Impact of Neutrino Flavor Conversions on Neutron Star Merger Dynamics, Ejecta, Nucleosynthesis, and Multi-Messenger Signals

TL;DR

This work investigates how neutrino flavor conversions, including fast flavor instabilities and potential beyond-Standard-Model effects, influence binary neutron star mergers by embedding a BGK-based flavor relaxation operator into dynamical GRHD with truncated moment neutrino transport. It implements MB and MX flavor-equilibrium prescriptions alongside an ELN-based fast-instability trigger, across two equations of state (DD2, SFHo) and different remnant lifetimes. The results show that flavor conversions push ejecta toward more neutron-rich $Y_e$ regimes and substantially boost heavy-element yields, particularly in low-density, near-equatorial outflows, while also modifying neutrino and gravitational-wave luminosities; the effects depend strongly on where mixing occurs and on the remnant type. Fast flavor instabilities persist in global simulations and the subgrid relaxation timescales interact with thermodynamic equilibration to shape ejecta composition, underscoring the need for more comprehensive transport and instability modeling to predict observables. Overall, the study highlights the potential significance of flavor conversions in neutron star mergers and provides a roadmap for improving the theoretical treatment of neutrino flavor dynamics in multi-messenger astrophysics.

Abstract

We present numerical relativity simulations of binary neutron star mergers incorporating neutrino flavor transformations triggered by fast flavor instability, quantum many-body effects, or potential beyond standard model physics. In both long-lived and short-lived remnant scenarios, neutrino flavor conversions modify species-dependent neutrino luminosities and mean energies, and drive the matter towards more neutron rich conditions. They produce up to $300\%$ more neutron rich ejecta and significantly boost the r-process yields, especially in low-density, near-equatorial outflows. We identify regions unstable to fast flavor instabilities and find that these instabilities persist despite flavor conversions. We further test the sensitivity to the equilibration timescale of the flavor conversions, finding that slower flavor conversions can interact with thermodynamic equilibration, and increase the neutron richness of the ejecta. Flavor conversions may also contribute to stronger gravitational wave and neutrino emissions, pointing to a correlation between neutrino transport and merger dynamics. These results highlight the potential impact of flavor conversions while motivating future work to improve on theoretical understanding of flavor instabilities in global simulations.

Figures (13)

• Figure 1: Simulation snapshots in the equatorial (xy) and meridional (xz) planes for the SFHo-11 and the SFHo-NM models at $\sim$5, 10 and 35 ms after merger on the left (right) four panels, respectively. The top-row panels of each plot show electron fractions (upper half) and rest mass densities (lower half). The second-row panels of each plot display electron neutrino number densities (upper half) and heavy-lepton neutrino number densities (lower half). The contour lines denote where the density is at $10^{11}$, $10^{12}$ and $10^{13}$$\mathrm{g/cm}^3$. The $\sim$5 ms post-merger snapshots correspond to dynamics just before collapse to BHs, while in $\sim$10 ms post-merger snapshots the collapses have happened. The region with lapse function $\alpha \leq 0.15$ (black circles) denotes the approximate location of the BH apparent horizon. Finally, the $\sim$35 ms post-merger snapshots depicts the states of the system at which the remnant BHs enter the accretion phases. Due to the uncertainties of the collapse times, SFHo-11 and SFHo-NM simulations appear to have very different patterns of density distributions after the collapses, which are not due to the neutrino flavor conversion effects. In terms of the composition of the disk and ejecta, we find similar results as in Qiu:2025kgy. In different stages of the remnant evolutions, the disk in the SFHo-11 mixing model is more neutron rich than that of the SFHo-NM model. Due to flavor conversion effects transforming electron (anti-)neutrinos to heavy-lepton neutrinos, we also find that electron neutrinos are more abundant in the SFHo-NM than those of SFHo-11, and vice versa for heavy-lepton neutrinos.
• Figure 2: Post-merger total neutrino luminosity, and luminosity for $\nu_e$, $\bar{\nu}_e$, and $\nu_x$ ($= \nu_\mu + \nu_\tau$), respectively, from left to right panels in each row. Note that $\nu_x$ luminosity does not contain the contributions from heavy-lepton anti-neutrinos. The upper row shows results of simulations considering DD2 EoS and the lower row shows those of SFHo EoS. We mark the collapse times for the SFHo runs with dashed lines. Neutrino flavor conversions $\nu_e, \bar{\nu}_e \rightarrow \nu_x, \bar{\nu}_x$ generically increase the $\nu_x$ luminosities, while decreasing the electron (anti-)neutrino luminosities. Among the three mixing models with DD2 EoS, the DD2-13 simulation has the highest net luminosities compared to the DD2-11 and the DD2-FFI simulations. This is because it includes additional flavor conversion effects in the inner disk regions between density of $10^{11}$ and $10^{13}\mathrm{g/cm}^3$ compared to the DD2-11, and does not filter out the FFI-stable regions as compared to the DD2-FFI. For the SFHo runs, we do not observe a clear hierarchy, due to differences in the collapse times and remnant disk masses. At late times, the SFHo-11 has higher luminosities for $\nu_x$ while the other mixing models are generally comparable. We note that such differences are likely not results of the neutrino flavor conversion effects, though.
• Figure 3: Post-merger neutrino mean energy for $\nu_e$, $\bar{\nu}_e$, and $\nu_x$, respectively, from left to right panels in each row. The upper row shows results of simulations considering DD2 EoS and the lower row shows those of SFHo EoS. We mark the collapse times for the SFHo runs with dashed lines. For the no-mixing models, heavy-lepton neutrinos generally have higher mean energies, indicating their smaller decoupling radii compared to the electron (anti)neutrinos. For the neutrino mixing models DD2(SFHo)-11, DD2(SFHo)-13 and DD2(SFHo)-FFI, the mean energies of all flavors are comparable, because flavor conversions can exchange energy of different neutrinos. The mean energy of $\nu_e$ seems a bit smaller, but it could be due to the fact that high energy $\nu_e$ are absorbed more easily by the neutron rich material and the spectrum tends to become a bit softer. The mixing models mostly have higher $\nu_e$ and $\bar{\nu}_e$ mean energies and lower $\nu_x$ mean energy compared to the no-mixing model.
• Figure 4: Upper (lower) left panels: histograms of the electron fraction distributions of the ejecta for the four DD2 (SFHo) models. Upper (lower) right panel: the relative abundances of nuclei of mass number $A$ formed in the ejecta of four DD2 (SFHo) models. The relative neutron richness of the ejecta follows the order: DD2-11, DD2-13, DD2-FFI, and DD2-NM. Notably, the ordering between the DD2(SFHo)-11 and DD2(SFHo)-13 is reversed for the two different EoS's. On the right plots, we normalize the yields at $A=85$. The yields relative to $A=85$ for lanthanides and heavy elements are one order of magnitude higher in the DD2-11 simulation than the DD2-NM simulation. Such difference shrinks in the SFHo simulations, but is still up to a factor of five between the SFHo-11 and SFHo-NM heavy element yields.
• Figure 5: Upper (lower) panels: equilibrium electron fractions for the post-merger ejecta in the four DD2 (SFHo) models. We mark the collapse times for the SFHo runs with dashed lines. The curves are smoothed using convolution with a 0.5 ms square window. The equilibrium $Y_e$ is generally lower in the no-mixing simulation, i.e., the DD2(SFHo)-NM, compared to the mixing models, for most of the time in the DD2 simulations and the time before collapses in the SFHo simulations.