Astronomy&Astrophysicsmanuscriptno.meusinger (cid:13)c ESO2010 January18,2010 J004457+4123 (Sharov21): not a remarkable nova in M31 but a background quasar with a spectacular UV flare H.Meusinger1,M.Henze2,K.Birkle3,4,W.Pietsch2,B.Williams5,D.Hatzidimitriou6,7,R.Nesci8,H.Mandel4, S.Ertel9,A.Hinze10,andT.Berthold11 1 Thu¨ringerLandessternwarteTautenburg,Sternwarte5,D–07778Tautenburg,Germany,e-mail:[email protected] 2 Max-Planck-Institutfu¨rextraterrestrischePhysik,Giessenbachstraße,D–85748,Garching,Germany 3 Max-Planck-Institutfu¨rAstronomie,Ko¨nigstuhl17,D–69117Heidelberg,Germany 0 4 ZAH,LandessternwarteHeidelberg,Ko¨nigstuhl12,Universita¨tHeidelberg,D–69117Heidelberg,Germany 1 5 DepartmentofAstronomy,Box351580,UniversityofWashington,Seattle,WA98195,USA 0 6 DepartmentofAstrophysics,AstronomyandMechanics,FacultyofPhysics,UniversityofAthens,Panepistimiopolis,GR15784 2 Zografos,Athens,Greece 7 IESL,FoundationforResearchandTechnology,GR71110Heraklion,Crete,Greece n 8 DepartmentofPhysics,UniversityofRomaLaSapienza,Rome,Italy a 9 Institutfu¨rTheor.PhysikundAstrophysik,Christian-Albrechts-Universita¨tzuKiel,Leibnizstraße15,D–24118Kiel,Germany J 10 AstronomischesInstitut,Universita¨tBern,Sidlerstraße5,CH–3012Bern,Switzerland 8 11 SternwarteSonneberg,Sternwartestr.32,D–96515Sonneberg,Germany 1 Received/Accepted ] O ABSTRACT C Aims. WeannouncethediscoveryofaquasarbehindthediskofM31,whichwaspreviouslyclassifiedasaremarkablenovainour . h neighbourgalaxy.ItisshownheretobeaquasarwithasinglestrongflarewheretheUVfluxhasincreasedbyafactorof∼20.The p presentpaperisprimarilyaimedattheremarkableoutburstofJ004457+4123(Sharov21),withthefirstpartfocussedontheoptical - spectroscopyandtheimprovementinthephotometricdatabase. o Methods. We exploited the archives of photographic plates and CCD observations from 15 wide-field telescopes and performed r targettednewobservations.Inthesecondpart,wetrytofittheflarebymodelsof(1)gravitationalmicrolensingduetoastarinM31 t s and(2)atidaldisruptionevent(TDE)ofastarclosetothesupermassiveblackholeofthequasar. a Results.BoththeopticalspectrumandthebroadbandspectralenergydistributionofSharov21areshowntobeverysimilartothat [ ofnormal,radio-quiettype1quasars.Wepresentphotometricdatacoveringmorethanacenturyandresultinginalong-termlight 1 curvethatisdenselysampledoverthepastfivedecades.Thevariabilityofthequasarischaracterizedbyagroundstatewithtypical v fluctuationamplitudesof∼0.2magaroundB¯ ∼ 20.5,superimposedbyasingularflareof∼2yrduration(observerframe)withthe 1 maximumat1992.81.Thetotalenergyintheflareisatleastthreeordersofmagnitudeshigherthantheradiatedenergyofthemost 9 luminoussupernovae,providedthatitcomesfromanintrinsicprocessandtheenergyisradiatedisotropically.Theprofileoftheflare 9 lightcurveisasymmetricshowinginparticularasuddenincreasebeforethemaximum,whereasthedecreasingpartcanberoughly 2 approximated by at−5/3 power law.Bothpropertiesappear tosupport thestandard TDEscenariowherea∼ 10M⊙ giant star was . shreddedinthetidalfieldofa∼2...5108M⊙blackhole.Theshortfallbacktimederivedfromtheobservedlightcurverequiresan 1 ultra-closeencounterwherethepericentreofthestellarorbitisdeepwithinthetidaldisruptionradius.Thissimplemodelneglects, 0 however,theinfluenceofthemassiveaccretiondisk,aswellasgeneral-relativisticeffectsontheorbitofthetidaldebris.Gravitational 0 microlensingprobablyprovidesanalternativeexplanation,althoughtheprobabilityofsuchahighamplificationeventisverylow. 1 : Keywords.Quasars:general–Quasars:individual:J004457+4123–Galaxies:individual:M31–Gravitationallensing–Blackhole v physics i X r a 1. Introduction of the source, as is the case for quasars, takes a long time. This also holds for recurrent events, e.g. novae, with intrinsi- cally shortertime scaleswhereit isnecessaryto coveralso the Temporalvariability is one of the most conspicuousproperties wide gaps between the single events. Presently, data mining in forseveralclassesofinterestingastrophysicalobjects.Owingto archives,mostnotablyintheplatearchivesfromlargeSchmidt theunprecedentedcombinationofskycoverageandphotometric telescopes, remains the only approach if the light curves have accuracy,discoveriesfromtheLargeSynopticSurveyTelescope tocoveratimeintervalofdecadesintherestframe,atleastfor (LSST), the Panoramic Survey Telescope & Rapid Response high-redshiftquasars. System (Pan-STARRS), the Palomar-QUEST (PQ) survey, or thePalomarTransientFactory(PTF)willprovidegreatadvances In the context of the search for and the identification of in the understanding of variable processes, especially of rare optically variable star-like sources in the field of the bright transient phenomena (Gezari et al. 2008; Strubbe & Quataert Local Group spiral galaxy M31 (Pietsch et al. 2005a; Henze 2009; Quimby et al. 2009). However, given the nature of the et al. 2008), our interest was pointed towardthe apparentnova problem,the creation of the observationaldatabase for investi- J004457+4123,originallydiscoveredbyNedialkovetal.(1996) gatingthevariabilityonlongtimescalesinthereferenceframe and describedin more detailby Sharovet al. (1998). The light 2 H.Meusingeretal.:AspectacularUVquasarflare curve presented by these authors clearly shows a strong bump 2. Observationaldata with a maximum brightening by more than 3mag in the year 2.1.Spectroscopy 1992 while the source remained constant both in the 23 years before and the 5 years after. Sharov et al. suggest that it is a The optical spectrum was obtained with the Double Imaging “remarkablenova” in M31, but underscorethat it “differs dra- Spectrograph (DIS) on the 3.5-m telescope at Apache Point matically from typical representativesof this class of objects”. Observatory (APO) in New Mexico, USA during a campaign Following the terminology of these authors (nova 21), we de- tofollow-upX-raysourcesinM31.Twoexposuresweretaken note the object as Sharov21 throughoutthis paper. A possible foratotalof4500seconds.Forthebluespectralrange(3200to X-raycounterpartwas first discussedby Pietsch et al. (2005b), 5500Å),the B400reflectancegratingwasused witha disper- whosearchedforsupersoftX-raycounterpartsofopticalnovae sionof1.85Å perpixelyieldinga nominalresolutionofabout inM31andidentifiedSharov21withthesource[PFH2005]601 7Åincombinationwitha1′.′5entranceslit.TheR300grating, oftheircatalogueofXMM-NewtonEPICX-raysources(Pietsch witha dispersionof2.26Å perpixel,givesa resolutionof8Å et al. 2005a) and with the hardROSAT source[SHL2001]306 fromthecatalogueofSupperetal.(2001).Basedonthehardness for the red part (5000 to 10100 Å). The spectra were taken at oftheX-raysourceandthepeculiaropticallightcurve,Pietsch UT0300on2007-11-09.Theobservingconditionswereexcel- et al. (2005b) speculate that Sharov21 “may not be a nova at lentthroughthenight. all”.Herewepresent,forthefirsttime,opticalfollow-upspec- The spectra were reduced, wavelength calibrated, and flux troscopywhichrevealsSharov21tobeaquasar. calibrated using the standard IRAF routines (ccdproc, iden- Fromtheverybeginningoftheinvestigationofactivegalac- tify,sunsfunc,apextract,apall).Wavelengthcalibrationwasper- ticnuclei(AGN),variabilityisknowntobeadiagnosticproperty formedusingHeNeArlampexposurestakenjustbeforetheob- of this object class and has been successfully used as a crite- jectexposures,andfluxcalibrationwasperformedusingaspec- rionfortheselectionofquasarcandidatesinanumberofstud- trumofthespectrophotometricstandardstarBD+28-4211. ies (e.g., Kron & Chiu 1981; Majewski et al. 1991; Hawkins & Ve´ron 1993; Meusinger et al. 2002, 2003; Rengstorf et al. 2.2.Opticalphotometry:long-termlightcurve 2004). AGN originally misclassified as variable stars are nei- therunprecedentednorunexpected.Themostfamouscaseisthe The light curve published by Sharov et al. (1998) is based on prototypicalblazarBLLac,discoveredbyCunoHoffmeisterin Bbandobservationstakenwithfourtelescopesbetween1969.0 1930.However,themisclassification of a luminousquasarasa and 1997.7 with a good coverage of the outburst phase. The novaishighlyremarkablebecauseitindicatesasingular,strong presentstudyisaimedatanextendedandbettersampledlong- outburstwhichpointstowardarareandinterestingtransientphe- term light curve. We exploited several data archives and com- nomenon. binedtheresultswith thedataavailablein the literature.Inad- It has long been understood that the observed flux varia- dition, targetted new observationsfor another 16 epochs in the tionsofAGNsholdkeystothestructureoftheradiationsource. years2006to2009weretakenwiththeCCDSchmidtcameraof Thephysicalmechanismsbehindthesefluctuationsarehowever theTautenburg2mtelescopeandwiththefocalreducercamera still poorly understood. Frequently discussed scenarios for the CAFOSatthe2.2mtelescopeonCalarAlto1,Spain.Mostofthe origin of the observed optical/UV broad-band long-term (non- archival photographicplates were digitized in the frame of the blazar)variabilityrelatedtomassiveorsupermassiveblackholes presentwork usingthe TautenburgPlate Scanner(Brunzendorf ingalaxycentresincludevariousprocessessuchasinstabilities &Meusinger,1999)fortheTautenburgSchmidtplates,thehigh- and non-linear oscillations of the accretion disk (Taam & Lin qualitycommercialscannerattheAsiagoobservatory(Barbieri 1984; Abramovicz et al. 1989; Honma et al. 1991; Kawaguchi etal.2003)fortheAsiagoplates,andtheMicrotekScanMaker et al. 1998), multiple supernovae in the starburst environment 9800XL for the Sonneberg astrograph plates. The Calar Alto (Terlevich et al. 1992; Cid Fernandes et al. 1997), microlens- plateswerescannedfortheHeidelbergDigitizedAstronomical ing of the accretion disk or the broad line region by com- Plates (HDAP) project using a Heidelberger Druckmaschinen pact foreground objects (Chang & Refsdal, 1979; Irwin et al. NexscanF4100professionalscannerandareavailablefromthe 1989; Hawkins 1993; Schneider 1993; Lewis & Irwin 1996; GermanAstrophysicalVirtualObservatory(GAVO)2. Zackrisson et al. 2003), the disruption of a star which passes Asummaryofallusedobservationsfromthelastsixdecades within the tidal radius of the supermassive black hole (Hills is givenin Table1 (CFHT = Canada FranceHawaiiTelescope, 1975; Rees 1988, 1990; Komossa & Bade 1999; Komossa & INT = Isaac Newton Telescope, WFS = wide-field survey). N t Meritt2008;Gezarietal.2008),star-starcollisionsinthedense is the total number of all single exposures, N the number of e circumnuclear environment (Torricelli-Ciamponi et al. 2000), epochsin the lightcurve where the quasar has been measured. and interactions of the components in a supermassive binary The last columngivesthe source of the photometricreduction: blackhole(Sillanpa¨a¨ etal.1988;Lehto&Valtonen1996;Katz (1)thiswork,(2)Vilardelletal.(2006),(3)Monetalal.(2003), 1997;Liu&Chen2007). (4)Masseyetal.(2006),(5)Sharovetal.(1998). Thepresentpaperisaimedatthehighlypeculiarlightcurve Altogether, the light curve data pool contains more than of the quasar Sharov21 which is worth detailed investigation. 1100 single observations from 15 telescopes. Included are the We present the optical spectrum and a significantly improved B magnitudespublishedbySharovetal. (1998) for84epochs, light curve and discuss possible scenarios for the strong out- by the Local Group Galaxies Survey (LGGS; Massey et al. burst. The observations are described in Sect.2. The spectrum 2006) for 2 epochs, and by Vilardell et al. (2006) binned here andotherbasicpropertiesareanalysedinSect.3.Theoutburstis into 6 epochs. For the other observations, the photometric re- thesubjectofSect.4.Twomodelsarediscussedindetail:gravi- tationalmicrolensingandastellartidaldisruptionevent;alterna- 1 The Calar Alto Observatory of the Centro Astrono´mico Hispano tivescenariosarebrieflysummarizedaswell.Section5givesthe Alema´n,Almer´ıa,Spain,isoperatedjointlybytheMax-Planck-Institut conclusions.StandardcosmologicalparametersH0 =71kms−1 fu¨rAstronomieandtheInstitutodeAstrof´ısicadeAndaluc´ıa(CSIC). Mpc−1,Ωm =0.27,ΩΛ =0.73areusedthroughoutthepaper. 2 http://dc.zah.uni-heidelberg.de H.Meusingeretal.:AspectacularUVquasarflare 3 Table1.Observationalmaterialfortheconstructionofthelight notregularlydistributedbutshowstrongclustering.Westacked curve. the images taken with the same telescope within typically sev- eral days to a few weeks applying a quality-weighting factor telescope Nt Ne years source (Froebrich & Meusinger 2000) to obtain deeper images with improvedsignal-to-noiseratio. Thisprocedurewas notapplied (a)Digitizedphotographicplates: howeverfortheoutburstphasein1992whereweareinterested AsiagoSchmidt 24 8 1968.8–1993.1 (1) inahightemporalresolution. CalarAltoSchmidt 43 15 1983.0–2000.7 (1) CalarAlto1.2m 8 5 1976.7–1982.6 (1) The final light curve (Fig.1) comprises magnitudes at 221 PalomarSchmidt 4 4 1948.7–1989.7 (1),(3) detection epochs but still suffers from several gaps. In particu- Sonneberg40cm 9 1 1992.2–1992.6 (1) lar, no data are available for the early rising phase of the out- TautenburgSchmidt 362 77 1961.5–1997.0 (1) burst between March and August 1992. We checked the Wide FieldPlateDatebase3butfoundnoentriesforthistime.Alsothe (b)CCDobservations: search in the plate archives of the Baldone Schmidt telescope CalarAlto2.2m 3 2 2008.7–2009.3 (1) (Alksniset al. 1998) and of the 100/300cm Schmidttelescope CFHT3.6m 1 1 1993.8 (1) oftheKvistabergObservatoryrevealednoobservationsofM31 INT(WFS) 5 1 1998.8 (1) duringthattime. INT 522 6 1999.7–2003.7 (2) KittPeak4m 10 2 2000.8,2001.7 (4) Skinakas60cm 5 1 2007.6 (1) TautenburgSchmidt 30 14 2006.1–2009.7 (1) Table2.Detectionlimitsonplatestakenbefore1950. (c)OriginaldatafromSharovetal.(1998): telescope plate date m pg,lim fourothertelescopes ∼150 84 1969.0–1997.7 (5) yyyy-mm-dd BruceastrographLHK 23a 1900-09-14 18.8 BruceastrographLHK 265a 1901-08-18 18.0 Yerkes24inchreflector ? 1901-09-18 19.5 BruceastrographLHK 649a 1903-01-15 18.0 duction was done in the frame of the present study. We used BruceastrographLHK 842a 1903-09-27 18.0 the Source Extractor package (Bertin 1996) for object selec- BruceastrographLHK 1384a 1905-12-26 18.0 tion, background correction, and relative photometry, and the WaltzreflectorLHK 198 1907-11-02 17.5 LGGScatalogueforthephotometriccalibration.Thereduction WaltzreflectorLHK 369 1908-08-20 18.3 wasperformedunderESOMIDAS.Becauseofthestronglyin- WaltzreflectorLHK 603 1909-10-19 18.5 homogeneousbackgroundacrossthediskofM31(Henzeetal. WaltzreflectorLHK 4484 1934-09-17 18.5 2008),thephotometriccalibrationwasdonelocallyona8′×8′ BruceastrographLHK 7163a 1949-09-20 18.2 subimagearoundSharov21wheretypically∼100±50calibra- tionstarsfromtheLGGSwereidentified.Notethatthemagni- tudesgivenbySharovetal.(1998)havealsobeenderivedfrom standard stars close to the target. The blue magnitude for the It is useful to check also older historical observations of Palomar POSS1 plate (1953-10-09) is taken from the USNO- the Sharov21 field which are not deep enough to detect the B1.0 catalogue (Monet et al. 2003) whereas the magnitudes quasar in its faint stage but would allow discovering a previ- given there for the POSS2 plate do notagree with the impres- ousoutburst.Table2liststhoseobservationsfromthe40/200cm sion from the visual inspection of the images. We re-reduced Bruce double-astrograph and the 72 cm Waltz reflector of the the image cutouts from the Digitized Sky Survey (DSS2) and LandessternwarteHeidelberg-Ko¨nigstuhl(LHK)with mpg,lim ∼ derived B = 20.73 ± 0.23 and R = 19.66 ± 0.25. An early 18.Areproductionofaplatetakenin1901withthe24inchre- deep plate taken with the 1.2 m Samuel Oschin Telescope on flector at Yerkesobservatoryis shown by Hubble(1929); from 1948-09-29isshownintheHubbleAtlasofGalaxies(Sandage thevisualinspectionweestimateadetectionthresholdm∼19.5. 1961). The visual inspection shows that Sharov21 is detected Each of these observations excludes the occurrence of a flare withB=20.3±0.3. similar to that of 1992 for at least several tens of days around With very few exceptions the observations were made theirdatesofexposure. through filters reproducing the Johnson UBVR system. About 80% of the data points in the light curve are from observa- tions in the B band. From several pairs of observations, taken 3. GeneralpropertiesofSharov21 atnearlythesameepochbutwithdifferentfilters,wederivere- lations between colour indices and the B band magnitude.The In Sect.3.4 below we demonstrate that the previous classifica- results (Fig.3) are used to obtain B from observations in the tion of Sharov21 as a remarkable nova in M31 has to be re- other bands. Because quasar variability is usually not achro- jected. Our optical spectrum, presented below, clearly reveals matic(seeSect.3),suchcolourrelationsaremoresuitablethan thesourcetobeaquasar.Themostimportantpropertiesofthis single-epochcolourindices.Photographicmagnitudesm from quasararesummarizedinTable3.t isthetimeintervalforthe pg 1/2 bluesensitiveemulsionswithoutfilterweretransformedbyB∼ declinefrommaximumfluxtohalfthemaximum.Remarks:(1) m +0.1. position from LGGS, (2) ground state/maximum, (3) observer pg Therearetwobasiclimitationswithregardtothefinalsetof frame/quasarrestframe,(4) basedon Civ line, (5)mean value data. First, with B > 20 for most of the time, Sharov21 is too forthegroundstate. fainttobedetectedineveryplatearchive.Ahighfractionofob- servationsyields thereforeupperlimits only.In other cases the 3 Both the WFPDB version available via VizieR, (Tsvetkov et al. detectionsareclosetotheplatelimitresultinginrelativelylarge 1997),andtheupdatedversionavailableviathesearchbrowserdevel- photometric errors. Second, the epochs of the observationsare opedinSofia(http://draco.skyarchive.org/search)wereused. 4 H.Meusingeretal.:AspectacularUVquasarflare Fig.1. Long-termBlightcurvefromthedatasummarizedinTable1(symbolspluserrorbars;nophotometricerrorsavailablefor thedatafromSharovetal.(1998)).Dottedverticallines:upperlimitsfromimagesonwhichtheobjectisnotdetected.Forlucidity, onlyasmallfractionoftheupperlimitdataisshown. Table3.BasicpropertiesofSharov21(remarks:seetext). Measuredandderivedquantities Remark RA(2000) 00h44m57s.94 (1) Dec(2000) +41◦23′43′.′9 (1) redshiftz 2.109 projecteddistancefromM31centre 26′ apparentmagnitudeB 20.5/17.2 (2) foregrounddustreddeningE(B−V) 0.2mag absolutemagnitudeM −27.5/−30.7 (2) B dateofthemaximum(year/JD) 1992.81/2448918 t decline(days) 15/5 (3) 1/2 blackholemassM 5108M (4) bh ⊙ log(L /ergs−1) 46.6 (5) bol EddingtonratioL /L 0.60 (5) bol edd 3.1.Opticallightcurveandgeneralremarksonvariability Fig.2. Thelightcurveintheoutburst.Opensymbols:uncertain The light curve from the data discussed in Sect.2 is shown in data, dashed lines: upper limits, dotted horizontal line: ground Fig.1.Sharovetal.(1998)notethatthequasarwas“nearlycon- state. stant from 1969 through 1991 with B ≈ 20.5 and returned to this value one or two yearsafter the outburst”. Comparedwith the original data from Sharov et al., we (1) basically confirm To evaluate the variability in the ground state we com- theirfinding,(2)extendthe coveredtimeintervalbyaboutone pare Sharov21 with quasars from the Tautenburg-Calar Alto decadeineachtimedirection,and(3)fillsomebroadgaps(e.g., VariabilityandProperMotionSurvey(VPMS;Meusingeretal. betweentheyears1984and1990andbetweenthebeginningof 2002,2003).ForSharov21wehaveaBstandarddeviationσ = B 1993 and 1994). Based on the better sampling of our data, in- 0.27magfromtheTautenburgdata(0.26magfromalldata).For cludingtheupperlimits,thepossibilityofoutburstsinintervals theVPMSquasarswithsimilarredshifts(z=2.1±0.2)andcom- that were not covered by observations is thus significantly re- parable(extinction-corrected)meanmagnitudes(B¯ =19.7±0.2) duced. Hence, we conclude that the light curve can be divided we have σ = 0.26 mag in the VPMS field around M3 (8 B into (a) the faint state (with B¯ = 20.52), which can be con- quasars)and0.29maginthefieldaroundM92(6quasars).We sidered as the ground state, and (b) a single outburst, or flare, conclude that the flux variability of Sharov21, in its ground lasting ∼ 2yr (JD ∼ 2448500...2449300) where the quasar state,isnotunusuallystrong. was3.3magbrighterin themaximum.Theflare (Fig.2) shows VariationsoftheBbandfluxofSharov21arecorrelatedwith aslightlyasymmetricprofilewiththreephases:(1)agradualin- colourchanges.Theobservedrelations(Fig.3)arequalitatively crease between JD ∼2448500 and 2448880 with a gap in the in agreement with the typical properties of quasars. A harden- lightcurvebetweenMarchandAugust1992,followedby(2)an ing of the optical/UV continuumduring the bright phase is in- abruptrise to the maximumat JD ∼2448918,and (3) a quasi- dicatedbymulti-frequencymonitoringofselectedAGNs(Cutri exponential decline to the ground state at JD ∼2449300. The etal.1985;Edelsonetal.1990;Paltani&Courvoisier1994)as interpretationoftheoutburstwillbethesubjectofSect.4. well as by statistical studies of AGN ensemble variability (Di H.Meusingeretal.:AspectacularUVquasarflare 5 2006; their table1), an inclination angle i = 77◦, and a posi- tion angleof the major axisof 35◦ fromthe RC3, its projected distancefromthecentreof26′(5.7kpc)correspondstoagalac- tocentric distance of R = 16 kpc in the midplane of M31. A radiusof∼ 30kpcisarealisticassumptionfortheextentofthe brightdisk(e.g.Racine1991;Fergusonetal.2002;Irwinetal. 2005). The “Catalogue of Quasars and Active Galactic Nuclei (12thEd.)”(Ve´ron-Cetty&Ve´ron2006)listsonlythreequasars within100′ fromthecentreofM31.However,Sharov21is, to ourknowledge,thefirstquasardetectedbehindthediskofM31. Fig.3. ColourindicesofSharov21asafunctionofitsBmag- nitudefromquasi-simultaneousobservationsinbothbands. Fig.4. Map of the known quasars in the M31 field. Open squares:quasarsfromVe´ron-Cetty&Ve´ron(2006),framedas- Clementeetal. 1996;Cristianietal.1997;Tre`veseetal.2001; terisk:Sharov21. VandenBerketal.2004).Thistrendhasbeenconfirmedalsoby multi-epochspectroscopyofquasarsfromtheSloanDigitalSky Survey(SDSS) where it was shown that the emission lines are considerablylessvariablethanthecontinuum,beingstrongerin 3.3.Foregroundreddening thefaintstage,relativetothecontinuum,thaninthebrightphase A foreground spiral galaxy is expected to produce a substan- (Wilhiteetal.2005).ForSharov21,theUbandisdominatedby tialreddeningofabackgroundquasar(O¨stmanetal.2006).The thestrongLymanα/Nvline(Fig.5).ThecontributionoftheCiv “GalacticDustExtinctionService”4oftheNASA/IPACInfrared linetothefluxintheBbandismuchsmaller,andtheotherbands Science Archive, which is based on the method pioneered by arenearlypurecontinuum.ThecolourindicesB−V andB−R Schlegel et al. (1998), provides E(B− V) = 0.20mag for the are hence expected to become bluer when the quasar becomes position of Sharov21. Individual reddening values for a large brighter,whileU −Bbecomesredderatthesametime.Wilhite set of globular clusters in M31 were derived by Barmby et al. et al. present the colour differencesbetween the brightand the (2000)andFanetal.(2008)yieldingE(B−V)=0.12±0.03mag faintphaseasafunctionofredshift(theirFig.14)indicatingthat forthesixclusterswithin6′ fromSharov21.Thereddeningof ∆(u−g)hasalocalminimumatz≈2while∆(g−r)hasapeak the quasar must be stronger since only a fraction of the clus- at the same redshift. Sesar et al. (2007) found a similar result ters is expectedto be located behind the disk of M31. Finally, from the multi-epochphotometricdata of quasarsin the SDSS a simple model for the radial dependence of the extinction in stripe82withrespecttothegandrbands.Inanongoingstudy M31 derived by Hatano et al (1997) yields A = 0.68; i.e., ofquasarvariabilityintheSDSSstripe82(Meusingeretal.,in B E(B− V) = 0.17 mag for the standard Milky Way extinction preparation)weconfirmedtherelationshownbySesaretal.and curve(Savage&Mathis1979),whichseemstobevalidalsofor derivedcorrespondingrelationsfortheotherbandswhicharein M31(Barmby2000).HereweadoptE(B−V) = 0.20magfor linewiththeresultsfromWilhiteetal.(2005). thetotalforegroundreddeningofSharov21. 3.2.PositionbehindM31 3.4.Opticalspectrum Figure4revealsthatSharov21isseenthroughthediskofM31. The optical spectrum is shown in Fig.5. For comparison the The ellipses represent the D isophotes of M31, M32, and 25 compositespectrumof“normal”quasarsfromtheSloanDigital NGC205, respectively, according to the RC3 (de Vaucouleurs et al. 1991). For a distance of 750kpc for M31 (Vilardellet al 4 http://irsa.ipac.caltech.edu/applocations/DUST/ 6 H.Meusingeretal.:AspectacularUVquasarflare Sky Survey (SDSS; Vanden Berk et al. 2001) is plotted (thin viouslynotrelatedtoSharov21.Thenon-detectionintheBraun smooth curve) shifted to the redshift of Sharov21, the spectra survey implies an upper limit for the radio-loudnessparameter arenormalizedatλ4500Å.We derivedaredshiftofz = 2.109 R, i.e. the ratio of the 5 GHz radio flux density to the 2500 both from the fit of the SDSS composite and directly from the Å optical flux density in the quasar rest frame (Stocke et al. wavelengthsofthenarrowcomponentsoftheLymanαandNv 1992), logR∗ < 0.7 for a radio spectral index α = −0.3 and R lines.ComparedwiththeSDSScompositeSharov21hasared- logR∗ < 0.9 for α = −0.5. With the threshold logR∗ ≥ 1 R der continuum. We de-reddened the spectrum for foreground forradio-loudAGNs(e.g.,Whiteetal.2000)Sharov21isnota (z = 0) extinction adopting the Milky Way extinction curve. radio-loudflat-spectrumquasar. GoodagreementwiththemeanSDSSquasarspectrumisfound We re-analysed the archival XMM-Newton and ROSAT forE(B−V)=0.2mag(Fig.5,bottom),whichisperfectlyinline data. For computing fluxes from instrument dependent count withthereddeningvaluefromtheNASA/IPACInfraredScience rates,weusedanabsorbedpower-lawmodelwithagenericpho- Archive(seeabove). tonindexof1.7(seealsoPietschetal.2005a).We adoptedthe The de-reddened spectrum of Sharov21 is that of a typi- foreground extinction of E(B− V) = 0.20mag, derived form caltype1 quasar. Thereis noevidenceof unusualspectralfea- theopticaldata,whichtranslatestoa N of1.11021 cm−2 fol- H turesindicatinga peculiarnatureofSharov21.Comparedwith lowing Predehl et al. (1995). Based on this spectral model, we the SDSS composite the reddening corrected spectrum shows used the source count rates given by Pietsch et al. (2005a) for a stronger Fe bump at λ ∼ 7300...8300 Å (observer frame) [PFH2005] 601 in the XMM-Newton observation0151580401 whichpointstowardsarelativelyhighEddingtonratio(Donget (2003-02-06)toestimateanunabsorbedfluxof(5.8±1.9) 10−14 al. 2009). The Ciii] λ1909Å line appearsslightly weaker,but ergs−1cm−2inthe(0.2-10.0)keVbandandthemonochromatic notethatthelinecoincideswiththeNaiforegroundabsorption fluxF =(4.5±1.5)10−9Jyatν=1.71017Hz. ν at λλ5890,5896 Å. We notice further a weak unidentified ab- TheROSATdataoftheSharov21field5consistsof12obser- sorptionlineatthepositionoftheNivλ1486Åline. vationswiththePositionSensitiveProportionalCounter(PSPC) and3observationswiththeHighResolutionImager(HRI).The data analysis was done under ESO MIDAS within the EXSAS context. We performedsource detection aroundthe position of Sharov21 on the original event files and computed count rates and 3σ upper limits for all observations. There is just one 3σ detection of an X-ray source in the data set, identical with [SHL2001]306,whichissupplementedbyupperlimitsforthe rest of the observations. The flux estimated from the detection is consistent, within the errors, with the XMM-Newton data. Although the ROSAT observations were performed around JD 2449000,i.e. duringthe decline of the UV flare, no significant X-rayvariabilityisdetected.Notehowever,thattheonlyROSAT detectionisa veryfaintoff-axisPSPCdetectionandduetothe largepositionalerrorcircleofthisinstrumentwecannotassume adoubtlesscorrelationwith[PFH2005]601. Fig.5.Observed(top)andforegroundextinction-corrected(bot- tom) optical spectrum of Sharov21 (observer frame), not cor- rectedfortelluricabsorption. 3.5.Otherwavelengthregimes M31 has been observed at radio wavelengths both as part of larger surveys and as the focus of dedicated programmes (see e.g.,Gelfandetal.2004;theirTable1).Innocase,aradiocoun- terpart was detected at the position of Sharov21. With a flux density limit of 0.15 mJy, the VLA survey by Braun (1990) is Fig.6. SED of Sharov21 in the restframe(open symbolswith thedeepestoneat20cmcontinuum(1.465GHz).Atlowerradio downwardarrows:upperlimits).ForcomparisonthemeanSED frequencies a deep survey of the M31 field was performed by (Elvisetal.1994)isshownforradio-quiet(solid)andradio-loud Gelfand et al. (2004, 2005) with the VLA yielding a flux den- quasars(dotted),normalizedatλ1415Å. sity limit of3mJy at325MHz.Within d ∼ 2′ fromSharov21, the 325MHz catalogue lists the sources GLG043, 045, 050, and 051. However, with d > 1′ all four radio sources are ob- 5 seehttp://www.xray.mpe.mpg.de/cgi-bin/rosat/seq-browser H.Meusingeretal.:AspectacularUVquasarflare 7 WealsocheckedimagesfromtheDeepImagingSurveywith holemass wouldrequireanunusuallyhighEddingtonratio for the Galaxy Evolution Explorer (GALEX) in the near ultravio- Sharov21. let (λ ∼ 2270 Å) and in the far ultraviolet (λ ∼ 1520 Å), eff eff butno counterpartcouldbe identifiedwithin a radiusof ∼ 5′′. For detections in the near infrared we searched in the 2MASS catalogue(Skrutskieetal.2006),againwithoutaclear-cutiden- 4. Theoutburst tificationofacounterpart.Fromsimplestatisticsofthesources aroundthepositionofSharov21afluxlimitof1µJyisestimated 4.1.Sharov21-likeUVflaresarerare fortheultravioletbandsandof0.25mJyintheKband.InFig.6 wecomparetheextinction-correctedbroadbandspectralenergy UV Flux variations by one or a few magnitudes are observed distribution(SED)ofSharov21withthemeanSEDfornormal, bothinlow-luminosityAGNslikeNGC5548(Ulrichetal.1997, nonblazarquasarsfromElvisetal.(1994). andreferencestherein)orinblazars(Sect.4.4).Contrarytothe burstof Sharov21,the B bandflux of highredshiftradio-quiet quasarstypicallyvariesbyafewtenthsofamagnitudeasisob- 3.6.BlackholemassandEddingtonratio servedinthegroundstateofSharov21.Thesecondremarkable differenceisthefactthatthestrongactivityofourquasarislim- The black hole mass is estimated from the Civ line width as a itedtoashorttimeintervalof∼1year(observerframe).During measureofproxyforthevelocitydispersionofthebroademis- theremaining∼47yrcoveredbythelightcurve(i.e.,theground sionline gasin combinationwith a radius-luminosity(R-L)re- state),themeanfluxvariationisafactorof∼16smallerthanthe lationshipfortheemissionregion.Significantprogresshasbeen maximumfluctuationin the outburst.In contrast,bothstrongly achievedoverthelastyearswiththecalibrationoftheR-Lrela- variablelow-luminosityAGNslikeNGC5548andopticallyvi- tion(Vestergaard2002,2009;Corbettetal.2003;Warneretal. olently variable (OVV) quasars show a more or less steady up 2003;Petersonetal.2004;Vestergaard&Peterson2006). anddownvariation. We use equation (8) from Vestergaard & Peterson (2006) which is based on the line dispersion σ (Civ) and line the monochromatic continuum luminosity λL at 1350Å. λ Following the method outlined by Peterson et al. (2004) and Vestergaard & Peterson (2006) we obtain σ (Civ) = (2.9± line 0.5)103kms−1.Theluminosityofthecontinuumatλ=1350Å (restframe) is derived from the extinction-corrected mean B band flux in the ground state after correction for the contribu- tion from the emission lines (∼ 12%) andassuming a standard power-law continuum F ∝ λ−(2+α) with α = −0.5. We find λ λL (1350Å) = 1.071046 erg s−1 which finally yields a black λ hole mass of M ∼ 5108M . Vestergaard & Peterson (2006) bh ⊙ give a standard deviation of ±0.33 dex for the scaling relation. Allowingfurtherfortheuncertaintiesofthelinewidthandofthe continuumluminosity,thetotaluncertaintyof M isroughlya bh factorof3. Adopting the bolometric correction k (1350Å) = 4 from bc Richards et al. (2006; their Fig.12), the monochromatic lumi- nosity from above corresponds to L = λL ·k = 4.31046 bol λ bc ergs−1 ∼ 1013L and to an Eddington ratio ǫ ≡ L /L = ⊙ bol edd 3.3108(M /M )−1 ∼0.6forthegroundstate.Theoverwhelm- bh ⊙ ingmajorityofSDSSquasarswithz∼2radiateatbetween10% and100%oftheirEddingtonluminosity(seeVestergaard2009) with a narrow ǫ interval at high luminosities (Kollmeier et al. 2006; Shen et al. 2008; Onken & Kollmeier 2008). According to Onken & Kollmeier, quasars with log(L /ergs−1) > 46.5 bol andz = 1.1...2.2arecharacterizedbyhlog ǫi = −0.65±0.35 whereas ǫ is smaller and shows a broaderdistribution for low- luminosityAGNs.Gavignaudetal.(2008)findlog ǫ =−0.97+ 0.28[log(L /ergs−1) − 45] over the interval log(L /L ) = bol bol ⊙ 45...48.5.AscanbeseenfromtheirFig.4,avalueofǫ ∼ 0.6 isrelativelyhighbutnotunusualforquasarsoftheredshiftand luminosityofSharov21.However,inthe maximumofthe out- burst, where the B bandflux is a factor of ∼ 20 higher,the lu- minosityofSharov21correspondstoahighlysuper-Eddington regime. Fig.7. Comparison of the variability of Sharov21 with the It has been questioned for a couple of reasons whether the VPMS quasars. Top: maximum fluctuation ∆Bmax vs. standard massestimatesfromscalingrelationsover-predicttheblackhole deviationσ(framedasterisk:Sharov21;σreferstotheground massesbyaboutanorderofmagnitude.Forargumentsinfavour state). Bottom: single object structure functions for Sharov21 ofthemassestimateswerefertothediscussionbyVestergaard (solid line at the top) and the VPMS quasars (thin solid lines), (2004). In addition we note that a significantly lower black andVPMSsample-averagedstructurefunction(dottedline). 8 H.Meusingeretal.:AspectacularUVquasarflare TogetanideahowusualorunusualtheoutburstofSharov21 ing maximum amplitudes > 2 mag up to 3.7 mag; all of them is, we first consider the light curves of the quasars from the areblazars.Nearlyallofthe14800spectroscopicallyidentified VariabilityandProperMotionSurvey(VPMS;Meusingeretal. quasarsinthedatabasehavejumps< 1mag;thehighestvalue 2002,2003).Thetimebaselineoftheobservationsisnearlythe is1.8mag. same(from∼1960to2008)asforSharov21,thenumbersofob- servationsper quasar are howevermuch lower, typically ∼ 50. TheVPMSquasarsampleis highlycompletedownto B ∼ 20. 4.2.Amicrolensingevent? Among the 321 AGNs in the VPMS there are 10 AGNs with AstheflareofSharov21appearstobeasingularfeatureinthe maximum fluctuations ∆B = B − B ≥ 2 mag. For max max min longlightcurve,itistemptingtospeculatethatitoriginatesfrom the majority of them (80%), the variability is characterized by arareevent.Herewefirstdiscussmicrolensing. amoreorlessmonotonicvariationoverthetimebaseline.Only Chang & Refsdal (1979) first suggested that the flux of a for two AGNs the maximum fluctuations can be attributed to (macrolensed)quasarcanbeaffectedbyastarcrossingcloseto burst-likefeatures.One is the stronglyvariableobjectCCBoo, the line of sight with time scales of the order of a few months a Seyfert galaxy at z = 0.17 (Margon & Deutsch 1997), the toseveralyears.Ontheobservationalside,quasarmicrolensing other one is a radio-quiet quasar at z = 1.08. However, with wasfirst identifiedby Irwinetal.(1989). Sincethen,consider- ∆B = 2.46magand2.50mag,respectively,the burstampli- max able progress has been made in microlensing simulations, and tudesareconsiderablysmallerthanforSharov21andthereare observationsofsignificantmicrolensinghavebeenreportedina strongfluctuationsalsoinotherpartsofthelightcurves. numberofsystems(e.g.,Peltetal1998;Koopmansetal.2000; InFig.7wecomparevariabilitypropertiesofSharov21with Chaeetal.2001;Wisotzkietal.2003,2004;Chartasetal.2004; thoseoftheVPMSquasars.Thetoppanelshowsthemaximum Eigenbrodetal.2006;Paraficz2006;Sluseetal.2007). fluctuationamplitudeversusstandarddeviationoftheBmagni- tudesinthegroundstateforSharov21,comparedwith98VPMS quasars of comparableredshifts (z = 2.1±0.5). As we cannot reasonablydistinguishbetweenagroundstateandahigherstate forthemajorityofthequasarswesimplyusethestandarddevi- ationofalldatainthelightcurveasaproxy.(Anaturalconse- quenceistheincreaseofσwiththemaximumamplitude.)The lightcurvesofthetwostronglyvariablyVPMSquasarsmarked byopensquaresinFig.7areclearlydominatedbysmoothlong- termvariationsoverdecades.Apopularstatisticaltoolforthein- vestigationofquasarvariabilityisthefirstorderstructurefunc- tion S2(τ) = h[m(t +τ)−m(t)]2i (e.g., Simonetti et al. 1985; t Kawaguchietal.1998)whereτisthetime-lagbetweentwoob- servationsin the quasar restframe and the angular bracketsde- note the time-average. The structure function represents a sort ofrunningvariance(asafunctionofthetime-lag)andcontains therewithinformationaboutthetimescalesoftheinvolvedvari- abilityprocesses.ThemostimportantconclusionfromFig.7is thattheflareofSharov21issingularandwithoutcomparisonin thelong-termvariabilitydataoftheVPMSquasarsample. Excellent data for the statistical study of quasar variabil- ity has been provided from stripe 82 of the Sloan Digital Sky Survey (SDSS) for ∼ 104 quasars in five colour bands over ∼ 7 years(e.g.,Sesar etal. 2007). Using the SDSS quasarcat- alogue (Schneider et al. 2007) we identified 8311 quasars in the LightandMotionCurveCatalogue(LMCC;Bramich etal. 2008) from ∼ 249 square degrees of the SDSS stripe 82. No Fig.8. The crowded field towards Sharov21 on the 10′′× high-redshift quasars with z > 2 were found with amplitudes 10′′cutout from the 4-m KPNO image in the I band (N up, E in the u and g bands > 1.5 mag. Allowing for the whole red- left). shift range, the two SDSS quasars with the highest amplitudes in the g band are SDSSJ001130.0+005751.8 (z = 1.49) and SDSSJ211817.37+001316.8(z = 0.46) with ∆g = 3.2 and Sharov21 is seen throughM31, the high star density close max 2.7mag,respectively.Botharebrightpolarizedflat-spectrumra- to the line of sight is illustrated by Fig.8. The quasar is the diosources(Jacksonetal.2007;Sowards-Emmerdetal.2005) brightest, slightly elongated object in the centre, all other ob- and their variability is hence characterized by blazar activity jectsaremostlikelystarsinM31.Notealsothatthequasarap- (Sect.4.4). Interestingly, both show (1) a trend of reddening pearsslightlyelongatedwhichpointstowardsanobjectatadis- whentheybecomebrighterand(2)atrendofincreasingintrinsic tance<∼ 0′.′5.Unfortunately,therearenoarchivalHubbleSpace variability(fordefinitionsee Sesar et al. 2007) with increasing Telescopeobservationsofthefield.Extensiveimagingwasper- wavelength.Sucha behaviouris oppositetotypicalradio-quiet formed with the WFPC2 in 2008 to create an accurate map of quasarsandalsotoSharov21(Sect.3.1). M31microlensing(PI:A.Crotts)whereSharov21is,however, ResultsfromthePalomar-QUESTSurveywererecentlypre- severalarcsecondsoutofthefield.Adeepimageofthefieldwas sented by Bauer et al. (2009). 3113 objects were identified in takenwithSubprime-Camatthe8mSubarutelescopein2004, 7200squaredegreeswithfluctuationamplitudes> 0.4magon butthe quasar lies exactlyin the gapbetweenthe two adjacent time scales up to ∼ 3.5 yr. There are only a few objects show- fields6and7. H.Meusingeretal.:AspectacularUVquasarflare 9 FromtheM31massmodel(Geehanetal.2006)weestimate astellarcolumndensityofΣ ∼170M pc−2towardsthequasar. ∗ ⊙ Thisvalueis∼ 2.510−4smallerthanthecriticalsurfacedensity Σ = Dc2/(4πG)with D = D /(D D )andD = D −D , crit S L LS LS S L whereD andD aretheangulardistancesofthesourceandthe S L lense, respectively,G is the gravitational constant. The optical depth,i.e.theprobabilityforthequasartofallintotheEinstein radiusofastarinM31,isτ=Σ /Σ ∼2.510−4. ∗ crit In the case of quasar microlensingby stars in a foreground galaxywithhighopticaldepththelensesdonotactindividually andthelightcurveiscomplex.Forlowopticaldepth(τ <∼ 0.5), however,microlensingcanbestudiedinthesingle-starapproxi- mation(Paczyn´ski1986),whichisappliedhere.Moreprecisely, the assumptionsare madethat boththe lens andthe sourceare point-likeandthattherelativemotionofthelensislinear.Then thelightcurveisgivenbythemagnification F (t) u(t)2+2 µ(t)≡ ν,obs = , (1) F¯ν,gs u(t)pu(t)2+4 where u(t) is the angular distance between source and lens in unitsoftheEinsteinangleΘ ,andF (t)andF¯ aretheob- E ν,obs ν,gs servedmonochromaticfluxdensity attime t and the meanflux density in the ground state, respectively, in the B band. As D S and D are given, the light curve depends only on the mass L M , the relative transverse velocity v of the lens, and the im- L t pact parameter, i.e., the minimum distance u between lens min and source. The light curve of a high magnification event can be significantly modified by the finite size of the source. This, however,occurs for u ∼ R /R , with R = Θ D , whereas min ∗ E E E L wehaveu ∼103R /R .Furthermore,forascaleof8.4kpc/′′ min ∗ E attheredshiftofSharov21andassumingthatthesourcesizeof theUVradiationofthequasaris<∼1013m,thesourcehasanan- Fig.9.Lightcurveµ(t)ofSharov21arounditsmaximum(noer- gular diameter abouttwo ordersof magnitudesmaller than the rorbarsforlucidity,opensymbols:uncertaindata).Overplotted minimumimpactparameterandcanbeconsideredaspoint-like. arethefittingcurves(solid)formicrolensingbyonesinglestar The transverse velocity is determined by the motion of the (top)andtwosinglestarsofthemassesM andM (bottom). L,1 L,2 lens in M31, the proper motion of M31 with respect to the barycentre of the Local Group (LG), the motion of the Sun aroundtheGalacticcentre,themotionoftheGalaxyaroundthe Local Group barycentre, and the motion of the LG relative to via the multiplicative magnification approximation (Vietri & the cosmic microwave background (CMB). Since the first two Ostriker 1983) µ = µ1 · µ2. A good fit to the shoulder in the effects are poorly constrained, we consider here for simplicity light curve between 400 and 50 days before the maximum is only the velocity of the LG with respect to the CMB. With achievedfor(M∗,1,M∗,2) = (0.3M⊙,0.1M⊙)ifvt = 300kms−1 vLG−CMB = 612 kms−1 towards (l,b) = (270◦,29◦) (Loeb & and (1.2M⊙,0.4M⊙) if vt = 600 kms−1, respectively, with Narayan2008)wehavevt ∼300kms−1. (umin,1,umin,2 = 0.055,0.8)inbothcases(Fig.9).Theprojected Forthesimplifyingassumptionthatalllenseshavethesame linear separation of the two lenses is ∼ 35 AU. Both the mass massandvelocity,theEinsteinradiuscrossingtimet =Θ /v is ratioandtheseparationarenotatypicalofabinarystar. E E t constantandtheeventrateforquasarmicrolensingisestimated AlthoughmicrolensingbystarsinM31appearstobeaplau- byΓ∼2N τ/(πt )(e.g.,Mao2008),whereN isthenumberof sibleexplanationfortheflareofSharov21,thereareseriousob- q E q quasars.Given∼20quasarswithB< 20persquaredegreeand jections. First, the probabilityfor magnificationas strong as in asurfaceareaof∼ 2squaredegreeswithinthe25magisophote Sharov21 is very low. With regard to his model with τ = 0.1, of M31, we have N ∼ 40 and Γ ∼ 1 per century. Hence, the Paczyn´ski(1986)pointsoutthat“appreciableincreasesininten- q discovery of a microlensed, faint quasar behind M31 over an sityarethereoncepermillennium...thereisnotmuchhopein intervalofhalfacenturyisnotunlikely. detectingintensitychangesduetomicrolensingatsuchlowop- Whenthetranversevelocityandthedistancesarefixed,the ticaldepth”.Second,colourindexvariationsareexpectedinthe maximumamplificationis determinedbythe impactparameter case of quasar microlensing (Wambsganss & Paczyn´ski 1991; u and the time scale t is determined by M . For v = 300 Yonehara et al. 2008) if the source appears extended. For the min E ∗ t kms−1thelightcurveisbestfittedwithM ∼0.3M andu ∼ point source-point lens constellation, however, an achromatic ∗ ⊙ min 0.048 (Fig.9, top). A higher velocity requires a higher stellar lightcurveisexpected.Here,thesourcecanbeconsideredpoint- mass. The fit is not perfect because the observed light curve like,butB−Rwasconsiderablysmallerintheflarecomparedto shows a weak but clearly indicated asymmetry (Fig.2) which thefaintstate(Fig.3).Third,astheangularseparationofthetwo hints on deviations from the single-lens hypothesis. Therefore lensesis∼Θ ,themagnificationpatternisexpectedtobemore E,1 weaddasecondlenswhichmoveswiththesamev butcrosses complex then for the single-star approximation.Detailed mod- t the line of sight ∼ 180 days earlier. For simplicity, each com- ellingwithamoreconsistenttreatmentofthebinarylensprob- ponent is treated as a single star, the light curve is computed lem is clearly necessary but is beyondthe scope of the present 10 H.Meusingeretal.:AspectacularUVquasarflare paper. It will be particularly interesting to see if such models a by the orbital period P we have E = −1(2πGM /P)2/3 for 2 bh finda naturalinterpretationforthe steep rise of the fluxbefore M ≫ M . After one orbit the bound elements will return to bh ∗ thepeak. R andefficientlyloosetheirenergyandangularmomentumvia p stream-streamcollisionandfinallybeaccretedtotheblackhole. The consequence is a luminous flare at t with a peak in the 0 4.3.Astellartidaldisruptionevent? UV/X-rays and a spectrum which is characterized by a black- An alternative process which is rare and produces a strong bodytemperatureof a thick disk or a sphericalenvelopeat the UV/X-ray flare is the disruption of a star by the strong tidal tidalradiusRtfromEq.(2): forces of a massive black hole (Lidskii & Ozernoi 1979; Rees 1/4 −1/6 1fa9r8m8,a1in9l9y0)in. Tthidealcodnistreuxpttoiofndeovremnatsnt(TbDlaEcks)hwoleerseidnisncounss-eAdGsNo Teff =4πσLR2∗ MMb∗h . (3) galaxiesor in low-luminosityAGNs (e.g., Phinney 1989; Rees 1990)and,morerecently,ofrecoilingblackholes(Komossa& This scenario implicates two important conclusions for the Meritt 2008). At least a fraction of low-luminosity AGNs ap- lightcurve.First, iftheenergydistributionisuniformthemass pearstobepoweredbystellartidaldisruptions(Komossaetal. returnrateisdeterminedbytherelationbetweenenergyandpe- 2004;Milosavljevic´ etal.2006).Forhigh-luminosityAGNsthe riodoftheorbitdM/dP ∝ dE/dP ∝ P−5/3. Assumingthe time situationismorecomplicated.TDEsareexpectedtoberarefor scale of the transformation of the orbital energy into radiation black hole masses higher than a critical mass M > M ≈ isshort,i.e.theluminosityoftheflarefollowstheaccretionrate bh crit 108M where the gravitation (Schwarzschild) radius, R , ex- andthelatterisgivenbythemassdistributionofthereturntimes, ⊙ S ceedsthetidaldisruptionradiusofsolarmassstarssothatsuch thefluxdensityintheflareshouldbe stars are swallowed whole without disruption (Hills 1975; see also Chen et al. 2008; Gezari et al. 2008). On the other hand, F ∝∆t−5/3, (4) it seems possible thata massive, self-gravitatingaccretiondisk where∆t = t−t isthetimesincethefirstpassageoftheperi- brings more stars into loss-cone orbits and enhances therefore d centre. Numerical simulations have shown that this ‘standard’ thetidaldisruptionrate(Syeretal.1991;Donleyetal.2002). ∆t−5/3 light curve is a good approximation, at least for later Until now, about a dozen TDE candidates in non-AGN stages (Evans & Kochanek 1989). Close to the peak luminos- galaxies have been found from X-ray surveys(Komossa 2002; ity the light curvecan be substantially shallower (Lodatoet al. Komossaetal.2009;Esquej2007,2008;Cappellutietal.2009) 2008). and also in the UV/Optical (Renzini et al. 1995; Gezari et al. As another consequence, the time ∆t = t − t the most- 0 0 d 2006, 2008). None of the events detected so far were found to tightlyboundmaterialneedstofallbacktoR isdirectlyrelated p be related to a high-luminosity AGN. Such a detection would tothemassoftheblackhole be interesting because it provides an opportunity to check the HbpaeosnwidceovtniedraM,labsdhinwsorhuteipcdthiobcnyanGthbeezeoarerysiteaimtsaatlht.ee(di2ri0np0dr8ee)dp,ietcnhtdieoeennxstilsygteeinnnectrehailoslfycvadaseer--. 10∆−t40yr ∼β−3MMb∗2hkR33∗1/2, β≡ RRpt (5) ious mechanisms for the UV variability of quasars makes the withM ,M ,R insolarunits;kdependsonthespin-upstateof bh ∗ ∗ interpretationofaUVflaresubjecttocarefulanalysis.Inpartic- thestarwithk ∼ 3forthelikelycasethatthestarisspunupto ularitisnecessarytomakesurethattheobservedflareisnotjust nearbreak-upspinandk ∼1ifspin-upisnegligible. amoreorlessusualfeatureinastronglyvariablelightcurve. Inwhatfollows,we shallcheckwhetherthetotalenergyin ThetheoreticalframefortheinterpretationofTDEshasbeen theoutburstoftheSharov21lightcurve,thedeclineoftheout- setwiththepioneeringworkbyHills(1975),Lacyetal.(1982), burst,andthetimescalesareconsistentwiththeTDEscenario. Rees(1988),Phinneyetal. (1989),Evans& Kochanek(1989), Theenergyreleaseintheflareisrelatedtothemassofthestar, and more recently by, among others, Magorrian & Tremaine (1999), Ulmer (1999), Ayal et al. (2000), Ivanov & Novikov E =ηf M c2, (6) flare acc ∗ (2001),Menou&Quataert(2001),Lietal.(2002),Bogdanovic´ where η is the efficiencyof convertingmass to radiatedenergy etal.(2004),Wang&Merritt(2004),Chenetal.(2008),Lodato and f isthefractionofmassofthe staraccretedtothe black etal.(2008).Westartwithashortsummaryofthebasictheory. acc hole. For simplicity we assume ηf = 0.1, which might be AstarofmassM andradiusR ,originallyinhydrostaticequi- acc ⋆ ⋆ quitehigh(seeLietal.2002)butisnotimplausible,namelyfor libriumandonaparabolicorbit,passingatthetimet withinthe d f ∼ 0.5followingRees(1988)andη ∼ 0.1...0.4depending tidaldisruptionradius acc onthespinoftheblackhole.E isobtainedbyintegratingthe flare R =R (M /M )1/3 (2) bolometric luminosity over the flare where Lbol(t) is computed t ∗ bh ∗ fromthemonochromaticfluxdensity,F (t)= F (t)−F¯ , ν,flare ν,obs ν,gs of the black hole will be torn apart by the tidal forces. The with Fν,obs(t)and F¯ν,gs intherestframe.Ablackbodyspectrum change in the black hole potential over the star will produce a is assumed with Teff as a free parameter. The lowest possible spreadinthespecificenergyofthegas.Aftertheencounter,the stellarmassis M∗ ∼ 6M⊙ (Eflare ∼ 21054 erg)correspondingto gasparticleswill havea considerablywidenedenergydistribu- Teff ∼ 2104 K where the black body spectrum peaks in the B tion.FollowingRees(1988),abouthalfofthemasshasnegative band,i.e.,atλ = 4400Å/(1+z)restframe.Forsuchaspectrum energy and is thus gravitationally bound to the black hole, the a colour index B− R ∼ 0.5 mag is expected after foreground otherhalfisunboundandwillbeejectedfromthesystem.The reddening,whichisinlinewithB−RobservedforSharov21in boundelementsmoveinhighlyexcentricellipticalorbitsofthe theflare(Fig.3). size2a = R +R ∼ R whereR andR arethepericentricand WithT ,L ,andM fixed,thestellarradiusisconstrained p a a p a eff bol ∗ theapocentricdistance,respectively.WiththeKeplerianrelation byEq.(3)foragivenblackholemass M .InFig.10(top),the bh betweentheenergyandthesizeoftheorbitandafterreplacing resulting (M ,R ) combinations are compared with the stellar ∗ ∗