Observations of galaxies that formed early in the Universe's history reveal much lower dust levels than are found in sources from a slightly later era. It seems that galaxies underwent rapid change during a relatively short period. See Letter p.455
The study of the most distant galaxies, observed as they were about 1 billion years after the Big Bang, is crucial for our understanding of the star-forming activity and the physical processes at work in these young systems. Writing in this issue (page 455), Capak et al.1 present a study of nine such galaxies using a linked-up telescope array. They find that the dust and gas properties in these systems hint at an interstellar medium (ISM) that is much less evolved than in galaxies about 2 billion years older. This suggests that there was a rapid change in the overall properties of galaxies during the early life of the Universe.
The expansion of the Universe shifts the ultraviolet (UV) light emitted by newly formed stars in remote systems to longer (visible and near-infrared) wavelengths that, unlike UV light, can be observed by ground-based telescopes. The most distant objects known today are detected as a result of a break in the continuum of their redshifted spectra at wavelengths of around 0.1 micrometres; this is due to the absorption of UV photons by neutral hydrogen in the intergalactic medium. The absorption occurs for photons with energies corresponding to wavelengths shorter than the Lyman-α line of hydrogen (1,216 nm), and galaxies whose distances have been estimated by this method are known as Lyman break galaxies (LBGs). The most comprehensive surveys undertaken so far have led to detections of very young LBGs that formed approximately 0.5 billion years after the Big Bang2.
The presence of interstellar dust complicates the study of galaxies, and affects measurements of fundamental, observationally derived properties such as the star-formation rate. This is because dust is efficient at absorbing the energetic UV photons (a proxy for the star-formation rate) that are emitted by young stars and at re-emitting their energy in the infrared domain, at wavelengths longer than 5 μm. This is a complex process that depends not only on the amount of dust present, but also on its distribution relative to the stars and on its composition3. Overall, dust substantially reduces the intensity of stellar light reaching the telescopes4.
A straightforward method to account for the UV light produced in galaxies involves observing the radiant energy that is absorbed by dust, is re-emitted and is then redshifted in the far-infrared and submillimetre domains. However, the low sensitivity of detectors, combined with the poor spatial resolution achieved by single-dish telescopes, make surveys of high-redshift galaxies at these wavelengths less efficient than those at optical or near-infrared wavelengths. Even the Herschel Space Observatory, which detected5 the infrared emission from dust in galaxies at redshifts of up to 2–3 (corresponding to a time roughly 2 billion to 3 billion years after the Big Bang), was able to detect only hyper-luminous sources at much larger distances6.
Given that directly measuring the long-wavelength emission from dust is so challenging, astronomers resort to empirical relations to derive dust's infrared luminosity. One such relation links this luminosity to the stellar UV luminosity7 for a representative sample of nearby, actively star-forming galaxies, for which both luminosities have been accurately measured. Unfortunately, this recipe is not universally applicable because it depends on the properties of the ISM (such as the composition of dust and its distribution relative to the stars), as well as on the stellar populations in the galaxies8. Despite these caveats, however, it is extensively used to estimate the level of obscuration of the stellar UV light by ISM dust for galaxies across a wide redshift range. It will therefore be important to check the validity of this relationship — especially for young, high-redshift systems. A critical evaluation could also yield clues to the properties of the ISM at those early times.
Capak et al. used the Atacama Large Millimetre Array (ALMA), which was designed to overcome both the resolution and sensitivity problems (Fig. 1). Being an interferometer (a series of telescopes linked up to combine astronomical observations), ALMA has a small field of view that is suitable for observing well-centred sources, and it can detect the weak submillimetre emission originating from dust in ordinary galaxies at high redshifts. The authors used 20 of ALMA's antennas in unison to observe the dust and gas emissions of 9 typical LBGs located at redshifts 5–6; these correspond to a time when the Universe was about 1 billion years old.
Capak and colleagues selected their sample from the Cosmic Evolution Survey field, a two-square-degree area that has been extensively observed by most of the major telescopes, from the ground and from space. ALMA detected the thermal dust emission in four galaxies, and an ISM spectral line emitted from gaseous carbon at a wavelength of 158 μm in all nine of them; the carbon feature is the dominant ISM emission line of galaxies in the far-infrared domain. Such a high detection rate is outstanding, because previous attempts failed to simultaneously detect the carbon feature and thermal dust emission9.
The authors' study argues for a very low dust content and stellar-light obscuration in these systems. The four galaxies whose thermal dust emission was detected may harbour an ISM similar to that of the Small Magellanic Cloud (a satellite galaxy of the Milky Way), which is characterized by a low abundance of elements heavier than helium. The upper limits put on the dust emission of the remaining five sources call for an even more extreme situation with a much lower infrared emission. That seems to be at odds with the observed UV luminosity of these systems. The enhanced carbon emission-line intensities also suggest low dust levels relative to the gas present in these early galaxies, although other explanations cannot be excluded.
An immediate consequence of these findings is that the classical calculations used to derive obscuration due to dust from the observed UV continuum luminosity are unlikely to be valid for LBGs in the early Universe. As a result, the star-formation rate considered to be appropriate for these galaxy types is likely to have been overestimated by factors of between two and four in previous studies. Last but not least, this pioneering work paves the way for future observational campaigns. Although observing the low levels of dust emission from large samples of high-redshift galaxies may prove challenging even for ALMA, Capak and co-workers' finding of enhanced carbon emission lines should become a useful tool in the study of star-forming galaxies at those early epochs.