1. Complete Ionisation of the Neutral Gas in the Hosts of High Redshift AGN As Traced Through HI and MgII Absorption

2. In absorption, the 21-cm transition of neutral hydrogen (HI): Traces the cool component of the neutral gas. That is, the raw material for star formation, which in turn forms planets and all heavy elements. Unlike the Lyman-  transition of HI (which traces all of the neutral gas), can be observed at z = 0 by ground-based telescopes (cf. z > 1.7). Unlike 21-cm emission, can be readily detected at z > 0.2, since absorption strength only dependent upon column density and background flux.

3. N HI.f/T spin = 1.823 x 10 18  dv N HI – total hydrogen column density [atoms cm -2 ] f – flux intercepted by gas (covering factor) T spin – spin temperature of the gas [K] Before 2008 ~50% detection rate attributed to orientation of absorbing torus/disk of dense gas Observable – integrated optical depth of profile Curran et al. (2008) Where is the cold neutral gas in high redshift radio galaxies and quasars? Absorption by gas in disk No absorption along this sight-line All new searches non-detections and still only one detection at z > 3 (Uson et al. 1991)

4. High redshifts are the most distant and so bias towards most UV luminous objects (accretion around supermassive BHs) -> gas is excited/ionised to beyond detection Curran et al., (2008), Curran & Whiting (2012)

5. Why is there a critical luminosity at all?Why is there a critical luminosity at all? Why is this 10 56 ionising photons sec -1 (L UV 10 23 W Hz -1 )?Why is this ≈ 10 56 ionising photons sec -1 (L UV ≈ 10 23 W Hz -1 )? Searches are generally limited to column densities of N HI 10 18 (T spin /f) cm -2. So is the neutral hydrogen ionised to just below the sensitivity thresholds of current large radio telescopes?Searches are generally limited to column densities of N HI ≈ 10 18 (T spin /f) cm -2. So is the neutral hydrogen ionised to just below the sensitivity thresholds of current large radio telescopes? That is, could the Square Kilometre Array readily detect the reservoir for star formation currently missing in high redshift radio galaxies and quasars?

6. Photo-ionisation rate Recombination rate n p, n e – ion densities [cm -3 ]   – radiative recombination rate of HI [cm 3 s -1 ] r ion – extent of ionisation (Strömgren sphere)

7. n 0 = 10 cm -3,  = 1.27 x 10 -12 cm 3 s -1 [i.e. T = 2000 K] 3 x 10 56 photons sec -1 (L 912 ≈ 10 23 W Hz -1 ) => R = 2.9 kpc, cf. ≈ 3 kpc HI in Milky Way (Kalberla et al., 2007). Strömgren sphere is infinite for a finite luminosity. For an exponential density distribution there is a critical luminosity, above which all of the gas is ionised. Distribution of neutral hydrogen in Milky Way (Kalberla & Kerp, 2009)

8. Observational verification using MgII Singly ionised magnesium (Mg + /MgII): Has a similar ionisation potential to HI (15.0 cf. 13.6 eV). Is a commonly detected transition at 0.2 < z < 2.2 and a proxy for HI at redshifts where Lyman-  transition inaccessible to ground-based telescopes (z < 1.7). Thus, may also expect a critical luminosity above which associated MgII absorption is not detected

9. Publicly available catalogues, in which associated systems are included and with SDSS photometry, give 16,269 QSO sight-lines searched for MgII absorption – 8098 of which have at least one strong (W r λ2796 ≥ 1 Å, Zhu & Ménard 2013) absorption system. After matching each QSO with GALEX NUV and FUV photometry and correcting for Galactic extinction (Schlegel et al., 1998): Rest-frame 912 < λ < 1216 Å corrected for Lyman-α Forest Rest-frame λ < 912 Å corrected for Lyman Limit Systems (due to presence of strong Lyman-α absorption) A broken power law with a break at λ = 1200 Å (Shull et al., 2012 and references therein) was fit to the corrected photometry

10. Out of the 9517 Mg II absorbers (towards 8098 QSO sight-lines), there are 296 associated (Δv < 3000 km s -1 ) systems all of which are below Q MgII = 6.2 x 10 56 ionising photons sec -1 (cf. Q HI = 3 x 10 56 sec -1, not corrected for forest or LLSs) Detections as a whole = 56.018 ± 0.005 Associated detections = 55.67 ± 0.03 Standard Gaussian fit to associated gives tail > 1.7σ missing (6.15% of area) => P(τ) = (1 - 0.0615) 296 = 6.93 x 10 -9 => S(τ) = 5.79σ Kolmogorov-Smirnov test gives => P(τ) = 1.01 x 10 -5 [S(τ) = 4.40σ] that associated and non-associated drawn from same sample 4.11% detection rate of associated absorption at Q MgII 6.2 x 10 56 sec -1

11. Conclusions Associated MgII absorption is not detected towards QSOs with ionsing photon rates of Q > 6 x 10 56 sec -1. This compares well with the Q > 3 x 10 56 sec - 1 (L UV ≈ 10 23 W Hz -1 ) found from the HI 21- cm observations, suggesting that the HI is ionised rather than just excited to the upper hyperfine level. Applying Q = 6.2 x 10 56 sec -1 to the model gives (for n 0 = 13 cm -3 ) a scale length of R = 3.09 kpc, cf. 3.15 kpc for the Milky Way indicating that this is sufficient to ionise all of the gas in a large spiral galaxy. This confirms our hypothesis that the dearth of associated HI 21-cm absorption at z ≥ 3 is due to a selection effect, where the most UV luminous AGN ionise all the atomic gas in their host. This ionisation of the star-forming reservoir has implications for how the AGN could suppress star formation within the host and suggests that we are unlikely to ever detect associated HI 21-cm, Lyman-α or MgII absorption in the most luminous (L UV ≈ 10 23 W Hz -1 ) objects.