Filling The Pockets: The Spherical Nature of 3D Deflagration in Thermonuclear Supernovae

We investigate thermonuclear explosions within the delayed detonation framework. While spherical delayed detonation models generally reproduce key observational features, a fundamental inconsistency emerges in three dimensions: 3D hydrodynamic simulations exhibit insufficient white dwarf expansion during the deflagration phase. We identify the early deflagration stage, when the burning is dominated by the laminar speed, as a critical phase and explore potential solutions using three dimensional magnetohydrodynamic simulations performed with the FLASH code. In hydrodynamical simulations, the early deflagration phase produces large pockets of unburned C/O, leading to inefficient burning. Much of the released energy is deposited into buoyantly rising plumes rather than into the global pre-expansion of the white dwarf, which is required to produce the partially burned layers characteristic of SNe Ia. In contrast, when preexisting turbulent velocity fields and strong magnetic fields, on scales expected from the smoldering phase, are included, the effective burning approaches that in spherical models. Both turbulence and magnetic fields promote the entrainment of burned material into unburned pockets, addressing a long-standing problem in multi-dimensional deflagration models. The resulting streaks of burned material enable the conductive ignition of the surrounding unburned fuel. The dominant effect is not a change in the small-scale flame physics (~10^{-3} cm), but rather enhanced mixing between burned and unburned material. As expected, this mechanism is most efficient when the turbulent length scales are smaller than those of the unburned plumes.

💡 Research Summary

This paper tackles a long‑standing discrepancy between one‑dimensional (spherical) delayed‑detonation models of Type Ia supernovae and fully three‑dimensional (3D) hydrodynamic simulations. While spherical models naturally produce the required pre‑expansion of the white dwarf (WD) during the early deflagration phase—thereby yielding the observed partially burned layers and spectra—3D simulations have historically shown that most of the nuclear energy is channeled into buoyantly rising plumes rather than into a global expansion. Consequently, large pockets of unburned carbon‑oxygen (C/O) remain between the plumes, and the overall expansion is insufficient to match observations.

The authors identify the laminar‑speed‑dominated early deflagration as the critical stage and explore two physical ingredients that could remedy the problem: (1) pre‑existing turbulence generated during the smoldering phase, and (2) strong magnetic fields that may be present in the WD interior. Observational hints (late‑time light curves, line profiles) and dynamo theory suggest magnetic fields of 10⁶–10⁷ G, potentially amplified up to 10⁹–10¹¹ G before runaway. Turbulence estimates from previous work (Höflich & Stein 2002) give rms velocities of ~200 km s⁻¹ with eddy scales of ~50 km.

Using the FLASH 4.8 code, the authors perform a suite of 3D magnetohydrodynamic (MHD) simulations. The WD model originates from a near‑Chandrasekhar mass progenitor (central density ≈10⁹ g cm⁻³, radius ≈2000 km). The flame is treated with an ADR (advection‑diffusion‑reaction) scheme employing a sharpened Kolmogorov‑Petrovsky (sKPP) reaction term, propagating at a fixed laminar speed of 200 km s⁻¹. Nuclear energy release is approximated by a constant rate based on NSE burning to ⁵⁶Ni. Grid refinement reaches a minimum cell size of 1.2 km (l_max = 10), sufficient to resolve the 50 km turbulent eddies but not the microscopic flame thickness (~10⁻³ cm). The initial conditions include synthetic turbulent velocity fields (rms up to 200 km s⁻¹) and magnetic fields (rms up to 7 × 10¹¹ G), with two key variations: the diffusion radius of the turbulence (40 km vs. 80 km) and the presence/absence of magnetic fields.

Four representative runs are discussed: a pure hydrodynamic case (D300V30B0), a turbulent‑only case (D80V170B0), and two turbulent‑plus‑magnetic cases with different turbulence diffusion radii (D80V170B12 and D40V170B12). All simulations are evolved to 0.5 s, at which point the flame front has reached ≲1000 km, well within the static outer boundary (1500 km).

Results show that in the pure hydro run, buoyant plumes leave ≈100 km wide pockets of unburned material, and the laminar flame cannot penetrate these gaps, leading to poor global pre‑expansion. Adding turbulence alone improves mixing: the turbulent eddies drag burned material into the pockets, reducing the inter‑plume spacing. However, the most dramatic improvement occurs when both turbulence and a strong magnetic field are present. The Lorentz force acts perpendicular to plume motion, curving the flame front and forcing burned material into the gaps. In the D40V170B12 run (small‑scale turbulence, B ≈ 10¹² G), the pockets are almost completely filled, and the flame propagates in a manner that closely resembles a spherical deflagration, achieving the pre‑expansion required by observations.

The authors conclude that the dominant factor is not a modification of the microscopic flame physics (which operates on ~10⁻³ cm scales) but rather enhanced macroscopic mixing between burned and unburned material. Efficient mixing is achieved when the turbulent length scale is smaller than the characteristic spacing of the buoyant plumes. Strong magnetic fields further aid mixing by providing a transverse restoring force that redirects plume trajectories and promotes entrainment of burned ash into the unburned pockets.

This work therefore provides a plausible solution to the “missing pre‑expansion” problem in 3D Type Ia supernova models. It emphasizes that realistic initial conditions—specifically, turbulence inherited from the smoldering phase and magnetic fields approaching equipartition—are essential for reproducing the spherical nature of early deflagration inferred from observations. Future modeling efforts should therefore incorporate detailed pre‑runaway turbulence spectra and magnetic field configurations to achieve more accurate predictions of SN Ia light curves, spectra, and nucleosynthetic yields.