Giant planet & Brown dwarf formation

Gilles Chabrier, Markus Janson, Anders Johansen, Roman Rafikov (Lyon/Exeter) (Princeton) (Lund) (Princeton) BROWN DWARFS OBSERVATIONS

• Brown dwarf census and mass function

• Brown dwarf vs young stars properties

• Brown dwarf and Giant planet as star or brown dwarf companions

• Proto- and Pre- Brown dwarf cores Brown dwarf mass function

T Y Reid et al. ’04 + Cruz et al. ’07 <10MJup Burgasser ’04 L Burningham et al. ’10 n(Teff) α=0 Gizis et al. ’00 Metchev et al. ’08 Reylé et al. ’10 Kirkpatrick et al.’12 T L Reylé et al. ’10 CFHBS Allen et al. ’05 n(MJ) Reid et al. ’04 + Cruz et al. ’07 Burgasser ’04 Burningham et al. ’10

Chabrier, ASSL 2005 NBD/N* = 1/4 - 1/3 WISE (now !)~1/5

Chabrier ’05 Young clusters and SFR: system IMF across the HB limit

field system IMF

0.01 Msol 5 Msol 0.01 Msol 5 Msol

Jeffries, EAS 2012

+ many «free floating» BDs discovered down to ~a few MJup (< 10 MJup) Multiplicity frequency continuous from the stellar to substellar regime Observed isolated Pre-Brown dwarf core

Oph B-11 M~30 MJup R<460 AU n~ 107-108 cm-3

Fig. 1. Interferometric 3.2 mm continuum image of the candidate pre-brown dwarf Oph B-11. The angular resolution (or synthesized half-power beamwidth) achieved with the CD array of the IRAM PdBI is 7.8’’"2.8’’ as shown at the upper left. The rms noise is ! ! 0.05 mJy/beam; the contours are -2! (dotted), 2!, 3!, 4!, 5!, and 6!. The SCUBA position (21) of Oph B-11 is marked by a cross whose size reflects the +-3’’ pointing uncertainty of JCMT. The PdBI continuum peak is located at RA(J2000) = 16h27m13s.96 +- 0.02s, DEC(J2000) = -24o28’29.3’’ +- 0.6’’, or at &% = -0.6”+-0.2”, &' = +1.75’’+-0.6” with respect to the SCUBA position. It has a peak flux density S3.2mm, peak ! 0.32 mJy/beam (6.2! detection). André et al., 2012, Science 337, 69 Young BDs vs young star properties (Luhman et al. 2007; Luhman ARA&A 2012) same radial velocity dispersion same spatial distribution in young clusters (see e.g. Bayo et al. ’11) consistent with the same IMF wide binary BDs form in low-density environments (e.g. Taurus) accretion + disk signature (large blue/UV excess, large asymetric emission lines, Hα => natural extension of CTTs disk fraction around BDs ~40-60% similar to stars timescales for accretion around BDs ~similar to stars ~1-10 Myrs presence of outflows

Observations of isolated proto-BDs and pre-BD BD and star formation : common dominant mechanism GIANT PLANETS OBSERVATIONS

• Correlations with stellar mass and metallicity

• Constraints from direct imaging Metallicity

Giant planet frequency is strongly dependent on stellar host metallicity

Johnson et al. (2010)

Expected from core-accretion but hardly from disk G.I. Correlation is weaker for lower- mass planets but stronger for lower-mass stars Trends in mass and radius

Mayor et al. (2011), Howard et al. (2013)

Planets increase with decreasing size/mass,

Stellar companions increase with increasing mass

Expected from core-accretion GP Mass dist’n ≠ BD Mass dist’n around ~ 30 MJup Statistical constraints from D.I. (with caveats!) apply both to BD’s and planets !

Lafreniere et al. (2007) Janson et al. (2012)

<23% of stars have >2 MJ planets at 25-450 AU <9% of stars have >5 MJ planets at 25-450 AU

<10% of stars host ~Jupiter-mass objects formed by disk instability Problem for C.A. : planets on wide orbits

(Rafikov 2010)

• HR 8799 (4 planets at 14.5, 24, 38, 68 AU) • Fomalhaut (1 controversial planet at 113 AU) • No way to form the cores of these planets within the lifetime of the protoplanetary disc by standard core accretion Pebble accretion • Most planetesimals are simply scattered by the protoplanet • Pebbles spiral in towards the protoplanet due to gas friction

Planetesimal is scattered by protoplanet • Pebbles are accreted from the entire Hill sphere •Growth rate by planetesimal accretion :

Pebble spirals towards protoplanet due to gas friction

•Growth rate by pebble accretion :

Ormel & Klahr 2010; Lambrechts & Johansen 2012 Time-scale of pebble accretion 0.10 0.10 Ω−1 Ω−1 5.0 0.05 t=0.0 t=120 > p Core growth to 10 M⊕

Σ 0.00 0.05 z/H /< 8

p −0.05

Σ −0.10 10 0.0 − 0.10 − 0.05 0 0.05 0.10 Planetesimals 0.00 x/H 7 y/H 10 −0.05 106 /yr

−0.10 t 0.10 5 Ω−1 Ω−1 ∆ 10 t=131 t=134 Fragments Pebbles 0.05 104

0.00 3 y/H 10 −1 0 1 2 −0.05 10 10 10 10 r −0.10 /AU −0.10 −0.05 0.00 0.05 0.10 −0.05 0.00 0.05 0.10 x/H x/H • Pebble accretion speeds up core formation by a factor 1,000 at 5 AU and 10,000 at 50 AU (Lambrechts & Johansen 2012; Ormel & Klahr 2010; Morbidelli & Nesvorny 2012) • Cores form well within the lifetime of the protoplanetary gas disc, even at large orbital distances • Requires large planetesimal seeds, consistent with turbulence-aided planetesimal formation (talk by A. Johansen et al.) (see poster 2H027 by Lambrechts et al.) Observations of a «dust trap» in a disk between 45-90 AU (Van der Marel et al., Science ’13) GJ 504 b (Kuzuhara et al. 2013)

Mp~3-8 MJup a=43 AU metal rich [Fe/H]=0.28

Pebble accretion ? In situ accretion (Rafikov 2004,2011) ? G.I. ? (cooling ? migration ?) FORMATION THEORIES Disk Fragmentation

BD’s :Vorobyov & Basu; Stamatellos & Whitworth; Bate et al.; GP’s : Mayer et al.; Boss

- Disk massive and cool enough to become gravitationally unstable, according to the Toomre and Gammie criteria cS  1 Prot Q = < 1 t ⇠ ⇡G⌃ cool . ⇡ 2 Comparison of the PdBI maps with MHD simulations Maury et al. 2010 Hydrodynamical simulations produce too much extended (+ multiple) structures if compared to the observations. Obs

MHD simulations ? Taurus Perseus Hennebelle & Teyssier (2008) MHD simulations : produce PdB-A synthetic images with No B typical FWHM ~ 0.2’’ - 0.6’’

Similar to Class 0 PdB-A sources observed ! w/ B

need B to produce compact, single PdB-A sources.

Massive extended disks prone to fragmentation not observed ! Rodriguez et al. ’05

Isolated, massive, compact (Rout~20 AU) disk around class 0 consistent w/ MHD disks ! Could be unstable but instability does not last long (≲104 yr) and most/all fragments quickly (~a few 103 yr) end up in the star (Machida et al. 2011) Thermodynamical constraints Rafikov 2005

GI planet formation

fragmentation

As a result, giant planet formation by GI requires

( ~ 100 MMSN) !

!!!

Requires rather extreme disk conditions •Gravitational coupling to the massive disk quickly migrates them into the star (Vorobyov & Basu 2005, Machida 2011, Baruteau et al. 2013) Constraints on BD formation/ejection by disk instability: magnetic field No B (pure hydro): no outflow -> too massive disks : inconsistent w/ observations (Maury et al. ’10) B : outflow + magnetic breaking (Price & Bate ’09, Commerçon et al. ’10; Machida et al. ’11, Masson ) Constraints on disk instability from class-0 disks Class-0 lasts ~ 105 yr (Evans et al., Enoch et al.) => most Class-0 objects should have massive disks: not observed !

Low-mass binaries with BD-mass companions

e.g. BD LHS6343c (Johnson et al. 2011) (other examples: Joergens 2008, Irwin et al., 2010, Bachelet et al., 2012) MA=0.37 Msol MC=0.063 Msol MDISK ≃ 0.04 Msol not enough mass in the disk to form LHS6343c formation by disk fragment’n wide BD binaries : (eg 2MASS J12583501/J12583798 6700 AU (Radigan et al. ’09)

Constraints on disk instability from direct imaging

Lafrenière et al. (2010), Johnson et al. (2010), Janson et al. (2011, 2012) statistics of companions around 85 ABFGKM stars in disks from 5 to 700AU companion frequency <10% at 99% C.L. Ejection, competitive accretion Reipurth & Clarke; Bate, Bonnell and coll.

- Collapse of a cloud -> starts forming small N-body clusters of small (~10-3 Msol) gravitationnally bound entities

- then accrete mass, M ˙ M 2 (Bondi-Hoyle) or M ˙ M 1 . 5 (tides) / / - dynamical interactions responsible for merging (-> high-mass stars) or ejection (LMS/BD)

• dynamical interactions are essential to build the IMF • all stars form in clusters • the IMF is determined by the gas-to-star conversion (no correspondence with the CMF) Bate 2012: fragmentation of a cloud w/ radiative feedback

IMF determined at the very begining of the simulations ? Would bring support to the gravoturbulent scenario !

4 -3 500 Msol; L=0.4 pc ; n ~ 3 10 pc ; Mach# =14

~ 4x denser and 4x more turbulent than (standard) Larson’s relations ! => overfavor dynamical int’ns !! idem NGC1333 : N*+BD (simus) = 191 Msol N*+BD (observed) ~ 50 Msol (Scholz et al. ’12)

André et al. ’09, <σ > ~ 0.4 km/s Walsh et al. ’07, ⌧ ⌧ collapse ⌧ collisions Muench et al. ’07, suggest little dynamical evolution during prestellar core formation Gutermuth et al. ’09, Bressert et al. ’10 <σ > ~ 0.4 km/s Other issues w/ competitive accretion / ejection scenario

- how do you preserve the correspondence between CMF and IMF ?

- how do you form BDs in low-density environments (eg Taurus) ?

- how do you preserve wide BD binaries ? (eg 2MASS J12583501/J12583798 6700 AU (Radigan et al. ’09) 2MASS 04414489+23015113 1700 AU (Todorov et al. ’10)

- how do you preserve disks, outflows in ejected embryos (proto-BDs) ?

- how do you end up having a universal IMF ? (Bate 2009 : feedback from LMS leads to a narrow range of Jeans mass in the irradiated gas )

- how do you form BD’s in isolation (eg Fu Tau A,B Luhman et al. ’09) ?

- all stars must form in clusters : contradicts obs (Bressert ‘10, ‘12) + formation of isolated massive stars (Bontemps et al. ’10, Bestenlehner et al. ’11, Bressert ’12) + formation of isolated proto-BDs (Palau et al. 12) and pre-BD cores (André et al. ’12) Gravoturbulent fragmentation Padoan & Nordlund; Hennebelle & Chabrier; Hopkins - Large-scale turbulence (density PDF lognormal) sets up a lognormal distribution of overdense regions at all scales in the cloud -> leads to the CO clump dist’n (can be filamentary (Inutsuka ’01)

- Virial: regions M(R) with |Egrav (R)| > (Eth+Erms(R)+Emag) collapse (introduces a scale dependence) -> leads to the core mass function (CMF)

- magneto-centrifugally outflows expel ~30-50% of the mass -> leads to the star mass function (IMF)

• turbulence sets up the density fluctuations at the very early stages of star formation • the IMF derives from the CMF and is imprinted in the very cloud conditions (, T, Mach)

Chabrier system IMF

7 HC analytical M IMF

5 M⇠

Hennebelle & Chabrier ’09, ’13

27 gravoturbulent fragmentation •supported by IMF (so far !) determined by the prestellar Core MF (André et al. 2010, Konyves et al. ‘12,10)

• «naturally» explains «universality» of the IMF (power spectrum of turbulence) • can lead to binary properties consistent w/ observations (Jumper & Fisher ’12) emerging observations of fragmentation at the Class-0 phase • can explain both the unbound clumps and bound cores MF’s (HC08,09) • fragmentation of cores seems to be limited (Bontemps et al. ’10; Longmore et al. ’11) • emerging observations of isolated proto-BD and pre-BD cores VeLLO L1148–IRS (Kauffmann et al. ’11); IRAS 16253-2429 (Wiseman et al.) J042118 + J041757 1-5 MJup (Palau et al. ’12); L328-IRS (Poster by C-W Lee et al.)

Oph B-11 30 MJup <460 AU n~107-108 cm-3 (André et al. ’12) support the idea that a BD forms like a star Observations and statistical properties of multiple Systems

About 50% of the stars are binaries (Duquenoy & Mayor 91) There are evidences that the fragmentation occurs very early when the star is still accreting.

Evidences for fragmentation Multiple embedded in the Class0 objects IRAS4A Young stellar objects (Observations realised with the in ρOph and Serpens BIMA interferometer)

NGC1333 IRAS4 - BIMA - 2.7 mm

5’’ 3’’

A2

A1 0.5’’ 1’’ Bonnell & Bate ’94 Looney, Mundy, Welch (2000) Machida et al. ’08 Duchêne et al. 2003 Conclusion 1

• Can star/BD’s form dominantly by G.I. or competitive accretion/ejection ? - not supported - so far - by observations - face many issues - IMF should vary appreciably w/ environment - need more realistic initial/physical conditions - can produce some objects (eg Delorme et al. ’13)

• Gravo-turbulent fragmentation - same theory yields CO-clump MF and bound core CMF - emerging population of isolated pre- and proto-BD cores - the CMF-to-IMF part still to be fully understood (magneto-centrifugally driven outflows (Matzner & McKee) - binary formation at the 1st and 2nd core stages still to be fully understood - might be complicated by the «filament condition» (talk by Hacar) (Inutsuka 2001 + chapter byAndré et al)

• Brown dwarfs & stars share many properties (notably the IMF) BD and star formation : common dominant mechanism ejection, disk frag’n, photoevaporation might play some role but are not essential for BD formation Conclusion 2 • Overwhelming majority of discovered exoplanets consistent w/ predictions from Core Accretion (ex.: Hat P 2b (9 MJup), Hat P 20b (7 MJup), Mcore > 200 MEarth (Leconte et al. ’09, ’11)

• Formation by G.I. requires very peculiar (extreme ?) conditions. Adding migration makes the problem even more accute (requires some «fine tuning»). Even if G.I. occurs, survival of the fragment is still en issue

• G.I. might be operational at large orbital distances. However: * scarcity of objects at such distances (constraints from direct imaging) * C.A. might work («pebble» accretion) * other possibilities (outward migration, planet scattering,....) ... stay tuned ! Future projects (SPHERE, GPI, CHARIS) will help

• ! Kappa And b (~10-20 MJup, 40 AU) companion to a B9 star w/ q=M2/M1<1% 2M 1207 b (Chauvin et al. ’05, Mamajek et al. ’05) (~4 MJup) companion to a BD w/ q=20% => existence of non D-burning BD’s / D-burning planets not unlikely BD and GP mass domain overlap Deuterium burning plays no role IAU definition scientificallY meaningless Effect of an increase of metallicity (factor 5) on spectra (Chabrier et al. 2007)

• deeper CO absorption spectrum calculated by T. Barman • enhanced K-band flux

Fortney et al. 2008e

Texte