Witnessing the onset of environmental quenching at z~1-2. Results from 3D-HST

Transcription

1 Witnessing the onset of environmental quenching at z~1-2. Results from 3D-HST Matteo Fossati USM/MPE Munich with D.Wilman, J.T. Mendel, R. Saglia, A. Beifiori, A. Galametz, R. Bender, N.M. Forster Schreiber, R. Genzel, P. Van Dokkum, & KMOS3D team Paris - 15 December 2016

2 The build-up of a hierarchical Universe Merger tree from Millennium simulation Halo Star forming satellite galaxy Subhalo Central Galaxy Passive satellite galaxy

3 Zooming-in on a galaxy ACCRETION Stars Molecular gas COOLING Properties characterizing galaxies: Central/satellite Halo Mass Stellar Mass Star Formation Rate (Colors) Gas masses (various phases) Redshift Warm-Hot gas Dark Matter Halo

4 3D-HST a game changer survey at z= DHST (PI P. Van Dokkum) grism spectroscopy in deep HST multi-band CANDELS fields. Accurate redshifts (σ ~1000 km/s), especially for emission line galaxies but also using continuum features f λ [10 18 erg s 1 cm 2 Å 1 ] a) z = 1.773, m 140 = 21.9 Hδ Hγ Hβ [OIII] λ f λ [10 18 erg s 1 cm 2 Å 1 ] b) z = 4.656, m 140 = 21.9 CII Ne IV Mg II λ F140W<24 mag f λ [10 18 erg s 1 cm 2 Å 1 ] c) z = 2.083, m 140 = λ f λ [10 19 erg s 1 cm 2 Å 1 ] d) z = 1.905, m 140 = λ Brammer et al. 2012, Momcheva et al. 2016

5 Computing galaxy densities Density is computed within cylinders centered on each galaxy: Mpc Radius +/ Km/s in velocity Accurate edge corrections are applied using extended photometric catalogues (cyan)

6 Calibrating environment Density is not a physical parameter. We calibrate it into Halo Mass. We use lightcones from the Henriques+2015 SAM, selected and processed to reproduce the number density and redshift accuracy of 3D-HST. N D HST MOCKS 11.0 log(m ) Σ 0.75 [Mpc 2 ] N

7 Calibrating environment Density is not a physical parameter. We calibrate it into Halo Mass. We use lightcones from the Henriques+2015 SAM, selected and processed to reproduce the number density and redshift accuracy of 3D-HST. Observational sample Density (0.75 Mpc aperture) Stellar Mass Mass Rank (most massive in the aperture?) Redshift Redshfit accuracy

8 Calibrating environment Density is not a physical parameter. We calibrate it into Halo Mass. We use lightcones from the Henriques+2015 SAM, selected and processed to reproduce the number density and redshift accuracy of 3D-HST. Observational sample Density (0.75 Mpc aperture) Stellar Mass Mass Rank (most massive in the aperture?) Redshift Redshfit accuracy Mock sample Density (0.75 Mpc aperture) Stellar Mass Mass Rank Redshift Redshfit accuracy Halo mass Central/satellite status For each 3D-HST galaxy we obtain a probability of being central/satellite (Pcen, Psat) and Halo mass PDFs given each type

9 Calibrating environment COSMOS (z=0.92) UDS (z=0.64) AEGIS (z=1.53) Central (P=0.91) Central (P=0.20) Central (P=0.54) 10 1 Satellite (P=0.09) Satellite (P=0.80) Satellite (P=0.46) PDF log(m h /M ) log(m h /M ) log(m h /M ) Most likely central Most likely satellite Uncertain classification Environment catalogue available at:

10 What number (or fraction) of galaxies become passive as a direct result of their surrounding environments? What is the main quenching mechanism? How long does it take for a satellite to be quenched?

11 Tracing evolution of the satellite population ( ) Locally, satellites are more likely to be passive than central galaxies at all stellar masses over at least 10 billion years of cosmic time Wetzel et al. 2013

12 Tracing evolution of the satellite population Satellites are more likely to be passive (in UVJ diagram) than central galaxies at all stellar masses over at least 10 billion years of cosmic time Passive Fraction log(m h /M ) < 13.0 Pure Centrals Pure Satellites Contaminated Centrals Contaminated Satellites <z 0.80 log(m h /M ) > log(m h /M ) < Pure Centrals Passive Fraction Pure Satellites Contaminated Centrals Contaminated Satellites 0.80 <z 1.20 log(m h /M ) > 13.0 Passive Fraction log(m h /M ) < 13.0 Pure Centrals Pure Satellites Contaminated Centrals Contaminated Satellites 1.20 <z 1.80 log(m h /M ) > Increasing redshift 0.0 Fossati et al. ApJ in press

13 Deconstructing the observed passive fraction ( ) What is the excess probability that a galaxy is passive as a result of its environment? Wetzel et al. 2013

14 Deconstructing the observed passive fraction ( ) What is the excess probability that a galaxy is passive as a result of its environment? Fraction of passive centrals Fraction of passive satellites fq sat excess = f Q, sat now f Q, cen now fa, cen now Fraction of star-forming centrals ~35 % of satellites became passive as a result of their environment Wetzel et al. 2013

15 Evolution of the conversion fraction Conversion Fraction log(m h /M ) < 13.0 zcosmos (Knobel + 13) zcosmos (Kovac + 14) This work 0.50 <z 0.80 log(m h /M ) > 13.0 Quenching efficiency declining with redshift, see also Darvish+16, Nantais+16, Muzzin+12 log(m1.0 /M ) Conversion Fraction log(m h /M ) < 13.0 GEEC2 (Balogh + 16) GCLASS (Balogh + 16) This work 0.80 <z 1.20 log(m h /M ) > 13.0 Conversion Fraction log(m h /M ) < <z 1.80 log(m h /M ) > 13.0 Fossati et al. ApJ in press

16 From conversion fraction to quenching timescale By assuming that the earliest accreted satellites also quench first, we can estimate how long it takes (on average) for galaxies to become passive Conversion Fraction Conversion Fraction <z <z log(m h /Mlog(M ) > < h /M 13.0 ) < 13.0 log(m h /M log(m ) > h /M 13.0 ) > < zCOSMOS zcosmos (Knobel + (Knobel 13) + 13) zcosmos zcosmos (Kovac + (Kovac 14) + 14) log(m h /M ) < This workthis work 0.6 This work Conversion Conversion Fraction Fraction 0.50 <z log(m log(m /M ) /M ) 0.0 log(m log(m /M ) /M ) Time since infall (Gyr) zcosmos (Knobel + 13) zcosmos (Kovac + 14) zcosmos (Knobel + 13) zcosmos (Kovac + 14) This work 0.50 <z l

17 From conversion fraction to quenching timescale By assuming that the earliest accreted satellites also quench first, we can estimate how long it takes (on average) for galaxies to become passive Conversion Fraction Conversion Fraction <z <z log(m h /Mlog(M ) > < h /M 13.0 ) < 13.0 log(m h /M log(m ) > h /M 13.0 ) > < zCOSMOS zcosmos (Knobel + (Knobel 13) + 13) zcosmos zcosmos (Kovac + (Kovac 14) + 14) log(m h /M ) < This workthis work 0.6 This work Conversion Conversion Fraction Fraction 0.50 <z log(m log(m /M ) /M ) 0.0 log(m log(m /M ) /M ) Time since infall (Gyr) zcosmos (Knobel + 13) zcosmos (Kovac + 14) zcosmos (Knobel + 13) zcosmos (Kovac + 14) This work Tquench 0.50 <z l

18 Quenching timescales are long! Quenching timescale (Gyr) Quenching timescale (Gyr) log(m h /M ) < 13.0 Satellite galaxies continue forming stars for 2-5 Gyrs after infall 0.50 <z log(m h /M ) > 13.0 Quenching times are independent of halo mass disfavoring rapid removal of atomic+cold gas via dynamical processes log(m h /M ) < <z log(m h /M ) > 13.0 Quenching timescale (Gyr) log(m h /M ) < <z 1.20 log(m h /M ) > 13.0 GEEC2 (Balogh + 16) GCLASS (Balogh + 16) This work Fossati et al. ApJ in press

19 Reproducing observations with fast vs. slow central galaxy satellite galaxy delay+quench slow quench infall quenching starts quenching starts adapted from Wetzel et al. 2013

20 Delayed suppression of satellite star formation Slow quenching over-produces transition galaxies Mixed delay+quenching model Wetzel et al. 2013

21 Delayed suppression of satellite star formation Fraction < < 10.5 Delay+Rapid τ f =0.6 Slow quenching 3D-HST satellites MS from Whitaker et al using same 3D-HST dataset Slow quenching over-produces transition galaxies < < 11.0 Delay+Rapid τ f =0.4 Mixed delay+quenching model fits satellite data (Red) Fraction MS log(offset from MS) Fossati et al. ApJ in press

22 Delayed suppression of satellite star formation Td/Tf Age Universe (Gyr) = = = Redshift Long quenching times require a significant residual gas reservoir (not only molecular)

23 A schematic picture for satellite quenching Dark matter Warm gas reservoir Molecular gas Stars

24 A schematic picture for satellite quenching 1.On infall, satellite galaxies lose their connection to the cosmic web, but retain significant gas reservoirs (both atomic and molecular)

25 A schematic picture for satellite quenching 1.On infall, satellite galaxies lose their connection to the cosmic web, but retain significant gas reservoirs (both atomic and molecular) 2. DELAY PHASE Over time, star formation slowly eats through the available molecular gas, which is replenished by cooling from the larger gas reservoir, depleting the overall gas supply but maintaining relatively normal star formation.

26 A schematic picture for satellite quenching 1.On infall, satellite galaxies lose their connection to the cosmic web, but retain significant gas reservoirs (both atomic and molecular) 2. DELAY PHASE Over time, star formation slowly eats through the available molecular gas, which is replenished by cooling from the larger gas reservoir, depleting the overall gas supply but maintaining relatively normal star formation. 3. FADING PHASE Once only molecular gas remains, the star-formation rate starts to decline as molecular gas is used up, resulting in a fading of star formation on relatively short timescales overconsumption model (McGee et al. 2014)

27 Conclusions and Reconciling to cluster observations - Our data supports a scenario where satellite quenching is driven by exhaustion of a multiphase gas reservoir over a long quenching time. This holds up to z=1.5, above this redshift the quenching time is too long to produce a significant population of quenched satellites. - How to reconcile with spectacular observations of rapid gas stripping in local clusters or statistical studies in high-z clusters? Despite being the most conspicuous environments, galaxy clusters are relatively rare, and probably not the dominant environment for most galaxies over their lifetimes Fumagalli, MF et al. 2014

28 VESTIGE: a large program to observe environmental effects in Virgo Boselli, MF et al. 2016