1. GeV GRBs Gabriele Ghisellini With the collaboration of: Giancarlo Ghirlanda, Lara Nava, Annalisa Celotti

2. EGRET – GRB 940217 18 GeV 1.5 hours It lasts much longer It starts during the prompt at lower energies The most energetic photon arrives late Prompt or afterglow? Compactness argument??

3. GRB 090510 ShortShort Very hardVery hard z=0.903z=0.903 Detected by the LAT up to 31 GeV!!Detected by the LAT up to 31 GeV!! Well defined timingWell defined timing Fermi-LAT

4. 0.6s 0.5s Time since trigger (precursor) precursor 8-260 keV 0.26-5 MeV LAT all > 100 MeV > 1 GeV 31 GeV Abdo et al 2009 Delay between GBM and LAT. Due to Lorentz invariance violation?

5. Different component 30 GeV0.1 GeV 1 2 3 3 4 Average Time resolved 0.5-1s  F( ) [erg/cm 2 /s] Energy [keV] Abdo et al 2009 If LAT and GBM radiation are cospatial:  >1000 to avoid photon-photon absorption 090510

6. t dec ~ 0.4 (1+z) (E k53 /n) 1/3 33 33 8/3 second s

7. Different component 30 GeV0.1 GeV 1 2 3 3 4 Average Time resolved 0.5-1s  F( ) [erg/cm 2 /s] Energy [keV] Abdo et al 2009 If LAT and GBM radiation are cospatial:  >1000 to avoid photon-photon absorption If  >1000: deceleration of the fireball occurs early  early afterglow! (see also Kumar & Barniol Duran 2009) 090510

8. No matter the origin of the GeV emission, the bulk Lorentz factor must be large

9. Ghirlanda+ 2010 t2t2t2t2 t -1.5 Fermi-LAT background level T*=0.6s

10. 0.1-1 GeV >1 GeV T-T* [s] Ghirlanda+ 2010 ~MeV and ~GeV emission are NOT cospatial. But the ~GeV emission is… No measurable 0.1-10 GeV delay in arrival time: t delay <0.2 s  Strong limit to quantum gravity  M QG > 4.7 M Planck

11. GG+ 2010 LAT GRBs LAT GRBs

12. GG+ 2010 LAT GRBs LAT GRBs background background

14. short z no z

15. Log  Log  F Log  F    GBM LAT Band PL Time integrated spectra

16.  vs   vs   Log  Log  F Log  F    GBM LAT  -values consistent with Zhang+ 2011

17. The 8 brightest LAT GRBs z=2, assumed z=1, assumed z=2, assumed

18. t - 10/7 Radiative!Radiative! Spectrum and decay: afterglow; L GeV ~L bol The 4 brightest LAT GRBs

19. t - 10/7 Radiative?Radiative? The 4 brightest LAT GRBs

20. e From Beloborodov (2002)

21. e

22. e+ e- e

23. e+ e- e

24. LAT GBM Opt

25. Time [s] Time [s] 0.1 1 10 10 2 10 3 1 10 10 2 10 3

26. Problems Fast variability of the GeV emission (Abdo+ 2009)

27. ”…the observed large amplitude variability on short timescales (≈90 ms) in the LAT data, which is usually attributed to prompt emission, argues against such models.” Abdo+ 2009, ApJ, 706, L138 FERMI observations of GRB 090902B: a distinct spectral component in the prompt and delayed emission 090902B

28. 5s Counts per bin

29. Problems Fast variability of the GeV emission (Abdo+ 2009). No evidence Simultaneous GBM-LAT spikes (Ackermann+ 2011; Zhang+ 2011

30. Ackermann+ 2011 090926A

31. e+ e- e L EC L syn ~ L ,iso,54 R 17  3   B,-1 n 42

32. Problems Fast variability of the GeV emission (Abdo+ 2009). No evidence Simultaneous GBM-LAT spikes (Ackermann+ 2011; Zhang+ 2011 EC scattering of prompt photons? Numbers are ok LAT spectra on the extrapolation of GBM spectra (Zhang+ 2011; with exceptions) if fitted together (but LAT emission lasts longer…) Highest energy photons that arrive after the peak of the LAT light curve are too energetic to be synchro(Piran & Nakar 2010).

33. GG+ 2010 LAT GRBs LAT GRBs 13 GeV 33 GeV

34. Problems Fast variability of the GeV emission (Abdo+ 2009). No evidence Simultaneous GBM-LAT spikes (Ackermann+ 2011; Zhang+ 2011 EC scattering of prompt photons? Numbers are ok LAT spectra on the extrapolation of GBM spectra (Zhang+ 2011; with exceptions) if fitted together (but LAT emission lasts longer…) Highest energy photons that arrive after the peak of the LAT light curve are too energetic to be synchro (Piran & Nakar 2010). Very few, possible additional component (SSC)?

35. Bulk Lorentz factors  =2000  = 630  = 670  = 900

36. t dec ~ 420 (1+z) (E k54 /n) 1/3 22 22 8/3 seconds A factor ~10 3 dimmer in luminosity, but if nearby… GeV detected GRBS could be the ones with the largest Lorentz factors… For smaller  … If pair enrichment is required, GeV detected GRBs could be the ones with E peak (1+z)>m e c 2 If E peak < 511 keV and t -1 : adiabatic because of no pairs

37. Ghirlanda 2009090510 511 keV

38. Conclusions GeV preferentially in E peak >511 keV GRBsGeV preferentially in E peak >511 keV GRBs GeV when  is large  early onset of the afterglow  very brightGeV when  is large  early onset of the afterglow  very bright Large E Aft : helps to understand E prompt /E AftLarge E Aft : helps to understand E prompt /E Aft

40. Internal shocks: relative kinetic energy of the shells External shocks: entire kinetic energy of the fireball Afterglows should be more energetic than the prompt

41. Different component 30 GeV0.1 GeV 1 2 3 3 4 Average Time resolved 0.5-1s  F( ) [erg/cm 2 /s] Energy [keV] Abdo et al 2009 If LAT and GBM radiation are cospatial:  >1000 to avoid photon-photon absorption If  >1000: deceleration of the fireball occurs early  early afterglow! If  >1000: large electron energies  synchrotron afterglow!

42. E afterglow < E prompt E afterglow ~ 0.1 E prompt Willingale+ 2007

43. X-ray and optical often behave differently Late prompt? X-ray optical

44. We expected the opposite, if the efficiency of prompt is ~ 0.1. Why is the afterglow so faint? Can it be hidden in some “unexplored” frequency range, i.e. GeV-TeV?

47. E aft ~ E prompt /10 Willingale+ 2007

48. In GRB 080916C (Abdo et al. 2009a), there is evidence that the spectrum from 8 keV to 10 GeV can be described by the same Band function (i.e. two smoothly connected power laws), suggesting that the LAT flux has the same origin of the low energy flux. On the other hand, the flux level of the LAT emission, its spectrum and its long lasting nature match the expectations from a forward shock, leading Kumar & Barniol–Duran (2009) to prefer the “standard afterglow” interpretation (see also Razzaque, Dermer & Finke 2009 for an hadronic model; Zhang & Peer 2009 for a magnetically dominated fireball model and Zou et al. 2009 for a synchrotron self–Compton origin). In the short bursts GRB 090510 the spectrum in the LAT energy range is not the extrapolation of the flux from lower energies, but is harder, leading Abdo et al. (2009b) to propose a synchrotron self–Compton interpretation for its origin. Instead we (Ghirlanda, Ghisellini & Nava 2009) proposed that the LAT flux is afterglow synchrotron emission, on the basis of its time profile and spectrum (see also Gao et al. 2009; De Pasquale et al. 2009). Finally, the LAT flux of GRB 090902B decays as t−1.5 (Abdo et al. 2009c), it lasts longer than the flux detected by the GBM, and its spectrum is harder than the extrapolation from lower frequen- cies, making it a good candidate for an afterglow interpretation, despite the arguments against put forward by Abdo et al. (2009c), that we will discuss in this paper. Moreover, in GRB 090902B there is evidence of a soft excess (observed in the GBM spectrum below 50 keV) which is spectrally consistent with the extrapolation at these energies of the LAT spectrum. Interpretations

49. Ackermann Ackermann 2010: 090510 coincidental peaks in GBM and LAT. SSC code to explain LAT: disfavored, afterglow has less problems. Confusing. Too many indices. De Pasquale De Pasquale 2010 : 090510 Curva di luce e confronto con Swift Ackermann Ackermann 2011: 090926A: break a 1.4 GeV. Confusione sugli alpha del solo LAT: ripido nel time integrated (come noi) e piattozzo nel time resolved. The delay timescale of the extra spectral component would correspond to the time needed for the forward shock to sweep up material and brighten (Kumar & Barniol Duran 2009; Ghisellini et al. 2010; Razzaque 2010). The rapid variability observed in GRB 090926A is contrary to expectations from an external shock model, unless it is produced by emission from a small portion of the blast wave within the Doppler beaming cone. This could occur, for instance, if the external medium is clumpy on length scale ≈Γf cΔT /(1 + z) ≃ 1012 (Γf /103 )(ΔT /0.2 s) cm, where Γf is the Lorentz factor of the forward shock and ΔT is the pulse duration (Dermer & Mitman 1999; Dermer 2008). Cenko 2010: Cenko 2010: analysis of afterglows of a few LAT bursts. Ioka 2010 Ioka 2010 fa tutto, ma non ho capito nente… Kumar- Barniol Duran 2010: Kumar- Barniol Duran 2010: fanno LAT e resto dell’afterglow, con closure relations… + calcolo del flusso external shock a 100 MeV + confronto 100 MeV / X-ray e ottico. Tutto adiabatico. B molto molto piccolo (1e-5). Dicono che se fosse radiative si sballerebbe l’X early.

50. Kumar Barniol Duran 2009: Kumar Barniol Duran 2009: I primi a dire external shock. Il lavoro e’ complicato. B~2e-5 Gauss, non ho capito perche’. Larsson+ 2011: Larsson+ 2011: There have been many suggestions for the origin of the extra component, including external shocks (Ghisellini et al. 2010; Kumar & Barniol Duran 2010), hadronic processes (Asano et al. 2009; Razzaque 2010), Compton upscattering of a photospheric component (Toma et al. 2010) as well as a combination of different emission mechanisms (Pe’er et al. 2010). Liu 2010: Liu 2010: A partially radiative blast wave model, which though is able to produce a sufficiently steep decay slope, can not explain the broadband data of GRB 090902B. The two-component jet model can. Maxham, Zhang 2011 Maxham, Zhang 2011: Detailed modelling adia/radia: Fit is good only after the peak. I think they do not include pairs. In any case fit is reasonably good, even if not perfect. McBreen+ 2010: McBreen+ 2010: GROND data for 4 LAT: they go on Amati, but not on Ghirlanda (no jet break or too late) Toma+ 2010: Toma+ 2010: photospheric emission scattered by relat. e- in internal shocks (ma come fanno a farla durare piu’ del prompt? E poi anche loo dicno che ci sono problemi nella parte a bassa energia, piu’soft di un BB ma piu’ hard di un sincro coolato).

51. Wang+ 2010: Wang+ 2010: importance of KN: at early times suppresses the IC cooling, at later times it becomes more important  synchro decays faster because at late times it competes with IC. Zhang+ 2011 Zhang+ 2011: strongly favors internal origin: time resolved GBM+LAT fits yield a single component (LAT on the extrapolation of beta). If LAT data are fitted separately, the slopes are all consistent with us within the errors (that they do not give…)