Santorini August 29, 2005
Supernovae, Gamma-Ray Bursts and Stellar Rotation
Brief Overview • The SN-GRB connection – observational • Progenitor evolution – what makes the difference between ordinary SNae and GRBs? • Central engine and diagnostics • Outlook
Paciesas et al (2002) Briggs et al (2002) Koveliotou (2002) Shortest 6 ms GRB 910711 Longest ~2000 s GRB 971208
Since 1997 we have known: • “Long-soft” bursts are at cosmological distances and are associated with star forming regions in starbursting galaxies 33 bursts Zeh et al. (2004)
Frail et al. (2001) As a relativistic jet decelerates we see a larger fraction of the emitting surface until we see the edges of the jet. These leads to a panchromatic break slope of the the afterglow light curve.
Djorgovski et al (2002) almost always galaxies experiencing an unusual rate of star formation
Fruchter et al (2005) submitted to ApJ. The green circles show GRB locations to an accuracy of 0.15 arc sec. Conclusion: GRBs trace starformation even more than the average core-collapse supernova. They are thus to be associated with the most massive stars. They also occur in young, small, star forming galaxies that might be metal poor.
Smoking Gun Number 1 SN 1998bw/GRB 980425 NTT image (May 1, 1998) of SN 1998bw in the barred spiral galaxy ESO 184-G82[Galama et al, A&A S, 138, 465, (1999)] WFC error box (8') for GRB 980425 and two NFI x-ray sources. The IPN error arc is also shown. Type Ic supernova, d = 40 Mpc Modeled as the 3 x 1052 erg explosion of a massive CO star (Iwamoto et al 1998; Woosley, Eastman, & Schmidt 1999) GRB 8 x 1047 erg; 23 s a very unusual supernova!
Smoking gun number 2 GRB030329/ SN2003dh z = 0.1685 One of the brightest GRBs ever – HETE2 Chornock et al. Eracleous et al., Hjorth et al., Kawabata et al., Stanek et al.
Lpeak implies (again) ~0.5 solar masses of 56Ni Exceptionally --- bright fast high velocity radio bright Supernova simultaneous with the GRB (+- 2 days).
Zeh, Klose, & Hartmann (2004) Results • among 36 GRBs with optical afterglows (end of 2002), 21 have a sufficient data quality and a known redshift • nine late time bumps are found in afterglow light curves
Conclusions So Far • GRBs (at least a lot of those of the long-soft variety) come from the deaths of massive stars • At least some of these eject about 0.5 solar masses of56Ni – an important diagnostic of the central engine • The supernovae may be, on the average, hyperenergetic (~1052 erg) and asymmetric. They are Type Ib/c. • Unlike ordinary supernovae, those that make GRBs eject an appreciable – and highly variable - fraction of their energy in relativistic ejecta (G > 200). This ejection is beamed to ~1/300 of the sky. • The fraction of all supernova-like events that make GRBs is small
While all (long-soft) GRBs may occur simultaneously with supernovae of Type Ic, the converse is not true – at least not locally. • Lack of relativistic ejecta or strong radio emission in most supernovae (Soderberg et al. 2004) • Lack of temporal correlation of GRBs with most supernovae • Lack of broad spectral lines in most supernovae Estimates are that typically no more than a few percentof supernovae (of any Type) are GRBs. e.g., Soderberg, Frail, and Wieringa (2004) < 6%
The GRB rate is a very small fraction of the total supernova rate Madau, della Valle, & Panagia, MNRAS, 1998 Supernova rate per 16 arc min squared per year ~20 This corresponds to an all sky supernova rate of 6 SN/sec For comparison the universal GRB rate is about 3 /day * 300 forbeaming or ~ 0.02 GRB/sec
Today, after times when over 150 GRB models could be “defended”, only two are left standing (for long-soft bursts): • The collapsar model • The millisecond magnetar model Both rely on the existence of situations where some fraction of massive stars die with an unsually large amount of rotation. The degree of rotation and the distribution of angular momentum is what distinguishes GRBs from ordinary supernovae.
Common theme (and a potential difficulty): Need iron core rotation at death to correspond to a pulsar of < 5 ms period if rotation and B-fields are to matter at all. Need a period of ~ 1 ms or less to make GRBs. This is much faster than observed in common pulsars. To make a disk around a 3 solar mass black hole need j ~ 5 x 1016 cm2 sec-1
Calculations agree that without magnetic torques it is easy to make GRBs This is plenty of angular momentum to make either a ms neutron star or a collapsar. Heger, Langer, & Woosley (2002)
But if include WR mass loss and magnetic fields the answer is greatly altered.... no mass loss or B-field In the absence of mass loss and magnetic fields, there would be abundant progenitors. Unfortunately nature has both. 15 solar mass helium core born rotating rigidly at f times break up with mass loss with mass loss and B-fields
Much of the spin down occurs as the star evolves from H depletion toHe ignition, i.e. as a RSG. Heger, Woosley, & Spruit (2004)
Good news for pulsars Bad news for GRBs! Heger, Woosley, & Spruit (2004) using magnetic torques as derived inSpruit (2002)
But must all massive stars pass through a red or blue supergiant phase? Consider stars with the upper 5% of rotation on the main sequence and standard rotational mixing parameters.
never a red giant vrot = 400 km/s He C,Ne He H C,Ne O Si O
PreSN He-depl GRB C-depl H 8 ms pulsar
Woosley & Heger astroph - 0508175 And so maybe …. GRBs come from single stars on the high- velocity tail of the rotational velocity distribution Such stars mix completely on the main sequence The WR mass loss rate is low (because of metallicity). The mass loss rate may be small when the star dies as well. Upper bound in metallicity for making GRBs is uncertain but around 0.1 solar. This means the fraction of massive star deaths at low metallicity that make GRBs is higher. See also Yoon and Langer astroph - 0508242
Effect of Mass Losson Burst Properties The wind mass required to decelerate a relativistic jet of equivalent isotropic energy E and Lorentz factor G is the mass loss rate times the time before the burst For typical GRB (equivalent isotropic) energies, E53 = 1 the relativistic jet with G ~ 100 gives up its energyat around 1015 cm.
Mass Loss Burning Phase Duration Wind Radius Probed by t 108 x t Hydrogen 19 My Optical observation Helium 0.5 My Optical observation Carbon/ 3400 y < 1019 cm SN Ib, GRB afterglow Neon Oxygen 7 mo < 2 x 1015 cm GRB Si 2 weeks < 2 x 1014 cm GRB For helium burning and beyond the wind radius is taken to be 1000 km/s times the duration of the burning phase. No WR star has ever been observed in any of the burning phasesmost appropriate to GRBs.
Summary • Credible, though uncertain models can give – approximately – the observed rotation rate of young pulsars for stars that become red supergiants and have their differential rotation partially braked by internal magnetic torques. • Using the same torques, but reducing WR mass loss by a factor of a few can give credible GRB progenitors if the stars thoroughly mix during hydrogen and helium burning. This may occur if the stars rotate on the main sequence considerably faster than usual – about 35% Keplerian, 400 km/s. • GRBs will be favored by low metallicity. The threshold metallicity for making a GRB depends critically upon the size and Z-dependence of the mass loss rate. Recent results (Vink et al 2005) sugegsting a Z0.86 scaling for WR star winds rather than Z0.5 as previously assumed are very encouraging.
Observations suggest that Z is low! Fynbo, Jakobsson, & Moeller, et al, 2003, A&ApL, 406, L63 “On the Ly-alpha emission from gamma-ray burst host galaxies: evidence for low metallicities” Gorosabel J., Perez-Ramirez D., Sollerman J., et al., 2005, A&A in press, astro-ph/0507488 “The GRB 030329 host: a blue low metallicity subluminous galaxy with intense star formation” Sollerman J., Ostlin G., Fynbo J. P. U., et al., New Astronomy, in press, astro-ph/0506686 “On the nature of nearby GRB/SN host galaxies”
Models for the Central Engine
When Massive Stars Die, How Do They Explode? Neutron Star + Neutrinos Neutron Star + Rotation Black Hole + Rotation Colgate and White (1966) Arnett Wilson Bethe Janka Herant Burrows Fryer Mezzacappa etc. Bodenheimer and Woosley (1983) Woosley (1993) MacFadyen and Woosley (1999) Narayan and Piran (2004) Hoyle (1946) Fowler and Hoyle (1964) LeBlanc and Wilson (1970) Ostriker and Gunn (1971) Bisnovatyi-Kogan (1971) Meier Wheeler Usov Thompson etc All of the above? 10 20 35
It is clear that very many massive stars, probably most, produce neutron stars when they die. A good model should explain: Supernova energies, light curves and spectra Neutron star masses Pulsar rotation rates Pulsar peculiar velocities Neutron star magnetic field strengths Nucleosynthesis Remnant morphology The standard neutron star – neutrino model shows great promise for explaining all of these.
But not everyone agrees: Healthy explosions without rotation or magnetic fields. Neutrino powered.Crude neutrino physics. 2D and 3D. Marginal failures. 2D with improved neutrino transport. Full sphere shows importance of dipole. Rotation alters convective flow in the neutrino-powered model and makes it weaker Very strong B-fields develop from differential rotation. Rotation powered explosions. B ~ 1016 gauss 3D models. Estimate B-field much smaller than Akiyama et al. Neutrino transport dominates B-fields more like Akiyama in realistic model but rotational energy dissipates in neutrinos. Rotation triggers neutrino powered explosion • Burrows, Hayes and Fryxell (1995) Fryer (1999); Fryer and Warren (2002) • Bruen et al (2001); Liebendorfer et al (2001) Burras et al (2003); Shenk et al. (2004) • Fryer and Heger (2000) • Akiyama et al (2003) • Fryer and Warren (2003) • Thompson, Quataert, and Burrows (2004)
Yet we all agree that rotation must be an essential ingredient in making a GRB. Today, after times when over 150 GRB models could be “defended”, only two are left standing (for long-soft bursts): • The collapsar model • The millisecond magnetar model Both rely on the existence of situations where some fraction of massive stars die with an unsually large amount of rotation.
Collapsar Basics • Wolf-Rayet Star – no hydrogen envelope – about 1 solar radius. • Collapse time scale tens of seconds • Rapid rotation – j ~ 1016 erg s • Black hole ~ 3 solar masses accretes about a solar mass MacFadyen & Woosley (1999)
The star collapses and forms a disk (log j > 16.5) In the vicinity of the rotational axis of the black hole, by a variety of possible processes, energy is deposited. 7.6 s after core collapse; high viscosity case.
The alternative: Wheeler, Yi, Hoeflich, and Wang (2001) Usov (1992, 1994, 1999) Thompson (2003) The ms Magnetar Model:
Challenges for the Neutron Star Model • The neutron star does not form promptly and develop its full field strength and rotation rate. It takes tKH ~ 3 s. During this time the accretion rate is several tenths of a solar mass per second. The proton-neutron star – at least in many cases becomes a black hole • Producing enough 56Ni. A jet alone cannot produce much56Ni or even blow up the star effectively. To make the 56Ni hydrodynamically, energy of ~1052 erg must be deposited nearly isotropically in less than 1 second. But then the neutron star must somehow produce a highly collimated flow for a minimum of 10 seconds after that.
The passage of a jet through a massive star will always lead to an explosion. All GRBs from massivestars will be accompanied by a “supernova”. But the explosion induced by the jet has very low energy and does not make much 56Ni, Zhang et al (2003)
The disk wind: MacFadyen & Woosley (2001) Neglecting electron capture in the disk Also recent work by Narayan and Piran (2004), Igumenshchev et al (2003), “NDAF”
If energy is not maintained at the base for a jet crossing time the jet dies. (This is also why one must have a WR-staras the GRB progenitor, BSGs do not work) Zhang et al (in preparation)
Three versions of this calculation now completed (11/04) in 3D (thanks to Columbia) minimum Power time to break out 3 x 1050 erg s-1 7 sec3 x 1049 erg s-1 16 sec 3 x 1048 erg s-1 25 sec (roughly E1/4) The central engine must stay on at least this long. Low energy events are possible, but should have long duration Structure at break out similar. Still being analyzed.
So, the jet must be maintained at least 10 seconds (and better yet, 30) in order that relativistic matter escape the star. The collapsar satisfies these twin constraints of 56Ni production and long time scale by a) staying on for a hydrodynamic time scale for the whole star and b) making the 56Ni by advection through a disk, not by a spherical shock. Can the magnetar model do it? Emit about 1052 erg first two seconds and 1051 erg after 10 s. Is that enough?