part of Saturn’s year, the atmospheric scale heights may be higher in the north than in the south. Winds in the thermosphere may further alter the latitude density profiles of H and H2. Although there is strong evidence for the low-altitude dust being biased toward the equator on revs 288 and 289, that signature is less evident or even absent in later revs.

Our kinetic, collisional model of the dust transport qualitatively matches the observed densities, latitude distribution, and velocities but does not quantitatively match. We have so far treated the dust as neutral in the transport calculations; however, this is not an adequate assumption for the charged component of the dust. As discussed above, for the relevant size range (~1- to 3-nm radius), the most probable charge state at higher altitudes is neutral and the charging times are hours to days. Even deeper in the ionosphere, at the 1700-km level, this is still the case, although the charging times (especially the times to charge negatively) decrease to less than an hour.

Comparing dust at different altitudes

At higher altitudes (3000 km and above), any charged dust particles are picked up by the corotational electric field felt by the particles owing to the difference between their Keplerian velocity and the corotation velocity, which forms the rest frame of the magnetic field. The combined gravitational and electromagnetic forces quickly destabilize the dust particles orbits, so they precipitate along the magnetic field into the atmosphere, away from the equator (3032). The INCA measurements appear to measure this population, as the higher-latitude wings of the equatorially confined population of dust is nearly 100% charged, that is, very likely the charged particles that have recently converted from neutrals and are in the process of precipitating along the magnetic field.

At the lower altitudes, collisions with the neutral atmosphere will damp electromagnetic effects, and charged dust is not expected to move guided by the field for any great distance. However, the dust-grain charge states may still play a role in their transport. Our transport calculations predict reduced, but still substantial, vertical velocities for the neutral dust. Having been scattered to the point that their net horizontal velocity is zero in the atmosphere rest frame, they still are influenced by the force of gravity. At these lower altitudes of ~1700 km, the atmospheric density is sufficiently high that the dust particles no longer free fall but should reach a terminal velocity. That terminal velocity is still on the order of 1 to 2 km s−1 in the altitude range from 1800 to 1700 km where diffusive spatial distributions ordered by scale height are observed. Using the dust flux continuity calculations (Fig. 6), we predict a dust density of ~1 or 2 cm−3 at ~1700 km, whereas INMS measurements (25) suggest dust densities of several thousand per cubic centimeter. We cannot address this directly with the CHEMS data because this instrument only measures the positively charged fraction of the dust.

Charging can also slow the dust as it falls through the atmosphere. Charged dust not only experiences the drag force of kinetic collisions with atmospheric atoms and molecules but will also be stopped by electromagnetic V × B forces, where V is particle velocity relative to magnetic field B. Previous studies of these forces on charged grains (33) were restricted to much larger grains, for which the ratio of electromagnetic to gravitational forces is very different. If sufficient dust grains are charged for sufficient lengths of time, much higher densities could build up. The charging times shorten as the ionospheric density increases, so a larger dust fraction may charge at low altitudes relative to the higher altitudes. Most of that charging will be negative, and CHEMS cannot measure such grains.

Comparison with other instruments

The MIMI-CHEMS measurements can be readily compared to those from the RPWS and INMS instruments, which made observations concurrent with those of MIMI during the closest approaches (25). RPWS obtained electron density, and INMS directly measured light ion densities of H+ and H3+ during closest approach.

Figure 8A shows the relatively close agreement between the light ion density and the plasma electron density farther from the equator (which also corresponds to higher altitudes, see Fig. 6) and the strong departure of the two densities near the equator (lower altitudes). In Fig. 8B, we take the difference between the electron density and the light ion density and overplot the CHEMS T2 dust counting rate for the E/q steps with high dust efficiencies, multiplied by an arbitrary constant. Charge conservation requires that the total ion density equal the total electron density Ne. However, near closest approach, Ne lies above the total light ion density of H+ and H3+ by nearly an order of magnitude. The missing ions are thought to be more massive molecules that are too energetic to be detected by INMS at this ram velocity (25). It is evident that the density of the positive dust grains measured by CHEMS is directly proportional to the heavier ion densities required for charge neutrality. The CHEMS measurement does not address this missing positive charge; the CHEMS counting rates—when converted to a positive dust grain density accounting for the CHEMS geometric factor; ESA passband, which is an estimate of efficiency; and the ram velocity—yield a dust-grain density of about 0.002 cm−3, which is far too low to account for the required positive charge density (>4000 cm−3 at the peak). But the observation that the positive dust-grain density scales linearly with the undetected ionospheric positive charge suggests a common transport path and a common source in the D-ring [compare (25)].

Fig. 8 Rev 288 MIMI-CHEMS, RPWS, and INMS results compared.

(A) Electron density (black curve) as determined by the RPWS instrument by using characteristic plasma resonances and light ion (H+ and H3+) density as measured by INMS. The difference between the two curves indicates the presence of a heavier ion constituent not measured by INMS under these conditions, which must be present to achieve overall charge neutrality. (B) The difference in the two densities shown in (A), with the CHEMS T2 rates (background corrected), multiplied by a constant and overplotted in blue dots. The correspondence between the two curves indicates a proportionality between the dust measured by CHEMMS and the plasma heavy ion population.

Summary and conclusions

Dust grains with masses from ~8000 to >40,000 u have been detected in situ inside Saturn’s rings. Modeling their distribution indicates that dust is precipitating into Saturn’s equatorial atmosphere directly from the inner edge of the D-ring and probably from the D68 ringlet. This dust population is tightly confined near the equator at higher altitudes (~3000 km), as directly observed by the MIMI-INCA instrument. At this ~3000-km altitude, the dust velocity distribution is measured to be sub-Keplerian speed (between circular Keplerian and about 70% Keplerian, peaking at 85% Keplerian). The dust population has a charged component that comprises ≥20% of the population in the equatorial plane, reaching 100% of the population on the higher-latitude (beyond ±~5°) wings of the spatial distribution, owing to spatial filtering of the original population by the Lorentz force.

The dust is precipitating through atmospheric drag, a process previously invoked for the interaction between Uranus’s ring and its atmosphere (34). At the highest altitudes, the nature of this drag is kinetic collisions between the dust grains and exospheric H. Many of these collisions (hundreds per dust particle) will result in H atoms with escape or near-escape velocities, in a specific direction (tangential to the atmosphere at the point of collision and close to the equator in direction) and at altitudes where the mean free path is very large. This is expected to produce a durable, high-altitude, equatorially confined escaping population of neutral H, as well as some H2 by the same process. An analogy of this population is sawdust ejected from a circular saw as it cuts through wood. We suggest that this scenario is consistent with the previously observed equatorial hydrogen plume (35).

MIMI-CHEMS measured the positively charged fraction of the dust at lower altitudes (~1700 km), where the densities are presumably much higher. The dust at this altitude is organized by a scale height, suggesting diffusion-dominated transport, unlike the dust observations at 3000 km. The source for this dust still appears to be near the equator. Kinetic transport calculations predict lower densities than are inferred for the dust population at this altitude (25). We argue that dust charging plays a role in slowing the fall of the dust through the atmosphere. The CHEMS densities of positively charged dust grains are shown to be directly proportional to the heavy positive-ion population that is inferred from the difference between the light ion density and the electron density measured at the lowest altitudes on revs 288 and 292.

Supplementary Materials

Materials and Methods

Supplementary Text

Fig. S1

Reference (37)

References and Notes

  1. Instrument descriptions and data analyses methods are available as supplementary materials.
Acknowledgments: We thank W. Ip, S. Kempf, and M. Horyani for helpful discussions. We thank L. Esposito for alerting us to a similar analysis of ring material precipitation at Uranus. Funding: This research was supported in part by the NASA Office of Space Science under task order 003 of contract NAS5-97271 between NASA Goddard Space Flight Center and the Johns Hopkins University. The research at the University of Iowa is supported by NASA through contract 1415150 with the Jet Propulsion Laboratory. The Swedish National Space Board (SNSB) supports the RPWS–Langmuir probe instrument on board Cassini. Author contributions: D.G.M. conducted the primary analysis of the MIMI, CHEMS, and INCA data and developed and ran the particle-atmosphere collision code. M.E.P. contributed to the understanding of the atmospheric interactions. D.C.H. provided expertise on the CHEMS instrument and its response to dust and contributed to reviewing and editing the text. J.F.C. provided extensive editorial services. J.H.We. contributed ideas and refinements. P.K. provided help and expertise in understanding the radiation environment and the background responses of the instruments. H.T.S. contributed expertise in understanding the responses of the CAPS and CHEMS sensors and understanding of neutral atmosphere interactions with dust. J.H.Wa. provided insights and helped to organize a workshop to analyze the proximal orbit measurements. R.P. provided detailed analysis of and graphics for the INMS sensor data that was used in Fig. 8 and the discussion of that data. H.-W.H. provided expertise in the transport of charged dust in the Saturn-ring-atmosphere system. J.-E.W., M.W.M., and L.Z.H. provided expertise on the RPWS–Langmuir probe data as well as on the ionosphere ion and electron densities. A.M.P. and W.S.K. provided analysis and insights into the ionospheric densities, as determined from the upper hybrid frequency, as well as discussions regarding the RPWS response to dust impacts at the ring plane crossings. Competing interests: The authors declare no competing interests. Data and materials availability: MIMI data are available from the Planetary Data System archive at We used the uncalibrated dataset from the period 9 May 2017 to 15 September 2017. The dust kinetic model is available at Zenodo (36).

Stay Connected to Science

Navigate This Article