Astronomy&Astrophysicsmanuscriptno.aamontecarloivarxiv (cid:13)c ESO2012 January6,2012 Extrasolar planet population synthesis IV. Correlations with disk metallicity, mass and lifetime C.Mordasini1,Y.Alibert2,3,W.Benz2,H.Klahr1,andT.Henning1 1 Max-Planck-Institutfu¨rAstronomie,Ko¨nigstuhl17,D-69117Heidelberg,Germany 2 PhysikalischesInstitut,UniversityofBern,Sidlerstrasse5,CH-3012Bern,Switzerland 3 InstitutUTINAM,CNRS-UMR6213,ObservatoiredeBesanc¸on,BP1615,25010Besanc¸onCedex,France Received27.05.2011/Accepted27.12.2011 ABSTRACT 2 1 Context.Thisisthefourthpaperinaseriesshowingtheresultsofplanetpopulationsynthesiscalculations.InPaperI,wepresented 0 our methods. In Paper II, we compared statistically the synthetic and the observed planetary population. Paper III addressed the 2 influencesofthestellarmassonthepopulation. Aims.Ourgoalinthisfourthpaperistosystematicallystudytheeffectsofimportantdiskproperties,namelydiskmetallicity,mass n andlifetimeonfundamentalpropertiesofplanetslikemassandsemimajoraxis. a Methods.Foralargenumberofprotoplanetarydiskswhichhavepropertiesfollowingdistributionsderivedfromobservations,we J calculateapopulationofplanetswithourformationmodel.Themodelisbasedontheclassicalcoreaccretionparadigmbutincludes 4 self-consistentlyplanetmigrationanddiskevolution. ] dRieffseureltnst.rWoleesfi:nFdoarhviegrhymlaregtaellnicuimtiebse,rgoifanctorprlealnaetitosnasr:eRmeograerdfirnegquthenetp.lFaonrethairgyhinMitial,mgiaasnstfpulnacnteiotsna,rmeemtaolrliecimtya,sMsivdiesk.Fanodrlτodnisgkτhave, P disk disk giantplanetsarebothmorefrequentandmassive.Atlowmetallicities,verymassivegiantplanetscannotform,butotherwisegiant E planetmassandmetallicityarenearlyuncorrelated.Incontrast,(maximal)planetmassesanddiskgasmassesarecorrelated.The h. formationofgiantplanetsispossibleforinitialplanetesimalsurfacedensitiesΣS ofatleast6g/cm2 at5.2AU.Thesweetspotfor p giantplanetformationisat∼ 5AU.In-andoutsidethisdistance,higherΣS arenecessary.Lowmetallicitiescanbecompensated - byhigh Mdisk,andviceversa,butnotadinfinitum.Atlowmetallicities,giantplanetsonlyformoutsidetheiceline,whileathigh o metallicities,giantplanetformationoccursthroughoutthedisk.Theextentofmigrationincreaseswith M andτ andusually disk disk r decreaseswithmetallicity.Noclearcorrelationofmetallicityandthesemimajoraxisdistributionofgiantplanetsexistsbecausein t s lowmetallicitydisks,planetsstartfurtherout,butmigratemore,whilethecontraryappliesforhighmetallicities.Thefinalsemimajor a axisdistributioncontainsanimprintoftheiceline.Close-inlowmassplanetshavealowermeanmetallicitythanHotJupiters.The [ frequencyofgiantplanetsvariesapproximatelyasM1.2 andτ2 . disk disk Conclusions.Thepropertiesofprotoplanetarydisks-theinitialandboundaryconditionsforplanetformation-aredecisiveforthe 1 propertiesofplanets,andleavemanyimprintsinthepopulation. v 6 Keywords.Stars:planetarysystems–Stars:planetarysystems:formation–Stars:planetarysystems:protoplanetarydisks–Planets 3 andsatellites:formation–Solarsystem:formation 0 1 . 1 1. Introduction (2010),hereafterPaperIII,wediscussedtheinfluenceofthestel- 0 larmassonthesyntheticplanetarypopulation,studyingforex- 2 Thenumberofknownextrasolarplanetshasgrownlargeenough ampletheeffectontheplanetaryinitialmassfunction,thesemi- 1 tolookatthestatisticalpropertiesofthepopulationasawhole, major axis distribution or the “metallicity effect” (the increase : rather than at the properties of single objects, and to compare v ofthedetectionprobabilityofgiantplanetswithmetallicity).In theactualpopulationwithasyntheticpopulationobtainedfrom i thisPaperIV,wefocusbackontosolar-likestars. X a theoretical planet formation model. In this way all discov- Overthelastyears,observationalconsiderableprogresswas r eredplanetsserve(providedthedetectionbiasisknown)tocon- a strainthemodel,andtoimproveourunderstandingofplanetfor- achieved in the characterization of the end products of plane- tary formation process, i.e. the planet themselves. Progress has mation. We used our extended core accretion formation model also been substantial in the characterization of the initial and (Alibert et al. 2005a) to generate in a Monte Carlo way popu- boundary conditions for this process, i.e. the properties of pro- lations of synthetic extrasolar planets (Mordasini et al. 2009a, toplanetary disks. This was made possible to a large measure hereafter Paper I). Then, we compared the detectable synthetic thankstonewobservationalfacilitieslikeSpitzer(e.g.Fangetal. planetswithanobservationalcomparisonsampleofactualexo- 2009).Observationsofdisksaroundyoungstarshaveprovided planets using statistical methods (Mordasini et al. 2009b, here- uswithknowledgeofthedistributionsofdiskmasses(Beckwith after Paper II). We found that we could reproduce in a statisti- &Sargent1996;Andrewsetal.2009),disksizes(McCaughrean cally significant way some of the most important properties of & Odell 1996; Andrews et al. 2010) and lifetimes (Haisch et the observed extrasolar giant planets. Finally, in Alibert et al. al.2001;Fedeleetal.2010).Additionally,correlationsbetween Send offprint requests to: Christoph MORDASINI, e-mail: disk properties and stellar mass were discovered (Kennedy & [email protected] Kenyon 2009; Mamajek 2009), which have important implica- 2 C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV tionsfortheformationofplanetsaroundstarsofdifferentmasses themareknown.Foractualexoplanetsthisisobviouslynotthe (Currie2009,PaperIII).AswasshowninPaperI,thediversity case.Observationally,thehoststarsmetallicityonlycanbede- of extrasolar planets is a direct consequence of the diversity of termined and assumed to be a proxy for the disks dust-to-gas protoplanetary disks, which means that the properties of disks ratio f . Other fundamental parameters of the disks, such as D/G are critical in defining the outcome of planet formation taking theirinitialgasmass(whichtogetherwith f setstheabsolute D/G placeinsuchdisks. amount of solids), and their lifetime, cannot be observationally In this article we focus on correlations between disk and determined. It is therefore no surprise that the observationally planetary properties. We study systematically the influences of inferredcorrelationsinvolveessentiallyonlythemetallicity. diskmetallicity[Fe/H],disk(gas)mass M andlifetimeτ disk disk Arecentlydiscussedcorrelationthatcouldpartiallychange onimportant,observablepropertiesofsyntheticplanetsbycom- thisisthepossibleenhancedlithiumdepletionmeasuredinSun- putingalargenumberofmodelsofplanetformation. likestarshostingplanets(Israelianetal.2009,butseeBaumann etal.(2010)foranopposedview).Thisisbecausesuchadeple- 1.1. Observedcorrelations tioncouldresultfromapositivecorrelationofdisklifetimeand likelihood of forming giant planets (Bouvier 2008). The influ- From an observational point of view, a number of correlations enceofthedisklifetimeτ ontheoccurrenceofgiantplanets, disk were inferred in the past years, with various degrees of signif- asobservedinourmodels,isdiscussedinSect.8.3. icance (see Udry & Santos 2007 or Mayor et al. (2011) for an In general however, only probability distributions for disk overview).Weaddressinthispaperthefollowingcorrelations: masses and lifetimes can be inferred from the observations of (1) The clearest correlation is the link between the stellar starformingregions.Ourstudyallowstoacertaindegreetonar- metallicity (to first order an indicator of the initial dust-to-gas- rowdowntheseprobabilitydistributionsforaspecificstar-planet ratio in the disk, see Santos et al. 2003) and the likelihood of system,asnotalltypesofplanetscanforminalltypesofdisks. detectingagiantplanet.This“metallicityeffect”isobservation- ally known for a long time and very well established for solar- like stars by numerous studies (e.g. Gonzalez 1997; Fischer & Valenti2005,Udry&Santos2007).The“metallicityeffect”was 1.3. Earlierworks studiedasanobservationalconstraintinPaperIIandisherefur- Severalstudieshavealreadyaddressedcorrelationsbetweendisk theraddressedinSect.4.2,4.3and8.1. and planetary properties: Ida & Lin (2004b) studied the influ- (2) Interestingly, lower mass Neptunian and Super-Earth encesofthemetallicityasmodeledbythedust-to-gasratio.They planets seem in contrast not to be found preferentially around find that the normalized mass and semimajor axis distributions high [Fe/H] stars (Mayor & Udry 2008; Sousa et al. 2008; ofthepotentiallyobservable(giant)syntheticplanetsarerather Ghezzietal.2010).WestudythisinSect.4.1and7. independent of [Fe/H], while the frequency of giant planets in- (3) Already much less compelling is a possible absence of creases with it, in agreement with observations. Kornet et al. very massive planets (in a mass range of about 5 to 20 Jupiter (2005)modeledtheevolutionofthesolidsfromdustsizetoplan- masses, M ) orbiting within a few AUs low metallicity solar- etesimalsand,usingthefinalplanetesimalsurfacedensity,esti- likestarsasfoundinradialvelocity(RV)searches.Suchapossi- (cid:88) mated the giant planet formation capability. Such an approach bleabsencewasnotedbyUdryetal.(2002),Santosetal.(2003) alsoreproducesthe“metallicityeffect”.Dodson-Robinsonetal. andFischer&Valenti(2005).ThisisdiscussedinSect.4.3.3. (2006) derived a fit for the time until gas runaway accretion is (4) In contrast, no secure correlations were found between triggeredbyaplanetarycoreatafixeddistanceof5.2AU,asa stellarmetallicityandplanetarysemimajoraxis,atleastamong function of the planetesimal surface density. They used this fit RV detections (Udry & Santos 2007; Valenti & Fischer 2008; to estimate the frequency of giant planets as a function of sev- Ammler-vonEiffetal.2009),eventhoughpossiblecorrelations eral disk properties. We show our results regarding this subject were discussed in the literature. Sozzetti (2004) for example in Sect. 8, comparing with their results. Matsuo et al. (2007) studiedwhetherstarswithPegasiplanetshaveparticularlyhigh usedparameterizedformationmodelstodeterminethediskpa- [Fe/H], even among planet hosts. Correlations between migra- rameterswheregiantplanetformationispossibleeitherbycore tion, semimajor axis and metallicity are addressed in Sect. 6.1 accretion or direct gravitational collapse. We present a similar and6.3. studyinSect.5. (5)GiantplanetsaremorefrequentlyfoundorbitingAstars than orbiting solar-like stars, and are also more massive (Lovis & Mayor 2007; Bowler et al. 2010) on average. In Paper III it 1.4. Structureofthepaper was shown that this correlation can be best reproduced if one assumes a roughly linear scaling between disk (gas) mass and Thepaperisorganizedasfollows:Section2describesthemeth- stellarmass.Thisimplies,atleastpartially,acorrelationbetween ods used to obtain the result concerning the synthetic popula- diskmassandplanetmass(Sect.4.1.2and4.4). tionpresentedinSect.3.Section4studiesvariousaspectsofthe Inaddition,weaddressbelowseveralothercorrelationsthat impact of disk properties on the mass of extrasolar planets, in- appear in the models but that have not yet been reported in the cluding the planetary initial mass function. Section 5 analyzes observation, but which could become observable in the future under which disk conditions giant planets can form. Section 6 withbetterinstrumentsandlarger,morecompletedatabases. addresses the correlationsof disk properties with planetary mi- gration and the final semimajor axis distribution. In Sect. 7 we study the metallicity of close-in planets. Section 8 shows how 1.2. Specialroleof[Fe/H] diskpropertiesdeterminethefractionofstarswithgiantplanets, Populationsynthesisisapowerfultooltoinvestigatesuchcorre- whileSect.9assestheconsequencesifdiskmassesandlifetimes lations, because all properties of the (numerical) parent disks, arecorrelated.Finally,inSect.10wesummarizetheresultsand and those of the synthetic planet populations emerging from presentourconclusions. C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV 3 2. Methods tential correlations arising during disk formation itself e.g. via opacityeffects.ThefactthatwedrawΣ and M˙ independently As our approach to planet formation and population synthesis 0 w implies,onaverage,longerdisklifetimeτ formoremassive wasdescribedindetailsinPaperI,welimithereourselvestoa disk disks. The importance of this coupling is discussed in sect. 9. shortoverview. We do not consider here the influence of varying initial disk radii(Kornetetal.2005).Thiswillbeconsideredinfuturework, 2.1. GeneralProcedure takingintoaccountrecentobservationalresults(Andrewsetal. 2009,2010). To obtain a synthetic population of planets, we proceed in five steps:(1)Theprobabilitydistributionsfortheinitialconditions are derived from observations of protoplanetary disks. (2) A 2.3. PlanetFormationModel largenumbersofsetsofinitialconditionsaredrawnfromthese Weuseaslightlymodifiedversion(seeMordasinietal.2009a) probabilitydistributionsinaMonteCarlofashion.(3)Thecorre- oftheextendedcoreaccretionformationmodeldescribedinde- spondingfinaloutcomesoftheplanetformationprocess(plane- tailsinAlibertetal.(2005a).AsinPollacketal.(1996),wecom- tarymassandposition)arecomputedusingourplanetformation putetheevolutionoftheplanetarycoreandenvelopestructure, model. This results in a population of synthetic planets, most butincludediskevolutionusingtheαformalism,andplanetary of them however undetectable with current observational tech- migration(isothermaltypeIandtypeII).Wehavebeenableto niques. (4) To obtain the subset of the potentially observable show(Alibertetal.2005b)thatthemodelreproducesmanyob- syntheticplanets,weapplyanappropriatedetectionbias.Since servationalconstraintsimposedbyourowngiantplanets. mostplanetshavebeendiscoveredbyradialvelocitytechniques, weuseabiasbasedonthevelocityamplitudeofthehoststar.(5) Correlations between the initial (disk) conditions and the plan- 2.3.1. Relevantmodelassumptions etary properties are searched and compared to observations (if existing). Forthedisk-planetcorrelationsdiscussedinthiswork,anumber ofmodelassumptionswerefoundtobeparticularlyrelevantand havedirectlyvisibleconsequencesinthecorrelations. 2.2. InitialConditions-Probabilitydistributions The first one concerns the structure of the protoplanetary We use four Monte Carlo variables to specify the initial con- disk, and specifically the location of the iceline. In the nomi- ditions. (1) The dust-to-gas-ratio f which we link to [Fe/H] nal model, the position of the ice line is an increasing function D/G as [Fe/H]=log(f /f ), where f is the dust-to-gas ra- ofdiskmassduetoviscousdissipation(PaperIandIII;Minet D/G D/G,(cid:12) D/G,(cid:12) tio of the solar nebula for which we assume a value of 0.04. al.2011).Itisfoundthatthishasimportantconsequencesboth The choice of this value was discussed in details in Paper I. on the mass of giant planets (Sect. 4.3.1) and their formation Here we only briefly mention that the factor two to three by location(Sect.5.2.1). which 0.04 is larger than the measured photospheric Z of the A second assumption which is directly relevant for the sun (Lodders 2003) is a first order representation of the effects massesofgiantplanets(Sect.4.3.2,4.4)isthatweassumethat ofdustevolutionanddrift.Thesemechanismsleadtoanincrease gap formation does no reduce the gas accretion rate of giant ofthe“planetesimal” f comparedtotheoriginal“dust” f planets (Lubow et al. 1999). As explained in Paper I, the rea- D/G D/G intheinner,planetformingpartsofthediskbyasimilarfactor sonforthisisthatKley&Dirksen(2006)haveshownthatwhen through the advection of material from the outer disk (Kornet agiantplanetbecomessufficientlymassive,thedisk-planetsys- et al. 2004). The probability of occurrence of a given [Fe/H] temcanundergoaneccentricinstability.Theplanetthenleaves is derived from the metallicity distribution of the FGK stars in the clean parts of the gap, resulting in a substantial increase of the CORALIE planet search sample (Udry et al. 2000) which the gas accretion rate. This means that the gas accretion rate is is representative for the [Fe/H] distribution of the target stars obtained for low mass planets (Mcore (cid:46) 10M⊕) by solving the inthevariousmajorradialvelocitysearchcampaigns(PaperI). planetarystructureequationswhileformoremassiveplanetsin Thus, we assume that the observed stellar metallicity is a good the disk limited, runaway gas accretion phase it is equal to the indicator of the primordial disk metallicity (Santos et al. 2003, rate at which gas viscously flows towards the star in the disk but see also Pasquini 2007). (2) The initial gas surface density M˙enve = M˙disk = 3πνΣ where ν is the disk viscosity, and Σ the Σ at a = 5.2 AU which we link to the initial disk gas mass gassurfacedensity. 0 0 √ √ M = 4πΣ a3/2( a − a ).Theprobabilitydistribution Athirdsettingwhichisrelevantforthecorrelationofmigra- disk 0 0 max min of circumstellar disk masses is derived from the observations tion and metallicity (6.1.2) is that we assume that the planetes- of the ρ Ophiuchi star formating region (Beckwith & Sargent imal accretion rates are independent of the migration rate. The 1996).(3)Therateofphoto-evaporationM˙w.Thedistributionof M˙core is calculated in the same way as in Pollack et al. (1996). M˙ isconstrainedbytheobserveddiskagedistribution:Wehave Possibleshepherdingeffectsarenotincludedforthereasonsex- w adjustedourM˙ distributiontoobtaintogetherwithourvalueof plained in Paper I. This means that cores in low solid surface w the viscosity parameter α a distribution of disk lifetimes τ densitydisksstillcangrowrelativelymassivebymigration. disk thatisingoodagreementwiththeobserveddistribution(Haisch Fourth, we assume for type II migration that as soon as the etal.2001).(4)Theinitialsemimajoraxisoftheplanetaryseed planet is more massive than the local disk mass (M > Σa2 planet a .Thedistributionofthestartingpositionsofplanetaryem- where a is the semimajor axis), the migration rate is given as start bryosisnotconstrainedbyobservationsandonlytheoreticalar- −(3ν/a)×(Σa2/M )p. In the nominal model, we use p = 1 planet gumentscanbeused.Theyindicate(e.g.Mordasinietal.2009a) (“fully suppressed” case, Armitage 2007). This is important to that runaway bodies should emerge with a uniform distribution understandtheabsenceofastrongcorrelationof[Fe/H]andthe inlog(a ),asalreadyadoptedbyIda&Lin(2004a). semimajoraxis(Sect.6.1.1,6.3.3) start Inaddition,weassumethattheseMonteCarlovariablesare Notethatwesimplifytheproblemsignificantlybyassuming independent variables. Hence, we do not take into account po- thatonlyasingleplanetcanforminagivendisk.Thiswouldbe 4 C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV correct in the limit that proto-planets can form within a single system without influencing each other. This is clearly an ideal- ization, but in the context of this study this is not necessarily a disadvantage. It allowsone to see clearly disk-planet correla- tions,whichotherwisemightgetpartiallyblurredduetotheran- domcharacterofthe(gravitational)interactionsbetweenseveral protoplanets.Thommesetal.(2008)discussextensivelysomeof theeffectsinducedbytheconcurrentformationofseveralplan- etsinonedisk. Thepopulationdiscussedhereisessentiallyobtained,except if otherwise mentioned, with identical parameters and Monte CarlodistributionsasthenominalpopulationpresentedinPaper IandII.Inparticular,thismeansthatthestellarmassisequalto onesolarmass M ,thegasdiskviscosityparameterαis0.007, (cid:12) andthetypeImigrationefficiencyfactor f is0.001. I However,thereisonerelevantaspectinwhichtheprocedure used here differs from the one used in paper I and II. We draw the initial semimajor axis of the starting seed strictly uniform in log(a ) and disregard the additional criteria mentioned in start PaperI.Thiseliminatessomecorrelationspresentalreadyinthe initial conditions and thus makes it easier to identify the influ- enceofdiskpropertiesontheplanetaryproperties. Asapracticalunitforthediskgasmasseswedefine,inanal- ogyto[Fe/H],arelativelogarithmicunit,denoted[M /M ] = log(Σ /Σ ), where we assume an initial gas surfacDe deSnNsity Fig.1. Mass-distance diagram of the synthetic population ana- at 5.20AU0,SΣN = 200 g/cm2. This is somewhat more than a lyzed in this work. The dashed curve shows the feeding limit 0,SN (cf.PaperI):planetsreachingthislimithavebeenarbitrarilyset plain MMSN values - Hayashi’s (1981) value would be about 145 g/cm2 - but with this choice [M /M ] covers, similar as to0.1AU. D SN [Fe/H],anearlysymmetricrangearoundzerobetween-0.6and +0.7forthevaluesofΣ consideredhere(50-1000g/cm2,cor- 0 responding to disk masses between about 0.004 and 0.09 M ). (cid:12) As the disk gas mass M and the initial gas surface density Σ M⊕ only. Further growth from these masses to the final masses D 0 by giant impacts (e.g. Marcus et al. 2010) will then only oc- aredirectlyproportionaltoeachother,thesetwotermsareoften cur after the damping influence of the gas disk is gone (Ida & usedinaninterchangeableway. Lin2010),aphasewhichishowevernotincludedinourmodel at the moment. This effect would populate the depleted region from“below”. 3. Mass-DistanceDiagram Second,thehighobservedfrequencycouldbeindicativeof To provide an overview of the synthetic planet population we quiteefficienttypeImigration,atleastinsomepartsofthedisk. analyzeinthisworkweplotinFig.1themass-distancediagram This would populate the depleted region from “outside”. Large oftheentirepopulation.Inthefigure,anumberofstructuresal- quantitiesoflowmass,close-inplanetsarefoundinsomesyn- ready discussed in Paper I can be identified again, like the low theticpopulationstoo,butonlyforhightypeIefficiencyfactors mass “failed cores”, the approximately Neptunian mass “hori- (f ∼ 0.1 − 1) as shown in Fig. 11 of paper II. On the other I zontal branch” or the massive “outer group” planets outside a hand can the distribution of semimajor axes of giant planets at fewAU.Wealsoseethattherearenoverymassiveplanetsclose- larger distances only be reproduced when the type I migration in. This is a consequence of the fact that planets with a mass rate (as found from linear theories for isothermal disks) is re- largerthanthelocaldiskmassmigrateatareducedratebecause ducedbyasignificantfactor(f (cid:46)0.01,PaperII;Schlaufmanet I theirinertiaistoolargeforthefiniteangularmomentumfluxin al.2009).HereweareusingthesamelowtypeIefficiencyfactor thedisk.Theirabsenceatsmalladependsupontheassumedde- f =0.001whichleadinPaperIItothebestreproductionofthe I greeofreductionofthetypeIImigrationrateoncetheplanetis observed properties of giant planets. Unfortunately, this causes massive(Syer&Clarke1995). inthesametimenecessarilyalsothedepletedregion. Theanswertothisdilemmaisprobablyasignificantlymore complex migration pattern than what can be mimicked with 3.1. Paucityoflowmass,close-inplanets global(independentofplanetmassanddistance)efficiencyfac- There is also a relative paucity of low mass (4 (cid:46) M (cid:46) 10M⊕) torslike fI forisothermaltypeIratesasdonehere.Recentstud- close-in(a(cid:46)0.3AU)planets.Recentobservations(e.g.Howard iesoftypeImigrationdroppingtheofteninappropriateassump- etal.2010;Mayoretal.2011)ratherindicateahighfrequency tion of isothermality (e.g. Paardekooper et al. 2010; Masset & oflowmassSuperEarthplanetsclosetotheirparentstar. Casoli 2010) indeed find complex patterns with rapid in- and Concerningthisissue,wefirstremindthatourmodelshows outward migration. First exploratory planet population synthe- the state of (proto-)planets at the moment when the disk disap- ses using such updated type I migration models (Mordasini et pears, and not the final one after billions of years of evolution. al.2011b)findthattypeImigrationisdirectedoutwardinsome Atthisstage,thelowmassprotoplanetsatsmalldistanceshave partsofthedisk,andinwardinothers,whichleadstotheexis- amassaboutequalthelocal(solid)isolationmass(typeImigra- tenceofconvergencezones(migrationtraps)whichcanresultin tionisstronglyreducedinthissimulation),whichisoforder0.1 a pile-up of many low mass planets in certain parts of the disk C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV 5 Fig.2.Planetaryinitialmassfunctionatthemomentwhenthegaseousdiskdisappears.Notethatthemodelsdoesn’tincludeany growthprocesseslikegiantimpactsafterdiskdispersalwhichcansignificantlymodifythemassdistributionafterwards,especially forlowmassplanetswith M (cid:46) 10M .Left:PIMFbinnedaccordingtometallicity.Center:PIMFbinnedaccordingtothegasdisk ⊕ mass.Left:PIMFbinnedaccordingtothedisklifetime.Themeaningofthedifferentlines,andthelimitingvalues,areindicatedin theplot. (cf.Lyraetal.2010;Sandoretal.2011).Thisimportantsubject 4.1.1. PIMFasfunctionofmetallicity willbeaddressedindedicatedwork(Dittkristetal.inprep.). A third mechanism occurs when several planets form and The left panel shows that the metallicity just modifies the im- migrate concurrently. Then, lower mass planets can be pushed portance of the three basic families of planets visible in the closetothestarafterbeingcapturedinmeanmotionresonances PIMF(Jovianplanets,Neptunianplanets,proto-terrestrialplan- of a more massive, more rapidly migrating outer planet. This ets), without however changing the general shape of the PIMF willbestudiedinoncomingsimulationsallowingtheformation (in contrast to the other two disk properties). For giant planets ofmanyplanetsperdisk(Alibertetal.inprep.) this means that in a high metallicity environment, more giant planets form, but their mass distribution is similar. The reason forthisisrelatedtothefactthatacertain[Fe/H]actsasathresh- 4. Mass oldforgiantplanetformation(thresholdsolidsurfacedensityto 4.1. Planetaryinitialmassfunction(PIMF) reachacriticalcoremass,seeSect.5),butisnotimportantinde- terminingthefinaltotalmass,becauseintheend,gasmakesup Acentraloutcomeofpopulationsynthesisistheplanetaryinitial formostofthemassofagiantplanet,andnotsolids.Therefore, massfunctionPIMF(PaperIIandIII).Figure2showsthePIMF a high metallicity mainly allows a larger number of high mass of all synthetic planets with a mass larger than 1 M , binned ⊕ planets,butnotofahighermass(exceptforverylargemasses, intolow,mediumandhighmetallicity(leftpanel),diskgasmass seesect.4.3).Thisincreaseinfrequencyisofcoursetheunder- (central panel) and disk lifetime (right panel). The diving val- lyingreasonfortheobservedmetallicityeffect,i.e.theincrease ues are indicated in the plot and are chosen for all three cases ofthedetectionrateofgiantplanetswith[Fe/H](seeSec.8.1). in a way that the central bin contains about 70% of the plan- ets,andtheothertwoabout15%each.Werecallthatourmod- Moving down in mass, we see that in the Neptunian mass els start with an initial seed mass of 0.6 M , so that we cannot domain, the dependence of planet frequency on metallicity is ⊕ reliably make predictions about the exact form of the PIMF in weak. This means that no metallicity effect is predicted in this the (cid:46) 10M domain as mentioned in Paper I and II. Note that domain. This is in good agreement with recent observations ⊕ each bin has been normalized individually, so that the absolute (Mayoretal.2011).Movingfurtherdowninmasstotheproto- heightofdifferentbins(e.g.lowvs.mediummetallicity)cannot terrestrial domain, the lines are inverted relative to the giant becomparedinabsoluteterms,buttakingintoaccountthatthe planet domain, which means that an inverse metallicity effect initialdistributionof[Fe/H]followsthedistributionobservedin occurs(lowmassplanetsaremorefrequentatlow[Fe/H]com- the solar neighborhood. However, one can directly see the rel- paredtohigh[Fe/H]).Concerningthislastpoint,wemusttake ative importance of a given planetary type within a bin (e.g. at into account that in our model, only one embryo can form per highmetallicitytherearemoreJovianthanNeptunianplanets,in disk.Thesituationthatinhigh[Fe/H]disks,bothagiantplanet contrast to the low [Fe/H] case.). One recognizes again a num- anda“byproduct”terrestrialplanetformisthereforenotpossi- ber of features discussed in Paper I, like the high mass tail in ble.Thiscouldartificiallystrengthenthe[Fe/H]-lowmassplanet theSuper-Jupiterdomain,thegiant’splateauintheJovianmass anti-correlation,sothatmoresecurepredictionswillbepossible regime, the minimum at about 30-40 M corresponding to the with models dropping the one-embryo-per-disk simplification ⊕ planetarydesert(Ida&Lin2004a),theNeptunianbumpandthe (Alibert et al. in prep.). Observationally, future high precision strongrisetowardssmallmasses. observations(e.g.withESPRESSO)willtestthisprediction. 6 C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV Fig.3.Planetarymassfunctionbinnedac- cording to [Fe/H] as in Fig. 2, left panel, butnowonlyforthesub-populationofpo- tentially detectable synthetic planets with an RV precision of 10 m/s (left panel) and 1 m/s (right panel). Green dashed linesagainrepresenthigh,bluesolidlines mediumandreddottedlineslowmetallic- ities,wherethethresholdmetallicitiesare thesameasinFig.2. Onefurthernotesthatforallthreebins,thehighestpeakin longer lifetimes, cores will be able to grow to the critical mass themassfunctionisfoundfortheproto-terrestrialplanets,with andaccretegasinarunawayfashionevenforrelativelylowsolid the clearest dominance at low metallicity. For high metallicity, surface density. Therefore, the disk lifetime acts as a threshold the second highest peak occurs for the giant planets, whereas for giant planet formation, similar to the metallicity. The life- for intermediate and even more clearly for low metallicity, the time of disks however also affects the total mass of the planets Neptunianplanetsareresponsibleforthesecondhighestpeak. similartothediskmass,as,tofirstorder,theplanet’sfinalmass will be equal to the accretion rate times the duration of the ac- cretion phase. This twofold effect is clearly seen in the figure. 4.1.2. PIMFasfunctionofdiskgasmass Giantplanetsformedinlong-liveddisksarebothnumerousand ofahighermass.Comparedtometallicityanddiskmasses,the Thecentralpanelshowsthatthediskmassdistributiondirectly affects the shape of the PIMF and not only the height of the disk lifetimes has thus a more complex influence on the result- ingplanetpopulation.Thedisklifetimebothscalesanddistorts peaks of the distribution as metallicity does. For giant plan- the shape of the PIMF. Similarly to the primordial disk mass, ets,ahigh[M /M ]shiftstheformationofgiantplanetsfrom D SN it is difficult to deduce the disk lifetime for any give observed lowermasstohighermasses(comparetheblueandgreenline). For lower disk masses, massive giant planets ((cid:38) 6M ) can- system.Wecomebacktothatinsect.8. notformatall.Inthiscase,thereissimplynotenoughgaseous (cid:88) materialavailable.Thedistributionofintermediate-massplanets (30M (cid:46) M (cid:46) 1M ), on the other hand, is little affected. We 4.2. Observabledistributionasfunctionof[Fe/H] ⊕ concludethatinourmodelthereisadirectcorrelationbetween (cid:88) It is interesting to look whether some of the correlations be- the (maximal) mass of giant planets and the disk gas masses. tween disk properties and the underlying mass function (of all Thisisessentiallyduetothefactthatgiantplanetsaccretemost planets) discussed in the section above can already be seen in of their mass in a regime where their accretion rate is propor- the observational data we have today, which represent only a tionaltothediskmass,seethediscussioninSection4.4. smallfractionofallexistingplanets.Forthis,weplotinFig.3 In Paper III, we showed that, in our alpha-disk model, the themasshistogrambinnedaccordingto[Fe/H],butnowinclud- disk mass has to be scaled roughly linearly with M in order ∗ ingonlyplanetsdetectablebya10-yeardurationradialvelocity to reproduce the observed correlation between the star’s accre- (RV) survey with a precision of either 10 m/s (left panel) or 1 tion rate and its mass. As shown in this paper, this translated m/sprecision(rightpanel).Notethat,likeasinPaperIII,weuse into the formation of planets of a higher mass orbiting stars of averysimplevelocityamplitudecut-offcriteriumtodetermine a larger mass, in good agreement with observation (Lovis & detectability. Mayor2007).Eventhoughwedidnotvarythestellarmassinthe present study, the fact that more massive planets form in disks takenfromthehigh-massendofthedistributionstemsfromthe 4.2.1. 10m/sradialvelocityprecision samereason.Unfortunately,inpracticewecannotinferthepri- mordial disk mass for any actually detected exoplanet and so The figure shows that at a precision of 10 m/s, the mass distri- thiscorrelationisdifficulttotestobservationalincontrasttothe butionhasaverysimilarshapeforallthreemetallicitybins,as metallicitycorrelation.Thepanelfurthershowsthattherelative expected from the discussion above and the previous work by importanceofNeptune-likeplanetsstronglydependsonthedisk Ida&Lin(2004b).However,acloserlookindicatesthatthelow gasmass.Neptune-likeplanetsformparticularlyeasilyindisks [Fe/H]binhasasomewhatnarrowerdistribution,andthatthere withsmallprimordialdiskmasseswhiletheydonotseemtobe isacertainsystematicdifferenceattheuppermassend(seethe abletoforminmassivedisks.Thisagainisaconsequenceofthe nextsection).However,themediumandhighmetallicitybinsto sameeffectdiscussedaboveandalreadypointedoutinPaperIII. whichmostofthepresentdayknownplanetpopulationbelongs, areverysimilar.Itisthereforenotsurprisingthat,giventherel- ativelysmallnumberofobservedplanetscomparedtothelarge 4.1.3. PIMFasfunctionofdisklifetime numberofsyntheticplanetsusedhere,nosignificantcorrelation ThethirdpanelattherightofFig.2finallyshowsthedisklife- betweentheshapeofthemassdistributionofgiantplanetsand time. The disk lifetime has a twofold influence. In disks with thehoststarmetallicityhasbeennoticedthusfar. C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV 7 4.2.2. 1m/sradialvelocityprecision Ataprecisionof1m/sonthecontrary,theinfluenceof[Fe/H] on the PIMF becomes visible as measured by the relative fre- quencyofNeptunianversusJovianplanets,atrendthathasbeen observed(Udryetal.2006).Wefindthattheratioofthenumber ofNeptune-masstoJupiter-massplanetsstronglycorrelateswith themetallicity.Forthelowmetallicity,theNeptune-massplan- etsarealmostasnumerousasJupiter-massplanetswhileforhigh metallicitydiskstheyaresignificantlylessfrequent.Thiscorre- sponds well with the observed result (Sousa et al. 2008). The numerical values of the ratios of synthetic Jupiter-to-Neptune numberofplanets(dividingmassdefinedat30M )obtainedfor ⊕ a 10 year search at 1 m/s is in our nominal simulation 4.7, 5.8 and 7.6 for the low, medium and high [Fe/H] bin respectively. ThesevaluesaredifficultdirectlywithSousaetal.(2008)dueto severalreasons(differing[Fe/H]bins,differingprimarymasses, differingsurveyduration).However,theseauthorsfindvaluesfor theJupiter-to-Neptuneratioof1.6-5for[Fe/H]<−0.15,and7.5- 30for[Fe/H]>0.15.Notethattheratiosevenat1m/sprecision arehigherthanthoseinthefullunderlyingdistributionwithout detectionbias.Thereasonisthatevenat1m/smanyNeptune- likeplanetsremainundetected.Finally,wepointoutthatdueto our assumption of a single planet-per-disk simplification, it is verylikelythatourratioisanupperlimit.Ahintinthesamedi- Fig.4. The mass of synthetic planets (in Jupiter masses M ) as J rectioncanalsobeinferredwhencomparingoursyntheticresult a function of [Fe/H] for the nominal population (small dots). withtheobservedoneshowninFig.10ofMayoretal.(2011). There is an absence of very massive planets around low metal- Otherwise,observedandsyntheticresultaresimilar. licitystars.Intheplot,linesindicateapproximativelytheregion From the two plots we also understand that the decrease of ofhighmassandlowmetallicitywherenosyntheticgiantplan- theobservedmassdistributionat10m/stowardssmallmasses, etsarefoundinothernon-nominalpopulations:withaniceline whilebeingclearlytheconsequenceoftheobservationaldetec- fixedto2.7AU(dottedline),orwiththeeffectofgapformation tionbiasagainstlowmassplanets,isamplifiedbytheactualtrue onthegasaccretionratemodeledasinVeras&Armitage(2004) decreaseoftheunderlyingmassdistribution.Thisdegreeofde- (solidline).ThedashedlineshowsthelimitgiveninFischer& pletionofintermediatemassplanets(30−100M⊕)ispotentially Valenti(2005). an important constraint on the gas accretion rate at the begin- ning of runaway gas accretion. In Mordasini et al. (2011a), we presentedadetaileddiscussionaboutvariousassumptionsabout 4.3.1. Effectcausedbytheicelineposition gasaccretionthatdirectlydeterminethisportionofthetheoreti- A less pronounced absence of massive planets at low metallic- calPIMF. ity can also be obtained if we fix the position of the ice line at 2.7AUindependentlyof[M /M ],asexpectedforanoptically D SN 4.3. (Maximal)planetarymassesandmetallicity thindiskirradiatedfroma1M star(Ida&Lin2004a).Thisdif- (cid:12) ferenceisduetoaninterestingchainofcorrelations: Toemphasizetheimportanceofthemetallicity(measuredinour 1)Werecallthatinthenominalcase,thepositionoftheiceline model by the ratio of dust-to-gas in the disk) on the resulting isanincreasingfunctionofdiskmass. high-endofplanetarymasses,weshowinFigure4themassof 2)Atlowmetallicities,highdiskgasmassesareneededforgiant synthetic planets (in units of Jupiter masses) as a function of planet formation (Matsuo et al. 2007; Sect. 5.3). Thus, the ice metallicity. While we have indicated in Sect. 4.1 that metallic- linepositioninlowmetallicitydisksforminggiantplanetswill itydoesnotchangesignificantlythedistributionofthemassfor typicallybelocatedatlargedistances. thebulkofthepopulation,weseeherethatthemetallicitydeter- 3)Hence,thetypicalstartingpositionofgiantplanets-to-bewill minesthemaximummassaplanet1 cangrowtoinagivendisk, alsobelargeatlow[Fe/H](a (cid:38)a ,seeSect.5.2). in particular for subsolar metallicities. This can be understood start ice 4)This distant starting position implies a slower growth of the asfollows.Togrowtoaveryhighplanetarymass,acriticalcore core (e.g. Paper I), disadvantageous for successful giant planet mustformbeforeviscousevolutionandphotoevaporationofthe formation. diskhavehadtimetosignificantlydepletethediskmass.Sucha Thus,apositivecorrelationofthediskgasmassandtheiceline veryearlystartisnotpossibleinalowmetallicityenvironment positionrendersgiantplanetformationatlow[Fe/H]evenmore inwhichgrowinglargecorestakeslonger. difficult,andmakesitimpossibletopopulatetheupperleftcor- This second order correlation via the core formation ner of Fig. 4 with the nominal model. It also means that there timescalehasbeencheckedandconfirmedbyartificiallyreduc- is a difference between a metal poor-gas rich and a metal rich- ingthestartingtimeoftheseedembryosbyafactoroftwo.In gas poor environment, even if the solid surface density beyond this test, the maximal mass of planets at [Fe/H]=-0.4 increases theicelineareinprinciplethesame.Alow[Fe/H]cannotinall toabout18M comparedto∼7M inthenominalcase. circumstancesbecompensatedahigh[M /M ]. D SN 1 Weherecal(cid:88)lforsimplicityallobje(cid:88)ctsthatforminthepopulation In the case with an ice line independent of [MD/MSN], this synthesisplanets,eveniftheirmassislargerthanthedeuteriumburning complexchainofcorrelationsbetween[Fe/H],[MD/MSN],aice, limitatabout13M . astart andtstart isbroken,andplanetswithabout10 M canstill (cid:88) (cid:88) 8 C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV format[Fe/H]=-0.5,andthatmaximalmassesbecomeindepen- dentof[Fe/H]alreadyfor[Fe/H](cid:38)−0.3.Theupperlimitofpos- siblemassesisindicatedforthiscasebythedottedlineinFigure 4. It is clear that this chain of correlations depends quite sensi- tively on a number of specific model assumptions, making it a lessrobustprediction. 4.3.2. Verymassiveplanets Theabsenceofverymassivecompanionsformedbycoreaccre- tion at very low metallicities is also seen in the models of Ida &Lin(2004b)andMatsuoetal.(2007).However,comparedto theseworks,planetsofsignificantlyhighermasscanforminour simulations.Thisisadirectconsequenceofthefactthatwedo not limit gas accretion due to gap formation, as mentioned in Sect.2.3.1.Ifweinsteadlimitinanon-nominalpopulationthe gas accretion rate by giant planets due to gap formation by us- ingthefitofVeras&Armitage(2004),themaximalmassofall planetsinthepopulationarereducedto∼7M .Theabsenceof themostmassiveobjectsatlow[Fe/H]wouldremainsimilarbut with weaker dependence of the maximal mas(cid:88)s on [Fe/H] when scaled to smaller absolute masses. The approximative limiting envelopeforsuchapopulationisalsoshowninFig.4asasolid line. This envelope is now similar to the one of Matsuo et al. Fig.5. Observed and synthetic planetary masses as a function (2007).Wethusseethatwhilethetendencytowardsanabsence of metallicity. Black dots are synthetic planets in the nominal ofverymassiveplanetsatlow[Fe/H]isageneralpredictionof model, while solid, dashed and dotted lines again indicate the thecoreaccretiontheory,thequantitativeresultsdependonspe- limitingenvelopeofnon-nominalmodels,asinFig.4.Redstars cificmodelsettings. areobservedRVcompanionsaroundstarswith0.8< M /M < ∗ (cid:12) Whenstudyingthefigureoneshouldkeepinmindthatvery 1.2.GreensquaresareobservedRVplanetsaroundstarswitha massive planets ((cid:38) 10M ) are in fact very rare outcomes in masslargerthan1.2M .Forbothcases,apointinthemiddleof (cid:12) our simulations, in agreement with the observed “brown dwarf thesymbolindicatesthatthestarisasubgiantorgiant.Bluetri- (cid:88) desert”asdiscussedinPaperII.Figure4mayprovidethesome- anglesshowplanetsaroundHR8799.Namesofrelevantobjects whatmisleadingimpressionthatthesearecommonobjectswhile aregivenintheplot. in fact these objects appear only because the underlying syn- thetic population is extremely large (about 200 000 initial con- ditions). Even if such a population vastly exceeds the actually tribute,stillsuffersfromlownumberstatistics(seealsoSozzetti observedpopulation,thesehighnumbersofplanetsarerequired &Desidera2010). to investigate the different correlations some depending on two In Fig. 5, red stars symbolize planets detected by the ra- variables(likediskmassand[Fe/H]). dialvelocitymethodaroundprimarieswithamassbetween0.8 and 1.2 M , which are the most relevant cases (in the model (cid:12) M = 1M ).Greensquaresarecompanionstostarsmoremas- 4.3.3. Comparisonwithobservation ∗ (cid:12) sivethanthis,alsodetectedbyradialvelocity.Starswhichhave The absence of very high mass planets (or of a very high to- evolvedoffthemainsequence(ifknown)aremarkedinthefig- tal mass locked up in planets in multi-planet systems) at low urewithadot.Notethatstellarevolutioncouldbeofrelevance metallicitieswasnotedalsoobservationallyinUdryetal.(2002), here, as it might invalidate our underlying assumption that the Santos et al. (2003) and Fischer & Valenti (2005). The dashed photospheric composition measured today correlates with the line in Fig. 4 is taken from the latter work. It is clear that such bulkcompositionofthediskmaterialatformation.Thisassump- an absence could also simply be a small number effect: Giant tioncouldbeinvalidincaseofenhancedheavyelementsettling, planets around low [Fe/H] hosts are rare, and high mass giant which for the sun seems to have already lead to a reduction of planets((cid:38) 5M )arerareatallmetallicities,whichcouldcom- thepresentdayZascomparedtotheprimordialZ0bymorethan bine into the observed paucity. It is interesting to compare our 10%(Lodders2003). (cid:88) results with the observational database especially since the lat- Looking at Fig. 5, we see that the bulk of all observation- ter has been significantly extended since the above mentioned ally detected planets falls into regions of the plot where syn- papershavebeenpublished. thetic planets are also found. Hence, from this observation we Such a comparison is presented in Fig. 5 which shows the could conclude that core accretion can account for almost all massesofactualandsyntheticplanetsasafunctionofmetallic- planets currently known. This conclusion was already reached ity.ObservationaldatawastakenfromJ.SchneidersExtrasolar byMatsuoetal.(2007).Itshouldbenotedthatsurveysthatlook Planet Encyclopedia. Note that this database does not include atverymetalpoorstarsdoexist(Santosetal.2009). companions larger than about 20 M . However, such compan- ions are anyway extremely rare, both in the model (only 0.1% (cid:88) 4.3.4. Relevantindividualobjects of the synthetic planets have a mass larger than 20 M ) but alsoobservationally.Asaresult,ourobservationalknowledgeof However,eightRVplanetsorbitingfivestarsclearlylieoutside (cid:88) thismassrange,whereseveralformationmechanismcouldcon- theregionwheregiantplanetsformorbitinga1 M starinour (cid:12) C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV 9 nominal model. This group is characterized by a much higher probably has a mass larger than 1.2 M (1.5±0.3M ), so the (cid:12) (cid:12) fractionofgiantstarsandofstarsmoremassivethan1.2M than synthetic population does not apply directly, but it is neverthe- ∗ thefullsample.Italsohasahighmultiplicityforgiantplanets. lessinterestingtonotethattheseplanetscometolieinaregion Assomeoftheobjectsarequitepeculiar,itisworthtodiscuss inthe[Fe/H]-massplanewhichisatorclosetothelimitwhere individuallysomeoftheseobjects. giantplanetscanoriginatefromcoreaccretion.Thisisafinding HD155358 has two planetary companions of rather low independentfromthefactthatthelargesemimajoraxesofthese mass ((cid:46) 1M ) at 0.6 and 1.2 AU (Cochran et al. 2007). The planetsmakescoreaccretionastheformationmechanismdiffi- starhasamassof0.87±0.07M anda[Fe/H]= −0.68±0.07. cult(Dodson-Robinsonetal.2009),ifnoadditionalmechanism (cid:12) (cid:88) HD155358 is a particular star (Fuhrmann & Bernkopf 2008). causingoutwarddisplacementlikescatteringisacting. It is a very old, thick disk subgiant with a chemical composi- tion very far from scaled solar composition. It is significantly enriched in alpha-chain nuclei. [Mg/H] is for example with - 4.3.5. Summaryconcerningthe[Fe/H]-M correlation 0.36 much higher than [Fe/H]. What matters for the ability to Insummaryweseethatthereareonlyextremelyfewexamples formasufficientlymassivecoreisthesurfacedensityofallcon- of bona fide, massive companions (5 (cid:46) M (cid:46) 20M ) orbiting densible elements beyond the ice line which can attribute im- solar-like, main sequence stars at small orbital distances which portant quantities of matter, in particular the α elements O, Si do not fall into the [Fe/H]-M parameter space cove(cid:88)red by our and Mg (Dodson-Robinson et al. 2006; Gonzalez 2009). Due implementationofthecoreaccretionmodel. to the enrichment in α elements, the disk around HD155358 The solid line in Fig. 5 shows also the limiting envelope of wasdepletedinplanetesimalsmaybeonlybyafactor2belowa syntheticplanetarymassesderivedinthecasewherethegasac- [Fe/H]=0solarcompositiondisk,equivalenttoa[Fe/H]∼ −0.3 cretionrateduetogapformationislimitedbyusingthesimple if it were to have a scaled solar composition. At such a value, one parameter fit of Veras & Armitage (2004) to hydrodynam- formingtheanywaylowmassgiantplanetsiscertainlypossible ical simulations of Lubow et al. (1999). It is clear that such a withthecoreaccretionmechanism. simple fit can only be a rough approximation of the real effect HD114762 with its Msini = 11M , a=0.36 AU compan- and that e.g. disk viscosity also has an influence on the degree ion (Latham et al. 1989) was discussed in this context already (cid:88) ofquenchingofgasaccretion(Lissaueretal.2009).Butthesig- byUdryetal.(2002).Ithasbeenoftenconsideredthatthesys- nificant number of planets lying above the solid line indicates temisseennearlypoleon,sothatthecompanionmightinfact that mechanisms like the eccentric instability (Kley & Dirksen bealateMdwarf(Cochranetal.1991;Hale1995).Analterna- 2006)thatallowgrowthbeyondthegapbarrierseemtoplayan tivehypothesisisbasedonthefindingthatHD114762ischem- importantroleinnature. ically and evolutionary very similar to HD155358 (Fuhrmann & Bernkopf 2008) which would mean that its effective surface Wefinallynotethatthemostmassivesyntheticplanetscould also, at least partially, be an artifact of the one embryo per densityofplanetesimalswasmuchhigherthanonewouldinfer fromthelow[Fe/H]andscaledsolarcomposition. disk approximation. In a disk where several protoplanets con- HD47536 is an old K1III, [Fe/H]=-0.68 giant which is or- currently grow, planets compete for gas and ejection can even removesomeplanets(Thommesetal.2008).Inthissense,look- bited by one (Setiawan et al. 2003), possibly two (Setiawan et ing at the total mass in the system should be a more adequate al.2008)quitemassivecompanions(5and7 M )insideafew quantitythanindividualplanetmasses. AU. While Setiawan et al. (2003) originally quoted a possible (cid:88) stellarmassofabout1to3 M ,daSilvaetal.(2006)morere- (cid:12) cently determined a mass of 0.94±0.08M . With this primary ∗ 4.4. Planetarymassesanddiskmass mass, the object falls clearly out of the envelope of synthetic planetsinourmodel.Inparticularifitstwo-planetconfiguration Figure6showsplanetarymassesasafunctionoftheinitialdisk is confirmed, this is an interesting object to study with forma- gasmass,againinunitsofJupitermassestofocusonthemassive tion models, as one then cannot invoke, due to orbital stability, planets.Herewefindadifferentbehaviorthanfor[Fe/H]:Asex- a nearly pole on orientation as a possible explanation. Also for pectedfromthediscussionoftheinfluenceof[M /M ]onthe D SN thedirectcollapsemodelthiscaseisprobablynotobvioustoex- PIMF, we see that there is a positive correlation of disk mass plain due to the small semimajor axis of the planets (about 1.5 andplanetarymassoverthefulldomainofdiskmassesconsid- AUfortheinnerplanet),seee.g.Boley(2009). eredhere.Thisisillustratedbythedottedlinecorrespondingtoa BD+202457isaK2IIgiant,withanestimatedmassof2.8± massthatisequalhalftheinitialdiskmass.Thisshowsthatthere 1.5M ,averylowmetallicityof[Fe/H]=−1.00±0.07withtwo is,forthedomainof[M /M ] (cid:38) −0.2anapproximativelylin- (cid:12) D SN verymassivecompanionsintightorbits(Niedzielskietal.2009). ear correlation between the maximal planet mass and the disk Eveniftheverydifferentprimarymassmakesacomparisonwith mass. The reason for this is that giant planets accrete in our theresultsheredifficult,thesecompanionsseemtobefarfrom model(seeMordasinietal.2009a)mostoftheirmassinthedisk thepossibleparameterspaceforcoreaccretion. limited accretion regime where the planetary gas accretion rate Theobjectaroundγ1 LeoA(HIP50583)wasfoundtohave isassumedtobethesameasinthedisk(M˙env = M˙disk =3πνΣ), a projected mass of about 8.8 M (Han et al. 2010). The orbit which is itself proportional to the initial disk mass for a given ofthecompanionwashoweverrecentlyastrometricallydetected distance and moment in time, at least if the disk evolution is by Reffert & Quirrenbach (2011(cid:88)) in the Hipparcos data. Their closetoself-similaritysolutions(Hartmannetal.1998). resultsindicateaclearlylargeractualmassofabout66M (with Thedifferentinfluenceof[Fe/H]anddiskgasmassesonthe substantial incertitude). This moves the companion out of the massofgiantplanetscanbequantified:Forplanetswithamass (cid:88) relevantmassdomain. larger than 100 M , one finds that the median planetary mass ⊕ The plot also includes HR 8799 with a measured [Fe/H]=- growsforanincreaseof[Fe/H]from-0.45to0.45onlybyabout 0.47 (which might however not reflect the initial metallicity, afactor1.5(from512to747M ).Foranincreaseoverthesame ⊕ Maroisetal.2008)andplanetsdetectedbydirectimagingi.e.at [M /M ]domainincontrast,themediangiantplanetmassin- D SN largeorbitaldistances(Maroisetal.2008,2010).HR8799also creasesfrom182to814M i.e.byaboutafactor4.5. ⊕ 10 C.Mordasinietal.:ExtrasolarplanetpopulationsynthesisIV value of M˙ = 0.04M˙ is hit. The consequence is as shown env disk by the figure that the clear correlation of disk mass and planet massisnowbroken,asexpectedfromthefunctionalformofthe equation: One finds that the maximal mass now depends only weakly on the disk mass, approximatively as (M /M )1/4 for D SN M (cid:38) M .Aweakcorrelationremainsbecauseofthefloorac- D SN cretion rate. At lower disk masses, the correlation is somewhat stronger,forthesamereasonasforthenominalpopulation. Thefactthatforthenominalpopulation,maximalmassesare linearly proportional to disk masses is interesting as one there- fore expects to see an imprint of the disk gas mass distribution ontheplanetarymassdistribution.Thisisreminiscentofthesit- uationforstars,wheretheprotostellarcoremassfunctionCMF hasaremarkablysimilarfunctionalformasthestellarIMF(e.g. McKee & Ostriker 2007). This question will be addressed fur- therinforthcomingwork. 5. Diskconditionsleadingtogiantplanetformation 5.1. Minimalsolidsurfacedensity Withthepopulationsynthesiscalculations,onecanstudyapos- teriori which combinations of disk properties allow the forma- tion of giant planets, and in particular which are the most ex- Fig.6. The mass of synthetic planets (in Jupiter masses) as a tremeones.ComparablestudieshavebeendoneinKornetetal. function of the initial disk mass for the nominal population (2006), Dodson-Robinson et al. (2006), Ikoma et al. (2000) or (black dots). The dotted line indicates a mass half as large as Thommesetal.(2008). the initial gas disk mass, showing the linear correlation of disk Figure 7 shows the relative planetesimal surface density massandmaximalplanetmassformostofthedomain.Thesolid Σ /Σ at 5.2 AU, where Σ = f Σ is the planetesimal S S,SN S D/G 0 lineshowstheupperlimitingenvelopeinanon-nominalpopula- surface density in a given disk and Σ = f Σ = 8 tionwherethelimitingeffectofgapformationonthegasaccre- g/cm2 is the planetesimal surface denSs,iStNy in oDu/rGs,(cid:12)ola0r,SNnebula tionrateistakenintoaccount,usingthefitofVeras&Armitage reference disk, which lead to the formation of a giant planet (2004). (M ≥ 300M ),asafunctionofthestartingpositionoftheseed ⊕ embryo. The core accretion rate M˙ ∝ Σ (e.g. Alibert et al. core S 2005a), therefore this quantity is within the core accretion for- For[M /M ] (cid:46) −0.2,themaximalmassesdecreasefaster mationparadigmacentralcontrolparameter.Bluefilledcircles D SN then linearly with decreasing [M /M ] and are thus smaller representssyntheticplanetsofthenominalpopulationinallfour D SN than indicated by the dotted line. This is because at such low panels.Avalueofe.g.ΣS/ΣS,SN = 3thuscorrespondstoadisk disk masses, the low resulting solid surface Σ = f Σ den- with roughly three times more solids than the amount we as- S D/G 0 sities becomes important in a second order effect, so that the sumeforthesolarnebula.Inalldisks,thesolidsurfacedensity final planetary mass is no more just determined by the amount scalesasa−1.5 withajumpofafactor4attheiceline,therefore ofgasthatcanbeaccreted,butalsobythetimeneededtoform this value corresponds for a starting position of the embryo of asupercriticalcore.Thisisaneffectthatisanalogoustothede- e.g.astart = 10AU(whichisoutsidetheicelineforallΣ0 con- pendenceofthemaximalmassseenatlow[Fe/H].Notethatthe sidered here) to a local planetesimal surface density at astart of exactvalueofthehighestefficiencyofconvertingdiskgasinto about3×8×(10/5.2)−1.5 =9g/cm2. planetaryenvelopematerial(hereabout0.5)dependsontheex- act treatment of the back-reaction of the planet’s accretion on 5.1.1. Sweetspotforgiantplanetformation the disk, but is found to lie in a domain of 0.3 to 0.5. The lin- earcorrelationbetweendiskmassandmaximalplanetarymass The plot shows that the sweet spot for giant planet formation is however not affected by different treatments. This indicates (i.e. the lowest solid surface densities which allow that) occurs thatapositivecorrelationofdiskmassandgiantplanetmassis at about 5 AU, at a value of about 0.75, corresponding to 6 astablepredictionofthecoreaccretiontheory,whilethequanti- g/cm2. This value is similar to the result of Dodson-Robinson tativedegreeismodeldependent.Forthelargemajorityofgiant et al. (2006) who find that a solid surface density of about 6.5 planets (M ≥ 1M ) the efficiency of converting disk gas into g/cm2 (also at 5.2 AU) is needed to bring an embryo to run- planetary material is lower and to order of magnitude 0.1, but away gas accretion in 7 Myr, which corresponds to the longest (cid:88) with a very wide spread of possible values, depending on e.g. living disks in our population (Paper I). At a starting position the photoevaporation rate, the metallicity or the disk gas mass of1AU(whichisinsidetheiceline),aΣ /Σ ofatleast10 S S,SN itself. isneeded,correspondingtoalocalplanetesimalsurfacedensity ThesolidlineinFigure6indicatesagaintheapproximative of about 0.25×10×8×(1/5.2)−1.5 ≈ 240 g/cm2, which is in upper envelope obtained for the non-nominal population using good agreement with Kornet et al. (2006). We see that both at gas accretion rates limited because of gap formation according smallerandatlargerdistances,moremassivedisksofplanetesi- totheVeras&Armitage(2004)fit.Inthiscase,theplanetarygas malsareneeded.Thisisinagreementwiththeworksmentioned accretion rate is still proportional to M˙ but decreases expo- earlier.Theabruptincreasebyaboutafactor4atabout3-4AU disk nentiallyasexp(−M /1.5M )withplanetmassuntilafloor simplycorrespondstotheincreasenecessarytocompensatethe planet (cid:88)