Balloon-borne Large-Aperture Submillimeter Telescope

 Home  Results  Flights  Science  Instrument  Collaborators  Links  Press  Contact  BLAST The Movie
 Overview  Extragalactic Science  Galactic Science  Polarization Science

Cosmic History of Galaxy Formation


The light emitted from massive star forming galaxies in the very early Universe results in a bright, nearly uniform cosmic background, seen at both optical and far-infrared wavelengths. These optical and infrared backgrounds (COB and CIB, respectively) are nearly the same brightness, and this suggests that a significant fraction of the UV and optical photons emitted by the brightest stars are absorbed by the surrounding clouds of dust from which the stars formed. The dust grains are heated to temperatures in the range 20–50 Kelvin, then re-radiate this thermal energy at longer wavelengths. In order to understand the cosmic history of star formation in more detail, a great deal of observational effort has been invested to spatially resolve the extragalactic background into the different populations of quiescent galaxies, starburst galaxies and active galactic nuclei (AGN).

Since the late 1990s, the first ground-based submillimeter and millimeter wavelength surveys (with SCUBA on the 15-m JCMT, and MAMBO on the 30-m IRAM, respectively) have detected a population of luminous (LFIR > 3x1012 LSUN) high-redshift (z~2.5 ± 0.6) starburst galaxies that produce 10–50% of the measured diffuse extragalactic background at wavelengths ~1mm. In contrast, the Spitzer Space Telescope has determined that the background at shorter wavelengths (24–160µm) is predominantly produced by less-luminous galaxies closer to z=1. The peak of the CIB occurs at a wavelength of ~200µm, apparently coinciding with a shift from the bulk of star-formation in rare high-luminosity sources at large redshifts to more numerous low-luminosity sources at lower redshifts. This phenomena, generally referred to as cosmic downsizing has now been observed in extra-galactic surveys at a wide range of wavelengths.

In 2006, BLAST conducted the first deep extra-galactic surveys at wavelengths 250–500µm capable of measuring large numbers of star-forming galaxies (~1000), and their contributions to the CIB. Data across this crucial wavelength range are required to understand the details of cosmic downsizing traced by the thermal emission from star-forming galaxies — our new data provide strong parameter-free constraints for the history of galaxy formation.


BLAST conducted two large un-biased extra-galactic surveys, one centered over the ECDF-S/GOODS-S region covering approximately 9 deg2 (Figure 1), and a second near to the south ecliptic pole (SEP) covering approximately 8 deg2. We have analyzed the ECDF-S/GOODS-S map (BLAST GOODS-S, or BGS) first, as it is one of the most well-studied fields in the southern sky, with the deep multi-wavelength data sets required to identify and characterize individual sources. The initial results from BGS are given in Devlin et al. (2009), in which we present the maps (Figure 1), measurements of the source counts (Figure 2), and determine that most of the CIB at BLAST wavelengths is produced by faint sources detected at 24µm with Spitzer.

Figure 1: This is a signal-to-noise ratio map of the total intensity in BGS, including data from all three bands (Devlin et al. 2009) — white peaks correspond to a signficance of greater than ~5-sigma. Roughly half of the time was spent on a shallow wide exposure (BGS-Wide) covering ~9 deg2 that overlaps with the ECDF-S Spitzer SWIRE coverage (grey contour). The other half of the time was spent in the deep inner square degree (BGS-Deep), denoted by the dashed yellow line, and surpasses the extra-galactic point-source confusion limit. Despite the significantly smaller area, BGS-Deep completely encompasses the Chandra/VLA/FIDEL/LABOCA coverage of the ECDF-S (blue contour), the Chandra 2 Ms exposure (green contour), and GOODS-S (purple contour).

Figure 2: Measurements of the Euclidean-normalized BLAST differential source counts compared to data at other wavelengths that bracket BLAST (Devlin et al. 2009).

Owing to the extensive observing time already invested in BGS at other wavelengths, there already exist vast catalogues of galaxies selected in the optical and IR, as well as the radio — many with spectroscopic or optical photometric redshift estimates. We have determined that the CIB is completely produced by sources in the deep 24µm FIDEL catalogue using a stacking analysis (Devlin et al. 2009, Marsden et al. 2009, see Figure 3). We have then proceeded to identify spectroscopic and photometric redshifts for all of the sources in FIDEL to obtain the first measurements of the star-formation rate history of galaxies that produce the bulk of the CIB (Figure 4) without requiring any assumptions about the spectra of individual galaxies.

Figure 3: BLAST measurements of the CIB made by stacking (averaging) the flux density at positions of sources from deep Spitzer 24µm catalogues (Marsden et al. 2009). These measurements are more precise than earlier data from FIRAS on the COBE satellite (Fixsen), and bridge the spectral gap between Spitzer measurements at 160µm (Dole) and SCUBA at 850µm.

Figure 4: The cosmic history of star-formation rate in infrared-luminous galaxies (black circles) to redshift z~2 (Pascale et al. 2009). These are the first such measurements that do not depend on assumptions about the spectra of individual galaxies.

In addition to the stacking analyses, we have also begun to study the properties of individual bright galaxies selected in the BLAST bands (Dye et al. 2009). Figure 5 shows the distribution of luminosity (and star-formation rate) as a function of redshift for this initial small sample of galaxies. Clearly the BLAST surveys detect a combination of lower-luminosity objects at z<1 common in Spitzer surveys, as well as the ultra-luminous galaxies at z>1 discovered in SCUBA surveys. For all of these sources, BLAST is presently unique in its ability to constrain bolometric far-infrared luminosity and temperatures; in the past, the rest-frame spectra of such sources have always been assumed to resemble objects of similar luminosities in the local Universe.

Figure 5: The luminosity (and star-formation rate) of bright BLAST sources as a function redshift (Dye et al. 2009). The lines indicate detection limits based on the depths of the BLAST catalogues in each band and at a range of dust temperatures. Once all of the BLAST data are analyzed, it will be possible to directly measure the evolution of the bolometric far-infrared luminosity function.

By operating at altitudes above most of the varying atmospheric water vapour emission that hinders ground-based observations, and using detectors with extremely stable long-term noise characteristics, BLAST maps faithfully reproduce structure on angular scales significantly larger than the projected array footprint on the sky (~15 arcmin). This has enabled us to perform the first multi-band measurement of the angular clustering in the CIB (Viero et al. 2009). In Figure 6, we show the angular power spectrum of the BGS maps in all three bands, clearly exhibiting a significant excess at scales 10–40 arcmin. In this initial work, we have used a simple halo model to describe the expected clustering signal due to large-scale structure in the Universe, and we are able to measure the bias of the luminous infrared galaxy population.

Figure 6: The angular power spectrum of the BGS maps in each BLAST band (symbols), compared to un-clustered Poisson noise simulations (horizontal dashed lines). BLAST unequivocally detects excess power in these maps at scales 10–40 arcmin which we use to infer the large-scale clustering of infrared-bright galaxies (Viero et al. 2009). We rule out the possibility that this excess is due to the clumpy structure of dust in our own galaxy since even the most pessimistic estimates (dotted line and shaded envelope) show that the contribution to the power spectrum in this field is completely subdominant to the Poisson noise.


Send questions or comments to Mark Devlin