INTEGRAL SPI All-Sky View in Soft Gamma Rays: Study of Point Source and Galactic Diffuse Emissions
L. Bouchet, E.Jourdain, J. P.Roques, A. Strong, R. Diehl, F. Lebrun, R. Terrier
aa r X i v : . [ a s t r o - ph ] J a n INTEGRAL
SPI all-sky view in soft γ -rays: Study of pointsource and Galactic diffuse emissions L. Bouchet, E.Jourdain, J.P.Roques
CESR–CNRS, 9 Av. du Colonel Roche, 31028 Toulouse Cedex 04, France
A. Strong, R. Diehl
Max-Planck-Institut f¨ur extraterrestrische Physik, Postfach 1603, 85740 Garching,Germany andF. Lebrun, R. Terrier
DSM/DAPNIA/SAp, CEA-Saclay, 91191 Gif-sur-Yvette, FranceAPC, UMR 7164, 10 rue A. Domon et L. Duquet, 75205 Paris Cedex 13, FranceReceived ; accepted
ABSTRACT
We have processed the data accumulated with
INTEGRAL
SPI during 4 years( ∼
51 Ms) to study the Galactic “diffuse” emission morphology in the 20 keV to8 MeV energy range. To achieve this objective, we have derived simultaneouslyan all-sky census of emitting sources and images of the Galactic Ridge (GR)emission. In the central radian, the resolved point source emission amounts to88%, 91% and 68% of the total emission in the 25-50, 50-100 and 100-300 keVdomains respectively. We have compared the GR emission spatial distributionto those obtained from CO and NIR maps, and quantified our results throughlatitude and longitude profiles. Below 50 keV, the SPI data are better traced bythe latter, supporting a stellar origin for this emission. Furthermore, we foundthat the GR emission spectrum follows a power law with a photon index ∼ ∼ ◦ by ∼ ◦ potentiallyassociated with the disk or halo surrounding the central regions of our Galaxy. 2 – Subject headings:
Galaxy: general— Galaxy: structure — gamma rays: obser-vations — surveys — (ISM): cosmic rays — ISM:general
1. Introduction
The soft γ -ray GR Emission ( >
20 keV) has been previously studied essentially withthe CGRO and GRANAT missions (Purcell et al., 1996, Skibo et al., 1997, Kinzer, Purcell& Kurfess, 1999). The main conclusion was that point sources explain at least 50 % ofthe total emission up to ∼
200 keV, allowing the possibility to anticipate that unresolvedsources could explain a major part of what was seen as “diffuse continuum emission”. Soonafter its launch, first INTEGRAL results indicated that the GR diffuse emission is less than15 % of the total in the 20-200 keV domain (Lebrun et al., 2004, Terrier et al., 2004, Stronget al., 2005, Bouchet et al., 2005). In more recent works, Revnivtsev et al. (2006), withRXTE PCA data, and Krivonos et al. (2007), using the data collected with the
INTEGRAL
IBIS telescope, suggest that the GRXE (Galactic Ridge X-ray Emission) between 3 and ∼
60 keV is explained in terms of sources belonging to the population of accreting white dwarfsbinaries.In a previous paper using one year of SPI data (Bouchet et al., 2005, hereafter paper I),we had used Galactic tracer morphologies to extract the spectrum of the diffuse continuumemission as well as that of the positronium/annihilation in the central radian of our Galaxy.We thus derived the relative contributions of these emissions along with that of the pointsource emission to the total Galactic emission. In this paper, we present a global view of thesoft γ -ray sky emission based on a larger amount of data covering the whole sky and performan imaging analysis of the diffuse component, with an estimate of its spectrum between 20keV and 8 MeV.
2. Instrument and Observations
The ESA’s INTEGRAL (INTErnational Gamma-Ray Astrophysics Laboratory) obser-vatory was launched from Baikonour, Kazakhstan, on 2002 October 17. The spectrometerSPI (Vedrenne et al., 2003) observes the sky in the 20 keV - 8 MeV energy range with an Based on observations with INTEGRAL, an ESA project with instruments and science data centrefunded by ESA member states (especially the PI countries: Denmark, France, Germany, Italy, Spain, andSwitzerland), Czech Republic and Poland with participation of Russia and USA. with a thicknessof 7 cm. In addition to its spectroscopic capability, SPI can image the sky with a spatialresolution of 2 . ◦ (FWHM ) over a field of view of 30 ◦ , thanks to a coded mask located 1.7 mabove the detection plane. Despite such a modest angular resolution, it is possible to locateintense sources with an accuracy of few arc minutes (Dubath et al., 2005). The assemblyis surrounded by a 5-cm thick BGO shield which stops and measures the flux of particlesarriving from outside the FoV. The instrument in-fligh performance is given in Roques et al.(2003).Due to the small number of detectors, SPI imaging capability relies on a specific observa-tional strategy, based on a dithering procedure (Jensen et al., 2003): the pointing directionvaries around the target by steps of 2 ◦ within a 5 × ∼
85% of the time outside the radiation belts.We have analysed observations recorded from 2002, February to 2006 May, covering theentire sky (Fig. 1). The central region of the Galaxy, within − ◦ ≤ l ≤ ◦ , − ◦ ≤ b ≤ ◦ , is thus particularly well scanned.Data polluted by solar flares and radiation belt entries are excluded. After image anal-ysis and cleaning, we obtained 51 × seconds of effective observing time.
3. Technical considerations and algorithm descriptions
The analysis method used in this paper is similar to the one described in our ’paper I’,with some significant enhancements however. We first summarize the analysis method basics,consisting of an iterative procedure between source position determination (with SPIROS)and flux extraction (with Time-Model-Fit) steps.- Source catalogues are iteratively produced using the SPIROS software (Skinner &Connell, 2003), in several energy bands.An iterative search is performed for point sources, starting from the strongest one identified,and successively adding weaker ones in successive iterations. Each time a source is found ina search of trial positions on the sky, it is added to a working source catalogue, and thenused as baseline knowledge for the next source search in the following iteration.In order to avoid unnecessary artifacts from spatial-resolution limits of our instrument, foreach newly-found source an identification step is introduced, and, for sufficient proximity,the (more accurate) position of the identified source is used instead. 4 –In crowded regions, due to the 2.6 ◦ SPI angular resolution, the brightest sources are consid-ered as representative of the local emissions. For the central degree of the Galaxy good fitshave been obtained in paper I, considering 1E 1740-2942.7 only. The better statistic of ournew set of data requires a more complex description of this region. Thus two other (softer)sources have been considered below 200 keV, tentatively identified with SLX 1744-299 andIGR J17475-2822.- Using the SPIROS output catalogues, source fluxes and diffuse emission intensitydistribution are derived with Time-Model-Fit (see Paper I) algorithm.Time-model-fit algorithm is a model fitting procedure based on the likelihood/ χ statistics.We thus model the sky successively with a set of three models for the large-scale diffuseand extended emission plus the sources accumulated in the search, allowing for intensityvariations down to the time scale of individual pointings ( ∼ • The matrix associated with the imaging transfer function is sparse with only a fewpercents of non-zero elements. We take advantage of this property by minimizing thestorage and speeding-up the numerous matrix-vector operations through a CompressedColumn Storage scheme (Duff et al. 1989). In addition, the transfer function related toa variable source exhibits a peculiar structure which allows a direct inversion of a partof the matrix. Thanks to these optimisations the algorithm is able to handle manymore observations together with more free parameters e.g. for source variability. • The background determination has been refined: We decompose background variabilityto global intensity variation and into the pattern of rates among the 19 detectors of ourcamera. We retain freedom of variability for global background intensity as describedabove, and allow the general background count rate pattern in the SPI camera to varyon a much more restrictive time scale (on the order of months). We find that over 6months of data, no pattern variations are noticeable. With this compromise of allowing 5 –high variability for intensity parameters of model components, and modest variabilityallowance for the background pattern, we obtain adequate fits to our measurementswith a minimum number of parameters. • The total energy redistribution matrix is included for the final flux extraction, withthe proper spectral shape taken into account for each source and diffuse component. • The algorithm is able to handle a sky described by cells of various sizes whose intensitiesare determined together with the other parameters. The use of ”large” cells is suitableto gather flux from diffuse emissions in a model independent way.This hybrid reconstruction algorithm takes the best of methods to map point source andextended emissions simultaneously . It is the basis for the results presented in this paper.
4. Mapping the sky
The GR emission is difficult to measure since its surface brightness and its signal-to-noise ratio are low. Moreover, as the distribution of a population of weak unresolved pointsources formally mimics an extended structure, the goal of any diffuse emission study con-sists in estimating always better the source component in order to derive upper limits forthe GR emission.In paper I, the small amount of data, covering only the central radian of the Galaxy, justifiedthe global model-fitting approach to estimate the different contributions : The GR Emissionwas tentatively described with a CO map while the annihilation process (511 keV line andpositronium continuum) was modeled by an azimuthally symmetric Gaussian of 8 ◦ FWHM(following previous works by Kn¨odlseder et al., 2005).In this paper, we attempt to estimate a model-independent GR emission morphology, by ex-tracting local fluxes without spatial model. For this, the sky will be represented by ”large”cells of fixed sizes (tested values range from 2 ◦ to 16 ◦ ) whose intensities are fitted to the datathrough the likelihood maximisation method, together with the ponctual sources ones.This represents the best trade off between a priori (model dependent) information intro-duced in the model-fitting algorithm (positions of a catalog of known sources plus variabilitytimescale of each source) and model independent results (fluxes of sources and of cells of theextended emission).The first step consists thus to build a catalog of individual sources as complete as possi-ble, following the procedure used in Paper I. Then, large cells are added in the convergenceprocess to obtain the diffuse emission distribution. 6 – We used the method described in paper I to generate source catalogues: the SPIROSsoftware delivered in the INTEGRAL OSA (Off-line Scientific Analysis) package is usediteratively in conjunction with our hybrid algorithm. The latter calculates variable sourceflux contributions using Galactic tracers models for spatial morphologies of the interstellaremissions (8 ◦ axisymmetric Gaussian for the annihilation emission, DIRBE 4.9 µ and COmaps for the GR continuum below and above 120 keV respectively),for which a normalisationfactor is adjusted during the fitting process. The extended emission and variable point sourcecontributions are then removed from the data set provided to SPIROS for the next iteration.It should be noted that the point source fluxes remain the same (within error bars) whateverthe GR continuum distribution model is used (CO or DIRBE), as expected given the lowsurface brightness of this component.As explained in paper I, to minimize the error bars (maximize the signal-to-noise ratios),we need to describe the sky with a minimum number of free parameters. We thus define asky model for each of the energy bands. In practice, 12 sources (marked with ’*’ in tableI) have been considered as variable for the 25-50 keV band. Above 50 keV, the reduced χ being sufficiently close to 1, only Cyg X-1 is set variable up to 200 keV.Figure 2 displays the resulting sky images sum of point sources and “diffuse” componentsin different energy bands and illustrates their evolution. At low energy, the sky emission isdominated by sources while a ”diffuse/extended” structure appears above 200-300 keV, ina domain corresponding to the annihilation radiation. We will come back to it in the nextsection, dedicated to the diffuse emissions. Above 511 keV, sources again dominate the skyemission.The resulting catalogues contain 173, 79, 30 and 12 sources detected above ∼ . σ inthe 25-50 keV and the 50-100 keV bands and above ∼ . σ in the 100-200 keV and 200-600keV ones (see Table I). All of them except one are associated (within 1 ◦ ) with at least oneIBIS source (Bird et al., 2006).Above 600 keV only the Crab Nebula, Cyg X-1, GRS1915+115 and GRS1758-258 aredetected, the 2 former being still emitting above 2 MeV.Note that the reported fluxes are four years averaged values, and that this analysis is notoptimised for any particular source, since all data are adjusted simultaneously. An individualsource analysis should be based on a restricted number of exposures, selected on the basisof the pointing direction (typically less than ∼ ◦ relatively to the source direction), andrequires a detailed study of the source time evolution. 7 – Once the contribution of the individual sources has been independently estimated, wecan study the unresolved component. The a priori information will thus now be introducedin the source terms. To determine the spatial distribution of the Galactic Ridge emission,we have considered the region | l | < ◦ , | b | < ◦ and divided it into cells of size δl = 16 ◦ x δb = 2 . ◦ , the 511 keV line case being treated apart. These cell or pixel sizes havebeen chosen a posteriori to optimize the signal-to-noise ratio per cell while being sufficientlysmall to follow the observed spatial variations. We use the a priori information on thesource positions obtained in the previous step, whereas the “diffuse” pixel cells, sources andbackground intensities are to be fitted to the data. The number of unknowns is high butreasonable compared to the data and the problem is easily tractable by a simple likelihoodoptimisation to determine all the corresponding intensities and error bars.Figure 3 displays the images obtained through this method for the “diffuse” compo-nent(s) in different energy bands. The last one contains the Al 1.8 MeV line but it is clearthat its contribution to the large band flux is quite negligible.To quantify more easily the behavior of the diffuse emission, we present our results interms of longitude and latitude profiles (figures 4 and 5). They have been built by integratingthe flux measured for | b | < . ◦ or | l | < ◦ , in longitude and latitude bins respectively. Wecan then compare them to those obtained from CO (Dame et al. 2001) and NIR Dirbe 4.9 µ corrected from reddening (http://lambda.gsfc.nasa.gov) maps.Both models agree grossly with SPI longitude profiles, showing a slowly decreasing in-tensity of the GR emission toward high longitudes.The latitude profiles are not so similar: Up to ∼
100 keV, the GR high energy and NIRemissions seem to be concentrated in a region slightly more extended than the CO one.This is supported by calculating a χ between the resulting sky images and the proposedmodels. We obtained χ values of 74 and 86 (37 dof) in the 25-50 keV energy band, whilein the 50-100 keV energy band, chisquare values are of 50 and 53 for DIRBE and CO maps,respectively. This fits in with the current understanding of a stellar origin for the GR emis-sion in the X ray domain ( <
50 keV) proposed by Krivonos et al. (2007). However, atenergies below 50 keV, systematic effects from instrumental properties and/or variability ofthe sources provides limitations to fit quality.The signal to noise ratio has a mininum in the 600 keV-1.5 MeV domain, prohibiting 8 –any serious analysis but at high energies ( > δl = 5 ◦ x δb = 5 ◦ (fig. 6). In this narrowenergy range, the emission is almost entirely due to the annihilation line, the contribution ofthe other components being negligible. The emission profiles (figures 7 and 8) are peakedtowards the Galactic Centre. This structure, corresponding to the bulge, has been modeledwith an axisymmetric Gaussian of 8.0 ◦ ± . ◦ in good agreement with previous INTEGRAL
SPI results (e. g. Kn¨odlseder et al. (2005)). However, the map together with the profilessuggest that the emission is not limited to that structure but exhibits an additional extentedcomponent revealed by the long exposure.To quantify this result, we attempted to describe the extended spatial distributionsuperimposed to the central bulge, with various models. Simple geomerical shapes (ie two-dimensional Gaussians) have been tested as well as maps obtained in other wavelengthes(CO and DIRBE maps). The best results are obtained with NIR data (more precisely the3.5 to 240 µ maps, too close to be differentiated in these tests) which, independently, happento be good tracers of the Al line emission (Kn¨odlseder et al., 1999).For a sky model consisting of the 240 µ map plus a 8 ◦ axisymmetric Gaussian, we obtainfluxes of 1 . ± . × − photons cm − s − , in the extended structure and 0 . ± . × − photons cm − s − in the central one. Even though it is difficult to describe the emissionin more detail, this represents a good indication for a bulge/disk structure.Coming back to the central bulge emission, we also tested several possibilities as we suspect ageometry more complex than a single axisymetric gaussian. A precise study of its morphologywould require a specific analysis, but first investigations show that the fluxes correspondingto each structure vary within the error bars. The spectral shape of the diffuse continuum remains of prime importance to determineits origin. As this emission is concentrated in the central regions, the spectral analysis hasbeen restricted to the Galactic central radian ( | l | < ◦ and | b | < ◦ ).The “diffuse” continuum spectrum shown in fig 9 has been extracted assuming a NIR 4.9 µ spatial distribution (up to 120 keV) and a CO map (above 120 keV) and fitted with 3 9 –components. • The diffuse spectrum (apart from positronium) is fitted by a power law of index 1 . ± .
25, with a 100 keV flux of 4 . ± . × − photons cm − s − keV − . • The additional component below ∼
50 keV presents a curved shape, which can bemodeled by a power law of index fixed to 0 with an exponential cutoff at 7 . ± . ± . × − photons cm − s − keV − . These parameters differfrom those derived in Paper I, since many more sources have been identified, and thusremoved from ’diffuse” component. This remaining ”diffuse” low energy component,with a central radian luminosity of ∼ × erg.s − , is likely to correspond tothe population of accreting magnetic white dwarfs proposed to provide a dominantcontribution to the Galactic X-ray emission (Krivonos el al., 2007). • The third component is due to the positronium/annihilation emission with its charac-teristic shape. We extract it using a 8 ◦ Gaussian spatial distribution and obtained a511 keV line flux of 8 . ± . × − photons cm − s − . A fit to the data allows us todetermine a positronium flux of 3 . ± . × − photons cm − s − corresponding toa positronium fraction (as defined by Brown and Leventhal, 1987) of F p =0 . ± . • The 511 keV emission component reveals a disk component in addition to the well-known bright bulge. We estimate a bulge-to-disk ratio of ∼ .
5. Sources vs diffuse emission contribution
In Fig 9 we compare the different contributions to the Galactic emission : The total pointsource emission spectrum has been built by adding spectra from all sources detected between | l | < ◦ and | b | < ◦ . It can be roughly described between 20 keV and 1 MeV by a powerlaw with a photon index of 2 .
67 and a flux at 50 keV of 2 . × − photons cm − s − keV − .The diffuse emission is represented together with its 3 components.¿From the analysis presented above, we can deduce ratios of combined emission ofdetected sources to the total emission of 88%, 91% and 68% in the 25-50, 50-100 and 100-300 keV bands respectively. These values represent lower limits as a population of weakunresolved sources may be confused with a diffuse emitting structure. 10 –
6. Discussion
The unresolved/extended emission observed in the soft γ -ray domain comprises threeseparate components: The electron-positron interactions are known to produce a large scaleemission. In another hand, an interstellar emission is expected due to high energy particlestravelling inside the whole galaxy. The third component has been identified more recently,with a contribution exponentially decreasing above 10 keV. We discuss each of them in thefollowing sections. e − /e + interaction Between 300 and 511 keV, the annihilation process plays the major role.We have determined a 511 keV flux of 8.7 ± . × − photons cm − s − and a positroniumfraction F p = 0 . ± .
05. Comparison with previous results is quite satisfying : for example,Kinzer et al. (2001) reported a positronium fraction of 0 . ± .
04 in OSSE data, while inde-pendent SPI data analyses led to F p = 0 . ± .
06 (Churazov et al. 2005), F p = 0 . ± . F p = 0 . ± .
09 (Weidenspointner et al., 2006) and F p = 0 . ± .
022 (Jean etal. 2006). Fluxes values are more difficult to compare as they depend on the assumed spatialdistribution but they range around 1 × − photons cm − s − in the quoted references.Indeed, the 511 keV line emission spatial distribution is essentially concentrated within thecentral region of the Galaxy but its morphology may be more complex than the proposedaxisymetric ( ∼ ◦ ) Gaussian distribution. The regions surrounding this central part containa significant flux ( ∼ . ± . × − photons cm − s − ) which is not compatible with thissimple spatial distribution. As a function of longitude, this emission seems to extend symet-rically up to ∼ − ◦ , while in latitude, it extends ∼ − ◦ . This emission could forexample correspond to the disk component, as already suggested by the OSSE team (Kinzeret al. 1999). Among several models, the NIR emission distributions, otherwise consideredas good Al tracers, reveal themselves to give the best description of the SPI data in thisenergy domain. The mentionned link with the Al emission leads to a straightforward in-terpretation: the Al decay produces 1809 keV photons simultaneously with positrons andthus 511 keV photons in such a way that F lux (511 keV ) = 0 . × (2 − . × f p ) × F lux (1809 keV )where fp is a positronium fraction of the annihilation process.Using the COMPTEL measurement F(1809 keV) = 7 − × − photons cm − s − (Kn¨odlseder,1999) and assuming fp= 0.98, we expect a 511 keV flux of 3 − × − photons cm − s − .We thus conclude that 20-30% of the 511 keV flux emitted in the disk/halo structure couldbe explained by the Al decay. The remaining ∼
75% of the observed disk/halo emission 11 –require another origin and will be refined by adding future observations.Previous ratio estimates of the bulge-to-disk flux ratios have been obtained with OSSE/CGRO,varying from 0.2 to 3.3 depending upon whether the bulge component features a halo (whichleads to a large ratio) or not (Milne et al. 2000, Kinzer et al., 2001). Benefiting from a moreuniform Galactic plane coverage than the OSSE data, the SPI data allow to better constrainthis parameter. Indeed, even though the bulge-to-disk flux ratio depends on the assumedbulge and disk shapes, tests with several representative configurations lead to a bulge-to-diskflux ratio estimation of ∼ <
50 keV component
This low energy component deserves a short discussion: it has been found to follow thesame spatial distribution as the NIR DIRBE 4.9 µ emission (see fig. 5). Together with thesoft spectral shape, this corroborates the interpretation of the Galactic ridge emission be-tween 20 and 60 keV by Krivonos et al. (2007) in terms of a population of accreting magneticwhite dwarfs, which present a spectral cutoff around 30-50 keV (Suleimanov et al, 2005).Our observed luminosity of ∼ × erg s − between 20 and 60 keV in the central radian isin good agreement with the estimation of 1.23 ± × erg s − given by Krivonos et al.(2007) on the basis of the proposed model. However, it is obviously beyond the SPI capacities(and objectives) to resolve this emission since the potential objects which could explain itare too faint, even thought the global emission is clearly detected as a large scale distribution. Above 50 keV, we detected an emission following a power law, which joins the highenergy points previously measured by
CGRO
OSSE (Kinzer et al. 1999) in the MeV re-gion (fig 10). Our photon index of ∼ . ± .
25 is quite compatible with the ∼
7. Conclusions
The ”diffuse emission” generic term gathers a set of various processus depending on theenergy domain, and making the determination of its origin rather difficult. In the soft γ -raydomain, the diffuse emission represents a complex problem, as its presence is swamped withthe individual sources emission and difficult to disentangle.To investigate it, we have developed an imaging algorithm dedicated to extended structures.This code first determines and takes into account the point sources emitting in the sameenergy domain, allowing us to then better estimate the geometry of the regions producingdiffuse emissions.The SPI all-sky survey analysis reveals that 173 sources can be identified in the 25-50 keV 13 –with 30 of them emitting a significant flux ( > ∼ σ ) above 100 keV. It is clear now thatthe sources dominate the Galactic emission up to ∼
300 keV (Fig. 9). Finally, upper limitson the diffuse emission are estimated to be one tenth of the total emission below 100 keVand one third in the 100-300 keV band.The e − /e + annihilation produces an emission which has already been the subject ofmany studies. The main point emerging from our analysis concerns the spatial distributionof this emission since a large structure, potentially associated with the Galactic disk/halohas been found to contain a significant fraction of the 511 keV line total flux. While a deeperanalysis is required to refine this result, it is clear that the ∼ ◦ Gaussian distribution previ-oulsy reported does not represent the complexity of the 511 keV line emission morphology.Once the source and annihilation process contributions have been taken into account, wecan access to the ”diffuse” Galactic emission, whose origin remains problematic.Its morphology is of prime importance as it is thought to trace the CR electron population,observed through bremsstrahlung or Compton emission. The
INTEGRAL
SPI data allowus to investigate this particular topic, and we present for the first time images of the diffuseGalactic emission up to a few MeV.This diffuse emission is clearly detected between 50 keV and 2 MeV with a power law rela-tively hard (photon index around 1.55), and an additional component, much steeper, requiredbelow 50 keV.Toward high energy, the SPI spectrum joins with the OSSE and EGRET measurements. Alltogether, they can be compared to various models to understand the nature of this emis-sion. The inverse Compton interaction is a good candidate and will be investigated in aforthcoming paper. Other interstellar processes have been invoked to explain the emissionbelow 1 MeV (Dogiel et al., 2002; Masai et al., 2002; Schr¨oder et al., 2005). In another side,unresolved point sources are the most probable origin below 50 keV (Krivonos et al., 2007)and are also a possibility at higher energies as AXPs or other pulsars (see for exemple Kuiperet al. 2004; Strong, 2007). Observations of such sources up to a few MeV would allow toquantify their potential contribution and bring a new piece to the Galactic diffuse emissionriddle.
Acknowledgments
The
INTEGRAL
SPI project has been completed under the responsibility and leadershipof CNES. We are grateful to ASI, CEA, CNES, DLR, ESA, INTA, NASA and OSTC forsupport. 14 –AWS is supported by the German Bundesministerium f¨ur Bildung, Wissenschaft, Forschungund Technologie (BMBF/DLR) under contract No. FKZ 50 OG 0502.
REFERENCES
Bazzano A., Stephen J.B., Fiocchi M. et al., 2006, ApJ, 649, L9Bird A.J., Malizia A., Bazzano A. et al., 2006, ApJS, 170, 175Bouchet L., Roques J. P., Mandrou P. et al., 2005, ApJ, 635, 1115 (Paper I)Brown B. & Leventhal M., 1987, ApJ, 319, 637Churazov E., Sunyaev R., Sazonov S. et al, 2005, MNRAS, 357, 1377Dame T. M., Hartmann D. & Thaddeus P., 2001, ApJ, 547, 792Dean A.J., Hill, A. B., Stephen J. B., et al, 2005, A&A, 443, 485Dogiel V. A., Sch¨onfelder V. and Strong, A. W., 2002, A&A, 382,730Dubath P., Kn¨odlseder J., Skinner G.K., et al., 2005, MNRAS, 357, 420Duff I., Grimes R. and J. Lewis, Sparse matrix test problems, ACM Trans. Math. Soft., 15(1989), pp. 1-14.Jean P., , et al., 2006, A&A, 445, 579Jensen P.L., Clausen K., Cassi C., et al., 2003, A&A, 411, L7Kinzer R.L., Purcell W.R., Kurfess J.D. et al., 1999, ApJ, 515, 215Kinzer R.L., Milnes P.A., Kurfess J.D. et al., 2001, ApJ, 559, 282Kn¨odlseder J., 1999, PhD, Universit´e P. Sabatier, Toulouse, FranceKn¨odlseder J., Bennett K., Bloemen H., et al., 1999, A&A, 344, 68Kn¨odlseder J., Pierre J., Lonjou V., et al., 2005, A&A, 441, 513Krivonos R., Revnivtsev M., Churazov E. et al., 2007, A&A, 463, 957Kuiper, L., Hermsen W. and Mendez M., 2004, ApJ, 613, 1173Lebrun F., Terrier R., Bazzano A. et al., 2004, Nature, 428, 293 15 –Masai K., Dogiel V. A., Inoue H. et al., 2002, ApJ, 581, 1071Milne, P. A., Kurfess, J. D., Kinzer et al., 2000, AIP Conf. Proc., 510, 211Moskalenko I. V. et al., 2007, AIP Conf Proc 921, 490Porter T.A., Moskalenko I. V., Strong A. W., Orlando E. and Bouchet L., 2007, ApJ,submittedPurcell W.R., Bouchet L., Johnson W.N. et al., 1996, A&AS, 120, 389Revnivtsev M., Sazonov S., Gilfanov M. et al., 2006, A&A, 452, 169Roques J.P., Schanne S., Von Kienlin A., et al, 2003, A&A, 411, L91Schr¨oder R., Schlickeiser R. and Strong A. W., 2005. A&A, 442, L45Skibo J. G., Johnson W. N., Kurfess J. D. et al., 1997, ApJ, 483, L95Skinner G. & Connell P., 2003, A&A, 411, L123Strong A. W. & Moskalenko I. V., 1998, ApJ, 509, 212Strong A. W., Moskalenko I. V. and Reimer O., 2004, ApJ 613, 962Strong A. W., Diehl R., Halloin H. et al., 2005, A&A, 444, 495Strong A. W., 2007, Astrophysics and Space Science, 309, 35 Proceedings of Conference’The multi-messenger approach to high-energy gamma-ray sources’, Barcelona, 2006.astro-ph/0609359v3Strong A. W., Moskalenko I. V. and Ptuskin V. S., 2007, Ann. Rev. Nuc. Part. Sci. 57, 285Suleimanov V., Revnivtsev M., & Ritter H., 2005, A&A, 435, 191Terrier R., Lebrun F., Belanger G. et al., 2004, Proc. 5th INTEGRAL Workshop, ESASP-552, p513; astro-ph/0405207Vedrenne G., Roques J.P., Schonfelder V., et al, 2003, A&A, 411, L63Weidenspointner G., Kn¨odlseder J., Pierre J. et al, 2006, A&A, 450, 1013
This preprint was prepared with the AAS L A TEX macros v5.0.
16 –Fig. 1.— 25-50 keV
INTEGRAL
SPI exposure map. Units are in cm s. This map takesinto account the differential sensitivity of SPI accross its field of view. 17 –Fig. 2.— Sky-maps in the 25-50 keV (top-left), 50-100 keV (top-right), 100-200 keV (bottom-left) and 200-600 (bottom-right) energy bands. The images are scaled logarithmically with arainbow color map (the scale of colors ranges from black (weakest intensity) to red (strongestintensity)). The scale is saturated to reveal the weakest sources. There are some systematicsdue to strong variable sources such as Cyg X-1 combined with the finite precision of SPIresponse, especially in the 25-50 keV band. We take into account this systematic by using ahigher threshold in this domain. 18 –Fig. 3.— Diffuse emission morphology in different energy bands (significance maps): 25-50keV, 50-100 keV, 100-200 keV, 200-600 keV and 1.8-7.8 MeV from up to bottom. 19 –Fig. 4.— Longitude profiles in different energy bands for | b | ≤ . ◦ (except for the last band).Dotted and dashed lines correspond respectively to the NIR 4.9 µ and CO maps. For the200-600 keV band, the annihilation contribution (8 c irc gaussian distribution) is representedby the solide line. 20 –Fig. 5.— Latitude profiles in different energy bands for | l | ≤ ◦ (except for the last band).Dotted and dashed lines correspond respectively to the NIR 4.9 µ and CO maps. For the200-600 keV band, the annihilation contribution (8 c irc gaussian distribution) is representedby the solide line. 21 –Fig. 6.— map of the significance of the 511 keV line emissionFig. 7.— Longitude profile in the 511 keV line for | b | ≤ ◦ . Dotted-dashed line correspondsto a 8 ◦ axisymetric Gaussian. Dotted line corresponds to the extended distribution (240 µ map model). The sum of both has been integrated on the same bins as the data (histogram)to compare to them.Fig. 8.— Latitude profile in the 511 keV line for | l | ≤ ◦ . Dotted-dashed line correspondsto a 8 ◦ axisymetric Gaussian. Dotted line corresponds to the extended distribution (240 µ map model). The sum of both has been integrated on the same bins as the data (histogram)to compare to them. 22 –Fig. 9.— Spectra of the different emission components in the central radian of the Galaxy,for | l | ≤ ◦ and | b | ≤ ◦ : Sum of sources (stars), annihilation spectrum (long dashed line)and total “diffuse” emission (solid line). “Diffuse” contiunuum components are described bya power law (dotted line) and power law plus exponential cutoff (dashed line). 23 –Fig. 10.— SPI all sky diffuse emission spectrum (triangles) compared to the broad bandspectrum compiled by Krivonos et al. (2007). Small diamonds are RXTE
PCA ( <
17 keV)and
INTEGRAL
IBIS ( >
17 keV) data, the big ones
CGRO
OSSE & EGRET data. Thedotted line represent the SPI Galactic Ridge emission best fit. 24 –Table 1. Sources catalogue
Name l b 25 - 50 keV 50 - 100 keV 100 - 200 keV 200 - 600 keVdeg deg mCrab mCrab mCrab mCrabA0535+26 ∗ ± ± < ± ± ± ± ± ± < ± ± ± < ± ± < ± ± ± < ± ± < ± ± < ∗ ± ± ± < ± ± ± ± ± < ± ± < ± ± < ± ± < ± ± ± < ± ± < ± ± < ± < ± < ± < ± ± ± < ± ± < ± < ± ± < ± < ± ± < ± < ± < ∗ ± ± < ± ± < ± ± ± < ± ± < ± ± ± ± ± < ± ± < ± ± ± ± ± ± ± < ± ± < ± ± ± < + ± < ± ± ± < + ± ± ± < ± ± ± ± ± < ± < ± ± < < ± ± < ± ± ± ±
25 –Table 1—Continued
Name l b 25 - 50 keV 50 - 100 keV 100 - 200 keV 200 - 600 keVdeg deg mCrab mCrab mCrab mCrab4U 1538-522 327.43 2.2 12.8 ± < ± < ± ± < ± ± < + ± ± ± < ± ± ± < ± ± ± < a ± ± ± ± a ± ± < ± ± ± ± ± ± < ∗∗∗ ± ± < + ± ± < ± ± ± < ± ± < ± ± ± ± ± ± ± < ± < ± ± < ∗ ± ± ± ± ± < ± < ± ± ± ± ± < ± ± ± ± ± < ± ± ± < ± ± ± ± ± < ± ± < ± ± ± ± + ± ± < ± ± < ± ± < + ± ± < ± ± < ∗ ± ± ± ± ± ± ± < ± ± ± < ∗ ± ± ± < ± ± ± ± ± ± < ± ± ± < ± < ± ± < ± ± < ± ± < ± ± ± <
26 –Table 1—Continued
Name l b 25 - 50 keV 50 - 100 keV 100 - 200 keV 200 - 600 keVdeg deg mCrab mCrab mCrab mCrabGX 1+4 1.93 4.8 48.3 ± ± ± < ± ± < ± ± < ± ± ± ± ± < ± ± < ± < ± ± < ± ± ± ± ± < ± ± < ± < ± ± ± ± ± < ± ± < b ± ± ± < ± ± ± < ± ± ± < + ± ± < ± ± < ± ± < ± ± < ± < ± ± < ± ± < ± ± ± < ± ± < ± ± < + ± ± < ± ± ± ∗ ± ± ± < ± ± < ± ± < + ± ± ± < + ± ± < + ± < ± < ± < ± ± ± < ± ± ± < ± ± ± < ± ± < ± ± < ∗ ± ± ± < ± ± < ± ± < ± < ± ± <
27 –Table 1—Continued
Name l b 25 - 50 keV 50 - 100 keV 100 - 200 keV 200 - 600 keVdeg deg mCrab mCrab mCrab mCrab4U 1909+07 41.90 -0.8 11.1 ± ± < + ± ± < ± ± < ∗ ± ± ± ± ± ± < ± ± < ± ± < ± ± < ± ± ± < ∗∗ ± ± ± ± ± < ± ± < ± ± ± ± ± < ± < ± < ± ± ± < ± ± < ± < γ Cas 123.55 -2.2 3.0 ± ± < ± ± < ± ± < ± ± < ± ± < ± ± < ∗ ± ± < ± ± ± ± ± ± ± < ± ± ± ∼
28 –Table 2. The best fit parameters for the spectra presented on fig 9 index1 Ecut F keV index2 F keV F keV F posit f p × − ph × − ph × − ph × − ph keV cm − s − keV − cm − s − keV − cm − s − cm − s − Diffuse 1 0 (fixed) 7.5 ± ± ± ± ± ± ± ± ±±