Evolution of metals and dust in the universe
Pushpa Khare, 1
V. P.Kulkarni, 2
J. T. Lauroesch, 3
S. M. Fall, 4
D. G. York, 5
D. E. Welty, 5
A. P. S.Crotts, 6
J. W. Truran, 5
and
O.Nakamura 7
1
PhysicsDepartment, UtkalUniversity, Bhubaneswar, 751004,India
2
Departmentof PhysicsandAstronomy,UniversityofSouthCarolina,Columbia,SC29208,USA
3
Departmentof PhysicsandAstronomy,Northwestern University,Evanston,IL60208, USA
4
SpaceTelescopeScience Institute,Baltimore,MD21218, USA
5
Departmentof AstronomyandAstrophysics,Universityof Chicago,Chicago,IL60637,USA
6
Departmentof Astronomy,ColumbiaUniversity,NewYork, NY10027, USA
7
SchoolofPhysicsandAstronomy,UniversityofNottingham,NottinghamNG72RD,UK
Abstract. With theaimofdeterminingtheevolutionofmetals anddustin
theuniverse,wehaveobserved8DampedLyman-alphaAbsorbers(DLAs)with
theMultipleMirrorTelescopeandtheHubbleSpaceTelescopeat0:1<z<1:5,
includingseveralabsorbersdiscoveredin theSloanDigitalSky Survey. These
measurementshavemorethandoubledthesampleofZnandCrmeasurements
atz<1:5andaddedthreemeasurementsatz<0:4,wherenoneexistedbefore.
In contradiction with thepredictions of most chemical evolution models, our
data suggest that the global mean metallicity of DLAs, as measured by the
gasphaseabundanceofZn,atbestevolvesweaklywithredshiftovertherange
0:09 < z < 3:9 and does notseem to rise to the solar level evenat very low
redshifts. Thedust content, as determined by [Cr/Zn], does notshow much
changewithredshift.
Keywords: cosmology: observations{galaxies: abundances{galaxies: evolu-
tion{quasars: absorptionlines
Basedon
Khare,P.,Kulkarni,V.P.,Lauroesch,J.T.,York,D.G.,Crotts,A.P.S.andNakamura,O.,2004,
ApJ,616,86.
Kulkarni,V.P.,Fall,S.M.,Lauroesch,J.T.,York,D.G.,Welty,D.E.,Khare,P.andTruran,J.W.,2005,
1. Introduction
Observations of the UV luminosity density of galaxies in the redshift surveysand the
Hubble Deep Field studies indicate that the global average star formation rate and,
therefore, the chemical enrichment rate of galaxies was high at z 1.5. Most cosmic
chemical evolution models (e.g., Malaney and Chaboyer 1996; Pei, Fall, and Hauser
1999;Somerville,PrimackandFaber2001)thus,predictthattheglobalmeaninterstellar
metallicity in galaxies risesfrom nearly zeroat high redshifts to nearly solarat z =0.
Itisofgreatimportancetodeterminewhethertheobservedglobalmeanmetallicities of
galaxiesagreewithpredictionsofthecosmic chemicalevolutionmodels.
DampedLyman-alphaabsorptionsystems(QSOabsorptionsystemswithHIcolumn
density210 20
cm 2
;hereafterDLAs)havebeenusedtodeterminetheevolutionof
thegasphaseabundancesintheuniversebecause(1)itisbelievedthatDLAsconstitute
mostof theneutralgasin thegalaxies athighredshifts,enoughtoform allstarsvisible
today (Wolfe et al. 1995), (2) in the absence of selection eects, DLAs are expected
toprovideanunbiasedsampleof normalgalaxiesastheyare selectedonlythroughthe
presenceoflargeamountsofneutralhydrogen,(3)DLAshavesuÆcientlylargeamountsof
neutralhydrogensothationizationcorrectionsforH,Zn,Cretcarerelativelysmall,and
(4)DLAsareourprincipalsourceofinformation aboutthemetalcontentofinterstellar
matteringalaxiesover80%oftheageoftheuniverse.
AlargenumberofelementshavebeenobservedinDLAs. Inparticular,Fe,Zn,Si,S
andOhavebeenusedasprobesofthemetallicityinthesesystems. Wepreferto useZn
todeterminethetotal(gas+solidphase)metallicityinDLAsbecause(1)ZntracksFein
mostGalacticstars,(2)itisrelativelyundepletedoninterstellardustgrains,and(3)the
linesofthedominantionizationspecies,Zn II,aremostoftenunsaturated. Abundances
of depleted elements such as Cr, Fe orNi relativeto Zn probe the dust content of the
absorbers.
Theissue ofmetallicityevolutionin DLAshasnotbeenfullyunderstood, themain
reason for the uncertainty being thesmall numberof measurements, especially at z <
1.5(Kulkarniand Fall 2002). This isbecause,thoughfor 0:6<z <1:3, theZn II lines
canbeaccessed withground-basedtelescopes,theyliein blue wavelengthswhere many
spectrographs havelowersensitivity. Also, theLy forz <1:5and theZn II lines for
z<0:5canonlybeaccessedwithspacetelescopes.
WiththeaimofenhancingtheDLAsampleatz<1:5wehaveobserved8DLAswith
the6.5mMultipleMirrorTelescope(MMT)andtheHubbleSpaceTelescope(HST),with
0.09<z<1:36,theresultsforwhichwepresenthere. Theobservationsarepresentedin
Table1. TargetDLAsandabundances.
QSO z
em z
abs
[Zn/H] [Cr/Zn]
Q0738+313 0.635 0.0912 < 1:1 >-0.1
Q0738+313 0.635 0.2212 <-0.7 >-0.7
Q0827+243 0.939 0.5247 <-0.0 ...
Q0952+179 1.472 0.2378 <-1.0 >-0.6
Q0933+733 2.525 1.4790 -1.6 -0.2
SDSSJ110729.03+004811.1 1.392 0.7405 -0.6 -0.3
SDSSJ172739.03+530229.1 1.444 0.9449 -0.5 -0.4
SDSSJ172739.03+530229.1 1.444 1.0311 -1.4 -0.3
2. Observations
OurMMTsampleoriginallyconsistedof11quasarabsorbersat0:09<z<1:5,forwhich
eitheradampedLylinewasobservedinHST spectra,orastrongDLAwasexpected
onthebasisofMgIIorFeIIlines(RaoandTurnshek2000)availableintheSDSSEarly
DataReleasespectra(Schneider etal. 2002). Inparticular,wechose systemsthathave
W rest
MgII2796
>1:0
A, W rest
FeII2344
>0:8
A and someotherindicator of highN
HI
(Fe I,CI,
MgI,MnII,Ni II,Ca II,CrII,Zn IIorSiII 1808). Later, HSTobservations ofLy
linesinthesesystems(HSTGOprojectNo. 9382;PI:S.Rao)conrmedtheDLAnature
of several of these absorberswhile 3 others were found to be sub-DLAs (systems with
HIcolumn density between10 19
210 20
cm 2
). ForHST observations, the sample
consistedofconrmedDLAswith0.09<z<0.52. Table1liststherelevantpropertiesof
theobserved\classicalDLAs"i.e. systemswithlogN
HI
20:3. Datareductionwasdone
withImageReductionandAnalysis FacilityandSpace TelescopeScience DataAnalysis
System. ForMMT data theZn II 2026 line and Mg I 2026line aswell as theZn
II 2062lineand CrII 2062line areblendedtogether. A prolettinganalysis was
performedforthese linesalongwiththelines2056and2066ofCrIIsimultaneously,
assumingthevelocitydispersionparametertobethesameforallthese ions. Wherever
necessary,thecolumn densityofMgIwasobtainedfromtheequivalentwidthofMgI
2853lineobservedintheSDSSspectra,usingthebvalueobtainedfortheunblendedCr
IIlines.
3. Results and discussion
TheabundancesofZnandthoseofCrrelativetoZnfortheDLAsinoursamplearegiven
inTable1. Formostofthesystems,evenforthoseatthelowestredshifts,theabundances
of Zn are sub-solar. The only super-solarabundance that we have obtainedis for the
systemwithredshiftof0.716towardSDSSJ1323-0021whichhasneutralhydrogencolumn
20 2
To examine the implications of our data for the mean metallicity-redshift relation
of DLAs, we combined our data with other Zn measurements at higher redshifts. We
only included the classical DLAs in our study and, also, excluded systems with radial
velocitieswithin3000kms 1
fromthequasaremissionredshifts,sincethesesystemsmay
be associatedwith theQSOsandmaynotrepresentinterveningpopulationofgalaxies.
Thecombinedsample containsa totalof 87 DLAs with Zn detections or limits, in the
redshiftrange0:09<z<3:90. Wehavenormalizedthemalltothesamesetofoscillator
strengthsandsolarabundancesthat wehaveusedforourdata.
0 1 2 3 4
Redshift
−3
−2
−1 0
log (Z/Zsolar)
PFH99 PF95 MC96 SPF01
−3
−2
−1 0
log (Z/Zsolar)
PFH99 PF95 MC96 SPF01
Maximum Limits
Minimum Limits
Figure 1. Theglobalmetallicity-redshiftrelationdeducedfromdampedLy-absorbers. The
circlesshowthelogarithm ofthe NHI-weightedmean Znmetallicity relative tothe solar value
vs. redshift forthe sampleof87DLAs. Thelled circlesinthetopand bottompanelsshow,
respectively, the metallicity for the maximum-limitsand minimum-limits cases. The unlled
circlesinthe bottom panelshowthe metallicity obtained byusingelements otherthanZnto
constrain the metallicities in casesof Zn limits. Vertical error bars denote 1 uncertainties.
Horizontal bars show the bin extent. Eachpoint is plotted at the median redshiftinits bin.
Theshort-dashed,long-dashed,solid,andshort-dash-long-dashedcurvesshow,respectively,the
meanmetallicity(not correctedfor dustobscuration) inthecosmicchemicalevolution models
ofPeiand Fall (1995),Malaney and Chaboyer(1996),Pei etal. (1999), andSomervilleetal.
(2001).
Of the87systemsin thecombinedsample,51are detectionsand36 arelimits. We
asdetections, anda\minimum-limits" sampletreatingthe Zn limitsas zeros. For any
individual system, these cases cover the full range of possible values the Zn column
densitiescantake. Forthesakeofpictorialillustration,wedividedeachofthesesamples
intosixredshiftbinswith14or15absorberseach. ResultsareshowninFigure1. Clearly,
theN
HI
-weightedmean metallicityof DLAsdoesnotrisefast enoughto reach solaror
near-solarmetallicitiesatz=0.
Forthebinneddata(withsixbinsintherange0:09<z<3:90),thelinearregression
slope of the logarithm of the metallicity vs. redshift relation is 0:180:06 for the
maximum-limits sample and 0:220:08 for the minimum-limits sample. The linear
regressionintercept of the metallicity-redshift relation (which corresponds to predicted
present-day metallicity) is 0:740:15 for the maximum-limits sample, 0:750:18
fortheminimum-limitssample. TheresultschangeverylittleiftheZnupperlimitsare
replacedwiththeconstraintsfromotherelements.
0 1 2 3 4
Redshift
−2
−1.5
−1
−0.5 0 0.5 1
[<Cr/Zn>]
DLAs Sun
Galactic Halo Clouds Galactic Warm Clouds Galactic Cold Clouds
Figure2. BinnedunweightedmeanoflogarithmicabundanceofCrrelativetoZnh[Cr/Zn]ivs.
redshiftfor DLAsfromoursampleand otherdata fromtheliterature. Thezero leveldenotes
thesolarlevel.
The refractoryelementssuch as Si, Ti, Cr, Mn, Fe, Co, and Ni are expected to be
depletediftheDLAscontaindust,whilevolatileelementslikeZnarealmostundepleted
ininterstellarclouds. [Cr/Zn]is,therefore,takentobeameasureoftheamountofdust.
ofthe dustcontentwecombinedourdata withthose from theliterature. Asbefore we
only used the classical DLAs which were separated by > 3000 km s 1
from the QSO
redshift. We,also, excluded caseswhere onlyupperorlowerlimitsare available forCr
and/orZn, and scaledthe measurementsto thesameset of oscillator strengthsfor the
ZnIIandCrIIlinesusedbyus. InFig. 2,weplot[Cr/Zn]vs. redshiftforthiscombined
sample of 41 DLAs. Our MMT data have provided 4 new [Cr/Zn] measurements for
DLAs at z < 1:5 and nearly doubled the existing sample of [Cr/Zn] measurementsat
theseredshifts. InadditionourMMTdatahaveprovided[Cr/Zn]ratioforonecandidate
DLAand threesub-DLAs . A linearregressiont totheunbinned[Cr/Zn]vs. redshift
datagivesaslopeof0:070:01,andaninterceptof 0:480:02implyingaslowdecrease
in[Cr/Zn]withdecreasingredshift.
Tosummarize,inallthe3DLAsatz<0:4forwhichourobservationsoermeaningful
constraintsonthe Zn abundance, the metallicities appear to besubstantiallysub-solar
(10 20%solar). TheglobalmeanmetallicityofDLAsshowsatmostaslowincrease
withdecreasingredshiftanddoesnotappeartoriseuptosolarornear-solarvaluesatz=
0in contradictionwith thechemical evolutionmodels. Thissuggeststhat theobserved
DLAs, especially those at low redshifts, probablydier from the generalpopulation of
galaxiesrepresentedbytheglobalstarformationhistoryoftheuniverse.
ThisweakevolutionintheDLAglobalmeanmetallicitycouldbepartiallyexplained
bythefactthatoursampleatredshiftssmallerthan0.6seemstobedominatedbydwarf
orlowsurface brightnessgalaxies. It is possible that current DLA samples, especially
thoseat lowredshifts,arebiasedagainstmoreenrichedgalaxies becausethelattermay
cause more dust obscuration of the background quasars. We note, however, that the
low-redshift samples are still small, and the fact that mostof the DLAs in our sample
are metal-poorcouldbean eect ofsmall numberstatistics. Metallicity measurements
of more low-redshift DLAs are necessary to further improve the statistics of the mean
metallicity-redshift relationand to studythescatter aroundthemean. Observations of
low-redshift DLAs towardoptically faint quasars are especially necessaryin the future
to improve the constraints on the mean metallicity-redshift relation for DLAs, and to
quantify potentialdustselectioneects.
References
Kulkarni,V.P.,andFall,S.M.,2002,ApJ,580,732.
Malaney,R.A.,andChaboyer,B.,1996,ApJ,462,57.
Pei,Y.C.,andFall,S.M.,1995,ApJ,454,69.
Pei,Y.C.,Fall,S.M.,andHauser,M.G.,1999,ApJ,522,604.
Rao,S.M.,andTurnshek,D.A.,2000,ApJS,130,1.
Schneider,D.P.etal.,2002,AJ,123,567.
Somerville,R.S.,Primack,J.R.,andFaber,S.M.2001,MNRAS,320,504.
Wolfe,A.M.,Lanzetta,K.M.,Foltz,C.B.andChaee,F.H.1995,ApJ,454,698.