Adding the latest data (red squares) and applying a survival analysis, in order to utilise the upper limits to the non-detections (unfilled symbols), we confirm a significant anti-correlation (3.63σ, Curran, 2020).

Comparing the 21-cm absorption strength with the total atomic hydrogen column density yields the spin temperature of the gas, which is significantly higher than Galactic values. Furthermore,  while the temperature in the Milky Way exhibits a constant 250 - 400 K out to 25 kpc, we find a peak in excess of  3000 K at r ~ 15 kpc, which is where the ionised gas (HII) regions reside in spiral galaxies. From the neutral gas density distribution, we find that the Strömgren spheres around young, hot (OB) stars in the outer disk have radii comparable to the width of the disk, supporting the hypothesis that the HII regions are the result of star formation in the spiral arms.

Complete Ionisation of the Gas in High Redshift Galaxies

Cold neutral gas, the reservoir for star formation, is readily detected through the 21-cm transition of hydrogen (HI) within the hosts of active sources (radio galaxies/quasars) at redshifts of z < 1 (back to half the age of the Universe). However, above this redshift the detection rate drops drastically, which we attribute to the intense ultra-violet radiation emanating from matter accreting onto the super-massive black hole, located at the centre of each galaxy, ionising all of the gas in the host (Curran et al., 2008). It is this radiation which allows quasars to be seen over the entire span of the Universe - over its several billion year journey towards us, the radiation is redshifted into visible light and dimmed until these massive, luminous distant objects appear as stars (or Quasi-Stellar Objects, QSOs).

The traditional optical selection of sources at high redshift selects objects above a critical luminosity (Curran et al., 2017,2019), where all of the gas in the host galaxy may be ionised by the UV radiation from the black hole accretion (Curran & Whiting, 2012). This would have profound implications for how the black hole affects its environment, inhibiting star formation, as well as presenting the issue of how the black hole is continually fed. Furthermore, since our own Milky Way has a population of old stars dating from these early epochs in addition to the near-by Universe being abundant in gas-rich galaxies, a critical UV luminosity implies that there must be an unknown population of gas-rich galaxies in the distant Universe hidden from optical spectroscopy.

Fuelling Star Formation over the History of the Universe

While the above describes the neutral gas in the hosts of active galaxies, the radiation can also be absorbed by a foreground galaxy lying along the same sight-line. Unlike the active galaxies, the neutral gas detected in these intervening galaxies is sufficiently dense to saturate the absorption spectrum, giving the so-called Damped Lyman-alpha Absorbers (DLAs), which may contain up to 80% of the baryonic matter in the Universe.

The abundance of neutral gas in DLAs presents a major issue for theories of cosmic star formation and galaxy evolution, in that the star formation density is known to exhibit a steep cosmic evolution, peaking at redshifts of z ~ 2 (the dotted trace and right-hand axis on the plot), whereas the abundance of neutral gas exhibits little evolution (the broken trace and left-hand axis).

Damped Lyman-alpha absorption (and at lower redshift, 21-cm emission) by neutral hydrogen traces all of the neutral gas, however, only the very coldest gas can initiate star formation. Comparison of the strength of the 21-cm absorption with the total hydrogen column density can yield the fraction of cold gas degenerate with the coverage of the background flux. By correcting for the geometry effects introduced by an expanding Universe (Curran, 2012) and applying the known evolution of galaxy sizes, we find that the cold gas fraction may trace the star formation history (the error bars in the plot, Curran, 2017). This could finally address the long-standing issue of the disparity between the star formation history and material available to fuel it. As seen from the plot, the high redshift bin shows a departure from the star formation rate density. This, however, comprises one detection and 23 limits (via a survival analysis) and so further detections of 21-cm absorption are required at redshifts of z > 3 (Curran, 2019).

Machine Classification of Absorption Spectra

As described above, the use of an optical redshift biases towards sources in which the gas within  the host is likely to be ionised. Optical selection of targets also favours sight-lines poor in molecular gas, as evident through their optical - near infrared colours (Curran et. al, 2006, 2011). Millimetre wave-band molecular absorption transitions are insensitive  to the value of the fine structure constant, and thus provide

valuable “anchor lines” with which to calibrate the shift of the 21-cm transition of hydrogen (see Curran, Kanekar & Darling, 2004). Molecular transitions also provide excellent diagnostics of the physics (density and temperature) and chemistry of the galaxy’s interstellar medium (e.g. Curran et al., 2001).

This is just one line-of-sight with a six-antenna prototype and the full First Large Absorption Survey in HI (FLASH) on ASKAP will search 150,000 sight-lines with 36 antennas. I have therefore been exploring machine learning techniques with which to determine the nature  of the absorbing galaxy in the absence of an optical spectrum.  From the ~50 of each intervening and associated absorption, we obtain a > 80% accuracy in the prediction of absorber type. As more spectra are added to the classifier, we expect this accuracy to increase and thus provide an invaluable tool in determining the populations of active and quiescent galaxies discovered through surveys with the next generations of large radio telescopes (Curran 2021).

Geometry Effects on HI 21-cm Absorption

For associated absorption, it has been suggested that column density of atomic gas is anti-correlated with the projected extent of the radio source. That is,  the smaller the radio source, the denser the absorbing medium, which could be consistent with either the ‘frustration scenario’ (van Breugel, Miley & Heckman 1984), where the jets are confined, or the ‘youth scenario’ (Fanti et al. 1995), where the gas has yet to be expelled.

However, the column density can only be obtained from 21-cm emission or Lyman-alpha absorption and so all of these studies assume a single spin temperature for the gas and a covering factor in each absorber. Hence,  rather than column density, it is the integrated optical depth which is anti-correlated with the extent of the radio source. I have shown that this was driven by the optical depth of the absorption (by removing the velocity dispersion measure information) and that an optical depth--source size anti-correlation is expected purely through the observed optical depth, which is defined by the actual optical depth and covering factor (Curran et al. 2013). Hence, the optical depth (or “column density”) and extent of the radio source are not independent parameters and the observed anti-correlation is just what would be expected from geometry alone.

Hence, future surveys should dispense optical pre-selection, using blind spectral scans towards unidentified radio sources. Without an optical spectrum, however, we require another means to determine whether the absorption is associated or intervening, For example, our recent detection obtained by scanning with the six-element test array of the Australian Square Kilometre Array Pathfinder (ASKAP), had photometric redshifts which spanned z  = 0.347 – 0.63 (a difference of 2 Gyr in look-back time). Thus, we required 1.5 hours of Director’s Discretionary Time on Gemini-South to confirm that absorption is associated with  zopt = 0.44230 ± 0.00022 cf. zHI = 0.44129230 ± 0.00000040 (Allison 2015 - the spectrum below shows just one eighth of the band, which spans over 3 Gyr in look-back time).

Ionisation in the Outskirts of Galaxies

Although it is well established that there is a decrease in total gas column density with galactocentric radius in spiral galaxies, in Curran et al., 2016 we showed, for the first time, a decrease in 21-cm absorption strength with impact parameter.

Development of a Radio Photometric Redshift

1. •Optical classification of high redshift source searched in HI 21-cm (atomic) absorption and CO (molecular) emission - this would allow us to determine how much of the ionising UV radiation arises from young stars in comparison to the black hole accretion disk (Curran et al., 2016).
   

2. •Measuring the polarisation of quasar light by the ionised gas by comparing the Faraday rotation measure of the radio source with the ionising photon rate/UV luminosity.
   

3. •A potentially novel technique for weighing super-massive black holes in the centres of galaxies  (Curran et al., 2016).
   

Variation of the Fundamental Constants

State-of-the-art observations with the world’s largest optical telescopes of the ultra-violet transitions in hundreds of distant galaxies absorbing the light from more distant Quasi-Stellar Objects (QSOs), provide strong evidence for a change in the fine structure constant, α, over the last 13 billion years. Such a variation would have a profound impact on our understanding of the Universe and  offer an experimental test of current Grand Unified theories of physics, which invoke extra spatial dimensions, the evolution of which would naturally result in variation of the fundamental “constants” of Nature. Any successful Grand Unified theory should predict why the constants have the values they have,  resolving the prominent ``fine tuning'' issue, where if the fundamental constants  (which are dependent upon 27 independent parameters) were even slightly different than the observed values, life could not appear in the Universe.

Naturally, the detection of such a variation is controversial and given that the optical data are of the highest standard possible (to the point that the frequencies of the transitions at 12 billion light-years was two orders of magnitudes more accurate than in the laboratory), other independent avenues must be explored. This is where radio spectroscopy comes in - not only are radio receivers phase-locked to a maser giving highly accurate redshifts, but the 21-centimetre spin-flip transition of neutral hydrogen is 30 times more sensitive to a change in α than the ionic transitions in the optical band. Furthermore, where 21-cm absorption is detected, other species follow, such as molecular lines in the millimetre band and the hydroxyl (OH) radical in the decimetre band, all of which have a variety of dependences on the constants (including the electron-proton mass ratio). Thus, the inter-comparison of all of these transitions over all redshifts, will give accurate measurements of several fundamental constants over the history of the Universe (Curran et al. 2004, 2011).

‘Oumuamua as Evidence of Extraterrestrial Intelligence

Being able to estimate the redshift of a continuum radio source from its spectral energy distribution (SED) will prove invaluable to the cosmological applications of the next generation of continuum surveys with the Square Kilometre Array and its precursors (e.g the Evolutionary Map of the Universe). Furthermore, as discussed above, obtaining a redshift from an optical spectrum selects against sources which contain cold, neutral (star-forming) gas. However, due to their relatively featureless radio SEDs, this is proving elusive (M. Majic’s Honours project). While still exploring this avenue, we have recently shown that optical, NIR & UV photometry (which can be detected much deeper than the corresponding spectroscopy) can be applied to radio sources (Curran, 2020 , Curran et al. 2021, Curran et al. 2022).

The plot on the left shows the result of using deep learning (neural networks) trained on a large  optical sample (SDSS) and validated on a radio sample (FIRST), from which we see accurate predictions over the last 11 billion years of cosmic history. The plots on the right show the optical band SEDs of the training (optical) and validation (radio) samples and the distributions of radio fluxes. From this we see that the training is valid over a wide range of radio-loudness. Thus, this method promises much potential in obtaining the redshifts of sources detected with the Square Kilometre Array, which are faint to yield an optical spectrum.

Other Projects

In 2017 the first known visitor from outside the Solar System, ‘Oumuamua, was observed passing close to the Sun. Since no cometary tail was visible there has been much controversy over the origin of its small non-gravitational acceleration, with an artificial  origin being proposed. Specifically, that ‘Oumuamua is a light sail sent as a probe by an alien civilization. While  the cause of the acceleration has been disputed, if a light sail, I have shown that the fastest speed attainable (the terminal velocity)  is less than one kilometre per second (Curran, 2021). While this is nearly three times the speed of sound, it means that ‘Oumuamua would take two million years to travel from Proxima Centauri, the nearest extrasolar star, and half a billion years from its projected point of origin. This is still very much in our corner of the Galaxy, but at a time when life first emerged from the oceans on Earth. Thus, it is very unlikely that  ‘Oumuamua was sent by an alien intelligence, but probably just an unusually shaped rock which wandered in the Solar System. Due to the terminal velocity, relativistic velocities are not attainable with a sail powered by starlight. This has profound implications for the travel times of current conceptual solar (light) sails.