COSMOGRAIL: the COSmological MOnitoring of GRAvItational Lenses III. Redshift of the lensing galaxy in eight gravitationally lensed quasars

0.42 MB·English

by  A. Eigenbrod

Astronomy&Astrophysicsmanuscriptno.4454 (cid:13)c ESO2008
February5,2008
COSMOGRAIL: the COSmological MOnitoring of
⋆
GRAvItational Lenses III.
6
0 Redshift of the lensing galaxy in eight gravitationally lensed quasars
0
2
A.Eigenbrod1,F.Courbin1,G.Meylan1,C.Vuissoz1,andP.Magain2
n
a
J
1 Laboratoired’Astrophysique,EcolePolytechniqueFe´de´raledeLausanne(EPFL),Observatoire,1290Sauverny,Switzerland
9 2 Institutd’AstrophysiqueetdeGe´ophysique,Universite´deLie`ge,Alle´edu6aouˆt17,Sart-Tilman,Bat.B5C,B-4000Lie`ge,Belgium

2
Received...;accepted...
v
6
2 ABSTRACT
0
1 Aims.WemeasuretheredshiftofthelensinggalaxyineightgravitationallylensedquasarsinviewofdeterminingtheHubbleparameterH
0
1 fromthetimedelaymethod.
5 Methods.DeepVLT/FORS1spectraoflensedquasarsarespatiallydeconvolvedinordertoseparatethespectrumofthelensinggalaxiesfrom
0
the glare of the much brighter quasar images. A new observing strategy is devised. It involves observations in Multi-Object-Spectroscopy
/
h (MOS)whichallowsthesimultaneousobservationofthetargetandofseveralPSFandfluxcalibrationstars.Theadvantageofthismethod
p overtraditionallong-slitobservationsisamuchmorereliableextractionandfluxcalibrationofthespectra.
- Results.Forthefirsttimewemeasuretheredshiftofthelensinggalaxyinthreemultiply-imagedquasars:SDSSJ1138+0314(z =0.445),
o lens
SDSSJ1226−0006(z = 0.517),SDSSJ1335+0118(z = 0.440),andwegiveatentativeestimateoftheredshiftofthelensinggalaxyin
r lens lens
st Q1355−2257 (zlens= 0.701).Weconfirmfourpreviouslymeasuredredshifts:HE0047−1756 (zlens= 0.407), HE0230−2130 (zlens= 0.523),
HE 0435−1223 (z = 0.454) and WFI J2033−4723 (z = 0.661). In addition, we determine the redshift of the second lensing galaxy in
a lens lens
: HE 0230−2130 (zlens= 0.526). The spectra of all lens galaxies are typical for early-type galaxies, except for the second lensing galaxy in
v
HE0230−2130whichdisplaysprominent[OII]emission.
i
X
r
a Key words. Gravitational lensing: time delay, quasar, microlensing — Cosmology: cosmological parameters, Hubble constant. Quasars:
general. Quasars:individual HE0047−1756, HE0230−2130, HE0435−1223, SDSSJ1138+0314, SDSSJ1226−0006, SDSSJ1335+0118,
Q1355−2257,WFIJ2033−4723.
1. Introduction through the determination of detailed mass maps for lensing
galaxies,especiallytheirdarkmattercontent.
Gravitationally lensed quasars have become a truly efficient
The second most important application of quasar lens-
source of new astrophysical applications, since the discovery
ing involves microlensing by stars in the lensing galaxy.
of the first case by Walsh et al. (1979). These objects are po-
Microlenses produce chromatic magnification events in the
tentially useful for measuring the Hubble parameter H from
0 lightcurveoftheimagesofthesource,astheycrosstheirline
thetimedelaybetweentheirlensedimages(Refsdal1964),as-
of sight. The amplitude, durationand frequencyof the events
sumingamodelforthemassdistributioninthelensinggalaxy.
dependonthetransversevelocityofstarsinthelensinggalaxy,
Conversely,foranassumedormeasuredvalueofH ,themass
0 on their surface density, and on the relative sizes of the mi-
distributioninthelenscanbereconstructedfromthetimede-
crolensing caustics with respect to the regions of the source
lay measurement. The study of lensed quasars is therefore a
affected by microlensing. Photometric monitoring of lensed
“no-lose” game, either because of its cosmological implica-
quasars in several bands or, better, spectrophotometric moni-
tions (H ) or for the study of galaxy formationand evolution
0 toring can therefore yield constraints on the energy profile of
quasaraccretiondisksandonthesize ofthevariousemission
⋆ Based on observations made withthe ESO-VLTUnit Telescope
lineregions(e.g.Agol&Krolik1999,Mineshige&Yonehara
2 Kueyen (Cerro Paranal, Chile; Proposals 074.A-0563 and 075.A-
1999,Abajasetal.2002).
0377, PI: G. Meylan). and on archive data taken with the ESO-
VLTUnitTelescope1Antu(CerroParanal,Chile;Proposals064.O- All the above applications of quasar lensing require the
0259(A),PI:L.Wisotzki) knowledgeof the redshift of the lensing galaxy,which is fre-
2 A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars
Table1.JournaloftheobservationsofHE0047−1756.Grism Table 5. Journal of the observations for SDSS J1226−0006.
and filter: G300V+GG435. HR collimator: 0.1′′ per pixel. Grism and filter: G300V+GG435. SR collimator: 0.2′′ per
Slitletswidth:1.0′′(R=210at5900Å). pixel.Slitletswidth:1.0′′(R=400at5900Å).
ID Date Seeing[′′] Airmass Weather ID Date Seeing[′′] Airmass Weather
1 18/07/2005 0.49 1.281 Photometric 1 16/05/2005 0.85 1.109 Photometric
2 18/07/2005 0.53 1.191 Photometric 2 16/05/2005 0.84 1.099 Photometric
3 16/05/2005 0.92 1.105 Photometric
4 16/05/2005 1.02 1.125 Photometric
5 16/05/2005 0.78 1.221 Photometric
Table2.JournaloftheobservationsofHE0230−2130.Grism
6 16/05/2005 0.89 1.299 Photometric
and filter: G600R+GG435. SR collimator: 0.2′′ per pixel.
7 16/05/2005 0.82 1.422 Photometric
Slitletswidth:0.5′′(R=1910at6200Å). 8 16/05/2005 0.88 1.576 Photometric
ID Date Seeing[′′] Airmass Weather
1 15/12/2004 0.60 1.166 Lightclouds Table 6. Journal of the observations for SDSS J1335+0118.
2 15/12/2004 0.58 1.248 Lightclouds Grism and filter: G300V+GG435. HR collimator: 0.1′′ per
3 01/03/2005 0.78 1.800 Photometric pixel.Slitletswidth:1.0′′(R=210at5900Å).
ID Date Seeing[′′] Airmass Weather
Table3.JournaloftheobservationsforHE0435−1223.Grism
1 03/02/2005 0.73 1.167 Photometric
and filter: G300V+GG435. HR collimator: 0.1′′ per pixel.
2 03/02/2005 0.71 1.133 Photometric
Slitletswidth:1.0′′(R=210at5900Å).
3 03/03/2005 0.69 1.112 Photometric
4 03/03/2005 0.77 1.123 Photometric
5 03/03/2005 0.68 1.153 Photometric
ID Date Seeing[′′] Airmass Weather 6 03/03/2005 0.62 1.198 Photometric
1 11/10/2004 0.47 1.024 Photometric
2 11/10/2004 0.45 1.028 Photometric
3 12/10/2004 0.46 1.024 Photometric
4 12/10/2004 0.53 1.028 Photometric Table 7.JournaloftheobservationsforQ1355−2257.Grism
5 11/11/2004 0.57 1.093 Photometric and filter: G300V+GG435. HR collimator: 0.1′′ per pixel.
6 11/11/2004 0.56 1.145 Photometric Slitletswidth:1.0′′(R=210at5900Å).
ID Date Seeing[′′] Airmass Weather
Table 4. Journal of the observations for SDSS J1138+0314.
1 05/03/2005 0.68 1.016 Photometric
Grism and filter: G300V+GG435. HR collimator: 0.1′′ per
2 05/03/2005 0.73 1.040 Photometric
pixel.Slitletswidth:1.0′′(R=210at5900Å).
3 20/03/2005 0.63 1.038 Photometric
4 20/03/2005 0.54 1.015 Photometric
5 20/03/2005 0.57 1.105 Photometric
ID Date Seeing[′′] Airmass Weather
6 20/03/2005 0.56 1.166 Photometric
1 10/05/2005 0.82 1.191 Photometric
2 11/05/2005 0.70 1.155 Photometric
3 11/05/2005 0.67 1.133 Photometric
4 11/05/2005 0.66 1.132 Photometric Table 8.JournalofobservationsforWFI J2033−4723.Grism
5 11/05/2005 0.64 1.148 Photometric and filter: G300V+GG435. HR collimator: 0.1′′ per pixel.
Slitletswidth:1.4′′(R=160at5900Å).
ID Date Seeing[′′] Airmass Weather
quentlyhiddenin the glare of the quasar images. The present 1 13/05/2005 0.50 1.256 Lightclouds
paperispartofalargerefforttocarryoutlongtermphotometric 2 13/05/2005 0.58 1.198 Lightclouds
3 13/05/2005 0.60 1.148 Lightclouds
monitoringoflensedquasarsinthecontextofCOSMOGRAIL
4 13/05/2005 0.48 1.117 Lightclouds
(e.g. Eigenbrod et al. 2005a). In this paper we focus on the
5 13/05/2005 0.53 1.095 Lightclouds
determination of the redshifts of the lensing galaxies in sev-
eralgravitationallylensedquasars,usingdeepspectraobtained
with the ESO Very Large Telescope (VLT). The targets were
simplyselectedinfunctionoftheirvisibilityduringtheperiod
ofobservation.
A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars 3
Epoch 1 Epoch 1 Epoch 3 Epoch 5
A
A
B
G
D G
B C
Fig.1.HE0047−1756.Slitwidth:1.0′′.MaskPA:10◦
Fig.4.SDSSJ1138+0314.Slitwidth:1.0′′.MaskPA:−84◦
Epoch 1 Epoch 3 Epoch 1 Epoch 2 Epoch 4 Epoch 6 Epoch 8
C B
G2
B G
D
A
G1
A
Fig.2.HE0230−2130.Slitwidth:0.5′′.MaskPA:−60◦ Fig.5.SDSSJ1226−0006.Slitwidth:1.0′′.MaskPA:−90◦
Epoch 1 Epoch 3 Epoch 5 Most of our targets are observed with the high-resolution
collimator, allowing us to observe simultaneously eight ob-
D
G jects over a field of view of 3.4′ × 3.4′ with a pixel scale of
0.1′′. However, because few suitable PSF stars are visible in
thevicinityofSDSSJ1226−0006andHE0230−2130,theob-
C A servationsforthesetwoobjectsusethestandard-resolutioncol-
B
limator,whichhasafieldofviewof6.8′×6.8′andapixelsize
of0.2′′.
WeusetheGG435ordersortingfilterincombinationwith
the G300V grism for all objects, except HE 0230−2130 for
Fig.3.HE0435−1223.Slitwidth:1.0′′.MaskPA:−164◦
whichweusetheG600Rgrism.TheG300Vgrismgivesause-
fulwavelengthrange4450 <λ< 8650Åandascaleof2.69Å
perpixelinthespectraldirection.Thissetuphasaspectralres-
olutionR=λ/∆λ≃200atthecentralwavelengthλ= 5900Å
2. VLTSpectroscopy for a 1.0′′ slit width in the case of the high resolution colli-
mator.Thechoiceofthisgrismfavorsspectralcoveragerather
2.1.Observations
thanspectralresolutionas theobservationsare aimedatmea-
We present observations for eight gravitationally lensed suringunknownlensredshifts.ThecombinationoftheG600R
quasars, in order to determine the redshift of the lensing grism with the GG435 filter used for HE 0230−2130 has a
galaxy. The targets are HE 0047−1756, HE 0230−2130, wavelengthrangeof5250 <λ< 7450Åwithapixelscaleof
HE 0435−1223, SDSS J1138+0314, SDSS J1335+0118, 1.08Åinthespectraldirection.Thisgivesahigherspectralres-
Q 1355−2257(also known as CTQ 327), SDSS J1226−0006 olutionofR ≃ 1200atthecentralwavelengthλ = 6270Åfor
andWFIJ2033−4723. aslitwidthof1.0′′andwiththestandardresolutioncollimator.
OurspectroscopicobservationsareacquiredwiththeFOcal We choose slitlets of various widths, depending on the
Reducer and low dispersion Spectrograph(FORS1), mounted brightnessofthetargetandontheconfigurationofthequasar
on the ESO Very Large Telescope. Very importantly, all the images. The slit width is chosen so that it matches the see-
observations are carried out in the MOS mode (Multi Object ingrequestedfortheservice-modeobservationsandminimizes
Spectroscopy).Thisstrategyallowsthesimultaneousobserva- lateral contamination by the quasar images. Our observing
tionofthemaintargetandofseveralstarsusedbothasfluxcal- sequences consist of a short acquisition image, an “image-
ibratorsandasreferencePSFtospatiallydeconvolvethedata. through-slit”check,followedby oneor two consecutivedeep
4 A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars
on18October1999forprogram064.O-0259(A).Theyusedthe
longslitmodewiththestandardresolutioncollimator(0.2′′per
Epoch 1 Epoch 3 Epoch 5
pixel), the G600R grism, and the order sorting filter GG435.
A
TheusefulwavelengthrangeisthesameasfortheMOSobser-
vations,i.e.,5250 < λ< 7450Åwithapixelscaleof1.08Å
G
inthespectraldirection.
B
2.2.ReductionandDeconvolution
We carry out the standard bias subtraction and flat field cor-
rectionofthespectrausingIRAF1.Theflatfieldiscreatedfor
Fig.6.SDSSJ1335+0118.Slitwidth:1.0′′.MaskPA:43◦
each slitlet fromfive dome exposuresusing cosmic ray rejec-
tion.Itisthennormalizedbyaveraging45linesalongthespa-
tialdirection,rejectingthe20highestand20lowestpixels.The
resultisthenblockreplicatedtomatchthephysicalsizeofthe
Epoch 1 Epoch 3 Epoch 5
individualflatfields.
A Wavelength calibration is obtained from numerous emis-
sionlinesinthespectrumofHelium-Argonlamps.Thewave-
length solution is fitted in two dimensions to each slitlet of
G
the MOS mask. The fit uses a fifth order Chebyshev polyno-
B
mialalongthespectraldirectionandafourthorderChebyshev
polynomialfitalongthespatialdirection.Eachspectrumisin-
terpolated following this fit using a cubic interpolation. This
procedureensures that the sky lines are well aligned with the
Fig.7.Q1355−2257.Slitwidth:1.0′′.MaskPA:−78◦ columnsoftheCCDafterwavelengthcalibration.
The sky background is then removed by fitting and sub-
tracting a second order Chebyshev polynomial in the spatial
directionto the areas of the spectrumthat are not illuminated
Epoch 1 Epoch 3 Epoch 5 bytheobject.
B Finally, we remove the cosmic rays as follows. First we
shift the spectra in order to align them spatially (this shift is
A2
G only a few tenths of a pixel). Second, we create a combined
spectrum for each object from all exposures, removing the
lower and higher pixels, after applying appropriate flux scal-
C A1
ing. The combined spectrum obtained in that way is cosmic
raycleanedandisusedasareferencetemplatetocleanthein-
dividualspectra.ForeachMOSmasktheobjectslitletandthe
PSFslitletsarereducedexactlyinthesameway.
Fig.8.WFIJ2033−4723.Slitwidth:1.4′′.MaskPA:−82◦
Afluxcross-calibrationofthespectratakenatdifferentair-
massesordatesisneededbeforecombiningthemintoonefinal
spectrum.ThisisdoneefficientlyusingthespectraofthePSF
stars as references, as described in Eigenbrod et al. (2005b).
spectroscopic exposures. All individual exposures for all ob-
Thereferencestarsareassumedtobenon-variableandaratio
jects are 1400 s long. The journals of the observations are
spectrum is created for each star, i.e. we divide the spectrum
given in Tables 1 to 8. The through-slitimages are displayed
ofthestarbythespectrumofthesamestarintheotherexpo-
inFigs.1to8,wheretheepochsrefertotheexposurenumbers
sures. Thisis doneforat leasttwo stars in eachmask andwe
inTables1to8.
checkthattheresponsecurvesderivedusingdifferentstarsare
For every object we center at least three slitlets on fore-
compatible. The dispersion between the response curves ob-
groundstarsandoneslitletonthelensinggalaxy.Themaskis
tainedinthatwayisabout2%.Ameancorrectioncurveisthen
rotatedto a Position Angle thatavoidsclippingof anyquasar
computedandappliedtoeachtwo-dimensionalspectrum.The
image.Thisismandatorytocarryoutspatialdeconvolutionof
individualspectraforeachobjectarefinallycombined.
thespectra.ThespectraofthePSFstarsarealsousedtocross-
The archive data of HE 0230−2130 consist of one sin-
calibrate the flux scale as the data are taken at different air-
glelongslitspectrum.Thebiassubtraction,flatfielding,wave-
masses and at different dates. Three spectrophotometricstan-
dardstarsareusedtocarryouttherelativefluxcalibration,i.e.
1 IRAF is distributed by the National Optical Astronomy
EG21,LTT3218,andLTT6248.
Observatories,whichareoperatedbytheAssociationofUniversities
In additionto ourowndata,we retrieveVLT/FORS1data for Research in Astronomy, Inc., under cooperative agreement with
fromthearchiveforHE0230−2130.Thesedatawereacquired theNationalScienceFoundation.
A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars 5
Table9.Redshiftvaluesdeterminedforthelensinggalaxiesin
theeightgravitationallenses.Onlyatentativeredshiftisgiven
forQ1355−2257.SeeSection4formoredetailsaboutthesec-
ondlensG2inHE0230−2130.
Object z
lens
HE0047−1756 0.407±0.001
HE0230−2130,G1 0.523±0.001
HE0230−2130,G2 0.526±0.002
HE0435−1223 0.454±0.001
SDSSJ1138+0314 0.445±0.001
SDSSJ1226−0006 0.517±0.001
SDSSJ1335+0118 0.440±0.001
Q1355−2257 0.701(?)
WFIJ2033−4723 0.661±0.001
lengthcalibrationandbackgroundremovalaredoneinexactly
thesamewayasfortheMOSspectra.Thecosmicrayremoval
is done using the IRAF packages for single-image data. The
flux calibration is done using three standard stars: G93−48, Fig.9.SpectrumofthelensinHE0047−1756.Thetotalinte-
LTT 7987, LTT 9239,taken on the same night as the science gration time is 2800s. The template spectrum of a redshifted
frame. ellipticalgalaxyisshownforcomparison(Kinneyetal.1996).
2.3.DeconvolutionandextractionoftheMOSspectra
Even thoughthe seeing values are good for most spectra, the pointsource(quasarimages)channels.OnlyourVLTspectrum
lensinggalaxyisoftencloseenoughtothebrighterquasarim- ofthelensinggalaxyofHE0230−2130issufferingfromresid-
agestobeaffectedbysignificantcontaminationfromthewings ualsofthequasarbroademissionlines,probablyduetolateral
of the PSF. For this reason, the spectral version of the MCS contamination by images A and B of the quasar. This addi-
deconvolution algorithm (Magain et al. 1998, Courbin et al. tionalsourceofcontaminationiscircumventedbysubtracting
2000) is used in orderto separate the spectrumof the lensing ascaledversionofthespectrumofquasarimageCtothespec-
galaxyfromthespectraofthequasarimages.TheMCSalgo- trum ofthe lensing galaxy.Theflux calibrationof thispartic-
rithmusesthespatialinformationcontainedinthespectrumof ular lens might therefore be less accurate than for the other
areferencePSF,whichisobtainedfromtheslitletspositioned objects.Note,however,thatthisprocedurewasonlyappliedto
ontheisolatedstars.ThefinalnormalizedPSFisacombination HE0230−2130.
ofatleastthreedifferentPSFspectra.Thedeconvolvedspectra
arenotonlysharpenedinthespatialdirection,butalsodecom-
3. Notesonindividualobjects
posed into a “point-sourcechannel” containing the spectra of
the quasar images, and a “extended channel” containing the HE 0047−1756: A doubly imaged quasar discovered by
spectraofeverythingintheimagewhichisnotapoint-source, Wisotzkietal.(2004).Ithasaredshiftofz = 1.67andamax-
i.e.inthiscasethespectrumofthelensinggalaxy. imum image separation of 1.44′′. The redshift of the lensing
The deconvolved spectra of the lensing galaxies are ex- galaxy has recently been measured by Ofek et al. (2005) at
tracted and smoothed with a 10 Å box. Fig. 9 to 16 display z =0.408.Weconfirmthisresult,withz =0.407±0.001,
lens lens
the finalone-dimensionalspectra,where theCa IIH &K ab- and presenta muchhighersignal-to-noisespectrumin Fig. 9.
sorptionlinesareobvious,aswellasthe4000Å Balmerbreak, Anellipticalgalaxytemplatematcheswellthespectrumofthe
and the G-bandtypicalfor CH absorption.In some cases, we lens.
identifya few more featuresthatare labeled in the individual HE0230−2130:Thisquadruplyimagedz = 2.162quasar
figures.Theidentifiedlinesare usedto determinethe redshift was discovered by Wisotzki et al. (1999). It has a maximum
of the lensing galaxiesgivenin Table 9. We computethe 1-σ imageseparationof2.15′′andtwolensinggalaxies.Themain
errorasthestandarddeviationbetweenallthemeasurementsof lensinggalaxy,G1,islocatedbetweenthefourquasarimages.
theindividuallines.Theabsenceofemissionlinesinallspectra Asecond,fainterlensislocatedoutsidetheareadefinedbythe
indicatesthattheobservedlensinggalaxiesaregas-poorearly- quasarimages,closetothefaintquasarimageD(Fig.17).The
typegalaxies. spectrumofG1isshowninFig.10.Ourredshift(z =0.523±
lens
In mostcases, notrace ofthe quasar broademission lines 0.001) is in very good agreementwith the result z = 0.522
lens
is seen in the spectrum of the lensinggalaxy,indicativeof an of Ofek et al. (2005). The spectrum matches well that of an
accuratedecompositionofthedataintotheextended(lens)and early-typegalaxy.The spectrum of the second lensing galaxy
6 A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars
Fig.10.SpectrumofthelensinggalaxyG1inHE0230−2130, Fig.12. Spectrum of the lens in SDSS J1138+0314. The to-
asobtainedbycombiningthedataforthethreeepochs,i.e.ato- tal integration time is 7000 s. The absorption feature marked
talintegrationtimeof4200s.Ascaledversionofthespectrum by the black triangle is probably residual light of the quasar
ofquasarimageCissubtractedtothespectrumofthelensing images. The C III] emission of the quasar falls exactlyat this
galaxy,inordertoremovelateralcontaminationbythequasar wavelength.
imagesAandB(seetext).
Fig.13.Spectrumofthe lensinSDSS J1226−0006.Thetotal
Fig.11.Spectrumof the lensin HE 0435−1223.Thetotalin- integrationtimeis11200s.
tegrationtimeis8400s.Thetemplatespectrumofaredshifted
S0galaxyisshown(Kinneyetal.1996).
HE0435−1223:Thisquadruplyimagedquasar,discovered
byWisotzkietal.(2002),hasaredshiftofz=1.689andhasa
G2 is extracted from the archive long-slit spectrum presented maximumimageseparationof2.56′′,hencewithlittlecontam-
inSection4. inationofthelensspectrumbythequasarimages.Theredshift
A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars 7
Fig.14.Spectrumofthe lensinSDSS J1335+0118.Thetotal Fig.16. Spectrum of the lens in WFI J2033−4723. The total
integrationtimeis8400s. integrationtimeis7000s.
SDSS J1138+0314: This quadruply imaged quasar was
discovered in the course of the Sloan Digital Sky Survey
(SDSS)byBurlesetal.(2005).Thislensedquasarhasamaxi-
mumimageseparationof1.46′′.Weobtainhighsignal-to-noise
spectra of the quasar and determine its redshift z = 2.438.
Thereisverylittledoubtthatthissystemislensed.Thelensing
galaxyisseenonarchivalHST/NICMOSimages.Wemeasure
z =0.445±0.001(Fig.12).
lens
SDSSJ1226−0006:Adoublyimagedquasaratz = 1.120
foundinthecourseoftheSloanDigitalSkySurvey(SDSS)by
Inada et al. (2005). This system is doubtlessly lensed, as the
lensinggalaxyisseenonarchivalHST/ACSimages,between
bothquasarimages,atonly0.4′′awayfromimageA.Wemea-
surez =0.516±0.001(Fig.13).Thespectrumofthelensing
lens
galaxyiswellmatchedbythespectrumofanellipticalgalaxy.
SDSSJ1335+0118:Adoublyimagedquasarwitha1.56′′
separation,discoveredbyOgurietal.(2004).Thequasarisat
z = 1.57and the spectra of the quasar imagesshow evidence
of strong absorption systems at lower redshifts. Based on the
color of the galaxy,Oguriet al. (2004) concludethat the lens
Fig.15. Spectrum of the lens in Q 1355−2257. The total in- galaxyis consistentwith anearly-typegalaxyatz <∼ 0.5.Our
tegrationtime is 8400s. Giventhe low signal-to-noiseof this spectrum (Fig. 14) is indeed that of an early-type galaxy. Its
spectrumwecannotsecurelydeterminethelensredshift. redshiftzlens=0.440±0.001hasnotbeenmeasuredbefore.
Q1355−2257(CTQ327):Thisdoublyimagedquasarhas
a redshift of z = 1.373 and was discovered by Morgan et al.
(2003).Thequasarimagesareseparatedby1.23′′andthered-
shift of the lensing galaxieshasbeen estimated by Morganet
ofthelensinggalaxyhasalreadybeenmeasured(Morganetal. al. (2003) to lie in the redshiftrange 0.4 < z < 0.6. The
lens
2005,Ofeketal.2005)atz = 0.455.Weconfirmthisresult, extractionofthespectrumofthelensinggalaxyisparticularly
lens
withz =0.454±0.001,andpresentamuchhighersignal-to- difficultbecausethegalaxyliesatonly0.29′′awayfromquasar
lens
noisespectruminFig.11.AS0galaxytemplatematcheswell image B and because it is 7 magnitudesfainter. We show the
thespectrumofthelens. deconvolvedspectrumofthelensinggalaxyin Fig.15.Given
8 A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars
G2 0.7" 0.5"
C
D
G1
B N
A
1"
E
Fig.17. Right: HST image of HE 0230−2130 taken with the
WFPC2 instrument in the F814W filter. The pixel scale is
0.05′′.Middle:sameimagebutwithapositionangleof−160o.
We show in overlay a 0.7′′ slit which corresponds to the ob-
servationalsetupusedtotakethelong-slitspectrum(program
064.O-0259(A)archive data). Left: for comparison, we show
the slit used for our MOS observations. The position angle
is −60o and the slit has a width of 0.5′′. Negative angles are
countedclockwisefromNorth.
5600 A 5750 A
Fig.19. Spectrum of the second lensing galaxy G2 in
HE0230−2130,obtainedfromtheFORS1long-slitspectrum.
2’’ The exposure time is 3000 s. Note the prominent[OII] emis-
[OII] sionlines,absentfromthespectrumofG1(Fig.10).
λ
Fig.18. Spectra of quasar image B and D of HE 0230−2130
(slitasinthemiddlepanelofFig.17.Astrongemissionfeature
isseen“below”thequasarimageD.Itextendswellbeyondthe spectrum from the brighter quasar image B directly. During
area delimited by the quasar images, and corresponds to the this first step, the PSF is constructed from the upper half of
[OII]emissionofgalaxyG2. the spectrum in Fig. 18 containing the spectrum of image B.
We then deconvolve the original data, we take the extended-
channel part of the deconvolution, reconvolve it with the ap-
thelowsignal-to-noiseofthisspectrumwecannotsecurelyde- proximatePSF andsubtractitfromthedata.Thisprovidesus
terminethelensredshift;wecanonlygiveatentativeestimate withaspectrumofthequasarimagethatismuchlessaffected
ofz =0.701. bythelensinggalaxy.AsecondPSFisconstructedfromthese
lens
WFI J2033−4723: Morgan et al. (2004) discovered this new lens-cleaned spectra and a new deconvolution is carried
quadruply lensed quasar with maximum image separation of out. We perform this cycle twice to obtain the final spectrum
2.53′′ and redshiftz = 1.66.Ofek et al. (2005) recently mea- presentedinFig.19.Thisspectrumhasalowersignal-to-noise
sured the lens redshiftz = 0.658. The lensing galaxyspec- than ourMOSspectra andless accuratefluxcalibrationbutit
lens
trumshowninFig.16ismatchedbyanellipticalorS0galaxy allowstomeasurewelltheredshiftoflensG2,whichdisplays
templatewitharedshiftofz =0.661±0.001.Ourmeasured prominent[OII] emission, contraryto lens G1. Fig. 18 shows
lens
redshiftis compatiblewith butslighlyhigherthanthe one re- thattheemissioncomesfromanobjectlyingoutsidetheregion
portedbyOfeketal.(2005). delimitedbythequasarimage.Itthereforeclearlycorresponds
tolensG2.
Theredshiftof lensG2 is z = 0.526±0.002.Itis mea-
4. ThesecondlensinHE0230−2130 lens
sured using the [OII] emission, the Ca II H & K absorption
The long-slit archive FORS1 spectrum of HE 0230−2130 is lines, the G-band, and the hydrogen Hθ and Hη absorption
taken with the slit centered on the quasar images B and D lines. The deconvolved spectrum of G2 is shown in Fig. 19.
(middlepanelinFig.17).Thetwo-dimensionalsky-subtracted Although the flux calibration is not optimal without a good
spectrumisshowninFig.18,whereahintofanemissionline knowledgeofthePSF,thespectrumresemblesthatofaSaspi-
canalreadybeseenatthespatiallocationoflensG2alongthe ralgalaxy.
slit. As can be seen in the middle panel of Fig. 17, G2 is not
AsnoPSFstarisavailablethedatahavetobedeconvolved wellcenteredintheslitbutliesat0.4′′awayfromtheslitcen-
inadifferentwaythantheMOSspectra.Weproceedinaniter- ter. This small spatial misalignment of the object within the
ativeway.First,wecutthespectrumintwoalongthespectral slitmimicsaspectralshiftof∼ 2Åtothered.Thistranslates
direction and determine a first estimate of the reference PSF intoaredshiftchangeoflessthan∆z = 0.0004at6000Åand
A.Eigenbrodetal.:COSMOGRAILIII:Redshiftofthelensinggalaxyineightgravitationallylensedquasars 9
has no effect on the redshift determination of galaxy G2. We Morgan,N.D.,Gregg,M.D.,Wisotzki,L.,etal.2004,AJ,126,696
conclude that the observed difference in redshift ∆z = 0.003 Morgan, N.D.,Cardwell, J.A.R.,Schechter, P.L.,etal. 2004, AJ,
betweengalaxyG1andG2isreal.Ittranslatesintoavelocity 127,2617
differenceof∆v=900±450kms−1typicalforagalaxygroup. Morgan, N. D., Kochanek, C. S., Pevunova, O., & Schechter, P. L.
HE0230−2130mightthereforebelensedbyagroupofphysi- 2005,AJ,129,2531
Ofek,E.O.,Maoz, D.,Rix,H.-W.,Kochanek, C.S.,&Falco,E.E.
callyrelatedgalaxiesofwhichG1andG2aretwoofthemain
2005,astro-ph/0510465
members.
Oguri,M.,Inada,N.,Castander,F.J.,etal.2004,PASJ,56,399
Refsdal,S.1964,MNRAS,128,307
5. SummaryandConclusions Saha,P.,etal.2005,submittedtoA&A
Walsh,D.,Carswell,R.F.,&Weymann,R.J.1979,Nature,279,381
Wepresentherethepreviouslyunknownredshiftsofthelens- Wisotzki,L.,Christlieb,N.,Liu,M.C.,etal.1999,A&A,348,L41
ing galaxies in three gravitationally lensed quasars and con- Wisotzki,L.,Schechter,P.L.,Bradt,H.V.,Heinmu¨ller,J.,&Reimers,
firm four others, already presented in Ofek et al. (2005). D.etal.2002,A&A,395,17
We also measure the redshift of a second lensing galaxy in Wisotzki, L., Schechter, P. L.,Chen, H.-W.,et al. 2004, A&A, 419,
HE 0230−2130and give a tentative estimate of the lens red- L31
shiftinQ1355−2257.
TheMOSmodeinwhichallobservationsaretakenandthe
subsequentobservationofseveralPSFstarsiscrucialtocarry
outareliabledecontaminationofthelensspectrumbythoseof
the quasar images. The PSF stars are also used to carry out a
veryaccuratefluxcalibrationofthespectra.
Contrary to long-slit observations where no PSF stars are
available (Ofek et al. 2005), we do not need to iteratively re-
move a scaled version of the quasar spectra from the data.
Because microlensing can produce significant differences be-
tweenthespectraofthequasarimages,suchaproceduremay
resultinbiasedcontinuumslopesandwrongconclusionsabout
the presenceof dustin the lens. For thisreasonwe fullybase
our extraction on a spatial decomposition using independent
PSFspectra.
Wefindthatallthelensinggalaxiesinoursampleareearly-
type ellipticals or S0, exceptfor the second lensing galaxyin
HE0230−2130,whichdisplaysprominent[OII]emission.And
we do notfind anyevidenceforsignificantextinctionbydust
intheirinterstellarmedium.
Acknowledgements. TheauthorsareverygratefultotheESOstaffat
Paranalfortheparticularcarepaidtotheslitalignmentnecessaryto
perform the spectra deconvolutions. PMacknowledge financial sup-
port from PRODEX (Belgium). COSMOGRAIL is financially sup-
portedbytheSwissNationalScienceFoundation(SNSF).Theimage
shown in Fig. 17 was obtained with the NASA/ESA Hubble Space
Telescope(Program#9744,PI:C.S.Kochanek) andextractedfrom
thedataarchivesattheSpaceTelescopeScienceInstitute,whichisop-
eratedbytheAssociationofUniversitiesforResearchinAstronomy,
Inc.,underNASAcontractNAS5-26555.
References
Abajas,C.,Mediavilla,E.,Mun˜oz,J.A.,Popovic´,L.Cˇ.,&Oscoz,A.
2002,ApJ,576,640
Agol,E.,&Krolik,J.1999,ApJ,524,49
Burles,S.,etal.2005,inpreparation
Courbin,F.,Magain,P.,Kirkove,M.,&Sohy,S.2000,ApJ,529,1136
Eigenbrod,A.,Courbin,F.,Vuissoz,C.,etal.2005a,A&A,436,25
Eigenbrod,A.,Courbin,F.,Dye,S.,etal.2005b,astro-ph/0510641
Inada,N.,etal.2005,inpreparation
Kinney,A.L.,Calzetti,D.,Bohlin,R.C.,etal.1996,ApJ,467,38
Magain,P.,Courbin,F.,&Sohy,S.1998,ApJ,494,452
Mineshige,S.,&Yonehara,A.1999,PASJ,51,497