Report

The Ionosphere of Europa from Galileo Radio Occultations

See allHide authors and affiliations

Science  18 Jul 1997:
Vol. 277, Issue 5324, pp. 355-358
DOI: 10.1126/science.277.5324.355

Abstract

The Galileo spacecraft performed six radio occultation observations of Jupiter's Galilean satellite Europa during its tour of the jovian system. In five of the six instances, these occultations revealed the presence of a tenuous ionosphere on Europa, with an average maximum electron density of nearly 104 per cubic centimeter near the surface and a plasma scale height of about 240 ± 40 kilometers from the surface to 300 kilometers and of 440 ± 60 kilometers above 300 kilometers. Such an ionosphere could be produced by solar photoionization and jovian magnetospheric particle impact in an atmosphere having a surface density of about 108 electrons per cubic centimeter. If this atmosphere is composed primarily of O2, then the principal ion is O2 + and the neutral atmosphere temperature implied by the 240-kilometer scale height is about 600 kelvin. If it is composed of H2O, the principal ion is H3O+ and the neutral temperature is about 340 kelvin. In either case, these temperatures are much higher than those observed on Europa's surface, and an external heating source from the jovian magnetosphere is required.

The Galileo spacecraft is in orbit about Jupiter, studying Jupiter and the Galilean satellites (Io, Europa, Ganymede, and Callisto). Of these, only Io was known to have an atmosphere, which was observed by a radio occultation (1) of Pioneer 10 in 1973 (2). Observations and theoretical considerations suggest that Europa may have an atmosphere that originates from frozen surface water ice, most likely produced by particle impact (3, 4). Therefore, any atmosphere must consist of some mixture of H2O, H, H2, OH, O2, and O. Early Pioneer 10 observations suggested an atomic oxygen column density of around 1 × 1013cm−2 (5), but these measurements may have been contaminated. Recent Hubble Space Telescope (HST) observations of the 1304–1356 Å lines of atomic oxygen were used to deduce an O2 column density of about 1.5 × 1015cm−2 (6). Recent Galileo measurements at Ganymede and Callisto indicated the presence of atomic hydrogen, with densities on the order of 1 × 104 cm−3(7), which might suggest similar abundances at Europa.

A search for an atmosphere on Europa was carried out when Galileo was occulted by Europa three times: on 19 December 1996 (E4) and on 20 and 25 February 1997 (E6a and E6b). The parameters defining the properties of these occultations are given in Table1. For a few minutes before and after the occultations, the S band (2.295 GHz, or about 13 cm wavelength) radio signal from Galileo traversed regions above Europa's surface in which one could observe the effects of refraction by an atmosphere, or more precisely, an ionosphere (a layer of ions and electrons produced in tenuous regions of the atmosphere by photoionization and magnetospheric particle impact), should one exist on Europa. Thus, the three occultations provided six opportunities to search for Europa's ionosphere. It was necessary to use the low-gain antenna to transmit the signal from Galileo, as the high-gain antenna failed to deploy properly. This meant that the signal level received by the Deep Space Net (DSN) stations at Canberra, Australia (E4), and Goldstone, California (E6a and E6b), was reduced by a factor of almost 1000 (30 dB). Also, only one frequency, S band, could be used rather than two coherently related frequencies at S band and X band (8.415 GHz, or 3.56-cm wavelength). The resulting signal-to-noise ratio, even when the 70-m-diameter antennas of the DSN were used, was on the order of 14 dB in a 1-Hz bandwidth. Five of the six occultation measurements were conducted in the one-way mode, with the downlink frequency referenced to the on-board ultrastable oscillator (USO) (8) having a stability of about one part in 1012, except for the E6a entry, which was done in the two-way mode in which the frequency transmitted by Galileo is referenced to a frequency standard at the DSN station.

Table 1

Geometry of the Galileo occultations by Europa, listing the date, distance from the limb, latitude, longitude, solar zenith angle (SZA), and the magnetospheric ram angle. The ram angle is defined as the angle between the vector pointing from the center of Europa to the “sub-ram” point and the one pointing from the center to the occultation point. Because all occultations occurred at near-equatorial latitudes, this angle can be approximated by the difference in longitude between the occultation point and 270°, which is the sub-ram longitude.

View this table:

The received signals were translated in frequency down to an audio frequency range, filtered in a bandpass of 0 to 2500 Hz, and digitized to 8-bit samples at a rate of 5000 samples per second. The digital data were processed with a digital phase-locked loop program analyzing 1-s data blocks to obtain time series of the frequency and signal strength. Removing the programmable local oscillator (PLO) (9) “steering function” results in a time series of “sky frequency,” which is what one would have observed without a PLO. This time series was compared to a time series of predicted frequency computed from a precise spacecraft ephemeris, and their difference forms the frequency residuals, which are the basic data used in determining the profiles of electron density above the surface of Europa. Ideally, these residuals should have a zero mean on the baseline, which is the portion of the data that is away from the influence of possible ionospheric refraction effects. In reality, because of drift in the USO, effects of the long propagation path through the interplanetary medium, and imperfect knowledge of the frequency transmitted by Galileo and of the spacecraft trajectory, this baseline has not only a nonzero mean but also a slope, which over periods of tens of minutes can be approximated by a linear frequency drift. The bias and linear drift in the residuals were removed by fitting of a straight line to the baseline data. When this was done, the baseline was seen to contain many frequency oscillations, some of a magnitude similar to that of the frequency deviations that occurred when the radio link was near the surface and presumably were caused by propagation through Europa's ionosphere. When the fitted residuals were integrated with respect to time to produce phase residuals, a number of linear features that abruptly changed back and forth from positive to negative slopes were observed. These linear features were clearly not random but were caused by abrupt positive and negative frequency steps on the order of a few millihertz, which probably are caused by gradients in the interplanetary plasma between the spacecraft and Earth. These frequency steps were computed from the magnitude and duration of the linear phase features and were removed throughout each data set. This procedure reduced the peak-to-peak excursions of the residual phase on the baseline by about a factor of 10, making the final results far less dependent on where on the baseline the linear function fit is applied. This step is crucial to the analysis, because the maximum frequency excursions, presumably due to a Europa ionosphere, are only on the order of 20 to 40 mHz.

The corrected residuals were then converted to refraction angle by application of the Europa-centered trajectory information derived from the Galileo navigation ephemeris, with iterative corrections for light propagation time between the spacecraft and the limb of Europa and from the limb to the DSN stations. These data were then inverted with the use of well-known techniques (10-12) to obtain the refractivity, or (n – 1) × 106, as a function of altitude. This is directly related to the electron density through a constant that is dependent on the inverse square of the frequency. The geometry of each measurement is illustrated in Table 1. The occultation locations, which are depicted by the diagrams in Figs. 1 through 3, are widely different because each occultation occurred when Europa was in a different place in its orbit about Jupiter. All occultations occurred within about 5° of the terminator.

The electron density profiles show the presence of ionization in the Europa atmosphere in five of the six measurements. The observation when no ionosphere was detected—E6a exit (Fig.2)—occurred at a ram angle of nearly 180°, near the middle of the wake region, where the electron density could be expected to be depleted. In contrast, the observations during the E4 occultation (Fig. 1) were made at ram angles close to 90°, where the line of sight lay along the wake direction, thus perhaps enhancing the electron density and giving rise to effects caused by the absence of spherical symmetry. The maximum electron densities in the other cases ranged from about 6 × 103cm−3 to about 11 × 103 cm−3 and occurred at or near the surface of Europa. There are significant differences between individual measurements, so for the purpose of modeling, an average electron density profile was computed, which is shown in Fig. 4. The average electron density has a maximum of about 9 × 103 ± 4 × 103cm−3 and a plasma scale height of about 240 ± 40 km below 300-km altitude and about 440 ± 60 km above 300- km altitude. These electron densities may be the result of photoionization, impact ionization, or a combination of both. The various likely scenarios and the resulting implications for the presence of an atmosphere around Europa are discussed in the following paragraphs.

Figure 1

The E4 occultation of Galileo by Europa. The locations of the occultation points relative to the magnetospheric ram direction and the terminator are shown by the inset diagram. The error bars in Figs. through 3 represent three times the standard deviation of the baseline data computed from the altitude of the first data point down to 1000 km. The apparent enhancement of ionization up to about 500 km in the entry observation may be an effect of looking along the wake direction.

Figure 2

The E6a occultation. The lack of an observable ionosphere in the exit observation is most probably due to its location, which is almost in the middle of the anti–ram wake region, which may be depleted of ionization.

Figure 3

The E6b occultation. The similarity of the entry and exit observations is puzzling, because the former lies on the ram side and the latter lies well within the wake region.

Figure 4

Average electron density profile computed from the five observations in which an ionosphere was detected. The error bars represent the square root of the sum of the squares of the standard deviation of the mean and the standard deviations of the data along the baseline of each data set. The average profile (line with dots) is best described by a model having a plasma scale height of 240 ± 40 km below 300 km and 440 ± 60 km above that altitude (smooth solid lines).

First we consider the case of an ionosphere with an electron density of 1 × 104 cm−3 near the surface. The photoabsorption cross sections at extreme ultraviolet (EUV) wavelengths of H2O, O2, O, H2 ,OH, and H, which are the likely neutral atmospheric constituents, are not very different; a value of ∼3 × 10−17 cm2is a reasonable approximation. Therefore, if the ionosphere is in a region where the optical depth is less than unity, the column density above the surface needs to be less than ∼3 × 1016cm−2. If the atmospheric scale height is ∼20 km (4), then the surface density has to be less than 1.5 × 1010 cm−3 in order to ensure an optical depth of less than unity. An upper limit on the column density of ∼3 × 1015 cm−2 was also suggested by Johnson (3). He argued that the atmosphere is created by slow (∼100 eV) ion sputtering of the surface, then these ions must be able to reach the surface, thus setting an upper limit on the column density. The HST observations (6) suggest that an atmosphere consisting primarily of O2 should be considered first, under the assumption that chemical processes dominate at the altitude region of maximum electron density. Assuming dissociative recombination as the dominant loss process for the O2 + ions, the neutral density required to produce an electron density of ne at zero optical depth is [O2] = ne 2 α/ξ, where α is the dissociative recombination rate and ξ is the ionization frequency. Taking the terrestrial photoionization frequency of about 9 × 10−7s−1 (13) adjusted for Jupiter distance (∼3.3 × 10−8 s−1), a recombination rate constant of 1 × 10−7 cm3 s−1, and an electron density of 1 × 104 cm−3, one gets an O2 density of ∼3 × 108 cm−3. This required density is reduced if additional ionization due to electron impact is considered. Schreier et al.(14) used measured plasma parameters from Voyager and estimated a total electron impact ionization rate of 1.88 × 10−7 s−1. This ionization rate is more than a factor of 5 larger than the rate for photoionization and thus leads to a required molecular oxygen density that is smaller, namely ∼5 × 107 cm−3.

If the major neutral constituent is not O2 but one of the other likely candidate constituents such as H2O, H, H2, OH, and O, the atmospheric density requirement does not change very much. Any of the other three molecular candidates will not change the above requirement significantly, because neither the dissociative recombination rate coefficients nor the ionization frequencies are significantly different. For example, if H2O is taken to be the dominant neutral molecular species, which leads to H3O+as the main ion species, the required neutral density is still on the order of 108 cm−3. If the neutral atmosphere consists of a mixture of atomic and molecular species, the conclusion is very similar, because both H+ and O+ can charge-exchange rapidly with one of the molecular neutral species, forming molecular ions.

The measured electron density scale height is on the order of 240 km in the lowest 300 km, which implies a neutral gas scale height of 120 km in a region where chemical processes dominate. This neutral scale height corresponds to a gas temperature of about 340 K for H2O and 610 K for O2. These temperatures are significantly higher than the apparent surface temperature of about 130 K (15); therefore, they require further consideration. A first-order estimate of the energy required, as an energy inflow at the top of the atmosphere, to maintain this elevated atmospheric temperature is on the order of 1010 eV cm−2s−1, which is more than a factor of 100 less than that which could be obtained from the surrounding jovian magnetosphere. This does not necessarily imply the presence of such a heat source, but it is energetically well within the realm of possibility. We also note that the upper limit of the solar column heating rate at Europa is about 1 × 109 eV cm−2 s−1.

Next, let us look at the issue of optical depth and column density in light of the atmospheric density values that have just been obtained. If the larger number density requirement is imposed, corresponding to photoionization (3 × 108cm−3), column densities of about 6 × 101 4 cm−2 and 3.6 × 1015 cm−2 are obtained for assumed neutral atmosphere scale heights of 20 and 120 km, respectively. Even the larger column density, if reduced by the factor of 5 corresponding to additional electron impact ionization, brings the required column density below the value set by Hall et al. (6). It should also be pointed out that these column densities do lead to an optical depth of less than unity, which was assumed at the beginning of this discussion. It is now appropriate to address the validity of the assumption of chemical equilibrium conditions. The chemical time constant near the surface (∼500 s), given the observed scale height, is significantly smaller than the diffusive time constant (∼104 s); thus, the assumption is a good one.

The near-surface atmospheric densities and temperatures that were found to be consistent with the ionospheric observations, namely a number density of ∼108 cm−3 and a temperature of ∼340 to 600 K, correspond to a surface pressure of 0.01 nbar. This value is comparable to about the limit set by the Voyager UV spectrometer for Ganymede (16), but no comparable limits are available for Europa.

A very unlikely scenario would lead to a density requirement that is significantly lower than any discussed here so far. This is the case of an atmosphere that consists solely of atomic hydrogen or oxygen, with no molecular species being present. The required atmospheric density is very different in this case, because the loss of atomic ions by either radiative recombination, corresponding to a lifetime of about 107 s, or diffusion to the surface, with a lifetime of about 106 s, are both very slow processes. In this scenario, transport processes related to Europa's interaction with Jupiter's magnetosphere are likely to be most important.

Finally, it should be pointed out that the external magnetospheric thermal plasma and magnetic pressures on Europa's ionosphere are about 2 × 10−8 and 2 × 10−7 Nm−2, respectively. Even if the electron and ion temperatures are allowed to be four times greater than the neutral gas temperature, the peak ionospheric thermal plasma pressure, calculated with the measured peak electron density, is still significantly less than the external one. This means that the ionosphere will be strongly magnetized and coupled to the magnetosphere.

REFERENCES AND NOTES

View Abstract

Navigate This Article