SEARCH FOR GAMMA-RAY EMISSION FROM X-RAY-SELECTED SEYFERT GALAXIES WITH FERMI-LAT

The all-sky observations of celestial objects by the Large Area Telescope (LAT; Atwood et al. 2009) aboard the Fermi Gamma-ray Space Telescope confirmed that, in addition to gamma-ray bursts, there are at least two more general classes of bright extragalactic sources of γ-rays (Abdo et al. 2010b). One class comprises active galactic nuclei (AGNs) with powerful relativistic jets, including blazars, radio-loud Narrow-line Seyfert 1 galaxies (NLS1s), and radio galaxies, which produce beamed high-energy emission via inverse Compton scattering of soft photon fields on ultrarelativistic jet electrons. The other class consists of nearby galaxies with prominent starburst systems, which produce diffuse, un-beamed γ-ray emission resulting from the interactions of cosmic-ray particles with the interstellar medium (ISM). An important question arises whether those are the only classes of extragalactic γ-ray sources. This question motivated us to search systematically for GeV emission from Seyfert galaxies using Fermi-LAT. Seyfert galaxies constitute the most numerous class of AGNs in the local universe (local number density ∼10⁻⁴ Mpc⁻³), but at the same time, lack, in general, ultrarelativistic collimated outflows or starburst regions (e.g., Osterbrock 1989). Seyferts, hosted by late-type galaxies, were originally identified in the optical regime by the presence of strong emission lines from highly ionized gas in their spectra (Seyfert 1943). They are believed to harbor super-massive ($\mathcal {M}_{\rm BH} \sim 10^6\hbox{--}10^9 \, M_{\odot }$; see e.g., Ho 2002) black holes in their galactic centers and are powered by the infalling matter which forms accretion disks emitting intense optical/UV continuum radiation.

Seyfert galaxies are generally much weaker radio emitters than radio quasars or radio galaxies, but they are not "radio silent." In addition to the diffuse radio continuum originating in the ISM of their late-type hosts, about half of the nearby Seyferts possess compact non-thermal radio cores (Ulvestad & Wilson 1989; Kukula et al. 1995; Ho & Ulvestad 2001), which are often accompanied by arcsecond-scale jets and jet-like features (e.g., Middelberg et al. 2004; Gallimore et al. 2006; Lal et al. 2011). However, the radio cores and jets witnessed in Seyferts are generally very different from those observed in radio quasars or radio galaxies. In particular, radio cores in Seyfert galaxies are characterized by only modest brightness temperatures (see, e.g., Ulvestad & Ho 2001; Ho 2008) and, in general, show no indication of relativistic beaming.

Some Seyfert galaxies also display broad permitted emission lines in their spectra. The presence of such lines anticorrelates with the obscuration of unresolved cores in the optical-to-soft X-ray regime. This fact led to the idea that objects either possessing or lacking broad lines are intrinsically the same, differing only in the orientation of the central engine to the line of sight due to the selective absorption of the core emission by the anisotropically distributed circumnuclear dust (the so-called unification scheme; Antonucci 1993). Hence broad-line Seyferts are called "unobscured" (or "type 1"), while Seyferts without broad emission lines in their spectra are called "obscured" (or "type 2"). One should note that there are several intermediate classes of Seyferts with respect to the nuclear obscuration (types 1.2, 1.5, etc.; Osterbrock 1977), as well as the objects that intrinsically lack broad emission lines (NLS1s; e.g., Pogge 2000; Foschini et al. 2011a).

While optical information is needed for proper classification of an astrophysical source as an AGN, X-ray characteristics are equally important in understanding the physics of central engines in active galaxies. Seyfert galaxies are ubiquitous X-ray emitters (e.g., Ho et al. 2001; Terashima & Wilson 2003; Cappi et al. 2006), and the class is generally known to be particularly bright in the hard X-ray regime (above 10 keV; see, e.g., Tueller et al. 2008; Beckmann et al. 2009). This hard X-ray emission, typically in a form of a power-law continuum (photon indices Γ_X ≃ 2) cutting-off around a few hundred keV (Gondek et al. 1996; Zdziarski et al. 2000), is well understood as being due to optical/UV disk emission reprocessed in the clumpy, hot, but predominantly thermal coronae of accretion disks (see, e.g., Poutanen 1998; Zdziarski 1999 for reviews).

On the other hand, spectral properties of Seyfert galaxies in the γ-ray regime (and especially at high and very high energy γ-rays, i.e., at GeV–TeV photon energy ranges) are basically unknown because of the limited sensitivity of past γ-ray instruments. The upper limits derived using observations by the Imaging Compton Telescope (COMPTEL; Schoenfelder et al. 1993) on board the Compton Gamma-Ray Observatory (CGRO) are consistent with no significant emission component around 1 MeV (Maisack et al. 1995, 1997). Similarly, observations with the Energetic Gamma-Ray Experiment Telescope (EGRET; Thompson et al. 1993) did not result in any detection of Seyfert galaxies (individually, or as a class by means of a stacking analysis) above 100 MeV (Lin et al. 1993; Cillis et al. 2004). One might therefore conclude that, despite some expectations (see Section 5.2) and unlike jet dominated sources (blazars, radio-loud NLS1s, or nearby radio galaxies), Seyferts are particularly "γ-ray quiet." This issue can now be addressed more robustly using the Fermi-LAT instrument simply because of its unprecedented sensitivity to photons on the GeV range.

Fermi-LAT has already discovered or confirmed a number of different classes and types of non-blazar γ-ray emitting AGNs, such as NLS1s (Abdo et al. 2009b), low-power FR I radio galaxies (Abdo et al. 2010c), high-power broad-line radio galaxies (Kataoka et al. 2011), and sources hosting "reborn" compact radio structures (McConville et al. 2011). However, all these targets appear to posses relativistic jets aligned relatively close to the line of sight. Nearby starburst systems have been detected by Fermi-LAT as well (Abdo et al. 2010b; Lenain et al. 2010; Ackermann et al. 2012). However, radio-quiet Seyfert galaxies lacking a circumnuclear starburst have never been significantly detected as γ-ray sources. We note that, parallel to our studies, Teng et al. (2011) have reported their analysis of 491 Seyfert galaxies included in the Swift Burst Alert Telescope (BAT) catalog using 2.1 years accumulation of Fermi-LAT data in the 1–100 GeV energy range. Teng et al. (2011) found only two objects in their sample, NGC 1068 and NGC 4945, to be significantly detected in the 1–100 GeV energy range. Those two sources have already been reported as γ-ray emitters in the first Fermi-LAT Catalog (Abdo et al. 2010b) and discussed in more detail by Lenain et al. (2010), but their GeV emission most likely originates in the ISM of the host galaxies (Ackermann et al. 2012). In this paper, we report on a systematic and detailed investigation of the γ-ray properties of hard X-ray-selected Seyfert galaxies classified as radio-quiet objects, utilizing the three-year accumulation of the Fermi-LAT data from 0.1 to 100 GeV, and report flux limits for individual sources. We also discuss the derived upper limits compared with fluxes in other wavebands for each source. The paper is organized as follows: in Section 2, we discuss the sample selection; the Fermi-LAT data analysis and the results are presented in Sections 3 and 4, respectively; the final discussion of our results are given in Section 5.

Observations in hard X-rays are useful for selecting a complete and unbiased sample of Seyfert galaxies because hard X-ray emission is a clear and common signature of AGN activity, as described in the previous section. By contrast, the optical-to-soft X-ray emission of Seyfert galaxies may be subject to severe obscuration by circumnuclear dust, depending upon the orientation of the source to the line of sight. The BAT (Barthelmy et al. 2005) on board the Swift satellite has provided all-sky survey data in the hard X-ray band with unprecedented high sensitivity, which are well suited for our investigation given the similar observational strategies of Swift-BAT and Fermi-LAT. During the last five years of the Swift-BAT observations, about seven hundred AGNs and galaxies were detected above 15 keV (Baumgartner et al. 2010; Cusumano et al. 2010). Notably, Seyferts outnumber the other classes of AGNs detected in the hard X-ray band.

For this project we have selected a sample of the hard X-ray brightest Seyfert galaxies using the most recent version of the publicly available Swift-BAT 58 month catalog,⁵⁷ restricting the analysis to sources with average 14–195 keV fluxes equal to or greater than 2.5 × 10⁻¹¹ erg cm⁻² s⁻¹ . Such hard X-ray flux selection returns 179 non-blazar type AGNs which are classified as either "galaxies" or "Seyfert galaxies" in the Swift-BAT 58 month catalog. From these, we excluded sources located close to the Galactic plane, specifically those within Galactic latitudes |b| < 10° for the Galactic longitudes |l| > 20°, and |b| < 20° for −20° < l < 20°, because Fermi-LAT sensitivity is reduced toward the Galactic plane due to substantial foreground emission related to the ISM of our Galaxy and presence of numerous Galactic γ-ray emitters (Abdo et al. 2010b). All selected sources are also included in another independent Swift-BAT catalog: the Palermo Swift-BAT 54 month catalog (Cusumano et al. 2010).

The constructed sample is contaminated by several objects with bright relativistic jets such as nearby radio galaxies (e.g., Centaurus A) and radio-loud quasars, which can also be classified also as Seyferts based on their emission line spectral properties in the optical band. All such sources should be removed from the analyzed sample, since those AGNs are physically distinct from "classical" Seyferts. In principle, this could be accomplished by investigating the "radio loudness" parameter for the selected targets, i.e., the ratio between the monochromatic 5 GHz radio and B-band optical fluxes, R_rB ≡ F_5 GHz/F_B. This parameter is often used to distinguish radio-loud (R_rB >10) from radio-quiet (R_rB <10) quasars, according to the criteria proposed by Kellermann et al. (1989), and is widely accepted as a useful proxy for the jet production efficiency. However, such an interpretation holds only if the radio fluxes correspond strictly to the jet emission, and the B-band optical fluxes are mainly due to the accretion disk emission. Both the total optical and radio fluxes in Seyferts can be dominated by host galaxies. If no careful subtraction of the starlight emission is performed, all Seyfert galaxies appear to be radio quiet (with R_rB <10). Yet when the starlight emission is carefully subtracted, many "classical" Seyfert galaxies (especially those accreting at lower rates) formally become radio loud, even if core radio fluxes are used instead of the total radio fluxes, as demonstrated first by Ho & Peng (2001), Ho (2002), and discussed further by Sikora et al. (2007). Another problem is that if one is dealing with a mixture of type 1 and type 2 Seyferts, the intrinsic nuclear optical fluxes may be extremely difficult to determine for the obscured (type 2) objects.

For these reasons, we conclude that the standard definition of the radio loudness parameter is not well suited for our purposes. Instead, we use the "hard X-ray radio loudness parameter," R_rX, defined as the dimensionless ratio of monochromatic radio (1.4 GHz) energy flux density to integrated hard X-ray (14–195 keV) energy flux density,

$$\begin{equation} R_{\rm rX} = \frac{[\nu F_{\nu }]_{\rm 1.4\,GHz}}{F_{\rm 14-195\,keV}} . \end{equation} \tag{ 1 }$$

An analogous X-ray radio loudness parameter was first introduced for Seyfert galaxies and low-ionization nuclear emission-line regions (LINERs) by Terashima & Wilson (2003), and discussed further by Panessa et al. (2007). However, those authors used X-ray data from a lower (medium) photon energy range 2–10 keV, rather than the hard X-ray fluxes considered in this work. Our choice of using the hard X-ray fluxes from the Swift-BAT catalog has an advantage of minimizing the effect of a possible absorption of the X-ray emission in obscured (type 2) objects. At the same time, the typical X-ray photon indices of unobscured Seyfert galaxies Γ_X ≲ 2 within the medium range (e.g., Zhou & Zhang 2010, claiming Γ_X ≈ 1.74 ± 0.02 for the 2–10 keV band) and Γ_X ≳ 2 at hard X-rays (e.g., Ajello et al. 2008, reporting Γ_X ≈ 2.23 ± 0.11 in the 14–195 keV band) imply roughly comparable intrinsic energy flux densities in both X-ray regimes. Therefore, the radio loudness parameters evaluated using the definition introduced here and the definition of Terashima & Wilson (2003) or Panessa et al. (2007) should be roughly equivalent. On the other hand, in the case of very Compton-thick objects with an intrinsic absorption column density N_H of more than 10^24.5 cm⁻², even hard X-ray fluxes in the 14–195 keV are affected by absorption (see, e.g., Gilli et al. 2007); hence the radio loudness parameters provided for such sources have to be taken with caution.

In order to evaluate the radio loudness parameter R_rX for all the analyzed objects, we gather their total radio fluxes from the literature including catalogs such as NRAO VLA Sky Survey (NVSS; Condon et al. 1998), the VLA Faint Images of the Radio Sky at Twenty-cm (FIRST; Becker et al. 2003), or Parkes Catalogue 1990 (PKSCAT90; Wright & Otrupcek 1990) (see Table 1). We use the 1.4 GHz fluxes because the data in this band have much better coverage than at 5 GHz. In the case of sources for which 1.4 GHz fluxes are not available, we use measurements at other frequencies (ν = 0.843 or 4.86 GHz) (Mauch et al. 2008; Miller et al. 1993), and convert those fluxes F_ν to fluxes at 1.4 GHz as [νF_ν]_1.4 GHz = (ν_1.4 GHz/ν)^{1 − α} [νF_ν] assuming a universal radio spectral index α = 0.7 for non-blazar type AGNs. We note that among the analyzed objects there are seven Seyfert galaxies for which radio data are not available in the literature; thus, the radio loudness parameters for these cannot be evaluated.

Figure 1 shows a histogram of the R_rX distribution for the Seyfert galaxies selected from the Swift-BAT 58 month catalog after the flux and position cuts described above (yellow bars in the figure). For comparison, we also plot the distribution of the R_rX parameter derived for classical "radio-loud" AGNs, which are dominated by the beamed emission of relativistic jets. These latter sources are similarly selected from the Swift-BAT 58 month catalog, based on the provided BAT classification (blazar or radio-quasar)⁵⁸ and hard X-ray fluxes ⩾2.5 × 10⁻¹¹ erg cm⁻² s⁻¹ . The selected blazars and radio-quasars (blue bars in Figure 1) are characterized by higher values of the radio loudness parameter and different R_rX distribution when compared to the analyzed population of Seyferts. As indicated in Figure 1, the critical value R_rX = 10⁻⁴ may be used to differentiate between the truly radio-loud and radio-quiet objects, and this cut is applied in our analysis below.

Zoom In Zoom Out Reset image size

Terashima & Wilson (2003) and Panessa et al. (2007) found that R_rB ∼10⁵ R_rX for Seyferts and low-luminosity AGNs using their medium X-ray fluxes. Since the intrinsic energy flux densities in the medium and hard X-ray regimes are expected to be comparable for Seyferts (see discussion above), the hard X-ray loudness parameter value R_rX = 10⁻⁴ roughly corresponds to the "classical" radio loudness parameter R_rB = 10. This simple conversion may not be correct in all cases, though. In particular, in the comparison sample of blazars, there are four objects characterized by the "standard" radio-loudness parameters R_rB >10 but R_rX <10⁻⁴, namely, 2MASS J16561677–3302127, QSO B0033+595, Mrk 421, and QSO B0229+200. Three of them are well-known "high-frequency peaked" BL Lac objects (HBLs) for which the X-ray fluxes are uniquely dominated by the synchrotron emission of highly relativistic jets, and as a result their X-ray-defined radio loudness parameters are low. Those are, however, exceptional objects in the Swift-BAT catalog. On the other hand, for some particularly low-luminosity spiral-hosted AGNs (such as LINERs), the evaluated X-ray loudness parameters are R_rX >10⁻⁴, even though such sources lack relativistic jets. This is simply due to the fact that the total radio emission of such AGNs is heavily dominated by the ISM, and is therefore relatively pronounced, while the total accretion-related X-ray emission is particularly low due to very low accretion rates in their nuclei. As a result, the evaluated X-ray loudness parameters for low-luminosity AGNs accreting at low rates are high (see in this context Terashima & Wilson 2003; Ho & Peng 2001).

In order to check our final sample against contamination by objects containing prominent relativistic jets, first we check 12 sources with relatively high R_rX values: 10^−4.5 < R_rX < 10^−4.0. Among them, eight sources are obscured Seyferts (types 1.8–2), for which relatively high values of R_rX could result from the absorption of the X-ray continuum rather than prominent jet activity. The other four sources are types 1–1.5, namely, Mrk 6, Mrk 1501, NGC 7469, and NGC 4051, for which no prominent relativistic jet is confirmed except for Mrk 1501. Only four sources (Mrk 1501, Mrk 348, NGC 3516, and NGC 7213) among our final sample of 120 sources have counterparts in the CRATES catalog (Healey et al. 2007), which provides a flux-limited all-sky survey of radio core emission. This suggests that most sources in our sample do not have a bright radio core and even the three CRATES Seyferts aside from Mrk 1501 do not display signatures of compact relativistic jets. Therefore, only Mrk 1501 in our sample shows peculiar features, and, in fact, the source is known as a "radio-intermediate" source (e.g., Miller et al. 1993). This galaxy is still worth including in our final sample to address a possible connection between "classical" radio-loud and radio-quiet AGNs. We have thereby confirmed that our sample consists of "radio-quiet" Seyfert galaxies with a single peculiar "radio-intermediate" Seyfert object, Mrk 1501.

Finally, we note that two starburst galaxies, NGC 1068 and NGC 4945, which are at the same time high accretion-rate Seyferts (e.g., Lodato & Bertin 2003), and which have been recently detected by Fermi-LAT (Abdo et al. 2010b; Lenain et al. 2010; Ackermann et al. 2012), do not survive the applied cut in the radio loudness parameter, and therefore are not included in the analyzed sample. Both sources are however established Compton-thick objects, with nuclear hydrogen column densities N_H > 10^24.5 cm⁻² (e.g., Burlon et al. 2011 and references therein). As noted above, the hard X-ray fluxes of such objects are expected to be affected by nuclear obscuration, and as a result their X-ray radio loudness parameters may—when uncorrected for the absorption—formally read as R_rX >10⁻⁴. Yet the GeV emission detected from those two sources most likely originates in the ISM of the galactic hosts, as discussed in detail in Ackermann et al. (2012), even though Lenain et al. (2010) claimed a dominant jet contribution for NGC 1068.

Summarizing, 120 sources are selected for the analysis accordingly to the following criteria.

• 1.  
  Hard X-ray fluxes F_14-195 keV ⩾ 2.5 × 10⁻¹¹ erg cm⁻² s⁻¹ in the Swift-BAT 58 month catalog.
• 2.  
  Spectral classification as "galaxies" or "Seyfert galaxies" in the Swift-BAT 58 month catalog.
• 3.  
  Hard X-ray radio loudness parameters R_rX <10⁻⁴.
• 4.  
  Galactic coordinates |b| > 10° for |l| > 20° and |b| > 20° for −20° < l < 20°.

Table 1 provides source information for the constructed sample of objects including 62 Seyferts of types 1–1.5, 55 Seyferts of types 1.8–2, and three low-luminosity Seyferts classified as "galaxies" in the Swift-BAT catalog. The selected sample includes several radio-quiet NLS1s, such as NGC 4051, NGC 5506, and NGC 7314. We emphasize once more that the applied cut in the hard X-ray-defined radio loudness parameter results in the rejection of not only truly radio-loud AGNs, but also some Compton-thick Seyferts or low-luminosity low-accretion rate AGNs.

Fermi-LAT is a pair-production telescope with large effective area (6500 cm² on axis for >1 GeV photons) and large field of view (2.4 sr at 1 GeV), sensitive to γ rays in the energy range from 20 MeV to >300 GeV. Full details of the instrument, as well as of the on board and ground data processing, are provided in Atwood et al. (2009). Information regarding on-orbit calibration procedures is given in Abdo et al. (2009a). Fermi-LAT normally operates in a scanning "sky-survey" mode, which provides a full-sky coverage every two orbits (3 hr). For operational reasons, the standard rocking angle (defined as the angle between the zenith and the center of the LAT field of view) for survey mode was increased from 35° to 50° on 2009 September 3.

The data used in this work comprise three years of Fermi-LAT observations carried out between 2008 August 4 and 2011 August 5, corresponding to the interval from 239557414 to 334195202 in Mission Elapsed Time (MET). We performed the analysis following the LAT standard analysis procedure⁵⁹ using the LAT analysis software, ScienceTools v9r25v1, together with the P7SOURCE_V6 instrument response functions. We discard events with zenith angles >100° and exclude time periods when the spacecraft rocking angle relative to zenith exceeded 52° to avoid contamination of γ rays produced in the Earth's atmosphere. Events are extracted within a 15° × 15° region of interest (RoI) centered on the position of each object in our sample (listed in Table 2). For our analysis, we accept the events with estimated energies in the range between 100 MeV and 100 GeV.

Table 2. Results of the Fermi-LAT Data Analysis for the Selected Sample of Seyfert Galaxies

Name	R.A.	Decl.	TS	UL: $\mathcal {F}(\rm {>}0.1\,GeV)$	UL2: $\mathcal {F}(\rm {>}0.1\,GeV)$	UL: log Lγ	UL: Lγ/LX
	(deg)	(deg)		(10−9 ph cm−2 s−1)	(10−9 ph cm−2 s−1)	(erg s−1)
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Mrk 1501	2.6292	10.9749	0.0	4.3	2.6	43.6	0.065
NGC 235A	10.7200	−23.5410	9.1	7.7	5.2	42.6	0.070
Mrk 348	12.1964	31.9570	0.0	2.0	1.3	41.6	0.0055
Mrk 1148	12.9783	17.4329	0.0	3.0	1.6	43.1	0.043
Mrk 352	14.9720	31.8269	0.0	3.2	2.1	41.8	0.048
Mrk 1152	18.4587	−14.8456	1.3	3.3	1.9	43.0	0.051
Fairall 9	20.9408	−58.8057	0.0	2.3	1.5	42.7	0.020
NGC 526A	20.9766	−35.0654	0.0	1.4	0.82	41.7	0.010
ESO 297-018	24.6548	−40.0114	2.5	4.8	3.1	42.5	0.030
NGC 788	30.2769	−6.8155	2.7	4.6	2.9	41.9	0.024
Mrk 1018	31.5666	−0.2914	0.7	3.9	2.1	42.8	0.052
NGC 931	37.0603	31.3117	6.4	9.4	5.5	42.1	0.067
NGC 973	38.5838	32.5056	0.0	3.5	1.9	41.8	0.054
NGC 985	38.6574	−8.7876	0.0	2.6	1.8	42.7	0.037
ESO 416-G002	38.8061	−29.6047	8.5	4.9	3.6	43.2	0.082
ESO 198-024	39.5821	−52.1923	0.0	3.2	2.0	42.8	0.049
2MASX J02485937+2630391	42.2472	26.5109	0.0	3.5	2.1	43.1	0.043
MCG-02-08-014	43.0975	−8.5104	0.9	2.5	2.0	41.8	0.041
NGC 1142	43.8008	−0.1836	0.0	1.8	1.1	42.1	0.0082
ESO 417-G006	44.0898	−32.1856	0.0	1.8	1.2	41.6	0.027
NGC 1194	45.9546	−1.1037	6.2	3.9	3.1	41.8	0.045
RX J0311.3-2046	47.8284	−20.7717	0.0	3.3	2.0	43.2	0.052
NGC 1365	53.4016	−36.1404	1.1	4.6	2.6	40.9	0.031
ESO 548-G081	55.5155	−21.2444	0.0	5.2	2.6	42.0	0.054
ESO 549-G049	60.6070	−18.0480	1.1	4.7	2.5	42.5	0.079
UGC 03142	70.9450	28.9718	1.6	9.0	6.3	42.6	0.079
2MASX J04440903+2813003	71.0376	28.2168	7.4	11.9	8.0	42.1	0.085
MCG-01-13-025	72.9229	−3.8094	3.2	3.1	2.3	41.9	0.043
CGCG 420-015	73.3573	4.0616	0.0	3.6	2.1	42.5	0.059
2MASX J05054575-2351139	76.4405	−23.8539	4.2	7.8	4.2	43.0	0.055
CGCG 468-002NED01	77.0820	17.3630	0.0	6.1	3.0	42.2	0.10
IRAS 05078+1626	77.6896	16.4989	0.0	3.1	1.7	42.0	0.015
2MASX J05151978+1854515	78.8324	18.9143	0.0	5.4	3.1	...	0.071
Ark 120	79.0476	−0.1498	0.0	2.4	1.6	42.4	0.016
ESO 362-18	79.8993	−32.6578	7.1	8.0	4.6	42.1	0.068
2MASX J05442257+5907361	86.0941	59.1267	0.0	2.9	1.9	43.1	0.048
NGC 2110	88.0474	−7.4562	0.1	5.2	2.2	41.4	0.0076
MCG+08-11-011	88.7234	46.4393	0.0	3.3	1.9	42.1	0.011
2MASX J05580206-3820043	89.5083	−38.3346	0.0	7.1	3.5	42.9	0.11
ESO 005-G004	91.4235	−86.6319	0.0	2.8	1.6	40.8	0.033
ESO 121-IG028	95.9400	−60.9790	0.0	2.2	1.2	42.6	0.036
ESO 426-G002	95.9434	−32.2166	0.7	5.9	3.4	42.5	0.095
ESO 490-IG026	100.0487	−25.8954	0.1	3.4	2.6	42.3	0.039
2MASX J06411806+3249313	100.3252	32.8254	0.5	4.7	3.3	43.0	0.055
Mrk 6	103.0511	74.4271	0.0	3.7	2.3	42.1	0.026
Mrk 79	115.6367	49.8097	2.5	6.1	3.2	42.5	0.057
2MASX J07595347+2323241	119.9728	23.3901	0.3	3.9	2.2	42.5	0.052
IC 0486	120.0874	26.6135	10.6	9.3	6.3	42.8	0.11
Mrk 1210	121.0244	5.1138	4.4	7.8	4.5	42.2	0.063
Fairall 272	125.7546	−4.9349	0.1	3.3	2.1	42.2	0.030
Mrk 704	139.6084	16.3053	0.0	3.5	2.0	42.5	0.046
MCG-01-24-012	140.1927	−8.0561	0.0	2.3	1.5	42.0	0.024
MCG+04-22-042	140.9292	22.9090	0.0	2.4	1.6	42.4	0.025
Mrk 110	141.3036	52.2863	0.0	3.0	1.9	42.6	0.023
MCG-05-23-016	146.9173	−30.9489	0.0	2.4	1.5	41.2	0.0053
NGC 3081	149.8731	−22.8263	0.0	4.2	2.5	41.2	0.021
ESO 263-G013	152.4509	−42.8112	0.0	5.6	3.6	42.8	0.073
NGC 3227	155.8774	19.8651	3.1	5.2	2.6	41.3	0.020
NGC 3281	157.9670	−34.8537	3.3	8.0	5.1	41.9	0.040
2MASS J10315431-1416514	157.9763	−14.2809	8.9	7.5	5.2	43.8	0.094
NGC 3393	162.0977	−25.1621	8.2	5.8	4.3	42.0	0.098
Mrk 417	162.3789	22.9644	1.2	6.5	3.7	42.9	0.084
NGC 3516	166.6979	72.5686	4.3	6.8	3.4	41.7	0.024
NGC 3783	174.7572	−37.7386	3.3	7.5	4.7	41.4	0.017
UGC 06728	176.3168	79.6815	0.0	2.3	1.4	41.1	0.037
2MASX J11454045-1827149	176.4186	−18.4543	0.6	5.1	2.8	42.8	0.045
NGC 4051	180.7901	44.5313	0.0	2.7	1.7	40.6	0.031
Ark 347	181.1237	20.3162	5.0	6.7	3.1	42.6	0.10
NGC 4138	182.3741	43.6853	1.5	4.3	2.7	40.6	0.061
NGC 4151	182.6357	39.4057	0.0	3.1	2.1	40.3	0.0025
NGC 4235	184.2912	7.1916	0.7	5.3	3.2	41.4	0.073
NGC 4388	186.4448	12.6621	0.0	3.5	2.1	40.7	0.0055
NGC 4395	186.4538	33.5468	0.0	3.3	2.1	39.6	0.055
NGC 4507	188.9026	−39.9093	1.1	6.8	4.1	42.1	0.016
ESO 506-G027	189.7275	−27.3078	0.0	3.0	1.8	42.3	0.014
LEDA 170194	189.7762	−16.1797	0.3	5.3	3.1	42.9	0.052
NGC 4593	189.9143	−5.3443	5.1	7.6	3.6	41.7	0.037
NGC 4686	191.6661	54.5342	0.0	2.0	1.3	41.8	0.032
SBS 1301+540	195.9978	53.7917	0.3	3.9	2.5	42.6	0.049
NGC 4939	196.0600	−10.3396	0.1	4.0	1.9	41.4	0.067
ESO 323-G077	196.6089	−40.4146	0.0a	6.5	5.3	42.1	0.066
			26.7	[Flux] 8.2 ± 2.7b, [Index] Γ = 2.21 ± 0.14b		42.3	0.11
NGC 4992	197.2733	11.6341	0.0	3.9	2.7	42.4	0.030
MCG-06-30-015	203.9741	−34.2956	0.0	4.2	2.4	41.2	0.029
NGC 5252	204.5665	4.5426	0.0	2.7	1.4	42.2	0.010
IC 4329A	207.3303	−30.3094	2.5	7.6	4.6	42.4	0.011
Mrk 279	208.2644	69.3082	3.6	5.1	3.0	42.7	0.050
NGC 5506	213.3119	−3.2075	0.0	4.0	2.4	41.0	0.0071
NGC 5548	214.4981	25.1368	0.0	3.7	2.4	42.1	0.022
ESO 511-G030	214.8434	−26.6447	0.0	2.6	1.6	42.2	0.025
Mrk 817	219.0920	58.7943	0.0	2.4	1.6	42.4	0.038
NGC 5728	220.5997	−17.2532	2.2	7.5	3.9	41.4	0.035
IC 4518A	224.4216	−43.1321	0.6	7.3	3.5	42.4	0.11
Mrk 841	226.0050	10.4378	0.0	31.2	14.9	43.7	0.37
2MASX J15115979-2119015	227.9992	−21.3171	7.3	11.7	7.5	43.4	0.16
2MASX J15144217-8123377	228.6751	−81.3939	5.3	12.2	7.4	43.8	0.17
MCG-01-40-001	233.3363	−8.7005	0.4	6.5	3.4	42.6	0.086
NGC 5995	237.1040	−13.7578	0.0	2.9	2.0	42.3	0.030
Mrk 1498	247.0169	51.7754	1.1	6.0	3.3	43.3	0.061
NGC 6240	253.2454	2.4009	0.0	4.2	2.4	42.4	0.027
NGC 6300	259.2478	−62.8206	5.8	12.2	6.5	41.0	0.054
2MASX J18074992+1120494	271.9580	11.3470	4.7	11.6	6.0	...	0.18
ESO 103-035	279.5848	−65.4276	0.3	2.9	2.3	41.7	0.011
Fairall 51	281.2249	−62.3648	0.5	5.9	3.7	42.1	0.060
ESO 141-G055	290.3090	−58.6703	0.0	2.5	1.8	42.6	0.020
NGC 6814	295.6694	−10.3235	0.0c	6.6	3.8	41.2	0.038
			25.6	[Flux] 16 ± 5d, [Index] Γ = 2.50 ± 0.15d		41.6	0.093
NGC 6860	302.1954	−61.1002	0.0	2.6	1.7	41.8	0.022
Mrk 509	311.0406	−10.7235	0.9	3.5	2.7	42.6	0.016
6dFJ2132022-334254	323.0092	−33.7150	0.0	2.1	1.3	42.3	0.020
1RXS J213623.1-622400	324.0963	−62.4002	4.3	4.8	3.6	43.2	0.071
Mrk 520	330.1724	10.5524	0.0	3.1	2.2	42.3	0.042
NGC 7172	330.5080	−31.8698	2.8	4.5	3.4	41.4	0.011
NGC 7213	332.3177	−47.1667	4.5	6.7	3.5	40.9	0.065
NGC 7314	338.9426	−26.0503	2.5	6.0	3.3	40.9	0.051
NGC 7319	339.0148	33.9757	0.0	2.3	1.5	42.1	0.025
Mrk 915	339.1938	−12.5452	0.0	4.1	2.4	42.4	0.055
MR 2251-178	343.5242	−17.5819	0.0	2.0	1.4	42.9	0.0085
NGC 7469	345.8151	8.8740	0.0	2.4	1.6	41.8	0.015
Mrk 926	346.1811	−8.6857	0.0	2.5	1.5	42.7	0.0094
NGC 7582	349.5979	−42.3706	0.3	5.8	3.5	41.0	0.031
NGC 7603	349.7359	0.2440	0.0	3.3	2.0	42.4	0.029

Notes. Column 1: source name from the Swift-BAT catalog; Column 2: right ascension J2000; Column 3: declination J2000; Column 4: TS of the γ-ray event excess using Fermi-LAT data above 0.2 GeV at the source position; Column 5: 95% C.L. upper limits for the photon flux above 0.1 GeV assuming photon index Γ = 2.5; Column 6: 95% C.L. upper limits for the photon flux above 0.1 GeV assuming photon index Γ = 2.2; Column 7: 95% C.L. upper limits for γ-ray luminosity in a range between 0.1 and 10 GeV based on the results of Column (5), assuming photon index Γ = 2.5; Column 8: upper limits for γ-ray-to-hard X-ray luminosity based on the results of Column (5), assuming photon index Γ = 2.5. ^aThe case in which the association of ESO 323-G077 with 2FGL J1306.9–4028 is actually the result of a chance spatial coincidence. 2FGL J1306.9–4028 is included as a background source in the model. ^bAssuming that ESO 323-G077 is detected by the LAT as 2FGL J1306.9–4028. ^cThe case in which the association of NGC 6814 with 2FGL J1942.5–1024 is actually the result of a chance spatial coincidence. 2FGL J1942.5–1024 is included as a background source in the model. ^dAssuming that NGC 6814 is detected by the LAT as 2FGL J1942.5–1024.

Download table as:  ASCIITypeset images: 1 2 3

Gamma-ray fluxes and spectra are determined by performing a binned maximum likelihood fit of model parameters with gtlike for events binned in direction and energy. The target objects themselves are modeled as point sources with simple power-law photon spectra $d\mathcal {F}/dE = N \times (E/E_0)^{-\Gamma }$. The background model applied here includes standard models for the isotropic and Galactic diffuse emission components.⁶⁰ In addition, the model includes point sources representing all γ-ray emitters within each RoI based on the Second Fermi-LAT Catalog (2FGL; Abdo et al. 2012). We examine the significances of γ-ray signals for the analyzed sources by means of their test statistic (TS) values based on the likelihood ratio test (Mattox et al. 1996). If no significant γ-ray excess above background is detected, we derive a 95% confidence level (CL) upper limit for the integrated photon flux above 100 MeV $\mathcal {F}(>{\rm 100\,MeV}) = \int _{\rm 100\,MeV} dE \, (d \mathcal {F}/dE)$ using the Bayesian method (Helene 1983) with a fixed photon index Γ. Here, we assume two values for the photon index: Γ = 2.5, corresponding to the average photon index for the flux-limited sample of "flat-spectrum radio quasars" included in the 1st Fermi-LAT AGN catalog (1LAC; Abdo et al. 2010a), and Γ = 2.2, corresponding to the typical γ-ray photon index of LAT-detected starburst galaxies (Ackermann et al. 2012). These values should be considered as examples only because spectral properties of Seyfert galaxies around GeV photon energies are unknown.

Our analysis results are summarized in Table 2. We require that two conditions be met in order to claim the detection of γ-ray emission from a target AGN. First, a significant γ-ray excess above backgrounds with TS > 25 must be present at the location of the Seyfert as given in the table.⁶¹ Second, we require a positional coincidence defined here as a target AGN existing within the 95 % confidence localization region of the γ-ray excess. Following these criteria, we did not find any significant γ-ray detections among the 120 Seyfert galaxies in our sample, with possible exceptions of ESO 323–G077 and NGC 6814.

TS values above 25 were obtained at the optically determined locations of ESO 323–G077 and NGC 6814. The γ-ray source 2FGL J1306.9–4028 has been associated with ESO 323–G077 with a probability of 0.8, and 2FGL J1942.5–1024 has been associated with NGC 6814 with 0.91 probability according to the 2FGL catalog and the Second LAT AGN Catalog (2LAC; Abdo et al. 2012; Ackermann et al. 2011). The 2FGL catalog warns, however, that "we expect up to ∼2 false positives among the Seyfert galaxy associations⁶² (cf. Table 8)." We consider two possibilities in this work. First, we analyze the RoI under the assumption that ESO 323–G077/NGC 6814 is detected by the LAT as 2FGL J1306.9–4028/2FGL J1942.5–1024. Second, we consider the case in which the proposed associations are actually the result of chance spatial coincidences. In the second case, we compute a flux upper limit at the position of ESO 323–G077/NGC 6814 with 2FGL J1306.9–4028/2FGL J1942.5–1024 included as a background source in the model for the RoI. The proposed associations may be reinforced by a more precise localization given additional exposure or confirmed by the identification of correlated variability with another waveband. We confirm no significant variability for both 2FGL J1306.9–4028 and 2FGL J1942.5–1024 during the observation period, and find that the spectral shapes are consistent with a simple power law. No blazar in the Roma-BZCAT catalog (Massaro et al. 2009) nor any flat-spectrum radio source in the CRATES catalog (Healey et al. 2007) can be found within 04 of 2FGL J1306.9–4028 and 2FGL J1942.5–1024.

Figure 2 shows the distribution of resulting upper limits for integrated photon fluxes above 100 MeV, $\mathcal {F}(>{\rm 100\,MeV})$. For instance, when we assume a photon index of 2.5, the mean value of the γ-ray upper limit from the analyzed Seyferts is ≃ 4 × 10⁻⁹ photons cm⁻² s⁻¹ , and the upper limits derived for several objects are as low as ≃ 1 × 10⁻⁹ photons cm⁻² s⁻¹ . The mean upper limit found with Fermi-LAT data is therefore more than two orders of magnitude lower than the upper limits derived for the brightest Seyferts based on the SAS 2 and COS B data (Bignami et al. 1979; Pollock et al. 1981, respectively), more than an order of magnitude lower than the analogous EGRET upper limits (0.5–1.5) × 10⁻⁷ photons cm⁻² s⁻¹ (Lin et al. 1993), and close to the lower bound of the effective upper limits from the EGRET stacking data analysis for the brightest 32 Seyfert objects (0.3–1.5) × 10⁻⁸ photons cm⁻² s⁻¹ (Cillis et al. 2004). We note here that Teng et al. (2011) estimated a typical flux upper limit of ∼1 × 10⁻¹⁰  photons cm⁻² s⁻¹ above 1 GeV for a single source. This is consistent with our results covering the bandpass 0.1–100 GeV when re-scaling our γ-ray upper limits to their bandpass, assuming a power-law spectral model with photon index 2.5.

Zoom In Zoom Out Reset image size

5.1. Multiwavelength Comparison

The left panel of Figure 3 compares hard X-ray (14–195 keV) energy fluxes to upper limits for the γ-ray (0.1–10 GeV) energy fluxes⁶³ for the analyzed sample of Seyfert galaxies (denoted in the figure by black open circles), together with a corresponding luminosity–luminosity (L_γ–L_X) plot in the right panel. We discuss the γ-ray results based on the LAT upper limits derived with an assumed photon index of 2.5. Dotted lines in the figures from top left to bottom right denote the ratios between the γ-ray and hard X-ray energy fluxes (or luminosities) F_0.1-10 GeV/F_14-195 keV = 1, 0.1, and 0.01, respectively. The distribution of the ratio between the γ-ray and hard X-ray luminosities is reported in Figure 4. As shown in the figure, for most of the analyzed objects the upper limits for this ratio are below 10 %, and for several particular sources are even below 1 %. The main conclusion here is that our investigation of the Fermi-LAT data indicate that there is no emission component around GeV photon energies in Seyfert objects down to the level of L_γ/L_X < 0.1 in most cases.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It is instructive to locate intriguing targets from the sample in the parameter space of both panels of Figure 3, namely, ESO 323–G077 and NGC 6814, the radio-intermediate quasar Mrk 1501, and the brightest hard-X-ray Seyfert galaxy in the sample NGC 4151. For comparison, two LAT-detected starburst galaxies showing Seyfert activity (NGC 1068 and NGC 4945) are plotted in the figure as well. The multifrequency data together with the Fermi-LAT fluxes for these are taken from Ackermann et al. (2012; see also Table 3). As shown, although NGC 4151 is the brightest hard X-ray source among the analyzed Seyfert galaxies, its intrinsic hard X-ray luminosity is relatively modest, L_X ∼ 10⁴³ erg s⁻¹. Importantly, the γ-ray-to-hard X-ray luminosity ratio for this Seyfert is the lowest among our sample, L_γ/L_X ∼ 0.0025. This can be compared with ESO 323–G077 and NGC 6814, for which the X-ray luminosities in the BAT range are comparable to that of NGC 4151, but for which the luminosity ratios would be L_γ/L_X ∼ 0.11 and 0.093 in the case of the associations with 2FGL J1306.9–4028 and 2FGL J1942.5–1024, respectively. Mrk 1501 is yet a different case, being characterized by a relatively low X-ray flux but high X-ray luminosity, L_X ≳ 10⁴⁴ erg s⁻¹. Indeed, this is the most distant object in the compiled sample. The two starburst galaxies included here for comparison, NGC 4945 and NGC 1068, are characterized by low hard X-ray luminosities, L_X ∼ 10⁴² erg s⁻¹, and γ-ray-to-hard X-ray luminosity ratios L_γ/L_X ∼ 0.1.

Table 3. Starburst Galaxies Discussed in the Paper

Name	dL	$\mathcal {F}({\rm {>}0.1\,GeV})$	log Lγ	F14-195 keV	log LX	log LR	log RrX	log LFIR	log LMIR
	(Mpc)	(10−9 ph cm−2 s−1)	(erg s−1)	(10−11 cgs)	(erg s−1)	(erg s−1)		(erg s−1)	(erg s−1)
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
NGC 253	2.5	12.6 ± 2.0	39.8	...	...	37.77	...	43.38	42.75
M82	3.4	15.4 ± 1.9	40.2	...	...	38.17	...	43.83	...
NGC 4945	3.7	8.5 ± 2.8	40.1	30.10	41.69	38.18	−3.51	43.53	42.68
NGC 1068	16.7	6.4 ± 2.0	41.2	3.38	42.05	39.37	−2.68	44.35	44.44

Notes. Column 1: source name; Column 2: luminosity distance; Column 3: γ-ray photon flux above 0.1 GeV taken from Ackermann et al. (2012); Column 4: γ-ray luminosity above 0.1 GeV taken from Ackermann et al. (2012); Column 5: 14–195 keV energy flux from the Swift-BAT 58 month catalog; Column 6: 14–195 keV luminosity; Column 7: 1.4 GHz radio luminosity; Column 8: hard X-ray radio loudness parameter; Column 9: FIR luminosity at 90 μm from the AKARI–FIS data for NGC 1068, and at 60 μm from the IRAS data for others; Column 10: MIR luminosity at 9 μm from the AKARI–IRC data.

Download table as:  ASCIITypeset image

Fermi-LAT upper limits derived for the analyzed Seyferts can be also compared with infrared fluxes measured by the AKARI satellite. Here we use AKARI 9 μm data (Ishihara et al. 2010) and 90 μm data (Yamamura et al. 2010) with a "good" quality (FQUAL = 3), which are available for 65 and 73 sources from the analyzed sample, respectively. In the left and right panels of Figure 5 we present the corresponding luminosity–luminosity plots, including for comparison, the LAT-detected starburst galaxies NGC 1068, NGC 4945, NGC 253, and M 82 utilizing the Fermi-LAT data analysis presented in Ackermann et al. (2012; see also Table 3). However, AKARI 90 μm data for NGC 4945, NGC 253, and M 82 are flagged as "bad" quality (FQUAL = 1). Hence, for these sources we instead use IRAS 60 μm data (Sanders et al. 2003), which can be considered as comparable to the AKARI 90 μm data according to Yamamura et al. (2010).

Zoom In Zoom Out Reset image size

The far-infrared fluxes of Seyferts galaxies ("FIR"; 90 μm data) are expected to be dominated by thermal dust emission related to the star-forming activity of the galactic hosts, while mid-infrared fluxes ("MIR"; 9 μm data) may originate substantially from circumnuclear dust heated by accretion disk emission, i.e., AGN activity. For most of the analyzed Seyferts the upper limits for the L_γ/L_FIR and L_γ/L_MIR ratios are in the range 0.01–0.1 (cf. black dotted lines in the plots). At the same time, the LAT-detected starburst galaxies are characterized by L_γ/L_FIR ≲ 0.001. This suggests that detection of ISM emission in the GeV photon energy range from the bulk of the hard X-ray selected Seyfert objects—emission analogous to that observed in nearby star-forming galaxies—would require increasing the sensitivity of the Fermi-LAT survey by roughly an order of magnitude. Yet, in the analyzed sample there are also some outliers with particularly high FIR luminosities and Fermi-LAT upper limits low enough to already probe the GeV fluxes close to the expected level of the ISM-related γ-ray emission.

A similar conclusion can be drawn from Figure 6, where we plot radio 1.4 GHz luminosities versus upper limits for the γ-ray luminosities derived for the analyzed Seyferts, also including the comparison sample of starburst galaxies. The thick dotted cyan line represents the best-fit power-law relation between the radio and GeV luminosities for star-forming and local galaxies discussed in Ackermann et al. (2012). The upper limits for the ratio L_γ/L_R in Seyfert sources are on average more than an order of magnitude above the γ-ray-to-radio luminosity ratios characterizing nearby star-forming galaxies. However, the relation between GeV and radio fluxes may not be expected to follow the trend established in star-forming systems for Seyferts in which AGN jet activity could contribute a substantial fraction of the total observed radio flux. Kataoka et al. (2011) argued that the jet-related γ-ray emission of Seyfert galaxies is expected to be below the flux levels probed at present by Fermi–LAT, at least for the majority of sources, and that conclusion is consistent with the upper limits presented in this work: γ-ray-to-radio luminosity ratio L_γ/L_R < 10⁴, on average.

Zoom In Zoom Out Reset image size

The issue of excess radio emission related to the jet activity in Seyfert galaxies may be addressed by looking at the ratio of FIR and radio luminosities for the considered targets, since a relatively tight FIR–radio correlation has been established for non-active (and therefore not jetted) galaxies (see, e.g., Yun et al. 2001). Seyferts with particularly low FIR-to-radio luminosity ratios are likely characterized by prominent jet activity. In Figure 7 we plot L_FIR/L_R versus upper limits for the γ-ray luminosities for the analyzed sample. As expected, there are many hard X-ray-selected Seyferts which are characterized by much lower L_FIR/L_R ratios than those in nearby starburst galaxies. For instance, NGC 4151, which has a relatively prominent pc-scale and kpc-scale jet (Mundell et al. 2003; Ulvestad et al. 2005), shows one of the lowest L_FIR/L_R ratio among the samples. In the case of ESO 323–G077 and NGC 6814—which are not characterized by any outstanding radio or infrared luminosity when compared with the other Seyferts included in the sample—the ratio of the FIR and radio luminosities is very similar to that observed in the LAT-detected starburst galaxies, implying that ESO 323–G077 and NGC 6814 obey the FIR-radio correlation well, and hence that there is not much room for jet activity in these sources. This is in agreement with a non-detection of a compact radio core in ESO 323–G077 by high-resolution very long baseline interferometry (VLBI) radio observations (Corbett et al. 2003), and with the presence of only weak steep-spectrum radio core in NGC 6814 (Ulvestad & Wilson 1984).

Zoom In Zoom Out Reset image size

5.2. Possible γ-ray Emission Components in Seyfert Galaxies

The Fermi-LAT upper limits derived for the Seyfert galaxies in our sample probe the γ-ray luminosity range L_γ/L_X < 0.1, and even <0.01 in some cases. Since hard X-ray luminosity is expected to constitute about 10% of the bolometric AGN-related luminosity of a typical Seyfert galaxy (see Ho 2008), the results indicate that there is no emission component in Seyfert spectra at GeV photon energies down to the level of 1% of the bolometric AGN-related luminosity, or even 0.1% for several objects. The results impose important constraints on any model of high-energy radiation produced by Seyfert-type AGNs.

There are several scenarios discussed in the literature in this context. For example, as noted above, the star-forming activity taking place in the host galaxies of Seyfert objects should result in non-negligible production of γ-rays in the ISM. This inevitable emission component in Seyfert spectra is expected to be analogous to that observed by Fermi-LAT in a few nearby star-forming galaxies (Ackermann et al. 2012), and as such is expected to scale with the FIR and with the diffuse radio luminosities of the host galaxies. However, the flux level probed by the Fermi-LAT in three years of all-sky survey does not allow for the detection of such diffuse emission for the majority of Seyferts, with a possible exception for the most nearby and actively star-forming targets. Teng et al. (2011) also mentioned that the lack of detection of more distant Seyfert galaxies is likely a Fermi-LAT sensitivity issue based on their results of stacking analysis of 215 undetected Seyfert objects. They derived the upper limit in the 1–100 GeV energy range from the stacking analysis to be ∼3 × 10⁴¹ erg s⁻¹ assuming the median redshift of the 215 stacked objects (z ∼ 0.031), but it is still approximately 3 and 18 times the γ-ray luminosities at 1–100 GeV of NGC 1068 and NGC 4945, respectively.

In the previous section, we also commented on a possible contribution of radio jets to the γ-ray emission of Seyfert sources. Unlike the ISM-related γ-ray output, this jet-related emission component in Seyfert spectra at GeV photon energies is a subject of speculation (see, e.g., Lenain et al. 2010; Kataoka et al. 2011). Four AGNs classified as NLS1s have been recently detected by Fermi-LAT and their observed high-energy radiation was established to be due to the jet activity (Abdo et al. 2009b; Foschini et al. 2011b). However, those objects are very different systems from the ones analyzed in this work, possessing flat-spectrum and high-brightness temperature radio cores, and therefore relativistic compact jets resembling blazar sources rather than sub-relativistic outflows observed in radio-quiet Seyferts (e.g., Foschini et al. 2009). Since the targets studied in this paper should be considered as being representative for the whole population of such "classical" Seyferts, our analysis indicates that any jet-related γ-ray emission component in this type of AGN, even if present, is not as prominent as in radio galaxies or blazars.

Another possible emission site of γ-rays in Seyfert galaxies could be disk coronae, where the bulk of observed hard X-ray emission from Seyferts is produced. The first models for such emission involved non-thermal electron populations, and predicted power-law tails in Seyfert spectra extending to at least MeV photon energies (e.g., Zdziarski & Lightman 1985; Svensson 1987). However, detections of spectral cutoffs in the hard X-ray continua around a few hundred keV photon energies in Seyfert galaxies favor emission models involving dominant thermal electron populations (see the discussion in Poutanen 1998; Zdziarski 1999). Still, the available observational constraints do not exclude the presence of non-thermal power-law tails in Seyfert spectra in the MeV range, albeit with a much reduced normalization, constituting not more than ∼10 % of total energy radiated in the X-ray regime (Johnson et al. 1997; Wardziński & Zdziarski 2001; Lubiński et al. 2010). Under this assumption, the observationally allowed luminosity ratio between the MeV (0.1–10 MeV) and the X-ray bands, L_0.1-10 MeV/L_X < 0.1, together with a simple scaling of the 0.1–10 GeV luminosity L_γ = 10^{3 (2 − Γ)} L_0.1-10 MeV, formally implies an expected ratio L_γ/L_X < 0.003 for Γ = 2.5. The current Fermi-LAT sensitivity can hardly probe such levels of γ-ray emission at the moment. In addition, even if the high-energy emission were to originate in or near the accretion disk, the opacity of γ rays to pair production via interaction with X-rays produced by the accretion disk might prevent those γ rays from escaping. Nevertheless, as pointed out by Inoue et al. (2008), the presence of such an emission component at the maximum allowed level would explain the observed extragalactic MeV background radiation in terms of a dominant contribution from Seyfert galaxies while Teng et al. (2011) suggest that the radio-quiet Seyfert galaxies are not a significant source of the extragalactic γ-ray background above 1 GeV based on their analysis results of no γ-ray detection from the radio-quiet Seyfert galaxies at that energy range.

Finally, we discuss a possible mechanism of producing GeV photon in Seyfert galaxies by proton–proton interactions in the innermost parts of their accretion disks. Such a possibility was discussed previously in the context of Galactic black hole systems (Shapiro et al. 1976; Mahadevan et al. 1997; Oka & Manmoto 2003) and was applied recently to the case of active galaxies by Niedzwiecki et al. (2009). Although the current model predictions are still preliminary, this hadronic process was anticipated to result in a significant emission component in the 0.1–10 GeV range, possibly constituting ≳ 10 % of the disk/disk corona X-ray luminosity in the case of a particular (preferred) range of the accretion rate, typically corresponding to advection-dominated ("hot") accretion flow, and of a maximally spinning black hole. That is because for a Kerr black hole the innermost stable orbit of the accretion disk can be located much closer to the event horizon, and hence the number density of the matter within the innermost parts of the accretion disk as well as the proton temperature is increased, leading to enhanced proton–proton interactions above the threshold for the pion production. The Fermi-LAT upper limits derived in this work for the sample of the hard X-ray–brightest Seyfert objects (mostly L_γ/L_X < 0.1, and <0.01 in several particular cases) could be useful to constrain the model parameters and, ultimately, to determine the spin distribution for supermassive black holes hosted by Seyfert-type AGNs. In this context, we find no prominent GeV emission component that could be related to hadronic interactions within accretion flows surrounding Kerr black holes for the whole analyzed sample, with the possible exceptions of ESO 323–G077 and NGC 6814.

As emphasized above, we cannot rule out the possibility that the associations of 2FGL J1306.9–4028 with ESO 323–G077 and 2FGL J1942.5–1024 with NGC 6814 are due to chance spatial coincidences, but if the Seyfert objects are conclusively established as γ-ray emitters, then neither the star forming nor the jet activity in these objects can be considered as origins of the γ-ray emission. Figure 8 shows the broadband spectral energy distribution of ESO 323–G077 (red data points), including the Fermi-LAT spectrum of 2FGL J1306.9–4028 (magenta data points), as well as NGC 6814 (dark green data points), and 2FGL J1942.5–1024 (green data points). For comparison, in the figure we also plot the broadband spectral energy distribution of the starburst galaxy NGC 1068 (blue data points). As shown in the figure and discussed in the previous section, the γ-ray-to-far-infrared luminosity ratio for ESO 323–G077/2FGL J1306.9–4028 is much larger than for NGC 1068, which seems to exclude the possibility that the γ-ray emission from ESO 323–G077/2FGL J1306.9–4028 is attributed to the star-forming activity within the host of the analyzed Seyfert. On one hand, no compact radio core is found in ESO 323–G077 even by high-resolution (<005) VLBI observations, and the resultant upper limit for the radio core emission is 1.3 mJy at 2.3 GHz (Corbett et al. 2002, 2003). This finding challenges the jet hypothesis for the origin of γ rays, if ESO 323–G077 is associated with 2FGL J1306.9–4028. Therefore, either the association of 2FGL J1306.9–4028 with ESO 323–G077 is due to a chance positional coincidence, or ESO 323–G077 is an exceptional source for which the γ-ray radiative output is dominated by an emission component not typically observed among other Seyferts. The same reasoning may be applied in the case of NGC 6814.

Zoom In Zoom Out Reset image size

In this paper, we report on a search for γ-ray emission from a sample of Seyfert galaxies selected via their hard X-ray fluxes, specifically for sources with 14–195 keV fluxes above 2.5 × 10⁻¹¹ erg cm⁻² s⁻¹ as determined using the Swift-BAT 58 month catalog, utilizing the three-year accumulation of Fermi-LAT data. We exclude "radio-loud" objects from the sample by selecting only those sources for which the parameter R_rX—the ratio of νF_ν radio flux at 1.4 GHz frequency to the hard X-ray flux in the 14–195 keV band—is less than 10⁻⁴. The selection criteria leave us with a well-defined sample of 120 "radio-quiet" Seyfert galaxies. The two nearby type-2 Seyferts which are detected by the Fermi-LAT, NGC 1068 and NGC 4945, are not included in the analysis. In a companion paper by Ackermann et al. (2012), we argue that the γ-ray emission of those two sources is more likely attributed to cosmic-ray interactions in the ISM of their host galaxies.

Generally, "radio-quiet" Seyfert galaxies selected by their hard X-ray flux are not detected in the γ-ray band covered by the Fermi-LAT. We report photon flux upper limits for all the sources included in our sample: the typical limit is ∼4 × 10⁻⁹ photons cm⁻² s⁻¹ in the energy range above 100 MeV. We find two possible associations of γ-ray sources with objects in our sample, ESO 323–G077 and NGC 6814, but caution that chance spatial coincidences with these objects cannot be ruled out.

FIR fluxes of the objects considered here, provided by the AKARI satellite, indicate the upper limits for the L_γ/L_FIR luminosity ratios in the range of 0.01–0.1. At the same time, the LAT-detected starburst galaxies are characterized by L_γ/L_FIR ≲ 0.001. This suggests that detection of ISM emission in the GeV photon energy range from the bulk of the hard X-ray-selected Seyfert objects—emission analogous to that observed in nearby star-forming galaxies—would require increasing the sensitivity of the Fermi-LAT survey by roughly an order of magnitude. Similarly, the derived upper limits for the γ-ray-to-radio luminosity ratio, L_γ/L_R < 10⁴ on average, support the conclusion by Kataoka et al. (2011) that the jet-related γ-ray emission of Seyfert galaxies is generally expected to be below the flux levels probed at present by Fermi-LAT.

The resultant Fermi-LAT upper limits yield the ratio of γ-ray to X-ray luminosities L_γ/L_X < 0.1, and even <0.01 in some cases. In general, coronae of accretion disks including non-thermal electron populations can be considered as plausible sites of the γ-ray production. Our analysis allows for the presence of such a broadband power-law emission component extending from MeV to GeV range, but constituting not more than 10% of the thermal radiative output of the disks and disk coronae.

Finally, γ-ray photons may be produced in Seyfert galaxies by proton–proton interactions in the innermost parts of their accretion disks. Although the current model predictions are still preliminary, this hadronic process was anticipated to result in a significant emission component in the 0.1–10 GeV range, possibly constituting ≳ 10% of the disk/disk corona X-ray luminosity in the case of a maximally spinning black hole. The upper limits derived in this paper indicate that no prominent GeV emission component that could be related to the hadronic interactions within accretion flows is found for the whole analyzed sample, with the possible exceptions of ESO 323–G077 and NGC 6814.

The Fermi-LAT Collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the LAT as well as scientific data analysis. These include the National Aeronautics and Space Administration and the Department of Energy in the United States, the Commissariat à l'Energie Atomique and the Centre National de la Recherche Scientifique/Institut National de Physique Nucléaire et de Physique des Particules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Nucleare in Italy, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the High Energy Accelerator Research Organization (KEK) and the Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallenberg Foundation, the Swedish Research Council and the Swedish National Space Board in Sweden. Additional support for science analysis during the operations phase is gratefully acknowledged from the Istituto Nazionale di Astrofisica in Italy and the Centre National d'Études Spatiales in France.

This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

M. H. is supported by the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists. K. B. is supported by a Stanford Graduate Fellowship. We thank the anonymous referee for the valuable comments which helped to improve the paper.

Facilities: Fermi - Fermi Gamma-Ray Space Telescope (formerly GLAST), Swift - Swift Gamma-Ray Burst Mission, AKARI -