Great Debate on GRB Composition: A Case for Poynting Flux Dominated GRB Jets Bing Zhang Department of Physics and Astronomy University of Nevada, Las Vegas March 6, 2011 In “Prompt Activity of Gamma-Ray Bursts” Raleigh, North Carolina Reference: Zhang & Yan (2011, ApJ, 726, 90)

Sherlock Holmes

GRB C (Abdo et al. 2009, Science) Fingerprint/footprint/ Smoking gun:

Standard Fireball Shock Model central photosphere internal shocks external shocks engine (reverse) (forward) GRB prompt emission: from internal shocks and photosphere Afterglow: from external shocks

Predicted spectra Meszaros & Rees (00) Daigne & Mochkovitch (02) Zhang & Meszaros (02, unpublished)

Expected photosphere emission from a fireball (Zhang & Pe’er 09) Sigma: ratio between Poynting flux and baryonic flux:  = L P /L b : at least ~ 20, 15 for GRB C Confirmed by Fan (2010) with a wider parameter space study.

The simplest fireball model does not work! Modified fireball models

Modified Fireball Model (1) central photosphere internal shocks external shocks engine (reverse) (forward) GRB prompt emission: from internal shocks and photosphere Afterglow: from external shocks

Magnetic acceleration?   High  initially, but quickly low  ? – MHD model, magnetic pressure gradient accelerate ejecta, Poynting flux can be (partially) converted to kinetic energy (Vlahakis & Konigl 2003; Komissarov et al. 2009; Tchekhovskoy, Narayan & McKinney 2010; Granot, Komissarov & Spitkovsky 2011; Lyubarsky 2010) – The conversion efficiency is low – With external confinement (e.g. stellar envelope), the efficiency can be higher, but the flow can still have a moderately high σ in the emission region. Talks by Narayan, Tchekhovsky, McKinney, Giannios …

Difficulties/issues of the internal shock model Missing photosphere problem Low efficiency Fast cooling problem Electron number excess problem Ep – Eiso (Liso) correlation inconsistency

Modified Fireball Model (2) central photosphere internal shocks external shocks engine (reverse) (forward) GRB prompt emission: from internal shocks and photosphere Afterglow: from external shocks

The band function is emission from the photosphere Dissipated photosphere with upscattering (Thompson 1994; Rees & Meszaros 2005; Ghisellini et al. 2007; Pe’er et al. 2006; Giannios 2008; Beloborodov 2010; Lazzati & Begelman 2010; Pe’er & Ryde 2010; Ioka 2010; Metzger et al. 2011) Two problems: – Cannot reach > 1 GeV – Low energy spectral index is too hard ? ? α ~ -1 α ~ ( ) Beloborodov (2010); Mizuta et al. (2010); Deng & Zhang (poster)

Superposition (Toma et al. 2010; Li 2009)? – Contrived fine-tuning – Seems not supported by data (Binbin’s talk) ? ? The band function is emission from the photosphere

Synchrotron + photosphere (Giannios 2008; Vurm’s talk; Beloborodov’s talk)? Predict bright optical emission Prompt optical data (Yost et al.) do not support this possibility (Shen & Zhang 09) The band function is emission from the photosphere Giannios (2008)

Sherlock Holmes

Preece’s talk Fingerprint/footprint/ Smoking gun:

Lu, Hou & Liang 2010

Diverse Ejecta Composition: Thermal emission in GRB B! Ryde et al. (2010); B.-B. Zhang et al. (2011) - Bin-Bin’s talk Ryde et al. (2010); B.-B. Zhang et al. (2011) - Bin-Bin’s talk This is a Paczynski-Goodman “fireball”!

GRB B (probably also ) At least GRB B is a fireball (probably also ) Rare: 2 of 17 LAT GRBs (B.-B. Zhang’s talk) Photosphere emission looks quasi-thermal. Band function is not superposition of photosphere emission Photosphere model must address diversity of Comptonization

Back-up slides

Constrain Emission Site It is very difficult to constrain the sub-MeV/MeV emission radius R  directly There are three ways to constrain R  – The emission radius of X-ray steep decay phase can be estimated. If R x = R , then R  can be constrained – The emission radius of prompt optical emission R opt can be constrained by the self-absorption limit. If R opt = R , then R  can be constrained – The emission radius of the GeV photons R GeV can be constrained by the pair-production limit. If R GeV = R , then R  can be constrained

Method One: X-Rays t tail Kumar et al. 07 Lyutikov, 06 > cm R ,X > cm If the steep-decay phase of the X-ray tail is defined by the high- latitude emission, one has: GRB tail jjjj R R  j 2 /2

Method Two: Optical Vestrand et al. 2006a,b “Tracking” optical band detection constraints the self- absorption frequency and, hence, the emission radius GRB A R ,opt > several cm Shen & Zhang (08):

GRB B: naked-eye GRB (Racusin et al. 2008) Lightcurve: optical roughly traces gamma-rays Spectrum: two distinct spectral components

Syn + SSC model for GRB B (Racusin et al. 2008; Kumar & Panaitescu 2008) E syn ~20 eV E SSC 1st ~650 keV E SSC 2st ~25 GeV E E 2 N(E) Klein-Nishina cut-off Y ~ 10 Y 2 ~100 R ,opt ~ cm

Method Three: GeV Pair cutoff feature depends on both bulk Lorentz factor (Baring & Harding 1997; Lithwick & Sari 2001) and the unknown emission radius (Gupta & Zhang 2008)  Gupta & Zhang 2008

Radius constraints (Zhang & Pe’er 09) Emission must come from a large radius far above the photosphere.

A Poyting-Flux-Dominated Flow: Kill Three Birds with One Stone Invoking a Poynting flux dominated flow can explain the lack of the three expected features – Non-detection of the pair cutoff feature is consistent with a large energy dissipation radius – Non-detection of the SSC feature is naturally expected, since in a Poynting flux dominated flow, the SSC power is expected to be much less that the synchrotron power – Non-detection of the photosphere thermal component is consistent with the picture, since most energy can be retained in the form of Poynting flux energy rather than thermal energy

Counter-arguments: Hide the thermal component Change R 0 ? – R 0 = c  t ~ 3  10 9 cm (based on the observed minimum variability, and the collapsar scenario) – If R 0 is smaller (10 6 cm - not observed, not favored for a massive star progenitor), the thermal temperature is higher, may be hidden below the non-thermal component. – But it does not work - a Poynting flux dominated flow is still needed to hide the thermal component (Fan 2010)

A Model in the High-σ Regime: The ICMART Model (Internal Collision-induced MAgnetic Reconnection & Turbulence) (Zhang & Yan 2011, ApJ, 726, 90) Basic Assumptions: The central engine launches a high-σ flow. The σ is still ~ (10-100) at R ~ cm. The central engine is intermittent, launching an outflow with variable Lorentz factors (less variable in σ).

(a) Initial collisions only distort magnetic fields (b) Finally a collision triggers fast turbulent reconnection - An ICMART event (a broad pulse in GRB lightcurve) ICMART Model Zhang & Yan (2010)

central engine R ~ 10 7 cm  =  0 >> 1 photosphere R ~ cm    0 early collisions R ~ cm  ~ ICMART region R ~ cm  ini ~  end  1 External shock R ~ cm   1 GRB Distance Scales in the ICMART Model Emission suppressed At most 1/(1+ σ ) energy released At most 1/(1+ σ ) energy released 1 /(1+ σ end ) energy released

GRB ejecta is turbulent in nature Reynold’s number: Magnetic Reynold’s number: Magnetic fields can be highly distorted and turbulent if turbulent condition is satisfied

λL

Turbulent Reconnection is needed to power GRBs In order to reach GRB luminosity, the effective global reconnection rate has to be close to c. Relativistic Sweet-Parker reconnection speed is << c (Lyubarsky 2005). Turbulent reconnection (Lazarian & Vishniac 1999) can increase reconnection speed by a factor L/λ.

Multiple collisions can distort field lines and eventually trigger turbulence in a high-σ flow Required condition from the observations (reach GRB luminosity): Condition for relativistic turbulence (I): relativistic shock Condition for relativistic turbulence (II): relativistic reconnection outflow

Features of the ICMART model Carries the merits of the internal shock model (variability related to central engine) Overcomes the difficulties of the internal shock model (carries the merits of the EM model) – High efficiency ~ 50% – Electron number problem naturally solved (electron number is intrinsically small) – Turbulent heating may overcome fast cooling problem – Amati relation more naturally interpreted (larger R, smaller , easier to have reconnection “avalanche”) – No missing photosphere problem Zhang & Yan (2011)

Two-Component Variability in the ICMART Model slow variability component related to central engine fast variability component related to turbulence Consistent with data: Shen & Song (03) Vetere et al. (06) Poster: H. Gao, B.-B. Zhang & B. Zhang

General picture 15/17 LAT GRBs are Band only 2/17 with extra PL component Applicability of ICMART: at least GRB C, probably most Band-only GRBs