Coupling small and large scales: how massive and dying stars drive the formation of a galaxy P. Monaco, Trieste University and INAF-Osservatorio Astronomico di Trieste Micro-physics of star formation and feedback (SF&FB) SF&FB on large scales Implementations in semi-analytic models Implementations in simulations Conclusions
Galaxy formation efficiency must be a strong function of halo mass stellar feedback AGN feedback
1. The micro-physics of SF&FB
A matter of scales galaxy formation feedback star formation > kpc pc - kpc < pc baryons in dark matter halo cloud formation feedback from massive stars inside the cloud
Massive and dying stars Physical process: Energy budget: SN explosions Ionising radiation Stellar winds Radiation pressure 10 51 erg each >8 Msun star + type Ia SNe up to 10 50 erg each >10 Msun star up to 10 50 erg each >10 Msun star ~10 52 erg for a ~10 Msun star
Early feedback, radiation pressure Molecular cloud are destroyed by massive stars (HII regions etc.) SNe explode in pre-heated gas need for early feedback Radiation pressure has access to a high energy budget but the efficiency of radiation pressure goes like v/c good to push a thick outflowing shell (e.g. Monaco & Fontanot 2005)
Efficiency of feedback Stars are born in clouds / clusters SNII explode when the cloud has almost being destroyed by massive stars Correlated type II SNe create an expanding super-bubble (SB) SBs expand in the hottest phases SBs heat the ISM in the adiabatic stage SBs cool the ISM in the snowplough stage SBs end by pressure confinement or by blowing out of the disc Feedback efficiency is set by the stage in which the SB ends adiabatic stage snowplough stage Monaco (2004), Lagos et al. (2013) The many pathways momentum-driven to galaxy growth, stageprato, June 2015
2. Star formation on large scales
cosmological inflow (cooling of hot gas or cold flow) galaxy wind / outflow galaxy wind / fountain AM is conserved: gas disc Schmidt-Kennicutt law SBs blow out while adiabatic star formation with a given IMF
(one of the several) Schmidt-Kennicutt relation gas consumption time-scale surface density of SFR (Msun/yr/kpc 2 ) Bigiel et al. 2008 total gas surface density (Msun/pc 2 )
Other Schmidt-Kennicutt relations almost no correlation ~unity slope HI gas surface density molecular gas surface density Wong & Blitz (2002), Bigiel et al. (2008)
The quest for the molecular fraction Blitz & Rosolowsky (2008): the molecular fraction is regulated by (external) pressure Krumholtz+ (2009) models: molecular gas must self-shield from UV metallicity dependence Gnedin & Kravtsov (2010) model: fit to a set of simulations with radiative transfer and molecular chemistry RT simulations with a molecular network: the full monty
strong inflow, or disk instability, or galaxy merger massive outflow AM is not conserved: compact star-forming clump Schmidt-Kennicutt law SBs confined? starburst with a given IMF
A double relation at high redshift? Daddi et al. 2010, Genzel et al. 2010
The dynamical Schmidt-Kennicutt relation Daddi et al. 2010
Observation of warm outflows wind mass load Steidel et al. 2010 Chisholm et al. 2014
Observations of molecular outflows Cicone et al. 2014
3. Implementations of SF&FB in semi-analytic models
Semi analytic models: the prototype Dekel & Silk (1986) Einject = ε Eavail > Mgas Vc 2 deavail/dt = 10 51 erg ηsn SFR SFR = Mgas / τ tff critical velocity: Vc ~ 100 km/s
Semi analytic models: the classical implementation of SF&FB Cole et al. 2000 dmstar/dt = ε Mgas / τdyn dmej/dt / dmstar/dt = (Vc/V0) α the wind is: put in the hot halo and allowed to cool and fall back ejected from the halo and reaccreted later
Quiescient / starburst star formation Somerville et al. 2001
More sophisticated schemes adiabatic stage snowplough stage e.g. Monaco (2004), Lagos (2013): compute the efficiency of feedback by studying the evolution of a super-bubble and its break-out from the ISM to the halo momentum-driven stage
A comparison of models Knebe et al. 2015
A problem with small galaxies Small galaxies (<10 11 Msun) in these models form too early: they are too passive at z<3 they are already in place at z=1 they are too old at z=0 Model Data Fontanot, PM et al. 2007, 2009; Weinmann et al. 2011
Stellar mass downsizing Without error in stellar mass Log Stellar mass density (M sun Mpc -3 ) 10-10.5 10.5-11 11-11.5 11.5-12 in Log M * With 0.25 dex error in stellar mass Fontanot et al. 2009
Suppressing star formation with Dekel & Silk-like SN feedback? SN energy budget per unit mass defines a velocity: V 2 sn = ε sn Esn / Mstar,sn ~ εsn 700 km/s massive outflow if: V 2 sn > 2 Vc 2 at z~0 it must be minimal for: Vc~200 km/s but circular velocity scales with halo mass as: G Mh ~ Vc 2 R this corresponds to ~10 11 Msun at z~5 Lo Faro et al. 2009
Is reaccretion the key? Henriques et al. 2013
4. Implementations of SF&FB in simulations
Sub-grid SF&FB in simulations: the prototype Katz, Weinberg & Hernquist (1996) Criteria for star formation: convergent flow Jeans unstable density threshold (>0.1 H-atoms cm -3 ) overdensity threshold (>55 rhob) Star formation: + stochastic generation of star particles SN feedback: 10 51 erg per SN as thermal energy to neighbours Energy is quickly RADIATED AWAY!
Sub-resolution SF&FB in simulations Kinetic winds to improve efficiency (Navarro & Steinmetz 00) Effective model (Springel & Hernquist 03) Blastwave feedback (Stinson+ 06) Momentum-driven winds (Oppenheimer & Dave` 06) Sticky particles (Booth+ 07) Hypernovae (Kobayashi+ 07) Effective equation of state (Schaye & Dalla Vecchia 08) Scaling with halo circular velocity (Tescari+ 09, Okamoto+ 10, Oser+ 10) Density estimation of hot gas (Scannapieco+ 10) Multi-Phase Particle integrator (Murante+ 10) Early feedback (Stinson+ 13) Heating to a critical temperature (Schaye & Della Vecchia 13) Accelerating wind (Barai+ 2013) Resolving feedback (Ceverino & Klypin 09, Gnedin & Kravtsov 12) Radiation pressure (Hopkins+ 14) Superbubble feedback (Keller+ 2015)
The effective equation of state Disc thickness is determined by the Jeans length Surface density of star formation follows a Schmidt-Kennicutt relation star forming particles stay on an effective equation of state this determines the Schmidt- Kennicutt relation figura Schaye & Dalla Vecchia 2008
Kinetic winds Assume outflow rate is proportional to SFR: Scale kinetic energy with available SN energy: Kick particles with this velocity: Choose wind particles with a stochastic algorithm Decouple them - or not - while they flow away Springel & Hernquist 2003; Schaye & Dalla Vecchia 2008
Blastwave feedback Energy from SNe emerges to resolved scales from blasts after some time: Cooling is switched off for te yr in particles heated by SN feedback Strong winds are triggered provided the density threshold for star formation (and then resolution!) is high
Momentum driven winds This is a phenomenological approach: produce winds that obey known scaling relations of momentum-driven winds (Murray et al. 2005): fl is a luminosity factor, sigma the velocity dispersion of dark matter particles Oppenheimer & Dave 2006
Cold gas Ṁ cold =Ṁ cool -Ṁ * -Ṁ evap atomic hydrogen computed on the cold phase Murante et al. 2010 molecular hydrogen Ṁ cool = M hot /t cool Ṁ * = f * f mol M cold /t dyn star formation evaporation cooling Ṁ evap = f evap Ṁ * Ṁ rest = f rest Ṁ * computed on the hot phase restoration Ṁ star =Ṁ * -Ṁ rest Stars Hot gas Ṁ hot =-Ṁ cool +Ṁ rest +Ṁ evap
The Aquila comparison project Scannapieco et al. 2012
Keller+ 2015 Marinacci+ 2014 Hopkins+ 2014 Volgesberger+ 2014 Christensen+ 2014 Schaye+ 2015 Stinson+ 2013 Murante+ 2015 Keller+ 2015
Keller+ 2015 Marinacci+ 2014 Hopkins+ 2014 Volgesberger+ 2014 Christensen+ 2014 Schaye+ 2015 All these simulations use different sub-grid physics: are models degenerate? Stinson+ 2013 Murante+ 2015 Keller+ 2015
Outflow measurement technique Transform galaxy coordinates s.t. cold gas disk is rotating in X-Y plane Select gas particles: lying inside either cylinder moving at a high-velocity, v z > V limit,outflow if (z*v z > 0) Outflow if (z*v z < 0) Inflow 05-Oct-14 P. Barai, INAF-OATS Barai et al. 2015
Selection of outflow particles Fixed value: Reveals correlations. Adopt this for outflow velocity Escape velocity: Measures escape outside halo. Adopt this for mass outflow rate
Observation : Martin (2005), Grimes et al. (2009), Banerji et al. (2011), Bordoloi et al. (2013) - positive correlation of outflow speed with galaxy mass and SFR.
Conclusions SF&FB is a very complex problem, the emergent behaviour may be simple but we still don t understand all details However, simple recipes often do not work! Stellar feedback must be strong at high redshift and mild in local spiral galaxies Stellar feedback must produce outflows to limit the mass of galaxies beyond the knee of the luminosity function Different sub-grid physical prescriptions can give similar results, a detailed study of outflows may remove this degeneracy