Core collapse supernovae: influence of magnetic fields Martin Obergaulinger, Pablo Cerdá-Durán, Ewald Müller Max-Planck-Institut für Astrophysik Magnetohydrodynamic effects in supernova infall R≈R0 / EG≈EG0, Erot≈2Erot0 , Emag≈ Emag0 ● (if collapse approx. homologous) ● poloidal circulation (due to centrifugal force) ()~-2 ● linear field amplification (toroidal field) by differential rotation ● ● ● 1 ● ● ● ● Up to 2000 only a few numerical studies: LeBlanc & Wilson ('70) Bisnovatyi-Kogan etal ('76) Müller & Hillebrandt ('79) Ohnishi ('83) Symbalisty ('84) ● Heger etal '05: =0.05s-1, R0= 5x10-4 Meier, Epstein, Arnett & Schramm '76 observed WD fields: < 10-10 M0 ● two parameters determine evolution: ● ● R0 = Erot0 / EG0 ● and 1 ● M0 = Emag0 / EG0 ● ● ● ● ● model C (LeBlanc & Wilson '70): 7 Msun Fe-Ni core R0 = 0.0025 ⇔ = 0.42 rad/s M0 = 0.00025 ⇔ B = 9.2 1011Gauss ● ● ● ● additional MHD processes of importance - Thompson & Murray ('01) (semi-analytic) field amplification by rapid convective overturn in PNS for ~10 s ; if turbulent = magnetic stresses Bequi ~ 1015G - Wheeler etal ('00, '02) (semi-analytic) asymmetric CCSNe & MHD jets (like in AGNs); field amplification by differential rotation and possibly by - dynamo - Akiyama etal ('03) (1D HD) importance of Magneto Rotational Instability strong MHD activity even for moderate initial conditions: Bsat ~1015...1016 G 1 ● - Tokyo group (2D/3D MHD) Yamada, Kotake, Sawai, parameter studies with polytropic & microTakiwaki, Nishimura physical EOS; neutrino leakage scheme; 2004-2008 B-field contribution to GW signal; nonuniform & offset B-fields MRI in action ? ● - Moiseenko etal ('06, '08) 2D MHD - MRI found? ● - Scheidegger etal ('08) 3D MHD ● - Mezzacappa etal ('08) compressional folding by SASI ● – – ● 2D MHD simulations: Obergaulinger, Aloy & Müller 2006 (TVD MHD code of Pen, Arras & Wong 2003 based on the relaxing TVD scheme of Jin & Xin 1995 ) ● ● ● Initial models: - rigidly & diff. rotating polytropes (as in Zwerger & Müller '97) ● 1 ● - approx. rel. gravity ● ● ● - B-field from toroidal current loop in equat. plane (100≤r/km≤800) central field strength 1010 – 1013 Gauss (realist. B < 109 Gauss) ● Evolution of central density and rotation rate non-magnetized slowly rotating Newtonian model bounces due to EOS Obergaulinger, Aloy & Müller '06 central density rotation rate ● ● 1 ● ● ● ● ● ● ● ----- 1010 Gauss ----- 1012 Gauss ----- 1013 Gauss ● Snapshot (49 ms post-bounce) of a model with a strong initial B-field & rapid differential rotation --> collimated outflow (Obergaulinger, Aloy & Müller 2006) ● ● gas & magnetic pressure ● 1 ● ● ● ● ● velocity & final velocity (total energy --> kinetic energy) ● Gravitational wave quadrupole amplitude TOV (solid line) vs Newtonian (dashed line) gravity for a rapidly rotating model Obergaulinger, Aloy & Müller '06 ● ● ● 1 ● ● ● ● ● ● ● ---- 1010 Gauss ----- 1013 Gauss ● Gravitational wave spectrum of a rapidly rotating initial model Obergaulinger, Aloy & Müller '06 ● ● ● ● 1 ● ● ● ● ● ---- 1010 Gauss ----- 1013 Gauss ● enhancement of spectral power at low frequencies in strong field model (right) is due to collimated outflow (jet) ● MHD studies have shown ● ● ● ● - efficient angular momentum transport by B-fields possible --> affects structure & rotation rate of PNS - conversion of rotational energy into thermal energy --> additional (isotropic) pressure contribution - weak, but realistic initial fields (B < 1011G) do neither change collapse dynamics nor resulting GW signal - strong initial fields (B ≥ 1012G) slow down core efficiently, change dynamics qualitatively, modify GW signal, and cause bipolar, mildly relativistic outflows ● ● Still waiting exploration: - MHD models with realistic microphysics and neutrino-driven convection - secular evolution ? ● - 3D effects, MHD instabilities (MRI) & possible MHD dynamo A closer look to field amplification mechanisms in core collapse supernovae ● ● ● ● compressional amplification feeds off kinetic energy of infall emag ~ - b2/2 div v amplifies field energy by a factor of 100...1000 during collapse works irrespective of field strength and geometry 13. III. 2008 MRI in core-collapse supernovae ● ● energy source: differential rotation ● requires poloidal seed field ● 13. III. 2008 winding by differential rotation creates & amplifies toroidal field component linear in time; time scale set by rotational period MRI in core-collapse supernovae ● ● ● ● turbulent dynamo energy source: turbulent kinetic energy turbulence excited by different effects genuinely 3D effect 13. III. 2008 MRI in core-collapse supernovae ● The magneto-rotational instability (I) ● ● main & original application: accretion disks (Balbus & Hawley 1991ff; based on Velikhov 1959 & Chandrasekhar 1960) ● MRI-driven turbulence can explain disk viscosity? -viscosity (Shakura-Sunyaev); see, however, King et al. 2007 ● large mean Maxwell stresses transport angular momentum outward (dependence on hydrodynamic background state?) ● energy source is differential rotation ● local linear instability with exponential growth 13. III. 2008 MRI in core-collapse supernovae ● ● The magneto-rotational instability (I) Properties of the MRI - stability criteria from analysis of wave-like (WKB) perturbations exp[i (k*r + t)] - accretion disks (Balbus & Hawley, 1991 ff) - weak fields |vA| << min { cS, |v| } - incompressible gas (Boussinesq approx.) - angular velocity constant on cylinders - thin Keplerian disk and polytropic EOS ● The magneto-rotational instability (II) ● ● MRI in SN cores quite different from MRI in accretion disks ● results from accretion disks may not apply, because - sub-Keplerian rotation - general rotation law (,z) - pressure forces are important (important global gradients) - thermal stratification (entropy gradients!) - different microphysics (e.g., radiative heating/cooling) - dynamic background ===> necessity for a MRI more general stability analysis in core-collapse supernovae 13. III. 2008 (Balbus 1995, Urpin 1996) ● ● Properties of the MRI ---> stability criterium: ℜ = ∂2 > 0 if criterium is violated, modes with wavenumbers are unstable! 2 growth rate of the fastest growing mode (Balbus & Hawley '92) independent of B; at wavenumbers close to the critical one ● ● Properties of the MRI general dispersion relation (Balbus 1995, Urpin 1996, Obergaulinger, Cerdá-Durán & Müller 2008 ● Properties of the MRI ● convenient quantity for stability analysis: general dispersion relation Obergaulinger, Cerdá-Durán & Müller (2008) modes with wavenumbers smaller than the critical wavenumber are unstable and grow Two branches of unstable modes (Urpin 1996): Alfven modes: C< 0 Buoyant modes: C < - 4 magneto-shear regime: -4<C< 0 mixed regime: -8<C<-4 magneto-convective regime: C<-8 ● ● Physical and technical problems ● dispersion relation imposes severe resolution problem (very difficult to resolve the MRI properly) ● relevant physical length scales 101 km ... 103 km – global scales: – scale heights of physical quantities: Hydrodynamics 13. III. 2008 MRI in core-collapse supernovae 1 km ● ● Physical and technical problems ● ● dispersion relation imposes severe resolution problem impossible to resolve properly the MRI relevant physical length scales 101 km ... 103 km – global scales: – scale heights of physical quantities: 1 km – MRI wavelength 1 cm ... 10 m magnetic field 13. III. 2008 MRI in core-collapse supernovae ● ● Physical and technical problems ● ● dispersion relation imposes severe resolution problem impossible to resolve properly the MRI relevant physical length scales 101 km ... 103 km – global scales: – scale heights of physical quantities: 1 km – MRI wavelength 1 cm ... 10 m – viscous and resistive scales 13. III. 2008 MRI in core-collapse supernovae ≪ 1 cm ● ● Physical and technical problems ● MRI requires 3d simulations ● numerical grid can cover a factor of 102...104 in length scales ● impossible to cover all scales appropriately ● where to place the numerical grid? ● global simulations do not resolve MRI properly, but capture global dynamics 13. III. 2008 MRI in core-collapse supernovae ● ● Physical and technical problems ● MRI requires 3d simulations ● numerical grid can cover a factor of 102...104 in length scales ● impossible to cover all scales appropriately ● where to place the numerical grid? – local simulations (more or less) well resolved, but neglect global dynamics 13. III. 2008 MRI in core-collapse supernovae ● ● Global simulations ● simulate a large region of the core at a fairly coarse resolution – 2d axisymmetric or 3d – include detailed physics – use detailed initial models ● MRI found; study its implications for dynamics ● BUT: MRI is underresolved ! => use stronger initial fields, relying on efficient amplification 13. III. 2008 MRI in core-collapse supernovae ● ● Global simulations magnetic energy 13. III. 2008 MRI in core-collapse supernovae ● Local simulations ● ● ● simulate a small box with a fine resolution – 2d, 3d Cartesian / spherical / cylindrical – simplified physics – representative initial conditions – choice of boundary conditions may be crucial periodic (shearing box), reflecting, outflow resolve the MRI for weak initial fields 13. III. 2008 MRI in core-collapse supernovae ● ● Local simulations ● problems: neglect of global dynamics may distort results – large parameter space ● ● ● ● ● – background models (rotation, stratification) initial field strength and geometry transport coefficients grid size resolution unclear dependence of the saturation state on the numerical parameters (e.g., grid size, boundary treatment) (Umurhan & Regev, 2008) 13. III. 2008 MRI in core-collapse supernovae ● Semi-global simulations ● ● local simulations of MRI in SN cores (Obergaulinger etal '2008) with shearing disk boundary conditions allowing for the inclusion of global gradients (Klahr & Bodenheimer 2003) ● box (edge length 1km; located 15 km from center of star) taken from an idealized model of a post-collapse core ● simplified EOS, no neutrino transport ● variation of the initial magnetic field strength and geometry – ● new high-resolution finite-volume flux-conservative MHD code [ TVD-PLM, MP5 or WENO4, with approx. Riemann solvers (MUSTA) and CT ] 13. III. 2008 MRI in core-collapse supernovae ● Semi-global simulations ● development of channel flows in an axisymmetric model with an initially uniform B-field 13. III. 2008 MRI in core-collapse supernovae ● Semi-global simulations ● ● ● MRI grows rapidly (within a fraction of a millisecond) saturation depends on the field geometry, the grid, and the dimensionality – 2d, uniform B-field: occurrence of coherent channel flows – 2d, vanishing net B-flux: turbulence – 3d: channel flows unstable against parasitic instabilities, turbulence develops; channels may reorganise spontaneously ● saturation level set by rotational equipartition (~ 1015 Gauss) ● Maxwell stresses transport angular momentum efficiently MRI in core-collapse supernovae 13. III. 2008 ● Evolution of magnetic field strength in semi-global 3d models ● (1km)3, 1003 0 = 15.5 km b0 = 2e13 G t = 16 ms: channel flows grow 0 = 1900 s-1 = -1.25 t = 27 ms: channels are disrupted t = 43 ms: reappearance of coherent flow 13. III. 2008 MRI in core-collapse supernovae volume rendering of |B| (blue to green) iso- surface [ z] ● ● Conclusions ● simulation of MRI hindered by resolution requirements ● combination of global and local simulations suggested ● semi-global simulations – confirm efficient amplification of magnetic stresses by the MRI, turbulence, and angular momentum transport – preliminary analysis leaves open several issues ● ● ● ● detailed dependence on initial conditions possible dependence on the numerics effects of neglected physics? MRI studies shouldMRIbe extended to GRMHD in core-collapse supernovae 13. III. 2008