We propose to obtain medium-band imaging on 25 filters (FWHM~17 nm), covering the GOODS-N field down to the 26.5 mag (AB) in each filter. Our goal is to probe the rest-frame UV spectral range (250-350 nm) of galaxies at redshift z>1.0, counting with enough spectral resolution to detect and accurately measure the Mg(UV) absorption, which is a distinctive, necessary and sufficient feature of massive quiescent Early Type Galaxies (ETGs) at high redshift. A positive detection of the feature with our spectro-photometric technique will confirm that the galaxies are quiescent and provide us with an estimate of the age of the last star formation burst and the galaxy redshift (with an accuracy Δz/(1+z)<0.02). OSIRIS provides imaging capabilities up to ~950 nm, which means that we can probe the the Mg absorption in the redshift range z=1.0-2.3.

The resolution of this observing strategy (R=35-50) is similar to that used by the Grism ACS Program for Extragalactic Science (GRAPES; see Daddi et al. 2005) and the Probing Evolution And Reionization Spectroscopically project (PEARS; PI: Malhotra) to study quiescent ETGs through HST grism slitless spectroscopy. Our project will cover 10 times more area than the one used in Daddi et al. (who used the UDF), and reach 1.5 mag fainter (Daddi et al.'s sources have 22.5

Combining the proposed observations with the already existing ultra-deep multi-λ data in GOODS-N, we will be able to perform very detailed SED fitting in order to: (1) select an unbiased sample of quiescent ETGs at z=1.0-2.3 (distinguishable from lower redshift ETGs with a prominent Balmer/4000 Angstrom break and CaII H&K absorption lines, according to our simulations, see Figs. 2 and 3); (2) derive highly reliable photometric redshifts Δ(z)/(1+z)<0.02) for the whole galaxy sample with m[500-950]<26.5 using the well sampled SEDs (see Wolf et al. 2004, Ilbert et al. 2008, Fig. 5); (3) study in detail the physical properties of the quiescent ETG sample: stellar population age, stellar masses, residual star formation, traces of AGN; and (4) analyze the extinction properties of the red dusty galaxies at z>1.

Back to top

Passive galaxies

Observing strategy of our program. We show the averaged spectrum of the 13 GMASS quiescent ETGs in GOODS-S from Cimatti et al. (2008) at 4 different redshifts, together with the profiles of the proposed filter system. The emission line spectrum of the sky is also depicted in green. Using this combination of 25 order sorter filters (FWHM17 nm), we will be able to probe the prominent absorption feature placed at λ=265–295 nm, distinctive of passively evolving galaxies (marked in red) with a resolution R~50. The detection of this feature will allow us to obtain an estimate of the age of the last star formation burst, jointly with the galaxy redshift, with an accuracy better than Δz=0.02. We also mark the region covered with narrow-band tunable filters by the OTELO Guaranteed Time Project, which would allow to build medium-width bands useful for our project in the 920 nm atmospheric window.

The GMASS stacked spectrum of quiescent ETGs in GOODS-S at a resolution R=50 (solid green), scaled to the observed magnitude (obtained from the RIz observations) at λrest=250,315 nm of the whole GMASS galaxy sample (in blue). The dashed black lines show the expected rest-frame SEDs in the 25 filters for the faintest and the brightest passive galaxy drawn from the GMASS sample (Cimatti et al. 2008) and the UDF sample (Daddi et al. 2005a). These magnitudes were derived from the best fitting templates to the observed SEDs of these galaxies (see Fig 3.; Pérez-González et al. 2005, 2008a). The associated error bars were estimated assuming SNR~3 for the proposed depth of the observations (I=26.5 mag). It is clearly demonstrated that our experiment will be able to detect quiescent ETGs at z>1.4 via their Mg absorption. The thick black line shows the SED of typical star forming galaxy with a large reddening A(V)=1.5 mag. Note that it can be easily distinguished from the GMASS spectra because of the lack any feature. The shaded regions depict the windows used in the index definition that we have used to estimate the SNR required to obtain a significant detection of the feature.

Back to top

High quality photo-z's

Figure extracted from Ilbert et al. (2008) showing the high quality photometric redshifts which can be obtained with a medium-band imaging survey in the optical. The number of outliers is extremely low (a few percent) and the average accuracy of the photo-z’s is Δ(z)/(1+z)=0.007 for I<22.5 galaxies. Our survey of GOODS-N will go up to 3 mag deeper then Ilbert et al.’s data. We will count with a better spectral resolution since we will use 25 ~17 nm width filters instead of the 12 20–30 nm width filters used in Ilbert et al (see also Wolf et al. 2004). We expect to obtain high quality photometric redshifts (Δ(z)/(1 + z)<0.02) for all sources in the GOODS-N field at z<1 and I<26 (more than 10,000 galaxies, according to the Subaru data available in the field, Capak et al. 2005). See Benítez et al. (2009) for details on the goodness of photometric redshifts in medium-band imaging surveys.

Back to top

Emission line galaxies

Spectra (black lines) of several types of galaxies (a QSO, a Seyfert 2 galaxy, a starburst, and an elliptical from the atlas of Kinney & Calzetti 1994) degraded to the resolution of our instrumental configuration (R~50; color lines) for z=0.5 and z=1.5. We demonstrate that we will be able to detect bright emission lines, specially for AGN, and discontinuities (such as the 4000 °A break) and absorptions (such as the Mg absorption) for galaxies at different redshifts.

Back to top

Spectro-photometric indices

Example of the detection of the Mg absorption feature with our experiment for a 1 Gyr old stellar population. We have defined an index to detect the absorption at z=1.5 with one band on the expected wavelength of the absorption and 4 bands to measure the continuum at both sides of the feature. The index detects the absorption with high reliability. Other absorptions may mimic the Mg one (e.g., the FeII absorption at 250 nm, but the continuum coverage of the spectral energy distribution allow us to use indices defined with contiguous bands and the accurate photometric redshifts we will obtain to differentiate among different features.

Our observations will provide SEDs with a spectral resolution of R~30 at 500 nm and R~50 at 950 nm. This resolution is not only enough to detect the Mg(UV) absorption feature for z=1-2 galaxies, but also to robustly measure spectral indices such as D(4000) or the Balmer break for z<1.2 galaxies (see Balogh et al. 1999), which provide reliable information about the ages of the stellar populations (Kaufmann et al. 2003).

Back to top