Observed supermassive black holes in the early universe have several proposed formation channels, in part because most of these channels are difficult to probe. One of the more promising channels, the direct collapse of a supermassive star, has several possible probes including the explosion of a helium-core supermassive star triggered by a general relativistic instability. We develop a straightforward method for evaluating the general relativistic radial instability without simplifying assumptions and apply it to population III supermassive stars taken from a post Newtonian stellar evolution code. This method is more accurate than previous determinations and it finds that the instability occurs earlier in the evolutionary life of the star. Using the results of the stability analysis, we perform 1D general relativistic hydrodynamical simulations and we find two general relativistic instability supernovae fueled by alpha capture reactions as well as several lower mass pulsations, analogous to the puslational pair instability process. The mass range for the events (2.6-3.0 × 104 M⊙) is lower than had been suggested by previous works (5.5 × 104 M⊙) because the instability occurs earlier in the star’s evolution. The explosion may be visible to, among others, JWST, while the discovery of the pulsations opens up additional possibilities for observation.

Since the discovery of GW190521, several proposals have been put forward to explain the formation of a black hole (BH) in the mass gap caused by (pulsational) pair-instability (PPI), M = 65–130 M ⊙. We calculate the mass ejection of Population III stars by the PPI process using a stellar evolution and hydrodynamical code. If a relatively small, but reasonable, value is adopted for the overshooting parameter, the stars do not become red supergiants during the PPI phase. We show that in this case most of the hydrogen envelope remains after the mass ejection by PPI. We find that the BH mass could be at most around 110 M ⊙ below the mass range of pair-instability supernovae.

The assembly of supermassive black holes poses a challenge primarily because of observed quasars at high redshift, but additionally because of the current lack of observations of intermediate mass black holes. One plausible scenario for creating supermassive black holes is direct collapse triggered by the merger of two gas rich galaxies. This scenario allows the creation of supermassive stars with up to solar metallicity, where the enhanced metallicity is enabled by extremely rapid accretion. We investigate the behavior of metal enriched supermassive protostars which collapse due to the general relativistic radial instability. These stars are rich in both hydrogen and metals and thus may explode due to the CNO cycle (carbon-nitrogen-oxygen) and the rp process (rapid proton capture). We perform a suite of 1D general relativistic hydrodynamical simulations coupled to a 153 isotope nuclear network with the effects of neutrino cooling. We determine the mass and metallicity ranges for an explosion. We then post process using a 514 isotope network which captures the full rp process. We present nucleosynthesis and lightcurves for selected models. These events are characterized by enhanced nitrogen, suppressed light elements (8 ≥ A ≥ 14), and low mass p nuclides and they are visible to JWST and other near infrared surveys as decades-long transients. Finally, we provide an estimate for the number of currently ongoing explosions in the Universe.

We investigate the possibility of a supernova in supermassive (5 × 104 M⊙) population III stars induced by a general relativistic instability occurring in the helium burning phase. This explosion could occur via rapid helium burning during an early contraction of the isentropic core. Such an explosion would be visible to future telescopes and could disrupt the proposed direct collapse formation channel for early Universe supermassive black holes. We simulate first the stellar evolution from hydrogen burning using a 1D stellar evolution code with a post-Newtonian approximation; at the point of dynamical collapse, we switch to a 1D (general relativistic) hydrodynamic code with the Misner-Sharpe metric. In opposition to a previous study, we do not find an explosion in the non-rotating case, although our model is close to exploding for a similar mass to the explosion in the previous study. When we include slow rotation, we find one exploding model, and we conclude that there likely exist additional exploding models, though they may be rare.