Detection of Gamma Rays from a Starburst Galaxy

See allHide authors and affiliations

Science  20 Nov 2009:
Vol. 326, Issue 5956, pp. 1080-1082
DOI: 10.1126/science.1178826


Starburst galaxies exhibit in their central regions a highly increased rate of supernovae, the remnants of which are thought to accelerate energetic cosmic rays up to energies of ~1015 electron volts. We report the detection of gamma rays—tracers of such cosmic rays—from the starburst galaxy NGC 253 using the High Energy Stereoscopic System (H.E.S.S.) array of imaging atmospheric Cherenkov telescopes. The gamma-ray flux above 220 billion electron volts is F = (5.5 ± 1.0stat ± 2.8sys) × 10−13 cm−2 s−1, implying a cosmic-ray density about three orders of magnitude larger than that in the center of the Milky Way. The fraction of cosmic-ray energy channeled into gamma rays in this starburst environment is five times as large as that in our Galaxy.

Starburst galaxies are characterized by a boosted formation rate of massive stars and an increased rate of supernovae in localized regions, which also exhibit very high densities of gas and of radiation fields. Their optical and infrared (IR) luminosity is dominated by radiation from numerous young massive stars, most of which later explode as supernovae. Given that most cosmic rays in normal galaxies are expected to be accelerated in supernova remnants (1), starburst regions represent a favorable environment for the acceleration of cosmic rays, resulting in cosmic-ray energy densities orders of magnitude higher compared with the local value in our Galaxy [e.g., (2)]. Cosmic-ray protons can produce gamma radiation by inelastic collisions with ambient gas particles and subsequent π0-decay. Primary and secondary cosmic-ray electrons can also produce gamma radiation by Bremsstrahlung and up-scattering of low-energy photons from massive stars or from ambient radiation fields. Starburst galaxies are therefore considered promising sources of gamma-ray emission (3, 4). Here, we report the detection of very high energy (VHE) (>100 GeV) gamma rays from the starburst galaxy NGC 253.

NGC 253, at a distance of 2.6 to 3.9 Mpc (57), is one of the closest spiral galaxies outside the Local Group. It is similar to our Milky Way in its overall star formation rate. Its nucleus, however, is a starburst region (8) of very small spatial extent (a few 100 pc), characterized by a very high star formation rate per volume and thus also by a very large mechanical energy production in the form of supernova explosions. Star formation activity is estimated to have been going on for 20 to 30 million years (8) and can therefore be considered to be in a steady state for the time scales governing cosmic-ray transport. A supernova rate of ~0.1 year−1 has been inferred for the entire galaxy from radio (9) and infrared (10) observations. The rate is most pronounced in the central starburst region, where a conservative estimate yields a rate of supernovae ~0.03 year−1, which is comparable to that in our Galaxy (8). This suggests a very high local cosmic-ray energy density. The mean density of the interstellar gas in the central starburst region is n ≅ 600 protons cm−3 (11), which is almost three orders of magnitude higher than the average density of the gas in the Milky Way. The thermal continuum emission of NGC 253 peaks in the far infrared (FIR) energy band at ~100 μm, with a luminosity that is ~5 times the total radiation from our own Galaxy (12). This FIR emission originates from interstellar dust, which reprocesses starlight from the numerous young massive stars. The emission is highly concentrated toward the small central starburst nucleus. Therefore, the density of the radiation field in the starburst region is about a factor 105 larger than the average value in the inner 100 pc around the Galactic Center. The activity of NGC 253 has been shown to be of a pure starburst nature and not due to an active supermassive black hole (13, 14). Observations of radio (15, 16) and thermal x-ray emission (17, 18) show a hot diffuse halo, consistent with the existence of a galactic wind extending out to ~9 kpc from the galactic plane that transports matter and cosmic rays from the nucleus to intergalactic space and reaches asymptotically a bulk speed of ~900 km/s (19).

Given its proximity and its extraordinary properties, NGC 253 was predicted to emit gamma rays at a detectable level (4). Recent calculations give similar results (20, 21). Previously, only upper limits have been reported in the gamma-ray range, in the MeV-GeV range by the Energetic Gamma-Ray Experiment Telescope (EGRET) (22), and in the TeV range by the High Energy Stereoscopic System (H.E.S.S.) [based on 28 hours of observation (23)] and by the Collaboration of Australia and Nippon (Japan) for a Gamma-Ray Observatory in the Outback (CANGAROO) III (24). We report the result of continued observations of NGC 253 with the H.E.S.S. telescope system with a much larger data sample. [See the Supporting Online Material (SOM) for a description of the experiment and the detection technique.]

We obtained observations in 2005, 2007, and 2008. After rejecting those data that did not have the required quality, we analyzed 119 hours of live-time data. Even with this extremely long exposure, the measured VHE gamma-ray flux of NGC 253 is at the limit of the H.E.S.S. sensitivity. Thus, advanced image analysis techniques were required to extract a significant signal on top of the uniform background of local cosmic rays impinging on Earth’s atmosphere—only one out of 105 recorded air showers represents a gamma ray from NGC 253. We used the Model Analysis technique (25) (SOM), based on which we detected an excess of 247 events from the direction of NGC 253 above 220 GeV, corresponding to a statistical significance of 5.2 standard deviations (Fig. 1). The signal is steady and stable (a fit over the period of 3 years to a constant has a chance probability of 47%). The source position is αJ2000 = 0h47m33s.6 ±30s, δJ2000 = −25°18′8′′±27′′ consistent with the position of the optical center of NGC 253 (αJ2000 = 0h47m33s.1, δJ2000 = −25°17′18′′). The distribution of excess events is consistent with the point spread function of the H.E.S.S. instrument, implying a source size of less than 4.2′ (at a 1σ confidence level) (see Fig. 2 for a comparison of the angular distribution of the gamma events with a point-like simulated signal). The integral flux of the source above the threshold of 220 GeV is F(>220 GeV) = (5.5 ± 1.0stat ± 2.8 sys) × 10−13 cm−2 s−1. This corresponds to 0.3% of the VHE gamma-ray flux from the Crab Nebula (26); given the well-known uncertainties in the diffusion part of the particle transport properties, as well as the only approximate knowledge of the starburst parameters, it is consistent with the original prediction (4) (Fig. 3).

Fig. 1

A smoothed map of VHE gamma-ray excess of the 1.5° by 1.5° region around NGC 253. A Gaussian with root mean square of 4.2′ was used to smooth the map in order to reduce the effect of fluctuations. The star shows the optical center of NGC 253. The inlay represents an image of a Monte Carlo simulated point source (i.e., the point-spread function of the instrument). The white contours represent the optical emission of the whole NGC 253, demonstrating that the VHE emission originates in the nucleus and not in the disk. The contours correspond to constant surface brightness of 25 magnitudes arc sec−2—traditionally used to illustrate the extent of the optical galaxy—and 23.94 magnitudes arc sec−2 according to (34).

Fig. 2

Reconstructed directions of the gamma-ray–like events around NGC 253. θ denotes the angular distance between the arrival direction and the position of the object. The background estimated from off-source regions is uniform in the θ2 representation and has been subtracted here. The signal is consistent with a H.E.S.S. point source (blue dashed line), corresponding to θ < 4.2′ or < 3.2 kpc at a distance of 2.6 Mpc.

Fig. 3

The observed integral flux of gamma rays from NGC 253 (red point) is compared to theoretical estimates (4, 20, 23). The solid line corresponds to the prediction by (4). The dashed line corresponds to the model (20). The gray-shaded band denotes the estimate (21). The error of the H.E.S.S. measurement includes systematic errors.

As an external galaxy detected in gamma rays that, as a key property, does not contain a massive black hole of sizeable associated luminosity, NGC 253 is a member of a class of gamma-ray emitters external to the Milky Way and the associated Large Magellanic Cloud (LMC). These gamma-ray emitters apparently produce their own cosmic-ray population. Except for the starburst, NGC 253 is a normal galaxy. So far, only the LMC, a small and close satellite of the Milky Way, was detected in gamma rays with the EGRET instrument (27). In contrast, there exists a class of external galaxies detected in gamma rays whose emission is—according to present knowledge—exclusively due to an active galactic nucleus (AGN), driven by a supermassive black hole in their center. Their physical characteristics are quite distinct from normal galaxies and not the subject of the discussion here.

The detection of NGC 253 in VHE gamma rays implies a high energy density of cosmic rays in this system. One can calculate a corresponding cosmic-ray density directly from the H.E.S.S. observations. Assuming a dominant hadronic origin of the gamma-ray emission, the spatial density Np(>Ep) of the gamma-ray generating protons in the starburst region with an energy exceeding Ep ≈ 220/0.17 GeV ≈ 1300 GeV is about 4.9 × 10−12 cm−3 for the measured gamma-ray flux above 220 GeV, independent of the distance to NGC 253. This is about 2000 times as large as the corresponding Galactic cosmic-ray number density at the solar system and about 1400 times as high as the density at the center of our Galaxy (28). Taking EpNp(>Ep) as a rough measure of the energy density of cosmic rays above energy Ep in NGC 253, EpNp(>Ep) ≈ 6.4 eV cm−3 for Ep > 1300 GeV. This is larger than the entire cosmic-ray energy density in the Galaxy near the solar system, which is dominated by GeV-particles.

Gamma-ray production represents one channel for conversion and loss of cosmic rays at TeV energies. The time between inelastic collisions of hadronic cosmic rays and target protons and nuclei at Ep ≈ 1300 GeV is of the order of 105 years for a mean gas density of about 600 protons cm−3. These collisional losses compete with two other processes in starbursts: spatial losses of particles convected out of the considered region by the wind, and diffusive losses (see the SOM for a summary of the cosmic-ray transport characteristics in NGC 253). Because of the very high gas density in the nucleus of NGC 253, the ratio of hadronic gamma-ray production to energy loss by transport is considerably higher than for a galaxy like ours. In the Milky Way, the ~1300-GeV gamma-ray–generating charged particles encounter about 0.6 g cm−2 of matter before they escape, extrapolating results from (29). Their mean free path for inelastic nuclear collisions is equivalent to about 56 g cm−2. Therefore, the Galactic ratio of gamma-ray production probability to the escape probability of 1300 GeV particles is about 10−2. If the cosmic-ray energy production in the starburst region of NGC 253 is in equilibrium with losses caused by nuclear collisions, then, for the measured gas density and supernova rate—together with an assumed cosmic-ray production efficiency of 1050 erg per event and a production spectrum ∝ E−2.1 (3, 23)—the expected integral gamma-ray flux above 220 GeV would be ≈10−11 cm−2 s−1. The observed flux is smaller than this calorimetric flux by a factor of ≈5 × 10−2—again independent of the distance. Therefore, the starburst region is only mildly calorimetric. For a comparison, see (23, 30). Nevertheless, the numbers imply that the conversion efficiency of protons into gamma rays in the starburst region of NGC 253 exceeds that in our Galaxy by almost an order of magnitude. This comparatively high efficiency has another consequence: Assuming that the remaining structure of NGC 253 is about the same as in our Galaxy, then the starburst nucleus is about 5 times as bright in VHE gamma rays as in the associated galaxy, and the starburst nucleus should outshine the rest of NGC 253. This is consistent with the detection of a H.E.S.S. point source (Fig. 1).

Given these results, one may ask whether they have a wider importance regarding the nonthermal particle population in the universe. A starburst galaxy such as NGC 253 is a potential model for a phase of galaxy formation as well as for two-body galaxy-galaxy interactions, especially in the dense environment of large galaxy clusters. High-energy gamma-ray emission as a result of these processes should accompany the thermal IR emission of such luminous infrared galaxies. The galactic winds present in these systems are expected to massively populate intergalactic space with nucleosynthesis products and cosmic rays.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S5


References and Notes

  1. The following description is based on a distance of 2.6 Mpc; the final conclusions are, however, independent of distance.
  2. The Model Analysis is based on a comparison and fit of observed air shower images with a precomputed library of images (31); another analysis technique, based on a machine learning algorithm called boosted decision trees (BDT) (32), was used to verify the results. It was trained with simulated gamma rays and with real cosmic-ray data from background fields (33). Both algorithms yield an improvement by a factor 1.5 to 1.7 in the statistical significance of faint sources compared with the standard image analysis (26), as verified with a number of other gamma-ray sources. The Model Analysis result is found to be consistent with the one obtained with the BDT analysis. Details are given in the SOM.
  3. The support of the Namibian authorities and of the University of Namibia in facilitating the construction and operation of H.E.S.S. is gratefully acknowledged, as is the support by the German Ministry for Education and Research (BMBF), the Max Planck Society, the French Ministry for Research, the CNRS-IN2P3 and the Astroparticle Interdisciplinary Programme of the CNRS, the UK Science and Technology Facilities Council (STFC), the IPNP of Charles University, the Polish Ministry of Science and Higher Education, the South African Department of Science and Technology and National Research Foundation, and the University of Namibia. U. Barres de Almeida is supported by the CAPES Coordenação de Aperfeiçoamento de Pessoal de Nível Superior Foundation, Ministry of Education of Brazil. The European Associated Laboratory for Gamma-Ray Astronomy is jointly supported by CNRS and Max-Planck-Gesellschaft. We appreciate the excellent work of the technical support staff in Berlin, Durham, Hamburg, Heidelberg, Palaiseau, Paris, Saclay, and Namibia in the construction and operation of the equipment.
View Abstract

Navigate This Article