Host galaxies of long gamma-ray bursts in the Millennium Simulation

(abridged) In this work, we investigate the nature of the host galaxies of long Gamma-Ray bursts (LGRBs) using a galaxy catalogue constructed from the Millennium Simulation. We developed an LGRB synthetic model based on the hypothesis that these events originate at the end of the life of massive stars following the collapsar model, with the possibility of including a constraint on the metallicity of the progenitor star. A complete observability pipeline was designed to calculate a probability estimation for a galaxy to be observationally identified as a host for LGRBs detected by present observational facilities. This new tool allows us to build an observable host galaxy catalogue which is required to reproduce the current stellar mass distribution of observed hosts. This observability pipeline predicts that the minimum mass for the progenitor stars should be ~75 solar masses in order to be able to reproduce BATSE observations. Systems in our observable catalogue are able to reproduce the observed properties of host galaxies, namely stellar masses, colours, luminosity, star formation activity and metallicities as a function of redshift. At z>2, our model predicts that the observable host galaxies would be very similar to the global galaxy population. We found that ~88 per cent of the observable host galaxies with mean gas metallicity lower than 0.6 solar have stellar masses in the range 10^8.5-10^10.3 solar masses in excellent agreement with observations. Interestingly, in our model observable host galaxies remain mainly within this mass range regardless of redshift, since lower stellar mass systems would have a low probability of being observed while more massive ones would be too metal-rich. Observable host galaxies are predicted to preferentially inhabit dark matter haloes in the range 10^11-10^11.5 solar masses, with a weak dependence on redshift.

💡 Research Summary

In this paper the authors investigate the nature of host galaxies of long gamma‑ray bursts (LGRBs) by coupling a semi‑analytic galaxy catalogue derived from the Millennium Simulation with a synthetic LGRB production model. Two progenitor scenarios are explored, both based on the collapsar model. Scenario I assumes that all stars above a minimum mass (m_min) can produce an LGRB, without any metallicity restriction, thus treating LGRBs as unbiased tracers of star formation. Scenario II adds a metallicity cut‑off Z_C, allowing only massive stars in galaxies whose cold‑gas metallicity is below Z_C to become LGRB progenitors. Three values of Z_C (0.1, 0.3, 0.6 Z⊙) are examined (II.1, II.2, II.3).

The authors construct an “observability pipeline” to translate intrinsic LGRB rates into rates that would be detected by real instruments, specifically the BATSE off‑line search (Stern et al. 2001). BATSE detected 3 475 long bursts over 6.37 yr of live time, corresponding to a full‑sky rate of ≈ 814 yr⁻¹ above its detection threshold. By adjusting the sole free parameter m_min, the authors match the simulated observable LGRB rate to this BATSE rate. The best fit is obtained for m_min ≈ 75 M⊙, indicating that only the most massive stars can account for the observed LGRB frequency.

Using the De Lucia & Blaizot (2007) semi‑analytic catalogue, the authors extract for each galaxy its stellar mass, star‑formation rate (SFR), cold‑gas metallicity, colours, luminosities, and host halo mass. The cold‑gas metallicity is taken as a proxy for the metallicity of the progenitor stars. Type 2 satellite galaxies are excluded from the host‑galaxy analysis to avoid artefacts, but they are retained for environmental studies.

Key results for the observable host population are:

1. Stellar mass distribution – 88 % of observable hosts have stellar masses between 10⁸·⁵ and 10¹⁰·³ M⊙. This range reproduces the observed host‑galaxy mass distribution (Sav 2009) and remains essentially unchanged with redshift. Low‑mass galaxies have a low probability of being observed because their SFRs are modest, while high‑mass galaxies are typically too metal‑rich to satisfy the metallicity cut‑off.

2. Metallicity – When a metallicity cut‑off of Z_C = 0.6 Z⊙ is applied (scenario II.3), 88 % of observable hosts have mean cold‑gas metallicities below this threshold, in excellent agreement with spectroscopic measurements of real hosts.

3. Host halo mass – Observable hosts preferentially reside in dark‑matter haloes of 10¹¹–10¹¹·⁵ M⊙, with only a weak dependence on redshift. This halo mass range corresponds to the regime where gas cooling and star formation are efficient but feedback does not suppress star formation completely.

4. Colours, luminosities and SFR – Hosts are generally blue, luminous in the rest‑frame UV, and exhibit high specific SFRs, consistent with the picture that LGRBs trace recent massive star formation. At redshifts z > 2 the predicted host properties converge toward those of the overall galaxy population, implying that LGRBs become unbiased tracers of star formation in the early Universe.

5. Environment – Hosts are found in slightly overdense regions compared to a random galaxy sample. At high redshift the probability of a host having a close companion is modestly enhanced, echoing observational claims of a non‑negligible fraction of interacting host galaxies.

6. Impact of observability bias – Without the observability pipeline, the raw simulation would predict many low‑mass, low‑metallicity hosts that are not seen in current surveys. Incorporating detection probabilities reconciles the model with the observed host sample, demonstrating the necessity of accounting for selection effects.

Overall, the study shows that (i) a minimum progenitor mass of ≈ 75 M⊙ is required to reproduce the BATSE LGRB rate, (ii) metallicity constraints naturally limit observable hosts to a narrow stellar‑mass band, (iii) host galaxies occupy a characteristic halo‑mass range that is largely redshift‑independent, and (iv) at z > 2 LGRB hosts are statistically indistinguishable from the general galaxy population, supporting their use as probes of star formation in the early Universe. The authors also provide a methodological framework—combining large‑scale cosmological simulations with a realistic observability model—that can be applied to future GRB surveys (e.g., Swift, SVOM) to refine our understanding of GRB progenitors and their cosmological applications.