Mass ejection from neutron-star mergers in numerical relativity

Transcription

1 Mass ejection from neutron-star mergers in numerical relativity Masaru Shibata Center for Gravitational Physics, Yukawa Institute for Theoretical Physics, Kyoto University

2 I. Brief introduction Outline II. Typical scenarios for NS mergers (both for NS-NS and BH-NS) III. Dynamical mass ejection IV. Early MHD/viscous ejection from NS-NS V. Viscous (+neutrino-assisted) disk wind VI. Summary

3 I Introduction (not necessary?) Why mass ejection from NS binaries is important? 1. Electromagnetic counterparts of NS merger: Key for confirming gravitational-wave detection (talks by Tanaka & Cowperthwaite ) 2. Possible site of r-process nucleosynthesis (talks by Foucart & Hotoke) Jet ISM Shock (Afterglow) Optical (hours days) Radio (weeks years) Ejecta ISM Shock Radio (years) θobs GRB (t ~ s) Kilonova Optical (t ~ 1 day) θj Merger Ejecta Tidal Tail & Disk Wind v ~ c BH Metzger & Berger 2012

4 In the following, I focus on Ejecta mass M eject Electron fraction Y e Light curve i band 200 Mpc Abundance pattern Observed magnitude m 4m 8m Days after the merger Tanaka & Hotoke 2013 Korobkin et al. 2012

5 II A Typical scenarios for NS-NS merger Constraints from radio-telescope observations: 1. Approximately 2-solar-mass NSs exist (Demorest ea 2010, Antoniadis ea 2013) à equation of state (EOS) for NS has to be stiff 2. Typical total mass of compact binary neutron stars à ~ 2.73±0.15 solar mass (by Pulsar timing obs.)

6 Mass-radius relation for various EOS Mass (solar) WD-NS binary J Radius is still unconstrained Strong constraint = EOS is stiff km 12 km 14 km Radius (km)

7 Compact NS-NS system in our galaxy Ø Total Mass of NS in compact NS-NS is likely to be in a narrow range, m 2.73±0.15 M sun Orbital period Eccentricity Each mass lifetime PSR P(day) e M(M sun ) M 1 M 2 T GW 1. B B B C J J J J ~5 8. A ~ yrs

8 II A Typical scenarios for NS-NS merger Constraints from radio-telescope observations: 1. Approximately 2-solar-mass NSs exist (Demorest ea 2010, Antoniadis ea 2013) à equation of state (EOS) for NS has to be stiff 2. Typical total mass of compact binary neutron stars à ~ 2.73±0.15 solar mass (by Pulsar timing obs.) Numerical relativity simulations have shown that Ø Merger results typically in high-mass neutron stars (not BH) (Shibata et al. 2005, recently many works.)

9 Possible outcomes of NS-NS mergers M thr > ~2.8M sun Depends strongly on EOS Likely typical cases for M = M sun

10 Mass ejection history for MNS formation Dynamical ejection (Sec. III) (determined by dynamical timescale of NS) Time after merger ms MHD/viscous ejection (Sec. IV) (by viscous timescale of remnant MNS) Long-term viscous ejection (V) (by viscous timescale of disk) Neutrino irradiation (for neutrino emission timescale) (minor effects but could play an assist) Recombination (Fernandez-Metzger 13)

11 II B Scenarios for BH-NS merger Almost no observational constraints but for black hole mass likely >~5M sun à Wide parameter space has to be explored Fate = two possibilities: 1. Tidal disruption of NS 2. Simple plunge of NS into BH

12 Condition for tidal disruption ( ) < ( BH tidal force) For tidal disruption, Self gravity of NS M M ( NS BH αr ) NS α >1 r 3 < ( αr ) 2 NS ( ) 1 M BH r ISCO 3 M NS M BH 2 αr NS M NS 3 For tidal disruption v Large NS Radius or v Small BH mass or v High corotation spin is necessary ISCO B H B H αr NS NS R NS NS

13 For tidal disruption of plausible BH-NS with M NS =1.35M sun, R NS ~ 12 km, & M BH > 6 M sun High BH spin is necessary > ~ 0.5 Foucart et al. ( 13,14, ); Kyutoku et al. ( 15) M BH r ISCO 3 7M NS M BH 2 R NS 6M NS (M BH r ISCO 9M BH ) 3 α Ø Natural conclusion: BH-disk systems formed as a remnant should have a high BH spin

14 Mass ejection history for BH-NS (in the presence of tidal disruption of NS) Time after merger ms Dynamical ejection (Sec. III) (determined by dynamical timescale of system) Long-term MHD/viscous ejection (Sec. V) (by viscous timescale of disk) (Fernandez-Metzger 13, Just+ 15, ) Neutrino irradiation (would be minor)

15 III Dynamical mass ejection

16 NS-NS: Neutrino-radiation hydro simulation Soft EOS (SFHo, R~11.9 km): M sun Rest-mass density Orbital plane Neutrino luminosity ν e ν e ν others Total mass ~ 0.01 M solar x-z plane Sekiguchi et al

17 NS-NS: Neutrino-radiation hydro simulation Stiff EOS (DD2, R~13.2 km): M sun Rest-mass density Orbital plane Neutrino luminosity ν e ν e ν others Total mass ~ 10-3 M solar x-z plane Sekiguchi et al

18 Ejecta mass depends on EOS : NS-NS case Soft EOS à strong gravity à SHOCK à high-mass ejection SLy APR4 Steiner ALF2 Mass ratio Tidal effect is major H4 MS1 Total mass = 2.7 solar mass Error bar for 1 < Q < 1.25 Radius of 1.35 solar mass NS Hotokezaka+ PRD 13 (See also Bauswein+ 13; Bernuzzi + 15)

19 Summary for dynamical ejecta in NR Ejecta mass depends significantly on NS EOS & mass Nearly equal mass (M tot ~ 2.7M sun ) Unequal mass: m 1 /m 2 < 0.9 (M tot ~ 2.7M sun ) Small total mass system (< 2.6M sun ) Soft EOS (R=11-12 km) HMNS à BH M eje ~10-2 M sun HMNS à BH M eje ~10-2 M sun MNS (long lived) M eje ~10-3 M sun Stiff EOS (R=13-15km) MNS (long lived) M eje ~10-3 M sun MNS (long lived) M eje ~ M sun MNS (long lived) M eje ~10-3 M sun Ø Typical velocity: c Foucart et al 16 Shibata unpublished Sekighichi+ 17

20 High Neutrino-radiation temperature hydrodynamics γγ e + e +, n + simulation e + p +ν e SFHo (R~11.9 km): M sun Neutrino irradiation n +ν p + e Electron fraction (x-y) ν e Y e Neutrino luminosity ν e ν others Electron fraction (x-z) Sekiguchi et al. (2016) Green = neutron rich 20

21 Fraction of mass Electron fraction profile: Broad solar case t - t M-6 [ms] Sekiguchi et al PRD SFHo DD2 TM Ye Ø Average depends on EOS but typically peak at Ø Broad distribution irrespective of EOS Ø Similar results by Radice+16, Lehner+15,16

22 Neutrino-radiation hydrodynamics simulation SFHo (R~11.9 km): M sun Y e More neutron-rich except for disk surrounding BH Green = neutron rich Sekiguchi et al. (2017 hopefully)

23 Mass fraction Electron fraction distribution: Broad irrespective of EOS and mass à Good for producing a variety of r-elements Mass fraction Mass fraction Asymmetric binary SFHo DD Soft EOS Stiff EOS 10-4 See also Radice Electron fraction (Ye) Electron fraction Y e Sekiguchi+ 16

24 Neutrino irradiation: subdominant effect Ejecta mass Electron fraction Hea/ng on Hea/ng off Hea/ng on Hea/ng off Sekiguchi Neutrino irradiation from MNS increases Ø the ejecta mass by ~ solar mass Ø Average value of Y e by ~ 0.03 ü Note that neutrino luminosity decreases in ~100 ms See also, Perego et al. 2014; Goriely et al. 2015; Martin et al. 2015; Foucart et al. 2016

25 BH-NS merger (SFHo EOS: density) M BH =5.4M sun, M NS =1.35M sun, a BH =0.75 Mass ejection occurs by tidal force of BH Kyutoku et al. hopefully 2017

26 M *ejecta (solar mass) BH-NS with NS mass 1.35M sun M NS =1.35 solar mass Soft EOS < ~0.01 M sun Data: Kyutoku et al Stiff EOS results in high mass > 0.01 M sun M BH =9.45, a=0.75 M BH =9.45, a=0.50 M BH =6.75, a=0.50 Higher spin Radius of R1.35 (km) solar mass NS High BH spin is important for mass ejection

27 BH-NS merger (SFHo EOS: electron frac) M BH =5.4M sun, M NS =1.35M sun, a BH =0.75 Very neutron rich Y e <~ 0.1 Kyutoku et al. hopefully 2017

28 Electron fraction of ejecta Δ M ej / M ej SFHo DD2 TM1 R=11.9 km R=13.2 km R=14.5 km (dm ej /dy e ) / M ej electron fraction Quite low electron fraction irrespective of EOS (Foucart et al., 13, 14, 15, Kyutoku+ hopefully 17) Likely to primarily produce heavy r-elements

29 Dynamical ejecta properties in NR u Mass: NS-NS: ~ M sun depending on each mass & EOS: Soft EOS & ~2.7 M sun is favorable (Hotoke+ 13, Sekiguchi+ 15,16, Radice+ 16, Lehner+ 15,16) BH-NS: M sun : Stiff EOS is favorable; high BH spin is also the key (Foucart , Kyutoku+15): -- M eject ~ M disk u Electron fraction NS-NS: Broad distribution of Y e with average <Y e > ~ : For asymmetric case, <Y e > could be < 0.2 BH-NS: Peak at Y e < 0.1 (Foucart , Kyutoku+ 17) u Typical velocity: c; max could be ~ 0.8 c

30 IV Early Viscous/MHD ejecta for NS-NS MHD/viscous effects are likely to play a role (Fernandez-Metzger , Just et al. 15.) But, previous simulations are studied only for torus surrounding BH (or very artificial NS) Realistic remnants = MNS + torus, for which no wellresolved MHD or viscous simulations MNS of differential rotation has potential for mass ejection

31 Physical state for the merger remnants Remnant MNS are magnetized & differentially rotating à subject to MHD instabilities MHD simulations (e.g., Price & Rosswog, 07, Kiuchi et al. 14, 15) suggest that magnetic fields would be significantly amplified by Kelvin-Helmholtz instability à turbulence may be induced

32 High-resolution GRMHD for NS-NS Kiuchi et al Δx=17.5m Kelvin-Helmholtz instability: τ KH Δx à Magnetic field should be amplified by winding à Quick angular momentum transport? (not yet seen)

33 Magnetic energy: Resolution dependence B field would Please be pay amplified attention in only Δt << to 1 blue ms curves à turbulence? E B [erg] m m 150m m m 110m m 70 35m 70m Still NOT convergent B max =10 13 G Higher resolution Purely hydrodynamics or radiation hydrodynamics is 10not 45 likely to be appropriate for this problem τ KH Δx t - t mrg [ms] Kiuch et al. 2015

34 Shear motion at the merger à huge number of vortexes are formed and magnetic field is quickly amplified à further shear motion à turbulence à turbulent (effectively global) viscosity

35 For post-merger dynamics, Obviously more resolved MHD simulation is needed à But it is not feasible due to the restriction of the computational resources (in future we have to do) One alternative for exploring the possibilities is viscous hydrodynamics (Radice 17, Shibata et al. 17) ü Note that we do not know whether viscous hydrodynamics can precisely describe turbulence fluid

36 Viscous neutrino radiation hydrodynamics for post-merger MNS τ ν R2 (S. ν = 1 ( RΩ ) 2 e ~ 10 α 1 v ms Fujibayashi α et 2 al. in preparation) ν Ω e c s 0.01 Employ covariant & causal GR viscous hydro (Israel & Steward) Initial condition: Merger remnant of M sun NS-NS Alpha viscosity; ν =α v c s 2 Ω -1 with α v = 0.01 EOS: DD2 (R NS = 13.2 km) à Dynamical ejecta mass ~ M sun Wide km km Density in x-z plane

37 Evolution of angular velocity Relax to uniform rota/on in viscous /mescale ~ 10 ms Fujibayashi et al. in preparation Play a role in the late-time viscous ejection Kinetic energy of ~10 52 erg is released à early viscous ejection

38 Ejecta mass and Y e distribution t < ms: Differential rotation of MNS à rigid rotation à viscous heating à ejecta of mass > M sun Viscous driven Neutrino driven Mass per bin [M Ȯ ] ms, α=0.02 α=0.01 α=0.00 Y e > ~0.25 This depends on initial condition Y e Fujibayashi et al. in prep.

39 Viscous hydrodynamics for post-merger MNS (S. Fujibayashi et al. in preparation) Electron fraction Wide km km

40 Dynamical + MHD/viscous ejecta in NR Total ejecta mass could be ~0.01 M sun or more Nearly equal mass (M tot ~ 2.7M sun ) Unequal mass: m 1 /m 2 < 0.9 (M tot ~ 2.7M sun ) Small total mass system (< 2.6M sun ) Soft EOS (R=11-12 km) MNS à BH M eje ~10-2 M sun MNS à BH M eje ~10-2 M sun MNS (long lived) M eje ~?? Stiff EOS (R=13-15km) MNS (long lived) M eje ~10-2 M sun MNS (long lived) M eje ~10-2 M sun MNS (long lived) M eje ~?? To be studied Ø <Y e > ~ (likely) Ø Y e has a wide distribution à Good for nucleo-synthesis

41 V Long-term viscous disk wind Ø Studies have been done mostly for BH-disk systems (Fernandez-Metzger, 13-15, Just+ 15, Siegel-Metzger 17; Natural model for BH-NS merger) 10 20% of mass of disk surrounding a spinning BH is likely to be ejected by viscous ejection Due to Y e freeze-out in the absence of strong neutrino sources, low Y e matter could be ejected

42 Basic Picture (Fernandez-Metzger 13,14, Just 15, ) Neutrino irradiated ejection à Y e is increased (weak effect for BH-NS) Low Y e ~ 0.1 BH Low Y e ~ 0.1 Viscous ejection of mass 10 20% of torus mass Y e freeze out à Low Y e is preserved (good r-process)

43 Concern ü Initial disk model is rather artificial, in particular, j=const angular momentum distribution is often used, but it s unphysical, and in this case, torus becomes geometrically thick leading to easy ejection: j=const torus High entropy ~ Kepler disk More realistic Overes/mated mass ejec/on? Overes/mated neutrino hea/ng?

44 Throughout mass ejection of BH-NS merger For tidal disruption of NS, high BH spin is necessary à remnant should be high-spin BH + disk Ø Dynamical ejecta: M eject ~ M disk (e.g., Kyutoku+ 15) Ø Viscous ejecta from disk could be ~ M disk à Comparable to dynamical ejecta ² Dynamical ejecta has small Y e < 0.1 (e.g., Forcart+, 14) ² Viscous ejecta is also likely to give Y e ~ because of the absence of strong neutrino sources and resulting freeze-out effect (Fernandez-Metzger 13, 14, Just + 15, Siegel-Metzger 17) à Likely to be a strong site for the r-process nucleosyn. Conclusion seems to be robust

45 Long-term viscous disk wind: NS-NS case Ø Remnant MNS-disk systems have been studied only with artificial treatments of MNS The presence of a strong neutrino emitter like MNS would change Y e significantly (Metzger-Fernanndez 13, Perego+ 14, Fujibayashi+ 17) ü Caution: Luminosity of MNS decreases with time Low-Y e disk initial condition may not be realistic for MNS-disk system Need more realistic studies from NR merger simulation

46 u NS-NS: IV Summary Dynamical + subsequent short-term MHD/viscous ejection are likely to provide ejecta mass of > 0.01 M sun irrespective of EOS and each mass of binary Y e is mildly low & broadly distributed: good Long-term evolution of post-merger MNS-torus:??? u BH-NS: likely robust conclusion Dynamical ejection could provide M sun, in the case of TD and resulting Y e is low < 0.1 Post-merger BH-torus could also eject mass 20 50% of disk mass by viscous effect à M eje >~0.01 M sun : Y e could also be mildly low ~