Sources of ionising radiation (Reionisation)

From Eric's Wiki

Sources of ionising radiation

From where do the ionising photons come? There are three candidates: stars, AGN, and shocked gas, with the stars and the AGN the more likely sources.

Stars in galaxies provide more than enough ionising photons, but the photons have a hard time getting out. In order to reionise the IGM by z=6 to 7, the escape fraction of ionising photons from galaxies must be ≥ 0.1 assuming a Saltpeter IMF. The SFE (star formation efficiency) derived from observed values of the SFR over z = 0 to 5 is consistent with a universal value. At low redshift, only a small fraction of hydrogen-ionising radiation from massive stars can escape absorption by the IGM of the host galaxy. For the milky way, the escape fraction is 0.06 (DSF00). Bianchi (et al 2001) find the escape fraction to be <=10% at z=3. See also: Giallongo et al 2002, Steidel et al 2001, Hurwitz et al 1997, Deharveng et al 2001. At higher redshift, galaxies were more compact (is this true?) and gas rich, hence even fewer ionising photons would have escaped. But if the star-formation was off-centre and had a high efficiency, then the fraction could be close to unity (Ricotti & Shull, 2000) (Wood & Loeb, 2000).

Meiksin (2005) points out that, whatever did the reionisation, the sources should not have just disappeared just as reionisation was completed. This would require a conspiracy, since there is no reason for feedback to shut off the sources. So at z∼6, we should be seeing the sources. This constraint removes Pop II stars from being the source.

Pop III stars could not have ionised the universe, since they don't last long enough before they've polluted the universe with metals, leading to Pop II stars. (Meiksin 2005).

However, galaxies still have their supporters. Stiavelli et al (2004) (SFP04) find the observed mean surface brightness of galaxies at z∼6 is sufficient for reionization if the sources are hot and metal poor.

How about stars not in galaxies? Ricotti04 argues globular cluster formation, if the escape fraction of photons is close to unity, would lead to sufficient photons to reionise the ICM at z=6.

How about first objects? FC04: Pop III stars will be hotter and hence have more ionizing photons. Constraints can be put on the shape of the IMF by the optical depth of the CMB (see Kogut03). FC04 rule out the very massive stars only (VMS) IMF and favour a top-heavy IMF since a VMS would produce too many ionizing photons and ionise the universe at too early a time. But early reionisation is not all that bad, and may bring consistency between the QSO spectra and the WMAP polarisation data. Indeed, the need for some high-redshift reionisation has openned the door for more exotic physics. Cosmic strings could cause the perturbations to collapse gas into stars at high redshifts(PV04). This would help explain the low optical depth and high infered redshift of reionisation from Kogut03.

CF04: The PopIII stars were very massive (Coorey et al 2004, eg) but were eclipsed by PopII stars rapidly at z=9. (Finish reading this paper. I've printed it out.)

The x-ray emission from AGN and free-free emission from shocked gas may also contribute. The ionisation cross section for x-ray photons is not very high. However, HM96 point out that HeII is very efficient at converting HeII ionising x-ray photons into HI ionising UV photons. A significant fraction of absorbed source photons are reradiated as Lyα line emission, 2-photon continuum, and recombination continuum radiation. ( HM96 has a good coverage of cosmological radiative transfer). Their Fig 5c charts the evolution of the input spectrum as it passes through a medium.

The broad spectral shape of the emission from the different sources will obviously alter their importance; more high-energy photons means more ionisation. The relative numbers of HI and HeII ionising photons will alter the ratio of HI and HeII column densities, small as they may be. Since this is a conceivably measurable quantity, there is hope of reconstructing the background radiation the gas sees, constraining the possible sources.

Quasars are sufficient to provide the ionising field. MM93 find a deficit of absorption systems with N_HI > 10¹⁵cm ^{− 2} which reduces the opacity at high redshift by a factor of 1.5 to 3 over previous estimates. The mean metagalactic intensity at 912 Å from QSO's is then around . This is enough to provide the ionising photons. Indeed, it would require only to ionise the IGM.

QSO's cannot be responsible for reionisation. Meiksin (2005) shows there just weren't enough of them.

Quasars have shallow spectra (α=-0.5), meaning lots of high-energy photons. Scott04 looked at 85 nearby (z < 0.67) observed with FUSE. After accounting for all sources of systematic error (galactic absorption in particular), they find in the EUV spectrum α=-0.56 (+0.4, -0.3). This is significantly harder than the result in the EUV spectrum by HST (α=-1.76 (± 0.12)) (Telfer et al 2002). The differences between the wavelength pass of FUSE and HST and the redshifts mean they sample the same rest wavelengths. But the FUSE suffers less from IGM absorption. The difference in the result may be a different treatment (or parameters in the treatment) of the correction for Lyα absorbers. The other possibility is evolution. Scott04 claim to find a hardening of spectra with decreasing redshift. In any case, there is a large scatter among the slopes, from α=-4 to α=2 (cf. the Baldwin effect) Note that an α≥2 integrates into an infinite number of ionising photons, so clearly the value of α is variable over the weavlength range. In Scott04, α is measured at λ=1000Å.

Quasars (read AGN) beam their ionising luminosity. That means numberdensities must assume a beaming factor. If the beaming factor evolves with time without being accounted for, there will be a false evolution in the number density.

QSO's and other sources are likely to be biased towards higher-density regions (Kauffmann & Haehnelt 2000). See also Cen03b, Ciardi, Stoehr & White (2003), Furlanetto et al (2004), Gnedind & Prada (2004), Santos (2004), WL04a.

Miniquasars, formed during the first round of star formation that produces the PopIII stars, have been suggested as a source (Madau04). They form in mini DM halos at z > 20. Since quasars have a hard spectrum, it doesn't take as many to get the job done.

Could a contribution to reionisation be due to the decay of exotic particles? Dark matter is likely an exotic particle and there is lots of it. WMAP polarization data imply a high Compton optical depth suggesting reionisation was well under way at redshift much larger than z∼6. SDSS data are not consistent with the photons coming from stars. An early bout of star formation of high-mass stars might provide the early reionisation. KK04 suggest decaying 30 eV particles which are consistent with WMAP and SDSS. 30 eV particles emit 15 eV photons which is conveniently close to the 13.6 eV ionisation potential of HI. They look at a range of particle lifetimes of 1014 to 1016 s, with a preference for the shorter lifetime.

Another source of UV photons: the shock-heated gas produced during structure formation. MFWB04 call this contribution thermal photons and find that there are enough to produce and sustain HeII reionisation at z∼6 and are comparable in numbers to those from QSOs at z∼3. This implies, among other problems, that the escape fraction of ionising photons from galaxies must be < 0.03. Kim, Cristiani, & D'Odorico 2001 found a QSO ionising background does not reproduce the redshift evolution of absorbers. Bianchi, Cristiani, & Kim try to fit it with a mix of QSOs and galaxies, using an escape fraction of 0.1 for the galaxies. The actual escape fraction is not well know, but believed to be < 0.04.

What is the shape of the ionising spectrum? The shape of the ionising spectrum constrains the sources, so finding the shape is pretty important. The shape can, however, be modified by frequency-dependent attenuation (absorption). The shape is characterised by the spectral index η such that I(ν) = I₀(ν / ν₀)η. The spectral index is generally < 0. A very low value, (eg. -1.5) is a soft spectrum, while a higher value (eg. -0.5) is a hard spectrum. HeII is sensitive to the spectral index (HM96 & Miralda-Escudé93): τ_HeII / τ_HI = η / 4 or α_HeII / α_HIΓ_HI / Γ_HeIIn_He / n_H = η = 1.7S (Giroux et al 1995) (Note: η in these two equations is not the spectral index, but the column density ratio. But S is the spectral softness parameter (which is?)) HAR97 looked at the HeII Lyα forest (HeII is hydrogen-like, so has all the lines of HI, just shifted in energy) in the spectrum of a z=3.285 quasar. They find HeII absorption with τ>1.3 in redshift intervals without HI absorption. They claim this is evidence of a soft spectrum and te IGM helium is doubly ionised by z=3.3.

There is a wide range of values for Γ_HI reported from simulations due to the sensitivity of the input cosmology (h, T, τ_eff). This is similar to my findings. BHVS04 studied the mean metagalactic ionisation rate in the fashion similar to Meiksin et al but varied cosmological parameters. Also, they used the fitting formula of (Schaye 2003) to get Γ from τ_eff and the simulation data. They find Γ_HI = [1.3±0.6,0.9±0.3,1.0±0.4] for z = [2,3,4]. The errors are dominated by uncertainty in the temperature of the low-density gas. The ionisation rate is twice that expected from the emissivity of QSO's. As simulation box size increases, attenuation lengths increase as more gas is drawn into larger structures, depleting the voids which dominate the filling-factor. This decreases Γ_HI. This effect is minimal for box sizes larger than 50 Mpc. Finer mass resolution has a similar effect, as tighter halos (and smaller SPH smoothing lengths) draw gas into denser structures. They used Gadget with up to 4003 gas particles with convergence at the 7% level for a doubling of the linear resolution.

The abundance of HeII is sensitive to the ionising spectrum. Too soft and He remains neutral. Too hard and HeII becomes HeIII. The ionising spectrum is a function of the source spectrum, modified by frequency-sensitive absorption, with the net outcome the hardening of the spectrum (Madau & ?). The source spectrum is obviously dependent on the source: star formation produces a soft spectrum, AGN a hard spectrum.

The background spectrum must be evolving. Songaila (1998) reports an abrupt change in the CIV/SiIV ratio at z∼3 but Kim, Cristiani, & D'Odorico don't see it. MFWB04 suggest a change would be due to HeII reionisation. Ricotti, Gnedin, & Shull find a change a z∼3 as well.

Reimers04 confirm the result of Kriss et al 2001 that the ratio of HeII/HI varies over many orders of magnitude with ranging from 1 to > 1000. They also observe a correlation between small column densities and high ratios. They observed a QSO in the redshift range of 2.3 < z < 2.7 with FUSE. η is effectively a measure of the background UV spectral index. Tefler (et al 2002) point out that the scatter is too large to be explained by different QSO spectral indices since the scales of fluctuations are smaller than the typical distances between QSOs. Fortunately, FR04 determines that part of the scatter is an artifact of the analysis.

The reionisation spectrum must have been hard (α_S∼0.5), regardless of the source density (Meiksin 2005).

For a soft spectrum, helium and hydrogen would have been ionised together. For a hard spectrum, helium would be ionised before hydrogen (Meiksin 2005, Fig 6).

Limits on the star formation rate of Pop III stars have been inferred from the GRB rate. Murakami05 finds Pop III could have formed continuously from z = 4 to 12.

QSO's are not sufficient to explain the ionisation state of the ISM/WHIM. BHVS05 find from simulations of the metagalactic hydrogen ionisation rate that a significant contribution from star formation regions is required. The errors in the simulations are dominated by lack of knowledge of the thermal state of the gas.

Star formation as a source for reionisation is difficult because there simply wasn't much. Bunker05 find a 6 times drop in SF between z=3 and 6 (6 is less). However, they find a population of stars formed between z=8 and 14. SE05 note SF decreases monotonically with redshift from z=3 to z=7 but agrees that there is a need for an earlier burst since old populations are observed at high redshift. HCCK05 also find that the luminosity function has evolved little between z=6.5 and 5.7 implying that the stellar populations must be mature at high redshift.

Miniquasars can't be the only answer to the sources of reionisation. SHF05 note that since we don't see them in the x-ray background, they must be either rare or short-lived. In either case, they can't contribute significantly to reionisation.

Willott05 report an unsuccessfull search for z=6 QSO's, allowing them to set a limit on the QSO contribution to the ionising flux of less that 30% of the SF flux.

Decay of sterile neutrinos, a WDM candidate, cannot be a significant source of reionising photons (MF05).

Panagia05 reports on z=6.5 galaxy which had a starburst at about =15±5.

Nagao05 found a z=6.33 galaxy with plenty of Lyα emission but no emission from HeII. This rules out an AGN or Pop III stars. So the stars must have been formed earlier.

Cen05 argues that dwarf galaxies in clusters were the source of ionising photons. Dwarf galaxies are old and faint. There is plenty to argue that they formed prior to reionisation. Most importantly, they have a different luminosity function from other galaxies, cluster or otherwise. Secondly, reionisation should have injected enough entropy to prevent their formation. Cen05 calculates that for every hydrogen atom in a cluster, an ionising photon was produced in a dwarf galaxy, assuming f_esc = 0.1. Since clusters have only %20 of the mass, the book-keeping is a factor of 5 short. But mergers of dwarf galaxies would mean there must have been more dwarf galaxies before. Also, the calculation ignores the contributions of larger galaxies. The picture is attractive because it also explains the lack of dwarf galaxies in the field population (I didn't know that was a problem).

Using simulations, TLE98 showed that the low-redshift evolution of the Lyα forest is tied to the evolution of the QSO luminosity function. In particular, for z < 2, the forest hasn't evolved much and the QSO's have disappeared.

BIBF05 find significant evolution in the UV luminosity function for galaxies at z∼6. Galaxies at z∼3 are brighter than galaxies at z∼6. The luminosity density is also lower at z∼6 by a factor of 0.68. But there are still enough UV flux to reionise the universe.

Mesinger05 found the ionising luminosity of SDSS J1030+0524 to be 2 to 10 times lower than expected and at a level that precludes quasar-dominated reinoisation.

BI05 found >500 z∼6 galaxies in deep HST fields. With these, they found evidence of evolution of the luminosity function for galaxies. In the past, galaxies were fainter in agreement with hierarchical formation.

High-z galaxies are a possible source, if you can find them. Panagia05b reports on a z=6.5 galaxy in the Hubble Deep Field. He argues it must have been producing sufficient ionisation to account for at least some of the ionisation detected by WMAP at z>10. Using simulations, DFO05 predict enough galaxies to account for reionisation.

Using simulations with radiative transfer that properly models the temperature, TM06 show the final gas temperature is sensitive to the shape of the source spectrum. Moreso, the order in which gas is ionised has a profound effect since the mean molecular weight drops by 45% during hydrogen ionisation while it drops only 7% during helium ionisation.

What causes the ionisation: local sources or the diffuse ionisation field? Does the answer depend on the nature of the structure? Low density gas will have a low density of sources but be easily ionised by a weak field. High density gas will have a high local density of sources and will require a strong field to overcome the enhanced recombination rates. Hence, low density gas will preferentially be ionised by the diffuse field and high density gas by local sources. Using simulations, KG06 confirm this trend.

Galaxies produce enough photons to account for reionisation, but they also absorb most of the photons, leaving too few for reionisation. So an important parameter controlling galaxies as a source is the escape fraction. IID06 collected the data of others to show that the escape fraction was probably higher in the past (about 0.1) than now (about 0.01 or less). A higher fraction improves the potential for galaxies to be the sources.

--Etittley 14:55, 24 June 2007 (BST)

Back to Reionisation