Exploring the Early Universe

A major goal of contemporary astrophysics and cosmology is to achieve a broad understanding of the formation of the first collapsed structures (Pop III and early Pop II stars, black holes and galaxies) during the first billion years in the life of the universe. This is intimately connected to the reionization of the IGM and build up of global metallicity. The latter is very poorly constrained, and even in the JWST era will rely on crude emission line diagnostics for only the brightest galaxies. Regarding reionization, measurements of the Thomson scattering optical depth to the microwave background by the Planck satellite now suggest it substantially occurred in the redshift range z ~ 7.8 – 8.8 (e.g., Planck collaboration. 2016), whereas the observations of the Gunn-Peterson trough in the spectra of distant quasars and galaxies indicate it was largely finished by z ~ 6.5 (e.g., Schenker et al. 2014). Statistical measurements of the fluctuations in the redshifted 21 cm line of neutral hydrogen by experiments such as LOFAR and SKA are expected to soon provide further constraints on the time history (e.g, Patil et al. 2014). The central question, however, remains whether it was predominantly radiation from massive stars that both brought about and sustained this phase change, or whether more exotic mechanisms must be sought?

Solving this problem largely splits into two subsidiary issues: how much massive star formation was occurring as a function of redshift? And, on average, what proportion of the ionizing radiation produced by these massive stars escaped from the immediate environs of their host galaxies? The former can be extrapolated based on observed candidate z > 7 galaxies found in HST deep fields, but two very significant uncertainties are, firstly, the completeness and cleanness of the photometric redshift samples at z > 7, and, secondly, the poorly constrained form and faint-end behaviour of the galaxy luminosity function (at stellar masses <~ 108 M), especially since galaxies below the HST detection limit very likely dominate the star-formation budget. Even though some constraints on fainter galaxies can be obtained through observations of lensing clusters (e.g. Atek et al. 2015 ApJ 814 69), which will be improved further by JWST, simulations suggest star formation was likely occurring in considerably fainter systems still (Liu et al. 2016). The second problem, that of the Lyman continuum escape fraction, is even more difficult since it cannot be determined directly at these redshifts, and lower redshift studies have generally found rather low values of fesc of only a few percent (e.g., Nestor et al. 2013), likely insufficient to drive reionization unless fesc increases significantly at early times and/or for smaller galaxies (e.g., Robertson et al. 2013).

GRBs and their host galaxies provide several powerful alternative routes to answering these key questions, potentially probing star formation, metal enrichment and galaxy evolution even preceding reionization.


  • 2

  • 3

Figure 2. Redshift distribution of GRBs as of Nov. 2011 (Coward et al. 2011). Right: The cumulative distribution of GRBs with redshift determination as a function of the redshift for Swift (in 10 yr) and THESEUS (in 3 yr). We note that these predictions are conservative in so far as they reproduce the current GRB rate as a function of redshift. However, with our sensitivity, we can detect a GRB of  Eiso ~ 1053 erg (corresponding to the median of the GRB radiated energy distribution) up to redhsift 12. Indeed, our poor knowledge of the GRB rate - SFR connection does not preclude the existence of a sizeble number of GRBs at such high redshifts, in keeping with recent models of pop III stars.


Global star formation from GRB rate as a function of redshift

Long-duration GRBs are produced by massive stars, and so track star formation, and in particular the populations of UV-bright stars responsible for the bulk of ionizing radiation production. Although there is evidence at low redshift that GRBs are disfavoured in high metallicity environments, since high-z star formation is predominantly at low metallicity (e.g., Salvaterra et al. 2013; Perley et al. 2016) it is likely that the GRB rate to massive star formation rate is approximately constant beyond z ~ 3. Thus simply establishing the GRB N(z), and accounting for the instrumental selection function, provides a direct tracer of the global star formation rate density as a function of cosmic time. Analyses of this sort have consistently pointed to a higher SFR density at redshifts z > 6 than traditionally inferred from UV rest-frame galaxy studies (Figure 3 left), which rely on counting star-forming galaxies and attempting to account for galaxies below the detection threshold. Although this discrepancy has been alleviated by the growing realisation of the extremely steep faint-end slope of the galaxy LF at z > 6, it still appears that this steep slope must continue to very faint magnitudes, MAB ~ -11 in order to provide consistency with GRB counts and indeed to achieve reionization (something that can only be quantified via a full census of the GRB population).


The high-z galaxy luminosity function

As discussed above, this is a key issue for understanding of reionization since, to the depth achieved in the Hubble Ultra-deep Field (HUDF), it appears that the faint-end of the LF at z > 6 approaches a power-law of slope α = 2. Thus the value of the total luminosity integral depends sensitively on the choice of low-luminosity cut-off (and indeed the assumption of continued power-law form for the LF). By conducting deep searches for the hosts of GRBs at high-z we can directly estimate the ratio of star-formation occurring in detectable and undetectable galaxies, with the sole assumption that GRB-rate is proportional to star-formation rate (Figure 3 left). Although currently limited by small-number statistics, early application of this technique has confirmed that the majority of star formation at z > ~ 6 occurred in galaxies below the effective detection limit of HST (Tanvir et al. 2012; McGuire et al. 2016). Since the exact position and redshift of the galaxy is known via the GRB afterglow, follow-up observations are more efficient than equivalent deep field searches for Lyman-break galaxies.


  • 4

  • 5

Figure 3. Left: Star Formation Rate density as a function of redshift as derived from rest-frame UV surveys (gray points and green hatched region for corresponding model) and from GRBs based on different assumptions (GRB rate to SFR ratio, GRB progenitor metallicity) (Robertson & Ellis 2012). Yellow points show corresponding estimates expected from THESEUS. Right: Absorption-line based metallicities [M/H] as a function of redshift of Damped Lyman-alpha absorbers, for GRB-DLAs (blue symbols) and emission line metallicities for GRB hosts (red symbols) [adapted from Sparre et al. 2014]. GRBs are essential to probe evolution of ISM metallicities in the first billion years of cosmic history.


The Lyman continuum escape fraction

Direct observations of escaping Lyman continuum radiation is essentially impossible at z > 6, especially for the small galaxies responsible for the bulk of star formation. GRBs provide a powerful alternative since high-S/N afterglow spectroscopy reveals the neutral hydrogen column along our line-of-sight to the GRB. Since the opacity of the medium to FUV photons depends on this column, a statistical sample of afterglows can be used to infer the average escape fraction over many lines of sight, specifically to the locations of massive stars dominating global ionizing radiation production. Useful constraints have so far only been possible at z = 2-4, indicating an upper limit of fesc< 7.5% (e.g., Fynbo et al. 2009), but future observations of z > 6 GRBs, particularly in the era of 30 m ground based telescopes, promise to provide a much more precise constraints during the epoch of reionization (see below and Figure 4).


  • 6


Figure 4: The UV luminosity density from stars at z~8 and average escape fraction <fesc> are insufficient to sustain reionization unless the galaxy luminosity function steepens to magnitudes fainter than MUV=-13 (grey hatched region), and/or <fesc> is much higher than that typically found at z~3 (grey shaded region). Even in the late 2020s, <fesc> at these redshifts will be largely unconstrained by direct observations. The green contours show the 1- 2-σ expectations for a sample of 25 GRBs at z~7-9 for which deep spectroscopy provides the host neutral column and deep imaging constrains the fraction of star formation occurring in hosts below the JWST limit (Robertson et al. 2013 ApJ 768 71). The input parameters were log10(ρUV)=26.44 and <fesc>=0.23, close to the (red) borderline for maintaining reionization by stars.


The build-up of metals

Bright GRB afterglows with their intrinsic power-law spectra provide ideal backlights for measuring not only the hydrogen column, but also obtaining exquisite abundances and gas kinematics probing to the hearts of their host galaxies (e.g., Hartoog et al. 2015). Further, the imprint of the local dust law, and in some cases observation of H2 molecular absorption, provides further detailed evidence of the state of the host ISM (e.g. Friis et al. 2015). Thus they can be used to monitor cosmic metal enrichment and chemical evolution to early times, and search for evidence of the nucleosynthetic products of even earlier generations of stars. In the late 2020s, taking advantage of the availability of 30m class ground-based telescopes (and possibly Athena), superb abundance determinations will be possible through simultaneous measurement metal absorption lines and modelling the red-wing of Lyman-alpha to determine host HI column density, potentially even many days post-burst (e.g. Figure 5). We emphasize that using the THESEUS on-board NIR spectroscopy capabilities, will provide the redshifts and luminosity measurements that are essential to optimising the time-critical follow-up observations using the highly expensive next-generation facilities, allowing us to select the highest priority targets and use the most appropriate telescope and instrument.


  • 7


Figure 5. Simulated E-ELT 30 min spectrum of a faint GRB afterglow observed after ~1 day. The S/N provides exquisite abundance determinations from metal absorption lines, while fitting the Ly-a damping wing simultaneously fixes the IGM neutral fraction and the host HI column density, as illustrated by the two extreme models, a pure 100% neutral IGM (green) and best-fit host absorption with a fully ionized IGM (red).



Topology of reionization

In practice, it is expected that reionization should proceed in a patchy way, with ionized bubbles being created first around the highest density peaks where the first galaxies occur, expanding and ultimately filling the whole IGM. The topology of the growing network of ionized regions reflects the character of the early structure formation and the ionizing radiation field. With high-S/N afterglow spectroscopy the Ly-alpha red damping wing can be decomposed into contributions due to the host galaxy and the IGM. The latter provides the hydrogen neutral fraction and so measures the progress of reionization local to the burst. With samples of several tens of GRBs at z > 7-8, we can begin to statistically investigate the average and variance of the reionization process as a function of redshift (e.g., McQuinn et al. 2008).


Population III stars

In addition to possible evidence for population III chemical enrichment, it has been argued that pop III stars may also produce collapsar-like jetted explosions, which are likely to be of longer duration than typical long-GRBs, and may be detected through surveys with longer dwell times (Meszaros & Rees 2010). To date no direct evidence of this connection has been observationally established. The multiwavelength properties of GRBs with a Pop-III

progenitor are only predicted on the expected large masses, zero metallicity of these stars. Even the detection of a single GRB from a popIII progenitor would put fundamental constraints on the unknown properties of the first stars.


The role of THESEUS

Our detailed simulations indicate that THESEUS will detect between 30 and 80 GRBs at z > 6 over a three year mission, with between 10 and 25 of these at z > 8 (and several at z > 10). The on-board follow-up capability will mean that redshifts are estimated for the large majority of these, and powerful next generation ground- and space-based telescopes available in this era will lead to extremely deep host searches and high-S/N afterglow spectroscopy for many (e.g. using JWST, if still operating, E-ELT, ATHENA etc.). To illustrate the potential of such a sample, we simulate in Fig. 1.3 the precision in constraining the product of the UV luminosity density and average escape fraction, ρ UV  fesc, that would be obtained with 25 GRBs at 7<z<9 having high-S/N afterglow spectroscopy and (3hr) JWST depth host searches (for definiteness the ρUV axis corresponds to z=8). This will provide a much clearer answer to the question of whether stars were the dominant contributors to reionization. In addition, this sample will allow us to map abundance patterns across the whole range of star forming galaxies in the early universe, providing multiple windows on the nature of the first generations of stars.

As an example of the great relevance of THESEUS in the context of the next generation large facilities (e.g., SKA, CTA, E-ELT, ATHENA), we highlight here the THESEUS synergy with Athena. Two of the primary science goals for Athena are: (1) Locate the missing baryons in the Universe by probing the Warm Hot Intergalactic Medium (the WHIM); this requires about 10 bright GRBs per year. (2) Probe the first generation of stars by finding high redshift GRB; this requires about 5 high-redshift GRBs per year. THESEUS will enable Athena to achieve these goals by greatly increasing the rate of GRBs found per year with good localisations and redshifts and the X-ray band of the THESEUS SXI will find a greater proportion of high-redshift GRBs than previous missions. Hence THESEUS will: (1) localise bright GRBs with sufficient accuracy using the SXI to enable a rapid repointing of the Athena X-IFU for X-ray spectroscopy of the WHIM, and (2) find high-redshift GRBs using the SXI and XGS and accurately localising them using the IRT for redshift determination on-board and on the ground to provide reliable high-redshift targets for Athena. Many of the other transients found by THESEUS, such as tidal disruption events and flaring binaries will also be high-value targetsfor Athena.