Bologna The size evolution of early-type galaxies since z=2 P. Saracco 1, M. Longhetti 1, with the contribution of S. Andreon 1, A. Mignano 1, G. Feulner 2, N. Drory 2, U. Hopp 2, R. Bender 2 1 INAF – Osservatorio Astronomico di Brera, Milano 2 Max Planck Institute and University of Munchen

Bologna Outline of the talk Small/compact Early-Type Galaxies (ETGs) at z>1: first evidence A morphologycal study of a sample of 10 ETGs at 1.2<z<1.7: size evolution of ETGs required The population of ETGs at 1<z<2: new clues on their formation and evolution ? Summary and conclusions

Small Small size, high-density ETGs: first evidence Daddi et al. (2005) Hubble UDF - 7 ETGs z>1.4 HST-ACS obs., FWHM~0.12”, F850W filter, λ rest <3000 Ǻ Bologna

Cassata et al. (2005) K20 + GOODS data HST-ACS observations, F850W λ rest <3000 Ǻ Further evidence Bologna Trujillo et al. (2006) IR ground based observations FWHM~1.0 arcsec redshift Re [Kpc] MassRe [Kpc]

Are ETGs at z>1 really more compact/denser than local counterparts ? These results were based on HST optical observations sampling the blue and UV rest-frame of the galaxies sensitive to k-correction and star formation and/or HST optical observations sampling the blue and UV rest-frame of the galaxies sensitive to k-correction and star formation and/or seeing limited ground-based observations seeing limited ground-based observations Doubts on the reliability of the estimate of R e Doubts on the reliability of the comparison high-z vs low-z High-resolution near-IR obs. sampling λ rest ~6500 Ǻ for a reliable comparison between high-z and low-z ETGs. Bologna

Effective radius r e (arcsec) and mean surface brightness (SB) e within r e from Sersic profile fitting n=4 de Vaucouleurs profile n=1 exponential profile galfit (Peng et al. 2002) to perform the fitting after the convolution with the NIC2 PSFs “/pixel NIC2 images models residuals z=1.34 z=1.40 z=1.7 n=3.2 n=4.5 n=2.7 Bologna HST-NICMOS observations in the F160W (λ~1.6 µm) filter of a sample of 10 ETGs at 1.2<z<1.7. (Longhetti et al. 2007) Data sampling the rest-frame R-band (λ rest ~6500 Ǻ) at z~1.4, at a spatial resolution <0.8 kpc (FWHM~0.12 “)

It is a scaling relation between the effective radius R e [Kpc] and the mean SB e [mag/arcsec 2 ] Any deviation from the KR at z=0 should reflects the evolution of e due luminosity evolution. The ETGs follow this tight relation with  ~3 up to z~1.  is found to vary reflecting the luminosity evolution. Expected KR at z=1.5 passive luminosity evolution (maximum evolution expected for early-types). Observed KR at z=0. Expected locus for z<1.5 early-type galaxies in case of luminosity evolution. The Kormendy relation in the R-band Bologna

It is a scaling relation between the effective radius R e [Kpc] and the mean SB e [mag/arcsec 2 ] The ETGs follow this tight relation with  ~3 up to z~1.  is found to vary reflecting the luminosity evolution. Expected KR at z=1.5 passive luminosity evolution (maximum evolution expected for early-types). Observed KR at z=0. Expected locus for z<1.5 early-type galaxies in case of luminosity evolution. The Kormendy relation in the R-band The SB exceeds by ~1 mag the one expected in the case of PLE for constant R e, i.e. luminosity evolution does not account for the observed SB of ETGs at high-z. luminosity evolution does not account for the observed SB of ETGs at high-z. (Longhetti et al. 2007) Bologna

Are ETGs at z>1 really more compact/denser than local counterparts ? These results are based on HST near-IR observations sampling the red rest-frame of the galaxies NOT sensitive to k-correction and star formation and/or NO seeing limited ground-based observations NO doubts on the reliability of the estimate of R e High-z ETGs (at least some of them) are more compact then their local counterparts. (Longhetti et al. 2007) Bologna

GMASS sample 13 ETGs 1.4<z<2 Spectroscopic data Morphology based on HST-ACS obs. F850W (λ rest ~3000 Ǻ) (Cimatti et al. 2008) The Kormendy relation in the B-band Bologna

Literature and HST archive research Aim – to collect a large (larger than 10…!) sample of ETGs at z>1 with spectroscopic confimation of the spectral type; HST-NICMOS observations in the F160W filter; multiwavelength coverage (optical + near-IR) in order to study the population of ETGs at 1<z<2 from an homogeneous set of data and a uniform analysis covering a larger interval in luminosity; defining the scaling relations at z~1.5 (Kormendy, size-luminosity/mass relations) Sample 10 ETGs 1.2<z<1.7 from TESIS (Saracco et al. 2005; Longhetti et al. 2005) + 10 ETGs 1.4<z<1.9 from GDDS (Abraham et al. 2004; McCarthy et al. 2005) + 6 ETGs z~1.27 from RDCS (Stanford et al.1997; van Dokkum et al ETGs 1<z<1.8 from HDF-N (Stanford et al. 2004) + 2 ETGs z=1.4,1.9 from GMASS H-UDF (Daddi et al. 2005; Cimatti et al. 2008) + 1 ETGs z= W091 (Dunlop et al. 1996; Waddington et al. 2002) = 32 ETGs 1<z<2, 17.0<K<20, HST-NICMOS observations F160W NIC2 (0.075 ”/pixel) for 14 galaxies NIC3 (0.2 “/pixel) for 18 galaxies FWHM ~ 0.12 arcsec Bologna

Physical properties of ETGs Morphological parameters effective radius and surface brightness derived as in Longhetti et al. (2007); Simulations done also for NIC3 images 0.16 and 0.32 kpc at z~1.5 Absolute magnitudes, stellar masses, ages Fit to the observed SEDs (BVRIzJHK F160W) at fixed z Charlot and Bruzual models (2007, CB07) IMF=Chabrier SFHs τ=0.1,0.3,0.6 Gyr (best-fit τ<0.3 Gyr for 28 out of 32) Metallicity Z ☼,0.4 Z ☼ (best-fit Z ☼ ) A V <0.6 mag (best-fit A V <0.3 for 24 out of 32 ) Bologna

The Kormendy relation in the R-band z=0 z~1.5 The ETGs at z~1.5 are placed on the [ e,R e ] plane according to the KR. z~1.5 ETGs follow the same KR of ETGs at z=0 but with a different zero-point. Saracco et al Bologna

Luminosity evolution Only 40% (13 gal) of the sample occupies the KR at z=0. The remaining 60% (19 gal) does not match the local KR, the SB exceeds by mag the one expected. Two distinct populations ? Each ETG evolves from z=z gal to z=0 according to its own SFH. Saracco et al Bologna

Two distinct populations ?Two distinct populations ! Saracco et al. 2008

Bologna Two distinct populations of ETGs at 1<z<2 Old ETGs, ~3.5 Gyr, =1.5  z f >5 Their stellar population formed in the early universe. Pure luminosity evolution does not account for their high SB. The evolution of their size must be invoked. Young ETGs, ~1.2 Gyr, ~1.5  z f ~2.5 Their stellar population formed much later than the stellar population of Old ETGs. Pure luminosity evolution from z gal to z=0 brings them onto the local KR.

Bologna Size-Luminosity/Mass relations SDSS Shen et al. (2003) Size-LuminositySize- Mass

Bologna Size-Luminosity (S-L) relation Saracco et al Young Old R e of oETGs is times smaller than - the local ETGs and - the yETGs with comparable luminosity.

Bologna Size-Mass (S-M) relation Saracco et al R e of Old ETGs is times smaller than - the local ETGs and - the yETGs with comparable stellar mass. Old ETGs are times denser ! Young - 9 out of 13 (70%) follow the S-M relation Old - 4 out of 19 (20%) follow the S-M relation

Constraining the formation and the evolution of ETGs Bologna Two distinct populations of ETGs at z~1-2 1.How did these two populations evolve from z~2 to z=0 to match the properties of the local ETGs ? 2.Which assembly history did they follow to have the properties shown at z~1.5-2 ?

Tracing the evolution at z<2 Bologna oETGs Luminosity evolution DOES NOT bring them onto the local Kormendy and S-L relations. They DO NOT match the local S-M relation. They are 2.6(±0.5) times smaller than their local counterparts. They must change their structure. Size evolution from z~2 to z=0 is required to move them onto the local scaling relations.

Tracing the evolution at z<2 Bologna oETGs Size evolution often used to advocate the merging processes the ETGs should experience in the hierarchical paradigm of galaxy formation. Dissipation-less (“dry”) merging is the most obvious and efficient mechanism to increase the size of galaxies. The size of ETGs increases according to the relation Boylan-Kolchin et al Khochfar and Silk 2006 Nipoti et al Ciotti et al. 2007

Tracing the evolution at z<2 Bologna oETGs - Merging would produce too much ETGs with M>10 11 M sun : we should observe 3 times more ETGs with M>4-5x10 11 M sun. - Why α=1.3 ? Merging cannot be the mechanism with which oETGs increase their size at z<2. Alternative mechanism(s) leaving nearly unchanged the mass and relaxing the system: 1.interactions between galaxies (e.g. close encounters) 2.minor or “satellite” merging (Naab et al. 2007): M 1 :M 2 = 0.1:1 Efficiency can be constrained from simulations.

Tracing the evolution at z<2 Bologna yETGs Luminosity evolution brings them on the local Kormendy and S-L relations. They match the local S-M relation. No size evolution is required. To move them along the S-M, α~0.6  M f ~5M i No evidence of merging at z<2. The build-up of yETGs was already completed at z~2.

Constraining the path at z>2 - Toward the formation of ETGs Bologna oETGs ~3.5 Gyr, =1.5  z f >5 (Age Univ. 4.2 Gyr at z=1.5)  To build-up M sun SFR>>100 M sun /yr Size times smaller  mechanism(s) acting at z>2 must be capable to produce galaxies 5-10 times more compact (15-30 times denser) than local ones Gas-rich merging with high fraction of stars formed during the merger in a violent starburst can produce highly compact ETGs (Khochfar et al. 2008; Naab et al. 2007). BUT t merger >3 Gyr

Constraining the path at z>2 - Toward the formation of ETGs Bologna yETGs ~1.2 Gyr, ~1.5  z f >2.5 Constraints on the mechanism(s) acting at z>2 less stringent: They can increase their mass and enlarge their size by subsequent mergers (major and minor/satellite) and through starburts till z~2.5 (contrary to oETGs). Different progenitors oETGs: we should see them as they are (younger) till z~3-3.5 yETGs: in the phase of merging, or star forming and interacting with other galaxies at z>2.5

Bologna Two distinct populations of ETGs at z~1-2 whose stellar populations differ in age by about 2 Gyr Young ETGs: No size/mass evolution is required. Old ETGs: Strong size evolution is required at z<2. The system must relaxes from high to low redshift  oETGs must show higher central velocity dispersion than local ETGs and than yETGs. Key observational test: measuring the velocity dispersion of oETGs. ESO-P82 VLT-FORS2: spectra of 10 oETGs, 10 hrs/spec Observations started in November 2008…we shall see! Summary and conclusions

Bologna Mean age vs stellar mass 5% Stellar mass

The evolution of the zero point α Zero point α of the KR derived from various samples at different redshifts. The curves show the expected evolution of α for different formation redshift zf. Luminosity evolution + Evolution of R e Our sample Luminosity evolution SFH tau=0.6 Gyr, solar metallicity, Chabrier IMF Longhetti et al Bologna

Luminosity evolution of Young and Old ETGs Saracco et al. 2008

Absolute magnitudes Bologna

Morphological study of a sample of 10 ETGs at 1.2<z<1.7 based on HST-NICMOS observations in the F160W (λ~1.6 µm) filter (Longhetti et al. 2007) NICMOS data - NIC2 camera (0.075 “/pixel) sampling the rest-frame R- band (λ rest ~6500 Ǻ) at z~1.4, at a resolution <0.8 kpc (FWHM~0.12 arcsec) Sample - K<18.5, spectroscopic confirmation of the spectral type from TESIS ( TNG EROs Spectroscopic Identification Survey; Saracco et al. 2003, 2005; Longhetti et al ). Bologna

Estimating the mean age of the stellar population 5% Stellar mass0.5 Gyr old 95% stellar mass, 4 Gyr old B V R I z J H K

Bologna Size-density and mass-density relations Saracco et al. 2008

100 simulated galaxies magnitudes F160W and r e assigned randomly in the ranges 19<F160W<21 and 0.1< r e <0.5 arcsec (1-5 Kpc at z~1.4); axial ratio b/a and position angle PA in the ranges 0.4<b/a<1 and 0<PA<180 Bologna Simulations Real galaxies Simulated De Vaucouleurs profile To assess the robustness of the results we applied the same fitting procedure to a set of simulated galaxies

NIC3 images (0.2 “/pixel) GDDS sample. z=1.65 z=1.73 z=1.85 NIC3 images (0.2 “/pixel) HDFS-NICMOS z=1.55 z phot =1.94 Bologna