X-ray observations of high redshift active galactic nuclei and galaxy clusters

(1)

X-Ray Observations of High Redshift

Active Galactic Nuclei and Galaxy

Clusters

by

Tim othy W askett

A thesis subm itted to the University of Wales

for the degree of Doctor of Philosophy

(2)

UMI Number: U 5 8 4 6 6 6

INFORMATION TO ALL U SE R S

The quality of this reproduction is d ep en d en t upon the quality of the copy subm itted. In the unlikely even t that the author did not sen d a com plete m anuscript and there are m issing p a g e s, th e se will be noted. Also, if material had to be rem oved,

a note will indicate the deletion.

Dissertation Publishing

UMI U 5 8 4 6 6 6

Published by ProQ uest LLC 2013. Copyright in the Dissertation held by the Author. Microform Edition © ProQ uest LLC.

ProQ uest LLC

789 East E isenhow er Parkway P.O. Box 1346

(3)

DECLARATION

This work has not previously been accepted in substance for any degree and is not being concurrently subm itted in candidature for any degree.

Signed ... D a t e ...

STATEMENT 1

The work presented in this thesis is all my own work carried out under the su pervision of Prof. Steve Eales with the exception of the following: I reduced all the XMM-Newton d a ta in this thesis myself but did not take the d a ta personally. The 3-h and 10-h field d a ta were made available to me by Prof. W alter Gear but all the other X M M d a ta in this thesis were obtained by me from the X M M pub lic archive. The Canada-France Deep Fields catalogue used to identify the X-ray sources was m ade available to me by Dr. Henry McCracken. Prof. Steve Eales performed the calculation to obtain the sub-mm flux of the X-ray sources in chap ter 3. The Chandra d a ta used in chapter 4 was reduced by Dr. Paul N andra and Elise Laird who subsequently provided me w ith the source list. The photom etric redshifts calculated using the CFD F photom etric redshift code were supplied by M ark Brodwin.

Signed ... D a t e ...

STATEMENT 2

I hereby give consent for my thesis, if accepted, to be available for photocopying and for inter-library loan, and for the title and sum m ary to be made available to outside organisations.

(4)

ACKNOWLEDGEMENTS

There are many people who gave a me great deal of help and support throughout my PhD and I would like to express my gratitude to them here. The most important person has always been my wife, Kat, without whom I would not have had the excuse to leave the department at a reasonable time each day. Aside from that, she always found the time to keep me fairly sane, even when she had enough of her own work to deal with. As always she is the best friend I have.

On the technical side one man has made all the difference. Dr. Dave Nutter is not only the most useful person I know, for fixing problems with software and teaching me new tricks, but is also a good friend. Without his help this thesis would have been much longer coming and would not have looked nearly so good. The guy surely knows everything there is to know. I cannot express how grateful I am to him.

The rest of the terminal room crowd, and other PhD students, have helped make my PhD time a lot of fun. They are, in no particular order: Kris Wojciechowski, Haley Morgan, Hannah Loebl, Neal Potter, Douglas Haig, Ian Bacchus, Melanie Bowden, Owen Davies, David Hubber, Sarah Roberts, Dean Trolley, Edward Gomez, Diego Garcia, Catherine Vlahakis, Iain Brown, Robbie Auld, Bruce Sibthorpe and probably a few more names that temporarily escape me. They are the best reason to come into the department every day, if only for the bizarre conversations we have at coffee time, Keyball and Yetisports.

My family has always given me their unconditional, if slightly bewildered, love and support for which I thank them. In particular my parents Malcolm and Janet, and my two sisters Lou and Sue. At last I can tell them that I have a proper job.

I’d also like to thank the computer support team who have been (mostly) very efficient at solving all those little problems that crop up all the time, but which threaten to stop all PhD progress indefinitely unless dealt with.

A special note of thanks goes to Dr. Mat Page for all his scientific help over the course of my PhD; he has saved me from many a moment of confusion.

Finally I’d like to express my gratitude to my supervisor, Prof. Steve Eales, for giving me the freedom to try out new things and to investigate subjects that interested me rather than forcing me down a particular road. If he had been more pushy this thesis would probably make a lot more sense but I would not have enjoyed the work nearly so much.

This thesis is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA.

(5)

ABSTRACT

X-ray surveys of three Canada-France Redshift Survey (CFRS) fields using

XMM-Newton are presented, with the aim of studying the Active Galactic Nuclei

(AGN) and galaxy cluster populations in these fields. The X-ray sources detected in these surveys resolve 51% of the X-ray background (XRB) in the 0.5 — 10 keV X-ray band.

The relation between the X-ray and sub-mm extra-galactic backgrounds is investigated using a combination of X-ray d a ta and sub-mm data. The X-ray properties of the sub-m m sources and visa versa indicate th a t the XRB is domi nated by accretion onto super-massive black holes, while the sub-mm background is dom inated by dust-obscured star formation.

X-ray sources are identified with optical objects using the Canada-France Deep Fields (CFDF) survey, which covers the m ajority of two fields. The redshift dis tribution of the AGN shows a clear peak at z ~ 0.7.

The 2-point angular correlation function, W(9), is calculated for the identified AGN but no significant clustering is detected. However, the results are consis ten t w ith X-ray selected AGN being good tracers of th e normal, inactive galaxy population.

The environments of m oderate luminosity AGN at z ~ 0.5 are investigated, using the clustering am plitude measure B gq and close pair counts. W hen compared to a control sample of equivalent inactive galaxies no difference is found between the respective environments. Minor mergers w ith low mass companions is therefore the most likely mechanism by which these AGN are fuelled.

A new m ethod for finding high redshift, optically selected, galaxy clusters is presented and is compared to X-ray selection. It is found th a t most optically selected clusters may have lower th a n expected X-ray luminosities suggesting th a t they are dynamically young compared to X-ray selected clusters.

(6)

C o n ten ts

1 A n In trod u ction to X -ray Sources 1

1.1 The Cosmic X-Ray Background - X R B ... 1

1.2 Active Galactic Nuclei - A G N ... 4

1.2.1 AGN X-ray Continuum E m i s s i o n ... 6

1.2.2 The Current Observational P i c t u r e ... 9

1.2.3 This T h e s i s ... 12

1.3 Galaxy C lu s te r s ... 13

1.3.1 The Evolution of Cluster X-ray P r o p e r t i e s ... 16

1.3.2 The Cooling Flow P ro b le m ... 18

1.3.3 The Relative Evolution of X -ray/O ptical P r o p e r tie s ... 20

1.3.4 This T h e s i s ... 21

2 X M M - N e w t o n D ata A cq u isition and R ed u ction 23 2.1 In tr o d u c tio n ... 23

2.2 XMM-Newton O v e r v ie w ... 24

2.3 X-ray D a t a ... 25

2.4 X-ray D ata R e d u c t i o n ... 26

(7)

jj C O N T E N T S 2.4.1 C reating Event F i l e s ... 26 2.4.2 Filtering the D a t a ... 27 2.5 Source D e t e c t i o n ... 31 2.5.1 Energy B a n d s ... 33 2.5.2 D etection S t a g e s ... 34 2.5.3 Energy Conversion F a c to r s ... 39 2.6 Notes on X-ray S p e c tr a ... 40 2.7 Basic R e s u l t s ... 42 2.7.1 3-h and 14-h F i e l d s ... 42 2.7.2 10-h F i e l d ... 46 3 47 3.1 I n tr o d u c tio n ... 47 3.2 SCUBA s o u r c e s ... 50

3.2.1 X-ray properties of the SCUBA s o u rc e s ... 51

3.2.2 S tatistical a n a l y s i s ... 56

3.3 sub-m m properties of X-ray s o u r c e s ... 61

3.4 D is c u s s io n ... 63

3.4.1 AGN Verses S ta r - f o r m a tio n ... 63

3.4.2 E xtra-G alactic Background R a d ia tio n ... 66

3.5 Concluding R e m a r k s ... 68

4 71 4.1 I n tr o d u c tio n ... 71

(8)

C O N T E N T S_____________________________________________________________ iii

4.2 O ptical I d e n tif ic a tio n s ... 72

4.2.1 XMM ... 72

4.2.2 The Chandra Training S e t ... 74

4.3 X-ray to O ptical Flux R a tio s ... 77

4.4 Photom etric R e d s h i f t s ... 81

4.5 Absolute M agnitudes - Galaxy T y p e s ... 83

4.7 C a t a l o g u e ... 89

4.8 A p p e n d i x ... 113

4.8.1 BPZ Photom etric Redshift Estim ation C o d e ...113

4.8.2 C FD F Photom etric Redshift Estim ation C o d e ... 117

5 T w o P oin t A ngular C orrelation F unction o f A G N 121 5.1 I n tro d u c tio n ...121

5.2 Calculating W { 9 ) ...123

5.3 Generating a Random P o p u la tio n ...125

5.3.1 Sensitivity M a p ...126

5.3.2 X-ray P o p u la tio n ... 127

5.4 R e s u lts ...129

5.5 D is c u ss io n ... 131

6 Searching for G alaxy C lusters 137 6.1 I n tro d u c tio n ...138

(9)

iv C O N T E N T S

6.2.1 Extended Sources D etected by E M L d e te c t ... 141

6.2.2 Gaussian Sm oothing T e c h n i q u e ...142

6.2.3 M ultiresolution Wavelet Filtering T e c h n iq u e ...144

6.2.4 Results for the Lockman H o l e ... 145

6.2.5 Results for CFRS fie ld s ...152

6.2.6 Future W o rk ...155

6.3 O ptical Cluster D etection ... 156

6.3.1 My A lg o rith m ... 157 6.3.2 Points to N o te ... 161 6.3.3 R e s u lts ... 162 6.3.4 M easuring th e Cluster R ic h n e s s ... 168 6.3.5 Correcting for In c o m p le te n e s s ... 170 6.4 D is c u s s io n ... 172

6.5 Chandra Deep Field - S o u t h ... 174

6.5.1 An A lternative S t a t i s t i c ... 176

6.5.2 Extended X-ray Sources D etected by C h a n d r a ... 180

6.5.3 XM M-Newton D a t a ... 185

6.5.4 Im plications for X-ray Cluster S e a rc h e s ... 191

6.5.5 F uture W o rk ...193

7 T h e E n vironm ents o f A G N 195 7.1 In tr o d u c tio n ... 195

(10)

C O N T E N T S v 7.1.2 Previous Work ... 199 7.1.3 Unbiased Tracers of A G N ...202 7.2 Selection of AGN S a m p le ... 204 7.3 Calculation of B gq ...207 7.3.1 Control S a m p l e ...208 7.3.2 Correction for In c o m p le te n e ss ... 208 7.4 R e s u lts ...209 7.4.1 Kolmogorov-Smirnov T e s ts ... 211 7.5 Close Companions ...212 7.6 D is c u ss io n ... 214

7.6.1 Implications for AGN Fuelling M ech an ism s...214

7.6.2 The Richest Environments in the 14-h F i e l d ... 218

7.7 Future W o rk ...221

8 Sum m ary and C onclusions 223 8.1 Thesis Sum m ary ...223

(11)

(12)

List o f F ig u res

1.1 A plot of th e extra-galactic background radiation field... 3

1.2 A schematic representation of a (radio-loud) AGN... 5

1.3 AGN continuum emission... 8

1.4 The effect of varying column densities of intervening HI on th e X-ray continuum of an AGN... 10

2.1 Unfiltered PN im age... 28

2.2 A partially filtered PN image... 29

2.3 An example of a rate curve... 31

2.4 Fully filtered PN image... 32

2.5 Exposure m ap for the 3-h PN exposure in the soft b an d ... 35

2.6 Detector mask for the 3-h PN exposure... 36

2.7 Background m ap for the 3-h PN soft band exposure... 38

2.8 False colour X-ray image of the 3-h field... 43

2.9 As figure 2.8 but for the 14-h field... 44

2.10 Differential source counts for the combined 3-h and 14-h field X-ray sources... 45

2.11 As figure 2.8 but for the 10-h field... 46

(13)

L IS T OF F IG U R E S

3.1 Smoothed, soft X-ray image of the 3-h CUDSS region... 51

3.2 Smoothed, soft X-ray image of th e 14-h CUDSS region... 52

3.3 Histograms showing the distribution of m ean X -ray counts associ ated w ith each artificial SCUBA source... 58

3.4 As for figure 3.3 b ut for the 14-h field... 59

3.5 X-ray to sub-mm flux ratios for the SCUBA sources and th e X-ray sources... 64

4.1 Soft X-ray to optical flux ratio for all sources detected in b o th the 3 and 14-h fields... 78

4.2 As for figure 4.1 b ut showing the h ard X -ray to optical flux ratio. . 79

4.3 As for figure 4.1 b ut showing the 0.5-7 keV X-ray to R band optical flux ratio as used in McHardy et al. (2003). Unidentified sources are now plotted at R = 27... 80

4.4 Redshift distribution of the identified X-ray sources... 83

4.5 Absolute Iab m agnitude vs. redshift for the identified sources. . . . 85

4.6 M edian values for the galaxy types in figure 4 . 5 ... 86

4.7 Absolute Ia b m agnitude vs. X-ray lum inosity... 87

4.8 Absolute Ia b m agnitude vs. to ta l X -ray flux... 88

4.9 Photom etric vs. spectroscopic redshifts...114

4.10 C FD F vs. BPZ photom etric redshift estim ates... 117

4.11 Photom etric vs. spectroscopic redshifts for X-ray sources...120

5.1 Sensitivity m aps for the 3-h exposure...127

5.2 An illustration of how the random ly generated source population is distributed w ithin the X M M FoV... 128

(14)

L IS T OF FIG U R E S______________________________________________________ ix

5.4 W{9) for th e to tal bright sample...133

6.1 Comparison of X-ray images, raw, smoothed and filtered... 145

6.2 Source detection on smoothed or filtered X-ray im ages... 147

6.3 Histogram showing the distribution of the CLASS-STAR param eters for the sources detected in the M R/1 filtered Lockman hole X M M im age...148

6.4 Close-up of the interacting clusters in the Lockman hole... 149

6.5 Close-up of the interacting clusters in the Lockman hole using only MOS d a ta ... 150

6.6 Figure 1 from Hashimoto et al. (2002) showing the ROSAT contours overlayed on a V R I colour image...152

6.7 Filtered X-ray image of the 14-h field... 153

6.10 Example of the effect changing zero-points on the photom etric red shift estim ation... 159

6.11 Slice through the 14-h d a ta cube at z ~ 0.88... 163

6.12 A 3D view through the 14-h d a ta cube... 165

6.13 An alternative 3D view through the 14-h d a ta cube... 165

6.14 I band image of the 14-h field over-density at z — 0.88... 167

6.15 Num ber counts of the full and reduced CFD F c a ta lo g u e ... 171

6.16 R band image of th e COMBO-17 field, coincident w ith the Chan dra Deep Field-South, overlayed with a colour representation of the galaxy distribution... 175

6.17 Comparison of the two statistics used to quantify structure in the data-cube technique... 178

(15)

X L IS T OF F IG U R E S

6.18 S tatistic B iso-surfaces projected onto th e COM BO -17 R b an d im age of the C D F-S... 182 6.19 Zoomed in portion of figure 6.18... 185 6.20 X M M soft band image of the C D F-S...186 6.21 Figure 13 from Osmond & Ponm an (2004) showing th e L x — Tx

relation for galaxy groups and clusters...188 6.22 R band CDF-S image centred on cluster 1...189 6.23 As figure 6.22 b ut centred on cluster 4 ...190

7.1 Luminosity vs. redshift for the 31 AGN in th e environm ent study

s a m p l e ...205 7.2 Hard X-ray luminosity function, from U eda et al. (2003)... 206

7.3 Clustering am plitude B gq of galaxies around th e 14-h AGN environ

m ent sam ple 211

7.4 Histogram s of the number of com panion galaxies... 215 7.5 Slice through the 14-h d a ta cube a t z — 0.48... 219 7.6 M R /1 filtered 14-h field X-ray image showing th e two Abell 1 regions.220

(16)

List o f T ables

2.1 Sum m ary table for the 3 X M M fields... 40

4.1 Sum m ary of ID statistics... 74

4.2 X-ray properties of the 3-h field X M M sources... 91

4.2 ( C o n tin u e d ) ... 92

4.2 ( C o n tin u e d ) ... 93

4.2 ( C o n tin u e d ) ... 94

4.3 X-ray properties of the 14-h field X M M sources... 95

4.3 ( C o n tin u e d ) ... 96

4.3 (C o n tin u e d ) ... 97

4.3 (C o n tin u e d )... 98

4.4 ID properties of the 3-h field X M M source ID s... 99

4.4 ( C o n tin u e d ) ...100

4.4 ( C o n tin u e d ) ...101

4.5 As table 4.4 b ut for the 14-h field... 102

4.5 ( C o n tin u e d ) ...103

4.5 ( C o n tin u e d ) ...104

(17)

xii L IS T OF T A B L E S

4.5 ( C o n tin u e d ) ... 105

4.6 O ptical properties of the 3-h field X M M source ID s... 106

4.6 ( C o n tin u e d ) ... 107

4.6 ( C o n tin u e d ) ... 108

4.7 As table 4.6 b u t for the 14-h field...109

4.7 ( C o n tin u e d ) ... 110

4.7 ( C o n tin u e d ) ... I l l 4.7 ( C o n tin u e d ) ... 112

5.1 EC F values for converting to full b and flux... 126

6.1 Relevant quantities for figure 6.17... 179

6.2 Properties of th e over-densities shown in figure 6.20... 191

7.1 Clustering am plitude results for the AGN sam ple... 210

7.2 Clustering am plitude averages... 212

7.3 K-S tests on the clustering results... 212

(18)

C h a p ter 1

A n In tro d u c tio n to X -ray S ou rces

Astronomy is a vast and interesting science. W hat other subject can boast the entire Universe as its playground? This thesis represents b u t a tiny p art of the huge endeavour to understand the Universe, b u t I hope th a t the reader will find it interesting nonetheless.

1.1 T he C osm ic X -R a y B ackground - X R B

X-ray astronomy is the observational counterpart to high energy astrophysics. In astronom ical term s these fields are relatively recent additions to the science be cause of th e opacity of the E arths atm osphere to X-rays, which has, until the age of space instrum entation, m ade it impossible to study such things. X-rays are energetic photons th a t are produced by high tem perature phenomena. They range in energy from roughly 0.12 — 50 keV, or in wavelength from roughly 0.25 — 100

A

(although the exact definition is somewhat debatable) and fill the gap between

(19)

2 C H A P T E R 1. A N IN T R O D U C T IO N T O X - R A Y SO U R C E S

extrem e ultraviolet at lower energies and gam m a rays a t even higher energies.

Like light from all regions of th e electrom agnetic spectrum X-rays originate b o th from w ithin our own Milky Way G alaxy and also from beyond - th e so-called extra-galactic background radiation field. In a sense, extra-galactic astronom y is th e study of this background radiation field and, w ith th e exception of th e CMB, to fully understand the origin of a particular background one m ust first resolve it into its constituent sources.

Figure 1.1 shows a recent com pilation of d a ta for th e extra-galactic back ground. Each peak is dom inated by different physical processes: th e Cosmic Mi crowave Background (CMB) has th e highest energy density and is m ade up of redshifted photons from the surface of last scattering, th e echo of the big bang as it were; the Far-Infrared Background (FIB, or sometim es Cosmic IR Background, CIRB) is the result of dust obscuration, which reprocesses higher energy photons from e.g. stars into the far-IR and sub-m m regime; th e optical and UV back grounds are essentially starlight while the origin of the G am m a-R ay Background

(GRB) is still something of a mystery b u t probably includes highly energetic forms of super-novae, am ongst other possibilities.

The X-Ray Background (XRB) contains significantly less energy density th an either th e FIB or th e optical background b u t it is no less im p o rtan t to understand, if we are to fully appreciate th e complexities of the Universe. Very soon after it was discovered it was found th a t the XRB can be resolved into individual sources; from th en on extra-galactic X-ray astronom y and high energy astrophysics took off, w ith a prim ary aim to understand the n atu re of these sources and to explain the origin of the XRB.

(20)

1.1. TH E CO SM IC X - R A Y BA C K G RO U N D - X R B 3

log(X)

(fji

m )

o

-4

6

Sh m CJ

100

1 CMB

/ \

_i XRB FIB GR B ■2 8

100

10

1

0.1 io- 2

io- 2

12

16

20 log

v

(Hz)

24

Figure 1.1: A plot of the extra-galactic background radiation field, taken from Bragaglia et al. (2000). CMB = Cosmic Microwave Background, FIB = Far-Infrared Background, XRB = X-Ray Background, GRB = Gamma-Ray Background. The optical and ultra violet backgrounds are still somewhat uncertain in their exact form but they lie between the FIB and the XRB.

The spectrum of the XRB holds many clues as to its origin. It has a rath er simple shape, significantly less complicated th an the optical background, which still does not have a reliable m easurem ent despite the relative m atu rity of optical astronomy. Broadly speaking th e spectrum of the XRB is a bum p w ith a peak in energy density at ~ 30 keV (the XRB ‘bum p’ in figure 1.1). In the energy range 3 — 20 keV, the XRB can be well approxim ated by the following function:

(21)

where a is the spectral index, which determ ines th e slope of th e power law model. A lthough the norm alisation is still uncertain to w ithin ~ 10% th e spectral index has been well determ ined to be a ~ 0.4 (C om astri et al., 1995). Therefore, over this fairly lim ited energy range, th e XRB increases slowly in energy density, since

v l u oc E 1~a = E 0,6 (the left hand p art of the XRB ‘b u m p ’ in figure 1.1). At slightly

lower energies (0.5 — 2 keV) the XRB is well fit w ith an o th er power law of spectral index a ~ 1.0. The current generation of X -ray telescopes detect X-rays in the energy range ~ 0.5 — 10 keV, and over this range th e XRB is often approxim ated to a power law w ith a = 0.4.

The rem ainder of this introduction gives some background to the sources th a t make up the XRB; the sources th a t form the subject of this thesis.

1.2 A c tiv e G alactic N u clei - A G N

T he vast m ajority of the sources th a t make up th e XRB are point-like and it was found th a t alm ost all were associated w ith the central regions of galaxies.

T he term Active Galactic Nucleus (AGN) is now used to describe w hat is, in effect, an accreting super-massive black hole (SBH) residing in th e centre of a galaxy. A theoretical picture has gradually emerged from th e various observa tions of AGN so th a t we now have a reasonable understanding of the mechanisms involved (e.g. Antonucci, 1993). Broadly speaking an AGN consists of the SBH itself which is fed by an optically thick accretion disc of m aterial. Beyond this, and in the same plane, is a torus of obscuring gas and dust th a t causes the AGN to assume a different appearance depending on the orientation of th e observer’s view into the central engine. Along the axis of ro tatio n jets of high energy plasm a

(22)

1.2. A C T IV E G A L A C T IC NU CLEI - A G N 5

Figure 1.2: An idealised schematic representation (not to scale) of a radio-loud AGN from Urry & Padovani (1995), representing the current unification scheme that is thought to apply to essentially all AGN. The accretion disc and hot corona occupy a tiny region in the centre of this diagram, surrounded by the much larger obscuring torus. High velocity clouds orbit very close in to the central engine and so the broad emission lines produced by them are not visible when the observers’ line of sight runs through the obscuring torus. The low velocity clouds that are responsible for the narrow line emission are visible even at low viewing angles. The radio jets axe ejected along the rotation axis and they can travel well outside the confines of the host galaxy, where they impact with the inter-galactic medium causing radio-lobes. Radio-quiet AGN do not have such powerful radio jets and so have much less radio emission. ~ 3/4 of all AGN are radio-quiet.

(23)

are sometimes formed, which carry away th e m ajo rity of th e angular m om entum lost from the in-falling accretion disc m aterial; they can be powerful radio sources. Various other phenom ena, such as broad and narrow emission line regions, arise when, respectively, high and low velocity gas clouds orbiting out of th e plane of th e accretion disc are illum inated by the rad iatio n from th e central engine.

Figure 1.2 shows a schematic representation of this ‘unification scheme’ to dem onstrate how this picture fits together geometrically. A lthough this figure shows a radio-loud AGN there is essentially no difference in th e central engines of radio-loud and radio-quiet AGN; the radio emission seems to be an addition rather th a n a fundam ental difference (Antonucci, 1993).

1.2.1 A G N X -ray C on tin u u m E m issio n

Differential ro tatio n in the accretion disc arises because th e inner p arts orbit the black hole more quickly th a n th e outer p a rts (from io2 = G M / r 3: where uj is the angular velocity of an element of accretion disc at a radius r from a black hole of mass M ). The resultant friction between neighbouring layers causes th e m aterial in th e disc (m ostly gas and some dust) to gradually lose angular m om entum and so spiral in tow ards the central black hole w ith the resu ltan t loss of potential energy converted into heat. This not only results in th e u ltim ate consum ption of the m aterial in the disc by the black hole, adding to the black hole’s mass, but also produces a characteristic accretion disc therm al spectrum . T he tem p eratu re across the face of the accretion disc is a function of its radius from th e black hole, the closer the hotter, and so th e overall spectrum is essentially the superposition of a range of black body curves at different tem peratures across th e whole surface of the disc. This prim ary emission from the accretion disc is seen as an optical/U V

(24)

1.2. A C T IV E G A L A C T IC N U C LE I - A G N 7

continuum.

Further to this prim ary emission, observations in the X-ray regime reveal other components th a t contribute to a typical AGN continuum. Various models have been used to explain this X-ray emission b u t the most successful can be summarised as follows (H aardt, Maraschi, & Ghisellini, 1997): Above (and below, depending on your orientation) th e accretion disc is a hot corona th a t is heated by some mechanism, possibly m agnetic dissipation processes, to very high tem peratures

( kT = 30 — 300 keV). The free electrons in this hot corona reprocesses the prim ary

accretion disc radiation into an X-ray continuum via inverse Com pton scattering. P art of this X-ray radiation is modified when it is reflected back off the accretion disc, via Com pton scattering, to form a ‘reflected com ponent’; the ultim ate X-ray emission is then a combination of the prim ary Comptonised emission from the hot corona plus the reflected component from the accretion disc (e.g. H aardt et al., 1997). The X-ray source may be in the form of an optically th in parallel layer over the whole accretion disc, or be composed of a number of smaller ‘hot sp o ts’. Figure 1.3 shows the continuum features typical of AGN in general, the right hand side of the plot being the most relevant to this discussion.

There is likely to be a certain am ount of feedback between th e hard X-ray em itting corona and the softer therm alised disc w ith an exchange of photons be tween the two. Not all the hard X-rays from the corona incident on the disc are reflected, and in fact the m ajority (89 — 90%) is therm alised in the disc and re em itted where it is then Comptonised again by the hot corona, and so on (H aardt et al., 1997). The reprocessed thermalised radiation is therefore thought to be responsible for the UV bum p and the reprocessed reflected radiation is responsible for the Com pton bum p at ~ 30 keV (see figure 1.3). At higher energies ( ~ 200 keV) an exponential cut-off causes the continuum emission to drop off rapidly. This

(25)

8 C H A P T E R 1. A N IN T R O D U C T IO N TO X - R A Y SO U RCES

100 urn 10 urn l i r a 1000 A 100 A 1.24 keV 12.4 keV 124 keV

1.0 AGN continuum Starburst component Dusty torus Accretion disc Hot corona Reflection from disc 0.5 0.0 -1.0 -1.5 12 14 16 18 20 log v (Hz)

Figure 1.3: AGN continuum emission, taken from Manners (2002) and based on the mean of a sample of mainly UV excess quasars from Elvis et al. (1994).

is due to energy distribution of the Comptonising electrons: if the distribution is therm al the electrons follow a Maxwell-Boltzmann energy distribution and so the number of electrons of a given energy follows:

f ( e ) = A

where T is the tem perature of the electrons. Above an energy of e ~ k T the number of electrons capable of up-scattering continuum photons drops off expo nentially causing the X-ray continuum to do the same. Therefore, the energy of the exponential cut-off indicates the tem perature of the electrons in the hot corona.

(26)

contin-1.2. A C T IV E G A L A C T IC N U C LE I - A G N 9

uum th at, to first order, can be approxim ated by a power law w ith spectral index

a ~ 0.8 —1.0 (see th e long-dashed line in figure 1.3): a consequence of the feedback

between the disc and corona. This is significantly softer th a n th e XRB spectrum so if the XRB is composed prim arily of AGN then there is a clear discrepancy between the two. However, as X-ray surveys probe deeper into the XRB and resolve more and more of it into point sources it has become clear th a t there is more to AGN X- ray emission th a n simply th e accretion disc and hot corona. Obscuring m aterial in the torus (typically neutral hydrogen gas, HI) between the AGN and an observer acts to suppress th e intrinsic X-ray emission via photo-electric absorption. The low energy (soft) X-rays are absorbed more heavily th a n the high energy (hard) X-rays however, so what starts out as a rath er soft intrinsic spectrum becomes a much harder observed spectrum. Therefore, many of the AGN th a t make up the bulk of the XRB actually have a significant am ount of intrinsic absorption, which results in the XRB having a harder spectrum th an th a t of a typical unabsorbed AGN (see figure 1.4 for a dem onstration of this effect).

1.2.2 T h e C urrent O b servation al P ictu re

Deep exposures w ith the most recent and powerful X-ray observatories, XMM-

Newton and Chandra (e.g Barger et al., 2003; Giacconi et ah, 2002; Mainieri et ah,

2002; M cHardy et ah, 2003; Page et ah, 2003), have built on the deepest R O S A T X-ray surveys (e.g M cHardy et ah, 1998; Hasinger et ah, 1998) by going deeper and to higher X-ray energies with b e tter positional accuracy. This has opened up the study of faint X-ray sources, such as high redshift AGN, and has also revealed X-ray emission from otherwise norm al galaxies at more modest redshifts (Hornschemeier et ah, 2003). These surveys have now resolved the m ajority of th e XRB in the soft

(27)

10 C H A P T E R 1. A N IN T R O D U C T IO N T O X - R A Y SO U R C E S 1 . 0 0) > 0.1 10.0 0.1 1.0 Energy (keV)

Figure 1.4: The effect of varying column densities of intervening HI on the X-ray con tinuum of an AGN (taken from Manners (2002)). The higher the column density of HI the higher the energy of X-rays th at are ‘wiped o ut’ by the photo-electric absorption.

(0.5 — 2 keV) X -ray b and w ith a small fraction left unaccounted for in the hard (2 — 10 keV) b and (M oretti et al., 2003).

The n atu re of th e XRB at these X -ray energies is now well on th e way to being understood b u t there is still a small discrepancy in th e h ard X -ray band. The peak of th e XRB lies a t ~ 30 keV and th e sources th a t dom inate th e 0.5 — 10 keV band are not sufficiently hard to be th e dom inant sources contributing to the XRB peak. This indicates th a t a population of very faint sources, w ith very hard spectra, make up th e rem aining fraction of th e XRB in th e h ard band, and would also be expected to contribute a much greater fraction to the X R B nearer its peak (M oretti et al., 2003). Such h ard sources are m ost likely a result of extrem ely high obscuration.

(28)

1.2. A C T IV E G A L A C T IC N U C LEI - A G N 11

The radiation absorbed by the obscuring m aterial must be re-em itted at longer wavelengths and the possibility of the Far-IR /Sub-m m background being somehow connected w ith the XRB is discussed in many papers (e.g. Almaini, Lawrence, & Boyle, 1999). However, current X -ray/Sub-m m surveys suggest th a t the two backgrounds are only loosely related (e.g. W askett et al., 2003; Alexander et al., 2003; Severgnini et al., 2000). Future instrum entation w ith higher energy limits are likely required to fully explain the XRB and the nature of the sources th a t dom inate its peak.

At present though, the emphasis must be turned to those sources th a t we can observe easily w ith th e current instrum entation. QSOs and type-I AGN dom inate the softest X-ray energies with an increasing contribution from more obscured type- II AGN becoming im portant at higher energies (e.g. Gilli, Salvati, & Hasinger, 2001). Identifying the optical counterparts to these sources is crucial for a full understanding of their properties and a great deal of effort has been expended in obtaining this inform ation (e.g. Barger et al., 2003; McHardy et al., 2003).

For example, one of the most useful quantities th a t can be derived for a popu lation of sources is the luminosity function. This reveals much about the nature of

0

a population and determ ining its evolution w ith redshift can shed light on how the population as a whole changes over time. The X-ray luminosity function (XLF) has begun to be investigated in depth by several groups (Cowie et al., 2003; Steffen et al., 2003; Ueda et al., 2003). Both Ueda et al. (2003) and Steffen et al. (2003) find th a t the evolution of the XLF is a function of luminosity. The population of X-ray sources with L x (2 — 10 k e V ) > 3 x 1043 erg s-1 is dom inated by type-I AGN, and th e number-density of these sources increases w ith redshift out to z ~ 2 — 3. At lower X-ray luminosities however, the fraction of type-II AGN increases rapidly with decreasing X-ray luminosity. The number-density of these sources appears to

(29)

peak a t z < 1.

A lthough Chandra is b e tte r suited for identifying X-ray sources w ith optical counterparts { X M M has a resolution of ~ 6" full w idth half m axim um (FW HM) cf. ~ 0.5" for Chandra), X M M has greater sensitivity and a larger field of view (FoV), m aking it b e tte r for large area surveys.

1.2.3 T h is T h e sis

The m ajority of th is thesis concerns the point sources detected in a medium-deep

X M M survey com posed of two separate exposures ( ~ 0.4 square degrees). Most of

these point sources are AGN. I study th e sub-m m properties of a selection of X-ray sources (chapter 3), an d visa versa, and th en quantify th e ability of such a survey to identify X -ray sources w ith optical co unterparts (chapter 4). I also estim ate redshifts for th e identified sources using photom etric redshift codes. These allow a quick, and reasonably reliable, way of obtaining redshifts for objects w ith m ulti b and photom etry. A lthough not as accurate as spectroscopy these techniques are becoming widely used as a short-cut for large surveys, where statistical properties are fairly insensitive to the accuracy of individual redshift m easurem ents (Csabai et al., 2003; F ontana et al., 2000; Kashikawa et al., 2003; M obasher et al., 2004). These m ethods can also be used on objects fainter th a n th e spectroscopic limit, where m any X -ray source counterparts reside (Alexander et al., 2001). I test two photom etric redshift estim ation codes on th e X-ray source IDs and obtain a robust redshift distribution for those sources th a t could be identified reliably, while placing lim its on th e properties of those th a t could not.

(30)

1.3. G A L A X Y C L U ST E R S 13

X M M surveys w ith the aim of understanding how AGN are distributed relative

to the normal galaxy population. In chapter 7 I investigate the environments of m oderate luminosity AGN, near the peak of th e AGN redshift distribution, to see if the mechanisms by which they are fuelled are influenced by the environments of the AGN host galaxies.

The accretion history of the Universe is dom inated by AGN w ith m oderate accretion rates a t z < 1. To fully understand th e processes th a t lead to the produc tion of the XRB we must understand th e mechanisms th a t lead to this accretion. Many of the AGN in this thesis are members of th a t im portant population and so hold fundam ental clues th a t could help explain the XRB phenomenon. If we can understand the origins and causes of accretion onto SBHs then we will be a step closer to knowing how this fits in with th e rest of the Universe.

1.3 G alaxy C lu sters

Chapter 6 is a slight deviation from the rest of the thesis as it does not deal w ith AGN but with th e other population of X-ray sources th a t contributes to the XRB, albeit only a small fraction ( ~ 5% in the soft band) - galaxy clusters.

Clusters are the end point of large scale structure formation. In the hierarchical picture of structure form ation small things form first from fluctuations in the dark m atter density distribution. These small over-densities, or ‘dark m atter halos’ merge w ith each other to form larger conglomerations of dark m atter and so on. The process continues until we are left in the present day Universe w ith a very clumpy distribution of m atter on all different mass scales. Baryonic m atter, in the form of stars and galaxies etc., is caught up in the overwhelming gravity of the

(31)

dom inant dark m a tte r so th a t the distribution of m ass is effectively traced by the light em itted from this ‘ordinary’ m atter.

Galaxy clusters are effectively the end point of a whole m erger tree history and are th e largest self gravitating systems in th e Universe. W ith masses up to > 1015 Mq th e dark m atter halos can encompass thousands of individual galaxies w ithin th e potential well. Along w ith the optically visible galaxies clusters are also perm eated w ith a very tenuous b ut also extrem ely hot ionised gas th a t sits in hydrostatic equilibrium w ith the dark m a tte r distribution. T he m ass of th e gas is typically similar to th a t associated w ith the galaxies and together they make up only a small fraction ( ~ 1/10 — 1/5) of th e to ta l mass of a cluster.

The mass distribution of dark m atter halos a t any given point in tim e can be described the Press-Schechter formalism (Press k Schechter, 1974), which ef fectively gives a prescription for how th e dum piness of th e Universe evolves. The only assum ption needed for this formalism is th a t an expanding cosmology is per m eated w ith a self gravitating ‘gas’ th a t experiences no forces other th a n its own gravity (basically dark m atter). Prom an initially ‘grainy’ m a tte r d istribution this formalism predicts how the smaller m atter condensations should merge together to form the larger ones, and this eventually results in a self-similar behaviour whereby th e distribution of m atter has forgotten the form of th e initial p ertu rb atio n s th a t caused it to sta rt collapsing in th e first place. It is very successful a t reproducing the observed present day dum piness of the mass distrib u tio n from galaxy to cluster and super-cluster scales.

However, when observing galaxy clusters th e mass is not an im m ediately ob servable quantity so we need something th a t we can observe th a t can th en be used to estim ate the mass. Fortunately there are several ways in which we can estim ate

(32)

1.3. G A L A X Y C L U ST E R S 15

the mass of a cluster:

Velocity dispersion of cluster galaxies. If the galaxies th a t make up a cluster are in dynam ical equilibrium with the underlying potential well then the to tal mass of the system can be estim ated from their line of sight velocity dispersion, a. For an isothermal distribution, M ci oc a 2.

X-ray properties. Rich clusters of galaxies produce extended X-ray emis sion from th e hot intra-cluster gas trap p ed in the potential well, via therm al brem sstrahlung radiation. Assuming this gas is in hydrostatic equilibrium w ith the cluster potential the X-ray properties are well behaved. The emis- sivity of the ionised gas scales as e oc p2asT 1/2, so the to tal luminosity of a cluster becomes L x oc p2gasR zclT 1/ 2 oc M gaspgasT 1/ 2. The luminosity and

tem perature can be estim ated from the X-ray emission and so the density can be estim ated using the size of the cluster. Since X-ray luminosity is most sensitive to the gas density the m ajority of the emission comes from the dense core of the cluster. The distribution of gas can then be used to calculate the to tal mass of the cluster, w ithin a given radius, by assuming th a t the gas is in hydrostatic equilibrium w ith the dark m atter:

p m pG \ d m r a m r J

A

where T is the tem perature and p m p is the mean particle mass of the gas. • G ravitational lensing. Much more difficult to achieve, this technique relies on

detecting background galaxies th a t have been gravitationally lensed by the mass of the cluster. The am ount of distortion experienced by the background galaxy depends on the surface mass over-density of th e cluster.

(33)

All three m ethods give results th a t are consistent w ith each other, giving added certainty to the mass determ inations.

1.3.1 T h e E v o lu tio n o f C lu ster X -ra y P r o p e r tie s

If we accept th a t th e hierarchical picture of stru ctu re form ation is th e way in which clusters form, th en they must still be forming today. As each sequential level of th e hierarchy collapses th en previous sub-structure is erased, leaving a central core which is essentially in quasi-equilibrium, while further out th e cluster continues to grow through accum ulation of in-falling m atter. If a sphere were placed around th e core, separating these two distinct regions, th en th e density w ithin th e sphere is always ab o u t 200 tim es the background density, w hatever th e epoch. W ithin th is framework clusters should, therefore, be quite easy to un d erstan d as, at their m ost basic level, they are only governed by a couple of processes. Kaiser (1986) used th e assum ption th a t the therm odynam ics of th e intra-cluster m edium (ICM) is only determ ined by gravitational processes and th a t th e emission from the hot gas is pure brem sstrahlung, to predict how the X -ray properties of clusters should evolve as we look to higher redshifts. The prediction of th e resu ltan t model is th a t clusters are self-similar, so th a t the properties of one cluster can simply be appropriately scaled to predict the properties of a cluster of different mass. This is the ‘self-similar’ model.

According to this model the characteristic cluster quantities are predicted to scale as follows: p* oc (1 + z)3, M* oc (1 + z )~3, R* oc (1 + z ) ~ 2, T* oc (1 + z ) ~ \

L*x oc (1 + z ) ~ 1/ 2 (assuming th a t the spectrum of density fluctuation in th e early

Universe is a power law i.e. scale free-initial conditions, and has a spectral index

(34)

1.3. G A L A X Y C L U ST E R S 17

cooler and slightly less luminous. This means th a t the low redshift cluster scaling relations, predicted from assuming hydrostatic equilibrium and the virial theorem, should evolve w ith redshift also.

The key relations are then: L x oc M p T 1/ 2, L x oc Tx ( 1 + z )3^2 and L x oc M 4/3( 1 + z ) 7/ 2. We also have the relations between the velocity dispersion of the cluster galaxies and the properties of the hot gas e.g. T oc cr2.

However, observations have revealed th a t real galaxy clusters do not follow these idealised scaling relations, so the self-similar model m ust be wrong to some degree. In reality, rath er th an following the relation L x oc Tx clusters scale more like L x oc Tx .

The solution to this discrepancy lies in the physics missing from the self-similar model. Galaxy clusters are not simply subject to gravitational processes, such as adiabatic compression during collapse and shock heating, there is a great deal more going on. Kaiser (1991) introduced pre-heating into the self-similar model, which increases th e entropy of th e ICM (defined as S = T / n 2/3 where n is the fully ionised gas density) and prevent it from becoming so dense. This decreases the X-ray luminosity, especially for poorer clusters, where the extra added entropy is com parable to the self-similar entropy, and results in the much steeper L x — Tx relation.

The exact source of this extra heating during the form ation of a cluster is not known b u t there are a num ber of candidates. Supernovae explosions could provide some heat b u t it is not enough to account for the observed relation. Heating by a central AGN is a more likely mechanism, as much more energy is available

(35)

A significant consequence of this pre-heating is th a t high redshift clusters do not produce nearly as much X-ray emission as predicted by Kaiser (1986). In the original self-similar theory the X-ray luminosity function of clusters should evolve in a positive way tow ards high redshift and the comoving X -ray emissivity from clusters is predicted to increase, e oc (1 + z)5/ 2 (for n = —1), so th a t high redshift clusters should be very numerous and easy to find. However, this is not th e case and in fact negative evolution has been observed m eaning th a t high redshift clusters are much harder to find th a n initially expected.

1.3.2 T h e C o o lin g F low P ro b lem

Continual input of energy by a central AGN may also be th e best explanation for th e so-called ‘cooling flow problem ’. P u t simply, th is aspect of cluster evolution is a question of energetics. T he hot gas in a cluster em its X-rays and so should cool down as th e energy is carried away. The cooling rate is sensitive to the density of th e gas because, as described above, th e emission increases as p2, so th e cooling ra te is fastest in th e centre of th e cluster. Now, if the gas in the centre of a cluster is cooling quickly i.e. th e cooling tim e is significantly less th a n th e Hubble time, th en the hydrostatic support in th e centre of a cluster decreases as the tem perature drops. This results in an inflow of gas leading to an increase in the central gas density. Because th e gas is now denser it will rad iate more efficiently, see above, and so will cool more quickly. t cooi oc T 2 at constant pressure so th e cooler the gas gets th e quicker it cools.

This runaway cooling effect should m ean th a t all th e gas in th e cores of rich clusters (where th e cooling tim e is less th a n th e Hubble tim e) should continue to cool until it has accreted into a central region, usually a giant elliptical cD galaxy.

(36)

1.3. G A L A X Y C L U ST E R S 19

The inflow rate is given by M x — (2L cooi f i m ) / ( 5 k T ) (Fabian, 1994) and can be as high as 500 M 0 y r-1 in some clusters. W hy is it then th a t we do not see this extraordinary accretion rate in every cluster, and why does the gas in the cores of clusters appear to stop cooling once it reaches about 1/3 of the virial tem perature? Some clusters do have cooling flows and are observed to have large in-flow rates b ut other clusters should be experiencing much more powerful cooling flows th an they are observed to have. This is the cooling flow problem and is an outstanding problem in cluster evolution.

As m entioned at the sta rt of this section a possible resolution of this problem could be found by assuming a continual heating of the gas in the cores of clusters by some mechanism th a t is capable of injecting vast quantities of energy into the ICM. Powerful AGN are the obvious candidate as the energy injection rate can easily outweigh the radiative cooling of the cluster gas. Also, the cooling flow itself may provide the necessary fuel to keep an AGN active and so a sort of feedback is set up between the cluster gas and the central AGN.

A nother possibility is therm al conduction th a t draws energy in to the centre of the cluster from the hotter outer layers of gas to prevent the tem perature in th e centre dropping enough to start a significant cooling flow. However, because conduction rate increases w ith tem perature it should become less efficient ju st as radiative emission is becoming more so i.e. in the cooler cores. Therefore, if an equilibrium is established between conduction and radiation it will be unstable leading inevitably to one process dom inating over the other, i.e. either a cooling flow or an isotherm al gas distribution. In reality most clusters experience a sort of half-way house in th a t they have a slightly cooler core th a n th e bulk of the X-ray gas, w ith m oderate in-fall rates b ut not a powerful cooling flow. This issue has not yet been fully resolved.

(37)

1.3.3 T h e R e la tiv e E v o lu tio n o f X -r a y /O p tic a l P r o p e r tie s

There is - and needs to be - some distinction betw een galaxy clusters detected via X -ray m eans and ones discovered by optical m ethods. A lthough th e original definition of galaxy clusters arose because th ey were first discovered in optical surveys as over-densities of galaxies, the galaxies them selves are only a sym ptom of th e underlying phenom enon.

D uring th e form ation of a cluster the galaxies and th e hot gas go down very different evolutionary paths. Therefore, a cluster detected as an over-density of optical galaxies m ay not necessarily be detected by X -ray m ethods.

W hen a cluster is first collapsing (or when several smaller dark m a tter clumps come together, in th e hierarchical picture) the m em ber galaxies to be have a high line of sight velocity dispersion due to the rapid in-fall and there is no clear sepa ratio n of different galaxy types. Only once the cluster has had tim e to relax do the galaxies settle down into a well behaved dynam ical state. The equipartition of en ergy between large and small galaxies eventually causes the most massive galaxies, such as giant ellipticals, to fall into the central regions of th e cluster w ith a small velocity dispersion, while th e smaller galaxies rem ain more widely d istrib u ted with a higher velocity dispersion.

So a cluster evolves from a very mixed up, irregular, high a state into a more relaxed, regular, lower a sta te w ith a clear m ass-density relation. O ther evolutionary effects also conspire to suppress star-form at ion in th e galaxies w ith th e highest environm ental density (i.e. the cluster core) so w hat we end up w ith is a core of massive, red, elliptical galaxies surrounded by a halo of less-massive, blue, spiral galaxies. The more evolved a cluster becomes th e lower th e fraction of

(38)

1.3. G A L A X Y C L U ST E R S 21

spiral galaxies becomes and the more relaxed and regular it appears.

The gas, on the other hand, has yet to fully virialise w ith the dark m atter distribution in a young cluster. The outcome of this is th a t although there is an over-density of optical galaxies at the position of the cluster the gas has not achieved its full potential X-ray luminosity. Only when the hot gas has become virialised and has reached the higher central density does the X-ray luminosity and tem perature sta rt to behave in the ways described in section 1.3.

The two effects of high a and low L x mean th a t dynamically young clusters do not follow the expected L — a relation found for local evolved clusters (e.g. Lubin, Mulchaey, & Postm an, 2004). X-ray surveys are therefore most sensitive to dynamically old, evolved clusters while optical techniques can also be sensitive to the younger cluster population. At high redshift there is a higher proportion of young clusters so optically selected samples show a significant difference in the X -ray/optical relations to those found for X-ray selected samples.

1.3.4 T his T h esis

The work in chapter 6 concentrates on m ethods for finding high redshift galaxy clusters. Because different m ethods for finding clusters have different selection effects I employ both X-ray and optical techniques to find clusters w ithin my data. Comparing different m ethods for finding clusters is im portant because evolutionary effects may alter the sensitivity of a particular cluster detection m ethod in ways th a t we don’t yet understand. Any bias th a t affects one technique may be overcome by the use of another technique th a t doesn’t suffer from the same bias.

(39)

tions from our expectations can we truly characterise the selection functions of different cluster finding m ethods. By relying on one m ethod alone we would risk m isinterpreting our findings and missing im p o rtan t factors th a t influence our un derstanding of cluster evolution.

T hroughout this thesis I assume an H 0 of 75 km s-1 M pc-1 and a concordance Universe w ith Dm = 0-3 and Da = 0.7, unless otherwise stated.

(40)

C h ap ter 2

X M M - N e w to n

D a ta A cq u isitio n

and R ed u ctio n

2.1 In tro d u ctio n

As this thesis is based on d a ta taken by the XMM-Newton X-ray space telescope I will outline the d a ta and its basic reduction in this chapter, along w ith th e source detection procedure I used to obtain the catalogues presented in chapter 4. O ther d a ta is used at various stages throughout this thesis b ut I was not involved in its acquisition or reduction. Therefore, I will not present a discussion of these other d a ta here except where it relates directly to this thesis.

(41)

24 C H A P T E R 2. XM M -NEW TON D A T A A C Q U IS IT IO N A N D R E D U C T IO N

2.2 X M M - N e w t o n

O verview

XM M-Newton is an X -ray telescope th a t was launched into a high altitude, long

period orbit in late 1999. O rdinary optics simply can ’t reflect X-rays so like all X -ray telescopes X M M uses special m irror assemblies th a t comprise a series of nested grazing-incidence mirrors, which focus th e X-rays onto th e detectors. There are three m irror assemblies (modules) in to tal, each one focussing X-rays onto a prim ary imaging instrum ent. The imaging cam eras are designed to detect X-rays in th e range ~ 0.15 — 15 keV, although th e range is often restricted to less th an this for useful scientific analysis.

All three imaging cameras are CCD designs, two of which are identical and are essentially th e same technology used to make optical CCDs. These two are the M etal Oxide Semi-conductor CCDs (MOS) and they each receive ~ 44% of the light from their respective m irror modules, th e rest being diverted to Reflection G rating Spectrom eters (RGS) for high resolution X -ray spectroscopy (I offer no further details on th e RGS instrum ents here as they are irrelevant to this work a p a rt from the reduction in the flux the two MOS instrum ents experience because of them ). The rem aining X-ray CCD is of a different design to th e MOS cameras and sits in the unobstructed beam of the th ird and final m irror module. This CCD is of the pn design, which offers superior sensitivity over th e MOS design, particularly at higher photon energies.

All three cam eras observe approxim ately th e same area of sky in each exposure and operate simultaneously, as do the two RGS instrum ents. In addition to the X -ray instrum entation a small optical/U V telescope, the O ptical M onitor (OM), also observes p a rt of the same field as the prim ary cam eras and is used principally for obtaining sim ultaneous UV d a ta of the X-ray sources under study. This si

(42)

2.3. X - R A Y D ATA 25

multaneous operation of all the instrum entation on X M M gives it a big advantage over other X-ray telescopes, for example Chandra, which tend to offer a suite of instrum entation of which only one can be used during any given exposure.

2.3 X -ray D a ta

Two main X M M surveys are considered in this work, X-ray surveys of the Canada- France Redshift Survey (CFRS) 3 and 14-h (also known as the G roth Strip) fields (Lilly et al., 1995b). The 14-h X M M d a ta was first presented in Miyaji &; Griffiths

(2001).

The d a ta for the 3-h field were taken on 17th February 2001 by XMM-Newton over a period of 51.5 ks, using the thin optical blocking filters and in full frame imaging mode. All three prim ary instrum ents gathered d a ta (MOS 1, MOS 2 & PN) as well as the OM telescope. This field is centred on R.A. 03:02:38.60 Dec. +00:07:40.0.

The 14hr field d a ta was obtained from the public archive after the proprietary period had expired, to extend the coverage of available X M M d a ta for the CFRS fields. This d a ta was first presented in Miyaji &; Griffiths (2001) and later in Miyaji et al. (2003) and Miyaji et al. (2004). Of the several available exposures of this field, one was selected th a t most closely m atched the exposure of th e 3-h field. The exposure was taken over 56.1 ks, using thin filters, and is centred on R.A. 14:17:12.0 Dec. +52:24:00.0.

In addition to the two m ain surveys presented in this thesis a th ird survey is also presented, albeit in less detail because of the lack of deep optical coverage for

(43)

26 C H A P T E R 2. XM M -NEW TON D A TA A C Q U IS IT IO N A N D R E D U C T IO N

this field. This th ird field is coincident w ith the 10-h CFRS field and was similarly surveyed by X M M for 50.8 ks, using the th in filter for th e PN instrum ent and the m edium filter for th e two MOS instrum ents. This field is centred on R.A 10:00:40.4 Dec. +25:14:20.0. I do not discuss the O ptical M onitor d a ta here, for any of the above surveys.

2.4 X -ray D a ta R ed u ctio n

T he XMM-Newton raw d a ta were processed using version 5.3 of an ensemble of tasks collectively titled the Science Analysis System (SAS). These tasks allow re running of basic pipeline processes as well as fu rth er d a ta reduction tasks.

2.4.1 C rea tin g E ven t F iles

T he raw d a ta files are labelled correspond to th e different instrum ents on board

X M M . PN refers to th e PN instrum ent, M l to th e MOS1 in strum ent etc. Each

instrum ent has several files, each corresponding to one of th e CCD chips on the arrays th a t make up th e detector (7 for each MOS instrum ent and 12 for the PN), plus a few housekeeping files. These need to be processed into a single calibrated photon event file which can then be used to create images, ra te curves etc. To do this there are two tasks th a t need to be run.

epproc

emproc

(44)

2.4. X - R A Y D ATA R E D U C T IO N 27

two event files, one for each MOS instrum ent. In addition, they create an attitu d e history file which is im portant for many other tasks. The photon event files record the time, position and energy of each photon incident on the detectors and are the basis for all further processing. These tasks also remove hot and flickering pixels and columns which would otherwise contam inate the data.

Figure 2.1 shows an image produced by using every event in the PN event file for the 3-h field. This example illustrates how heavily contam inated the event file is before it is properly filtered.

2.4.2 F ilterin g th e D a ta

B asic F iltering and im age G eneration

It is possible to create an image directly from the initial event files, however this will not produce useful results. Filtering of the event files is essential to obtain usable data, and so this is the next step. Non X-ray associated events such as cosmic rays create pattern s on th e detectors th a t look different from th e im pacting of X-rays. These events can be flagged and filtered out easily. On the other hand soft protons, produced by the sun and projected towards E arth in solar flares, produce pattern s th a t look identical to X-rays, so these events need more careful attention.

The first p art of the filtering process involves removing all the events th a t do not look like X-rays. This is simply a m atter of screening out events from the event file th a t have flags indicating a non-X-ray event p attern. The event file is left containing only X-rays and soft protons. At this stage it is also prudent to

(45)

28 C H APTE R 2. XMM-NEWTON DATA A C Q U ISIT IO N A N D R E D U C T IO N

Figure 2.1: Unfiltered PN image. The arrangement of the PN CCD chips can be clearly seen in this image, which is aligned in R.A and Dec coordinates. The contrast is loga rithmically scaled.

reduce the number of events in the file by retaining only those w ith energies in a sensible range. In this case I have chosen 0.5 — 10.0 keV, despite the fact th a t the d ata includes a much broader range of energies ( ~ 0.15 — 15 keV). This is mainly because the instrum ents operate most effectively in th e m id-range bu t it also removes some low energy Galactic contamination. The lowest energy is also most strongly attenuated by Galactic absorption, so anything below 0.5 keV is