An absence of neutrinos associated with cosmic-ray acceleration in [ggr]-ray bursts
R. Abbasi,M. Ahlers,K. Andeen,J. Auffenberg,M. Baker,S. BenZvi,D. Chirkin,P. Desiati,J. C. Díaz-Vélez,J. P. Dumm,J. Eisch,O. Fadiran,J. Feintzeig,L. Gladstone,S. Grullon,F. Halzen,K. Hoshina,J. Jacobsen,A. Karle,M. Krasberg,N. Kurahashi,H. Landsman,R. Maruyama,M. Merck,R. Morse,A. O’Murchadha,B. Riedel,J. P. Rodrigues,M. Santander,S. Toscano,J. van Santen,Ch. Weaver,C. Wendt,S. Westerhoff,N. Whitehorn,Y. Abdou,M. Carson,F. Descamps,G. de Vries-Uiterweerd,T. Feusels,A. Haj Ismail,D. Ryckbosch,A. Van Overloop,T. Abu-Zayyad,J. Madsen,G. M. Spiczak,M. Ackermann,J. L. Bazo Alba,M. L. Benabderrahmane,J. Berdermann,E. Bernardini,A. H. Cruz Silva,R. Franke,T. Glüsenkamp,D. Góra,K. Han,E. Jacobi,F. Kislat,R. Lauer,E. Middell,R. Nahnhauer,A. Schönwald,C. Spiering,A. Stößl,D. Tosi,M. Walter,R. Wischnewski,J. P. Yanez,J. Adams,A. M. Brown,S. Hickford,J. A. Aguilar,T. Montaruli,D. Altmann,M. Bissok,J. Blumenthal,D. J. Boersma,S. Euler,D. Heinen,B. Hoffmann,J.-P. Hülβ,K. Laihem,L. Paul,A. Schukraft,M. Schunck,M. Vehring,M. Wallraff,C. H. Wiebusch,X. Bai,P. Berghaus,P. A. Evenson,T. K. Gaisser,S. Hussain,T. Kuwabara,B. Ruzybayev,D. Seckel,T. Stanev,A. Tamburro,S. Tilav,C. Xu,S. W. Barwick,J. W. Nam,A. Silvestri,G. Yodh,R. Bay,M. V. D’Agostino,K. Filimonov,L. Gerhardt,C. Ha,S. R. Klein,S. Miarecki,P. B. Price,K. Woschnagg,K. Beattie,A. Goldschmidt,H. S. Matis,D. R. Nygren,G. T. Przybylski,T. Stezelberger,R. G. Stokstad,J. J. Beatty,J. C. Davis,C. Rott,M. Stamatikos,S. Bechet,D. Bertrand,M. Dierckxsens,K. Hanson,D. Heereman,A. Marotta,T. Meures,J. K. Becker,J. Dreyer,A. Fedynitch,M. Olivo,S. Schöneberg,K.-H. Becker,D. Bindig,T. Fischer-Wasels,M. Gurtner,K. Helbing,T. Karg,S. Kopper,U. Naumann,J. Posselt,A. Schultes,B. Semburg,M. Bell,K. S. Caballero-Mora,D. F. Cowen,T. DeYoung,M. Dunkman,D. J. Koskinen,M. J. Larson,P. Mészáros,D. Rutledge,M. W. E. Smith,R. Wasserman,D. Berley,E. Blaufuss,B. Christy,R. W. Ellsworth,J. A. Goodman,R. Hellauer,K. D. Hoffman,W. Huelsnitz,K. Meagher,A. Olivas,P. Redl,M. Richman,T. Schmidt,G. W. Sullivan,H. Wissing,D. Z. Besson,C. Bohm,M. Danninger,C. Finley,S. Flis,P. O. Hulth,K. Hultqvist,H. Johansson,S. H. Seo,C. Walck,M. Wolf,M. Zoll,D. Bose,L. Brayeur,S. Buitink,M. Casier,C. De Clercq,J. Kunnen,M. Labare,A. Rizzo,E. A. Strahler,N. van Eijndhoven,S. Böser,T. Degner,A. Franckowiak,A. Homeier,M. Kowalski,S. Panknin,L. Schulte,M. Stüer,M. Voge,O. Botner,O. Engdegård,A. Hallgren,J. Miller,C. Pérez de los Heros,R. Ström,H. Taavola,F. Clevermann,J.-H. Köhne,N. Milke,D. Pieloth,W. Rhode,T. Ruhe,S. Cohen,M. Ribordy,C. Colnard,A. Groß,S. Odrowski,E. Resconi,O. Schulz,Y. Sestayo,P. Mészáros,S. M. Movit,J. Daughhetee,I. Taboada,A. R. Fazely,S. Ter-Antonyan,X. W. Xu,J. Gallagher,D. Grant,S. C. Nowicki,T. R. Wood,T. Griesel,L. Köpke,T. Kowarik,G. Kroll,J. Lünemann,A. Piegsa,F. Rothmaier,H.-G. Sander,K. Schatto,K. Wiebe,G. C. Hill,A. Ishihara,K. Mase,S. Yoshida,G. S. Japaridze,A. Kappes,H. Kolanoski,T. Waldenmaier,J. Kiryluk,G. Kohnen,K. Rawlins,E. Resconi,S. Sarkar,S. Seunarine,P. A. Toale,D. R. Williams,D. L. Xu,P. Zarzhitsky
DOI: https://doi.org/10.1038/nature11068
IF: 64.8
2012-01-01
Nature
Abstract:Neutrinos from GRBs are produced in the decay of charged pions produced in interactions between high-energy protons and the intense γ-ray background within the GRB fireball, for example in the Δ-resonance process p + γ Δ+ n + π+ (p, proton; γ, photon (here γ-ray); Δ+, delta baryon; n, neutron; π+, pion). When these pions decay via π+ µ+νµ and , they produce a flux of high-energy muon neutrinos (νμ) and electron neutrinos (νe), coincident with the γ-rays, and peaking at energies of several hundred tera-electronvolts (TeV)4, 11 (μ+, antimuon; e+, positron). Such a flux should be detectable using km3-scale instruments like the IceCube neutrino telescope12, 13 (Supplementary Fig. 1). Owing to maximal mixing between muon neutrinos and tau neutrinos, neutrinos from pion decay in and around GRBs will arrive at Earth in an equal mixture of flavours. We focus here only on muons produced in νµ charged-current interactions. As the downgoing cosmic-ray muon background presents challenges for the identification of neutrino-induced muons, we achieve our highest sensitivity for upgoing neutrinos (from sources in the northern sky). However, the tight constraint of spatial and temporal coincidence with a GRB allows some sensitivity even in the southern sky. One of the two analyses presented here therefore includes Southern Hemisphere GRBs during the 2009–10 IceCube run. The results presented here were obtained while IceCube was under construction, using 40 and 59 of the 86 photomultiplier strings of the final detector (Supplementary Fig. 1), which took data from April 2008 to May 2009 and from May 2009 until May 2010, respectively. During the 59-string data-taking period, 190 GRBs were observed and reported by γ-ray observatory satellites via the GRB Coordinates Network14, with 105 in the northern sky. Of those GRBs, 9 were not included in our catalogue owing to detector downtime associated with construction and calibration. Two additional GRBs were included from test runs before the start of the official 59-string run. 117 northern-sky GRBs were included from the 40-string period7 to compute the final combined result. GRB positions were taken from the satellite with the smallest reported error, which is typically smaller than the IceCube resolution. The GRB γ-ray emission start (Tstart) and stop (Tstop) times were taken by finding the earliest and latest time reported for γ-ray emission. As in our previous study7, we conducted two analyses of the IceCube data. In a model-dependent search, we examine data during the period of γ-ray emission reported by any satellite for neutrinos with the energy spectrum predicted from the γ-ray spectra of individual GRBs6, 9. The model-independent analysis searches more generically for neutrinos on wider timescales, up to the limit of sensitivity to small numbers of events at ±1 day, or with different spectra. Both analyses follow the methods used in our previous work7, with the exception of slightly changed event selection and the addition of the Southern Hemisphere to the model-independent search. Owing to the large background of downgoing muons from the southern sky, the Southern Hemisphere analysis is sensitive mainly to higher-energy events (Supplementary Fig. 3). Systematic uncertainties from detector effects have been included in the reported limits from both analyses, and were estimated by varying the simulated detector response and recomputing the limit, with the dominant factor being the efficiency of the detector’s optical sensors. In the 59-string portion of the model-dependent analysis, no events were found to be both on-source and on time (within 10° of a GRB and between Tstart and Tstop). From the individual burst spectra6, 9 with an assumed ratio of energy in protons to energy in electrons εp/εe = 10 (ref. 6), 8.4 signal events were predicted from the combined 2-year data set and a final upper limit (90% confidence) of 0.27 times the predicted flux can be set (Fig. 1). This corresponds to a 90% upper limit on εp/εe of 2.7, with other parameters held fixed, and includes a 6% systematic uncertainty from detector effects. In the model-independent analysis, two candidate events were observed at low significance, one 30 s after GRB 091026A (event 1) and another 14 h before GRB 091230A (most theories predict neutrinos within a few minutes of the burst). Subsequent examination showed they had both triggered several tanks in the IceTop surface air shower array, and are thus very probably muons from cosmic-ray air showers. In Fig. 2 are shown limits from this analysis on the normalization of generic power-law muon neutrino spectra expected from shock acceleration at Earth as a function of the size of the time window |Δt|, which is the difference between the neutrino arrival time and the first reported satellite trigger time. As a cross-check on both results, the limit from this analysis on the average individual burst spectra6, 9 during the time window corresponding to the median duration of the bursts in the sample (28 s) was 0.24 times the predicted flux, within 10% of the model-dependent analysis. Assuming that the GRBs in our catalogue are a representative sample of a total of 667 per year (ref. 7), we can scale the emission from our catalogue to the emission of all GRBs. The resulting limits can then be compared to the expected neutrino rates from models that assume that GRBs are the main sources of ultra-high-energy cosmic rays4, 8, 10, with sampling biases of the same order as model uncertainties in the flux predictions15, 16. Limits from the model-independent analysis on fluxes of this type are shown in Fig. 3. These limits exclude all tested models4, 8, 9, 10 with their standard parameters and uncertainties on those parameters (Figs 1, 3). The models are different formulations of the same fireball phenomenology, producing neutrinos at proton–photon (p–γ) interactions in internal shocks. The remaining parameter spaces available to each model therefore have similar characteristics: either a low density of high-energy protons, below that required to explain the cosmic rays, or a low efficiency of neutrino production. In the GRB fireball, protons are believed to be accelerated stochastically in collisions of internal shocks in the expanding GRB. The neutrino flux is proportional to the rate of p–γ interactions, and so to the proton content of the burst by a model-dependent factor. Assuming a model-dependent proton ejection efficiency, the proton content can in turn be related to the measured flux of high-energy cosmic rays if GRBs are the cosmic-ray sources. Limits on the neutrino flux for cosmic-ray-normalized models are shown in Fig. 3; each model prediction has been normalized to a value consistent with the observed ultra-high-energy cosmic-ray flux. The proton density can also be expressed as a fraction of the observed burst energy, directly limiting the average proton content of the bursts in our catalogue (Fig. 4). An alternative is to reduce the neutrino production efficiency, for example by modifying the physics included in the predictions16, 17 or by increasing the bulk Lorentz boost factor, Γ. Increasing Γ increases the proton energy threshold for pion production in the observer frame, thereby reducing the neutrino flux owing to the lower proton density at higher energies. Astrophysical lower limits on Γ are established by pair production arguments9, but the upper limit is less clear. Although it is possible that Γ may take values of up to 1,000 in some unusual bursts, the average value is probably lower (usually assumed6, 9 to be around 300) and the non-thermal γ-ray spectra from the bursts set a weak constraint that ≲2,000 (ref. 18). For all considered models, with uniform fixed proton content, very high average values of Γ are required to be compatible with our limits (Figs 3, 4). In the case of models where cosmic rays escape from the GRB fireball as neutrons8, 10, the neutrons and neutrinos are created in the same p–γ interactions, directly relating the cosmic-ray and neutrino fluxes and removing many uncertainties in the flux calculation. In these models, Γ also sets the threshold energy for production of cosmic rays. The requirement that the extragalactic cosmic rays be produced in GRBs therefore does set a strong upper limit on Γ: increasing it beyond ~3,000 causes the proton flux from GRBs to disagree with the measured cosmic-ray flux above 4 × 1018 eV, where extragalactic cosmic rays are believed to be dominant. Limits on Γ in neutron-origin models from this analysis (≳2000, Fig. 3) are very close to this point, and as a result all such models—in which all extragalactic cosmic rays are emitted from GRBs as neutrons—are now largely ruled out. Although the precise constraints are model-dependent, the general conclusion is the same for all the versions of fireball phenomenology we have considered here: either the proton density in GRB fireballs is substantially below the level required to explain the highest-energy cosmic rays or the physics in GRB shocks is significantly different from that included in current models. In either case, our current theories of cosmic-ray and neutrino production in GRBs will need to be revisited. Download references We acknowledge support from the following agencies: US NSF, Office of Polar Programs, and US NSF, Physics Division; University of Wisconsin Alumni Research Foundation; the GLOW and OSG grids; US DOE, NERSCC; the LONI grid; NSERC, Canada; Swedish Research Council, Swedish Polar Research Secretariat, SNIC, K. and A. Wallenberg Foundation, Sweden; German Ministry for Education and Research, Deutsche Forschungsgemeinschaft; Research Department of Plasmas with Complex Interactions (Bochum), Germany; FSR, FWO Odysseus, IWT, BELSPO, Belgium; University of Oxford, UK; Marsden Fund, New Zealand; Australian Research Council; JSPS, Japan; SNSF, Switzerland. J.P.R was supported by the Capes Foundation, Brazil; N.W. by the NSF GRFP. We thank S. Hümmer, E. Waxman and W. Winter for discussions. Subscribe to comments