The role of massive halos in the cosmic star formation history A&A, 574, A105 (2015) A&A, 579, A132 (2015) A. Biviano, INAF-Osservatorio Astronomico di Trieste
Introduction The cosmic z-dependent Star Formation Rate Density SFRD(z) with z up to z~1, and does not decrease before z~2-3 (e.g. Madau+98, Reddy+08) SFRD dominated by LIRGs at high-z, not at low-z (e.g. Le Floc'h+05) Group number density with time, more and more galaxies experience the group environment. Is this related to in cosmic SFRD(z)? Let's measure the contribution of halos of different mass to SFRD(z) Two ingredients needed: 1) IR luminosity functions (LFs) 2) Halo masses
The Data Chandra & XMM X-ray selected groups with LX (Finoguenov+15) COSMOS Spitzer @ 24 μm (Le Floc'h+09, Sanders+07) Herschel PEP PACS @ 100 160 μm (Lutz+11, Magnelli+13) CDFN, ECDFS, GOODS: Spitzer 24 μm Fidel Pgm (Magnelli+09) Combined PACS PEP (Lutz+11) & GOODS-Herschel (Elbaz+11) @ 70, 100, 160 μm (Magnelli+13) IR sources associated with optical (Capak+07, Cardamone+10, Berta+10 for COSMOS, CDFS, CDFN) via Max Lik Redshifts for COSMOS (SDSS or zcosmos via VLT/VIMOS, Lilly+07; Keck/DEIMOS, Scoville+; Magellan/IMACS, Trump+07; MMT, Prescott+06); CDFS (Cardamone+10, Silverman+10, ACES: Cooper+12, GMASS: Cimatti+08); CDFN (Barger+08)
The Sample 38 X-ray selected groups at 0<z 1.05 (Finoguenov+15) + the group of Kurk+08 at z=1.6 + stack of GAMA low-z optically selected groups (Robotham+11) made by Guo+14 + 9 clusters from Popesso+12 + Coma cluster of Bai+06 + 3 stacks of LoCuSS clusters by Haines+13 Spectroscopic completeness 60% down to 0.06 mjy @ 24 μm Cluster/group membership from projected phase-space distribution (l.o.s. velocities vs. clustercentric distances) using CLEAN algorithm (Mamon+13)
The group and clusters masses Derived from LX via scaling relation (Leauthaud+10), or for the most distant group and for clusters from velocity dispersion, via scaling relation (Mamon+13), or taken from the literature Define three subsamples: clusters: 2 1014 M rich groups: 5 1013 M - 2 1014 M poor groups: 6 1012 M - 5 1013 M
The galaxy IR luminosities Derived from SED fitting of the main sequence and starburst templates of Elbaz+11 to PACS (70, 100, 160 μm) and MIPS (24 μm) fluxes. If only MIPS available, adopt the main sequence template. Integrate over the 8-1000 μm spectral range of the template. SFR from LIR via Kennicutt's (1998) relation The IR luminosity functions (LFs) Derived from composite using Colless' (1989) method for groups. Taken from the literature for clusters and the global population
The group IR LFs vs. z Both LIR* and the normalization increase with z Models: Solid: Schechter (1976) Dashed: Saunders+90 (also seen in the general population, e.g. Magnelli+09, Gruppioni+13) (Guo +14) i.e. star-formation is a more common and more intense phenomenon among group galaxies at higher z
The group vs. global IR LFs To compare with the global IR LF of galaxies in all environments we must evaluate the IR LF of group galaxies per unit comoving volume. We do not have a complete group sample down to a certain mass. We then consider the IR LFs of groups in mass bins, and multiply them by the theoretical estimate of the comoving number density of DM halos with masses in the same bins (WMAP9 cosmology, Hinshaw+13; using Planck cosmology instead does not change much our results):
The group vs. global IR LFs Cmp with global IR LF: @ z~1 the group <SFR> is higher than the mean for all galaxies, and groups contribute a substantial fraction of the global SF galaxy population Guo +14 Sanders+03 Groups (points and best-fit) vs. Field (Gruppioni+13, Magnelli+13)
The comoving number density of IR galaxies as f=f(z,environment) Integrate IR LF comoving number density of IR galaxies in different LIR ranges Stars: groups; dots: global (Sanders+03, Magnelli+13) Strong z-evolution of (U)LIRG number density; faster in groups than in the whole galasy population At z~1, 50% IR galaxies (and 100% ULIRGs) are in groups Dashed lines: (1+z)α best-fits
The SFR density as f=f(z,environment), SFRD(z) Split the groups by mass, and consider also... massive clusters field Massive groups and clusters contribute <10% and <1%, resp. (low-sf and few) Low-mass groups make most of the cosmic SFR density at z 1.
Galaxies of different masses in halos of different masses Use the model of Guo+13 to estimate the fraction of galaxies of different masses: a) b) c) d) a) >1011 M b) 1010.5-1011 M c) 1010 1010.5 M d) 109-1010 M that belong to halos of different masses At z> 1 massive galaxies belong to groups
Observed vs. simulated SFRD(z) vs. De Lucia+06 Cmp with the SAM of De Lucia+06 applied to Millennium SAM predicts too low SFRD(z) and small (if negligible) contribution of groups to the cosmic SFRD at all z Over- and too-earlyquenching Dots: simulations Shadings: observations
Observed vs. simulated SFRD(z) Cmp with the Hydro sims of Genel+14 (Illustris) Excellent agreement with total SFRD(z) but group contribution still underpredicted (but less so) Overquenching by AGN radio-mode in massive galaxies? vs. Genel+14 Dots: simulations Shadings: observations
Observed vs. simulated SFRD(z) vs. Bethermin+13 Cmp with the abundance mathing model of Bethermin+13 Dots: simulations Shadings: observations Total SFRD(z) group <SFRD> Use observed evolution of SFR-M* (quenching is not immediate in groups) (field)
Summary and conclusions IR observations galaxy SFR; X-ray observations group masses spectroscopy membership (and cluster masses) Differential evolution of group IR LF vs. global IR LF: (U)LIRGs avoid groups at low-z, but prefer groups at z 1. SFRD(z) dominated by (low-mass) groups at z~1; partly due to mass segregation and SFR-M* relation. Clusters contribute <1%: they are too rare. Cmp with models: rapid (<1 Gyr) quenching of SF upon accretion onto groups is ruled out.