The Formation & Evolution of Galaxies Making up the Cirb : Fir/submm Extragalactic Surveys from Dome C

The science case for FIR/Submm surveys of the extragalactic sky to be carried from Dome C are reviewed. The main questions concerning the formation and evolution of galaxies making up the CIRB are outlined and opportunities to exploit Dome C unique observing conditions through single-dish observations are discussed. 1 FIR/Submm Extragalactic Astronomy The wide spectral region between the optical and radio windows has enjoyed an increasing popularity ever since the IRAS mission (Neugebauer et al. 1984) opened up the 10-100 μm wavelength range to a systematic investigation in 1983. In its short lifetime the satellite discovered a sizeable population of galaxies emitting more than 95% of their total luminosity in the infrared, showed that the starburst phenomenon is ubiquitous, and that starburst galaxies must undergo an extreme evolution with redshift if they are to accounf for the observed Cosmic Infrared Background (CIRB). The first homogeneous all-sky survey ever, IRAS has not only had a strong impact on both Galactic (Beichman 1987) and extragalactic (Soifer et al. 1987), but also demonstrated how image atlases and source catalogs of outstanding quality (Beichman et al. 1988) can have a long-lasting impact. Over the following 25 years, our knowledge of the properties of infrared galaxies and their relationship with optical galaxies has improved by leaps and bounds, helped by wide-ranging technological breakthroughs as much as by high-altitude ground-based observatories, air-borne, baloon-borne, or satellite-borne instrumentatin, gradually overcoming the hardships imposed by atmospheric absorption. In this context, the loosely defined FIR/Submm wavelength range (30-1000 μm) has played a central role in two of the most striking intervened discoveries, namely 1 Department of Astronomy, University of Padova, Vicolo Osservatorio 3, I-35122, Padova, Italia 2 [email protected] c © EDP Sciences 2008 DOI: (will be inserted later) 2 Title : will be set by the publisher Fig. 1. Broad-band (UV through radio) SED of prototype local starburst galaxy M82. the discovery of an isotropic 100-500 μm background of likely cosmological origins with COBE in 1996 and SCUBA 850 μm blank-field surveys implying sustanined star formation activity well above a redshift of 2 in 1998. These two elements, confirmed by several other pieces of evidence, strengthened the importance of the FIR/Submm in disentangling the star formation history of the Universe up to the highest redshifts and in evaluating the relative importance of black hole accretion and star formation on the overall cosmic energy budget. This is because the SEDs of bolometrically-bright high-redshift starbursts are welldescribed by local starburst spectral templates such as M82, whose emission peaks at 30-100 μm (See Figure 1). This, combined with an abundant high-redshift starburst population, causes the CIRB to peak at about 200 μm (See Figure 2), with most of the highest-redshift sources still observable at longer wavelengths due to the favourable K-correction setting in. In the process, a few questions have emerged and largely remained unanswered, e.g. the origin of massive ellipticals observed at least up to z ∼ 2, the mechanisms regulating the co-evolution of black holes and their host galaxies and the role of merging. Over the last decade, through coordinated observing programs of the most popular cosmological fields, the ISO, Spitzer & Akari satellites have contributed to shed new light on these issues, although the limited angular resolution provided by their < 1m telescope largely confined them to wavelengths shorter than 100 μm. It will thus be left to Herschel, the first 4-m class telescope in space whose launch is expected in early 2009, to open up the 70-500 μm range to photometric surveys over large areas and, to a more limited extent, to spectroscopic investigation at roughly the same wavelengths. The SPICA mission, if Give a shorter title using \runningtitle 3 Fig. 2. Broad-band cosmic background spectrum, superposed on M82 SED. The FIR/Submm peak, enclosing close to 50% of the overall energy, lies at about 200 μm. From Lagache et al. 2005. approved, will then leverage the advantage of an Herschel-size cold telescope by carrying out deep spectroscopy over the 30-200 mum range, while JWST will do so over the complementary 5-30 μm range. A number of ground-based Submm dishes will complement these efforts with longer-wavelength observations, building on pioneering efforts by CSO, SCUBA &MAMBO. ALMA will finally provide outstanding interferometric capabilities throughout the 0.3-10mm range, althoughits mapping capabilities will be somewhat limited and will therefore mainly be used as a spectroscopic follow-up instrument. One must then ask, guided by state-of-theart models reproducing the statistics provided by previous observing efforts (See e.g. Figure 3), whether a Dome C FIR/Submm observatory may be a competitive player in this crowded field. In so doing, we refer to the proposed design for an ALMA-like winterized 12m antenna (Olmi et al. 2007, Olmi et al. 2008) equipped with a 10,000-pixel filled bolometer array operating at 200/350/450 μm and characterized by the parameters reported in Table 1. These assumptions represent realistic estimates of NEFD ranges to be expected at Dome C and will be used by the ARENA working group established to carry out a detailed assessment of FIR/Submm science cases throughout 2008. This instrumental configuration will be hereafter referred to as the Antarctic Submm Observatory (ASO) concept. 4 Title : will be set by the publisher Fig. 3. 70 & 160 μm differential counts from several Spitzer observing programs, superposed on models for long-wavelength extragalactic populations by Franceschini et al. 2008. Coloured lines are the individual populations while the black line is the total. The upturn in the counts is interpreted as a strong evolution with redshift in the number and the luminosity of the populations. From Vaccari et al. 2008 Table 1. Instrumental parameters for an ALMA-like winterized 12m antenna equipped with a 10,000-pixel filled bolometer array operating at 200/300/450 μm, or the ASO concept. The NEFD ranges roughly correspond to the expected 25% and 75% PWV percentiles and on realistic assumptions for telescope aperture efficiency and optical transmission. and bolometer absorption. Courtesy of Vincet Minier. λ [μm NEFD [mJy/beam] Beam (λ/D) [arcsec] FOV [arcmin] 200 μm 500–1000 3.4 2.8× 2.8 350 μm 100–200 6.0 5.0× 5.0 450 μm 100–200 7.8 6.5× 6.5 2 Atmospheric Transparency at Dome C Interest in Dome C as a FIR/Submm observing site owes much to its atmospheric transparency at these wavelengths. While we still lack a consistent collation of FIR/Submm site testing data, Dome C seems to provide a clear advantage throughout the 200, 350 and 450 μm atmospheric windows, although concerns about both the stability and the low overall levels of atmospheric transmission still stand (Minier et al. 2007, Minier et al. 2008). The 200 μm window appears particularly interesting over the long term, since this is where Herschel & SPICA sensitivity will be hampered by confusion due to their < 4m telescope size, and where the advantage over Chilean Submm sites seems more compelling. Give a shorter title using \runningtitle 5 3 Extragalactic Confusion Limits When instrumental and natural backgrounds are kept down to manageable levels, the sensitivity of FIR/Submm observations soon becomes limited by the sky background caused both by structure in the solar system and in the Galaxy and by an abundant population of point-like extragalactic sources. While several components to this ”noise” must in general be considered, for deep cosmological surveys of high-latitude fields the contribution from extragalactic sources, also knows as extragalactic confusion, is by far the dominant one. The flux levels at which confusion sets in can be estimated on the basis of models for the properties of long-wavelength extragalactic populations such as those by Franceschini et al. 2008 used in Figure 3 using two complementary approaches. The first approach (hereafter labelled Fluctuations and discussed by Franceschini et al. 1989 follows from an analysis of the cell-to-cell fluctuations due to randomly distributed unresolved sources. This is carried out by modelling the telescope deflection probability distribution using a Gaussian beam profile with a given FWHM and then computing the rms signal (σ) due to sources below the confusion limit. The implicit relation between σ and the FWHM is solved iteratively by fixing ? and deriving the FWHM. In our case, confusion is assumed to set in at the 4 σ level. The second approach (hereafter labelled Counts, and discussed by Franceschini et al. 2001 follows from the maximum number of resolved sources to be found within each beam, and sets confusion at the 30 beams / source level. Roughly speaking, while the former approach follows the trend of source counts fainter than the confusion limit, the latter follows from source counts brighter than that. The degree of consistency between the two approaches thus depends on the slope of source counts below the confusion limit, while providing a measure of the robustness of the derived estimates. In the present FIR/Submm case, given the rather steep slope of the counts at the wavelengths and flux levels under consideration, we adopt the ”Fluctuations” approach, which provides a more accurate (and conservative) estimate of confusion under the circumstances. Figure 4 provides a comparison of expected confusion limits for Herschel (3.6m) and ASO (12m). The intersection of coloured curves describing fluctuation levels with coloured lines corresponding to beam sizes at various wavelengths provides an estimate of confusion limits. The limits for the 250/350/500mum Herschel bands consistently fall at fluxes of 30 mJy while ASO would be able to reach down to 5 mJy at 350 & 450 μm and down to 1 mJy at 200 μm. The above confusion estimates set fundamental limits on the detectability of fainter sources (although it is estimated that multi-wavelength source extraction techniques might push down this limit by a factor of∼ 2). It is therefore interesting to estimate which percentage of the integrated CIRB would be resolved into sources at each wavelength should we carry out confusion-limited surveys with ASO and other facilities. Figure 5 details such a comparison, showing how ASO would be able to resolve virtually all of the 200 μm CIRB and about half of the 350 & 450 μm CIRB into discrete sources, while the fraction resolved by Herschel (but 6 Title : will be set by the publisher Fig. 4. Expected confusion limits for Herschel (left) and ASO (right) channels on the basis of the models by Franceschini et al. 2008. Similar colours are used for similar wavelengths. See text for details. Fig. 5. CIRB Resolution from Ground & Space. Percentage of CIRB resolved by Spitzer, Herschel & ASO as a function of wavelength. See text for details. also by SPICA, which if approved would share Herschel 3.6m mirror size) rapidly decreases with wavelength as we move above 100 μm. Give a shorter title using \runningtitle 7 4 Practical Sensitivities As seen previously, Dome C therefore offers an attractive opportunity for deep observations at Submm wavelengths in atmospheric windows which very rarely open up at even the best astronomical observing sites. In order to fully exploit this opportunity for cosmological surveys, however, the sensitivity of the facility to be put in place at Dome C must be such that it allows to actually achieve the above confusion limits over practical observing times. Unfortunately, when compared to space instrumentation such as Herschel, ASO will suffer from a much higher natural background emission which will roughly compensate the advantage offered by its larger aperture than Herschel’s. Using parameters from Table 1, and namely the average NEFD values, the time required to cover 50 arcmin (or one third of each of the GOODS fields) down to 1 mJy/beam rms would be of order 1000 hr at 200 μm but only of order 10 hr for 350 & 450 μm. Given that the 200 μm window will only open some of the time, the viability of the ASO concept for fully-fledged cosmological surveys at 200 μm appears to be questionable at the very least. The mapping speed, i.e. the time required to cover a given area down to a given flux, scales as D. Even more dramatically, the time required to reach a given flux limit scales as D, although the sharper PSF provided by a larger dish means than an array detector with a given number of elements will Nyquist-sample a smaller field of view, hence the more limited quadratic effect on the mapping speed. For this reason, while 350 and 450 μm surveys could still be routinely and effectively carried out, the cosmological impact of a 200 μm channel operating at a 12m antenna will be somewhat limited by its reduced sensitivity, and a substantially larger dish, e.g. the 25m currently envisaged by the CCAT project, would be required to suitably fulfill the promise of the 200 μm Dome C window. 5 Simulated FIR/Submm Surveys from Dome C Give the limitations to its sensitivity outlined above, it is probably helpful to currently think of the ASO concept as a 350 & 450 μm surveyor with 200 μm follow-up capabilities. According to Table 1, mapping 1 deg down to 1 mJy (i.e. a 5σ limit equal to the confusion limit of 5 mJy) would take ASO about 900 & 500 hr at 350 & 450 micron, respectively. At a speculative 18 observing hr per day, such a massive 1,400 hr program could be accomplished within a mere three months if, as suggested by Figure 3 of Minier et al. 2007, Dome C atmospheric transmission at 450 μm remains above 50% for 90% of the time. By itself, this programme would allow to image all ultra-deep fields targeted by Herschel Key Programmes with a much-improved angular resolution and down to a flux fainter by a factor of 2.5 or more. Figure 6 shows how this would allow not only to greatly reduce confusion and therefore multi-wavelength identification problems, but also reach down to below 102 solar luminosities and up to redshifts above 2.5 to greatly expand the parameter space available for exploration. The 450 μm channel would then be even more efficient at detecting high-redshift galaxies. 8 Title : will be set by the publisher Fig. 6. Simulated 350 μm flux-limited surveys over a 1 deg area. Three different flux ranges are chosen, roughly representing Herschel confusion limit (25 mJy), Herschel ultradeep survey limit and ASO confusion limit (5 mJy). For clarity, only one in two sources is plotted at 15 mJy and one in five at 5 mJy.