Table Of Content

Astronomy&Astrophysicsmanuscriptno.6285 (cid:13)c ESO2008 February5,2008 2003–2005 INTEGRAL and XMM-Newton observations of 3C 273 M.Chernyakova1,2,A.Neronov1,2,T.J.-L.Courvoisier1,2,M.Tu¨rler1,2,S.Soldi1,2,V.Beckmann3,P.Lubinski4,1, R.Walter1,2,K.L.Page5,M.Stuhlinger6,R.Staubert7,andI.M.McHardy8 1 INTEGRALScienceDataCenter,Chemind’E´cogia16,1290Versoix,Switzerlande-mail:[email protected] 2 GenevaObservatory,UniversityofGeneva,51ch.desMaillettes,CH-1290Sauverny,Switzerland 3 NASAGoddardSpaceFlightCenter,ExplorationoftheUniverseDivision,Code661,Greenbelt,MD20771,USA 7 4 NicolausCopernicusAstronomicalCenter,Bartycka18,00-716Warszawa,Poland 0 5 UKSwiftScienceDataCentre,Dept.ofPhysics&Astronomy,UniversityofLeicester,UniversityRoad,LeicesterLE17RH 0 6 XMM-Newton Science Operations Centre, European Space Astronomy Centre (ESAC) - Villafranca, P.O. Box 50727 - 28080 2 Madrid-Spain 7 Institutfu¨rAstronomieundAstrophysik-Astronomie,72076Tu¨bingen,Germany n 8 SchoolofPhysics&AstronomyUniversityofSouthamptonHighfield,SouthamptonSO171BJ,U.K. a J 9 ABSTRACT 2 Aims. Theaimofthispaperistostudytheevolutionofthebroadbandspectrumofoneofthebrightestandnearestquasars3C273. 1 Methods. Weanalyzethedataobtainedduringquasi-simultaneousINTEGRALandXMM-Newtonmonitoringoftheblazar3C273 v in2003–2005intheUV,X-rayandsoftγ-raybandsandstudytheresultsinthecontextofthelong-termevolutionofthesource. 5 Results.The0.2-100keVspectrumofthesourceiswellfittedbyacombinationofasoftcut-offpowerlawandahardpowerlaw. 2 Noimprovement ofthefitisachievedifonereplacesthesoftcut-off power lawbyeitherablackbody, oradiskreflectionmodel. 8 During the observation period the source has reached the historically softest state in the hard X-ray domain with a photon index 1 Γ=1.82±0.01.ComparingourdatawithavailablearchivedX-raydatafrompreviousyears,wefindasecularevolutionofthesource 0 towardsofterX-rayemission(thephotonindexhasincreasedby∆Γ ≃ 0.3−0.4overthelastthirtyyears).Wearguethatexisting 7 theoreticalmodelshavetobesignificantlymodifiedtoaccountfortheobservedspectralevolutionofthesource. 0 / Keywords.quasars–individual–3C273 h p - o 1. Introduction spectrum, as proposed by Grandi&Palumbo (2004), is not r observedinourdataset. t 3C 273 is a radio loud quasar, with a jet showing superlumi- s a nalmotion,discoveredattheverybeginningofquasarresearch. : Beingoneofthebrightestandnearest(z=0.158)quasars,3C273 Walter&Courvoisier (1992) have found an anticorrelation v wasintensivelystudiedatdifferentwavelengths(seeCourvoisier between the X-ray spectral slope and the ratio of X-ray to ul- i X (1998)forareview). traviolet (UV) luminosity of 3C 273 using quasi-simultaneous datafromEXOSAT,GINGAandIUEobtainedinthe80-s.They r In the X-ray band 3C 273 is thought to be quite a unique a haveinterpretedthiscorrelationasevidencethattheX-rayemis- source in the sense that in its X-ray spectrum the “blazar-like” sionisduetothermalComptonizationoftheUVflux.Thepres- emissionfromthepc-scalejetmixeswiththe“nuclear”compo- ence of optical monitors on-board of both XMM-Newton and nent, which reveals itself through the presence of a large blue- INTEGRAL enables us to analyze correlations between the big bump and a so-called “soft excess” below ∼ 1 keV. Both are blue bump variability and the variability in X-rays. Repeating typicalforemissionfromthenucleiofSeyfertgalaxies.During the analysis of Walter&Courvoisier (1992) with our data set the observationperiod discussed in this paper the radio to mil- wedonotfindthepreviouslyobservedX-rayslope–UV/X-ray limetercomponent,usuallyassociatedwithajet,hasbeeninits flux correlation. The disappearance of this correlation is even loweststate.Thus,ourdataprovideanopportunitytostudythe moresurprisingbecauseoftheappearanceofapreviouslyunob- “nuclear”componentinmoredetailsandtotesttheoreticalmod- servedcorrelationbetweentheUVandtheX-rayfluxfromthe elsfortheemissionof3C273. system. Taking into accountthe fact that the typicalvariability Surprisingly, in spite of the low level of the “jet-like” timescalesintheUVandtheX-raybandsaredifferent(months contribution, our data do not reveal the presence of typical inUVanddaysinX-ray),truesimultaneityoftheobservations “Seyfert-like”featuresinthespectrum.Infact,the0.2-100keV couldbeimportant.OurXMM-NewtonandINTEGRALobserva- spectrum is well fitted by a simple model composed of two tionsarethefirstobservationsinwhichUVandX-raydatawere power laws or a combination of a hard power law with a soft takenatthesametime. cut-off power law. We find that more sophisticated models, which include a black-body for description of the soft excess, It is possible, however, that appearance/disappearance of or a disk reflection component give comparable or worse the UV–X-ray correlation reflects a real evolution of the spec- fits to the data, than the above simple model. Moreover, the tral state of the source during the last 30 years. Collecting the “disentanglement” of Seyfert and jet components of the X-ray archivalX-raydata forabout30 yearsof observations,we find thatthe sourceindeedevolvestowardsa softerX-raystate. We Sendoffprintrequeststo:M.Chernyakova findthatthesourcehasreacheditshistoricallysoftestX-raystate 2 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 Table1.INTEGRALobservationsof3C273. Table2.XMM-Newtonobservationsof3C273. Period Date Revolution OnTime PNExposureTime Period T Rev OnTime ISGRI/SPI/JEM-X (ks) (ks) start (ks) Exposure(ks) 1 2003-01-06 563 8.95 5.97 1a 2003-01-05 28 120.8 2 2003-07-07 655 58.56 40.60 1b 2003-01-11 30 10.6 3 2003-12-14 735 8.91 5.77 1c 2003-01-17 32 106.8 4 2004-06-30 835 62.91 13.93 1 182.5/182.4/12.4 5 2005-07-10 1023 28.11 19.33 2a 2003-06-01 77 57.0 2b 2003-06-04 78 52.6 2c 2003-06-16 82 33.5 2e 2003-07-06 89–90 227.1 2f 2003-07-18 93 20.5 2g 2003-07-23 94 14.0 scientific analysis (OSA) package 5.1 distributed by ISDC1 2 308.9/310.5/40.3 (Courvoisieretal.2003b). 3 2004-01-01148–149 26.4 152.3/NA/13.8 Mostofthetime,observationsweredoneinthe5×5dither- 4 2004-06-23 207 92.7 67.1/NA/29.2 5a 2005-05-28320–321 278.3 ing mode. The staring mode was used only during revolutions 5b 2005-07-09 334 102.5 30,148,and149.ItisthereforenotpossibletouseSPIdatadur- 5 269.1/367.2/68.7 ingthethirdperiod.TherearealsonoSPIdataduringthefourth period,astheinstrumentwasintheannealingmodeduringthis observation. Fortheanalysisofthe INTEGRAL/JEM-Xobservationswe with photonindexΓ = 1.82±0.01in June 2003.Thisspectral have selected science windows for which the source was less slope is (by ∆Γ ≃ 0.3−0.4) softer than the typical slope ob- than 3.5◦ from the center of the field of view. During the first served in the 70-s and 80-s. This is definitely larger than the three periods JEM-X 2 was operating, while during the fourth typicalshorttime scale fluctuationsof the slope which are ata andfifthperioditwasswitchedoff,anddataweretakenwiththe levelof∆Γ≃0.1. JEM-X1telescope. It is important to note that spectral evolution of the source To minimize the influence of the flares in the surrounding should not be ignored in theoretical modelling of nuclear and backgroundofthechargedparticleswehavenotusedpointings jet emission from 3C 273. For example, the very hard spectral in which the detector countrate exceeds the mean of the given stateofthesourceinthelate70-shasledLightman&Zdziarski periodbymorethan5%. (1987) to the conclusion that “compact source” type models dominated by pair production can not be applied to 3C 273 in spiteofitshighapparentcompactness.However,thisconclusion 2.2.XMM-Newtonobservations isnotapplicableforthepresentstateofthesource.Similarly,the mostpopularmodelforthehardX-rayspectrumofthesource, ThelogoftheXMM-Newtonobservationsanalyzedinthispaper inverseComptonemission from the pc-scale jet, is satisfactory ispresentedinTable2.Amongalltheobservationsofthesource forthepresentstateofthesource,butcannotexplaintheΓ≃1.4 donewithXMM-Newtonwehaveselectedthosethatweredone spectralindexobservedin70-s. whileINTEGRALwasobserving. The paper is organizedas follows: in Section 2 we present thesequencesofobservationsandmethodsusedfordatareduc- TheXMM-NewtonObservationDataFiles(ODFs)wereob- tion and analysis. In section 3 we present the results obtained, tained from the on-line Science Archive2. The data were pro- anddiscussthem in Section4. We thengivea summaryof our cessed and the event-lists filtered using  within the analysisinthelastpartofthepaper. ScienceAnalysisSoftware(),version6.5andthemostup-to- datecalibrationfiles. AllEPICobservationswereperformedin theSmallWindowMode.Inallobservationsthesourcewasob- servedwiththeMOS1,MOS2andthePNdetectors.AsthePN 2. ObservationsandDataAnalysis detector,incontrasttoMOS1andMOS2,isnotaffectedbypile- up in our observations we concentrate our analysis on the PN 2.1.INTEGRALobservations data only. Patterns 0-4 were used in the analysis. Background SincethelaunchofINTEGRAL(Winkleretal.2003)onOctober spectra were extracted from areas of the sky close to 3C 273 17, 2002, 3C 273 was regularly monitored. Details of the data showingnoobvioussources.ARFGENandRMFGENwerethen usedinthecurrentanalysisaregiveninTable1.Thepreliminary runtoproducethecorrespondingARFandRMFfiles. analysisofthe2003datawasdonebyCourvoisieretal.(2003a). Opticalandultravioletfluxeswereobtainedwiththeoptical AdetailedstudyofthemultiwavelengthspectrumofJune2004 monitor(OM)ofXMM-Newtonwithomichainfollowingthein- waspresentedbyTu¨rleretal.(2006). structionsoftheSASWatchoutpage3andusingthevaluespro- DuringtheINTEGRALobservations3C273wasallthetime vided for an AGN spectral type. The resulting magnitudes are inaverylowstate,withafluxofseveralmilliCrabs(seeSection given in Table 3. No OM data for the selected filters are avail- 3.2).Toobtainbetterstatisticsforspectralanalysiswehavecom- ableforPeriod3. bineddatatakenwithinafewweeks,whichresultedinfivedata sets. The combined exposure times for each data set and each instrumenton-boardofINTEGRALarelistedinTable1. 1 http://isdc.unige.ch/?Soft+download The data for all INTEGRAL instruments were analysed 2 http://xmm.vilspa.esa.es/external/xmm data acc/xsa/index.shtml in a standard way with the latest version of the off-line 3 http://xmm.esac.esa.int/sas/documentation/watchout/uvflux.shtml M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 3 3 2 1 0 1 0 -1 Fig.1.toppanel:2003–2005IBIS/ISGRIlightcurvesof3C 273in20-60keV (uppercurve,black)and60-100keV (lowercurve, red).Thedataarebinnedin1-dintervals.Foreachtimeframethedategivenatthebottomcorrespondstothefirstdatapoint,tick intervalis1day.bottompanel:HardnessratioF /F . (60−100keV) (20−60keV) Table3.OpticalandUVmagnitudesof3C273duringeachob- 30 servationalperiod. 20 Period1 Period2 Period4 Period5 V 12.53±0.01 12.69±0.01 12.57±0.01 10 B 12.95±0.01 12.86±0.01 U 11.62±0.01 11.76±0.01 11.69±0.01 UVW1 11.32±0.01 11.28±0.01 11.44±0.01 11.30±0.01 0 UVM2 11.34±0.01 11.37±0.01 11.19±0.01 UVW2 11.14±0.01 11.36±0.01 11.16±0.01 0.2 0.15 3. Results 0.1 06.01.03 07.07.03 14.12.03 30.06.04 20.07.05 3.1.TimingAnalysis Fig.2.toppanel:2003–2005XMM/PNlightcurvesof3C273in In the top panel of Figure 1 we show the 3C 273 IBIS/ISGRI 0.2–1keV(uppercurve,black)and3–10keV(lowercurve,red). lightcurves in the 20-60 keV (black points) and in the 60-100 The data are binned in quarter of day intervals. bottom panel: keV(redpoints)energyranges.Inthisfigurethedataarebinned HardnessratioF /F . (3−10keV) (0.2−1keV) inonedayintervals.ISGRIdatashowthatthehardX-rayemis- sionfromthesystemisvariableonatleastaonedaytimescale. Inordertoinvestigatethenatureofthisvariabilitywehavepro- ducedthe“flux-flux”diagramforthe20-60keVand60-100keV energybands,shownin thebottomrightpanelofFig. 8. Large ofthe sourcein the20-100keV bandis mostlydueto changes error bars of the single-day flux measurementsdo not allow to inthenormalization,butnotintheshapeofthespectrum. seethegeneraltrendofthedependencebetweenthe20-60keV Fig.2showstheXMM-Newtonlightcurvesin0.2–1keV(up- and 60-100 keV flux. However, rebinning the data by building per curve, black) and 3–10 keV (lower curve, red) bands. One the average of three adjacent (along the 20-60 keV flux axis) can see significantchangesof the 0.2–1keV flux fromdata set originaldata pointsin the flux-flux plotwe find that the whole to data set, but the variabilityat the time scale of single obser- data set has a roughly linear behaviour (Spearman correlation vation, ∼ 10 ksec is negligible. The plot of the hardness ratio coefficient is equal to R = 0.89, with a probability of random (bottom panel of Fig. 2) shows that the 0.2-10 keV spectrum coincidenceP =7×10−3).Thisindicatesthatthevariability hardenstowardtheendoftheobservationperiod. rand 4 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 Table4.Parametersofthesinglepower-lawmodelin2-100keV energyrange.Thefitwasdoneleavingfreeintercalibrationfac- torsforISGRI(C),SPI(C )andJEM-X(C )withrespecttothe Period #1 i s j PNcamera.Theparameteruncertaintiescorrespondto1σlevel. Period #2 Period1 Period2 Period3Period4Period5 Period #3 Powerlaw Γ 1.81+0.01 1.84+0.01 1.70+0.011.75+0.011.66+0.01 Period #4 −0.01 −0.01 −0.01 −0.01 −0.01 F∗ 8.79+0.1710.10+0.048.04+0.136.72+0.108.54+0.09 2−10 −0.16 −0.07 −0.12 −0.09 −0.09 Period #5 Intercalibrationfactors C 1.11+0.09 1.35+0.05 1.33+0.090.30+0.110.67+0.05 j −0.09 −0.05 −0.09 −0.11 −0.05 C 0.92+0.05 1.22+0.03 1.42+0.110.98+0.081.23+0.04 i −0.05 −0.03 −0.11 −0.07 −0.04 C 1.39+0.26 1.49+0.16 – – 1.67+0.15 s −0.25 −0.16 −0.14 Statistics χ2 920 1673 950 1139 1419 (dof) (920) (1632) (1153) (1205) (1390) ⋆Inunitsof10−11ergcm−2s−1. 0.001 0.01 0.1 1 10 100 Energy, keV 11 0.9 Fig.3.Broadbandspectraof3C273foreachperiod.Dataarefit- 0.8 tedwiththesimplephenomenologicalmodeldescribedinTable 0.7 5.Thenewdataarecomparedtotheaverageflux(greyline)and the 1-σ variability range (grey area) based on historic data of 0.6 3C273(Tu¨rleretal.1999). 0.5 0.4 3.2.SpectralAnalysis OurINTEGRALandXMM-Newtonobservationsprovideavery 0.3 broadband(5decadesinenergy)setofquasi-simultaneousdata which spans from the UV (∼ 1 eV) up to the soft gamma- 3 ray (∼ 100 keV) band. The spectra of the source for all the 2 1 five selected data sets are shown in Fig. 3. Taking into ac- 0 count a wide range of theoretical models proposed for the X- -1 ray spectrum of 3C 273we adopta “simple to complex”strat- -2 -3 egy in the spectral analysis and attempt first to find a “mini- mal model” which can provide a satisfactory representation of 4 5 6 7 Energy, keV thedataandthenconsiderhowchangingand/oraddingvarious theoreticallymotivatedcomponentsimprovesorworsensthefit. The hydrogen column density was fixed to the galactic value N =1.79×1020cm−2(Dickey&Lockman1990). Fig.4. A 3C 273 spectrum in the 4-7 keV energy band as ob- H served with XMM/PN in July 2003 (second data set). Data are fitted with a single power law. A broad iron line is marginally 3.2.1. Phenomenologicalmodel visibleintheresiduals(seetext). Datainthe2–100keVenergyrangearewellfittedbyasimple powerlaw,seeTable4.InouranalysiswehavelefttheISGRI, SPIandJEM-Xintercalibrationfactorsfreetovarywithrespect below 2 keV and describes the so-called ”soft excess”. A hard tothePNcamera.ThesignificantvariationoftheJEM-Xinter- power law with photon index Γ dominates in the hard X-ray 2 calibrationfactorreflectsaknownchangeofthegainparameters band. An F-test shows that a slightly better fit to the data can oftheinstrument.Itis possibleto fitallthedatawiththe same beachievedifoneusesacut-offpowerlawinsteadofthesingle SPI(C = 1.41)andISGRI(C = 1.19)intercalibrationfactors, powerlawforthedescriptionofthesoftemission. s i but we have preferred not to do this, as the XMM-Newton and Surprisingly,wefindthatalreadythissimplemodelprovides INTEGRAL data are only quasi-simultaneous,and the different a verygoodfit to the combinedINTEGRAL and XMM-Newton valueofthefluxmeasuredbyXMM-NewtonandINTEGRALcan data in all the observations except the second one. In the sec- beduetothephysicalchangeofthesourcefluxonaonemonth ond data set the addition of a broadredshiftedemission line at scale.Thefitoftheseconddatasetcanbeimprovedbythead- E =6.65±0.07keVintherestframeoftheobjectmakesthe Fe ditionofabroademissionlineasdiscussedbelow. fit acceptable.Thisline can be identifiedwith a fluorescentK α The simplestsatisfactoryfit tothe 0.2-100keVdata canbe ironline.Abroadironlinewasseeninpreviousobservationsof achieved with a model consisting of two power laws. A soft 3C273bydifferentinstruments,seee.g.Yaqoob&Serlemitsos powerlawwithaphotonindexΓ dominatesintheenergyband (2000).The largewidth andmarginaldetectionofthe line (see 1 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 5 Table 5. Parameters of the phenomenologicalmodel fit to the Table 6. Parametersofthe modelwhich includesa blackbody data.Whentheuncertaintiesarenotquoted,thevalueofthepa- todescribethesoftexcess. rameterwasfixed. Period1 Period2 Period3 Period4 Period5 Period1 Period2 Period3Period4 Period5 Blackbody Cut-offpowerlawdescribingSoftexcess T(eV) 78.9+1.5 87.2+0.8 79.0+1.8 78.0+2.0 82.0+1.0 −1.6 −0.7 −1.2 −1.2 −1.0 Γ 2.46+0.14 2.02+0.08 2.37+0.122.36+0.12 2.32+0.10 F∗ 1.93+0.05 1.08+0.01 1.21+0.03 0.94+0.02 1.42+0.02 1 −0.16 −0.08 −0.13 −0.13 −0.11 0.2−1 −0.06 −0.01 −0.03 −0.03 −0.03 E 0.65+0.20 0.52+0.06 0.73+0.240.65+0.16 0.58+0.10 Powerlaw F⋆1 3.60+−10..0184 2.11+−00..1065 2.34+−00..51951.89+−00..4192 2.36+−00..2048 Γ 1.892+0.0071.877+0.0011.787+0.0051.815+0.0051.719+0.004 0.2−1 Sec−o0.n76dpower−0l.a26wdesc−r0i.b41ingha−r0d.35tail −0.39 F∗ 8.48+−00..00506 9.88+−00..00101 7.75+−00..00306 6.52+−00..00206 8.32+−00..00303 2−10 −0.04 −0.01 −0.04 −0.03 −0.02 Γ 1.79+0.02 1.82+0.01 1.67+0.021.73+0.01 1.63+0.01 Gaussianline F2⋆−210 8.76−+−000...210192 10.03−−+000.0..101068.02−+−000...1207216.70−+−000...110201 8.53−+−000...100081 EFe 6.65 6.72−+00..0064 6.65 6.65 6.65 Gaussianline σFe 0.10 0.10 0.10 0.10 0.10 E 6.65 6.65+0.07 6.65 6.65 6.65 n∗∗ 4.56+1.64 4.89+0.37 3.54+1.29 2.49+0.94 2.35+0.90 Fe −0.07 Fe −1.57 −1.10 −1.26 −0.95 −0.85 σ 1.98+0.55 0.31+0.07 0.39+0.220.50+0.92 2.35+2.97 Intercalibrationfactors I⋆FF⋆ee 42.50−−+013.223..98302Int6e.r1c−+−a110l...i430b6ra5t.i2o9n−+−f230a...531c112to3r.1s4−+−230...79591017.14−−+115.2.38.0331 CCij 11..1138−+−+0000....00115600 11..3349−+−+0000....00003355 11..8401−+−+0000....11113200 10..2410−+−+0000....00118812 10..4760−+−+0000....00003466 Cj 1.11−+00..0099 1.33−+00..0055 1.31−+00..00990.37−+00..1100 0.66−+00..0055 Cs 1.72−+00..3311 1.65−+00..1188 – – 1.98−+00..1177 C 0.87+0.05 1.16+0.03 1.38+0.110.97+0.07 1.19+0.37 Statistics Ci 1.32+−00..2045 1.43+−00..1053 –−0.10 –−0.07 1.62+−00..1345 χ2 1445 2659 1543 1799 2161 s −0.24 −0.15 −0.13 Statistics (dof) (1310) (2024) (1515) (1630) (1780) χ2 1305 2174 1360 1594 1832 Statisticswithoutironlinetakenintoaccount (df) (1308) (2022) (1519) (1634) (1778) χ2 1453 2710 1551 1807 2169 nl Statisticswithoutironlinetakenintoaccount (dof) (1311) (2026) (1516) (1631) (1781) χ2 1316 2204 1364 1596 1836 nl (df) (1310) (2025) (1521) (1636) (1780) ⋆Inunitsof10−11ergcm−2s−1. ⋆⋆Inunitsof10−5photonskeV−1cm−2s−1. ⋆in10−11ergcm−2s−1units. ⋆⋆⋆in10−5photonscm−2s−1units. Table 7. Parameters of the model consisting of emission from thecentralsourcereflectedfromtherelativistically-blurredpho- toionizeddiscandajet Figure4)donotallowustofirmlydistinguishionisedandnon- Period1 Period2 Period3 Period4 Period5 ionisediron.Inallotherspectratheadditionofanironlinedoes Directandreflectedemissionfromthecentralsource notsignificantlyimprovethefit.Forcompletenessweshowthe ξ+ 2977+1565 10000+0 3170+336010000+0 2570+2391 degradationofthefitifnolineisincluded.Thecompletelistof F∗ 5.78+−01.17216 4.31+0−.23418 4.59+−01.46303 4.05+−02.64353 3.88+−04.1106 the fit parameters for our phenomenologicalmodel is given in 0Γ.2−1 2.83+−00..0772 2.56+−00..0316 2.56+−00..0589 2.48+−00..0379 2.76+−00..0345 Table5. R(%1) 2.39+−30..9088 20.02+−107.092.51 2.61+−20..0018 11.38−+01.00.969 2.69+−30..4011 −1.00 −2.39 −1.56 −7.11 −1.15 PowerlawdescribingJetemission Γ 1.70+0.03 1.75+0.01 1.53+0.04 1.55+0.05 1.54+0.03 3.2.2. Modellingofthesoftexcess F∗2 8.05+−00..2072 9.23+−00..2052 6.88+−00..6003 5.76+−20..6036 8.02+−20..6082 2−10 −0.57 −0.36 −0.75 −2.80 −0.38 Gaussianline The “soft excess” below 1 keV is typical of Seyfert galaxies. E 6.65 6.73+0.05 6.65 6.65 6.65 Severalcompetingmodelsforthephysicaloriginofthiscompo- Fe −0.06 n∗∗ 3.81+1.57 3.06+0.62 2.28+1.37 0.20+0.93 1.42+0.94 nent are available. In particular, the most simple “phenomeno- Fe −1.56 −0.63 −1.35 −0.20 −0.86 Intercalibrationfactors logical”modelsofthesoftexcessincludea(multi-temperature) C 1.10+0.04 1.19+0.02 1.28+0.05 0.35+0.05 0.66+0.04 blackbodycomponent.However,the originof suchblackbody Cj 0.77+−00..0095 1.17+−00..0042 1.12+−00..0594 0.76+−00..1005 1.08+−00..0052 i −0.09 −0.05 −0.09 −0.10 −0.05 componentisnotclear,becausetheinferredtemperatureisusu- C 1.16+0.21 1.30+0.14 – – 1.35+0.12 s −0.21 −0.14 −0.11 allytoohightobeaccountedforstandardaccretiondisks.More Statistics complicated models include e.g. reflection of external X-ray χ2 1399 2239 1349 1516 1819 emission from the disk. Attempts to fit the soft excess in our (dof) (1410) (2002) (1507) (1542) (1758) XMM-Newtondatawithdifferentmodelsgivethefollowingre- Statisticswithoutironlinetakenintoaccount sults. χ2nl 1405 2262 1353 1516 1822 (dof) (1412) (2004) (1509) (1543) (1759) Ablackbodymodelforthesoftexcessdoesnotprovidean acceptablefittothedata.Best fitparametersforthismodelare +inunitsofergcms−1. listed in Table 6.The disk reflection component calculated us- ⋆inunitsof10−11ergcm−2s−1. ingthekdblurandreflionmodelsofXSPEC(Ross&Fabian ⋆⋆inunitsof10−4photonskeV−1cm−2s−1. 2005; Crummyetal. 2006) providesa better descriptionof the softexcess,thantheblackbodymodel.Amongtheparametersof the reflionmodelwe fit onlythe spectralslope of the power law emission of the central source, fluxes of the direct and re- alsoreportthemeasuredratioRofthereflectedfluxtothetotal flected components, and the ionization parameter of the gas ξ fluxexpressedinpercents. (ξ = 4πF/n , where F is the illuminatingenergy flux, and n One should note, however, that the amount of reflection H H is the hydrogennumberdensity in the illuminatedlayer;in the found in each data set, except for data set number 2, is at the reflion model ξ is restricted to 1 < ξ < 10000 erg cm s−1). level of several percent. Thus, a better fit (compared to black- ParametersofthemodelfitaregiveninTable7.InthisTablewe body)to the data is explainednotbythe carefulaccountof re- 6 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 flectionfeatures,butbyalargecontributionofthedirectpower Table 8. Black body, power law, Seyfert-like component and law illuminating the disk, so that the disk reflection model is GaussianlinefittoPN,JEM-X,SPIandISGRI3C273spectra. not significantly different from the previous phenomenological ThefitwasdoneleavingfreeintercalibrationfactorsforISGRI model. To summarize, we find that neither the black body nor (C), SPI (C ) and JEM-X (C ) with respect to the PN camera. i s j thediskreflectionprovidebetterdescriptionsofthesoftexcess Theparameteruncertaintiescorrespondto1σlevel. thanthe(cut-off)powerlaw. Period1 Period2 Period3 Period4 Period5 Compton reflectionis characterizedby a ”reflectionhump” Blackbody atabout30keV.Tolookforthepresenceofareflectioncompo- F∗ 1.01+0.05 <0.004 0.53+0.03 0.33+0.02 0.79+0.03 1 −0.06 −0.03 −0.03 −0.02 nentin the spectrumwe have also tried to fit the spectra above Powerlaw 2 keV with either a simple powerlaw or a directplusreflected F∗ 5.25+0.09 6.32+0.03 5.86+0.07 4.76+0.04 6.74+0.05 2 −0.09 −0.03 −0.06 −0.05 −0.04 power law (the reflion model of XSPEC) and compare the Seyfertcomponent qualityofthefits. Althoughwe findthatthe additionofthere- F∗ 3.47+0.04 3.86+0.01 2.10+0.03 1.96+0.02 1.71+0.02 s −0.04 −0.01 −0.03 −0.02 −0.02 flectedcomponentatthelevelofabout20%improvesthequality Gaussianline ofthefits,theF-testshowsthattheprobabilitythatthisimprove- n∗F∗e 6.45−+31..5882 3.03−+10..0969 4.45−+12..5714 1.78−+11..7185 2.92−+11..0786 Intercalibrationfactors mentiscoincidentalisatthelevelof10%,sothatnodefinitive C 1.09+0.09 1.26+0.05 1.39+0.10 0.28+0.11 0.65+0.05 conclusionaboutthepresenceofareflectionhumpinthespec- j −0.09 −0.05 −0.10 −0.11 −0.05 C 0.68+0.03 0.80+0.02 1.16+0.09 0.74+0.05 1.05+0.02 trumcanbemade.Oneshouldnotethatthelargestuncertaintyis i −0.03 −0.02 −0.09 −0.05 −0.02 C 1.01+0.18 0.97+0.11 – – 1.42+0.12 duetothefactthattheISGRIinstrumentissensitiveonlyabove s −0.19 −0.11 −0.12 Statistics ∼ 20 keV, while XMM-Newton data are available only below χ2 1340 2342 1383 1561 1913 ∼10keVand,infact,thereisadegeneracybetweenthenormal- (dof) (1311) (2026) (1531) (1565) (1781) ization of the reflection hump and the ISGRI – XMM-Newton Statisticswithoutironlinetakenintoaccount intercalibrationfactor. χ2 1345 2350 1389 1562 1918 nl (dof) (1312) (2027) (1532) (1566) (1782) A combination of black-body and reflection components, suggested as a best-fit model for the Beppo-SAX data by ⋆inunitsof10−11ergcm−2s−1. Grandi&Palumbo(2004)doesnotchangetheaboveconclusion ⋆⋆inunitsof10−5photonskeV−1cm−2s−1. (seebelow),leavingthe(cut-off)powerlawtobethebestmodel forthesoftexcess. 3.2.4. Multibandfluxandspectralslopecorrelations 3.2.3. DisentanglingtheJetandSeyfertcontributions Comptonization of big blue bump photons was proposed as a plausible mechanismof productionof the hardX-ray emission Grandi&Palumbo(2004)(GP)haveputforwardastrongclaim from 3C 273 by Walter&Courvoisier (1992). The motivation thatSeyfertand“jet” componentsof the X-rayspectrumcould forthismodelwasalinearcorrelationbetweentheX-rayspec- bedisentangledinthecaseof3C273sothattheobservedvari- tral slope above2 keV and the logarithmof the ratio of the 2– ations of the X-ray spectral shape are successfully described 10keVfluxtotheultravioletandsoftexcesscountratesfoundin by variations in the normalizations of the template “jet” and EXOSAT,GINGAandIUEdata.ThepresenceoftheUVcamera “Seyfert” components. We have tested this hypothesis with on board of XMM-Newton and of the Optical Monitor Camera our data. For this we repeat the analysis of Grandi&Palumbo (OMC) on board of INTEGRAL enables us to test this correla- (2004).First, wefindthe“template”modelsforSeyfertandjet tionbetweentheUVandX-rayfluxesandslopes.Fig.6shows componentsbyfittingsimultaneouslyallthe5datasetswiththe the values of the power law index in 2–10keV band, Γ , plot- 2 samemodelwhichconsistsofahardpowerlaw(jet)andblack- ted against the logarithm of the ratio of X-ray and UV fluxes bodypluspartiallyreflectedcut-offpowerlaw(Seyfertcompo- (F /νF ).4 2−10keV 10eV nent).The best fit is achievedwith the followingparameters:a Figure 6 clearly shows that the correlations seen in blackbodytemperatureT =75eV,aSeyfertcomponentpower EXOSAT/GINGAX-raydataandIUEUV dataarenotpresent bb lawslopeΓ =2.25,areflectioncomponentfractionRC =0.77, in our data set. For completeness we have extended the anal- s ajetpowerlawslopeΓ =1.5,aninclinationofthediskφ=18◦. ysis for the whole set of publicly available XMM-Newton data j We assumed solar abundances of elements and fixed the posi- on3C273toproducethisFigure(fordetailsseeSoldietal.,in tionandwidthoftheironlineatthevaluesfoundinthesecond preparation). dataset: E = 6.65,σ = 0.3keV.Thenwe frozethe above InFig. 6 two importantfactsare apparent.First, the scatter Fe Fe parameters and left only the normalizations of the black body, oftheXMM-Newtondatapointsalongthex-axisismuchsmaller reflection and jet power law componentsvariable. With such a than that of the EXOSAT/GINGAand IUE points. Thismeans “restricted model” we fit each data set separately and find the thatintherecentpasttheX-rayandUVfluxeshavevariedina normalizationsof the three “template” components.The fluxes correlatedway.LefttoppanelofFig.8showsthatthisisindeed ineachcomponentfoundinthiswayarelistedinTable8. thecase(SpearmancorrelationcoefficientisequaltoR = 0.93, withaprobabilityofrandomcoincidenceP =3×10−3).This Afterthisanalysis,andfollowingGrandi&Palumbo(2004), rand is a quite surprising result, because previoussearches for such we study the correlations between e.g. Seyfert componentand correlationsgave negativeresults. Howeveranothercorrelation iron line strength or between fluxes in Seyfert and black body typicalforSeyfertgalaxiesbetweenthehardX-rayspectralin- component, which were found in the Beppo-SAX data. Fig. 5 dexΓ andtheX-rayflux(Zdziarskietal.(2003),andreferences (whichisananalogofFig.1ofGP)showsthatnosuchcorrela- 2 tionsarepresentinourdataset. Wediscusstheimplicationsof 4 We have extrapolated the UV spectrum obtained with the XMM- thisresultbelow. Newtonopticalmonitorupto10eVusingthetime-averagedUVspec- trumof3C273. M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 7 7 1 8 6 6 0.5 4 5 2 0 4 0 1 2 3 4 1 2 3 4 1 2 3 4 Fig.5.Thedependanceofthe0.2-2keVblack-bodyemission(left),the2-10keVjetlikeflux(middle),andtheironlineflux(right) onthe2-10keVSeyfert-likeflux.DatapointsaretakenfromTable8 GINGAandIUEdataand“appearance”ofX-ray–UVfluxcor- relationsnotdetectedbefore. Having found a previously undetected correlation between UVandX-rayflux,wehavesearchedforthepresenceofaddi- tionalcorrelations,suchasbetweenthe0.2-1keVand3-10keV fluxes(toprightpanelofFig.8),0.2-1keVand20-60keVfluxes (bottom left panel of Fig. 8) and 20-60 keV and 60-100 keV fluxes(bottomrightpanelofFig.8).Onecanseeaclearcorrela- tionbetween20-60keVand60-100keVfluxes,indicatingthat bothbandsaredominatedbyoneandthesamephysicalcompo- nent(e.g.powerlawemissionfromthejet).Ontheotherside,no or only marginal correlation in the two other cases shows that severalphysicalmechanismscontributetotheemissioninthese energybands. 4. Long-termevolutionoftheX-rayspectrumof 3C273 TotesttheevolutionoftheX-rayspectrumoverthelast30year periodwehavecollectedthe archivalX-raydatafromdifferent missions. The evolution of the hard (2-10 keV) spectral index found from the archival data5 (Tu¨rleretal. 1999) is shown in Fig. 9. One clearly sees both significant short-term variations of the spectral index with a scatter of ∆Γ ≃ 0.1 and a long- Fig.6. 2 – 10 keV spectral photon index of the phemenologi- termtrendofsofteningofthespectrumbyasmuchas∆Γ≃ 0.4 cal model as a functionof the logarithmof the ratio of the 2– overthelast30yr.Itisinterestingtonotethatduringour2003- 10keV to ultraviolet fluxes. Old Ginga and EXOSAT data are 2005 monitoring campaign the source has reached its histor- shownwithredtriangles.Thelineindicatesthebestcorrelation ically softest hard X-ray spectrum characterized by a photon foundinWalter&Courvoisier(1992).NewXMM-Newtondata index Γ = 1.82 ± 0.01. It is possible that still softer spectral areshownwithblackcircles.WehaveextrapolatedtheUVspec- states were found at the very beginning of X-ray observations trumobtainedwiththeXMMopticalmonitorupto10eVusing (UHURU, ARIEL5/CandOSO8/A data points),butthe uncer- thetime-averagedUVspectrumof3C273. tanityofthosemeasurementsaretoolargetobeconclusive. 5. Discussion therein),isnotobserved(seeFig.7).Asimilarresultwasfound byPageetal.(2004). Mostofthetheoreticalmodelsof3C273suggestthatthesource Second,thevaluesofthehardX-rayspectralindexinXMM- spectrumcontainstwomajorcontributions,onefromapc-scale Newtonobservationsaresystematicallyhigherthanthoseofthe jetandtheotherfromaSeyfert-likenucleus.Theverylowmil- EXOSAT/GINGAdata. This indicatesthatthe spectralstate of limeter flux measured during our observations suggested that the sourcehasevolvedoverthe 30yearsofX-rayobservations the jet contribution has reached its historical minimum (see (inthenextsectionwediscussthisindetails).Thismaybelinked to the “disappearance” of the correlations seen in EXOSAT, 5 Availablefrom3C273’sDatabase:http://isdc.unige.ch/3c273/ 8 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 Fig.8.Flux-fluxplotsfordifferentenergybands.OntherightbottompanelonedayaveragedINTEGRALdataareshown.Toreduce thescatterthedatawererebinnedbybuildingtheaverageofthreeadjacentoriginaldatapointsinxdirection. Tu¨rleretal.(2006)).However,thisdidnotresultintheclearap- lation is seen at such low iron line flux values. However, it is pearanceoftheSeyfert-likecontribution.Thismeanseitherthat importanttonotethatnodirectcomparisonbetweenourFig.5 theSeyfert-likecontributionhasdecreasedsimultaneouslywith andFig.1ofGrandi&Palumbo(2004)canbedone,becauseof thejetcontribution(which,becauseofthedifferenceinthesize thedifferencesinthetemplatemodelsderivedfromBeppo-SAX scalesandproductionsitescanhappenonlybyachancecoinci- andINTEGRAL-XMM-Newtondatasets. dence),orthattheobservedemissionjustcannotbedecomposed In fact, the analysis of the secular evolution of the source intoSeyfertandjetcontributions. presented aboveshows that both Seyfertand jet spectra evolve withtime.ThiscanbeillustratedbytheresultspresentedinFig. Our data show thatthe separationof Seyfertand jet contri- 6, wherewe plotthehardX-rayspectralslopeas a functionof butionstothespectrumcouldnotbedonebyfixinga”template” theratiooftheX-raypowerlawfluxtotheUVflux.Thisratio for each of them. Namely, although the typical ranges of vari- isapartfromaconstantfactortheratiooftheComptonizedflux ations of soft and hard X-ray fluxes in our data set are simi- to the sourcephotonfluxin a modelin whichthe X-raypower lar to the ones of Beppo-SAX data set, the ”template” models lawisduetotheComptonizationofthe softphotonsbyather- for the Seyfert, jet and black body components found in our malelectronpopulation.Thepointsgivenbytheredcrosseswith data set are significantly different. In particular, the additional largeerrorbarswereobtainedbytheEXOSATandIUEmissions black bodycomponentis notrequiredby the data and thus the in the 80-s and were analysed by Walter&Courvoisier (1992) resultingblackbodyfluxshownintheleftpanelofFig.5issys- who deduced that in the frame of the thermal Comptonization tematicallylowerthanthevaluesreportedinGrandi&Palumbo model, the electron temperature was about 1MeV or some- (2004). Our data show that the correlation between the black what less. They also deduced that the optical depth of the bodyandSeyfertcomponentsisabsentatsuchlow blackbody Comptonizing medium was 0.1–0.2 and that it covered a cou- fluxvalues.Thesameistrueforthenon-observationofthecor- pleofpercentofthesoftphotonsource. relationbetweentheSeyfertcomponentfluxandthestrengthof theironline:sincetheadditionoftheironlineisnotstatistically The XMM-Newton data shown in Figure 6 by the black required in any data sets except for the second one, no corre- pointswithsmallerrorbarsclearlydonotmatchthecorrelation M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 9 4 3 2 1 2 UHURU EXOSAT ARIEL5/C GINGA XMM-Newton/PN OSO8/A ASCA This Work HEAO-1 RXTE 1.8 HEAO2/MPC BeppoSAX/MECS BALL/AIT XMM-Newton/PN 1.6 1.4 1.2 1 1970 1980 1990 2000 Time Fig.9.Evolutionofthefluxdensityat1keV,asextrapolatedfromthefitdoneabove2keV,in10−2photons/keV/cm2/s(toppanel) andhard X-rayspectralindex(bottompanel)overthe 30 yearsofobservations.UHURU, Ariel 5/C, OSO 8/A,HEAO-1,HEAO- 2/MPC, and BALL/AIT data points are taken from Malagutietal. (1994); EXOSAT and Ginga data are taken from Turneretal. (1990);ASCAdataarefromCappietal.(1998);RXTE,Beppo-SAX,andpartofXMM-NewtondataarefromSoldietal.(inprepa- ration). foundinearlieryears.Thisisparticularlytrueforthespectraob- Comptonization region when the source is very weak. The re- tainedduringperiods1and2,whenthesoftX-rayemissionwas centdataalsoshowthatthehardX-rayfluxiscorrelatedwiththe muchbrighterthanattheotherepochs,whiletheultravioletflux softflux(seeFigure8),aneffectnotobservedatearlierepochs was similar. This indicates a much harder blue bump for these (Courvoisieretal.1990). epochs.MorephotonswerethereforeavailabletoComptoncool the hot electrons than inferred from the ultraviolet flux shown The correlation between the optical/UV and the X-ray flux in Figure 5. This therefore leads to the softening in the X-ray ischaracteristicforSeyfertgalaxies.Thisrelationshipisparticu- emission.Alternatively,onecanalsoinferamodificationofthe larlyclearinthelong-termbehaviourofNGC5548(Uttleyetal. 2003), but was also seen in other Seyfert galaxies like for 10 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 emission from an electron distribution cooled via synchrotron andinverseComptonenergylosswhichcanproduceonlyspec- trawithΓ≥1.5. Thisproblemcan,however,berelaxedifoneassumesthatin additiontoelectronstheemission regioncontainsalso protons. IfthemainsourceoftheseedphotonsfortheinverseCompton scatteringisUVphotonsfromthebigbluebump,thehardX-ray inverseComptonemissionoriginatesfromapopulationofelec- tronswith moderateenergies E ∼ 1−10 MeV. In this energy e rangecoolingcouldbedominatedbyionization,ratherthanin- verse Compton/synchrotronlosses. The Coulomb loss leads to the hardeningof the electron spectrum, compared to the injec- tion spectrum by ∆Γ ≃ −1. This will result in a hardeningof e the inverse Compton spectrum below the so-called ”Coulomb break” (see e.g. Aharonian (2004)). A simple estimate shows that if the proton density in the gamma-ray emission region is n ∼ 1010cm−3, the Coulomb break is situated in the hard X- p ray band which can provide the explanationof the harder than Γ= 1.5spectrumofthesourceinlate70-s.Highmatterdensity in the emission region obviously favors the ”compact source” modelscomparedtothejetmodels:theassumptionaboutalarge (1010cm−3)matterdensitiesatthepcdistancesfromthenucleus neededtoexplainthehardspectruminthejetmodelframework looks somewhat extreme (for comparison, the density of dust andmoleculargasintheCentralMolecularZone(CMZ)ofthe Fig.7.ThedependenceofthehardX-rayphotonindex(derived Galaxyis estimatedto be 104 cm−3, see e.g.Morris&Serabyn from fitting the data in 2-100 keV energy range with a single (1996)).Theconclusionthattheobservedevolutionofthehard powerlaw,seetable4)onthe2–10keVX-rayflux. X-ray spectral index favors the compact source models should also be taken with caution.The reasonfor this is thatthe pres- instance NGC 4051 (Shemmeretal. (2003), and references ence of a Seyfert-like contribution to the hard X-ray spectrum therein). The data of 3C 273 presented here do also show the canaffectthevalueofthespectralindex. appearance of such a trend (see Fig. 8), apparently suggesting Ifthegamma-rayemissionof3C273originatesinacompact aSeyfert-likebehaviorof3C273in2003–2005.However,itis sourcethesynchrotronradiationfromelectronswhichupscatter clearfromtheabovediscussionthatinthecaseof3C273con- thebluebumpphotonstotheGeVenergybandcouldcontribute clusions of this sort should be taken with caution: there is no alsotothesoftX-rayspectrumofthesource.Themagneticfield clear way to disentangle the Seyfert and jet contributions. For strength in the γ-ray emission region is poorly constrained by example, another characteristic of Seyfert galaxies, a correla- theobservations(see,however,Courvoisieretal.(1988)forre- tionbetweenthehardX-rayspectralindexΓ andtheX-rayflux strictionsonthemagneticfieldstrengthduringflaringactivityof 2 (Zdziarskietal.(2003),andreferencestherein),isnotobserved thesourceinthelate80-s).Assumingthattheenergydensityof (seeFig.7). magneticfieldinthecompactemissionregioniscomparableto The secularevolutionofthe hardX-rayspectralindexover theenergydensityofradiation, the last 30 years,clearly seen in Fig. 9, showsthatthe jet con- tributionalsoevolveswithtime.WhetherthehardX-rayfluxis U ≃2×102 L 1016cm 2erg/cm3 (1) generated in the jet, or it is due to the nuclear related physics ph "1046erg/s#" R # (i.e. from the ”compact source” (Lightman&Zdziarski 1987)) isstillcompletelyundecided.”Jet”and”compactsource”mod- (weestimatethesizeofemissionregiontobeR∼1016cmfrom elsdifferintheirapproachtotheso-calledcompactnessproblem. the observedintra-dayvariabilityof the source)oneobtainsan Theessence ofthecompactnessproblemisthatthenaivelyes- estimate timatedopticaldepthofthesourceforγ-rayswithenergiesE , γ L 1/2 1016cm such that pairs are produced in the interactions with soft pho- B≃102 G (2) tonsofenergiesǫ ≥ 2m2c4/E ,ishigherthan1.The”compact "1046erg/s# " R # s e γ source”modelsself-consistentlyaccountfortheeffectofthepair Thesynchrotronradiationfrom1-10GeVelectronsinsuchmag- productioncascadeinthesource(Lightman&Zdziarski1987). neticfieldisemittedattheenergies The”jet”modelsgetridofthecompactnessproblembyassum- ingrelativisticbeamingoftheγ-rayemission. B E 2 Γ ≃T1h.e4otboseΓrv≃ed1e.8vooluvteironthoefltahseth3a0rdyXea-rrsaycosupledcthraellpintdoexdifsrtoinm- ǫs =3(cid:20)102G(cid:21)(cid:20)10GeeV(cid:21) keV (3) guishbetweenthetwopossiblemodelsofthehigh-energyemis- Thenon-observationofacut-offintheGeVband(Collmaretal. sion. Thepointis thatalthoughthe present-dayvaluesΓ > 1.5 2000) impliesthat the high-energycut-offin the electronspec- areeasilyexplainedinbothmodels,theveryhardspectrumob- trumis atleastat energiesE ≥ 1GeV which,in turn,implies e servedisthelate70-sisnot.Inbothmodelsthegamma-raysare thatthecut-offinthesynchrotronspectrumshouldbeatleastin producedviainverseComptonemissionfromhigh-energyelec- the soft X-rayband.Taking into accountthatthe bestfit to the trons.ThehardspectralindexΓ≃1.4observedin70-sand80-s ”softX-rayexcess”inourdataisgivenbyasimplecut-offpower (see Fig. 9) is too hard for the optically thin inverse Compton lawmodel,wefindthatitispossiblethattheobservedsoftX-ray

ebook img

2003--2005 INTEGRAL and XMM-Newton observations of 3C 273 PDF

0.34 MB·English
by  M. Chernyakova
#additional_collections #journals #arxiv
Save to my drive
Quick download
Download
Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview 2003--2005 INTEGRAL and XMM-Newton observations of 3C 273

Astronomy&Astrophysicsmanuscriptno.6285 (cid:13)c ESO2008 February5,2008 2003–2005 INTEGRAL and XMM-Newton observations of 3C 273 M.Chernyakova1,2,A.Neronov1,2,T.J.-L.Courvoisier1,2,M.Tu¨rler1,2,S.Soldi1,2,V.Beckmann3,P.Lubinski4,1, R.Walter1,2,K.L.Page5,M.Stuhlinger6,R.Staubert7,andI.M.McHardy8 1 INTEGRALScienceDataCenter,Chemind’E´cogia16,1290Versoix,Switzerlande-mail:[email protected] 2 GenevaObservatory,UniversityofGeneva,51ch.desMaillettes,CH-1290Sauverny,Switzerland 3 NASAGoddardSpaceFlightCenter,ExplorationoftheUniverseDivision,Code661,Greenbelt,MD20771,USA 7 4 NicolausCopernicusAstronomicalCenter,Bartycka18,00-716Warszawa,Poland 0 5 UKSwiftScienceDataCentre,Dept.ofPhysics&Astronomy,UniversityofLeicester,UniversityRoad,LeicesterLE17RH 0 6 XMM-Newton Science Operations Centre, European Space Astronomy Centre (ESAC) - Villafranca, P.O. Box 50727 - 28080 2 Madrid-Spain 7 Institutfu¨rAstronomieundAstrophysik-Astronomie,72076Tu¨bingen,Germany n 8 SchoolofPhysics&AstronomyUniversityofSouthamptonHighfield,SouthamptonSO171BJ,U.K. a J 9 ABSTRACT 2 Aims. Theaimofthispaperistostudytheevolutionofthebroadbandspectrumofoneofthebrightestandnearestquasars3C273. 1 Methods. Weanalyzethedataobtainedduringquasi-simultaneousINTEGRALandXMM-Newtonmonitoringoftheblazar3C273 v in2003–2005intheUV,X-rayandsoftγ-raybandsandstudytheresultsinthecontextofthelong-termevolutionofthesource. 5 Results.The0.2-100keVspectrumofthesourceiswellfittedbyacombinationofasoftcut-offpowerlawandahardpowerlaw. 2 Noimprovement ofthefitisachievedifonereplacesthesoftcut-off power lawbyeitherablackbody, oradiskreflectionmodel. 8 During the observation period the source has reached the historically softest state in the hard X-ray domain with a photon index 1 Γ=1.82±0.01.ComparingourdatawithavailablearchivedX-raydatafrompreviousyears,wefindasecularevolutionofthesource 0 towardsofterX-rayemission(thephotonindexhasincreasedby∆Γ ≃ 0.3−0.4overthelastthirtyyears).Wearguethatexisting 7 theoreticalmodelshavetobesignificantlymodifiedtoaccountfortheobservedspectralevolutionofthesource. 0 / Keywords.quasars–individual–3C273 h p - o 1. Introduction spectrum, as proposed by Grandi&Palumbo (2004), is not r observedinourdataset. t 3C 273 is a radio loud quasar, with a jet showing superlumi- s a nalmotion,discoveredattheverybeginningofquasarresearch. : Beingoneofthebrightestandnearest(z=0.158)quasars,3C273 Walter&Courvoisier (1992) have found an anticorrelation v wasintensivelystudiedatdifferentwavelengths(seeCourvoisier between the X-ray spectral slope and the ratio of X-ray to ul- i X (1998)forareview). traviolet (UV) luminosity of 3C 273 using quasi-simultaneous datafromEXOSAT,GINGAandIUEobtainedinthe80-s.They r In the X-ray band 3C 273 is thought to be quite a unique a haveinterpretedthiscorrelationasevidencethattheX-rayemis- source in the sense that in its X-ray spectrum the “blazar-like” sionisduetothermalComptonizationoftheUVflux.Thepres- emissionfromthepc-scalejetmixeswiththe“nuclear”compo- ence of optical monitors on-board of both XMM-Newton and nent, which reveals itself through the presence of a large blue- INTEGRAL enables us to analyze correlations between the big bump and a so-called “soft excess” below ∼ 1 keV. Both are blue bump variability and the variability in X-rays. Repeating typicalforemissionfromthenucleiofSeyfertgalaxies.During the analysis of Walter&Courvoisier (1992) with our data set the observationperiod discussed in this paper the radio to mil- wedonotfindthepreviouslyobservedX-rayslope–UV/X-ray limetercomponent,usuallyassociatedwithajet,hasbeeninits flux correlation. The disappearance of this correlation is even loweststate.Thus,ourdataprovideanopportunitytostudythe moresurprisingbecauseoftheappearanceofapreviouslyunob- “nuclear”componentinmoredetailsandtotesttheoreticalmod- servedcorrelationbetweentheUVandtheX-rayfluxfromthe elsfortheemissionof3C273. system. Taking into accountthe fact that the typicalvariability Surprisingly, in spite of the low level of the “jet-like” timescalesintheUVandtheX-raybandsaredifferent(months contribution, our data do not reveal the presence of typical inUVanddaysinX-ray),truesimultaneityoftheobservations “Seyfert-like”featuresinthespectrum.Infact,the0.2-100keV couldbeimportant.OurXMM-NewtonandINTEGRALobserva- spectrum is well fitted by a simple model composed of two tionsarethefirstobservationsinwhichUVandX-raydatawere power laws or a combination of a hard power law with a soft takenatthesametime. cut-off power law. We find that more sophisticated models, which include a black-body for description of the soft excess, It is possible, however, that appearance/disappearance of or a disk reflection component give comparable or worse the UV–X-ray correlation reflects a real evolution of the spec- fits to the data, than the above simple model. Moreover, the tral state of the source during the last 30 years. Collecting the “disentanglement” of Seyfert and jet components of the X-ray archivalX-raydata forabout30 yearsof observations,we find thatthe sourceindeedevolvestowardsa softerX-raystate. We Sendoffprintrequeststo:M.Chernyakova findthatthesourcehasreacheditshistoricallysoftestX-raystate 2 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 Table1.INTEGRALobservationsof3C273. Table2.XMM-Newtonobservationsof3C273. Period Date Revolution OnTime PNExposureTime Period T Rev OnTime ISGRI/SPI/JEM-X (ks) (ks) start (ks) Exposure(ks) 1 2003-01-06 563 8.95 5.97 1a 2003-01-05 28 120.8 2 2003-07-07 655 58.56 40.60 1b 2003-01-11 30 10.6 3 2003-12-14 735 8.91 5.77 1c 2003-01-17 32 106.8 4 2004-06-30 835 62.91 13.93 1 182.5/182.4/12.4 5 2005-07-10 1023 28.11 19.33 2a 2003-06-01 77 57.0 2b 2003-06-04 78 52.6 2c 2003-06-16 82 33.5 2e 2003-07-06 89–90 227.1 2f 2003-07-18 93 20.5 2g 2003-07-23 94 14.0 scientific analysis (OSA) package 5.1 distributed by ISDC1 2 308.9/310.5/40.3 (Courvoisieretal.2003b). 3 2004-01-01148–149 26.4 152.3/NA/13.8 Mostofthetime,observationsweredoneinthe5×5dither- 4 2004-06-23 207 92.7 67.1/NA/29.2 5a 2005-05-28320–321 278.3 ing mode. The staring mode was used only during revolutions 5b 2005-07-09 334 102.5 30,148,and149.ItisthereforenotpossibletouseSPIdatadur- 5 269.1/367.2/68.7 ingthethirdperiod.TherearealsonoSPIdataduringthefourth period,astheinstrumentwasintheannealingmodeduringthis observation. Fortheanalysisofthe INTEGRAL/JEM-Xobservationswe with photonindexΓ = 1.82±0.01in June 2003.Thisspectral have selected science windows for which the source was less slope is (by ∆Γ ≃ 0.3−0.4) softer than the typical slope ob- than 3.5◦ from the center of the field of view. During the first served in the 70-s and 80-s. This is definitely larger than the three periods JEM-X 2 was operating, while during the fourth typicalshorttime scale fluctuationsof the slope which are ata andfifthperioditwasswitchedoff,anddataweretakenwiththe levelof∆Γ≃0.1. JEM-X1telescope. It is important to note that spectral evolution of the source To minimize the influence of the flares in the surrounding should not be ignored in theoretical modelling of nuclear and backgroundofthechargedparticleswehavenotusedpointings jet emission from 3C 273. For example, the very hard spectral in which the detector countrate exceeds the mean of the given stateofthesourceinthelate70-shasledLightman&Zdziarski periodbymorethan5%. (1987) to the conclusion that “compact source” type models dominated by pair production can not be applied to 3C 273 in spiteofitshighapparentcompactness.However,thisconclusion 2.2.XMM-Newtonobservations isnotapplicableforthepresentstateofthesource.Similarly,the mostpopularmodelforthehardX-rayspectrumofthesource, ThelogoftheXMM-Newtonobservationsanalyzedinthispaper inverseComptonemission from the pc-scale jet, is satisfactory ispresentedinTable2.Amongalltheobservationsofthesource forthepresentstateofthesource,butcannotexplaintheΓ≃1.4 donewithXMM-Newtonwehaveselectedthosethatweredone spectralindexobservedin70-s. whileINTEGRALwasobserving. The paper is organizedas follows: in Section 2 we present thesequencesofobservationsandmethodsusedfordatareduc- TheXMM-NewtonObservationDataFiles(ODFs)wereob- tion and analysis. In section 3 we present the results obtained, tained from the on-line Science Archive2. The data were pro- anddiscussthem in Section4. We thengivea summaryof our cessed and the event-lists filtered using  within the analysisinthelastpartofthepaper. ScienceAnalysisSoftware(),version6.5andthemostup-to- datecalibrationfiles. AllEPICobservationswereperformedin theSmallWindowMode.Inallobservationsthesourcewasob- servedwiththeMOS1,MOS2andthePNdetectors.AsthePN 2. ObservationsandDataAnalysis detector,incontrasttoMOS1andMOS2,isnotaffectedbypile- up in our observations we concentrate our analysis on the PN 2.1.INTEGRALobservations data only. Patterns 0-4 were used in the analysis. Background SincethelaunchofINTEGRAL(Winkleretal.2003)onOctober spectra were extracted from areas of the sky close to 3C 273 17, 2002, 3C 273 was regularly monitored. Details of the data showingnoobvioussources.ARFGENandRMFGENwerethen usedinthecurrentanalysisaregiveninTable1.Thepreliminary runtoproducethecorrespondingARFandRMFfiles. analysisofthe2003datawasdonebyCourvoisieretal.(2003a). Opticalandultravioletfluxeswereobtainedwiththeoptical AdetailedstudyofthemultiwavelengthspectrumofJune2004 monitor(OM)ofXMM-Newtonwithomichainfollowingthein- waspresentedbyTu¨rleretal.(2006). structionsoftheSASWatchoutpage3andusingthevaluespro- DuringtheINTEGRALobservations3C273wasallthetime vided for an AGN spectral type. The resulting magnitudes are inaverylowstate,withafluxofseveralmilliCrabs(seeSection given in Table 3. No OM data for the selected filters are avail- 3.2).Toobtainbetterstatisticsforspectralanalysiswehavecom- ableforPeriod3. bineddatatakenwithinafewweeks,whichresultedinfivedata sets. The combined exposure times for each data set and each instrumenton-boardofINTEGRALarelistedinTable1. 1 http://isdc.unige.ch/?Soft+download The data for all INTEGRAL instruments were analysed 2 http://xmm.vilspa.esa.es/external/xmm data acc/xsa/index.shtml in a standard way with the latest version of the off-line 3 http://xmm.esac.esa.int/sas/documentation/watchout/uvflux.shtml M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 3 3 2 1 0 1 0 -1 Fig.1.toppanel:2003–2005IBIS/ISGRIlightcurvesof3C 273in20-60keV (uppercurve,black)and60-100keV (lowercurve, red).Thedataarebinnedin1-dintervals.Foreachtimeframethedategivenatthebottomcorrespondstothefirstdatapoint,tick intervalis1day.bottompanel:HardnessratioF /F . (60−100keV) (20−60keV) Table3.OpticalandUVmagnitudesof3C273duringeachob- 30 servationalperiod. 20 Period1 Period2 Period4 Period5 V 12.53±0.01 12.69±0.01 12.57±0.01 10 B 12.95±0.01 12.86±0.01 U 11.62±0.01 11.76±0.01 11.69±0.01 UVW1 11.32±0.01 11.28±0.01 11.44±0.01 11.30±0.01 0 UVM2 11.34±0.01 11.37±0.01 11.19±0.01 UVW2 11.14±0.01 11.36±0.01 11.16±0.01 0.2 0.15 3. Results 0.1 06.01.03 07.07.03 14.12.03 30.06.04 20.07.05 3.1.TimingAnalysis Fig.2.toppanel:2003–2005XMM/PNlightcurvesof3C273in In the top panel of Figure 1 we show the 3C 273 IBIS/ISGRI 0.2–1keV(uppercurve,black)and3–10keV(lowercurve,red). lightcurves in the 20-60 keV (black points) and in the 60-100 The data are binned in quarter of day intervals. bottom panel: keV(redpoints)energyranges.Inthisfigurethedataarebinned HardnessratioF /F . (3−10keV) (0.2−1keV) inonedayintervals.ISGRIdatashowthatthehardX-rayemis- sionfromthesystemisvariableonatleastaonedaytimescale. Inordertoinvestigatethenatureofthisvariabilitywehavepro- ducedthe“flux-flux”diagramforthe20-60keVand60-100keV energybands,shownin thebottomrightpanelofFig. 8. Large ofthe sourcein the20-100keV bandis mostlydueto changes error bars of the single-day flux measurementsdo not allow to inthenormalization,butnotintheshapeofthespectrum. seethegeneraltrendofthedependencebetweenthe20-60keV Fig.2showstheXMM-Newtonlightcurvesin0.2–1keV(up- and 60-100 keV flux. However, rebinning the data by building per curve, black) and 3–10 keV (lower curve, red) bands. One the average of three adjacent (along the 20-60 keV flux axis) can see significantchangesof the 0.2–1keV flux fromdata set originaldata pointsin the flux-flux plotwe find that the whole to data set, but the variabilityat the time scale of single obser- data set has a roughly linear behaviour (Spearman correlation vation, ∼ 10 ksec is negligible. The plot of the hardness ratio coefficient is equal to R = 0.89, with a probability of random (bottom panel of Fig. 2) shows that the 0.2-10 keV spectrum coincidenceP =7×10−3).Thisindicatesthatthevariability hardenstowardtheendoftheobservationperiod. rand 4 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 Table4.Parametersofthesinglepower-lawmodelin2-100keV energyrange.Thefitwasdoneleavingfreeintercalibrationfac- torsforISGRI(C),SPI(C )andJEM-X(C )withrespecttothe Period #1 i s j PNcamera.Theparameteruncertaintiescorrespondto1σlevel. Period #2 Period1 Period2 Period3Period4Period5 Period #3 Powerlaw Γ 1.81+0.01 1.84+0.01 1.70+0.011.75+0.011.66+0.01 Period #4 −0.01 −0.01 −0.01 −0.01 −0.01 F∗ 8.79+0.1710.10+0.048.04+0.136.72+0.108.54+0.09 2−10 −0.16 −0.07 −0.12 −0.09 −0.09 Period #5 Intercalibrationfactors C 1.11+0.09 1.35+0.05 1.33+0.090.30+0.110.67+0.05 j −0.09 −0.05 −0.09 −0.11 −0.05 C 0.92+0.05 1.22+0.03 1.42+0.110.98+0.081.23+0.04 i −0.05 −0.03 −0.11 −0.07 −0.04 C 1.39+0.26 1.49+0.16 – – 1.67+0.15 s −0.25 −0.16 −0.14 Statistics χ2 920 1673 950 1139 1419 (dof) (920) (1632) (1153) (1205) (1390) ⋆Inunitsof10−11ergcm−2s−1. 0.001 0.01 0.1 1 10 100 Energy, keV 11 0.9 Fig.3.Broadbandspectraof3C273foreachperiod.Dataarefit- 0.8 tedwiththesimplephenomenologicalmodeldescribedinTable 0.7 5.Thenewdataarecomparedtotheaverageflux(greyline)and the 1-σ variability range (grey area) based on historic data of 0.6 3C273(Tu¨rleretal.1999). 0.5 0.4 3.2.SpectralAnalysis OurINTEGRALandXMM-Newtonobservationsprovideavery 0.3 broadband(5decadesinenergy)setofquasi-simultaneousdata which spans from the UV (∼ 1 eV) up to the soft gamma- 3 ray (∼ 100 keV) band. The spectra of the source for all the 2 1 five selected data sets are shown in Fig. 3. Taking into ac- 0 count a wide range of theoretical models proposed for the X- -1 ray spectrum of 3C 273we adopta “simple to complex”strat- -2 -3 egy in the spectral analysis and attempt first to find a “mini- mal model” which can provide a satisfactory representation of 4 5 6 7 Energy, keV thedataandthenconsiderhowchangingand/oraddingvarious theoreticallymotivatedcomponentsimprovesorworsensthefit. The hydrogen column density was fixed to the galactic value N =1.79×1020cm−2(Dickey&Lockman1990). Fig.4. A 3C 273 spectrum in the 4-7 keV energy band as ob- H served with XMM/PN in July 2003 (second data set). Data are fitted with a single power law. A broad iron line is marginally 3.2.1. Phenomenologicalmodel visibleintheresiduals(seetext). Datainthe2–100keVenergyrangearewellfittedbyasimple powerlaw,seeTable4.InouranalysiswehavelefttheISGRI, SPIandJEM-Xintercalibrationfactorsfreetovarywithrespect below 2 keV and describes the so-called ”soft excess”. A hard tothePNcamera.ThesignificantvariationoftheJEM-Xinter- power law with photon index Γ dominates in the hard X-ray 2 calibrationfactorreflectsaknownchangeofthegainparameters band. An F-test shows that a slightly better fit to the data can oftheinstrument.Itis possibleto fitallthedatawiththe same beachievedifoneusesacut-offpowerlawinsteadofthesingle SPI(C = 1.41)andISGRI(C = 1.19)intercalibrationfactors, powerlawforthedescriptionofthesoftemission. s i but we have preferred not to do this, as the XMM-Newton and Surprisingly,wefindthatalreadythissimplemodelprovides INTEGRAL data are only quasi-simultaneous,and the different a verygoodfit to the combinedINTEGRAL and XMM-Newton valueofthefluxmeasuredbyXMM-NewtonandINTEGRALcan data in all the observations except the second one. In the sec- beduetothephysicalchangeofthesourcefluxonaonemonth ond data set the addition of a broadredshiftedemission line at scale.Thefitoftheseconddatasetcanbeimprovedbythead- E =6.65±0.07keVintherestframeoftheobjectmakesthe Fe ditionofabroademissionlineasdiscussedbelow. fit acceptable.Thisline can be identifiedwith a fluorescentK α The simplestsatisfactoryfit tothe 0.2-100keVdata canbe ironline.Abroadironlinewasseeninpreviousobservationsof achieved with a model consisting of two power laws. A soft 3C273bydifferentinstruments,seee.g.Yaqoob&Serlemitsos powerlawwithaphotonindexΓ dominatesintheenergyband (2000).The largewidth andmarginaldetectionofthe line (see 1 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 5 Table 5. Parameters of the phenomenologicalmodel fit to the Table 6. Parametersofthe modelwhich includesa blackbody data.Whentheuncertaintiesarenotquoted,thevalueofthepa- todescribethesoftexcess. rameterwasfixed. Period1 Period2 Period3 Period4 Period5 Period1 Period2 Period3Period4 Period5 Blackbody Cut-offpowerlawdescribingSoftexcess T(eV) 78.9+1.5 87.2+0.8 79.0+1.8 78.0+2.0 82.0+1.0 −1.6 −0.7 −1.2 −1.2 −1.0 Γ 2.46+0.14 2.02+0.08 2.37+0.122.36+0.12 2.32+0.10 F∗ 1.93+0.05 1.08+0.01 1.21+0.03 0.94+0.02 1.42+0.02 1 −0.16 −0.08 −0.13 −0.13 −0.11 0.2−1 −0.06 −0.01 −0.03 −0.03 −0.03 E 0.65+0.20 0.52+0.06 0.73+0.240.65+0.16 0.58+0.10 Powerlaw F⋆1 3.60+−10..0184 2.11+−00..1065 2.34+−00..51951.89+−00..4192 2.36+−00..2048 Γ 1.892+0.0071.877+0.0011.787+0.0051.815+0.0051.719+0.004 0.2−1 Sec−o0.n76dpower−0l.a26wdesc−r0i.b41ingha−r0d.35tail −0.39 F∗ 8.48+−00..00506 9.88+−00..00101 7.75+−00..00306 6.52+−00..00206 8.32+−00..00303 2−10 −0.04 −0.01 −0.04 −0.03 −0.02 Γ 1.79+0.02 1.82+0.01 1.67+0.021.73+0.01 1.63+0.01 Gaussianline F2⋆−210 8.76−+−000...210192 10.03−−+000.0..101068.02−+−000...1207216.70−+−000...110201 8.53−+−000...100081 EFe 6.65 6.72−+00..0064 6.65 6.65 6.65 Gaussianline σFe 0.10 0.10 0.10 0.10 0.10 E 6.65 6.65+0.07 6.65 6.65 6.65 n∗∗ 4.56+1.64 4.89+0.37 3.54+1.29 2.49+0.94 2.35+0.90 Fe −0.07 Fe −1.57 −1.10 −1.26 −0.95 −0.85 σ 1.98+0.55 0.31+0.07 0.39+0.220.50+0.92 2.35+2.97 Intercalibrationfactors I⋆FF⋆ee 42.50−−+013.223..98302Int6e.r1c−+−a110l...i430b6ra5t.i2o9n−+−f230a...531c112to3r.1s4−+−230...79591017.14−−+115.2.38.0331 CCij 11..1138−+−+0000....00115600 11..3349−+−+0000....00003355 11..8401−+−+0000....11113200 10..2410−+−+0000....00118812 10..4760−+−+0000....00003466 Cj 1.11−+00..0099 1.33−+00..0055 1.31−+00..00990.37−+00..1100 0.66−+00..0055 Cs 1.72−+00..3311 1.65−+00..1188 – – 1.98−+00..1177 C 0.87+0.05 1.16+0.03 1.38+0.110.97+0.07 1.19+0.37 Statistics Ci 1.32+−00..2045 1.43+−00..1053 –−0.10 –−0.07 1.62+−00..1345 χ2 1445 2659 1543 1799 2161 s −0.24 −0.15 −0.13 Statistics (dof) (1310) (2024) (1515) (1630) (1780) χ2 1305 2174 1360 1594 1832 Statisticswithoutironlinetakenintoaccount (df) (1308) (2022) (1519) (1634) (1778) χ2 1453 2710 1551 1807 2169 nl Statisticswithoutironlinetakenintoaccount (dof) (1311) (2026) (1516) (1631) (1781) χ2 1316 2204 1364 1596 1836 nl (df) (1310) (2025) (1521) (1636) (1780) ⋆Inunitsof10−11ergcm−2s−1. ⋆⋆Inunitsof10−5photonskeV−1cm−2s−1. ⋆in10−11ergcm−2s−1units. ⋆⋆⋆in10−5photonscm−2s−1units. Table 7. Parameters of the model consisting of emission from thecentralsourcereflectedfromtherelativistically-blurredpho- toionizeddiscandajet Figure4)donotallowustofirmlydistinguishionisedandnon- Period1 Period2 Period3 Period4 Period5 ionisediron.Inallotherspectratheadditionofanironlinedoes Directandreflectedemissionfromthecentralsource notsignificantlyimprovethefit.Forcompletenessweshowthe ξ+ 2977+1565 10000+0 3170+336010000+0 2570+2391 degradationofthefitifnolineisincluded.Thecompletelistof F∗ 5.78+−01.17216 4.31+0−.23418 4.59+−01.46303 4.05+−02.64353 3.88+−04.1106 the fit parameters for our phenomenologicalmodel is given in 0Γ.2−1 2.83+−00..0772 2.56+−00..0316 2.56+−00..0589 2.48+−00..0379 2.76+−00..0345 Table5. R(%1) 2.39+−30..9088 20.02+−107.092.51 2.61+−20..0018 11.38−+01.00.969 2.69+−30..4011 −1.00 −2.39 −1.56 −7.11 −1.15 PowerlawdescribingJetemission Γ 1.70+0.03 1.75+0.01 1.53+0.04 1.55+0.05 1.54+0.03 3.2.2. Modellingofthesoftexcess F∗2 8.05+−00..2072 9.23+−00..2052 6.88+−00..6003 5.76+−20..6036 8.02+−20..6082 2−10 −0.57 −0.36 −0.75 −2.80 −0.38 Gaussianline The “soft excess” below 1 keV is typical of Seyfert galaxies. E 6.65 6.73+0.05 6.65 6.65 6.65 Severalcompetingmodelsforthephysicaloriginofthiscompo- Fe −0.06 n∗∗ 3.81+1.57 3.06+0.62 2.28+1.37 0.20+0.93 1.42+0.94 nent are available. In particular, the most simple “phenomeno- Fe −1.56 −0.63 −1.35 −0.20 −0.86 Intercalibrationfactors logical”modelsofthesoftexcessincludea(multi-temperature) C 1.10+0.04 1.19+0.02 1.28+0.05 0.35+0.05 0.66+0.04 blackbodycomponent.However,the originof suchblackbody Cj 0.77+−00..0095 1.17+−00..0042 1.12+−00..0594 0.76+−00..1005 1.08+−00..0052 i −0.09 −0.05 −0.09 −0.10 −0.05 componentisnotclear,becausetheinferredtemperatureisusu- C 1.16+0.21 1.30+0.14 – – 1.35+0.12 s −0.21 −0.14 −0.11 allytoohightobeaccountedforstandardaccretiondisks.More Statistics complicated models include e.g. reflection of external X-ray χ2 1399 2239 1349 1516 1819 emission from the disk. Attempts to fit the soft excess in our (dof) (1410) (2002) (1507) (1542) (1758) XMM-Newtondatawithdifferentmodelsgivethefollowingre- Statisticswithoutironlinetakenintoaccount sults. χ2nl 1405 2262 1353 1516 1822 (dof) (1412) (2004) (1509) (1543) (1759) Ablackbodymodelforthesoftexcessdoesnotprovidean acceptablefittothedata.Best fitparametersforthismodelare +inunitsofergcms−1. listed in Table 6.The disk reflection component calculated us- ⋆inunitsof10−11ergcm−2s−1. ingthekdblurandreflionmodelsofXSPEC(Ross&Fabian ⋆⋆inunitsof10−4photonskeV−1cm−2s−1. 2005; Crummyetal. 2006) providesa better descriptionof the softexcess,thantheblackbodymodel.Amongtheparametersof the reflionmodelwe fit onlythe spectralslope of the power law emission of the central source, fluxes of the direct and re- alsoreportthemeasuredratioRofthereflectedfluxtothetotal flected components, and the ionization parameter of the gas ξ fluxexpressedinpercents. (ξ = 4πF/n , where F is the illuminatingenergy flux, and n One should note, however, that the amount of reflection H H is the hydrogennumberdensity in the illuminatedlayer;in the found in each data set, except for data set number 2, is at the reflion model ξ is restricted to 1 < ξ < 10000 erg cm s−1). level of several percent. Thus, a better fit (compared to black- ParametersofthemodelfitaregiveninTable7.InthisTablewe body)to the data is explainednotbythe carefulaccountof re- 6 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 flectionfeatures,butbyalargecontributionofthedirectpower Table 8. Black body, power law, Seyfert-like component and law illuminating the disk, so that the disk reflection model is GaussianlinefittoPN,JEM-X,SPIandISGRI3C273spectra. not significantly different from the previous phenomenological ThefitwasdoneleavingfreeintercalibrationfactorsforISGRI model. To summarize, we find that neither the black body nor (C), SPI (C ) and JEM-X (C ) with respect to the PN camera. i s j thediskreflectionprovidebetterdescriptionsofthesoftexcess Theparameteruncertaintiescorrespondto1σlevel. thanthe(cut-off)powerlaw. Period1 Period2 Period3 Period4 Period5 Compton reflectionis characterizedby a ”reflectionhump” Blackbody atabout30keV.Tolookforthepresenceofareflectioncompo- F∗ 1.01+0.05 <0.004 0.53+0.03 0.33+0.02 0.79+0.03 1 −0.06 −0.03 −0.03 −0.02 nentin the spectrumwe have also tried to fit the spectra above Powerlaw 2 keV with either a simple powerlaw or a directplusreflected F∗ 5.25+0.09 6.32+0.03 5.86+0.07 4.76+0.04 6.74+0.05 2 −0.09 −0.03 −0.06 −0.05 −0.04 power law (the reflion model of XSPEC) and compare the Seyfertcomponent qualityofthefits. Althoughwe findthatthe additionofthere- F∗ 3.47+0.04 3.86+0.01 2.10+0.03 1.96+0.02 1.71+0.02 s −0.04 −0.01 −0.03 −0.02 −0.02 flectedcomponentatthelevelofabout20%improvesthequality Gaussianline ofthefits,theF-testshowsthattheprobabilitythatthisimprove- n∗F∗e 6.45−+31..5882 3.03−+10..0969 4.45−+12..5714 1.78−+11..7185 2.92−+11..0786 Intercalibrationfactors mentiscoincidentalisatthelevelof10%,sothatnodefinitive C 1.09+0.09 1.26+0.05 1.39+0.10 0.28+0.11 0.65+0.05 conclusionaboutthepresenceofareflectionhumpinthespec- j −0.09 −0.05 −0.10 −0.11 −0.05 C 0.68+0.03 0.80+0.02 1.16+0.09 0.74+0.05 1.05+0.02 trumcanbemade.Oneshouldnotethatthelargestuncertaintyis i −0.03 −0.02 −0.09 −0.05 −0.02 C 1.01+0.18 0.97+0.11 – – 1.42+0.12 duetothefactthattheISGRIinstrumentissensitiveonlyabove s −0.19 −0.11 −0.12 Statistics ∼ 20 keV, while XMM-Newton data are available only below χ2 1340 2342 1383 1561 1913 ∼10keVand,infact,thereisadegeneracybetweenthenormal- (dof) (1311) (2026) (1531) (1565) (1781) ization of the reflection hump and the ISGRI – XMM-Newton Statisticswithoutironlinetakenintoaccount intercalibrationfactor. χ2 1345 2350 1389 1562 1918 nl (dof) (1312) (2027) (1532) (1566) (1782) A combination of black-body and reflection components, suggested as a best-fit model for the Beppo-SAX data by ⋆inunitsof10−11ergcm−2s−1. Grandi&Palumbo(2004)doesnotchangetheaboveconclusion ⋆⋆inunitsof10−5photonskeV−1cm−2s−1. (seebelow),leavingthe(cut-off)powerlawtobethebestmodel forthesoftexcess. 3.2.4. Multibandfluxandspectralslopecorrelations 3.2.3. DisentanglingtheJetandSeyfertcontributions Comptonization of big blue bump photons was proposed as a plausible mechanismof productionof the hardX-ray emission Grandi&Palumbo(2004)(GP)haveputforwardastrongclaim from 3C 273 by Walter&Courvoisier (1992). The motivation thatSeyfertand“jet” componentsof the X-rayspectrumcould forthismodelwasalinearcorrelationbetweentheX-rayspec- bedisentangledinthecaseof3C273sothattheobservedvari- tral slope above2 keV and the logarithmof the ratio of the 2– ations of the X-ray spectral shape are successfully described 10keVfluxtotheultravioletandsoftexcesscountratesfoundin by variations in the normalizations of the template “jet” and EXOSAT,GINGAandIUEdata.ThepresenceoftheUVcamera “Seyfert” components. We have tested this hypothesis with on board of XMM-Newton and of the Optical Monitor Camera our data. For this we repeat the analysis of Grandi&Palumbo (OMC) on board of INTEGRAL enables us to test this correla- (2004).First, wefindthe“template”modelsforSeyfertandjet tionbetweentheUVandX-rayfluxesandslopes.Fig.6shows componentsbyfittingsimultaneouslyallthe5datasetswiththe the values of the power law index in 2–10keV band, Γ , plot- 2 samemodelwhichconsistsofahardpowerlaw(jet)andblack- ted against the logarithm of the ratio of X-ray and UV fluxes bodypluspartiallyreflectedcut-offpowerlaw(Seyfertcompo- (F /νF ).4 2−10keV 10eV nent).The best fit is achievedwith the followingparameters:a Figure 6 clearly shows that the correlations seen in blackbodytemperatureT =75eV,aSeyfertcomponentpower EXOSAT/GINGAX-raydataandIUEUV dataarenotpresent bb lawslopeΓ =2.25,areflectioncomponentfractionRC =0.77, in our data set. For completeness we have extended the anal- s ajetpowerlawslopeΓ =1.5,aninclinationofthediskφ=18◦. ysis for the whole set of publicly available XMM-Newton data j We assumed solar abundances of elements and fixed the posi- on3C273toproducethisFigure(fordetailsseeSoldietal.,in tionandwidthoftheironlineatthevaluesfoundinthesecond preparation). dataset: E = 6.65,σ = 0.3keV.Thenwe frozethe above InFig. 6 two importantfactsare apparent.First, the scatter Fe Fe parameters and left only the normalizations of the black body, oftheXMM-Newtondatapointsalongthex-axisismuchsmaller reflection and jet power law componentsvariable. With such a than that of the EXOSAT/GINGAand IUE points. Thismeans “restricted model” we fit each data set separately and find the thatintherecentpasttheX-rayandUVfluxeshavevariedina normalizationsof the three “template” components.The fluxes correlatedway.LefttoppanelofFig.8showsthatthisisindeed ineachcomponentfoundinthiswayarelistedinTable8. thecase(SpearmancorrelationcoefficientisequaltoR = 0.93, withaprobabilityofrandomcoincidenceP =3×10−3).This Afterthisanalysis,andfollowingGrandi&Palumbo(2004), rand is a quite surprising result, because previoussearches for such we study the correlations between e.g. Seyfert componentand correlationsgave negativeresults. Howeveranothercorrelation iron line strength or between fluxes in Seyfert and black body typicalforSeyfertgalaxiesbetweenthehardX-rayspectralin- component, which were found in the Beppo-SAX data. Fig. 5 dexΓ andtheX-rayflux(Zdziarskietal.(2003),andreferences (whichisananalogofFig.1ofGP)showsthatnosuchcorrela- 2 tionsarepresentinourdataset. Wediscusstheimplicationsof 4 We have extrapolated the UV spectrum obtained with the XMM- thisresultbelow. Newtonopticalmonitorupto10eVusingthetime-averagedUVspec- trumof3C273. M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 7 7 1 8 6 6 0.5 4 5 2 0 4 0 1 2 3 4 1 2 3 4 1 2 3 4 Fig.5.Thedependanceofthe0.2-2keVblack-bodyemission(left),the2-10keVjetlikeflux(middle),andtheironlineflux(right) onthe2-10keVSeyfert-likeflux.DatapointsaretakenfromTable8 GINGAandIUEdataand“appearance”ofX-ray–UVfluxcor- relationsnotdetectedbefore. Having found a previously undetected correlation between UVandX-rayflux,wehavesearchedforthepresenceofaddi- tionalcorrelations,suchasbetweenthe0.2-1keVand3-10keV fluxes(toprightpanelofFig.8),0.2-1keVand20-60keVfluxes (bottom left panel of Fig. 8) and 20-60 keV and 60-100 keV fluxes(bottomrightpanelofFig.8).Onecanseeaclearcorrela- tionbetween20-60keVand60-100keVfluxes,indicatingthat bothbandsaredominatedbyoneandthesamephysicalcompo- nent(e.g.powerlawemissionfromthejet).Ontheotherside,no or only marginal correlation in the two other cases shows that severalphysicalmechanismscontributetotheemissioninthese energybands. 4. Long-termevolutionoftheX-rayspectrumof 3C273 TotesttheevolutionoftheX-rayspectrumoverthelast30year periodwehavecollectedthe archivalX-raydatafromdifferent missions. The evolution of the hard (2-10 keV) spectral index found from the archival data5 (Tu¨rleretal. 1999) is shown in Fig. 9. One clearly sees both significant short-term variations of the spectral index with a scatter of ∆Γ ≃ 0.1 and a long- Fig.6. 2 – 10 keV spectral photon index of the phemenologi- termtrendofsofteningofthespectrumbyasmuchas∆Γ≃ 0.4 cal model as a functionof the logarithmof the ratio of the 2– overthelast30yr.Itisinterestingtonotethatduringour2003- 10keV to ultraviolet fluxes. Old Ginga and EXOSAT data are 2005 monitoring campaign the source has reached its histor- shownwithredtriangles.Thelineindicatesthebestcorrelation ically softest hard X-ray spectrum characterized by a photon foundinWalter&Courvoisier(1992).NewXMM-Newtondata index Γ = 1.82 ± 0.01. It is possible that still softer spectral areshownwithblackcircles.WehaveextrapolatedtheUVspec- states were found at the very beginning of X-ray observations trumobtainedwiththeXMMopticalmonitorupto10eVusing (UHURU, ARIEL5/CandOSO8/A data points),butthe uncer- thetime-averagedUVspectrumof3C273. tanityofthosemeasurementsaretoolargetobeconclusive. 5. Discussion therein),isnotobserved(seeFig.7).Asimilarresultwasfound byPageetal.(2004). Mostofthetheoreticalmodelsof3C273suggestthatthesource Second,thevaluesofthehardX-rayspectralindexinXMM- spectrumcontainstwomajorcontributions,onefromapc-scale Newtonobservationsaresystematicallyhigherthanthoseofthe jetandtheotherfromaSeyfert-likenucleus.Theverylowmil- EXOSAT/GINGAdata. This indicatesthatthe spectralstate of limeter flux measured during our observations suggested that the sourcehasevolvedoverthe 30yearsofX-rayobservations the jet contribution has reached its historical minimum (see (inthenextsectionwediscussthisindetails).Thismaybelinked to the “disappearance” of the correlations seen in EXOSAT, 5 Availablefrom3C273’sDatabase:http://isdc.unige.ch/3c273/ 8 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 Fig.8.Flux-fluxplotsfordifferentenergybands.OntherightbottompanelonedayaveragedINTEGRALdataareshown.Toreduce thescatterthedatawererebinnedbybuildingtheaverageofthreeadjacentoriginaldatapointsinxdirection. Tu¨rleretal.(2006)).However,thisdidnotresultintheclearap- lation is seen at such low iron line flux values. However, it is pearanceoftheSeyfert-likecontribution.Thismeanseitherthat importanttonotethatnodirectcomparisonbetweenourFig.5 theSeyfert-likecontributionhasdecreasedsimultaneouslywith andFig.1ofGrandi&Palumbo(2004)canbedone,becauseof thejetcontribution(which,becauseofthedifferenceinthesize thedifferencesinthetemplatemodelsderivedfromBeppo-SAX scalesandproductionsitescanhappenonlybyachancecoinci- andINTEGRAL-XMM-Newtondatasets. dence),orthattheobservedemissionjustcannotbedecomposed In fact, the analysis of the secular evolution of the source intoSeyfertandjetcontributions. presented aboveshows that both Seyfertand jet spectra evolve withtime.ThiscanbeillustratedbytheresultspresentedinFig. Our data show thatthe separationof Seyfertand jet contri- 6, wherewe plotthehardX-rayspectralslopeas a functionof butionstothespectrumcouldnotbedonebyfixinga”template” theratiooftheX-raypowerlawfluxtotheUVflux.Thisratio for each of them. Namely, although the typical ranges of vari- isapartfromaconstantfactortheratiooftheComptonizedflux ations of soft and hard X-ray fluxes in our data set are simi- to the sourcephotonfluxin a modelin whichthe X-raypower lar to the ones of Beppo-SAX data set, the ”template” models lawisduetotheComptonizationofthe softphotonsbyather- for the Seyfert, jet and black body components found in our malelectronpopulation.Thepointsgivenbytheredcrosseswith data set are significantly different. In particular, the additional largeerrorbarswereobtainedbytheEXOSATandIUEmissions black bodycomponentis notrequiredby the data and thus the in the 80-s and were analysed by Walter&Courvoisier (1992) resultingblackbodyfluxshownintheleftpanelofFig.5issys- who deduced that in the frame of the thermal Comptonization tematicallylowerthanthevaluesreportedinGrandi&Palumbo model, the electron temperature was about 1MeV or some- (2004). Our data show that the correlation between the black what less. They also deduced that the optical depth of the bodyandSeyfertcomponentsisabsentatsuchlow blackbody Comptonizing medium was 0.1–0.2 and that it covered a cou- fluxvalues.Thesameistrueforthenon-observationofthecor- pleofpercentofthesoftphotonsource. relationbetweentheSeyfertcomponentfluxandthestrengthof theironline:sincetheadditionoftheironlineisnotstatistically The XMM-Newton data shown in Figure 6 by the black required in any data sets except for the second one, no corre- pointswithsmallerrorbarsclearlydonotmatchthecorrelation M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 9 4 3 2 1 2 UHURU EXOSAT ARIEL5/C GINGA XMM-Newton/PN OSO8/A ASCA This Work HEAO-1 RXTE 1.8 HEAO2/MPC BeppoSAX/MECS BALL/AIT XMM-Newton/PN 1.6 1.4 1.2 1 1970 1980 1990 2000 Time Fig.9.Evolutionofthefluxdensityat1keV,asextrapolatedfromthefitdoneabove2keV,in10−2photons/keV/cm2/s(toppanel) andhard X-rayspectralindex(bottompanel)overthe 30 yearsofobservations.UHURU, Ariel 5/C, OSO 8/A,HEAO-1,HEAO- 2/MPC, and BALL/AIT data points are taken from Malagutietal. (1994); EXOSAT and Ginga data are taken from Turneretal. (1990);ASCAdataarefromCappietal.(1998);RXTE,Beppo-SAX,andpartofXMM-NewtondataarefromSoldietal.(inprepa- ration). foundinearlieryears.Thisisparticularlytrueforthespectraob- Comptonization region when the source is very weak. The re- tainedduringperiods1and2,whenthesoftX-rayemissionwas centdataalsoshowthatthehardX-rayfluxiscorrelatedwiththe muchbrighterthanattheotherepochs,whiletheultravioletflux softflux(seeFigure8),aneffectnotobservedatearlierepochs was similar. This indicates a much harder blue bump for these (Courvoisieretal.1990). epochs.MorephotonswerethereforeavailabletoComptoncool the hot electrons than inferred from the ultraviolet flux shown The correlation between the optical/UV and the X-ray flux in Figure 5. This therefore leads to the softening in the X-ray ischaracteristicforSeyfertgalaxies.Thisrelationshipisparticu- emission.Alternatively,onecanalsoinferamodificationofthe larlyclearinthelong-termbehaviourofNGC5548(Uttleyetal. 2003), but was also seen in other Seyfert galaxies like for 10 M.Chernyakovaetal.:2003–2005INTEGRALandXMM-Newtonobservationsof3C273 emission from an electron distribution cooled via synchrotron andinverseComptonenergylosswhichcanproduceonlyspec- trawithΓ≥1.5. Thisproblemcan,however,berelaxedifoneassumesthatin additiontoelectronstheemission regioncontainsalso protons. IfthemainsourceoftheseedphotonsfortheinverseCompton scatteringisUVphotonsfromthebigbluebump,thehardX-ray inverseComptonemissionoriginatesfromapopulationofelec- tronswith moderateenergies E ∼ 1−10 MeV. In this energy e rangecoolingcouldbedominatedbyionization,ratherthanin- verse Compton/synchrotronlosses. The Coulomb loss leads to the hardeningof the electron spectrum, compared to the injec- tion spectrum by ∆Γ ≃ −1. This will result in a hardeningof e the inverse Compton spectrum below the so-called ”Coulomb break” (see e.g. Aharonian (2004)). A simple estimate shows that if the proton density in the gamma-ray emission region is n ∼ 1010cm−3, the Coulomb break is situated in the hard X- p ray band which can provide the explanationof the harder than Γ= 1.5spectrumofthesourceinlate70-s.Highmatterdensity in the emission region obviously favors the ”compact source” modelscomparedtothejetmodels:theassumptionaboutalarge (1010cm−3)matterdensitiesatthepcdistancesfromthenucleus neededtoexplainthehardspectruminthejetmodelframework looks somewhat extreme (for comparison, the density of dust andmoleculargasintheCentralMolecularZone(CMZ)ofthe Fig.7.ThedependenceofthehardX-rayphotonindex(derived Galaxyis estimatedto be 104 cm−3, see e.g.Morris&Serabyn from fitting the data in 2-100 keV energy range with a single (1996)).Theconclusionthattheobservedevolutionofthehard powerlaw,seetable4)onthe2–10keVX-rayflux. X-ray spectral index favors the compact source models should also be taken with caution.The reasonfor this is thatthe pres- instance NGC 4051 (Shemmeretal. (2003), and references ence of a Seyfert-like contribution to the hard X-ray spectrum therein). The data of 3C 273 presented here do also show the canaffectthevalueofthespectralindex. appearance of such a trend (see Fig. 8), apparently suggesting Ifthegamma-rayemissionof3C273originatesinacompact aSeyfert-likebehaviorof3C273in2003–2005.However,itis sourcethesynchrotronradiationfromelectronswhichupscatter clearfromtheabovediscussionthatinthecaseof3C273con- thebluebumpphotonstotheGeVenergybandcouldcontribute clusions of this sort should be taken with caution: there is no alsotothesoftX-rayspectrumofthesource.Themagneticfield clear way to disentangle the Seyfert and jet contributions. For strength in the γ-ray emission region is poorly constrained by example, another characteristic of Seyfert galaxies, a correla- theobservations(see,however,Courvoisieretal.(1988)forre- tionbetweenthehardX-rayspectralindexΓ andtheX-rayflux strictionsonthemagneticfieldstrengthduringflaringactivityof 2 (Zdziarskietal.(2003),andreferencestherein),isnotobserved thesourceinthelate80-s).Assumingthattheenergydensityof (seeFig.7). magneticfieldinthecompactemissionregioniscomparableto The secularevolutionofthe hardX-rayspectralindexover theenergydensityofradiation, the last 30 years,clearly seen in Fig. 9, showsthatthe jet con- tributionalsoevolveswithtime.WhetherthehardX-rayfluxis U ≃2×102 L 1016cm 2erg/cm3 (1) generated in the jet, or it is due to the nuclear related physics ph "1046erg/s#" R # (i.e. from the ”compact source” (Lightman&Zdziarski 1987)) isstillcompletelyundecided.”Jet”and”compactsource”mod- (weestimatethesizeofemissionregiontobeR∼1016cmfrom elsdifferintheirapproachtotheso-calledcompactnessproblem. the observedintra-dayvariabilityof the source)oneobtainsan Theessence ofthecompactnessproblemisthatthenaivelyes- estimate timatedopticaldepthofthesourceforγ-rayswithenergiesE , γ L 1/2 1016cm such that pairs are produced in the interactions with soft pho- B≃102 G (2) tonsofenergiesǫ ≥ 2m2c4/E ,ishigherthan1.The”compact "1046erg/s# " R # s e γ source”modelsself-consistentlyaccountfortheeffectofthepair Thesynchrotronradiationfrom1-10GeVelectronsinsuchmag- productioncascadeinthesource(Lightman&Zdziarski1987). neticfieldisemittedattheenergies The”jet”modelsgetridofthecompactnessproblembyassum- ingrelativisticbeamingoftheγ-rayemission. B E 2 Γ ≃T1h.e4otboseΓrv≃ed1e.8vooluvteironthoefltahseth3a0rdyXea-rrsaycosupledcthraellpintdoexdifsrtoinm- ǫs =3(cid:20)102G(cid:21)(cid:20)10GeeV(cid:21) keV (3) guishbetweenthetwopossiblemodelsofthehigh-energyemis- Thenon-observationofacut-offintheGeVband(Collmaretal. sion. Thepointis thatalthoughthe present-dayvaluesΓ > 1.5 2000) impliesthat the high-energycut-offin the electronspec- areeasilyexplainedinbothmodels,theveryhardspectrumob- trumis atleastat energiesE ≥ 1GeV which,in turn,implies e servedisthelate70-sisnot.Inbothmodelsthegamma-raysare thatthecut-offinthesynchrotronspectrumshouldbeatleastin producedviainverseComptonemissionfromhigh-energyelec- the soft X-rayband.Taking into accountthatthe bestfit to the trons.ThehardspectralindexΓ≃1.4observedin70-sand80-s ”softX-rayexcess”inourdataisgivenbyasimplecut-offpower (see Fig. 9) is too hard for the optically thin inverse Compton lawmodel,wefindthatitispossiblethattheobservedsoftX-ray

See more

The list of books you might like

Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.
DMCA & Copyright: Dear all, most of the website is community built, users are uploading hundred of books everyday, which makes really hard for us to identify copyrighted material, please contact us if you want any material removed.
Copyright @ PDFdrive.to, 2022 | We love our users