Resumen de Multi-dimensional simulations of mixing in classical novae
Jordi Casanova Bustamante
Classical nova explosions are stellar explosions that take place in close binary systems with an energy release only exceeded by gamma-ray bursts and supernova explosions. Matter from the white dwarf flows through the inner lagrangian point and spirals in towards the white dwarf for about 10^4-10^5 years, forming an accretion disk around it. Ultimately, part of this hydrogen-rich matter piles-up on top of the compact object and becomes partially degenerate due to the high densities attained. Consequently, temperature is allowed to rise, but the envelope does not experience any expansion. Actually, this is the key mechanism that controls the subsequent phases and powers a thermonuclear runaway, which is followed by an ejection of part of the accreted envelope. The ejecta are enriched with the products from the nuclear processes, presenting a final metallicity much above solar. This model, introduced in the early 70s, is a solid theory that can account for the gross scenario of nova explosions. Nevertheless, the theory relies on the fact that a mixing episode with matter from the white dwarf core has to take place at the core-envelope interface to successfully account for the high metallicities inferred from observations. During the past 40 years, theoreticians have performed many one-dimensional simulations, which can reproduce the abundances in the ejecta and other important observational properties. However, these calculations performed in spherical symmetry cannot study the mixing process, since they exclude a suite of very important multi-dimensional effects, such as convection. Therefore, multi-dimensional calculations are required to shed light into the mixing episode. In this thesis we have performed two- and three- dimensional simulations of CO novae to study the mixing mechanisms operating at the core-envelope interface, how convection sets in and how the deflagration spreads over the domain, by means of the Eulerian, parallelized, hydrodynamical FLASH code. The two-dimensional results show how convection sets in at the innermost envelope layers, after the appearance of temperature fluctuations that arise from the interface. Convection, in turn, powers the formation of kelvin-Helmholtz instabilities, which efficiently dredge-up 12C from the core and carry it into the envelope, reproducing correctly the high metallicity found in the ejecta. This result solves the controversy generated by the two existing two-dimensional calculations up-to-date. We have also realized a sensitivity study to analyze the impact of some initial parameters, such as the temperature perturbation, resolution of the simulations and the size of the computational domain. The results point out that these parameters have a negligible impact on the degree of mixing and, therefore, the calculations are not affected by numerical artifacts. Although two-dimensional calculations can quantitatively reproduce the mixing episode, they cannot describe correctly the convective pattern due to conservation of vorticity, which translates into recombination of the convective cells. Therefore, we have extended the work to three dimensions and performed the first three-dimesional model of mixing in classical novae up-to-date. These calculations can successfully reproduce the intermittency present in turbulent convection, with an energy cascade into smaller scales which clearly fulfills the Kolmogorov theory, while the thermonuclear runaway continues propagating with almost spherical symmetry. Mixing proceeds through the filamentary structure powered by robust kelvin-Helmholtz instabilitites that arise from the interface, resulting in a CNO enhancement which agrees with observations. This convective profile also generates density contrasts that could be the origin of the inhomogeneous distribution of chemical species.
