The onset of star and galaxy formation marks the time at which the early Universe emerged from the so-called "Dark Ages" and is referred to as the epoch of re-ionization (EOR). Uncovering the exact time and circumstances that gave rise to this event is one of the most important outstanding issues in cosmology, and has been the subject of much recent theoretical investigation. The LOFAR design will provide the much needed observational evidence that can specify the physical conditions during this important stage in the Universe, when the first stars and compact objects were formed.
After recombination of the elements at a redshift of ~1100, the baryonic matter in the Universe remained neutral until the first stars and galaxies started to form. At least three different types of sources have been suggested as contributing to the radiation that was responsible for reionization. These are: 1) emission from the first generation of stars, 2) radiation released in the collapse of the first galaxy-sized halos, and 3) emission from an early generation of luminous quasars. At the moment it is still unclear which sources or combination of sources or processes dominated. In fact, it is quite likely that actual measurement of the time at which re-ionization occurred will help us determine which sources played the dominant roles.
Theoretical simulations show that the process of re-ionization was probably quite rapid and could have taken place between z ~ 6 and 20 (Gnedin & Ostriker 1997; Tozzi et al. 2000). The present observational constraints are not restrictive. The EOR is not more recent than z ~ 5.8, since there is no distinctive absorption in the spectra of the highest redshift quasars due to the presence of a neutral intergalactic medium (Songaila et al. 1999; Zheng et al. 2000). The EOR could not have been earlier than z ~ 30, or there would be a suppression of the first Doppler peak in the angular fluctuation spectrum of the Cosmic Microwave Background (Tegmark & Zaldarriaga 2000; De Bernardis et al. 2000). Recent measurements of the temperature of the intergalactic medium in the Lyman alpha Forest clouds at redshift three suggest that the EOR must have occured after z ~ 10 or the clouds would have become too cool (Hui, in preparation). One further indirect constraint comes from the apparent behaviour of Dark Matter on small scales that is implied by the absence of the predicted number of intergalactic mini-halos and sub-halos of galaxies that are predicted to exist at z = 0 by simulations (Spergel & Steinhardt 2000; Kamionkowski & Liddle 2000). If these small scale structures are absent at present, the implication is that the localized star forming regions will form more slowly, which means later in the z ~ 6 to 20 window than had earlier been predicted.
The radiation signature that LOFAR may be able to detect was emitted in the period immediately before full re-ionization. The signal is expected to be similar in all directions, i.e. it is a global signal. In the cool, still neutral regions of the Universe, the medium was heated by the ionizing sources (stars, quasars or mini-halos) and the hydrogen spin temperature decoupled from the CMB emission (from the Big Bang). This effect caused a small step in the temperature of the background radiation. The predicted spectroscopic signature is generated at the rest frequency of the neutral hydrogen line (1420 MHz), but redshifted to LOFAR frequencies by the expansion of the Universe. Therefore, the exact frequency at which the temperature step is detected is linked to the time in the past at which it occurred.
Figure 2.2 Simulations of the expected brightness temperature of the cosmic background in the vicinity of the hydrogen reionization edge as a function of observing frequency. Three cases are shown for the HI step, corresponding to different models of how the process took place. All three produce an effect that is capable of being detected by LOFAR (from Shaver et al.1999).
To investigate this transition phase LOFAR will be equipped with dipoles optimized for the 110-250 MHz band. Because the transition is expected to occur globally, i.e. in all directions at about the same time, the LOFAR collecting area at these frequencies in principle need not be very large. The expected signal, about 15-20 mK in brightness temperature, with a spectral width of about 5-10 MHz, does not depend on aperture size.
Calibration of this faint signal, however, will dictate a telescope with a substantial collecting area (cf. Shaver et al., 1999). Collecting area and resolution are also important to deal with a crucial aspect of this experiment; which is to identify, model, and remove foreground sources that contaminate the signal. One such contaminant is the population of discrete radio sources (faint radio galaxies and, especially, starburst galaxies which make up the bulk of the faint source population) . The longer baselines of LOFAR are needed to assist in the identification and spectral characterization of discrete sources in the field(s) of view being observed for the spectral decrement. In both cases - use of the inner portion of LOFAR to search for the spectral decrement and exploiting the high angular resolution of LOFAR to identify foreground contaminants - the broad band nature of LOFAR will be essential. A second contaminant is the diffuse nonthermal Galactic foreground emission which is responsible for the bulk of the radio noise from the sky at LOFAR frequencies. Fortunately, this diffuse emission has very little structure on the 0.5 degree angular scales at which a global signal will be sought. Faint Galactic foreground fluctuations that do exist are expected to be spectrally broad, thus permitting them to be modeled.
As was pointed out above, the epoch of reionization, and therefore the frequency of the spectral decrement, is uncertain. The broad band nature of LOFAR would enable a search for this effect over a finely spaced grid of frequencies, ensuring detection of the EOR transition, provided it falls in the redshift range 5–12.
Prior to full re-ionization the intergalactic medium was most likely a mixture of neutral, partially ionized, and fully ionized structures. It is believed that the low-density regions will be fully ionized first, followed by regions with higher and higher densities. A patchwork of neutral hydrogen emission, and possibly absorption against the cosmic background radiation (about 30 K at these redshifts), will result in structures up to a degree in size. Rather than being a global, all-sky feature, this patchwork of emitting and absorbing structures will appear on smaller angular scales (3'- 30') and in narrow bandwidths (few MHz). While remaining an extremely challenging project, the detection and imaging of these small-scale structures with LOFAR is a tantalizing possibility within range of the thermal noise of LOFAR (cf. Figure 2.3). Long integration times (approaching weeks or more) may be required. However, LOFAR's multi-beaming capability enables the simultaneous imaging of large areas of sky, effectively permitting very long integrations.
As discussed above, in the description of the search for a 'global' signal, the biggest hurdles to overcome are the removal of discrete and diffuse foreground emission components. Fortunately, both of these contaminants have broad spectral energy distributions, much broader than the few MHz spectral features in the re-ionization signal. Moreover, any residual Galactic signal is expected to show a rather non-uniform distribution over the Galaxy and should not show a preference for a particular spectral range. These are powerful discriminators between the 'uninteresting' contaminants and the real cosmological signals.
The output of this experiment should be a large set of narrow-band images over a wide area of the sky (hundreds of square degrees) , and over a wide frequency range, containing fluctuations due to HI emission and or absorption. The observed spatial and spectral fluctuation signals can be analyzed via standard power spectrum analysis techniques and can be compared with theoretical models for a range of cosmological models and sources of re-ionization.
Figure 2.3 Comparison of the predicted brightness fluctuations as a function of angular scale with the sensitivity attainable after integrations of 100 and 1000 hrs with a compact sub-aperture of LOFAR. The simulationed fluctuation spectra show the amplitude of the 3 sigma peaks in sky brightness computed by Tozzi et al (1999), for two different cosmological models. The sensitivity lines represent a confidence level of 5 times the noise level attained after the indicated integration time. The flat portion of the sensitivity curves at the right side of the plot indicate the response of a fully filled aperture of diameter equal 300m to brightness fluctuation on angular scales greater than 22 arcminutes. Dispersing the same number of elements over a wider area to obtain better angular resolution causes the sensitivity to surface brightess to fall off as denoted by the diagonal line, so that once the aperture is diluted over a 1.5 km diameter area, the instrument should still detect the background peaks at 5 times the noise level on angular scales of 6 arcminutes after 1000 hours of integration.
References