Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN
P. Severgnini, A. Caccianiga, R. Della Ceca, V. Braito, C. Vignali, V. La Parola, A. Moretti
AAstronomy & Astrophysics manuscript no. ct˙suzaku˙arXiv c (cid:13)
ESO 2018November 6, 2018
Suzaku and SWIFT-BAT observations of a newly discoveredCompton-thick AGN
P. Severgnini, A. Caccianiga, R. Della Ceca, V. Braito, C. Vignali, , V. La Parola and A. Moretti INAF– Osservatorio Astronomico di Brera, via Brera 28, 20121 Milano, Italye-mail: [email protected] Department of Physics and Astronomy, Leicester University, Leicester LE1 7RH, UK Dipartimento di Astronomia, Universita’ degli Studi di Bologna, via Ranzani 1, 40127 Bologna, Italy INAF–Osservatorio Astronomico di Bologna, via Ranzani 1, 40127 Bologna, Italy INAF–Istituto di Astrofisica Spaziale e Fisica Cosmica di Palermo, via U. La Malfa 153, 90146 Palermo, ItalyReceived ...; accepted ...
ABSTRACT
Context.
Obscured AGN are fundamental to understand the history of Super Massive Black Hole growth and their influence on galaxyformation. However, the Compton-thick AGN ( N H > cm − ) population is basically unconstrained, with less than few dozen con-firmed Compton-thick AGN found and studied so far. A way to select heavily obscured AGN is to compare the X-ray emission below10 keV (which is strongly depressed) with the emission from other bands less a ff ected by the absorption, i.e. the IR band. To this end,we have cross-correlated the 2XMM catalogue with the IRAS Point Source catalogue and, by using the X-ray to infrared flux ratioand X-ray colors, we selected a well defined sample of Compton-thick AGN candidates at z < Aims.
The aim of this work is to confirm the nature and to study one of these local Compton–thick AGN candidates, the nearby(z = + N H ) and thus the intrinsic lumi-nosity. Methods.
To this end we obtained deep (100 ks) Suzaku observations (AO4 call) and performed a joint fit with SWIFT–BAT data.We analyzed
XMM-Newton , Suzaku and SWIFT–BAT data and we present here the results of this broad–band (0.4-100 keV) spectralanalysis.
Results.
We found that the broad–band X–ray emission of IRAS 04507 + cm − ) to be well reproduced, thus confirming the Compton–thick nature of this source. In particular, the most probable scenario isthat of a mildly ( N H ∼ × cm − , L(2–10 keV) ∼ × erg s − ) Compton–thick AGN. Key words. galaxies: active – galaxies: individual: IRAS 04507 +
1. Introduction
On the basis of the most accredited X-ray background synthe-sis models (Gilli et al. 1997, Treister et al. 2009, Ballantyneet al 2006), obscured AGN ( N H > cm − ) dominate the en-tire AGN population. For this reason, their space density at dif-ferent redshifts is one of the main ingredients in setting theevolutionary properties of Super Massive Black Hole (SMBH).Unfortunately, the absorption of the obscuring medium (com-posed of gas and dust) along the line of sight does not allowus to easily detect them and study their nuclear properties inthe UV / optical bands and in the soft X-ray bands (energy be-low few keV). This is particularly true for the most obscuredsources, the so called Compton–thick AGN, that are hidden by alarge amount of circum-nuclear obscuring matter along the lineof sight (column density of N H > cm − ).Compton–thick AGN are generally divided in two mainclasses of sources: mildly Compton-thick with N H of the orderof a few times 10 cm − and heavily Compton-thick with N H above ∼ cm − . The primary radiation is strongly suppressedat low energies in case of mildly Compton-thick AGN, emergingonly above 10 keV, while in the case of heavily Compton-thickAGN the primary radiation is strongly depressed also above 10keV, due to the compton down-scattering e ff ect (Matt 1996). Inboth classes of sources, the spectrum below 10 keV is dominated by a pure reflection component (i.e. the continuum emission re-flected by the putative torus), that is much fainter than the di-rect one. For this reason, even for intrinsically luminous objects,it is generally di ffi cult to accumulate enough counts below 10keV to allow us a reliable spectral analysis and thus to assessthe nature of these sources. Furthermore, even when the spectralanalysis is possible, the absorption cut-o ff is only marginally de-tectable below 10 keV (or completely outside the observed en-ergy window). This fact does not allow to correctly measure theintrinsic absorption and thus to estimate the nuclear luminosity.Moreover, as we will show in this paper, in spite of the di ff er-ent values of intrinsic N H , the shape of Compton-thin AGN with5 × cm − < N H < cm − and Compton-thick AGN spectrabelow 10 keV may appear very similar and indistinguishable,specially in cases of low counting statistics. This makes evenmore di ffi cult to adequately probe the heavily obscured AGNpopulation. Often the presence of Compton thick matter is in-ferred through indirect arguments, such as the presence of astrong iron emission line at 6.4 keV; for high values of N H , theequivalent width of this line is expected to be high, reaching val-ues as a few keV (Matt et al. 1996; Murphy & Yaqoob 2009).To properly set the presence of a Compton-thick source andmeasure the amount of absorption, hard X–ray data above 10keV are needed (Comastri et al. 2010). These data allow to detectthe absorption cut-o ff in the very hard X–ray domain typical of a r X i v : . [ a s t r o - ph . C O ] O c t Severgnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN large amount of intrinsic N H ( > cm − ). For the vast majorityof Compton–thick AGN without high energy data, only a con-servative lower limit on the intrinsic column density and nuclearluminosity can be placed. Extreme examples of such limitationsare NGC 6240 (Vignati et al. 1999), Mrk 231 (Braito et al. 2004)and Arp 299 (Della Ceca et al. 2002), where only Beppo-SAXPDS (energy range 15-300 keV) observations were able to revealthe intrinsic AGN emission, allowing the direct measure of the N H and the intrinsic AGN luminosity.A reliable estimate of the number of Compton–thick AGNrequires first of all the use of e ffi cient methods able to se-lect large and possibly complete samples of Compton–thickAGN candidates. As a second step, broad-band X–ray spec-tra (from a few to hundreds keV) are needed to confirm theirCompton–thick nature. However, we lack such a statistical sam-ple and, more importantly, only a limited fraction of the puta-tive Compton thick AGN have been confirmed through broad–band X–ray spectroscopy (see Comastri et al. 2004; Della Cecaet al.2008a, Awaki et al. 2000, Teng et al. 2009, Braito et al.2009). For this reason, indirect arguments are generally used toestimate their density in the local Universe (Della Ceca et al.2008b).In Severgnini et al. (2009) we found that the (2–10 keV)to 24 µ m flux ratio vs. X–ray colors can be used as an e ffi -cient technique to select Compton–thick candidates, at least inthe local Universe. In particular, after having cross–correlatedthe IRAS Point Source Catalog v2.1 (PSC) with the bright end(F − keV > − erg cm − s − ) of the incremental version of the2XMM catalogue (Watson et al. 2009), we found that 85% of the46 extragalactic sources with F(2-10 keV) / ( ν µ m F24 µ m) < > − + + =
71 km s − Mpc − , Ω λ = Ω M =
2. IRAS 04507+0358
This source is optically classified as Seyfert 2 galaxy at z = µ m) lumi-nosity following the prescription reported in Sanders & Mirabel(1996) and Risaliti et al. (2000). We found L IR (cid:39) × ergs − (cid:39) × L (cid:12) , in agreement with the estimate reported inFraquelli et al. (2003). HR4 is defined using the two following bands: 2-4.5 keV and 4.5–12 keV: HR4 = CTS (4 . − keV ) − CTS (2 − . keV ) CTS (4 . − keV ) + CTS (2 − . keV ) , where CTS are the vignettingcorrected count rates in the energy ranges reported in bracket. SeeWatson et al. 2009 for details. XMM-Newton archival data
The 2–10 keV spectrum used in the analysis described here(Observation ID = ∼
11 ksec of good time expo-sure in the MOS camera only) was taken from the
XMM–Newton archive as one of the products of the 2XMM catalogue (Watsonet al. 2009). The data are grouped with a minimum of 20 countsper channel. The X–ray emission of IRAS 04507 + ∼ − erg cm − s − ) could be explained with two for-mally acceptable models: a transmission dominated ( N H < cm − ) and a reflection ( N H >> cm − ) dominated scenario(see Fig. 1). In both cases, two thermal components are requiredto fit the softer part of the spectrum (below 2 keV). We usedtwo mekal models (Mewe et al. 1985, 1986; Liedahl et al. 1995)with the abundances of Wilms et al. (2000), typical of the inter–stellar medium. We found k T = + . − . and k T = + . − . keV(if we use the Compton–thin model) and k T = + . − . keV (ifwe use the Compton–thick model). These thermal componentsare most probably associated to star-forming activity. As dis-cussed in Section 3, we have estimated a star formation rate(SFR) of about 10 M (cid:12) / yr in our source on the basis of the in-frared luminosity. For these values of SFR, the presence of amulti–temperature components in the X–ray spectra is generallyobserved (e.g. M 82, Persic et al. 2004).As for the data above 2 keV, a clear roll-over is present inthe spectrum, which, in the Compton-thin hypothesis, could beinterpreted as a signature of the photoelectric cuto ff . We usedtwo power–laws with the same photon index fixed to Γ= N H = (4 ± × cm − ( χ / d.o.f. = /
52, left panel of Fig. 1). A similarly goodfit could be obtained also by replacing the absorbed power–law with a reflection component ( pexrav model, Magdziarz &Zdziarski 1995). We call this model the ”heavily Compton-thickhypothesis” ( χ / d.o.f. = /
53, right panel of Fig. 1). Even inthis case, we used the same slope ( Γ= =Ω/ π )equal to 1 and the inclination angle to the mean value of 60 ◦ .We kept the abundances of Wilms et al. (2000). Independentlyfrom the assumed model for the underlying continuum, a strongnarrow Fe line is detected. We fixed the energy line at 6.4keV and we found an EW = + . − . keV ( σ = ± = + . − . keV ( σ = + . − . keV) inthe heavily Compton-thick hypothesis.Apart from the strong iron line that could be considered asan indirect hint for the presence of a Compton–thick source,this analysis shows that the XMM-Newton data alone can notdistinguish between a Compton-thin or Compton-thick scenar-ios. With the aim of assessing the nature of this object andestimating its intrinsic properties (e.g. N H and luminosity), wehave obtained 100 ks of Suzaku observation in the A04 call. Wehave combined here this Suzaku data with SWIFT–BAT spec-trum accumulated during the first 54–months of observations(Cusumano et. 2010). IRAS 04507 + evergnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN 3 Fig. 1.
XMM–Newton (MOS1 – black data points in the electronic version and MOS2 – red data points in the electronic version)unfolded spectrum of IRAS 04507 + N H = (4 ± × cm − ) absorber (e). The two power-laws have the same Γ fixed to 1.9. A prominent iron line is also present (d). A similarly good fit could be obtained by using a cold reflection componentinstead of the absorbed power–law (e component in the left panel).of the four X-ray Imaging Spectrometers (XIS, Koyama et al.2007) were working: two front-illuminated (XIS0 and XIS3)and one back-illuminated (XIS1) CCDs. For our analysis weused the 0.4–8 keV and 0.4–10 keV data obtained by XIS1 andXIS0 + XIS3 respectively, combined with the HXD–PIN data.This latter is a non imaging hard X-ray detector (Takahashi et al.2007) covering the 12–70 keV energy band. IRAS 04507 + . The e ff ective expo-sure time after data cleaning are 83.7 ks for each of the XISsource’s and 77.6 ks for the HXD-PIN. XIS spectra: were extracted from a circular region of 2.3arcmin of radius centered on the source. Background spectrawere extracted from two circular regions with the same radius of The screening procedure filter all events within the South AtlanticAnomaly (SAA) as well as with an Earth elevation angle (ELV) < ◦ andEarth day-time elevation angles (DYE ELV) less than 20 ◦ . Furthermorealso data within 256s of the SAA were excluded from the XIS andwithin 500s of the SAA for the HXD. Cut-o ff rigidity (COR) criteriaof > > the source region but o ff set from the source and the calibrationsources. The XIS response (rmfs) and ancillary response (arfs)files were produced using the latest calibration files available andthe ftools tasks xisrmfgen and xissimarfgen , respectively. Thenet count rates observed with the three XIS in the 0.4-10 keVband are 0.034 ± .
001 cts / s (XIS0), 0.048 ± .
013 cts / s (XIS1),0.040 ± .
001 cts / s (XIS3).The spectra from the two front-illuminated CCD were thencombined, while the back–illuminated CCD spectrum was keptseparate and then fitted simultaneously. The net XIS source spec-tra were then binned in order to have a minimum signal–to–noiseratio (S / N) (cid:39) HXD–PIN spectrum:
At the time of writing, two instru-mental background (called non–X–ray background, NXB) fileshave been released, the quick and the tuned one. We tested bothof them and we found that the tuned background count rate inthe 15–70 keV is 5% higher than the count rate of the quickone in the same energy range. To estimate which is the eventfile that provides the most reliable estimate of the real NXB, wecompared them with the data taken during periods of Earth oc-
Severgnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN cultation in the same energy range, 15-70 keV. Since the Earthis known to be dark in hard X–rays and during the Earth oc-cultation we do not measure the emission of the source and ofthe cosmic X–ray background, the occulted data should give agood representation of the actual NXB rate. Thus, after havingcorrected the Earth occultation data for the dead-time and takeninto account only the events with an Earth elevation angle lowerthan -5 ◦ , we extracted the Earth occultation spectrum and light–curve in the 15–70 keV an 15–50 keV, respectively.The comparison between the occulted data and the quick (bkgA) and tuned (bkgD) backgrounds is shown in Fig. 2. The tuned background level is systematically higher than the quick background level and than the Earth occultation level. This dif-ference becomes larger at E >
35 keV. The light-curve compar-ison is shown in Fig. 3. The tuned background spectrum andlight-curve are nearly 7% above the data taken during the Earthoccultation.As a final check, we compared Suzaku flux with Swift-BATdata. We found that with the quick background, the Suzaku fluxis close to the Swift one, i.e. F(15–70 keV)
S uzaku = × − ergcm − s − and F(15–70 keV) BAT = × − erg cm − s − (assum-ing the same model for the two instruments). We thus decidedto use the quick background file and combined it with the cos-mic X–ray background, the latter was parametrized by us usingthe prescription of Boldt (1987) and Gruber et al. (1999). Thebackground-corrected count rate in the 15 −
70 keV is 0.033 cts / s( ∼ The SWIFT–BAT data were processed using the Bat Imagersoftware (Segreto et al. 2010). IRAS 04507 + / N ∼
12 in the 14-150 keV energy range. In order to pro-duce the spectrum of IRAS 04507 + The XIS, the HXD–PIN and the BAT spectra have been fittedsimultaneously covering a wide energy range (0.4–100 keV)using Xspec version 12.5.0. We tied together XIS, HXD andBAT parameters. We left free to vary the BAT / XIS cross-normalization, while for the HXD / XIS instruments we assumeda cross–calibration of 1.18 (Manabu et al. 2007; Maeda et al.2008 ). In all the models described here we have used the abun-dance of Wilms et al. (2000) and the Galactic hydrogen columndensity along the line of sight (from Dickey & Lockman 1990), N H (Galactic) = × cm − .Since the aim of this work is to explore the physical proper-ties of the nuclear regions of IRAS 04507 + N H of the obscuring matter and the de-absorbed X–ray luminos- http: // / suzaku / doc / suzakumemo / suzakumemo-2007-11.pdf;http: // / suzaku / doc / suzakumemo / suzakumemo-2008-06.pdf Fig. 2.
Comparison between the quick (bkgA, red triangles) andthe tuned (bkgD, green crosses) backgrounds with the Earth oc-cultation spectra (black squares). The BkgD spectrum is aboveboth the bkgA and the Earth data. The tuned background clearlyoverpredicts the real NXB, in particular above 35 keV.
Fig. 3.
Comparison between light curves extracted in the 15-50keV energy range. The Earth light curves are corrected for thedead time. In the upper panel we show the di ff erence betweenthe Earth and the tuned background (bkgD) light curves, whilein the lower panel we show the di ff erence between the Earth andthe quick background (bkgA) light curves.ity of the AGN), a detailed analysis of the spectrum below 2keV will be not discussed here. Starting from the results ob-tained with XMM–Newton data, we have fitted this part usingtwo thermal components. We left the temperatures free to varyand we found that the two temperatures are in good agreementwith those found with
XMM-Newton data: k T = + . − . and k T = + . − . . evergnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN 5 Fig. 4.
Suzaku XIS-FI (black data points in the electronicversion), XIS-BI (red in the electronic version), HXD–PIN(green in the electronic version) and SWIFT–BAT (blue in theelectronic version) data of IRAS 04507 + Upper panel:
Unfolded Compton–thin model. The model consists of: two ther-mal emission components ( a and b components), a scatteredcomponent ( c ), a Fe emission line at ∼ d ) and an ab-sorbed power–law with a N H fixed to 4 × cm − ( e ). Lowerpanel:
Ratio between data and the Compton–thin best–fit model.As for the
XMM-Newton spectra, to fit the data above 2keV, we used a simple basic model for obscured AGN, i.e. twopower–laws with the same slope. One of the two power–lawsis absorbed only by Galactic column density and representsthe scattered component; the second power-law represents theprimary X–ray emission seen through the cold absorber (theputative torus). To reproduce this absorbed power–law, we usedthe model by Yaqoob 1997 ( plcabs in Xspec), which partiallytakes into account for the Compton down-scattering. The Feemission line was modeled with a Gaussian profile.
Compton-thin hypothesis:
As a first step, we tested theCompton–thin hypothesis by fixing the slope of the power–lawsto Γ= N H = × cm − , i.e.the values found with the XMM–Newton data (see Fig. 4, χ / d.o.f. = / Fig. 5.
Residuals, with respect to a single power-law, of theXIS data / model (XIS-FI - black data points in the electronicversion and XIS-BI-red points in the electronic version) ofIRAS 04507 + Γ= + . − . and N H = + . − . × cm − ( χ / d.o.f. = / Mildly Compton-thick hypothesis:
To account for theprominent Fe emission line in the spectrum and for the excessdetected above 10 keV, we added to the model a component,which represents the emission reflected from neutral materialinto our line of sight. In particular, we added the pexrav modelleaving open the possibility that the reflected emission could bepartially obscured by the torus itself. We fixed the inclinationangle to its default value (i = ◦ ) and the normalization equalto that of the intrinsic powerlaw. With the addition of this newcomponent, that we found to be absorbed ( N H = (3.3 ± . × cm − ), the model provides a relatively good fit to the 0.4–100keV spectrum ( χ / d.o.f. = / / XIScross-normalization of 1.1 + . − . . The primary X–ray continuumhas an intrinsic slope of Γ= ± N H = (1.45 ± . × cm − . We found that the scattering frac-tion is less than 1%, and the reflection fraction is R = + . − . . Theslightly steep photon index, which is also not well constrained(ranging from 1.9 to 2.9), is not so striking if we take into ac-count the complexity of the model.The Suzaku data confirm the presence of a Fe K α line at6.37 ± σ<
50 eV), but with a lower equivalent width(EW = + − eV) with respect to the value measured with the XMM–Newton data, although consistent considering the errors.In order to investigate the presence of other possible features, weinspected the residuals around 6.4 keV with respect to a singlepower-law model (see Fig. 5); there are clear residuals, both redand blue–wards the K α line. Some excesses in the data / modelratio are present around 5.8 keV (observer’s frame) and between6.5 and 7 keV, while at E > Severgnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN the first excess is most probably due to a bad subtraction of theemission lines present in the calibration source (i.e. Mn K α α σ<
50 eV)line centered at ∼ ∼
100 eV) could beactually present (see Fig. 6) even if it is not statistically required( χ / d.o.f. = / α emission line (rest-frame energy E = β emission line (rest-frame energy E = α line, as it is expected for the Fe K β emission line (see e.g. Leahy & Creighton 1993).As reported in Table 1, the best–fit values obtained withboth models (one and two emission lines) are consistent withinthe errors. We refer to these models as mildly Compton-thickAGN models. In Table 1, the lines at ∼ ∼ and E respectively. The 2–10 keV flux, correctedfor the Galactic absorption along the line of sight, is F(2–10keV) = × − erg cm − s − and, once corrected also for theamount of intrinsic absorption, the deabsorbed 2–10 keV lumi-nosity of the AGN is 7 × erg s − .To further test the above models, we modeled the two com-ponents associated with cold reflection (the 6.4 keV Gaussianemission and the Compton-scattered continuum from neutralmaterial) using a single and self-consistent model: reflionx (Ross& Fabian 2005). In this model, an optically thick disc is illumi-nated by a power–law, producing florescence lines and contin-uum emission. The parameters include the Fe abundance, ion-ization parameter ξ (defined as ξ = π Ftot / n H , where Ftot isthe total illuminating flux and n H is the density of the reflec-tor) and the incident power-law photon index Γ . Being the ob-served Fe iron K α emission line unresolved in the Suzaku spec-trum, no additional velocity broadening was applied to the re-flected spectrum. We found that the data are well reproduced bythis model ( χ / d.o.f. = / N H = (2.7 ± . × cm − , while the transmitted component isabsorbed by a larger column density N H = (1.3 ± . × cm − .We found a value for the photon index that is consistent with theprevious ones ( Γ= ± ξ = + − erg cm s − (the lowest value allowed by the model is ξ = − ), in agreement with iron atoms typically in a low–ionization state corresponding to Fe I–XVII. Best fit parametersare reported in Table 1 (mildly Compton-thick AGN models - reflionx ).Finally, we checked the self-consistence of the mildlyCompton–thick hypothesis by using also the recent paper pub-lished by Murphy & Yaqoob (2009). They calculated Green’sfunctions that may be used to produce spectral-fitting routines tomodel the putative neutral toroidal X-ray reprocessor in AGNsfor an arbitrary input spectrum. In their calculation the repro-cessed continuum and fluorescent line emission due to Fe K α , FeK β and Ni K α are treated self-consistently. On the basis of theEW obtained for the Fe K α emission line ( ∼
450 eV) by our anal-ysis and by assuming an inclination angle between the observer’sline of sight and the symmetry axis of the torus larger than60 ◦ , the model predicts an N H (cid:39) × cm − . The mildlyCompton–thick hypothesis is thus fully supported also by this recent model proposed in the case of neutral toroidal X-ray re-processor in AGNs. Heavily Compton-thick hypothesis:
For completeness, wetested also the heavily ( N H > cm − ) Compton–thick AGNhypothesis. In this case, the emission is completely dominatedby the scattered component at low energies and only by the re-flected component ( reflionx or pexrav model) at high energies.By using disc-reflection models ( reflionx or pexrav models), wefound that this scenario is statistically acceptable. As an exam-ple, we report in Fig. 7 and Table 1 the results found with the reflionx model ( χ / d.o.f. = / reflionx in Table 1). However, an intrinsic columndensity higher than 10 cm − is not supported by the model ofMurphy & Yaqoob (2009) for Fe K α EW of order of 450 eV.Moreover, on the basis of this model, the observed luminosityshould be just less than few percents of the intrinsic one, by im-plying an intrinsic luminosity of L(2–10 keV) ≥ erg s − . Thisvalue of X–ray luminosity exceeds the infrared luminosity mea-sured for this source (L IR = × erg s − ), which, under theassumption that most of the optical and ultraviolet radiation isabsorbed by a dusty torus surrounding the nuclear source, is agood proxy of the bolometric luminosity. Therefore, on the ba-sis of the Murphy & Yaqoob (2009) model, this scenario is notacceptable from the physical point of view.
3. Discussion and Conclusion
IRAS 04507 + XMM-Newton andXIS data below 10 keV alone, we could not distinguish be-tween Compton-thin or Compton-thick scenarios. However, theCompton-thin model clearly under-predicts higher energy data.In particular, we found that the most favorite scenario is thatof a mildly Compton–thick AGN with N H = × cm − and L(2–10 keV) = × erg s − . This luminosity is in fullagreement with the observed infrared luminosity and with thatpredicted for the torus by Fraquelli et al. (2003). These authors,after having estimated the rate of the ionizing photons emittedby the AGN and assuming an opening angle of 30 ◦ for the ion-ization cones, predicted an infrared luminosity for the torus ofL IR = × erg s − . This value is about 20% of the infraredluminosity (L IR (cid:39) × erg s − , see also Sect. 2). The remain-ing fraction of the infrared luminosity (2.2 × erg s − ) is mostprobably due to star-formation activity and it is roughly in agree-ment with the soft X–ray luminosity of the thermal componentsof our source (L(0.5–2 keV) = × erg s − ) considering theL(soft–X)-L(FIR) relation presented in Persic et al. (2004). Toestimate the relevant SFR of our source, we adopted the follow-ing Kennicutt et al. (1998) relation: SFR = (L FIR / × L (cid:12) ) M (cid:12) yr − and we considered only the infrared luminosity produced bystar–formation activity ( ∼ × erg s − ). We found a SFR ofabout 10 M (cid:12) / yr.As for the heavily Compton–thick hypothesis, by using disc–reflection models it is possible to well reproduce the broad–bandspectrum of IRAS 04507 + evergnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN 7 Table 1.
Best–fit values of the mildly and heavily Compton-thick AGN models
Mildly Compton-thick AGN models
Model Γ N H Refl. Scatt. E σ E σ χ / d.o.f. L(2–10 keV)[10 cm − ] frac. frac. keV eV keV eV 10 erg s − One line 2.4 ± ±
10 0.7 + . − . <
1% 6.37 ± . <
50 – – 265.5 /
190 70Two lines 2.4 ± + − + . − . <
1% 6.37 ± . <
50 6.92 + . − . <
50 255.9 /
188 70Model Γ N H ξ E σ χ / d.o.f. L(2–10 keV)[10 cm − ] erg cm s − keV eV 10 erg s − Reflionx 2.5 ± ±
30 29 + − + . − . <
50 226.3 /
190 50
Heavily Compton-thick AGN model
Model Γ N H ξ E σ χ / d.o.f. L(2–10 keV)[10 cm − ] erg cm s − keV eV 10 erg s − Reflionx 2.2 ± . > a + − + . − . <
50 218.9 / > a This value of N H is not an output of the fit, but it is an assumption that we made in this model. We note that this source could not be found as a possi-ble Compton-thick AGN by using only the data below 10 keV,nor by using the two-dimensional diagnostic tool for reflection-dominated Compton-thick object proposed by Bassani et al.(1999). This latter diagram shows that Compton-thick AGNshould be characterized by high K α iron line equivalent width(EW >
300 eV) and by a 2-10 keV flux normalized to the [OIII]optical-line flux (T parameter) typically lower than 0.5. In thisestimate, the [OIII] optical-line flux should be corrected forGalactic and intrinsic extinction. From the X–ray analysis per-formed on IRAS 04507 + α line anEW of ∼
450 eV. In order to estimate the T parameter, we con-sidered the [OIII] line flux (already corrected for the Galacticextinction) reported by Cid Fernandes et al. (2001). After thecorrection for the intrinsic extinction , we found T =
2, largerthan the typical value adopted to select Compton-thick candi-dates ( << / or diagnostic diagrams based on the Fe K α line EW vs. 2–10 keV to [OIII] flux ratio, a significant fraction of mildly orheavily Compton–thick AGN could be missed. Whereas someexamples of Compton–thick AGN with low value of Fe K α line equivalent width are indeed already present in the literature(e.g. Awaki et al. 2000; Ueda et al. 2007; Braito et al. 2009),IRAS 04507 + / F24 µ m vs. HR diagnostic diagram, also shows the impor- For the intrinsic extinction, we applied the following formula:Fcor[OIII] = Fobs[OIII][(H α / H β )obs / (H α / H β ) ] . (Bassani et al.1999), where Fcor[OIII] is the extinction-corrected flux of [OIII]5007,and Fobs[OIII] the observed flux of [OIII]5007. We used and observedline ratio of H α / H β = α / H β ) = tance of using a wide X–ray spectral coverage in order to con-strain the intrinsic column density in this type of sources even inthe case of mildly Compton-thick AGN.With the aim of establishing the nature of other Compton-tick AGN candidates found through our diagram, we have re-cently obtained in the Suzaku AO5 call 100 ksec of observationfor MCG-03-58-007, while the broad–band X–ray analysis forother twelve sources with a SWIFT–BAT counterpart is alreadyongoing and will be the subject of a forthcoming paper. The fi-nal aim is to better constrain and define the Compton–thick AGNpopulation and hence to better estimate their space density. Acknowledgements.
The authors acknowledge financial support from ASI(grant n. I / / /
0, COFIS contract and grant n. I / / / / his useful comments. References
Awaki, H. Ueno, S., Taniguchi, Y., 2000, AdSpR, 25, 797Awaki, H. Terashima, Y., Higaki, Y., Fukazawa, Y. 2009 PASJ, 61, 317Ballantyne, D. R., Everett, J. E., Murray, N. 2006, ApJ, 639, 740Bassani, L., Dadina, M., Maiolino, R., et al. 1999, ApJS, 121, 473Boldt, E. 1987, Phys. Rep., 146, 215Braito, V., Della Ceca, R., Piconcelli, E., et al. 2004, A&A, 420, 79Braito, V., Reeves, J. N., Della Ceca, R., et al. 2009, A&A, 504, 53Caccianiga, A.,Severgnini, P., Braito, V., et al. 2004, A&A, 416, 901Cid Fernandes, R., Heckman, T., Schmi tt, H., Gonz´alez Delgado, R. M., Storchi-Bergmann, T. 2001, ApJ, 558, 81Comastri, A. 2004, in Astrophys. Space Sci. Libr. 308, Supermassive BlackHoles in the Distant Universe, ed. A. J. Barger, 245Comastri, A., Iwasawa, K., Gilli, R., et al. 2010, ApJ, 717, 787Cusumano, G., La Parola, V., Segreto, A., et al. 2009, AIPC, 1126, 104Cusumano, G., La Parola, V., Segreto, A., et al. 2010, A&A in press[arXiv:1009.0522]Della Ceca, R., Ballo, L., Tavecchio, F., et al. 2002, ApJ 581, 9Della Ceca, R., Caccianiga, A., Severgnini, P., et al. 2008, A&A 487, 119Della Ceca, R., Severgnini, P., Caccianiga, A., et al. 2008, MmSAI, 79, 65Dickey, J. M., Lockman, F. J. 1990, ARA&A, 28, 215Fraquelli, H. A., Storchi–Bergmann, T., Levenson, N.A., 2003, MNRAS, 341,449Gilli, R., Comastri, A., Hasinger, G. 2007, A&A, 463, 79Gruber, D. E., Matteson, J. L., Peterson, L. E., Jung, G. V. 1999, ApJ, 520, 124Kennicutt, R. C. Jr. 1998, ApJ, 498, 541Koyama, K., Tsunemi, H., Dotani, T., et al. 2007, PASJ, 59, 23Leahy D.A., Creighton J. 1993, MNRAS, 263,314
Severgnini et al.: Suzaku and SWIFT-BAT observations of a newly discovered Compton-thick AGN
Fig. 6.
Suzaku XIS-FI (black data points in the electronicversion), XIS-BI (red in the electronic version), HXD–PIN(green in the electronic version) and SWIFT–BAT (blue in theelectronic version) data of IRAS 04507 + Upper panel:
Unfolded mildly Compton–thick model. The spectral compo-nents are: a) and b) two thermal emission components; c) scat-tered AGN component; d) narrow Gaussian line at 6.37 keV;e) absorbed power–law AGN component; f) pure reflectionAGN component; g) narrow Gaussian line at ∼ Lowerpanel:
Ratio between data and the mildly Compton–thick best–fit model.
Liedahl, D. A., Osterheld, A. L., Goldstein, W. H. 1995, ApJ, 438, L115Magdziarz, P., Zdziarski, A. A. 1995, MNRAS, 273, 837Mateos, S., Carrera, F. J., Page, M. J., et al. 2010, A&A, 510, 35Matt G., 1996, Proc. of the conference ”Roentgenstrahlung from the Universe”,MPE Report n.263, p.479Mewe, R., Gronenschild, E. H. B. M., van den Oord, G. H. J. 1985, A&AS, 62,197Mewe, R., Lemen, J. R., van den Oord, G. H. J. 1986, A&AS, 65, 511Mitsuda, K., Bautz, M., Inoue, H., et al. 2007, PASJ, 59, 1Murphy, K. D., Yaqoob, T.,2009, MNRAS, 397, 1549Page, K. L., Reeves, J. N., O’Brien, P. T., Turner, M. J. L., Worrall, D. M. 2004,MNRAS, 353, 133Persic, M., Rephaeli, Y., Braito, V., et al. 2004, A&A, 419, 849Reeves, J. N., Turner, M. J. L. 2000, MNRAS, 316, 234Risaliti, G., Gilli, R., Maiolino, R., Salvati, M., 2000, A&A, 357, 13Ross R. R., Fabian A. C., 2005, MNRAS, 358, 211Sanders, D. B., Mirabel, I. F., 1996, ARA&A, 34, 749Segreto, A., Cusumano, G., Ferrigno, C., et al. 2010, A&A 510, 47
Fig. 7.
Suzaku XIS-FI (black data points in the electronic ver-sion), XIS-BI (red in the electronic version), HXD–PIN (greenin the electronic version) and SWIFT–BAT (blue in the elec-tronic version) data of IRAS 04507 + Upper panel:
Theunfolded heavily Compton–thick model ( reflionx , see Table 1)is overplotted on the data. The spectral components are: a) andb) two thermal emission components; c) scattered AGN com-ponent; d) narrow Gaussian line at ∼ reflionx ). Lower panel: