High-redshift star formation in the Hubble Deep Field revealed by a submillimetre-wavelength survey

Abstract

In the local Universe, most galaxies are dominated by stars, with less than ten per cent of their visible mass in the form of gas. Determining when most of these stars formed is one of the central issues of observational cosmology. Optical and ultraviolet observations of high-redshift galaxies (particularly those in the Hubble Deep Field) have been interpreted as indicating that the peak of star formation occurred between redshifts of 1 and 1.5. But it is known that star formation takes place in dense clouds, and is often hidden at optical wavelengths because of extinction by dust in the clouds. Here we report a deep submillimetre-wavelength survey of the Hubble Deep Field; these wavelengths trace directly the emission from dust that has been warmed by massive star-formation activity. The combined radiation of the five most significant detections accounts for 30-50 per cent of the previously unresolved background emission in this area. Four of these sources appear to be galaxies in the redshift range 2 < z < 4, which, assuming these objects have properties comparable to local dust-enshrouded starburst galaxies, implies a star-formation rate during that period about a factor of five higher than that inferred from the optical and ultraviolet observations. Recent years have seen the first meaningful attempts to determine the global star-formation history of the Universe, using the combined information provided by deep redshift surveys (for example, the Canada France Redshift Survey 1 ) reaching z Ϸ 1, and the statistics of Lyman-limit galaxies 2 at higher redshifts in, for example, the Hubble Deep Field (HDF) 3-5 . The results 6 imply that the starformation and metal-production rates were about 10 times greater at z Ϸ 1 than in the local Universe, that they peaked at a redshift in the range z Ϸ 1-1:5, and that they declined to values comparable to those observed at the present day at z Ϸ 4.These conclusions, derived from optical-ultraviolet data, may however be misleading, because the absorbing effects of dust within distant galaxies undergoing massive star-formation may have distorted our picture of the evolution of the high-redshift Universe in two ways. First, the star-formation rate (SFR) in known highredshift objects is inevitably underestimated unless some correction for dust obscuration is included in deriving the rest-frame ultraviolet luminosity. Second, it is possible that an entire population of heavily dust-enshrouded high-redshift objects, as expected in some models of elliptical galaxy formation 7 , have gone undetected in the optical-ultraviolet surveys. The extent of the former remains controversial 8-11 , while the possibility of the latter has until now been impossible to investigate. Submillimetre cosmologyAt high redshifts (z Ͼ 1), the strongly-peaked far-infrared radiation emitted by star-formation regions in distant galaxies is redshifted into the submillimetre waveband, and the steep spectral index of this emission on the long-wavelength side of the peak, at l Ϸ 100 m in the rest-frame, result...

Dusty star-forming galaxies at high redshift

“…The first few submillimeter deep-field maps which were published (e.g. Smail et al, 1997;Barger et al, 1998;Hughes et al, 1998) detected several galaxies at 850µm within several square arcminutes. A Figure 5: A schematic spectral energy distribution for a dusty star forming galaxy at z = 2.…”

“…HDF850.1 HDF850.1 dates back to the original SCUBA survey of the HDF-N by Hughes et al (1998). The multi-wavelength counterpart identification has been challenging, despite many attempts (Downes et al, 1999;Dunlop et al, 2004;Wagg et al, 2007;Cowie et al, 2009), highlighting the challenge at identifying and studying properties of extreme DSFGs at very high redshifts.…”

“…Smail et al, 1997;Hughes et al, 1998;Barger et al, 1998), although their properties were hard to study due to the large beamsize of Scuba's observations (∼15 at 850µm; at cosmological redshifts, roughly 120 kpc) and the difficulty in identifying multi-wavelength counterparts. However, at the time it was already thought that these submillimeter sources would be predominantly located at high-redshift.…”

“…P(D) method allows a constraint on the faint-end counts below confusion and has been used widely in sub-mm maps since the SCUBA surveys (Hughes et al, 1998). Patanchon et al (2009) expanded the technique with a parameterized functional form for faint-end counts with knots and slopes.…”

The intense starburst HDF 850.1 in a galaxy overdensity at z ≈ 5.2 in the Hubble Deep Field

“…From the spatial offset (total offset=0.9", i.e. radius: 0.45", or r=2.8 kpc) and the FWHM of the line, we derive an approximate dynamical mass of M dyn ~ 3.4 x 10 10 M sun / (sin i) 2 where i is the (unknown) inclination of the system (using M dyn sin 2 i = 1.3 x (FWHM/2) 2 r/G, where G is the gravitational constant 30 ). These deep HST images of the HDF fail to reveal the (rest-frame) UV/optical counterpart of the galaxy that is forming stars at a rate of ~850 M sun yr -1 .…”