diffuse optical light in galaxy clusters. i. abell 3888

17 Pages·2005·1.04 MB·English

View metadata, citation and similar papers at core.ac.uk brought to you by CORE
provided by University of Queensland eSpace
TheAstronomicalJournal,131:168–184,2006January
#2006.TheAmericanAstronomicalSociety.Allrightsreserved.PrintedinU.S.A.
DIFFUSE OPTICAL LIGHT IN GALAXY CLUSTERS. I. ABELL 3888
J. E. Krick and R. A. Bernstein
AstronomyDepartment,UniversityofMichigan,AnnArbor,MI48109;[email protected],[email protected]
and
K. A. Pimbblet
DepartmentofPhysics,UniversityofQueensland,Brisbane,QLD4072,Australia;[email protected]
Received2005February10;accepted2005September18
ABSTRACT
We are undertaking a program to measure the characteristics of the intracluster light (ICL; total flux, profile,
color,andsubstructure)inasampleof10galaxyclusterswitharangeofclustermass,morphology,andredshift.
We present here the methods and results for the first cluster in that sample, A3888. We have identified an ICL
component in A3888 in Vand r that contains 13%(cid:1)5% of the total cluster light and extends to 700 h(cid:2)1 kpc
70
((cid:3)0.3r )fromthecenterofthecluster.TheICLcolorinoursmallestradialbinisV (cid:2)r¼0:3(cid:1)0:1,similartothe
200
centralclusterellipticalgalaxies.TheICLisredderthanthegalaxiesat400 h(cid:2)1 kpc<r<700 h(cid:2)1 kpc,although
70 70
theuncertaintyinanyoneradialbinishigh.BasedonacomparisonofV (cid:2)rcolorwithsimplestellarmodels,the
ICLcontainsa componentthatformedmorethan7Gyrago(atz>1)witha high-metallicity(1:0 Z <Z P
(cid:4) ICL
2:5 Z )andamorecentralizedcomponentthatcontainsstarsformedwithinthepast5Gyr(atz(cid:3)1).Theprofileof
(cid:4)
theICLcanberoughlyfittedbyashallowexponentialintheouterregionsandasteeperexponentialinthecentral
region.Wealsofindaconcentrationofdiffuselightaroundasmallgroupofgalaxies1.4h(cid:2)1Mpcfromthecenter
70
of the cluster. In addition, we find three low surface brightness features near the cluster center that are blue
(V (cid:2)r¼0:0)andcontainatotalfluxof0.1M .BasedontheseobservationsandX-rayandgalaxymorphology,we
(cid:5)
suggest that this cluster is entering a phase of significant merging of galaxy groups in the core, whereupon we
expecttheICLfractiontogrowsignificantlywiththeformationofacDgalaxy,aswellastheinfallofgroups.
Key words: cosmology: observations — galaxies: clusters: individual (A3888) — galaxies: evolution —
galaxies: interactions — galaxies: photometry
1. INTRODUCTION TheICL isafossilremnantofclusterformationandevolu-
tionandcanbeusedtostudythedominantphysicalprocesses
Galaxy clusters contain a population of stars that are not involvedingalaxyevolutioninclusters.Hierarchicaldarkmat-
members of individual galaxies but are bound to the cluster tersimulationssuggestthatgalaxiesfallingintodenseregions
potential, producing diffuse intracluster light (ICL). This ICL would lose most of their mass. When mechanisms such as ra-
componenthasbeenfoundinanumberofclustersthroughsur- diative cooling and star formation are included in the simu-
facebrightnessmeasurementsanddirectdetectionsofresolved lations,galaxiescomposedofacentraldensecoreofstarsretain
starsincludingplanetarynebulae,redgiants,supernovae(SNe), mostoftheirstellarmassthroughoutclusterinfallbutlosesome
novae, and globular cluster systems. These investigations in- stars to the cluster potential. State-of-the-art simulations are
dicatethattheopticalICLcomprisesbetween5%and50%of abletopredicttheexistenceofthisintraclusterpopulation,but
the total optical cluster luminosity (see Feldmeier et al. 2004; basicquestionsastoitspropertiescanonlybeansweredbyunder-
Gonzalezetal.2005;Zibettietal.2005,andreferencestherein). standingwhichphysicalmechanismsareimportant.Thiswork
Conclusionsonthecolorofthe ICL vary widelyfromblueto seekstodiscoverwhenandhowintraclusterstarsareformedby
red,withandwithoutcolorgradients(Schombert1988;Mackie studyingthetotalflux,profileshape,color,andsubstructurein
1992; Gonzalez et al. 2000; Zibetti et al. 2005). Current mea- theICLasafunctionofclustermass,morphology,andredshift.
surements of the shape of the ICL generally favor a double ObservationsofthetotalfluxintheICLoverasampleofclus-
deVaucouleursprofilesuchthatonefunctionfitsthebrightest terswillallowustoidentifytheeffectsofclusterenvironmenton
clustergalaxy(BCG)andthesecondfunctionfitstheextended galaxyevolution.Forexample,ahigh-massclustershouldhavea
envelope (Gonzalez et al. 2000; Bernstein et al. 1995; Zibetti higherICLfractionthanlow-massclustersiframpressurestrip-
et al. 2005). Examples of tidal features such as plumes and pingorharassmentaredominantmechanisms.Infact,simulations
bridgesarefoundinmultipleclustersasevidenceofinteractions bybothLin &Mohr(2004) and Muranteetal. (2004)predict
thataddstarstotheICL(Gregg&West1998;Calca´neo-Rolda´n a strong relation between ICL fraction and mass. If, however,
et al. 2000; Feldmeier et al. 2004). Long-slit spectroscopy of galaxy-galaxymergingisthedominantmechanismandmostof
A2199 shows that the intracluster stars there have the same thegalaxyevolutionhappensearlyoninclustercollapse,thenthe
velocitydispersionastheclustergalaxies(Kelsonetal.2002), ICLshouldnotcorrelatedirectlywithcurrentclustermass.The
confirming that, in at least one cluster, the intracluster stars existenceofacDgalaxyinaclusterisevidenceofarichmerger
arenotboundtotheindividualgalaxiesbuttracetheoverallclus- history,andtherefore,morphologyshouldalsocorrelatestrongly
ter potential. Conversely, intracluster planetary nebula studies withICLfraction.TheICLfractionwillbeaffectedbyredshift,
showevidenceforlessrelaxedvelocities(Arnaboldietal.2004; sincewithtimecomesanincreasednumberofinteractions.
Gerhardetal.2005).Thereisnoconsensusonthevelocitydis- Observations of the color and substructure of the ICL will
tributionofintraclusterstars. helpidentifytheorigin,formationepoch,metallicity,andpossibly
168
DIFFUSE OPTICAL LIGHT IN GALAXY CLUSTERS. I. 169
progenitormorphologiesofclustergalaxies.Forexample,ifthe a statistical sample representative of a wide range in cluster
ICLisasredorredderthanthebrightclusterellipticalgalaxies, characteristics,namely,redshift,morphology,spatialprojected
itislikelytobearemnantfromtheearlyepochsofclusterfor- density(richness),andX-rayluminosity(mass).
mationwithlittlerecentaccretionoftidallydisruptedsystems. To the extent possible, we also selected clusters for which
IftheICLisbluerthanthegalaxies,thensomerecentaccretion massestimatesandmembershipinformationareavailableinthe
hasoccurred,eitherfromellipticalgalaxieswithlowmetallicity literature.Forexample,inadditiontopublishedX-raymasses,
orspiralgalaxieswithyoungerstellarpopulations.Whilemul- threeoftheclustershavemassestimatesfromgravitationallens-
tiple mechanisms are likely to play a role in the complicated ingmeasurements.Publishedredshiftsurveysprovidevelocity
process of the formation and evolution of clusters, important dispersionsandmembershipinformationforallbuttwoclusters
constraintscancomefromICLmeasurementinclusterswitha inthesurvey.Thosetwoclusters,aswellasfiveotherswithsmall
widerangeofproperties. numbersofpublishedvelocities,wereincludedinaredshiftsur-
In addition to constraining galaxy evolution, the ICL is an vey we undertook with IMACS on Magellan I (Baade). With
important baryonic component in clusters. The ICL, which is theseadditionalobservationsthephysicalpropertiesofallclusters
typically not included in the baryon census, will contribute to inoursamplecanbecomparedtotheICLcharacteristics.Table1
thebaryonfractionofclusters,andthatcontributionislikelyto liststherelevantinformationfortheclustersample.
changewithtime.IftheICLfractiondoesindeedevolvewith The sample is divided into a ‘‘low’’ (0:05<z<0:1) and a
redshiftandisasignificantfractionofthetotalclusterlight,it ‘‘high’’(0:15<z<0:3)redshiftrangeobservedwiththe1m
willsystematicallybiastheinferredredshiftdependenceofthe Swopeand2.5mDuPonttelescopes,respectively.Thebottom
baryonfraction.RecentworkbyAllenetal.(2004)hasuseda end of the redshift range is limited by the field of view of the
change in baryon fraction with redshift of only a few percent 1mtelescopeanddetector,whichcorrespondsto0:9 h(cid:2)1 Mpc ;
70
to constrain cosmological parameters. When doing such pre- 1:4 h(cid:2)1Mpc(14A8;22A8)atz¼0:05.Thisfieldofviewallows
70
cisioncosmologyitwillbenecessarytoincludeICLintheclus- ustomeasuretheICL,aswellasoff-clusterbackgroundfluxin
terlightbudget. thesameimageforallclustersinthesample.Thetopendofthe
TheICLmayalsoplayanimportantroleintheglobalprop- redshiftrangereflectsX-raydataavailabilityandtheincreasing
erties of the intracluster medium (ICM). It has recently been difficulty of measuring diffuse sources at high redshift due to
suggestedthatanintraclusterstellarpopulation(ICSP)canac- (1þz)4 surfacebrightnessdimming.
countforatleastsomeamountofheatingandmetalenrichment
2.1. A3888
oftheICM(Zaritskyetal.2004;Lin&Mohr2004;Domainko
etal.2004).ConsideringonlySNewithingalaxies,thefullmetal- Thispaperfocusesononeclusterinoursample,A3888,which
licityoftheICMcannoteasilybeaccountedfor(Lin&Mohr isarichnessclass2clusteratz¼0:15(Abelletal.1989).This
2004).However,sinceintraclusterSNeareinsituintheICM, Bautz-Morgan(B/M)typeI-IIclusterhasnocDgalaxy;instead,
they contribute directly to the metallicity of the ICM and will thecorecomprisesthreedistinctsubclumpsofmultiplegalaxies
thereforehaveadirectimpactonitsabundance.Althoughthese each.Atleasttwogalaxiesineachoftheseclumpsareconfirmed
authorsfindthattheICLcannotfullyaccountforthehighabun- members (Teague et al. 1990; Pimbblet et al. 2002). On large
dance of the ICM ((cid:3)0.3 Z ), further studies are warranted to scales (286, 535, and 714 h(cid:2)1 kpc) Girardi et al. (1997) find a
(cid:4) 70
quantifyjusthowmanyintraclusterSNethereare.Evenifthe unimodal distribution for this cluster, with no detected sub-
ICSP cannot account for the full metallicity of the ICM, it is structuresineitherthegalaxyspatialorvelocitydistribution.The
possible that this population is responsible for the metallicity projected spatial distribution of galaxies in A3888 is slightly
gradient found in clusters. If true, a correlation between ICL elongated,withanellipticityof0.43(Struble&Ftaclas1994).
fluxandabundancegradientsinclustersshouldexist. X-raysurfacebrightnessesobtainedfromXMM-Newtonobser-
Inthispaperwepresentthemethodsofthissurvey,aswellas vations also indicate an elongated, single-peaked distribution
measurementsofthecolor,totalflux,andprofileshapeforthe for the hot gas. The cluster contains an X-ray-bright Seyfert I
first cluster in our sample, A3888. In x2 we discuss the char- galaxy located at a projected distance of roughly 600 h(cid:2)1 kpc
70
acteristicsoftheentiresample.Detailsoftheobservationsand fromtheclustercenter(Reimersetal.1996).
reductionarepresentedinxx3and4,includingflat-fieldingand ThemassofA3888canbeestimatedfromtwodifferentsets
skybackground–subtractionmethods.Inx5wediscussobject ofobservations.Reiprich&Bo¨hringer(2002)calculategravi-
detectionandremoval,aswellasclustermembership.Inx6we tational mass based on pointed ROSAT PSPC count rates and
describeourresults,followedbyadiscussionofaccuracylimits theROSAT-ASCAL -T relation(Markevitch1998).Assuming
X X
inx7.Inx8wepresentadiscussionoftheresults.Conclusions an isothermal distribution and employing hydrostatic equilib-
aresummarizedinx9. rium,theyfindM ¼25:5þ10:8 ;1014 h(cid:2)1 M ,wherer ¼
200 (cid:2)7:4 70 (cid:4) 200
Throughout this paper we use H ¼70 km s(cid:2)1 Mpc(cid:2)1, 2:8 h(cid:2)1 Mpc, which is defined as the radius within which the
0 70
(cid:1) ¼0:3,and (cid:1) ¼0:7, which gives3.5 kpc arcsec(cid:2)1atthe meanmassdensityisequalto200timesthecriticaldensity.Ina
M (cid:2)
distanceofA3888. complementarymethod,masscanbedeterminedfrompublished
galaxyvelocitydispersions.Basedonredshiftsfor50member
2. THE SAMPLE
galaxieslocatedwithinaradiusof3.11h(cid:2)1Mpc(Teagueetal.
70
The 10 galaxy clusters in our survey were selected to meet 1990) and using the method described by Girardi & Mezzetti
two general criteria. First, each cluster has a published X-ray (2001), we find that the mass of A3888 within r is M ¼
200 200
luminosity that guarantees the presence of a cluster and pro- 40:2þ10:6 ;1014 h(cid:2)1 M . For the purpose of this work, these
(cid:2)7:4 70 (cid:4)
videsanestimateofthecluster’smass.Second,allareathigh two mass estimates are in good agreement, particularly since
Galacticand ecliptic latitudealong linesof sightwithlowH i thisclusteriselongatedandlikelynotindynamicequilibrium.
columndensity.Thisminimizescomplicationsduetoscattered
3. OBSERVATIONS
light from Galactic stars and zodiacal dust and from variable
extinctionacrosstheclusterfield.Oftheclustersthatmeetthe Observationsfortheentiresampleof10clustershavebeencom-
abovequalifications,weselected10clustersasthebeginningof pleted.Thehigh-redshiftobservationsweremadewiththeduPont
TABLE1
ClusterCharacteristics
R.A. Decl. l b Richness L (cid:1) No.ofConfirmed n(Hi)a Lensing
X v
ClusterName (J2000.0) (J2000.0) (deg) (deg) z B/M Class (1044ergss(cid:2)1) (kms(cid:2)1) Members (1020cm(cid:2)2) Measurement
A4059b........................... 2357 (cid:2)3440 356.8 (cid:2)76.06 0.048c I 1 3.09a 845þ280d 45e,f 1.1 ...
(cid:2)140
A3880b........................... 2228 (cid:2)3034 17.99 (cid:2)58.50 0.058c II 0 1.86a 827þ120g 122h,i,j 1.1 ...
(cid:2)79
A2734b........................... 0011 (cid:2)2852 19.46 (cid:2)80.98 0.062c III 1 2.55a 628þ61g 127k,f,j 1.8 ...
(cid:2)57
A2556b........................... 2313 (cid:2)2138 41.37 (cid:2)66.97 0.087c II-III 1 2.47a 1247þ249d 5l,m,n 2.0 ...
(cid:2)249
A4010b........................... 2331 (cid:2)3630 359.0 (cid:2)70.60 0.096c I-II 1 5.55a 625þ127g 30k,f 1.4 ...
(cid:2)95
A3888............................. 2234 (cid:2)3743 3.96 (cid:2)59.40 0.151c I-II 2 14.52a 1102þ137o 69p 1.2 ...
(cid:2)107
A3984b........................... 2315 (cid:2)3748 359.0 (cid:2)67.19 0.181c II-III 2 9.18a ... ...i 1.7 ...
A0141b........................... 0106 (cid:2)2435 175.3 (cid:2)85.93 0.23c III 3 12.62a ... ...q 1.8 Dahleetal.(2002)
AC118............................ 0014 (cid:2)30 2 8.90 (cid:2)81.24 0.308c III 3 22.05a 1947þ292r 363s,t,r 1.7 Smailetal.(1991)
(cid:2)201
AC114............................ 2258 (cid:2)3447 8.32 (cid:2)64.78 0.31q II-III 2 12.7u 1388þ128o 380t,v,r 2.0 Smailetal.(1991)
(cid:2)71
Note.—Unitsofrightascensionarehoursandminutes,andunitsofdeclinationaredegreesandarcminutes.
a FromEbelingetal.(1996).
b Clustersforwhichwehavedoneaphotometricandspectroscopicsurveyforadditionalmembershipinformation(seex2).
c FromStruble&Rood(1999).
d FromWuetal.(1999).
e FromChenetal.(1998).
f FromMazureetal.(1996).
g FromGirardietal.(1998).
h FromStein(1996).
i FromCollinsetal.(1995).
j FromDeProprisetal.(2002).
k FromdenHartog(1995).
l FromCiardulloetal.(1985).
mFromKowalskietal.(1983).
n FromBatuskietal.(1999).
o FromGirardi&Mezzetti(2001).
p FromTeagueetal.(1990).
q FromAbelletal.(1989).
r FromCouch&Sharples(1987).
s FromBusarelloetal.(2002).
t FromCouch&Newell(1984).
u FromDeFilippisetal.(2004).
v FromCouchetal.(2001).
DIFFUSE OPTICAL LIGHT IN GALAXY CLUSTERS. I. 171
2.5mtelescopeatLasCampanasObservatory.Weusedthethinned, mean dark count is 0.6 counts per 900 s, which is less than
2048;2048TektronixTek5CCDwitha3e(cid:2)count(cid:2)1gainand 0.08%oftheskylevelandisthereforenotsignificant.However,
7e(cid:2)readnoise.Thepixelscaleis0B259pixel(cid:2)1(15(cid:2)mpixels), evenatthissmallcountlevelthereissomeverticalstructurein
sothatthefullfieldofviewis8A8onaside,correspondingto thedarkthatamountsto1countper900soverthewholeimage.
1.8h(cid:2)1 Mpcperframe.Dataweretakenintwofilters,Gunn-r Toremove thislarge-scalestructure from thedataimages, the
70
(k ¼65508)andV(k ¼54008).Thesefilterswereselected combined dark frame was median-smoothed over 9;9 pixels
0 0
toprovidesomecolorconstraintonthestellarpopulationsinthe (2B3), scaled by the exposure time, and subtracted from the
ICLwhileavoidingflat-fieldingdifficultiesatlongerwavelengths programframes.Small-scalevariationswerenotpresentinthe
andprohibitiveskybrightnessatshorterwavelengths. dark.Errorsinboththebiasanddarksubtractionduetostruc-
Observingrunsoccurredon1998August19–25,1999Sep- ture in the residuals are an additive offset to the background
tember2–10,and2000September22–27.Specifically,A3888 level.Theseareincludedinourfinalerrorbudgetbasedonan
wasobservedonthenightsof1999September2and8and2000 empiricalmeasurementofthestabilityofthebackgroundlevel
September22–25.Bothobservingrunstookplaceinthedays inthefinalcombinedimage(seex7).
leadinguptonewMoon.Thenightof1999September2 was
the only nonphotometric night, and only three cluster images 4.2.Flat-Fielding
were taken on that night. These were individually tied to the
The accuracy of any low surface brightness (LSB) mea-
photometricdata.Theaverageseeingduringthe1999and2000
surementislimitedbyfluctuationsonthebackgroundlevel.A
runswas1B77and0B93,respectively.Acrossbothrunsweob-
major contributor to those fluctuations is the large-scale flat-
servedA3888foranaverageof5hrineachband.Inadditionto
fieldingaccuracy.Pixel-to-pixelsensitivityvariationswerecor-
theclusterframes,night-skyflatswereobtainedinnearby,off-
rected in all cluster and night-sky flat images using nightly,
cluster, ‘‘blank’’ regions of the sky with total exposure times
high-S/N, median-combined dome flats with 70,000–90,000
roughly equal to one-third of the integration times on cluster
totalcounts.Afterthisstep,alarge-scaleilluminationpatternof
targets. Night-sky flats were taken in all Moon conditions.
order 1% remained across the chip. This was removed using
TypicalV-andr-bandskylevelsduringtherunwere21.3and
combinednight-skyflatsof‘‘blank’’regionsofthesky.Tomake
21.1magarcsec(cid:2)2,respectively.
thesenight-skyflats,objectsintheindividualblank-skyframes
Clusterimageswereditheredbyone-thirdofthefieldofview
werefirstmaskedbeforecombination.WeusedSExtractor(Bertin
between exposures. The large overlap from the dithering pat-
& Arnouts 1996) to identify all sources with a minimum of
terngivesusampleareaforlinkingbackgroundvaluesfromthe
6pixelsandatotalfluxof2(cid:1)abovetheskybackground.Mask
neighboring cluster images. Observing the cluster in multiple
sizeswereincreasedby4–7pixelsoverthesemimajorandsemi-
positionsonthechipisbeneficialbecauseoncombinationlarge-
minor axes from the object catalogs to insure object rejection.
scale flat-fielding fluctuations are reduced. Integration times
The masked images were then median-combined with 2 (cid:1) re-
weretypically900sasacompromisebetweensignal-to-noise
jection. This produced an image with no evident residual flux
ratio(S/N)andmoderatingthenumberofsaturatedstars.
fromsourcesandkeptthelarge-scaleilluminationpatternintact.
Observationsofthelow-redshiftclusterswillbediscussedin Fluctuationsarelessthan0.1%peaktopeakon1000 scales.The
afuturepaper.
finalcombinednight-skyflatswerethenmedian-smoothedover
4. REDUCTION 7;7 pixels (200), normalized, and divided into the program im-
ages. The illumination pattern was stable among images taken
In order to create a single, mosaicked image of the cluster
duringthesameMoonphase.Programimageswerecorrected
withauniformbackgroundlevelandaccurateresolved-source
onlywith night-sky flats taken in conditionsofsimilar Moon.
fluxes,theimagesmustbebiasanddarksubtracted,flat-fielded,
The contribution of flat-fielding to our final error budget is
fluxcalibrated,backgroundsubtracted,extinctioncorrected,and
measuredempirically,asdescribedinx7.
registeredbeforecombining.Theseissuesaredealtwithasde-
scribedbelow. 4.3. Nonlinearity
4.1. Bias and Dark Subtraction AlthoughtheICLmeasurementisbasedonalownumberof
counts,photometriccalibrationisbasedonbrightstandardstars.
Preprocessingofthedata,includingoverscan,bias,anddark
AccuratecalibrationisthendependentontheCCDhavingalinear
subtraction, was done in the standard manner using mainly
responsetoflux.ToascertainwhetherTek5waslinearwithflux
IRAFtasks.Theaveragebiaslevelwasstableat(cid:3)800counts,
over a wide dynamic range, a consecutive chain of dome-flat
changingby1%throughoutthenight.Thereisstructureinthe
imagesweretakenwithexposuretimesof2–100s,corresponding
biasintheformofrandomfluctuations,aswellasahighlyre-
to approximately 300–15,000 counts pixel(cid:2)1. Multiple passes
peatable, large-scale ramping in the first 500 pixels of every
throughtheexposuretimesequence(increasinganddecreasing)
row.Toremovethisstructure,wefirstfitaneighth-orderpoly-
weremadetoruleoutanyeffectsfromfluctuatinglampflux.We
nomialto140overscancolumnsandsubtractthatfit,columnby
find that the Tek5 CCD does have an approximately 2% non-
column,fromeachimage.Wefurtheraveragetogether10bias
linearity,whichwefittedwithasecond-orderpolynomialandcor-
frames per night with 3 (cid:1) cosmic-ray (CR) rejection and then
rected for in all the data. The same functional fit was found for
boxcarsmoothintheverticaldirectionbeforesubtractingfrom
both the 1999 and the 2000 data. Note that the exposure times
thedata.Wechoosetosmoothintheverticaldirectionbecause
usedforallobservationsarelongenoughthatshutterperformance
wehavealreadyremovedverticalstructureinthepreviouspro-
is not a problem. The uncertainty in the linearity correction is
cessingstep.Testreductionofthebiasframesthemselveswith
incorporatedinthetotalphotometricuncertaintydiscussedbelow.
thisprocedurerevealsnoremainingvisiblestructure,andeach
framehasameanlevelof0countstowithin(cid:1)0.05counts.
4.4. Photometric Calibration
Twenty-five dark exposures were taken per observing run.
We averaged these together with a 3 (cid:1) rejection to look for Photometriccalibrationwasperformedintheusualmanner.
structure or significant count levels in the dark current. The Fiftytoseventystandardstars(Landolt1992;Jorgensen1994)
172 KRICK, BERNSTEIN, & PIMBBLET Vol. 131
were observed per night per filter over a range of air masses.
Stellar magnitudes were measured with an aperture size of
5timestheFWHM,wheretheFWHMofthestarsintheimages
wasdeterminedusingSExtractor.Wechoosethisaperturesize
asacompromisebetweenaperturecorrectionandaddedback-
ground noise. Photometric nights were analyzed together; so-
lutionswerefoundineachfilterforazeropointandextinction
coefficientwithanrmsof0.03mag(r)in1999Septemberand
0.02mag(randV)in2000September.Theseuncertaintiesare
a small contribution to our final error budget but are included
forcompletenessasdiscussedinx7.Observingthesamecluster
fieldforlongperiodsthroughoutthenightallowsustomeasure
anextinction coefficient from stars inthecluster fields,which
wefindisfullyconsistentwiththe extinction coefficientmea-
suredfromthestandardstars.Thethreeexposurestakeninnon-
photometricconditionswereindividuallytiedtothephotometric
datausingroughly10starswelldistributedaroundeachframe
tofindtheeffectiveextinctionforthatframe.Wefindastandard
deviationof0.03withineachframe,withnospatialgradientin
theresiduals.
We have compared our V- and r-band magnitudes for hun-
dreds of galaxies in the cluster with R-band magnitudes from
theLasCampanasAATRichClusterSurvey(LARCS;Pimbblet
etal.2002).TothedetectionlimitoftheLARCSphotometry,and Fig.1.—Central(cid:3)1Mpc(4A9)ofA3888intheVband.Thegrayscaleis
adoptingasingleaveragegalaxycolortoconvertbetweenfilters, linearovertherange20.4–29.5magarcsec(cid:2)2.Notethethreemaingroupsof
galaxiesnearthecenterofthecluster.Adozenobjectsinthisimagearestars;the
thetwosamplesareconsistentwithanrmsscatterof0.07mag.
restaregalaxies.
4.5. Sky Background Subtraction
Animportantissueforaccuratesurfacebrightnessmeasure- 4.6. Extinction Correction
mentistheaccurateidentificationofthebackgroundskylevel. Afterbackgroundsubtraction,allfluxintheframeoriginates
Theoff-clusterbackgroundlevelinanyimageisacombination abovetheatmosphereandissubjecttoatmosphericextinction
ofatmosphericemission(airglow)andlightfromextraterrestrial (largeanglescatteringoutofthelineofsight).Thisisequally
sources(zodiacallight,moonlight,starlight,andstarlightscat- true of resolved sources and diffuse sources less than several
tered off of Galactic dust). Zodiacal light comes from solar degreesinextent.Whileextinctioncorrectionsareusuallyap-
photons scattered off of ecliptic dust and is therefore concen- pliedtoindividualresolvedsources,thatisnotpossiblewiththe
tratedintheeclipticplane,which,alongwiththeGalacticplane, diffuseICL.Wecorrectentireclusterimagesforthisextinction
we were careful to avoid in sample selection, so the extrater- bymultiplyingeachindividualimageby10(cid:3)(cid:4)/2.5,where(cid:4)isthe
restrialbackgroundlightwillnotvaryspatially.Lightfromthe airmassand(cid:3)isthefittedextinctiontermfromthephotometric
extraterrestrial sources will also be scattered into the field of solution. This multiplicative correction is between 1.06 and
viewbytheatmosphere.Airglowisemissionfromtherecom- 1.29fortheair-massrangeofourA3888observations.
binationofelectronsintheEarth’satmospherethatwereexcited
during the day by solar photons, and as such is a function of 4.7. Registration and Distortion
solaractivity,timeelapsedsincesunset,andgeomagneticlatitude
Tocombineimageswealignall41individualframestoone
(Leinertetal.1998).Airglowandatmosphericscatteringvary
centralreferenceframe.SExtractorpositionsofapproximately
throughoutthenight,moonlightvariesfromnighttonight,and
10starsineachframeareusedasinputcoordinatestotheIRAF
zodiacallightvariesfromyeartoyear.Thebackgroundvalues
tasksgeomapandgeotran,whichfindandapplyx-andy-shifts
fromframetoframecorrespondinglyvarytemporallybyupto
androtationsbetweenimages.Thegeotransolutionisaccurate
10%throughoutonerunand20%fromyeartoyear.
to 0.01 pixels (rms). As an independent check of registration
Due to the temporalvariations in the background, it is nec-
accuracy,weconfirmthatthecentercoordinatesofstarsinthe
essarytolinktheoff-clusterbackgroundsfromadjacentframes
original images, as compared to the combined image, are the
tocreateonesinglebackgroundofzerocountsfortheentireclus-
sametowithin0.01pixels.Thisuncertaintyisnegligibleforour
termosaicbeforeaveragingtogetherframes.Todeterminethe
measurement,whichismadeonmuchlargerscales.Inaddition,
backgroundoneachindividualframewemeasureaveragecounts
inapproximatelytwenty20;20pixelregionsacrosstheframe. theellipticitiesofindividualstarsdonotchangewithimagecom-
bination,suggestingthatnosystematicerrorsinregistrationexist.
Regionsarechosenindividuallybyhandtobearepresentative
Stellarellipticitiesalsoshownovariationacrosstheframe,sug-
sample of all areas of the frame that are more distant than
0:8 h(cid:2)1Mpcfromthecenterofthecluster.Thisiswellbeyond gestingthattherearenosignificantimagedistortions.
70
the radius at which ICL components have been identified in
4.8. Image Combination
otherclusters(Feldmeieretal.2002;Gonzalezetal.2005)and
isalsobeyondtheradiusatwhichwedetectanydiffuselightin Afterpreprocessing,backgroundsubtraction,extinctioncor-
A3888.Theaverageofthesebackgroundregionsforeachframe rection, and registration, we combined the images using the
issubtractedfromthedata,bringingeveryframetoazeroback- IRAFroutineimcombinewithrejectionsof(cid:2)3.5and+4.5(cid:1).
ground.Theaccuracyofthebackgroundsubtractionisdiscussed Thisrangewaschosenasacompromisebetweenrejectingthe
inx7. CRsandallowingforsomeseeingvariationsinthepeakfluxof
No. 1, 2006 DIFFUSE OPTICAL LIGHT IN GALAXY CLUSTERS. I. 173
stars.Intotal,16and25900sexposuresintheVandrbands,
respectively,wereaveragedtogether.Thefinalcombinedimage
is 4096 pixels (3:6 h(cid:2)1 Mpc) on a side. The central region
70
(approximately 1 h(cid:2)1 Mpc on a side) of the final combined
70
V-bandimageisshowninFigure1.
5. ANALYSIS
5.1. Object Detection
We use SExtractor both to find all objects in the combined
framesandtodeterminetheirshapeparameters.Thedetection
thresholdinboththeVandrimageswasdefinedsuchthatob-
jectshaveaminimumofsixcontiguouspixels,eachofwhich
aregreaterthan1.5(cid:1)abovethebackgroundskylevel.Thiscor-
respondstoaminimumsurfacebrightnessof26.0magarcsec(cid:2)2
in V and 26.4 mag arcsec(cid:2)2 in r. The faintest object in the
cataloghasatotalmagnitudeof27.0maginVand27.4magin
r;however,wearecompleteonlyto24.8maginVand24.5mag
in r. We choose these parameters as a compromise between
detectingfaintobjectsinhigh-S/Nregionsandrejectingnoise
fluctuations in low-S/N regions. Shape parameters are deter-
minedinSExtractorusingonlythosepixelsabovethedetection
threshold. Fig. 2.—PSFofthe100inch(2.5m)duPonttelescopeatLasCampanas
Observatory.They-axisshowssurfacebrightnessscaledtocorrespondtothe
5.2. Object Removal and Masking totalfluxofazero-magnitudestar.Theprofilewithin500 wasmeasuredfrom
unsaturatedstarsandcanbeaffectedbyseeing.Theouterprofilewasmeasured
TomeasuretheICLweremovealldetectedobjectsfromthe fromtwostarswithsupersaturatedcoresimagedintwodifferentpositionsonthe
framebyeithersubtractionofananalyticalprofileormasking. CCDontwodifferentobservingruns.Thebumpintheprofileat10000islikely
Detailsofthisprocessaredescribedbelow. relatedtothepositionofthestarinthefocalplane.Ifthecoreofastarisimaged
offtheCCD,itsprofiledoesnotshowthisfeature,suggestingthatthefeatureis
5.2.1.Stars causedbyareflectionofftheCCDitself.Theoutersurfacebrightnessprofile
decreasesasroughlyr(cid:2)3,shownbythesolidline.Anr(cid:2)2.0profileisplottedto
Scattered light in the telescope and atmosphere produce an showtherangeinslopes.
extendedpoint-spreadfunction(PSF)forallobjects.Tocorrect
forthiseffect,wedeterminethetelescopePSFusingtheprofiles stellarprofilewiththatmagnitudeandproducealargemaskto
of a collection of stars from supersaturated 4 mag stars to un- covertheinnerregionsandanybleeding.We canaffordtobe
saturated 14 mag stars. The radial profiles of these stars were liberalwithoursaturatedstarmasking,sincethereareveryfew
fittedtogethertoformonePSFsuchthattheextremelysaturated saturatedstarsandnoneofthemarenearthecenteroftheclus-
star was used to create the profile at large radii and the unsat- ter,whereweneedtheunmaskedareatomeasureanypossible
uratedstarswereusedfortheinnerportionoftheprofile.This ICL.
allowsustocreateanaccuratePSFtoaradiusof70,shownin
5.2.2.Galaxies
Figure2.
TheinnerregionofthePSF iswellfittedbyaMoffatfunc- TomakeanICLmeasurementwewouldideallyliketosub-
tion. The outer region is well fitted by r(cid:2)3. There is a small tractascaledanalyticalprofileforeachgalaxythatwouldleave
additionalhalooflightatroughly5000–10000 (200–400pixels) noresidualsandwouldallowustorecovertheareaonthesky
around stars imaged on the CCD. Images of saturated stars covered by cluster galaxies. We have attempted to do this us-
located off the field of view of the detector do not show this ingthreepubliclyavailablealgorithms:GIM2D(Simardetal.
halo, indicating that it is due to a reflection of light off of the 2002),GALFIT(Pengetal.2002),andtheIRAFtaskellipse
CCDitself.Wefindthatroughly1%ofthetotalfluxinthestar (Jedrzejewski1987).Withthesealgorithms,wehaveemployeda
is in this halo. There are 13 saturated stars within 3.8 Mpc of widerangeofsurfacebrightnessprofiles,includingdeVaucouleurs,
A3888 ranging from 11.6 to 15.2 V magnitudes. The nearest Se´rsic, exponential profiles, and combinations thereof. In ad-
three saturated stars are 0.6, 0.8, and 1.0 h(cid:2)1 Mpc from the dition,wehaveusediterativetechniquestoalternatelyfitand
70
clustercenterandhave14.6,13.4,and11.6Vmagnitudes,re- removegalaxiesincrowdedfields.Thetechnicalchallengesin
spectively. These stars do not directly contribute to the ICL fitting the galaxies, including galaxy deblending, PSF effects
measurement because they are not near enough to the center, anddeconvolution,two-dimensionalprofilefitting,andspeed
donothaveverybrightmagnitudes,andthePSFdoesnotput inperformingmanyFouriertransforms,havebeenpreviously
verymuchpowerintothewings.Wedoacarefuljobofback- discussed by several groups (see, e.g., Peng et al. 2002 for a
groundsubtraction,bytyingtooff-clusterflux,sothatthePSF review).
alsodoesnotaffectthebackgroundmeasurement. Figure 3 shows representative results of modeling three
For each individual, nonsaturated star we subtract a scaled galaxiesusingGALFIT:oneisolatedgalaxyandtwogalaxiesin
r(cid:2)3profilefromtheframeinadditiontomaskingtheinner3000 increasingly dense regions. These examples show that the al-
oftheprofile(theregionthatfollowsaMoffatprofile).Sincewe gorithmsperformwellforisolatedgalaxiesbutfailforgalaxies
do not have accurate magnitudes for the saturated stars in our nearthecoreeitherbecauseofthedifficultyindeblendingmany
owndata,andtobeascautiousaspossiblewiththePSFwings, overlappinggalaxyprofilesorbecausetheindividualgalaxies
we have assumed the brightest possible magnitudes for these insuchdenseregionsdonotfollowsimpleanalyticalprofiles.
stars given the full USNO catalog errors. We then subtract a Itisnotclearwhattheprofilesshouldbeofgalaxiesdeepinthe
174 KRICK, BERNSTEIN, & PIMBBLET Vol. 131
Fig.3.—Imagesofobservedgalaxies,GALFITmodels,andmodel-subtractedresidualimagesshownfromlefttorightforthreedifferentgalaxiesinA3888.All
imagesareshownatthesamesurfacebrightnesslevels.Thetoprowshowsafairlyisolatedgalaxyintheouterregionsofthecluster,whichiswellmodeledbyGALFIT.The
middleandbottomrowsshowgalaxiesinincreasinglydenserenvironments,depictingwellthelimitationsofgalaxymodelingalgorithmsforgalaxiesinverydenseregions.
potentialwellsofclusters(Trujilloetal.2001;Feldmeieretal. We use the same masks for both bands so that all galaxies
2004).ThefactthatA3888isnotarelaxedclusterclearlymakes aremaskedtothesameradius,therebyensuringaself-consistent
galaxy subtraction more difficult near the core than it would measurementoftheICLcolor.Weusether-bandimagetode-
be in a cD cluster; A3888 has three main brightness peaks, finethemasks,asithasa deeperdetection threshold(and thus
whichcontain3,7,and12galaxycoresintheirdensestregions, largerdetectionareas)thantheV-bandcatalog.Objectsareiden-
respectively. tified using SExtractor, and masks are based on the isophotal
Asitisnotpossibletocleanlyfitthegalaxiesinthiscluster detection area with a threshold of 26.4 mag arcsec(cid:2)2 (1.5 (cid:1)
suchthattheresiduals(positiveornegative)donotinterferewith abovesky).Tobeconservativeinrejection,wescalethesemi-
theICLmeasurement,wehavechosentomaskthegalaxies.This majorandsemiminoraxesidentifiedbySExtractortoincrease
givesusawell-definedmeasurementoftheICLattheexpense theareaofeachgalaxymaskbyamultiplicativefactorof 2–2.3,
offorfeitingsomearea.Althoughwecouldmodelandsubtract dependingonthemagnitudeofthegalaxy.Toexploretheeffect
themoreisolatedgalaxiesintheouterregionsofthecluster,itis ofmasksizeontheprofileshapeoftheICL,wemaketwoad-
intheseregionsthatwecangenerouslymaskthegalaxiesand ditionalimageswithmasksizesthatare30%smallerand30%
stillhaveenoughpixelsforanICLmeasurement.Notethatwe largerthantheoriginalmasks.WethenmeasuretheICLthree
do not replace masked pixels. Masked regions are simply re- times with the three versions of mask sizes. Additional minor
movedentirelyfromtheICLmeasurement. maskingisdonebyhandtoremoveanyremainingfluxassociated
No. 1, 2006 DIFFUSE OPTICAL LIGHT IN GALAXY CLUSTERS. I. 175
Fig. 5.—Color-magnitudediagramofgalaxiesinA3888.Allgalaxiesde-
tected in our data are plotted with gray symbols. Those galaxies that have
membershipinformationintheliteratureareoverplottedwithopentriangles
(members)orsquares(nonmembers).Theredsequenceisclearlyvisible.Solid
linesindicateabiweightfittotheredsequencewith1(cid:1)uncertainties.
Fig.4.—Fullymasked,finalV-bandimageofthecentral1.5h(cid:2)1Mpc(7A3)of higher redshift background galaxies, and not as concentrated
70
A3888,smoothedtoaidinthevisualidentificationofsurfacebrightnesslevels. towardthecenteroftheclusterasallgalaxies.Thenumberof
Masksareshownintheirintermediatesize;seex5.2.2.Largecircularmaskscorre- thoseveryredgalaxiesperprojectedareais38%(cid:1)11%higher
spondtothelocationsofbrightstars.Thesixgray-scalelevelsshowsurfacebright- within400kpcthanwithout.Althoughsomeofthesegalaxies
nesslevelsofupto28.5,27.7,27.2,26.7,andbrighterthan26.2magarcsec(cid:2)2.
are undoubtedly members of the cluster, their spatial distribu-
Theellipseisophotesareoverlaidfrom6500to19000.Thetidalfeature,C,also
showninFig.11,isclearlyvisibleatcenternearthebottomoftheimage. tiondoesnotallowustomakeconclusivestatementsabouttheir
membership.Approximately42%ofthegalaxiesintheimage
areidentifiedasmembersbythismethod.Ofthegalaxieswith
with resolved objects. These few regions are associated with spectroscopically determined velocities, 78% of the 55 con-
small overlapping sources that are not correctly deblended by firmedmembersareincludedinthecut;54%ofthe13known
SExtractor. nonmembers are also included. The red cluster sequence is a
The total masked area within the central 1.2 Mpc of the goodtoolforidentifyingclusters,butitisnotaperfectmethod
clusterineachofthethreemasksizesis34%,41%,and49%, of determining membership, as it is unable to cleanly distin-
respectively. The masked fraction is much higher in the very guishbetweenmemberandnonmembergalaxies.
centeroftheclusterandreachesnearly100%intheinner3000. Wemeasurethetotalfluxinallgalaxiesidentifiedasmem-
Theincreaseinthemaskedfractionisnotdirectlyproportional bers using corrected isophotal magnitudes from SExtractor.
to the increase in mask size because the masks often overlap. For these, SExtractor assumes a Gaussian profile to infer the
Figure4showsthefinalV-bandimagewithintermediate-sized fluxbeyondtheisophotaldetectionthreshold,correspondingto
masks. 26.0Vmagarcsec(cid:2)2and26.4rmagarcsec(cid:2)2.Asexpected,the
corrected magnitudes are brighter than the isophotal magni-
5.3. Cluster Membership and Flux
tudesbyafullmagnitudeatthefaintendofourdetectionlimit.
An interesting characteristic of the ICL lies in its compari- Thetotalfluxingalaxieswithin700h(cid:2)1kpcofthecenterofthe
70
sontoclusterpropertiesthatincludetheclustergalaxiesthem- cluster, as determined from the same galaxy catalog that was
selves.Wecomparetwomethodsbelowformeasuringcluster used for galaxy masking, is 3:9;1012 L in the V band and
(cid:4)
membership and flux: (1) we identify member galaxies using 4:9;1012L intherband.Weexpecttheerroronthetotalflux
(cid:4)
ourowntwo-bandphotometry;and(2)weintegratethefluxina fromthisestimatetobegreaterthan30%,whichismainlydue
publishedgalaxyluminosityfunctionforthiscluster. touncertaintyinthemembershipdetermination.
Some published velocities are available in the literature We can also determine cluster flux using the Driver et al.
(Teagueetal.1990;Pimbbletetal.2002)andcanbeusedtoex- (1998)luminositydistributionforthiscluster,whichisbasedon
plicitlyidentifymembergalaxies.However,theseredshiftsur- a statistical background subtraction of noncluster galaxies. It
veysarenotcompletetoourdetectionthresholdandtherefore wouldbepossibletodothiswithourowndata;however,Driver
cannot provide membership information for all detected gal- etal.(1998)havemoreuniform,large-areacoveragetoseveral
axies.Alternatively,wecanestimateclustermembershipusing magnitudesbelowM attheredshiftofthecluster.Inaddition,
(cid:5)
acolor-magnituderelation(Fig.5)fromourVandGunn-rim- theauthorspaycarefulattentiontoobservingbackgroundfields
ages. There is a clear red sequence of galaxies in which the thatareupto750 fromtheclustercenter,atapproximatelythe
brightest galaxies have V (cid:2)r¼0:3(cid:1)0:15. Those galaxies sameairmass,seeing,exposuretime,andUTastheclusterfields.
thatliewithin1(cid:1)ofabiweightfittotheredsequencearetaken Consequently,thebackgroundfieldshavethesamenoisechar-
tobeclustermembers(functionalformtakenfromBeersetal. acteristics and detection threshold as the cluster images and
1990).Theslopeoftheredsequenceis0.1mag(color)/4mag samplethesamelarge-scalestructures.Theycanthereforebeused
(galaxy r-magnitude). Those galaxies that are redder than the to reliably determine the contamination of the cluster fields.
redsequencearebothgenerallyfainter,implyingthattheyare Bernsteinetal.(1995)giveacarefulaccountofthesignificant
176 KRICK, BERNSTEIN, & PIMBBLET Vol. 131
considerationsinusingthismethod,allofwhicharetakeninto
accountbyDriveretal.(1998).
We explore one minor effect not discussed by Driver et al.
(1998): the effect of gravitational lensing on the background
galaxycounts.Therearetwocompetingeffectsthatchangethe
number and brightness of galaxies behind the cluster as com-
paredtobackgroundgalaxycountsinanoff-clusterfield.First,
magnificationofthebackgroundgalaxieswillartificiallyinflate
thebackgroundcountsbehindthecluster,resultinginanunder-
estimationofclustergalaxyflux.Second,allbackgroundobjects
behind the cluster will appear radially more distant from the
cluster center, which will artificially decrease the background
counts, resulting in an overestimate oftheclustergalaxy flux.
The change from an overall magnification to demagnification
happensatz’0:5.FollowingthemethodofBroadhurstetal.
(1995) to determine the strength of the demagnification for
A3888atz¼0:15,wefindanegligibledegradationintheV-and
r-band flux (<0.2%) and therefore do not correct for it in the Fig.6.—SurfacebrightnessprofileoftheICL,aswellasthetotalclusterlight
Driveretal.(1998)backgroundcounts. plottedasafunctionofdistancealongthesemimajoraxisinarcseconds.Theaxis
Driveretal.(1998)usetheirR-bandluminositydistribution atthetopofthefigureindicatesthecorrespondingphysicalscaleinh(cid:2)1Mpc.We
70
to determine a dwarf-to-giant ratio; however, we choose to fit plotbothV-andr-banddatatogetherforcomparison.Thebottomtwolineson
theplotaretheICLprofiles;ther-bandlightissurroundedbyfilledshading,and
it with a classical Schechter function (M(cid:5) ¼(cid:2)22:82(cid:1)0:28;
R theVbandissurroundedbyhatchedshading.Theshadingsshowthedifference
(cid:5)¼(cid:2)0:97(cid:1)0:09; and (cid:4)2(cid:6) ¼0:71), which can then be used in ICL profiles produced by increasing or decreasing the area of the galaxy
todeterminealuminositydensityforthecluster.Wenotethat masks,asdiscussedinx5.2.2.Thetoptwolines(withoutshading)representthe
theluminositydistributionisnotperfectlyfittedbyaSchechter totalclusterlightasmeasuredinthesameellipticalisophotesastheICL;the
dashedlinerepresentstheV-bandlight,andthesolidlinerepresentstherband.
functionatthebrightend,duemainlytoasmallnumberofex-
Alsoshownarethecumulative1(cid:1)errorsforbothbandsasdiscussedinx7and
tremely bright galaxies, as is typical of clusters. Adopting a
summarizedinTable3.
volume equal to that over which we are able to measure the
ICL,1.4Mpc3,andintegratingtheluminosityfunctiondownto
veryfaintdwarfgalaxies,M ¼(cid:2)11,thetotalluminosityfrom NotethatwearenotabletomeasuretheICLatradiismaller
R
galaxiesintheclusterisð5:9(cid:1)0:94Þ;1012 L intheRband. than6000becausethatregionisheavilymasked.MostotherICL
(cid:4)
Given galaxy colors from Fukugita et al. (1995), the total lu- measurementsfocusonthisinnerregion,leavinglittleoverlap
minosityfromgalaxiesinA3888isð3:4(cid:1)0:6Þ;1012 L inV betweenthissurveyandpreviousworkinotherclusters.Inclus-
(cid:4)
andð4:3(cid:1)0:7Þ;1012L intherband.Thedifferencebetween terscontainingacDgalaxy,thediffusecomponentofthecluster
(cid:4)
this value of total flux and that determined from our color- hasbeenfoundtoblendsmoothlyintothecDenvelope,andmask-
magnitudeestimateofmembershipislikelyduetouncertainties inginthecoreofsuchclustersisnotnecessary(seemostrecently
in our membership identification and differences in the detec- Gonzalezetal.2005).
tionthresholdsofthetwosurveys.Althoughthetwoestimates We identify the surface brightness profile of the total clus-
aregenerallyconsistent,weadoptthetotalfluxasderivedfrom terlight(i.e.,includingresolvedgalaxies)forcomparisonwith
the luminosity distribution throughout the remainder of the the ICL within the same radial extent. To do this, we make a
paper. new‘‘cluster’’image,withcolor-determined,nonmembergal-
axiesmaskedout(seex5.3).Asurfacebrightnessprofileofthe
6. RESULTS cluster light is then measured from this image using the same
elliptical isophotes as were used in the ICL profile measure-
6.1. Surface Brightness Profile
ment. This profile, in contrast to the ICL, is quite irregular,
After subtracting the stars and masking the galaxies, we fit reflectingtheclusteringofgalaxies.Substructureinthegalaxy
the resulting image with the IRAF routine ellipse, a two- distributionisanindicationofayoungdynamicalageforthis
dimensional,interactive,isophote-fittingalgorithm.Again,the cluster.
maskedpixelsarecompletelyexcludedinthisprocedure.There Figure6showsthesurfacebrightnessprofilesoftheICL,as
arethreefreeparametersintheisophotefitting:center,position wellasthetotalclusterlightasafunctionofsemimajoraxisin
angle, and ellipticity. We fix the center (R:A:¼22h34m26s:0, both the Vand r bands. Results based on all three versions of
decl:¼(cid:2)37(cid:6)44007B2 [J2000.0]) and position angle ((cid:2)70(cid:6)) to masksize(asdiscussedinx5.2.2)areshown.Theuncertainty
valuesfoundbyellipsebasedontheinnerisophotesandlet intheICLsurfacebrightnessisdominatedbytheaccuracywith
the ellipticity vary as a function of radius. Fitted ellipticities which the background level can be identified, as discussed in
rangefrom0.2to0.5.Allowingthecenterandpositionangleto x7.ErrorbarsinFigure6showthecumulativeuncertaintiesin
varyresultsinworsefits.Stablefitsarefoundfrom6000to25000. Table3.
FromthefittedisophotesweidentifyafairlysmoothICLprofile Two characteristics are evident from the surface brightness
over the range of 26 to approximately 29 mag arcsec(cid:2)2. The profiles.First,theinnerregion(200–400h(cid:2)1kpc)hasanotably
70
erroronthemeanwithineachellipticalisophoteisnegligible, steeperprofilethantheouterregion.Whiletheentireprofilecan
asdiscussedinx7.Itispossiblethatthedifferentseeinginthe beadequatelydescribedwithinthe1(cid:1)uncertaintiesbyasingle
V-andr-bandimagescouldunevenlyaffecttheprofiles.Toad- exponential,adoubleexponentialgivesabetterfitintherband
dressthisissue,theV-andr-bandimageshavebeenconvolved ((cid:4)2improvesby50%)andamarginallybetterfitintheVband.
(cid:6)
tothesameseeingandthesurfacebrightnessprofilesremeasured. ThesefitsareshowninFigures7and8.Wehavealsofittedthe
Nosignificantchangewasfoundintheprofiles. ICLprofilewithdeVaucouleursandSe´rsicprofiles.Acceptable
No. 1, 2006 DIFFUSE OPTICAL LIGHT IN GALAXY CLUSTERS. I. 177
Fig.7.—V-bandICLand2(cid:1)errorbarsoverplottedwithexponentialfits.The
best-fitsingle(dashedline)anddouble(grayline)exponentialsareshown.
fits can be found; however, the best-fit values are unphysical,
Fig.9.—X-raycontourstakenfromXMM-Newtonarchivaldataoverlaidon
namely, they have high exponents for the Se´rsic and unrealis-
ourV-bandopticalimage.Logarithmiccontoursareshownfrom1to20counts.
ticallylargeeffectiveradiiforthedeVaucouleursprofiles.The Thebrightpointsource600h(cid:2)1 kpcfromtheclustercenterisaSeyfertI
70
secondgeneralcharacteristicoftheICListhatitismorecon- galaxy.
centratedthanthegalaxies,whichistosaythattheICLfallsoff
morerapidlywithradiusthanthegalaxylight.
diffuseICLhasanaveragecolorofV (cid:2)r’0:3(cid:1)0:1.Beyond
400 h(cid:2)1 kpc the ICL becomes increasingly redder such that
6.2. Spatial Distribution 70
by700h(cid:2)1kpctheICLhasanaveragecolorofV (cid:2)r’0:7(cid:1)0:4.
70
TheICLisalignedtowithin10(cid:6) ofthepositionangleofthe Theonlycharacteristiccolorofthegalaxieswehavetocompare
hotintraclustergas.Figure9showscontoursofXMM-Newton with the ICL is the red-sequence color (V (cid:2)r¼0:3(cid:1)0:15).
archivalobservationsoverlaidonouropticalimage.Weinter- We have no definitive membership information for those gal-
pret the alignment of the diffuse ICL with the hot gas in the axies off the red sequence. The color of the ICL in the inner
clusterasanindicationthatweareindeedmeasuringlightthat 400 h(cid:2)1 kpc is roughly equivalent to the red elliptical galaxies
70
followsthegravitationalpotentialofthecluster.Inaddition,the residing in the same part of the cluster but significantly redder
ICL radial surface brightness profile is significantly different thanseveraltidalfeatureswedetect(seex6.5).Thecolorofthe
fromthegalaxysurfacebrightnessprofileinbothVandr,sug- ICL beyond 400 h(cid:2)1 kpc is redder than the red-sequence gal-
70
gestingthattheICLcomponentisatleastinpartdistinctfrom axies.ThecolorofthediffuseICLcanbeapproximatedasasim-
theindividualgalaxiesinthecluster. plelinearfunctionofradius,withaslopeof+0.1per100h(cid:2)1kpc
70
anday-interceptof(cid:2)0.1.Figure10showsthecolorprofileand
6.3.Color corresponding1(cid:1)errorbars.Whilethisfitisclearlysimplistic,
We measure an average V (cid:2)r color of the ICL by binning
together three to four pointsfrom the ICL radial profile. Be-
tween200and400h(cid:2)1 kpc,theinnermostmeasuredradii,the
70
Fig. 10.—ICL color vs. radius in coarse radial bins basedon thesurface
brightnessinVandr,asshowninFig.6.Theloweraxisshowstheradiusin
arcseconds,andtheupperaxisshowstheradiusinmegaparsecs.Thedashedline
isthebest-fitlinearfunction.Theaveragecolorsoftheredclustersequenceand
Fig.8.—SameasFig.7,butintherband. thetidalfeaturesarealsoshownforcomparison.