Evidence for a second component in the high-energy core emission from Centaurus A?
aa r X i v : . [ a s t r o - ph . H E ] F e b ApJ submitted
Evidence for a second component in the high-energy coreemission from Centaurus A?
N. Sahakyan ICRANet, Piazz della Repubblica 10, I-65122 Pescara, ItalyInstitute for Physical Research, NAS of Armenia, Ashtarak-2, 0203, Armenia
R. Yang
Key Laboratory of Dark Matter and Space Astronomy, Purple Mountain Observatory, CAS,Nanjing, 210008, China
F.A. Aharonian
Dublin Institute for Advanced Studies, 31 Fitzwilliam Place, Dublin 2, Ireland andF.M. Rieger
Max-Planck-Institut f¨ur Kernphysik, P.O. Box 103980, 69029 Heidelberg, Germany
ABSTRACT
We report on an analysis of
Fermi -LAT data from four year of observationsof the nearby radio galaxy Centaurus A (Cen A). The increased photon statisticsresults in a detection of high-energy ( >
100 MeV) γ -rays up to 50 GeV from thecore of Cen A, with a detection significance of about 44 σ . The average gamma-rayspectrum of the core reveals evidence for a possible deviation from a simple power-law. A likelihood analysis with a broken power-law model shows that the photonindex becomes harder above E b ≃ = 2 . ± .
03 belowto Γ = 2 . ± .
20 above. This hardening could be caused by the contributionof an additional high-energy component beyond the common synchrotron-selfCompton jet emission. A variability analysis of the light curve with 15-, 30-, and60-day bins does not provide evidence for variability for any of the components.Indications for a possible variability of the observed flux are found on 45-daytime scale, but the statistics do not allow us to make a definite conclusion in thisregards. We compare our results with the spectrum reported by H.E.S.S. in theTeV energy range and discuss possible origins for the hardening observed. 2 –
Subject headings:
Galaxies: active – galaxies: individual (Cen A)
1. Introduction
The prominent radio galaxy Centaurus A (NGC 5128), at a distance of ≃ . ′ ≃ . ∼ ◦ and oriented primar-ily in the north-south direction (Feain et al. 2011). Optical images reveal the bright hostgalaxy bulge ( ∼ ′ bulge radius) and the famous, warped dark lane of gas, dust and youngstars ( ∼ ′ in east-west extension) which obscures the inner part of the galaxy. Chan-dra X-ray observations show a one-sided, kpc-scale (up to ∼ . > ∼ ◦ and characterized by moderate (radio) bulk flow speed < ∼ . c (Tingay et al. 1998;Hardcastle et al. 2003; M¨uller et al. 2011). Its bolometric luminosity of L ∼ erg/s(Meisenheimer et al. 2007; van der Wolk et al. 2010), accompanied by indications for thelack of a dust torus, is thought to be powered by gas accretion onto a supermassive blackhole of mass M BH ≃ (3 − × M ⊙ (Marconi et al. 2006; Cappellari et al. 2009).At MeV energies, Cen A has been observed with both OSSE (0.05-4 MeV) and COMP-TEL (0.75-30 MeV) onboard the Compton Gamma-Ray Observatory (CGRO) in the period1991-1995 (Steinle et al. 1998). An agreement of the OSSE spectrum with the COMP-TEL one in the transition region around 1 MeV, and correlated variability has been found(Steinle et al. 1998). At higher energies, a marginal (3 σ ) detection of gamma-rays from thecore of Cen A was reported with EGRET (0.1-1.0 GeV), but due its large angular resolu-tion the association with the core remained rather uncertain (Hartman et al. 1999). Unlikethe initial variability (month-type?) seen at lower energies, the flux detected by EGRETappeared stable during the whole period of CGRO observation (Sreekumar et al. 1999).At VHE ( >
100 GeV) energies, Cen A has also been detected (with a significance of 5 σ )by the H.E.S.S. array based on observations in 2004-2008. The results show an average VHEspectrum compatible with a power law of photon index Γ = 2 . ± . stat ± . syst and an 3 –integral flux F ( E >
250 GeV) = (1 . ± . stat ) × − cm − s − (Aharonian et al. 2009).No evidence for variability has been found in the H.E.S.S. data set, but given the weak signalno certain conclusions can be drawn. Fermi -LAT has reported the detection of high energy (HE, >
100 MeV) gamma-rays from both the core (i.e., within 0 . ◦ ) and the giant radio lobes (Abdo et al. 2010b;Abdo et al. 2010a). The analysis of 10 months of data revealed a point-like HE emissionregion coincident with the position of the radio core of Cen A, the emission being well de-scribed by a power-law function with a photon index ≈ .
7, quite similar to the one in theVHE regime. Also, no variability (on 15 d and 30 d time scales) has been found. A simpleextrapolation of the HE power-law spectrum to the VHE regime, however, fails to account forthe TeV core flux as measured by H.E.S.S., which could indicate the need for an additionalcontribution towards the highest energies. The giant lobes were detected with a significanceof 5 σ and 8 σ for the northern and the southern structure, respectively (Abdo et al. 2010a).The HE lobe spectrum could be described by a power-law function extending up to 2 or 3GeV with photon indices of Γ ≈ .
6. A recent analysis of a three times larger data set hasconfirmed the existence of these HE lobes, but also showed that the HE emission extendswell beyond the WMAP radio image (Yang et al. 2012).The apparent lack of significant variability features at GeV and TeV energies has sofar precluded robust inferences as to the physical origin of the core emission in Cen A.Unfortunately, the resolutions of current gamma-ray instruments is not sufficient to localizethe gamma-ray emitting region(s) either: The angular resolutions of both the H.E.S.S. array( ∼ . ◦ ) and Fermi -LAT (0 . ◦ -1 ◦ , depending on energy) correspond to linear sizes of thegamma-ray emitting region(s) of about 5 kpc or larger. This ∼ Fermi -LAT data.
2. Fermi-LAT Data Analysis2.1. Data Extraction
Fermi -LAT on board the Fermi satellite is a pair-conversion telescope designed to detecthigh-energy γ -rays in the energy range 20 MeV - 300 GeV (Atwood et al. 2009). It constantlyscans the entire sky every three hours and is always in survey mode.For the present analysis we use publicly available Fermi -LAT ∼ gtselect and gtmktime tools and retained only events belongingto the class 2, as is recommended by the Fermi/LAT science team . To reject atmosphericgamma-rays from the Earth’s limb, events with zenith angle <
100 deg are selected. Thestandard binned maximum likelihood analysis is performed using events in the energy range0.1–100 GeV extracted from a 10 ◦ region centered on the location of Cen A, which is referredto as ’region of interest’ (ROI). The fitting model includes diffuse emission components andgamma-ray sources within ROI which are not associated with Cen A (the model file is createdbased on Fermi -LAT second catalog (Noland et al. 2011). In the model file, the giant radiolobes were modeled using templates from WMAP-k band observation of the source whichis extracted from NASA’s SkyView. Although our previous results showed that the HEextension and the radio lobe regions do not perfectly match (Yang et al. 2012), this doesnot affect the central parts of relevance here. The background was parameterized with thefiles gal 2yearp7v6 v0.fits and iso p7v6source.txt and the normalizations of both componentswere allowed to vary freely during the spectral point fitting.
Initially the continuum gamma-ray emission of the core of Cen A is modeled with asingle power law. The normalization and power-law index are considered as free parametersthen the binned likelihood analysis is performed. From a binned gtlike analysis, the best-fitpower-law parameters for the core of Cen A are (cid:18) dNdE (cid:19) P = (2 . ± . × − (cid:18) E
100 MeV (cid:19) − . ± . . (1)This corresponds to an integral flux of F γ = (1 . ± . × − photon cm − s − , (2)with only statistical errors taken into account. The test statistic (defined as TS = 2(log L -log L ), where L and L are the likelihoods when the source is included or not) is T S = 1978above 100 MeV, corresponding to a ≈ σ detection significance. The results are consistentwith the parameters found in (Abdo et al. 2010b), namely photon index Γ = 2 . ± . http://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Cicerone/Cicerone Data Exploration/Data preparation.html . ± . × − ph cm − s − above 100MeV (model B). Figure 1 shows the spectrum of the core of Cen A obtained by separatelyrunning gtlike for 12 energy bands, where the dashed line shows the best-fit power-lawfunction for the data given in Eq. (1). For the highest energy bin (56.2-100 GeV), an upperlimit is shown. The spectrum shows a tendency for a deviation from a single power-law à à à à à à à àæ æ æ æ æ æ æ æ æ æ æ æ - - - - - - - H E H GeV LL l og H d N (cid:144) d E H c m - s - G e V - LL Fig. 1.— Average high-energy gamma-ray ( >
100 MeV) spectrum of the core of Cen A(black points - this work) as compared to the one based on the initial 10 month data set(blue squares - Abdo et al. 2010b). The dashed black line shows the power-law functiondetermined from the gtlike. The blue and the red line show power-law fits to the energybands below and above E b ≃ χ fit of the power-law modelto the data gives a relatively poor fit with χ = 39 . P ( χ ) < × − . In order to investigate this in more detail, the corespectrum is modeled with a broken power-law model and gtlike tool is retried. The best-fitbroken power-law parameters are (cid:18) dNdE (cid:19) BP = (1 . ± . × − (cid:18) EE b (cid:19) − Γ , , (3) 6 –and F γ = (1 . ± . × − photon cm − s − , (4)with Γ = 2 . ± .
02 and Γ = 2 . ± .
14 below and above E b = (4 . ± .
09) GeV,respectively. In order to compare the power-law and the broken-power-law model, a loglikelihood ratio test between the models is applied. The test statistic is twice the differencein these log-likelihoods, which gives 9 for this case. Note that the probability distributionof the test statistic can be approximated by a χ distribution with 2 dof, corresponding todifferent degrees of freedom between the two functions. The results give P ( χ ) = 0 . gtlike tool is separately applied to these two energy bands. The photon index and flux between100 MeV and 4 GeV are Γ = 2 . ± .
02 and F γ = (1 . ± . × − photon cm − s − ,respectively, and the test statistics gives TS=1944. The result is shown with a blue line inFig. 1. On the other hand, for the energy range (4-100) GeV we obtain Γ = 2 . ± . F γ = (4 . ± . × − photon cm − s − , respectively, and a TS value of 124.4,corresponding to a ≈ σ detection significance. This component is depicted with a red linein Fig. 1.
3. Temporal Variability
Variability, if present, could provide important constraints on the emitting region(s).An observed HE flux variation on time scale t var , for example, would limit the (intrinsic)size of the gamma-ray production region to R ′ ≤ δ D z ct var . However, previous HE and VHEgamma-ray observations of the core of Cen A with Fermi -LAT (Abdo et al. 2010b) andH.E.S.S. (Aharonian et al. 2009) did not find evidence for significant variability. Here weinvestigate whether the longer (4 yr) data set employed changes this situation. We thusdivide the whole data set (from August 4th 2008 to October 1st 2012) into different timebins and generate light curves using the unbinned likelihood analysis with gtlike . Due tolimited photon statistics the shortest time scale that one can probe is 15 days. In ouranalysis we generate light curves in 15, 30, 45 and 60 day bins. The normalization of thecore and background point sources are treated as free parameters, but the photon indices ofall sources and the normalization of the lobes are fixed to the values obtained in 100 MeV- 7 –100 GeV energy range for the whole time period. Since no variability is expected for theunderlying background diffuse emission, the normalization of both background componentsis fixed to the values obtained for the whole time period. To search for variability, a χ
54 800 55 000 55 200 55 400 55 600 55 800 56 000 56 2001.01.52.02.5 time H MJD L F l ux @ - ph c m - s - D Fig. 2.— Gamma-ray light curve from August 4th 2008 to October 1st 2012. The bin sizeis 45 day. The background diffuse emission (both galactic and extragalactic) is fixed to thebest-fit parameters obtained for the overall time fit. While some variability may be present,limited statistics do not yet allow to make definite conclusions.test was performed. The result for the light curve with 15 day bins is χ /d.o.f. = 1 .
22 andthe probability is P ( χ ) = 0 .
07. For the light curves with 30 day and 60 day bins we find χ /d.o.f. = 1 .
37 and χ /d.o.f. = 1 .
32, corresponding to P ( χ ) = 0 .
04 and P ( χ ) = 0 . χ /d.o.f. ≈ .
61 and P ( χ ) = 0 .
4. Discussion and Conclusion
In the case of high-frequency-peaked BL Lac objects, homogeneous leptonic synchrotron-self-Compton (SSC) jet models often provide reasonable descriptions of their overall spectralenergy distributions (SEDs). For Cen A, however, classical one-zone SSC models (under theproviso of modest Doppler beaming) are unable to satisfactorily account for its core SED upto the highest energies (cf. Chiaberge et al. 2001; Lenain et al. 2008; Abdo et al. 2010b). It 8 –seems thus well possible, that an additional component contributes to the observed emissionat these energies (e.g., Lenain et al. 2008; Rieger & Aharonian 2009). The results presentedhere indeed provides support for such a consideration. Our analysis of the 4 yr-data set æ æ æ æ æ æ æ æ æ æ æ à à à à à à - - - - - -
10 log H E H GeV LL l og H E d N (cid:144) d E H e r g c m - s - LL Fig. 3.— Gamma-ray spectrum for the core of Cen A from high (
Fermi -LAT , this work)to very high (H.E.S.S., blue squares) energies. The blue bowtie represents a power-law withphoton index 2 .
74, and the red bowtie a power-law with photon index 2 .
09. The dashed linesshow extrapolations of these models to higher energies. The power-law extrapolation of thelow-energy component (blue lines) would under-predict the fluxes observed at TeV energies.reveals that the HE core spectrum of Cen A shows a ”break” with photon index changingfrom ≃ . ≃ . E b ≃ . F γ = (1 . ± . × − photon cm − s − for the component below4 GeV corresponds to an apparent (isotropic) γ -ray luminosity of L γ (0 . − ≃ erg s − . The component above 4 GeV, on the other hand, is characterized by an isotropicHE luminosity of L γ ( > ≃ . × erg s − . This is an order of magnitude less whencompared with the first component, but still larger than the VHE luminosity reported byH.E.S.S. L γ ( >
250 GeV) = 2 . × erg s − (Aharonian et al. 2009). All luminosities are 9 –below the Eddington luminosity corresponding to the black hole mass in Cen A; nevertheless,they are still quite impressive when compared with the other nearby radio galaxy M87containing a much more massive black hole.Figure 3 shows the gamma-ray spectrum for the core of Cen A up to TeV energies.As one can see, the flux expected based on a power-law extrapolation of the low-energycomponent (below the break) clearly falls below the TeV flux reported by H.E.S.S.. Althoughthe uncertainties in the photon index are large, it is clear that the spectrum becomes harderabove 4 GeV. Remarkably, a simple extrapolation of the second (above the break) high-energy component to TeV energies could potentially allow one to match the average H.E.S.S.spectrum. These spectral considerations support the conclusion that we may actually bedealing with two (or perhaps even more) components contributing to the HE gamma-raycore spectrum of Cen A. Our analysis of the HE light curves provides some weak indicationfor a possible variability on 45 day time scale, but the statistics are not sufficient to drawclear inferences.The limited angular resolution ( ∼ γ -ray induced pair-cascades in a torus-like re-gion (at ∼ r s ) (e.g. Roustazadeh & B¨ottcher 2011) (v) secondary Compton up-scatteringof host galaxy starlight (Stawarz et al. 2006) or (vi) inverse-Compton (IC) processes in thekpc-scale jet (e.g. Hardcastle & Croston 2011). What concerns the more compact scenarios(i)-(iv) just mentioned: Opacity considerations do not a priori exclude a near-BH-origin, butcould potentially affect the spectrum towards highest energies (e.g. Rieger 2011). A SSCmulti-blob VHE contribution, on the other hand, requires the soft gamma-rays to be due tosynchrotron instead of IC processes, in which case correlated variability might be expected.Photo-meson ( pγ ) interactions with, e.g., UV or IR background photons ( n γ ) require thepresence of UHECR protons, which seems feasible for Cen A. However, as the mean freepaths λ ∼ / ( σ pγ n γ K p ) of protons through the relevant photon fields are comparativelylarge, usually only a modest fraction of the proton energy can be converted into secondary 10 –particles. Models of this type thus tend to need an injection power in high-energy protonsexceeding the average jet power of ∼ − erg/s (e.g. Yang et al. 2012). The efficiencyof IC-supported pair cascades in Cen A, on the other hand, appears constrained by lowaccretion modes and the possible absence of a dust torus. Considering the more extendedscenarios (v)-(vi): Partial absorption ( ∼ γ -rayluminosity ≈ erg / s above 100 MeV is larger by two orders of magnitude than the γ -rayluminosity of the Milky Way, which could be related to a higher rate of cosmic-ray productionand a more effective confinement in the case of NGC 5128. Moreover, gamma-rays might alsobe produced in a diluted R halo ∼
30 kpc (halo) region of this galaxy. Despite the low densityof gas, gamma-ray production on characteristic timescale t pp ≈ × ( n/ − cm − ) − yr canbe effective, even for a relatively fast diffusion of cosmic rays in this region. More specifically,the efficiency could be close to one, if the diffusion coefficient at multi-GeV energies doesnot exceed D ∼ R /t pp ∼ cm /s. This seems an interesting possibility, especiallyfor the second (hard) HE component with photon index close to 2 .
1, in the context of itssimilarity to the gamma-ray spectrum of the so-called ’Fermi Bubbles’ around the center ofour Galaxy (Su et al. 201)). The much higher luminosity (by ∼ − REFERENCES
Aharonian, F. et al. (H.E.S.S. Collaboration) 2009, ApJ, 695, L40Aharonian, F. et al. (H.E.S.S. Collaboration) 2006, Science, 314, 1424Atwood, W., Abdo, A., Ackermann, M. et al., 2009,ApJ, 697, 1071Abdo, A.A. et al. (Fermi-LAT) 2010b, ApJ, 719, 1433Abdo, A.A. et al. (Fermi-LAT) 2010a, Science, 328, 725Cappellari, M. et al. 2009, MNRAS, 394, 660Chiaberge, M., Capetti, A. & Celotti, A. 2001, MNRAS, 324, L33Feain, I.J., Cornwell, T.J., Ekers, R.D. et al., ApJ, 740, 17,Fanaroff, B., & Riley, J., MNRAS, 1974, 167, 31PHarris, G.L.H., Rejkuba, M., & Harris, W.E. 2010, PASA, 27, 457Hartman, R., Bertsch, D., Bloom, S., et al., ApJS, 1999,123, 79Meisenheimer K., Tristram K.R.W., Jaffe W., et al., 2007, A&A, 471, 453M¨uller, C., Kadler, M., Ojha, R. et al. 2011, A&A, 530, L11Morganti, R., Fosbury, R., Hook, R. et al., MNRAS, 1992, 256, 1PHardcastle, M.J. et al. 2003, ApJ, 593, 169Hardcastle, M.J. and Croston, J.H. 2011, MNRAS, 415, 133Israel, F.P. 1998, A&A Rev., 8, 237Kachelrieß, Ostapchenko, S. & Tomas, R. 2010, PASA, 27, 482Kraft, R., Forman, W. et al., ApJ, 2002, 569, 54Lenain, J.-P. et al. 2008, A&A, 478, 111Nolan, P. L., Abdo, A. A., Ackermann, M., et al., 2012, ApJS, 199, 31Marconi, A. et al. 2006, A&A, 448, 921Meisenheimer K., Tristram K.R.W., Jaffe W., et al., 2007, A&A, 471, 453 12 –Rieger, F.M. & Aharonian 2009, A&A, 506, L41Rieger, F.M. 2011, IJMPD 20, 1547Roustazadeh, P., & B¨ottcher, M. 2011, ApJ, 728, 134Sreekumar, P. et al. 1999, APh, 11, 221Steinle, H., Bennett, K., Bloemen, H., et al., A&A, 1998, 330, 97Stawarz, L., Aharonian F., Wagner, S., & Ostrowski, M. 2006, MNRAS, 371, 1705Sahu, S., Zhang, B. & Fraija, N. 2012, PRD, 85, 043012Su, M., Slatyer, T.R., & Finkbeiner, D.P. 2010, ApJ, 724, 1044Tingay, S. et al. 1998, AJ, 115, 960van der Wolk, G. et al. 2010, A&A, 511, A64Yang, R., Sahakyan, N., et al., 2012, A&A 542, A19