A&A 381, 834-840 (2002)
C. Vignali1, - A. Comastri1
Osservatorio Astronomico di Bologna, via Ranzani 1, 40127 Bologna, Italy
Received 31 August 2001 / Accepted 29 October 2001
We present the BeppoSAX broad-band X-ray spectrum of the Seyfert 2 galaxy NGC 3281. The source high-energy spectrum is characterized by the nuclear transmitted component, with an absorbing column density of 21024 cm-2, while the MECS spectrum is reflection-dominated, with a prominent ( -1.2 keV) iron K emission line. The source is detected at only the 5 significance level in the LECS band, because of the strong obscuration which hampers at low energies the direct view of the active nucleus harbored in NGC 3281. BeppoSAX results are consistent with the scenario where NGC 3281 is inclined more than 60 with respect to the line-of-sight. Combining the value obtained from the present X-ray analysis with the measurement, a about 50 times the Galactic value is derived.
Key words: galaxies: Seyferts: individual: NGC 3281 - galaxies: nuclei - galaxies: active - X-rays: galaxies
According to the unified models (e.g., Antonucci 1993), Seyfert 2 galaxies are powered by the same engine of Seyfert 1s but are seen at medium/high inclination angles with respect to the line-of-sight, thus being absorbed by thick matter in the equatorial plane. This picture, at least at its first order approximation, has found strong support from many observational evidences, including optical spectropolarimetry (e.g., Antonucci & Miller 1985; Miller & Goodrich 1990), near-infrared spectroscopy (e.g., Goodrich et al. 1994) and X-ray spectroscopy (e.g., Turner et al. 1997a,b, 1998). For column densities lower than 1024 cm-2, the X-rays above a few keV are able to penetrate the absorbing medium (likely to be associated to the torus) and to be detectable as direct continuum emission (Compton-thin sources). If the absorbing column density exceeds cm-2, then the nuclear radiation is obscured up to about 10 keV and the source is called Compton-thick (see Matt et al. 2000 and Matt 2001 for the most recent reviews). While in a handful of Compton-thick sources the column density has been directly measured (e.g., NGC 6240: Vignati et al. 1999; NGC 4945: Iwasawa et al. 1993; Done et al. 1996; Guainazzi et al. 2000; Madejski et al. 2000; Circinus: Matt et al. 1999a), in the majority of them the column density is so large to totally obscure the nucleus even in the hard X-ray band (e.g., NGC 1068: Matt et al. 1997; NGC 7674: Malaguti et al. 1998) or their flux at high energies is too low to allow a detailed spectral analysis (Salvati et al. 1997; Maiolino et al. 1998).
Emission from Compton-thick galaxies can be detected by means of the reflection caused by the visible inner surface of the torus and/or via scattering by the material responsible for the broad lines observed in polarized light (e.g., Antonucci & Miller 1985; Tran 1995). In the X-ray band, the spectrum of Compton-thick sources is generally characterized by an iron K line with extremely large equivalent width ( keV) and a reflection-dominated continuum flatter than the intrinsic spectrum (e.g., Maiolino et al. 1998). In the last few years, thanks to the high capabilities of observing Compton-thick sources with broad-band instruments like those onboard ASCA (e.g., Matt et al. 1996a; Collinge & Brandt 2000), RXTE (e.g., Madejski et al. 2000), BeppoSAX (e.g., Malaguti et al. 1998; Matt et al. 1999a; Cappi et al. 1999; Vignati et al. 1999; Guainazzi et al. 2000; Franceschini et al. 2000; Iwasawa et al. 2001) and, more recently, XMM-Newton (Guainazzi et al., in preparation), much has be learned about their X-ray properties. However, there are still several astrophysical reasons why Compton-thick sources deserve to be studied in detail. About 50% of the active galactic nuclei (AGNs) in the local Universe are obscured by Compton-thick matter (e.g., Maiolino et al. 1998; Risaliti et al. 1999). Even though their contribution to the X-ray background (XRB) is not energetically dominant (unless the number of such objects is highly underestimated), they may constitute however an important ingredient for the IR and the sub-mm backgrounds, where most of the absorbed radiation is re-emitted by dust (Fabian & Iwasawa 1999; Brusa et al. 2001). Furthermore, Compton-thick sources allow to study those spectral components that otherwise would be dominated by the nuclear continuum emission (Matt et al. 1996b; Bianchi et al. 2001).
In this regard, NGC 3281 is an extremely interesting object. ASCA observation, in a relatively short exposure time (15 ks), revealed a complex X-ray spectrum with a prominent iron line (equivalent width of 480 +770-230 eV) and a large absorbing column density ( cm-2, Simpson 1998). The extinction to the nucleus inferred from the near-infrared studies, mag, combined with the measurement derived from the X-ray analysis, gives a ratio which is more than one order of magnitude larger than the Galactic value ( cm-2 mag-1, assuming E(B-V), Bohlin et al. 1978). Simpson (1998) has explained this finding as due to a dense cloud in the foreground of both the X-ray and the infrared emitting regions that obscures the whole X-ray source but only a fraction of the much larger IR region. There is however increasing evidence from local AGNs (Risaliti et al. 2000; Maiolino et al. 2001a; Severgnini et al. 2001) and optical/X-ray selected quasars (Risaliti et al. 2001; Fiore et al. 2000; Comastri et al. 2001) that the ratio can be much higher than the Galactic value. This ratio may also affect the fraction of nuclear radiation reprocessed and re-emitted by dust at far-infrared and sub-millimeter wavelengths.
The Italian-Dutch satellite BeppoSAX (Boella et al. 1997a) carries four co-aligned Narrow-Field Instruments (hereafter NFI), two of which are gas scintillation proportional counters with imaging capabilities: the Low Energy Concentrator Spectrometer (LECS: 0.1-10 keV, Parmar et al. 1997) and the Medium Energy Concentrator Spectrometer (MECS: 1.3-10 keV, Boella et al. 1997b). The remaining two instruments are the High Pressure Gas Scintillation Proportional Counter (HPGSPC: 4-120 keV, Manzo et al. 1997), which was switched off during the observation, and the Phoswich Detector System (PDS: 13-200 keV, Frontera et al. 1997).
NGC 3281 was observed by BeppoSAX from May 20 to May 23 2000. Standard reduction techniques and screening criteria were applied in order to produce useful linearized and equalized event files. The resulting net exposure times are about 71.3 ks, 13.5 ks and 35 ks for MECS, LECS and PDS, respectively. Spectra from MECS and LECS were extracted from circular regions of radius 4 around the source centroid, while background spectra were extracted from both blank-sky event files and from source-free regions in the target field-of-view, with no apparent difference between the two spectra. The background subtraction in PDS light curves and spectra was performed by plain subtraction of the "off-source'' from the "on-source'' products. The systematic uncertainties of this method are lower than 0.03 s-1 in the full PDS sensitive energy bandpass (Guainazzi & Matteuzzi 1997). The background-subtracted count rates are , and counts per second for MECS (1.3-10 keV band), LECS (0.5-4.5 keV) and PDS (13-100 keV), respectively.
The choice of the extracting regions is such to ensure that the groups of galaxies populating NGC 3281 field-of-view (Ferguson & Sandage 1990) and apparent in MECS image have a very poor contribution (if any) to the source spectrum in the E<10 keV energy range. The PDS band, owing to the soft, presumably thermal spectra characterizing these groups of galaxies, is totally unaffected. A flat-spectrum radio-loud quasar is present in PDS field-of-view (PKS 1030-357). The extrapolation of its radio flux to the X-ray band by means of a power law well matches the ROSAT All Sky Survey (RASS) detection and gives a very poor contribution, of the order of a few 10-3, to the 20-100 keV flux. No known bright source is present in the PDS field-of-view. The probability of a serendipitous source with a flux equal or larger than NGC 3281 is (assuming the ASCA 2-10 keV from Cagnoni et al. 1998).
No significant flux variability has been revealed either in the MECS or in the PDS light curve.
The spectral analysis was carried out through the XSPEC program (version 11, Arnaud 1996). Errors are quoted at 90% confidence level for one interesting parameter ( , Avni 1976), and energies are reported in the source rest frame. The Anders & Grevesse (1989) abundances have been used throughout the paper, as well as a cosmology with H0=70 kms-1 Mpc-1 and q0=0.5 (unless otherwise stated).
NGC 3281 is detected in the LECS at only the significance level, the cause being the short exposure time in the low-energy spectrometer (about 13.5 ks) and presumably the large absorption obscuring the source, which previous ROSAT and Einstein IPC (Fabbiano et al. 1992) observations have already suggested. Due to the low quality of the data in the LECS band, in the following only the MECS and PDS data will be presented. After an introduction to the broad-band X-ray spectrum of NGC 3281 (Sect. 3.1) using the 1.3-100 keV MECS+PDS data, we will focus on the 3-100 keV X-ray properties (Sect. 3.2). Then the results will be discussed and put into a physically more exhaustive scenario (Sect. 4).
To allow for differences in the absolute flux calibration of the MECS and PDS detectors,
the normalizations of the two instruments were allowed to vary within the fiducial values
reported by Fiore et al. (1999).
A column density of
cm-2, due to absorption by the Galaxy
(Dickey & Lockman 1990), is included in all of the models.
|Figure 1: Residuals of the MECS+PDS 1.3-100 keV spectrum for a model consisting of a single power law with and Galactic absorption. Spectral complexity in the soft X-ray band is clearly present, as well as a strong Fe K emission line at about 6.4 keV and an upturn of the spectrum in the PDS band.|
|Open with DEXTER|
|Figure 2: The residuals obtained assuming a transmission plus iron K emission line model.|
|Open with DEXTER|
|Figure 3: The residuals obtained assuming a reflection plus iron K emission line model.|
|Open with DEXTER|
Therefore, owing to the flat spectrum and the presence of a strong Fe K emission line, we tried to fit the overall broad-band spectrum with a pure-reflection model ( PEXRAV in XSPEC, Magdziarz & Zdziarski 1995). Although this model provides a statistically significant improvement in the spectral fit with respect to the transmission model ( with the same number of d.o.f., corresponding to an improvement >99.99% according to the F-test), it still represents a bad parameterization of NGC 3281 X-ray emission. Indeed, large residuals appear evident (Fig. 3), both in the MECS and in the PDS band. An additional absorbed power-law component is likely to be responsible for most of the X-ray emission in the PDS band (see Sect. 3.2).
However, neither the transmission nor the reflection model (or a combination of the two) are able to reproduce the soft ( keV) X-ray spectrum of NGC 3281. The deficit of counts in the 2-4 keV energy range, coupled with the excess of counts below 2 keV (Fig. 1), indicates that a more complex model is required. Thermal emission and/or absorption plus emission from a hot, photo-ionized medium could be relevant for NGC 3281, as well as for Mrk 3, a Compton-thick galaxy which shows a similar spectral behavior (Iwasawa 1995; Cappi et al. 1999). A soft X-ray emission was indeed revealed by ASCA (Simpson 1998) thanks to the low-energy spectral coverage down to 0.5 keV provided by the SIS instrument. Unfortunately, the nature of this component cannot actually be addressed by the present LECS data due to the source low counting statistics in this instrument. It should be noted that soft X-ray components are not uncommon in obscured AGNs, as recently observed in Compton-thick sources (Guainazzi et al. 1999) or derived from the X-ray analysis of a large sample of hard X-ray selected AGNs (Vignali et al. 2001 and references therein).
In the following we will discuss only the 3-100 keV spectrum of NGC 3281, demanding the study of the E<3 keV spectral complexity to other better-suited X-ray instruments, as those onboard Chandra and XMM-Newton.
Supported by the broad-band spectral analysis presented in Sect. 3.1,
a power-law component was added to the transmission plus iron line model
to mitigate the residuals in the MECS band below 5 keV (model (a) in Table 1).
While the nuclear transmitted component is characterized by a slope (
consistent, within the errors, with that typical of AGNs (e.g., Nandra & Pounds 1994; Piro et al. 2000),
the extremely flat
power law below 10 keV (
and the prominent iron line
prompted us to fit the overall broad-band spectrum with the following transmission plus reflection model ((b)
in Table 1)
|(a)||0.78 +0.59-0.27||2.07 +0.32-0.27||1.51 +0.26-0.24||3.42 +5.91-2.00||6.36 +0.06-0.07||2340 +750-710||78.9/65|
|(b)||1.51 +0.20-0.19||2.24 +2.03-1.13||7.41 +2.60-2.04||1180 +400-361||69.7/66|
|(c)||1.98 +0.04-0.21||1.96 +0.19-0.05||5.26 +3.29-0.40||6.84 +0.96-2.04||526 +128-144||83.3/66|
We also tried a different scenario, in which the reflecting matter is partially covered. This would represent
a situation in which the line-of-sight towards the nucleus is heavily blocked due to absorption, but the optical depth diminishes
at higher torus latitudes, thus allowing transmission of some of the radiation reflected by the far side of the torus.
With the present BeppoSAX data, no apparent difference between an unabsorbed reflection and an absorbed one was found
in either the X-ray spectral fits or in the derived parameters.
|Figure 4: BeppoSAX 68, 90 and 99% confidence contours relative to the power-law slope vs. the column density (model (b) in Table 1).|
|Open with DEXTER|
Column densities of the order of 1024 cm-2 imply that the photons which eventually emerge may have been suffered one or more Compton scatterings. Therefore we adopted the model of Matt et al. (1999b), based on Monte-Carlo simulations, which properly treats the photon transfer in very high-column density matter, including the Compton down-scattering and the Klein-Nishina decline, and assuming a spherical geometry for the distribution of the matter (model (c) in Table 1). By this model the column density increases ( cm-2), while both the solid angle subtended by the reflecting matter to the nucleus ( ) and the iron K line equivalent width (500 eV) decrease. This is due to the increased strength of the nuclear transmitted component with respect to the reflected one, as shown in Table 1. Even though the fit obtained with such a model is statistically worse ( ), it must be considered physically more correct than those described above, due to the particular treatment of the photon transfer required in this heavily absorbed environment.
Only a lower limit (60 keV) was derived for the high-energy cut-off. The observed 2-10 keV and 20-100 keV fluxes of NGC 3281 derived from the best-fit model are and erg cm-2s-1, respectively. These fluxes correspond to 2-10 keV and 20-100 keV intrinsic luminosities of and ergs-1, respectively ( and ergs-1 adopting H0=50 kms-1 Mpc-1 and q0=0). The 2-10 keV flux obtained with ASCA data in January 1996 observation (Simpson 1998) is fully consistent with that derived by BeppoSAX.
Thanks to its broad-band capabilities, BeppoSAX has been able to reveal the X-ray spectrum of the Sey 2 galaxy NGC 3281 being more complex than previously derived from ASCA observation (due to its limited bandpass). Below 10 keV, the typical imprints of the reflected continuum (i.e., a flat slope and a moderately strong iron K emission line) do clearly appear, and the presence of a heavily absorbed component is suggested. The reflecting matter appears to be neutral or only marginally ionized, since there is no clear evidence of ionized iron features.
In the PDS band the nuclear, heavily absorbed continuum ( cm-2) is able to emerge and to reveal the true nature of NGC 3281, being a Compton-thick source. Due to the broad-band spectral coverage, both the reflected and the transmitted components have been constrained in a comprehensive picture in which also Compton down-scattering has been taken into account. According to this scenario, the strong absorption is due to transmission through the rim of the torus, while the unabsorbed (or mildly absorbed) reflection component, directly viewed, is due to reprocessing from the torus inner surface. This picture agrees with previous findings for Compton-thick sources (e.g., Matt et al. 2000).
The equivalent width of the iron K line, coupled to the absorption column density, allows to constrain the inclination angle of NGC 3281 with respect to the line-of-sight to be higher than 60 (see Fig. 3 in Ghisellini et al. 1994), assuming typical values for the half-opening angle of the torus and its location in the equatorial plane of the galaxy. A similar value for the inclination angle was suggested by Storchi-Bergmann (1992) from optical observations of this galaxy, in particular from the conical morphology of the high-excitation gas revealed by [O III] emission. It is worth noting that the conical morphology, even though not peculiar of Compton-thick objects, is expected if large amounts of gas hide the nuclear emission, and was seen in many Sey 2 galaxies (Wilson & Tsvetanov 1994). Interesting enough, the [O III] flux ( erg cm-2s-1, Storchi-Bergmann et al. 1992) is about one-third of the 2-10 keV flux derived by BeppoSAX. The ratio is extremely high if compared to that obtained by Mulchaey et al. (1994) for a sample of Seyfert 2 galaxies, whose average observed ratio is 0.02. This provides further support for the hypothesis of obscuration of most of the direct X-rays from NGC 3281 below 10 keV.
The derived ratio is about 50 times higher than the Galactic value, and a factor of 2 higher than that derived by Simpson (1998) from the analysis of ASCA data. The high ratio was originally explained by Simpson (1998) assuming a dense cloud which obscures the entire X-ray source but only partially the more extended infrared emission. An alternative explanation has been recently provided by broad-band studies conducted on AGNs. There is, in fact, growing evidence (e.g., Severgnini et al. 2001; Maiolino et al. 2001a; Risaliti et al. 2001; Vignali 2001) that such high values of are not exceptions, but seem to be a common property of many AGNs at low and high redshift along three orders of magnitude in (from 1021 up to 1024 cm-2, Guainazzi et al. 2001). Maiolino and collaborators (2001b) have tentatively explained this result assuming larger grain sizes and, as a consequence, a flatter extinction curve. On the contrary, this anti-correlation between X-ray absorption and optical reddening seems not to ubiquitously apply to Compton-thick sources (Guainazzi et al. 2001). This result has also cosmological implications. Indeed, if a significant fraction of the X-ray background is not ascribed only to X-ray unabsorbed, broad-line quasars (e.g., Schmidt et al. 1998; Lehmann et al. 2001), but actually also to broad-line, dust-free X-ray obscured quasars (e.g., Comastri et al. 2001; Fiore et al. 2001), then the contribution to the IR background of the sources making the hard XRB would be smaller than previously stated (e.g., Fabian & Iwasawa 1999).
Even though more studies are required to address the above issues, it is however interesting to note that only with a broad-band instrument as the PDS onboard BeppoSAX it is possible to reveal the nuclear X-ray component in heavily absorbed objects, and to constrain the column density in a accurate way. It should be noted, however, that all three models give a fairly good representation of the 3-100 keV spectrum.
BeppoSAX observation of the Seyfert 2 galaxy NGC 3281 has allowed to reveal, for the first time, its Compton-thick nature. The E<10 keV spectrum is reflection-dominated, with a relatively strong ( -1.2 keV) iron K emission line. The PDS band reveals the presence of the heavily absorbed nuclear continuum, the absorption column density being of the order of 1.5- cm-2. This finding also justifies the extremely high value observed in NGC 3281 when compared to the average value for Seyfert 2 galaxies. The ratio is extremely high (about 50 times the Galactic value), thus adding NGC 3281 to the list of AGNs characterized by dust properties different from the Galactic ones.
The present BeppoSAX results fit well into the unified models scenarios for the Seyfert galaxies and highlight the potentialities of the broad-band X-ray spectroscopy with BeppoSAX to study the buried, heavily absorbed Seyferts, as well as to discover new Compton-thick galaxies.
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. CV wishes to thank G. Matt who allows to use his spectral code for transmission and Compton scattering, G. Brunetti for useful discussions, and the referee for helpful comments. The authors gratefully acknowledge support from the Italian Space Agency, under the contracts ASI 00/IR/103/AS and ASI-ARS-99-75, and from the Ministry for University and Research (MURST) under grant Cofin-00-02-36. CV also acknowledges financial support of NASA LTSA grant NAG5-8107.