ebook img

Constraining DM properties with SPI PDF

1.1 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Constraining DM properties with SPI

Mon.Not.R.Astron.Soc.000,1–15(2007) Printed5January2009 (MNLATEXstylefilev2.2) CERN-PH-TH/2007-202 Constraining DM properties with SPI Alexey Boyarsky1⋆, Denys Malyshev2,3, Andrey Neronov4⋆, Oleg Ruchayskiy5 9 0 1CERN,PH/TH,CH-1211Geneve23,Switzerland 0 2BogolyubovInstituteforTheoreticalPhysics,Kiev,03780,Ukraine 2 3DublinInstituteforAdvancedStudies,31FitzwilliamPlace,Dublin2,Ireland 4INTEGRALScienceDataCenter,Chemind’E´cogia16,1290Versoix,Switzerland n GenevaObservatory,51ch.desMaillettes,CH-1290Sauverny,Switzerland a 5E´colePolytechniqueFe´de´raledeLausanne,InstituteofTheoreticalPhysics, J FSB/ITP/LPPC,BSP720,CH-1015,Lausanne,Switzerland 3 ] h p - o ABSTRACT r Using the high-resolution spectrometer SPI on board the International Gamma-Ray Astro- t physics Laboratory (INTEGRAL), we search for a spectral line produced by a dark matter s a (DM)particlewithamassintherange40keV <MDM <14MeV,decayingintheDMhalo [ oftheMilkyWay.TodistinguishtheDMdecaylinefromnumerousinstrumentallinesfound intheSPIbackgroundspectrum,westudythedependenceoftheintensityofthelinesignal 2 ontheoffsetoftheSPIpointingfromthedirectiontowardtheGalacticCentre.Afteracritical v 2 analysis of the uncertaintiesof the DM density profile in the inner Galaxy,we find that the 2 intensityoftheDMdecaylineshoulddecreasebyatleastafactorof3whentheoffsetfrom 9 the Galactic Centre increases from0◦ to 180◦. We find that such a pronouncedvariationof 4 the line flux acrossthe sky is notobservedfor any line, detected with a significance higher . than3σintheSPIbackgroundspectrum.PossibleDMdecayoriginisnotruledoutonlyfor 0 1 theunidentifiedspectrallines,havinglow(∼ 3σ)significanceorcoincidinginpositionwith 7 theinstrumentalones.Intheenergyintervalfrom20keVto7MeV,wederiverestrictionson 0 theDMdecaylineflux,impliedbythe(non-)detectionoftheDMdecayline.Foraparticular v: DMcandidate,thesterileneutrinoofmassMDM,wederiveaboundonthemixingangle. Xi Keywords: methods:dataanalysis–techniques:spectroscopic–Galaxy:halo–darkmatter; r a 1 INTRODUCTION pact halo objects (MACHOs) – constitute dominant fraction of mass in the halo (Alcocketal. 2000; Lasserreetal. 2000; Alard DarkmatterintheUniverse 1999). The only non-baryonic DM candidate in the SM can- There is a vast body of evidence, suggesting that the large frac- didates – (left-handed) neutrino – is ruled out from the large tionofmatterintheUniverseexistsintheformoftheDarkmat- scale structure (LSS) considerations (see e.g. Bondetal. 1980; ter(DM).However,whilethetotaldensityoftheDMismeasured Hannestad&Raffelt2004;Crottyetal.2004). withaveryhighprecision(ΩDMh2 = 0.105+−00..000079,Spergeletal. What are the properties of a successful DM candidate? 2007),littleisknownaboutitspropertiesapartfromthis.Thepos- First of all, this particle should be massive. Many extensions sibilitythattheDMiscomposedoftheStandardModel(SM)par- of the SM present the DM candidates with the masses rang- ticleshasbeenruledout foralongtimealready. Indeed, theDM ing from 10−10 eV (massive gravitons, Dubovskyetal. 2005) cannot be made out of baryons, as producing such an amount of and 1∼0−6 eV (axions) to hundreds of GeV (WIMPs) and ∼ baryonic matter would require drastic modifications of the sce- evento1013GeV(WIMPZILLA,Kuzmin&Tkachev1998,1999; nario of the Big Bang nucleosynthesis (BBN), which otherwise Chungetal.1999).ForareviewofparticlephysicsDMcandidates successfully describes the abundance of light elements (see for seee.g.Bergstrom(2000);Bertoneetal.(2005);Carretal.(2006). example Dar 1995). Recent microlensing experiments rule out Secondly, there should exist mechanisms of DM production thepossibility that another type of baryonic DM– massive com- withthecorrectabundances.Theproductionmechanisminpartic- ular determines the velocity distribution of particles in the early Universe.Thisvelocitydistributioncan,inprinciple,beprobedex- ⋆ OnleaveofabsencefromBogolyubovInstituteforTheoreticalPhysics, perimentally.Namely, ifduringthestructureformationepoch the Kiev,Ukraine DM particles have velocities, comparable to the speed of sound 2 A. Boyarskyet al. in the baryon-photon plasma, they “erase” density fluctuations at Bertoneetal. (2007); Zhangetal. (2007) (see also the book scales, smaller than the distance, they have traveled (called the byKhlopov1997). free-streaming length). To differentiate various models in accor- dancewiththisproperty,theDMcandidateswiththenegligibleve- locitydispersion(and,correspondingly, free-streaming)arecalled cold DM (CDM), while those with the free-streaming of the or- der of 1Mpc are considered to be warm (WDM).1 It is pos- SterileneutrinoDM ∼ sible to constrain the free-streaming length of a particular DM It was noticed long ago that the right-handed (or as it is often candidate by probing the structure of the Universe at galaxy-size calledsterile)neutrinowiththemassinthekeVrangewouldrep- scales.ThiscanbedonethroughtheanalysisoftheLyman-αfor- resent a viable DM candidate (Dodelson&Widrow 1994). Such estdata(Huietal.1997).Lyman-αanalysisputsanupper bound a neutrino would interact withthe rest of the matter only via the on the free-streaming of the DM particles (Hansenetal. 2002; quadraticmixingwithleft-handed(active)neutrinosandtherefore Vieletal. 2005; Seljaketal. 2006; Vieletal. 2006; Vieletal. (althoughnotstable)couldhavecosmologicallylonglife-time.At 2007).Itshouldbenotedhoweverthatcurrentlyexistinginterpreta- thesametime,itcouldbeproducedintheearlyUniversewiththe tionoftheLyman-αdataismodel-dependent,as,apartfromanum- correctabundances(Dodelson&Widrow1994;Shi&Fuller1999; berofastrophysicalassumptions(seeHuietal.1997)andcompli- Shaposhnikov&Tkachev2006).Oneofthedecaychannelsofthe catedhydrodynamicsimulations,itreliesonaprioriassumptions unstablesterileneutrinos includesemission ofphotons of theen- aboutthevelocitydistributionoftheDMparticles. ergyequaltohalfofthesterileneutrinorestenergy.Thispotentially A way to differentiate between CDM and WDM models providesapossibilitytoobservethedecaysofDMsterileneutrinos would be to compare the numerical simulations of the DM dis- viadetectionofacharacteristicspectrallineinthespectraofastro- tribution in the Milky Way-type galaxies with the actual obser- physicalobjectswithlargeDMconcentration. vations.However, theresolutionoftheN-bodysimulationsisnot RecentlythisDMcandidatehasattractedmuchattention(see yet sufficient to answer the questions about e.g. the DM density e.g.Shaposhnikov(2007)andreferencestherein).Itwasfoundthat profiles in dwarf satellite galaxies. Moreover, most of the simu- averymodestandnaturalextensionoftheSMby3right-handed lationsincludeonlycollisionlessDMparticles,anddonot model neutrinos(makingtheSMmoresymmetricasallSMfermions,in- thebaryonsandtheirfeedbackonthegalaxystructureformation. cludingneutrino,wouldhavenowtheirleftandrighthandedcoun- TheseproblemsarenotsolvedevenfortheCDMsimulations,and terparts)providedaviableextensionofthetheory,capableofsolv- WDMsimulationshaveadditionalseriousdifficulties.Fromanob- ingseveral“beyondtheSM”problems.Firstofall,suchanexten- servationalpointofview,ithasbeenarguedforsometimealready sionmakes neutrinosmassiveandthusperhaps providesthesim- thatthereisadiscrepancybetweenCDMsimulationsandobserva- plestandthemostnaturalexplanationofthephenomenonof“neu- tions(seee.g.Moore1994;Mooreetal.1999;Klypinetal.1999; trino oscillations” (see e.g. Foglietal. (2006); Strumia&Vissani Bodeetal.2001;Avila-Reeseetal.2001;Goerdtetal.2006)Ithas (2006); Giunti (2007) for reviews). The smallness of neutrino beenclaimedrecentlythatanumberofrecentobservationsofdwarf masses in this model (called νMSM in Asaka&Shaposhnikov satellite galaxies of the Milky way and Andromeda galaxy seem 2005)isachievedbytheusualsee-sawmechanismwithMajorana to indicate the existence of the smallest scale at which the DM massesofright-handedneutrinosbeingbelowelectroweakscale.2 exists (Gilmoreetal. 2006, 2007; Gilmore 2007; Koposovetal. Secondly, if two heavier sterile neutrinos (N and N ) 2007). However, this statement and the interpretation of the ob- 2 3 are almost degenerate in mass and have their masses between servations are still subject to debate (Klimentowskietal. 2007; (100)MeVand (20)GeV,theνMSMprovidesthemechanism Penarrubiaetal. 2007; Strigarietal. 2007; Simon&Geha 2007). O O ofgeneratingthebaryonasymmetryoftheUniverse.Thirdly,the ThereforeitistooearlytosaywhatkindofDMmodelsisfavoured lighteststerileneutrinoN canhavearbitrarymassandarbitrarily bycomparingsimulationsandobservations. 1 weakcouplingwiththe(active)neutrinosector.Atthesametime, UsuallyitisalsonecessaryfortheDMcandidatetobestable. itcanbeproducedintheearlyUniverseinthecorrectamounts.It Forthemostpopular DMcandidate–weaklyinteractingmassive representsthereforetheDMparticleintheνMSM.Thus,altogether particles(WIMPs),thisisrelatedtothefactthattheparticlesof ∼ theνMSMrepresents(arguably)thesimplestextensionoftheSM, electroweakmass,havingweakstrengthinteractionwithSMmatter capableofexplainingthreeimportantquestions:originandsmall- (requiredtoproducethecorrectamountofDM),woulddecaytoo nessofneutrinomasses,baryonasymmetryintheUniverseandthe fastandwouldnotbe“dark”.If,however,theDMparticleinteracts existenceoftheDM. withtheSMmoreweaklythanWIMPs,itcouldwellhaveafinite (althoughcosmologicallylong)lifetime. There exist several unstable (decaying) DM candidates e.g. gravitino (Borganietal. 1996; Baltz&Murayama 2003; Roszkowskietal. 2005; Cerdenoetal. 2006; Cembranosetal. ExistingrestrictionsonsterileneutrinoDMparameters. 2006; Lolaetal. 2007). In this paper we will concentrate mainly on one candidate, the sterile neutrino (although our results Whatarethecurrentrestrictionsonparameters(massandmixing) will be applicable for any type of decaying DM). Constraints of sterile neutrino DM? First of all sterile neutrino mass should on the decaying DM were analyzed in deRujula&Glashow (1980); Berezhianietal. (1987); Doroshkevichetal. (1989); Berezhianietal. (1990); Berezhiani&Khlopov (1990); 2 The fact that the νMSM does not introduce any new scale above the 1 Theleft-handed neutrino wouldrepresent hotDMinthisterminology, electroweakone,makesthistheoryespeciallyappealingfromthepointof i.e.theDMwiththefree-streaminglength≫1Mpc. viewofitsexperimentalverification/falsification. ConstrainingDM propertieswithSPI 3 satisfytheuniversalTremaine-Gunnlowerbound:3M &300 DM − 500eV.4 Next,asthesterileneutrinopossessesthe(two-body)radiative decaychannel:N ν+γ,theemittedphotonwouldcarrytheen- 1 → ergyE =M /2.Alargefluxofsuchphotonsisexpectedfrom γ DM thelargeconcentrationsoftheDMsterileneutrinos,likegalaxies orgalaxyclusters. Recently an extensive search of the DM decay line in the region of masses M . 20 keV was conducted, using the DM data of Chandra (Riemer-Sørensenetal. 2006; Boyarskyetal. 2006d; Abazajianetal. 2007) and XMM-Newton (Boyarskyetal. 2006a,b,c; Watsonetal. 2006; Boyarskyetal. 2007). The re- gion of soft X-ray (down to energies 0.2 keV) was explored by Boyarskyetal. (2007) with the use of the wide field of view spectrometer (McCammonetal. 2002). The non-observation of the DM decay line in X-ray, combined with the first principles calculation of DM production in the early Universe (Asakaetal. 2007), implies that the Dodelson&Widrow (1994) (DW) sce- nario can work only if the sterile neutrino mass is below 4 keV (Boyarskyetal. 2008). If one takes into account recent lower bound on the mass of sterile neutrino DM in the DW sce- nario M > 5.6 keV (Vieletal. 2007), it seems that the pos- DM sibility that all the DM is produced via DW scenario is ruled Figure1.ComparisonofsensitivitytowardsthesearchofthenarrowDM decaylinefordifferent instrumentswiththewideFoV.Diagonalstraight out(Boyarskyetal.2008).Thepossibilitythatonlyfractionofthe linesshowtheimprovementofsensitivity(byafactor,markedontheline) DMisproduced viaDWmechanismremainsopen(Palazzoetal. ascomparedwiththeHEAO-IA4lowenergydetector(LED),takenasa 2007). reference. There are other viable mechanisms of DM production, in- cludinge.g.resonantoscillationproductioninthepresenceoflep- tonasymmetries(Shi&Fuller1999).SterileneutrinoDMcanbe When the preparation of this paper was at its final producedbythedecayoflightinflaton(Shaposhnikov&Tkachev stage, Yu¨kseletal. (2008, hereafter Y07) published their work, 2006) or in a similar model with the different choice of parame- whichused theresultsof Teegarden&Watanabe(2006,hereafter ters(Kusenko2006;Petraki&Kusenko2007).Thesemechanisms T06)toplacerestrictionsontheparametersofsterileneutrinoDM are currently not constrained and remain valid for DM particles intherange40 700keV.WediscussitinmoredetailsinSec- − withthemassesinthekeVrangeandabove. tion6. The search for the DM decay line signal produced by ster- ileneutrinos withmassesabove 20 keV iscomplicated bythe ∼ SPIspectrometer absence of the focusing optics telescopes (similar to Chandra or XMM-Newton) in the hard X-ray and γ-ray domain of the spec- Theabsenceofthefocusingopticssignificantlyreducesthesensi- trum. For example, the existing restrictions in the 20 100 keV tivityof thetelescopesoperating inthehard X-ray/softγ-rayen- − mass range (Boyarskyetal. 2006a,c) are derived from the obser- ergyband. Most of theinstrumentsoperating inthisenergy band vationsofdiffuseX-raybackground,withthehelpofnon-imaging usecollimatorsand/orcodedmaskstodistinguishsignalsfromthe instruments,HEAO-I(Gruberetal.1999).Thecurrentstatusofas- sourcesontheskyfromtheinstrumentalbackground.Contraryto trophysicalobservationsinsummarizedinRuchayskiy(2007). thefocusingopticstelescopes,boththesourceandbackgroundsig- In this paper we use the spectrometer SPI on board of IN- nalsarecollectedfromtheentiredetector,whichsignificantlyin- TEGRAL satellite to place restrictions on parameters of decay- creasestheirreduciblebackground. ing DM in the mass range 40 keV 14 MeV. This range Thefocusingopticsenablestosignificantlyreducetheback- − of masses is interesting, for example, the sterile neutrinos, pro- groundonlyinthestudiesofpointsources.Ifthesourceunderin- duced in the early Universe in the presence of large lepton vestigationoccupiesalargefractionofthesky(e.g.theentireMilky asymmetries (Shi&Fuller 1999) or through the inflaton de- Waygalaxy),theperformanceofthefocusingandnon-focusingin- cay(Shaposhnikov&Tkachev2006).Itisalsorelevantforthecase strumentswiththesamedetectorcollectionareaare,infact,com- ofgravitinoDM(Pagels&Primack1982;Bondetal.1982). parable. In the case of an extended source, emitting a narrow spec- tralline,anefficientwayofreductionofinstrumentalbackground 3 In its simplest form the Tremaine-Gunn bound comes from the is via the improvement of the spectral resolution of the instru- fact that for the fermions there is a maximal density in the phase ment(inthecaseofabroadcontinuumbackgroundspectrum,the space (Tremaine&Gunn 1979; Dalcanton&Hogan 2001) and therefore number of background counts attheenergyof thelineispropor- theobservedphase-spacedensityinvariousDMdominatedsystemsshould tionaltothespectralresolution∆E).Thebestpossiblesensitivity belessthatthis(massdependent)bound. isachievedwhenthespectralresolutionreachestheintrinsicwidth 4 AstrongerlowerboundfromLy-α(Seljaketal.2006;Vieletal.2006; ofthespectralline(seeFig.1forthecaseofwideFoVinstruments Vieletal. 2007) can be obtained in the case of the particular produc- tion mechanisms – the Dodelson-Widrow scenario (Dodelson&Widrow andBoyarskyetal.(2007)forthecaseofnarrowFoVinstruments). 1994).Forotherpossibleproductionmechanisms(e.g.Shi&Fuller1999; In the case of the line produced by the DM decaying in the Shaposhnikov&Tkachev2006)theLy-αconstraintsshouldbereanalyzed. MilkyWayhalo,thelinewidthisdeterminedbytheDopplerbroad- 4 A. Boyarskyet al. R H Figure3.TheeffectiveareaoftheSPIdetectorforanon-axissource,asa functionofthephotonenergy.Theplotisproducedbycollectivetheon-axis effectiveareasofthe17SPIdetectorsfromtheinstrumentalcharacteristics files. R det 2 THEEXPECTEDSIGNALFROMTHEDMDECAYIN THEHALOOFTHEMILKYWAY. Theexpected surface brightness of theDM decay lineinagiven Figure2.ThegeometryoftheSPIFoV. directionontheskyisafunctionoftheangulardistanceφbetween thegivendirectionontheskyandthedirectiontowardstheGalactic center(GC).ItcanbecalculatedbytakingtheintegraloftheDM eningbytherandommotionoftheDMparticles.Thevelocitydis- densityprofileρDM(r)alongthelineofsight(“columndensity”) persionoftheDMmotioninthehaloisabouttherotationvelocity ∞ oftheGalacticdisk,v 200km/s.ThismeansthatDopplerbroad- (φ)= dzρ r2 2zr cosφ+z2 , (4) eningoftheDMdecay∼lineisabout SDM Z DM„q ⊙− ⊙ « 0 ∆E v 10−3 . (1) wherer⊙ ≃ 8.5kpcisthedistancefromtheSolarsystemtothe E ∼ c ≃ GC.Angleφisrelatedtothegalacticcoordinates(b,l)via Thus,theoptimalspectralresolutionofaninstrumentsearchingfor cosφ=cosb cosl. (5) theDMdecaylineproducedbytheMilkyWayDMhaloshouldbe Thus, the galactic center corresponds to φ = 0◦, the anti-center ∆E 10−3E. φ = 180◦,andthedirectionperpendiculartothegalacticplaneto ≃ Such optimal spectral resolution isalmost achieved withthe φ=90◦.TheexpectedDMfluxisgiventhenby spectrometer SPIonboard ofINTEGRALsatellite,whichhasthe dF (φ) Γ E maximal spectral resolving power of E/∆E 500 and works DM = DM γ (φ), (6) ≃ dΩ 4πM SDM inthe energy range 20 keV –8 MeV (Vedrenneetal.2003).SPI DM isa“codedmask”typeinstrumentwithanarrayof19hexagonal whereΓDMistheDMdecayrate shaped Ge detectors (of which only 17 are operating at the mo- Ingeneral, thesurface brightnessFDM(φ)isvariable across ment). thetelescopeFoV.Thisisespeciallytrueforawidefieldofview TheSPItelescope consists of acoded maskinscribed intoa (FoV) instruments (like SPI). In order to calculate the detector circle of the radius R = 39 cm, placed at the height H = countrate,onehastointegrateflux(6)overtheFoVandoverthe mask 171cmabovethedetectorplaneandofthedetector,whichhasthe (effective)detectorareaandthendividebytheenergyofthepho- shape of a hexagon inscribed into a circle of the radius Rdet tons,Eγ =MDM/2: ≃ 15.3cm(seeFig.2).Theportionoftheskyvisiblefromeachpoint A (E α,β)dF hoafsthteheSrPefIodreeteacntgourl(atrhdeisaom-ceatellredfullycodedfieldofview,FCFOV) R=FZoZVdαdβ eff Eγγ| dΩDM`φ(α,β)´, (7) where (α,β) are the angular coordinates in the FoV, A is the eff R R ΘFCFOV =2arctan» maskH− det–≈16◦ , (2) teifofnec(tαiv,eβa)r.eaatenergyEγforthephotons,comingfromthedirec- TheeffectiveareaoftheSPIdetector(whichisdeterminedby whiletheportionoftheskyvisiblebyatleastsomeofthedetectors thetransparencyofthemaskandthequantumefficiencyofthede- (thepartiallycodedfieldofview,PCFOV)is tector)changeswiththephotonenergy.Foranon-axispointsource, Θ =2arctan Rmask+Rdet 35◦ . (3) dF/dΩ(α,β)=f0δ(α)δ(β), PCFOV » H –≈ theintegralofEq.(7)reducestof A ,whereA (E )is 0 eff,on eff,on γ The solid angle spanned by the cone with this opening angle is thedetectoreffectiveareaforanon-axissource.Itsdependenceon energyE isshownonFig.3.5 Ω =2π 1 cos(Θ /2) 0.29(seeFig.2).Widefield γ PCFOV “ − PCFOV ”≃ ofviewmakestheSPItelescopesuitableforthestudyofthevery extendedsources,liketheMilkyWayDMhalo. 5 The on-axis effective area is calculated by summing the energy- ConstrainingDM propertieswithSPI 5 2.1 ModelingtheDMhalooftheGalaxy TheDMhalooftheGalaxyhasbeenextensivelystudied(seee.g. 2 SPI effective area [cm ] at 100 keV Kravtsovetal.1998;Klypinetal.2002;Battagliaetal.2005).Var- 25 160 iousDMprofiles,usedtofitobservedvelocitydistributions,differ 20 themostintheGCregion. 140 It was shown in Klypinetal. (2002); Battagliaetal. (2005) 15 g]g] 120 that the DM halo of the MW can be described by the Navarro- ee 10 Frenk-White(NFW)profile(Navarroetal.1997) dd gle [gle [ 5 100 ρ (r)= ρsrs3 , (11) anan 0 80 NFW r(r+rs)2 r r ntente -5 60 withparameters, given inTable 1.Therelationbetween virial pa- cece rametersandρs,rs caneasilybefound(seee.g.theAppendixA ---10 OffOff 40 ofBoyarskyetal.2007). -15 To explore the uncertainty of the DM density profile in the 20 inner part oftheGalaxy, wealsodescribe theDMdistributionin -20 theMWviaanisothermalprofile(Bahcall&Soneira1980): -25 0 v2 1 ρ -25-20-15-10 -5 0 5 10 15 20 25 ρ (r)= h = 0 . (12) iso 4πG r2+r2 1+(r/r )2 N c c OOffff--cceenntteerr aannggllee [[ddeegg]] ThefollowingparametersofisothermalprofilereproducetheDM contribution to the (outer parts of) Galaxy rotation curve v = h 170 km/sec and r = 4kpc (Boyarskyetal. 2006c, 2007) (i.e. c Figure 4. Dependence of the effective area onthe off-axis position ofa 2 2 ρ = 1.2 106 keV vh 4kpc ).Theseparametersare (point)source. 0 × cm3h170km/si h rc i consistent withthose, from favoredNFWmodels of Klypinetal. (2002);Battagliaetal.(2005),i.e.forφ > 90◦ thedifferencebe- Inthegeneralcaseofextendedsources,evaluationofthede- tweenisothermalmodelandNFWwithpreferredparameterswas tector count rate (7) analytically is not possible because of the completely negligible (less than 5%) – c.f. FIG.5. Both types of complicateddependenceoftheeffectiveareaontheoff-axisangle modelsprovidethelocalDMdensityatthepositionoftheSunto (shownonFig.4).Inthesimplestcaseofanextendedsourcewith beρDM(r⊙)≃0.22GeV/cm3,whichisclosetotheexistingesti- a constant surface brightness dF (φ)/dΩ = f = const, the mates(Kuijken&Gilmore1989c,a,b,1991;Gilmoreetal.1989). DM ext integralofEq.(7)reducestothemultiplicationbythesolidangle The DM flux from a given direction φ, measured by an ob- Ω 0.29andtheeffectivearea,averagedovertheFoV: server on Earth (distance r⊙ 8.5kpc from the GC), is given PCFOV ≃ by ≃ 1 Aeff,ext(Eγ)= ΩPCFOVFZoZVdαdβAeff(Eγ|α,β) Siso(φ)=ρ0Rrc2 ×8< πa2r+ctaanrctan“Rr⊙Rcosφ,”, ccoossφφ><00 , ≈κ(Eγ)Aeff,on(Eγ). (8) “r⊙|cosφ|” (13) : Thenumericalfactorκ(Eγ)dependsontheenergyandhastobe whereR= r2+r2 sin2φandρ r 1.5 1028keV/cm2. calculated via a numerical integration over the energy dependent q c ⊙ 0 c ≃ × TheuncertaintyoftheDMradialdensityprofileintheinner off-axis response map of the SPIdetector. A reasonably accurate Galaxystemsfromthedifficultyofseparationbetweenvisibleand numericalapproximationtoκ(E )isgivenby γ DMcontributionstotheinnerGalaxyrotationcurve.6 Inorderto κ(E) 0.165(E/keV)0.11 . (9) getthemostconservativelimitonthecolumndensityoftheDMin ≈ thedirectionoftheGC,onecanassumethefollowing“rigidlower Onecanseethatκ 1inalltheenergyinterval.Thisisexplained bound”:whiletheDMoutsidether isdescribedbythe“maximal ≪ ⊙ bythefactthatthedetectorareavisiblefromagivendirectionon disk” model (model A of Klypinetal. 2002), for r 6 r DM 2 ⊙ thesky strongly decreases withthe increase of the off-axis angle densityremainsconstant (sothat thetotalDMmasswithinr is ⊙ of this direction, so that the sky-averaged effective area is much thesameasinthemodelA ofKlypinetal.2002).Thisgives 2 smallerthantheon-axiseffectiveareaofthedetector.Substituting M keV (9), (8) into(7) onefinds that for anextended source of constant ρmin 3.9 106 ⊙ =0.146 106 . (14) surfacebrightnessthedetectorcountrateis DM ≃ × kpc3 × cm3 The surface brightness profile on the “constant density” model cts 1keV R =2.73 10−5 (10) is shown in black dashed line on the Fig.5. One can see that ext × s » Eγ – the difference between the maximal (φ = 0◦) and the minimal Aeff,ext(Eγ) (dFDM/dΩ)ext φ = 180◦) column densities is 3.4 (as compared to 6 for ∼ ∼ ×» 150cm2 –»10−15erg/(cm2ssr)– 6 WhenquotingresultsofKlypinetal.(2002),wedonottaketheeffects ofbaryoncompressiononDMintoaccount.WhiletheseeffectsmakeDM dependenton-axiseffectiveareasofeachofthe17operatingdetectorsof distributioninthecoreoftheMWdenser,anysuchcomputationisstrongly SPI,extractedfromtheinstrument’scharacteristicsfiles. modeldependent. 6 A. Boyarskyet al. Table1.Best-fitparametersofNFWmodeloftheMWDMhalo.Max.diskmodelmaximizesamountofbaryonicmatterintheinner3kpcoftheMWhalo (MDM/(Mdisk+Mbulge)=0.4forthemodelA2andMDM/(Mdisk+Mbulge)=0.14inthemodelB2). References Mvir[M⊙] rvir[kpc] Concentration rs[kpc] ρs[M⊙/kpc3] Klypinetal.(2002),favoredmodels(A1orB1) 1.0×1012 258 12 21.5 4.9×106 Klypinetal.(2002),Max.diskmodelsA2 0.71×1012 230 5 46 0.6×106 Klypinetal.(2002),Max.diskmodelsB2 0.71×1012 230 10 23 3.1×106 Battagliaetal.(2005) 0.8+1.2×1012 255 18 14.2 11.2×106 −0.2 ThelowerboundontheDMdecaylinerateinSPIpointingstoward theinner Galaxy iscalculated bysubstituting thecolumn density =1028keV/cm2(seeFig.5)intoEqs.(10),(16) S cts (φ) sin22θ M 4 Fmin ≃3.0×10−6cm2s»102S8DkMeV/cm2–»10−10 –»1kDeMV– (17) Theapproximationoftheconstantsurfacebrightnessworkswell, iftheextendedsourcehasacoreoftheangulardiameterexceeding thesizeoftheSPIpartiallycodedFoV(Θ 17◦ maximal PCFOV ≈ off-axisangle).Takingisothermalprofiletheangularsizeoftheflat coreoftheextendedsourceis φ =arctan(r /r ) 25◦ , (18) core c ⊙ ≃ whichsatisfiesthisconstraint. 3 STRATEGYOFSEARCHFORTHEDMDECAYLINE WITHSPI TheMWhalocontributiontotheDMdecaysignalrepresentsthe all-skysource.Indeed,astheresultsofSection2.1show,thevari- Figure5.ExpectedcolumndensityforvariousDMprofiles:favoredNFW abilityofthesignalovertheskymaybeaslowasthefactor 3. profile(redthick solidline); NFWprofilewiththemaximaldisk(model ∼ A2,seeTable1)–bluesolidline;cored(isothermal)profile–greenthick ThismakesthestrategyofsearchoftheDMdecaysignaldifferent dashedline;constantdensitywithinr⊙–blackdashedline). from any other types of astrophysical sources: the point sources, diffusesources(e.g. 10◦Gaussianprofilefore+e−annihilation ∼ region,Kno¨dlsederetal.(2005))oreventhesearchforDManni- isothermal model). For comparison we show on Fig. 5 expected hilationsignal(seee.g.Tasitsiomietal.2004;Boehmetal.2004; DM flux (6) for various profiles. The minimal column density is Diemandetal.2007;Sa´nchez-Condeetal.2006;Carretal.2006). ofcoursetheoneinthedirectionofanti-center: (φ = 180◦) Theproblemgetsexacerbatedbythefactthatduringitsmo- S ≃ 0.33 1028 keV/cm2.Weseethatevenfortheminimalprofile tion,SPIisirradiatedbythechargedhigh-energy particles(parti- S(φ<×30◦)>1028keV/cm2. clesfromEarthradiationbelt,Solarwind,cosmicTeVphotons).As aresult,thematerials(evendetectorsthemselves)usedforSPIcon- structionstarttoradiateindifferentenergyregions(seesubsection 2.2 DMdecaylinecountrate 3.2).AsaresultanySPIspectrumconsistsofabroadcontinuum, In the case of the Majorana sterile neutrinos of mass M the whichisacombination oftheskyandinstrumental backgrounds, DM DMdecaywidthisgivenby(Pal&Wolfenstein1982;Bargeretal. andofasetoftheinstrumentalbackgroundlines(Attie´etal.2003; 1995):7 Diehletal.2003;Jeanetal.2003;Weidenspointneretal.2003).In ordertodetectaspectrallineproducedbyanastrophysicalsource Γ 1.3 10−32 sin22θ MDM 5 s−1. (15) one has to be able to (a) separate the continuum and line contri- DM ≃ × »10−10 –»1keV– butionstothespectrumand(b)separatetheinstrumentalandsky Substituting(15)to(6)wefind signalcontributionstothelinesfound. Onecanexpectthreeapriorisituations: dF DM(φ) 8.3 10−15erg/(cm2ssr) dΩ ≃ × (I) DM decay line is strong (its equivalent width much larger (16) thanthespectralresolution)andatitspositionthereareno sin22θ M 5 (φ) otherstronglines(ofeitherinstrumentalorastrophysicalori- DM SDM gin).Suchaline,duetoitspresenceinanySPIspectrumand ×»10−10 –»1keV– »1028keV/cm2– itslowvariabilityovertheskycaninprinciplebeconfused withsomeunknowninstrumentalline. 7 ThequotedvalueofΓ isfortheMajoranasterileneutrino.Incaseof (II) DMlineisweak( 3 4σ detectionoverthecontinuum) DM ∼ − Diracparticlethisvalueis2timessmaller (c.f.Pal&Wolfenstein1982; butitspositionalsodoesnotcoincidewithanyinstrumental Bargeretal.1995). line. ConstrainingDM propertieswithSPI 7 (III) DM decay line coincides with some instrumental line. To information. There exist various “background tracers” (Gedetec- beablefindsuchalineweneedtomodelSPIinstrumental torssaturationrates,anti-coincidence shieldrates,ratesofcertain background. background lines, see Jeanetal. (2003); Teegardenetal. (2004); Teegarden&Watanabe(2006)andrefs.therein). Tobeabletoworkeffectivelywithallthesesituations,weneedto findthewaytoseparatethesourceandbackgroundcontributions. 3.3 Searchingforthelines TobeabletodetectstrongDMline,whichisnotcloseinposition 3.1 Imaging toanyinstrumentalline(caseIabove),weusedthemodificationof To distinguish source and background contribution to the signal, themethodofbackgroundsubtraction,describedinTW06.TW06 one often uses imaging capabilities of an instrument. If the size looked for γ-ray lines, assuming different types of sources, from of a point or even an extended source on the sky is smaller than the point sources to the very diffuse sources (10◦ Gaussian, 30◦ the size of the SPI FoV one can (at least, to some extent) use flat,etc.)TW06showedthatthestrongbackgroundlineat198keV the imaging capabilities of the SPI instrument. In this case the canbeusedasabackgroundtracer,ifbackgroundobservationsare coded mask, placed above the detector, partially screens the in- matchedcloseintimetothecorresponding“source”ones.Thisal- dividual detectors from the source, so that the source at a given lowedTW06tocancelallstronginstrumentallineswiththepreci- positionontheskyproduces different count ratesindifferent de- sionbetterthan1%.TW06detectednoemissionlineinsuchback- tectors. One can find the source flux by comparing the ratios of groundsubtractedspectrum(apartfromthe511keVand1809keV) the actual count rates in different modules of the detector to the withthesignificanceabove3.5σ. ones predicted by the degree of screening of the modules by the WeadoptthefollowingmodificationoftheTW06method: mask (see Dubathetal. 2005; Skinner&Connell 2003). It is a – AstheDMdecaysignalremainsnearlyconstantwithincen- challenge, however, to use the imaging capabilities of the SPI to tral30 50◦,themethodofTW06,ifapplieddirectly,couldcancel separatetheastrophysicalsignalfromtheinstrumentalbackground mosto−ftheDMsignal.8Wethereforesubtractthedata(renormal- ifthesizeoftheextendedsourceiscomparabletothesizeofthe izedbythestrengthof198keVline)inthedirectionawayfromthe SPIFoV(seee.g.Kno¨dlsederetal.2005;Allain&Roques2006; GC(off-GCangleφ > 120◦)fromtheON-GCdataset(theangle Weidenspointneretal. 2007, and refs. therein). Therefore for our φ613◦). analysiswedidnotuseanyimagingcapabilitiesofSPI,andtopro- – Intheresulting“ON–OFF”spectrumweperformthesearch duce spectra fromsome point inthe sky wejust collectedall the forthelinewiththesignificancehigherthan3σ. photons,arrivingintheSPIFoV. This procedure allows to eliminate strong instrumental lines withtheprecisionbetterthanfewpercents.Atthesametimeany strongDMlinewouldremaininthe“ON–OFF”spectrum.Indeed, 3.2 SPIbackgroundmodeling even for the flattest profile (Section 2.1), the strength of the DM signal in the OFF dataset is at least 60% weaker that of the ON Intheabsence ofimaging, theseparationof theinstrumental and dataset.Thereforewesee,thatthemodification,describedabove, astrophysicalcontributionstothelinespectrumrequiressomesort is indeed well suited for searching of the strong DM decay line of background modeling (see e.g. Weidenspointneretal. 2003; (caseI). Teegardenetal.2004; Teegarden&Watanabe2006).Namely, for However,thismethoddoesnotworkwellfortheweak(3 4σ) thebackgroundmodelingwecanusethefactthatforanyDMdis- − lines,orforthelines,whosepositioncoincideswithsomeinstru- tributionmodeltheintensityoftheDMdecaylinechangesbyafac- mentalline(casesII-IIIabove).Indeed,inthiscaseitisnotpos- tor>3betweenthepointingstowardstheGalacticcenter(φ 0◦) sible to tell whether the remaining line is the residual of the in- ∼ andanti-center(φ 180◦,seeSec.2.1).Ontheotherhand,ifthe strumentaloneorhastheastrophysicalorigin.Belowwewilluse ∼ lineisofpurelyinstrumentalorigin,thereisnoa-priorireasonwhy analternativemethodofanalysisofthedetectedlines,suitablefor thestrengthofthelineinthebackgroundspectraofthepointings casesIIandIII. towardse.g.theGalacticAnti-centershouldbedifferentfromthe strength of the line in the spectra of the pointings toward e.g the GC.Thus,onepossiblewaytodistinguishbetweentheDMdecay 3.4 Analyzingacandidateline andinstrumentaloriginofthelineistostudythevariationsofthe line’sstrengthdependingonitsskyposition(inthesimplestcase– Havingdetectedanumberoflineswiththesignificanceof3σand onthe“off-GC”angleφ,ofthepointing,Eq.(5)). above,weshoulddecidewhichofthemcanbeconsideredas“DM Thesituationbecomesmorecomplicatedduetothefactthat decaylinecandidates”.Tothisendwedothefollowing. the instrumental background (and thus the intensity of the in- a) Wecomparelinefluxforeachoftheselineswiththeflux strumental lines) experiences great variability in time (depend- ofthesamelineinthe“ON”spectrum.Wedecidethatthelineis ing on the position in orbit, solar flares and the solar activity a“DMlinecandidate”ifthecancellationofthefluxbetweenON period, degradation of the detectors, etc., c.f. Jeanetal. 2003; andOFFdatasetswasworsethan10%.9 Teegardenetal. 2004). As observations of different parts of the skycanbesignificantlyseparatedintime,oneneedstouse“back- 8 ForexampleforthemostconservativeDMdistributionmodel,thediffer- ground tracers” to find the correct spatial dependence of the line enceofDMsignalsatφ=0◦andφ=30◦ismere8%. intensity(Jeanetal.2003;Teegarden&Watanabe2006).Without 9 In principle, the DM line in ON–OFF spectrum should not cancel by somesort of “renormalization” procedure, which correctstheab- morethan∼ 40%,whilethebackgroundinstrumentallineshouldcancel solute value of the line flux using a measurement of a specific betterthan1%.Thusthechoiceofthethresholdtobearound10%ensures characteristicsof theSPIinstrument asa“calibrator”of the flux, that noDMdecay line was thrown away whilemostoftheinstrumental theφdependence foranyofthedetectedlinescontainsnouseful linesdisappeared. 8 A. Boyarskyet al. b) Forany“DMcandidateline”weconstructits“spatialpro- file”(asdescribed indetailsin thenext Section) tocheck for the ON-OFF possibility of it to be a DM decay line (we also construct distri- 0.003 ON spectrum × 0.01 bution of the line flux over the sky for all the unidentified lines fromWeidenspointneretal.2003).Sincethecolumndensityofthe V] ke 0.002 DtuhaMelddiienrcertchetaeiosdeniroteofcwtthiaoerndlittnohewesaGtrradelnatghctethicGwaCintthsih-tcoheuenlitdnecrb,reeoahnsieginshgheorauntlhgdalneseφteh.aaWtgoerfaddino- ate [cts/sec/ 0.001 notmakeanyspecificassumptionabouttheDMdensityprofileand nt r donottrytofitthecandidatelinespatialprofiletoanyparticular ou 0 C model,butratherlookifthereisageneraltrendofdecreasingin- tensityofthelinewiththeincreasingoff-GCangle. -0.001 50 100 150 200 250 E [keV] 4 DATAREDUCTION Figure 6. Comparison of ON–OFF spectrum (thick solid line) with the 4.1 ONdataset 0.01× theONspectrum afterthesliding window (thin dashedline). It canbeseenthattheinstrumentallinesaresubtractedwiththeprecisionbet- Duringitsalmost 5yearsinorbitINTEGRALhasintensivelyob- terthan1%. served the inner part of the Galaxy (Galactic Center, Galactic BulgeandtheinnerpartoftheGalacticPlane)andcollectedabout 4.2 ON–OFFdataset T 10MsofexposuretimeintheGCregion.Inouranalysis exp ∼ oftheinnerGalaxyweusedthepubliclyavailabledata(asofJuly Most of the lines found in the continuum subtracted background 2007)fromallINTEGRALpointingsatwhichtheangleofftheGC spectrum are of the instrumental origin. To remove them, we wasat most 13◦ and for whichtheSPIexposure timewas larger matchedeachScWintheONdatasetwiththepointingawayfrom than1ksec.Thiscriteriaselects5355pointings(or“ScienceWin- the GC (galactic coordinate φ > 120◦) – OFF pointing. As de- dows”,ScW),withtotalexposuretimeof12.2Ms,spreadoverthe scribedbyTW06,the198keVlinecanserveasgoodbackground periodfromFebruary2003,tillApril,25,2006.Wecallthisdataset tracer if the time duration between ON and OFF observations is "ON"dataset. 6 20 days. We were able to match 3688 ON-OFF pairs. For For each of the analyzed ScWs, we have extracted photon eachON–OFFpairweintroducednormalizingcoefficientnforthe (event) lists from spi-oper.fitsfiles and applied additional OFF spectrum in such a way that the strong instrumental line at energy correction to convert the channel number into photon en- 198keVcancelscompletelyaftersubtractionoftheOFFspectrum ergy, using spi gain cor tool from standard Offline Analysis multiplied by the factor n from the ON spectrum. After that we Software (OSA). We have binned the events into narrow energy subtracted(renormalized)OFFScWfromthecorrespondingScW bins of the size ∆E = 0.5 keV to generate the background from theON dataset. Thisallowed us to remove most prominent bin countsspectraineachScW,eachrevolutionand,subsequently,in instrumental lines with the precision better than 1% (c.f. Fig. 6). theentiredataset. To avoid contributions of strong astrophysical sources (such as Wethenappliedthe“slidingspectralwindow”methodtothe e.g. Crab) we threw out all pairs with negative total flux at 20– linesearch(asdescribede.g.inTW06toproduceacontinuumsub- 40keVrangeaftersubtraction.Takingaverageover2456remain- tractedspectrumofthe"ON"dataset.Namely,ateachgivenenergy ing“good”pairswereceivedthespectrumalmostfreefromback- E ,onedefinesanenergyintervalE 2∆E <E <E +2∆E, groundatenergiesabove200keV.Atlow(<200keV)energieswe 0 0 0 − where ∆E is the SPI spectral resolution at a given energy, as a foundcontinuumcomponent, whichcanbefittedwiththesimple ”linesignal”energyband.Forthe(energydependent)∆Eweused powerlaw: approximateformulafromSPI/INTEGRALgroundcalibrationof E α FWHM(Attie´etal.2003): F(E)=F (20) 0»100keV– ∆E(E)=F1+F2√E+F3E (19) Parametersofthisbackgroundwerefoundtobe wEhiesrienFk1eV=.F1o.r54E,F=21=034k.e6V·1F0W−3H,MF3 =26.3.0k·eV10.−4 andenergy F0 =(4.95±0.05)×10−5cts/s/cm2 (21) ≈ α= (2.264 0.003) ForeachenergybincenteredatanenergyE wehavedefined − ± 0 thetwoadjacentenergyintervals,E 4∆E <E <E 2∆E Thiscontinuumrepresentstheresidualcontributionfromalltheset 0 0 andE +2∆E <E <E +4∆E,a−ndpostulatedthatthe−sumof oftheastrophysicalsourcespresentintheGalacticBulge. 0 0 thecountratesinthesetwoadjacentenergybandsgivesthemea- sure of the continuum count rate in the energy band around E . 0 4.3 Systematicerror Subtractingthesumofthecountratesintheadjacentenergybands fromthecountrateinthe”linesignal”energyband,wehavecal- Toestimatethesystematicerrorofour“ON-minus-OFF”dataset, culatedthecontinuumsubtractedcountrateatagivenenergyE . wecomputedbackgroundaroundthe“tracerline”of198keV.We 0 Doingsuchprocedureatalltheenergies20keV<E <8MeV, found that it does not vanish. Thus, we estimated the systematic 0 wehaveproduceda“continuumsubtracted”SPIbackgroundspec- error as the error in thenormalization coefficient n which would trum.Inthisspectrumwewereabletoidentifymostoftheknown makethebackgroundzerowithinsystematicuncertainty.Thiscor- instrumentallines(Weidenspointneretal.2003). rectionδncanbefoundasfollows.Letnbethecoefficient,needed ConstrainingDM propertieswithSPI 9 Phase Revolutions(start-stop) 180 1 042–092 096–140∗ 160 2 140-205 140 209-215∗ 3 215-277 120 4 282-326 5 330-395 deg] 100 6 400-446 φ [ 80 Table2.Splittingrevolutionsintophasesincorrespondencewithannealing 60 phasesandbreakageofthedetectors(revolutions,markedwiththe∗). 40 20 tocancelfluxinthe198keVlineinONandOFFspectra: 0 0 100 200 300 400 500 600 n=F /F (22) ON OFF Revolution whereF ,F –fluxesin198keVlineinONandOFFScWs ON OFF Figure7.Positionontheskyasafunction ofrevolution over6yearsof correspondingly. Theremainingnon-zero δF fluxintheadjacent INTEGRAL observations. The periods of annealing phases are shown in to line position in ON-OFFspectrum, determines the uncertainty solidverticallines.Twodashedlinesindicatetherevolution,duringwhich ofthecoefficient: 2NDand17thof19SPIdetectorshavefailed. δF δn= (23) F ON We found, that average value of δn is equal to δn = 1.1 0.009 h i h i · 10−3, n 1.So,oursystematicerroroffinalON–OFFspectra atenerghyEi∼is1.1 10−3FOFF(E) 1.1 10−3FON.Weaddthis 0.008 · ≈ · systematic uncertainty to the flux of ON–OFF spectrum in every V] energybin. ke 0.007 4.4 Obtaining3σrestrictions ate [cts/sec/ 0.006 Attheenergiesatwhichnolinesweredetected(i.e.the“continuum unt r 0.005 o subtracted”countratedidnotdeviatebymorethan3σfromzero) C weobtainedthe3σ upperlimitonthepossiblefluxfromtheDM 0.004 decay.Above 200keVthefluxintheenergybiniszerowithin ∼ statisticalerrors,therefore3σupperlimitfluxisgivenbystatistical 0.003 plussystematicerrors.Below200keVweputstatisticalrestrictions 0 20 40 60 80 100 120 140 160 180 φ [deg] abovepowerlawcontinuumflux(20),describedintheSection4.2. UsingEq.(17)onecanderivetherestrictiononthesterileneutrino Figure8.ScatterofthefluxdatapointsforthelineatE=1068keVasa mixingangle,impliedbythisupperlimit.Oneshouldalsotakeinto functionoftheoff-GCangle. accountthatthesubtractionoftheOFFobservationsledtothere- ductionof theexpected DMsignal. Takingthemost conservative “minimal”model,describedinSection2.1,weseethatthesubtrac- tionoftheOFFsignalleadstoabout40%decreaseoftheexpected possibletodistinguishbetweentheinstrumentalandDMdecayori- DMsignal.10Theresulting3σboundisshownonFig.10. ginofalinebyfittingF(φ)withaknownprofilecalculatedfrom the radial DM density profile. However, the details of the radial DM density profile inthe inner Galaxy are highly uncertain, and 4.5 PossibleDMcandidates thispreventsusfromdirectlyfittingthemodelprofiletothedata. WeadoptedasimplecriterionwhichselectsaDMdecaycandidate WhenanalyzingON–OFFspectrum,wefoundthatalmostalllines, line:theratiooffluxes presentinONspectrumcancelwithprecisionbetterthanfewper- cents. We found 21 lines (see Table 3) that did not cancel by at F(0◦) = > 3. (24) least90%,(includingknownlinesat511keVand1809keV).Apart R F(180◦) Rmin ≃ fromthese2linesallotherlinesaredetectedwithlowsignificance 3 4σ. where min istheratiooftheDMdecaylinefluxesfromtheGC − As discussed in Sections 3.3–3.4, we took all these lines as andtheRGalacticanti-centerinthe“minimalDMcontent”modelof possible DM candidates and analyzed the dependence of theline DMdistribution. fluxesF(φ)ontheoff-GCangleφofthepointing.IftheDMdis- Sincetheobservationsatdifferentoff-GCanglesaredonedur- tributionintheinner partoftheGalaxywereknown, itwouldbe ingdifferenttimeperiods,toproperlystudythedependenceofthe lineflux on the off-GC angle φ one should take into account the timevariabilityof the response of the SPIdetectors. Several fac- 10 Toestimatethis,wetookthemaximalcolumndensityforOFFobserva- torshavetobetaken intoaccount. First,theSPIinstrument goes tionsatφ=120◦−17◦. throughaso-called“annealing”phase–heatingofthedetectorsto 10 A. Boyarskyet al. 6.1×10-2 1021 6.0×10-2 2m] 6.0×10-2 1020 ne flux [cts/sec/c 55..89××1100--22 τLife-time [yr] 1019 EXCLUDED Li 5.8×10-2 5.8×10-2 1018 0 20 40 60 80 100 120 140 160 180 20 100 1000 7000 Off-center angle φ [deg] Eγ [keV] Figure9.Dependenceoftheintensityofthepositronannihilation lineat Figure11.Life-timeoftheradiatively decayingDMasafunctionofthe E =511keVontheoff-GCangle.Thesolidlineshowsfittothedatain emittedphotonenergy.Regionbelowthecurveisexcluded. theformconst+Ne−φ2/(2σ2). 5 RESULTS WeanalyzedthespectrumofSPIandfoundthatnoneofthestrong recoverfromaradiativedamage.11Next,twoofthe19SPIdetector (i.e.detectedwithsignificanceabove5σ)linescanbeinterpreted have“died”.12Thefaileddetectorsalsoaffecttheresponseoftheir asthatofthedecayingDM.Thisconclusionwasbasedonthefact neighbors. Tomarginalizetheeffectsofthechangingresponseof thatvariabilityoftheselinesovertheskyislessthan10%(when theSPIdetector,wesplittheentiredatasetinto7periods,asshown movingfromGCtotheanti-center,seeFig.8).Atthesametimefor onFig.7.TheintervalsaresummarizedintheTable2.Asbothde- anyrealisticDMmodelsuchavariabilitywouldbegreaterthanat tector failuresoccurred soon after theend of anannealing phase, least60%.Thus,weexcludethepossibilitythatoneofthespectral wechose toignorerevolutions136 through 140and209 through lines,detectedintheSPIbackgroundspectrumisaDMdecayline. 215. The period 096–140 does not cover theessential part of the The non-detection of a DM decay line in the entire energy skyandthereforeweskipit,leavingonly6periods. rangeoftheSPIdetectorhasenabledustoputanupperlimitonthe Foreachoftheperiods,shownintheTable2,weplotthedis- parametersoftheDMparticles.Inparticular,the3σ upperbound tributionofthelinefluxasafunctionoftheoff-GCangleφ.The onthemixingangleofthesterileneutrinoDMinthemassrange resultsaresummarizedonFig.13,p.16.Onecanseethatnoneof 40keV–7MeVisshownonFig.10,p.12. theselinesexhibitscleartrendofdecreasingfromφ=0◦towards OurresultsareapplicabletoanydecayingDM.Tothisendwe φ = 180◦. For each line (and each phase) we also compute the alsopresenttherestrictionsontheDMlife-time(withrespecttothe averagefluxF¯,standarddeviationσ fromtheaverage,minimum F radiativedecay)asafunctionoftheenergyofemittedphoton.The (F )andmaximum(F ).Ouranalysisshowsthat(a)95 100% min max − correspondingexclusionplotisshownonFig.11.Forexample,the ofallpointsliewithin3σ fromtheaverage(thus,thedataiscon- F gravitinocandecayintotheneutrinoandphoton(similarlytothe sistent with having flat spatial profile) and (b) the scatter of the caseofsterileneutrino)insupersymmetrictheorieswithbrokenR- data(F F )ismuchlessthanitsmeanvalue F .Therefore, max− min h i parity.Suchaninteractionisgeneratedviatheloopeffects(seee.g. noneofthemcannotoriginateentirelyfromaDMdecay.Thecor- Borganietal. 1996; Lolaetal. 2007). The restrictions on Fig. 11 respondingnumbersforeachlineandeachphasearesummarized improve existingbounds on thelife-timeof such a gravitinoDM inTable4,page17. byseveralordersofmagnitude(c.f.Borganietal.1996). The positron annihilation lineat E = 511keV illustratesa Topresentourresultsintheformlessdependentonaparticu- situation, when a line of astrophysical origin is superimposed on larmodelofDMdistributionintheMW,weshowthe3σsensitivity top of the strong instrumental line. In this case, the data can be towardsthelinesearchonFig.12.Note,thattheseresultsshould fittedbytheconstant,plussomefunction,depending onassumed beusedwithcare,asthesensitivitydependsontheassumedspatial shapeofthesource.Fig.9showsthedependenceofthefluxofthe profileofthesource(becausetheeffectiveareadecreaseswiththe 511keVlineontheoff-GCangleoftheSPIpointing.Onecansee off-axisangle,seediscussioninSection2).Theresults,presented that for the pointings with theoff-GC angle lessthan 20◦ (about onFig.12arevalidforanextendedsourcewiththesurfacebright- the size of the PCFOV of SPI) the 511 keV line flux contains a nesswhichvariesontheangularscaleslargerthan(orcomparable contributionfromaskysourceatthepositionoftheGC,whilefor to)thesizeoftheSPIfieldofview(blacksolidline).Thisplotis thepointingatlargeroff-GCanglestheastrophysicalsourceisnot analogoustotheFig.9ofTeegarden&Watanabe(2006)(TW06). visibleandtheonlycontributioncomesfromtheinstrumentalline, However,adirectcomparisonoftheFig.9ofTW06andFig.12is whosefluxdoesnotdependontheoff-GCangleofthepointing. notpossible,sinceTW06haveassumedadifferentmorphologyof theextendedsource(10◦Gaussian).Explicitlytakingintoaccount thedependenceoftheeffectiveareaoftheSPIdetectorontheoff- axis angle (see Section 2), one can find that in order to make a 11 FordetailsseeSPIUserManual: direct comparison between the two figures, one has to “re-scale” http://isdc.unige.ch/Instrument/spi/doc/spi um. theresultsof Fig.9of TW06by an(energydependent) factor of 12 Detector#2atrevolution140anddetector#17atrevolutions214-215. 1.5.Thisfactorconvertsthesensitivityfortheline,producedby ≈

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.