Non-spherical core collapse supernovae

Abstract

Abstract.We have performed two-dimensional simulations of core collapse supernovae that encompass shock revival by neutrino heating, neutrino-driven convection, explosive nucleosynthesis, the growth of Rayleigh-Taylor instabilities, and the propagation of newly formed metal clumps through the exploding star. A simulation of a type II explosion in a 15 M blue supergiant progenitor is presented, that confirms our earlier type II models and extends their validity to times as late as 5.5 hours after core bounce. We also study a type Ib-like explosion, by simply removing the hydrogen envelope of the progenitor model. This allows for a first comparison of type II and type Ib evolution. We present evidence that the hydrodynamics of core collapse supernovae beyond shock revival differs markedly from the results of simulations that have followed the Rayleigh-Taylor mixing starting from ad hoc energy deposition schemes to initiate the explosion. We find iron group elements to be synthesized in an anisotropic, dense, low-entropy shell that expands with velocities of ∼17 000 km s −1 shortly after shock revival. The growth of Rayleigh-Taylor instabilities at the Si/O and (C+O)/He composition interfaces of the progenitor, seeded by the flow-structures resulting from neutrino-driven convection, leads to a fragmentation of this shell into metal-rich "clumps". This fragmentation starts already ∼20 s after core bounce and is complete within the first few minutes of the explosion. During this time the clumps are slowed down by drag, and by the positive pressure gradient in the unstable layers. However, at t ≈ 300 s they decouple from the flow and start to propagate ballistically and subsonically through the He core, with the maximum velocities of metals remaining constant at ∼3500−5500 km s −1 . This early "clump decoupling" leads to significantly higher 56 Ni velocities at t = 300 s than in one-dimensional models of the explosion, demonstrating that multi-dimensional effects which are at work within the first minutes, and which have been neglected in previous studies (especially in those which dealt with the mixing in type II supernovae), are crucial. Despite comparably high initial maximum nickel velocities in both our type II and our type Ib-like model, we find that there are large differences in the final maximum nickel velocities between both cases. In the "type Ib" model the maximum velocities of metals remain frozen in at ∼3500−5500 km s −1 for t ≥ 300 s, while in the type II model they drop significantly for t > 1500 s. In the latter case, the massive hydrogen envelope of the progenitor forces the supernova shock to slow down strongly, leaving behind a reverse shock and a dense helium shell (or "wall") below the He/H interface. After penetrating into this dense material the metal-rich clumps possess supersonic speeds, before they are slowed down by drag forces to ∼1200 km s −1 at a time of 20 000 s post-bounce. While, due to this deceleration, the maximum velocities of iron-group elements in SN 1987 A cannot be reproduced ...

Nucleosynthesis-relevant conditions in neutrino-driven supernova outflows

“…Neutrino luminosities at the boundary are chosen such that the explosion energy is around 1.5 B. A detailed description of the numerical method, initial models, and boundary treatment can be found in Paper I, Scheck et al (2006), and Kifonidis et al (2003).…”

Simulation of the growth of the 3D Rayleigh-Taylor instability in supernova remnants using an expanding reference frame

“…The result can strongly deform the shock surface and form ripples or bubble-like regions of local subdensity. Local changes to the solver were proposed (Kifonidis et al 2003), based on an algorithm for the detection of the shock. Reconstruction of the interface values of the state variables is performed, but in the cells detected by that algorithm a more diffusive solver is used, such as HLL, HLLC or Lax-Friedrichs (Toro 1999), while in all the remaining regions a conventional Riemann exact solver or an approximate Roe solver can be used.…”

“…In the case of supernova core-collapse simulations, a possible solution has been to compute the intercell flux in the zones of strong shock with an Einfeldt solver (Quirk 1994(Quirk , 1997, while all the other grids in the computational box are treated with a less diffusive Riemann solver (Kifonidis et al 2003). We preferred not to introduce a solver-switching algorithm to avoid introducing other spurious effects possibly arising from the matching of the intercell fluxes.…”

An axis-free overset grid in spherical polar coordinates for simulating 3D self-gravitating flows

“…In addition to the severe restriction of the time step size, boundary conditions that have to be imposed at the symmetry axis θ ∈ [0, π] flaw the simulations near the axis by causing undesired numerical artifacts in 2D axisymmetric simulations, as e.g., jet-like flow features Present address: Max-Planck-Institut für Plasmaphysik, Boltzmannstraße 2, 85748 Garching, Germany. (Kifonidis et al 2003). In 3D simulations, the axis represents a coordinate singularity that almost unavoidably will leave its mark on the flow near or across the axis.…”