Table Of ContentMon.Not.R.Astron.Soc.000,000–000 (0000) Printed1February2008 (MNLATEXstylefilev1.4)
BeppoSAX Observations of 2-Jy Lobe-dominated Broad-Line
Sources. I. The Discovery of a Hard X-ray Component
⋆
Paolo Padovani1,2,3, , Raffaella Morganti4, Joachim Siebert5,
Fausto Vagnetti3, Andrea Cimatti6
1 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore MD. 21218, USA
2 Affiliated to the Astrophysics Division, Space Science Department, European Space Agency
3 Dipartimento di Fisica, II Universit`adi Roma “TorVergata”, Viadella Ricerca Scientifica1, I-00133 Roma, Italy
9
4 Istituto di Radioastronomia, Via Gobetti 101, I-40129 Bologna, Italy
9
5 Max-Planck-Institut fu¨r Extraterrestrische Physik, Giessenbachstrasse, D-85740 Garching bei Mu¨nchen, Germany
9
6 Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, I-50125 Firenze, Italy
1
n
a
Accepted ,Received
J
1
1
ABSTRACT
1
We present new BeppoSAX LECS, MECS and PDS observations of five lobe-
v
dominated,broad-lineactivegalacticnucleiselectedfromthe 2-Jysampleofsouthern
9
radio sources. These include three radio quasars and two broad-line radio galaxies.
2
1 ROSAT PSPCdata,availablefor allthe objects,arealsousedto better constrainthe
1 spectralshapeinthesoftX-rayband.Thecollecteddatacovertheenergyrange0.1−10
0 keV, reaching ∼50 keV for one source (Pictor A). The main result from the spectral
9 fits is that all sourceshaveahardX-rayspectrumwith energyindex αx ∼0.75inthe
9 2−10 keV range. This is at variance with the situation at lower energies where these
h/ sourcesexhibitsteeperspectra.Spectralbreaks∆αx ∼0.5at1−2keVcharacterizein
p facttheoverallX-rayspectraofourobjects.Theflat,high-energyslopeisverysimilar
- to that displayed by flat-spectrum/core-dominated quasars, which suggests that the
o
same emission mechanism (most likely inverse Compton) produces the hard X-ray
r
t spectra in both classes. Finally, a (weak) thermal component is also present at low
s
energies in the two broad-line radio galaxies included in our study.
a
:
v Key words: galaxies: active – X-ray: observations
i
X
r
a
1 INTRODUCTION radio galaxies (BLRG) have a still uncertain place. They
couldrepresenteitherobjectsintermediatebetweenquasars
There is abundant evidence that strong anisotropies play
and radio galaxies, (i.e., with the nucleus only partly ob-
a major role in the observed characteristics of radio loud
scuredandthebroademissionlinesjustbecomingvisibleat
active galactic nuclei (AGN; see Antonucci 1993 and Urry
the edge of the obscuring torus) or low-redshift, low-power
&Padovani1995forareview).Radiojetsareinfactknown
equivalent of quasars.
tobestronglyaffectedbyrelativisticbeaming,whilepartof
The scenario described above makes a number of pre-
theopticalemissioninsomeclassesofobjectsislikelytobe
dictionsabouttheX-rayemission oftheseradio-loud AGN.
absorbedbyathickdiskortorusaroundtheactivenucleus.
Moreover, the hard X-ray band, that is less affected by ab-
A unification of all high-power radio sources has been
sorption,isessentialforacompleteknowledgeoftheintrin-
suggested (Barthel 1989; Urry & Padovani 1995 and ref-
sic natureof these objects.
erences therein) and according to this scheme, the lobe-
AlthoughtheX-rayspectrumcanbeverycomplex,the
dominated, steep-spectrum radio quasars (SSRQ) and the
presence of a nuclear, likely beamed X-ray component in
core-dominated, flat-spectrum radio quasars (FSRQ) are
quasarsisquitewellestablishedinparticularforFSRQand
believed to be increasingly aligned versions of Fanaroff-
blazars(Wilkes&Elvis1987;Shastrietal.1993;Sambruna
Riley typeII (FRII;Fanaroff & Riley 1974) radio galaxies.
Within this scheme, broad-line (FWHM ∼> 2000 km s−1) etal.1994).Therearemainlytwoargumentstosupportthis:
1) the tendency for radio loud AGN to have systematically
flatter X-ray slopes than radio quiet ones; 2) the fact that
thesoftX-rayslopedecreaseswithcoredominance(Shastri
⋆ Email:[email protected] et al. 1993) and increases with radio spectral index (Fiore
(cid:13)c 0000RAS
2 P. Padovani et al.
et al. 1998). Both theseresults areexplained with thepres- 2 THE SAMPLE
enceofaradio-linkedsynchrotronself-Compton component
The lobe-dominated, broad-line objects studied in this pa-
oftheX-rayemissionthatislikelytobebeamed.Thiscom-
perbelongtoacompletesubsampleofthe2Jycatalogueof
ponentwouldbedominantintheFSRQ.InSSRQ,inwhich
radio sources (Wall & Peacock 1985). This subsample, de-
“blazar-like”, non-thermal emission is probably less impor- ◦
finedbyredshiftz<0.7anddeclinationδ<10 ,includes88
tant because of the larger angle w.r.t. the line of sight, the
objects and is complete down to afluxdensity level of 2Jy
“UVbump”wouldbestronger(aseffectivelyobserved:e.g.,
at 2.7 GHz. Opticalspectra are available for all thesources
Wills et al. 1995) and the steeper soft X-ray component
together with accurate measurements of the [O III]λ5007,
would represent its high-energy tail.
[O II]λ3727 and Hβ emission line fluxes (Tadhunter et al.
AlthoughanuclearX-raycomponenthasbeendetected
1993, 1998). Estimates of the core dominance parameter
alsoinradiogalaxies, itappearstobemuchweakerthanin
radio quasars (consistent with the idea that radio galaxies R [R = Score/(Stot−Score)] have been derived from both
arcsec-resolution images and higher resolution data (Mor-
haveanobscurednucleus).Forexample,theX-rayspectrum
gantietal.1993,1997).AstudyofthesoftX-raycharacter-
of Cygnus A (Ueno et al. 1994) is consistent with a typical
isticsoftheobjectsinthesamplehasbeencarriedoutusing
quasarspectrumabsorbedbyahighcolumndensityofcold
the ROSAT All-Sky Survey and/or ROSAT PSPC pointed
gas along the line of sight. On the other hand, in the case
observations (Siebert et al. 1996). For most of the objects,
of the broad-line radio galaxy 3C 390.3 (Inda et al. 1994),
however, nouseful X-ray spectral information is available.
its hard X-ray spectrum can be described by a relatively
The 2 Jy subsample described above contains 16 lobe-
flat,unabsorbedpower-law.Thiswouldsuggest thatBLRG
dominated, broad-line objects (excluding compact steep-
might well be the low-redshift counterpart of radio quasars
and therefore aligned within ∼<40◦ (as predicted byunified spectrumsources,whoserelation tootherclasses isstillnot
clear). From those,wehaveselected the10sources with es-
schemes: see e.g., Urry & Padovani 1995). timatedfluxinthe0.1–10keVbandlargerthan2×10−12
From the above it is clear that a spectral X-ray study erg cm−2 s−1 for an X-ray spectral study with the Bep-
of lobe-dominated, broad-lineradio sources (includingboth
†
poSAX satellite . Here we present the results obtained for
SSRQandBLRG),coveringalargeX-raybandisnecessary
the 5 objects so far observed in Cycle 1. The list of objects
for a number of reasons. Namely: 1) to study the hard X-
and their basic characteristics are given in Table 1, which
ray properties of lobe-dominated, broad-line radio sources,
presents the source name, position, redshift, optical magni-
atpresentnotwellknown;2)toinvestigateifthedifference
in the soft X-ray spectra of SSRQ and FSRQ apply also to tude V, 2.7 GHz radio flux, radio spectral index αr (taken
from Wall & Peacock 1985), core dominance parameter R
thehardX-rayband.Thedetectionofaflattercomponentin
SSRQwillbeextremelyimportantforourunderstandingof at 2.3 GHz, Galactic NH and classification.
theemission processesinthisclassofobjects;3)toincrease
thenumberofBLRGforwhichtheX-rayspectrumisknown
in detail in order to disentangle the real nature of BLRG 3 OBSERVATIONS AND DATA ANALYSIS
andinvestigateiftheX-rayspectraofBLRGandSSRQare
AcompletedescriptionoftheBeppoSAXmissionisgivenby
similar.
In this paper we present BeppoSAX observations of Boellaetal.(1997a).Therelevantinstrumentsforourobser-
vations are the coaligned Narrow Field Instruments (NFI),
fivelobe-dominated, broad-line radio sources, namely three
which include one Low Energy Concentrator Spectrometer
SSRQandtwoBLRG(wefollowthecommonlyadopteddef-
(LECS; Parmar et al. 1997) sensitive in the 0.1 – 10 keV
inition of lobe-dominated source, which implies a value of
band; three identical Medium Energy Concentrator Spec-
the core dominance parameter R < 1). The sample is well
trometers(MECS;Boellaetal.1997b),coveringthe1.5–10
defined (i.e., it is not a compilation of known hard X-ray
keV band; and the Phoswich Detector System (PDS; Fron-
sources) and it is extracted from the 2-Jy sample of radio
tera et al. 1997), coaligned with the LECS and the MECS.
sources for which a wealth of radio and optical information
isavailable.TheuniquecapabilityoftheBeppoSAXsatellite ThePDSinstrumentismadeupoffourunits,andwasoper-
atedincollimatorrockingmode,withapairofunitspointing
(Boellaetal.1997a)ofperformingsimultaneousbroad-band
atthesourceandtheotherpairpointingatthebackground,
X-ray (0.1−200 keV) studies is particularly well suited for
the two pairs switching on and off source every 96 seconds.
a detailed analysis of the X-ray energy spectrum of these
The net source spectra have been obtained by subtracting
sources.
the ‘off’ to the ‘on’ counts. A journal of the observations is
In§2wepresentoursample,§3discussestheobserva-
given in Table 2.
tions and the data analysis, while § 4 describes the results
of our spectral fits to the BeppoSAX data. In § 5 we also The data analysis was based on the linearized, cleaned
event files obtained from the BeppoSAX Science Data Cen-
examine the ROSAT PSPC data of our sources to better
ter (SDC) on-line archive (Giommi & Fiore 1997) and on
constrain the fits at low energies, in § 6 we combine the
analysis of the BeppoSAX and ROSAT data while in § 7 the XIMAGE package (Giommi et al. 1991) upgraded to
supporttheanalysis of BeppoSAX data. Thedata from the
we briefly comment on the lack of iron lines in our spec-
tra. Finally, § 8 discusses our results and § 9 summarizes
ourconclusions. Throughout this paperspectral indices are
†
written Sν ∝ν−α. lecNteodteobthjeactt,saasreex∼pe3ct0edtiminesanmyorfleuxlu-lmiminitoeudssinamthpele,Xt-hraey1b0asned-
than the 6 sources which did not make the X-ray flux cut. Our
sample is then biased towards the most X-ray luminous lobe-
dominated,broad-linesourcesinthe2-Jysample.
(cid:13)c 0000RAS,MNRAS000,000–000
BeppoSAX Observations of 2-Jy Lobe-dominated Broad-Line Sources 3
Table 1. Sample Properties
Name RA(J2000) Dec(J2000) z V F2.7GHz α2.7−5GHz R2.3GHz GalacticNH Class
Jy 1020 cm−2
OF−109 040748.4 −121136 0.574 14.9 2.35 0.42 0.57 3.81 QSO
PictorA 051949.0 −454646 0.035 15.8 29.00 1.07 0.03 4.15 BLRG
OM−161 113910.6 −135043 0.554 16.2 2.80 0.65 0.16 3.59 QSO
PHL1657 213744.1 −143255 0.200 15.5 2.00 0.63 0.06 4.45 QSO
PKS2152−69 215705.7 −694123 0.027 13.8 19.27 0.71 0.03 2.50 BLRG
three MECS instruments were merged in one single event sections weretakenfrom Morrison andMcCammon (1983).
filebySDC.TheLECSdataabove4keVwerenotuseddue For one set of fits NH was fixed at the Galactic value, de-
tocalibration uncertaintiesin thisbandthat havenot been rived from Elvis, Lockman & Wilkes (1989) for PHL 1657
completely solved at this time (Orr et al. 1998). As recom- andfromthenhprogramatHEASARC(basedonDickey&
mendedbytheSDC,LECSdatahavebeenthenfittedonly Lockman 1990), for the remaining objects. NH was also set
in the 0.1−4 keV range, while MECS data were fitted in free to vary to check for internal absorption and/or indica-
the1.8−10.5 keV range. tions of a “soft-excess.”
Spectra were accumulated for each observation using Our results are presented in Table 3, which gives the
the SAXSELECT tool, with 8.5 and 4 arcmin extraction name of the source in column (1), the energy index αx
radii for the LECS and MECS respectively, which include andreducedchi-squaredandnumberofdegreesoffreedom,
more than 90% of the flux. The count rates given in Ta- χ2ν(dof), in columns (2)-(3) for the fixed-NH fits; columns
ble 2 were obtained using XIMAGE and refer to channels (4)-(6)giveNH,αx andχ2ν(dof)forthefree-NH fits.Finally,
10 to 950 for the LECS and 36 to 220 for the MECS. in column (7) we report the unabsorbed X-ray flux in the
The BeppoSAX images were also checked for the presence 0.1−4.0keVrange(multipliedbyanormalization constant
of serendipitous sources which could affect the data anal- derived from the combined LECS plus MECS fits: see next
ysis. The ROSAT public images of our sources were also section). The errors quoted on the fit parameters are the
inspected(seebelow).Most ofourobjectshaveatleast one 90% uncertainties for one and two interesting parameters
source within the LECS extraction radius in the ROSAT for Galactic and free NH respectively.
PSPC and/or MECS images but at a flux level which is TworesultsareimmediatelyapparentfromTable3:the
at maximum 10% of the target (and in most cases below fittedenergyindicesareflat(αx <1);andthefittedNH val-
3%). Serendipitous sources in the field are then unlikely to uesareconsistent with theGalactic ones.Thisis confirmed
affect our results at a significant level. We also looked for by an F-test which shows that the addition of NH as a free
variability on timescales of 500 and 1,000 seconds for each parameter does not result in a significant improvement in
observation, with null results. theχ2 values.WewillthenassumeGalacticNH inthecom-
TheLECSandMECS backgroundislow,althoughnot bined LECS and MECS fits. For the two objects without
uniformlydistributedacrossthedetectors,andratherstable. LECSdatathisassumption isalsojustifiedbythefactthat
For this reason, and in particular for the spectral analysis, thefit tothe MECS data is not strongly dependenton NH.
it is better to evaluate the background from blank fields, The spectrum of PHL 1657 is more complicated than
rather than in annuli around the source. Background files a simple power-law: the residuals show a clear excess at
accumulatedfromblankfields,obtainedfromtheSDCpub- E ∼<0.7keV.Indeedabrokenpower-lawmodelsignificantly
lic ftp site, were then used. improvesthefit (see below). Weaker“soft-excesses” cannot
beexcludedinthetwoothersources(seebelow)sothefluxes
giveninTable3,basedonasinglepower-lawfittothedata,
are almost certainly underestimated.
4 SPECTRAL FITS
SpectralanalysiswasperformedwiththeXSPEC9.00pack-
age, using the response matrices released by SDC in early 4.2 LECS and MECS Data
1997. The spectra were rebinned such that each new bin
Our results from the jointly fitted LECS and MECS data
contains at least 20 counts (using the command GRPPHA
assumingasinglepower-lawmodelwithGalacticabsorption
within FTOOLS). Various checks using some of the rebin-
arepresentedinTable4,whichgivesthenameofthesource
ning files provided by SDC haveshown that our results are
independentoftheadoptedrebinningwithintheuncertain- in column (1), αx and χ2ν(dof) in columns (2)-(3), and the
unabsorbed X-ray flux in the 2−10 keV range in column
ties.TheX-rayspectraofoursourcesareshowninFigure1
(whichincludesbothBeppoSAXandROSATdata:seeSect. (4). The errors quoted on αx are 90% uncertainties.
DuetotheuncertaintiesinthecalibrationoftheLECS
6).
instrument, the LECS/MECS normalization has been let
freetovary.Theresultingvalues,inthe0.6–0.8range,are
consistent with theexpected one (Giommi & Fiore, private
4.1 LECS Data: Constraining NH
communication).
At first, we fitted the LECS data with a single power-law The striking result is that all the sources have rela-
modelwithGalacticandfreeabsorption.Theabsorbingcol- tively flat X-ray energy indices. The mean value is hαxi =
umnwasparameterizedintermsofNH,theHIcolumnden- 0.75±0.02 (here and in the following we give the standard
sity,with heavierelementsfixedat solar abundances.Cross deviation of the mean). This implies that the spectra are
(cid:13)c 0000RAS,MNRAS000,000–000
4 P. Padovani et al.
Table 2. Journal of observations
Name LECS LECS MECS MECS Observingdate
exposure(s) countrate(10−2 s−1) exposure(s) countrate(10−2 s−1)
OF−109 ... ... 16,864 7.8±0.3 22-SEP-1996
PictorA 5,271 16.5±0.7 14,798 26.4±0.5 12/13-OCT-1996
OM−161 13,782 3.3±0.2 28,248 5.7±0.2 11/12-JAN-1997
PHL1657 4,704 9.3±0.6 17,429 18.0±0.4 29-OCT-1996
PKS2152−69 ... ... 9,367 15.1±0.5 29-SEP-1996
Table 3. LECS spectral fits
Galactic NH Free NH
Name αax χ2ν(dof) NbH αbx χ2ν(dof) Flux(0.1–4.0keV)c
1020 cm−2 10−12 ergcm−2 s−1
PictorA 0.69+−00..1144 0.78(30) 2.75+−21..7595 0.60+−00..2232 0.75(29) 17.6±0.7
OM−161 0.77+−00..2202 0.77(24) 3.70+−72..8406 0.77+−00..4306 0.80(23) 3.8±0.2
PHL1657d 0.83+−00..2234 1.01(16) 2.75+−31..4988 0.69+−00..3386 0.97(15) 13.5±0.9
a Quotederrorscorrespondto90%uncertainties foroneinterestingparameter.
b Quotederrorscorrespondto90%uncertainties fortwointerestingparameters.
c Unabsorbedflux.1σ statisticalerrorsforbest-fitmodel(modeluncertainties notincluded).LECSfluxeshavebeenmultipliedbya
normalizationconstant derivedfromthecombinedLECSandMECSfits(∼0.6−1−0.8−1:seetext).
d AnF-testshowsthatabrokenpower-lawmodelimprovesthefitatthe98.1%level.Best-fitparametersare(forGalacticNH):
αS=1.2,αH=0.3,Ebreak=1.6keV,χ2ν(dof)=0.84(14).
still raising in a ν−fν plot, and therefore that thepeak of 5 ROSAT PSPC DATA
theX-rayemissionintheBeppoSAXbandisatE >10keV.
All our objects were found to have ROSAT PSPC data:
Table4alsoreports(inthefootnotes) thefitstotheMECS
namely, OF −109, Pictor A, PHL 1657 and PKS 2152−69
data for the three sources with both LECS and MECS ob-
wereall targets ofROSATobservations, whiledatafor OM
servations. The energy indices in the 1.8−10.5 keV range
−161 were extracted from the ROSATAll-SkySurvey.
haveameanvaluehαxi=0.74±0.03, basically thesameas
in thewhole 0.1−10.5 keVband.
As mentioned in the Introduction, the spectra of the 5.1 Data Analysis
class of sources under study are generally steep at lower
X-ray energies (and there is indeed strong evidence for a In the analysis of the pointed PSPC observations, we first
steeper X-ray component in the LECS data of PHL 1657). determined the centroid X-ray position by fitting a two-
dimensional Gaussian to the X-ray image. Source counts
Wethentriedtofitabrokenpower-lawmodeltoourdata.A
significantimprovementinthefit(96.5%level)wasobtained were then extracted from a circular region with 3 arcmin
only for PHL 1657, whose residuals again showed a clear radius around the centroid source position. The local back-
excessatE ∼<0.7keV.Thebest-fitparametersareαS =1.3, ground was determined from an annulus with inner radius
αH =0.76±0.12, and Ebreak=0.9 keV. 5 arcmin and outer radius 8 arcmin. If any X-ray sources
weredetectedinthebackgroundregion,theywerefirstsub-
Thefactthattheotherfoursourcesshownosignificant
tracted from thedata.
evidence for a concave spectrum needs to be investigated
The source countsfrom OM −161 were extracted from
with more data at soft X-ray energies. Hence the need to
a circular region with radius 5 arcmin from the All-Sky
resort to ROSATPSPC data (see Sect. 5).
Survey data. The larger extraction radius compared to the
pointed PSPC observations accounts for the larger point
spread function in the Survey. The local background was
4.3 The PDS Detection of Pictor A
determined from two source-free regions with radius 5 ar-
Onlythebrightestsourceofoursample,PictorA,hasbeen cmin,displacedfromthesourcepositionalongthescanning
detectedbythePDSinstrument(upto∼50keV;seeFig.1) direction of the satellite duringthe All-SkySurvey.
despitetherelativeshortexposuretime(6.8 ks).Thecount Afterbackgroundsubtraction,thedatawerevignetting
rate is 0.18±0.05 ct/s, that is the significance of the de- anddeadtimecorrectedandfinallybinnedintopulseheight
tection is about 3.6 σ. Given the relatively small statistics, channels. Only channels 12-240 were used in the spectral
it is hard to constrain the high energy (E > 10 keV) spec- analysis, due to existing calibration uncertainties at lower
∼
trumofPictorA.AparametrizationoftheMECSandPDS energies. Thepulse height spectra were rebinned toachieve
data with asingle power-law model gives a best fit value of a constant signal-to-noise ratio in each spectral bin, which
αx perfectly consistent with that derived from MECS data ranged from 3 to 6, depending on the total number of pho-
only.Abrokenpower-lawmodel,withthesoft energyindex tons.
fixedtothevalueobtained from theMECS data (seeTable
4),givesnosignificantimprovementinthefit(andthehard
5.2 Spectral Fits
energyindexisconsistentwiththesoft one).Therefore, the
PDS data appear to lie on the extrapolation of the lower AsfortheLECSdata,wefittedtheROSATPSPCdatawith
energy data.Theremight bea slight excessin theresiduals asinglepower-lawmodelwithGalacticandfreeabsorption.
above 10 keV but as described above this is not significant Our results are presented in Table 5, which gives the name
and does not warrant more complicated models. ofthesourceincolumn(1),theROSATobservationrequest
(cid:13)c 0000RAS,MNRAS000,000–000
BeppoSAX Observations of 2-Jy Lobe-dominated Broad-Line Sources 5
Figure 1. Spectra of the combined BeppoSAX and ROSAT
PSPC data for our sources. The data are fitted with a single
power-law model with Galactic absorption (free absorption for
PKS2152−69).NotethattheROSATcounts[lowerspectra]have
been normalized by the PSPC geometric area of 1,141 cm2 and
shouldbereadas“normalizedcounts/sec/keV/cm2.”
Table 4. LECS and MECS spectral fits
Name αax χ2ν(dof) Flux(2–10keV)b
10−12 ergcm−2 s−1
OF−109c 0.81+−00..1144 1.17(46) 3.7±0.1
PictorAd 0.68+−00..0067 0.78(128) 13.8±0.3
OM−161e 0.77+−00..1111 0.81(80) 2.5±0.1
PHL1657f,g 0.78+−00..0088 0.82(105) 8.6±0.2
PKS2152−69c 0.70+−00..1122 1.04(48) 7.5±0.2
a AssumingGalacticNH.
bUnabsorbedflux.1σstatisticalerrorsforbest-fitmodel(model
uncertainties notincluded).
(ROR) number in column (2), αx and χ2ν(dof) in columns c OnlyMECSdataavailable.
e(tχ3hr2νr)e(o-d(ru4osn)fqa)fubofoosrotrterhtdbheeoefidnfxrXteehde-e-r-NaNfiyHtHflpfifiuattrxssa;.micnFoeilttnuheamrelslny0as,.r1i(en5−t)ch-o2(el7.u49)m0gk%nievV(ue8nNr)caHwenr,egtαear.xeinpTatoinherdets gfed MMMAEEEnCCCFSSS-otooennnsllltyyysfififihttto:::wαααsxxx==t=ha000t...767777a+−+−+−00b0000......r110000o327788k,,,eχχχn2ν2ν2ν(((pdddoooowfff))e)r===-la000w...877489m(((59o8578d)))el improves
foroneandtwointerestingparametersforGalacticandfree
the fit at the 96.5% level. Best-fit parameters are: αS = 1.3,
NH respectively. αH=0.76±0.12,Ebreak=0.9keV,χ2ν(dof)= 0.81(103).
The main results of the ROSAT PSPC fits are the fol-
lowing: 1. the fitted energy indices are steeper than the
(cid:13)c 0000RAS,MNRAS000,000–000
6 P. Padovani et al.
MECS (plus LECS) ones (with the exception of OM −161 in the PSF (dueto residual wobble motion, attitude uncer-
for which the ROSAT αx has large uncertainties); 2. there tainties,etc.)thesefractionsagreeverywellwiththeresults
isnoevidenceforinterveningabsorptionabovetheGalactic from thespectral decomposition (5 and 10% respectively).
value in our sources, with the exception of PKS 2152−69, The gas temperatures we find (< 1 keV) are very rea-
forwhichtheF-testshowsthattheadditionofNH asafree sonable for gas associated with an elliptical galaxy (e.g.,
parameterresultsinasignificantimprovement(98.6%level) Forman et al. 1985). To check that the observed luminosi-
inthegoodnessofthefit(thefittedNH isabout50%higher ties are also physically plausible, we performed the follow-
than the Galactic value); 3. the single power-law fit is not ing test. There is a well-known strong correlation between
great, although still acceptable, for Pictor A (Pχ2 ∼ 5%) X-ray luminosity and absolute blue magnitude for ellipti-
and PHL 1657 (Pχ2 ∼7%). calgalaxies(e.g.,Formanetal.1985,1994).Integratedblue
The mean difference between the ROSAT PSPC and magnitudes for Pictor A and PKS 2152−69, obtained from
MECS (plusLECS) energy indices(excludingOM−161) is NED,implyMB ≃−20.7and−22respectively.The0.5−4.5
0.44±0.11,clearlyindicativeofaflatteningathighenergies, keV luminosities for the thermal components of the two
with theemergence of a hard component. sources are L0.5−4.5 ≃ 4×1042 erg s−1 for Pictor A, with
a rather large 90% error range (1042−1043) while for PKS
2152−69wegetL0.5−4.5 ≃2×1042ergs−1(90%errorrange:
5.3 A Thermal Component in the X-ray Spectra 3×1041−8×1042).Thesenumbers,compared against Fig.
of Pictor A and PKS 2152−69
4 of Forman et al. (1994), show that while theX-ray power
Pictor A and PKS 2152−69 are relatively nearby objects in the thermal component of PKS 2152−69 is not unusual
(z ≤ 0.035). The ROSAT PSPC images show evidence for for its optical power, that of Pictor A is about an order
an extendedcomponent on a scale ∼>50” for PKS 2152−69 of magnitude larger than the maximum values of elliptical
and ∼>70” for Pictor A, which correspond to about 40 and galaxiesofthesameabsolutemagnitude.Itthenseemsthat
70kpcrespectively.PKS2152-69isalsoseenextendedfrom theintrinsicpowerofthethermalcomponentistoolargeto
ROSATHRIdata(Fosbury,privatecommunication).Early- beassociated with thegalaxy.
type galaxies are known to have diffuse emission from hot Astherelativelylowgastemperatureinferredfrom the
gas on these scales (Forman, Jones & Tucker 1985), so we dataisalsotypicalofsmallgroups,onecouldspeculatethat
addedathermalcomponent(Raymond&Smith1977)tothe most of thethermalemission in Pictor Ais associated with
power-law model, assuming solar abundances (our results a group associated with this source. An inspection of Fig.
are only weakly dependent on the adopted abundances). 8 of Ponman et al. (1996), which reports the X-ray lumi-
Our results are reported in Table 6 which gives the name nosity – temperature relation for Hickson’s groups, shows
of the source in column (1), αx in column (2), thegas tem- that, within the rather large errors, Pictor A might fall in
perature(inkeV)incolumn(3),χ2(dof)incolumn(4),the the correct portion of the plot, although the best fit values
ratio between the thermal and nonν-thermal components in (Lx≃4×1042 ergs−1,kT =0.55 keV),would putitabove
the 0.1−2.4 keV range in column (5), and finally the F- the observed correlation. However, the richness of the envi-
test probability in column (6). The errors quoted on the ronmentofthissourceisverylow(Zirbel1997),inconsistent
fit parameters are the90% uncertainties for two interesting even with a small group. It might therefore be speculated
parameters. Galactic NH as been assumed (see above). In thatPictorAisanotherexampleforaso-called fossilgroup
both cases this addition results in a significantly improved (Ponman et al. 1994), i.e. a single elliptical galaxy that is
fit (> 99.9% level) over a single power-law model. (Note considered to be the result of a merging process of a com-
thataRaymond-Smithmodelbyitselfgivesextremelypoor pactgroup.ThismergingisbelievednottoaffecttheX-ray
fitstothedata.)Withtheadditionofathermalcomponent haloofthegroup(Ponman&Bertram1993) andthegalax-
theneedforabsorptionabovetheGalacticvalue,whichwas ies formed in this way will still show the extended thermal
indicatedforPKS2152−69andsuggestedforPictorAvan- emission component of the intra-group gas although they
ishes;freeNH fitsnowdonotresultinasignificantimprove- appear isolated.
ment in thegoodness of thefits. In summary, while the thermal component in PKS
AsacheckofourresultswealsofittedthespectraofOF 2152−69 is consistent with emission from a hot corona
−109andPHL1657with apower-lawplusthermalcompo- aroundthegalaxy,inthecaseofPictorAtheintrinsicpower
nent. No need for an extra component was found, which is of this component is too high. However, the luminosity of
consistentwiththefactthatthesetwosourcesareathigher thethermalcomponentisconsistentwiththatofacompact
redshift(i.e.,theputativethermalcomponentiscompletely groupof galaxies. SincePictorAappearstobeisolated, we
swamped by thestronger non-thermalemission). might have anotherexample of a fossil group.
It is interesting to note that the dominant component We found no physically meaningful evidence for the
intheX-rayemissionofourtwonearestsourcesisdefinitely presence of a thermal component in the LECS spectra of
non-thermal, but nevertheless the data indicate a 5−10% thethreesources for which we havethe relevant data.
contribution from thermal emission. This is confirmed by
an analysis of thePSPC images. The relevanceof extended
emission was in fact estimated by subtracting the point
6 ROSAT AND BeppoSAX DATA: THE WHOLE
spreadfunction(PSF)fromtheradialprofileofthesources.
PICTURE
For Pictor A, thefraction of photons above thePSF is 4%,
while for PKS 2152−69 is 13% (for the two other sources The last step is to put BeppoSAX and ROSAT PSPC data
withPSPCpointeddatathesefractionsarelessthan1%,as togethertobetterconstrain theshapeoftheX-rayspectra,
expected).Giventhestatisticalandsystematicuncertainties especially at low energies. Two of our sources, in fact, have
(cid:13)c 0000RAS,MNRAS000,000–000
BeppoSAX Observations of 2-Jy Lobe-dominated Broad-Line Sources 7
Table 5. ROSAT PSPC spectral fits
Galactic NH Free NH
Name ROR# αax χ2ν(dof) NbH αbx χ2ν(dof) Flux(0.1–2.4keV)c
1020 cm−2 10−12 ergcm−2 s−1
OF−109 701072 1.30+−00..0076 1.04(26) 3.65+−00..7772 1.26+−00..2233 1.08(25) 11.9±0.3
PictorA 700057 0.80+−00..0055 1.46(30) 4.83+−00..6685 0.96+−00..1167 1.34(29)d 17.7±0.3
OM−161 ... 0.80+−00..5850 0.65(5) 6.2 1.3 0.73(4)e 3.9±0.8
PHL1657 701542 1.42+−00..0065 1.49(20) 4.86+−00..7752 1.53+−00..2221 1.50(19) 23.7±0.5
PKS2152−69 701154 0.86+−00..0099 1.02(25) 3.72+−11..2120 1.22+−00..3365 0.83(24)f 6.4±0.2
a Quotederrorscorrespondto90%uncertainties foroneinterestingparameter.
b Quotederrorscorrespondto90%uncertainties fortwointerestingparameters.
c Unabsorbedflux.1σ statisticalerrorsforbest-fitmodel(modeluncertainties notincluded).
d Thereductionintheχ2 valueobtainedwiththefreeNH fitissignificantatthe93.7%levelaccordingtotheF-test.
e Theerrorsonthebest-fitparametersareessentiallyundeterminedduetothelowphotonstatistics.
f Thereductionintheχ2 valueobtainedwiththefreeNH fitissignificantatthe98.6%levelaccordingtotheF-test.
Table 6. ROSAT PSPC spectral fits: thermal and non-thermal components
Name αax kTa χ2ν(dof) RS/power-lawb P(F-test)c
keV
PictorA 0.79+−00..0180 0.55+−00..4300 1.13(28) 0.05 99.97%
PKS2152−69 0.91+−00..1199 0.59+−00..5307 0.73(23) 0.10 99.97%
a Quotederrorscorrespondto90%uncertainties fortwointerestingparameters.GalacticNH assumed.
b FluxratiooftheRaymond-Smithandpower-lawcomponents inthe0.1−2.4keVband.
c Probabilitythatthedecreaseinχ2 duetotheadditionofthethermalcomponent tothepower-lawmodelissignificant.
noLECS data, while for theremaining threeonly less than fallatrelativelylowenergiesexplainwhytheenergyindices
10LECS binsareavailablebelow1keV.TheROSATeffec- derived from the LECS fits are basically the same as those
tiveareaislargerthantheLECSeffectiveareaintherange obtained from thecombined LECS and MECS fits.
of overlap, providingmore leverage in the soft X-rays. The addition of a thermal component in Pictor A and
As before, we left free the LECS/MECS normalization PKS 2152−69 improvessignificantly thefit (>99.8% level)
and, as before, the fitted values are consistent with the ex- even in the case of a broken power-law model. As for the
pected ones. We also left free the PSPC/MECS normaliza- single power-law plus thermal component, free NH fits do
tion, to allow for any X-ray variability, which is seen in at not result in a significant improvement in the goodness of
least some lobe-dominated broad-line sources (e.g., 390.3: thefits, so Galactic NH is assumed. Best-fit parameters are
Leighlyetal.1997;3C382:Barr&Giommi1992).However, αS = 0.77, αH = 0.66, Ebreak = 2.1 keV, kT = 0.39 keV,
thePSPC/MECS normalization was, onaverage, around 1, for Pictor A, and αS = 0.92, αH = 0.68, Ebreak = 1.4 keV,
withamaximumexcursionof30−40%.Nostrongvariability kT =0.57 keV, for PKS 2152−69. These are perfectly con-
is then present between theROSATand BeppoSAX data. sistent with those obtained from the broken power-law fit
to the BeppoSAX and ROSAT PSPC data and with the
As it turned out, in all cases for which we had enough
temperaturesderivedfromtheROSATPSPCdata,butthe
statistics at low energies (i.e., excluding OM −161) a bro-
uncertaintiesarenowpoorly determinedbecauseof therel-
ken power-law model resulted in a significantly improved
atively large numberof parameters.
fit (≥ 99.9% level) over a single power-law model over the
whole 0.1 –10.5 keVrange. Ourresults are reported in Ta-
ble7 which givesthenameof thesourcein column (1),αS,
αH, and Ebreak in columns (2)-(4), χ2ν(dof) in column (5)
7 IRON LINES?
and finally the F-test probability in column (6). The er-
rors quotedon thefit parametersare the90% uncertainties A number of AGN exhibit in their X-ray spectra iron Kα
for three interesting parameters. Based on the LECS and lines which are characteristic of relativistic effects in an ac-
ROSATPSPCresults, Galactic NH asbeen assumed forall cretion disk surrounding a central black hole (e.g., Nan-
sources apart from PKS 2152−69. Thecombined datawith dra & Pounds 1994). It appears that radio-loud AGN have
the best fit single power-law model (to show the spectral weakeriron linesthanradio-quiet ones,although some low-
concavity) are shown in Figure 1. luminosity,radio-loudsourcesareknowntohavestrongiron
As can be seen from the Table, the model parameters lines (Nandraet al. 1997).
areextremelywelldetermined.Notsurprisingly,theαS val- We searched for Fe Kα emission in our MECS spectra:
ues are very similar to the ROSAT PSPC energy indices, none was found. The 90% upper limits on the equivalent
while the αH values are basically the same as the MECS width (in the source rest frame) of an unresolved iron line
(plusLECS) energyindices. Thespectra areobviously con- (σ =0) at energy 6.4 keV are the following: OF −109: 380
cave,withhαS−αHi=0.49±0.09andenergybreaksaround eV;PictorA:150eV;OM−161:300eV;PHL1657:170eV;
1.5 keV (Pictor A has a break at about 4 keV but with a PKS 2152−69: 400 eV. Note that for any broader line the
largeerrorduetotherelativelysmalldifferencebetweenthe limitsarecorrespondinglyhigher.OurresultforPictorAis
softandthehardspectralindices).Thefactthatthebreaks consistentwiththeupperlimitof100eVgivenbyEracleous
(cid:13)c 0000RAS,MNRAS000,000–000
8 P. Padovani et al.
Table 7. ROSAT PSPC, BeppoSAX LECS and MECS spectral fits
Name αa αa Ea χ2(dof) P(F-test)b
S H break ν
keV
OF−109 1.33+−00..1110 0.80+−00..2201 1.32+−10..3397 1.06(71) >99.99%
PictorA 0.79+−00..0077 0.56+−00..2318 3.92+−43..8180 0.91(157) 99.90%
OM−161 ... 0.77+−00..1111 ... 0.79(86) ...
PHL1657 1.42+−00..0099 0.75+−00..1122 1.45+−00..5209 0.96(124) >99.99%
PKS2152−69c 1.23+−00..5471 0.70+−00..1188 1.69+−20..0715 0.98(71) >99.99%
a Quotederrorscorrespondto90%uncertainties forthreeinterestingparametersforallsourcesapartfromOM−161(oneinteresting
parameter).GalacticNH assumedforallsourcesapartfromPKS2152−69.
b Probabilitythatthedecreaseinχ2 duetotheadditionoftwoparameters(fromasinglepower-lawfittoabrokenpower-lawfit)is
significant.
c FitwithfreeNH=3.8+−10..09×1020 cm−2.
& Halpern (1998) based on a ∼ 65 ks ASCA observation, The objects in our sample are lobe dominated and
and our limit for PHL 1657 agrees with the upper limit of therefore we investigated if inverse Compton scattering of
140 eV given by Williams et al. (1992) based on a 15 ks cosmic microwave background photonsintotheX-rayband
Ginga observation. by relativistic electrons in the diffuse radio lobes could be
We note that the upper limits on iron lines in our responsiblefortheobservedX-rayemission (seee.g.,Harris
sources are not very stringent and still consistent with val- &Grindlay1979; Feigelsonetal.1995).Wefindthatthisis
uesfoundinotherlobe-dominatedbroad-linedsources(e.g., not the case and the derived X-ray emission is almost two
Inda et al. 1994; Lawson & Turner 1997; Wo´zniak et al. orders of magnitude lower than observed. This is further
1998). supported by the fact that no strong resolved components
have been found for our objects by ROSAT. The hotspot
in PictorA isknown tohaveassociated X-rayemission but
this is very weak and indeed only marginally detected by
8 DISCUSSION Einstein (R¨oser & Meisenheimer 1987; Perley et al. 1997).
Wehavepresentedthefirstsystematic,hardX-raystudyof Thehard X-raycomponentpresent in FSRQisusually
a well-defined sample of lobe-dominated, broad-line AGN. interpretedasduetoinverseComptonemission,mostlikely
Our main result is that this class of objects has a hard X- duetoacombination ofsynchrotronself-Compton emission
rayspectrumwithαx ∼0.75atE ∼>1−2keV.Inaddition, (with the same population of relativistic electrons produc-
wealsodetectathermalemissioncomponentpresentatlow ing the synchrotron radiation and then scattering them to
energies inthespectraof twoBLRGs,butwefindthatthis higher energies) and Comptonization of external radiation
component contributes only ∼<10% of thetotal flux. (possibly emitted by material being accreted by the cen-
Hard X-ray emission is also present in core-dominated tral object; see e.g., Sikora, Begelman & Rees 1994). As
radio-loud quasars, and the detection of similarly flat spec- the hard X-ray emission in our sources has a similar, flat
tra in our sample of lobe-dominated AGN has important slope, it seems natural to attribute it to the same emission
implications for our understanding of the relation between mechanism. The smaller effect of Doppler boosting for lobe
thetwo classes. dominated sources would then make this component to ap-
In order to make a more quantitative comparison be- pearandbecomedominantonlyathighenergies,andthisis
tweenthehardX-rayspectraofthetwoclasseswesearched exactlywhatisshownbyourdata.Thisisfurtherconfirmed
the literature for a study of core-dominated/flat-spectrum andclearlyshowninFigure2,whereweplotthe2−10keV
radio quasars similar to ours, i.e., based on a well-defined, spectralindexversusthecoredominanceparameterR.(The
homogeneoussample.Surprisingly,wefoundnone.Wethen open triangles indicate the SSRQ and BLRG found in the
decided to collect all the information we could find on the literature, in some of thepapers listed above for FSRQ).It
2−10 keV spectra of FSRQ, excluding objects with large is evident from the figure that no correlation is present be-
(>0.5) uncertaintieson theX-rayspectral index.Thedata tween the two quantities. This seems in disagreement with
comefromGingaandEXOSAT/MEobservationspublished the correlation claimed by Lawson et al. (1992) (based on
in Makino (1989), Ohashi et al. (1989, 1992), Lawson et al. EXOSAT/MEdata)andLawson&Turner(1997)(basedon
(1992), Saxton et al. (1993), Williams et al. (1992), Sam- Gingadata),theonlyprevioushardX-raystudieswhichin-
bruna et al. (1994) and Lawson & Turner (1997; αx in the cludedsomesteep-spectrumradioquasars.Wenotethatthe
2−18keVrange).Theresultingsampleincludes15objects spectral indices derived from EXOSAT data had relatively
andischaracterized byhαxi=0.70±0.06, perfectly consis- largeerrorsandthatoursampleofFSRQ,SSRQandBLRG
tent with our results. islarger thanthesamplesusedbyLawson etal.(1992) and
We stress again that the sample of FSRQ is heteroge- Lawson & Turner (1997) (26 vs. 18 and 15 objects respec-
neouswhereas oursample,although relatively small,iswell tively)especiallyasfaraslobe-dominatedbroad-linesources
definedandhasverywelldeterminedspectralindices.How- areconcerned(wherewehavebasicallydoubledthenumber
ever,withinthelimitsintroducedbythebiaseslikelypresent of available sources). Moreover, the correlation claimed by
in the FSRQ sample, lobe-dominated and core-dominated Lawson & Turner (1997) becomes significant only by ex-
broad-line AGN appear to have practically identical hard cluding the three BLRG in their sample (the exclusion of
X-ray spectra in the 2−10/18 keVregion. the BLRG has no effect on the lack of correlation between
(cid:13)c 0000RAS,MNRAS000,000–000
BeppoSAX Observations of 2-Jy Lobe-dominated Broad-Line Sources 9
appeartorequireaslightlyflatterspectralindexthangiven
by these authors (0.77±0.03, from the SIS and GIS fits)
whileour2−10fluxissimilar tothat derivedfrom theSIS
(but∼10%smallerthanthe2−10keVfluxestimatedfrom
theGIS).
AsdiscussedintheIntroduction,variouspreviousstud-
ies had found that SSRQdisplayed asteep soft X-rayspec-
trum (see, e.g., Fiore et al. 1998). In fact, despite the hard
component at higher energies, we nevertheless observe a
steeper spectrum at lower energies. In the whole ROSAT
band we find αx = 1.19 ± 0.13 (excluding OM−161, for
which the ROSAT αx has large uncertainties), which is in-
termediate between the values obtained for SSRQ by Fiore
etal.(1998)between0.4−2.4keV(αx=1.14)and0.1−0.8
keV (αx = 1.37). Our best fits to the whole 0.1−10 keV
rangeindeedrequireaspectralbreak∆αx ∼0.5betweenthe
softandhardenergyslopesatabout1−2keV.Thedisper-
sion in the energy indices is larger for the soft component.
Wefindσ(αS)=0.28whileσ(αH)=0.10,whichmightsug-
gestamorehomogeneousmechanismathigherenergies.We
notethatFiore et al. (1998) also found aconcavespectrum
(α0.1−0.8 keV−α0.4−2.4 keV ≃0.2)forradio-loudAGN(both
flat- and steep-spectrum) in theROSATband.
Figure 2. The 2−10 keV (2−18 keV for 10 sources from There are some concerns (R. Mushotzky, private com-
Lawson & Turner 1997) X-ray spectral index versus the core- munication)ofmiscalibrationbetweenROSAT,ononeside,
dominance parameter for our sources (filled symbols) and core- andBeppoSAX, ASCAand RXTEon theotherside,which
dominated flat-spectrum radio quasars (FSRQ: empty squares) could affect some of our conclusions. Namely, the inferred
andlobe-dominatedsteep-spectrumradioquasarsandbroad-line ROSAT spectral indices might be steeper than those de-
radiogalaxies(SSRQ+BLRG:triangles)fromtheliterature.Er- rived,inthesameband,byotherX-raysatellites(adetailed
ror bars represent 90% errors for our objects and 1 σ errors for
comparisonofsimultaneousASCA/RXTE/BeppoSAXspec-
mostoftheothersources.Seetextfordetails.
traof3C273isgivenbyYaqoobetal.inpreparation).The
spectral breaks we find in the spectra of our objects could
thenbepartlyduetothiseffect.Thisisclearlyanimportant
αx and Rin oursample). Larger homogeneous samples (es-
point,veryrelevantforX-rayastronomy,butwhichgoesbe-
peciallyofFSRQ)areclearlyrequiredinordertoinvestigate
yondthescopeofthispaper.Nevertheless,wecanstillcom-
this issue in more details.
ment on this as follows: 1. the “BeppoSAX only” spectrum
OurhardX-rayspectraarewellfittedbyasinglepower
ofPHL1657shows,byitself,significantevidenceofabreak,
law and we find no evidence for the hard excess often seen
with best fit parameters consistent (within the rather large
in low-luminosity AGN (e.g., Nandra & Pounds 1994) and
errors) with those obtained from the full ROSATand Bep-
interpreted as due to Compton reflection of the X-rays off
poSAX fit (see Sect. 4.2). At least in this source, then, the
opticallythickmaterial(Guilbert&Rees1988;Lightman&
evidenceforaspectralbreakis“ROSATindependent.”The
White1988).InthecaseofPictorA,thisisalsoconfirmedby
fact that this is not the case for the two other objects with
theresultsofEracleous&Halpern(1998)basedonalonger
LECS data, Pictor A and OM−161, can be explained by
ASCA observation. We note, however, that this reflection
therelativelysmallerbreakinthefirstobject andthesmall
component should normally be apparent above ∼ 10 keV
LECS statistics in the latter. In other words, the available
and that our data reach these energies only for Pictor A
evidence is consistent with breaks similar to those derived
(andeventhenwithrelativelysmallstatistics;seeSect.4.3).
fromthecombinedROSATandBeppoSAXfitstobepresent
Wo´zniak et al. (1998) have recently studied the X-
also in the LECS/MECS data; 2. our main result, that is,
ray (and soft γ-ray) spectra of BLRG using Ginga, ASCA,
the presence of a hard X-ray component in all our sources
OSSEandEXOSATdata.Theirobject list includes4lobe-
atE >1−2keV,isbasedonBeppoSAX dataandtherefore
dominatedBLRG,namely3C111,3C382,3C390.3,and3C ∼
clearly independentof any possible ROSATmiscalibration.
445.TheX-rayspectrahaveanenergyindexαx ∼0.7,with
some moderate absorption. Fe Kα lines have also been de- Onecouldalsoworryaboutpossiblemiscalibrationsbe-
tectedwithtypicalequivalentwidths∼100eV.AnyComp- tween different X-ray instruments affecting the (lack of )
ton reflection component is constrained to be weak and is correlation in Fig. 2. However,theresults of Wo´zniak et al.
unambiguously detected only in 3C 390.3. Our results are (1998)appeartoexcludethatpossibility.TheASCA,Ginga
consistent with their findings. andEXOSATX-rayspectraofthesourcesstudiedbythese
OurMECSresultsforPHL1657areinagreementwith authors, in fact, agree within the errors, particularly in the
theGingaenergyslopeobtainedbyWilliamsetal.(1992:see hard X-rayband.Given thegood cross-calibration between
theirTable3),whileour2−10keVfluxappearstobe∼35% BeppoSAX and ASCA,alarge miscalibration betweenBep-
smaller. Eracleous & Halpern (1998) reported on a ∼65 ks poSAX, Ginga and EXOSAT (the instruments used to ob-
ASCAobservationsofPictorA.OurLECSandMECSdata tain thedata used in Fig. 2) seems to beruled out.
(cid:13)c 0000RAS,MNRAS000,000–000
10 P. Padovani et al.
9 CONCLUSIONS Feigelson E.D., Laurent-Muehleisen S.A., Kollgaard R.I., 1995,
ApJ,449,L149
The main conclusions of this paper, which presents Bep-
FioreF.,ElvisM.,GiommiP.,Padovani P.,1998, ApJ,492,79
poSAXdataforawelldefinedsampleof2-Jysteep-spectrum FormanW.,JonesC.,TuckerA.,1985,ApJ,293,102
radio quasars and broad-line radio galaxies can be summa- FormanW.,JonesC.,TuckerA.,1994,ApJ,429,77
rized as below. FronteraF.etal.,1997,A&AS,122,357
All five lobe-dominated, broad-line sources included in GiommiP.,FioreF.,1997,in5thInternationalWorkshoponData
thisstudyhavebeen clearly detected upto10 keV (50keV AnalysisinAstronomy,Erice,inpress
forPictorA)anddisplayaflatX-rayspectrum(αx ∼0.75) GiommiP.,AngeliniL.,JacobsP.,TagliaferriG.,1991inWorrall
in the 2−10 keV range. One source (out of the three with D.M.,BiemesderferC.,BarnesJ.,eds,ASPConf.Ser.Vol.25,
AstronomicalDataAnalysisSoftwareandSystemsI,Astrom.
LECSandMECSdata,i.e.,reachingdownto0.1keV)shows
Soc.Pac.,SanFrancisco,p.100
significantevidenceofaspectralbreakatE ∼1keV.When
GuilbertP.W.,ReesM.J.,1988,MNRAS,233,475
ROSAT PSPC data, available for all five sources, are in-
HarrisD.E.,GrindlayJ.E.,1979,MNRAS,188,25
cluded in the fit, the evidence for concave overall spectra,
IndaM.etal.,1994,ApJ,420,143
with αsoft −αhard ∼ 0.5 and Ebreak ∼ 1.5 keV, becomes LawsonA.J.,TurnerM.J.L.,1997,MNRAS,288,920
highlysignificantforallobjectswithgood enoughstatistics LawsonA.J.,TurnerM.J.L.,WilliamsO.R.,StewartG.C.,Saxton
at low energies (i.e., excluding OM−161). No iron lines are R.D.,1992,MNRAS,259,743
detectedinourspectrabuttheupperlimitswederivearenot Leighly K.M., O’Brien P.T., Edelson R., George I.M., Malkan
very stringent (due to the relatively short exposure times). M.A., Matsuoka M., Mushotzky R.F., Peterson B.M., 1997,
The flat high-energy slope we find for our lobe-dominated ApJ,483,767
LightmanA.P.,WhiteT.R.,1988, ApJ,335
sources is consistent with the hard X-ray emission present
McDowellJ.C.etal.,1989,ApJ,345,L13
in core-dominated radio quasars. In fact, by collecting data
Morganti R., Killeen N.E.B., Tadhunter C.N., 1993, MNRAS,
from the literature on the X-ray spectra of radio quasars,
263,1023
we show that the available data are consistent with no de-
MorgantiR.,OosterlooT.A.,ReynoldsJ.E.,TadhunterC.N.,Mi-
pendence between the 2−10 keV spectral indices and the genesV.,1997MNRAS,284,541
core-dominance parameter, somewhat in contrast with the MorrisonR.,McCammonD.,1983,ApJ,270,119
situationatlowerenergies.Finally,athermalemissioncom- NandraK.,PoundsK.A.,1994, MNRAS,268,405
ponent is present at low energies in the spectra of the two Nandra K., George I.M., Mushotzky R.F., Turner T.J., Yaqoob
broad-lineradiogalaxies, although only atthe∼10% level. T.,1997,ApJ,488,L91
Three more targets have been approved as part of this OrrA.,Parmar,A.N.,Yaqoob T.,Guainazzi M.,1998, inScarsi
BeppoSAX observingprogram (oneat alowerpriority).We L., Bradt H., Giommi P., Fiore F., eds, The Active X-ray
Sky:ResultsfromBeppoSAX andRossi-XTE,Elsevier,Am-
will bepresentingresults ontheseadditional objects, anda
sterdam,p.497
morethoroughdiscussionoftheimplicationsofourresultsin
ParmarA.etal.,1997,A&AS,122,309.
termsofemissionprocesses,orientation,andmoregenerally
PerleyR.,Ro¨serH.-J.,MeisenheimerK.,1997, A&A,328,12
unified schemes in a future paper.
PonmanT.J.,BetramD.,1993,Nature363,51
Ponman T.J, Allan D.J., Jones L.R., Merrifield M., McHardy
I.M.,LehtoH.J.,LuppinoG.A.,1994, Nature369,462
ACKNOWLEDGEMENTS Ponman T.J., Bourner P.D.J., Ebeling H., Bo¨hringer H., 1996,
MNRAS,283,690
We acknowledge useful discussions with Alfonso Cavaliere, RaymondJ.C.,SmithB.W.,1977, ApJS,35,419
Andrea Comastri, Fabrizio Fiore, Paolo Giommi, Paola Ro¨serH.-J.,MeisenheimerK.,1987,ApJ,314,70
Grandi,RichardMushotzky,TahirYaqoob.WethankPaola Sambruna R., Barr P., Giommi P., Maraschi L., Tagliaferri G.,
Grandi also for her help with the analysis of the PDS data TrevesA.,1994, ApJS,95,371
Saxton R.D., Turner M.J.L., Williams O.R., Stewart G.C.,
ofPictorA.ThisresearchhasmadeuseoftheNASA/IPAC
OhashiT.,KiiT.,1993, MNRAS,262,63
ExtragalacticDatabase(NED),whichisoperatedbytheJet
Shastri P., Wilkes B.J., Elvis M., McDowell J., 1993, ApJ, 410,
Propulsion Laboratory, California Institute of Technology,
29
under contract with the National Aeronautics and Space
Siebert J., Brinkmann W., Morganti R., Tadhunter C.N.,
Administration. Danziger I.J., Fosbury R.A.E., di Serego Alighieri S., 1996,
MNRAS,279,1331
SikoraM.,BegelmanM.C.,ReesM.J.,1994, ApJ,421,153
Tadhunter C.N., Morganti R., di Serego Alighieri S., Fosbury
REFERENCES
R.A.E.,DanzigerI.J.,1993,MNRAS,263,999
Antonucci R.,1993,ARA&A,31,473 Tadhunter C.N., Morganti R., Robinson A., Dickson R., Villar-
BakerJ.C.,HunsteadR.W.,BrinkmannW.,1995,MNRAS,277, MartinM.,FosburyR.A.E.,1998,MNRAS,298,1035
553 Ueno,S.etal.,1994,ApJ,431,L1
BarrP.,GiommiP.,1992, MNRAS,255,495 UrryC.M.,Padovani P.,1995,PASP,107,803
BarthelP.D.,1989,ApJ,336,319 WallJ.V.,Peacock, J.A.,1985,MNRAS,216,173
BoellaG.etal.,1997a,A&AS,122,299 WilkesB.J.,ElvisM.,1987,ApJ,323,243
BoellaG.etal.,1997b,A&AS,122,327 WilliamsO.R.etal.,1992,ApJ,389,157
DickeyJ.M.,LockmanF.J.,1990,ARA&A,28,215 WillsB.J.etal.,1995,ApJ,447,139
ElvisM.,LockmanF.J.,WilkesB.J.,1989,AJ,97,777 Worrall D.M., Giommi P., Tananbaum H., Zamorani G., 1987,
Eracleous M., Halpern J.P., 1998, ApJ, in press (astro- ApJ,313,596
ph/9805253) Wo´zniakP.R.,ZdziarskiA.A.,SmithD.,MadejskiG.M.,Johnson
FanaroffB.L.,RileyJ.M.,1974,MNRAS,167,31P W.N.,1998,MNRAS,299,449
(cid:13)c 0000RAS,MNRAS000,000–000
