http://www.diva-portal.org

This is the published version of a paper published in Geophysical Research Letters.

Citation for the original published paper (version of record):

Roth, L., Ivchenko, N., Retherford, K D., Cunningham, N J., Feldman, P D. et al. (2016) Constraints on an at Ceres from Hubble Space Telescope observations. Geophysical Research Letters, 43(6): 2465-2472 https://doi.org/10.1002/2015GL067451

Access to the published version may require subscription.

N.B. When citing this work, cite the original published paper.

Permanent link to this version: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-187210 Geophysical Research Letters

RESEARCH LETTER Constraints on an exosphere at Ceres from Hubble Space 10.1002/2015GL067451 Telescope observations

Key Points: 1 1 2 3 • New unprecedented far Lorenz Roth , Nickolay Ivchenko , Kurt D. Retherford , Nathaniel J. Cunningham , observations at Ceres with the Hubble Paul D. Feldman4, Joachim Saur5, John R. Spencer6, and Darrell F. Strobel4,7 Space Telescope • Upper limits for oxygen emission are 1School of Electrical Engineering, KTH Royal Institute of Technology, Stockholm, Sweden, 2Southwest Research Institute, derived and related to upper limits for 3 oxygen abundance San Antonio, Texas, USA, Department of Physics and Astronomy, Nebraska Wesleyan University, Lincoln, Nebraska, USA, 4 5 • The found upper limit for O suggests Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, Maryland, USA, Institute of Geophysics 6 that H2O production is not higher and , University of Cologne, Cologne, Germany, Southwest Research Institute, Boulder, Colorado, USA, than during previous observations 7Department of and Planetary Science, The Johns Hopkins University, Baltimore, Maryland, USA

Correspondence to: L. Roth, Abstract We report far ultraviolet observations of Ceres obtained with the Cosmic Origin Spectrograph [email protected] (COS) of the Hubble Space Telescope in the search for atomic emissions from an exosphere. The derived brightnesses at the oxygen lines at 1304 Å and 1356 Å are consistent with zero signals within the 1𝜎

Citation: propagated statistical uncertainties. The OI 1304 Å brightness of 0.12 ± 0.20 Rayleighs can be explained by Roth, L., N. Ivchenko, K. D. Retherford, solar resonant scattering from an atomic oxygen column density of (8.2 ± 13.4)×1010 cm−2. Assuming that N. J. Cunningham, P. D. Feldman, O is produced by photodissociation of H O, we derive an upper limit for H O abundance and compare it to J. Saur, J. R. Spencer, and D. F. Strobel 2 2 (2016), Constraints on an exosphere at previous observations. Our upper limit is well above the expected O brightness for a tenuous sublimated 26 −1 Ceres from Hubble Space Telescope H2O exosphere, but it suggests that H2O production with a rate higher than 4 × 10 molecules s was not observations, Geophys. Res. Lett., present at the time of the COS observation. Additionally, we derive an extremely low geometric albedo of 43, doi:10.1002/2015GL067451. ∼1% in the 1300 Å to 1400 Å range.

Received 17 DEC 2015 Accepted 28 FEB 2015 1. Introduction Accepted article online 5 MAR 2016 Dwarf planet 1 Ceres is the largest object in the asteroid belt with a mean radius of ∼475 km [Thomas et al., 2005]. Despite being located in the relatively ice-poor asteroid belt, the shape and mean density of ∼2.2 g cm−3 may imply a differentiated body and a substantial part of its total mass is water ice [e.g., McCord et al., 2011]. Two decades ago, UV spectra taken by the International Ultraviolet Explorer (IUE) revealed emission features from the OH A-X (0,0) band at 3085 Å near Ceres presumably resulting from dissociation of water vapor [A’Hearn and Feldman, 1992]. Global water vapor production rates of (0.5–1.4) ×1026 molecules s−1 were esti- mated from the observed OH brightness integrated over distances roughly 7′′ –23′′ (arc sec) away from Ceres. Ground-based spectra with higher sensitivity taken in 2007 did not detect OH and constrained the produc- tion rate to less than ∼7 × 1025 molecules s−1 during the time of the observations [Rousselot et al., 2011]. Recently, water vapor was detected around Ceres by Herschel through absorption at submillimeter wave- lengths with a production rate of at least 1026 molecules s−1 [Küppers et al., 2014a], similar to the rate from the IUE detection. The water vapor appears to originate from localized sources and is hypothesized to come from comet-like sublimation of exposed ices, some form of cryovolcanic activity or impact events, but the sources and mechanisms are not yet understood [e.g., Hayne and Aharonson, 2015; Küppers et al., 2014b].

Exosphere modeling indicates that water sublimation might sustain a very tenuous global H2O exosphere [Fanale and Salvail, 1989], and recent 3-D simulations suggest peak H2O densities around the cold polar caps [Tuetal., 2014]. If sublimation derived, the source rate for sublimation of H2O and hence the exospheric density ultimately depend on the amount of ice that is either actually exposed at the surface or only covered by a very thin layer of nonice material [Fanale and Salvail, 1989; Titus, 2015; Formisano et al., 2015]. If surface ice is present due to recent impact or outgassing events, sublimation rates are expected to be higher by at least an order of magnitude [Titus, 2015; Hayne and Aharonson, 2015]. Since March 2015 NASA’s Dawn spacecraft is in orbit around Ceres. The main objectives of the Dawn mission

©2016. American Geophysical Union. are to measure and characterize Ceres’s body parameters, its chemical composition, interior structure, and All Rights Reserved. geological and thermal history [e.g., Russell and Raymond, 2011]. First results from Dawn’s Framing Camera

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 1 Geophysical Research Letters 10.1002/2015GL067451

Table 1. Details and Geometry Parameters of the COS Observations

Sun Earth Ceres True Exposure Start Exposure Used Exposure y Offset Subobserver Date Distance Distance Diameter Anomaly No Time Time Timea Ceres E longitude (UT) (AU) (AU) (arc sec) (deg) # ID (UT) (s) (s) (arc sec) (deg)b 2015-Aug-26 2.96 2.09 0.63 155 1 lcu701010 10:30 2361 30–2361 ∼0.0 329–304 2 lcu701020 11:55 2912 (580–2912)c ∼1.0 267–241 3 lcu701030 13:30 2912 580–2912 ∼0.0 204–178 4 lcu701040 15:06 2915 (580–2915)b ∼1.0 141–115 5 lcu701050 16:41 2912 580–2912 ∼0.0 78–52 aUsed exposure time after exposure start. The first portion of each exposure was cut because of high geocoronal background. bExposures 2 and 4 were used only for background emission determination. cThe given range corresponds to the planetocentric subobserver east longitude of Ceres at the start and end of the used exposure time after the Dawn postsurvey updated coordinate system (SPICE kernel “dawn_ceres_v05.tpc”).

revealed hazes over a bright pit on the floor of crater Occator, which are suggested to arise from sublimation of water ice [Nathues et al., 2015]. The Visible-Infrared Mapping (VIR) Spectrometer on Dawn, however, did not find detectable water ice on the surface [De Sanctis et al., 2015]. The VIR spectra instead indicate widespread ammoniated phyllosilicates across the surface. Global limb scans obtained at high solar phase angles also did not reveal any obvious plumes or global dust production [Nathues et al., 2015; Li et al., 2015]. As a complementary observation, we used the Cosmic Origins Spectrograph (COS) of the Hubble Space Telescope to search for atomic emissions from Ceres exosphere at far ultraviolet (FUV) wavelengths. The aim of the observations is to determine or constrain, in particular, the brightnesses of the FUV oxygen emissions of the transitions at 1304 Å and 1356 Å. Considering various processes like solar fluorescence and electron impact excitation of oxygen bearing species, we use the derived brightnesses to constrain exospheric abundances.

2. Observations We observed Ceres at far ultraviolet (FUV) wavelengths using COS during a five-orbit visit by Hubble Space Telescope (HST) on 26 August 2015 (Table 1). Five exposures were taken, one per orbit, with the G130M grating and a central wavelength of 1291 Å, providing wavelength coverage of 1290–1430 Å on detector segment A and 1135–1275 Å on segment B. Our focus lies on the oxygen 1304 Å and 1356 Å emission, and we thus analyze detector segment A. The observations were scheduled such that the planetocentric longitude of the Occator crater (239∘E) associated with a potential water vapor source region [Küppers et al., 2014a], and active sublimation [Nathues et al., 2015] was on the observed hemisphere through ∼60% of the total exposure time (see Table 1). As a first step, we cropped those parts of each exposure that are contaminated by scattered light from the geocorona. The observations were obtained 4 weeks after Ceres’s opposition, and HST was on the Earth’s dayside at the beginning of each observing orbit. We used the -- electrodynamics-TAG data on the temporal variation of the OI 1304 Å emission to monitor the geocoronal background, which dropped down to a constant nightside level after about one fifth of each orbit. Only the low-geocorona portions of each exposure were used. Because the first exposure started later during the orbit, it was less contaminated, see Table 1. Ceres was centered along the cross dispersion (or vertical) axis (y axis) within the 2.5 arc sec diameter pinhole aperture of COS during exposures 1, 3, and 5 and offset by about y = 1′′ toward the upper aperture edge in exposures 2 and 4, see Figure 1. We fitted a Gaussian plus a linear function to the cross dispersion y profiles of all counts integrated along the dispersion axis (see plots to the right in Figure 1) in order to determine Ceres’s y location. The linear function is added to the Gaussian to account for trending background counts. In exposures 1, 3, and 5 (Figure 1, top) the peak of the Gaussian (solid green) coincides roughly with the nominal aperture center (solid red). In exposures 2 and 4 (Figure 1, bottom) the fitted peak is at an offset of ∼1′′ and thus close to the nominal upper aperture edge (dashed red). Before assembling the exposures, we determined Ceres’ y location in the individual exposures yielding similar or identical offset for exposures 1, 3, and 5 and 2 and 4, respectively (as also planned initially). The exposures with matching locations were then superposed (Figure 1).

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 2 Geophysical Research Letters 10.1002/2015GL067451

Figure 1. Spectral images and profiles of integrated counts in the dispersion and cross-dispersion direction assembled from exposures obtained during HST orbits (top) 1, 3, and 5 and (bottom) 2 and 4. The vertical center of Ceres (green solid) was located near the aperture center (red solid) during orbits 1, 3, and 5 and close to the upper aperture edge (red dashed) during orbits 2 and 4. Note that the COS FUV spectral trace is not flat along the y axis. Between 0 and 5 counts per pixel are detected in the images. The spectra above show the y-integrated counts for bins of 32 pixels along the dispersion axis (1 x pixel = 0.01 Å). The profiles to the right show the counts integrated along the dispersion (x)axis per y pixel and the fitted profile for Ceres (green dashed).

Because Ceres is located near the aperture edge in exposures 2 and 4, portions of the Ceres signal are vignetted or missed in these two exposures. The count rate (counts/s) at the strong solar CII 1335 Å line, which can be used as a diagnostic for the brightness of sunlight reflected off the surface, is 40% lower in exposures 2 and 4 compared to the center exposures (1,3, and 5). Hence, about 40% of the signal from Ceres is potentially lost due to the offset in exposures 2 and 4. We continue using only the assembled image from exposures 1, 3, and 5 (Figure 1, bottom) and integrate over the entire nominal aperture along the cross-dispersion axis to obtain a spectrum. The spectrum from the lower aperture half from exposures 2 and 4, where no or only a very weak signal from Ceres is detected, is used for the background emission. This half-aperture background is then subtracted twice from the aperture-wide Ceres signal to account for the different width of the regions on the detector. We thereby neglect the slightly higher gain sag and thus lower sensitivity in the upper aperture half at the current lifetime position of the aperture on the COS detector, which might lead to an oversubtraction of the background. If Ceres-related signal is present in the lower aperture half of exposures 2 and 4, it will also falsely increase the subtracted background. Thus, the applied background correction possibly prevents a detection but might also lead to a slightly underestimated upper limit for the resulting nondetection. The background-corrected spectral photon flux (black) and the propagated uncertainty (grey) are displayed in 32 pixel bins in Figure 2. Peak fluxes from sunlight reflected off the surface are found at the oxygen lines at 1304 Å (and weaker near 1356 Å) and carbon and silicon ion doublets around 1335 Å (C II) and 1397 Å (Si IV). Removal of the reflected sunlight is performed in analogy to the method applied by Cunningham et al. [2015] to COS spectra of Jupiter’s moon . Because the solar spectrum varies with solar longitude, we used the Solar Radiation and Climate Experiment/Solar Stellar Irradiance Comparison Experiment (SORCE/SOLSTICE) spectrum [McClintock et al., 2005] from 28 August to match the sub-Ceres solar longitude from the day of the observation. We note that the solar activity level hardly changed between 26 and 28 August and using the spectrum from 26 August yields almost identical results. The SOLSTICE spectrum is recorded at 1.0 Å full width at half maximum (FWHM) spectral resolution, lower than COS, and we therefore used the SOLSTICE spectrum to modify a higher-resolution solar spectrum from the Solar Ultraviolet Measurements of Emitted Radiation (SUMER) spectral atlas of Curdtetal.[2001], adjusting the SUMER continuum and individual emission line strengths with an automated simplex routine so that it matched the SOLSTICE spectrum when degraded to SOLSTICE resolution.

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 3 Geophysical Research Letters 10.1002/2015GL067451

Figure 2. Combined full-aperture spectrum of exposures 1, 3, and 5 corrected for sky background measured in the lower aperture half during exposures 2 and 4. The solar spectrum (red) is scaled to the C II 1335 Å doublet and subtracted (blue). No residual emission outside the propagated uncertainties (light grey) are detected at 1304 Å or 1356 Å. The slight surplus at the solar Si IV 1394 Å line can be attributed to differing line shapes and solar coronal variations.

The corrected high-resolution solar spectrum is adjusted to the COS resolution through a convolution with a Gaussian profile with FWHM of Ceres’s diameter, which corresponds to 0.22 Å on the dispersion axis. We then

calculate the geometric albedo from the measured flux FCeres by

𝜋 2 FCeres dSun p = f𝛼 , (1) F⊙ ΩCeres

where dSun is Ceres- distance, ΩCeres is the solid angle of Ceres as seen from Earth, and F⊙ is the solar flux at 1 AU. We assumed unity for the unknown photometric phase function f𝛼, which is a reasonably good esti- mate for these data obtained near opposition. A very low geometric albedo of 0.9% ± 0.1% is derived at the solar CII 1335 Å doublet. A similar but slightly higher albedo of 1.1% ± 0.1% is found at the stronger line of the Si IV doublet at 1393.8 Å. The measured Ceres signal might be underestimated because of (1) potential vignetting from a small x offset and (2) an overestimation of the subtracted background as discussed ear- lier. Therefore, the derived albedo might underestimate the actual FUV albedo. The slightly higher albedo at 1393.8 Å compared to CII 1335 Å could be simply statistical fluctuations or solar activity variations between the used solar spectrum and the observation. The solar spectrum is then adjusted to the measured CII 1335 Å with the derived albedo and subtracted from the measured spectrum. Residual emissions are shown in blue at an offset to the zero line in Figure 2. Small peaks near the solar CII and Si IV lines are attributed to slight line shape differences of the solar spectrum and the data. Integrating over the oxygen emissions at 1304 Å and 1356 Å, we derive brightnesses normalized to the entire aperture size (with a diameter 2.5 arc sec compared to Ceres’s diameter of 0.63 arc sec) of 0.12±0.20 R (1304 Å) and 0.02 ± 0.12 R (1356 Å), respectively.

3. Discussion 3.1. FUV Albedo Our analysis reveals a very low FUV albedo in the 1300 to 1400 Å range of about 1%. Higher UV albedos with values between 6% and 12% around 1620 Å and 3% near 2230 Å were measured in previous HST imaging observations [Parker et al., 2002; Li et al., 2006]. Measured albedos from Ceres or other asteroids in our wave- length range below 1400 Å are barely available in the literature, because of the difficulty to measure the mostly low reflectances of small bodies from Earth orbit. Observations by the Rosetta Alice UV spectrograph revealed higher far UV albedos of ∼5% at asteroids (2867) Šteins and (21) Lutetia [A’Hearn et al., 2010; Stern et al., 2011], which are, however, not carbonaceous asteroids like Ceres. Low FUV albedos similar to the value derived here were found at icy moons of Saturn [e.g., Hendrix et al., 2010, 2012] and Jupiter [e.g., Hall et al., 1998; Feldman et al., 2000; Strobel et al., 2002; Roth et al., 2014]. In particular, the reflectivity of the Saturnian moon Phoebe

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 4 Geophysical Research Letters 10.1002/2015GL067451

yields very similar values near 1% at the same wavelengths [Hendrix and Hansen, 2008]. Like mentioned in section 1, however, the most recent results from the Dawn mission suggest that exposed water ice on the surface is not present at high surface concentrations [De Sanctis et al., 2015]. Various minerals also have very low FUV reflectivity and might account for Ceres’s low FUV albedo [Wagner et al., 1987]. Parker et al. [2002] suggest UV reflectance properties that are more Lambertian than for the Moon; their higher albedos near 1620 Å are difficult to reconcile with the present measurement without invoking the relatively low data quality described therein. A more detailed FUV reflectivity analysis that also addresses a potential change in reflec- tivity across the 1300 Å to 1400 Å range in our spectrum requires a wider spectrum like the recently obtained HST Space Telescope Imaging Spectrograph observations by Hendrix and Vilas [2015]. 3.2. Constraints on Exospheric O Abundance We calculate the theoretical emission yields for solar fluorescence (resonant scattering) by atomic oxygen, O,

and for electron impact (dissociative) excitation of O, O2,H2O, and CO2 in order to estimate the contributions and to relate the oxygen brightnesses and upper limits to exospheric abundances. For OI 1304 Å, a resonant scattering rate or g factor of 1.5 × 10−6 s−1 is derived based on the measured solar flux around 1304 Å of 1.15 × 109 photons/cm2/s. Here we assumed atomic oxygen with a temperature of T = 200 K in approximate thermal equilibrium with the surface. To estimate emission yields from excitation by impacting solar wind electrons, we checked the solar wind conditions 7 to 10 days (travel time from 1 AU to 2.96 AU is 8.5 days at 400 km/s) before the observations. Based on the measured densities between 3 and 8 cm−3 (at standard moderate velocities between 330 and 530 km/s), we assume a density of 6 cm−3 at 1 AU, which is then adjusted to Ceres’s distance assuming a density variation with inverse distance squared [e.g., Belcher et al., 1993] yielding 0.7 cm−3 at 2.96 AU. For

the temperature of the core solar wind electron population at Ceres we use kBTe = 5 eV based on Ulysses measurements [Issautier et al., 1998]. Taking the electron impact cross sections for O, O2,CO2, and H2O[Kanik et al., 2001, 2003; Ajello, 1971; Makarov et al., 2004], we find excitation rates between 5 × 10−10 s−1 and 3 × 10−13 s−1. These rates are all more than 3 orders of magnitude below the resonant scattering rate. Recently, FUV emissions measured by the Rosetta Alice spectrograph at comet 67P/Churyumov-Gerasimenko

were attributed to photoelectron excitation of H2O, when the comet was at similar heliocentric distance [Feldman et al., 2015]. Because the photoelectron production rate is proportional to the neutral density, emis- sion yields will be large only in localized high density regimes (like measured by Alice over ∼1kmfromthe comet). Average emissions from photoelectron excitation in a larger field of view of several hundreds of kilometers like in the case of the COS observations will be small. Thus, required column densities to produce detectable emissions are orders above the expected abundances, and we do not further consider electron impact excitation. The insignificant brightness and upper limits of the 1356 Å doublet that can only be produced by electron impact do not allow meaningful constraints on abundances. . . 10 −2 The measured OI 1304 Å brightness is then converted to a column density of NO =(8 0 ± 13 3)×10 cm using the resonant scattering rate estimated above. The signal integrated over the entire aperture corre- . . 27 sponds to a total abundance of (9 3 ± 15 2)×10 O atoms within 1.25 arc sec or 1900 km (4 RCeres, Ceres radii) around the aperture center. 3.3. Comparison and Implications First, we compare our derived O column density directly to the OH column density from the IUE observations [A’Hearn and Feldman, 1992]. IUE measured OH column densities of 1.4 × 1011 cm−2 and 2.8 × 1011 cm−2 in 4 4 two exposures with field of views that covered areas about 1 × 10 km (20 RCeres)to3 × 10 km (60 RCeres)away from Ceres, see Figure 3. For the case of a comet-like exosphere that expands radially at a constant velocity (neglecting gravity) and 2 without any loss, the H2O density decreases by 1/r with distance r. The radial decrease of secondary species around comets, like OH and O derived from H2O, can assume even flatter profiles [e.g., Morgenthaler et al., 2001]. The results from Tu et al. [2014] for Ceres suggest, however, a considerably steeper decrease for the H2O profile with a H2O density decrease of 3 orders of magnitude over 1 RC above the surface. Assuming now that most of the O and OH is produced from H2O near the surface, the line-of-sight O and OH column densities decrease with ∼1/r. For such profiles, the column densities of the same species in the region observed by COS should then be at least ∼15 times higher compared to the IUE field of view for the assumed shallow profile.

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 5 Geophysical Research Letters 10.1002/2015GL067451

Figure 3. Comparison of the OH column density measured by IUE with the COS constraints on O. The line-of-sight OH column density profile for a radially outflowing O/OH exosphere (∝1/r, solid black) is adjusted to the mean value measured by IUE (blue solid with measured range) in a large field of view (vertical dotted blue). An ∼8 times lower O column density is expected from photodissociation of H2O (dashed black). The upper limit for O from COS (solid red with error) is below the expected O abundance in the COS field of view (vertical dotted red). For less steep profiles (grey solid and dashed), our O brightness and upper limit are consistent with the IUE OH detection.

Now photodissociation to produce atomic O from H2O is yet less effective by a factor of ∼8 compared to pho- todissociation of H2O into OH (and H) [Huebner et al., 1992]. Thus, we might expect a roughly 15∕8 ≈ 2 times higher column density for O in our COS observations compared to OH in the IUE field of view. The inferred O column density is instead ∼2.5 times lower than the IUE OH column densities (Figure 3). If the line-of-sight column density decrease is, however, less steep than 1∕r, our results√ will be less constraining. As an example for a flatter profile, we have plotted the O and OH densities for a 1∕ r profile in Figure 3 (grey). For this profile, our O range is well consistent with the IUE detection of OH.

Second, we relate the abundance of O in the COS field of view to an H2O abundance. By applying an analytical approach, we convert this to an H2O source rate, which can be compared to previous results. The photodis- 𝜅 . −7 −1 sociation rate to produce O from H2Ois dissoc = 1 7 × 10 s at 2.96 AU [Huebner et al., 1992]. Assuming a velocity component of 0.05 km s−1 perpendicular to the line of sight (i.e., the O atoms leave the field of view 𝜏 . 4 moving at this speed), the lifetime of O atoms in the COS field of view is O = 3 6 × 10 s. Loss from ionization of O can be neglected on this time scale. The measured number of atomic oxygen O within the COS field of . . 30 view then converts to a total number of the parent molecule H2Oof(1 4 ± 2 4)×10 . We estimate an H2O exosphere at Ceres that matches the found upper limit using classical exosphere theory to calculate ballistic

and escaping H2O molecules from the sunlit hemisphere assuming hemispherical symmetry. For the sunlit hemisphere, we assume that there are random patches of exposed ice at a surface temperature . −8 of 200 K. If these patches constitute only 1 5 × 10 of the total sunlit surface area, then the H2O sublimation . 26 −1 5 −3 flux is Q = 1 5 × 10 molecules s ; the average H2O surface density is 9 × 10 cm with the exobase at the . 26 −1 surface and an escape rate of 0 9 × 10 s . From the calculated distribution of H2O molecules, we compute . 30 the total number of H2O in the COS field of view to be 1 4 × 10 molecules matching the H2O abundance in the field of view derived from the 0.12 R OI 1304 Å brightness. Since the calculation is linear in sublimation

rate and H2O number density for the range considered, a factor of 2.7 higher abundance of ice patches yields . 26 a 2.7 times larger sublimation flux of Q = 4 0 × 10 , which corresponds to the upper limit for H2O abundance in the field of view. For comparison, production rates on the order of 1024 to 1025 molecules s−1 from subsurface ice sublimation

were studied by FanaleandSalvail [1989] and Tuetal. [2014]. The H2O and O abundance from such sublimation rates are well below our upper limits. Higher source rates of Q ∼ 1 − 2 × 1026 molecules s−1 were derived from the IUE and Herschel observations [A’HearnandFeldman, 1992; Küppersetal., 2014a]. These rates are still within 𝜎 26 −1 the 1 upper limit derived from our observations. H2O source rates higher than 4 × 10 s would lead to a brighter O emission than detected by COS. On the other hand, the direct comparison to the OH abundance

measured by IUE near Ceres A’Hearn and Feldman [1992] suggests that the H2O production is similar or slightly lower than during the IUE detection, i.e., lower than 2 × 1026 molecules s−1.

4. Conclusion

We have derived exospheric O and H2O upper limits that suggest that strong water vapor production was not present during the COS observations. If H2O was actively produced on the observed hemisphere at a rate

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 6 Geophysical Research Letters 10.1002/2015GL067451

higher than 1.5 × 1026 molecules s−1, then a higher OI 1304 Å signal would be expected from solar resonant scattering. Ceres was at a distance of 2.96 AU approaching aphelion at the time of the COS observations. No previous attempts to constrain water vapor at Ceres were made in this orbital phase. Successful detections by Herschel and IUE occurred before and after perihelion, while no water was detected during two observa- tions in the orbital phase after aphelion including a Herschel observation at 2.94 AU [Rousselot et al., 2011; Küppers et al., 2014a].

The O abundances derived here are in agreement with a tenuous H2O exosphere. Our estimated upper limit 26 −1 for the H2O production rate of 4×10 molecules s is comparable to the rates derived from previous Herschel and IUE detections of H2O and OH, respectively.

Acknowledgments References We thank Marty Snow of LASP for full-resolution SOLSTICE spectra and A’Hearn, M. F., and P. D. Feldman (1992), Water vaporization on Ceres, Icarus, 98, 54–60, doi:10.1016/0019-1035(92)90206-M. Emmanuel Chané for helping to assess A’Hearn, M. F., et al. (2010), The far-ultraviolet albedo of Šteins measured with Rosetta-ALICE, Planet. Space Sci., 58, 1088–1096, solar wind conditions. This work is doi:10.1016/j.pss.2010.03.005. based on observations made with the Ajello, J. M. (1971), Emission cross sections of CO2 by electron impact in the interval 1260-4500 Å. II, J. Chem. Phys., 55, 3169–3177, NASA/ESA Hubble Space Telescope, doi:10.1063/1.1676564. which is operated by the Associa- Belcher, J. W., A. J. Lazarus, R. L. McNutt Jr., and G. S. Gordon Jr. (1993) Solar wind conditions in the outer heliosphere and the distance to the tion of Universities for Research in termination shock, J. Geophys. Res., 98, 15,177–15,183, doi:10.1029/93JA01178. Astronomy, Inc., under NASA con- Cunningham, N. J., J. R. Spencer, P. D. Feldman, D. F. Strobel, K. France, and S. N. Osterman (2015), Detection of Callisto’s oxygen atmosphere tract NAS 5-26555. These observations with the Hubble Space Telescope, Icarus, 254, 178–189. are associated with program 14106. Curdt, W., P. Brekke, U. Feldman, K. Wilhelm, B. N. Dwivedi, U. Schühle, and P. Lemaire (2001), The SUMER spectral atlas of solar-disk features, L.R. appreciates the support from the Astron. Astrophys., 375, 591–613, doi:10.1051/0004-6361:20010364. Swedish VINNOVA agency and the De Sanctis, M. C., et al. (2015), Ammoniated phyllosilicates with a likely outer solar system origin on (1) ceres, Nature, 528(7581), 241–244. Göran Gustafsson stiftelse. Fanale, F. P., and J. R. Salvail (1989), The water regime of asteroid (1) Ceres, Icarus, 82, 97–110, doi:10.1016/0019-1035(89)90026-2. Feldman, P.D., M. A. McGrath, D. F. Strobel, H. W. Moos, K. D. Retherford, and B. C. Wolven (2000), HST/STIS ultraviolet imaging of polar aurora on Ganymede, Astrophys. J., 535, 1085–1090, doi:10.1086/308889. Feldman, P. D., et al. (2015), Measurements of the near-nucleus coma of comet 67P/Churyumov-Gerasimenko with the Alice far-ultraviolet spectrograph on Rosetta, Astron. Astrophys., 583, A8, doi:10.1051/0004-6361/201525925. Formisano, M., M. C. De Sanctis, G. Magni, C. Federico, and M. T. Capria (2015), Ceres water regime: Surface temperature, water sublimation and transient exo(atmo)sphere, Mon. Not. R. Astron. Soc., 455, 1892–1904. Hall, D. T., P. D. Feldman, M. A. McGrath, and D. F. Strobel (1998), The far-ultraviolet oxygen airglow of Europa and Ganymede, Astrophys. J., 499(5), 475–481. Hayne, P. O., and O. Aharonson (2015), Thermal stability of ice on Ceres with rough topography, J. Geophys. Res. Planets, 120, 1567–1584, doi:10.1002/2015JE004887. Hendrix, A. R., and C. J. Hansen (2008), Ultraviolet observations of Phoebe from the Cassini UVIS, Icarus, 193, 323–333, doi:10.1016/j.icarus.2007.06.030. Hendrix, A. R., and F. Vilas (2015), First Spectral Coverage of two regions on Ceres in the far-UV, AAS/Division for Planetary Sciences Meeting Abstracts 103.06 presented at 47th Annual DPS Meeting National Harbor, Md., Nov. Hendrix, A. R., C. J. Hansen, and G. M. Holsclaw (2010), The ultraviolet reflectance of Enceladus: Implications for surface composition, Icarus, 206, 608–617, doi:10.1016/j.icarus.2009.11.007. Hendrix, A. R., T. A. Cassidy, B. J. Buratti, C. Paranicas, C. J. Hansen, B. Teolis, E. Roussos, E. Todd Bradley, P. Kollmann, and R. E. Johnson (2012), Mimas’ far-UV albedo: Spatial variations, Icarus, 220, 922–931, doi:10.1016/j.icarus.2012.06.012. Huebner, W. F., J. J. Keady, and S. P. Lyon (1992), Solar photo rates for planetary atmospheres and atmospheric pollutants, Astrophys. Space Sci., 195, 1–289, doi:10.1007/BF00644558. Issautier, K., N. Meyer-Vernet, M. Moncuquet, and S. Hoang (1998), Solar wind radial and latitudinal structure—Electron density and core temperature from ULYSSES thermal noise spectroscopy, J. Geophys. Res., 103, 1969–1979, doi:10.1029/97JA02661. Kanik, I., P. V. Johnson, M. B. Das, M. A. Khakoo, and S. S. Tayal (2001), Electron-impact studies of atomic oxygen: I. Differential and integral cross sections; experiment and theory, J. Phys. B, 34, 2647–2665. Kanik, I., C. Noren, O. P. Makarov, P. Vatti Palle, J. M. Ajello, and D. E. Shemansky (2003), Electron impact dissociative excitation of O2:2. Absolute emission cross sections of the OI(130.4 nm) and OI(135.6 nm) lines, J. Geophys. Res., 108(E11), 5126, doi:10.1029/2000JE001423. Küppers, M., et al. (2014a), Localized sources of water vapour on the dwarf planet (1)Ceres, Nature, 505, 525–527, doi:10.1038/nature12918. Küppers, M., et al. (2014b), Is (1) Ceres a small planet or a large comet?, Abstracts P52A-03 presented at 2014 Fall Meeting, AGU, San Francisco, Calif., 15–19 Dec. Li, J.-Y., L. A. McFadden, J. W. Parker, E. F. Young, S. A. Stern, P. C. Thomas, C. T. Russell, and M. V. Sykes (2006), Photometric analysis of 1 Ceres and surface mapping from HST observations, Icarus, 182, 143–160, doi:10.1016/j.icarus.2005.12.012. Li, J.-Y., et al. (2015), Search for dust around ceres, EPSC Abstracts vol. 10 EPSC2015-385 presented at European Planetary Science Congress 2015, Nantes, France, 28 Sept.–2 Oct. Makarov, O. P., J. M. Ajello, P. Vatti Palle, I. Kanik, M. C. Festou, and A. Bhardwaj (2004), Kinetic energy distributions and line profile measurements of dissociation products of water upon electron impact, J. Geophys. Res., 109, A09303, doi:10.1029/2002JA009353. McClintock, W. E., G. J. Rottman, and T. N. Woods (2005), Solar-Stellar Irradiance Comparison Experiment II (Solstice II): Instrument concept and design, Sol. Phys., 230, 225–258, doi:10.1007/s11207-005-7432-x. McCord, T. B., J. Castillo-Rogez, and A. Rivkin (2011), Ceres: Its origin, evolution and structure and Dawn’s potential contribution, Space Sci. Rev., 163, 63–76, doi:10.1007/s11214-010-9729-9. Morgenthaler, J. P., W. M. Harris, F. Scherb, C. M. Anderson, R. J. Oliversen, N. E. Doane, M. R. Combi, M. L. Marconi, and W. H. Smyth (2001), Large-Aperture [O I] 6300 Å Photometry of Comet Hale-Bopp: Implications for the Photochemistry of OH, Astrophys. J., 563, 451–461, doi:10.1086/323773. Nathues, A., et al. (2015), Sublimation in bright spots on (1) Ceres, Nature, 528(7581), 237–240.

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 7 Geophysical Research Letters 10.1002/2015GL067451

Parker, J. W., S. A. Stern, P. C. Thomas, M. C. Festou, W. J. Merline, E. F. Young, R. P. Binzel, and L. A. Lebofsky (2002), Analysis of the first disk-resolved images of Ceres from ultraviolet observations with the Hubble Space Telescope, Astron. J., 123, 549–557, doi:10.1086/338093. Roth, L., K. D. Retherford, J. Saur, D. F. Strobel, P. D. Feldman, M. A. McGrath, and F. Nimmo (2014), Orbital apocenter is not a sufficient condition for HST/STIS detection of Europa’s water vapor aurora, Proc. Natl. Acad. Sci. USA., 111, E5123–E5132. Rousselot, P., E. Jehin, J. Manfroid, O. Mousis, C. Dumas, B. Carry, U. Marboeuf, and J.-M. Zucconi (2011), A search for water vaporization on Ceres, Astron. J., 142, 125, doi:10.1088/0004-6256/142/4/125. Russell, C. T., and C. A. Raymond (2011), The Dawn mission to vesta and ceres, Space Sci. Rev., 163, 3–23, doi:10.1007/s11214-011-9836-2. Stern, S. A., et al. (2011), Ultraviolet Discoveries at Asteroid (21) Lutetia by the Rosetta Alice Ultraviolet Spectrograph, Astron. J., 141, 199, doi:10.1088/0004-6256/141/6/199. Strobel, D. F., J. Saur, P. D. Feldman, and M. A. McGrath (2002), Hubble Space Telescope space telescope imaging spectrograph search for an atmosphere on Callisto: A Jovian unipolar inductor, Astrophys. J., 581, L51–L54. Thomas, P. C., J. W. Parker, L. A. McFadden, C. T. Russell, S. A. Stern, M. V. Sykes, and E. F. Young (2005), Differentiation of the asteroid Ceres as revealed by its shape, Nature, 437, 224–226, doi:10.1038/nature03938. Titus, T. N. (2015), Ceres: Predictions for near-surface water ice stability and implications for plume generating processes, Geophys. Res. Lett., 42, 2130–2136, doi:10.1002/2015GL063240. Tu, L., W.-H. Ip, and Y.-C. Wang (2014), A sublimation-driven exospheric model of Ceres, Planet. Space Sci., 104, 157–162, doi:10.1016/j.pss.2014.09.002. Wagner, J. K., B. W. Hapke, and E. N. Wells (1987), Atlas of reflectance spectra of terrestrial, lunar, and meteoritic powders and frosts from 92 to 1800 NM, Icarus, 69, 14–28, doi:10.1016/0019-1035(87)90003-0.

ROTH ET AL. HST EXOSPHERE OBSERVATION AT CERES 8