Orianne ROOS CEA-Saclay Collaborators : F. Bournaud, J. Gabor, S. Juneau
Bachelor of Physics, Master of Astrophysics Université de Strasbourg PhD, Université Paris-Diderot Observatoire de Strasbourg Les Grands Moulins How did I end up here? M2 internship in SAp, then PhD
«Modeling the link between AGN and star formation in primeval galaxies» Primeval/high-redshift galaxies AGN, AGN feedback Link with star formation HPC, simulations Stellar feedback
Some galaxies suddenly stop forming stars (they get quenched and then they die ) : why? Ideal culprit : central supermassive black hole SMBH mass related to bulge mass -> co-evolution? Energy generated by SMBHs in active phases (AGNs) theoretically able to blow away all gas of the host AGNs needed to reproduce observed number of stars in simulations => Link between AGN feedback and SF quenching?
Halo (gas and globular clusters) To turn the Earth into a black hole, you would need to squeeze it into a 9 mm radius sphere!! Bulge (stars) Disk (gas, stars, dust, ) Supermassive black hole 10^9 solar masses Radius : 3 x 10^9 km
Halo (gas and globular clusters) Bulge (stars) Disk (gas, stars, dust, ) Supermassive black hole 10^9 solar masses Radius : 3 x 10^9 km
8
Local disk galaxies look like spirals Clumpy, gas-rich high-redshift disk-galaxies Gabor & Bournaud, 2013
z band H band z - H z band H band z - H Typical star-forming galaxies at z ~ 2. (Guo et al 2012) 10
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
From Urry & Padovani 1995, modified (Schematic) Structure of an AGN : Active Galactic Nucleus Unified AGN Model : BLR (< 3 ly) NLR (> 300 ly) Radio jets (>>3 kly) Emission cone
Ground-based Optical/Radio image AGN HST image Artist view Reynolds et al 2014
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
A clump of gas falls onto the AGN Thermal feedback : HEATING in the simulation AGN absorbs energy due to accretion and re-emits fraction of it, heating central region Creation of a hot and diffuse outflow of gas Radiative feedback : IONIZATION added a posteriori Energy of photons emitted by AGN is so high that encountered gas is ionized
From Urry & Padovani 1995, modified Structure of an AGN : Active Galactic Nucleus Unified AGN Model : BLR (< 3 ly) NLR (> 300 ly) Radio jets (>>3 kly) Emission cone
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
AGN feedback vs star formation : Stars form into cold and dense clumps of gas WHEREAS AGN feedback dilutes and heats gas Is there an impact on star formation?
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
High Performance Computing More than 5000 computation nodes 5 PB of disks ( 100 GB/s of bandwith) 10 PB of bands 1 PB of cache
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
Do AGN (outflows) quench SF? Gas temperature Max. resolution : 6 pc GMCs are resolved. High-velocity AGN-driven outflows Mass outflow rates : ~ 10-100 % of SFR (~ 30 M_sun/yr). 1 snapshot = 1 moment in the simulation Gabor & Bournaud 2014
Simulation Do AGN (outflows) of a high-redshift quench SF disk-galaxy? Gas temperature AGN outflows do Max. resolution : 6 pc GMCs not quench are resolved. star formation. High-velocity AGN-driven outflows but : Do AGNs Mass outflow rates : ~ quench 10-100 % SF of SFR? (~ 30 M_sun/yr). AGN photoionization 1 snapshot = 1 moment in the simulation Gabor & Bournaud 2014
Does AGN radiation quench SF? Gas density map of the disk seen edge-on What happens in the disk?? RT with Cloudy (Ferland et al 2013) (cannot probe positive feedback) 3000 lines cast isotropically from location of BH Galactic disk edge-on Zoom in Simulation with standard thermal AGN feedback (Gabor & Bournaud 2013) + RT post-processing (Roos et al. 2015 ApJ 800 19)
log Initial r SFR [M O yr -1 kpc -3 ] log Final r SFR [M O yr -1 kpc -3 ] log log Initial Initial r temperature [K] SFR [M O yr -1 kpc -3 ] log Rel. temperature change log Final r SFR [M O yr -1 kpc -3 ] log log Initial Gas temperature density [cm -3 [K] ] log Rel. temperature change log Fraction of neutral H log Gas dens log Fraction of n Before RT -6 L_AGN 10 L_AGN 100 L_AGN Before RT rho_sfr 0-2 -4 610 4 8 2 06-2 -4 4-6 2 10-6 8-8 -10 6-12 -14 4-16 2-18 -6 f_ion L AGN 10 x L AGN 100 x L AGN -6-8 -10-12 0-2 6-4 4-6 2-8 -10 0-12 -2-6 6-8 4-10 -12 2-14 0-16 -2-18 -6-8 Even at a QSO -10 luminosity, the galactic disk remains completely neutral! Edge-on view -12 super-powerful -14 AGN -16-18 -8-10 -12-14 -16 L_AGN = 10^44.5 erg/s -18
Is the impact of AGN feedback luminosity-dependent? With a higher AGN luminosity : - Regions ionized by AGN enlarged - Most heated gas is in the gaseous halo - Galactic disk remains neutral even for QSO luminosity Let s focus on the central region : 600 x 600 pc (face-on view)
log Initial r SFR [M O yr -1 kpc -3 ] log Final r SFR [M O yr -1 kpc -3 ] log log Initial r temperature [K] SFR [M O yr -1 kpc -3 ] log Rel. temperature change log Final r SFR [M O yr -1 kpc -3 ] log log Initial Gas density temperature [cm -3 ] [K] log Rel. temperature change log Fraction of neutral H log Gas dens log Fraction o n Before RT -6 L_AGN 10 L_AGN 100 L_AGN Before RT rho_sfr 0-2 -4 610 4 8 2 06-2 4-4 -6 2 10-6 -8 8-10 6-12 -14 4-16 2-18 -6 f_ion L AGN 10 x L AGN 100 x L AGN -6-8 -10-12 0-2 6-4 4-6 2-8 -10 0-12 -2-6 6-8 4-10 -12 2-14 0-16 -2-18 -6-8 Even at a QSO luminosity, the bulk of the star forming clumps is not affected! -10-8 -10-12 super-powerful -14 AGN -16 Face-on view zoom in -18-12 -14-16 L_AGN = 10^44.5 erg/s -18
log Initial r SFR [M O yr -1 kpc -3 ] log Final r [M yr -1 kpc -3 ] log Initial temperature [K] log Final temperature [K] log Gas density [cm log Fraction of neutr 2 Does AGN radiation 0 quench SF? -2-4 -6-8 -4-6 10 - Diffuse star-forming 8 regions are prevented from forming stars 8 - The stronger the AGN, the more efficiently gas is heated/ionized 6-10 -12 10 6 4-100 But pc dense star-forming clumps shield themselves 2-6 -8-10 -12-14 4 2-6 -8-10 -12-14 -16-18 -16-18 -> Major contributors to the total SFR are not affected. Roos et al. 2015 ApJ 800 19
SFR [M ] O yr -1 ] (> 10 cm -3 ) ) 40 35 Instantaneous reduction of the total SFR due to AGN photo-ionization SFR reduction is not significant Density of affected gas is small 30 25 20 15 10 5 No feedback Thermal FB Initial FB + RT : 1 x L AGN 10 x L AGN 100 x L AGN 0 20 40 60 80 Time [Myrs]
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
Physical Origins of Galactic Outflows Do AGN+stellar outflows quench star formation? in z~2 star-forming galaxies Work in progress! 11 Mh of computation time, very high resolution (1.5 pc), AGN feedback + stellar feedback, suite of 7 simulations
Stars-driven outflows : High mass outflow rate Limited velocity (100-500 km/s) Multiple SNe and young stars per galaxy Bournaud et al. 2014, AGN-driven outflows : Low outflow density High velocity (3 000-30 000 km/s) 1 AGN in the center of the galaxy (usually) Gabor & Bournaud 2014,
Example : 10 kpc M*=3.5x10 10 Outflows generated by : - Supernovae (kin. energy) - Young stars (rad. pressure) - Both (no AGN)
Three runs with different feedback processes, all evolved for 80 Myr. Supernovae OB stars SNe + OB stars 10 kpc M*=3.5x10 10 Outflow rates 2.6 6.3 34.1 Outflows from SNe + OB stars >> Outflows from SNe + Outflows from OB stars See also Hopkins+14 Gas density of simulated disk seen edge-on
Three runs with different feedback processes, all evolved for 80 Myr. Supernovae OB stars SNe + OB stars 10 kpc M*=3.5x10 10 Accurate modeling of stellar feedback is crucial for the outflow parameters!! Outflow rates 2.6 6.3 34.1 Outflows from SNe + OB stars >> Outflows from SNe + Outflows from OB stars See also Hopkins+14 Gas density of simulated disk seen edge-on
Thermal energy injection from AGN Ionization from AGN Thermal energy injection from SNe Kinetic energy injection from SNe Radiative pressure from OB stars High space and mass resolution : probe GMCs and avoid numerical coupling + half resolution : check convergence
Supernovae OB stars SNe + OB stars 10 kpc M*=3.5x10 10 Thermal AGN feedback 10 kpc Outflow rates 2.6 6.3 34.1 First time ever. Let s POGO! (a) + RT postprocessing ~30 Gabor & Bournaud 2013 (b) Combined effect of all feedback models may lead to very powerful and fast winds with high mass loading. M1 : Mgas = 15 1E9 Msun M2 : Mgas = 49 1E9 Msun M3 : Mgas = 115 1E9 Msun M3 : AGN + stellar FB M1, M2 : AGN ; stellar FB ; AGN + stellar FB
Beginning Stellar outflows AGN + stellar outflows AGN outflows t1 < t2 < t3
Beginning Stellar outflows AGN + stellar outflows t1 < t2 < t3 With M1 and M2 : Study outflow characteristics as a function of time Identify outflow pattern for AGN FB only and stellar FB only Study outflow pattern for AGN + stellar FB : - = sum? - > sum? - < sum? Nature of expelled gas : impact on SF?
Study 3 galaxy masses : < Outflow rate > Very high? *-dominated? AGN-dominated? High Mstar M1 M2 M3 Mass loading compatible with observed IGM density?
If such dense UFOs are produced : Evolve until steady state is reached Study mass loading, expelled gas, Impact on star formation? Impact on IGM? UFO = ultra-fast outflow
If such dense UFOs are NOT produced : What other wind sources could produce UFOs? What kills them? What other non-linear effects are we missing?
Primeval/high-redshift galaxies AGN AGN feedback Link with star formation HPC To kill or not to kill? Stellar feedback
SUMMARY : For further details, see Roos et al. 2015 ApJ 800 19. - AGNs, through ionization and outflows, affect the gas of the galaxy, but mostly the non star-forming phase (halo : ionization cone). - Total instantaneous SFR is marginally reduced by thermal and radiative AGN feedback, even at QSO luminosities (< 4 %) AGNs do not kill galaxies - In association with outflows : + Gabor & Bournaud 2014 Frequent AGNs in high-z SFGs seem to be a promising mechanism to regulate/remove the mass of galaxies, without impacting star formation (up to 100 Myr-scale), i.e. preserving the steady-state evolution with relatively constant SFHs. Would this conclusion change with accurate modeling of stellar and AGN winds? Stay tuned for the POGO project!
Thank you for your attention.
log Initial temperature log Gas [K] density log [cm Gas -3 ] log Initial r log Initial density [cm -3 ] log Initial r SFR [M O yr -1 temperature kpc -3 [K] SFR [M O yr -1 kpc -3 ] ] Orianne Roos, with S. Juneau, F. Bournaud and J. Gabor see Roos et al. 2015 log Final r log Final log Final r log Final temperature [K] SFR [M O yr -1 temperature kpc -3 [K] SFR [M O yr -1 kpc -3 ] ] log Fraction of neutral log Fraction H of neutral 4-2 hermal and radiative 2 AGN feedback in high-z galaxies -4 n Before RT 0 6-2 4-4 2-6 010-2 8-4 mainly affect diffuse 4 gas ρ_sfr -6 6 10 8 2 6-6 -8 4-10 2-12 -6-14 -8-16 -10-18 L AGN -12 and have very little impact on the star-forming phase of the ISM. -14-16 Reduction of SFR < 4% -18 f_ion 10 x L AGN 100 x L AGN L_AGN = 10^44.5 erg/s -6 0-8 -2-10 -4-12 -6 10-8 8-10 -12 6 10 Edge-on view -16 suppressed, but major contributors to the SFR are left unaffecte -18 4 8 Face-on view zoom in Diffuse and extended star-forming regions around the AGN are 2 6-6 -8 4-10 2-12 -6-14 -8-16 -10-18 -12-14
n Before RT L_AGN 10 L_AGN 100 L_AGN f_ion T rho_sfr Edge-on view L_AGN = 10^44.5
Simulation of a high-redshift disk galaxy Typical density profile of a LOP in the disk
Large-scale spectral synthesis code : Ferland et al, 2013 Computes radiative transfer and molecular chemistry along 1D lines Divides each line into thin zones Balances recombination and ionization processes Input : ionization source and density profile Output : ionization fraction, temperature, line emission
AGN SEDs X-rays UV optical FIR X-rays UV optical FIR
Seyfert- and QSO-matched SEDs
Emission cone : From STScI, modified by G. Rieke Mean half-opening angle : ~ 30 at low-z (Müller-Sanchez+2011)