Sagittarius dwarf spheroidal galaxy observed by H.E.S.S
G. Lamanna, C. Farnier, A. Jacholkowska, M. Kieffer, C. Trichard
aa r X i v : . [ a s t r o - ph . H E ] J u l RD I NTERNATIONAL C OSMIC R AY C ONFERENCE , R
IO DE J ANEIRO T HE A STROPARTICLE P HYSICS C ONFERENCE
Sagittarius dwarf spheroidal galaxy observed by H.E.S.S.
G. L
AMANNA , C. F ARNIER , A. J ACHOLKOWSKA , M. K IEFFER , C. T RICHARD FOR THE
H.E.S.S.C
OLLABORATION . LAPP, Universit´e de Savoie, CNRS/IN2P3, F-74941 Annecy-le-Vieux, France. OKC, Physics Department, Stockholm University, AlbaNova SE-10691 Stockholm, Sweden. LPNHE, Universit´e Paris VI et Paris VII, CNRS/IN2P3, 4 Place Jussieu, F-75252, Paris Cedex 5, France.
Abstract:
Dwarf spheroidal galaxies are characterized by a large measured mass-to-light ratio and are notexpected to be the site of high-luminosity non-thermal high-energy gamma-ray emissions. Therefore they areamong the most promising candidates for indirect searches of dark matter particle annihilation signals in gammarays. The Sagittarius dwarf spheroidal galaxy has been regularly observed by the High Energy StereoscopicSystem (H.E.S.S.) of Cherenkov telescopes for more than 90 hours, searching for TeV gamma-ray emission fromannihilation of dark matter particles. In absence of a significant signal, new constraints on the annihilation cross-section of the dark matter particles applicable for Majorana Weakly Interacting Massive Particles (WIMPs) arederived.
Keywords: dark matter, gamma rays, H.E.S.S., dwarf galaxy
A large number of observations from Galactic to cosmolog-ical scales support the explanation that Dark Matter (DM)would be composed by a new type of particle although it-s nature remains unknown. The proposal of WIMP pre-dicted by theories beyond the Standard Model of particlephysics provides a relic abundance accounting for the in-ferred amount of DM [1]. The WIMP search is conductedin three ways: by particle production at the LHC, probingthe theories of Standard Model extension; by searching fornuclear recoil signals experiments, probing the WIMP s-cattering cross section; by indirect searches of a signal inthe secondary products of WIMP annihilation, probing thecorresponding cross section.Potential spectral signatures in gamma rays can be clas-sified in mainly strong spectral features and ambiguous sig-nals. The first class is for instance due to annihilation in-to gg or Z g producing a sharp line spectrum with a photonenergy depending on the WIMP mass, e.g. the Supersym-metric (SUSY) neutralino hypothesis. Unfortunately, theseprocesses are loop-suppressed and therefore very rare. Tosome extent more ambiguous are signals due to continuumemission from pion decay resulting from the WIMPs anni-hilation in pairs of leptons or quarks. The number of gam-ma rays finally originated by WIMP annihilation dependsquadratically on the DM density along the line of sight ofthe observer. This motivates a number of promising target-s for indirect DM searches, namely those with known den-sity enhancements, in particular the Galactic Centre andclose-by dwarf galaxies and galaxy clusters. More specif-ically, assuming the L CDM cosmological model, the hier-archical collapse of small over-densities are formed by D-M structures which may also host smaller satellite struc-tures and it has been proposed that dwarf spheroidal galax-ies may have formed within some of these sub-halos host-ed in the larger Milky Way DM halo [2].
The Sagittarius Dwarf Spheroidal Galaxy (SgrD) was dis-covered in 1994 and it is located at RA = 18 h m s ,Dec = -30 ◦ ′ ′′ in equatorial coordinates (J2000.0),at a distance of about 24 kpc from the Sun [3]. SgrD isone of the nearby dwarf spheroidal companion galaxiesof the Milky Way. Characterised by an extremely low sur-face brightness it has been estimated that SgrD orbits theMilky Way within less than a billion years and thereforeit must have already passed through the galactic plane atleast about ten times. Although its high level of tidal dis-ruption, the hypothesis of a large DM content would be re-sponsible to hold onto so many of its stars and for so long. The High Energy Stereoscopic System (H.E.S.S.) is an ar-ray of Imaging Atmospheric Cherenkov Telescopes (IACT-s) the working principle of which is based on the detec-tion of faint Cherenkov light from the gamma-ray-inducedair showers in the atmosphere. Located in the KhomasHighland in Namibia, 1800 m above sea level, H.E.S.S.began operation in 2003 with four 13 m dish-diametertelescopes, equipped with cameras containing 960 photo-multiplier tubes (PMTs) and operated in a coincidencemode, in which at least two of them must have been trig-gered for each event within a coincidence window of 60 n-s. Since September 2012 one additional larger 28 m dish-diameter telescope (and a camera with 2048 PMTs) has s-tarted operation enabling to achieve a larger sensitivity ofthe full array in the range of about 100 GeV to 10 TeVwhile increasing the lower energy domain to ∼
30 GeV byexploiting the larger telescope in standalone mode.The H.E.S.S. collaboration observed SgrD searching fora potential emission of very-high-energy (VHE) gammarays during 11 h in 2006. In absence of any signal an up-per limit on the gamma-ray flux and a constraint on thevelocity-weighted annihilation cross-section in the frame agittarius dwarf spheroidal galaxy observed by H.E.S.S.33 RD I NTERNATIONAL C OSMIC R AY C ONFERENCE , R
IO DE J ANEIRO2013
IO DE J ANEIRO2013
RA-287.35 -280.38 D ec ( d e g ) -33-32-31-30-29-28 -4-3-2-10123418h50m19h00m Sgr−dSph Significance Map -4 -2 0 2 4110 / ndf c – – – c – – – c – – – Significance distribution
Fig. 1 : Significance sky map: no excess observed at the target position (left panel). Significance distributions related to theon-source and off-source events which are well overlapped and compatible each other (right panel).of SUSY models as well as extra-dimensions models werepublished [4]. A deeper observation of SgrD was conduct-ed accumulating further 90 hours of observation in 2007-2012 using the four 12 m dish telescopes. In the followingthe procedure and the results of the analysis of these newobservations are described and discussed.
H.E.S.S. has observed SgrD from 2006 to 2012. The ob-servations were performed in wobble mode, i.e. with thetarget offset by 0.7 ◦ to 1.1 ◦ from the pointing direction,enabling simultaneous background estimation and in thesame field of view. The data used for the analysis were tak-en at different zenith angles spanning from few degrees upto 45 ◦ and an average value of ∼ ◦ over the full data sam-ple. Some standard quality selection cuts were applied re-sulting in a reduction of the data of about 13% correspond-ing to a total 91.5 live-hours data analysed.The X e f f analysis was employed for the selection ofgamma-ray events and for the suppression of cosmic-raybackground events. X e f f denotes a multivariate analysismethod developed to improve signal-to-background dis-crimination, which is important in searches for weak sig-nals [5]. The X e f f method improves the separation ofgamma and cosmic-ray events compared to the standardH.E.S.S. analysis [6], by exploiting the complementary dis-criminating variables of three reconstruction methods inuse in the H.E.S.S. analysis. The resulting unique discrim-inating variable X e f f acts as an event-by-event gamma-misidentification probability estimator. The definition ofthe X e f f probability function follows the relation: X e f f ( d i ) = h (cid:213) j H j ( d i )( − h ) (cid:213) j G j ( d i ) + h (cid:213) j H j ( d i ) , (1)where d i are the discriminating variables (indexed by i ) ofthree reconstruction methods (indexed by j ); G j ( d i ) and H j ( d i ) are the one-dimensional probability density func-tions ( p.d.f. s) for events identified as gamma-ray-like ( G )and hadron-like ( H ) (the product of individual p.d.f. s re-places the global multi-dimensional ones since the vari-ables d i are highly uncorrelated, for gamma-ray events inparticular); h is the misidentified fraction of the gamma class of events (i.e. the relative background fraction). Thefinal gamma-ray event selection was achieved with the setof cuts adapted to the detection of faint sources ( h = . X e f f , cut = .
3) and a cut in the reconstructed image chargerequiring more than 60 photo-electrons. The gamma-raysignal was searched at the target position within an angu-lar size of q ≤ ◦ . No significant gamma-ray excess was found above the es-timated background at the nominal position of SgrD norin the camera field of view as shown by the significancesky-map and the Gaussian distribution compatible with thebackground in figure 1 as well as presented by the flat q distribution in figure 2. The target position is chosen ac-cording to the photometric measurements of the SgrD lu-minous cusp showing that the position of the centre corre-sponds to the centre of the globular cluster M54 [11]: RA= 18 h m s , Dec = -30 ◦ ′ ′′ in equatorial coor-dinates (J2000.0).A 95% confidence level (C.L.) upper limit (N C . L . g ) onthe total numbers of observed gamma-ray events can be ) (deg q Sgr-dSph: 91.5 live hours=13) a ON=2181 OFF=27958 (=0.6 S/B=0.0 s , g Fig. 2 : q radial distribution of the ON events for gamma-ray-like events from the SgrD target position. Estimatedbackground is also shown by black markers overlapped tothe histogram. No excess is seen at small q value.
1. The three reconstruction methods are referred to as Hillas [7],Model [8] and 3D-model [9, 10].agittarius dwarf spheroidal galaxy observed by H.E.S.S.33 RD I NTERNATIONAL C OSMIC R AY C ONFERENCE , R
IO DE J ANEIRO2013
IO DE J ANEIRO2013 thus deduced using the Feldman & Cousins [12] method.This limit is computed using the number of events reg-istered on target region ( ON =2181), those correspondingto the background ( OFF =27958) normalized by the ratio( a =13) between the on-source and the off-source livetimes.The result is N C . L . g = 123. Effects of uncertainties onbackground and a can be safely neglected.Such a limit can be used to constrain indirectly theWIMP annihilation cross-section in a specific modelingcontext. In order to do so it is worth to remind that thedifferential gamma-ray flux ( d F g / dE g ) due to DM particleannihilation depends on the following items: • the particle physics coupling of the DM particle; • the intrinsic DM density distribution in the source; • the field of view DW within which the signal is inte-grated along the line of sight by the observer.It is usually factorised into two terms: d F g dE g ( E g , DW ) = F pp ( E g ) × J ( DW ) DW , (2)where the first factor ( F pp ) takes into account the par-ticle physics model for the WIMP annihilation, while thesecond factor (called J -factor in the following) accountsfor the astrophysics model and denotes the DM density dis-tribution in the source.The particle physics factor is given by: F pp = d F g dE g = p h s ann v i m c × dN g dE g , (3)where h s ann v i is the velocity-averaged annihilation crosssection, m c is the WIMP particle mass, and dN g / dE g isthe differential gamma-ray spectrum summed over the w-hole final states weighted by their corresponding branch-ing ratios. In this work, for simplicity and homogeneitywith previous similar estimates, all channels are taken intoaccount by using a parametrised average spectrum (fromBergstr¨om et al. [13]): dN g dE g = m c dN g dx = m c . e − . x x . , (4)where x = E g / m c .In the astrophysical factor, the integral along the lineof sight of the squared density of the DM distribution inthe object is averaged over the instrument solid angle ofthe integration region. For this analysis with H.E.S.S., DW = 2 × − sr since we are looking for an almost point-like signal considering the point-spread function of theinstrument: J = DW Z DW Z r DM ( l , W ) dld W . (5)Two different models of the DM halo of SgrD were con-sidered in the analysis published by the H.E.S.S. collabora-tion in 2008 [4]. They turned out to be too optimistic sincethe dwarf disruption by tidal winds was ignored. Recentmeasurements (by Niederste-Ostholt et al. [14]) have madepossible to scale down the previous density estimate sinceonly 30-to-50% of the luminosity of SgrD is assumed cur-rently still bound to the remnant core. Recently new mod-eling of the SgrD density which takes into account tides have been published [18]. Such new densities are consid-ered in this work. For completeness the approach followedand the results obtained in [18] are here summarised. A D-M halo described by NFW profile [17] is first considered: r NFW ( r ) = r s ( r / r s )( + r / r s ) , (6)where parameter values: r s = 1.3 kpc is a scale radius and r s = 1.1 × M ⊙ pc is a characteristic density.An isothermal profile for DM halo is also considered. Itis proposed by Pe˜narrubia et al. [15] assuming that SgrD isa late-type, rotating disc galaxy. In this model, the galaxyis composed of a stellar disk and a DM component withthe following density distribution: r ISO ( r ) = m h a p / r cut exp [ − ( r / r cut ) ]( r c + r ) , (7)where m h is the halo mass, r c is the core radius and a ≃ r c = 0.45 kpc. Assuming an initial luminosity of ∼ L ⊙ [14] and a mass-to-light ratio of 24 [16], thetotal mass of the halo is found to be m h = 2.4 × M ⊙ .To take into account the lost of the outer halo envelope dueto tidal disruption by the Milky Way, a truncation in theDM density profile is imposed at r cut = 12 r c = 5.4 kpc. Itroughly corresponds to the tidal radius of a satellite galaxyat pericenter of 15 kpc with a mass ∼ × M ⊙ .The J -factors can be computed using the equation 5.They are summarised for both approaches in table 1 as ex-tracted from [18]. It is important to remind that accordingto the authors of [18] the uncertainties on the halo profileparameters are still large and can affect the value of theastrophysical factor by a factor of 2. Finally the 95% C.L.upper limit on the velocity-weighted annihilation cross-section as a function of the WIMP particle mass and for agiven halo profile is computed as: h s v i C . L . min = p J ( DW ) DW × m DM N C . L . g , tot T obs m DM R A e f f ( E g ) dN g dE g ( E g ) dE g , (8)where T obs is the observation time dedicated to SgrD (91.5hours) and A e f f ( E ) is the H.E.S.S. effective area. The ex-clusion curves for the SUSY neutralino are shown in figure3 referring to the two halo profiles as in table 1. The exclu-sion limits depend on the particle mass and the best sensi-tivity is reached at 1-2 TeV with the value of ∼ × − cm s − still well above the thermal value of h s v i ∼ × − cm s − [19]. As the two astrophysical J -factorsare compatible one each other it results that the two limitsare of the same order of magnitude.Dark Matter density profile J -factor (GeV cm − )Isothermal 0.88 × NFW 1.00 × Table 1 : Astrophysics J-factor computed for two differentdark matter halo profi les [18]. agittarius dwarf spheroidal galaxy observed by H.E.S.S.33 RD I NTERNATIONAL C OSMIC R AY C ONFERENCE , R
IO DE J ANEIRO2013
IO DE J ANEIRO2013 (TeV) c m -1
10 1 10 ) - s v > ( c m s < -28 -27 -26 -25 -24 -23 -22 -21 NFWIsothermalNMSSM models
Fig. 3 : Exclusion limit at 95% C.L. on the velocity-weighted annihilation cross-section versus the DM particle mass for aNFW (green dotted line) and an Isothermal (red line) DM density profiles. NMSSM models scan is shown (blue markers).In order to compare the H.E.S.S. exclusion limit to thepredictions of a set of realistic particle physics models, weperformed a scan of the Next-to-Minimal Supersymmet-ric Standard Model (NMSSM)[20] that constitutes one ofthe most well-motivated extensions of the Standard Mod-el, both theoretically and phenomenologically, especiallyin light of the recent discovery of a Higgs-like particle atthe LHC [22, 23]. The scan was performed using the pub-licly available code NMSSMTools 3.2.4 [21, 24, 25] withparameter values and ranges as: tan b ∈ [1.5, 60]; M , M ∈ [100, 2000]GeV; M ∈ [500, 3000]GeV; l , k ∈ [0.1, 0.8]; A k ∈ [-2000, 0]; m eff ∈ [100, 2000]; M A ∈ [10, 2000] andmaking several discrete choices for A U , A D = A E , m L = m E = m Q , m U and m D depending on the scanned subre-gion. Models with a Higgs particle in the mass range [119;130]GeV were retained. For computing of the predictedrelic density according to WMAP-7 and thermally aver-aged self-annihilation cross-sections at zero velocity foreach considered model, the micrOMEGAs-3.1 [26] wasused imposing a loose relic density constraint of W DM h ∈ [0.087, 0.14], thus resulting in more than 9000 viablepoints. As a conclusion from this scan, the capability toprovide a complementary constraint to LHC seems to belimited for the somewhat less conventional models as in-ferred from figure 3, however in agreement with mostly s-tudied constraint MSSM models. It would be not necessar-ily the case for alternative models, e.g. leptophilic modelswith light mediators [27], to be studied further. The Sagittarius dwarf spheroidal galaxy has been observedwith H.E.S.S. for more than 90 hours, thus enabling ef-ficient search for a dark matter signal coming from thisMilky Way satellite. The absence of a signal leads to con-straints on the velocity weighted mean annihilation cross-section as a function of WIMP mass which were comparedto NMSSM models. The dark matter halo modeling hascrucial impact on the derived limits; an increase in detec-tor sensitivity is mandatory to improve the search potential(as expected with the Cherenkov Telescope Array).
References [1] G. Bertone et al., Phys. Rep. (2005) 405, 279-390.[2] A.V. Kravtsov et al., Astroph. J. (2004) 609, 482-497.[3] S.R. Majewski, Astroph. J. (2003) 599, 1082-1115.[4] F. Aharonian et al., Astropa. Phys. (2008) 29: 55-62,Erratum (2010) 33: 274-275.[5] F. Dubois, G. Lamanna, A. Jacholkowska, Astrop.Phys. (2009), 32, 73.[6] F. Aharonian et al., A&A (2006) 457, 899.[7] A.M. Hillas, Proc. 19th ICRC, La Jolla (1985), 3, 445.[8] M. De Naurois, L. Rolland, Astrop. Ph., 32, 231(2009).[9] M. Lemoine-Goumard, B. Degrange, M. Tluczykont,Astroph. J. (2006), 25, 195.[10] M. Naumann-God´o, M. Lemoine-Goumard & B.Degrange, Astropa. Phys. (2009), 31, 421.[11] L. Monaco et al., MNRAS (1995) 356(4):1396-1402.[12] G.J. Feldman, R.D. Cousins, Phys. Rev. D (1998)57(7): 3873-3889.[13] L. Bergstr¨om, P. Ullio, J. H. Buckley, Astropa. Phys.(1998), 9, 137.[14] M. Niederste-Ostholt et al., Astroph. J. (2010) 712,516.[15] J. Pe˜narrubia et al., MNRAS (2010) 408, L26.[16] M.L. Mateo, A&A (1998) 36: 435-506.[17] J.F. Navarro, C. S. Frenk, D. M. White, Astroph. J.,(1996) 462, 563.[18] A. Viana et al., Astroph. J., (2012) 77, 746.[19] G. Jungman, M. Kamionkowski, K. Griest, Phys.Rep. 267, 195-373 (1996).[20] U. Ellwanger et al., Phys. Rept., 496, 1-77 (2010).[21] U. Ellwanger and C. Hugonie, Comp. Phys. Com.,175, 290 (2006).[22] J. Cao et al., JHEP 1203, 86, arXiv:1202.5821(2012).[23] S.F. King et al., Nucl.Phys. B870, 323 (2013).[24] D. Das et al., Comp. Phys. Com., 183, 774 (2012).[25] M. Muhlleitner, Comp. Phys. Com., 168, 46 (2005).
2. Study conducted within the French ANR-
DMAstroLHC project by A. Goudelis, P.D. Serpico and G. Lamannaagittarius dwarf spheroidal galaxy observed by H.E.S.S.33 RD I NTERNATIONAL C OSMIC R AY C ONFERENCE , R
IO DE J ANEIRO2013