2009 Fermi Symposium, Washington, D.C., Nov. 2-5 1 Diffuse Gamma-ray Observations of the Orion Molecular Clouds AkiraOkumura∗ DepartmentofPhysics,UniversityofTokyo,7-3-1Hongo,Bunkyo-ku,Tokyo,113-0033,Japan TuneKamae† SLACNationalAcceleratorLaboratory,Stanford,CA94025,USA onbehalfoftheFermi/LATCollaboration Wereportonapreliminaryanalysisofthediffusegamma-rayobservationsoflocalgiantmolecularcloudsOrionA andBwiththeLargeAreaTelescopeonboardtheFermiGamma-raySpaceTelescope. Thegamma-rayemission 0 ofthecloudsiswellexplainedbyhadronicandelectromagneticinteractionsbetweencosmicraysandnucleiinthe 1 clouds. Consequently,weobtainthetotalmassesoftheOrionAandBcloudstobe(80.6±7.5±4.8)×103M⊙ 0 and (39.5±5.2±2.6)×103M⊙, respectively, for the distance to the clouds of 400 pc and the Galactic CR 2 spectrum predicted by GALPROP on the local observations of CRs. The structure of molecular clouds has been extensively studied by radio telescopes, especially using the line intensity of CO molecules (WCO) and n a constant conversion factor from WCO to N(H2) (≡ XCO). However, this factor is found to be significantly a differentforOrionAandB:1.76±0.04±0.02and1.27±0.06±0.01,respectively. J 5 1 1. INTRODUCTION because they located about only ∼400 pc away from the Sun, their total mass is of the order of 105M , ] ⊙ E Diffuse emission of > 100 MeV gamma rays in the and their Galactic coordinates (ℓ ≃ 210◦,b ≃ −15◦) H Galaxyismainlyinducedbyhadronicinteractionsbe- are far from the Galactic plane and the Galactic cen- . tweenthe Galactic cosmic rays(CRs) andinterstellar ter. h medium(ISM),viathe productionofπ0 particlesand p In the study of MCs, the biggest difficulty is that their subsequent decay into photons. This emission - H , the main component of MCs, cannot be ob- o can be used to study CRs and the structure of ISM, 2 served directly. Hence, CO molecules which are the r because its emission can be written as the product t second abundant molecules, have been widely used s of the CR flux and the density of ISM, and because a as a tracer of H2 distribution with a conventional [ thegamma-rayspectralshapepreservestheCRprop- factor X ≡ N(H )/W which converts veloc- erties. Observations of the emission have mainly two CO 2 CO ity integrated CO line intensity, W to H col- 2 uniqueadvantages. First,wecanstudyCRsatthelo- umn density, N(H ). For example, CXO = (21.8 ± v 2 CO 0 cation of the gamma-ray emission, while direct mea- 0.3)×1020 cm−2(K km s−1)−1 was derived by com- surements of CRs at the Earth cannot extract their 6 paring a CO survey with HI and dust observations 8 positional information which is already lost during over a Galactic scale [9]. This factor has been also 3 propagation in the interstellar magnetic fields of the determined by diffuse gamma-ray observations, e.g. 2. Galaxy. Second, π0 emission is not affected by the (2.3 ± 0.3)×1020 cm−2(K km s−1)−1 by COS-B[10], 1 gas condition, such as the temperature of interstellar and (1.35 ± 0.15)×1020 cm−2(K km s−1)−1 for the 9 dust, and the CO to H2 ratio which are often used to Orion region by EGRET [6]. However, due to the 0 estimate the column density of ISM. limitedangularresolutionandphotonstatisticsofthe : Since the early stage of gamma-ray astronomy, v precedingtelescopes,theycouldnotstudyX forin- CO i diffuse emission from HI gas and molecular clouds dividual clouds. Besides, a mystery of “GeV excess” X (MCs) has been extensively studied to understand whichshowsadisagreementbetweentheobserveddif- r GalacticCRs andthe ISM(e.g. [1,2]). Among them, a fuse gamma-ray spectrum and the local CR spectra, the Orion A and the Orion B MCs are two of the was reported by the EGRET [2]. best targets, and have been well studied by pioneer- ing gamma-ray telescopes [3, 4, 5, 6]. This is because Since the EGRET era, much progress has been made in gamma-ray observations and studies on the theyareconsideredtobethearchetypesofgiantMCs located near the Earth, and have been surveyed by Orion A/B clouds. The most important progress is the launch of the Large Area Telescope (LAT) on- radio telescopes especially using the line emission of CO molecules (e.g. [7, 8]). From the analysis side, boardtheFermiGamma-raySpaceTelescope (Fermi) [11]. Its large effective area (∼ 1 m2) and wide en- theirgamma-rayfluxesarestrongenoughtobedistin- guished from other diffuse emission components (HI ergy band (20 MeV to > 300 GeV) make it possible gas,inverseCompton, extragalacticdiffuse emission), to study π0 gamma rays with large photon statistics above1GeV. Inaddition,theenergy-dependentLAT angular resolution is a few times better than that of EGRET.As a result,we areable to resolvethe struc- ∗Electronicaddress: [email protected] ture of the Orion clouds on a scale of ∼ 1 ◦ using †Electronicaddress: [email protected] higher-energy photons of better angular resolutions, eConf C091122 2 2009 Fermi Symposium, Washington, D.C., Nov. 2-5 mental response function (IRF) using ScienceTools b = 0° 5500 xelxel v9r15p4. pipi 4455 nts/nts/ uu oo 4400 cc b = -10° Mono. R2 3355 3. ANALYSIS 3300 3.1. EXTRACTION OF GAMMA-RAY 2255 b = -20° Orion B EMISSION FROM ORION A AND B 2200 Orion A 1155 To study the gamma-ray emission associated with the Orion MCs (molecular gas), other emission com- 1100 b = -30° ponents must be subtracted from Fig. 1. The emis- °032 = l °022 =l ° l= 210 l =° 200 l = 1°90 0505 sdioomniinnatnhteornegeisonarceondsiiffstusseofπse0vegraamlmcoam-rpaoyneanntds.eTlehce- tronbremsstrahlungemissioninduced byinteractions Figure 1: Gamma-ray (200 MeV to 20 GeV) count map between CRs and HI/H2 gas. In addition, inverse of the Orion region binned in 0.25◦×0.25◦ pixels with Compton(IC)scatteringofCRelectronsoffinterstel- Hammer-Aitoff projection. The coordinate centerand lar radiation fields exits in the region. Extragalac- theradius of theregion are (ℓ,b)=(211◦,−17◦) and 20◦, tic diffuse emission, and residual instrumental back- respectively. In addition to theGalactic plane, three ground are classified as diffuse emission. We refer to molecular clouds, Orion A, Orion B and Monoceros R2 the latter two components as “isotropic component”. are visible as bright extended sources. TheLATdetectionofsomepointsourcesintheregion hasbeen reported. However,there is no intense point source overlapping the Orion MCs [19]. andalsoabletocomparethegamma-rayenergyspec- In order to extract the Orion A and B, we mod- trum with the predicted one based on the CR spec- eledotheremissioncomponentsusingacomputerpro- tra observed at the Earth. π0 gamma-ray emissiv- gramcalledGALPROPwhichcalculatesCRflux and ity of hadronic interactions was modeled in several diffuse gamma-ray emission in the Galaxy [20, 21]. articles using recent results of accelerators and the- It uses the data of HI and CO surveys to calcu- oretical studies (e.g. [12, 13, 14]), so that the mass late the target mass of CR interaction [9, 22]1. Its of the clouds can be calculated backwards from the calculation is known to be consistent with the LAT gamma-ray emission and the distance to the clouds gamma-rayobservations in a galactic scale except for which was measured accurately from recent MASER the Galactic plane, when reasonable input parame- observations of the Orion nebula in the Orion A ters basedon many observations andexperiments are [15, 16, 17, 18]. given [23, 24, 25]. Each set of input parameters is referred to as a GALDEF file. In this analysis, we use GALDEF 54 77Xvarh7Sin which CR proton and 2. OBSERVATIONS AND DATA electron fluxes are scaled by factors of 1.15 and 1.75, respectively, to reproduce the LAT observations. The data used in this analysiswere obtained in the BeforewesubtractedtheHIcontributioninthere- nominal all-sky survey mode of the LAT between 4 gion, its normalization was estimated in nearby two August2008and15August2009. AmongalltheLAT 10◦ radius regions near the Orion MCs where no events,weselectedonesclassifiedasP6 V3 DIFFUSE known large MCs exists. The regions are centered class [11]. The reconstructedenergyrangeand zenith at(ℓ,b)=(230◦,−16◦) and(210◦,−32◦), and the ob- anglewerelimitedto200MeV–20GeV,and<105◦, tained fluxes are 1.15 and 0.98 times those predicted respectively. Gamma rays in a circular region of ra- by GALPROP, respectively. Thus we multiply 1.07 dius 20◦ centered at (ℓ,b) = (211◦,−17◦) were then totheGALPROPpredictionforgamma-rayintensity selected for later analyses. from HI gas, and add 8% to the systematic uncer- All events in the selected data are binned in 160× taintyofHIsubtractionprocess. ThescaledHIemis- 160 equalarea pixels with the Hammer-Aitoff projec- sionwasconvolvedwiththeLATIRF,andsubtracted tionasshowninFig.1. TheOrionAandBareeasily from the pixelized data in each energy bin. In addi- recognizedbytheircountexcessfromthesurrounding tion to this, IC emission was also subtracted. Since diffuse emission and point sources. Thereconstructedenergiesoftheeventsarebinned in a logarithmic series of 16 between 200 MeV and 20 GeV. The LAT exposure of each energy bin was 1Weassumed aconstant spintemperature of125 K forthe calculated from the LAT pointing history and instru- HImap. eConf C091122 2009 Fermi Symposium, Washington, D.C., Nov. 2-5 3 ×10-6 ×10-6 0° 90 -2-1srs 0° 90 -2-1srs 80 -2m 80 -2m c c s s n n 70 o 70 o -10° hot -10° hot 60 p 60 p de 50 de 50 u u Latit -20° 40 Latit -20° 40 30 30 20 20 -30° -30° 10 10 °032 °022 °210 °200 1°90 0 °032 °022 °210 °200 1°90 0 Longitude Longitude (a)Extractedgamma-rayintensity (b)ModeledWCO-basedintensity Figure 2: (a) Same as Fig. 1, but shown after subtracting theother diffuse components. The unit is integrated photon intensity between 200 MeV and 20 GeV. Solid lines show thedefinitions of the regions of Orion A and B. Rectangles with dashed lines show theregions which are used to estimate theisotropic component. A straight dot-dashed line at ℓ=212◦ is theboundaryof two separated regions in Orion A.(b)A modeled gamma-ray intensity map simulated from a constant XCO =1.5×1020 cm−2(K km s−1)−1, WCO map [9], theLATresponse, theemissivity of π0 gamma and electron bremsstrahlung, and theCR fluxpredicted byGALPROP.(These figures are still PRELIMINARY) the ICcontributionissmallerthanotheremission,its sumed,whichare∼8%smallerthantheobservedCR uncertainty does not affect our results. flux at the Earth [26, 27, 28]. Gamma-ray inclusive Finally, the isotropic component was estimated cross sections for proton-He, alpha-H, alpha-He, and from surrounding regions (background regions) heavier CR metals are scaled using the method de- around the Orion MCs, then subtracted from the scribed in [14, 29]. A factor 1.02 as a contribution region-of-interest. Fig. 2a shows the gamma-ray in- from ISM metal was finally multiplied to the calcu- tensity map after the subtraction method described lated π0 emissivity [13]. The bremsstrahlung spec- above . It was divided by the LAT exposure, and re- trum was calculated using GALPROP. binned in 1◦×1◦ pixels. We defined the boundary of Theobtainedfitresultsshowgoodagreementswith Orion A and B with solid lines in Fig. 2a. the π0 spectrum. This means that gamma-ray emis- sion from MCs is dominated by known physical pro- cesses, i.e. interactions between CRs and the gas. 3.2. ENERGY SPECTRA AND TOTAL Therefore, we are able to calculate the total mass of MASSES OF THE CLOUDS theclouds. AssumingthedistancetotheOrionclouds to be 400 pc [15, 16, 17, 18], the mass of the OrionA The energy spectra associated with the Orion A andBareestimatedtobe(80.6±7.5(stat)±4.8(HI))× and B clouds are shown in Fig. 3. All photons in 103M⊙ and (39.5±5.2(stat)±2.6(HI))×103M⊙, re- the boundary regions in Fig. 2a are collected for each spectively. The second terms indicate statistical er- cloud. They are fitted with π0 and bremsstrahlung rors of the fit, and the third ones are systematic er- components with two free normalization parameters. rors introduced by the uncertainty of HI subtraction Here, we calculated π0 gamma-ray emissivity using a explained in subsection 3.1. recent parameterized model [12]. Input CR spectra at the Orion region2 predicted by GALPROP are as- 3.3. CORRELATION BETWEEN GAMMA-RAY AND CO INTENSITY 2Corresponding to a cylindrical Galactic location (R = Assumption of a constant XCO in small scales (∼ 8.5kpc,Z=−0.14kpc). 10 pc to ∼ 100 pc) or Galactic scale has been widely eConf C091122 4 2009 Fermi Symposium, Washington, D.C., Nov. 2-5 Pionic Orion A Total (80.6 ± 7.5 ± 4.8) × 103 M Electron Bremsstrahlung 400 χ2/ndf = 16.7/10 10-4 Orion A Total 1) -V (39.5 ± 5.2 ± 2.6) × 103 M 400 Me χ2/ndf = 9.3/10 1 -s 2 -m c 2 V e M ( 10-5 E d / N d 2 E 10-6 102 103 104 Energy (MeV) Figure 3: Energy spectra of theOrion A (black) and B (red) fitted with modeled spectra of π0 gamma (dashed) and electron bremsstrahlung(dotted). Statistical errorsareshown withbars,and thesystematic errorsoftheLATresponse are shown with shaded area. Polygons of solid lines show the systematic uncertainties of HI subtraction process. (This figureis still PRELIMINARY) used in studies of MCs (e.g. [6, 9]), while gradient of 0.04(stat)±0.02(HI))×1020 and (1.27±0.06(stat)± XCO isalsodiscussed[30,31]. However,thereisroom 0.01(HI))×1020, respectively. toreconsiderthissimpleassumptionthatW canbe CO used to trace the structure of MCs [32, 33]. Utilizingthegoodangularresolutionandlargepho- 4. DISCUSSION ton statistics of the LAT, the correlation between gamma-ray intensity and a W map can be stud- CO Weobtainedtheenergyspectraofthe OrionAand ied in a scale of ∼ 1◦. Fig. 2b shows a mod- B, and showed that they can be explained by the CR eled gamma-ray intensity map based on a CO survey interactions with the nuclei in the gas. Thus, a lin- and GALPROP calculation with a constant X of CO ear correlation between gamma-ray intensity and the 1.5 ×1020 cm−2(K km s−1)−1, where the LAT IRF is column density of the clouds are expected, because convolved. gamma-ray emission is not affected by environmen- Fig. 4 shows the pixel-by-pixel correlation between tal condition of the gas. In fact, the correlation be- theobservedgamma-rayintensity(Fig.2a;x)andthe tween gamma-rayintensity and the W -based mod- CO WCO model (Fig. 2b; y). The best fit results by a lin- eledmapholdsalinearityforroughlyonedecadewith ear function (y = p0 +p1x) are also shown. If the only a few 10% deviations. This implies that cosmic gamma-rayintensity canbe presentedby the product rays of energy above ∼ 1 GeV can penetrate dense ofa constantCR flux anda constantXCO, the slopes cores of molecular clouds. However, the correlation of the best-fit functions become constant. However, slopes in the Orion region was found to be not con- wefoundthattheobtainedslopesaresignificantlydif- stant as shown in Fig. 4. ferentfortheOrionAandBclouds,whiletheseclouds There are some possible interpretations of the dif- arethoughttobeinacommonenvironmentsincetheir ferent correlationslopes. In the analysis, we assumed birth. In Fig. 4a, two additional linear functions are aconstantCRfluxinthe Orionregion. However,ifit shown: one is fitted with the larger longitude region issignificantlydifferentinthe threeseparatedregions inOrionA(ℓ>212◦),andtheotherisforthesmaller inFig.2a,thegamma-rayintensityalsovariesaccord- longitude region (212◦ >ℓ) (see Fig. 2a). ing to the CR flux variation which might be caused Since we modeled the vertical values using a con- by the strong magnetic field in molecular clouds. stant X of 1.5 ×1020 cm−2(K km s−1)−1, the cor- On the contrary, if the CR flux is almost constant CO responding X of Orion A and B are (1.76 ± in the region, we need to consider nonuniformity of CO eConf C091122 2009 Fermi Symposium, Washington, D.C., Nov. 2-5 5 (a) (b) 70 70 χ2 / ndf 257 / 74 χχ22 // nnddff 6699..99 // 2288 60 60 p0 1.2 ± 0.313 pp00 --00..779966 ±± 11..0022 -1-1-2)srsm50 p1 0.85 ± 0.0186 -1-1-2)srsm50 pp11 11..1188 ±± 00..00551166 c40 c40 -6 0 -6 0 1 1 e (30 e (30 ns ns o o esp20 esp20 R R × × co 10 co 10 W W 0 0 -10 -10 -10 0 10 20 30 40 50 60 70 -10 0 10 20 30 40 50 60 70 Observed (10-6 cm-2s-1sr-1) Observed (10-6 cm-2s-1sr-1) Figure 4: Correlations between themeasured gamma-ray intensity (x) and a model prediction byWCO (y)assuming XCO =1.5×1020 cm−2(K km s−1)−1. Data points are fitted with a linear function y=p0+p1x. (a) Orion A. Circles correspond to thedata in pixels of ℓ>212◦. (b) Orion B. (This figures are still PRELIMINARY) X in the region. While the CO line (J = 1−0) Acknowledgments CO is the de fact standard of mass tracers of molecular clouds, the CO to H2 ratio can be varied by the con- The Fermi LAT Collaboration acknowledges gen- dition of each cloud. By comparing WCO maps with erous ongoing support from a number of agencies dustobservationsorgamma-rayobservations[32,33], and institutes that have supported both the develop- itisknownthatthereexistsgaswhichisnottracedby ment and the operation of the LAT as well as scien- CO or HI observations. The nonuniformity of XCO tific data analysis. These include the National Aero- in the Orion region can also be explained by this. In nautics and Space Administration and the Depart- fact, IR emissionfrom interstellardust andvisual ex- ment of Energy in the United States, the Commis- tinction by dust are stronger in the left half of Orion sariat `a l’Energie Atomique and the Centre National A than that expected from CO observations [34, 35]. de la Recherche Scientifique / Institut National de Therefore,if there exist H2 molecules not fully traced Physique Nucl´eaire et de Physique des Particules in byWCO,buttracedbygamma-rayanddust,theXCO France, the Agenzia Spaziale Italiana and the Isti- variation shown in Fig. 4 can be explained. In addi- tutoNazionalediFisicaNucleareinItaly,theMinistry tiontoH2,itispossiblethatapartofHIgaswasnot of Education, Culture, Sports, Science and Technol- subtracted adequately, because its spin temperature ogy (MEXT), High Energy Accelerator Research Or- is relatively low in cold molecular clouds compared ganization (KEK) and Japan Aerospace Exploration to surrounding diffuse HI region. While we assumed Agency (JAXA) in Japan, and the K. A. Wallenberg a constant spin temperature of 125 K in the region, Foundation, the Swedish Research Council and the optically thick HI gas whose temperature is low may Swedish National Space Board in Sweden. not have been corrected properly and contribute to Additional support for science analysis during the the gamma-rayemission, especially in the left part of operations phase is gratefully acknowledged from the Orion A. Istituto NazionalediAstrofisicainItaly andthe Cen- tre National d’E´tudes Spatiales in France. While the gamma-ray emission from HI and H2 cannotbe distinguishedfromeachother,ourobserva- tionsrevealednon-linearrelationbetweengamma-ray and W maps, which will enable us to understand CO thenatureofmolecularcloudsindetail,inadditionto References the Galactic CRs. More photon statistics in future is expected to show the smaller structure of the clouds. [1] W. L. Kraushaar, G. W. Clark, G. P. Garmire, et al., Astrophys. Jour. 177, 341 (1972). [2] S. D. Hunter, D. L. Bertsch, J. R. Catelli, et al., Astrophys. Jour. 481, 205 (1997). [3] P. A. Caraveo, K. Bennett, G. F. Bignami, et al., Astron. Astrophys. 91, L3 (1980). eConf C091122 6 2009 Fermi Symposium, Washington, D.C., Nov. 2-5 [4] J.B.G.M.Bloemen,P.A.Caraveo,W.Hermsen, [22] P.M. W.Kalberla,W. B.Burton,D. Hartmann, et al., Astron. Astrophys. 139, 37 (1984). et al., Astron. Astrophys. 440, 775 (2005). [5] S. W. Digel, S. D. Hunter, and R. Mukherjee, As- [23] A. A. Abdo, M. Ackermann, M. Ajello, et al. trophys. Jour. 441, 270 (1995). (Fermi LAT Collaboration), Physical Review Let- [6] S.W. Digel, E.Aprile, S.D. Hunter, et al.,Astro- ters 103, 251101 (pages 6) (2009). phys. Jour. 520, 196 (1999). [24] A. W. Strong, Proc. 31st Int. Cosmic Ray Conf. [7] R. J. Maddalena, M. Morris, J. Moscowitz, et al., (2009). Astrophys. Jour. 303, 375 (1986). [25] A. A. Abdo, M. Ackermann, M. Ajello, et al., [8] B. A. Wilson, T. M. Dame, M. R. W. Masheder, Astrophys. Jour. 703, 1249 (2009). et al., Astron. Astrophys. 430, 523 (2005). [26] T. Sanuki, M. Motoki, H. Matsumoto, et al., As- [9] T. M. Dame, D. Hartmann, andP. Thaddeus,As- trophys. Jour. 545, 1135 (2000). trophys. Jour. 547, 792 (2001). [27] J. Alcaraz, D. Alvisi, B. Alpat, et al., [10] A.W.Strong,J.B.G.M.Bloemen,T.M.Dame, Phys. Lett. B 472, 215 (2000). et al., Astron. Astrophys. 207, 1 (1988). [28] J. Alcaraz, B. Alpat, G. Ambrosi, et al., [11] W. B. Atwood, A. A. Abdo, M. Ackermann, Phys. Lett. B 494, 193 (2000). et al., Astrophys. Jour. 697, 1071 (2009). [29] M. Mori, Astrophys. Jour. 478, 225 (1997). [12] T.Kamae,N.Karlsson,T.Mizuno,et al.,Astro- [30] N. Arimoto, Y. Sofue, and T. Tsujimoto, phys. Jour. 647, 692 (2006). Pub. Astron. Soc. Japan 48, 275 (1996). [13] M. Mori, Astropar. Phys. 31, 341 (2009). [31] A.W.Strong,I.V.Moskalenko,O.Reimer,etal., [14] T. K. Gaisser and R. K. Schaefer, Astro- Astron. Astrophys. 422, L47 (2004). phys. Jour. 394, 174 (1992). [32] F. Boulanger, L. Bronfman, T. M. Dame, et al., [15] T. Hirota, T. Bushimata, Y. K. Choi, et al., Astron. Astrophys. 332, 273 (1998). Pub. Astron. Soc. Japan 59, 897 (2007). [33] I. A. Greiner, J.-M. Casandjian, and R. Terrier, [16] K. M. Sandstrom, J. E. G. Peek, G. C. Bower, Science 307, 1292 (2005). et al., Astrophys. Jour. 667, 1161 (2007). [34] D. J. Schlegel, D. P. Finkbeiner, and M. Davis, [17] K. M. Menten, M. J. Reid, J. Forbrich, et al., Astrophys. Jour. 500, 525 (1998). Astron. Astrophys. 474, 515 (2007). [35] K. Dobashi, H. Uehara, R. Kandori, et al., [18] M.K.Kim,T.Hirota,M.Honma,etal.,Pub.As- Pub. Astron. Soc. Japan 57, S1 (2005). tron. Soc. Japan 60, 991 (2008). [19] A. A. Abdo, M. Ackermann, M. Ajello, et al., Astrophys. Jour. Suppl. Ser. 183, 46 (2009). [20] A. W. Strong and I. V. Moskalenko, Astro- References phys. Jour. 509, 212 (1998). [21] A. W. Strong, I. V. Moskalenko, and O. Reimer, Astrophys. Jour. 537, 763 (2000). [1] eConf C091122