The Formation and Evolution of Planetary Systems: Placing Our Solar System in Context with Spitzer Michael R. Meyer Steward Observatory, The University of Arizona [email protected] 7 Lynne A. Hillenbrand 0 0 California Institute of Technology 2 Dana Backman n a SOFIA/SETI Institute J 3 Steve Beckwith 1 Space Telescope Science Institute v 8 Department of Physics and Astronomy, Johns Hopkins University 5 0 Jeroen Bouwman 1 0 Max–Planck–Institut fu¨r Astronomie, Heidelberg 7 0 Tim Brooke / h p California Institute of Technology - o John Carpenter r t s California Institute of Technology a : v Martin Cohen i X Radio Astronomy Laboratory, University of California–Berkeley r a Stephanie Cortes Steward Observatory, The University of Arizona Nathan Crockett National Optical Astronomy Observatories Uma Gorti NASA-Ames Research Center, Theory Branch Thomas Henning Max–Planck–Institut fu¨r Astronomie, Heidelberg Dean Hines Space Science Institute – 2 – David Hollenbach NASA-Ames Research Center, Theory Branch Jinyoung Serena Kim Steward Observatory, The University of Arizona Jonathan Lunine Lunar and Planetary Laboratory, The University of Arizona Renu Malhotra Lunar and Planetary Laboratory, The University of Arizona Eric Mamajek Harvard–Smithsonian Center for Astrophysics Stanimir Metchev The University of California, Los Angeles Amaya Moro-Martin Department of Astrophysical Science, Princeton University Pat Morris Herschel Science Center, IPAC Joan Najita National Optical Astronomy Observatories Deborah Padgett Spitzer Science Center Ilaria Pascucci Steward Observatory, The University of Arizona Jens Rodmann Max–Planck–Institut fu¨r Astronomie, Heidelberg Wayne Schlingman Steward Observatory, The University of Arizona Murray Silverstone Steward Observatory, The University of Arizona David Soderblom Space Telescope Science Institute – 3 – John Stauffer Spitzer Science Center Elizabeth Stobie Steward Observatory, The University of Arizona Steve Strom National Optical Astronomy Observatories Dan Watson Department of Physics and Astronomy, The University of Rochester Stuart Weidenschilling Planetary Science Institute Sebastian Wolf Max–Planck–Institut fu¨r Astronomie, Heidelberg Erick Young Steward Observatory, The University of Arizona ABSTRACT WeprovideanoverviewoftheSpitzerLegacyProgram“FormationandEvolutionofPlanetary Systems” (FEPS) which was proposed in 2000, begun in 2001, and executed aboard the Spitzer SpaceTelescopebetween2003and2006. ThisprogramexploitsthesensitivityofSpitzertocarry out mid-infrared spectrophotometric observations of solar-type stars. With a sample of ∼328 starsranginginagefrom∼3Myrto∼3Gyr,wetracetheevolutionofcircumstellargasanddust fromprimordialplanet-building stagesin youngcircumstellardisksthroughto oldercollisionally generateddebrisdisks. Whencompleted,ourprogramwillhelpdefine thetimescalesoverwhich terrestrialand gas giant planets are built, constrain the frequency of planetesimal collisions as a function of time, and establish the diversity of mature planetary architectures. In addition to the observational program, we have coordinated a concomitant theoretical effortaimedatunderstanding the dynamics ofcircumstellardust with andwithout the effects of embeddedplanets,dustspectralenergydistributions,andatomicandmoleculargaslineemission. Togetherwiththeobservations,these effortswillprovideastronomicalcontextforunderstanding whether our Solar System – and its habitable planet – is a common or a rare circumstance. Additional information about the FEPS project can be found on the team website: \protect\vrule width0pt\protect\href{http://feps.as.arizona.edu/}{http://feps.as.arizona.edu/} Subject headings: infrared: general, space vehicles: Spitzer Space Telescope, surveys, stars: circumstellar matter – 4 – 1. Introduction The Spitzer Space Telescope (Werner et al. 2004), formerly SIRTF the Space InfraRed Telescope Facility, is an 85 cm cryogenic space observatory in earth–trailing orbit. The observatory was launched in August of 2003 and has an estimated mission lifetime of 5+ years. There are three science instruments on–board: IRAC (Fazio et al. 2004),IRS (Houck et al. 2004),and MIPS (Rieke et al. 2004)which together provide the capability for imaging and spectroscopy from 3.6–160 µm. The Legacy Science Program was establishedbeforelaunchwithtwogoals: toenablelargescaleprogramsofbroadscientificandpublicinterest, andtoprovideaccesstouniformandcoherentdatasetsasrapidlyaspossibleinsupportofGeneralObserver (GO)proposals,giventhelimitedlifetimeofthemission. TheFormationandEvolutionofPlanetarySystems (FEPS) Spitzer Legacy Science Program is one of six such original programs (for descriptions of the others see Evans et al. 2003; Benjamin et al. 2003; Kennicutt et al. 2003; Londsdale et al. 2003; Dickenson et al. 2003)anduses350hoursofSpitzerobservingtime. FEPSbuildsupontherichheritageofSpitzer’sancestors in space: the international all-sky mid-infrared survey telescope IRAS (the Infrared Astronomical Satellite, 1983–1985)and ESA’s pointed mission ISO (the Infrared Space Observatory,1995–1999),and complements Guaranteed Time Observer (GTO) and GO programs also being pursued with Spitzer. Inasinglesentence,FEPSisacomprehensivestudyoftheevolutionofgasanddustinthecircumstellar environment. The scientific motivation for FEPS lies in the fragmented but compelling evidence for dusty circumstellarmaterialsurroundingstarsspanning a wide rangeofages,fromyoung pre-mainsequence stars to those as old as, and even older than, the Sun. At young ages, incontrovertible evidence assembled over the past three decades (based on data from ultraviolet through millimeter wavelengths), suggests that most stars are surrounded at birth by accretion disksthatareremnantsofthestarformationprocessitself(e.g. BeckwithandSargent,1996). Therevelations provided by IRAS, and later ISO, led to a nearly complete census of optically–thick disks within 100-200 pc, and in the case of ISO revealed their rich dust mineralogy and gas content (see Lorenzetti 2005 and Molster & Kemper 2005 for reviews). That at least some of these disks build planets has become clear from radial velocity and photometric studies revealing M sini=0.02-15 M planets orbiting well over one J hundred nearby stars (e.g. Marcy et al. 2005). At older ages, IRAS and ISO revealed the presence around dozens of main sequence stars of micron-sized grains. These dusty “debris” disks are produced in collisions between asteroid-likebodies with orbits that are dynamically stirred by planets (e.g. Lagrangeet al. 2000). Subsequently, several of these disks were spatially resolved at optical, infrared, and millimeter wavelengths (e.g. Kalas,Liu,&Mathews2004;Weinbergeretal. 1999;Greavesetal. 1998)revealingstructureconsistent with the planetary perturber interpretation. Theconnectionsbetweenplanets,debrisdisks,andthedustyandgaseousdisksnear-ubiquitouslyfound in association with recently formed young stars are tantalizing, but not yet unequivocally established. Un- derstanding the evolution of young circumstellar dust and gas disks as they transition through the planet– building phase requires the hundred-fold enhancement in sensitivity and increased photometric accuracy offered by Spitzer at mid- and far-infrared wavlengths. For main sequence stars, while IRAS discovered the prototypical debris disks (see Backman & Paresce 1993 for a review) and ISO made additional surveys (see de Muizon 2005 for a review), neither IRAS nor ISO was sensitive enough to detect dust in solar systems older than a few hundred Myr for any but the nearesttens of stars 1. Spitzer, by contrastcan detect orders of magnitude smaller dust masses: for a solar-type star at 30 pc, down to ∼ 1020 kg or ∼ 10−5 M in Earth 1TheIRASandISOobservatories wereabletostudyrepresentativesamplesofA-typestars,butnotG-typestars. – 5 – micron to sub-millimeter size grains at 50 K, only an order of magnitude above the dust mass inferred for our own present-day Kuiper Belt, and ∼ 1017 kg or ∼ 10−8 M at 150 K, only an order of magnitude Earth above the dust mass in our present-day asteroid belt plus zodiacal cloud. The FEPS program is designed to study circumstellar dust properties around a representative sample ofsolar-typestars. Includedare328starschosentoprobethesuspecteddirectlinkbetweendiskscommonly foundaroundpre–mainsequencestars<3.0Myroldandour4.56GyroldSunandSolarSystem. Specifically, we trace the evolution of circumstellar material at ages 3-10 Myr when stellar accretion from the disk terminates, to 10-100 Myr when planets achieve their final masses via coalescence of solids and accretion of remnantmoleculargas,to100-1000Myrwhenthefinalarchitectureofsolarsystemstakesformandfrequent collisions between remnant planetesimals produce copious quantities of dust, and finally, to 1-3 Gyr mature systems in which planet-driven activity of planetesimals continues to generate detectable dust. Our sample is distributed uniformly in log-age from 3 Myr to 3 Gyr. We probe the full range of dust disk optical– depths diagnostic of the major phases of planet system formation and evolution, including primordial disks (those dominated by ISM grains in the process of agglomeratinginto planetesimals) and debris disks (those dominated by collisionally generated dust) like our own. Ourstrategyistoobtainforall328starsinoursamplecarefullycalibratedspectralenergydistributions (SEDs) using all three Spitzer intstruments. A high-resolution spectroscopic survey limited to the younger targets establishes the gas content. In addition to insight into problems of fundamental scientific and philosophicalinterest,theFEPSLegacyScienceProgramprovidesarichdatabaseforfollow-upobservations with Spitzer, with existing and future ground-based facilities, as well as SIM and JWST, and eventually TPF. FEPS complements and motivates many existing Spitzer GTO and GO programs. 2. Science Strategy We take advantage of Spitzer’s unprecedented mid-infrared sensitivity and hence its unique ability to detect the photospheres of solar-type stars out to distances of several tens of pc. Spitzer observations of excesses above those photospheres are indicative of dust located at a range of orbital separations, from analogs of the terrestrial zone, to the gas-giant zone, to the Kuiper belt zone in our own Solar System. Such observations are important in the search for exo-solar planetary systems – either those in formation from primordial dust and gas disks, or those which later perturb planetesimals into crossing orbits that collisionally cascade to produce dusty debris disks. For a given system, the mass in small grains to which Spitzer is sensitive is first expected to decrease with time as planet formation begins, then increase on a relatively rapid time scale (few Myr) as the debris phase begins, and finally decrease on a much longer time scale (many Gyr) as the disk slowly grinds itself down and grains are removedvia radiative and mechanical effects. With the goal of understanding how common or rare the evolutionary path taken by our Solar System might have been, we have initiated a Spitzer survey of F–G–K (solar-type) stars. First, we study the formationofplanetaryembryosinasurveyofpost–accretioncircumstellardustdisks. Weaimtounderstand theevolutionofdiskproperties(massandradialstructure)anddustproperties(sizeandcomposition)during the main phase of planet–building and early solar system evolution from 3–100 Myr. Second, we study the growth of gas giants in a sensitive search for warm molecular gas at the level of >2 × 10−4 M⊙[H2] at 70–200 K, in a sub–sample of the targets from our dust disk survey. Our goal is to constrain directly the time available for embryonic planets to accrete gas envelopes. Third, we investigate mature solar system – 6 – evolutionbytracing100Myrto3Gyrolddustdisksgeneratedthroughcollisionsofplanetesimals. Through our analysis we hope to infer the locations and masses of giantplanets >0.05 M ≈ 1 M through Jupiter Uranus their action on the remnant disk. Our large sample enables us to measure the mean properties of evolving dust disks and discover the dispersioninevolutionarytimescales. Further,wecansearchforrelationsbetweeninferreddustevolutionary path and stellar properties such as metallicity and multiplicity. In the following subsections we detail our science strategy. 2.1. Formation of Planetary Embryos Our experiment begins as the disks are making the transition from optically thick to thin, the point at which all of the disk’s mass first becomes detectable through observation. The FEPS goals are to: • constrain the initial structure and composition of post-accretion, optically thin disks; • trace the evolution of disk structure, composition, and mass over time; • characterize the time scales over which primordial disks dissipate and debris disks arise; • measure changes in the dust particle size distribution due to coagulation of interstellar grains at early stages and shattering associated with high-speed planetesimal collisions at later early-debris stages; • infer the presence of newly formed planets at orbital radii of 0.3-30 AU. Photometric observations from 3.6-160µm with Spitzer probe temperatures (radii) encompassing the entire system of planets in our Solar System. In Figure 1 we show the mass sensitivity of Spitzer as a function of wavelength, indicating the mass in small grains that Spitzer can detect as a function of orbital radius. Detailed spectrophotometry in the range 5.3–40 µm permits a search for gaps in disks caused by the dynamicalinteractionof young gas giantplanets and the particulate disk from 0.2–10AU. This extends from just outside the innermost radii of the exo-solar “hot Jupiters” (thought to have suffered significant orbital migrationin a viscous accretion disk) to the gas giants of our Solar System (thought to have formed beyond the “ice line”). Mid–infrared spectroscopic observations are sensitive to dust properties including size distribution and composition. They thus probe physical conditions in the disk. From observations in the 5.3–40µm spectral regionwe determine the relative importance of broadfeatures attributed to amorphous silicates (ubiquitous in the ISM) compared to numerous narrow features due to crystalline dust (observed only in circumstellar environments). In this way, we can look for evidence of, e.g. radial mixing in the disk since the tem- perature required to anneal grains (> 1000K) is substantial higher than that inferred for the continuum emitting material (∼ 300K). Further, the shape and strength of specific mid-infrared spectroscopic features provideconstraintsonthe fractionalcontributionofeachgrainpopulationtothe totalopacity,necessaryfor estimating dust mass surface densities (see section 4.2). 2.2. Growth of Gas Giants Next,wehaveundertakenthemostsensitivesurveytodateofatomicandmoleculargasinpost–accretion disksystems. Inordertocharacterizegasdissipationandtoplacelimitsonthetimeavailableforgiantplanet formationwe obtainhigh spectralresolution(R=600)data from10–37µm with the IRS of35 starsselected – 7 – from our dust disk survey sample. The data include the S(0) 28.2 µm, S(1) 17.0 µm, and S(2) 12.3 µm H lines as well as strong atomic lines such as [SI] 25.23 µm, [Si II] 34.8 µm, and [FeII] 26 µm (Gorti & 2 Hollenbach2004;Hollenbachetal. 2005). Wefocusonthepost-accretionepochsfrom3–100Myrtoexamine whether gas disks persist after disk accretion onto the star has ceased and planetesimal agglomeration has removed the dust disk, potentially providing “nucleation sites” for gas giant planet formation. Understanding gas–dustdynamics is crucialto our ability to derive the time scales importantin planet formation and evolution. The dust and gas experiments being conducted at young ages (<100 Myr) have an important synergy in furthering this understanding because dust dynamics are controlled by gas drag rather than radiation pressure when the gas-to-dustmass ratio is >0.1, while it is the presence of dust that mediates gas heating and therefore detectability. If the gas to dust ratio is low (the dust opacity per gas particlehigh)andthegasanddustareatsimilartemperatures,thedetectionofgaslinesbySpitzerbecomes difficult due to the small ratio of line to continuum. However,the theoreticalmodels of Gorti & Hollenbach (2004) show that in many instances the dust opacity is sufficiently low and the gas temperature sufficiently high(>100K)thatsmallquantities(<0.1M )ofgas,ifpresent,canbedetectedbySpitzeraroundnearby J disks with optically thin dust (see section 4.4). 2.3. Mature Solar System Evolution Finally, we conduct a study of second generation “debris disks”. The presence of any observable cir- cumstellar dust around stars older than the maximum lifetime of a primordial dust disk (the sum of the to–be–determined gas dissipation time scale and the characteristic Poynting–Robertson drag time scale) provides compelling evidence not only for large reservoirs of planetesimals colliding to produce the dust, butalsoforthe existenceofmassiveplanetarybodies thatdynamicallyperturbplanetesimalorbitsinducing frequent collisions. We have undertaken the first comprehensive survey of F5–K5 stars with ages 100 Myr to 3 Gyr that is sensitive to dust disks comparableto those characteristic of our ownSolar System throughoutits evolution. We chart the history of our Solar System from 100–300 Myr, the last phase of terrestrial and ice giant ((Uranus-andNeptune-like)planet–building,through0.3–1Gyr,bracketingthe “lateheavybombardment” impact peak which might have had an effect on the early evolution of life on Earth, and finally over 1.0– 3.0 Gyr, examining the diversity of evolutionary paths among mature planetary system. Spectroscopic observations in the range 5.3–40 µm enable diagnosis of gaps caused by giant planets and estimates of dust size and composition which translate directly into constraints on the mass opacity coeffients for the dust (Miyake & Nakagawa,1993) as well as Poynting-Robertsondrag time scales (Backman e& Parsece, 1993). 3. Survey Preparation and Execution 3.1. Observing Strategy A complete and uniform set of Spitzer photometric and spectroscopic observations are obtained for all stars in our dust disk evolution sample, as described below. To derive statistically meaningful results on the disk and dust properties, we observe ∼50 stars in each of 6 logarithmically spaced age bins from 3 Myr (connecting our Legacy program to that of Evans et al.) to 3 Gyr (beyond which there is strong emphasis byGTOsondebrisdiskscience; Beichmanetal. 2005). Ourtargetsspananarrowmassrange(0.7-2.2M⊙) – 8 – andare proximateenoughto enable a complete census for circumstellardust comparableto our model solar system as a function of age. We measure the stellar photosphere at SNR>30 for λ 3.6–24 µm, SNR > 5 at 70 µm (or SNR > 5 for 5 × the current zodiacal dust emission) with broadband photometry from IRAC and MIPS (subject to calibration uncertainties). To identify gaps in the dust distribution created by the presence of giant planets from 0.2–10AU, we require relative spectrophotometry with SNR>30 from5.3–11 µm, and with SNR ∼ 6–12 at wavelengths between 20–30 µm with the IRS. Asub-sampleof35starscomprisesourgasdiskevolutionstudywiththehighresolutionmodeoftheIRS. Thissamplespansarangeofspectraltype(F3–K5),age(3–100Myr),activity(L /L ∼10−3−10−5),and x bol a wide range of infraredexcess emission, with some preference for optically–thin excess in the mid–infrared. Fourteen stars were chosen for first-look observations and enable us to explore the limits implied by null results and guide our choice of follow–up observations for additional stars drawn from our dust disk survey. Our goal is to collect data capable of realizing the fundamental limits imposed by instrument stability and systematic calibrationuncertainties. Integrationtimes are chosen accordingto each star’s distance, age and spectraltype to reachuniform SNR at the photospheric limits, as specified above– thereby providinga complete census of dust disks for our targets. 3.2. Sample Selection The source list for FEPS consists of young near-solar analogs, stars ranging in mass from 0.7-2.2 M⊙ though strongly peaked at 1 M⊙ (see Figure 2), and spanning ages 3 Myr to 3 Gyr (our Sun is 4.56 Gyr old). The stars are drawn from three recently assembled samples. First, Soderblom et al. (1999) have produced a well-characterized set of ∼5000 solar-type stars spread over the entire sky (see also Henry et al. 1996) having parallaxes that place the stars within 60 pc, (B-V) colorsbetween0.52and0.81(F8-K0spectraltypes),andlocationintheHertzsprung-Russelldiagramwithin 1.0 mag of the solar-metallicity Zero-Age Main Sequence. This sample is fully complete out to 50 pc. The age distribution in such a volume-limited region around the Sun is roughly flat in linear age units out to 2.5 Gyr, at which point heating by the Galactic disk has increased the scale height of older stars and thus removedthemfromthevolume-limitedsample. Fromthiscatalogwehaveselectedasamplewithagesbased on the R’ chromosphericactivity index from ∼100Myr to 3 Gyr. However,being locatedmore than 100 HK pcfromthenearestsitesofrecentstarformation,theimmediatesolarneighborhoodisdeficientinstarswith ages youngerthan 100Myr. Hence the volume limit was extended in order to identify largeenough samples of young stars for the FEPS project. We have conducted a new (e.g. Mamajek et al., 2002) and literature-based examination to identify stars whose ages are in the range 3-100 Myr. These were selected as having (B-V) colors between 0.58 and 1.15 or spectral types G0-K0, strong x-ray emission, kinematics appropriate for the young galactic disk, and high lithium abundance compared to the 120 Myr old Pleiades. Young stars are copious coronal x-ray emitters and a large body of literature demonstrates the connection between x-ray emission, chromospheric activity, stellar rotation, and age. The surface density distribution of x-ray sources detected by the ROSAT all-sky survey reveals a concentration of objects coincident with Gould’s Belt, a feature in the distant solar neighborhood (50-1000pc) comprised of an expanding ring of atomic and molecular gas of which nearly all star-forming regions within 1 kpc are a part. These x-ray sources are thought to be the dispersed low-mass counterparts to a series of 1-100 Myr-old open clusters and extant and fossil OB associations that delineate Gould’s Belt (e.g. Torra et al. 2000; Guillout 1998). Proper motion data enable us to select the nearest – 9 – Fig. 1.— Spitzer sensitivity to mass in small grains as a function of radius in a hypothetical circumstellar disk surrounding a sun–like star at a distance of 30 pc for integration times typical for FEPS. For emission fromsmallgrainsinradiativeequilibrium,eachradiusinthediskcorrespondstoaspecificdusttemperature. From simple blackbody considerations, shorter Spitzer wavelengths probe warmer dust (at smaller orbital radii) while longer Spitzer wavelengths probe cooler dust (at larger orbital radii), as indicated. Fig. 2.— Distribution of masses for stars in the FEPS sample. The range spans 0.7–2.2 M⊙ though it is strongly peaked at 1.0 M⊙. – 10 – of these young, x-ray-emitting stars with space motions consistent with those of higher mass stars having measured parallax, and hence estimate their distances. Follow-up optical spectroscopy of these x-ray + proper motion selected stars is used to confirm youth and determine photospheric properties. A total of ∼600 field stars are x-ray-selectedcandidate young stars. Finally,starsinnearbywell-studiedopenclusters[IC2602(55Myr),AlphaPer(90Myr),Pleiades(125 Myr), Hyades (650 Myr)] serve to “benchmark” our field star results by providing samples nearly identical in age, composition, and birth environment. We considered all known members of these clusters meeting our targetted mass / B-V color / spectral type range that were not part of GTO samples. From this large parent sample, stars were selected for potential observationwith Spitzer if they met all ofthe followingadditionalcriteria. The criteriawerechosentoensuresufficientsignal-to-noiseonthe stellar photosphere out to 24 µm with Spitzer and thus accurate characterization of the underlying photosphere both observationally and with stellar models. • K <10 mag (young <100 Myr x-ray selected and cluster samples) or K <6.75 mag (older 0.1-3 Gyr Hipparcos + R’HK selected sample) • 24 µm background <1.70 mJy/arcsec2 (x-ray selected samples) or <1.54 mJy/arcsec2 (Hipparcos + R’HK sample) • 70 µm background <0.76 mJy/arcsec2 • galactic latitude |b|>5◦ (stars in IC 2602 were permitted to violate this criterion) • quality 2MASS JHK photometry, with no flags • no projected 2MASS companions closer than 5” • no projected2MASS companionscloser than15” unless they are both bluer in J-K and fainter in K by >3 mag than the Spitzer target. These critera were applied uniformly to our parent sample though in the cases of a few exceptional stars (such as the IC 2602 sample) some criteria were violated. Next, targets appearing on Spitzer GTO programswereremovedfromthe sourcelist. Also,to a limited degree,starsidentified throughspectroscopy orhighresolutionimagingliteraturepublished throughMarch2001asbeing binary,with companionscloser than 2”, were removed. These cases were all either spectroscopicbinaries or visual binaries with small delta magnitudes and the literature search was not exhaustive. Subsequent investigation using AO imaging have uncovered additional binary systems with larger delta magnitudes remaining within the FEPS sample (e.g. Metchev & Hillenbrand, 2004). Finally,amongstthestarsinourparentsampleolderthan∼600Myr,approximately1/2werearbitrarily removed from our program in order to even out the age bins and bring the observing program within the allocated number of Spitzer hours (350). Based on pre-Spitzer spectral energy distributions assembled from the literature, 2MASS, IRAS, ISO, and ancillary observations conducted to date, several (5-7) of the youngest stars in our programshow some hint of circumstellar material. Because the stars were randomly selected based on their kinematic and activitypropertiesasderivedfromopticalinformation,thesepreviouslyknowndustexcessesdonotbiasthe program.