General Relativistic Simulations of Accretion Induced Collapse of Neutron Stars to Black Holes

Neutron stars (NSs) in the astrophysical universe are often surrounded by accretion disks. Accretion of matter onto an NS may increase its mass above the maximum value allowed by its equation of state, inducing its collapse to a black hole (BH). Here we study this process for the first time, in three-dimensions, and in full general relativity. By considering three initial NS configurations, each with and without a surrounding disk (of mass ~7% M_{NS}), we investigate the effect of the accretion disk on the dynamics of the collapse and its imprint on both the gravitational wave (GW) and electromagnetic (EM) signals that can be emitted by these sources. We show in particular that, even if the GW signal is similar for the accretion induced collapse (AIC) and the collapse of an NS in vacuum (and detectable only for Galactic sources), the EM counterpart could allow us to discriminate between these two types of events. In fact, our simulations show that, while the collapse of an NS in vacuum leaves no appreciable baryonic matter outside the event horizon, an AIC is followed by a phase of rapid accretion of the surviving disk onto the newly formed BH. The post-collapse accretion rates, on the order of ~10^{-2} M_{sun} s^{-1}, make these events tantalizing candidates as engines of short gamma-ray bursts.

💡 Research Summary

This paper presents the first fully general‑relativistic three‑dimensional simulations of the accretion‑induced collapse (AIC) of a neutron star (NS) into a spinning black hole (BH) when the NS is surrounded by a modest mass torus. The authors construct three rotating NS equilibrium models using the RNS code and, for each, generate a companion torus with a mass of roughly 7 % of the NS mass using the TORERO code. The torus is built assuming a Kerr spacetime; its self‑gravity is neglected, which is justified because the torus mass is a small fraction of the central object. Collapse is triggered by a 0.1 % reduction of the pressure, mimicking the effect of continued accretion that pushes the NS over its maximum stable mass.

The evolution is performed with the Whisky GRMHD code (ideal‑fluid, Γ = 2 EOS) coupled to the Ccatie spacetime solver, employing the Carpet driver for fixed‑mesh refinement (seven refinement levels, finest resolution ≈0.15 km). Symmetry conditions (reflection across the equatorial plane and π‑symmetry across the x = 0 plane) reduce computational cost, but test runs without these symmetries confirm that the dynamics remain essentially axisymmetric. Magnetic fields are omitted, a limitation acknowledged for future work.

Results show a clear dichotomy between the vacuum collapse (no torus) and the torus‑aided collapse. In the vacuum case the NS collapses almost perfectly axisymmetrically; the apparent horizon forms and all baryonic matter is quickly swallowed, leaving essentially zero mass outside the BH. When a torus is present, the BH still forms promptly, but the torus survives the collapse and continues to orbit and accrete onto the BH. The early post‑collapse accretion rate is of order 10⁻² M⊙ s⁻¹, corresponding to a retained torus mass of ~7 % of the original NS mass. The torus also induces small oscillations in the NS density prior to collapse, slightly delaying horizon formation.

Gravitational‑wave emission is dominated by the l = 2, m = 0 mode, reflecting the near‑axisymmetric nature of the collapse. The waveforms for torus‑bearing and torus‑free models are virtually indistinguishable, apart from minor pre‑collapse oscillations in the former. Consequently, with current Advanced LIGO/Virgo detectors such events would be observable only if they occur within the Milky Way, while a third‑generation detector such as the Einstein Telescope could detect them out to ~1 Mpc.

The electromagnetic counterpart is far more distinctive. The surviving torus supplies a sustained mass inflow onto the newly formed BH at ~10⁻² M⊙ s⁻¹. Assuming a modest conversion efficiency of 10 % from accreted mass to jet power, the resulting luminosity would be ≈10⁵¹ erg s⁻¹, comparable to the peak γ‑ray luminosities of short gamma‑ray bursts (sGRBs). If realistic magnetorotational turbulence (α ≈ 0.1) operates, the effective accretion timescale shortens dramatically, potentially raising the accretion rate to ~1 M⊙ s⁻¹ and further enhancing the jet power. Because the vacuum collapse leaves no material for post‑collapse accretion, the presence of a bright, short‑duration γ‑ray signal would be a clear discriminator between AIC and direct NS‑to‑BH collapse.

The study highlights several avenues for future research: inclusion of magnetic fields to assess jet launching mechanisms, exploration of a broader range of torus masses and angular‑momentum profiles, and use of more realistic equations of state. Multi‑messenger observations—gravitational waves combined with prompt γ‑ray, X‑ray, or radio afterglows—will be essential to identify AIC events and to evaluate their contribution to the observed population of short GRBs.