HAL Id: tel-00003969
https://tel.archives-ouvertes.fr/tel-00003969
Submitted on 12 Dec 2003
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Propriétés des absorbants Lyman-alpha à grand décalage
spectral
Céline Péroux
To cite this version:
Céline Péroux. Propriétés des absorbants Lyman-alpha à grand décalage spectral. Astrophysics
[astro-ph]. University of Cambridge, 2001. English. �tel-00003969�
at High-Redshift C eline P eroux Institute of Astronomy & Fitzwilliam College September, 2001
De laration
Thisthesis,entitledProperties of Lyman- Absorbersat High-Redshift, issubmittedfor
the degree of Do tor of Philosophy at the University of Cambridge. The resear h
de-s ribedwasperformedintheInstituteofAstronomybetweenO tober1998and
Septem-ber2001. Thework ontainedinthisdissertationisoriginal,ex eptwhereexpli it
refer-en eto theresultsofothersisgiven. Partsofthiswork,whi hareindi atedinthetext,
wereperformedin ollaborationandsome oftheresultshaveappeared orwillappearin
publishedform:
Celine Peroux, Mike Irwin, Ri hard G. M Mahon & LisaJ. Storrie-Lombardi,
The Evolution andSpa e Densityof Damped Lyman-alpha Galaxies,2000, in
pro- eedings of the Euro onferen e: "The Evolution of Galaxies. I - Observational
Clues"ed. J.M. Vil hez,G. Stasinska andE. Perez.
Celine Peroux, LisaJ. Storrie-Lombardi, Ri hard G. M Mahon & Mike Irwin,
Absorption Systems in the Spe tra of 66z >
4 Quasars,2001, AJ,121,1799.
Celine Peroux, Ri hard G. M Mahon, Mike Irwin & LisaJ. Storrie-Lombardi,
Cosmologi al Evolution of the Universe Neutral Gas Mass Measured by Quasar
AbsorptionSysems,2001,inthepro eedingsofthe"Cosmi Evolution" onferen e,
held at l'Institutd'Astrophysiquede Paris, November13-17, 2000.
Celine Peroux, Ri hard G. M Mahon, Lisa J. Storrie-Lombardi & Mike Irwin,
The Natureof High-Redshift Damped Ly- Absorbers,2001,MNRAS, submitted.
Celine Peroux, Mike Irwin, Ri hard G. M Mahon & LisaJ. Storrie-Lombardi,
Statisti al Properties of DLAs and sub-DLAs,2001, inpro eedings ofthe
"Chem-i alEnri hmentof Intra luster andIntergala ti medium"Vul ano Workshop,ed.
Fran es aMatteu i.
The lengthofthisworkdoesnotex eed 60,000words.
Celine Peroux
Summary
Properties of Lyman- Absorbers at High-Redshift
Celine Peroux
Inre ent years, anextremely su essfulmethod to observationally studyearly
stagesof galaxy formationhasbeenprovidedbythe studyof quasar absorbers. Quasar
absorption lines are systems inter epting our line-of-sight to a given quasar and thus
produ e a feature in the quasar spe trum. Damped Lyman- systems (hereafter
DLAs) have N (H I) > 210 20
atoms m
2
, and were originally thought to be the
pre ursorsofpresentdaydiskgalaxiesbutthereiseviden ethattheymaybedominated
by gas-ri h proto-dwarf galaxies representing the basi building blo ks of hierar hi al
growthofstru ture. Sin etheirdete tionisindependentoftheirsize,shape,and overing
fa tor,they provideaunbiasedmethod withwhi h to studyearlygalaxies. DLAsare a
subsetofLyman-limitSystems(hereafterLLS)whi hhavehydrogen olumndensities
N(H I)>1:610 17
atoms m 2
. At z<1,they areprobablyasso iatedwithgala ti
halos. Finally,theLyman-forest is omposedofmanysmall olumndensitysystems
rangingfromN (HI) =10 12 to 1:610 17 atoms m 2 .
This thesis presents a sample of 66 bright z >
4 quasars observed with the 4
m Cerro Tololo Inter-Ameri an Observatory teles ope and the 4.2 m William Hershel
teles ope. The rst part of the study on entrates on the quasars themselves via the
tting of quasar ontinua and the measurement of ontinuum depression parameters
hara terising the mean absorption a ross the Lyman- forest. The quasar spe tra
are then analysed to investigate the absorption systems they ontain. This led to the
dis overy of 26 new DLAs, 34 LLSand many asso iated metal lines whi h enables the
analysisoftheevolution ofthe olumndensitydistribution,f(N;z),andthetotal mass
in high- olumn density neutral hydrogen quasar absorbers. The observed number of
LLS per unit redshift is used to onstrain f(N;z) below the DLA limit in the range
N(HI) = 1:610 17 to 210 20 atoms m 2
. The joint analysis shows unambiguously
thatf(N;z) deviatessigni antlyfromasingle powerlawandthata -lawdistribution
of the form f(N;z) = (f =N )(N=N ) exp( N=N
) provides a better des ription of
the observations. These results are further used to determine the amount of neutral
gas ontained inbothDLAs and in systems withN(HI) 210 19
atoms m
2
(\sub-DLAs"). Inthe redshiftrange2 { 3,85% of theneutral H I + HeII mass density isin
DLAs,howeverwe ndthatat z>3.5thisfra tion dropsto 55%andthattheremaining
neutral gas mass lies in sub-DLAs. After orre tion of the observed mass in H I for
this\missing"neutralgasthe omovingmassdensitynolongershowsanyeviden e fora
de reaseovertherangez=2{5. The hangeinthe olumndensitydistributionsupports
a pi ture, where at z>3.5, we may be dire tly observing the formation of high olumn
density neutral hydrogen systems from lower olumn density units. Finally, predi tions
on the redshift evolution of the sub-DLAs number density are presented. Preliminary
Contents
Contents v
List of Figures viii
List of Tables x
1 Introdu tion 2
1.1 S ienti Ba kground . . . 2
1.1.1 Quasars: Cosmologi al Lighthouses . . . 2
1.1.2 Quasar Absorbers: UnrevealingtheStru turesof theUniverse . . 5
1.2 FindingHigh-RedshiftQuasars . . . 6
1.2.1 The 2 nd APM Quasar Sample . . . 6
1.2.2 OtherHigh-RedshiftQuasars . . . 9
1.3 Theory ofQuasar Absorbers. . . 10
1.3.1 Classi ationof QuasarAbsorbers . . . 10
1.3.2 AbsorptionLine Formationand ColumnDensityDetermination. . 22
1.4 S ienti Motivation . . . 27 1.4.1 ThesisMotivation . . . 27 1.4.2 ThesisOutline . . . 27 2 The z > 4 Quasar Sample 29 2.1 Introdu tion. . . 29 2.2 Observations . . . 30 2.2.1 WHTRuns . . . 30 2.2.2 CTIO Runs . . . 33 2.2.3 Ke kObservation. . . 33 2.3 TheData . . . 33 2.3.1 Data Redu tion. . . 33 2.3.2 Quasar Spe tra . . . 34
2.4 Redshiftand MagnitudeMeasurements . . . 50
2.4.1 RedshiftMeasurements . . . 50
2.4.2 MagnitudeMeasurements . . . 75
2.5 Notes onIndividual Obje ts . . . 79
2.6 Summary . . . 86
3 Quasar Continua 87 3.1 QuasarContinuum Fitting. . . 87
3.1.1 Introdu tion . . . 87
3.1.4 Analysis . . . 93
3.2 Dete ting Dust inQuasarAbsorbers . . . 96
3.3 Continuum Depression . . . 102 3.3.1 Introdu tion . . . 102 3.3.2 Methodology . . . 103 3.3.3 Measurements . . . 103 3.3.4 Analysis . . . 106 3.4 Summary . . . 110
4 LymanLimit Systems Analysis 114 4.1 Introdu tion. . . 114 4.1.1 Ba kground . . . 114 4.1.2 PreviousSamples . . . 115 4.2 New High-RedshiftLLS . . . 115 4.2.1 LLSdete tion . . . 115 4.2.2 TheSampleof LLS . . . 120 4.3 LLSAnalysis . . . 124 4.3.1 LLSProperties . . . 124 4.3.2 Dis ussion. . . 129 4.4 Summary . . . 129
5 Damped Lyman- Systems Analysis 133 5.1 Introdu tion. . . 133
5.1.1 Ba kground . . . 133
5.1.2 PreviousSamples . . . 134
5.1.3 PreviousResults . . . 135
5.2 New DLA Sample . . . 135
5.2.1 Survey'sSensitivity . . . 135
5.2.2 DLADete tion . . . 137
5.2.3 OtherLinesat theDLAs' Redshift . . . 151
5.3 Metal Systems . . . 151
5.4 DLA Analysis . . . 154
5.4.1 DLAProperties. . . 154
5.4.2 NumberDensityofDLAs . . . 154
5.5 Summary . . . 156
6 Quasar Absorbers: a Study of the History of the Universe 159 6.1 ColumnDensityDistribution . . . 159
6.1.1 Introdu tion . . . 159
6.1.2 PreviousWork . . . 162
6.1.3 Results . . . 164
6.1.4 ComparisonwithModels . . . 168
6.2 Cosmologi alEvolutionof NeutralGas Mass . . . 172
6.2.1 Introdu tion . . . 172
6.2.2 PreviousWork . . . 173
6.2.3 Results . . . 173
6.2.4 Models . . . 187
7 Con lusions and Future Work 192
7.1 Con lusions . . . 192
7.2 Future Work . . . 193
Bibliography 2
A Normalised Quasar Spe tra 13
B Metal systems 27
List of Figures
1 Quasar spe trumovera largewavelength range . . . 3
2 Fra tional look-ba ktime . . . 4
3 Cartoonrepresentationof a quasarsight line . . . 7
4 Colour-magnitude diagram . . . 8
5 Typi alquasar spe trum . . . 11
6 MilkyWayabsorptionfeatures in3C273 spe tra . . . 13
7 Quasar absorbers ross-se tions . . . 13
8 Light elements abundan es . . . 15
9 Simulationsof theLyman-forest . . . 17
10 The z6:28 Sloanquasar . . . 19
11 Expe tedquasar spe trumat z em >z reionisation . . . 20
12 Formation ofVoigt prole . . . 24
13 Variousregimes ofthe urve-of-growth . . . 26
1 Sky lines . . . 35
2 B-star uxstandard and orre tionforatmospheri features . . . 36
3 Fluxed spe tra of allthe observed quasars . . . 38
4 Main quasaremission lines. . . 76
5 Filters usedinvarious surveys. . . 80
6 APM versusspe tralmagnitudes . . . 81
1 F() median omposite spe trum . . . 90
2 F() ompositespe trum . . . 90
3 Comparisonof ompositespe tra fromvarious surveys . . . 91
4 Median ompositespe trumforvarioussurveys . . . 91
5 Continuum slope asa fun tionof thequasaremission redshift . . . 97
6 Continuum slopes: omparing our measurements with Storrie-Lombardi (1994) . . . 97
7 Number ofquasars withorwithoutDLAs asa fun tionof ontinuumslope100 8 KS teston thedistribution ofslopesof quasarwithand withoutabsorbers101 9 Continuum depressionversusthe ontinuumslope . . . 107
11 Continuumdepression parameterversusemission redshift . . . 111
1 ExampleofLLSdete tion/non-dete tion . . . 117
2 LLSnumberdensityandlogarithmi likelihoodparametersestimators ex- ludingsystems within3000 km s 1 of z em . . . 125
3 LLSnumberdensityandlogarithmi likelihoodparametersestimators in- ludingsystems within3000 km s 1 of z em . . . 126
4 Logarithmi numberdensityofLLS . . . 128
5 Cumulativenumberof LLS . . . 130
6 Comparisonofobserved numberdensityof LLSwith\mini-halo"models. 131 1 Surveysensitivityfun tion. . . 136
2 FiguresillustratingDLA olumndensitymeasurements omparison . . . . 139
3 Correspondingspe tra . . . 140
4 DLAdete tion . . . 141
5 BRJ0307 4945 DLA . . . 143
6 Exampleofmediumresolution \lo al" ontinuum t . . . 144
7 ExampleofDLA andidates . . . 152
8 ColumndensitydistributionwithredshiftandnumberofDLAsofagiven olumndensity . . . 155
9 Numberdensityof DLAs . . . 157
1 Distan eintervalasa fun tionofredshift . . . 161
2 The olumn densitydistributionof quasarabsorbersat <z>2:8 . . . . 163
3 Cumulativenumberof absorbers . . . 165
4 Columndensitydistributionof quasarabsorbers forvarious redshiftranges169 5 Dierential olumndensitydistributionforz>3:5 . . . 170
6 SPHsimulationsofthe olumndensityof absorbers . . . 171
7 The\ osmi G-dwarfproblem" . . . 174
8 Massintegralplotfortwodierent redshiftranges . . . 176
9 Numberdensityof DLAs and sub-DLAs . . . 177
10 DLA inanon-zero -Universe . . . 179
11 DLA indierent osmologi al models . . . 180
12 DLA withdierent valuesof theHubble onstant . . . 181
13 DLA freefrom dustbias . . . 185
14 Comparisonofobserved DLA withmodels . . . 188
15 DLA and thestar formationrateevolutionwithredshift . . . 191
List of Tables
1 Journal ofObservations . . . 31
2 Quasar RedshiftMeasurements . . . 73
3 Quasar MagnitudeMeasurements . . . 77
1 Continuum slope () measurements of the quasar sample presented in Chapter 2 . . . 94
2 SDSS ontinuumslope () measurements . . . 98
3 Continuum slope() measurementsof the1 st APMsurvey . . . 99
4 Continuum depressionmeasurements . . . 105
5 Continuum depressionmeasurementsof the1 st APM survey . . . 108
6 Continuum depressionmeasurementsfrom SDSS . . . 113
1 New survey forLyman-limitSystems . . . 118
2 Lyman-limit Systemspreviouslyknown . . . 121
3 LLSnumberdensityredshiftevolution parameters . . . 128
1 MediumandhighresolutionDLA olumndensitymeasurements omparison138 2 SurveyforDamped Lyman AbsorptionSystems . . . 145
3 Metal LinesRest Wavelengths. . . 153
1 Parametert to the olumndensitydistribution . . . 166
2 DLA and DLA+sub DLA values . . . 182
1 Identi ationof Metal AbsorptionLines . . . 18
1 Quasar WithDamped Lyman-alphaAbsorbers- thiswork . . . 48
2 Quasar WithDamped Lyman-alphaAbsorbers- datafrom theliterature 49 1 Quasar Without DampedLyman-alpha Absorbers - thiswork . . . 55
Chapter 1
Introdu tion
`Il n'y a pas de sentiment plus ommunement partage, que de vouloir ^etre
dierent des autres'
Jean-Paul Sartre
In thisintrodu toryChapter, I re all histori al developments inthe dis overy
of both quasars (Se tion 1.1.1) and quasar absorbers (Se tion 1.1.2). In se tion 1.2, I
des ribethete hniqueusedtondthequasarswhi hmakeupthesamplestudiedinthis
thesis. IthendetailthetheoryofabsorptionlinesandVoigtproletting(Se tion1.3.1)
as well as the urve of growth (Se tion 1.3.2). In the last part, I dene the s ienti
motivations for the proje t (Se tion 1.4.1) and emphasise the stru ture of the thesis
(Se tion 1.4.2).
1.1 S ienti Ba kground
1.1.1 Quasars: Cosmologi al Lighthouses
The Third Cambridge (3C) Catalogue prepared in the early 1960s, listed
ra-dio sour es of whi h 10 were extremely small in size(less than 1 ar se ). At the same
time, Hazard et al. (1963) developed a new method, using lunar o ultation, to
a u-rately determine the position of radio sour es and hen e allow for opti al follow-up of
these ompa tobje ts. ThisenabledS hmidt(1963)totakeaspe trumof3C273,whi h
showedastar-likeobje tlyingattheunexpe tedlyhigh(atthetime)redshiftofz=0:16,
implying a distan e of 1 Gp and an opti al luminosity 510 12
L
. Other
simi-lar obje ts were subsequentlyobserved and they be ame known as quasi-stellar obje ts
(QSO)orquasars.
Su h obje ts were (and still are) a hallenge to theorists: how an so mu h
energy omewithsu hrapidvariabilityoutofsu ha ompa tregionandbedistributed
Fig. 2.| Fra tionallook-ba ktime asa fun tionofredshift. M =1:0, =0:0(solid line), M =0:0and
=0:0(dashedline)and M
=0:3and
beams arisesurroundedbyan a retion dis . In any ase, theluminous, ompa t
emis-sion from quasars has made them ideal bea ons from the early Universe. It is indeed
be ause they are extremely luminous that quasars are among the youngest obje ts
ob-served in the Universe. They have now been dete ted out to redshifts of z > 5. The
urrent re ordholder is SDSS1030+0524 (Figure10) at z 6:28 (Fan et al., 2001b), a
redshiftwhi h orresponds to a look-ba k time of over 90% of the age of the Universe
(Figure2).
1.1.2 Quasar Absorbers: Unrevealing the Stru tures of the Universe
Inadditionto beinginterestingobje tsintheirownright,quasars allowforthe
dete tionof mu h fainter systems,observed inabsorptionin theirspe tra.
The nature ofabsorption lineswastheobje t of mu hdebate inthe years
fol-lowingtheir dis overy (e.g. Pettini, 1998). A urrentlya epted hypothesis states that
the lines are intervening and arise in osmologi ally distributed gas between the
ba k-groundquasarand theobserver. Thisimpliesthattheobservedredshiftis osmologi al,
duetotheexpansionoftheUniverse. However,sin ethedensityoflinesasafun tionof
redshiftisobserved toin rease, thisimpliesthatthe ross-se tion oftheabsorbers were
higherinthepast. Earlierhypothesessuggested thattheabsorptionlinesystems
repre-sent material eje ted from thequasar itself. The redshiftwould thenbe kinemati and
represent a Doppler shift from the eje ted material. This would imply a large eje tion
velo ityrequiringavastamountofenergyandalsoame hanismto onnetheeje tato
narrowvelo itydispersions,two on i tingrequirementsdiÆ ulttoa hieve. Atpresent,
the ommonly a eptedpi turefavours theinterveninghypothesis thanksto threemain
arguments:
1. Sargent et al. (1980) analysed the rst homogeneous sample of Ly forest lines
in six quasars. They show that neither the line density nor the equivalent width
distribution vary from quasar to quasar, or with redshift, along a given
line-of-sight. Theyalsofoundno lusteringonalls alesfrom300to 30,000kms 1
. This
la k of any orrelation between line properties and eje tion velo ity is extremely
diÆ ult to explainwithan eje tion model.
2. Boksenberg&Sargent(1978) observed,inemission,thegalaxyNGC3067
respon-sible for the Ca II absorption inthe spe trum of the quasar 3C 232, welloutside
theextentofthegalaxy. Thiswasthersteviden ethatitistheextendedgaseous
halo of the galaxy whi h is responsible for that absorption line observed in the
quasar spe trum. Similarly, more re ent work (e.g.Steidel et al., 1995)showthat
andidate galaxiesareasso iatedwithalmostallof thelowredshiftMgIIsystems.
3. Shaver &Robertson(1983)dis overed ommon absorptionsystemsinthespe tra
of quasar pairs providing indisputable eviden e in favour of the intervening
absorbers (Wolfe et al., 1993; Petitjean et al., 1998). The observation of quasar
pairs, or groups of quasars with small angular separations on the sky, indi ates
thatthe hara teristi sizeof theabsorbersislargerthantheproje tedseparation
of the lines of sight. At present, the number of suitable groups of quasars is too
smallto providedetailed onstraintsonthestru tureofLy omplexes. However,
despite this, Ly- absorbers provide a powerful way to study the orrelation of
baryoni matter at high redshiftand hen e a unique han e to study how galaxy
formation is related to the distribution and dynami s of the underlying matter
eld.
Thisplusothereviden ehasleadtothestandardpi tureofquasarabsorbersas
systemsalongtheline-of-sightbetweentheobserverandaluminousba kgroundquasar.
A artoon representationof thisphenomenonis shownon Figure 3.
1.2 Finding High-Redshift Quasars
Anysurvey for quasar absorbers starts with a survey for quasars. I have not
been dire tly involved in the sear h for the high-redshift quasars whi h make up the
sample presented in this thesis, but for ompleteness and later referen e, this se tion
des ribesthemethodologyusedtondthequasarsample. Mostofthequasarspresented
inthisthesishavebeenopti allysele tedwhi h,be auseofreddeningee ts, meansthat
we maypotentiallymissthemostdustyinterveningobje ts oralternativelythosewhi h
ontain more absorbers along their line-of-sight. Ellison et al. (2001 ) have re ently
undertakenobservationsofradio-sele tedquasarswiththeaimofpin-pointingtheee t
ofdustonquasarabsorbersurveys. InadditionPei&Fall(1995)haveusedself- onsistent
losed-box/in ow-out owgala ti modelstoshowthatthefra tionofmissingDLAsdue
to dust at z = 3 ranges from 23% to 38%. The following se tions brie y des ribe the
methods used to ndthe quasar sample. Storrie-Lombardiet al. (2001) provides more
detailsonndingtheAPMquasarsandFanetal.(1999);Warrenetal.(1991);Kenne k
et al. (1995a); Storrie-Lombardi et al. (1996 ); Zi kgraf et al. (1997); Kenne k et al.
(1995b); Henry et al. (1994); Hook (prep); Hall et al. (1996) give more informationon
thedis overyof theremaining quasars.
1.2.1 The 2
nd
APM Quasar Sample
Reviews on various te hniques used to nd quasars are des ribed in Hook
(1994); Kembhavi &Narlikar(1999). TheAPM quasarswerefoundusingamulti olour
te hnique whi h is the high-redshift (z > 2:2) ounterpart to the Ultra-Violet ex ess
(UVX) method (Sandage 1965). Irwin et al. (1991) developed the so- alled \BRX"
methodwheretheymadeuseofthefa tthatforz>3:9quasars,theregionofthe
spe -trumabsorbed byneutral hydrogen isredshiftedto theB band. A majorityof quasars
Fig. 3.| This artoon illustrates a quasar sight line along whi h various obje ts give
Fig. 4.|AB j
R olour-magnitudediagramforatypi alhighlatitudeUKSTeldused
intheAPMsurvey. Everydete tedB J
,Rmat hedpairofobje ts lassiedasstellaron
in the olour- olour diagram that sele tion in the single olour B R is suÆ ient to
identify them with high probability. Figure 4 shows a B j
,R olour-magnitude diagram
for a typi al high latitude UKST eld. Every dete ted B j
,R mat hed pair of obje ts
lassied as stellar on the R plate is plotted as a small dot. Overlaid as lled ir les
are the omplete southern sample of BRX-sele ted quasars whi h form the 2 nd
part of
theAPM quasar survey (Storrie-Lombardiet al.,1996 , 2001). Oftheroughly 250,000
pairedobje ts on ea h high latitude UKST eld, two-thirds are lassied as stellar on
theR plate and roughly 50,000 of these are brighterthan R =19 { 19.5, the rangefor
theRmagnitudelimitofthesurvey. TheredboundaryforBRX andidatesele tionwas
set to approximately B j
{R = 2.5 to 3 to follow the red extremity of the stellar lo us.
Theee tsof oloursele tiononsample ompletenesshavebeenthoroughlyinvestigated
over the past few years (Warren et al., 1994; Storrie-Lombardi et al., 1996 ; Kenne k
et al.,1995a).
The plates ame from the generi southern sky survey material taken by the
UKS hmidtteles ope(UKST)andweremeasuredandanalysedattheAutomatedPlate
Measuring 1
(APM) fa ility in Cambridge. The total area of Southern high latitude
sky surveyed is roughly 8000 square degrees from a total of 328 UKST elds. Low
resolution( > 10
A)spe tra werethenobtainedto identifythequasars,primarilyat the
LasCampanasObservatory(Storrie-Lombardiet al.,2001). Thiste hniquewasusedto
nd26 ofthequasars whi hmake upthesamplepresentedinthisthesis,theremaining
APMdis overedquasarswereobservedusingtheINTandWHT.Thequasarsdis overed
aspartof theAPM surveys arelabelled\BR"or\BRI"a ordingly.
1.2.2 Other High-Redshift Quasars
Thirty-two of the other quasars making up our 66 z >
4 sample are from
the Se ond Palomar Observatory Sky Survey. Kenne k et al. (1995a,b) 2
arried out
a multi olour survey with the aim of determining the spa e density of bright quasars
(M B
< 27) at z>4. The quasars dis overed aspart ofthe Se ondPalomar Observ
a-torySkySurvey arelabelled\PSS".
Four of the quasars presented in thisthesis are part of the Sloan Sky Digital
Survey(SDSS)whi hmakesuseofthededi atedApa he Point3.5mteles ope. The
sur-veyisa jointproje tofTheUniversityofChi ago,Fermilab,theInstituteforAdvan ed
Study,theJapan Parti ipation Group,The JohnsHopkins University, the
Max-Plan k-InstituteforAstronomy(MPIA),theMax-Plan k-InstituteforAstrophysi s(MPA),New
Mexi oStateUniversity,Prin etonUniversity,theUnitedStatesNavalObservatory, and
theUniversityof Washington. Itaimstomapindetailone-quarteroftheentiresky,
de-terminingthepositionsand absolutebrightnessesofmorethan100 millionobje ts. The
quasars found are often faint and thus not the most appropriate for absorber sear hes
with4m lass teles ope.
1
http://www.ast. am.a .uk/mike/apm at/ 2
Of the four remaining quasars, one is radio-sele ted from the
Parkes-MIT-NRAOsurveyandislabelled\PMN"(Hook,prep)andoneisX-raysele tedandlabelled
\RX" (Henry et al., 1994). The two last obje ts areopti allysele ted byWarren et al.
(1991) and by Hallet al. (1996) as part of theDeep Multi olor Survey (DMS).All the
quasars making upthe samplestudied inthisthesis arelistedinTable2.2.1.
1.3 Theory of Quasar Absorbers
1.3.1 Classi ation of Quasar Absorbers
Quasarabsorptionfeatures anbeorderedinto ategorieshavingdierent
har-a teristi s asillustratedon thetypi al quasarspe trumshowninFigure 5:
I) z abs
<z em
Forabsorptionsystems with wavelength(z abs
) belowthe quasarLy emission
line(z em
),Lynds(1970)suggested thattheabsorptionlineswere aused byLy
transi-tions. Thiswaslater onrmed by two arguments: rstly, thepresen e of theLyman-
line,whenobservable(Baldwinetal.,1974;Oemler&Lynds,1975),andse ondlybe ause
the number density of absorbers blueward of the quasar emission line is mu h greater
thanredward. Sin e theirdete tion isindependent oftheirluminosityand morphology,
these absorbersprovideaunbiasedmethod withwhi h to studyearlygalaxies.
Inafew ases,theLyabsorptionlineisobservedatz abs
>z em
. Theobserved
orrespondingvelo ityrarelyex eeds 2 000 kms 1
in good agreement withtheoreti al
predi tions(Loeb&Eisenstein,1995),althoughquasarredshiftmeasurementsare
some-times subje t to systemati osets. The proje ted velo ity of these louds is dire ted
towards the quasaritself. It ismost probablyasso iated with gasfalling into thebla k
holepotentialwellalthoughmoredetailedstudiesofthisproblemarerequiredtoa hieve
a betterunderstanding ofthisphenomenon.
Ly absorbers are sub-divided into three lasses a ording to their olumn
density, thenumberof hydrogenatomsperunitareaalongtheline-of-sightbetweenthe
observerandthequasar ( ommonlyexpressed inatoms m 2
). Thereforealow olumn
density loud ould eitherbe a small loud with highdensityora large loud withlow
density. They thus probe media spanning the range from voids through to halos and
disks ofbothdwarfandnormal (proto)galaxies.
1. Damped Lyman- systems (hereafter DLAs) have N(HI) > 210
20
atoms
m 2
. Thisdenitionis somewhat arti ial sin edamped wings appearforlower
olumn densities (N(H i) > 10 19
m 2
; see next Se tion and Figure 13). This
1600 1800 2000 2200 2400 2600 2800 3000 3200
0
1
2
3
4
5
Fig. 5.| Typi al spe trum of a quasar, showing the quasar ontinuum,emission lines,
andtheabsorptionlinesprodu edbygalaxiesandintergala ti materialthatliebetween
thequasarand theobserver. This spe trumof thez=1:34 quasarPKS0454+039 was
obtained with the Faint Obje t Spe trograph on the Hubble Spa e Teles ope. The
emissionlinesat2400
Aand2850
AareLyman-andLyman-. Thetwostrongest
at lowresolutionandthistresholdmadethem relativelyunambiguoustopi kout.
The equivalent width is w obs (z 2.5) > 17.5 A for N(H i) > 10 20 m 2 . The
probabilitythat su hastrongabsorptionfeature istheresultofblendingis small.
It is lear that this denition may introdu e a systemati bias in the dis ussion
of the nature of these DLAs and the work presented in this thesis generalises
the denition in order to ompare the properties of systems with 10 19 < N(HI) <210 20 andN(HI) >210 20 .
DLAswereoriginallythoughtto be thepre ursorsofpresentdayL
diskgalaxies,
but there is growing eviden e that they may instead be dominated by gas-ri h
proto-dwarf galaxies representing the basi building blo ks of hierar hi al growth
ofstru ture. HypothesesforthemorphologyofDLAsrangefromlargedisksystems
(Pro haska&Wolfe,1998),tolowsurfa ebrightnessgalaxies(Jimenezetal.,1999;
Bowen et al.,2001;O'Neil,2001)and dwarfgalaxies(Matteu i etal.,1997). The
Milky Way itself is dete ted in the spe trum of the low-redshift quasar 3C 273
(Figure 6). At low redshift, some of the galaxies that are responsible for the
DLA absorption an be dire tly identied (Le Brun et al., 1997; Fynbo et al.,
1999). These galaxies area heterogeneouspopulation: they are notjust themost
luminousgalaxies,butin ludedwarfandlowsurfa ebrightnessgalaxies,andthere
are many ases where no galaxy has been identied to sensitive dete tion limits.
The size of one DLA, observed in PKS0458{020, hasbeen measured from 21 m
absorption observations (see Figure 7) and is found to be greater that 8h 1
kp
(Briggset al., 1989).
The kinemati stru ture of the absorption proles of neutral and low ionisation
spe ies is onsistent with the rotation of a thi k disk(Pro haska & Wolfe, 1997),
so it is possible that these are the z = 3 progenitors of normal spiral galaxies.
However, thissignatureisnotunique. It ouldalso bethe onsequen eofdire ted
infall in an hierar hi al stru ture formation s enario (Ledoux et al., 1998).
Al-though the exa t nature of the quasar HI absorbers is not known, they form a
sampleofsystemsunbiasedasregardsluminosity,spe i morphology,oremission
line strength, thus enabling studies of metalli ity and HI evolution over a large
redshiftrange.
ForN(HI) >10 19
atoms m
2
, theopti aldepth at theLymanlimitis not
ne es-sarily largeenough for the interior to be self-shielded from the external radiation
eld and thusmight notbepredominantlyneutralHI gas. The orre tion forthe
ionisingfra tionissofarunknownandpreventsreliablemetalli itymeasurements
and estimates of the total hydrogen ontent inthese systems. Nevertheless,these
quasarabsorptionlinesareapowerfuldiagnosti toolforinvestigatingthe hemi al
ompositionof highredshiftgalaxies.
Fe abundan e isoftenused to measurestellar metalli itiesandit wouldbe
Fig. 6.| 3C 273 spe trum showing Ly 1216
A absorptionline dueto the MilkyWay.
Thisillustratesthe ommonnatureofhigh olumndensityinterveningabsorbers(Bah all
et al.,1991).
Fig. 7.| Quasarabsorbers ross-se tionsshowing theimpa tparameters at whi h
advantagethat they exhibitarange of rest wavelengths and soareoften dete ted
(Savaglio& et al.,2000). However, the problem withstudying metal abundan es
is the ompli ation of dust depletion, the pro ess by whi h parti les are removed
from the gas phase via ondensation onto grains. Although the abundan es of
Zn and Fe tra k ea h other losely down to metalli ities of < 0:01Z
(where Z
refers to the solar abundan e) inGala ti stars, in the lo al inter-stellar medium
anoverabundan eof Znrelative toFeisoftenobserved. Thisisdueto dierential
depletion onto grains, su h that whilst Fe is usually heavily depleted, very little
Zn is seen to be in orporated into dust (Pettini et al., 1997). For these reasons,
Zn (as well as Cr) are usually adopted as the metalli ity indi ator of hoi e for
quasar absorbers metal abundan es. It therefore follows that the relative
abun-dan es of [Zn/Cr℄ and [Zn/Fe℄ will provide an estimate of the fra tions of these
refra toryelementswhi haremissingfrom thegas-phase. Su h studiesshowthat
at z abs
> 1:5, DLAs are generally metal poor, typi ally 1/10 of solar, with small
amountsof dustdepletion. Theselow abundan esseemto persistforall redshifts
observed, withno sign of metalli ityevolution when olumn density weightedZn
abundan esare onsidered.
DLA star formationhistories an thusbe derived from metal studies. -elements
(O, Si, S, Ar) are believed to be produ ed in Type II supernovae (SN II) after a
relativelyshortlifetimeof10 7
years. MostoftheFe-peakelements(Fe,Zn) ome
fromthelonger-lived(10 9
years)progenitorsofTypeIasupernovae(SNIa). The
dierent lifetimes of supernovae progenitors means that an over-abundan e of
is observed relative to Fe at low [Fe/H℄. More re ently, mole ules have also been
observed in DLAs. Measuring mole ules at high-redshift is important be ause
they dominate the ooling fun tion of neutral metal-poor gas. Petitjean et al.
(2000) have dete tedmole ularhydrogen(H 2
) in11 absorbersand usethese data
to dedu e that most of the DLA systems arise in warm and diuse neutral gas.
Thesemeasurementsallowinvestigationof thepro essesof dustformationaswell
as oolingand photodisso iationfrom therst stars(Levshakovet al., 2000).
2. Lyman-limit Systems (hereafter LLS) have hydrogen olumn densities N(HI)
>1:610 17
atoms m 2
andareopti allythi kat theLymanlimitduetotheHI
photo-ionisation: H o + !H + +e (1.1)
where the photon energy, 13.6 eV, orresponds to 912
A rest wavelength. These
absorbers areeasily identiableby theirdistin tive breaksignaturein the quasar
Fig. 8.|Abundan esexpe tedforthelightnu lei 4 He,D, 3 Heand 7 Li(toptobottom)
al ulated in standard Big Bang Nu leosynthesis. The 95% onden e intervals are
shown by the verti al widths of the abundan e predi tions. The horizontal s ale is
expressed inunits of thebaryon densityor riti al density fora Hubble onstant of 65
where is the HI photo-ionisation ross-se tion 6:810 18 m 2 . For 912 = 1
(opti ally thi k), N(HI) has to be = 1:610 17
atoms m
2
. The opti al depth
below 912 A is proportional to 912 (=912) 3 . For example, if a LLS lies at z =3
with a olumn density N(HI) = 10 18
atoms m
2
, the ontinuum rea hes zero
( 912
= 6) at 3648
A and the ux only re overs at around 2000
A ( = 1). The
dete tion of these systemsis thusfairlyeasy, even inmediumresolutionspe tra.
\Grey"LLShaveopti aldepth <1andthusprodu eapartialbreakinthequasar
spe trum. Be ause of theirrelativelysmall olumndensity, they provide some of
the best andidates for measurement of the primordial abundan e of deuterium
(see Figure 8 and Molaro et al., 1999; Levshakov et al., 2000; O'Meara et al.,
2001, and referen esherein). The rst synthesis of light elements (D, He and Li)
took pla e in the early Universe and heavier elements have then been produ ed
through stellarnu leosynthesis. High-resolution observations of quasar absorbers
anbeusedto determinetheprimordialabundan esofelementsformedintheBig
Bang, whi h provides a fundamental tool for testing the Big Bang theory and a
unique measure of the baryoni density of the Universe, b
h 2
. Note that it has
beendemonstrated re ently(D'Odori o et al., 2001; Pettini &Bowen, 2001)that
DLAs and sub-DLAs an also be used for D measurements. Pettini & Bowen
(2001) omputed theweighted mean of the6 D measurements urrentlyavailable
at high-redshiftand found:
D=H =(2:20:2)10 5
(1.3)
The ionisationstate of LLSs an be onstrainedbymeasuringthe olumn density
of the same ion in dierent ionising states (i.e. Fe +
and Fe ++
) and omparing
theCLOUDYsoftwarepa kage (Pro haska,1999). Inaddition,LLSprovide
infor-mation on the ionisingphotons of theintergala ti medium. The exa t nature of
LLSis not known but, at z<1, they are probablyasso iatedwith gala ti halos
(Steidelet al.,1994).
3. Finally, the Lyman- Forest is omposed of manylow olumn density systems
rangingfromN(HI)=10 12
to1:610 17
atoms m 2
. It is urrentlybelievedthat
theabsorption intheLy forest is ausednot by individual, onned louds, but
byagraduallyvaryingdensityeld hara terizedbyoverdensesheetsandlaments
and extensive, underdensevoids whi h evolve with time(Carswell& Rees, 1987),
as shown in Figure 9. Rau h (1998) provides an ex ellent review of our urrent
knowledge of theLy forest.
Theobservationofmultiplelinesofsighthasbeensu essfulindeterminingthesize
of the Ly louds(see x1.1.2). Inparti ular, D'Odori o et al. (1998) analysedall
thedataavailableatthetimeand on ludedthatLyforest loudshaveatypi al
size of 350h 1
kp for a spheri al geometry or to 400h 1
PG1634+706
z=3.63
z=1.33
Q1422+2309
z=3
z=1
Fig. 9.|Illustrationofstru tureevolutionofintergala ti gasfromhightolowredshift.
Theupper spe trumof a z=3:6 quasaris a Ke k/HIRES observation,whilethe lower
spe trumisaFOS/HSTobservationsofaz=1:3quasar. Higherredshiftquasarsshowa
mu hthi kerforestofLyman-lines. Sli es throughN{body/hydrodynami simulation
resultsat the two epo hs z =3 and z = 1 are shown inthe right{hand panels. Three
ontour levelsare shown: 10 11 m 2 (dottedlines),10 12 m 2
(solid lines)and 10 13
m 2
geometry. Line ofsight lusteringdistributionanalysis isalsopossiblewitha
two-point orrelation fun tion. Thereareseveral ontradi toryresultsintheliterature
but the onsensus is for weak lustering on small s ales, with little eviden e for
orrelationsoverlarger (v>300 km s 1
) s ales.
At rst, the Ly forest was thought to ontain pristine material, but tra es of
metals were dete ted later on (Meyer & York, 1987). Ellison et al. (2000), have
used two dierent methods to dete t CIV in the forest: in the rst approa h, a
highsignal-to-noiseratiosta ked spe trumprodu edby ombiningthedataoftwo
quasarsleadstonodete tionofsigni antmetalsandinthese ondapproa h,
mea-surementsof individualpixel opti al depthsshow thatthere are indeedCIV lines
in the Ly forest. This work illustrates the diÆ ulties en ountered in dete ting
metals inthe Lyforest. These observationsmake possiblethestudyof hemi al
enri hment in the inter-gala ti medium and hen e the epo h at whi h the rst
generationsofstarsformedandthendistributedtheirmetalsintothesurrounding
environment (Ferraraet al.,2000).
Gunn-Peterson Ee t
ObservingtheLyforestathigh-redshiftsgivesusdire tindi ationsofthe
ionisa-tion state oftheearly Universe. We see some ontinuumbluewardofthe quasar's
Ly emission ontrary to what one would expe t from a neutral medium (Gunn
& Peterson, 1965). This indi ates that the inter-gala ti medium (IGM) is
pre-dominantlyionised hydrogeneveninthehighest-redshiftquasars,althoughre ent
observationsatz6showtheabsen eof uxintheforestoveralargeregion
(Fig-ure10 Be ker et al.,2001;Djorgovski et al.,2001). Sin e thehighenergyphotons
requiredtofullyioniseheliumaremu hrarerthanthose apableofre-ionising
hy-drogen,aGunn-PetersonHe troughispredi tedto bepresent atlowerwavelength
thanthe orresponding Hfeature (Outram, 1999). Gunn-Peterson absorptionhas
been dete ted for the He II Ly ( rest
= 303:7822
A) at z 3 (Jakobsen et al.,
1994;Reimersetal.,1997). Similarly,ifweweretoobservethespe trumofaquasar
before there-ionisation of theUniverse, the ux blueward of thequasar emission
isexpe tedtobealmostfullyabsorbedleadingtotheso- alled\Lyman-prairie"
(Loeb, 1999).
Proximity Ee t
Anotherobservational hara teristi oftheLyforestisthede reasinglinedensity
at theimmedidatevi inityofthe quasarrstdis overed byCarswellet al.(1982).
ThisisprobablyduetothequasarLyman ontinuumradiationboostingthe
meta-gala ti ionising eld J
. Measurements of the so- alled \proximity ee t" have
beenused to inferthe uxof the ionisingba kground, J
'10 21
(Bajtlik et al.,
1988;Willigeret al.,1994). ThesemeasurementsofJ
suggestthatknownquasars
Fig. 10.|Opti aland near-IR spe trumof SDSS1030+0524, thehighest-redshift(z
6:28) quasar urrently known dis overed as part of the Sloan Digital Sky Survey (Fan
etal.,2001b). AKe kspe trumof theobje t(Be ker etal.,2001) showstheabsen eof
uxoveralargeregion,bluewardofthelyman-emissionline. Thissuggests apossible
rstdete tion ofthe ompleteGunn-Petersonee t ( ausedbyneutralhydrogeninthe
DETERMINING THE REIONIZATION REDSHIFT
HI
HII
Spectrum
SOURCE
λ
α
(1+ zreion
)
Lyα forest
Lyβ
Prarie
)
(1+ zs
λ
β
Ly
α Prarie
Ly
α
λ (1+
α
zs
)
1+ zreion
1+ zs
1 <
<
λ
α
λ
β
= 1.18
Fig. 11.| Expe ted quasar spe trum at z
em
slightly above the redshift of the
II) z abs
>z em
Fewer absorption lines are observed at wavelengths longward of the quasar
emissionfeature. These arenot dueto neutral hydrogen but areasso iatedwith metal
systems. C, N,O, Si, Fe, Al, Mg, butalso Ni, Zn, Cr,S and more re entlyCo (Ellison
et al.,2001a) are dete ted. These areoften multiplesystems orresponding to dierent
loud omponents whi h an be resolved in high-resolution data. Magnesium II and
Carbon IV are hara teristi doublets and are easily dete table even in low resolution
quasarspe tra:
1. CIV doublet (restframe 1548
A and 1551
A):Simulationshave shownthat
theobservedCIVkinemati stru tureand olumndensities anbewellreprodu ed
by merging of proto-gala ti lumps (Haehnelt et al., 1996). Compa t halos of
hot gas with temperatures lose to 10 5
K seem to su essfully a ount for the
observed multi- omponent nature of the CIV absorbers. Based upon the number
densityof absorbers, dN=dz, whi h gives the average number of systems perunit
redshift path, the sky-proje ted ross se tions an be al ulated for an assumed
galaxy luminosity fun tion. The CIV absorption-sele ted systems (to rest-frame
dete tion sensitivity of 0.4
A) are thus inferred to have a 70 kp diameter (see
Figure 7). The methodsusedto studytheproperties of CIV systems an be split
intotwodierentapproa hes. Therstapproa h onsistsofobservingCIVinhigh
signal-to-noise ratio, high-resolution spe tra and determining its olumn density
down to log N(CIV) 11:5 atom m
2
. This allows (i) detailedkinemati s and
temperature studies (Rau h et al., 1996) whi h shows that CIV omponents may
be the building blo ks of future normal galaxies; (ii) the determination of the
lowend of the CIV olumndensity distribution(see Ly forest se tion); (iii)the
study of velo ity stru ture within the halos (Petitjean & Bergeron, 1994; Crotts
et al., 1997), but is limited to a few lines of sight. The se ond approa h onsists
of studying a statisti ally signi ant sample of absorbers by onstru ting a large
homogenous sample(Sargent etal.,1988;Steidel,1990a). Thismethodhasproved
su essful at des ribing the number density and lustering properties of metals,
but previousstudies were restri ted to therange z <
3:5 (Sargent et al., 1988)or
were inhomogenous (Quashno k et al., 1996). The data presented in this thesis
provide a sample of 80 CIV absorbers in the redshift range 3:0 < z < 4:5 whi h
ould beused to studythe hara teristi sand evolution ofthe gala ti halos.
Along with the CIV doublet, another very important transition is the CII line.
The temperature of the Cosmologi al Ba kground Radiation, T CMB
an also be
measured using the abundan e ratio of ex ited states of CII. This puts a dire t
onstraintontheBigBang,theoryalthoughonlyonemeasurementhasbeenmade
so far: Srianand et al. (2000) have derived 6:0< T CMB
<14 K at z 2:3 when
9.1 Kisexpe ted intheHotBig Bang osmology.
2. MgII doublet (rest frame 2796
A and 2803
have been dete ted in emission at z 0:6 (Bergeron & Boisse, 1991). The most
extensive MgII surveys have been ondu ted by Lanzetta et al. (1987); Sargent
etal.(1988);Petitjean&Bergeron(1990);Steidel&Sargent(1992);Chur hilletal.
(2000a,b). Usingsimilarassumptionsasinthepreviousparagraph(numberdensity
of absorbersand S he tergalaxy luminosityfun tion),MgII systems arefoundto
be40kp indiameter(dete tionsensitivityof0.3
A)or60kp (dete tionsensitivity
of0.02
A).The MgIIabsorbinggalaxiesappeartobe onsistent witha normal0.7
L
Sbgalaxy havinga roughly onstant star formationrate sin ez1. The data
presentedinthisthesisprovideasampleof48MgIIabsorbersintheredshiftrange
1:3<z<2:2. An ex ellent review of our urrent knowledge on thetopi of MgII
absorbers ompiled by ChrisChur hill(1999) is available on-line at the following
address: http://www.astro.psu.edu./users/ w /qsogroup/mgii-over.html
III)z abs
z em
Asso iated systems are, by denition, at the same redshift as the quasar.
They ould beexplained by(Petitjean, 1999):
galaxies whi hare partof thequasar luster
gasfrom thegalaxy withinwhi h thea tive gala ti nu lei(AGN) is embedded
gaseje ted bythequasar itself
Be ausetheyare loseto thequasar,these systemsareoftenmoreionisedthan
other metallines andhave higherheavy element abundan es(Petitjeanet al.,1994).
IV) Broad Absorption Lines
Broad Absorption Lines (hereafter BALs) are observed in about 10% of
all quasars (Weymann et al., 1991). They are hara terised by troughs with out ow
velo itiesupto 60,000 kms 1
bluewardofthe quasaremission redshift. Theyare often
highlyionisedandhavehighmetalli itiessuggestingthatBALs haveare onne tedwith
the nu lear region of the quasar (the a tive gala ti nu lei). Although some of these
absorption lines may belong to external galaxies lose in velo ity spa e to the quasar,
there is dire t eviden e from time-variabilityof the linestrength (Hamannet al.,1995;
Barlow et al., 1997; Vilkoviskij & Irwin, 2001) or partial overage of the ba kground
sour e (Ganguly et al., 1999) that many of these lines are physi ally asso iated with
the quasar. BALs have previously been thought notto o ur inradio-loud quasars but
re ent dis overies byBe keret al. (2000) indi atethatthisis notalwaysthe ase. This
nding is problemati for simpleunied models in whi h BAL quasars are a subset of
quasars seen nearlyedge-on and thusraises further questionsaboutthenature of these
Absorption Line Prole
The spe tral uxintensity, F(), an be expressedinterms ofthe unabsorbed
ontinuumintensity, F 0
(),and thefrequen y:
F()=F 0
()e ()
(1.4)
where(),theopti al depth,is expressedasa fun tionof numberdensityof atoms, n
(ornumberofatomspersurfa earea, i.e. olumn densityN) and ross-se tion ():
()= Z
+1
0
n()ds=N() (1.5)
Two dierent pro esses lead to the line broadening that gives absorption
fea-turestheir hara teristi prole:
Lorentzian Prole
Natural (damping) broadening of an absorptionlineis due to the intrinsi
un-dertainty E in the energyof the upperatomi level as expressed bythe Un ertainty
Prin iple: Eth. Thisleadsto a Lorentz prole:
L ()= e 2 m e ! f os =4 2 ( 0 ) 2 +( =4) 2 (1.6) where m e
and e are the mass and harge of an ele tron respe tively, is the
speedof light, f os
is the transition os illator strength, 0
is the entral frequen y and
is the total damping onstant, i.e. the re ipro al of the mean lifetime of the upper
energystate.
Gaussian Prole
Within the loud that we are observing via quasar absorbers, the ions may
have a hara teristi radialvelo ityrelativetotheobserver,resultinginaDoppler-shift.
Theseinternalmotions an be hara terisedasa Gaussianvelo itydistribution:
P(v)= 1 b p e (v=b) 2 (1.7)
wheretheDopplerwidth,b,isdetermined by ontributionsfrom boththermal
andturbulent motionswithinthe absorbing loud:
b= q b 2 thermal +b 2 turbul ent = s 2kT m ion +b 2 turbul ent (1.8)
V2
V1
V2
The Gaussian function is centered at V=0
The Lorentziean functions are centered
at the velocity of the absorbing atom
V1
Absorption at V1
Atoms 1 dominate -> lorentzian shape
Absorption at V2
1 2
1
2
1: gauss(0)*lorentz(V1)
2: gauss(V1)*lorentz(0)
---Atoms 2 dominate -> gaussian shape
1: gauss(0)*lorentz(V2)
2: gauss(V2)*lorentz(0)
---Fig. 12.| Formationof Voigtproles fromthe onvolution ofGaussian andLorentzian
fun tions. At V1,theLorentzianfun tionfallsomoreslowlyatlarge thanthe
Gaus-sianprolethatdes ribesDopplerbroadening. Itisthusthelatterwhi hdominatesthe
absorptionprole. However, at V2,theprominentdampingwings ompletelydominate
the outer parts of the line prole leading to a Lorentzian shape. This orresponds to
Voigt Prole
Convolvingthenatural (Lorentz prole,see equation 1.6)and Doppler (Gauss
prole,see equation1.7) broadening produ esa Voigt prolewithan opti aldepth,:
()= p e 2 m e Nf os b H(a;u) (1.9) where: H(a;u)= a Z +1 1 e y 2 dy (u y) 2 +a 2 (1.10) and: a= 4 ; u= ( 0 ) 0 b ; y =v=b (1.11)
The ore of the Voigt fun tion is thus Gaussian, while the extended wings
of the prole are Lorentzian. This is illustrated by Figure 12: at V1, the Lorentzian
fun tionfallsomoreslowlyatlarge thantheGaussianprolethatdes ribesDoppler
broadening. It is thus the latter whi h dominates the absorption prole. However, at
V2,theprominentdampingwings ompletelydominatetheouterpartsofthelineprole
leadingtoa Lorentzian shape. This orrespondsto veryhigh olumndensityabsorbers,
theso- alledDamped Lyman-systems.
The opti aldepth at theline entreis thengiven by:
( 0 )= p e 2 m e Nf os 0 b =1:49710 15 N( m 2 )f os 0 ( A ) b(kms 1 ) (1.12)
Absorption linesin quasar spe tra are ommonly tted withtheoreti al Voigt
proles,althoughthisse tionshowsthatthisisbasedontheassumptionthatthevelo ity
distributionofthe atomsisdes ribed bya Gaussianfun tion.
The Curve of Growth
Atmediumspe tralresolution,itisnotalwayspossibleto taVoigt proleto
theabsorption feature. In that ase, we make use of the urve of growth whi h relates
theequivalent width of theabsorbers with its olumn density N (the equivalent width
istraditionallyusedalthoughmeasuringtheFullWidthHalf Maximum, FWHM,ofthe
linewouldbemoreappropriateasitislessdependentuponthe ontinuumposition). In
general, theequivalentwidth ofan absorptionline, W(),isdenedas:
W obs ()= Z F 0 F() F 0 d obs = Z (1 e () )d obs (1.13)
wheretheobserved equivalent widthis:
W ()=W
rest
-200 -150 -100 -50
0
50 100 150 200
0
.5
1
-100 -50
0
50 100
.4
.6
.8
1
1.2
-300
-200
-100
0
100
200
300
0
.5
1
12 13 14 15 16 17 18 19 20 21
-2
-1
0
1
Fig. 13.| Illustrationofthedierentregimesof the urve ofgrowth. Themiddlepanel
shows the urve ofgrowth fortheHI Lyman-transition,relatingtheequivalentwidth,
W(),oftheabsorptionproleits olumndensity,N(HI).Thedierent urvesrepresent
four dierent values of the Doppler parameter: b = 13, 23, 53, and 93 kms 1
. The
upper panel shows absorption proles with Doppler parameter b = 23 kms 1
for the
seriesofneutralhydrogen olumndensities N(HI)=10 12 {10 20 atoms m 2 . Thethi k
(thin) urves orrespondtothelled(open)pointsontheb=23 kms 1
urveofgrowth
(middle panel), starting at N(HI) = 10 12 atoms m 2 . For N(HI) <10 13 atoms m 2 ,
known as the linear part of the urve of growth, theequivalent width does notdepend
on b. Thelowerleftpanelshowsthat, at xedN(HI), thedepthoftheproleissmaller
forlargeDopplerparameter, b,su hthattheequivalentwidthremains onstant. Onthe
atpartofthe urveofgrowth,prolesaresaturated andtheequivalentwidthin reases
withbfor onstantN(HI).ForN(HI) > 10
19:5
atoms m 2
,theproledevelopsdamping
The equivalent widthof an absorptionlineis thusindependent of thespe tral
resolutionsin eitisanintegralover. TheHILy urveofgrowthisshowninFigure13.
Therearethree distin tregimes:
1. The Linear Part. The lines in this regime are unsaturated and orrespond to
absorbers with small olumn densities (N(HI) <10 13
atoms m
2
). Be ause the
feature is opti ally thin, the equivalenth width is not dependent on the Doppler
parameter b: N =1:1310 20 W() 2 0 f os (1.15)
2. The Flat Part. The lines in this regime are saturated and dominated by the
Doppler ontribution (see Se tion 1.3.2 above). Their olumndensity, N, depend
on theDoppler parameterb at agiven equivalenth widthW():
W() 2b 0 s ln 0:5 e 2 N 0 f os m e b (1.16)
In order to reliably determine the olumn density of su h absorption systems,
higher-order Lyman series lines whi h have smaller os illator strength, f, (and
thuslieon thelinear partof the urve ofgrowth)arerequired.
3. The Damping Part. The lines in this regime are saturated and dominated
by the Lorentzian damping wings (see Se tion 1.3.2 above). They orrespond to
high olumn densities (N(HI) > 10
19
atoms m
2
) and their equivalent width is
proportionalto the olumndensityindependentlyof b-value:
N(HI)=1:8810 18 W 2 0 ( A ) m 2 : (1.17) 1.4 S ienti Motivation 1.4.1 Thesis Motivation
Oneofthefundamentalphenomenastillpoorlyunderstoodin osmologyisthe
detailed pro ess of the origin of stru ture formation. Signi ant theoreti al progress
hasbeenre entlya hieved withthedevelopmentofSmoothParti leHydrodynami and
Semi-Analyti alsimulations. ButobservationallyprobingtheearlyUniverseis ru ially
dependentuponnew methodsto over omethenatural hallengesofhigh-redshift(z4)
observations and upon the advan ement of teles ope-related te hnologies. In re ent
years, an extremely su essful method to observationally study early stages of galaxy
formationhasbeenprovidedbythe studyofquasar absorbers.
statisti-by the apparent de it of high olumn density systems in the early Universe
(Storrie-Lombardiet al.,1996a;Storrie-Lombardi&Wolfe,2000). Inparti ularwe aimto study
in more detail the evolution with redshift of the olumn density distribution, number
density, and omovingmassdensityofhigh olumndensityHI absorptionsystems. The
aim of our new survey for quasar absorbers is to better understand the high-redshift
endof the massdensityof neutral hydrogen by signi antlyimprovingthestatisti s at
z >
3:5. Severalfundamentalquestionsremainin luding: lo atingtheepo hofDLA
as-sembly; larifyingtherelationshipbetweenLymanlimitsystems and dampedabsorbers
via a detailedstudyofthe olumndensitydistributionof quasarabsorbers andits
evo-lution with redshift; and measuring the total amount of neutral hydrogen ontained in
quasarabsorbersand studyinghowthisvarieswithredshift. Thisthesisemphasizes the
impa t ofournew surveyon these issues.
1.4.2 Thesis Outline
Thisthesis is organisedas follows. In Chapter 2,we providethedetailsof the
set-up for ea h observational run, des ribe the data redu tion and present the quasar
spe tra together with redshift and magnitude measurements. Chapter 3 presents
on-tinuumttingand ontinuumdepressionmeasurementsofallourquasarspe tra. These
measurements are then ompared with the most re ent simulations. The sample of
Lyman-limitsystems isintrodu edinChapter4 whi halso in ludesan analysisof their
numberdensityand olumndensitydistribution. DampedLy-absorbersarepresented
inChapter 5 togetherwith astudy oftheirproperties. Metal absorbersdete ted inthe
quasar spe tra are also listed. The osmologi al neutral gas mass evolution and
impli- ations of our results for theories of stru ture formation are detailedin Chapter 6. In
Chapter 7, the on lusions of this thesis are summarised and a brief dis ussion of the
extensionof thiswork to on-goingand future proje tsis given.
Thiswork assumesH 0 =65kms 1 Mp 1 , M =0:3 and =0:7, although
pre-Chapter 2
The z
>
4 Quasar Sample
`La souran e estdans la solitude qui l'a ompagne'
AndreMalraux
This Chapter details the observations of the 66 z >
4 quasar studied in this
thesiswhi hwereundertakeninfourdierentruns. Alltheobservationswere arriedout
duringtwoobserving runsat the 4.2mWilliam Hers helteles ope (WHT) of theIsaa
NewtonGroupofteles opesintheCanaryIslands(Se tion2.2.1)andtwoobservingruns
at the Blan o 4 m teles ope at the Cerro Tololo Inter-Ameri an Observatory (CTIO)
in Chile (Se tion 2.2.2). Se tion 2.3.1 gives details on the data redu tion pro ess and
theredu edspe tra arepresentedinSe tion 2.3.2. The redshiftand magnitudeof ea h
obje t have been measured and are tabulated in Se tion 2.4.2 and 2.4.3, respe tively.
Se tion2.5givesdetailson individualquasarspe tra.
2.1 Introdu tion
This hapter presentsthehigh signal-to-noise,5
Aresolution (FWHM) spe tra of 66
z >
4 bright quasars obtained with the 4 m Cerro Tololo Inter-Ameri an Observatory
and 4.2mWilliam Hershel teles opes 1
. The primary goalof these observations wasto
undertake a new survey for intervening absorption systems dete ted in the spe tra of
ba kgroundquasars. We lookforbothLyman-LimitSystems ( olumn densities N HI 1:610 17 atoms m 2
- see Chapter 4) and Damped Ly Systems ( olumn densities
N HI 210 20 atoms m 2
-seeChapter 5). Ten ofthequasarspresentedhereexhibit
intrinsi broadabsorptionlines (BAL).
1
ThissampleisbasedonobservationsobtainedattheWilliamHers helTeles opewhi hisoperated
onthe island of La Palma by the Isaa Newton Groupinthe Spanish Observatorio delRoque de los
Mu ha hosoftheInstitutodeAstrosi a deCanarias,onobservationsmadeattheCerroTololo
Intra-Ameri anObservatorywhi hisoperatedby theAsso iationofUniversities forResear hinAstronomy,
undera ooperative agreement withthe National S ien eFoundationas part of the NationalOpti al
AstronomyObservatoriesandondataobtainedatthe W.M.Ke kObservatory,whi hisoperatedasa
s ienti partnershipamongtheCaliforniaInstituteofTe hnology,theUniversityofCaliforniaandthe
30 CHAPTER2. THE Z 4QUASAR SAMPLE
Any survey for quasar absorbers begins with a sear h for bright quasars and
so onstitutesan ambitiousobservationalprogram. Weobservedsixty-sixz >
4quasars
dis overed by various groups (Fan et al., 1999; Warren et al., 1991; Kenne k et al.,
1995a; Storrie-Lombardi et al., 1996 ; Zi kgraf et al., 1997; Kenne k et al., 1995b;
Henry et al., 1994; Hook, prep; Hall et al., 1996) almost all of whi h have not been
previouslystudied at su h resolution( 5
A)and signal-to-noise(ranging from 10{30).
We obtained opti al spe tra at the 4.2 m William Hers hel Teles ope for the
north-ern quasars and at the 4 m Cerro TololoInter-Ameri an Observatory for the southern
obje ts. More information about z >
4 quasars is available at the following URL:
http://www.ast. am.a .uk/quasars.
2.2 Observations
Highsignal-to-noiseopti alspe trophotometrywasobtained overing
approxi-mately3500
A to 9000
A, theexa t range dependingon whi h instrumentwas usedfor
theobservations. A journalofthe observationsispresentedinTable 2.2.1.
2.2.1 WHT Runs
Thirty-one(in ludingthemis lassiedz=1.90quasarPSSJ0052+2405)quasars
wereobservedattheWHTduring1998September22-24and1999Mar h18-19. The
in-tegrationtimesweretypi ally1800{3600se onds. WeusedtheISISdouble-spe trograph
whi h onsistsof two independent armsfed via adi hroi allowingforblue and red
ob-servationstobe arriedoutsimultaneously. Gratingswith158linesmm 1
andadi hroi
to splitthelight at5700
A wereused. Thisgivesadispersionof 2.89 A pixel
1 inthe
red arm and 1.62 A pixel
1
in the blue. The gratings were arranged so that the blue
partof thespe trumwas entered on 4500
Awhiletheredwas entered on 7000
A.On
thebluearm athinned oatedEnglishEle tri Valve(EEV)20484096 CCDwith13.5
mpixelswasusedasdete tor. Ontheredarm athinned oatedTektronix10241024
CCD with 24 m pixels was used. All the narrow-slit observations were taken with a
slitwidthof 1.2{1.5ar se whilethewide-slit observationswere arried outwitha slit
widthof 5{ 7ar se . Blind-osettingfrom bright15{17 th
magnitudestellardu ials
wasusedto positionthequasarsintheslitpartlytosavea quisitiontimeandpartly
Table 1: JournalofObservations
Quasar Teles ope Observing Exp. Time Wide Ref
Name Date B/R
a
(se s) Slit
PSSJ0003+2730 WHT 1998 Sep22 3600/3600 yes 1
BRJ0006 6208 CTIO 1998 O t 14 3600 yes 3
BRJ0030 5129 CTIO 1998 O t 15 3600 yes 3
PSSJ0034+1639 WHT 1998 Sep22 3600/3600 yes 1
SDSSJ0035+0040 CTIO 1999 O t 10 8100 yes 4
PSSJ0052+2405 WHT 1998 Sep23 3600/3600 yes 2
QJ0054 2742 CTIO 1999 O t 12 2700 yes 5
PSSJ0106+2601 WHT 1998 Sep24 3600/3600 yes 1 PSSJ0131+0633 CTIO 1999 O t 12 3600 no 2 PSSJ0133+0400 CTIO 1999 O t 12 3600 no 2 PSSJ0134+3307 WHT 1998 Sep22 3600/3600 yes 1 PSSJ0137+2837 WHT 1998 Sep24 5200/3600 yes 1 PSSJ0152+0735 WHT 1998 Sep24 3600/3600 yes 1 PSSJ0207+0940 CTIO 1999 O t 12 3600 no 2 PSSJ0209+0517 CTIO 1999 O t 12 2400 no 2 SDSSJ0211 0009 CTIO 1999 O t 10 8100 no 4
BRJ0234 1806 CTIO 1999 O t 09 5400 yes 3
PSSJ0248+1802 WHT 1998 Sep22 3600/3600 yes 6
BRJ0301 5537 CTIO 1998 O t 16 3600 yes 3
BRJ0307 4945 CTIO 1998 O t 14 5400 yes 3
SDSSJ0310 0014 CTIO 1999 O t 9 9900 no 4
BRJ0311 1722 CTIO 1999 O t 11 3600 yes 3
PSSJ0320+0208 CTIO 1999 O t 12 3600 no 2
BRJ0324 2918 CTIO 1999 O t 11 3600 yes 3
BRJ0334 1612 WHT 1998 Sep23 2740/1650 no 3
SDSSJ0338+0021 Ke k 1999 Feb 17 3000/3600 no 4
BRJ0355 3811 CTIO 1998 O t 15 3600 yes 3
BRJ0403 1703 WHT 1999 Sep19 1800/1800 no 7
BRJ0415 4357 CTIO 1998 O t 16 5400 yes 3
BRJ0419 5716 CTIO 1998 O t 14 3600 yes 3
BRJ0426 2202 CTIO 1999 O t 11 3000 yes 3
PMNJ0525 3343 CTIO 1998 O t 15 3600 yes 3
BRJ0529 3526 CTIO 1998 O t 14 5400 yes 3
BRJ0529 3552 CTIO 1998 O t 15 3600 yes 3
BRJ0714 6455 CTIO 1998 O t 15 3600 yes 3
PSSJ0747+4434 WHT 1998 Sep22 1800/1800 no 1 RXJ1028 0844 WHT 1999 Mar 19 2700/2700 yes 8 PSSJ1048+4407 WHT 1999 Mar 19 2700/2700 yes 9 PSSJ1057+4555 WHT 1999 Mar 19 1800/1800 yes 9 PSSJ1159+1337 WHT 1999 Mar 18 2700/2700 yes 1 PSSJ1253 0228 WHT 1999 Mar 18 2700/1800 yes 2 BRJ1310 1740 WHT 1999 Mar 19 2700/2700 yes 3 BRJ1330 2522 WHT 1999 Mar 19 2700/2700 yes 3
32 CHAPTER2. THE Z 4QUASAR SAMPLE PSSJ1438+2538 WHT 1999 Mar18 2700/2700 yes 9 PSSJ1456+2007 WHT 1999 Mar18 2700/2700 yes 1 BRJ1603+0721 WHT 1999 Mar19 2700/2700 yes 3 PSSJ1618+4125 WHT 1999 Mar19 2700/2700 yes 1 PSSJ1633+1411 WHT 1999 Mar19 1800/1800 yes 2 PSSJ1646+5514 WHT 1998 Sep23 3600/3600 yes 1 PSSJ1721+3256 WHT 1998 Sep24 1800/3600 yes 1
RX J1759+6638 WHT 99/98 Mar 19/Sep23 6300/6300 yes 10
PSSJ1802+5616 WHT 1999 Sep14 1800 no 2
BRJ2017 4019 CTIO 1998 O t14 3600 no 3
PSSJ2122 0014 WHT 1998 Sep22 3600/3600 no 1
BRJ2131 4429 CTIO 1998 O t16 1800 no 3
PMN J2134 0419 CTIO 1999 O t12 5400 yes 11
PSSJ2154+0335 WHT 1999 Sep14 1800 no 2
PSSJ2155+1358 CTIO 1999 O t10 3600 yes 2
BRJ2216 6714 CTIO 1999 O t09 3600 yes 3
PSSJ2241+1352 CTIO 1999 O t11 3600 yes 2
DMSB2247 0209 WHT 1998 Sep24 5400/3600 yes 12
PSSJ2315+0921 CTIO 1999 O t11 3600 yes 2
BRJ2317 4345 CTIO 1998 O t14 3600 yes 3
BRJ2328 4513 CTIO 1998 O t15 3600 yes 3
PSSJ2344+0342 CTIO 1999 O t11 3600 yes 2
BRJ2349 3712 CTIO 1999 O t09 3600 yes 3
a
For thequasars observed at the WHT the B/R designationsgive theexposure times
throughtheblue andred armsofthe ISISspe trograph.
Notes:
The quasarprexes indi ationthefollowingorigin: BR= APMsurveyobje ts sele ted
byB J
-R olorex ess; PSS=Se ond PalomarSky Survey; PMN = Parkes-MIT-NRAO
radio-sele ted obje ts; RX = X-ray sele ted; SDSS = Sloan Digital Sky Survey; and
DMS=Deep Multi olor Survey.
Referen es:
(1)Stern et al. 2000;
(2)G. Djorgovski's wwwpageat http://www.astro. alte h.edu/george/z4.qsos;
(3)Storrie-Lombardiet al.2000;
(4)Fanet al.1999;
(5)Warren,Hewett, &Osmer1991;
(6)Kenne ket al. 1995a;
(7)Storrie-Lombardiet al.1996a;
(8)Zi kgraf et al. 1997;
(9)Kenne ket al. 1995b;
(10) Henryet al. 1994;
2.2.2 CTIO Runs
Thirty-ve quasars were observed at CTIO during1998 O tober14 { 16, and
1999O tober9{12. Thetypi alexposuretimewas3600se ondsforthebrighterobje ts
(R18{19
th
mag) butsubstantiallylongertimes wereusedforthefainterSloanDigital
Sky Survey quasars. We used the R-C spe trograph with the 316 lines mm 1
grating,
entered at 6050
A and overing the range 3000
A 9100
A. This set-up resulted in a
dispersionof 1.98 A pixel
1
. Thedete tor usedwasa Loral30721024 CCDdete tor.
Thenarrowandwideslitobservationsweretakenwith1{1.5ar se and5ar se widths,
respe tively. Be ause of the substantial wavelength overage available with thisset-up
we used a WG345 blo kinglter (with 50 % transmission at 3450
A) to minimize the
se ond order ontamination from thestandard stars above 7000
A. The ontamination
isnegligibleforthe quasarsasmosthave no ux below4500
A butae tsthestandard
stars that have substantial ux at 3500
A. Appropriate measures, as dis ussed in the
data redu tion se tion, have been taken so that this set-up does not modify the ux
alibration at the red end of the spe tra. Using two instrumental set-ups in order to
ompletely remove the se ond order ontamination problem would have resulted in a
60{80 %in reaseintherequiredobserving time.
2.2.3 Ke k Observation
The observations of SDSS J0338+0022 were taken at the Ke k Observatory
withtheLowResolutionImagingSpe trometer(LRIS)on10February1999 witha1".0
wideslit and 400`/mmgrating blazedat 8500
A (spe tral resolution 8.1
A),and on 17
February 1999 with a 1".5 wide slit,with boththe 400`/mm grating blazedat 8500 A
(spe tral resolution 12.3
A), and withthe 300`/mm gratingblazed at 5000
A (spe tral
resolution17.3
A). Seeingwas0".7 0".9 FWHMon therstnight,and 0".7 1".0 on
the se ond night; observations were made under dark, photometri onditions in ea h
instan e. Asequen eof threeexposures,shiftedby10 00
alongtheslit,wastaken inea h
ongurationwiththepositionangleoftheslitsettotheparalla ti angleforthemiddle
exposure. This yielded net integrations of 3600 s (5542{9350 A; 8.1 A resolution; 1".0 slit),3600 s (5799{9609 A; 12.4
A resolution;1".5 slit), and 3000 s (3673{8708 A; 17.3
Aresolution;1".5slit). A GG495blo kinglterwasusedto suppressse ondorder blue
light. Moreobservational detailsaregiven inSongailaet al. (1999).
2.3 The Data
2.3.1 Data Redu tion
Thedataredu tionwasundertakenusingtheIRAF 2
softwarepa kage. Be ause
2
IRAFisdistributedbytheNationalOpti alAstronomyObservatories,whi hareoperatedbythe