Deciphering the CIB 12 Oct 2012 Banyuls MODELING COUNTS AND CIBA WITH MAIN SEQUENCE AND STARBURST GALAXIES Matthieu Béthermin CEA Saclay In collaboration with Mark Sargent, Emanuele Daddi, Georgios Magdis, Olivier Doré, Guilaine Lagache and many others Material at

INTRODUCTION Star formation history: when and in which galaxies do stars form in the Universe? Which are the dominant processes of star formation in the Universe? In which dark matter halos do stars form?

OUTLINE 2 modes of star formation? IR SEDs of high-z galaxies. A new source counts model Putting IR galaxies into dark matter halos CIB fluctuation modeling

2 STAR-FORMATION MODES (2SFM) MODELING FRAMEWORK

NORMAL STAR-FORMING GALAXIES AND STARBURSTS Local Universe: most of the star formation in disk galaxies. However, higher SFR (higher IR luminosity) are mergers-induced starburst. Two sequences of star-formation for normal and starburst galaxies: star formation efficiency higher in mergers. Daddi+10

A “MAIN SEQUENCE” OF STAR-FORMING GALAXIES Strong correlation between SFR and stellar mass in star forming galaxies. Evolution of this correlation with redshift. Suggest a quiet smooth star formation history in the bulk of the star-forming galaxies. Correlation between stellar mass and SFR at z~2 (Daddi+07)

Distribution of sSFR at fixed M ✭ at z~2 (Rodighiero+11): Cross-section through main sequence at fixed M ✭ DISTRIBUTION OF SPECIFIC STAR FORMATION RATE AROUND THE “MAIN SEQUENCE” SFR-M ✭ relation at z~2 (Rodighiero+11): -> Self similar distribution of sSFR Excess due to starbursts

MAIN SEQUENCE-STARBURST DECOMPOSITION Sargent+12

Evolution of stellar mass function of star- forming galaxies: constant M* Φ* constant to z = 1; then declining Evolution of specific SFR (=SFR/M ✭ ) for main sequence galaxies EVOLUTION OF THE OTHER INGREDIENTS OF THE MODEL Sargent+12

RECONSTRUCTION OF THE IR LF FROM MASS FUNCTION AND SSFR DISTRIBUTION Mass function SFR function Animation from Mark Sargent

CONTRIBUTION OF MS AND SB GALAXIES TO THE INFRARED LUMINOSITY FUNCTIONS Combining 3 elements, we reconstruct the IR LF: -Mass function of star-forming objects -Evolution of the main sequence -Distribution of sSFR around the MS Main sequence Starburst Contribution of MS and SB modes to the bolometric infrared luminosity function from z=0 et z=2 (Sargent+12)

EVOLUTION OF SPECTRAL ENERGY DISTRIBUTION OF MAIN-SEQUENCE GALAXIES

FEW VARIATION OF SEDS OF MAIN-SEQUENCE GALAXIES WITH STELLAR MASS Mean spectral energy distribution of z~2 star-forming galaxies measured by stacking for several stellar mass bins (Magdis+12, resubmitted) -Mean SEDs of MS galaxies at z~2 (sample built using a BzK criterion and removing PACS detected starbursts). -We detect few variations of the SED with stellar mass.

FEW VARIATION OF SEDS OF MAIN SEQUENCE GALAXIES WITH SSFR Mean spectral energy distribution of z~2 star-forming galaxies measured by stacking for several sSFR bins (Magdis+12) -Mean SEDs of MS galaxies at z~2 (sample built using a BzK criterion and removing PACS detected starbursts). -We detect few evolution of the SED with stellar mass. -There are also few variations of the SED with sSFR/sSFR MS.

GAZ FRACTION AS A FUNCTION OF DISTANCE TO THE MAIN SEQUENCE Gas fraction as a function of the distance to the main sequence (Magdis+12) -Dust mass can be estimated from Mdust measured from Draine&Li model. -Gas mass estimated from dust mass and gas to dust ratio versus Z relation (Magdis et al. 2011) -Main sequence galaxies above the central relation seems to contain more gas than the galaxies below the sequence.

SED TEMPLATES OF MS AND SB GALAXIES Plots from Béthermin+12c Data from Magdis+12 Evolution of (radiation field in Draine&Li model) with redshift. -From Magdis+12, we derive an SED library for MS and SB. -We also assume a scatter on of 0.2 dex.

SOURCE COUNTS MODEL

SED TEMPLATES OF MS AND SB GALAXIES sSFR assumed to be flat at z>2.5. Evolution of SF mass function extrapolated at z>2. Simple prescription for AGN contamination in the mid-IR using Mullaney+11 templates and Aird+12 results. Dust attenuation increasing with stellar mass following Pannella+09. Take into account the magnification of a small fraction of distant by strong galaxy-galaxy lensing (computed based on Hezaveh+11 lensing model). Evolution of the various parameters of the model with redshift (Béthermin+12c)

MID-IR TO FAR-IR COUNTS FROM OUR FIDUCIAL MODEL Number counts are globally well reproduced using fiducial parameters (no fine tuning). The starburst (dashed line) have a very variable contribution depending on the flux regime and the wavelength. Comparison between the model and the observations (Béthermin+12c)

MID-IR TO FAR-IR COUNTS FROM OUR FIDUCIAL MODEL Number counts are globally well reproduced using fiducial parameters (no fine tuning). The starburst (dashed line) have a very variable contribution depending on the flux regime and the wavelength. Reproduce also counts per redshift slice. Comparison between the model and the observations (Béthermin+12c)

PUTTING IR GALAXIES INTO DARK MATTER HALOS

PRINCIPLE OF ABUNDANCE MATCHING TECHNIQUE Number density LIR = K -1 f(M * ) = K -1 h(M h ) Number density M * = h(M h ) Number density MhMh Assume a monotonic relation between infrared luminosity stellar mass and halo mass

MAIN SEQUENCE RECOVERED BY A SIMPLE ABUNDANCE MATCHING Relation between specific star formation rate and the stellar mass (Béthermin+12a)

LINK BETWEEN SFR AND HALO MASS Top: Ratio between SFR and halo mass as a function of halo mass Bottom: Differential contribution to SFR density as a function of halo mass. (Béthermin+12a)

CIB FLUCTUATION MODELING

COMBINING ABUNDANCE MATCHING AND POPULATION MODEL SFR-Mstar model for star forming galaxies (Béthermin+12c) Mass function spitted in star- forming and quenched galaxies (adapted from Ilbert+09) Halo mass function including sub-structures (Tinker+08,Tinker+09) Abundance matching Star forming Neglect the IR outputs of quenched galaxies Quenched Main hypotheses:- same Mstar-Mhalo relation in main and sub-structures - same Mstar-Mhalo relation for SF and quenched galaxies - the probability to be quenched depends only on the halo mass (no environnemental quenching) - starburst and main-sequence lies in the same halos PRELIMINARY

SFR TO HALO MASS RATIO SFR to halo mass ratio at various redshift used in our CIB model (Béthermin+ in prep) PRELIMINARY

COMPUTATION OF THE POWER SPECTRUM 2 halo term: 1 halo term: PRELIMINARY

PRELIMINARY RESULTS CIB power spectrum and galaxy counts (Béthermin et al. in prep.) PRELIMINARY

PRELIMINARY RESULTS CIB power spectrum and galaxy counts (Béthermin+ in prep.) Slightly modified evolution model: -sSFR in (1+z) instead of constant at z>2.5 (as suggested by e.g. Starck+12) -compensated by a decrease in density in (1+z) -0.8 instead of (1+z) -0.4 at z>2.5. Compatible with the observations. BUT, many other possible explanations… PRELIMINARY

SFR TO HALO MASS RATIO SFR to halo mass ratio at various redshift used in our CIB model (Béthermin+ in prep) PRELIMINARY

STAR FORMATION EFFICIENCY SFR to baryonic accretion rate ratio at various redshift used in our CIB model (Béthermin+ in prep) Salpeter IMF PRELIMINARY Baryon accretion rate deduced from Fakhouri+10

STAR FORMATION EFFICIENCY SFR to baryonic accretion rate ratio at various redshift used in our CIB model (Béthermin+ in prep) Chabrier IMF PRELIMINARY Baryon accretion rate deduced from Fakhouri+10

SUMMARY Evidences of two main modes of star-formation (main sequence and starburst). MS dominates the SFR budget at z<2. A simple model (2SFM: 2 star formation mode) based on our knowledge about MS and SB galaxies reproduce nicely the IR LF. If we use SEDs updated with Herschel data with a temperature of MS increasing with redshift, we are also able to nicely reproduce the infrared source counts, showing the predictive power of this simple approach. From simple prescription based on abundance matching, we can make puzzlingly accurate predictions of CIB fluctuations, reinforcing the clues of a maximum star formation efficiency around a halo mass of M sun. Material at