1: %\documentclass[preprint]{aastex}
2: %\documentclass[manuscript]{aastex}
3: \documentclass[12pt,preprint]{aastex}
4: %\usepackage{natbib}
5: %\usepackage{epsfig}
6: \def\comma{,~}
7:
8: \slugcomment{Version of 7 May 2007}
9: \shorttitle{Milagro upper limits on VHE emission from short GRBs}
10: \shortauthors{Atkins et al.}
11:
12: \begin{document}
13:
14: \title{Milagro Constraints on Very High Energy Emission from
15: Short Duration Gamma-Ray Bursts}
16: \author{
17: A.~A.~Abdo,\altaffilmark{\ref{msu}}
18: B.~T.~Allen,\altaffilmark{\ref{uci}}
19: D.~Berley,\altaffilmark{\ref{umcp}}
20: E.~Blaufuss,\altaffilmark{\ref{umcp}}
21: S.~Casanova,\altaffilmark{\ref{lanl}}
22: B.~L.~Dingus,\altaffilmark{\ref{lanl}}
23: R.~W.~Ellsworth,\altaffilmark{\ref{georgemason}}
24: M.~M.~Gonzalez,\altaffilmark{\ref{unam}}
25: J.~A.~Goodman,\altaffilmark{\ref{umcp}}
26: E.~Hays,\altaffilmark{\ref{umcp},\ref{hayscurrent1},\ref{hayscurrent2}}
27: C.~M.~Hoffman,\altaffilmark{\ref{lanl}}
28: B.~E.~Kolterman,\altaffilmark{\ref{nyu}}
29: C.~P.~Lansdell,\altaffilmark{\ref{umcp}}
30: J.~T.~Linnemann,\altaffilmark{\ref{msu}}
31: J.~E.~McEnery,\altaffilmark{\ref{mcenerycurrent}}
32: A.~I.~Mincer,\altaffilmark{\ref{nyu}}
33: P.~Nemethy,\altaffilmark{\ref{nyu}}
34: D.~Noyes,\altaffilmark{\ref{umcp}}
35: J.~M.~Ryan,\altaffilmark{\ref{unh}}
36: F.~W.~Samuelson,\altaffilmark{\ref{fda}}
37: P.~M.~Saz~Parkinson,\altaffilmark{\ref{ucsc}}
38: A.~Shoup,\altaffilmark{\ref{osu}}
39: G.~Sinnis,\altaffilmark{\ref{lanl}}
40: A.~J.~Smith,\altaffilmark{\ref{umcp}}
41: G.~W.~Sullivan,\altaffilmark{\ref{umcp}}
42: V.~Vasileiou,\altaffilmark{\ref{umcp}}
43: G.~P.~Walker,\altaffilmark{\ref{lanl}}
44: D.~A.~Williams,\altaffilmark{\ref{ucsc}}
45: X.~W.~Xu\altaffilmark{\ref{lanl}}
46: and
47: G.~B.~Yodh\altaffilmark{\ref{uci}}}
48:
49: \altaffiltext{1}{\label{msu} Department of Physics and Astronomy, Michigan State University, 3245 BioMedical Physical Sciences Building, East Lansing, MI 48824}
50: \altaffiltext{2}{\label{uci} Department of Physics and Astronomy, University of California, Irvine, CA 92697}
51: \altaffiltext{3}{\label{umcp} Department of Physics, University of Maryland, College Park, MD 20742}
52: \altaffiltext{4}{\label{lanl} Group P-23, Los Alamos National Laboratory, P.O. Box 1663, Los Alamos, NM 87545}
53: \altaffiltext{5}{\label{georgemason} Department of Physics and Astronomy, George Mason University, 4400 University Drive, Fairfax, VA 22030}
54: \altaffiltext{6}{\label{unam} Instituto de Astronom\'{i}a, Universidad Nacional Aut\'{o}noma de M\'{e}xico, D.F., M\'{e}xico, 04510}
55: \altaffiltext{7}{\label{hayscurrent1} Current address: High Energy Physics Division, Argonne National Laboratory, Argonne, IL 60439}
56: \altaffiltext{8}{\label{hayscurrent2} Current address: Enrico Fermi Institute, University of Chicago, Chicago, IL, 60637}
57: \altaffiltext{9}{\label{mcenerycurrent} NASA Goddard Space Flight Center, Greenbelt, MD 20771}
58: \altaffiltext{10}{\label{nyu} Department of Physics, New York University, 4 Washington Place, New York, NY 10003}
59: \altaffiltext{11}{\label{unh} Department of Physics, University of New Hampshire, Morse Hall, Durham, NH 03824} %-3525}
60: \altaffiltext{12}{\label{fda} Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, U.S. Food and Drug Administration.}
61: \altaffiltext{13}{\label{ucsc} Santa Cruz Institute for Particle Physics, University of California, 1156 High Street, Santa Cruz, CA 95064}
62: \altaffiltext{14}{\label{osu} Ohio State University, Lima, OH 45804}
63:
64: \begin{abstract}
65:
66: Recent rapid localizations of short, hard gamma-ray bursts (GRBs) by the Swift and HETE
67: satellites have led to the observation of the first afterglows and the measurement of the first
68: redshifts from this type of burst~\citep{2005Natur.437..845F,2005Natur.437..851G,
69: 2005Natur.437..855V,2005Natur.438..988B, 2005Natur.438..994B}.
70: Detection of $>$100 GeV counterparts would place powerful constraints on GRB mechanisms.
71: Seventeen short duration ($<$5 s) GRBs detected by satellites occurred within the field of view of
72: the Milagro gamma-ray observatory between 2000 January and 2006 December. We have searched the
73: Milagro data for $>$100 GeV counterparts to these GRBs and find no significant emission
74: correlated with these bursts.
75: Due to the absorption of high-energy gamma rays by the extragalactic background light (EBL),
76: detections are only expected for redshifts less than $\sim$0.5. While most long
77: duration GRBs occur at redshifts higher than 0.5, the opposite is thought to be true of short GRBs.
78: Lack of a detected VHE signal thus allows setting meaningful fluence limits.
79: One GRB in the sample (050509b) has a likely association with a galaxy
80: at a redshift of 0.225, while another (051103) has been tentatively linked to the nearby galaxy
81: M81. Fluence limits are corrected for EBL absorption, either using the known measured
82: redshift, or computing the corresponding absorption for a redshift of 0.1 and 0.5, as well as
83: for the case of z=0.
84:
85: \end{abstract}
86:
87: \keywords{gamma rays: bursts --- gamma rays: observations}
88:
89: Gamma-ray bursts (GRBs) have long been classified by their
90: durations into long and short bursts~\citep{1981Ap&SS..75...47M,1984Natur.308..434N}.
91: Later classification schemes took into account the combination of both the temporal and spectral
92: properties~\citep{1993ApJ...413L.101K} leading to what are currently known as
93: short, hard bursts and long, soft bursts. Recent classification schemes list as
94: many as ten different criteria to try and distinguish between these two
95: populations~\citep{2006astro.ph..5570D}. The fraction of bursts that fall in
96: each category is instrument-dependent, with BATSE finding approximately 25\% of bursts to be
97: ``short''\citep{1999ApJS..122..465P}, while the equivalent fraction for Swift is closer to
98: 10\%~\citep{2006AIPC..838...14G}. The discovery of the first X-ray afterglow from a long duration
99: GRB~\citep{1997Natur.387..783C} led to a rapid string of observations validating the fireball
100: shock model of GRBs~\citep{1992MNRAS.258P..41R,1993ApJ...405..278M}, culminating in the
101: observation of a GRB-supernova association~\citep{2003Natur.423..847H,2003ApJ...591L..17S}
102: confirming that at least some GRBs are related to the deaths of massive stars, as predicted by
103: the ``collapsar'' model~\citep{1993ApJ...405..273W}.
104:
105: Until recently, however, all the observations of afterglows (and therefore, most of the
106: understanding about GRBs) came from long duration GRBs. The first detection of the afterglow of
107: a short, hard burst -- for GRB 050509b~\citep{2005Natur.437..851G} -- was followed by
108: others~\citep{2005Natur.437..845F,2005Natur.437..855V,2005Natur.438..988B}, and there are now
109: approximately half a dozen measured redshifts for short, hard bursts~\citep{2006AIPC..838...25H}.
110: Although some of these redshifts are less secure than others, their average ($\sim$ 0.3--0.5)
111: is significantly lower than the typical redshift of long duration bursts. The location of several
112: of these short bursts in old galaxies with little star formation, unlike the association of long
113: GRBs with active star-forming regions, seems to rule out the collapsar model for these bursts and
114: favors instead merger models involving binary neutron stars or black hole-neutron star systems as
115: the progenitors for these bursts. One predicted consequence of these
116: models~\citep{2006ApJ...650..998R} is that the neutron-rich outflows expected from these mergers
117: would lead to pion decay photons at $\sim$60 GeV which could be detected by Milagro.
118:
119: The detection
120: of gamma rays in the GeV-TeV regime is affected by the extragalactic infrared background light
121: (EBL)~\citep{nikishov61}. The amount of gamma-ray absorption due to the EBL
122: is not well determined, though it is a strong function of redshift and energy. One
123: model~\citep{primack05}, recently validated by HESS observations~\citep{2006Natur.440.1018A},
124: predicts an optical depth of roughly unity to 500 GeV (10 TeV) gamma rays from a redshift of
125: 0.2 (0.05). The significantly lower redshift of short duration GRBs compared to long duration
126: ones makes them particularly suitable candidates for very high energy (VHE) emission studies, such
127: as possible with the Milagro detector. On the other hand, their much lower luminosity means their
128: possible emission at higher energies is also expected to be substantially lower than the brighter,
129: long duration bursts.
130:
131: Previous searches for VHE emission from GRBs, both long and short, have produced no conclusive
132: detection to date. Milagrito, a prototype of Milagro, reported evidence for emission above
133: 650 GeV from GRB 970417a, with a (post-trials) probability of 1.5$\times10^{-3}$ of being a
134: background fluctuation~\citep{atkins00a,atkins03}. More recent Milagro searches have yielded no
135: conclusive detection~\citep{2005ApJ...630..996A,2006astro.ph.11457S}.
136: Evidence at about the 3 sigma level from the
137: HEGRA AIROBICC array has been published for emission above 20 TeV from GRB 920925c~\citep{padilla98}.
138: Follow-up observations above 200 GeV by the Whipple atmospheric Cherenkov
139: telescope~\citep{connaughton97,2007ApJ...655..396H} did not find any high energy afterglow from
140: the GRBs observed. Recently, the MAGIC group have reported upper limits on the gamma-ray flux
141: in the 85-1000 GeV energy range from the 9 GRBs\footnote[1]{Unfortunately, 4 out of the 9 GRBs that
142: MAGIC observed had measured redshifts in excess of 3.5, making it virtually impossible for any VHE
143: gamma rays to reach Earth.} they observed in their first year of operations, including the afterglow of the
144: the short duration HETE burst 060121~\citep{2006astro.ph.12548A}. The MAGIC list includes
145: GRB 050713a, for which they had the fastest response so far, beginning their observations 40 s
146: after the burst onset~\citep{2006ApJ...641L...9A}.
147: Because searches carried out with atmospheric Cherenkov telescopes, like MAGIC or Whipple, involve
148: slewing a telescope to the right location in the
149: sky and are limited by their relatively small fields of view and duty cycles, Milagro is the best suited
150: instrument for observing the shortest GRBs at very high energies.
151:
152: In this paper we place limits on the VHE emission from short duration\footnote[2]{The term ``short
153: duration'' is used in the paper to refer only to the duration of the burst being less than 5 seconds,
154: while the term ``short, hard'' burst is used in the usual more narrow sense found in the literature,
155: based on the timing and spectral properties of the burst, as introduced by
156: ~\cite{1993ApJ...413L.101K}.} GRBs which might help constrain models of their progenitors. We
157: selected all known bursts detected by satellites which occurred
158: in the Milagro field of view and had a duration of 5 seconds or less. This duration was
159: chosen, rather than 2 seconds, in part due to the recent work of \cite{2006astro.ph..5570D}, but
160: also in order to be more inclusive. In the following
161: section we describe the detector, Milagro, which was used to perform the search. We describe in
162: some detail the new, low-energy-threshold trigger which was especially designed to increase
163: Milagro's sensitivity to GRB detections. In section 3, the sample of short duration GRBs analyzed
164: in the paper is presented, with a special emphasis on GRB 050509b, the most promising candidate
165: in the sample. Section 4 describes the analysis carried out to search for emission, both prompt
166: and delayed. Finally, in Section 5 we discuss the main results and summarize our conclusions.
167:
168: \section{The Milagro Observatory}
169:
170: Milagro is a TeV gamma-ray detector which uses the water Cherenkov technique to detect
171: extensive air showers produced by VHE gamma rays as they traverse the Earth's
172: atmosphere~\citep{atkins00b}. Milagro is located in the Jemez Mountains of northern
173: New Mexico (35.9$^\circ$ N, 106.7$^\circ$ W) at an altitude of 2630 m above sea level,
174: and has a field of view of $\sim$2 sr and a duty cycle of over 90\%, making it an
175: ideal all-sky monitor of transient phenomena at very high energies, such as GRBs.
176: The effective area and energy threshold of Milagro are a function of
177: zenith angle, due to the increased atmospheric overburden at larger zenith
178: angles, which tends to attenuate the particles in the air shower before they reach the ground.
179: The sensitivity of Milagro varies slowly with zenith angle from 0 to $\sim$30 degrees and then
180: decreases more rapidly~\citep{2005ApJ...630..996A}.
181:
182: For the data sample used in this analysis, the typical single shower angular resolution is approximately
183: 0.7 degrees; however, at lower energies there are fewer photomultiplier tubes hit so the angular
184: resolution is about 1 degree. The energy response of Milagro is rather broad
185: with no clear point to define as an instrument threshold. To obtain a rough guide of the range of
186: energies to which Milagro is sensitive, we consider a
187: power-law spectrum with a differential photon index, $\alpha$, of -2.4. The energy (E$_{5}$) above
188: which 95\% of the triggered events from such a spectrum are obtained is approximately 350 GeV,
189: the energy (E$_{95}$) below which 95\% of the triggered events occur is 30 TeV, and the median
190: energy is 3 TeV. This illustrates the breadth of the energy response of Milagro, showing that
191: the Milagro detector has significant sensitivity below energies of several hundred GeV.
192:
193: The Milagro sensitivity as a function of energy can be understood as a simple
194: consequence of one dimensional cascade shower theory.
195: The fluctuations in the amount of energy reaching a certain detector level from a gamma-ray
196: shower arise primarily because
197: of variations in the depth of the first interaction, which follows a probability
198: distribution $P\sim e^{-\frac{9}{7}X}$, where $X$ is the depth of the interaction in radiation lengths.
199: According to Approximation B~\citep{1941RvMP...13..240R}, after shower maximum ($>$10 km above
200: sea level for the energy range of Milagro, well above the altitude of the Milagro detector),
201: the average number of particles in a gamma-ray shower, as well
202: as the amount of energy, decreases exponentially as shower particles
203: are absorbed by the atmosphere.
204: From the longitudinal shower profile obtained in Approximation B,
205: the number of radiation lengths deeper in the atmosphere, $X$, which a gamma-ray
206: of energy $E$ must penetrate in order to deposit energy at the ground equivalent to that of a typical
207: shower of higher energy $E_{thr}$ is given by $X\simeq2\,ln(E_{thr}/E)$.
208: So the probability that a gamma ray shower of energy $E$ has a certain minimum amount of energy reaching
209: the ground is given approximately by $P(E)\sim(\frac{E}{E_{thr}})^{2.6}$.
210: In other words, the low energy effective area scales like a power law in energy. Figure~\ref{fig1}
211: shows that the effective area of Milagro does, indeed, follow this power law.
212: As seen from Figure~\ref{fig1}, the ratio of the effective area at 100 GeV vs 1 TeV is $\sim$0.005,
213: roughly what is predicted by the previous formula. The effective area of Milagro at a median energy
214: of $\sim$4 TeV has been confirmed by the measurement of the flux from the Crab, in agreement with
215: atmospheric Cherenkov telescope measurements. For more details on Milagro see \cite{atkins03b}.
216:
217: During the period covered by these observations, the Milagro trigger underwent a significant
218: upgrade. Until 2002, the Milagro trigger consisted of a simple multiplicity count of
219: the number of photomultiplier tubes hit out of the 450 in the top layer of
220: the pond. This threshold was set to between 50 and 70 tubes hit within a 200 ns
221: time window to maintain the trigger rate at $\sim$1400-1600 Hz, the maximum sustained rate that can
222: be handled by the Milagro data acquisition system with a reasonable deadtime ($\sim$5\%)
223: \footnote{The deadtime to record single triggers depends instead on the digitization time, which
224: scales with the number of hit PMTs, and is $<$50 $\mu$s. Triggers separated by as little
225: as 30 $\mu$s are routinely recorded.}.
226: A lower trigger threshold would lower the energy
227: threshold of Milagro, thus making it more sensitive to GRBs. Based on the knowledge
228: that most of the increase in the rate as the multiplicity requirement is lowered comes from single
229: muon events which produce enough light to trigger the instrument but cannot be fit to a
230: shower plane, a new programmable trigger was custom-designed for Milagro.
231: It is known from Monte Carlo simulations that gamma-ray events can be
232: reconstructed with as few as 20 tubes hit. A high angle muon traveling
233: across the pond nearly horizontally produces light which arrives over
234: a longer time period than the shower particles, so by making a cut on the
235: time development of the event, it is possible to eliminate these muon events. A custom
236: VME trigger module was built, allowing the use of multiple
237: trigger conditions and including the rise time of the pulse representing the number of
238: struck tubes in the top layer as one of the triggering criteria. The new trigger
239: greatly increased the number of low energy showers detected, while maintaining
240: a manageable overall trigger rate and dead time. Figure~\ref{fig1} shows the effective
241: area of Milagro to gamma rays as a function of energy for three different zenith angles.
242: Figure~\ref{fig2} shows the significant increase in sensitivity gained from the new
243: trigger, relative to the old simple multiplicity trigger, especially at energies below
244: 100 GeV, where detection of GRBs is most likely. The VME trigger was installed in January
245: 2002 and became fully operational on 19 March 2002. The column labeled ``Notes'' in
246: Table~\ref{grb_table} identifies the bursts in our sample for which the VME trigger was in
247: operation.
248:
249: \section{The GRB sample}
250: There is no sharp cutoff point between long duration and short
251: duration bursts; these two populations of GRBs have overlapping distributions in duration.
252: Although earlier studies determined that an effective T90 (duration over which the cumulative
253: counts over the background increase from 5\% to 95\% of the total) cut for separating short
254: from long bursts should be approximately 2 seconds~\citep{1993ApJ...413L.101K}, more recent
255: work~\citep{2006astro.ph..5570D} suggests that bursts shorter than {\em five} seconds
256: have a higher probability of belonging to the short duration class than the long
257: duration one, so we have chosen to include GRBs with
258: durations up to 5 seconds in this list of ``short duration'' bursts.
259:
260: In the 7 years since Milagro began operations (2000 January to 2006 December), there have
261: been approximately 100 known GRBs detected by satellites which have been in the Milagro field
262: of view. Of these, seventeen had measured durations of five seconds or shorter. Many of the bursts
263: in this study were detected by the Interplanetary Network
264: (IPN\footnote[3]{See {\tt http://www.ssl.berkeley.edu/ipn3/}}), and their locations were not
265: immediately known to experiments on the ground, making it very unlikely that a redshift
266: could be determined. More recent bursts detected by Swift and HETE have benefited from
267: extensive multi-wavelength observations from the ground and are therefore far better
268: studied. One burst in our sample (GRB 001204) was obtained from the BeppoSAX
269: GRBM catalog~\citep{guidorzi}.
270:
271: Table~\ref{grb_table} lists the sample of 17 bursts that we
272: analyzed for this paper. Four of the bursts in the sample (000330, 000408, 000424, and 010104)
273: were presented in an earlier paper summarizing the first two years of Milagro observations
274: of GRBs~\citep{2005ApJ...630..996A} and are included here for completeness. One of these bursts
275: (GRB 010104) has recently been found to have occurred at a significantly different location
276: than previously thought~\citep{hurley}, so we take this opportunity to present our results on
277: this burst at the new location.
278: The first column of the table gives the GRB name, which, following the usual convention,
279: represents the UTC date (YYMMDD) on which the burst took place. The second column gives the
280: instrument(s) that detected the burst. We list the IPN as an instrument, although it consists of a
281: network of many satellites, a different set of which may detect any given burst.
282: The third column gives the time of the burst,
283: represented by the UTC second of the day. Column four gives the coordinates (right ascension
284: and declination, in degrees) of the burst. All the bursts listed in the table except for one
285: (GRB 000330) were localized to an error region significantly smaller than the Milagro angular resolution.
286: For GRB 000330, the position error was approximately 5 degrees, so the upper limit was computed using
287: the most significant bin within that region, as described in~\cite{2005ApJ...630..996A}.
288: For one burst, GRB 000607, the coordinates are not known unambiguously;
289: the IPN sometimes determines two possible error regions and in this case only one of them was in
290: the field of view of Milagro.
291: The fifth column gives the duration of the burst, as reported by the different instrument teams.
292: Column six lists the zenith angle of the burst at Milagro, in degrees. We include only bursts for
293: which the zenith angle was less than approximately 50$^{\circ}$.
294: The effective area of Milagro at zenith angles greater than 50$^{\circ}$ becomes negligible in
295: the energy range where we expect GRB emission to be detectable (e.g. $<$ 1 TeV). Column seven
296: gives the value of the redshift, if measured.
297:
298: For those bursts with no measured redshift, we take into account the effect of absorption in
299: computing the upper limits by considering two different redshifts: z=0.5 and z=0.1. We also
300: give limits for the case z=0 (i.e. nearby bursts).
301: By their very nature, short duration bursts are much more difficult to localize than long
302: duration bursts. In addition to being very brief events, they also tend to be much less
303: luminous than long duration GRBs, making it much more challenging to obtain
304: redshifts from these bursts than from long GRBs. GRB 040924, detected by
305: HETE~\citep{Fenimore}, was the first short duration burst to have a measured
306: redshift~\citep{wiersema}, although its spectrum was considered
307: too soft to be part of the short, hard population and it has been speculated that it may belong
308: to the short duration tail of the long duration GRB population~\citep{2005ApJ...628L..93H}.
309: GRB 050509b was the first short, hard burst for which an afterglow was detected. As it is the
310: most interesting burst in the sample, we describe it in more detail in the following paragraph.
311: The remaining columns of Table~\ref{grb_table} present the Milagro results, which we describe
312: later.
313:
314: The detection of an X-ray afterglow from GRB 050509b by Swift~\citep{2005Natur.437..851G}
315: represented the first time such an event had been observed from a short, hard burst. A
316: low probability ($\sim5\times10^{-3}$) of chance alignment suggests that this burst may be
317: associated with a bright elliptical galaxy at a redshift of 0.225~\citep{2006ApJ...638..354B}.
318: Subsequent detections of short, hard bursts~\citep{2005Natur.438..994B} have made this
319: association more plausible and point to an origin of these bursts in
320: regions of low star formation, thus disfavoring the collapsar model invoked for explaining the
321: long duration bursts. At 10 degrees, the zenith angle of this burst is the most favorable
322: in the list of 17 short bursts, and one of the most favorable of all bursts to have occurred in
323: the Milagro field of view. Its redshift of 0.225 is the second or third lowest of those GRBs
324: with known
325: redshift in the Milagro field of view (depending on whether or not one believes GRB 051103 is
326: associated with the nearby satellite galaxy M81), again, making it a very promising candidate.
327: The 15--150 keV fluence of this burst, however, was measured by Swift to be
328: (9.5 $\pm$ 2.5) $\times 10^{-9}$ erg cm$^{-2}$, making it one of the dimmest bursts
329: detected by Swift~\citep{2005Natur.437..851G} and about forty times dimmer than the
330: next dimmest short duration burst in this sample. If the VHE emission of GRBs scales with the
331: fluence measured at the lower energies, this would dampen significantly the expectations of
332: detecting such emission from this burst.
333:
334: \section{Data Analysis}
335:
336: A search for an excess of events above those expected from the background was
337: made for each of the 17 bursts in the sample. The total number of events falling within
338: a circular bin of radius $1.6^{\circ}$ at the location of the burst was summed for the
339: duration of the burst. An estimate of the number of background events was then made by
340: characterizing the angular distribution of the background using two hours of data
341: surrounding the burst, as described in ~\cite{atkins03b}. Figure~\ref{fig3} shows the
342: rate of background events detected by Milagro in a $1.6^{\circ}$ bin as a function of
343: zenith angle. This background rate is a function of the trigger settings and the particular
344: conditions of the detector on the given day and varies slightly from burst to burst.
345: The significance of the excess (or deficit) for each burst was evaluated using
346: equation [17] of~\cite{lima}. The 99\% confidence upper limits on the number of signal
347: events detected, $\mathrm{N_{UL}}$, given the observed $\mathrm{N_{ON}}$ and the predicted
348: background $\mathrm{N_{OFF}}$, is computed using the Feldman-Cousins
349: prescription~\citep{feldman-cousins}. This upper limit on the number of gamma-ray events is then
350: converted into an upper limit on the fluence. Using the effective area of Milagro,
351: $A_{eff}$, and assuming a differential power-law photon spectrum, we integrate in the
352: appropriate energy range and solve for the normalization constant. We chose a spectrum of the
353: form $dN/dE=KE^{-2.4}$ photons/TeV/m$^2$. The spectrum of a GRB has never been measured
354: above 100 GeV, so we must make an assumption of a suitable spectrum for
355: evaluating the limits. The average spectrum of the four brightest bursts observed by EGRET has a
356: differential power law spectrum with index 1.95$\pm$0.25 over the energy range 30 MeV to 10 GeV,
357: showing no sign of a cut-off, though only 4 gamma rays were detected above 1 GeV~\citep{dingus01}.
358: The choice of 2.4 as the spectral index in the Milagro energy range allows for some softening of
359: the spectrum at higher energy.
360:
361: The normalization factor $K$ can be calculated by solving the equation
362: $N_{UL}=\int{A_{eff}(dN/dE)e^{-\tau_{EBL}}dE}$, where $\tau_{EBL}$ represents the optical depth
363: due to the EBL. Finally, we integrate the photon spectrum
364: multiplied by the energy to obtain the corresponding value for the total fluence:
365: $F=\int{E(dN/dE)dE}$, integrating from 0.05 to 5 TeV. For bursts of known
366: (albeit uncertain) redshift (040924, 050509b, 051103, and 051221a), we use the
367: optical depths predicted by~\cite{primack05} and take these into account in computing the preceding
368: integrals, thus obtaining a more realistic upper limit which factors in the correct absorption
369: due to the EBL. For the remaining bursts, we compute the upper limits assuming three possible
370: values of the redshift: 0.5, 0.1, and 0.0.
371:
372: In addition to searching for prompt emission from these bursts, we also searched for extended
373: emission over a period of 312 seconds following the reported trigger time. This timescale is
374: motivated both by the observations of late-time (several hundred seconds after the GRB trigger)
375: X-ray flares during some GRB afterglows~\citep{2006ApJ...641.1010F,2005Sci...309.1833B},
376: which are predicted by some to emit in the GeV-TeV regime (e.g.~\cite{2006ApJ...641L..89W}), as well as
377: by the discovery of a second higher energy component in GRB 941017. While the T90 for that burst
378: was 77 s, the second, higher energy component (which has a fluence more than three times greater
379: than the fluence in the BATSE energy range alone) had a duration of approximately 211
380: seconds~\citep{gonzalez03}.
381:
382: \section{Results and Discussion}
383:
384: None of the bursts in the sample showed significant VHE emission, either prompt or
385: delayed. Column nine of Table~\ref{grb_table} gives the 99\% upper limits on the fluence, computed
386: as described in the previous section over the duration (given in column five) of the burst. For
387: comparison, we give the measured fluence in the keV band in column eight. Most models of
388: VHE emission predict it should be correlated to the lower energy emission. In column
389: ten, we give the 99\% upper limits on the fluence computed over a duration of
390: 312 seconds from the trigger time.
391:
392: The localization of several short, hard bursts to old, low-star-forming galaxies has led
393: to the speculation that their origins may be related to binary
394: mergers, possibly double neutron star systems or black-hole neutron star binaries.
395: \cite{2006ApJ...650..998R} propose that in such a scenario, the accretion of neutron star material
396: would lead to the emission of a neutron-rich jet, which would emit $\pi^0$ decay photons in the
397: 100 GeV range. Several parameters and assumptions are important in this model, including the
398: total isotropic-equivalent energy outflow of the burst, the total energy to mass flow ratio,
399: $\eta$, and the initial neutron to proton number density ratio, $\xi_0$.
400:
401: Of the bursts considered in the sample, GRB 050509b is the most promising candidate, given its
402: known low redshift and its optimal zenith angle at Milagro. The attenuation due to the IR
403: background in this case is not very significant. Using the \cite{primack05} model, the
404: corresponding optical depth for the resulting 60 GeV photons at z=0.225 would be $\sim$0.04,
405: leading to an attenuation of less than 5\%. Using the \cite{2006ApJ...650..998R} model with their
406: standard parameters, $\eta$=316, and $\xi_0$=10, and using the measured isotropic luminosity in
407: gamma rays, $E_{iso}$, the predicted flux from this GRB would be
408: 2.3$\times10^{-7}$ cm$^{-2}$s$^{-1}$ gamma rays of energy $\sim$60 GeV~\citep{razzaque}. The
409: effective area of Milagro is approximately 90,000 cm$^2$ at 60 GeV for the given zenith angle
410: of this burst, yielding approximately 0.02 per second, or less than 3$\times10^{-3}$ events for
411: the duration (0.128s) of the burst, making this burst clearly undetectable. The next best
412: candidate is GRB 061210. Despite having a much larger $E_{iso}$ than 050509b (about twenty
413: times larger), this burst, assuming a redshift of 0.41 has a predicted flux of pion-decay photons
414: comparable to 050509b~\cite{razzaque}. Given the less favorable zenith angle of this burst
415: and the fact that the VME trigger was not operating at the time this burst took place, the effective
416: area of Milagro for these events is approximately an order of magnitude lower than for the case of
417: 050509b. As discussed below, GRB 051103 might have been an SGR outburst in M81. If it is not
418: an SGR outburst, but a binary merger at very low redshift, the model by \cite{2006ApJ...650..998R}
419: would predict a significant detection of this burst in Milagro, had it
420: occurred at a zenith angle $\leq$20 degrees, instead of at 50 degrees. This is despite having
421: a very low $E_{iso}$, more than an order of magnitude less than GRB 050509b.
422:
423: %Unfortunately,
424: %GRB 050509b was one of the dimmest bursts ever recorded. Its $E_{iso}$ of
425: %8.6$\times10^{48}$ erg s$^{-1}$ is several orders of magnitude below the typical burst. The same
426: %burst, with an $E_{iso}$ $\sim$2000 times greater (like that of GRB 040924, for example),
427: %would have been detectable by Milagro.
428:
429: It has been suggested that a fraction of short duration GRBs could be due to soft gamma-ray
430: repeaters (SGRs) in nearby galaxies. There is some debate as to the exact fraction such objects
431: could represent, with estimates ranging from more than 1\%~\citep{ofek07} of short GRBs, to less
432: than 40\%~\citep{lazzati}. We have presented upper limits at three different redshifts, including
433: the case of z=0 which would be appropriate for bursts happening nearby. Indeed, the bright
434: GRB 051103 detected by the IPN has been found to be consistent with an SGR flare originating in
435: the nearby M81 galaxy group~\citep{ofek06}. Assuming this to be the location of the burst, we
436: obtain a Milagro TeV upper limit (1.9$\times10^{-5}$erg cm$^{-2}$) which is lower than the
437: IPN measured fluence of 2.3$\times10^{-5}$erg cm$^{-2}$.
438:
439: In conclusion, we have searched the Milagro data for prompt and delayed GeV--TeV emission from a
440: collection of seventeen short duration ($<$ 5 s) GRBs which occurred in Milagro's field of view
441: in the seven years since Milagro began operations in 2000. This represents the most comprehensive
442: search for very high energy emission from short GRBs ever performed. Due to the short duration and
443: low rate of short bursts, such observations must carried out by an experiment like Milagro with its
444: large field of view of $\sim$2 sr and high duty cycle. While no emission was detected from any of these
445: short bursts, HAWC~\citep{HAWC},a next-generation version of Milagro, would have more than 15 times
446: the sensitivity. The GLAST Gamma-ray Burst Monitor with its BATSE like field of view of over
447: 2$\pi$ sr will detect many bright, short GRBs and simultaneous observations of the GLAST Large
448: Area Telescope and HAWC will provide prompt spectra from keV-TeV energies to further our
449: understanding of short GRBs.
450:
451:
452: \acknowledgements
453: We are grateful to Kevin Hurley for providing us with the details of the IPN bursts and
454: for useful discussions regarding the use of such data. We thank Cristiano Guidorzi for discussions
455: regarding BeppoSAX data and Soeb Razzaque for help in comparing our data to his model. We are
456: also grateful to James Bullock for sending us optical depth data from the \cite{primack05}
457: EBL model. We have used GCN Notices to select raw data for archiving and use in this search, and
458: we are grateful for the hard work of the GCN team, especially Scott Barthelmy. We acknowledge
459: Scott Delay and Michael Schneider for their dedicated efforts in the construction and
460: maintenance of the Milagro experiment. This work has been supported by the National Science
461: Foundation (under grants
462: %-0096256, %UW-Madison
463: %-0097315, %LANL via UMD
464: PHY-0245234, %UCI; previous is PHY-0070933
465: -0302000, %UMD
466: -0400424, %Milagro Operations; previous is PHY-0075326
467: -0504201, %NYU current; previous are PHY-9901496 & PHY-0206656
468: -0601080, %UCSC; previous are PHY-0070927 & PHY-0245143
469: and
470: ATM-0002744) %UNH
471: the US Department of Energy (Office of High-Energy Physics and
472: Office of Nuclear Physics), Los Alamos National Laboratory, the University of
473: California, and the Institute of Geophysics and Planetary Physics.
474:
475:
476:
477: \def \atel {The Astronomer's Telegram}
478: \def \apj {ApJ}
479: \def \aj {AJ}
480: \def \apjl {ApJL}
481: \def \mnras {MNRAS}
482: \def \iaucirc {IAUCIRC}
483: \def \em { }
484: \def \aap {A\&A}
485: \def \nat {Nature}
486: \def \araa {Anual Review of Astronomy and Astrophysics}
487:
488: %\begin{thebibliography}{99}
489:
490: \begin{thebibliography}{10}
491: \bibliographystyle{plain}
492:
493: \bibitem[Aharonian et al.(2006)]{2006Natur.440.1018A} Aharonian, F., et
494: al.\ 2006, \nat, 440, 1018
495:
496: \bibitem[Albert et al.(2006a)]{2006astro.ph.12548A} Albert, J., et al.\
497: 2006, ArXiv Astrophysics e-prints, arXiv:astro-ph/0612548
498:
499: \bibitem[Albert et al.(2006b)]{2006ApJ...641L...9A} Albert, J., et al.\
500: 2006, \apjl, 641, L9
501:
502: \bibitem[Atkins et al.(2000a)]{atkins00a} Atkins, R. et al. 2000a, \apjl, 533, L119
503:
504: \bibitem[Atkins et al.(2000b)]{atkins00b} Atkins, R. et al. 2000b, Nucl.\ Instrum.\ Meth.\ A, 449, 478
505:
506: \bibitem[Atkins et al.(2003a)]{atkins03} Atkins, R. et al. 2003a, \apj, 583, 824
507:
508: \bibitem[Atkins et al.(2003b)]{atkins03b} Atkins, R. et al. 2003b, \apj, 595, 803
509:
510: \bibitem[Atkins et al.(2005)]{2005ApJ...630..996A} Atkins, R., et al.\
511: 2005, \apj, 630, 996
512:
513: \bibitem[Barthelmy et al.(2005)]{2005Natur.438..994B} Barthelmy, S.~D., et
514: al.\ 2005, \nat, 438, 994
515:
516: \bibitem[Berger et al.(2005)]{2005Natur.438..988B} Berger, E., et al.\
517: 2005, \nat, 438, 988
518:
519: \bibitem[Bloom et al.(2006)]{2006ApJ...638..354B} Bloom, J.~S., et al.\
520: 2006, \apj, 638, 354
521:
522: \bibitem[Burrows et al.(2005)]{2005Sci...309.1833B} Burrows, D.~N., et al.\
523: 2005, Science, 309, 1833
524:
525: \bibitem[Connaughton et al.(1997)]{connaughton97} Connaughton, V., et al. 1997, \apj, 479, 859
526:
527: \bibitem[Costa et al.(1997)]{1997Natur.387..783C} Costa, E., et al.\ 1997, \nat, 387, 783
528:
529: \bibitem[Dingus(2001)]{dingus01} Dingus, B. L., 2001, in \emph{High Energy Gamma Ray Astronomy}, ed. F. A. Aharonian and H. J. Volk, 2001, AIP, 558, 383
530:
531: \bibitem[Dingus(2006)]{HAWC} Dingus, B. L., 2006, HAWC (High Altitude Water Cherenkov) Observatory
532: for Surveying the TeV Sky, \emph{Proceedings from the First GLAST Symposium}, 5-8 February 2007, Stanford
533: University, Stanford, CA.
534:
535: \bibitem[Donaghy et al.(2006)]{2006astro.ph..5570D} Donaghy, T.~Q., et al.\
536: 2006, ArXiv Astrophysics e-prints, arXiv:astro-ph/0605570
537:
538: \bibitem[Falcone et al.(2006)]{2006ApJ...641.1010F} Falcone, A.~D., et al.\
539: 2006, \apj, 641, 1010
540:
541: \bibitem[Feldman \& Cousins(1998)]{feldman-cousins} Feldman, G.~J.,\& Cousins, R.~D.\ 1998,Phys.Rev.D 57,3873
542:
543: \bibitem[Fenimore et al.(2004)]{Fenimore} Fenimore, E.~E., et al.\ 2004, GCN 2735
544:
545: \bibitem[Fox et al.(2005)]{2005Natur.437..845F} Fox, D.~B., et al.\ 2005,
546: \nat, 437, 845
547:
548: \bibitem[Guidorzi (2001)]{guidorzi} Guidorzi, C.\ 2001, PhD Thesis, Universit\`{a} degli Studi di Ferrara
549:
550:
551: \bibitem[Gehrels et al.(2005)]{2005Natur.437..851G} Gehrels, N., et al.\
552: 2005, \nat, 437, 851
553:
554: \bibitem[Gehrels \& The Swift Team(2006)]{2006AIPC..838...14G} Gehrels, N.,
555: \& The Swift Team 2006, AIP Conference Series, 838, 14
556:
557: \bibitem[Gonzalez et al.(2003)]{gonzalez03} Gonzalez, M.~M.\ et al. 2003, Nature, 424, 749
558:
559: \bibitem[Hjorth et al.(2003)]{2003Natur.423..847H} Hjorth, J., et al.\
560: 2003, \nat, 423, 847
561:
562: \bibitem[Hopman et al.(2006)]{2006ApJ...643L..91H} Hopman, C., Guetta, D.,
563: Waxman, E., \& Portegies Zwart, S.\ 2006, \apjl, 643, L91
564:
565: \bibitem[Horan et al.(2007)]{2007ApJ...655..396H} Horan, D., et al.\ 2007,
566: \apj, 655, 396
567:
568: \bibitem[Huang et al.(2005)]{2005ApJ...628L..93H} Huang, K.~Y., et al.\
569: 2005, \apjl, 628, L93
570:
571: \bibitem[Hurley(2006)]{2006AIPC..838...25H} Hurley, K.\ 2006, American Institute of Physics Conference Series, 838, 25
572:
573: \bibitem[Hurley(2007)]{hurley} Hurley, K.\ 2007, private communication
574:
575: \bibitem[Kouveliotou et al.(1993)]{1993ApJ...413L.101K} Kouveliotou, C.,
576: Meegan, C.~A., Fishman, G.~J., Bhat, N.~P., Briggs, M.~S., Koshut, T.~M.,
577: Paciesas, W.~S., \& Pendleton, G.~N.\ 1993, \apjl, 413, L101
578:
579: \bibitem[Lazzati et al.(2005)]{lazzati} Lazzati, D., Ghirlanda,
580: G., \& Ghisellini, G.\ 2005, \mnras, 362, L8
581:
582: \bibitem[Li \& Ma(1983)]{lima} Li, T.\ P., \& Ma, Y.\ Q.\ 1983, \apj, 272, 317
583:
584: \bibitem[Mazets \& Golenetskii(1981)]{1981Ap&SS..75...47M} Mazets, E.~P.,
585: \& Golenetskii, S.~V.\ 1981, \apss, 75, 47
586:
587: \bibitem[Meszaros \& Rees(1993)]{1993ApJ...405..278M} Meszaros, P., \&
588: Rees, M.~J.\ 1993, \apj, 405, 278
589:
590: \bibitem[Nikishov (1961)]{nikishov61} Nikishov, A. I. 1961, Zh.\ Eksp.\ i Teor.\ Fiz., 41, 549
591: (English translation: Sov. Phys. JETP 14, 392 [1962]) %\[English trans. Sov. Phys. JETP, 14, 393 (1962)\]
592:
593: \bibitem[Norris et al.(1984)]{1984Natur.308..434N} Norris, J.~P., Cline,
594: T.~L., Desai, U.~D., \& Teegarden, B.~J.\ 1984, \nat, 308, 434
595:
596: \bibitem[Noyes, D. (2005)]{noyes} Noyes, D., 2005, PhD Thesis, University of Maryland, College Park
597:
598: \bibitem[Ofek et al.(2006)]{ofek06} Ofek, E.~O., et al.\ 2006, \apj, 652, 507
599:
600: \bibitem[Ofek(2007)]{ofek07} Ofek, E.~O.\ 2007, \apj, 659, 339
601:
602: \bibitem[Paciesas et al.(1999)]{1999ApJS..122..465P} Paciesas, W.~S., et
603: al.\ 1999, \apjs, 122, 465
604:
605: \bibitem[Padilla et al.(1998)]{padilla98} Padilla, L., et al. 1998, \aap, 337, 43
606:
607: \bibitem[Primack et al.(2005)]{primack05} Primack, J. R., Bullock, J. S., \& Somerville, R. S. 2005, AIP Conf. Proc., 745, 23.
608:
609: \bibitem[Prochaska et al.(2006)]{2006ApJ...642..989P} Prochaska, J.~X., et
610: al.\ 2006, \apj, 642, 989
611:
612: \bibitem[Razzaque(2006)]{razzaque} Razzaque, S.\ 2006, private communication
613:
614: \bibitem[Razzaque \& M{\'e}sz{\'a}ros(2006)]{2006ApJ...650..998R} Razzaque,
615: S., \& M{\'e}sz{\'a}ros, P.\ 2006, \apj, 650, 998
616:
617: \bibitem[Rees \& Meszaros(1992)]{1992MNRAS.258P..41R} Rees, M.~J., \&
618: Meszaros, P.\ 1992, \mnras, 258, 41P
619:
620: \bibitem[Rossi \& Greisen(1941)]{1941RvMP...13..240R} Rossi, B., \&
621: Greisen, K.\ 1941, Reviews of Modern Physics, 13, 240
622:
623: \bibitem[Saz Parkinson(2006)]{2006astro.ph.11457S} Saz Parkinson, P.~M.\
624: 2006, ArXiv Astrophysics e-prints, arXiv:astro-ph/0611457
625:
626: \bibitem[Stanek et al.(2003)]{2003ApJ...591L..17S} Stanek, K.~Z., et al.\
627: 2003, \apjl, 591, L17
628:
629: \bibitem[Villasenor et al.(2005)]{2005Natur.437..855V} Villasenor, J.~S.,
630: et al.\ 2005, \nat, 437, 855
631:
632: \bibitem[Wang et al.(2006)]{2006ApJ...641L..89W} Wang, X.-Y., Li, Z., \&
633: M{\'e}sz{\'a}ros, P.\ 2006, \apjl, 641, L89
634:
635: \bibitem[Wiersema et al.(2004)]{wiersema} Wiersema, K., et al.\ 2004, GCN 2800
636:
637: \bibitem[Woosley(1993)]{1993ApJ...405..273W} Woosley, S.~E.\ 1993, \apj,
638: 405, 273
639:
640: \end{thebibliography}
641:
642:
643: \clearpage
644: \thispagestyle{empty}
645: \input{tab1}
646:
647: \clearpage
648:
649: \begin{figure}[htbp]
650: \centering
651: \rotatebox{0}{\resizebox{!}{12cm}{
652: \includegraphics{f1.eps}}}
653: \caption{Effective area of Milagro for gamma rays as a function of energy for three
654: different zenith angles. The straight line is a power law $E^{2.6}$ (with arbitrary
655: normalization). The different curves (in decreasing order of thickness) reflect the
656: effective area for zenith angles of 10$^\circ$, 30$^\circ$, and 45$^\circ$ (roughly corresponding
657: to GRBs 050509b, 050505, and 040924). The figure illustrates the decrease in effective area with
658: zenith angle. The limited number of simulated showers at the highest energies results in fluctuations
659: in the curves above 10$^4$ GeV.
660: \label{fig1}}
661: \end{figure}
662:
663: \begin{figure}[htbp]
664: \centering
665: \rotatebox{0}{\resizebox{!}{12cm}{
666: \includegraphics{f2.eps}}}
667: \caption{Relative increase in effective area between the simple (55 tube) multiplicity trigger
668: and the VME programmable trigger, as applied to GRB 050509b. The figure shows an increase
669: in effective area using the new trigger of more than 50\% at 1 TeV and around 150\% at 100 GeV,
670: relative to the old trigger.\label{fig2}}
671: \end{figure}
672:
673: \begin{figure}[htbp]
674: \centering
675: \rotatebox{0}{\resizebox{!}{12cm}{
676: \includegraphics{f3.eps}}}
677: \caption{Number of background events per second detected in a circular bin of radius 1.6 degrees, as a
678: function of zenith angle. The background rate depends on the analysis cuts used as well as the detector
679: configuration and atmospheric conditions on a particular day. The figure was made with data taken
680: within one hour of GRB 050509b.\label{fig3}}
681: \end{figure}
682:
683:
684: \end{document}
685: