Astronomy&Astrophysicsmanuscriptno.NGC5548UV (cid:13)c ESO2014 November11,2014 Anatomy of the AGN in NGC 5548: II. The Spatial, Temporal and Physical Nature of the Outflow from HST/COS Observations N.Arav1,C.Chamberlain1,G.A.Kriss2,3,J.S.Kaastra4,5,M.Cappi6,M.Mehdipour4,7,P.-O.Petrucci8,9,K.C. Steenbrugge10,11,E.Behar12,S.Bianchi13,R.Boissay14,G.Branduardi-Raymont7,E.Costantini4,J.C.Ely2,J.Ebrero4, L.diGesu4,F.A.Harrison15,S.Kaspi12,J.Malzac16,17,B.DeMarco18,G.Matt13,K.P.Nandra18,S.Paltani14, B.M.Peterson19,20,C.Pinto21,G.Ponti18,F.PozoNun˜ez22,A.DeRosa23,H.Seta24,F.Ursini8,9,C.P.deVries4, D.J.Walton15,andM.Whewell7 4 1 1 DepartmentofPhysics,VirginiaTech,Blacksburg,VA24061,USA. 0 2 SpaceTelescopeScienceInstitute,3700SanMartinDrive,Baltimore,MD21218,USA. 2 3 DepartmentofPhysicsandAstronomy,TheJohnsHopkinsUniversity,Baltimore,MD21218,USA. 4 SRONNetherlandsInstituteforSpaceResearch,Sorbonnelaan2,3584CAUtrecht,theNetherlands. v 5 LeidenObservatory,LeidenUniversity,PostOfficeBox9513,2300RALeiden,Netherlands. o 6 INAF-IASFBologna,ViaGobetti101,I-40129Bologna,Italy. N 7 MullardSpaceScienceLaboratory,UniversityCollegeLondon,HolmburySt.Mary,Dorking,Surrey,RH56NT,UK. 8 Univ.GrenobleAlpes,IPAG,F-38000Grenoble,France. 8 9 CNRS,IPAG,F-38000Grenoble,France. 10 InstitutodeAstronom´ıa,UniversidadCato´licadelNorte,AvenidaAngamos0610,Casilla1280,Antofagasta,Chile. ] A 11 DepartmentofPhysics,UniversityofOxford,KebleRoad,Oxford,OX13RH,UK. 12 DepartmentofPhysics,Technion-IsraelInstituteofTechnology,32000Haifa,Israel. G 13 DipartimentodiMatematicaeFisica,Universita`degliStudiRomaTre,viadellaVascaNavale84,00146Roma,Italy. . 14 DepartmentofAstronomy,UniversityofGeneva,16Ch.d’Ecogia,1290Versoix,Switzerland. h 15 CahillCenterforAstronomyandAstrophysics,CaliforniaInstituteofTechnology,Pasadena,CA91125,USA. p 16 Universite´deToulouse,UPS-OMP,IRAP,Toulouse,France. - o 17 CNRS,IRAP,9Av.colonelRoche,BP44346,31028ToulouseCedex4,France. r 18 Max-Planck-Institutfu¨rextraterrestrischePhysik,Giessenbachstrasse,D-85748Garching,Germany. t 19 DepartmentofAstronomy,TheOhioStateUniversity,140W18thAvenue,Columbus,OH43210,USA. s a 20 CenterforCosmology &AstroParticlePhysics,TheOhioStateUniversity,191WestWoodruff Avenue, Columbus,OH43210, [ USA. 21 InstituteofAstronomy,UniversityofCambridge,MadingleyRd,Cambridge,CB30HA,UK. 1 22 AstronomischesInstitut,Ruhr-Universita¨tBochum,Universita¨tsstraße150,44801,Bochum,Germany. v 23 INAF/IAPS-ViaFossodelCavaliere100,I-00133Roma,Italy. 7 24 ResearchCenterforMeasurementinAdvancedScience,FacultyofScience,RikkyoUniversity3-34-1Nishi-Ikebukuro,Toshima- 5 ku,Tokyo,Japan. 1 2 November11,2014 . 1 ABSTRACT 1 4 Context.AGNoutflowsarethoughttoinfluencetheevolutionoftheirhostgalaxiesandsupermassiveblackholes.Ourdeepmul- 1 tiwavelengthcampaignonNGC5548revealedanunusuallystrongX-rayobscuration. Theresultingdramaticdecreaseinincident : v ionizingfluxontheoutflow,allowedustoconstructacomprehensivephysical,spatialandtemporalpictureforthelong-studiedAGN i windinthisobject. X Aims.Todeterminethedistanceoftheoutflowingcomponentsfromthecentralsource;theirtotalcolumndensityandthemechanism r responsiblefortheobservedabsorptiontroughvariability. a Methods.WestudytheUVspectraacquiredduringthecampaignaswellasfromfourpreviousepochs,wheretheoutflowsarede- tectedasblue-shiftedabsorptiontroughsinthespectraoftheobject.Ourprincipalanalysistoolsareioniccolumndensityextraction techniques,photoinizationmodelsbasedonthecodeCLOUDY,andcollisionalexcitationsimulations. Results.Asimplemodelbasedonafixedtotalcolumn-densityabsorber,reactingtochangesinionizingillumination,matchesthe verydifferentionizationstatesseeninfivespectroscopicepochsspanning16years.Themainoutflowcomponentissituatedat3.5±1 pcfromthecentralsource.Threeothercomponentsaresituatedbetween5-70pcandtwoarefurtherthan100pc.Thewealthofob- servationalconstraintsandthedisparaterelationshipoftheobservedX-rayandUVfluxbetweendifferentepochsmakeourphysical modelaleadingcontenderforinterpretingtroughvariabilitydataofquasaroutflows. Conclusions.Thiscampaign,incombinationwithpriordata,yieldsthefirstsimplemodelthatcanexplainthephysicalcharacteristics andthesubstantialvariabilityobservedinanAGNoutflow. Keywords.galaxies:Seyfert–galaxies:active–X-rays:galaxies 1 1. Introduction HST observationsofNGC5548thatare alsoused inthisanal- ysis. The five observationsfrom the summer of 2013 were op- AGN outflows are detected as blueshifted absorption troughs, timally weighted to produce an average spectrum that we use with respect to the object systemic redshift. Such outflows for our analysis. Kaastraetal. (2014) describe the data reduc- in powerful quasars can expel sufficient gas from their host tionprocessfromthecalibrationofthedatatotheproductionof galaxies to halt star formation, limit their growth and lead to thisaveragespectrum. the co-evolution of the size of the host and the mass of its The2013averageHST/COSspectrumwithallidentifiedab- central super massive black holes (e.g., Ostrikeretal. 2010; sorptionfeaturesisshowninFigureA1.AsdescribedinKaastra Hopkins&Elvis2010;Soker&Meiron2011;Ciottietal.2010; etal.(2014),wemodeledtheemissionfromNGC5548usinga Faucher-Gigue`reetal. 2012; Borguetetal. 2013; Aravetal. reddenedpowerlaw(withextinctionfixedatE(B−V) = 0.02, 2013). Therefore, deciphering the properties of AGN outflows Schlegeletal. 1998), weak Feii emission longward of 1550 Å iscrucialfortestingtheirroleingalaxyevolution. intherestframe,broadandnarrowemissionlinesmodeledwith Nearby bright AGN are excellent laboratories for study- severalGaussiancomponents,blue-shiftedbroadabsorptionon ing these outflows as they yield: a) high-resolution UV data, all permitted transitions in NGC 5548, and a Galactic damped which allow us to study the outflow kinematics and can yield Lyα absorption line. Using this emission model, we normal- diagnostics for their distance from the central source; and b) ized the average spectrum to facilitate our analysis of the nar- high quality X-ray spectra that give the physical conditions rowintrinsicabsorptionlinesinNGC5548.Figure1showsnor- for the bulk of the outflowingmaterial (e.g., Steenbruggeetal. malized spectra for absorption lines produced by Siiii λ1206, 2005;Gabeletal.2005;Aravetal.2007;Costantinietal.2007; Siiv λλ1394,1403, Civ λλ1548,1550, Nv λλ1238,1242, and Kaastraetal.2012).Thus,suchobservationsareavitalstepping Ly α as a functionof rest-framevelocity relative to a systemic stone for quantifying outflows from the luminous (but distant) redshift of z = 0.017175 (deVaucouleursetal. 1991) via the quasars,forwhichhighqualityX-raydataarenotavailable. NASA/IPACExtragalacticDatabase(NED). Forthesereasons,weembarkedonadeepmultiwavelength AsshowninTableA.1,high-resolutionUVspectraofNGC campaign on the prototypical AGN outflow seen in the inten- 5548usingHSTcoveranadditionalfourepochsstretchingback sively studied Seyfert 1 galaxy NGC 5548. For the past 16 to 1998. We use the calibrated data sets for each of these ob- years, this outflow has shown 6 kinematic components in the servations as obtained from the Mikulski Archive for Space UV band (labeled in descending order of velocity, following Telescopes (MAST) at the Space Telescope Science Institute Crenshawetal.2003),andtheirassociatedX-raywarmabsorber (STScI). We compare the strengths of the narrow UV absorp- (WA). Our 2013 campaign revealed a new X-ray obscurer ac- tiontroughsforeachoftheseepochswithournewdatasetfrom companied by broad UV absorption (analyzed in Kaastraetal. 2013inFigs.A2andA3. 2014).Theappearanceoftheobscurerallowsustoderiveacom- prehensive physical picture of the long-term observed outflow, whichwereporthere. 3. Physicalandtemporalcharacteristicsof The plan of the paper is as follows: In § 2 we describe the Component1 observationsanddatareduction;in§ 3weanalyzethekeycom- ponentof the outflow;in § 4 we discuss the remaining5 com- The key for building a coherent picture of the long-seen out- ponents;in § 5 we connectthe results of the UV analysis with flow is component 1: the strongest and highest velocity out- thoseoftheX-raywarmabsorberofthesameoutflow;andin§6 flow component(centeredat−1160km s−1). Dueto the strong wecompareourresultswithpreviousstudies,discusstheimpli- suppressionof incidentionizing flux by the obscurer,the 2013 cation of our results to the variability of AGN outflow troughs HST/COS data of component 1 show a wealth of absorption in general, and elaborate on the connection between the X-ray troughsfromionsneverbeforeobservedintheNGC5548out- obscurer and the persisting outflow; In § 7 we summarize our flow. These data allow us to decipher the physical characteris- results. ticsofthiscomponent.In§3.1weusethecolumndensitymea- surementsofPv,Piii,Feiii,andSiiiasinputinphotoionization models,andderivethetotalhydrogencolumndensityforcom- 2. ObservationsandDataReduction ponent1oflog(N ) = 21.5+0.4 cm−2,andanionizationparam- H −0.2 Our 2013 multiwavelength campaign on NGC 5548 included eteroflog(UH) = −1.5+−00..42.In§3.2weusethecolumndensity coordinated observations using XMM-Newton, HST, Swift, measurements of Ciii* and Siiii* to infer the electron number INTEGRAL, and NuSTAR. Kaastraetal. (2014) describe the densitylog(ne)=4.8±0.1cm−3,whichcombinedwiththevalue overall structure of the campaign. A full log of all the obser- oftheincidentUH yieldsadistanceR=3.5±1parsecbetween vations is given by Mehdipouretal (2014). Here we present a component1andthecentralsource.In§3.3weconstructasim- moredetailedanalysisoftheUVobservationsweobtainedusing plemodelbasedonafixedtotalcolumn-densityabsorber,react- theCosmicOriginsSpectrograph(COS)(Greenetal.2012)on- ing to changes in ionizing illumination, that matches the very boardHST.We obtainedfiveCOSobservationssimultaneously differentionizationstatesseeninfiveHSThigh-resolutionspec- with five of the XMM-Newton observationsbetween 2013-06- troscopicepochsspanning16years. 22 and 2013-08-01. Each two-orbit observation used gratings G130MandG160Matmultiplecentralwavelengthsettingsand 3.1.TotalColumnDensity(N )andIonizationParameter H multiplefocal-planepositions(FP-POS)tocoverthewavelength (U ) H rangefrom1132Åto1801Åataresolvingpowerof∼15,000. TableA.1liststheobservationdatesoftheindividualvisits,the In the Appendix we describe the methods we use to derive exposuretimes, andthecontinuumfluxmeasuredat1350Å in the ionic column densities (Nion) from the outflow absorption therestframe,aswellascorrespondinginformationforarchival troughs. Table 2 gives the Nion measurements for all observed troughsofall6outflowcomponentsinthe5HSTepochs(span- Sendoffprintrequeststo:[email protected] ning 16 years) of high-resolution UV spectroscopy. The upper N.Aravetal.:NGC5548 Fig.2.TheadoptedNGC5548SEDs.InblackweshowtheSED for NGC 5548 in 2002 at an unobscured X-ray flux (hereafter “highSED,”fromSteenbruggeetal.2005).Inblueweshowthe SED appropriateto the 2013epoch,whereforthesame fluxat 1000 Å, the new X-ray obscurer reduced the ionizing flux by a factor of 17 between 1 Ry and 1 keV (obscured SED). The ionization potentials for destruction of some of the prominent observedspeciesareshownbytheverticallines. the “obscured”SED in Figure 2, and furtherjustify its specific shape in § 3.3 For abundances, we use pure proto-Solar abun- dancesgivenbyLoddersetal.(2009). WiththechoiceofSEDandchemicalabundances,twomain Fig.1.Intrinsicabsorptionfeaturesinthe2013COSspectrumof parametersgovernthephotoionizationstructureoftheabsorber: NGC5548.Normalizedrelativefluxesareplottedasafunction the total hydrogencolumn density (N ) and the ionization pa- ofvelocityrelativetothesystemicredshiftofz=0.017175,top H to bottom: Siiii λ1206, Siiv λλ1394,1403,Civ λλ1548,1550, rameter Nv λλ1238,1242,andLy α, as a functionofrest-frameveloc- Q U ≡ H (1) ity.Forthedoubletsthe redandbluecomponentsareshownin H 4πR2n c redandblue,respectively.Dottedverticallinesindicatetheve- H locitiesoftheabsorptioncomponentsnumberedasinCrenshaw where QH is the rate of hydrogen-ionizingphotons emitted by etal.(2003). the object, c is the speed of light, R is the distance from the centralsourcetotheabsorberandn isthetotalhydrogennum- H berdensity.Wemodelthephotoionizationstructureandpredict the resulting ionic column densities by self-consistently solv- andlowerlimitswerederivedusingtheApparentOpticalDepth ing the ionization and thermal balance equations with version (AOD)method.Allthereportedmeasurementsweredoneusing 13.01 of the spectral synthesis code Cloudy, last described in thePartialCovering(PC)method.WeusedthePower-Law(PL) Ferlandetal.(2013).Weassumeaplane-parallelgeometryfora methodonlyontheCiii*inordertoquantifythesystematicer- gasofconstantn . rorduetodifferentmeasuringmethods(see§3.2andFig.4). TofindthepaHirof(U ,N )thatbestpredictsthesetofob- H H TheNionwemeasurearearesultoftheionizationstructureof served column densities, we vary UH and NH in 0.1 dex steps theoutflowingmaterial,andcanbecomparedtophotoionization to generate a grid of models (followingthe same approachde- models to determine the physicalcharacteristics of the absorb- scribedinBorguetetal.2012b)andperformaminimizationof ing gas. As boundary conditions for the photoionization mod- thefunction els, we need to specify the choice of incident Spectral Energy Distribution(SED),andthechemicalabundancesoftheoutflow- χ2 = log(Ni,mod)−log(Ni,obs) 2 (2) inggas. log(N )−log N ±σ ! Wemakethesimple(andprobablyover-restrictive)assump- Xi i,obs i,obs i (cid:0) (cid:1) tion that the shape of the SED emitted from the accretion disk where,forioni, N andN aretheobservedandmodeled i,obs i,mod did not change over the 16 years of high-resolution UV spec- columndensities,respectively,andσ istheerrorintheobserved i troscopy.Specifically,we assume that the emittedSED hasthe columndensity.Themeasurementerrorsarenotsymmetric.We sameshapeithadwhenweobtainedsimultaneousX-ray/UVob- use the positive error (+σ) when log(N ) > log(N ) and i i,mod i,obs servationsin2002(the “high”SED in Figure2, determinedby thenegativeerror(−σ)whenlog(N )<log(N ). i i,mod i,obs Steenbruggeetal. 2005). In 2013 the obscurer absorbed much Theionizationsolutionforcomponent1atthe2013epochis ofthesoftionizingphotonfluxfromthisSEDbeforeitreached showninfigure3.WeonlyshowconstraintsfromN measure- ion component 1. We model the incident SED on component 1 as ments,andnotethatallthelowerlimitsreportedinTable2are 3 N.Aravetal.:NGC5548 FeIII 22 N ) H PV g ( o L PIII 21 SiII -2 -1 0 Log (U ) H Fig.3.A photoionizationphaseplotshowingtheionizationso- lution for component1 epoch2013.We use the obscuredSED andassumedproto-solarmetalicity(Loddersetal. 2009).Solid linesandassociatedcoloredbandsrepresentthelocusofU ,N H H models, which predict the measured N , and their 1σ uncer- ion tainties.Theblackdotisthebestsolutionandissurroundedby a1σ χ2contour. satisfiedbythissolution.Wefindlog(N ) = 21.5+0.4cm−2,and H −0.2 anionizationparameteroflog(U )=−1.5+0.4,wheretheerrors H −0.2 arestronglycorrelatedasillustratedbythe1σχ2contour. 3.2.NumberDensityandDistance As shown in Figure 4, we detect absorption troughs from the Ciii* 1175 Å multiplet, arising from the metastable 3P levels j ofthe2s2pterm.AsdetailedinGabeletal.(2005),theexcited Ciii* 1175Å multipletcomprisessix linesarisingfromthreeJ levels. The J=0 and J=2 levels have significantly lower radia- tivetransitionprobabilitiestothegroundstatethantheJ=1level and are thus populated at much lower densities than the latter. In particular, figure 5 in Borguetetal. (2012a) shows that the relative populations of the three levels are a sensitive probe to a wide range of n while being insensitive to temperature.The e twohigh-S/NtroughsfromtheJ=2levelallowustoaccurately accountforthemildsaturationinthesetroughsandthereforeto derivereliableN forbothlevels.Fromthesemeasurementswe ion inferlog(n ) = 4.8±0.1cm−3 (see Fig.4).We alsodetecttwo e shallow troughs from the same metastable level of Siiii* (see panelbofFig.4),whichweusetomeasureanindependentand consistent value for ne (see panel c of Fig. 4). The collisional Fig.4. a)AbsorptionspectrumfortheCiii* 1175Å multiplet. excitationsimulationsshowninFigure4cwereperformedusing The 2013 COS spectrum shows clear and relatively unblended version7.1.3ofCHIANTI(Dereetal.1997;Landietal.2013), individualtroughsfromtheJ=2andJ=0levels,butnocontribu- withatemperatureof10,000K(similartothepredictedtemper- tionfromtheJ=1levelthatispopulatedathigherdensities(see atureofourCLOUDYmodelforcomponent1duringthe2013 panelc).b)AbsorptionspectrumfortheSiiii*1298Åmultiplet epoch). (similar in level structure to the Ciii* 1175 Å multiplet). The In addition to the Ciii* and Siiii* troughs, the 2013 COS 2013COS spectrum shows shallow but highlysignificant indi- spectraofcomponent1showseveraladditionaltroughsfromex- vidualtroughsfromtheJ=2andJ=0levels,butagainnocontri- citedstates:Cii*, Siii*,Piii*, Siii* andFeiii* (aswellastheir bution from the J=1 level that is populated at higher densities. associatedresonancetransitions).Carefulmeasurementofthose c) Ciii* andSiiii* levelpopulationratios, theoryandmeasure- troughsshowthatinthesecases,thededucedn isalowerlimit e ments.ThecomputedpopulationsfortheJ=2/J=0andJ=1/J=0 thatislargerthanthecriticaldensityoftheinvolvedexcitedand areplottedasafunctionofelectronnumberdensityforbothions zero energy levels. In all cases the critical densities are below (seetextforelaboration).Thecrossesshowthemeasuredratios fortheJ=2/J=0ratioofbothions.FromtheCiii*ratiosweinfer 4 log(ne) = 4.8±0.1cm−3,wheretheerrorincludesbothstatisti- calandsystematiceffects.Thisvalueisfullyconsistentwiththe oneinferredfromtheSiiii*ratio,whereinthecaseofSiiii*the statisticalerrorislargersincethetroughsaremuchshallower. N.Aravetal.:NGC5548 log(n ) = 4.8 cm−3, thus they are consistent with the n mea- ualintensityofthecomponent1Civtroughcannotbeattributed e e surementwederivefromtheCiii*andSiiii*troughs. to new materialappearingdue to transverse motionat this dis- As can be seen from the definition of the ionization pa- tance.Wenotethat25%motionacrosstheprojectedsizeofthe rameter U (Equation 1), knowledge of the hydrogen number Civ BLR is a highly conservative limit for two reasons: 1) at H density n for a given U and N allows us to derive the dis- certainvelocitiestherearechangesof50%intheresidualinten- tance R. OHur photoionizaHtionmodHels show that for component sityinthecomponent1Civtroughbetweenthe2011and2013 1,log(U )=−1.5andsincen ≃1.2n (asisthecaseforhighly epochs;andintheelapsing2years,transversemotionwillonly ionizedpHlasma),n =5.3×10e4cm−3.HTodeterminetheQ that cover3%oftheprojectedsizeoftheCivBLR;2)materialthat H H affectscomponent1, wefirstcalculatethe bolometricluminos- movesawayfromthecentralsourceundertheinfluenceofradial ity using the average flux at 1350Å for visits 1-5 in 2013, the forces should conserve its angular momentum. Therefore, if it redshift of the object and the obscured SED (see Fig. 2). We movedtodistancesthataremuchlargercomparedwithitsinitial find Lbol = 2.6 × 1044 erg s−1 and from it, QH = 6.9 × 1052 distance,itsv⊥ ≪ vkep atitscurrentdistance.Weconcludethat s−1.Therefore,equation(1)yieldsR = 3.5+1.0pc,wheretheer- evenunderfavorableconditions,thetransversemotionmodelof rorisdeterminedfrompropagatingtheerro−rs1.2ofthecontributing gas into or out of the line of sight cannotexplain the observed quantities. behaviorofcomponent1overthe5observedepochs. Assuming the canonical 50% global covering factor for Can changes of the ionizing flux experienced by the out- Seyfertoutflows(Crenshawetal.1999),andusingequation(1) flowinggasexplaintheobservedtroughchanges?Weconstruct in Aravetal. (2013) we find thatthe massflux associated with suchamodelunderthesimplestandrestrictiveassumptionthat the UV manifestation of component 1 is 1.0+−20..05 solar masses the NH ofcomponent1 didnotchangeoverthe 16yearsspan- per year, and that the kinetic luminosity is 4+8 × 1041 erg s−1. ningthe5highresolutionUVspectralepochs.Furthermore,for WenotethatmostoftheNH inthevariousou−tfl2owcomponents log(ne)=4.8(cm−3),theabsorbershouldreacttochangesinin- isassociatedwiththehigherionizationX-rayphaseofthe out- cidentionizingfluxontime-scalesof 5days(seeeq.3here,and flow.Therefore,wedeferafulldiscussionofthetotalmassflux discussioninAravetal. 2012).Therefore,forcomponent1 we andkineticluminosityoftheoutflowtoafuturepaperthatwill usetherestrictiveassumptionofasimplephotoionizationequi- presentacombinedanalysisoftheUVandX-raydatasets. librium,determinedbythefluxlevelofthespecificobservation. In 1998 the AGN was in a high flux level of F=6 (mea- sured at 1350 Å rest-frame and given in units of 10−14 3.3.ModelingtheTemporalBehavioroftheOutflow ergs s−1 cm−2 Å−1). At that epoch the absorber only showed The absorption troughs of component1 change drastically be- a Lyα trough necessitating log(UH) ∼> 0.1 (otherwise a Nv tweenthe fiveHST high-resolutionspectroscopicepochsspan- troughwouldbedetected,seeFig.5).In2002theAGNwasina ning 16 years (see figures A2 and A3). After finding the lo- mediumfluxlevelwithF=2,atwhichtimetheabsorbershowed cation and physical characteristics of component 1 using the Civ andNv troughsin additionto Lyα. In 2004the AGN was 2013 data, the next step is to derive a self-consistent temporal atahistoricallylowfluxofF=0.25.InthatepochaSiiiitrough picture for this component. There are two general models that appearedinadditiontotheCiv,NvandLyαtroughs;however, explain trough variability in AGN outflows (e.g., Barlowetal. aCiitroughdidnotappear.ThecombinationoftheSiiiiandCii 1992;Gabeletal.2003;Capellupoetal.2012;Aravetal.2012; constraints necessitates −1.3 < log(UH) < −1.15 (see Fig. 5). FilizAketal. 2013, and references therein). One model at- The change in log(UH) required by the photoionization mod- tributesthetroughvariabilitytochangesoftheionizingfluxex- elsagreesremarkablywellwiththechangeinfluxbetweenthe periencedbytheoutflowinggas.Initssimplestform,thismodel 1998and2004epochsaslog(F2004/F1998) =−1.4.Thus,acon- assumesthatthetotalNH alongthelineofsightdoesnotchange stant NH absorberyieldsanexcellentfitforthe absorptionfea- as a function of time. A second model invokes material mov- tures from two epochs with the same spectral energy distribu- ingacrossthelineofsight,whichingeneralcauseschangesof tion(highSEDinFigureFig.5),butwithverydifferentUH val- N along the line of sight as a function of time to explain the ues.Comparisonofseveralkeytroughsbetweenthe5epochsis H observedtroughchanges. showninFigs.A2andA3.The1350Åfluxmeasurements,plus In the case of component 1, we have enough constraints observation details are given in Table A.1 and the derived col- to exclude the model of material moving across our line of umndensitiesforalltheoutflowfeaturesaregiveninTableA.2. sight.The outflowis situatedat3.5 pcfromthe centralsource, In 2013 the AGN flux was F=3. With this flux level and which combined with the estimated mass of the black hole in assuming the same SED, the U value should have been 50% H NGC5548(4×107 solarmasses;Pancoastetal.2013),yieldsa higher than in 2002, and we would expect to see only Civ, Keplerianspeedof1.9×107 cms−1 atthatdistance.Ascanbe NvandLyαtroughs.Insteadwe alsodetectSiiii, Cii, Siii and seen from figure A1, 2/3 of the emission at the wavelength of Alii. Therefore, the incident SED for component 1 must have component1 arises fromthe Civ BroadEmission Line (BEL). changed,andindeedthe2013softX-rayfluxis25timeslower Therefore,thetransversemotionmodelcruciallydependsonthe comparedtothatofthe2002epoch(seeFig.1inKaastraetal. ability of gas clouds to cross most of the projected size of the 2014).Thisdropiscausedbythenewlyobservedobscurerclose BroadLineRegion(BLR)inthetimespanningtheobservations totheAGN,whichdoesnotfullycoverthesource(Kaastraetal. epochs. Reverberation studies (Koristaetal. 1995) give the di- 2014).WefoundagoodmatchtotheUVabsorptionandsoftX- ameteroftheCivBLRas15lightdaysor3.9×1016cm,which rayfluxwithanSEDthatissimilartothatofthehighfluxone for v⊥ = 1.9×107 cm s−1, yieldsa crossing time of 2.0×109 longwardof1Ry,butabruptlydropsto6%ofthatfluxbetween seconds,or65years. 1Ryand1keV(seeFig.2).Thispictureisconsistentwiththe Thus, in the 16 years between our epochs, material that transmitted flux resulting from the low-ionization, partial cov- movesattheKeplerianvelocity,3.5pcawayfromtheNGC5548 eringmodeloftheobscurerderivedinKaastraetal.(2014).To black hole, will cross only about 25% of the projected size of complete the UV picture for component1, in 2011NGC 5548 the Civ BLR. Therefore, the much larger change in the resid- showed F=6, equal to that of the 1998 epoch. However, Civ 5 N.Aravetal.:NGC5548 notbetweenthe2013/7/24and2013/8/1ones,whenwe dosee changes in component 1. Using the formalism given in § 4 of Aravetal. (2012), we can deduce the n of these components e fromtheobservedtime lags.Supposean absorberin photoion- izationequilibriumexperiencesasuddenchangeintheincident 22 ionizing flux such that I(t > 0) = (1 + f)I(t = 0), where ) NH −1≤ f ≤∞.Thenthetimiescaleforchangeintiheionicfraction ( g isgivenby: o 2004 2002 1998 L 2013 2011 n α −1 t∗ = −fαn i+1 − i−1 , (3) " i e n α !# 21 CII 2004 SiIII 2004 SiIII 2011 CIV 2002 NV 1998 HI 1998 owsefhpaearrgeaitvαeeienipseotlchehmesreseinhctooiwmnibinoigninatirztoiaoutnigohrnactsehtaaognfegieoisn.aFinradonmtdhenthi7eisd4at0hyesdsafreyapscattrihaoatn-t ingepochswithnochange,wecandeduce3.5 < log(n ) < 4.5 -2 -1 0 e Log (U ) (cm−3)forbothcomponents,otherwisetheirtroughswouldnot H react to changes in incident ionizing flux in the observed way, Fig.5.A photoionizationphaseplotshowingtheionizationso- despitethelargechangesinincidentionizingfluxoverthatpe- lutionsforcomponent1forall5epochs.The2013epochsolu- riod.AssumingasimilarU tothatofcomponent1(seediscus- H tionisidenticaltotheoneshowninFig.3.Forthe1998,2002, sionbelow),thisn rangeyieldsdistancesof5<R<15parsec. e and 2004 epochs we used the high SED (see Fig. 2) with the Similarly,component6showsCivandNvtroughsin2011but same abundances, and their ionization solutions are shown in not in 2002. This nine years timescale yields R < 100 parsec. bluecrosses.DashedlinesrepresentN lower-limitthatallow ion Component2andcomponent4donotshowchangesintheUV thephase-spaceabovetheline,whilethedottedlinesareupper- absorptionbetweenanyoftheepochs.Therefore,wecanderive limits that allow the phase-spacebelow the line (onlythe most alowerlimitfortheirdistanceofR>130parsec. restrictive constraints are shown). N is fixed for all epochsat H ConstrainingN andU :Itisnotfeasibletoputphysically H H the value determined from the 2013 solution. For 1998, 2002 interestingconstraintsoncomponents2,4and6.First,theyonly and2004thedifferenceinUH valuesisdeterminedbytheratio showtroughsfromCiv,Nv,andLyα,which(basedonouranal- of fluxesat 1350Åand the actual value is anchoredby the ob- ysisofcomponent1)areprobablyhighlysaturated.Second,the servedNionconstraints.Asexplainedin§3.3,theUH positionof ionizationtime-scalesofcomponents2and4arelargerthan16 the2011epochislesstightlyconstrained.Thesolutionforeach years.Therefore,evenifaU canbededucedfromthemeasure- H epochsatisfiesalltheNionconstraintsforthatepoch. ments,itwillonlybearepresentativeaveragevalueforaperiod oftimelargerthan16years. Figure 6 shows the N –U phase plot for component 3 H H based on the N reported in Table A.2 (the Lyα and Civ and Nv are clearly seen in the 2011 epoch but not any of the N lowerlimitsioanretriviallysatisfiedbythelowerlimitshown ion other ionic species seen in 2013. To explain this situation we for the Nv N ). The phase plot constraints given by the N ion ion assume that the obscurer was present at the 2011 epoch, but it measurementsare mostly parallel to each other. Therefore,the onlyblockedsomewherebetween50–90%oftheemittedioniz- N –U constraints are rather loose, allowing a narrow strip H H ingradiationbetween1Ryand1keV.Thepossiblepresenceof from log(N ) = 19.6 and log(U ) = −2, to log(N ) = 21.5 H H H a weaker X-rayobscureris also suggestedbybroadabsorption and log(U ) = −1.1. If we take the most probable value of H onthebluewingoftheCIVemissionlinein2011thatisweaker log(U ) = −1.3, the distance estimate for component 3 will H thanthatseenin2013. dropby30%comparedwiththeestimateof5 < R < 15parsec, Tosummarize,asimplemodelbasedonafixedtotalcolumn- whichusedthelog(U )=−1.5ofcomponent1.Forcomponent H density absorber, reacting to changes in ionizing illumination, 5, the situation is rather similar, as the detected Siiii and Siiv matches the very different phenomenology seen in all high- allowanarrowstripfromlog(N ) = 19.2andlog(U ) = −1.8, H H resolutionUVspectraofcomponent1spanning16years.Figure to log(N ) = 20.7andlog(U ) = −1.2,while the lowestχ2 is H H 7givesaschematicillustrationofthetemporalmodel. achievedatlog(N )=20.7andlog(U )=−1.3. H H 4. Components2-6 5. ComparisonwiththeWarmAbsorberAnalysis Whatarethedistancesandphysicalconditions(N andU )for Howdothephysicalcharacteristicsinferredfromtheoutflows’ H H the other 5 outflow components(2-6)? As we show below, we UV diagnostics compare to the properties of the X-ray mani- can derive distances (or interesting limits of) for all these UV festation of the outflow known as the Warm Absorber? Since components and some loose constraints on the N and U of the soft X-ray flux in our 2013 data is very low due to the ap- H H components3and5.Figure7givesthevelocitiesanddistances pearanceoftheobscurer,wecannotcharacterizetheWAthatis ofall6components,aswellasaschematicillustrationfortheir connectedwiththe6UVoutflowcomponentsatthatepoch.Our temporalbehavior. main inferences about the WA are from the 2002 epoch when Constraining the Distances: components2-6 do notshow weobtainedsimultaneousX-rayandUVspectraoftheoutflow absorptionfromexcitedlevels(Exceptforcomponent3,whose (when no obscurer was present) that gave a much higher soft Cii/Cii*troughsonlyyieldalowerlimitforn .).However,com- X-rayflux (comparedwith the 2013observations)and allowed e ponents 3 and 5 show clear variations in their Siiii and Siiv a detail modeling of the WA in that epoch (Steenbruggeetal. troughs between our 2013/6/22and 2013/8/1observations, but 2005; Kaastraetal. 2014). Due to the inherent complications 6 N.Aravetal.:NGC5548 Table1.ComparisonbetweentheUVandWAcomponents 22 Component Velocitya log(N ) log(U ) H H (kms−1) (cm−2) (2002) UV1b -1160 21.5+0.4 −0.4+0.4 + SiIV UV2 -720 —−0.2 —−0.2 ) NH 21 UV3 -640 19.6−21.5 (−0.9)−0.0 ( g SiIII UV4 -475 — — o UV5 -300 19.2−20.7 (−0.7)−(−0.1) L UV6 -40 — — WAAc −588±34 20.30±0.12 −0.82±0.08 CII WAB −547±31 20.85±0.06 −0.09±0.05 20 WAC −1148±20 21.18±0.08 0.55±0.03 NV WAD −255±25 21.03±0.07 0.76±0.03 WAE −792±25 21.45±0.12 1.34±0.08 WAF −1221±25 21.76±0.13 1.53±0.05 avelocitycentroidofthecomponent -2 -1 0 Log (U ) bUVcomponents1-6arearrangedbydecreasingabsolutevelocity. H cParametersforWarmAbsorbercomponentsA-FarefromTableS2 Fig.6. A photoionization phase plot showing the ionization ofKaastraetal.(2014).Theyarearrangedbyincreasingionization solution for component 3 epoch 2013. As for component 1, parameter. we use the obscured SED and assumed proto-solar metalicity (Loddersetal. 2009). Solid lines and associated colored bands represent the locus of UH,NH models, which predict the mea- theWAareatlowenoughionizationstatestogiverisetotheUV sured Nion, and their 1σ uncertainties, while the dashed line is observedmaterial. thelowerlimitontheNvcolumndensitythatpermitsthephase- 3. Assuming constant N for the UV componentsand com- spaceaboveit.Thebluecrossisthebestχ2 solutionandissur- H ponentsAandBoftheWA. roundedbya1σ χ2bluecontour. Our temporalmodel for component1 has a constant N value H inalltheobservedepochs,including2002.Themodelalsopre- dicts the U of the 2002 epoch (see Fig. 5). We can therefore H compare the predictions of this model to the results of the re- analysisofthe2002WA(Kaastraetal.2014),providedthatthe of comparinganalyses on differentspectral regions(X-rayand N forcomponentsAandBoftheWAalsodidnotchangeover H UV) separated by 11 years(2002and 2013),of a clearly time- the 11 years between the epochs. We note that since UV com- dependent phenomenon, we defer a full comparison to a later ponent1 is the closest to the central source, the assumption of paper (Ebrero et al 2015). Here we outline some of the main constant N for the other UV components, over this 11 years H points in such a comparison, based on the analysis presented time period, is reasonable (see discussion in § 3.3). Therefore, here and the published analysis of the WA (Steenbruggeetal. we use the same ionization assumptionsfor UV components3 2005; Kaastraetal. 2014) and discuss some of the similarities and5asforcomponent1.Thatis,their N isfixedtothe2013 H andchallengesofsuchacombinedanalysis. valueandtheirlog(U ) = log(U ) +1.1,whicharethe H 2002 H 2013 1.KinematicSimilarity.Thereiskinematiccorrespondence valueswe list in Table 1. We do nothave empiricalconstraints betweentheUVabsorptiontroughsincomponents1–5andthe onthedistancesofWAcomponentsA-Ffromthecentralsource. sixionizationcomponents(A–F)oftheX-rayWA(Kaastraetal. 4.ComparingUVcomponents1and3tocomponentsAand 2014, see Table 1 here). X-ray components F and C span the BoftheWA. width of UV component 1. X-ray component E matches UV Using proto-Solar abundances (Loddersetal. 2009), our 2002 component2. The lowest ionization X-ray components, A and model prediction for UV component1 has log(N ) = 21.5+0.4 H −0.2 B comprise the full width of the blended UV troughs of com- cm−2, and an ionization parameter of log(U ) = −0.4+0.4 (see H −0.2 ponents 3 and 4 . Finally, X-ray component D kinematically § 3.1 and Fig. 5). This model gives a good match for the UV matchesUVcomponent5.However,asweshowinpoint2be- dataofthatepoch(2002)anditslog(U )isinbetweenthoseof H low, this kinematic matching is physically problematic as the WAcomponentsA[log(U ) = −0.8]andB[log(U ) = −0.1]. H H ionizationparametersofWAcomponentsC,D,EandFaretoo However, there are two inconsistencies between the models. hightoproduceobservedtroughsfromCivandNvthatareob- First,componentsAandBhaveatotallog(N )=20.95±0.1or H servedinalltheUVcomponents. about2σdisagreementwiththatofUVcomponent1.Thisdis- 2.Comparingsimilarionizationphases. crepancyis mainlydueto the limit onthe Ovii Nion thatarises WenotethattheX-rayanalysisoftheWAintheChandra2002 from the bound-free edge of this ion in the X-ray data. In the observations uses a different ionization parameter (ξ) than the WAmodel,about95%oftheOviiNion arisesfromcomponents U we use here; where ξ ≡ L/(n r2) (erg cm) with n being A andB. Second,the reportedvelocitycentroidsfor WA com- thHehydrogennumberdensity,LtheHionizingluminositybHetween ponents A and B (−588± 34 km s−1 and −547± 31 km s−1, 13.6eVand13.6keVandrthedistancefromthecentralsource. respectively)arein disagreementwith the velocitycentroidsof ForthehighSED,log(U ) = log(ξ)+1.6.InTable1, we give UVcomponent1(−1160kms−1)andits300kms−1width. H thelog(U )fortheWAcomponents.FromtheWAanalysisand UVcomponent3hasavelocitycentroidat–640kms−1and H figure5here,wededucethat90%oftheWAmaterial(compo- awidthof∼200kms−1,andthereforeisabettervelocitymatch nentsC,D,EandFinTableS2ofKaastraetal.2014)isintoo with WA components A and B. The large uncertainties in the high an ionization stage to produce measurable lines from the inferredN andU forUVcomponent3(seeFig.6andTable H H UVobservedions(e.g.,Civ,Nv).OnlycomponentAandBof 1), make these values consistent with the N and U deduced H H 7 N.Aravetal.:NGC5548 for WA components A and B. However, the uncertainties also 6. Discussion allow UV component 3 to have a negligible N compared to H WAcomponentsAandB. 6.1.ComparisonWithPreviousWork We notethatwiththecurrentanalyses,thebettertheagree- Howdotheseresultscomparewiththepreviousextensivework mentbetweenUVcomponent3andWA componentsA andB, ontheenduringoutflowinNGC5548?Forthefirsttimeasim- theworseisthedisagreementbetweenUVcomponent1andWA ple model of a constant NH absorber yields a physical picture componentsAandB.ThisisbecauseUVcomponent1already thatisconsistentwiththesubstantialtroughvariationseeninall predictshighervaluesofN andOviiN thanaremeasuredin epochsofhighresolutionUVspectralobservations.Thetrough H ion WA components A and B, and the kinematics of the deduced changesareexplainedsolely bytheobserveddifferencesinthe Ovii N disagree considerably. In points 5 and 6 below we incidentionizingflux.Inaddition,wedeterminerobustdistances ion identifytwopossiblewaystoalleviateandeveneliminatethese for(orlimitson)all6kinematiccomponents.Ourresultsdiffer apparentdiscrepancies. considerablyfromthosepreviouslyfoundforthisoutflow(e.g., Crenshawetal. 2009). This is due to the powerful diagnostics 5. Existence of considerable Ovii N at the velocity of ion that were revealedduring the 2013 campaign,recognizingthat UV component 1. The 2002 X-ray spectra presented by manyoftheobservedtroughsarehighlysaturated(e.g.,theex- Steenbruggeetal. (2005) consist of two different data sets that istence of similar depth Civ and Pv troughs in component 1; wereacquiredinthesameweek:170ksobservationstakenwith seediscussioninBorguetetal.2012a)andthefactthatthepre- the High EnergyTransmissionGratingSpectrometer(HETGS) vious work did not account for the emergence of Siiii troughs and 340 ks observations with the Low Energy Transmission associatedwithcomponents1and3inthe2004data. Grating Spectrometer (LETGS). Figure 2 in Steenbruggeetal. (2005)showssomeofthelowionizationWAtroughsinvelocity presentation,wherethedottedlinesshowthepositionoftheUV 6.2.ImplicationsforBALVariabilityStudies components (with somewhat different velocity values than we use here due to the use of a slightly differentsystemic redshift Our multiwavelength campaign has significant consequences fortheobject).TheLETGSdataoftheOviiandOvtroughsare for studies of absorption trough variability in quasar outflows, consistentwithonemainkinematiccomponentmatchingtheve- and in particular for the intensive studies of trough variabil- locityofUVcomponent3.However,themorenoisybuthigher ityinBroadAbsorptionLine(BAL)quasars(e.g.,Barlowetal. resolutionHETGS data for the same transitions, show two sub 1992;Capellupoetal.2012;FilizAketal.2013,andreferences troughsone correspondingto UV component1 and one to UV therein).As discussed in § 3.3, the two main proposedmecha- component3.Therefore,itispossiblethatmuchoftheOviiN nisms for trough variability are (1) reaction to changes in ion- ion isassociatedwithUVcomponent1. izing flux of a constant absorber (which is the model we suc- cessfullyusetoexplaintheNGC5548outflowtroughchanges); 6. Abundances considerations: As mentioned in § 3, for the and(2)absorbermotionacrossthelineofsight(e.g.,Gabeletal. UVanalysis,weusepureproto-Solarabundances(Loddersetal. 2003),whichaswedemonstrated,cannotexplainthevariability 2009), which well-matchthe measured Nion from the UV data. oftheNGC5548outflow(see§3.3). But these models produce considerably more Ovii N in the ion In some BAL cases the rest-frame UV flux around 1350 2002epoch,thanthemeasuredOvii N inthewarmabsorber ion Å does not change appreciably between the studied epochs data. However, the N (and therefore also the Ovii N ) of H ion whilesignificanttroughvariabilityisobserved.Thisbehavioris UV component1 is critically dependenton the assumed phos- takenasanargumentagainstmechanism(1)(e.g.,Barlowetal. phorus abundance. Ionization models with all elements having 1992; FilizAketal. 2013) as the ionizing flux (below 912Å) proto-Solar abundances except phosphorus, for which we as- is assumed to correlate with the longer wavelength UV flux. sume twice proto-Solar abundance, preserve the fit to the UV However,oursimultaneousUV/X-rayobservationsshowaclear data(at1/3the N value)andatthesametimematchtheOvii H casewheretheionizingphotonfluxdropsbyafactorof25be- N measured in the X-ray warm absorber at the 2002 epoch. ion tween the 2002 and 2013 epochs, while the 1350 Å UV flux Larger over-abundances of phosphorus further reduce the N valueandthereforethepredictedOviiN forthe2002epoch.H actuallyincreasesby50%. ion ThespectroscopicdatasetofNGC5548giveshighS/Nand But are such assumed abundances physically reasonable? high-resolutionspectraofmanyUVabsorptiontroughs,andsi- AGN outflows are known to have abundances higher by fac- multaneously yields the crucial soft X-ray flux that is respon- tors of two (Aravetal. 2007) and even ten (Gabeletal. 2006) sible for the ionization of the outflow. Such data can constrain comparedwiththeproto-Solarvalues.Inparticular,phosphorus troughvariabilitymodelsfarbetterthanthestandardvariability abundance in AGN outflows, relative to other metals, can be a datasetswhereoneortworest-frameUVBALareobservedat factorofseveralhigherthaninproto-Solarabundances(see§4.1 two (andlessfrequentlyinmore)epochs(e.g.,Capellupoetal. in Aravetal. 2001). Furthermore, the theoretical expectations 2012;FilizAketal.2013). forthevalueofchemicalabundancesinanAGNenvironmentas a function of metalicity are highly varied. The leading models can differ about relative abundances values by factors of three 6.3.ImplicationsfortheX-rayObscurer ormore(e.g.,comparingthevaluesofHamann&Ferland1993; What do our results about the enduring outflow tell us about Balleroetal.2008). theX-rayobscurerandthebroadUVabsorptiondiscoveredby WeconcludethatifroughlyhalfoftheobservedOviiN is Kaastraetal.(2014)?ThederivedtransmissionfortheX-rayob- ion associatedwithUVcomponent1(asdiscussedinpoint5),and scurerisconsistentwiththeSEDrequiredforthe2013spectrum, iftherelativeabundanceofphosphorustooxygenistwicesolar thus showing that the obscurer is closer to the super massive or larger,than the N , U and velocity distribution of the WA blackholesthan3.5pc.,andthatitsshadowinfluencesthecon- H H andUVoutflowinggascanbeconsistent. ditionsinthemoredistantnarrowUVabsorbers. 8 N.Aravetal.:NGC5548 Fig.7.Anillustrationofthephysical,spatialandtemporalconditionsoftheoutflowsseeninNGC5548.Alongthetimeaxiswe show the behavior of the emission source at the 5 UV epochs and give its UV flux values (measured at 1350 Å rest-frame and giveninunitsof10−14 ergss−1 cm−2 Å−1).Theobscurerissituatedatroughly0.01pcfromtheemissionsourceandisonlyseen in2011and2013(itismuchstrongerin2013).Outflowcomponent1showsthemostdramaticchangesinitsabsorptiontroughs. Differentobservedionicspeciesarerepresentedascoloredzoneswithintheabsorbers.Thetroughchangesarefullyexplainedby ourphysicalmodelshowninFigure5.Usingcomponent1Ciii*troughs,whichareonlyseeninthe2013epoch,wedetermineits numberdensity(seeFigure4)tobelog(n ) = 4.8±0.1cm−3,andthereforeitsdistance,R=3.5parsec.Thedistancesfortheother e componentsare discussed in § 4. Dimmer clouds represent epochs where components2-6 did not show new absorption species comparedwiththe2002epoch. 7. Summary of log(N ) = 21.5+0.4 cm−2, an ionization parameter of H −0.2 log(U ) = −1.5+0.4, and an electron number density of In 2013 we executed the most comprehensive multiwave- H −0.2 log(n ) = 4.8±0.1cm−3.Thiscomponentprobablycarries length spectroscopic campaign on any AGN to date, directed e thelargestN associatedwiththeUVoutflow.See§3.1and at NGC 5548. This paper presents the analysis’ results from H ourHST/COSdataofthe oftenobservedUVoutflow,whichis §3.2forelaboration. 2. For component 1 a simple model based on a fixed total detectedin 6 distinctkinematiccomponents.Ourcampaignre- column-densityabsorber,reactingtochangesinionizingil- vealedanunusuallystrongX-rayobscuration.Theresultingdra- lumination,matchestheverydifferentionizationstatesseen maticdecreaseinincidentionizingfluxontheoutflowallowed atfivespectroscopicepochsspanning16years.See§3.3for us to construct a comprehensivephysical, spatial and temporal elaboration. pictureforthewell-studiedAGNwindinthisobject.Ourmain 3. Thewealthofobservationalconstraintsmakesourchanges- findingsarelistedbelow(seeFig.7foragraphicillustrationof of-ionization model a leading contender for interpreting ourresults). troughvariabilitydataofquasaroutflows,inparticularBroad AbsorptionLine(BAL)variability.See§6.2forelaboration. 1. Ourbestconstraintsareobtainedforcomponent1(thehigh- est velocity component): It is situated at 3.5±1 pc from 4. Components3and5aresituatedbetween5-15pcfromthe centralsource,component6iscloserthan100pcandcom- the central source, has a total hydrogen column density 9 N.Aravetal.:NGC5548 ponents2and4arefurtheroutthan130pc.See§4forelab- deVaucouleurs,G.,deVaucouleurs, A.,Corwin,Jr.,H.G.,etal.1991,Third oration. ReferenceCatalogueofBrightGalaxies.VolumeI:Explanationsandrefer- 5. A detailed comparison of the physical characteristics in- ences.VolumeII:Dataforgalaxiesbetween0hand12h.VolumeIII:Datafor galaxiesbetween12hand24h. ferred from the outflows’ UV diagnosticswith those of the Dere,K.P.,Landi,E.,Mason,H.E.,MonsignoriFossi,B.C.,&Young,P.R. X-ray Warm Absorber is deferred to a future paper. Here 1997,A&AS,125,149 we outline some of the main points in such a comparison, Edmonds,D.,Borguet,B.,Arav,N.,etal.2011,ApJ,739,7 anddiscusssomeofthesimilaritiesandchallengesofsucha Faucher-Gigue`re,C.-A.,Quataert,E.,&Murray,N.2012,MNRAS,420,1347 combinedanalysis.See§5forelaboration. Ferland, G.J.,Porter, R. L.,van Hoof, P.A.M., et al. 2013, Rev. Mexicana Astron.Astrofis.,49,137 FilizAk,N.,Brandt,W.N.,Hall,P.B.,etal.2013,ApJ,777,168 Acknowledgements. This work was supported by NASA through grants for Gabel,J.R.,Arav,N.,&Kim,T.2006,ApJ,646,742 HSTprogramnumber13184fromtheSpaceTelescopeScienceInstitute,which Gabel,J.R.,Crenshaw,D.M.,Kraemer,S.B.,etal.2003,ApJ,583,178 is operated by the Association of Universities for Research in Astronomy, Gabel,J.R.,Kraemer,S.B.,Crenshaw,D.M.,etal.2005,ApJ,631,741 Incorporated, under NASA contract NAS5-26555. SRON is supported finan- Green,J.C.,Froning,C.S.,Osterman,S.,etal.2012,ApJ,744,60 ciallybyNWO,theNetherlandsOrganizationforScientificResearch.M.M.ac- Hamann,F.,Barlow,T.A.,Junkkarinen,V.,&Burbidge,E.M.1997,ApJ,478, knowledgesthesupportofaStudentshipEnhancementProgrammeawardedby 80 theUKScience&TechnologyFacilitiesCouncil(STFC).P.-O.P.andF.U.thanks Hamann,F.&Ferland,G.1993,ApJ,418,11 financial supportfromtheCNESandfromtheCNRS/PICS.F.U.acknowledg Hopkins,P.F.&Elvis,M.2010,MNRAS,401,7 PhDfundingfromtheVINCIprogramoftheFrench-ItalianUniversity.K.C.S. Kaastra,J.S.,Detmers,R.G.,Mehdipour,M.,etal.2012,A&A,539,A117 wantstoacknowledge financial supportfromtheFondoFortalecimiento dela Kaastraetal.,J.S.2014,Science ProductividadCientficaVRIDT2013.E.B.issupportedbygrantsfromIsrael’s Korista,K.T.,Alloin,D.,Barr,P.,etal.1995,ApJS,97,285 MoST, ISF (1163/10), and I-CORE program (1937/12). J.M. acknowledges Landi,E.,Young,P.R.,Dere,K.P.,DelZanna,G.,&Mason,H.E.2013,ApJ, fundingfromCNRS/PNHEandCNRS/PICSinFrance.G.M.andF.U.acknowl- 763,86 edge financial support from the Italian Space Agency under grant ASI/INAF Lodders,K.,Palme,H.,&Gail,H.-P.2009,in”Landolt-Bo¨rnstein -GroupVI I/037/12/0-011/13. B.M.P. acknowledges support from the US NSF through AstronomyandAstrophysicsNumericalDataandFunctionalRelationships grant AST-1008882. M.C, S.B, G.Mand A. D. R., acknowledge INAF/PICS inScienceandTechnologyVolume,ed.J.E.Tru¨mper,44 support.G.P.acknowledgessupportviaanEUMarieCurieIntra-Europeanfel- Mehdipouretal,M.2014 lowshipundercontractno.FP-PEOPLE-2012-IEF-331095.M.W.acknowledges Ostriker,J.P.,Choi,E.,Ciotti,L.,Novak,G.S.,&Proga,D.2010,ApJ,722, the support of aPhD studentship awarded bythe UK Science & Technology 642 FacilitiesCouncil(STFC). Pancoast,A.,Brewer,B.J.,Treu,T.,etal.2013,ArXive-prints The data used in this research are stored in the public archives of the Schlegel,D.J.,Finkbeiner,D.P.,&Davis,M.1998,ApJ,500,525 satellites thatareinvolved. WethanktheInternational SpaceScience Institute Soker,N.&Meiron,Y.2011,MNRAS,411,1803 (ISSI)inBern forsupport. This workisbasedonobservations obtained with Steenbrugge,K.C.,Kaastra,J.S.,Crenshaw,D.M.,etal.2005,A&A,434,569 XMM-Newton,anESAsciencemissionwithinstrumentsandcontributionsdi- rectly funded byESAMember States andthe USA (NASA).It is alsobased on observations with INTEGRAL, an ESA project with instrument and sci- ence data center funded by ESA member states (especially the PI countries: Denmark, France, Germany, Italy, Switzerland, Spain), Czech Republic, and Poland and with the participation of Russia and the USA. This work made useofdatasuppliedbytheUKSwiftScienceDataCentreattheUniversityif Leicester.ThisresearchmadeuseoftheChandraTransmissionGratingCatalog andarchive(http://tgcat.mit.edu).Thisresearchhasmadeuseofdataobtained withtheNuSTARmission,aprojectledbytheCaliforniaInstituteofTechnology (Caltech), managed by the Jet Propulsion Laboratory (JPL) and funded by NASA,andhasutilizedtheNuSTARDataAnalysisSoftware(NUSTARDAS) jointly developed bytheASIScience DataCenter (ASDC,Italy)andCaltech (USA).Figure7wascreatedbyAnnFeildfromSTScI. References Arav,N.,Becker,R.H.,Laurent-Muehleisen,S.A.,etal.1999,ApJ,524,566 Arav, N., Borguet, B., Chamberlain, C., Edmonds, D., & Danforth, C. 2013, MNRAS,436,3286 Arav,N.,deKool,M.,Korista,K.T.,etal.2001,ApJ,561,118 Arav,N.,Edmonds,D.,Borguet,B.,etal.2012,A&A,544,A33 Arav,N.,Gabel,J.R.,Korista,K.T.,etal.2007,ApJ,658,829 Arav,N.,Kaastra,J.,Kriss,G.A.,etal.2005,ApJ,620,665 Arav,N.,Korista,K.T.,&deKool,M.2002,ApJ,566,699 Arav,N.,Moe,M.,Costantini,E.,etal.2008,ApJ,681,954 Ballero,S.K.,Matteucci,F.,Ciotti,L.,Calura,F.,&Padovani,P.2008,A&A, 478,335 Barlow,T.A.,Junkkarinen,V.T.,Burbidge,E.M.,etal.1992,ApJ,397,81 Borguet,B.C.J.,Arav,N.,Edmonds,D.,Chamberlain,C.,&Benn,C.2013, ApJ,762,49 Borguet,B.C.J.,Edmonds,D.,Arav,N.,Benn,C.,&Chamberlain,C.2012a, ApJ,758,69 Borguet,B.C.J.,Edmonds,D.,Arav,N.,Dunn,J.,&Kriss,G.A.2012b,ApJ, 751,107 Capellupo,D.M.,Hamann,F.,Shields,J.C.,Rodr´ıguezHidalgo,P.,&Barlow, T.A.2012,MNRAS,422,3249 Ciotti,L.,Ostriker,J.P.,&Proga,D.2010,ApJ,717,708 Costantini,E.,Kaastra,J.S.,Arav,N.,etal.2007,A&A,461,121 Crenshaw,D.M.,Kraemer,S.B.,Boggess,A.,etal.1999,ApJ,516,750 Crenshaw,D.M.,Kraemer,S.B.,Gabel,J.R.,etal.2003,ApJ,594,116 Crenshaw,D.M.,Kraemer,S.B.,Schmitt,H.R.,etal.2009,ApJ,698,281 deKool,M.,Korista,K.T.,&Arav,N.2002,ApJ,580,54 10
Description: