19 9 6MNRAS.283. .111S 2 + 1 1234 Accepted 1996July1.ReceivedJune25;inoriginalformMarch4 ^Astrophysics, OxfordUniversity,RebleRoad,0X13RH ^Institute ofAstronomy,MadingleyRoad,CambridgeCB3OHA Lick Observatory,UniversityofCalifornia,SantaCruz,CA95064,USA Mon. Not.R.Astron.Soc.283,777-797(1996) Forbidden Feemissionfromactivegalaxies The term‘activegalaxy’isoftenusedtodescribeanygalaxy SpaceTelescopeScienceInstitute,3700SanMartinDrive,Baltimore,MD21218,USA being duetoaburstofmassivestarformation.However,the that possessesemissionlines.Oneofthefundamentalprob- Chris Simpson,*DuncanA.Forbes,tAmandaC.BakerandMartinJ.Ward$ cesses (e.g.,photoionizationfromasupermassiveblackhole lems inactivegalaxyresearchisdeterminingthedominant lines attributedtopurelystellarprocesses,theiractivity , shocksfromsupernovaremnants)ornon-stellarpro- due tostellarprocesses(e.g.,photoionizationfromOB excitation mechanismoftheline-emittinggas,whetherit is Objects classifiedasstarburstgalaxieshavetheiremission ^Present address:X-rayAstronomy Group,DepartmentofPhysics sity ofBirmingham,Edgbaston, Birmingham B152TT. f Presentaddress:SchoolofPhysics andSpaceResearch,Univer- Drive, Pasadena,CA9409,USA. and surroundingaccretiondisc,shocksfromaradiojet). 1 INTRODUCTION and Astronomy, University ofLeicester,Leicester LEI7RH. * Presentaddress:JetPropulsion Laboratory,4800OakGrove © 1996RAS © Royal Astronomical Society • Provided by the NASA Astrophysics Data System + -1 ABSTRACT intheirsamplealsoappearedtohaveshock-excited[Fen]emission, forbidden Feemissioninactivegalaxies,whichisstillunderdebate.Inastudyof We presentintermediate-resolution(~400kms)near-infraredspectraforeight photoionization ofmaterialwithnormalISMabundancesmustbethedominant combine thesedatawiththoseintheliteraturetoinvestigateoriginof Key words:galaxies:active-nucleiSeyfertstarburst cent) ofthe[Fen]emissionmaybeshock-excited. grain destructionthatisaninevitableconsequenceofshocks.Togetherwiththefact However, wearguethattheobservedlineratiosareinconsistentwithsubstantial tightly correlatedwithradiopowerthanthehydrogenrecombinationemission. objects providessomesupportforthefindingbyForbes&Wardthat[Fen]ismore starburst galaxiesisshock-excitedbysupernovaremnants.Interestingly,theSeyfert 26 activegalaxies,Forbes&Wardprovidedevidencethat[Fen]emissionin Seyfert andthreestarburstgalaxies,coveringthe[Fen]1.257-|imPa/?lines.We excitation mechanism,althoughwesuggestthatasignificantcontribution(~20per that thedispersionin[Fen]-radiocorrelationisstillfairlylarge,wededuce although theradiojetswerebelievedtoberesponsible.Analysisofoursample46 - :galaxiesradiocontinuum:galaxies. wood &Oliva1988,1990;Goldader etal.1995),withthe windows (e.g.Kawara,Nishida &Taniguchi1988;Moor- galaxies classifiedaslow-ionizationnuclearemission by Goodrich,Veilleux &Hill(1994,hereafter GVH94)and bands. SpectraintheJwindow (1.25pm)havebeentaken processes atwork.Muchofthespectralworktodatehas stellar andnon-stellarprocessesmayco-exist,although it regions (LINERs)(Heckman1980)orSeyfertsinwhich primary linesofinterestbeing [Fen]1.644pm,Hu=l-0 concentrated ontheH(1.65-pm) andK(2.2-pm)spectral lower thanintheoptical.Inaddition,thereareavarietyof near-infrared wavelengths,wheretheextinctionismuch and sothereareobviousbenefitsinstudyingsuchgalaxiesat et al.1992). can beexplainedbypurelystellarprocesses(e.g.Terlevich has beenarguedthateventhepropertiesofSeyfertgalaxies situation becomesmorecomplicatedwhenweconsider S(l) 2.121pm,Bry2.166pm and theCOstellarabsorption spectral linesthatcanbeusedasdiagnosticsofthephysical 2 The centralregionsofactivegalaxiescanbequitedusty, 19 9 6MNRAS.283. .111S -1 with supernovaremnants(SNRs),fromspectroscopic burst andSeyfertgalaxies,althoughtheirsamplecontained bes &Ward(1993,hereafterFW93)foundthatthestrength photoionization fromthenuclearsource.Interestingly,For- galaxies, asmightexcitationfromtheradiojetsand/or lines simultaneously. times strongerthanthe1.644-pmline)andPaß1.282-pm of coveringthe[Fen]1.257-pm(whichisintrinsically1.35 This rangeisreasonablyconsistentwiththestarburst potentially usefuldiagnostic.Instarburstgalaxiestheratio mechanism fortheproductionof[Fen]emissioninSeyfert only afewpureSeyferts.Theysuggestedthatthemostlikely of the[Fen]lineemissionincentral(~kpc)region line mapping(Lesteretal.1990)andbydirectimaging studies (Moorwood&Oliva1988;Greenhouseetal.1991), in starburstgalaxiesisproducedbyfastshocksassociated Puxley etal.(1994).Thisspectralregionhastheadvantage 778 CSimpsonetal. models byColina(1993),whichpredictratios0.1-1.4. galaxies isshock-excitationfromtheradiojetsand/or sion, andthatthesamecorrelationisobeyedforbothstar- active galaxiesistightlycorrelatedwiththe6-cmradioemis- between [Fen],Paßandradioemissionwithinthevarious ferent rangeofratiosmightbeexpectedforSeyfertgalaxies unaffected byextinction,unliketheratio[Fen]1.644pm/ the ratio[Fen]1.257pm/Paß,sincetheirwavelengthsare ionization fromhotstars.Abetterdiagnostic,however,is the hydrogenrecombinationemissionisproducedbyphoto- Moorwood &Danziger1989).Thisimpliesthatthe[Fen] Higher ratiosareobtainedforindividualSNRsinour SNRs. Bry. Assumingintrinsicratiosof[Fen]1.644pm/1.257 close enoughtogetherthattheirratioisessentially emission ispredominantlyduetotheSNRshocks,whereas andintheLargeMagellanicCloud(LMC)(Oliva, (Forbes etal.1993).ThisprocessmayalsooccurinSeyfert photoionization andshock-excitationscenariostothedata groups ofobjects.Wethencomparethepredictions meaningful statisticalanalysis.Afterdiscussingthegalaxies galaxies, whichwecombinewiththedataofFW93and if non-stellarprocessesareatwork. starburst galaxiesis0.1-0.5(fromColina’smodels).Adif- band). Theslitwasgenerally aligned alongtheradioaxisof 2 OBSERVATIONS in oursampleindividually,weexaminetherelations GVH94 toproduceasampleuponwhichwecanperform tion wasmade with theslitrotatedthrough90° (i.e.,perpen- the source,althoughinsome casesanadditionalobserva- pixel, withaspectralresolution of766(/band)or510(H and discusstheimplications. (case B),theexpectedrangeof[Fen]1.257pm/Paßfor [Fe n]1.644pm/Bryisobservedtobeintherange0.5-2.0. seeing wasabout1.0arcsec.Our spatialscalewas1.5arcsec 3.9-m UnitedKingdomInfrared Telescope(UKIRT).A Observations of11galaxiesweremadeonthenightsut pm =0.74(Nussbaumer&Storey1988)andPaß/Bry5.88 1993 August19-21usingtheCGS4spectrometeron 1.5-arcsec slitwasusedforall observations,andthetypical The ratioof[Fen]toahydrogenrecombinationlineis It hasbeenfairlywellestablishedthatthe[Fen]emission In thispaper,wepresentnear-infraredspectraof11 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System were createdandaddedtogether.Inordertoremoveatmo- was amaximum.Exposuretimeswerenormally40min,with between regionsofcontinuumoneithersidethelineand formed simultaneouslywiththeremovalofatmospheric model blackbodyoftheappropriatetemperature.The heads. Thesetworows,onwhichthesourcesignalwas the slittoallowskysubtractionwithoutadditionalover- the sourcebeingnodded18arcsec(12detectorrows)along the fluxmeasuredonapre-determinedrowofdetector Fig. 1. the spectrumofatypicalatmosphericandfluxstandardin observed andusedtocalibratethespectrum.Wepresent UKIRT fluxstandards,socalibrationcouldbeper- resulting spectrawerethendividedintotheobjectspectra. absorption features(e.g.,Paß),anddividedthembya spheric features,weobservednearbystars(generallyof spectra obtainedfromtheserowsasthe‘nuclearspectra’. strongest, willbereferredtoasthe‘nuclearrows’,and dithering thetelescopeanddetermininglocationwhere dicular totheradioaxis).Eachsourcewasacquiredby flux fromapointsourcecouldbecontainedintherows measuring thefluxabovethisinferredcontinuumlevel. Many ofthestarsusedforatmosphericcorrectionarealso spectral typeG8III),interpolatedovertheirintrinsic flux. Incaseswherethesignal-to-noiseratiowasinsuffu- the nuclearspectrum.Forspectrawherelineswereclearly the lineswereweaklydetectedatnucleus,andnot absorption features.Inothercases,afluxstandardwasalso 3.1 NGC253 3 RESULTS the nucleuswasbelowestimateddetectionthreshold. method wasused,thefluxpredictedinrowsadjacentto line-emitting regionisunresolved,andshouldreducethe its fluxinthenuclearrow.Thismethodassumesthat measured fromthestandardstarobservedclosestintimeto cient todetectlinesoff-nucleus,wemultipliedthefluxmea- methods, dependingonthestrengthoflinesdetectedin such errors,wehavedeterminedthelinefluxesbytwo detected atallintheadjacentrows.Inanattempttoavoid systematic errorsintoourfluxmeasurementsincaseswhere tion, itwasdeterminedthatasmuch40percentofthe From theprofileofstandardstarsinspatialdirec- Table 1. errors inourline-fluxmeasurementstobenomorethan object-sky pairsshowslittlevariation,weestimatethe nights, andtherelativephotometrybetweensuccessive errors causedbylightleakage.Inallcaseswherethissecond sured inthenuclearspectrumbyratiooftotalflux summed fromeverydetectorrowtoprovidethetotalline detected inrowsadjacenttothenucleus,linefluxeswere adjacent tothe‘nuclear’rows.Thiswouldintroducelarge NGC253 isanearbystarburst galaxy.Thenear-infrared Since theseeingappearedtoberelativelystableonallthree in severalhotspots. Theradiocontinuum (Ulvestad& continuum (Forbes,Ward&DePoy 1991)andlineemission (Forbes etal.1993)inthecentral regionsareconcentrated 15-20 percent.Wepresenttheextractedlinefluxesin After biassubtractionandflat-fielding,object-skypairs Line fluxesweredeterminedbylinearlyinterpolating © 1996RAS,MNRAS 283,777-797 19 9 6MNRAS.283. .111S 4-3 -1 we alsonotethepresenceoflinesat1.295,1.321and1.329 Antonucci 1991)followsasimilarmorphology,withnumer- predicted byanelectrondensityof«=10cm(Thomp- to the1.257-fimlineareingoodagreementwithvalues tion tothestrong,narrow[Fen]1.257-pmandPaßlines ous embeddedcompactsources(SNRsandHnregions). Figure 1.PartiallyreducedspectrumofBS8551,aKOIIIstarusedforfluxcalibrationandtheremoval0Hatmospheric © 1996RAS,MNRAS 283,777-797 absorption. ThisparticularstarwasusedinthereductionofNGC7682. son 1995;theratiosarerelativelyinsensitivetoelectron (which havesimilarvelocitywidths,FWHM&35Qkms), pm, whichweidentifyasadditional[Fen]lines.Theirratios e 2 We presentourspectrumofNGC253inFig.2.Inaddi- © Royal Astronomical Society • Provided by the NASA Astrophysics Data System 18-2 UW84a =Ulvestad&Wilson(1984a),UW84b(1984b), Table 1Emission-lineandradiofluxes(inunitsof10~WmmJyrespectively) of -0.7). UW89 =Ulvestad&Wilson(1989;extrapolatedfrom20cmassuming aspectralindex Condon (1991),N88=Norris(1988),PR81Pearson&Readhead(1981), CD78 =Condon&Dressel(1978),C82etal.(1982),GC91Gregory radio structure.Radiofluxreferences:AU88=Antonucci&Ulvestad(1988), for thegalaxiesobserved.LinefluxesinparenthesesaretakenfromForbes&Ward (1993). Thepositionangleofthefirst(oronly)observationis Arp 102B Mkn 1066 Mkn 573 NGC 7682 NGC 6240 NGC 7672 NGC 5929 NGC 1275 NGC 1068 NGC 253 1.257pm1.644 Name ActivitySlitPA[Fen]Pa/? SB SB SI S2 SB S2 S2 S2 S2 S2 S2 140° 125° 157° 162° 98° 95° 50° 35° 95° 20° 62° 72° 30° 55° 417 241 33 38 12 13 95 52 62 14 7.2 7.2 1.9 1.8 + 46 Forbidden Feemissionfromactivegalaxies779 the ratioof1.644-pmlinefluxfromMoorwood&Oliva nucleus andhotspot‘A’.Itisthereforenotsurprisingthat few percentofthe1.257-pmline(GVH94). conditions. Wealsonotethatthe[Fen]1.270-pmline temperature), although,likeThompson,wefindthatthe slit, althoughourslitwillhaveencompassedboththe et al.(1993)thatwehavenotincludedallthelinefluxinour (aD-a) maybepresentatitspredictedstrengthofa 1.295-pm lineisweakerthanpredictedbythesephysical < 1.3 451 514 1/2 33 50 41 48 50 9.4 We notefromthe[Fen]1.644-pmlineimagebyForbes 3.8 9.3 4.2 9.5 8.1 (161) (473) (318) 80 7.8 36300 1090 Sqc 207 156 120 26 CD78 80 46 35.5 5.5 UW84a 3.1 UW89 UW84b GC91 Ref. AU88 N88 C82 C82 C82 P90 19 9 6MNRAS.283. .111S 2 2 more. Figure 2./-bandspectrumofNGC253.Three[Fen]linesareclearlydetected,andweakfeaturesmaybepresentatthelocation oftwo Figure 3./-bandspectrumofNGC1068.Notethatthereisstillsome residualseeingrippleinthespectrum,whichisparticularlynoticeable rather insecure,aswellmakingthe1.321-pmlineseembroader andmorecomplexthanitstrueprofile. redward ofPa/?,andthattheH01.329-pmlinehasbeenpoorlyremoved. Theseinaccuraciesmakethedetectionof[Fen]1.295pmappear 3.2 NGC1068 passed alltheemission-linepeaks. 780 C.Simpsonetal. 0.74, sincetheir6x6-arcsecaperturewillhaveencom- radio jetsareseenextending outtoseveralarcseconds NGC 1068isanearby,luminous Seyfert2galaxy.Twin arcsec) is1.32,muchhigherthanthetheoreticalvalueof (1988) toour1.257-pmlineflux(extractedfrom12xL5 nuclear source,ithasaclassical narrow-lineregion(NLR) and acircumnuclear starburstringofradius ~1kpc.Imag- (Wilson &Ulvestad1983). As wellasthenon-thermal 2 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System wavelength (/¿m) wavelength (/¿m) -1 with theNLRandcentralcollimatedpartofnorth- we observeditforonly16min.Its/-bandspectrumispre- sented inFig.3.Apartfrom theweak[Fen]1.295-and ing ofthe[Fen]emissionrevealsittobespatiallycoincident be pureHei1.253 pmandPa/?.Sincethere appearstobe these linesasuncontaminated [Fen]1.257-pmemission uncertainties of+0.0004pm). Weidentifythesecondof lengths of1.2514,1.2568and 1.2806pm(withestimated east radiojet(Blietzetal.1994). (with FWHM&95Qkms), but theothersaretooblueto 1.321-pm lines,threestronglines areobservedatrestwave- Since NGC1068isextremelybrightinthenearinfrared, © 1996RAS,MNRAS 283,777-797 w r" Forbidden Fe + emission from active galaxies 781 r" oo CMoo

ID

Figure 4. Spectra extracted from the central 6 arcsec of NGC 1068 (the slit was aligned at /M =30°). The dotted lines indicate the velocity structure of the proposed line identifications. The wavelengths of the lines increase by 26 Â from SW to NE, corresponding to 600 km s“1. The dotted lines pass through the data at the expected wavelengths of the lines listed, based on the [Fe n] 1.257-pm line.

Figure 5. /-band spectrum of NGC 1275 (3C 84). We note that the apparent difference in profiles of the [Fe n] 1.257-jim and Pa/? lines may be caused by He 11.279 pm contaminating the blue wing of Pa/?. Although the shoulder on the Pa/? line suggests this, the fit to the line profile is not better when two Gaussians are used, rather than one. Higher resolution data are needed. There is clearly a broad base to the Paß line.

He i 1.279-pm emission north-east of the nucleus (see Fig. ionization potential of 379.1 eV, would not be expected to 4), we believe that this is contaminating the nuclear Paß be observed off-nucleus. This notion is supported by the fact line. However, the profile of the 1.2806-pm line is extremely that the wavelengths of the lines off-nucleus have measured well fitted by a single Gaussian (FWHM&1350 km s-1) and wavelengths consistent with pure [Fe n], He i and Paß emis- poorly fitted by two lines, suggesting that a third line, or an sion, although at slightly different velocities from the sys- additional velocity component, is present. The 1.2514-pm temic velocity. line may be contaminated by [S ix] 1.252 pm which, with an Our proposed line identifications and inferred velocity © 1996 RAS, MNRAS 283, 777-797

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System 19 9 6MNRAS.283. .111S region studied,exceptforafeatureobserved1.5arcsec south-west ofthenucleusatawavelength1.289pm.Since structure identifyallthesignificantemissionfeaturesin 782 CSimpsonetal attempting toidentifythisasPaßresultsintheotherlines peaking-up. from thestandardstarinsamedetectorrows,andapproxima- line fluxinNGC1275asafunctionofpositionalongtheradioaxis Figure 6.Plotof[Fen]1.257-pm(solidline)andPaß(dashed tely representstheseeingprofile.Thereismarginalevidencefor asymmetry oftheperpendicularprofileiscausedbyimperfect (top) andperpendiculartoit(bottom).Thedottedlineistheflux [Fe n]beingextendedalongtheradioaxis.Wesuspectthat Figure 7./-band spectrumofNGC5929.Theline profilesappeartobenon-Gaussian, althoughthisisnotavery significantresult. © Royal Astronomical Society • Provided by the NASA Astrophysics Data System -4 -2024 Offset (arcsec) -1 17-2 2 46 -1 17-2 1 3.3 NGC1275 being mismatched,wearemoreinclinedtobelievethatitis which isdominatedbyextranuclearemission.Ourspatial The peculiargiantellipticalNGC1275liesatthecentreof wavelength of1.295pm,whichappearsinthespectrum galaxy interactionmayhavepromotedstarformationand material viaacoolingflow(Fabianetal.1981).Aprevious galaxy emitshardX-raysandisthoughttobeaccreting the Perseuscluster.Itisaparticularlystrongradiosource caused bypoorremovalofatmosphericH0absorption. pared withtheotheremissionlinesexcept[Ni]25199, km s)andnarrow(~500components.Rudyetal. on differentspatialscales(e.g.Pedlaretal.1990).The both directions.Wesuspectthattheapparentasymmetryof measured an[Fen]1.257-pmfluxof9.5x10“Wmin in ourspectrum(Fig.5).Kawara&Taniguchimeasuredan & Taniguchi(1993)andRudyetal.(1993),isalsoseen al. (1987).Strong[Fen]emissionhasbeenfoundbyKawara shock-excited thestrongHemissiondetectedbyFischeret given byRudyetal.(1993).Wesuggestthatthisisthe[Fen] information fromthetwoslitpositionsindicates,however, arcsec aperture,andfoundBrytoconsistofbroad(~2000 (also calledPerseusAor3C84)withstructurethatvaries inferred byKawara&Taniguchi. of anarcsecondeastthenominalposition.Ourspectra imperfect peaking-up,withthetruenucleusbeingafraction the emissionwhenslitwasorientedeast-westisdueto see thatthe[Fen]emissionisatbestmarginallyresolvedin that thisisnottrueforthe[Fen]emission.FromFig.6,we an 8.0-arcsecaperture,andnoteditsnarrowprofilecom- aD-a transition,sinceitisthestrongestofother indicate anFWHMof800kmsfor[Fen]comparedto [Fe n]1.644-pmfluxof5.6x10~Wmina2.52.0- [Fe n]linescoveredbyourspectrum,anditsmeasuredflux 1730 kms"forPaß.Thesearesimilartothelinewidths 2 2 5/2 We alsonoticeaweakfeaturetotheredofPaß,atrest © 1996RAS,MNRAS 283,777-797 19 9 6MNRAS.283. .111S Hv =5-30(3)and6-4Q(l)linesobservedbyElsteon&Maloney (1990)and[Fen]1.599pm. Figure 8./-bandspectrumofNGC6240(southernnucleus;seetext).Notetheslightshouldersonbluewings1.257-pm andFaß lines, whichmaybeweakHeiemission. 3.4 NGC5929 removal oftheatmosphericC0absorption,andabroadfeature around 1.645pm(observedwavelength).Thismaybeacombinationofthe Figure 9.//-bandspectrumofNGC6240.The[Fen]1.644-pmline isprominent.Alsovisiblearetwofeaturesprobablycausedbyimperfect position angle of62°,withahotspotseparation of1.25 form Arp90).Itisatripleradio source,extendedalonga close neighbourNGC5930(Lewis &Bowen1993;thepah- NGC 5929isaSeyfert2galaxy whichisinteractingwithits (Thompson 1995).Weagainnotethatweak[Fen] © 1996RAS,MNRAS 283,777-797 ( ~15percentofthe1.257-pmline)isclosetothatexpected 1.270-pm emissionmaybepresentinourspectrum. 2 2 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System T o + _1 1 Forbidden Feemissionfromactivegalaxies783 between theopticalemission-linegasandradiohotspots The near-infraredlineemission inthisobject(Fig.7)is km s). broader (FWHM=650km s") thanPaß(FWHM=m marginally resolved,consistent withitbeingofsimilar matter ejectedfromtheactive nucleus(Boweretal.1994). Pa/? ratioisfairlyhigh(1.6at the nucleus).The[Fen]lineis suggests thatthestructureofNLRmaybegovernedby extent. Bothlinesareweak(EW^6k), althoughthe[Fen]/ arcsec (Ulvestad&Wilson1984b).Positionalcoincidence 19 9 6MNRAS.283. .111S Figure 10./-bandspectrumofNGC7672.Duetothefaintnessthisobject,signal-to-noiseratioisratherlow.Wedetect the [Fen] 784 CSimpsonetal. Figure 11./-bandspectrumofNGC7682.Notetheverynarrow [Fen]andPaßlines.Thefeatureobservedat1.329pmiscausedby The infrared-luminousgalaxyNGC6240hasrecently 3.5 NGC6240 undergone amergerandrevealstwonucleiseparatedby2.2 imperfect removaloftheatmosphericH0absorption. 1.257-pm line,andnotethattheremaybeaverymarginal(<2ci)excessoffluxatthelocationPaß,althoughitisformallyundetected. of morethan2, andsuggestthatthisisdue toanextended line fluxmadebydifferentauthors havedifferedbyafactor Werf etal.notethatmeasurements ofthe[Fen]1.644-pm nuclei. Slowshocksattheinterface ofthecollidinggalaxies enormous Hemission(vander Werfetal.1993).Vander arcsec. Boththe[Fen](vanderWerfetal.1993)andradio (i.e., betweenthetwonuclei) arethoughttogiverisethe (Bales etal.1990)emission areconcentratedatthetwo 2 2 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System -1 FWHM~ 1000 kms. nuclei, whichwouldbelostinthebackgroundfluctuations the twonucleieasily.Wetherefore present/-and//-band two nuclei)agreeswiththelineimagingbyvanderWerfet include anyemissionperpendiculartothelinejoining of theirimage.Our1.644-pmlineflux(whichdoesnot component, inadditiontothestrongemissionfromtwo strongest lineandcontinuum emission,whichisthe spectra (Figs8and9respectively) oftherowcontaining al., supportingtheirinterpretation. southern nucleus.Both[Fen] andPaßareverybroad,with Our 1.5-arcsecpixelscaledoes notenableustoseparate © 1996RAS,MNRAS 283,777-797 19 9 6MNRAS.283. .111S _1 wavelengths of1.270and1.295pm.Althoughtheyare Figure 12./-bandspectrumofMrk573.He11.253pmisstrong,andtheshoulderonbluewingPaßmaybe11.279pm.[Fe ii]1.271 blue wingsofthe[Fen]1.257-pmandPaßlines.Sincesuch 3.6 NGC7672 line fluxes. that thereisasmallamountofHeicontaminationinour measurement couldalsobeconsistenteitherwithzero However, theuncertaintiesarefairlylarge,andour reddening oíA=\3magtothe[Fen]-emittingregion. NGC 7672isalinear,possiblytriple,radiosourceextending rather noisy,wetentativelyidentifythemasadditional extinction orthe2.4magderivedfromHa/Hßratio pm maybepresent,butitisexpectedtoweakerthanthe1.295-pmline,whichnotobserved. 3.7 NGC7682 We prefertoplaceanupperlimitonthefluxfromthisline, a shoulderisnotobservedonthe1.644-pmline,wesuspect The Seyfert2galaxyNGC7682 appearstobephysically trum hinderssuchasearch. however, andwethereforefindalowerlimittothe[Fen]/ km s),andtheremaybeaverymarginaldetectionofPaß. the faintestgalaxy(atJ)inoursample,anditsspectrum (Morris &Ward1988).Wenoteaslightshoulderonthe gas connectingthetwogalaxies (Durret&Warin1990). emission, althoughthelowsignal-to-noiseratioofspec- Paß ratioof1.38.Wefindnoevidenceforextendedline present ournuclearspectrum ofNGC7682inFig.11.The However, near-infraredimaging byMcLeod&Rieke (shown inFig.10)isthenoisiest.However,[Fen] [Fe n]lines. associated withNGC7679,as thereisastreamofionized 1.257-pm lineisclearlydetected(with © 1996RAS,MNRAS 283,777-797 (1995) showsnosignofdisturbance inthegalaxy.We [Fe n]lineisunresolved atourresolution,and thePaßline v ~6 arcsecalongPA=95°(Ulvestad&Wilson1989).Itis Finally, aswithNGC1275,wenoteweakfeaturesatrest Our 1.644pm/1.257pmratioof0.84impliesaneffective © Royal Astronomical Society • Provided by the NASA Astrophysics Data System + -1 -1 46 -1 1 Forbidden Feemissionfromactivegalaxies785 very lowoff-nucleus. wing ofthePaßline,whichcouldbeHei1.279pm.How- 3.8 Mrk573 The isolatedSeyfert2galaxyMrk573appearstohave undergone arecentmerger,sinceitpossessestwistedopti- is barelyresolved(FTTHM^250kms).Neitherline predicted tobestronger. that Heilinesarethemostlikelyexplanationsfortwo km s.Theratioofthesecomponentsis[Fen]/Paß~3, they areadditionalvelocitycomponentsof[Fen]andPaß, NGC 1068,appearstoshowstrongHe11.253-|imemission tated bytheradiojets(Capettietal.1996).Mrk573,like NGC 5929,themorphologyofitsNLRappearstobedic- tion of2.8arcsec(Ulvestad&Wilson1984a)and,like cal isophotesandaringofHaemissionwhichispresumably extended although,onceagain,thesignal-to-noiseratiois 3.9 Mrk1066 length of1.270pm,consistentwiththe[Fen]aD-a to it.[Fen]ismuchbroader{FWHM=A59kms)than features. NeitherthePaßnor[Fen]emission is suggestive ofshock-excitedgas.However,nosuchvelocity coming fromanoutflowwitharadialvelocityof~1000 ever, analternativeexplanationforthesetwolinesisthat associated withstarformation(Heisler&Vader1994).Itis line, althoughwedonotsee the1.295-pmline,whichis extended, eitherparalleltotheradioaxisorperpendicular component isseenintheopticalspectrumgivenbyDurret about theopticalcentreofgalaxy,withahotspotsepara- a tripleradiosourceatPA=125°,symmetricallyarranged emission islinear, probablytriple,andextended alonga Mrk 1066(seeFig.13)isa Seyfert 2galaxywhoseradio Paß (FWM=250kms“). (1994) and,whilstitcouldbeobscuredbydust,webelieve (Fig. 12).Inaddition,thereisaslightshoulderontheblue 1/2 /2 We againnotethepresenceofafeatureatrestwave- 19 9 6MNRAS.283. .111S -1 1 third ofthePaßemissionarisingoff-nucleus.Wenote position angleof134°(Ulvestad&Wilson1989).Ithasbeen 786 CSimpsonetal feature whichmaybethe1.295-pmline.Inaddition,thereis presence ofthe[Fen]1.321-pmline,andarathernoisy spatially extendedinthisdirection,withapproximatelyone- emission-line gasisextendedalongtheradioaxisto extensively studiedbyBoweretal.(1995),whofindthatthe although thismaybecausedbyimperfectremovalofthe north-west. Wefindthatboththe[Fen]andPaßlinesare sion lineshavekms. Paß lineinthedividingstandard.Bothprominentemis- a weakemissionlinenearrestwavelengthof1.271pm, Figure 13./-bandspectrumofMrk1066.Theemissionlinesarestrongandfairlynarrow(FWHM&A50kms).[Fen]1.321pm isclearly detected, andthe1.271-1.295-pmlinesmayalsobepresent. Figure 14./-bandspectrumofArp 102B.[Fen]1.257pmisstronglydetected,andthenarrowPaß lineismarginallydetected.Theremay Paß line,which is expectedtooccupytheentire spectrum redwardofthiswavelength. also beaweakemissionfeature at thelocationof[Fen]1.272pm.Therisetored1.27pm (observedwavelength)maybethebroad © Royal Astronomical Society • Provided by the NASA Astrophysics Data System -1 -1 weak narrowPaß(theratio[Feii]/Paß^3.4,andisonly with extremelybroad,double-peakedpermittedlines Arp 102BisaSeyfert1galaxywhichformsaninteracting 3.10 Arp102B pair withArp102A.Our/-bandspectrum(Fig.14)shows to bemorethan10timesstrongerthenarrow strong [Fen]1.257-pmemission(FIF7/M^750kms),and here). Arp102Bis,however,abroad-lineradiogalaxy, surpassed byMrk176inthesampleofgalaxiesstudied (FJFZ/ælSOOO kms;Chen,Halpern&Filippenko 1989). AlthoughthebroadPaßcomponentisexpected © 1996RAS,MNRAS 283,777-797 19 9 6MNRAS.283. .111S present. component, itsbreadthmakesitimpossibletodetect inclusion ofsomebroadPa/?emission.Alternatively, is thatthe1.257-pmfluxhasbeenoverestimatedby only 0.6,lessthanthetheoreticalvalueof0.74.Since of thebroadline.Again,[Fen]1.270-pmlinemaybe emission featureobservedat1.278pmmaybethebluepeak assume thatoneofourlinefluxesisinerror.Onepossibility due toreddening(whichwouldincreasethisvalue),wemust observed valuecandeviatefromthetheoreticaloneonly spectrum (Fig.15),althoughthe1.644pm/1.257pmratiois against thenoiseofcontinuum.However,weak Figure 16./-bandspectrumofArp 220.Both[Fen]1.257-pmandPa/?aredetected.Theremaybe aslightshoulderonthebluewingof emitting material. © 1996RAS,MNRAS 283,777-797 [Fe n]line,whichcouldbedueto He11.253-pmemission,orthe[Fen]linecomingfromgaswhich iskinematicallydistinctfromthePa/?- We alsodetectthe[Fen]1.644-pmlineinour//-band © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Figure 15.//-bandspectrumofArp102B.Onlythe[Fen]1.644-pmlineisdetected. + Forbidden Feemissionfromactivegalaxies787 we maynothavepeakeduponthesamelocationafter whose bolometricluminosityisdominatedbyfar-infrared 3.11 Arp220 Arp 220issimilartoNGC6240inthatitamergingsystem present (Armusetal.1995).AlsolikeNGC6240,and observations. unlike theotherstarburstgalaxiesstudiedhere,ithasaratio there isdebateoverwhetheranactivenucleusalso observing anatmosphericstandardbetweenthetwo emission, believedtobepoweredbyastarburst,although [Fe n]/Pa/?>1(Fig.16).Then]lineissomewhat 19 9 6MNRAS.283. .111S 1 1 -182 AyttlX mag,implyingthatonly~8percentofthe/-band within 25percentofours.Goodagreementistobe which allowthemtoinferthereddening[Feii]-emit- broader (FWHM=550kms")thanPa£{FWHM=3W km s"). work byincludingourobjects,aswellthosefromGVH94 paper wheretheextinctionissevereenoughtohaveastrong flux istransmitted.Thistheonlyobjectstudiedinthis ting region,fromthe1.644pm/1257pmratio.Theyderive our east-westslit.Theyalsopresent/-and//-bandspectra, 788 C.Simpsonetal As wementionedearlier,FW93reportedastrongcorrela- plots presentedinthenexttwosections. this pointwhenwediscussthelocationofArp220in expected, sincethereisverylittleemissionperpendicularto 4.1 Literaturedata 4 ANALYSISANDDISCUSSION effect onthenear-infraredlinefluxes.Weshallreturnto for bothstarburstandSeyfertgalaxies,althoughtheir tion betweenthe[Fen]and6-cmfluxesluminosities, by vigorousstarformation. fluxes to1.257-pmassumingalineratioof0.74,and sample ofSeyfertswasrathersmall.Wecanexpandontheir [Fe n]1.257-|imandPaßlines,theirfluxesagreeto Wilson 1993),andNGC253,18086240aredominated now classifiedasaSeyfert1galaxy,followingtheappear- ticularly helpfulinthisanalysis,wherewearecomparing objects listedasLINERs,sincethisdesignationisnotpar- Bry fluxestoPaßassumingcaseBrecombination,i.e., emission fromstarburstandSeyfertgalaxies.NGC1097is Paß/Bry =5.88.Inaddition,wehavereclassifiedtheir (see Table2).WehaveconvertedFW93’s[Fen]1.644-pm ance ofbroadBalmerlines(Storchi-Bergmann,Baldwin& aperture effectsthatmaybepresentinourdata.FW93were Table 2.Lineandradiofluxes(inunitsof10Wmmly Wilson (1984a),UW84b=Ulvestad&(1984b), Note thatonlynarrow-linecomponentsareused,andobjects respectively) forthegalaxiesobservedbyGoodrichetal.(1994). nucci &Goodrich(1995). UW89 =Ulvestad&Wilson(1989),UAG95Ulvestad,Anto- classified asSeyfert1or2onthebasisofabroadPaßcomponent. Radio fluxreferences:E87=Edelson(1987),UW84Ulvestad& Mkn 78 Mkn 3 MCG -05-23-16 Mkn 1388 Mkn 477 Mkn 463E Mkn 266 Mkn 176 NGC 5728 NGC 5506 NGC 5252 NGC 4388 NGC 3081 NGC 2992 Armus etal.(1995)presentimagesofArp220inthe Before weexaminethedata,shallpausetodiscuss 1.257pm Name Activity[Fen] © Royal Astronomical Society • Provided by the NASA Astrophysics Data System SI S2 52 51 S2 S2 S2 S2 S2 S2 S2 S2 52 51 285 164 20.4 77.6 77.0 29.9 12.2 30.8 73.7 31.3 86.2 16.5 19.6 7.8 851 104 155 Paß Sc 6 30.1 95.2 21.8 80.9 54.6 29.2 19.8 19.5 13.7 16.8 3.8 361 100 160 77 25.0 33.4 12 10.4 16.2 6.0 3.5 4.6 5.3 0.9 UW84b UW84b UW84b UW84b UAG95 UW84a UW84a UW84a UW84a UW84a UW89 UW89 Ref. E87 E87 -1 -1 1 we useFW93’sdataincaseswheretheobjectiscommonto well aspossible.SincethelinefluxesreportedbyMoorwood much extendedemission,andwebelieveouraperture more thanonesample.Thefactthatweorientedourslit copy thatweandGVH94haveperformed.Forthisreason, tures, thiswaseasiercomparedwiththelong-slitspectros- careful tomatchtheiremission-lineandradioaperturesas flux of40mlyat13cm(Royetal.1994).Therefore,unless mly inalargebeam(Condon&Dressel1978)andcore match inaperturesize.NGC7682hasafluxat6cmof26 radio-emitting nucleiofArp220,sotheaperturematch radio sources.ThisisthecaseforNGC5929,7672, the totallinefluxprovidesagoodaperturematchforlinear & Oliva(1988)weregenerallytakenthroughsquareaper- unresolved. match tobeagoodone.ThecoreinArp102Bis the radiospectralindexisextremelysteep,therecannotbe extended inNGC1275,sousingthecoreradiofluxisafair here isalsogood.Thelineemissionappearsonlymarginally Mrk 573and1066.Inaddition,ourslitcoveredboth along theradioaxesofourtargetsensuresthatextracting cases, evenwhentheradioemissionisextendedinaneast- arcsec) encompassestheentireradioemissioninmany Wilson &Sramek(1981)revealthatthe3-arcsecslitof the threeremainingsourceswhicharenotalsoin MCG —05-23-16.NGC5252isextendednorth-south,so Mrk 3,176,266,Mrk463E,477and west direction.ThisisthecaseforNGC3081,5728, Hummel, vanGorkom&Kotanyi(1983)andUlvestad, cant east-westextent,althoughtheradiomapspublishedby sample ofFW93,two(NGC4388andMrk78)havesignifi- GVH94’s slitalsoencompassesalltheradioemission.Of galaxies havebeenimagedinthe[Fen]1.644-jimline,itis ments arelowerthanthoseofFW93.Sinceallthreethese reddening tothe[Fen]-emittingregion,allourmeasure- FW93: NGC253,1068and6240.Assumingthatthereisno radio structureofMrk1388intheliterature. emission. Wehavebeenunabletofindanyreferencethe GVH94 hasencompassedatleasttwo-thirdsoftheradio This lossisdifficulttoquantify,andwecannotruleoutan // =50kmsMpc andq=0. Wright (1990)andRixetal.(1990),whichagreewellwith fluxes aregreatlyinexcessofthoseBlanco,Ward& NGC 5506issimilarlyinbothsamples,butGVH94’sline is commontobothFW93andGVH94,thelinefluxes included intheaperturemeasurementscollatedbyFW93. direction perpendiculartoourslit,whichwouldhavebeen apparent thatwehavemissedsignificantemissioninthe additional effectduetoreddening.Inaddition,NGC2992 4.2 [Fen]-radiocorrelation Moorwood &Olivavalue. Moorwood &Oliva(1988).Wethereforechoosetousethe each other,andareconsistentwiththefluxreportedin (both ofthe1.257-jimline)areinexcellentagreement. In Fig.17,wepresentfluxand luminosityplotsofthe[Fen] We usethedistanceofFW93where available;otherwiseweadopt 1.257-(im and6-cmradioemission forourobjects,andthose o () GVH94 usedanorth-southslit,althoughitswidth(3 There arethreeobjectswhichwehaveincommonwith © 1996RAS,MNRAS 283,777-797 19 9 6MNRAS.283. .111S Figure 17.(a)Plotof[Fen]1.257-^imlineversus6-cmradioflux.Circlesrepresentstarburstgalaxies,squaresareSeyferts,andtriangles composite objects.OpensymbolsaredatafromFW93,solidourdata,andcrossestheSeyfertgalaxiesGVH94.The the regressionlines,whichwehavedeterminedusing solid lineistheleast-squaresregressionforentiresample,afterexclusionofNGC1275:log/=0.73+0.59logS,(70.49. brighter’ natureofactivegalacticnuclei.FW93arguedthat method ofIsobe,Feigelson&Nelson(1986),takinginto of FW93andGVH94.Inthesefigures,wehavealsoplotted that powersSeyferts. object isnotaclassicalSeyfertgalaxy,anditsstrongradio account theupperlimitsfor[Fen]emission.Wehave (b) As(a),butforluminosities.ThesolidlineislogL=11.65+0.68,<70.44. galaxies. present studyistodeterminewhethertheirresultholdsfor is tobeexpectedgiventhegeneral'brighterobjectsare emission isprobablynotproducedbythesamemechanism excluded NGC1275(3C84)fromthedataset,sincethis radio emission,orHandemission.Onegoalofthe scatter waslowerthanforthecorrelationbetweenPaßand the [Feii]-radiocorrelationwasaprimaryone,since region ofArp220willcauseittoliewellbelowthecorrela- Although Arp220ismoreheavilyobscuredthantheother fast shocksassociatedwithsupernovaremnants(SNRs). lation obeyedbytheotherstarburstgalaxiesonlybecauseof other starburstgalaxies.Indeed,ifweinverttheargument tion betweenemission-lineandradiostrengthobeyedbythe a largersamplewhichincludesmanymoreSeyfert the sameforArp220asotherstarburstgalaxies,i.e., its largereddening,wederiveanextinctionoM=12±1,in and assumethatArp220liesbelowthe[Fen]-radiocorre- point inSection4.4,butnotethatitdoesnotaffectthe excellent agreementwiththatderivedbyArmusetal. by starburstgalaxiesandcomposite objects(wherethestar- trends discussedbelow. galaxies inoursample,dustobscurationmayhaveasimilar, burst componentissignificant). Therelationshipbetween albeit lesser,effectontheentiresample.Weaddressthis (1995). Thissuggeststhatthe[Fen]emissionmechanism is fast radiativeshocks,caused bythesynchrotron-emitting these twoquantitiesisbelieved tobeaprimaryonein FW93 maybeduetothefactthat theirdatasetisdominated [Fen]6cm starburst galaxies,sincethe[Fe n]emissionisproducedby [Fen]6cm SNRs. Thisissupported bythemuchreduced scatterifwe 2 © 1996RAS,MNRAS 283,777-797 K There isclearlyacorrelationinbothplots,althoughthis We notethatthehighreddeningtoline-emitting The tightnessofthe[Fen]-radiocorrelationobservedby © Royal Astronomical Society • Provided by the NASA Astrophysics Data System log S(mJy) 6cm + Forbidden Feemissionfromactivegalaxies789 Arp 220andNGC1275,alsothecompositeobjects). FW93 (weobtaincj=0.26)oronlytheir10starburstgalaxies examine justthe15starburstandcompositegalaxiesof for thestarburstsand0.69+0.11Seyferts,indicating The resultingslopes(inlogluminosityspace)are0.95+0.07 the extrauncertaintyintroducedbylargeextinction. be expectedthattheslopeforstarburstgalaxiesiscloseto for theSeyfertgalaxies.(Theequationsofregression that theslopeissomewhatsteeperforstarburststhan divided intostarburstandSeyfertgalaxies(excluding for Seyfertgalaxiesisconsistentwith0.7,whichtheexpo- unity, sincethe[Fen]andradioluminosityarebothpropor- lines arelogL„]=3.33+0.95forthestarbursts (90percent(andthereforepromoteshydrogen 40 000-Kblackbodyproducesahydrogenionizationfrac- for power-lawandhot-starphotoionization.Whilea the [Oi]/Haratio.Thisisbecauseremovingdustgrains iron toFe,andsolimitstheproductionof[Fen]lines. recombination lineproduction),italsoionizesmostofthe ratios muchlargerthanareobserved.Secondly,asMouriet 794 C.Simpsonetal the incidentcontinuumbyabsorbingultravioletphotons) makes thePIZsmaller(dustisveryeffectiveathardening suppresses H,hardercontinuabeingmoreeffectiveat ionized byapower-lawcontinuumbothenhancesFeand al. (1993)note,theionizationfractionsarevastlydifferent abundance power-lawphotoionizationproduces[Fen]/Paß phase ironabundancemorethancounterbalancesthis compensated forbythemodestincreaseintotaloxygen and reducestheionizationfractionno/nmorethancanbe depletion increasesthe[Feii]/Paßratio,itactually¿fec/mses effect. abundance (afactorof2).Asimilarthinghappenstothe Fc¥e 0 Several thingsareimmediatelyapparent.First,solar- © Royal Astronomical Society • Provided by the NASA Astrophysics Data System -14_ n Table 5.Ionizationfractionsandlineratiosfordifferentphotoionizationmodels.PL1.4 and PL1.0arepower-lawionizingcontinuawithiS,ocv',v’°,respectively,BB40is a blackbodywithT=40000K. v log nuUModelAbundancesn+/nFe^n+/H[Fen]/Pa/?[Oi]/Ha Fe 4.0 4.0 2.5 2.5 2.5 3.0 3.0 2.5 -3.0 3.0 4.0 3.0 4.0 -2.5 -3.0 -3.0 -2.5 -2.5 -2.5 -3.0 -3.0 -3.0 -2.5 -2.5 PL1.4 PL1.4 BB40 BB40 PL1.4 PL1.4 PL1.0 PL1.4 BB40 PL1.0 BB40 PL1.0 PL1.0 PL1.4 PL1.4 PL1.0 BB40 PL1.0 BB40 PL1.4 BB40 PL1.0 PL1.0 PL1.4 BB40 PL1.0 BB40 PL1.4 PL1.4 BB40 PL1.0 PL1.4 BB40 PL1.0 BB40 PL1.0 Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion Orion 0 0 0 0 0 0 O 0 0 0 0 O 0 0 0 0 0 0 0.91 0.77 0.26 0.94 0.94 0.38 0.92 0.83 0.12 0.90 0.18 0.89 0.75 0.27 0.94 0.92 0.88 0.12 0.86 0.13 0.90 0.90 0.41 0.91 0.95 0.33 0.94 0.81 0.83 0.24 0.92 0.11 0.86 0.93 0.14 0.93 from themixingcurve,implyingthatactiveSeyfert galaxies liebetweenthestarburstsandSeyferts,alsoaway regions andSNRs.Itisseenthatthestarburstgalaxieslieon the contributionfromPIZ(eitherthroughshocksorX- nation emission)butanorderofmagnitudemore[Oi] twice asmuch[Fen](normalizedtothehydrogenrecombi- near-solar value.AlthoughSeyfertgalaxieshavemore[Oi] lines aresosimilarintermsofthephysicalconditionswhich ratios unlikeeitherofthestellarprocesses. nucleus producesacombinationof[Oi]/Haand[Fen]/Pa/? this curve,buttheSeyfertgalaxiesdonot.Thecomposite and thelocusofmixingcurveforacombinationHn H iiregionandthefiveSNRslistedinMourietal.(1990), and GVH94.Tothisplot,wehaveaddeddatafortheOrion Such adiagramwasalsopresentedbyMourietal.(1990) sample of[Fen]/Paßversus[Oi]/Ha(listedinTable3). scenarios, wherethegas-phaseabundanceisincreasedtoa note) actuallypresentsaproblemforshock-excitation are requiredtoproducethem(asMourietal.1993also and [Fen]comparedwithstarbursts,theyonlyhaveabout (again normalizedtohydrogenemission).Sinceincreasing The factthat[Oi]26300andthenear-infrared[Fen] In Fig.21weshowanemission-lineratiodiagramforour 0.14 0.96 0.32 0.94 0.09 0.11 0.95 0.28 0.15 0.98 0.34 0.94 0.10 0.21 0.97 0.21 0.19 0.97 0.11 0.22 0.13 0.95 0.21 0.31 0.25 0.09 0.94 0.08 0.97 0.08 0.11 0.98 0.16 0.93 0.07 0.15 13.30 12.60 15.17 13.24 13.35 14.46 4.71 4.63 4.68 0.75 0.09 0.87 0.19 2.35 0.22 2.31 3.59 0.06 0.96 0.96 6.66 2.12 3.34 7.24 0.10 0.85 7.39 0.13 2.03 2.31 2.10 1.76 1.09 1.12 1.09 1.59 © 1996RAS,MNRAS 283,777-797 0.23 0.006 0.006 0.62 0.26 0.006 0.013 0.005 0.24 0.11 0.012 0.29 0.003 0.37 0.098 0.011 0.59 0.27 0.13 0.16 0.33 0.56 0.55 0.006 0.40 0.62 0.007 0.41 0.16 0.005 0.22 0.14 0.52 0.014 0.69 0.55 19 9 6MNRAS.283. .111S Figure 21.Plotof[Fen]1.257\imfPaßversus[Oi]/16300/Ha,forgalaxiesfromthecombinedsample,OrionHnregion,andSNRs. The producing SNR-likeratios).Theseratios(2and10)arealso the Seyfertsdonot. grains, sothisproblemcanbesolvedbykeepingiron the ratiosbywhichoxygenandironaredepletedonto nent withoutmakingthe[Fen]/Pa/?ratiotoolarge(i.e., impossible toaddasufficientlylarge[Oi]-emittingcompo- ray photoionization)willfavourboththeselinesequally,itis dashed lineisthemixingcurveforacombinationofHnregionsandSNRs.Notethatstarburstgalaxiesliealong curve, but low. ThisreductioninthesizeofPIZisexactlywhat the sizeofPIZ),wecanreproduceOrionnebula depleted. Infact,ifwereducetheemissionfrom[Oi]and fluxes foranyironlinespredictedbytheirmodel,sowe nizing shocksofthekindproposedbySutherland,Bicknell happens whentheionizingcontinuumismadesofter. emission-line ratios,wherethegas-phaseironabundanceis producing theobserved[Fen]/Pa/?ratios.However, it cannot sayforcertainwhethersuchshocksarecapableof Unfortunately, Sutherlandetal.didnotpresenttheline explaining theclosespatialcorrelationbetweenradio sion-line ratios.Thesehavetheadvantageofnaturally & Dopita(1993)maybeabletoproducetheobservedemis- land etal.’snotation),ashappens inSNRs,andthatthe the factthatnear-infrared [Fen]lineswouldbepro- and [Fen]emissionseeninNGC1068(Blietzetal.1994). other linesformedhere([O i],[Nn],[Sn])havesimilar duced mainlyinthedensepost-shock gas(zoneFinSuther- SNRs wouldbeproduced.We reachthisconclusionfrom seems probablethatratios similar tothoseobservedin [Fe n]inSeyfertgalaxiesbyafactor~60(e.g.,reducing strengths inSNRs andtheemission-linefilaments ofCen- © 1996RAS,MNRAS 283,777-797 As analternativetopower-lawphotoionization,autoio- © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Oh QQ_ 0 O