The Carnegie-Spitzer-IMACS Survey

1 Introduction 

The Carnegie-Spitzer-IMACS Survey (CSI) has been designed to address one of the most dramatic and least understood features of galaxy evolution — the remarkably rapid decline in cosmic star formation since z~1.5. It is during this extended epoch of galaxy maturation that galaxy groups and clusters have also emerged as a conspicuous feature of the landscape. The CSI Survey is uniquely able to link together the evolution of individual galaxies with these features of large-scale structure growth.  

In our ambitious spectrophotometric redshift survey of distant galaxies, we strike a balance between the aforementioned three factors: (1) high completeness to moderate redshift (z~1.5), (2) spectral resolution intermediate between conventional photometric and spectroscopic surveys (combining the efficiency of imaging surveys with a spectral resolution high enough to resolve large-scale structure and prominent emission-lines); and (3) an unprecedented area of 15 sq. degs for a z~1 survey. This gives the CSI Survey a volume comparable to the SDSS, with a selection method that efficiently traces stellar mass over 2/3 the age of the universe (0.4<z<1.5) — spanning the critical redshift range where cosmic star formation precipitously drops, and groups and clusters become prominent. As the redshift survey with the largest unbiased volume at z=1, CSI will allow us to comprehensively address the interplay between environment, galaxy mass buildup, and star formation at these redshifts. 

Figure 1: (top left) Apparent magnitudes in r, i, and 3.6μm as a function of redshift, for passively evolving old stellar populations with stellar masses 5M* and M*/2 at z<1. The magnitude limits for DEEP2 and PRIMUS are drawn. Such optical flux limits only cover the most massive passively evolving systems or those galaxies with young unattenuated stellar populations, biasing galaxy samples at early times. The shallow dependence of the 3.6icron magnitude on redshift yields a selection with significantly less bias against old systems. (bottom left) Limiting stellar mass of faint galaxy surveys by redshift. By using the IRAC 3.6μm band, the CSI survey (the solid orange line) traces stellar mass more uniformly than samples selected in the optical. Our sample reaches stellar masses equivalent to the present day Milky Way out to z=1.4, almost order-of-magnitude lower than DEEP2, and half the present day Milky Way at z=0.9, a factor of two deeper than DEEP2 or PRIMUS. (bottom right) Volumes probed with complete, unbiased samples for several redshift surveys as functions of limiting stellar mass. When the 15 deg is completed, CSI’s volume will be more than an order of magnitude larger than DEEP2. The volume traced by the first 5.3 degs (see Kelson et al. 2012, submitted) is shown by the dotted line.  

2 The limitations of optically-selected surveys 

Any effective probe of galaxy assembly must sample a wide range of masses in order to trace evolutionary connections between large and small systems. The build-up of massive red sequence galaxies may be driven by mergers with sub- systems, so it is essential to trace evolution to masses below to mitigate against the differential growth of the high- and low-mass populations. This trade-off between depth and area noted above has led to a dichotomy in redshift surveys. Very deep programs, such as the Gemini Deep Survey or the Galaxy Mass Assembly Ultra-deep Spectroscopic Survey could not cover enough volume to robustly sample the evolution of the high mass population at z<1, while other surveys that have traded depth for area do not reach below with high fidelity.  

An old stellar population at z=1 with a stellar mass of corresponds to roughly i=23 mag and r=24 mag in the optical, as shown in Figure 1(top left). The extreme optical faintness of such galaxies results in many of them being missed in optically-selected surveys despite their relatively high masses. For example, the selection limits for DEEP2 and PRIMUS are shown in this figure by the blue and green dashed lines, respectively. At z>1 these programs required galaxies to be extremely massive or have their optical light dominated by young stellar populations in order to be fall within the survey selection. At best it can be difficult to trace the evolution of the most massive galaxies with such surveys. At worst, if not properly accounted for, color-dependent selection effects introduce biases and systematic effects. Extremely deep optical limits can be taken as one valid approach to ameliorating this problem, with the side effect of an overwhelming number of low-mass star-forming dwarfs dominating one’s source catalog. 

3 The potential of an IRAC-selected survey 

The CSI Survey exploits the wide field and sensitivity of the Spitzer Space Telescope with the Infrared Array Camera to take advantage of this property of the integrated near-IR light from stellar populations in z~1 galaxies. Combined with the insensitivity to internal and Galactic extinction, selection at 3.6μm closely mirrors selection by stellar mass. Figure 1 (top left) directly compares the evolution of 3.6μ with the optical r and i bands: the CSI selection wavelength has a dependence on redshift that is much shallower than surveys selected in the optical.  

Even so, massive galaxies exhibit a much flatter trend of observed magnitudes with redshift in the IR than in the optical, and this weaker dependence on galaxy mass afforded by 3.6μ-selection of the sample is a key feature of the CSI Survey. Our goal has been to make a spectrophotometric survey to characterize the galaxy populations and environments to z=1.5, unbiased at all redshifts down to a stellar mass of 4x10¹⁰ M_☉. As can be read from Figure 1, this mass limit corresponds at z=1 to r=25 mag or i=24 mag, with equivalent limits of r=26 mag and i=25 mag to reach this mass limit at z=1.2. Our current spectroscopic reduction and analysis, described below, is reaching an effective photometric limit of r=26 mag (the dashed orange line). 

Figure 1 (bottom left) plots the limiting mass as a function of redshift for CSI (solid orange) and other redshift surveys. The depth of CSI in stellar mass is substantially less sensitive to redshift compared to the others due to the IRAC 3.6μm selection, varying by a factor ~3 over z=0.5-1.5, compared to 1-2 dex over the same redshift range for optically-selected surveys. The result is that CSI samples nearly uniformly by stellar mass for the full, large volume of the survey, and to a lookback time of 9 Gyr. 

This critical point is illustrated in Figure 1 (bottom right), plotting the depth in stellar mass against the volume probed with complete, unbiased samples. DEEP2, shown in violet (the line colors are the same as in the bottom left), is limited in both area and mass depth. PRIMUS’s 9 sq. degs is limited in depth, and thus only probes an unbiased volume comparable to our first 5 sq. degs (see Kelson et al. 2012, submitted). When CSI reaches its goal of 15 degs, the survey will cover an unbiased volume equal to the SDSS with similar depth in stellar mass. The large areas available from legacy Spitzer IRAC surveys — both wide and deep, and the freedom from foreground and internal extinction at this wavelength, allows the construction of uniformly deep, homogeneous photometric samples for spectroscopic follow-up. 

CSI reaches factors of 2-6 deeper than DEEP2 in mass, over an area ultimately 8 times wider, and redshifts sufficiently accurate to characterize environments by directly identifying groups and clusters. With such data we aim to make the first group catalog at z=1 that is statistically comparable to SDSS, and thus enable the first detailed environmental characterizations of galaxies at a time when the universe was less than half its current age.