Research

A selection of my works are described below. A full list of my publications is here.

Fast Radio Bursts

• Emission Mechanism

In Lu, Kumar & Zhang (2020), we propose that FRBs are generated by Alfven waves propagating in the neutron star (NS) magnetosphere to large distances where they become charge starved, and charge clumps are accelerated by an electric field and coherently generate curvature emission in the radio band.

The figure to the right shows Alfven waves launched from the magnetic foot points propagate along the field lines to distances much larger than the NS radius where the charge density is too low to sustain the current associated with the wave. As a result of charge starvation, a strong electric field component parallel to the background B-field develops, and charge clumps are accelerated to high Lorentz factors and coherently produce curvature emission in the radio band. The FRB emission is beamed into the solid angle spanned by the orange arrow. The Alfven waves launched far from the poles are trapped by field lines that do not extend to sufficiently large distances, and a pair fireball is formed which emits hard X-rays visible from a large fraction of the sky. This model provides a unified picture for faint bursts like FRB 200428 (Bochenek et al. 2020, CHIME Collaboration 2020) as well as the bright bursts seen at cosmological distances.

• Source Objects


In Lu, Beniamini & Kumar (2022), we propose that the FRB source in a globular cluster of M81 (Kirsten et al. 2022) is a NS with strong B-fields > 10^13 G (likely at the level of magnetars).

The figure to the right shows the joint constraints on the P (spin period) and Pdot (period derivative) of the M81-FRB source object, based on (i) the non-detection of X-rays by Chandra, (ii) the age limit (> 10^4 yr, from the event rate), and (iii) the requirement of a magnetic energy reservoir. The white region is allowed by observations --- the source object is magnetar-like, but somewhat older than Galactic magnetars with an age of ~10^5 yrs. It is still unknown how the old (~10 Gyr) stellar population in the globular cluster can produce such a magnetar-like object. A likely channel is the merger of two massive (~1 Msun) white dwarfs, which did not generate a type-Ia supernova but the merger remnant collapses into a strongly magnetized NS.

• Polarization


In Lu, Kumar & Narayan (2019), we study the propagation of FRB waves through the magnetosphere of a strongly magnetized NS. It is shown that the observed electric (polarization) vector is determined by the B-field orientation at the polarization-limiting radius located far from the NS surface where the dipolar field dominates (as shown in the right figure), independent of magnetic configuration of the emission region which is much closer by. This is because, as the waves propagate in an inhomogeneous plasma, the electric vector undergoes adiabatic walking, which keeps its X-mode eigenmode nature --- the E-field stays perpendicular to the k-B plane (k and B are the wavevector and background B-field). Therefore, the polarization angles of repeating bursts should be periodic as modulated by the NS spin, even if the burst occurrence may not be periodic. Such a periodicity should show up once we have a sufficient number of bright bursts from a given repeater.

Tidal Disruption Events

Main-sequence stars are shredded by tidal forces when they get sufficiently close to a supermassive black hole (BH). After disruption, the star is stretched into a long thin stream (of aspect ratio ~1000) in highly eccentric orbits (1-e ~ 0.01) which undergo general relativistic precession. Such precession causes the stream to self-intersect. Then, the shocked gas expanding from the intersection point will eventually form an accretion disk, which powers multi-wavelength emission. A few dozen of these tidal disruption event (TDE) candidates have been found in recent surveys designed to find variable sources in the optical/UV and soft X-rays.

Global numerical simulation of the hydrodynamic evolution of the thin fallback stream is prohibitively expensive. In Lu & Bonnerot (2020), we made a key simplification of the problem by first calculating the location where the stream self-intersects according to general relativistic geodesic motion and then performing hydrodynamic simulations of the stream collision process in a local region near the intersection point. We found that the self-crossing shock redistributes energy and angular momentum of the gas in the fall-back stream. For sufficiently deeply penetrating orbits (pericenter radius Rp <~ 15Rg), the shocks at the intersection are able to unbind a large fraction (up to 50%) of the fallback gas. We call the unbound gas the “collision-induced outflow” or CIO. We found that the CIO covers a large fraction of the sky viewed from the BH and is able to reprocess the hard (EUV/soft X-ray) photons from the disk into the optical band. This provides an explanation of the bright optical emission seen in many TDEs. Viewing angle effects then cause the X-ray luminosity to vary by orders of magnitude from one event to another.

A schematic picture of the stream-collision process is show to the right.

In Bonnerot, Lu & Hopkins (2021), we conduct the first radiation hydrodynamics simulation of a realistic TDE (a 0.5Msun main-sequence star disrupted by a 2.5x10^6 Msun BH), using the code GIZMO. The multi-panel figure to the right shows the time evolution of the radiation temperature defined based on the local radiation energy density (U_rad = a T_r^4) inside the equatorial plane. The time unit is t_{min} = 40d --- the minimum period of the bound stellar debris. In the t=0 panel (and all others), the BH is represented by a white circle in the lower half of the figure, and the circle above that is the position of the self-crossing shock where we inject gas into the computational domain. We found rapid formation of an accretion disk after the onset of stream self-crossing. Moreover, the region near the self-crossing point has very little radiation as a result of strong shielding, so the gas there can maintain a low ionization fraction and hence high opacity --- this provides the necessary ingredients for reprocessing of EUV/X-ray photons into the optical band.

Formation of GW190814 (most likely a BBH merger)

The LIGO-Virgo Collaboration recently reported a puzzling event, GW190814, with component masses of 23 and 2.6 Msun --- a surprising mass ratio of q ~ 0.1. In Lu, Beniamini & Bonnerot (2021), we propose a 2nd-generation merger scenario where the 2.6Msun object was from a previous binary neutron star (BNS) coalescence and the remnant was able to merge again with the 23 Msun BH tertiary. This occurs when the remnant (most likely a low-mass BH) receives a kick of about 100 km/s in the direction of the "loss cone" shaped like a flying saucer, as shown in the figure to the right. This model was motivated by the relatively small rate (1 to 23 per Gpc^3 per yr) inferred from GW190814 and the secondary mass being close to the total masses of known BNS systems. We show that about 1% of the BNS coalescence occurring in triple systems should give rise to 2nd-generation mergers, provided that a massive BH tertiary is located at a separation less than a few AU. Since the total BNS coalescence rate is of the order 10^3 per Gpc^3 per yr, this model requires that at least 10% BNS mergers occur in triple systems. Since the typical delay time for the 2nd-generation merger is about a Hubble time, these BNS mergers in triples occurred in the distant past when the Universe was less metal-enriched. Low-metallicity (< 0.1 solar) is also favored for the formation of the 23 Msun BH. This model has a number of testable predictions, one of which is that the secondary of GW190814-like events should have dimensionless spin of about 0.7. The other one is that future extreme mass-ratio (q ~ 0.1) events should also have seconary mass in the narrow range of 2.5 to 2.8 Msun (based on Galactic BNS systems).

Neutron Star Mergers and Gamma-Ray Bursts

The overall picture of GW170817/AT2017gfo, as inferred from its gravitational wave (GW) emission, gamma-ray flash, kilonova, and broadband afterglow from a relativistic off-axis jet, confirmed the conjecture that neutron star (NS) mergers are a significant contributor to the r-process nucleosynthesis as well as the sources of short gamma-ray bursts (GRBs).

In Lu & Quataert (2022), we study the long-term (>> 10 sec) evolution of the accretion disk in NS mergers (previous works focused on short-term evolution within the first few seconds). It is shown that the radioactive heating by r-process nuclei eventually exceeds the disk's binding energy roughly 10^2 sec after the merger. This causes the disk to rapidly evaporate leaving nothing behind. The figure to the right schematically shows the evolution of the disk mass with time since merger. At t ~ 1 sec, the disk loses most of its mass due to nuclear recombination forming alpha-particles. At t ~ 10^2 sec, radioactive decay causes another episode of mass loss from the disk.

We suggest that the extended emission (EE) observed in many short GRBs is powered by late-time disk accretion and that the steep flux decline seen at the end of the EE is due to rapid disk evaporation (which shuts off the jet).

We also suggest that the jets from NS mergers have two components --- a short-duration narrow one corresponding to the prompt gamma-ray emission and a long-lasting wide component producing the EE. Observers at different viewing angles (marked by 1, 2, 3 in the figure to the right) would see different phenomena. This leads to a prediction that "orphan EE" (without the prompt gamma-rays, from viewing angle "2") may be a promising electromagnetic counterpart for NS mergers observable by future wide-field X-ray surveys.

The long-lived disk produces a slow ejecta component (~0.01 c). We predict that future JWST near-IR spectroscopy of nearby (<~ 100 Mpc) NS mergers will detect narrow line features a few weeks after the merger, which provides a powerful probe of the atomic species formed in these events.