Astron.Astrophys.359,457–470(2000) ASTRONOMY AND ASTROPHYSICS α Structure of the Mg and damped Lyman- systems II (cid:63) along the line of sight to APM 08279+5255 P.Petitjean1,2,B.Aracil1,R.Srianand3,andR.Ibata4 1 Institutd’AstrophysiquedeParis–CNRS,98bisBoulevardArago,75014Paris,France 2 UACNRS173–DAEC,ObservatoiredeParis-Meudon,92195MeudonCedex,France 3 IUCAA,PostBag4,GaneshKhind,Pune411007,India 4Max-Plank-Institutu¨frAstronomie,Ko¨nigstuhl17,69117Heidelberg,Germany Received27January2000/Accepted9May2000 Abstract. Astudyoftheabsorptionsystemstowardthegravi- absorberislargerthan200h−1 pc.ColumndensitiesofAlii, 75 tationallylensedquasarAPM08279+5255ispresented. Feii,Siii,CiiandOiindicateabundancesrelativetosolarof MostoftheMgiisystemsintheredshiftrangez (cid:24)1.2–2.07, −2.31,−2.26,−2.10,−2.35and−2.37for,respectively,Fe,Al, although saturated, show large residuals at the bottom of the Si,CandO(forlogN(Hi)=20.3).Thesesurprizinglysimilar lines. The most likely interpretation is that individual clouds valuesindicatethattheamountofdustinthecloudisverysmall withinMgiihalosdocoveronlyoneofthetwobrightestQSO asareanydeviationsfromrelativesolarabundances.Itseems images.Theseparationbetweenthetwolinesofsightdecreases likelythattheupperlimitsfoundforthezincmetallicityofsev- from1.7to0.7h−751kpc(qo=0.5,zlens=1)betweenz=1.22and eraldampedLyman-αsystemsazt >3inprevioussurveysis z=2.07.ThisrevealsthatMgii halosaremadeofacollectionindicativeofatruecosmologicalevolutionofthemetallicityin ofcloudsofradiussmallerthanabout1h−1kpc. individualsystems. 75 TwostrongMgiiabsorbersatzabs=1.062and1.181arestud- ied in detail. This is the first time that the Na iλ3303 doubletKey words: galaxies: quasars: absorption lines – galaxies: isdetectedinsuchhighredshiftsystems.Togetherwiththede-quasars:individual:APM08279+5255 tectionoftheMgiλ2852transition,thisstronglyconstrainsthe physicalcharacteristicsofthegas.TheN(Nai)/N(Mgi)ratio isfoundtobelargerthanunity,implyingthatthegasiscooland neutral. The Doppler parameters measured in individual and 1. Introduction welldetachedcomponentsareprobablyassmallas1kms−1. ThegravitationallylensedBroadAbsorptionLine(BAL)QSO i,Caii,Mgi,Tiii,MniiandFeii APM 08279+5255 (zem = 3.911) has been given much atten- observedatzabs=1.1801areveryclosetothatobservedalong tion since its discovery by Irwin et al. (1998), as it is one of thelineofsighttowards23OriinourGalaxy.Theshapeofthe themostluminousobjectsintheuniverseevenaftercorrection QSOcontinuumisconsistentwithattenuationbydustatz(cid:24)1 for the gravitational lensing induced amplification. Adaptive- (AV(cid:24)0.5mag).AltogetheritisfoundthattheHicolumnden- opticsimaginghasrevealedtwomaincomponents(Ledouxet sityatz=1isoftheorderof1to51021cm−2,thecorresponding al. 1998b), separated by 0.378(cid:6)0.001 arcsec as measured on metallicityisintherange1–0.3 Z(cid:12),theoveralldust-to-metal HST/NICMOS data (Ibata et al. 1999), and of relative bright- ratioisabouthalfthatinourGalaxyandtherelativedepletion nessFB/FA =0.773(cid:6)0.007.TheHSTimagesrevealalsothe ofiron,titanium,manganeseandcalciumissimilartowhatis presence of a third object C with FC/FA = 0.175(cid:6)0.008, lo- observedincoolgasinthediskofourGalaxy.Theobjectsasso- cated in between A and B and almost aligned with them. The ciatedwiththesetwosystemscouldbothcontributetothelens point-spread-functionmodelfitsonthethreeobjectsareconsis- together with another possible strong system at zabs = 1.1727 tent with the three components being point-sources, and their andthestrongLyman-αsystematzabs=2.974. colorsaresimilarwithintheuncertainties.Thereisnotraceof The probable damped Lyman-α system at zabs = 2.974 has thelensingobjectuptomagnitudeV =23. 19.8 < log N(H i) < 20.3. The transverse dimension of the A high S/N ratio high-resolution spectrum of APM 08279+5255wasobtainedattheKecktelescope(Ellisonetal. Sendoffprintrequeststo:P.Petitjean (cid:63) Based on observations collected at the W.M. Keck Observatory, 1999a,b), and made available to the astronomical community. Thisspectrum,thoughcomplicatedbythecombinationoflight whichisoperatedasascientificpartnershipamongtheCaliforniaIn- travelingalongthreedifferentsightlines,isauniquelaboratory stitute of Technology, the University of California and the National Aeronautics and Space Administration. The Observatory was made forstudyingtheinterveningandassociatedabsorptionsystems. possiblebythegenerousfinancialsupportoftheW.M.KeckFounda- InthispaperwestudythestructureofsixinterveningMgii tion. systemsat1.2<z<2.07andthephysicalcharacteristicsofthe 458 P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 gasintwoverystrongMgiisystemsdetectedatzabs=1.06and 1 1 1 1.18,which,weargue,aredampedLyman-αsystemsandmay well reveal the lensing galaxies. We also comment on a third probabledampedLyman-αsystematzabs =2.974.Thispaper 0.5 0.5 0.5 is organized as follows: the data are described in Sect.2; the structureoftheinterveningMgiisystemsisinvestigatedusing thecoveringfactoranalysisinSect.3;wedemonstratethatthe 1 1 1 Mgiisystemsatzabs=1.06and1.18aredampedLyman-αsys- temsinSect.4;wediscussaprobabledampedLyman-αsystem at zabs = 2.974 in Sect.5. We adopt Ho = 75 km s−1 Mpc−1 0.5 0.5 0.5 andqo=0.5throughoutthepaper. 1 1 1 2. Data AhighS/Nratio,high-resolutionspectrumofthezem =3.911 0.5 0.5 0.5 quasarAPM08279+5255wasobtainedwiththeHIRESechelle spectrograph at the 10m Keck-I telescope (Ellison et al. 1999a,b). This data was made public together with a low- -50 0 50 -50 0 50 -50 0 50 resolution spectrum of the quasar and a high-resolution spec- trumofastandardstar.Wehavecorrectedthehigh-resolution spectrumofAPM08279+5255forsmalldiscontinuitiesinthe 1 1 1 continuum,whichareprobablyduetotheinappropriatemerging ofdifferentorders.Thesediscontinuitieshavebeenrecognized bycomparingthehighandlow-resolutionspectra.Thelatteris 0.5 0.5 0.5 alsousedfornormalizationofthehigh-resolutiondata.Atmo- spheric absorption features were identified from the standard star spectrum. Voigt profile fitting of the absorption features 1 1 1 have been performed using the context FITLYMAN (Fontana &Ballester1995)oftheEuropeanSouthernObservatorydata 0.5 0.5 0.5 reductionpackageMIDASandthecodeVPFIT(Carswelletal. 1987). We have measured the final spectral resolution by fit- tingthenarrowatmosphericabsorptionlineswhicharefreeof blending. We find FWHM (cid:24) 8 km s−1 (b (cid:24) 4.8 km s−1) at 1 1 1 6900A˚,R=37500,andusethisvaluethroughoutthepaper. 0.5 0.5 0.5 3. Structureoftheinterveningz>1Mgiiabsorbers Ellison et al. (1999b) have already noted that the two lines of -50 0 50 -50 0 50 -50 0 50 someoftheMgiiλλ2796,2803doubletscannotbefittedwith thesamecolumndensityandDopplerparameter.Ascanbeseen Fig.1.Coveringfactor(bottompanels)ofsixMgiisystemsobserved onFig.1,mostofthesystemshaveadoubletratioclosetounity alongthelineofsighttoAPM08279+5255calculatedfromthepro- inspiteofhavingresidualintensitiesinthenormalizedspectrum filesoftheMgiiλ2796(toppanels)andMgiiλ2803(middlepanels) closeto0.5. absorptions. If the background source is a point source and if the Mg iiλ2796 absorption line is resolved, then the residual in- tensityofthenormalizedspectrummeasuredatanyvelocityv from different images, it is possible that the column densities withrespecttothecentroidoftheline,isequaltoe−τ(v),where are different along different lines of sight and, as a limiting τ(v) is the optical depth at v and the residual intensity of the case,thattheabsorbingclouddoesnotcoveralltheimages.We Mgiiλ2803lineise−τ(v)/2.ItisapparentfromFig.1thatthe discussthesedifferentpossibilitiesbelow. aboveconditionisnotfulfilledformostoftheMgiidoublets. Therearetwopossibilitiestoexplainthis.Ifthelinesarenot 3.1. Unresolvednarrow-components resolved,themeasuredresidualintensityisaffectedbyconvolu- tionofthetrueabsorptionprofilewiththeinstrumentalprofile. If the absorption profiles are made up of unresolved compo- Thiscanintroduceartificialresidualsatthebottomofthesat- nents, the observed residual intensity does not correspond to uratedabsorptionfeatures(seee.g.Lespine&Petitjean1997). the real optical depth. Given the resolution of the spectrum Alternatively, as the observed light is a combination of light (R (cid:24) 37500), a saturated Mg iiλ2796 line can have a resid- P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 459 Fig.2.BestfitoftheMgiidoubletatzabs=1.5523withfivenarrow components(b<1.5kms−1)andacoveringfactorf =1(reduced χ2of1.5). ual intensity in the normalized spectrum of 0.5 if its Doppler parameterissmallerthan1.5kms−1.Inthatcase,theresidual intensityoftheMgiiλ2803lineisintherange0.5–0.6,depend- ing on the actual column density. The two equivalent widths differbynomorethan20%(seee.g.Lespine&Petitjean1997). The system at zabs = 1.5497 can be indeed fitted this way using10componentswithbvaluesintherange1.1–1.7kms−1. Inthatcasethewelldetachedcloudintheredwing(seeFig.1) is fitted with two adhoc nearly identical components though theMgiiλ2796profileisperfectlyfittedwithasingleresolved componentmodel.Howevertheone-componentmodelcannot explain the strengths of the two Mg ii lines without invoking partialcoverage(seebelow). Weusethezabs =1.5523systemtoillustratethecase.The Fobigta.i3n.eBdewstitfihtsfivoeftnhaerrzoawbs(b=<1.515.523kmMgs−ii1)ancodmFpeoinieanbtssoarnpdtiofn=lin0e.6s finalspectrumhasadefectinthecenteroftheMgiiλ2796line. (top panel, reduced χ2 of 0.9) and three broader (b > 2.5 km s−1) WehavethereforeusedthreeindividualexposuresofhighS/N componentsandf =0.45(bottompanel,reducedχ2 of0.6).Thefit ratio (Ellison private communication) to correct for this. The parametersaregiveninTable1. finalopticaldepthvariationsfromonespectrumtotheotheris about 2%. Fig.2 shows the best fit to the doublet considering fullcoverage(f=1).Fivenarrowcomponents(b<1.5kms−1) f =0.45andreducedχ2 =0.6(seeFig.3andTable1).Theb are needed. Although good (reduced χ2 of 1.5), the fit is not valuesarelargerthan2.5kms−1relaxingtherestrictiononthe completely satisfactory. We have fitted consistently the Mg ii temperature. together with the Fe ii lines considering that only one of the A statistically acceptable fit is difficult to find for the brightestsourcesiscovered.Withfivecomponents,agoodfit systems at zabs = 1.221 and 1.5523. The red wing of the (reduced χ2 of 0.9), shown on Fig.3, is obtained if B is not Mg iiλ2803 line at zabs = 1.221 is blended with another ab- covered(f =0.6).Detailsofthesubcomponentparametersare sorptionfeaturewhichwefounddifficulttoidentify.Itcouldbe given in Table 1. As the Mg ii doublet ratio is close to one, Mgiλ2852atzabs =1.1727.Thissystemispossiblydetected small b values are needed even with the assumption of partial byFeiiλλ2344,2382,Mniiλ2576andCaiiλ3934(seeFig.4). coverage. Theotherlinesareeitherbelowthedetectionlimitorblended. ItshouldbenotedhoweverthataDopplerparametersmaller Confirmation of the Ca ii lines, which are free of blending, than 1.5 km s−1 corresponds to a temperature smaller than would be particularly important as, probably, the presence of 3500K,asurprizinglysmalltemperatureforthisgaswhichis this additional system, together with the presence of the three mostprobablyionized.Allowingforasmallernumberofcom- dampedsystemsatzabs =1.062,1.181and2.974(seebelow) ponents, we can find a good fit with only three components, shouldbetakenintoaccountinanymodelofthelens. 460 P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 &Petitjean2000).Giventheuncertainties,ifonlyoneline-of- 1 sight is completely absorbed, the covering factor should be in 1 therange0.4–0.6.Ofcourse,itislargerifthesecondline-of- 0.9 0.5 sight is not completely clear. We have computed the covering 0.8 Ca II 3934 Fe II 2344 factor for the Mg ii systems using the method described by 0.7 10 1 Srianand & Shankaranarayanan (1999). This assumes that the 0.9 lines are resolved. It can be seen in Fig.1 that for the three 0.5 0.8 systemswithzabs<1.7,thecoveringfactorrangesbetween0.5 Ca II 3969 Fe II 2374 and0.6whereasforthesystemswithz>1.7,thecoveringfactor 0.71 10 islarger(butalwayslessthan0.8)withthepossibleexception 0.5 0.5 of the zabs = 2.0668 system. The latter system is quite weak howeveranduncertaintiesarelarge.Thevaluesofthecovering Mg I 2852 Fe II 2382 10 0 factorforthethreelowerredshiftsystemssuggestthattheclouds 1 cover one of the two brightest background sources only. This 0.9 0.5 hastobeinvestigatedinmoredetail,though. 0.8 Mg II 2803 Ti II 3242 110 0.711 3.2.2. Opticaldepthsalongdifferentsightlines 00..55 00..55 In this section we investigate the effect of the optical depth Mg II 2796 Mn II 2576 beingdifferentalongdifferentsightlines.Inordertomakeour 00 00 -200 0 200 -200 0 200 analysissimpler,weconsideronlytwoimages,AandB,with fractionalfluxcontributionsF1=0.6andF2=0.4.Supposethe Fig.4.ProbableMgiisystematzabs =1.1727.Mostofthelinesare optical depth along the two sight lines are τ1 and τ2 then the blendedbutthesimilarityoftheFeii,MgiiandCaiiprofilessupports measuredresidualintensitiesare, theidentification.Verticaldottedlinesindicatethepositionsofthefour probablecomponents. R(2796) = F1e−τ1 +F2e−τ2 (1) R(2803) = F1e−τ1/2+F2e−τ2/2 Table1.Fitparametersforthezabs=1.5523system Theresidualintensitiescanbewrittenas(Srianand&Shankara- narayanan1999), z bb (cid:6) N(MgII)a (cid:6) N(FeII)a (cid:6) R(2796) = 1−f +fe−τ (2) Bestfitwithfivecomponents;f =0.6 R(2803) = 1−f +fe−τ/2 1.552165 1.7 0.5 11.51 0.15 1.552219 1.1 0.5 11.88 0.05 11.67 0.12 where, f and τ are the resulting covering factor and optical 1.552270 1.3 0.2 14.47 0.27 12.79 0.05 depthandtherefore, 1.552338 3.4 0.2 12.47 0.10 12.23 0.06 1.552399 0.8 0.1 14.67 0.09 12.34 0.07 [1−R(2803)]2 f = (3) Bestfitwiththreecomponents;f =0.45 1+R(2796)−2R(2803) 1.552170 2.9 0.5 11.67 0.10 11.29 0.11 FromEqs.(1)and(3)onecanderivef asafunctionofF1,F2, 1.552274 4.2 0.6 13.02 0.18 12.82 0.09 1.552378 4.6 0.7 12.85 0.20 12.44 0.07 τ1andτ2.Thisanalysisassumesthattheabsorptionprofilesare resolvedintheHIRESspectrum.AsF1andF2areknownfrom baliongkamritsh−m1ofthecolumndensityincm−2 observation,thecoveringfactordependsonτ1andτ2only. In order to investigate the parameter space we have com- putedandplottedonFig5,thecoveringfactor(paneld),resid- 3.2. Partialcoveringfactor ualintensitiesofthelines(panelsaandb)andtheirratio(panel c)asafunctionofτ2fordifferentvaluesofτ1.Inpanels(a)and 3.2.1. Computingacoveringfactor (b),thetwohorizontaldottedlinesshowtherangeofobserved Wecaninterprettheobservationsintermsofacoveringfactor values for the Mg iisystems with zabs<1.7. In panel (c) the whichisthefractionofthebackgroundsourcecoveredbythe dottedlinesgivethemeasuredvaluesforthethreelow-redshift absorbing cloud. The relative brightness of the three sources Mgiisystems. are FA,B,C/Ftot = 0.513, 0.397 and 0.090 (Ibata et al. 1999). As expected, when τ1 = τ2, f = 1 (albeit with various Ifoneline-of-sightiscompletelyabsorbed(conditionimposed ratiosofresidualintensities)andforτ1 =/ τ2 thecoveringfac- by the fact that the doublets are saturated) and the other free tors are less than 1.0. When τ1 (cid:21) τ2 (respectively τ1 (cid:20) τ2), of absorption, then the covering factor is 0.40, 0.49, 0.51 and the covering factor is in the range 0.6–1.0 (respectively 0.4– 0.60if,respectively,Bonly,B+C,AonlyandA+Carecovered 1.0).Conversely,theobservedresidualintensities,R(2796)and (thecaseConlyisveryunlikelyasCisclosetoAandlocated R(2803),togetherwiththemeasuredcoveringfactor,f,canbe in between A and B, Ibata et al. 1999; see however Srianand usedtoconstrainτ1andτ2. P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 461 111111 111111 000000......888888 000000......888888 000000......666666 000000......666666 000000......444444 000000......444444 000000......222222 000000 111111 222222 333333 444444 555555 000000 111111 222222 333333 444444 555555 111111 111111 000000......888888 000000......888888 000000......666666 000000......666666 000000......444444 000000......444444 000000 111111 222222 333333 444444 555555 000000 111111 222222 333333 444444 555555 Fig.5. Panels a and b give, respectively, the observed residuals R(2796) and R(2803) in the normalized spectrum and in the center of the Mgiiλλ2796,2803linesasafunctionofτ2fordifferentvaluesofτ1.τ1andτ2aretheopticaldepthsinthecenteroftheMgiiλ2796linealong, respectively,lineofsightnumberonetowardsimageAandlineofsightnumbertwotowardsimageB(seeEqs.(1)and(2)).Itisassumedthat thefractionalfluxcontributionsaref1 =0.6andf2 =0.4.Thetwohorizontaldottedlinesshowtherangeofobservedvalues.Panelcgives theratiooftheresidualintensitiesofthetwoabsorptionlinesasafunctionofτ2fordifferentvaluesofτ1.Thedottedlinesshowtheobserved valuesinthethreesystemsatzabs<1.7.Inpaneldthederivedcoveringfactorf (Eq.(3))isshownasafunctionofτ1fordifferentvaluesof τ2.Measuredvaluesareindicatedbydottedhorizontallines. The covering factor estimates for zabs< 1.7 systems are addition,whenconsideringtotalequivalentwidths,theconse- in the range 0.5 and 0.6 (see Fig.1). This, together with the quence of contamination by weak lines is small unlike in the observed residual intensities, indicates that the absorbing gas case of the analysis of the residual intensities. The constrains issaturatedalongonesightlineonlywithopticaldepthratios aremuchweakerhowever. as large as ten. For example, the well detached component in Let us assume that the absorption is saturated along theredwingofthezabs =1.5497systemhasf = 0.6(with line of sight number one and optically thin along line of a typical error of 0.02), a residual intensity ratio (cid:24)0.84 (with sight number two. Therefore Wreal(2796) = Wreal(2803) and 1 1 a typical error of 0.02) and R(2796)’0.40 at the core of the Wreal(2796)=2(cid:2)Wreal(2803).Thecombinedobservedequiv- 2 2 line.Thisimpliesthatthecontributiontothisabsorptioncomes alentwidthratiois: mainlyfromthelineofsighttowardA+Cwiththeopticaldepth alongBbeingmorethananorderofmagnitudesmaller. Wobs(2803) 1 + W1real(2796)F1 = 2 W2real(2796)F2 (4) Wobs(2796) 1+ W1real(2796)F1 3.3. Equivalentwidthratio W2real(2796)F2 Theprecisedeterminationofthestrengthoftheabsorptionlines where F1 and F2 are the fractional flux contributions of the along different lines of sight should await HST/STIS spectro- two distinct background sources. In Fig.6, we have plotted scopic observations. Though we derived some information on Wreal(2796) /Wreal(2796) versus Wobs(2803) /Wobs(2796) 1 2 this in the previous section using the absorption line residual for F1/F2 = 0.65 (1 is B; 2 is A+C), 1 (1 is A or B+C; 2 is intensities,theresultsdependcruciallyontheassumptionthat B+C or A) and 1.5 (1 is A+C; 2 is B). The vertical dashed- theabsorptionlinesareresolved.Wecancomplementthepre- dotted lines correspond to the observed doublet ratios of the viousanalysisusingthetotalequivalentwidthsandtheirratios systemsatzabs =1.221,1.5523andofthereddestcomponent withoutmakinganyassumptionaboutthespectralresolutionIn ofthezabs=1.5497system. 462 P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 (qo =0.5)betweenz =1.22andz =2.04andismorethanan order of magnitude smaller than the radius of Mg ii halos at intermediateredshift.Althoughevolutionisprobable,itwould be really surprizing that the Mg ii systems studied here with Wr <0.5A˚ atzabs >1.2havecharacteristicdimensionsmore thananorderofmagnitudesmallerthanwhatisderivedatlower redshift.Iftrue,thiswouldsuggestthatthestructureoftheMgii halosattheseredshiftsdifferssubstantiallyfromthatatlower redshifts. Amorelikelyexplanationoftheseobservationsisthatthe halos are composed of a collection of clouds (see Petitjean & Bergeron 1990; Srianand & Khare 1994) and that individual cloudscoveronlyonesightline.Thenumberdensityofclouds is large enough so that the total covering factor of the halo is closetoone,consistentwithobservationsofassociatedgalaxies. Fig.6.EquivalentwidthratiooftheMgiiλ2796absorptionlinealong However,individualclouds,regularlyspreadoverthevelocity profile by kinematics, cover only one image of the lens. The twodifferentline-of-sights1and2versustheobserveddoubletratio forthreedifferentfluxratiosF1/F2=0.65(solidline),1(dashed-dotted numberdensityofcloudsisnotlargeenoughfortheabsorption line)and1.5(dashedline).TheMgiidoubletisassumedtobesaturated material to cover the two lines of sight at all velocities. The alongline-of-sight1andopticallythinalongline-of-sight2.Thedou- distance over which the optical depth, and hence the column bletratiosobservedforsixMgiisystemstowardAPM08279+5255 density of Mg ii, changes by at least one order of magnitude areshownasverticaldashed-dottedanddottedlinesforzabs<1.5and at zabs<1.7 is smaller than (cid:24) 1h−751 kpc. In contrast, the two zabs>1.5respectively.Redshiftsareindicatednexttothelines. strongMgiisystemsatzabs =1.06and1.18(seebelow)have covering factor equal to one (the lines are saturated and go to thezerolevel)overmorethan200kms−1.Theselattersystems From Fig.6, it can be seen that the zabs = 1.221 system arelikelytoariseduetoabsorptionthroughcentralregionsof (doubletratioof0.96)requirestheratiooftheequivalentwidths galaxieswherethenumberofcloudsissolargethatsaturated ofMgiiλ2796alongthetwolinesofsighttobelargerthan8. absorptionoccursalongbothlinesofsightwhatevertheradius andthecolumndensitiesalongthetwosightlinesdifferbymore oftheindividualcloudsmightbe. thananorderofmagnitude. 4. ThetwoMgiisystemsatzabs=1.062and1.181 3.4. Dimensionofindividualclouds ThepresenceofastrongMgiisystem(Wrλ2803(cid:24)2.4A˚)at Rauchetal.(1999)haveobservedstrongvariationsofCiiand zabs(cid:24)1.18wasalreadymentionedbyIrwinetal.(1998).There Siiiabsorptionsatz=3.538alongtwosightlinesseparatedby is an additional even stronger Mg ii system at zabs = 1.062 only13h−1pc.However,asthevelocitydifferencebetweenthe (Wrλ2803 (cid:24) 3.3 A˚). Although the Mg ii lines are redshifted quasarandthesystemisonly6000kms−1,itcannotbeexcluded in the Lyman-α forest and may be blended with Lyman-α in- that the latter system is somehow associated with the quasar. terveningabsorptions,theexistenceofthesystemisconfirmed Variationsofthestrengthofmetallinesystemshavealsobeen by numerous lines redshifted redward of the quasar Lyman-α reported along adjacent lines of sight with larger separations emission.Asthetwobrightestimagesofthelensedquasarhave (5–10 kpc) by Monier et al. (1998) and Lopez et al. (1999). similarmagnitudes,itisexpectedthatthelinesofsighttoboth Each time however a damped Lyman-α system is seen along imagespassthroughthecoreofthelensingobjectwherestrong one of the sightlines. Contrary to these previous studies, the Mg ii absorption is likely to occur. The gravitational lensing Mgiisystemsweexamineherearemostlikelytobeassociated maythusresultfromthecumulativeeffectofthetwogalaxies withhalosofinterveninggalaxies. associatedwiththesetwoabsorbingsystemstogetherwiththe From the detection of associated galaxies, radii of the or- objectsresponsibleforthepossiblesystematzabs=1.1727(see derof35h−1 kpchavebeenderivedforMgiihalosproducing Sect.3.1)andtheotherdampedLyαsystematzabs=2.974.Ab- absorptionswithequivalentwidthsWr>0.3A˚ atz<1(Berg- sorptionsfromMgii,Feii,Caii,Mnii,TiiiandNaiareseen eron & Boisse´ 1991, Steidel 1993). Dimensions of the same inbothsystems(seeFigs.7,8). orderhavebeenderivedfromthestudyofMgiisystemsseen along two lines of sight separated by 3 arcsec (Smette et al. 1995). The latter authors find a lower limit of 22 h−1 kpc for 4.1. Naiabsorptions 50 theradiusofMgiiabsorberswithWr>0.3A˚ at0.5<z<1.3. In both Mg ii systems, the weak Na iλλ3303,3303 doublet is IfweassumethatthelensinggalaxyofAPM08279+5255 detected (see Figs.7 and 9). The fact that the two lines of the is at zlens (cid:24) 1 (see next section), the separation between the doubletareseenwithconsistentstrengthsgivesconfidencethat twolinesofsighttoAandBdecreasesfrom1.7to0.7h−1kpc theidentificationiscorrect.Wehaveidentifiedlinesfromother 75 P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 463 Fig.7.Absorptioninafewtransitionsonavelocityscalewithorigin Fig.8.Absorptionsinafewtransitionsonavelocityscalewithorigin at zabs = 1.06230. Vertical dashed lines mark velocity components at zabs = 1.18070. Vertical dashed lines mark velocity components discussedinthetext. discussedinthetext. sitiesofotherspeciesthatarefoundtobesurprisinglycloseto metallinesystemsinthevicinityofthedoublet.Thespectraof what is observed in typical interstellar clouds (see below and APM08279+5255andthestandardstararecomparedinFig.9 Table 2). This leaves little doubt that the systems are indeed to rule out the possibility that the absorption features are of dampedLyman-αsystems. atmosphericorigin. In our Galaxy, such high Na i column densities are seen The column densities obtained by Voigt profile fitting are only in dense and cool gas (see below). The typical b values large,logN(Nai)=12.9and13.5at,respectively,zabs=1.0626 oftheNaidiffusecomponentsinboththelocalandlow-halo and1.1801.Theseparationofthetwoprincipallinesofsightis (cid:24)1.9h−1 kpc at z (cid:24) 1. It is therefore possible that the clouds gasisabout0.7kms−1correspondingtoT <500K(Weltyet 75 al.1994).Althoughtheconclusionisveryuncertaingiventhe seen by their Na i absorptions do not cover the two brightest resolutionofthespectrum(R(cid:24)37500)andthedoublenature linesofsight.Inthatcase,however,thecolumndensitycould of the background source, the lines we observe are consistent beevenlargerbyafactoroftwo(seeSect.4.4). withbvaluesassmallas1kms−1 (seebelow).Fromthis,we Using the measurements by Sembach et al. (1993) and deriveanupperlimitonthetemperatureofT <2000K. Diplas&Savage(1994),Bowenetal.(1995)findthat,inour Galaxy, log N(H i) = 0.688 log N(Na i) + 12.16. Note that thiscorrelationholdsuptocolumndensitieslogN(Hi)>21 4.2. Commentsoneachsystem (see e.g. Ferlet et al. 1985). Applying this correlation for the zabs=1.0626and1.1801absorbersgivesneutralhydrogencol- 4.2.1. zabs=1.062 umndensitiesoftheorderoflogN(Hi)(cid:24)21.0and21.4respec- Asubsetoftheabsorptionsdetectedinthissystemisshownin tively,forgas-phasemetallicitiescomparabletowhatisseenin Fig.7.Itcanbeseenthatstrong(butunsaturated)absorptions the interstellar medium in our Galaxy (the neutral hydrogen from Ti ii and Ca ii are detected. The profile of the Ti ii ab- columndensitiescouldbeevenlargerifthemetallicityinthese sorptionisspreadoverabout250kms−1butdoesnotshowany intermediate-redshift systems is smaller than in our Galaxy). edge leading pattern (Prochaska & Wolfe 1998). We have se- SuchlargevaluesforN(Hi)aresupportedbythecolumnden- lectedtoexaminetwocomponentsatzabs=1.0613and1.0631 464 P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 4.2.2. zabs=1.181 Asubsetoftheabsorptionsdetectedinthissystemisshownin Fig.8. The profile of the Mg i absorption is spread over more than 200 km s−1 but, as for the previous system, it does not show any edge leading pattern. A number of absorption lines areopticallythinormoderatelysaturatedandreliablecolumn densitiescanbederivedeventhoughdifficultiesarisefrommost of the components being badly defined (see Fig.8). We have selected for study two subcomponents which are clearly seen in all absorption profiles at zabs = 1.1799 and 1.1801. They areindicatedonFig.8byverticaldashedlines,andthecolumn densitiesobtainedfromVoigt-profilefittingaregiveninTable2. Dopplerparametershavebeenconsideredtobeidenticalforall species.Forzabs=1.1799wefindb=2.5kms−1. Forthecomponentatzabs =1.1801inwhichNaiabsorp- tion is detected, the Doppler parameter is estimated by fitting thewell-definedlinesofthesodiumdoubletafterhavingtaken intoaccounttheeffectofNaiλ3303.3beingpartiallyblended withanatmosphericfeature(seeFig.9).Weobtainabestvalue b = 1.1+1.0 km s−1. The column densities derived using the −0.5 twovaluesb(cid:6)1σdifferbylargefactors.Wehavethereforere- finedthedeterminationofbandN usingthefollowingindirect argument. TheratiooftheMgitotheNaicolumndensitiesinneutral gascanbewritten N(Mgi) Mgi Na δ Z = (cid:2) (cid:2) Mg (cid:2) Mg, (5) N(Nai) Mg Nai δ Z Na Na where δ is the depletion of the element due to the presence Fig.9. Portions of the APM 08279+5255 spectrum centered on the expectedpositionsofNaiλ3303.3(toppanel),Naiλ3303.9(middle of dust and Z the abundance. Assuming that (i) the relative panel)andMgiλ2852(bottompanel)atzabs=1.181.TheNaicom- abundance of Na to Mg is solar, ZMg/ZNa = 19, (ii) the rel- ponent at zabs = 1.1801, detected by the two lines of the doublet is ative depletion into dust-grains is δMg/δNa > 0.3 (Savage & indicatedbyaverticaldashedline.Thespectrumofthestandardstar Sembach 1996) and (iii) (Mg i/Mg ii)(cid:2)(Na ii/Na i) > 0.15 isoverplottedinthemiddlepanel.Metallinesfromothersystemsare in cold and neutral gas (Pe´quignot & Aldrovandi 1986), we indicatedbycrosses. derive N(Mg i)/N(Na i) (cid:21) 0.8. This estimation is certainly very approximate. However, this simple argument shows that the latter ratio cannot be much smaller than 0.5. If we as- because they show well detached absorptions in Ti ii, Mn ii, sumeb=1.5kms−1,thenwefindN(Mgi)/N(Nai)=0.045 Caiiand/orFeiiplusthecomponentatzabs=1.0626inwhich which is definitively too small. We therefore have fitted the we see Na i. Column densities are listed in Table 2. We have absorption lines decreasing b from 1.5 to 0.5 km s−1 to adjustedthebestvaluesforbfromthefittothelinesthatarefree find the largest N(Mg i)/N(Na i) ratio. We find a maximum ofanyblending,consideringforsimplicitythatallspecieshave N(Mgi)/N(Nai)=0.3forb=0.8kms−1,logN(Mgi)=13.0 thesameDopplerparameterandassumingcompletecoverage. andlogN(Nai)=13.5. We find b = 1.5, 1.5 and 1.9 for, respectively, zabs = 1.0613, 1.0626and1.0631. 4.3. Physicalstateofthegas Caiisnotdetectedandthe3σupperlimitonthecolumnden- sityinthethreecomponentswehaveselectedis<10.43,10.43 Table 2 contains the column densities measured in the five and10.20.TheMgiiλ2796,2803andFeiiλλλ2382,2600,2586 subcomponents defined above. The penultimate column gives lines are badly saturated. Moreover, they are redshifted in the forcomparisonthecolumndensitiesmeasuredbyWeltyetal. Lymanα forest, which prevents any fit of the lines. However, (1999)intheneutralgastoward23Ori.Thisgasisfoundtohave the unsaturated Fe iiλ2260 features detected at zabs = 1.0613 temperatureT (cid:24)100K,hydrogendensitynH (cid:24)10–15cm−3 and1.0626(seeFig.7)giveareliableestimateoftheFeiicol- (andthereforetotalthicknessof12–16pc)andelectronicden- umn density, log N(Fe ii) = 13.90(cid:6)0.80 and 14.10(cid:6)0.80 for sityne(cid:24)0.15(cid:6)0.05cm−3.ThelastcolumnofTable2givesthe b=1.5kms−1,thecontinuumbeingadjustedlocally.Thenon- measurementsforthestrongestcomponentofthewarmneutral detection of Fe iiλ2367 at zabs = 1.0631 gives an upper limit gastowardµCol(Howk&Savage1999).Thisgasisfoundto logN(Feii)<14.60. have T (cid:24) 6000–7000 K and ne (cid:24) 0.3 cm−3. The two sets of P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 465 Table2.Columndensitiesa Redshift 1.0613 1.0626 1.0631 1.1799 1.1801 23Orib µCold HI 20.74 19.86 H2 18.30 15.50 NaI <11.90 12.91(cid:6)0.04 <12.80 <12.40 13.50(cid:6)0.11 13.36 11.6e MgI 11.6:c bl bl 11.61(cid:6)0.06 13.0(cid:6)0.74 13.81 12.55 MgII 15.68 15.08 CaI <10.43 <10.43 <10.20 <10.80 <10.33 10.20 CaII 11.31(cid:6)0.11 bl 11.71(cid:6)0.03 11.33(cid:6)0.39 11.79(cid:6)0.35 12.10 12.19 TiII 11.80(cid:6)0.05 <12.00 11.90(cid:6)0.02 <11.66 <11.20 11.23 11.78 MnII bl bl 12.70(cid:6)0.40c bl bl 13.15 12.48 FeI <11.50 <11.63 <11.40 <11.40 <11.50 11.34 FeII 13.90(cid:6)0.80 <14.10 <14.60 13.36(cid:6)0.07 <15.00 14.38 14.13 CH+ <12.80 <12.70 <13.00 <13.20 <13.30 13.06 CH <13.40 <13.60 <13.50 <13.40 <13.60 12.69 alogarithmof,incm−2;Dopplerparametersaretakentobeb=1.5,1.5,1.9,2.5,0.8kms−1forthefivecomponentsrespectively;thecomponents areassumedtocoverallimages; bWeltyetal.(1999); cThecontinuumisfittedlocally; dHowketal.(1999); eHobbs(1978).Asign“bl”meansthatnomeasurementispossibleduetoblendingeffects columndensitiesaresimilarexceptforN(Nai)whichismuch FromN(Hi)=N(Nai)(cid:2)(Na/Nai)/Z(Na)andusingthetwo smallertowardµCol.Thisillustratesdirectlythatatleastinthe upperlimitsontheNai/Naiiratiosderivedabove,wecanwrite componentatz(cid:24)1.1802towardAPM08579+5255whereNai N(Hi)>19.9and19.5/(Z(Na)/Z(cid:12)(Na))/δ(Na)atzabs=1.0626 isdetected,thegasismostlikelytobeneutralandcold.More- and1.1801respectively,whereδ(Na)isthefractionofsodium over,notealsothatinourGalaxy,aratioN(Nai)/N(Caii)>1, remaininginthegasphaseafterdepletionintodust-grains.This asobservedatzabs=1.1802towardAPM08579+5255,ischar- factor is equal to about 0.1 in the ISM (Savage & Sembach acteristicofcoldgasinthedisk.Indeed,alongthelineofsight 1996).Thisaddssupporttoargumentspresentedpreviouslythat to the LMC, such large ratios are observed only at the sys- thesesystemsaredamped.Toillustratethediscussion,simple temicvelocitiesoftheLMCandtheGalaxy;gasinbetweenhas photo-ionizationmodelsusingthecodeCloudy(Ferland1996) N(Nai)/N(Caii)<1(Vidal-Madjaretal.1987,Vladiloetal. have been constructed. The absorbing cloud is modelled as a 1993). planeparallelslabwithuniformdensity,solarchemicalcompo- FromtheupperlimitonCaiandFeiwecanderive,inthe sitionandneutralhydrogencolumndensity5(cid:2)1020cm−2.The componentswhereN(Caii)andN(Feii)aremeasured,alower elementsareconsideredtobedepletedintodust-grainsasinthe limitfortheelectronicdensityforagivenionizingfield.Indeed, coolcloudobservedtowardζOph(Savage&Sembach1996). ne=(Xo/X+)(cid:2)(Γ/α),whereΓisthephotoionizationrateand TheshapeoftheUVfluxistakentobeapower-lawFν/ν−1.0. αtherecombinationcoefficient.Thecorrespondingelectronic Theresultingcolumndensitiesofvariousspeciesalongaline- densitiesforaGalacticionizingfield(Pe´quignot&Aldrovandi of-sightperpendiculartotheslabaregivenversustheionizing 1986),arene<0.13and3cm−3forFeandCaatzabs=1.0626 parameterinFig.10.Asdiscussedabove,theN(Mgi)/N(Nai) and1.1801respectively.Notethatthedeterminationofneinthe ratio can be smaller than one only if magnesium is more de- interstellarmediumfromtheratioofsinglyionizedtoneutral pletedintodustgrainsthansodium.Notethateverymodelthat speciesishighlyuncertainprobablybecauseofcontamination producesenoughNaihastemperaturelessthan100K. ofthesinglyionizedcolumndensitydeterminationbyadjacent Finally, we do not detect any CHλ4300 and CH+λ4232 components(Weltyetal.1999).Writingthesamerelationfor absorption.ThelimitsonthecolumndensitiesarelogN(CH) sodiumandequatingtheexpressionofne obtainedforsodium <13.5andlogN(CH+)<13.0atbothzabs=1.06and1.18(see and iron or calcium leads to upper limits on the Na i/Na ii Table2).ThisisjustwhatwouldbeexpectedinourGalaxyalong ratiowhichdependsonlyontheshapeoftheionizingspectrum anotherwisesimilarlineofsight.Indeed,alongthelineofsight and not on its absolute value. With the only assumption that to23Ori,logN(CH+)=13.06andlogN(CH)=12.69(Welty theionizingspectrumhasthesameshapeasinourGalaxyand et al. 1999; see Table 2). More generally, the column density using the coefficients derived from Pe´quignot & Aldrovandi ofCHisobservedtoincreasefrom1.5to7.5(cid:2)1013 cm−2 for (1986;seeWeltyetal.1999),wefindlogNai/Naii<−1.3and linesofsightwithEB−V increasingfrom0.5to1.5(Gredelet 0.1 from the constraints obtained on Fe and Ca in the 1.0626 al.1993).Itwouldbeofprimeinteresttoobtainbetterdatain and1.1801systemsrespectively. thiswavelengthrangetobetterconstrainthemoleculecolumn densities. 466 P.Petitjeanetal.:StructureoftheMgiianddampedLyman-αsystemsalongthelineofsighttoAPM08279+5255 16 blendat(cid:1)v =−60kms−1 (seeFig.8).Forthecomponentat zabs=1.1801,weobtainlogN(Mgi)=13.8. Notethatinthiscase,theNaiandMgicolumndensitiesare nearlyidenticaltowhatisobservedtoward23Ori(seeTable2). We therefore conclude that N(Mg i)/N(Na i) (cid:24) 1 is a robust measurementinthissystem. 14 Mg i is the line with the largest saturation among those used to derive column densities quoted in Table 2. Therefore, the column densities derived from weaker lines should not differ fromwhatisquotedinTable2bymorethanafactoroftwo. 4.5. Metallicityanddustcontent 12 In the following we use the conventional definition [X/H] = log(Z/Z(cid:12)), with Z(X) the metallicity of species X relative to hydrogen. Ca ii and Ti ii are both detected at zabs = 1.0613 and 1.0631. The column densities are consistent with what is seenintheinterstellarmediumofourGalaxy(Stokes1978;see Table2).However,logN(Caii)/N(Tiii)(cid:24)−0.5and−0.2at, 10-4 -3.5 -3 -2.5 -2 respectively,zabs =1.0613and1.0631whentherelativesolar log Ionization parameter metallicityislogZ(cid:12)(Ca)−logZ(cid:12)(Ti)=1.38.Variousexpla- nationsforthisdiscrepancycanbeinvoked,amongstthemthe Fig.10.ResultsofphotoionizationmodelswithZ =Z(cid:12),logN(Hi) mostlikelyare:(i)CalciumismostlyintheformofCaiii,(ii) =20.7,constantdensityandplaneparallelgeometry.Depletioninto dust-grainsisassumedtobethesameasinthecoolcloudtowardζOph Calciumismoredepletedintodust-grainsthanTitanium.Note (Savage&Sembach1996).Columndensitiesaregivenversusioniza- that the relative metallicities [Ca/Fe] and [Ti/Fe] are both ob- tionparameter;theshapeoftheUVfluxisapower-lawFν /ν−1.0. servedtobe(cid:24)+0.3for[Fe/H]<−1inlate-typestars(The´venin 1998).NotealsothattheCaiii/CaiiratioderivedinourGalaxy isintherange5–10whichismuchsmallerthanthediscrepancy mentioned above. This favors the explanation that Calcium is 4.4. Consequenceofpartialcoveringfactor heavilydepletedintodust-grains(seebelow). FromthedetectionoftheNaiλλ3303.3,3303.9doublet,wecan In the zabs = 1.0631 component, Mn ii is also seen derivethatthetwocomponentsatzabs=1.0626and1.1801arise and log N(Mn ii)/N(Ti ii) (cid:24) +0.8. As the solar metal- incold,denseandneutralgas(seeSect.4.3).Itisthereforepos- licity of Mn is −6.47, the relative solar metallicity is siblethatthedimensionofthecloudislessthan(cid:24)1.9kpcwhich log Z(cid:12)(Mn) − log Z(cid:12)(Ti) = = +0.6. This is consistent with istheseparationofthelinesofsighttothetwobrightestimages similardepletionofMnandTiasobservedinwarmgas(Sav- attheredshiftoftheabsorberassumingthatthelensingobject age&Sembach1996).Therefore,inthiscomponent,wecannot isatthesameredshift.Indeed,largevariationsofNaicolumn ruleoutthatthelowCaiicolumndensityisduetoionization. densityhavebeenreportedinthenearbyinterstellarmediumon In the zabs = 1.0613 component, Fe ii is also seen with verysmallscales(Meyer&Lauroesch1999).Wehavetherefore log N(Fe ii)/N(Ti ii) (cid:24) +2.1. The solar metallicity of iron investigatedtheimpactonthecolumndensitymeasurementsof and titanium are, respectively, −4.49 and −7.07 and the rel- the assumption that the cloud covers the three images. If it is ative solar metallicity is log Z(cid:12)(Fe) − log Z(cid:12)(Ti) = +2.58. the case that the cloud covers only one of the images the col- Thereisnodifferentialionizationcorrectionforthesetwoele- umn density is larger. The discrepancy cannot be very large, ments.Thediscrepancy,(cid:24)+0.5dex,betweenthetworatioscan however,asmostofthelinesusedforcolumndensitydetermi- only be explained by a larger depletion of titanium compared nationareweak.Thestronglinesarecompletelysaturatedand to iron into dust-grains by (cid:24)0.5 dex as in the cool gas of the blended, which prevents in any case any determination of the ISM(Savage&Sembach1996).Indeed,fromnucleosynthesis lineparameters. alone, we would expect titanium to be enhanced compared to WehaveconsideredtheMgiandNailinesinthezabs=1.1801 iron (The´venin 1998) contrary to what is observed. We there- component. It can be seen on Fig.8 that if only one image is fore conclude that depletion into dust-grains is present in this covered, it cannot be the brightest as the residual normalized system.ThelowCaiicolumndensitycanindeedbeexplained fluxintheMgiabsorptionissmallerthan0.5.Asthefluxratio byalargedepletionofcalciumintodust-grainsasisobserved ofthetwobrightestcomponentsis1.2,weartificiallyplacedthe intheISM. zerolevelat0.4onthescaleofFig.8.Voigtprofilefittingofthe The fact that Ti ii is not detected in the zabs = 1.18 system NaidoubletgiveslogN(Nai)=13.7andb=1kms−1which is surprising, although the limit on the column density is not is within a factor of two of what has been derived previously stringentandonlyafactoroffoursmallerthanwhatisseenin (seeTable2).FourcomponentshavebeenusedtofittheMgi thezabs=1.06components.
Description: