RHESSI Spectral Fits of Swift GRBs

0.1 MB·

by  Eric C. Bellm

RHESSI Spectral Fits of Swift GRBs
Eric C. Bellm∗,†, Mark E. Bandstra∗,†, Steven E. Boggs∗,†, Wojtek
Hajdas∗∗, Kevin Hurley∗, David M. Smith‡ and Claudia Wigger∗∗
∗UCBerkeleySpaceSciencesLaboratory,7GaussWay,Berkeley,CA94720-7450,USA
8
†DepartmentofPhysics,UCBerkeley
0
0 ∗∗PaulScherrerInstitute,VilligenPSI,Switzerland
2 ‡SantaCruzInstituteforParticlePhysics,SantaCruz,CA,USA

n
a Abstract. One of the challenges of the Swift era has been accurately determining E for the
peak
J
promptGRB emission. RHESSI, which is sensitive from 30 keV to 17 MeV, can extend spectral

6 coverage above the Swift-BAT bandpass. Using the public Swift data, we present results of joint
1 spectralfitsfor26burstsco-observedbyRHESSIandSwift-BATthroughMay2007.Wecompare

] these fits to estimates of Epeak which rely on BAT data alone. A Bayesian Epeak estimator gives
h bettercorrespondencewithourmeasuredresultsthananestimatorrelyingoncorrelationswiththe
p Swiftpowerlawindices.
-
o Keywords: gamma-rays:bursts
r PACS: 98.70.Rz;95.85.Pw
t
s
a
[
GRB PROMPT EMISSION SPECTROSCOPY WITH RHESSI

1
v
The Reuven Ramaty High-Energy Solar Spectroscopic Imager (RHESSI) [1] is a dedi-
7
1 cated solarobservatory.RHESSI’s ninegermaniumdetectors are sensitivefrom 30keV
4 to17MeV,withexcellentresolutioninenergy(1-5keV)andtime(1binarym s)[2].Each
2
of the nine coaxial detectors is electronically segmented into front and rear segments.
.
1
Because the detectors are unshielded, RHESSI observes emission from astrophysical
0
8 sourceslikeGRBs witha∼2p field ofview.
0 We perform Monte Carlo simulations using the MGEANT package [3] to determine
:
v RHESSI’s response to off-axis sources like GRBs. RHESSI’s response varies with off
i
X axis angle, so we create responses every 15 degrees. For each response, we simulate
monoenergeticphotonsin 192 logarithmicenergy bins ranging from 30 keV – 30 MeV.
r
a
Since RHESSI’s per-detector response also varies during the spacecraft’s four second
spin period, we bin theannular response in six azimuthal bins and weight these bins by
thetotalburstlightcurvetocreate thefinal response.
The RHESSI data are extracted in SSW-IDL. We fit and subtract a time-varying
background,allowingforpossibleperiodicmodulationwiththespinperiod.Weperform
spectral fitting in ISIS [4], a forward-fitting package analogous to XSPEC1, which is
extensively programmable and allows computation of rigorous fluence error estimates
viaexplorationoftheparameterspace.
Energetic charged particles over time have caused radiation damage to RHESSI’s
1 http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/
germaniumdetectors.Thisradiationdamagecausesbroadenedspectrallinesduetohole
trapping and a general loss of active volume. In this work, we restrict our analysis to
those detectors which do not exhibit signs of radiation damage. At the time of Swift’s
launch, six RHESSI segments were usable for spectroscopy; by May 2007 only two
segments remained undamaged. In November 2007, RHESSI underwent an annealing
procedure to reverse the effects of the radiation damage. The anneal restored some lost
sensitivity,but analysis of future bursts will require more sophisticated modelingof the
remainingeffects ofradation damageon RHESSI’s spectral response.
RHESSI-BAT JOINT FITS
WeattemptedsimultaneousspectralfittingforallRHESSI-observedGRBsappearingin
the first BAT Catalog [5]. Of 46 candidate bursts, 26 had sufficient RHESSI counts for
spectral analysis and produced acceptable joint fits. We selected analysis time intervals
manually from the RHESSI lightcurve using a S/N criterion. The resulting intervals
wereusuallyshorterthanthoseusedintheBATCatalog.WegeneratedBATspectraand
responsesforourintervalswiththestandardanalysisprocedures.2
Typically, the joint fits did not require a normalization offset between the RHESSI
andBATdata.(Ofthefourburststhatneededanoffsetforacceptablefitting,onewasin
aperiodofhighlymodulatedRHESSIbackground,andthreewereafterDecember2006
when radiation damage was becoming severe.) For the RHESSI data, we generally fit
overthefull 30 keV-17 MeV energy band. Forbursts comingfrom therear ofRHESSI,
we raised the lower energy bound to ∼60 keV, as the additional passive material of the
RHESSI cryostat can influence the low energy data. For GRB 061007, we omitted the
RHESSI data above 3 MeV; there were no significant counts above that level, but the
residualsshowedsystematicdeviationwhich biasedthefit.
OurresultsshowgoodcorrespondencewithcomparablefitsreportedbyKonus-Wind
and Suzaku-WAM, which also are sensitive in the MeV range [e.g., 6, 7, 8, 9, 10]. For
16 of the 26 bursts, the joint RHESSI-BAT best fit found additional model parameters
(E and/orb )compared totheBAT-onlyfit.
peak
Wereport theresultsofthejointfits inTable1.
TESTING E ESTIMATORS
peak
The peak energy E of the prompt GRB spectrum is crucial to determining overall
peak
burst energetics, and it plays a key role in several proposed luminosity indicators.
The narrow passband of Swift-BAT prevents determination of E for many bursts.
peak
Accordingly, a number of attempts have been made to infer E from the BAT data
peak
alone. Butler et al. [11] used a Bayesian fit method with priors determined from the
BATSE catalog to estimate E . Zhang et al. [12] derived an E – BAT power law
peak peak
index correlation using hardness ratios (see also [13]). In Figure 1, we compare the
2 http://swift.gsfc.nasa.gov/docs/swift/analysis/threads/bat_threads.html
FIGURE 1. Comparison of the measured RHESSI+BAT E values with those predicted by [11]
peak
(left)andthosefoundwiththecorrelationof[12](right).Pointsmarkedwithacrosshavefitb >−2,and
henceE isonlyaformalvalue.PointsintherightplotmarkedwithadiamondhaveBAT-onlypower
peak
lawindicesoutsidetherange(−2.3<a <−1.2)likelytoyieldaccuratepredictionsofE [13].The
peak
overplotlinesrepresentequalityofthemeasuredandpredictedvalues.
predictionsofthesemodelsto ourjointfit results.
TheBayesmodelshowsgoodcorrespondencewiththemeasuredvalues.Above∼600
keV,thepredictedvaluesofE tendtobelow,althoughtheirerrorbarsreachnearthe
peak
measured values. This deviation is exaggerated somewhat, as our data are for shorter,
moreintense(and typicallyharder)burst intervalsthan usedin[11].
ComparisonoftheZhangetal.correlation-predictedE toourmeasuredRHESSI-
peak
Swift values showsthat this correlation appears to systematicallyunderpredict themea-
suredvalueofE , especiallyat highenergy.
peak
ACKNOWLEDGMENTS
Thisworkwas supportedbytheSwift AO-3GIgrant NNX08AE86G.
REFERENCES
1. R.P.Linetal.,SolarPhys.210,3–32(2002).
2. D.M.Smithetal.,SolarPhys.210,33–60(2002).
3. S.J.Sturner,H.Seifert,C.Shrader,andB.J.Teegarden,“MGEANT-AGEANT-BasedMulti-Purpose
SimulationPackageforGamma-RayAstronomyMissions,”inAmericanInstituteofPhysicsConfer-
ence Series, edited by M. L. McConnell, and J. M. Ryan, 2000, vol. 510 of American Institute of
PhysicsConferenceSeries,pp.814.
4. J. C. Houck,“ISIS:The InteractiveSpectralInterpretationSystem,” in High ResolutionX-raySpec-
troscopywithXMM-NewtonandChandra,editedbyG.Branduardi-Raymont,2002.
5. T.Sakamotoetal.,ArXive-prints707(2007),0707.4626.
11 1 11 98 76 TABLE1. Bestjointfitparameters.SpectralmodelsconsideredareaBandfunction[14],cutoffpowerlaw
43 2 10 .. ..
.. . .. (CPL),andsimplepowerlaw(PL).Errorsarequotedatthe90%confidencelevel.Fluenceisinthe15keV–10
L a NSKNSH
D.Bandetal.,ApJ413,281–292(1993).B.Zhangetal.,ApJ655,989–1001(2007),.arXiv:astro-ph/061017728(2007),.arXiv:astro-ph/0612238B.Zhang,B.-B.Zhang,E.-W.Liang,N.Gehrels,D.N.Burrows,andP.Mészáros,A.rXiv:0706.1275N.R.Butler,D.Kocevski,J.S.Bloom,andJ.L.Curtis,ApJ671,65K.L.Pageetal.,ApJ663,1125–1138(2007),.arXiv:0704.1609etwork5837,1(2006)..Golenetskii,R.Aptekar,E.Mazets,V.Pal’Shin,D.Frederiks,andT.Cline,GR.Yamaokaetal.,GRBCoordinatesNetwork5724,1(2006).etwork5722,1(2006)..Golenetskii,R.Aptekar,E.Mazets,V.Pal’Shin,D.Frederiks,andT.Cline,GR.A.Krimmetal.,ApJ648,1117–1124(2006),.arXiv:astro-ph/0605507 M00000000000000000000000e66666666665565555555544VG1110000000110000000001111065954412118787532122Rb22010001212112101221222aB16742818101170723569443nAAABBABd,anAdna---an512-----------1----060l40250003113a60130249y5.............l...0......81s48.928623617y02(5332531iss–s–––––––––––––––e––––––i–scI1n11)261311322238t2112i313322t2300197527465410me....................3.5778..89621829487493r0057evsaalrerBeMlBBBBBBBBBBBeCCCCCCCCaPPPPsoaaaaaaaaaaatPPPPPPPPtiLLLLnnnnnnnnnnndvLLLLLLLLFdddddddddddeeilttoth-----------------------e01011011011111011100110.......................B8563683086452055016701956870816750264781357207Aa+−++−++−−+−+++−−−++−+−−+−+−++−−+−+−+−+−+−+−+−T000000000000000000000000000000000000000000000.............................................t000001312220030020101111011000000111111746300r446442514305736765759340976464744103120250254igge2r7721612222t41116(4987243573Eikm87764811602418577p78e6781787..........eeV71++−++−−++−−++−−+−+−.......a.+−+−+−+−+−+−+−..k2112196612411881743)+−+−4221155964374707042752114228787886633040949921..9294.16..05658.1766036............................... -----------22111131222...........3498991761177878823220b++−−+−+−++−−+−++−−+−+−0010010000000000000000......................32242626112425121825122412670317165029940064 (10−F6l2ue46re63122221462318g41n244631412962053513727s..c67...................../e070768896784080733198c+−+−m++−+−++−−+−+++−−−++−+−−+−+−++−−+−+−+−+−+−22118580074546121711404004011334746115411338732962.............................................134430592721479248383638662827921173282112510) 01001100101000110000001c.......................80980098093899009979892n242274452625960232029131
pJL 6–67 BC BC 061222A 81.9– 87.9 CPL -0.80+−−000...101757 269.+−−317147.66.. 7.5−+−461...468 0.91
655,L 7(20 oordin oordin 007700252008 02..12–– 3128..12 CCPPLL --11..2047++−−0000....00008776 421893.+−.+−1164816796.... 4220..33+−+−6443....4996 00..6965
2 0 a a
5– 7), tes tes