Infrared Glow from Early Cosmic Times
The last decade has seen rapid progress in a major area of cosmology ? the development and the origin of cosmic structures in our 14 billion year old Universe. The earliest epoch one can probe directly with photons corresponds to when the bulk of the photons of the cosmic microwave background radiation last scattered on the (still partially) ionized plasma. This corresponds to the time the Universe was only 400,000 year old. The imprint of the structure of that surface on the cosmic microwave background has now been accurately measured by the NASA?s WMAP satellite  on angular scales exceeding a few arcminutes. In the opposite direction, going from the present epochs, powerful ground- and space-based telescopes now (almost) routinely detect objects out to the first billion year of the Universe?s evolution (see review in ref ). These galaxies and their structures have grown out of the tiny (about a few parts in 100,000) fluctuations that were already present at the last scattering epoch.
As one looks back in time, one observes stellar populations that have already been enriched with metals, such as the nearby stellar Populations I and II (our Sun is a Population I star). This means that these stellar populations were preceded by the first stars, the metal-free Population III. These objects have never been found, but were a subject of lively theoretical debates for some 20+ years. Theoretical discussions of what these first objects ought to have been ranged from them being mostly low-mass unevolved stars (so-called “Jupiters”) through very massive stars to even black holes of various masses. It was clear, however, that they cannot be like the ordinary stars in the mass range between 0.1 to 10 solar masses for then they would still be around today contrary to observations. With the establishment over the last decade of the now standard “concordance” cosmological model (dark-energy dominated cold-dark-matter model) and advances in numerical computations of the physics of collapse of the first objects there, a broad consensus has emerged that the first luminous sources were very massive stars that started forming in any significant quantities when the Universe was ~ 100 million years old and continued to form in very small first mini-galaxies (perhaps as small a few million solar masses) for the next several hundred million years (see review in ref ).
Such small objects so far away are not accessible to any of the currently operating telescopes. However, there may be another way to find them: being so bright intrinsically, they could produce a significant background radiation which could be detectable today through its imprint in the cosmic infrared background (CIB) . (The reason for the infrared part of the background is that emissions are absorbed by the intergalactic medium below the Lyman limit, which corresponds to the capture of photons by the lowest energy level of the hydrogen atom and is shifted by cosmic expansion to the near-infrared wavelengths greater than 1 micrometer if the objects formed at such early times). Indeed, there were numerous claims of excess in the measured levels of the CIB over that from “ordinary” galaxies [5,6,7] (see review in ref ), but the problem in such measurements is that it is very difficult to remove the contributions and contributors from the various intervening sources to reliably isolate the truly early component. On the other hand, it was pointed out independently in [9,10] that the background radiation produced by the first luminous objects should have a distinct and measurable structure. There are simple intuitive reasons why the first luminous objects should have produced significant CIB anisotropies: 1) if massive, each unit mass of the Population III stars should have produced significantly more luminosity than the present day stellar populations; 2) the cosmic time spanned by the epochs corresponding to these stars is short (a few hundred million years at most), so the relative fluctuations in their CIB integrated over such a “narrow” slab would be larger; and 3) within the concordance cosmological model, these objects had to form at high peaks of the density fluctuations and their clustering properties would then be strongly amplified.
Imagine you want to see very distant, and therefore faint, fireworks from an illuminated city. In order to do this, you have to remove city lights one by one until what is left in the city produces less illumination than the fireworks on the horizon. Once you have removed enough of the foreground lights, you will see the overall glow from the fireworks. But even in that case, you are limited by the resolving power of your binoculars, or telescope. In practice, even after you have removed enough foreground lights, what you will see would be only the large scale contours of the fireworks, not every individual speckle. Still, the overall intensity of the glow and its angular structure can tell you a lot about the luminosities of these fireworks, their distribution and even – after some additional interpretation – their epochs (or the distance they are at).
A couple of years ago we (myself, Rick Arendt, John Mather and Harvey Moseley – all at the Goddard Space Flight Center) began a program to detect CIB fluctuations from early populations using deep data from NASA’s Spitzer Space Telescope. The data was not obtained by us; it was obtained in the course of other programs. We used the part of the data these programs did not need or want, so the project turned out to be very cost-effective!
The original dataset presented by us in 2005 (ref. ) came from a field of about 5 by 10 arcminutes with ~8 hours per pixel of integration during the in-orbit-calibration (IOC) of the Telescope and covering four channels from 3.5 to 8 micrometers. We used the individual exposures and assembled them into the final images correcting for the detector variability with time (basically using detector parameters at the time of the observation). In the assembled image we removed the intervening galaxies down to fairly low fluxes, but in a way that allowed us to robustly compute the clustering properties of the remaining diffuse light. All the steps in the measurement were verified with detailed numerical simulations – a necessary step in any such analysis. We then measured the contributions from the zodiacal light produced by dust in the Solar System (it is time-varying as the Earth moves in its orbit and we used some other data observed at times separated by ~ six months) and Galactic dust. Both of these were shown to be small except for a possible pollution of the measured signal by Galactic dust in the 8 micrometer channel. What we were left with was the signal that was significantly larger than that from the remaining galaxies. It also had to originate in sources fainter (statistically more distant) than the remaining ordinary galaxies, suggesting its likely origin in the first stars (see independent overview of these results in ref. ).
But that was only one field and to definitively demonstrate the cosmological origin of the measured signal, we had to show that it is isotropic on the sky. As the Spitzer telescope continued its operations, more and deeper datasets became available. One such dataset was obtained in the course of one of its observing programs and it covered four additional directions on the sky each of which was also integrated for a significantly longer time (~25 hours) than the IOC field. The analysis [13,14] of the new dataset confirmed our original measurements, but also made significant advances and the new data allowed a more definitive interpretation in terms of the nature of the populations producing these fluctuations (an independent overview of these results in given in ref ). The additional directions have confirmed that the signal is the same in all five directions and thus is almost certainly cosmological. The longer integration allowed us to eliminate the intervening galaxies down to still fainter limits than in the IOC field and to better isolate the remaining contribution on the intermediate scales. Finally, each of the new fields has been observed at two epochs ~ six months apart and this allowed us to further rule out any significant zodiacal light contribution to the signal.
What does the measured signal tell us? What we measured is the angular spectrum of CIB fluctuations produced by populations below the flux threshold we reached for the data. The signal has two parts: 1) small scales are dominated by the shot-noise (white noise like what you see on an empty screen of an old TV set) produced by the remaining sources occasionally entering the beam, and 2) the large scale part is produced by clustering of the emitters and reflects the primordial density field out of which these emitters formed. The measurement suggests that the signal is produced by populations that have a fairly prominent clustering component, but only at most a small level of the shot noise. The clustering component is of most interest cosmologically as it reflects the light emitted by the distant sources. The net flux associated with these emissions can then be estimated assuming the density field of the concordance cosmological model. It turns out to be significant, although is probably below the levels claimed in the various earlier CIB excess studies. But it is the shot noise component that now enabled us to pin the nature of these sources more firmly: the low levels of the shot noise suggest that the signal is produced by the populations which are individually very faint and are thus likely located within the first billion years of the Universe’s evolution. This is where one expects to find the first stars. The spectral energy distribution of the fluctuations is flat to slowly rising with increasing wavelength, which again is a signature of the emissions from the first luminous objects. Furthermore, these sources had to radiate at much higher luminosities per unit mass than the present-day stars in order produce the required CIB in the cosmic time available.
What we cannot say, however, is whether these emissions were produced by stars or accreting black holes. These two kinds of astrophysical objects release their energies via completely different mechanisms: in stars the luminosity is produced by nuclear fusion, while in accreting black holes it is produced by viscous dissipation processes in the accretion disk around the central black hole. Data such as what we used in these studies cannot really discriminate between the two mechanisms, so we cannot determine which type of objects were responsible for the emissions at these early times. To make such a determination one would need high resolution spectral measurements, which are not feasible with the present day instruments.
Additional information, images and original articles can be obtained from http://www.kashlinsky.info
1. Bennett, C.L. et al 2003, Ap.J.Suppl., 148,1
2. Ellis, R. 2007, in `First Light in Universe', Saas-Fee Advanced Course 36, Swiss Soc. Astrophys. Astron., in press. (astro-ph/0701024)
3. Bromm, V. & Larson, R. 2004, Ann. Rev. Astron. Astrophys., 42, 79
4. Santos, M., Bromm, V. & Kamionkowski, M. 2002, Mon. Not. R. Astr. Soc., 336, 1082
5. Dwek, E. & Arendt, R. 1998, Ap.J., 508, L9
6. Kashlinsky, A. & Odenwald, S. 2000, Ap.J., 528, 74
7. Matsumoto, T. et al. 2005, Ap.J., 626, 31
8. Kashlinsky, A. 2005, Phys. Rep., 409, 361
9. Kashlinsky, A., Arendt, R., Gardner, J.P., Mather, J.C. & Moseley, S.H. 2004, Ap.J., 608, 1
10. Cooray, A, Bock, J., Keating, B., Lange, A. & Matsumoto, T. 2004, Ap.J.,606, 611
11. Kashlinsky, A., Arendt, R., Mather, J.C. & Moseley, S.H. 2005, Nature, 438, 45
12. Ellis, R. S. 2005, Nature, 438, 39 (News & Views)
13. Kashlinsky, A., Arendt, R., Mather, J.C. & Moseley, S.H. 2007, Ap.J., 654, L1
14. Kashlinsky, A., Arendt, R., Mather, J.C. & Moseley, S.H. 2007, Ap.J., 654, L5
15. Hogan, C.J. 2007, Nature, 445, 37 (News & Views)