Multi-dimensional Models of Mixed H/He Flames in X-ray Bursts

Eric T. Johnson and Michael Zingale · 2025

Read Multi-dimensional Models of Mixed H/He Flames in X-ray Bursts by Eric T. Johnson and Michael Zingale (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-13879.txt.

Introduction

Eric Johnson and Michael Zingale, two authors writing in 2025, simulate flames on neutron stars. X-ray bursts are runaway burning in a layer of hydrogen and helium on such a star. How the mixed flame spreads, and how hot it runs, depends on the starting mix.

Two-dimensional fluid simulations varied the hydrogen share of the layer. More hydrogen made the flame weaker and slower. Whether a flame formed at all, and whether the base ran away, hinged on the starting model and the reaction network. A second flame lit late in the runs, by a route that depended on the mix.

Catalog record

CREATOREric T. Johnson and Michael Zingale
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-13879

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Multi-dimensional Models of Mixed H/He Flames in X-ray Bursts Eric T. Johnson and Michael Zingale Abstract We investigate the properties of mixed H/He flames in X-ray bursts using 2D hydrodynamic simulations. We find that as the initial hydrogen abundance of the atmosphere increases, the flame is less energetic and propagates slower. The simulation outcome, whether a flame forms and whether there’s runaway burning at the base of the atmosphere, is very sensitive to the initial model and the nuclear reaction network used. We also see that at late times a secondary flame ignites, with the ignition mechanism dependent on the composition. \journalinfo The Open Journal of Astrophysics \submitted Version May 20, 2025 keywords X-ray bursts, nucleosynthesis, flames, hydrodynamics 1 Introduction Type I X-ray bursts (XRBs) are transient outbursts from a low-mass X-ray binary system, triggered by thermonuclear burning on the surface of a neutron star ( Galloway & Keek, 2021 ) . Depending on the accretion rate, the built-up fuel may be pure helium, or a mix of hydrogen and helium with varying proportions. One-dimensional numerical simulations can accurately reproduce burst durations and recurrence times, light curves, and nucleosynthesis yields ( Woosley et al., 2004 ; Heger et al., 2007 ; Johnston et al., 2018 ) . However, some features such as burst oscillations and light curve rise times depend on the initial flame dynamics, which are inherently multidimensional ( Watts, 2012 ) . As computational resources have become more powerful, more attention has been given to 2D and 3D simulations. Spitkovsky et al. (2002) used the shallow-water approximation to model flame propagation, and showed that the Coriolis force is important for flame dynamics. However, this approximation does not model the vertical structure of the flame, so detailed investigations of flame dynamics and nucleosynthesis are not possible. Cavecchi et al. (2013) and Cavecchi et al. (2015) performed the first vertically resolved flame simulations, using a method that enforces vertical hydrostatic equilibrium. These simulations used pure helium and showed an enhancement of the burning rate from the multi-dimensional flow pattern. Follow-on full three-dimensional simulations find enhancement of the flame speed by a factor of 10 due to instabilities in the flame front ( Cavecchi & Spitkovsky, 2019 ) . More recently, the formation of a hotspot, which may lead to the ignition of a burning front, was explored using 2D heat transport simulation by Goodwin et al. (2021) . Previous work by our group has focused on pure helium flames, using the full compressible Euler equations modeled by the Castro simulation code ( Almgren et al., 2010 ; Almgren et al., 2020 ) . Eiden et al. (2020) introduced the simulation setup using a 2D axisymmetric domain, and investigated boosted flames by artificially increasing the reaction rates and thermal conductivity. Harpole et al. (2021) looked at the effects of rotation rate and thermal structure, using more realistic simulations without boosting. These calculations show the formation of a steady flame and explored how the flame ignition depends on the rotation rate. These resolved simulations were enabled by porting our code to run on GPUs ( Katz et al., 2020 ) . These calculations also established the resolution requirements of {\sim}$20\text{\,}\mathrm{cm}\text{/}\mathrm{}$ . Zingale et al. (2023) compared the 2D simulations to a full 3D simulation with the same parameters, and found that the two matched reasonably well, validating the use of axisymmetric geometry to explore the burning in the absence of turbulence. Finally, Chen et al. (2023) explored the sensitivity of the flame to different reaction networks and found that including proton-capture reactions to bypass the slower 12 C ( \alpha , \gamma ) 16 O reaction caused a burst of acceleration at early times. In this paper, we explore flames propagating through a mixed H/He atmosphere. The addition of hydrogen enables addition nuclear pathways, creating a much more complex nucleosynthesis process than the pure He flames done previously. In section 2 we describe our numerical method, in section 3 we explore the size of the reaction network, in section 4 we discuss our results, with an emphasis on the nucleosynthesis, and finally in section 5 , we conclude. 2 Numerical Methodology All simulations are done with the Castro simulation code, which solves the compressible Euler equations together with thermal diffusion, reactions, and rotation (just the Coriolis force for this application). The equations are supplemented with a general stellar equation of state from Timmes & Swesty (2000) . We use the same initial model formulation and simulation parameters as Zingale et al. (2023) . For reference, we use a peak temperature of 1.4\text{\times}{10}^{9}\text{\,}\mathrm{K} in the hot region behind the flame and 2\text{\times}{10}^{8}\text{\,}\mathrm{K} in the cool region ahead of the flame, a neutron star crust temperature of 2\text{\times}{10}^{8}\text{\,}\mathrm{K} , a minimum temperature of 8\text{\times}{10}^{6}\text{\,}\mathrm{K} at the top of the isentropic region, and a density of 3.43\text{\times}{10}^{6}\text{\,}\mathrm{g}\text{/}{\mathrm{cm}}^{3} at the interface between the neutron star and the atmosphere. The main changes are the initial model composition and the nuclear reaction networks used. The atmosphere is composed of uniformly mixed 1 H and 4 He , along with 1% of 14 O and 15 O distributed according to their \beta^{+} -decay lifetimes to help start CNO burning. Recently, Guichandut et al. (2024) explored a similar model starting from a layered H/He atmosphere, and found the convective flows could mix the composition efficiently. Here, we assume that this mixing has already occurred before the flame ignition. We use a two-dimensional axisymmetric domain of $196\,608\text{\,}\mathrm{cm}$\times$24\,576\text{\,}\mathrm{cm}$ with a coarse grid of 768\times 192 zones and 2 levels of refinement (first by a factor of 4 then by a factor of 2). This gives a fine-grid resolution of 32\text{\,}\mathrm{cm} in the radial direction and 16\text{\,}\mathrm{cm} in the vertical direction (the same resolution as in Zingale et al. 2023 ). The refined grids are static, with the finest resolution (level 2) present below z=$4096\text{\,}\mathrm{cm}$ and the second finest (level 1) below z=$5120\text{\,}\mathrm{cm}$ . This puts the initial atmosphere completely at the finest resolution. We use a reflecting boundary at bottom of the domain, along with the well-balanced pressure reconstruction described in Zingale (2024) . At the top of the domain, we include an isothermal buffer region into which the flame can expand. We also apply a sponge source term which suppresses the velocities in low-density regions ( Eiden et al., 2020 , described in) , as we are not concerned with this material. Table 1: Overview of the reaction networks used name nuclei rates notes rprox 10 14 based on Appendix C of Wallace & Woosley (1981) , more details in Malone et al. (2014) aprox19 19 78 described in Weaver et al. (1978) CNO_extras 21 67 based on MESA ( Paxton et al., 2011 ) , only goes up to 24 Mg CNO_He_burn 33 113 CNO_extras + subch_simple from Chen et al. (2023) Figure 1: Structure of the CNO_He_burn network (links to 1 H and 4 He are not drawn). The light gray nodes and dotted lines represent the ( \alpha , p)(p, \gamma ) approximation, with those nuclei approximated out of the network. We explore several different reaction networks, summarized in Table 1 , with the largest two ( CNO_extras and CNO_He_burn ) created using pynucastro ( Smith et al.,

Read the complete text