DOI: 10.1126/science.1204054, 1396 (2011);332 Science

, et al.Michael F. A'HearnEPOXI at Comet Hartley 2

This copy is for your personal, non-commercial use only.

clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

  here.following the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

  ): June 17, 2011 www.sciencemag.org (this infomation is current as of

The following resources related to this article are available online at

http://www.sciencemag.org/content/332/6036/1396.full.htmlversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org/content/suppl/2011/06/15/332.6036.1396.DC1.html can be found at: Supporting Online Material

http://www.sciencemag.org/content/332/6036/1396.full.html#ref-list-1, 5 of which can be accessed free:cites 51 articlesThis article

http://www.sciencemag.org/cgi/collection/planet_sciPlanetary Science

subject collections:This article appears in the following

registered trademark of AAAS. is aScience2011 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

June

17,

201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

EPOXI at Comet Hartley 2Michael F. A’Hearn,1* Michael J. S. Belton,2 W. Alan Delamere,3 Lori M. Feaga,1

Donald Hampton,4 Jochen Kissel,5 Kenneth P. Klaasen,6 Lucy A. McFadden,1,7

Karen J. Meech,8 H. Jay Melosh,9,10 Peter H. Schultz,11 Jessica M. Sunshine,1

Peter C. Thomas,12 Joseph Veverka,12 Dennis D. Wellnitz,1 Donald K. Yeomans,6

Sebastien Besse,1 Dennis Bodewits,1 Timothy J. Bowling,10 Brian T. Carcich,12

Steven M. Collins,6 Tony L. Farnham,1 Olivier Groussin,13 Brendan Hermalyn,11

Michael S. Kelley,1,14 Jian-Yang Li,1 Don J. Lindler,15 Carey M. Lisse,16

Stephanie A. McLaughlin,1 Frédéric Merlin,1,17 Silvia Protopapa,1

James E. Richardson,10 Jade L. Williams1

Understanding how comets work—what drives their activity—is crucial to the use of comets instudying the early solar system. EPOXI (Extrasolar Planet Observation and Deep Impact ExtendedInvestigation) flew past comet 103P/Hartley 2, one with an unusually small but very active nucleus,taking both images and spectra. Unlike large, relatively inactive nuclei, this nucleus is outgassingprimarily because of CO2, which drags chunks of ice out of the nucleus. It also shows substantialdifferences in the relative abundance of volatiles from various parts of the nucleus.

Comets are the fundamental building blocksof the giant planets and may be an im-portant source of water and organics on

Earth. On 4 July 2005, the Deep Impact mis-sion carried out an impact experiment on comet9P/Tempel 1 (1, 2) to study differences betweenthe comet’s surface and the interior. Althoughthe impactor spacecraft was destroyed, the flybyspacecraft and its instruments remained healthyin its 3-year, heliocentric orbit after completionof the mission. The Deep Impact flyby spacecraftwas retargeted to comet 103P/Hartley 2 as part ofan extended mission named EPOXI (ExtrasolarPlanet Observation and Deep Impact ExtendedInvestigation).

The flyby spacecraft carries the High Resolu-tion Instrument (HRI), which combines a visible-wavelength camera with a pixel size of 2 mradand a set of filters with a near-infrared (near-IR)(1.05 to 4.85 mm) spectrometer with an entranceslit of 10 mrad by 256 mrad, with 512 spatialpixels along the slit. Spectral maps were createdby scanning the slit across the comet while tak-ing a sequence of spectra. The Medium Resolu-tion Instrument (MRI) has a pixel size of 10 mradand a different but overlapping set of visible-wavelength filters (3, 4).

Encounter with Hartley 2The closest approach to Hartley 2 was 694 km at13:59:47.31 UTC on 4 November 2010, 1 weekafter perihelion passage and at 1.064 astronom-ical units (AU) from the Sun. Flyby speed was12.3 km s–1, and the spacecraft flew under thecomet with a somewhat northward trajectory in asolar system reference frame. Because instru-ments are body-mounted on the spacecraft, thespacecraft rotated to keep the instruments pointedat the comet. Observations of the comet werecarried out for 2 months on approach (5 Sep-tember to 4 November) and for 3 weeks on de-parture (4 to 26 November), during which morethan 105 images and spectra were obtained.

Prior remote sensing showed that Hartley 2’snucleus has an average radius 1/5 that of cometTempel 1’s nucleus (5, 6), yet it releases more gasper unit time at perihelion, even when allowingfor the smaller perihelion distance of Hartley 2(1.059 versus 1.506 AU). This puts Hartley 2 in adifferent class of activity than that of Tempel 1 orany of the other comets visited by spacecraft(fig. S1). The two comets have very differentsurface topography (Fig. 1), but whether the dif-ferent topography is related to the hyperactivity isstill being investigated.

The NucleusSpin state and variations. The rotation state of thenucleus distinguishes the morning from the eve-ning terminator and, in comparison with longer-term coma observations, allows the number andrelative strengths of active areas to be determined.Knowledge of the nuclear spin can also put con-straints on internal distribution of mass in thenucleus, internal energy dissipation, and themag-nitude of the net torque.

The large variations in brightness in Fig. 2,reduced to a measure of the amount of dustleaving the nucleus, show a period of roughly 18hours, but the spacing of peaks in the light curveshows a clear pattern that repeats every three cy-cles. We interpret this [supporting online mate-rial (SOM) text] as an excited state of rotation,with each cycle corresponding to precession ofthe long axis of the nucleus around the angularmomentum vector, with a period of 18.34 hoursat encounter. The pattern of three cycles is dueto an approximate commensurability betweenthis precession and the roll around the long axiswith a period of 27.79 hours (55.42 hours isalso possible; the ambiguity does not affect anyconclusions in this paper). The orientation ofthe angular momentum vector is not yet tightlyconstrained but is within 10° of being perpen-dicular to the long axis. This excited state alsoimplies a nodding motion of the long axis rel-ative to the angular momentum vector, but theobserved near-axial symmetry of the shape lim-its this to an amplitude of <1°. The precessionperiod is increasing at 0.1% per period nearperihelion, which is an unusually high but notunprecedented rate of change for a comet. Theroll period is decreasing. These changes are

RESEARCHARTICLES

1Department of Astronomy, University of Maryland, CollegePark, MD 20742-2421 USA. 2Belton Space Exploration Ini-tiatives LLC, 430 South Randolph Way, Tucson, AZ 85716 USA.3Delamere Support Services, 525 Mapleton Avenue, Boulder,CO 80304, USA. 4Geophysical Institute, University of Alaska–Fairbanks, 903 Koyukuk Drive, Fairbanks, AK 99775–7320, USA.5Max-Planck-Institut für Sonnensystemforschung, Max-Planck-Strasse 2, 37191 Katlenburg-Lindau, Germany. 6Jet PropulsionLaboratory, 4800 Oak Grove Drive, Pasadena CA 91109, USA.7Code 600, NASA Goddard Space Flight Center, Greenbelt, MD20771, USA. 8Institute for Astronomy, University of Hawaii,2680 Woodlawn Drive, Honolulu, HI 96822, USA. 9Lunar andPlanetary Library, University of Arizona, 1629 East UniversityBoulevard, Tucson, AZ 85721–0092, USA. 10Department ofEarth and Atmospheric Sciences, Purdue University, 550 Sta-dium Mall Drive, West Lafayette, IN 47907, USA. 11Departmentof Geological Sciences, Brown University, Providence, RI 02912,USA. 12Department of Astronomy, 312 Space Sciences Build-ing, Cornell University, Ithaca, NY 14853, USA. 13Laboratoired’Astrophysique de Marseille, Universitéde Provence and CNRS,13013 Marseille, France. 14Planetary Science Division, NASAHeadquarters, Mail Suite 3V71, 300 E Street SW,Washington,DC 20546, USA. 15Sigma Space Corporation, 4600 Forbes Bou-levard, Lanham, MD 20706, USA. 16Johns Hopkins University–Applied Physics Laboratory, 11100 Johns Hopkins Road, Laurel,MD 20723, USA. 17LESIA, Observatoire de Paris, UniversitéParis7, Batiment 17, 5 place Jules Janssen, Meudon Principal Cedex92195, France.

*To whom correspondence should be addressed. E-mail:ma@astro.umd.edu

Fig. 1. Comparison of asmall part of (left) Tem-pel 1 with (right) Hartley2 at approximately thesame image scale andwith nearly identical in-struments. (Left) Impac-tor Targeting Sensor (ITS)image iv9000675, 9.1 mpixel–1. (Right) MRI im-age mv5004032, 8.5 mpixel–1. Sun is to the right.

17 JUNE 2011 VOL 332 SCIENCE www.sciencemag.org1396

on

June

17,

201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

presumed to be due to torques produced by theoutgassing.

Nuclear shape. The nucleus is bi-lobed in shape(Fig. 1), with a maximum length of 2.33 km. Theshape is well constrained by stereo viewing ofnearly half the object and for much of the re-mainder sampled by silhouettes against the lightscattered from the coma (Table 1). This bi-lobedshape is crudely similar to that of comet Borrelly(7) but is both relatively and absolutely smoother.The rotation is slow enough that gravity is suf-ficient to hold the two lobes together for any bulkdensity of >100 kg m–3.

Constraints on density. The fast (12.3 km s–1)flyby did not permit determination of the nuclearmass from the spacecraft’s trajectory. The smoothshape of the “waist” region connecting the twolobes might indicate material collecting in a grav-itational low, such as observed on asteroid (25143)Itokawa (8). Collection could proceed by in-fallingmaterial landing in the gravitational low and/orby in situ fluidization of regolith induced by out-flowing gas (9). If some form of frictionless,fluidized flow is responsible for the formation ormodification of this region, it should represent a“flat” surface so that it lies along an equipotentialwith respect to the combined forces of both grav-ity and rotation. Under these assumptions, whichmay not be valid, the density of the nucleus canbe estimated by fitting potential contours to theobserved geometry of the waist (SOM text).

We assumed internal homogeneity, a preces-sion period of 18.34 hours, and a wide range ofdensities. The variance is minimized for a bulkdensity of 220 kg m−3 (fig. S4). Even this min-imum residual leaves large-scale slopes of up to afew degrees relative to the equipotential. A rea-sonable lower limit for the density is 180 kg m−3

because the waist is no longer a gravitational lowfor lower densities. The upper limit is not welldetermined, but a four-times increase above thebest-fit density to 880 kg m−3 requires modestporosity for pure ice and substantial porosity forplausible rock and ice mixtures.

Any interpretation rests on the character ofthe surface at the waist, which is mottled onhorizontal scales of 10 to 30 m and has someisolated cases of local relief >10 m. The mot-tling, local topography, and generally gradationalboundary distinguishes the waist from the best-observed flow on Tempel 1 (1) and the pondedmaterials on (433) Eros (10) and Itokawa (8), atleast locally, but does not rule out the possibilitythat the overall shape of the waist is approxi-mately an equipotential. If the surface in this re-gion approaches an equipotential andwas formedby flows or by deposition similar to those on otherobjects, it has subsequently evolved, suffering sev-eral meters of erosional etching. These facts sug-gest that the equipotential assumption may not bereliable and that, unlike the case for Tempel 1 inwhich a different and more direct approach waspossible (11), the density might be considerablyhigher than deduced under our assumptions.

Geology of surface. The nucleus has two pri-mary terrain types (Fig. 3A): knobby terrain char-acterized by rounded to angular elevated formsup to 50 m high and 80 m wide and relativelysmooth regions occupying both the waist (Fig.3C) between the two lobes and parts of the largerlobe (Fig. 3B). The elevated forms appear to bethe larger members of a population, with mostexamples near or below our practical mappingresolution (~12 m) (Fig. 3D). The smoother areashave darker central regions, are elongate, andare partially bounded by strings of the elevated

forms that make up the knobby terrain. There ismarginally resolvedmottling of the smooth regionson the larger lobe and better resolved mottlingand local topography of 10- to 30-m scale in thewaist. The darkest regions of the larger lobe areslightly more sharply bounded than is the darkerband within the waist. The elevated forms thatconstitute the rough terrain are in some areasaligned along boundaries of albedomarkings andfollow much of the southern edge of the waist.Many of the elevated forms exhibit two to threetimes higher albedo than the average, which is amuch greater range than seen onTempel 1.Ragged,somewhat sinuous, narrow depressions are visi-ble at high-incidence angles near the southern end,about 10 m deep, up to 90 m wide, and extend-ing for over 250 m.

The average geometric albedo of the nucleusis ~4%. Within the larger lobe area are severalroughly equidimensional spots <80 m across thatappear even darker than the larger, more elon-gated “dark” areas mentioned above (Fig. 3E).The darkest unshadowed spots are less than halfthe average brightness. Although we cannot ruleout steep-sided holes for the dark spots near theterminator, they generally occur in regions thatare smooth in stereo and show no shadow sig-natures. Thus, local albedos span at least a factorof 4, compared with <2 on Tempel 1.

Jets occur in all terrains but are clustered inthe rough topography of the smaller lobe andmid- to northern part of the larger lobe (Fig. 3).Such clustering of jets has also been seen atcomet 1P/Halley (12). At least some jets appearto originate at or near large, bright, elevated forms.Jets also originate beyond the terminator in areaswith no direct sunlight, such as along the loweredge of the larger lobe in Fig. 4. Even our res-olution of 10 to 12 m is not sufficient to clearlyresolve the morphology of the sources of the jets.

This comet lacks a population of depressions,such as those that dominate 81P/Wild 2 (13) orthose scattered acrossTempel 1 (1,14). The knobbyterrain is similar to some of the rougher areason Tempel 1. The smoother regions on Hartley2 do not show the striations that are suggestiveof flow markings on the best-observed such re-gion on Tempel 1, and they are more gradation-ally bounded. The combination of terrains is verydifferent fromTempel 1 orWild 2, and this cometlacks exposures of thick, internal layers that wereprominent on Tempel 1.

Nuclear ActivityCN anomaly. Gaseous CN abundances were mea-sured routinely from the start of observations on5 September (SOM text). The long-term CN gasproduction gradually increased from 6 × 1024 s–1

on 5 September to a peak of 3 × 1025 s–1 at peri-helion (28 October), after which it decreasedagain to ~2.4 × 1025 s–1at closest approach and1.2 × 1025 s–1 on 25 November (fig. S5). Duringmost of the encounter, the CN production variedperiodically with the precession of the nucleus, asdid the grains and other gases.

Fig. 2. Variation of visible flux with time in a 191-km square aperture at the comet. Flux is proportionalto the amount of dust leaving the nucleus if physical parameters of the dust do not change (no changes indust color or other properties appear in our data nor are any reported by Earth-based observers). Thedotted box is expanded later (Fig. 7). Near closest approach, the flux from the nucleus is too large to allowaccurate photometry of the coma.

www.sciencemag.org SCIENCE VOL 332 17 JUNE 2011 1397

RESEARCH ARTICLES

on

June

17,

201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

The production of CN also exhibited an anom-alous increase, by a factor of ~7, between 9 and17 September, after which it slowly decreased,returning to the long-term trend line by 24 Sep-tember (fig. S5). There was a very weak increasein dust release [DA(q)fr < 10 cm] above its trendin that same period. There was no correspondingincrease of water (15). Integrating over the periodof 9 to 24 September and subtracting a baselinegas production rate of 7 × 1024 s−1, we found that~2 × 1031 CN radicals (~800 tons) were releasedin the anomaly.

The long-duration, gradual increase and de-crease of gaseous emission without a correspond-ing increase in the dust production is atypical ofcometary outbursts, which have sudden onsetsand are usually accompanied by considerabledust. The increase is unlike the activity observedat 9P/Tempel 1 or the behavior of any other com-et. The spatial profile of CN during the anomalywas very different from that during the rest ofthe observations and suggests formation of CNfrom some extended source other than photo-dissociation ofHCN. This could be grains too darkto scatter much sunlight, such as HCN polymers(16) or the CHON grains found at 1P/Halley (17).They would need to be lifted by something abun-dant and volatile.

Large chunks. Radar observations in Octobershowed an ensemble of particles greater than afew centimeters (18, 19). Although clouds oflarge particles had been reported previously fromradar measurements of other comets (20), thelocation relative to the nucleus and the com-position are not known. Previous searches withremote sensing for an icy grain halo in cometshave rarely been successful, with the only detec-tions being at large heliocentric distances (21–23).

At EPOXI’s close approach, individualchunks were seen near the nucleus in manyimages from both MRI and HRI (Fig. 4). Themotion of the spacecraft allowed determinationof the positions of individual chunks and theirmotions.A sample of 50 chunks has been followedin many different MRI images and, other thanone at 28 km, all were found to be within 15 kmof the large end of the nucleus. The motions areall very slow, with 80%moving at <0.5m s−1 andthe fastest moving at <2 m s−1. This implies min-imum life times of 104 s for many of the chunks.

Escape velocity is poorly defined near the surfacebecause of the rotating, elongated shape as wellas the uncertainty in the mass of the nucleus. Weestimate that 10 to 20% of the chunks are mov-ing at less than their local escape velocity.

Sizes were estimated from the brightness ofthe chunks (SOM text). The apparent flux fromeach of >104 individual chunks ranges from~10−13 to ≲10−11 W m−2 mm−1 in the HRI imagein Fig. 4, which is similar to that measured in theMRI image taken close in time (Fig. 4). Sam-pling below 10−12Wm−2 mm−1 is incomplete.Weconsidered two extreme cases for the scatteringproperties: icy chunks scattering with the albedoand phase function of Europa (24, 25) and dirtychunks scattering with the albedo and phasefunction of the nucleus of comet Tempel 1 (26).The range of measured fluxes corresponds toradii of 10 to 150 cm if they are dirty chunks and1 to 15 cm if they are ice (nearly pure). Below theminimum size, the chunks blend into the back-ground of unresolved, smaller chunks or grains.Because meter-sized objects are at the extreme ofwhat can be lifted by gas drag and because thesmaller, unresolved chunks are demonstrably icy(see below), we argue that the largest chunks areicy and roughly 10 to 20 cm in radius.

The size distribution implied by the fluxes ofthe discrete chunks is unusually steep. Most ofthe mass and most of the cross section are in thesmallest grains. The discrete chunks contributeroughly 4% of the total surface brightness in theinnermost coma (<5 km), and those ≳5 cm (com-pleteness limit) are widely spaced at 4 × 10−8 m−3.If we extrapolate the size distribution (SOM text),>100% of the surface brightness is accounted forjust with chunks >0.5mm. The total cross sectionof discrete chunks, roughly the same size chunks

as ones to which radar is sensitive, is very smallas compared with that detected by radar (18, 19),which presumably is detecting much darkerchunks over a much larger field of view.

Heterogeneity of dominant volatiles. Spectralscans of the comet were obtained from 1 Octoberto 26 November, including several in which thenucleus was spatially resolved. Figure 5 is from ascan taken 7 min after closest approach, with anMRI image taken at nearly the same time. Redboxes show regions where we have extracted thetwo spectra shown in Fig. 6, both of which havehad the continuum manually removed. The ratioof the H2O band to the CO2 band varies spatiallyby 2.9 times. In these maps (Fig. 5), the emissionbands of H2O and CO2 are both somewhat op-tically thick, implying that variations in columndensity could be larger than in brightness.

The maps show a water vapor–rich regionextending roughly perpendicular to the waist ofthe nucleus and presumably arising from thewaist. This region has relatively little CO2, rel-atively little gaseous organics, and thus far, nodetectable water ice. The region of the jets off theend of the smaller lobe of the nucleus is rich inCO2, organics, and water ice but has a lower col-umn density of water vapor than above the waist.There is substantial ice in jets emanating frombeyond the terminator along the lower edge ofthe larger lobe of the nucleus. The boundariesand the direct association with the major units ofthe nucleus seen here are dramatic and suggestvery different histories for the waist and the re-mainder of the nucleus. The coincidence of strongabsorption bands of ice with jets that are bright inthe continuum suggests that jets are bright whenhighly reflective ice is present in the jets and con-versely, that jets are usually fainter than the nucleus

Fig. 3. (A) Hartley 2, im-age mv6000002, ~7 mpixel–1. Sun is to the right,and the positive rotation-al pole is at the smaller(right) end. The view isfrom latitude ~–33°. (B)Relatively smooth regionof larger lobe. (C) Thewaistbetween the two lobes.(D) Knobby terrain, char-acterized by rounded toangular elevated forms upto 50 m high and 80 mwide. (E) An example of aroughly equidimensionalspot <80 m across thatappears even darker thanthe larger,more elongated“dark” areas.

Table 1. Properties of Hartley 2’s nucleus.

Volume 0.82 T 0.08 km3

Diameter 0.69 to 2.33 kmSurface gravity 0.0013 to 0.0033 cm s−2

Precession period 18.34 T 0.04 hours(November 4)

Geometric albedo 4% (average)Mean radius 0.58 T 0.02 kmCross section 0.43 to 1.59 km2

(Assumed density 220 kg m−3)Roll period 27.79 or 55.42 T 0.1 hours

17 JUNE 2011 VOL 332 SCIENCE www.sciencemag.org1398

RESEARCH ARTICLES

on

June

17,

201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

because they are of optically thin, relatively darkmaterial.

Theoretical calculations of scattering by icygrains (SOM text) show that the predominantscattering grains must be smaller than 10 mm.However, >100% of the surface brightness canbe accounted for by extrapolating the chunks to asize of 0.4 mm, implying that the chunks arefluffy aggregates or clusters of ~1-mm solid grains.Either most of the aggregates of order 1 mm havebroken up, or they mimic the scattering of thesmall grains. This result is very similar to the resultobtained at Tempel 1 after the impact (no ice wasobserved before the impact). Those grains werepredominantly micrometer-sized (27). The sim-ilarity between excavated material from Tempel1 and ambient outgassing fromHartley 2 suggeststhat the constituent grains of solid ice are on orderof a micrometer in most comets. On the basis ofcalculations of life times (28–30) for the <10-mmsolid components, the ice must be nearly pure forthe grains to persist.

The detection of strong absorption by ice,the detection of very large chunks in the coma,the concentration of all species other than H2Ovapor away from the waist of the nucleus, andthe relatively smooth surface of the waist leadus to suggest that the material at the waist hasbeen redeposited as a mixture of dirty grains andfluffy, icy aggregates that have not yet sublimed.The warmth of the dirty grains then leads to sub-limation of the icy grains just below the surface.We conclude that this aspect of the chemical het-erogeneity of the nucleus of Hartley 2 is probablyevolutionary.

To determine the absolute abundance ratio,we considered a spectral map made three ro-tations (55 hours) earlier, when both the preces-sion and roll orientations were the same. Spectrawere extracted from 120- and 600-km boxes,both centered on the brightest pixel of thermalemission (a better proxy for the nucleus than areflected light center). In an aperture of 600 by600 km centered on the nucleus (fig. S8), andassuming an outflow speed of 0.5 km s–1, wefound average production rates Q(H2O) = 1.0 ×1028 s−1 and Q(CO2) = 2.0 × 1027 s−1 for ~20%fraction of CO2. This is higher than the fractionobtained in previous measurements of the globalproduction of CO2 in this comet (31–33).

In Fig. 7, we compare a portion of the vi-sual light curve with the variation of CO2 andH2O from the spectral scans. The scale is arbi-trary, so only relative variations are meaningful.The CO2/H2O ratio varies by a factor of 2 be-tween maxima and minima. The lower portionof Fig. 7 shows images of the CO2 and the H2Ofrom the spectral maps. The red line indicatesthe position of the nucleus as defined by thepeak thermal pixel. Close inspection shows thatCO2 is more sunward (up in the figure) thanH2O near the maxima, reflecting the differentspatial distributions. This suggests that theCO2/H2O ratio is less in the large lobe of thenucleus than in the small lobe, but this is a verytentative conclusion until the rotational state isfully understood. If true, this heterogeneity is al-

most certainly primordial, unlike the ambig-uous interpretation for the heterogeneity ofTempel 1 (34).

Summary and ConclusionsComet 103P/Hartley 2 differs inmanyways from9P/Tempel 1 and is an ideal example of hyper-active comets, ones that produce more H2O perunit time than should be possible by sublimationfrom the small surface area of their nuclei. Super-volatiles, specifically CO2 in the case of Hartley2, are the primary drivers of activity. The super-volatiles drag out chunks of nearly pure water-ice,which then sublime to provide a large fraction ofthe total H2O gaseous output of the comet. Otherhyperactive comets include 46P/Wirtanen and21P/Giacobini-Zinner.

Fig. 5. Relative spatial distribution in the coma of Hartley 2. The red boxes (5 by 5 pixels; 52 m pixel–1)indicate regions sampled to produce the spectra in Fig. 6. Panels labeled CO2, Organics, and H2O Vaporare maps of the total flux in the relevant emission bands. The panel labeled H2O Ice is a map of the depthof the ice absorption feature at 3 mm. Each panel has been individually linearly stretched. All spectralimages are from a scan at E+7 min, hi5006000. Sun is to the right.

Fig. 4. (Left) Original HRI image(left, hv5004024, E-66s, range915 km). (Middle) Deconvolvedimage. (Right) MRI context im-age (mv5004029) showing thelocation of the HRI field abovethe large lobe of the nucleus. Ar-rows indicate projected direc-tions to the Sun and Earth.

www.sciencemag.org SCIENCE VOL 332 17 JUNE 2011 1399

RESEARCH ARTICLES

on

June

17,

201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

In Hartley 2, H2O sublimes from the waistwith amuch lower content of CO2 and barely anytrace of icy grains. We tentatively interpret thewaist as a secondary deposit of material, althoughthe mechanism of redeposition remains unclear.The most likely mechanism involves fallback of

both refractories and icy chunks from the periph-ery of the active regions. CO2/H2O varies by afactor 2, probably between one end and the other.

From HST measurements (35), CO is <0.3%,implying a ratio of CO2/CO that is >60, whichis a far more oxidized environment than is con-

sistent with any model of the outer protoplan-etary disk. This is also different from Tempel1 (~1) (36) and Halley (<1) (37). A large anom-aly in CN, too slow to be called an outburst, isunexplained.

References and Notes1. M. F. A’Hearn et al., Science 310, 258 (2005).2. M. F. A’Hearn, M. R. Combi, Eds., Icarus 191, 1 (2007).3. D. L. Hampton et al., Space Sci. Rev. 117, 43

(2005).4. K. P. Klaasen et al., Rev. Sci. Instrum. 79, 091301

(2008).5. O. Groussin, P. Lamy, L. Jorda, I. Toth, Astron. Astrophys.

419, 375 (2004).6. C. M. Lisse et al., Publ. Astron. Soc. Pac. 121, 968

(2009).7. L. A. Soderblom et al., Science 296, 1087 (2002).8. A. Fujiwara et al., Science 312, 1330 (2006).9. M. J. S. Belton, J. Melosh, Icarus 200, 280 (2009).

10. M. S. Robinson, P. C. Thomas, J. Veverka, S. Murchie,B. Carcich, Nature 413, 396 (2001).

11. J. E. Richardson, H. Melosh, C. Lisse, B. Carcich, Icarus190, 357 (2007).

12. H. U. Keller et al., Astron. Astrophys. 187, 107 (1987).13. D. E. Brownlee et al., Science 304, 1764 (2004).14. P. C. Thomas et al., Icarus 187, 4 (2007).15. M. R. Combi et al., Astrophys. J. 734, L6 (2011).16. C. N. Matthews, R. Ludicky, Adv. Space Res. 12, 21

(1992).17. J. Kissel et al., Nature 321, 336 (1986).18. J. K. Harmon et al., IAU Circ. 9179, 1 (2010).19. J. K. Harmon et al., Astrophys. J. 734, L2 (2011).20. J. K. Harmon et al., in Comets II, M. C. Festou et al., Eds.

(Univ. of Arizona Press, Tucson, AZ, 2005), p. 274.21. M. F. A’Hearn, E. Dwek, A. T. Tokunaga, Astrophys. J.

248, L147 (1981).22. H. Campins, G. H. Rieke, M. J. Lebofsky, Nature 301, 405

(1983).23. M. S. Hanner, Astrophys. J. 277, L78 (1984).24. W. M. Grundy et al., New Horizons Team, Science 318,

234 (2007).25. B. Buratti, J. Veverka, Icarus 55, 93 (1984).26. J.-Y. Li et al., Icarus 187, 141 (2007).27. J. M. Sunshine et al., Icarus 190, 284 (2007).28. H. Patashnick, G. Rupprecht, Icarus 30, 402 (1977).29. M. S. Hanner, Icarus 47, 342 (1981).30. H. I. M. Lichtenegger, N. I. Kömle, Icarus 90, 319

(1991).31. H. A. Weaver et al., Astrophys. J. 422, 374 (1994).32. L. Colangeli et al., Astron. Astrophys. 343, L87 (1999).33. J. Crovisier et al., ESA SP 427, 161 (1999).34. L. M. Feaga, M. Ahearn, J. Sunshine, O. Groussin,

T. Farnham, Icarus 190, 345 (2007).35. H. A. Weaver et al., Astrophys. J. 734, L5 (2011).36. L. M. Feaga et al., Bull. Am. Astron. Soc. 38, 450

(2007).37. D. Bockelée-Morvan et al., in Comets II, M. C. Festou et al.,

Eds. (Univ. of Arizona Press, Tucson, AZ, 2005), p. 391.Acknowledgments: Data from EPOXI will be released

through NASA’s Planetary Data System in 2011. ImageIDs in this paper are part of the final ID in the archive.This work was supported by NASA’s Discovery Programcontract NNM07AA99C to the University of Maryland andtask order NMO711002 to the Jet Propulsion Laboratory.The work was supported by the home institutions ofseveral of the scientists, particularly by the Universityof Maryland. The contributions of O. Groussin andF. Merlin to this project were funded in part by theCentre National d’Etudes Spatiales.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/332/6036/1396/DC1SOM TextFigs. S1 to S8Table S1

9 February 2011; accepted 5 May 201110.1126/science.1204054

Fig. 6. Absolute spectra in the coma of Hartley 2 at CA+7 min. The black curve is the water-rich but ice-free spectrum from the box in the H2O vapor panel in Fig. 5, above the waist of the nucleus. The red curveis the ice-rich spectrum from the box in the H2O ice panel in Fig. 5, the jet region near the small end of thenucleus.

Fig. 7. Light curve in visible light compared with the light curve in H2O and CO2. The curve for grains isan expanded version of the box in Fig. 2. The data points for gas are derived from spectral maps. The lowerportion shows an expanded view, with images of the CO2 and H2O for each point in the dotted box. Thecentroid of reflected continuum in each scan has been aligned with the red, horizontal line. Sun is up inthe small images of gas.

17 JUNE 2011 VOL 332 SCIENCE www.sciencemag.org1400

RESEARCH ARTICLES

on

June

17,

201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from